paperID
stringlengths 36
36
| pwc_id
stringlengths 8
47
| arxiv_id
stringlengths 6
16
⌀ | nips_id
float64 | url_abs
stringlengths 18
329
| url_pdf
stringlengths 18
742
| title
stringlengths 8
325
| abstract
stringlengths 1
7.27k
⌀ | authors
stringlengths 2
7.06k
| published
stringlengths 10
10
⌀ | conference
stringlengths 12
47
⌀ | conference_url_abs
stringlengths 16
198
⌀ | conference_url_pdf
stringlengths 27
199
⌀ | proceeding
stringlengths 6
47
⌀ | taskID
stringlengths 7
1.44k
| areaID
stringclasses 688
values | embedding
stringlengths 9.26k
12.5k
| umap_embedding
stringlengths 29
44
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
f4a37d10-3b56-4856-b932-e72e7018eaa0
|
optimizing-readability-using-genetic
|
2301.00374
| null |
https://arxiv.org/abs/2301.00374v1
|
https://arxiv.org/pdf/2301.00374v1.pdf
|
Optimizing Readability Using Genetic Algorithms
|
This research presents ORUGA, a method that tries to automatically optimize the readability of any text in English. The core idea behind the method is that certain factors affect the readability of a text, some of which are quantifiable (number of words, syllables, presence or absence of adverbs, and so on). The nature of these factors allows us to implement a genetic learning strategy to replace some existing words with their most suitable synonyms to facilitate optimization. In addition, this research seeks to preserve both the original text's content and form through multi-objective optimization techniques. In this way, neither the text's syntactic structure nor the semantic content of the original message is significantly distorted. An exhaustive study on a substantial number and diversity of texts confirms that our method was able to optimize the degree of readability in all cases without significantly altering their form or meaning. The source code of this approach is available at https://github.com/jorge-martinez-gil/oruga.
|
['Jorge Martinez-Gil']
|
2023-01-01
| null | null | null | null |
['readability-optimization']
|
['natural-language-processing']
|
[ 1.73193276e-01 2.28036389e-01 -1.65111169e-01 -9.55327824e-02
-2.31281370e-01 -6.40375376e-01 4.20259595e-01 5.41396201e-01
-4.29936737e-01 7.12398589e-01 5.18135548e-01 -2.92450428e-01
-3.11202049e-01 -8.43926251e-01 -4.28842455e-01 -4.04438347e-01
6.16595089e-01 1.65184200e-01 -9.72711071e-02 -5.46041429e-01
8.70789886e-01 2.86913246e-01 -1.88699770e+00 6.74247891e-02
1.08955026e+00 4.41627473e-01 5.12980163e-01 5.81046939e-01
-4.46713775e-01 5.18613636e-01 -6.63118303e-01 -3.50122422e-01
2.60388255e-02 -6.29043341e-01 -9.28053319e-01 8.39381106e-03
1.00026771e-01 -3.87917012e-02 -1.45627230e-01 1.11049986e+00
4.65380400e-01 9.12530124e-02 7.17269957e-01 -6.88973248e-01
-8.71875107e-01 8.22780609e-01 -8.70098695e-02 2.70926446e-01
7.68679440e-01 -9.97037441e-02 1.16502798e+00 -6.14218652e-01
3.75940293e-01 1.11068022e+00 2.85887390e-01 4.15242128e-02
-9.54731882e-01 -2.45376676e-01 -9.03558917e-03 2.27836192e-01
-1.53610003e+00 -3.97827208e-01 6.62537813e-01 -3.90326768e-01
9.27332401e-01 6.00438416e-01 8.52359056e-01 7.42688894e-01
6.82134211e-01 2.73786396e-01 1.17017090e+00 -1.16549015e+00
2.07745448e-01 3.67775738e-01 2.13529855e-01 6.69674337e-01
3.97706836e-01 -6.17286116e-02 -4.58994001e-01 1.32066831e-01
8.50705802e-02 -4.50608373e-01 -2.83854574e-01 7.88723454e-02
-1.05574453e+00 8.58922243e-01 -1.63348362e-01 8.46123457e-01
-1.72658652e-01 -4.64481950e-01 2.39484817e-01 3.97910237e-01
2.84534812e-01 9.56784010e-01 -4.90330875e-01 -4.55440283e-01
-7.81455636e-01 5.93125522e-02 1.05244172e+00 6.70007586e-01
7.02422798e-01 -1.15409203e-01 -2.80052394e-01 7.77331352e-01
2.74525434e-01 6.35932207e-01 8.33278775e-01 -5.20168126e-01
3.88807654e-01 9.36480165e-01 1.08505115e-02 -1.42649603e+00
-3.54867011e-01 -3.02536488e-01 -3.76034468e-01 -8.39319918e-03
3.36024851e-01 -1.53534636e-01 -5.87162375e-01 1.42594266e+00
7.75108412e-02 -7.44907558e-01 -1.98636815e-01 5.96854806e-01
8.37350011e-01 7.78547347e-01 -1.98383242e-01 -3.40469152e-01
1.39614499e+00 -8.15178156e-01 -1.07975793e+00 -3.71563137e-01
3.92913878e-01 -1.07649457e+00 1.43361378e+00 1.24489129e-01
-1.11826813e+00 -4.54635561e-01 -1.14377189e+00 -7.38790073e-03
-6.18975520e-01 3.22988957e-01 2.09198475e-01 1.01091635e+00
-8.12444270e-01 3.48540306e-01 -3.68878245e-01 -3.89991939e-01
-3.62323403e-01 2.52086997e-01 -4.63803485e-02 3.92459184e-01
-1.18671453e+00 1.24703109e+00 7.32888281e-01 -4.23015989e-02
-2.93685887e-02 -1.51429147e-01 -7.49904692e-01 1.51581541e-01
3.96591842e-01 -4.38926846e-01 9.73975420e-01 -1.41722882e+00
-1.71955693e+00 8.61234665e-01 -3.37310791e-01 1.73725590e-01
3.65563661e-01 -4.33274060e-02 -4.36654210e-01 -5.82332676e-03
3.70904803e-02 1.80499017e-01 7.85411656e-01 -8.75839293e-01
-5.68897069e-01 -2.68235505e-01 3.63686755e-02 4.53228354e-01
-7.30709195e-01 1.54838696e-01 -3.81311893e-01 -8.63301754e-01
-9.54800323e-02 -8.29691648e-01 2.06962362e-01 -3.82083625e-01
-4.34371263e-01 -2.65133262e-01 2.68993795e-01 -1.14774680e+00
1.76380956e+00 -2.06365037e+00 4.11066324e-01 4.05413568e-01
2.03741509e-02 1.96331859e-01 -7.14740455e-02 6.52468860e-01
1.79984450e-01 4.49239105e-01 -3.59983295e-02 3.38493347e-01
1.04309954e-01 2.04160050e-01 4.10284847e-01 2.43267924e-01
-6.99581951e-02 7.57823348e-01 -7.67168403e-01 -4.33180511e-01
1.39219388e-01 3.74235719e-01 -4.20684218e-01 7.13898465e-02
-2.14202940e-01 5.96919879e-02 -3.75367761e-01 4.86837775e-01
3.87530714e-01 -1.30184263e-01 1.71053544e-01 1.07902363e-01
-3.75427991e-01 5.97455025e-01 -1.25806737e+00 1.05056345e+00
-4.83541071e-01 7.07683086e-01 -3.03002059e-01 -5.46150446e-01
9.74210143e-01 2.03951985e-01 2.33429447e-01 -1.11363482e+00
3.79946053e-01 1.70083418e-01 1.39562547e-01 -7.85974562e-01
8.74168932e-01 2.51161456e-01 1.72932714e-03 5.64706683e-01
-5.14158189e-01 -2.80053407e-01 7.75064528e-01 -9.67454985e-02
6.01255059e-01 -7.41632506e-02 8.34527969e-01 -4.67335165e-01
8.06924522e-01 -1.04518376e-01 3.83224696e-01 5.88847280e-01
9.68836546e-02 4.54331547e-01 5.09214461e-01 1.32046789e-01
-1.31898522e+00 -7.97199905e-01 -2.08955407e-01 1.05245996e+00
1.43849868e-02 -5.01962066e-01 -1.02984583e+00 -7.20492527e-02
-2.12077990e-01 1.21208584e+00 -5.29595196e-01 -2.96057045e-01
-3.85577083e-01 -5.87340891e-01 3.38939577e-01 -7.69318687e-03
1.91968113e-01 -1.09912896e+00 -7.00298309e-01 2.49455541e-01
-5.36159754e-01 -4.99581933e-01 -4.29945707e-01 3.42639200e-02
-6.17897868e-01 -8.28249753e-01 -4.06633824e-01 -7.54089475e-01
7.78462529e-01 -5.87422848e-02 1.01144075e+00 5.71920633e-01
-9.71925184e-02 1.77143335e-01 -9.17187810e-01 -5.31544566e-01
-8.86254489e-01 4.29886192e-01 -3.87765199e-01 -3.74693155e-01
3.19303066e-01 -8.47776085e-02 -5.24861291e-02 1.14006981e-01
-1.07916808e+00 -1.33101180e-01 5.32246709e-01 6.21027172e-01
3.27494085e-01 5.24235070e-01 3.01729143e-01 -5.64464271e-01
1.25524521e+00 -2.47476906e-01 -5.24300873e-01 4.93958026e-01
-9.65450704e-01 2.38390639e-01 5.22380829e-01 -3.02333802e-01
-8.29211712e-01 -2.38707691e-01 -2.67757058e-01 6.96132064e-01
-1.38666406e-01 7.94194639e-01 -3.35072219e-01 8.03744197e-02
5.86240411e-01 3.79731685e-01 1.86331689e-01 -3.01207691e-01
1.40954956e-01 8.61487031e-01 6.35269806e-02 -2.99938798e-01
6.54761374e-01 -1.22726455e-01 -4.70649689e-01 -9.65076029e-01
-5.78055918e-01 -2.69602954e-01 -8.36072683e-01 -3.12068760e-01
6.31145537e-01 -5.76117933e-01 -2.18056053e-01 5.78117013e-01
-8.15394163e-01 -9.11251456e-02 -1.39713660e-02 3.65171283e-01
-2.28328720e-01 4.35671002e-01 -2.59813696e-01 -6.22669578e-01
-2.38755748e-01 -1.08227396e+00 5.78162372e-01 4.08215940e-01
-7.18384683e-01 -1.14732444e+00 -1.80713400e-01 6.07067466e-01
4.61786181e-01 -1.79929569e-01 1.44552243e+00 -5.49233198e-01
-5.99235408e-02 -1.57640785e-01 9.39002931e-02 3.61328185e-01
4.53822404e-01 3.77167106e-01 -3.37880194e-01 -2.10685000e-01
9.49190408e-02 1.38158351e-01 2.39694238e-01 1.58362940e-01
6.52207971e-01 -6.62656546e-01 1.59563318e-01 2.11343989e-01
1.43921697e+00 3.91477793e-01 8.03068936e-01 9.00147080e-01
3.61620039e-01 4.88673568e-01 6.29351258e-01 7.16108680e-01
3.42891425e-01 7.11190045e-01 2.59509772e-01 1.94433540e-01
-5.45909666e-02 -9.27431956e-02 5.94322145e-01 1.03289378e+00
7.20747933e-03 -6.04515254e-01 -8.73548329e-01 3.92688096e-01
-1.30662870e+00 -9.48466420e-01 -2.00941041e-01 2.04928827e+00
9.41162705e-01 1.32440120e-01 5.23917079e-02 4.06120121e-01
7.39876568e-01 8.39251652e-02 -2.99005639e-02 -1.00833261e+00
-3.81152749e-01 -7.98870772e-02 3.85833234e-01 6.96401477e-01
-7.00330794e-01 7.48783231e-01 6.35878801e+00 6.70329928e-01
-1.13981867e+00 -3.65764290e-01 4.36533391e-01 -3.66730541e-02
-7.10610926e-01 -1.01145804e-01 -7.63048470e-01 7.32501030e-01
8.71117353e-01 -7.22765625e-01 7.12449789e-01 3.41856718e-01
7.64102936e-01 -5.05784214e-01 -5.43677092e-01 4.82728630e-01
4.07328814e-01 -9.53107595e-01 4.35832739e-02 -9.99970213e-02
6.88662767e-01 -5.94765842e-01 4.05357480e-02 -8.27719271e-02
-2.82641858e-01 -1.03403389e+00 1.17450082e+00 6.26737297e-01
3.77932817e-01 -7.78554142e-01 8.78982902e-01 3.66083205e-01
-6.23617768e-01 -1.75676450e-01 -2.55731672e-01 -3.52207273e-01
-1.17149755e-01 6.28088295e-01 -5.45458198e-01 2.94848204e-01
5.31922698e-01 2.06765398e-01 -1.09332502e+00 9.82638597e-01
-6.40794516e-01 6.43187821e-01 -1.84712723e-01 -7.31498837e-01
6.66584149e-02 -4.02141690e-01 8.19067419e-01 1.24508059e+00
5.23044467e-01 -4.57624458e-02 1.56541262e-03 7.33844161e-01
1.65449619e-01 8.62121165e-01 -2.75264680e-01 -3.91835392e-01
6.30921125e-01 8.46587777e-01 -8.37642789e-01 -6.26572117e-04
-3.12026978e-01 8.32645357e-01 1.51051626e-01 1.70793667e-01
-7.22632051e-01 -5.24390817e-01 1.57061890e-01 2.22601697e-01
2.97855437e-01 -2.40126066e-02 -8.15000534e-01 -8.80055904e-01
3.31616402e-01 -1.51399302e+00 7.18949884e-02 -6.16011620e-01
-8.00218761e-01 4.44048554e-01 -1.66469291e-01 -7.80881226e-01
-5.80231622e-02 -4.67716604e-01 -4.42848861e-01 9.27820981e-01
-1.06735861e+00 -8.49367082e-01 -1.33209482e-01 2.07037315e-01
5.30218899e-01 -3.38209152e-01 6.45973325e-01 9.31906328e-02
-6.82153940e-01 5.83722293e-01 5.05851090e-01 -2.73055471e-02
5.74943960e-01 -1.14846504e+00 -8.81444588e-02 9.78006840e-01
-1.11229979e-01 6.03982806e-01 1.00785482e+00 -7.27187872e-01
-1.08891010e+00 -4.47257966e-01 1.72496605e+00 -8.78704116e-02
5.96516907e-01 -5.14602661e-02 -6.20406747e-01 2.26881161e-01
6.47125363e-01 -1.28386784e+00 6.43279791e-01 1.52486935e-01
1.91980246e-02 8.27208534e-02 -9.80625749e-01 8.85892510e-01
3.95801187e-01 -4.32360291e-01 -9.12203133e-01 3.01565528e-01
3.74555111e-01 -2.49436304e-01 -7.46652186e-01 -6.36887699e-02
4.17567849e-01 -1.18469620e+00 5.18503666e-01 -9.85221192e-02
5.84357381e-01 -2.86145627e-01 -3.62100825e-02 -1.42130637e+00
-5.47310531e-01 -4.44858879e-01 2.74767548e-01 1.24291813e+00
8.39260995e-01 -7.43682563e-01 4.39799756e-01 7.15957701e-01
-4.50694636e-02 -4.86329168e-01 -7.05855310e-01 -7.23573267e-01
2.08670959e-01 -4.62945085e-03 6.90060437e-01 7.02545106e-01
2.95368731e-01 3.53978634e-01 -2.38027751e-01 -1.01058759e-01
3.32334489e-02 -9.27382112e-02 3.87685537e-01 -8.65317106e-01
-1.87564358e-01 -7.99610317e-01 -3.01861644e-01 -5.18954694e-01
1.75924048e-01 -8.39895010e-01 -1.32536143e-02 -1.54484522e+00
2.46488273e-01 1.76265817e-02 2.82910708e-02 4.08567667e-01
-5.01237273e-01 -1.56843856e-01 4.51862276e-01 -4.40283380e-02
-2.20854566e-01 3.45786631e-01 1.27507067e+00 -1.78073242e-01
-4.88541216e-01 7.42071718e-02 -9.71708655e-01 4.12075698e-01
1.27036273e+00 -5.64596117e-01 -4.04135615e-01 -5.32044351e-01
6.49587870e-01 -3.15602243e-01 -1.82373170e-02 -7.44561315e-01
1.19408660e-01 -4.64196354e-01 1.56000406e-01 -2.39283130e-01
-6.44932315e-02 -7.81194985e-01 3.13203305e-01 5.61711788e-01
-4.91911560e-01 6.58751607e-01 2.19868392e-01 -5.08616455e-02
-2.47600466e-01 -1.02841079e+00 6.91652477e-01 -3.62412818e-02
-7.10995138e-01 -4.27777976e-01 -1.02973294e+00 9.12229344e-02
1.01352727e+00 -5.25249183e-01 -6.06947541e-02 -3.95653993e-01
-3.54164392e-01 -8.61277580e-02 7.79739499e-01 6.46282494e-01
2.72711009e-01 -8.98848236e-01 -7.88226664e-01 1.72160015e-01
-1.40483370e-02 -6.92985654e-01 -1.62050575e-01 5.64922690e-01
-9.67353761e-01 5.49619913e-01 -4.14804012e-01 8.13639760e-02
-1.48001540e+00 4.02100563e-01 2.86859840e-01 -1.36733249e-01
-2.84336597e-01 4.37667042e-01 -5.64812601e-01 -3.70839387e-01
9.89878029e-02 -1.79239184e-01 -5.02562284e-01 4.31790382e-01
4.79121953e-01 5.72551072e-01 1.89222619e-01 -7.64032364e-01
-1.42296508e-01 4.58422571e-01 1.03618152e-01 -7.95166790e-02
1.13272321e+00 -6.16356730e-01 -5.11025369e-01 4.76321906e-01
1.07719600e+00 5.30576944e-01 -4.63034302e-01 1.98108181e-01
1.28223345e-01 -6.77025080e-01 1.61055997e-01 -1.02570117e+00
-7.45691121e-01 2.87159652e-01 3.62612456e-01 2.90382266e-01
1.17197919e+00 -2.95636445e-01 4.97549057e-01 4.54647273e-01
-1.47925377e-01 -1.62316334e+00 -1.26492251e-02 9.76893783e-01
8.56344342e-01 -9.33411837e-01 1.05410255e-01 -3.10615748e-01
-5.84490001e-01 1.33698046e+00 2.99842358e-01 5.47309339e-01
5.57183981e-01 1.87424347e-01 2.40304768e-01 1.15151726e-01
-4.48310971e-01 -7.34520257e-02 4.40226644e-01 4.82595474e-01
8.39539170e-01 1.01218216e-01 -1.24581265e+00 1.52198508e-01
-8.08085263e-01 -3.18163306e-01 8.23020577e-01 7.97869384e-01
-7.55916417e-01 -1.28895843e+00 -6.31927371e-01 4.44875568e-01
-5.01504481e-01 -2.35847250e-01 -7.11047709e-01 6.63997769e-01
1.70992255e-01 1.36542392e+00 -1.63310915e-01 -3.18990916e-01
3.50865930e-01 1.78236231e-01 3.12153608e-01 -5.21594644e-01
-7.94215798e-01 6.42522201e-02 2.11002678e-01 3.89033146e-02
-1.65336907e-01 -9.00592983e-01 -1.08111489e+00 -7.59398997e-01
-4.24669415e-01 2.63348609e-01 8.16038847e-01 1.14670932e+00
3.08524072e-02 4.40019161e-01 5.10570943e-01 -9.19694006e-02
-6.40283346e-01 -7.84641147e-01 -5.06906569e-01 4.10283893e-01
1.94408260e-02 -2.90673137e-01 -3.73073906e-01 1.10041797e-01]
|
[11.0912446975708, 10.106895446777344]
|
5c3dfb49-ca42-4e9a-8fc5-d86275b85f2e
|
an-efficient-transformer-for-simultaneous
|
2306.04927
| null |
https://arxiv.org/abs/2306.04927v1
|
https://arxiv.org/pdf/2306.04927v1.pdf
|
An Efficient Transformer for Simultaneous Learning of BEV and Lane Representations in 3D Lane Detection
|
Accurately detecting lane lines in 3D space is crucial for autonomous driving. Existing methods usually first transform image-view features into bird-eye-view (BEV) by aid of inverse perspective mapping (IPM), and then detect lane lines based on the BEV features. However, IPM ignores the changes in road height, leading to inaccurate view transformations. Additionally, the two separate stages of the process can cause cumulative errors and increased complexity. To address these limitations, we propose an efficient transformer for 3D lane detection. Different from the vanilla transformer, our model contains a decomposed cross-attention mechanism to simultaneously learn lane and BEV representations. The mechanism decomposes the cross-attention between image-view and BEV features into the one between image-view and lane features, and the one between lane and BEV features, both of which are supervised with ground-truth lane lines. Our method obtains 2D and 3D lane predictions by applying the lane features to the image-view and BEV features, respectively. This allows for a more accurate view transformation than IPM-based methods, as the view transformation is learned from data with a supervised cross-attention. Additionally, the cross-attention between lane and BEV features enables them to adjust to each other, resulting in more accurate lane detection than the two separate stages. Finally, the decomposed cross-attention is more efficient than the original one. Experimental results on OpenLane and ONCE-3DLanes demonstrate the state-of-the-art performance of our method.
|
['Mingming Gong', 'Guoqi Qian', 'Bo Du', 'Kate Smith-Miles', 'Ziye Chen']
|
2023-06-08
| null | null | null | null |
['3d-lane-detection', 'lane-detection']
|
['computer-vision', 'computer-vision']
|
[-2.96471845e-02 1.16168462e-01 -1.35128379e-01 -7.15233266e-01
-5.43568611e-01 -4.26543742e-01 7.11569071e-01 -4.14366513e-01
-1.85953110e-01 2.97879726e-01 -8.26053508e-03 -5.31821311e-01
2.66958326e-01 -1.00760818e+00 -9.26992834e-01 -5.54001451e-01
5.79871833e-01 2.15945140e-01 6.27868414e-01 -4.57186520e-01
4.29389656e-01 6.37802303e-01 -1.70997715e+00 -3.40584517e-02
9.87273514e-01 8.53616178e-01 4.31485593e-01 4.50081259e-01
-1.97561890e-01 7.18307137e-01 8.43729228e-02 -5.52755259e-02
4.00359660e-01 -5.60686663e-02 -1.21102586e-01 4.45612282e-01
7.50790536e-01 -6.57554209e-01 -5.45816839e-01 1.00775516e+00
1.08962476e-01 7.51594380e-02 5.86485326e-01 -1.48182714e+00
-3.94360125e-01 -2.76429236e-01 -9.76137877e-01 -3.25536951e-02
3.32124770e-01 2.58448422e-01 7.62617052e-01 -1.16900146e+00
4.24901903e-01 1.44748497e+00 5.20622194e-01 1.34349450e-01
-1.08179772e+00 -7.85334349e-01 4.85259622e-01 4.44671959e-01
-1.08546662e+00 -4.34364408e-01 1.10486519e+00 -6.61176443e-01
7.79883385e-01 -1.75188128e-02 7.37140477e-01 5.67650497e-01
5.54350734e-01 7.27014959e-01 1.20979798e+00 -2.16589421e-01
-3.74896437e-01 2.87505776e-01 1.14942603e-01 1.02169073e+00
2.43900597e-01 4.43929821e-01 -1.63390115e-01 4.80807364e-01
9.90866542e-01 3.27867955e-01 -3.97833556e-01 -9.45578754e-01
-1.30772829e+00 8.59305680e-01 5.18603325e-01 -2.97806561e-01
-1.28812104e-01 -7.59867951e-02 1.57187402e-01 1.14224486e-01
3.34868252e-01 1.63377106e-01 -2.38253906e-01 2.23290175e-01
-5.02174318e-01 2.93967482e-02 2.61510521e-01 1.17181635e+00
1.33554053e+00 6.39411509e-02 2.47764006e-01 6.47270083e-01
5.54573059e-01 9.17859197e-01 1.66286856e-01 -9.66591060e-01
8.22312474e-01 6.80474281e-01 1.13062516e-01 -1.36743379e+00
-4.95330811e-01 -3.40930820e-01 -7.05124855e-01 5.35222292e-01
2.88549632e-01 1.62563965e-01 -1.00650513e+00 1.43712294e+00
2.48327300e-01 3.55975889e-02 1.02414444e-01 9.74512815e-01
5.38547337e-01 6.86766803e-01 -5.20431221e-01 3.52566317e-02
1.25773871e+00 -1.25606501e+00 -7.66909361e-01 -8.48499179e-01
6.32312775e-01 -7.76942730e-01 9.79974806e-01 -4.05822657e-02
-7.81594872e-01 -9.14911687e-01 -1.26172543e+00 -4.60177481e-01
-3.96494627e-01 2.56050587e-01 3.59318376e-01 1.16077691e-01
-8.15729797e-01 -1.63541690e-01 -4.52299744e-01 -2.07262143e-01
-1.77217424e-02 1.07240215e-01 -5.79816818e-01 -1.73962697e-01
-1.24600899e+00 1.01969385e+00 1.22083858e-01 4.19068664e-01
-5.62771618e-01 -5.58774471e-01 -1.38393486e+00 -8.18756297e-02
5.56675732e-01 -6.84150577e-01 9.21678185e-01 -6.06845975e-01
-1.54274881e+00 9.69495654e-01 -4.99687672e-01 -3.04014266e-01
7.02988803e-01 -2.46728331e-01 -4.24997568e-01 -3.24825458e-02
4.57916647e-01 7.38267779e-01 8.03031743e-01 -1.64869380e+00
-1.17901957e+00 -4.03942019e-01 1.48077935e-01 4.57011163e-01
3.53482515e-01 -8.33480954e-01 -5.96724987e-01 -8.82003754e-02
7.83169091e-01 -8.76866639e-01 -1.74863786e-01 2.50529237e-02
-3.82881075e-01 8.73260871e-02 1.04970181e+00 -5.61690509e-01
9.25179303e-01 -2.12963200e+00 -1.06787302e-01 5.02736010e-02
4.05268490e-01 -1.25071153e-01 -6.73149079e-02 1.71037734e-01
-1.22021280e-01 -2.20634952e-01 4.87558432e-02 -2.94793230e-02
-2.74470657e-01 1.87482253e-01 -6.38877392e-01 5.38364649e-01
1.73798397e-01 8.48425627e-01 -8.80503535e-01 -2.93618530e-01
6.92718029e-01 2.00639620e-01 -4.69337702e-01 2.65255809e-01
1.80440277e-01 4.95752722e-01 -4.80803549e-01 2.32058570e-01
9.56797063e-01 1.24595627e-01 -2.86129475e-01 -5.31368256e-01
-6.24870718e-01 2.97303736e-01 -8.57246280e-01 1.26642549e+00
-7.88637578e-01 8.45084727e-01 -1.30993590e-01 -7.76381254e-01
1.33670306e+00 -2.17669323e-01 1.61016777e-01 -9.39957440e-01
6.44608885e-02 1.27158016e-01 -1.16173796e-01 -4.58793968e-01
4.47102487e-01 4.65642475e-02 -2.77460158e-01 2.20129341e-02
-2.64793724e-01 -4.38872814e-01 -1.66153997e-01 2.10234299e-02
3.67061079e-01 2.96708733e-01 3.94432515e-01 4.89271954e-02
9.60688055e-01 -6.20724969e-02 7.21737027e-01 4.35317785e-01
-1.99165016e-01 4.38642263e-01 5.90791821e-01 -6.76291823e-01
-9.96628225e-01 -1.08138943e+00 4.61244956e-02 4.38053727e-01
8.50978374e-01 6.44084513e-02 -5.28412282e-01 -8.71690214e-01
2.41448030e-01 8.93302441e-01 -5.52833855e-01 -3.43045652e-01
-6.82179689e-01 -9.46675520e-03 1.62277073e-02 5.93120515e-01
8.86619151e-01 -4.96997535e-01 -7.85759628e-01 9.74060223e-02
-1.75917566e-01 -1.36682594e+00 -8.73934031e-01 -9.14728492e-02
-7.43039727e-01 -1.02266240e+00 -2.34213412e-01 -6.93433940e-01
8.14585149e-01 1.01902819e+00 7.69550204e-01 -1.85028717e-01
2.61497974e-01 2.35179871e-01 -1.03179060e-01 -3.87137771e-01
-2.68088847e-01 -1.42976761e-01 1.03047289e-01 3.54661137e-01
3.61759722e-01 -5.11649013e-01 -5.20222664e-01 6.53780401e-01
-2.10074201e-01 7.54004955e-01 8.32039654e-01 1.00701904e+00
7.17660308e-01 -3.04354966e-01 2.57128656e-01 -8.96543443e-01
-5.63931577e-02 -2.86374629e-01 -1.09470665e+00 -1.32362589e-01
-7.92248547e-01 8.70615523e-03 6.78471208e-01 5.93276881e-02
-1.22404838e+00 3.86911482e-01 -3.30417156e-02 -8.25744450e-01
-2.05457270e-01 1.63677111e-01 -4.95167404e-01 3.13487276e-02
1.35284945e-01 4.99212027e-01 2.74891734e-01 -1.78952023e-01
4.86761034e-01 5.42113185e-01 6.10242426e-01 7.35488757e-02
1.21385622e+00 7.56526530e-01 5.18466495e-02 -8.01317751e-01
-9.85140204e-01 -4.29377675e-01 -1.12675619e+00 -5.43625653e-01
1.08719754e+00 -1.07471347e+00 -5.54042041e-01 4.52256143e-01
-9.98988509e-01 -6.89971596e-02 -6.24624640e-02 5.80496371e-01
-7.69698024e-01 4.16392893e-01 -2.18586415e-01 -6.11860037e-01
1.82561457e-01 -1.48160315e+00 1.27554929e+00 2.60637015e-01
3.66805822e-01 -8.97551894e-01 -2.83342987e-01 5.02239108e-01
-4.04225178e-02 7.64839724e-02 9.22452271e-01 -1.50869310e-01
-1.02061307e+00 -2.87595391e-01 -5.33519924e-01 2.14526638e-01
1.93466604e-01 -1.32338822e-01 -1.08174920e+00 -1.35003045e-01
3.42445225e-02 1.82062596e-01 9.84743953e-01 3.51108342e-01
7.80366480e-01 2.50488054e-02 -5.92364073e-01 9.53427255e-01
1.23280144e+00 4.89655674e-01 5.54779947e-01 6.34395301e-01
1.09102929e+00 7.24122643e-01 1.06002665e+00 -7.77497068e-02
9.13190424e-01 8.01188767e-01 6.26282215e-01 -5.42955160e-01
5.34132682e-02 -7.69702733e-01 4.16941226e-01 8.62463117e-01
9.19562131e-02 8.51388350e-02 -9.04920936e-01 3.95782858e-01
-1.82920837e+00 -9.67544377e-01 -4.49742138e-01 2.30195570e+00
6.11715019e-02 4.68481302e-01 -1.75695688e-01 -4.65862453e-02
5.45321405e-01 4.32059258e-01 -7.55429387e-01 -4.14786816e-01
1.28315076e-01 -9.57722008e-01 8.91253769e-01 8.44549119e-01
-1.14102376e+00 9.86363351e-01 5.09169436e+00 4.93531168e-01
-1.36922729e+00 -1.02693081e-01 5.26276350e-01 3.94191474e-01
-4.87556130e-01 1.86766520e-01 -1.13547444e+00 3.31085056e-01
2.79916734e-01 7.72633543e-03 3.95479739e-01 8.50366950e-01
4.86415386e-01 -1.07674919e-01 -1.01960671e+00 9.99658465e-01
3.18364471e-01 -1.11696684e+00 4.83410768e-02 2.57231474e-01
5.12438953e-01 3.85858454e-02 -7.01598004e-02 3.76015484e-01
6.80815056e-02 -5.46819329e-01 1.10497046e+00 6.56787217e-01
8.62278640e-01 -6.63558304e-01 7.77996361e-01 6.24905050e-01
-1.57296014e+00 -1.73328131e-01 -2.90346444e-01 -2.14241687e-02
4.13678229e-01 4.70730722e-01 -7.58707762e-01 7.13855743e-01
5.16809881e-01 1.13465071e+00 -5.58936238e-01 4.57145780e-01
-3.05344462e-01 8.45793560e-02 -1.13558061e-01 2.98489749e-01
3.58386755e-01 -7.10828304e-01 7.89250612e-01 7.68365502e-01
2.83150941e-01 -1.46785602e-01 3.86040717e-01 9.79326785e-01
3.96798700e-01 -2.40518570e-01 -1.22825432e+00 7.26717710e-01
4.18895185e-01 1.19259512e+00 -3.66691589e-01 -3.13412309e-01
-8.16158712e-01 7.62665391e-01 3.55134666e-01 4.98378903e-01
-9.57757711e-01 -5.09003580e-01 7.30410278e-01 5.53143740e-01
3.16722125e-01 -3.24131906e-01 -1.77039832e-01 -1.17987335e+00
1.13188326e-01 -3.72348726e-01 -1.07433610e-01 -1.03477693e+00
-1.09778106e+00 5.76526642e-01 1.14858188e-01 -1.90057039e+00
-1.72433585e-01 -5.76201677e-01 -4.82490838e-01 1.02182400e+00
-2.05525208e+00 -1.32306910e+00 -6.06307507e-01 1.78363636e-01
8.90787840e-01 -6.59010187e-03 1.72242805e-01 1.16707899e-01
-4.57772613e-01 3.29937369e-01 -1.05827451e-01 4.91308095e-03
8.07309270e-01 -1.15761328e+00 5.55760860e-01 8.32427561e-01
-2.66732514e-01 1.87918246e-01 4.05900449e-01 -4.72876906e-01
-1.31666553e+00 -1.36771631e+00 7.44943857e-01 -5.92695475e-01
4.56180632e-01 -4.58067417e-01 -8.34279001e-01 9.64091003e-01
-9.58406404e-02 4.65146564e-02 -3.46696451e-02 -4.15815800e-01
-2.95320511e-01 -5.36703885e-01 -6.23821616e-01 6.19962215e-01
1.13918149e+00 -4.92747277e-01 -6.43277645e-01 -4.90233041e-02
5.82642674e-01 -7.59530604e-01 -4.33304906e-01 5.18751800e-01
6.80820882e-01 -1.24612331e+00 9.35498953e-01 -8.65034834e-02
2.94830322e-01 -8.35619211e-01 -5.58973327e-02 -1.34824920e+00
-5.48304319e-01 -1.06570497e-01 2.81299710e-01 1.11412573e+00
3.98533076e-01 -1.21783042e+00 4.04352754e-01 1.62953824e-01
-5.11409163e-01 -8.57330978e-01 -7.04005778e-01 -4.95437413e-01
-2.15507403e-01 -2.68125385e-01 4.83062834e-01 7.36524642e-01
-4.52550828e-01 7.08024561e-01 -3.45512569e-01 5.38859189e-01
5.95375299e-01 4.69194502e-01 1.25083804e+00 -1.27896369e+00
4.22755405e-02 -3.73311698e-01 -5.45299232e-01 -1.67667425e+00
2.67781377e-01 -8.06043506e-01 4.07724202e-01 -1.50626802e+00
-7.08094388e-02 -3.12537462e-01 9.38557759e-02 2.11670265e-01
-2.69919336e-01 -1.04069345e-01 2.33464748e-01 2.26475209e-01
-1.41416758e-01 6.87455475e-01 1.56825173e+00 -5.84490001e-02
-1.94062546e-01 3.51631641e-02 -4.03229088e-01 1.03183579e+00
5.79152048e-01 -4.83667515e-02 -5.78873158e-01 -5.38330734e-01
-1.29121393e-01 2.04378143e-01 3.56018543e-01 -9.16681707e-01
2.07671970e-01 -2.36709937e-01 4.98341292e-01 -1.21590126e+00
4.14099634e-01 -9.56076980e-01 -2.71419197e-01 3.52905273e-01
7.14337230e-02 3.26268524e-01 1.62831977e-01 8.58137488e-01
-3.89501512e-01 2.79945433e-01 8.16977143e-01 -4.89748754e-02
-1.19535804e+00 3.28790069e-01 -3.87569636e-01 -2.56732464e-01
1.25656950e+00 -7.38382876e-01 -4.11937714e-01 -6.24198020e-01
-2.60211319e-01 7.05525577e-01 7.34936595e-01 7.43667841e-01
8.24914217e-01 -1.39596522e+00 -4.29348379e-01 9.59406912e-01
3.11507016e-01 3.67999136e-01 3.89364272e-01 1.15664089e+00
-5.02444327e-01 6.31852090e-01 -2.22950667e-01 -9.79697406e-01
-1.00670922e+00 7.91447937e-01 5.76511085e-01 -1.41434833e-01
-8.23249936e-01 2.40647003e-01 1.06154275e+00 -7.26185918e-01
-2.14920759e-01 -4.31015790e-01 -4.07748818e-01 -7.51746520e-02
2.75856614e-01 2.77099013e-01 -2.06752822e-01 -1.03836381e+00
-1.57339677e-01 1.33865345e+00 -6.91296533e-02 3.45114875e-03
9.72750723e-01 -6.72301888e-01 1.38188124e-01 7.29011178e-01
1.18551779e+00 2.32511565e-01 -1.75115931e+00 -2.72248127e-02
-3.95585477e-01 -6.68617904e-01 2.49036029e-01 -2.57963240e-01
-1.01282907e+00 1.29356730e+00 5.97773731e-01 2.11936831e-02
1.02375925e+00 -3.39181602e-01 7.92015254e-01 2.28097424e-01
4.39729780e-01 -7.43932843e-01 -2.52124488e-01 6.54796004e-01
9.20373380e-01 -1.35041451e+00 -1.61618501e-01 -9.18141484e-01
-6.11994147e-01 1.25728035e+00 1.10953617e+00 -1.79035425e-01
4.76624161e-01 2.55830847e-02 3.20889145e-01 -9.06120539e-02
-6.28641248e-01 -1.28338128e-01 4.17175084e-01 5.56809485e-01
-3.57989110e-02 1.92900598e-02 1.90947056e-01 2.97977149e-01
-8.48518685e-02 -3.09642375e-01 5.32453537e-01 5.72355807e-01
-5.69290519e-01 -6.48850083e-01 -3.80474865e-01 4.73612905e-01
4.10398573e-01 1.91539735e-01 -1.21334288e-02 1.16721475e+00
2.01019585e-01 7.01746464e-01 4.21636194e-01 -7.08542049e-01
9.16737020e-01 -1.81390926e-01 1.20077118e-01 -4.98976201e-01
3.34212065e-01 8.43372643e-02 1.97969340e-02 -7.33620822e-01
-1.76128540e-02 -6.05490446e-01 -1.28115892e+00 -9.99242440e-02
-3.87129754e-01 -1.67045027e-01 4.04510617e-01 8.96603048e-01
4.49900597e-01 5.66315472e-01 1.18738317e+00 -1.16172493e+00
-2.50149548e-01 -5.62931180e-01 -4.11484063e-01 2.60235399e-01
6.62621737e-01 -1.13298678e+00 -5.11992991e-01 -4.73739319e-02]
|
[8.02640438079834, -1.7432641983032227]
|
e740e720-7b47-458c-8b75-3b613df02073
|
robustness-disparities-in-face-detection
|
2211.15937
| null |
https://arxiv.org/abs/2211.15937v1
|
https://arxiv.org/pdf/2211.15937v1.pdf
|
Robustness Disparities in Face Detection
|
Facial analysis systems have been deployed by large companies and critiqued by scholars and activists for the past decade. Many existing algorithmic audits examine the performance of these systems on later stage elements of facial analysis systems like facial recognition and age, emotion, or perceived gender prediction; however, a core component to these systems has been vastly understudied from a fairness perspective: face detection, sometimes called face localization. Since face detection is a pre-requisite step in facial analysis systems, the bias we observe in face detection will flow downstream to the other components like facial recognition and emotion prediction. Additionally, no prior work has focused on the robustness of these systems under various perturbations and corruptions, which leaves open the question of how various people are impacted by these phenomena. We present the first of its kind detailed benchmark of face detection systems, specifically examining the robustness to noise of commercial and academic models. We use both standard and recently released academic facial datasets to quantitatively analyze trends in face detection robustness. Across all the datasets and systems, we generally find that photos of individuals who are $\textit{masculine presenting}$, $\textit{older}$, of $\textit{darker skin type}$, or have $\textit{dim lighting}$ are more susceptible to errors than their counterparts in other identities.
|
['John P. Dickerson', 'Tom Goldstein', 'George Z. Wei', 'Samuel Dooley']
|
2022-11-29
| null | null | null | null |
['face-detection', 'gender-prediction']
|
['computer-vision', 'computer-vision']
|
[ 1.50319338e-01 1.58655345e-01 7.46506527e-02 -5.44999480e-01
-1.97855100e-01 -6.16443753e-01 5.17427206e-01 -1.97246760e-01
-2.55017936e-01 4.15905356e-01 -1.99372336e-01 -1.89775378e-01
1.10911705e-01 -5.41961670e-01 -2.96653718e-01 -6.77364469e-01
-1.54867098e-01 -2.21220683e-02 -3.51509333e-01 -1.05800807e-01
3.88444632e-01 6.30164087e-01 -1.85037553e+00 3.82378958e-02
2.33155072e-01 1.26356769e+00 -9.33634460e-01 4.37993377e-01
4.21124041e-01 5.44027805e-01 -6.73180819e-01 -1.16655850e+00
4.11995977e-01 -2.36618102e-01 -3.81168306e-01 2.35861307e-03
1.08554852e+00 -5.74743152e-01 -1.38234422e-01 1.36623693e+00
8.17644060e-01 -3.08376491e-01 5.56059062e-01 -1.49804592e+00
-5.86868346e-01 3.22142601e-01 -1.03018486e+00 2.03381866e-01
5.03176987e-01 4.15613592e-01 5.75399160e-01 -1.06768596e+00
4.65966910e-01 1.76874483e+00 9.06797886e-01 7.44557440e-01
-1.21594894e+00 -1.51629293e+00 -4.52630036e-02 -9.22219604e-02
-1.67732430e+00 -1.26246071e+00 5.14470577e-01 -5.72693586e-01
3.57551217e-01 3.71085316e-01 2.89275408e-01 1.15086198e+00
-4.83376980e-02 1.21876143e-01 1.25162315e+00 -2.70599931e-01
-2.08636418e-01 3.85458767e-01 -1.28685944e-02 9.74974573e-01
4.61458027e-01 8.72557387e-02 -6.25103831e-01 -4.50016856e-01
4.99101132e-01 -3.32770050e-01 1.07086271e-01 3.03707421e-01
-3.58131051e-01 7.35990405e-01 -1.32536113e-01 -1.72199622e-01
-8.90607461e-02 1.10750206e-01 4.24788803e-01 3.31666529e-01
6.64910078e-01 2.16377154e-01 -1.92948446e-01 -1.13172077e-01
-1.00618470e+00 2.35486984e-01 5.77841520e-01 6.53958142e-01
7.60548830e-01 1.98951215e-01 1.39562292e-02 8.83942664e-01
3.90881747e-01 5.82130849e-01 4.60175797e-02 -1.34109747e+00
-4.96196263e-02 6.76058829e-01 3.19942869e-02 -1.45104456e+00
-3.09859276e-01 7.18404651e-02 -5.10072231e-01 6.02349758e-01
5.69380105e-01 -4.42801774e-01 -7.51113415e-01 1.73990393e+00
3.69271189e-01 -2.29924366e-01 -4.78210896e-01 6.52098417e-01
8.57705593e-01 -9.23144165e-03 4.90657806e-01 -3.53284389e-01
1.52146316e+00 -8.87891576e-02 -3.38259846e-01 -3.01832080e-01
1.96325094e-01 -1.04233873e+00 6.81545258e-01 4.00397480e-01
-1.17202044e+00 -3.90218914e-01 -8.02772403e-01 2.51776218e-01
-1.84081003e-01 1.01213768e-01 5.26223779e-01 1.61358559e+00
-1.10020363e+00 6.48312628e-01 -2.89587796e-01 -6.63369536e-01
9.25664902e-01 6.08468652e-01 -6.47968650e-01 6.46583140e-02
-7.71231711e-01 8.34870338e-01 -4.43545431e-01 7.86050335e-02
-7.08556473e-01 -6.49147689e-01 -6.91794693e-01 -3.20321649e-01
1.28546312e-01 -7.30844438e-02 9.85820174e-01 -1.46891487e+00
-1.12472475e+00 1.54495740e+00 -2.52092421e-01 3.80093157e-02
5.62609673e-01 6.86458722e-02 -8.03412318e-01 3.40491831e-02
1.34674171e-02 7.23193705e-01 1.40372801e+00 -1.07448173e+00
-3.34046811e-01 -8.67120683e-01 -2.02268332e-01 -9.96212512e-02
-4.60160226e-01 9.49404299e-01 -1.42713815e-01 -4.59517896e-01
-1.28888249e-01 -9.78630960e-01 3.88211817e-01 3.91175598e-01
-3.00772280e-01 -2.12481827e-01 9.92205262e-01 -7.55530357e-01
1.22621226e+00 -2.27450061e+00 -6.06592894e-01 4.61133301e-01
2.88911939e-01 3.46954465e-01 1.51622579e-01 2.66777650e-02
-4.94414151e-01 7.01680481e-01 4.11607027e-01 -4.45696145e-01
1.76820278e-01 -2.66018450e-01 5.31137735e-03 9.93896723e-01
1.39729887e-01 3.84241998e-01 -4.45422053e-01 -6.63278937e-01
-1.80096909e-01 3.69411439e-01 -5.70669770e-01 -1.66068032e-01
4.46773142e-01 5.27263097e-02 -1.74844071e-01 1.37416351e+00
9.46881056e-01 2.24805593e-01 1.16554476e-01 -1.15114026e-01
-2.17538163e-01 -2.72851080e-01 -1.07314074e+00 6.33902729e-01
3.50921005e-01 7.70356655e-01 5.58193743e-01 -5.14331281e-01
9.50970650e-01 3.34964246e-01 4.36852396e-01 -4.12954539e-01
4.80520159e-01 1.08740516e-01 3.92797083e-01 -3.05058002e-01
5.02795398e-01 -2.52918750e-01 2.83771515e-01 4.82972413e-01
-1.62847608e-01 1.84257537e-01 6.03988208e-02 -7.86766224e-03
9.61890459e-01 -4.01761271e-02 -4.43139412e-02 -2.93161601e-01
2.89430708e-01 -4.34848666e-01 7.12460935e-01 3.86517942e-01
-1.07284439e+00 6.42842948e-01 8.89998794e-01 -1.99988350e-01
-9.56575811e-01 -8.67489934e-01 -3.42954069e-01 1.45425212e+00
-2.92945385e-01 -4.05056626e-01 -1.08760834e+00 -4.89914834e-01
3.83304775e-01 1.28405392e-01 -9.28210258e-01 -2.31551990e-01
-1.18867934e-01 -9.09859657e-01 1.15203369e+00 3.00655663e-01
3.19987983e-01 -8.74187231e-01 -5.32422721e-01 -4.60101157e-01
2.39188090e-01 -8.89472604e-01 -2.94165730e-01 -4.78755295e-01
-4.34859276e-01 -1.25523782e+00 -5.23389876e-01 -3.67565960e-01
9.20645595e-01 -1.17341332e-01 1.06647444e+00 6.07955217e-01
-5.98999798e-01 6.25824094e-01 -1.56937972e-01 -9.35034931e-01
-2.37155676e-01 -3.60702217e-01 5.46541333e-01 4.05626863e-01
7.62335896e-01 -3.32598388e-01 -8.41813982e-01 6.38710439e-01
-5.10643005e-01 -6.29067600e-01 3.30568939e-01 3.44613582e-01
-1.77011684e-01 -6.69396073e-02 5.22221267e-01 -8.51509809e-01
4.67867136e-01 -3.10283452e-01 -3.01211596e-01 1.12862065e-01
-7.07013071e-01 -6.56226754e-01 7.66096637e-02 -3.48985761e-01
-9.98598635e-01 5.64132109e-02 3.97105962e-02 -4.78607744e-01
-2.68703401e-01 -6.57735020e-02 -1.71316907e-01 -4.84581411e-01
8.50303054e-01 -1.87614188e-01 4.72403288e-01 -1.95684865e-01
-1.79907233e-01 7.63371408e-01 4.01393622e-01 -5.37765682e-01
9.64055002e-01 5.85755527e-01 -1.24277763e-01 -9.87449765e-01
-2.04026654e-01 -6.17738813e-02 -4.67353135e-01 -6.99844778e-01
4.77021396e-01 -8.94527197e-01 -1.33029270e+00 7.83040285e-01
-7.47486830e-01 3.67199332e-02 2.58586764e-01 -3.11201047e-02
-2.73302346e-02 4.21761185e-01 -4.49173152e-01 -1.25364423e+00
-3.66727620e-01 -1.08119619e+00 9.25262809e-01 3.48057270e-01
-7.17691362e-01 -4.51760352e-01 -4.60837632e-01 5.17797351e-01
3.47846657e-01 3.57206672e-01 7.51403034e-01 -3.03372741e-01
5.16318381e-02 -5.07192969e-01 -4.00471002e-01 3.69462729e-01
2.04328746e-01 9.24995244e-01 -1.46141005e+00 -4.10804182e-01
-2.95324177e-01 -4.69403505e-01 4.76486713e-01 2.66351998e-01
1.14643335e+00 -2.75750101e-01 -2.27409244e-01 4.71575797e-01
8.89932275e-01 4.74245399e-02 8.00431430e-01 -8.83271452e-03
2.02377677e-01 1.10558939e+00 3.96280140e-01 6.84714615e-01
3.89823504e-02 3.34285885e-01 4.94781047e-01 -9.64375772e-03
1.64283112e-01 -4.16035280e-02 6.09102130e-01 -2.00001448e-01
-4.47794080e-01 2.38199532e-01 -8.90569448e-01 1.70272201e-01
-1.22803164e+00 -1.26024723e+00 1.39139637e-01 2.10345268e+00
6.84858739e-01 -2.50270944e-02 4.57835495e-01 1.40572965e-01
1.01531875e+00 6.61706403e-02 -5.41381896e-01 -5.65853715e-01
-8.48494694e-02 2.99333334e-01 4.00782824e-01 6.34045452e-02
-1.13007712e+00 8.63794029e-01 7.11151648e+00 5.06273627e-01
-1.25177670e+00 -1.63736716e-01 1.37765718e+00 -3.51893783e-01
1.88200057e-01 -1.39731377e-01 -8.04202020e-01 4.74921495e-01
7.82648563e-01 -2.16492936e-01 3.46939921e-01 9.98442173e-01
3.12088698e-01 -3.60582739e-01 -1.11497509e+00 1.13995838e+00
2.44786635e-01 -8.03148627e-01 -1.77699894e-01 2.83585221e-01
4.29375350e-01 -4.88127321e-01 7.22804070e-01 2.05994949e-01
1.51623785e-01 -1.42287624e+00 8.89329493e-01 3.70512933e-01
1.27262104e+00 -7.54258513e-01 3.37808639e-01 -1.94937453e-01
-8.62220824e-01 -2.73376703e-01 -1.97481707e-01 -2.97187656e-01
-3.55465561e-01 3.21497083e-01 -5.32793164e-01 -1.21865392e-01
1.09909511e+00 3.03135365e-01 -8.80470097e-01 3.91967088e-01
1.54391497e-01 5.48248291e-01 -3.26303393e-01 2.45321974e-01
-3.61957997e-01 -1.42502517e-03 2.89965689e-01 9.80566204e-01
2.01182440e-01 3.01532239e-01 -3.14526945e-01 6.95103526e-01
-3.30129445e-01 6.78821281e-02 -5.17167509e-01 -2.52834916e-01
5.61532617e-01 1.58589542e+00 -8.08954835e-01 -3.02631818e-02
-4.46207136e-01 5.39127648e-01 9.44231171e-03 2.11176381e-01
-6.67731285e-01 -3.71997207e-02 1.21399379e+00 5.52234530e-01
-2.49812171e-01 1.73559263e-01 -4.76789415e-01 -6.38739407e-01
-7.11698979e-02 -1.25139308e+00 4.02411073e-01 -6.65365517e-01
-1.22279060e+00 2.68516392e-01 -3.39565277e-01 -7.28896797e-01
-4.77435738e-02 -7.88143158e-01 -5.87996960e-01 9.41152692e-01
-1.00899792e+00 -1.01089442e+00 -3.46051693e-01 6.94365561e-01
-6.54415190e-02 -4.52321202e-01 7.30425000e-01 4.15474355e-01
-1.09013975e+00 1.19116104e+00 -3.88447523e-01 5.57478905e-01
1.16345251e+00 -7.05419660e-01 4.10220064e-02 8.39203894e-01
-3.44710469e-01 1.02780831e+00 7.35754669e-01 -6.53266072e-01
-1.28205264e+00 -8.06151032e-01 5.49308181e-01 -7.49088287e-01
3.92778099e-01 -3.28925282e-01 -4.10593361e-01 7.78315783e-01
4.25616186e-03 1.97565444e-02 8.80802989e-01 3.06984425e-01
-5.51074564e-01 -3.34497720e-01 -1.56649792e+00 6.73780084e-01
9.91679549e-01 -6.71106577e-01 2.33087186e-02 -1.08670155e-02
-2.47296765e-01 -5.52340075e-02 -8.93069446e-01 6.10898435e-01
1.17073607e+00 -1.43321145e+00 8.34150493e-01 -5.10948896e-01
4.39315408e-01 1.86929896e-01 -2.59778704e-02 -6.75222218e-01
-3.64801884e-01 -8.00213933e-01 2.47669592e-01 1.68464637e+00
3.46835971e-01 -5.17469227e-01 1.06102133e+00 1.32975125e+00
3.60791266e-01 -4.92475569e-01 -8.46868396e-01 -3.93336356e-01
9.76319388e-02 -2.92837024e-01 3.73107523e-01 1.10605490e+00
-1.40105039e-01 -2.35530868e-01 -3.67539942e-01 1.20314993e-01
7.10606098e-01 -3.88675064e-01 9.60682869e-01 -1.29975462e+00
2.96709746e-01 -8.14381003e-01 -6.44829810e-01 -4.37185960e-03
2.51046777e-01 -4.03671563e-01 -3.01791012e-01 -5.91218591e-01
3.66972804e-01 -3.60914201e-01 -3.39766592e-02 6.42342031e-01
-9.38545987e-02 8.41193140e-01 9.36946571e-02 1.16076574e-01
-1.48465723e-01 -2.83295903e-02 6.19623721e-01 6.71900958e-02
3.85956705e-01 -1.20871767e-01 -1.08213270e+00 1.12938774e+00
4.75001067e-01 -2.56849349e-01 -1.80012360e-02 -1.23017028e-01
3.70196164e-01 -4.61837858e-01 5.36190391e-01 -8.16939831e-01
-3.23801162e-03 -2.36653820e-01 8.09976518e-01 -1.04170933e-01
4.92395878e-01 -5.85765600e-01 1.89452499e-01 3.36632550e-01
1.08178481e-01 2.60398000e-01 2.08883375e-01 -1.72633138e-02
1.26139820e-01 -4.27073017e-02 1.21663165e+00 -1.88741773e-01
-5.16718686e-01 4.27261651e-01 -3.54068577e-01 -8.12032372e-02
1.08470690e+00 -6.86566055e-01 -4.61531907e-01 -5.42032659e-01
-6.03184640e-01 -6.39154315e-02 8.50619137e-01 3.33756447e-01
2.91598022e-01 -1.06998098e+00 -8.57959092e-01 3.69616777e-01
1.34420618e-01 -7.83201456e-01 1.46499112e-01 1.05208206e+00
-4.08366859e-01 -1.84122249e-01 -3.99359971e-01 -3.69631380e-01
-1.78062165e+00 2.54056573e-01 5.99134982e-01 6.93737328e-01
1.69430435e-01 1.16613317e+00 7.85551220e-02 -9.10835043e-02
2.39153802e-01 4.13333654e-01 -2.14642175e-02 5.43911099e-01
6.04149878e-01 8.25347424e-01 2.50003859e-02 -1.12333071e+00
-5.83265722e-01 3.55812728e-01 -8.97300467e-02 4.31592129e-02
9.37897623e-01 -8.48126262e-02 -4.19524908e-01 -1.03114970e-01
9.64763463e-01 1.45979241e-01 -1.00565386e+00 2.62151152e-01
-1.29538104e-01 -8.59395146e-01 -3.39331031e-01 -6.98076844e-01
-1.32107329e+00 6.42752707e-01 8.45183551e-01 1.44908607e-01
1.03620481e+00 -2.08215326e-01 1.79009721e-01 -6.12884760e-02
1.92043155e-01 -1.41994131e+00 7.12792650e-02 2.40677521e-01
7.52020895e-01 -1.43395007e+00 3.81251842e-01 -4.27676529e-01
-3.66978347e-01 8.66190434e-01 8.44748855e-01 1.97363183e-01
8.04328978e-01 1.63354576e-01 2.42468134e-01 -2.93915033e-01
-4.68209565e-01 7.11797178e-02 4.25298847e-02 5.09077787e-01
6.55458868e-01 -1.86022054e-02 -1.98517799e-01 5.64763248e-01
-4.42837805e-01 -2.02832296e-01 3.01526725e-01 7.14833796e-01
-2.37611011e-01 -8.45837653e-01 -7.57042766e-01 7.07588196e-01
-1.02843153e+00 9.67417434e-02 -9.82391238e-01 6.74204469e-01
4.81312633e-01 1.04694235e+00 1.95714980e-01 -4.57720071e-01
1.88430876e-01 4.93588179e-01 4.48884279e-01 -4.90573376e-01
-7.57155120e-01 -8.24050531e-02 2.01948658e-01 -5.80104589e-01
-4.36923862e-01 -1.20298350e+00 -7.94251919e-01 -1.07423949e+00
-1.42305151e-01 -3.47371817e-01 4.95915681e-01 6.56403840e-01
3.65067005e-01 -2.70853937e-01 6.91502810e-01 -8.50610614e-01
-3.86070400e-01 -9.39315498e-01 -8.72114778e-01 6.47898674e-01
9.20334607e-02 -7.92370737e-01 -4.09286350e-01 6.21542595e-02]
|
[13.038195610046387, 1.2085245847702026]
|
1623b724-cf1f-4fed-86ea-56b0150f31c1
|
ace-vc-adaptive-and-controllable-voice
|
2302.08137
| null |
https://arxiv.org/abs/2302.08137v1
|
https://arxiv.org/pdf/2302.08137v1.pdf
|
ACE-VC: Adaptive and Controllable Voice Conversion using Explicitly Disentangled Self-supervised Speech Representations
|
In this work, we propose a zero-shot voice conversion method using speech representations trained with self-supervised learning. First, we develop a multi-task model to decompose a speech utterance into features such as linguistic content, speaker characteristics, and speaking style. To disentangle content and speaker representations, we propose a training strategy based on Siamese networks that encourages similarity between the content representations of the original and pitch-shifted audio. Next, we develop a synthesis model with pitch and duration predictors that can effectively reconstruct the speech signal from its decomposed representation. Our framework allows controllable and speaker-adaptive synthesis to perform zero-shot any-to-any voice conversion achieving state-of-the-art results on metrics evaluating speaker similarity, intelligibility, and naturalness. Using just 10 seconds of data for a target speaker, our framework can perform voice swapping and achieves a speaker verification EER of 5.5% for seen speakers and 8.4% for unseen speakers.
|
['Boris Ginsburg', 'Jason Li', 'Jocelyn Huang', 'Paarth Neekhara', 'Shehzeen Hussain']
|
2023-02-16
| null | null | null | null |
['voice-conversion', 'voice-conversion', 'speaker-verification']
|
['audio', 'speech', 'speech']
|
[ 2.19161838e-01 2.85996556e-01 -6.23819195e-02 -4.69410449e-01
-1.29486048e+00 -5.15698612e-01 3.99865210e-01 -1.79527014e-01
6.36798143e-02 3.74721229e-01 7.14828789e-01 -1.79198518e-01
1.97601169e-01 -3.82724941e-01 -4.77381319e-01 -5.28170824e-01
1.24101363e-01 1.57217532e-01 -2.79489517e-01 -3.59685093e-01
-2.37786531e-01 3.88823062e-01 -1.86956978e+00 2.64910907e-01
7.37490714e-01 9.54499245e-01 1.22836798e-01 1.10390210e+00
-1.09589301e-01 3.79293740e-01 -8.78992617e-01 -2.25546747e-01
8.52012634e-02 -6.90393746e-01 -5.55535734e-01 1.78925619e-01
5.23434758e-01 -1.64989784e-01 -4.07768935e-01 8.17390382e-01
9.43184912e-01 3.44659448e-01 5.41561782e-01 -1.07151890e+00
-7.53769159e-01 9.07427549e-01 3.39611173e-02 1.46694779e-01
6.30180001e-01 1.19704209e-01 1.12763703e+00 -9.60682750e-01
2.45804071e-01 1.37214231e+00 5.32402694e-01 8.97190928e-01
-1.51960051e+00 -8.07069778e-01 3.93443624e-04 -1.40812844e-02
-1.17933977e+00 -1.32566845e+00 8.93038034e-01 -2.40330279e-01
8.69357407e-01 5.21691382e-01 3.87945682e-01 1.30344558e+00
-7.16572776e-02 5.54253459e-01 7.25220203e-01 -5.70707917e-01
1.39551803e-01 2.21373335e-01 -1.97479293e-01 4.67569381e-01
-6.44681454e-01 4.03385341e-01 -8.53316903e-01 -5.81745878e-02
3.82967323e-01 -3.80434126e-01 -4.03670162e-01 4.42845114e-02
-1.19010222e+00 7.55571008e-01 -2.09965706e-02 3.01385522e-01
-2.20095530e-01 -1.14510603e-01 4.10665274e-01 5.81810594e-01
4.69067037e-01 5.39897561e-01 -3.01378071e-01 -2.52009124e-01
-1.06369090e+00 -9.07482766e-03 8.77969682e-01 8.56217384e-01
2.44578063e-01 8.34528446e-01 -4.71215636e-01 1.27613676e+00
1.38309255e-01 7.57807851e-01 9.27713275e-01 -1.05622602e+00
2.14756295e-01 -2.75839955e-01 -1.36670455e-01 -5.07596850e-01
1.03914654e-02 -3.91728103e-01 -5.85455239e-01 9.84321013e-02
-1.00314863e-01 -2.59641558e-01 -8.37802768e-01 1.99344862e+00
8.67711008e-02 3.30137312e-01 4.03693140e-01 5.84260821e-01
1.04256618e+00 8.94763052e-01 -2.55373091e-01 -5.35252333e-01
1.22446883e+00 -1.33891737e+00 -1.11795688e+00 -1.37443155e-01
-1.85123026e-01 -8.23209107e-01 1.48584390e+00 2.55730510e-01
-1.34168446e+00 -9.36336160e-01 -1.12936866e+00 2.66004503e-01
6.29954855e-04 1.27662331e-01 -2.71989089e-02 9.16755319e-01
-1.06649518e+00 6.21974587e-01 -5.24256527e-01 -3.32979970e-02
-1.50507957e-01 1.32427931e-01 -1.38788015e-01 4.45782870e-01
-1.24040592e+00 5.75969815e-01 -9.73285809e-02 -5.20089865e-01
-1.19293070e+00 -9.01179790e-01 -9.09882426e-01 4.24771845e-01
6.64575845e-02 -4.94915366e-01 1.65617049e+00 -9.21362340e-01
-2.44898748e+00 4.71733868e-01 -4.40668821e-01 -4.37078059e-01
1.61878735e-01 -1.28273562e-01 -1.18210030e+00 2.28607059e-01
7.02903494e-02 5.84418297e-01 1.42731535e+00 -1.08086777e+00
-4.68589902e-01 8.52457285e-02 -5.79021335e-01 2.28440717e-01
-6.95573747e-01 1.25760913e-01 -1.31906435e-01 -1.09076440e+00
3.92344370e-02 -8.11995208e-01 3.52022648e-01 -1.14442497e-01
-5.32716453e-01 -4.87379394e-02 7.99480081e-01 -9.11427557e-01
9.93526101e-01 -2.53416705e+00 2.71022767e-01 -2.20579386e-01
-7.91717023e-02 9.56055596e-02 -5.40321827e-01 1.11921966e-01
-2.10372001e-01 -5.86770028e-02 -6.37711361e-02 -8.37528884e-01
-6.97649969e-03 -1.96791828e-01 -7.18746901e-01 1.48988187e-01
1.86363012e-01 3.77105415e-01 -7.03757107e-01 -1.61064163e-01
1.65452406e-01 8.65402818e-01 -7.84104347e-01 5.79009056e-01
5.47856577e-02 4.55164671e-01 3.56758296e-01 5.36883354e-01
3.15168947e-01 3.92167687e-01 1.20690120e-02 -1.44471884e-01
-5.16338311e-02 7.28563607e-01 -9.00003910e-01 1.66023326e+00
-9.92399871e-01 8.27733815e-01 3.97638500e-01 -3.79383445e-01
1.15292799e+00 9.23061788e-01 2.60811687e-01 -3.93737614e-01
1.00404277e-01 9.09428373e-02 -6.32808134e-02 -1.99650079e-01
4.03347760e-01 -3.59777629e-01 9.29051358e-03 4.13729191e-01
5.25596201e-01 -6.19235873e-01 -2.98460692e-01 -3.57513666e-01
7.64609456e-01 -3.65881056e-01 9.65035930e-02 -2.64894795e-02
4.87442404e-01 -8.94806504e-01 3.97702843e-01 3.94939303e-01
-3.73142809e-01 7.92491317e-01 1.44315474e-02 1.88685164e-01
-8.56118858e-01 -1.62782216e+00 2.10587829e-02 1.55622041e+00
-2.88551182e-01 -2.01405376e-01 -8.09511304e-01 1.34441657e-02
-1.27086878e-01 1.26861894e+00 -2.55880952e-01 -6.30570471e-01
-4.49969023e-01 3.30214538e-02 8.49915802e-01 3.36550772e-01
6.25510141e-02 -9.86167192e-01 -8.53902660e-03 3.21565062e-01
-3.63270909e-01 -1.02391171e+00 -1.11199212e+00 1.68874696e-01
-5.19099355e-01 -1.73590869e-01 -7.29666173e-01 -9.73694205e-01
1.56081185e-01 2.04157203e-01 7.86104202e-01 -6.24331951e-01
2.17451481e-03 2.36320585e-01 -1.52401775e-01 -1.86120078e-01
-1.10623157e+00 1.14173405e-02 7.10817814e-01 2.92997092e-01
-2.68146008e-01 -8.90633464e-01 -2.56582677e-01 3.07396621e-01
-5.01999199e-01 -1.91661328e-01 2.73214638e-01 8.14902902e-01
3.76589179e-01 -1.37351409e-01 1.13669670e+00 -1.39806345e-01
9.22003031e-01 -2.09538952e-01 -1.90111682e-01 2.05847487e-01
-5.91310382e-01 1.00547910e-01 9.18070138e-01 -8.44052255e-01
-1.12084985e+00 4.37214188e-02 -2.01790243e-01 -8.28645289e-01
-2.35994145e-01 -1.29986880e-02 -4.63114589e-01 2.66331285e-01
7.32712388e-01 4.15102333e-01 3.24985385e-01 -4.37958270e-01
9.04767275e-01 1.22335434e+00 1.06251144e+00 -3.62296581e-01
6.70675516e-01 3.27978507e-02 -7.43648052e-01 -1.17733383e+00
-6.00476146e-01 -3.09824467e-01 -3.47290695e-01 -3.46402340e-02
7.18964875e-01 -1.03813744e+00 -6.45366490e-01 2.18421817e-01
-1.08897805e+00 -1.64572269e-01 -5.27692556e-01 5.47706902e-01
-8.22337985e-01 1.83925509e-01 -6.02495492e-01 -9.70067799e-01
-6.19398952e-01 -1.24885213e+00 1.12709165e+00 -2.82394662e-02
-5.54349780e-01 -7.19776094e-01 1.88733354e-01 4.95788813e-01
7.55754352e-01 -3.37020636e-01 7.32328176e-01 -7.74575651e-01
-1.28393965e-02 2.49881689e-02 4.40739602e-01 6.93239629e-01
7.35436738e-01 4.32057567e-02 -1.49120498e+00 -4.51324373e-01
2.29334682e-01 -2.49837801e-01 6.83377683e-01 3.26479644e-01
9.56651330e-01 -6.77143216e-01 1.13612510e-01 6.89925909e-01
5.90143621e-01 2.58434951e-01 1.81558594e-01 -3.42518121e-01
4.29382890e-01 6.67480171e-01 2.50711222e-03 4.50578779e-01
1.09377839e-01 7.73239851e-01 -7.56606832e-02 8.50845277e-02
-7.11383998e-01 -5.08440554e-01 7.88491011e-01 1.30884278e+00
4.91187960e-01 -2.80741245e-01 -5.07209480e-01 6.10154152e-01
-9.75803435e-01 -1.11368990e+00 7.03489184e-01 2.18650484e+00
1.13226116e+00 1.62950069e-01 3.35332543e-01 2.99350321e-01
9.65412259e-01 3.56324911e-01 -6.88866198e-01 -5.82909107e-01
1.07234177e-04 3.30323100e-01 -9.83276069e-02 8.68344367e-01
-9.80592489e-01 9.07495499e-01 7.00065374e+00 6.76144600e-01
-1.46427751e+00 8.93948674e-02 4.12389517e-01 -6.89370453e-01
-5.02730727e-01 -5.81318498e-01 -6.37524545e-01 3.26555341e-01
1.62699831e+00 -6.11938119e-01 9.55565453e-01 7.95985341e-01
4.44910496e-01 8.01439703e-01 -1.37266898e+00 1.00796938e+00
5.24569869e-01 -1.08902514e+00 5.21567501e-02 -2.09354714e-01
5.19088447e-01 -1.54066220e-01 5.37763178e-01 4.50397670e-01
7.37167299e-02 -1.04358125e+00 1.04190779e+00 3.47726464e-01
1.27481484e+00 -6.79406703e-01 2.29273271e-02 1.37851983e-01
-1.12222624e+00 -1.37977287e-01 1.38565451e-01 3.21964741e-01
2.15225667e-01 2.24278629e-01 -1.04387939e+00 2.03384712e-01
4.34669793e-01 2.84566045e-01 -7.72823095e-02 5.94509423e-01
-1.41975880e-01 7.75981784e-01 -7.45271444e-02 1.02102630e-01
-3.80591869e-01 2.80124843e-01 9.28282619e-01 1.13068819e+00
5.56687772e-01 -1.85411766e-01 7.40206912e-02 8.62237155e-01
-4.08578902e-01 7.21308291e-02 -3.85471702e-01 -1.45261288e-01
1.01934826e+00 9.61934566e-01 -1.06434092e-01 -2.52279580e-01
-1.26942188e-01 1.20607460e+00 6.38568699e-02 3.92019749e-01
-7.02241063e-01 -7.13778853e-01 1.10032916e+00 -1.14715897e-01
3.69789809e-01 -2.08415110e-02 -1.82714552e-01 -1.09664023e+00
-1.15990780e-01 -1.12403202e+00 -2.06414655e-01 -6.59408092e-01
-1.21467686e+00 1.06291115e+00 -2.78725833e-01 -1.16582847e+00
-8.51771057e-01 -1.74014196e-01 -8.46355498e-01 1.04206729e+00
-1.24459171e+00 -8.17704499e-01 7.50582367e-02 4.12704468e-01
1.10147560e+00 -8.05952311e-01 1.24078274e+00 1.20380208e-01
-4.74650562e-01 1.11412144e+00 1.34278551e-01 -7.21410587e-02
8.49633396e-01 -1.05136991e+00 7.74726689e-01 6.09539390e-01
3.54839057e-01 3.90469342e-01 8.67963374e-01 -1.72442809e-01
-1.12731671e+00 -1.09342921e+00 9.55278337e-01 -1.28501385e-01
5.45547664e-01 -4.20538247e-01 -9.26420331e-01 3.13420951e-01
3.71694148e-01 -2.45351531e-02 1.09240890e+00 1.80526137e-01
-6.40862465e-01 -3.17951709e-01 -1.10477245e+00 6.92978501e-01
7.68550217e-01 -1.12818611e+00 -9.36966956e-01 1.03092380e-01
1.42254174e+00 -2.23998383e-01 -7.41617441e-01 -1.51696457e-02
5.33183038e-01 -8.27410161e-01 1.05979311e+00 -5.13260841e-01
1.21900223e-01 6.54705986e-02 -4.56280172e-01 -1.71078062e+00
-2.36102968e-01 -1.08402979e+00 -1.55196235e-01 1.45203280e+00
6.67397857e-01 -4.83334780e-01 3.40510756e-01 1.93764269e-01
-3.82966310e-01 -3.10715467e-01 -1.09097028e+00 -1.07664526e+00
1.15063563e-01 -3.38793367e-01 7.03949153e-01 8.20967615e-01
1.72495022e-01 6.75386667e-01 -4.39484954e-01 2.72843689e-01
4.76144105e-01 1.36481538e-01 4.24824834e-01 -1.08178377e+00
-6.05929077e-01 -5.13232946e-01 8.29415172e-02 -8.62118661e-01
6.77642345e-01 -9.48330522e-01 3.47770542e-01 -9.92031693e-01
-3.40302050e-01 1.90270320e-01 -2.59055436e-01 4.11332786e-01
-3.34201683e-03 -2.43624404e-01 3.38795662e-01 1.10522062e-02
1.24589121e-03 1.08521807e+00 9.06511009e-01 -3.96293819e-01
-6.53856993e-01 3.00525397e-01 -7.42891252e-01 4.82213020e-01
8.49112749e-01 -2.09000051e-01 -5.47660053e-01 -2.52259851e-01
-7.80464172e-01 6.66493893e-01 5.83859310e-02 -1.08367825e+00
6.33368045e-02 -3.88005227e-02 -5.33669256e-02 -2.14342907e-01
8.73392999e-01 -3.99492413e-01 -1.80113107e-01 3.16529483e-01
-9.37937260e-01 -2.80965656e-01 3.33274424e-01 4.58953619e-01
-4.90769446e-01 6.01934344e-02 9.90373790e-01 2.66389370e-01
-2.26777829e-02 8.15851167e-02 -4.87106502e-01 -4.65633161e-03
5.97565651e-01 1.41473919e-01 -2.81745587e-02 -8.30134690e-01
-9.80475247e-01 -1.76230341e-01 -7.77860880e-02 8.66767049e-01
7.77551055e-01 -1.56240857e+00 -9.98012483e-01 6.92237496e-01
2.88882311e-02 -6.45756125e-01 1.95836410e-01 5.52090257e-02
1.82947338e-01 3.46721828e-01 -1.02111839e-01 -4.16673362e-01
-1.40549827e+00 4.70756233e-01 4.56227690e-01 4.47901338e-01
-3.67551506e-01 9.33749795e-01 -7.43403658e-02 -5.80311477e-01
5.20123780e-01 -5.20407796e-01 8.57835710e-02 9.73637775e-02
5.89538753e-01 3.06838751e-01 1.63076952e-01 -7.54641891e-01
-2.73204148e-01 1.26979813e-01 6.99613616e-02 -7.74124742e-01
1.11072195e+00 -2.40869015e-01 3.76983762e-01 8.06005239e-01
1.46648240e+00 4.13194984e-01 -1.25374258e+00 -1.87334269e-01
-4.96782631e-01 -1.93570152e-01 3.05110842e-01 -8.80693972e-01
-8.07073593e-01 9.78959739e-01 7.47170210e-01 3.14412594e-01
1.03809166e+00 8.63066390e-02 1.00223386e+00 3.37979019e-01
-1.40763506e-01 -1.00278568e+00 3.86642307e-01 5.75200737e-01
1.26027071e+00 -1.03823066e+00 -6.78436995e-01 -2.30005592e-01
-7.70338058e-01 9.22792912e-01 2.71421582e-01 2.21878454e-01
6.73440039e-01 3.13946873e-01 3.05031091e-01 4.53857273e-01
-9.55832720e-01 7.85865728e-03 5.11961639e-01 6.95712745e-01
6.17580116e-01 3.24645072e-01 5.49168587e-01 7.61423051e-01
-9.98555183e-01 -5.19844115e-01 3.24055135e-01 1.62180349e-01
-6.07029855e-01 -8.85788560e-01 -5.04401147e-01 8.33238512e-02
-1.46690711e-01 -2.66039491e-01 -3.80902439e-01 -9.05822031e-04
-4.20093119e-01 1.38212419e+00 2.43988186e-01 -6.87481642e-01
5.91085613e-01 4.74546939e-01 2.03364477e-01 -7.71547854e-01
-5.25588095e-01 3.89799058e-01 8.98429826e-02 -2.37095684e-01
8.71374831e-02 -6.95272565e-01 -1.20940769e+00 -7.90986270e-02
-2.51945317e-01 1.09388888e-01 8.43351960e-01 6.61725104e-01
5.06620109e-01 9.90067065e-01 1.28448689e+00 -8.55009496e-01
-1.07927454e+00 -1.09246397e+00 -5.81490397e-01 1.55350968e-01
9.15789962e-01 -2.97554195e-01 -7.65964329e-01 1.86343595e-01]
|
[14.950693130493164, 6.564146995544434]
|
08e8fc03-9d2b-4215-bf9c-ad6574eb55eb
|
a-topic-coverage-approach-to-evaluation-of
|
2012.06274
| null |
https://arxiv.org/abs/2012.06274v3
|
https://arxiv.org/pdf/2012.06274v3.pdf
|
A Topic Coverage Approach to Evaluation of Topic Models
|
Topic models are widely used unsupervised models capable of learning topics - weighted lists of words and documents - from large collections of text documents. When topic models are used for discovery of topics in text collections, a question that arises naturally is how well the model-induced topics correspond to topics of interest to the analyst. In this paper we revisit and extend a so far neglected approach to topic model evaluation based on measuring topic coverage - computationally matching model topics with a set of reference topics that models are expected to uncover. The approach is well suited for analyzing models' performance in topic discovery and for large-scale analysis of both topic models and measures of model quality. We propose new measures of coverage and evaluate, in a series of experiments, different types of topic models on two distinct text domains for which interest for topic discovery exists. The experiments include evaluation of model quality, analysis of coverage of distinct topic categories, and the analysis of the relationship between coverage and other methods of topic model evaluation. The paper contributes a new supervised measure of coverage, and the first unsupervised measure of coverage. The supervised measure achieves topic matching accuracy close to human agreement. The unsupervised measure correlates highly with the supervised one (Spearman's $\rho \geq 0.95$). Other contributions include insights into both topic models and different methods of model evaluation, and the datasets and code for facilitating future research on topic coverage.
|
['Jan Šnajder', 'Jelena Repar', 'Strahil Ristov', 'Damir Korenčić']
|
2020-12-11
|
a-topic-coverage-approach-to-evaluation-of-1
|
https://ieeexplore.ieee.org/abstract/document/9526605
|
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9526605
| null |
['topic-coverage']
|
['natural-language-processing']
|
[ 1.08578138e-01 5.42227983e-01 -5.85756242e-01 -5.31379044e-01
-1.28403533e+00 -4.46958482e-01 1.01939917e+00 6.97468340e-01
-1.12817474e-01 6.63099468e-01 4.66381341e-01 -1.62811890e-01
-5.70855916e-01 -1.06275773e+00 -3.68735105e-01 -5.19643366e-01
-2.75004029e-01 1.10743558e+00 4.53789532e-01 1.36420548e-01
7.01592743e-01 -9.07782614e-02 -1.77700460e+00 2.46540129e-01
1.04039192e+00 6.74586356e-01 3.78442973e-01 4.48328108e-01
-5.62711596e-01 3.27713609e-01 -8.33239317e-01 -2.93668844e-02
-2.19401702e-01 -2.63062924e-01 -9.08862531e-01 3.33231121e-01
3.12598646e-01 2.24884525e-01 2.84219205e-01 7.96090841e-01
-8.79711378e-03 1.21658541e-01 1.20726395e+00 -1.37082767e+00
1.32778794e-01 8.75510097e-01 -3.81347179e-01 3.56897146e-01
6.49506509e-01 -4.33296084e-01 1.17490470e+00 -7.05620348e-01
9.72727418e-01 1.32261717e+00 6.14800572e-01 -2.47118063e-02
-1.30833411e+00 -5.83305359e-01 1.48516417e-01 -2.58120805e-01
-1.41702974e+00 -1.35341421e-01 5.93382835e-01 -9.33548391e-01
7.31342554e-01 3.91750276e-01 5.75259924e-01 6.66499078e-01
3.60054702e-01 5.99518120e-01 1.10658836e+00 -5.77798367e-01
4.44056511e-01 9.81564224e-01 8.96157980e-01 1.04088485e-01
6.50023460e-01 -1.47342801e-01 -5.78719735e-01 -8.47165048e-01
2.86583811e-01 -1.18234023e-01 -1.67668134e-01 -4.85020339e-01
-1.17009163e+00 1.39670062e+00 -1.58672765e-01 6.76705599e-01
-4.54199433e-01 -2.07879469e-01 3.18158627e-01 1.71640396e-01
1.33384454e+00 8.76351297e-01 -4.44688439e-01 -2.25864518e-02
-1.50813055e+00 8.30758989e-01 1.10469735e+00 1.15422380e+00
8.82534683e-01 -3.71883214e-01 -2.20631436e-01 8.11615288e-01
5.59547246e-01 4.10719305e-01 7.61295855e-01 -4.29261416e-01
2.61465997e-01 9.09351408e-01 2.72201151e-01 -9.74651217e-01
-4.08086807e-01 -3.34178060e-01 -1.33283094e-01 -3.04187447e-01
2.08633944e-01 1.41058400e-01 -6.99136734e-01 1.41180348e+00
2.88859457e-01 -3.73799175e-01 3.29576433e-03 2.18970656e-01
7.81128347e-01 9.26100373e-01 2.72293866e-01 -6.63094997e-01
1.56522393e+00 -6.22052372e-01 -7.20731795e-01 1.11031838e-01
8.05306554e-01 -9.09960747e-01 9.79496539e-01 3.91642839e-01
-1.18872452e+00 -4.03730005e-01 -6.50822520e-01 2.82497585e-01
-5.90686083e-01 -1.83033586e-01 6.01421237e-01 7.59699166e-01
-1.02771437e+00 3.18462670e-01 -6.02035522e-01 -8.79015326e-01
1.54660702e-01 2.49621958e-01 9.62434411e-02 1.44801930e-01
-1.17696083e+00 7.54240990e-01 7.33676195e-01 -1.02201807e+00
-9.85369444e-01 -9.61643755e-01 -6.68389082e-01 3.01709205e-01
2.35989600e-01 -2.48109430e-01 1.32440794e+00 -6.03715181e-01
-6.57290339e-01 9.55498040e-01 -5.13247490e-01 -6.95084393e-01
6.78430498e-02 1.40868640e-02 -4.88740444e-01 2.17268720e-01
6.54706538e-01 6.33085549e-01 5.05516291e-01 -1.49288201e+00
-1.03645849e+00 -1.92247003e-01 -1.68754026e-01 8.69165063e-02
-5.22099316e-01 3.09882581e-01 -2.62900829e-01 -5.16946971e-01
3.25904191e-01 -5.77004611e-01 -1.95801154e-01 -5.74220955e-01
-3.91909182e-01 -7.80224144e-01 8.09319079e-01 -4.27940905e-01
1.57439947e+00 -1.65330720e+00 -4.11302507e-01 5.81359208e-01
3.41320634e-01 -3.66545200e-01 4.14485991e-01 8.30094337e-01
-1.42845050e-01 4.90381300e-01 -1.52171791e-01 -2.37311289e-01
1.49023244e-02 -3.28605711e-01 -7.67918408e-01 3.25141132e-01
-2.77843922e-01 4.72951502e-01 -6.34666741e-01 -9.27132308e-01
-4.16801311e-03 1.56948745e-01 -3.26871276e-01 5.67165799e-02
-5.68674922e-01 -2.51516581e-01 -5.36282003e-01 4.50861305e-01
4.66055393e-01 -4.50545996e-01 3.04543469e-02 3.40864271e-01
-1.84523731e-01 7.33950496e-01 -1.16722453e+00 1.32413983e+00
-4.43526804e-02 9.05242026e-01 -5.15431345e-01 -7.62091041e-01
1.28467286e+00 6.01948082e-01 7.97080755e-01 -3.41986455e-02
-5.60905598e-02 1.41200647e-01 -2.27668330e-01 -1.30057573e-01
9.54826772e-01 -2.00529799e-01 -1.48053274e-01 1.06677330e+00
2.76422977e-01 -3.83512944e-01 4.46163237e-01 5.53847015e-01
7.07618237e-01 -4.62696284e-01 5.94874740e-01 -1.15876150e+00
1.49111645e-02 6.62710309e-01 -7.80483335e-02 8.70853484e-01
2.32722178e-01 3.83749127e-01 7.01672554e-01 -3.66177499e-01
-1.08579826e+00 -7.98495889e-01 -7.09999561e-01 1.23678029e+00
1.80971790e-02 -7.25719333e-01 -8.31158757e-01 -4.66221511e-01
-1.86507180e-01 9.73305047e-01 -8.44550908e-01 2.62374401e-01
-4.22698222e-02 -8.94802630e-01 1.88246220e-01 1.49682850e-01
-5.01009859e-02 -8.37716758e-01 -6.06763363e-01 1.93000391e-01
-4.55708206e-01 -4.93363857e-01 -1.23822540e-01 1.40429974e-01
-1.33689618e+00 -1.11854649e+00 -8.19563568e-01 -6.65101290e-01
5.45837224e-01 4.60174441e-01 1.47685111e+00 -5.63949794e-02
7.44528174e-02 5.09484947e-01 -4.24856722e-01 -9.77316916e-01
-5.08593321e-01 4.75440323e-01 2.10268293e-02 -4.79120731e-01
9.49025512e-01 -2.99194843e-01 -2.98927784e-01 5.21637559e-01
-1.06420290e+00 -3.12530786e-01 9.17068720e-02 6.17387831e-01
3.47871184e-01 4.24167961e-01 6.03530645e-01 -1.30596054e+00
1.04193890e+00 -8.26928258e-01 -5.71660399e-01 2.97637343e-01
-1.23749673e+00 -2.02677861e-01 -3.53996605e-01 -2.76309609e-01
-1.12067878e+00 -5.73058188e-01 4.37016964e-01 1.55429259e-01
-3.27450335e-01 7.26159453e-01 1.00512542e-01 5.52171111e-01
1.02626193e+00 -2.55993218e-03 -1.42864659e-01 -3.61783177e-01
8.88332650e-02 6.73591316e-01 -1.61216199e-01 -5.09260714e-01
5.52903771e-01 5.12911081e-01 -5.56341171e-01 -1.01430357e+00
-7.23696887e-01 -1.30132008e+00 -4.25709546e-01 -2.05740839e-01
4.67962205e-01 -1.05196750e+00 2.43176911e-02 -2.33880747e-02
-1.18403614e+00 -4.57794219e-02 -6.49376154e-01 6.70863152e-01
-6.57922208e-01 7.26426542e-02 -9.01747122e-02 -1.00365257e+00
-3.98373842e-01 -8.21449041e-01 1.13060737e+00 7.62761459e-02
-8.45286548e-01 -1.52331603e+00 5.64375520e-01 -3.63279358e-02
3.12210500e-01 4.68525961e-02 9.92191195e-01 -1.32231820e+00
-2.58001029e-01 -3.68644178e-01 -3.20578106e-02 -3.35258305e-01
2.54952181e-02 -1.42542452e-01 -1.14276862e+00 -3.59475464e-01
2.69027889e-01 -2.47606629e-04 9.82127607e-01 9.04067457e-01
6.11671150e-01 -2.91507334e-01 -1.15389168e+00 -1.78155243e-01
1.45071805e+00 3.85891825e-01 5.86034179e-01 5.73978722e-01
-8.26256201e-02 1.06983507e+00 9.62930620e-01 3.56994152e-01
1.79590419e-01 6.38944685e-01 -1.87372312e-01 -7.03876242e-02
4.39594328e-01 -1.47151038e-01 6.61534891e-02 7.63801277e-01
1.38690591e-01 -4.35536742e-01 -1.27171576e+00 1.03570735e+00
-1.64748514e+00 -1.00241995e+00 -2.77355343e-01 2.18840027e+00
6.97033882e-01 4.99982774e-01 5.05941570e-01 1.86836332e-01
8.16840351e-01 5.93177937e-02 3.10151298e-02 -1.90865681e-01
6.52489066e-02 -3.12580094e-02 4.72210646e-02 6.67450726e-01
-1.15287936e+00 7.45324731e-01 7.51605558e+00 9.69298899e-01
-4.68079805e-01 2.03814656e-01 6.86425209e-01 1.72939226e-01
-7.23440289e-01 2.60309517e-01 -1.21574461e+00 3.18874836e-01
1.22314453e+00 -7.99788952e-01 -7.23017991e-01 1.29564273e+00
3.08709830e-01 -6.41052604e-01 -1.12357938e+00 2.24204198e-01
2.23971456e-01 -1.34269929e+00 1.62213728e-01 5.45672357e-01
1.13242662e+00 -2.43318588e-01 -1.22044943e-01 2.95262903e-01
3.82839262e-01 -7.92596459e-01 3.82689625e-01 6.21346295e-01
5.26817918e-01 -5.99859595e-01 8.78902256e-01 3.75009030e-01
-9.74527180e-01 2.05265939e-01 -5.83613575e-01 7.73947015e-02
8.03223476e-02 9.33356702e-01 -1.22028220e+00 2.34224081e-01
6.74082160e-01 5.03561497e-01 -3.82504046e-01 1.24351418e+00
2.92681575e-01 9.80315745e-01 -3.83582115e-01 -1.36322126e-01
2.40494162e-01 1.12140514e-01 7.82620370e-01 1.44484735e+00
3.21086287e-01 -1.89933464e-01 1.84589788e-01 9.52150524e-01
4.08488750e-01 4.79184002e-01 -8.69938552e-01 1.30199045e-01
6.44230843e-01 9.61174011e-01 -1.20699370e+00 -7.57434726e-01
-2.30943952e-02 -1.43079445e-01 -2.93013424e-01 1.73580110e-01
-3.86224538e-01 -3.40984285e-01 2.84455478e-01 5.97195685e-01
-7.61418566e-02 5.12381457e-02 -5.30824959e-01 -8.51937592e-01
-2.40224808e-01 -6.58511043e-01 4.38169479e-01 -4.84342843e-01
-1.08288264e+00 5.80644190e-01 9.97229636e-01 -1.17472661e+00
-5.32779396e-01 -1.07743196e-01 -1.01802373e+00 8.40987742e-01
-1.10448217e+00 -8.23717594e-01 -2.15284288e-01 2.67121464e-01
9.26279128e-01 -3.37312549e-01 8.56525064e-01 -3.25750470e-01
2.19859719e-01 1.91980973e-01 3.97617072e-01 -4.40128714e-01
6.52149260e-01 -1.37905300e+00 4.03127342e-01 3.33580106e-01
1.93771839e-01 7.70030677e-01 1.04776216e+00 -9.69315231e-01
-3.67250741e-01 -1.02586901e+00 1.34211969e+00 -5.90294957e-01
5.81863821e-01 -4.10661578e-01 -1.14479578e+00 5.32454550e-01
2.65019983e-01 -1.15810812e+00 1.08819950e+00 7.36420870e-01
-7.95304999e-02 1.87202170e-01 -1.00963485e+00 2.64133643e-02
2.28885189e-01 -2.53050566e-01 -9.14651573e-01 8.74543965e-01
9.64765728e-01 1.71061710e-01 -9.99796867e-01 7.80570209e-02
4.20531183e-01 -7.49462366e-01 7.07524478e-01 -4.87102032e-01
5.08211970e-01 2.57459402e-01 -1.31264374e-01 -1.11626565e+00
-1.71522930e-01 -4.20693040e-01 1.66606247e-01 1.47710752e+00
8.35525453e-01 -4.73759681e-01 9.69626844e-01 6.52440429e-01
1.33669361e-01 -7.37846673e-01 -6.56413078e-01 -6.57015204e-01
3.35987359e-01 -6.07031882e-01 5.61417937e-01 1.01552641e+00
5.33614099e-01 2.91510940e-01 1.69849142e-01 -9.66739282e-02
7.43771791e-01 2.42414773e-01 8.69733214e-01 -1.84833717e+00
1.55123487e-01 -4.84523892e-01 -3.35081428e-01 -6.71704412e-01
-1.81171820e-01 -5.36313176e-01 8.21575522e-02 -1.66206813e+00
5.66671073e-01 -5.82457542e-01 3.45172696e-02 -7.76224881e-02
-3.88344191e-02 -2.46393934e-01 -1.50721699e-01 7.87216485e-01
-7.07806945e-01 2.79989183e-01 4.99261588e-01 -2.01926172e-01
-5.91554821e-01 3.23744595e-01 -7.52375007e-01 9.57532585e-01
8.83535445e-01 -9.45490599e-01 -6.36598825e-01 1.66636929e-01
1.68932617e-01 -1.19491659e-01 7.61743784e-02 -1.03247547e+00
2.69900739e-01 -2.16650128e-01 8.31714049e-02 -1.17446077e+00
9.95789543e-02 -6.15077257e-01 8.57951418e-02 2.99860716e-01
-8.35001588e-01 -1.46098346e-01 1.04853876e-01 6.18519008e-01
-4.35854226e-01 -6.35430515e-01 4.42832142e-01 -3.36341888e-01
-3.59291136e-01 -4.95777815e-04 -6.95774496e-01 1.50263369e-01
1.03841269e+00 -3.15318316e-01 -2.46015415e-01 -6.48193598e-01
-7.50967324e-01 1.81620583e-01 4.10436213e-01 2.51771927e-01
3.29859793e-01 -1.06280708e+00 -8.19214165e-01 -1.74981773e-01
4.91859108e-01 -1.27027482e-01 -1.24996841e-01 6.17379546e-01
-1.34509310e-01 1.13442171e+00 3.41925561e-01 -8.45671296e-01
-1.15937173e+00 4.51227695e-01 -1.06209017e-01 -8.54356289e-01
-2.92592794e-01 5.58049858e-01 7.39825904e-01 -2.88567036e-01
2.74443775e-01 -3.87725353e-01 -5.13958275e-01 4.74794775e-01
5.80909610e-01 5.59842050e-01 -1.15685545e-01 -4.44665700e-01
-7.23916478e-03 4.20707166e-01 -4.61940974e-01 -4.07542378e-01
1.15478706e+00 -2.18588844e-01 -1.48771212e-01 9.37753320e-01
9.09784794e-01 -3.80077094e-01 -5.76495111e-01 -3.56050164e-01
5.81705630e-01 -2.43199646e-01 -7.53644705e-02 -6.15924180e-01
-2.57320315e-01 7.23446131e-01 5.39535105e-01 1.03581738e+00
8.12375009e-01 4.65314150e-01 1.28378766e-02 2.71640629e-01
2.50453234e-01 -1.13644862e+00 1.11815996e-01 2.91478008e-01
8.53593349e-01 -1.31711340e+00 4.57769006e-01 -5.59911489e-01
-3.48748028e-01 8.87663603e-01 2.99973935e-01 -7.21636117e-02
1.26050043e+00 -4.21376079e-02 -2.16968715e-01 -8.96730602e-01
-9.00099099e-01 4.63165008e-02 7.33912945e-01 6.39479637e-01
6.15914762e-01 1.11021399e-01 -7.30901659e-01 6.13905132e-01
-5.74339211e-01 -3.36077452e-01 3.79221439e-01 7.82417178e-01
-1.16686928e+00 -7.44173467e-01 -6.23495817e-01 8.92052472e-01
-5.53701997e-01 -6.73478991e-02 -3.82080197e-01 1.23235762e+00
-1.25145912e-01 1.21204174e+00 2.77660102e-01 -1.23631723e-01
-4.50142920e-02 4.01699215e-01 -3.97903442e-01 -1.01671481e+00
-5.20047128e-01 6.38155162e-01 1.02224141e-01 -1.03050672e-01
-6.99166298e-01 -9.24996078e-01 -6.01130664e-01 8.31324831e-02
-1.09120619e+00 1.04654408e+00 8.89438391e-01 8.64395618e-01
6.07511327e-02 2.54547060e-01 5.47167122e-01 -4.39013392e-01
-1.21756554e-01 -1.63339245e+00 -9.97951269e-01 2.12259784e-01
-2.02488065e-01 -7.51907468e-01 -4.42845255e-01 3.33532602e-01]
|
[10.374818801879883, 7.102914333343506]
|
00986646-9f08-4268-a63f-eff289a218c6
|
progressive-bilateral-context-driven-model
|
2009.03098
| null |
https://arxiv.org/abs/2009.03098v1
|
https://arxiv.org/pdf/2009.03098v1.pdf
|
Progressive Bilateral-Context Driven Model for Post-Processing Person Re-Identification
|
Most existing person re-identification methods compute pairwise similarity by extracting robust visual features and learning the discriminative metric. Owing to visual ambiguities, these content-based methods that determine the pairwise relationship only based on the similarity between them, inevitably produce a suboptimal ranking list. Instead, the pairwise similarity can be estimated more accurately along the geodesic path of the underlying data manifold by exploring the rich contextual information of the sample. In this paper, we propose a lightweight post-processing person re-identification method in which the pairwise measure is determined by the relationship between the sample and the counterpart's context in an unsupervised way. We translate the point-to-point comparison into the bilateral point-to-set comparison. The sample's context is composed of its neighbor samples with two different definition ways: the first order context and the second order context, which are used to compute the pairwise similarity in sequence, resulting in a progressive post-processing model. The experiments on four large-scale person re-identification benchmark datasets indicate that (1) the proposed method can consistently achieve higher accuracies by serving as a post-processing procedure after the content-based person re-identification methods, showing its state-of-the-art results, (2) the proposed lightweight method only needs about 6 milliseconds for optimizing the ranking results of one sample, showing its high-efficiency. Code is available at: https://github.com/123ci/PBCmodel.
|
['Silong Peng', 'Arjan Kuijper', 'Xiyuan Hu', 'Min Cao', 'Hao Dou', 'Chen Chen']
|
2020-09-07
| null | null | null | null |
['large-scale-person-re-identification']
|
['computer-vision']
|
[-5.79666197e-02 -6.42739415e-01 1.51317388e-01 -4.78678912e-01
-4.73904163e-01 -6.13912582e-01 6.49079859e-01 4.13206488e-01
-7.80497313e-01 2.93381125e-01 3.71983767e-01 1.71955660e-01
-1.53524175e-01 -5.70505500e-01 -1.29816920e-01 -6.62749112e-01
4.78929393e-02 4.81683582e-01 1.53139666e-01 -9.03593823e-02
5.31641841e-01 2.87404180e-01 -1.62443972e+00 -9.03244540e-02
8.20866406e-01 7.33702242e-01 6.78582042e-02 4.31901902e-01
3.38608725e-03 -1.09773256e-01 -5.10161281e-01 -6.10053658e-01
4.17966127e-01 -5.47013998e-01 -6.67695642e-01 -1.25360116e-01
6.52340531e-01 -2.53075063e-01 -2.18372166e-01 1.28856659e+00
7.43518829e-01 3.69561374e-01 4.48397577e-01 -1.08760178e+00
-5.28640568e-01 6.96202368e-02 -6.73408687e-01 1.89414218e-01
7.66208649e-01 8.95559192e-02 9.73997772e-01 -1.00995815e+00
4.96211767e-01 1.26695132e+00 7.23899186e-01 4.73006845e-01
-1.19880962e+00 -6.50534153e-01 1.65070206e-01 5.08534193e-01
-1.79589880e+00 -3.66835803e-01 8.00461054e-01 -4.83835548e-01
5.12925029e-01 5.43034017e-01 8.91546369e-01 7.41585970e-01
-3.77678424e-01 4.28714693e-01 9.56879199e-01 -2.39071131e-01
4.24049795e-02 -4.31412235e-02 3.80147547e-01 4.94737238e-01
2.46921197e-01 9.83235836e-02 -4.68570173e-01 -3.15637589e-01
4.73442644e-01 1.80225655e-01 -2.37736478e-01 -3.74423295e-01
-1.38262296e+00 5.32967091e-01 4.69536752e-01 2.72750944e-01
-1.21271843e-02 -1.73421204e-01 4.02321517e-01 1.09617949e-01
1.49880692e-01 1.13569334e-01 1.30366758e-01 -2.32213289e-01
-9.52165484e-01 4.26442981e-01 5.14117599e-01 7.21403658e-01
8.27136159e-01 -6.44393623e-01 -2.00610697e-01 9.57396448e-01
3.96121591e-01 3.62461120e-01 5.82217395e-01 -5.88511705e-01
5.04646122e-01 6.84146285e-01 3.36872876e-01 -1.45004749e+00
-3.31582040e-01 -3.48178148e-01 -8.95507812e-01 -3.98038030e-02
7.98236370e-01 1.84780940e-01 -3.93347412e-01 1.76812291e+00
5.68297327e-01 3.46670747e-01 -2.51862466e-01 1.12633240e+00
7.16350198e-01 3.71320933e-01 -2.91220695e-02 -1.59214541e-01
1.45623124e+00 -8.16962063e-01 -4.02540565e-01 6.05834983e-02
4.71775502e-01 -7.24977374e-01 9.23148453e-01 1.62818134e-01
-8.22932780e-01 -8.11766863e-01 -1.07736492e+00 -5.70468903e-02
-3.73698175e-01 3.17419142e-01 6.59412965e-02 6.48425579e-01
-1.08170056e+00 6.29504442e-01 -4.47557449e-01 -5.97988009e-01
-6.91261813e-02 2.29531482e-01 -5.74305356e-01 3.82657424e-02
-1.00707114e+00 5.92442811e-01 3.25666636e-01 3.02785486e-01
-2.37678289e-01 -5.17169356e-01 -6.72334731e-01 -2.92866081e-02
2.00965852e-01 -6.35127187e-01 6.32974923e-01 -7.35829830e-01
-1.24575984e+00 9.81114805e-01 -5.69795549e-01 7.80418608e-03
8.53731036e-01 -1.92410246e-01 -4.47431296e-01 1.91992387e-01
2.63732910e-01 5.08164883e-01 6.58610046e-01 -1.31095278e+00
-7.61212170e-01 -5.81052184e-01 -1.33176848e-01 4.61529881e-01
-4.32962745e-01 6.55563027e-02 -8.70324969e-01 -8.07084680e-01
3.04891825e-01 -1.00939143e+00 -2.89022941e-02 4.28490266e-02
-4.51089978e-01 -5.38392901e-01 6.71775460e-01 -9.05104995e-01
1.38488626e+00 -2.32460427e+00 2.17461303e-01 5.12397885e-01
2.69748449e-01 2.33230233e-01 -1.28056929e-01 5.35591781e-01
-1.40861616e-01 1.65081974e-02 -2.81667709e-01 -7.55107045e-01
-2.63429843e-02 -4.39705789e-01 5.98608851e-02 7.57031083e-01
-2.57068515e-01 6.91104591e-01 -1.08789790e+00 -6.81834400e-01
3.35609585e-01 4.49110836e-01 -3.08088392e-01 1.81860402e-01
5.83114922e-01 5.70619881e-01 -2.14266285e-01 4.70759481e-01
7.21787632e-01 -1.90849435e-02 2.72771448e-01 -5.99149108e-01
-2.03772932e-01 -1.01099990e-01 -1.55574203e+00 1.70458198e+00
7.71537125e-02 3.71118903e-01 -2.56413311e-01 -7.17130303e-01
1.02804792e+00 -1.10832110e-01 5.61957300e-01 -6.05200171e-01
-1.18634284e-01 1.86948642e-01 -1.62416905e-01 -2.43336603e-01
5.93319356e-01 2.32080996e-01 2.53749266e-02 4.67596769e-01
-4.20190752e-01 6.19800150e-01 4.02369797e-01 1.94186404e-01
5.05545914e-01 1.64093256e-01 4.02264148e-01 -3.28213632e-01
9.53155100e-01 -2.99996674e-01 6.49102688e-01 5.61693311e-01
-4.97858167e-01 7.93954909e-01 8.22658613e-02 -4.93720323e-01
-1.06755173e+00 -1.17216980e+00 -1.30264163e-01 7.67406225e-01
7.33973205e-01 -7.82633722e-01 -9.32543933e-01 -5.89500785e-01
2.76052766e-02 3.23919952e-01 -5.02364993e-01 1.76893845e-02
-5.87003529e-01 -5.56096375e-01 4.05994385e-01 4.33278233e-01
8.93536329e-01 -5.29443204e-01 -3.86147469e-01 -2.85185017e-02
-5.78980744e-01 -8.75003755e-01 -1.15807354e+00 -8.07353616e-01
-6.58634305e-01 -1.13608694e+00 -1.00602651e+00 -9.08227324e-01
8.82058978e-01 6.09604537e-01 6.02145731e-01 4.43989873e-01
-2.05906004e-01 4.87605721e-01 -3.00430030e-01 3.21972221e-01
1.85931116e-01 -2.98830718e-01 4.87015605e-01 4.25282598e-01
6.94941700e-01 -5.05891979e-01 -1.04190719e+00 7.30484068e-01
-4.13716882e-01 3.31923738e-02 1.75296783e-01 8.77620161e-01
7.08297014e-01 -2.34891456e-02 2.68527478e-01 -2.20951006e-01
6.72269523e-01 -1.46986261e-01 -3.28143299e-01 4.78725612e-01
-6.49302065e-01 -1.19933069e-01 4.26043600e-01 -4.59587991e-01
-7.89523661e-01 1.01429194e-01 1.07813910e-01 -1.24018535e-01
-8.82114097e-02 2.47355565e-01 -2.58798152e-01 -9.64036509e-02
3.36025000e-01 4.79506940e-01 5.41137122e-02 -6.55785382e-01
2.25038171e-01 7.54227579e-01 7.69912541e-01 -5.94183445e-01
9.89599407e-01 6.21559381e-01 -6.05014861e-02 -7.22305179e-01
-3.01919699e-01 -8.47796917e-01 -1.06714475e+00 -3.63457441e-01
8.27642441e-01 -8.59922290e-01 -1.07546580e+00 6.20753050e-01
-1.02480233e+00 1.55453250e-01 3.64774540e-02 5.33579290e-01
-1.09957390e-01 1.13816905e+00 -3.53354126e-01 -7.33967900e-01
-4.23210979e-01 -1.05640745e+00 1.00902534e+00 3.06337982e-01
-4.06997591e-01 -7.55737305e-01 2.33450696e-01 4.05985683e-01
4.56848275e-03 7.55790025e-02 5.72315991e-01 -5.81281364e-01
-2.40974009e-01 -3.61840695e-01 -4.15782720e-01 2.97134724e-02
3.18134844e-01 -1.52970314e-01 -7.77428687e-01 -7.03383803e-01
-3.63544524e-01 3.49039465e-01 5.74007630e-01 -2.56186929e-02
1.04846001e+00 -2.03134060e-01 -4.28320050e-01 7.63301194e-01
1.32560182e+00 -6.49624839e-02 5.75802445e-01 4.69870478e-01
1.04130065e+00 8.72924089e-01 6.11908972e-01 4.90775019e-01
6.43315852e-01 1.10564864e+00 3.18139270e-02 1.97009280e-01
-1.23850532e-01 -4.77273911e-01 3.17325681e-01 7.89101243e-01
-3.43022138e-01 1.38199419e-01 -7.77027726e-01 4.10697520e-01
-1.99339008e+00 -1.23232186e+00 -2.06706554e-01 2.83821392e+00
5.69607437e-01 -3.24362248e-01 6.29597127e-01 2.10540920e-01
1.14044774e+00 1.69340465e-02 -4.11774188e-01 1.43570334e-01
6.24217391e-02 -4.54695106e-01 2.54060835e-01 6.08294725e-01
-1.09261167e+00 6.25541389e-01 5.46077490e+00 8.38947713e-01
-7.72246063e-01 -1.33248463e-01 3.82412046e-01 -7.44786710e-02
-1.16062321e-01 1.50711248e-02 -8.90469491e-01 8.65393937e-01
4.47565079e-01 -2.97211796e-01 3.85323673e-01 5.21521807e-01
2.70213813e-01 -1.35750875e-01 -1.25467229e+00 1.70114577e+00
2.86498249e-01 -8.72035623e-01 1.85640097e-01 2.47849807e-01
2.55139560e-01 -5.83012044e-01 2.39014067e-02 -2.29229957e-01
-2.74137169e-01 -7.78385222e-01 8.08237553e-01 6.82518244e-01
8.39703381e-01 -8.51969004e-01 7.10082293e-01 8.84479582e-02
-1.77437830e+00 -4.92855385e-02 -3.16013366e-01 1.77497114e-03
1.58686399e-01 4.00660932e-01 -4.03669149e-01 7.35661328e-01
9.95390892e-01 9.93392825e-01 -7.56661713e-01 1.22833395e+00
6.53884187e-02 1.14498764e-01 -2.22313911e-01 1.00497610e-03
-2.00978965e-01 -7.83337951e-01 8.13644528e-01 1.30763960e+00
4.82590348e-01 3.74822356e-02 2.65763521e-01 6.84974730e-01
2.38662153e-01 3.52998018e-01 -1.50967002e-01 3.65569502e-01
5.49321413e-01 1.24960887e+00 -6.67820990e-01 -2.90156007e-01
-2.80212194e-01 1.37147939e+00 3.04242700e-01 4.20423895e-01
-6.96953058e-01 -4.43435371e-01 7.65747666e-01 8.36439151e-03
1.49833038e-02 -3.95704865e-01 -2.95338124e-01 -1.12660336e+00
4.23158586e-01 -7.86827981e-01 6.01231694e-01 -4.04771596e-01
-1.57329559e+00 4.87807333e-01 9.32986736e-02 -1.50987458e+00
-2.00149179e-01 -4.84566212e-01 -5.72318077e-01 1.08216488e+00
-1.03948951e+00 -1.04065061e+00 -6.85022056e-01 7.76352465e-01
2.12095588e-01 -1.90083399e-01 7.31653631e-01 3.83367360e-01
-6.46005988e-01 1.04752028e+00 1.69549197e-01 4.37962472e-01
9.50705409e-01 -1.17818534e+00 3.98171514e-01 1.08724689e+00
6.21892922e-02 1.04879451e+00 5.70113182e-01 -7.29400933e-01
-1.18609178e+00 -7.23545909e-01 1.04128551e+00 -5.06381929e-01
3.72518122e-01 -3.79409462e-01 -8.33819926e-01 2.68904775e-01
-2.44367480e-01 -2.23767385e-01 9.14968491e-01 1.81630030e-01
-6.24887049e-01 -4.19205189e-01 -1.19406533e+00 8.21843147e-01
1.47305489e+00 -6.73908889e-01 -5.49541056e-01 -3.88670154e-02
2.77422428e-01 -4.38547134e-02 -8.44664931e-01 1.37387022e-01
8.98882210e-01 -1.05538154e+00 1.26037574e+00 -1.59851834e-01
-5.28630577e-02 -8.24065030e-01 -1.33681104e-01 -9.93575752e-01
-5.78798652e-01 -5.58436394e-01 4.66172546e-02 1.42090368e+00
-1.49375081e-01 -9.22539413e-01 5.41072726e-01 6.19251490e-01
3.29887033e-01 -5.31562388e-01 -1.02543688e+00 -7.48673856e-01
-3.80053669e-01 -1.89811159e-02 7.83319890e-01 7.75261521e-01
1.28490269e-01 -2.80420464e-02 -4.35599238e-01 2.45770648e-01
1.05136824e+00 1.74261451e-01 7.75198877e-01 -1.29744363e+00
-2.20801055e-01 -5.73413074e-01 -7.92980134e-01 -1.30866969e+00
-4.35728207e-02 -8.12087893e-01 -9.10719037e-02 -1.27930403e+00
6.79724574e-01 -5.22952974e-01 -3.70261729e-01 2.04357460e-01
-5.25493085e-01 3.40471566e-01 4.23257232e-01 6.86456978e-01
-5.15631020e-01 4.56763536e-01 8.92063498e-01 -1.84478879e-01
-1.90510124e-01 -3.86321060e-02 -6.45660281e-01 5.23252010e-01
6.90423846e-01 -7.88114741e-02 -2.87796766e-01 -2.46106356e-01
-5.94340032e-03 -4.49294120e-01 7.50817180e-01 -1.17034435e+00
4.91516113e-01 2.37525720e-02 5.96127331e-01 -6.04620278e-01
3.72153401e-01 -6.11750960e-01 2.94419318e-01 3.80184054e-01
-2.03585252e-01 3.45597386e-01 -1.69367284e-01 6.53141022e-01
-1.80569530e-01 -2.28972241e-01 6.31211579e-01 6.34225458e-03
-8.11060667e-01 4.76617008e-01 1.96103051e-01 -2.28571132e-01
9.35311556e-01 -6.31424665e-01 -1.74133807e-01 -3.81787658e-01
-4.94262546e-01 2.09827974e-01 8.59969974e-01 4.47614819e-01
7.98571408e-01 -1.55012214e+00 -8.44213367e-01 1.12051114e-01
2.48199880e-01 -4.12086993e-01 3.53739113e-01 8.29612553e-01
-3.49602163e-01 1.10911153e-01 -1.82760194e-01 -7.30401814e-01
-1.64838541e+00 5.73430896e-01 3.85110289e-01 -3.32744941e-02
-6.47968948e-01 6.00873053e-01 3.03036541e-01 -3.15689594e-01
1.48185194e-01 2.65363574e-01 -4.43259507e-01 1.96530640e-01
8.82627726e-01 7.46182978e-01 -3.20687592e-01 -1.17029750e+00
-6.76770091e-01 1.12857974e+00 -1.29707456e-02 -2.39452913e-01
8.44350696e-01 -4.63228077e-01 -3.18292707e-01 3.23285311e-01
1.42031562e+00 1.45162955e-01 -1.16870320e+00 -2.98196703e-01
1.67605339e-03 -8.81485522e-01 -3.83514494e-01 -4.27821338e-01
-7.93899357e-01 7.17899740e-01 1.02230835e+00 -9.42294374e-02
1.04333866e+00 -2.87836492e-01 7.71349847e-01 1.17316484e-01
5.50669849e-01 -1.17039132e+00 3.13118957e-02 1.85453326e-01
8.00249040e-01 -1.23916912e+00 1.82838425e-01 -5.11559427e-01
-5.49218237e-01 1.05347288e+00 4.66199756e-01 -1.35610218e-03
5.25163233e-01 -4.02999729e-01 -6.46181554e-02 1.68117374e-01
1.51267096e-01 -1.94079295e-01 6.02450967e-01 6.81166589e-01
2.54439890e-01 2.04855725e-01 -5.55549324e-01 5.00632823e-01
-3.43999386e-01 -3.30298901e-01 -7.96464533e-02 4.65512842e-01
-1.97790593e-01 -1.22046733e+00 -4.86891568e-01 2.07998559e-01
3.48630315e-03 -6.63137957e-02 -3.50409895e-01 3.98120850e-01
1.31577447e-01 1.15663552e+00 1.75972089e-01 -6.52830064e-01
3.24786246e-01 -1.66723400e-01 3.54788840e-01 -2.15487048e-01
-4.28211629e-01 -1.96821362e-01 -1.08694099e-01 -7.04979479e-01
-3.84895682e-01 -1.05999684e+00 -1.11472952e+00 -5.13207197e-01
2.46715057e-03 1.55356154e-01 4.35859382e-01 8.13130200e-01
4.88369495e-01 -1.71776652e-01 7.25753069e-01 -8.98222029e-01
-5.13076544e-01 -8.10976028e-01 -3.64256144e-01 1.00612843e+00
1.45925075e-01 -7.17833042e-01 -3.74000907e-01 4.70520668e-02]
|
[14.783207893371582, 1.0392581224441528]
|
56a521f2-3d96-46c8-b8be-61686feeb509
|
learning-to-adapt-multi-view-stereo-by-self
|
2009.13278
| null |
https://arxiv.org/abs/2009.13278v1
|
https://arxiv.org/pdf/2009.13278v1.pdf
|
Learning to Adapt Multi-View Stereo by Self-Supervision
|
3D scene reconstruction from multiple views is an important classical problem in computer vision. Deep learning based approaches have recently demonstrated impressive reconstruction results. When training such models, self-supervised methods are favourable since they do not rely on ground truth data which would be needed for supervised training and is often difficult to obtain. Moreover, learned multi-view stereo reconstruction is prone to environment changes and should robustly generalise to different domains. We propose an adaptive learning approach for multi-view stereo which trains a deep neural network for improved adaptability to new target domains. We use model-agnostic meta-learning (MAML) to train base parameters which, in turn, are adapted for multi-view stereo on new domains through self-supervised training. Our evaluations demonstrate that the proposed adaptation method is effective in learning self-supervised multi-view stereo reconstruction in new domains.
|
['Jörg Stückler', 'Arijit Mallick', 'Hendrik Lensch']
|
2020-09-28
| null | null | null | null |
['3d-scene-reconstruction']
|
['computer-vision']
|
[ 2.54274637e-01 -1.54490396e-01 4.05447856e-02 -6.05454445e-01
-6.56749904e-01 -4.09956694e-01 7.43179321e-01 -1.50313869e-01
-4.36940223e-01 7.09764779e-01 6.20421357e-02 2.28059828e-01
-7.51852021e-02 -9.01401401e-01 -1.02562916e+00 -6.28203452e-01
3.76228124e-01 9.07494068e-01 4.70088720e-01 -3.94983888e-01
1.97156876e-01 6.95980012e-01 -1.84864461e+00 4.45848465e-01
5.98771632e-01 6.15396678e-01 8.27441692e-01 5.92052639e-01
6.14271387e-02 7.04346418e-01 -1.90425873e-01 -9.05151218e-02
3.99429858e-01 -2.48459175e-01 -8.47077489e-01 6.24446034e-01
5.96068084e-01 -4.16524470e-01 -1.71795711e-01 9.15259361e-01
5.00784576e-01 2.06243157e-01 6.12541497e-01 -6.59684539e-01
-1.65923014e-01 4.08925815e-03 -3.42699945e-01 2.71063805e-01
4.01168853e-01 1.59335732e-01 5.70567310e-01 -8.37925017e-01
9.05461311e-01 1.14037466e+00 7.14553952e-01 7.31835425e-01
-1.33133531e+00 -2.22993925e-01 1.54898679e-02 3.96693408e-01
-7.99107909e-01 -5.17094016e-01 1.03173637e+00 -3.33933204e-01
1.01542878e+00 -9.62973461e-02 8.42814028e-01 1.31239772e+00
2.40933329e-01 7.48262465e-01 1.46362638e+00 -4.29211169e-01
2.82058120e-01 3.62579912e-01 -6.04248941e-01 4.42265242e-01
5.03599420e-02 5.94534993e-01 -5.50113261e-01 2.90198654e-01
1.02786982e+00 3.03056687e-02 -9.19824839e-02 -1.06615341e+00
-1.34339666e+00 7.71973968e-01 4.99906331e-01 1.79018125e-01
-4.98684525e-01 -1.84795186e-01 5.63945591e-01 4.56027359e-01
4.71301526e-01 4.47449803e-01 -6.21197402e-01 1.56438872e-01
-6.85629606e-01 7.14873709e-03 3.29493850e-01 8.76740098e-01
9.09404576e-01 2.84827352e-01 7.24199951e-01 1.12669420e+00
1.32718265e-01 5.74868560e-01 7.61324644e-01 -1.20851338e+00
3.55405986e-01 6.58227682e-01 -1.23003125e-01 -7.62239337e-01
-5.10821164e-01 -4.65878159e-01 -1.04517770e+00 5.10817170e-01
1.70377046e-01 1.71303689e-01 -9.09334123e-01 1.42095256e+00
5.28887451e-01 -8.96601379e-02 3.71987313e-01 6.70183718e-01
8.20955813e-01 3.94179672e-01 -3.36802006e-01 -9.36217308e-02
5.35060525e-01 -9.39250708e-01 -1.83809906e-01 -6.11469626e-01
3.13562155e-01 -6.81119978e-01 8.31289828e-01 6.39828742e-01
-1.13102472e+00 -9.88229036e-01 -9.33291852e-01 2.83906430e-01
-3.83289814e-01 -4.27923441e-01 4.32363003e-01 3.42972428e-01
-9.62802529e-01 6.90608084e-01 -6.75237119e-01 -7.12273717e-01
1.97132349e-01 4.74210471e-01 -8.28059018e-01 -2.44444251e-01
-8.37463140e-01 1.02016211e+00 8.35156381e-01 -1.50521785e-01
-1.07147288e+00 -2.51538008e-01 -1.05665243e+00 -4.23025310e-01
3.14247906e-01 -1.10453475e+00 1.07831573e+00 -1.31460190e+00
-1.76715124e+00 1.21621323e+00 3.79989594e-02 -4.45234984e-01
5.22163510e-01 -2.82884892e-02 -3.91945541e-01 3.19904864e-01
7.73271844e-02 6.09321952e-01 1.27413797e+00 -1.75031173e+00
-7.66033828e-01 -5.34422576e-01 1.71217903e-01 5.55097938e-01
-1.45207822e-01 -5.48995614e-01 -1.73405439e-01 -3.22058141e-01
4.96852458e-01 -8.68232131e-01 -3.71176392e-01 -3.47046226e-01
-2.35809255e-02 3.64059329e-01 7.67207921e-01 -3.17475080e-01
4.71474051e-01 -1.78548229e+00 3.78435463e-01 -2.05075201e-02
-3.93068120e-02 2.30808854e-01 -8.99227187e-02 3.92907321e-01
-1.11624852e-01 -6.13530457e-01 -2.38062777e-02 -1.84645921e-01
-4.50642496e-01 6.18040860e-01 -1.99021790e-02 3.49409610e-01
-1.33735076e-01 5.69632292e-01 -9.85009670e-01 -3.69396746e-01
7.42597103e-01 2.85598665e-01 -7.19999075e-01 5.20640671e-01
-2.49358341e-01 1.16268194e+00 -2.06023064e-02 3.45964819e-01
6.71219110e-01 -4.19324547e-01 3.09412807e-01 -3.18170667e-01
3.00606526e-02 -1.58355832e-01 -1.16993248e+00 2.04747915e+00
-1.10771048e+00 3.12928677e-01 -9.63485911e-02 -1.31503952e+00
1.09276783e+00 1.37603998e-01 4.73742723e-01 -8.19891095e-01
8.00798312e-02 1.60505489e-01 -2.29563206e-01 -6.26449525e-01
3.58683825e-01 -5.88609576e-01 1.64291114e-01 3.02274466e-01
4.24599886e-01 -7.33867824e-01 -1.54643983e-01 -2.64973909e-01
6.54782236e-01 3.95217538e-01 6.62219107e-01 7.27936327e-02
9.27919090e-01 6.83334172e-02 5.47919571e-01 6.27717137e-01
1.10668801e-01 8.22727621e-01 -5.01896679e-01 -1.08121860e+00
-1.36968064e+00 -1.02234256e+00 -1.38724878e-01 9.01046813e-01
2.53793925e-01 1.37092590e-01 -3.85611594e-01 -8.43767464e-01
-1.19347103e-01 5.31681597e-01 -4.08539623e-01 -2.95902975e-02
-5.60507298e-01 -4.69353348e-01 -2.85617560e-01 5.23825824e-01
7.71049559e-01 -1.13536453e+00 -8.87847006e-01 4.61327910e-01
-1.52755782e-01 -1.28648221e+00 9.26904082e-02 4.12353605e-01
-1.49597621e+00 -1.21167207e+00 -7.91503727e-01 -7.74075568e-01
8.32773626e-01 5.96006632e-01 1.19498956e+00 -2.42540389e-01
1.02677956e-01 1.01864696e+00 -2.05323771e-01 -1.99259475e-01
-8.97623122e-01 2.88777761e-02 3.15469474e-01 1.33054048e-01
2.01351270e-01 -1.04192626e+00 -4.36580598e-01 5.89533567e-01
-1.02007294e+00 3.23386967e-01 6.86102390e-01 1.30576193e+00
8.94724607e-01 -1.37386983e-02 4.32078093e-01 -1.06379700e+00
8.59744996e-02 -2.30578735e-01 -9.00446534e-01 2.36859024e-01
-4.96557325e-01 1.43638656e-01 9.22708690e-01 -2.94671595e-01
-1.38152266e+00 5.04369259e-01 -2.89931893e-01 -5.61374962e-01
-7.49951661e-01 2.80853659e-01 -2.04037264e-01 -5.17301440e-01
8.88957143e-01 5.47265053e-01 -6.13260409e-03 -4.34076071e-01
1.81237280e-01 5.00937164e-01 4.73882169e-01 -3.74250740e-01
8.81549478e-01 8.30484748e-01 1.28722042e-01 -9.14890170e-01
-1.01055503e+00 -5.13959825e-01 -1.32241881e+00 -4.50529426e-01
7.21513629e-01 -1.17551637e+00 -2.84051180e-01 6.87875032e-01
-9.59642649e-01 -4.52068716e-01 -3.21818262e-01 5.01072466e-01
-1.18675184e+00 5.51573455e-01 -8.21768939e-02 -4.49757665e-01
-6.95191994e-02 -9.68580484e-01 1.06769633e+00 1.74965143e-01
4.56960835e-02 -1.66073835e+00 2.89020658e-01 6.99914753e-01
2.68956304e-01 3.21045756e-01 6.56265557e-01 -4.66074944e-01
-5.57427526e-01 -1.35342345e-01 7.45439827e-02 5.98616600e-01
2.81194240e-01 -5.13388515e-01 -1.11131120e+00 -6.06291652e-01
3.54541063e-01 -7.68554926e-01 6.81337118e-01 3.39663357e-01
1.00210595e+00 -2.06519826e-03 -1.62077740e-01 9.02223647e-01
1.72033131e+00 4.99982014e-02 3.82118434e-01 9.42850292e-01
6.21322095e-01 5.53118646e-01 6.15186036e-01 4.41568106e-01
5.04110754e-01 8.09024513e-01 7.90064454e-01 -1.26981497e-01
3.48079614e-02 -1.87004417e-01 1.13929287e-01 8.60856771e-01
-3.04930627e-01 2.03851581e-01 -9.40687597e-01 5.62211692e-01
-1.69740820e+00 -1.12294912e+00 1.48380145e-01 2.21530724e+00
3.67293179e-01 2.65797198e-01 7.20742047e-02 6.63905516e-02
4.31459844e-01 5.37787192e-02 -8.32572281e-01 -2.31041878e-01
-2.34410927e-01 2.93942541e-02 4.48756099e-01 3.15592021e-01
-1.18664467e+00 9.39959109e-01 6.19044113e+00 2.45761558e-01
-1.22037232e+00 1.46045595e-01 2.83440024e-01 1.44104019e-01
-2.26377189e-01 -1.00631483e-01 -5.57643831e-01 3.78544815e-02
5.65349340e-01 1.31895766e-01 4.23863173e-01 9.82478261e-01
1.32862981e-02 -3.16376984e-01 -1.14749467e+00 1.30546618e+00
2.68891543e-01 -1.40275884e+00 2.41757587e-01 5.02154790e-03
1.20864201e+00 3.31064373e-01 2.05736142e-02 6.55664951e-02
4.49078292e-01 -5.79028010e-01 4.09007907e-01 4.63256389e-01
4.70488101e-01 -6.65909946e-01 7.05813825e-01 8.05189729e-01
-8.67557049e-01 -3.15541506e-01 -5.97103655e-01 1.44489512e-01
2.78168797e-01 3.01990241e-01 -8.47913384e-01 8.03942800e-01
7.58405387e-01 1.14271522e+00 -5.29225290e-01 9.45775986e-01
-7.74400756e-02 -2.93086004e-02 -1.09244592e-01 5.34043252e-01
2.01488301e-01 -1.40130579e-01 8.04968059e-01 7.23958969e-01
3.19477856e-01 -2.87111402e-01 1.40017614e-01 3.69041234e-01
3.51633847e-01 -2.55934000e-02 -1.23339760e+00 5.50807595e-01
-7.25364909e-02 9.10071552e-01 -4.85700667e-01 -3.42508018e-01
-5.92744172e-01 1.15772855e+00 4.54735070e-01 2.49150753e-01
-2.52423048e-01 4.17678922e-01 2.42952958e-01 2.24037349e-01
5.42581677e-01 -1.96773082e-01 -7.49188811e-02 -1.51358688e+00
-3.36428732e-01 -1.00589943e+00 5.18354774e-01 -8.63111377e-01
-1.44866502e+00 8.78505051e-01 1.30873889e-01 -1.79235697e+00
-7.20664561e-01 -6.32302225e-01 -2.02159971e-01 4.69044685e-01
-1.77387810e+00 -1.29389238e+00 -5.02288401e-01 9.97103393e-01
9.15052354e-01 -7.89500356e-01 9.67165232e-01 -2.81357579e-02
-2.90149245e-02 2.25432351e-01 4.59605396e-01 -3.15164298e-01
7.66616404e-01 -1.31520391e+00 1.33399591e-01 6.48931980e-01
2.24379793e-01 2.34144777e-01 7.04404414e-01 -4.10839140e-01
-1.39718258e+00 -1.12796617e+00 4.00962263e-01 -3.01973611e-01
2.73900658e-01 -2.68079210e-02 -8.22797716e-01 6.27331257e-01
2.94406593e-01 1.51776284e-01 6.39426351e-01 9.36555564e-02
-1.99868396e-01 -4.18690026e-01 -1.23369479e+00 3.12711746e-01
1.13922954e+00 -4.82568383e-01 -6.48388565e-01 2.42005855e-01
3.72522250e-02 -6.20192468e-01 -8.51138413e-01 5.35268188e-01
4.93840694e-01 -1.73245084e+00 1.22043478e+00 -5.96227467e-01
3.61846387e-01 -4.75836024e-02 -3.96096259e-01 -1.65270710e+00
-3.80191177e-01 -6.95991665e-02 9.73988548e-02 6.74393117e-01
6.37597740e-02 -6.53393984e-01 9.38082099e-01 5.15187420e-02
-7.61676729e-02 -3.75984162e-01 -9.07805860e-01 -1.02840877e+00
-8.93788189e-02 -2.07303792e-01 4.58347201e-01 1.05851758e+00
-4.24385995e-01 4.76240575e-01 -6.03239715e-01 3.13012064e-01
9.93450880e-01 4.90289509e-01 1.27512527e+00 -1.50320554e+00
-7.27096915e-01 -5.60293579e-03 -5.74482501e-01 -9.32049513e-01
3.63314092e-01 -8.36264491e-01 -9.28066894e-02 -1.57097650e+00
1.86697453e-01 -2.94653028e-01 -1.86352521e-01 1.18079655e-01
1.36595413e-01 1.99918404e-01 8.36396441e-02 2.73575842e-01
-6.71001971e-01 7.16534317e-01 1.50615084e+00 -1.74774542e-01
-3.06142122e-01 4.02013868e-01 -1.76409930e-01 1.03647578e+00
8.54888380e-01 -3.18735778e-01 -4.97932523e-01 -6.84240162e-01
2.42696196e-01 2.32299849e-01 3.04264784e-01 -1.36308074e+00
1.74308687e-01 -2.45973691e-01 5.67230046e-01 -5.72525620e-01
6.35673702e-01 -1.35075629e+00 3.55829030e-01 3.83746237e-01
-2.80910218e-03 7.23513663e-02 2.36591384e-01 7.71286964e-01
-4.63943481e-01 -3.04271072e-01 1.10465133e+00 -8.99646759e-01
-1.21601248e+00 1.84925199e-01 -2.96166182e-01 -1.68969706e-01
8.74531329e-01 -7.41000950e-01 2.88045675e-01 -5.92778802e-01
-1.10617089e+00 -3.34725678e-02 8.64493251e-01 3.61592591e-01
9.53611314e-01 -1.36635876e+00 -5.22695363e-01 4.40310031e-01
4.47799206e-01 2.59551018e-01 5.29480159e-01 4.92049187e-01
-5.76372683e-01 1.57322615e-01 -7.83497214e-01 -1.11535430e+00
-1.24955022e+00 7.22344100e-01 4.94123161e-01 -3.68457526e-01
-6.85023189e-01 5.40831089e-01 3.04272741e-01 -1.01081729e+00
-6.02628402e-02 1.57597333e-01 -2.97418058e-01 -1.91017106e-01
1.64095595e-01 1.83717236e-01 7.64537677e-02 -7.93152392e-01
-7.06582144e-03 8.97165120e-01 -3.70267592e-02 -1.32669806e-01
1.84713411e+00 -4.53990251e-01 1.40615523e-01 6.17015958e-01
1.06631362e+00 -4.35190529e-01 -1.49911976e+00 -6.63668513e-01
-9.44583640e-02 -6.30268037e-01 1.18836373e-01 -5.88989496e-01
-1.00749040e+00 1.00125074e+00 7.34672487e-01 -9.92150977e-02
1.33307159e+00 -1.30175695e-01 6.28397763e-01 6.50884032e-01
7.66815126e-01 -1.29927146e+00 5.02833426e-01 7.07693815e-01
9.47755694e-01 -1.75784421e+00 1.00435719e-01 -1.28743164e-02
-5.87868392e-01 1.58775580e+00 8.39426994e-01 -3.58527228e-02
6.15130067e-01 -1.21448077e-01 2.63103604e-01 -1.65668815e-01
-6.16288483e-01 -2.82318830e-01 1.38721824e-01 1.05985355e+00
-1.29707873e-01 -3.68852943e-01 4.24503565e-01 -2.30339110e-01
-3.81620630e-04 -1.18625544e-01 5.44728696e-01 9.35218573e-01
-4.52899694e-01 -1.31901431e+00 -4.36220497e-01 2.37387314e-01
1.58051282e-01 2.93919951e-01 -1.04960725e-01 7.42112637e-01
1.59274399e-01 5.98790288e-01 -1.77441940e-01 -3.54579985e-01
4.83152807e-01 -9.55724865e-02 9.96805012e-01 -8.29447865e-01
-4.16696876e-01 6.54261038e-02 -9.79386941e-02 -4.80044425e-01
-1.02327454e+00 -8.48529637e-01 -6.07882798e-01 -7.43650720e-02
-1.02099136e-01 -3.33391398e-01 4.91117746e-01 1.04229546e+00
2.77615905e-01 2.37716630e-01 1.07774544e+00 -1.18188334e+00
-3.99049163e-01 -6.13524795e-01 -4.27919239e-01 6.40621006e-01
3.99650395e-01 -7.37555623e-01 -4.23249565e-02 3.10058832e-01]
|
[8.665449142456055, -2.428774356842041]
|
cdb6432b-c164-4196-8bf3-efa9fbeaf03b
|
stochastic-package-queries-in-probabilistic
|
2103.06784
| null |
https://arxiv.org/abs/2103.06784v1
|
https://arxiv.org/pdf/2103.06784v1.pdf
|
Stochastic Package Queries in Probabilistic Databases
|
We provide methods for in-database support of decision making under uncertainty. Many important decision problems correspond to selecting a package (bag of tuples in a relational database) that jointly satisfy a set of constraints while minimizing some overall cost function; in most real-world problems, the data is uncertain. We provide methods for specifying -- via a SQL extension -- and processing stochastic package queries (SPQs), in order to solve optimization problems over uncertain data, right where the data resides. Prior work in stochastic programming uses Monte Carlo methods where the original stochastic optimization problem is approximated by a large deterministic optimization problem that incorporates many scenarios, i.e., sample realizations of the uncertain data values. For large database tables, however, a huge number of scenarios is required, leading to poor performance and, often, failure of the solver software. We therefore provide a novel SummarySearch algorithm that, instead of trying to solve a large deterministic problem, seamlessly approximates it via a sequence of smaller problems defined over carefully crafted summaries of the scenarios that accelerate convergence to a feasible and near-optimal solution. Experimental results on our prototype system show that SummarySearch can be orders of magnitude faster than prior methods at finding feasible and high-quality packages.
|
['Alexandra Meliou', 'Peter J. Haas', 'Azza Abouzied', 'Nishant Yadav', 'Matteo Brucato']
|
2021-03-11
| null | null | null | null |
['decision-making-under-uncertainty', 'decision-making-under-uncertainty']
|
['medical', 'reasoning']
|
[-1.75503656e-01 1.58824712e-01 -3.96524407e-02 -1.05279160e+00
-1.47484660e+00 -8.50335538e-01 6.69079646e-02 4.67133552e-01
-1.37022678e-02 8.90578032e-01 1.16715722e-01 -6.38434589e-01
-3.26437354e-01 -1.07209134e+00 -1.03780949e+00 -3.34174156e-01
-1.46876961e-01 1.23286510e+00 4.69979905e-02 1.48251757e-01
2.47417495e-01 4.71207619e-01 -1.72984791e+00 4.47353870e-01
9.05273616e-01 1.19550908e+00 -7.57373348e-02 5.37812233e-01
-4.46351945e-01 3.85567188e-01 -8.76070797e-01 -6.73959732e-01
4.11391973e-01 -2.55506746e-02 -4.50094372e-01 1.01439156e-01
1.77901253e-01 -4.45983082e-01 3.74126524e-01 9.66589153e-01
3.55677843e-01 2.44514152e-01 4.37465370e-01 -1.31250513e+00
6.98519424e-02 9.95319903e-01 -4.62860972e-01 -1.72378689e-01
5.72265506e-01 1.93747371e-01 1.00326133e+00 -7.48311937e-01
5.11386335e-01 1.47307229e+00 4.81036544e-01 -1.62386522e-01
-1.61072242e+00 -2.24159911e-01 2.85447836e-01 -4.26737666e-01
-1.61928141e+00 -5.04963040e-01 2.97219992e-01 -1.46467462e-01
1.20374203e+00 7.47419715e-01 3.53996456e-01 3.72297227e-01
3.27152789e-01 5.72483659e-01 5.87518096e-01 -3.50628123e-02
1.00542176e+00 5.41996956e-01 -2.09234860e-02 7.89279193e-02
5.99590361e-01 -5.50123081e-02 -6.65603995e-01 -9.23241735e-01
1.23942509e-01 -4.84563597e-02 7.16277361e-02 -5.10027766e-01
-8.79927635e-01 7.44958639e-01 -3.13788444e-01 -4.37662274e-01
-6.24496162e-01 2.84181923e-01 2.92173266e-01 2.79851347e-01
2.85679072e-01 3.58318210e-01 -8.28042448e-01 -3.14062774e-01
-1.07226527e+00 9.57006633e-01 1.57355058e+00 1.40605402e+00
5.12922645e-01 -1.47566304e-01 -2.42380053e-01 3.41362208e-01
4.64889914e-01 5.73624432e-01 -4.23077017e-01 -1.20710456e+00
9.89881992e-01 3.63678038e-01 8.02791774e-01 -8.13289165e-01
2.03562938e-02 -1.38564453e-01 -2.72485852e-01 2.00943753e-01
4.95520651e-01 -1.90786481e-01 -5.36846459e-01 1.43689406e+00
5.51467180e-01 -3.39057893e-01 1.71141714e-01 5.67623556e-01
1.04708157e-01 9.44869578e-01 -8.54906067e-03 -6.81966901e-01
1.08151233e+00 -2.11541429e-02 -7.30707765e-01 -2.05905929e-01
3.14661354e-01 -7.50124097e-01 8.41238678e-01 8.65299046e-01
-1.46023715e+00 4.55014920e-03 -1.16984093e+00 3.27212870e-01
-1.84686780e-01 -3.65100741e-01 8.80963087e-01 9.76658225e-01
-7.85809338e-01 6.48144782e-01 -8.97649348e-01 1.62801474e-01
1.42501369e-01 4.76226568e-01 1.17935136e-01 -2.20586672e-01
-8.29693258e-01 7.32696772e-01 4.29780573e-01 1.94102734e-01
-5.35610557e-01 -9.34803605e-01 -6.93425536e-01 3.00212324e-01
1.00279510e+00 -6.27832174e-01 1.26093531e+00 1.48961172e-02
-1.14531255e+00 4.45434809e-01 -2.32210308e-01 -3.01639169e-01
6.49388313e-01 -1.21361315e-01 -1.86230555e-01 -5.27068138e-01
3.24596427e-02 3.35069071e-03 3.28575611e-01 -1.27794206e+00
-5.63205540e-01 -6.86066806e-01 -2.31988933e-02 2.57856399e-01
2.55130887e-01 3.49030584e-01 -7.19307542e-01 -2.90926427e-01
3.89406830e-01 -7.05394983e-01 -8.21107924e-01 -2.70806849e-01
-7.68440843e-01 -6.50690198e-02 6.89334124e-02 -5.04450977e-01
1.58307695e+00 -1.92546463e+00 2.05358345e-04 8.36380899e-01
-2.59242296e-01 -5.86277723e-01 5.34788907e-01 6.25344872e-01
8.15545842e-02 4.28083390e-01 -4.18451518e-01 -2.24832952e-01
4.05939549e-01 4.22860801e-01 -6.37821436e-01 3.43071312e-01
6.08491823e-02 4.28652257e-01 -4.52290267e-01 -4.52013701e-01
-4.64046188e-02 -3.42117280e-01 -7.92933643e-01 2.79421717e-01
-1.09172034e+00 -2.86383599e-01 -5.84130585e-01 9.15349245e-01
1.01691592e+00 1.06047746e-02 5.69966674e-01 3.39500308e-02
1.02273850e-02 3.44565630e-01 -2.33310103e+00 1.43318021e+00
-3.69090170e-01 -3.02083701e-01 2.70488888e-01 -6.66822076e-01
8.84261489e-01 -1.45277828e-01 4.94020641e-01 1.89106643e-01
-2.12208301e-01 4.20620412e-01 -5.46645105e-01 -2.69372731e-01
7.64132321e-01 -3.44066322e-02 -6.69573843e-01 6.49495959e-01
-5.27117133e-01 -8.57951820e-01 2.46465206e-01 2.65045881e-01
1.18990362e+00 7.50791430e-02 3.04439574e-01 -1.92721024e-01
-1.02456689e-01 3.30307633e-01 9.82390165e-01 1.07891130e+00
3.30664635e-01 7.31985629e-01 9.98893917e-01 -3.45765740e-01
-7.73042500e-01 -1.25679398e+00 -3.52291197e-01 7.63190091e-01
-7.10410476e-02 -5.43582559e-01 -4.56241965e-01 -3.11983764e-01
8.18303347e-01 1.38436663e+00 -8.83248895e-02 2.44709045e-01
-4.10166115e-01 -1.18320513e+00 -1.06784314e-01 4.71206725e-01
-6.24881089e-02 -5.03860831e-01 -4.90524411e-01 7.44008482e-01
1.30147457e-01 -8.66727471e-01 -4.51359391e-01 4.05770481e-01
-9.57402170e-01 -8.66035342e-01 1.95190057e-01 1.04207478e-01
5.83926320e-01 -3.21952015e-01 1.56876945e+00 -3.47406954e-01
-3.37421745e-01 1.17594130e-01 3.23292315e-01 -6.62584364e-01
-2.68522859e-01 -6.04200065e-01 5.91227897e-02 -2.49405220e-01
3.90256524e-01 -4.70942020e-01 -1.32994941e-02 4.17726129e-01
-8.66199493e-01 -5.59604883e-01 1.36195451e-01 6.63580835e-01
1.17777085e+00 6.54642880e-01 2.07896799e-01 -1.18933964e+00
6.92146778e-01 -6.76343203e-01 -1.25839758e+00 7.32829273e-01
-7.71898031e-01 4.16245311e-01 2.45167017e-01 -9.41716880e-02
-1.14735830e+00 3.70358020e-01 2.48415932e-01 -1.77297667e-01
1.37717247e-01 8.46408486e-01 -7.04472005e-01 6.11766458e-01
6.20231152e-01 -1.17617838e-01 -1.02252543e-01 -2.98701286e-01
4.99991477e-01 6.16725564e-01 5.51661193e-01 -1.25590038e+00
3.48650366e-01 2.25283250e-01 -5.36692627e-02 -1.33302063e-01
-6.26359940e-01 -1.58055738e-01 -7.64371529e-02 1.80970952e-01
2.97993124e-01 -8.44843864e-01 -9.24469769e-01 -2.62891855e-02
-1.06714940e+00 5.00806570e-02 -5.71105659e-01 2.88648725e-01
-7.78330684e-01 -1.05943725e-01 -5.87398335e-02 -1.38439929e+00
7.00861886e-02 -1.37338686e+00 1.05445504e+00 -3.27896364e-02
-3.93199623e-01 -4.62035567e-01 -1.34776816e-01 1.85349599e-01
3.03983122e-01 5.54744720e-01 1.02060366e+00 -8.85997117e-01
-1.27598524e+00 -5.16575456e-01 1.55570224e-01 -4.37191315e-02
-2.95526832e-01 4.95167643e-01 -3.91274512e-01 -1.32198697e-02
2.96744347e-01 -1.25190422e-01 3.43620121e-01 6.29831195e-01
1.41837096e+00 -4.93200451e-01 -5.31133950e-01 5.59873223e-01
1.59041810e+00 4.14714038e-01 3.69810492e-01 -2.65315063e-02
-6.47177771e-02 8.04221272e-01 1.21253300e+00 1.33681059e+00
5.10514736e-01 7.48105228e-01 3.04283053e-01 6.25338614e-01
9.14589465e-01 -1.74041718e-01 6.94018155e-02 -5.29740937e-02
4.88710642e-01 -3.74441743e-01 -9.99613047e-01 4.87931520e-01
-2.11527491e+00 -8.53444695e-01 1.06255412e-01 2.65408254e+00
1.11468279e+00 4.90468025e-01 1.99389085e-02 -2.26137295e-01
5.24864018e-01 -6.22508191e-02 -1.07949769e+00 -6.68700695e-01
3.23126167e-02 -1.01965882e-01 7.26945281e-01 6.09572709e-01
-7.38043725e-01 3.50042939e-01 6.66067219e+00 6.11299336e-01
-3.27355444e-01 -4.35218602e-01 1.03608441e+00 -6.17000937e-01
-1.07267964e+00 1.32024094e-01 -1.24550819e+00 5.80259621e-01
1.34651768e+00 -7.15752423e-01 6.40133560e-01 1.26956975e+00
2.86942393e-01 -6.64645255e-01 -1.60223043e+00 9.44355428e-01
-3.70196939e-01 -1.62129104e+00 -2.86197633e-01 1.93594500e-01
7.38705039e-01 -3.92174661e-01 -5.31907491e-02 9.44003686e-02
1.13865983e+00 -9.70887959e-01 8.31848681e-01 6.85587168e-01
5.83635986e-01 -1.18894720e+00 7.58735001e-01 3.56563985e-01
-7.37721026e-01 -2.21762881e-01 -4.50374186e-01 3.59804630e-01
3.25991094e-01 1.32358909e+00 -7.71706104e-01 4.44786042e-01
9.73754346e-01 -9.29792784e-03 -5.02969623e-02 8.10891032e-01
3.37459147e-01 2.53551215e-01 -1.10217404e+00 -1.48187846e-01
-3.25581610e-01 -3.67999643e-01 3.28245401e-01 8.50160599e-01
4.02937919e-01 4.03199822e-01 3.30485165e-01 1.29776108e+00
2.10973158e-01 -5.84756508e-02 -6.87948465e-01 -1.67865865e-02
1.10026920e+00 6.41645372e-01 -4.70338255e-01 -2.99600810e-01
-1.06115744e-01 2.25509182e-01 8.16130489e-02 4.42830831e-01
-6.91555679e-01 -4.30514067e-01 9.19169724e-01 3.36465053e-02
4.05969769e-01 -1.97851911e-01 -1.04907167e+00 -8.88256490e-01
6.31837785e-01 -1.17120528e+00 6.81656897e-01 -5.14247239e-01
-1.12570274e+00 3.41161728e-01 3.94432247e-01 -8.63673747e-01
-9.06955123e-01 -2.36929432e-01 -4.29790646e-01 1.30004966e+00
-5.33777416e-01 -1.97770298e-01 1.86381355e-01 2.13305578e-01
1.69732317e-01 -1.12174973e-01 7.19258249e-01 -2.41159379e-01
-5.34255564e-01 5.17251015e-01 4.56280142e-01 -8.12584341e-01
5.54274082e-01 -1.52309334e+00 4.69369859e-01 7.37115800e-01
-3.62868339e-01 1.02975905e+00 1.27301812e+00 -1.02610195e+00
-1.94871140e+00 -7.58737206e-01 8.03079844e-01 -6.56139553e-01
5.74024618e-01 -4.75107610e-01 -7.78819561e-01 5.80748558e-01
-3.47294629e-01 1.08941384e-02 6.44256473e-01 2.82272905e-01
-7.54501000e-02 -6.82677865e-01 -1.82423985e+00 3.66361618e-01
7.00821638e-01 -1.71301559e-01 -2.31822044e-01 5.86393952e-01
8.63146782e-01 -8.13022375e-01 -1.24148619e+00 2.96380758e-01
5.31197786e-01 -9.22693014e-01 9.93068874e-01 -6.90004289e-01
2.14877963e-01 -5.39935112e-01 -7.87245929e-01 -1.15792465e+00
4.30903494e-01 -9.61327732e-01 -2.46381566e-01 1.29886663e+00
6.86811090e-01 -7.64723480e-01 9.45308506e-01 1.87628663e+00
1.38574630e-01 -9.92430210e-01 -9.34409082e-01 -6.45122945e-01
-4.10178363e-01 -8.35661232e-01 1.31018817e+00 3.92172515e-01
-1.92044064e-01 -3.77880365e-01 -8.89105722e-03 6.06990933e-01
1.02059245e+00 6.51930988e-01 8.03895831e-01 -8.77960801e-01
-6.88766062e-01 -2.26089135e-02 -8.17432702e-02 -7.62285531e-01
-2.70920545e-01 -3.07360381e-01 3.40664238e-01 -1.26049411e+00
-1.10737137e-01 -9.99449193e-01 7.48262927e-02 1.28913030e-01
-2.09468275e-01 -8.97729516e-01 -9.76743102e-02 -1.23794630e-01
-6.32644594e-01 2.45060220e-01 3.92750412e-01 -9.05880425e-03
-5.63829422e-01 6.52849436e-01 -1.14513707e+00 3.22320044e-01
4.09655929e-01 -6.54892981e-01 -4.91705954e-01 -1.95551202e-01
6.15463078e-01 1.00562859e+00 -2.44360745e-01 -6.13924980e-01
4.52012390e-01 -6.77779496e-01 3.32642406e-01 -1.33279109e+00
2.19739273e-01 -8.05739105e-01 9.44924355e-01 1.04895994e-01
-4.62748975e-01 3.49440366e-01 1.96934521e-01 8.24975312e-01
-3.65094721e-01 -4.71858263e-01 2.55228311e-01 -2.35427186e-01
-1.22360028e-01 2.18777269e-01 -2.25113496e-01 1.38317063e-01
1.07834542e+00 5.04475422e-02 -6.94904402e-02 -4.37729239e-01
-8.91097426e-01 8.19898009e-01 5.85352540e-01 -8.67262408e-02
4.83834863e-01 -1.07514095e+00 -6.39291644e-01 -1.50617305e-02
-1.53807569e-02 6.66975439e-01 -7.84513541e-03 1.17596880e-01
-5.49288750e-01 3.65788341e-01 4.30556774e-01 -6.11001372e-01
-8.55923593e-01 5.30289590e-01 2.28721261e-01 -4.17406827e-01
-6.02463447e-03 1.06262326e+00 -3.78956079e-01 -5.17059863e-01
5.81644535e-01 -5.37883937e-01 5.90120971e-01 7.03693703e-02
5.47996104e-01 3.49561810e-01 4.01288569e-01 4.95586336e-01
-5.00925958e-01 -8.62757191e-02 -5.81241734e-02 -4.42923576e-01
1.56866431e+00 -1.25044152e-01 -4.68209863e-01 5.37411749e-01
6.73241675e-01 2.09609643e-01 -1.16592741e+00 -2.70584047e-01
3.98004800e-01 -7.98992455e-01 -2.48137355e-01 -1.01178980e+00
-7.24348485e-01 2.95745403e-01 1.09974027e-01 2.38269284e-01
1.05086291e+00 -1.56648774e-02 1.96036816e-01 7.17368484e-01
8.04344475e-01 -1.15875256e+00 -6.13202691e-01 -7.85422921e-02
1.14780533e+00 -1.16266143e+00 5.06223679e-01 -5.09550154e-01
-8.17914009e-01 1.05165458e+00 5.34241617e-01 2.07298711e-01
9.19513881e-01 9.08748925e-01 -3.15845042e-01 -1.01986401e-01
-1.28874993e+00 3.95713657e-01 -2.32029811e-01 2.47801423e-01
-7.51069980e-03 3.85309786e-01 2.10608542e-02 1.11925507e+00
-4.55885321e-01 -8.90748128e-02 5.94604731e-01 1.23729193e+00
-3.54112685e-01 -1.12220705e+00 -8.23855221e-01 9.41071749e-01
-4.80493128e-01 -1.15067204e-02 1.28674820e-01 4.03531849e-01
-1.85152501e-01 1.12282217e+00 1.18969560e-01 -1.41284004e-01
7.23601937e-01 -4.12176503e-03 1.24091655e-01 -8.98277164e-01
-4.77585346e-01 3.82520370e-02 6.86903954e-01 -9.18429971e-01
4.46458936e-01 -1.25244284e+00 -1.19060397e+00 -4.44737285e-01
-1.60250187e-01 3.02142024e-01 1.09817743e+00 5.86691856e-01
3.90760958e-01 2.10733399e-01 7.33008206e-01 -4.15121436e-01
-1.53547060e+00 -3.17647934e-01 -9.42890227e-01 -2.02241063e-01
-1.29881591e-01 -4.18624550e-01 -2.24456534e-01 -1.89882219e-01]
|
[5.73463773727417, 3.6269540786743164]
|
84c53a12-8307-4439-ac09-d05d84c748bb
|
how-does-it-detect-a-malicious-app-explaining
|
2111.05108
| null |
https://arxiv.org/abs/2111.05108v1
|
https://arxiv.org/pdf/2111.05108v1.pdf
|
"How Does It Detect A Malicious App?" Explaining the Predictions of AI-based Android Malware Detector
|
AI methods have been proven to yield impressive performance on Android malware detection. However, most AI-based methods make predictions of suspicious samples in a black-box manner without transparency on models' inference. The expectation on models' explainability and transparency by cyber security and AI practitioners to assure the trustworthiness increases. In this article, we present a novel model-agnostic explanation method for AI models applied for Android malware detection. Our proposed method identifies and quantifies the data features relevance to the predictions by two steps: i) data perturbation that generates the synthetic data by manipulating features' values; and ii) optimization of features attribution values to seek significant changes of prediction scores on the perturbed data with minimal feature values changes. The proposed method is validated by three experiments. We firstly demonstrate that our proposed model explanation method can aid in discovering how AI models are evaded by adversarial samples quantitatively. In the following experiments, we compare the explainability and fidelity of our proposed method with state-of-the-arts, respectively.
|
['Vrizlynn L. L. Thing', 'Zhi Lu']
|
2021-11-06
| null | null | null | null |
['android-malware-detection']
|
['miscellaneous']
|
[ 5.89878440e-01 7.20282137e-01 -2.46549487e-01 -1.92952976e-01
-2.84466952e-01 -6.80235505e-01 9.93719637e-01 -2.69095600e-01
3.75702202e-01 6.28691018e-01 -1.09858088e-01 -4.49248761e-01
-4.97438349e-02 -5.31802535e-01 -9.68160331e-01 -3.90882164e-01
-2.52013624e-01 1.81139991e-01 5.14624314e-03 -2.13158906e-01
6.37928724e-01 2.60152549e-01 -1.51694071e+00 5.23429155e-01
1.12663054e+00 1.13082695e+00 -4.86908734e-01 7.99345016e-01
4.25319105e-01 8.61014605e-01 -7.31732905e-01 -7.70970166e-01
6.24984145e-01 -2.73877442e-01 -5.03392339e-01 -1.08993031e-01
8.72534588e-02 -3.32841188e-01 -1.41629934e-01 1.35723281e+00
-1.37133926e-01 -4.97644752e-01 7.54554689e-01 -1.92462516e+00
-1.03263390e+00 5.09067416e-01 -3.48926663e-01 5.25090424e-03
4.24041539e-01 7.26421356e-01 6.49587810e-01 -5.08733332e-01
3.42347294e-01 1.23470056e+00 6.73705339e-01 9.54492331e-01
-1.03200579e+00 -9.65516329e-01 5.13451695e-02 2.53277451e-01
-9.84149694e-01 -4.92121667e-01 9.38538313e-01 -5.20760655e-01
7.40237057e-01 7.72093475e-01 6.03075504e-01 1.40219879e+00
6.37820423e-01 4.77014452e-01 1.28552496e+00 -1.00544699e-01
4.48991030e-01 8.17758560e-01 1.55724943e-01 8.10686707e-01
7.68846929e-01 5.53519607e-01 -4.30109620e-01 -6.75976396e-01
-2.02350989e-02 1.33764282e-01 -7.44018098e-03 -8.65380168e-02
-7.67035663e-01 9.88625705e-01 4.08390760e-01 -1.47969142e-01
-5.13533831e-01 2.17187464e-01 4.03952777e-01 1.62258625e-01
3.39391410e-01 8.66807222e-01 -5.79390287e-01 -5.55628240e-02
-6.31627202e-01 1.87797442e-01 6.90352857e-01 8.76703203e-01
4.28546429e-01 3.59342813e-01 4.17091884e-03 -2.12022766e-01
5.92713118e-01 6.50971770e-01 7.05412328e-01 -7.36632884e-01
1.59818068e-01 8.71290982e-01 2.08495051e-01 -1.52670658e+00
1.73650756e-01 -2.73627251e-01 -5.28667569e-01 2.93174416e-01
6.10992908e-02 -2.29818225e-01 -6.68904543e-01 1.58439136e+00
3.34495872e-01 4.34432715e-01 3.51783723e-01 5.89870036e-01
4.67328340e-01 5.55145979e-01 6.53362721e-02 -3.85125607e-01
1.18785369e+00 -7.40253031e-01 -6.70731723e-01 -2.97441989e-01
5.37513196e-01 -4.24447656e-01 1.34499300e+00 4.95385021e-01
-7.25228727e-01 -4.23584104e-01 -1.04245842e+00 5.40149093e-01
-3.20565760e-01 2.36577261e-02 6.06196582e-01 1.11981213e+00
-5.86741030e-01 6.41591370e-01 -6.94488823e-01 -1.48487270e-01
8.30656767e-01 7.65486777e-01 -6.42936230e-02 6.33674800e-01
-1.09194291e+00 6.32040560e-01 1.63620442e-01 -7.98066109e-02
-1.34638345e+00 -6.66685224e-01 -4.95717168e-01 1.62875012e-01
3.38992804e-01 -4.07916546e-01 9.81090367e-01 -1.26759851e+00
-1.43339217e+00 4.77142185e-01 -8.23968723e-02 -9.90961313e-01
5.16308427e-01 -2.73795813e-01 -4.21121150e-01 -2.47395471e-01
-2.03529522e-01 5.01660705e-01 1.29954970e+00 -1.73482072e+00
-2.55261600e-01 -3.69049102e-01 -1.50318332e-02 -3.81600887e-01
-4.63944256e-01 -1.28077656e-01 4.17917848e-01 -3.83475631e-01
-3.81789118e-01 -1.19379318e+00 -2.12569430e-01 -6.09549209e-02
-9.19446766e-01 2.72436649e-01 1.28481114e+00 -7.50076711e-01
1.19223046e+00 -1.96008313e+00 -3.35856974e-01 4.22083199e-01
4.64386791e-01 6.70602202e-01 5.34141734e-02 2.32847050e-01
1.14400350e-01 7.75446594e-01 -3.44242454e-01 -3.55275124e-02
6.45042136e-02 -5.02467565e-02 -8.60228062e-01 2.63748974e-01
4.41269994e-01 9.61537004e-01 -6.79379523e-01 -3.71483445e-01
-6.31565973e-02 5.48790455e-01 -8.25897515e-01 2.70809382e-01
-4.85441864e-01 4.40740079e-01 -7.49765694e-01 6.81953967e-01
5.82108557e-01 -9.19693187e-02 8.24891031e-02 1.08616296e-02
3.38989288e-01 1.22314900e-01 -5.22450566e-01 5.88816166e-01
-9.11755785e-02 6.78118944e-01 -4.24210101e-01 -5.33487737e-01
1.07743943e+00 9.99230146e-02 1.60374463e-01 -1.81309924e-01
3.95237386e-01 1.12811523e-02 5.15433788e-01 -4.76884127e-01
3.09993386e-01 3.22201729e-01 -2.61501037e-02 6.82601988e-01
-6.20762885e-01 -1.62402213e-01 -5.23983836e-01 6.95383549e-02
1.12140393e+00 -1.21913612e-01 6.21926606e-01 -2.40757197e-01
7.71376252e-01 2.26565957e-01 4.44879323e-01 7.74254560e-01
-5.52354515e-01 -1.51707157e-01 7.62398481e-01 -5.97614110e-01
-9.42085087e-01 -7.42460668e-01 2.16126680e-01 4.53847796e-01
1.94925413e-01 -3.81596118e-01 -1.36242473e+00 -1.32380939e+00
5.47374487e-02 1.17226410e+00 -1.23284602e+00 -7.42034495e-01
-2.12097242e-01 -6.94648921e-01 6.85178459e-01 -8.60752631e-03
5.26958108e-01 -1.19702387e+00 -8.63928795e-01 -4.79540408e-01
2.98015326e-01 -8.74840736e-01 -2.15925649e-01 -2.54128426e-01
-7.02274382e-01 -1.28632236e+00 3.74627709e-01 -9.14996713e-02
1.05017745e+00 6.00608252e-02 6.15862966e-01 5.38200617e-01
3.87365073e-02 1.45007759e-01 -3.66668701e-01 -1.10335600e+00
-1.05189252e+00 -7.35565200e-02 4.55923319e-01 1.66848004e-01
5.35427570e-01 -2.31450260e-01 -5.88277161e-01 4.05587912e-01
-8.93015802e-01 -3.09725502e-03 6.99146569e-01 6.45978451e-01
3.41824800e-01 1.14312299e-01 5.61072648e-01 -1.33100462e+00
1.16631246e+00 -6.86582983e-01 -4.68783319e-01 2.77778208e-01
-1.16255939e+00 1.75419614e-01 1.06539011e+00 -7.55588293e-01
-7.91753829e-01 1.67439982e-01 3.52721125e-01 -3.58015716e-01
-1.60219148e-01 8.96861777e-02 -2.32832119e-01 -4.27083433e-01
1.15485561e+00 3.05927932e-01 3.73665929e-01 6.27284274e-02
3.59574765e-01 8.13526869e-01 2.30559602e-01 -2.67342031e-01
1.27616775e+00 3.78291965e-01 9.91502330e-02 -3.65011811e-01
-4.92227584e-01 5.41676939e-01 -3.44462484e-01 -1.49874076e-01
6.28871858e-01 -3.56981009e-01 -9.33324099e-01 9.57749262e-02
-1.13629758e+00 -9.11465585e-02 -2.23584205e-01 -5.02387621e-02
-4.57731307e-01 2.05821216e-01 -1.82762370e-01 -1.02261567e+00
-6.73266768e-01 -1.53646684e+00 8.88716102e-01 1.80157438e-01
-5.01672447e-01 -7.42069602e-01 -4.69279811e-02 3.49815965e-01
4.01415348e-01 4.57240552e-01 8.91616583e-01 -1.35403907e+00
-4.12236243e-01 -6.24877334e-01 -1.42947227e-01 1.80336371e-01
2.92273194e-01 4.76744831e-01 -1.38649189e+00 -5.92048615e-02
2.97019541e-01 1.63210984e-02 2.49276578e-01 1.20727383e-01
1.30401409e+00 -1.30002987e+00 -5.14770567e-01 2.67651320e-01
8.84598732e-01 3.85224760e-01 6.73405886e-01 2.84577191e-01
6.39930665e-01 5.46382129e-01 8.16707253e-01 4.21235681e-01
-1.28759872e-02 5.72658718e-01 1.05196393e+00 1.97014898e-01
3.22622687e-01 -4.34685111e-01 6.08906806e-01 4.09392655e-01
1.31041661e-01 -2.27066070e-01 -7.56764770e-01 9.58809182e-02
-1.71629250e+00 -9.16348636e-01 -1.22887924e-01 2.13693380e+00
5.60007691e-01 4.57280606e-01 1.98459327e-01 3.55745524e-01
5.71744204e-01 -1.34882510e-01 -8.58134747e-01 -9.82197165e-01
3.79910439e-01 -1.99493691e-01 4.06025589e-01 5.33187211e-01
-8.58607948e-01 9.96757269e-01 6.37325191e+00 5.77059388e-01
-1.06090820e+00 1.54326530e-02 9.24044847e-01 1.73256233e-01
-5.65135598e-01 1.47385389e-01 -4.77501482e-01 6.91955805e-01
1.14908004e+00 -4.66758549e-01 5.93243957e-01 1.27362144e+00
3.99550796e-01 4.54382837e-01 -1.09568584e+00 4.90987271e-01
-1.24248052e-02 -1.29218662e+00 4.73963231e-01 3.78374428e-01
8.66376877e-01 -6.05002880e-01 6.87087178e-01 1.22297600e-01
2.40625054e-01 -1.22185159e+00 6.49846673e-01 6.28694773e-01
4.26584631e-01 -7.02072382e-01 8.03022265e-01 4.03505504e-01
-4.98824418e-01 -5.33779860e-01 -2.67301500e-01 -2.72634596e-01
-5.83728611e-01 2.53285795e-01 -1.53486133e+00 -1.41093368e-02
2.17823565e-01 4.48386967e-01 -9.14428115e-01 3.97938490e-01
-3.79077435e-01 1.08498406e+00 9.02518928e-02 -5.35476446e-01
7.71599496e-03 1.28628433e-01 6.73886061e-01 8.45806420e-01
1.84763804e-01 -6.58157989e-02 -3.18312407e-01 1.31434774e+00
4.20562699e-02 -1.40333265e-01 -1.09080219e+00 -4.17530090e-01
7.13452458e-01 1.00328529e+00 -3.40916097e-01 -2.79590726e-01
1.12757742e-01 6.85367763e-01 -8.50424021e-02 8.13532770e-02
-1.10542381e+00 -3.07601038e-02 7.53135562e-01 3.66713285e-01
-1.59918293e-01 1.39554650e-01 -8.28001142e-01 -7.96292424e-01
-1.13694981e-01 -1.42624569e+00 -1.32093271e-02 -6.78927839e-01
-1.06933904e+00 9.61352110e-01 -1.53574375e-02 -1.38548243e+00
-6.33418262e-01 -5.53851366e-01 -9.47755218e-01 3.36839586e-01
-9.88234222e-01 -1.30401814e+00 -3.66383255e-01 3.65790546e-01
5.20357966e-01 -6.99692786e-01 8.20796072e-01 -4.41099733e-01
-6.37704253e-01 8.73315215e-01 -3.08768630e-01 -2.70833910e-01
1.37721539e-01 -8.36652815e-01 8.64447773e-01 1.04734778e+00
8.19214731e-02 7.98818886e-01 1.20610535e+00 -1.10754025e+00
-1.48900771e+00 -1.16213012e+00 4.08734798e-01 -9.19581771e-01
7.34604597e-01 -3.68333131e-01 -8.19876313e-01 8.09396148e-01
4.13765311e-02 -3.51478346e-02 5.68183839e-01 -4.24284935e-01
-4.81026262e-01 2.16796100e-02 -1.74737227e+00 7.63528168e-01
6.30456269e-01 -2.85360754e-01 -1.69703037e-01 4.34654295e-01
9.41810846e-01 1.78659540e-02 -4.68595952e-01 3.95006269e-01
5.94069004e-01 -9.10504282e-01 7.90500641e-01 -1.13661087e+00
5.83817661e-01 -2.62626529e-01 -2.07553327e-01 -1.13087022e+00
5.21275587e-03 -1.05642176e+00 -5.42953551e-01 9.33166087e-01
6.87103331e-01 -7.61320233e-01 8.86775970e-01 9.42052782e-01
1.41704991e-01 -1.02186704e+00 -5.89457810e-01 -5.48733950e-01
-1.43696561e-01 -4.62212503e-01 9.91992235e-01 8.10163677e-01
4.83392412e-03 -2.21018627e-01 -5.81005394e-01 5.12450039e-01
7.29172051e-01 -3.37125599e-01 1.21204400e+00 -1.10330188e+00
-2.86381602e-01 -2.92978644e-01 -7.11937428e-01 -1.70105278e-01
1.80549875e-01 -4.73261178e-01 -1.22912899e-01 -6.52723193e-01
3.97910506e-01 -1.55799344e-01 5.49095869e-02 4.61625755e-01
-3.45336914e-01 1.66062102e-01 3.38432670e-01 4.64086324e-01
-3.08032304e-01 3.70668799e-01 9.06917393e-01 -1.25107512e-01
-3.64205927e-01 1.04910575e-01 -1.09291840e+00 9.58468080e-01
1.14606571e+00 -8.84444535e-01 -7.59944439e-01 4.30902429e-02
6.70387298e-02 -3.75587463e-01 6.92476749e-01 -1.02669394e+00
-2.12231487e-01 -4.76676285e-01 2.30002135e-01 1.91625342e-01
2.64528841e-01 -1.01320696e+00 2.49292463e-01 9.34130251e-01
-5.71789384e-01 3.43855545e-02 2.87515402e-01 7.45009780e-01
3.49793494e-01 -1.15994595e-01 6.54930055e-01 2.13591337e-01
-1.17970370e-01 1.67356625e-01 -3.33885819e-01 -2.58775234e-01
1.42546594e+00 -3.11094940e-01 -6.28813386e-01 -6.87197506e-01
-3.84840280e-01 -2.82503873e-01 5.26785791e-01 4.48226601e-01
7.41154194e-01 -1.13537729e+00 -5.59263885e-01 4.83719915e-01
-1.63261387e-02 -7.45375276e-01 -1.64613083e-01 6.18209362e-01
-3.54724854e-01 4.53864902e-01 -3.82057160e-01 -4.84299004e-01
-1.50638700e+00 9.34447229e-01 2.71465629e-01 -2.97443151e-01
-6.25869110e-02 5.07565022e-01 2.47274742e-01 -3.71987015e-01
-3.90483513e-02 -1.14225566e-01 -1.13838978e-01 -8.79974723e-01
4.69759911e-01 2.96142429e-01 -4.58541125e-01 -7.80861735e-01
-3.87520999e-01 2.27002084e-01 -2.15422019e-01 1.41263649e-01
1.03145480e+00 1.67868957e-01 2.52175033e-01 -6.12834841e-03
7.13919580e-01 2.84558713e-01 -1.27986014e+00 3.79560292e-01
-9.02032554e-02 -7.29459167e-01 -3.29731971e-01 -1.08665192e+00
-8.90377522e-01 8.43468249e-01 7.76045322e-01 7.45120168e-01
1.02175009e+00 -3.74168605e-01 6.65426791e-01 2.88253665e-01
1.07712127e-01 -4.31075007e-01 1.49321526e-01 -2.59288073e-01
1.02812004e+00 -1.54139698e+00 1.35190338e-01 -5.61494827e-01
-1.08506727e+00 8.15844655e-01 8.27987075e-01 -2.07168803e-01
5.83820462e-01 -5.02724899e-03 -1.63652822e-01 -3.26139748e-01
-9.75769341e-01 5.12034893e-01 6.73407137e-01 1.01829231e+00
-7.64252916e-02 2.35764995e-01 -1.16202898e-01 1.10532403e+00
-5.31430542e-01 -1.63113609e-01 6.93601608e-01 5.25377393e-01
-4.38940585e-01 -6.96922064e-01 -5.00978470e-01 5.89844108e-01
-3.97926331e-01 1.49691687e-03 -1.32180834e+00 9.49846745e-01
1.25895545e-01 1.23483801e+00 -4.15751398e-01 -1.27156389e+00
8.67508575e-02 -1.64873093e-01 -1.39060318e-01 -1.85213074e-01
-8.09835076e-01 -6.10883057e-01 4.97632474e-03 -5.58864713e-01
1.39144659e-01 -6.48225486e-01 -1.21351552e+00 -5.36677003e-01
-3.81930649e-01 2.60232419e-01 9.33953881e-01 1.02504349e+00
9.23961818e-01 3.35560650e-01 9.41565156e-01 -5.08209586e-01
-7.29391098e-01 -9.14306939e-01 8.08303207e-02 5.14025867e-01
2.45810792e-01 -5.93645036e-01 -7.03120470e-01 1.06207468e-01]
|
[14.32205581665039, 9.63951587677002]
|
26e10d25-83e9-45f4-b2eb-f956318244f1
|
latent-dirichlet-allocation
| null | null |
https://dl.acm.org/doi/10.5555/944919.944937#d7400906e1
|
https://www.jmlr.org/papers/volume3/blei03a/blei03a.pdf
|
Latent Dirichlet Allocation
|
We describe latent Dirichlet allocation (LDA), a generative probabilistic model for collections of
discrete data such as text corpora. LDA is a three-level hierarchical Bayesian model, in which each
item of a collection is modeled as a finite mixture over an underlying set of topics. Each topic is, in
turn, modeled as an infinite mixture over an underlying set of topic probabilities. In the context of
text modeling, the topic probabilities provide an explicit representation of a document. We present
efficient approximate inference techniques based on variational methods and an EM algorithm for
empirical Bayes parameter estimation. We report results in document modeling, text classification,
and collaborative filtering, comparing to a mixture of unigrams model and the probabilistic LSI
model.
|
['Michael I. Jordan', 'Andrew Y. Ng', 'David M. Blei']
|
2003-01-01
| null | null | null | null |
['text-categorization']
|
['natural-language-processing']
|
[-2.22093254e-01 1.57873750e-01 -3.62712145e-01 -4.85418946e-01
-8.20491433e-01 -4.48432833e-01 1.20698082e+00 1.38445109e-01
-6.67977259e-02 5.74205279e-01 6.04583323e-01 -8.11698735e-02
-9.37932432e-02 -1.09276378e+00 -2.72778600e-01 -6.90747380e-01
-2.96553094e-02 1.37073529e+00 1.06018916e-01 2.92366683e-01
2.61029065e-01 1.25562614e-02 -1.35895789e+00 1.36717901e-01
5.62974155e-01 4.47723597e-01 2.75201470e-01 7.49912858e-01
-8.65963042e-01 4.32673782e-01 -6.99803472e-01 -6.11234367e-01
-3.95352095e-01 -1.27761289e-01 -7.26917624e-01 3.63088191e-01
6.72944933e-02 -5.07306576e-01 -1.83113739e-01 8.91268969e-01
-2.02689636e-02 3.82878393e-01 1.65093565e+00 -1.49510777e+00
-4.39038992e-01 9.22677934e-01 -5.39291620e-01 -2.45409980e-02
3.47198308e-01 -5.36822438e-01 1.25402617e+00 -1.00098252e+00
5.99273205e-01 1.81454062e+00 3.44723314e-01 1.09317645e-01
-1.69338834e+00 -2.95445561e-01 3.25988919e-01 -3.59241873e-01
-1.49077117e+00 -1.07545711e-01 6.20573759e-01 -9.01017427e-01
7.44474530e-01 -1.97320804e-01 7.13182449e-01 1.10645568e+00
4.97990340e-01 1.28754520e+00 5.87313175e-01 -4.79104161e-01
6.55647695e-01 3.81076396e-01 7.36646116e-01 2.13765487e-01
4.08288598e-01 -6.12448514e-01 -5.97587407e-01 -1.27771485e+00
7.10332394e-01 3.17566931e-01 1.63933203e-01 -6.22759163e-01
-7.38262534e-01 1.38533747e+00 -6.20433629e-01 2.19780326e-01
-6.58511758e-01 2.13585511e-01 1.14160590e-01 -3.56457472e-01
9.30853963e-01 -3.57897162e-01 -5.35337478e-02 -6.82241693e-02
-1.39418542e+00 4.93875355e-01 1.41735852e+00 1.02128005e+00
6.83116019e-01 -1.90010607e-01 -1.93335876e-01 9.59174037e-01
1.21315515e+00 6.33981824e-01 2.36314565e-01 -1.00138426e+00
-1.52632957e-02 1.27915725e-01 5.99227726e-01 -7.31643736e-01
1.15336269e-01 6.87174648e-02 -6.00198686e-01 -1.66545063e-01
2.00401857e-01 8.04526284e-02 -7.26990044e-01 1.64954793e+00
3.27447981e-01 -9.36095137e-03 -1.96493655e-01 1.48862496e-01
4.56790179e-01 1.24573684e+00 5.64361036e-01 -6.47226632e-01
1.40746856e+00 -4.03597116e-01 -1.15500808e+00 3.62838507e-01
1.84704557e-01 -8.28773081e-01 5.64810812e-01 6.84303224e-01
-1.35385716e+00 -1.84930995e-01 -4.24503952e-01 -1.36445329e-01
-3.58056605e-01 -8.66998211e-02 6.81728959e-01 9.23977494e-01
-1.17679942e+00 1.88251629e-01 -1.14658558e+00 -4.79585081e-01
1.29928321e-01 9.82261077e-02 2.23727569e-01 -1.48173347e-01
-1.05556333e+00 4.60645765e-01 2.32439712e-01 -5.60865045e-01
-1.14697325e+00 -6.20726168e-01 -6.24699533e-01 3.55582237e-01
-3.41481000e-01 -8.42000306e-01 1.38784444e+00 -1.12619892e-01
-1.40041840e+00 7.16658354e-01 -9.22486842e-01 -3.89731288e-01
1.97616935e-01 -2.27623552e-01 -2.41880819e-01 2.02025294e-01
1.59167707e-01 5.16937435e-01 9.98316705e-01 -1.58271575e+00
-6.48951709e-01 -3.23717475e-01 -2.43983179e-01 2.21081853e-01
-3.37358683e-01 2.14119166e-01 -4.93246168e-01 -6.27674878e-01
4.26770031e-01 -8.26321661e-01 -2.02373371e-01 -5.05989075e-01
-3.69556367e-01 -8.89272571e-01 7.85001218e-01 -7.19623327e-01
1.46130157e+00 -2.01235485e+00 1.17390215e-01 2.51258016e-01
3.56940806e-01 -7.79226303e-01 4.21778142e-01 8.00495625e-01
3.06025326e-01 3.54805321e-01 -1.80931404e-01 -8.58180940e-01
4.66286987e-01 3.40201110e-01 -7.78090477e-01 5.68903625e-01
-4.95853096e-01 1.92265168e-01 -7.58697629e-01 -7.26635695e-01
9.92251933e-02 5.99545002e-01 -6.26763701e-01 1.45880058e-01
-5.97899079e-01 -1.84624448e-01 -4.71978098e-01 1.30606160e-01
8.63259137e-01 -4.16237921e-01 4.45494384e-01 3.26048017e-01
-6.71246722e-02 4.41820115e-01 -1.41839731e+00 1.50694799e+00
-2.74478793e-01 5.67159712e-01 1.67673796e-01 -4.76908296e-01
7.84225225e-01 8.68661344e-01 8.06252599e-01 7.43980944e-01
-8.22275281e-02 -3.10187876e-01 -5.20918489e-01 -1.69093534e-02
9.87740874e-01 -1.23181731e-01 -3.67366135e-01 1.38376617e+00
3.89689058e-01 -3.34525883e-01 2.39795923e-01 9.04112279e-01
5.30691326e-01 -5.80424592e-02 1.65628821e-01 -5.92402101e-01
-2.45497629e-01 -1.33270649e-02 1.88703373e-01 1.21004295e+00
2.29034320e-01 2.12176397e-01 4.58315462e-01 -1.78876981e-01
-1.21934593e+00 -1.69950664e+00 -6.95986986e-01 1.47263825e+00
-1.93515390e-01 -7.86368012e-01 -6.99646831e-01 -5.01264855e-02
-4.35598940e-03 1.20453203e+00 -4.67256099e-01 3.80608529e-01
8.38247500e-03 -1.27108729e+00 9.80765447e-02 1.85919657e-01
9.66644958e-02 -5.08040309e-01 2.81222817e-02 4.01894033e-01
-3.77153635e-01 -5.95743716e-01 -2.23041356e-01 -7.09691942e-02
-1.14914536e+00 -4.97999012e-01 -8.27424467e-01 -4.47624892e-01
4.79698598e-01 1.47940174e-01 1.39472961e+00 -5.39886475e-01
7.32629746e-03 1.03638983e+00 -4.99207564e-02 -4.06796038e-01
-7.53692508e-01 -9.07786265e-02 3.75517994e-01 -8.71748999e-02
9.41337705e-01 -5.35748422e-01 -3.71031493e-01 1.76295370e-01
-1.00190353e+00 -2.38138124e-01 -5.95439598e-02 5.65386057e-01
3.71695757e-01 4.00393784e-01 -1.05947824e-02 -8.98503065e-01
9.62518156e-01 -1.11497450e+00 -4.56468403e-01 2.61081427e-01
-4.52244401e-01 -1.91756263e-01 -4.30947304e-01 -5.72285295e-01
-1.70211983e+00 -3.40375721e-01 2.65416771e-01 -1.48793578e-01
-5.87005019e-01 5.18831491e-01 -2.04867780e-01 8.71094704e-01
3.46154898e-01 3.14783156e-01 -2.69798815e-01 -7.37225711e-01
6.25104010e-01 1.04617250e+00 6.22122101e-02 -9.88672376e-01
1.48328468e-01 7.06504405e-01 -2.75875688e-01 -1.19021773e+00
-7.65287280e-01 -8.97669256e-01 -9.09012139e-01 -1.07797354e-01
1.04532874e+00 -1.06014633e+00 -4.64237422e-01 4.59907860e-01
-1.31600904e+00 -1.24875590e-01 -1.01053737e-01 8.00480604e-01
-5.94414532e-01 3.78191948e-01 -8.75235319e-01 -1.41358769e+00
-4.19053212e-02 -8.18130374e-01 1.13985264e+00 -2.86576282e-02
-5.88087201e-01 -1.59646547e+00 4.49771017e-01 1.83172554e-01
2.33401388e-01 -3.80282402e-01 1.09714079e+00 -9.25521016e-01
-3.88635010e-01 -2.15125829e-01 1.01792790e-01 -5.58857732e-02
7.42663816e-02 4.55790728e-01 -7.64634490e-01 -3.08731049e-01
2.29906648e-01 -2.30608806e-02 8.43361735e-01 8.97757113e-01
7.54139304e-01 -1.55164897e-01 -8.71212602e-01 -2.24304408e-01
1.22664690e+00 2.63042510e-01 3.70472699e-01 -1.56657696e-01
3.19936901e-01 4.56939906e-01 9.01088715e-02 9.78304565e-01
6.83585286e-01 3.82198036e-01 -1.69627026e-01 3.43651116e-01
5.31005561e-01 -2.68051356e-01 1.42692298e-01 8.10046554e-01
1.24506168e-01 -6.46316588e-01 -1.13226962e+00 7.97975361e-01
-1.65443444e+00 -1.24152100e+00 -4.80453491e-01 2.11759853e+00
9.09384906e-01 1.05282888e-02 3.46475691e-01 -2.17913449e-01
1.07891071e+00 -3.62609215e-02 -1.33729130e-01 -1.57139793e-01
7.07665309e-02 -9.35936421e-02 2.22006813e-03 7.05337703e-01
-1.05306840e+00 8.34222019e-01 8.25884342e+00 8.31875741e-01
-1.20110765e-01 4.04190868e-01 3.50678444e-01 -5.94155267e-02
-6.34072542e-01 1.92094594e-01 -1.38768888e+00 5.85509717e-01
1.11194825e+00 -5.09694576e-01 -3.26567553e-02 8.44868541e-01
2.04434186e-01 -5.39251387e-01 -1.05440795e+00 4.93180811e-01
1.17795337e-02 -1.01134574e+00 5.02817988e-01 7.17148423e-01
1.03660989e+00 2.16598362e-02 1.31692678e-01 2.50804871e-01
1.40268314e+00 -4.64029938e-01 5.39155543e-01 7.97779977e-01
1.31108165e-01 -6.51690364e-01 3.06201369e-01 6.14478827e-01
-7.53993630e-01 1.95717499e-01 -6.43232942e-01 1.09293371e-01
5.34875333e-01 9.73152041e-01 -6.79395974e-01 -5.47915809e-02
8.89656901e-01 2.77124971e-01 3.42236042e-01 9.96432424e-01
4.63946164e-02 1.23302317e+00 -6.41186774e-01 3.08587849e-02
1.59170568e-01 -5.47277391e-01 8.21058095e-01 1.36223698e+00
2.78643250e-01 3.28363866e-01 4.35588598e-01 1.04708731e+00
6.35795891e-02 1.31009504e-01 -4.62531835e-01 3.84813063e-02
7.81323671e-01 9.48760748e-01 -9.34526443e-01 -9.47440445e-01
-3.46677989e-01 5.74357569e-01 -8.25856850e-02 6.32649720e-01
-3.67158592e-01 2.45910317e-01 5.21790922e-01 7.61959329e-02
3.47625941e-01 -4.98461127e-01 -1.63515195e-01 -1.36999488e+00
-6.57937169e-01 -2.50175864e-01 6.74966097e-01 -8.22014272e-01
-1.68159974e+00 2.38213584e-01 7.71248698e-01 -7.48671293e-01
-5.69964170e-01 -2.06668928e-01 -7.62537181e-01 1.17037642e+00
-7.07716942e-01 -9.02267039e-01 1.68599084e-01 5.30761600e-01
6.08720541e-01 -1.90216199e-01 1.12956810e+00 -3.23103279e-01
-1.87616631e-01 -1.56388730e-01 8.39975476e-01 -1.90883473e-01
5.20471513e-01 -1.50723326e+00 3.91696006e-01 2.04701543e-01
5.25484122e-02 1.22637749e+00 1.00582433e+00 -9.50306296e-01
-8.05594087e-01 -7.28211939e-01 9.42740440e-01 -6.68718874e-01
7.67045438e-01 -5.42785823e-01 -1.06648815e+00 1.00770617e+00
4.21708941e-01 -8.42978597e-01 1.56257498e+00 6.57436430e-01
4.65297792e-03 7.33826995e-01 -1.17490673e+00 3.49506587e-01
3.46755892e-01 -4.15111959e-01 -8.84213030e-01 6.70613706e-01
5.23148417e-01 1.69803292e-01 -1.27577448e+00 -3.98024112e-01
5.91339409e-01 -6.40693247e-01 7.94826150e-01 -6.23865306e-01
2.23950103e-01 6.23325296e-02 -4.85271841e-01 -1.17708552e+00
-6.08350813e-01 -5.15046239e-01 -3.30206156e-01 1.50516129e+00
1.94754928e-01 -1.93725213e-01 7.77706981e-01 1.16867280e+00
4.45798695e-01 -7.92340413e-02 -7.06786454e-01 -4.94776219e-01
4.45886850e-01 -7.45102882e-01 5.22471786e-01 9.51708496e-01
2.40596771e-01 2.36710638e-01 -2.50072867e-01 8.65722969e-02
1.32706821e+00 2.03527644e-01 6.62313879e-01 -1.76980889e+00
-2.73537695e-01 -3.94968569e-01 1.76445786e-02 -1.34327292e+00
3.23457062e-01 -5.15760839e-01 3.80106904e-02 -1.85034394e+00
7.34088004e-01 -4.50737327e-01 2.47641906e-01 -1.04135409e-01
1.40956670e-01 -2.73122162e-01 -3.17902654e-01 8.41592908e-01
-6.45565331e-01 6.02715969e-01 4.68977690e-01 -8.62164274e-02
-2.64686167e-01 3.68432760e-01 -1.86151713e-01 1.04998219e+00
5.43298185e-01 -7.00724602e-01 -3.80650669e-01 -2.51785874e-01
3.31669927e-01 3.39200228e-01 -7.68351629e-02 -2.75671631e-01
3.89991820e-01 -1.19569674e-01 3.33197027e-01 -1.30345798e+00
5.96208513e-01 -4.69506174e-01 3.17930073e-01 2.52211448e-02
-4.45077598e-01 -4.66119051e-01 -5.00671268e-02 9.43910539e-01
-1.69361874e-01 -5.30878067e-01 4.93512541e-01 -2.89025486e-01
-2.42837295e-01 1.74653932e-01 -1.32691956e+00 -3.09971929e-01
6.02150261e-01 2.60036364e-02 1.32783160e-01 -8.25014889e-01
-1.44199312e+00 7.21671656e-02 4.07614946e-01 2.75669359e-02
2.27325171e-01 -1.27110922e+00 -7.85349548e-01 -2.34204948e-01
-2.32393667e-01 -8.97877514e-02 3.50073040e-01 3.20462465e-01
1.21523336e-01 4.95483071e-01 2.97198832e-01 -7.84803629e-01
-9.66680288e-01 4.28153306e-01 -2.18738467e-02 -4.10266161e-01
-2.32445806e-01 6.72880352e-01 5.53429186e-01 -1.44463256e-01
3.26262087e-01 1.31644174e-01 -3.52869123e-01 3.04573298e-01
6.26273632e-01 6.64225280e-01 -4.43180144e-01 -4.74392802e-01
-1.45909656e-02 9.60519910e-02 -2.51426011e-01 -7.72903979e-01
1.13172972e+00 -5.77679276e-01 -5.79591155e-01 1.10506606e+00
8.90614331e-01 -2.72667050e-01 -9.36484635e-01 -5.16949534e-01
3.91245494e-03 -2.46710017e-01 4.25582618e-01 -3.42819005e-01
-2.23335087e-01 7.98069954e-01 1.72717169e-01 8.17093730e-01
3.37861657e-01 4.60835248e-01 2.86469042e-01 2.11260259e-01
4.06221688e-01 -9.97796297e-01 -1.54159352e-01 3.93465757e-01
5.97117960e-01 -1.05456746e+00 1.17307931e-01 -3.89015675e-01
-4.24960315e-01 1.01995909e+00 -1.76705554e-01 -7.01832250e-02
1.48615873e+00 5.04043460e-01 -5.63292325e-01 -4.32411313e-01
-1.03463149e+00 1.83196977e-01 3.07173759e-01 4.65366215e-01
7.53167331e-01 1.17668808e-01 -1.52293295e-01 8.90438795e-01
-2.77129501e-01 -9.46057513e-02 6.04455948e-01 8.33438396e-01
-8.81483793e-01 -9.48937654e-01 -8.20094764e-01 6.47983253e-01
-7.13158488e-01 -1.64541975e-01 -9.73894522e-02 4.14853036e-01
-3.12100470e-01 1.20571494e+00 7.52932310e-01 3.75879526e-01
-4.38982844e-01 5.89182913e-01 2.53543317e-01 -9.76083696e-01
1.81221485e-01 7.81091750e-01 -1.26355514e-01 1.96710065e-01
-3.66810530e-01 -1.35341132e+00 -8.73085439e-01 -4.72382128e-01
-5.04468381e-01 7.07087040e-01 9.46966469e-01 8.52717459e-01
-2.80632917e-02 3.59876528e-02 3.77073646e-01 -8.92232120e-01
-4.93646979e-01 -1.38772285e+00 -1.22439301e+00 7.65802115e-02
-2.84118116e-01 -6.73519850e-01 -4.93462086e-01 6.79137647e-01]
|
[10.333670616149902, 6.902393817901611]
|
cd11aebe-6d64-4685-9142-0c58af1ed07e
|
complex-politics-a-quantitative-semantic-and
|
1510.03797
| null |
http://arxiv.org/abs/1510.03797v1
|
http://arxiv.org/pdf/1510.03797v1.pdf
|
Complex Politics: A Quantitative Semantic and Topological Analysis of UK House of Commons Debates
|
This study is a first, exploratory attempt to use quantitative semantics
techniques and topological analysis to analyze systemic patterns arising in a
complex political system. In particular, we use a rich data set covering all
speeches and debates in the UK House of Commons between 1975 and 2014. By the
use of dynamic topic modeling (DTM) and topological data analysis (TDA) we show
that both members and parties feature specific roles within the system,
consistent over time, and extract global patterns indicating levels of
political cohesion. Our results provide a wide array of novel hypotheses about
the complex dynamics of political systems, with valuable policy applications.
|
['Bahattin Tolga Oztan', 'María Pereda', 'Stefano Gurciullo', 'Slava Mikhaylov', 'Federico Battiston', 'Sebastian Poledna', 'Michael Smallegan', 'Alice Patania', 'Alexander Herzog', 'Peter John', 'Daniel Hedblom']
|
2015-10-13
| null | null | null | null |
['dynamic-topic-modeling']
|
['natural-language-processing']
|
[-1.90341979e-01 4.27472085e-01 -2.85383672e-01 -4.12537828e-02
-2.28890821e-01 -9.72862601e-01 1.58391941e+00 8.50200534e-01
-1.09294221e-01 4.26923186e-01 1.11070716e+00 -9.01490390e-01
-9.15520608e-01 -9.08041239e-01 -1.61045745e-01 -3.81777197e-01
-5.05726814e-01 6.85201943e-01 4.93505061e-01 -6.90090001e-01
3.60113084e-01 3.64299595e-01 -9.01426256e-01 2.78823465e-01
5.99071205e-01 5.86069763e-01 -1.16299435e-01 3.26744407e-01
-3.69225919e-01 9.05107081e-01 -5.83119452e-01 -2.85106897e-01
6.27587512e-02 -1.72802597e-01 -9.72461104e-01 -6.69719875e-02
1.68478042e-01 5.00598490e-01 -6.02013230e-01 8.10284913e-01
1.76386014e-01 1.40221909e-01 5.82728267e-01 -9.33028817e-01
-8.46418589e-02 7.48398304e-01 -5.34499824e-01 8.96115482e-01
4.88013357e-01 -1.69084281e-01 1.11684561e+00 -5.10593176e-01
1.49991608e+00 1.66395593e+00 5.97644329e-01 -5.41727781e-01
-1.33647442e+00 -2.26183861e-01 2.19716087e-01 -1.45297036e-01
-8.98292899e-01 -4.60042834e-01 8.35344613e-01 -8.92872214e-01
6.95957482e-01 3.88851911e-01 9.80564833e-01 7.86884844e-01
3.76653135e-01 9.87765044e-02 1.75483978e+00 -5.39721131e-01
1.37075171e-01 -1.91339195e-01 4.31555361e-01 4.70736504e-01
1.10123657e-01 -2.35161498e-01 -4.13079381e-01 -7.93157756e-01
5.71140289e-01 -1.68035299e-01 1.39869049e-01 -3.34580690e-01
-1.33268833e+00 9.87025201e-01 4.34048682e-01 8.53294373e-01
-5.00139952e-01 -2.87284004e-03 5.30847430e-01 5.43868661e-01
8.65957201e-01 4.16479439e-01 -2.76655167e-01 -5.61091602e-01
-9.71078098e-01 4.06790555e-01 1.00402963e+00 1.91229567e-01
3.60179842e-01 -6.29666805e-01 2.91790873e-01 5.20978034e-01
3.73662800e-01 3.68823946e-01 3.67803462e-02 -1.16102695e+00
6.47242904e-01 8.46667707e-01 -7.87530467e-02 -1.88935018e+00
-4.89633650e-01 -5.35050519e-02 -5.38856030e-01 -1.54122621e-01
4.52725023e-01 1.05161592e-01 -4.87803727e-01 1.72993672e+00
5.44588745e-01 -2.46304363e-01 -3.01550686e-01 2.43692324e-01
6.13418579e-01 7.39833355e-01 1.67299539e-01 -5.81837058e-01
1.44547832e+00 -1.34315327e-01 -8.06740344e-01 1.82697639e-01
4.65028793e-01 -6.96077049e-01 7.86935031e-01 -9.10882279e-02
-1.01734185e+00 1.35507807e-01 -4.65728581e-01 2.60732770e-01
-7.00024784e-01 -8.10916126e-01 6.70906901e-01 3.50766450e-01
-1.13143790e+00 3.95555854e-01 -8.16450894e-01 -9.09485877e-01
3.85560006e-01 -2.06807047e-01 -2.25455508e-01 4.25126284e-01
-1.31287825e+00 1.07803345e+00 -1.42354757e-01 -2.50548601e-01
-2.80451089e-01 -4.24796373e-01 -6.40273452e-01 -1.82056308e-01
4.00828362e-01 -2.85330623e-01 8.40207279e-01 -3.06366682e-01
-1.03937852e+00 9.59256291e-01 3.98983881e-02 -5.17873824e-01
5.75693429e-01 3.45866978e-01 -4.07366842e-01 1.26664817e-01
5.20497203e-01 -1.92509860e-01 8.19538608e-02 -1.06140876e+00
-3.56220931e-01 -4.40283507e-01 1.31433755e-01 1.27729803e-01
-1.78807542e-01 5.71833968e-01 4.97219265e-02 -7.56190419e-01
4.76472110e-01 -6.79391682e-01 -2.84490258e-01 -6.40808225e-01
-3.26925039e-01 -4.10060644e-01 7.70067394e-01 -8.50245416e-01
1.79535794e+00 -2.13242483e+00 5.49918234e-01 6.94422901e-01
5.31962514e-01 -4.38006312e-01 5.10572612e-01 1.16099012e+00
1.24161899e-01 6.16008759e-01 -2.45268896e-01 1.99207097e-01
1.21490151e-01 2.25842059e-01 -8.23930919e-01 8.17752481e-01
-4.97925162e-01 8.65518749e-01 -7.53028631e-01 -7.20278084e-01
2.84511685e-01 -1.29040033e-01 -3.11306655e-01 -8.20200384e-01
-1.56099647e-01 2.44611755e-01 -6.02950037e-01 5.80815375e-01
1.74317360e-01 -2.77695715e-01 9.92521584e-01 2.71520674e-01
-7.89874196e-01 6.18633032e-01 -8.58626306e-01 1.22584653e+00
-1.54847398e-01 1.35931253e+00 3.91561508e-01 -1.10193467e+00
4.60648596e-01 2.14802697e-01 6.53957963e-01 -7.80828655e-01
2.46992961e-01 -1.06505059e-01 2.29947716e-01 -8.57815892e-02
4.23379958e-01 -4.42155041e-02 -7.84861207e-01 8.37843359e-01
-1.27367690e-01 -2.15416610e-01 1.37930527e-01 7.57436454e-01
1.21080983e+00 -5.22892416e-01 5.59730291e-01 -1.18436396e+00
-2.29298677e-02 4.03894603e-01 4.37113494e-01 6.61284089e-01
-3.79141241e-01 -1.54374897e-01 9.96865809e-01 -7.87316620e-01
-1.18139887e+00 -9.73369837e-01 -5.27635396e-01 1.04974663e+00
1.21244617e-01 -9.05588269e-01 -3.91516775e-01 -9.56021026e-02
4.30818535e-02 3.95279765e-01 -9.49417830e-01 5.74744523e-01
-8.99501741e-01 -7.36834228e-01 1.63143635e-01 -2.31299430e-01
2.71657139e-01 -1.03516090e+00 -7.08124816e-01 2.00072885e-01
-4.16021287e-01 -8.58940840e-01 6.21141493e-02 -5.45831025e-02
-8.75935853e-01 -1.31918108e+00 -1.36838153e-01 -4.21114713e-01
2.69404680e-01 6.34100139e-02 1.13652122e+00 -2.69491822e-01
-3.85402173e-01 3.41634333e-01 4.63373698e-02 -2.42754489e-01
-5.15002608e-01 1.07915357e-01 3.35004359e-01 -4.60895956e-01
-1.51013494e-01 -7.64735043e-01 -2.06893489e-01 2.95459628e-01
-5.68239689e-01 -7.98841789e-02 -1.70709834e-01 4.52064633e-01
8.68735369e-03 4.82584506e-01 2.00726047e-01 -7.39147961e-01
1.26453626e+00 -8.93210530e-01 -5.05428493e-01 3.42392623e-01
-5.02870321e-01 -4.28652227e-01 -1.34988397e-01 -7.92725012e-02
-9.48839128e-01 -9.69620466e-01 7.30948150e-01 4.30641532e-01
2.29761109e-01 1.06315207e+00 2.43286178e-01 2.66697973e-01
7.44709373e-01 -1.47333249e-01 1.40739396e-01 -4.91477340e-01
7.59456038e-01 5.48628390e-01 2.07655236e-01 -8.39905500e-01
6.30433977e-01 1.03461277e+00 -1.36139438e-01 -1.19781339e+00
-2.82288820e-01 -3.73900145e-01 -7.11967826e-01 -6.93666458e-01
5.79018414e-01 -7.02090979e-01 -5.54689348e-01 1.92322746e-01
-7.44121790e-01 -4.79090273e-01 -4.30807680e-01 7.56815001e-02
-5.41972876e-01 3.13944876e-01 -6.22256219e-01 -7.41637766e-01
1.93398952e-01 -5.88671923e-01 5.65593600e-01 -6.99395835e-02
-9.42375600e-01 -1.36828697e+00 6.91227794e-01 2.30403975e-01
5.23300767e-01 9.23758745e-01 1.08482826e+00 -3.66358459e-01
-2.03936428e-01 2.27518499e-01 2.72734165e-02 -8.31812978e-01
4.38902885e-01 5.40314555e-01 -1.20054446e-01 -1.50571004e-01
7.26864189e-02 2.21821651e-01 5.93468726e-01 6.87413871e-01
1.51843652e-01 -7.98084617e-01 -8.00586641e-01 1.73798073e-02
1.14101398e+00 1.07032336e-01 2.51932234e-01 1.12311578e+00
5.18962517e-02 1.18293011e+00 4.31567073e-01 5.06741464e-01
5.96225619e-01 8.08542311e-01 -1.04784876e-01 1.50932461e-01
1.16731741e-01 -1.60288543e-01 -5.46969585e-02 9.41852152e-01
-1.00955762e-01 1.34264678e-01 -1.64659393e+00 9.08340096e-01
-2.13761163e+00 -1.30221212e+00 -3.73731494e-01 1.55381525e+00
7.47307539e-01 4.25282031e-01 4.94904846e-01 -1.90872565e-01
7.96988428e-01 7.37794459e-01 2.28284061e-01 -4.28850234e-01
-2.77730942e-01 7.44033605e-02 4.66910392e-01 7.12558270e-01
-1.00029063e+00 9.67130721e-01 8.06516171e+00 5.83902359e-01
-6.70113027e-01 1.35401800e-01 5.82327187e-01 -1.31902874e-01
-4.87522364e-01 4.50714886e-01 4.54040393e-02 3.77308935e-01
1.08148944e+00 -6.14345670e-01 1.52634189e-01 4.77507323e-01
5.40463388e-01 -3.44118506e-01 7.23203197e-02 2.11717859e-01
-4.90401685e-01 -1.81872761e+00 -3.40979129e-01 5.84845841e-01
8.79944384e-01 1.86036646e-01 -2.28066836e-02 -2.19363555e-01
1.03455830e+00 -7.48792827e-01 1.03050363e+00 5.89638352e-01
5.07033765e-01 -4.90634859e-01 1.45134717e-01 2.75906444e-01
-1.02683914e+00 -3.20164025e-01 1.71777546e-01 -4.89269882e-01
6.21394873e-01 7.24296868e-01 -4.03834641e-01 4.79212373e-01
8.65348637e-01 6.69588268e-01 -3.00672710e-01 4.19133335e-01
5.78068458e-02 7.27738023e-01 -5.77977717e-01 2.40602449e-01
4.66942430e-01 -4.38281715e-01 1.03177190e+00 1.05250776e+00
-1.16932429e-01 3.65223020e-01 2.54773702e-02 3.15162778e-01
2.63127446e-01 5.51132783e-02 -1.00747561e+00 -4.01057839e-01
7.27722406e-01 1.02789903e+00 -1.39622200e+00 -3.45093012e-01
-6.84013665e-02 7.66809508e-02 1.30919904e-01 1.77053794e-01
-4.45743978e-01 9.18411165e-02 7.56360948e-01 6.33242488e-01
-2.19730541e-01 -8.00475597e-01 -4.68106955e-01 -1.07521498e+00
-2.09996179e-01 -7.95475125e-01 5.62583208e-01 -1.55346438e-01
-1.07826066e+00 2.27252990e-01 4.76600647e-01 -5.60300589e-01
-6.21895045e-02 7.68073872e-02 -8.66947293e-01 4.55282539e-01
-4.67539668e-01 -9.32028592e-01 2.10499793e-01 3.55322808e-01
3.00459564e-01 -1.69919163e-01 4.52631503e-01 -2.37490460e-01
-4.74405378e-01 -6.09083891e-01 5.04375875e-01 -1.51358083e-01
1.82371348e-01 -1.11539018e+00 7.51545846e-01 5.23945510e-01
1.22148745e-01 9.96898711e-01 9.68537509e-01 -9.85364974e-01
-9.94236529e-01 -3.05082887e-01 1.23254514e+00 -9.00480986e-01
1.52770889e+00 -6.41985714e-01 -4.22426939e-01 6.72626853e-01
5.50656199e-01 -6.04647219e-01 7.17995405e-01 6.36850297e-01
-6.04650564e-02 5.29596984e-01 -1.00142813e+00 6.20825946e-01
1.06170928e+00 -6.80719793e-01 -1.19652128e+00 4.17243838e-01
6.23177707e-01 6.53984398e-02 -1.12083876e+00 2.31947869e-01
5.61196744e-01 -6.55288458e-01 1.01789951e+00 -6.26727343e-01
2.83257663e-01 -8.09182897e-02 -4.47167784e-01 -1.05535781e+00
-6.41919613e-01 -7.89249063e-01 9.48549584e-02 1.22649336e+00
2.01475084e-01 -8.09151530e-01 2.73084223e-01 4.46831942e-01
4.05896604e-01 -4.82642710e-01 -1.44215465e+00 -2.05579713e-01
2.17612311e-01 -1.87743142e-01 3.00626427e-01 1.56311011e+00
6.53136849e-01 -2.79343091e-02 1.81870610e-01 -2.60152876e-01
5.46622455e-01 4.12343293e-01 6.96079016e-01 -1.63128436e+00
2.84968704e-01 -9.35606420e-01 -5.49988925e-01 -1.92003518e-01
-5.54176755e-02 -6.06399715e-01 -6.26536667e-01 -1.75263071e+00
4.72895391e-02 -6.36325359e-01 1.69036523e-01 2.33053397e-02
5.72324574e-01 -2.92701036e-01 2.46179298e-01 1.00163257e+00
-4.18801427e-01 3.00140589e-01 5.72152436e-01 -8.82856920e-02
-4.05289710e-01 -3.51972967e-01 -9.10828710e-01 7.76700139e-01
5.40916920e-01 -3.99067044e-01 -1.39898092e-01 1.21211320e-01
7.50113964e-01 1.55393630e-01 3.21162105e-01 -3.83958131e-01
1.56197056e-01 -6.79359555e-01 -1.78507999e-01 -5.26834846e-01
-8.96132812e-02 -5.02429426e-01 5.00434995e-01 7.25395620e-01
-2.11053774e-01 2.85459638e-01 3.27149004e-01 7.75994360e-01
-3.71678352e-01 5.79344451e-01 3.41330588e-01 -1.55531600e-01
-4.31687027e-01 -2.66817510e-01 -6.95152044e-01 1.09031938e-01
1.11162782e+00 9.99444500e-02 -5.72116256e-01 -4.94522214e-01
-1.12878215e+00 2.41020456e-01 8.18071365e-01 3.42425913e-01
-1.27073035e-01 -1.33692002e+00 -6.59727156e-01 -5.97294450e-01
-2.76088297e-01 -4.38435018e-01 1.22164376e-01 9.09512699e-01
-6.43895984e-01 6.23772204e-01 -5.42050460e-03 -2.93963850e-01
-1.06773126e+00 2.20548660e-01 9.73629355e-02 -4.54994202e-01
-9.86387312e-01 3.85749400e-01 1.13776937e-01 -5.28154731e-01
-3.75854850e-01 -4.25060727e-02 -4.43246454e-01 9.65653658e-01
-6.80451049e-03 4.77863461e-01 -6.68449521e-01 -1.02893090e+00
-6.71502054e-01 3.95067900e-01 4.13946539e-01 -7.99730361e-01
1.71166539e+00 -6.16939187e-01 -7.00225592e-01 9.16057229e-01
7.00977683e-01 1.84415728e-01 -6.98340654e-01 -4.32061493e-01
6.47768855e-01 -4.94932860e-01 -2.09033564e-01 -4.57137734e-01
-4.11585510e-01 2.98588514e-01 5.23566008e-02 1.24617147e+00
5.51554441e-01 5.87667167e-01 3.12524452e-03 -1.55458912e-01
3.34460586e-01 -1.28411400e+00 -2.03901693e-01 4.88276750e-01
1.07753015e+00 -5.29674292e-01 3.47194076e-01 -3.26156050e-01
-3.84133786e-01 8.18348110e-01 -1.78488851e-01 -1.32423446e-01
1.11659622e+00 6.71485951e-03 -2.22282574e-01 -1.14970922e+00
-7.80946732e-01 1.33653462e-01 6.94784801e-03 1.19788267e-01
2.22930700e-01 4.59250242e-01 -8.14856887e-01 2.16984540e-01
-5.92434108e-01 -6.35625720e-01 5.58426023e-01 1.16685891e+00
-6.19857013e-01 -8.02411735e-01 -3.91380876e-01 3.83951992e-01
-4.67920065e-01 -3.04924753e-02 -9.38750327e-01 1.18947482e+00
-3.91319662e-01 7.59816349e-01 2.88470060e-01 -4.96551767e-02
9.53148752e-02 1.01500832e-01 1.07861862e-01 -2.78926164e-01
-5.50768673e-01 7.44187608e-02 3.73285532e-01 -3.52535218e-01
-6.17581069e-01 -1.20607591e+00 -8.01303864e-01 -8.09282124e-01
-3.19046199e-01 3.13438118e-01 8.25868845e-01 8.31640661e-01
4.22773659e-01 4.88006026e-01 6.54826581e-01 -2.35312730e-01
6.08013198e-02 -1.22531164e+00 -6.26028180e-01 2.51319975e-01
1.96902901e-02 -8.76894772e-01 -4.48091537e-01 -2.26444095e-01]
|
[8.888388633728027, 9.848833084106445]
|
17f3f17a-2897-433a-94a5-2576cf3f6f34
|
topics-in-the-haystack-extracting-and
|
2303.17324
| null |
https://arxiv.org/abs/2303.17324v1
|
https://arxiv.org/pdf/2303.17324v1.pdf
|
Topics in the Haystack: Extracting and Evaluating Topics beyond Coherence
|
Extracting and identifying latent topics in large text corpora has gained increasing importance in Natural Language Processing (NLP). Most models, whether probabilistic models similar to Latent Dirichlet Allocation (LDA) or neural topic models, follow the same underlying approach of topic interpretability and topic extraction. We propose a method that incorporates a deeper understanding of both sentence and document themes, and goes beyond simply analyzing word frequencies in the data. This allows our model to detect latent topics that may include uncommon words or neologisms, as well as words not present in the documents themselves. Additionally, we propose several new evaluation metrics based on intruder words and similarity measures in the semantic space. We present correlation coefficients with human identification of intruder words and achieve near-human level results at the word-intrusion task. We demonstrate the competitive performance of our method with a large benchmark study, and achieve superior results compared to state-of-the-art topic modeling and document clustering models.
|
['Benjamin Säfken', 'Elisabeth Bergherr', 'Arik Reuter', 'Quentin Seifert', 'Anton Thielmann']
|
2023-03-30
| null | null | null | null |
['topic-models']
|
['natural-language-processing']
|
[-2.57480234e-01 9.35675129e-02 -3.04572374e-01 -3.39610457e-01
-7.06469536e-01 -5.74176729e-01 1.03477001e+00 6.26714408e-01
-2.90485263e-01 2.64822304e-01 7.03026474e-01 -1.93854049e-01
-5.66050224e-02 -7.35597134e-01 7.76521266e-02 -5.34700334e-01
-2.68873066e-01 8.15789461e-01 1.54663548e-01 6.63414970e-02
6.00749850e-01 6.45645186e-02 -1.14855433e+00 3.87047172e-01
7.32702553e-01 5.16015172e-01 -1.14262573e-01 2.98520952e-01
-7.36471653e-01 3.78620058e-01 -8.57351959e-01 -3.50973845e-01
-3.55706573e-01 -4.83024828e-02 -8.55894268e-01 3.48329931e-01
-3.14854197e-02 -9.55607668e-02 -1.80625558e-01 9.00894582e-01
1.59306690e-01 1.75800636e-01 1.05139685e+00 -1.35204208e+00
-6.18217766e-01 8.29444110e-01 -7.96427190e-01 4.06023771e-01
3.41771215e-01 -2.33222723e-01 1.31116629e+00 -1.10391891e+00
4.96262938e-01 1.70001030e+00 5.31044483e-01 3.11606675e-01
-1.39606845e+00 -8.36982012e-01 4.82117206e-01 2.59520680e-01
-1.39590597e+00 -1.44666687e-01 9.50020373e-01 -7.34485686e-01
1.10289943e+00 -4.58889157e-02 2.13310599e-01 1.58446777e+00
3.76274377e-01 8.95240247e-01 7.10637391e-01 -4.88879591e-01
5.53144991e-01 4.73500162e-01 9.14348125e-01 1.85407162e-01
3.64136934e-01 -5.77429831e-01 -6.70160770e-01 -8.91316652e-01
1.23994194e-01 2.73729056e-01 -7.64794722e-02 -4.39636201e-01
-9.37401295e-01 1.62014925e+00 -2.61047214e-01 5.23594677e-01
-5.49006283e-01 -2.02518627e-01 5.14270067e-01 -2.48114988e-01
1.23634577e+00 4.95099634e-01 -2.01022938e-01 -1.66337386e-01
-1.25199449e+00 1.62869051e-01 1.10535991e+00 7.86321104e-01
6.13218665e-01 -2.25321189e-01 -1.27822414e-01 1.00710166e+00
6.34982646e-01 1.87178612e-01 9.08409417e-01 -5.21946549e-01
3.98702741e-01 5.93284428e-01 7.71179199e-02 -1.34669435e+00
-2.81403780e-01 -1.09381765e-01 -5.25427818e-01 -3.96398693e-01
-2.04560697e-01 -1.11234725e-01 -7.51771212e-01 1.55165350e+00
1.76965654e-01 8.60266909e-02 -1.34036690e-03 1.66517898e-01
4.63836998e-01 9.95274544e-01 6.11255705e-01 -5.59596419e-01
1.79884338e+00 -5.85529864e-01 -1.07752621e+00 -2.96977013e-01
5.15172839e-01 -7.30920315e-01 1.03258240e+00 5.20183325e-01
-6.32127404e-01 -1.76639602e-01 -8.06815922e-01 1.14015557e-01
-5.95082283e-01 -4.24878746e-01 8.10644150e-01 1.06842506e+00
-8.50369632e-01 2.95217305e-01 -8.91384602e-01 -7.73541272e-01
3.70391130e-01 3.21596228e-02 -4.14237268e-02 1.89492300e-01
-1.41491950e+00 5.50095081e-01 5.50411761e-01 -5.94341040e-01
-9.75236118e-01 -7.53552914e-01 -8.15022767e-01 2.83366352e-01
3.76710087e-01 -2.91142434e-01 1.03224707e+00 -1.64976776e-01
-1.12610209e+00 7.05411434e-01 -5.74126959e-01 -6.23726785e-01
-1.09961890e-01 -5.79870641e-01 -4.45114702e-01 2.94417381e-01
3.51097435e-01 6.01789415e-01 8.87707412e-01 -1.32585251e+00
-4.57621008e-01 -4.97303963e-01 -4.65046078e-01 6.35889471e-02
-1.06974733e+00 4.20324177e-01 -2.66040683e-01 -7.64917552e-01
1.72047481e-01 -7.39054859e-01 -1.81422353e-01 -5.00262499e-01
-7.69981086e-01 -9.10597861e-01 1.18761480e+00 -5.09393871e-01
1.41969478e+00 -2.16916966e+00 -2.34249547e-01 3.06562215e-01
6.20880425e-01 -1.15260005e-01 2.63565958e-01 6.96165085e-01
-1.68641284e-01 4.61771339e-01 -1.16259448e-01 -8.22450936e-01
2.95720875e-01 -1.33684082e-02 -1.21070027e+00 3.97489756e-01
-1.25175819e-01 5.40091217e-01 -6.87497139e-01 -5.50800979e-01
-7.29201138e-02 4.16971594e-01 -6.52987599e-01 3.96157764e-02
-1.62466839e-01 -2.90917337e-01 -4.35979694e-01 2.58034855e-01
3.78425688e-01 -3.10802281e-01 2.28221714e-01 3.23608458e-01
7.68994018e-02 7.11512506e-01 -8.18379223e-01 1.50384271e+00
-2.52173573e-01 8.32757413e-01 -3.59870881e-01 -8.29225361e-01
9.81040001e-01 6.09655201e-01 5.92165053e-01 4.44063358e-02
7.93494508e-02 -4.29064512e-01 -4.14375067e-01 -1.34579346e-01
8.59152973e-01 -2.65208721e-01 -3.25344801e-01 9.82181549e-01
2.68754989e-01 8.26733336e-02 1.65443301e-01 8.02094758e-01
9.35449302e-01 -6.85041606e-01 4.56058115e-01 -6.45188034e-01
-4.41873558e-02 9.37745497e-02 2.82124341e-01 8.77741516e-01
-1.09936245e-01 2.81192571e-01 8.28324139e-01 -2.60385126e-01
-8.18949282e-01 -1.35887754e+00 -3.77579063e-01 1.36418056e+00
1.52774481e-02 -1.05766714e+00 -7.31434345e-01 -6.91267788e-01
-9.98988524e-02 1.28033292e+00 -7.85378754e-01 -1.93214297e-01
-2.05246620e-02 -1.09810019e+00 5.42385757e-01 3.91999245e-01
6.05005808e-02 -9.46222782e-01 -2.66367108e-01 2.76438743e-01
-4.34463620e-01 -1.10054445e+00 -1.97513700e-01 7.63061568e-02
-9.29502666e-01 -5.44348657e-01 -3.47982496e-01 -2.92994142e-01
4.55455899e-01 3.48936558e-01 9.94951785e-01 -4.34015036e-01
-4.02108163e-01 4.33935642e-01 -4.54021782e-01 -6.61971986e-01
-2.06329703e-01 1.97069600e-01 5.03815472e-01 -1.11748032e-01
1.17183936e+00 -6.41795754e-01 -3.39061350e-01 3.75152349e-01
-1.02226186e+00 -4.06307548e-01 9.91725698e-02 4.46159601e-01
-1.06428057e-01 4.61081415e-01 5.22722483e-01 -1.15178812e+00
1.09227514e+00 -9.61848438e-01 -6.53136596e-02 1.77088268e-02
-8.41003239e-01 -2.05032766e-01 6.24813586e-02 -5.88428319e-01
-1.06904471e+00 -5.51608622e-01 2.34078720e-01 -2.95121104e-01
-6.10807657e-01 4.05421883e-01 -2.13194802e-01 8.30908000e-01
6.81433201e-01 2.55699754e-01 -3.63724351e-01 -5.49221218e-01
6.44204080e-01 8.26047421e-01 1.30504698e-01 -4.44686145e-01
7.69109666e-01 9.51261044e-01 -7.88511813e-01 -1.19525599e+00
-9.24346983e-01 -1.19290555e+00 -6.22651517e-01 1.12794787e-01
7.92988241e-01 -9.34020519e-01 -3.24215859e-01 1.67508736e-01
-1.58316958e+00 2.94705778e-01 -2.44989052e-01 4.74726796e-01
-2.97363281e-01 7.09707737e-01 -7.27933884e-01 -1.07217324e+00
-4.33706194e-01 -7.41666496e-01 1.23883975e+00 -2.40551949e-01
-1.12866509e+00 -1.35094607e+00 4.50794369e-01 1.58194274e-01
3.11420947e-01 -1.65533811e-01 1.35896695e+00 -1.48394394e+00
-3.70719545e-02 -2.85499364e-01 -7.18482137e-02 -7.22373351e-02
2.74372786e-01 -2.15411171e-01 -1.04098737e+00 -2.86119312e-01
5.48032939e-01 -1.61751453e-02 1.13427937e+00 3.18109900e-01
8.82791519e-01 -5.46406209e-01 -8.94081771e-01 4.91503477e-02
1.03698313e+00 2.31852159e-01 6.35245442e-01 2.32799217e-01
5.35973191e-01 1.00395644e+00 1.99311838e-01 6.89085484e-01
2.50512481e-01 4.90176708e-01 -1.03387691e-01 1.49948463e-01
6.18357837e-01 -2.38706291e-01 4.80159909e-01 8.42542171e-01
5.73668897e-01 -5.00461340e-01 -1.35823095e+00 9.08336639e-01
-1.63715279e+00 -1.15713918e+00 -1.24646276e-01 1.81138933e+00
8.00816119e-01 3.71642977e-01 1.74927279e-01 2.17220247e-01
9.42807257e-01 3.43237728e-01 -1.88087150e-01 -4.35319185e-01
2.00259924e-01 -2.12136451e-02 -8.79167542e-02 5.54838598e-01
-1.30659401e+00 1.19871485e+00 7.13988256e+00 1.08359158e+00
-5.91696262e-01 2.82471508e-01 6.73015773e-01 -3.66808660e-02
-3.61675233e-01 -7.42664561e-02 -1.28257906e+00 4.13432449e-01
1.10370314e+00 -5.37554681e-01 -2.13024333e-01 1.31678104e+00
1.65171102e-01 -5.82466871e-02 -1.04621756e+00 6.60868943e-01
6.62693918e-01 -9.07290161e-01 4.72966641e-01 4.16511208e-01
7.64202416e-01 -2.21936181e-01 3.10030073e-01 3.42358232e-01
5.52011251e-01 -7.91526794e-01 2.57659376e-01 2.09806506e-02
1.66413531e-01 -7.11424887e-01 5.86760223e-01 4.21183914e-01
-8.03459942e-01 -1.01268971e-02 -5.86935818e-01 1.56616449e-01
3.12463671e-01 9.79679763e-01 -1.14425659e+00 1.31440669e-01
6.17685974e-01 6.56792164e-01 -4.97842789e-01 7.00632572e-01
-2.53661722e-01 1.08161950e+00 -3.18764180e-01 -3.73151042e-02
1.83990046e-01 1.16146564e-01 8.38750541e-01 1.44892442e+00
7.65888616e-02 -1.09687768e-01 1.56028688e-01 9.57899988e-01
1.30461887e-01 4.33387667e-01 -7.33211815e-01 -3.49118322e-01
5.75762331e-01 1.28353691e+00 -1.24386907e+00 -6.49872661e-01
-1.25796974e-01 8.06267977e-01 -2.04495676e-02 4.29837555e-01
-6.31670475e-01 -3.80875438e-01 7.70123482e-01 2.01612756e-01
2.47376725e-01 -4.89252597e-01 -4.53958839e-01 -1.20919943e+00
-2.41120458e-01 -5.00023305e-01 5.80408990e-01 -3.58932018e-01
-1.76541710e+00 7.59117305e-01 5.21813989e-01 -9.22925174e-01
-4.07824636e-01 -3.17893118e-01 -9.72349226e-01 5.10537803e-01
-1.02731431e+00 -8.17177594e-01 -3.30769387e-03 4.92907196e-01
9.11609113e-01 -3.35133076e-01 9.83111680e-01 -1.26109496e-01
-3.79111856e-01 2.98146993e-01 2.28320226e-01 1.26651630e-01
6.82310402e-01 -1.34076643e+00 8.31723928e-01 6.45118058e-01
4.05715704e-01 1.34383869e+00 8.85151565e-01 -9.05731261e-01
-6.28920496e-01 -9.65299845e-01 1.15125310e+00 -9.60474312e-01
1.09934986e+00 -9.68253791e-01 -1.08367121e+00 7.01551497e-01
3.87860060e-01 -9.11501944e-01 1.35354650e+00 9.44269001e-01
-6.55705214e-01 6.11559629e-01 -6.80577278e-01 5.51465750e-01
5.91208220e-01 -6.95404708e-01 -1.18971252e+00 5.82013369e-01
9.69419777e-01 5.45670331e-01 -3.24768037e-01 -2.58747116e-03
2.29057744e-01 -5.65892994e-01 1.05103302e+00 -9.66595650e-01
4.05428022e-01 1.56402349e-01 -4.64681387e-02 -1.29174590e+00
-4.07447219e-01 -6.58680499e-01 -2.50878304e-01 1.61651838e+00
3.90672833e-01 -5.13790607e-01 9.13265884e-01 6.75337017e-01
2.62617648e-01 -2.87684172e-01 -8.21245193e-01 -7.66597271e-01
2.52206296e-01 -7.62960851e-01 1.80237964e-01 1.26990116e+00
7.43355215e-01 7.49458730e-01 -2.95004338e-01 1.87388361e-01
9.06036198e-01 3.05490941e-02 5.42221010e-01 -1.74748611e+00
3.18087637e-02 -4.59722430e-01 -3.88269216e-01 -8.83368969e-01
5.26394010e-01 -6.14699721e-01 7.68652558e-02 -1.47033560e+00
6.05800927e-01 1.07063539e-01 -1.53218165e-01 3.58133465e-01
-3.63150090e-01 -8.38983804e-03 -1.64181709e-01 6.26484752e-01
-1.05695796e+00 8.63215387e-01 1.13408335e-01 -9.42023620e-02
-3.20206821e-01 -2.65096277e-01 -9.21829641e-01 1.10198677e+00
8.29979837e-01 -9.52124119e-01 -4.14170980e-01 1.26252053e-02
-1.21792033e-03 -4.39185530e-01 5.95767200e-02 -8.39066744e-01
3.79061729e-01 1.21671982e-01 1.38296753e-01 -8.42693985e-01
4.70121294e-01 -5.87010086e-01 -4.88240361e-01 2.45254472e-01
-6.07414126e-01 -1.16632439e-01 2.04117775e-01 9.90739107e-01
-2.44755149e-01 -1.04656719e-01 4.72066492e-01 4.00664508e-02
-4.14153546e-01 1.01890504e-01 -1.05149364e+00 -4.43528593e-02
1.00349402e+00 1.47737479e-02 -4.46502537e-01 -6.42586291e-01
-7.75801837e-01 1.49557739e-01 1.10140242e-01 7.84704626e-01
6.31868124e-01 -1.05865860e+00 -5.73741198e-01 -3.71813886e-02
3.09241205e-01 -5.38450837e-01 3.44104916e-01 4.65734690e-01
3.61775570e-02 8.97825480e-01 2.79979110e-01 -5.69341540e-01
-1.18645728e+00 8.14615428e-01 -4.30732012e-01 -5.35262406e-01
-5.39918423e-01 6.34773910e-01 8.76029313e-01 -8.96911621e-02
3.37717623e-01 1.47807837e-01 -5.30770838e-01 5.51040471e-01
8.87789726e-01 2.63605893e-01 -2.10108921e-01 -4.41817671e-01
-3.71455282e-01 1.72214851e-01 -6.99222386e-01 -3.86986494e-01
1.21062779e+00 -2.55403519e-01 -2.72271723e-01 8.46202910e-01
1.18312228e+00 -1.21641733e-01 -5.96991837e-01 -5.14124453e-01
6.96225822e-01 -3.58543783e-01 1.05544470e-01 -2.59756237e-01
-2.46764734e-01 1.16600513e+00 3.47507745e-01 7.07382202e-01
4.68621045e-01 3.96722525e-01 7.67498791e-01 5.45485914e-01
1.81581736e-01 -1.07480407e+00 5.03478169e-01 5.50167084e-01
5.68791270e-01 -1.12089062e+00 2.05554828e-01 -6.73770487e-01
-7.34322250e-01 8.50574076e-01 3.16933274e-01 -6.47353828e-02
9.99130309e-01 6.76195398e-02 2.40675062e-02 -6.71347380e-01
-1.04542017e+00 8.88869166e-02 2.87460953e-01 3.92498136e-01
4.71954435e-01 -6.09446578e-02 -2.51709968e-01 7.84326017e-01
-3.83918911e-01 -7.63464749e-01 3.55282158e-01 6.89662635e-01
-8.47049236e-01 -9.28380191e-01 -4.93662000e-01 3.59723657e-01
-8.55714381e-01 -3.15181851e-01 -6.36715233e-01 4.89968985e-01
-4.53445196e-01 1.26359499e+00 3.73537868e-01 -2.02517584e-01
-1.79827869e-01 7.11480200e-01 -4.96717125e-01 -1.16968763e+00
-2.79456049e-01 4.26552445e-01 -7.63485208e-02 -3.32151055e-01
-1.19997270e-01 -9.07677174e-01 -9.33597267e-01 -2.17581436e-01
-5.49814999e-01 9.10048962e-01 9.55600917e-01 1.10577524e+00
2.89449334e-01 2.92698503e-01 5.76978862e-01 -4.93960857e-01
-2.57683694e-01 -1.33354127e+00 -8.56768906e-01 3.75042588e-01
-2.32596919e-01 -6.57070100e-01 -6.25766456e-01 2.92654037e-01]
|
[10.393796920776367, 6.996838092803955]
|
62e744ca-9b1f-49d6-9e89-098ae5a3ac06
|
vuldeepecker-a-deep-learning-based-system-for
|
1801.01681
| null |
http://arxiv.org/abs/1801.01681v1
|
http://arxiv.org/pdf/1801.01681v1.pdf
|
VulDeePecker: A Deep Learning-Based System for Vulnerability Detection
|
The automatic detection of software vulnerabilities is an important research
problem. However, existing solutions to this problem rely on human experts to
define features and often miss many vulnerabilities (i.e., incurring high false
negative rate). In this paper, we initiate the study of using deep
learning-based vulnerability detection to relieve human experts from the
tedious and subjective task of manually defining features. Since deep learning
is motivated to deal with problems that are very different from the problem of
vulnerability detection, we need some guiding principles for applying deep
learning to vulnerability detection. In particular, we need to find
representations of software programs that are suitable for deep learning. For
this purpose, we propose using code gadgets to represent programs and then
transform them into vectors, where a code gadget is a number of (not
necessarily consecutive) lines of code that are semantically related to each
other. This leads to the design and implementation of a deep learning-based
vulnerability detection system, called Vulnerability Deep Pecker
(VulDeePecker). In order to evaluate VulDeePecker, we present the first
vulnerability dataset for deep learning approaches. Experimental results show
that VulDeePecker can achieve much fewer false negatives (with reasonable false
positives) than other approaches. We further apply VulDeePecker to 3 software
products (namely Xen, Seamonkey, and Libav) and detect 4 vulnerabilities, which
are not reported in the National Vulnerability Database but were "silently"
patched by the vendors when releasing later versions of these products; in
contrast, these vulnerabilities are almost entirely missed by the other
vulnerability detection systems we experimented with.
|
['Sujuan Wang', 'Zhijun Deng', 'Yuyi Zhong', 'Xinyu Ou', 'Hai Jin', 'Deqing Zou', 'Zhen Li', 'Shouhuai Xu']
|
2018-01-05
| null | null | null | null |
['vulnerability-detection']
|
['miscellaneous']
|
[-5.08490443e-01 -2.43145540e-01 -1.27728134e-01 -2.94567227e-01
-5.44827938e-01 -9.38645601e-01 1.32538527e-01 3.57171863e-01
-8.83490499e-03 1.46870136e-01 -2.08764136e-01 -9.98650491e-01
1.26541823e-01 -1.05862963e+00 -4.60067034e-01 -3.02094072e-01
-3.05154383e-01 -2.06288353e-01 3.61386955e-01 -4.25810784e-01
4.42162424e-01 4.42037761e-01 -1.34746075e+00 4.80841547e-01
6.01137161e-01 6.16392553e-01 -2.48533309e-01 5.08274198e-01
-3.13207805e-01 7.32888699e-01 -8.45750809e-01 -7.96503901e-01
4.03921127e-01 6.30393028e-02 -9.05691564e-01 -6.19276345e-01
3.79953206e-01 -4.65818495e-01 -4.41923775e-02 1.48752522e+00
2.95305811e-02 -3.08874547e-01 4.24738109e-01 -1.54689407e+00
-7.38528609e-01 8.50555301e-01 -7.67529666e-01 2.47619465e-01
2.48583049e-01 8.13784227e-02 1.18745089e+00 -6.82423055e-01
2.40279496e-01 1.10273421e+00 9.33856308e-01 6.56062663e-01
-9.93839741e-01 -6.82696402e-01 9.86106321e-02 2.08899155e-02
-1.33804703e+00 -1.20694742e-01 7.82297850e-01 -8.72753501e-01
1.24961960e+00 4.40351695e-01 1.59721494e-01 1.20849192e+00
4.58535284e-01 3.53712380e-01 6.04605138e-01 -3.73073786e-01
3.17083448e-01 4.03459519e-01 8.54062974e-01 6.32641137e-01
5.90905190e-01 3.49931628e-01 4.13182944e-01 -7.73000658e-01
3.76939327e-01 1.40998259e-01 -1.89076934e-03 -3.57668430e-01
-4.33839053e-01 1.25453818e+00 4.15045828e-01 3.84971797e-01
-1.63265318e-01 1.28838450e-01 8.63546968e-01 6.56813622e-01
-1.13624677e-01 7.10539281e-01 -6.53477490e-01 -9.86148044e-02
-6.70821905e-01 2.18438998e-01 9.47602749e-01 5.68852901e-01
9.16754723e-01 3.60323995e-01 1.63655281e-01 5.40658057e-01
3.08229148e-01 -2.20724251e-04 3.19249034e-01 -3.58025461e-01
4.88984466e-01 8.98324430e-01 -6.45406023e-02 -1.43678510e+00
-3.10063094e-01 -1.27020478e-01 -4.82504457e-01 6.39945984e-01
2.25343853e-02 -3.57914656e-01 -7.95708656e-01 1.56418288e+00
-1.14237070e-02 -1.48671255e-01 2.28463933e-01 5.05189478e-01
6.61260545e-01 5.13283074e-01 1.87341198e-01 2.37841561e-01
1.15361643e+00 -6.38018012e-01 -2.05410630e-01 -7.86207914e-02
1.08686090e+00 -5.88209331e-01 1.23822975e+00 5.49210191e-01
-3.52781981e-01 -4.81386214e-01 -1.17881739e+00 4.68352884e-01
-7.46943474e-01 -2.01400280e-01 1.01497757e+00 1.34528244e+00
-1.05902743e+00 4.11913872e-01 -7.19328523e-01 -6.90057129e-02
2.74733782e-01 3.35241377e-01 -4.37856168e-01 2.76978523e-01
-1.31542695e+00 7.29552865e-01 5.45430243e-01 -2.14574650e-01
-1.16284442e+00 -4.79716510e-01 -1.04515803e+00 4.03649062e-01
4.81998056e-01 9.24067721e-02 1.14800608e+00 -1.14435196e+00
-6.97459340e-01 5.83817244e-01 3.80040795e-01 -3.48133087e-01
-2.40195338e-02 -2.68910229e-01 -8.58537972e-01 -2.14378238e-01
-9.95461196e-02 -1.22747190e-01 7.44912565e-01 -1.24427223e+00
-4.97105718e-01 -3.99368210e-03 7.37535119e-01 -7.36208379e-01
-9.48845804e-01 4.76938844e-01 -1.69086203e-01 -5.62598944e-01
-4.65946227e-01 -7.30848670e-01 -3.06371808e-01 -4.64676231e-01
-6.80992544e-01 -6.75316602e-02 7.54479349e-01 -6.68503881e-01
1.56795895e+00 -2.21114111e+00 -3.03808302e-01 5.41534305e-01
5.31572104e-01 8.99913073e-01 -5.25224686e-01 4.32834417e-01
-5.66441536e-01 6.46863520e-01 -3.26705247e-01 1.66311875e-01
2.83272285e-02 -2.08696872e-02 -7.17055261e-01 2.55226105e-01
2.52350748e-01 7.96589613e-01 -8.12547505e-01 -7.17943162e-02
-3.53446491e-02 3.04361582e-01 -7.09223986e-01 1.75491452e-01
-2.67980635e-01 -3.46141249e-01 -6.43817425e-01 9.47024882e-01
7.67294109e-01 1.49343207e-01 7.02584162e-02 1.20427489e-01
-1.29348323e-01 1.52005747e-01 -9.42224205e-01 9.46485341e-01
-5.83680689e-01 5.14930725e-01 -2.01387540e-01 -9.24770534e-01
1.13540339e+00 2.68552870e-01 8.77513438e-02 -5.88119149e-01
7.20892772e-02 1.21950857e-01 6.28075153e-02 -5.95823288e-01
3.74894887e-01 1.16986856e-01 -5.60743809e-01 6.97121978e-01
-1.85627475e-01 3.47619504e-01 -4.03400660e-02 2.07119286e-01
1.51342928e+00 -3.93187374e-01 3.01938862e-01 -1.98577553e-01
6.84363008e-01 -6.07221164e-02 8.13651800e-01 7.40171373e-01
-2.01448575e-01 1.99251875e-01 1.21349382e+00 -9.16453600e-01
-8.26316655e-01 -1.00472617e+00 1.68897524e-01 8.20051432e-01
-1.94977835e-01 -8.48848224e-01 -1.02152467e+00 -1.33921516e+00
6.01220271e-03 6.84782326e-01 -7.61281431e-01 -5.78355730e-01
-4.53374863e-01 -5.37009478e-01 8.48936796e-01 7.72007227e-01
1.04897417e-01 -1.12640238e+00 -8.31145048e-01 1.82235435e-01
2.68936217e-01 -5.96038699e-01 -3.51821899e-01 2.10916132e-01
-3.73155475e-01 -1.31955040e+00 -1.39544070e-01 -6.02657735e-01
4.43032801e-01 1.13080546e-01 1.27912366e+00 5.83637178e-01
-6.15182757e-01 1.22170888e-01 -5.86332738e-01 -2.10440442e-01
-6.78787708e-01 7.62211755e-02 -5.36677949e-02 -2.71941960e-01
7.95923054e-01 -4.01720494e-01 -8.29503834e-02 1.79219753e-01
-1.12313628e+00 -8.40205908e-01 6.42650187e-01 7.47769296e-01
2.21472025e-01 4.84476328e-01 6.70144677e-01 -1.01683676e+00
8.97085428e-01 -7.93974638e-01 -1.05271125e+00 4.15998429e-01
-4.25528467e-01 2.37206556e-02 8.13855946e-01 -4.59638029e-01
-8.52838814e-01 5.87390624e-02 -5.72333634e-01 -4.62175578e-01
-6.85152635e-02 9.02416885e-01 -4.22358125e-01 -4.69175369e-01
8.52460027e-01 -1.61986068e-01 -3.41988325e-01 -4.94200379e-01
1.99441433e-01 5.62311590e-01 2.64447242e-01 -6.12420022e-01
9.49474156e-01 -1.76261425e-01 -4.10656452e-01 -5.36329508e-01
-2.35017285e-01 -5.02931466e-03 -3.63104343e-01 2.61738598e-01
4.85016286e-01 -4.88346130e-01 -6.73253536e-01 4.33679909e-01
-1.07066083e+00 -1.39770851e-01 1.12022750e-01 -1.05344132e-01
-8.19908977e-02 7.76578665e-01 -6.47260845e-01 -6.62475049e-01
-5.22859752e-01 -1.35818684e+00 4.43245709e-01 -3.99998687e-02
-4.81593490e-01 -9.25721049e-01 3.97419542e-01 -2.35815197e-01
5.45966148e-01 4.81764585e-01 1.58470345e+00 -8.34250510e-01
-1.81871802e-01 -5.15551329e-01 -1.21153355e-01 7.32374310e-01
1.92318156e-01 4.75249171e-01 -9.72781301e-01 -3.28011185e-01
1.29257724e-01 -2.23160043e-01 6.34137273e-01 -1.02990627e-01
1.21498859e+00 -3.48760039e-01 -2.82673627e-01 6.46754801e-01
1.68981290e+00 4.49470133e-01 7.18585193e-01 6.02188468e-01
8.55511069e-01 7.85407960e-01 4.84507591e-01 4.18060035e-01
1.21058725e-01 6.85723484e-01 8.54204834e-01 -1.66056268e-02
4.25654143e-01 -9.55309160e-03 5.97093225e-01 2.97831386e-01
3.24135363e-01 -7.80059025e-02 -1.28844631e+00 6.00652337e-01
-1.47536016e+00 -7.62439132e-01 -2.86952645e-01 2.25648975e+00
6.05420411e-01 3.18797737e-01 1.92847550e-01 8.11473802e-02
6.00518167e-01 -2.75927912e-02 -2.94180244e-01 -1.17709374e+00
2.61874199e-01 3.80335480e-01 1.73856825e-01 3.31461459e-01
-1.27541089e+00 1.15154791e+00 6.27318525e+00 5.37388563e-01
-1.24116528e+00 4.26120646e-02 3.09013128e-01 2.50225604e-01
-5.77148795e-01 2.28745043e-01 -7.67812431e-01 3.89742315e-01
1.09668171e+00 -1.89525962e-01 1.17397979e-01 1.32733667e+00
-4.46121454e-01 3.04821402e-01 -1.19328439e+00 6.02202535e-01
-1.09953612e-01 -1.08531392e+00 -1.14589490e-01 2.68993899e-02
4.13605124e-01 -2.29352415e-01 2.64411658e-01 7.10490465e-01
5.42889833e-01 -1.03466392e+00 5.80964088e-01 8.22406858e-02
4.94671822e-01 -1.20853519e+00 9.38748419e-01 1.39667895e-02
-1.07928669e+00 -6.14923656e-01 -5.34765244e-01 1.04386270e-01
-2.71799773e-01 6.94366455e-01 -5.82221210e-01 3.31761301e-01
8.75918746e-01 2.41751045e-01 -7.94566095e-01 9.71376657e-01
-3.47401232e-01 4.85244811e-01 1.20446637e-01 1.71759322e-01
3.46537173e-01 4.08313125e-01 2.99257904e-01 1.21143270e+00
3.65651190e-01 -3.86540502e-01 1.21815234e-01 1.21322477e+00
2.14197561e-02 -9.70760435e-02 -8.09011996e-01 -3.14666420e-01
4.36115950e-01 1.38561010e+00 -5.91372728e-01 -7.27675185e-02
-7.84242690e-01 5.37127793e-01 3.56093645e-01 2.90212870e-01
-8.64622712e-01 -9.17230427e-01 1.14643121e+00 -9.91935134e-02
3.67919266e-01 -3.37938294e-02 -1.50817826e-01 -1.12934244e+00
2.09342465e-01 -1.19260108e+00 4.68348026e-01 -1.71244696e-01
-1.32923019e+00 1.13003469e+00 -5.38757332e-02 -1.00855541e+00
-2.66274512e-01 -6.01410687e-01 -1.21168113e+00 1.06853533e+00
-1.30998766e+00 -9.58779573e-01 -6.74336702e-02 5.39467454e-01
1.10577039e-01 -4.50766325e-01 1.03628254e+00 3.64111841e-01
-7.31470168e-01 1.09375405e+00 -1.40087888e-01 4.04979497e-01
3.83009911e-01 -1.07710052e+00 1.14552629e+00 1.13981485e+00
3.85451684e-04 1.10189188e+00 3.66735637e-01 -9.33158994e-01
-1.18592548e+00 -1.14340246e+00 7.25089252e-01 -5.92787325e-01
8.18841994e-01 -5.24847746e-01 -1.34614825e+00 7.42590725e-01
-9.23253968e-03 -1.67859830e-02 8.84446383e-01 1.95703834e-01
-1.00904703e+00 1.58966735e-01 -1.30168378e+00 5.96349180e-01
4.32889313e-01 -7.00602829e-01 -6.32848442e-01 -1.35650828e-01
7.26663888e-01 1.31857753e-01 -6.41003668e-01 3.03428262e-01
3.92201990e-01 -1.15087509e+00 9.96244729e-01 -8.09861898e-01
3.06330651e-01 -1.34778202e-01 -1.83056563e-01 -1.12742639e+00
-3.21362972e-01 -3.19931537e-01 -1.91178516e-01 1.56236327e+00
3.71850640e-01 -7.72686839e-01 6.24080837e-01 6.71729624e-01
-1.31371155e-01 -6.87675476e-01 -5.85651517e-01 -1.04007328e+00
4.61489648e-01 -4.43421960e-01 1.06033242e+00 1.30389202e+00
1.46857798e-01 -1.90471336e-01 -2.22145811e-01 4.84638661e-01
4.78792638e-01 9.42455828e-02 5.34622133e-01 -1.25207067e+00
-5.57638168e-01 -5.58681190e-01 -6.84609473e-01 -1.38009921e-01
4.93396938e-01 -5.97878277e-01 3.47452750e-03 -9.04497743e-01
2.00982556e-01 -4.30301011e-01 -6.10582292e-01 1.07233548e+00
-1.52315944e-01 -1.10501967e-01 -1.35692239e-01 -3.49172838e-02
-1.52901307e-01 4.41035256e-02 1.88150615e-01 -3.70661974e-01
-2.07259119e-01 1.52671501e-01 -1.12185228e+00 8.52370977e-01
9.05428350e-01 -6.46698833e-01 -4.55431670e-01 -4.40649122e-01
4.26915288e-01 -1.16818167e-01 3.45443666e-01 -8.79677355e-01
-1.77539989e-01 -1.70548782e-01 -4.36420692e-03 -2.76052654e-01
-2.59975880e-01 -6.85309708e-01 3.50845717e-02 5.84031522e-01
-2.51127779e-01 4.20197248e-01 5.70512295e-01 6.70298785e-02
-1.79746240e-01 -8.38805676e-01 6.42887831e-01 -1.56175151e-01
-1.27628529e+00 2.67356694e-01 -6.56947792e-01 -2.47925818e-01
1.26338983e+00 8.88974443e-02 -5.32950342e-01 1.14509180e-01
-4.57782358e-01 6.90206513e-02 6.81634784e-01 8.14044476e-01
9.08187449e-01 -1.18227446e+00 -3.18325758e-01 4.98417318e-01
3.61525893e-01 -6.87089860e-01 2.35005111e-01 8.84506106e-02
-4.82868075e-01 2.23959208e-01 -5.36185086e-01 6.64793849e-02
-1.48319829e+00 1.17587388e+00 2.66239941e-01 -2.15685800e-01
-6.13521993e-01 8.76428902e-01 2.90950716e-01 -5.84566534e-01
3.27932417e-01 -1.67701542e-01 -5.26560903e-01 -1.36742562e-01
8.25355411e-01 1.14610538e-01 2.77915418e-01 -4.46023047e-01
-6.27436101e-01 4.55405414e-01 -6.04623556e-01 4.13058251e-01
1.29752755e+00 6.06076717e-01 -3.72673899e-01 -1.36156529e-01
1.40173233e+00 2.76745539e-02 -8.86401653e-01 1.05060555e-01
4.85391617e-01 -6.34650052e-01 2.75736824e-02 -9.09475863e-01
-1.52012622e+00 1.22600996e+00 8.49307418e-01 4.93100673e-01
1.03634453e+00 -1.32504806e-01 8.62821698e-01 5.01714826e-01
5.62096238e-01 -5.15129685e-01 1.16388045e-01 6.16491079e-01
5.68751276e-01 -1.03305590e+00 -3.07709068e-01 -3.14368039e-01
-4.64155495e-01 1.19882572e+00 9.62370038e-01 -1.24833724e-02
5.46070933e-01 6.04638755e-01 9.84176025e-02 -3.35080802e-01
-5.53748310e-01 1.72238708e-01 -3.17560509e-02 9.08179104e-01
2.79646218e-01 1.99394107e-01 -2.14596540e-02 9.70819533e-01
1.40599102e-01 -3.77465129e-01 8.28061342e-01 1.10594654e+00
-4.72932041e-01 -1.53481495e+00 -4.02886957e-01 3.51202548e-01
-6.81407869e-01 -2.82072842e-01 -6.18654966e-01 6.23328924e-01
8.57277289e-02 8.29320371e-01 -2.64676660e-01 -1.01308048e+00
4.29288179e-01 -2.10416958e-01 -5.62227750e-03 -9.29316044e-01
-9.88229632e-01 -3.93071264e-01 7.29966164e-03 -6.57111406e-01
5.68237424e-01 -2.96835423e-01 -1.05175948e+00 -3.41399610e-01
-1.30411044e-01 2.87549943e-01 4.39179689e-01 5.14393508e-01
2.55385071e-01 4.73549932e-01 7.38889337e-01 -4.24224377e-01
-8.47772896e-01 -3.77431065e-01 -5.02871275e-01 3.15567702e-01
4.66143221e-01 -7.12399840e-01 -5.02652526e-01 -2.38154396e-01]
|
[7.054413795471191, 7.778486728668213]
|
70f3ac43-877c-4b3f-95e7-97b1dc5e9129
|
multi-person-pose-estimation-with-enhanced-1
|
2003.10238
| null |
https://arxiv.org/abs/2003.10238v1
|
https://arxiv.org/pdf/2003.10238v1.pdf
|
Multi-Person Pose Estimation with Enhanced Feature Aggregation and Selection
|
We propose a novel Enhanced Feature Aggregation and Selection network (EFASNet) for multi-person 2D human pose estimation. Due to enhanced feature representation, our method can well handle crowded, cluttered and occluded scenes. More specifically, a Feature Aggregation and Selection Module (FASM), which constructs hierarchical multi-scale feature aggregation and makes the aggregated features discriminative, is proposed to get more accurate fine-grained representation, leading to more precise joint locations. Then, we perform a simple Feature Fusion (FF) strategy which effectively fuses high-resolution spatial features and low-resolution semantic features to obtain more reliable context information for well-estimated joints. Finally, we build a Dense Upsampling Convolution (DUC) module to generate more precise prediction, which can recover missing joint details that are usually unavailable in common upsampling process. As a result, the predicted keypoint heatmaps are more accurate. Comprehensive experiments demonstrate that the proposed approach outperforms the state-of-the-art methods and achieves the superior performance over three benchmark datasets: the recent big dataset CrowdPose, the COCO keypoint detection dataset and the MPII Human Pose dataset. Our code will be released upon acceptance.
|
['Xixia Xu', 'Qi Zou', 'Xue Lin']
|
2020-03-20
| null | null | null | null |
['2d-human-pose-estimation']
|
['computer-vision']
|
[-2.90761203e-01 -5.18141627e-01 1.43873349e-01 -4.60624337e-01
-7.73175657e-01 1.40424535e-01 4.26180899e-01 -1.08974576e-01
-4.79490489e-01 7.17154980e-01 7.35780895e-01 6.53089225e-01
-1.79935805e-02 -6.92938328e-01 -5.31527579e-01 -4.96797174e-01
1.23053407e-02 4.78109181e-01 5.78139186e-01 -3.47701460e-01
-1.92709744e-01 3.92683923e-01 -1.74795711e+00 1.44188449e-01
9.74664509e-01 1.09290159e+00 2.69197673e-01 2.42126137e-01
1.21315226e-01 3.62807304e-01 -6.12242520e-01 -2.92388529e-01
1.03309534e-01 1.26186281e-01 -4.91865426e-01 4.20578271e-02
6.13669097e-01 -5.51205218e-01 -5.89737892e-01 8.77485514e-01
8.36902857e-01 4.25281703e-01 4.70310062e-01 -1.15704501e+00
-3.21290076e-01 -7.05901235e-02 -8.62132072e-01 8.71098638e-02
7.97374487e-01 3.42020571e-01 7.55209148e-01 -1.12826180e+00
5.20416439e-01 1.87286222e+00 6.89194441e-01 2.28789106e-01
-6.90052688e-01 -7.03037858e-01 3.89456302e-01 3.20787758e-01
-1.63521838e+00 -2.62188166e-01 7.37279236e-01 -2.65530348e-01
5.74873507e-01 2.68037468e-01 1.06750154e+00 9.68418300e-01
1.99942783e-01 1.15154147e+00 9.66842949e-01 -2.24670637e-02
-1.80499703e-01 -4.79740739e-01 -1.26830429e-01 9.03804660e-01
4.09064114e-01 -4.60514091e-02 -7.03108668e-01 -3.37675512e-01
1.04318190e+00 5.76030195e-01 -3.47464323e-01 -2.39824608e-01
-1.45978737e+00 5.75507462e-01 8.66156459e-01 8.34280699e-02
-7.51713753e-01 2.04647779e-01 3.34299445e-01 -2.75704294e-01
3.34040433e-01 -1.13868512e-01 -3.48582119e-01 -7.40849748e-02
-8.32446992e-01 9.75692689e-01 2.28873581e-01 8.74461949e-01
8.74384463e-01 -3.70442122e-01 -7.67971039e-01 9.54022825e-01
5.78169525e-01 6.65901363e-01 4.11249429e-01 -8.89221191e-01
7.11727560e-01 7.35057175e-01 3.08206469e-01 -1.34150767e+00
-7.13358819e-01 -5.09566247e-01 -8.25673103e-01 -1.59349710e-01
2.77241945e-01 -1.01307131e-01 -9.59996402e-01 1.50960577e+00
8.32862079e-01 2.07001343e-01 -2.89719611e-01 1.40766418e+00
9.61598277e-01 4.71816361e-01 1.35525554e-01 1.01196438e-01
1.78752959e+00 -1.00849485e+00 -6.95694029e-01 -2.21137524e-01
7.51617551e-02 -6.99436665e-01 9.11308825e-01 2.14458674e-01
-7.87329495e-01 -9.72983479e-01 -1.01379299e+00 -2.96800047e-01
-9.12660062e-02 3.57211292e-01 9.34886456e-01 8.62629935e-02
-4.55015182e-01 4.11479264e-01 -8.53014410e-01 -1.86865374e-01
6.03476584e-01 2.69949317e-01 -5.60543478e-01 -4.28259373e-01
-1.37932789e+00 6.34430170e-01 2.56683826e-01 4.35328960e-01
-5.63049614e-01 -5.38979709e-01 -1.08978248e+00 -2.33974814e-01
5.83024502e-01 -9.90694523e-01 1.02954400e+00 -2.35523507e-01
-1.21621871e+00 3.01427782e-01 -3.18764389e-01 -7.10688680e-02
7.50242889e-01 -7.97600567e-01 -2.78645337e-01 2.75697440e-01
4.79023933e-01 7.20613241e-01 7.73739696e-01 -9.35493410e-01
-7.81797528e-01 -7.68411458e-01 -2.08083481e-01 5.28284967e-01
-4.40108031e-02 -1.00515611e-01 -8.32688570e-01 -9.05078173e-01
2.29635000e-01 -8.39354873e-01 -3.54166359e-01 5.27472682e-02
-5.65041959e-01 -4.36724842e-01 6.38207316e-01 -9.02989566e-01
1.19391906e+00 -1.82589078e+00 3.20099980e-01 3.10747623e-01
5.16989291e-01 1.84844270e-01 1.21056005e-01 1.64390653e-01
3.38613868e-01 -5.86753905e-01 5.77266216e-02 -5.00697196e-01
8.08517188e-02 1.49738155e-02 2.32783332e-01 4.85306293e-01
6.49893805e-02 1.06042516e+00 -6.75395906e-01 -8.20595205e-01
3.69134486e-01 7.78108478e-01 -5.19026279e-01 2.04016253e-01
3.40219564e-03 5.54209650e-01 -8.90799046e-01 8.32395792e-01
8.18786025e-01 -1.14040777e-01 -4.60040599e-01 -5.78857422e-01
4.99130115e-02 -2.07046092e-01 -1.65397918e+00 2.01121926e+00
-1.04061961e-01 -2.63686590e-02 -1.77793413e-01 -3.88448745e-01
1.01618838e+00 3.49696912e-03 4.30281252e-01 -4.90948379e-01
1.57194972e-01 7.91475624e-02 -4.69907314e-01 -4.93097454e-01
6.10724747e-01 2.32529968e-01 -3.58627439e-01 -7.92417377e-02
-3.41668278e-02 3.38037163e-01 -2.06163377e-02 9.73638296e-02
7.68523932e-01 4.08426046e-01 2.10471168e-01 -1.28237382e-01
7.74899542e-01 -2.86998570e-01 9.99939620e-01 3.44407320e-01
-4.21713918e-01 6.34803534e-01 9.16781425e-02 -6.70316100e-01
-7.44517624e-01 -9.99348044e-01 1.21902622e-01 1.13046527e+00
3.68429720e-01 -6.37116969e-01 -6.90101743e-01 -5.81301987e-01
3.26601982e-01 -1.55264795e-01 -6.36641860e-01 -4.29255925e-02
-8.74010146e-01 -5.87128401e-01 3.11364830e-01 8.25707257e-01
1.07389474e+00 -8.16793859e-01 -4.55070019e-01 1.58326685e-01
-5.73659062e-01 -9.47579801e-01 -8.33164752e-01 -6.60636306e-01
-4.02677059e-01 -1.12904298e+00 -1.17025483e+00 -6.92884207e-01
4.39951658e-01 3.13511461e-01 7.01379418e-01 1.57391563e-01
-4.16711897e-01 1.36792094e-01 -2.62460291e-01 -1.85734078e-01
5.84991217e-01 -5.12495115e-02 4.06953096e-01 1.11061394e-01
4.71932650e-01 -4.49959874e-01 -9.74643886e-01 2.10212559e-01
-4.49876934e-01 1.33074643e-02 7.61529744e-01 8.28801334e-01
7.87817240e-01 -3.15358378e-02 3.74312520e-01 -3.27500045e-01
5.74543715e-01 -9.84369814e-02 -1.99576005e-01 2.26299793e-01
-5.27132128e-04 -3.71047482e-02 4.06365484e-01 -3.29885781e-01
-1.27205062e+00 2.56144881e-01 -2.86919594e-01 -4.74517375e-01
-2.96346456e-01 1.95437104e-01 -5.06870329e-01 -4.38567512e-02
4.53553587e-01 3.16235632e-01 -4.64598602e-03 -8.62798929e-01
1.96548745e-01 5.79301119e-01 7.43924618e-01 -7.81473994e-01
8.78097475e-01 4.74495858e-01 5.74664026e-02 -6.10239863e-01
-7.78549612e-01 -4.69102979e-01 -8.60194147e-01 -2.00264975e-01
1.02051139e+00 -1.40738785e+00 -8.24604690e-01 7.96299398e-01
-1.18863559e+00 3.68497163e-01 8.38736910e-03 6.64878726e-01
-3.83924961e-01 4.78812188e-01 -7.74555624e-01 -7.30510354e-01
-4.27909553e-01 -1.16366243e+00 1.55594730e+00 6.90558076e-01
-7.21170232e-02 -4.24739659e-01 1.61746924e-03 4.54320163e-01
1.27424732e-01 5.94292462e-01 1.56554416e-01 -3.36008072e-01
-5.46387851e-01 -2.91363358e-01 -2.67903954e-01 -3.19965519e-02
-4.06712815e-02 -2.17517674e-01 -6.31709337e-01 -3.31583738e-01
-5.38476408e-01 -3.15017998e-01 1.02990818e+00 3.95771295e-01
1.00660491e+00 -8.18118975e-02 -6.31367743e-01 5.67162752e-01
8.71822357e-01 -4.51033443e-01 4.70940232e-01 1.65502459e-01
1.11778235e+00 4.72403824e-01 1.18512690e+00 7.20747948e-01
9.08972800e-01 9.07604396e-01 7.04410598e-02 -5.92717119e-02
-9.96265039e-02 -5.56261659e-01 -3.02644391e-02 5.87722719e-01
-5.17974079e-01 4.30975556e-01 -5.66686749e-01 2.57906497e-01
-2.12873721e+00 -8.62424731e-01 2.89098807e-02 1.86193454e+00
5.84636807e-01 3.45396921e-02 5.86337268e-01 1.20060649e-02
9.07320917e-01 3.18224132e-01 -4.31689560e-01 6.35496140e-01
-2.62736063e-02 5.38867451e-02 1.45161420e-01 2.83753067e-01
-1.41035938e+00 8.98923278e-01 5.34447956e+00 1.07920969e+00
-5.65561533e-01 1.15650102e-01 2.91252345e-01 -1.28261253e-01
-4.76052947e-02 -3.59164149e-01 -9.80006158e-01 6.43228948e-01
4.05507348e-02 1.95243999e-01 3.12172454e-02 9.71078455e-01
7.25502968e-02 -1.71026468e-01 -6.16273105e-01 1.31710041e+00
-3.79790412e-03 -1.05653119e+00 5.86898886e-02 9.79379490e-02
4.16480422e-01 -2.69449055e-01 -1.77743167e-01 3.84326488e-01
7.59870112e-02 -8.52663338e-01 6.99016690e-01 1.14863098e+00
5.76503277e-01 -1.16916251e+00 8.18889976e-01 3.25268954e-01
-1.89446461e+00 -2.10434183e-01 -5.77141523e-01 -1.19666465e-01
2.87871063e-01 7.62296200e-01 -1.69026479e-01 7.19343781e-01
9.06445384e-01 7.01651096e-01 -7.58051813e-01 1.29358351e+00
-2.08245769e-01 -1.38363346e-01 -4.38451111e-01 -1.77311629e-01
-6.68795258e-02 2.36629859e-01 4.54756171e-01 1.02373838e+00
1.29009783e-01 3.16874206e-01 8.59546065e-01 3.72290701e-01
1.97114646e-01 2.15824798e-01 3.09740081e-02 5.73922038e-01
7.52587914e-01 1.43400860e+00 -4.00081962e-01 -4.31045979e-01
-2.65776545e-01 1.20064425e+00 5.46677768e-01 2.47330815e-01
-7.97230005e-01 -5.79354703e-01 1.01103437e+00 1.27577096e-01
3.17370921e-01 -4.02308971e-01 2.89693475e-01 -1.40723193e+00
3.59503239e-01 -6.88021302e-01 5.05835354e-01 -5.90670645e-01
-1.33887780e+00 4.13743079e-01 1.17133051e-01 -1.18494439e+00
-1.51776135e-01 -3.30893993e-01 -5.13822079e-01 9.34839189e-01
-1.23459816e+00 -1.58046687e+00 -7.99096644e-01 9.64210868e-01
3.66902024e-01 -5.65853640e-02 5.03155410e-01 4.56888527e-01
-5.93385994e-01 6.69417202e-01 -4.84985054e-01 4.19645041e-01
8.29752207e-01 -9.92958307e-01 3.26673150e-01 6.22552872e-01
-3.87198210e-01 7.51788199e-01 4.65650827e-01 -1.03538442e+00
-1.30093873e+00 -1.19941866e+00 5.74976861e-01 -2.75132686e-01
1.79205220e-02 -2.90250421e-01 -6.90305352e-01 5.07295728e-01
-4.56419289e-01 3.65097255e-01 3.90832484e-01 -2.32545962e-03
-2.22997487e-01 -2.65127033e-01 -1.18348539e+00 3.98363829e-01
1.39296448e+00 -1.56564891e-01 -8.12163830e-01 2.92144090e-01
9.45212960e-01 -6.92606688e-01 -1.07880580e+00 5.05288005e-01
7.92088211e-01 -8.92921269e-01 1.28792429e+00 -3.39932233e-01
1.79234251e-01 -6.53445661e-01 -2.30418146e-01 -1.19305801e+00
-7.52135575e-01 -4.04087722e-01 -3.80312622e-01 1.10813272e+00
-3.01293731e-01 -4.82418507e-01 7.09714293e-01 4.77227092e-01
4.69501540e-02 -9.20741439e-01 -1.05737078e+00 -4.83516574e-01
-5.02569914e-01 -1.42096147e-01 1.11918414e+00 4.62731898e-01
-2.66834319e-01 2.25389659e-01 -6.79628670e-01 4.48379703e-02
9.03985023e-01 1.43346086e-01 1.11326122e+00 -1.42986119e+00
-2.12229311e-01 -5.17142266e-02 -7.85293639e-01 -1.35611188e+00
-4.15543169e-02 -3.00915122e-01 -1.68946963e-02 -1.53889501e+00
2.86664337e-01 -3.87652874e-01 -1.63763762e-01 3.25625777e-01
-6.87471628e-01 4.21613067e-01 2.90942699e-01 2.99590766e-01
-1.03287423e+00 9.38294530e-01 1.60427725e+00 1.27892002e-01
-1.27846420e-01 7.66800418e-02 -5.40487587e-01 8.56522441e-01
4.19936001e-01 3.05078514e-02 -2.44169217e-02 -1.12486571e-01
-2.99556524e-01 7.62944389e-03 7.60719717e-01 -1.57480347e+00
2.26150766e-01 -1.49537116e-01 1.36394858e+00 -1.04752946e+00
7.35643268e-01 -5.18015742e-01 3.68469581e-02 5.82971036e-01
1.07320817e-02 -1.22217871e-02 -2.11251408e-01 7.11022258e-01
-1.46226704e-01 4.36175138e-01 4.71643358e-01 -3.20732147e-01
-8.88029933e-01 8.94840956e-01 2.67271012e-01 5.10415658e-02
9.67552602e-01 -1.00494809e-01 -1.14718713e-01 -2.30460092e-01
-8.07829082e-01 5.02432227e-01 3.89738142e-01 6.85595274e-01
8.79193902e-01 -1.95110250e+00 -7.96649456e-01 2.97967613e-01
1.99656755e-01 3.26854020e-01 7.19857216e-01 8.40016842e-01
-3.61290872e-01 2.22160771e-01 -3.09957802e-01 -6.71554267e-01
-1.18898809e+00 3.09238255e-01 1.80199206e-01 -2.29249984e-01
-8.25942576e-01 8.70525420e-01 1.91991068e-02 -3.99452299e-01
8.92665312e-02 -1.48703471e-01 -3.06968719e-01 3.95687371e-02
1.01488602e+00 6.50823653e-01 -2.77467757e-01 -1.19089389e+00
-7.33624578e-01 8.64428520e-01 -6.27348945e-02 8.15143734e-02
1.28469527e+00 -2.14863002e-01 1.00507773e-02 5.34375310e-02
1.17312706e+00 3.47313397e-02 -1.58694744e+00 -6.14046693e-01
-5.30561864e-01 -8.46155167e-01 -2.63891309e-01 -4.23258513e-01
-1.02283192e+00 7.69436598e-01 6.47111952e-01 -3.75575781e-01
9.95936632e-01 6.64187893e-02 1.20597267e+00 3.20630163e-01
5.93776345e-01 -1.33465087e+00 2.29404882e-01 4.29987282e-01
9.82078195e-01 -1.01143181e+00 3.76580715e-01 -6.16527021e-01
-7.29852438e-01 9.59601879e-01 1.04549921e+00 -4.35770154e-01
4.56357002e-01 -3.53374332e-02 -2.97262192e-01 -1.96379572e-01
-2.58687526e-01 -4.82964694e-01 6.17833436e-01 8.22583616e-01
1.52171910e-01 3.45715135e-01 -2.81359524e-01 1.17164075e+00
-3.80910337e-01 8.60009044e-02 -2.25324363e-01 9.00403142e-01
-8.17926288e-01 -7.37810254e-01 -8.73791397e-01 5.59526742e-01
-1.60322174e-01 2.76291281e-01 -7.37646520e-02 6.95971370e-01
5.38234293e-01 6.74427748e-01 -2.92958301e-02 -6.87298417e-01
5.58153987e-01 -1.36440068e-01 5.49995840e-01 -2.81890243e-01
-3.09209585e-01 2.14099661e-01 6.05480373e-02 -1.02853739e+00
-4.16665733e-01 -8.35547984e-01 -1.34131181e+00 -2.74127722e-01
-1.61766067e-01 3.26791033e-02 1.09909117e-01 1.13122928e+00
4.30393696e-01 6.40020728e-01 2.06319273e-01 -1.39195001e+00
-3.89402330e-01 -1.11573207e+00 -4.80054498e-01 5.60320735e-01
2.41751105e-01 -1.37800872e+00 3.03842902e-01 -4.40602005e-01]
|
[7.14971399307251, -0.7900003790855408]
|
4028744c-0922-4a0d-b247-3b3dc1c0967a
|
from-depth-data-to-head-pose-estimation-a
|
1703.03624
| null |
http://arxiv.org/abs/1703.03624v1
|
http://arxiv.org/pdf/1703.03624v1.pdf
|
From Depth Data to Head Pose Estimation: a Siamese approach
|
The correct estimation of the head pose is a problem of the great importance
for many applications. For instance, it is an enabling technology in automotive
for driver attention monitoring. In this paper, we tackle the pose estimation
problem through a deep learning network working in regression manner.
Traditional methods usually rely on visual facial features, such as facial
landmarks or nose tip position. In contrast, we exploit a Convolutional Neural
Network (CNN) to perform head pose estimation directly from depth data. We
exploit a Siamese architecture and we propose a novel loss function to improve
the learning of the regression network layer. The system has been tested on two
public datasets, Biwi Kinect Head Pose and ICT-3DHP database. The reported
results demonstrate the improvement in accuracy with respect to current
state-of-the-art approaches and the real time capabilities of the overall
framework.
|
['Guido Borghi', 'Roberto Vezzani', 'Rita Cucchiara', 'Marco Venturelli']
|
2017-03-10
| null | null | null | null |
['head-pose-estimation', 'driver-attention-monitoring']
|
['computer-vision', 'computer-vision']
|
[-3.05798650e-01 2.29899302e-01 -6.12510741e-02 -8.09688151e-01
-8.21363509e-01 4.81194444e-02 4.99115527e-01 -2.14185581e-01
-1.07166290e+00 6.29857898e-01 1.09655631e-03 5.32880947e-02
5.69321625e-02 -3.93681020e-01 -7.32205331e-01 -6.89011216e-01
1.79084629e-01 5.51534712e-01 1.34690329e-01 -2.38205269e-01
2.43107021e-01 8.96541178e-01 -1.85803401e+00 -3.33362937e-01
2.66163856e-01 1.38585615e+00 -2.84893274e-01 3.33435267e-01
1.64472342e-01 4.62206900e-01 -3.72732610e-01 -4.30440158e-01
2.76362985e-01 8.06871131e-02 -4.11500931e-01 -6.44921735e-02
6.70224190e-01 -2.60417670e-01 -3.75997901e-01 9.38885152e-01
8.88921320e-01 1.89989105e-01 4.35478330e-01 -1.34480178e+00
3.14229369e-01 -1.44899145e-01 -6.25012219e-01 1.18855409e-01
4.04447019e-01 1.03427000e-01 5.78554749e-01 -8.32128763e-01
4.54748183e-01 1.03156710e+00 7.85431504e-01 7.91015983e-01
-7.02688694e-01 -8.99078369e-01 -7.08614066e-02 6.80132210e-01
-1.67219424e+00 -8.59537363e-01 8.92730892e-01 -2.45556921e-01
7.32305467e-01 -3.38525921e-02 4.91409838e-01 8.62385392e-01
1.93167314e-01 6.57022417e-01 9.79152918e-01 -1.55430198e-01
2.76098251e-01 -1.16986316e-02 -1.03965662e-01 8.60026062e-01
1.00633346e-01 1.31228164e-01 -7.83230066e-01 2.60577500e-01
3.68398547e-01 5.80235310e-02 -1.36655390e-01 -5.59037685e-01
-4.65084970e-01 9.01635289e-01 7.33125448e-01 -2.90171779e-03
-5.52812636e-01 2.48851374e-01 4.13244694e-01 -6.93258345e-02
5.87012827e-01 -9.61284190e-02 -3.89262050e-01 -3.53608370e-01
-1.17734158e+00 4.56245780e-01 8.23459446e-01 6.62705243e-01
5.65644622e-01 -5.56274503e-02 8.39737132e-02 6.61495864e-01
7.01625764e-01 4.45253193e-01 6.41874552e-01 -6.68980718e-01
4.36134547e-01 5.02655387e-01 -2.32554898e-01 -7.75507629e-01
-8.81478429e-01 -2.81979322e-01 -4.49279785e-01 6.64641500e-01
4.22442406e-01 -1.96701646e-01 -9.31009710e-01 1.51296675e+00
6.93582654e-01 2.42184579e-01 -2.72641599e-01 1.12655342e+00
1.12685883e+00 7.82128572e-02 -9.42218825e-02 5.93099631e-02
1.33197534e+00 -9.34991717e-01 -6.64735019e-01 -3.12997729e-01
2.00128838e-01 -5.28664649e-01 5.44316769e-01 6.00876808e-01
-1.08761644e+00 -4.86713946e-01 -1.05858767e+00 -1.16651103e-01
-5.59506536e-01 2.46488839e-01 2.58789957e-01 8.66291106e-01
-1.00306940e+00 4.13549960e-01 -1.04266596e+00 -3.60731959e-01
5.34143507e-01 9.67523694e-01 -8.02050114e-01 -4.30416763e-02
-7.91915536e-01 1.08318913e+00 1.12528712e-01 6.53223693e-01
-5.66760600e-01 -4.61711794e-01 -1.05310488e+00 -1.52647540e-01
2.64334381e-01 -3.23647976e-01 1.30882442e+00 -6.58122182e-01
-2.02008295e+00 1.09647405e+00 -3.69508654e-01 -5.93962550e-01
8.64769161e-01 -5.29070854e-01 -7.60377198e-02 -3.92397270e-02
-2.47771546e-01 6.83832765e-01 9.66972351e-01 -8.00162017e-01
-6.37511075e-01 -8.41129065e-01 -2.03389898e-01 -1.40969558e-02
-2.30128378e-01 1.22296266e-01 -6.44883335e-01 -3.30143347e-02
1.80850163e-01 -1.16747546e+00 -4.98685613e-02 1.82399988e-01
-4.34326887e-01 -3.62831473e-01 8.56448531e-01 -8.05209696e-01
7.69369185e-01 -1.94735169e+00 1.37382105e-01 1.73442915e-01
3.20029140e-01 4.23026055e-01 2.72509009e-01 -1.77164838e-01
-1.43701851e-01 -5.62424123e-01 -1.27148390e-01 -1.01586735e+00
9.13612619e-02 1.15801610e-01 3.90936196e-01 1.06948805e+00
1.72323152e-01 8.57716322e-01 -3.71772766e-01 -4.05391783e-01
3.00627828e-01 9.12071288e-01 -5.01517892e-01 3.65200251e-01
1.75741017e-01 5.37774444e-01 -1.22633360e-01 4.95187372e-01
8.94609213e-01 5.57161033e-01 -3.55583727e-01 -2.01935858e-01
-2.02356711e-01 1.64804608e-01 -1.06463742e+00 1.73842072e+00
-4.90698844e-01 6.92056894e-01 3.23180109e-01 -8.06763768e-01
9.42686141e-01 2.85241485e-01 3.55595171e-01 -8.48568857e-01
4.91476417e-01 2.40799844e-01 -1.94194783e-02 -7.02579200e-01
3.23221833e-01 -3.17066044e-01 8.39504227e-02 7.29368925e-02
1.97004825e-01 3.11561991e-02 -2.97614485e-01 -6.33908272e-01
7.83154309e-01 1.75690547e-01 2.86946356e-01 1.00094983e-02
1.00892341e+00 -5.03200769e-01 4.53459084e-01 -3.69626353e-03
-5.62207222e-01 8.07240486e-01 4.76193190e-01 -3.92733455e-01
-6.92628741e-01 -6.75305784e-01 -1.76882058e-01 9.95877266e-01
-2.91796029e-01 2.34475769e-02 -1.02484560e+00 -7.18925834e-01
9.95011032e-02 4.34409171e-01 -9.11835790e-01 -2.16936827e-01
-8.89348924e-01 -4.73583430e-01 6.35603786e-01 7.76510179e-01
5.22250235e-01 -1.03208804e+00 -8.11127782e-01 -9.52664316e-02
2.68928140e-01 -1.40201521e+00 -2.84830004e-01 1.95680842e-01
-6.61155939e-01 -9.42208946e-01 -7.89473057e-01 -5.22892892e-01
4.83337194e-01 -3.12832981e-01 7.84972787e-01 -1.06164411e-01
-2.98089236e-01 2.72062302e-01 -2.34692320e-02 -7.06173897e-01
1.14650451e-01 4.19447422e-01 1.89145982e-01 2.51396269e-01
7.68050790e-01 -3.60828191e-01 -7.40846038e-01 9.69443023e-02
-5.19432127e-01 -5.19421101e-01 5.95575571e-01 5.47387838e-01
3.92721653e-01 -3.54988396e-01 4.81251836e-01 -7.01195896e-01
3.75317782e-01 -2.66039521e-01 -8.33088994e-01 -2.96363443e-01
-4.34919447e-01 1.16196059e-01 2.59754121e-01 -3.32603836e-03
-8.50460947e-01 5.88921726e-01 -8.38054121e-01 -5.25125086e-01
-4.29044276e-01 2.48149604e-01 -4.02463704e-01 -4.12279427e-01
2.30350316e-01 -3.16086859e-02 3.43144774e-01 -5.55631697e-01
8.73330049e-04 6.05610847e-01 7.10506856e-01 -1.37168646e-01
6.86934471e-01 4.86778915e-01 5.28255045e-01 -8.46413910e-01
-6.84299827e-01 -7.76058614e-01 -1.02304733e+00 -2.12528810e-01
1.02805686e+00 -8.00234497e-01 -1.18268776e+00 6.17845476e-01
-1.16539013e+00 4.48755398e-02 1.27235457e-01 5.54490924e-01
-5.83216846e-01 5.87089956e-02 -1.19581670e-01 -9.34411764e-01
-4.41953182e-01 -1.35290051e+00 1.38930559e+00 3.94313544e-01
3.51040661e-02 -9.23539102e-01 1.22321770e-01 4.92750287e-01
4.64353651e-01 5.21747828e-01 1.62737608e-01 -6.81261122e-01
-2.20526233e-01 -7.88051188e-01 -5.73624969e-02 2.66665787e-01
-1.30553395e-01 -2.91358024e-01 -1.57970309e+00 -2.95802534e-01
4.63943519e-02 -3.65282685e-01 7.54647911e-01 4.09571707e-01
1.11564469e+00 2.30424814e-02 -8.27095658e-02 8.31114471e-01
1.23109770e+00 -6.22427203e-02 6.76099539e-01 4.27675128e-01
7.53902853e-01 8.86990488e-01 5.07652700e-01 5.37876904e-01
6.54165864e-01 1.00511444e+00 7.25263119e-01 -9.07372311e-03
-5.36390282e-02 -7.29085580e-02 2.73448437e-01 4.82520968e-01
-3.29144478e-01 3.92542005e-01 -9.32652235e-01 3.13059837e-01
-1.74983537e+00 -5.21187007e-01 2.88940370e-02 2.42637563e+00
5.12747347e-01 2.20346764e-01 4.05043215e-01 3.52253139e-01
3.03626090e-01 9.09524504e-03 -6.13627791e-01 -4.29340541e-01
4.17554885e-01 5.59634387e-01 6.67645454e-01 4.66314852e-01
-1.15357172e+00 9.33309317e-01 5.48368597e+00 4.30459678e-01
-1.65254498e+00 3.54690135e-01 3.31689805e-01 -4.32770312e-01
5.09455025e-01 -7.43501008e-01 -1.15910351e+00 2.47064456e-01
1.09475529e+00 2.63298959e-01 5.86515255e-02 9.56761718e-01
3.42764705e-01 -1.33440435e-01 -1.15805328e+00 1.29305112e+00
5.50248981e-01 -7.44961917e-01 -6.95492089e-01 2.02516824e-01
1.41517103e-01 1.69886917e-01 1.20210379e-01 2.77014375e-01
-4.63639140e-01 -1.40215981e+00 7.33321369e-01 6.03034735e-01
6.48919702e-01 -1.16131485e+00 1.10560870e+00 3.09402496e-01
-1.18177104e+00 3.99449421e-03 -2.19442919e-01 -8.07627738e-02
1.68562397e-01 1.67800546e-01 -1.00635278e+00 2.11275458e-01
7.38863051e-01 3.34929585e-01 -7.47882783e-01 1.31430924e+00
-3.63802522e-01 3.42282772e-01 -3.65214884e-01 -5.84607907e-02
1.59192264e-01 1.44666314e-01 3.45039099e-01 1.12251341e+00
1.32318005e-01 -2.36457512e-01 -1.19497135e-01 5.58879137e-01
-1.44035831e-01 5.12488149e-02 -6.75792277e-01 5.06448627e-01
-5.11432700e-02 1.39253485e+00 -4.60607558e-01 1.69890210e-01
-3.74525875e-01 7.88652480e-01 4.04392868e-01 -4.16850671e-02
-9.22035158e-01 -3.88808280e-01 9.01648760e-01 3.16174418e-01
3.77859145e-01 -3.06041151e-01 -2.47446522e-01 -6.93270624e-01
1.90230951e-01 -5.90216279e-01 1.45626292e-02 -3.38790894e-01
-8.24297726e-01 7.37350643e-01 -9.18476135e-02 -1.07841516e+00
-5.12358308e-01 -1.00122190e+00 -6.91369534e-01 8.61557901e-01
-1.95697498e+00 -1.22989619e+00 -5.81756830e-01 7.05425024e-01
4.66526955e-01 -1.03669614e-01 4.93757814e-01 6.09130442e-01
-7.50764072e-01 1.07321846e+00 -3.74257565e-01 2.58482516e-01
6.51607513e-01 -1.07457721e+00 3.59314591e-01 5.03716826e-01
-2.09339499e-01 4.26654726e-01 7.71072567e-01 -2.04893202e-01
-1.50813007e+00 -9.31268036e-01 9.12603259e-01 -3.71208668e-01
1.97016031e-01 -4.34656799e-01 -6.59040034e-01 4.77968514e-01
1.22353122e-01 4.41809505e-01 5.22377610e-01 -6.90050721e-02
-1.60406426e-01 -4.12780136e-01 -1.42141449e+00 5.24562411e-02
7.08107531e-01 -4.39333439e-01 -4.44445044e-01 9.78820026e-02
1.62000850e-01 -7.48193085e-01 -5.69204152e-01 3.70871276e-01
8.79119158e-01 -1.10018492e+00 9.03863966e-01 -5.51173627e-01
7.66612515e-02 -1.65041655e-01 3.91393527e-02 -1.09189391e+00
1.96759373e-01 -3.90707493e-01 -1.91484898e-01 9.55217421e-01
2.46150166e-01 -6.78628147e-01 1.31799650e+00 8.21273446e-01
3.67392902e-03 -9.19321835e-01 -1.45482516e+00 -4.32903916e-01
2.30811629e-03 -5.06650031e-01 4.69144702e-01 2.81401426e-01
-3.20160419e-01 2.76786685e-01 -1.72256887e-01 1.32667735e-01
7.26639509e-01 -5.09429991e-01 9.55185294e-01 -1.47477531e+00
1.15111127e-01 -4.35147852e-01 -1.09991539e+00 -7.20269322e-01
6.28976583e-01 -4.76518840e-01 4.03192312e-01 -1.13042068e+00
-1.24009661e-01 -4.84369285e-02 -2.18621060e-01 3.40056717e-01
7.93139488e-02 5.39097190e-01 1.27499446e-01 -3.80403727e-01
-4.72651631e-01 6.84088528e-01 8.38473499e-01 6.95685595e-02
-5.27996682e-02 4.55271870e-01 -2.45544657e-01 9.22627032e-01
7.01448500e-01 -4.52986926e-01 -2.64467504e-02 -2.55360663e-01
-2.50857547e-02 -7.48631135e-02 4.75403994e-01 -1.32727182e+00
7.19206035e-01 3.27644318e-01 5.04327655e-01 -6.43738449e-01
8.49682570e-01 -9.36363280e-01 -3.94318163e-01 4.67026591e-01
-4.49354500e-02 2.36964419e-01 3.61113548e-01 2.29389235e-01
-3.75302166e-01 -2.15548888e-01 1.00623691e+00 1.79054141e-01
-6.62569463e-01 4.82332408e-01 -4.00095480e-03 -1.63057745e-01
1.01461363e+00 -3.04750204e-01 1.73234448e-01 -4.41796422e-01
-5.31894922e-01 5.02292886e-02 2.03601971e-01 4.66390580e-01
8.86709869e-01 -1.25066829e+00 -5.95198274e-01 5.14639318e-01
1.36662424e-01 7.96472728e-02 -1.82618588e-01 1.21283424e+00
-5.84901392e-01 5.09815454e-01 -3.16707790e-01 -7.09939957e-01
-1.56114304e+00 1.97562277e-01 7.53482401e-01 1.04016878e-01
-4.70488608e-01 9.90079284e-01 -4.42573242e-02 -6.06011569e-01
6.28541350e-01 -2.80382276e-01 -6.53089464e-01 1.47566482e-01
6.90458357e-01 4.55177873e-01 7.32658148e-01 -1.24948668e+00
-7.01805890e-01 8.46821368e-01 1.51685560e-02 -1.51801914e-01
1.48744190e+00 7.48544233e-03 2.81648580e-02 2.33211920e-01
1.62152672e+00 -9.18460414e-02 -1.32260168e+00 -7.82257225e-03
2.25761950e-01 -3.26861262e-01 3.90162349e-01 -3.12313676e-01
-1.44665682e+00 1.20724535e+00 1.13446343e+00 -3.79314154e-01
9.30564880e-01 -1.35640040e-01 8.18249166e-01 5.07296205e-01
2.81945437e-01 -1.01886141e+00 -1.48977429e-01 5.17541647e-01
8.45402896e-01 -1.59923410e+00 -6.71575591e-02 -1.76822647e-01
-4.79467571e-01 1.22910190e+00 7.15995967e-01 -2.40201592e-01
9.40121830e-01 2.80233949e-01 3.04479033e-01 -2.77269661e-01
-3.28483880e-01 -5.52234411e-01 6.02690578e-01 5.19181430e-01
7.05456436e-01 -1.40659928e-01 -1.53844222e-01 3.86503428e-01
-4.23140764e-01 2.23898605e-01 5.53591289e-02 8.98677111e-01
-2.96264082e-01 -9.48072314e-01 -4.60028619e-01 1.34981886e-01
-6.27219975e-01 3.46434802e-01 -3.35474670e-01 8.25193882e-01
2.66839355e-01 7.79914618e-01 -1.97517034e-02 -2.84015536e-01
7.00953603e-01 2.23756522e-01 5.11635303e-01 -4.65049565e-01
-6.27180219e-01 -1.32643119e-01 -9.08685252e-02 -8.84066403e-01
-4.96044189e-01 -7.70134151e-01 -1.16252983e+00 -1.60956234e-01
-1.90448537e-01 -1.93350986e-01 1.19609475e+00 1.14464116e+00
4.99458648e-02 5.37521422e-01 5.67539036e-01 -1.34854627e+00
-5.48041403e-01 -1.02893758e+00 -5.13962984e-01 1.60543516e-01
7.56608367e-01 -1.07788551e+00 -6.95476755e-02 -2.60597438e-01]
|
[13.659453392028809, 0.28994593024253845]
|
b34cb75c-4f01-48be-a44c-fdd404ac20e8
|
idiapers-causal-news-corpus-2022-extracting
|
2209.03891
| null |
https://arxiv.org/abs/2209.03891v2
|
https://arxiv.org/pdf/2209.03891v2.pdf
|
IDIAPers @ Causal News Corpus 2022: Extracting Cause-Effect-Signal Triplets via Pre-trained Autoregressive Language Model
|
In this paper, we describe our shared task submissions for Subtask 2 in CASE-2022, Event Causality Identification with Casual News Corpus. The challenge focused on the automatic detection of all cause-effect-signal spans present in the sentence from news-media. We detect cause-effect-signal spans in a sentence using T5 -- a pre-trained autoregressive language model. We iteratively identify all cause-effect-signal span triplets, always conditioning the prediction of the next triplet on the previously predicted ones. To predict the triplet itself, we consider different causal relationships such as cause$\rightarrow$effect$\rightarrow$signal. Each triplet component is generated via a language model conditioned on the sentence, the previous parts of the current triplet, and previously predicted triplets. Despite training on an extremely small dataset of 160 samples, our approach achieved competitive performance, being placed second in the competition. Furthermore, we show that assuming either cause$\rightarrow$effect or effect$\rightarrow$cause order achieves similar results.
|
['Pavel Smrz', 'Petr Motlicek', 'Sergio Burdisso', 'Esaú Villatoro-Tello', 'Juan Zuluaga-Gomez', 'Muskaan Singh', 'Martin Fajcik']
|
2022-09-08
| null | null | null | null |
['event-causality-identification']
|
['natural-language-processing']
|
[ 2.85323262e-01 1.49598196e-01 -7.44069293e-02 -3.79934788e-01
-1.17667699e+00 -6.03411913e-01 7.41048872e-01 6.24137938e-01
-2.92819530e-01 1.02003789e+00 7.94020772e-01 -4.60338384e-01
-1.21479735e-01 -5.72295666e-01 -1.08683527e+00 -2.79015899e-01
-6.00753367e-01 3.06435049e-01 2.66804665e-01 -6.91470727e-02
2.16992229e-01 -1.57978069e-02 -1.23488116e+00 9.46237266e-01
1.46205455e-01 7.86452770e-01 1.14107870e-01 8.34816873e-01
3.86466160e-02 1.43197048e+00 -6.87466681e-01 -2.65856475e-01
-1.31228760e-01 -8.29237223e-01 -8.88557851e-01 -5.20251393e-01
3.23356450e-01 6.51955977e-02 -2.81916708e-01 4.61500943e-01
3.01364839e-01 1.06239691e-01 7.58729696e-01 -1.05835748e+00
-1.11322284e-01 1.28299332e+00 -6.68555021e-01 9.04258609e-01
8.92036498e-01 -1.79660603e-01 1.48429179e+00 -9.26499546e-01
1.00559628e+00 1.12384009e+00 6.60870314e-01 -5.88281825e-03
-1.37935495e+00 -7.11758733e-01 4.58261043e-01 3.80652070e-01
-1.09800136e+00 -2.07705751e-01 5.87867081e-01 -7.20729172e-01
1.56420588e+00 4.02545869e-01 4.09181178e-01 1.28874445e+00
5.96380949e-01 6.73364937e-01 7.52858698e-01 -4.04980361e-01
1.00630030e-01 -4.58751231e-01 4.29347157e-01 4.22224492e-01
-4.16209430e-01 9.72153805e-03 -9.51696277e-01 -4.62265521e-01
1.51364535e-01 -4.70449895e-01 -4.48684618e-02 5.31832814e-01
-1.41046786e+00 6.39939189e-01 2.05470011e-01 2.91902035e-01
-7.50764668e-01 5.21379769e-01 6.26628160e-01 1.89034939e-01
6.43779933e-01 1.86164930e-01 -8.51343572e-01 -1.87432930e-01
-9.96310532e-01 5.85324526e-01 7.06706464e-01 6.90651000e-01
3.19432527e-01 -3.73490453e-01 -6.09562278e-01 5.35597384e-01
1.14608712e-01 1.10809647e-01 5.21816649e-02 -4.54555452e-01
7.64502287e-01 3.37312609e-01 2.06458509e-01 -8.97552490e-01
-8.86662304e-01 -3.80341321e-01 -4.01351422e-01 -4.34041977e-01
6.26923919e-01 -6.06640697e-01 -4.84140158e-01 1.90712953e+00
2.36504108e-01 4.86342907e-01 -2.83502132e-01 5.99875748e-01
6.87135994e-01 1.09978819e+00 4.48206455e-01 -7.53200650e-01
1.54057288e+00 -2.92683035e-01 -7.30314672e-01 -2.54491061e-01
6.99482620e-01 -1.03429377e+00 7.50416815e-01 5.01828313e-01
-9.27657723e-01 -1.66906595e-01 -7.51547098e-01 1.74720325e-02
-1.08104363e-01 8.36310983e-02 4.22915906e-01 -9.26143229e-02
-5.88375211e-01 4.42323864e-01 -3.17505568e-01 -4.98030037e-02
-2.21470427e-02 -1.67675287e-01 -9.47601274e-02 2.99261332e-01
-1.76952684e+00 8.09419513e-01 3.24217469e-01 -1.91527888e-01
-1.15535092e+00 -1.23813224e+00 -6.54551387e-01 3.71167436e-02
5.83458722e-01 -3.90681177e-01 1.36686552e+00 -3.31395745e-01
-5.97884476e-01 8.67682278e-01 -6.53017282e-01 -9.06316221e-01
4.01323825e-01 -4.79255259e-01 -7.95914114e-01 -1.09242156e-01
4.06922996e-01 1.06422290e-01 6.69992924e-01 -7.49628365e-01
-8.81183684e-01 -1.49080247e-01 -3.43541741e-01 -1.33373320e-01
3.70240480e-01 8.29797745e-01 -1.42703518e-01 -7.05681980e-01
8.34340453e-02 -6.37400031e-01 -1.75721630e-01 -7.88797736e-01
-8.98064613e-01 -5.07663250e-01 1.91140249e-01 -8.29119563e-01
1.63628864e+00 -1.98385644e+00 -1.28061160e-01 6.57963976e-02
3.73089500e-02 -5.32153547e-01 1.30625501e-01 7.84153879e-01
-7.59599209e-01 8.53862166e-02 8.26650783e-02 -1.10661224e-01
-4.24560048e-02 -2.41299480e-01 -9.37673271e-01 4.08004820e-01
3.38489234e-01 4.81430560e-01 -1.08517885e+00 -4.12082285e-01
-1.95900425e-01 5.05787916e-02 -3.89434665e-01 4.04777899e-02
-6.61766171e-01 3.11337888e-01 -3.22901338e-01 3.27355452e-02
1.09028302e-01 -2.08525971e-01 2.22479820e-01 -1.32701039e-01
-4.00033295e-01 1.03680944e+00 -1.03084993e+00 1.24215317e+00
-2.42838427e-01 7.44404674e-01 -2.88191378e-01 -8.92141700e-01
8.44962656e-01 6.95030391e-01 5.91396809e-01 -6.12592816e-01
-9.90208797e-03 6.49206266e-02 1.22198939e-01 -6.21734023e-01
3.88532043e-01 -5.01820326e-01 -5.76777756e-01 5.79401016e-01
-4.78863763e-03 9.56518501e-02 6.72681212e-01 5.13496876e-01
1.49388611e+00 6.64968714e-02 3.98173869e-01 -1.01234049e-01
2.17239931e-01 -9.05726328e-02 6.11554563e-01 9.23654020e-01
1.06396221e-01 6.93893135e-01 1.30234063e+00 -5.96823812e-01
-8.59658659e-01 -1.04293442e+00 -9.72243324e-02 1.20452082e+00
-3.38345230e-01 -8.61860693e-01 -2.99453437e-01 -7.21118927e-01
-1.93566173e-01 1.49188530e+00 -8.54411304e-01 2.02339172e-01
-9.61886704e-01 -1.03287613e+00 6.75549448e-01 4.46085185e-01
-1.67837933e-01 -1.19164503e+00 -5.93544006e-01 6.58773661e-01
-7.46949255e-01 -9.38353062e-01 -4.79574114e-01 5.13231933e-01
-2.12498635e-01 -1.14238143e+00 -1.91939846e-01 -3.88333946e-01
1.02087013e-01 -5.55320084e-01 1.40273786e+00 -4.54459608e-01
-3.55064809e-01 -1.16759382e-01 -2.36342818e-01 -6.05353475e-01
-4.00082558e-01 -3.64811450e-01 -1.15748018e-01 7.96760023e-02
3.29560459e-01 -3.96506071e-01 -4.09066349e-01 4.04587090e-02
-4.25005287e-01 -1.32979661e-01 1.12017795e-01 6.96727097e-01
6.76701665e-01 1.60060331e-01 6.85994744e-01 -7.13777006e-01
5.22180140e-01 -8.35932791e-01 -3.36510777e-01 5.33474833e-02
-2.04839915e-01 -1.98198557e-02 6.73939049e-01 -3.18487316e-01
-1.03067803e+00 -1.08384252e-01 -2.43465990e-01 1.69348493e-01
-2.24563599e-01 1.00731385e+00 5.66170663e-02 1.35370934e+00
9.47365344e-01 4.39186729e-02 -6.42229974e-01 -3.82655770e-01
2.50490338e-01 2.01409068e-02 6.21404946e-01 -3.21371078e-01
1.80181399e-01 2.43093431e-01 -5.84446155e-02 -4.51237589e-01
-1.06693912e+00 -3.70147526e-01 -2.89084285e-01 -3.89939576e-01
8.72985542e-01 -1.04244673e+00 -5.87559879e-01 2.18045581e-02
-1.59405148e+00 -2.95412749e-01 -7.14966878e-02 6.04464114e-01
-3.25120240e-01 -1.01811059e-01 -6.56507194e-01 -8.86353314e-01
-1.82995930e-01 -6.77967489e-01 8.86773050e-01 -2.77402639e-01
-1.05559349e+00 -6.80111945e-01 1.31593183e-01 6.84029981e-02
-3.24883103e-01 4.64022547e-01 9.50162232e-01 -9.97153282e-01
-1.12171322e-01 -5.04194856e-01 -6.59988374e-02 -3.14285994e-01
-2.22761735e-01 1.38020605e-01 -5.76373041e-01 3.13365906e-01
-8.36692229e-02 -1.75015330e-02 9.79291260e-01 7.27563560e-01
9.21521962e-01 -3.85780215e-01 -5.83867013e-01 -2.22291797e-01
8.82963777e-01 1.38672903e-01 4.94336396e-01 2.58147359e-01
3.23175520e-01 6.59013033e-01 4.66470778e-01 7.97127724e-01
3.82505357e-01 6.11677170e-01 2.75413871e-01 1.16713829e-01
-4.79381196e-02 -5.40883839e-01 5.29071510e-01 -2.78259683e-02
2.51940101e-01 -5.23354232e-01 -1.04408109e+00 8.66357088e-01
-1.59708631e+00 -1.38951743e+00 -1.01415598e+00 1.93880880e+00
1.14242852e+00 6.56297266e-01 2.09735602e-01 1.24358326e-01
8.13144088e-01 2.67900914e-01 1.08865416e-02 -4.23915952e-01
-2.55080879e-01 4.33990313e-03 2.11752892e-01 5.76248944e-01
-1.14985335e+00 5.86241007e-01 6.52874660e+00 8.54795516e-01
-7.58065820e-01 1.52952343e-01 7.29883611e-01 -4.98137385e-01
-4.08420771e-01 1.51501447e-01 -9.83886123e-01 7.62980402e-01
1.17408180e+00 -2.12159157e-01 -6.07943013e-02 3.97240043e-01
8.75825405e-01 -4.82738972e-01 -1.27652574e+00 4.65729713e-01
-5.30837514e-02 -1.47100115e+00 -2.37632424e-01 -1.44128069e-01
6.03173196e-01 1.25180468e-01 -2.33204409e-01 3.18459421e-01
4.04989898e-01 -8.15083802e-01 1.14995575e+00 5.36393344e-01
5.73221862e-01 -6.04981601e-01 5.19956231e-01 4.96021032e-01
-1.00147378e+00 -8.47020373e-02 1.85695797e-01 -5.89435101e-01
7.01262653e-01 1.27041531e+00 -1.12978458e+00 6.33192778e-01
6.43869042e-01 6.29320264e-01 -7.27355555e-02 6.75396144e-01
-6.23295844e-01 1.33530450e+00 -3.64655077e-01 -2.06872791e-01
-2.31950823e-02 3.78155053e-01 9.44477737e-01 1.66943419e+00
1.78390846e-01 3.36234212e-01 9.17937234e-02 9.04567838e-01
-2.98085772e-02 1.32227287e-01 -3.76849174e-01 1.97626799e-01
2.89310783e-01 7.76248693e-01 -8.72821331e-01 -6.22421622e-01
-1.77932367e-01 4.49033499e-01 2.23754242e-01 2.13678494e-01
-1.15475202e+00 -1.40259832e-01 7.37176463e-02 2.56740302e-01
2.27500975e-01 1.49725571e-01 -5.24525046e-01 -9.17475104e-01
-9.09212306e-02 -5.19431770e-01 9.20253694e-01 -1.01561344e+00
-1.20022857e+00 5.25383472e-01 2.74163246e-01 -9.22854841e-01
-5.12054324e-01 -7.11333081e-02 -9.05431926e-01 1.04639852e+00
-8.47773015e-01 -7.20739543e-01 4.85992491e-01 2.92243481e-01
5.82228065e-01 2.08153725e-01 7.94325709e-01 2.01876029e-01
-5.12286782e-01 2.61719853e-01 -3.90740126e-01 1.82444826e-01
8.80769849e-01 -1.26200676e+00 3.13235253e-01 1.08539855e+00
2.89357781e-01 4.14717704e-01 1.34545898e+00 -1.16163325e+00
-7.02975929e-01 -8.61885488e-01 1.87114358e+00 -6.59864962e-01
1.14179587e+00 -4.28376883e-01 -6.90513194e-01 1.00822985e+00
3.06953639e-01 -3.15732718e-01 5.74935973e-01 6.21700168e-01
-5.44233024e-01 4.82292213e-02 -5.91990173e-01 4.54126507e-01
8.00710261e-01 -2.80402303e-01 -9.32364285e-01 5.25133610e-01
7.22179413e-01 -4.87212658e-01 -5.80705285e-01 3.53671372e-01
3.95510823e-01 -8.32794189e-01 9.08006847e-01 -7.22281992e-01
1.00805533e+00 -2.67718166e-01 -4.75337319e-02 -1.15319169e+00
-3.82806480e-01 -6.96891963e-01 -7.34354928e-02 1.33746433e+00
1.14481139e+00 -1.42651871e-01 2.49767646e-01 4.41050112e-01
-3.46225649e-01 -5.32045484e-01 -1.13863695e+00 -4.58768785e-01
9.20381173e-02 -1.10525155e+00 3.04385841e-01 7.69824982e-01
4.14230496e-01 7.93302596e-01 -5.73714316e-01 2.55375415e-01
4.10786510e-01 2.49836266e-01 3.20417523e-01 -8.28533292e-01
-4.77201521e-01 -4.02705371e-01 2.95477271e-01 -8.31877947e-01
2.46856804e-03 -8.16045225e-01 3.26693892e-01 -1.50297737e+00
4.05073792e-01 1.69262860e-03 -3.90130401e-01 5.78875422e-01
-5.40796578e-01 -9.11258608e-02 -2.88558099e-02 7.80543254e-04
-6.02003634e-01 1.59541607e-01 7.22915292e-01 -7.05899522e-02
-2.28882983e-01 3.40234451e-02 -3.98937464e-01 7.69826829e-01
5.73151469e-01 -8.36849332e-01 -1.63461819e-01 -1.36051163e-01
7.24265814e-01 4.94118392e-01 5.67452967e-01 -5.54145932e-01
1.74717098e-01 -1.77180260e-01 4.65193987e-01 -1.16037190e+00
-2.59195305e-02 -2.32335761e-01 4.06792432e-01 3.25270116e-01
-8.45173061e-01 1.39298126e-01 3.23069304e-01 6.50280595e-01
-1.18641287e-01 -3.30125503e-02 1.78215235e-01 3.26225795e-02
-4.35367703e-01 -2.89337695e-01 -8.38204622e-01 1.85972854e-01
8.83365989e-01 4.88217562e-01 -2.81967878e-01 -3.58443797e-01
-1.02748418e+00 1.97073132e-01 -5.37914515e-01 4.97366369e-01
4.03608948e-01 -9.70713079e-01 -1.29266012e+00 -2.30600253e-01
-5.47421398e-04 -3.19603354e-01 5.33392608e-01 1.10500753e+00
6.87822774e-02 4.30785984e-01 5.56199312e-01 -3.38661134e-01
-1.14775479e+00 6.60048664e-01 6.24258704e-02 -8.58232677e-01
-5.47242999e-01 1.26074994e+00 2.08318830e-01 1.62980985e-02
-2.20280550e-02 -5.35905600e-01 -2.23167315e-01 4.56178039e-01
8.50639641e-01 2.68557847e-01 6.60471097e-02 -4.55472529e-01
-6.51120305e-01 -1.95553601e-02 -6.77412674e-02 -5.00272214e-01
1.26448369e+00 1.15105413e-01 -3.32687438e-01 8.18072557e-01
1.12989247e+00 1.79831386e-01 -9.27711606e-01 -8.76084119e-02
3.19428056e-01 1.01710692e-01 -5.81069551e-02 -1.28543603e+00
-3.30935955e-01 2.73032248e-01 -1.64025053e-01 5.35384834e-01
8.70003402e-01 5.78544438e-01 4.91124898e-01 -1.87654123e-01
2.49756902e-01 -9.78473127e-01 -1.66443922e-02 7.20953405e-01
1.37012565e+00 -7.42460251e-01 1.03113011e-01 -4.34907138e-01
-4.80508536e-01 8.28602970e-01 1.77416086e-01 -2.47783005e-01
6.64523363e-01 3.88977826e-01 -2.68348813e-01 -4.42628860e-01
-1.36299801e+00 1.00788735e-01 3.84165198e-01 -7.61838630e-02
8.91583681e-01 1.18185118e-01 -7.80694246e-01 9.50734913e-01
-3.86733294e-01 -8.61919150e-02 4.04223919e-01 4.61210549e-01
-2.08343148e-01 -7.59908020e-01 -6.67198718e-01 6.28593624e-01
-8.82723927e-01 -4.76917773e-01 -3.75065833e-01 4.79435921e-01
2.76677847e-01 1.30702090e+00 1.34893075e-01 -1.82842791e-01
6.58137918e-01 3.77604157e-01 1.67733282e-01 -5.42695522e-01
-7.83513010e-01 6.13476038e-01 7.24715471e-01 -5.28222382e-01
-8.14976469e-02 -1.22730684e+00 -1.65619493e+00 5.15483804e-02
3.82637829e-02 2.00133771e-01 4.45916772e-01 1.14895904e+00
2.77059495e-01 7.71856308e-01 4.68916833e-01 -3.51086766e-01
-2.83959717e-01 -1.26526594e+00 -4.64139849e-01 3.53906542e-01
3.81153733e-01 -6.06648147e-01 -3.99532914e-01 5.37267029e-01]
|
[9.0769624710083, 9.248492240905762]
|
e87cdf60-531a-4894-8802-973af69eb7dc
|
dynamic-scene-graph-generation-via
| null | null |
http://openaccess.thecvf.com//content/CVPR2022/html/Li_Dynamic_Scene_Graph_Generation_via_Anticipatory_Pre-Training_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Li_Dynamic_Scene_Graph_Generation_via_Anticipatory_Pre-Training_CVPR_2022_paper.pdf
|
Dynamic Scene Graph Generation via Anticipatory Pre-Training
|
Humans can not only see the collection of objects in visual scenes, but also identify the relationship between objects. The visual relationship in the scene can be abstracted into the semantic representation of triple <subject, predicate, object> and thus results in a scene graph, which can convey a lot of information for visual understanding. Due to the motion of objects, the visual relationship between two objects in videos may vary, which makes the task of dynamically generating scene graphs from videos more complicated and challenging than the conventional image-based static scene graph generation. Inspired by the ability of humans to infer the visual relationship, we propose a novel anticipatory pre-training paradigm based on Transformer to explicitly model the temporal correlation of visual relationships in different frames to improve dynamic scene graph generation. In pre-training stage, the model predicts the visual relationships of current frame based on the previous frames by extracting intra-frame spatial information with a spatial encoder and inter-frame temporal correlations with a temporal encoder. In the fine-tuning stage, we reuse the spatial encoder and the temporal decoder and combine the information of the current frame to predict the visual relationship. Extensive experiments demonstrate that our method achieves state-of-the-art performance on Action Genome dataset.
|
['Changsheng Xu', 'Xiaoshan Yang', 'Yiming Li']
|
2022-01-01
| null | null | null |
cvpr-2022-1
|
['scene-graph-generation']
|
['computer-vision']
|
[ 4.27037716e-01 -1.65027946e-01 -6.46266416e-02 -5.45008838e-01
8.49744529e-02 -4.55453783e-01 4.27554935e-01 1.92335114e-01
-1.67549908e-01 3.47825110e-01 5.20047426e-01 6.20541312e-02
1.60840526e-01 -7.30848491e-01 -8.89115930e-01 -4.88381237e-01
5.92228808e-02 -4.48209606e-02 7.14926839e-01 -1.31030425e-01
2.08406925e-01 1.19071282e-01 -1.51142919e+00 9.67643201e-01
4.37825978e-01 9.46780145e-01 9.16284382e-01 6.09056175e-01
-2.66380340e-01 1.28000271e+00 -5.79686105e-01 1.87073573e-02
-1.36554733e-01 -7.41466165e-01 -8.30479383e-01 5.83678961e-01
2.98484951e-01 -4.69670981e-01 -5.90411425e-01 8.79125297e-01
2.12493569e-01 3.65130633e-01 2.06830457e-01 -1.23385000e+00
-6.06569588e-01 4.74582314e-01 -6.42591715e-01 4.87289488e-01
7.29399502e-01 3.62680197e-01 8.68566692e-01 -5.52837849e-01
9.25591052e-01 1.53716063e+00 -8.35730284e-02 4.04964387e-01
-1.02299368e+00 -5.07801712e-01 6.73337102e-01 8.61379802e-01
-1.10154104e+00 -2.98407614e-01 8.70902896e-01 -6.40741408e-01
8.77620041e-01 -4.41882247e-03 1.18809593e+00 7.87011504e-01
1.50293067e-01 8.96714449e-01 6.68719709e-01 -2.19407752e-01
-2.28074454e-02 -4.24551219e-01 -2.07116127e-01 7.36567914e-01
-2.18172297e-01 -7.67058181e-03 -8.34998250e-01 3.89300734e-01
8.97656858e-01 2.02178314e-01 -4.58158284e-01 -4.80314791e-01
-1.56861901e+00 3.49517852e-01 7.50798166e-01 2.55472094e-01
-4.37769413e-01 2.83300906e-01 2.90022939e-01 -5.86664155e-02
-4.55531217e-02 7.04189092e-02 -3.23886544e-01 5.33819534e-02
-4.86750960e-01 1.33332565e-01 3.12617958e-01 9.70932364e-01
6.83298171e-01 -1.41232297e-01 -5.26269197e-01 4.60110873e-01
4.11235064e-01 1.85631916e-01 3.50955814e-01 -1.03040540e+00
6.57192707e-01 8.48434091e-01 -4.16600518e-02 -1.36299098e+00
-1.71488121e-01 3.61643136e-02 -7.92549372e-01 -2.35192358e-01
3.14895630e-01 1.61783665e-01 -8.60661745e-01 1.71599913e+00
4.97836918e-01 5.21596909e-01 -7.24232644e-02 9.98286545e-01
1.01757836e+00 1.01813281e+00 2.50133038e-01 -5.09277403e-01
1.56534886e+00 -9.47332501e-01 -8.46973956e-01 -4.99812663e-01
3.56017470e-01 -5.62943816e-01 8.01651776e-01 9.83449742e-02
-9.97645557e-01 -9.43933010e-01 -7.84989893e-01 -3.30987483e-01
-1.28415629e-01 2.19257511e-02 7.35280693e-01 -3.36169779e-01
-7.52747178e-01 2.45417386e-01 -7.65411735e-01 -3.50806057e-01
4.50156063e-01 1.37899658e-02 -3.96084189e-01 -2.67075419e-01
-1.12602282e+00 4.32333082e-01 9.75056410e-01 1.81610242e-01
-9.26262200e-01 -5.23933649e-01 -1.09975111e+00 3.51795815e-02
5.62544584e-01 -8.13143611e-01 8.63181651e-01 -1.14563048e+00
-1.05170321e+00 6.68694019e-01 -5.88862598e-01 -2.01061919e-01
1.63709834e-01 -1.98485013e-02 -3.64593506e-01 4.13507700e-01
5.36944941e-02 9.84870672e-01 7.23593831e-01 -1.26438725e+00
-9.73084629e-01 -1.92543745e-01 3.29918861e-01 5.19317567e-01
8.19055736e-02 1.36785701e-01 -9.96310115e-01 -5.47658145e-01
2.80676991e-01 -8.12400281e-01 -1.00040153e-01 2.15096295e-01
-3.25882763e-01 -2.15218708e-01 1.02710783e+00 -6.88287258e-01
1.19887388e+00 -2.46969247e+00 3.49628657e-01 -1.83852151e-01
1.13205500e-01 1.66293025e-01 -1.46005958e-01 3.21313322e-01
-3.42835605e-01 -2.78392196e-01 1.48258824e-02 1.35227993e-01
-5.35340488e-01 3.74128044e-01 -3.01760018e-01 1.91793919e-01
1.02617987e-01 9.52429473e-01 -1.41367090e+00 -7.21127868e-01
4.02930975e-01 4.91782814e-01 -5.50953329e-01 5.48949301e-01
-6.17961168e-01 8.47801864e-01 -6.05252266e-01 2.93325484e-01
2.85334915e-01 -6.11878037e-01 3.49244833e-01 -7.39245236e-01
8.78635943e-02 2.32529417e-01 -1.05510426e+00 2.04368377e+00
-2.78108474e-02 7.40231276e-01 -5.41946530e-01 -1.09257066e+00
7.02822566e-01 2.40273193e-01 4.89392579e-01 -8.61007094e-01
-3.51055712e-02 -5.06488860e-01 2.23801747e-01 -8.43516827e-01
2.12213010e-01 2.46869609e-01 4.44734395e-02 1.65525377e-01
5.40851578e-02 6.17002919e-02 6.04100168e-01 4.67640400e-01
7.25402236e-01 4.83076513e-01 4.09866452e-01 1.91974968e-01
7.71642625e-01 -1.11584984e-01 7.71195531e-01 2.68935382e-01
-8.18265416e-03 5.69544077e-01 6.87885642e-01 -6.17839217e-01
-6.67840481e-01 -1.07874203e+00 3.98519486e-01 9.67034519e-01
7.28694022e-01 -6.83290839e-01 -4.55826998e-01 -5.04670918e-01
-2.61530221e-01 6.20912790e-01 -6.74285412e-01 -3.17438096e-01
-7.29005039e-01 -3.62851083e-01 -2.48258665e-01 5.91411889e-01
7.19947577e-01 -1.36362302e+00 -8.13492060e-01 2.50674397e-01
-7.05247283e-01 -1.61586320e+00 -7.02536464e-01 -3.36973548e-01
-6.53387547e-01 -1.27093136e+00 -2.38835275e-01 -8.97752762e-01
9.21273351e-01 5.99829376e-01 1.03713953e+00 3.38923901e-01
-2.88871557e-01 2.00226545e-01 -4.68583465e-01 -6.73178509e-02
-2.27258638e-01 -6.48652971e-01 -3.47203046e-01 3.44491303e-01
5.12176305e-02 -4.57751274e-01 -9.06602263e-01 3.36621970e-01
-8.96854699e-01 8.84117186e-01 3.58505934e-01 6.61104262e-01
7.04614758e-01 2.76160210e-01 1.23336772e-02 -6.86231196e-01
-5.08791395e-02 -2.56748259e-01 -4.48969185e-01 4.10460562e-01
1.75514311e-01 9.84600559e-02 5.82305849e-01 -4.94456112e-01
-1.18959987e+00 3.54560077e-01 4.19648975e-01 -5.18125176e-01
-8.06690976e-02 4.62083399e-01 -3.67863804e-01 5.27802289e-01
1.61509857e-01 5.30574262e-01 -2.43378058e-01 -1.26356408e-01
5.53999245e-01 9.53413546e-02 7.96341240e-01 -2.95969307e-01
5.81860304e-01 6.12488091e-01 6.46125972e-02 -5.26122689e-01
-1.10377467e+00 -4.40295219e-01 -1.10239148e+00 -5.24891376e-01
1.35077071e+00 -1.09384501e+00 -7.38375425e-01 3.03517908e-01
-1.41005623e+00 -3.50881010e-01 -1.13328688e-01 4.38530833e-01
-5.70290446e-01 4.06995833e-01 -2.70531714e-01 -3.00835520e-01
1.27815500e-01 -1.20780468e+00 1.07765353e+00 3.91440690e-01
-2.00293198e-01 -8.63680184e-01 -3.70807409e-01 3.71654630e-01
-2.40185633e-01 2.97722846e-01 1.11038148e+00 -4.92457673e-02
-1.07255805e+00 2.41715312e-01 -4.17247623e-01 2.81249117e-02
3.70256335e-01 1.10631153e-01 -4.23354119e-01 -9.97158736e-02
-2.85609990e-01 3.65170538e-02 6.99840426e-01 3.58662695e-01
1.45006740e+00 -2.86541820e-01 -5.02685308e-01 6.10285163e-01
1.29784703e+00 5.04904151e-01 7.63733685e-01 4.43442017e-02
1.14882660e+00 7.53953099e-01 8.18989635e-01 4.21530873e-01
6.84970379e-01 8.80385041e-01 3.88353199e-01 1.07273318e-01
-4.11340296e-01 -5.08365512e-01 3.71091515e-01 7.06779659e-01
-1.41772687e-01 -2.57198513e-01 -6.56911552e-01 4.29306209e-01
-2.17566967e+00 -1.37038279e+00 -1.21756136e-01 2.01331973e+00
8.77896905e-01 1.64677273e-03 3.43532749e-02 -2.16591299e-01
7.61984706e-01 3.92137796e-01 -5.11751294e-01 1.15195468e-01
1.36979287e-02 -4.34978873e-01 1.08725168e-01 2.81439722e-01
-8.68239999e-01 1.07379067e+00 5.80773783e+00 4.89975125e-01
-1.11641812e+00 -2.70346820e-01 6.60420775e-01 -8.75631571e-02
-7.26235583e-02 2.71910757e-01 -5.63497484e-01 6.63362026e-01
3.32024157e-01 -3.51902515e-01 4.33503121e-01 4.50630039e-01
4.64931786e-01 -3.62827122e-01 -1.34905708e+00 1.19886315e+00
1.07596181e-01 -1.29736900e+00 4.53117281e-01 -1.90412953e-01
6.33615613e-01 -4.30072576e-01 -2.04635769e-01 -2.52556168e-02
1.82439879e-01 -8.39047909e-01 8.43956828e-01 7.33622372e-01
5.52423477e-01 -4.94920880e-01 3.15799624e-01 3.85246277e-01
-1.73176467e+00 -1.69558853e-01 -3.32452655e-01 -1.40681580e-01
4.21394318e-01 2.82048881e-01 -7.02996194e-01 6.52195036e-01
7.82617092e-01 1.35662460e+00 -5.75332224e-01 8.87397349e-01
-4.99012798e-01 1.98800832e-01 6.40775934e-02 1.66217819e-01
5.47242463e-02 -2.18630329e-01 3.20286661e-01 9.50572550e-01
1.97960988e-01 5.56418955e-01 3.45318228e-01 8.25969577e-01
2.67829746e-01 -1.42222688e-01 -4.80123580e-01 -6.30703047e-02
3.62009108e-01 1.08389342e+00 -7.92019367e-01 -6.28433168e-01
-5.65438032e-01 1.08157444e+00 2.48461410e-01 4.66684252e-01
-9.07406569e-01 1.40057117e-01 3.52847189e-01 1.44370601e-01
4.85668004e-01 -3.41178238e-01 2.95978010e-01 -1.20952344e+00
6.68565854e-02 -8.60535562e-01 6.11862481e-01 -1.31868279e+00
-7.63694942e-01 3.82100821e-01 3.02032650e-01 -1.40823066e+00
-4.13030684e-01 -3.18662494e-01 -5.61883450e-01 5.32313883e-01
-1.15504014e+00 -1.12269008e+00 -7.31308818e-01 7.15973675e-01
7.96993434e-01 1.32459372e-01 4.16053832e-01 -5.94448484e-02
-3.95357728e-01 3.23621072e-02 -5.13265550e-01 2.53334314e-01
4.30992305e-01 -1.01009011e+00 4.01086003e-01 1.02087379e+00
3.44312131e-01 3.08691055e-01 5.30831099e-01 -7.50782609e-01
-1.39659321e+00 -1.17090297e+00 7.99504757e-01 -2.70874768e-01
5.07247686e-01 -3.59658122e-01 -9.70118761e-01 7.00592935e-01
1.51899353e-01 2.02244937e-01 4.83572364e-01 -3.20583731e-01
-3.16489339e-01 -1.73755109e-01 -5.08390188e-01 7.83572197e-01
1.50653863e+00 -5.87026119e-01 -6.82211280e-01 3.14023733e-01
9.16759908e-01 -6.23878062e-01 -5.08663654e-01 4.02195305e-01
5.21429062e-01 -1.07719111e+00 1.27626240e+00 -5.98802865e-01
6.91036642e-01 -7.51473308e-01 -6.30684569e-02 -1.09414959e+00
-5.33100069e-01 -4.03898150e-01 -1.42351016e-01 1.15252650e+00
-9.14786682e-02 -4.52884436e-02 4.79126990e-01 4.57636803e-01
9.42719281e-02 -5.24235964e-01 -5.61446130e-01 -4.34160471e-01
-7.76532829e-01 -1.95705846e-01 4.34777647e-01 8.72608840e-01
7.04605281e-02 5.02588034e-01 -4.32035863e-01 2.39010483e-01
3.78559679e-01 4.35148954e-01 8.92655849e-01 -8.13168764e-01
-4.84520048e-01 -9.08059105e-02 -7.19369531e-01 -1.48098052e+00
1.53664947e-01 -6.85860515e-01 1.82226434e-01 -1.84764051e+00
5.21127939e-01 5.07057421e-02 -2.03945547e-01 5.09302497e-01
-5.67604005e-01 -2.33901236e-02 4.96301323e-01 1.37252182e-01
-9.84874189e-01 5.06546199e-01 1.84266925e+00 -2.38697648e-01
-1.19134389e-01 -5.11260509e-01 -4.03117478e-01 7.53353417e-01
2.69031823e-01 -2.26751462e-01 -9.08744335e-01 -5.91125607e-01
1.16261818e-01 5.00450552e-01 5.52842379e-01 -1.04914534e+00
2.87824243e-01 -5.48076749e-01 7.17003524e-01 -7.54136920e-01
3.48994553e-01 -8.45167220e-01 3.47026587e-01 4.53458905e-01
-3.85949403e-01 1.36594906e-01 7.04961494e-02 7.66352713e-01
-4.56564009e-01 1.45849302e-01 5.94747245e-01 -3.53422493e-01
-1.39765823e+00 5.09890556e-01 -1.45136848e-01 -2.53559258e-02
1.15647280e+00 -3.06768835e-01 -3.72207940e-01 -4.77278352e-01
-8.92347217e-01 4.14570838e-01 4.21024859e-01 6.66688442e-01
8.71419907e-01 -1.26893008e+00 -4.58933264e-01 2.04224616e-01
2.48985350e-01 3.61753285e-01 6.46118641e-01 6.39545381e-01
-3.66140991e-01 5.50056137e-02 -4.27112281e-01 -8.64731491e-01
-1.51144612e+00 8.27821314e-01 2.28869259e-01 7.46965379e-05
-7.95581698e-01 8.09431195e-01 1.12124991e+00 4.65366900e-01
4.37908396e-02 -3.93459469e-01 -5.65170705e-01 -5.30425645e-02
7.83193529e-01 -1.89449891e-01 -5.95320880e-01 -9.57968771e-01
-3.21441889e-01 7.06512392e-01 -9.50583145e-02 1.49572194e-01
1.14283645e+00 -3.25516492e-01 -3.29937369e-01 6.79387808e-01
1.22070765e+00 -3.42204124e-01 -1.52232718e+00 -3.59465927e-01
-3.48612487e-01 -8.37017298e-01 -2.37296656e-01 -5.69541335e-01
-1.25844252e+00 8.17879736e-01 2.80893505e-01 -4.80691418e-02
1.36769223e+00 2.52090931e-01 6.63742900e-01 8.19015503e-02
3.01284939e-01 -8.12620819e-01 7.26262748e-01 4.53552634e-01
8.89845669e-01 -1.02338421e+00 2.17988551e-01 -9.38058555e-01
-8.95783603e-01 1.12070489e+00 8.41180861e-01 5.73404878e-02
4.88795370e-01 -8.69676769e-02 -1.94597080e-01 -2.69676298e-01
-1.03912032e+00 -3.32138062e-01 5.47678232e-01 7.51853824e-01
3.56044561e-01 -2.61432290e-01 3.41994502e-02 1.44079626e-01
8.90712291e-02 6.81679249e-02 2.90857881e-01 7.95715034e-01
-3.32538992e-01 -9.83548760e-01 3.75307165e-02 1.49839327e-01
-1.08643591e-01 -4.41688076e-02 -2.13545054e-01 5.65063953e-01
4.26956564e-01 8.43499362e-01 5.24244905e-01 -3.19888592e-01
2.91854411e-01 -2.09586188e-01 6.60557210e-01 -6.55049443e-01
-9.53721777e-02 2.59220839e-01 -7.12632667e-03 -7.99713671e-01
-8.69384825e-01 -8.00423622e-01 -1.72561479e+00 1.36572510e-01
1.59854174e-01 -9.36404020e-02 1.59799993e-01 1.11522377e+00
3.96600962e-01 9.52876151e-01 5.66210926e-01 -7.30165064e-01
4.06809598e-01 -6.24878049e-01 -2.74244308e-01 1.01945484e+00
2.84453839e-01 -7.50593305e-01 1.54096037e-01 8.26095760e-01]
|
[9.017190933227539, 0.7610204219818115]
|
2c797ca6-4486-44e2-9818-283d7c090348
|
human-joint-kinematics-diffusion-refinement
|
2210.05976
| null |
https://arxiv.org/abs/2210.05976v2
|
https://arxiv.org/pdf/2210.05976v2.pdf
|
Human Joint Kinematics Diffusion-Refinement for Stochastic Motion Prediction
|
Stochastic human motion prediction aims to forecast multiple plausible future motions given a single pose sequence from the past. Most previous works focus on designing elaborate losses to improve the accuracy, while the diversity is typically characterized by randomly sampling a set of latent variables from the latent prior, which is then decoded into possible motions. This joint training of sampling and decoding, however, suffers from posterior collapse as the learned latent variables tend to be ignored by a strong decoder, leading to limited diversity. Alternatively, inspired by the diffusion process in nonequilibrium thermodynamics, we propose MotionDiff, a diffusion probabilistic model to treat the kinematics of human joints as heated particles, which will diffuse from original states to a noise distribution. This process offers a natural way to obtain the "whitened" latents without any trainable parameters, and human motion prediction can be regarded as the reverse diffusion process that converts the noise distribution into realistic future motions conditioned on the observed sequence. Specifically, MotionDiff consists of two parts: a spatial-temporal transformer-based diffusion network to generate diverse yet plausible motions, and a graph convolutional network to further refine the outputs. Experimental results on two datasets demonstrate that our model yields the competitive performance in terms of both accuracy and diversity.
|
['Shengxiang Hu', 'Xiaoning Sun', 'Weiqing Li', 'Jianfeng Lu', 'Bin Li', 'Huaijiang Sun', 'Dong Wei']
|
2022-10-12
| null | null | null | null |
['stochastic-human-motion-prediction']
|
['computer-vision']
|
[ 1.19841471e-01 2.49317437e-01 2.37446725e-02 -5.13593331e-02
-5.79064012e-01 -7.22406730e-02 7.98416913e-01 -7.73117840e-01
-2.07260817e-01 7.89858162e-01 6.51382506e-01 1.73709556e-01
2.76476294e-01 -8.89785230e-01 -9.21515822e-01 -1.32532883e+00
3.28479201e-01 6.55279934e-01 3.57472241e-01 -1.10181056e-01
-2.44824603e-01 1.24879338e-01 -1.16651845e+00 -1.08491100e-01
9.13759232e-01 5.49836457e-01 5.19569039e-01 6.92666948e-01
4.05404344e-02 1.03556979e+00 -4.19916272e-01 -4.35815305e-01
9.17685255e-02 -8.95458519e-01 -5.21179438e-01 9.49584842e-02
-3.25292826e-01 -4.92577404e-01 -9.37404633e-01 1.09345627e+00
5.39392531e-01 4.64298725e-01 7.95655906e-01 -1.01771510e+00
-9.49860632e-01 7.21392035e-01 -4.04361874e-01 -3.37133735e-01
3.02716792e-01 7.00744450e-01 8.21723461e-01 -5.49633682e-01
8.46427023e-01 1.58089960e+00 4.05422330e-01 1.07316780e+00
-1.22119749e+00 -5.03995657e-01 1.41721755e-01 7.67475888e-02
-1.14903283e+00 -2.43248656e-01 9.98936415e-01 -4.77614194e-01
5.84441125e-01 5.95340952e-02 1.02194071e+00 1.82375252e+00
5.84137082e-01 1.10030162e+00 5.17032087e-01 6.36000112e-02
4.54220474e-01 -4.96215761e-01 -3.61123502e-01 6.89155877e-01
-4.31241170e-02 1.49534822e-01 -4.96090710e-01 -1.99596673e-01
1.01609290e+00 7.87754431e-02 -4.61038470e-01 -3.38758290e-01
-1.41177034e+00 7.36929417e-01 3.05183560e-01 -9.93174613e-02
-6.51631832e-01 3.76147240e-01 1.89441890e-01 -1.18542314e-01
4.56765264e-01 1.73452899e-01 -3.69491130e-02 -2.45000705e-01
-9.35878694e-01 6.52691722e-01 6.89418137e-01 8.66990209e-01
6.79526627e-01 1.72422022e-01 -5.70230365e-01 5.20447969e-01
6.00188971e-01 7.20975399e-01 6.48905337e-01 -1.24989474e+00
5.12687147e-01 1.34092674e-01 3.20902139e-01 -1.10090137e+00
-6.35134801e-02 -3.19401234e-01 -1.32009900e+00 -2.91826148e-02
2.78912604e-01 -4.23144877e-01 -1.15587997e+00 2.07081389e+00
3.06570709e-01 3.05138260e-01 -1.09604925e-01 1.34979784e+00
2.85742164e-01 1.22453249e+00 -9.74259153e-03 -3.21966648e-01
7.17436492e-01 -1.30021286e+00 -8.54287744e-01 -1.68561816e-01
2.42506236e-01 -4.68298554e-01 8.86874020e-01 3.86741847e-01
-1.28174758e+00 -7.14506030e-01 -9.11149502e-01 -2.40247354e-01
4.17718381e-01 -1.87855631e-01 4.43678498e-01 1.85367078e-01
-1.01316869e+00 8.24830115e-01 -1.41056192e+00 -6.07898161e-02
1.90937817e-01 7.04888105e-02 1.22095458e-02 -1.18700065e-01
-1.37779832e+00 6.59736693e-01 3.52830797e-01 4.60896015e-01
-1.28695476e+00 -2.00205624e-01 -6.73875153e-01 -9.22346711e-02
2.67123222e-01 -1.47347975e+00 9.45298016e-01 -6.84765875e-01
-2.06326318e+00 1.72247648e-01 -3.44429553e-01 -3.67871076e-01
1.10865009e+00 -3.17364693e-01 -9.45118070e-02 -2.05778945e-02
8.41824412e-02 9.62101698e-01 1.11148357e+00 -9.58562255e-01
-3.70110810e-01 -4.48819846e-02 -2.06450433e-01 3.79860461e-01
-6.07804172e-02 -6.63284600e-01 -8.85537624e-01 -9.09591854e-01
1.57470420e-01 -1.26633286e+00 -6.73632801e-01 -3.71386930e-02
-5.15375078e-01 -1.21396273e-01 5.58428586e-01 -8.01877737e-01
1.10723019e+00 -1.72261214e+00 1.06386566e+00 -5.85404485e-02
2.53489405e-01 -1.14386417e-01 -1.09748006e-01 3.20334643e-01
3.16651165e-01 -2.88157295e-02 -4.32196945e-01 -5.23196518e-01
1.75992996e-01 4.86149341e-01 -6.03243470e-01 3.60186487e-01
3.81431095e-02 1.17920375e+00 -1.15428805e+00 -3.44771832e-01
1.80246994e-01 6.68824434e-01 -6.35661662e-01 4.37653661e-01
-6.92925632e-01 9.93182003e-01 -7.49397039e-01 7.93754682e-02
5.58205128e-01 -2.89975524e-01 7.21828938e-02 1.67111754e-01
2.71438837e-01 8.80445614e-02 -9.17287469e-01 2.02815461e+00
7.90403876e-03 1.18141815e-01 -2.57992297e-01 -7.25453794e-01
9.81438637e-01 2.89868593e-01 4.87049133e-01 -2.45828986e-01
2.95010637e-02 9.10773277e-02 -2.13644400e-01 -5.76709867e-01
6.28499627e-01 -2.88613409e-01 -1.05339505e-01 3.03283632e-01
-1.77617908e-01 -1.88878059e-01 -3.56445871e-02 8.58500302e-02
9.56315160e-01 6.36920094e-01 -4.73014176e-01 1.28652140e-01
4.88816679e-01 -6.28220662e-02 1.19154191e+00 6.09962881e-01
-9.57645997e-02 9.35465217e-01 5.45749187e-01 -3.83689225e-01
-1.36587465e+00 -1.27530038e+00 6.00796759e-01 4.68737066e-01
4.16183382e-01 -1.91320941e-01 -8.09569180e-01 -3.12368184e-01
-3.02309364e-01 7.53648281e-01 -5.25413811e-01 -6.17984951e-01
-9.58668530e-01 -8.72583508e-01 2.86187559e-01 3.84077400e-01
5.14157712e-01 -1.25893998e+00 -3.87342721e-01 4.79607105e-01
-5.60040355e-01 -7.14740813e-01 -7.42499471e-01 -3.75320941e-01
-8.47596705e-01 -5.23292899e-01 -1.41577864e+00 -6.84631109e-01
4.21767175e-01 -1.37396038e-01 7.85091460e-01 -1.19013421e-01
8.00298378e-02 2.14153118e-02 -1.88279912e-01 1.41586483e-01
-6.57648265e-01 8.38233083e-02 1.49210334e-01 1.85728595e-01
2.56162733e-02 -6.67403340e-01 -1.00700045e+00 5.16372137e-02
-9.46464658e-01 4.89670008e-01 6.60966158e-01 1.07223272e+00
6.58123493e-01 6.75557330e-02 1.63679242e-01 -4.22028095e-01
4.71707135e-01 -7.07336664e-01 -2.35945940e-01 9.65292007e-02
-2.49212235e-01 4.31086272e-01 7.71882832e-01 -7.38660455e-01
-1.33225584e+00 2.79078364e-01 -3.18142474e-01 -7.55495131e-01
4.83837798e-02 3.28263700e-01 -2.91078180e-01 4.96893644e-01
3.21782380e-01 7.29274392e-01 8.82932022e-02 -4.77454513e-01
6.89869046e-01 1.02436185e-01 6.44567311e-01 -7.05800235e-01
7.40049124e-01 6.15970194e-01 4.06111479e-02 -6.32978201e-01
-4.96810317e-01 9.57456902e-02 -5.02312660e-01 -2.64871031e-01
1.29805088e+00 -9.02680933e-01 -5.28490245e-01 9.46906447e-01
-1.32068551e+00 -5.35008550e-01 -3.15252066e-01 7.17858613e-01
-7.68657148e-01 6.75266922e-01 -1.12489891e+00 -7.95629263e-01
-1.97743103e-01 -1.29893064e+00 1.14128709e+00 3.14826429e-01
-3.31895888e-01 -8.92030001e-01 2.20561087e-01 1.15577243e-01
9.42575708e-02 3.94789249e-01 8.34706008e-01 1.32821947e-01
-9.57378626e-01 -3.08235623e-02 4.12698537e-01 1.83373451e-01
-7.98123032e-02 3.50745507e-02 -5.40799201e-01 -3.34255099e-01
2.76796818e-01 -2.32461706e-01 1.12160802e+00 5.57568848e-01
9.19197142e-01 -4.04189259e-01 -5.55666327e-01 7.17707992e-01
9.14991200e-01 9.45372060e-02 8.55123401e-01 6.01375522e-03
1.04745793e+00 5.44938982e-01 3.59649658e-01 4.77339149e-01
4.80873525e-01 3.96492481e-01 2.42895812e-01 3.06593120e-01
-1.31277919e-01 -8.53592098e-01 6.34895861e-01 1.33142853e+00
-2.39282459e-01 -4.96775866e-01 -7.31979549e-01 4.33398098e-01
-2.16204262e+00 -1.11215770e+00 -1.73962247e-02 1.87826502e+00
7.81196475e-01 2.53925860e-01 1.08769357e-01 -2.52908111e-01
8.42196882e-01 5.76276422e-01 -9.71292853e-01 2.30332464e-01
-1.10121973e-01 -2.31583148e-01 2.61449009e-01 6.10569835e-01
-7.30050862e-01 1.01517045e+00 6.04205608e+00 9.94800508e-01
-1.13395178e+00 5.98671008e-03 5.93058109e-01 -2.25595608e-01
-7.81649590e-01 1.09258018e-01 -6.16616607e-01 8.18941474e-01
6.17965043e-01 -8.44596401e-02 4.12821501e-01 6.34106755e-01
4.62056160e-01 2.13324025e-01 -7.48510897e-01 8.70902598e-01
-3.12256962e-01 -1.25892544e+00 5.32795668e-01 4.99838702e-02
8.74418855e-01 -8.88882503e-02 2.01643437e-01 3.36362839e-01
6.95158422e-01 -8.64324510e-01 1.10936928e+00 1.28232408e+00
3.74450237e-01 -6.34151936e-01 2.93487102e-01 8.89485776e-01
-1.09906054e+00 1.10549286e-01 -6.28109097e-01 -1.15280434e-01
6.86997533e-01 6.15066767e-01 -2.78289735e-01 4.99827892e-01
4.38508421e-01 9.03704882e-01 -4.37784530e-02 6.96950734e-01
-5.15232623e-01 6.89132571e-01 -2.17334479e-01 -2.83685803e-01
1.80566669e-01 -6.36457026e-01 8.39616060e-01 6.75846159e-01
5.67730367e-01 7.63788298e-02 2.59990841e-01 1.27057385e+00
1.03399128e-01 -3.60839397e-01 -3.20960462e-01 3.32012102e-02
3.12158585e-01 8.06091487e-01 -5.72893441e-01 -3.95083964e-01
-3.87124419e-02 1.57735598e+00 2.88410991e-01 7.54615843e-01
-1.10320139e+00 -4.63962369e-02 5.14995039e-01 -1.44903421e-01
3.76118243e-01 -5.99379718e-01 -7.84752741e-02 -1.54310513e+00
1.63403854e-01 -4.99256641e-01 6.13645129e-02 -7.92705357e-01
-1.34062064e+00 5.44479370e-01 -1.64811343e-01 -1.26793087e+00
-5.15193284e-01 -1.46235049e-01 -6.18605852e-01 1.16073143e+00
-9.58870232e-01 -1.08483934e+00 -4.16112989e-02 2.82815516e-01
6.75099075e-01 1.42710775e-01 3.99889946e-01 -1.35584362e-02
-6.03121579e-01 2.50927538e-01 2.39244297e-01 -4.72806469e-02
4.11805719e-01 -1.18711519e+00 8.79383743e-01 8.45745385e-01
-1.88606903e-01 4.60149050e-01 9.81449306e-01 -1.11794984e+00
-1.40397418e+00 -1.23196959e+00 6.85190737e-01 -4.27426040e-01
5.93652070e-01 -2.95044810e-01 -1.12374580e+00 5.65213442e-01
-5.22146337e-02 -2.31315002e-01 3.18450783e-03 -5.43500721e-01
2.26191774e-01 2.86963880e-01 -6.23643100e-01 1.04562593e+00
1.39272082e+00 -3.80475134e-01 -7.14547336e-01 3.13958466e-01
9.30213332e-01 -5.99971771e-01 -6.77482307e-01 2.51768023e-01
5.53950846e-01 -8.17743123e-01 8.33221078e-01 -4.13446009e-01
7.00975120e-01 -3.78892988e-01 1.85755521e-01 -1.42359889e+00
-6.21104062e-01 -1.01645935e+00 -3.69437546e-01 1.09465015e+00
2.19583109e-01 -4.01643008e-01 1.12853479e+00 7.99080968e-01
-2.78265215e-02 -8.00038815e-01 -7.65230417e-01 -7.73050129e-01
3.91402721e-01 -3.18966389e-01 6.67071879e-01 6.06070101e-01
-4.83091146e-01 3.28509420e-01 -9.47168231e-01 8.97700489e-02
6.98694229e-01 2.93512805e-03 8.68526042e-01 -7.25331306e-01
-6.32189691e-01 -3.78572136e-01 -1.40138373e-01 -2.04431105e+00
1.41598448e-01 -7.08597958e-01 5.62059402e-01 -1.58116925e+00
2.40911230e-01 -1.35961115e-01 1.10878617e-01 -1.77407444e-01
-5.14778912e-01 -3.33287120e-01 1.15065321e-01 7.62582541e-01
-4.51842576e-01 1.50348210e+00 1.92026520e+00 -1.99347094e-01
-2.22775936e-01 4.40089367e-02 -1.21665910e-01 8.55033338e-01
4.13175046e-01 -5.19989133e-01 -5.88740826e-01 -6.29194617e-01
1.33727640e-01 5.69613934e-01 3.80483061e-01 -1.05154729e+00
3.74593675e-01 -4.26726043e-01 4.07079935e-01 -6.82938755e-01
5.91137886e-01 -3.54946166e-01 5.66059351e-01 7.66196728e-01
-3.69897217e-01 -1.23052448e-01 -4.74395245e-01 1.17637217e+00
2.58279935e-04 1.44417986e-01 5.43725908e-01 -2.98482776e-01
-6.00250602e-01 7.47147918e-01 -6.41281903e-01 -2.47218888e-02
8.20369482e-01 -8.99436697e-02 -3.42843868e-02 -6.39776886e-01
-9.79831874e-01 3.65944624e-01 6.58116519e-01 5.45051455e-01
5.68767548e-01 -1.62742329e+00 -7.17670441e-01 6.26789778e-03
-5.18905699e-01 4.82835859e-01 5.01315653e-01 5.68864107e-01
-6.98224127e-01 4.58508395e-02 -8.79690200e-02 -7.52410889e-01
-4.48491126e-01 6.87919855e-01 3.59208763e-01 -3.20717216e-01
-9.55343485e-01 9.45089579e-01 3.74553502e-01 -3.02161098e-01
3.27207185e-02 -3.61791790e-01 2.86706872e-02 -3.91308248e-01
2.33381584e-01 5.21117389e-01 -7.52754867e-01 -7.32384026e-01
-2.93621924e-02 5.34335792e-01 1.84338108e-01 -4.76413071e-01
1.10320950e+00 -4.32468086e-01 -1.62293352e-02 6.16638303e-01
1.16956449e+00 -2.77928799e-01 -1.94277048e+00 -1.26916632e-01
-1.35409683e-01 -2.13325337e-01 -4.01449412e-01 -2.22927675e-01
-9.77759063e-01 1.03584564e+00 1.70645326e-01 1.07017141e-02
8.53813767e-01 -1.55119181e-01 1.50211787e+00 1.94911659e-01
4.88665551e-01 -1.09137535e+00 2.19134733e-01 6.39018655e-01
8.81035566e-01 -8.26856196e-01 -3.34208190e-01 -2.52466202e-01
-8.50797355e-01 1.05530095e+00 5.99868000e-01 -3.85411859e-01
6.17700875e-01 -1.20271869e-01 -1.12175383e-01 8.89147371e-02
-8.47070754e-01 1.00310780e-01 2.66000152e-01 4.98617887e-01
2.89225281e-04 5.79446219e-02 -2.38255963e-01 7.91548073e-01
-2.98948109e-01 2.41740569e-01 2.76866674e-01 5.49619436e-01
-3.75031143e-01 -9.27684844e-01 -2.64476746e-01 7.30629414e-02
-1.02350786e-01 2.46524602e-01 -1.11555956e-01 3.91225040e-01
1.46751776e-01 5.78244805e-01 -6.21107593e-02 -5.79728365e-01
1.03232734e-01 -3.93135473e-02 2.47555435e-01 -3.35667998e-01
6.32470325e-02 3.69642466e-01 -2.90120751e-01 -5.60773015e-01
-2.77475357e-01 -6.99520886e-01 -1.30224574e+00 -3.96919787e-01
-8.28576926e-03 1.72340110e-01 -1.07048349e-02 9.56264734e-01
2.57216185e-01 6.08562946e-01 4.68868762e-01 -1.03366303e+00
-7.00024605e-01 -8.93406987e-01 -5.11027038e-01 7.06552088e-01
2.77952939e-01 -5.87102711e-01 -4.12056565e-01 2.59335518e-01]
|
[7.313658237457275, -0.12049538642168045]
|
6109d261-9050-4829-b304-b46f31c29630
|
evaluating-and-improving-the-coreference
|
2302.08464
| null |
https://arxiv.org/abs/2302.08464v1
|
https://arxiv.org/pdf/2302.08464v1.pdf
|
Evaluating and Improving the Coreference Capabilities of Machine Translation Models
|
Machine translation (MT) requires a wide range of linguistic capabilities, which current end-to-end models are expected to learn implicitly by observing aligned sentences in bilingual corpora. In this work, we ask: \emph{How well do MT models learn coreference resolution from implicit signal?} To answer this question, we develop an evaluation methodology that derives coreference clusters from MT output and evaluates them without requiring annotations in the target language. We further evaluate several prominent open-source and commercial MT systems, translating from English to six target languages, and compare them to state-of-the-art coreference resolvers on three challenging benchmarks. Our results show that the monolingual resolvers greatly outperform MT models. Motivated by this result, we experiment with different methods for incorporating the output of coreference resolution models in MT, showing improvement over strong baselines.
|
['Gabriel Stanovsky', 'Omri Abend', 'Arie Cattan', 'Asaf Yehudai']
|
2023-02-16
| null | null | null | null |
['coreference-resolution']
|
['natural-language-processing']
|
[ 2.82856107e-01 3.68356407e-01 -7.01353669e-01 -5.96964777e-01
-1.57146049e+00 -1.15120482e+00 8.54367912e-01 -4.03817862e-01
-4.79450732e-01 1.18618166e+00 6.63855672e-01 -5.00560939e-01
2.43897438e-01 -1.77252859e-01 -7.44201422e-01 -2.00075403e-01
2.95804471e-01 1.47142422e+00 -4.98877019e-02 -6.98547781e-01
1.22527555e-01 1.33667707e-01 -8.98837328e-01 7.35302985e-01
1.22053063e+00 -4.00503576e-02 2.70226002e-01 2.62887448e-01
-1.69914424e-01 6.59307122e-01 -4.95328605e-01 -9.45124090e-01
6.04148768e-02 -6.17069662e-01 -1.52925193e+00 -6.94755316e-01
7.94191599e-01 1.65248603e-01 -1.92644060e-01 1.36653185e+00
5.06897211e-01 -1.34357676e-01 3.77032846e-01 -8.37977231e-01
-5.52596807e-01 1.41729105e+00 -4.16510999e-01 5.89856982e-01
7.87282228e-01 -4.66075260e-04 1.42042792e+00 -1.07815635e+00
1.27703702e+00 1.46994638e+00 4.03677970e-01 7.46652246e-01
-1.51300192e+00 -9.67348158e-01 -8.75057653e-02 4.49573129e-01
-1.08456409e+00 -8.71326327e-01 5.80488622e-01 -1.69130802e-01
1.27446783e+00 5.02533793e-01 -3.22896451e-01 1.53750134e+00
-1.84682440e-02 8.45430672e-01 1.29342175e+00 -6.42439187e-01
-5.63328922e-01 4.90104742e-02 2.16046423e-01 2.77808964e-01
-4.79734968e-03 5.51256478e-01 -8.44636261e-01 -7.83985481e-02
1.98262140e-01 -8.27969790e-01 -6.87054157e-01 -2.01930091e-01
-1.61528873e+00 7.26595640e-01 3.74342024e-01 5.44142842e-01
-9.64274779e-02 -4.89093542e-01 4.52777177e-01 7.38737762e-01
1.52940243e-01 7.08116174e-01 -5.97749531e-01 -9.32329670e-02
-9.90517139e-01 8.68894011e-02 9.61661816e-01 1.27324271e+00
6.48156524e-01 -5.93029201e-01 -2.63631731e-01 9.38943923e-01
-4.13344540e-02 8.62693012e-01 3.40189725e-01 -1.17126071e+00
1.22868121e+00 2.25614384e-01 1.66234523e-01 -3.45519662e-01
-2.90854752e-01 -4.83404070e-01 -4.45623457e-01 -3.18655252e-01
4.38061386e-01 -9.00078267e-02 -3.12563151e-01 2.20900536e+00
2.55922247e-02 -1.39948860e-01 5.07680118e-01 1.05718017e+00
8.22187662e-01 3.87302339e-01 7.11615803e-03 -7.49431729e-01
1.23546600e+00 -1.30047798e+00 -9.49592352e-01 -4.10712302e-01
7.43312478e-01 -1.25843358e+00 1.01444137e+00 5.08318953e-02
-1.25494266e+00 -5.22209346e-01 -8.11616957e-01 -1.90678880e-01
1.34041727e-01 3.73647735e-02 2.61991501e-01 8.37900043e-02
-1.01655161e+00 5.81270635e-01 -7.25102186e-01 -6.81178212e-01
-3.36702913e-01 5.82220018e-01 -5.33383250e-01 -1.81131065e-02
-1.66117883e+00 1.61107731e+00 4.22341675e-01 -4.55188900e-02
-7.03587711e-01 -4.66503680e-01 -7.01009393e-01 -2.03478977e-01
2.18606368e-01 -8.18986833e-01 1.83530390e+00 -1.20320487e+00
-1.37758338e+00 1.40902686e+00 -6.29481256e-01 -3.91338766e-01
3.57506871e-01 -6.56109154e-01 -6.02799773e-01 1.54780725e-03
5.82645416e-01 8.21463764e-01 1.61783338e-01 -1.22253191e+00
-8.87355149e-01 -1.97220504e-01 -2.08268836e-02 5.73769808e-01
1.27773806e-01 5.63657284e-01 -2.73930937e-01 -3.47853899e-01
-5.64230345e-02 -1.08382559e+00 1.17156431e-01 -9.70303595e-01
-4.63267773e-01 -3.20212245e-01 4.97156620e-01 -6.88753963e-01
1.01057112e+00 -1.85126090e+00 6.55391634e-01 -2.69099116e-01
-2.98114747e-01 2.51280367e-01 -4.71072555e-01 4.63324606e-01
-1.27229840e-01 -1.28796190e-01 -3.62074345e-01 -1.94953650e-01
5.30208014e-02 3.81852031e-01 -7.28394628e-01 1.47247881e-01
2.20233008e-01 1.00646520e+00 -1.30629241e+00 -7.82066584e-01
-1.60965234e-01 -3.04681379e-02 -4.58636522e-01 2.59537458e-01
-2.39562333e-01 9.18395936e-01 -2.22870141e-01 4.14700031e-01
4.54386026e-01 8.36605951e-02 9.05236065e-01 -2.17713907e-01
-2.90506631e-01 1.22952187e+00 -6.15756631e-01 2.19109297e+00
-3.48727703e-01 6.76877081e-01 1.22405678e-01 -8.23189080e-01
8.21774960e-01 6.95181608e-01 5.54199107e-02 -8.59738946e-01
-4.16936204e-02 6.53189242e-01 4.22892839e-01 -3.14081967e-01
5.43124735e-01 -2.55708963e-01 -2.93814003e-01 6.96218848e-01
4.45096493e-01 -3.15042697e-02 2.48664945e-01 1.79702938e-01
8.79744887e-01 4.27990735e-01 2.26902157e-01 -5.87034643e-01
8.04992318e-01 7.91502953e-01 8.70199382e-01 5.95140755e-01
-2.01545015e-01 4.10867333e-01 6.50718212e-02 -2.20077500e-01
-7.38012791e-01 -1.34887850e+00 -1.01202950e-01 1.27273226e+00
2.74831772e-01 -2.89196670e-01 -8.44748616e-01 -9.02064919e-01
-5.21065712e-01 9.70072746e-01 -1.78996637e-01 9.93257388e-02
-1.49024761e+00 -6.11055434e-01 8.95794392e-01 3.92124802e-01
1.98511988e-01 -1.37933028e+00 -1.10580117e-01 2.90381819e-01
-1.39963138e+00 -1.33280444e+00 -8.56724739e-01 1.54869169e-01
-1.00596571e+00 -1.20774019e+00 -2.34951079e-01 -1.25463569e+00
1.91881016e-01 2.33890980e-01 1.85432923e+00 -2.13179305e-01
4.82935995e-01 -4.74648289e-02 -2.31328800e-01 -6.84511214e-02
-7.21804082e-01 6.69127643e-01 3.23883444e-01 -6.98002815e-01
1.08698344e+00 -5.85386336e-01 -1.01013832e-01 4.07314837e-01
-2.39443913e-01 1.05639949e-01 7.08737254e-01 9.48176324e-01
5.08726537e-01 -9.03156817e-01 4.23113823e-01 -1.02147901e+00
6.72279119e-01 -2.94671506e-01 -4.39617366e-01 4.69327152e-01
-5.09718359e-01 4.15687442e-01 3.39949101e-01 -4.06472534e-01
-1.34800887e+00 -1.29533336e-01 -7.18268901e-02 -2.29278117e-01
5.43337734e-03 4.50387716e-01 -3.35433573e-01 3.55932534e-01
8.05976212e-01 -2.11316790e-03 -3.06304455e-01 -5.98461807e-01
7.16690361e-01 7.27197707e-01 1.37268412e+00 -1.18970156e+00
6.97790325e-01 3.86632532e-02 -4.51443315e-01 -8.89504254e-02
-1.28385103e+00 -5.66388607e-01 -1.08033478e+00 2.98780620e-01
6.07076705e-01 -1.04119253e+00 -3.14988554e-01 -2.71110266e-01
-1.70620215e+00 -2.85756052e-01 1.06335416e-01 8.03979158e-01
-8.02027524e-01 2.58857548e-01 -9.67722893e-01 5.42142382e-03
-9.00855422e-01 -1.28645408e+00 8.77527475e-01 -1.28057860e-02
-9.76384699e-01 -8.68146420e-01 7.73646355e-01 7.27891982e-01
1.11189894e-01 -1.77875698e-01 9.49666977e-01 -8.51572573e-01
-3.66174728e-01 6.54382586e-01 -5.65334596e-02 -1.63896203e-01
5.12539856e-02 -3.91687214e-01 -8.65710139e-01 -3.73030901e-01
6.01924844e-02 -4.00431186e-01 5.46115398e-01 5.35048768e-02
-1.08793275e-02 -3.28077823e-01 -7.00223923e-01 6.62769020e-01
1.11194980e+00 -1.91016614e-01 4.48728085e-01 6.77063942e-01
4.05254304e-01 7.70985067e-01 1.04300261e+00 -6.18209660e-01
5.48794031e-01 1.01487625e+00 1.26661032e-01 8.88575837e-02
-3.39888364e-01 -3.06246191e-01 7.91748047e-01 1.36037171e+00
-1.03408396e-01 5.68966605e-02 -9.88994956e-01 7.49723852e-01
-1.98300803e+00 -1.27933168e+00 -4.82482553e-01 1.96569645e+00
1.61250865e+00 6.53552637e-02 -1.53828412e-01 -4.81474608e-01
1.04030335e+00 -5.21055013e-02 -1.69705629e-01 -5.96640468e-01
-4.56441462e-01 5.76303720e-01 6.03592321e-02 1.02713597e+00
-1.01858509e+00 1.65149260e+00 6.68076229e+00 2.73502201e-01
-1.15327919e+00 4.74791259e-01 -1.58772811e-01 -6.07669093e-02
-2.30329826e-01 3.81264627e-01 -1.03745639e+00 4.35323939e-02
1.13353658e+00 -2.71645188e-01 4.75905806e-01 3.28067422e-01
3.33906226e-02 4.62575972e-01 -1.75736725e+00 6.92243695e-01
5.25883697e-02 -1.03997517e+00 1.44080520e-01 -1.62668124e-01
8.21875513e-01 7.56412208e-01 -1.82100788e-01 5.99826157e-01
9.14478838e-01 -1.01618469e+00 6.70054555e-01 9.60213840e-02
9.40467775e-01 -4.72923934e-01 8.65617812e-01 4.96346503e-01
-8.84180307e-01 2.62815922e-01 -3.47218424e-01 -2.74576955e-02
5.04219294e-01 2.00739931e-02 -7.97429264e-01 9.31590497e-01
4.83461797e-01 5.15404999e-01 -1.84952483e-01 6.37102306e-01
-9.31046307e-01 6.10789001e-01 -6.10897541e-02 4.90463525e-01
2.43280157e-01 -1.79257527e-01 8.93597543e-01 1.68768668e+00
1.61015466e-01 1.82114854e-01 5.41553386e-02 9.45267975e-01
-4.68622476e-01 8.57915357e-02 -5.70050478e-01 2.87826955e-01
9.20206428e-01 1.18520427e+00 9.88650993e-02 -2.98232049e-01
-4.69780892e-01 1.10240006e+00 7.71654963e-01 2.87021399e-01
-6.08627856e-01 -3.27912532e-02 8.32970262e-01 -3.37414950e-01
-1.23320311e-01 4.76165414e-02 -1.45765468e-01 -1.48998702e+00
-6.98683411e-02 -1.56533730e+00 7.62245595e-01 -5.65440357e-01
-1.26523423e+00 8.86054218e-01 -6.56978041e-02 -1.14113641e+00
-7.41677701e-01 -3.00319850e-01 -6.10936880e-01 1.12909210e+00
-1.56065416e+00 -1.21652448e+00 2.80735523e-01 6.39261782e-01
7.17774808e-01 -2.52975449e-02 1.09744585e+00 3.71892244e-01
-1.61173448e-01 7.12479532e-01 -9.45732668e-02 5.46658993e-01
1.49963200e+00 -1.14947593e+00 7.03589737e-01 9.57626343e-01
6.72507405e-01 9.82453287e-01 1.03965294e+00 -5.35443962e-01
-1.19536197e+00 -8.51651430e-01 1.85942721e+00 -1.06439972e+00
6.57010972e-01 -1.57238707e-01 -8.39024723e-01 1.27054203e+00
8.72832358e-01 -4.48737711e-01 5.92836738e-01 7.30759561e-01
-5.73641062e-01 2.17703581e-01 -8.44850898e-01 5.58334708e-01
1.54635608e+00 -8.00489366e-01 -1.81734359e+00 1.18621692e-01
7.34319866e-01 -6.23218000e-01 -8.12669158e-01 9.22326744e-01
1.97085917e-01 -5.67111552e-01 9.07930076e-01 -9.75598812e-01
4.36779261e-01 -2.45509967e-01 -3.22901964e-01 -1.60421610e+00
-4.82480675e-01 -7.20569670e-01 -7.58891867e-04 1.40550661e+00
8.39502573e-01 -1.81265116e-01 3.22459310e-01 8.83185565e-02
-3.33045393e-01 7.46919066e-02 -9.96308625e-01 -6.67907834e-01
6.93618715e-01 5.80409169e-02 4.11758780e-01 1.45259917e+00
6.71492517e-01 1.21482384e+00 -1.26742810e-01 3.29256862e-01
8.32448542e-01 6.96221471e-01 7.85856545e-01 -1.28743982e+00
-2.45869100e-01 -4.89842147e-01 3.69155496e-01 -1.07476473e+00
8.50238621e-01 -1.38193750e+00 2.06433222e-01 -1.13228297e+00
6.33927166e-01 -2.59786993e-01 -2.13394910e-01 4.88710582e-01
-4.52071398e-01 3.19407284e-01 1.69824436e-01 7.97686815e-01
-6.53271258e-01 3.06823641e-01 1.09721470e+00 -3.61487478e-01
-9.38557014e-02 -2.77597815e-01 -7.48623908e-01 5.56932807e-01
6.02095425e-01 -6.75368309e-01 -3.20625454e-02 -1.08929479e+00
2.37684958e-02 2.68648475e-01 -2.13960454e-01 -4.23529565e-01
3.56704623e-01 -2.80185848e-01 -1.65965974e-01 -6.58147633e-01
-5.49226552e-02 -2.90495157e-01 1.17869452e-01 3.90644491e-01
-4.78478760e-01 3.08226049e-01 1.79699555e-01 -2.42437378e-01
-5.86345315e-01 -1.90784961e-01 6.25825942e-01 -2.04772502e-01
-5.57825625e-01 -1.95620805e-01 -3.12955268e-02 6.54078007e-01
3.42085302e-01 4.15306062e-01 -7.57111371e-01 -6.96667433e-02
-7.19782293e-01 4.73761052e-01 5.33947766e-01 8.99617076e-01
-7.43501587e-03 -1.38730133e+00 -1.19609201e+00 -3.39600652e-01
2.69530267e-01 -3.59397173e-01 -5.89231253e-01 1.08256853e+00
-6.80224821e-02 9.86892879e-01 -2.61782557e-01 -7.62197137e-01
-1.43166995e+00 6.19979084e-01 4.35903341e-01 -6.16743028e-01
-3.89930427e-01 5.17448187e-01 1.91716149e-01 -1.26694381e+00
1.10525405e-02 1.45741224e-01 -1.59767464e-01 -2.02752039e-01
4.13142264e-01 -2.86896527e-03 6.42438605e-02 -1.03467250e+00
-6.09469831e-01 5.26127815e-01 -2.91613311e-01 -5.40576875e-01
1.02364707e+00 -3.72177392e-01 -3.19739074e-01 2.26938277e-01
7.03209281e-01 2.68637508e-01 -6.08894467e-01 -7.69112289e-01
8.20831120e-01 -7.47825727e-02 -4.77220923e-01 -1.20038021e+00
-5.64177930e-01 8.09532225e-01 2.53680021e-01 -6.56869233e-01
7.97205389e-01 3.18976820e-01 8.37680876e-01 6.42539740e-01
7.68855691e-01 -1.03927279e+00 -3.70843798e-01 1.14895034e+00
8.13576221e-01 -1.30868971e+00 -4.24390405e-01 -4.55376595e-01
-5.49767077e-01 8.72040153e-01 6.46583140e-01 2.16445357e-01
-1.41262010e-01 3.32495630e-01 7.03671217e-01 -1.07854985e-01
-9.10498142e-01 -2.91934401e-01 2.55092412e-01 5.35386384e-01
1.00881433e+00 2.17624620e-01 -8.81597817e-01 5.70198178e-01
-6.72625721e-01 -1.01760663e-01 1.59051985e-01 5.72335541e-01
-3.43774378e-01 -1.79292822e+00 -4.72707450e-01 -3.40188056e-01
-8.10802817e-01 -5.99343598e-01 -9.58234489e-01 7.99037695e-01
-5.01409061e-02 1.11518812e+00 -9.95444134e-02 -1.92432895e-01
3.94251913e-01 2.32117116e-01 7.63809264e-01 -6.82548106e-01
-8.78920197e-01 -8.84092674e-02 6.14816010e-01 -5.30362427e-01
-7.07338452e-01 -1.04518676e+00 -1.25987542e+00 -1.76358819e-01
-3.06981802e-01 7.42230535e-01 1.45265058e-01 1.21709478e+00
1.54877484e-01 3.07514076e-03 4.12285507e-01 -5.38068235e-01
-7.88427472e-01 -1.37064385e+00 3.45262378e-01 6.41140759e-01
9.93055999e-02 -4.22068506e-01 -2.37068370e-01 1.55834019e-01]
|
[9.273039817810059, 9.560680389404297]
|
a37ec121-9a6e-4966-a405-1d841eee76b3
|
hgr-net-a-fusion-network-for-hand-gesture
|
1806.05653
| null |
https://arxiv.org/abs/1806.05653v3
|
https://arxiv.org/pdf/1806.05653v3.pdf
|
HGR-Net: A Fusion Network for Hand Gesture Segmentation and Recognition
|
We propose a two-stage convolutional neural network (CNN) architecture for robust recognition of hand gestures, called HGR-Net, where the first stage performs accurate semantic segmentation to determine hand regions, and the second stage identifies the gesture. The segmentation stage architecture is based on the combination of fully convolutional residual network and atrous spatial pyramid pooling. Although the segmentation sub-network is trained without depth information, it is particularly robust against challenges such as illumination variations and complex backgrounds. The recognition stage deploys a two-stream CNN, which fuses the information from the red-green-blue and segmented images by combining their deep representations in a fully connected layer before classification. Extensive experiments on public datasets show that our architecture achieves almost as good as state-of-the-art performance in segmentation and recognition of static hand gestures, at a fraction of training time, run time, and model size. Our method can operate at an average of 23 ms per frame.
|
['Amirhossein Dadashzadeh', 'Majid Mirmehdi', 'Alireza Tavakoli Targhi', 'Maryam Tahmasbi']
|
2018-06-14
| null | null | null | null |
['hand-segmentation']
|
['computer-vision']
|
[ 4.79266495e-01 -2.83764690e-01 -3.14432919e-01 -3.53302807e-01
-6.61040068e-01 -5.75272739e-01 1.52545825e-01 -5.85194886e-01
-7.39549220e-01 2.97507234e-02 -1.64915640e-02 -2.55873561e-01
4.42679137e-01 -5.18929303e-01 -5.83353400e-01 -7.33140469e-01
1.67229176e-01 3.23244005e-01 6.42904401e-01 8.36330280e-02
2.13947430e-01 9.68509972e-01 -1.38519824e+00 4.52668190e-01
4.08913374e-01 1.19897878e+00 -2.95496103e-03 1.10218894e+00
-2.45790437e-01 1.04773998e+00 -7.68199146e-01 -8.05110559e-02
5.52081287e-01 -4.33317199e-02 -1.04382837e+00 9.82574373e-02
3.81082088e-01 -1.00153565e+00 -8.76208365e-01 7.40287423e-01
8.23381722e-01 1.49223641e-01 2.01847821e-01 -9.34249163e-01
-2.94459999e-01 4.54790771e-01 -6.08423114e-01 1.53891832e-01
2.33059853e-01 6.71755850e-01 6.45266175e-01 -8.44121516e-01
5.42791724e-01 1.40475011e+00 5.42340398e-01 7.80183792e-01
-1.06157005e+00 -7.49118924e-01 3.49280864e-01 -1.74326882e-01
-1.32738054e+00 -2.30430886e-01 3.60667378e-01 -3.62696826e-01
1.07291234e+00 1.51411369e-01 5.57414651e-01 1.10534239e+00
-3.09477836e-01 1.24255073e+00 1.00812829e+00 -1.71129480e-01
6.93689063e-02 -7.84087002e-01 5.01778185e-01 7.85329401e-01
-2.71686502e-02 -2.92466357e-02 -3.75083059e-01 2.68249840e-01
1.28294146e+00 3.34073752e-01 8.51732492e-03 2.61523932e-01
-1.05351233e+00 2.34322906e-01 7.62469828e-01 1.24680161e-01
-6.09943748e-01 4.00959969e-01 1.05591543e-01 1.19656650e-02
2.14468054e-02 -4.77295928e-02 -4.75115836e-01 -1.86986461e-01
-1.23976171e+00 5.56870550e-02 7.62991965e-01 8.65372002e-01
5.55287421e-01 -1.36630451e-02 -4.52149063e-01 6.60778701e-01
3.31214577e-01 6.26372337e-01 1.73854724e-01 -9.59675074e-01
5.95730066e-01 7.62008607e-01 -1.39732137e-01 -4.42218930e-01
-4.27920729e-01 -7.74082243e-02 -5.98201156e-01 5.42224586e-01
8.18893254e-01 -4.27518547e-01 -1.85815966e+00 1.23588061e+00
1.45716295e-01 6.95054457e-02 -2.63056874e-01 1.36501014e+00
1.10234690e+00 3.56760204e-01 4.19637620e-01 5.74999630e-01
1.23244476e+00 -1.21659744e+00 -2.27908239e-01 -3.40699345e-01
-1.02624372e-01 -7.89021373e-01 8.96816611e-01 1.91508397e-01
-1.25668228e+00 -6.93428934e-01 -7.68043578e-01 -4.23119515e-01
-2.61890560e-01 4.05437410e-01 6.18438721e-01 6.81012392e-01
-1.03793669e+00 5.36163390e-01 -1.27649450e+00 -3.71681899e-01
7.78165817e-01 5.98717391e-01 -1.43428266e-01 -8.98051262e-02
-4.29169446e-01 4.07373697e-01 3.29097152e-01 5.80641508e-01
-8.86680186e-01 -4.69844550e-01 -5.77432990e-01 1.12899639e-01
2.02614427e-01 -1.53492883e-01 1.19556212e+00 -1.09023046e+00
-1.70534313e+00 7.74779618e-01 -4.74445403e-01 -5.79957217e-02
7.61259377e-01 -3.42066050e-01 3.47572193e-02 4.09173369e-01
-2.66216278e-01 8.90241385e-01 8.90207887e-01 -1.07088411e+00
-6.64366484e-01 -6.48825407e-01 1.48887902e-01 3.86725590e-02
9.77815390e-02 5.33550203e-01 -1.19790685e+00 -7.30330706e-01
4.72164482e-01 -9.64102209e-01 -3.01945031e-01 -4.48415987e-02
-6.21770740e-01 -3.73702012e-02 1.03664982e+00 -1.10633743e+00
9.80446935e-01 -1.92468417e+00 1.90432873e-02 5.40602386e-01
1.06393307e-01 6.02791607e-01 -4.85991001e-01 -5.78770600e-02
7.24643171e-02 1.76449344e-02 -8.85326117e-02 -2.06243709e-01
-1.56219557e-01 5.29320836e-02 -2.88956493e-01 2.30320871e-01
3.88943672e-01 1.27851665e+00 -6.39099240e-01 -4.27225500e-01
3.42279702e-01 7.45539188e-01 -2.84767359e-01 5.20845652e-01
-3.30404103e-01 6.75889432e-01 -3.98856997e-01 1.19595444e+00
5.24751008e-01 -3.66121650e-01 2.87572235e-01 -1.67159691e-01
-4.94679026e-02 3.34940195e-01 -1.19809353e+00 1.75529552e+00
1.46217796e-03 8.17102313e-01 3.51329774e-01 -5.23197293e-01
5.87211370e-01 2.59247154e-01 4.48446989e-01 -5.02636969e-01
3.88421774e-01 5.99860698e-02 -3.94971669e-02 -5.47307611e-01
2.97415853e-01 5.56644857e-01 2.55142599e-01 7.03494549e-01
-4.90545807e-03 3.10425907e-01 9.17754918e-02 -8.88509303e-02
1.14663959e+00 3.83448690e-01 -3.69143248e-01 2.37362683e-01
2.66316921e-01 -5.75626269e-02 3.68968636e-01 7.12573886e-01
-3.00388545e-01 8.50529194e-01 5.04156172e-01 -6.89798057e-01
-7.82370448e-01 -8.93422484e-01 2.92316854e-01 1.47982645e+00
2.69546390e-01 1.50405377e-01 -1.12561059e+00 -6.91039860e-01
-4.70064208e-02 -7.95429200e-02 -6.34244323e-01 4.53276157e-01
-1.11779046e+00 -5.34697294e-01 9.82926428e-01 1.29946959e+00
1.00507796e+00 -1.42259765e+00 -1.03928256e+00 5.41572310e-02
8.16914588e-02 -1.07341349e+00 -5.01698673e-01 1.13587812e-01
-7.09655344e-01 -1.44435775e+00 -1.16040516e+00 -9.35289741e-01
7.25923479e-01 1.42649844e-01 7.51293600e-01 4.67563599e-01
-6.95646167e-01 1.76775798e-01 -3.88562918e-01 -2.12340415e-01
5.03774658e-02 3.49718481e-01 -5.95182121e-01 -2.44453996e-02
6.36689126e-01 -2.61798799e-01 -9.54291165e-01 3.28037322e-01
-6.77303791e-01 -8.62790272e-02 7.76018083e-01 6.33869290e-01
4.83177781e-01 -5.05526483e-01 4.59093377e-02 -4.29755062e-01
2.06333265e-01 1.44629925e-01 -5.23809612e-01 4.31943774e-01
-1.01540729e-01 -8.14397335e-02 2.36079082e-01 -4.10970896e-01
-1.09066713e+00 6.67416096e-01 -3.60849023e-01 -3.96873683e-01
-5.07368028e-01 -1.33028463e-01 -4.79636528e-02 -3.13127905e-01
3.54532301e-01 8.40529427e-02 -1.36081669e-02 -6.67899430e-01
4.62714881e-01 1.09637392e+00 8.36053431e-01 -4.65624720e-01
5.36192238e-01 5.57535648e-01 -3.41554314e-01 -7.20044374e-01
-6.22836173e-01 -5.93403459e-01 -1.12801158e+00 -2.76697814e-01
1.17488015e+00 -6.57606483e-01 -8.94137025e-01 1.06021595e+00
-1.27384114e+00 -9.10223544e-01 -6.76476508e-02 1.92857638e-01
-1.51645780e-01 1.29827783e-01 -1.06226468e+00 -8.29982519e-01
-6.27541602e-01 -1.39052129e+00 1.33166790e+00 6.13440454e-01
2.03834642e-02 -4.20645803e-01 -4.20963615e-01 5.28327763e-01
4.09997821e-01 2.81377822e-01 5.18054843e-01 -6.64814830e-01
-9.84334528e-01 -2.17902035e-01 -7.53201246e-01 2.83303171e-01
8.22682157e-02 1.34665653e-01 -1.28205585e+00 -3.83848965e-01
-7.79421926e-01 -5.02221406e-01 1.25882280e+00 4.69726235e-01
1.59016252e+00 2.59474777e-02 -1.87822804e-01 8.46821070e-01
1.24294388e+00 4.03951526e-01 8.60489428e-01 -1.82356965e-02
1.09561872e+00 4.43744868e-01 1.94997981e-01 2.15037808e-01
2.01361895e-01 2.79556006e-01 3.61452729e-01 -6.08642578e-01
-5.94238818e-01 2.63994392e-02 -1.15285143e-02 1.26490116e-01
-5.89806855e-01 5.01178838e-02 -1.10936618e+00 4.18734610e-01
-1.74208498e+00 -9.24025714e-01 1.96677193e-01 1.93660581e+00
7.56553948e-01 -2.41480127e-01 5.20446897e-01 3.19493711e-02
7.37518609e-01 3.89152318e-01 -9.06475425e-01 -9.85188261e-02
-6.70842901e-02 8.68760824e-01 7.19331563e-01 3.72876257e-01
-1.37397742e+00 1.59146225e+00 7.11471176e+00 3.56771290e-01
-1.48461318e+00 -1.07537054e-01 6.20273948e-01 -2.29743302e-01
4.30195242e-01 -4.50119346e-01 -6.52553439e-01 2.06426576e-01
4.71293628e-01 8.78538907e-01 7.75255919e-01 9.22062516e-01
-8.33508000e-02 2.38827914e-02 -8.00110519e-01 9.58778977e-01
1.49218170e-02 -1.10389602e+00 -4.62385006e-02 -9.19102877e-02
7.17384517e-01 3.94913524e-01 -1.07425906e-01 3.03436890e-02
4.99726683e-01 -1.44190407e+00 7.13903725e-01 2.60946214e-01
1.06526947e+00 -5.96896589e-01 7.15437710e-01 9.95637011e-03
-1.43435478e+00 -1.06034264e-01 1.10722534e-01 2.88046561e-02
-7.84998015e-02 -1.02464490e-01 -5.68462849e-01 6.06174767e-02
9.36888993e-01 4.10050273e-01 -4.79092151e-01 8.23318541e-01
-5.42364955e-01 6.53415442e-01 -2.82005012e-01 7.92581737e-02
3.18845421e-01 2.48813376e-01 9.89036039e-02 1.60476160e+00
-3.06738555e-01 5.83410323e-01 3.03312719e-01 7.41990745e-01
-1.59797251e-01 -3.05509508e-01 6.22064173e-02 -2.89928883e-01
4.71668780e-01 1.14331698e+00 -1.17196500e+00 -5.29078722e-01
-2.98833519e-01 1.36334455e+00 1.93360329e-01 7.91327536e-01
-5.99939764e-01 -7.82387197e-01 7.84223318e-01 -3.72154683e-01
5.73205948e-01 -3.18396568e-01 -8.50631654e-01 -1.03618038e+00
2.10553959e-01 -8.36054802e-01 3.06540728e-01 -3.88569921e-01
-8.66559923e-01 5.10755002e-01 -5.62837660e-01 -5.58945239e-01
-4.90473509e-02 -1.00249076e+00 -5.97068846e-01 9.28535044e-01
-1.64583611e+00 -1.29887211e+00 -8.02477896e-01 7.85947800e-01
6.61444128e-01 -4.59743850e-02 7.67257214e-01 2.75882095e-01
-9.09200847e-01 5.94417989e-01 -3.89830351e-01 1.13162005e+00
3.74875963e-01 -1.24318302e+00 8.96235704e-01 1.05165315e+00
-1.48709267e-01 6.69378638e-01 -1.10234223e-01 -7.69646943e-01
-1.36666977e+00 -1.17420483e+00 3.96564394e-01 -2.73567051e-01
1.77812114e-01 -2.00323790e-01 -7.01936424e-01 8.93174410e-01
-2.62688249e-02 2.11049512e-01 3.91395658e-01 -4.01020646e-02
-8.53088856e-01 2.04894617e-01 -1.10852098e+00 5.24271309e-01
1.24408817e+00 -5.49310625e-01 -5.52363098e-01 1.94913879e-01
3.44730437e-01 -9.41943228e-01 -6.79568291e-01 3.42617601e-01
1.12120259e+00 -7.63375163e-01 1.10574102e+00 -7.40856290e-01
3.43442082e-01 -1.66898459e-01 -1.00291274e-01 -5.03012776e-01
-1.04443565e-01 -6.99252665e-01 -1.36772960e-01 7.34320343e-01
2.47385502e-01 -3.25108588e-01 1.20827472e+00 9.93375421e-01
2.02278495e-01 -7.89210796e-01 -6.69056058e-01 -6.45717680e-01
-1.58256546e-01 -4.26280022e-01 7.12264597e-01 4.19561118e-01
-3.67666304e-01 -9.39144641e-02 7.85046816e-02 2.20716283e-01
5.52359641e-01 2.10772514e-01 9.03542936e-01 -9.94453847e-01
-7.69043639e-02 -6.52946174e-01 -3.15194279e-01 -1.46229124e+00
9.71393511e-02 -5.75306773e-01 3.61510217e-01 -1.74981594e+00
2.71166086e-01 -4.06964630e-01 -2.99701005e-01 1.00726318e+00
-2.41853282e-01 6.78493857e-01 4.74072069e-01 3.71945381e-01
-5.93931496e-01 -1.31581903e-01 1.19004810e+00 -2.83141881e-01
-6.34528756e-01 -1.10699706e-01 -3.10059398e-01 7.98071682e-01
7.28531301e-01 -2.51973681e-02 1.55475304e-01 -8.75781596e-01
-7.22374022e-01 -2.48367608e-01 6.01193845e-01 -1.04468656e+00
3.62448514e-01 -1.65585130e-01 6.62115097e-01 -8.04533780e-01
3.42604488e-01 -7.39176631e-01 -2.95025557e-01 6.46996856e-01
-4.99021292e-01 -1.25770614e-01 1.09185696e-01 2.33869106e-01
-3.90225053e-02 3.72101843e-01 8.62480223e-01 -1.77180231e-01
-9.48790610e-01 4.47879702e-01 -1.46627575e-01 -2.10623384e-01
7.54438877e-01 -4.38507199e-01 -3.91386837e-01 -6.59527630e-02
-6.47302091e-01 2.27301210e-01 2.30873153e-01 5.01152098e-01
5.63693583e-01 -8.53812575e-01 -4.42893058e-01 3.39580476e-01
-4.04616743e-01 3.42171729e-01 1.45773575e-01 5.76243520e-01
-1.00196242e+00 4.82981861e-01 -2.44032159e-01 -7.84894705e-01
-1.45989454e+00 -6.30995408e-02 3.88794512e-01 6.75609633e-02
-7.78504670e-01 1.24510503e+00 -2.57482648e-01 -2.45704606e-01
9.46235776e-01 -5.41715682e-01 1.34884268e-01 -1.80187702e-01
9.43887532e-01 6.17549896e-01 -2.71935156e-03 -7.89133489e-01
-5.54884076e-01 8.83837283e-01 -5.98142557e-02 -3.69246423e-01
1.18107867e+00 4.72351074e-01 -1.91870257e-01 -9.54464450e-02
1.02951646e+00 -4.91886199e-01 -1.78451931e+00 -1.07294723e-01
-2.19840854e-02 -4.44602281e-01 6.36559650e-02 -1.27529418e+00
-1.47538304e+00 1.06649351e+00 6.94069207e-01 -3.40827823e-01
1.25562155e+00 -6.09504580e-02 1.24937499e+00 5.22446990e-01
3.19081545e-01 -1.18592238e+00 5.89863323e-02 7.45888054e-01
7.23330319e-01 -1.07911015e+00 -3.12683642e-01 -3.75366837e-01
-3.67798567e-01 1.36777186e+00 6.78723097e-01 -3.62926185e-01
4.87050593e-01 5.75938642e-01 5.63946366e-01 -7.22229108e-02
-7.39478245e-02 -6.54775023e-01 3.83673012e-01 5.17684698e-01
3.88349533e-01 9.74455848e-02 9.96375978e-02 4.45689946e-01
7.95335248e-02 1.31169349e-01 -3.25811237e-01 1.19245327e+00
-2.89034516e-01 -8.96914124e-01 -3.83979321e-01 3.16497475e-01
-5.34129560e-01 6.66077361e-02 -8.14620674e-01 7.83701658e-01
3.89718451e-02 9.65211332e-01 2.23760754e-01 -7.35111237e-01
4.50303465e-01 1.99357599e-01 6.59714162e-01 -5.76101661e-01
-1.12911618e+00 2.67592639e-01 -3.02433401e-01 -1.18446410e+00
-5.20848453e-01 -4.24353451e-01 -1.61800706e+00 -2.35747144e-01
-1.46355987e-01 -5.66376865e-01 8.01855206e-01 1.05743778e+00
3.74435157e-01 6.32632077e-01 2.95637429e-01 -1.43827510e+00
-3.92580777e-01 -9.43217933e-01 -3.88084024e-01 2.47358456e-01
4.15750206e-01 -2.74602026e-01 1.06251158e-01 2.07346827e-01]
|
[6.628302097320557, -0.5812715888023376]
|
2924d1c7-4722-4899-b515-23d98699c98c
|
automatic-controllable-product-copywriting
|
2206.10103
| null |
https://arxiv.org/abs/2206.10103v1
|
https://arxiv.org/pdf/2206.10103v1.pdf
|
Automatic Controllable Product Copywriting for E-Commerce
|
Automatic product description generation for e-commerce has witnessed significant advancement in the past decade. Product copywriting aims to attract users' interest and improve user experience by highlighting product characteristics with textual descriptions. As the services provided by e-commerce platforms become diverse, it is necessary to adapt the patterns of automatically-generated descriptions dynamically. In this paper, we report our experience in deploying an E-commerce Prefix-based Controllable Copywriting Generation (EPCCG) system into the JD.com e-commerce product recommendation platform. The development of the system contains two main components: 1) copywriting aspect extraction; 2) weakly supervised aspect labeling; 3) text generation with a prefix-based language model; 4) copywriting quality control. We conduct experiments to validate the effectiveness of the proposed EPCCG. In addition, we introduce the deployed architecture which cooperates with the EPCCG into the real-time JD.com e-commerce recommendation platform and the significant payoff since deployment.
|
['Lingfei Wu', 'Bo Long', 'Yun Xiao', 'Meng Jiang', 'Qingkai Zeng', 'Xiaojie Guo']
|
2022-06-21
| null | null | null | null |
['product-recommendation', 'aspect-extraction']
|
['miscellaneous', 'natural-language-processing']
|
[-3.49572524e-02 1.78435922e-01 -3.56983662e-01 -5.18066943e-01
-6.50402546e-01 -8.56886029e-01 6.10865593e-01 2.35376030e-01
1.46857455e-01 1.51174301e-02 3.45472723e-01 -2.23898560e-01
-6.71124235e-02 -9.21216846e-01 -2.56914258e-01 -3.53875346e-02
1.55009627e-01 8.02435100e-01 2.07592860e-01 -6.06357634e-01
4.46938366e-01 3.31678808e-01 -1.63405323e+00 8.15507829e-01
4.92385566e-01 1.01352942e+00 1.83485076e-01 5.24594426e-01
-6.89734161e-01 3.47304195e-01 -3.60083520e-01 -8.65066230e-01
5.07841825e-01 -2.93446571e-01 -5.55580020e-01 9.28407550e-01
-2.56842691e-02 -1.52173638e-01 4.42870617e-01 9.69252467e-01
1.59951717e-01 -5.65559901e-02 3.22966009e-01 -1.42357314e+00
-7.64362276e-01 1.05231106e+00 -2.53990322e-01 -2.96336383e-01
3.73762250e-01 7.48596787e-02 1.26079977e+00 -9.34086084e-01
1.01393926e+00 9.35250401e-01 4.04437155e-01 2.44253218e-01
-8.62609327e-01 -1.96983233e-01 3.30380738e-01 -1.56808361e-01
-1.25008309e+00 -3.19564849e-01 9.20482874e-01 -4.22826409e-01
1.32271183e+00 2.96000183e-01 7.47949421e-01 5.82863390e-01
1.03387952e-01 9.83848393e-01 6.50550187e-01 -5.35205007e-01
3.38736385e-01 9.73936200e-01 4.22001988e-01 5.41030645e-01
2.54556566e-01 -3.02666843e-01 -4.12197649e-01 -2.56853580e-01
6.59915209e-01 -8.11656862e-02 3.69007200e-01 -3.31466943e-01
-7.29567468e-01 7.72993147e-01 -2.27565452e-01 3.29701334e-01
-6.82611763e-01 -2.55182445e-01 4.98258173e-01 3.16359431e-01
5.11423290e-01 5.99016786e-01 -7.49627471e-01 -2.07606882e-01
-6.96696222e-01 4.82566357e-01 1.27106917e+00 2.13371396e+00
2.97316074e-01 -1.48295611e-01 -7.24095851e-02 9.91986930e-01
7.45085597e-01 2.28623495e-01 4.36293155e-01 -7.00667918e-01
3.06051731e-01 8.08226585e-01 2.96476632e-01 -9.12326217e-01
-9.97379199e-02 -7.27377415e-01 -1.14219025e-01 -2.12155357e-01
-1.20714031e-01 -9.04835947e-03 -4.33531404e-01 9.20767784e-01
1.72500536e-01 -6.63571358e-01 6.97516352e-02 7.05049336e-01
7.56269813e-01 6.55831099e-01 2.25464609e-02 -2.49849424e-01
1.61264038e+00 -1.24469888e+00 -9.14302289e-01 1.05904996e-01
7.05063105e-01 -1.30723417e+00 1.31150115e+00 5.53535819e-01
-1.22045481e+00 -5.73392093e-01 -1.17358160e+00 2.60987043e-01
-4.78867978e-01 3.70202363e-01 8.76230359e-01 9.39319074e-01
-6.52476251e-01 2.04415679e-01 -5.04683375e-01 -6.22229874e-01
-1.54322041e-02 3.63482058e-01 8.23326781e-02 1.15227915e-01
-8.58614564e-01 3.14267099e-01 3.47156733e-01 -3.22759926e-01
-3.40343237e-01 -5.41188002e-01 -8.87111843e-01 1.40256315e-01
4.83065784e-01 -6.25975609e-01 1.69545949e+00 -1.20464182e+00
-1.66436374e+00 6.78530216e-01 1.95970848e-01 -7.48510957e-02
2.24669978e-01 -2.08410963e-01 -9.84812498e-01 -1.00119904e-01
1.77451894e-01 4.34085757e-01 5.00868380e-01 -1.44873190e+00
-1.07266319e+00 -3.69117320e-01 1.24505974e-01 1.19282760e-01
-2.56924808e-01 2.96764463e-01 -8.78222346e-01 -7.48492002e-01
-2.21301448e-02 -7.59913445e-01 -2.37292305e-01 -4.01376516e-01
-1.92078382e-01 -3.11285049e-01 7.43802369e-01 -3.78985435e-01
1.47655594e+00 -2.08928084e+00 -5.86201131e-01 3.13968748e-01
-9.44326445e-02 1.71858892e-01 -3.60009223e-01 7.75626183e-01
3.53287369e-01 2.21593276e-01 3.84832859e-01 -1.43610388e-01
5.12522399e-01 -5.24601564e-02 -1.33293360e-01 -1.88927636e-01
1.45285085e-01 7.88255274e-01 -6.71039879e-01 -3.75855148e-01
-5.55578060e-02 1.05196834e-01 -8.80575359e-01 9.60287303e-02
-8.20907652e-01 -3.63165706e-01 -6.84297979e-01 9.06225383e-01
6.12593293e-01 -1.61272466e-01 4.38091367e-01 -3.65311444e-01
-3.99912298e-01 2.98140317e-01 -1.21069062e+00 1.53619742e+00
-7.07996130e-01 -1.24591284e-01 2.88561247e-02 -8.75939280e-02
1.14842749e+00 2.80782789e-01 5.09038031e-01 -7.92408705e-01
3.33516270e-01 2.97732443e-01 -2.73978919e-01 -7.75667012e-01
1.10722351e+00 -2.98494268e-02 -2.33160153e-01 7.03135610e-01
1.50302395e-01 -1.47983611e-01 6.89510822e-01 2.59338617e-01
7.68021882e-01 5.18299162e-01 4.21967208e-01 -1.32294029e-01
5.43487072e-01 5.48242748e-01 3.57317328e-01 2.72790372e-01
9.98705626e-02 4.01193351e-01 3.39110464e-01 -2.98377097e-01
-1.12573457e+00 -6.78335547e-01 5.32502457e-02 9.23067689e-01
5.65333851e-02 -9.27566528e-01 -6.76842034e-01 -9.64066505e-01
-1.49177954e-01 1.10519826e+00 -1.25005960e-01 1.16762087e-01
-2.74705112e-01 -4.28740591e-01 -2.69927103e-02 4.32327121e-01
5.66762805e-01 -1.08813345e+00 -2.36895442e-01 6.35052681e-01
4.89256578e-03 -1.09067380e+00 -1.01260543e+00 -1.65481895e-01
-7.50557184e-01 -5.32757580e-01 -2.77565747e-01 -9.42360878e-01
5.37981689e-01 4.06187207e-01 1.33834422e+00 -2.55334586e-01
5.67729659e-02 5.75974941e-01 -8.92798662e-01 -4.29150790e-01
-6.91578448e-01 3.08633059e-01 -3.45821738e-01 1.03747249e-01
8.99945140e-01 -2.19105035e-01 -4.76618409e-01 5.82057297e-01
-1.08788085e+00 1.62776317e-02 7.44338214e-01 3.48761261e-01
7.78318882e-01 4.76918131e-01 6.27819359e-01 -1.47255254e+00
1.02625430e+00 -3.97843063e-01 -7.26893067e-01 2.74272799e-01
-1.22022903e+00 -1.45831436e-01 5.13312638e-01 -2.65610784e-01
-1.33200073e+00 7.78854117e-02 -4.81027961e-01 3.35565656e-01
-1.50979742e-01 9.55087245e-01 -5.21055043e-01 5.99560142e-01
2.93977708e-01 -1.11862095e-02 -2.27878720e-01 -6.64496303e-01
6.33164763e-01 1.22470272e+00 1.06980324e-01 -2.74064302e-01
2.93607980e-01 5.04869744e-02 -5.90317547e-01 -5.76493621e-01
-5.45414865e-01 -9.26316082e-01 -5.55843651e-01 -1.48669600e-01
3.96379054e-01 -6.11217022e-01 -6.05889559e-01 3.00498419e-02
-1.11704946e+00 -1.98008530e-02 -5.64230263e-01 2.45870844e-01
-4.82894123e-01 8.73699188e-02 -7.38533616e-01 -6.44116938e-01
-5.30949593e-01 -9.49084699e-01 1.08846676e+00 -1.20513849e-02
-6.35782540e-01 -6.92528725e-01 -5.51680401e-02 6.91150904e-01
5.07512212e-01 -2.29073659e-01 1.09848666e+00 -9.94355679e-01
-5.86022913e-01 -5.15814245e-01 -1.33150727e-01 3.55416358e-01
1.61268830e-01 1.66160583e-01 -5.05853474e-01 4.82142456e-02
-4.11322387e-03 3.68496239e-01 -4.16747630e-02 1.19287863e-01
4.38938528e-01 -6.52791023e-01 -1.75275415e-01 2.77850896e-01
1.52052462e+00 6.36580825e-01 5.44242740e-01 5.87763965e-01
3.27377021e-01 9.48611915e-01 1.18083990e+00 6.71162188e-01
4.28826302e-01 1.06129968e+00 2.98406035e-02 1.28325477e-01
-2.75236040e-01 -5.91014385e-01 4.71914709e-01 1.25107563e+00
2.29847446e-01 -2.86373138e-01 -3.98083925e-01 4.44687903e-01
-1.82921243e+00 -7.19091475e-01 -2.65541971e-01 1.81819618e+00
4.83878702e-01 3.17510068e-01 2.42142990e-01 -2.02703744e-01
2.98121423e-01 -2.59673864e-01 -3.53813171e-01 -7.76279032e-01
1.18577778e-01 -3.46421450e-02 3.22575659e-01 2.84623832e-01
-8.42038929e-01 8.45402539e-01 5.67072010e+00 6.14473760e-01
-8.13536763e-01 2.70196944e-01 3.30539733e-01 9.95836630e-02
-7.83565104e-01 -1.98476717e-01 -1.19717276e+00 2.11425394e-01
8.60630751e-01 -3.86495292e-01 2.55546957e-01 1.35453093e+00
3.99436057e-01 3.76521647e-01 -9.18933153e-01 7.51327813e-01
3.68634820e-01 -1.45998120e+00 4.37011957e-01 9.17194337e-02
6.75543487e-01 -3.62885177e-01 -8.13986082e-03 3.85683328e-01
3.16200405e-01 -3.03871810e-01 1.06155109e+00 -8.20867792e-02
6.27911747e-01 -9.17537928e-01 1.00753796e+00 1.05921648e-01
-1.19702303e+00 6.24715984e-02 -2.11793080e-01 2.93011516e-01
4.38360751e-01 5.50339401e-01 -1.10693038e+00 4.80032325e-01
3.59928846e-01 6.08585358e-01 -2.04696506e-01 6.60897732e-01
-2.05122456e-02 3.69192094e-01 1.81989312e-01 -3.51186723e-01
1.47424936e-01 -6.27729714e-01 4.54927415e-01 1.41301775e+00
4.80538011e-01 8.39798525e-02 5.40410839e-02 1.05968392e+00
-9.73938704e-02 7.26871967e-01 -3.06590378e-01 -5.44951677e-01
1.76391721e-01 1.51908350e+00 -1.05780780e+00 -2.98322707e-01
-7.23773718e-01 9.38134789e-01 -4.23422515e-01 1.37927964e-01
-5.66908836e-01 -4.95753944e-01 5.30562580e-01 5.83694220e-01
6.29055440e-01 -1.51022330e-01 -3.67818117e-01 -9.57600951e-01
1.46413147e-01 -1.15161133e+00 2.46277630e-01 -8.04471314e-01
-1.35470140e+00 1.08550060e+00 -1.65555388e-01 -1.43184662e+00
-3.62955213e-01 -3.61879826e-01 -5.17928563e-02 5.86183071e-01
-1.08453262e+00 -1.33036494e+00 -6.71913698e-02 8.28554779e-02
1.12289834e+00 -3.73663604e-01 9.18473303e-01 5.61480284e-01
-2.12679967e-01 4.62583572e-01 -1.71313927e-01 -1.99813530e-01
3.11093897e-01 -1.04180372e+00 7.05825806e-01 7.15006709e-01
2.28542253e-01 9.83025849e-01 7.43205249e-01 -9.06865656e-01
-1.63262331e+00 -1.18529809e+00 1.39398718e+00 -2.23500386e-01
6.94060326e-01 -5.40905595e-01 -2.44433224e-01 7.61936128e-01
4.51948285e-01 -6.16372347e-01 1.01801872e+00 2.52151072e-01
-3.91718805e-01 -3.45923841e-01 -1.41233301e+00 7.14078426e-01
9.12587523e-01 -1.51573539e-01 -3.90839010e-01 4.25700724e-01
8.05767477e-01 2.15897020e-02 -9.87167418e-01 -6.70600543e-03
7.38169789e-01 -6.46673799e-01 5.20296276e-01 -5.00814378e-01
3.76915395e-01 -4.95778769e-01 -2.94540048e-01 -1.09710336e+00
-4.67754662e-01 -7.64339507e-01 3.03017557e-01 1.79681432e+00
9.72817242e-01 -3.85682762e-01 8.41912091e-01 8.80927563e-01
-2.46601567e-01 -5.38114727e-01 3.58749144e-02 -6.50753498e-01
-5.88094890e-01 -6.13900423e-01 1.04076016e+00 6.18825316e-01
5.39723217e-01 6.83743477e-01 9.86832976e-02 9.57736373e-02
3.12985688e-01 5.04136443e-01 7.32148647e-01 -7.02060521e-01
-8.92244399e-01 -4.14797038e-01 -1.83576658e-01 -1.29333854e+00
-5.15041530e-01 -1.07398856e+00 -2.42322460e-02 -1.51560724e+00
3.48217875e-01 -5.51277578e-01 4.56063114e-02 1.38845041e-01
4.78541791e-01 9.09057185e-02 1.35964766e-01 2.17266291e-01
-7.33318925e-01 1.28265098e-01 1.06928861e+00 -2.34547794e-01
-5.47821820e-01 4.74054843e-01 -1.07962561e+00 4.24405843e-01
8.44420910e-01 -4.95487630e-01 -7.89746165e-01 -1.00947417e-01
6.97015405e-01 -1.19495369e-01 -4.71254319e-01 -2.88940668e-01
3.73571366e-02 1.18475974e-01 -1.91293672e-01 -6.16819501e-01
-1.75228968e-01 -1.16126561e+00 4.56898659e-01 2.11812198e-01
-5.65546751e-01 4.45383668e-01 2.45924164e-02 1.07278846e-01
-3.04713100e-01 -5.34108043e-01 3.35009366e-01 -2.88454533e-01
-6.21481836e-01 9.06575322e-02 -7.58238316e-01 -4.99730855e-01
9.19327557e-01 -1.35385484e-01 -2.24463910e-01 -3.05330008e-01
-6.59591496e-01 -1.86113939e-01 3.90936255e-01 8.02785814e-01
4.66094077e-01 -1.23107672e+00 -4.21901226e-01 4.46240753e-01
6.28957450e-01 -6.69767797e-01 -1.67181566e-01 3.35526168e-01
-6.08856142e-01 6.65802300e-01 -2.72879358e-02 -1.99887738e-01
-1.33255887e+00 8.60538006e-01 -5.56929447e-02 -4.83544320e-01
-4.17423964e-01 4.18428540e-01 -2.78132588e-01 -4.17047918e-01
7.85387084e-02 -3.55601549e-01 -4.52587366e-01 -1.40596375e-01
6.92931831e-01 1.85440168e-01 4.60189462e-01 -4.30688471e-01
3.96520346e-02 2.01417267e-01 -4.90773737e-01 -5.12638748e-01
1.50059450e+00 -5.12556016e-01 -3.22459117e-02 2.06214398e-01
1.00200748e+00 2.10807607e-01 -8.23225439e-01 -1.00076616e-01
4.50235009e-01 -4.01850104e-01 1.26086310e-01 -1.18413579e+00
-1.28264201e+00 7.61839822e-02 3.18251163e-01 5.68584144e-01
1.05352890e+00 2.01835886e-01 1.02079296e+00 1.16571613e-01
7.96271086e-01 -1.22108209e+00 -8.47780108e-02 1.65908068e-01
9.48434472e-01 -8.46082270e-01 -2.45537743e-01 -9.64425206e-01
-1.02642405e+00 1.05486465e+00 1.97291821e-01 1.61820114e-01
9.59424019e-01 6.52053118e-01 3.73991281e-01 -6.26018047e-01
-1.01174223e+00 -1.88282058e-01 2.30402172e-01 5.80653071e-01
8.29360127e-01 6.19068630e-02 -7.64854133e-01 1.21014857e+00
-1.61271751e-01 3.01654458e-01 5.88440776e-01 9.08558130e-01
-3.29867452e-01 -1.67749333e+00 1.27184466e-01 3.87061149e-01
-5.04361928e-01 -2.21112549e-01 -5.34954488e-01 6.40295088e-01
1.57263950e-01 1.33247793e+00 -2.20459104e-01 -3.72696787e-01
8.96729827e-01 -1.10917896e-01 3.20442110e-01 -9.96700883e-01
-1.00575757e+00 6.50701940e-01 7.14079976e-01 -4.41328853e-01
-3.99868399e-01 -8.65667045e-01 -1.18943632e+00 -1.70217395e-01
-3.39907289e-01 5.24654090e-01 9.85093772e-01 5.01749396e-01
6.78396106e-01 3.87324691e-01 6.58520937e-01 -1.64832771e-01
-4.31827068e-01 -9.45624530e-01 -1.05245554e+00 3.94192547e-01
-5.24338245e-01 -5.22018932e-02 3.97970714e-02 6.84572935e-01]
|
[10.045573234558105, 6.079237461090088]
|
bc267fc3-0083-4f2e-9e18-c9391a63a7ac
|
semi-detr-semi-supervised-object-detection
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_Semi-DETR_Semi-Supervised_Object_Detection_With_Detection_Transformers_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_Semi-DETR_Semi-Supervised_Object_Detection_With_Detection_Transformers_CVPR_2023_paper.pdf
|
Semi-DETR: Semi-Supervised Object Detection With Detection Transformers
|
We analyze the DETR-based framework on semi-supervised object detection (SSOD) and observe that (1) the one-to-one assignment strategy generates incorrect matching when the pseudo ground-truth bounding box is inaccurate, leading to training inefficiency; (2) DETR-based detectors lack deterministic correspondence between the input query and its prediction output, which hinders the applicability of the consistency-based regularization widely used in current SSOD methods. We present Semi-DETR, the first transformer-based end-to-end semi-supervised object detector, to tackle these problems. Specifically, we propose a Stage-wise Hybrid Matching strategy that com- bines the one-to-many assignment and one-to-one assignment strategies to improve the training efficiency of the first stage and thus provide high-quality pseudo labels for the training of the second stage. Besides, we introduce a Cross-view Query Consistency method to learn the semantic feature invariance of object queries from different views while avoiding the need to find deterministic query correspondence. Furthermore, we propose a Cost-based Pseudo Label Mining module to dynamically mine more pseudo boxes based on the matching cost of pseudo ground truth bounding boxes for consistency training. Extensive experiments on all SSOD settings of both COCO and Pascal VOC benchmark datasets show that our Semi-DETR method outperforms all state-of-the-art methods by clear margins.
|
['Guanbin Li', 'Jingdong Wang', 'Errui Ding', 'Junyu Han', 'Xiao Tan', 'Kuo Wang', 'Wei zhang', 'Xiangru Lin', 'Jiacheng Zhang']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['semi-supervised-object-detection', 'pseudo-label']
|
['computer-vision', 'miscellaneous']
|
[ 9.28946808e-02 -1.51094526e-01 -3.64215046e-01 -6.18935466e-01
-1.00680268e+00 -4.60903913e-01 3.35638255e-01 -5.81689663e-02
-2.60603219e-01 1.89588457e-01 -3.78044546e-01 -9.05607454e-03
4.22619581e-02 -5.75097322e-01 -9.69741702e-01 -4.79631215e-01
5.32667220e-01 7.37280071e-01 9.15194750e-01 4.09307852e-02
-8.81191641e-02 2.91638345e-01 -1.83676302e+00 4.83373016e-01
7.93559313e-01 1.36854005e+00 2.62989968e-01 1.02173232e-01
-1.12003855e-01 6.23529136e-01 -2.28829205e-01 -4.93771881e-01
5.37886441e-01 -4.13258195e-01 -6.14196360e-01 3.67619604e-01
1.02509582e+00 -2.57869631e-01 -1.88198030e-01 1.25888228e+00
2.67098695e-01 -8.58816504e-02 6.06831312e-01 -1.37517083e+00
-3.66022825e-01 1.68568566e-01 -7.30175078e-01 3.18860337e-02
7.60066882e-02 -4.14268253e-03 1.17730474e+00 -1.42345238e+00
7.25390792e-01 1.17750633e+00 6.48273587e-01 3.42738599e-01
-1.35053980e+00 -6.64265275e-01 2.02922598e-01 1.15880869e-01
-1.58209431e+00 -2.89885640e-01 7.19242513e-01 -4.59566593e-01
7.43283391e-01 2.16227025e-01 4.02235508e-01 7.80933321e-01
-9.52926874e-02 9.42608953e-01 1.05405891e+00 -3.09834003e-01
1.43068746e-01 4.21755731e-01 1.56253710e-01 9.83822048e-01
3.71697992e-01 1.50519833e-01 -5.67038238e-01 -2.78853863e-01
4.98566926e-01 1.44810513e-01 -3.03575490e-02 -1.06697154e+00
-1.05405474e+00 5.79029024e-01 5.58085382e-01 6.51875734e-02
7.21742806e-04 -4.34394106e-02 3.99044454e-01 2.94557631e-01
2.78399855e-01 1.23122893e-01 -3.98649096e-01 4.85904396e-01
-1.08032084e+00 3.04890931e-01 4.85340923e-01 1.33862162e+00
9.44174707e-01 -3.10786545e-01 -3.18432271e-01 8.46002519e-01
3.92448813e-01 5.29724300e-01 2.73725539e-01 -7.10238993e-01
6.57173514e-01 1.03044748e+00 7.10582510e-02 -9.51463223e-01
-2.47433886e-01 -5.52749991e-01 -6.38876259e-01 8.97200685e-03
4.21156615e-01 5.93432844e-01 -8.85304749e-01 1.63211846e+00
6.34928167e-01 1.84361674e-02 -1.88359901e-01 1.02253449e+00
7.21643090e-01 1.43734634e-01 -5.11047617e-02 -1.72575831e-01
1.48519099e+00 -1.26553118e+00 -5.49162626e-01 -4.09943461e-01
9.61705625e-01 -7.51271546e-01 1.19208109e+00 1.09414734e-01
-7.39273310e-01 -8.74797344e-01 -1.17770302e+00 -1.55325523e-02
-2.17356160e-01 3.70039672e-01 3.00591171e-01 4.14638966e-01
-4.54779327e-01 2.99912244e-01 -6.95159614e-01 -1.66873306e-01
5.48319101e-01 2.34253511e-01 -4.35874730e-01 -2.76258290e-01
-7.59994090e-01 7.31012106e-01 5.54019213e-01 1.99674461e-02
-7.02610493e-01 -5.00571012e-01 -7.42027700e-01 -1.49156496e-01
8.71076286e-01 -5.49339890e-01 1.12306416e+00 -8.82451892e-01
-8.10761034e-01 1.18391347e+00 -3.50963414e-01 -2.60690004e-01
7.34717071e-01 -9.19039994e-02 -2.91827798e-01 1.06332749e-02
4.87512231e-01 7.43837416e-01 9.69781697e-01 -1.30562365e+00
-8.69526625e-01 -6.52462780e-01 -2.72005588e-01 2.34380513e-01
-1.17012493e-01 -1.89833686e-01 -8.34141374e-01 -5.18804193e-01
8.17345142e-01 -1.05234945e+00 8.83452594e-02 4.33966845e-01
-4.98618573e-01 -4.55999285e-01 1.00292051e+00 -5.75156286e-02
1.11485875e+00 -2.35401058e+00 -5.67957498e-02 1.26456574e-01
1.40292853e-01 1.69043168e-01 -4.45365980e-02 3.81266512e-02
2.38132086e-02 -3.66643131e-01 -1.82813033e-01 -7.35120177e-01
-4.87134457e-02 1.97231784e-01 -3.56448531e-01 5.62061310e-01
2.35495672e-01 7.50689805e-01 -8.80240560e-01 -7.88540721e-01
1.44480512e-01 -5.12229651e-02 -4.44572955e-01 4.24883008e-01
-3.60622644e-01 7.33153373e-02 -2.47051045e-01 8.07256579e-01
9.31654334e-01 -4.95361656e-01 4.02801484e-02 -5.93723714e-01
9.09175426e-02 2.09324926e-01 -1.45358539e+00 1.55385149e+00
-2.03394204e-01 1.22640021e-01 -2.22011894e-01 -9.07214344e-01
9.27277923e-01 -3.41230929e-02 3.88690114e-01 -7.24950790e-01
-6.92981035e-02 4.95320231e-01 -3.71556789e-01 -2.07222238e-01
3.41736555e-01 8.94554034e-02 -3.14614810e-02 3.17070514e-01
9.64353308e-02 -4.64702351e-03 1.85859486e-01 2.25403845e-01
7.26447463e-01 3.92915159e-01 2.91749299e-01 -1.86592594e-01
6.36604249e-01 2.10274741e-01 8.03839684e-01 7.27552891e-01
-2.84619659e-01 8.70299220e-01 2.85036862e-01 -4.23324794e-01
-9.04813588e-01 -9.66859102e-01 -4.09834832e-01 1.13614798e+00
6.49428904e-01 -3.39858294e-01 -6.56416297e-01 -1.15312684e+00
2.31975645e-01 4.87872720e-01 -5.53631783e-01 -1.37147889e-01
-3.37590486e-01 -4.71691608e-01 3.64498228e-01 6.60040796e-01
6.31008625e-01 -7.12000370e-01 -6.11455023e-01 1.57124355e-01
-3.22077334e-01 -1.43365216e+00 -8.24452698e-01 4.84933764e-01
-9.12122905e-01 -1.29438436e+00 -2.79336572e-01 -8.43306005e-01
9.73913372e-01 6.65759504e-01 1.03434384e+00 3.17585692e-02
-1.93465069e-01 1.26070559e-01 -1.93575978e-01 -3.02391082e-01
-3.57881933e-01 -1.83738321e-01 5.40215373e-02 1.51567042e-01
6.86347306e-01 -4.94796708e-02 -4.89764929e-01 8.90069187e-01
-8.67273152e-01 7.24203587e-02 4.62301075e-01 1.00821185e+00
1.25930500e+00 -2.68653601e-01 2.65511751e-01 -1.04251695e+00
-1.77288234e-01 -1.72266185e-01 -1.01905215e+00 4.88131016e-01
-1.11626494e+00 2.31954128e-01 4.30782318e-01 -5.44619441e-01
-7.05646217e-01 5.38242042e-01 2.14914888e-01 -8.59559119e-01
1.01560928e-01 -7.68027678e-02 -2.61745483e-01 -3.39275114e-02
7.72316456e-01 3.75808179e-01 -1.15838028e-01 -4.56219584e-01
2.60994196e-01 5.62320530e-01 7.28462696e-01 -2.99081117e-01
9.01696444e-01 8.01622212e-01 -5.97642772e-02 -2.80770332e-01
-1.34967446e+00 -9.78063583e-01 -7.51490772e-01 3.59447580e-03
7.10159183e-01 -1.15799987e+00 -2.70799100e-01 3.90984952e-01
-9.95547175e-01 1.11302391e-01 -3.35830808e-01 2.85748094e-01
-5.01616418e-01 3.94678414e-01 -3.25124621e-01 -5.70323467e-01
-3.28306258e-01 -1.22636187e+00 1.44971073e+00 5.03716506e-02
1.03052959e-01 -3.84553552e-01 -5.66651374e-02 5.77961206e-01
6.17443658e-02 -9.71250162e-02 6.65642798e-01 -8.31802070e-01
-8.47066641e-01 -1.86243460e-01 -6.10367358e-01 3.96962821e-01
-7.75759853e-03 -2.77423322e-01 -1.04829681e+00 -4.44707692e-01
2.93210596e-02 -6.36128187e-01 8.50444078e-01 -7.37975612e-02
1.01152730e+00 -1.28258407e-01 -3.71500582e-01 6.11100316e-01
1.52778709e+00 -2.98868477e-01 3.41283947e-01 2.89188355e-01
8.05193841e-01 5.92938602e-01 1.08789432e+00 3.33172441e-01
5.28798342e-01 1.07423162e+00 5.78620017e-01 -8.16870332e-02
-2.65490204e-01 -6.13473892e-01 2.53274232e-01 4.37264413e-01
6.61777973e-01 2.75369752e-02 -5.98399341e-01 5.74428558e-01
-1.98724782e+00 -7.04742432e-01 -2.85848975e-01 2.42690969e+00
7.21474409e-01 4.48266327e-01 1.87689081e-01 2.85431612e-02
7.12624848e-01 5.72883375e-02 -6.22070670e-01 1.79182485e-01
-1.27730370e-01 -2.49652445e-01 3.83273393e-01 1.94133464e-02
-1.20274758e+00 1.00762892e+00 4.90331984e+00 9.89363849e-01
-9.61392760e-01 3.19213092e-01 3.87983859e-01 1.56244516e-01
-7.29973242e-02 2.20669433e-01 -1.22372091e+00 3.38905483e-01
2.80044377e-01 3.39620382e-01 9.88071188e-02 1.35915709e+00
-1.98128775e-01 -3.27377677e-01 -1.52059984e+00 1.12516963e+00
1.55247912e-01 -1.08014631e+00 1.28181830e-01 -1.91128314e-01
8.44685793e-01 3.79572883e-02 -9.26819965e-02 3.17778230e-01
-2.48929799e-01 -3.97270530e-01 1.04039192e+00 1.64429367e-01
8.56572568e-01 -4.26464528e-01 7.34677076e-01 6.18263423e-01
-1.45091283e+00 6.23798184e-03 -5.59342325e-01 5.81875801e-01
-8.47698897e-02 6.35396421e-01 -7.84785032e-01 5.68399012e-01
8.52618873e-01 4.24819410e-01 -8.97857308e-01 8.65566313e-01
-1.02746114e-02 3.44571590e-01 -4.71468985e-01 2.40502015e-01
1.04807757e-01 -3.43733653e-02 5.21918178e-01 1.12943578e+00
-4.96268682e-02 -2.22758755e-01 3.38175714e-01 1.07656169e+00
-9.17229280e-02 3.91255170e-02 -4.81402338e-01 3.79027247e-01
6.25880122e-01 1.06507659e+00 -7.34879434e-01 -2.90028036e-01
-5.70377767e-01 9.70924973e-01 5.20977557e-01 2.39105113e-02
-8.35501492e-01 -1.62072688e-01 9.42974091e-02 2.88713425e-01
5.35843730e-01 2.03806162e-02 -3.86944532e-01 -1.15900421e+00
5.31060874e-01 -7.87454903e-01 6.22344971e-01 -4.80387449e-01
-1.35247529e+00 5.50880075e-01 8.33127797e-02 -1.52419281e+00
-3.74338813e-02 -2.91334093e-01 -2.82701492e-01 4.78495687e-01
-1.52794945e+00 -1.38794148e+00 -5.53376496e-01 5.24870515e-01
6.35329247e-01 -5.14567308e-02 5.15171409e-01 3.91373932e-01
-5.65241516e-01 8.95687103e-01 -7.07643479e-02 1.73883781e-01
8.83011878e-01 -1.02436066e+00 2.75528193e-01 8.61144841e-01
4.81881171e-01 3.12071264e-01 3.91450554e-01 -5.83478332e-01
-1.23791265e+00 -1.27889585e+00 9.22652781e-01 -5.80488920e-01
2.43847877e-01 -5.08348048e-01 -1.11021173e+00 4.83663917e-01
-5.29723465e-01 7.21344352e-01 3.21575791e-01 -9.32758525e-02
-7.20509708e-01 -4.83214110e-01 -1.13348806e+00 2.90754735e-01
1.15864289e+00 -6.62882447e-01 -5.67793787e-01 4.87967223e-01
7.02106476e-01 -6.14561796e-01 -3.79808992e-01 8.69821906e-01
5.27081132e-01 -1.25768554e+00 7.91528583e-01 -3.09832633e-01
6.46851659e-02 -7.04034984e-01 -3.75377506e-01 -4.61732656e-01
-2.29500964e-01 -1.25814870e-01 -3.56742978e-01 1.21379364e+00
2.99731404e-01 -3.98060083e-01 9.45486486e-01 3.58387232e-01
-9.37703475e-02 -8.74540985e-01 -1.11241174e+00 -1.14702380e+00
-5.35432220e-01 -4.35492694e-01 3.81744653e-01 7.65289843e-01
-5.74377716e-01 2.62777030e-01 -1.59830064e-01 3.29775274e-01
7.97860146e-01 4.75912094e-01 9.54898953e-01 -9.97784376e-01
-5.10279596e-01 -8.23496506e-02 -4.31073397e-01 -1.45914733e+00
-5.39808981e-02 -8.08192492e-01 2.94305444e-01 -1.02181852e+00
4.97416764e-01 -8.42245996e-01 -2.11583361e-01 4.99144971e-01
-2.52269000e-01 4.53988761e-01 1.24719746e-01 6.83303714e-01
-9.61892426e-01 5.65844297e-01 1.00492871e+00 -5.04540950e-02
-1.90053359e-01 1.81619927e-01 -2.61870682e-01 6.09470963e-01
1.82063892e-01 -1.11592507e+00 -4.62553501e-01 -2.20724940e-01
1.60464764e-01 -1.83791950e-01 5.47336042e-01 -1.04904830e+00
2.96538919e-01 -8.96332413e-02 1.57297447e-01 -1.01238370e+00
1.67902887e-01 -1.15021563e+00 -4.26069908e-02 3.94205093e-01
-2.04368681e-01 -5.96975163e-02 -5.20909801e-02 9.69896376e-01
-3.03903878e-01 -1.94880217e-01 1.09355319e+00 9.67288986e-02
-6.82599247e-01 2.67629176e-01 3.14821035e-01 2.98844069e-01
1.12365258e+00 -3.25614870e-01 -1.24730311e-01 7.02045187e-02
-3.46930623e-01 4.90332067e-01 5.33625782e-01 4.08688277e-01
5.06065249e-01 -1.42323911e+00 -4.48506296e-01 3.78708839e-01
7.52579749e-01 4.85756099e-01 6.83111995e-02 8.91646326e-01
-3.35579425e-01 4.15356427e-01 1.58037007e-01 -1.21456981e+00
-1.29945600e+00 6.79964304e-01 4.48885322e-01 -3.71348262e-01
-4.49704617e-01 9.90751863e-01 3.57235730e-01 -4.47890490e-01
5.30032873e-01 -2.58741409e-01 3.00639629e-01 -1.13220088e-01
3.04681182e-01 2.73658812e-01 3.50415438e-01 -6.32698536e-01
-6.23285890e-01 5.56844711e-01 -3.06102067e-01 2.28567123e-02
8.45328927e-01 -1.36768863e-01 2.62613818e-02 5.54680884e-01
1.25685692e+00 -2.20333531e-01 -1.32269812e+00 -7.16025293e-01
1.97804511e-01 -6.39666140e-01 -8.59731063e-02 -4.97737408e-01
-7.92129278e-01 6.94954574e-01 9.08291161e-01 -2.65366256e-01
9.84277606e-01 2.84142435e-01 7.52072394e-01 5.33957422e-01
5.46282470e-01 -1.30610585e+00 3.30609977e-01 2.41493255e-01
6.91234291e-01 -1.48527515e+00 1.72592267e-01 -7.12477624e-01
-6.89648449e-01 8.95403504e-01 1.14881897e+00 5.96051179e-02
5.13133526e-01 8.85723308e-02 -4.03047688e-02 -2.89017171e-01
-6.88518822e-01 -2.35337481e-01 7.27609694e-01 2.35538676e-01
-5.34003153e-02 -2.24010736e-01 -7.44211450e-02 3.57372016e-01
3.01732302e-01 -1.46225497e-01 -1.03741519e-01 9.96576309e-01
-4.59883749e-01 -1.05088735e+00 -3.33450496e-01 4.48094159e-01
-3.82558331e-02 1.44751564e-01 -3.51347893e-01 7.47230172e-01
3.71879756e-01 6.62687957e-01 -3.91420349e-02 -4.01523620e-01
6.04964018e-01 1.24243140e-01 4.36150700e-01 -7.71221519e-01
-3.76999617e-01 1.22597426e-01 -2.35644355e-01 -7.56619215e-01
-4.81428415e-01 -5.86349845e-01 -1.27865410e+00 2.49059066e-01
-1.04376292e+00 -1.59126699e-01 4.31382537e-01 8.72294664e-01
4.67023015e-01 4.98664379e-02 7.15093493e-01 -4.63375270e-01
-1.11679351e+00 -8.16923201e-01 -3.92217696e-01 7.64936864e-01
2.62524277e-01 -9.12118793e-01 -2.95873135e-01 -9.47024673e-02]
|
[9.21371078491211, 1.2996522188186646]
|
61f67ab0-a1f7-4d5c-91c9-7a5a61ff7a7e
|
end-to-end-resume-parsing-and-finding
|
1910.03089
| null |
https://arxiv.org/abs/1910.03089v2
|
https://arxiv.org/pdf/1910.03089v2.pdf
|
End-to-End Resume Parsing and Finding Candidates for a Job Description using BERT
|
The ever-increasing number of applications to job positions presents a challenge for employers to find suitable candidates manually. We present an end-to-end solution for ranking candidates based on their suitability to a job description. We accomplish this in two stages. First, we build a resume parser which extracts complete information from candidate resumes. This parser is made available to the public in the form of a web application. Second, we use BERT sentence pair classification to perform ranking based on their suitability to the job description. To approximate the job description, we use the description of past job experiences by a candidate as mentioned in his resume. Our dataset comprises resumes in LinkedIn format and general non-LinkedIn formats. We parse the LinkedIn resumes with 100\% accuracy and establish a strong baseline of 73\% accuracy for candidate suitability.
|
['Ajit Kumar', 'Vedant Bhatia', 'Rajiv Ratn Shah', 'Prateek Rawat']
|
2019-09-30
| null | null | null | null |
['sentence-pair-classification']
|
['natural-language-processing']
|
[ 2.33107582e-01 2.61076599e-01 -7.03730106e-01 -8.67006183e-01
-1.51590943e+00 -9.04353261e-01 6.06091261e-01 6.59278989e-01
-4.69032317e-01 9.23132300e-01 4.42838490e-01 -4.75235343e-01
-5.67759633e-01 -7.32916415e-01 -6.03218019e-01 3.92656744e-01
5.52620471e-01 9.02662158e-01 1.24426112e-01 -4.32440609e-01
2.96666831e-01 3.45252901e-01 -1.47900999e+00 5.50589800e-01
8.25059116e-01 7.07755923e-01 3.12439054e-01 7.51992762e-01
-1.22545786e-01 5.53668916e-01 -8.40033650e-01 -1.22089219e+00
1.83904227e-02 -4.66654092e-01 -1.29609764e+00 -4.92032915e-02
5.72007358e-01 2.15814799e-01 -2.95250297e-01 5.49559176e-01
4.73993212e-01 3.27241540e-01 2.79536366e-01 -8.61501932e-01
-2.35812560e-01 7.25422263e-01 2.53907740e-01 2.78656721e-01
1.04698193e+00 -3.57360959e-01 1.10428870e+00 -8.65722895e-01
1.12756968e+00 8.54543865e-01 6.58396304e-01 2.38231078e-01
-1.10923612e+00 -3.58824760e-01 -3.07467610e-01 -1.45227313e-01
-1.11338365e+00 -8.30280125e-01 6.07909381e-01 -4.29738730e-01
1.07662368e+00 6.22302294e-01 3.40943903e-01 8.06984246e-01
3.56835514e-01 2.83332080e-01 1.09487534e+00 -6.29599094e-01
-1.91444397e-01 1.61016077e-01 4.18647110e-01 6.37996495e-01
1.73397083e-02 -1.59096256e-01 -8.42189968e-01 -4.67455268e-01
1.44641742e-01 -2.93934435e-01 3.80995005e-01 2.29063511e-01
-1.19076574e+00 2.99914211e-01 1.44316673e-01 5.09101003e-02
-2.74900883e-01 -1.26995668e-01 3.56065840e-01 4.46524322e-01
4.03745145e-01 8.83006692e-01 -2.85132527e-01 -6.57970190e-01
-1.05901980e+00 6.97333574e-01 9.19349909e-01 1.23985493e+00
7.32393563e-01 -6.84291184e-01 -4.37346518e-01 9.60558057e-01
-1.27994880e-01 2.02017397e-01 3.10262859e-01 -1.17434609e+00
7.86186278e-01 5.76421976e-01 3.69250953e-01 -8.05510223e-01
-3.87892038e-01 -8.62762809e-01 -2.62086779e-01 -2.84211844e-01
3.64919126e-01 -5.11157513e-02 -4.79897022e-01 1.33234334e+00
1.74818076e-02 -3.77730519e-01 1.15136787e-01 5.65286934e-01
9.33761954e-01 3.62378925e-01 8.17584470e-02 -3.75368893e-01
1.42329597e+00 -7.76244044e-01 -7.34751225e-01 -5.02211571e-01
5.61118245e-01 -1.09755325e+00 9.31369781e-01 8.52894932e-02
-1.47032058e+00 -1.09942114e+00 -7.09036410e-01 -3.02160561e-01
3.21914218e-02 4.59221512e-01 5.54686368e-01 3.94751042e-01
-9.23584223e-01 9.91562307e-01 -6.30513489e-01 -4.44734275e-01
-2.39902169e-01 6.00963414e-01 -4.83925343e-01 -2.49971151e-02
-1.27018690e+00 8.63689482e-01 5.20725012e-01 -1.19741075e-01
1.93437561e-02 -1.00143336e-01 -1.00178516e+00 6.06500618e-02
2.53065109e-01 -7.82121062e-01 1.56561244e+00 -5.50981998e-01
-9.61956441e-01 1.12873709e+00 -4.45135921e-01 -2.15922669e-01
2.60811239e-01 1.07367545e-01 -6.70935273e-01 -1.26497298e-01
6.93029583e-01 3.22445840e-01 2.17842773e-01 -9.00190771e-01
-9.53641295e-01 -2.36507788e-01 1.42645836e-01 4.41386789e-01
-2.36410081e-01 4.31465536e-01 -9.44828928e-01 -3.65740269e-01
3.05643082e-01 -9.83755648e-01 -9.75214764e-02 -1.02528322e+00
-3.92733932e-01 -6.59189224e-01 -1.02882594e-01 -9.42307651e-01
1.65650594e+00 -1.88724625e+00 -2.64987528e-01 1.87792659e-01
-8.70966762e-02 -4.50892150e-02 4.55989912e-02 8.73130381e-01
1.98236406e-01 2.17462599e-01 1.31285220e-01 -4.24531817e-01
3.02321732e-01 -3.04656494e-02 -2.59493113e-01 2.92959679e-02
4.50148992e-02 1.03676808e+00 -1.15013599e+00 -8.82430136e-01
-3.62265915e-01 -2.76662767e-01 -2.01910257e-01 2.44845435e-01
2.19635814e-01 9.07992572e-02 -5.22230387e-01 7.59884298e-01
1.22951083e-01 1.39403835e-01 6.29087567e-01 2.46947885e-01
-1.91053063e-01 1.04003274e+00 -6.91649318e-01 1.78918827e+00
-4.44393575e-01 3.51605386e-01 2.17208683e-01 -6.52981877e-01
1.24126220e+00 2.12456912e-01 3.00878972e-01 -4.62514430e-01
-3.87623221e-01 5.75097978e-01 -2.75995612e-01 -4.39243525e-01
1.10133469e+00 -1.15297504e-01 -1.02434659e+00 2.09067583e-01
-7.44211599e-02 -2.75846511e-01 6.97096825e-01 2.17571199e-01
1.41113365e+00 4.68671918e-01 2.42990509e-01 -1.41764656e-01
2.36839965e-01 5.48017919e-01 6.47710741e-01 1.00902188e+00
1.96441710e-02 4.22923565e-01 4.97636527e-01 -4.79359418e-01
-9.25659955e-01 -8.94807398e-01 -1.17026672e-01 1.15605772e+00
-5.09909429e-02 -8.92840266e-01 -4.18319583e-01 -8.95490170e-01
-9.82324257e-02 6.41574323e-01 1.25089854e-01 2.62604766e-02
-7.24755645e-01 -1.08529523e-01 4.53946471e-01 4.56403226e-01
2.06366867e-01 -1.10564101e+00 -2.24689469e-01 3.80308062e-01
-6.04135811e-01 -1.02802372e+00 -5.73288441e-01 2.83742547e-01
-5.73573112e-01 -6.09658659e-01 -2.19035417e-01 -1.08465230e+00
5.34266829e-01 -1.21908590e-01 1.62127030e+00 1.42732784e-01
1.65926829e-01 1.36585206e-01 -2.81147659e-01 -1.02092385e-01
-5.82453132e-01 7.70005584e-01 1.64966077e-01 -7.39096880e-01
3.57988894e-01 -2.76173472e-01 -1.15799576e-01 2.46877566e-01
-4.63325113e-01 -1.00489736e-01 6.35012567e-01 6.87079072e-01
7.05953658e-01 -9.10082757e-02 8.21500719e-01 -1.27278757e+00
1.23522699e+00 -4.42464918e-01 -2.40822062e-01 4.22147155e-01
-9.11007881e-01 -5.78381047e-02 5.38213551e-01 1.58860996e-01
-9.48405504e-01 6.18647277e-01 -6.13870561e-01 2.67189682e-01
-2.89251953e-01 1.03868723e+00 -3.67080905e-02 4.78556037e-01
7.07023382e-01 -1.35230020e-01 -3.62785906e-01 -5.90963364e-01
1.25284791e-01 1.15572250e+00 1.22217989e+00 -7.38461733e-01
7.78478503e-01 -5.67178488e-01 3.24354991e-02 -1.33554667e-01
-1.05108345e+00 -7.49153197e-01 -6.94559634e-01 -1.10001221e-01
5.24139762e-01 -6.87461913e-01 -7.63876200e-01 -3.12130213e-01
-9.88192916e-01 -2.61958353e-02 -2.69761413e-01 5.21077812e-01
-5.06737232e-01 3.24591100e-01 -4.91611958e-01 -6.31762505e-01
-1.18990548e-01 -7.39645422e-01 1.00005710e+00 2.44578630e-01
-7.20157564e-01 -5.95125973e-01 -8.73314142e-02 1.04832625e+00
-5.06507978e-02 8.94086994e-03 8.14202726e-01 -1.00612223e+00
-1.16948761e-01 -7.87865639e-01 1.82523981e-01 -1.90244645e-01
-9.45322141e-02 -3.42966706e-01 -4.45071757e-01 -1.83110461e-01
-3.06015879e-01 -4.29510951e-01 6.80530250e-01 -4.32424210e-02
8.66987467e-01 -3.43792677e-01 -5.62994838e-01 3.72499853e-01
9.09219503e-01 -3.60622740e-04 3.14259231e-01 4.59386498e-01
2.18332708e-01 9.17903781e-01 1.20504320e+00 2.46577531e-01
5.39702475e-01 1.02684033e+00 -1.14135884e-01 1.78599775e-01
-4.69702445e-02 -5.90584397e-01 3.79036784e-01 8.45631897e-01
3.91199589e-02 -5.07278383e-01 -1.14125896e+00 5.22846878e-01
-1.80480170e+00 -1.15682113e+00 -3.06730747e-01 2.26662183e+00
9.70040500e-01 6.97246075e-01 3.06951016e-01 -5.12272194e-02
7.56096900e-01 -1.93246379e-01 -6.15495369e-02 -7.39158988e-01
1.67458743e-01 2.14571819e-01 4.27380294e-01 6.60304606e-01
-9.83754277e-01 8.86081636e-01 6.98299599e+00 7.54398882e-01
-5.03443301e-01 1.33529648e-01 8.15456271e-01 -1.57668255e-02
-3.79845589e-01 3.27807099e-01 -1.25535059e+00 4.59662914e-01
1.45443761e+00 -4.70671356e-01 3.08347553e-01 7.63484478e-01
-7.02664480e-02 3.66376899e-02 -1.02039087e+00 5.77606976e-01
-5.71134910e-02 -1.46754587e+00 -5.00763178e-01 4.90126237e-02
5.01485348e-01 -2.64724106e-01 -2.19987303e-01 8.10003698e-01
6.34984598e-02 -8.91055465e-01 7.14371443e-01 8.26617837e-01
1.07274401e+00 -9.01344001e-01 9.41092134e-01 7.27087557e-01
-1.05280650e+00 -1.23783849e-01 -3.39635044e-01 -4.90983307e-01
2.49893814e-01 5.37593961e-01 -1.31866539e+00 8.53465021e-01
2.63755500e-01 2.85709679e-01 -7.00383186e-01 7.58970380e-01
-5.46110943e-02 6.83980048e-01 1.24831880e-02 4.24412191e-02
-2.64541835e-01 -1.00957647e-01 3.01427424e-01 1.31529403e+00
4.88743931e-01 -8.19347426e-02 7.03572631e-01 4.90933239e-01
-4.95849043e-01 1.88786000e-01 -5.69927096e-01 -2.49765992e-01
8.11895549e-01 1.29412317e+00 -5.65153241e-01 -3.05720896e-01
-2.18823284e-01 1.10419405e+00 4.01211828e-01 1.41067116e-03
-3.42360884e-01 -1.15113854e+00 1.66749984e-01 5.33825696e-01
-3.45955610e-01 -3.32297951e-01 -3.13628465e-02 -5.68260372e-01
4.30869222e-01 -7.52160668e-01 4.17204201e-01 -6.98217571e-01
-1.11904037e+00 9.67905939e-01 1.83758512e-01 -9.87764299e-01
-9.02594030e-01 -2.80545473e-01 -3.49603087e-01 1.14093375e+00
-1.07299984e+00 -1.10422504e+00 -1.19229950e-01 -1.89385265e-01
5.15944302e-01 -5.33503667e-02 8.73814464e-01 4.26325291e-01
-3.12678695e-01 5.21560490e-01 -1.29628390e-01 1.40657857e-01
1.08117127e+00 -1.34529567e+00 9.03851449e-01 6.42865956e-01
1.69449672e-01 1.01015019e+00 5.84390223e-01 -1.19158924e+00
-1.09562492e+00 -9.41974223e-01 2.38926697e+00 -1.01183808e+00
2.28567451e-01 -3.68126929e-01 -2.97779858e-01 6.29993439e-01
4.24637832e-02 9.18483187e-04 6.18227541e-01 6.00767314e-01
4.23715472e-01 -2.07883596e-01 -7.07237959e-01 9.09867436e-02
1.33200359e+00 -8.04362535e-01 -6.94066823e-01 7.79462516e-01
4.84539062e-01 -8.78744304e-01 -1.22137415e+00 1.88358754e-01
3.84266317e-01 -7.01909006e-01 8.47494185e-01 -6.93260252e-01
4.80308741e-01 3.04874592e-02 1.13489650e-01 -9.78427827e-01
-6.06552899e-01 -1.15009773e+00 2.66854882e-01 1.56246412e+00
7.73546636e-01 -5.36521040e-02 8.13259363e-01 1.09032166e+00
-7.50755370e-01 -8.56283069e-01 -9.23377156e-01 -8.00170720e-01
-3.23183537e-01 -2.90762305e-01 7.49550462e-01 6.82179987e-01
5.44395506e-01 6.70852125e-01 -2.63110131e-01 -3.56073558e-01
3.82968932e-02 6.47402823e-01 4.76525187e-01 -1.56903720e+00
-1.73063666e-01 -1.53405875e-01 -1.92996919e-01 -9.06444252e-01
5.16922235e-01 -1.51032889e+00 9.39995423e-02 -2.05439115e+00
3.35177869e-01 -5.98918915e-01 -1.91106200e-01 5.22308648e-01
-2.57502824e-01 4.09924239e-01 5.28891310e-02 4.87630039e-01
-9.31706488e-01 -9.19097885e-02 7.52505124e-01 1.47121191e-01
-2.72115052e-01 6.57796025e-01 -8.26070964e-01 3.91774952e-01
7.54510343e-01 -7.27192044e-01 -9.55429152e-02 -2.85737246e-01
7.54715383e-01 6.85066104e-01 7.80847203e-03 -1.18918288e+00
2.52911150e-01 -3.30463439e-01 4.12273735e-01 -5.87867677e-01
3.60640287e-01 -2.45421857e-01 2.78151631e-01 1.02847040e-01
-7.98133373e-01 6.52850568e-01 -1.07062481e-01 3.20129693e-01
-3.02664638e-01 -7.07281351e-01 4.06249985e-02 -1.01648100e-01
-3.62730116e-01 -1.36174694e-01 -4.06982541e-01 1.61947589e-02
5.31577826e-01 -2.02392891e-01 -2.22597420e-01 -4.74255830e-01
-8.90825748e-01 8.72911811e-02 7.01129198e-01 3.59841853e-01
3.22240680e-01 -1.36825418e+00 -6.89770520e-01 -3.27969342e-02
2.12640554e-01 -5.34908175e-01 -3.86563748e-01 5.89104712e-01
-3.27393532e-01 6.48999870e-01 -1.68247730e-01 1.30875766e-01
-1.43192327e+00 3.15538824e-01 -1.96485415e-01 -4.13000286e-01
-1.33204028e-01 8.04772079e-01 -7.44212031e-01 -7.64078617e-01
9.70950425e-02 1.15435543e-02 -2.04976127e-01 -1.10257626e-01
3.73497903e-02 3.29249412e-01 4.48686749e-01 -7.68405139e-01
-5.66962421e-01 1.43220976e-01 3.01608682e-01 -3.83968651e-01
1.08462155e+00 -1.15562975e-01 -1.37342393e-01 2.94184417e-01
9.34701860e-01 7.55759060e-01 -6.03547633e-01 -4.45717666e-03
5.43695033e-01 -5.58551431e-01 -2.68909603e-01 -9.24592018e-01
-1.86539158e-01 1.91046551e-01 8.82674679e-02 4.15955514e-01
7.95961857e-01 1.70566961e-01 9.27568853e-01 7.94720054e-01
8.24495926e-02 -1.31771815e+00 -3.51423204e-01 6.65432692e-01
5.71619391e-01 -1.23828363e+00 1.69648692e-01 -4.64854240e-01
-8.44983399e-01 1.10811222e+00 4.62619901e-01 4.80504274e-01
9.97054353e-02 -6.55480549e-02 5.56913838e-02 -2.24945307e-01
-8.23577821e-01 -1.80984125e-01 6.75762653e-01 3.13711643e-01
1.11658406e+00 6.08650558e-02 -9.74291682e-01 1.05626202e+00
-6.47891998e-01 -2.73796152e-02 3.70343298e-01 9.61037993e-01
-4.89507854e-01 -1.92219055e+00 -1.25327542e-01 7.57101834e-01
-8.42153966e-01 -5.44342473e-02 -8.51329446e-01 2.80251980e-01
2.71076739e-01 1.42480290e+00 -1.29847929e-01 -7.29413152e-01
6.16811633e-01 3.61199319e-01 1.69456005e-01 -1.09661961e+00
-1.19816267e+00 -1.53016806e-01 1.04604983e+00 -2.73408353e-01
-1.18783591e-02 -8.97379756e-01 -1.33832228e+00 -1.61463127e-01
-1.20998926e-01 8.01491201e-01 7.60824919e-01 7.97241449e-01
7.47524798e-01 3.50171357e-01 4.82751548e-01 -3.94887596e-01
-7.81913817e-01 -7.92945206e-01 -3.17669660e-01 5.76696277e-01
-1.50135905e-01 -1.70718774e-01 3.55362982e-01 5.84017895e-02]
|
[9.85311222076416, 9.217659950256348]
|
27123412-0c7c-46c7-a4e0-cda3201c2e19
|
a-generalized-framework-for-video-instance
|
2211.08834
| null |
https://arxiv.org/abs/2211.08834v2
|
https://arxiv.org/pdf/2211.08834v2.pdf
|
A Generalized Framework for Video Instance Segmentation
|
The handling of long videos with complex and occluded sequences has recently emerged as a new challenge in the video instance segmentation (VIS) community. However, existing methods have limitations in addressing this challenge. We argue that the biggest bottleneck in current approaches is the discrepancy between training and inference. To effectively bridge this gap, we propose a Generalized framework for VIS, namely GenVIS, that achieves state-of-the-art performance on challenging benchmarks without designing complicated architectures or requiring extra post-processing. The key contribution of GenVIS is the learning strategy, which includes a query-based training pipeline for sequential learning with a novel target label assignment. Additionally, we introduce a memory that effectively acquires information from previous states. Thanks to the new perspective, which focuses on building relationships between separate frames or clips, GenVIS can be flexibly executed in both online and semi-online manner. We evaluate our approach on popular VIS benchmarks, achieving state-of-the-art results on YouTube-VIS 2019/2021/2022 and Occluded VIS (OVIS). Notably, we greatly outperform the state-of-the-art on the long VIS benchmark (OVIS), improving 5.6 AP with ResNet-50 backbone. Code is available at https://github.com/miranheo/GenVIS.
|
['Seon Joo Kim', 'Joon-Young Lee', 'Seoung Wug Oh', 'Hanjung Kim', 'Jeongseok Hyun', 'Sukjun Hwang', 'Miran Heo']
|
2022-11-16
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Heo_A_Generalized_Framework_for_Video_Instance_Segmentation_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Heo_A_Generalized_Framework_for_Video_Instance_Segmentation_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['video-instance-segmentation']
|
['computer-vision']
|
[ 6.87215338e-03 -2.22254664e-01 -5.24217844e-01 -1.85354829e-01
-9.33827162e-01 -5.94974399e-01 3.63877296e-01 -2.54653335e-01
-5.49734533e-01 4.95266855e-01 1.20449271e-02 -2.39199191e-01
1.95722416e-01 -5.47626495e-01 -9.71067548e-01 -3.61395150e-01
-3.28506939e-02 2.69148439e-01 7.65828133e-01 -7.73831606e-02
2.73186415e-02 2.29960576e-01 -1.56461918e+00 4.18992370e-01
6.36868894e-01 1.12532675e+00 1.94405019e-01 7.03403950e-01
-1.10855356e-01 9.23886716e-01 -2.57211983e-01 -5.58472693e-01
4.81221259e-01 -3.30349654e-01 -9.45159137e-01 1.46839291e-01
8.70993435e-01 -5.47517717e-01 -6.59297526e-01 8.75713706e-01
5.70314169e-01 9.95550305e-02 1.60991326e-01 -1.20477569e+00
-1.23172566e-01 6.63439512e-01 -6.37210846e-01 3.90275389e-01
1.44886464e-01 4.34686452e-01 1.04603195e+00 -8.00298035e-01
7.61687219e-01 1.07256520e+00 6.47504210e-01 4.63039279e-01
-1.10138571e+00 -4.67553407e-01 5.66115677e-01 5.65542877e-01
-1.37090838e+00 -6.16110861e-01 4.20690149e-01 -4.98819292e-01
8.33085716e-01 2.63825804e-01 7.53746927e-01 1.19530666e+00
-2.76642531e-01 1.33321977e+00 9.27694798e-01 -2.11607981e-02
1.03575051e-01 -2.99341500e-01 1.65353656e-01 7.69744873e-01
-7.53486007e-02 9.63611007e-02 -6.56204462e-01 2.86555976e-01
6.33626819e-01 2.02087406e-02 -3.57999474e-01 -3.21327388e-01
-1.15368915e+00 5.11052608e-01 2.62035370e-01 2.31201202e-02
-3.28907251e-01 4.16367352e-01 7.83524573e-01 1.77788794e-01
5.41987360e-01 6.65499717e-02 -6.85343206e-01 -5.42312205e-01
-1.36529541e+00 2.09449545e-01 7.71837831e-01 1.00646603e+00
6.39288425e-01 -3.26303914e-02 -5.16298473e-01 6.92109406e-01
1.77465275e-01 4.55471873e-01 1.08093105e-01 -1.25250340e+00
4.49916899e-01 2.06326708e-01 -1.66866809e-01 -5.08424759e-01
-2.30507344e-01 -6.09665215e-01 -4.75214124e-01 3.04681491e-02
5.10248065e-01 -1.31719664e-01 -1.08060372e+00 1.67897081e+00
4.43786830e-01 8.00781667e-01 -1.98193580e-01 9.47002769e-01
9.23638284e-01 6.81125760e-01 1.08828187e-01 -1.84807330e-01
1.26344073e+00 -1.73433733e+00 -5.07771969e-01 -3.08856308e-01
4.67833072e-01 -6.11999154e-01 1.12939668e+00 5.14349461e-01
-1.34886050e+00 -6.13061011e-01 -8.40948939e-01 -1.34981588e-01
-3.06619834e-02 2.77148262e-02 5.76185226e-01 3.49056184e-01
-9.89123821e-01 6.28637791e-01 -1.16760373e+00 -2.95185834e-01
7.23587215e-01 2.30898023e-01 -1.10328786e-01 -1.51030287e-01
-8.12611103e-01 3.77270430e-01 3.02977651e-01 1.67092577e-01
-1.08430505e+00 -8.97265613e-01 -8.42501938e-01 4.72744182e-02
1.10982919e+00 -5.78956187e-01 1.37204456e+00 -1.10818362e+00
-1.65252030e+00 7.46436357e-01 -1.99943751e-01 -5.36084235e-01
9.31495786e-01 -5.82955718e-01 -1.64619058e-01 4.11502153e-01
-8.51345435e-03 7.88578153e-01 7.65728652e-01 -1.13391018e+00
-7.62650907e-01 -8.19846913e-02 3.81627858e-01 5.69285005e-02
-2.35446453e-01 1.19788870e-01 -1.44688976e+00 -6.00693464e-01
-2.61215806e-01 -1.04252374e+00 -9.27956626e-02 1.69312328e-01
-3.44413638e-01 -1.35631442e-01 8.50868046e-01 -6.22377872e-01
1.41472816e+00 -2.24113655e+00 2.38497227e-01 -1.34900600e-01
2.59122938e-01 7.40358829e-01 -3.62988651e-01 2.98907012e-01
2.09872231e-01 -1.50939031e-02 -7.81302303e-02 -7.19588876e-01
-1.44139016e-02 2.32430577e-01 -1.65199518e-01 3.87558609e-01
2.73309052e-02 1.13278818e+00 -8.67263377e-01 -6.19333148e-01
3.00230205e-01 4.38609689e-01 -7.13768661e-01 1.99004054e-01
-4.43239599e-01 4.36632842e-01 -2.16465861e-01 7.40996182e-01
6.47090137e-01 -4.46863800e-01 2.05000117e-01 -3.69253039e-01
-1.96378827e-01 2.30803281e-01 -1.23944640e+00 2.05782533e+00
-2.49336123e-01 7.43072927e-01 1.46009788e-01 -1.02468586e+00
3.40207726e-01 1.90130770e-01 5.77031314e-01 -6.69491529e-01
1.43771306e-01 1.00632668e-01 -2.96325684e-01 -7.46215701e-01
4.64990407e-01 5.35598397e-01 2.80579865e-01 2.31147677e-01
2.02649415e-01 2.97478944e-01 7.75210679e-01 2.54125744e-01
1.03319860e+00 6.88090086e-01 -5.67355752e-02 -9.94999483e-02
5.64776003e-01 -7.51792863e-02 8.66091788e-01 7.71228671e-01
-4.23509389e-01 6.84980929e-01 6.99034631e-01 -3.91336411e-01
-7.91480184e-01 -1.01367140e+00 7.10309148e-02 1.22242224e+00
3.29470247e-01 -7.76134133e-01 -8.91115665e-01 -8.96856427e-01
-1.51115060e-01 3.36024612e-01 -4.66407865e-01 2.22572386e-01
-8.09813499e-01 -3.36970836e-01 4.80480969e-01 6.95889890e-01
5.89190722e-01 -9.47566330e-01 -7.35526502e-01 2.27369800e-01
-3.23435545e-01 -1.57792771e+00 -6.07429504e-01 -2.15030447e-01
-7.81789362e-01 -1.19929790e+00 -6.64008975e-01 -5.70089042e-01
2.92614609e-01 4.61723834e-01 1.32304668e+00 2.41332218e-01
-2.11735636e-01 4.26204175e-01 -5.11393249e-01 1.24019496e-02
-1.03454337e-01 4.70625639e-01 -3.97366643e-01 -5.37422597e-02
4.65484671e-02 -4.02816713e-01 -9.20091867e-01 4.86419231e-01
-9.53011811e-01 4.94720578e-01 3.93128812e-01 5.72179675e-01
7.98126936e-01 -4.36651349e-01 4.81531382e-01 -1.00635946e+00
-2.02462703e-01 -5.11498630e-01 -7.37398386e-01 1.97658837e-01
-3.93421799e-01 -2.00624719e-01 4.71741945e-01 -3.46294284e-01
-9.13874149e-01 5.79343960e-02 -4.46648449e-01 -5.55369914e-01
-9.52999368e-02 5.24940968e-01 7.47657521e-03 4.17841040e-02
2.67301947e-01 8.15828294e-02 -3.26721556e-02 -5.39997101e-01
4.01719689e-01 4.97600496e-01 6.47579312e-01 -6.48284793e-01
5.45088470e-01 6.76059663e-01 -2.13485867e-01 -5.76037586e-01
-1.13964176e+00 -6.08160377e-01 -5.37609339e-01 -4.80065405e-01
8.90995443e-01 -1.17006588e+00 -7.82549441e-01 6.05434895e-01
-9.66454983e-01 -8.50289762e-01 -2.58207440e-01 2.91114450e-01
-5.53946853e-01 4.76058185e-01 -1.02864790e+00 -2.91619062e-01
-3.84455055e-01 -1.46930420e+00 1.02949035e+00 2.15184376e-01
7.25234821e-02 -7.80127466e-01 -1.29684478e-01 6.71857357e-01
4.56006229e-01 1.83349952e-01 3.09926033e-01 -4.72302556e-01
-1.00805485e+00 4.03785221e-02 -4.35756654e-01 5.57527065e-01
-2.22583354e-01 2.06792191e-01 -9.99370039e-01 -5.07453203e-01
-4.11766052e-01 -5.07655621e-01 1.18565392e+00 4.11641449e-01
1.40527451e+00 3.23559456e-02 -2.04606116e-01 1.00995517e+00
1.39538682e+00 -4.50789258e-02 6.88882709e-01 4.08465892e-01
8.80921841e-01 2.77306467e-01 8.47956717e-01 4.04026747e-01
6.66357696e-01 8.46337557e-01 5.37901461e-01 -1.27223969e-01
-4.29531425e-01 -1.66545451e-01 4.48334455e-01 7.95238137e-01
-2.25538462e-01 -3.79553854e-01 -8.02720428e-01 5.52867413e-01
-2.16577053e+00 -9.01059985e-01 -1.88013822e-01 2.02309966e+00
7.04961777e-01 2.33969152e-01 2.58570313e-01 -7.36319423e-02
5.29990017e-01 6.02663159e-01 -6.10439599e-01 8.16680640e-02
-5.24908826e-02 2.31432840e-01 5.19188285e-01 4.50576007e-01
-1.24816334e+00 1.29248536e+00 5.70074892e+00 1.03895032e+00
-1.31344986e+00 3.05660933e-01 7.42795467e-01 -3.70180249e-01
5.56367673e-02 2.24139541e-02 -8.18916857e-01 5.75837731e-01
8.40971112e-01 8.44687447e-02 5.39658964e-01 7.18105078e-01
3.73460382e-01 -1.49895236e-01 -1.14129996e+00 1.08492148e+00
1.37090072e-01 -1.66352129e+00 -1.29883572e-01 -1.10915929e-01
8.14338744e-01 4.99898463e-01 -7.58620203e-02 3.98268372e-01
-1.57147154e-01 -6.75887167e-01 1.02233362e+00 3.45595241e-01
9.42672253e-01 -4.77573156e-01 6.12484336e-01 1.36177763e-01
-1.44836247e+00 5.58266044e-02 -2.40096208e-02 2.57192808e-03
5.06678998e-01 4.14859504e-01 -1.32979199e-01 6.93006277e-01
8.75471294e-01 1.13280606e+00 -5.73365331e-01 1.27439129e+00
-3.26341242e-01 9.19206500e-01 -2.94790953e-01 4.26557124e-01
3.65445316e-01 -1.59271210e-01 4.13411707e-01 1.44384670e+00
1.14940003e-01 -5.87858930e-02 5.69528043e-01 5.52653372e-01
-2.58801341e-01 -7.09996969e-02 -2.01912466e-02 5.05893901e-02
4.02832150e-01 1.30258179e+00 -8.32417905e-01 -6.62198365e-01
-7.01213837e-01 9.59335327e-01 2.61785835e-01 5.17598450e-01
-1.24758482e+00 1.41636565e-01 6.56980157e-01 1.59581885e-01
7.85782099e-01 -3.01216215e-01 3.77618484e-02 -1.42532706e+00
1.27930418e-01 -1.07776761e+00 4.29969221e-01 -5.01478791e-01
-9.27016616e-01 6.08427584e-01 8.41623470e-02 -1.10337496e+00
-9.36801173e-03 -4.44230407e-01 -3.51257324e-01 2.70006090e-01
-1.89960837e+00 -1.13286376e+00 -5.78863621e-01 5.70832551e-01
9.68536019e-01 8.52507427e-02 2.75068671e-01 8.55862319e-01
-8.95359933e-01 5.80291808e-01 -7.03532845e-02 1.77642986e-01
7.96475470e-01 -1.02519453e+00 6.17702901e-01 1.04841280e+00
2.57361174e-01 9.94692836e-03 4.41928357e-01 -4.18055505e-01
-1.50422299e+00 -1.20692742e+00 4.45771843e-01 -2.11304337e-01
6.83652520e-01 -4.26045418e-01 -8.54095042e-01 8.41092646e-01
2.53377467e-01 5.04582465e-01 4.03452307e-01 -1.20357007e-01
-4.14620280e-01 -6.92966804e-02 -5.97462177e-01 5.51114917e-01
1.60235035e+00 -3.94721717e-01 -3.08868848e-02 3.60068947e-01
8.29797566e-01 -8.87336731e-01 -7.53012300e-01 3.75924498e-01
4.58950609e-01 -1.25231767e+00 9.77696240e-01 -4.28420991e-01
5.56583762e-01 -3.49452406e-01 -8.53951201e-02 -8.60037923e-01
-1.10197239e-01 -8.40496719e-01 -4.67898011e-01 1.16206789e+00
1.80471778e-01 -4.29408967e-01 9.37157631e-01 2.76895761e-01
-3.01409721e-01 -1.12962222e+00 -7.27378070e-01 -8.67859364e-01
-2.47202709e-01 -7.93309391e-01 3.18495542e-01 6.11918390e-01
-5.31890929e-01 1.94494560e-01 -5.24168730e-01 2.68463343e-02
5.40756702e-01 1.77299887e-01 9.99998629e-01 -8.62564743e-01
-5.99774718e-01 -3.67789477e-01 -3.16057205e-01 -1.63780260e+00
2.58158356e-01 -6.97937787e-01 -7.95964226e-02 -1.51025259e+00
3.08301061e-01 -3.21676522e-01 -3.06399643e-01 5.95324039e-01
-2.67777175e-01 5.39654613e-01 7.48479605e-01 2.25417271e-01
-1.20543182e+00 5.08641362e-01 1.22413588e+00 2.98244995e-03
-9.89370272e-02 -7.30041564e-02 -4.07781422e-01 8.31450522e-01
6.14708424e-01 -3.62731218e-01 -3.40343982e-01 -7.55492687e-01
1.00566357e-01 4.91571128e-02 4.60294217e-01 -1.04637504e+00
1.98060825e-01 7.32385088e-04 -2.34204710e-01 -6.34942353e-01
3.91012639e-01 -6.33988857e-01 7.36626610e-02 4.42041248e-01
-1.56489104e-01 2.43667560e-03 2.31384218e-01 4.41729635e-01
-3.63398552e-01 -1.35364503e-01 7.07988203e-01 1.14255264e-01
-1.18425322e+00 7.27584481e-01 4.73691104e-03 4.94876862e-01
1.22343802e+00 -9.22858790e-02 -5.40959001e-01 -2.05362871e-01
-5.33563614e-01 5.30842304e-01 5.23448408e-01 4.46950197e-01
3.73176336e-01 -9.87294614e-01 -7.32493460e-01 -4.19688076e-02
-3.62296887e-02 9.13830474e-02 5.14480948e-01 1.18258798e+00
-8.17413211e-01 1.06527232e-01 3.56622078e-02 -8.29742551e-01
-1.38063931e+00 4.67321545e-01 2.80929357e-01 -3.49160522e-01
-8.17452073e-01 8.64147723e-01 1.75758004e-01 -9.28264111e-02
3.97196442e-01 -1.90167457e-01 1.61837172e-02 9.98666883e-02
5.23406863e-01 5.66491067e-01 -1.33592170e-02 -5.43351233e-01
-2.57484794e-01 5.08707285e-01 -1.91938847e-01 3.57380696e-02
1.38538218e+00 -2.41118401e-01 6.42189011e-02 3.20858240e-01
1.15477753e+00 -9.33842659e-02 -1.90212119e+00 -4.91940558e-01
-1.64458767e-01 -7.28951275e-01 8.35464671e-02 -7.21147537e-01
-1.58656669e+00 7.33702540e-01 4.84540999e-01 -3.48954083e-04
1.05455804e+00 9.28509831e-02 1.22573566e+00 1.28178120e-01
2.85645872e-01 -1.09572423e+00 7.81706572e-02 5.29343426e-01
5.35064161e-01 -1.42928004e+00 2.65941638e-02 -6.51558280e-01
-5.86392760e-01 9.71225977e-01 5.59717655e-01 -1.05270967e-01
5.30801892e-01 2.70989418e-01 1.96811184e-01 -5.15985340e-02
-8.27449620e-01 -3.43680710e-01 2.98082620e-01 1.47706389e-01
4.38176394e-01 -2.00938731e-01 -3.83455813e-01 2.97844023e-01
2.43749395e-01 2.68723190e-01 4.03549910e-01 1.01122606e+00
-1.63779035e-01 -1.09323776e+00 4.60698381e-02 2.89187968e-01
-6.75496161e-01 -2.46970564e-01 3.00455779e-01 8.24636877e-01
8.78401101e-02 7.78733611e-01 3.33622918e-02 -1.62183762e-01
3.55411410e-01 -2.63463914e-01 3.90848607e-01 -4.52430487e-01
-5.08012712e-01 3.02734673e-01 1.99911788e-01 -1.27399445e+00
-7.05144107e-01 -7.93722689e-01 -1.16968429e+00 -3.78317744e-01
-1.37533247e-01 -1.07046425e-01 4.82321858e-01 9.49657321e-01
6.38495922e-01 7.71911085e-01 2.71750391e-01 -1.06460881e+00
-4.46737975e-01 -6.05061352e-01 -1.93252876e-01 5.06940722e-01
3.05119514e-01 -5.83026707e-01 -1.12521201e-01 2.78809726e-01]
|
[9.18776798248291, 0.010488499887287617]
|
4f05afb1-9883-4ba6-b7e7-91528d6dec21
|
katakomba-tools-and-benchmarks-for-data
|
2306.08772
| null |
https://arxiv.org/abs/2306.08772v1
|
https://arxiv.org/pdf/2306.08772v1.pdf
|
Katakomba: Tools and Benchmarks for Data-Driven NetHack
|
NetHack is known as the frontier of reinforcement learning research where learning-based methods still need to catch up to rule-based solutions. One of the promising directions for a breakthrough is using pre-collected datasets similar to recent developments in robotics, recommender systems, and more under the umbrella of offline reinforcement learning (ORL). Recently, a large-scale NetHack dataset was released; while it was a necessary step forward, it has yet to gain wide adoption in the ORL community. In this work, we argue that there are three major obstacles for adoption: tool-wise, implementation-wise, and benchmark-wise. To address them, we develop an open-source library that provides workflow fundamentals familiar to the ORL community: pre-defined D4RL-style tasks, uncluttered baseline implementations, and reliable evaluation tools with accompanying configs and logs synced to the cloud.
|
['Sergey Kolesnikov', 'Denis Tarasov', 'Alexander Nikulin', 'Vladislav Kurenkov']
|
2023-06-14
| null | null | null | null |
['nethack', 'd4rl']
|
['playing-games', 'robots']
|
[-1.52293622e-01 -9.34841931e-02 -2.97076672e-01 -4.36844677e-01
-5.27756631e-01 -7.51098156e-01 6.93064988e-01 2.40596429e-01
-6.09058797e-01 6.93599463e-01 2.12570474e-01 -3.28543603e-01
-4.95974123e-01 -5.83743930e-01 -6.13996983e-01 -3.37756723e-01
-2.04478994e-01 7.33114660e-01 3.46271247e-01 -5.98646522e-01
4.95533824e-01 3.97732615e-01 -1.71656024e+00 5.53936362e-02
5.99400997e-01 8.23469222e-01 1.68165907e-01 5.86193502e-01
-3.66903916e-02 1.03224361e+00 -4.59397346e-01 -2.17900693e-01
4.76831526e-01 -1.91914231e-01 -9.72972095e-01 -3.00488740e-01
5.03168344e-01 -5.11621356e-01 -1.90488786e-01 7.40123749e-01
6.04353607e-01 2.52490103e-01 4.65046018e-02 -1.46045768e+00
-4.97628003e-01 7.74290025e-01 -1.01264767e-01 1.31792054e-02
3.32792073e-01 5.34222126e-01 8.33878934e-01 -4.62707192e-01
6.64196134e-01 1.00221634e+00 6.61079884e-01 6.76623583e-01
-9.81999516e-01 -4.87674773e-01 1.44756332e-01 3.42583567e-01
-7.38907456e-01 -4.79157478e-01 4.48261350e-01 -4.56409872e-01
1.12068367e+00 3.18628997e-02 8.50606024e-01 1.47025478e+00
-2.07309335e-01 1.04601550e+00 1.14643967e+00 -3.93256694e-01
5.39940953e-01 4.04273979e-02 9.49320272e-02 6.58846915e-01
2.37746686e-01 4.68222320e-01 -6.51514769e-01 4.53606248e-02
8.03105891e-01 -6.21043332e-02 8.01787376e-02 -8.20081234e-01
-1.33161879e+00 6.36108577e-01 3.58460128e-01 3.12895417e-01
-3.69883567e-01 3.48568588e-01 4.95519638e-01 6.53009295e-01
1.24661148e-01 7.81169057e-01 -5.98877668e-01 -8.51963401e-01
-8.98912251e-01 8.56298625e-01 8.48218799e-01 1.01861632e+00
7.49277949e-01 1.11308414e-02 3.52918659e-03 7.24852443e-01
2.28940144e-01 1.37805536e-01 3.89340073e-01 -1.41361570e+00
2.77753651e-01 6.14252508e-01 4.11046118e-01 -4.76145208e-01
-5.51299751e-01 -3.86321723e-01 -2.95814984e-02 5.46477437e-01
6.56239152e-01 -4.08921033e-01 -5.45980752e-01 1.51241875e+00
4.25713480e-01 8.35678726e-02 -4.41033356e-02 1.13879645e+00
4.86975193e-01 1.91264004e-01 -8.79550651e-02 4.00154926e-02
7.38839149e-01 -1.33090878e+00 -3.86461705e-01 -3.08617890e-01
6.03034377e-01 -7.12975502e-01 1.32497561e+00 6.59328341e-01
-8.10033083e-01 -3.95602375e-01 -1.23802626e+00 1.56629220e-01
-5.45619190e-01 -2.34699845e-01 9.93656278e-01 6.91498935e-01
-1.12510705e+00 8.99617255e-01 -9.15415049e-01 -8.43488634e-01
1.68289348e-01 2.07125008e-01 -3.15840006e-01 -3.27056080e-01
-7.88251400e-01 1.26764083e+00 3.57525676e-01 -1.78247109e-01
-1.17415679e+00 -6.91242099e-01 -4.92425263e-01 -3.14514101e-01
9.08907175e-01 -3.96212608e-01 1.77018392e+00 -7.99059212e-01
-1.90399694e+00 4.45241809e-01 6.71212852e-01 -4.43814784e-01
6.21689677e-01 -4.67001289e-01 -2.73475528e-01 -3.49889815e-01
-1.16640694e-01 5.78450441e-01 6.31430805e-01 -1.03260553e+00
-7.00711727e-01 -3.66118491e-01 5.12397647e-01 4.23356563e-01
-2.15790108e-01 -1.45465769e-02 -2.86053121e-01 -3.30706030e-01
-3.79769355e-01 -1.05421662e+00 -3.33127081e-01 -5.73288016e-02
1.76128834e-01 -3.56809855e-01 6.42584026e-01 -2.21857116e-01
9.87634003e-01 -2.02830458e+00 5.52149117e-02 -9.02077556e-02
-8.30478147e-02 4.65630442e-01 -4.54603314e-01 8.72168541e-01
3.26899678e-01 -2.90728986e-01 5.97706661e-02 -3.28613743e-02
3.99270296e-01 4.55795437e-01 -5.71947917e-02 3.04166466e-01
1.78807769e-02 6.45331621e-01 -1.57926428e+00 -9.48181897e-02
4.89376754e-01 -4.84669879e-02 -7.85420716e-01 2.23850101e-01
-7.17874944e-01 3.15712184e-01 -2.97416538e-01 6.44440114e-01
2.90989161e-01 8.04500952e-02 2.81409532e-01 1.51613414e-01
-4.66750532e-01 5.25163889e-01 -1.44833851e+00 2.18968916e+00
-4.03903335e-01 3.17143500e-01 7.95112178e-02 -9.88623738e-01
7.69505978e-01 1.45552397e-01 5.95799267e-01 -5.46772897e-01
-7.53525272e-02 2.38633528e-01 1.43948883e-01 -6.21372521e-01
7.29295373e-01 2.83017874e-01 1.22741446e-01 5.15871227e-01
5.36308467e-01 -2.50784010e-01 4.63536918e-01 -4.92587052e-02
1.51487708e+00 9.36604857e-01 2.32565686e-01 -7.63488635e-02
-7.44449766e-03 3.75939280e-01 5.25946558e-01 9.17219758e-01
-3.50654155e-01 1.27199218e-01 2.33667254e-01 -5.75805187e-01
-7.34385252e-01 -8.94892395e-01 1.37326047e-01 1.54828012e+00
-5.06687164e-02 -7.36112475e-01 -7.97369838e-01 -9.67921197e-01
1.98222801e-01 7.78844357e-01 -4.26499635e-01 1.17835343e-01
-3.22298765e-01 -3.38947713e-01 6.73940301e-01 4.01607275e-01
4.56548929e-01 -1.36782157e+00 -8.97292435e-01 4.94720697e-01
3.05009186e-01 -8.42408717e-01 -7.00044259e-02 3.94729376e-01
-8.16727161e-01 -1.19931269e+00 -4.61035013e-01 -3.90425414e-01
1.67078018e-01 3.87533128e-01 1.22742176e+00 -6.53405637e-02
-1.13431424e-01 7.62004316e-01 -6.67506754e-01 -5.42502284e-01
-2.19525650e-01 2.00210392e-01 1.40153989e-01 -6.60747945e-01
3.25979561e-01 -6.16414905e-01 -6.24894559e-01 3.37549418e-01
-7.45414257e-01 -1.77142382e-01 5.64653039e-01 5.65529466e-01
2.88225591e-01 -2.20078617e-01 6.91391706e-01 -8.19797635e-01
7.96525180e-01 -6.31524801e-01 -7.32400179e-01 2.38151103e-01
-1.07700217e+00 2.91552767e-02 6.47327960e-01 -2.60961145e-01
-8.04314137e-01 -3.02220266e-02 -2.22701520e-01 -3.66511703e-01
-4.56334680e-01 6.80048168e-01 1.61201522e-01 1.00058960e-02
8.52511466e-01 1.74862519e-02 7.25176260e-02 -6.45249724e-01
6.97340190e-01 5.98438919e-01 3.24286491e-01 -9.63112712e-01
4.62693661e-01 1.14473984e-01 -3.68809104e-01 -5.93546867e-01
-8.20100367e-01 -4.19549704e-01 -1.88723147e-01 -4.18461710e-01
3.19321007e-01 -5.68494439e-01 -8.24623168e-01 2.36534104e-01
-5.74100614e-01 -1.06080639e+00 -6.05865598e-01 4.84085411e-01
-7.75397956e-01 3.54718305e-02 -4.25344527e-01 -6.42999470e-01
-1.23911925e-01 -1.27083588e+00 5.60946763e-01 3.70268553e-01
-1.53350949e-01 -7.64769375e-01 5.79936802e-01 4.37014848e-01
6.74697280e-01 -9.04486328e-02 7.62788415e-01 -8.69282544e-01
-4.59622025e-01 -1.29116043e-01 2.36639697e-02 4.75881726e-01
-1.49751842e-01 7.60171339e-02 -8.19577157e-01 -3.12099963e-01
-3.74574780e-01 -8.53460610e-01 3.49346757e-01 2.33732518e-02
9.81984019e-01 -4.80830967e-02 8.87115076e-02 3.36604774e-01
1.35562003e+00 1.91041544e-01 3.52093577e-01 6.20978713e-01
2.85287172e-01 4.12802130e-01 8.81643951e-01 4.63739932e-01
6.52901411e-01 7.10067689e-01 6.20153069e-01 2.17566550e-01
-4.04687114e-02 -4.84314501e-01 5.92442393e-01 8.30992997e-01
-2.56174296e-01 4.34104912e-02 -9.47216332e-01 3.70622933e-01
-2.19092035e+00 -1.12361360e+00 1.20977171e-01 2.17618632e+00
5.90438664e-01 1.90069318e-01 4.91170287e-01 -5.05510420e-02
4.65657562e-02 8.81833434e-02 -8.82665455e-01 -4.36773419e-01
2.59942234e-01 2.74267346e-01 3.27903271e-01 2.70145714e-01
-8.70022774e-01 1.07939577e+00 6.61190796e+00 6.44835293e-01
-1.22526538e+00 1.26972869e-01 1.51375346e-02 -9.87588614e-02
-7.32774585e-02 3.54800791e-01 -7.09908068e-01 1.63436592e-01
9.35219646e-01 -1.33606404e-01 1.13069963e+00 1.35474157e+00
2.03318715e-01 -2.71772444e-01 -1.30734932e+00 8.28117669e-01
-1.26492932e-01 -1.37960255e+00 -4.69086170e-01 -5.22028171e-02
5.50143600e-01 8.76119494e-01 -1.05974004e-01 1.03496635e+00
8.89853239e-01 -7.66788363e-01 7.63435960e-01 4.23024803e-01
3.96917671e-01 -3.99504095e-01 4.70429122e-01 5.42420685e-01
-6.14532232e-01 -1.12425067e-01 -3.95277798e-01 -4.39829856e-01
-3.67285252e-01 1.54580384e-01 -9.50711489e-01 6.59230292e-01
7.49763191e-01 9.64868903e-01 -5.55299699e-01 1.22718692e+00
-2.08432898e-01 6.58051670e-01 -2.42202267e-01 -1.90456331e-01
3.84407163e-01 -2.36366317e-01 2.49326199e-01 1.06669652e+00
6.47437647e-02 -4.32686150e-01 5.44036031e-01 4.47904825e-01
1.91705674e-02 1.14886492e-01 -7.12715924e-01 -1.98162064e-01
5.67985535e-01 1.59932935e+00 -5.73115468e-01 -1.08287120e-02
-6.20951653e-01 7.20932901e-01 5.89282572e-01 1.35043144e-01
-6.83695376e-01 -1.36400148e-01 9.49082971e-01 5.23657687e-02
-3.30384709e-02 -4.01476353e-01 -4.30204682e-02 -1.05641651e+00
-1.19535029e-01 -1.37103987e+00 2.75169939e-01 -6.61579251e-01
-1.31696773e+00 1.66085616e-01 8.32861438e-02 -1.06564021e+00
-3.43774587e-01 -6.50833011e-01 -3.37119401e-01 3.30640733e-01
-1.44166768e+00 -8.77793491e-01 -3.40602309e-01 4.23812389e-01
7.02966511e-01 -2.96611607e-01 1.16445792e+00 3.78587246e-01
-5.22299290e-01 3.41196179e-01 2.44389087e-01 -9.28637534e-02
8.91353011e-01 -1.36937821e+00 3.05498242e-01 5.18670380e-01
4.75902796e-01 5.81651866e-01 6.96848750e-01 -3.18027645e-01
-1.74928534e+00 -8.45819712e-01 2.13468567e-01 -5.40048182e-01
8.43374133e-01 -3.08893412e-01 -5.07494807e-01 6.86744571e-01
4.14081514e-01 -5.86615317e-03 4.10380960e-01 4.99054700e-01
-2.95884430e-01 -3.38542521e-01 -9.32942450e-01 5.51028967e-01
1.16279137e+00 -3.16077203e-01 -5.05775273e-01 4.34804499e-01
3.67644787e-01 -7.16743112e-01 -1.03637636e+00 3.08907777e-01
7.47647882e-01 -1.15873349e+00 6.92924619e-01 -5.56563258e-01
3.15134555e-01 -3.50674957e-01 -2.71265864e-01 -1.64106476e+00
-1.87924802e-01 -1.00770628e+00 -7.84323644e-03 1.16780734e+00
3.54865253e-01 -3.25149029e-01 8.74031782e-01 3.29814255e-01
-3.25338870e-01 -9.13428307e-01 -5.51322818e-01 -8.34934771e-01
6.42811805e-02 -6.64555252e-01 6.40515745e-01 1.06201828e+00
3.72191489e-01 3.98912877e-01 -3.06624144e-01 -2.80531853e-01
3.23656529e-01 3.43741961e-02 1.24628317e+00 -1.15776169e+00
-6.41367674e-01 -4.33182657e-01 -1.39497489e-01 -1.05443633e+00
-4.85977111e-03 -9.51602697e-01 1.17362469e-01 -1.78020573e+00
-2.30891183e-01 -8.80492985e-01 -6.10351920e-01 7.73217857e-01
2.33620450e-01 -1.29823953e-01 3.44581634e-01 2.32426878e-02
-9.88023639e-01 3.75617355e-01 1.07858980e+00 7.01303557e-02
-2.54118025e-01 1.92166883e-02 -7.02793896e-01 6.76516116e-01
8.94241750e-01 -5.06839454e-01 -7.04263806e-01 -6.74476206e-01
3.56470436e-01 5.15800379e-02 2.12755337e-01 -1.49479127e+00
3.07104230e-01 -4.68754828e-01 4.63773571e-02 -1.78466395e-01
2.47466341e-01 -7.28112042e-01 -1.32734150e-01 8.72815922e-02
-5.21869540e-01 1.71847746e-01 2.20410615e-01 4.31410998e-01
1.31488398e-01 -2.92942762e-01 6.09453321e-01 -2.49193385e-01
-9.28634644e-01 2.05406845e-01 -3.57058316e-01 3.00658226e-01
1.04234731e+00 -6.79138629e-03 -5.77662289e-01 -2.82262295e-01
-3.84116888e-01 3.81827056e-01 5.70382774e-01 6.76369309e-01
3.55815411e-01 -8.01831603e-01 -4.89443451e-01 -1.90981142e-02
3.41492265e-01 -2.54117787e-01 -1.20945200e-01 7.04816937e-01
-3.80879045e-01 2.88793921e-01 -4.78554398e-01 -2.85225928e-01
-7.74290144e-01 4.49618816e-01 3.53751838e-01 -2.91117042e-01
-5.93666852e-01 5.99412322e-01 -7.01122165e-01 -1.07128751e+00
4.22078431e-01 -3.70965272e-01 8.88564438e-02 -2.53611147e-01
4.50917333e-01 4.21401054e-01 4.57321972e-01 1.30147189e-01
-1.95797145e-01 -1.20058559e-01 -8.12699497e-02 -2.41629049e-01
1.70722997e+00 3.02509546e-01 2.77758300e-01 7.70411670e-01
3.57300550e-01 -1.60964116e-01 -1.53224969e+00 -1.90223575e-01
4.66869265e-01 -3.24139118e-01 1.67388037e-01 -1.36529446e+00
-7.00303376e-01 6.55255079e-01 6.02028966e-01 2.54736155e-01
7.50838816e-01 -1.59856856e-01 4.41944629e-01 8.95197153e-01
8.05215955e-01 -1.53231180e+00 -1.66457519e-02 6.77890956e-01
8.03041041e-01 -1.15330958e+00 7.89121464e-02 2.55370975e-01
-6.34155869e-01 1.03026354e+00 8.56732428e-01 -3.40129942e-01
5.47415435e-01 3.15638393e-01 1.57834649e-01 -9.04445946e-02
-1.23011994e+00 -5.28412461e-01 -3.37165982e-01 8.91425490e-01
7.04174161e-01 7.33203441e-03 -2.09886342e-01 3.88225198e-01
-1.95785642e-01 5.24611294e-01 3.98577273e-01 1.16465855e+00
-3.87707472e-01 -1.62017560e+00 -3.05329002e-02 4.63498890e-01
-2.03278765e-01 1.04378171e-01 -1.90491617e-01 9.68713105e-01
1.93225399e-01 1.04974186e+00 -2.81476825e-01 -5.39211988e-01
6.00328505e-01 1.00103080e-01 9.59826112e-01 -7.49944568e-01
-9.97830808e-01 -3.71483624e-01 3.73323560e-01 -8.60908389e-01
-3.66236061e-01 -5.62318325e-01 -1.05841863e+00 -3.05505812e-01
-1.33472225e-02 1.42979324e-01 1.06784987e+00 9.95839179e-01
4.10142839e-01 2.95791954e-01 2.15517178e-01 -1.02593577e+00
-8.10422003e-01 -9.12456751e-01 -3.33481789e-01 2.28977963e-01
-1.29463345e-01 -8.52167547e-01 1.71305493e-01 -2.94435441e-01]
|
[4.076313018798828, 1.546043872833252]
|
26fc6b82-128e-4bdc-9be7-cabf4a0c6e76
|
multimodal-representation-learning-of
|
2304.07675
| null |
https://arxiv.org/abs/2304.07675v1
|
https://arxiv.org/pdf/2304.07675v1.pdf
|
Multimodal Representation Learning of Cardiovascular Magnetic Resonance Imaging
|
Self-supervised learning is crucial for clinical imaging applications, given the lack of explicit labels in healthcare. However, conventional approaches that rely on precise vision-language alignment are not always feasible in complex clinical imaging modalities, such as cardiac magnetic resonance (CMR). CMR provides a comprehensive visualization of cardiac anatomy, physiology, and microstructure, making it challenging to interpret. Additionally, CMR reports require synthesizing information from sequences of images and different views, resulting in potentially weak alignment between the study and diagnosis report pair. To overcome these challenges, we propose \textbf{CMRformer}, a multimodal learning framework to jointly learn sequences of CMR images and associated cardiologist's reports. Moreover, one of the major obstacles to improving CMR study is the lack of large, publicly available datasets. To bridge this gap, we collected a large \textbf{CMR dataset}, which consists of 13,787 studies from clinical cases. By utilizing our proposed CMRformer and our collected dataset, we achieved remarkable performance in real-world clinical tasks, such as CMR image retrieval and diagnosis report retrieval. Furthermore, the learned representations are evaluated to be practically helpful for downstream applications, such as disease classification. Our work could potentially expedite progress in the CMR study and lead to more accurate and effective diagnosis and treatment.
|
['David Chen', 'Ding Zhao', 'Douglas Weber', 'Debbie Kwon', 'Byung-Hak Kim', 'Christopher Nguyen', 'Pohao Chen', 'Wilson Tang', 'Jiacheng Zhu', 'Jaehyun Lee', 'Makiya Nakashima', 'Peide Huang', 'JieLin Qiu']
|
2023-04-16
| null | null | null | null |
['anatomy']
|
['miscellaneous']
|
[ 2.70123720e-01 -2.12701201e-01 -3.04723233e-01 -3.95038068e-01
-1.18067229e+00 -6.76111758e-01 1.14693724e-01 4.70782131e-01
-1.40085161e-01 7.47873545e-01 3.38202178e-01 -4.45810229e-01
-2.61677742e-01 -3.24459821e-01 -2.99899340e-01 -7.12292910e-01
-2.00609177e-01 4.63341027e-01 -8.88495147e-02 3.53731543e-01
1.86677405e-03 3.60954434e-01 -8.12327445e-01 5.17460763e-01
5.62600553e-01 9.72465098e-01 6.16733670e-01 5.97270310e-01
3.63808870e-02 1.00781751e+00 -3.83073598e-01 -1.42643258e-01
-4.49907221e-02 -6.34820402e-01 -7.53132522e-01 2.52420425e-01
4.08902854e-01 -4.64610487e-01 -4.59288746e-01 8.41125965e-01
8.79382133e-01 -2.05553994e-01 5.54131269e-01 -7.58094311e-01
-8.10893834e-01 6.46467865e-01 -7.16736674e-01 7.89638221e-01
6.20452911e-02 4.06933516e-01 7.80481219e-01 -7.67042696e-01
8.19219112e-01 4.62204099e-01 4.20873940e-01 3.82945031e-01
-1.07468343e+00 -5.87265193e-01 1.73432156e-02 3.17107230e-01
-1.23663044e+00 -3.98674160e-01 8.37024212e-01 -7.01830029e-01
5.74741542e-01 3.48765105e-01 6.96360230e-01 8.10738862e-01
1.12339854e-01 4.99523640e-01 1.06709123e+00 -1.70541987e-01
-7.85621777e-02 -8.46126303e-02 -1.19726323e-01 8.84454072e-01
2.47628018e-01 -1.66472480e-01 -3.70162845e-01 -2.15410754e-01
1.06883335e+00 4.15611655e-01 -6.55854285e-01 -4.19040233e-01
-2.04565477e+00 5.30656874e-01 4.96434808e-01 5.25055051e-01
-5.74909925e-01 -7.30890632e-02 6.16008759e-01 1.88058168e-01
2.82850266e-01 2.76471943e-01 8.64376873e-03 1.89631104e-01
-1.00205588e+00 -2.46016994e-01 3.07529896e-01 7.41362512e-01
3.19134556e-02 -2.33124606e-02 -2.23506913e-01 1.10450721e+00
3.70786816e-01 6.69340312e-01 6.01386666e-01 -1.02366757e+00
6.61890626e-01 3.57713193e-01 -2.12344348e-01 -1.26576769e+00
-5.53830683e-01 -6.24215066e-01 -1.45468271e+00 -1.81511864e-01
2.98884362e-01 2.47335713e-02 -6.54412985e-01 1.40635097e+00
4.41152044e-02 2.72565216e-01 -6.10564947e-02 1.56342041e+00
1.34361291e+00 3.64140600e-01 7.97684789e-02 -6.44128919e-01
1.41781795e+00 -9.16044831e-01 -5.90968013e-01 2.66272910e-02
8.11929464e-01 -7.92671084e-01 7.53179312e-01 1.12190567e-01
-9.98057425e-01 -3.87589663e-01 -8.89706612e-01 3.31299067e-01
2.61413962e-01 4.90041852e-01 8.16461682e-01 6.16549924e-02
-8.91145647e-01 8.38278085e-02 -9.11597610e-01 -1.68059573e-01
7.12767243e-01 1.66739807e-01 -6.67079508e-01 -4.23928589e-01
-8.64274621e-01 7.41943300e-01 2.94952959e-01 5.45266926e-01
-8.87399077e-01 -8.69364977e-01 -7.91263402e-01 -2.29160085e-01
4.39276814e-01 -9.33494151e-01 8.67424011e-01 -3.83159488e-01
-9.14730608e-01 1.43601096e+00 1.20060910e-02 -2.80225605e-01
4.91463482e-01 1.53698787e-01 -4.87681270e-01 6.18390679e-01
2.18529314e-01 5.74690640e-01 7.60321259e-01 -1.09411800e+00
-2.57603079e-01 -4.77270365e-01 -1.06394611e-01 1.20060086e-01
6.55544316e-03 -1.65341701e-02 -4.04375046e-01 -8.93209457e-01
5.04225612e-01 -9.29203212e-01 -4.37025517e-01 1.85905427e-01
-3.78435373e-01 1.79093987e-01 3.54775220e-01 -9.08090353e-01
1.14112365e+00 -2.19469333e+00 5.92991300e-02 -4.78293672e-02
8.88288856e-01 2.62631088e-01 2.99706738e-02 -5.46291918e-02
-4.36589390e-01 1.90319389e-01 -2.42942289e-01 -7.64229670e-02
-6.92503512e-01 -6.43193498e-02 -1.63827911e-01 5.85066020e-01
2.03141928e-01 1.07821035e+00 -9.87514913e-01 -9.01665866e-01
2.81996965e-01 2.67679513e-01 -2.64385521e-01 3.81065756e-01
1.85287908e-01 1.24356461e+00 -5.65243661e-01 6.78439260e-01
3.24126452e-01 -8.43783379e-01 3.89156789e-01 -5.61445653e-01
1.44723594e-01 1.48567855e-01 -7.38713682e-01 2.09427667e+00
-4.82177258e-01 5.18813014e-01 -4.71837744e-02 -1.25659573e+00
7.92200148e-01 5.49822211e-01 1.04421675e+00 -7.58841813e-01
-4.39747982e-02 1.71165466e-01 1.46513894e-01 -8.69777977e-01
-1.03527643e-01 -2.20692515e-01 2.22765982e-01 7.58763850e-01
-1.89738885e-01 -1.05511457e-01 1.06106550e-01 2.40134835e-01
1.14085817e+00 -3.14319074e-01 2.62563527e-01 1.46236405e-01
6.24155819e-01 6.04556799e-02 6.22161865e-01 8.32496703e-01
-5.61767459e-01 1.09730124e+00 2.76176602e-01 -8.07088196e-01
-9.47847188e-01 -1.21258616e+00 -1.70405626e-01 4.83459890e-01
-8.10186416e-02 -2.70923138e-01 -7.05376193e-02 -7.66411424e-01
-2.97359705e-01 -1.26974344e-01 -1.91636533e-01 2.73107681e-02
-8.27926219e-01 -7.95558155e-01 3.60161036e-01 5.92289627e-01
2.40852594e-01 -9.95424211e-01 -6.72230363e-01 3.91874492e-01
-6.44927144e-01 -1.43947768e+00 -6.97195947e-01 -2.05556095e-01
-1.24012232e+00 -1.21722949e+00 -1.17916977e+00 -7.92632937e-01
6.57543004e-01 2.78426677e-01 1.29757214e+00 4.10771698e-01
-7.15413034e-01 3.94084692e-01 -2.60122418e-01 -5.89362942e-02
-3.48525196e-01 7.13554397e-02 -9.07809958e-02 -4.02362347e-02
-3.18955183e-01 -5.34678817e-01 -9.33744490e-01 -1.16531402e-02
-7.75818110e-01 1.82153285e-01 8.02064538e-01 1.13262475e+00
8.58579695e-01 -2.94242829e-01 7.15556324e-01 -9.63657260e-01
2.90303767e-01 -4.56507772e-01 -2.49634907e-01 5.57130933e-01
-4.33717698e-01 -2.28816971e-01 2.28748947e-01 -2.25926369e-01
-8.02076876e-01 -9.27795619e-02 9.94352400e-02 -5.99164128e-01
-1.82779193e-01 7.80119717e-01 4.08895254e-01 1.23367891e-01
3.69817197e-01 3.85170639e-01 2.22634792e-01 -2.34002247e-01
1.14613585e-01 6.51465058e-01 6.70263052e-01 -3.73852611e-01
3.87571186e-01 5.31844258e-01 1.71678007e-01 -6.55475020e-01
-9.37605917e-01 -5.94956040e-01 -7.31642067e-01 -3.33637863e-01
8.50219488e-01 -8.92497718e-01 -6.25188708e-01 1.22384988e-01
-1.17593300e+00 2.29953811e-01 1.21584255e-02 8.18038166e-01
-4.70142484e-01 5.36718905e-01 -8.58523607e-01 -4.27556545e-01
-7.17252910e-01 -1.38456118e+00 8.55639756e-01 3.98900248e-02
-1.33426622e-01 -1.07277560e+00 -8.62144306e-02 7.58642912e-01
4.42837745e-01 3.16150188e-01 1.16707098e+00 -3.50280404e-01
-8.15047264e-01 1.50247514e-01 -4.30524439e-01 3.25627178e-01
4.85473871e-01 -2.74865985e-01 -5.22005498e-01 -3.24275494e-01
9.02456492e-02 -4.03078556e-01 9.25361812e-01 4.68717873e-01
1.26829159e+00 8.23879167e-02 -1.04791664e-01 6.24094188e-01
1.11433971e+00 3.29618305e-01 2.54241407e-01 4.55418378e-02
1.00604272e+00 4.60032374e-01 5.25912344e-01 4.47525591e-01
5.89968860e-01 5.85562110e-01 2.20477611e-01 -4.85211551e-01
-3.24515760e-01 2.73666456e-02 -2.27628708e-01 1.54711235e+00
-2.26793304e-01 2.63938725e-01 -1.34917760e+00 6.21999145e-01
-1.74480748e+00 -6.68739736e-01 -4.69502099e-02 1.94825339e+00
8.89388859e-01 -1.10842317e-01 -1.57757461e-01 -1.40013248e-01
6.83024764e-01 2.39673063e-01 -5.70659637e-01 2.34034359e-01
-7.41084740e-02 4.61892299e-02 1.27191499e-01 3.26756090e-02
-1.24712729e+00 3.92490774e-01 6.32137823e+00 2.15206534e-01
-1.58444130e+00 4.26331699e-01 8.27839136e-01 -1.11233138e-01
-2.22011730e-01 -2.68584430e-01 -1.76177859e-01 3.45238954e-01
4.56453115e-01 -1.01178080e-01 7.58639947e-02 3.65892500e-01
3.65612060e-01 -2.69036964e-02 -1.21055543e+00 1.54222667e+00
3.23067218e-01 -1.75389028e+00 1.02101743e-01 -1.40190601e-01
5.14329851e-01 3.75953875e-02 7.25262165e-02 -2.30706893e-02
-2.49585941e-01 -1.04811609e+00 1.25451386e-01 7.25862563e-01
1.05203795e+00 -2.45012075e-01 8.38701963e-01 7.78299943e-02
-9.53826368e-01 3.52959126e-01 -1.34964108e-01 3.34316075e-01
2.08020002e-01 8.21176350e-01 -9.36883926e-01 8.90558481e-01
5.34275591e-01 1.09887064e+00 -4.25071985e-01 1.20135820e+00
7.40640983e-02 6.93708241e-01 1.70035046e-02 5.16223609e-01
-1.59372061e-01 -1.47456020e-01 5.30551136e-01 9.26966012e-01
1.76947162e-01 4.03625250e-01 6.62680686e-01 9.35174048e-01
3.14911385e-03 2.62673438e-01 -6.98396742e-01 -2.91399211e-01
2.42418557e-01 1.34874439e+00 -9.13604975e-01 -3.86798441e-01
-5.67211628e-01 6.66381598e-01 2.34516650e-01 3.37554008e-01
-7.93175280e-01 2.84803063e-01 1.00299038e-01 2.39899203e-01
-3.48665938e-02 -2.93619394e-01 -3.02960306e-01 -1.51743007e+00
2.38290668e-01 -8.62476468e-01 6.71803951e-01 -6.96232259e-01
-1.49780917e+00 7.02747524e-01 -2.60024637e-01 -1.47688925e+00
-2.34595224e-01 -2.12042451e-01 -2.82476425e-01 9.04167175e-01
-1.55920935e+00 -1.18001056e+00 -3.38806391e-01 3.88127476e-01
3.72987241e-01 -3.68562073e-01 8.62461627e-01 7.52293944e-01
-5.76648831e-01 3.62526953e-01 -1.88191414e-01 2.83617437e-01
9.64981377e-01 -9.02178168e-01 -3.25118899e-01 3.88544261e-01
2.72381544e-01 7.84559727e-01 -3.22497450e-02 -4.85377669e-01
-1.28119600e+00 -9.90240216e-01 5.58988988e-01 -3.61418247e-01
6.34002209e-01 2.21885026e-01 -9.25578415e-01 5.47997832e-01
-7.43086115e-02 6.00152969e-01 8.95341337e-01 -2.53525496e-01
-1.99233070e-01 -3.16831678e-01 -8.04992199e-01 4.03956443e-01
9.54085410e-01 -7.69496143e-01 -6.32716954e-01 4.58082169e-01
5.60059547e-01 -7.51907527e-01 -1.36409748e+00 6.55963421e-01
5.40916145e-01 -6.13761425e-01 1.01175606e+00 -6.92105830e-01
7.18173265e-01 -3.09238315e-01 -2.65699695e-03 -1.00351965e+00
-1.72635391e-01 -4.99161109e-02 3.82624543e-03 1.12480319e+00
4.21921492e-01 -5.50288200e-01 5.74799299e-01 2.89455861e-01
-6.44898713e-02 -1.05035186e+00 -8.31543446e-01 -2.11410373e-01
-3.85414436e-02 -4.06678826e-01 1.31009653e-01 1.30875075e+00
-2.60469973e-01 4.04728651e-02 -2.52096534e-01 1.95072040e-01
7.40146101e-01 5.16642690e-01 2.57754326e-01 -9.16171134e-01
-4.29921806e-01 -3.38048637e-01 -4.85134542e-01 -6.56072021e-01
4.15566750e-03 -1.25549555e+00 -1.36634707e-01 -1.67618394e+00
7.50702500e-01 -8.85355532e-01 -8.17765832e-01 4.61091429e-01
-2.90932268e-01 5.61354280e-01 2.34765381e-01 7.19047725e-01
-8.01296771e-01 2.17201322e-01 1.64790583e+00 -5.07721782e-01
1.49114206e-01 -1.53755739e-01 -6.25167787e-01 6.22398853e-01
5.64600766e-01 -5.45751214e-01 -3.90038490e-01 -5.59610307e-01
-1.20601878e-01 7.56230533e-01 5.79678833e-01 -7.77397513e-01
2.50887752e-01 -2.90092104e-03 4.54926312e-01 -6.38401568e-01
-1.85495894e-02 -6.23671830e-01 1.36656106e-01 4.55100745e-01
-4.49578047e-01 2.54789919e-01 -8.25451612e-02 3.64079773e-01
-5.11593759e-01 5.09936847e-02 5.27294993e-01 -5.75271726e-01
-4.95088786e-01 5.67886412e-01 -9.29249078e-02 3.07768941e-01
9.83768880e-01 2.43886739e-01 -2.27152348e-01 -2.58382410e-01
-1.11565161e+00 2.79044032e-01 -5.28714387e-03 4.78035301e-01
9.58037317e-01 -1.23060131e+00 -1.01009881e+00 -5.25250658e-02
2.25153908e-01 2.06513792e-01 7.48091102e-01 1.52011263e+00
-5.88141143e-01 6.68662429e-01 -1.26136348e-01 -1.16826463e+00
-1.18333137e+00 3.82063091e-01 3.00886095e-01 -4.84090745e-01
-9.88021791e-01 2.75176287e-01 2.87247121e-01 -3.60292763e-01
1.61219344e-01 -2.76744902e-01 -1.80271342e-01 -9.38148201e-02
6.23026788e-01 -1.99199334e-01 1.78077728e-01 -7.01824009e-01
-6.14155710e-01 6.06968880e-01 -4.44169998e-01 1.12203360e-01
1.36968851e+00 -3.90014678e-01 -1.85315728e-01 5.43641329e-01
1.00433767e+00 -2.82972485e-01 -7.59722531e-01 -4.47434485e-01
-1.33627206e-01 -3.54285687e-01 2.29509041e-01 -7.90336490e-01
-1.58829939e+00 9.63541090e-01 7.75261879e-01 -2.15808302e-01
1.00537956e+00 1.39627367e-01 8.87990177e-01 2.02844203e-01
5.48851848e-01 -4.09141034e-01 1.31084263e-01 4.43854406e-02
8.69100928e-01 -1.76182914e+00 7.03022256e-02 -4.80961174e-01
-8.63543987e-01 1.00556242e+00 3.56981516e-01 3.23976308e-01
4.84065771e-01 1.37294248e-01 5.89958191e-01 -3.20555061e-01
-7.80347347e-01 3.41541283e-02 4.08390045e-01 5.35383582e-01
8.33315253e-01 1.55125499e-01 -3.27078640e-01 5.97004533e-01
3.11961561e-01 1.68376788e-01 3.40755075e-01 9.45504010e-01
1.11026399e-01 -9.50904667e-01 -3.11998785e-01 6.76286817e-01
-6.87363625e-01 -2.02776998e-01 -1.78534184e-02 3.85201007e-01
3.60591784e-02 6.84130251e-01 -3.44208032e-01 -9.89849046e-02
2.21628100e-01 -1.08481072e-01 5.69483399e-01 -7.70029426e-01
-2.86115676e-01 4.48338129e-02 -1.65811747e-01 -3.21583271e-01
-5.67772031e-01 -6.26938522e-01 -1.29194033e+00 5.29341064e-02
2.83088423e-02 -6.07333705e-02 4.51123238e-01 1.03086150e+00
4.01706040e-01 8.40572357e-01 6.50164127e-01 -3.71063113e-01
-3.38125437e-01 -7.25038171e-01 -4.75592464e-01 6.46502137e-01
4.38855171e-01 -6.43464029e-01 1.36478081e-01 5.13924897e-01]
|
[14.885512351989746, -1.8815430402755737]
|
39783990-e168-4aa4-b1d5-1dfa87069ebe
|
model-based-offline-reinforcement-learning
|
2210.06692
| null |
https://arxiv.org/abs/2210.06692v2
|
https://arxiv.org/pdf/2210.06692v2.pdf
|
Model-Based Offline Reinforcement Learning with Pessimism-Modulated Dynamics Belief
|
Model-based offline reinforcement learning (RL) aims to find highly rewarding policy, by leveraging a previously collected static dataset and a dynamics model. While the dynamics model learned through reuse of the static dataset, its generalization ability hopefully promotes policy learning if properly utilized. To that end, several works propose to quantify the uncertainty of predicted dynamics, and explicitly apply it to penalize reward. However, as the dynamics and the reward are intrinsically different factors in context of MDP, characterizing the impact of dynamics uncertainty through reward penalty may incur unexpected tradeoff between model utilization and risk avoidance. In this work, we instead maintain a belief distribution over dynamics, and evaluate/optimize policy through biased sampling from the belief. The sampling procedure, biased towards pessimism, is derived based on an alternating Markov game formulation of offline RL. We formally show that the biased sampling naturally induces an updated dynamics belief with policy-dependent reweighting factor, termed Pessimism-Modulated Dynamics Belief. To improve policy, we devise an iterative regularized policy optimization algorithm for the game, with guarantee of monotonous improvement under certain condition. To make practical, we further devise an offline RL algorithm to approximately find the solution. Empirical results show that the proposed approach achieves state-of-the-art performance on a wide range of benchmark tasks.
|
['Yanhui Geng', 'Yunfeng Shao', 'Kaiyang Guo']
|
2022-10-13
| null | null | null | null |
['d4rl']
|
['robots']
|
[-8.81704465e-02 2.14080229e-01 -7.99446225e-01 -4.71615382e-02
-5.98295093e-01 -5.09937823e-01 4.46194410e-01 6.92744926e-02
-6.44514978e-01 9.64796603e-01 2.66164280e-02 -3.81346345e-01
-3.72487605e-01 -7.98744321e-01 -7.77468085e-01 -8.78919482e-01
-4.60454911e-01 2.91450441e-01 -2.90275775e-02 -1.73088521e-01
4.18723792e-01 4.76843230e-02 -1.15620041e+00 -2.76223540e-01
1.16363239e+00 1.20647347e+00 3.90194952e-01 4.01913732e-01
1.94979832e-01 1.12020993e+00 -4.26978886e-01 -2.95313537e-01
5.15140295e-01 -3.89300883e-01 -5.36977232e-01 9.15510952e-02
-5.06670177e-01 -6.94954693e-01 -1.78389654e-01 1.18762577e+00
2.96315342e-01 3.79449129e-01 4.31929171e-01 -1.20722592e+00
-3.14659566e-01 8.99270117e-01 -6.01230562e-01 4.24730033e-01
4.21563573e-02 6.14098608e-01 1.03474903e+00 -1.67935044e-01
4.01481837e-01 1.16845679e+00 1.30834892e-01 5.62740147e-01
-1.21304166e+00 -5.34816265e-01 7.95053840e-01 1.93177953e-01
-1.14187515e+00 -7.92257190e-02 8.81731749e-01 -2.29918092e-01
5.24847567e-01 2.14410499e-02 9.48417544e-01 1.21635771e+00
4.54372704e-01 1.06718147e+00 1.45863998e+00 -4.70667854e-02
7.10788727e-01 2.15554699e-01 -2.16976821e-01 5.14090002e-01
2.38273710e-01 8.66061985e-01 -2.66867042e-01 -2.78811276e-01
7.94041812e-01 1.41902640e-02 -2.54877418e-01 -6.54979765e-01
-6.91569746e-01 7.56072104e-01 1.48208901e-01 -3.53170127e-01
-6.72304273e-01 3.50764513e-01 3.48223627e-01 5.54310024e-01
3.49659592e-01 3.87088001e-01 -5.46219885e-01 -6.32105649e-01
-4.88272816e-01 5.47381878e-01 6.60411239e-01 6.06030703e-01
5.27421832e-01 2.39093721e-01 -5.17617226e-01 4.53203917e-01
3.15420181e-01 3.91861022e-01 4.84373420e-01 -1.04202008e+00
5.57950616e-01 3.22657526e-01 7.85201669e-01 -9.70866084e-01
-7.06261471e-02 -6.14809632e-01 -4.11420524e-01 2.96469420e-01
3.78295392e-01 -3.29497725e-01 -5.04042923e-01 2.04197669e+00
4.82227445e-01 2.78079659e-01 4.10229750e-02 9.54568386e-01
-4.06992733e-01 5.40697098e-01 -7.84426462e-03 -8.88447106e-01
8.19009781e-01 -7.01624036e-01 -9.04305160e-01 -2.22403072e-02
2.81153828e-01 -6.26788810e-02 1.30323160e+00 5.73823929e-01
-1.14705420e+00 -1.18083283e-01 -9.94241059e-01 7.87104607e-01
2.95850337e-01 -1.25351056e-01 2.40186974e-01 6.07944429e-01
-5.93847871e-01 9.77751434e-01 -1.06597137e+00 3.68976444e-02
2.84956813e-01 2.42304593e-01 5.26213169e-01 3.92732143e-01
-1.17138565e+00 9.91306722e-01 5.82793891e-01 -4.27175574e-02
-1.68998075e+00 -6.69369817e-01 -4.01048064e-01 -1.10883698e-01
1.24248219e+00 -4.42756653e-01 1.62979650e+00 -7.93832183e-01
-2.15898204e+00 1.01364143e-01 2.62209982e-01 -9.71215189e-01
9.04516339e-01 -3.29804838e-01 -1.84579685e-01 1.21837649e-02
-2.49829397e-01 1.50811866e-01 1.34436488e+00 -1.38506079e+00
-7.65889645e-01 -7.94701055e-02 5.40381074e-01 4.46120888e-01
-4.40155417e-01 -4.12991226e-01 9.76265073e-02 -6.10687912e-01
-5.80041289e-01 -8.66963983e-01 -6.41393661e-01 -3.91652942e-01
-1.46413103e-01 -5.86225241e-02 5.50470769e-01 -4.41458672e-01
1.66304529e+00 -1.85423827e+00 1.15080081e-01 3.23169649e-01
1.23516321e-01 -1.84000712e-02 1.71337724e-01 3.94970030e-01
4.46121573e-01 -5.69312600e-03 -3.18544030e-01 -2.38760114e-02
1.44194230e-01 4.77194726e-01 -8.07460248e-01 5.23928404e-01
1.90442689e-02 7.52769113e-01 -1.19617867e+00 -2.74706066e-01
1.22517884e-01 -2.20279068e-01 -7.66130269e-01 4.98492181e-01
-6.41686022e-01 5.02983987e-01 -9.49318290e-01 4.93772477e-01
5.01324117e-01 -5.00377156e-02 3.92674983e-01 1.49112195e-01
-1.98065788e-01 6.22281730e-02 -1.28130853e+00 1.23520446e+00
-5.22545040e-01 -3.13605458e-01 1.33072928e-01 -1.30435359e+00
8.55590820e-01 5.24125807e-02 5.51329136e-01 -5.41643798e-01
3.96025062e-01 2.77391728e-02 1.56865809e-02 -3.99344265e-01
4.81998563e-01 -1.56193256e-01 -2.28066631e-02 6.08145952e-01
-2.74348021e-01 4.37783729e-03 -1.23404033e-01 3.54356016e-03
8.85498405e-01 4.28299278e-01 5.77454746e-01 -3.87426049e-01
3.76992613e-01 -1.85682736e-02 7.53850639e-01 1.04253304e+00
-6.94237769e-01 -3.10804278e-01 8.76083016e-01 -2.79146224e-01
-7.62368977e-01 -8.90343070e-01 2.02280283e-01 1.00537360e+00
3.59283715e-01 -1.43116161e-01 -5.26230276e-01 -9.45009887e-01
1.98739931e-01 8.80456150e-01 -7.48596668e-01 -4.66572434e-01
-4.55222368e-01 -7.77298212e-01 3.54296006e-02 2.20947459e-01
4.52416331e-01 -8.64817142e-01 -1.02041519e+00 5.31500161e-01
1.25688955e-01 -6.42891288e-01 -5.57783067e-01 7.95585737e-02
-7.90977478e-01 -8.92202973e-01 -6.11926854e-01 3.20617780e-02
4.60389853e-01 -4.12193709e-04 8.24309647e-01 -3.28197151e-01
2.30901867e-01 6.96260333e-01 -2.68447429e-01 -3.44685018e-01
-3.80367935e-01 -1.27240598e-01 4.43619072e-01 1.33201003e-01
-1.02423310e-01 -6.50771201e-01 -8.55688751e-01 2.13740766e-01
-7.32591629e-01 -1.20899260e-01 5.40595174e-01 1.05902267e+00
7.20534146e-01 1.69661760e-01 7.10861802e-01 -5.18439174e-01
1.10697377e+00 -7.93568850e-01 -1.00726008e+00 2.98872143e-01
-1.15350974e+00 5.11954606e-01 9.09798026e-01 -7.83701003e-01
-1.18187141e+00 -1.80391878e-01 3.42447549e-01 -7.66975343e-01
3.24735254e-01 5.62557638e-01 6.55905455e-02 1.54445872e-01
4.08515066e-01 4.42559659e-01 3.08711559e-01 -2.18569621e-01
5.18236756e-01 3.75400305e-01 1.68549970e-01 -1.30587173e+00
6.66377723e-01 4.61061001e-01 -4.91202250e-02 -2.22889483e-02
-9.06479776e-01 9.74641070e-02 1.48889333e-01 -5.85535526e-01
2.90789992e-01 -6.22134328e-01 -1.23729396e+00 1.06299482e-01
-5.62564135e-01 -5.70656180e-01 -4.93040979e-01 4.25150692e-01
-1.08985829e+00 2.54613161e-01 -2.65518516e-01 -1.56731033e+00
-1.56948522e-01 -1.12801611e+00 4.40663576e-01 2.50094354e-01
2.24500343e-01 -8.18765879e-01 3.68777275e-01 -1.63271010e-01
3.56221110e-01 3.00622523e-01 6.41709089e-01 -3.37378442e-01
-6.84177518e-01 2.42906138e-01 2.17503726e-01 3.22439522e-01
3.28679569e-04 -2.40460724e-01 -5.82288682e-01 -6.11138165e-01
4.02646065e-01 -5.39671838e-01 6.31237745e-01 4.00830507e-01
1.43449271e+00 -8.81751597e-01 -6.96666688e-02 2.53245652e-01
1.51324761e+00 6.40479982e-01 2.25765273e-01 5.41058838e-01
1.82948276e-01 4.92173165e-01 1.13883448e+00 1.29591548e+00
4.59601223e-01 4.45112675e-01 8.70163441e-01 7.10794091e-01
7.64907718e-01 -6.31096244e-01 7.77489066e-01 4.72126245e-01
-1.23356774e-01 1.39944591e-02 -5.83432794e-01 4.44545239e-01
-2.19132495e+00 -1.01232648e+00 7.89559364e-01 2.60172462e+00
1.18325329e+00 4.58703220e-01 5.57645679e-01 -2.15772033e-01
5.19615412e-01 2.35138312e-01 -1.22669363e+00 -4.19793248e-01
3.15453827e-01 -7.98819810e-02 7.87664592e-01 5.23556709e-01
-7.35509932e-01 9.32053089e-01 6.24727774e+00 1.21958113e+00
-1.18665266e+00 -1.04259126e-01 8.65701616e-01 -3.50007474e-01
-3.74930710e-01 1.69358119e-01 -7.27166831e-01 6.55119956e-01
9.86711442e-01 -7.82328069e-01 9.66053307e-01 1.16930842e+00
7.59599447e-01 -1.11184902e-01 -8.21206689e-01 5.69375098e-01
-6.39315903e-01 -1.12340283e+00 -1.94096804e-01 2.13811085e-01
6.68504834e-01 -3.13429207e-01 3.25711906e-01 7.17520058e-01
7.12563872e-01 -7.45676816e-01 1.00443184e+00 6.94813311e-01
4.00640130e-01 -1.16625595e+00 2.92201549e-01 7.76004374e-01
-9.89065588e-01 -6.84126675e-01 -3.16198826e-01 -3.73908788e-01
6.08517379e-02 3.68406564e-01 -5.97984016e-01 4.81943101e-01
3.18163693e-01 6.85301125e-01 3.29145230e-02 7.28201568e-01
-1.60409972e-01 7.36888528e-01 -2.17971325e-01 -4.23571289e-01
4.62687254e-01 -5.57165265e-01 7.75014162e-01 5.91323972e-01
2.34213352e-01 5.08623123e-02 6.03159010e-01 9.48122442e-01
2.40421176e-01 5.95170148e-02 -4.47828114e-01 -2.23774120e-01
6.81392014e-01 1.03033233e+00 -4.09884870e-01 -1.73877433e-01
9.31799188e-02 5.54453015e-01 5.49137056e-01 3.32761198e-01
-1.14558113e+00 7.15217441e-02 7.25153089e-01 -2.71046430e-01
3.06575596e-01 -2.25359365e-01 -9.11283046e-02 -1.07026470e+00
5.93406744e-02 -9.67206478e-01 3.79082799e-01 -7.52584264e-02
-1.31491458e+00 1.88568592e-01 2.55400032e-01 -1.39920235e+00
-5.17085731e-01 -2.57205725e-01 -6.83014929e-01 5.15633523e-01
-1.61709571e+00 -4.45750654e-01 3.62755984e-01 5.35647511e-01
4.76471961e-01 -7.63695389e-02 3.51663679e-01 -1.87108278e-01
-8.26306105e-01 6.09608710e-01 3.56694728e-01 -5.05412638e-01
4.12506461e-01 -1.30994475e+00 -2.00081065e-01 5.91792881e-01
-4.85478878e-01 4.40677077e-01 1.01205063e+00 -5.81849158e-01
-1.59113050e+00 -1.08023393e+00 -1.74135983e-01 -8.33067764e-03
1.00985599e+00 5.00415117e-02 -6.94739997e-01 3.38781297e-01
-8.38393271e-02 -9.91177633e-02 1.16546012e-01 -2.62078643e-01
3.20219137e-02 -2.04694986e-01 -1.32445049e+00 9.43932712e-01
1.00500548e+00 -1.18417598e-01 -4.20977741e-01 1.90057233e-02
1.06716323e+00 -5.64374804e-01 -8.98925304e-01 4.45088804e-01
5.11383057e-01 -7.24741757e-01 7.19201624e-01 -8.35321605e-01
3.48667145e-01 -1.55934975e-01 -1.79113701e-01 -1.50152850e+00
-6.32014424e-02 -1.21948767e+00 -9.21753407e-01 8.92704189e-01
1.14251457e-01 -6.22528672e-01 7.66535997e-01 5.31594932e-01
8.80299881e-02 -1.35636544e+00 -1.01367819e+00 -1.25813413e+00
1.53572276e-01 -4.58492935e-01 6.28211260e-01 6.46971643e-01
2.10583791e-01 -2.90358424e-01 -9.09300268e-01 1.45345554e-01
7.95086145e-01 1.87650084e-01 6.10838592e-01 -4.97629434e-01
-8.30769837e-01 -5.28705895e-01 2.34549046e-01 -1.32803476e+00
1.89851955e-01 -3.70611906e-01 7.60308877e-02 -8.74664962e-01
7.74311647e-02 -4.83446091e-01 -6.36198342e-01 1.84747219e-01
-3.16769302e-01 -8.09268236e-01 2.28815645e-01 2.31506750e-01
-7.18107104e-01 1.16714501e+00 1.56938994e+00 3.21938321e-02
-5.59568167e-01 2.63583899e-01 -7.42967308e-01 5.28903246e-01
9.43728447e-01 -4.02642250e-01 -8.62913430e-01 1.34099424e-01
1.42846420e-01 5.47568023e-01 7.00026378e-02 -6.18986130e-01
-1.10158741e-01 -1.07271445e+00 -3.60012293e-01 -4.89255309e-01
8.77090394e-02 -6.25422180e-01 -1.37766659e-01 8.53676677e-01
-6.37069941e-01 -1.61810443e-01 -7.70919994e-02 1.23593485e+00
1.30312845e-01 -1.56108573e-01 7.31090903e-01 -1.35672152e-01
-5.92574954e-01 6.46936476e-01 -4.78066981e-01 2.47306764e-01
1.11353087e+00 1.92296296e-01 -5.19404002e-02 -5.20150006e-01
-5.29153526e-01 5.90674102e-01 1.79298565e-01 2.49240085e-01
4.63327408e-01 -1.15853131e+00 -2.90707886e-01 -2.42974296e-01
-1.84361041e-01 -3.93492848e-01 1.89518169e-01 7.62294054e-01
1.89442337e-01 1.76057875e-01 -1.21648848e-01 -2.89771348e-01
-4.64755625e-01 7.33386219e-01 4.75157797e-01 -8.20244431e-01
-4.09587473e-01 4.27831352e-01 -1.74266368e-01 -1.28203675e-01
4.15132791e-01 -5.74345946e-01 -9.14771184e-02 -7.74569660e-02
5.00945866e-01 5.29815078e-01 -3.45055372e-01 1.31903172e-01
-2.34138109e-02 2.63854116e-02 -1.21139057e-01 -4.54478413e-01
9.89969075e-01 -4.17123467e-01 4.07489538e-01 4.83438522e-01
6.64092362e-01 -2.67858982e-01 -2.01435733e+00 -3.89304817e-01
1.81437507e-01 -6.15169287e-01 1.53016269e-01 -8.15708518e-01
-8.06126952e-01 3.10466528e-01 6.93615913e-01 4.36807722e-01
1.06642389e+00 -5.13247490e-01 7.04685092e-01 4.68519539e-01
8.02796721e-01 -1.40169692e+00 1.91866085e-01 5.73001862e-01
7.11609542e-01 -1.19178367e+00 -1.14261374e-01 1.63540393e-01
-1.02572107e+00 8.57139468e-01 8.61447453e-01 -3.96934390e-01
6.44915581e-01 1.40073344e-01 -3.94936144e-01 2.39642009e-01
-1.16042960e+00 -2.26626605e-01 -1.66701242e-01 4.62054759e-01
-1.98215082e-01 3.17088246e-01 -7.52658904e-01 9.85605061e-01
3.68821099e-02 2.28453472e-01 5.29046714e-01 1.21407878e+00
-6.94894016e-01 -1.02701068e+00 -3.73119295e-01 4.26471680e-01
-4.05042380e-01 2.33745411e-01 5.88434376e-02 5.82421362e-01
-2.33015269e-01 7.49282718e-01 -1.30401149e-01 -4.25992578e-01
1.06362000e-01 -2.67198831e-01 4.43910331e-01 -1.73560143e-01
-3.89747679e-01 1.36715397e-01 -5.06210029e-02 -8.27438116e-01
-2.02920467e-01 -5.20675659e-01 -1.16821146e+00 -2.61513382e-01
-1.51514560e-01 1.98830247e-01 3.39663595e-01 1.05546796e+00
3.69335502e-01 3.57129097e-01 1.24406850e+00 -5.56587219e-01
-1.74254894e+00 -6.96834862e-01 -6.53643847e-01 3.38718668e-02
2.94927239e-01 -1.13245559e+00 -4.25119400e-01 -4.38370079e-01]
|
[4.255938529968262, 2.439300298690796]
|
63546088-d202-42e9-95b2-3759b0f9a1bc
|
use-of-speech-impairment-severity-for
|
2305.10659
| null |
https://arxiv.org/abs/2305.10659v1
|
https://arxiv.org/pdf/2305.10659v1.pdf
|
Use of Speech Impairment Severity for Dysarthric Speech Recognition
|
A key challenge in dysarthric speech recognition is the speaker-level diversity attributed to both speaker-identity associated factors such as gender, and speech impairment severity. Most prior researches on addressing this issue focused on using speaker-identity only. To this end, this paper proposes a novel set of techniques to use both severity and speaker-identity in dysarthric speech recognition: a) multitask training incorporating severity prediction error; b) speaker-severity aware auxiliary feature adaptation; and c) structured LHUC transforms separately conditioned on speaker-identity and severity. Experiments conducted on UASpeech suggest incorporating additional speech impairment severity into state-of-the-art hybrid DNN, E2E Conformer and pre-trained Wav2vec 2.0 ASR systems produced statistically significant WER reductions up to 4.78% (14.03% relative). Using the best system the lowest published WER of 17.82% (51.25% on very low intelligibility) was obtained on UASpeech.
|
['Xunying Liu', 'Xurong Xie', 'Jianwei Yu', 'Guinan Li', 'Mingyu Cui', 'Jiajun Deng', 'Shujie Hu', 'Tianzi Wang', 'Zengrui Jin', 'Mengzhe Geng']
|
2023-05-18
| null | null | null | null |
['severity-prediction']
|
['computer-vision']
|
[ 1.93772446e-02 1.30553365e-01 1.35336831e-01 -4.58399147e-01
-1.28095698e+00 -3.46859545e-01 4.66545999e-01 -4.44131464e-01
-5.13532221e-01 5.52468359e-01 1.02576661e+00 -1.90502480e-01
-4.52522226e-02 -1.22372307e-01 -9.17156935e-02 -6.36904359e-01
2.06468552e-01 3.74709487e-01 -1.00803494e-01 -4.91730094e-01
4.04160209e-02 4.26694244e-01 -1.90349603e+00 3.02997053e-01
7.91961968e-01 7.07527816e-01 4.50222492e-01 1.00336540e+00
-1.26308262e-01 6.00753844e-01 -1.15170145e+00 1.56903602e-02
1.35678068e-01 -4.77250785e-01 -6.71320319e-01 -1.92223787e-01
7.51422703e-01 -3.39212000e-01 -6.07322633e-01 1.06920767e+00
1.17142379e+00 2.31417418e-01 4.61026222e-01 -5.44721723e-01
-9.05972242e-01 8.66416216e-01 -5.39775863e-02 7.62338459e-01
2.92170823e-01 1.05113633e-01 8.53717268e-01 -8.90906215e-01
2.72947758e-01 1.35188997e+00 3.68265301e-01 1.22048032e+00
-9.99150872e-01 -8.83541882e-01 -2.17205867e-01 7.71044552e-01
-1.31502080e+00 -1.10352099e+00 7.73588598e-01 -4.43138897e-01
1.30699825e+00 5.24452984e-01 3.03701609e-01 1.27886903e+00
-3.99268627e-01 4.05905604e-01 1.31104481e+00 -5.97742975e-01
-2.13469821e-03 -3.43052112e-02 4.37200367e-01 2.73662299e-01
-3.28525364e-01 2.08008856e-01 -9.04981673e-01 -3.76700647e-02
4.42099899e-01 -8.54259670e-01 -5.37269056e-01 6.39842510e-01
-9.31485713e-01 6.31098747e-01 -3.18319768e-01 6.09830320e-01
-2.81684548e-01 -1.01761006e-01 6.19788051e-01 8.12975526e-01
5.80889881e-01 3.38103652e-01 -6.76316619e-01 -5.76480746e-01
-9.42995906e-01 1.67351663e-01 6.01020336e-01 6.86249912e-01
1.91180203e-02 9.25010204e-01 -4.58475292e-01 1.78817153e+00
1.97896346e-01 6.01145327e-01 1.08377099e+00 -7.35830307e-01
5.17702639e-01 -2.32132655e-02 -4.72736567e-01 -9.12050977e-02
-2.62557387e-01 -4.97802138e-01 -2.85870820e-01 2.65218437e-01
3.18900585e-01 -2.00255141e-01 -1.42090821e+00 2.06396079e+00
-5.79012521e-02 -3.31884697e-02 2.43088394e-01 8.71765852e-01
1.03955388e+00 7.21574426e-01 2.42162049e-01 -2.65304893e-01
1.29900146e+00 -9.00420845e-01 -1.11779666e+00 -2.13618398e-01
5.59578955e-01 -1.00029039e+00 1.27294493e+00 4.88151878e-01
-1.29544652e+00 -6.44286156e-01 -1.03394234e+00 -4.51930426e-02
-2.14771807e-01 4.25087005e-01 -1.83907896e-01 1.44787288e+00
-1.47136295e+00 3.25370610e-01 -4.63563263e-01 -2.40829229e-01
5.80971278e-02 5.12712300e-01 -3.55653077e-01 1.87426016e-01
-1.22992694e+00 1.28500295e+00 -1.21658444e-02 -3.69797349e-01
-7.97522962e-01 -9.52828169e-01 -7.02864110e-01 5.59839197e-02
-8.45750868e-02 -2.77034044e-01 1.31595445e+00 -6.81759000e-01
-2.01755691e+00 9.70121741e-01 -4.39200640e-01 -4.39121127e-01
3.69569212e-01 -4.90856260e-01 -1.04494202e+00 -9.37443972e-02
-3.60396266e-01 5.75465523e-02 8.35875869e-01 -1.03053141e+00
-5.79803109e-01 -8.45963478e-01 -5.13580859e-01 5.82804024e-01
-6.06008291e-01 6.33193970e-01 -7.89265633e-02 -8.15076888e-01
-1.73230618e-02 -7.61175096e-01 4.46052134e-01 -5.15273511e-01
-1.67312548e-01 -5.94634831e-01 1.06828809e+00 -1.75826383e+00
1.38790071e+00 -2.02656722e+00 3.51099759e-01 -7.52738118e-02
2.82046665e-02 7.55261302e-01 -2.04299957e-01 3.23514156e-02
-2.06633925e-01 6.79285452e-02 -7.31823817e-02 -5.35385370e-01
6.28565699e-02 9.57937166e-02 6.65058522e-03 4.41243976e-01
-9.78992432e-02 3.01766545e-01 -3.96735430e-01 -2.74115764e-02
4.65924442e-01 7.52049863e-01 -4.83085483e-01 4.26788658e-01
4.64213490e-01 2.79454947e-01 2.60362625e-01 5.25128782e-01
5.54019868e-01 8.95244122e-01 -1.31823331e-01 -7.55971521e-02
-3.30715448e-01 6.02432668e-01 -9.22510445e-01 1.50612140e+00
-8.47817004e-01 7.62665212e-01 5.37454247e-01 -6.84913099e-01
9.87026930e-01 6.58238947e-01 5.77102862e-02 -6.00264072e-01
2.22503290e-01 4.86252993e-01 6.27022147e-01 -5.44373333e-01
3.03661078e-01 -3.44390929e-01 2.17715889e-01 -1.11500233e-01
6.35563552e-01 -2.53234655e-01 -2.13557079e-01 -4.23497856e-01
1.29896498e+00 -5.18955827e-01 1.08718015e-01 -5.59590340e-01
9.88955975e-01 -7.20197856e-01 3.92129213e-01 5.45145333e-01
-5.98327100e-01 1.02278566e+00 -4.45511900e-02 2.41488010e-01
-1.16443288e+00 -1.32847786e+00 -2.36447439e-01 1.30908406e+00
-6.83337986e-01 -1.58935994e-01 -8.95644307e-01 -2.10561365e-01
-1.91952139e-01 1.23173618e+00 -5.25223076e-01 -2.90856421e-01
-8.22933614e-01 -3.58908743e-01 9.80419636e-01 5.17049789e-01
1.93635359e-01 -9.60026264e-01 2.65112430e-01 7.66012520e-02
-4.23791781e-02 -1.00999403e+00 -6.14741921e-01 -5.48855169e-03
-3.67841393e-01 -5.26505053e-01 -1.21420598e+00 -9.24380481e-01
-1.08471222e-01 -2.31479645e-01 5.72333336e-01 -3.80979747e-01
-1.32163912e-01 4.58340526e-01 -4.58351016e-01 -3.35721642e-01
-8.91922355e-01 1.67015698e-02 6.80081785e-01 -2.10544571e-01
4.33103979e-01 -6.58834398e-01 -2.46433124e-01 7.15806186e-02
-3.17647845e-01 -5.58060288e-01 4.94910687e-01 9.19150293e-01
1.94844797e-01 -3.47774237e-01 9.89499748e-01 -4.52457547e-01
9.99681115e-01 -3.00477535e-01 -4.21024393e-03 1.22369997e-01
-5.57732642e-01 -1.83504581e-01 5.37879407e-01 -4.61962521e-01
-1.50909543e+00 -2.56350249e-01 -8.29681396e-01 -5.44167280e-01
-5.60825825e-01 -5.91668859e-03 -4.26034242e-01 8.57428089e-02
6.92073524e-01 5.40543020e-01 2.27531970e-01 -7.27887213e-01
2.43496820e-01 1.56913066e+00 8.01516175e-01 -4.16129917e-01
4.84268695e-01 -1.57683060e-01 -7.27944076e-01 -1.71369612e+00
-3.60016227e-01 -6.57866120e-01 -3.57824385e-01 -1.80373102e-01
9.02861655e-01 -9.59604979e-01 -2.84517348e-01 8.35591912e-01
-1.12257004e+00 -2.72101223e-01 -1.81492656e-01 7.96190381e-01
-5.73538184e-01 2.90033370e-01 -5.49355805e-01 -9.66452420e-01
-7.82014132e-01 -1.28139305e+00 6.77512467e-01 -1.27219290e-01
-5.17533958e-01 -5.89649677e-01 1.62096754e-01 9.15378928e-01
7.98458815e-01 -4.43170309e-01 9.38257575e-01 -1.02667856e+00
3.44359607e-01 6.22116495e-03 -6.06301473e-03 1.14418757e+00
4.63704407e-01 -4.39433396e-01 -1.33727825e+00 -2.17611507e-01
3.36502731e-01 -1.44999608e-01 5.02385676e-01 4.27385807e-01
1.07772505e+00 -1.87148422e-01 4.71954703e-01 4.30308640e-01
7.83738315e-01 4.98372197e-01 6.86854720e-01 -1.35831945e-02
6.80069029e-01 6.67817593e-01 -2.95566232e-03 2.60014594e-01
2.17703477e-01 9.31641638e-01 -2.76875496e-01 3.42054486e-01
-1.16817915e+00 1.17200881e-01 7.14938521e-01 1.49873340e+00
-3.71920168e-01 -1.26456216e-01 -9.72912550e-01 8.00150573e-01
-1.03242660e+00 -9.84941125e-01 -1.64601147e-01 2.20310593e+00
9.47972834e-01 -2.21366137e-01 1.46963701e-01 4.13429976e-01
1.03649962e+00 3.34346354e-01 -3.82819802e-01 -8.47164631e-01
-4.81010437e-01 6.04059815e-01 4.44400817e-01 6.35768592e-01
-6.65950596e-01 1.05768728e+00 5.88706636e+00 1.10463762e+00
-1.14407492e+00 5.03088355e-01 1.95297673e-01 -5.33068240e-01
-1.33676395e-01 -7.50155568e-01 -8.01027775e-01 4.87815499e-01
1.41952705e+00 -3.57018709e-01 6.61096632e-01 7.39706218e-01
2.80156344e-01 5.45865715e-01 -7.60934174e-01 1.28916800e+00
6.48687005e-01 -7.13258445e-01 7.23243877e-02 3.71975042e-02
4.53820378e-01 4.30238754e-01 1.71490714e-01 4.70332682e-01
7.00420747e-03 -1.10617507e+00 8.60525966e-01 1.20784923e-01
1.09005404e+00 -8.69764984e-01 6.28585398e-01 -5.24367690e-02
-7.14233279e-01 -1.66964680e-01 -1.91290602e-01 1.10101148e-01
1.65633529e-01 4.61650006e-02 -8.60258400e-01 1.23369209e-01
6.92441940e-01 1.14654265e-01 -2.04357654e-01 7.20182776e-01
-3.05305809e-01 1.18865740e+00 -3.12147379e-01 1.39550469e-03
-6.53739944e-02 3.33294153e-01 1.03952348e+00 1.38572907e+00
7.06853271e-01 1.53874576e-01 -5.62322140e-01 3.46239269e-01
5.71190268e-02 4.28629756e-01 -1.91171482e-01 9.99183655e-02
6.70115829e-01 6.21276140e-01 1.71864659e-01 -1.29214525e-01
-6.54381871e-01 1.03858733e+00 3.45163494e-01 2.42508635e-01
-3.67028594e-01 -3.63205552e-01 1.26498532e+00 1.18652306e-01
1.16244920e-01 -2.57648408e-01 -5.01750469e-01 -8.51783574e-01
1.19563676e-02 -1.00785565e+00 1.78292662e-01 -4.72765326e-01
-1.21962869e+00 9.12252486e-01 -2.73181915e-01 -7.22119749e-01
-2.91034102e-01 -8.42195988e-01 -6.95603371e-01 1.33606339e+00
-1.19908655e+00 -1.06508839e+00 3.03749349e-02 6.55085742e-01
1.24150288e+00 -8.65860403e-01 1.03081679e+00 5.83260059e-01
-5.67725360e-01 1.05527282e+00 1.05206497e-01 -1.30170345e-01
7.32405663e-01 -1.39039063e+00 2.82201141e-01 8.20325017e-01
-1.50580779e-01 3.35718036e-01 8.54228854e-01 -5.21914244e-01
-1.07688046e+00 -8.27409387e-01 1.05683672e+00 -3.46613526e-01
7.00073659e-01 -3.38542074e-01 -1.02085388e+00 3.67761463e-01
2.86949545e-01 -4.26379979e-01 8.14948082e-01 2.55076945e-01
-3.75269622e-01 -4.55178693e-02 -1.26874387e+00 7.60230064e-01
1.33283114e+00 -7.52406418e-01 -1.02341425e+00 4.57518771e-02
6.69793546e-01 -3.41222912e-01 -9.51046526e-01 3.87793183e-01
4.68854666e-01 -6.99665964e-01 9.44473207e-01 -3.97181213e-01
1.29449904e-01 7.47028813e-02 -6.04720771e-01 -1.71513116e+00
-3.14442664e-01 -5.88159680e-01 -1.97715089e-01 1.49352646e+00
4.47343171e-01 -6.53160334e-01 5.33098042e-01 5.88069022e-01
-8.56179833e-01 -4.40263987e-01 -1.41232097e+00 -1.05805326e+00
4.26294237e-01 -5.49370229e-01 3.40574682e-01 8.77157807e-01
7.99802481e-04 9.86194313e-02 -4.06687438e-01 3.30218732e-01
4.38910246e-01 -8.66605341e-01 3.29328001e-01 -1.02961707e+00
-2.93251239e-02 -7.26041198e-01 -6.58506989e-01 -4.41731334e-01
4.98298675e-01 -8.63329709e-01 2.41944622e-02 -1.30085897e+00
-2.92084545e-01 4.25268598e-02 -3.69794816e-01 3.64058942e-01
-2.09087849e-01 -1.40071241e-02 1.53103188e-01 -1.75933391e-01
3.43272924e-01 8.19840431e-01 7.79701412e-01 -1.79021284e-02
-4.71090138e-01 1.08099364e-01 -5.09679496e-01 6.70485795e-01
1.30194139e+00 -2.83931226e-01 -4.88638759e-01 -4.81383830e-01
-7.97080278e-01 1.44524872e-01 -1.07364096e-01 -1.27898395e+00
9.93069559e-02 2.70983577e-01 2.95801628e-02 -3.44165027e-01
8.33211601e-01 -1.23743370e-01 -3.14820290e-01 2.90129215e-01
-3.84470284e-01 -2.55345374e-01 3.18705410e-01 1.07171521e-01
-2.63486713e-01 -2.64053404e-01 1.01960790e+00 1.66208521e-02
-7.92373717e-01 -4.28762510e-02 -7.63813317e-01 2.34321013e-01
6.11556292e-01 -1.06382601e-01 -3.63564759e-01 -4.83099699e-01
-9.07261968e-01 -2.81113774e-01 -1.16951577e-01 9.56750870e-01
6.12558603e-01 -1.17008412e+00 -1.16638970e+00 2.30973646e-01
4.31386717e-02 -8.26305628e-01 7.28531182e-01 3.48347694e-01
-1.90254614e-01 4.14240688e-01 -2.61522442e-01 -8.42559412e-02
-1.92910957e+00 -1.86110780e-01 3.27108324e-01 2.59458840e-01
-4.16396141e-01 1.28247249e+00 -9.27631930e-02 -6.44448876e-01
3.67091179e-01 6.56997934e-02 -4.51487482e-01 1.76616088e-01
6.38875663e-01 1.05010331e+00 4.91067439e-01 -1.20014024e+00
-4.89389181e-01 3.70859385e-01 -3.02102178e-01 -6.61158741e-01
1.22843885e+00 -2.50589609e-01 2.08313316e-01 4.86697584e-01
1.28480077e+00 3.33002687e-01 -6.34587109e-01 -1.63641706e-01
-2.55065262e-01 -2.78806746e-01 6.01480961e-01 -1.28773820e+00
-9.23069656e-01 8.76157582e-01 1.05975056e+00 -1.57107844e-03
1.07882702e+00 1.75196499e-01 9.07281637e-01 1.61563843e-01
3.12683880e-02 -1.51923859e+00 -4.05525684e-01 7.53012002e-01
1.42720139e+00 -7.46235073e-01 -4.73895788e-01 -1.86595187e-01
-6.46944046e-01 9.48023796e-01 7.37156391e-01 -1.91592854e-02
7.42209256e-01 1.36701927e-01 6.14202440e-01 2.43247986e-01
-4.70120758e-01 -3.63047689e-01 4.95039225e-01 8.84256303e-01
7.63907552e-01 2.40353286e-01 -6.83869660e-01 1.01170790e+00
-1.04365885e+00 -4.92050380e-01 4.86039907e-01 3.18047225e-01
-4.52671170e-01 -1.08579326e+00 -4.28489357e-01 6.67770267e-01
-9.22354043e-01 -2.55532056e-01 -3.71779591e-01 3.81744981e-01
8.61565396e-02 1.16341293e+00 8.73312503e-02 -6.13609135e-01
7.42365003e-01 5.25425255e-01 4.54280525e-01 -8.25698793e-01
-7.71134079e-01 3.04552875e-02 6.30163074e-01 -3.28401655e-01
5.43118902e-02 -9.44017053e-01 -1.03552401e+00 3.50074992e-02
-1.40676469e-01 -1.02338560e-01 1.08566034e+00 9.61935222e-01
1.58552364e-01 8.70707810e-01 3.14471871e-01 -5.90536237e-01
-7.72930622e-01 -1.65744281e+00 -7.39531159e-01 2.68996626e-01
5.53594470e-01 -5.23930848e-01 -8.69484782e-01 -1.98538482e-01]
|
[14.53078842163086, 6.429051399230957]
|
8fd396e1-2b60-4b7f-927a-f6c7c8b5d1a1
|
tell-me-something-new-a-new-framework-for
|
1805.07483
| null |
http://arxiv.org/abs/1805.07483v2
|
http://arxiv.org/pdf/1805.07483v2.pdf
|
Tell Me Something New: A New Framework for Asynchronous Parallel Learning
|
We present a novel approach for parallel computation in the context of
machine learning that we call "Tell Me Something New" (TMSN). This approach
involves a set of independent workers that use broadcast to update each other
when they observe "something new". TMSN does not require synchronization or a
head node and is highly resilient against failing machines or laggards. We
demonstrate the utility of TMSN by applying it to learning boosted trees. We
show that our implementation is 10 times faster than XGBoost and LightGBM on
the splice-site prediction problem.
|
['Yoav Freund', 'Julaiti Alafate']
|
2018-05-19
| null | null | null | null |
['splice-site-prediction']
|
['medical']
|
[ 1.66807100e-01 1.53663278e-01 -3.08404654e-01 -7.62908518e-01
-1.14868212e+00 -2.27520093e-01 5.00089526e-01 4.65653419e-01
-6.60504878e-01 9.63434696e-01 -7.62379393e-02 -5.72412729e-01
-4.28719930e-02 -7.89984703e-01 -1.09143591e+00 -9.09137845e-01
-2.06970379e-01 1.11477184e+00 5.53749979e-01 -1.85804844e-01
2.42631957e-01 1.00323834e-01 -1.30651748e+00 6.41470492e-01
-2.35738270e-02 4.79267418e-01 -1.51005080e-02 9.69566524e-01
-3.64665836e-02 9.43919122e-01 -8.18681777e-01 -5.46199501e-01
3.22623044e-01 -3.38581830e-01 -1.58031249e+00 -4.47363973e-01
2.37985730e-01 6.06149435e-02 3.30077171e-01 6.33673429e-01
6.31046653e-01 1.18388332e-01 8.54775459e-02 -1.19893324e+00
1.36371225e-01 1.20124376e+00 -9.65778708e-01 4.61037874e-01
1.75880298e-01 2.98491754e-02 1.15622365e+00 -4.17128116e-01
9.26369011e-01 1.27432585e+00 1.07776701e+00 4.37074989e-01
-1.77891314e+00 -3.33704561e-01 5.52192032e-02 2.67854512e-01
-9.18558836e-01 -3.93997908e-01 2.26729363e-01 -8.77688155e-02
1.31697655e+00 6.83823824e-01 5.31318724e-01 5.99553704e-01
3.40194315e-01 9.01673436e-01 1.32363713e+00 -8.99872720e-01
5.57154417e-01 -2.90002376e-01 6.39166951e-01 8.09615493e-01
9.57489107e-03 9.81981680e-02 -1.25108457e+00 -8.55816722e-01
8.14798921e-02 -1.32857794e-02 1.80920646e-01 -4.91474904e-02
-1.18113577e+00 8.06122124e-01 2.28508160e-01 4.15762812e-02
-3.39541435e-01 6.52830899e-01 6.69761717e-01 6.89242780e-01
6.25180304e-01 7.92061687e-02 -7.96122313e-01 -2.46962354e-01
-7.46218979e-01 1.63996313e-02 1.03664303e+00 6.40513122e-01
8.93776715e-01 -3.66091013e-01 2.18353406e-01 6.10416234e-01
-7.89542720e-02 1.03853494e-01 6.12113118e-01 -9.30005848e-01
-8.55207965e-02 8.22175369e-02 -7.36938640e-02 -8.93700942e-02
-7.89265811e-01 -4.03086305e-01 -7.79413104e-01 3.24487478e-01
2.09748775e-01 -6.69951364e-02 -7.90848970e-01 1.63557625e+00
8.65524948e-01 5.44763766e-02 -1.55355573e-01 6.95521235e-01
3.80857080e-01 5.16475081e-01 1.68434814e-01 -2.66932040e-01
1.21608639e+00 -9.87792313e-01 -3.00342679e-01 -1.94149688e-01
1.00226951e+00 -6.74372494e-01 6.94131315e-01 6.37147069e-01
-1.21394348e+00 -3.06474745e-01 -7.98010945e-01 -1.59708425e-01
-9.65061188e-02 -5.01509666e-01 1.32431126e+00 7.65920937e-01
-1.32542026e+00 7.52486348e-01 -1.35007179e+00 -2.39017010e-01
1.81145698e-01 4.97589439e-01 -3.47044885e-01 2.52687782e-01
-7.00991273e-01 9.62714314e-01 3.69377464e-01 -1.14081226e-01
-9.55892205e-01 -6.16690874e-01 1.81745104e-02 -1.74002890e-02
1.20288759e-01 -9.80506182e-01 1.81689060e+00 -9.19184506e-01
-1.54314828e+00 1.01363182e+00 -5.40160477e-01 -7.43474305e-01
4.87102032e-01 -2.37726167e-01 1.55037731e-01 -3.86826009e-01
-4.09139246e-02 3.05848271e-01 5.51726878e-01 -7.49623120e-01
-1.18588865e+00 -7.47157097e-01 -3.30006897e-01 -1.12426892e-01
-3.05739753e-02 3.01284015e-01 -1.56391725e-01 -3.27025801e-01
4.69950289e-01 -1.06357145e+00 -3.93939316e-01 -4.56284910e-01
-6.58891857e-01 -5.27404010e-01 3.65559459e-01 -3.09708953e-01
8.54390442e-01 -1.88951743e+00 2.16887340e-01 4.45977092e-01
2.81760335e-01 -1.87745243e-01 -4.75447588e-02 7.60613203e-01
1.34246662e-01 -1.07755242e-02 2.79593952e-02 -2.04359546e-01
-1.67749405e-01 5.80147743e-01 -2.87438333e-01 5.13766110e-01
-3.68226349e-01 5.69474459e-01 -8.57391417e-01 -3.19632381e-01
-4.34802294e-01 -1.29800513e-01 -4.93805766e-01 -7.10891038e-02
-4.77549553e-01 1.85295045e-01 -2.71300524e-01 5.00480056e-01
5.82752824e-01 -5.68517029e-01 8.92852008e-01 5.14199913e-01
5.17841242e-02 5.94639599e-01 -1.01691496e+00 1.63341761e+00
-1.93249688e-01 3.54046404e-01 1.51456073e-01 -9.89098191e-01
8.50444853e-01 2.39166752e-01 9.05970186e-02 -1.52680308e-01
-2.96334416e-01 1.42539993e-01 -1.44755349e-01 -3.88854891e-02
1.98597312e-01 -3.69079649e-01 7.52911046e-02 9.93300080e-01
2.22799685e-02 1.41643792e-01 -2.24679634e-02 4.11593139e-01
1.84508801e+00 -7.29689896e-02 5.27563512e-01 -3.65928233e-01
2.46753842e-02 2.03062788e-01 1.00424480e+00 1.25467885e+00
1.21503912e-01 -6.12867512e-02 6.11328721e-01 -1.21167505e+00
-8.81099224e-01 -1.08901167e+00 8.44790637e-02 2.08340073e+00
-4.10156846e-01 -8.26616466e-01 -5.07911026e-01 -7.12800264e-01
2.08281949e-01 4.34214056e-01 -4.45770383e-01 2.60150969e-01
-5.74157059e-01 -1.15763450e+00 4.08090651e-01 4.99807864e-01
2.05909446e-01 -8.08956981e-01 -6.43159926e-01 3.67658168e-01
3.28464434e-02 -2.27706015e-01 -3.49880941e-02 8.84565175e-01
-1.48137796e+00 -8.06984901e-01 -3.51377800e-02 -5.64381242e-01
4.63614106e-01 2.01871663e-01 1.39428926e+00 3.91894102e-01
-2.22318769e-01 -2.03444883e-01 -1.81154013e-01 -6.38867855e-01
-7.02947497e-01 5.64430416e-01 -6.26024753e-02 -5.12692571e-01
7.27765024e-01 -9.02217627e-01 -3.75558466e-01 3.19668204e-01
-5.89137077e-01 3.32921267e-01 5.05979300e-01 1.23579443e+00
6.34603381e-01 -1.05897397e-01 3.21858644e-01 -1.79200709e+00
4.12390560e-01 -4.57740724e-01 -8.55871677e-01 3.00134540e-01
-8.77177000e-01 2.90830344e-01 4.45681632e-01 -2.15330243e-01
-9.72240627e-01 4.12886232e-01 -1.29521593e-01 3.42283577e-01
2.79377438e-02 4.46989536e-01 4.01581854e-01 2.17625741e-02
9.97786164e-01 6.44488409e-02 1.34423412e-02 -1.00614798e+00
3.05073857e-01 5.13501108e-01 4.74339545e-01 -7.94323504e-01
4.73742664e-01 6.77669108e-01 3.91170174e-01 -4.37533259e-01
-5.87710083e-01 -4.30624932e-01 -3.38662028e-01 8.12565908e-02
1.77548259e-01 -6.72948182e-01 -1.30970311e+00 4.78386611e-01
-1.14741969e+00 -4.60186690e-01 -3.46223176e-01 2.34293163e-01
-5.44481933e-01 2.44080767e-01 -1.23145497e+00 -6.15261376e-01
-7.38955021e-01 -6.26471817e-01 8.65839064e-01 -7.66005367e-02
-3.18522751e-01 -6.26770675e-01 6.69947445e-01 6.06736004e-01
3.43237400e-01 -1.95462435e-01 1.08402073e+00 -9.27589893e-01
-6.04515374e-01 -1.26514718e-01 3.39971989e-01 1.84694454e-02
-4.44722742e-01 -2.01815501e-01 -1.03514779e+00 -5.68761826e-01
-6.90311268e-02 -5.03740966e-01 1.11960435e+00 1.41160965e-01
1.09376156e+00 -4.50384349e-01 -6.36172056e-01 5.61881065e-01
1.33768225e+00 -1.75516322e-01 4.60381925e-01 6.62344694e-01
1.61668792e-01 4.04392898e-01 3.26421559e-01 6.59511328e-01
2.84340709e-01 5.11644304e-01 9.08737108e-02 -2.96447963e-01
1.61981076e-01 -1.88893706e-01 2.19216496e-01 8.27502012e-01
-4.56270605e-01 5.05666658e-02 -1.14315975e+00 1.92334235e-01
-2.17277932e+00 -8.43154788e-01 -3.96086931e-01 1.91445839e+00
1.24308479e+00 1.87405095e-01 1.44889414e-01 6.33766130e-02
7.07857728e-01 -2.35735670e-01 -4.60608840e-01 -8.30472112e-01
-1.41787082e-02 7.29850054e-01 6.29794478e-01 4.98759091e-01
-7.47652769e-01 1.09095371e+00 7.72719049e+00 8.75588715e-01
-7.10985303e-01 7.88533390e-01 8.69175971e-01 -3.05004597e-01
-9.60616246e-02 5.93206346e-01 -9.14789677e-01 2.59616017e-01
1.14337528e+00 1.55442446e-01 3.94450098e-01 1.02773595e+00
-1.68643326e-01 -5.15383005e-01 -1.24536598e+00 4.55886692e-01
-4.64460969e-01 -1.63673389e+00 -3.85371447e-01 -2.29666643e-02
8.04804742e-01 6.67464018e-01 -3.82773906e-01 1.36406405e-03
1.37816739e+00 -6.26927674e-01 4.47814167e-01 5.02741933e-01
1.77311182e-01 -6.13217056e-01 6.99611783e-01 5.82307339e-01
-4.98043299e-01 8.48421678e-02 -6.76874578e-01 -4.42115724e-01
-2.98425592e-02 9.65699792e-01 -1.23798025e+00 3.92640740e-01
9.86978769e-01 -1.46149427e-01 -5.06485701e-01 9.09235179e-01
-4.08915341e-01 1.03972197e+00 -7.70398796e-01 -1.95935175e-01
-1.48896143e-01 1.90463945e-01 3.25344503e-01 1.23265481e+00
-2.58559398e-02 -6.53734878e-02 2.32710809e-01 2.01820925e-01
-2.50149876e-01 6.42289873e-03 -1.90154258e-02 7.50414729e-01
5.79498529e-01 1.12297916e+00 -6.75314367e-01 -6.30344748e-01
-1.53283224e-01 8.15671325e-01 7.51238942e-01 2.05216259e-02
-1.18998148e-01 -1.03969023e-01 3.39953840e-01 -1.75368384e-01
2.42409319e-01 2.96723127e-01 -2.21841499e-01 -8.29734325e-01
-8.27664509e-02 -1.11767173e+00 1.12914538e+00 -6.35085583e-01
-1.58847380e+00 4.41454321e-01 -4.21436638e-01 -3.95260990e-01
-3.32845449e-01 -3.40797544e-01 -5.98470330e-01 5.82145274e-01
-8.45932066e-01 -1.04706252e+00 1.78448126e-01 4.40580130e-01
1.81706548e-01 -6.22998923e-02 1.06838071e+00 -3.57794940e-01
-4.71798956e-01 5.23767948e-01 4.76133287e-01 -3.85687828e-01
7.17062593e-01 -1.38976669e+00 8.77028346e-01 6.55173004e-01
3.44637573e-01 6.56575561e-01 8.36568177e-01 -5.70729077e-01
-1.60083640e+00 -6.43834591e-01 1.16889894e+00 -4.72306341e-01
3.26563805e-01 -4.64253485e-01 -7.52666593e-01 1.28579247e+00
4.04925235e-02 6.93407655e-02 9.69974637e-01 1.17775643e+00
-5.75869620e-01 -5.47435999e-01 -9.03688908e-01 9.61608365e-02
9.13511217e-01 -3.39233398e-01 -4.90508944e-01 8.69124174e-01
3.88056338e-01 -3.35170835e-01 -9.85134006e-01 1.61644906e-01
5.29017985e-01 -1.14499998e+00 5.88981807e-01 -1.22208321e+00
-4.18511555e-02 -7.91517720e-02 -5.30903274e-03 -1.15450513e+00
-5.60826957e-01 -1.19578564e+00 2.47857627e-02 9.02769029e-01
6.26134932e-01 -9.94322836e-01 1.17627370e+00 4.41708416e-01
1.47693232e-01 -4.04039681e-01 -1.36208105e+00 -7.66687512e-01
1.07963741e-01 -3.89858603e-01 7.92153120e-01 9.55501616e-01
4.72132355e-01 5.72143793e-01 -6.01414800e-01 -2.08796002e-02
5.48443913e-01 2.93746352e-01 1.26549888e+00 -1.29042828e+00
-9.46622789e-01 -4.03943360e-02 -4.73756075e-01 -7.44845331e-01
-1.67354897e-01 -1.18751979e+00 7.06188679e-02 -9.63891566e-01
9.14620697e-01 -6.33034706e-01 -5.63005567e-01 1.09054267e+00
-1.93292663e-01 1.55296504e-01 -1.10504925e-01 5.95311761e-01
-8.28231931e-01 -1.81899235e-01 3.21117282e-01 1.07516795e-01
-2.15179976e-02 3.43933493e-01 -3.77072483e-01 6.50432229e-01
8.05671632e-01 -9.84718323e-01 2.95002937e-01 -5.08788824e-01
7.79833496e-01 1.48347020e-01 1.78788677e-01 -6.67026341e-01
5.75342357e-01 -5.27230427e-02 3.85141432e-01 -5.75499833e-01
1.16469100e-01 -2.51834989e-01 7.15096235e-01 9.08836186e-01
-5.48928499e-01 3.87008131e-01 -7.47586414e-02 6.47731543e-01
1.57062083e-01 -2.92712152e-01 5.58622181e-01 -2.26501822e-01
-2.97632068e-01 -5.92754595e-02 -6.09857917e-01 -2.49525055e-01
7.25381911e-01 4.15223271e-01 -5.62362731e-01 -3.08822244e-01
-6.76658869e-01 2.59022564e-01 3.91304374e-01 -2.88400799e-01
1.57139316e-01 -7.56062567e-01 -8.47966790e-01 1.54362172e-02
-8.46172422e-02 -2.75556117e-01 5.96150868e-02 8.28488052e-01
-6.79854214e-01 2.33416617e-01 -1.09991267e-01 -6.20016515e-01
-1.65460038e+00 5.60929656e-01 6.72973022e-02 -4.84916687e-01
-6.86640918e-01 1.25317168e+00 -2.97934383e-01 -6.17186844e-01
1.81818753e-01 1.27756342e-01 8.75539064e-01 -2.57997632e-01
5.11117876e-01 6.57813430e-01 3.55349571e-01 2.39928216e-01
-3.86224002e-01 -1.10628128e-01 -4.77764368e-01 -4.17652369e-01
1.56816614e+00 2.50074834e-01 -8.34712088e-01 5.76898038e-01
7.62599409e-01 -7.65800774e-02 -6.62964642e-01 -6.99186444e-01
6.56394362e-01 -4.12367284e-01 7.57131949e-02 -9.57839727e-01
-7.52662420e-01 5.97407699e-01 6.47380054e-01 4.61365245e-02
1.02419889e+00 2.32326478e-01 7.76793242e-01 1.04239714e+00
7.41935432e-01 -1.30912530e+00 -4.43109602e-01 2.95073181e-01
2.29135811e-01 -9.15984213e-01 3.39254409e-01 -1.77570105e-01
-1.60437524e-01 1.09323359e+00 2.86181271e-01 3.41897339e-01
4.13495272e-01 6.03767097e-01 1.88080326e-01 -1.77656665e-01
-1.65967095e+00 4.74662222e-02 -5.47820568e-01 4.56309438e-01
3.48118901e-01 4.74023730e-01 -6.83929980e-01 1.99559659e-01
-4.02746469e-01 1.26260728e-01 2.11359397e-01 1.45494950e+00
-6.45998478e-01 -1.59048700e+00 -4.45104122e-01 5.66093862e-01
-7.10311353e-01 -2.95835193e-02 -4.63336349e-01 5.87182164e-01
-2.34874189e-02 6.99213266e-01 2.40664519e-02 -4.93807346e-01
-2.61221737e-01 4.79717702e-01 4.83094662e-01 -5.22624195e-01
-1.14554489e+00 9.23103914e-02 4.69891965e-01 -5.71232200e-01
-2.80506432e-01 -6.93546653e-01 -1.15015566e+00 -9.97123480e-01
-6.00555360e-01 3.62303942e-01 9.20363843e-01 7.85202265e-01
4.64136988e-01 -6.84420159e-03 7.11254954e-01 -1.78257614e-01
-1.03213191e+00 -9.52291489e-01 -5.77537239e-01 6.94035739e-02
-5.16280234e-02 9.14179087e-02 -1.46409959e-01 -7.86484331e-02]
|
[8.39270305633545, 4.300658702850342]
|
68cd3cad-4eff-4da6-b596-1562a53869da
|
a-trio-method-for-retinal-vessel-segmentation
|
2209.11230
| null |
https://arxiv.org/abs/2209.11230v1
|
https://arxiv.org/pdf/2209.11230v1.pdf
|
A Trio-Method for Retinal Vessel Segmentation using Image Processing
|
Inner Retinal neurons are a most essential part of the retina and they are supplied with blood via retinal vessels. This paper primarily focuses on the segmentation of retinal vessels using a triple preprocessing approach. DRIVE database was taken into consideration and preprocessed by Gabor Filtering, Gaussian Blur, and Edge Detection by Sobel and Pruning. Segmentation was driven out by 2 proposed U-Net architectures. Both the architectures were compared in terms of all the standard performance metrics. Preprocessing generated varied interesting results which impacted the results shown by the UNet architectures for segmentation. This real-time deployment can help in the efficient pre-processing of images with better segmentation and detection.
|
['Manoj Sahni', 'Vinayak Singh', 'Mahendra Kumar Gourisaria']
|
2022-09-19
| null | null | null | null |
['edge-detection']
|
['computer-vision']
|
[-1.26522973e-01 2.61056393e-01 4.08461303e-01 -3.16699088e-01
2.81573921e-01 -5.47392011e-01 3.59521598e-01 4.72529233e-02
-8.71986806e-01 6.82335913e-01 -5.94266132e-02 -4.40001577e-01
-2.05439068e-02 -8.42793107e-01 -2.76286572e-01 -5.29648185e-01
-1.27213359e-01 -4.10648398e-02 7.82632470e-01 -5.07909097e-02
6.85451269e-01 9.68815446e-01 -1.71439743e+00 2.39552453e-01
6.53445601e-01 1.19263268e+00 -2.12651372e-01 1.32525182e+00
-2.50943214e-01 5.09495318e-01 -5.06669044e-01 -2.13483423e-01
8.56555700e-01 -2.57736892e-01 -7.59702981e-01 1.99947625e-01
6.82803154e-01 -1.92970514e-01 -2.12415308e-01 1.09373689e+00
7.54854858e-01 3.05996602e-03 7.17262685e-01 -7.03264654e-01
-3.92305166e-01 1.82265133e-01 -5.45962274e-01 1.02066052e+00
-1.79155022e-01 3.99705976e-01 2.55395889e-01 -5.16610384e-01
6.52793109e-01 1.27213717e+00 4.55329508e-01 2.64970869e-01
-1.03088617e+00 1.28015671e-02 -3.65438223e-01 2.07338020e-01
-1.06752658e+00 -5.91711640e-01 9.58410576e-02 -6.65962875e-01
1.21640384e+00 2.41197199e-01 6.21933162e-01 -7.57210776e-02
3.11244905e-01 2.67526537e-01 1.45981956e+00 -5.80974579e-01
1.05674557e-01 4.02279526e-01 7.08717644e-01 7.36224771e-01
4.36886340e-01 1.60279974e-01 1.45058453e-01 5.05254507e-01
1.10295701e+00 -4.84126538e-01 -8.34407583e-02 3.25203925e-01
-7.62130916e-01 4.73529428e-01 4.60199416e-01 3.56584668e-01
-4.48631436e-01 1.56527773e-01 5.27411938e-01 4.03325438e-01
-8.29106010e-03 3.55431199e-01 -1.68524206e-01 -3.26575227e-02
-1.04865050e+00 -2.07143396e-01 6.16281271e-01 7.86895514e-01
4.33214426e-01 6.01076148e-03 -3.73144239e-01 4.28116590e-01
3.83012742e-01 -1.13661978e-02 5.11909962e-01 -9.16413188e-01
-1.83190942e-01 9.43994164e-01 4.89240475e-02 -5.83085179e-01
-5.73326766e-01 -4.92968261e-01 -4.91244435e-01 1.16274655e+00
9.20498550e-01 -7.41391718e-01 -1.51188111e+00 6.16581917e-01
3.31117243e-01 7.24098980e-02 1.25163704e-01 9.57859635e-01
1.18708360e+00 3.90192002e-01 -6.36434034e-02 1.06498636e-01
1.57537842e+00 -8.48758936e-01 -7.72427142e-01 3.35384935e-01
1.89795166e-01 -1.21609139e+00 3.77438724e-01 4.35225099e-01
-1.37689793e+00 -8.03256333e-01 -1.17514825e+00 -3.13496292e-01
-7.24228203e-01 4.61053461e-01 4.86447871e-01 1.12539315e+00
-1.38382411e+00 5.65515757e-01 -7.75245428e-01 -7.04732299e-01
9.11739171e-01 5.52229047e-01 -2.96878278e-01 3.56002599e-01
-3.26019883e-01 1.07597387e+00 1.97387427e-01 2.63736069e-01
-5.33983231e-01 -2.38658771e-01 -3.28632176e-01 -1.60248011e-01
-1.64032996e-01 -6.15619004e-01 1.02405000e+00 -6.56610250e-01
-1.44398105e+00 9.36213970e-01 -1.27822503e-01 -1.14581764e+00
8.23226929e-01 8.63302872e-02 -2.10802630e-01 5.23762047e-01
-4.47944403e-01 8.69729936e-01 5.19027114e-01 -6.72142982e-01
-1.13064408e+00 -3.16802830e-01 -1.58469118e-02 -7.66528919e-02
3.20325583e-01 3.11256111e-01 -3.23075920e-01 -2.28210852e-01
1.20637335e-01 -3.09164643e-01 -2.01634496e-01 9.00629684e-02
-3.63180369e-01 -8.39241967e-02 6.99872673e-01 -7.81945646e-01
1.09001696e+00 -1.92335188e+00 -5.70028543e-01 1.62326187e-01
2.90058911e-01 6.52710795e-01 8.50760564e-02 5.95269762e-02
-1.91666663e-01 2.81773746e-01 3.47223401e-01 -1.60579272e-02
-4.75815564e-01 -5.30178361e-02 1.21176496e-01 6.15334749e-01
4.15642560e-01 5.84232748e-01 -2.85082012e-01 -7.13446319e-01
2.79676437e-01 6.18697524e-01 -3.12726110e-01 -5.07693477e-02
1.20330386e-01 1.79838255e-01 -5.60212545e-02 8.12455118e-01
6.34183228e-01 2.89765239e-01 -6.70554638e-01 -5.22914886e-01
-7.70381570e-01 -3.27263355e-01 -1.22921431e+00 1.06189477e+00
1.09407216e-01 1.18176520e+00 -1.05920069e-01 -9.00498807e-01
1.19060743e+00 2.06196040e-01 1.57698721e-01 -5.13852179e-01
8.53372574e-01 1.81665733e-01 6.51382864e-01 -8.84457052e-01
2.21142918e-01 3.20306033e-01 8.97993565e-01 1.18423007e-01
1.40501466e-02 6.76733181e-02 6.39214575e-01 -1.77595451e-01
6.17996693e-01 3.53758723e-01 3.38273078e-01 -3.89313847e-01
6.02440059e-01 3.61731052e-01 3.64626825e-01 4.63588208e-01
-6.09277785e-01 6.32914484e-01 8.16801190e-01 -5.79247475e-01
-1.08830488e+00 -6.44818306e-01 -6.58551633e-01 1.58084482e-01
-5.79731613e-02 3.15224171e-01 -1.09584391e+00 -3.09600860e-01
-1.65481940e-01 2.06150845e-01 -7.84024298e-01 3.68378133e-01
-1.58649623e-01 -8.91649663e-01 6.15910649e-01 2.48234376e-01
9.85785425e-01 -9.19661939e-01 -1.44218826e+00 2.52601027e-01
8.40062857e-01 -8.39076102e-01 -2.90061813e-02 2.24890590e-01
-1.24645174e+00 -1.32715166e+00 -8.43723655e-01 -8.23163211e-01
9.07880127e-01 1.46030605e-01 6.94429457e-01 -5.97228296e-02
-1.14522564e+00 -7.03423545e-02 -1.17130309e-01 -7.88141251e-01
-1.86628133e-01 -5.32777071e-01 -3.83814573e-01 1.42154351e-01
8.07580769e-01 -4.97243971e-01 -9.24277842e-01 1.65711924e-01
-5.55090308e-01 -4.61583853e-01 8.41797650e-01 3.43464047e-01
6.62418723e-01 3.65622640e-01 2.61478961e-01 -8.51347804e-01
6.37931883e-01 5.61142340e-02 -9.83917713e-01 1.77106932e-02
-4.38814074e-01 -1.69370413e-01 8.45002905e-02 -1.03126004e-01
-8.26481104e-01 2.92858761e-02 2.25592911e-01 -1.10471897e-01
-5.64597785e-01 -8.27089399e-02 2.46993035e-01 -7.76563108e-01
9.92151618e-01 -2.22192809e-01 3.59241337e-01 -5.19580781e-01
2.48755708e-01 9.14228261e-01 6.58284962e-01 1.93694994e-01
1.99799404e-01 5.79623520e-01 2.29424894e-01 -1.12835813e+00
-2.21231267e-01 -5.53199232e-01 -8.37608993e-01 -3.60291749e-01
1.23196507e+00 -4.29957300e-01 -6.10892832e-01 7.70168781e-01
-1.19749773e+00 -7.48621970e-02 -2.76141226e-01 5.26022553e-01
-1.54769048e-01 1.99479580e-01 -6.28082871e-01 -1.03504837e+00
-4.79592532e-01 -1.15408230e+00 4.47050557e-02 1.14566231e+00
2.17386588e-01 -8.26695442e-01 -2.29520008e-01 2.56316394e-01
6.74899757e-01 3.84414792e-01 7.81275749e-01 -5.72407961e-01
-6.75703764e-01 -2.87496895e-01 -9.39640164e-01 5.94917655e-01
-6.07867464e-02 8.56344581e-01 -1.20669198e+00 3.20638031e-01
-2.45169163e-01 3.38385671e-01 9.69979346e-01 1.02333713e+00
5.94566286e-01 1.00751169e-01 -1.12357765e-01 6.99470878e-01
1.87779880e+00 6.94411039e-01 1.01369536e+00 3.45760256e-01
7.97799230e-02 7.05958426e-01 3.99404675e-01 1.79945275e-01
-1.05707034e-01 -6.77803680e-02 5.61473608e-01 -5.20469069e-01
-7.10126102e-01 6.78755522e-01 -2.77873814e-01 -2.11889576e-02
-6.74086213e-01 9.01942700e-02 -7.38491952e-01 7.87232399e-01
-1.48529863e+00 -7.22871363e-01 -8.05139482e-01 2.15344071e+00
5.62327862e-01 3.93277705e-01 3.71666074e-01 1.69474646e-01
7.25699842e-01 -6.95449471e-01 -4.87304062e-01 -7.28531003e-01
-1.17634542e-01 7.26516962e-01 1.06209695e+00 5.07744312e-01
-1.18311524e+00 8.43078613e-01 7.10430813e+00 9.90319904e-03
-1.34958255e+00 -8.09689537e-02 9.76971567e-01 -1.40112132e-01
6.30858243e-01 -1.15016535e-01 -1.04984081e+00 3.48428518e-01
8.44403625e-01 5.89008257e-02 3.98583785e-02 4.78729993e-01
5.75697243e-01 -8.28259170e-01 -5.54700911e-01 9.15273428e-01
-3.48997742e-01 -1.49789429e+00 -1.93096861e-01 5.99449314e-02
7.37203717e-01 1.34755224e-01 -2.03432351e-01 -1.40196458e-01
1.44696946e-03 -1.08177638e+00 2.27399617e-01 1.03087199e+00
4.92723882e-01 -6.69895351e-01 1.14597726e+00 -3.29143494e-01
-6.31665885e-01 5.34200459e-04 -8.25847924e-01 -1.71362236e-01
-2.84635760e-02 6.33945048e-01 -9.30256009e-01 7.57910833e-02
7.44741797e-01 3.32006216e-01 -8.25698733e-01 2.23519754e+00
2.02536047e-01 5.26073098e-01 -3.68454218e-01 -4.74474654e-02
3.21518898e-01 -6.02435648e-01 6.69131815e-01 1.33639324e+00
2.33348504e-01 -6.86868466e-03 -4.60068971e-01 8.83766413e-01
2.73822486e-01 4.39244270e-01 -4.74807769e-01 -1.48201315e-02
1.90686002e-01 1.50782990e+00 -1.31263578e+00 -3.98023605e-01
-5.66403687e-01 4.18300003e-01 -3.62831265e-01 4.84350592e-01
-4.43273187e-01 -1.08394861e+00 6.21319592e-01 4.05044496e-01
2.10937083e-01 -6.58971891e-02 -7.01638818e-01 -3.58288467e-01
-4.62635368e-01 -1.65678352e-01 1.11147352e-01 -5.60728490e-01
-7.65185654e-01 9.24764991e-01 -3.43765676e-01 -8.66019368e-01
3.66860420e-01 -1.06882298e+00 -8.84843111e-01 1.53518295e+00
-1.67671990e+00 -8.49107146e-01 -5.13405263e-01 3.28070670e-01
4.30399120e-01 -6.22257173e-01 3.12963575e-01 2.48700008e-01
-8.59760106e-01 1.50750086e-01 -3.87219816e-01 2.40147486e-01
6.17830575e-01 -1.48948419e+00 2.37542152e-01 1.42654610e+00
-1.63375974e-01 6.63120508e-01 6.65588796e-01 -5.99539280e-01
-8.42622578e-01 -1.00057125e+00 5.08644879e-01 -1.76182743e-02
4.42717761e-01 3.70217413e-01 -6.22120261e-01 3.38712364e-01
7.38615870e-01 2.50774860e-01 3.63435715e-01 -5.10575533e-01
3.77376884e-01 -1.06338732e-01 -1.38452339e+00 3.89870971e-01
3.39190304e-01 4.57508937e-02 -5.53447068e-01 2.25588024e-01
1.52895898e-01 -4.25397962e-01 -8.09209824e-01 -1.04603730e-01
3.96590889e-01 -1.34934115e+00 4.96195436e-01 -7.33920872e-01
2.14138180e-01 -5.23109317e-01 3.75858307e-01 -7.72046745e-01
-2.80192420e-02 -9.02977109e-01 2.72842228e-01 1.15902913e+00
5.96301377e-01 -8.45246077e-01 9.35380995e-01 6.38537705e-01
-9.67534482e-02 -7.11462677e-01 -5.97145140e-01 -3.57101023e-01
-3.97895634e-01 -7.25408569e-02 1.65648267e-01 3.33953321e-01
-5.49161613e-01 1.26352042e-01 3.60651016e-01 4.07166369e-02
6.46235824e-01 -3.21046442e-01 5.05601406e-01 -1.30038846e+00
1.10292912e-01 -8.27528059e-01 -1.12457979e+00 -5.48440933e-01
-9.22019303e-01 -3.23925942e-01 -3.41725647e-01 -1.88091946e+00
-4.27094132e-01 -1.82888746e-01 -6.06483221e-02 2.79131263e-01
2.48534545e-01 5.97183406e-01 -1.97904900e-01 -1.88524574e-01
1.68041170e-01 -4.03760910e-01 1.38789356e+00 3.18370372e-01
-4.26093251e-01 3.53463620e-01 -5.26435018e-01 6.35963321e-01
8.28658700e-01 7.09100068e-02 -3.91394466e-01 -2.23611474e-01
-1.06949449e-01 -2.29477942e-01 3.17671061e-01 -1.42501652e+00
5.76180100e-01 4.24827576e-01 3.46404105e-01 -6.84581339e-01
-1.27772436e-01 -7.65827298e-01 -3.63908738e-01 5.57296395e-01
2.47684750e-03 -1.03977643e-01 3.87084126e-01 2.76870102e-01
-3.57215196e-01 -5.25550544e-01 1.42695367e+00 -3.10458183e-01
-8.49034786e-01 -2.84928251e-02 -4.76067692e-01 -2.99837410e-01
1.50588632e+00 -1.13061786e+00 -5.22413552e-01 1.30405977e-01
-1.16713679e+00 1.14308745e-01 3.19946170e-01 -2.27411911e-01
3.90994519e-01 -5.27884185e-01 -7.99565732e-01 3.41045767e-01
-4.61712539e-01 9.94863138e-02 1.79564416e-01 1.28415537e+00
-1.47323918e+00 5.63207686e-01 -1.01110339e+00 -4.16932821e-01
-1.64405048e+00 2.70367384e-01 9.16710496e-01 4.82791871e-01
-5.56066632e-01 1.13634872e+00 -3.97681206e-01 5.63696146e-01
4.07913834e-01 -8.52369010e-01 -1.12554252e+00 3.09371680e-01
6.89596653e-01 8.30470026e-01 1.78195775e-01 -2.33701259e-01
5.22013834e-05 6.77535653e-01 -1.03118852e-01 -4.18014675e-02
1.14154637e+00 -2.39435866e-01 -4.64440227e-01 8.81220549e-02
7.23615110e-01 -7.84745216e-02 -1.27806950e+00 1.25464842e-01
1.12269580e-01 -4.85455155e-01 7.05316305e-01 -1.03906250e+00
-1.35200834e+00 1.05994952e+00 1.17262363e+00 3.79732847e-01
1.22943532e+00 -6.80575907e-01 4.32262808e-01 1.17718950e-01
-1.38431996e-01 -1.13304973e+00 -4.93496388e-01 2.35452518e-01
4.78840947e-01 -1.09908700e+00 -2.02050041e-02 -5.37149191e-01
-4.36819762e-01 1.64738619e+00 6.83900476e-01 -5.40647447e-01
7.65269637e-01 3.95018101e-01 4.74416405e-01 -3.60841811e-01
-4.78677362e-01 -7.69193888e-01 4.79821116e-01 8.81546915e-01
7.68919289e-01 -2.62639523e-01 -6.30688369e-01 2.36990452e-01
9.94814336e-02 2.75554746e-01 1.01651514e+00 6.44597292e-01
-8.34848046e-01 -9.01740611e-01 -3.17576110e-01 7.91911781e-01
-8.44765961e-01 1.57141294e-02 -3.63135666e-01 8.94344211e-01
4.91386712e-01 8.32490742e-01 2.92555124e-01 1.70678183e-01
3.06125551e-01 -1.48840785e-01 4.64110076e-01 -6.50345325e-01
-8.99399996e-01 2.86149114e-01 2.35466376e-01 -5.03615022e-01
-3.29426855e-01 -4.99639273e-01 -1.32545543e+00 9.85840783e-02
-4.77055795e-02 -9.04251710e-02 1.04224801e+00 5.29643655e-01
2.89757311e-01 7.65927374e-01 1.06264874e-01 -3.78749013e-01
-1.95572630e-01 -1.21777058e+00 -7.38525629e-01 -7.86284506e-02
3.36289436e-01 -3.15569609e-01 -1.61480814e-01 5.26994407e-01]
|
[15.813056945800781, -3.967346429824829]
|
02ed0504-5a59-4b5d-a72a-50540bde0150
|
metagraspnet-a-large-scale-benchmark-dataset
|
2112.14663
| null |
https://arxiv.org/abs/2112.14663v3
|
https://arxiv.org/pdf/2112.14663v3.pdf
|
MetaGraspNet_v0: A Large-Scale Benchmark Dataset for Vision-driven Robotic Grasping via Physics-based Metaverse Synthesis
|
There has been increasing interest in smart factories powered by robotics systems to tackle repetitive, laborious tasks. One impactful yet challenging task in robotics-powered smart factory applications is robotic grasping: using robotic arms to grasp objects autonomously in different settings. Robotic grasping requires a variety of computer vision tasks such as object detection, segmentation, grasp prediction, pick planning, etc. While significant progress has been made in leveraging of machine learning for robotic grasping, particularly with deep learning, a big challenge remains in the need for large-scale, high-quality RGBD datasets that cover a wide diversity of scenarios and permutations. To tackle this big, diverse data problem, we are inspired by the recent rise in the concept of metaverse, which has greatly closed the gap between virtual worlds and the physical world. Metaverses allow us to create digital twins of real-world manufacturing scenarios and to virtually create different scenarios from which large volumes of data can be generated for training models. In this paper, we present MetaGraspNet: a large-scale benchmark dataset for vision-driven robotic grasping via physics-based metaverse synthesis. The proposed dataset contains 100,000 images and 25 different object types and is split into 5 difficulties to evaluate object detection and segmentation model performance in different grasping scenarios. We also propose a new layout-weighted performance metric alongside the dataset for evaluating object detection and segmentation performance in a manner that is more appropriate for robotic grasp applications compared to existing general-purpose performance metrics. Our benchmark dataset is available open-source on Kaggle, with the first phase consisting of detailed object detection, segmentation, layout annotations, and a layout-weighted performance metric script.
|
['Alexander Wong', 'Maximilian Gilles', 'E. Zhixuan Zeng', 'Yuhao Chen']
|
2021-12-29
| null | null | null | null |
['robotic-grasping']
|
['robots']
|
[ 7.70489573e-02 -3.46640408e-01 3.87960603e-03 -1.71501160e-01
-2.84696341e-01 -7.52007425e-01 3.44352394e-01 -9.76574048e-02
5.92352357e-04 2.05807462e-01 -2.57350147e-01 -1.09798536e-01
-5.70875406e-01 -8.32553148e-01 -9.16430593e-01 -6.46720409e-01
-2.18688712e-01 8.87423396e-01 4.75619316e-01 -3.37515235e-01
4.30830240e-01 9.96091545e-01 -1.97978115e+00 5.17187774e-01
5.05181372e-01 1.43956244e+00 1.07274461e+00 5.97818077e-01
-2.15754732e-01 6.87502995e-02 -3.62790555e-01 -1.01061799e-01
7.30392277e-01 3.40883315e-01 -6.84892654e-01 -1.72320439e-03
1.08083963e-01 -2.97648668e-01 -2.41053030e-01 6.26052499e-01
4.33624446e-01 -2.06381798e-01 5.76016307e-01 -1.59183013e+00
-5.04411757e-01 5.77180982e-01 -2.41349772e-01 -4.84215111e-01
2.19599858e-01 7.15581596e-01 7.03918457e-01 -7.93899536e-01
8.06244433e-01 1.60416758e+00 3.34500372e-01 5.03756642e-01
-8.37640941e-01 -4.27611679e-01 -1.15754969e-01 2.45620593e-01
-5.46900570e-01 1.73142985e-01 8.36392820e-01 -4.30443823e-01
7.44970918e-01 5.32309934e-02 6.91813588e-01 1.44754159e+00
3.33511472e-01 8.58084798e-01 1.03361070e+00 -2.26901740e-01
4.04632568e-01 -3.32573920e-01 -1.04504742e-01 6.07211947e-01
3.55439633e-01 8.62549432e-03 -1.52855650e-01 1.43303454e-01
1.01490235e+00 2.13769391e-01 5.07774428e-02 -1.09998190e+00
-1.90021753e+00 2.97666699e-01 7.04666972e-01 1.83883369e-01
-6.08155668e-01 6.19986773e-01 4.55492109e-01 1.36082247e-01
-3.85267884e-01 8.06871712e-01 -9.09689605e-01 -8.95122513e-02
-2.01617792e-01 6.24428332e-01 9.73919809e-01 1.63569844e+00
5.51670194e-01 -1.91068560e-01 -3.57360393e-01 7.58852839e-01
2.34766275e-01 8.03628266e-01 4.59078439e-02 -1.21814275e+00
4.88557309e-01 8.69987845e-01 4.11738455e-01 -6.57666147e-01
-5.72302043e-01 2.57096499e-01 -4.49646413e-01 4.94699329e-01
6.40980482e-01 2.46427849e-01 -1.26469171e+00 1.10076213e+00
3.66739124e-01 -3.91011387e-01 -1.40305653e-01 1.13115191e+00
8.28736365e-01 3.66312176e-01 -3.65231372e-02 4.40218240e-01
1.39215398e+00 -1.05524576e+00 -2.69497573e-01 6.09014928e-02
1.07307322e-01 -9.20060575e-01 1.20816255e+00 7.78506160e-01
-8.65330100e-01 -4.86397833e-01 -9.98750687e-01 -3.66499238e-02
-7.18318105e-01 1.73321843e-01 1.14052355e+00 2.87227809e-01
-4.26674902e-01 8.55037153e-01 -7.61466384e-01 -5.26379585e-01
7.77204812e-01 3.99947256e-01 -2.03621596e-01 -4.87805694e-01
-4.68007237e-01 1.10113096e+00 6.86334372e-01 8.00862685e-02
-1.17701972e+00 -6.35688841e-01 -2.97755152e-01 -1.45062178e-01
7.31030941e-01 -5.00677943e-01 1.19797039e+00 -2.68350005e-01
-1.60663319e+00 6.10009730e-01 5.98099649e-01 -7.54148960e-02
6.60319924e-01 -2.52832741e-01 1.56763479e-01 1.19154364e-01
-1.68301180e-01 8.18642020e-01 9.29910481e-01 -1.78831232e+00
-4.82423365e-01 -4.08995032e-01 2.32818648e-01 -2.68787742e-01
1.42009228e-01 -2.14307562e-01 -4.63394254e-01 -3.82848114e-01
2.05706805e-01 -1.11230946e+00 -1.51872084e-01 3.05076361e-01
-5.17763674e-01 -4.56138521e-01 1.35089910e+00 -2.80382067e-01
-9.50127691e-02 -1.78790307e+00 3.55411470e-01 -6.74991310e-02
-2.26598471e-01 3.25855285e-01 -4.22238022e-01 6.83961451e-01
4.04959351e-01 -1.98765576e-01 -5.39020374e-02 1.76394545e-02
4.63432401e-01 3.94409686e-01 -3.77304167e-01 7.26224715e-03
2.49672204e-01 1.13146949e+00 -1.04883027e+00 -3.69465828e-01
7.09567368e-01 1.50774792e-01 -3.66671860e-01 1.58801481e-01
-9.24306750e-01 6.05313718e-01 -7.84658134e-01 1.45085216e+00
8.26623261e-01 7.41711706e-02 -2.82366332e-02 -5.63859046e-01
-1.40129015e-01 -3.54397357e-01 -1.11372900e+00 2.20359468e+00
-6.52270257e-01 1.82493687e-01 4.34912533e-01 -9.61597741e-01
1.15313983e+00 -3.67056988e-02 7.76441753e-01 -5.15450656e-01
4.23356295e-01 4.96392846e-01 -6.69510588e-02 -8.93006623e-01
5.52396953e-01 3.95980030e-01 -8.88058320e-02 -4.30985028e-03
1.58362478e-01 -9.63014126e-01 3.62498194e-01 -1.77117139e-01
1.23478794e+00 8.46412301e-01 -5.14008701e-01 -3.47432315e-01
-5.84289804e-02 6.46573186e-01 1.55873418e-01 6.68081939e-01
-1.37161866e-01 6.40657902e-01 3.19750547e-01 -5.72135270e-01
-1.34933674e+00 -1.37290466e+00 -1.05790310e-01 7.87045181e-01
6.78643942e-01 2.82327056e-01 -5.95139444e-01 -3.94212633e-01
6.73573732e-01 4.66450870e-01 -2.45627582e-01 1.10989153e-01
-6.81860447e-01 -3.65218848e-01 4.68787327e-02 5.27046680e-01
5.59946835e-01 -1.74728465e+00 -1.29648113e+00 2.69588619e-01
2.53605783e-01 -1.42102146e+00 3.76045644e-01 1.64025247e-01
-8.82567167e-01 -1.51606071e+00 -7.50133216e-01 -8.27813387e-01
3.82825345e-01 4.89566445e-01 1.06387842e+00 -5.23376130e-02
-1.06778264e+00 6.95416689e-01 -7.95189798e-01 -9.03364301e-01
-4.59169209e-01 9.18999985e-02 5.90416491e-02 -6.31125152e-01
-6.03927560e-02 -4.07144397e-01 -7.87241280e-01 5.47974467e-01
-9.03965354e-01 1.05007611e-01 1.01593184e+00 5.17045856e-01
4.76762146e-01 -2.67203331e-01 7.02201366e-01 -2.99712643e-02
3.61850679e-01 -3.76366377e-01 -7.62940705e-01 4.91943538e-01
-1.27383381e-01 -1.03557430e-01 4.05154765e-01 -5.56940377e-01
-7.36851692e-01 1.29312187e-01 2.63334513e-01 -7.17521548e-01
-2.99560934e-01 6.65889606e-02 -2.16661572e-01 -7.42377862e-02
3.09174597e-01 -1.70469895e-01 1.12659775e-01 -6.25208259e-01
6.36708081e-01 7.61890531e-01 5.26249111e-01 -1.06987786e+00
5.69793284e-01 3.92535299e-01 3.32272530e-01 -8.40532422e-01
-3.03371429e-01 -3.53759944e-01 -8.72699201e-01 -3.81417125e-01
8.09544504e-01 -2.65632451e-01 -1.15823829e+00 7.50984848e-01
-1.36777520e+00 -6.98145211e-01 -3.75847101e-01 3.85375530e-01
-9.81291831e-01 9.37962830e-02 -3.22202742e-01 -6.15495861e-01
-4.00461018e-01 -1.59648347e+00 1.38942838e+00 8.92741457e-02
2.93609470e-01 -1.32404163e-01 -5.00130534e-01 3.98592800e-01
4.75202054e-01 7.51047075e-01 9.92744088e-01 -1.92987740e-01
-1.21020341e+00 -2.25795344e-01 -5.92101336e-01 3.04864556e-01
2.16999412e-01 1.48009002e-01 -6.18126690e-01 -2.14384258e-01
-3.83591712e-01 -6.85349762e-01 5.93625665e-01 1.23419441e-01
1.60520148e+00 2.33907953e-01 -5.60603976e-01 1.84942514e-01
1.49695098e+00 2.27307633e-01 5.39753079e-01 2.90562361e-01
8.00856709e-01 7.31385946e-01 9.57576632e-01 2.90398628e-01
7.24142939e-02 8.82402718e-01 1.15456486e+00 3.46810341e-01
-2.53763258e-01 9.12513435e-02 -1.53052926e-01 3.65169525e-01
-3.76204640e-01 -1.63363844e-01 -1.09335434e+00 5.76856673e-01
-1.87353551e+00 -7.02877164e-01 -1.96265042e-01 1.88146901e+00
3.86026710e-01 -8.53595808e-02 -1.07010089e-01 9.21956524e-02
4.64711875e-01 -3.33419889e-01 -8.97779584e-01 -3.54294181e-01
2.00951442e-01 2.79713094e-01 5.92174172e-01 -3.32784295e-01
-9.27475691e-01 1.05035853e+00 5.13111734e+00 6.29842401e-01
-1.14572227e+00 -5.71880564e-02 -9.86418352e-02 9.45349112e-02
1.82933331e-01 -1.91980869e-01 -3.10725152e-01 4.39996779e-01
2.29016900e-01 3.18941534e-01 5.61015785e-01 1.37847030e+00
-2.51605399e-02 -3.26675951e-01 -1.34474432e+00 1.01028264e+00
-3.31518590e-01 -1.40078294e+00 5.79541996e-02 -5.81709389e-03
6.88435256e-01 3.02515119e-01 -2.12465197e-01 2.09177420e-01
3.46319109e-01 -8.07256877e-01 1.01679504e+00 6.14272952e-01
5.17153323e-01 -3.04512888e-01 5.08921146e-01 1.96924567e-01
-1.06099916e+00 -4.84705329e-01 -4.73380774e-01 2.14350268e-01
5.38023151e-02 4.99090433e-01 -9.11739349e-01 6.83976054e-01
1.12027156e+00 2.43989855e-01 -9.05854031e-02 1.35120595e+00
6.57331720e-02 -2.33607695e-01 -1.65142521e-01 -5.19261003e-01
1.99441701e-01 -7.61823282e-02 5.76874971e-01 8.97658229e-01
4.36287284e-01 -1.35308221e-01 3.73527378e-01 1.17650747e+00
3.38359997e-02 -3.16763639e-01 -6.31129563e-01 -1.03922166e-01
5.04101932e-01 1.46097922e+00 -1.04872525e+00 -6.32207096e-02
-5.20448759e-02 8.32099438e-01 2.67874333e-03 6.71753809e-02
-7.15090871e-01 -4.11122084e-01 5.71064770e-01 -7.07271276e-03
4.90232021e-01 -8.09883118e-01 -4.34878200e-01 -6.94899440e-01
2.21889675e-01 -6.19113863e-01 -4.48617220e-01 -1.12138045e+00
-1.22674811e+00 1.88448116e-01 3.66605490e-01 -1.08885062e+00
1.59160376e-01 -1.50980127e+00 -2.72821337e-01 4.91840810e-01
-1.31600249e+00 -1.52391744e+00 -8.83776724e-01 2.34061241e-01
9.65126574e-01 -5.87688759e-02 6.90456331e-01 -6.50002286e-02
-1.40717626e-01 -2.32595652e-01 9.97015834e-02 -1.37836680e-01
4.37555909e-01 -1.08180976e+00 2.86122054e-01 1.52261063e-01
-3.29560667e-01 2.88466543e-01 6.21155858e-01 -5.69142044e-01
-2.50742078e+00 -1.02788329e+00 -2.87934363e-01 -6.64002717e-01
6.37583435e-01 -5.33040345e-01 -5.26527405e-01 2.87202299e-01
-1.36694074e-01 1.86565802e-01 -4.37899172e-01 -4.45348024e-01
-9.00847539e-02 -1.92339584e-01 -1.51780081e+00 3.72089744e-01
1.38436961e+00 1.70921713e-01 -4.80683386e-01 7.14935184e-01
7.42090404e-01 -6.16885245e-01 -1.03347933e+00 8.77516687e-01
8.84246647e-01 -7.54348636e-01 1.19924068e+00 -4.13607359e-01
7.19881177e-01 -1.40433878e-01 -4.16470855e-01 -1.10312474e+00
7.24361986e-02 -3.30242991e-01 -1.81806177e-01 9.37800825e-01
3.27898376e-02 -4.44948703e-01 8.36774468e-01 4.96325433e-01
-6.46314800e-01 -9.96578038e-01 -8.84327471e-01 -1.15045571e+00
7.01526552e-02 -3.57855260e-01 9.06661332e-01 5.30217886e-01
-1.47178128e-01 -5.33930004e-01 2.25139871e-01 5.57817519e-02
7.83724666e-01 6.09280288e-01 1.12838876e+00 -1.49569678e+00
1.21744312e-01 -5.13494968e-01 -4.34786558e-01 -6.60961270e-01
-2.11675569e-01 -7.76744008e-01 4.28087592e-01 -2.14902711e+00
3.53851393e-02 -1.09095621e+00 4.99267466e-02 5.54415584e-01
6.03637218e-01 -2.34198868e-02 5.24020851e-01 3.30968350e-01
-4.43272740e-01 4.87751186e-01 1.89259017e+00 -3.80272597e-01
-9.41044390e-02 -1.84640571e-01 9.41635072e-02 5.37913084e-01
8.27636063e-01 -1.19636599e-02 -1.04691245e-01 -6.04394734e-01
-1.66993022e-01 -5.85004836e-02 7.87520826e-01 -1.32893467e+00
-8.90525728e-02 -4.74647164e-01 3.19883674e-01 -7.83105016e-01
5.60840189e-01 -1.11934233e+00 -8.69453419e-03 7.77566254e-01
1.73872352e-01 -1.35433495e-01 2.43412420e-01 3.70112956e-01
4.18014854e-01 -2.64523089e-01 5.44010460e-01 -6.11173928e-01
-1.15822053e+00 3.26355249e-01 2.42509678e-01 -4.93442893e-01
1.51664734e+00 -2.60990173e-01 -5.21567762e-01 4.62519556e-01
-5.94731688e-01 3.09485734e-01 6.71369612e-01 1.12616217e+00
5.93165278e-01 -1.16898382e+00 -3.71704459e-01 -7.13106468e-02
2.17278928e-01 4.68644083e-01 6.71319142e-02 3.56627673e-01
-8.89512300e-01 4.37110007e-01 -7.17996895e-01 -9.00759578e-01
-7.28511631e-01 7.69968152e-01 -9.10851732e-02 2.18946055e-01
-7.72191525e-01 5.50752342e-01 -4.23627764e-01 -7.38333464e-01
3.26183170e-01 -9.16454136e-01 3.67240459e-01 -3.52982759e-01
3.02775241e-02 7.29111791e-01 1.98578060e-01 -1.09383345e-01
-4.02711868e-01 5.70934236e-01 3.77157629e-01 3.34529161e-01
1.59607756e+00 4.56407040e-01 -1.74701005e-01 2.69444734e-01
7.58928418e-01 -5.57625294e-01 -1.51394784e+00 2.81086564e-01
3.50523517e-02 -4.27201778e-01 -2.27903813e-01 -1.23566914e+00
-9.89720464e-01 7.67276943e-01 7.46355355e-01 2.97851503e-01
5.73067963e-01 3.72921169e-01 8.25064957e-01 8.53947401e-01
1.34264064e+00 -1.25820017e+00 5.86643219e-01 5.24472535e-01
1.62257516e+00 -1.41401255e+00 -3.91504243e-02 -5.96293211e-01
-2.88055003e-01 1.30219507e+00 7.01140106e-01 -2.75125176e-01
3.46772522e-01 3.03769708e-01 -1.17855869e-01 -5.03216863e-01
-1.09874547e-01 -3.97658572e-02 -1.16269449e-02 8.68812084e-01
-2.67609775e-01 2.62133837e-01 4.57068346e-02 1.91252261e-01
-2.24246588e-02 2.74726689e-01 2.05856770e-01 1.43037486e+00
-5.84606469e-01 -1.22094059e+00 -5.17067850e-01 5.22558689e-01
1.30487129e-01 6.04091823e-01 -1.95307821e-01 9.86978412e-01
2.66512811e-01 8.15778255e-01 -8.12174305e-02 -3.43215793e-01
5.66066086e-01 -1.04515724e-01 1.02029669e+00 -3.91295344e-01
-3.05170000e-01 -6.28783584e-01 -1.23405911e-01 -9.23564255e-01
-4.16732281e-01 -5.69065630e-01 -1.39321065e+00 1.84917506e-02
-2.78103858e-01 -5.37388682e-01 1.69907141e+00 8.08205009e-01
5.35711765e-01 6.57401860e-01 3.99240136e-01 -1.92501390e+00
-7.19982862e-01 -9.67485130e-01 -3.83357197e-01 3.63976896e-01
-1.52451485e-01 -1.18426943e+00 1.28369957e-01 -2.86876351e-01]
|
[5.770267963409424, -0.852952241897583]
|
7afe7440-4520-41cc-aa42-9248b0ca324b
|
self-supervised-anomaly-detection-a-survey
|
2205.05173
| null |
https://arxiv.org/abs/2205.05173v2
|
https://arxiv.org/pdf/2205.05173v2.pdf
|
Self-Supervised Anomaly Detection: A Survey and Outlook
|
Over the past few years, anomaly detection, a subfield of machine learning that is mainly concerned with the detection of rare events, witnessed an immense improvement following the unprecedented growth of deep learning models. Recently, the emergence of self-supervised learning has sparked the development of new anomaly detection algorithms that surpassed state-of-the-art accuracy by a significant margin. This paper aims to review the current approaches in self-supervised anomaly detection. We present technical details of the common approaches and discuss their strengths and drawbacks. We also compare the performance of these models against each other and other state-of-the-art anomaly detection models. Finally, we discuss a variety of new directions for improving the existing algorithms.
|
['Narges Armanfard', 'Thi Kieu Khanh Ho', 'Hadi Hojjati']
|
2022-05-10
| null | null | null | null |
['self-supervised-anomaly-detection', 'supervised-anomaly-detection']
|
['computer-vision', 'computer-vision']
|
[-2.01732576e-01 -1.58726186e-01 1.54316396e-01 -5.41806817e-01
-2.84053832e-01 -1.54278249e-01 7.95974016e-01 7.78488100e-01
-3.73534054e-01 4.67263103e-01 -1.36986151e-01 -1.98085919e-01
-1.04144670e-01 -7.53119826e-01 -2.21856922e-01 -6.57534361e-01
-5.95707357e-01 4.16387200e-01 4.18154091e-01 -2.17627436e-01
3.08918893e-01 8.13806236e-01 -1.94719005e+00 1.01964705e-01
8.46381903e-01 1.36743248e+00 -9.97137785e-01 5.03922462e-01
-6.47562921e-01 7.75526762e-01 -6.65312529e-01 -4.01348710e-01
1.48746207e-01 -3.69183242e-01 -5.67837656e-01 -1.07053600e-01
6.19913638e-01 -3.31276596e-01 -6.39837086e-01 1.05470097e+00
2.56685019e-01 -1.71568543e-02 5.75158358e-01 -1.69226944e+00
-3.56983930e-01 2.49842435e-01 -5.34914911e-01 1.01697135e+00
1.20669618e-01 -3.13717425e-02 9.66598213e-01 -6.55909479e-01
1.35607913e-01 8.30931902e-01 7.89232671e-01 6.53187037e-01
-1.02797222e+00 -5.09483933e-01 3.92287314e-01 5.50999224e-01
-1.05251324e+00 -2.68327087e-01 7.57252574e-01 -2.03722030e-01
1.25610745e+00 1.84221238e-01 5.60479462e-01 1.14390945e+00
3.26418251e-01 9.97548401e-01 7.45991528e-01 -4.21192974e-01
4.52253014e-01 -3.26711714e-01 3.56328964e-01 8.31816196e-01
5.12039661e-01 2.56407320e-01 -4.15542960e-01 -7.05419302e-01
5.21783888e-01 3.16516489e-01 4.74425435e-01 -5.08041263e-01
-7.75483310e-01 1.02856791e+00 -5.04769646e-02 6.72676027e-01
-2.39970356e-01 -2.88407087e-01 9.56066430e-01 4.34887826e-01
6.04499698e-01 5.10305524e-01 -5.44221938e-01 -3.13031554e-01
-8.86660814e-01 2.49468610e-01 7.62401104e-01 4.40735489e-01
4.08579022e-01 6.70693040e-01 1.46447733e-01 6.87572777e-01
1.70823634e-01 1.11216068e-01 5.53857803e-01 -3.85720342e-01
-1.42257735e-01 9.00143445e-01 -2.47163191e-01 -8.18542123e-01
-5.92741549e-01 -6.80200100e-01 -1.08730614e+00 3.54686350e-01
4.97359812e-01 4.00171168e-02 -9.18284178e-01 1.24640512e+00
1.93450168e-01 4.07387882e-01 -5.05576096e-02 3.72066498e-01
2.77691394e-01 2.66620606e-01 7.85086900e-02 -5.98407276e-02
7.63099194e-01 -6.77173138e-01 -8.86712909e-01 -3.63225728e-01
7.62565792e-01 -5.19175947e-01 6.28216386e-01 6.95186615e-01
-6.82350218e-01 -1.16847463e-01 -1.03645599e+00 4.78418976e-01
-6.52626634e-01 -4.74029750e-01 9.52713966e-01 7.63019264e-01
-7.24816978e-01 8.29470456e-01 -1.18499339e+00 -7.06000686e-01
7.12182760e-01 2.08679274e-01 -3.74163568e-01 3.06319416e-01
-1.06945634e+00 9.11366999e-01 3.34934771e-01 -2.43496701e-01
-7.21809566e-01 -4.21929926e-01 -7.49320388e-01 -1.22412205e-01
3.94041747e-01 -1.26885399e-01 1.27952719e+00 -8.15081418e-01
-9.49729860e-01 1.07411754e+00 -6.58639595e-02 -8.81610990e-01
3.62226695e-01 -6.79748535e-01 -1.27608490e+00 -1.06707573e-01
-5.80723770e-02 -1.74509972e-01 7.16469169e-01 -5.82246006e-01
-1.15804052e+00 -5.76437116e-01 -4.75986987e-01 -3.58911693e-01
-4.12391692e-01 2.57271171e-01 4.96347919e-02 -7.33409345e-01
2.87560016e-01 -5.50612152e-01 -4.51460391e-01 9.55276713e-02
-3.81399274e-01 -5.72497427e-01 1.20813465e+00 -1.39610469e-01
1.33972180e+00 -2.50372243e+00 -4.14639950e-01 3.95320177e-01
2.62059003e-01 4.80591118e-01 2.06081375e-01 5.43168783e-01
-4.74663556e-01 -1.85906544e-01 -7.10481405e-01 -4.37453538e-01
-2.32190296e-01 5.21780849e-01 -5.85104883e-01 7.39042044e-01
3.45114946e-01 6.12959802e-01 -1.00479198e+00 -1.29558921e-01
3.65536243e-01 -8.73817131e-02 -3.11815232e-01 3.80686074e-01
-4.23457213e-02 4.73575503e-01 -3.35630298e-01 9.51634645e-01
5.15603423e-01 3.77033129e-02 -2.19783291e-01 4.56173956e-01
-2.68094420e-01 2.79273272e-01 -1.20056307e+00 1.32554042e+00
2.09621012e-01 6.09752953e-01 -2.44267568e-01 -1.58790243e+00
1.01517534e+00 1.70329139e-01 9.21448708e-01 -6.66092932e-01
-1.15709957e-02 5.13837218e-01 2.99466461e-01 -4.08786148e-01
1.14872508e-01 7.84348026e-02 4.75179590e-03 6.02702081e-01
2.68449813e-01 4.32754010e-01 1.94951996e-01 7.29745850e-02
1.50943196e+00 -3.63983512e-01 7.52923429e-01 -8.36601108e-02
7.77177751e-01 -1.13138370e-01 6.07444227e-01 1.06554997e+00
-6.83155477e-01 4.34814394e-01 3.80335242e-01 -1.17469430e+00
-8.31126392e-01 -1.32410395e+00 -3.49533081e-01 1.00243795e+00
-3.35368514e-01 -4.23621923e-01 -5.56884766e-01 -1.11951053e+00
2.79347926e-01 7.32032001e-01 -6.29604757e-01 -3.82454842e-01
-6.18957698e-01 -1.16257215e+00 9.18298960e-01 6.95171773e-01
7.04864621e-01 -1.24628556e+00 -5.97651780e-01 2.84106910e-01
1.08012073e-01 -1.07149339e+00 4.31991160e-01 3.60435009e-01
-1.22860074e+00 -1.12740922e+00 -1.26197606e-01 -3.69768113e-01
5.87778151e-01 -1.11404188e-01 1.30053449e+00 3.07583570e-01
-7.64605284e-01 4.62465674e-01 -4.23222393e-01 -9.32976246e-01
-2.82566130e-01 6.13891259e-02 5.56006610e-01 2.45719761e-01
1.15743256e+00 -6.99183464e-01 -4.24883038e-01 5.41485213e-02
-1.08071399e+00 -9.63625729e-01 4.09734458e-01 5.38125575e-01
4.09318417e-01 1.07349321e-01 7.09362149e-01 -8.73748183e-01
4.49252546e-01 -7.93776214e-01 -4.60399806e-01 -1.52138337e-01
-9.83753204e-01 8.10018107e-02 5.30707777e-01 -2.00554639e-01
-8.35599124e-01 -1.56199113e-01 -5.28917432e-01 -1.54414877e-01
-9.40497041e-01 1.35858595e-01 2.07451016e-01 -1.48652375e-01
8.92043352e-01 3.98984283e-01 7.29251429e-02 -7.20910907e-01
1.05713867e-01 5.91444612e-01 5.62458873e-01 -2.34987482e-01
8.15557003e-01 7.07109332e-01 -5.14593609e-02 -1.05347288e+00
-9.62504387e-01 -6.00863874e-01 -6.49133325e-01 -2.91259009e-02
4.72469002e-01 -3.36115301e-01 -1.45296872e-01 1.17197013e+00
-7.90446460e-01 1.25705063e-01 -6.69775784e-01 1.28620699e-01
-2.40859985e-01 5.93625307e-01 -6.22570097e-01 -9.40749586e-01
-3.28883946e-01 -6.20357394e-01 6.17727399e-01 9.21459422e-02
-4.53507364e-01 -1.07659352e+00 5.67592919e-01 -2.46240422e-01
7.55009592e-01 3.06752533e-01 8.28403592e-01 -1.66475451e+00
-1.34286627e-01 -7.43368745e-01 6.19693175e-02 6.38727605e-01
2.57704198e-01 -3.96755291e-04 -1.02979231e+00 -1.49747640e-01
1.56549305e-01 1.07474066e-01 8.38452995e-01 1.56186864e-01
1.58990443e+00 1.65684782e-02 -3.49108964e-01 6.04252875e-01
1.12432480e+00 3.09758455e-01 6.76407635e-01 7.68040240e-01
4.95325267e-01 2.65733153e-01 4.20143634e-01 5.24447381e-01
4.35402989e-02 1.74427807e-01 8.21671247e-01 -1.48104966e-01
4.68346000e-01 1.72037527e-01 -2.97507475e-04 5.12777448e-01
-1.00255981e-01 6.29703850e-02 -1.24378526e+00 6.73517883e-01
-1.98361313e+00 -1.15513623e+00 -1.97578505e-01 2.17852497e+00
2.88151622e-01 5.20918250e-01 4.28458929e-01 6.17267251e-01
5.03995419e-01 2.42470831e-01 -8.04428697e-01 -6.81670129e-01
-2.42983028e-01 3.37886930e-01 2.29675248e-01 -1.04073882e-02
-1.66347027e+00 8.60548317e-01 7.98939323e+00 3.99463177e-01
-1.06998658e+00 -1.80531278e-01 4.83453363e-01 1.45013019e-01
2.31826469e-01 -3.71775925e-01 -5.82222760e-01 4.83460307e-01
1.09009814e+00 -9.97998472e-03 -3.30942422e-02 1.24440801e+00
-2.97094226e-01 -1.60362914e-01 -1.19286239e+00 8.26928198e-01
2.66230673e-01 -1.12579322e+00 1.29182667e-01 -1.13935187e-01
5.01249254e-01 4.68881220e-01 1.55230360e-02 3.46320033e-01
6.24846369e-02 -8.60809147e-01 9.59887262e-03 3.63259852e-01
1.97185695e-01 -9.31878626e-01 9.63856995e-01 2.18015477e-01
-8.52643669e-01 -4.46357518e-01 -1.79825619e-01 -3.69663477e-01
3.31900865e-02 1.08321083e+00 -2.68940270e-01 3.70351642e-01
1.10754287e+00 7.49807954e-01 -6.85224295e-01 1.40226686e+00
-2.08232347e-02 9.45525527e-01 -4.26764131e-01 3.30473721e-01
4.74950314e-01 -1.48349598e-01 9.23991919e-01 1.23962462e+00
4.90400232e-02 -1.48973301e-01 1.59555316e-01 4.40474570e-01
3.16889226e-01 1.04064293e-01 -8.50028813e-01 -2.39002928e-02
1.98064491e-01 1.06844270e+00 -5.58223724e-01 -4.00777072e-01
-9.16959405e-01 9.12883759e-01 3.54394674e-01 6.89448565e-02
-5.98591387e-01 -5.43409050e-01 1.16265059e+00 1.71343565e-01
-3.21817361e-02 -2.06530213e-01 -3.82788748e-01 -1.26041067e+00
-2.35011503e-02 -1.01781058e+00 9.55519676e-01 1.82568610e-01
-1.91565251e+00 6.12654030e-01 -1.15025919e-02 -1.17316806e+00
-3.73327136e-01 -6.85413361e-01 -1.07793713e+00 3.13318908e-01
-1.27897465e+00 -4.46012795e-01 -1.21141396e-01 6.33244693e-01
4.02483553e-01 -8.39730024e-01 1.27790749e+00 3.38154912e-01
-7.77579486e-01 7.27614939e-01 3.49007189e-01 5.26656091e-01
7.61155307e-01 -1.31955683e+00 1.07238805e+00 1.11247373e+00
2.88328111e-01 4.32962567e-01 7.71401107e-01 -5.73842883e-01
-6.89239323e-01 -9.55045938e-01 7.92608738e-01 -4.99030650e-01
9.91441607e-01 -2.36064836e-01 -1.48816919e+00 8.26065898e-01
1.02571561e-03 4.85187441e-01 1.03092372e+00 4.71621573e-01
-4.94491786e-01 -2.27721646e-01 -1.51685226e+00 4.78735119e-01
1.00287759e+00 -3.98904681e-01 -9.24727261e-01 1.27645344e-01
-3.22781801e-02 -1.04989186e-01 -3.59787494e-01 6.25069261e-01
3.06378722e-01 -1.37372744e+00 8.36296558e-01 -1.03291500e+00
-4.77241818e-04 -2.14514527e-02 1.11197308e-01 -1.23440862e+00
-3.94102484e-01 -5.63663602e-01 -8.91098559e-01 7.80582368e-01
1.42387757e-02 -1.14586484e+00 9.69646513e-01 5.27717829e-01
-1.41277179e-01 -8.24022889e-01 -1.14309216e+00 -1.02447701e+00
-2.37669257e-04 -5.94923377e-01 4.77559924e-01 1.16383517e+00
2.77089387e-01 -2.33423561e-01 -2.00669050e-01 6.11904413e-02
9.10548627e-01 -2.58588970e-01 5.05308270e-01 -1.91438639e+00
1.15515769e-01 -5.28296292e-01 -1.24423409e+00 -5.45313478e-01
6.03160746e-02 -6.42593920e-01 -3.95707905e-01 -1.03226471e+00
-9.38933641e-02 -9.29262489e-02 -9.84978735e-01 7.22065091e-01
-2.12820530e-01 2.96271831e-01 -4.62705553e-01 -9.01283994e-02
-7.82326102e-01 3.65705281e-01 2.08344653e-01 1.87123299e-01
-1.51403964e-01 2.12999925e-01 -5.79129040e-01 1.17354989e+00
1.21310556e+00 -4.75711584e-01 -9.69202816e-03 -1.37911797e-01
6.20000027e-02 -7.69846678e-01 1.20533481e-01 -1.57660496e+00
2.23428741e-01 1.44797340e-01 5.93965530e-01 -6.13524556e-01
-1.51221856e-01 -8.49863112e-01 -6.03411198e-01 5.70971310e-01
-1.35728925e-01 5.10049701e-01 4.63491231e-01 7.34700501e-01
-4.28209692e-01 -2.99938228e-02 9.27953482e-01 -7.47897997e-02
-9.30897176e-01 4.55735922e-01 -7.83306122e-01 2.02625290e-01
1.18111181e+00 -5.55071011e-02 -1.36015862e-01 -1.80956349e-01
-8.11560571e-01 1.57747880e-01 8.50969478e-02 7.46224880e-01
6.04002953e-01 -1.16657758e+00 -6.19879603e-01 6.21086240e-01
4.59869385e-01 -2.93437868e-01 1.92310229e-01 7.44857311e-01
-3.42562199e-01 2.35297531e-01 -4.47715670e-01 -6.64183557e-01
-1.16574824e+00 4.93510962e-01 5.03771007e-01 -3.39460433e-01
-1.02183890e+00 5.47787488e-01 -1.62308306e-01 -3.46305728e-01
5.23146629e-01 1.38678357e-01 -1.94122002e-01 -2.76702315e-01
8.69575262e-01 7.45493591e-01 3.06413472e-01 -3.15123111e-01
-6.43123925e-01 1.50084756e-02 -5.68508208e-01 3.17110628e-01
1.34519577e+00 2.27193192e-01 -2.53057867e-01 7.72190034e-01
7.26901710e-01 -3.62567514e-01 -6.65564358e-01 -4.22398686e-01
5.45544446e-01 -4.21396255e-01 -1.12631749e-02 -7.54451871e-01
-1.12414336e+00 8.02441061e-01 9.88988578e-01 6.12644970e-01
1.13159454e+00 -7.83538595e-02 7.53082633e-01 6.67523205e-01
3.43302041e-01 -1.25514400e+00 1.10831439e-01 9.07634735e-01
4.67969716e-01 -1.54806852e+00 -1.35276467e-01 -6.09488450e-02
-3.13650995e-01 1.16483915e+00 8.39246929e-01 -4.97930110e-01
9.80502367e-01 3.20488811e-01 1.60379395e-01 -3.08696836e-01
-3.53223681e-01 -2.91479826e-01 2.62192100e-01 6.60279930e-01
5.41941226e-01 -2.43927598e-01 -2.15894040e-02 3.56212586e-01
1.91713780e-01 -3.02251965e-01 2.70053566e-01 1.18391156e+00
-6.27066612e-01 -1.24250770e+00 -2.95414358e-01 9.86526906e-01
-9.47640777e-01 7.44289234e-02 -5.18451214e-01 6.67854130e-01
1.57938851e-03 9.02564943e-01 6.30282164e-01 -1.53242230e-01
5.44926941e-01 6.62266612e-01 9.35257748e-02 -5.46123028e-01
-4.23416495e-01 -4.08309996e-01 -1.72765329e-01 -9.08244848e-01
-7.22135529e-02 -7.42597520e-01 -1.25832283e+00 -3.52443606e-01
-1.23702036e-02 2.03327879e-01 2.93137223e-01 1.26978743e+00
3.67008507e-01 3.93029392e-01 4.40284044e-01 -4.96960700e-01
-5.88824511e-01 -9.02787626e-01 -7.63854027e-01 3.94138306e-01
6.87185287e-01 -6.15836740e-01 -7.85221875e-01 -5.69158196e-01]
|
[7.494948387145996, 2.5547075271606445]
|
d522161b-e5a7-4c14-9f37-a1927df4785b
|
decomposition-based-generation-process-for
|
2204.03845
| null |
https://arxiv.org/abs/2204.03845v3
|
https://arxiv.org/pdf/2204.03845v3.pdf
|
Decompositional Generation Process for Instance-Dependent Partial Label Learning
|
Partial label learning (PLL) is a typical weakly supervised learning problem, where each training example is associated with a set of candidate labels among which only one is true. Most existing PLL approaches assume that the incorrect labels in each training example are randomly picked as the candidate labels and model the generation process of the candidate labels in a simple way. However, these approaches usually do not perform as well as expected due to the fact that the generation process of the candidate labels is always instance-dependent. Therefore, it deserves to be modeled in a refined way. In this paper, we consider instance-dependent PLL and assume that the generation process of the candidate labels could decompose into two sequential parts, where the correct label emerges first in the mind of the annotator but then the incorrect labels related to the feature are also selected with the correct label as candidate labels due to uncertainty of labeling. Motivated by this consideration, we propose a novel PLL method that performs Maximum A Posterior (MAP) based on an explicitly modeled generation process of candidate labels via decomposed probability distribution models. Extensive experiments on manually corrupted benchmark datasets and real-world datasets validate the effectiveness of the proposed method. Source code is available at https://github.com/palm-ml/idgp.
|
['Xin Geng', 'Ning Xu', 'Congyu Qiao']
|
2022-04-08
| null | null | null | null |
['partial-label-learning']
|
['methodology']
|
[ 2.90394604e-01 3.73098910e-01 -3.24410588e-01 -6.45581841e-01
-9.84328866e-01 -6.36721253e-01 6.63522124e-01 5.03250718e-01
-2.64424622e-01 8.69748116e-01 -2.88387984e-01 4.41851504e-02
-1.34202570e-01 -7.21656382e-01 -8.77084732e-01 -9.84477103e-01
4.62272823e-01 8.84244442e-01 3.16799611e-01 4.21962172e-01
1.27762243e-01 -2.26177201e-02 -1.67747724e+00 3.38291734e-01
8.61748934e-01 8.75602663e-01 6.62652403e-02 1.54759735e-01
-2.06474245e-01 8.42103601e-01 -6.53640509e-01 -3.68339151e-01
1.47381052e-01 -4.50465202e-01 -7.90109396e-01 3.17266494e-01
2.59807080e-01 9.89823863e-02 4.29040611e-01 1.35969996e+00
2.81969100e-01 -3.86985089e-03 7.35967278e-01 -1.37463784e+00
-1.57615408e-01 8.48471105e-01 -5.01480162e-01 -3.49659652e-01
1.26094505e-01 -1.01490535e-01 1.19214964e+00 -9.53515232e-01
4.75373030e-01 9.93071735e-01 4.26757783e-01 5.10854304e-01
-1.21990812e+00 -7.55658865e-01 3.52409184e-01 2.20800862e-01
-1.55625403e+00 -1.80360019e-01 9.31220174e-01 -4.62693691e-01
6.71468079e-02 9.69188288e-02 2.16795594e-01 8.89665842e-01
-6.94268420e-02 9.64233875e-01 1.42460668e+00 -5.87867081e-01
5.06034076e-01 6.61069095e-01 6.78604126e-01 6.04613662e-01
2.32313633e-01 -4.56197858e-02 -3.93924981e-01 -4.37939495e-01
1.10996731e-01 -2.32962593e-01 -3.24268371e-01 -4.10135359e-01
-1.09066045e+00 6.93513870e-01 1.09360941e-01 2.35280171e-01
-3.17922503e-01 -2.64672898e-02 1.93263799e-01 -5.73583581e-02
5.08787692e-01 8.25663581e-02 -6.07780278e-01 3.05652320e-01
-1.04249656e+00 1.92924351e-01 8.71038377e-01 1.00044334e+00
9.11809981e-01 -4.58117276e-01 -2.23878667e-01 7.85355687e-01
6.29048407e-01 1.78925812e-01 3.46806526e-01 -7.25044847e-01
1.95631728e-01 5.98730981e-01 5.15959561e-01 -7.45972753e-01
-2.74246603e-01 -6.60796583e-01 -5.47537386e-01 1.65360212e-01
7.48547673e-01 -8.65972936e-02 -8.29079926e-01 1.89716578e+00
6.88237071e-01 4.40095991e-01 6.17080145e-02 7.51884341e-01
7.54558086e-01 8.12491000e-01 2.63569564e-01 -5.29100478e-01
1.26764083e+00 -1.02724481e+00 -8.57417166e-01 -9.69583541e-02
4.69613612e-01 -7.59209096e-01 8.62013102e-01 6.67119324e-01
-6.82544291e-01 -6.72288656e-01 -7.77683675e-01 4.01209205e-01
-1.20509833e-01 6.86158538e-01 2.07246870e-01 4.49862689e-01
-5.92333674e-01 4.38972443e-01 -7.05325663e-01 -1.40236944e-01
2.05723673e-01 2.54489541e-01 -1.19994558e-01 -2.00284328e-02
-1.16972339e+00 6.13809645e-01 7.25169539e-01 2.60878384e-01
-8.81040335e-01 -2.56661803e-01 -4.81294841e-01 1.08053768e-02
7.17481136e-01 -1.98874727e-01 1.33541191e+00 -1.14296782e+00
-1.22032571e+00 7.67163277e-01 -3.56387287e-01 -2.25046873e-01
7.15505064e-01 -2.34395489e-02 -2.39230782e-01 -7.86441714e-02
2.39272267e-01 5.24447143e-01 8.57552528e-01 -1.72345817e+00
-9.97506082e-01 -1.36850148e-01 1.07100748e-01 1.59641892e-01
-1.07247885e-02 -1.91932648e-01 -4.09961194e-01 -4.81768936e-01
2.02453256e-01 -1.13056672e+00 2.31052451e-02 -1.54629812e-01
-5.90172231e-01 -7.21064806e-01 6.27810121e-01 -3.12838644e-01
1.14835572e+00 -2.15035319e+00 -1.46468744e-01 2.44189903e-01
5.45420609e-02 1.39072552e-01 1.63291425e-01 2.07027972e-01
-7.19457343e-02 5.87615371e-02 -3.31974387e-01 -4.95252848e-01
1.23936042e-01 2.20722318e-01 -2.56157100e-01 4.91762310e-01
6.10117428e-02 3.05149376e-01 -1.07300723e+00 -6.47512138e-01
1.55969784e-01 3.64284188e-01 -6.85157627e-02 3.09168786e-01
-5.59628487e-01 6.80104375e-01 -6.85166359e-01 5.48340619e-01
8.19444180e-01 -4.36004162e-01 2.50379413e-01 -3.66015762e-01
8.34718868e-02 4.91147041e-02 -1.63252366e+00 1.14461052e+00
-1.65348053e-01 -1.31842509e-01 -1.77536309e-01 -1.06947136e+00
9.26134229e-01 5.39142370e-01 4.55981553e-01 -6.00337982e-04
9.52487737e-02 4.24642742e-01 -1.38016284e-01 -3.65942478e-01
1.11403339e-01 -4.39143717e-01 -1.18463963e-01 6.55168951e-01
1.34224087e-01 1.94984153e-02 2.54239112e-01 1.04954906e-01
6.10292912e-01 4.90431815e-01 3.96189511e-01 -1.96242929e-01
7.96589911e-01 -2.59204097e-02 9.52790737e-01 7.43011296e-01
-2.47031301e-01 6.05622768e-01 5.41510105e-01 -2.57048428e-01
-6.94573343e-01 -7.64328539e-01 -4.06151950e-01 8.64914298e-01
3.35034817e-01 -3.99429858e-01 -7.97015250e-01 -1.22807431e+00
-3.21454108e-01 1.07675624e+00 -5.70359111e-01 3.08649819e-02
-2.08512858e-01 -7.88812935e-01 2.06955671e-01 1.34003237e-01
4.06072497e-01 -1.23138022e+00 -4.55178529e-01 3.16256016e-01
-3.39830071e-01 -9.16376829e-01 -1.24097750e-01 4.87021506e-01
-5.23620307e-01 -1.20519161e+00 -3.23661804e-01 -7.89476633e-01
1.01472509e+00 -1.97033897e-01 9.31490958e-01 1.18849382e-01
3.18927914e-01 9.28243399e-02 -5.59099674e-01 -3.50627691e-01
-7.27075934e-01 -4.27613966e-02 6.54648840e-02 5.51136076e-01
3.68101716e-01 -1.70724928e-01 -2.08908737e-01 4.33309495e-01
-8.59809279e-01 9.08404663e-02 4.50318962e-01 1.02168977e+00
1.03295696e+00 6.85898900e-01 7.22429156e-01 -1.48827958e+00
3.41572076e-01 -7.16355920e-01 -6.49319649e-01 5.97564459e-01
-6.44844949e-01 3.48982550e-02 8.59941840e-01 -5.87430358e-01
-1.30714691e+00 4.07040387e-01 -4.07000296e-02 -1.76864147e-01
-6.47945940e-01 5.91073334e-01 -4.58885491e-01 3.43630701e-01
4.15950030e-01 1.34545997e-01 -4.07485247e-01 -5.57713032e-01
9.42512602e-02 6.45119488e-01 1.12778448e-01 -8.67562830e-01
6.79670811e-01 2.88723022e-01 -7.38562271e-02 -2.73119718e-01
-1.31461489e+00 -3.42851341e-01 -6.30094230e-01 -3.84914428e-01
4.87583667e-01 -7.88990319e-01 -2.21522391e-01 5.33695936e-01
-1.01204467e+00 -1.42466664e-01 -2.87534565e-01 3.83003592e-01
-4.00153250e-01 2.31589064e-01 -3.69751453e-01 -8.09487700e-01
-5.38029224e-02 -1.45706129e+00 9.78196144e-01 3.50318581e-01
-2.18030483e-01 -8.71146560e-01 -7.87307993e-02 3.84916782e-01
-1.59881741e-01 2.00167343e-01 9.67984676e-01 -1.05177748e+00
-4.94243234e-01 -5.11749029e-01 -4.90282625e-02 4.71217602e-01
2.73903012e-01 3.67550775e-02 -1.13656628e+00 -2.28481978e-01
1.93993270e-01 -4.42753315e-01 6.27559304e-01 3.67788851e-01
9.85318542e-01 -2.21078321e-01 -4.16847587e-01 -7.49028549e-02
1.45761275e+00 1.61852077e-01 8.12778175e-02 7.00558573e-02
5.71314692e-01 7.84958363e-01 1.11451638e+00 4.82199192e-01
3.75799596e-01 6.62829280e-01 4.55004364e-01 2.45736226e-01
6.68954626e-02 -2.73095280e-01 2.76798338e-01 7.17076957e-01
2.24385634e-01 -5.39655924e-01 -7.59439647e-01 4.05253679e-01
-2.10290337e+00 -7.00250685e-01 -2.78947711e-01 2.36439562e+00
1.01585913e+00 2.93970913e-01 -1.45749435e-01 3.70994151e-01
1.07420933e+00 -8.57599080e-02 -5.05690932e-01 7.78505802e-02
5.24254814e-02 -7.58524165e-02 1.05063789e-01 5.52459776e-01
-1.24731386e+00 7.15468347e-01 4.54784060e+00 1.07645619e+00
-8.71756613e-01 3.68521065e-01 8.18284929e-01 1.90578610e-01
-2.69519389e-01 3.61353815e-01 -1.18255305e+00 7.38378525e-01
6.19610786e-01 -7.10686967e-02 -5.68531118e-02 8.17348838e-01
1.19688042e-01 -4.21025485e-01 -1.22575641e+00 5.79683006e-01
-3.67209949e-02 -7.25854933e-01 -7.00988770e-02 -9.16770399e-02
9.04775441e-01 -3.53249192e-01 -6.16314337e-02 2.32363999e-01
3.03081810e-01 -6.67397559e-01 1.17236173e+00 6.64959073e-01
4.35300499e-01 -8.29988956e-01 1.02415955e+00 8.95316899e-01
-1.05565929e+00 -1.53290909e-02 -3.35252643e-01 1.29793748e-01
1.28027573e-01 1.09372139e+00 -8.45266283e-01 6.10399485e-01
4.16295230e-01 5.69341600e-01 -5.52387297e-01 1.01790130e+00
-7.79880702e-01 1.11926043e+00 -4.46355939e-01 6.28767237e-02
2.84984764e-02 -3.04405726e-02 3.72296870e-01 9.22730267e-01
3.34071487e-01 -3.70760709e-02 5.23851752e-01 8.83997262e-01
-6.84979036e-02 2.82518655e-01 -1.89370230e-01 3.29124063e-01
6.49619639e-01 1.36129951e+00 -9.95707095e-01 -4.09490734e-01
-3.71827006e-01 5.60530663e-01 2.92434543e-01 3.08612496e-01
-9.87902641e-01 1.60087928e-01 -1.62583441e-01 2.46862806e-02
9.86701474e-02 3.34262520e-01 -6.57602325e-02 -1.05068922e+00
1.64826363e-01 -7.20376074e-01 5.03568411e-01 -5.29599309e-01
-1.48723865e+00 5.55690467e-01 1.34371206e-01 -1.47125399e+00
-1.34695753e-01 -2.00848863e-01 -3.72721732e-01 7.18219817e-01
-1.55927014e+00 -1.17628944e+00 -2.20470041e-01 3.28692704e-01
5.89972079e-01 2.30990857e-01 7.56479502e-01 1.29861310e-01
-5.04286170e-01 4.49700356e-01 -4.61768322e-02 -1.57195687e-01
8.77000690e-01 -1.27224576e+00 -2.95542091e-01 7.61868715e-01
2.99467862e-01 3.84657383e-01 8.76001954e-01 -6.74780846e-01
-5.69952369e-01 -1.25318134e+00 1.18535829e+00 -2.94923306e-01
4.02145326e-01 -1.65935546e-01 -1.08862007e+00 6.72156990e-01
5.54369204e-02 2.22080678e-01 7.66254187e-01 -1.76053196e-01
-1.36762500e-01 -9.31131095e-02 -1.23179650e+00 1.56369507e-01
4.30529982e-01 -1.41462609e-01 -4.97982949e-01 5.43458819e-01
3.24029446e-01 -2.39181593e-01 -6.00889325e-01 6.29521668e-01
2.07787856e-01 -7.67175615e-01 4.47373897e-01 1.58413418e-03
1.88722894e-01 -9.08958256e-01 9.86458268e-03 -1.18952930e+00
-1.20879419e-01 5.55438884e-02 -1.44215703e-01 1.66730011e+00
5.87902308e-01 -4.85797971e-01 6.95765138e-01 7.12921858e-01
1.52699515e-01 -8.06660712e-01 -8.77646804e-01 -6.00032508e-01
-1.77980229e-01 -4.22485888e-01 5.34529328e-01 1.03533328e+00
-9.93721858e-02 2.59706020e-01 -5.21054983e-01 4.70713228e-01
8.50828648e-01 4.26104277e-01 4.34114844e-01 -1.45667124e+00
-4.10266072e-01 -1.21323340e-01 2.02002749e-02 -8.97003531e-01
3.92793596e-01 -9.21281576e-01 4.86403048e-01 -1.44595110e+00
3.99949014e-01 -9.83143747e-01 -5.38561940e-01 7.23993063e-01
-3.91214013e-01 1.14751771e-01 -4.80245240e-02 4.30327058e-01
-7.22878575e-01 4.52081740e-01 9.62395012e-01 -2.80172508e-02
5.48209175e-02 5.43235719e-01 -4.85505998e-01 9.83431935e-01
8.61493289e-01 -8.62347960e-01 -5.54881275e-01 -1.42147332e-01
1.94063410e-01 1.92307141e-02 2.21643493e-01 -8.39894295e-01
3.01071107e-01 -5.22920638e-02 7.43671730e-02 -5.44420719e-01
7.59807155e-02 -1.10270834e+00 3.62994522e-01 2.66915530e-01
-6.47186399e-01 -5.56155682e-01 -1.93184257e-01 6.69849813e-01
-2.77883828e-01 -7.55164564e-01 9.52881813e-01 -8.41181725e-02
-5.79607308e-01 1.33971468e-01 -2.39609957e-01 -1.73728347e-01
1.19047797e+00 6.83955401e-02 -8.24790895e-02 -2.68021882e-01
-1.05907786e+00 2.48237506e-01 3.30139518e-01 5.70530817e-02
3.28619838e-01 -1.23800611e+00 -8.00023854e-01 -5.65498061e-02
2.68673867e-01 3.01793933e-01 1.15639724e-01 6.87189937e-01
3.43737826e-02 1.77831978e-01 4.09107238e-01 -6.37838900e-01
-1.23689449e+00 6.23129666e-01 1.39553115e-01 -4.34774309e-01
-2.79044271e-01 9.03127015e-01 2.33542323e-01 -6.35777652e-01
3.68679732e-01 -1.02902045e-02 -5.42622328e-01 1.84623554e-01
3.11223239e-01 8.67241621e-02 -6.74566999e-02 -9.02366161e-01
-2.54291892e-01 5.62386572e-01 -3.98861356e-02 8.70632976e-02
1.08404183e+00 -1.82051286e-01 -2.25761592e-01 7.92546690e-01
7.07960784e-01 3.63198593e-02 -1.28640437e+00 -5.15577078e-01
2.24677965e-01 -3.10735196e-01 -8.75703245e-02 -9.29240465e-01
-1.05606675e+00 6.09231889e-01 4.53512311e-01 1.50470257e-01
1.14152813e+00 1.71191365e-01 3.43270987e-01 4.75065410e-02
7.14325726e-01 -1.08139610e+00 -1.17910087e-01 1.63754076e-01
4.81161892e-01 -1.26818633e+00 -1.12641335e-01 -6.61510825e-01
-5.94447911e-01 1.07350898e+00 8.53382707e-01 2.92533301e-02
7.97247469e-01 1.49772197e-01 2.41262577e-02 1.39422733e-02
-8.56754601e-01 8.06836188e-02 2.20578954e-01 1.06494203e-01
4.63008523e-01 1.74912065e-01 -6.78485036e-01 9.47559595e-01
2.41522372e-01 4.40416783e-02 4.77182835e-01 8.58944654e-01
-4.32383567e-01 -1.39102459e+00 -4.06242013e-01 2.56433189e-01
-3.78237516e-01 5.09751290e-02 -1.17199533e-01 3.98250461e-01
8.30796242e-01 1.04178166e+00 -2.62580812e-01 -2.67773956e-01
1.20862693e-01 3.82093906e-01 3.00579488e-01 -1.01552856e+00
-3.80288422e-01 3.25726062e-01 5.57604656e-02 -1.48753896e-01
-7.15085685e-01 -8.34294558e-01 -1.41625106e+00 2.97230750e-01
-7.84889877e-01 5.45800328e-01 4.10023242e-01 1.15360188e+00
-3.18355346e-03 2.93857843e-01 7.35424876e-01 -5.39712727e-01
-6.17935300e-01 -1.02379215e+00 -7.87005186e-01 4.60989237e-01
3.68784778e-02 -7.99776077e-01 -6.08871043e-01 2.94084132e-01]
|
[9.433905601501465, 4.044144153594971]
|
4c9a5741-1044-4025-878c-fd0e73671824
|
hybrid-loss-for-learning-single-image-based
|
1812.07134
| null |
http://arxiv.org/abs/1812.07134v1
|
http://arxiv.org/pdf/1812.07134v1.pdf
|
Hybrid Loss for Learning Single-Image-based HDR Reconstruction
|
This paper tackles high-dynamic-range (HDR) image reconstruction given only a
single low-dynamic-range (LDR) image as input. While the existing methods focus
on minimizing the mean-squared-error (MSE) between the target and reconstructed
images, we minimize a hybrid loss that consists of perceptual and adversarial
losses in addition to HDR-reconstruction loss. The reconstruction loss instead
of MSE is more suitable for HDR since it puts more weight on both over- and
under- exposed areas. It makes the reconstruction faithful to the input.
Perceptual loss enables the networks to utilize knowledge about objects and
image structure for recovering the intensity gradients of saturated and grossly
quantized areas. Adversarial loss helps to select the most plausible appearance
from multiple solutions. The hybrid loss that combines all the three losses is
calculated in logarithmic space of image intensity so that the outputs retain a
large dynamic range and meanwhile the learning becomes tractable. Comparative
experiments conducted with other state-of-the-art methods demonstrated that our
method produces a leap in image quality.
|
['ShaoDi You', 'Rei Kawakami', 'Kenta Moriwaki', 'Takeshi Naemura', 'Ryota Yoshihashi']
|
2018-12-18
| null | null | null | null |
['single-image-based-hdr-reconstruction', 'hdr-reconstruction']
|
['computer-vision', 'computer-vision']
|
[ 6.17564857e-01 -3.87200639e-02 1.32535025e-02 -2.80477524e-01
-8.57590318e-01 -1.94948897e-01 1.36843905e-01 -2.53157288e-01
-5.19982159e-01 5.98037958e-01 -1.81726396e-01 1.53397210e-02
2.81487219e-03 -8.68082464e-01 -8.18492591e-01 -9.03660297e-01
1.13396101e-01 -2.27221191e-01 4.39994663e-01 -1.59967259e-01
2.00515136e-01 7.71710336e-01 -1.48591447e+00 1.19549878e-01
9.64300573e-01 1.34925508e+00 5.72315097e-01 5.39391816e-01
2.66477495e-01 9.67258215e-01 -3.70577484e-01 -4.72736865e-01
6.53118789e-01 -4.27275181e-01 -3.18467259e-01 4.82658669e-02
4.89820778e-01 -5.00107288e-01 -7.25520432e-01 1.41459954e+00
6.11173809e-01 1.66607231e-01 5.96313596e-01 -9.70076561e-01
-8.93653929e-01 3.91628146e-02 -8.99012625e-01 2.59686679e-01
2.50277102e-01 3.59243333e-01 7.47702956e-01 -7.72018850e-01
5.74607313e-01 1.26628745e+00 5.08428752e-01 3.43527764e-01
-1.41009545e+00 -6.28197849e-01 3.77756767e-02 2.18389481e-01
-1.36682439e+00 -3.38937551e-01 8.78257275e-01 -4.18741889e-02
6.02006316e-01 3.16042393e-01 4.11929190e-01 6.34922802e-01
4.50289160e-01 4.40616578e-01 1.31779552e+00 -3.94298226e-01
-5.51742315e-02 1.14586979e-01 -5.47183931e-01 7.15101957e-01
-9.20542926e-02 3.82282585e-01 -2.50305831e-01 2.92468548e-01
1.17541897e+00 1.87172115e-01 -6.92006707e-01 -3.93782914e-01
-1.04324210e+00 5.51573932e-01 8.17298889e-01 1.74742222e-01
-3.41509670e-01 4.70984057e-02 -5.36064617e-02 4.38787043e-01
1.72852710e-01 3.06248665e-01 -1.66601449e-01 3.63889456e-01
-8.44459891e-01 -7.92879984e-02 2.47191682e-01 5.72582841e-01
9.24223006e-01 2.79824257e-01 -6.26480803e-02 1.06130993e+00
3.24108601e-01 7.22234070e-01 2.36945719e-01 -1.22507346e+00
4.71118093e-01 3.95553589e-01 1.41032085e-01 -1.24473524e+00
5.98414475e-03 -4.00010288e-01 -1.23703134e+00 7.33891308e-01
3.67842644e-01 2.31977433e-01 -1.02708173e+00 1.76721787e+00
9.18825790e-02 1.51726268e-02 1.05767421e-01 1.25118649e+00
5.38245142e-01 9.52728033e-01 5.03928587e-02 -4.06556845e-01
9.73182321e-01 -8.31360221e-01 -7.48972714e-01 -4.24706787e-01
-4.02691543e-01 -7.81956673e-01 1.24701297e+00 3.46405804e-01
-1.64044929e+00 -8.31937432e-01 -1.30270696e+00 -2.91180253e-01
-1.43034592e-01 -1.76352099e-01 8.18547904e-02 2.87328064e-01
-1.10139453e+00 7.83451676e-01 -3.84839982e-01 2.29854301e-01
3.09457451e-01 2.53331631e-01 -3.69255930e-01 -4.06449586e-01
-1.18514645e+00 9.65475261e-01 1.75044730e-01 1.00426257e-01
-8.37352872e-01 -7.23113418e-01 -6.93742275e-01 -2.16664430e-02
2.12049469e-01 -5.05805492e-01 6.89668894e-01 -1.32247412e+00
-1.59542871e+00 9.33394790e-01 2.15991870e-01 -3.41099381e-01
7.37212121e-01 1.65497616e-01 -4.58189338e-01 4.99670327e-01
-2.02491403e-01 7.55197525e-01 1.26366675e+00 -1.64733422e+00
-5.42201221e-01 -3.57278913e-01 -1.32678330e-01 4.00353640e-01
-7.68340081e-02 -2.07387403e-01 -4.21896666e-01 -8.25933933e-01
3.03056747e-01 -6.74253166e-01 -1.21011771e-01 6.01124287e-01
-1.77868232e-01 3.65742117e-01 8.31627250e-01 -8.45824957e-01
1.11615348e+00 -2.24044013e+00 2.35759452e-01 8.85038748e-02
1.84270829e-01 2.16854632e-01 -2.56731272e-01 1.28638297e-01
-1.73524022e-01 -9.74441245e-02 -3.66239250e-01 -4.78079394e-02
-2.70665795e-01 1.40229106e-01 -4.78737801e-01 7.62149870e-01
4.05388996e-02 8.50807309e-01 -7.66907334e-01 -6.27178013e-01
4.87108022e-01 7.76491880e-01 -1.59947246e-01 4.34471488e-01
1.11196123e-01 4.60248262e-01 -1.25239685e-01 4.27873731e-01
1.01913643e+00 -1.67065635e-01 -1.01009190e-01 -4.96838361e-01
3.71230803e-02 -3.81882131e-01 -1.00986624e+00 1.51283312e+00
-6.89942420e-01 4.63437766e-01 1.31102055e-01 -7.25362420e-01
1.09440708e+00 -1.18979052e-01 4.00137752e-01 -1.13491654e+00
5.59774274e-03 2.03977928e-01 -1.60287917e-01 -3.35295707e-01
2.99623787e-01 -4.42050695e-01 3.40596586e-02 1.84955612e-01
-2.67746925e-01 -2.53648341e-01 -1.74528033e-01 1.96986124e-02
6.79140449e-01 4.93601784e-02 3.23664367e-01 1.11504130e-01
5.78291714e-01 -6.08326972e-01 6.16770864e-01 5.97510338e-01
-2.11523160e-01 9.32178795e-01 2.49053881e-01 -2.38639802e-01
-1.33330655e+00 -1.65139580e+00 -1.70576036e-01 8.50939214e-01
7.01485038e-01 3.86726767e-01 -3.98002863e-01 -4.09407228e-01
-2.40560696e-01 5.60721219e-01 -4.49523628e-01 -2.96471149e-01
-6.19370818e-01 -3.67920995e-01 3.43009889e-01 1.99611053e-01
9.47278678e-01 -1.18512022e+00 -6.03509903e-01 -8.90561491e-02
-4.17611957e-01 -9.56904352e-01 -6.43888474e-01 1.24407589e-01
-8.70321512e-01 -8.12382519e-01 -9.20197308e-01 -8.59526098e-01
7.07322896e-01 3.58118683e-01 1.10414708e+00 4.78140563e-02
-4.67636824e-01 2.50080414e-02 -2.33669087e-01 2.31978241e-02
-4.94259924e-01 -4.50532764e-01 -3.09585094e-01 1.37886167e-01
-2.72532105e-01 -6.90208793e-01 -8.96334052e-01 4.07805651e-01
-1.10981250e+00 1.55856786e-02 9.04610336e-01 9.24833655e-01
1.08921611e+00 4.72242057e-01 4.29926664e-01 -5.39529860e-01
1.15161210e-01 -1.75376326e-01 -6.67259693e-01 3.92321497e-01
-7.65395164e-01 -5.23967072e-02 9.01945293e-01 -6.08293474e-01
-1.11908996e+00 8.86427686e-02 -1.82354674e-01 -7.60151088e-01
2.29094431e-01 -3.49321753e-01 -2.57019013e-01 -3.76683444e-01
3.89516354e-01 6.04462981e-01 9.21045467e-02 -2.16994628e-01
3.46788287e-01 4.66326356e-01 7.92051375e-01 -1.82294294e-01
9.86438215e-01 5.24079382e-01 1.11316144e-01 -6.82200909e-01
-7.58636534e-01 -2.01144502e-01 -3.06895614e-01 -3.24891567e-01
8.14445019e-01 -8.93009245e-01 -6.76202357e-01 6.20529771e-01
-7.71886528e-01 -2.77761370e-01 -5.48412561e-01 2.43292585e-01
-7.38453507e-01 4.49970663e-01 -5.74492753e-01 -7.51115620e-01
-3.30850363e-01 -1.16560924e+00 9.42583621e-01 3.27649564e-01
5.40957212e-01 -7.55061686e-01 -2.71602154e-01 1.37128279e-01
5.09477198e-01 4.51067626e-01 9.75688756e-01 2.65338302e-01
-8.19299579e-01 -2.02985425e-02 -5.18063962e-01 7.64439940e-01
-5.92620671e-02 -2.66174197e-01 -9.20781910e-01 -5.20747185e-01
2.53022701e-01 -4.76307303e-01 9.52087402e-01 4.52910870e-01
1.48556674e+00 -5.80081522e-01 1.62310198e-01 9.58238661e-01
1.98396301e+00 1.41083568e-01 1.39752769e+00 2.48602331e-01
6.19733036e-01 5.16783178e-01 7.94860423e-01 1.65622458e-01
3.65818329e-02 8.07038069e-01 6.20338917e-01 -4.98226017e-01
-3.97052020e-01 -3.37880015e-01 4.02680427e-01 4.34459627e-01
1.34916827e-01 -3.64241332e-01 -3.50498587e-01 2.32658669e-01
-1.38974380e+00 -1.16349053e+00 3.37394208e-01 2.39322495e+00
1.08562481e+00 7.19842911e-02 -1.17938600e-01 2.52994090e-01
8.06470335e-01 4.52507019e-01 -9.84465003e-01 -2.84812808e-01
-4.12291825e-01 1.97696015e-02 6.47791505e-01 4.61811662e-01
-9.39410925e-01 5.20096064e-01 6.43600225e+00 1.16676116e+00
-1.27924621e+00 -2.78966855e-02 1.07228959e+00 8.41027964e-03
-3.48804444e-01 -2.94124126e-01 -4.04027849e-01 6.37546599e-01
6.52895212e-01 6.35960847e-02 7.75476754e-01 6.75696969e-01
5.97844869e-02 -1.33357763e-01 -6.87159300e-01 1.17993867e+00
1.01564545e-02 -1.00238347e+00 3.33868451e-02 1.08899392e-01
9.13360238e-01 -1.40372634e-01 6.62931383e-01 -1.61183551e-01
6.06134022e-03 -1.20432603e+00 9.27079201e-01 7.76080310e-01
1.33447492e+00 -9.34848249e-01 5.42890429e-01 1.91770568e-01
-1.14749348e+00 -4.05278087e-01 -6.52844846e-01 4.28048402e-01
7.26308897e-02 5.93993425e-01 -1.72974825e-01 3.25084209e-01
8.04114163e-01 5.85201323e-01 -4.73720789e-01 7.54931629e-01
-2.57063419e-01 -1.32232204e-01 -7.07632676e-02 5.22277474e-01
-3.46058793e-02 -3.49967539e-01 5.18723965e-01 9.88039255e-01
1.89733684e-01 1.31649956e-01 2.02007473e-01 9.17958915e-01
-2.13425159e-01 -5.79362595e-03 -5.56657970e-01 2.83904552e-01
4.77272928e-01 1.20524967e+00 -5.97276688e-01 -2.60372721e-02
-1.30333036e-01 1.33186209e+00 2.50740826e-01 4.15542752e-01
-8.44436944e-01 -4.30810720e-01 3.35284233e-01 5.10592699e-01
3.70677829e-01 4.85214144e-02 -2.88030118e-01 -9.11487162e-01
3.33290733e-02 -8.57074618e-01 1.89786971e-01 -9.81742740e-01
-1.39881814e+00 6.83527291e-01 -2.27697864e-01 -1.48847461e+00
3.17930069e-04 -3.09503525e-01 -3.85723799e-01 9.30767596e-01
-1.92560410e+00 -9.17428851e-01 -4.47071910e-01 6.35038078e-01
4.87802207e-01 5.37609830e-02 4.04298037e-01 4.81538922e-01
-3.66978973e-01 7.96477973e-01 1.50225520e-01 -1.58902556e-01
7.73779333e-01 -1.18212068e+00 -3.67302671e-02 9.80214179e-01
-2.93415815e-01 3.12105417e-01 6.41164839e-01 -3.36536288e-01
-1.18835342e+00 -1.10261726e+00 3.46676469e-01 1.38682231e-01
3.01777869e-01 5.55758439e-02 -1.05073678e+00 2.81640738e-01
1.03581414e-01 3.04024458e-01 1.43514097e-01 -7.39211321e-01
-5.18332303e-01 -6.15894556e-01 -1.62464535e+00 5.33359349e-01
7.87730277e-01 -6.04160011e-01 -8.01770017e-02 -6.65309057e-02
7.94492900e-01 -3.23676407e-01 -1.19367552e+00 3.25465858e-01
5.64511597e-01 -1.28291678e+00 1.42886698e+00 1.69934690e-01
6.10366762e-01 -5.57242930e-01 -3.77107680e-01 -1.08709180e+00
-3.40707511e-01 -4.77045834e-01 -8.78003463e-02 1.03422749e+00
1.89421579e-01 -4.10184979e-01 2.53910840e-01 3.02504569e-01
1.76094025e-01 -1.00933838e+00 -8.94931138e-01 -5.96898913e-01
-8.38123187e-02 5.37890717e-02 2.09736198e-01 5.44636190e-01
-5.95669150e-01 1.46846443e-01 -7.01188564e-01 1.13972172e-01
1.02481008e+00 2.03605831e-01 3.10217947e-01 -8.35318685e-01
-3.89979005e-01 -2.21261740e-01 -3.67080837e-01 -1.12256849e+00
1.02441087e-02 -6.97400391e-01 3.74300033e-01 -1.30203903e+00
3.54925841e-01 -5.61115444e-01 -4.40335959e-01 2.48033434e-01
-1.56920850e-01 7.99178779e-01 3.51627588e-01 3.51898700e-01
-5.00005126e-01 5.64412594e-01 1.80532670e+00 -1.01611845e-01
-1.27019957e-01 -1.57787055e-01 -6.63813531e-01 5.61564386e-01
6.96533322e-01 -4.18549746e-01 -4.81104583e-01 -3.28215420e-01
2.54836306e-02 2.60174483e-01 5.40834785e-01 -9.54312205e-01
6.26690760e-02 -2.32795790e-01 9.06429410e-01 -4.35034573e-01
4.34730560e-01 -9.43152785e-01 2.79704005e-01 4.14251298e-01
-4.91279483e-01 -1.72003895e-01 -1.59110472e-01 7.92304039e-01
-2.03385398e-01 -5.71026541e-02 1.93261659e+00 -2.14414597e-01
-7.09338188e-01 3.62502098e-01 1.63225010e-01 -4.25966792e-02
1.10858691e+00 -5.29970109e-01 -2.22896650e-01 -5.46240807e-01
-4.07731771e-01 -2.15607807e-01 8.18981588e-01 2.90115386e-01
1.10356259e+00 -1.22039592e+00 -6.80818737e-01 3.72312754e-01
-1.98012069e-01 -1.43705666e-01 5.99173307e-01 5.51430941e-01
-6.90175772e-01 -3.56438458e-02 -6.16290510e-01 -4.09032434e-01
-1.13959157e+00 7.52302229e-01 5.22147894e-01 -4.53008711e-01
-9.14368987e-01 6.45415843e-01 3.23540390e-01 -1.59360375e-02
3.77665251e-01 1.98361903e-01 -3.13415471e-03 -3.29289168e-01
6.91469729e-01 4.69148040e-01 -3.66521746e-01 -8.14964175e-01
-1.48310512e-01 8.68359804e-01 -8.26741010e-02 -1.47735298e-01
1.20968187e+00 -6.12275183e-01 -5.08697331e-02 3.32543463e-01
1.53195131e+00 -1.20065875e-01 -1.81066155e+00 -2.21086174e-01
-6.91716552e-01 -1.02044117e+00 4.77122128e-01 -9.66475844e-01
-1.42008698e+00 8.29489470e-01 1.10775757e+00 7.27256015e-02
1.82186544e+00 -2.94326633e-01 9.57208693e-01 -4.31478396e-02
3.62188816e-01 -9.14435983e-01 5.33192754e-01 2.84416433e-02
9.88436103e-01 -1.10968292e+00 1.64432317e-01 -2.38376409e-01
-7.83089101e-01 9.74798203e-01 5.45964837e-01 -4.06631768e-01
4.83121067e-01 2.03751177e-01 3.59908193e-02 3.16764981e-01
-4.01348710e-01 -2.68707741e-02 3.12154680e-01 7.92608082e-01
-9.15828273e-02 -2.85238773e-01 -9.02239308e-02 -8.63560736e-02
1.55903056e-01 -1.35038033e-01 4.63262796e-01 4.18483496e-01
-5.80413640e-01 -7.87628174e-01 -2.92132467e-01 2.96952397e-01
-6.59227729e-01 -3.48306149e-02 1.76139012e-01 6.08171642e-01
2.83011079e-01 8.92718613e-01 5.27056195e-02 -2.63359606e-01
3.37340653e-01 -6.01556063e-01 6.45925224e-01 -1.20012335e-01
-2.83074528e-01 1.24202348e-01 -5.83224118e-01 -7.93542147e-01
-3.65749478e-01 -7.73652345e-02 -1.15864730e+00 -4.93894130e-01
-1.41770959e-01 -3.17899823e-01 6.38725936e-01 3.48479807e-01
3.99150029e-02 6.20358467e-01 1.12461793e+00 -9.30123746e-01
-6.18753076e-01 -5.61241448e-01 -1.01405549e+00 5.22436380e-01
6.14874721e-01 -3.76775175e-01 -6.21454000e-01 9.39898267e-02]
|
[10.98436164855957, -2.159470319747925]
|
9951263d-cf14-40ab-a1cd-816f2dc67d8f
|
inverting-ransac-global-model-detection-via
| null | null |
http://openaccess.thecvf.com/content_cvpr_2015/html/Litman_Inverting_RANSAC_Global_2015_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2015/papers/Litman_Inverting_RANSAC_Global_2015_CVPR_paper.pdf
|
Inverting RANSAC: Global Model Detection via Inlier Rate Estimation
|
This work presents a novel approach for detecting inliers in a given set of correspondences (matches). It does so without explicitly identifying any consensus set, based on a method for inlier rate estimation (IRE). Given such an estimator for the inlier rate, we also present an algorithm that detects a globally optimal transformation. We provide a theoretical analysis of the IRE method using a stochastic generative model on the continuous spaces of matches and transformations. This model allows rigorous investigation of the limits of our IRE method for the case of 2D-translation, further giving bounds and insights for the more general case. Our theoretical analysis is validated empirically and is shown to hold in practice for the more general case of 2D-affinities. In addition, we show that the combined framework works on challenging cases of 2D-homography estimation, with very few and possibly noisy inliers, where RANSAC generally fails.
|
['Simon Korman', 'Shai Avidan', 'Roee Litman', 'Alexander Bronstein']
|
2015-06-01
| null | null | null |
cvpr-2015-6
|
['homography-estimation']
|
['computer-vision']
|
[ 1.16405822e-01 2.96906251e-02 -5.28799221e-02 1.34886548e-01
-1.05362678e+00 -7.13406801e-01 6.02794707e-01 -9.33000743e-02
3.99426147e-02 5.55394173e-01 1.57948554e-01 -1.21535279e-01
-3.21084410e-01 -5.42952180e-01 -8.09360266e-01 -8.23658645e-01
1.63946092e-01 8.14494193e-01 7.15626255e-02 -3.08091283e-01
4.47658628e-01 7.57196546e-01 -1.21136653e+00 -5.49264848e-01
6.15256250e-01 6.43865883e-01 -1.43854618e-01 8.81028056e-01
5.09869933e-01 6.67061731e-02 -3.29648286e-01 -4.98283505e-01
8.56928289e-01 -6.85074329e-01 -5.64529896e-01 4.76835787e-01
8.86518061e-01 -2.26236492e-01 -1.15792669e-01 1.07328784e+00
5.31352103e-01 1.20954305e-01 7.31595159e-01 -1.21337306e+00
4.88663018e-02 1.58076018e-01 -5.18163741e-01 -1.99471042e-01
7.92234838e-01 -2.92331487e-01 1.07485521e+00 -1.18673015e+00
6.91150367e-01 1.09385955e+00 1.18879724e+00 1.53584071e-02
-1.38502264e+00 -2.84090906e-01 -4.42572951e-01 -2.94355065e-01
-1.47919476e+00 -5.35131574e-01 6.85854554e-01 -6.83909416e-01
4.36086476e-01 4.11939502e-01 7.56364524e-01 4.51983601e-01
1.14832252e-01 5.21132767e-01 7.69529879e-01 -6.97993934e-01
1.40745997e-01 -1.53131396e-01 1.23083264e-01 5.80249906e-01
5.85009515e-01 3.37558895e-01 -4.09808218e-01 -4.24442172e-01
1.26417172e+00 -1.25470772e-01 -3.37971210e-01 -9.75827575e-01
-1.57804883e+00 8.03261578e-01 2.16037199e-01 3.45281780e-01
-2.04381064e-01 1.65988669e-01 1.98668420e-01 4.14683849e-01
4.18791801e-01 5.60450435e-01 3.16482298e-02 -6.66686445e-02
-1.24201810e+00 8.65248367e-02 1.12708616e+00 1.35490966e+00
8.19179296e-01 -2.61192769e-01 3.02962542e-01 5.03666580e-01
3.31845954e-02 6.53381586e-01 -1.88739896e-02 -1.04276443e+00
3.34211707e-01 1.85672462e-01 3.02203238e-01 -1.25573123e+00
-1.64828598e-01 -6.18755817e-01 -6.27016902e-01 4.23304588e-02
5.54156959e-01 1.67955503e-01 -3.50273132e-01 1.45230317e+00
4.71885055e-01 2.94561356e-01 -1.33174703e-01 8.72896612e-01
1.60595834e-01 2.27626652e-01 -9.44082916e-01 -3.95989567e-01
7.49796033e-01 -6.09226882e-01 -4.50924546e-01 2.30552942e-01
5.26141226e-01 -1.20287085e+00 5.65646112e-01 3.62150997e-01
-1.00535429e+00 -4.72207874e-01 -1.08178353e+00 -1.06953233e-01
2.73358345e-01 3.16233009e-01 2.48630628e-01 5.62467873e-01
-1.21365535e+00 7.11007237e-01 -7.71782815e-01 -8.84055912e-01
-1.65581420e-01 6.01787329e-01 -3.44456047e-01 1.11334056e-01
-4.95202690e-01 9.10763919e-01 1.20437853e-01 2.57076085e-01
-4.02353972e-01 -4.14273411e-01 -7.80046582e-01 -1.69003859e-01
2.90827781e-01 -9.52003717e-01 1.11004674e+00 -6.68101728e-01
-1.40341938e+00 1.04009521e+00 -3.89942765e-01 -3.30581576e-01
1.03567326e+00 -3.04841965e-01 3.03826958e-01 1.20689332e-01
2.75276452e-01 1.30462632e-01 8.43959570e-01 -1.38435149e+00
-1.57604873e-01 -4.58613604e-01 -7.59853870e-02 1.12993494e-01
1.65127099e-01 -1.37821808e-01 -3.97849023e-01 -6.29274368e-01
8.90381634e-01 -1.16625035e+00 -2.18484983e-01 4.79968041e-02
-6.05796099e-01 6.48293644e-02 3.88305694e-01 -4.55550641e-01
9.46380138e-01 -2.18011475e+00 2.53502220e-01 7.94517934e-01
1.01655267e-01 -3.73468280e-01 1.57199278e-01 8.59853148e-01
-1.13266431e-01 -2.51494557e-01 -1.96393251e-01 -4.63683397e-01
4.89808060e-02 1.48774654e-01 -6.05469644e-01 1.21120143e+00
-2.33796060e-01 4.99463022e-01 -9.37468469e-01 -4.59237307e-01
5.31443179e-01 3.59640658e-01 -5.07909834e-01 -1.04585819e-01
4.60802525e-01 4.48059916e-01 -2.84923047e-01 4.73483920e-01
6.36096597e-01 -1.52061328e-01 5.72984032e-02 -1.97333157e-01
-2.60846198e-01 3.30645926e-02 -1.70890522e+00 1.55512500e+00
-4.05169487e-01 7.97266304e-01 1.55529216e-01 -7.68589735e-01
1.19802690e+00 2.94277310e-01 7.18340456e-01 1.64253712e-01
3.57943535e-01 8.04863870e-01 -2.28240743e-01 -4.65776324e-02
5.03111899e-01 -3.49041164e-01 4.46573757e-02 4.36106920e-01
6.63255528e-03 -3.84649664e-01 6.71079159e-02 1.13135181e-01
8.70463610e-01 1.97319657e-01 7.61694252e-01 -5.67768753e-01
5.56760371e-01 4.58041541e-02 3.77058804e-01 8.31294298e-01
9.80397803e-04 1.04298186e+00 1.58277974e-01 -2.84542382e-01
-1.39515138e+00 -9.09874797e-01 -3.94794434e-01 1.87853470e-01
5.42400837e-01 -4.30916578e-01 -5.08117318e-01 -1.71853364e-01
2.49134362e-01 2.57519990e-01 -4.87849236e-01 2.98326463e-01
-6.08895421e-01 -4.02924180e-01 1.57688111e-01 3.07571709e-01
2.64755934e-01 -1.62581414e-01 -6.21803164e-01 1.94440588e-01
-9.59217474e-02 -1.14597476e+00 -5.90664864e-01 -9.71425027e-02
-1.21725452e+00 -1.49621010e+00 -7.04262257e-01 -6.97584271e-01
8.80716324e-01 4.13390815e-01 9.80100453e-01 1.76837370e-01
5.67165203e-02 7.58215606e-01 -7.67639056e-02 4.89938669e-02
-5.61829746e-01 -1.25963911e-01 2.29119644e-01 5.89435920e-02
1.48984656e-01 -5.48998058e-01 -4.02142376e-01 8.64060104e-01
-4.64443237e-01 -1.11062616e-01 1.47034943e-01 9.33349252e-01
6.22932971e-01 -1.64779350e-01 -2.52198223e-02 -5.61601758e-01
4.07906502e-01 -1.02613494e-01 -9.96668458e-01 2.44092941e-01
-3.76032978e-01 3.06832395e-03 3.44805151e-01 -1.55242160e-01
-4.81617182e-01 5.61841369e-01 1.92756534e-01 -6.31016552e-01
1.18490703e-01 1.34330019e-01 8.56219530e-02 -6.15956485e-01
6.34853184e-01 9.57231596e-02 7.61530772e-02 -4.51357573e-01
2.58273065e-01 3.68057936e-01 9.82667387e-01 -7.16038465e-01
1.33592439e+00 8.60435545e-01 6.13018513e-01 -8.89058113e-01
-4.69072461e-01 -1.00444281e+00 -1.18149292e+00 -1.45504475e-01
2.47312069e-01 -8.62893343e-01 -6.75478101e-01 7.19603449e-02
-1.16192055e+00 2.17326537e-01 -3.92527431e-01 8.29301536e-01
-1.34688020e+00 7.01061368e-01 -3.28432053e-01 -1.03332055e+00
3.36948526e-03 -1.30820036e+00 1.38983774e+00 -2.68971205e-01
-4.11529034e-01 -1.23539996e+00 5.06986082e-01 1.21967699e-02
-6.83102533e-02 4.53598440e-01 3.88259888e-01 -6.42665803e-01
-8.41626287e-01 -4.40258741e-01 9.09048244e-02 3.58966812e-02
3.82258557e-02 2.77841210e-01 -7.48429537e-01 -4.45758373e-01
2.75792450e-01 1.91268101e-01 5.69263697e-01 5.34752607e-01
3.99726659e-01 -1.46050289e-01 -4.94008332e-01 6.77769005e-01
1.57445347e+00 -1.19704194e-01 5.70206165e-01 4.45315868e-01
5.35647273e-01 4.26002324e-01 7.06411362e-01 3.45511556e-01
5.04798219e-02 1.00454593e+00 3.41778904e-01 -9.64327827e-02
-1.57427266e-02 -3.25240552e-01 8.78593624e-02 8.84661913e-01
-3.00824106e-01 1.27237484e-01 -7.19937921e-01 6.63989246e-01
-2.03467107e+00 -7.90240467e-01 -4.17176872e-01 2.74714518e+00
4.57880706e-01 -2.68432885e-01 3.28342527e-01 4.14886564e-01
1.15411925e+00 -3.39657009e-01 -1.91901326e-01 -1.62659481e-01
-2.10702494e-01 1.87230498e-01 8.34074438e-01 7.55089402e-01
-9.93650317e-01 5.34294486e-01 7.40073538e+00 5.13207972e-01
-6.00863695e-01 -1.32290855e-01 -2.76957061e-02 3.06664616e-01
-2.22874820e-01 5.17209053e-01 -8.32977712e-01 1.79027602e-01
3.29997152e-01 -2.84512579e-01 2.07798630e-01 7.34773636e-01
1.53479487e-01 -2.11889744e-01 -1.44434571e+00 1.30987024e+00
2.14597747e-01 -1.26953137e+00 -2.31440783e-01 3.98760855e-01
1.05350256e+00 -2.69089997e-01 -1.41781479e-01 -3.29325885e-01
2.24428013e-01 -4.39425260e-01 5.29400289e-01 3.76526386e-01
7.26739883e-01 -5.30685365e-01 7.08633900e-01 3.92975539e-01
-9.61665154e-01 1.89765304e-01 -3.14913303e-01 1.46061346e-01
1.72148794e-01 6.48669720e-01 -9.33566570e-01 9.30797875e-01
8.69080573e-02 9.94597375e-01 -2.94060677e-01 1.61219800e+00
-1.00478694e-01 2.16324821e-01 -8.62259388e-01 3.20158780e-01
-5.90478890e-02 -6.34680152e-01 1.01937616e+00 9.79150653e-01
7.65437961e-01 1.52286470e-01 1.72003597e-01 4.80467319e-01
1.74681574e-01 2.25062355e-01 -8.24687302e-01 5.49784541e-01
4.51452315e-01 7.99333274e-01 -8.04813862e-01 -1.60680220e-01
-5.21297343e-02 7.72291899e-01 -2.28073701e-01 1.49235070e-01
-4.31157738e-01 -2.20921919e-01 2.74517775e-01 2.28305086e-01
2.19207719e-01 -5.14741182e-01 -5.32953262e-01 -1.43194878e+00
1.48413941e-01 -7.66653061e-01 3.19837540e-01 -6.25384390e-01
-1.05377245e+00 1.88725531e-01 1.65336251e-01 -1.77451432e+00
-4.86569971e-01 -3.79715770e-01 -6.60360754e-01 5.60909629e-01
-9.94435251e-01 -8.21794450e-01 -1.56672791e-01 4.90795046e-01
4.24646020e-01 1.17155135e-01 4.49820131e-01 1.22714721e-01
-1.35495484e-01 4.18179452e-01 3.90383422e-01 -1.00219831e-01
8.03470731e-01 -1.19190609e+00 2.36210883e-01 1.14206016e+00
3.38749558e-01 8.60141993e-01 1.18097627e+00 -5.66100478e-01
-1.55926716e+00 -5.31164289e-01 1.04385960e+00 -5.83517194e-01
7.23854959e-01 -1.91442385e-01 -4.58991051e-01 9.65162396e-01
-2.15805978e-01 -2.76623785e-01 3.13560516e-01 1.54327199e-01
-1.04658566e-01 3.94785367e-02 -1.18073428e+00 3.52321416e-01
1.09411180e+00 -5.00056446e-01 -7.08512127e-01 6.43773615e-01
8.70574489e-02 -8.00586224e-01 -8.26703310e-01 4.28764343e-01
5.68219781e-01 -1.12150991e+00 1.06180704e+00 -3.16989943e-02
6.48954585e-02 -5.48875272e-01 -4.13581818e-01 -9.88362908e-01
-3.01717855e-02 -1.11201847e+00 1.19553871e-01 7.76162684e-01
-7.44499862e-02 -6.64339066e-01 6.15664542e-01 1.20323516e-01
-3.15970890e-02 -4.06248033e-01 -1.21994388e+00 -1.15730333e+00
-2.74051130e-01 -2.59661645e-01 3.88561875e-01 9.04233575e-01
-1.77340955e-01 1.17345396e-02 -3.79554003e-01 2.07651585e-01
9.83601809e-01 3.94931674e-01 1.27226579e+00 -1.39343047e+00
-2.94099480e-01 -2.11573213e-01 -9.84939933e-01 -1.45240903e+00
3.93306799e-02 -6.77363634e-01 4.73735183e-02 -1.09565151e+00
2.74686553e-02 -3.61285537e-01 2.89402306e-01 -1.74506798e-01
2.05042720e-01 3.63695115e-01 2.35788852e-01 7.48510361e-01
-2.57920057e-01 3.06582093e-01 1.00101054e+00 2.36150131e-01
-1.64149091e-01 2.84644246e-01 -1.96392164e-01 7.03103900e-01
4.42381233e-01 -3.90408367e-01 -1.26478463e-01 -1.09808028e-01
2.66743094e-01 1.01732776e-01 6.66464269e-01 -1.10536873e+00
3.65540475e-01 1.42603666e-01 2.67892450e-01 -9.74089682e-01
3.12878042e-01 -9.47641909e-01 6.26636624e-01 4.13676500e-01
-1.63251087e-01 2.27117181e-01 -2.22173899e-01 6.07980609e-01
-2.29401559e-01 -3.60463351e-01 6.61627352e-01 6.86799958e-02
-2.99083829e-01 1.31943941e-01 3.29064220e-01 -2.93154329e-01
9.71008539e-01 -7.94587553e-01 7.96941295e-02 -7.55519807e-01
-9.53653514e-01 7.96205178e-02 1.01221025e+00 -4.02625166e-02
4.00477350e-01 -1.57538366e+00 -7.03555286e-01 3.25623304e-01
8.30834433e-02 -1.54952630e-01 -3.52816582e-01 1.23056889e+00
-7.71209717e-01 3.14806879e-01 1.62967131e-01 -1.13936210e+00
-1.26385689e+00 4.38091993e-01 4.33387965e-01 -8.90683085e-02
-6.28502369e-01 3.93582433e-01 5.05381003e-02 -2.42546126e-01
2.22212095e-02 -2.06151918e-01 3.97545546e-01 -1.42004564e-01
6.68730438e-02 6.02254152e-01 2.08943337e-01 -1.07354724e+00
-2.40908325e-01 1.16924143e+00 4.55899239e-01 -2.55882859e-01
1.17839789e+00 -3.87091130e-01 -2.59441704e-01 4.86088753e-01
1.28679812e+00 4.04088378e-01 -1.04291093e+00 -2.41740510e-01
-4.30777222e-02 -8.63633692e-01 -3.52371246e-01 2.90468242e-02
-7.96519339e-01 5.94904900e-01 2.11423934e-01 1.67582631e-01
9.61090565e-01 -2.55085877e-03 4.98007566e-01 4.24305588e-01
6.93244934e-01 -7.77834296e-01 -3.31424683e-01 4.26361769e-01
9.55963075e-01 -1.03036547e+00 2.70156235e-01 -7.62716591e-01
9.25576538e-02 1.64185262e+00 -6.89386651e-02 -6.17282033e-01
5.06631792e-01 3.26145887e-02 -1.80929199e-01 -1.84642032e-01
-1.36489347e-01 -1.32762238e-01 3.21444631e-01 4.47802275e-01
2.95628071e-01 -1.37843817e-01 -4.46101904e-01 -4.09766823e-01
-6.74979925e-01 -1.36050388e-01 6.64205551e-01 7.22156584e-01
-3.71224195e-01 -1.11249363e+00 -9.42871094e-01 -7.07274005e-02
-3.29553075e-02 1.23092823e-01 -4.93142337e-01 1.14889288e+00
-2.82740116e-01 8.01535964e-01 1.83892865e-02 -1.71875834e-01
3.64352226e-01 -1.03059620e-01 8.55482519e-01 -4.19253141e-01
-5.59070647e-01 6.06064975e-01 -1.31845221e-01 -4.79943216e-01
-7.69367278e-01 -9.77658808e-01 -7.68307030e-01 -5.06846011e-01
-6.84005439e-01 3.72824162e-01 5.96902132e-01 8.05372953e-01
1.59300864e-01 -3.03609043e-01 9.97608423e-01 -8.78602684e-01
-7.64586985e-01 -5.03679931e-01 -6.68079674e-01 3.38585496e-01
6.65025115e-01 -7.35714912e-01 -7.05206394e-01 3.11202630e-02]
|
[7.937934875488281, -2.3525514602661133]
|
a10e046f-67dc-4fa4-835a-36924cb13e99
|
deepnnk-explaining-deep-models-and-their
|
2007.10505
| null |
https://arxiv.org/abs/2007.10505v1
|
https://arxiv.org/pdf/2007.10505v1.pdf
|
DeepNNK: Explaining deep models and their generalization using polytope interpolation
|
Modern machine learning systems based on neural networks have shown great success in learning complex data patterns while being able to make good predictions on unseen data points. However, the limited interpretability of these systems hinders further progress and application to several domains in the real world. This predicament is exemplified by time consuming model selection and the difficulties faced in predictive explainability, especially in the presence of adversarial examples. In this paper, we take a step towards better understanding of neural networks by introducing a local polytope interpolation method. The proposed Deep Non Negative Kernel regression (NNK) interpolation framework is non parametric, theoretically simple and geometrically intuitive. We demonstrate instance based explainability for deep learning models and develop a method to identify models with good generalization properties using leave one out estimation. Finally, we draw a rationalization to adversarial and generative examples which are inevitable from an interpolation view of machine learning.
|
['Sarath Shekkizhar', 'Antonio Ortega']
|
2020-07-20
| null | null | null | null |
['interpretability-techniques-for-deep-learning']
|
['miscellaneous']
|
[ 3.00864935e-01 5.26049256e-01 -1.03017651e-01 -3.88205945e-01
-5.91790199e-01 -5.38054049e-01 7.86681354e-01 -2.01308802e-01
1.05289882e-02 1.01024354e+00 -3.01627576e-01 -3.82107884e-01
-3.67248327e-01 -7.44277716e-01 -1.14844036e+00 -6.69797838e-01
-2.37658277e-01 7.33238876e-01 -2.67297834e-01 -2.98868120e-01
4.91667204e-02 6.60326064e-01 -1.18864620e+00 1.55216634e-01
1.04544234e+00 8.30615997e-01 -2.05141097e-01 5.57131410e-01
1.28228903e-01 6.95392787e-01 -2.78446943e-01 -5.36120772e-01
4.20389324e-01 -2.79406279e-01 -6.31740570e-01 -1.69838235e-01
6.00741029e-01 -8.66562575e-02 -8.59381407e-02 7.75060833e-01
2.03035429e-01 3.95415984e-02 1.06782830e+00 -1.79234457e+00
-1.14763141e+00 2.27795318e-01 -1.17001824e-01 -1.73261479e-01
-8.75380635e-02 -8.32341313e-02 7.78580666e-01 -9.56026673e-01
3.73784065e-01 1.07804084e+00 1.07442939e+00 8.51355255e-01
-1.60564041e+00 -4.98018295e-01 -2.19602764e-01 1.73613206e-01
-1.19675767e+00 -1.29326284e-01 8.51270318e-01 -5.07069707e-01
7.03320384e-01 4.76824373e-01 4.47715282e-01 1.11383295e+00
2.94043005e-01 6.06941342e-01 1.11275721e+00 -4.29355502e-01
2.83377737e-01 4.51813579e-01 -7.65825287e-02 7.52809107e-01
-7.59090530e-03 4.01955396e-01 -3.03853780e-01 -1.41385660e-01
1.03726315e+00 1.00589477e-01 -2.05051616e-01 -6.65977538e-01
-9.38393295e-01 9.90611732e-01 8.28658223e-01 7.60044530e-02
-1.54031262e-01 3.34699243e-01 3.03768426e-01 4.16478872e-01
5.69854200e-01 5.55756211e-01 -6.71077788e-01 3.88296843e-01
-8.82213771e-01 3.19315016e-01 6.95924282e-01 1.01779032e+00
8.06537747e-01 3.40645432e-01 3.20635200e-01 5.95267713e-01
-1.73855461e-02 1.58294335e-01 2.69401222e-01 -8.16435099e-01
2.30126739e-01 4.51161772e-01 1.41658083e-01 -9.40255880e-01
-3.98021370e-01 -6.70352280e-01 -1.23475802e+00 6.98903441e-01
5.55654764e-01 1.42751381e-01 -7.62405396e-01 1.63859975e+00
1.65523246e-01 2.86007375e-01 2.26349622e-01 7.32699394e-01
3.38928610e-01 5.52868009e-01 2.57504165e-01 3.04332171e-02
7.84657180e-01 -8.93593967e-01 -3.35474342e-01 9.04963613e-02
7.47590125e-01 -3.55649143e-01 1.18992352e+00 5.62223136e-01
-9.08843756e-01 -5.52099228e-01 -8.87154102e-01 -2.64312059e-01
-5.12307763e-01 1.35547549e-01 7.53571570e-01 4.87385899e-01
-9.21160519e-01 9.40883040e-01 -7.69071400e-01 -8.90835002e-02
6.74110353e-01 7.55078554e-01 -5.32831013e-01 5.65193146e-02
-1.12010217e+00 1.03681815e+00 4.80945200e-01 2.41971076e-01
-7.00910807e-01 -9.52136278e-01 -7.63534963e-01 2.69261189e-02
1.17941715e-01 -8.13201547e-01 8.42224181e-01 -1.32970750e+00
-1.19280136e+00 6.64777160e-01 -1.22904696e-01 -1.01345813e+00
8.21886122e-01 -2.70374477e-01 -3.01294476e-01 -2.90714324e-01
-3.80673438e-01 8.26024532e-01 1.03146935e+00 -1.27773130e+00
-1.34231791e-01 -3.52403671e-01 7.41146207e-02 -1.61244109e-01
-9.46978256e-02 -3.66574109e-01 4.07060862e-01 -7.51570880e-01
5.00787869e-02 -1.12045693e+00 -4.73808765e-01 3.63762647e-01
-5.63942611e-01 -1.62625894e-01 8.02846730e-01 -5.29882312e-01
6.23281240e-01 -1.85056770e+00 1.73972473e-01 2.85123318e-01
3.62018466e-01 3.26972485e-01 6.48247078e-02 3.04915130e-01
-4.32106555e-01 3.46466303e-01 -3.57466340e-01 -2.72266626e-01
1.55415550e-01 4.62605774e-01 -7.32972562e-01 4.56895113e-01
5.93350768e-01 1.04684317e+00 -5.69233000e-01 -2.27256104e-01
2.34763637e-01 6.60083354e-01 -5.73012948e-01 1.03857391e-01
-3.77452791e-01 7.37433910e-01 -3.60250950e-01 3.86753052e-01
7.02221155e-01 -7.31262639e-02 -4.74053949e-01 1.08206756e-01
2.27325946e-01 -4.76477593e-02 -8.67592156e-01 1.26509643e+00
-7.19050109e-01 9.17743146e-01 -3.64109010e-01 -1.27629626e+00
9.52790141e-01 3.76193136e-01 4.85029444e-02 -2.33174831e-01
-1.76987454e-01 3.91413391e-01 -1.64386481e-01 -3.70357454e-01
3.82534415e-01 -5.51869869e-01 1.20353319e-01 7.46475607e-02
-1.23013154e-01 -3.85544673e-02 -4.64723825e-01 -1.30503982e-01
5.85321367e-01 1.80977285e-01 3.71953458e-01 -4.44272786e-01
5.11500299e-01 2.53729522e-01 3.23440611e-01 5.90708256e-01
-4.75210138e-03 7.79885292e-01 4.94362682e-01 -1.06326270e+00
-1.43500698e+00 -1.13353467e+00 -4.13221419e-01 7.81006336e-01
-4.78515737e-02 1.10068694e-01 -7.33832598e-01 -6.42980337e-01
8.95112902e-02 9.09446597e-01 -1.01065445e+00 -1.54245540e-01
-6.01942897e-01 -3.97288710e-01 6.62755013e-01 7.74516582e-01
2.86840111e-01 -1.00265777e+00 -9.64540094e-02 4.26883921e-02
1.39854029e-01 -8.99478912e-01 5.29591329e-02 2.49794677e-01
-1.05743480e+00 -9.16908920e-01 -7.27000713e-01 -7.63888717e-01
9.26086843e-01 -2.19882503e-01 1.13444948e+00 -6.99176965e-03
-2.72537977e-01 2.26036444e-01 1.74743354e-01 -7.32935727e-01
-6.75184786e-01 9.47983861e-02 2.16152176e-01 -3.05294871e-01
2.47229353e-01 -6.01341546e-01 -4.48422760e-01 4.58989441e-01
-1.01845264e+00 2.82728404e-01 3.33929807e-01 1.05225217e+00
8.13201070e-01 1.24645494e-02 8.78119111e-01 -9.46323454e-01
5.65685868e-01 -5.84926307e-01 -6.33721530e-01 3.54470938e-01
-6.61296427e-01 2.73021132e-01 1.26224458e+00 -4.07463789e-01
-8.31722021e-01 2.86543518e-01 -1.17483720e-01 -5.66205263e-01
-3.51097077e-01 1.10604450e-01 1.30604357e-01 -4.06766444e-01
1.11245120e+00 1.86443776e-01 1.21397272e-01 -3.21145147e-01
3.45830262e-01 1.37236118e-01 4.91952330e-01 -6.30650043e-01
1.21615446e+00 5.51438570e-01 5.93267262e-01 -7.78493285e-01
-7.09115207e-01 5.82138598e-02 -8.28611195e-01 -1.32526994e-01
5.65756261e-01 -7.01136291e-01 -5.13378084e-01 1.34379596e-01
-1.13104784e+00 -2.97703981e-01 -4.65846986e-01 2.83675313e-01
-9.75807309e-01 5.06994016e-02 -3.06074709e-01 -8.38428319e-01
-1.85121015e-01 -1.17703032e+00 8.31853449e-01 1.74091756e-02
-3.65355104e-01 -1.48759413e+00 -1.81297675e-01 3.37773532e-01
3.94959718e-01 6.41228199e-01 1.10080612e+00 -9.22174931e-01
-8.06698203e-01 -4.39267695e-01 -1.29465982e-01 5.29784203e-01
-2.07165211e-01 4.18754593e-02 -1.10875297e+00 4.14648205e-02
-1.24097662e-02 -3.81766289e-01 7.98434019e-01 4.23189878e-01
1.45725513e+00 -5.82337439e-01 -4.03317288e-02 7.11116195e-01
1.53150356e+00 -2.73456335e-01 5.76343536e-01 3.19126368e-01
8.32288027e-01 8.00559223e-01 4.03071016e-01 7.66359940e-02
-3.42569575e-02 5.98649144e-01 7.44052470e-01 -2.84101874e-01
1.69648081e-01 -3.74747843e-01 -1.71564501e-02 2.04603463e-01
-2.79622674e-01 5.69365807e-02 -8.73337030e-01 4.40993011e-01
-1.77150524e+00 -8.98916423e-01 -5.39375365e-01 2.22253942e+00
5.36073625e-01 8.40365812e-02 7.76650161e-02 2.51499504e-01
5.12837887e-01 -3.68865252e-01 -5.74799001e-01 -6.84344649e-01
-2.28750110e-01 2.06595361e-01 5.24880350e-01 5.58798373e-01
-1.05641794e+00 7.86360621e-01 6.20008326e+00 9.83914733e-01
-9.95271981e-01 -1.23255037e-01 9.08275306e-01 2.09028512e-01
-5.84431291e-01 -2.02396847e-02 -5.22134840e-01 1.37182221e-01
7.68151879e-01 -2.81328503e-02 5.52298009e-01 1.17997408e+00
2.31130570e-01 3.15298080e-01 -1.42149103e+00 7.31608331e-01
-2.77379066e-01 -1.61655962e+00 6.07532747e-02 1.47793666e-01
9.12152708e-01 -1.69653475e-01 5.80670595e-01 2.04263628e-01
1.00570932e-01 -1.73565102e+00 4.48279709e-01 5.64661920e-01
5.91764688e-01 -9.39072907e-01 4.73676771e-01 6.78331614e-01
-6.32884562e-01 9.85372812e-03 -6.07984900e-01 -1.54491216e-01
-2.28573248e-01 3.89147520e-01 -1.26743042e+00 3.21817607e-01
2.62890726e-01 4.33739752e-01 -3.09692860e-01 9.01472986e-01
-1.77224532e-01 5.59652209e-01 -4.00984704e-01 1.00497402e-01
3.05404037e-01 -2.39858866e-01 5.65828025e-01 8.73623669e-01
3.23517948e-01 -1.06325962e-01 -2.30207965e-01 1.31019199e+00
-4.87287492e-02 8.12041163e-02 -9.99177873e-01 4.09648657e-01
-1.16808549e-01 9.94570136e-01 -4.63100463e-01 4.29316051e-02
-2.21356917e-02 8.40333521e-01 3.72439057e-01 4.64103580e-01
-9.49256957e-01 8.99650380e-02 5.52101672e-01 1.96647942e-01
-9.24471244e-02 -2.85656571e-01 -8.05955231e-01 -1.03718674e+00
2.13302821e-01 -5.98370910e-01 -2.36288458e-02 -7.31588066e-01
-1.31473768e+00 6.81289732e-01 -1.99294146e-02 -1.30523002e+00
-5.44296622e-01 -8.57132018e-01 -7.88987994e-01 1.04887486e+00
-1.44419396e+00 -1.35706472e+00 -4.14565131e-02 5.32357633e-01
4.86084729e-01 -2.45366454e-01 1.08808601e+00 -5.44595392e-03
-2.80739754e-01 7.14492023e-01 3.94984543e-01 -1.70157537e-01
3.91429931e-01 -1.41073012e+00 4.89496559e-01 4.21812952e-01
2.06253603e-01 6.01896346e-01 9.89042103e-01 -3.02140683e-01
-1.14922798e+00 -1.27519178e+00 8.69420528e-01 -7.35580087e-01
5.98987997e-01 -2.73344100e-01 -1.15184271e+00 7.12168813e-01
-1.94756567e-01 2.06948921e-01 6.30330980e-01 1.41927108e-01
-3.68698806e-01 -2.12208293e-02 -1.43635297e+00 7.08815336e-01
7.03012526e-01 -3.96791667e-01 -3.33425075e-01 5.65839112e-01
5.84933877e-01 -3.12906981e-01 -8.35100949e-01 4.12048429e-01
6.35271728e-01 -9.64640141e-01 1.20933414e+00 -1.07632494e+00
8.16393077e-01 -1.12362370e-01 -4.43170406e-02 -1.31505442e+00
4.92358347e-03 -6.99412346e-01 -8.43509883e-02 9.55693901e-01
6.12653673e-01 -6.33367121e-01 1.12630785e+00 1.11423337e+00
1.18163945e-02 -1.25396252e+00 -1.03897536e+00 -9.68119323e-01
5.97815156e-01 -5.92554808e-01 4.38551426e-01 1.05166328e+00
-2.05090687e-01 -3.83900329e-02 -7.44864643e-01 2.22968251e-01
6.99767709e-01 -1.71842188e-01 8.28556180e-01 -1.30547631e+00
-2.87654787e-01 -1.46609202e-01 -7.40248203e-01 -6.79230869e-01
4.26245421e-01 -8.29219759e-01 -3.34903479e-01 -1.10576296e+00
-9.05479491e-02 -8.25690866e-01 -4.25999649e-02 3.64916652e-01
1.19921319e-01 4.82880563e-01 4.96951900e-02 4.22728986e-01
-1.60671435e-02 4.82228070e-01 1.34987092e+00 -8.29471275e-02
-8.86986107e-02 4.03889358e-01 -4.11814749e-01 1.00061309e+00
1.10373175e+00 -4.65278089e-01 -5.12931585e-01 -4.26707655e-01
2.94884533e-01 7.44030252e-02 9.21348214e-01 -1.01262522e+00
3.25961299e-02 -1.52350605e-01 6.84194088e-01 -2.91242570e-01
5.35300255e-01 -1.21635425e+00 4.07877773e-01 4.12676394e-01
-6.90514386e-01 -4.12141979e-02 1.95158556e-01 7.91544259e-01
-9.81971174e-02 -3.13946575e-01 8.50753367e-01 -3.09017487e-03
-4.53923762e-01 4.90710169e-01 6.25144243e-02 -7.00916797e-02
1.10372913e+00 -5.09279013e-01 -8.84396583e-02 -5.04739583e-01
-1.15995586e+00 -1.25350565e-01 5.45261383e-01 2.75409520e-01
5.30876815e-01 -1.44591355e+00 -6.80411220e-01 4.26561147e-01
9.28568989e-02 6.35966733e-02 3.17013174e-01 6.31874025e-01
-6.80619359e-01 5.15635252e-01 -2.58344293e-01 -5.04798651e-01
-1.11309385e+00 6.54726028e-01 4.75390106e-01 -2.04065397e-01
-4.93544281e-01 7.67924130e-01 4.71146137e-01 -4.49328840e-01
1.39623880e-01 -4.00997639e-01 1.09694283e-02 -5.49929261e-01
1.52947769e-01 4.02029306e-01 -3.67706195e-02 -6.48981988e-01
6.16266355e-02 3.61696869e-01 4.90887463e-02 2.48855546e-01
1.38345265e+00 3.14253509e-01 6.88551292e-02 3.14728945e-01
1.41542554e+00 -2.95973778e-01 -1.43469238e+00 -5.06352969e-02
-1.54701680e-01 -3.32202166e-01 -4.29588795e-01 -6.46589875e-01
-7.23456562e-01 1.25243664e+00 3.52980644e-01 5.77909589e-01
9.99906778e-01 -4.01417650e-02 6.41978085e-01 5.21433890e-01
1.59606770e-01 -6.94858015e-01 -2.87997901e-01 3.47795039e-01
1.18393683e+00 -1.39912915e+00 -2.32694000e-01 -5.68087935e-01
-6.18816316e-01 1.38289702e+00 3.39017242e-01 -5.90572715e-01
5.86486042e-01 -8.32800269e-02 -1.28982693e-01 1.12797402e-01
-5.87967694e-01 3.55025262e-01 5.95973015e-01 9.89675939e-01
2.17225447e-01 2.38474727e-01 5.08879311e-02 3.82527560e-01
-4.70033497e-01 -1.20806381e-01 4.47419226e-01 2.25618884e-01
-2.35341430e-01 -1.14652371e+00 -4.01079506e-01 3.11164647e-01
-4.77835506e-01 -1.70824215e-01 -3.99843216e-01 1.17494237e+00
3.62707563e-02 3.61868292e-01 -1.69732124e-01 -2.70463884e-01
3.40266190e-02 1.45766601e-01 3.94278854e-01 -2.19110876e-01
-4.49233830e-01 -4.03741479e-01 -7.87160546e-02 -9.04743969e-02
-1.92641303e-01 -3.91733706e-01 -1.07824504e+00 -4.14269537e-01
-2.76141286e-01 -4.42899428e-02 8.38273466e-01 1.11298716e+00
2.99040526e-01 3.43655735e-01 5.00446796e-01 -8.85533392e-01
-8.13859046e-01 -5.39880633e-01 -4.15465981e-01 3.50702524e-01
5.73732257e-01 -4.58367586e-01 -4.17626530e-01 2.46896148e-01]
|
[8.593213081359863, 4.649456977844238]
|
c8620b11-7e22-4c18-a815-3c7e62bf2fa1
|
learning-enriched-illuminants-for-cross-and
|
2203.11068
| null |
https://arxiv.org/abs/2203.11068v1
|
https://arxiv.org/pdf/2203.11068v1.pdf
|
Learning Enriched Illuminants for Cross and Single Sensor Color Constancy
|
Color constancy aims to restore the constant colors of a scene under different illuminants. However, due to the existence of camera spectral sensitivity, the network trained on a certain sensor, cannot work well on others. Also, since the training datasets are collected in certain environments, the diversity of illuminants is limited for complex real world prediction. In this paper, we tackle these problems via two aspects. First, we propose cross-sensor self-supervised training to train the network. In detail, we consider both the general sRGB images and the white-balanced RAW images from current available datasets as the white-balanced agents. Then, we train the network by randomly sampling the artificial illuminants in a sensor-independent manner for scene relighting and supervision. Second, we analyze a previous cascaded framework and present a more compact and accurate model by sharing the backbone parameters with learning attention specifically. Experiments show that our cross-sensor model and single-sensor model outperform other state-of-the-art methods by a large margin on cross and single sensor evaluations, respectively, with only 16% parameters of the previous best model.
|
['Houqiang Li', 'Xu Jia', 'Wengang Zhou', 'Jianzhuang Liu', 'Chi-Man Pun', 'Zhendong Wang', 'Xiaodong Cun']
|
2022-03-21
| null | null | null | null |
['color-constancy']
|
['computer-vision']
|
[ 5.97310483e-01 -2.82009065e-01 -8.15309510e-02 -3.14125806e-01
-4.60195303e-01 -5.15629828e-01 3.84224981e-01 -3.83170784e-01
-3.77966672e-01 6.23899341e-01 4.86599877e-02 -9.26216394e-02
2.11377993e-01 -6.29561782e-01 -1.16687393e+00 -1.00639749e+00
3.92509490e-01 -1.57907959e-02 3.57269794e-01 -3.47688258e-01
-1.20629944e-01 2.59033531e-01 -1.54189420e+00 4.95182984e-02
9.68421340e-01 1.07094204e+00 4.87454116e-01 7.31104672e-01
2.84491390e-01 1.22919202e+00 -6.32831812e-01 -2.55330116e-01
5.43075860e-01 -4.95179713e-01 -2.82948434e-01 4.15718913e-01
6.06794000e-01 -6.51699007e-01 -5.41350722e-01 1.19680238e+00
4.86894786e-01 1.40907258e-01 1.31415024e-01 -1.21941996e+00
-9.59954679e-01 5.14982879e-01 -7.09241450e-01 -2.23579526e-01
7.30246827e-02 3.96192402e-01 8.12929809e-01 -5.42992711e-01
4.23532635e-01 1.04911447e+00 3.03654552e-01 8.07908237e-01
-1.12540174e+00 -6.81227088e-01 6.18316352e-01 8.02548602e-02
-9.59466279e-01 -5.43405652e-01 1.05511880e+00 -1.26985401e-01
5.37115753e-01 5.88286631e-02 6.81618571e-01 1.43482578e+00
-2.13492841e-01 7.95295238e-01 1.45092392e+00 -1.76536351e-01
2.46874258e-01 8.20963457e-03 -2.48716071e-01 6.59678638e-01
4.34130222e-01 4.32550728e-01 -5.98031402e-01 4.18183982e-01
7.82646418e-01 1.53230980e-01 -6.04686916e-01 -4.83745277e-01
-1.37484336e+00 4.66592431e-01 9.11854923e-01 -3.64087150e-02
-1.53436705e-01 3.87646586e-01 3.99220660e-02 1.71034023e-01
3.93390089e-01 2.46430650e-01 -4.57061440e-01 3.10674727e-01
-4.26080734e-01 -1.57740727e-01 3.69026452e-01 1.00400937e+00
8.84782553e-01 4.11194175e-01 1.31864637e-01 6.65510833e-01
1.21024303e-01 9.06669199e-01 1.91374689e-01 -1.01603782e+00
5.58351159e-01 3.30926210e-01 3.60685915e-01 -8.71222198e-01
-4.32603687e-01 -7.02963293e-01 -1.20916975e+00 3.94874275e-01
3.33187222e-01 -3.28929216e-01 -1.05212593e+00 1.82710171e+00
2.60463476e-01 3.58809888e-01 2.37272903e-01 1.24377453e+00
7.16754198e-01 7.66822636e-01 -2.45917261e-01 -7.57188350e-02
8.86782765e-01 -1.53681648e+00 -6.64495587e-01 -6.59316659e-01
1.01530507e-01 -6.23433292e-01 1.32983339e+00 6.49685383e-01
-7.70472109e-01 -7.90322721e-01 -1.29637074e+00 -1.33859977e-01
-3.55416089e-01 2.42212608e-01 8.35852146e-01 5.98346770e-01
-1.05351496e+00 3.21010113e-01 -6.55826926e-01 -2.72808641e-01
4.28621143e-01 -1.47132203e-01 -1.14471033e-01 -5.35272360e-01
-9.04411972e-01 7.41182446e-01 2.88825631e-01 1.77802294e-01
-1.27274489e+00 -5.35267472e-01 -7.31769741e-01 -2.11394966e-01
5.38727224e-01 -6.03626311e-01 9.95449007e-01 -1.32498050e+00
-1.81455696e+00 5.67255139e-01 3.78779359e-02 -2.31409490e-01
6.14111006e-01 -3.40976268e-01 -5.13361037e-01 4.37390208e-02
-1.83681592e-01 7.17898786e-01 9.28330004e-01 -1.93861580e+00
-5.62502027e-01 -2.45260224e-01 5.34543097e-01 4.11184371e-01
-2.55339265e-01 -3.27593684e-01 -6.97735608e-01 -3.82063776e-01
-3.34381796e-02 -8.19093406e-01 -3.44213009e-01 1.90361932e-01
-6.26365006e-01 4.01313543e-01 5.61697781e-01 -4.76451457e-01
5.42054594e-01 -2.19065738e+00 8.74787867e-02 -1.19056284e-01
1.44159228e-01 9.11623091e-02 -5.13962090e-01 1.06296025e-01
-7.20146969e-02 -3.48148823e-01 -3.03953052e-01 -6.03011370e-01
-1.15058392e-01 3.62425625e-01 -4.93884623e-01 6.38376594e-01
7.18608797e-02 7.13092148e-01 -1.04922867e+00 -2.94022352e-01
5.00187337e-01 5.98686159e-01 -2.52039909e-01 5.11623979e-01
-4.24234658e-01 7.27128029e-01 -2.39125997e-01 6.88306510e-01
8.61873269e-01 -2.65322387e-01 1.39329106e-01 -5.19775212e-01
-9.67419147e-02 -4.48816344e-02 -1.07567418e+00 2.13540745e+00
-4.75471139e-01 6.31410956e-01 3.28180045e-02 -9.36568379e-01
8.20383072e-01 -1.35397837e-01 3.35901082e-01 -1.13831198e+00
1.19954757e-01 1.42591462e-01 -2.74305135e-01 -5.42874396e-01
4.81414258e-01 1.27047852e-01 1.18594617e-01 3.50260407e-01
-3.11026692e-01 -2.30666310e-01 1.10915348e-01 -1.75962541e-02
8.28249395e-01 5.12655377e-01 -5.03016859e-02 2.01815739e-01
1.94234654e-01 -2.15174571e-01 6.88941479e-01 8.41839910e-01
-2.12147519e-01 9.52591300e-01 1.07192308e-01 -3.65626425e-01
-9.69117701e-01 -9.93360758e-01 1.15417160e-01 1.12275398e+00
8.32325280e-01 1.35801569e-01 -6.70703650e-01 -4.90914136e-01
-1.72420815e-01 6.56573176e-01 -5.93934000e-01 -1.09492682e-01
-3.75840634e-01 -9.58041489e-01 1.97160795e-01 4.79598522e-01
1.03284466e+00 -8.89486015e-01 -9.31701660e-01 -1.20450750e-01
-3.11133802e-01 -1.43144727e+00 -2.88633317e-01 2.43633285e-01
-4.14932311e-01 -1.15400600e+00 -6.07906103e-01 -5.24400890e-01
5.54729819e-01 1.01038325e+00 1.18431854e+00 1.62081778e-01
-7.23013580e-02 3.18438828e-01 -5.21039844e-01 -4.30246621e-01
-1.05092727e-01 -1.02929361e-01 -2.18356580e-01 2.92069167e-01
-3.57282385e-02 -3.93740684e-01 -9.33845460e-01 1.97312370e-01
-1.16015923e+00 5.25109828e-01 6.43465459e-01 7.31233895e-01
3.79024029e-01 1.30035236e-01 1.87897041e-01 -8.51195872e-01
9.88871381e-02 -1.66141301e-01 -8.15434396e-01 4.59249794e-01
-3.77266020e-01 -1.47536859e-01 7.85189629e-01 -4.03729886e-01
-1.25282836e+00 1.61569372e-01 2.83306181e-01 -4.44345564e-01
-1.46156475e-01 1.25922756e-02 -3.99690986e-01 -2.29322329e-01
6.39847815e-01 2.51769722e-01 -3.59660208e-01 -3.67217422e-01
6.15366817e-01 4.78537202e-01 8.08797240e-01 -5.27443528e-01
1.14969814e+00 9.25825238e-01 -5.15837222e-02 -4.07143116e-01
-1.31357586e+00 -1.83991566e-01 -2.91558295e-01 -4.29176509e-01
8.96030307e-01 -1.21194601e+00 -6.75918460e-01 9.49553490e-01
-1.12556326e+00 -9.23027635e-01 -3.31108421e-01 2.73252785e-01
-3.83926243e-01 2.46869132e-01 -3.93911391e-01 -8.12139809e-01
-8.32128227e-02 -1.12898302e+00 1.25737703e+00 4.02390629e-01
6.69032812e-01 -7.22905099e-01 3.43714394e-02 4.44880813e-01
4.37986076e-01 4.25690174e-01 5.08577704e-01 4.39754948e-02
-9.60692525e-01 1.69036299e-01 -6.43271089e-01 5.82397640e-01
3.37829739e-01 -7.35114142e-02 -1.20390642e+00 -4.30039138e-01
-8.37076828e-02 -5.93841910e-01 1.17088044e+00 2.13187724e-01
1.33981824e+00 8.60903412e-03 -1.61401946e-02 1.01512492e+00
1.81891060e+00 1.14249915e-01 5.10023773e-01 4.70753342e-01
1.02920234e+00 4.79597598e-01 5.44441879e-01 2.94272512e-01
5.49718022e-01 3.93897504e-01 1.12304747e+00 -6.11268342e-01
-3.14957231e-01 -3.52515489e-01 3.96811008e-01 3.80774379e-01
-2.04557180e-02 -5.93836188e-01 -4.38637942e-01 4.86833692e-01
-1.82967937e+00 -7.03225911e-01 1.66410103e-01 2.24576974e+00
6.42271459e-01 -7.43731484e-03 -9.03870538e-02 -2.93728411e-02
6.92407429e-01 6.65786922e-01 -1.04793835e+00 2.24610969e-01
-7.91025519e-01 -1.98938951e-01 9.72814977e-01 4.28523719e-01
-1.06048441e+00 9.81538415e-01 6.16553593e+00 3.75659794e-01
-1.25055099e+00 1.46557167e-01 8.39844227e-01 -1.79017797e-01
-3.63715559e-01 2.73060482e-02 -3.38453591e-01 5.17461717e-01
4.52300787e-01 5.35465181e-01 1.09439993e+00 6.05554283e-01
3.94939817e-02 -2.17692912e-01 -9.80046391e-01 1.16899621e+00
4.16042477e-01 -1.10103750e+00 -1.63152084e-01 -3.36135298e-01
1.23826563e+00 3.90838027e-01 2.88600564e-01 -8.89456421e-02
7.72287905e-01 -7.57811069e-01 8.84144127e-01 5.34760594e-01
8.55989158e-01 -3.55926454e-01 4.07040626e-01 1.72622338e-01
-8.61711383e-01 -2.50088602e-01 -6.10020578e-01 3.54084675e-03
1.31347775e-01 6.83349013e-01 1.35072265e-02 7.78323650e-01
8.79720867e-01 1.05765319e+00 -5.44721365e-01 8.32974553e-01
-4.58438486e-01 3.57775450e-01 -2.92161316e-01 1.67355403e-01
1.15944631e-01 -4.15612787e-01 1.40791550e-01 8.43369961e-01
2.10561946e-01 -2.94883326e-02 1.75790757e-01 7.60043800e-01
-1.86461508e-01 -4.94755208e-01 -4.44522232e-01 3.78679901e-01
2.89148420e-01 1.25192595e+00 -3.23118091e-01 -2.99984783e-01
-6.42537117e-01 1.15210986e+00 4.29358035e-01 9.66800213e-01
-1.04732144e+00 -2.07400188e-01 6.46501958e-01 -2.73199052e-01
2.22539067e-01 -1.43369839e-01 -2.45892286e-01 -1.54240274e+00
9.03343707e-02 -9.31860507e-01 2.15658799e-01 -1.43901837e+00
-1.31483436e+00 6.71779037e-01 -3.11438531e-01 -1.33865798e+00
3.07627469e-01 -8.24153543e-01 -4.78288263e-01 4.86871600e-01
-2.33748627e+00 -1.43584335e+00 -9.78930354e-01 7.98699081e-01
4.48260397e-01 -3.43758762e-02 4.39200729e-01 2.40946800e-01
-8.80553246e-01 2.41502866e-01 3.89816016e-01 1.66954011e-01
8.99800301e-01 -1.18734825e+00 2.50988573e-01 1.24064612e+00
3.21051106e-02 1.93192035e-01 7.44007051e-01 -3.15226406e-01
-1.60149121e+00 -1.27771914e+00 1.62653685e-01 -3.17337736e-02
6.40890777e-01 -4.14050639e-01 -5.79769611e-01 7.52146363e-01
5.15479922e-01 2.94557154e-01 1.96023971e-01 -2.31678426e-01
-8.02889407e-01 -6.98390722e-01 -9.47670639e-01 7.06115186e-01
1.34575760e+00 -3.16813797e-01 -4.10108902e-02 6.02386117e-01
9.92941320e-01 -6.17021322e-01 -4.09432769e-01 2.29844645e-01
3.00313532e-01 -1.31610322e+00 1.02927709e+00 -3.64585638e-01
5.80163062e-01 -5.88124931e-01 -4.09018844e-01 -1.57732856e+00
-1.04717188e-01 -7.12900400e-01 5.17748334e-02 1.02300596e+00
1.80300623e-01 -8.72352302e-01 5.08849382e-01 3.93436521e-01
-1.11968659e-01 -3.11618447e-01 -7.04281867e-01 -5.69617569e-01
-2.34817773e-01 -3.99259806e-01 8.47052574e-01 7.77339876e-01
-5.99097729e-01 2.88251996e-01 -9.16144192e-01 4.07874316e-01
9.93768752e-01 4.44999665e-01 9.57232952e-01 -7.83056021e-01
-4.90270734e-01 -2.82060742e-01 1.13261538e-02 -1.35751247e+00
1.22204488e-02 -2.91059107e-01 3.58572811e-01 -1.45709038e+00
4.61380869e-01 -4.73542124e-01 -4.47911680e-01 4.06684279e-01
-3.35776061e-01 3.86130035e-01 3.84999007e-01 1.03466377e-01
-9.10920262e-01 7.45366514e-01 1.46677363e+00 -3.71082127e-01
1.20204277e-01 -3.39066684e-01 -8.64544451e-01 4.70067710e-01
7.91825414e-01 -6.10497519e-02 -6.38722956e-01 -1.12733269e+00
4.17662561e-01 -3.43084216e-01 5.87886453e-01 -1.02242792e+00
2.14347914e-01 -4.65495050e-01 5.00371814e-01 -4.20672894e-01
4.70853835e-01 -1.12373424e+00 2.74003227e-03 2.28272840e-01
-3.17263633e-01 -1.36135012e-01 -1.73726395e-01 7.21438348e-01
8.58798623e-02 1.05798811e-01 9.39018190e-01 -5.23350835e-02
-6.99421883e-01 4.18979853e-01 1.28323659e-01 -6.03920184e-02
8.78431022e-01 -1.38499901e-01 -9.25549507e-01 -5.24528623e-01
-2.14261990e-02 1.58432320e-01 8.03317010e-01 4.14206743e-01
5.19354343e-01 -1.16583812e+00 -5.41321754e-01 1.90314874e-01
2.92246014e-01 5.42183936e-01 5.29989421e-01 4.62275863e-01
-4.35529470e-01 -3.41644958e-02 -2.68216103e-01 -5.52414060e-01
-7.35449970e-01 8.84229243e-01 5.06707430e-01 6.92343786e-02
-5.61599612e-01 6.91947579e-01 4.88233536e-01 -4.91723180e-01
3.22994888e-01 -4.31283593e-01 3.72366756e-02 -4.48672354e-01
4.97809023e-01 1.90575033e-01 -1.92875788e-01 -5.11628449e-01
-7.19956262e-03 8.83561790e-01 2.68966854e-01 6.47561774e-02
1.39644837e+00 -4.44557935e-01 6.25622738e-03 5.30716002e-01
1.05909121e+00 -1.68953881e-01 -2.03026128e+00 -5.41684031e-01
-7.29658246e-01 -6.65359914e-01 1.74204573e-01 -1.01363504e+00
-1.75300014e+00 8.04703236e-01 5.61040282e-01 1.05425470e-01
1.65962124e+00 -1.66680828e-01 7.00050414e-01 4.87928629e-01
3.45485210e-01 -1.12977445e+00 2.78721988e-01 1.38568044e-01
6.57922268e-01 -1.58644485e+00 -8.98014940e-03 -3.78434986e-01
-7.96584427e-01 8.52037609e-01 9.11433041e-01 -1.37210265e-01
2.14221820e-01 1.87089190e-01 4.12831783e-01 7.80262724e-02
-6.29361033e-01 -4.17840660e-01 -1.56965442e-02 8.18433166e-01
3.39642689e-02 4.36174497e-02 4.81973469e-01 7.20275640e-02
9.91441011e-02 -1.95916265e-01 6.05916977e-01 6.35820925e-01
-3.64877224e-01 -6.42466366e-01 -3.03283691e-01 -4.31635343e-02
8.30716193e-02 -3.95454541e-02 -2.76787847e-01 7.36128151e-01
1.52394146e-01 1.28116453e+00 2.01023798e-02 -4.04383332e-01
3.05332810e-01 -6.49901748e-01 5.18709719e-01 -1.21439554e-01
-1.89113766e-01 2.43915059e-02 -1.12766571e-01 -8.24624121e-01
-9.02396023e-01 -4.68092591e-01 -8.05679500e-01 -3.35198551e-01
-1.78697512e-01 -4.61297333e-01 7.44941592e-01 7.21322000e-01
2.37172574e-01 7.72551656e-01 1.07763028e+00 -1.03646815e+00
-4.65371072e-01 -8.02031517e-01 -5.79976737e-01 5.81674874e-01
7.06468105e-01 -5.04642248e-01 -4.33893412e-01 7.15029538e-02]
|
[10.490192413330078, -2.6015408039093018]
|
5220b8bd-a871-4019-9441-8c75ca81bb96
|
modeling-a-hidden-dynamical-system-using
|
1904.05172
| null |
https://arxiv.org/abs/1904.05172v2
|
https://arxiv.org/pdf/1904.05172v2.pdf
|
Modeling a Hidden Dynamical System Using Energy Minimization and Kernel Density Estimates
|
In this paper we develop a kernel density estimation (KDE) approach to modeling and forecasting recurrent trajectories on a compact manifold. For the purposes of this paper, a trajectory is a sequence of coordinates in a phase space defined by an underlying hidden dynamical system. Our work is inspired by earlier work on the use of KDE to detect shipping anomalies using high-density, high-quality automated information system (AIS) data as well as our own earlier work in trajectory modeling. We focus specifically on the sparse, noisy trajectory reconstruction problem in which the data are (i) sparsely sampled and (ii) subject to an imperfect observer that introduces noise. Under certain regularity assumptions, we show that the constructed estimator minimizes a specific energy function defined over the trajectory as the number of samples obtained grows.
|
['Trevor K. Karn', 'Steven Petrone', 'Christopher Griffin']
|
2019-04-08
| null | null | null | null |
['trajectory-modeling']
|
['time-series']
|
[-3.84364545e-01 -4.63657305e-02 -8.19795728e-02 4.97811846e-02
-9.31394398e-01 -3.86826277e-01 6.22403979e-01 -5.19582219e-02
1.06551304e-01 6.60357594e-01 3.11282843e-01 -2.55465031e-01
-2.56471723e-01 -6.53859556e-01 -7.26499677e-01 -8.25011849e-01
-4.91867125e-01 4.23893273e-01 4.25254628e-02 1.87817901e-01
-5.25307730e-02 5.65020919e-01 -1.07119310e+00 -5.11385798e-01
7.00048804e-01 6.78628087e-01 2.58242333e-04 1.07483780e+00
1.97983131e-01 1.09176600e+00 -5.04451394e-01 -1.67212933e-01
3.50421667e-02 -5.85697234e-01 -4.81265962e-01 1.87846214e-01
-1.73038766e-01 -7.78083950e-02 -8.12669754e-01 1.06104815e+00
-5.52869737e-02 4.05397356e-01 1.19692540e+00 -1.47380030e+00
-4.76867437e-01 8.67900997e-02 -7.22961426e-02 6.12670779e-01
2.99008992e-02 -1.06521070e-01 4.04187858e-01 -9.77119803e-01
4.74394023e-01 6.45744622e-01 1.09762383e+00 2.68241942e-01
-1.36490595e+00 -1.55229300e-01 -7.81693459e-02 6.33665249e-02
-1.68537164e+00 -4.49687004e-01 7.25742221e-01 -8.62788677e-01
6.86331630e-01 1.13711022e-01 5.33331156e-01 1.04411960e+00
3.45678180e-01 6.32708430e-01 4.18869168e-01 -1.68507963e-01
6.50970161e-01 2.59916455e-01 2.83971459e-01 6.22531414e-01
3.27741086e-01 2.59790063e-01 -3.26032102e-01 -6.42548561e-01
6.32806420e-01 3.44096303e-01 -2.18907326e-01 -1.90943003e-01
-9.28653240e-01 1.08667111e+00 -2.09536716e-01 3.92651767e-01
-8.06095243e-01 1.13901258e-01 -6.80782944e-02 4.23675388e-01
8.88060212e-01 7.64417276e-02 -1.99866891e-01 -5.45490623e-01
-1.27429497e+00 2.62824744e-01 1.14200020e+00 1.00261140e+00
8.12231541e-01 3.47785205e-01 2.72313416e-01 1.14861272e-01
3.61558616e-01 7.45980620e-01 1.07143328e-01 -9.66688991e-01
3.22818846e-01 1.26337647e-01 5.88852584e-01 -1.12221909e+00
-3.23850572e-01 -3.13276350e-01 -9.32568908e-01 -1.98309988e-01
4.03454304e-01 -6.98713481e-01 -4.19025242e-01 1.43553066e+00
2.57941633e-01 1.03100836e+00 3.49723786e-01 6.19474232e-01
-9.79865901e-03 1.22029281e+00 -3.45717132e-01 -3.96653473e-01
4.30061817e-01 -4.08267885e-01 -9.18210626e-01 4.27527279e-01
7.21653163e-01 -4.22080904e-01 2.39881307e-01 2.41908863e-01
-8.85908723e-01 -1.72967046e-01 -8.18421543e-01 4.23154652e-01
-1.85301155e-01 1.24492265e-01 1.26070259e-02 4.20456022e-01
-1.09275043e+00 6.99166834e-01 -1.38753629e+00 -3.92565668e-01
1.62907586e-01 -1.89478546e-01 -1.67897314e-01 8.97220969e-02
-8.46116245e-01 8.76824975e-01 -2.72313416e-01 9.77270156e-02
-1.12277853e+00 -9.36301827e-01 -8.94323170e-01 -1.23770453e-01
-1.37464821e-01 -3.23609501e-01 1.13424301e+00 -5.33541620e-01
-1.18293226e+00 1.94270775e-01 -5.89223742e-01 -7.26556659e-01
3.39069307e-01 -2.00673491e-01 -8.03664207e-01 2.32101277e-01
1.18685268e-01 -3.16467345e-01 1.05383563e+00 -9.78991151e-01
-4.48359191e-01 -1.48818821e-01 -7.05741882e-01 -2.82962620e-01
3.12179979e-02 -2.29134277e-01 1.68104887e-01 -6.27890110e-01
-2.57425457e-02 -1.15514219e+00 -2.80610740e-01 -3.07451963e-01
-2.16589093e-01 -1.20566443e-01 1.23086548e+00 -1.02915215e+00
1.54428291e+00 -2.17106771e+00 3.01782459e-01 3.28419507e-01
1.70347422e-01 8.54047462e-02 3.18995178e-01 9.17460322e-01
1.47971168e-01 -2.16741517e-01 -6.41453862e-01 -6.71160400e-01
-3.35768387e-02 1.76816955e-01 -6.97572589e-01 1.24682403e+00
2.77702719e-01 5.95595717e-01 -8.57769191e-01 2.62038712e-03
2.53934205e-01 4.24963385e-01 -2.47840255e-01 3.11735511e-01
1.79306537e-01 5.75930774e-01 -3.57967794e-01 3.84319872e-01
5.05568147e-01 -2.43814677e-01 -4.07259852e-01 2.83981383e-01
-4.33631986e-01 -1.89447209e-01 -1.27505827e+00 1.40042663e+00
-2.67315894e-01 1.10405576e+00 1.25972167e-01 -1.08007038e+00
8.77027869e-01 6.17842615e-01 7.54967928e-01 1.31401345e-01
-8.14301521e-03 1.28568813e-01 -5.25938630e-01 -5.21908164e-01
4.80663627e-01 -3.09753180e-01 -1.92953900e-01 4.98779118e-01
7.54212663e-02 2.99152851e-01 -2.54448652e-01 4.40178216e-01
1.43852901e+00 -4.16049391e-01 2.64678299e-01 -3.66574347e-01
4.56880108e-02 3.57254863e-01 4.19246286e-01 6.76443517e-01
-1.91199526e-01 5.15015125e-01 4.58222717e-01 -3.52435023e-01
-1.48650146e+00 -1.15710199e+00 -3.42852890e-01 1.69267625e-01
-1.09712645e-01 -3.18528950e-01 -6.69165254e-01 -8.44875947e-02
7.56259337e-02 9.36203182e-01 -6.37344241e-01 -1.94044411e-01
-4.83296394e-01 -8.62343013e-01 5.21205008e-01 4.54202592e-01
2.12357372e-01 -6.89134896e-01 -3.80891740e-01 4.52421278e-01
-4.05042991e-02 -9.22549069e-01 -6.31008148e-01 -2.28168771e-01
-8.83184075e-01 -1.00078011e+00 -6.83980107e-01 -5.48843384e-01
6.87781930e-01 1.26417041e-01 8.61934185e-01 -3.75390768e-01
-9.65669379e-02 9.92960334e-01 -2.64704078e-01 -3.73442113e-01
-4.74075377e-01 -7.86776319e-02 3.48668635e-01 3.71236503e-01
5.36695302e-01 -4.51511770e-01 -2.15778321e-01 1.28215343e-01
-8.24406505e-01 -4.10080969e-01 7.70835057e-02 6.10149622e-01
2.94386744e-01 4.09920841e-01 8.45668972e-01 -4.90177065e-01
6.73679829e-01 -1.54603827e+00 -8.59706402e-01 -3.59422853e-03
-4.36426818e-01 -6.50597289e-02 4.11519915e-01 -4.19816107e-01
-7.93849945e-01 1.67341635e-01 1.87540904e-01 -9.01687443e-01
-3.31596196e-01 5.24230361e-01 3.05721402e-01 7.36186355e-02
5.54391503e-01 5.64650893e-01 2.52191573e-01 -4.12092119e-01
2.41139501e-01 6.53236032e-01 3.31764579e-01 -1.96607232e-01
9.09690082e-01 6.18701100e-01 1.90550715e-01 -1.34804046e+00
-4.78341758e-01 -7.79226840e-01 -7.58326173e-01 -2.53658116e-01
7.68211961e-01 -9.98505235e-01 -7.62805104e-01 4.11181271e-01
-1.00518537e+00 -5.55074096e-01 -5.72942376e-01 8.84672165e-01
-7.62010217e-01 1.26532674e-01 -7.71806479e-01 -1.44061494e+00
1.47350475e-01 -5.01443505e-01 1.14628434e+00 4.28995714e-02
-3.23076904e-01 -1.63026655e+00 9.25515056e-01 -4.85585302e-01
6.11156940e-01 5.44882953e-01 3.66620362e-01 -4.28887129e-01
-6.15674615e-01 -5.48253715e-01 2.00310901e-01 3.45650136e-01
9.13861617e-02 -1.85972378e-02 -5.50962150e-01 -2.12901831e-01
5.73764920e-01 2.83905357e-01 5.37538826e-01 6.75369263e-01
4.55968469e-01 -6.27332091e-01 -4.70802248e-01 5.35146594e-01
1.39693689e+00 1.07435599e-01 3.81011724e-01 -1.48253590e-01
5.88420093e-01 5.14208198e-01 2.97320753e-01 6.36265934e-01
5.45100331e-01 2.57178724e-01 -9.19466466e-02 2.06992298e-01
5.21982014e-01 -4.04500306e-01 5.50036788e-01 1.05910027e+00
3.67489517e-01 -1.42700464e-01 -8.85953307e-01 1.24544334e+00
-2.17116499e+00 -1.44934392e+00 -3.64682227e-01 2.05148053e+00
3.38893294e-01 -4.73425865e-01 6.01040363e-01 -7.25783557e-02
7.22472787e-01 -1.01929884e-02 -5.99360108e-01 -4.47334163e-02
-3.58273722e-02 -1.53606206e-01 7.29152560e-01 6.67549491e-01
-1.00029624e+00 3.94999325e-01 7.13768911e+00 4.45132583e-01
-7.50523865e-01 1.74934000e-01 1.98937654e-01 2.37655197e-03
-8.23188424e-02 -7.19254389e-02 -6.96557283e-01 7.22990513e-01
1.75169492e+00 -5.72704673e-01 4.29862022e-01 9.12252247e-01
3.21527332e-01 -6.23582602e-02 -7.97221243e-01 8.05677056e-01
-1.57991592e-02 -1.48520243e+00 -7.22352386e-01 2.65572488e-01
9.17244613e-01 1.98738903e-01 2.44764119e-01 1.25136226e-01
4.81959879e-01 -1.05226684e+00 5.91274738e-01 1.19608307e+00
5.05834401e-01 -9.65498269e-01 6.28211081e-01 8.29055607e-01
-1.22011101e+00 1.86592881e-02 -3.45440239e-01 -1.98241442e-01
6.38614655e-01 7.49862194e-01 -8.43986630e-01 2.96723932e-01
3.55884314e-01 1.19555473e+00 1.36988238e-01 1.09027016e+00
3.34722608e-01 1.37353396e+00 -5.22230446e-01 7.12607726e-02
4.48406696e-01 -5.57833493e-01 1.05388212e+00 1.25582385e+00
7.68555045e-01 4.32570934e-01 1.90962359e-01 9.93403256e-01
3.42594445e-01 -4.15161520e-01 -1.17192018e+00 -1.27866270e-03
4.62491661e-01 9.87600207e-01 -4.58860219e-01 -3.53276849e-01
-6.04897320e-01 9.01444614e-01 1.03389934e-01 7.12074161e-01
-6.42696619e-01 -1.76074713e-01 9.33514416e-01 2.71371335e-01
5.52322388e-01 -7.60997295e-01 5.43687157e-02 -1.40608227e+00
-1.39844701e-01 2.19975971e-02 8.76258761e-02 -2.56657302e-01
-1.65065646e+00 4.52575207e-01 2.80287266e-01 -1.35120833e+00
-5.83366990e-01 -9.57045332e-02 -8.44222426e-01 9.82811570e-01
-1.05337644e+00 -4.32171524e-01 3.14186990e-01 6.70622349e-01
3.54237825e-01 -2.30747893e-01 8.63739371e-01 2.12106794e-01
-6.24300122e-01 -3.76043953e-02 8.79878044e-01 1.84107140e-01
-2.26894185e-01 -1.39017332e+00 6.57361686e-01 1.08737874e+00
2.49163270e-01 4.90123928e-01 9.67996955e-01 -9.83478189e-01
-1.38732564e+00 -1.56921685e+00 7.61056900e-01 -8.91830921e-01
9.49007690e-01 -2.68870920e-01 -1.10543001e+00 1.06662297e+00
-1.13622926e-01 2.15028688e-01 7.91547000e-01 -2.09427878e-01
3.83726597e-01 4.48693365e-01 -1.31838775e+00 3.83878127e-02
5.21948636e-01 -6.24197841e-01 -5.07455230e-01 3.41109425e-01
4.14511859e-01 -2.10441947e-01 -1.06772840e+00 -3.55043225e-02
3.45673934e-02 -4.38676924e-01 5.64784408e-01 -7.80586779e-01
-3.58184934e-01 -4.53903198e-01 -3.09373170e-01 -1.44679868e+00
-4.99444336e-01 -1.08236933e+00 -6.94471359e-01 8.78942013e-01
1.38386726e-01 -5.40684700e-01 8.46911073e-01 7.21306801e-01
-8.60332921e-02 -5.87481380e-01 -1.24898088e+00 -8.85567307e-01
1.50125567e-02 -5.05814791e-01 3.93797100e-01 7.76327014e-01
2.90742815e-01 -4.53975499e-02 -7.81018376e-01 7.77344108e-01
9.46893513e-01 -1.80368811e-01 6.30583644e-01 -1.35020769e+00
7.95854181e-02 1.58898905e-01 -5.75513005e-01 -9.81420696e-01
3.30995888e-01 -5.04338622e-01 2.88694710e-01 -1.06274939e+00
-1.76002905e-01 -3.60917836e-01 2.68944800e-01 -5.14709294e-01
2.64788628e-01 -2.51611739e-01 -8.64779204e-02 4.95325565e-01
-5.90856433e-01 8.29118907e-01 2.67291576e-01 3.84197563e-01
-2.32930899e-01 5.98319411e-01 -2.04526037e-01 7.50522017e-01
7.69588649e-01 -9.04164970e-01 -3.74298155e-01 1.19306065e-01
-2.52262205e-02 6.46005750e-01 5.79942822e-01 -9.20980453e-01
6.33995056e-01 6.74147252e-03 1.50522903e-01 -9.05046821e-01
3.28236878e-01 -9.56328034e-01 5.63044906e-01 2.67760128e-01
-9.29215699e-02 4.81106102e-01 3.78896929e-02 1.21309507e+00
-3.27428967e-01 -2.49985501e-01 5.77454150e-01 8.41644332e-02
-4.72705275e-01 4.64610100e-01 -9.48997855e-01 1.31793693e-01
1.18448353e+00 1.51072055e-01 -4.24414650e-02 -8.46027732e-01
-9.38380837e-01 1.22489020e-01 5.33604741e-01 -1.01656340e-01
7.14862168e-01 -1.52004242e+00 -7.72866726e-01 2.26017430e-01
-6.47594109e-02 -1.52804345e-01 4.15229142e-01 1.01489806e+00
-3.29634756e-01 6.29574716e-01 4.80983555e-01 -6.26368880e-01
-5.52188218e-01 5.74522793e-01 3.89278799e-01 1.02060605e-02
-1.05393183e+00 4.66330469e-01 -4.63342965e-01 -1.32937342e-01
1.29765421e-01 -3.70356232e-01 6.80923983e-02 -5.31231277e-02
7.70685971e-01 9.67246830e-01 -1.97718531e-01 -1.04630828e+00
-8.28391314e-02 2.71311134e-01 3.25391531e-01 -4.46971506e-01
1.28599083e+00 -3.89751852e-01 8.27800781e-02 1.26249564e+00
1.51063883e+00 -1.61397785e-01 -1.58578563e+00 -3.90687674e-01
2.66714305e-01 -2.83220679e-01 -2.29071565e-02 -4.58981842e-03
-5.91246903e-01 4.90738481e-01 4.15253103e-01 7.86864161e-01
6.62586689e-01 2.22790956e-01 1.01244354e+00 1.44605115e-01
3.41469675e-01 -8.45862031e-01 -2.99284339e-01 4.30879176e-01
4.11007166e-01 -9.47987199e-01 -3.99367273e-01 9.54984576e-02
-7.14041591e-01 1.08274579e+00 -2.52888769e-01 -8.18499863e-01
1.53480494e+00 4.19809192e-01 -3.28680336e-01 -4.07796562e-01
-8.94467235e-01 -1.49245352e-01 2.96577513e-01 6.64945602e-01
-1.32689923e-01 2.17306420e-01 4.08534944e-01 4.79989231e-01
-1.03844225e-01 9.07486677e-02 7.63521314e-01 7.88012385e-01
-6.12420797e-01 -3.08493137e-01 -2.93271273e-01 5.47868073e-01
-2.64151275e-01 2.26611897e-01 1.05995223e-01 5.51998138e-01
-3.36121500e-01 9.87813354e-01 4.02290583e-01 -1.40311956e-01
2.50492394e-01 1.90691859e-01 -1.28462734e-02 -5.01937032e-01
7.92741701e-02 -2.25599334e-01 -2.04894111e-01 -4.60795701e-01
-4.14072812e-01 -1.41175497e+00 -1.00005996e+00 -6.45400167e-01
-1.65305987e-01 5.71403146e-01 6.31133497e-01 9.82214928e-01
4.02738750e-01 2.19986156e-01 8.52656543e-01 -6.40387058e-01
-4.41713750e-01 -1.03923798e+00 -1.18832707e+00 -3.56485508e-02
1.03071141e+00 -5.32234550e-01 -9.03363049e-01 1.34732842e-01]
|
[6.758241653442383, 3.4801242351531982]
|
3689ee43-4c43-438d-95ca-6b372731c140
|
predicting-hurricane-evacuation-decisions
|
2303.06557
| null |
https://arxiv.org/abs/2303.06557v1
|
https://arxiv.org/pdf/2303.06557v1.pdf
|
Predicting Hurricane Evacuation Decisions with Interpretable Machine Learning Models
|
The aggravating effects of climate change and the growing population in hurricane-prone areas escalate the challenges in large-scale hurricane evacuations. While hurricane preparedness and response strategies vastly rely on the accuracy and timeliness of the predicted households' evacuation decisions, current studies featuring psychological-driven linear models leave some significant limitations in practice. Hence, the present study proposes a new methodology for predicting households' evacuation decisions constructed by easily accessible demographic and resource-related predictors compared to current models with a high reliance on psychological factors. Meanwhile, an enhanced logistic regression (ELR) model that could automatically account for nonlinearities (i.e., univariate and bivariate threshold effects) by an interpretable machine learning approach is developed to secure the accuracy of the results. Specifically, low-depth decision trees are selected for nonlinearity detection to identify the critical thresholds, build a transparent model structure, and solidify the robustness. Then, an empirical dataset collected after Hurricanes Katrina and Rita is hired to examine the practicability of the new methodology. The results indicate that the enhanced logistic regression (ELR) model has the most convincing performance in explaining the variation of the households' evacuation decision in model fit and prediction capability compared to previous linear models. It suggests that the proposed methodology could provide a new tool and framework for the emergency management authorities to improve the estimation of evacuation traffic demands in a timely and accurate manner.
|
['Xilei Zhao', 'Shih-Kai Huang', 'Yuran Sun']
|
2023-03-12
| null | null | null | null |
['interpretable-machine-learning']
|
['methodology']
|
[-2.90818304e-01 -2.14119732e-01 -4.67583388e-02 -2.92777449e-01
-1.73814252e-01 -9.37394425e-02 2.15131208e-01 4.36466962e-01
-5.46603262e-01 8.76557529e-01 6.14866674e-01 -7.77580738e-01
-5.69327831e-01 -1.04907000e+00 -7.14225844e-02 -8.70160341e-01
-3.25035602e-01 3.70705009e-01 -4.18121248e-01 -6.76805675e-01
1.13034524e-01 6.29617929e-01 -1.31884229e+00 -3.30761552e-01
1.17919326e+00 5.42528450e-01 1.38017535e-01 2.43110865e-01
5.66828728e-01 5.70703328e-01 -2.87194967e-01 -3.46959010e-02
1.17678091e-01 -6.52393550e-02 -3.50744069e-01 -3.89024794e-01
-8.28575075e-01 -7.58898497e-01 -3.86141837e-01 2.49296635e-01
8.65731657e-01 4.60602790e-01 8.43811572e-01 -1.29012156e+00
-5.01353443e-01 6.46530271e-01 -1.54911295e-01 3.82188022e-01
5.95719934e-01 3.54997694e-01 4.22867000e-01 -7.75344133e-01
6.63460568e-02 1.27805400e+00 8.49155664e-01 6.47843257e-02
-1.15915966e+00 -1.07433832e+00 2.08418414e-01 1.60495475e-01
-1.53766322e+00 -2.43987411e-01 7.35580623e-01 -6.35068834e-01
9.09171402e-01 4.78482872e-01 7.56645918e-01 9.18510258e-01
1.78336158e-01 6.46290183e-03 1.32899356e+00 -3.84182706e-02
2.04177409e-01 3.39498729e-01 3.15044284e-01 1.61823824e-01
6.42041862e-01 5.00177801e-01 -1.53036058e-01 -3.24962378e-01
4.10911858e-01 3.31227362e-01 -1.74828440e-01 5.37843347e-01
-6.68563366e-01 1.05407679e+00 5.20512402e-01 2.49407962e-01
-7.10435212e-01 -5.16802907e-01 2.32757226e-01 1.61104091e-02
4.45165634e-01 5.44473469e-01 -4.53410506e-01 1.06484629e-02
-1.03835428e+00 3.50310892e-01 5.07153630e-01 4.68764752e-01
5.71938097e-01 3.26281160e-01 2.94659156e-02 3.67079526e-01
1.35355741e-01 1.02064168e+00 1.97610572e-01 -2.63322562e-01
4.68868375e-01 7.97436655e-01 1.57883465e-01 -1.80181181e+00
-1.17708647e+00 -4.20820445e-01 -9.94423985e-01 -2.03792736e-01
1.05597153e-01 -5.58520555e-01 -3.79789114e-01 1.53467500e+00
2.81302214e-01 -6.03884012e-02 3.41520049e-02 8.45814824e-01
7.10047245e-01 6.94050431e-01 5.38277686e-01 -5.10758877e-01
1.12095046e+00 -1.41568139e-01 -8.75034034e-01 -2.39660621e-01
5.88817060e-01 -1.72621697e-01 1.02500820e+00 -7.81646520e-02
-7.65792251e-01 -4.71408457e-01 -6.27481341e-01 3.58407199e-01
-3.52640033e-01 -1.44134298e-01 6.56037688e-01 6.44048810e-01
-7.10128069e-01 1.01702563e-01 -8.03080440e-01 -4.51630354e-01
-2.20485236e-02 4.41707164e-01 -7.75821581e-02 2.37994373e-01
-1.63765144e+00 1.15195882e+00 3.17327887e-01 5.30847788e-01
-5.11549771e-01 -4.98982191e-01 -9.54552174e-01 1.77174136e-01
2.78801052e-03 -6.46119714e-01 3.65335524e-01 -2.30150446e-01
-5.91302574e-01 2.48817489e-01 -1.48614869e-01 -4.47001040e-01
3.72544169e-01 1.77930333e-02 -6.14036739e-01 5.50843403e-02
2.05637008e-01 3.93022031e-01 3.38048249e-01 -1.30000854e+00
-5.45319021e-01 -5.05268037e-01 -2.83143282e-01 2.66091168e-01
-4.44809437e-01 2.71869779e-01 7.09376931e-01 -5.37844002e-01
3.04918904e-02 -6.96468651e-01 -5.83093226e-01 -9.59249616e-01
-7.80801103e-02 -3.27475108e-02 4.28220540e-01 -1.07895947e+00
2.09342980e+00 -2.12345600e+00 -4.86034989e-01 3.95122111e-01
7.94300139e-02 8.30512345e-02 3.43341321e-01 8.98836315e-01
1.01128623e-01 1.86357901e-01 -4.48639631e-01 4.94700894e-02
-1.75373126e-02 1.13062486e-01 -5.02551556e-01 3.49527925e-01
3.28715950e-01 7.18747675e-01 -7.22371459e-01 -4.50694144e-01
4.27951485e-01 4.47164893e-01 -4.09228325e-01 3.19835722e-01
7.29489505e-01 7.84750283e-01 -5.68771362e-01 6.27462924e-01
7.39272058e-01 3.77588630e-01 -1.05917841e-01 3.30413908e-01
-5.46212673e-01 2.57295310e-01 -8.12875450e-01 4.09627736e-01
-1.68915987e-01 3.49761456e-01 -1.94922183e-02 -8.02454829e-01
1.50577676e+00 3.65147352e-01 4.79422748e-01 -7.50642657e-01
3.00280184e-01 8.12039375e-02 -3.36037278e-02 -9.95323956e-01
5.60678780e-01 -3.93611193e-01 -4.46657449e-01 3.35762531e-01
-8.24765146e-01 -3.81437838e-02 -2.70773292e-01 -1.83311537e-01
7.52893746e-01 -4.78533983e-01 5.10046899e-01 -4.60813344e-01
3.87823135e-01 1.91606924e-01 6.63692057e-01 4.74519581e-01
-4.15341645e-01 5.09699658e-02 1.17537156e-01 -7.29319453e-01
-8.36024404e-01 -6.62638009e-01 -5.38901627e-01 8.98642421e-01
-3.13676968e-02 -1.42272022e-02 -5.98951578e-01 2.84719728e-02
-3.59885804e-02 1.34158134e+00 -5.32098293e-01 -3.18646640e-01
-6.50012195e-01 -1.44282091e+00 7.20709860e-01 5.77427328e-01
4.53500390e-01 -9.67292845e-01 -9.72590744e-01 1.60598367e-01
-5.26069820e-01 -9.23917830e-01 2.23447144e-01 1.40113771e-01
-7.53670514e-01 -9.50988591e-01 -3.61459218e-02 -4.77489382e-01
7.99410045e-01 5.28232157e-01 5.63246131e-01 2.85102934e-01
1.48348212e-01 1.53493479e-01 -3.35898876e-01 -4.18109804e-01
-7.49892071e-02 1.78290457e-01 4.63037819e-01 -2.32356742e-01
7.30701029e-01 -7.75862694e-01 -6.09978855e-01 4.55877513e-01
-6.89161837e-01 -6.44586161e-02 4.35063571e-01 6.17385268e-01
-1.83738098e-02 5.60431778e-01 1.18008173e+00 -1.70150831e-01
1.01770866e+00 -1.05059850e+00 -2.30113700e-01 -2.28738427e-01
-1.04716754e+00 -3.17638785e-01 6.85474873e-01 -3.94216686e-01
-1.16906691e+00 -3.32557052e-01 -1.94296107e-01 3.26156199e-01
-7.04145610e-01 8.33021581e-01 3.98381017e-02 3.45057338e-01
4.94305938e-01 1.78938463e-01 -2.73119390e-01 -2.70834416e-01
-3.05658311e-01 8.18830967e-01 3.82061303e-01 -4.32794958e-01
1.07179105e+00 2.46993661e-01 4.20328788e-03 -7.93417633e-01
-3.26217860e-01 -4.43432182e-01 -7.62763619e-01 -4.62144494e-01
9.04030859e-01 -1.05919504e+00 -8.96728218e-01 3.15841228e-01
-6.55874014e-01 -3.42839926e-01 2.92415917e-01 8.71039271e-01
-1.10796988e-01 -5.09300223e-03 -3.59774768e-01 -1.60483229e+00
-2.87673384e-01 -1.06998956e+00 3.97602022e-01 1.29986256e-01
-5.62648416e-01 -1.15970993e+00 1.94122881e-01 4.90457147e-01
6.71315193e-01 6.74539745e-01 1.15202224e+00 -7.88074434e-01
-1.91927433e-01 -2.18052506e-01 -5.88677414e-02 -3.59488726e-01
2.70073246e-02 -1.83522124e-02 -7.24469602e-01 -3.43851626e-01
2.09571376e-01 -5.61810918e-02 4.60308194e-01 4.73701626e-01
5.04819572e-01 -7.32393444e-01 -3.69441777e-01 4.59774524e-01
1.09246624e+00 5.45798600e-01 5.27213871e-01 7.16330707e-01
2.37181425e-01 1.05375457e+00 6.67950094e-01 8.11552405e-01
1.08764410e+00 3.97858977e-01 2.91405827e-01 -3.96942556e-01
5.49808919e-01 -5.47285557e-01 4.40900475e-01 7.04775870e-01
-2.69302458e-01 2.07815655e-02 -1.30012047e+00 4.76882935e-01
-1.69625854e+00 -1.18405652e+00 -4.86672640e-01 2.22308636e+00
3.86897802e-01 -1.99858863e-02 2.45777696e-01 4.71612781e-01
5.46823382e-01 -1.62110459e-02 -1.82293028e-01 -6.32447124e-01
-2.71618783e-01 -2.57261366e-01 5.19897461e-01 6.43820941e-01
-7.98887193e-01 7.61508465e-01 6.93205023e+00 4.12680358e-01
-9.66238618e-01 -2.51230627e-01 9.77538049e-01 1.62310109e-01
-3.03653240e-01 -2.31614467e-02 -8.12428355e-01 3.60595614e-01
1.26053643e+00 -1.62732020e-01 4.58169341e-01 6.24723852e-01
1.27072048e+00 -2.76885599e-01 -3.42636317e-01 3.98637563e-01
-1.56728134e-01 -5.84141850e-01 -2.12911427e-01 2.20251933e-01
3.66305798e-01 -5.80093384e-01 1.06679328e-01 5.08336306e-01
9.96141732e-02 -1.12162101e+00 4.34117556e-01 6.28624201e-01
6.13324940e-01 -9.25783098e-01 1.11069405e+00 7.33373761e-01
-1.12544477e+00 -5.95729053e-01 -2.59404749e-01 -1.05671203e+00
5.07335126e-01 3.28846186e-01 -1.29601264e+00 3.10269624e-01
7.13257968e-01 2.23315135e-01 -6.10551298e-01 7.32296526e-01
-7.42938221e-02 1.11667597e+00 -2.38959745e-01 -3.04465787e-03
3.29254210e-01 -3.28509510e-01 3.89472991e-01 1.28814197e+00
3.60658646e-01 6.81178510e-01 8.90215188e-02 6.62842929e-01
7.56209135e-01 3.73519748e-01 -8.89275610e-01 4.16693360e-01
7.37904906e-01 1.05712187e+00 -5.98493099e-01 1.82442851e-02
-3.02789181e-01 2.22777072e-02 3.41301076e-02 5.15825450e-01
-9.53244448e-01 4.87207472e-02 4.10894513e-01 5.24046302e-01
-2.36708686e-01 -4.08310920e-01 -9.09549236e-01 -5.95335126e-01
-2.50869960e-01 -5.74507833e-01 3.19619924e-01 -4.63672310e-01
-1.06451857e+00 6.26191020e-01 5.30234098e-01 -8.33606362e-01
-2.75635064e-01 -7.02722967e-02 -8.49722207e-01 7.80573010e-01
-1.52364981e+00 -1.13842070e+00 -3.95626992e-01 6.58192992e-01
1.62310079e-01 1.75370857e-01 9.54007626e-01 2.90280461e-01
-9.41727698e-01 4.96768355e-01 -8.36965367e-02 -1.52455494e-02
3.99453104e-01 -7.12149978e-01 -9.38814227e-03 8.24740648e-01
-9.90980089e-01 5.98180473e-01 8.88720453e-01 -1.05610335e+00
-9.92451072e-01 -9.04972911e-01 1.39171946e+00 -1.93832085e-01
6.17984712e-01 -3.56291413e-01 -9.55032229e-01 4.82150704e-01
-2.15821087e-01 -9.42368150e-01 9.60394800e-01 7.36866146e-02
2.30743691e-01 -2.04641283e-01 -1.22587001e+00 6.93622589e-01
6.38453066e-01 -4.24186699e-02 -6.55224442e-01 -4.29566130e-02
6.61300659e-01 2.16240212e-01 -8.48266006e-01 5.93795359e-01
4.94235188e-01 -1.01404655e+00 8.40628147e-01 -6.01918459e-01
1.10400826e-01 -2.43377741e-02 -8.42400715e-02 -8.88528824e-01
-7.73692787e-01 -4.87337708e-01 4.44578767e-01 1.18813491e+00
5.55616081e-01 -8.95760894e-01 1.41227469e-01 1.47729421e+00
-1.32957458e-01 -8.29549670e-01 -9.69435275e-01 -3.85431916e-01
1.77920908e-01 -5.17166317e-01 1.04880822e+00 9.78804171e-01
2.69888729e-01 1.43584326e-01 -6.74391389e-01 5.51611900e-01
2.47545615e-01 -2.81578451e-01 7.73320436e-01 -1.33725679e+00
4.01723772e-01 -2.64290929e-01 -1.51202515e-01 -4.10056144e-01
1.43442944e-01 -3.96140128e-01 -2.96155334e-01 -1.38610280e+00
1.64080977e-01 -8.77677739e-01 -1.50759786e-01 5.20740986e-01
-4.59453672e-01 -2.02707261e-01 -2.56624967e-02 1.93516180e-01
1.35519594e-01 7.68460035e-01 8.19989145e-01 1.73692048e-01
-7.80579567e-01 3.74233156e-01 -8.85748386e-01 5.99390566e-01
1.08632791e+00 -4.61095482e-01 -3.84602487e-01 -8.04418698e-02
3.42833012e-01 4.39081401e-01 4.25902009e-01 -8.17158699e-01
2.98130751e-01 -7.16719627e-01 3.93355370e-01 -6.85355484e-01
1.11361474e-01 -1.12795031e+00 5.38010538e-01 6.89040363e-01
-1.86564907e-01 7.63518989e-01 3.26970816e-01 1.18441939e-01
-1.29734222e-02 1.04738638e-01 2.23469973e-01 3.13741535e-01
-8.08402896e-02 1.74928799e-01 -9.26953495e-01 -3.57918203e-01
1.10453832e+00 -4.64638591e-01 -1.79947853e-01 -5.92240930e-01
-5.88859856e-01 5.45568228e-01 3.70909214e-01 2.01958150e-01
8.15008998e-01 -9.98684645e-01 -9.41154420e-01 5.05406797e-01
-3.36952716e-01 -2.51212209e-01 4.07765329e-01 1.28031421e+00
-3.25936139e-01 6.71283424e-01 -2.91254252e-01 8.80234614e-02
-7.34019399e-01 7.64366806e-01 7.57215843e-02 -2.72538781e-01
-4.57619399e-01 1.10896207e-01 9.43921432e-02 -3.85692179e-01
-3.98720102e-03 -3.19042236e-01 -5.36933541e-01 3.33031088e-01
4.82764721e-01 9.52847242e-01 -3.40848446e-01 -1.13052166e+00
-4.73824382e-01 2.88684905e-01 6.52112067e-01 -8.13859850e-02
1.44975805e+00 -6.67144358e-01 -1.81308091e-01 4.05314386e-01
5.29769480e-01 4.28652130e-02 -1.09189689e+00 3.41184825e-01
-1.49544150e-01 -3.96486491e-01 -4.60019745e-02 -8.11500072e-01
-5.50143719e-01 7.54766762e-01 2.74127275e-01 2.78295547e-01
1.48489344e+00 -4.63217914e-01 8.50210786e-01 3.89954120e-01
2.43780196e-01 -9.76992965e-01 -6.50734544e-01 2.29985923e-01
8.08527291e-01 -1.32041967e+00 -1.61449373e-01 9.95088462e-03
-9.10474896e-01 1.01722085e+00 2.82989085e-01 2.17253104e-01
7.26981580e-01 2.46967703e-01 -1.12377480e-02 2.41766516e-02
-6.35468543e-01 -1.71818361e-01 2.10292675e-02 8.13691020e-01
8.12658593e-02 6.25400364e-01 -6.93446755e-01 1.26135671e+00
-7.16377616e-01 -3.57095242e-01 3.73212814e-01 3.31334651e-01
-7.02873647e-01 -4.73361105e-01 -7.98069239e-01 2.80881494e-01
-7.38004297e-02 -2.84555227e-01 -2.80245423e-01 1.00277734e+00
1.38352692e-01 1.62979972e+00 -9.45934653e-02 -7.72386491e-01
2.70208746e-01 8.47811028e-02 -6.42608821e-01 4.82937805e-02
-8.70499790e-01 1.21333219e-01 1.00532815e-01 -1.16369806e-01
-1.69991091e-01 -8.06305766e-01 -1.33050203e+00 -8.38995993e-01
-2.69424796e-01 4.28348333e-01 3.72737914e-01 1.10085499e+00
2.51237959e-01 1.26583770e-01 1.11514080e+00 -7.37740159e-01
-3.18987519e-01 -1.02652335e+00 -4.35123146e-01 2.43783236e-01
3.36477518e-01 -7.52226114e-01 -6.99013770e-01 -3.92060012e-01]
|
[6.572482109069824, 2.0208423137664795]
|
e894db83-da47-431a-8fa5-73d035887181
|
cross-lingual-text-classification-of
|
2108.13620
| null |
https://arxiv.org/abs/2108.13620v1
|
https://arxiv.org/pdf/2108.13620v1.pdf
|
Cross-Lingual Text Classification of Transliterated Hindi and Malayalam
|
Transliteration is very common on social media, but transliterated text is not adequately handled by modern neural models for various NLP tasks. In this work, we combine data augmentation approaches with a Teacher-Student training scheme to address this issue in a cross-lingual transfer setting for fine-tuning state-of-the-art pre-trained multilingual language models such as mBERT and XLM-R. We evaluate our method on transliterated Hindi and Malayalam, also introducing new datasets for benchmarking on real-world scenarios: one on sentiment classification in transliterated Malayalam, and another on crisis tweet classification in transliterated Hindi and Malayalam (related to the 2013 North India and 2018 Kerala floods). Our method yielded an average improvement of +5.6% on mBERT and +4.7% on XLM-R in F1 scores over their strong baselines.
|
['Huzefa Rangwala', 'Hemant Purohit', 'Antonios Anastasopoulos', 'Jitin Krishnan']
|
2021-08-31
| null | null | null | null |
['transliteration']
|
['natural-language-processing']
|
[ 2.27913707e-02 9.68458652e-02 -2.64595836e-01 -5.79522371e-01
-1.25370586e+00 -6.88674629e-01 9.40077186e-01 1.19214006e-01
-8.67366612e-01 8.65137398e-01 2.92461783e-01 -1.02145720e+00
6.09358788e-01 -5.70551157e-01 -9.39486921e-01 -2.79258251e-01
1.66259959e-01 9.79654074e-01 -4.39907700e-01 -7.60060906e-01
-1.39076129e-01 7.05659389e-03 -5.12771010e-01 5.14356971e-01
1.24953449e+00 5.54497123e-01 -7.70409107e-02 5.47544360e-01
-1.38950244e-01 9.61907744e-01 -4.92166579e-01 -7.15779185e-01
1.37619883e-01 -3.10886592e-01 -1.05414569e+00 -3.45026314e-01
7.36731291e-01 -1.04022093e-01 -1.12556420e-01 6.03012919e-01
5.75901210e-01 -2.40450084e-01 8.75050426e-01 -6.26225471e-01
-1.22370338e+00 1.16272891e+00 -6.28719628e-01 3.01825732e-01
8.32426623e-02 -1.88119829e-01 9.39416289e-01 -1.17560911e+00
6.61589563e-01 1.19415295e+00 8.71945202e-01 3.81238014e-01
-1.11462176e+00 -8.78775537e-01 7.72221833e-02 6.56519756e-02
-1.04482877e+00 -4.50165749e-01 3.32032353e-01 -4.08072352e-01
1.27892458e+00 -7.06737069e-03 7.60490447e-02 1.23125565e+00
2.79223084e-01 1.11952448e+00 1.54367101e+00 -6.26458883e-01
-3.42664123e-01 6.06664836e-01 -3.76087204e-02 6.13120377e-01
-1.76210582e-01 -2.70310998e-01 -6.72931492e-01 9.39229503e-02
-6.11534938e-02 -3.75012517e-01 2.25930810e-02 5.47883153e-01
-1.26511943e+00 1.24029088e+00 2.18453199e-01 2.61893004e-01
-1.59070492e-01 -1.01017788e-01 6.03390336e-01 8.87232542e-01
1.46330273e+00 3.04296851e-01 -1.12838757e+00 -2.21825570e-01
-9.43906784e-01 -2.04737531e-03 7.54324198e-01 8.52691948e-01
7.60575056e-01 3.10147047e-01 1.63055420e-01 1.09088182e+00
2.02824727e-01 1.29509687e+00 7.52180278e-01 -2.40273669e-01
1.25097990e+00 4.32021439e-01 -2.63901591e-01 -4.75816548e-01
-4.36760575e-01 -4.70562011e-01 -7.48799920e-01 -2.42902324e-01
3.89493018e-01 -6.88636482e-01 -1.12177145e+00 1.67582488e+00
8.73745605e-02 -3.23790759e-01 4.58256751e-01 4.55233186e-01
7.54323900e-01 1.03740978e+00 1.09925382e-01 3.99202555e-02
1.15292084e+00 -1.24418294e+00 -5.05208135e-01 -6.98572934e-01
1.16074657e+00 -1.05442846e+00 1.58641970e+00 2.73033112e-01
-1.04317832e+00 -3.60419124e-01 -7.16873586e-01 -2.19272345e-01
-8.11335921e-01 3.53639185e-01 3.28468233e-01 7.03428507e-01
-9.08872962e-01 2.81655312e-01 -8.86795044e-01 -4.81730342e-01
1.59752339e-01 2.43987814e-01 -5.16657352e-01 -3.75373065e-02
-1.61659288e+00 1.40823650e+00 2.19789177e-01 6.62315935e-02
-8.10612917e-01 -9.11069393e-01 -9.72808421e-01 -4.38123226e-01
-1.56522527e-01 2.63442136e-02 1.34217727e+00 -9.22709584e-01
-1.55275095e+00 1.20037019e+00 -1.56348318e-01 -4.45115000e-01
6.38909638e-01 -6.06955409e-01 -5.45625091e-01 -3.69870752e-01
2.34445900e-01 3.20269227e-01 5.63856542e-01 -8.18415642e-01
-7.88516939e-01 -2.52603054e-01 -1.91823304e-01 3.77178848e-01
-7.77545094e-01 4.96637940e-01 -4.57943194e-02 -6.40054941e-01
-2.83030510e-01 -9.89097416e-01 -7.36769512e-02 -1.06955087e+00
-1.81919336e-01 -1.90403104e-01 8.79451632e-01 -1.29619253e+00
1.04098701e+00 -1.52654994e+00 7.08531812e-02 -4.63454016e-02
-4.08204019e-01 4.87526476e-01 -3.70572120e-01 6.18181944e-01
-4.70561311e-02 2.13176563e-01 -3.06809366e-01 -8.77384305e-01
3.66518535e-02 3.41197461e-01 -6.01817906e-01 4.89288479e-01
4.65981483e-01 1.13908672e+00 -7.75758743e-01 -9.45708454e-02
-7.93479234e-02 5.10023654e-01 -3.44254732e-01 -2.42217425e-02
-2.79932097e-02 7.57886410e-01 -9.59517062e-02 6.40015185e-01
5.09139717e-01 8.86874422e-02 1.28906697e-01 2.75322735e-01
-2.65930384e-01 8.42792928e-01 -2.61120677e-01 1.62352192e+00
-1.11733186e+00 6.80253506e-01 -1.66902408e-01 -8.59712124e-01
8.65727842e-01 5.12855232e-01 2.32928291e-01 -1.03941333e+00
2.28457049e-01 7.11853504e-01 -1.53580800e-01 -3.74562621e-01
8.13783526e-01 -1.38112366e-01 -5.37352860e-01 9.30169702e-01
2.09854320e-01 -3.29738587e-01 2.16300160e-01 1.95043340e-01
5.96050382e-01 1.95218563e-01 4.49892506e-02 -5.50806344e-01
5.80657899e-01 -8.36422890e-02 1.57687724e-01 5.95382869e-01
1.80414706e-01 5.12314081e-01 4.51857559e-02 -5.71107745e-01
-1.05929756e+00 -7.71276593e-01 -1.67533204e-01 1.69808745e+00
-8.38014722e-01 -3.59761655e-01 -7.47715175e-01 -1.04785872e+00
-2.84349978e-01 8.19932818e-01 -6.74791336e-01 1.06448419e-01
-1.04480529e+00 -1.48128057e+00 7.36974955e-01 2.10565746e-01
4.66816545e-01 -1.22420514e+00 -1.51808262e-02 2.85453022e-01
-7.30234981e-01 -1.32084250e+00 -5.69992602e-01 5.15127301e-01
-5.09943247e-01 -4.46826577e-01 -7.89565742e-01 -8.97493839e-01
4.62402344e-01 -2.93560594e-01 1.33559191e+00 -5.97734988e-01
2.87360311e-01 -8.43537748e-02 -5.63634336e-01 -8.45318496e-01
-8.42551649e-01 9.37488854e-01 2.11974829e-01 -8.39996040e-02
4.74942565e-01 -3.78118604e-01 -1.21824704e-01 -2.36947052e-02
-6.46772742e-01 -1.89529527e-02 4.61158127e-01 5.97192526e-01
1.23340473e-01 -4.16521639e-01 8.47328246e-01 -1.38447583e+00
6.49599969e-01 -7.63817310e-01 -4.10703331e-01 2.80006975e-01
-7.91892469e-01 -1.65947631e-01 9.33477640e-01 -5.10575414e-01
-1.11515892e+00 -2.73458570e-01 -3.17855239e-01 3.83213431e-01
2.51615494e-01 1.16644621e+00 2.30035409e-01 9.70193520e-02
8.78534377e-01 3.21346879e-01 -5.19371510e-01 -4.40223157e-01
5.48986197e-01 1.16540956e+00 5.31437039e-01 -6.47827208e-01
9.42437112e-01 2.24766016e-01 -5.63617766e-01 -7.01116800e-01
-1.17776334e+00 -2.18329653e-01 -8.00742865e-01 1.03870943e-01
7.49634206e-01 -1.29452229e+00 -8.84683952e-02 7.66867399e-01
-1.14617813e+00 -9.88109648e-01 -5.60026132e-02 4.22046185e-01
-2.55259097e-01 2.84422729e-02 -1.04719365e+00 -5.26693463e-01
-9.29383397e-01 -1.04708445e+00 1.08011949e+00 -2.21102789e-01
-1.89524457e-01 -1.55165577e+00 5.44379652e-01 7.07655609e-01
8.10682297e-01 2.58458331e-02 9.95185852e-01 -8.06764245e-01
2.30828986e-01 -1.76985949e-01 -1.23271577e-01 6.33964956e-01
3.14159721e-01 -2.63873935e-01 -9.47592974e-01 -6.16876125e-01
-1.64563179e-01 -8.73487234e-01 8.11682403e-01 2.42128316e-03
3.90496314e-01 -5.32191038e-01 2.55287081e-01 4.71027017e-01
1.03030336e+00 -1.99618980e-01 4.27678853e-01 5.58837235e-01
8.86086822e-01 5.86304784e-01 4.37724620e-01 1.53779387e-01
9.94930446e-01 3.32819909e-01 -6.89804694e-03 -5.14026225e-01
-8.97631571e-02 -8.37531462e-02 1.10989738e+00 1.51083970e+00
1.26516774e-01 -3.94217819e-01 -1.37380934e+00 8.33657086e-01
-1.67786002e+00 -3.17447066e-01 -3.45023334e-01 1.89923537e+00
1.40542865e+00 -9.98988077e-02 -1.28906116e-01 -1.51773885e-01
1.62412196e-01 3.83346900e-02 -1.90161511e-01 -9.10516322e-01
-4.14993137e-01 7.19219744e-01 5.93974471e-01 8.68134856e-01
-1.26252806e+00 1.64770985e+00 5.77840090e+00 8.11233044e-01
-1.58549368e+00 5.76755822e-01 8.18495154e-01 1.39505133e-01
-1.56506374e-01 -1.74107715e-01 -1.01427817e+00 1.75362259e-01
1.66235316e+00 6.74592704e-02 4.09629613e-01 4.63748783e-01
2.77541637e-01 2.19717354e-01 -7.95649171e-01 4.19430047e-01
3.55668455e-01 -9.94431198e-01 -3.58143076e-02 1.53571246e-02
1.35247517e+00 1.07244682e+00 5.69302499e-01 8.68072331e-01
6.60641372e-01 -1.24136686e+00 7.19389915e-01 -1.92403048e-01
9.85048831e-01 -8.08591723e-01 8.45690131e-01 4.48083550e-01
-7.76025832e-01 3.71310979e-01 -2.55476743e-01 -4.98275459e-02
1.16152570e-01 4.34727162e-01 -1.15434754e+00 6.47422850e-01
6.98095262e-01 9.53065991e-01 -6.60297513e-01 -6.99775591e-02
-5.66992521e-01 1.26657259e+00 -3.66319478e-01 2.23006606e-01
7.16810226e-01 -2.05366313e-01 8.11084807e-02 1.70824254e+00
2.75197119e-01 -5.26745796e-01 2.16931641e-01 3.19766134e-01
-3.67954373e-01 6.79718316e-01 -3.22103411e-01 -1.03640378e-01
6.76786061e-03 1.38117778e+00 -2.59587109e-01 -4.04606581e-01
-6.42814100e-01 1.11244273e+00 6.43699944e-01 3.77788514e-01
-9.23623443e-01 -2.88991511e-01 1.76352113e-01 4.21940945e-02
2.73355115e-02 -2.90624142e-01 -1.00505121e-01 -1.53120315e+00
-1.08685687e-01 -1.33888769e+00 4.18663979e-01 -3.37127537e-01
-1.50113225e+00 1.09982729e+00 -2.72093415e-01 -8.34173203e-01
-4.86998081e-01 -7.30714321e-01 -3.93286407e-01 1.17318511e+00
-2.07076812e+00 -1.99585605e+00 3.02604288e-01 6.35771513e-01
5.44773936e-01 -4.78219658e-01 1.14781773e+00 7.98866987e-01
-4.87470925e-01 9.23781872e-01 4.53581542e-01 3.85265231e-01
1.19352627e+00 -1.41272855e+00 1.04257119e+00 8.75699878e-01
2.42475495e-01 5.32928407e-01 4.65531260e-01 -5.59814930e-01
-1.00031137e+00 -1.59209085e+00 1.77653801e+00 -8.32350433e-01
1.18577671e+00 -7.56169736e-01 -7.10492790e-01 1.30981052e+00
5.35249054e-01 -3.61274570e-01 7.01124549e-01 4.62581187e-01
-5.52664638e-01 1.18886121e-01 -8.81535888e-01 6.56518519e-01
4.10407275e-01 -7.63170302e-01 -3.58365208e-01 9.21516657e-01
5.67561209e-01 -3.83845687e-01 -9.26959813e-01 4.32911992e-01
3.83610725e-01 -2.58092970e-01 5.98731637e-01 -9.00380850e-01
6.60629272e-01 1.60416394e-01 -2.40380123e-01 -1.82828486e+00
3.00638806e-02 -7.40990341e-01 2.51413792e-01 1.37799025e+00
1.15508604e+00 -8.55967104e-01 6.46759450e-01 1.60514880e-02
-2.89188713e-01 -5.54557562e-01 -9.47003186e-01 -6.48780763e-01
1.03146279e+00 -5.08572161e-01 1.59562379e-01 1.43841541e+00
9.27303061e-02 7.72212029e-01 -7.01040864e-01 -1.31968528e-01
3.17398161e-01 -2.53093928e-01 8.38762403e-01 -6.82096004e-01
-4.61077057e-02 -2.00260520e-01 2.13238850e-01 -7.62327075e-01
6.12476647e-01 -1.47327697e+00 -1.49178222e-01 -1.31459320e+00
5.09428233e-02 -5.56530774e-01 -1.50275320e-01 8.97405565e-01
-1.74980640e-01 8.51364136e-01 -1.05364667e-03 4.45182063e-02
-2.23880783e-01 5.29664636e-01 8.49439144e-01 -2.87412375e-01
-1.96171612e-01 -3.36318687e-02 -6.23694420e-01 6.54845238e-01
9.30102348e-01 -7.30532944e-01 -1.83272660e-01 -1.15462160e+00
4.67607945e-01 -2.99026847e-01 -3.99466395e-01 -6.23033822e-01
-2.40294382e-01 -8.43017250e-02 2.00367365e-02 -4.29922372e-01
2.27969989e-01 -4.77630019e-01 -4.82691616e-01 4.03807521e-01
-4.57649082e-01 4.27783191e-01 3.26960087e-01 -1.81180894e-01
-1.92055970e-01 9.38843861e-02 7.28957057e-01 -7.30897635e-02
-1.31659657e-01 4.04043496e-01 -7.55338311e-01 3.47989440e-01
4.70023513e-01 4.65611339e-01 -6.47080481e-01 -4.91206646e-01
-4.56647336e-01 3.47664833e-01 -1.89824160e-02 5.54707110e-01
1.73114851e-01 -1.09384191e+00 -1.67972970e+00 1.93358645e-01
1.73367798e-01 -2.62226999e-01 -2.54957974e-01 1.02772367e+00
-6.39626503e-01 6.02004230e-01 1.64076779e-02 -3.53302002e-01
-9.52582121e-01 -6.69487715e-02 2.98794150e-01 -9.87567365e-01
-8.93716514e-02 8.91405106e-01 -4.61511947e-02 -1.68795383e+00
-1.77798748e-01 -3.87736261e-01 -6.09896295e-02 1.50516748e-01
2.93475479e-01 1.55331835e-01 5.22296250e-01 -9.61777925e-01
-3.08988720e-01 5.35381019e-01 -4.42181855e-01 -4.76444840e-01
1.45119607e+00 -2.73842454e-01 -3.22845161e-01 7.00702548e-01
1.44125843e+00 4.44556832e-01 -4.95110303e-01 -5.40918648e-01
6.91249818e-02 3.11155796e-01 -2.10656039e-02 -1.32119358e+00
-9.48960364e-01 9.91751909e-01 2.59529620e-01 -1.62834525e-01
8.31125021e-01 -1.22384325e-01 1.08459437e+00 7.15521693e-01
1.33936390e-01 -1.16647303e+00 -1.79164201e-01 1.31946123e+00
8.64945114e-01 -1.62585878e+00 -1.91097498e-01 1.32708192e-01
-8.52861702e-01 9.76727843e-01 3.48685354e-01 4.00058739e-02
7.41090298e-01 2.91880250e-01 8.43477249e-01 2.16746479e-02
-6.07182860e-01 2.34724700e-01 3.57564956e-01 2.71438330e-01
9.56829250e-01 1.66176826e-01 -2.11083755e-01 2.87943244e-01
-8.71680677e-01 -4.47994351e-01 5.51570714e-01 8.12029183e-01
-9.92547944e-02 -1.24859834e+00 -1.52392119e-01 2.03758538e-01
-1.04307938e+00 -7.98499405e-01 -3.26442748e-01 7.94792891e-01
-4.54524625e-03 1.14814842e+00 -3.01318150e-02 -2.95042306e-01
1.56525671e-01 2.23518133e-01 2.69742519e-01 -7.82428920e-01
-1.14055002e+00 -2.02937215e-03 3.05150270e-01 -5.10013923e-02
-3.12804461e-01 -8.02066684e-01 -1.08293200e+00 -3.59737992e-01
-2.25393236e-01 2.72683918e-01 1.07454610e+00 1.10990226e+00
-1.15828723e-01 3.17249089e-01 6.65929973e-01 -5.09437561e-01
-5.23431659e-01 -1.55644143e+00 -1.59972981e-01 2.90340632e-01
3.43548328e-01 2.54064173e-01 -3.30397129e-01 6.90484419e-02]
|
[11.063369750976562, 9.990811347961426]
|
fa65d2a5-ac77-4f50-9eef-f3c73f8eab77
|
diffdock-pp-rigid-protein-protein-docking
|
2304.03889
| null |
https://arxiv.org/abs/2304.03889v1
|
https://arxiv.org/pdf/2304.03889v1.pdf
|
DiffDock-PP: Rigid Protein-Protein Docking with Diffusion Models
|
Understanding how proteins structurally interact is crucial to modern biology, with applications in drug discovery and protein design. Recent machine learning methods have formulated protein-small molecule docking as a generative problem with significant performance boosts over both traditional and deep learning baselines. In this work, we propose a similar approach for rigid protein-protein docking: DiffDock-PP is a diffusion generative model that learns to translate and rotate unbound protein structures into their bound conformations. We achieve state-of-the-art performance on DIPS with a median C-RMSD of 4.85, outperforming all considered baselines. Additionally, DiffDock-PP is faster than all search-based methods and generates reliable confidence estimates for its predictions. Our code is publicly available at $\texttt{https://github.com/ketatam/DiffDock-PP}$
|
['Tommi S. Jaakkola', 'Regina Barzilay', 'Céline Marquet', 'Gabriele Corso', 'Menghua Wu', 'Hannes Stärk', 'Ruslan Mammadov', 'Cedrik Laue', 'Mohamed Amine Ketata']
|
2023-04-08
| null | null | null | null |
['drug-discovery', 'protein-design']
|
['medical', 'medical']
|
[-2.32464626e-01 2.08892092e-01 -1.20057143e-01 -3.47152114e-01
-1.03027809e+00 -8.86645257e-01 3.90940458e-01 1.00629732e-01
-1.70825347e-01 1.57605326e+00 1.81904420e-01 -6.25684679e-01
2.24137977e-01 -5.09427607e-01 -1.36372507e+00 -1.22657752e+00
-7.06950426e-02 9.69250739e-01 2.00775024e-02 -1.36201397e-01
1.40995860e-01 5.96758127e-01 -6.36169493e-01 2.42808849e-01
9.67557251e-01 1.51243523e-01 2.35115185e-01 5.53773522e-01
2.61911482e-01 2.10257158e-01 -3.15738648e-01 -6.61794662e-01
-9.72600430e-02 -5.70113778e-01 -9.55203116e-01 -7.80847132e-01
2.07187645e-02 1.88488945e-01 -2.18080848e-01 7.34595597e-01
1.02528632e+00 1.30831838e-01 9.30353343e-01 -5.02584040e-01
-9.68064368e-01 9.60064754e-02 -4.30922657e-01 6.33398741e-02
5.05878329e-01 5.48272669e-01 1.01628816e+00 -1.21036589e+00
1.05628705e+00 1.08078885e+00 5.53923547e-01 7.22839892e-01
-1.79715431e+00 -6.90331936e-01 -7.33070746e-02 4.03408110e-02
-1.50684154e+00 -1.13373674e-01 7.31186047e-02 -5.56197107e-01
1.56151366e+00 1.49007723e-01 5.31642377e-01 1.31806958e+00
8.12157929e-01 3.82982731e-01 9.24597859e-01 2.29394026e-02
4.55275595e-01 -4.96868998e-01 -1.24866955e-01 5.67144156e-01
3.01235765e-01 -6.50198683e-02 -5.92209995e-01 -7.04519153e-01
6.23636246e-01 1.08152933e-01 -4.49976534e-01 -4.60758179e-01
-1.10270238e+00 8.98301899e-01 6.56124592e-01 -1.92650110e-02
-7.21723914e-01 1.99847385e-01 -1.31757960e-01 -1.33899733e-01
4.46825594e-01 5.25910676e-01 -8.08659315e-01 -1.91951156e-01
-4.90965426e-01 8.35755408e-01 1.02376366e+00 7.62522936e-01
5.69356680e-01 -4.52356875e-01 -1.02404140e-01 4.54382539e-01
3.08870703e-01 3.44490767e-01 4.53345180e-02 -6.06489003e-01
4.68512811e-02 1.38654783e-01 5.38044035e-01 -4.03344512e-01
-4.75289851e-01 -2.25644037e-01 -6.75859392e-01 2.26868361e-01
3.35580945e-01 -3.00053060e-01 -1.02740669e+00 1.76837468e+00
5.09506643e-01 1.64112449e-01 2.36594200e-01 8.16442370e-01
8.31970155e-01 8.18700075e-01 3.68885994e-01 -3.01532924e-01
1.18935180e+00 -8.68943214e-01 -3.81196171e-01 1.50527090e-01
5.52444160e-01 -9.34098542e-01 8.12009692e-01 3.85859370e-01
-1.08739007e+00 -1.88468471e-01 -8.53274882e-01 -1.34431005e-01
-2.35212356e-01 -1.44972160e-01 8.38005364e-01 2.66766280e-01
-1.13369167e+00 9.52281952e-01 -1.13141096e+00 -2.30495051e-01
4.24113005e-01 8.62597406e-01 -5.03957748e-01 1.40390933e-01
-1.04674888e+00 1.01567686e+00 4.75403547e-01 -1.19094104e-01
-9.95238125e-01 -8.56707275e-01 -3.12702388e-01 -1.77625060e-01
-4.40043621e-02 -1.10975468e+00 1.38145137e+00 -3.03175926e-01
-1.67793941e+00 8.64497840e-01 -6.30708814e-01 -4.92746949e-01
3.09464127e-01 -4.30948168e-01 1.35451868e-01 -3.94278854e-01
-1.83607325e-01 7.66274035e-01 3.54597084e-02 -1.01843262e+00
1.13240607e-01 -3.79101723e-01 -2.21577123e-01 4.15898442e-01
6.63136482e-01 -1.88413821e-02 -2.20985368e-01 -4.06874895e-01
4.16062586e-02 -1.22115958e+00 -6.55844808e-01 -4.15617198e-01
-6.19479120e-01 -3.07626128e-01 1.10197656e-01 -3.53218853e-01
8.44284415e-01 -1.39857769e+00 8.26829791e-01 2.58594692e-01
4.26202059e-01 4.82945859e-01 6.91518784e-02 1.00918519e+00
-3.76010835e-01 -6.23830333e-02 4.19776700e-02 -1.80758424e-02
1.51378270e-02 9.10621211e-02 -2.16268554e-01 6.76762521e-01
8.39038193e-02 1.28149581e+00 -8.25993180e-01 2.40682378e-01
-1.16478810e-02 9.61973071e-01 -6.50482893e-01 1.72940806e-01
-6.37938082e-01 9.10666347e-01 -5.98946035e-01 6.03589177e-01
8.10338020e-01 -7.65649557e-01 7.47098267e-01 -9.88982022e-02
-1.73908900e-02 4.02087986e-01 -5.25712729e-01 1.68613791e+00
2.27873519e-01 -4.98297960e-02 -3.53625387e-01 -6.32690966e-01
8.80137444e-01 3.28253746e-01 4.59212482e-01 -2.45465904e-01
3.82771194e-02 3.52457136e-01 1.87278211e-01 -2.56539118e-02
4.38471213e-02 -1.79012999e-01 3.08790594e-01 1.13975532e-01
1.35760128e-01 9.02561098e-03 -5.17473705e-02 1.57180250e-01
1.07690465e+00 7.07442105e-01 4.07080382e-01 -3.99098754e-01
2.70509928e-01 1.55441761e-01 4.59319055e-01 3.18881273e-01
2.56533802e-01 4.14742976e-01 5.18272340e-01 -6.43827021e-01
-1.15133429e+00 -1.08684194e+00 -2.12367728e-01 1.01975441e+00
5.21157458e-02 -7.00973868e-01 -1.02717805e+00 -3.29310328e-01
1.93798378e-01 3.92775714e-01 -5.15794694e-01 -2.15648293e-01
-4.52729195e-01 -1.36738360e+00 4.23333973e-01 2.90242791e-01
-3.12817872e-01 -1.14247072e+00 7.69690052e-02 5.46236336e-01
7.40713393e-03 -4.03576910e-01 -5.16522288e-01 7.96493292e-01
-7.46455848e-01 -1.11190546e+00 -1.02090681e+00 -7.10310936e-01
6.38860822e-01 -9.28714573e-02 1.21906304e+00 -2.13797674e-01
-3.85841936e-01 -4.53671515e-01 -7.67658725e-02 -4.18097287e-01
-3.35858583e-01 2.06904918e-01 2.34464809e-01 -6.06574476e-01
8.55330408e-01 -8.87553453e-01 -1.16093218e+00 2.38608554e-01
-5.62013030e-01 9.51262116e-02 5.68052173e-01 8.53123903e-01
1.26371539e+00 -8.08678210e-01 2.57764459e-01 -9.39947724e-01
8.07807088e-01 -5.24746895e-01 -6.00508511e-01 8.49855989e-02
-6.83854640e-01 4.15304095e-01 5.07391810e-01 -2.76010901e-01
-6.61193252e-01 4.49082583e-01 -7.08195627e-01 -1.32020727e-01
-2.32218847e-01 5.30227840e-01 -1.36757329e-01 -2.53387362e-01
8.39716554e-01 3.82813722e-01 -4.27164510e-02 -7.35649765e-01
3.54260862e-01 1.86817616e-01 2.95387685e-01 -9.11116302e-01
5.50458848e-01 1.35533318e-01 8.09782371e-02 -4.17032212e-01
-7.11249650e-01 -3.14138651e-01 -5.80926716e-01 4.11797732e-01
1.03199983e+00 -9.53851104e-01 -1.30505574e+00 3.53402197e-01
-1.23312569e+00 -5.61490178e-01 2.54532486e-01 4.93166834e-01
-7.66330421e-01 3.73810232e-01 -7.71182716e-01 -1.84930190e-01
-6.56770885e-01 -1.47651446e+00 1.07357943e+00 2.78773218e-01
-4.22592580e-01 -8.95548284e-01 7.65463948e-01 3.95695686e-01
2.57926077e-01 6.78461850e-01 7.80846775e-01 -8.60167325e-01
-7.09713459e-01 2.20475830e-02 1.64302990e-01 -1.48364410e-01
2.72794645e-02 -2.30568834e-02 -6.71097159e-01 -4.97908622e-01
-5.86533964e-01 -5.23908794e-01 8.25298309e-01 7.94596255e-01
7.58319676e-01 -3.83292675e-01 -7.03693271e-01 8.62047017e-01
1.20868886e+00 4.05772448e-01 8.84292006e-01 3.91469061e-01
5.65008402e-01 -6.88842759e-02 5.12634933e-01 4.06590372e-01
1.90254852e-01 8.78753603e-01 4.49703276e-01 -2.73074925e-01
2.42276415e-01 -3.04763019e-01 2.15914428e-01 3.18014473e-01
-7.28487849e-01 -4.89180654e-01 -1.08814669e+00 -1.70072690e-01
-2.06054950e+00 -1.04931772e+00 -3.35694969e-01 2.09832907e+00
1.53281856e+00 -6.40666410e-02 1.41621962e-01 -7.41311252e-01
3.90675217e-01 -2.15873718e-01 -1.05424190e+00 -2.05479994e-01
-1.95641011e-01 8.44936550e-01 4.90871102e-01 7.99297214e-01
-9.71495032e-01 1.35685086e+00 6.53236437e+00 7.25686431e-01
-9.97963488e-01 9.47421324e-03 7.41148710e-01 -1.42441332e-01
-1.87824443e-01 9.15602371e-02 -1.10936499e+00 3.51312041e-01
1.11986446e+00 -1.43824384e-01 3.23329329e-01 7.47595906e-01
4.47666347e-01 1.14486672e-01 -1.12717819e+00 6.47188902e-01
-5.59886396e-01 -1.81621206e+00 2.52901483e-02 5.10187566e-01
7.67819703e-01 5.73293030e-01 1.19880699e-01 -2.52074562e-02
8.16650450e-01 -1.49584854e+00 2.22964779e-01 6.52188122e-01
7.94402659e-01 -8.21413159e-01 6.07811630e-01 1.51389465e-01
-8.71483803e-01 8.46533597e-01 -6.47293568e-01 1.81694716e-01
6.79648668e-02 5.12310624e-01 -1.02303231e+00 4.66773421e-01
4.83072162e-01 5.79704404e-01 -8.18752050e-02 9.20459270e-01
-2.46208876e-01 4.63460237e-01 -2.05041811e-01 -2.10962877e-01
1.65543094e-01 -6.07403994e-01 2.77019203e-01 1.23788846e+00
3.92365493e-02 4.49082792e-01 2.91470140e-01 9.37618077e-01
-3.33420217e-01 1.91816196e-01 -2.34661549e-01 -2.75342818e-02
4.21443164e-01 9.08723712e-01 -5.79149365e-01 -1.66161403e-01
-2.61723828e-02 1.22188187e+00 4.96312499e-01 5.53708494e-01
-1.21658146e+00 -3.04270178e-01 1.34469640e+00 1.25589505e-01
4.74445522e-01 -2.20054537e-01 5.45266032e-01 -1.11237431e+00
-2.03658625e-01 -8.91203642e-01 -5.06156832e-02 -6.74217582e-01
-1.19467342e+00 5.06014645e-01 -3.89898449e-01 -6.43355131e-01
-1.91837549e-02 -8.12221706e-01 -4.18806463e-01 1.39127028e+00
-1.25410771e+00 -1.08205080e+00 4.94870171e-02 4.83087331e-01
2.52487868e-01 4.11819145e-02 1.23849678e+00 -2.35666856e-02
-4.30987388e-01 5.22839606e-01 9.67227519e-01 -3.72677922e-01
1.03694808e+00 -1.37726736e+00 1.00602961e+00 3.02773237e-01
-1.41429245e-01 1.24109232e+00 1.12598085e+00 -9.37341690e-01
-1.61073446e+00 -1.03280091e+00 7.22751737e-01 -7.57273078e-01
4.59128737e-01 -3.53846192e-01 -1.03007913e+00 6.49881899e-01
2.82944113e-01 -5.79964370e-02 1.14161623e+00 6.36970401e-02
-3.12243372e-01 5.08740544e-01 -9.86791372e-01 5.16203225e-01
1.06068349e+00 -2.83676356e-01 -2.62007207e-01 9.65737343e-01
7.04277575e-01 -9.20407236e-01 -1.10729218e+00 1.32159248e-01
5.48838377e-01 -8.52117717e-01 1.31977224e+00 -1.08197880e+00
-1.36216998e-03 -5.10399699e-01 3.42871062e-02 -1.14981639e+00
-7.53878891e-01 -1.00034130e+00 -2.32338145e-01 3.64408195e-01
7.71937013e-01 -6.96337223e-01 9.36037183e-01 4.08958435e-01
-2.09006116e-01 -1.25956464e+00 -7.72752106e-01 -5.19326806e-01
6.16168201e-01 3.97525340e-01 3.59577209e-01 7.42631435e-01
4.83958274e-02 5.47608852e-01 -4.19713199e-01 1.39396355e-01
4.09063160e-01 1.26215041e-01 8.28593850e-01 -1.14735401e+00
-6.72230363e-01 -8.43804628e-02 -2.45407447e-01 -1.23442793e+00
7.58484900e-02 -1.04473543e+00 -2.40765110e-01 -1.72200227e+00
5.71587741e-01 3.93060520e-02 -1.81938663e-01 7.36761510e-01
-1.54749423e-01 2.26151615e-01 -3.06841642e-01 4.05241638e-01
-5.79895377e-01 5.87126076e-01 1.21958005e+00 9.22626704e-02
-2.60237128e-01 1.06891662e-01 -6.83008552e-01 3.31499845e-01
9.61142957e-01 -5.80798447e-01 -5.22931665e-02 1.60929605e-01
2.85147101e-01 -1.54015630e-01 1.15265243e-01 -6.32411659e-01
-1.09941177e-01 -4.46855307e-01 5.83994746e-01 -5.36130607e-01
4.20840174e-01 -1.66095331e-01 9.40548003e-01 8.02373648e-01
5.13559580e-03 2.07075000e-01 2.72596031e-01 6.33700967e-01
2.13603571e-01 2.83931166e-01 8.95325720e-01 -2.42417634e-01
2.90482324e-02 5.59392035e-01 -4.37265277e-01 1.41704176e-02
1.09816730e+00 6.60873670e-03 -3.80320966e-01 -7.84038901e-02
-1.07606161e+00 -2.03211345e-02 8.49791348e-01 4.71897647e-02
4.71561790e-01 -9.56104219e-01 -5.78110516e-01 -1.76897570e-01
-1.02875665e-01 3.26022767e-02 -2.41516214e-02 8.17399681e-01
-1.13702285e+00 9.39345777e-01 7.02736899e-02 -4.33110505e-01
-1.47404659e+00 4.48502451e-01 6.55930459e-01 -2.66525984e-01
-3.79628986e-01 1.11330283e+00 3.29279810e-01 -4.75904495e-01
1.09043993e-01 -2.27502286e-01 2.97587723e-01 -5.62590897e-01
4.11547273e-01 -1.13886699e-01 1.33445233e-01 -5.55453837e-01
-5.31995952e-01 5.44043541e-01 -5.22743285e-01 5.19354880e-01
1.62891579e+00 5.03518760e-01 -2.19912082e-01 -1.94077685e-01
1.00987864e+00 -1.01003490e-01 -1.33350599e+00 -8.85853320e-02
-1.12986816e-02 -2.71316934e-02 -4.65463638e-01 -1.10161209e+00
-3.83787185e-01 4.42741781e-01 5.52353919e-01 -6.17304146e-01
4.65892851e-01 3.24809611e-01 6.09376013e-01 8.52927804e-01
5.47238588e-01 -3.26530218e-01 -8.30819458e-02 5.53648114e-01
7.16295123e-01 -1.23752534e+00 1.74332649e-01 -1.30950406e-01
-6.75673246e-01 9.69191849e-01 4.40854549e-01 -7.12506846e-02
4.53572780e-01 2.55766809e-01 -1.94678102e-02 -5.05391657e-01
-8.60448599e-01 1.11889355e-01 2.50537604e-01 5.43812275e-01
1.15304279e+00 2.12597638e-01 -4.96669143e-01 6.53582692e-01
-4.85783704e-02 1.13587640e-01 -3.03689600e-03 9.59331989e-01
-4.96791601e-01 -1.77349842e+00 -1.24927416e-01 -1.27811968e-01
-7.15979397e-01 -4.19342935e-01 -9.57749426e-01 4.95620310e-01
-1.78557307e-01 3.66289407e-01 -5.46528578e-01 -1.48934692e-01
2.81885833e-01 2.15023786e-01 6.54828608e-01 -6.03507876e-01
-5.43026805e-01 2.25681216e-01 -2.23876342e-01 -6.31460667e-01
-3.68922263e-01 -4.94973123e-01 -1.71994996e+00 -7.13662744e-01
-4.36682880e-01 7.76941538e-01 4.16848511e-01 4.54071075e-01
1.09106398e+00 3.06398273e-01 8.68547857e-02 -1.17996383e+00
-6.38130426e-01 -7.76197791e-01 -2.96877444e-01 -7.44381174e-02
1.35367259e-01 -5.31785905e-01 2.34270208e-02 1.54549807e-01]
|
[4.8602190017700195, 5.591073989868164]
|
e8c3e7b7-e723-4ecd-907b-dd09922fa180
|
bayesian-community-detection-for-networks
|
2203.02090
| null |
https://arxiv.org/abs/2203.02090v2
|
https://arxiv.org/pdf/2203.02090v2.pdf
|
Bayesian community detection for networks with covariates
|
The increasing prevalence of network data in a vast variety of fields and the need to extract useful information out of them have spurred fast developments in related models and algorithms. Among the various learning tasks with network data, community detection, the discovery of node clusters or "communities," has arguably received the most attention in the scientific community. In many real-world applications, the network data often come with additional information in the form of node or edge covariates that should ideally be leveraged for inference. In this paper, we add to a limited literature on community detection for networks with covariates by proposing a Bayesian stochastic block model with a covariate-dependent random partition prior. Under our prior, the covariates are explicitly expressed in specifying the prior distribution on the cluster membership. Our model has the flexibility of modeling uncertainties of all the parameter estimates including the community membership. Importantly, and unlike the majority of existing methods, our model has the ability to learn the number of the communities via posterior inference without having to assume it to be known. Our model can be applied to community detection in both dense and sparse networks, with both categorical and continuous covariates, and our MCMC algorithm is very efficient with good mixing properties. We demonstrate the superior performance of our model over existing models in a comprehensive simulation study and an application to two real datasets.
|
['Lizhen Lin', 'Nathaniel Josephs', 'Arash Amini', 'Luyi Shen']
|
2022-03-04
| null | null | null | null |
['stochastic-block-model']
|
['graphs']
|
[ 2.67131507e-01 -1.90357417e-02 -3.29325289e-01 -3.38135272e-01
-3.16858530e-01 -4.92136687e-01 5.63703537e-01 3.51313591e-01
-1.45167291e-01 7.76245356e-01 2.06018433e-01 -3.66363674e-01
-5.47046006e-01 -1.00143194e+00 -4.09326255e-01 -9.23330545e-01
-6.07265890e-01 7.87229836e-01 1.54692635e-01 3.22732389e-01
2.71895267e-02 1.11523904e-01 -9.08661067e-01 -2.04824999e-01
8.17625821e-01 4.26514357e-01 2.04403102e-01 3.89307827e-01
1.17131181e-01 6.19758308e-01 -3.19382489e-01 -3.74605477e-01
1.23175383e-01 -1.14896171e-01 -5.35577238e-01 3.71633798e-01
-1.85291991e-01 -1.47904456e-01 -4.67253864e-01 1.07562971e+00
3.30563903e-01 -2.45202065e-01 9.12463844e-01 -1.46095788e+00
-1.63353309e-01 9.84686911e-01 -1.17002106e+00 1.88203931e-01
-1.46250976e-02 -2.08964661e-01 1.24393845e+00 -5.90769172e-01
5.23031831e-01 1.39176214e+00 8.44391942e-01 -5.51901758e-02
-1.85153234e+00 -7.68568337e-01 2.04051957e-01 7.00102672e-02
-1.71366322e+00 -2.11262658e-01 7.90013969e-01 -9.10086334e-01
2.07566231e-01 2.35464354e-03 5.41413963e-01 8.85637224e-01
-1.19442724e-01 5.27030945e-01 9.86618221e-01 -2.41565168e-01
4.67725933e-01 -4.26412784e-02 1.11380309e-01 3.88659298e-01
6.96367800e-01 -1.33105725e-01 -2.97659844e-01 -7.09887743e-01
9.65850532e-01 4.06584591e-01 -8.98017064e-02 -7.29473293e-01
-1.06808889e+00 1.18491089e+00 2.56441325e-01 -3.66773494e-02
-4.02213246e-01 3.06896955e-01 1.08009547e-01 -9.05305296e-02
6.02265656e-01 -2.08732575e-01 -1.16794586e-01 2.73036063e-01
-1.14648736e+00 6.05142256e-03 9.64358509e-01 8.61926854e-01
8.22914124e-01 -3.91481131e-01 1.68863442e-02 7.36812592e-01
6.40907049e-01 5.12651801e-01 -3.08541924e-01 -9.38897967e-01
3.85082930e-01 5.69535494e-01 1.60819113e-01 -1.22949362e+00
-2.29073822e-01 -6.30308568e-01 -1.56607509e+00 -1.50866747e-01
4.84938651e-01 -3.30688000e-01 -6.70697391e-01 1.95222521e+00
4.69680816e-01 6.05958343e-01 -4.38051969e-01 4.16534245e-01
3.40347290e-01 4.39893067e-01 -1.26537964e-01 -3.47174197e-01
1.14878404e+00 -3.34527701e-01 -5.70668697e-01 -1.85811505e-01
2.21465915e-01 -4.94123518e-01 2.18912095e-01 2.21437931e-01
-6.43310547e-01 9.40609316e-04 -6.73227608e-01 3.96694690e-01
6.85930550e-02 -2.68771619e-01 9.33890998e-01 6.16449296e-01
-1.10564947e+00 3.69975507e-01 -1.00771987e+00 -6.64035380e-01
5.40610850e-01 3.98585916e-01 -1.68800354e-01 -4.96921152e-01
-9.24945772e-01 3.39621961e-01 1.51851341e-01 1.84984654e-01
-8.02687407e-01 -5.48665762e-01 -5.69005668e-01 3.04073721e-01
7.49378145e-01 -8.79664838e-01 7.42594898e-01 -5.23004830e-01
-7.33856738e-01 4.78275508e-01 -3.15908968e-01 -3.69437724e-01
3.90526354e-01 4.22523379e-01 -1.09050088e-01 1.24578252e-01
3.59943151e-01 3.56442362e-01 5.76196074e-01 -1.11136878e+00
-4.83555287e-01 -2.67298788e-01 -6.09659329e-02 -7.44746104e-02
-2.77528465e-01 1.46560837e-02 -5.53573847e-01 -5.87694883e-01
2.95661628e-01 -9.29329157e-01 -5.86201727e-01 2.65814424e-01
-5.72784841e-01 -1.46618158e-01 6.57166183e-01 -3.74713928e-01
1.24923551e+00 -1.97213149e+00 3.09009373e-01 7.79457688e-01
7.92731702e-01 -3.94785434e-01 8.03994983e-02 8.15678477e-01
2.44773254e-01 3.64718705e-01 -5.33610284e-01 -4.39235389e-01
-1.32051423e-01 3.74554127e-01 9.02159363e-02 8.51920247e-01
6.96445853e-02 4.65038627e-01 -1.02897990e+00 -6.14712119e-01
2.95630991e-02 5.89367509e-01 -5.72789967e-01 -2.09560897e-02
1.98992833e-01 4.75977629e-01 -5.55340171e-01 4.38378632e-01
6.13345504e-01 -1.04897988e+00 7.82722712e-01 2.96262741e-01
1.81873918e-01 -1.38221666e-01 -1.68983710e+00 1.00551593e+00
1.90715685e-01 5.74947715e-01 7.77065635e-01 -1.16070604e+00
5.90606928e-01 4.99335915e-01 7.42595613e-01 3.22641522e-01
-8.45585167e-02 -5.70046417e-02 3.92254561e-01 -1.55811056e-01
-6.75209090e-02 -1.02379568e-01 -9.26641840e-03 6.75524592e-01
-3.21539015e-01 1.90031454e-01 4.22943980e-01 6.66302800e-01
1.45757043e+00 -6.70945346e-01 3.94045860e-01 -4.31097537e-01
2.10962936e-01 -1.98672742e-01 8.19556117e-01 1.13652420e+00
1.65617466e-02 3.14491779e-01 7.29964435e-01 1.10460192e-01
-1.06441236e+00 -1.07934511e+00 -2.80268908e-01 6.01354897e-01
4.01676893e-02 -2.59126216e-01 -3.88445616e-01 -1.53952986e-01
3.25854093e-01 -1.40190929e-01 -7.36588955e-01 2.30250284e-01
-1.35606617e-01 -1.19007874e+00 2.69885451e-01 3.95720154e-01
1.35080978e-01 -6.15815461e-01 2.09632009e-01 3.57528806e-01
-3.60082924e-01 -9.73046601e-01 -4.11513537e-01 1.76222041e-01
-1.08500755e+00 -1.37693381e+00 -3.79729003e-01 -6.90125525e-01
8.23394179e-01 3.96257848e-01 1.08786786e+00 2.62641937e-01
-3.44124496e-01 3.30170214e-01 -1.21644787e-01 -1.02316156e-01
-2.41862163e-01 1.58382773e-01 1.06433839e-01 2.64212996e-01
4.11919028e-01 -9.86084104e-01 -5.35886943e-01 2.63364673e-01
-7.51057923e-01 1.53520331e-01 5.70247293e-01 7.80821681e-01
2.73590952e-01 4.20074105e-01 6.80333734e-01 -1.09534419e+00
4.08704638e-01 -1.13980949e+00 -6.61874413e-01 9.48411152e-02
-4.26814020e-01 -1.77025154e-01 1.48845375e-01 -4.96880293e-01
-7.78658807e-01 -1.28731877e-01 4.21670824e-01 -1.44628450e-01
-9.28180106e-03 1.08714175e+00 -2.58215904e-01 2.40939021e-01
1.57083422e-01 -2.88515329e-01 1.73587590e-01 -5.95858037e-01
2.75063097e-01 6.30531788e-01 3.40621531e-01 -6.53959095e-01
9.50993896e-01 8.44725430e-01 3.54502082e-01 -8.65317643e-01
-6.05782568e-01 -8.26077640e-01 -9.62120712e-01 -1.82202294e-01
3.83493990e-01 -1.16557801e+00 -7.54348993e-01 4.50458407e-01
-7.89070129e-01 -1.92224354e-01 1.24860115e-01 5.98374724e-01
-1.94501117e-01 5.97597659e-01 -7.86920071e-01 -1.14527154e+00
3.27845737e-02 -9.21566486e-01 6.57827914e-01 -6.94194213e-02
-1.94258854e-01 -1.47577739e+00 8.44431221e-02 1.50350019e-01
2.01606318e-01 3.13417792e-01 1.03444207e+00 -5.70212841e-01
-9.34786141e-01 -3.64291787e-01 -4.74706143e-01 -1.11317985e-01
3.57921928e-01 1.87094390e-01 -5.98718107e-01 -6.00987196e-01
-1.81870490e-01 6.81729317e-02 9.12873387e-01 9.01191711e-01
9.83169019e-01 -1.94554299e-01 -9.30354118e-01 4.60736871e-01
1.36300695e+00 -2.69903034e-01 2.80775994e-01 -2.29985923e-01
6.55163348e-01 6.55031800e-01 4.45903800e-02 8.19020689e-01
5.09350657e-01 4.89358068e-01 5.32952189e-01 -1.26873583e-01
2.20196068e-01 -2.51742959e-01 -7.64493048e-02 9.99104798e-01
1.75400332e-01 -4.37463671e-01 -1.14503276e+00 7.79179096e-01
-2.10912657e+00 -1.22675025e+00 -4.81991976e-01 2.29090691e+00
1.01780260e+00 -3.20733339e-02 2.29843363e-01 1.39770746e-01
1.36907423e+00 -2.78662033e-02 -5.81095874e-01 5.79039454e-01
-1.83931589e-01 -1.66085795e-01 5.27471900e-01 5.98642409e-01
-1.14662528e+00 3.91911507e-01 6.78659964e+00 6.37358963e-01
-4.32898700e-01 -9.22528654e-02 7.16312528e-01 -9.23904590e-03
-1.75836504e-01 4.46837902e-01 -7.18779922e-01 4.63853538e-01
7.12038219e-01 -2.42928863e-01 3.96966964e-01 4.44406778e-01
5.40548801e-01 -2.71997422e-01 -1.23373103e+00 6.24580562e-01
-2.83200413e-01 -1.09259105e+00 -3.95793200e-01 7.23958731e-01
9.92550790e-01 1.05317578e-01 -2.56538540e-01 -1.02420002e-01
1.15102601e+00 -1.02576351e+00 1.60922900e-01 4.66794789e-01
8.35160077e-01 -6.23996139e-01 6.17499530e-01 5.76240480e-01
-1.22775185e+00 -1.69694945e-01 -3.29529285e-01 -4.55984235e-01
2.97637433e-01 1.15787268e+00 -9.09591019e-01 2.35733956e-01
5.89364171e-01 1.13450348e+00 -3.72732192e-01 1.55174077e+00
-1.29130021e-01 1.12615764e+00 -7.59136736e-01 1.23502925e-01
-1.20945930e-01 -3.97401929e-01 5.00436664e-01 8.64379406e-01
2.71397710e-01 -3.57995778e-02 6.22299075e-01 7.63313591e-01
-1.06006794e-01 -1.31132588e-01 -4.90491271e-01 -3.28552239e-02
1.12918425e+00 1.29253995e+00 -1.04739118e+00 -3.04362476e-01
-5.59882283e-01 2.54034221e-01 3.63854527e-01 5.60932159e-01
-4.06482726e-01 -8.61938521e-02 5.64874232e-01 4.48805422e-01
4.65472728e-01 -2.98112243e-01 -2.73681879e-01 -1.09464622e+00
-2.50408739e-01 -6.15948856e-01 4.76244271e-01 -2.45491728e-01
-1.94347394e+00 -2.07685515e-01 2.56685019e-01 -6.93497717e-01
-1.65772304e-01 -3.43600482e-01 -7.18303502e-01 1.01176500e+00
-1.13787889e+00 -1.00508749e+00 -1.11698568e-01 4.67141747e-01
-9.34201330e-02 1.42401695e-01 4.13108438e-01 4.45591122e-01
-7.63626397e-01 3.48590650e-02 7.45591760e-01 4.94896024e-01
6.92813873e-01 -1.16342258e+00 2.54431486e-01 8.23081851e-01
-5.63455895e-02 8.18461239e-01 6.93104446e-01 -1.06260824e+00
-9.50755715e-01 -1.01683104e+00 7.31487870e-01 -3.21864009e-01
9.52419817e-01 -7.28691339e-01 -8.75407577e-01 8.29456270e-01
-1.59748703e-01 -9.65223589e-04 8.90029371e-01 4.81742173e-01
-3.17491531e-01 1.24053825e-02 -1.01024091e+00 2.79594332e-01
9.71341074e-01 -3.28377306e-01 -4.46374193e-02 2.54023045e-01
3.66169065e-01 4.04018283e-01 -9.84052658e-01 3.98220390e-01
4.03887421e-01 -6.38128519e-01 9.91342306e-01 -3.29854339e-01
1.42532408e-01 -4.15359586e-01 -1.31873950e-01 -1.02614737e+00
-6.97827578e-01 -4.88282919e-01 -2.71301061e-01 1.42681098e+00
1.29262313e-01 -4.79981244e-01 9.98194933e-01 4.92313802e-01
5.36852360e-01 -4.46234107e-01 -1.03871024e+00 -5.21937966e-01
6.78971484e-02 -7.79085532e-02 5.57838738e-01 1.05685222e+00
-1.56431925e-02 4.81756240e-01 -5.68990469e-01 3.97348970e-01
1.23983049e+00 1.44716054e-01 7.98744440e-01 -2.23062587e+00
-5.44429600e-01 -4.76908833e-01 -2.49361575e-01 -1.01818919e+00
1.01354763e-01 -9.00613308e-01 -3.61037999e-02 -1.66145205e+00
1.10143971e+00 -7.61493027e-01 9.60799232e-02 3.34394455e-01
-3.67916375e-01 6.35249615e-02 -2.04762250e-01 4.63533312e-01
-6.18116140e-01 4.09899741e-01 8.22658420e-01 -1.14475161e-01
3.52662168e-02 4.09782916e-01 -6.74241364e-01 7.44948149e-01
5.45483112e-01 -7.50978649e-01 -3.93464059e-01 -8.49299431e-02
4.27371949e-01 2.65296638e-01 5.40023267e-01 -6.32002592e-01
5.97977102e-01 -2.76173413e-01 3.67343903e-01 -6.94711149e-01
3.06197822e-01 -8.71902049e-01 5.55899501e-01 4.47683752e-01
-1.86246023e-01 -6.27451241e-02 -4.33778346e-01 1.22231507e+00
6.97333217e-02 -2.12484866e-01 5.85309803e-01 -6.91637248e-02
-1.06723055e-01 5.59419453e-01 -5.85464954e-01 1.74602211e-01
8.02507877e-01 -1.17762111e-01 -2.24783465e-01 -7.18272388e-01
-8.04611504e-01 7.06990659e-01 5.72661996e-01 -4.08023521e-02
2.66529888e-01 -1.05697024e+00 -1.00198305e+00 -3.49193402e-02
-7.73008466e-02 1.72528714e-01 1.10370390e-01 9.56213534e-01
5.66511340e-02 7.55941272e-02 3.65341812e-01 -8.62160087e-01
-1.28093994e+00 6.70916200e-01 1.45024536e-02 -4.26846385e-01
-3.36117148e-01 4.52750891e-01 2.18628794e-01 -4.10828173e-01
3.15742403e-01 2.46351585e-01 -4.12103720e-03 1.42823607e-01
3.69024605e-01 5.73632896e-01 -5.37921607e-01 -4.89817470e-01
-2.88177609e-01 1.19596869e-01 -3.84514928e-02 -8.74526650e-02
1.51203418e+00 -4.89637703e-01 -6.51823580e-01 6.95623279e-01
7.80422866e-01 6.94074705e-02 -1.24844587e+00 -7.33289957e-01
2.65646338e-01 -3.80583137e-01 -1.07972786e-01 -5.00028253e-01
-1.04262018e+00 8.72699738e-01 2.44399104e-02 4.10469681e-01
6.02518201e-01 3.20852727e-01 -6.43237913e-03 9.31273922e-02
4.97050464e-01 -6.36531174e-01 -5.19274063e-02 3.68838698e-01
1.93164304e-01 -1.32236767e+00 3.40248644e-01 -7.13706732e-01
5.26319444e-03 6.09933853e-01 2.88160205e-01 -1.68295596e-02
1.15929663e+00 3.89271647e-01 -4.35161650e-01 -2.37409979e-01
-8.27258229e-01 -7.70767704e-02 2.98368111e-02 7.59884894e-01
3.58742267e-01 8.52534473e-02 -7.13573620e-02 5.05975425e-01
2.16627300e-01 -2.71524489e-01 7.75682926e-01 6.49837136e-01
-5.77715755e-01 -1.28955603e+00 -4.58297312e-01 9.24455881e-01
-5.55876970e-01 -2.25169286e-01 -2.82198131e-01 6.25364363e-01
-1.94995463e-01 1.19097543e+00 2.01762706e-01 1.92192137e-01
-3.05498987e-01 -1.78420216e-01 8.81452411e-02 -1.01501787e+00
-1.71599388e-01 3.15127075e-01 -8.39402229e-02 8.66630599e-02
-6.24280274e-01 -9.67890143e-01 -6.93374813e-01 -8.03520620e-01
-6.94482386e-01 3.27245981e-01 3.72170210e-01 8.66041064e-01
3.01830560e-01 2.15087593e-01 6.91494703e-01 -7.24496603e-01
-5.27774334e-01 -1.09787071e+00 -1.04380631e+00 2.03525499e-01
3.57174188e-01 -7.81456769e-01 -5.85413337e-01 -3.76338139e-02]
|
[6.961796760559082, 5.1672043800354]
|
f822da47-4d3d-46b3-a502-7ea24317644b
|
reconstructing-the-somatotopic-organization
|
2306.05623
| null |
https://arxiv.org/abs/2306.05623v2
|
https://arxiv.org/pdf/2306.05623v2.pdf
|
Reconstructing the somatotopic organization of the corticospinal tract remains a challenge for modern tractography methods
|
The corticospinal tract (CST) is a critically important white matter fiber tract in the human brain that enables control of voluntary movements of the body. Diffusion MRI tractography is the only method that enables the study of the anatomy and variability of the CST pathway in human health. In this work, we explored the performance of six widely used tractography methods for reconstructing the CST and its somatotopic organization. We perform experiments using diffusion MRI data from the Human Connectome Project. Four quantitative measurements including reconstruction rate, the WM-GM interface coverage, anatomical distribution of streamlines, and correlation with cortical volumes to assess the advantages and limitations of each method. Overall, we conclude that while current tractography methods have made progress toward the well-known challenge of improving the reconstruction of the lateral projections of the CST, the overall problem of performing a comprehensive CST reconstruction, including clinically important projections in the lateral (hand and face area) and medial portions (leg area), remains an important challenge for diffusion MRI tractography.
|
["Lauren J. O'Donnell", 'Alexandra J. Golby', 'Ron Kikinis', 'Nikos Makris', 'Yogesh Rathi', 'Erickson Torio', 'Jarrett Rushmore', 'Yuanjing Feng', 'Yiang Pan', 'Fan Zhang', 'Jianzhong He']
|
2023-06-09
| null | null | null | null |
['anatomy']
|
['miscellaneous']
|
[-3.66792530e-01 -1.85782805e-01 -1.63213283e-01 -1.40514210e-01
1.81647837e-01 -9.94080722e-01 4.16660041e-01 -3.92964482e-01
-6.55334651e-01 7.35416889e-01 8.87711346e-01 -6.53354049e-01
-1.91422880e-01 -2.49262288e-01 -1.38371840e-01 -4.99100477e-01
-7.30653524e-01 7.48059154e-01 2.32389405e-01 -7.99625274e-03
3.52579504e-01 9.63173509e-01 -3.76013875e-01 3.16979550e-02
9.77762640e-01 5.44459760e-01 4.98957813e-01 3.07504624e-01
-3.13566536e-01 3.14858973e-01 -4.75600846e-02 1.22659262e-02
3.02901596e-01 -7.80579805e-01 -1.16766477e+00 -2.20755935e-01
1.28191784e-01 -4.34138179e-01 -4.36881721e-01 1.07523060e+00
6.37934685e-01 3.20748389e-02 5.96876681e-01 -5.14957368e-01
-4.27970141e-01 6.45948827e-01 -3.67439806e-01 1.12103641e+00
8.90098736e-02 4.96225476e-01 4.34235811e-01 -6.99069917e-01
1.47138572e+00 8.45299602e-01 5.23787618e-01 6.06063008e-01
-1.50878358e+00 -6.21198714e-01 9.67307463e-02 1.33501813e-01
-1.00069118e+00 -3.08579087e-01 6.26682639e-02 -9.77321267e-01
1.28556156e+00 5.17239012e-02 1.52023065e+00 7.74768233e-01
7.30229676e-01 2.93734312e-01 1.81117964e+00 1.27243578e-01
2.53754649e-02 -5.17886102e-01 1.69116974e-01 2.82818079e-01
9.42015350e-02 2.93322474e-01 -1.92292422e-01 -6.71886057e-02
1.36604548e+00 -3.03358406e-01 -6.22831821e-01 -4.47733253e-01
-1.89320540e+00 4.58859324e-01 6.92618966e-01 1.00376415e+00
-3.99951965e-01 2.34689593e-01 4.33836818e-01 2.72536933e-01
1.33416459e-01 2.42572784e-01 -2.69809961e-01 -7.32175410e-02
-1.25832546e+00 1.80713132e-01 3.85667086e-01 4.78662103e-01
-1.99538693e-01 3.51597220e-02 -1.96161196e-01 6.42911077e-01
4.05461639e-01 3.50475580e-01 6.07897341e-01 -1.14122498e+00
5.70735157e-01 2.57969230e-01 -1.56232819e-01 -3.91389549e-01
-1.12219083e+00 -6.71745658e-01 -7.39221096e-01 5.52224338e-01
6.89145386e-01 -5.83792269e-01 -8.74793947e-01 1.45026863e+00
3.54400873e-01 -6.45089746e-01 -8.00213695e-01 1.35508239e+00
1.54657125e-01 -2.38931134e-01 3.31778437e-01 -1.21864177e-01
9.93376613e-01 -7.07182348e-01 -5.34205854e-01 -3.50187421e-01
5.56815624e-01 -8.22528183e-01 6.43463135e-01 -4.67562005e-02
-1.42685211e+00 2.59579252e-02 -7.08881736e-01 8.59188959e-02
1.24418430e-01 -1.43340126e-01 5.36239982e-01 3.60170245e-01
-1.30738556e+00 8.23833227e-01 -1.33000779e+00 -5.42154551e-01
4.44880009e-01 3.46091032e-01 -7.18182147e-01 -2.23328415e-02
-7.58080900e-01 1.73159182e+00 2.63327688e-01 6.35877028e-02
-7.43365943e-01 -9.87275243e-01 -4.75354977e-02 -5.76821947e-03
-2.52961874e-01 -8.37613523e-01 8.63842070e-01 -3.95754397e-01
-1.11518371e+00 8.78039956e-01 -5.54403514e-02 -1.41719326e-01
6.77785933e-01 2.37451017e-01 -3.83576721e-01 5.19360423e-01
8.24455395e-02 6.79103196e-01 2.67419070e-01 -7.39534855e-01
-1.86239127e-02 -8.25339794e-01 -5.95833004e-01 2.06524327e-01
3.55121195e-01 4.10802722e-01 3.79319400e-01 -9.25917387e-01
8.75610411e-01 -9.84830022e-01 -3.77904803e-01 5.26791632e-01
-2.53902704e-01 2.46130720e-01 1.32929394e-02 -1.34884048e+00
7.02727079e-01 -1.89056945e+00 4.40224737e-01 3.72241199e-01
7.49717414e-01 -1.11151464e-01 -2.86717229e-02 6.65344298e-01
-5.01872361e-01 1.44041076e-01 -2.26361483e-01 4.56003577e-01
-3.88699174e-01 -1.95985720e-01 1.14886880e-01 1.28213930e+00
-5.53131402e-01 1.12814200e+00 -7.01307237e-01 -1.65069953e-01
-1.33018708e-02 2.85805881e-01 -3.34066480e-01 -1.31545752e-01
4.47640955e-01 1.23792219e+00 -2.65473694e-01 1.65495604e-01
6.78641140e-01 1.32250041e-01 6.18339598e-01 -2.55371422e-01
-5.10168910e-01 5.76420367e-01 -7.71430790e-01 2.07049179e+00
6.12212457e-02 6.25821114e-01 4.49254870e-01 -7.98267901e-01
4.96694535e-01 5.13538361e-01 8.76101911e-01 -1.01840985e+00
1.50697365e-01 4.37556952e-01 1.26455903e+00 -4.23323989e-01
-6.14702165e-01 -6.30584538e-01 7.04472244e-01 9.01788950e-01
3.69894743e-01 3.24822869e-03 2.27653548e-01 2.05438197e-01
1.08922815e+00 2.32404470e-01 2.17134226e-02 -1.06989861e+00
9.98796243e-03 2.87225634e-01 7.17452690e-02 8.81610364e-02
-5.33268213e-01 5.12461066e-01 4.60044712e-01 -1.27994075e-01
-1.24123383e+00 -1.68877780e+00 -3.27467710e-01 2.51692355e-01
-3.28387320e-01 2.34930679e-01 -8.58011186e-01 -2.39239618e-01
1.21639468e-01 5.54599702e-01 -5.14587760e-01 1.02416217e-01
-8.78648639e-01 -6.49133205e-01 4.85683292e-01 6.05272472e-01
3.18738371e-01 -1.07660127e+00 -5.86775303e-01 4.75950807e-01
-3.71384412e-01 -9.23771441e-01 -5.82333148e-01 -1.65375695e-02
-1.52291811e+00 -1.11605477e+00 -1.45544863e+00 -6.93985641e-01
7.78719068e-01 3.80232558e-02 6.89461827e-01 5.32829612e-02
-4.47976589e-01 6.66577667e-02 1.01291992e-01 4.69222337e-01
4.31340002e-03 -1.17095031e-01 1.91113483e-02 -8.51518214e-01
-1.11275017e-01 -1.14003885e+00 -1.19407725e+00 4.25617456e-01
-2.72051334e-01 3.63846496e-02 4.39264625e-01 3.95033658e-01
3.48740906e-01 -7.52231538e-01 4.78528351e-01 -2.78395712e-01
1.01685727e+00 -6.26407743e-01 -1.07161239e-01 1.37888297e-01
-5.84166527e-01 -1.39398083e-01 7.46996850e-02 -2.70080328e-01
-9.97566581e-01 -4.63942319e-01 -2.91208059e-01 2.88458288e-01
-5.99732585e-02 3.49819869e-01 3.90857160e-01 -5.58003724e-01
6.97389066e-01 3.05976093e-01 4.07247990e-01 -6.32349193e-01
3.98840964e-01 -7.23049510e-03 4.72762823e-01 -3.49682152e-01
3.48582149e-01 7.19728649e-01 -9.56927463e-02 -4.59538937e-01
-4.01983373e-02 -5.64170063e-01 -1.34692252e+00 -5.30896187e-01
8.91694546e-01 -3.43738019e-01 -6.81813121e-01 -1.36144117e-01
-9.66769338e-01 -6.83806181e-01 -2.20179692e-01 1.06692433e+00
-6.72293663e-01 4.09063816e-01 -1.02295327e+00 -2.10616156e-01
-6.53241992e-01 -1.25937665e+00 2.13941038e-01 -4.50841576e-01
-7.59535611e-01 -1.12431681e+00 3.91549230e-01 1.06507465e-02
7.63023973e-01 2.48459399e-01 1.27302194e+00 -1.15210466e-01
-2.06556618e-01 3.54020596e-01 -2.69796252e-01 -3.83264534e-02
-5.54888733e-02 -4.91802543e-01 -1.67690322e-01 -4.24985021e-01
3.22817594e-01 1.91641524e-01 6.30867004e-01 7.34620333e-01
4.53794509e-01 9.92162898e-02 -3.49911332e-01 5.12481093e-01
1.53005469e+00 1.80738330e-01 4.44468260e-01 2.63000309e-01
3.17634434e-01 9.41448271e-01 -5.76685630e-02 -3.35693628e-01
1.21102221e-01 2.15277597e-01 3.05265537e-03 3.41260470e-02
-9.61269200e-01 2.14979827e-01 -1.26252621e-02 1.21712029e+00
-6.67705476e-01 5.12468457e-01 -1.11127687e+00 7.81170547e-01
-1.18652999e+00 -7.95086265e-01 -5.56403279e-01 1.92956090e+00
7.60156512e-01 9.23201889e-02 4.83710855e-01 -2.62816966e-01
4.16714996e-01 -2.53129452e-01 -7.01186359e-01 -2.29187697e-01
1.11566149e-01 2.07543209e-01 7.18675196e-01 4.99693602e-01
-9.48988497e-02 5.16003907e-01 8.58128071e+00 1.44724026e-01
-1.24608874e+00 7.12941170e-01 -9.52640623e-02 -4.64072883e-01
-4.66728091e-01 -1.64132535e-01 -3.21910113e-01 3.39423448e-01
7.59600341e-01 -1.43454060e-01 1.12902987e+00 1.70738608e-01
5.82087398e-01 -3.70021105e-01 -6.86901152e-01 3.44052345e-01
-5.61254382e-01 -1.30148077e+00 -5.59745908e-01 4.18483585e-01
7.64788210e-01 9.79297042e-01 -8.95408913e-02 -3.18021446e-01
1.73413709e-01 -1.07665288e+00 7.90437758e-01 6.89887524e-01
1.04623520e+00 -2.80000627e-01 1.72509104e-01 3.01748902e-01
-8.37121308e-01 4.22005542e-02 -8.48375931e-02 -4.25128229e-02
6.30463481e-01 6.50089204e-01 -4.13678080e-01 -7.41667449e-02
6.43360078e-01 3.02641898e-01 -5.82311861e-02 1.65729713e+00
-2.85376549e-01 4.54832256e-01 -7.69014284e-02 4.33071822e-01
2.88723260e-01 -3.68746072e-01 7.16329992e-01 1.03297353e+00
2.50632465e-01 1.37865603e-01 -2.05100343e-01 1.26322329e+00
1.94574237e-01 1.22158132e-01 -2.40996420e-01 -3.18221211e-01
1.86803430e-01 1.27795863e+00 -1.25378919e+00 -1.44539448e-02
-1.75151557e-01 6.97841287e-01 6.12713039e-01 7.76016474e-01
-2.01538429e-01 3.06868374e-01 6.60721898e-01 3.75681460e-01
2.68246382e-02 -1.05077493e+00 -8.12839866e-01 -9.45619702e-01
4.39088307e-02 -6.30467713e-01 -9.00092162e-03 -6.65412009e-01
-1.09202039e+00 4.47438270e-01 -1.96644470e-01 -6.79778874e-01
-1.21807389e-01 -4.72240895e-01 -5.36394894e-01 1.63938713e+00
-1.02199233e+00 -6.86731875e-01 9.58412364e-02 7.06023633e-01
2.25255579e-01 3.99009854e-01 7.49701977e-01 3.64587337e-01
-4.89180461e-02 -5.14817797e-02 2.80723006e-01 8.02545249e-02
3.64393950e-01 -1.22657704e+00 6.00612760e-01 6.67869747e-01
-3.05107087e-01 1.20025563e+00 6.69501424e-01 -1.24513018e+00
-9.96018291e-01 -5.75796187e-01 6.04395926e-01 -9.62698236e-02
1.12482154e+00 -2.32712626e-01 -6.12972558e-01 1.03060424e+00
1.83773398e-01 -1.52802691e-01 5.78707993e-01 -1.10323235e-01
1.29845977e-01 4.88656104e-01 -1.27977622e+00 6.87877893e-01
1.39908648e+00 -3.46018016e-01 -8.22729051e-01 4.08015460e-01
2.29828924e-01 -3.06895524e-01 -1.44154549e+00 -4.71015573e-02
8.34967077e-01 -7.41851985e-01 1.12936246e+00 -5.35398960e-01
2.51013309e-01 -5.09818569e-02 4.58040416e-01 -1.70223594e+00
-6.50166988e-01 -1.56380832e-02 6.54620305e-02 5.71407676e-01
2.04300269e-01 -7.87992001e-01 6.28714383e-01 5.46741009e-01
-2.93343246e-01 -6.88277543e-01 -1.13149297e+00 -8.27010930e-01
8.13001156e-01 3.20690088e-02 1.19281746e-01 8.78145874e-01
5.50909162e-01 -1.06767468e-01 2.72340745e-01 -5.85390449e-01
7.19404161e-01 6.06535235e-03 -3.07023376e-01 -8.78815889e-01
8.49420112e-03 -9.27089453e-01 -1.47748902e-01 -7.66953707e-01
-3.41953546e-01 -1.57788599e+00 -2.76110142e-01 -2.36303520e+00
1.11803822e-01 -3.81438226e-01 1.84691623e-01 -2.28784177e-02
4.29663718e-01 4.77355495e-02 2.30365157e-01 4.61820006e-01
5.00416994e-01 1.63754553e-01 2.38610649e+00 3.47040445e-01
1.14308767e-01 -4.69080597e-01 -4.28427339e-01 7.22587764e-01
1.05599535e+00 -5.43679714e-01 -3.76342297e-01 -7.20291317e-01
-2.07715318e-01 2.66468078e-01 2.16692969e-01 -8.30868363e-01
3.56990546e-01 -3.71200889e-02 8.62636745e-01 -4.44796026e-01
7.32256100e-02 -7.48956323e-01 1.68138877e-01 1.10918677e+00
-2.23720506e-01 2.89878577e-01 -1.62264463e-02 -4.41332161e-02
3.67464244e-01 -2.40779314e-02 8.63090038e-01 -6.71735346e-01
-3.01568121e-01 2.94747293e-01 -1.01821911e+00 3.50939512e-01
8.82799029e-01 -1.81898415e-01 -2.95447826e-01 3.58276702e-02
-1.57395160e+00 1.60225481e-01 3.53493363e-01 1.03562973e-01
4.69780564e-01 -1.30647993e+00 -6.12180531e-01 -8.78481418e-02
-5.77603757e-01 -9.49958324e-01 4.47107345e-01 1.87246704e+00
-8.95962596e-01 7.96418130e-01 -1.03174710e+00 -3.35779756e-01
-6.08603001e-01 5.23420572e-01 8.17328274e-01 -2.18259647e-01
-1.27449811e+00 6.46996617e-01 -1.05593570e-01 -6.11606598e-01
-1.08154655e-01 -4.24812317e-01 -2.95221806e-01 -1.80147395e-01
4.24378932e-01 7.26633430e-01 -1.24482274e-01 -7.17243731e-01
-3.29075485e-01 4.45079654e-01 3.98564517e-01 -5.40466607e-01
1.19867146e+00 -6.57266378e-01 -6.76358700e-01 2.50884682e-01
9.57133651e-01 -1.90210134e-01 -8.93938541e-01 1.88253492e-01
4.50702012e-02 -1.66551396e-01 1.98959857e-01 -1.36537385e+00
-1.22997677e+00 1.21603048e+00 8.61446381e-01 -7.71030188e-02
5.89127183e-01 -9.68093649e-02 9.71690178e-01 -4.77527887e-01
5.53222895e-01 -8.89065027e-01 -5.43729722e-01 1.76347315e-01
1.28588033e+00 -4.10943449e-01 -1.07190803e-01 -4.06837583e-01
-3.94201517e-01 9.94679749e-01 5.05292833e-01 -3.79684776e-01
9.24546719e-01 3.67173582e-01 -9.30131041e-03 -7.05215156e-01
-2.71268189e-02 -6.47512674e-02 5.01375973e-01 7.05130160e-01
7.67647982e-01 2.11873561e-01 -1.32496047e+00 3.87517852e-04
-3.69678319e-01 3.64353955e-02 2.37410083e-01 6.97153449e-01
-5.18220246e-01 -9.62196052e-01 -1.84021071e-01 8.80684495e-01
-5.43483555e-01 -2.75006384e-01 -2.99231291e-01 7.79513836e-01
-1.29778255e-02 5.26883006e-01 -1.29205316e-01 1.46426663e-01
2.52238661e-01 1.27013549e-01 9.70582306e-01 -6.50160968e-01
-8.96876156e-01 3.28035623e-01 1.56719714e-01 -7.73117900e-01
-1.76245704e-01 -1.03764033e+00 -1.81263733e+00 -4.79619235e-01
1.18116193e-01 -1.97313949e-01 1.01550198e+00 1.39094496e+00
-4.81506921e-02 7.07681417e-01 -1.61977097e-01 -7.47776270e-01
-3.29058021e-01 -1.15426350e+00 -1.01293087e+00 5.59750348e-02
1.51660204e-01 -7.52050340e-01 8.16806871e-03 -3.87627810e-01]
|
[14.011828422546387, -2.2482261657714844]
|
9a2b6815-ef4e-413c-80bd-963213968ea4
|
acrobat-a-multi-stain-breast-cancer
|
2211.13621
| null |
https://arxiv.org/abs/2211.13621v1
|
https://arxiv.org/pdf/2211.13621v1.pdf
|
ACROBAT -- a multi-stain breast cancer histological whole-slide-image data set from routine diagnostics for computational pathology
|
The analysis of FFPE tissue sections stained with haematoxylin and eosin (H&E) or immunohistochemistry (IHC) is an essential part of the pathologic assessment of surgically resected breast cancer specimens. IHC staining has been broadly adopted into diagnostic guidelines and routine workflows to manually assess status and scoring of several established biomarkers, including ER, PGR, HER2 and KI67. However, this is a task that can also be facilitated by computational pathology image analysis methods. The research in computational pathology has recently made numerous substantial advances, often based on publicly available whole slide image (WSI) data sets. However, the field is still considerably limited by the sparsity of public data sets. In particular, there are no large, high quality publicly available data sets with WSIs of matching IHC and H&E-stained tissue sections. Here, we publish the currently largest publicly available data set of WSIs of tissue sections from surgical resection specimens from female primary breast cancer patients with matched WSIs of corresponding H&E and IHC-stained tissue, consisting of 4,212 WSIs from 1,153 patients. The primary purpose of the data set was to facilitate the ACROBAT WSI registration challenge, aiming at accurately aligning H&E and IHC images. For research in the area of image registration, automatic quantitative feedback on registration algorithm performance remains available through the ACROBAT challenge website, based on more than 37,000 manually annotated landmark pairs from 13 annotators. Beyond registration, this data set has the potential to enable many different avenues of computational pathology research, including stain-guided learning, virtual staining, unsupervised pre-training, artefact detection and stain-independent models.
|
['Mattias Rantalainen', 'Pekka Ruusuvuori', 'Johan Hartman', 'Anne-Vibeke Laenkholm', 'Leena Latonen', 'Kajsa Ledesma Eriksson', 'Abhinav Sharma', 'Sandra Kristiane Sinius Pouplier', 'Yanbo Feng', 'Dusan Rasic', 'Aino Kuusela', 'Sonja Koivukoski', 'Constance Boissin', 'Kimmo Kartasalo', 'Circe Carr', 'Leslie Solorzano', 'Masi Valkonen', 'Philippe Weitz']
|
2022-11-24
| null | null | null | null |
['unsupervised-pre-training']
|
['methodology']
|
[ 2.91996479e-01 1.85167730e-01 -4.94066685e-01 -1.66673526e-01
-1.47441304e+00 -7.60262907e-01 1.31934196e-01 8.47533345e-01
-6.77432954e-01 4.93091375e-01 1.67754859e-01 -6.22036278e-01
-1.91461354e-01 -5.51440358e-01 -1.24979824e-01 -1.23079157e+00
-8.20879489e-02 8.75112474e-01 7.79528618e-02 -1.84948072e-02
-1.17473245e-01 8.52004170e-01 -8.57698441e-01 1.50025859e-01
2.18846634e-01 8.06363702e-01 1.33109719e-01 9.03319240e-01
-9.27890316e-02 2.67111778e-01 -5.24339750e-02 -5.18206239e-01
6.07436486e-02 -1.48244530e-01 -8.18246543e-01 -3.26718271e-01
2.04052791e-01 1.68272480e-01 -1.01809949e-01 9.21901345e-01
6.60160840e-01 -6.32398963e-01 6.61187708e-01 -1.04092407e+00
-2.49274131e-02 3.86684984e-01 -7.74731338e-01 4.21771795e-01
-7.86402747e-02 3.97413999e-01 1.03367031e+00 -6.08110309e-01
1.28606093e+00 3.93398046e-01 8.93412471e-01 2.86178589e-01
-1.37424827e+00 -7.57838488e-01 -9.21112597e-01 3.51489373e-02
-1.66538680e+00 -2.10459158e-01 2.37518936e-01 -5.42294085e-01
6.89062536e-01 5.83074093e-01 8.81995380e-01 2.66736954e-01
4.68440413e-01 4.71265972e-01 1.26653099e+00 -4.54700410e-01
-7.56659964e-03 1.91342264e-01 7.21674263e-02 7.62304008e-01
3.32797974e-01 7.42772548e-03 -1.04248613e-01 -3.34106028e-01
5.05389154e-01 1.87559649e-01 -3.06248814e-01 -2.32676715e-01
-1.33864415e+00 5.53935766e-01 5.07762134e-01 6.73268616e-01
8.40317905e-02 -2.69792944e-01 7.64913678e-01 3.09411883e-01
2.97708988e-01 2.45098427e-01 -2.00014651e-01 2.74829119e-01
-1.02967894e+00 -1.60774067e-01 4.42175657e-01 3.84142309e-01
6.41263604e-01 -8.70228767e-01 1.13557270e-02 7.12625980e-01
4.08632934e-01 5.33958673e-01 7.58809984e-01 -3.55012327e-01
-8.37318450e-02 8.34332824e-01 -1.41032293e-01 -8.90437663e-01
-1.08970904e+00 -2.36830607e-01 -1.13201737e+00 1.90587699e-01
7.02022552e-01 3.50538343e-01 -8.28800738e-01 1.22193050e+00
4.07752663e-01 -2.30133057e-01 -9.41536501e-02 9.32423353e-01
8.81108940e-01 -1.22563303e-01 3.71060759e-01 -1.20721266e-01
1.67509830e+00 -4.50705588e-01 -6.01237953e-01 1.68323696e-01
1.38111043e+00 -8.42668176e-01 8.21221054e-01 -4.26280200e-02
-7.89711297e-01 7.22276345e-02 -9.90791023e-01 -8.86284411e-02
-7.60952294e-01 3.35104644e-01 8.90114665e-01 4.44577456e-01
-1.25348818e+00 1.36104792e-01 -1.10469174e+00 -8.42596471e-01
9.40495253e-01 7.18285978e-01 -1.28100657e+00 -7.45446384e-02
-6.09812379e-01 1.06174028e+00 2.10064221e-02 2.00075656e-01
-5.35750866e-01 -1.12539279e+00 -6.44220352e-01 -4.32152212e-01
-1.82120696e-01 -3.76390696e-01 9.06595290e-01 -7.46881843e-01
-9.50513482e-01 1.91920316e+00 -1.81011364e-01 -2.43597180e-01
2.41170764e-01 9.89862740e-01 -4.48749423e-01 2.63042748e-01
1.96358293e-01 5.02131402e-01 -1.85312331e-01 -9.30932105e-01
-7.43903160e-01 -8.15809786e-01 -6.49755716e-01 -1.03288889e-01
-1.52367368e-01 4.56576385e-02 -3.10143560e-01 -3.35504681e-01
-1.66131333e-02 -9.96927321e-01 -3.88638109e-01 4.81053948e-01
-3.87965769e-01 9.99249816e-02 4.54241157e-01 -8.41895640e-01
8.78126681e-01 -2.43896794e+00 -2.37119734e-01 7.55071342e-01
4.38904613e-01 1.79725528e-01 9.21311136e-03 4.16525155e-02
-2.14175358e-01 5.72815895e-01 3.24041359e-02 -2.37569004e-01
4.28928137e-02 -5.50547056e-02 4.38639253e-01 1.21058774e+00
-5.70285842e-02 1.21192431e+00 -9.09703255e-01 -9.20402229e-01
-4.40039337e-02 2.45597258e-01 -2.23321766e-02 1.03911363e-01
6.09248400e-01 3.60139489e-01 2.81545706e-02 8.80039573e-01
4.58353966e-01 -4.72118497e-01 5.60143828e-01 -4.66762573e-01
2.57289279e-02 -2.56688237e-01 -6.24525726e-01 1.44085872e+00
-1.71996206e-01 8.03468585e-01 5.22731245e-01 -6.60994172e-01
6.45410419e-01 4.92590278e-01 7.27752805e-01 -4.87738520e-01
3.22367251e-01 4.83116657e-01 2.12931842e-01 -2.96764135e-01
-3.54000479e-02 -5.76510847e-01 1.62270874e-01 3.39200795e-01
3.14055681e-02 -2.12616235e-01 3.24973792e-01 -5.29690124e-02
1.60441351e+00 -5.93929231e-01 6.71995223e-01 -3.66198242e-01
5.27392387e-01 5.76508045e-01 4.16734636e-01 1.49070904e-01
-7.42680550e-01 6.37714148e-01 4.79485154e-01 -3.74653399e-01
-1.18299973e+00 -1.12239468e+00 -7.14062691e-01 4.71897185e-01
-4.80922833e-02 -6.41702041e-02 -1.19146988e-01 -4.91351187e-01
9.12681520e-02 -3.09153020e-01 -1.18137634e+00 2.30617180e-01
-2.31115714e-01 -1.24955332e+00 8.45577955e-01 3.24722111e-01
-1.11725871e-02 -7.50535667e-01 -2.94682980e-01 1.40502304e-01
4.06592488e-02 -8.04008007e-01 -3.21373791e-01 5.62232375e-01
-7.59949625e-01 -1.64932048e+00 -8.00726175e-01 -1.14353704e+00
1.24532950e+00 1.74281403e-01 9.97617185e-01 5.89918852e-01
-1.06252706e+00 1.49205089e-01 -2.18297854e-01 -5.65578163e-01
-6.90468431e-01 5.30854799e-02 -3.83788019e-01 -8.92915353e-02
5.09947181e-01 -2.86887228e-01 -6.93675637e-01 3.72801483e-01
-9.93885636e-01 5.91803230e-02 1.15517795e+00 1.04642701e+00
1.19391835e+00 -1.77958295e-01 2.67237067e-01 -1.11840785e+00
-4.93139029e-03 -4.39081132e-01 -4.50854778e-01 3.76951784e-01
-3.90210271e-01 -5.82246661e-01 4.07945275e-01 6.50252700e-02
-7.85977364e-01 -5.30030280e-02 -2.49979660e-01 1.93729848e-01
-2.14103162e-01 8.40232015e-01 3.37656677e-01 -4.92451787e-01
6.28099144e-01 -2.56667286e-01 7.20850050e-01 1.65018499e-01
-3.43976378e-01 6.20709479e-01 7.09557354e-01 1.90403104e-01
8.97804558e-01 9.26256537e-01 4.08619255e-01 -4.88295078e-01
-4.38023865e-01 -1.19087875e+00 -6.23469412e-01 9.31659713e-02
7.96020508e-01 -6.71657324e-01 -6.56478822e-01 2.82398641e-01
-3.41199577e-01 -7.15991378e-01 -1.36265576e-01 4.08434629e-01
-2.96268046e-01 8.36115777e-02 -8.84857297e-01 -3.16227488e-02
-6.38373733e-01 -1.17704368e+00 1.17415750e+00 2.52663225e-01
-5.73434889e-01 -1.29407537e+00 5.74975431e-01 2.95090854e-01
5.54260254e-01 5.75857878e-01 1.12461329e+00 -7.44156897e-01
6.90365955e-03 -8.51413906e-01 -2.89905697e-01 -3.13827455e-01
3.90172690e-01 5.99998176e-01 -6.93619251e-01 -2.41993964e-01
-6.73311710e-01 -2.11090386e-01 5.51741064e-01 3.28804553e-01
7.18839049e-01 -6.49517700e-02 -9.94090259e-01 8.39113772e-01
1.74602580e+00 -1.73793864e-02 6.95426702e-01 5.82296252e-01
3.46283644e-01 6.99747860e-01 6.85398638e-01 8.22610483e-02
2.97446132e-01 4.75587279e-01 3.67171109e-01 -8.28059256e-01
-1.47844732e-01 9.45645869e-02 -3.99815023e-01 2.24342778e-01
-1.17296413e-01 1.31560832e-01 -1.42344809e+00 9.01218653e-01
-1.13181281e+00 -7.59856105e-01 -2.20677987e-01 1.95465434e+00
1.10343432e+00 -1.05842836e-01 -1.74108967e-01 3.48672688e-01
5.60419679e-01 -3.63947153e-01 -1.12393171e-01 3.24661355e-03
-2.35618562e-01 3.17434281e-01 7.52054334e-01 2.72675157e-01
-1.09891081e+00 4.06032711e-01 6.49814510e+00 7.62082994e-01
-1.29360652e+00 2.05771346e-02 1.09223127e+00 2.24320367e-02
-8.45952556e-02 -1.98520362e-01 -6.43927634e-01 1.64161593e-01
6.86676502e-01 -3.11666757e-01 -6.60010278e-02 3.76854181e-01
2.82295495e-01 -4.64021772e-01 -1.05710542e+00 8.09375584e-01
-2.50540704e-01 -1.74284804e+00 -4.10409063e-01 5.25047004e-01
5.58136463e-01 2.35841960e-01 1.51547007e-02 -9.74428952e-02
2.68262118e-01 -1.21650922e+00 -1.11402363e-01 3.96789640e-01
1.42481065e+00 -3.88735175e-01 1.77010202e+00 -1.49618968e-01
-8.80006969e-01 3.94880474e-01 -7.02468604e-02 5.97177386e-01
-2.44357675e-01 5.52015007e-01 -1.54449499e+00 3.84031653e-01
7.37245798e-01 6.12087071e-01 -9.33571517e-01 1.08191657e+00
1.96037427e-01 5.37694633e-01 -2.75962353e-01 3.18293869e-02
-1.29086956e-01 1.58502668e-01 -2.23547744e-04 1.33408296e+00
2.03725830e-01 1.93595305e-01 -2.59847730e-01 2.64187783e-01
2.04113632e-01 4.08931047e-01 -2.59877235e-01 -9.20901895e-02
3.90792847e-01 2.23097348e+00 -1.28185058e+00 -2.57798396e-02
-5.11031449e-01 2.62938797e-01 2.63934374e-01 1.23988114e-01
-5.18525958e-01 -1.20371938e-01 3.86650860e-01 4.56268668e-01
-3.17662627e-01 2.14550421e-01 -3.49800795e-01 -4.99613971e-01
-5.81304431e-01 -6.43202126e-01 7.62786329e-01 -4.77839500e-01
-1.48269272e+00 2.76791543e-01 -3.72449785e-01 -1.18643653e+00
1.37483850e-01 -9.27020907e-01 -6.39984012e-01 8.22647095e-01
-1.75232363e+00 -1.38210821e+00 -6.25620186e-01 1.88303456e-01
-3.88642788e-01 -1.73891298e-02 1.32471991e+00 2.05968097e-01
-4.49926883e-01 7.94441342e-01 3.75668377e-01 4.75988984e-01
1.15284812e+00 -1.34905314e+00 -5.32353520e-01 -1.78749692e-02
-4.43186283e-01 4.60763276e-01 4.91281092e-01 -3.14229071e-01
-1.38602233e+00 -1.12815988e+00 7.54739225e-01 -4.55356479e-01
1.01892555e+00 9.40445587e-02 -5.96407473e-01 9.09815609e-01
-4.18454930e-02 6.13630116e-01 1.87873006e+00 -2.30856150e-01
7.43992776e-02 -2.16584966e-01 -1.48261166e+00 5.42445123e-01
3.70019674e-01 -2.78304905e-01 4.53711152e-02 3.95998001e-01
-3.72628331e-01 -6.16128147e-01 -1.59456944e+00 4.79641765e-01
7.40930974e-01 -5.71521223e-01 7.42125034e-01 -1.20054863e-01
1.17900282e-01 -4.24342692e-01 1.35777310e-01 -1.07665551e+00
-4.00858402e-01 -6.48186132e-02 8.39208305e-01 1.06306863e+00
7.24253118e-01 -7.37297714e-01 1.02651477e+00 6.11952484e-01
-2.04321682e-01 -8.85134518e-01 -1.11390817e+00 -2.46092528e-01
1.01749904e-01 1.72319606e-01 3.70431930e-01 6.91774189e-01
5.68597794e-01 -3.72435719e-01 7.77305841e-01 -5.43868616e-02
4.02450323e-01 -1.66163757e-01 9.38632965e-01 -1.10665500e+00
-7.56192431e-02 -8.50049436e-01 -9.81328905e-01 1.75440103e-01
7.18654916e-02 -1.39513290e+00 -7.01970980e-02 -1.59292912e+00
7.12923646e-01 -7.29148746e-01 -3.41467857e-01 7.89969921e-01
-1.77798316e-01 9.26437259e-01 -4.33467627e-01 6.02323592e-01
-5.46269715e-01 -2.51953840e-01 1.42415154e+00 -4.64549035e-01
1.09803423e-01 -4.79776204e-01 -7.87685573e-01 6.22563958e-01
7.30547130e-01 -5.33221483e-01 2.64232993e-01 3.02632660e-01
2.14565709e-01 -7.76877999e-02 5.99483550e-01 -7.62291908e-01
3.51770252e-01 -8.44435841e-02 6.61137342e-01 -5.75085044e-01
-1.12326495e-01 -9.23170090e-01 6.69045925e-01 5.76668620e-01
-2.38080353e-01 -1.30876482e-01 3.06361645e-01 2.04772815e-01
-3.92369121e-01 2.66011991e-02 7.86704779e-01 -1.15597874e-01
-3.55031312e-01 4.80146497e-01 -4.10177112e-01 -3.61503422e-01
1.26772940e+00 -5.12249470e-01 -6.54851079e-01 1.35696247e-01
-7.53729820e-01 2.52398908e-01 8.36405039e-01 -3.23498279e-01
1.67738393e-01 -1.20860839e+00 -9.00763690e-01 -8.19451436e-02
6.01746619e-01 2.41272375e-01 4.68170851e-01 1.65047979e+00
-8.73809457e-01 4.69193667e-01 -2.37341747e-01 -6.57803237e-01
-1.57063317e+00 3.51469278e-01 3.99685919e-01 -1.05768871e+00
-3.56141031e-01 7.78444529e-01 1.19661815e-01 -3.88680220e-01
-1.12420753e-01 8.14772397e-02 -2.70520627e-01 -1.05214929e-02
5.73834121e-01 2.48003732e-02 3.61405313e-01 -7.84402132e-01
-4.15242404e-01 3.76625240e-01 -3.00844282e-01 1.59841776e-01
1.25185025e+00 1.68040454e-01 -5.61382532e-01 3.45885992e-01
1.50767136e+00 1.22023024e-01 -5.84744215e-01 -1.06945746e-02
5.22618033e-02 -2.31328383e-01 5.09977005e-02 -8.69126379e-01
-1.11235309e+00 2.68234879e-01 8.29420090e-01 -3.70346382e-02
1.08828366e+00 2.19807103e-01 3.74239057e-01 -2.27820128e-01
3.18634540e-01 -7.57920504e-01 -3.68649125e-01 -1.16986178e-01
5.73999345e-01 -1.60646069e+00 4.34007421e-02 -5.65158546e-01
-3.01534720e-02 1.14889431e+00 2.93501735e-01 4.14359011e-02
6.92590654e-01 8.40091169e-01 5.59000969e-01 -4.86947775e-01
-5.31857967e-01 -1.98255554e-02 1.90750994e-02 7.24467993e-01
8.66926372e-01 2.20617637e-01 -5.59274971e-01 8.79935205e-01
-2.78551847e-01 9.69092697e-02 2.73382068e-01 8.39677632e-01
-5.74403070e-02 -1.00069416e+00 -2.52289861e-01 8.47045362e-01
-8.00168931e-01 9.15308595e-02 -3.95556211e-01 1.34053385e+00
-2.29183361e-01 4.87302423e-01 5.70704453e-02 6.36654124e-02
1.31341487e-01 -3.27254266e-01 2.95176804e-01 -4.48541909e-01
-7.16160119e-01 -5.84053807e-02 -8.78686830e-02 -1.35380149e-01
-6.29573226e-01 -8.22933495e-01 -1.51639569e+00 -3.14789176e-01
-3.12414676e-01 1.93672404e-01 6.16057336e-01 8.00002396e-01
3.44740860e-02 4.50767159e-01 5.30102313e-01 -6.04444265e-01
7.04317316e-02 -8.79362404e-01 -8.95576954e-01 4.39927787e-01
2.80854672e-01 -4.03264374e-01 -6.77164674e-01 2.90167689e-01]
|
[15.090856552124023, -3.064809799194336]
|
bd5b3913-0fd7-4c5d-8682-1ffa673b08ad
|
adversarial-skill-networks-unsupervised-robot
|
1910.09430
| null |
https://arxiv.org/abs/1910.09430v2
|
https://arxiv.org/pdf/1910.09430v2.pdf
|
Adversarial Skill Networks: Unsupervised Robot Skill Learning from Video
|
Key challenges for the deployment of reinforcement learning (RL) agents in the real world are the discovery, representation and reuse of skills in the absence of a reward function. To this end, we propose a novel approach to learn a task-agnostic skill embedding space from unlabeled multi-view videos. Our method learns a general skill embedding independently from the task context by using an adversarial loss. We combine a metric learning loss, which utilizes temporal video coherence to learn a state representation, with an entropy regularized adversarial skill-transfer loss. The metric learning loss learns a disentangled representation by attracting simultaneous viewpoints of the same observations and repelling visually similar frames from temporal neighbors. The adversarial skill-transfer loss enhances re-usability of learned skill embeddings over multiple task domains. We show that the learned embedding enables training of continuous control policies to solve novel tasks that require the interpolation of previously seen skills. Our extensive evaluation with both simulation and real world data demonstrates the effectiveness of our method in learning transferable skills from unlabeled interaction videos and composing them for new tasks. Code, pretrained models and dataset are available at http://robotskills.cs.uni-freiburg.de
|
['Wolfram Burgard', 'Markus Merklinger', 'Oier Mees', 'Gabriel Kalweit']
|
2019-10-21
| null | null | null | null |
['video-alignment']
|
['computer-vision']
|
[ 2.40627959e-01 1.01231970e-01 1.43792229e-02 -6.89952299e-02
-5.78259051e-01 -8.99935484e-01 5.52210987e-01 -5.25175154e-01
-6.21088386e-01 1.07840598e+00 1.96658954e-01 3.00449431e-01
-2.49320582e-01 -3.23130935e-01 -1.11761427e+00 -8.89313877e-01
-4.57804471e-01 3.60374600e-01 1.21336676e-01 -3.56471390e-01
7.49926791e-02 4.08631861e-01 -1.58691108e+00 -2.71459785e-03
7.63142288e-01 7.72755146e-01 5.43698609e-01 8.61897945e-01
6.85004413e-01 1.34814465e+00 -4.79399711e-01 5.73618822e-02
6.99277520e-01 -4.95103002e-01 -7.36122429e-01 -2.71181893e-02
5.91426253e-01 -6.05685353e-01 -8.36323857e-01 9.04784203e-01
2.84011066e-01 6.65699244e-01 5.57475805e-01 -1.43009508e+00
-7.69791186e-01 6.34806678e-02 -4.24936712e-02 1.41649753e-01
5.71996570e-01 6.06986284e-01 9.10548866e-01 -5.19185781e-01
9.74511206e-01 1.11467934e+00 3.69946122e-01 1.13303041e+00
-1.35737574e+00 -7.05523908e-01 1.63914114e-01 1.86357662e-01
-8.92641783e-01 -8.55052099e-02 6.89240396e-01 -6.13647819e-01
9.02284682e-01 -6.15519546e-02 9.10667479e-01 1.68506372e+00
2.94589847e-01 8.21137846e-01 1.33543324e+00 -8.90858695e-02
1.57923862e-01 8.46931711e-02 -5.72767556e-01 9.62641597e-01
4.69285622e-02 9.02102232e-01 -7.37343788e-01 4.47997451e-02
1.27032864e+00 2.86007643e-01 -4.16853994e-01 -1.27011597e+00
-1.39110518e+00 7.98199952e-01 5.88154435e-01 7.68591464e-02
-2.27189332e-01 6.73431516e-01 3.96795869e-01 8.92973363e-01
1.86088130e-01 9.27480400e-01 -5.91519117e-01 -2.25108087e-01
-4.27372575e-01 4.16239202e-01 6.95252299e-01 1.02593613e+00
6.95130527e-01 2.99987674e-01 -1.94751754e-01 2.23850399e-01
-6.53694198e-03 8.25943470e-01 5.16802788e-01 -1.54552126e+00
3.84424627e-01 2.61609852e-01 3.08901846e-01 -5.34572721e-01
-1.20891780e-01 -1.79700866e-01 -2.44154707e-01 1.06291533e+00
3.68969738e-01 -4.04461473e-01 -8.76225829e-01 2.24565363e+00
3.71193811e-02 6.50032699e-01 3.62620324e-01 9.81267631e-01
1.07212067e-01 4.48691726e-01 1.56502961e-03 1.20608583e-01
6.98793471e-01 -1.11691356e+00 -5.22953928e-01 -3.49707663e-01
2.51386404e-01 -1.88481197e-01 1.20091701e+00 2.99022704e-01
-1.16384459e+00 -6.36896729e-01 -1.11113977e+00 2.24650018e-02
-2.15636984e-01 -2.63178796e-01 3.49577606e-01 -4.78486391e-03
-1.02156293e+00 1.03390610e+00 -1.04902720e+00 -3.58166575e-01
4.48886812e-01 5.39328992e-01 -7.89401829e-01 3.44791263e-02
-1.15116143e+00 1.17891943e+00 3.52857977e-01 -4.41074491e-01
-1.75090289e+00 -6.97685003e-01 -1.15891767e+00 -3.05872634e-02
5.27256787e-01 -8.08169842e-01 1.21385276e+00 -1.47915578e+00
-1.88972831e+00 7.40742445e-01 5.41008770e-01 -5.46573699e-01
5.43320119e-01 -4.18636411e-01 -1.24886177e-01 4.80682373e-01
1.24156751e-01 6.42405629e-01 1.30196905e+00 -1.40544581e+00
-5.61260521e-01 -2.32815862e-01 5.70079684e-01 6.57916367e-01
-3.27869475e-01 -5.91915786e-01 2.65295595e-01 -7.74251223e-01
-6.29018009e-01 -1.21825373e+00 -1.88524872e-01 2.37241566e-01
2.32075095e-01 1.95546985e-01 9.22430754e-01 -5.73458195e-01
4.59522098e-01 -2.18511200e+00 1.08707702e+00 -7.31673539e-02
2.04433605e-01 -1.46216722e-02 -4.37217236e-01 5.27120650e-01
-4.62555103e-02 -5.17405570e-01 -6.49384335e-02 -1.10060677e-01
-1.53837297e-02 4.91914243e-01 -4.49787319e-01 5.55272579e-01
2.72739142e-01 1.02710080e+00 -1.56634545e+00 -1.70198709e-01
2.33608574e-01 4.57524836e-01 -7.34380722e-01 7.35812962e-01
-3.17613572e-01 1.16787755e+00 -5.23873329e-01 1.30245015e-01
-5.00672907e-02 -1.40577719e-01 3.21136713e-01 1.69955626e-01
2.48335391e-01 -4.56916951e-02 -8.79215181e-01 2.40183902e+00
-6.62792087e-01 5.59482634e-01 2.17925206e-01 -1.04925013e+00
6.38277650e-01 3.52340937e-01 5.87102652e-01 -4.24672365e-01
4.89395708e-02 3.06612812e-02 -2.01083466e-01 -6.76527619e-01
2.49148220e-01 -2.30856508e-01 -3.71821940e-01 4.48314607e-01
8.87456059e-01 -6.23332620e-01 3.72914150e-02 1.96490005e-01
1.30271542e+00 8.97017360e-01 1.82332605e-01 -1.79739103e-01
2.61487126e-01 -6.08889619e-03 3.92437130e-01 6.09074235e-01
-3.66596878e-01 2.65272141e-01 2.79844046e-01 -3.58011156e-01
-1.02605474e+00 -1.59592664e+00 4.42662179e-01 1.27813721e+00
2.57919431e-01 6.09725937e-02 -4.25092787e-01 -1.22460687e+00
2.68484622e-01 4.93483484e-01 -9.89859462e-01 -5.92466474e-01
-8.44986260e-01 5.06110668e-01 3.31834137e-01 6.40160620e-01
3.24976504e-01 -1.48061860e+00 -1.12269735e+00 -8.09297264e-02
7.23771527e-02 -9.47021544e-01 -7.07813859e-01 3.25771868e-01
-7.91910946e-01 -1.27235675e+00 -8.23100507e-01 -9.86203551e-01
6.76298678e-01 1.01426639e-01 1.12982810e+00 -2.64881939e-01
-2.56794721e-01 1.30061483e+00 -2.28254840e-01 -4.18216474e-02
-4.17040646e-01 -2.79418021e-01 5.60340643e-01 -3.08097929e-01
3.03717200e-02 -8.00194442e-01 -6.28718019e-01 1.67237848e-01
-8.20468187e-01 -2.01445282e-01 4.22996372e-01 1.35819805e+00
3.88582677e-01 -4.44783002e-01 4.46719229e-01 -5.58981776e-01
6.06884778e-01 -4.90531921e-01 -7.44197845e-01 1.43396989e-01
-3.38795304e-01 4.04119164e-01 7.13064790e-01 -9.12151754e-01
-9.95458961e-01 3.38441730e-01 4.45800900e-01 -1.02850032e+00
-1.33371025e-01 4.35550921e-02 2.83660591e-01 -1.95727214e-01
7.87076235e-01 2.81912178e-01 4.76257622e-01 -1.07364401e-01
6.26926303e-01 6.62731845e-03 5.99066556e-01 -8.01631749e-01
1.11403525e+00 5.65687060e-01 4.32109758e-02 -3.55838448e-01
-7.08957195e-01 -1.88606799e-01 -6.28168762e-01 -3.12630564e-01
1.03864658e+00 -1.09407187e+00 -8.77720714e-01 1.32001266e-01
-6.68543041e-01 -8.65646064e-01 -1.01914954e+00 7.64505923e-01
-1.45320702e+00 1.67541847e-01 -4.65478778e-01 -4.69812870e-01
6.78333640e-02 -1.14819932e+00 8.76062393e-01 2.04723567e-01
-8.46946388e-02 -1.10868764e+00 5.31324625e-01 1.48670152e-01
2.37476647e-01 4.59761500e-01 3.99418235e-01 -4.62033927e-01
-6.75337911e-01 1.16031915e-01 3.41259629e-01 6.63913429e-01
1.36365965e-01 -6.17197037e-01 -7.76545107e-01 -9.10257638e-01
1.01197064e-01 -1.10881662e+00 7.25978792e-01 1.41183093e-01
9.09469843e-01 -2.45565563e-01 -1.90699436e-02 5.72124898e-01
1.33906126e+00 1.30993053e-01 3.69289219e-01 2.37041384e-01
6.18521392e-01 6.09944820e-01 7.71186054e-01 2.98947334e-01
1.13072775e-01 5.53862751e-01 7.38451123e-01 3.06511015e-01
-6.30220491e-03 -6.23924136e-01 8.28043818e-01 5.90648115e-01
-4.02111650e-01 1.81933507e-01 -4.41233456e-01 5.05066037e-01
-1.87179554e+00 -1.34608483e+00 9.40097332e-01 2.15726185e+00
8.62550318e-01 -3.50946188e-02 1.89025104e-01 -3.89184445e-01
3.39055419e-01 6.79820850e-02 -1.00462246e+00 -2.54148841e-01
3.21015388e-01 4.71114367e-01 4.92997587e-01 5.53673029e-01
-1.16784370e+00 1.00590479e+00 5.75779057e+00 4.15449440e-01
-1.03119159e+00 2.55353540e-01 -1.10116534e-01 -4.82987016e-01
-5.04916832e-02 -2.53962070e-01 -1.65669844e-01 2.93866515e-01
5.86637020e-01 -3.11025053e-01 9.94892359e-01 9.05680418e-01
-4.91248444e-02 1.98993266e-01 -1.38503647e+00 7.50080228e-01
9.19924006e-02 -1.08001840e+00 -2.67068774e-01 -1.13947671e-02
9.81377184e-01 1.19172625e-01 5.14095783e-01 7.03252971e-01
8.76319408e-01 -1.06166840e+00 7.37542272e-01 6.05428576e-01
1.08090007e+00 -5.50395429e-01 2.14590654e-01 2.25678131e-01
-8.97830009e-01 -4.75978434e-01 -1.14180945e-01 -2.27810472e-01
-3.34478356e-02 -5.18134058e-01 -5.41229248e-01 4.16046202e-01
6.69250727e-01 1.15580297e+00 -8.92455205e-02 4.48372483e-01
-5.12843966e-01 1.61647528e-01 9.57953557e-03 8.37636590e-02
4.03603435e-01 -2.13106453e-01 7.24747777e-01 7.03048170e-01
1.76644757e-01 -1.11890443e-01 3.53616148e-01 7.69527555e-01
-5.11488738e-03 -4.28850323e-01 -1.23974574e+00 -1.09523632e-01
1.34810060e-01 9.68126595e-01 -1.46738783e-01 -6.38732612e-02
-4.15648699e-01 1.32735169e+00 7.74965286e-01 6.77960932e-01
-8.69650483e-01 -1.57003626e-01 1.11276960e+00 -2.05106556e-01
4.17302251e-01 -4.00302112e-01 4.06023115e-01 -1.44548929e+00
-1.47053495e-01 -9.92366672e-01 2.81174272e-01 -7.16739476e-01
-1.31816280e+00 5.01899421e-01 1.44699246e-01 -1.64187098e+00
-5.78387082e-01 -7.64486492e-01 -3.89041334e-01 6.50923610e-01
-1.40362990e+00 -1.09916568e+00 -3.59577954e-01 1.02101433e+00
6.31453335e-01 -6.63577318e-01 9.21859682e-01 -1.93794906e-01
-5.23406900e-02 4.20555919e-01 2.31955886e-01 -3.90247256e-02
8.89468729e-01 -1.62964034e+00 -2.04834738e-03 4.69823152e-01
1.84795767e-01 3.68770868e-01 7.99716890e-01 -4.35280859e-01
-1.30609882e+00 -8.42554271e-01 -4.77036051e-02 -7.35596299e-01
9.09741938e-01 -4.36476469e-01 -4.74096179e-01 1.17270482e+00
4.38387811e-01 1.67102173e-01 3.63497794e-01 -3.41218174e-01
-4.79267746e-01 8.52957275e-03 -1.27997339e+00 7.20846415e-01
1.22933638e+00 -7.41686285e-01 -9.44121420e-01 2.48888791e-01
9.44091022e-01 -4.00416791e-01 -8.71450901e-01 1.37010887e-01
7.05462635e-01 -7.58815169e-01 1.22628653e+00 -1.11500072e+00
6.06380403e-01 -3.71611528e-02 -1.69725157e-02 -1.85688269e+00
-3.15434307e-01 -7.93862224e-01 -3.76944840e-01 2.94287980e-01
7.79535174e-02 -6.09977126e-01 6.92846775e-01 1.89543784e-01
-6.71408325e-02 -5.91630638e-01 -9.67028797e-01 -1.20783937e+00
3.44403177e-01 2.61837035e-01 1.07301123e-01 9.58699644e-01
2.93956578e-01 1.52701065e-01 -7.56537318e-01 8.65029618e-02
7.23660350e-01 5.99102080e-02 7.50430226e-01 -9.76450920e-01
-6.79052472e-01 -9.39441007e-03 -5.15120566e-01 -1.08385456e+00
6.74026072e-01 -8.49983037e-01 8.31669718e-02 -1.05397356e+00
5.18889688e-02 -1.90529868e-01 -4.98671204e-01 4.02977318e-01
6.93203434e-02 -4.10954133e-02 4.51390475e-01 8.67186859e-02
-7.40188658e-01 9.55704570e-01 1.89690518e+00 -1.13274775e-01
-4.00085077e-02 -1.26237899e-01 -1.86517790e-01 6.11218154e-01
1.00580597e+00 -6.01010025e-01 -8.92796993e-01 -5.31637132e-01
-4.75886352e-02 1.19590648e-01 6.39732897e-01 -1.12260783e+00
-5.54142818e-02 -3.84917706e-01 1.97734848e-01 2.86452174e-01
8.10913205e-01 -1.23439646e+00 -1.68628976e-01 8.21414769e-01
-4.78809297e-01 1.54918775e-01 1.90004215e-01 1.05412400e+00
-1.27544673e-02 -1.87174141e-01 7.77380347e-01 -4.09535885e-01
-8.79249036e-01 2.58662105e-01 -2.61680067e-01 4.67125893e-01
1.57461214e+00 -5.28690889e-02 -1.50043502e-01 -5.52434802e-01
-1.18609679e+00 4.01164830e-01 6.09768212e-01 6.23192728e-01
7.38424361e-01 -1.40624332e+00 -4.83946174e-01 4.18035716e-01
9.98256728e-02 -3.86876017e-01 1.55800521e-01 4.43373561e-01
-4.85628575e-01 6.41044229e-02 -1.01398444e+00 -2.70679861e-01
-1.03789771e+00 8.99032772e-01 4.86511528e-01 -4.13339049e-01
-6.56062424e-01 7.93998420e-01 4.58020329e-01 -6.59576476e-01
2.93362826e-01 -1.57406434e-01 -8.84101093e-02 -4.48692739e-01
1.76628351e-01 3.50908250e-01 -6.45843387e-01 -3.40994954e-01
-5.31858690e-02 3.47908348e-01 9.77494642e-02 -4.22793150e-01
1.28315723e+00 1.36279641e-02 4.08725977e-01 4.92077678e-01
1.10229146e+00 -3.09911251e-01 -2.25031447e+00 -1.96891010e-01
-3.21530253e-01 -5.96752107e-01 -2.94365466e-01 -8.20952117e-01
-9.43021178e-01 7.34756410e-01 7.17631161e-01 -2.42165159e-02
9.65878129e-01 3.34723890e-02 4.50473458e-01 5.90277493e-01
7.54252553e-01 -1.19058144e+00 1.04915440e+00 6.06696546e-01
1.19788170e+00 -1.30529690e+00 -3.80292684e-01 2.13261753e-01
-1.08502066e+00 8.46871674e-01 9.30438340e-01 -8.18069160e-01
5.95204473e-01 2.54029393e-01 -1.19735114e-02 -1.60893381e-01
-7.46369660e-01 -3.49209547e-01 1.25143617e-01 9.74256635e-01
-1.33990765e-01 -1.46058397e-02 1.22012995e-01 2.09182411e-01
1.83629751e-01 4.35215607e-02 4.57858086e-01 1.23910177e+00
-2.49426693e-01 -9.82526124e-01 1.20911829e-01 2.11688295e-01
-1.70404360e-01 2.26810068e-01 -8.79054666e-02 8.37933660e-01
-1.29520865e-02 5.28079808e-01 -1.34576587e-02 -5.28152525e-01
3.85269135e-01 4.82090078e-02 1.19198799e+00 -9.44360077e-01
-5.28133035e-01 -3.72093767e-01 -2.14415044e-01 -9.65152681e-01
-4.85666215e-01 -6.95936441e-01 -1.08748233e+00 -9.44689382e-03
2.12813571e-01 7.01202452e-02 1.77861363e-01 6.34546876e-01
3.15111816e-01 6.11983955e-01 7.00906754e-01 -1.24642944e+00
-1.16823399e+00 -7.92776883e-01 -6.88840926e-01 8.33819926e-01
6.67010486e-01 -1.22945607e+00 -3.78377616e-01 2.58475929e-01]
|
[4.481469631195068, 0.8401564359664917]
|
e49feab6-5447-4b91-a772-cfaa1440de1d
|
rethinking-optical-flow-from-geometric
|
2303.08384
| null |
https://arxiv.org/abs/2303.08384v1
|
https://arxiv.org/pdf/2303.08384v1.pdf
|
Rethinking Optical Flow from Geometric Matching Consistent Perspective
|
Optical flow estimation is a challenging problem remaining unsolved. Recent deep learning based optical flow models have achieved considerable success. However, these models often train networks from the scratch on standard optical flow data, which restricts their ability to robustly and geometrically match image features. In this paper, we propose a rethinking to previous optical flow estimation. We particularly leverage Geometric Image Matching (GIM) as a pre-training task for the optical flow estimation (MatchFlow) with better feature representations, as GIM shares some common challenges as optical flow estimation, and with massive labeled real-world data. Thus, matching static scenes helps to learn more fundamental feature correlations of objects and scenes with consistent displacements. Specifically, the proposed MatchFlow model employs a QuadTree attention-based network pre-trained on MegaDepth to extract coarse features for further flow regression. Extensive experiments show that our model has great cross-dataset generalization. Our method achieves 11.5% and 10.1% error reduction from GMA on Sintel clean pass and KITTI test set. At the time of anonymous submission, our MatchFlow(G) enjoys state-of-the-art performance on Sintel clean and final pass compared to published approaches with comparable computation and memory footprint. Codes and models will be released in https://github.com/DQiaole/MatchFlow.
|
['Yanwei Fu', 'Chenjie Cao', 'Qiaole Dong']
|
2023-03-15
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Dong_Rethinking_Optical_Flow_From_Geometric_Matching_Consistent_Perspective_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Dong_Rethinking_Optical_Flow_From_Geometric_Matching_Consistent_Perspective_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['geometric-matching']
|
['computer-vision']
|
[-2.63700604e-01 -5.24162292e-01 -4.07311231e-01 -1.87807411e-01
-4.28911299e-01 -4.14662153e-01 3.84656876e-01 -4.33044881e-01
-3.02575082e-01 7.32187331e-01 3.13364923e-01 -1.14259496e-01
-3.34817581e-02 -7.76446164e-01 -6.47937655e-01 -4.10266370e-01
-2.10878342e-01 1.52956441e-01 1.98972404e-01 -7.04297200e-02
5.06959558e-01 4.34480131e-01 -1.25749207e+00 4.84416783e-02
1.03356516e+00 9.73275185e-01 2.15050623e-01 9.77567613e-01
-3.74603085e-02 1.30725920e+00 -2.74086505e-01 -4.27478224e-01
6.40045822e-01 -4.36369926e-01 -1.20304930e+00 -6.51446283e-02
1.43286276e+00 -9.08277035e-01 -9.96214271e-01 8.55026066e-01
3.80246818e-01 4.46919352e-01 4.18950558e-01 -1.48027861e+00
-7.62913525e-01 1.21921249e-01 -5.56632578e-01 6.59642339e-01
2.91956127e-01 5.99194229e-01 1.07837868e+00 -9.15842235e-01
7.22705364e-01 1.27989364e+00 4.84135807e-01 5.55062056e-01
-9.74109769e-01 -7.61491716e-01 7.80929849e-02 5.89932561e-01
-1.13866103e+00 -5.69017112e-01 5.34417093e-01 -6.05695665e-01
8.81177425e-01 -8.26366842e-02 7.70455837e-01 8.26872587e-01
2.15052009e-01 8.77898514e-01 7.01091588e-01 -1.93554186e-03
-1.21417351e-01 -3.02246690e-01 -4.81755361e-02 1.06928480e+00
1.97440997e-01 3.65509272e-01 -6.96383476e-01 3.64392430e-01
1.06921601e+00 7.57447816e-03 -6.88211262e-01 -3.02906364e-01
-1.13518798e+00 7.60416329e-01 1.02221262e+00 -4.36452441e-02
-8.73459280e-02 5.27073979e-01 2.57689893e-01 1.32351950e-01
3.17289114e-01 3.47214639e-01 -2.39709675e-01 -3.27438831e-01
-9.50441658e-01 3.69605124e-01 5.89177847e-01 9.68780398e-01
1.13669419e+00 1.81183517e-01 -1.20572336e-01 4.76851195e-01
2.54614532e-01 6.17609680e-01 4.45622891e-01 -1.33628404e+00
6.95535362e-01 3.60805809e-01 6.49689287e-02 -1.37183189e+00
-2.98685670e-01 -4.53730285e-01 -9.55741584e-01 2.11426586e-01
7.56720066e-01 -5.21592982e-02 -7.52733707e-01 1.54780805e+00
3.49836707e-01 8.85245204e-01 -6.12810031e-02 1.31221902e+00
8.08163345e-01 7.36365795e-01 -1.96395606e-01 1.65308595e-01
8.44981551e-01 -1.59561443e+00 -4.72176135e-01 -3.54700148e-01
7.69656837e-01 -9.69858468e-01 8.68862331e-01 1.89106554e-01
-1.21823657e+00 -7.94891477e-01 -8.87275219e-01 -5.47820568e-01
-3.93752195e-03 -9.52507481e-02 9.54762399e-01 3.97429526e-01
-9.47776616e-01 8.57129216e-01 -9.26623166e-01 -2.28248522e-01
8.60973895e-01 2.06397682e-01 -4.23708081e-01 -4.12849337e-01
-9.27145600e-01 5.99354267e-01 6.43240809e-02 3.09375435e-01
-9.06253874e-01 -1.18985415e+00 -9.87676859e-01 -1.47739165e-02
2.67434806e-01 -1.10117292e+00 1.05245268e+00 -6.18359804e-01
-1.55382836e+00 6.34721637e-01 -3.27487946e-01 -4.67702925e-01
7.43558943e-01 -6.39254928e-01 -2.66394075e-02 4.13788766e-01
2.52850085e-01 9.37236786e-01 8.22475553e-01 -7.60139287e-01
-8.18166792e-01 -5.91569394e-02 1.92467093e-01 -1.32933892e-02
-8.23785514e-02 -3.75831068e-01 -4.71613020e-01 -5.67480981e-01
-5.58246709e-02 -8.94860029e-01 -1.63583249e-01 4.85701859e-01
-2.38330856e-01 4.73773815e-02 7.46315777e-01 -4.24518943e-01
1.12141335e+00 -1.80748558e+00 8.59432679e-04 -2.55583942e-01
4.75060046e-01 5.96299946e-01 -3.90339792e-01 1.43058509e-01
1.89800665e-01 -7.80872032e-02 -2.25147679e-01 -5.31923532e-01
-3.59726310e-01 1.43966541e-01 -3.81990045e-01 6.90497637e-01
3.48610789e-01 1.09925747e+00 -1.20450056e+00 -5.78026235e-01
6.60753906e-01 4.66591179e-01 -1.07605052e+00 4.18931663e-01
1.00592189e-01 7.48932302e-01 -2.89688408e-01 4.95036751e-01
9.85931814e-01 -4.13472533e-01 -3.50210488e-01 -4.33167547e-01
-1.29169539e-01 3.83399904e-01 -1.13813055e+00 2.10989881e+00
-6.38320386e-01 1.01746285e+00 -2.45000571e-01 -8.89294207e-01
7.73269475e-01 -1.44543752e-01 7.19532669e-01 -7.48488069e-01
1.41899958e-01 1.58474699e-01 1.09910652e-01 -5.63222706e-01
5.81798673e-01 2.88610846e-01 4.58636731e-01 2.67183334e-01
3.63960028e-01 -1.56193554e-01 4.45864916e-01 2.70163536e-01
9.50643480e-01 3.95707458e-01 8.41203257e-02 -2.13176221e-01
7.62305498e-01 -8.92245322e-02 8.06606889e-01 7.28569806e-01
-6.98660195e-01 8.74598980e-01 1.77670717e-01 -8.31276000e-01
-8.16203713e-01 -1.09626627e+00 -1.32376015e-01 4.51643646e-01
5.62453866e-01 -4.44563597e-01 -4.48755741e-01 -6.75396502e-01
7.93849677e-02 1.05131539e-02 -5.85368335e-01 1.75517946e-02
-9.96748269e-01 -5.16206145e-01 2.44038224e-01 4.46607947e-01
1.00265574e+00 -9.68034089e-01 -5.73124349e-01 2.34966949e-01
-3.29320043e-01 -1.49720323e+00 -8.47499669e-01 -5.65419018e-01
-8.62359107e-01 -1.45195925e+00 -5.93889713e-01 -7.01053441e-01
4.82156038e-01 6.33266389e-01 1.24151742e+00 3.81003588e-01
-5.07501066e-01 1.41577214e-01 -2.30285868e-01 1.51701912e-01
2.95561161e-02 2.69367307e-01 -2.62825757e-01 1.94117412e-01
3.59998904e-02 -6.21097267e-01 -1.21660364e+00 3.55412960e-01
-6.45294189e-01 4.42091413e-02 3.19967389e-01 8.86444211e-01
2.28809670e-01 -5.80781460e-01 1.21034369e-01 -6.22360349e-01
6.51640147e-02 -4.08511430e-01 -7.11796820e-01 -7.52003789e-02
-4.95047271e-01 4.32480425e-02 7.51871049e-01 -7.23231956e-02
-9.80078757e-01 -8.93032625e-02 7.56240310e-03 -8.04300785e-01
3.95557210e-02 3.17304358e-02 1.57729924e-01 -4.47093040e-01
5.08724809e-01 1.11279283e-02 -1.27246873e-02 -1.59059882e-01
3.95614833e-01 2.71611631e-01 7.69066751e-01 -4.13404882e-01
9.50416446e-01 8.29295456e-01 3.20252389e-01 -6.57839000e-01
-1.05519390e+00 -5.46827495e-01 -7.77363718e-01 -2.49478340e-01
7.63149083e-01 -8.92441511e-01 -1.06601441e+00 7.12229788e-01
-1.16101205e+00 -5.82052588e-01 -5.40122353e-02 8.43218744e-01
-6.13857985e-01 4.45088148e-01 -6.95606887e-01 -3.26307595e-01
-4.18137491e-01 -1.15728831e+00 8.80571306e-01 5.58277786e-01
3.64499316e-02 -1.27911413e+00 2.79390633e-01 6.22926116e-01
4.98665661e-01 2.10283235e-01 1.17596902e-01 1.78132296e-01
-1.26586866e+00 2.09847867e-01 -6.33049130e-01 3.24265033e-01
3.15242141e-01 2.30152711e-01 -9.85255718e-01 -4.37981933e-01
-3.35403949e-01 -3.78353268e-01 1.21931934e+00 6.17794216e-01
1.11674082e+00 -4.41057011e-02 -1.74929261e-01 1.39405775e+00
1.53133345e+00 -5.52601032e-02 7.34058917e-01 1.82026535e-01
1.12865746e+00 3.67537707e-01 5.80497682e-01 4.77465361e-01
5.36863923e-01 5.61359048e-01 5.79982758e-01 -8.74280930e-02
-6.76366508e-01 -3.38207394e-01 2.04730198e-01 7.30991244e-01
-1.31276801e-01 -2.91662157e-01 -7.16820478e-01 5.91703534e-01
-1.91923928e+00 -1.11779809e+00 -2.19223469e-01 1.95775843e+00
4.33980882e-01 2.15951893e-02 -7.60760009e-02 -1.56977788e-01
4.13460970e-01 4.70633984e-01 -4.75458145e-01 -1.04498230e-01
-2.84411907e-02 3.92708570e-01 5.31618416e-01 1.01059771e+00
-1.25163054e+00 1.28015935e+00 5.11212540e+00 4.67770338e-01
-1.31212604e+00 -1.69880409e-02 6.61528885e-01 -2.57882059e-01
7.31039494e-02 1.74302876e-01 -8.63236189e-01 5.62180281e-01
4.65139329e-01 -1.27677262e-01 4.84573305e-01 4.78393674e-01
4.41504240e-01 -2.31375933e-01 -1.02829623e+00 1.31312180e+00
2.81912237e-02 -1.83428311e+00 6.62719607e-02 -3.50588225e-02
1.01397526e+00 2.83582598e-01 6.17146306e-03 1.07986175e-01
2.62559026e-01 -1.00139916e+00 4.42686319e-01 6.00929558e-01
5.69668889e-01 -5.42966962e-01 6.85627639e-01 -1.07770063e-01
-1.38558578e+00 -2.81088464e-02 -4.40559864e-01 -4.13412660e-01
4.65574890e-01 5.04689693e-01 -4.25549567e-01 6.97007358e-01
7.66663134e-01 1.51270127e+00 -5.46029866e-01 1.41619039e+00
-1.44454107e-01 4.72451985e-01 -1.32003531e-01 2.71544099e-01
4.47150826e-01 -3.47902924e-01 3.93229842e-01 1.06245160e+00
2.12216631e-01 -1.19477175e-01 3.63839984e-01 9.67340469e-01
-2.47533008e-01 8.80952850e-02 -4.51485962e-01 3.64000320e-01
2.87218690e-01 1.20767331e+00 -3.83788407e-01 -3.46534550e-01
-6.00105345e-01 9.31265593e-01 6.06330872e-01 3.49487126e-01
-8.06879878e-01 -3.15761209e-01 1.26491547e+00 2.94667836e-02
1.28684670e-01 -3.74282688e-01 -7.37081170e-02 -1.56685126e+00
7.03585073e-02 -3.15278590e-01 2.86187798e-01 -6.69661999e-01
-1.24090707e+00 5.44139862e-01 -2.53757805e-01 -1.49083877e+00
-2.39135280e-01 -6.27192080e-01 -7.99347579e-01 8.01434577e-01
-1.99021053e+00 -9.76797581e-01 -8.22695434e-01 6.99213982e-01
4.79847610e-01 -4.14884475e-04 3.12376559e-01 5.66908360e-01
-7.43392527e-01 5.26257515e-01 -2.16488287e-01 5.84946215e-01
9.13509905e-01 -1.11602402e+00 6.25465035e-01 1.18677604e+00
2.52186656e-01 3.29231739e-01 2.33122125e-01 -3.82867694e-01
-1.31656075e+00 -1.27797723e+00 7.60177135e-01 -4.44099039e-01
7.29679465e-01 -4.41114977e-02 -9.04862702e-01 6.41428530e-01
2.86915541e-01 8.32984030e-01 1.99510589e-01 -3.37267756e-01
-2.88261831e-01 -3.27473998e-01 -8.25991869e-01 5.38607299e-01
1.48517048e+00 -4.06888545e-01 -9.61580276e-02 2.77565509e-01
5.19993722e-01 -6.43054605e-01 -7.34570563e-01 2.99575567e-01
4.87928808e-01 -1.33877301e+00 1.02500296e+00 -5.63143432e-01
7.62476623e-01 -4.28275913e-01 1.29161358e-01 -1.19704890e+00
-3.06153715e-01 -1.07189727e+00 -2.83308834e-01 8.79909694e-01
-4.51283455e-02 -6.65625095e-01 1.04571009e+00 3.14088404e-01
-2.23043472e-01 -6.64032936e-01 -7.05726385e-01 -7.23341882e-01
2.04013065e-01 -5.01002789e-01 2.97324508e-01 9.78832603e-01
-3.33627373e-01 2.91542679e-01 -5.90729356e-01 1.12390779e-01
7.23536611e-01 3.70730072e-01 1.25078344e+00 -9.32355881e-01
-1.82252079e-01 -5.81682861e-01 -7.29799926e-01 -1.60520744e+00
3.57208908e-01 -1.08023000e+00 -2.27605224e-01 -1.42860401e+00
-7.66976774e-02 -4.83037412e-01 7.45097920e-02 2.39925891e-01
-3.78432810e-01 4.85538810e-01 4.87886667e-01 2.63213009e-01
-4.54665303e-01 7.10543573e-01 1.89064658e+00 -2.63724387e-01
-2.39237711e-01 -1.19357303e-01 -2.37854719e-01 3.94307554e-01
8.17634881e-01 -2.82984436e-01 -5.29753327e-01 -7.80271351e-01
-1.68048099e-01 8.13670382e-02 6.99988246e-01 -1.28854537e+00
2.57719308e-01 -2.63324648e-01 2.40966797e-01 -3.71205479e-01
1.37914270e-01 -4.12942827e-01 -2.95884043e-01 6.25461638e-01
-1.78977594e-01 9.70397368e-02 2.00885504e-01 2.95989215e-01
-4.08442914e-01 -7.93112814e-02 8.90395522e-01 -5.91911077e-02
-9.53913629e-01 9.97084439e-01 8.07615146e-02 5.74248850e-01
6.81507885e-01 -2.04780608e-01 -5.74301243e-01 -3.73467833e-01
-4.18853730e-01 3.80390376e-01 2.26415649e-01 5.80110610e-01
6.91754520e-01 -1.25108814e+00 -8.35057259e-01 3.80050182e-01
-4.57396917e-02 2.52845943e-01 4.31852430e-01 8.64212513e-01
-1.05173016e+00 4.17736501e-01 -4.84334797e-01 -8.23039472e-01
-8.20283353e-01 4.08892125e-01 5.78589141e-01 -1.18308261e-01
-7.16017902e-01 9.22829449e-01 3.23607713e-01 -3.11032962e-02
4.25827205e-02 -5.38320959e-01 1.16812445e-01 -2.21133322e-01
6.44405901e-01 5.70232928e-01 -6.34821132e-02 -6.48086190e-01
-3.15123230e-01 9.50107634e-01 -7.87849426e-02 2.48768941e-01
1.06837320e+00 -2.51084298e-01 6.62632510e-02 -1.33440718e-01
1.67348158e+00 -1.36259004e-01 -1.84314501e+00 -3.37166190e-01
-4.38058078e-01 -1.17077923e+00 1.65489003e-01 -2.39723265e-01
-1.76416409e+00 1.22432816e+00 5.40758193e-01 -3.68373960e-01
1.00985575e+00 -3.02287072e-01 1.10081339e+00 2.25930363e-01
1.51437536e-01 -6.25554025e-01 3.11471879e-01 6.02925956e-01
6.93593919e-01 -1.57959557e+00 -3.78465243e-02 -5.41151047e-01
-2.00871378e-01 1.32520473e+00 1.00833035e+00 -5.38346767e-01
7.57851779e-01 5.37781119e-02 1.74812645e-01 9.81280059e-02
-7.14870751e-01 -3.62361431e-01 3.23360354e-01 5.36308229e-01
3.09783757e-01 -2.15090826e-01 1.19084962e-01 -2.96580732e-01
-3.49185109e-01 2.22792104e-01 6.37986183e-01 7.26370335e-01
-1.29095048e-01 -9.13390160e-01 -5.72049096e-02 1.28966987e-01
-1.82845652e-01 -1.61091387e-01 1.58723235e-01 8.97685885e-01
2.98080649e-02 8.51221919e-01 4.01149511e-01 -2.45384306e-01
2.21085563e-01 -5.76034188e-01 6.86351955e-01 -2.02553958e-01
-3.72002184e-01 -3.16117585e-01 -2.94748515e-01 -1.09013224e+00
-6.09432638e-01 -4.95949745e-01 -1.22830212e+00 -7.22488701e-01
7.46663883e-02 -2.13018060e-01 1.27985269e-01 8.97141993e-01
4.47214186e-01 3.38529170e-01 8.44348848e-01 -1.00213134e+00
-1.17728926e-01 -7.52408803e-01 -1.04895800e-01 6.66033447e-01
7.71923840e-01 -7.54659235e-01 -3.85490268e-01 2.10722283e-01]
|
[8.76429557800293, -1.8457454442977905]
|
c70a1e6c-8d26-4ccf-b7fc-98ae11cb3d98
|
lightweight-estimation-of-hand-mesh-and
|
2303.14838
| null |
https://arxiv.org/abs/2303.14838v1
|
https://arxiv.org/pdf/2303.14838v1.pdf
|
Lightweight Estimation of Hand Mesh and Biomechanically Feasible Kinematic Parameters
|
3D hand pose estimation is a long-standing challenge in both robotics and computer vision communities due to its implicit depth ambiguity and often strong self-occlusion. Recently, in addition to the hand skeleton, jointly estimating hand pose and shape has gained more attraction. State-of-the-art methods adopt a model-free approach, estimating the vertices of the hand mesh directly and providing superior accuracy compared to traditional model-based methods directly regressing the parameters of the parametric hand mesh. However, with the large number of mesh vertices to estimate, these methods are often slow in inference. We propose an efficient variation of the previously proposed image-to-lixel approach to efficiently estimate hand meshes from the images. Leveraging recent developments in efficient neural architectures, we significantly reduce the computation complexity without sacrificing the estimation accuracy. Furthermore, we introduce an inverted kinematic(IK) network to translate the estimated hand mesh to a biomechanically feasible set of joint rotation parameters, which is necessary for applications that leverage pose estimation for controlling robotic hands. Finally, an optional post-processing module is proposed to refine the rotation and shape parameters to compensate for the error introduced by the IK net. Our Lite I2L Mesh Net achieves state-of-the-art joint and mesh estimation accuracy with less than $13\%$ of the total computational complexity of the original I2L hand mesh estimator. Adding the IK net and post-optimization modules can improve the accuracy slightly at a small computation cost, but more importantly, provide the kinematic parameters required for robotic applications.
|
['Yao Wang', 'Zhipeng Fan']
|
2023-03-26
| null | null | null | null |
['3d-hand-pose-estimation', '3d-hand-pose-estimation']
|
['computer-vision', 'graphs']
|
[-2.72153243e-02 -2.73940898e-03 -3.42793763e-01 3.78743708e-02
-6.46086216e-01 -4.55991894e-01 2.48796046e-02 -2.42939353e-01
-5.40492952e-01 5.70805371e-01 -1.42786264e-01 2.52933707e-02
-5.29233851e-02 -4.32185948e-01 -9.50153768e-01 -3.49118531e-01
2.60824561e-01 9.27122831e-01 1.71994939e-01 -4.01613154e-02
2.45240435e-01 7.72521198e-01 -1.47109449e+00 -3.23350310e-01
6.07399523e-01 1.04496658e+00 3.50625575e-01 5.63909590e-01
1.75694644e-01 2.42538452e-01 -1.03448190e-01 -2.18691885e-01
4.04799074e-01 4.46197018e-02 -7.08727241e-01 1.11809522e-02
6.00187480e-01 -7.45637059e-01 -4.47904259e-01 7.60268211e-01
8.79323184e-01 1.02016769e-01 7.52830625e-01 -9.88037884e-01
-5.79395965e-02 2.98126400e-01 -9.04751241e-01 -4.77591366e-01
3.63030881e-01 2.25500330e-01 8.14536870e-01 -1.02735996e+00
1.09847534e+00 1.15738845e+00 1.03666258e+00 4.85608608e-01
-1.36390603e+00 -6.65809095e-01 2.22536564e-01 4.41397578e-02
-1.63145673e+00 -3.25422257e-01 1.02867639e+00 -5.53848803e-01
9.45095479e-01 -6.44554868e-02 1.01254404e+00 9.03791606e-01
1.48819476e-01 7.48515606e-01 6.24694824e-01 -4.04015690e-01
-5.77213168e-02 -4.30158913e-01 -2.84794092e-01 1.07632804e+00
2.15371519e-01 -2.04884171e-01 -3.50968271e-01 -4.43544611e-02
1.57441902e+00 -9.85996500e-02 -9.17912498e-02 -8.03594828e-01
-1.26420486e+00 4.23216105e-01 6.18304074e-01 -1.54905915e-01
-4.93721217e-01 7.25191832e-01 2.26156756e-01 -2.36976549e-01
2.61360228e-01 4.41736102e-01 -6.52248502e-01 -2.64177710e-01
-9.13269043e-01 7.50312686e-01 8.55291903e-01 1.00460923e+00
5.27628958e-01 -5.70551977e-02 1.95774257e-01 7.23013759e-01
5.25798678e-01 5.31288683e-01 -2.01234952e-01 -1.37705398e+00
6.40677929e-01 6.74229801e-01 2.59747386e-01 -1.03444934e+00
-7.02590406e-01 -3.78909379e-01 -7.03769267e-01 4.02728915e-01
7.87473261e-01 -1.30403325e-01 -1.01944089e+00 1.66176462e+00
6.42375052e-01 -2.93951243e-01 -7.56785393e-01 1.07862401e+00
2.42375746e-01 -5.56588508e-02 -5.81485629e-02 1.13697872e-01
1.32834208e+00 -7.56661296e-01 -3.37374508e-01 -3.03535193e-01
1.06173359e-01 -7.87206948e-01 1.05422783e+00 4.55113798e-01
-1.22815633e+00 -2.85239518e-01 -8.83108199e-01 -4.95662808e-01
6.92160055e-02 4.74064678e-01 6.67243183e-01 2.47691929e-01
-7.30365634e-01 7.47157395e-01 -1.11624622e+00 -1.63305938e-01
4.54379737e-01 8.48023415e-01 -3.41832012e-01 1.68493852e-01
-6.36046112e-01 1.18820524e+00 1.13303721e-01 4.61137474e-01
-5.19002616e-01 -8.32892954e-01 -7.52113044e-01 -2.12134942e-01
6.41391039e-01 -1.03360987e+00 1.29928112e+00 -4.54961270e-01
-1.87031770e+00 6.47505105e-01 -2.18510687e-01 1.11879356e-01
1.13003111e+00 -4.12565082e-01 5.64545870e-01 2.37887353e-01
-1.35804370e-01 8.83543432e-01 1.01486897e+00 -1.14055598e+00
-1.33553460e-01 -5.91697812e-01 1.53751187e-02 2.49923930e-01
-1.19006902e-01 -4.69466001e-01 -8.39120090e-01 -7.82105684e-01
4.45093334e-01 -1.27365887e+00 -2.56201774e-01 8.52898240e-01
-2.59861261e-01 -2.54039675e-01 6.21873379e-01 -9.16872919e-01
8.43267560e-01 -1.68249142e+00 6.91336632e-01 3.98358226e-01
2.65488535e-01 1.46200974e-02 -5.73452823e-02 6.93335831e-02
4.16919440e-01 -2.47743428e-01 -2.33035535e-01 -3.34576696e-01
-9.61049572e-02 1.21783391e-01 1.02057651e-01 6.24085128e-01
1.75719112e-01 1.00117064e+00 -7.18158960e-01 -6.13829792e-01
2.82678902e-01 7.96720684e-01 -9.18809175e-01 3.47068273e-02
-5.81019223e-01 6.52289271e-01 -4.45506275e-01 7.00377941e-01
3.94031495e-01 -1.83100834e-01 4.32072937e-01 -6.73841357e-01
-5.72574288e-02 1.57286152e-02 -1.48598373e+00 2.19999361e+00
-6.10340953e-01 1.76154137e-01 6.36349499e-01 -7.02356994e-01
4.91440147e-01 3.90636712e-01 6.23682380e-01 1.60014946e-02
6.24304295e-01 4.94646847e-01 -9.62703228e-02 -4.23528224e-01
2.41995960e-01 -3.83076631e-02 1.88631952e-01 4.24589843e-01
-5.92172444e-02 -6.14159167e-01 -1.33975819e-01 -4.10967857e-01
6.36789322e-01 9.67125356e-01 1.25149563e-01 -2.12208435e-01
2.12229326e-01 4.63200733e-02 3.47355306e-01 2.06050158e-01
1.42551392e-01 6.24718606e-01 3.01611036e-01 -4.19884473e-01
-1.29451644e+00 -9.29332018e-01 -8.04123282e-02 8.19390655e-01
5.63260540e-03 -9.64979380e-02 -9.66401875e-01 -2.25667208e-01
5.44127524e-01 -1.24392072e-02 -4.12323475e-01 2.91010618e-01
-1.03260505e+00 -7.23693892e-02 5.16643643e-01 9.40688670e-01
2.68674523e-01 -8.78115356e-01 -7.28343725e-01 3.55191082e-01
-2.48495623e-01 -1.00836825e+00 -6.66522980e-01 -5.49956635e-02
-9.43960845e-01 -1.02455688e+00 -9.72893357e-01 -7.93889225e-01
7.32126653e-01 -1.37048766e-01 7.35083461e-01 8.59227777e-02
-6.40530705e-01 2.04675794e-01 1.29633486e-01 -3.09289753e-01
1.51264086e-01 5.44803143e-01 3.58914763e-01 -5.27071238e-01
-2.92935908e-01 -8.44898343e-01 -8.36409986e-01 4.60120946e-01
-4.87981379e-01 1.18112348e-01 5.17816484e-01 6.85611188e-01
8.25824380e-01 -4.64624047e-01 3.71712178e-01 -2.18083978e-01
2.09330887e-01 1.21693842e-01 -6.47856236e-01 -3.21633779e-02
-5.29340863e-01 3.07158560e-01 4.36253697e-01 -8.05146337e-01
-9.37208593e-01 5.66397846e-01 -2.34363884e-01 -6.65926218e-01
3.01950634e-01 4.13639396e-01 -1.16936741e-02 -4.07953709e-01
4.18420553e-01 -2.51572520e-01 5.95713973e-01 -7.13487029e-01
4.70950097e-01 5.60805142e-01 6.12806737e-01 -9.03646529e-01
5.25565147e-01 5.82705498e-01 3.39494020e-01 -7.82559633e-01
-7.06143737e-01 -2.27214724e-01 -1.01703358e+00 -3.10703129e-01
6.55606031e-01 -6.56011641e-01 -1.42688608e+00 5.97696781e-01
-1.42241406e+00 -4.18660074e-01 -7.59092672e-03 5.00639498e-01
-9.05190468e-01 4.52157855e-01 -6.88597679e-01 -7.65376151e-01
-6.20556235e-01 -1.41664374e+00 1.21667671e+00 -2.46552810e-01
-6.41395926e-01 -4.90862578e-01 -3.09195966e-01 3.92773598e-01
1.44823268e-01 3.29774857e-01 8.32461834e-01 3.88055205e-01
-6.33754551e-01 -4.49862540e-01 -2.21934989e-01 6.63752295e-03
-3.37008499e-02 -1.22585103e-01 -6.14420056e-01 -4.70845461e-01
-4.17452484e-01 -3.17375004e-01 3.83679420e-01 5.71852326e-01
1.17775369e+00 -1.20334931e-01 -2.19824791e-01 6.29281521e-01
1.20246184e+00 -2.83951551e-01 2.39556819e-01 2.85534322e-01
1.09733605e+00 6.66025698e-01 3.57177675e-01 4.54626888e-01
4.43719536e-01 9.85508323e-01 4.22610402e-01 1.97433889e-01
-4.16178733e-01 -2.67914802e-01 1.49401808e-02 6.93013668e-01
-7.47716129e-01 3.28183144e-01 -9.85876143e-01 3.82650703e-01
-1.80977273e+00 -4.34699416e-01 2.37250365e-02 2.11721134e+00
1.08377934e+00 -9.57703777e-03 3.25760424e-01 1.33656785e-01
4.18299288e-01 -1.89044118e-01 -1.01827621e+00 7.40437433e-02
4.42960054e-01 4.23222840e-01 5.88390768e-01 5.92511237e-01
-8.01389873e-01 1.14693487e+00 6.20588636e+00 6.10371232e-01
-1.14262974e+00 -1.28954828e-01 -1.03915364e-01 -3.25303793e-01
1.81072205e-01 -2.19089955e-01 -7.57312715e-01 4.19128090e-02
1.93455175e-01 2.96775073e-01 7.78594792e-01 9.75772738e-01
1.62677109e-01 -3.88868265e-02 -1.22595954e+00 1.08337379e+00
-3.10959183e-02 -1.15050435e+00 4.53394558e-03 2.50619233e-01
4.56454545e-01 -5.55849001e-02 -1.63536847e-01 -2.88720965e-01
-1.72138765e-01 -8.63533378e-01 1.18816710e+00 5.89222789e-01
1.19016361e+00 -6.93845928e-01 3.86569768e-01 4.14974421e-01
-1.42809820e+00 5.78311384e-02 -1.41664818e-01 -3.20167124e-01
3.37450266e-01 3.34967226e-01 -5.74031472e-01 1.82569504e-01
6.22253835e-01 3.90358329e-01 -5.47228940e-02 5.64823747e-01
-3.27341586e-01 -9.16741323e-03 -7.39088237e-01 6.76276311e-02
-2.13377222e-01 -9.64463204e-02 6.21879280e-01 7.80829966e-01
5.93224801e-02 1.75935887e-02 2.51540422e-01 9.78936136e-01
-1.78256482e-01 -1.26692921e-01 -1.21784039e-01 1.40860155e-01
4.73330408e-01 1.06400692e+00 -7.59124994e-01 -1.05983257e-01
1.16990283e-01 9.72587049e-01 5.37950695e-01 1.10847898e-01
-5.02326667e-01 -4.07807887e-01 6.17820084e-01 4.78624970e-01
3.03031266e-01 -8.07783008e-01 -7.24077046e-01 -1.14476097e+00
5.38391173e-01 -6.64125741e-01 -3.04645181e-01 -5.32787263e-01
-9.97448027e-01 1.41025946e-01 -6.49273992e-02 -9.85729158e-01
-4.19066459e-01 -8.27034175e-01 8.25063661e-02 8.33297491e-01
-1.18251312e+00 -1.39551187e+00 -4.05126780e-01 3.38461936e-01
5.01907885e-01 3.55683059e-01 7.26447821e-01 1.85572803e-01
-3.82130742e-01 6.38428330e-01 -2.81196952e-01 2.36641467e-01
6.29586041e-01 -9.30539846e-01 3.43364477e-01 3.32444221e-01
-3.63901287e-01 9.22965050e-01 6.32623315e-01 -8.27336013e-01
-2.06134248e+00 -6.72160864e-01 5.39870024e-01 -6.08836293e-01
5.15241265e-01 -4.07589793e-01 -5.68815887e-01 7.56376147e-01
-5.89424193e-01 4.33969125e-02 -4.01409576e-03 1.51027277e-01
-4.23080385e-01 -3.82657386e-02 -1.22587192e+00 7.87877023e-01
1.43524539e+00 -4.73929614e-01 -5.01422584e-01 1.32205859e-01
3.04142535e-01 -8.85166049e-01 -1.20956075e+00 4.13734794e-01
1.62218785e+00 -3.10205251e-01 1.31987298e+00 -2.37405971e-01
3.96089762e-01 -4.03060585e-01 -3.34287831e-03 -8.29137444e-01
-2.09252432e-01 -4.98219401e-01 -4.39670742e-01 8.79149973e-01
6.11817650e-02 -3.39379311e-01 1.24816859e+00 8.18880022e-01
1.56312153e-01 -1.06803763e+00 -1.07795203e+00 -7.00835168e-01
7.12485984e-02 -4.26338166e-01 2.70224929e-01 5.16418576e-01
-3.06656528e-02 1.83887780e-01 -3.11564535e-01 2.61090435e-02
9.40710664e-01 9.96519104e-02 9.41920459e-01 -1.38816571e+00
-3.07636142e-01 -4.92096037e-01 -2.87998408e-01 -1.30258203e+00
2.50091881e-01 -7.82320380e-01 3.94046605e-01 -1.58181179e+00
-3.58474962e-02 -7.01264262e-01 2.65408248e-01 5.99090993e-01
-5.23861572e-02 4.67075229e-01 2.55242467e-01 3.75328958e-01
8.32457468e-02 3.50184232e-01 1.47588861e+00 9.40636918e-02
-4.02467579e-01 -1.70541942e-01 -1.07628562e-01 1.12421668e+00
5.40645003e-01 -2.38794252e-01 -2.45416118e-03 -8.13958168e-01
1.76094964e-01 2.22431421e-01 5.32844424e-01 -8.31579626e-01
1.69375703e-01 -1.16484411e-01 4.77599412e-01 -5.20156860e-01
5.30150712e-01 -8.77951503e-01 2.91270226e-01 6.94499552e-01
-7.52663687e-02 -7.40487799e-02 4.02814597e-02 3.70516121e-01
3.93172264e-01 -4.27114256e-02 8.82286847e-01 -2.35297546e-01
-2.43453786e-01 5.28550208e-01 -1.21830642e-01 -2.62550861e-01
7.05152929e-01 -4.33370769e-01 3.23706239e-01 -1.36998653e-01
-6.40118778e-01 1.02570824e-01 6.83433473e-01 2.69574583e-01
4.39675122e-01 -1.10199559e+00 -3.73674154e-01 1.60270669e-02
-2.46663809e-01 6.16667449e-01 1.37934670e-01 1.01053941e+00
-8.74645412e-01 1.65600687e-01 -2.33621910e-01 -7.34089911e-01
-1.10900569e+00 2.53723085e-01 1.74559832e-01 -4.41248640e-02
-7.90234327e-01 8.16939950e-01 -3.34982038e-01 -6.93737984e-01
3.42846721e-01 -4.63515013e-01 3.45801830e-01 4.54630740e-02
1.49287358e-01 9.15277183e-01 3.73389386e-03 -5.02542555e-01
-4.61334705e-01 1.18044174e+00 1.95924670e-01 -1.89368159e-01
1.47469032e+00 1.30206019e-01 -3.03805918e-01 9.74455476e-02
1.12471271e+00 -1.15941033e-01 -1.54179704e+00 -3.07039339e-02
-1.41400293e-01 -2.09600866e-01 1.09649062e-01 -6.87759697e-01
-1.04993165e+00 9.12554741e-01 4.24696863e-01 -6.75763190e-01
5.17679214e-01 4.49975906e-03 9.80406761e-01 4.99547988e-01
6.88112915e-01 -1.18180501e+00 -9.02155563e-02 3.95459086e-01
1.24460673e+00 -8.85346949e-01 4.58785564e-01 -7.69848049e-01
-8.96611437e-02 1.27536404e+00 5.49894214e-01 -4.46781486e-01
6.15589678e-01 4.25960749e-01 -5.60698099e-02 -1.00965649e-01
6.13475330e-02 5.63648567e-02 4.60347146e-01 4.60813761e-01
4.11964834e-01 5.09707406e-02 -3.60561699e-01 3.36861521e-01
-2.53455669e-01 3.61603081e-01 -5.81891611e-02 1.14674044e+00
-2.52048761e-01 -1.15774000e+00 -4.29515660e-01 5.39830208e-01
-3.86118442e-01 2.09755987e-01 -1.52970895e-01 7.64258802e-01
3.82375829e-02 4.85514969e-01 -1.66492339e-03 -2.96550155e-01
5.76370895e-01 -3.78805697e-02 1.19359279e+00 -3.30009252e-01
-4.40892190e-01 3.15817147e-01 -1.32032678e-01 -7.64219224e-01
-3.24871957e-01 -5.77416360e-01 -1.38978064e+00 -2.94176489e-01
-5.33888698e-01 -4.76257622e-01 9.69044328e-01 1.00414419e+00
4.34909433e-01 3.60783309e-01 -4.10120115e-02 -1.96076083e+00
-7.79121459e-01 -9.87711370e-01 -4.06464398e-01 9.53261033e-02
1.77735358e-01 -1.34331226e+00 5.70642874e-02 5.78392446e-02]
|
[6.6884660720825195, -1.0028648376464844]
|
d68104f6-901c-4d03-b8fd-48731d431ff6
|
automated-audio-captioning-and-language-based
|
2207.04156
| null |
https://arxiv.org/abs/2207.04156v2
|
https://arxiv.org/pdf/2207.04156v2.pdf
|
Automated Audio Captioning and Language-Based Audio Retrieval
|
This project involved participation in the DCASE 2022 Competition (Task 6) which had two subtasks: (1) Automated Audio Captioning and (2) Language-Based Audio Retrieval. The first subtask involved the generation of a textual description for audio samples, while the goal of the second was to find audio samples within a fixed dataset that match a given description. For both subtasks, the Clotho dataset was used. The models were evaluated on BLEU1, BLEU2, BLEU3, ROUGEL, METEOR, CIDEr, SPICE, and SPIDEr scores for audio captioning and R1, R5, R10 and mARP10 scores for audio retrieval. We have conducted a handful of experiments that modify the baseline models for these tasks. Our final architecture for Automated Audio Captioning is close to the baseline performance, while our model for Language-Based Audio Retrieval has surpassed its counterpart.
|
['Ankit Shah', 'Iffanice Houndayi', 'Yi Song', 'Patrick Kollman', 'Hyejin Park', 'Clive Gomes']
|
2022-07-08
| null | null | null | null |
['audio-captioning']
|
['audio']
|
[ 1.53917909e-01 8.71105194e-02 6.04742169e-02 -2.30844766e-01
-2.10882354e+00 -9.58620548e-01 8.00952733e-01 3.16455930e-01
-2.42228299e-01 7.01216757e-01 8.17977130e-01 1.55326575e-01
1.46629969e-02 2.93515250e-03 -5.94078958e-01 -1.47331432e-01
-3.16161126e-01 6.64901495e-01 3.79294753e-01 -1.37738347e-01
2.77127743e-01 9.79318656e-03 -1.79076767e+00 8.64782155e-01
1.15727022e-01 1.47285676e+00 -1.57382153e-02 1.23077130e+00
5.08935824e-02 8.70112658e-01 -7.08501995e-01 -2.06422433e-01
-1.55386239e-01 -5.51218510e-01 -1.12375474e+00 -3.04704696e-01
5.37810802e-01 -1.81063861e-02 -3.04480195e-01 6.84098661e-01
9.20404017e-01 2.09895000e-01 5.90006173e-01 -1.58856070e+00
-3.69174033e-01 1.08095860e+00 -2.10065216e-01 1.66924030e-01
1.05676687e+00 5.10100536e-02 1.60956824e+00 -8.02576065e-01
3.92725378e-01 1.39587462e+00 5.38992941e-01 5.35330892e-01
-8.28331888e-01 -7.93801606e-01 -1.50755852e-01 2.25207657e-01
-1.61601174e+00 -7.70590842e-01 3.92411709e-01 -5.10887623e-01
8.47144425e-01 5.00520527e-01 3.99088085e-01 1.01288462e+00
-3.36204082e-01 8.00872028e-01 5.87552011e-01 -3.78082037e-01
2.54674584e-01 7.92645961e-02 3.00935619e-02 -2.84487475e-02
-5.51139593e-01 -2.34122813e-01 -9.23292220e-01 -3.74427050e-01
2.12149262e-01 -9.24610734e-01 -3.34066212e-01 2.66679943e-01
-1.52634847e+00 5.97647488e-01 3.13508883e-02 2.87221909e-01
-3.82423043e-01 4.81515676e-01 7.47185409e-01 4.56206560e-01
3.70049119e-01 8.19839001e-01 -3.13208789e-01 -5.53669572e-01
-1.19053185e+00 5.75526893e-01 7.64390767e-01 1.09866714e+00
2.52874285e-01 -2.05086008e-01 -8.11016202e-01 1.06432319e+00
3.63245785e-01 5.99201083e-01 5.34412086e-01 -1.13937271e+00
6.19934797e-01 -2.79251605e-01 4.55452830e-01 -5.82247257e-01
-1.69008285e-01 -2.23385468e-02 -3.07303429e-01 -5.66898823e-01
2.21700832e-01 -2.26831421e-01 -9.09965873e-01 1.74769044e+00
-2.57909417e-01 1.82393283e-01 -1.67111158e-02 9.66659963e-01
1.28372777e+00 1.23907065e+00 2.06826419e-01 -2.26308510e-01
1.39012206e+00 -1.10015523e+00 -8.34342301e-01 -6.20610639e-02
3.81888598e-01 -1.36866939e+00 1.12731695e+00 4.31207478e-01
-1.40268028e+00 -6.30438745e-01 -8.71873319e-01 5.73478080e-02
6.79054484e-02 5.28891124e-02 2.35244587e-01 4.84005034e-01
-1.39814150e+00 8.34591910e-02 -3.44167680e-01 -3.38190734e-01
-2.22912803e-01 6.64892606e-03 -2.08862543e-01 2.89466947e-01
-1.59233809e+00 2.28139013e-01 5.59474647e-01 -2.92631745e-01
-1.45565498e+00 -6.84259772e-01 -5.74744821e-01 1.77700937e-01
7.93315098e-02 -3.37972164e-01 1.94442701e+00 -7.51376748e-01
-1.33306634e+00 1.03748655e+00 1.71086803e-01 -7.29603827e-01
2.28150457e-01 -2.05099374e-01 -6.49205089e-01 4.10733819e-01
7.23216087e-02 1.38927305e+00 6.48942411e-01 -1.03942990e+00
-8.39891911e-01 1.37695909e-01 2.57092789e-02 4.20577049e-01
-9.11161229e-02 5.27787685e-01 -7.43313491e-01 -7.74883986e-01
-3.45379800e-01 -9.97008622e-01 4.03711379e-01 -2.89365351e-01
-4.06163573e-01 -3.58120590e-01 3.82828712e-01 -9.62913811e-01
1.57866609e+00 -2.33449674e+00 -7.95725882e-02 4.07746620e-02
-2.75333256e-01 -2.46560741e-02 -7.62075961e-01 6.60334587e-01
-2.20330119e-01 2.62907952e-01 1.96444497e-01 -1.65672854e-01
2.86243588e-01 -6.23169363e-01 -7.59489357e-01 -2.62111574e-01
5.24077390e-04 5.48648477e-01 -9.85087335e-01 -4.45710689e-01
-3.45080435e-01 3.44980508e-01 -4.88119572e-01 4.28037494e-01
-4.99379218e-01 3.81821506e-02 -1.37086496e-01 5.32989740e-01
7.22392127e-02 9.64249521e-02 -2.29233891e-01 -8.56603868e-03
2.17079539e-02 7.82791913e-01 -1.12308562e+00 1.72556567e+00
-5.79862297e-01 9.43689227e-01 2.59551138e-01 -2.57054746e-01
8.68278086e-01 1.05901289e+00 6.27844751e-01 -5.54035962e-01
-3.29445004e-01 3.95448625e-01 -5.37834346e-01 -4.38884586e-01
7.12166607e-01 3.61506641e-02 -5.36826015e-01 6.23825788e-01
1.43870309e-01 -5.28639734e-01 3.88325661e-01 3.42113882e-01
1.11720037e+00 -1.08069062e-01 -8.66910592e-02 5.14287315e-02
5.26429653e-01 -6.08385354e-02 -3.21061760e-02 7.00849593e-01
-2.07894772e-01 1.19753969e+00 5.03190517e-01 -1.90287024e-01
-1.05448043e+00 -1.13301015e+00 3.11535358e-01 1.41016018e+00
-1.49311438e-01 -6.28032506e-01 -8.03942084e-01 -3.20010871e-01
-3.34829450e-01 7.90014088e-01 -2.52819151e-01 -2.01403603e-01
-3.22259218e-01 -2.87449181e-01 1.21800148e+00 2.64782399e-01
2.02608690e-01 -1.43531954e+00 -1.63016945e-01 3.08051199e-01
-1.02892137e+00 -1.06128180e+00 -1.15378034e+00 -1.48648053e-01
-3.73849660e-01 -6.14622176e-01 -1.06869614e+00 -9.41655695e-01
-3.28307450e-01 -1.34448195e-03 1.42656934e+00 -4.13788170e-01
-6.34177029e-02 7.18327522e-01 -7.00859249e-01 -6.53467894e-01
-5.40815890e-01 4.29075688e-01 1.01598315e-01 -1.22637674e-01
2.62304187e-01 -3.56814146e-01 -4.15067971e-01 3.75865996e-01
-9.28441823e-01 -3.50819558e-01 3.60772729e-01 4.34062034e-01
5.81602335e-01 -4.28805530e-01 9.51838017e-01 -3.55969445e-04
1.07609594e+00 -3.28819782e-01 -3.45379174e-01 2.18877479e-01
-2.02280313e-01 -2.23029003e-01 2.78955340e-01 -5.00562847e-01
-5.16146243e-01 1.98950380e-01 -3.12370181e-01 -2.42649376e-01
-7.41868019e-02 6.00296915e-01 -8.97939578e-02 6.23507142e-01
7.11472511e-01 1.18961766e-01 -2.76832253e-01 -6.34321809e-01
2.14326397e-01 1.26412618e+00 8.74580562e-01 -5.27457356e-01
6.45739555e-01 -2.28011057e-01 -7.06332207e-01 -8.29958022e-01
-9.30038214e-01 -7.96513915e-01 -1.88639224e-01 -4.70098615e-01
7.98129559e-01 -1.47093379e+00 -3.37660879e-01 3.55235159e-01
-1.31571186e+00 -1.32980362e-01 -4.87805367e-01 6.21246934e-01
-8.41186345e-01 1.08999170e-01 -4.46445286e-01 -9.18186903e-01
-6.49426222e-01 -9.15008485e-01 1.45001400e+00 -9.39997360e-02
-7.68959880e-01 -3.65924031e-01 3.86831164e-01 6.40385389e-01
3.03408802e-01 -1.00770138e-01 8.05530131e-01 -9.87466037e-01
-2.99407154e-01 -3.80917728e-01 -2.12195069e-01 2.71018058e-01
-2.12536663e-01 -3.47129814e-02 -1.32662976e+00 -2.79384047e-01
-6.30511642e-01 -9.33436155e-01 7.50446200e-01 3.57033849e-01
1.13335133e+00 -3.37837011e-01 1.12149492e-01 -4.22757119e-02
7.50964165e-01 4.72656727e-01 7.06113040e-01 1.62598461e-01
1.01675637e-01 5.37280917e-01 8.53511691e-01 5.41251779e-01
2.26259679e-01 1.06430364e+00 1.92672059e-01 3.49672794e-01
-5.36388993e-01 -5.06017208e-01 7.39399254e-01 1.23179424e+00
3.52350175e-01 -6.14886582e-01 -1.03569627e+00 1.05335724e+00
-1.59878886e+00 -9.84655142e-01 1.50592342e-01 2.16882324e+00
9.85955358e-01 1.30315661e-01 6.31470084e-01 3.59527111e-01
8.07058990e-01 1.62138328e-01 -5.32926954e-02 -5.00747859e-01
1.31118238e-01 -7.68351927e-02 -1.08129673e-01 4.58573103e-01
-1.02443945e+00 5.39967537e-01 7.47254133e+00 9.85462189e-01
-8.29539239e-01 7.28109851e-02 5.35674632e-01 -2.96049029e-01
-3.80977184e-01 -2.75585145e-01 -5.23967266e-01 5.70608020e-01
1.69199860e+00 -3.79927129e-01 6.36853337e-01 4.41183925e-01
2.58624464e-01 8.54939297e-02 -1.33276844e+00 1.09530509e+00
7.65633583e-02 -1.07568252e+00 3.85616332e-01 -4.95479047e-01
4.23659146e-01 1.77343294e-01 1.30633384e-01 6.52597308e-01
-4.12494466e-02 -9.65237379e-01 1.10851538e+00 4.09486651e-01
1.19443369e+00 -7.24042296e-01 7.56647587e-01 -1.11879677e-01
-1.20873272e+00 1.84250280e-01 1.60677046e-01 3.06043446e-01
3.73041660e-01 3.80713701e-01 -1.10799861e+00 3.49572182e-01
7.62489021e-01 2.10855812e-01 -4.46691990e-01 1.49958229e+00
4.21532728e-02 8.50696623e-01 -3.34714502e-01 -2.34294638e-01
4.07140493e-01 5.66549301e-01 8.10924470e-01 1.49321651e+00
5.09676695e-01 -3.46890301e-01 1.11509740e-01 3.63462061e-01
-5.53625345e-01 2.92385191e-01 -1.56122983e-01 -4.61690545e-01
9.04970288e-01 8.59818995e-01 -4.61790293e-01 -2.62586713e-01
1.41024977e-01 6.11553133e-01 -5.35389125e-01 1.87666997e-01
-9.18632090e-01 -9.70365226e-01 3.70700568e-01 2.05351248e-01
1.18816100e-01 2.88313150e-01 1.60204098e-01 -7.59735584e-01
1.06501970e-02 -1.23165607e+00 6.36729896e-01 -1.35187221e+00
-1.02506196e+00 8.49243343e-01 2.35409975e-01 -1.38478017e+00
-7.04757929e-01 7.10261241e-02 -3.23447973e-01 7.87773073e-01
-1.18691146e+00 -7.31162250e-01 -2.07783580e-01 3.25726151e-01
8.43568921e-01 -4.18261588e-01 9.46119726e-01 7.90074110e-01
3.45354713e-02 7.25960135e-01 5.21588698e-03 -1.21463258e-02
1.17623675e+00 -1.15399468e+00 5.19516706e-01 2.39146620e-01
4.12543267e-01 -2.55399272e-02 1.03527927e+00 -3.63553286e-01
-8.40257108e-01 -1.18356407e+00 1.36189353e+00 -5.01590669e-01
9.25242543e-01 -3.62898588e-01 -4.96403813e-01 4.97469813e-01
1.14749573e-01 -5.03382087e-01 6.55081987e-01 -1.36728585e-02
-4.62404370e-01 -2.14391753e-01 -7.99423158e-01 2.50470132e-01
5.25044918e-01 -1.00468016e+00 -5.44310749e-01 6.19143605e-01
1.18478620e+00 -2.97504157e-01 -1.03982437e+00 4.31678519e-02
8.71048987e-01 -2.29212761e-01 9.17200446e-01 -5.34597456e-01
3.90664220e-01 -3.80189657e-01 -4.11204487e-01 -1.10375047e+00
-1.49916746e-02 -1.14695549e+00 7.72135183e-02 1.71127856e+00
9.53683197e-01 3.13383102e-01 4.27853674e-01 1.44303173e-01
-1.74766317e-01 -2.19094884e-02 -1.10488296e+00 -1.06393886e+00
-1.61630865e-02 -6.64984286e-01 6.65191948e-01 4.82564628e-01
2.66683009e-02 7.72284865e-01 -4.37954366e-01 -7.32600093e-02
1.84388861e-01 -1.87468573e-01 6.14171147e-01 -1.24015141e+00
-1.91027820e-01 -4.08792555e-01 -2.75855571e-01 -8.41836751e-01
-1.61867023e-01 -9.02888477e-01 4.76963341e-01 -1.54002988e+00
2.51749247e-01 -7.38766938e-02 -3.20105582e-01 5.08869648e-01
1.83919325e-01 6.25404894e-01 4.37763035e-01 2.35923335e-01
-1.09101808e+00 3.59580964e-01 7.09864497e-01 -5.61220646e-01
-4.38683927e-01 1.17606662e-01 -6.69385672e-01 1.45021111e-01
6.41252577e-01 -5.52602887e-01 -3.91367376e-01 -3.36942285e-01
3.66998613e-01 2.88508713e-01 5.74253406e-03 -1.24290943e+00
1.06502980e-01 1.78506598e-01 -2.81499594e-01 -7.21639514e-01
4.72662508e-01 -5.78199387e-01 2.11948276e-01 7.98585787e-02
-1.12237024e+00 2.37285152e-01 2.94095695e-01 3.90910208e-01
-6.96677089e-01 -3.54953021e-01 5.44805646e-01 1.43190116e-01
-4.20309037e-01 2.20366288e-03 -8.07100356e-01 3.65585178e-01
6.52658820e-01 1.97028741e-01 -7.00619537e-03 -1.17179298e+00
-9.19986725e-01 4.74630326e-01 -4.38167974e-02 8.91485631e-01
5.59111238e-01 -1.72460878e+00 -1.21955931e+00 -4.09604996e-01
5.58223784e-01 -3.27337354e-01 8.97305459e-03 5.08117497e-01
-3.78762782e-01 8.96545827e-01 -7.07502440e-02 -4.91449475e-01
-1.43106270e+00 2.69745469e-01 6.85310289e-02 -4.51540470e-01
3.07862367e-03 9.12055790e-01 -2.04527065e-01 -2.61486202e-01
1.03355086e+00 -2.01596290e-01 -2.42249265e-01 4.37119097e-01
9.10225451e-01 4.44522768e-01 1.83975086e-01 -6.72813952e-01
-3.61948550e-01 2.23779410e-01 -4.26540431e-03 -7.65854895e-01
1.15300238e+00 -2.38183171e-01 2.00366288e-01 6.77860081e-01
1.26625705e+00 -9.50195268e-02 -6.24887586e-01 -1.37331970e-02
2.60901779e-01 1.14841640e-01 1.22120060e-01 -1.22158039e+00
-5.45442343e-01 8.08225870e-01 6.24415517e-01 5.45609236e-01
1.20438135e+00 7.85444975e-02 1.09140825e+00 4.18066919e-01
2.32088134e-01 -1.25467563e+00 2.56889373e-01 7.14992285e-01
1.28286600e+00 -8.22443366e-01 -3.34720999e-01 5.42876944e-02
-6.87465549e-01 1.00741446e+00 2.26069793e-01 4.13872898e-01
2.10167408e-01 2.85516102e-02 1.69412047e-01 -5.12112416e-02
-1.16378152e+00 -1.25470221e-01 7.43993342e-01 5.57924569e-01
6.93681240e-01 -1.07756145e-02 -1.94023512e-02 6.60595179e-01
-3.67019981e-01 -1.78921092e-02 5.34527898e-01 5.23119807e-01
-5.60841441e-01 -7.81096160e-01 -5.31322241e-01 1.06750607e-01
-7.58670926e-01 -4.38584499e-02 -1.04949355e+00 5.22469163e-01
-4.42851275e-01 1.28907835e+00 8.11622813e-02 -6.75034463e-01
5.36139846e-01 3.10758650e-01 1.22120596e-01 -6.10545933e-01
-8.12188923e-01 2.95388460e-01 5.83639681e-01 -4.79418427e-01
-3.38652939e-01 -5.38842738e-01 -9.65390682e-01 1.50479317e-01
-1.52161837e-01 8.62214565e-01 7.17075586e-01 4.92736846e-01
2.58330166e-01 4.64283735e-01 7.65642941e-01 -7.52264023e-01
-4.15244848e-01 -1.28801608e+00 -5.16281068e-01 2.57782519e-01
4.80800182e-01 -3.82725745e-02 -4.23211843e-01 3.07314336e-01]
|
[15.286962509155273, 4.893754482269287]
|
2226dd92-5612-402d-ba8c-6681d33a3871
|
a-comprehensive-survey-on-affective-computing
|
2305.07665
| null |
https://arxiv.org/abs/2305.07665v1
|
https://arxiv.org/pdf/2305.07665v1.pdf
|
A Comprehensive Survey on Affective Computing; Challenges, Trends, Applications, and Future Directions
|
As the name suggests, affective computing aims to recognize human emotions, sentiments, and feelings. There is a wide range of fields that study affective computing, including languages, sociology, psychology, computer science, and physiology. However, no research has ever been done to determine how machine learning (ML) and mixed reality (XR) interact together. This paper discusses the significance of affective computing, as well as its ideas, conceptions, methods, and outcomes. By using approaches of ML and XR, we survey and discuss recent methodologies in affective computing. We survey the state-of-the-art approaches along with current affective data resources. Further, we discuss various applications where affective computing has a significant impact, which will aid future scholars in gaining a better understanding of its significance and practical relevance.
|
['Jong Weon Lee', 'Md. Jalil Piran', 'Imran Ullah Khan', 'Haseeb Ali Khan', 'Sitara Afzal']
|
2023-05-08
| null | null | null | null |
['mixed-reality']
|
['computer-vision']
|
[-1.80512667e-02 -1.43737584e-01 -2.07789734e-01 -6.15735471e-01
4.67989556e-02 -5.19464433e-01 3.97379667e-01 3.46228272e-01
-1.75512418e-01 5.88630259e-01 2.43858814e-01 2.84112155e-01
2.46895030e-01 -7.00838327e-01 3.07738125e-01 -5.83088875e-01
6.66442215e-02 -1.59586906e-01 -6.68217421e-01 -4.36498702e-01
3.53694499e-01 2.40092307e-01 -1.80597687e+00 2.20422596e-01
6.86205566e-01 1.18301654e+00 -3.11044693e-01 4.09525484e-01
-2.48713285e-01 9.57057238e-01 -6.06120706e-01 -8.38811576e-01
-3.67512256e-01 -3.48510265e-01 -7.98326015e-01 -1.96036443e-01
-4.90820944e-01 1.83406875e-01 1.71648428e-01 1.05175102e+00
6.31644607e-01 3.43484014e-01 5.12793839e-01 -1.45279121e+00
-1.00184906e+00 3.17112148e-01 -4.33291227e-01 -1.42191544e-01
7.72055864e-01 -1.94463462e-01 9.45392191e-01 -1.04429555e+00
3.21888983e-01 8.27280104e-01 6.12899125e-01 6.47367597e-01
-5.66636801e-01 -5.61567605e-01 1.71741351e-01 2.26702005e-01
-1.29614127e+00 -3.48281682e-01 9.37650502e-01 -3.62377226e-01
1.14745939e+00 4.88855213e-01 1.25283420e+00 9.13437009e-01
3.17553431e-01 9.70533311e-01 1.32963455e+00 -7.08292425e-01
4.08930421e-01 7.53569663e-01 3.60236079e-01 3.11022490e-01
-4.77290079e-02 -4.71253097e-01 -5.06371439e-01 -1.33957773e-01
2.64810681e-01 6.50614128e-02 1.22610793e-01 8.69471729e-02
-8.62714767e-01 6.61188483e-01 -7.68229812e-02 6.71464324e-01
-8.04026425e-01 -3.55396513e-03 5.59886813e-01 1.97390273e-01
7.74108648e-01 6.78626418e-01 -5.28999925e-01 -8.23758900e-01
-3.42387974e-01 -1.93410322e-01 9.51670766e-01 6.66882217e-01
6.65759444e-01 2.71484047e-01 3.62443924e-01 1.18874013e+00
2.36030534e-01 4.81576860e-01 4.15449709e-01 -1.08046556e+00
-6.70394719e-01 6.29392028e-01 -6.17976338e-02 -1.36866438e+00
-2.61859447e-01 1.70549184e-01 -7.59520888e-01 -9.02766064e-02
-5.06248415e-01 -4.86691505e-01 -1.67015806e-01 1.62028015e+00
1.06112510e-01 -3.73219363e-02 4.57471192e-01 1.01351321e+00
1.30805361e+00 5.73716819e-01 4.17233855e-01 -7.03290939e-01
1.61751091e+00 -6.21188521e-01 -1.32015419e+00 -5.37855744e-01
2.81463891e-01 -9.67875957e-01 9.75944281e-01 5.46664417e-01
-1.10942304e+00 -2.11747810e-01 -7.88224280e-01 1.29437178e-01
-7.96570182e-01 -1.06344640e-01 1.52727318e+00 1.07779968e+00
-1.10947168e+00 -6.20354973e-02 -4.96722639e-01 -8.07326257e-01
1.54332435e-02 3.54281604e-01 2.22473908e-02 4.43371505e-01
-1.60102308e+00 1.19259202e+00 -1.18186712e-01 2.16208631e-04
-1.15509098e-02 -8.12437907e-02 -6.93867028e-01 -1.96054280e-01
6.21870868e-02 -6.46652281e-01 1.11371589e+00 -1.61668861e+00
-1.82079089e+00 1.15024519e+00 -2.50693053e-01 2.84668386e-01
-4.71853137e-01 -3.57973665e-01 -1.02977288e+00 1.11916631e-01
-3.80427867e-01 3.54689121e-01 1.78164229e-01 -1.17189395e+00
-3.26638728e-01 -6.13103211e-01 1.07723363e-01 7.23017275e-01
-9.01610553e-01 5.34909904e-01 -4.56228495e-01 -1.74622610e-01
-9.77068022e-03 -8.67038727e-01 -4.81062889e-01 -4.82747227e-01
9.70269181e-03 -2.22221226e-01 3.83594215e-01 -3.82514467e-04
1.48437595e+00 -2.25464368e+00 -1.75425082e-01 3.22627693e-01
1.65397540e-01 9.79403406e-02 1.27620026e-01 8.67197573e-01
5.79423979e-02 2.84648627e-01 3.19863379e-01 -1.47343606e-01
2.70547327e-02 7.89916217e-02 -2.47627214e-01 1.47928759e-01
-2.51297951e-02 9.88183796e-01 -1.11783373e+00 -5.11746943e-01
4.03310537e-01 6.56154156e-01 -3.53054345e-01 3.62903297e-01
2.11528748e-01 1.76726282e-01 -7.19228923e-01 1.02283800e+00
2.29722515e-01 -2.41839245e-01 5.59816062e-01 -8.41089636e-02
-2.95593351e-01 -6.47867545e-02 -7.40429223e-01 1.12899661e+00
-6.52379274e-01 7.34247923e-01 -4.51710150e-02 -7.97365189e-01
1.15993142e+00 6.29448295e-01 8.97272348e-01 -8.14221978e-01
4.40412164e-01 1.41901998e-02 -3.54997814e-01 -9.24336612e-01
1.01331139e+00 -4.29227978e-01 -3.14503133e-01 5.70191741e-01
-3.21721464e-01 -2.38379136e-01 -1.91220134e-01 1.26081720e-01
8.64112258e-01 -1.11328028e-01 7.46859431e-01 -4.40775491e-02
4.00948375e-01 -2.40993187e-01 5.02889156e-01 2.17144325e-01
-6.98918581e-01 1.50539845e-01 2.47275308e-01 -4.29161578e-01
-3.04244846e-01 -6.66014612e-01 2.52588671e-02 1.48178220e+00
4.51710016e-01 -6.21693671e-01 -6.79747283e-01 3.61000635e-02
-3.21470976e-01 6.59470379e-01 -6.74762607e-01 -2.75359333e-01
1.04866624e-01 -1.15273988e+00 3.35761964e-01 4.44286704e-01
1.61892101e-01 -1.58429098e+00 -7.65959382e-01 -9.39960703e-02
-4.99669909e-01 -1.12101734e+00 3.31763744e-01 7.64458701e-02
-6.46840513e-01 -6.36048317e-01 3.17307711e-02 -5.42030931e-01
5.40576816e-01 2.81393439e-01 1.50818026e+00 9.08772871e-02
-4.84468453e-02 8.98091972e-01 -6.34255052e-01 -6.84644699e-01
2.47475319e-02 -3.11780572e-01 3.94645035e-01 -5.31878136e-02
1.07744598e+00 -5.49829781e-01 -6.10111773e-01 -8.76252428e-02
-8.19874942e-01 -1.84177279e-01 4.51786578e-01 2.39593357e-01
4.02840465e-01 -1.27421379e-01 9.26638424e-01 -1.12039733e+00
1.17158628e+00 -8.23888361e-01 4.82436001e-01 2.45581105e-01
-9.51034188e-01 -7.95976937e-01 3.31237137e-01 -2.24595875e-01
-1.08255720e+00 -2.44319186e-01 -2.72305399e-01 -1.71887323e-01
-2.19682395e-01 7.91423380e-01 -6.81634173e-02 1.17101088e-01
4.39909667e-01 -1.84779882e-01 -9.31133553e-02 1.56042889e-01
5.38700819e-01 1.21165931e+00 1.68303445e-01 -5.20761192e-01
-2.30055481e-01 3.43908995e-01 -5.43258846e-01 -1.00821352e+00
-8.61444235e-01 -5.55215776e-01 -2.48529404e-01 -9.04213786e-01
8.92102003e-01 -8.17806005e-01 -9.82284427e-01 3.05204481e-01
-8.58240187e-01 1.25009775e-01 -2.84673125e-01 5.59572399e-01
-5.42688310e-01 7.52879530e-02 -9.04833496e-01 -1.43184197e+00
-5.99357605e-01 -8.58959138e-01 6.33833170e-01 6.79007113e-01
-9.90831435e-01 -1.14722300e+00 1.40240192e-01 3.96729380e-01
5.48029959e-01 1.33760765e-01 6.80816054e-01 -5.29011488e-01
3.89972806e-01 -2.00140625e-01 1.45302147e-01 1.66500002e-01
1.65508315e-01 5.29092073e-01 -1.27380073e+00 1.90882996e-01
1.49936482e-01 -7.59800136e-01 7.81168416e-02 2.17902988e-01
9.72923517e-01 -9.97194052e-02 -6.82792515e-02 3.83885833e-03
1.42913961e+00 6.46562815e-01 8.34286690e-01 1.42854705e-01
1.28751948e-01 8.28809738e-01 1.08543885e+00 9.45446432e-01
8.42589319e-01 1.46389157e-01 3.38235885e-01 -1.91424370e-01
6.81297541e-01 2.71757007e-01 5.38054824e-01 1.66561079e+00
-3.87377381e-01 -1.26696005e-01 -8.93718839e-01 4.47143078e-01
-1.93892765e+00 -1.05823612e+00 -8.88062567e-02 1.60745525e+00
9.08100784e-01 -3.40812296e-01 -2.10730582e-01 8.29754695e-02
5.48408628e-01 1.83782965e-01 -3.24655801e-01 -1.29164040e+00
-1.59613833e-01 2.89780408e-01 -4.80884433e-01 1.80093825e-01
-9.74215090e-01 1.27873719e+00 7.06694698e+00 3.85781854e-01
-1.29536653e+00 -1.10279769e-01 8.97213578e-01 -1.66512688e-03
-5.07142782e-01 -1.08979762e-01 2.27169301e-02 1.55280009e-01
8.86697114e-01 -3.37040901e-01 5.13965845e-01 1.07230675e+00
2.47856513e-01 -4.96090800e-01 -8.07684481e-01 1.22932768e+00
3.74159425e-01 -7.70679712e-01 -4.20137674e-01 -3.11073601e-01
7.07119703e-01 -3.43757212e-01 2.02752486e-01 5.55761516e-01
5.20021841e-02 -9.48813319e-01 3.49757642e-01 8.19764793e-01
7.30820656e-01 -9.35094893e-01 9.33750868e-01 -7.91332722e-02
-9.98783588e-01 2.54193157e-01 -2.06529707e-01 -8.49181771e-01
-1.43747255e-01 7.87075639e-01 -2.50217468e-01 3.91074985e-01
8.51361275e-01 7.83056676e-01 -1.83390096e-01 1.67209595e-01
3.45559344e-02 4.04417038e-01 4.61273417e-02 -7.79564202e-01
-1.56539887e-01 -7.38324344e-01 4.06530686e-02 1.49413955e+00
-4.67885938e-03 8.86967063e-01 -2.19575353e-02 3.96256238e-01
8.32073838e-02 7.22871482e-01 -7.02162683e-01 -6.25865698e-01
7.18704104e-01 1.83197737e+00 -9.21104670e-01 -4.11991566e-01
-5.76549113e-01 9.23758447e-01 8.29976872e-02 2.14473575e-01
-7.12817907e-01 -3.44754815e-01 1.00887764e+00 -5.43938637e-01
-6.48337901e-01 1.36819715e-02 -7.88354516e-01 -1.10315943e+00
-3.57290983e-01 -9.26133513e-01 1.35818571e-01 -1.09256327e+00
-1.52685165e+00 7.37030149e-01 -4.75107878e-01 -1.00440049e+00
-2.41368815e-01 -2.33343825e-01 -5.46463609e-01 4.38311934e-01
-9.02295530e-01 -9.17127073e-01 -5.66685140e-01 3.51497769e-01
1.37979597e-01 -3.60208489e-02 1.57729781e+00 2.04499856e-01
-6.41421497e-01 9.07727778e-02 -1.98696390e-01 -1.20627470e-01
9.21752453e-01 -9.72446442e-01 -3.05500001e-01 1.62739113e-01
-2.97243625e-01 8.61585915e-01 7.75675714e-01 -3.45825315e-01
-1.95882857e+00 -5.02105296e-01 9.62831497e-01 -3.84113163e-01
5.64350605e-01 7.49198571e-02 -4.52248991e-01 5.02509713e-01
6.63477480e-01 -2.97912657e-01 1.66608334e+00 7.68216729e-01
1.02699220e-01 -7.55773112e-02 -1.20650947e+00 9.54449594e-01
4.58369285e-01 -7.41571009e-01 -2.46286422e-01 -6.71677738e-02
2.92984188e-01 5.11195026e-02 -1.09493041e+00 4.63560492e-01
1.10375714e+00 -1.16194975e+00 7.09059298e-01 -4.17433321e-01
6.34513617e-01 1.84639305e-01 -3.42807561e-01 -1.24361992e+00
-2.77001768e-01 -4.10539865e-01 -9.78834033e-02 1.30395103e+00
8.40980560e-02 -6.40333533e-01 6.27620220e-01 1.10590661e+00
-2.74836682e-02 -1.29988503e+00 -2.35791549e-01 1.12251438e-01
-1.22172266e-01 -8.62189054e-01 4.82506424e-01 1.63101971e+00
9.92908537e-01 6.41192138e-01 -2.97438711e-01 -3.56734931e-01
-3.08320206e-02 4.47905540e-01 6.07418776e-01 -1.09763193e+00
3.26101363e-01 -5.54283082e-01 -3.45090240e-01 -4.18756962e-01
2.45587364e-01 -3.71949226e-01 -6.60889074e-02 -1.56148219e+00
3.80903006e-01 -4.15330976e-01 -6.11666262e-01 4.31729138e-01
-3.69380027e-01 7.06124187e-01 1.05132230e-01 8.19483101e-02
-1.04674327e+00 6.28612816e-01 1.13492990e+00 4.14429814e-01
-2.92760074e-01 -2.90490776e-01 -1.32809055e+00 1.17004979e+00
1.18534410e+00 1.30255565e-01 -4.86471236e-01 5.60037307e-02
8.23004484e-01 1.26253337e-01 -4.82875705e-01 -6.02157950e-01
2.34273896e-01 -7.37848401e-01 4.26744163e-01 -2.77152210e-01
8.10955822e-01 -7.78304696e-01 2.92172730e-01 -7.36002028e-02
-2.18999028e-01 4.11886185e-01 -1.86038781e-02 -6.05241768e-02
-4.77592796e-01 -2.02627972e-01 6.50248468e-01 -1.39691696e-01
-9.52757061e-01 5.15689626e-02 -9.77410853e-01 9.65648964e-02
1.32296753e+00 -2.34284297e-01 -1.12639576e-01 -8.69355202e-01
-7.11990833e-01 2.00882673e-01 6.73143685e-01 4.99707103e-01
9.24559057e-01 -1.23945534e+00 -1.29589230e-01 -9.73315537e-03
2.48418659e-01 -7.06741035e-01 2.68725485e-01 8.54394674e-01
-1.73723891e-01 4.42715064e-02 -3.61130118e-01 -5.06586470e-02
-1.28100419e+00 3.56613725e-01 7.34889582e-02 3.10378186e-02
1.03928253e-01 6.66548550e-01 -2.66473815e-02 -4.36178356e-01
4.62465808e-02 5.18110394e-01 -7.59570956e-01 5.30783176e-01
4.05624807e-01 2.47803345e-01 -1.07274823e-01 -7.13800073e-01
-6.49412632e-01 4.11949247e-01 4.25575882e-01 -2.72270441e-01
1.16452074e+00 -4.84081060e-01 -8.08929443e-01 1.19446516e+00
7.98267066e-01 2.29059681e-01 3.22507210e-02 9.43028927e-02
5.63212810e-03 -1.68123066e-01 -5.25884563e-03 -9.58284199e-01
-1.11207783e+00 7.99452126e-01 4.95891631e-01 5.44914842e-01
1.52996373e+00 -1.15291290e-01 6.82464182e-01 3.76728952e-01
4.83214140e-01 -1.83952868e+00 3.82467806e-01 8.41722012e-01
5.26883483e-01 -1.10775137e+00 1.57412022e-01 -5.10003805e-01
-1.35833228e+00 1.06309438e+00 8.88813376e-01 1.06372073e-01
1.03427398e+00 4.89252895e-01 6.26715600e-01 -3.80914718e-01
-1.09093285e+00 -3.70157361e-01 -1.68404192e-01 4.21927512e-01
1.33228981e+00 5.03848553e-01 -4.76339877e-01 1.14136910e+00
-3.19189459e-01 1.28056765e-01 5.96886694e-01 1.17331898e+00
-2.97430515e-01 -1.02841747e+00 -1.51571363e-01 4.63421643e-01
-8.19097102e-01 -1.76057771e-01 -9.50020432e-01 2.24582866e-01
9.74210799e-02 1.32403350e+00 6.55343160e-02 -1.00371635e+00
1.23761408e-01 2.13070512e-01 2.23209262e-01 -5.83279788e-01
-1.01862979e+00 -1.68225095e-02 3.29791427e-01 -4.65470344e-01
-1.06078970e+00 -5.88471949e-01 -1.55519080e+00 -6.38748169e-01
-2.52090782e-01 4.83962357e-01 9.16611314e-01 8.30433130e-01
5.11058331e-01 4.31604087e-01 7.27729201e-01 -4.46900606e-01
3.81879449e-01 -8.64718258e-01 -6.78249359e-01 3.38931531e-01
-3.69528055e-01 -3.45917165e-01 -1.53086841e-01 -2.29570102e-02]
|
[13.02918815612793, 5.675216197967529]
|
97015013-d421-42ee-a70a-19189e488cf5
|
scalable-dynamic-topic-modeling-with
|
1610.07703
| null |
https://arxiv.org/abs/1610.07703v3
|
https://arxiv.org/pdf/1610.07703v3.pdf
|
Scalable Dynamic Topic Modeling with Clustered Latent Dirichlet Allocation (CLDA)
|
Topic modeling, a method for extracting the underlying themes from a collection of documents, is an increasingly important component of the design of intelligent systems enabling the sense-making of highly dynamic and diverse streams of text data. Traditional methods such as Dynamic Topic Modeling (DTM) do not lend themselves well to direct parallelization because of dependencies from one time step to another. In this paper, we introduce and empirically analyze Clustered Latent Dirichlet Allocation (CLDA), a method for extracting dynamic latent topics from a collection of documents. Our approach is based on data decomposition in which the data is partitioned into segments, followed by topic modeling on the individual segments. The resulting local models are then combined into a global solution using clustering. The decomposition and resulting parallelization leads to very fast runtime even on very large datasets. Our approach furthermore provides insight into how the composition of topics changes over time and can also be applied using other data partitioning strategies over any discrete features of the data, such as geographic features or classes of users. In this paper CLDA is applied successfully to seventeen years of NIPS conference papers (2,484 documents and 3,280,697 words), seventeen years of computer science journal abstracts (533,560 documents and 32,551,540 words), and to forty years of the PubMed corpus (4,025,978 documents and 273,853,980 words).
|
['Amy W. Apon', 'Alexander Herzog', 'Ilya Safro', 'Paul W. Wilson', 'Chris Gropp']
|
2016-10-25
| null | null | null | null |
['dynamic-topic-modeling']
|
['natural-language-processing']
|
[ 3.37357000e-02 -1.24095544e-01 -1.40672296e-01 -1.32122129e-01
-7.07565844e-01 -5.13124883e-01 8.63114774e-01 8.98614943e-01
-5.23331225e-01 6.08705223e-01 3.25438261e-01 -3.19196045e-01
-3.30831558e-01 -8.91569138e-01 -2.43873015e-01 -7.65001118e-01
-3.17160010e-01 1.03128755e+00 5.64836621e-01 2.39741474e-01
5.18843472e-01 2.00567156e-01 -1.59351099e+00 2.49765038e-01
6.21220112e-01 4.38744813e-01 2.64819950e-01 6.85877979e-01
-7.37914801e-01 -3.79650667e-02 -7.94246018e-01 1.26362979e-01
-3.44562083e-01 -1.87534258e-01 -7.91741252e-01 2.40392596e-01
-2.08738774e-01 7.13138878e-02 2.57363051e-01 7.18448758e-01
3.54471058e-01 1.92982793e-01 8.01790655e-01 -1.30114067e+00
2.92738318e-01 5.03449857e-01 -9.29197431e-01 1.34654626e-01
2.70156980e-01 -5.75917304e-01 8.37553978e-01 -8.85000229e-01
8.17714989e-01 1.21795619e+00 3.10650498e-01 1.46921188e-01
-1.31567395e+00 -4.84379530e-01 2.47915491e-01 2.08136123e-02
-1.24835658e+00 -1.43956795e-01 5.16115963e-01 -7.04542220e-01
8.42807472e-01 3.68866831e-01 6.39277339e-01 7.62714744e-01
2.98483521e-01 6.53727829e-01 7.99272835e-01 -5.51268578e-01
5.80541849e-01 4.16492373e-01 6.06682599e-01 1.18735574e-01
5.53632796e-01 -6.69919312e-01 -5.78988969e-01 -1.06411064e+00
2.16729894e-01 2.55986631e-01 -6.38388321e-02 -2.25026891e-01
-1.28043807e+00 9.75578070e-01 -4.67627585e-01 4.24877465e-01
-4.78393972e-01 -3.58761340e-01 5.37966311e-01 1.69376180e-01
7.62289643e-01 4.86655720e-02 -4.15442616e-01 -3.71356457e-01
-1.22465754e+00 5.19044220e-01 9.96885478e-01 8.81343722e-01
6.74156368e-01 -5.57324886e-01 1.79289997e-01 7.89967418e-01
4.61558908e-01 1.65525019e-01 5.88790536e-01 -5.58182418e-01
3.62352908e-01 6.59184396e-01 2.60553569e-01 -1.06021678e+00
-5.24612308e-01 2.37431660e-01 -6.96314752e-01 -2.41255224e-01
1.14711814e-01 -2.13745028e-01 -9.47370648e-01 1.51389956e+00
5.47842026e-01 -3.31355423e-01 5.45425899e-02 2.78960288e-01
5.66509128e-01 1.08983314e+00 2.45674446e-01 -6.49945319e-01
1.64296341e+00 -4.30408448e-01 -6.42626166e-01 2.41359204e-01
4.54060376e-01 -8.91418099e-01 4.62539196e-01 7.40939736e-01
-1.20958412e+00 -2.28817925e-01 -6.56722784e-01 3.71228270e-02
-6.32572532e-01 -2.26082325e-01 4.08202291e-01 4.68200296e-01
-1.10206294e+00 1.12536848e-01 -1.12431240e+00 -6.57232702e-01
2.51505017e-01 5.16371906e-01 6.31292164e-02 -9.86323878e-02
-9.91614699e-01 3.82864833e-01 3.95542979e-01 -5.16203344e-01
-5.39901257e-01 -7.69974947e-01 -3.58087897e-01 6.14439957e-02
3.29609841e-01 -4.42550778e-01 8.19993138e-01 -3.15379649e-01
-9.73646462e-01 5.58035076e-01 -7.19738066e-01 -2.80975223e-01
1.87458619e-01 1.01293832e-01 -1.79061145e-01 1.83356926e-01
3.41244221e-01 6.39328480e-01 5.18187940e-01 -9.59577322e-01
-8.52754653e-01 -5.85196435e-01 -3.99737418e-01 1.46548644e-01
-5.95792353e-01 2.55013913e-01 -7.40098059e-01 -6.00030303e-01
3.19765896e-01 -8.38290989e-01 -4.72545177e-01 -5.95989108e-01
-3.58880520e-01 -7.11717606e-01 8.99139404e-01 -6.30060434e-01
1.32430708e+00 -2.04673100e+00 3.82925391e-01 4.50070083e-01
4.68513876e-01 -2.10586146e-01 4.97599989e-01 9.75608885e-01
3.82418782e-02 7.82876015e-02 -2.12015778e-01 -5.10454595e-01
-7.46127740e-02 5.64154563e-03 -4.22969431e-01 3.85879397e-01
-3.80674988e-01 2.86230713e-01 -7.46611595e-01 -5.25125563e-01
1.70891043e-02 4.77983177e-01 -3.38042408e-01 -2.22241864e-01
-3.41779858e-01 7.60923699e-02 -5.59265614e-01 2.88997293e-01
4.52356040e-01 -2.43331835e-01 3.11422080e-01 3.68559122e-01
-3.48953724e-01 2.82569885e-01 -1.17078102e+00 1.61320972e+00
-2.27988258e-01 9.29305732e-01 -1.08728461e-01 -9.21688259e-01
9.59088862e-01 7.09641457e-01 1.16328657e+00 -1.27869800e-01
-5.44447973e-02 -4.80823033e-02 -3.01875889e-01 -1.24853119e-01
8.05923581e-01 -1.09050378e-01 -2.77052492e-01 9.47419047e-01
-1.36271566e-01 1.60410151e-01 5.44009745e-01 5.07807374e-01
1.08933794e+00 -6.24216795e-01 2.89829642e-01 -5.31826913e-01
5.28435670e-02 4.98253763e-01 3.76077384e-01 5.36913991e-01
2.40947917e-01 3.62834930e-01 8.03584754e-01 -5.21790743e-01
-1.18959868e+00 -8.28764260e-01 -4.48051035e-01 9.59635139e-01
-1.30153924e-01 -6.80077672e-01 -6.85350955e-01 -3.00340921e-01
-8.25960115e-02 4.65122342e-01 -6.08696938e-01 1.46870181e-01
-2.64715552e-01 -1.22560716e+00 1.18607409e-01 1.02702595e-01
1.92779563e-02 -9.23312545e-01 -8.24573755e-01 6.22864008e-01
-2.46857226e-01 -6.76050127e-01 -1.66792333e-01 1.61459789e-01
-1.10870278e+00 -7.65166402e-01 -9.09481764e-01 -4.36386734e-01
6.90751374e-01 3.82091492e-01 9.47631538e-01 -1.86070710e-01
-3.97739559e-01 3.44144046e-01 -2.42964208e-01 -5.88700294e-01
-2.45257616e-01 2.68936992e-01 2.11836502e-01 -1.83218062e-01
7.52298534e-01 -6.09119356e-01 -3.08767140e-01 2.03355506e-01
-1.14489281e+00 6.36038333e-02 1.79696709e-01 5.29666185e-01
3.96322876e-01 5.06087124e-01 5.29429674e-01 -1.15389812e+00
6.89947605e-01 -9.28821266e-01 -5.85358322e-01 1.34964854e-01
-8.08994174e-01 -2.68724561e-01 1.56427756e-01 -4.29441661e-01
-1.01723886e+00 -2.32414559e-01 2.17913032e-01 -1.60302312e-04
-3.98484230e-01 7.78696239e-01 -1.38921872e-01 7.61054456e-01
4.26060110e-01 2.58555740e-01 -1.45758092e-01 -6.90479875e-01
1.91695556e-01 1.03175676e+00 1.07210129e-01 -5.24830043e-01
2.19979823e-01 8.51377189e-01 -3.01373243e-01 -1.41069794e+00
-1.87188223e-01 -1.18990695e+00 -5.95339298e-01 4.07722630e-02
7.58702517e-01 -8.43266487e-01 -4.32772011e-01 4.02441084e-01
-1.13472593e+00 -1.38581604e-01 -2.25880012e-01 4.52873498e-01
-3.77413243e-01 2.96107858e-01 -4.44240212e-01 -9.10153270e-01
-3.25990528e-01 -8.18022132e-01 1.08511972e+00 -1.67072434e-02
-7.30791211e-01 -1.20230448e+00 3.89611006e-01 1.07218251e-01
2.38783672e-01 1.58521593e-01 1.19709766e+00 -8.71849298e-01
-1.76515967e-01 -3.77319276e-01 6.62023649e-02 -3.74913931e-01
2.58259267e-01 8.98269787e-02 -7.54167020e-01 -4.12845731e-01
8.25644284e-02 3.38719964e-01 8.26667070e-01 5.74296117e-01
8.64627004e-01 -1.75689831e-01 -1.03983307e+00 -3.94473337e-02
1.23829615e+00 7.23125458e-01 6.25880182e-01 3.82407576e-01
4.61562842e-01 9.28354144e-01 5.55176556e-01 8.94941986e-01
4.56210583e-01 5.57975829e-01 -1.28748462e-01 -1.30078271e-01
5.04097462e-01 1.91408202e-01 1.21972717e-01 8.85320306e-01
5.47468700e-02 -4.02431726e-01 -1.30977988e+00 1.06424296e+00
-1.68989742e+00 -7.16155708e-01 -3.77401352e-01 2.33652043e+00
9.46414948e-01 2.71090209e-01 3.86269003e-01 2.26972014e-01
8.00831735e-01 -1.34544879e-01 -3.39503318e-01 -4.04345542e-01
2.54389465e-01 -2.92294715e-02 3.04650337e-01 3.85763168e-01
-9.52788115e-01 6.22034788e-01 5.71120453e+00 8.23359907e-01
-9.17105317e-01 9.17610377e-02 5.67412257e-01 -2.22457543e-01
-3.79528672e-01 8.26153606e-02 -1.17983699e+00 5.95527411e-01
1.40877426e+00 -5.66923082e-01 -1.64927438e-01 6.50991559e-01
4.35295999e-01 -6.46380365e-01 -8.84758472e-01 6.17934585e-01
4.30374853e-02 -1.16012788e+00 5.68116046e-02 6.42619073e-01
9.13530350e-01 -3.88805941e-02 -7.90994987e-03 -1.54249445e-01
4.23922837e-01 -5.86857498e-01 3.35815877e-01 3.07632923e-01
3.72832268e-01 -7.97668755e-01 5.34413695e-01 6.68374181e-01
-7.99081683e-01 1.36995330e-01 -4.76404458e-01 2.09956020e-01
2.61899501e-01 1.13002348e+00 -1.07218981e+00 5.79627216e-01
8.44334662e-01 5.02093375e-01 -7.65283108e-02 1.09246922e+00
5.46780050e-01 8.49299908e-01 -7.47728646e-01 -2.27533668e-01
2.70993531e-01 -6.57528918e-03 6.40545487e-01 1.31361234e+00
2.35839605e-01 6.36049137e-02 6.58552423e-02 5.44641793e-01
2.64127105e-01 2.57493496e-01 -4.99161154e-01 -6.54367507e-02
4.27879095e-01 1.05063319e+00 -1.32597315e+00 -8.02422166e-01
-2.51118749e-01 4.50405091e-01 -1.65752441e-01 5.13359725e-01
-2.92488247e-01 -2.99037606e-01 4.37940747e-01 3.06972116e-01
3.31896931e-01 -4.52106655e-01 -2.71139652e-01 -9.44664240e-01
-1.08648986e-01 -5.76116681e-01 4.82979417e-01 -2.72837967e-01
-1.06892359e+00 4.89664584e-01 5.67279100e-01 -9.99160647e-01
-3.55904967e-01 -7.58709833e-02 -5.00915468e-01 1.01244640e+00
-9.82125878e-01 -6.70868576e-01 7.30303526e-02 3.57949436e-01
8.40756476e-01 -1.49169445e-01 7.20485866e-01 1.44237369e-01
-2.18844935e-01 -1.31646529e-01 6.02466404e-01 -2.46561363e-01
6.42827272e-01 -1.29948235e+00 4.99729663e-01 3.28704238e-01
3.73018049e-02 9.33178246e-01 7.93645442e-01 -7.95160651e-01
-9.67938960e-01 -8.45872998e-01 1.33504164e+00 -4.42693919e-01
5.63654721e-01 -5.39527476e-01 -1.19820607e+00 4.06854600e-01
1.85033411e-01 -7.12459981e-01 9.44505811e-01 3.65249068e-01
1.63774803e-01 1.96502313e-01 -8.07439208e-01 3.21324289e-01
1.59602091e-01 -2.93185152e-02 -4.89308804e-01 3.90272290e-01
6.88186347e-01 -1.75073937e-01 -1.01293516e+00 -6.53861240e-02
4.78331774e-01 -5.95022857e-01 7.12187529e-01 -5.82179785e-01
2.96816945e-01 -2.78007351e-02 2.28683511e-03 -9.56623852e-01
-1.73827223e-02 -6.33039594e-01 2.05990523e-01 1.39994490e+00
3.93197000e-01 -5.36973953e-01 8.90322328e-01 7.03454375e-01
1.13434106e-01 -8.43723714e-01 -8.56082499e-01 -2.59713650e-01
1.50490701e-01 -3.26605260e-01 4.79288489e-01 1.06161952e+00
2.44652942e-01 2.50542015e-01 -4.82131615e-02 -7.78368860e-02
7.17455328e-01 3.07425320e-01 6.69396341e-01 -1.62766314e+00
8.27941671e-03 -3.87381405e-01 -3.32870305e-01 -7.62193263e-01
-2.74980783e-01 -3.60343665e-01 -8.95012468e-02 -1.63833499e+00
4.27679539e-01 -7.50775099e-01 -4.43002805e-02 2.83156693e-01
-3.99007276e-02 -2.19777837e-01 -2.21296906e-01 6.73783541e-01
-5.01095176e-01 2.59471357e-01 4.91717368e-01 -2.39248034e-02
-6.47052765e-01 2.71909237e-01 -5.42992711e-01 7.24058986e-01
6.19868457e-01 -7.15390861e-01 -4.91604567e-01 -9.44599733e-02
1.76971942e-01 1.50010481e-01 -1.60736829e-01 -5.38014770e-01
5.54620147e-01 -9.55891386e-02 2.22977936e-01 -1.18306637e+00
3.03270996e-01 -6.17414534e-01 3.04991066e-01 3.41515720e-01
-3.15952629e-01 4.32290435e-02 3.54134619e-01 7.63872385e-01
-3.32906455e-01 -2.46553749e-01 1.57996446e-01 -1.59761816e-01
-4.38822865e-01 2.02778041e-01 -9.02104735e-01 -1.65259819e-02
1.22511923e+00 -2.26132512e-01 1.74758118e-02 -1.73743680e-01
-1.02072787e+00 4.07531053e-01 3.53502333e-01 3.63031536e-01
5.01375079e-01 -1.00860536e+00 -5.93243122e-01 9.17865708e-02
-1.12875082e-01 4.31002200e-01 2.85706431e-01 7.09689796e-01
-1.68320626e-01 8.50442946e-01 1.50882855e-01 -8.38884234e-01
-1.44562721e+00 5.15966177e-01 -5.13427436e-01 -3.46308291e-01
-6.75252557e-01 3.55121851e-01 2.08249688e-01 1.03489898e-01
3.53717029e-01 -5.24739698e-02 -3.97314429e-01 7.87450731e-01
5.42946577e-01 5.52829683e-01 2.48666197e-01 -3.44365418e-01
-3.81785095e-01 4.23709482e-01 -4.82932359e-01 -7.29792416e-01
1.60889840e+00 -3.32709044e-01 -4.20082986e-01 9.88444507e-01
1.13508070e+00 -2.14259177e-01 -6.84331536e-01 -1.89542085e-01
4.27641839e-01 -2.02645630e-01 -1.11024767e-01 -2.85356253e-01
-4.77122158e-01 9.02572870e-01 1.70528486e-01 6.89101040e-01
1.04813206e+00 3.24014306e-01 6.53688133e-01 3.19495238e-02
3.20605665e-01 -9.13058639e-01 -3.02724719e-01 1.81484669e-01
4.05048490e-01 -8.55773151e-01 1.76550612e-01 -2.47336462e-01
-3.89364809e-01 1.12093520e+00 -7.19162300e-02 2.88758129e-01
9.17515278e-01 2.42378026e-01 -1.36084810e-01 -4.51451659e-01
-1.16205835e+00 2.71554202e-01 2.02201173e-01 2.40366772e-01
3.68271619e-01 -1.94770582e-02 -5.96031845e-01 5.86961567e-01
-1.10619769e-01 -1.46958381e-01 5.56167364e-01 1.19111943e+00
-6.18352532e-01 -1.25919771e+00 -5.90670526e-01 6.04544163e-01
-6.77321851e-01 2.99378075e-02 -2.26116985e-01 6.82134151e-01
-5.94823211e-02 8.04100871e-01 4.02313352e-01 1.41085878e-01
-1.06082581e-01 2.83094823e-01 -1.29823983e-01 -8.79027247e-01
-1.74367458e-01 7.38581598e-01 -2.38958821e-01 3.00529990e-02
-4.25277740e-01 -1.35165966e+00 -1.32371211e+00 -2.05493420e-01
-3.14020932e-01 7.82039106e-01 1.08821023e+00 9.71467972e-01
3.87267768e-01 2.86487699e-01 4.62231606e-01 -6.36211395e-01
1.89249758e-02 -9.99036789e-01 -7.23691702e-01 -8.99274498e-02
1.93971664e-01 -5.83562672e-01 -2.31276289e-01 5.16038120e-01]
|
[10.316554069519043, 7.1036176681518555]
|
27b7ca41-421c-4a4f-a4ff-53954880d0f3
|
a-new-semi-supervised-inductive-transfer
|
2108.07930
| null |
https://arxiv.org/abs/2108.07930v2
|
https://arxiv.org/pdf/2108.07930v2.pdf
|
A new semi-supervised inductive transfer learning framework: Co-Transfer
|
In many practical data mining scenarios, such as network intrusion detection, Twitter spam detection, and computer-aided diagnosis, a source domain that is different from but related to a target domain is very common. In addition, a large amount of unlabeled data is available in both source and target domains, but labeling each of them is difficult, expensive, time-consuming, and sometime unnecessary. Therefore, it is very important and worthwhile to fully explore the labeled and unlabeled data in source and target domains to settle the task in target domain. In this paper, a new semi-supervised inductive transfer learning framework, named Co-Transfer is proposed. Co-Transfer first generates three TrAdaBoost classifiers for transfer learning from the source domain to the target domain, and meanwhile another three TrAdaBoost classifiers are generated for transfer learning from the target domain to the source domain, using bootstraped samples from the original labeled data. In each round of co-transfer, each group of TrAdaBoost classifiers are refined using the carefully labeled data. Finally, the group of TrAdaBoost classifiers learned to transfer from the source domain to the target domain produce the final hypothesis. Experiments results illustrate Co-Transfer can effectively exploit and reuse the labeled and unlabeled data in source and target domains.
|
['Zhe Yuan', 'Yimin Wen']
|
2021-08-18
| null | null | null | null |
['spam-detection']
|
['natural-language-processing']
|
[ 1.41018510e-01 4.04573046e-02 -3.42434853e-01 -5.29571533e-01
-6.10447943e-01 -3.19717944e-01 4.36018556e-01 1.57119602e-01
-3.51413906e-01 1.22452784e+00 -3.06932181e-01 -1.69151917e-01
-2.81046871e-02 -1.11117220e+00 -4.60159779e-01 -5.96173704e-01
2.55909473e-01 8.56610060e-01 6.17747545e-01 -9.29529816e-02
1.05811328e-01 2.82631785e-01 -1.18754363e+00 3.24937969e-01
1.20085704e+00 1.03601992e+00 8.15634504e-02 -5.62418140e-02
-4.22984779e-01 7.28870988e-01 -6.57964647e-01 -2.12354481e-01
1.26743808e-01 -8.14617872e-01 -9.27410245e-01 3.16540331e-01
-4.50360268e-01 -6.34772256e-02 1.54769585e-01 1.08689964e+00
1.21570654e-01 7.57575259e-02 9.61150110e-01 -1.57147360e+00
-6.78318322e-01 3.91444266e-01 -7.16112375e-01 5.59462719e-02
4.00028899e-02 -2.25733981e-01 3.77541214e-01 -9.33383405e-01
4.90319163e-01 1.18169463e+00 5.97851694e-01 4.38508034e-01
-8.76080394e-01 -1.31555355e+00 7.88001195e-02 2.09762365e-01
-1.22098565e+00 -6.27983827e-03 8.42425883e-01 -2.57951766e-01
6.39809296e-02 -2.15698287e-01 5.48206329e-01 1.10086763e+00
-2.51127064e-01 8.57330203e-01 1.43741882e+00 -5.64150095e-01
3.40473711e-01 9.26968038e-01 2.45585278e-01 3.53644460e-01
8.54846388e-02 2.42063254e-01 -2.08249182e-01 -2.98144549e-01
3.66889745e-01 3.89739037e-01 1.60551828e-03 -3.43154252e-01
-8.50044072e-01 1.09807765e+00 5.45303881e-01 5.85047007e-01
-2.53598601e-01 -8.21674109e-01 3.43694806e-01 5.91933608e-01
7.58204877e-01 2.36937746e-01 -7.65445054e-01 1.19518086e-01
-7.02473164e-01 -1.31826282e-01 8.14622939e-01 1.03431690e+00
1.38107526e+00 -3.30559343e-01 3.21422666e-01 8.42927098e-01
3.33408386e-01 3.25926721e-01 9.40847993e-01 -1.46968350e-01
4.28495020e-01 1.10221624e+00 6.12946823e-02 -6.52738154e-01
-1.13661118e-01 -2.58937031e-01 -8.21449101e-01 -7.00960979e-02
4.35392380e-01 -3.99911612e-01 -8.22768986e-01 1.38905823e+00
7.96840549e-01 5.35469890e-01 2.02437639e-01 6.39352500e-01
6.85350120e-01 7.04297781e-01 1.55325308e-01 -4.14814651e-01
9.20488238e-01 -9.58270192e-01 -3.56505752e-01 -3.85831922e-01
6.78627431e-01 -5.81046581e-01 7.29324996e-01 1.57315627e-01
-4.62288171e-01 -6.40384495e-01 -8.59753609e-01 3.16069931e-01
-5.99555790e-01 -1.68763101e-01 2.36875907e-01 4.78169143e-01
-3.91176552e-01 5.15517175e-01 -2.40597516e-01 -4.78868604e-01
6.28956676e-01 3.13126266e-01 -4.44782108e-01 -2.93835133e-01
-1.61861372e+00 7.43795037e-01 9.27874207e-01 -3.54389161e-01
-5.64474106e-01 -3.77008408e-01 -6.16058469e-01 -6.71229362e-02
3.78629863e-01 -7.97427297e-02 1.19728851e+00 -1.53349400e+00
-1.34995031e+00 8.62006366e-01 5.66977262e-02 -3.70094448e-01
5.09715438e-01 2.06690669e-01 -7.04715788e-01 -1.20026074e-01
4.03065026e-01 5.21706820e-01 1.09134018e+00 -1.23299217e+00
-1.07125247e+00 -4.73517299e-01 -3.99084687e-01 7.00131208e-02
-6.27570212e-01 -7.09888786e-02 -2.14237228e-01 -4.18574750e-01
-5.17130233e-02 -7.34070539e-01 -1.26115195e-02 -2.17577577e-01
-2.25332081e-01 -4.87862021e-01 1.47236812e+00 -3.18675518e-01
8.23777258e-01 -2.20051193e+00 -2.35723853e-01 4.46535558e-01
1.49992302e-01 7.60346055e-01 -3.62176672e-02 1.75929517e-01
-2.62061954e-01 -4.06375751e-02 -3.73043418e-01 3.26736182e-01
-5.38930178e-01 2.57341176e-01 -4.29432422e-01 2.06477404e-01
3.32460731e-01 6.72032893e-01 -1.09464884e+00 -6.56339526e-01
1.21558681e-01 -7.42322057e-02 -2.63543814e-01 3.82199645e-01
-1.48276493e-01 1.01358974e+00 -1.06978953e+00 5.22143126e-01
7.24709928e-01 -2.99969912e-01 -3.34992222e-02 1.81767806e-01
2.94019550e-01 9.90106538e-02 -8.93767834e-01 1.02880418e+00
-3.91677260e-01 3.29967588e-01 -3.00004452e-01 -1.58201790e+00
1.41233659e+00 2.01328769e-01 4.79944676e-01 -6.01086259e-01
3.54069829e-01 3.58910352e-01 1.92801189e-03 -4.75602508e-01
4.15491983e-02 -6.50674403e-01 -1.41573429e-01 7.17674911e-01
8.70846435e-02 -2.46023517e-02 -1.15890741e-01 -1.95881873e-01
6.79121315e-01 3.72637017e-03 4.20712531e-01 5.77782802e-02
8.17119360e-01 2.77734846e-01 7.01219559e-01 1.30494773e-01
-3.50238293e-01 2.66647100e-01 2.73380488e-01 -3.92374933e-01
-8.13875020e-01 -8.72013867e-01 -2.40594178e-01 1.19236636e+00
2.60854751e-01 5.98128289e-02 -6.88464284e-01 -1.38663530e+00
1.45021230e-01 6.30302608e-01 -7.68780887e-01 -4.96991664e-01
-2.75801271e-01 -5.85918427e-01 4.30554479e-01 4.39483494e-01
8.81052852e-01 -1.37447715e+00 -1.09028094e-01 3.24056983e-01
-1.30639240e-01 -7.66806841e-01 -1.73197657e-01 2.44306698e-01
-9.69253838e-01 -1.29102576e+00 -7.46384919e-01 -1.20743990e+00
9.41449940e-01 3.38409960e-01 7.94190884e-01 1.90259609e-02
1.10729612e-01 -2.00342804e-01 -7.73395658e-01 -7.54042745e-01
-6.01278901e-01 2.42870171e-02 -1.75013449e-02 3.27146858e-01
1.10578763e+00 -5.06566107e-01 -9.07884091e-02 7.75100470e-01
-8.08765054e-01 -2.30354607e-01 5.51747441e-01 1.10207129e+00
3.98884565e-01 3.89099181e-01 1.18091571e+00 -1.44576681e+00
6.02083683e-01 -1.01815832e+00 -2.60148942e-01 1.11053027e-01
-6.19588375e-01 -2.03987256e-01 8.77570808e-01 -9.85806942e-01
-1.16635120e+00 -1.87546089e-01 2.08105385e-01 -5.32304585e-01
-3.63820791e-01 7.46445537e-01 -3.63728374e-01 3.47933263e-01
1.05384290e+00 1.07502729e-01 3.02972585e-01 -3.78409982e-01
1.61762610e-01 1.27961266e+00 1.85726717e-01 -4.38148856e-01
1.12393081e+00 2.97062457e-01 -5.41753471e-01 -5.80563724e-01
-1.07240272e+00 -4.26904440e-01 -8.15867066e-01 1.91381164e-02
5.38872898e-01 -4.86732453e-01 1.52862919e-02 6.69492960e-01
-7.76497960e-01 -2.63320446e-01 -6.01968288e-01 4.69167858e-01
-1.55095965e-01 2.43484110e-01 -1.13689393e-01 -5.05572617e-01
-1.00501828e-01 -9.65503156e-01 6.12739801e-01 6.38278067e-01
-1.57230899e-01 -1.19108713e+00 6.32109307e-03 2.13295102e-01
1.91363797e-01 1.84116978e-02 9.75464702e-01 -1.57838464e+00
-1.31582171e-02 -7.45526969e-01 -4.61019158e-01 5.83331585e-01
6.55089378e-01 -4.67000365e-01 -8.35286736e-01 -3.72885734e-01
2.80949920e-01 -8.66074085e-01 5.63973069e-01 -4.01899256e-02
1.06936848e+00 -1.08657286e-01 -6.57090545e-01 9.41213369e-02
7.91843891e-01 5.43318868e-01 2.86433607e-01 3.74221921e-01
3.68080735e-01 1.07845974e+00 1.18001366e+00 1.73715100e-01
1.56061232e-01 6.92520440e-02 -1.21493556e-01 -2.24334300e-01
2.41205662e-01 -4.05175745e-01 3.57182562e-01 6.50393903e-01
4.13459778e-01 7.06769340e-03 -1.00413799e+00 4.95909929e-01
-1.70395041e+00 -7.43227720e-01 2.33280927e-01 2.20978379e+00
1.04738665e+00 3.29621792e-01 2.47876331e-01 2.24731922e-01
1.24167669e+00 -4.55926478e-01 -9.78047431e-01 4.11888771e-02
2.54998684e-01 3.34076822e-01 1.13275282e-01 1.53244808e-01
-1.22324324e+00 9.85020757e-01 5.21925259e+00 1.00375426e+00
-1.34849501e+00 1.38063848e-01 7.19552279e-01 4.61099505e-01
3.46202636e-04 1.66585192e-01 -7.13589609e-01 6.97810173e-01
1.00255835e+00 -5.97052634e-01 9.57283974e-02 1.17259848e+00
-1.86443374e-01 -5.50636090e-02 -9.93299782e-01 7.08010972e-01
-3.05730272e-02 -8.21170688e-01 -6.39047548e-02 1.00775711e-01
6.96125031e-01 -1.49520487e-01 -1.24460049e-01 8.02206933e-01
5.51900506e-01 -8.19353938e-01 1.01226434e-01 -2.39358526e-02
8.68691385e-01 -9.73977923e-01 9.72220778e-01 9.05902684e-01
-9.82191503e-01 -1.58930540e-01 -4.38200474e-01 1.28720015e-01
-2.10453227e-01 5.69688976e-01 -1.20187056e+00 5.49552500e-01
6.91382706e-01 9.59187329e-01 -3.77019644e-01 8.56754839e-01
-3.74108166e-01 7.14238226e-01 -4.81066108e-02 -6.15589917e-02
1.67896315e-01 -4.44220752e-01 5.00344336e-02 7.58487940e-01
3.73909026e-01 8.93456265e-02 6.31539524e-01 5.73579550e-01
-2.01452985e-01 2.59655207e-01 -7.54226267e-01 -6.54058233e-02
6.62924170e-01 1.08558643e+00 -6.13821268e-01 -6.05747879e-01
-3.86546344e-01 7.90156960e-01 3.68257046e-01 2.45477259e-01
-7.53546894e-01 -6.37226820e-01 -1.02248471e-02 2.49013856e-01
1.38739288e-01 5.17308116e-01 -9.27578807e-02 -1.03726745e+00
-2.69537568e-01 -8.11852753e-01 5.79526842e-01 -4.00493801e-01
-1.97495282e+00 6.32795870e-01 6.32489622e-02 -1.77950656e+00
-4.31672841e-01 -4.66060102e-01 -8.09094906e-01 1.03080094e+00
-1.54917133e+00 -1.13363004e+00 -3.76420617e-01 9.76864278e-01
4.35299903e-01 -6.21950626e-01 8.07780385e-01 2.68017858e-01
-3.22326481e-01 5.19331276e-01 2.51411170e-01 4.67833489e-01
1.04388034e+00 -8.67823005e-01 7.08981603e-02 3.83750230e-01
-2.74413317e-01 5.39503038e-01 1.95843548e-01 -6.81361318e-01
-5.68614364e-01 -1.51970553e+00 6.16451502e-01 -3.14744413e-02
7.23153830e-01 -5.27454317e-02 -1.47222638e+00 8.41593504e-01
-7.39508048e-02 1.58958092e-01 1.12011158e+00 1.21712185e-01
-3.29941213e-01 -5.35873584e-02 -1.51647687e+00 3.13636094e-01
4.04803723e-01 -4.19518769e-01 -1.08680022e+00 5.59448957e-01
3.80174130e-01 -2.19903037e-01 -5.46220005e-01 3.75562847e-01
1.74374193e-01 -7.49672353e-01 5.65480411e-01 -8.88648808e-01
3.54055524e-01 -1.66593939e-01 2.26990014e-01 -1.58101082e+00
-3.79256271e-02 -1.51819527e-01 2.61029512e-01 1.40928757e+00
4.09196794e-01 -1.01551747e+00 9.81417179e-01 3.25801373e-01
1.07311897e-01 -4.73727196e-01 -8.43424201e-01 -8.38108659e-01
5.29873312e-01 -1.98235676e-01 6.33717060e-01 1.43237329e+00
1.53573632e-01 5.82350492e-01 -1.37190178e-01 -1.26573771e-01
4.23064440e-01 4.80954289e-01 8.89638901e-01 -1.78081942e+00
-6.27341941e-02 5.50954081e-02 -1.54995367e-01 -9.93564427e-01
4.26928490e-01 -1.15473509e+00 6.61350042e-02 -1.19210756e+00
1.80820629e-01 -1.01916218e+00 -2.74370104e-01 7.36544907e-01
-3.21268380e-01 1.96448252e-01 -1.85483634e-01 5.24972975e-01
-3.12189490e-01 7.19004691e-01 1.34391403e+00 -1.32692575e-01
-5.24053693e-01 4.16781694e-01 -7.96035826e-01 9.86654043e-01
8.31647635e-01 -6.36531234e-01 -5.65932095e-01 1.95485070e-01
-5.66602767e-01 1.20634481e-01 -8.43194276e-02 -7.24522173e-01
9.79515612e-02 -2.83474654e-01 5.90494990e-01 -4.43813086e-01
1.28244236e-01 -1.00681531e+00 -3.76579940e-01 4.80880857e-01
-1.48144364e-01 -5.95934987e-01 6.08773679e-02 6.89484477e-01
-5.21200895e-01 -4.42625523e-01 1.11690104e+00 -3.82850841e-02
-8.49247575e-01 3.88651401e-01 -2.48343885e-01 3.29956353e-01
1.62213278e+00 -4.24603134e-01 -2.53637612e-01 -3.08381140e-01
-8.56943309e-01 4.52951252e-01 3.19663376e-01 5.15213013e-01
6.25800908e-01 -1.36646307e+00 -6.48828089e-01 5.31905293e-01
2.72570223e-01 2.24210471e-01 1.39840886e-01 5.23322999e-01
-2.02736501e-02 3.22326832e-02 -4.98053014e-01 -6.20635152e-01
-9.48361218e-01 8.63739312e-01 1.07205935e-01 -3.83738786e-01
-2.56166279e-01 6.56149387e-01 2.35563561e-01 -1.03064239e+00
-3.30944397e-02 2.05095381e-01 -4.43915814e-01 3.12855273e-01
4.79637206e-01 2.54159182e-01 -7.24745616e-02 -6.06965184e-01
-9.61844400e-02 2.79209405e-01 -4.76288080e-01 1.00960217e-01
1.16212225e+00 8.74130130e-02 -1.80977598e-01 3.86029035e-01
1.35713863e+00 -1.53406948e-01 -9.06912684e-01 -7.47056723e-01
2.72609919e-01 -3.48043352e-01 -4.32451904e-01 -6.92950964e-01
-8.89644325e-01 8.04832101e-01 4.37055081e-01 2.37941816e-01
1.29849362e+00 4.72351313e-02 7.81087637e-01 2.87180662e-01
6.50382519e-01 -9.95199442e-01 2.85238326e-01 5.77036858e-01
4.73461866e-01 -1.36426210e+00 -3.07501942e-01 -5.66730857e-01
-8.42856407e-01 8.96127939e-01 9.96475279e-01 -2.33407512e-01
9.82134581e-01 -2.97237299e-02 1.28999397e-01 1.81581423e-01
-3.87738526e-01 1.83772091e-02 8.24921429e-02 9.14999247e-01
1.92993432e-01 -1.40902504e-01 -1.16176307e-01 6.98905230e-01
1.67581346e-02 3.02038312e-01 9.92737412e-02 1.16777897e+00
-6.54207766e-01 -1.57879663e+00 -6.56250596e-01 7.91453838e-01
-8.84315223e-02 1.74915925e-01 -5.50808609e-01 8.13728690e-01
4.02003139e-01 1.09904301e+00 1.50491118e-01 -5.94965756e-01
2.55423754e-01 3.02228987e-01 -3.82204838e-02 -9.98660922e-01
-4.86212045e-01 -2.53472906e-02 -2.38303229e-01 1.80183321e-01
-4.37488675e-01 -3.41235906e-01 -1.40938747e+00 -4.82982621e-02
-5.36763191e-01 7.91314483e-01 1.39206260e-01 1.14506483e+00
1.47077456e-01 1.71685830e-01 1.10041440e+00 -5.07825673e-01
-5.45947254e-01 -1.26441622e+00 -7.75004148e-01 5.22995055e-01
8.09401870e-02 -8.53913546e-01 -2.96569347e-01 -3.08493227e-02]
|
[10.353095054626465, 3.1385281085968018]
|
bf79e989-06dd-40de-b17e-9ad10de02a14
|
a-unified-image-preprocessing-framework-for
|
2208.07110
| null |
https://arxiv.org/abs/2208.07110v1
|
https://arxiv.org/pdf/2208.07110v1.pdf
|
A Unified Image Preprocessing Framework For Image Compression
|
With the development of streaming media technology, increasing communication relies on sound and visual information, which puts a massive burden on online media. Data compression becomes increasingly important to reduce the volume of data transmission and storage. To further improve the efficiency of image compression, researchers utilize various image processing methods to compensate for the limitations of conventional codecs and advanced learning-based compression methods. Instead of modifying the image compression oriented approaches, we propose a unified image compression preprocessing framework, called Kuchen, which aims to further improve the performance of existing codecs. The framework consists of a hybrid data labeling system along with a learning-based backbone to simulate personalized preprocessing. As far as we know, this is the first exploration of setting a unified preprocessing benchmark in image compression tasks. Results demonstrate that the modern codecs optimized by our unified preprocessing framework constantly improve the efficiency of the state-of-the-art compression.
|
['Xiaocheng Li', 'Weihui Deng', 'Moqi Zhang']
|
2022-08-15
| null | null | null | null |
['data-compression']
|
['time-series']
|
[ 4.73752201e-01 -2.63864338e-01 -4.39960659e-01 -2.40901813e-01
-5.67978978e-01 4.93428446e-02 2.74065703e-01 3.78151238e-01
-5.47188878e-01 1.63106725e-01 2.87885278e-01 -3.35659862e-01
3.01158167e-02 -1.06166553e+00 -7.89024353e-01 -4.48664337e-01
-7.93371648e-02 1.90677688e-01 4.14832741e-01 -1.62488490e-01
3.43120784e-01 1.69512257e-01 -1.79178631e+00 6.17187560e-01
8.39181364e-01 1.18192458e+00 6.26351774e-01 5.86673439e-01
-4.54835206e-01 7.87321210e-01 -3.30267608e-01 -6.51425302e-01
3.07763934e-01 -4.52675462e-01 -8.43658507e-01 1.42526343e-01
1.51026770e-01 -6.94588900e-01 -7.47075558e-01 1.07872033e+00
3.77268225e-01 -1.54436752e-01 2.13656574e-01 -9.18525159e-01
-5.21744609e-01 1.07151604e+00 -6.60111070e-01 2.02762447e-02
2.06116870e-01 5.62273227e-02 7.47137129e-01 -4.25965250e-01
5.50720453e-01 8.74817848e-01 4.83206421e-01 4.23648059e-01
-8.41548324e-01 -4.69741553e-01 -1.62819937e-01 6.39739931e-01
-1.41901207e+00 -5.21926522e-01 5.76073945e-01 8.42027217e-02
6.20897174e-01 2.83652782e-01 8.09359074e-01 5.41521072e-01
-1.71705738e-01 8.15504432e-01 6.32656157e-01 -6.10116124e-01
2.06483841e-01 -1.62458688e-01 -8.11001137e-02 7.26384044e-01
2.98145264e-01 -3.01321656e-01 -5.45277834e-01 3.22892934e-01
4.33347315e-01 2.31461331e-01 -3.78157735e-01 -2.07194522e-01
-9.94017541e-01 5.08886456e-01 3.57482880e-01 2.79986531e-01
-5.10143265e-02 3.27339679e-01 6.33094132e-01 2.70318151e-01
4.32478577e-01 -2.23472640e-02 -2.89143056e-01 -4.08563644e-01
-1.36023021e+00 2.17434019e-01 6.19346619e-01 1.00598311e+00
7.59655237e-01 -3.09098095e-01 -2.05457091e-01 9.06114817e-01
2.09408209e-01 3.19403380e-01 7.02160895e-01 -1.25757170e+00
7.30139911e-01 7.44302034e-01 -5.15473068e-01 -9.49912846e-01
-3.82279009e-02 -5.29250264e-01 -9.69818950e-01 -1.77706733e-01
6.46276772e-02 3.92058730e-01 -6.14651263e-01 1.36231124e+00
8.15247223e-02 2.23587796e-01 -1.49344817e-01 6.61869109e-01
6.11081243e-01 9.35139775e-01 -3.20608392e-02 -4.62062150e-01
1.24273849e+00 -1.16021562e+00 -7.59147823e-01 1.95633471e-01
8.38425636e-01 -7.08872437e-01 1.37861061e+00 6.46837115e-01
-1.42694259e+00 -5.91031611e-01 -1.33175123e+00 -3.58355701e-01
-1.34423390e-01 -2.27451086e-01 6.74270689e-01 7.25462317e-01
-7.81563163e-01 7.41630971e-01 -9.95942354e-01 -6.31510764e-02
7.58420467e-01 -5.47678843e-02 -1.95063740e-01 -4.98195857e-01
-6.98734462e-01 3.16492409e-01 6.11912787e-01 -5.02358019e-01
-6.03526235e-01 -1.01850224e+00 -6.59348965e-01 4.69324648e-01
2.62590230e-01 -4.66333508e-01 1.16663432e+00 -7.60050356e-01
-1.45431387e+00 6.68208003e-01 3.41663108e-04 -8.05450976e-01
3.93118083e-01 -2.53017008e-01 -2.06645429e-01 4.94952232e-01
-4.35785681e-01 6.80518568e-01 7.46533692e-01 -1.18336606e+00
-6.55515373e-01 -5.78332916e-02 -1.59359783e-01 3.67756821e-02
-9.45775747e-01 -1.12297811e-01 -1.11964726e+00 -5.82677603e-01
-4.17396910e-02 -5.29052138e-01 -2.08527565e-01 1.83878735e-01
-8.28927606e-02 3.37274047e-03 1.09499037e+00 -5.21050692e-01
1.75623977e+00 -2.46496725e+00 1.21307753e-01 4.73746248e-02
3.26010078e-01 4.94441986e-01 -2.81428069e-01 4.50647324e-01
1.71392262e-01 1.36525258e-01 -4.82255757e-01 -6.96384549e-01
-2.98153870e-02 2.52887338e-01 -4.60758090e-01 2.09566221e-01
-3.30228478e-01 6.86638057e-01 -7.13950455e-01 -8.33378911e-01
7.37572536e-02 4.60487276e-01 -1.06710291e+00 2.69789428e-01
-3.53371769e-01 5.39765991e-02 -1.76497161e-01 4.70624298e-01
8.52127016e-01 -4.89333689e-01 1.17119018e-03 -3.28459203e-01
-1.03985406e-01 1.95972025e-01 -9.05883610e-01 2.23724842e+00
-4.60877687e-01 3.73833537e-01 -1.22790612e-01 -1.38060629e+00
4.42150623e-01 6.82360083e-02 6.89955652e-01 -8.90403986e-01
1.44446075e-01 7.13855922e-02 -2.25173488e-01 -5.90209126e-01
5.70520759e-01 4.32249278e-01 1.55190870e-01 4.43020046e-01
-1.44601958e-02 -1.61704466e-01 6.01407707e-01 5.04267991e-01
1.20992279e+00 5.38012721e-02 9.69572067e-02 1.64505586e-01
4.96565402e-01 -1.38477296e-01 2.89946437e-01 3.90418887e-01
-4.49083075e-02 6.88475549e-01 3.76079738e-01 -4.15895492e-01
-1.34064770e+00 -9.69372213e-01 -2.61478156e-01 1.04839921e+00
1.67559236e-01 -1.15008783e+00 -1.19537568e+00 -3.07877719e-01
-2.78637409e-01 4.25227523e-01 -2.47486327e-02 -1.80934712e-01
-4.70102698e-01 -6.99974954e-01 5.63444793e-01 1.00663707e-01
9.89841044e-01 -7.10889876e-01 -8.40330482e-01 5.88307530e-03
-4.66593087e-01 -1.27694392e+00 -4.45998251e-01 -2.02857181e-01
-1.05802107e+00 -8.50129724e-01 -5.72567642e-01 -5.38552105e-01
3.90453756e-01 6.09016836e-01 9.79157448e-01 6.43676400e-01
-2.72750020e-01 2.83636570e-01 -8.11894059e-01 -3.47710490e-01
-5.07026017e-01 3.85687768e-01 -3.35422993e-01 -1.60320371e-01
-7.17706699e-03 -8.80540311e-01 -8.76247644e-01 -1.13984689e-01
-1.44024169e+00 4.65619236e-01 5.95834792e-01 3.92205656e-01
7.49428570e-01 3.58574003e-01 2.00465903e-01 -8.13027084e-01
3.59181851e-01 -3.92747283e-01 -4.68396425e-01 1.93344280e-01
-8.79273415e-01 1.18450738e-01 7.44345784e-01 -1.99454948e-01
-8.28702271e-01 -1.89772453e-02 -1.99612468e-01 -3.73138964e-01
7.55073652e-02 7.09657550e-01 -4.59035784e-02 -7.59301633e-02
4.40527797e-01 4.67253268e-01 1.69045329e-02 -6.14161849e-01
4.29726869e-01 8.50036442e-01 7.59147823e-01 -4.70367700e-01
6.72066092e-01 4.40309644e-01 1.16186738e-01 -7.06421435e-01
-6.71603203e-01 -2.16928020e-01 -3.42224300e-01 -2.00084880e-01
6.12776518e-01 -9.15311933e-01 -6.17672622e-01 4.01101053e-01
-1.04680967e+00 -2.63982952e-01 -2.26701826e-01 2.79365003e-01
-5.00838935e-01 8.96333814e-01 -8.74401331e-01 -2.92663753e-01
-3.07502538e-01 -1.34473324e+00 8.22142184e-01 -4.69835475e-03
2.79511303e-01 -5.19121468e-01 -4.97006774e-02 4.76898879e-01
6.99450254e-01 -2.58870661e-01 1.10687423e+00 -1.29760727e-01
-9.62867081e-01 -9.44071785e-02 -3.99548382e-01 4.47517693e-01
-2.37032786e-01 -1.00077756e-01 -7.45788932e-01 -2.65112907e-01
-2.90288292e-02 -3.56473505e-01 1.17103601e+00 1.76118135e-01
2.15308094e+00 -4.20364588e-01 -1.61021367e-01 1.25386953e+00
1.58977723e+00 1.42988842e-02 1.10508275e+00 3.15487891e-01
5.34423947e-01 3.78756166e-01 1.43736050e-01 7.83963680e-01
4.90443259e-01 6.28541768e-01 6.04597569e-01 1.89457878e-01
-3.52104127e-01 -3.98904592e-01 1.82688221e-01 1.47098053e+00
-1.44126385e-01 -3.05068940e-01 -7.52483904e-01 2.59361029e-01
-1.55545378e+00 -9.68176126e-01 1.42580375e-01 2.20198035e+00
1.18104267e+00 3.56936827e-02 -1.42961353e-01 6.30755663e-01
4.07199353e-01 2.46331751e-01 -2.06686229e-01 -1.90411404e-01
5.68036400e-02 4.19022620e-01 6.38139486e-01 6.25795051e-02
-9.77404118e-01 6.22363150e-01 6.49516869e+00 1.27862549e+00
-1.02359092e+00 1.75358713e-01 6.30893469e-01 -1.48460627e-01
-3.48966926e-01 -4.98710945e-02 -6.30776107e-01 6.90612614e-01
1.28746021e+00 -3.51907402e-01 8.84460807e-01 1.04891086e+00
7.70564005e-02 -8.75290036e-02 -9.66805756e-01 1.44362402e+00
9.90699232e-02 -1.71471250e+00 3.10217708e-01 1.17972925e-01
5.81064165e-01 -5.05256020e-02 1.51266053e-01 1.13758616e-01
-1.72530010e-01 -8.15399945e-01 6.81925774e-01 3.56585860e-01
1.04316795e+00 -6.56198502e-01 5.82033515e-01 3.73409361e-01
-1.11416578e+00 -9.13581774e-02 -4.81786937e-01 1.28629310e-02
3.56594205e-01 8.82742643e-01 -1.66203871e-01 4.63516414e-01
8.71065199e-01 7.59380937e-01 -6.46689236e-01 1.30107224e+00
6.56477660e-02 8.34439397e-01 -2.10896105e-01 4.79855418e-01
-4.26882617e-02 -3.75132300e-02 4.79426757e-02 1.32839310e+00
5.87106645e-01 1.66307509e-01 8.90898407e-02 3.81348431e-01
-5.82245231e-01 3.87144089e-01 -3.51288766e-01 -1.66194901e-01
5.42692065e-01 1.16887856e+00 -7.57944047e-01 -4.51095670e-01
-6.44628227e-01 9.94855940e-01 2.31474027e-01 -1.98645033e-02
-9.03931081e-01 -3.44466150e-01 2.87109315e-01 2.35948816e-01
2.97825843e-01 -4.56380725e-01 -3.28126460e-01 -1.17484617e+00
8.04142207e-02 -9.47831035e-01 1.99595913e-01 -5.14560342e-01
-8.51900399e-01 2.80071169e-01 1.61354065e-01 -1.34310341e+00
2.16752350e-01 -2.41596133e-01 -2.46116295e-01 1.44604474e-01
-1.64254594e+00 -8.25906515e-01 -6.55273199e-01 6.09996557e-01
6.34844124e-01 -2.03719333e-01 6.57649577e-01 8.61182332e-01
-4.20474380e-01 7.28171945e-01 9.91867036e-02 -1.70377284e-01
7.19606280e-01 -6.55457616e-01 9.29628089e-02 9.04138327e-01
2.10523918e-01 3.58242631e-01 3.74098629e-01 -2.70865768e-01
-1.75844622e+00 -1.09186721e+00 5.68532884e-01 2.81641483e-01
3.47525001e-01 -2.90663600e-01 -1.08159649e+00 2.86388099e-01
3.02860379e-01 1.20393343e-01 6.96400046e-01 -3.82083505e-01
-4.93225485e-01 -5.97783506e-01 -9.00345683e-01 4.69610184e-01
1.27932096e+00 -4.67556506e-01 -1.50745884e-01 4.06199276e-01
1.25709403e+00 -3.80471110e-01 -7.57939994e-01 3.93626571e-01
3.62343162e-01 -1.12842727e+00 1.02748442e+00 -1.64614514e-01
1.15355349e+00 -9.13394541e-02 -3.77978563e-01 -8.21451545e-01
-2.59119719e-01 -4.23644751e-01 -4.98555660e-01 1.23125029e+00
7.23320395e-02 -1.39154091e-01 8.94151926e-01 2.59136826e-01
-3.67913067e-01 -7.93889940e-01 -8.16639006e-01 -5.78870714e-01
-2.87753552e-01 -8.31382215e-01 6.95873082e-01 6.12673044e-01
1.71107292e-01 -1.44799501e-01 -3.57156754e-01 -1.69883922e-01
6.98674977e-01 -7.27096796e-02 7.55700350e-01 -8.72703373e-01
-6.17200315e-01 -5.75648487e-01 -3.67103994e-01 -1.41407466e+00
-1.24479443e-01 -1.02890372e+00 -2.17961565e-01 -1.38776970e+00
5.10738313e-01 -4.63329732e-01 -2.34519616e-01 3.15028727e-01
6.52099624e-02 4.95323181e-01 4.60823983e-01 5.06766438e-01
-8.28468561e-01 6.83682382e-01 1.24214864e+00 -1.38762638e-01
1.31481290e-02 -3.09336185e-01 -6.20569527e-01 5.42455316e-01
6.25204325e-01 -4.15109605e-01 -7.03176081e-01 -9.37052727e-01
5.43646395e-01 3.96571569e-02 2.99150124e-02 -1.41488206e+00
5.59156537e-01 -8.50136299e-03 -2.60504987e-02 -5.17088890e-01
-2.43964139e-02 -9.26383317e-01 5.31698130e-02 5.94099760e-01
-4.57316935e-01 6.24137893e-02 -9.34383422e-02 4.28803027e-01
-4.68863934e-01 -3.07059318e-01 9.48409021e-01 -8.47684518e-02
-6.69975400e-01 6.18069768e-01 -3.40572000e-02 -1.37996525e-01
9.31201518e-01 2.49473262e-03 -4.23587382e-01 -3.72077405e-01
-5.43305762e-02 1.65431947e-02 5.44262707e-01 1.82239994e-01
7.73332059e-01 -1.19200075e+00 -6.05401993e-01 2.29427949e-01
-1.81715805e-02 2.90428214e-02 3.15355569e-01 6.74555898e-01
-1.00374234e+00 2.56565213e-01 -1.92146212e-01 -3.94918442e-01
-1.22315812e+00 8.29134464e-01 -1.63713977e-01 -3.32981646e-01
-7.96559036e-01 6.04224503e-01 -1.14128456e-01 2.17477962e-01
4.83998179e-01 -2.21842408e-01 -7.82719105e-02 -2.53005862e-01
9.09711540e-01 5.01880050e-01 1.96290627e-01 -1.80608884e-01
1.51653081e-01 4.87470925e-01 -8.04736167e-02 3.76960747e-02
1.42221236e+00 -4.47028428e-01 -3.68650049e-01 8.32299143e-02
1.28817677e+00 -1.69995893e-02 -1.09901035e+00 -2.73644954e-01
-2.70126849e-01 -7.64811695e-01 3.30383509e-01 -3.63959163e-01
-1.45637381e+00 9.74301219e-01 6.40458107e-01 2.02502757e-01
1.71811628e+00 -1.23625860e-01 1.35590792e+00 3.69856387e-01
5.16884387e-01 -1.08225584e+00 9.67020914e-02 3.17223012e-01
5.91344595e-01 -9.02644932e-01 3.29364955e-01 -4.40287143e-01
-2.24512815e-01 1.22778749e+00 1.88304812e-01 7.84050003e-02
7.60740340e-01 6.30306602e-01 -3.73926044e-01 2.20111132e-01
-8.09713423e-01 1.95179991e-02 6.64822161e-02 3.78779471e-01
4.69628394e-01 -2.06541866e-01 -7.02470660e-01 5.55700779e-01
-2.34480068e-01 3.26578259e-01 3.51727396e-01 7.81637073e-01
-6.97232127e-01 -1.54496396e+00 -5.44561855e-02 4.47248846e-01
-6.28274381e-01 -2.13511199e-01 5.06724775e-01 3.53974283e-01
3.31740826e-01 7.23271608e-01 1.46512032e-01 -5.68254828e-01
1.38222530e-01 -1.47075906e-01 4.63215292e-01 -3.44209373e-01
-2.25504115e-01 -7.32957199e-02 -3.45639288e-01 -8.29074800e-01
-6.50730312e-01 -3.03208172e-01 -1.21768415e+00 -7.18842506e-01
1.08452871e-01 -1.01589657e-01 8.89475167e-01 8.11894238e-01
4.72754210e-01 4.44725454e-01 5.38195491e-01 -8.32894385e-01
-4.45738137e-01 -6.22182012e-01 -3.01931560e-01 6.47943139e-01
-4.50080121e-03 -1.19043857e-01 -1.02416985e-01 6.33653283e-01]
|
[11.374841690063477, -1.5568381547927856]
|
d1773c71-3b45-44e0-be46-1e8db0ece881
|
rspnet-relative-speed-perception-for
|
2011.07949
| null |
https://arxiv.org/abs/2011.07949v2
|
https://arxiv.org/pdf/2011.07949v2.pdf
|
RSPNet: Relative Speed Perception for Unsupervised Video Representation Learning
|
We study unsupervised video representation learning that seeks to learn both motion and appearance features from unlabeled video only, which can be reused for downstream tasks such as action recognition. This task, however, is extremely challenging due to 1) the highly complex spatial-temporal information in videos; and 2) the lack of labeled data for training. Unlike the representation learning for static images, it is difficult to construct a suitable self-supervised task to well model both motion and appearance features. More recently, several attempts have been made to learn video representation through video playback speed prediction. However, it is non-trivial to obtain precise speed labels for the videos. More critically, the learnt models may tend to focus on motion pattern and thus may not learn appearance features well. In this paper, we observe that the relative playback speed is more consistent with motion pattern, and thus provide more effective and stable supervision for representation learning. Therefore, we propose a new way to perceive the playback speed and exploit the relative speed between two video clips as labels. In this way, we are able to well perceive speed and learn better motion features. Moreover, to ensure the learning of appearance features, we further propose an appearance-focused task, where we enforce the model to perceive the appearance difference between two video clips. We show that optimizing the two tasks jointly consistently improves the performance on two downstream tasks, namely action recognition and video retrieval. Remarkably, for action recognition on UCF101 dataset, we achieve 93.7% accuracy without the use of labeled data for pre-training, which outperforms the ImageNet supervised pre-trained model. Code and pre-trained models can be found at https://github.com/PeihaoChen/RSPNet.
|
['Chuang Gan', 'Mingkui Tan', 'Shilei Wen', 'Runhao Zeng', 'Xiang Long', 'Dongliang He', 'Deng Huang', 'Peihao Chen']
|
2020-10-27
| null | null | null | null |
['self-supervised-action-recognition']
|
['computer-vision']
|
[ 2.04493999e-01 -4.19410974e-01 -6.00114048e-01 -5.02227724e-01
-5.89056492e-01 -5.23992181e-01 3.69502515e-01 -6.06260858e-02
-4.04989928e-01 2.84463078e-01 -1.72768049e-02 -9.44160968e-02
-9.77867050e-04 -4.82606322e-01 -8.15116465e-01 -8.66708815e-01
-1.06580064e-01 9.36982222e-03 2.01346099e-01 2.36444753e-02
5.97942509e-02 3.33511561e-01 -1.71368396e+00 2.10213244e-01
5.96248984e-01 1.18724561e+00 3.58948976e-01 6.93581045e-01
2.51372337e-01 1.09464085e+00 -2.98453510e-01 1.46310717e-01
2.74990946e-01 -4.67938781e-01 -7.34270632e-01 4.15487021e-01
6.31234944e-01 -5.94300270e-01 -7.51592636e-01 9.19317424e-01
9.97698084e-02 3.60602647e-01 6.87212586e-01 -1.29847634e+00
-4.99038190e-01 1.45524979e-01 -6.86246097e-01 4.08417970e-01
2.96189815e-01 2.01924026e-01 8.70269060e-01 -5.80654442e-01
5.18925428e-01 9.28152084e-01 2.73600221e-01 6.04705572e-01
-8.88411939e-01 -6.63800120e-01 3.96624088e-01 6.58018172e-01
-1.28194702e+00 -5.78589618e-01 8.14234138e-01 -5.37186384e-01
5.08218825e-01 2.13279247e-01 7.43422925e-01 1.25422096e+00
-1.38677269e-01 1.09045720e+00 7.12705374e-01 -3.37954462e-01
7.53891543e-02 -7.96948895e-02 -1.26885280e-01 7.29477942e-01
-1.81511700e-01 1.05163716e-01 -4.64020818e-01 3.45247298e-01
1.00861049e+00 4.57484752e-01 -5.29128611e-01 -4.49271649e-01
-1.21138549e+00 4.65649158e-01 4.88707274e-01 3.95154357e-01
-1.61197707e-01 3.73811066e-01 3.59156609e-01 4.51562107e-01
3.57105732e-01 1.06632590e-01 -5.34315646e-01 -5.19236147e-01
-8.49417925e-01 -1.42204419e-01 2.88817972e-01 8.57460797e-01
9.23696697e-01 3.09422966e-02 -5.68375252e-02 7.44941294e-01
2.39857003e-01 5.71173429e-01 6.47332311e-01 -1.13482428e+00
4.93216544e-01 4.23052102e-01 3.27699743e-02 -1.21340740e+00
-2.71245360e-01 7.33434036e-02 -8.81800592e-01 -2.68772859e-02
7.25560904e-01 6.26200736e-02 -9.77512896e-01 1.82360876e+00
7.80633688e-02 5.24243057e-01 -7.64178336e-02 1.23452389e+00
5.08016586e-01 7.37615168e-01 1.12584203e-01 -4.52450305e-01
1.17021072e+00 -1.13761222e+00 -6.22456968e-01 -1.47468328e-01
8.00718069e-01 -5.13114750e-01 9.95949030e-01 3.93597811e-01
-7.98668087e-01 -8.66685152e-01 -9.77151871e-01 6.34209660e-04
-1.64942905e-01 3.88223916e-01 7.78165281e-01 2.90686160e-01
-8.26965153e-01 6.81346297e-01 -1.15687549e+00 -2.98509598e-01
2.82147408e-01 3.70544255e-01 -5.46209633e-01 -3.44003439e-01
-1.04119444e+00 5.56710243e-01 1.84780821e-01 1.03500128e-01
-9.44427073e-01 -4.46749598e-01 -9.33346331e-01 -7.50524923e-02
4.69774038e-01 -3.83594424e-01 1.16876090e+00 -1.62076187e+00
-1.60667408e+00 7.38321245e-01 -9.85711887e-02 -1.28558293e-01
5.85639656e-01 -1.45697221e-01 -3.13028455e-01 4.92308170e-01
-1.06543809e-01 6.49741590e-01 1.17687953e+00 -1.06929195e+00
-7.03365326e-01 -3.36506724e-01 3.77401382e-01 1.30947158e-01
-7.28453398e-01 -2.38400027e-01 -9.11998034e-01 -7.12317884e-01
-7.55089056e-03 -1.03095245e+00 -8.56439322e-02 1.81947142e-01
1.84766669e-02 -1.96099296e-01 9.72824633e-01 -6.01926804e-01
1.12913406e+00 -2.32305121e+00 2.60258645e-01 8.44179932e-03
1.37625799e-01 4.10722733e-01 -3.58058333e-01 1.77827552e-02
-1.55345351e-01 -5.03310189e-02 -6.25237226e-02 -3.09424073e-01
-3.23418409e-01 3.00759912e-01 -1.41154438e-01 6.14424825e-01
2.07949147e-01 8.78482640e-01 -1.16440117e+00 -4.48711902e-01
3.88957888e-01 4.30626124e-01 -4.46814507e-01 4.08667386e-01
-1.92528680e-01 7.50289977e-01 -6.37548685e-01 6.28095031e-01
3.85259062e-01 -3.54997426e-01 2.08123505e-01 -3.01829398e-01
1.38164058e-01 -2.08573416e-02 -9.80562270e-01 1.90257037e+00
-5.87780058e-01 8.22926521e-01 -1.08438969e-01 -1.50523579e+00
6.86725855e-01 3.18635821e-01 8.46342146e-01 -8.92321646e-01
8.74754116e-02 -6.21053092e-02 -1.73367679e-01 -9.19561386e-01
1.96335807e-01 -1.52806696e-02 8.68328810e-02 3.27106744e-01
7.86159933e-02 2.80385941e-01 2.10735977e-01 -4.47809584e-02
1.14070690e+00 3.90514851e-01 1.73395057e-03 1.74706638e-01
5.50919116e-01 -2.29203045e-01 7.35518456e-01 4.07520026e-01
-3.73647779e-01 6.98495150e-01 2.39929765e-01 -4.03241903e-01
-7.50403166e-01 -9.09842551e-01 3.82792987e-02 1.21199119e+00
4.43408370e-01 -4.62046117e-01 -4.75575626e-01 -8.42635572e-01
-2.11653009e-01 1.26891956e-01 -6.38081193e-01 -3.10286701e-01
-5.66138685e-01 -2.54027665e-01 2.84592628e-01 8.03059340e-01
3.23896527e-01 -9.10027504e-01 -5.71862280e-01 -2.81285271e-02
-3.11089635e-01 -1.22475481e+00 -6.84109747e-01 1.12031959e-02
-9.64970231e-01 -1.20071185e+00 -7.71482289e-01 -7.98530579e-01
7.83214569e-01 7.09222734e-01 7.32595980e-01 4.30249721e-01
-3.11718524e-01 6.06991529e-01 -6.80393755e-01 1.08956009e-01
-7.15337098e-02 -1.65630847e-01 1.83720097e-01 2.34770775e-01
2.47803554e-01 -6.01181805e-01 -8.32529962e-01 4.84439731e-01
-1.07062757e+00 1.70905262e-01 6.22925460e-01 7.65562057e-01
6.21251643e-01 3.26246880e-02 3.54528427e-01 -6.53470814e-01
-1.63168505e-01 -3.43410850e-01 -4.56289113e-01 2.93791801e-01
-3.44217867e-01 -4.29823436e-02 8.37795615e-01 -7.07144558e-01
-8.76899183e-01 3.46402675e-01 4.20296304e-02 -1.00477493e+00
-3.29433471e-01 3.44840705e-01 -2.54192144e-01 4.58624028e-02
3.41040134e-01 2.95825332e-01 1.77568465e-01 -3.61526459e-01
3.46610337e-01 6.96831167e-01 4.41207618e-01 -4.15124804e-01
7.35734701e-01 6.58041060e-01 -2.39120394e-01 -8.68257165e-01
-8.13990176e-01 -7.85114825e-01 -7.72557676e-01 -4.49715048e-01
8.99566114e-01 -9.57467854e-01 -5.80735922e-01 4.83695775e-01
-8.30095708e-01 -6.75016940e-01 -5.96981160e-02 7.34422624e-01
-9.05498326e-01 6.91416204e-01 -6.84472382e-01 -6.58439577e-01
1.69645268e-02 -1.05747688e+00 1.03487659e+00 1.60012737e-01
4.41460609e-02 -1.04824030e+00 -1.14529751e-01 5.43911874e-01
2.18354881e-01 9.00548249e-02 6.24295831e-01 -2.44429052e-01
-6.56766474e-01 -1.87450364e-01 -2.72755325e-01 5.35013795e-01
4.68260318e-01 1.66640237e-01 -9.51401889e-01 -4.51671958e-01
1.09769322e-03 -5.50590754e-01 1.01569295e+00 3.75938326e-01
1.54399252e+00 -2.07526535e-01 -2.14785293e-01 8.08674395e-01
1.23204267e+00 1.75334901e-01 7.52101481e-01 3.03302079e-01
9.07290697e-01 7.22446561e-01 9.53131437e-01 5.12045622e-01
2.49259382e-01 9.49498236e-01 2.93200046e-01 -7.64638484e-02
-4.88349423e-02 -2.96645045e-01 6.72179639e-01 9.18090880e-01
-4.92823303e-01 -1.19016081e-01 -6.13161325e-01 2.50393122e-01
-2.21790290e+00 -1.13599002e+00 1.71007931e-01 2.32601690e+00
6.66134775e-01 -1.02836274e-01 2.31643811e-01 1.45455733e-01
4.82973099e-01 2.41201311e-01 -6.13821268e-01 1.91634551e-01
1.84241861e-01 -1.96001187e-01 4.43055421e-01 2.24712715e-01
-1.37731731e+00 8.61087978e-01 5.26083660e+00 8.62563074e-01
-1.48563111e+00 6.55017514e-03 7.34888434e-01 -1.56630099e-01
-2.16754619e-02 -5.93876392e-02 -4.00521457e-01 6.83471382e-01
8.38335216e-01 1.03533603e-01 4.31155026e-01 7.39096880e-01
4.93853062e-01 7.35993609e-02 -1.36650670e+00 1.33212841e+00
1.37156680e-01 -9.45669234e-01 -3.69138569e-02 -3.50131840e-02
5.10884762e-01 -1.69399694e-01 -3.80537729e-03 4.65086460e-01
-2.85975903e-01 -1.00390995e+00 6.65190697e-01 5.56995034e-01
9.29082513e-01 -4.95745152e-01 4.22711879e-01 4.50095773e-01
-1.28711808e+00 -2.40624189e-01 -4.72498536e-01 -6.61536753e-02
8.45563337e-02 3.15851241e-01 -1.56776056e-01 4.64696318e-01
6.95916533e-01 1.35817587e+00 -5.15466988e-01 1.03222072e+00
-3.06370467e-01 5.45581341e-01 -1.18261926e-01 2.72712618e-01
1.68312624e-01 -3.21517199e-01 2.19191328e-01 1.09869003e+00
2.61894256e-01 1.46659032e-01 3.62680972e-01 3.59760076e-01
-2.94631496e-02 9.18443948e-02 -5.67031920e-01 -1.18831545e-01
2.40608174e-02 1.22140300e+00 -6.60648942e-01 -2.59619206e-01
-6.29698157e-01 1.14940107e+00 3.83489698e-01 5.85368931e-01
-1.00341713e+00 1.20340995e-01 7.53244996e-01 1.40196025e-01
3.79577726e-01 -3.93339664e-01 4.48991746e-01 -1.52806294e+00
6.34841397e-02 -8.45970154e-01 4.08693999e-01 -5.43927014e-01
-1.08775687e+00 4.39158529e-01 -1.66681334e-02 -1.71930265e+00
-4.15512443e-01 -7.42965519e-01 -5.86471498e-01 1.68751210e-01
-1.49728251e+00 -1.00840366e+00 -5.01772165e-01 7.06229806e-01
7.57535756e-01 -3.74209322e-02 5.42589068e-01 6.03778839e-01
-7.13690698e-01 6.18879139e-01 1.56335309e-01 2.88843274e-01
8.72615457e-01 -1.08896661e+00 -1.94951490e-01 6.85386300e-01
4.77789193e-01 2.77918786e-01 4.42253113e-01 -1.89378217e-01
-1.72912610e+00 -1.12618184e+00 2.97985822e-01 -3.91826928e-01
6.69439137e-01 -2.06625924e-01 -9.42344725e-01 5.24865568e-01
-1.45008102e-01 3.52839261e-01 6.73476279e-01 -6.43512085e-02
-4.52504069e-01 -3.33277524e-01 -5.48707068e-01 3.27685535e-01
1.17756999e+00 -5.65991879e-01 -3.06687504e-01 4.78053063e-01
3.71217996e-01 -1.20993808e-01 -7.64646709e-01 4.52715218e-01
6.04414761e-01 -8.44787955e-01 9.47906315e-01 -7.01614082e-01
5.57308018e-01 -3.65340322e-01 -1.16058923e-01 -1.07025838e+00
-3.05295765e-01 -2.99056828e-01 -3.05912822e-01 1.13551176e+00
1.09650701e-01 -2.46942088e-01 9.60156977e-01 4.79217559e-01
1.02509193e-01 -8.56684923e-01 -7.73373067e-01 -9.52233553e-01
-1.70466185e-01 -4.45153713e-01 -2.90863891e-03 9.49638963e-01
6.54406101e-02 2.47523695e-01 -5.91801047e-01 4.12892550e-02
3.30699235e-01 3.43888789e-01 7.25081861e-01 -8.37254345e-01
-5.89408934e-01 -4.63791639e-01 -7.47248113e-01 -1.61604929e+00
3.52645218e-01 -6.49835825e-01 2.89162815e-01 -1.25948226e+00
3.37140262e-01 -5.24142265e-01 -5.46817183e-01 5.54983079e-01
-1.57241151e-01 3.78827274e-01 2.99665451e-01 5.82564294e-01
-1.02572179e+00 6.96789205e-01 1.38881505e+00 -3.72009277e-01
-1.15044139e-01 1.05626985e-01 -3.37638021e-01 6.77724123e-01
7.25390494e-01 -1.51912019e-01 -6.86696649e-01 -5.57457387e-01
-2.01933295e-01 2.91672826e-01 2.41409734e-01 -9.65330124e-01
1.16692156e-01 -3.07890862e-01 4.64435488e-01 -1.21301532e-01
4.76408839e-01 -7.87027717e-01 -1.55692458e-01 3.03084165e-01
-2.18531027e-01 -1.36763453e-01 -1.73574537e-02 8.17840517e-01
-4.52175826e-01 -2.31050074e-01 6.29935801e-01 5.89092113e-02
-1.08446276e+00 6.64729297e-01 -3.94905359e-01 -1.90838724e-01
1.13890839e+00 -2.59755164e-01 -1.86091051e-01 -7.10979521e-01
-6.69298351e-01 3.44460219e-01 6.77757978e-01 6.98689401e-01
7.80678213e-01 -1.32751286e+00 -4.05375183e-01 6.10097088e-02
3.20963562e-01 -5.28630577e-02 4.18806434e-01 1.02833986e+00
-4.55559045e-01 2.58797526e-01 -2.49075592e-01 -7.28735149e-01
-1.18127942e+00 8.84269357e-01 1.51547849e-01 3.13583873e-02
-5.69069326e-01 6.15239203e-01 3.89720321e-01 7.64118731e-02
4.10747975e-01 -3.35108548e-01 -2.91927308e-01 4.37213294e-02
8.13411117e-01 1.70896292e-01 -1.59594715e-01 -7.82332659e-01
-3.03914517e-01 7.46528864e-01 -2.82002389e-02 2.98932999e-01
1.24921227e+00 -3.05563569e-01 2.54930764e-01 4.45380807e-01
1.64302623e+00 -3.17448139e-01 -1.73473465e+00 -1.49636716e-01
-1.03756778e-01 -6.86281145e-01 7.61620328e-02 -1.32628620e-01
-1.54695058e+00 9.47180033e-01 6.84025049e-01 1.34440148e-02
1.37523055e+00 -2.86912508e-02 7.67459810e-01 3.58999133e-01
3.84060591e-01 -1.03536463e+00 4.81601626e-01 2.77595639e-01
6.47663534e-01 -1.52160382e+00 -1.38859376e-01 -4.39169705e-01
-5.35420239e-01 1.23401606e+00 8.44946444e-01 -8.09612870e-03
5.50612271e-01 -3.57332230e-02 2.42383897e-01 -1.41468486e-02
-7.40375698e-01 -2.53472626e-01 3.84160310e-01 4.97608960e-01
4.38271433e-01 -6.12104163e-02 4.36484776e-02 2.99846292e-01
2.63466150e-01 -5.63864931e-02 2.35382512e-01 8.49371850e-01
-2.26509377e-01 -1.15393651e+00 -4.13560309e-02 4.50767875e-01
-2.04270706e-01 3.20594430e-01 -1.28483534e-01 5.51533282e-01
5.43102175e-02 9.19231951e-01 1.91516444e-01 -6.01297140e-01
2.10468724e-01 -2.48225510e-01 6.79394901e-01 -5.56747139e-01
8.79703835e-02 4.40628417e-02 -2.20618472e-01 -8.33245337e-01
-7.19562650e-01 -4.99414355e-01 -1.15134883e+00 -1.33405313e-01
-2.27252617e-01 3.46941724e-02 3.47343802e-01 9.12158608e-01
1.48807526e-01 3.08019191e-01 8.58331323e-01 -1.05691910e+00
-4.50555444e-01 -7.87998617e-01 -6.65666819e-01 9.08162057e-01
4.06997293e-01 -8.81060004e-01 -3.96741599e-01 4.54689711e-01]
|
[8.75222110748291, 0.7637948989868164]
|
e7f4ab57-44bb-4bd5-9bc1-a0b7ef501287
|
data-to-text-generation-with-variational
|
2202.13756
| null |
https://arxiv.org/abs/2202.13756v1
|
https://arxiv.org/pdf/2202.13756v1.pdf
|
Data-to-text Generation with Variational Sequential Planning
|
We consider the task of data-to-text generation, which aims to create textual output from non-linguistic input. We focus on generating long-form text, i.e., documents with multiple paragraphs, and propose a neural model enhanced with a planning component responsible for organizing high-level information in a coherent and meaningful way. We infer latent plans sequentially with a structured variational model, while interleaving the steps of planning and generation. Text is generated by conditioning on previous variational decisions and previously generated text. Experiments on two data-to-text benchmarks (RotoWire and MLB) show that our model outperforms strong baselines and is sample efficient in the face of limited training data (e.g., a few hundred instances).
|
['Mirella Lapata', 'Yao Fu', 'Ratish Puduppully']
|
2022-02-28
| null | null | null | null |
['data-to-text-generation']
|
['natural-language-processing']
|
[ 6.27766430e-01 1.00679684e+00 -2.89260775e-01 -2.27204829e-01
-1.32853425e+00 -6.07304096e-01 1.40097511e+00 1.03358127e-01
-2.03293070e-01 1.26068342e+00 1.02640820e+00 -3.05803835e-01
3.46319139e-01 -9.28093970e-01 -9.69483018e-01 -3.66205812e-01
3.60303789e-01 1.20750713e+00 -2.09416747e-01 -6.35130927e-02
2.63056904e-01 -1.49352834e-01 -1.29679871e+00 6.84030950e-01
8.52224469e-01 3.82936776e-01 4.12905306e-01 7.56386817e-01
-4.55303282e-01 9.52760398e-01 -8.11013162e-01 -2.68068910e-01
-4.37891297e-02 -7.20682740e-01 -1.03704929e+00 4.59773928e-01
1.30109131e-01 -5.39084613e-01 -1.03576398e-02 5.20417809e-01
2.83805043e-01 5.10084987e-01 1.15865016e+00 -9.10079777e-01
-6.19208395e-01 1.44213080e+00 -2.75217444e-01 -3.18414092e-01
5.05597293e-01 3.79335642e-01 1.18666649e+00 -9.42789674e-01
1.05971527e+00 1.46052623e+00 1.61565185e-01 9.49525356e-01
-1.57849586e+00 -2.41518244e-02 2.47881293e-01 -5.81380367e-01
-9.62335408e-01 -6.79404080e-01 3.90867501e-01 -4.89308804e-01
1.38542533e+00 3.28443609e-02 3.67312044e-01 1.62930071e+00
3.68701369e-01 1.12350047e+00 4.80385721e-01 -5.76227665e-01
3.17399204e-01 -1.48987487e-01 -1.47808120e-01 5.38168252e-01
1.69451341e-01 1.61376875e-02 -7.59828389e-01 -7.75210336e-02
5.71321070e-01 -3.79783064e-01 9.68719870e-02 2.66821474e-01
-1.50180531e+00 1.05178273e+00 4.79920991e-02 9.95226875e-02
-6.31245434e-01 4.51797873e-01 1.38394877e-01 -2.11229727e-01
5.96869171e-01 7.64812469e-01 -1.62451565e-01 -2.65581429e-01
-1.21824849e+00 8.09130132e-01 8.98927987e-01 1.45335364e+00
5.81961572e-01 2.24055767e-01 -1.01758993e+00 5.84612906e-01
3.49690437e-01 5.87229669e-01 6.23707533e-01 -8.84658217e-01
1.16968453e+00 2.76105106e-01 5.25607586e-01 -1.82373017e-01
-1.94042906e-01 2.45941833e-01 -8.71307194e-01 -1.50411114e-01
1.57484382e-01 -7.75686324e-01 -1.40553164e+00 1.49364150e+00
-5.18716611e-02 -3.12395662e-01 4.11441714e-01 3.98857921e-01
7.39532053e-01 1.34055531e+00 9.76707339e-02 -4.20171142e-01
1.00314236e+00 -1.24341619e+00 -8.50028813e-01 -6.26656711e-01
5.12289941e-01 -5.89292109e-01 1.03864670e+00 3.19019228e-01
-1.59396887e+00 -4.21534836e-01 -6.90420687e-01 -5.66855729e-01
-2.56790876e-01 1.25110433e-01 3.09727430e-01 7.87446722e-02
-1.15003145e+00 6.14926875e-01 -9.28468704e-01 -5.22590876e-02
4.28874731e-01 -2.04032971e-04 -2.67460402e-02 1.16368115e-01
-1.04794955e+00 4.94834393e-01 9.42100346e-01 -1.44679621e-01
-1.33987582e+00 -5.14360011e-01 -8.71218860e-01 1.24471262e-01
4.43452030e-01 -1.18328190e+00 1.84962058e+00 -3.47150892e-01
-1.77498305e+00 3.42289358e-01 -6.33738041e-01 -6.88742161e-01
7.56463826e-01 -2.56373405e-01 1.34285882e-01 -1.11625195e-01
3.53479654e-01 1.17061186e+00 8.26354206e-01 -1.29648089e+00
-7.08733022e-01 6.06198236e-02 -2.17420533e-01 4.64101493e-01
3.17115523e-02 -2.41865516e-01 -5.23759425e-01 -6.06180310e-01
-1.91809848e-01 -7.88583875e-01 -5.52902579e-01 -7.41153002e-01
-1.08089602e+00 -6.43399239e-01 2.41406262e-01 -6.90396488e-01
1.14138210e+00 -1.47154367e+00 5.31494558e-01 -1.14251319e-02
-1.77590568e-02 -2.94315964e-01 -1.70830309e-01 7.85053253e-01
2.28804141e-01 6.20434284e-01 -3.55603367e-01 -9.52177167e-01
3.63856226e-01 9.90102515e-02 -8.80370080e-01 -4.22102123e-01
4.25650448e-01 1.26315916e+00 -9.51914132e-01 -7.69693375e-01
-1.32579759e-01 4.02696282e-02 -6.36497974e-01 4.24246937e-01
-1.44674706e+00 3.49824727e-01 -6.54956341e-01 2.59071201e-01
-1.47800565e-01 -4.55739379e-01 2.22292453e-01 3.71342421e-01
-2.08187118e-01 7.60454655e-01 -8.75944614e-01 1.98725569e+00
-5.84488690e-01 5.26292026e-01 -4.09359545e-01 -2.57942408e-01
5.24325192e-01 7.04161882e-01 -3.60943936e-03 -3.15936387e-01
-1.37340203e-02 -5.68156354e-02 -5.62410355e-01 -3.67544949e-01
1.08579719e+00 -2.45216519e-01 -3.97965372e-01 1.11291218e+00
1.52157201e-02 -6.62043214e-01 8.25780034e-01 6.08506382e-01
8.43350232e-01 3.98637265e-01 3.28394860e-01 1.01085484e-01
-2.83410937e-01 3.58634055e-01 6.31585568e-02 1.07278907e+00
6.82335675e-01 7.61471391e-01 7.71330953e-01 -1.01872765e-01
-1.31227672e+00 -9.83024478e-01 3.04360777e-01 1.04245031e+00
-4.25832391e-01 -7.25174427e-01 -8.06837797e-01 -7.31712162e-01
-1.40482068e-01 1.59078455e+00 -6.10434473e-01 3.07934403e-01
-6.70381010e-01 -7.27925062e-01 4.38274473e-01 5.89490771e-01
-4.33280170e-02 -1.64436388e+00 -4.42318529e-01 5.74950695e-01
-4.83811080e-01 -9.61241245e-01 -5.97441733e-01 1.20118342e-01
-9.44564402e-01 -3.59702915e-01 -6.20237768e-01 -7.11943269e-01
8.36048126e-01 -2.53979355e-01 1.54845607e+00 -2.47515604e-01
1.20471276e-01 -6.48432672e-02 -1.68330118e-01 -5.45406222e-01
-9.63926196e-01 4.52288687e-01 -3.73303026e-01 -3.52993459e-01
-3.42356890e-01 -2.51813918e-01 -8.99263546e-02 -4.44486409e-01
-1.13501203e+00 9.43972111e-01 8.07424843e-01 1.05486524e+00
7.47547507e-01 -8.82430822e-02 4.39129263e-01 -1.23638034e+00
1.15405238e+00 -5.86395741e-01 -3.89320463e-01 2.81306177e-01
-5.95009327e-01 5.76127410e-01 6.86005414e-01 -2.01523796e-01
-1.61565423e+00 -3.37462947e-02 1.36111096e-01 1.41745046e-01
-3.60310763e-01 8.42695355e-01 -5.60579672e-02 1.34711254e+00
8.07945073e-01 3.78950417e-01 -3.64076823e-01 -2.66633123e-01
8.71448100e-01 5.11090457e-01 5.47620118e-01 -9.01672840e-01
7.53952861e-01 2.01369762e-01 -1.94340006e-01 -5.43267071e-01
-9.47631955e-01 1.36290476e-01 -5.02026200e-01 7.61324763e-02
9.92159486e-01 -8.09245050e-01 -1.55179858e-01 1.89646572e-01
-1.58609486e+00 -1.02624655e+00 -5.94674468e-01 1.18029058e-01
-8.55887592e-01 -1.44103885e-01 -7.88670719e-01 -8.94108593e-01
-6.50940239e-01 -8.05009425e-01 1.42363036e+00 9.12251100e-02
-6.88153625e-01 -1.12181377e+00 1.61167502e-01 2.14564323e-01
4.72416319e-02 3.22101533e-01 1.01380599e+00 -5.98424256e-01
-9.55690086e-01 6.71944916e-02 2.09662721e-01 -1.00891411e-01
-3.18815783e-02 1.86244413e-01 -8.01302671e-01 4.92939651e-02
-4.77937102e-01 -7.67485559e-01 1.05522418e+00 3.48939240e-01
9.79853392e-01 -1.00772882e+00 -4.52316046e-01 1.99396729e-01
9.96442914e-01 1.95392862e-01 4.84930992e-01 -1.68599129e-01
6.68108165e-01 7.48987198e-01 4.04089302e-01 5.61349630e-01
3.84986728e-01 3.87086570e-01 1.28045917e-01 9.52428877e-02
-7.36470819e-02 -8.60206008e-01 5.05060911e-01 5.49918830e-01
1.49829835e-01 -9.25464332e-01 -9.77495790e-01 7.44843781e-01
-1.99852192e+00 -1.16231012e+00 1.06991213e-02 1.62010968e+00
1.37931609e+00 3.97115946e-01 -1.89302012e-01 -2.71159679e-01
3.42163205e-01 4.55516368e-01 -4.68968600e-01 -3.18225294e-01
1.75944082e-02 2.24374846e-01 9.07657892e-02 9.48889256e-01
-8.80182564e-01 1.35066485e+00 6.77172709e+00 7.59603918e-01
-5.95238149e-01 -1.44053921e-01 7.40119457e-01 -6.12288356e-01
-1.01049960e+00 -2.17642039e-02 -1.02704859e+00 3.82076055e-01
1.19434583e+00 -5.52346408e-01 5.84384084e-01 6.49042547e-01
5.02535224e-01 -3.74024026e-02 -1.40881586e+00 2.93653518e-01
5.22998907e-02 -1.88813937e+00 6.42943203e-01 2.63612624e-02
1.38603318e+00 -1.26445452e-02 -8.20476785e-02 4.24288392e-01
1.20287526e+00 -1.30114245e+00 1.11810863e+00 6.25090241e-01
7.90896654e-01 -5.31305015e-01 3.26565504e-01 7.71800280e-01
-7.27991998e-01 1.83365226e-01 -1.44378707e-01 -4.97219972e-02
6.85393810e-01 4.86239135e-01 -1.31649256e+00 5.77144504e-01
8.73585343e-02 6.51585400e-01 -1.62074879e-01 1.57318369e-01
-6.37767911e-01 5.38990021e-01 -1.39353454e-01 -3.35272998e-01
4.48506117e-01 -9.80390608e-02 3.76387388e-01 1.17665672e+00
4.29555893e-01 1.62343472e-01 3.17887336e-01 1.41964424e+00
-4.78201926e-01 3.18472981e-02 -6.27442658e-01 -4.74613398e-01
3.63653868e-01 1.05214500e+00 -5.03552616e-01 -8.09230626e-01
3.55828106e-02 8.94576132e-01 3.69642437e-01 5.49341202e-01
-4.61963147e-01 -2.61908084e-01 -3.43641751e-02 1.13771884e-02
3.06052119e-01 -1.97254196e-01 -6.81708097e-01 -1.05487502e+00
-5.03957979e-02 -8.73948276e-01 2.44157031e-01 -1.06155241e+00
-9.70619500e-01 6.50759041e-01 1.53063849e-01 -5.78671277e-01
-1.34240520e+00 -8.78141895e-02 -8.17024469e-01 1.21546435e+00
-1.16140890e+00 -1.14769804e+00 -1.48586771e-02 2.17261612e-01
1.08400691e+00 -1.43964231e-01 9.13972378e-01 -6.25799596e-01
-5.00695646e-01 2.00065985e-01 1.34253968e-03 4.41153869e-02
4.31568056e-01 -1.50062180e+00 1.14509666e+00 1.00365222e+00
3.77868742e-01 7.38496006e-01 8.33241105e-01 -1.02669466e+00
-1.13076842e+00 -1.31258869e+00 1.43852532e+00 -7.84729183e-01
5.01713693e-01 -6.34807110e-01 -3.64615679e-01 1.06434822e+00
7.48229146e-01 -8.77464712e-01 3.57495248e-01 -8.78015757e-02
8.35864618e-02 4.87233251e-01 -6.95401013e-01 1.10355484e+00
9.52823043e-01 -2.40477994e-01 -8.19068611e-01 6.46645308e-01
1.16074884e+00 -7.75154710e-01 -4.93221074e-01 -1.82163671e-01
1.57177761e-01 -4.55788463e-01 4.87695634e-01 -8.12244415e-01
1.26079369e+00 7.03893043e-03 2.15997249e-01 -1.53814948e+00
-1.02034777e-01 -1.19150615e+00 -3.17205071e-01 1.18608654e+00
1.04020393e+00 -7.77455047e-02 7.78873205e-01 6.68596089e-01
-4.48098570e-01 -5.69480360e-01 -5.77289522e-01 -3.79400313e-01
1.97827369e-01 -3.81429017e-01 6.70097172e-01 3.39987457e-01
1.25317708e-01 9.56045330e-01 -5.48697293e-01 -3.39799613e-01
4.16629344e-01 1.96914598e-01 9.34479356e-01 -9.11152005e-01
-4.10388350e-01 -4.51836079e-01 8.35238576e-01 -1.32497668e+00
3.01145881e-01 -1.00237250e+00 7.40287483e-01 -2.24029350e+00
2.30542734e-01 -8.73109400e-02 3.69148672e-01 6.86948180e-01
-3.81095827e-01 -2.72782713e-01 2.08727643e-01 2.34829322e-01
-5.16066372e-01 8.19023609e-01 1.36582541e+00 -2.95291513e-01
-4.64901835e-01 -5.70948198e-02 -1.03272033e+00 4.21546340e-01
6.78671539e-01 -5.25008619e-01 -6.15165949e-01 -7.14376330e-01
3.47066283e-01 5.29635787e-01 -4.87701222e-02 -6.04789555e-01
2.62465566e-01 -6.66036308e-01 3.79863828e-01 -8.06486249e-01
2.92420805e-01 -5.89475073e-02 -1.04627879e-02 1.94962606e-01
-1.14235187e+00 -4.78038229e-02 2.09705979e-02 6.01450443e-01
-2.23500822e-02 -3.64286512e-01 2.06238136e-01 -4.07892942e-01
-1.67147100e-01 3.09176654e-01 -7.07551479e-01 3.59714240e-01
7.30373859e-01 2.18951181e-01 -4.28135365e-01 -5.93013167e-01
-5.85385799e-01 4.56475347e-01 2.50947118e-01 4.21517432e-01
4.44734067e-01 -1.21936631e+00 -1.06566012e+00 1.14062361e-01
2.40280833e-02 6.73821926e-01 -1.11218266e-01 -2.97162347e-02
-2.77547807e-01 9.76791203e-01 3.11022252e-01 -2.24330455e-01
-6.10446155e-01 3.24015617e-01 -4.76151109e-02 -9.51632023e-01
-5.96796930e-01 6.93880498e-01 8.63502622e-02 -1.77156106e-01
2.68063575e-01 -6.91597879e-01 -4.58538420e-02 1.41137004e-01
6.01260483e-01 1.18737146e-01 -3.26363653e-01 -8.64671767e-02
3.50551873e-01 -1.47605643e-01 -2.24292755e-01 -1.03877306e+00
1.20432544e+00 7.87530467e-02 -5.09140193e-02 6.41284704e-01
6.93221450e-01 -4.34775109e-04 -1.55396485e+00 -1.92318007e-01
2.04223469e-01 1.62202105e-01 -1.85494646e-01 -8.97575259e-01
-3.65838617e-01 8.95420671e-01 -5.20282567e-01 1.93687215e-01
4.62852687e-01 1.85737908e-01 7.24834323e-01 6.57758534e-01
1.07702658e-01 -1.25247478e+00 5.99638462e-01 8.16778421e-01
1.27596629e+00 -1.03786051e+00 -5.72889931e-02 7.66686276e-02
-9.27843988e-01 9.94576693e-01 5.70200861e-01 2.18212023e-01
7.97399059e-02 2.85992682e-01 -2.06359342e-01 -2.27805860e-02
-1.55739617e+00 5.74459843e-02 4.31195498e-01 4.22794640e-01
5.89535534e-01 2.09595546e-01 1.38585195e-02 5.97249687e-01
-5.74475467e-01 6.22167960e-02 6.50164485e-01 9.59846497e-01
-5.76429725e-01 -1.17587376e+00 -1.65237933e-01 6.54016256e-01
-2.61613280e-01 -4.77478474e-01 -5.36654830e-01 5.77405453e-01
-3.18517298e-01 1.03704894e+00 2.91735590e-01 2.59915590e-02
3.01046614e-02 4.96840328e-01 2.04793230e-01 -1.30817258e+00
-5.17719150e-01 3.74422252e-01 4.86682445e-01 -3.54876280e-01
2.53964923e-02 -8.02048862e-01 -1.52174902e+00 -2.02794462e-01
1.77587569e-01 2.46289149e-01 5.51737607e-01 1.04669344e+00
3.57251406e-01 8.00379455e-01 2.70685196e-01 -9.71086323e-01
-7.48044670e-01 -1.31666219e+00 -2.55274385e-01 3.17471653e-01
2.07422018e-01 1.42688781e-01 2.81858575e-02 6.51674688e-01]
|
[11.69546890258789, 8.967421531677246]
|
2f3bfb02-a5c9-43cb-bbb9-e5a85782f79c
|
towards-smart-city-security-violence-and
|
2207.12850
| null |
https://arxiv.org/abs/2207.12850v6
|
https://arxiv.org/pdf/2207.12850v6.pdf
|
SSIVD-Net: A Novel Salient Super Image Classification & Detection Technique for Weaponized Violence
|
Detection of violence and weaponized violence in closed-circuit television (CCTV) footage requires a comprehensive approach. In this work, we introduce the \emph{Smart-City CCTV Violence Detection (SCVD)} dataset, specifically designed to facilitate the learning of weapon distribution in surveillance videos. To tackle the complexities of analyzing 3D surveillance video for violence recognition tasks, we propose a novel technique called, \emph{SSIVD-Net} (\textbf{S}alient-\textbf{S}uper-\textbf{I}mage for \textbf{V}iolence \textbf{D}etection). Our method reduces 3D video data complexity, dimensionality, and information loss while improving inference, performance, and explainability through the use of Salient-Super-Image representations. Considering the scalability and sustainability requirements of futuristic smart cities, the authors introduce the \emph{Salient-Classifier}, a novel architecture combining a kernelized approach with a residual learning strategy. We evaluate variations of SSIVD-Net and Salient Classifier on our SCVD dataset and benchmark against state-of-the-art (SOTA) models commonly employed in violence detection. Our approach exhibits significant improvements in detecting both weaponized and non-weaponized violence instances. By advancing the SOTA in violence detection, our work offers a practical and scalable solution suitable for real-world applications. The proposed methodology not only addresses the challenges of violence detection in CCTV footage but also contributes to the understanding of weapon distribution in smart surveillance. Ultimately, our research findings should enable smarter and more secure cities, as well as enhance public safety measures.
|
['Abdulmotaleb El Saddik', 'Mustaqeem Khan', 'Reem Alameeri', 'Li Zhiyuan', 'Toluwani Aremu']
|
2022-07-26
| null | null | null | null |
['video-classification']
|
['computer-vision']
|
[ 3.15015793e-01 -2.64438331e-01 -1.82701021e-01 2.94566788e-02
-7.43499994e-01 -5.39808095e-01 7.11663008e-01 -1.75671309e-01
-1.45109400e-01 2.93209523e-01 3.38161647e-01 -6.94752753e-01
-4.08454418e-01 -7.85620689e-01 -6.31180048e-01 -9.73143458e-01
1.66015714e-01 -9.40125585e-02 1.33221686e-01 -1.34698838e-01
1.22118786e-01 9.48058009e-01 -1.35153043e+00 3.37934524e-01
5.96535265e-01 1.15785837e+00 -1.14246272e-01 8.38362932e-01
5.06379902e-01 1.33198750e+00 -5.71081281e-01 -6.24680340e-01
2.22075403e-01 -3.21474932e-02 -6.59195840e-01 -1.39763191e-01
8.22484493e-01 -8.51845026e-01 -9.01701570e-01 8.76620352e-01
4.76969928e-01 3.84324230e-02 1.11051166e+00 -1.43834507e+00
-6.02699041e-01 2.05628008e-01 -8.82848024e-01 8.46208572e-01
4.54152852e-01 3.42501879e-01 5.35300732e-01 -6.13637805e-01
4.66857940e-01 1.29424155e+00 7.85679758e-01 6.92139030e-01
-6.48633063e-01 -9.24690485e-01 2.03567773e-01 4.18858230e-01
-1.29336429e+00 -1.37981772e-01 8.36800754e-01 -8.66472602e-01
1.05761492e+00 7.43814766e-01 6.79608166e-01 1.88407338e+00
2.06353456e-01 1.00333357e+00 6.96561456e-01 3.94915715e-02
9.82189626e-02 -3.43432799e-02 2.85144597e-01 9.71845269e-01
3.27696830e-01 3.54255795e-01 -3.38434167e-02 -3.89966339e-01
7.20868409e-01 2.87079155e-01 -1.65097669e-01 1.01408444e-01
-6.37443185e-01 1.02302885e+00 2.11739838e-01 2.13674232e-01
-3.75324368e-01 1.85158610e-01 7.71325648e-01 -1.44361719e-01
7.61035800e-01 -1.41987711e-01 -1.63380608e-01 -2.34133363e-01
-8.93887401e-01 2.45752990e-01 1.98595926e-01 6.84473336e-01
1.62973981e-02 3.91418189e-01 -3.91608179e-01 2.84631491e-01
5.38926609e-02 9.88201022e-01 -3.67599428e-01 -7.44858503e-01
9.50623274e-01 5.07912457e-01 -7.05176443e-02 -1.31834066e+00
-7.07348526e-01 -1.41392559e-01 -9.74496841e-01 1.90355569e-01
3.93910147e-02 -2.06786603e-01 -7.46585727e-01 1.27752888e+00
4.64429826e-01 5.44189632e-01 -1.38718933e-01 8.64641488e-01
1.10954201e+00 7.18743503e-01 2.80637920e-01 -2.11922750e-01
1.30876207e+00 -4.26557243e-01 -4.26234365e-01 2.77682871e-01
6.95920587e-01 -2.13113636e-01 6.79080248e-01 4.00161564e-01
-5.86597919e-01 -3.02407593e-01 -5.30398011e-01 8.76975507e-02
-3.47689629e-01 4.73343767e-02 4.11137462e-01 1.16250098e+00
-6.53680086e-01 5.34668937e-02 -6.49515986e-01 -2.15213090e-01
6.91262484e-01 1.87056273e-01 -2.36404434e-01 2.91419271e-02
-1.04960644e+00 1.00602043e+00 -9.07580077e-04 6.42037988e-02
-1.51168811e+00 -7.62104690e-01 -1.11097729e+00 -9.56150368e-02
5.90780139e-01 -3.81002396e-01 6.77918434e-01 -3.84065032e-01
-9.08690691e-01 7.75432646e-01 2.85217196e-01 -2.77976125e-01
3.69510949e-01 -1.73798814e-01 -5.22184074e-01 5.14605045e-01
-6.31535277e-02 1.23977594e-01 1.30865860e+00 -1.01805484e+00
-6.89216912e-01 -4.48436618e-01 2.21695423e-01 -3.79952043e-01
-4.77619380e-01 6.99950457e-01 1.09132282e-01 -7.45506227e-01
-4.68323678e-01 -7.89164066e-01 2.09725574e-01 -1.57468960e-01
-3.74090254e-01 -4.17907894e-01 1.53838825e+00 -1.02284944e+00
1.32195401e+00 -1.96698606e+00 1.07059948e-01 1.43089399e-01
5.76136351e-01 8.28881979e-01 4.46964521e-03 8.06773156e-02
-1.42723843e-01 8.18329025e-03 -9.68920588e-02 8.56355056e-02
-1.96708724e-01 2.08832920e-02 -4.37080085e-01 7.30338514e-01
3.02602082e-01 7.75866449e-01 -9.27589655e-01 -4.45885718e-01
7.80060947e-01 7.08286941e-01 -3.49800557e-01 -3.17757130e-02
2.53936976e-01 4.07332420e-01 -9.33677673e-01 1.00299394e+00
7.11226463e-01 1.64276674e-01 -4.04224604e-01 -7.12737367e-02
-1.46687344e-01 -2.90800720e-01 -8.69463682e-01 8.08602154e-01
-3.28298181e-01 7.22303092e-01 2.28627652e-01 -1.20397377e+00
8.49955678e-01 5.38448334e-01 7.31052637e-01 -6.10821843e-01
7.53019512e-01 -2.47070253e-01 -7.21362710e-01 -1.14618552e+00
3.60096514e-01 3.22959572e-01 -2.05552086e-01 1.01349294e-01
-2.50743032e-01 -4.17881981e-02 -1.92048132e-01 1.05220750e-01
1.41536796e+00 -6.36832565e-02 1.62595943e-01 -5.35994954e-03
6.46470070e-01 -1.24867044e-01 3.75912905e-01 7.12648511e-01
-6.23085320e-01 2.27310926e-01 7.22836316e-01 -8.76776099e-01
-8.39142203e-01 -1.00987124e+00 -1.83843717e-01 1.04880071e+00
-8.46083313e-02 -1.07356846e-01 -1.00006986e+00 -1.22150373e+00
-4.08013910e-02 8.66486311e-01 -7.53508270e-01 -2.16899917e-01
-6.31064415e-01 -7.38930523e-01 1.07998848e+00 6.74569249e-01
4.82139140e-01 -8.25270891e-01 -8.09777856e-01 -2.06769049e-01
-2.74307907e-01 -1.44536233e+00 -5.65241873e-02 -3.71609271e-01
-4.61569279e-01 -1.29801929e+00 -6.33717835e-01 -3.63972992e-01
4.14252102e-01 4.49721664e-01 7.37165928e-01 -9.46443081e-02
-6.69567943e-01 8.75072062e-01 -5.41510642e-01 -5.05405545e-01
-1.90495089e-01 -2.62547821e-01 2.24125504e-01 4.93383296e-02
6.01092458e-01 -1.45935223e-01 -4.55561817e-01 2.94045359e-01
-8.89673352e-01 -2.62717426e-01 -1.20256012e-02 4.00893390e-01
1.81947052e-02 2.49535725e-01 9.42816436e-02 -1.67686880e-01
5.27350962e-01 -7.98989594e-01 -8.07992458e-01 2.53892601e-01
3.60180996e-03 -5.95182776e-01 6.87158704e-01 -5.05453169e-01
-1.09994435e+00 -6.84656277e-02 -1.59432039e-01 -8.35882783e-01
-4.28285807e-01 -1.14628319e-02 2.93578468e-02 -1.57358110e-01
7.12469101e-01 2.76551172e-02 -5.64280570e-01 -4.06611204e-01
6.99356496e-02 1.03025568e+00 4.19722289e-01 -3.76293361e-01
9.37193394e-01 5.94697177e-01 3.00510317e-01 -1.22241163e+00
-6.64806306e-01 -5.50256312e-01 -5.01045346e-01 -6.81744695e-01
1.26930237e+00 -9.80843902e-01 -1.24668562e+00 5.24225712e-01
-1.17858279e+00 1.44559741e-01 5.93280904e-02 6.05614007e-01
-2.70819008e-01 6.43224895e-01 -5.21434128e-01 -1.44301438e+00
-5.16322553e-01 -1.25894809e+00 1.53419161e+00 -1.44146215e-02
-7.03163296e-02 -9.09623682e-01 -5.44248819e-02 9.33732748e-01
3.08296438e-02 7.42388248e-01 8.63679528e-01 -3.10726941e-01
-4.11100626e-01 -3.65496278e-01 -5.40805817e-01 4.89264399e-01
-5.50148189e-02 1.38038725e-01 -9.78710473e-01 -1.83626771e-01
-1.58360635e-03 -3.21525276e-01 9.07174230e-01 4.60060298e-01
1.15157664e+00 -4.01867390e-01 -2.24963218e-01 5.69384873e-01
1.03422666e+00 2.65886843e-01 6.16570413e-01 2.17354044e-01
1.05280828e+00 5.44530153e-01 5.21204531e-01 8.96643996e-01
3.76454443e-01 6.84446454e-01 8.86343837e-01 -1.17562175e-01
7.38939270e-02 6.19997084e-02 5.08932650e-01 3.32843482e-01
-7.28807569e-01 -4.26569223e-01 -1.17695129e+00 6.21545970e-01
-1.76759815e+00 -1.61033475e+00 -4.25578237e-01 1.61831439e+00
7.33437017e-02 -2.30298772e-01 2.32284874e-01 2.85280436e-01
6.93065524e-01 3.99190962e-01 -3.19095880e-01 -7.04230309e-01
6.81913942e-02 2.16385961e-01 4.62626576e-01 2.46670604e-01
-1.66487980e+00 7.71584511e-01 5.58914518e+00 1.09252751e+00
-1.18288255e+00 3.52171421e-01 5.18040478e-01 -4.42502767e-01
-1.86416693e-02 -5.83974719e-01 -8.36502433e-01 4.69119579e-01
7.94813812e-01 5.78561544e-01 3.29354912e-01 1.07489562e+00
6.71747506e-01 3.95480245e-02 -6.12042367e-01 1.13728893e+00
4.30318326e-01 -1.30181730e+00 1.93742648e-01 -3.41393389e-02
5.14511526e-01 -7.10442066e-02 2.36004665e-01 2.92273939e-01
9.36873630e-02 -8.67035270e-01 7.02134073e-01 4.04808491e-01
8.69398355e-01 -9.13059533e-01 8.14744771e-01 4.40269947e-01
-1.40445399e+00 -5.61623216e-01 -1.67761698e-01 8.60638916e-02
7.45394230e-02 2.50812948e-01 -5.65580070e-01 6.80190623e-01
1.00539017e+00 6.41598821e-01 -3.16994667e-01 4.29296792e-01
-1.98478594e-01 7.59629369e-01 -5.81500903e-02 -8.58388469e-02
6.16471767e-01 2.99007781e-02 7.31632471e-01 1.43310678e+00
1.96728587e-01 5.81245005e-01 2.26846099e-01 4.58932817e-01
2.93686628e-01 -1.84911400e-01 -1.07671404e+00 2.04832777e-01
5.72913066e-02 1.01878774e+00 -4.10935640e-01 -2.13592529e-01
-5.57095170e-01 6.13769770e-01 1.67158321e-01 2.67479599e-01
-1.39203203e+00 2.14706417e-02 7.44647861e-01 1.03484340e-01
4.04819191e-01 -1.40668601e-01 -1.92670807e-01 -1.07998836e+00
-7.08510876e-02 -8.23432326e-01 5.61625659e-01 -7.37384737e-01
-9.71626818e-01 4.80393171e-01 6.75877154e-01 -1.27140129e+00
-5.31632118e-02 -8.05902302e-01 -5.19123495e-01 2.59936571e-01
-1.36224139e+00 -1.64252234e+00 -3.77455443e-01 1.02890491e+00
7.17512131e-01 -4.80417669e-01 4.79348451e-01 3.26664627e-01
-9.36510742e-01 4.04008418e-01 5.70395496e-03 3.88075382e-01
-1.91169307e-01 -4.05892372e-01 3.17322314e-01 8.89600575e-01
-1.28958344e-01 6.59226477e-02 4.78148103e-01 -8.93064737e-01
-1.40674388e+00 -1.39113569e+00 5.10655403e-01 -7.80884922e-01
7.15797246e-01 -3.74981314e-01 -4.68563348e-01 6.44684911e-01
-1.81232646e-01 -1.36473067e-02 5.76574802e-01 -1.14628188e-01
-5.88408470e-01 -2.17958931e-02 -1.50318956e+00 5.28227687e-01
1.23174882e+00 -6.66158497e-01 -6.63737655e-01 5.26229322e-01
4.60316896e-01 -5.39842807e-02 -7.11416066e-01 4.50609893e-01
4.24773246e-01 -8.71750593e-01 1.17217457e+00 -8.39492261e-01
4.22047585e-01 1.89752474e-01 -1.90765113e-01 -5.75640976e-01
-5.31049192e-01 -5.62210798e-01 -5.87720513e-01 9.76857066e-01
-1.72797233e-01 -3.29098314e-01 7.49683261e-01 6.70688748e-01
-6.67775497e-02 -5.32679617e-01 -1.55945039e+00 -8.07353914e-01
-6.96497178e-03 -8.72239470e-01 3.68774623e-01 9.56130564e-01
-9.88512561e-02 -2.16178581e-01 -9.45530713e-01 4.93826598e-01
8.99320126e-01 -5.89782774e-01 5.61728656e-01 -1.10688961e+00
1.91057950e-01 -2.44634077e-01 -8.06979716e-01 -5.17099500e-01
1.92369089e-01 -4.65456307e-01 -5.27039707e-01 -1.17034662e+00
5.32858908e-01 2.80595366e-02 -2.63582349e-01 6.06348276e-01
-3.53141986e-02 3.47453028e-01 2.76949674e-01 -2.58993953e-01
-4.97511953e-01 4.06620950e-01 9.92996991e-01 -5.32708049e-01
-1.64038595e-02 5.48081771e-02 -2.15920433e-01 9.89629447e-01
7.66109526e-01 -5.54876745e-01 -4.61632162e-01 -5.37159383e-01
-1.98203623e-02 3.06740850e-01 1.07905138e+00 -9.07511413e-01
-9.79877338e-02 -4.00098503e-01 3.33950073e-01 -8.66342485e-01
5.83212495e-01 -1.12977970e+00 -7.79312104e-02 5.34220636e-01
1.75995395e-01 -7.27035031e-02 2.57612348e-01 4.18722361e-01
3.35192025e-01 -1.41802013e-01 6.53010011e-01 -1.15530733e-02
-5.73614597e-01 3.35676700e-01 -8.42475832e-01 -5.22228889e-02
1.57063067e+00 -3.79946530e-01 -6.71457410e-01 -2.76329756e-01
-4.00550306e-01 -7.26433918e-02 -1.35802478e-01 6.73409998e-01
9.95232761e-01 -1.13560474e+00 -8.50087106e-01 1.18486844e-01
1.92037807e-03 -4.54606891e-01 5.98094642e-01 8.70996296e-01
-5.18344104e-01 5.16101778e-01 -1.38830751e-01 -6.23430490e-01
-1.81760383e+00 8.70828390e-01 5.47780879e-02 -1.02131724e-01
-8.05209219e-01 7.33848810e-01 2.01302588e-01 -1.44270495e-01
4.49959427e-01 -3.71763438e-01 -4.86068547e-01 1.21036664e-01
8.44309926e-01 1.04626715e+00 -1.86119139e-01 -1.11455560e+00
-5.59029400e-01 8.07804346e-01 -7.33658224e-02 5.14748156e-01
1.38739729e+00 1.62849054e-01 3.09367180e-01 -2.75066882e-01
1.21466744e+00 -2.60695577e-01 -1.31335175e+00 1.97994545e-01
-2.73427188e-01 -6.34738028e-01 3.59454125e-01 -6.18792713e-01
-1.17924047e+00 7.89055467e-01 7.52519369e-01 1.45624533e-01
1.20515370e+00 1.30725965e-01 1.00642908e+00 3.59491140e-01
4.79224503e-01 -1.02307153e+00 8.49594995e-02 4.15190607e-01
8.78963411e-01 -1.21395981e+00 1.31477773e-01 -4.61094290e-01
-7.16238976e-01 9.37868476e-01 3.45134705e-01 6.16878644e-02
5.88961720e-01 2.94265181e-01 -1.57280236e-01 -5.25007606e-01
-2.87074775e-01 -1.39144436e-01 3.22966516e-01 1.00324976e+00
-3.48490387e-01 1.11237496e-01 1.15206093e-01 4.15896356e-01
2.28223279e-01 -1.41528651e-01 1.13788515e-01 9.84612644e-01
-4.03243661e-01 -2.72829026e-01 -8.70301962e-01 3.90823543e-01
-4.80366796e-01 3.31443846e-02 -6.20815516e-01 8.21897328e-01
4.81207699e-01 1.32290232e+00 3.65949944e-02 -7.08139896e-01
4.99803245e-01 -4.57169443e-01 3.92104566e-01 -1.08934537e-01
-9.23885584e-01 -2.38733307e-01 2.34252706e-01 -5.27109265e-01
-2.63357490e-01 -7.32708573e-01 -8.11578155e-01 -6.99982285e-01
-1.38523519e-01 -4.29906696e-01 4.67814386e-01 9.55797195e-01
8.65646899e-02 4.91727412e-01 9.00794923e-01 -8.87744248e-01
-4.83438998e-01 -6.35605395e-01 -4.68640476e-01 5.00582099e-01
6.07178211e-01 -9.37972844e-01 -3.47005725e-01 -1.55395031e-01]
|
[7.980055809020996, 0.7293256521224976]
|
b50c63ce-70c9-4bef-a5c7-dab1efa0687c
|
motion-policy-networks
|
2210.12209
| null |
https://arxiv.org/abs/2210.12209v1
|
https://arxiv.org/pdf/2210.12209v1.pdf
|
Motion Policy Networks
|
Collision-free motion generation in unknown environments is a core building block for robot manipulation. Generating such motions is challenging due to multiple objectives; not only should the solutions be optimal, the motion generator itself must be fast enough for real-time performance and reliable enough for practical deployment. A wide variety of methods have been proposed ranging from local controllers to global planners, often being combined to offset their shortcomings. We present an end-to-end neural model called Motion Policy Networks (M$\pi$Nets) to generate collision-free, smooth motion from just a single depth camera observation. M$\pi$Nets are trained on over 3 million motion planning problems in over 500,000 environments. Our experiments show that M$\pi$Nets are significantly faster than global planners while exhibiting the reactivity needed to deal with dynamic scenes. They are 46% better than prior neural planners and more robust than local control policies. Despite being only trained in simulation, M$\pi$Nets transfer well to the real robot with noisy partial point clouds. Code and data are publicly available at https://mpinets.github.io.
|
['Dieter Fox', 'Byron Boots', 'Bryan Peele', 'Clemens Eppner', 'Adithyavairan Murali', 'Adam Fishman']
|
2022-10-21
| null | null | null | null |
['robot-manipulation']
|
['robots']
|
[-1.25200748e-01 3.01183820e-01 -1.56669319e-01 5.98535547e-03
-8.13591838e-01 -4.88269866e-01 4.82221812e-01 -3.29526722e-01
-4.92960751e-01 8.82255256e-01 7.99369663e-02 -2.74288982e-01
-2.01907486e-01 -7.73793161e-01 -1.03311670e+00 -6.08959079e-01
-5.26754200e-01 9.84166741e-01 3.13236505e-01 -4.75856274e-01
1.21905923e-01 6.53746426e-01 -1.38639545e+00 -2.22893685e-01
6.36545420e-01 5.91773689e-01 8.04253995e-01 9.92392898e-01
3.17324758e-01 6.02070928e-01 -3.70771348e-01 3.63037765e-01
5.38363159e-01 -1.69545263e-01 -7.88167000e-01 -3.94242816e-02
1.36641890e-01 -6.35598481e-01 -5.50325692e-01 9.32264447e-01
6.44570649e-01 5.64100802e-01 5.60194671e-01 -1.36790085e+00
-2.80499637e-01 5.07333279e-01 -2.92556375e-01 -2.31722787e-01
3.42429876e-01 8.97784770e-01 9.01094913e-01 -6.53656304e-01
9.92065370e-01 1.46568334e+00 4.49422628e-01 7.47150600e-01
-9.27301645e-01 -4.41467166e-01 5.29675968e-02 -7.29468912e-02
-1.13223958e+00 -3.69604409e-01 2.43850514e-01 -3.24208707e-01
1.36885667e+00 -1.85401052e-01 4.85428154e-01 1.07867706e+00
7.43997276e-01 6.85905755e-01 1.89816684e-01 1.09151863e-01
3.34685564e-01 -6.96981668e-01 -4.84780490e-01 8.34149301e-01
2.75859743e-01 3.11288178e-01 -1.77280769e-01 -8.15336332e-02
1.19541574e+00 -2.80910522e-01 -3.37006956e-01 -5.48416972e-01
-1.72499704e+00 7.58515060e-01 6.93469107e-01 -9.96705145e-02
-5.62010825e-01 9.44811106e-01 1.48424044e-01 6.87364563e-02
-2.39777386e-01 9.14750755e-01 -4.56357181e-01 -4.31124717e-01
-1.98286787e-01 1.01802433e+00 8.48476171e-01 1.41046154e+00
6.66364968e-01 3.98205936e-01 1.56054616e-01 4.78954166e-01
2.10342795e-01 7.21004426e-01 4.58017051e-01 -1.84751999e+00
6.15020454e-01 1.37743384e-01 6.05410397e-01 -1.14586258e+00
-8.01198661e-01 -1.25351533e-01 -7.02034473e-01 7.52735436e-01
3.89315128e-01 -5.82050204e-01 -8.87622595e-01 1.70912349e+00
2.11320981e-01 -1.99891508e-01 1.25967562e-01 1.03784800e+00
4.57944006e-01 1.04769576e+00 -9.03967097e-02 2.17818603e-01
8.03313792e-01 -1.33454978e+00 -3.55948776e-01 -8.04576755e-01
6.69799805e-01 -6.45744264e-01 7.52339661e-01 4.26947743e-01
-1.38779223e+00 -4.40614820e-01 -9.68075156e-01 2.33616922e-02
7.35253096e-02 -7.48339146e-02 5.33701003e-01 -2.32058972e-01
-1.43486500e+00 9.68620479e-01 -1.26250303e+00 -3.70826155e-01
1.42503962e-01 6.37321770e-01 -4.25335974e-01 -6.11392409e-02
-7.71091819e-01 1.22612548e+00 5.04637480e-01 1.66222975e-01
-1.44035423e+00 -1.14772253e-01 -9.41231012e-01 -7.66721219e-02
4.30790156e-01 -1.00756979e+00 1.84237838e+00 -3.64893824e-01
-1.78533149e+00 1.21481739e-01 -9.82076600e-02 -4.52512443e-01
5.17443955e-01 -4.08029854e-01 2.61503100e-01 2.01773345e-01
3.63455921e-01 1.30970621e+00 4.44219798e-01 -1.35338378e+00
-7.54404008e-01 2.87357271e-01 -1.08936884e-01 5.87960839e-01
4.68883544e-01 -1.86543703e-01 -4.77240294e-01 -3.12340796e-01
1.52299404e-01 -1.37369156e+00 -9.95659769e-01 1.18718587e-01
-3.47328752e-01 -2.35227153e-01 7.57100523e-01 -3.57620925e-01
4.16439474e-01 -1.54741657e+00 5.26420176e-01 -8.92152637e-02
-1.11016147e-01 1.36510968e-01 -5.15931308e-01 6.39957428e-01
3.35632712e-01 -9.53672975e-02 -2.98073232e-01 -2.24303588e-01
2.28667691e-01 4.23159957e-01 -3.26042742e-01 5.23018837e-01
2.51654983e-01 1.09243417e+00 -1.17842579e+00 -1.70828596e-01
2.99050242e-01 2.12166816e-01 -6.62628174e-01 1.28231049e-01
-7.93868542e-01 6.10033035e-01 -5.85333943e-01 6.45387590e-01
1.84643745e-01 -1.64678708e-01 3.10130790e-03 4.69060808e-01
-2.48150662e-01 2.27339610e-01 -1.11798310e+00 1.98338783e+00
-3.57241005e-01 6.65492952e-01 5.17406762e-01 -7.13015795e-01
8.48853350e-01 2.91709006e-01 7.04614878e-01 -3.53785574e-01
3.39327008e-01 4.46207136e-01 1.49252370e-01 -4.79296803e-01
8.96531940e-01 1.18111596e-01 -3.70121568e-01 2.90735722e-01
-8.08904395e-02 -7.70546973e-01 3.19602609e-01 -7.52948374e-02
1.53895879e+00 5.30634701e-01 4.44434099e-02 -1.43503666e-01
-7.80651122e-02 6.94261551e-01 7.57484019e-01 7.96802461e-01
-2.65924960e-01 7.62973428e-01 1.94689274e-01 -4.83737022e-01
-1.23944151e+00 -1.00174046e+00 4.86768603e-01 7.27478921e-01
5.36058426e-01 -1.54965430e-01 -4.65890616e-01 -1.36888120e-02
-3.63611453e-03 5.19627392e-01 2.74799927e-03 -7.10504800e-02
-1.15935123e+00 -4.37169999e-01 4.79195237e-01 6.36629581e-01
4.73234713e-01 -1.52161229e+00 -1.11231852e+00 8.06050599e-01
-2.32930198e-01 -1.21660697e+00 -2.82055914e-01 1.37691617e-01
-8.96755815e-01 -9.60678518e-01 -6.27897322e-01 -8.68400514e-01
5.78900754e-01 4.40353125e-01 1.03742230e+00 1.40041739e-01
-2.60649353e-01 1.94163084e-01 -2.45889261e-01 -4.22452092e-01
-5.89857399e-01 2.46027306e-01 1.88574433e-01 -9.70719635e-01
-3.48771572e-01 -5.17089605e-01 -6.31282926e-01 5.24100304e-01
-6.70239210e-01 6.82504624e-02 1.01807582e+00 6.87871218e-01
6.09437883e-01 5.84351271e-02 3.28658313e-01 -2.21724063e-01
5.67519426e-01 -5.96516609e-01 -7.48395026e-01 -4.00608659e-01
-1.49740130e-01 1.19167298e-01 7.98080325e-01 -4.10101503e-01
-8.14073682e-01 4.62649554e-01 -3.24449927e-01 -4.84721839e-01
-4.28571284e-01 3.29570413e-01 -3.56776454e-02 -3.31337983e-03
8.26000154e-01 -1.02769777e-01 4.35570888e-02 -1.04414403e-01
5.02872944e-01 1.38289528e-02 8.59776199e-01 -8.09789479e-01
8.61052573e-01 5.16873479e-01 7.16879815e-02 -8.21766973e-01
-6.03245795e-02 -2.19874069e-01 -5.45444906e-01 -1.71371922e-01
9.08381104e-01 -7.28714824e-01 -7.94774592e-01 5.42708695e-01
-1.44173026e+00 -1.12080574e+00 -8.77014399e-02 4.73929226e-01
-1.22868562e+00 1.14035137e-01 -6.22997761e-01 -5.12201905e-01
-2.30402529e-01 -1.56688774e+00 1.16308546e+00 2.76687741e-01
-4.32940513e-01 -6.97090268e-01 1.18797384e-01 -2.12529123e-01
4.75805104e-01 7.13399231e-01 4.78729099e-01 6.12964630e-02
-1.00790811e+00 -2.17905924e-01 1.38285533e-01 -2.32087955e-01
8.35495889e-02 1.78282022e-01 -3.94077748e-01 -3.96050751e-01
-3.17521840e-01 -4.23621327e-01 6.15184486e-01 6.86707973e-01
8.00975084e-01 -4.77902174e-01 -6.54774666e-01 5.72691441e-01
1.26967347e+00 4.86464977e-01 5.26815295e-01 6.51129127e-01
6.92869723e-01 3.87403339e-01 8.42635930e-01 2.67623186e-01
5.01758575e-01 6.82285070e-01 9.47582304e-01 2.66526788e-01
7.61560425e-02 -1.44009039e-01 6.32273972e-01 4.27358091e-01
-1.29229203e-01 -4.11198169e-01 -1.28253543e+00 7.27290988e-01
-2.14018297e+00 -9.82003450e-01 -9.75111946e-02 1.86597490e+00
4.39222187e-01 2.21290559e-01 -4.01897244e-02 -4.11410600e-01
5.61814904e-01 1.36056289e-01 -7.45208204e-01 -4.10376400e-01
2.51077652e-01 9.45132151e-02 8.58063459e-01 6.64407730e-01
-1.06253862e+00 1.18133175e+00 6.15955877e+00 5.05617023e-01
-1.01602459e+00 -1.96403265e-01 1.36720046e-01 -1.88604161e-01
-4.89666592e-03 -8.28657008e-04 -7.32302904e-01 1.27052367e-01
9.36155379e-01 -6.22391403e-02 4.36015069e-01 1.21369565e+00
5.77620745e-01 -3.01924676e-01 -9.56796288e-01 5.51299214e-01
-4.05107260e-01 -1.43085742e+00 -3.46735299e-01 1.30627692e-01
8.83576870e-01 7.66199470e-01 -1.67875990e-01 3.06633949e-01
1.21024251e+00 -1.26761699e+00 9.60336268e-01 2.55674034e-01
4.46705699e-01 -8.48158002e-01 5.52198529e-01 8.18954408e-01
-1.18719184e+00 -1.05294935e-01 -6.38044238e-01 -2.64851481e-01
7.92411685e-01 7.58611187e-02 -1.01266456e+00 4.29189473e-01
6.41431272e-01 5.18456697e-01 -2.87298230e-03 1.14074230e+00
-3.54210138e-01 -2.29110438e-02 -5.87491632e-01 -2.10856125e-01
7.32007205e-01 -4.52592447e-02 7.99040318e-01 8.76191735e-01
5.41825831e-01 6.66085184e-02 5.00920653e-01 7.96642900e-01
2.37735674e-01 -5.28670907e-01 -9.95679617e-01 1.44380971e-03
5.04557014e-01 1.19292951e+00 -7.03834653e-01 -7.21610636e-02
8.17892551e-02 8.52442503e-01 3.86527687e-01 2.90960103e-01
-9.40380335e-01 -5.14940917e-01 1.07756257e+00 -2.12138191e-01
1.92859888e-01 -1.05939031e+00 -1.17049240e-01 -7.07911551e-01
-1.33883908e-01 -8.10748994e-01 -1.63836628e-01 -9.69985366e-01
-8.76203537e-01 6.08379185e-01 1.12501949e-01 -1.42525411e+00
-6.83391869e-01 -8.50365400e-01 -6.37215734e-01 7.03547597e-01
-1.30060613e+00 -7.71310627e-01 -4.22032833e-01 2.39198580e-01
7.76134849e-01 1.13376454e-01 8.28212798e-01 -1.23660550e-01
-2.57630795e-01 -9.11352858e-02 -8.78416374e-02 4.34520058e-02
4.89517599e-01 -1.17507148e+00 9.00454819e-01 8.37241650e-01
-3.67289245e-01 5.38676500e-01 9.63263452e-01 -7.61297345e-01
-1.82949996e+00 -1.29168260e+00 4.37577814e-01 -5.36651671e-01
5.17518640e-01 1.13230795e-01 -6.60399199e-01 8.79950106e-01
3.41795772e-01 -2.10101545e-01 -3.61556977e-01 -5.95641315e-01
3.81807715e-01 3.51837218e-01 -9.05739367e-01 9.90755916e-01
1.19078624e+00 2.28649944e-01 -3.40334117e-01 4.39168066e-01
8.70221257e-01 -1.13879669e+00 -6.23576403e-01 5.43455660e-01
3.20827454e-01 -7.74436235e-01 9.90801394e-01 -2.84789413e-01
6.67998552e-01 -4.48906362e-01 7.51748160e-02 -1.50110209e+00
-3.63118827e-01 -9.67884243e-01 3.67728062e-02 3.94219995e-01
5.56116819e-01 -5.70658028e-01 8.46917033e-01 5.38012147e-01
-7.29294538e-01 -6.28304780e-01 -9.20606494e-01 -1.04528916e+00
4.19255167e-01 -4.47131187e-01 4.16313589e-01 5.35058796e-01
-1.48124531e-01 9.95327830e-02 -2.63417900e-01 4.35059190e-01
3.82300735e-01 -1.07490882e-01 1.18756878e+00 -6.60697162e-01
-3.61565292e-01 -7.16665089e-01 -2.18873918e-01 -1.37576330e+00
2.25747481e-01 -6.62682772e-01 1.00511718e+00 -2.07416272e+00
-3.77680987e-01 -5.92939138e-01 5.49296319e-01 6.34768546e-01
1.52633026e-01 -6.44692853e-02 3.16146851e-01 3.79454792e-01
-3.74887228e-01 7.26145446e-01 1.63683558e+00 -1.08695790e-01
-3.93660754e-01 -2.22026519e-02 -2.72696525e-01 8.36867809e-01
1.16502833e+00 -2.48481944e-01 -4.79807943e-01 -9.93619919e-01
2.15990543e-01 4.23082650e-01 4.20212239e-01 -1.43353772e+00
4.57804173e-01 -7.00002670e-01 5.26384823e-02 -5.39444804e-01
6.44242883e-01 -5.35262048e-01 2.17366591e-01 8.05403054e-01
-1.58898026e-01 6.14416122e-01 1.97695136e-01 6.09004676e-01
9.71964281e-03 -1.85508698e-01 6.37678504e-01 -6.15345955e-01
-1.06479025e+00 4.84015673e-01 -7.64081419e-01 -1.42602637e-01
1.15470338e+00 3.59753966e-02 -4.83233780e-01 -6.90218806e-01
-6.17102802e-01 6.44767463e-01 7.10716367e-01 5.65695167e-01
7.45661259e-01 -1.21723747e+00 -5.76758623e-01 -2.76813239e-01
-1.97239071e-01 8.76373947e-01 4.23481539e-02 4.66997206e-01
-1.22894478e+00 5.72065771e-01 -2.94633210e-01 -7.36150324e-01
-4.88705695e-01 3.21031511e-01 3.66359532e-01 -1.09370008e-01
-9.03092563e-01 8.47120583e-01 2.97053363e-02 -7.94301450e-01
1.07261010e-01 -6.29264355e-01 2.93138325e-01 -7.66584218e-01
2.41635472e-01 3.87888342e-01 -2.61014104e-01 -5.87386370e-01
-1.37122616e-01 5.54380178e-01 3.88273448e-01 -5.86310148e-01
1.37408209e+00 6.81156591e-02 3.37530300e-02 -1.00031635e-02
8.48840833e-01 -4.27526742e-01 -1.78356755e+00 2.70239472e-01
-6.63812310e-02 -2.69845426e-01 -3.23506325e-01 -3.66105437e-01
-9.57792342e-01 5.98476529e-01 -7.49816149e-02 -1.82767659e-01
5.54320872e-01 -1.06133752e-01 1.10670018e+00 8.91185701e-01
9.32895780e-01 -1.14108896e+00 2.42119595e-01 1.03173542e+00
1.08140266e+00 -1.03833830e+00 -7.92615414e-02 -1.87893242e-01
-6.96264803e-01 1.08953202e+00 9.18095410e-01 -6.86819971e-01
1.46540537e-01 3.95572394e-01 1.22963995e-01 -1.20740838e-01
-9.80609536e-01 -6.88514486e-02 -2.01072291e-01 6.94183886e-01
-1.74627844e-02 -4.13001068e-02 9.38964486e-02 5.51915132e-02
-4.94721979e-01 -4.07753661e-02 7.85662293e-01 1.41158521e+00
-8.36997628e-01 -9.39828336e-01 -3.40254694e-01 6.93028197e-02
5.71291633e-02 5.10030508e-01 -1.12796836e-01 1.10081851e+00
-1.01531789e-01 9.77277577e-01 6.96237898e-03 -3.97938013e-01
3.30026776e-01 -3.68176907e-01 3.62387508e-01 -6.89889550e-01
-2.71897912e-01 -7.78589398e-02 3.43183756e-01 -1.04990935e+00
-2.16679648e-01 -7.42343128e-01 -1.90637553e+00 -5.24637759e-01
-3.83154079e-02 -1.78359717e-01 6.91688776e-01 7.31686294e-01
4.55810159e-01 5.66240311e-01 3.22792262e-01 -1.79965043e+00
-7.28907883e-01 -7.99222231e-01 -6.54189959e-02 -4.51597162e-02
3.73237461e-01 -6.53064787e-01 -1.09635264e-01 -6.92518800e-02]
|
[4.728422164916992, 0.9833574891090393]
|
ec61b8c8-8cf1-4cee-b068-aeff9aaec552
|
scale-aware-super-resolution-network-with
|
2305.19063
| null |
https://arxiv.org/abs/2305.19063v1
|
https://arxiv.org/pdf/2305.19063v1.pdf
|
Scale-aware Super-resolution Network with Dual Affinity Learning for Lesion Segmentation from Medical Images
|
Convolutional Neural Networks (CNNs) have shown remarkable progress in medical image segmentation. However, lesion segmentation remains a challenge to state-of-the-art CNN-based algorithms due to the variance in scales and shapes. On the one hand, tiny lesions are hard to be delineated precisely from the medical images which are often of low resolutions. On the other hand, segmenting large-size lesions requires large receptive fields, which exacerbates the first challenge. In this paper, we present a scale-aware super-resolution network to adaptively segment lesions of various sizes from the low-resolution medical images. Our proposed network contains dual branches to simultaneously conduct lesion mask super-resolution and lesion image super-resolution. The image super-resolution branch will provide more detailed features for the segmentation branch, i.e., the mask super-resolution branch, for fine-grained segmentation. Meanwhile, we introduce scale-aware dilated convolution blocks into the multi-task decoders to adaptively adjust the receptive fields of the convolutional kernels according to the lesion sizes. To guide the segmentation branch to learn from richer high-resolution features, we propose a feature affinity module and a scale affinity module to enhance the multi-task learning of the dual branches. On multiple challenging lesion segmentation datasets, our proposed network achieved consistent improvements compared to other state-of-the-art methods.
|
['Hao Chen', 'Pheng-Ann Heng', 'Huangjing Lin', 'Luyang Luo', 'Yanwen Li']
|
2023-05-30
| null | null | null | null |
['image-super-resolution', 'lesion-segmentation']
|
['computer-vision', 'medical']
|
[ 4.94173378e-01 6.35505989e-02 -3.18257898e-01 -2.83967823e-01
-1.05610955e+00 -1.62906751e-01 -3.10556944e-02 -1.91928893e-01
-4.14870530e-01 4.60370183e-01 2.70962179e-01 -1.03799673e-02
-7.64556900e-02 -8.20683360e-01 -3.42311591e-01 -8.19967628e-01
3.41647655e-01 3.20861548e-01 1.00451589e+00 -2.10992768e-01
1.34185255e-01 6.63768828e-01 -1.23765051e+00 8.36182952e-01
1.17355931e+00 8.17456901e-01 7.24932134e-01 6.35968506e-01
-4.55677271e-01 4.45319921e-01 -2.03458741e-01 1.71806276e-01
8.85728747e-02 -2.93196529e-01 -9.56910193e-01 2.22141683e-01
2.31173605e-01 -5.33305943e-01 -3.59189123e-01 1.24094737e+00
4.51732278e-01 -2.96836168e-01 6.21822536e-01 -3.51534665e-01
-9.12026525e-01 6.31713808e-01 -9.51915205e-01 8.20544839e-01
-3.76864791e-01 8.27275217e-02 4.86060888e-01 -5.29995143e-01
5.60512125e-01 1.10273981e+00 5.46900809e-01 7.15671539e-01
-1.05123234e+00 -6.50960326e-01 1.76037580e-01 5.69447391e-02
-1.39641988e+00 -9.06686038e-02 6.06634736e-01 -4.60797429e-01
6.28465474e-01 2.09397092e-01 3.66734803e-01 6.85708046e-01
2.29898185e-01 5.80018640e-01 1.23986685e+00 -2.23943703e-02
-8.51787105e-02 -1.44369662e-01 -1.41801685e-01 7.95038283e-01
2.10421115e-01 -2.28794608e-02 7.37638995e-02 1.83932170e-01
1.80985522e+00 2.18581706e-01 -3.07038158e-01 6.55224174e-02
-1.36525559e+00 7.98480153e-01 1.00251293e+00 7.22968102e-01
-4.80226159e-01 -3.97240147e-02 3.23198378e-01 -3.18151832e-01
3.98792028e-01 3.04012626e-01 -3.24735224e-01 3.69809687e-01
-8.88756037e-01 -2.18979076e-01 6.75976500e-02 5.71013927e-01
4.60856706e-01 -1.70962617e-01 -4.98272538e-01 1.06634986e+00
1.73886493e-02 -9.92577374e-02 7.67341495e-01 -6.13190949e-01
3.09093624e-01 9.15704966e-01 -2.68577635e-01 -6.37762427e-01
-7.70540535e-01 -8.19635630e-01 -1.28620744e+00 1.51413247e-01
5.50092399e-01 7.42832273e-02 -1.49514306e+00 1.29840171e+00
4.05074894e-01 1.99197605e-01 -1.92749977e-01 1.22729087e+00
9.82858896e-01 3.84926736e-01 1.41738921e-01 -2.21098781e-01
1.75309610e+00 -1.26289618e+00 -5.98527014e-01 -3.41027528e-01
5.30960619e-01 -7.41273999e-01 1.01455379e+00 -1.71359517e-02
-1.12434876e+00 -6.01681888e-01 -9.75322843e-01 -2.65036613e-01
-1.27060935e-01 2.61629522e-01 6.12306356e-01 2.92854249e-01
-9.81376946e-01 4.02019918e-01 -1.09094989e+00 -1.31316945e-01
9.55398798e-01 4.70654130e-01 -8.71420577e-02 5.42881014e-03
-1.03322768e+00 7.12121606e-01 4.44725990e-01 1.38275884e-02
-5.50246775e-01 -9.13139105e-01 -5.72950423e-01 -8.68351758e-02
2.60641396e-01 -6.18085206e-01 1.02044261e+00 -8.91386628e-01
-1.17463899e+00 1.12317014e+00 -1.92270905e-01 -1.96186930e-01
4.56926435e-01 2.96269476e-01 -4.55806583e-01 4.61803079e-01
2.67974645e-01 8.14630568e-01 8.43834519e-01 -1.08982551e+00
-1.07994998e+00 -4.30572569e-01 -7.44254813e-02 2.72012711e-01
-1.53704450e-01 -1.32021122e-02 -7.16239095e-01 -8.73318017e-01
5.17918169e-01 -5.65454900e-01 -7.44997323e-01 -2.55819838e-02
-4.13407326e-01 1.08655663e-02 7.90913105e-01 -7.71001279e-01
1.34753537e+00 -2.17196631e+00 2.14651182e-01 -2.30348445e-02
5.30412078e-01 2.21664980e-01 -7.54938647e-02 -5.65966725e-01
-1.15596376e-01 2.27305815e-01 -3.11973751e-01 1.17643625e-01
-5.42243540e-01 4.17488031e-02 1.81084454e-01 4.21910703e-01
2.92336404e-01 9.65653479e-01 -6.47957385e-01 -8.94424319e-01
2.25262314e-01 6.30431354e-01 -4.56153691e-01 1.10237636e-01
6.47479594e-02 8.39866042e-01 -8.07774544e-01 7.67745554e-01
7.87391663e-01 -6.43317342e-01 -2.87324935e-01 -7.76934564e-01
-2.81933874e-01 -1.64202183e-01 -1.03078020e+00 1.81817317e+00
-4.13015157e-01 1.97714478e-01 2.80702859e-01 -8.92532468e-01
6.47900939e-01 2.74025083e-01 6.19928598e-01 -6.95095181e-01
2.84409195e-01 4.48035866e-01 1.48663685e-01 -5.89102924e-01
1.65978506e-01 -2.98415273e-01 7.75544345e-02 2.34092489e-01
-2.62721032e-01 1.14417091e-01 2.23365381e-01 -1.60924673e-01
9.34111774e-01 -3.74202162e-01 2.23507091e-01 -1.36189654e-01
6.68064713e-01 4.54900339e-02 6.53625071e-01 4.48601395e-01
-3.53827208e-01 1.00116527e+00 3.31488490e-01 -5.86992443e-01
-1.14977956e+00 -1.14101613e+00 -4.25386488e-01 1.02883470e+00
3.18342894e-01 2.03665435e-01 -1.12295067e+00 -6.85579419e-01
-3.17409545e-01 -1.07829310e-01 -8.43679786e-01 -8.41051564e-02
-8.78474653e-01 -1.09462631e+00 4.88395184e-01 7.87714481e-01
8.06698978e-01 -1.12491846e+00 -6.42296135e-01 3.51346076e-01
-3.54924172e-01 -1.20946443e+00 -9.53697681e-01 1.62678495e-01
-9.81817067e-01 -9.01979208e-01 -1.02798378e+00 -1.31317794e+00
1.00475550e+00 2.80328751e-01 6.96564615e-01 2.80732542e-01
-7.17039883e-01 -4.48915809e-01 -1.85968503e-01 -5.89422323e-02
-4.07807797e-01 4.39311713e-01 -3.90988439e-01 -9.39028338e-02
9.86555442e-02 -5.89585185e-01 -9.39417958e-01 4.65313315e-01
-1.19880378e+00 5.17639279e-01 1.28926003e+00 8.78369808e-01
1.22791803e+00 3.85818630e-01 7.18888402e-01 -1.06121492e+00
4.67087239e-01 -1.83296517e-01 -4.41747427e-01 2.32218131e-01
-2.68803835e-01 9.66282412e-02 6.22148037e-01 -6.51313484e-01
-1.09554577e+00 2.63377219e-01 -1.58102512e-01 -2.35046059e-01
-3.72606039e-01 2.07118839e-01 -6.51745871e-02 -2.78340667e-01
6.90733790e-01 2.54310876e-01 -2.67650392e-02 -3.37864041e-01
2.95927942e-01 7.20914602e-01 7.38745749e-01 -1.33611754e-01
6.30413175e-01 8.24002981e-01 -1.47360086e-01 -5.01883030e-01
-1.09603572e+00 -5.72606087e-01 -9.90525305e-01 9.45465267e-02
1.37634087e+00 -7.52243042e-01 -2.39666432e-01 5.63686788e-01
-8.90237153e-01 -3.62634242e-01 -2.58018166e-01 3.87185127e-01
-4.77829188e-01 9.10322219e-02 -1.08346367e+00 -8.40835925e-03
-5.20118356e-01 -1.63684797e+00 1.14194238e+00 7.40531206e-01
2.41149679e-01 -7.14818239e-01 -3.29718381e-01 5.91967642e-01
6.86392009e-01 2.75063336e-01 1.15003085e+00 -2.82157421e-01
-7.06811547e-01 2.33637318e-02 -9.91690516e-01 1.78560585e-01
5.57688057e-01 -3.80355716e-01 -7.37030506e-01 -2.51201630e-01
-1.91591561e-01 -5.45087047e-02 1.10673094e+00 9.70407724e-01
1.52718735e+00 5.72107472e-02 -5.46441853e-01 1.16865218e+00
1.47046876e+00 -9.11919959e-03 5.48254132e-01 3.64307970e-01
9.12272930e-01 2.64154524e-01 4.18501705e-01 1.41214252e-01
2.44605288e-01 4.81865942e-01 2.68344641e-01 -7.45551169e-01
-6.34662747e-01 1.69316262e-01 -3.00698787e-01 7.07611382e-01
-1.77343026e-01 5.14900446e-01 -7.42386103e-01 6.20502532e-01
-1.55766463e+00 -6.74959779e-01 -9.65115651e-02 1.72499919e+00
1.11958146e+00 2.53611356e-01 2.67397370e-02 -2.14302227e-01
1.00482869e+00 1.87133566e-01 -9.82171476e-01 -1.17330756e-02
-4.66533266e-02 2.18125954e-01 7.23713100e-01 4.38664109e-01
-1.36301816e+00 1.11807334e+00 5.54830408e+00 1.28663635e+00
-1.30282593e+00 3.78103077e-01 7.17911601e-01 1.28199067e-03
-1.30926728e-01 -4.27552491e-01 -9.47856367e-01 2.90508687e-01
3.14352930e-01 9.34079662e-02 4.09877986e-01 5.62287271e-01
2.05435604e-02 9.70704574e-03 -7.42702961e-01 8.55778515e-01
-9.24137533e-02 -1.51182723e+00 1.10477328e-01 8.18712115e-02
8.96818399e-01 2.41347119e-01 2.49726459e-01 4.78134491e-02
2.78368797e-02 -1.29067016e+00 2.42150515e-01 3.58107448e-01
1.24025023e+00 -8.42410684e-01 6.53607666e-01 1.54183492e-01
-1.53476632e+00 -1.83073923e-01 -6.71944201e-01 6.17931426e-01
1.13385208e-01 6.14467323e-01 -6.26212180e-01 2.79090703e-01
6.55546963e-01 4.26392496e-01 -7.23011315e-01 1.03445995e+00
4.48240228e-02 3.24094832e-01 -5.36142141e-02 4.09395039e-01
4.13734645e-01 2.82182395e-02 2.07066536e-01 1.35733008e+00
1.13909833e-01 3.93639117e-01 3.20013851e-01 1.06944239e+00
-1.16555411e-02 1.07010372e-01 1.81267202e-01 2.31580585e-01
3.18430573e-01 1.68169320e+00 -1.22038925e+00 -2.61415303e-01
-4.53296542e-01 1.00809300e+00 4.87314314e-01 3.39672029e-01
-7.60644495e-01 -3.19009364e-01 4.96810198e-01 2.77487814e-01
3.65606427e-01 6.47238567e-02 -6.06141746e-01 -9.68659103e-01
-1.62164733e-01 -7.47346520e-01 5.63407779e-01 -2.78588831e-01
-1.21767700e+00 8.70942473e-01 -4.16554689e-01 -1.06508303e+00
3.37418258e-01 -4.18210506e-01 -4.23273116e-01 1.08634579e+00
-1.74535489e+00 -1.36189699e+00 -5.70387363e-01 5.95051467e-01
7.41223097e-01 1.16589487e-01 6.00015283e-01 3.88357610e-01
-6.44289136e-01 5.36996424e-01 -5.98916784e-02 3.93570811e-01
6.23585165e-01 -1.12687290e+00 2.05262870e-01 7.32481778e-01
-4.20454144e-01 2.26544470e-01 1.03039362e-01 -7.74762988e-01
-8.08502197e-01 -1.40068376e+00 1.43959507e-01 -4.81072590e-02
5.37530184e-01 -2.84330687e-03 -1.27025616e+00 3.30175042e-01
-3.46567720e-01 5.23297608e-01 3.87320518e-01 -2.90446430e-01
-1.46030203e-01 5.42186461e-02 -1.28956532e+00 5.63929677e-01
8.97663295e-01 -5.02136767e-01 -3.51293027e-01 4.01978552e-01
9.22816575e-01 -8.93037975e-01 -1.06495166e+00 6.17107511e-01
3.41055572e-01 -7.68829405e-01 1.24053335e+00 -3.71518254e-01
5.31690955e-01 -3.46681446e-01 1.52126864e-01 -1.12468457e+00
-9.79532421e-01 2.01713338e-01 2.51040131e-01 7.91764438e-01
4.65755820e-01 -4.59069967e-01 9.62908447e-01 2.00677797e-01
-4.33514625e-01 -1.20284748e+00 -9.32724178e-01 -2.80763119e-01
2.88119912e-01 4.38605063e-02 5.97234547e-01 6.72025383e-01
-4.05402243e-01 5.48067354e-02 8.06570426e-02 5.09175301e-01
6.64059818e-01 2.96018153e-01 8.34906008e-03 -9.49548185e-01
-1.53735921e-01 -9.02165353e-01 -1.95381552e-01 -8.82405818e-01
-2.51891583e-01 -1.04105568e+00 -4.00151499e-02 -1.97608244e+00
5.22202134e-01 -5.45514286e-01 -5.22923291e-01 2.81674117e-01
-4.46652263e-01 5.42020619e-01 -2.74277925e-01 3.69643956e-01
-4.90388811e-01 -4.25654799e-02 2.13427472e+00 -1.93688944e-01
-3.68492126e-01 -6.67907484e-03 -9.01656151e-01 9.47400033e-01
8.69850159e-01 -2.33033195e-01 -2.12785378e-01 -5.39235651e-01
-2.33988121e-01 2.03068316e-01 2.49086946e-01 -9.68442678e-01
3.96033674e-01 -2.83575147e-01 7.94357955e-01 -6.32662833e-01
6.35159686e-02 -5.02848625e-01 -2.72811651e-01 5.31959832e-01
-4.99564111e-01 -4.36454207e-01 1.74915820e-01 4.30618554e-01
-9.77016315e-02 1.63491187e-03 1.33721542e+00 -3.89223784e-01
-6.04230702e-01 7.69057095e-01 -3.33048761e-01 8.64044502e-02
9.99464273e-01 -3.15592229e-01 -3.59435707e-01 2.76384413e-01
-9.53522623e-01 1.86643362e-01 3.88911724e-01 4.19395894e-01
7.59653568e-01 -1.16470087e+00 -9.17674780e-01 2.47184977e-01
-1.18130958e-02 6.10458612e-01 8.41040730e-01 1.25304770e+00
-5.68025053e-01 3.87807131e-01 -4.19365227e-01 -6.62966013e-01
-1.17877460e+00 5.68203628e-01 7.36123383e-01 -6.38230801e-01
-8.23858857e-01 9.64959383e-01 7.64466763e-01 -1.36686862e-01
8.01510811e-02 -6.45036697e-01 -5.11689305e-01 -2.53077209e-01
8.50867450e-01 1.43974528e-01 3.94751504e-02 -7.14535475e-01
-2.28919700e-01 9.08340335e-01 -6.78233564e-01 3.26828420e-01
1.11290324e+00 -2.93168992e-01 -2.04327822e-01 -3.05571314e-02
1.09113848e+00 -2.71773607e-01 -1.37796235e+00 -5.85183382e-01
-1.99606940e-01 -2.88115829e-01 4.57918018e-01 -8.59476268e-01
-1.52080607e+00 6.85541987e-01 9.13638055e-01 -5.82251586e-02
1.32979679e+00 3.02721083e-01 1.25081837e+00 -2.12394997e-01
1.66741788e-01 -1.12738299e+00 3.81055009e-03 1.66409507e-01
6.66127801e-01 -1.26889479e+00 -1.31425560e-01 -8.68890226e-01
-4.87862468e-01 1.14682853e+00 9.69517708e-01 -1.45647243e-01
4.96513188e-01 5.66456616e-01 1.21580668e-01 -1.73482493e-01
-3.36816490e-01 -4.05355245e-01 5.18322468e-01 6.63234651e-01
4.00910378e-01 2.60162294e-01 -3.23793411e-01 9.68068123e-01
1.28766343e-01 6.14281744e-02 4.11062956e-01 4.37157899e-01
-8.15279245e-01 -8.27259064e-01 -4.71673459e-01 8.31033945e-01
-6.11232698e-01 -1.12778656e-01 4.03332524e-02 5.62769473e-01
4.52187777e-01 5.67633092e-01 1.53584421e-01 -1.60528362e-01
2.72742778e-01 -5.64737141e-01 4.76621538e-01 -7.80325592e-01
-5.26050329e-01 3.79500240e-01 -3.74972433e-01 -5.75638175e-01
-1.08415879e-01 -3.95372927e-01 -1.81335139e+00 1.64823905e-01
-2.41784304e-01 -2.00815514e-01 2.96681255e-01 9.45156991e-01
7.12729692e-02 1.18626678e+00 4.17287529e-01 -7.30141759e-01
-3.10507447e-01 -8.20050955e-01 -6.77393019e-01 2.01005220e-01
3.61400604e-01 -4.10707980e-01 1.90462694e-02 6.75551593e-02]
|
[14.633111000061035, -2.570448637008667]
|
e0c9464d-1d37-4fe7-a06e-8d8a411c5134
|
combining-explicit-and-implicit-1
|
2306.00342
| null |
https://arxiv.org/abs/2306.00342v1
|
https://arxiv.org/pdf/2306.00342v1.pdf
|
Combining Explicit and Implicit Regularization for Efficient Learning in Deep Networks
|
Works on implicit regularization have studied gradient trajectories during the optimization process to explain why deep networks favor certain kinds of solutions over others. In deep linear networks, it has been shown that gradient descent implicitly regularizes toward low-rank solutions on matrix completion/factorization tasks. Adding depth not only improves performance on these tasks but also acts as an accelerative pre-conditioning that further enhances this bias towards low-rankedness. Inspired by this, we propose an explicit penalty to mirror this implicit bias which only takes effect with certain adaptive gradient optimizers (e.g. Adam). This combination can enable a degenerate single-layer network to achieve low-rank approximations with generalization error comparable to deep linear networks, making depth no longer necessary for learning. The single-layer network also performs competitively or out-performs various approaches for matrix completion over a range of parameter and data regimes despite its simplicity. Together with an optimizer's inductive bias, our findings suggest that explicit regularization can play a role in designing different, desirable forms of regularization and that a more nuanced understanding of this interplay may be necessary.
|
['Dan Zhao']
|
2023-06-01
|
combining-explicit-and-implicit
|
https://papers.nips.cc/paper_files/paper/2022/hash/1419d8554191a65ea4f2d8e1057973e4-Abstract-Conference.html
|
https://papers.nips.cc/paper_files/paper/2022/file/1419d8554191a65ea4f2d8e1057973e4-Paper-Conference.pdf
|
neurips-2022-11
|
['matrix-completion']
|
['methodology']
|
[-7.45893046e-02 3.03111494e-01 -3.80571038e-01 -5.19809961e-01
-3.79891574e-01 -3.22717726e-01 6.37639463e-01 1.24445245e-01
-6.82890415e-01 5.54396451e-01 6.80349469e-01 -4.01687354e-01
-2.60708153e-01 -5.70302904e-01 -8.64554763e-01 -6.61770105e-01
-9.17680487e-02 2.92166799e-01 -2.93244153e-01 -3.83841962e-01
9.37031209e-02 5.52273929e-01 -1.09495664e+00 4.58911359e-01
8.40301216e-01 4.20552582e-01 3.40814069e-02 3.65766466e-01
-6.87241256e-02 8.48037839e-01 -3.50087658e-02 -6.03447676e-01
6.03601575e-01 -8.76082480e-02 -6.58182681e-01 -5.57851419e-02
6.00801766e-01 -3.13277632e-01 -4.82932389e-01 9.33222890e-01
2.92233050e-01 1.94922134e-01 7.31171906e-01 -7.18926311e-01
-7.60688066e-01 9.08177555e-01 -5.04026830e-01 1.55758113e-01
-7.90007785e-02 3.67494822e-01 1.47935998e+00 -1.13207197e+00
4.53440189e-01 1.25362957e+00 1.11541688e+00 5.37167072e-01
-1.82833242e+00 -4.37186509e-01 4.10536677e-01 -3.18672359e-01
-1.08640575e+00 -5.63596547e-01 7.55428910e-01 -3.37427497e-01
6.19543195e-01 1.79608703e-01 6.93417609e-01 1.06982863e+00
4.37455773e-02 8.71646225e-01 1.08180094e+00 -1.51899919e-01
-4.02518660e-02 2.77227581e-01 3.29825550e-01 6.57581866e-01
5.32449424e-01 5.77039160e-02 -4.58328187e-01 -1.94277257e-01
7.23229051e-01 2.18069311e-02 -3.43220145e-01 -5.44153988e-01
-9.24070537e-01 9.55660164e-01 7.28793323e-01 3.40497792e-01
-4.94341552e-01 4.39294606e-01 4.48153466e-01 5.61265945e-01
4.47999835e-01 1.05036819e+00 -4.20900404e-01 7.72095248e-02
-8.53641689e-01 2.64898628e-01 8.01325321e-01 4.15081590e-01
1.14302158e+00 2.59056330e-01 -2.05598548e-01 8.98572624e-01
2.03073114e-01 1.37973586e-02 1.70914695e-01 -9.68170166e-01
5.32289743e-01 6.65304601e-01 9.57628340e-02 -1.09966028e+00
-6.52191520e-01 -1.02262545e+00 -9.66497481e-01 3.79174411e-01
7.78560638e-01 -2.78352618e-01 -6.44533515e-01 2.02570987e+00
9.56504419e-02 -6.14458472e-02 -1.98278606e-01 1.36561906e+00
2.94894248e-01 4.43909734e-01 1.38320237e-01 -1.88200511e-02
1.00706506e+00 -7.20021367e-01 -3.70358467e-01 -3.41838896e-01
9.07681286e-01 -5.04936576e-01 1.66724908e+00 3.74555260e-01
-1.06381774e+00 -1.81962371e-01 -1.11422122e+00 -2.89383650e-01
-2.04964310e-01 2.91133076e-01 1.16793597e+00 6.17577255e-01
-1.14948368e+00 1.07867253e+00 -8.52932990e-01 -1.32581919e-01
3.91510189e-01 5.94418228e-01 -3.15066963e-01 -1.85942259e-02
-1.11554682e+00 8.86487544e-01 -4.28400971e-02 4.85525638e-01
-7.31241882e-01 -1.12935293e+00 -4.62174982e-01 1.05664812e-01
1.17446966e-01 -9.61605430e-01 7.57196903e-01 -1.50968552e+00
-1.41672647e+00 8.08544159e-01 1.23952199e-02 -6.54500008e-01
7.05544651e-01 -6.43801868e-01 1.29278436e-01 -6.06612153e-02
-2.18407735e-01 5.80161273e-01 1.08910584e+00 -1.22473633e+00
5.58090024e-02 -4.07529175e-01 4.50508147e-01 2.98758894e-01
-7.36372471e-01 -3.60353321e-01 -1.47625402e-01 -5.97953022e-01
1.71293736e-01 -1.08941936e+00 -5.86878717e-01 -1.03690363e-02
-2.96741903e-01 -1.27623469e-01 2.80959338e-01 -5.22023618e-01
1.26248682e+00 -1.85916376e+00 4.76771027e-01 4.52521980e-01
5.72184205e-01 2.83585399e-01 -4.12734836e-01 5.73900342e-01
-1.79962173e-01 2.21236885e-01 -1.68021530e-01 -6.30457103e-01
-9.19870734e-02 1.40157714e-01 -3.65137368e-01 7.67051816e-01
3.23028266e-01 8.07157695e-01 -8.44050050e-01 1.94101125e-01
1.21550299e-02 6.87779963e-01 -1.05044663e+00 -1.57361373e-01
-1.81204885e-01 5.55209279e-01 -4.32832330e-01 5.26423827e-02
5.70905924e-01 -3.56196761e-01 3.21763337e-01 -4.40195113e-01
-1.61124453e-01 5.90251267e-01 -1.18225956e+00 1.61994541e+00
-7.10621715e-01 8.12846184e-01 5.45489848e-01 -1.24984336e+00
6.12247229e-01 1.26896739e-01 2.58612305e-01 -4.57029760e-01
1.56897932e-01 3.51627529e-01 2.69340098e-01 -3.65478992e-01
5.92787564e-01 -2.97812790e-01 4.93240505e-01 6.22409046e-01
-2.53048062e-01 1.18272498e-01 2.79385865e-01 3.85678828e-01
1.07073772e+00 7.20080063e-02 -2.57360339e-01 -7.05847025e-01
5.83296895e-01 -1.53875470e-01 5.53397357e-01 8.80803108e-01
2.81986564e-01 5.23343921e-01 6.16426349e-01 -6.08095050e-01
-1.20566833e+00 -7.18945265e-01 -1.09277181e-01 1.51587021e+00
-2.11262226e-01 -6.04805648e-01 -4.75979984e-01 -3.36932421e-01
2.63451725e-01 4.77530718e-01 -7.72521019e-01 -4.21506226e-01
-7.98826516e-01 -9.84306931e-01 5.51849306e-01 3.25144440e-01
1.88217685e-01 -6.59409404e-01 -1.61864050e-02 5.43739796e-02
2.00274900e-01 -6.71235144e-01 -3.38916808e-01 4.37103838e-01
-1.30840993e+00 -8.53914142e-01 -6.49900317e-01 -4.85082835e-01
1.07913399e+00 2.01601222e-01 1.29215693e+00 3.23237002e-01
1.67688087e-01 3.28505993e-01 -6.67957142e-02 7.21530337e-03
-3.91195863e-01 4.57146913e-01 2.58317798e-01 1.14400767e-01
7.23243505e-02 -1.15882814e+00 -8.17537427e-01 1.38372928e-01
-9.38690245e-01 -3.72192971e-02 8.10873985e-01 9.68384743e-01
7.17344433e-02 -3.74901950e-01 5.66620886e-01 -1.38467288e+00
1.15431786e+00 -4.82572883e-01 -5.29872477e-01 -2.91378926e-02
-8.49462628e-01 6.97262645e-01 8.72677684e-01 -6.27761662e-01
-9.01698828e-01 1.99597217e-02 -8.13676640e-02 -5.50283909e-01
4.22907203e-01 7.46055961e-01 1.79468989e-01 -1.07303031e-01
1.24008501e+00 -9.89188626e-02 2.16244206e-01 -6.49694741e-01
6.74507320e-01 1.37300074e-01 2.64352094e-02 -8.97347987e-01
9.97517228e-01 6.34893656e-01 8.60159621e-02 -9.20198560e-01
-1.15959191e+00 -2.09306359e-01 -4.83411908e-01 1.49142131e-01
5.50500631e-01 -1.09476209e+00 -6.36097193e-01 4.34674323e-02
-9.50489759e-01 -6.03717089e-01 -3.09665442e-01 5.75251460e-01
-1.92505002e-01 1.20209940e-01 -8.52453291e-01 -6.28040314e-01
-1.63349673e-01 -9.93373036e-01 6.59522712e-01 -1.01471893e-01
-4.26132798e-01 -1.26490247e+00 1.11147106e-01 1.92167476e-01
5.85302293e-01 -1.69905633e-01 8.40915918e-01 -3.74726236e-01
-6.03678703e-01 -1.39295990e-02 -2.14743242e-01 5.97280979e-01
-1.12202190e-01 -8.53294581e-02 -9.50275242e-01 -4.86521661e-01
2.59795897e-02 -4.29695725e-01 1.23105574e+00 4.16446090e-01
8.89711380e-01 -5.17658234e-01 -4.60462552e-03 8.59851241e-01
1.17150080e+00 -7.22830653e-01 5.01503170e-01 2.29560331e-01
9.60501015e-01 6.56033933e-01 1.17320120e-01 2.50187159e-01
1.64594814e-01 4.49034274e-01 3.38763565e-01 -3.79939914e-01
3.69853713e-02 -2.61618465e-01 4.56709296e-01 6.92796946e-01
-1.84778661e-01 4.08066422e-01 -7.51306713e-01 3.06856930e-01
-1.85109103e+00 -7.41708994e-01 -2.89117277e-01 2.25881577e+00
1.01767182e+00 2.58497357e-01 1.96097910e-01 -3.25631239e-02
4.11657244e-01 4.08860832e-01 -6.04369402e-01 -4.03007030e-01
-2.12230176e-01 4.40714471e-02 6.38774216e-01 7.11774468e-01
-9.08340454e-01 9.24523234e-01 6.40542936e+00 5.70543587e-01
-1.46020675e+00 1.87325254e-02 7.70703435e-01 -2.59993553e-01
-7.60080218e-01 1.91613868e-01 -5.20001054e-01 5.26576564e-02
7.96568394e-01 1.68751955e-01 8.06477726e-01 6.59161985e-01
4.92775053e-01 7.57752880e-02 -1.42150164e+00 7.41763711e-01
-3.99747878e-01 -1.53687930e+00 1.30478024e-01 2.64724076e-01
8.13243032e-01 3.27952802e-01 3.21323961e-01 3.84781569e-01
4.59021926e-01 -1.24854028e+00 6.51313901e-01 4.72823858e-01
5.05002320e-01 -6.40744090e-01 3.32590848e-01 4.58752453e-01
-7.60369956e-01 -2.72521079e-01 -6.37535870e-01 -4.56964791e-01
3.34430635e-02 1.07650709e+00 -6.61292434e-01 1.17265014e-02
3.19441378e-01 7.60257363e-01 -4.57298487e-01 6.91436589e-01
-4.32799190e-01 7.74467230e-01 -4.49635029e-01 1.92164421e-01
4.39018339e-01 -6.73978925e-01 8.56196761e-01 1.26511788e+00
-1.57108009e-01 5.15964590e-02 -4.33466025e-02 1.14137399e+00
-3.05454493e-01 1.40806228e-01 -6.48162961e-01 -3.21177095e-01
1.25551596e-01 1.41471291e+00 -5.29007196e-01 -5.51369600e-02
-2.90477008e-01 7.64750361e-01 8.84288967e-01 7.02380121e-01
-5.96995473e-01 1.58487875e-02 9.91317689e-01 5.10902345e-01
1.06371373e-01 -5.44039667e-01 -6.98698282e-01 -1.36218357e+00
1.20846387e-02 -8.79478216e-01 2.01307833e-02 -3.50551337e-01
-1.19545269e+00 3.58153433e-01 -4.22586501e-01 -8.34811687e-01
9.56045315e-02 -5.84839165e-01 -6.40839875e-01 8.23682129e-01
-1.36790359e+00 -9.23020422e-01 9.38540325e-02 5.46364367e-01
1.29762739e-01 1.11044891e-01 3.50023091e-01 5.53307712e-01
-7.11278915e-01 6.41561270e-01 1.29650742e-01 2.15606857e-02
6.68451786e-01 -1.19190836e+00 7.75236487e-02 7.56876051e-01
3.23798418e-01 1.42526150e+00 1.06551576e+00 -4.96769994e-01
-1.75928938e+00 -9.70989704e-01 4.80821967e-01 -6.29680395e-01
8.60140979e-01 -4.25940663e-01 -1.04346430e+00 6.17440701e-01
-7.43198395e-03 -6.62439838e-02 4.40051734e-01 9.23982322e-01
-6.63952947e-01 -3.11676502e-01 -7.52289176e-01 1.01934004e+00
1.22787249e+00 -7.91529775e-01 -3.01015973e-01 4.47527826e-01
5.92264056e-01 -2.05693230e-01 -6.46160781e-01 2.14152947e-01
3.87941599e-01 -1.04184020e+00 1.09111249e+00 -9.26721931e-01
5.73504806e-01 -1.25046968e-01 7.20244050e-02 -1.34301758e+00
-5.52483141e-01 -8.81211102e-01 -8.46198723e-02 9.88396287e-01
7.50611603e-01 -7.51505911e-01 1.01745319e+00 8.90606165e-01
-2.43518695e-01 -1.07807624e+00 -5.25228441e-01 -6.83934987e-01
3.86770546e-01 -5.04412770e-01 1.23601824e-01 1.08803487e+00
-1.77818179e-01 5.18420517e-01 -6.00120306e-01 3.19917593e-03
3.83512616e-01 -2.56718636e-01 8.29837978e-01 -1.08827615e+00
-2.89002478e-01 -5.54958522e-01 7.37436563e-02 -1.31706405e+00
3.73019427e-01 -1.31000006e+00 -3.93271118e-01 -1.17049825e+00
-7.05825165e-02 -8.84620011e-01 -2.68077701e-01 3.75728130e-01
-1.35631099e-01 2.32862234e-01 2.40436479e-01 4.38744515e-01
-3.51780355e-01 6.04393125e-01 1.20159209e+00 3.55181023e-02
-5.60108423e-01 7.28821196e-03 -1.13926220e+00 7.01099396e-01
6.25133932e-01 -4.93034065e-01 -4.87371564e-01 -6.66538775e-01
9.32866871e-01 -3.99328381e-01 1.27482265e-01 -7.27887869e-01
9.29473490e-02 -1.06126256e-02 2.54095644e-01 1.93013892e-01
2.99819976e-01 -7.46605456e-01 -4.11629155e-02 4.51247245e-01
-7.16497302e-01 -4.39977013e-02 3.16619663e-03 5.24577260e-01
2.07816251e-02 -3.22307616e-01 7.21687138e-01 -1.53787956e-01
-2.17067808e-01 2.78806448e-01 -4.73040402e-01 3.24768096e-01
2.57082194e-01 -1.33063152e-01 7.45685995e-02 -4.96491343e-01
-8.11028540e-01 2.05928236e-01 4.67531323e-01 3.02749813e-01
2.35171959e-01 -1.22609460e+00 -7.58222222e-01 8.33739713e-02
-3.70413691e-01 -2.47494251e-01 -4.44802083e-02 1.35753286e+00
-3.41257215e-01 2.89918363e-01 1.73344463e-01 -5.49279332e-01
-8.31920087e-01 1.80133522e-01 4.07443732e-01 -4.80949432e-01
-6.82860911e-01 9.55598712e-01 3.49367589e-01 -6.14623725e-01
1.44361764e-01 -3.77948076e-01 8.86346959e-03 2.10784540e-01
2.72529423e-01 3.23676407e-01 -5.96443284e-03 -2.92145818e-01
-1.36502579e-01 3.34427536e-01 -3.19988608e-01 -2.84723775e-03
1.59120917e+00 8.23603757e-03 -2.58913249e-01 4.28448498e-01
1.26867819e+00 3.43884796e-01 -1.49232471e+00 -2.85689682e-01
9.95683372e-02 -2.87303954e-01 7.53868595e-02 -4.19867903e-01
-1.22696269e+00 9.15698171e-01 1.32549688e-01 2.85800714e-02
7.28333473e-01 -3.21014822e-01 4.01453435e-01 7.35673726e-01
1.05157182e-01 -1.16413510e+00 2.33785793e-01 7.39005089e-01
1.06483722e+00 -1.36608219e+00 3.78641754e-01 -1.02548592e-01
-6.31126523e-01 1.05053329e+00 4.28858906e-01 -6.38061821e-01
5.95392883e-01 2.43868120e-03 -1.00456506e-01 -2.08125859e-01
-7.74477184e-01 5.23803756e-02 4.17903841e-01 2.73380995e-01
8.56638193e-01 -5.09686098e-02 -5.00397801e-01 3.09891313e-01
-1.85571715e-01 -2.52786070e-01 4.21299458e-01 4.66026872e-01
-2.58317202e-01 -1.27275777e+00 -2.19396397e-01 6.23067677e-01
-5.08022189e-01 -3.36640954e-01 -3.58086914e-01 7.60337472e-01
-1.89082876e-01 5.44319987e-01 -1.22042991e-01 -2.88831115e-01
1.47439986e-01 -2.54558120e-02 4.44797605e-01 -6.43850982e-01
-9.90627944e-01 -2.36609206e-01 2.96558768e-01 -6.82203770e-01
-1.44629553e-01 -5.59698403e-01 -1.13928282e+00 -4.46588695e-01
-1.67575002e-01 2.81931192e-01 5.32693982e-01 7.76540577e-01
4.53301549e-01 2.63765872e-01 4.28216547e-01 -1.02277112e+00
-9.29482698e-01 -7.87164211e-01 -3.42876524e-01 6.05342269e-01
5.42305827e-01 -5.27082622e-01 -7.16654062e-01 -3.42543513e-01]
|
[8.389641761779785, 3.629164457321167]
|
1ffc946c-47d0-4f8b-92c8-c8a27c9784a3
|
flexible-android-malware-detection-model
|
2210.14225
| null |
https://arxiv.org/abs/2210.14225v1
|
https://arxiv.org/pdf/2210.14225v1.pdf
|
Flexible Android Malware Detection Model based on Generative Adversarial Networks with Code Tensor
|
The behavior of malware threats is gradually increasing, heightened the need for malware detection. However, existing malware detection methods only target at the existing malicious samples, the detection of fresh malicious code and variants of malicious code is limited. In this paper, we propose a novel scheme that detects malware and its variants efficiently. Based on the idea of the generative adversarial networks (GANs), we obtain the `true' sample distribution that satisfies the characteristics of the real malware, use them to deceive the discriminator, thus achieve the defense against malicious code attacks and improve malware detection. Firstly, a new Android malware APK to image texture feature extraction segmentation method is proposed, which is called segment self-growing texture segmentation algorithm. Secondly, tensor singular value decomposition (tSVD) based on the low-tubal rank transforms malicious features with different sizes into a fixed third-order tensor uniformly, which is entered into the neural network for training and learning. Finally, a flexible Android malware detection model based on GANs with code tensor (MTFD-GANs) is proposed. Experiments show that the proposed model can generally surpass the traditional malware detection model, with a maximum improvement efficiency of 41.6\%. At the same time, the newly generated samples of the GANs generator greatly enrich the sample diversity. And retraining malware detector can effectively improve the detection efficiency and robustness of traditional models.
|
['Linxi Han', 'Fengyang Deng', 'Zhao Yang']
|
2022-10-25
| null | null | null | null |
['android-malware-detection']
|
['miscellaneous']
|
[ 2.31526002e-01 -5.57972848e-01 -2.01973483e-01 1.49707600e-01
-2.86701739e-01 -5.87526739e-01 4.60648775e-01 -7.60703444e-01
1.09238520e-01 1.79785788e-01 -8.25351253e-02 -3.59965295e-01
3.55941862e-01 -9.33979094e-01 -4.18423951e-01 -9.90119338e-01
-4.46593165e-02 2.24682257e-01 4.18034405e-01 -2.79368639e-01
1.86354995e-01 3.35910738e-01 -1.12215269e+00 4.29252177e-01
1.01958382e+00 1.05585706e+00 -8.83032084e-02 6.99801862e-01
-3.81392181e-01 7.79852033e-01 -8.18294883e-01 -4.91915226e-01
3.81197035e-01 -5.45983016e-01 -4.08003122e-01 8.03797320e-02
-2.23946691e-01 -6.74939871e-01 -5.71401715e-01 1.49910784e+00
1.74397260e-01 -2.55158842e-01 7.37670600e-01 -1.46256733e+00
-9.55098331e-01 5.48308492e-01 -9.75655019e-01 2.23616183e-01
1.16103880e-01 2.91259617e-01 3.77253115e-01 -4.92314845e-01
3.06800723e-01 1.51104987e+00 7.88859487e-01 9.72287238e-01
-8.30587268e-01 -9.62127805e-01 1.00744896e-01 1.05494894e-01
-1.20920694e+00 -2.61833612e-02 1.04612041e+00 -5.24982631e-01
4.26147014e-01 5.74372649e-01 8.04032147e-01 1.63455331e+00
3.66887331e-01 8.68680894e-01 1.08868814e+00 3.30038309e-01
-7.25246072e-02 1.24756554e-02 -3.54380012e-02 8.20952177e-01
3.92312437e-01 1.37205526e-01 3.45334947e-01 -4.10115242e-01
6.13117754e-01 7.48743474e-01 -1.78986162e-01 6.13027476e-02
-7.73269832e-01 1.10173059e+00 4.87137228e-01 4.20941174e-01
-9.80508998e-02 -4.10489477e-02 7.82185316e-01 1.20443948e-01
3.64276767e-01 4.67011407e-02 -3.01463157e-01 -1.45483285e-01
-7.38027036e-01 -4.20952812e-02 6.18635356e-01 6.32983327e-01
5.74667037e-01 6.90677941e-01 1.12867523e-02 5.55208623e-01
5.70248187e-01 9.83646393e-01 1.07404470e+00 -4.61491555e-01
1.60615563e-01 9.85085905e-01 -6.37525558e-01 -1.58348370e+00
8.92186388e-02 -3.27743739e-01 -1.04999673e+00 -6.50582239e-02
-1.37350366e-01 -2.70990759e-01 -1.01789021e+00 1.27939963e+00
4.47030008e-01 6.14054263e-01 1.01893865e-01 4.44932103e-01
6.07425869e-01 1.04139054e+00 -2.29852587e-01 -4.21760112e-01
1.30842125e+00 -6.66689992e-01 -5.18965960e-01 -3.22784893e-02
4.50994045e-01 -7.35179842e-01 1.12268817e+00 5.46397924e-01
-2.00296745e-01 -4.15586531e-01 -9.85183120e-01 4.88725007e-01
-2.83904731e-01 1.74404025e-01 6.81320846e-01 1.20369530e+00
-6.68188393e-01 5.94984517e-02 -8.52215469e-01 8.91609266e-02
7.80936241e-01 4.08829927e-01 1.90831814e-02 -1.27572671e-01
-9.35091913e-01 5.27349412e-02 5.14189482e-01 1.21596232e-01
-1.41460359e+00 -2.75263280e-01 -5.83602250e-01 -2.51013815e-01
4.87155259e-01 -2.06107885e-01 1.02428877e+00 -1.20366347e+00
-1.59642243e+00 3.64384055e-01 1.60147101e-01 -2.52232581e-01
4.73923713e-01 1.80685505e-01 -7.29540944e-01 1.25620617e-02
4.69693765e-02 1.56645074e-01 1.39921355e+00 -1.38065946e+00
-3.46722543e-01 -4.06044036e-01 -1.43186778e-01 -1.62989095e-01
-6.46204412e-01 1.12120964e-01 -1.35834634e-01 -6.97874129e-01
-7.68118352e-02 -1.07377172e+00 -5.55706173e-02 -6.65189981e-01
-6.33543968e-01 8.94473046e-02 1.85537601e+00 -1.05475366e+00
1.19633484e+00 -2.34866095e+00 2.09449381e-01 4.43733871e-01
4.88791972e-01 5.69101334e-01 1.33604864e-02 -7.33681023e-02
1.52106032e-01 4.55869406e-01 -4.65227008e-01 1.35213435e-01
-3.71623099e-01 4.33220536e-01 -4.79251802e-01 3.39088529e-01
8.49740654e-02 9.46146250e-01 -7.41038203e-01 -5.32845736e-01
-6.36971742e-02 4.83352512e-01 -6.43112123e-01 1.11493476e-01
-3.24295253e-01 5.05785406e-01 -8.77820253e-01 7.96200097e-01
9.65117097e-01 1.14288993e-01 -7.17210472e-02 -9.35917196e-04
3.18300486e-01 -2.95421928e-01 -8.79294753e-01 8.31407845e-01
-2.64348298e-01 1.43244430e-01 1.22836210e-01 -9.11742926e-01
1.14375818e+00 -1.09114729e-01 5.93064010e-01 -2.37394035e-01
5.53034663e-01 2.77319968e-01 3.58921647e-01 -7.14885235e-01
6.64213672e-02 1.79446101e-01 -5.41108586e-02 3.77776772e-01
-2.64964759e-01 1.79359376e-01 -1.73774868e-01 1.82389528e-01
1.13007069e+00 -1.02947734e-01 -2.25827515e-01 -4.40596007e-02
9.84820724e-01 -4.23915684e-02 5.61589241e-01 1.29398704e-01
-6.04043249e-04 -9.79938060e-02 8.80604267e-01 -2.72114575e-01
-8.83734345e-01 -9.82867360e-01 1.73708692e-01 8.77080858e-01
8.80704150e-02 -2.91229963e-01 -1.24601030e+00 -1.35767889e+00
-2.38713160e-01 2.50592530e-01 -7.40343869e-01 -7.28290617e-01
-6.79002404e-01 -1.08845401e+00 8.60960782e-01 2.75147527e-01
1.07569826e+00 -1.05259907e+00 -2.74684373e-02 -1.90782741e-01
-4.30489890e-02 -7.58885384e-01 -5.77705979e-01 -5.15454352e-01
-8.46222818e-01 -1.14959288e+00 -4.68105376e-01 -7.73436427e-01
7.81886220e-01 3.09340388e-01 3.13772500e-01 4.28400725e-01
-1.08428210e-01 2.18334481e-01 -5.49242258e-01 -2.11354986e-01
-8.72424901e-01 9.20308605e-02 1.39813378e-01 4.94538933e-01
2.93709785e-01 -3.99016082e-01 -2.83935338e-01 2.26831928e-01
-1.09367371e+00 -4.45936650e-01 8.01972628e-01 1.02878141e+00
3.08012962e-01 5.42484403e-01 2.35019267e-01 -9.35431957e-01
8.39219034e-01 -7.10752547e-01 -4.00954753e-01 -6.67013898e-02
-5.70790827e-01 -2.70511240e-01 1.17785060e+00 -1.08576083e+00
-1.03795648e+00 -1.47444367e-01 -3.47718149e-01 -8.45030665e-01
1.67730093e-01 2.87942261e-01 -6.21681213e-01 -5.04334688e-01
6.39567137e-01 8.57351601e-01 2.98658550e-01 -2.35190719e-01
2.11574018e-01 7.69354224e-01 2.25254416e-01 -4.38449502e-01
1.19557738e+00 3.34326744e-01 -7.14034587e-02 -7.20513284e-01
-1.71979830e-01 4.67444919e-02 -1.60364956e-01 -1.58985868e-01
9.11303222e-01 -4.34513390e-01 -8.49058211e-01 9.14637446e-01
-1.04175591e+00 -1.71559006e-02 4.56916094e-02 2.64810920e-01
-3.41669656e-02 6.26856506e-01 -9.61434603e-01 -7.30350673e-01
-4.61754203e-01 -1.58932269e+00 9.54166532e-01 1.59463063e-01
5.64117551e-01 -8.17659020e-01 -8.84151384e-02 2.26634085e-01
4.23150480e-01 5.62564313e-01 8.46063852e-01 -8.61161292e-01
-5.61525941e-01 -4.36764807e-01 -1.21967427e-01 7.52048612e-01
1.96486384e-01 4.38363373e-01 -7.44428098e-01 -2.92269170e-01
7.91372299e-01 4.90800813e-02 8.03353131e-01 8.15044492e-02
1.46101224e+00 -1.04351234e+00 -4.58063304e-01 9.29017663e-01
1.26122999e+00 7.90738285e-01 6.80802882e-01 6.90287650e-02
1.46160758e+00 2.81142801e-01 3.08079243e-01 1.48149848e-01
1.13403499e-01 1.85872421e-01 7.05798626e-01 1.58305526e-01
2.48462230e-01 -3.25415999e-01 7.79637873e-01 1.22559845e+00
-8.30801427e-02 1.08197153e-01 -7.65401065e-01 -7.99974427e-02
-1.40053678e+00 -1.14808798e+00 -2.63890654e-01 1.80581784e+00
6.23844922e-01 2.95759141e-01 2.94811636e-01 2.93251663e-01
7.58143902e-01 1.79100737e-01 -7.06499398e-01 -5.02322197e-01
2.35889524e-01 6.37763590e-02 5.17887056e-01 2.63088316e-01
-1.05024350e+00 9.27438736e-01 5.70734930e+00 1.51676011e+00
-1.49296057e+00 5.09700835e-01 7.13624299e-01 4.39869225e-01
-3.76399308e-01 -1.01563685e-01 -6.38656318e-01 1.03953314e+00
7.31922746e-01 4.33413982e-02 8.66114080e-01 1.34782135e+00
-1.04837254e-01 6.71721637e-01 -3.25041354e-01 9.88171875e-01
3.49828839e-01 -7.83257186e-01 2.96992064e-01 4.58753824e-01
6.53534234e-01 -2.76652932e-01 5.69569349e-01 6.38914466e-01
2.17754021e-01 -9.31221187e-01 1.34159699e-01 2.89313376e-01
8.80166054e-01 -1.08228004e+00 8.27398896e-01 3.12435478e-01
-1.35727465e+00 -5.82276046e-01 -4.93269622e-01 3.18218499e-01
-4.73271936e-01 4.48282570e-01 -8.25386703e-01 5.57608485e-01
3.55089068e-01 7.96611130e-01 -8.69288862e-01 4.56203789e-01
2.88847685e-02 9.69553649e-01 -1.39621660e-01 -2.51672506e-01
3.28765482e-01 -3.85003209e-01 5.94547808e-01 9.91320968e-01
4.34836268e-01 -2.50136346e-01 5.84663451e-01 7.54083276e-01
-3.92252840e-02 3.12845856e-01 -7.96257496e-01 -3.58845264e-01
2.98554450e-01 1.43496323e+00 -8.50885928e-01 -2.84633547e-01
-5.25613129e-02 9.59936619e-01 -1.58382729e-01 1.29458800e-01
-1.24480247e+00 -4.33166981e-01 4.21916246e-01 -1.06243752e-02
1.45411044e-01 -1.23248987e-01 -9.92920846e-02 -1.30723906e+00
-2.03463268e-02 -1.49958646e+00 2.22896248e-01 -2.12606579e-01
-1.32119095e+00 7.18614101e-01 -2.32877880e-01 -1.33262730e+00
-2.00902402e-01 -8.51804912e-01 -9.11511421e-01 4.49958980e-01
-7.42787421e-01 -1.58897758e+00 -4.00681406e-01 8.97291183e-01
3.84092599e-01 -6.71755672e-01 5.56111395e-01 3.82483304e-01
-8.62525940e-01 8.08423400e-01 2.95442998e-01 4.72893655e-01
-3.71820740e-02 -7.77293682e-01 2.72935957e-01 1.09610808e+00
-2.08825290e-01 7.44527161e-01 2.07395509e-01 -1.18459535e+00
-1.61089110e+00 -1.37582719e+00 -3.19526404e-01 -2.83825636e-01
7.90124953e-01 -4.81137633e-01 -9.81090307e-01 7.60025918e-01
-2.14156568e-01 -2.01789632e-01 5.43333054e-01 -4.97615218e-01
-6.67684078e-01 -3.02805990e-01 -1.58460689e+00 7.01433480e-01
8.88191760e-01 -4.35225636e-01 -1.83894113e-01 4.39276785e-01
1.08379340e+00 -4.01691616e-01 -6.10495269e-01 4.55846012e-01
3.26560974e-01 -8.02581251e-01 1.02614665e+00 -4.63747352e-01
3.80636483e-01 -3.71858925e-01 -2.83573885e-02 -8.29671741e-01
-2.48663962e-01 -5.30793130e-01 -4.76438582e-01 1.19330013e+00
-9.76067111e-02 -8.98562431e-01 8.22411060e-01 -2.37368867e-01
-1.33705437e-01 -1.01754451e+00 -9.26919639e-01 -8.21139097e-01
-7.14088678e-02 -1.95060953e-01 9.01061058e-01 1.18970847e+00
-5.65120280e-01 1.63704038e-01 -4.57075119e-01 -2.54207589e-02
4.71332699e-01 -3.51689309e-01 7.82264054e-01 -1.13657105e+00
-4.82487142e-01 -6.77911162e-01 -6.25276506e-01 -9.03563321e-01
1.60442799e-01 -1.02086973e+00 -2.21509621e-01 -8.48658144e-01
2.34964803e-01 -3.23589474e-01 -9.55970585e-03 2.84230739e-01
-2.16930360e-01 3.26130718e-01 -1.01627834e-01 4.15151834e-01
-1.72004908e-01 5.32811701e-01 1.50015938e+00 -4.48675185e-01
-2.92364001e-01 1.60676420e-01 -7.34921336e-01 8.08092892e-01
9.42559838e-01 -4.26937670e-01 -5.99051833e-01 1.95272248e-02
-2.00276330e-01 -2.94460267e-01 3.20720524e-01 -9.11195040e-01
-3.33273470e-01 -2.96056360e-01 4.98820782e-01 -2.76326776e-01
2.20836997e-01 -8.09351027e-01 9.47019830e-02 1.17567515e+00
2.67729700e-01 1.90181971e-01 9.40663517e-02 7.02495098e-01
1.55491561e-01 -2.68780559e-01 7.06668556e-01 -1.28095299e-01
-3.88782769e-01 7.58403599e-01 -4.12494212e-01 1.49254680e-01
1.30451155e+00 -3.32509011e-01 -3.95080417e-01 -9.64140706e-03
-2.49255419e-01 -1.87362283e-01 5.19574583e-01 3.63218963e-01
8.65375042e-01 -1.51124811e+00 -6.27181590e-01 4.60241348e-01
-3.42449903e-01 -1.39713079e-01 3.57960105e-01 7.50549853e-01
-7.67076254e-01 -2.93186784e-01 -3.08130026e-01 -7.81617045e-01
-1.25505674e+00 1.00115311e+00 3.16930056e-01 -4.30957139e-01
-3.39793146e-01 7.64474511e-01 7.02429861e-02 -4.02018189e-01
-3.33869368e-01 8.66192281e-02 -5.20571649e-01 -2.84011513e-01
4.81518179e-01 5.03447950e-01 -4.43368286e-01 -1.06609523e+00
-2.30282903e-01 5.67174375e-01 -3.12013566e-01 2.53589422e-01
8.84743631e-01 4.41264927e-01 -6.81524515e-01 1.03993952e-01
1.45335758e+00 5.86990714e-01 -7.85796165e-01 2.85832524e-01
-4.24020469e-01 -6.98251665e-01 -3.47567052e-01 -3.80427986e-01
-1.51142395e+00 9.16420698e-01 8.23444605e-01 7.06509352e-01
1.14261627e+00 -3.93566698e-01 1.31817436e+00 1.66743651e-01
1.92608029e-01 -3.73779237e-01 5.02489090e-01 4.32157129e-01
5.06469131e-01 -8.23116541e-01 -2.33392194e-01 -4.85524237e-01
-7.49834418e-01 1.05723703e+00 9.22307432e-01 -5.35978854e-01
4.93241012e-01 3.99004072e-01 -2.06372976e-01 -7.93327242e-02
-8.05138052e-02 3.16214859e-01 1.15269333e-01 9.87644374e-01
-2.18953013e-01 2.34333023e-01 -2.15062663e-01 7.36651957e-01
-3.63084614e-01 -5.26163578e-01 4.89324063e-01 4.96037751e-01
-4.25536394e-01 -1.11284113e+00 -5.77019334e-01 9.50373232e-01
-6.14947438e-01 -9.09227133e-02 -5.99088430e-01 4.41035688e-01
6.02810442e-01 7.88058996e-01 -2.74408698e-01 -1.43176031e+00
-9.43209305e-02 -1.47186965e-01 1.66012496e-01 -4.44530874e-01
-5.71429312e-01 5.58951013e-02 -5.79883158e-01 -2.91300207e-01
8.85111317e-02 -5.89586020e-01 -1.13392413e+00 -5.41803300e-01
-3.46204787e-01 1.99536726e-01 4.32246625e-01 7.35782385e-01
2.61314929e-01 6.90048337e-01 1.11990631e+00 -5.79895973e-01
-7.43109465e-01 -9.65100944e-01 -3.08577240e-01 3.64339352e-01
1.47809103e-01 -6.35493398e-01 -5.58667839e-01 -2.79787909e-02]
|
[14.40152645111084, 9.645242691040039]
|
cff96c2d-337b-4242-abf9-59b6e82e44bb
|
an-mrc-framework-for-semantic-role-labeling-1
| null | null |
https://openreview.net/forum?id=PgcGLPyh8f
|
https://openreview.net/pdf?id=PgcGLPyh8f
|
An MRC Framework for Semantic Role Labeling
|
Semantic Role Labeling (SRL) aims at recognizing the predicate-argument structure of a sentence and can be decomposed into two subtasks: predicate disambiguation and argument labeling. Prior work deals with these two tasks independently, which ignores the semantic connection between the two tasks. In this paper, we propose to use the machine reading comprehension (MRC) framework to bridge this gap. We formalize predicate disambiguation as multiple-choice machine reading comprehension, where the descriptions of candidate senses of a given predicate are used as options to select the correct sense. The chosen predicate sense is then used to determine the semantic roles for that predicate, and these semantic roles are used to construct the query for another MRC model for argument labeling. In this way, we are able to leverage both the predicate semantics and the semantic role semantics for argument labeling. We also propose to select a subset of all the possible semantic roles for computational efficiency. Experiments show that the proposed framework achieves state-of-the-art results on both span and dependency benchmarks.
|
['Anonymous']
|
2022-01-16
| null | null | null |
acl-arr-january-2022-1
|
['semantic-role-labeling']
|
['natural-language-processing']
|
[ 7.78192043e-01 5.18806577e-01 -4.05226469e-01 -5.48019767e-01
-7.70892382e-01 -8.87928784e-01 6.31948292e-01 8.65012288e-01
-4.85788882e-01 6.82273209e-01 6.44325018e-01 -4.74529386e-01
-2.37406775e-01 -9.96930778e-01 -4.84141022e-01 -5.07980585e-01
3.33106428e-01 5.18651128e-01 6.59337044e-01 -4.16446507e-01
4.03261364e-01 -3.18433940e-02 -1.79183519e+00 5.08834779e-01
8.78543019e-01 9.84742820e-01 4.45931882e-01 2.04468772e-01
-3.98150474e-01 9.51366067e-01 -4.78330225e-01 -2.15732947e-01
-2.07827002e-01 -5.17855465e-01 -1.53312087e+00 -1.93098277e-01
-6.38753250e-02 5.52230105e-02 3.77800733e-01 1.01922774e+00
2.26327389e-01 3.76815856e-01 1.31846935e-01 -1.07337427e+00
-1.42059326e-01 9.20534432e-01 -7.69464448e-02 3.28155309e-01
8.38861287e-01 -3.68477970e-01 1.88495421e+00 -6.85605049e-01
5.08696258e-01 1.46071887e+00 -1.39016822e-01 6.31664455e-01
-1.01932132e+00 -2.03495041e-01 6.75042152e-01 3.34751695e-01
-7.23023355e-01 -3.70134413e-01 8.83223534e-01 -1.95756406e-01
9.39604998e-01 2.87362009e-01 4.74424869e-01 6.98495388e-01
-4.84158725e-01 9.45485532e-01 1.03387940e+00 -7.68097222e-01
3.76181751e-01 -2.85967112e-01 9.16844189e-01 5.35407484e-01
8.45330395e-03 -3.90125304e-01 -6.62544012e-01 -3.64992172e-01
2.74315655e-01 -3.47853601e-01 -4.62540567e-01 -1.03551760e-01
-1.18799949e+00 9.30783629e-01 2.47389242e-01 2.99307525e-01
-3.54080379e-01 2.33347490e-02 4.12407964e-01 2.85408050e-01
2.18740627e-01 7.41082072e-01 -7.76767373e-01 -1.05146058e-01
-3.15141916e-01 4.37493324e-01 9.77664113e-01 4.91429240e-01
6.53365195e-01 -9.00352120e-01 -4.62929100e-01 9.27055061e-01
4.02318478e-01 1.40948609e-01 2.12012842e-01 -1.03240168e+00
6.41239405e-01 9.23429608e-01 3.50470275e-01 -8.00167024e-01
-4.59528655e-01 -9.71609727e-02 8.12180266e-02 -2.94793665e-01
5.53284347e-01 1.76130235e-01 -6.63546979e-01 2.11752462e+00
7.13718534e-01 2.17971355e-01 4.11549777e-01 9.78195310e-01
9.11540270e-01 5.39991140e-01 5.57308912e-01 -2.58408368e-01
1.92100394e+00 -9.39934909e-01 -5.69879055e-01 -4.95269418e-01
6.90777838e-01 -4.23180878e-01 1.25092506e+00 -7.37472102e-02
-9.86727655e-01 -1.63039044e-01 -9.67120409e-01 -2.91689515e-01
-1.26060039e-01 -5.16916439e-02 4.44508523e-01 7.42248818e-02
-7.44047344e-01 4.25724149e-01 -7.33877242e-01 -2.12560743e-01
2.01095879e-01 2.30173349e-01 -1.85832232e-01 -2.07116976e-01
-1.83341658e+00 7.53322542e-01 8.12242746e-01 -2.12431312e-01
-6.36878252e-01 -4.93957520e-01 -1.12070549e+00 2.48017192e-01
9.20825005e-01 -7.95657992e-01 1.31732047e+00 -7.68705845e-01
-1.08794796e+00 1.07098627e+00 -6.71446741e-01 -5.58973908e-01
-5.24675809e-02 -3.84348452e-01 -8.31124857e-02 3.73167545e-01
5.87079287e-01 4.79469150e-01 6.36806309e-01 -1.23453140e+00
-1.06167984e+00 -4.73113000e-01 8.93120706e-01 6.75363481e-01
1.23016194e-01 3.19633663e-01 -4.21023458e-01 -3.71425569e-01
5.66977501e-01 -7.50637591e-01 6.26697913e-02 -4.53465223e-01
-4.90449160e-01 -8.47695112e-01 5.18440247e-01 -5.38844407e-01
1.26446557e+00 -1.98906815e+00 2.72655964e-01 1.15796007e-01
3.76556277e-01 2.73410343e-02 -1.44115046e-01 3.88280898e-01
-1.84697866e-01 2.01911524e-01 -2.90344954e-01 -2.80918330e-01
3.53011303e-02 3.85633856e-01 -5.44124722e-01 1.77485775e-02
1.94521636e-01 7.77597368e-01 -1.23111868e+00 -2.41436243e-01
-1.30819455e-01 -1.28061086e-01 -3.28903705e-01 3.96031797e-01
-7.66982615e-01 5.57608902e-01 -9.97759521e-01 4.80995208e-01
4.36226696e-01 -3.58680516e-01 4.49523568e-01 -7.03919306e-02
1.60505161e-01 1.28585279e+00 -1.09625256e+00 1.46986580e+00
-3.06705147e-01 -1.78461358e-01 -1.88998808e-03 -1.03691936e+00
7.71290243e-01 8.48322436e-02 1.81760162e-01 -6.77446067e-01
-2.88763940e-02 4.14223909e-01 8.25168639e-02 -4.49950516e-01
2.11653531e-01 -2.37645924e-01 -4.00946021e-01 6.22784793e-01
-2.44740725e-01 2.32833177e-01 4.99296397e-01 1.68300837e-01
1.08208621e+00 3.15149397e-01 6.69216216e-01 -2.90584117e-01
1.00856650e+00 1.75925165e-01 7.02824712e-01 6.26798749e-01
5.07829450e-02 1.94438264e-01 9.19289768e-01 -2.10861757e-01
-4.15096313e-01 -1.02462447e+00 2.46461645e-01 1.52742684e+00
5.95646143e-01 -5.41281998e-01 -7.64659166e-01 -1.01688218e+00
-2.21797869e-01 9.77438092e-01 -5.06103575e-01 3.73787954e-02
-1.01302898e+00 -3.42463106e-01 2.56810755e-01 5.33743322e-01
4.09880996e-01 -1.32198560e+00 -1.06292951e+00 3.27075928e-01
-5.12056470e-01 -1.25682008e+00 -1.19061954e-02 3.00005347e-01
-5.08970320e-01 -1.60278141e+00 1.25022113e-01 -9.37829792e-01
4.91470754e-01 1.85754567e-01 1.36573792e+00 5.25604546e-01
4.21690494e-01 3.22197616e-01 -8.62545550e-01 -2.72460133e-01
-4.38296914e-01 8.33879188e-02 -2.63196439e-01 8.94498006e-02
4.81697530e-01 -4.21546966e-01 -5.23143530e-01 1.57397524e-01
-8.25102746e-01 1.99314550e-01 2.15939432e-01 8.21721911e-01
8.56395543e-01 1.53881818e-01 6.52536333e-01 -1.26690722e+00
7.86238670e-01 -5.29676914e-01 -4.89398390e-01 6.09211504e-01
-3.82498145e-01 4.43293452e-01 7.27383375e-01 2.07111537e-02
-1.26145649e+00 -2.16827765e-01 -2.41431341e-01 2.71990001e-01
-2.85998642e-01 8.79643917e-01 -6.51053846e-01 5.90387285e-01
2.26033688e-01 9.05394480e-02 -3.44244063e-01 -5.98184407e-01
2.71664172e-01 2.37049207e-01 4.45165277e-01 -1.16967309e+00
6.69473290e-01 3.13682258e-01 -3.65247279e-02 -3.88936102e-01
-1.68102944e+00 -6.38379574e-01 -4.80491847e-01 3.19270581e-01
1.10017502e+00 -7.87505269e-01 -6.59367323e-01 1.30771086e-01
-1.30882525e+00 -4.09366414e-02 -3.12202960e-01 7.80372322e-02
-4.68534946e-01 3.62546057e-01 -2.41567388e-01 -6.49746776e-01
-2.38037661e-01 -1.15253234e+00 1.34530377e+00 2.90084571e-01
-4.09267277e-01 -9.43941116e-01 -2.73053855e-01 6.50152266e-01
-1.50351534e-02 1.91458538e-01 1.62331367e+00 -1.26503098e+00
-3.74014169e-01 8.65023732e-02 -1.93547532e-01 6.26583993e-02
9.84625053e-03 -7.49356925e-01 -8.39250743e-01 3.68488729e-02
2.09834203e-01 -3.44165027e-01 8.91781032e-01 2.14816779e-02
1.04384720e+00 -3.25789005e-01 -3.72359037e-01 1.00705251e-01
1.20124280e+00 1.10580891e-01 4.60776001e-01 4.34240103e-01
5.98555863e-01 9.77054954e-01 1.23629892e+00 2.50008702e-01
7.40238130e-01 5.92221558e-01 4.26931053e-01 1.93821952e-01
1.29847690e-01 -5.92588186e-01 9.11250338e-02 1.99204385e-01
2.21751645e-01 -3.50471854e-01 -1.09649575e+00 5.42960167e-01
-2.04572868e+00 -7.20787764e-01 -2.00144246e-01 2.04362774e+00
9.97546494e-01 2.67733365e-01 -1.16024785e-01 3.92884493e-01
8.02056313e-01 4.76925761e-01 -4.17595267e-01 -5.16096912e-02
-6.89809769e-02 3.65988463e-01 3.62095572e-02 6.15666628e-01
-1.25948179e+00 1.16620266e+00 5.02055883e+00 7.18235552e-01
-5.00111938e-01 1.28348902e-01 4.77890581e-01 3.55397433e-01
-5.77774405e-01 5.93259513e-01 -9.35050011e-01 4.26689982e-02
5.76822460e-01 -8.06570575e-02 4.05198693e-01 5.32228053e-01
1.48086011e-01 -4.13692445e-01 -1.35721838e+00 5.19638598e-01
-3.66530381e-02 -1.16130722e+00 1.39950842e-01 -4.76210952e-01
1.16075031e-01 -4.02769595e-01 -4.54520673e-01 2.05389455e-01
1.59492880e-01 -9.55796301e-01 8.74737620e-01 1.20554544e-01
4.10020113e-01 -5.31242490e-01 6.23124659e-01 5.88527918e-01
-1.33657038e+00 -3.60009909e-01 -7.02572539e-02 -2.99328446e-01
2.29074985e-01 3.13467652e-01 -6.22059226e-01 5.71046650e-01
3.89205217e-01 5.45290291e-01 -2.80086219e-01 4.83452111e-01
-1.21651781e+00 7.55726039e-01 -2.73019075e-01 -1.01339407e-01
1.34366915e-01 7.65226707e-02 7.91329682e-01 6.63325846e-01
-1.84285283e-01 3.83614391e-01 7.51627922e-01 7.36287653e-01
-1.80152223e-01 1.49206609e-01 6.20448142e-02 -6.42770007e-02
8.61719668e-01 8.52590919e-01 -8.11062813e-01 -4.11899298e-01
-2.97027081e-01 7.95086563e-01 6.22208953e-01 9.79733169e-02
-6.83370709e-01 -5.05772717e-02 6.56963468e-01 1.82981476e-01
1.35351121e-01 6.08373769e-02 -2.40319073e-01 -1.12895274e+00
3.39959919e-01 -7.83084452e-01 1.05823994e+00 -6.33158565e-01
-1.15720129e+00 3.14945728e-01 3.68919075e-01 -7.44069636e-01
-1.57485962e-01 -4.63488638e-01 -6.78915381e-01 1.03986311e+00
-2.04294133e+00 -1.20986676e+00 -1.63422957e-01 3.97342563e-01
6.65585279e-01 4.30285931e-01 9.94257092e-01 -3.47869217e-01
-1.11202322e-01 1.11366615e-01 -7.00051963e-01 4.89070378e-02
2.20632493e-01 -1.43722856e+00 3.78951371e-01 9.19270217e-01
-1.66520715e-01 7.88094819e-01 6.27793074e-01 -5.81383646e-01
-1.03056836e+00 -8.68906677e-01 1.52720201e+00 -4.26565677e-01
5.65520406e-01 -2.32258126e-01 -1.06908309e+00 4.66701835e-01
-1.20847277e-01 -1.60699949e-01 8.49934459e-01 2.04011381e-01
-5.65536618e-01 3.43894452e-01 -9.94653344e-01 5.33703208e-01
1.25435603e+00 -4.21064854e-01 -1.22367573e+00 1.64864510e-01
1.24550426e+00 -3.76966715e-01 -5.13027310e-01 2.52426147e-01
1.75547704e-01 -6.44686639e-01 9.00250912e-01 -7.88189948e-01
4.71217990e-01 -5.98196805e-01 -3.71023387e-01 -1.10253167e+00
-2.07056701e-02 -3.74425799e-01 -7.59426272e-03 1.25599897e+00
6.87808454e-01 -6.25877142e-01 4.33673382e-01 7.47127652e-01
-1.27204120e-01 -8.61759245e-01 -9.34106648e-01 -3.29506099e-01
-1.78556800e-01 -4.61975843e-01 9.04632211e-01 7.76410103e-01
1.15750469e-01 1.03036416e+00 1.83304057e-01 3.71227086e-01
4.38060820e-01 7.75018632e-01 2.30426759e-01 -1.47437000e+00
-5.22209585e-01 -1.85925752e-01 6.77957088e-02 -1.28839827e+00
6.87985480e-01 -1.05155265e+00 1.62672877e-01 -1.84626222e+00
1.50135517e-01 -5.72934508e-01 -3.48991752e-01 8.95462751e-01
-7.61514068e-01 -4.16084647e-01 2.76198566e-01 3.32733750e-01
-7.51274228e-01 3.73685002e-01 1.13833570e+00 2.36100964e-02
-3.37351739e-01 1.82983670e-02 -1.14492261e+00 7.88918495e-01
7.77412951e-01 -5.01311064e-01 -7.46415079e-01 -5.09650767e-01
3.21393073e-01 2.89436758e-01 3.01385999e-01 -5.33840239e-01
2.18423996e-02 -4.86683458e-01 -1.96850687e-01 -1.39417708e-01
1.69781178e-01 -6.68466806e-01 -5.09937823e-01 9.46090072e-02
-8.69219005e-01 -1.17836125e-01 -2.56355882e-01 6.38789952e-01
-3.48112315e-01 -5.09509563e-01 5.50191402e-01 -2.45731577e-01
-1.00511646e+00 -1.05441026e-01 8.48769248e-02 4.71141458e-01
9.08379018e-01 5.00719137e-02 -3.59874606e-01 -9.61215720e-02
-9.81617451e-01 6.15013957e-01 3.48949581e-01 5.48664808e-01
6.10281348e-01 -9.14244056e-01 -6.27761960e-01 -5.32564670e-02
3.49376172e-01 3.28657120e-01 -5.70937656e-02 6.06058598e-01
-8.23298246e-02 3.15458506e-01 2.68385112e-01 -3.43337446e-01
-1.40891516e+00 3.80472064e-01 1.63235351e-01 -7.94455051e-01
-3.27265471e-01 9.70674813e-01 3.33748043e-01 -3.46346289e-01
8.15253612e-03 -2.32500508e-01 -8.41452539e-01 1.05821222e-01
6.21848762e-01 -4.13781963e-02 1.91823021e-02 -7.04890728e-01
-6.69730723e-01 3.50240201e-01 2.60497946e-02 -1.30001351e-01
1.17339861e+00 -2.67983079e-01 -5.63503683e-01 2.97363460e-01
8.81419182e-01 1.38425767e-01 -7.68905044e-01 -4.67290848e-01
6.27595186e-01 -2.94401109e-01 -2.55132437e-01 -9.96734619e-01
-4.25623178e-01 5.29196262e-01 -8.20598453e-02 2.84572184e-01
1.27146065e+00 5.48895955e-01 8.83776903e-01 3.29785466e-01
4.87959027e-01 -8.59364986e-01 -8.26997235e-02 9.28102970e-01
7.11731493e-01 -8.64594638e-01 -2.84779340e-01 -1.13870323e+00
-6.37362480e-01 1.06634796e+00 6.99437857e-01 1.37863189e-01
2.75717348e-01 -4.05444391e-02 -1.19925968e-01 -4.79139537e-01
-8.94463181e-01 -6.40978634e-01 2.59719491e-01 3.50750893e-01
6.04376793e-01 1.82729781e-01 -9.13881898e-01 8.68025184e-01
-1.98596403e-01 -4.02255684e-01 2.40343273e-01 1.31341136e+00
-7.69455850e-01 -1.55404365e+00 -2.79226035e-01 3.30480933e-01
-4.87470150e-01 -1.92848980e-01 -4.76753235e-01 3.52564901e-01
4.19164523e-02 1.28971541e+00 -1.32945031e-01 -1.08012080e-01
4.77392554e-01 3.90920252e-01 3.98749679e-01 -1.18045521e+00
-5.87700307e-01 -1.86447695e-01 6.72961235e-01 -5.42213798e-01
-5.80957115e-01 -5.49730718e-01 -1.94294608e+00 3.72027099e-01
-1.73702568e-01 4.93672043e-01 3.18131596e-01 1.45230615e+00
2.14705631e-01 3.86541903e-01 3.94037485e-01 -1.40446007e-01
-7.97668159e-01 -5.84505498e-01 -1.73229992e-01 8.09925258e-01
1.78523049e-01 -8.31061363e-01 -1.94793537e-01 -7.50754699e-02]
|
[10.23563289642334, 9.240662574768066]
|
5c861189-d0fb-4ef2-8f69-7bc3da003ea7
|
multi-conditional-latent-variable-model-for
| null | null |
http://openaccess.thecvf.com/content_iccv_2015/html/Eleftheriadis_Multi-Conditional_Latent_Variable_ICCV_2015_paper.html
|
http://openaccess.thecvf.com/content_iccv_2015/papers/Eleftheriadis_Multi-Conditional_Latent_Variable_ICCV_2015_paper.pdf
|
Multi-Conditional Latent Variable Model for Joint Facial Action Unit Detection
|
We propose a novel multi-conditional latent variable model for simultaneous facial feature fusion and detection of facial action units. In our approach we exploit the structure-discovery capabilities of generative models such as Gaussian processes, and the discriminative power of classifiers such as logistic function. This leads to superior performance compared to existing classifiers for the target task that exploit either the discriminative or generative property, but not both. The model learning is performed via an efficient, newly proposed Bayesian learning strategy based on Monte Carlo sampling. Consequently, the learned model is robust to data overfitting, regardless of the number of both input features and jointly estimated facial action units. Extensive qualitative and quantitative experimental evaluations are performed on three publicly available datasets (CK+, Shoulder-pain and DISFA). We show that the proposed model outperforms the state-of-the-art methods for the target task on (i) feature fusion, and (ii) multiple facial action unit detection.
|
['Ognjen Rudovic', 'Stefanos Eleftheriadis', 'Maja Pantic']
|
2015-12-01
| null | null | null |
iccv-2015-12
|
['action-unit-detection', 'facial-action-unit-detection']
|
['computer-vision', 'computer-vision']
|
[ 3.23782772e-01 -4.81154546e-02 -2.54382104e-01 -2.77971983e-01
-1.10579562e+00 -1.99836269e-01 9.80077028e-01 -2.78117150e-01
-3.17831576e-01 6.89505696e-01 -5.18840216e-02 3.75294477e-01
-2.01121002e-01 -3.25103819e-01 -4.36882496e-01 -1.28123891e+00
-8.83647725e-02 3.05756658e-01 1.03377968e-01 2.88568705e-01
8.37495178e-02 6.74926281e-01 -2.12210965e+00 1.91796884e-01
5.72903335e-01 1.36945486e+00 -8.86523202e-02 5.18206716e-01
3.23553860e-01 5.43228447e-01 -2.10600138e-01 -3.80409986e-01
3.79063524e-02 -2.21223310e-01 -3.26139867e-01 4.75152671e-01
3.15598667e-01 -3.07823777e-01 1.91927239e-01 9.56737697e-01
5.69240689e-01 4.16402146e-02 1.03854287e+00 -1.44663703e+00
-9.55903679e-02 5.47235310e-02 -7.55791545e-01 -8.76420289e-02
3.85929704e-01 3.68613303e-02 9.19442952e-01 -1.31993341e+00
4.98275995e-01 1.65854990e+00 4.84332472e-01 5.56961596e-01
-1.40794432e+00 -8.73713791e-01 1.87585622e-01 4.45438549e-02
-1.58853161e+00 -9.86437857e-01 6.45444274e-01 -5.07996321e-01
7.81637073e-01 -7.07710460e-02 5.08745670e-01 1.39527178e+00
3.72682035e-01 1.03499293e+00 1.35672390e+00 -3.13538820e-01
4.14835334e-01 -4.87469360e-02 -2.91943252e-01 9.61127341e-01
6.76961988e-02 2.64999032e-01 -8.12990904e-01 -6.60766780e-01
8.10947120e-01 5.15726879e-02 1.16926678e-01 -4.58031356e-01
-6.69337749e-01 9.52846467e-01 -2.04316929e-01 1.76603481e-01
-6.10088289e-01 3.99672151e-01 2.20309198e-01 -1.94856137e-01
6.43092632e-01 -3.00577313e-01 -1.82614267e-01 -2.40378842e-01
-1.21675968e+00 1.68632746e-01 6.20757461e-01 6.61167443e-01
7.07251191e-01 1.55307408e-02 -3.01839501e-01 6.45516396e-01
8.08713853e-01 5.83976567e-01 4.35521483e-01 -9.02021706e-01
-1.22504704e-01 4.62022811e-01 2.73573752e-02 -8.25892687e-01
-2.88804412e-01 -1.33699074e-01 -6.90493882e-01 5.32073677e-01
2.61482835e-01 -1.71630010e-01 -1.09981263e+00 1.77218211e+00
5.38940310e-01 4.55121398e-01 -2.13217996e-02 3.27406466e-01
7.11957395e-01 1.27928615e-01 2.34349832e-01 -5.41065514e-01
1.39311051e+00 -4.97836769e-01 -8.71027350e-01 1.03796124e-01
2.86508083e-01 -7.54764140e-01 2.70512491e-01 5.59741259e-01
-8.89479756e-01 -5.72509348e-01 -8.51847649e-01 3.12836349e-01
-2.79341247e-02 5.99418223e-01 8.84920418e-01 9.53200161e-01
-8.68078649e-01 4.52037811e-01 -1.18728602e+00 -3.02656829e-01
8.26772511e-01 5.79554617e-01 -6.05096698e-01 1.67704411e-02
-7.24796355e-01 5.21088183e-01 1.16474554e-01 2.28647679e-01
-1.14873421e+00 -2.31429309e-01 -8.22868288e-01 -3.15752812e-02
4.45471704e-01 -6.34284973e-01 1.06727147e+00 -9.22755003e-01
-1.78661692e+00 7.35791802e-01 -3.80847186e-01 -1.57743677e-01
5.95449865e-01 -3.95808429e-01 -1.27402410e-01 3.96412432e-01
-9.56381708e-02 7.18009889e-01 1.45049369e+00 -1.20267177e+00
-6.90085232e-01 -5.16184628e-01 -3.21690649e-01 1.96839198e-02
-1.54003292e-01 3.27770591e-01 -2.53884286e-01 -4.37067032e-01
7.43698999e-02 -8.41827691e-01 -1.30823314e-01 1.16991207e-01
-2.40628347e-01 -5.60661554e-01 6.37332976e-01 -4.21379060e-01
9.85868573e-01 -2.25810075e+00 1.45871326e-01 3.70943993e-01
1.16839208e-01 5.95342740e-02 4.95982170e-02 2.42808610e-01
-1.18009754e-01 -1.25337228e-01 1.26086086e-01 -1.03001404e+00
-1.13121727e-02 3.42123628e-01 -4.08601053e-02 8.04994524e-01
3.57438743e-01 6.99353814e-01 -6.67050123e-01 -7.78164208e-01
2.11463600e-01 7.21507788e-01 -4.44585383e-01 1.47106256e-02
1.23995490e-01 3.07230175e-01 -5.47501385e-01 1.13461840e+00
7.44372129e-01 -7.50821009e-02 1.23512410e-01 -8.71916041e-02
8.81596431e-02 -3.16669077e-01 -1.39045727e+00 1.59230435e+00
-1.82220697e-01 1.06841937e-01 9.81977880e-02 -6.48865342e-01
8.94803405e-01 6.78971529e-01 6.05977297e-01 -5.78231774e-02
3.48474264e-01 1.83471292e-01 -2.78375089e-01 -2.57836998e-01
-3.64386477e-02 -4.03311551e-01 1.82734244e-02 1.97367951e-01
8.12866449e-01 3.08278471e-01 1.39987797e-01 -7.21342415e-02
9.32458818e-01 3.55168432e-01 6.69736564e-01 -1.36544287e-01
5.32213748e-01 -8.18190396e-01 6.67730927e-01 8.28939795e-01
-3.08507949e-01 3.28190535e-01 6.66528463e-01 8.12416822e-02
-4.26645488e-01 -1.15363395e+00 -3.31833303e-01 1.19009030e+00
-3.44344676e-01 -4.18858945e-01 -7.40782201e-01 -7.38864422e-01
1.64386541e-01 3.05164129e-01 -9.75313246e-01 -2.64115781e-01
-1.24478646e-01 -8.00175607e-01 6.01559162e-01 7.11887062e-01
1.55313775e-01 -9.31535423e-01 -5.57386458e-01 1.58204705e-01
9.62541550e-02 -1.14148366e+00 1.37908431e-02 1.15939341e-01
-8.71987343e-01 -1.11082423e+00 -5.87892592e-01 -2.51988620e-01
5.61476827e-01 -6.17514364e-02 6.70655251e-01 -2.07510725e-01
-4.56627935e-01 7.06314266e-01 -2.90678680e-01 -4.56656396e-01
-3.26600879e-01 -4.01708305e-01 3.35087687e-01 7.89646864e-01
4.48926568e-01 -6.64565563e-01 -3.99678171e-01 3.86884540e-01
-8.25513124e-01 -1.30826458e-01 1.01833749e+00 9.94830012e-01
6.31792486e-01 -7.12203458e-02 4.36173618e-01 -5.33622146e-01
4.48936701e-01 -4.08081979e-01 -5.40748715e-01 2.79800743e-01
-4.22759593e-01 -3.89240915e-03 -1.76900178e-01 -6.31988823e-01
-1.26986659e+00 5.61553657e-01 -4.44477657e-03 -7.30758727e-01
-4.58659381e-01 3.58017623e-01 -3.67599577e-01 -1.17785871e-01
3.57814372e-01 2.04205289e-01 2.15204537e-01 -5.72678685e-01
3.20026398e-01 4.87166673e-01 2.80800551e-01 -4.47696030e-01
5.50916612e-01 8.65272105e-01 4.73748356e-01 -9.18540776e-01
-3.90924424e-01 -5.84740162e-01 -9.79792356e-01 -3.44743937e-01
8.80343556e-01 -1.02691472e+00 -8.74111831e-01 7.39576936e-01
-1.04940093e+00 3.26912791e-01 -1.73841026e-02 6.77814543e-01
-9.96088386e-01 3.54991108e-01 -3.63480389e-01 -1.45122254e+00
-2.21687093e-01 -1.22799265e+00 1.62497044e+00 1.92258313e-01
-1.62536785e-01 -8.17961156e-01 1.20607883e-01 1.79787368e-01
7.17895478e-02 5.40698707e-01 5.47677517e-01 -6.82306826e-01
-4.28235561e-01 -5.31893134e-01 -3.75660434e-02 4.68723506e-01
3.03031027e-01 3.98879528e-01 -1.31144941e+00 -2.72725433e-01
-5.46595454e-02 -4.83418643e-01 1.10429513e+00 4.50317740e-01
7.62279212e-01 -1.05736338e-01 -5.89825869e-01 2.76159078e-01
1.08315408e+00 -2.87839379e-02 5.52503526e-01 -3.21377158e-01
2.99291939e-01 5.40058434e-01 7.06692815e-01 8.74468029e-01
-5.73823005e-02 8.18041384e-01 4.94324267e-01 6.49872189e-03
5.47925755e-02 -2.13538334e-02 5.93362629e-01 1.32153839e-01
-5.78940392e-01 1.03424609e-01 -5.82852244e-01 3.78135681e-01
-2.15025759e+00 -1.00688863e+00 2.74787813e-01 2.00315285e+00
6.83432877e-01 4.41906825e-02 2.77878433e-01 2.04925805e-01
5.62643647e-01 -4.98561449e-02 -1.11748613e-01 -1.73304547e-02
-4.10293527e-02 4.76010114e-01 1.99292749e-01 1.43553555e-01
-1.36977530e+00 8.55935633e-01 6.53502321e+00 1.30400813e+00
-7.61125624e-01 3.62471372e-01 3.20171565e-01 -2.60046363e-01
2.55368263e-01 -1.10822149e-01 -1.16290963e+00 3.40419531e-01
6.83795333e-01 3.10485244e-01 -6.53911754e-02 8.82855296e-01
1.43608764e-01 -4.30918396e-01 -1.08831561e+00 1.09901607e+00
3.25397581e-01 -9.31536257e-01 -3.43298092e-02 3.82238746e-01
5.89133918e-01 -3.28216672e-01 2.19059840e-01 1.74808383e-01
1.77747488e-01 -1.12499583e+00 6.47366762e-01 1.01440823e+00
6.61881626e-01 -7.77921617e-01 7.94621229e-01 2.66339004e-01
-1.15100503e+00 -1.67765364e-01 -3.59379388e-02 2.85876215e-01
6.80940375e-02 4.97078717e-01 -6.57025456e-01 5.60117126e-01
5.16375005e-01 6.31971657e-01 -5.85539818e-01 9.93969202e-01
-4.27156657e-01 5.65145671e-01 -5.14428854e-01 -2.25840099e-02
7.60282278e-02 1.15126997e-01 6.10199869e-01 1.09236252e+00
1.72557831e-01 -2.15084240e-01 1.64960787e-01 7.28292286e-01
3.51245880e-01 8.65799934e-02 -5.27534425e-01 -1.41344979e-01
9.53388438e-02 1.58046913e+00 -8.14272463e-01 -1.93902165e-01
-4.46927011e-01 8.08503807e-01 2.17500702e-01 3.39980930e-01
-7.37619102e-01 8.36067945e-02 7.26053119e-01 -1.68495759e-01
7.44815826e-01 -2.34583065e-01 1.78709805e-01 -8.99142027e-01
-4.10548486e-02 -5.22325873e-01 3.43431503e-01 -4.14413661e-01
-1.26490080e+00 3.90834153e-01 4.33977336e-01 -1.11163211e+00
-7.49942780e-01 -7.75844574e-01 -4.64353412e-01 7.15150535e-01
-1.34938407e+00 -1.73371172e+00 -1.23858169e-01 7.55693495e-01
1.87295288e-01 -3.30909669e-01 1.01796794e+00 1.38455071e-02
-5.53171754e-01 7.51794875e-01 -2.77931001e-02 -7.71676153e-02
7.22796321e-01 -9.52972591e-01 -3.28619868e-01 6.35827839e-01
1.75900221e-01 5.30499876e-01 5.58430493e-01 -6.59940660e-01
-1.11178946e+00 -7.46332288e-01 5.84109724e-01 -4.24017370e-01
5.52998900e-01 -6.80694461e-01 -4.76780087e-01 4.55184013e-01
-2.11693838e-01 1.61560908e-01 1.03850794e+00 2.06342131e-01
-2.55045861e-01 -2.47585364e-02 -1.20342040e+00 2.10028633e-01
7.81959713e-01 -3.61150056e-01 -3.19617003e-01 1.40116960e-01
4.87228762e-03 3.53242474e-04 -8.12324524e-01 6.72885299e-01
1.10063899e+00 -1.04339850e+00 8.35822105e-01 -6.28897667e-01
9.67352167e-02 1.13242166e-02 -3.54559869e-01 -9.11471367e-01
-3.31909209e-01 -7.93293595e-01 -5.91302574e-01 1.38386750e+00
9.54972953e-02 -5.24882138e-01 5.96586943e-01 2.49632537e-01
3.08363676e-01 -9.72603023e-01 -1.47927439e+00 -7.48533309e-01
-3.63380522e-01 -3.52875769e-01 1.79571077e-01 3.42422068e-01
-3.29368889e-01 8.18532705e-02 -4.79253739e-01 1.71000496e-01
8.53069365e-01 3.29381414e-02 7.61382878e-01 -1.44835901e+00
-4.01395351e-01 -4.05952573e-01 -9.41645682e-01 -5.97029090e-01
3.29578757e-01 -4.58488464e-01 -6.89496323e-02 -1.03650045e+00
4.23795670e-01 5.44238202e-02 -3.74069631e-01 7.24674463e-01
-1.36494145e-01 3.26805443e-01 -6.20608591e-02 1.82841763e-01
-6.93073213e-01 8.32597792e-01 9.04174864e-01 4.36873823e-01
5.10450266e-03 3.63705724e-01 -5.40224195e-01 1.07622790e+00
3.67297471e-01 -5.36442220e-01 -3.95060033e-02 3.94233614e-01
-3.72876115e-02 2.57421583e-02 7.45820820e-01 -1.00054812e+00
1.12055808e-01 -1.46111742e-01 7.40407884e-01 -6.07430995e-01
9.19654906e-01 -6.52131796e-01 -8.37253109e-02 3.12640846e-01
1.41180381e-01 -4.66952860e-01 1.09266512e-01 8.99142504e-01
-2.16971070e-01 -6.83149919e-02 1.00890183e+00 7.85760358e-02
-5.03487289e-01 2.42754772e-01 -5.28548062e-01 -6.47605002e-01
1.21137655e+00 -3.34171534e-01 1.06183454e-01 -4.48440403e-01
-1.17694962e+00 -1.36528060e-01 2.30030529e-02 3.86606097e-01
7.49550343e-01 -1.48564148e+00 -7.34818161e-01 4.54216987e-01
2.59627402e-01 -5.06391585e-01 2.02872261e-01 1.39579582e+00
3.21868867e-01 2.30539978e-01 -2.74741322e-01 -8.42102885e-01
-1.71295333e+00 3.62734288e-01 2.35022709e-01 -3.12429965e-01
-1.25047415e-01 9.29722071e-01 3.80771875e-01 9.55195725e-02
2.15781882e-01 -1.27727583e-01 -3.55457067e-01 5.20437121e-01
5.54059267e-01 5.05817592e-01 3.10381465e-02 -9.63410020e-01
-4.89627540e-01 4.31387097e-01 -1.19687952e-01 -2.18832746e-01
1.20910227e+00 -7.43204430e-02 4.26323619e-03 5.11413932e-01
9.84527647e-01 -1.31171182e-01 -1.42593622e+00 -2.66040236e-01
-2.35256627e-01 -5.64942479e-01 1.22474611e-01 -5.68801343e-01
-7.69559741e-01 7.83294618e-01 9.30661976e-01 -2.24825457e-01
1.18352008e+00 2.68607736e-01 6.84211357e-03 2.47896090e-03
4.94911999e-01 -9.92461205e-01 3.17716181e-01 1.20267607e-01
8.16564858e-01 -1.19848907e+00 1.23057485e-01 -5.73915422e-01
-3.68463516e-01 1.08438122e+00 3.24972481e-01 -2.55850524e-01
9.28044736e-01 2.49661565e-01 -2.91034877e-01 -1.88552648e-01
-9.88400102e-01 -5.60173452e-01 5.76483846e-01 5.15090108e-01
1.70229018e-01 3.54008563e-02 -1.63307697e-01 7.58238256e-01
2.60806054e-01 1.84789896e-01 -1.40593937e-02 1.15655458e+00
-3.47735882e-01 -1.06097209e+00 -4.19180274e-01 3.58402252e-01
-5.34983039e-01 1.62025914e-01 -4.14291054e-01 8.41959834e-01
3.24281842e-01 9.93857682e-01 -1.21998116e-01 -1.76093847e-01
-3.71235870e-02 3.96929801e-01 8.97310615e-01 -5.16651034e-01
-1.41175002e-01 7.74674833e-01 -7.67211542e-02 -8.33559632e-01
-8.72088671e-01 -1.18176687e+00 -9.00786698e-01 2.24518389e-01
-7.33280897e-01 -3.45657825e-01 6.54277205e-01 1.19498682e+00
3.53239626e-01 1.23191021e-01 6.35995686e-01 -1.13371444e+00
-7.29238689e-01 -1.28302610e+00 -9.26331937e-01 2.00733021e-01
1.58088759e-01 -1.56299329e+00 -4.25733864e-01 2.14423984e-02]
|
[13.59305477142334, 1.6501468420028687]
|
e57c3f42-26b1-41e6-82fc-99d0eecbc9a4
|
multiview-hessian-discriminative-sparse
|
1307.3811
| null |
http://arxiv.org/abs/1307.3811v1
|
http://arxiv.org/pdf/1307.3811v1.pdf
|
Multiview Hessian Discriminative Sparse Coding for Image Annotation
|
Sparse coding represents a signal sparsely by using an overcomplete
dictionary, and obtains promising performance in practical computer vision
applications, especially for signal restoration tasks such as image denoising
and image inpainting. In recent years, many discriminative sparse coding
algorithms have been developed for classification problems, but they cannot
naturally handle visual data represented by multiview features. In addition,
existing sparse coding algorithms use graph Laplacian to model the local
geometry of the data distribution. It has been identified that Laplacian
regularization biases the solution towards a constant function which possibly
leads to poor extrapolating power. In this paper, we present multiview Hessian
discriminative sparse coding (mHDSC) which seamlessly integrates Hessian
regularization with discriminative sparse coding for multiview learning
problems. In particular, mHDSC exploits Hessian regularization to steer the
solution which varies smoothly along geodesics in the manifold, and treats the
label information as an additional view of feature for incorporating the
discriminative power for image annotation. We conduct extensive experiments on
PASCAL VOC'07 dataset and demonstrate the effectiveness of mHDSC for image
annotation.
|
['DaCheng Tao', 'Yuanyan Tang', 'Weifeng Liu', 'Jun Cheng']
|
2013-07-15
| null | null | null | null |
['multiview-learning']
|
['computer-vision']
|
[-1.06088175e-02 -1.73708200e-01 -3.60216588e-01 -6.05132341e-01
-8.03887725e-01 -3.54976624e-01 2.20761955e-01 -2.37875104e-01
7.92248622e-02 3.99008155e-01 5.69788337e-01 2.20417947e-01
8.75324756e-02 -2.59160548e-01 -6.85434282e-01 -8.30355644e-01
2.28342071e-01 7.62844831e-02 -1.10599175e-01 -6.14339188e-02
2.11630657e-01 2.15556040e-01 -1.17061639e+00 4.03688014e-01
7.58505523e-01 7.46789753e-01 4.80353355e-01 1.53344989e-01
-9.25048590e-02 8.59480143e-01 -2.30827972e-01 -6.67530596e-02
2.53006071e-01 -5.78134000e-01 -4.30886209e-01 4.63128388e-01
6.96965456e-01 -1.52195394e-01 -7.25123644e-01 1.40192473e+00
2.50597954e-01 -4.80751917e-02 6.40565276e-01 -1.30823791e+00
-9.69976008e-01 2.54659057e-01 -1.02982163e+00 9.20550898e-02
3.38715911e-01 -1.71768978e-01 1.02597344e+00 -1.24941194e+00
8.51362765e-01 1.35797703e+00 5.45843005e-01 3.56664151e-01
-1.34319127e+00 -3.84288907e-01 2.48211712e-01 2.91191608e-01
-1.57960093e+00 -3.80072087e-01 1.35492003e+00 -5.30783653e-01
4.87584084e-01 2.90252477e-01 6.39567196e-01 8.10100257e-01
1.41462460e-01 1.09540355e+00 8.85183871e-01 -2.70373970e-01
2.56442934e-01 4.89399135e-02 -1.79386064e-01 9.39100504e-01
-6.64071068e-02 -3.25720698e-01 -6.73203170e-01 -2.71530360e-01
8.92581820e-01 5.04162133e-01 -4.96048689e-01 -6.70209706e-01
-1.25106037e+00 1.15166008e+00 5.21981955e-01 2.90171176e-01
-1.71173096e-01 1.14476793e-01 3.86000991e-01 2.92854697e-01
5.12341559e-01 -1.13936625e-01 -4.91395518e-02 2.83851177e-01
-7.89551079e-01 4.31829772e-04 4.27227050e-01 1.28731740e+00
1.06396925e+00 3.92349601e-01 -1.07301898e-01 1.12110877e+00
4.74092484e-01 6.35470390e-01 6.15309477e-01 -1.46230996e+00
3.78921688e-01 4.44333583e-01 -2.09505752e-01 -1.57983315e+00
-4.60635088e-02 -2.67373353e-01 -1.17088342e+00 -2.97979683e-01
-1.45154297e-01 2.70413250e-01 -4.92932826e-01 1.49575889e+00
2.35059589e-01 3.40915531e-01 -1.29224166e-01 1.22704566e+00
8.51626158e-01 7.51873910e-01 -2.52189159e-01 -2.89027333e-01
8.80774438e-01 -8.69989336e-01 -8.69748414e-01 -1.79455757e-01
6.11469805e-01 -7.49118090e-01 1.04594111e+00 3.08914691e-01
-8.24626625e-01 -5.21406174e-01 -9.23107505e-01 -3.75890672e-01
2.66049802e-01 2.22525522e-01 6.63019538e-01 1.64694130e-01
-9.25380230e-01 1.49640307e-01 -8.63664210e-01 -7.46618360e-02
3.80946517e-01 1.05566494e-02 -6.40102625e-01 -6.69143438e-01
-7.81167448e-01 2.95726418e-01 -9.57173109e-02 5.42255007e-02
-1.00664163e+00 -5.18617272e-01 -1.37192619e+00 -8.18788931e-02
1.84262432e-02 -5.11384368e-01 6.23998523e-01 -9.42161441e-01
-1.16170478e+00 7.30211020e-01 -5.86459816e-01 -7.76849389e-02
1.10332981e-01 -1.99671052e-02 -1.73303649e-01 3.84709418e-01
5.58079183e-01 6.25819445e-01 1.22030449e+00 -1.36296988e+00
-1.29966646e-01 -4.90167171e-01 -1.66636825e-01 3.11042309e-01
-3.49144667e-01 -3.20540160e-01 -6.11094952e-01 -9.98245537e-01
6.27381682e-01 -8.50634098e-01 -3.33631933e-01 2.03921720e-01
-1.67707875e-01 -1.02797784e-01 1.30301833e+00 -6.39717996e-01
9.81767893e-01 -2.67826128e+00 7.46938944e-01 1.33190989e-01
1.78549156e-01 -2.76473671e-01 -2.25462183e-01 3.19147706e-01
-4.68005836e-02 -2.29222268e-01 -5.15282989e-01 -5.46003163e-01
-4.48105991e-01 5.46238601e-01 -2.78761327e-01 1.04396641e+00
-7.52692595e-02 8.00924838e-01 -1.02939308e+00 -7.27173567e-01
2.51740336e-01 6.73999190e-01 -8.47209632e-01 2.19967484e-01
5.52918985e-02 6.28275573e-01 -6.61883295e-01 7.64235258e-01
6.43616498e-01 -4.53403205e-01 7.33462274e-02 -4.82692957e-01
1.56708062e-02 -3.79490942e-01 -1.06686807e+00 2.50505781e+00
-3.89931023e-01 6.86814845e-01 4.60138321e-01 -1.43452871e+00
8.70610178e-01 1.25645339e-01 6.52770162e-01 -5.67192674e-01
-1.40198186e-01 1.08195387e-01 -6.13254011e-01 -4.75348324e-01
3.04611444e-01 -1.00301303e-01 3.09352964e-01 1.28961742e-01
1.84779137e-01 -2.56957948e-01 -1.94696769e-01 6.92139447e-01
7.15212941e-01 5.31023880e-03 1.06860131e-01 -3.24115008e-01
6.46110594e-01 -1.98980674e-01 8.75278175e-01 2.73197800e-01
-1.30889103e-01 1.09235251e+00 2.16185614e-01 -3.36897910e-01
-7.96059608e-01 -8.49053204e-01 -1.82456508e-01 8.04042041e-01
4.29352880e-01 -5.07131994e-01 -4.51158106e-01 -6.73471034e-01
-4.78103794e-02 4.17432845e-01 -3.80077094e-01 -2.61579216e-01
-4.83223855e-01 -4.28672612e-01 -1.85693894e-02 3.57504606e-01
4.31320339e-01 -5.43142676e-01 6.19211644e-02 5.96767664e-02
-5.69788635e-01 -1.11564946e+00 -9.35868263e-01 -3.10623460e-02
-9.89015102e-01 -9.29734468e-01 -8.41585875e-01 -1.21394324e+00
1.00539529e+00 1.12138677e+00 7.78694570e-01 1.13612503e-01
-4.66717154e-01 8.62968743e-01 -4.21879649e-01 1.79349571e-01
2.35269926e-02 -3.67948860e-01 -2.34633572e-02 4.51868325e-01
1.09649830e-01 -6.07878804e-01 -6.62608206e-01 3.84788930e-01
-9.97586489e-01 -1.39884889e-01 3.38909745e-01 1.17804253e+00
1.00384283e+00 -2.68027872e-01 5.10947168e-01 -8.92553389e-01
1.77897424e-01 -6.88365519e-01 -3.74036789e-01 1.07792385e-01
-3.40924799e-01 5.23465574e-02 6.91213667e-01 -1.73515171e-01
-1.00070179e+00 4.81393993e-01 -1.25982746e-01 -1.03140497e+00
1.58640191e-01 6.43561780e-01 -2.42342815e-01 -4.68686014e-01
4.68044788e-01 7.33902633e-01 3.52825403e-01 -5.73733926e-01
5.83248794e-01 5.33302188e-01 3.91136825e-01 -5.04123569e-01
7.02627182e-01 8.75759661e-01 5.81178442e-02 -1.21296918e+00
-1.05733430e+00 -9.47458208e-01 -5.14007330e-01 -1.52758494e-01
7.43155181e-01 -1.41183043e+00 -3.13055336e-01 7.43280947e-02
-1.03892326e+00 1.21044125e-02 -3.12123179e-01 5.82284391e-01
-6.54805243e-01 9.77483630e-01 -6.31403744e-01 -3.25943559e-01
9.05697793e-02 -1.19596589e+00 1.41948593e+00 -1.40765935e-01
3.81945223e-02 -1.25026393e+00 8.42561107e-03 5.16371965e-01
2.02371567e-01 7.18678236e-02 7.87892640e-01 6.62779287e-02
-5.92813432e-01 -1.51171803e-01 -1.69077575e-01 5.78443348e-01
1.17033891e-01 -4.62888777e-01 -7.16407299e-01 -6.33237839e-01
3.33197504e-01 -5.79638124e-01 9.81280804e-01 5.84997594e-01
1.45856714e+00 -2.67735094e-01 -2.28759736e-01 1.09072435e+00
1.42238164e+00 -1.79749876e-01 5.78128457e-01 -7.16648698e-02
1.24449313e+00 4.90099102e-01 5.30910611e-01 4.40986663e-01
4.38309282e-01 5.58886468e-01 5.49894214e-01 -1.27766103e-01
-1.22970343e-01 -3.97634149e-01 3.75748903e-01 1.42586899e+00
1.38004825e-01 2.32567802e-01 -3.63699824e-01 4.65339094e-01
-1.95119786e+00 -9.70387399e-01 -1.54314175e-01 1.96702707e+00
8.06051195e-01 -4.04411316e-01 -3.37488741e-01 -5.14333695e-02
5.91391742e-01 5.98928750e-01 -5.47803938e-01 5.11927642e-02
-3.10799330e-01 -2.78242826e-01 3.85532498e-01 6.56899333e-01
-9.38069105e-01 8.99595797e-01 6.29524851e+00 1.06229568e+00
-9.59831119e-01 2.74289906e-01 4.31168169e-01 1.78621933e-01
-6.81241989e-01 5.74005991e-02 -5.03287375e-01 4.04804349e-01
1.74743563e-01 1.04600675e-01 4.14840817e-01 1.07178736e+00
1.92378730e-01 2.14078084e-01 -9.40696180e-01 1.57915056e+00
4.76414323e-01 -1.37963700e+00 4.28199798e-01 -2.39134375e-02
1.06791580e+00 2.00020391e-02 3.00724238e-01 -7.90588558e-03
1.24963000e-02 -8.07154238e-01 5.32299340e-01 4.09725428e-01
7.92462349e-01 -5.85238159e-01 4.67745513e-01 3.76332819e-01
-1.23123252e+00 5.36421016e-02 -7.98847973e-01 1.68259785e-01
1.49788246e-01 7.03920662e-01 -2.61699766e-01 4.39799666e-01
5.71918428e-01 1.70660520e+00 -5.31735837e-01 7.39906251e-01
-1.37804076e-01 6.74283624e-01 -7.55807478e-03 5.49466789e-01
2.36193016e-01 -7.11154103e-01 6.80997193e-01 9.03339922e-01
2.69986361e-01 1.31209105e-01 6.10574365e-01 8.24612916e-01
-5.76468594e-02 2.97149807e-01 -9.01458263e-01 1.10430568e-01
2.64399976e-01 1.21784985e+00 -5.11400402e-01 -1.24811538e-01
-8.84381950e-01 1.32030547e+00 3.21681470e-01 7.21135676e-01
-5.10016620e-01 -6.51848540e-02 5.15947640e-01 1.69075541e-02
1.93267182e-01 -5.74443996e-01 -1.46697477e-01 -1.67865992e+00
1.81287657e-02 -8.76153052e-01 1.62475646e-01 -6.10949993e-01
-1.28842115e+00 9.24725607e-02 -2.79151559e-01 -1.45006001e+00
-5.15385205e-03 -2.36598805e-01 -4.85261977e-01 7.38786697e-01
-1.45547831e+00 -1.20943165e+00 -4.04570609e-01 1.06501508e+00
9.37113643e-01 -3.43960196e-01 5.52631557e-01 4.19932514e-01
-3.89198959e-01 3.28151107e-01 4.34406132e-01 3.83534767e-02
7.73973227e-01 -1.12502491e+00 -1.54442519e-01 6.54549599e-01
4.21390444e-01 6.72189176e-01 4.93150175e-01 -4.96104747e-01
-2.02530646e+00 -1.23239005e+00 4.46929693e-01 -1.39841512e-01
4.88294929e-01 -1.76290289e-01 -9.97983694e-01 7.94898570e-01
6.75782859e-02 4.93279219e-01 8.38217735e-01 -1.91822141e-01
-5.12413919e-01 -1.51293024e-01 -9.05978620e-01 2.36666903e-01
1.04442275e+00 -8.58787417e-01 -1.85506701e-01 6.09292626e-01
7.24808395e-01 -1.82320863e-01 -8.10351014e-01 1.32309869e-01
2.30003268e-01 -9.08269942e-01 1.11225092e+00 -3.45241934e-01
3.60708296e-01 -5.12216508e-01 -6.78571105e-01 -1.38399637e+00
-5.02466857e-01 -4.14446652e-01 -7.08428696e-02 1.04602432e+00
-1.79017201e-01 -3.68908316e-01 7.32922196e-01 7.47803822e-02
-3.67520124e-01 -6.37702465e-01 -8.68570685e-01 -5.86296558e-01
-1.80904940e-01 -1.15800887e-01 -1.24635160e-01 1.25130701e+00
-2.04972625e-01 2.79870212e-01 -8.16159248e-01 1.56085581e-01
1.09434724e+00 3.16142261e-01 6.03209972e-01 -1.00618303e+00
-2.39609584e-01 5.13326228e-02 -7.37454832e-01 -1.56116104e+00
5.38435340e-01 -1.30281174e+00 -6.18281551e-02 -1.34098160e+00
4.31496948e-01 -2.08226874e-01 2.68489625e-02 2.94268221e-01
6.11718697e-03 2.82394379e-01 6.70245364e-02 6.13619864e-01
-7.26988375e-01 1.02837574e+00 1.40546215e+00 -6.20966136e-01
-1.70322284e-02 -3.62535864e-01 -7.02583551e-01 8.15085411e-01
4.78962749e-01 -3.08433741e-01 -6.49232745e-01 -7.21020222e-01
8.15442055e-02 2.68495053e-01 3.56759518e-01 -5.18951476e-01
2.67167419e-01 -2.07595110e-01 3.39478761e-01 -3.35508823e-01
3.47130418e-01 -9.00548041e-01 2.87185814e-02 1.92849174e-01
-3.26908767e-01 -2.61280984e-01 -3.82819980e-01 1.17491734e+00
-8.61311913e-01 -1.44054100e-01 8.55001271e-01 -2.34804884e-01
-7.94664264e-01 5.41470528e-01 -1.16785266e-01 1.63089231e-01
7.79949725e-01 -5.72021864e-03 1.98305145e-01 -5.30664563e-01
-7.52482116e-01 3.78258079e-01 5.15944242e-01 2.68093616e-01
1.14070153e+00 -1.64083850e+00 -6.46299601e-01 4.98149484e-01
2.46995315e-01 1.06739260e-01 4.06753838e-01 8.65378380e-01
-5.38343072e-01 1.34982228e-01 3.86133157e-02 -9.36525226e-01
-1.12778342e+00 6.61820412e-01 2.72329226e-02 4.27346826e-01
-9.55671906e-01 8.47187877e-01 5.09528637e-01 -1.22186035e-01
3.80191863e-01 -1.04447871e-01 -1.57798305e-01 -6.00413345e-02
5.46532810e-01 3.39055032e-01 -3.08569312e-01 -1.02552712e+00
-2.36034572e-01 8.77359331e-01 -2.48158779e-02 1.54380322e-01
1.26880395e+00 -4.50241417e-01 -4.19011205e-01 4.72534657e-01
1.79481494e+00 3.72815728e-02 -1.32257020e+00 -4.07550931e-01
-2.24832311e-01 -7.59959877e-01 2.81236708e-01 4.59082574e-02
-1.43457580e+00 9.72872972e-01 3.01815420e-01 -7.65412748e-02
9.82095540e-01 9.00720954e-02 7.89284885e-01 2.99841076e-01
6.00986362e-01 -9.10651624e-01 3.58378857e-01 3.28520954e-01
1.24389219e+00 -1.60663176e+00 1.72292784e-01 -6.51579678e-01
-8.36336076e-01 9.94933665e-01 3.06751817e-01 -3.59538168e-01
9.18684840e-01 -1.27687827e-01 -1.25887662e-01 -2.38251656e-01
-3.63703519e-01 1.70223430e-01 2.91657180e-01 5.69954693e-01
4.17881399e-01 3.27901691e-02 -2.57227212e-01 1.84027195e-01
2.84252644e-01 -3.32020998e-01 3.76545191e-01 8.21899295e-01
-4.58770335e-01 -9.61017728e-01 -2.83361465e-01 2.78423429e-01
-2.67807752e-01 -1.56182135e-02 -2.47781143e-01 2.85855025e-01
-1.03575185e-01 7.66291559e-01 -2.17188582e-01 -1.55033231e-01
6.05022162e-02 -2.46750951e-01 3.87536496e-01 -7.68221438e-01
1.16227560e-01 5.33024132e-01 -5.06484151e-01 -8.40988576e-01
-6.72061384e-01 -8.50442946e-01 -1.24770820e+00 8.65788832e-02
-1.13252671e-02 3.77077907e-01 5.07636786e-01 6.56402707e-01
3.27699363e-01 8.32830369e-02 9.49947476e-01 -8.61652792e-01
-4.92700845e-01 -6.24806702e-01 -1.07085836e+00 5.93921542e-01
5.16710877e-01 -6.99631929e-01 -6.67056382e-01 4.10855025e-01]
|
[11.436014175415039, -1.6095242500305176]
|
89b94d37-47ea-4b2b-901d-3c893c0d02b7
|
u-net-with-hierarchical-bottleneck-attention
|
2107.04721
| null |
https://arxiv.org/abs/2107.04721v1
|
https://arxiv.org/pdf/2107.04721v1.pdf
|
U-Net with Hierarchical Bottleneck Attention for Landmark Detection in Fundus Images of the Degenerated Retina
|
Fundus photography has routinely been used to document the presence and severity of retinal degenerative diseases such as age-related macular degeneration (AMD), glaucoma, and diabetic retinopathy (DR) in clinical practice, for which the fovea and optic disc (OD) are important retinal landmarks. However, the occurrence of lesions, drusen, and other retinal abnormalities during retinal degeneration severely complicates automatic landmark detection and segmentation. Here we propose HBA-U-Net: a U-Net backbone enriched with hierarchical bottleneck attention. The network consists of a novel bottleneck attention block that combines and refines self-attention, channel attention, and relative-position attention to highlight retinal abnormalities that may be important for fovea and OD segmentation in the degenerated retina. HBA-U-Net achieved state-of-the-art results on fovea detection across datasets and eye conditions (ADAM: Euclidean Distance (ED) of 25.4 pixels, REFUGE: 32.5 pixels, IDRiD: 32.1 pixels), on OD segmentation for AMD (ADAM: Dice Coefficient (DC) of 0.947), and on OD detection for DR (IDRiD: ED of 20.5 pixels). Our results suggest that HBA-U-Net may be well suited for landmark detection in the presence of a variety of retinal degenerative diseases.
|
['Michael Beyeler', 'Jacob Granley', 'Ziming Qi', 'Shuyun Tang']
|
2021-07-09
| null | null | null | null |
['optic-disc-detection', 'fovea-detection']
|
['medical', 'medical']
|
[-8.40207338e-02 1.71008017e-02 4.66999114e-02 3.59653868e-02
-4.14566219e-01 -2.71254241e-01 1.13601692e-01 7.47486725e-02
-5.20816565e-01 6.21808350e-01 3.02217126e-01 -5.04696429e-01
-1.73803605e-02 -5.20136416e-01 -4.00059491e-01 -5.58248639e-01
-2.41697982e-01 -1.37808830e-01 4.76735801e-01 2.85659790e-01
4.47778493e-01 8.54528546e-01 -1.67334366e+00 1.81423485e-01
1.19721246e+00 1.10633850e+00 1.43194973e-01 9.32911694e-01
1.90276965e-01 4.64594811e-01 -5.01646519e-01 -2.60191500e-01
3.50943297e-01 -3.51855785e-01 -7.77817309e-01 -2.47606561e-02
1.08528936e+00 -7.70061016e-01 -5.89211583e-02 1.06095159e+00
9.97109950e-01 -2.17231870e-01 6.30111873e-01 -4.02728438e-01
-5.33073843e-01 -2.92596072e-01 -1.09659505e+00 1.01030934e+00
-2.41031855e-01 5.98488092e-01 5.35546064e-01 -6.16453767e-01
6.79691315e-01 1.11628699e+00 5.75470865e-01 3.79433095e-01
-1.08262110e+00 -1.68708071e-01 -1.86907217e-01 3.80287319e-01
-9.26078856e-01 -3.72287601e-01 -1.44982025e-01 -9.07205820e-01
1.06204915e+00 -2.75040907e-03 1.00390601e+00 2.93133110e-01
4.03428227e-01 4.21012074e-01 1.07180595e+00 -3.52221221e-01
1.47738401e-02 -5.61586320e-01 8.84723663e-02 5.62281311e-01
7.20071912e-01 1.49554342e-01 1.35710970e-01 1.35789871e-01
9.68466461e-01 -3.69969130e-01 -5.08725643e-01 2.08824858e-01
-9.79968786e-01 5.31064808e-01 6.69380307e-01 -2.67522722e-01
-5.90175033e-01 2.26486530e-02 3.64097476e-01 -1.49075955e-01
2.89796978e-01 6.74984515e-01 -1.09197766e-01 -1.18621953e-01
-6.03229880e-01 -8.11213686e-05 -1.72395289e-01 3.80252391e-01
3.64115864e-01 -2.97754198e-01 -5.50370693e-01 9.14572477e-01
1.97448656e-01 2.31681958e-01 4.80971038e-01 -1.25960219e+00
-1.94749981e-02 8.89770269e-01 2.30159074e-01 -2.82829583e-01
-6.15147412e-01 -6.22287810e-01 -8.24478686e-01 8.78288805e-01
6.60857022e-01 -4.50406522e-01 -1.18392026e+00 9.60796833e-01
3.04180712e-01 2.22667068e-01 -3.88856918e-01 1.29008257e+00
8.21866870e-01 -2.10352596e-02 -1.80305429e-02 -1.45852536e-01
1.29339123e+00 -9.18527722e-01 -4.41921771e-01 -2.80056685e-01
8.55435848e-01 -8.09291542e-01 1.07038736e+00 4.69741598e-02
-1.33091271e+00 -5.72399378e-01 -7.58912802e-01 -5.90967596e-01
6.94579557e-02 7.89319396e-01 3.06369871e-01 4.43199217e-01
-1.44823277e+00 2.26140663e-01 -6.18945718e-01 -5.61088502e-01
7.72869349e-01 3.19180250e-01 -2.45623380e-01 -2.13494927e-01
-3.51096243e-01 9.72806692e-01 -1.11245085e-02 2.50321269e-01
-2.48413652e-01 -7.69800901e-01 -5.51463068e-01 -2.94553846e-01
-9.54797044e-02 -1.38370907e+00 8.84784877e-01 -4.16342407e-01
-1.11430633e+00 1.21894932e+00 -5.08142412e-01 -7.10424304e-01
5.89189291e-01 -5.56218028e-01 -2.67370641e-01 6.18559122e-01
1.62177652e-01 7.78289914e-01 6.51360333e-01 -6.78433120e-01
-1.17808461e+00 -7.59047866e-01 1.21564969e-01 9.22533795e-02
3.05151880e-01 3.42212528e-01 -3.77030104e-01 -2.22540885e-01
-6.23518936e-02 -5.92688084e-01 -1.51665598e-01 6.24486208e-01
-5.62810421e-01 -3.05380076e-01 4.80088919e-01 -7.35629559e-01
1.34088767e+00 -1.98327160e+00 -4.61796165e-01 -4.20824103e-02
8.20545256e-01 1.01715803e+00 -1.72110602e-01 -4.07660902e-01
1.83696393e-02 3.16380411e-01 8.43881369e-02 3.92592698e-02
-6.66023612e-01 -3.08538556e-01 2.62336612e-01 6.73340976e-01
4.28292930e-01 9.72313464e-01 -9.92262959e-01 -3.55795532e-01
1.34293482e-01 6.34412825e-01 -3.10240448e-01 -4.12346460e-02
-2.83290297e-02 2.94711709e-01 -5.20736277e-02 8.59114051e-01
4.93418455e-01 -5.27708590e-01 -2.38712266e-01 -3.77053291e-01
-6.12496912e-01 3.17628197e-02 -8.15353692e-01 8.02862883e-01
7.57098049e-02 1.28237224e+00 -1.48970738e-01 -3.32755665e-03
7.11744368e-01 -2.15543598e-01 1.95194453e-01 -7.39146411e-01
3.71328682e-01 2.77448267e-01 6.50040746e-01 -7.22141027e-01
1.00518875e-01 4.25974637e-01 1.07256436e+00 1.88136622e-01
-5.54323673e-01 6.12443089e-01 5.29655457e-01 -3.02317828e-01
9.58831012e-01 -2.45018378e-01 3.99626940e-01 9.03402269e-02
4.69127357e-01 -1.88646123e-01 3.55853230e-01 4.22154397e-01
-5.44852197e-01 1.04776776e+00 9.77436602e-01 -5.01543224e-01
-1.22487485e+00 -8.23023260e-01 -5.30170202e-01 1.29789323e-01
5.76163344e-02 -2.39117056e-01 -5.51080465e-01 -3.03916961e-01
1.75424784e-01 -5.86014017e-02 -6.99877858e-01 6.03636131e-02
-2.53468573e-01 -8.26890647e-01 2.99690515e-01 5.87347090e-01
8.09122026e-01 -4.90406960e-01 -8.74458671e-01 4.38465811e-02
7.79469162e-02 -7.77046800e-01 -3.54374409e-01 -7.61517823e-01
-9.25094306e-01 -1.72211659e+00 -1.24769258e+00 -7.91583776e-01
9.63605225e-01 4.94406164e-01 7.73975492e-01 -8.50862451e-03
-1.06272864e+00 1.47500848e-02 7.57678598e-03 -3.41760516e-01
4.65135314e-02 -4.08616751e-01 -1.15717627e-01 3.06564450e-01
3.08871061e-01 -2.54910707e-01 -1.51537585e+00 4.37254965e-01
-3.41463149e-01 -3.30195755e-01 9.90528286e-01 6.83489501e-01
8.87646914e-01 -2.53582031e-01 1.27651736e-01 -4.86285001e-01
5.44207573e-01 -1.48056895e-01 -8.86596322e-01 9.44313407e-02
-6.27519548e-01 -5.68553448e-01 -5.96817546e-02 -3.11790109e-01
-4.57859576e-01 -4.14133579e-01 4.06151235e-01 -4.56493616e-01
-2.75690496e-01 2.70312458e-01 2.53782719e-01 -4.64556038e-01
1.02561235e+00 -3.02064776e-01 5.13135076e-01 -4.66609776e-01
1.36219457e-01 9.12516952e-01 7.60901928e-01 5.58239818e-02
-1.91469975e-02 6.96070969e-01 3.10580522e-01 -9.81711328e-01
-7.32298136e-01 -7.55896509e-01 -4.63826954e-01 -2.30829075e-01
1.02080595e+00 -8.61158013e-01 -8.74417245e-01 8.26381981e-01
-1.03705692e+00 -5.42444408e-01 -1.06876418e-01 8.19111705e-01
-1.54263437e-01 5.42287827e-01 -4.44892764e-01 -3.70178401e-01
-5.67008018e-01 -1.17608917e+00 7.06458628e-01 6.83311760e-01
-1.22363694e-01 -6.60778284e-01 -9.12914053e-02 5.02246916e-01
3.08722287e-01 3.13644081e-01 1.25789356e+00 1.37173146e-01
-6.75527096e-01 -7.03375116e-02 -1.03148258e+00 5.73206067e-01
1.86590463e-01 6.61083817e-01 -7.40143061e-01 -9.54546779e-02
-7.72940040e-01 1.57705247e-01 9.28956449e-01 1.32562137e+00
8.87459755e-01 4.48499173e-02 -3.17358047e-01 7.44128108e-01
1.21870124e+00 3.00375104e-01 1.18335271e+00 6.50111139e-01
3.35385531e-01 6.63532555e-01 4.75159764e-01 3.06323349e-01
2.07711026e-01 5.24830699e-01 6.50346041e-01 -6.59675956e-01
-9.17292535e-01 5.87223411e-01 -7.86523968e-02 -3.02343071e-01
-5.05157232e-01 8.55820030e-02 -1.23746037e+00 9.82404292e-01
-1.40635455e+00 -6.10624909e-01 -7.57971942e-01 2.42445374e+00
6.85099781e-01 -1.50955826e-01 3.96381110e-01 -4.73770529e-01
1.07629931e+00 -4.75903600e-01 -8.99618804e-01 -1.28354818e-01
-3.63498598e-01 1.13410182e-01 5.71399212e-01 2.89483100e-01
-1.02284491e+00 6.14652753e-01 6.42475224e+00 1.15708746e-02
-1.36505949e+00 -2.65782714e-01 7.03737378e-01 -5.20489693e-01
3.34312737e-01 -2.95056850e-01 -1.02025783e+00 5.78094840e-01
4.20788139e-01 9.89557281e-02 -3.18234265e-02 1.66180968e-01
5.87763906e-01 -5.73872387e-01 -5.56255579e-01 9.69808102e-01
-2.15958565e-01 -1.46952748e+00 1.89238787e-02 4.50463444e-01
8.22780550e-01 3.80062968e-01 2.46676117e-01 -5.81670225e-01
-1.93628237e-01 -1.05750835e+00 -2.23648190e-01 7.91130602e-01
1.37871838e+00 -5.48387229e-01 1.15920174e+00 -6.21116221e-01
-6.00374877e-01 -7.70457610e-02 -4.21014935e-01 1.45723581e-01
-1.15655810e-01 6.59000099e-01 -9.30441558e-01 -1.98271155e-01
9.19277906e-01 1.00165582e+00 -8.80641043e-01 2.41233826e+00
-2.40923494e-01 3.36837202e-01 -1.04966789e-01 3.24873477e-01
2.87106752e-01 -2.76005059e-01 1.02426648e+00 6.70417428e-01
4.33506846e-01 -2.03920770e-02 -4.37274039e-01 8.38627636e-01
5.92501014e-02 3.06427926e-01 -2.15215415e-01 2.96480358e-01
4.81400102e-01 9.03627992e-01 -4.13357913e-01 6.71278462e-02
-5.26076555e-01 4.23409730e-01 -1.38977408e-01 7.20740199e-01
-3.03467631e-01 -6.50032103e-01 1.18958986e+00 6.80391729e-01
2.34169275e-01 1.79991037e-01 -4.15351599e-01 -4.56864774e-01
3.02331656e-01 -5.05927145e-01 1.69779196e-01 -9.80016172e-01
-8.89070988e-01 4.28371787e-01 -6.81991458e-01 -1.30609488e+00
2.03411415e-01 -8.20924640e-01 -8.25651467e-01 1.30484354e+00
-1.90892029e+00 -8.73096466e-01 -5.32348752e-01 4.39284623e-01
-3.11345588e-02 -4.36424881e-01 3.10077876e-01 8.79137814e-02
-9.90536869e-01 4.04600769e-01 2.24293664e-01 1.83492333e-01
1.01786947e+00 -1.35441494e+00 3.82833898e-01 9.31268990e-01
-8.92620683e-01 7.61929214e-01 8.04355145e-02 -8.18909943e-01
-6.01155162e-01 -1.40243745e+00 7.04963863e-01 4.70909737e-02
5.76492846e-01 7.80564368e-01 -7.64521539e-01 4.89456505e-01
-1.51305005e-01 4.70304310e-01 6.34891689e-01 7.64350668e-02
-3.00056666e-01 -1.75849885e-01 -9.17913795e-01 9.41441178e-01
6.62881672e-01 -2.35073090e-01 -1.60460219e-01 3.43191624e-01
2.19738781e-01 -4.94000643e-01 -9.68188524e-01 3.68151367e-01
5.97826064e-01 -1.23657036e+00 9.58365440e-01 -6.02707982e-01
3.18048060e-01 -4.45337981e-01 4.52141345e-01 -7.20730662e-01
-4.15920019e-01 -9.37794983e-01 -2.50297129e-01 6.35798693e-01
2.57141858e-01 -9.47426677e-01 5.41174948e-01 3.94579917e-01
-4.68333125e-01 -8.00258398e-01 -8.04166794e-01 -4.12666708e-01
-1.82343736e-01 1.79480970e-01 3.08684204e-02 4.36253071e-01
-6.30697846e-01 -3.93106453e-02 3.10966969e-01 3.77657175e-01
6.59668744e-01 -1.25440687e-01 6.62676096e-01 -1.52232742e+00
3.47834677e-01 -9.99898195e-01 -9.78433430e-01 -7.94615030e-01
-4.23483908e-01 -5.58900595e-01 -6.09929860e-01 -2.10963321e+00
-2.03313142e-01 -1.54786604e-02 -2.86102116e-01 4.00488198e-01
-3.22544664e-01 3.24257225e-01 -2.69652992e-01 4.27405208e-01
-1.40541606e-02 -4.60248142e-02 1.73601937e+00 -2.92865317e-02
-7.50578046e-01 2.62803167e-01 -7.44070232e-01 6.55564725e-01
8.38724434e-01 1.77014917e-01 -3.31870556e-01 -3.70508969e-01
1.40656456e-01 -3.32926273e-01 9.11904395e-01 -1.07506585e+00
3.83106232e-01 3.04743588e-01 4.38830376e-01 -6.18573487e-01
-3.79188284e-02 -4.59555015e-02 -4.89951372e-01 4.85259771e-01
4.55934554e-02 -4.36121374e-01 2.95276552e-01 4.22855496e-01
-1.27235830e-01 1.54085189e-01 1.22332919e+00 3.60729359e-02
-7.35269606e-01 3.84967834e-01 -3.47956359e-01 3.15708518e-01
1.07274556e+00 -6.78248584e-01 -1.11203444e+00 5.85723110e-02
-7.79805243e-01 3.39263141e-01 5.65798283e-01 3.76886241e-02
8.54004443e-01 -6.81358337e-01 -1.03886449e+00 1.60608217e-01
2.53310800e-01 2.38968238e-01 4.62357432e-01 1.59527004e+00
-9.55121338e-01 4.85336363e-01 -5.50461888e-01 -7.37636685e-01
-1.54923868e+00 -9.10909176e-02 9.02499795e-01 3.49379241e-01
-9.81213331e-01 8.71157587e-01 8.50834399e-02 8.61440837e-01
4.10988659e-01 -6.43796027e-01 -3.98230731e-01 4.05113488e-01
1.00807810e+00 1.12468159e+00 1.48995558e-03 -3.24924827e-01
1.51237905e-01 1.07710540e+00 -5.48641384e-01 5.14380872e-01
9.01086092e-01 -5.15880942e-01 -5.93849182e-01 -2.09660083e-02
7.74991870e-01 -1.51518345e-01 -1.26057231e+00 -1.56006873e-01
-3.28226358e-01 -5.63853383e-01 6.42684877e-01 -1.08355367e+00
-9.31438565e-01 9.93682384e-01 1.13917136e+00 -2.03661680e-01
1.04804575e+00 -8.43634754e-02 9.27580535e-01 -5.52738942e-02
4.68819588e-03 -7.02071786e-01 -1.65249795e-01 2.55038261e-01
7.58066356e-01 -1.18037522e+00 1.33922463e-02 -4.62275028e-01
-3.11049521e-01 1.28921568e+00 8.28822196e-01 1.69846594e-01
4.14209902e-01 -4.35270190e-01 3.80384892e-01 -2.16610730e-01
-2.44757205e-01 -1.11031127e+00 7.69267082e-01 1.02258992e+00
3.15507740e-01 -1.59777209e-01 -3.05710852e-01 4.95556630e-02
1.71679974e-01 1.69261601e-02 7.79251873e-01 4.54212397e-01
-8.57092917e-01 -6.32523119e-01 -9.37393755e-02 9.57075298e-01
-5.40433884e-01 -2.48251572e-01 -3.89062673e-01 7.93615520e-01
3.38252276e-01 7.26878762e-01 4.84469593e-01 1.98397473e-01
1.61007315e-01 -4.18471962e-01 3.58874559e-01 -6.96888328e-01
-3.90740424e-01 3.86537224e-01 2.66376466e-01 -7.70890176e-01
-3.47426772e-01 -4.95536894e-01 -1.01304257e+00 -2.06998065e-02
5.06345667e-02 -6.81886315e-01 3.57885212e-01 4.05607104e-01
8.04129779e-01 3.18734437e-01 1.51205331e-01 -3.83657455e-01
1.84926361e-01 -9.55202341e-01 -9.01615262e-01 -1.77791893e-01
8.77394080e-01 -6.98957622e-01 -2.48086706e-01 6.95975050e-02]
|
[15.819815635681152, -3.9959566593170166]
|
fc480cd3-80d4-4288-acac-22fc1c23b49c
|
borderdet-border-feature-for-dense-object
|
2007.11056
| null |
https://arxiv.org/abs/2007.11056v3
|
https://arxiv.org/pdf/2007.11056v3.pdf
|
BorderDet: Border Feature for Dense Object Detection
|
Dense object detectors rely on the sliding-window paradigm that predicts the object over a regular grid of image. Meanwhile, the feature maps on the point of the grid are adopted to generate the bounding box predictions. The point feature is convenient to use but may lack the explicit border information for accurate localization. In this paper, We propose a simple and efficient operator called Border-Align to extract "border features" from the extreme point of the border to enhance the point feature. Based on the BorderAlign, we design a novel detection architecture called BorderDet, which explicitly exploits the border information for stronger classification and more accurate localization. With ResNet-50 backbone, our method improves single-stage detector FCOS by 2.8 AP gains (38.6 v.s. 41.4). With the ResNeXt-101-DCN backbone, our BorderDet obtains 50.3 AP, outperforming the existing state-of-the-art approaches. The code is available at (https://github.com/Megvii-BaseDetection/BorderDet).
|
['Jian Sun', 'Songtao Liu', 'Zeming Li', 'Yuchen Ma', 'Han Qiu']
|
2020-07-21
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2211_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123460528.pdf
|
eccv-2020-8
|
['dense-object-detection']
|
['computer-vision']
|
[-1.89069480e-01 -1.50340036e-01 -4.29588586e-01 -2.63299257e-01
-9.27967727e-01 -4.51672435e-01 4.93286341e-01 -4.62593101e-02
-4.50492144e-01 3.27735037e-01 -5.68680689e-02 -1.17142119e-01
2.85245597e-01 -6.99501693e-01 -7.15891421e-01 -5.11864960e-01
3.33690457e-02 2.33190414e-02 8.81868780e-01 -3.29619348e-02
2.83744007e-01 6.37939453e-01 -1.36049998e+00 4.04879779e-01
6.25783801e-01 1.53361118e+00 6.54744029e-01 2.76898593e-01
6.60214946e-02 2.05022678e-01 -3.30394566e-01 -1.31649166e-01
6.45025313e-01 3.93210165e-02 -1.57284081e-01 -3.84407014e-01
5.84466875e-01 -5.20353973e-01 -4.47110653e-01 1.13503385e+00
6.56392097e-01 -1.82638705e-01 3.94421428e-01 -1.02683461e+00
-4.53582525e-01 4.19660121e-01 -1.02756476e+00 5.69923341e-01
1.70190692e-01 1.02056459e-01 1.04794645e+00 -1.41603804e+00
5.65155745e-01 8.72050822e-01 9.51887608e-01 1.60987198e-01
-1.17629719e+00 -1.18609393e+00 2.52800673e-01 3.15584481e-01
-1.98025811e+00 -3.77012074e-01 5.91657281e-01 -1.99137911e-01
8.54912162e-01 3.42303440e-02 6.45698369e-01 8.40459645e-01
2.59953529e-01 6.99781299e-01 9.03680146e-01 -1.73104897e-01
2.29158513e-02 -9.09336656e-02 1.19146824e-01 7.40477383e-01
4.17134792e-01 2.98315555e-01 -5.11339843e-01 -6.96180388e-02
9.87776399e-01 2.59054601e-01 -1.63150832e-01 -2.77901858e-01
-1.18941808e+00 7.29784369e-01 1.12791216e+00 1.30995065e-01
-4.40673321e-01 1.09026067e-01 1.23303734e-01 -3.37940276e-01
2.99894273e-01 1.94599345e-01 -1.85991853e-01 2.73042768e-01
-1.13967383e+00 1.81352675e-01 3.09620172e-01 1.23251855e+00
8.41662645e-01 -1.90463543e-01 -3.01804990e-01 5.42339802e-01
3.17180574e-01 6.23280525e-01 2.66698152e-01 -7.95224249e-01
5.86945951e-01 7.56990194e-01 2.11890548e-01 -9.12921071e-01
-5.94674647e-01 -9.38317180e-01 -7.14478076e-01 1.01400912e-01
3.36010396e-01 -1.00535572e-01 -8.98611367e-01 1.20082295e+00
3.36881816e-01 5.32528937e-01 -4.18895870e-01 1.10686374e+00
6.14605606e-01 6.38180017e-01 -1.06039550e-02 2.37709850e-01
1.56722188e+00 -8.90517771e-01 -2.19891161e-01 -3.58553261e-01
6.32923841e-01 -7.01430261e-01 7.91961372e-01 1.43054157e-01
-8.04586172e-01 -7.18097925e-01 -1.17345202e+00 -4.07985933e-02
-1.50950536e-01 7.02091157e-01 5.58665872e-01 2.28374362e-01
-1.06793845e+00 3.02193344e-01 -9.69189584e-01 -1.94920674e-01
4.75850165e-01 4.15958583e-01 -1.94177613e-01 7.88899139e-02
-8.68337989e-01 7.40821660e-01 4.01119560e-01 -5.13795065e-03
-5.71399510e-01 -7.28854656e-01 -7.31703579e-01 1.73050106e-01
2.85265237e-01 -4.53168094e-01 1.20785761e+00 -3.78757209e-01
-1.00816011e+00 6.92946374e-01 -2.62845367e-01 -7.46980369e-01
6.06333792e-01 -2.18268603e-01 -4.45234269e-01 1.58116832e-01
5.31244099e-01 1.04585934e+00 5.83636582e-01 -8.42464924e-01
-1.07248557e+00 -4.61220413e-01 -5.71218133e-02 8.93828943e-02
-8.28577206e-02 1.45936375e-02 -7.71256983e-01 -6.43907070e-01
5.38104177e-01 -9.63261545e-01 -3.69562835e-01 1.83913231e-01
-5.67598283e-01 -2.15498388e-01 7.20664322e-01 -4.60608631e-01
1.38166606e+00 -2.38292193e+00 -8.07103753e-01 1.90744638e-01
3.96585822e-01 2.76146829e-01 6.96704760e-02 1.33612365e-01
1.30509377e-01 -5.14833927e-02 2.18538821e-01 -2.62667924e-01
-1.50332168e-01 -3.51104230e-01 -4.09738183e-01 7.71110594e-01
1.79543003e-01 7.95381904e-01 -6.35317862e-01 -3.94986182e-01
2.52192169e-01 4.97498035e-01 -8.64149153e-01 -1.69114292e-01
1.22247465e-01 3.25211644e-01 -7.51127243e-01 6.41373515e-01
9.96195436e-01 -5.34828484e-01 -1.26473770e-01 -5.97923338e-01
-6.53776050e-01 2.88477719e-01 -1.30300617e+00 1.73409390e+00
-1.57409266e-01 7.70758152e-01 -5.34098670e-02 -5.00010073e-01
1.17009711e+00 -1.32588044e-01 3.43499064e-01 -7.99373984e-01
1.85102805e-01 3.31444710e-01 2.72276513e-02 7.96035081e-02
5.13556063e-01 4.57272351e-01 -4.15142365e-02 -2.22319178e-03
-1.71422884e-01 4.67691720e-01 -7.40847588e-02 2.98193283e-02
1.22026622e+00 5.23926504e-02 4.82031763e-01 -3.91733348e-01
3.98398250e-01 1.35497257e-01 7.69977748e-01 1.04640067e+00
-4.57273483e-01 7.12969899e-01 1.25758812e-01 -4.42348570e-01
-9.77402210e-01 -1.16033041e+00 -4.73877013e-01 8.52888703e-01
5.78805804e-01 -5.51257253e-01 -7.51155019e-01 -6.24828935e-01
1.63273066e-01 4.84595746e-01 -4.06402439e-01 1.27011940e-01
-7.77794242e-01 -3.82136911e-01 5.07456243e-01 9.07651186e-01
1.00676095e+00 -6.53136075e-01 -6.51874304e-01 1.51974961e-01
-1.27812847e-01 -1.22599721e+00 -6.68931365e-01 3.84923406e-02
-6.47986650e-01 -8.51512015e-01 -4.72424537e-01 -7.64287770e-01
6.38337016e-01 6.83597505e-01 5.66891909e-01 3.61812338e-02
-2.93283701e-01 -1.15159430e-01 -3.22316945e-01 -2.92286932e-01
2.51454443e-01 2.98573613e-01 9.31746140e-02 -1.32033691e-01
4.67082173e-01 -4.36828703e-01 -1.11277497e+00 6.39645338e-01
-2.50798613e-01 2.56911606e-01 8.83577645e-01 6.41158700e-01
1.00438070e+00 -4.22068864e-01 3.75631750e-01 -4.07844931e-01
-6.66602775e-02 -4.31585968e-01 -9.98664558e-01 -2.79086307e-02
-6.05319738e-01 -3.30572575e-02 5.62488914e-01 -4.44666743e-01
-7.73608565e-01 4.15399432e-01 -2.08677858e-01 -4.09576267e-01
-1.64997894e-02 -1.34023294e-01 2.18092538e-02 -3.75017196e-01
7.56210208e-01 2.70406961e-01 -3.89170647e-01 -5.61244547e-01
3.28214735e-01 6.46426022e-01 5.87406099e-01 -1.03194498e-01
7.72692084e-01 9.22868013e-01 -1.12474702e-01 -5.46947360e-01
-9.01484311e-01 -6.91047192e-01 -5.45046270e-01 -5.62104397e-02
6.36520922e-01 -1.25028718e+00 -6.62549555e-01 8.71437937e-02
-1.22487891e+00 -1.00185759e-01 6.02735281e-02 6.47866488e-01
-3.55316103e-01 7.31156170e-02 -5.21370053e-01 -5.30802131e-01
-6.52323186e-01 -1.03626442e+00 1.24295473e+00 3.72254521e-01
3.70405382e-04 -2.57437199e-01 -1.79875612e-01 -3.75811905e-02
4.56646770e-01 -7.13120997e-02 6.61379620e-02 -6.98070884e-01
-9.70270693e-01 -5.24598360e-01 -7.60060191e-01 -1.10224448e-02
-2.74508834e-01 -3.24387640e-01 -1.03036344e+00 -1.90832481e-01
-2.12889448e-01 2.12267458e-01 9.94133472e-01 5.81339300e-01
1.12650740e+00 -1.26143929e-03 -9.27361250e-01 9.58588064e-01
1.41603065e+00 1.51771039e-01 3.53202909e-01 4.31244761e-01
4.60156262e-01 3.23387422e-02 8.08716416e-01 6.31149352e-01
4.24635261e-01 9.74220455e-01 4.28286225e-01 -6.47565871e-02
-3.13456535e-01 -4.76614565e-01 3.08874875e-01 2.99217284e-01
4.73508276e-02 6.49396097e-03 -9.56704378e-01 4.41765010e-01
-1.58849716e+00 -8.35945249e-01 -1.59204111e-01 2.02119970e+00
4.28630918e-01 4.64691550e-01 -3.48241888e-02 -3.48165035e-01
1.15551341e+00 1.44292802e-01 -6.06396735e-01 3.02898139e-01
1.10636182e-01 -1.00803167e-01 1.06411040e+00 2.38231942e-01
-1.21527612e+00 1.10772324e+00 5.67639780e+00 1.06860065e+00
-1.22131765e+00 2.43580535e-01 3.96060318e-01 -4.25339229e-02
2.77810633e-01 3.02246828e-02 -1.68050659e+00 5.56044281e-01
5.48166394e-01 8.41916427e-02 1.92000657e-01 1.19357371e+00
3.52776676e-01 -1.65719941e-01 -8.42922628e-01 1.16159678e+00
-1.31694049e-01 -1.55595458e+00 -2.38536507e-01 1.97915852e-01
5.48857391e-01 5.85340261e-01 1.41414240e-01 2.53708392e-01
-1.04917377e-01 -5.15155792e-01 1.01466525e+00 1.85554162e-01
9.52199161e-01 -4.52984810e-01 5.32625437e-01 4.93900627e-01
-1.72984815e+00 -1.57678351e-01 -5.04132867e-01 -4.55696508e-03
1.55915439e-01 5.98941505e-01 -1.03619254e+00 2.61109859e-01
8.59834671e-01 5.92362881e-01 -6.18896544e-01 1.40497971e+00
-3.16526324e-01 5.14946461e-01 -6.34414494e-01 2.78776810e-02
2.11610749e-01 8.33013579e-02 5.99812567e-01 1.40266144e+00
5.65402091e-01 9.56498235e-02 2.90760726e-01 9.14870858e-01
-1.29660172e-02 1.25835747e-01 -4.75644141e-01 6.91375732e-01
9.74053264e-01 1.36752379e+00 -9.79315698e-01 -3.51812571e-01
-4.71606642e-01 8.76500666e-01 4.03713465e-01 1.43909916e-01
-1.14623868e+00 -4.35332835e-01 5.34324706e-01 4.47528571e-01
7.43395925e-01 -3.08130383e-01 -4.23599720e-01 -9.98026013e-01
1.29733175e-01 -2.18876630e-01 2.49599040e-01 -6.28813803e-01
-9.50592875e-01 7.51491606e-01 -1.88381597e-01 -1.61447680e+00
5.95111065e-02 -6.45171762e-01 -6.18113220e-01 7.67533660e-01
-1.40921080e+00 -1.13638699e+00 -5.67990601e-01 4.49295878e-01
3.66255105e-01 1.09991446e-01 4.56165940e-01 4.67229933e-01
-5.95454454e-01 7.27905393e-01 3.21021653e-03 4.02958930e-01
6.67709708e-01 -7.66089380e-01 7.31493771e-01 9.73702490e-01
1.63801402e-01 5.94803214e-01 4.11782265e-01 -7.82924116e-01
-1.08936715e+00 -1.31230700e+00 6.81874990e-01 -2.52608538e-01
7.41279721e-01 -5.53037524e-01 -6.48389578e-01 7.83256710e-01
-2.13304773e-01 6.25713944e-01 1.71244964e-01 -2.46782780e-01
-4.68378901e-01 -4.30005044e-01 -9.98498023e-01 6.49218500e-01
1.40866292e+00 -3.33137035e-01 -2.43977919e-01 6.03718683e-02
5.70630133e-01 -4.96511877e-01 -6.35051906e-01 4.29699272e-01
6.12316847e-01 -1.07081878e+00 1.05310571e+00 1.72037870e-01
-1.78810954e-01 -6.09461844e-01 -2.01642469e-01 -8.60735178e-01
-6.95608258e-01 -3.29102635e-01 -8.28205794e-02 9.96271312e-01
4.92826909e-01 -8.76369894e-01 9.43922579e-01 4.19945866e-02
-2.84298778e-01 -8.76077890e-01 -1.08272338e+00 -9.38683033e-01
-4.07368034e-01 -5.85054994e-01 5.06655335e-01 4.81670886e-01
-2.18274504e-01 5.09628616e-02 7.01630339e-02 5.57763219e-01
6.45611107e-01 2.08503738e-01 4.91333157e-01 -1.01056230e+00
-5.92797771e-02 -4.68541533e-01 -7.14415312e-01 -1.71621585e+00
-1.58672586e-01 -8.36273789e-01 -2.29468918e-03 -1.45059490e+00
9.47153419e-02 -6.88846767e-01 -4.11657244e-01 5.06401896e-01
-2.11780909e-02 5.74032426e-01 3.39386672e-01 5.60943782e-01
-8.38778794e-01 3.90197188e-01 1.00156403e+00 1.94326520e-01
-2.06926644e-01 7.96515942e-02 -6.57834709e-01 9.91847813e-01
1.16709960e+00 -6.34617984e-01 6.33354113e-02 -2.50222594e-01
-1.90687597e-01 -3.02470744e-01 5.12860656e-01 -1.49722290e+00
6.26309037e-01 2.11265847e-01 7.78635263e-01 -1.05854166e+00
5.40394008e-01 -7.16855407e-01 -1.18545173e-02 5.84273696e-01
-3.86011712e-02 2.13744178e-01 1.54022798e-01 5.04166663e-01
2.91948002e-02 -7.57555813e-02 7.45889604e-01 1.10499211e-01
-1.02779174e+00 3.95857900e-01 1.31999236e-02 -7.74777979e-02
1.30241847e+00 -2.43097350e-01 -6.16554081e-01 -8.96508247e-03
-4.34080929e-01 2.43332759e-01 5.08373201e-01 2.25723267e-01
6.26626432e-01 -1.40615928e+00 -5.93351066e-01 3.15722018e-01
1.58781722e-01 -7.19064996e-02 1.96949542e-01 1.17412007e+00
-5.63671589e-01 6.15198314e-01 -3.88818495e-02 -7.74674654e-01
-9.32931840e-01 4.05060709e-01 3.33604276e-01 -8.09341595e-02
-9.82686937e-01 7.68209338e-01 4.78597462e-01 9.46107730e-02
1.56970516e-01 -4.86027390e-01 -1.21315673e-01 -8.21287781e-02
7.73091435e-01 3.95537525e-01 -8.54682252e-02 -6.73750520e-01
-6.62973523e-01 6.16761506e-01 -2.59403586e-01 -1.15205413e-02
9.61859703e-01 -6.38381764e-02 2.64887750e-01 -3.38863581e-02
1.00509202e+00 2.52691776e-01 -1.64398170e+00 -3.50041777e-01
6.31628111e-02 -4.31562960e-01 2.19490156e-01 -5.70045173e-01
-9.40551758e-01 5.73266983e-01 8.70235741e-01 1.95313971e-02
9.95412171e-01 2.69063264e-01 7.83611178e-01 1.52105302e-01
6.29434705e-01 -7.60484636e-01 -2.66505033e-01 4.65220243e-01
8.48773897e-01 -1.10013199e+00 -3.65588348e-03 -6.91604555e-01
-4.01762336e-01 8.90667796e-01 8.90909612e-01 -4.69780117e-01
6.79965913e-01 4.25788075e-01 -1.29007742e-01 -2.58419495e-02
-3.16786319e-01 -3.93776149e-01 3.51968408e-01 4.32037979e-01
1.19781874e-01 1.76765457e-01 -2.08554372e-01 8.13873470e-01
-1.77746430e-01 -2.01813444e-01 -3.26435007e-02 6.35646880e-01
-7.39469349e-01 -7.07013428e-01 -5.17651200e-01 5.35282910e-01
-3.65587860e-01 -3.85685027e-01 8.04815963e-02 7.80199945e-01
4.20585275e-01 7.01916099e-01 3.57519031e-01 -5.62016547e-01
3.56847018e-01 -4.05158699e-01 2.14271203e-01 -6.58998728e-01
-3.28276485e-01 1.91250697e-01 -6.26010969e-02 -9.06914532e-01
4.20934968e-02 -6.45997107e-01 -1.61749148e+00 -2.27030143e-01
-5.99965394e-01 -2.07524076e-01 7.21728027e-01 4.58823144e-01
9.22132254e-01 1.26950189e-01 4.68438864e-01 -1.12285280e+00
-6.73534632e-01 -1.01833916e+00 -5.19613266e-01 -1.39368355e-01
2.83584028e-01 -8.55955541e-01 -2.23768398e-01 -3.11337858e-01]
|
[8.640538215637207, -0.429386705160141]
|
24d65348-bb65-4ba7-99c0-8129795d4f82
|
open-world-detr-transformer-based-open-world
|
2212.02969
| null |
https://arxiv.org/abs/2212.02969v1
|
https://arxiv.org/pdf/2212.02969v1.pdf
|
Open World DETR: Transformer based Open World Object Detection
|
Open world object detection aims at detecting objects that are absent in the object classes of the training data as unknown objects without explicit supervision. Furthermore, the exact classes of the unknown objects must be identified without catastrophic forgetting of the previous known classes when the corresponding annotations of unknown objects are given incrementally. In this paper, we propose a two-stage training approach named Open World DETR for open world object detection based on Deformable DETR. In the first stage, we pre-train a model on the current annotated data to detect objects from the current known classes, and concurrently train an additional binary classifier to classify predictions into foreground or background classes. This helps the model to build an unbiased feature representations that can facilitate the detection of unknown classes in subsequent process. In the second stage, we fine-tune the class-specific components of the model with a multi-view self-labeling strategy and a consistency constraint. Furthermore, we alleviate catastrophic forgetting when the annotations of the unknown classes becomes available incrementally by using knowledge distillation and exemplar replay. Experimental results on PASCAL VOC and MS-COCO show that our proposed method outperforms other state-of-the-art open world object detection methods by a large margin.
|
['Gim Hee Lee', 'Mingli Ding', 'Yongqiang Zhang', 'Na Dong']
|
2022-12-06
| null | null | null | null |
['open-world-object-detection']
|
['computer-vision']
|
[ 9.52057391e-02 2.67376661e-01 -1.27755195e-01 -2.90123165e-01
-6.01951301e-01 -5.62041223e-01 2.71358520e-01 1.73278838e-01
-3.91440570e-01 9.42328334e-01 -3.68758112e-01 1.97884515e-01
3.15104306e-01 -8.19669783e-01 -9.68355656e-01 -8.21481824e-01
1.07476331e-01 9.03867900e-01 1.03698611e+00 1.75847441e-01
9.08691660e-02 4.33884174e-01 -1.80751240e+00 4.74519044e-01
8.97993147e-01 1.07813787e+00 4.41228777e-01 6.42884433e-01
-3.24612528e-01 8.66316199e-01 -4.54514861e-01 -4.23790365e-01
3.92878234e-01 8.93897936e-02 -8.07921767e-01 3.70271325e-01
6.04432762e-01 -5.28548777e-01 -6.98500574e-02 1.10932827e+00
2.75494277e-01 1.86659187e-01 4.92465377e-01 -1.17315102e+00
-7.40651786e-01 2.13725448e-01 -6.07199490e-01 5.16625106e-01
-1.33179901e-02 2.55206198e-01 6.37933731e-01 -1.11728859e+00
7.95778990e-01 1.05943441e+00 5.63202262e-01 5.84756255e-01
-9.79002535e-01 -8.26168418e-01 5.80355406e-01 4.71807927e-01
-1.54598498e+00 -4.66344208e-01 5.65429688e-01 -4.85134333e-01
5.68589628e-01 2.38524601e-01 5.94306529e-01 7.09970713e-01
8.33935514e-02 8.14253330e-01 8.79313409e-01 -3.38597298e-01
3.53726298e-01 7.14299142e-01 4.75976408e-01 9.46311533e-01
4.71925318e-01 -3.46547216e-02 -2.49924228e-01 -2.28196487e-01
4.26924139e-01 2.96920985e-01 -1.04388684e-01 -7.87041068e-01
-1.15022421e+00 5.37064552e-01 6.57079399e-01 -1.08636878e-01
-2.61265486e-01 -1.70245707e-01 3.68078828e-01 4.85263802e-02
7.57569313e-01 4.13740836e-02 -7.58467555e-01 5.08694828e-01
-6.45947933e-01 -5.89611568e-02 7.12892473e-01 1.15069354e+00
9.99858141e-01 -2.61324227e-01 -2.63773054e-01 6.84112906e-01
2.94250250e-01 3.25014263e-01 7.41831183e-01 -4.13001686e-01
3.34754646e-01 9.01729524e-01 3.34559649e-01 -5.45326293e-01
2.54207551e-02 -5.72681606e-01 -2.67387122e-01 2.35847563e-01
3.61166984e-01 1.68399826e-01 -1.29823411e+00 1.20471716e+00
1.12181032e+00 5.36853254e-01 2.17894509e-01 8.42283010e-01
8.47564578e-01 6.81318402e-01 -9.30420309e-03 -2.95422882e-01
1.46522701e+00 -1.19357395e+00 -5.27621865e-01 -3.00209492e-01
3.66892755e-01 -6.22745275e-01 7.61536419e-01 1.09752484e-01
-4.96732116e-01 -7.83092797e-01 -1.01232147e+00 2.11911500e-01
-5.78640938e-01 3.01707566e-01 4.23051536e-01 5.17047882e-01
-3.14649969e-01 3.35589379e-01 -8.76521528e-01 -2.37054050e-01
9.03267801e-01 1.89923584e-01 -3.78049076e-01 -1.97450548e-01
-7.71364927e-01 7.89047897e-01 1.10552216e+00 7.06151351e-02
-1.42556894e+00 -6.07420683e-01 -7.26634622e-01 -3.50309275e-02
9.66468990e-01 -4.34686929e-01 1.05601811e+00 -1.14397907e+00
-9.47876990e-01 8.48421276e-01 -1.16518557e-01 -4.74929154e-01
6.26544476e-01 -3.76074910e-01 -4.26958174e-01 -2.37765466e-03
2.95951307e-01 6.98321640e-01 1.26919341e+00 -1.45899594e+00
-1.17103636e+00 -4.66235071e-01 1.13790415e-01 2.50827253e-01
-2.42443815e-01 -3.05372328e-01 -4.21633154e-01 -3.46316338e-01
4.22462136e-01 -9.36803579e-01 -6.18600920e-02 4.98461485e-01
-4.10954654e-01 -2.99437374e-01 1.23071194e+00 -3.91928226e-01
7.93631852e-01 -2.16290307e+00 -9.93044153e-02 -1.34262979e-01
3.84992719e-01 4.36674356e-01 7.14579672e-02 -3.10302168e-01
3.29953171e-02 -4.02267098e-01 -8.16537216e-02 -9.53165740e-02
-4.53580439e-01 4.82881427e-01 -5.22357881e-01 5.20962238e-01
4.10420507e-01 5.92792869e-01 -1.00002384e+00 -7.32303381e-01
1.94996208e-01 1.46152332e-01 -3.15529555e-01 3.47839385e-01
-5.04367650e-01 3.87173176e-01 -4.07942921e-01 6.83511913e-01
8.84628475e-01 -3.28310907e-01 -1.47628605e-01 -1.56838611e-01
2.23005302e-02 -5.45081645e-02 -1.62111187e+00 1.12155342e+00
-1.32190421e-01 2.43545532e-01 -2.78443545e-01 -8.29672396e-01
6.99370682e-01 7.41923973e-02 -1.11254133e-01 -1.89066038e-01
2.48475634e-02 1.60273477e-01 3.63783017e-02 -5.36114275e-01
4.04571861e-01 8.43666047e-02 3.69369000e-01 3.06336582e-01
2.82682896e-01 1.89379022e-01 2.54265666e-01 1.43957719e-01
7.16469347e-01 2.83142000e-01 4.23852444e-01 -5.18575571e-02
5.98946393e-01 5.26852496e-02 8.93808067e-01 8.16306531e-01
-4.57547218e-01 5.57483137e-01 -2.46688835e-02 -9.44637358e-01
-9.06932533e-01 -1.01479399e+00 -3.82863522e-01 1.18927824e+00
4.40916806e-01 1.41143546e-01 -6.93060458e-01 -1.30729151e+00
6.82776934e-03 5.88073671e-01 -7.53517866e-01 -1.53899685e-01
-3.36715519e-01 -6.92961156e-01 1.00096963e-01 4.79676276e-01
6.33090973e-01 -1.05230665e+00 -8.93025041e-01 1.82864025e-01
-1.69730350e-01 -1.07183802e+00 -3.27696413e-01 4.05644774e-01
-8.15276206e-01 -1.48969638e+00 -6.93724692e-01 -9.36746299e-01
1.04296768e+00 3.92619640e-01 7.24771500e-01 3.14289937e-03
-5.95063627e-01 1.56390876e-01 -3.83174509e-01 -7.19696105e-01
-4.25935805e-01 -2.20402196e-01 2.17013583e-01 5.64096093e-01
2.80939490e-01 1.94848590e-02 -4.85234261e-01 4.59441394e-01
-6.20113850e-01 1.71935670e-02 4.89740133e-01 8.91916990e-01
1.01794612e+00 7.66563118e-02 5.28527915e-01 -1.23367798e+00
-2.23192036e-01 -5.61137855e-01 -7.05612361e-01 5.02342403e-01
-4.31757867e-01 4.53983545e-02 4.06842172e-01 -1.04942369e+00
-1.29585755e+00 4.60226834e-01 2.97354132e-01 -6.69836998e-01
-2.43715823e-01 -8.50977451e-02 -2.31603071e-01 -1.17509246e-01
7.78438032e-01 3.58617067e-01 -4.79273796e-01 -3.64979923e-01
5.01982629e-01 7.33060718e-01 5.76310873e-01 -3.42461497e-01
8.89821231e-01 7.34698713e-01 -5.91439307e-01 -5.16996861e-01
-1.30745828e+00 -6.67294919e-01 -9.48324323e-01 -2.10810706e-01
7.73956001e-01 -1.17779589e+00 -2.47303043e-02 3.96530777e-01
-1.06424749e+00 1.03795923e-01 -7.50913501e-01 2.67273307e-01
-3.30143541e-01 3.02069128e-01 -1.96007073e-01 -9.83455539e-01
-2.84137666e-01 -9.69948232e-01 1.13111162e+00 5.00433445e-01
1.62230730e-01 -5.20410597e-01 -1.53530523e-01 5.16081035e-01
-6.66943863e-02 1.04146808e-01 5.81434846e-01 -1.18469083e+00
-9.56067681e-01 -4.89420354e-01 -3.17056149e-01 3.14438373e-01
7.34765604e-02 -1.55820668e-01 -1.24301445e+00 -4.27416116e-01
1.13685220e-01 -3.18762392e-01 9.23725307e-01 4.22887783e-03
1.04774344e+00 -3.38499188e-01 -8.70462775e-01 3.66181910e-01
1.18322337e+00 1.95175692e-01 3.35804999e-01 1.73198327e-01
7.94168532e-01 3.57140929e-01 9.39060450e-01 4.26899523e-01
2.49176919e-01 2.78825581e-01 6.18400156e-01 9.04493704e-02
-2.72753447e-01 -2.72210062e-01 6.81919008e-02 2.52121598e-01
2.50516176e-01 -6.75559565e-02 -7.25570142e-01 8.66928577e-01
-1.86689472e+00 -8.03913653e-01 -7.24512115e-02 2.35939479e+00
8.69938910e-01 5.44676483e-01 -1.84323564e-01 2.19871029e-02
1.23421931e+00 -2.52090216e-01 -9.42643464e-01 2.09086895e-01
1.07976817e-01 -1.40200287e-01 4.60309178e-01 1.60490245e-01
-1.37793529e+00 1.05488169e+00 5.33711576e+00 7.26693451e-01
-8.76905203e-01 4.73744839e-01 6.56481087e-01 -1.05592206e-01
3.09922010e-01 1.88374698e-01 -1.23308384e+00 4.17464852e-01
4.63723600e-01 7.48960748e-02 1.06777862e-01 1.22325718e+00
-4.15368557e-01 -4.13462073e-01 -1.33740091e+00 7.97785282e-01
2.10304320e-01 -1.16945803e+00 5.77171147e-02 -4.32215333e-01
8.97123396e-01 3.50969881e-02 -2.86704868e-01 4.95708138e-01
2.23461032e-01 -2.70484596e-01 9.51090336e-01 5.61942160e-01
5.20097375e-01 -5.87732494e-01 6.78468943e-01 8.22236598e-01
-1.23333073e+00 -4.19163197e-01 -6.85797095e-01 2.38147727e-03
-1.96977809e-01 4.52408850e-01 -9.99790192e-01 3.10052216e-01
8.93797338e-01 4.39469993e-01 -6.98012114e-01 1.31134415e+00
-1.83150813e-01 5.36253095e-01 -3.75869691e-01 2.61489511e-01
-6.01595491e-02 3.10844690e-01 6.71511173e-01 7.63158143e-01
-1.23500995e-01 3.02715540e-01 5.22117078e-01 8.99591386e-01
-1.53284548e-02 -4.58219945e-02 -3.40717643e-01 2.67410338e-01
4.50675189e-01 1.37962115e+00 -1.06389368e+00 -6.94535732e-01
-3.95135164e-01 1.16493475e+00 6.43066108e-01 4.83578220e-02
-9.57554996e-01 -3.20734203e-01 2.31903300e-01 1.18571058e-01
6.42561793e-01 3.07801515e-01 6.90172762e-02 -1.35312545e+00
2.00477600e-01 -3.70881170e-01 6.98464572e-01 -8.49200189e-01
-1.32146108e+00 4.78244483e-01 -1.84610188e-01 -1.20111656e+00
2.45073795e-01 -4.99559283e-01 -4.17261362e-01 5.73999226e-01
-1.71344829e+00 -1.32047188e+00 -4.02902216e-01 4.66278553e-01
9.78333533e-01 -6.21707737e-02 7.30335593e-01 1.46473095e-01
-5.71718991e-01 3.67685765e-01 1.87320173e-01 2.35938370e-01
7.38189220e-01 -1.05113399e+00 1.77423865e-01 9.27665949e-01
2.03718230e-01 3.09399962e-01 5.51516294e-01 -1.01921833e+00
-9.06187952e-01 -1.55961251e+00 4.58963245e-01 -7.22489059e-01
3.99523288e-01 -3.76325667e-01 -1.27944112e+00 8.53125334e-01
-3.46523345e-01 9.20895994e-01 3.78590435e-01 -1.39495909e-01
-4.30411071e-01 -1.03969812e-01 -1.34208107e+00 2.06725925e-01
8.20976615e-01 -2.63631523e-01 -8.90920460e-01 7.13326037e-01
8.71051848e-01 -7.33292699e-01 -3.95324022e-01 4.00094301e-01
4.22895789e-01 -5.83966315e-01 9.03589427e-01 -8.46310616e-01
-6.38444424e-02 -7.07524359e-01 -1.21246263e-01 -1.03693438e+00
-2.11441666e-01 -1.58327341e-01 -4.78613466e-01 1.25788414e+00
1.79765865e-01 -6.93022430e-01 6.90037310e-01 6.52772248e-01
-2.51887832e-02 -5.25991440e-01 -1.05979753e+00 -7.93692648e-01
-3.70408207e-01 -5.75259440e-02 3.06550115e-01 9.58541274e-01
-4.97078627e-01 2.84262449e-01 -2.39760935e-01 6.71627641e-01
6.70197725e-01 3.95960033e-01 8.22876930e-01 -1.44336307e+00
-1.83208734e-01 3.60733986e-01 -6.18267953e-01 -8.29728901e-01
3.48745622e-02 -7.25142479e-01 2.15260163e-01 -1.15687990e+00
8.03567469e-01 -8.00014198e-01 -4.10355836e-01 7.04846203e-01
-6.30977035e-01 2.57073760e-01 4.10450362e-02 4.14841533e-01
-1.11281478e+00 5.57628453e-01 1.17491746e+00 -1.92164108e-01
-2.23568693e-01 1.77084044e-01 -4.19903457e-01 9.47644711e-01
3.97618592e-01 -8.46020937e-01 -2.65321016e-01 -1.66000396e-01
-2.54669696e-01 -1.89228252e-01 6.32334173e-01 -1.18166351e+00
2.17426911e-01 -2.03204349e-01 8.13876271e-01 -8.40306818e-01
2.64331013e-01 -1.09504414e+00 7.11171627e-02 6.54565692e-01
-1.34048685e-01 -6.04716718e-01 3.91077459e-01 1.20746398e+00
1.75330460e-01 -4.77240443e-01 1.00903153e+00 -1.93753049e-01
-1.09006619e+00 4.27621067e-01 -3.73461843e-02 1.41368717e-01
1.69618773e+00 -4.66323048e-01 -3.32567155e-01 3.78238946e-01
-1.06673467e+00 3.79912972e-01 3.13302994e-01 7.08358586e-01
5.82852244e-01 -1.01054823e+00 -5.54519475e-01 4.32281733e-01
5.44382095e-01 4.83246595e-01 2.94608980e-01 3.02469373e-01
-3.83908421e-01 -3.94761330e-04 -1.07261851e-01 -7.71452785e-01
-1.38405895e+00 9.84181762e-01 4.79835331e-01 -8.07898268e-02
-6.41667306e-01 1.02327764e+00 5.87179303e-01 -4.12024856e-01
3.31808627e-01 -2.05089182e-01 -2.16773763e-01 9.17437952e-03
6.92561746e-01 3.46944779e-01 4.58119288e-02 -6.15196466e-01
-3.68209958e-01 1.65325955e-01 -6.50794387e-01 4.69434321e-01
1.18712473e+00 -2.50025600e-01 9.25093368e-02 6.94816947e-01
6.75264895e-01 -2.29225412e-01 -1.57804823e+00 -6.13888681e-01
-1.67567968e-01 -7.27562308e-01 -1.09423287e-01 -9.50814545e-01
-8.21800947e-01 6.38480961e-01 1.18542206e+00 -1.65665820e-01
8.27595830e-01 2.10221782e-01 5.36660194e-01 5.15058756e-01
5.12564123e-01 -1.04387832e+00 1.05331004e-01 4.25034285e-01
4.96473789e-01 -1.53392792e+00 1.12701230e-01 -6.03046477e-01
-5.75626612e-01 1.01420939e+00 1.16098785e+00 -6.83997199e-02
5.31364679e-01 1.12938099e-01 -8.68204013e-02 1.09322220e-02
-6.67467415e-01 -2.50198036e-01 2.92610824e-01 5.90886056e-01
-2.97119468e-01 6.33487254e-02 2.41113514e-01 5.51379502e-01
3.66420507e-01 -5.95856197e-02 4.18998033e-01 1.09087789e+00
-9.75805342e-01 -5.69707632e-01 -6.17663801e-01 5.53498387e-01
-2.96166569e-01 -3.79344076e-02 -2.94930696e-01 5.45776010e-01
6.54383361e-01 5.01793325e-01 1.38101667e-01 -1.63616072e-02
2.13239193e-01 4.16833818e-01 3.40269029e-01 -9.73513961e-01
-2.19570696e-01 -2.13367298e-01 -1.46177322e-01 -2.11746395e-01
-1.66441157e-01 -6.25444293e-01 -1.32017696e+00 3.02121371e-01
-1.20082259e+00 -7.25119263e-02 3.38581711e-01 9.89435732e-01
3.77599746e-01 5.90686321e-01 3.70846838e-01 -8.71512711e-01
-6.40487194e-01 -8.16951036e-01 -2.72551864e-01 2.24910602e-01
5.02297640e-01 -9.37259555e-01 -1.17997974e-01 4.48048919e-01]
|
[9.366954803466797, 1.4811415672302246]
|
678d8818-d0fa-4da0-891b-5af3f80db6c8
|
sequence-guided-protein-structure
|
2007.06847
| null |
https://arxiv.org/abs/2007.06847v3
|
https://arxiv.org/pdf/2007.06847v3.pdf
|
Sequence-guided protein structure determination using graph convolutional and recurrent networks
|
Single particle, cryogenic electron microscopy (cryo-EM) experiments now routinely produce high-resolution data for large proteins and their complexes. Building an atomic model into a cryo-EM density map is challenging, particularly when no structure for the target protein is known a priori. Existing protocols for this type of task often rely on significant human intervention and can take hours to many days to produce an output. Here, we present a fully automated, template-free model building approach that is based entirely on neural networks. We use a graph convolutional network (GCN) to generate an embedding from a set of rotamer-based amino acid identities and candidate 3-dimensional C$\alpha$ locations. Starting from this embedding, we use a bidirectional long short-term memory (LSTM) module to order and label the candidate identities and atomic locations consistent with the input protein sequence to obtain a structural model. Our approach paves the way for determining protein structures from cryo-EM densities at a fraction of the time of existing approaches and without the need for human intervention.
|
['Saulo H. P. de Oliveira', 'Po-Nan Li', 'Soichi Wakatsuki', 'Henry van den Bedem']
|
2020-07-14
| null | null | null | null |
['cryogenic-electron-microscopy-cryo-em']
|
['computer-vision']
|
[ 3.34366918e-01 2.76169982e-02 4.22984302e-01 -5.51421344e-01
-7.79631913e-01 -5.20024240e-01 3.49353194e-01 5.21706641e-01
-7.41690338e-01 1.08717716e+00 -2.56405920e-01 -7.98758149e-01
1.98856056e-01 -6.32848978e-01 -1.00417113e+00 -7.51430988e-01
-9.69665870e-02 1.01857901e+00 1.20892107e-01 -4.52196822e-02
3.75177234e-01 7.92353094e-01 -1.12361228e+00 4.19164389e-01
6.40771568e-01 4.79215443e-01 8.61629426e-01 6.89765453e-01
-1.77299216e-01 5.52898645e-01 -2.01202124e-01 -1.16236754e-01
-5.86026497e-02 -5.27466059e-01 -1.17927492e+00 -6.02812432e-02
1.28347233e-01 -5.19340225e-02 4.44254726e-02 8.35182309e-01
5.48374295e-01 1.14585951e-01 6.16530895e-01 -3.73369247e-01
-6.56260610e-01 1.22684417e-02 -1.95489690e-01 1.19659171e-01
2.97662109e-01 4.14414883e-01 8.43418419e-01 -1.05115998e+00
1.28766465e+00 9.65076745e-01 7.10413516e-01 5.34754694e-01
-1.88510501e+00 -2.42524698e-01 4.06145416e-02 4.58840907e-01
-1.16334713e+00 -1.52828038e-01 5.39421380e-01 -5.31204641e-01
1.70567906e+00 -2.71761268e-01 6.87353730e-01 8.18807960e-01
5.39298534e-01 -9.76581872e-02 9.99782085e-01 -3.93276662e-01
3.57938886e-01 -2.64076769e-01 1.06119007e-01 8.96668673e-01
-6.51925653e-02 -2.03241810e-01 -1.55436054e-01 -4.49475855e-01
4.54886794e-01 1.95456892e-01 -1.89888388e-01 -6.82636976e-01
-1.02805805e+00 7.85143077e-01 6.93543315e-01 2.16284186e-01
-6.79437637e-01 4.90724407e-02 2.40191624e-01 1.76677331e-01
2.45873153e-01 6.21506035e-01 -4.79164600e-01 5.32237366e-02
-9.13296819e-01 2.36887574e-01 6.94217443e-01 3.72093678e-01
1.19396341e+00 -3.80040526e-01 7.65498221e-01 4.64008033e-01
3.11945826e-01 6.25086483e-04 2.24303946e-01 -6.17489934e-01
-7.78798088e-02 7.24293172e-01 3.12837839e-01 -5.40180743e-01
-6.86973453e-01 3.90966654e-01 -6.24675035e-01 4.59482014e-01
2.24919096e-01 -8.30564201e-02 -1.17440927e+00 1.63225019e+00
4.69492823e-01 -1.19269907e-01 -1.48738042e-01 8.77868056e-01
4.18409884e-01 8.71830404e-01 1.83770627e-01 -2.58719891e-01
1.14526141e+00 -6.10366166e-01 -1.53053015e-01 -3.05728596e-02
8.25346828e-01 -5.56092620e-01 6.36249959e-01 8.59199017e-02
-9.19804633e-01 -3.94522429e-01 -1.28101742e+00 -4.32952583e-01
-6.74416840e-01 -1.23184904e-01 3.67778331e-01 3.91634144e-02
-1.24636316e+00 1.13662529e+00 -1.13771284e+00 -2.90461361e-01
1.93595923e-02 8.32150578e-01 -9.76869106e-01 4.98178974e-02
-8.12987447e-01 1.44318140e+00 7.18091786e-01 1.30343422e-01
-9.11481678e-01 -5.70104539e-01 -8.65444362e-01 2.52180733e-02
3.15749012e-02 -6.50607526e-01 9.95754361e-01 -6.94552004e-01
-1.25340974e+00 9.43114638e-01 -6.21665895e-01 -5.58255970e-01
-4.39867452e-02 2.27438524e-01 5.74471727e-02 3.10558081e-01
-1.54726520e-01 9.12211359e-01 4.13787901e-01 -1.17926824e+00
-1.75993040e-01 -4.32058990e-01 -1.21934749e-01 -2.15110760e-02
5.85133791e-01 1.98934972e-01 1.77400917e-01 2.08853170e-01
3.43367606e-01 -9.60706532e-01 -7.04537392e-01 -6.56565130e-02
-1.54718176e-01 -1.14546172e-01 6.94521070e-01 -8.38220716e-01
6.37079000e-01 -1.53274870e+00 5.94348013e-01 3.16197366e-01
5.76227665e-01 4.63899463e-01 1.70973390e-02 8.19100559e-01
-3.92289430e-01 1.52826020e-02 -3.81027430e-01 -2.00272560e-01
-6.65333793e-02 4.19566320e-04 1.06873259e-01 6.25191391e-01
4.65386391e-01 9.24971402e-01 -6.61100447e-01 -2.27100044e-01
4.71627176e-01 8.33716035e-01 -4.65627044e-01 3.50376010e-01
-6.48483396e-01 5.79444826e-01 3.51480581e-02 1.88860252e-01
6.49947822e-01 -7.50151038e-01 1.04301941e+00 -3.08599234e-01
-3.02092135e-01 6.78595126e-01 -7.33559489e-01 1.57967520e+00
-2.80222595e-01 2.30280146e-01 2.55833477e-01 -1.04503238e+00
1.07742512e+00 1.80257931e-01 5.27544618e-01 -4.38718587e-01
5.27565740e-02 2.66092569e-01 3.28100890e-01 -6.77180961e-02
2.30747655e-01 -5.97653151e-01 3.84170383e-01 8.39636147e-01
4.12323594e-01 1.83700576e-01 1.88365474e-01 1.26370355e-01
1.12556756e+00 5.28511703e-01 3.14567327e-01 -3.01782340e-01
5.20234525e-01 1.91487610e-01 5.28653204e-01 1.93126470e-01
7.17976093e-02 5.77127635e-01 4.48893934e-01 -1.29987025e+00
-1.83183408e+00 -8.49489033e-01 8.06935057e-02 8.66919100e-01
-1.80711240e-01 -5.68654180e-01 -9.20318604e-01 -4.27432775e-01
-1.83480278e-01 7.23583773e-02 -5.47301829e-01 -7.52339587e-02
-8.11151981e-01 -8.34752262e-01 -5.24424091e-02 2.71189839e-01
-2.58297503e-01 -1.54752946e+00 -7.10240066e-01 7.45168984e-01
1.42427742e-01 -7.90691376e-01 -1.70894951e-01 8.77422214e-01
-8.21720302e-01 -1.18438697e+00 -5.08643568e-01 -9.25300300e-01
1.05053222e+00 -7.97176659e-02 1.10054135e+00 1.91029415e-01
-6.00545585e-01 -4.84410256e-01 6.67459369e-02 -7.77278170e-02
-4.39494401e-01 1.46136820e-01 2.98471451e-01 -2.18057975e-01
8.54025900e-01 -8.17729414e-01 -5.36895335e-01 9.84483585e-03
-7.46102631e-01 3.60755771e-01 4.33044106e-01 9.53126848e-01
1.09080589e+00 -4.81442869e-01 4.71246749e-01 -1.01576614e+00
6.13096714e-01 -3.14762592e-01 -7.89050877e-01 1.94933176e-01
-6.86784029e-01 4.66197103e-01 1.06132638e+00 -3.57173830e-02
-6.95611775e-01 6.18996322e-01 -4.84990090e-01 -3.26496065e-01
-3.81282955e-01 4.89602178e-01 -1.33156702e-01 -2.42499277e-01
5.35119951e-01 4.55291212e-01 2.23790243e-01 -7.20260262e-01
1.84618607e-01 4.19526070e-01 4.22944427e-01 -6.26584470e-01
3.39305371e-01 4.90342826e-01 6.26693517e-02 -7.12687910e-01
-3.84858191e-01 -3.10373038e-01 -1.27878726e+00 2.40971759e-01
9.42221820e-01 -5.82517445e-01 -1.05140638e+00 1.36545941e-01
-1.26978207e+00 -2.66364753e-01 2.99414903e-01 4.29338396e-01
-7.29738772e-01 6.00284934e-01 -7.15393841e-01 -4.15530562e-01
-6.38428867e-01 -1.46464217e+00 1.06400657e+00 -1.27192149e-02
-3.51777345e-01 -7.67596662e-01 3.49521607e-01 1.28117472e-01
2.63168007e-01 1.80998325e-01 1.35428119e+00 -4.28909451e-01
-7.01362550e-01 -3.35093611e-03 -2.63431072e-01 -3.03390510e-02
3.29622962e-02 -6.07390702e-02 -5.91215968e-01 -3.75380009e-01
-2.68692732e-01 -6.21956825e-01 7.95741558e-01 3.57037246e-01
7.25300670e-01 -9.46786702e-02 -3.45846027e-01 6.54274225e-01
1.46821642e+00 3.50934446e-01 6.18528128e-01 4.74053741e-01
7.55892754e-01 5.53478897e-01 1.59759060e-01 1.88373312e-01
1.18059814e-01 5.48886418e-01 3.38838756e-01 -2.58538187e-01
4.27085191e-01 -3.54340464e-01 -3.00804432e-03 5.24998546e-01
-3.14115942e-01 1.36199251e-01 -1.04648030e+00 4.64246213e-01
-1.68454909e+00 -9.82861876e-01 3.26624632e-01 1.97184730e+00
1.05050814e+00 -7.29689002e-03 1.58150613e-01 -1.89449191e-01
7.04588532e-01 -1.42150372e-01 -8.79604161e-01 -6.81741476e-01
1.03271894e-01 5.80536187e-01 3.66074860e-01 7.56023407e-01
-7.85968065e-01 1.00749946e+00 6.62317085e+00 1.91270169e-02
-1.40849352e+00 -1.58690095e-01 5.13383448e-01 -1.14601873e-01
-1.53712854e-01 3.51634949e-01 -7.48195231e-01 4.63981688e-01
1.50303876e+00 -1.03074443e-02 6.01293504e-01 7.65164495e-01
3.05618852e-01 -2.79679652e-02 -1.01116359e+00 9.43411469e-01
-4.43228006e-01 -1.91172945e+00 -1.48397386e-02 3.83532822e-01
2.73164958e-01 4.34618533e-01 -6.13423228e-01 -2.23714367e-01
5.74945271e-01 -1.31778800e+00 2.26057649e-01 4.45649683e-01
7.88185239e-01 -1.06176603e+00 6.28744900e-01 3.47770929e-01
-1.02165353e+00 5.28125226e-01 -8.63665938e-01 -2.91860867e-02
6.05108261e-01 2.91087151e-01 -1.04223251e+00 1.42314613e-01
4.62086707e-01 3.64285916e-01 -1.71280771e-01 5.09677768e-01
8.20241943e-02 2.03785300e-01 -3.67306590e-01 -5.88672534e-02
3.73867333e-01 -6.07956886e-01 -1.45704225e-02 1.09011602e+00
-3.95885110e-02 1.16507806e-01 2.56260008e-01 1.07493150e+00
-3.65680754e-01 -2.45319828e-02 -5.55719197e-01 -2.51005024e-01
2.16701180e-01 1.34153152e+00 -8.05259168e-01 -2.52574801e-01
-2.31350631e-01 9.86037135e-01 9.83932495e-01 3.04874815e-02
-6.10242248e-01 -4.59136397e-01 8.18701804e-01 2.46371418e-01
5.44348359e-01 -5.59566319e-01 3.35207224e-01 -9.16031837e-01
2.07874179e-02 -9.91210222e-01 -4.44392450e-02 -8.00844550e-01
-9.28637207e-01 7.37662613e-01 -5.22638261e-01 -4.63645607e-01
-4.37146902e-01 -8.54744792e-01 -5.05672157e-01 1.36866522e+00
-1.16688085e+00 -8.39194238e-01 2.42771924e-01 8.61500651e-02
4.52862941e-02 9.71568748e-02 1.26849735e+00 1.32999077e-01
-3.83031100e-01 4.22104672e-02 3.27618033e-01 -1.99964538e-01
4.79104638e-01 -1.33944225e+00 8.94595742e-01 5.40517986e-01
2.98070908e-02 1.17234671e+00 8.39938760e-01 -7.38273740e-01
-1.50622976e+00 -9.42378879e-01 1.21210277e+00 -3.95686120e-01
4.81043935e-01 -5.93458891e-01 -1.25480235e+00 7.15669036e-01
2.10985169e-01 4.58729453e-02 7.93541014e-01 1.95332672e-02
-2.74753958e-01 4.27263796e-01 -1.19776785e+00 3.56566697e-01
8.21022153e-01 -8.52568448e-01 -4.81656283e-01 4.23897892e-01
6.33006632e-01 -1.30508333e-01 -9.70285118e-01 2.17781544e-01
4.34489876e-01 -9.01700437e-01 9.45678115e-01 -1.17254877e+00
2.38981575e-01 -5.78865826e-01 7.58985132e-02 -1.09043801e+00
-5.95267653e-01 -6.64096594e-01 4.19662446e-02 2.56617159e-01
7.98145115e-01 -4.50123459e-01 1.00968242e+00 7.85001576e-01
-2.24688828e-01 -9.16706204e-01 -9.34920490e-01 -3.93555552e-01
1.51230246e-01 1.59266070e-01 5.44274509e-01 6.91967010e-01
2.23700210e-01 7.01415658e-01 -3.69711906e-01 1.53488191e-02
4.18514401e-01 3.91375095e-01 6.39393866e-01 -1.12862992e+00
-1.74251139e-01 2.31907055e-01 -5.05050719e-01 -8.15885901e-01
3.56625438e-01 -9.32153523e-01 7.46624619e-02 -1.68573523e+00
3.35274637e-01 -5.07273823e-02 -3.15969139e-01 5.58682561e-01
1.55474931e-01 1.70656443e-01 -1.01109341e-01 2.12596014e-01
-7.17118680e-01 4.16250378e-01 8.35677028e-01 5.05971164e-02
-1.10763967e-01 -5.62676609e-01 -3.65211099e-01 5.32240152e-01
7.67861903e-01 -6.59826696e-01 -1.40954286e-01 -1.67090118e-01
2.72160530e-01 -8.88938010e-02 1.93012401e-01 -8.97287846e-01
1.04187287e-01 -9.76532921e-02 5.20412564e-01 -7.40211308e-01
4.94076371e-01 -6.60669327e-01 5.16025603e-01 5.52057683e-01
-9.55941677e-02 5.03282070e-01 1.37920648e-01 4.29574966e-01
7.31001049e-02 -2.56834805e-01 1.13068771e+00 -5.65107405e-01
-4.45005566e-01 3.29018235e-01 -6.53628409e-01 -5.27172685e-01
7.34400630e-01 -1.45771772e-01 -2.46252671e-01 5.02397865e-02
-1.14615691e+00 -8.83481354e-02 1.06650639e+00 -1.04523584e-01
7.07789242e-01 -8.88352871e-01 -3.30382884e-01 3.94949973e-01
-2.44661160e-02 -8.30813870e-02 1.87187344e-01 5.40613413e-01
-1.18928885e+00 7.50477552e-01 -5.90415716e-01 -4.30607259e-01
-1.26867533e+00 8.10260296e-01 5.86752534e-01 -3.03859383e-01
-8.40866327e-01 4.75620955e-01 2.08294820e-02 -7.70211220e-01
-4.29062068e-01 -1.73374608e-01 5.83653152e-02 -3.77005428e-01
7.12172091e-01 -1.75231680e-01 2.89115310e-01 -9.30519938e-01
-3.38712811e-01 2.84230381e-01 -5.43043196e-01 1.08649559e-01
1.88471580e+00 6.93643615e-02 -2.96226531e-01 6.73362911e-02
1.30334640e+00 -6.05397701e-01 -1.42447090e+00 -9.92317498e-02
1.13770410e-01 3.60290445e-02 -2.32484877e-01 -4.91794020e-01
-4.86753941e-01 9.08878088e-01 5.87062538e-01 -2.43937731e-01
5.86558461e-01 -2.27510645e-05 9.21948791e-01 9.64915693e-01
5.52841544e-01 -1.03343701e+00 -4.95277226e-01 6.48099542e-01
4.27797049e-01 -1.23512149e+00 -8.78767297e-02 1.55740187e-01
-1.07307285e-01 1.19875288e+00 4.33324903e-01 -2.09581360e-01
4.19252753e-01 3.53283316e-01 1.20133765e-01 -5.90396523e-01
-1.11651552e+00 -9.16689858e-02 -1.56850711e-01 7.06481695e-01
7.27489769e-01 -9.81362537e-02 -2.26348102e-01 2.22635761e-01
-8.60535875e-02 -6.39544204e-02 3.68848264e-01 1.23236668e+00
-8.35394323e-01 -1.55114675e+00 -5.51744439e-02 3.67049605e-01
-5.76484799e-01 -2.06607476e-01 -8.34167898e-01 2.69372493e-01
-1.35819346e-01 4.73649442e-01 1.07615609e-02 -4.01641130e-01
4.67221066e-02 6.03529096e-01 6.04552269e-01 -7.17057407e-01
-4.92821276e-01 -7.97453150e-03 -5.45209870e-02 -4.82414722e-01
-3.48273546e-01 -3.90056700e-01 -1.64798212e+00 -7.17081785e-01
-3.00612867e-01 3.51953149e-01 8.67592692e-01 9.08329070e-01
9.58586097e-01 1.11993209e-01 3.03822249e-01 -1.30667257e+00
-1.29351974e-01 -9.64407027e-01 -4.74430203e-01 4.86240864e-01
1.74771622e-01 -5.05728543e-01 1.71724393e-03 2.42595419e-01]
|
[13.347379684448242, -3.0915119647979736]
|
8dfd1560-4e3e-4060-aee2-d67a71002e3f
|
video-text-tracking-for-dense-and-small-text
|
2304.00018
| null |
https://arxiv.org/abs/2304.00018v1
|
https://arxiv.org/pdf/2304.00018v1.pdf
|
Video text tracking for dense and small text based on pp-yoloe-r and sort algorithm
|
Although end-to-end video text spotting methods based on Transformer can model long-range dependencies and simplify the train process, it will lead to large computation cost with the increase of the frame size in the input video. Therefore, considering the resolution of ICDAR 2023 DSText is 1080 * 1920 and slicing the video frame into several areas will destroy the spatial correlation of text, we divided the small and dense text spotting into two tasks, text detection and tracking. For text detection, we adopt the PP-YOLOE-R which is proven effective in small object detection as our detection model. For text detection, we use the sort algorithm for high inference speed. Experiments on DSText dataset demonstrate that our method is competitive on small and dense text spotting.
|
['Hongen Liu']
|
2023-03-31
| null | null | null | null |
['text-spotting', 'small-object-detection']
|
['computer-vision', 'computer-vision']
|
[ 2.25758359e-01 -6.97720647e-01 -1.62353247e-01 -7.03874230e-02
-6.22167230e-01 -2.99412787e-01 4.10006195e-01 -2.28184149e-01
-6.64394259e-01 3.93889666e-01 1.24921583e-01 -5.16672790e-01
2.98614442e-01 -5.52826762e-01 -6.91306710e-01 -7.13934541e-01
4.52370644e-01 5.41032195e-01 7.93613374e-01 4.24050748e-01
2.90010601e-01 1.77820668e-01 -9.80069160e-01 6.19186997e-01
6.40686691e-01 1.04755700e+00 5.10514617e-01 1.03989601e+00
-3.64614010e-01 1.13705349e+00 -7.35969543e-01 -4.34055269e-01
1.84479997e-01 -2.31245801e-01 -4.30603683e-01 1.96216747e-01
7.45397210e-01 -8.76485825e-01 -9.16250825e-01 1.01641786e+00
6.01389587e-01 -8.55452716e-02 4.96052921e-01 -1.01336229e+00
-3.16329241e-01 7.39722908e-01 -1.08633471e+00 4.13161457e-01
3.92201006e-01 -1.75516650e-01 9.73192513e-01 -1.32306683e+00
5.41298091e-01 1.24188161e+00 7.74615884e-01 5.30879974e-01
-6.72099710e-01 -9.60109711e-01 2.87545621e-01 2.91768014e-01
-1.51798379e+00 -3.76691878e-01 3.47434312e-01 -1.12294696e-01
9.00196075e-01 3.88796419e-01 4.88995463e-01 1.02829611e+00
2.42447063e-01 1.46524668e+00 5.41273534e-01 -3.37936729e-01
-1.75682113e-01 -1.89786762e-01 4.87453416e-02 9.05514359e-01
2.64971346e-01 -2.08082944e-01 -9.22711968e-01 6.88155517e-02
8.59906197e-01 2.77291059e-01 -3.29649150e-01 3.42694938e-01
-1.43771648e+00 5.43048978e-01 -8.80612619e-03 1.42055258e-01
2.44283885e-01 3.32684219e-01 6.16483331e-01 2.98553109e-01
4.70168293e-01 -3.44396651e-01 -2.13216826e-01 -1.73761636e-01
-1.44813704e+00 -1.54346824e-01 3.85998905e-01 1.20178592e+00
1.79210186e-01 -1.63030535e-01 -4.21078920e-01 8.24700534e-01
3.95040214e-01 1.04277372e+00 3.00405383e-01 -1.95850521e-01
1.11713731e+00 4.64037031e-01 -2.33954713e-01 -9.01088595e-01
-2.72891372e-01 2.03919575e-01 -1.05788946e+00 -2.20213771e-01
5.69840848e-01 -3.17461163e-01 -1.19515729e+00 8.75733554e-01
-1.81541983e-02 2.85312027e-01 -5.43709159e-01 7.92320430e-01
6.78221881e-01 8.58019829e-01 2.69554798e-02 -2.66693205e-01
1.44457972e+00 -1.00017929e+00 -9.72914696e-01 -2.80728966e-01
8.56573224e-01 -9.37310874e-01 9.88789558e-01 4.60427284e-01
-8.80088329e-01 -3.02691728e-01 -7.16285348e-01 -3.75926077e-01
-3.06774080e-01 5.81014872e-01 3.13446373e-01 6.71492577e-01
-8.07478547e-01 -1.77670680e-02 -8.23390007e-01 -4.54387724e-01
7.29091048e-01 4.05627638e-01 4.82186899e-02 -2.82894135e-01
-1.09923804e+00 3.34058464e-01 4.44815248e-01 5.72195835e-02
-6.55867398e-01 -3.17339450e-01 -4.41023737e-01 2.40469843e-01
5.88967204e-01 -4.38244641e-01 7.64389634e-01 -5.77347338e-01
-1.32477939e+00 7.24240184e-01 -3.68854195e-01 -5.61789036e-01
8.64998221e-01 -3.61015975e-01 -3.41487229e-01 3.86593521e-01
1.87921040e-02 7.19160318e-01 1.39121354e+00 -3.59488338e-01
-1.18688118e+00 -2.58095324e-01 -4.89589632e-01 3.58161747e-01
-7.12391257e-01 4.45173591e-01 -1.15648544e+00 -9.87347782e-01
5.62775806e-02 -7.76646912e-01 2.56741732e-01 2.45640755e-01
-6.14024520e-01 -3.38381946e-01 1.49912524e+00 -6.91510081e-01
1.57242382e+00 -2.18545985e+00 -3.17532122e-01 2.11694241e-01
4.81542498e-01 1.49779111e-01 1.66136786e-01 -3.23132728e-03
1.32206112e-01 -3.31739476e-03 2.49530464e-01 -4.65225130e-01
-1.61090210e-01 -1.80115119e-01 -6.71126127e-01 6.29741073e-01
-3.29973787e-01 9.63556588e-01 -4.17106152e-01 -1.12184882e+00
5.22425830e-01 3.31112236e-01 -2.92913973e-01 -1.05464216e-02
-2.35489056e-01 -2.71411747e-01 -7.63779998e-01 7.90897846e-01
6.63299441e-01 -5.48957467e-01 -1.17624663e-01 -7.40340352e-02
2.00455636e-01 1.81951816e-03 -9.99004066e-01 1.52602696e+00
-1.93171613e-02 1.23480928e+00 -1.30648941e-01 -5.04139304e-01
5.85350990e-01 2.62688220e-01 5.25918126e-01 -6.99881673e-01
2.00282931e-01 -1.45595297e-01 -5.02338767e-01 -6.08413577e-01
6.53398931e-01 1.19807765e-01 2.59130239e-01 6.49174273e-01
-4.81615841e-01 2.36931160e-01 3.33247036e-01 4.87602890e-01
1.27275336e+00 -3.61833200e-02 -3.48907888e-01 8.15222263e-02
4.36576039e-01 -8.60114619e-02 1.64956018e-01 9.74007964e-01
3.89547162e-02 6.56968474e-01 5.10206103e-01 -4.83337402e-01
-9.62946713e-01 -7.17727005e-01 -3.19360077e-01 1.34592307e+00
3.53076875e-01 -7.16168046e-01 -6.77170038e-01 -1.02221894e+00
-1.47740826e-01 3.60152632e-01 -3.96108061e-01 2.47671917e-01
-9.89938259e-01 -1.12937665e+00 9.67641175e-01 6.55017257e-01
7.15571225e-01 -9.03204739e-01 -5.04595399e-01 2.03441978e-02
-3.63690704e-01 -1.52890754e+00 -1.16990125e+00 -3.92550696e-03
-8.23796093e-01 -1.14161921e+00 -1.05678058e+00 -9.55997825e-01
7.44299710e-01 5.03729522e-01 9.02399421e-01 1.57467276e-01
-4.66505677e-01 2.68878132e-01 -3.72246087e-01 -4.94787395e-02
-1.39795139e-01 -1.72157213e-01 -2.67290622e-01 -1.86721817e-01
5.93609273e-01 2.91732490e-01 -5.19732952e-01 4.64602560e-01
-9.58955109e-01 3.55339766e-01 5.28615355e-01 8.65055263e-01
3.15901667e-01 5.69601178e-01 -1.48811102e-01 -7.72736371e-01
4.05223548e-01 9.44262221e-02 -7.40583777e-01 4.68854994e-01
-6.31931961e-01 -3.06311071e-01 5.44546306e-01 -6.15817666e-01
-9.36508179e-01 3.28613102e-01 5.13376296e-02 -6.29022479e-01
1.88275531e-01 1.97606176e-01 1.43107817e-01 9.13538970e-03
1.89381450e-01 7.28129089e-01 -3.76396239e-01 -2.72303700e-01
8.73174705e-03 9.68269706e-01 3.18457186e-01 -1.88764289e-01
7.14125693e-01 5.89267552e-01 -2.02245399e-01 -1.12239623e+00
-4.79896158e-01 -7.13724554e-01 -3.78061146e-01 -1.17541514e-01
9.96739864e-01 -1.15108597e+00 -8.87686551e-01 8.22405219e-01
-1.09637988e+00 -3.46114278e-01 2.50639498e-01 3.21889669e-01
-2.62389809e-01 8.84929001e-01 -9.34980094e-01 -6.47718787e-01
-6.00739002e-01 -9.39630330e-01 1.56036377e+00 -2.34726280e-01
2.36894757e-01 -8.60256970e-01 -3.35238069e-01 2.95614481e-01
3.76636977e-03 -5.30276299e-01 7.36983657e-01 -4.02092874e-01
-8.40482891e-01 -4.79120970e-01 -6.82011247e-01 -5.86966760e-02
-1.78289369e-01 -7.27615901e-04 -6.64408207e-01 -3.29897344e-01
-8.97232890e-02 -8.65522921e-02 1.21007764e+00 5.99775314e-01
1.46457505e+00 -2.66888030e-02 -7.67574131e-01 5.23642063e-01
1.22078538e+00 1.62756681e-01 8.33654225e-01 2.49086291e-01
1.26841450e+00 7.54185170e-02 8.28374565e-01 4.15992796e-01
1.63766439e-03 5.65303028e-01 -8.71592760e-02 -2.29384989e-01
-3.12009662e-01 -2.41153404e-01 7.26697087e-01 4.18728292e-01
2.99747646e-01 -9.10516560e-01 -8.78497481e-01 2.89209727e-02
-2.00412607e+00 -1.09969842e+00 -3.17763567e-01 2.03103447e+00
6.13219619e-01 4.38693613e-01 1.73303157e-01 1.64117873e-01
1.17104423e+00 2.26396486e-01 -4.53556120e-01 2.93831825e-01
-2.38916159e-01 -1.78717345e-01 9.65812325e-01 3.43277872e-01
-1.07334399e+00 1.30779386e+00 6.76370335e+00 1.33662248e+00
-1.08080077e+00 -1.08967282e-01 8.71174455e-01 -4.64560449e-01
4.59890455e-01 -4.96762276e-01 -1.37986875e+00 8.32514942e-01
7.05698133e-01 1.18773438e-01 2.35938743e-01 5.26869416e-01
3.39096218e-01 -2.48953059e-01 -1.04492128e+00 1.51264930e+00
2.87371039e-01 -1.38539767e+00 2.66773760e-01 -2.16052204e-01
3.50196183e-01 1.61866859e-01 -9.21769366e-02 1.23276353e-01
-6.82678446e-02 -9.81725514e-01 5.51295877e-01 1.02217220e-01
1.12640405e+00 -4.94658470e-01 5.40262341e-01 2.47447222e-01
-1.39708865e+00 -4.66390094e-03 -5.06410122e-01 5.49119949e-01
7.97657967e-02 7.18006015e-01 -7.36759782e-01 -1.06989309e-01
9.06158268e-01 1.01296711e+00 -6.51449382e-01 1.02198410e+00
8.10955241e-02 6.37827873e-01 -7.68629909e-01 -2.55944103e-01
1.09134175e-01 -1.24155261e-01 4.72488254e-01 1.41551566e+00
3.23699206e-01 -1.49059847e-01 2.33696818e-01 4.62030888e-01
-3.79497230e-01 -1.03813531e-02 -3.51492107e-01 -4.18765768e-02
2.67398894e-01 8.76181960e-01 -1.34791791e+00 -6.60807967e-01
-7.08009541e-01 1.17556262e+00 -2.87868828e-01 3.70636672e-01
-1.16476440e+00 -4.83289331e-01 -1.30951658e-01 1.60258546e-01
7.19515085e-01 -8.15466493e-02 -4.27654803e-01 -1.35150647e+00
1.40495375e-01 -9.73062515e-01 5.23628652e-01 -8.50404143e-01
-8.22246909e-01 3.93539220e-01 -4.00046051e-01 -1.10518098e+00
2.27600962e-01 -9.35602009e-01 -4.42524523e-01 3.17986846e-01
-1.11936021e+00 -1.07885528e+00 -4.05143350e-01 9.39424992e-01
1.20624840e+00 -1.11868426e-01 2.93826342e-01 6.24909341e-01
-9.25762355e-01 8.92721355e-01 3.72043073e-01 6.56119585e-01
9.47404265e-01 -9.38641548e-01 5.68514645e-01 8.88331294e-01
3.16250265e-01 1.05900392e-02 5.05532444e-01 -1.10123730e+00
-1.56057680e+00 -1.06158614e+00 6.49940133e-01 -6.86110795e-01
8.18340778e-01 -7.01236129e-01 -8.13445389e-01 7.35157609e-01
-1.81243662e-02 1.41434699e-01 1.74578145e-01 -1.03679992e-01
-1.75707057e-01 8.00078288e-02 -7.51442075e-01 6.96171880e-01
9.78844941e-01 -5.75408340e-01 -3.11059386e-01 6.88072860e-01
3.99163038e-01 -6.15460336e-01 -4.18426216e-01 -3.17274034e-03
5.80833912e-01 -7.56581724e-01 8.61391842e-01 -3.38812359e-02
2.37902686e-01 -2.21952111e-01 1.32172048e-01 -3.12464893e-01
-3.93072376e-03 -6.00123584e-01 -3.07028234e-01 1.07326376e+00
2.45490551e-01 -2.39558727e-01 1.29924798e+00 4.77342695e-01
2.79843003e-01 -3.61861527e-01 -9.53647316e-01 -4.88914490e-01
-1.87131792e-01 -5.44054866e-01 2.69327104e-01 6.91101909e-01
1.81365222e-01 5.51347673e-01 -6.77136540e-01 1.67062916e-02
5.22985518e-01 3.44590247e-02 5.12758732e-01 -1.02078509e+00
-1.76536232e-01 -4.25420374e-01 -1.83467582e-01 -1.87531471e+00
-1.26121044e-02 -3.83291066e-01 1.94523737e-01 -1.19864047e+00
5.29294491e-01 -3.65085453e-01 -1.26026362e-01 3.59070957e-01
-3.65183681e-01 2.82101005e-01 1.04291700e-01 4.79088157e-01
-1.10220695e+00 3.80906194e-01 1.12951553e+00 -3.09937000e-01
6.00270517e-02 1.03803918e-01 -4.21780236e-02 7.68314660e-01
3.85043919e-01 -7.11487353e-01 -1.07429102e-01 -5.99639058e-01
4.92478371e-01 2.32547328e-01 1.11906193e-01 -8.07513058e-01
6.90149128e-01 1.67889833e-01 7.88489938e-01 -1.29174232e+00
1.04221404e-01 -1.15102828e+00 -4.85764444e-01 3.88438076e-01
-5.33405423e-01 -3.10048454e-05 2.24656224e-01 7.13841259e-01
1.50568169e-02 -2.35935926e-01 6.27573371e-01 2.95048416e-01
-6.18959427e-01 5.65919995e-01 -5.09184062e-01 -5.98351583e-02
8.60140622e-01 -3.24926138e-01 -5.62627614e-01 -2.61835039e-01
-2.98880011e-01 4.24100578e-01 2.97036141e-01 5.06694436e-01
8.02570522e-01 -9.72502530e-01 -5.80240905e-01 3.16058218e-01
-2.06684157e-01 1.01280831e-01 -8.71557463e-03 9.65467274e-01
-6.77707255e-01 6.67899966e-01 2.55868018e-01 -7.51280963e-01
-1.83255172e+00 6.27617359e-01 1.99820474e-01 -3.83265585e-01
-1.31916273e+00 8.20317626e-01 5.56522667e-01 3.70857000e-01
8.02992821e-01 -3.37241888e-01 -4.49362248e-02 -2.26095784e-02
9.19222057e-01 4.88605320e-01 -1.79224357e-01 -2.92568773e-01
-3.04416329e-01 7.94361591e-01 -5.77032447e-01 1.59134552e-01
8.21797013e-01 -5.48505127e-01 1.35182604e-01 3.19903381e-02
9.57301140e-01 7.10886195e-02 -1.31702995e+00 -1.91124588e-01
4.34811898e-02 -5.81036687e-01 5.51585793e-01 -6.16937459e-01
-1.16601634e+00 8.65756810e-01 7.75201499e-01 1.60321638e-01
1.13046992e+00 -1.77505493e-01 1.11954403e+00 8.20436954e-01
7.27264285e-02 -1.22136176e+00 9.61074978e-02 6.49496853e-01
2.41431758e-01 -1.33076227e+00 3.57167542e-01 -3.91591012e-01
-3.83739561e-01 1.32949233e+00 5.11957109e-01 1.17681146e-01
4.05681372e-01 8.01246822e-01 -5.24672568e-02 -1.96888387e-01
-7.65723228e-01 -4.22598980e-02 1.38245136e-01 1.29283801e-01
2.57748306e-01 -3.60889733e-01 2.49431774e-01 -1.84057876e-01
2.11875632e-01 9.79740173e-02 3.66349638e-01 6.58913791e-01
-6.65015578e-01 -7.56598651e-01 -5.80772281e-01 8.43227863e-01
-8.62945378e-01 -6.76341355e-01 -2.90368915e-01 7.21415341e-01
-2.69103199e-01 8.13912749e-01 3.30026239e-01 -2.40500733e-01
-1.06677614e-01 -6.38301745e-02 3.76026779e-01 -3.40603054e-01
-3.00481737e-01 6.99988186e-01 -7.70698786e-02 -3.02290142e-01
-7.52265006e-02 -5.35226107e-01 -1.46952438e+00 -6.90848112e-01
-7.83517003e-01 -1.40443444e-01 3.90697062e-01 1.08836138e+00
7.03762919e-02 4.57651615e-01 3.98037672e-01 -4.97129202e-01
-2.09142357e-01 -9.04050529e-01 -7.65230775e-01 1.49459809e-01
3.43532175e-01 -1.81490511e-01 -3.82280767e-01 4.96715307e-01]
|
[12.005133628845215, 2.191074848175049]
|
9242d0d9-9920-412d-84a0-6bebb572232b
|
trickvos-a-bag-of-tricks-for-video-object
|
2306.15377
| null |
https://arxiv.org/abs/2306.15377v2
|
https://arxiv.org/pdf/2306.15377v2.pdf
|
TrickVOS: A Bag of Tricks for Video Object Segmentation
|
Space-time memory (STM) network methods have been dominant in semi-supervised video object segmentation (SVOS) due to their remarkable performance. In this work, we identify three key aspects where we can improve such methods; i) supervisory signal, ii) pretraining and iii) spatial awareness. We then propose TrickVOS; a generic, method-agnostic bag of tricks addressing each aspect with i) a structure-aware hybrid loss, ii) a simple decoder pretraining regime and iii) a cheap tracker that imposes spatial constraints in model predictions. Finally, we propose a lightweight network and show that when trained with TrickVOS, it achieves competitive results to state-of-the-art methods on DAVIS and YouTube benchmarks, while being one of the first STM-based SVOS methods that can run in real-time on a mobile device.
|
['Albert Saa-Garriga', 'Bruno Manganelli', 'Anastasios Drosou', 'Armando Domi', 'Koskinas Ioannis', 'Mehmet Kerim Yucel', 'Konstantinos Georgiadis', 'Evangelos Skartados']
|
2023-06-27
| null | null | null | null |
['semi-supervised-video-object-segmentation', 'video-object-segmentation', 'video-semantic-segmentation']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 3.58568817e-01 1.06947534e-02 -7.09938526e-01 -1.91892162e-01
-8.07037771e-01 -5.71528196e-01 6.70002043e-01 -3.79568070e-01
-5.77510834e-01 4.89703089e-01 -5.24104163e-02 -5.33598483e-01
2.03255087e-01 -2.31933638e-01 -1.18265343e+00 -4.56711501e-01
8.91118348e-02 7.33130872e-01 9.44556892e-01 1.46485910e-01
6.52093366e-02 3.12280267e-01 -1.36289811e+00 2.63433963e-01
8.48274887e-01 1.34416413e+00 3.29018295e-01 6.82489216e-01
-3.19606438e-02 1.00855267e+00 -2.58339942e-01 -2.57427990e-01
3.13156694e-01 -3.60856324e-01 -9.84372258e-01 1.10121928e-01
8.95405829e-01 -3.41188341e-01 -3.03126305e-01 8.33492637e-01
3.39528322e-01 5.73709980e-02 5.49375415e-01 -1.03151858e+00
-2.13147953e-01 6.03476167e-01 -6.16778076e-01 4.08296704e-01
-1.50357764e-02 4.93718475e-01 7.17897773e-01 -8.04326653e-01
8.45072627e-01 1.08202660e+00 9.49581504e-01 7.49927223e-01
-1.34568417e+00 -3.52218777e-01 4.60119903e-01 2.42094398e-01
-1.11081254e+00 -6.21917784e-01 4.91919905e-01 -5.05391359e-01
1.06885707e+00 1.32001504e-01 7.38536775e-01 1.42801976e+00
9.76077765e-02 1.27091885e+00 1.06694484e+00 -2.17648521e-01
3.88000369e-01 1.26219392e-01 -1.77523643e-02 7.67841935e-01
-2.25691926e-02 -3.00500020e-02 -6.38389587e-01 2.68209070e-01
7.92152166e-01 -1.54733226e-01 -8.79036561e-02 -6.88388050e-01
-1.06857669e+00 5.74406147e-01 4.22048539e-01 8.48293379e-02
1.19202891e-02 5.95525563e-01 6.40779912e-01 7.74777085e-02
4.84351218e-01 1.42412782e-01 -6.01659536e-01 -3.34277004e-01
-1.41305709e+00 6.26881793e-02 5.87476552e-01 9.62049603e-01
3.43551517e-01 3.17836344e-01 -3.47764671e-01 5.23259342e-01
3.84622514e-01 3.51601124e-01 5.08662581e-01 -1.00071073e+00
7.01822639e-01 2.21021920e-01 -5.88809550e-02 -4.50761259e-01
-3.38640928e-01 -5.84017754e-01 -4.23735887e-01 1.77312285e-01
4.89184052e-01 -2.01376192e-02 -1.31628680e+00 1.76567626e+00
3.24037284e-01 7.60107100e-01 -2.33038977e-01 9.01337862e-01
8.40353906e-01 4.56975639e-01 -2.42011137e-02 1.45015838e-02
1.10178161e+00 -1.57458496e+00 -4.78436351e-01 -5.00184476e-01
6.49289668e-01 -3.88811707e-01 1.15425634e+00 3.96425754e-01
-1.27283299e+00 -6.07813776e-01 -1.19073486e+00 -1.18152902e-01
-3.28920752e-01 2.38299116e-01 7.20457494e-01 6.80582881e-01
-1.30203235e+00 1.02749014e+00 -1.34727085e+00 -3.93623054e-01
7.13950098e-01 5.66712618e-01 -5.02852201e-02 3.74878675e-01
-7.06664264e-01 7.37373114e-01 1.85607731e-01 -3.01138032e-03
-1.32381606e+00 -6.18252456e-01 -8.02612245e-01 -2.47312915e-02
8.82764041e-01 -8.14647675e-01 1.36996925e+00 -1.19255698e+00
-1.64370835e+00 9.74096954e-01 -2.69394279e-01 -1.05305862e+00
9.14637566e-01 -4.14569825e-01 -1.67201683e-01 3.48751396e-01
1.56629592e-01 9.77230012e-01 1.19952285e+00 -1.05879951e+00
-6.39710903e-01 -3.16438496e-01 -1.13148220e-01 6.28498495e-02
-3.23701710e-01 3.87094170e-02 -8.57166171e-01 -7.02618182e-01
-3.15705240e-02 -9.14305687e-01 -1.76246196e-01 2.36548573e-01
-7.38980055e-01 -1.46363452e-01 1.06236601e+00 -6.29765987e-01
1.10378265e+00 -1.95146096e+00 2.93361902e-01 -1.50705442e-01
3.75195354e-01 8.32212806e-01 -8.81124437e-02 1.05045922e-02
2.29948834e-01 7.79062957e-02 -1.11523300e-01 -9.17287171e-01
8.05111416e-03 1.00945525e-01 -3.28577310e-01 4.09127533e-01
9.73005593e-02 1.22733200e+00 -8.12199891e-01 -7.16893673e-01
2.98168778e-01 4.61044729e-01 -4.06367511e-01 -1.13592565e-01
-5.26352167e-01 5.35138309e-01 -2.13774055e-01 7.48722076e-01
5.00002861e-01 -4.37790781e-01 1.13736190e-01 -1.33447826e-01
-2.27771372e-01 4.52966422e-01 -1.04474008e+00 1.81946659e+00
-1.23126701e-01 8.51350486e-01 2.33001068e-01 -9.01203752e-01
3.55773926e-01 2.26749912e-01 2.89147288e-01 -6.27477944e-01
2.46940017e-01 2.39910930e-01 -4.62591529e-01 -2.52947599e-01
4.12489891e-01 3.83105427e-01 4.53027248e-01 1.95623502e-01
4.16621536e-01 2.81153262e-01 2.24057198e-01 2.81762302e-01
1.08967948e+00 6.68971896e-01 -8.63418169e-03 -3.28638554e-01
2.20859289e-01 -2.23462597e-01 6.83284521e-01 1.01023936e+00
-4.56957102e-01 7.65638649e-01 4.29567814e-01 -4.64166343e-01
-8.35680485e-01 -1.01612759e+00 -1.36234816e-02 1.16859877e+00
3.89987618e-01 -4.91127968e-01 -9.36447680e-01 -1.06464243e+00
-4.31304500e-02 3.87802720e-01 -6.86048150e-01 7.07998797e-02
-7.27942705e-01 -2.30907977e-01 6.01738513e-01 1.09508002e+00
6.70464754e-01 -8.04833889e-01 -7.18209445e-01 1.92052469e-01
1.14058308e-01 -1.40368354e+00 -7.82241285e-01 5.20236194e-01
-1.13787496e+00 -8.72475088e-01 -7.45501578e-01 -7.36338556e-01
2.78063357e-01 3.31098080e-01 1.05764127e+00 -1.47716552e-01
-1.11155480e-01 2.48235956e-01 -7.54192770e-02 -2.59293139e-01
-5.88314934e-03 3.61775845e-01 2.74993554e-02 -6.64648935e-02
8.70286673e-02 -5.99705279e-01 -6.93252206e-01 4.69528377e-01
-6.44730151e-01 2.66416162e-01 5.93334675e-01 7.10562170e-01
7.16644585e-01 -4.45989490e-01 1.76504463e-01 -1.02485204e+00
-9.23551545e-02 -2.06474200e-01 -8.43080878e-01 2.81216502e-01
-6.36902213e-01 -3.82369719e-02 5.44065237e-01 -4.69093889e-01
-9.59662735e-01 4.41230357e-01 -2.70762265e-01 -6.20300114e-01
-5.33731207e-02 8.13007057e-02 -6.86710253e-02 -4.57231671e-01
5.80063522e-01 3.67432415e-01 -5.77956364e-02 -6.71387732e-01
3.21054995e-01 3.09058249e-01 6.48778379e-01 -4.53431249e-01
7.29843676e-01 8.30509841e-01 -1.14156343e-01 -7.07117200e-01
-1.12040269e+00 -5.38199663e-01 -8.00492287e-01 -1.10741287e-01
1.09249079e+00 -9.67233419e-01 -5.75559556e-01 4.89855021e-01
-1.03077817e+00 -8.90464544e-01 -3.60895544e-01 2.18098015e-01
-8.35138083e-01 3.81483078e-01 -9.14505601e-01 -6.59053981e-01
-2.30446443e-01 -1.24739325e+00 1.30166149e+00 2.56957740e-01
-3.92170772e-02 -1.03731096e+00 -1.48741687e-02 5.08441210e-01
5.44456780e-01 2.69134194e-01 2.74205118e-01 -5.33918083e-01
-8.91099155e-01 1.68800369e-01 -3.61025453e-01 3.89702111e-01
-2.94358850e-01 -2.68279575e-02 -1.08342493e+00 -4.29405391e-01
-1.18605122e-01 -3.99528980e-01 1.38977289e+00 6.97891712e-01
1.17407238e+00 -2.85687417e-01 -7.01984763e-01 1.17788577e+00
1.25703776e+00 7.46029317e-02 4.61375475e-01 4.88365829e-01
1.03578603e+00 -1.01594031e-02 4.60930854e-01 4.27181385e-02
4.41011339e-01 1.01398373e+00 6.20204687e-01 -3.24955016e-01
-4.40442890e-01 -4.37471718e-01 4.81220752e-01 5.99085152e-01
-9.52515900e-02 -2.08415449e-01 -7.14169085e-01 6.74070418e-01
-2.22623563e+00 -7.52953112e-01 -7.68801719e-02 2.14033079e+00
5.20393074e-01 5.56265295e-01 5.13906300e-01 -5.90424947e-02
5.10876954e-01 5.82905650e-01 -8.28547060e-01 -2.63132811e-01
-1.78451166e-01 1.44464612e-01 7.00783968e-01 2.25021571e-01
-1.59024096e+00 1.33316052e+00 6.64807320e+00 9.17886674e-01
-1.35034871e+00 5.43014288e-01 6.35056257e-01 -2.38079697e-01
1.48277625e-01 -6.66187331e-02 -1.02040303e+00 6.30969048e-01
8.32870245e-01 5.31885207e-01 2.52795726e-01 1.15524876e+00
4.39394191e-02 -4.14504185e-02 -1.27632356e+00 9.38631713e-01
1.25420839e-01 -1.65739346e+00 -6.76721781e-02 6.98527023e-02
7.90733457e-01 5.12426436e-01 2.07036287e-01 1.99835479e-01
4.48840447e-02 -1.11238170e+00 1.23291206e+00 1.06266551e-01
7.85319626e-01 -3.64822477e-01 4.75369900e-01 2.59333253e-01
-1.40547884e+00 -7.82703981e-02 -2.25320324e-01 1.36316910e-01
4.66772377e-01 3.38272184e-01 -3.29901785e-01 2.76822716e-01
7.71710575e-01 9.03433919e-01 -5.82413137e-01 1.20388877e+00
-2.68356353e-01 1.03895819e+00 -4.23592955e-01 2.20425963e-01
5.52835763e-01 2.43865252e-01 7.09384322e-01 1.41503096e+00
1.04627525e-02 -4.57696825e-01 2.53702849e-01 7.88084030e-01
-1.62550136e-01 -2.94469297e-01 -4.97634381e-01 5.44761829e-02
3.24210346e-01 1.03196657e+00 -1.12132001e+00 -4.05503958e-01
-2.99290746e-01 1.14417231e+00 2.76094198e-01 3.64797741e-01
-1.07472444e+00 1.36562185e-02 4.96982783e-01 3.02206278e-01
9.38408017e-01 -2.37769634e-01 -4.71855730e-01 -1.23694587e+00
3.17448020e-01 -7.05797791e-01 2.25372270e-01 -5.59483349e-01
-7.90538490e-01 6.07149124e-01 -3.24338019e-01 -1.21143866e+00
-8.76148567e-02 -7.16269732e-01 -5.85532665e-01 2.63551950e-01
-1.66430056e+00 -1.41686785e+00 -6.40521795e-02 4.26517367e-01
7.31977463e-01 -9.54249129e-02 3.24013054e-01 4.54918474e-01
-7.80319154e-01 7.16715932e-01 1.81338698e-01 5.63027114e-02
4.91230071e-01 -1.44350350e+00 8.07699442e-01 9.53838289e-01
3.92336428e-01 4.61040080e-01 5.10576606e-01 -5.96486092e-01
-1.40241206e+00 -1.18296182e+00 6.05216086e-01 -7.31500387e-01
4.87233311e-01 -6.25288069e-01 -7.49177158e-01 1.15094578e+00
7.21145123e-02 3.03119391e-01 8.84900466e-02 7.51250936e-03
-3.30994785e-01 -1.06727511e-01 -8.47208381e-01 5.45715690e-01
1.43123913e+00 -4.77367759e-01 -3.19921255e-01 5.04395127e-01
8.34816098e-01 -8.96876276e-01 -3.44046891e-01 2.80020595e-01
6.88125193e-01 -1.22552836e+00 1.03359210e+00 -5.27391732e-01
2.40808174e-01 -4.07392353e-01 1.52925909e-01 -8.16328466e-01
-1.25202745e-01 -1.04774034e+00 -6.59987569e-01 1.06808090e+00
4.93932039e-01 -5.51399887e-01 1.23138571e+00 3.97871196e-01
-3.60060990e-01 -1.19946277e+00 -1.12091708e+00 -1.18573415e+00
-1.25479653e-01 -5.43956697e-01 2.83415884e-01 5.47637165e-01
-3.70134979e-01 3.58836144e-01 -6.90487862e-01 -2.51563042e-01
6.00575328e-01 -1.31276816e-01 7.50150204e-01 -1.03363025e+00
-7.05300272e-01 -5.11175871e-01 -4.78957742e-01 -1.83526647e+00
-8.25025886e-02 -5.94577491e-01 6.68963715e-02 -1.44708633e+00
2.85623789e-01 -2.41938278e-01 -8.25918242e-02 5.03377974e-01
1.45065352e-01 3.80410761e-01 3.64421189e-01 1.77778378e-01
-1.37490845e+00 4.05583918e-01 1.00325787e+00 -9.67969298e-02
-2.75745779e-01 1.47367492e-02 -4.83669072e-01 9.06543732e-01
4.28460538e-01 -4.85539764e-01 -4.05447304e-01 -6.68336511e-01
-6.05928898e-02 -1.77142411e-01 4.97164696e-01 -1.20272183e+00
4.56059396e-01 7.71602392e-02 2.82425106e-01 -6.67842388e-01
6.32659853e-01 -5.39825976e-01 -1.85313582e-01 4.83278960e-01
-2.26787940e-01 -2.15398371e-01 1.80778131e-01 5.55926800e-01
-9.20274779e-02 4.12347279e-02 8.86642694e-01 -5.91346482e-03
-9.35101986e-01 3.29794019e-01 -1.88108668e-01 2.22156852e-01
9.89524305e-01 -6.31186545e-01 -4.29900736e-01 -2.75000155e-01
-5.90686619e-01 3.24853957e-01 4.12740290e-01 3.37835699e-01
4.00469363e-01 -1.04483151e+00 -3.09880286e-01 -9.65952501e-02
-2.07593307e-01 4.43927459e-02 1.26584694e-01 1.22980011e+00
-4.59248334e-01 6.26516283e-01 7.81354383e-02 -1.03171062e+00
-1.35036886e+00 4.68948364e-01 3.05692017e-01 -4.50444192e-01
-8.22972357e-01 1.10854828e+00 3.04159015e-01 -2.84497231e-01
7.83900976e-01 -4.39721674e-01 4.10063900e-02 -1.63630560e-01
4.73042965e-01 3.61883610e-01 6.74147978e-02 -6.07054174e-01
-5.16227186e-01 7.50430703e-01 -1.78360760e-01 -7.47818351e-02
1.20243239e+00 -1.25542119e-01 3.70026231e-01 3.91009480e-01
9.91220415e-01 -3.87848884e-01 -1.99413097e+00 -2.92816788e-01
1.88914880e-01 -4.13860351e-01 2.07967982e-01 -7.19426274e-01
-1.18085539e+00 9.65360463e-01 7.07923770e-01 1.33012915e-02
9.74041939e-01 4.80970182e-02 1.16249537e+00 2.76476651e-01
3.11905831e-01 -1.34482396e+00 1.96233302e-01 5.78234792e-01
4.14556891e-01 -1.29814231e+00 -1.45695105e-01 -4.62832183e-01
-6.77701294e-01 8.17992389e-01 6.71147645e-01 -5.24230339e-02
5.78985870e-01 4.54768628e-01 3.35825980e-02 -2.21590716e-02
-7.28823543e-01 -4.17815298e-01 5.44678271e-01 6.05565310e-01
1.88493311e-01 -3.32527757e-01 3.53024527e-02 4.09157604e-01
2.61594564e-01 1.59772739e-01 -1.98417064e-02 9.80049431e-01
-4.13841039e-01 -8.20157826e-01 -1.46811241e-02 5.69095790e-01
-5.54224432e-01 -1.35331852e-02 -3.77363384e-01 7.46294439e-01
6.10615909e-02 6.17053509e-01 -1.43496946e-01 -3.15101534e-01
1.30840689e-01 -1.30778700e-01 4.78492260e-01 -4.78741139e-01
-6.45490408e-01 1.18421242e-01 9.74494666e-02 -1.03854060e+00
-3.75362366e-01 -6.79334402e-01 -1.10365582e+00 -1.50782853e-01
-4.32659209e-01 -2.03729615e-01 7.88761318e-01 1.29309452e+00
5.50435364e-01 5.26380599e-01 1.01723664e-01 -1.35346985e+00
-3.93367320e-01 -7.42371440e-01 -3.35702747e-01 -2.04330683e-02
4.05178607e-01 -6.76506758e-01 -1.71627685e-01 3.81525569e-02]
|
[9.063101768493652, -0.13000692427158356]
|
bd2400de-9465-4950-b980-cc50a36dd0cb
|
exploiting-semantic-epsilon-greedy
|
2201.10803
| null |
https://arxiv.org/abs/2201.10803v2
|
https://arxiv.org/pdf/2201.10803v2.pdf
|
Exploiting Semantic Epsilon Greedy Exploration Strategy in Multi-Agent Reinforcement Learning
|
Multi-agent reinforcement learning (MARL) can model many real world applications. However, many MARL approaches rely on epsilon greedy for exploration, which may discourage visiting advantageous states in hard scenarios. In this paper, we propose a new approach QMIX(SEG) for tackling MARL. It makes use of the value function factorization method QMIX to train per-agent policies and a novel Semantic Epsilon Greedy (SEG) exploration strategy. SEG is a simple extension to the conventional epsilon greedy exploration strategy, yet it is experimentally shown to greatly improve the performance of MARL. We first cluster actions into groups of actions with similar effects and then use the groups in a bi-level epsilon greedy exploration hierarchy for action selection. We argue that SEG facilitates semantic exploration by exploring in the space of groups of actions, which have richer semantic meanings than atomic actions. Experiments show that QMIX(SEG) largely outperforms QMIX and leads to strong performance competitive with current state-of-the-art MARL approaches on the StarCraft Multi-Agent Challenge (SMAC) benchmark.
|
['Ho-fung Leung', 'Hon Tik Tse']
|
2022-01-26
| null | null | null | null |
['smac-1', 'smac']
|
['playing-games', 'playing-games']
|
[-2.70937741e-01 8.29959586e-02 -4.72973913e-01 4.25044522e-02
-8.19518507e-01 -4.55999970e-01 7.40006804e-01 2.31284693e-01
-7.28936374e-01 1.08697009e+00 4.15853947e-01 -3.73445630e-01
-5.56369364e-01 -7.65453100e-01 -8.48240852e-01 -8.63666832e-01
-5.37011445e-01 8.94492328e-01 4.10733104e-01 -6.97514892e-01
3.87254715e-01 9.15936381e-02 -1.37594700e+00 4.07030702e-01
8.61976445e-01 6.77895963e-01 4.11983579e-01 5.97174525e-01
-9.64327678e-02 1.03031266e+00 -4.76790160e-01 1.44596428e-01
4.12755728e-01 -4.95964438e-01 -1.20822537e+00 -1.66243047e-01
-2.89700985e-01 -2.14006275e-01 4.94844653e-02 9.58637297e-01
3.79927397e-01 8.39688063e-01 1.45047233e-01 -1.44396985e+00
-4.19089079e-01 1.26663125e+00 -6.12388253e-01 8.24604630e-02
3.53616953e-01 3.29923838e-01 1.30675483e+00 -4.33917701e-01
6.97844505e-01 1.76382673e+00 4.16200697e-01 6.29086435e-01
-1.03245354e+00 -1.85466066e-01 7.54957497e-01 3.76448750e-01
-6.30126774e-01 1.05558947e-01 6.07621133e-01 1.45968452e-01
1.27272141e+00 1.59420744e-01 8.92095566e-01 9.88250911e-01
2.60143369e-01 1.45344508e+00 1.34662354e+00 -3.76644492e-01
9.50195193e-01 -4.66582537e-01 -1.90493405e-01 6.99576557e-01
-2.53124069e-02 4.84682143e-01 -8.40060234e-01 -5.46445370e-01
6.22853041e-01 -4.43706438e-02 1.92315280e-01 -8.85980487e-01
-1.39387393e+00 1.22840893e+00 5.72170675e-01 2.62815524e-02
-5.91275632e-01 6.13796949e-01 4.84234124e-01 4.28745538e-01
-6.22753128e-02 1.19820750e+00 -6.53938532e-01 -6.21681154e-01
-3.34366560e-01 6.17387354e-01 6.56929970e-01 5.99224806e-01
8.33781898e-01 2.28012055e-02 1.85363814e-02 6.20521963e-01
2.40874097e-01 1.79764166e-01 5.42401612e-01 -1.44599354e+00
1.93359986e-01 5.08874774e-01 4.16161299e-01 -2.82776594e-01
-6.40578389e-01 -2.43733048e-01 -1.90980732e-01 6.60555005e-01
5.32632880e-02 -3.13658148e-01 -7.84508824e-01 1.74151611e+00
4.97496963e-01 1.60413440e-02 4.98416215e-01 9.28278685e-01
2.33986482e-01 6.48995221e-01 2.80603439e-01 -9.43746418e-02
1.06203842e+00 -1.51720321e+00 -3.33343267e-01 -4.22394782e-01
8.73398244e-01 -1.36032984e-01 1.19285047e+00 6.78955555e-01
-1.10159934e+00 -2.02353939e-01 -8.28458548e-01 3.81965816e-01
-3.02710652e-01 -3.95508975e-01 1.23939013e+00 2.65511513e-01
-1.14101052e+00 8.61675739e-01 -1.26093829e+00 -1.44551069e-01
3.99433404e-01 2.37982184e-01 -4.24273424e-02 1.14087634e-01
-1.25394332e+00 9.80966389e-01 8.86701941e-01 -2.23394126e-01
-1.31915665e+00 -3.86756182e-01 -9.68221962e-01 -1.38294071e-01
1.10943651e+00 -4.01866198e-01 1.51602733e+00 -8.41803372e-01
-1.86351991e+00 5.92945255e-02 2.42905661e-01 -9.79856431e-01
4.85805333e-01 -5.89447439e-01 -7.28062727e-03 2.81411856e-01
2.21252054e-01 1.07344139e+00 4.91550744e-01 -1.25932086e+00
-9.76489305e-01 -3.12828198e-02 6.08319581e-01 5.87298334e-01
3.67345475e-02 -1.88888624e-01 -1.56995431e-02 -4.18092817e-01
-1.65693447e-01 -1.16340673e+00 -9.78540838e-01 -5.53500473e-01
-6.44864812e-02 -6.31198525e-01 4.35196072e-01 -5.71097471e-02
1.00409257e+00 -1.86712480e+00 5.63701689e-01 1.50298014e-01
5.16763479e-02 5.92761822e-02 -5.48088372e-01 8.01880002e-01
3.82909417e-01 -1.23266481e-01 1.60193667e-02 -3.38117063e-01
4.41313058e-01 7.84036100e-01 -3.35599810e-01 2.61922896e-01
-1.20223746e-01 1.14232910e+00 -1.55228508e+00 -2.65540093e-01
3.52620423e-01 -6.24788962e-02 -9.07157481e-01 2.64667552e-02
-9.09533381e-01 3.81248325e-01 -8.52608502e-01 6.92167997e-01
3.16356868e-01 -2.58221924e-01 4.13951546e-01 4.72431064e-01
-1.35122061e-01 4.13061380e-01 -1.18067467e+00 1.98303354e+00
-3.60653073e-01 8.50509927e-02 -1.56350300e-01 -8.49092364e-01
6.55548751e-01 -1.04316272e-01 7.28553116e-01 -9.34271038e-01
-5.79276755e-02 1.99756980e-01 -2.93808710e-02 1.15022352e-02
8.09813321e-01 -1.00005306e-01 -3.59042823e-01 5.47774315e-01
-6.01352789e-02 7.32054189e-02 5.43529332e-01 3.11780334e-01
1.19137907e+00 6.30337834e-01 4.56008404e-01 -5.35034060e-01
2.70223767e-01 1.87325805e-01 8.47335398e-01 1.32065105e+00
-3.14879984e-01 -2.19786033e-01 7.87997484e-01 -8.28462541e-01
-6.35027587e-01 -1.07657838e+00 4.70055848e-01 1.67039597e+00
6.08486116e-01 -7.49363899e-01 -5.20402312e-01 -1.06696188e+00
7.52463788e-02 9.47505713e-01 -7.93675005e-01 -1.14378273e-01
-7.65921175e-01 -8.82431805e-01 2.65603244e-01 7.27229357e-01
5.21306217e-01 -1.78593266e+00 -1.41545761e+00 4.60572511e-01
1.46263495e-01 -7.19943166e-01 -9.89820212e-02 6.13793194e-01
-7.55943894e-01 -1.06768668e+00 -3.25155020e-01 -4.85632628e-01
3.43397141e-01 6.28460646e-02 1.31651390e+00 -2.72153895e-02
3.98690552e-02 5.06420791e-01 -9.69382465e-01 -2.38271281e-01
-3.83215040e-01 1.39687195e-01 8.21958110e-02 -5.85039973e-01
2.66467988e-01 -3.70132625e-01 -7.09054649e-01 1.94703162e-01
-6.45581365e-01 2.15520874e-01 6.95414722e-01 1.06604230e+00
5.85584581e-01 2.07978398e-01 7.00110257e-01 -7.29167998e-01
6.82221234e-01 -4.41763788e-01 -6.85896039e-01 1.35857761e-01
-6.84295654e-01 5.88434398e-01 6.77290082e-01 -4.60190713e-01
-8.74865472e-01 -4.97386754e-02 -4.65489179e-02 -2.88199723e-01
4.05858010e-02 5.83479226e-01 2.03472972e-01 7.16097504e-02
4.54030544e-01 3.60127777e-01 5.28923981e-02 -5.00788093e-01
5.59044957e-01 -5.30937016e-02 2.22857490e-01 -1.06365633e+00
2.53387302e-01 4.58101094e-01 2.30932701e-02 -4.60456133e-01
-7.44696438e-01 -1.95386901e-01 5.12277670e-02 -1.09616388e-03
7.74979711e-01 -7.32294500e-01 -1.12408471e+00 2.14225858e-01
-5.03518283e-01 -1.01721311e+00 -5.70206165e-01 5.07212460e-01
-1.12158394e+00 3.12456131e-01 -6.35238230e-01 -8.93420935e-01
-9.80356336e-02 -1.36626005e+00 9.07696426e-01 3.60750675e-01
-6.29406571e-02 -9.04824138e-01 3.41179371e-01 1.04527995e-01
1.78000197e-01 1.54970214e-01 8.12235594e-01 -7.23635733e-01
-6.82976007e-01 5.68770885e-01 3.29346240e-01 -9.96866897e-02
-2.33448520e-01 -6.10001624e-01 -1.84346616e-01 -7.06142783e-01
-4.23805416e-01 -7.76297152e-01 1.02096617e+00 4.94039059e-01
9.30536509e-01 -3.13488156e-01 -2.94216573e-01 4.04785663e-01
1.42939782e+00 5.16858160e-01 5.63439131e-01 1.16640627e+00
2.76478738e-01 1.56807899e-01 1.21769047e+00 8.49558353e-01
5.64836681e-01 5.62185407e-01 1.02916253e+00 2.01032564e-01
3.12945545e-01 -3.93887877e-01 7.00162351e-01 1.27402052e-01
-9.74107906e-02 -2.46226326e-01 -7.62632132e-01 6.45333290e-01
-2.50220704e+00 -8.26330006e-01 5.54916084e-01 1.81055665e+00
7.61106312e-01 3.52179438e-01 5.04663348e-01 -4.26082641e-01
1.98853269e-01 3.49886209e-01 -8.37654769e-01 -7.42897153e-01
-1.29427820e-01 2.72191763e-01 5.48271298e-01 7.18769252e-01
-1.18835068e+00 1.32797611e+00 6.24575758e+00 9.27720487e-01
-5.31754315e-01 1.03835486e-01 2.28109941e-01 -2.44323641e-01
-3.32272887e-01 2.61106670e-01 -8.52149367e-01 3.89035970e-01
5.18091679e-01 1.61362901e-01 7.70821273e-01 1.18331945e+00
-8.99573937e-02 -4.39796656e-01 -9.80930328e-01 6.90144598e-01
-3.36438239e-01 -1.42187774e+00 6.27658218e-02 1.21512273e-02
9.80955184e-01 2.27325305e-01 -6.19929545e-02 7.63389826e-01
1.23343599e+00 -9.58454669e-01 7.62154937e-01 2.62022223e-02
-8.87142718e-02 -1.22348690e+00 5.62367439e-01 4.42101032e-01
-1.08388495e+00 -6.43567562e-01 -3.87757778e-01 -2.09614769e-01
2.43994519e-01 -1.09838940e-01 -6.59412444e-01 5.48960626e-01
7.69687593e-01 6.80592179e-01 -3.23172063e-01 8.47778320e-01
-4.34781730e-01 2.11342022e-01 -1.42104015e-01 -4.34796423e-01
1.17543757e+00 -2.61421263e-01 7.24235654e-01 7.82439351e-01
1.12694427e-01 -7.59502128e-02 9.27833378e-01 5.37846804e-01
2.90478826e-01 -2.60851711e-01 -2.00591952e-01 -2.18825936e-01
5.02087116e-01 9.11791861e-01 -7.39012182e-01 -3.10937136e-01
-2.69207269e-01 8.22937369e-01 5.08176386e-01 1.77075163e-01
-8.81873131e-01 -4.06008661e-02 8.71331811e-01 -3.64772677e-01
6.26024187e-01 -2.88213849e-01 1.58958018e-01 -8.99390876e-01
-2.80708045e-01 -1.23345292e+00 7.23324776e-01 -4.80982333e-01
-9.98948693e-01 6.14471972e-01 5.37898466e-02 -1.04182374e+00
-5.49087405e-01 -5.48099995e-01 -5.47436297e-01 3.03621382e-01
-1.63779390e+00 -9.95316148e-01 1.74419492e-01 4.16061401e-01
9.76202309e-01 -3.82938445e-01 8.12934577e-01 -5.02219975e-01
-2.53072888e-01 1.83000997e-01 2.11189374e-01 -3.18686992e-01
2.66073465e-01 -1.70398521e+00 4.93587434e-01 6.34267986e-01
6.30506724e-02 3.58527571e-01 8.24658871e-01 -7.88789451e-01
-1.50390196e+00 -7.99406588e-01 1.99484497e-01 -2.08569571e-01
8.46547902e-01 -3.45430300e-02 -5.45654356e-01 7.94881105e-01
5.02673268e-01 -3.16675484e-01 3.70191932e-01 4.45200592e-01
-9.36859194e-03 3.42478156e-01 -7.70150363e-01 8.92089248e-01
1.04604101e+00 4.78448682e-02 -9.17160213e-01 3.01755846e-01
7.04685807e-01 -4.50332135e-01 -7.10249841e-01 2.98787713e-01
3.69909912e-01 -1.13562429e+00 1.05417454e+00 -9.64275301e-01
3.62642556e-01 -4.03856725e-01 -1.65340275e-01 -1.80900192e+00
-4.39835548e-01 -1.13494217e+00 -5.00019848e-01 4.44553673e-01
2.51256406e-01 -6.46421552e-01 9.43965673e-01 7.02348202e-02
-2.64531821e-01 -1.22764361e+00 -8.92951012e-01 -1.14846873e+00
2.62286305e-01 -1.07977025e-01 7.65282094e-01 5.41619301e-01
3.39892298e-01 -1.76195595e-02 -5.73570490e-01 -1.13326516e-02
7.29568183e-01 4.54681307e-01 6.58328056e-01 -8.72188449e-01
-7.77974606e-01 -6.03783429e-01 2.63234209e-02 -1.16090143e+00
1.75413415e-01 -6.96111441e-01 1.94289148e-01 -1.65125191e+00
7.16279447e-02 -6.94227040e-01 -7.03481495e-01 6.73557699e-01
-1.50729984e-01 -2.67277420e-01 5.24256647e-01 4.12900280e-03
-1.49996781e+00 1.00733662e+00 1.24479914e+00 1.01938151e-01
-5.84052980e-01 -1.74360618e-01 -8.15323472e-01 9.19957161e-01
9.50736880e-01 -3.06313962e-01 -6.59889042e-01 -2.92563200e-01
7.23837972e-01 7.65743926e-02 1.99673161e-01 -9.24965024e-01
1.40992936e-03 -7.17564821e-01 1.09664671e-01 -4.96803790e-01
4.36769873e-01 -4.69642580e-01 -2.24474430e-01 8.86254609e-01
-5.68173051e-01 3.51072460e-01 1.24837607e-01 5.96982956e-01
-4.52250093e-02 -1.65174395e-01 7.22776949e-01 -5.09908438e-01
-1.36913514e+00 5.30443788e-02 -5.06632805e-01 1.94340169e-01
1.18092501e+00 -4.49085832e-02 -3.46966803e-01 -2.17788249e-01
-9.05325532e-01 9.57994103e-01 4.32626128e-01 4.45492059e-01
6.43882632e-01 -1.31516671e+00 -4.78249162e-01 1.04330882e-01
-5.18380739e-02 -8.41763020e-02 2.05676988e-01 7.58046210e-01
-2.65917629e-01 3.02787066e-01 -6.04070365e-01 -2.81290412e-01
-9.41906869e-01 7.36283243e-01 2.00050518e-01 -8.26202214e-01
-8.85347962e-01 8.87557089e-01 1.84665203e-01 -5.06065190e-01
2.30853990e-01 -2.30793759e-01 -1.48552522e-01 -1.68535873e-01
4.94605094e-01 5.78545570e-01 -5.22654355e-01 -1.38953188e-02
-4.73983228e-01 2.27394685e-01 -3.53129894e-01 -3.06242973e-01
1.62214947e+00 2.28099916e-02 9.28396210e-02 1.68105736e-01
4.66050386e-01 -2.86213577e-01 -1.62214184e+00 -1.26710817e-01
1.49822116e-01 -3.69338542e-01 -9.54504907e-02 -1.05766618e+00
-6.61062419e-01 3.44416499e-01 1.83789611e-01 1.45580783e-01
9.96333122e-01 2.24392071e-01 7.58535922e-01 8.51889193e-01
8.57154191e-01 -1.71141171e+00 4.10407841e-01 7.60599434e-01
8.48046064e-01 -1.12809134e+00 -6.73783347e-02 1.28031120e-01
-1.23240590e+00 9.04272914e-01 7.97591686e-01 -4.60316360e-01
2.27487579e-01 7.88472220e-02 -2.43474543e-01 -2.53602147e-01
-1.16069257e+00 -4.69746470e-01 -2.95276165e-01 5.03258109e-01
-1.83907270e-01 2.85973787e-01 -3.83170366e-01 3.68527979e-01
-1.79148659e-01 -1.53553471e-01 4.92626816e-01 1.44515765e+00
-8.51771653e-01 -1.37982178e+00 -1.26528174e-01 2.93287456e-01
-1.31790325e-01 1.27428817e-02 -1.75410002e-01 8.99026692e-01
1.72528084e-02 7.50658572e-01 2.26709060e-02 -1.75485134e-01
-2.39123963e-02 -1.66167811e-01 5.50294995e-01 -3.14544588e-01
-6.68269336e-01 1.42194003e-01 1.67713761e-01 -1.11158466e+00
-2.81308770e-01 -7.47865200e-01 -1.71620166e+00 -1.17556728e-01
1.66564196e-01 5.54698646e-01 2.66469955e-01 1.02410114e+00
4.65741903e-01 6.59994006e-01 4.81809258e-01 -7.04333603e-01
-1.02522588e+00 -5.96885562e-01 -5.55440187e-01 3.00228745e-01
3.90553176e-01 -1.18438601e+00 -2.55369488e-02 -6.58982456e-01]
|
[3.8789241313934326, 1.7781462669372559]
|
4843730c-1e28-4191-8665-553353f40f35
|
towards-cross-modality-medical-image
|
2010.01532
| null |
https://arxiv.org/abs/2010.01532v1
|
https://arxiv.org/pdf/2010.01532v1.pdf
|
Towards Cross-modality Medical Image Segmentation with Online Mutual Knowledge Distillation
|
The success of deep convolutional neural networks is partially attributed to the massive amount of annotated training data. However, in practice, medical data annotations are usually expensive and time-consuming to be obtained. Considering multi-modality data with the same anatomic structures are widely available in clinic routine, in this paper, we aim to exploit the prior knowledge (e.g., shape priors) learned from one modality (aka., assistant modality) to improve the segmentation performance on another modality (aka., target modality) to make up annotation scarcity. To alleviate the learning difficulties caused by modality-specific appearance discrepancy, we first present an Image Alignment Module (IAM) to narrow the appearance gap between assistant and target modality data.We then propose a novel Mutual Knowledge Distillation (MKD) scheme to thoroughly exploit the modality-shared knowledge to facilitate the target-modality segmentation. To be specific, we formulate our framework as an integration of two individual segmentors. Each segmentor not only explicitly extracts one modality knowledge from corresponding annotations, but also implicitly explores another modality knowledge from its counterpart in mutual-guided manner. The ensemble of two segmentors would further integrate the knowledge from both modalities and generate reliable segmentation results on target modality. Experimental results on the public multi-class cardiac segmentation data, i.e., MMWHS 2017, show that our method achieves large improvements on CT segmentation by utilizing additional MRI data and outperforms other state-of-the-art multi-modality learning methods.
|
['Pheng-Ann Heng', 'Shujun Wang', 'Lequan Yu', 'Kang Li']
|
2020-10-04
| null | null | null | null |
['cardiac-segmentation']
|
['medical']
|
[ 5.24044394e-01 4.26928550e-01 -4.15511072e-01 -3.89198929e-01
-1.19963765e+00 -5.05582571e-01 2.33005822e-01 1.16881415e-01
-4.06723440e-01 8.46640527e-01 1.31448656e-01 -2.98556089e-01
-1.08967930e-01 -5.64339578e-01 -6.92458570e-01 -9.62125778e-01
3.75521302e-01 5.65638006e-01 3.13048571e-01 7.37991109e-02
-2.19481051e-01 1.60587221e-01 -8.73811364e-01 2.84203202e-01
1.07638860e+00 9.72870767e-01 5.02892673e-01 2.75770128e-01
-1.81430846e-01 6.33301616e-01 -9.81141329e-02 -2.61674285e-01
1.55423850e-01 -5.08029938e-01 -1.20004725e+00 3.09089184e-01
2.16752589e-01 -2.37208545e-01 -2.85464704e-01 1.03748274e+00
5.76360881e-01 -5.46149770e-03 6.66558981e-01 -9.87878740e-01
-3.92032057e-01 7.65239120e-01 -8.98295879e-01 2.49959350e-01
-1.68706581e-01 2.55916864e-01 7.67442465e-01 -6.27612054e-01
7.04092264e-01 5.04429936e-01 7.10189581e-01 4.85753298e-01
-1.19751787e+00 -5.99743009e-01 2.18043178e-01 3.02030090e-02
-1.30303168e+00 -1.32674620e-01 1.01640272e+00 -3.98155987e-01
3.37080985e-01 1.99862629e-01 5.81031561e-01 9.35795605e-01
-1.64316781e-02 1.14180398e+00 1.09581447e+00 -2.41177380e-01
-1.44254640e-01 1.99455917e-02 6.09176122e-02 8.74646068e-01
-6.12273552e-02 -1.03552967e-01 -1.82703793e-01 -9.93840769e-02
9.06446576e-01 4.85455506e-02 -6.12613142e-01 -5.99605083e-01
-1.59979033e+00 5.54965913e-01 6.48686767e-01 5.32101095e-01
-5.53327918e-01 -2.43401125e-01 4.61667120e-01 -6.11000396e-02
2.60364860e-01 2.70895362e-01 -5.52671790e-01 1.23601168e-01
-9.47569668e-01 -3.18615735e-01 5.96340239e-01 8.46403182e-01
7.09427178e-01 -3.46254617e-01 -2.79015034e-01 9.73613560e-01
3.81349117e-01 4.14403498e-01 6.21905029e-01 -6.40429437e-01
4.93718594e-01 6.40621185e-01 -3.41763586e-01 -6.46655619e-01
-6.72632456e-01 -4.41142380e-01 -1.20155740e+00 -2.63503253e-01
6.82569444e-01 -2.04120159e-01 -1.28191686e+00 1.75269628e+00
5.76819539e-01 4.96797770e-01 -5.87819070e-02 1.02885091e+00
1.17975760e+00 1.51577160e-01 2.72264123e-01 -1.99669600e-01
1.48599410e+00 -1.24405980e+00 -5.14809012e-01 -1.32839844e-01
9.43325102e-01 -7.32102394e-01 7.61968017e-01 1.17746383e-01
-9.21757162e-01 -3.82925332e-01 -8.30778360e-01 2.36267805e-01
6.03008736e-03 2.94956923e-01 8.15079749e-01 3.93635392e-01
-7.30291843e-01 3.17717582e-01 -1.09976125e+00 -1.50870055e-01
8.46746325e-01 4.28715914e-01 -6.75575256e-01 -1.28058106e-01
-1.12152207e+00 8.47279310e-01 6.25603616e-01 2.85270035e-01
-8.06306481e-01 -1.17999136e+00 -8.44305992e-01 -3.45568120e-01
7.28761911e-01 -8.24236393e-01 1.23045671e+00 -1.09107041e+00
-1.26751137e+00 1.07590342e+00 1.22852057e-01 -9.22219530e-02
5.22369206e-01 1.34323716e-01 -3.21351886e-01 3.93198013e-01
2.11666614e-01 7.62624025e-01 6.27605379e-01 -1.46951079e+00
-5.30643523e-01 -3.32251102e-01 6.54409006e-02 3.42093229e-01
-1.41847983e-01 -4.11278367e-01 -8.18101287e-01 -7.02947915e-01
3.79979879e-01 -1.11386895e+00 -4.41141486e-01 1.04603283e-01
-7.46857703e-01 3.96889821e-02 5.94839692e-01 -8.45161617e-01
9.90230799e-01 -1.99659741e+00 3.63444984e-01 3.93071413e-01
4.40412641e-01 1.85555220e-01 -2.59028748e-02 -3.24200839e-01
-2.36661181e-01 -3.28329168e-02 -7.52579033e-01 -3.91143143e-01
-3.86390775e-01 5.56889355e-01 2.02545121e-01 4.14312363e-01
9.89173800e-02 1.01812172e+00 -9.46460009e-01 -1.05747318e+00
2.40203053e-01 3.55629444e-01 -5.42796552e-01 1.90340504e-01
7.46823922e-02 1.32737184e+00 -6.93672836e-01 9.43932116e-01
6.39762998e-01 -7.69266248e-01 3.82605970e-01 -7.88218379e-01
2.57110029e-01 -1.51345700e-01 -8.89082372e-01 2.42702699e+00
-5.60090125e-01 1.95618887e-02 7.09291622e-02 -1.25878298e+00
4.77279097e-01 6.00165665e-01 1.00002253e+00 -6.04384184e-01
2.68043578e-01 4.17806953e-01 1.95240185e-01 -6.09284580e-01
6.54638037e-02 -3.71041030e-01 2.84289941e-02 3.31070751e-01
3.14485669e-01 -5.18203415e-02 -7.91298747e-02 1.12715125e-01
8.40313375e-01 1.34825349e-01 2.73449630e-01 -1.36126027e-01
6.68515623e-01 -2.93500274e-02 8.11002493e-01 6.68156505e-01
-5.26406348e-01 8.64280820e-01 2.93508083e-01 -2.61482477e-01
-8.65220308e-01 -1.09569740e+00 -3.25202882e-01 7.23477662e-01
3.86831492e-01 6.15906715e-02 -5.07995009e-01 -1.23966074e+00
-2.02767491e-01 2.19256282e-01 -8.70059788e-01 -1.66358307e-01
-6.55028880e-01 -9.28267896e-01 6.70851469e-01 8.64120245e-01
5.29082716e-01 -9.61583853e-01 -4.39578176e-01 2.18207091e-01
-5.68788171e-01 -1.25065482e+00 -5.73875010e-01 1.78886414e-01
-1.03858817e+00 -1.18379831e+00 -1.19445980e+00 -7.60265291e-01
9.57308233e-01 -9.98791009e-02 1.25706148e+00 2.15157136e-01
-2.78817534e-01 4.64476943e-01 -3.90277028e-01 -1.07819766e-01
-2.23584354e-01 3.23445320e-01 -3.76867503e-01 6.60392046e-02
3.71326432e-02 -5.87351859e-01 -6.95475876e-01 3.54739130e-01
-9.42669928e-01 5.52036643e-01 9.19949234e-01 1.16642630e+00
9.41082418e-01 -1.41717225e-01 5.19134700e-01 -1.11447966e+00
1.68133173e-02 -5.89424789e-01 -1.76566616e-01 5.17751515e-01
-3.57664675e-01 -4.71830182e-02 1.91406965e-01 -4.99438226e-01
-1.30176389e+00 3.92441183e-01 -1.92899808e-01 -4.62694049e-01
-3.23946059e-01 9.01871920e-01 -9.72948298e-02 -1.41715616e-01
2.50005692e-01 3.43303740e-01 1.09661140e-01 -4.24866825e-01
3.40743512e-01 2.49818549e-01 7.04765260e-01 -8.86343896e-01
5.01088262e-01 4.70852256e-01 9.05587152e-02 -3.53704929e-01
-1.06758296e+00 -5.07322669e-01 -8.87264192e-01 -1.68264806e-01
1.20103562e+00 -7.53658891e-01 -3.61017108e-01 5.43869555e-01
-9.91202652e-01 -3.41388285e-01 -2.98503727e-01 6.36161327e-01
-4.14591968e-01 5.69036841e-01 -5.41152775e-01 -1.20858021e-01
-3.86413574e-01 -1.68988705e+00 1.19050586e+00 4.78358895e-01
1.28802225e-01 -1.27291989e+00 -7.51778763e-03 6.31609559e-01
2.09579602e-01 2.58159012e-01 8.24634016e-01 -8.74018073e-01
-5.11566997e-01 5.19792289e-02 -5.14512181e-01 2.74898082e-01
4.09070760e-01 -4.70204085e-01 -7.29768157e-01 -1.65161744e-01
-2.02895105e-01 -3.75597984e-01 8.09011340e-01 5.76809645e-01
1.41497803e+00 2.19621882e-01 -5.39765775e-01 6.53859854e-01
1.10523582e+00 -8.90630931e-02 3.81349266e-01 1.78620040e-01
1.16418576e+00 4.05470967e-01 5.47948718e-01 2.67456561e-01
6.31984651e-01 5.89865386e-01 4.63008434e-01 -6.40845120e-01
-2.61099041e-01 3.56352478e-02 -3.97064179e-01 1.00631928e+00
-2.87531704e-01 1.34354591e-01 -1.12991548e+00 6.69283450e-01
-1.86619437e+00 -2.92091399e-01 -5.35057299e-02 1.95106232e+00
1.27318943e+00 -1.52832314e-01 -6.41055703e-02 -2.04209998e-01
6.30648971e-01 -3.16975378e-02 -6.49402916e-01 3.91571641e-01
4.83126417e-02 6.54568225e-02 6.03944063e-01 1.00454122e-01
-1.40505850e+00 6.35196447e-01 5.29907560e+00 1.01338124e+00
-1.11689889e+00 4.26033914e-01 8.28174293e-01 1.30702019e-01
-2.74906307e-01 -5.85579462e-02 -3.96778584e-01 4.55907881e-01
3.61141205e-01 1.86796531e-01 3.24105248e-02 5.30371487e-01
-1.94184497e-01 -1.64474383e-01 -1.10317063e+00 8.18728209e-01
-5.81713691e-02 -1.33206940e+00 -1.17650837e-01 3.97088379e-02
9.66749430e-01 -6.08563982e-03 -5.19683063e-02 3.21716636e-01
-2.34111436e-02 -8.49697113e-01 1.89858243e-01 7.13022709e-01
7.70024002e-01 -5.73705614e-01 1.08649170e+00 2.27662757e-01
-1.26659238e+00 3.69557738e-01 3.62881608e-02 6.76621318e-01
4.33625579e-01 7.13390946e-01 -7.13343918e-01 1.16753972e+00
6.08552158e-01 8.43202174e-01 -4.21581835e-01 1.14144599e+00
-1.61567539e-01 6.09591365e-01 -3.37760717e-01 7.24484324e-01
3.19461793e-01 -1.44785210e-01 4.07864630e-01 9.62055385e-01
1.34349093e-01 2.54547417e-01 6.25319779e-01 8.28133106e-01
-1.57884002e-01 1.45342961e-01 -1.61415309e-01 7.16663972e-02
1.90130517e-01 1.58316112e+00 -8.42399955e-01 -5.54043233e-01
-6.55858159e-01 9.34260905e-01 1.20224021e-01 2.34937698e-01
-9.74406600e-01 1.50953427e-01 5.65821044e-02 -2.51463324e-01
1.94615960e-01 1.76212385e-01 -4.48383629e-01 -1.22312748e+00
-8.79024267e-02 -7.65843511e-01 6.85320079e-01 -5.32285929e-01
-1.61494052e+00 6.03614390e-01 3.72037515e-02 -1.37700868e+00
1.01168931e-01 -3.47310662e-01 -4.37905192e-01 1.04447496e+00
-1.73125386e+00 -1.67996514e+00 -4.11986202e-01 7.60412157e-01
3.48988831e-01 3.68653238e-02 7.16364026e-01 7.34116435e-01
-5.24518669e-01 7.98979878e-01 -2.52443850e-01 3.45181197e-01
8.79426777e-01 -1.24045932e+00 -4.02524948e-01 4.50010896e-01
-1.45416185e-01 5.13781548e-01 1.92485213e-01 -6.79522514e-01
-1.06806719e+00 -8.84250283e-01 3.27409863e-01 -3.11994672e-01
8.09712946e-01 3.87804508e-01 -1.08509433e+00 7.16930628e-01
7.22335279e-02 4.98978317e-01 1.01426232e+00 8.50062445e-02
-1.15043893e-01 1.52798131e-01 -1.14951563e+00 4.41154808e-01
8.65232944e-01 -4.81423855e-01 -5.70544064e-01 3.97318423e-01
6.66560948e-01 -9.69203830e-01 -1.37419033e+00 9.29290891e-01
4.44577128e-01 -5.60990989e-01 1.00797153e+00 -5.51062167e-01
4.99674141e-01 -3.52891505e-01 -6.63089082e-02 -1.09870446e+00
1.03224374e-01 -1.40237957e-01 -8.96507055e-02 1.18606353e+00
5.70180237e-01 -5.63021541e-01 7.50372410e-01 8.48277092e-01
-5.11263311e-01 -1.07026958e+00 -1.06703424e+00 -4.57617283e-01
2.61232644e-01 -3.55697691e-01 4.25933182e-01 1.35449481e+00
-1.26697853e-01 5.22933416e-02 -4.70168114e-01 2.37825155e-01
5.26505470e-01 3.36682290e-01 3.98402572e-01 -1.02660549e+00
-5.86579919e-01 -4.00934577e-01 -1.08796373e-01 -9.89679575e-01
1.30330533e-01 -1.18608427e+00 1.32910728e-01 -1.53056633e+00
7.31887162e-01 -8.23391199e-01 -7.65438795e-01 9.07079875e-01
-5.62074602e-01 5.70410311e-01 1.59219429e-01 3.46894920e-01
-6.82931006e-01 4.13478822e-01 1.89899731e+00 -2.01012999e-01
-9.30330902e-02 9.53054875e-02 -7.54524589e-01 9.10823464e-01
5.44629753e-01 -3.80763948e-01 -4.32937264e-01 -4.04569596e-01
-1.32101551e-01 3.28018844e-01 5.31085253e-01 -7.02655554e-01
2.72983640e-01 5.47882020e-02 3.70472610e-01 -6.00648165e-01
1.18635364e-01 -1.02543092e+00 -3.20620760e-02 2.40514562e-01
-3.02094102e-01 -4.36917424e-01 2.75299430e-01 6.25839353e-01
-3.39498311e-01 -2.29348883e-01 7.05554545e-01 -2.68255144e-01
-6.99559331e-01 6.72635317e-01 1.81039833e-02 1.32125854e-01
9.07631099e-01 -6.69849291e-02 -1.17857173e-01 3.83962430e-02
-1.27124941e+00 4.48109627e-01 1.52152374e-01 3.41723204e-01
3.94357234e-01 -1.40559471e+00 -5.65396965e-01 -1.14403799e-01
2.40945101e-01 4.29156333e-01 8.17281961e-01 1.77553213e+00
-8.60720947e-02 1.56841099e-01 -2.03190699e-01 -1.01656330e+00
-1.11211634e+00 3.24331760e-01 4.44101214e-01 -7.99678862e-01
-7.39230156e-01 8.71478438e-01 5.87962329e-01 -8.20490718e-01
5.58977528e-03 -1.27513051e-01 -2.14180470e-01 -5.22936620e-02
1.81097999e-01 -1.17716424e-01 1.22119054e-01 -8.46983254e-01
-5.58270872e-01 6.79041266e-01 -3.83156687e-01 1.76691815e-01
1.09267187e+00 -1.34505868e-01 -7.94458464e-02 2.13960662e-01
1.22605634e+00 -3.63829851e-01 -1.26671040e+00 -7.20801413e-01
-2.01881528e-01 -2.26274252e-01 2.39053920e-01 -1.22011876e+00
-1.69556940e+00 7.29915679e-01 7.78776109e-01 -3.29530329e-01
1.36411095e+00 2.60572881e-01 1.04837251e+00 2.77849799e-03
2.63770580e-01 -9.73170280e-01 -7.00956257e-03 1.21575184e-01
5.54700553e-01 -1.70526469e+00 2.17724452e-03 -6.86445892e-01
-1.00009596e+00 1.01662672e+00 7.51792133e-01 3.64349097e-01
7.19738662e-01 6.99690357e-02 1.49326891e-01 -3.73542219e-01
-7.20187500e-02 -2.38774925e-01 7.48685300e-01 4.89911139e-01
5.15464902e-01 2.44762689e-01 -3.08891296e-01 8.72069120e-01
4.20958519e-01 1.01053856e-01 2.94300844e-03 9.72677529e-01
-3.33431289e-02 -1.26584363e+00 -1.89735502e-01 5.11125743e-01
-4.65741068e-01 -2.02073559e-01 5.13193607e-02 8.10308814e-01
4.24714565e-01 5.26253581e-01 -2.31266543e-01 -1.42288148e-01
1.53616726e-01 -1.65637076e-01 6.41002595e-01 -4.82728571e-01
-5.41608572e-01 2.90163815e-01 -1.10243663e-01 -4.58377331e-01
-7.65915275e-01 -6.25689983e-01 -1.58076978e+00 2.52627492e-01
-3.68036985e-01 -1.71586245e-01 4.63479638e-01 1.41317046e+00
1.37994453e-01 1.03712821e+00 3.19270343e-01 -8.37883651e-01
-2.57785261e-01 -8.25485706e-01 -4.00006562e-01 5.01022756e-01
9.11326632e-02 -9.15962100e-01 1.02165733e-02 1.74609363e-01]
|
[14.63249397277832, -2.1840310096740723]
|
cdd62ae4-9d48-417d-b178-6f0e55b50493
|
a-pooling-based-scene-text-proposal-technique
|
1811.10003
| null |
http://arxiv.org/abs/1811.10003v1
|
http://arxiv.org/pdf/1811.10003v1.pdf
|
A pooling based scene text proposal technique for scene text reading in the wild
|
Automatic reading texts in scenes has attracted increasing interest in recent
years as texts often carry rich semantic information that is useful for scene
understanding. In this paper, we propose a novel scene text proposal technique
aiming for accurate reading texts in scenes. Inspired by the pooling layer in
the deep neural network architecture, a pooling based scene text proposal
technique is developed. A novel score function is designed which exploits the
histogram of oriented gradients and is capable of ranking the proposals
according to their probabilities of being text. An end-to-end scene text
reading system has also been developed by incorporating the proposed scene text
proposal technique where false alarms elimination and words recognition are
performed simultaneously. Extensive experiments over several public datasets
show that the proposed technique can handle multi-orientation and
multi-language scene texts and obtains outstanding proposal performance. The
developed end-to-end systems also achieve very competitive scene text spotting
and reading performance.
|
['Shangxuan Tian', 'Mounir Mokhtari', 'Shijian Lu', 'Nizar Ouarti', 'Dinh NguyenVan']
|
2018-11-25
| null | null | null | null |
['text-spotting']
|
['computer-vision']
|
[ 5.33622205e-01 -2.87840217e-01 2.08636209e-01 -7.03556240e-01
-6.81281567e-01 -1.35317177e-01 1.15625715e+00 5.65986991e-01
-8.33258152e-01 3.02962691e-01 4.97498333e-01 -5.88372350e-02
4.81639318e-02 -9.65308666e-01 -5.92168927e-01 -4.77090091e-01
7.34734535e-01 3.18590194e-01 6.40140533e-01 -1.28344417e-01
8.71494114e-01 1.29359454e-01 -1.58473587e+00 4.16681647e-01
7.86774218e-01 9.83607292e-01 1.02756774e+00 8.64286125e-01
-6.76740527e-01 9.11985695e-01 -7.29768276e-01 -3.13909858e-01
1.70145884e-01 -2.93454230e-01 -5.75796604e-01 3.48540336e-01
7.02117682e-01 -4.05615687e-01 -3.72858793e-01 1.15798593e+00
7.76819646e-01 5.36422968e-01 6.49250448e-01 -4.27537888e-01
-5.50458074e-01 6.81430280e-01 -7.25300312e-01 4.08876389e-01
5.32674909e-01 -1.98954880e-01 1.10932970e+00 -1.22113085e+00
2.83410609e-01 1.20897734e+00 2.00921446e-01 9.04175267e-03
-6.13895237e-01 -1.20914064e-01 1.33522570e-01 4.49094385e-01
-1.35504723e+00 -2.49082088e-01 8.46719384e-01 -7.79993460e-02
1.22976398e+00 4.62932944e-01 2.27753714e-01 7.14319229e-01
3.70317519e-01 1.35645759e+00 7.44752169e-01 -7.36537457e-01
2.31822371e-01 2.84257740e-01 3.88716280e-01 5.63317478e-01
1.36647299e-01 -7.31022179e-01 -7.07759917e-01 3.67961854e-01
5.46813548e-01 2.22386137e-01 -1.23834617e-01 -6.72085434e-02
-1.37989831e+00 5.95647812e-01 3.61046821e-01 4.79489893e-01
-4.16497588e-01 -1.52510986e-01 5.33188403e-01 -2.19089508e-01
4.16489571e-01 2.91227341e-01 -1.56982064e-01 1.20961793e-01
-1.16466415e+00 2.44772673e-01 4.20780331e-01 9.53846514e-01
4.53853846e-01 5.45595586e-02 -6.19969904e-01 1.21161556e+00
2.50339299e-01 7.33095825e-01 7.95042932e-01 4.32247184e-02
8.16889226e-01 6.54350519e-01 5.33529092e-03 -1.38723624e+00
-6.49271607e-01 -3.81674320e-01 -8.80125523e-01 -3.15970957e-01
1.37723297e-01 2.24690914e-01 -9.11387026e-01 9.69835162e-01
2.53739804e-01 -5.81151135e-02 1.39110446e-01 9.02175486e-01
1.02107966e+00 1.01743948e+00 6.03438504e-02 1.64496720e-01
1.64085281e+00 -1.13379765e+00 -8.71365011e-01 -2.00738385e-01
3.06862563e-01 -1.16651905e+00 1.32698810e+00 5.38206637e-01
-9.90401804e-01 -8.94720316e-01 -9.80387688e-01 -3.38525534e-01
-7.08500564e-01 7.90439725e-01 1.46918029e-01 5.31375527e-01
-1.02615190e+00 -1.25295706e-02 -5.22543132e-01 -8.53793442e-01
3.12015116e-01 1.07695721e-01 8.02745819e-02 -8.79231691e-02
-9.40186203e-01 7.29524434e-01 8.24121714e-01 1.59520164e-01
-4.91652340e-01 1.11384854e-01 -8.04883480e-01 3.59203815e-01
2.84152269e-01 -5.34652352e-01 1.11038697e+00 -8.40204358e-01
-1.57009614e+00 7.51605988e-01 -3.21831465e-01 -3.75484407e-01
5.54063499e-01 -4.76924956e-01 -5.90381861e-01 4.97271717e-01
3.28483164e-01 5.95650256e-01 1.14738286e+00 -7.67388940e-01
-9.92009878e-01 -3.96254182e-01 -2.40329251e-01 8.90904963e-01
-5.76789856e-01 1.60482466e-01 -6.58451378e-01 -8.37986112e-01
3.18230003e-01 -2.01774538e-01 1.13674970e-02 -2.30582759e-01
-6.99771881e-01 -4.74838346e-01 9.63483095e-01 -7.50293434e-01
9.49768066e-01 -1.88947952e+00 -2.99255073e-01 1.03937715e-01
4.07666229e-02 2.89456666e-01 8.12693611e-02 3.91159564e-01
3.55882704e-01 -3.14478457e-01 -4.99570668e-02 -4.49554026e-01
9.50664356e-02 -3.10620010e-01 -2.98503280e-01 3.38591635e-01
6.84775189e-02 6.95193052e-01 -5.83056211e-01 -8.12551856e-01
1.03269100e+00 3.32921028e-01 -2.06829384e-01 1.73882365e-01
-2.77321428e-01 -6.78426251e-02 -7.73969233e-01 4.10040051e-01
8.72616887e-01 -2.20627517e-01 -2.31133997e-01 -2.96480954e-03
-2.52889544e-01 8.63557234e-02 -1.10792291e+00 1.61024296e+00
-2.47625664e-01 9.26152766e-01 -2.81569183e-01 -1.12033165e+00
1.16714823e+00 -1.97794922e-02 -1.29302591e-01 -1.15585279e+00
5.34716308e-01 -2.22623199e-01 -6.11518443e-01 -6.39503181e-01
1.42636335e+00 1.98604971e-01 -7.93923661e-02 6.29962906e-02
-1.79838344e-01 -2.22928733e-01 1.91617906e-01 2.33027428e-01
6.27048850e-01 1.26533238e-02 4.12044346e-01 -3.22314084e-01
1.01163554e+00 2.10157424e-01 -1.45022690e-01 1.07707262e+00
-1.15861639e-01 7.08245456e-01 4.45579886e-02 -3.28466326e-01
-9.73122537e-01 -8.81620765e-01 -1.20141841e-01 1.29701531e+00
6.16233706e-01 -1.23087920e-01 -7.36153007e-01 -3.72310817e-01
-4.99468595e-01 1.01768327e+00 -3.99831086e-01 2.43953064e-01
-3.48711699e-01 -6.16331518e-01 4.08965141e-01 6.22065127e-01
1.18461657e+00 -1.21814334e+00 -7.71735847e-01 2.28419974e-01
-3.93156677e-01 -1.33745480e+00 -3.15911800e-01 -1.66509226e-02
-5.55429459e-01 -7.50732005e-01 -1.07881141e+00 -1.26257479e+00
7.37664402e-01 7.52376199e-01 8.04937422e-01 -3.16824257e-01
-4.23740804e-01 4.24647063e-01 -7.41092205e-01 -6.68897390e-01
-8.89797285e-02 8.66168663e-02 -2.57069379e-01 3.19077790e-01
6.22784615e-01 2.94634104e-01 -7.33616352e-01 8.82384703e-02
-1.06212676e+00 3.87933046e-01 6.02149427e-01 6.45198226e-01
4.22259241e-01 4.01014447e-01 4.17166352e-01 -5.68329871e-01
7.32175350e-01 -1.60851806e-01 -7.54038393e-01 3.80390018e-01
-2.19215229e-01 -1.58073023e-01 9.37170982e-01 -5.84938340e-02
-1.57362235e+00 2.68145707e-02 -2.52906173e-01 6.59571663e-02
-6.06592536e-01 2.35942587e-01 -6.83896765e-02 2.88914740e-01
4.90262002e-01 8.09715211e-01 -6.41829193e-01 -4.80299383e-01
3.47699463e-01 9.42528665e-01 5.44625282e-01 -1.03406243e-01
4.63539124e-01 5.28881192e-01 -5.06316900e-01 -1.49078238e+00
-6.96705520e-01 -1.14553094e+00 -6.14617944e-01 -3.40475529e-01
1.29953837e+00 -1.01854515e+00 -4.98829007e-01 8.32725167e-01
-1.10880864e+00 1.05286755e-01 1.32490963e-01 6.26527488e-01
-4.41240549e-01 6.71338797e-01 -3.15503180e-01 -8.66744399e-01
-6.15553975e-01 -1.19130874e+00 1.53701055e+00 4.94730771e-01
1.33028299e-01 -8.30640554e-01 -2.98089176e-01 4.09423351e-01
5.48141360e-01 -3.31602782e-01 6.53583944e-01 -9.11130965e-01
-4.70569670e-01 -4.15000200e-01 -6.81101263e-01 1.52781233e-01
-9.61442217e-02 -3.89206946e-01 -1.02908432e+00 -5.31958751e-02
1.19893163e-01 -8.96194205e-02 1.01219988e+00 6.73046768e-01
1.25310338e+00 3.96240465e-02 -2.07919970e-01 2.77108938e-01
1.53407598e+00 1.65920049e-01 7.47824550e-01 6.20826602e-01
9.30748880e-01 5.26354015e-01 8.42932701e-01 7.61054337e-01
2.97546297e-01 5.40076017e-01 2.24654987e-01 -3.54901820e-01
4.12976407e-02 -1.24111690e-01 2.67128110e-01 5.77144802e-01
5.91614604e-01 -8.15789819e-01 -7.90861189e-01 6.52035236e-01
-1.78355908e+00 -9.76177037e-01 -4.98358518e-01 1.83685613e+00
3.31855446e-01 3.67660105e-01 -1.78329527e-01 2.46867597e-01
1.01474643e+00 5.13982952e-01 -4.12154466e-01 -5.04476011e-01
-3.97832572e-01 2.65100241e-01 6.56692088e-01 3.01154017e-01
-1.42087424e+00 1.38445151e+00 5.50360632e+00 1.18074799e+00
-1.16936731e+00 -1.49644524e-01 6.75647140e-01 2.99743533e-01
3.10203820e-01 -4.10741270e-01 -1.04865086e+00 2.21098334e-01
3.59803557e-01 -6.99281394e-02 -1.47754818e-01 9.32530940e-01
5.53499639e-01 -7.12082863e-01 -3.72652918e-01 1.14872181e+00
7.00001240e-01 -1.17739820e+00 4.94303405e-01 -5.78958333e-01
6.68752909e-01 -1.66832376e-02 9.25183445e-02 9.72175822e-02
-4.53819707e-02 -6.84603870e-01 8.58212113e-01 4.94147032e-01
5.07512450e-01 -7.70281076e-01 8.49790454e-01 4.79044646e-01
-1.26582170e+00 3.02987769e-02 -6.65232599e-01 -1.41646981e-01
1.65303916e-01 4.18682426e-01 -1.36076462e+00 5.11884987e-01
6.92506075e-01 7.46769845e-01 -8.63785386e-01 1.10865176e+00
-1.43086806e-01 6.58574343e-01 -3.26638073e-01 -7.28586197e-01
5.83784282e-01 -5.30884154e-02 5.47332883e-01 1.45968902e+00
2.96695232e-01 -3.68310250e-02 1.34536833e-01 6.31188929e-01
3.51481996e-02 7.89606869e-01 -4.99618441e-01 2.02901930e-01
-5.05068526e-02 1.36622548e+00 -1.43320262e+00 -7.38683581e-01
-3.90653640e-01 1.29614830e+00 -1.16087496e-01 2.25213334e-01
-7.08683312e-01 -7.51359165e-01 -2.13801891e-01 -2.82176435e-01
5.74852109e-01 -2.19292879e-01 -4.90634769e-01 -1.14584219e+00
1.68487981e-01 -5.39960921e-01 1.59271374e-01 -1.01919162e+00
-9.29354250e-01 4.10417348e-01 -2.14835435e-01 -1.06954134e+00
3.82715493e-01 -7.40993023e-01 -6.29271150e-01 6.24229968e-01
-1.58446944e+00 -1.07643318e+00 -5.90422451e-01 8.31580043e-01
1.48262942e+00 -1.94722310e-01 3.90559971e-01 1.57646146e-02
-6.89753771e-01 5.63860655e-01 5.20060539e-01 1.27134860e-01
6.13200426e-01 -1.25761425e+00 4.08369452e-01 1.22728658e+00
2.23670229e-01 3.70813429e-01 7.13675022e-01 -6.33681595e-01
-1.18669891e+00 -1.29041207e+00 9.77960348e-01 -1.75038472e-01
3.27660322e-01 -4.59993452e-01 -7.41136611e-01 3.35573047e-01
4.64606047e-01 -7.58078814e-01 2.46997848e-01 -3.09650272e-01
2.36250069e-02 5.55296727e-02 -1.04967785e+00 8.07762444e-01
5.03563404e-01 -3.56017113e-01 -9.44721937e-01 5.82049847e-01
4.38621908e-01 -3.36254150e-01 -8.04165155e-02 -3.13677751e-02
2.78811634e-01 -9.33530688e-01 8.83396924e-01 1.11666128e-01
2.45010510e-01 -3.19695473e-01 -2.83771217e-01 -7.63690174e-01
-9.69699472e-02 -2.31296301e-01 5.22951722e-01 1.04189956e+00
1.84828088e-01 -4.45579648e-01 6.09070718e-01 2.58568764e-01
-1.64815515e-01 -1.85498029e-01 -5.59879124e-01 -3.94711375e-01
-4.10553843e-01 -4.66717899e-01 2.90997922e-01 6.86255991e-01
-7.31610209e-02 6.04992568e-01 -4.23019916e-01 1.79734558e-01
4.23340946e-01 -6.32937998e-02 7.70986915e-01 -9.26885664e-01
-1.55360863e-01 -6.94766521e-01 -4.93045837e-01 -1.67081177e+00
-1.10747546e-01 -7.58463740e-01 4.83426750e-01 -1.91090536e+00
4.49692994e-01 8.81767347e-02 -2.14129657e-01 6.58585578e-02
-5.29922724e-01 1.84676662e-01 1.78589836e-01 5.49402647e-02
-1.12284529e+00 7.40890920e-01 1.02534139e+00 -2.04652801e-01
-1.19384617e-01 -8.63793790e-02 -3.47251594e-01 7.36390591e-01
8.70774865e-01 3.61153111e-02 -3.08283508e-01 -6.78586423e-01
-4.34615510e-03 -2.04302803e-01 2.66412407e-01 -1.24914682e+00
4.49510843e-01 2.43756428e-01 6.80494010e-01 -1.47266066e+00
3.62262875e-01 -6.56016707e-01 -6.80254221e-01 5.18552773e-02
-7.14630842e-01 4.50946800e-02 1.31792635e-01 8.27652514e-01
-2.78571010e-01 -2.61753321e-01 7.19018281e-01 -1.84220225e-02
-1.19155228e+00 -2.59805441e-01 -6.21522427e-01 -2.38075301e-01
9.70694602e-01 -4.53605741e-01 -2.97037065e-01 -4.17736709e-01
-3.68230820e-01 2.64950305e-01 2.06281945e-01 5.50045490e-01
9.66337144e-01 -7.04500258e-01 -7.28885770e-01 6.21962994e-02
4.02284414e-01 -1.84739549e-02 3.82072717e-01 6.13160372e-01
-8.59995663e-01 7.45955586e-01 7.49179255e-03 -8.07359159e-01
-1.39876652e+00 2.79996514e-01 -2.61911843e-02 -2.97637165e-01
-8.23925495e-01 9.79667544e-01 4.30572212e-01 -1.79114074e-01
3.84848475e-01 -4.56724882e-01 -6.35232329e-01 -6.40216693e-02
7.55673051e-01 2.97168970e-01 1.42098576e-01 -9.88771379e-01
-6.88079670e-02 8.13916266e-01 -4.15479422e-01 -1.07522368e-01
1.05866635e+00 -5.41347384e-01 2.44407862e-01 4.11021262e-01
9.74681556e-01 -7.43784383e-02 -8.84017408e-01 -3.81051630e-01
3.55604105e-02 -4.43489701e-01 3.67384851e-01 -8.46096098e-01
-5.82928717e-01 1.19183910e+00 6.80323601e-01 1.85204595e-01
1.25763845e+00 -2.58734941e-01 9.72747803e-01 7.38747358e-01
1.90546423e-01 -1.48410356e+00 2.01466352e-01 6.85022116e-01
6.36591554e-01 -1.52384365e+00 -4.53908481e-02 -1.81322679e-01
-7.63785243e-01 1.18606734e+00 4.79138345e-01 -8.53338912e-02
2.29474708e-01 -1.99403301e-01 -6.98965089e-03 -3.29615921e-01
-1.14663310e-01 -5.34369528e-01 2.93416977e-01 3.00864667e-01
2.54634917e-01 1.83947086e-01 -4.24423516e-01 1.70598224e-01
-2.06433639e-01 -4.63955343e-01 4.56936955e-01 8.62580597e-01
-1.01404846e+00 -5.50798357e-01 -6.43036604e-01 6.33775473e-01
-5.63281953e-01 -3.28550488e-01 -4.26992744e-01 5.33126295e-01
-2.62228757e-01 1.24538147e+00 6.88233152e-02 -1.85980260e-01
3.53943110e-01 4.45504077e-02 8.83675441e-02 -4.66541260e-01
-5.64657152e-01 3.49247247e-01 -7.39724487e-02 -3.07507068e-01
-3.53421777e-01 -4.45452631e-01 -1.42935133e+00 1.35328844e-01
-7.61210084e-01 -1.71144500e-01 9.06059384e-01 1.01892543e+00
7.17250109e-02 5.87426066e-01 6.40042484e-01 -8.37785602e-01
-1.17730379e-01 -1.13572562e+00 -4.96738970e-01 4.09944385e-01
1.13121182e-01 -3.58118594e-01 -4.55775075e-02 2.06570402e-01]
|
[12.028962135314941, 2.2874255180358887]
|
7c8a72c3-6fa3-490c-b4fb-78c879f8cc32
|
flamingo-a-visual-language-model-for-few-shot-1
|
2204.14198
| null |
https://arxiv.org/abs/2204.14198v2
|
https://arxiv.org/pdf/2204.14198v2.pdf
|
Flamingo: a Visual Language Model for Few-Shot Learning
|
Building models that can be rapidly adapted to novel tasks using only a handful of annotated examples is an open challenge for multimodal machine learning research. We introduce Flamingo, a family of Visual Language Models (VLM) with this ability. We propose key architectural innovations to: (i) bridge powerful pretrained vision-only and language-only models, (ii) handle sequences of arbitrarily interleaved visual and textual data, and (iii) seamlessly ingest images or videos as inputs. Thanks to their flexibility, Flamingo models can be trained on large-scale multimodal web corpora containing arbitrarily interleaved text and images, which is key to endow them with in-context few-shot learning capabilities. We perform a thorough evaluation of our models, exploring and measuring their ability to rapidly adapt to a variety of image and video tasks. These include open-ended tasks such as visual question-answering, where the model is prompted with a question which it has to answer; captioning tasks, which evaluate the ability to describe a scene or an event; and close-ended tasks such as multiple-choice visual question-answering. For tasks lying anywhere on this spectrum, a single Flamingo model can achieve a new state of the art with few-shot learning, simply by prompting the model with task-specific examples. On numerous benchmarks, Flamingo outperforms models fine-tuned on thousands of times more task-specific data.
|
['Karen Simonyan', 'Andrew Zisserman', 'Oriol Vinyals', 'Ricardo Barreira', 'Mikolaj Binkowski', 'Sahand Sharifzadeh', 'Aida Nematzadeh', 'Andrew Brock', 'Sebastian Borgeaud', 'Jacob Menick', 'Marianne Monteiro', 'Sina Samangooei', 'Zhitao Gong', 'Tengda Han', 'Serkan Cabi', 'Eliza Rutherford', 'Roman Ring', 'Malcolm Reynolds', 'Katie Millican', 'Arthur Mensch', 'Karel Lenc', 'Yana Hasson', 'Iain Barr', 'Antoine Miech', 'Pauline Luc', 'Jeff Donahue', 'Jean-Baptiste Alayrac']
|
2022-04-29
|
flamingo-a-visual-language-model-for-few-shot
|
https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/tackling-multiple-tasks-with-a-single-visual-language-model/flamingo.pdf
|
https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/tackling-multiple-tasks-with-a-single-visual-language-model/flamingo.pdf
|
deepmind-2022-4
|
['generative-visual-question-answering', 'video-question-answering', 'zero-shot-cross-modal-retrieval']
|
['computer-vision', 'computer-vision', 'miscellaneous']
|
[ 1.96652368e-01 -2.35556170e-01 -3.18357408e-01 -2.24551693e-01
-9.66002166e-01 -7.68734097e-01 9.96869981e-01 6.83937967e-02
-7.23022699e-01 4.50936764e-01 3.63296390e-01 -3.45739663e-01
2.30330944e-01 -3.63334328e-01 -7.75788367e-01 -1.57366738e-01
-1.28989309e-01 5.45466721e-01 4.95432585e-01 -4.29521114e-01
1.84092280e-02 1.35139883e-01 -1.72577620e+00 7.88375795e-01
3.83702099e-01 1.11317468e+00 4.48158085e-01 1.16008747e+00
-4.03135151e-01 1.17526782e+00 -2.34750181e-01 -4.70438421e-01
-5.71394190e-02 -3.33156586e-01 -9.96624529e-01 3.30947995e-01
7.53490150e-01 -4.48184520e-01 -5.17537892e-01 5.22775233e-01
4.93023545e-01 5.37397444e-01 6.09978557e-01 -1.42865348e+00
-9.80391562e-01 1.98564798e-01 -2.73122400e-01 5.76389909e-01
8.14326823e-01 9.14090633e-01 1.12489116e+00 -1.24474192e+00
9.86289978e-01 1.35396874e+00 3.34617287e-01 8.63104403e-01
-1.22840345e+00 -2.29174510e-01 2.63066769e-01 6.62499189e-01
-1.02683783e+00 -7.28617370e-01 5.78010142e-01 -4.43916142e-01
1.31023777e+00 2.34920755e-01 4.60549533e-01 1.58196020e+00
-2.05305949e-01 1.04878557e+00 8.61374497e-01 -2.34716833e-01
2.64422119e-01 -1.95609443e-02 4.98168729e-02 7.65429616e-01
-3.91898036e-01 -1.04110777e-01 -8.00946593e-01 -3.17526311e-02
3.55216503e-01 5.92197403e-02 -2.13356331e-01 -4.86562669e-01
-1.41510117e+00 8.48446131e-01 3.89990062e-01 1.68222621e-01
-1.49863791e-02 2.72913128e-01 6.21748686e-01 4.96960342e-01
9.04962644e-02 6.17596447e-01 -3.59048784e-01 -1.71675354e-01
-6.90140784e-01 1.61061049e-01 6.92569315e-01 1.07239866e+00
9.05919254e-01 5.67756072e-02 -5.70638657e-01 9.79406595e-01
7.22203180e-02 4.56077605e-01 6.76313758e-01 -1.02418602e+00
5.91975808e-01 4.69768733e-01 1.20403960e-01 -4.71760452e-01
-3.58962029e-01 2.27166712e-01 -4.72056061e-01 1.43908754e-01
5.16983688e-01 -1.04351103e-01 -1.39853036e+00 1.65758693e+00
1.90686658e-01 4.86172140e-02 1.00535884e-01 8.31351519e-01
1.42549050e+00 9.27565098e-01 3.73664558e-01 2.58834269e-02
1.53065646e+00 -1.29636705e+00 -4.96482193e-01 -7.16082931e-01
5.24054050e-01 -6.94275558e-01 1.80278218e+00 -7.14294380e-03
-1.08089817e+00 -7.40818501e-01 -6.94945931e-01 -4.11145478e-01
-7.27979064e-01 -4.77387458e-01 4.48959351e-01 1.01274952e-01
-1.16194475e+00 2.32470214e-01 -5.20017982e-01 -7.71396160e-01
3.96713287e-01 -7.94688910e-02 -6.44395590e-01 -5.39262474e-01
-1.03776586e+00 9.47773218e-01 3.67097050e-01 -3.42521399e-01
-1.32641804e+00 -6.83887482e-01 -1.27144217e+00 1.89018413e-01
7.25209594e-01 -9.11746919e-01 1.45955634e+00 -1.01825869e+00
-1.09406734e+00 1.08483446e+00 -2.01271385e-01 -2.65804589e-01
5.36120951e-01 -3.98928020e-03 -4.77015972e-01 6.16367102e-01
4.52939682e-02 1.10963178e+00 1.07265806e+00 -1.30055618e+00
-4.77806538e-01 -6.60921112e-02 4.05557692e-01 2.41326556e-01
-3.87771100e-01 8.62078443e-02 -7.80294299e-01 -4.45468664e-01
-5.91379285e-01 -6.21030688e-01 -1.12359717e-01 3.76757711e-01
-7.35787600e-02 -3.16864699e-01 1.11751986e+00 -4.92961019e-01
8.44439030e-01 -2.02269340e+00 1.74589619e-01 -4.13809478e-01
2.55051821e-01 3.41892660e-01 -6.77254379e-01 7.65085518e-01
1.51744097e-01 6.56182840e-02 -1.02682933e-01 -4.72844273e-01
1.21689945e-01 3.65798056e-01 -2.91846454e-01 -1.62988193e-02
3.82480294e-01 1.58061254e+00 -1.15133357e+00 -7.61892855e-01
3.37154001e-01 2.02345163e-01 -3.17320466e-01 5.63391864e-01
-6.92800105e-01 1.54339224e-01 6.47967234e-02 7.47407436e-01
6.65793046e-02 -7.49153614e-01 -1.07141316e-01 -2.04701379e-01
1.89010784e-01 -4.52743053e-01 -8.66985440e-01 1.87573183e+00
-6.56528234e-01 9.90504503e-01 -1.23527497e-01 -7.15595245e-01
5.99757910e-01 3.17679137e-01 6.57979846e-02 -9.41018939e-01
-9.85223204e-02 -2.64638007e-01 -4.13405269e-01 -1.19030607e+00
3.58031809e-01 -9.64653790e-02 -1.81110770e-01 4.24663782e-01
7.63724864e-01 8.32964480e-03 4.89755422e-01 5.32349765e-01
1.20335674e+00 3.38070393e-02 4.16314811e-01 3.24557692e-01
2.51958460e-01 1.16917536e-01 -3.78205143e-02 1.03783596e+00
-3.99876118e-01 7.17794418e-01 4.24691975e-01 -7.68780291e-01
-1.28573847e+00 -1.22944522e+00 3.36463422e-01 1.86968970e+00
6.56733066e-02 -4.04585123e-01 -2.40537867e-01 -7.49243021e-01
2.44324077e-02 6.77621424e-01 -8.79216433e-01 2.01702844e-02
-2.88919866e-01 -2.89032429e-01 3.08933109e-01 6.95475638e-01
3.17100883e-01 -1.29248178e+00 -8.16305339e-01 -8.32506865e-02
-3.39337498e-01 -1.47835684e+00 -6.76328778e-01 6.35832399e-02
-4.08567488e-01 -1.13803589e+00 -9.95425522e-01 -9.76043701e-01
3.70929003e-01 4.24469978e-01 1.39002037e+00 1.24282628e-01
-4.53018248e-01 1.14091492e+00 -5.74684501e-01 -1.66884407e-01
-2.49405831e-01 -1.90953746e-01 -3.36747974e-01 2.74351835e-01
2.27204159e-01 -3.08366299e-01 -5.54535985e-01 1.69066027e-01
-1.19705331e+00 -8.40755086e-03 4.02099282e-01 9.58947122e-01
3.69405597e-01 -9.29368079e-01 4.82547522e-01 -6.71247005e-01
5.90766907e-01 -6.67902291e-01 -2.31356487e-01 6.84055507e-01
-4.64995466e-02 3.91443260e-02 5.89891374e-01 -8.88232052e-01
-8.93009067e-01 1.27521649e-01 -8.73009190e-02 -7.80632317e-01
-2.57344514e-01 3.76331061e-01 7.25007579e-02 -7.83288330e-02
8.73456478e-01 3.75508279e-01 -8.44413415e-02 -2.81200171e-01
9.44832802e-01 3.83175880e-01 8.77801061e-01 -3.53203535e-01
7.17735708e-01 5.61998069e-01 -2.98377007e-01 -1.12272274e+00
-8.77332211e-01 -7.57465601e-01 -6.17423713e-01 -4.49659258e-01
1.14281523e+00 -8.13883364e-01 -7.88859725e-01 1.85410991e-01
-1.11308265e+00 -7.29002297e-01 -2.73505002e-01 9.54020917e-02
-7.37374246e-01 4.59005654e-01 -4.47696745e-01 -6.47997141e-01
-2.11918250e-01 -9.52206969e-01 1.13590646e+00 3.54353040e-01
-2.40671486e-01 -1.16772711e+00 8.80811885e-02 4.81299430e-01
4.60055053e-01 2.76164711e-01 1.04132056e+00 -7.87448406e-01
-4.45508659e-01 -8.31843987e-02 -4.55754071e-01 5.11317924e-02
-3.14769685e-01 -1.96161523e-01 -1.08017063e+00 -4.80892211e-01
-3.52210432e-01 -1.17642725e+00 1.09249389e+00 -7.40112131e-03
1.10962343e+00 -2.10839823e-01 -2.73599625e-01 6.24104321e-01
1.41301751e+00 -3.22182067e-02 4.84749496e-01 2.22558662e-01
6.44272506e-01 4.72272009e-01 3.99504393e-01 3.80471289e-01
7.34271884e-01 6.85595632e-01 7.34540462e-01 -6.21664226e-02
-3.55354398e-01 -4.43266302e-01 2.38223210e-01 4.91754860e-01
5.02875596e-02 -3.53922933e-01 -1.19976890e+00 7.71933854e-01
-1.99074447e+00 -1.26812661e+00 3.15013349e-01 1.83463585e+00
6.38255298e-01 -6.50632903e-02 2.48606995e-01 -5.32971084e-01
4.16851878e-01 5.86309552e-01 -7.39646912e-01 -3.53071630e-01
-1.01450861e-01 1.97326317e-02 1.43737812e-02 4.56172138e-01
-1.18914914e+00 1.03375578e+00 6.59699631e+00 6.42335653e-01
-1.04468775e+00 3.84396613e-01 4.02044922e-01 -3.01108658e-01
-3.85722667e-01 -6.44694343e-02 -5.58981597e-01 3.35421860e-01
7.70531178e-01 -1.23888679e-01 5.70586324e-01 7.93267727e-01
-2.23627314e-01 -4.24771979e-02 -1.34135807e+00 1.30477035e+00
7.12537348e-01 -1.56369710e+00 4.19709355e-01 -3.83121818e-01
6.48051977e-01 2.23441288e-01 1.07123137e-01 6.85349286e-01
2.99165130e-01 -1.32683659e+00 4.91238415e-01 6.59997284e-01
1.01365960e+00 -3.12589824e-01 3.66411358e-01 3.00060123e-01
-1.21560395e+00 -4.33339000e-01 -2.75599986e-01 -4.47849855e-02
4.50458527e-01 -2.42517143e-01 -8.40082645e-01 2.19385430e-01
9.03899014e-01 5.55598855e-01 -1.01516020e+00 1.16265154e+00
-1.09027989e-01 3.17707151e-01 -1.20829055e-02 -1.76158085e-01
5.12933731e-01 4.55306381e-01 3.90409857e-01 1.30141997e+00
4.64882292e-02 6.09216578e-02 2.89132804e-01 5.61583519e-01
-2.37717047e-01 5.03614694e-02 -9.27223027e-01 -9.06175673e-02
3.64829481e-01 1.23156822e+00 -5.46725273e-01 -5.73025107e-01
-8.85255396e-01 1.12993813e+00 7.41613090e-01 7.55201161e-01
-6.91871524e-01 -3.87172848e-01 4.34546888e-01 3.52823883e-02
5.60069799e-01 -2.85985798e-01 3.36153954e-01 -1.36708987e+00
-1.08432055e-01 -9.22326386e-01 7.73849726e-01 -1.40913737e+00
-1.44198811e+00 7.14056015e-01 4.56517600e-02 -1.01140571e+00
-4.74228054e-01 -8.22788596e-01 -6.88813686e-01 3.90461504e-01
-1.59294224e+00 -1.47234464e+00 -6.77411616e-01 9.08113897e-01
1.04192030e+00 -2.63475329e-01 7.13263333e-01 1.54835775e-01
-1.81197271e-01 4.72136408e-01 -1.71766177e-01 2.49135077e-01
9.31847930e-01 -1.12864947e+00 4.97583032e-01 5.56043565e-01
5.76363862e-01 1.27116874e-01 5.94040155e-01 -2.77970403e-01
-1.62225091e+00 -1.05571270e+00 8.18578899e-01 -9.61868346e-01
8.80370855e-01 -6.03612721e-01 -1.07640386e+00 9.73475933e-01
5.05685687e-01 4.91680562e-01 7.26196468e-01 8.07454437e-02
-8.02661836e-01 9.40764621e-02 -7.25059628e-01 7.71849036e-01
1.13338363e+00 -9.29799318e-01 -9.22607839e-01 5.50423622e-01
8.91405582e-01 -2.98493236e-01 -6.34359002e-01 2.78662503e-01
5.67845821e-01 -7.85881519e-01 1.24940467e+00 -1.36192060e+00
7.07751155e-01 -3.07376906e-02 -3.19704115e-01 -1.16082716e+00
-2.56124765e-01 -6.24719977e-01 -5.12923717e-01 9.53626275e-01
3.78764421e-01 -9.65797231e-02 2.83530623e-01 5.37104726e-01
-1.84653662e-02 -6.71944439e-01 -7.43701398e-01 -7.72639394e-01
-1.97162688e-01 -4.20127988e-01 2.59330899e-01 9.23138618e-01
1.29576713e-01 7.69504070e-01 -6.72127485e-01 -1.64973497e-01
3.43650848e-01 6.30267859e-02 9.45186257e-01 -8.78830135e-01
-4.18914467e-01 -4.09946620e-01 -3.72857422e-01 -9.84601080e-01
8.66967812e-02 -9.07557666e-01 -8.31823871e-02 -1.74785113e+00
4.72672194e-01 3.41511518e-01 -1.97055131e-01 6.45597696e-01
-4.06276226e-01 5.10527015e-01 6.61561847e-01 8.94584432e-02
-1.36254716e+00 4.61774588e-01 1.13359046e+00 -4.68748212e-01
-1.37960747e-01 -4.24196392e-01 -4.16326940e-01 7.12800324e-01
3.00408304e-01 -5.35096750e-02 -5.33837616e-01 -6.94880247e-01
3.17567796e-01 2.80686855e-01 8.03125143e-01 -8.41717720e-01
2.72170842e-01 -3.44418228e-01 5.60300410e-01 -2.87234604e-01
6.96408272e-01 -5.41110039e-01 -2.66447753e-01 1.74153388e-01
-6.04723036e-01 3.16509128e-01 2.74851680e-01 7.50666261e-01
-2.30050683e-01 -2.68999815e-01 7.63262868e-01 -4.55692142e-01
-1.74617243e+00 4.21250790e-01 -2.56794512e-01 5.93174756e-01
1.22174704e+00 -7.79960379e-02 -7.25460947e-01 -7.58295596e-01
-1.01818287e+00 6.76088989e-01 5.20169914e-01 8.79803956e-01
8.79826367e-01 -1.34629619e+00 -5.91157913e-01 -1.82629541e-01
9.38637435e-01 -4.44691718e-01 6.14985645e-01 5.36787212e-01
-3.12922001e-01 3.37053686e-01 -3.12400371e-01 -7.36580372e-01
-1.27114773e+00 1.22268164e+00 7.00129792e-02 -8.37892853e-03
-6.07501090e-01 9.39163148e-01 2.44638607e-01 -3.89778912e-01
3.34252179e-01 7.18449950e-02 -2.88306326e-01 2.95958042e-01
8.81454706e-01 1.07908234e-01 -3.16108793e-01 -6.27013206e-01
-3.35440248e-01 4.76393282e-01 4.41315509e-02 -1.69513032e-01
1.08628023e+00 -2.68115133e-01 3.56495976e-01 7.47716367e-01
1.35139728e+00 -5.85302830e-01 -1.56939423e+00 -4.55447584e-01
-1.33696422e-01 -3.83857459e-01 -4.19486493e-01 -1.03410208e+00
-5.81825852e-01 1.27965868e+00 4.27126855e-01 7.26980641e-02
8.88913691e-01 4.96622235e-01 6.78700447e-01 7.06037521e-01
1.41707942e-01 -1.04765606e+00 9.85756814e-01 6.00775540e-01
9.68567431e-01 -1.78087831e+00 -3.52254957e-01 3.24369609e-01
-1.20290816e+00 1.02183115e+00 7.74346232e-01 2.02574745e-01
2.43846625e-01 -1.35386899e-01 2.30489671e-01 -1.48246482e-01
-1.34102309e+00 -7.05166221e-01 5.70586681e-01 8.98980498e-01
1.66639108e-02 -2.54153252e-01 2.75404692e-01 3.50482285e-01
2.68041968e-01 -4.79722135e-02 3.50197285e-01 8.62770140e-01
-5.94491184e-01 -6.61233902e-01 -1.98230088e-01 4.84907746e-01
8.48648921e-02 -1.10102490e-01 -3.92513424e-01 9.00355399e-01
-1.97573423e-01 9.69094813e-01 1.44090950e-01 -2.52263904e-01
2.70621121e-01 5.23870289e-01 4.20388877e-01 -7.77498484e-01
-2.53710091e-01 -3.43594313e-01 1.95887133e-01 -7.70079672e-01
-2.39753798e-01 -4.30567443e-01 -9.03975785e-01 3.17432694e-02
3.22554648e-01 -2.05231860e-01 3.41598928e-01 1.16736531e+00
3.87624800e-01 3.26068789e-01 2.44220331e-01 -1.05242944e+00
-4.85596180e-01 -7.64895856e-01 -4.16974872e-02 8.05755973e-01
6.19795680e-01 -5.26638806e-01 -1.19483978e-01 3.88085008e-01]
|
[10.692691802978516, 1.726918339729309]
|
75fda6c0-e8b7-4880-b90f-f2403673f906
|
robust-reference-based-super-resolution-with
| null | null |
http://openaccess.thecvf.com/content_CVPR_2020/html/Shim_Robust_Reference-Based_Super-Resolution_With_Similarity-Aware_Deformable_Convolution_CVPR_2020_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2020/papers/Shim_Robust_Reference-Based_Super-Resolution_With_Similarity-Aware_Deformable_Convolution_CVPR_2020_paper.pdf
|
Robust Reference-Based Super-Resolution With Similarity-Aware Deformable Convolution
|
In this paper, we propose a novel and efficient reference feature extraction module referred to as the Similarity Search and Extraction Network (SSEN) for reference-based super-resolution (RefSR) tasks. The proposed module extracts aligned relevant features from a reference image to increase the performance over single image super-resolution (SISR) methods. In contrast to conventional algorithms which utilize brute-force searches or optical flow estimations, the proposed algorithm is end-to-end trainable without any additional supervision or heavy computation, predicting the best match with a single network forward operation. Moreover, the proposed module is aware of not only the best matching position but also the relevancy of the best match. This makes our algorithm substantially robust when irrelevant reference images are given, overcoming the major cause of the performance degradation when using existing RefSR methods. Furthermore, our module can be utilized for self-similarity SR if no reference image is available. Experimental results demonstrate the superior performance of the proposed algorithm compared to previous works both quantitatively and qualitatively.
|
[' In So Kweon', ' Jinsun Park', 'Gyumin Shim']
|
2020-06-01
| null | null | null |
cvpr-2020-6
|
['reference-based-super-resolution']
|
['computer-vision']
|
[ 5.93021989e-01 -1.47570744e-01 -1.93085566e-01 -1.19352333e-01
-7.30726242e-01 -1.50249466e-01 4.44890797e-01 -1.72907308e-01
-5.92086434e-01 7.78263390e-01 1.89724445e-01 2.36318380e-01
-3.05386811e-01 -5.84790409e-01 -5.03384769e-01 -5.50831497e-01
2.48767540e-01 -1.30076278e-02 5.38706005e-01 -3.17291111e-01
4.67217773e-01 7.08215117e-01 -1.71080303e+00 -8.01467746e-02
1.01960933e+00 1.04265153e+00 7.89415300e-01 2.75778651e-01
2.21509635e-01 9.26096022e-01 -2.19829977e-01 -3.40729579e-02
4.08801317e-01 -3.54857743e-01 -7.32599080e-01 1.18709520e-01
8.15295219e-01 -5.41302085e-01 -5.48004210e-01 1.09423828e+00
5.90315282e-01 6.52133882e-01 1.49532378e-01 -6.44730926e-01
-6.87676013e-01 3.39171112e-01 -8.15211117e-01 6.66525006e-01
4.73860800e-01 2.70985682e-02 8.44917119e-01 -1.20169330e+00
9.71154392e-01 1.05662072e+00 3.55925441e-01 6.24143958e-01
-1.02746165e+00 -7.30085611e-01 1.84315220e-01 3.23014855e-01
-1.55240238e+00 -6.57497287e-01 1.00573063e+00 -1.20790169e-01
5.03457963e-01 3.58274356e-02 2.30217665e-01 7.59683490e-01
-1.29459068e-01 5.17506242e-01 1.18024075e+00 -2.97654152e-01
7.93003142e-02 8.16667825e-02 4.71278839e-02 7.94423580e-01
1.81252077e-01 3.25307369e-01 -7.15294302e-01 1.91371948e-01
1.24769020e+00 8.55212472e-03 -6.09997272e-01 -4.66707617e-01
-1.33840322e+00 4.74809796e-01 6.99485421e-01 4.24190074e-01
-5.61681390e-01 -1.08455360e-01 1.68054447e-01 7.20902532e-02
2.17311949e-01 3.43928099e-01 6.55165091e-02 8.98722336e-02
-1.28772211e+00 -2.22716499e-02 1.55871540e-01 8.00422847e-01
8.28839064e-01 2.62768596e-01 -7.04653263e-02 8.97402287e-01
-1.23411283e-01 1.58698007e-01 6.41475558e-01 -1.29094267e+00
4.45600837e-01 3.63905817e-01 5.04421651e-01 -1.31738985e+00
-2.01785669e-01 -8.82969379e-01 -1.04078019e+00 1.92913190e-01
1.31516024e-01 2.67087489e-01 -6.89625740e-01 1.56948912e+00
3.88569236e-01 5.35142899e-01 2.33526662e-01 1.41902578e+00
9.24492300e-01 3.21798861e-01 -3.36863875e-01 -3.89091939e-01
1.02216876e+00 -1.04863966e+00 -6.39107466e-01 -1.92736596e-01
9.99285355e-02 -8.79284084e-01 7.78610766e-01 1.03130266e-01
-1.23250782e+00 -9.09585953e-01 -1.08704937e+00 -1.55876234e-01
5.10768071e-02 2.36776829e-01 3.65417421e-01 5.47208451e-02
-1.09611320e+00 9.09470916e-01 -6.48643315e-01 -1.87520415e-01
3.77270430e-01 3.58482420e-01 -5.20063996e-01 -2.42598891e-01
-1.06746566e+00 9.35467720e-01 3.25972676e-01 2.98152804e-01
-6.90400481e-01 -6.86164320e-01 -7.61583984e-01 8.37336481e-02
5.11295557e-01 -7.01539457e-01 8.64380479e-01 -1.00814414e+00
-1.48842275e+00 5.41038632e-01 -5.58267117e-01 -5.01976371e-01
5.46112776e-01 -2.72776544e-01 -3.57801437e-01 5.25716960e-01
1.09106287e-01 4.19516742e-01 8.64381135e-01 -1.27427518e+00
-7.84075677e-01 -2.20285580e-01 1.38055459e-01 3.94014210e-01
-1.50039226e-01 1.29471302e-01 -3.23256701e-01 -6.67185485e-01
4.37342614e-01 -6.27052665e-01 -3.85566711e-01 1.34689614e-01
-9.56191272e-02 -3.98222506e-02 7.65431941e-01 -6.67148173e-01
1.13807213e+00 -2.13416600e+00 2.87316833e-02 5.12088686e-02
3.30503196e-01 4.77549374e-01 -3.25418770e-01 1.78884163e-01
-1.46533206e-01 -1.71334848e-01 -6.50274381e-02 -6.67510256e-02
-5.32695830e-01 -1.27731711e-01 -1.48701787e-01 5.09645939e-01
2.33616188e-01 5.93451619e-01 -1.06467974e+00 -7.12981045e-01
4.99219567e-01 7.75139987e-01 -2.53821254e-01 1.88869566e-01
3.42098206e-01 7.55897284e-01 -6.15982950e-01 3.81049842e-01
6.77178204e-01 -3.49869877e-01 -1.94021598e-01 -6.44986212e-01
-2.29111910e-01 2.18260393e-01 -1.48135865e+00 1.84796321e+00
-6.78120792e-01 4.42212671e-01 8.60671252e-02 -8.61775517e-01
1.27384090e+00 1.75983205e-01 5.36897659e-01 -9.35794771e-01
-1.37586117e-01 2.65636086e-01 -1.42286330e-01 -2.69831955e-01
6.91477239e-01 2.62447387e-01 5.44551790e-01 1.74956426e-01
-8.96539763e-02 5.75736880e-01 3.43181074e-01 2.81383216e-01
9.47456598e-01 3.67951840e-01 4.93669987e-01 2.78291516e-02
1.09932899e+00 -1.37804478e-01 8.84555101e-01 6.91480577e-01
-3.29758018e-01 6.42000377e-01 -3.57329339e-01 -5.22077084e-01
-1.02915359e+00 -9.86021817e-01 -3.73571962e-02 6.66052103e-01
7.13845670e-01 -2.59023815e-01 -2.94457227e-01 -4.24219877e-01
-3.22217077e-01 2.36542344e-01 -4.12995040e-01 6.67459220e-02
-9.23442721e-01 -2.44033054e-01 -1.21784322e-02 4.34049666e-01
9.28130507e-01 -8.96710217e-01 -7.87220955e-01 2.08573252e-01
-5.55452645e-01 -1.49541306e+00 -7.67161787e-01 -4.48876023e-01
-1.03341389e+00 -8.68327498e-01 -8.08916092e-01 -7.19037652e-01
7.47921705e-01 8.49378645e-01 8.23259294e-01 1.26082331e-01
-3.44446450e-01 1.24097779e-01 -2.61761725e-01 5.60165763e-01
-2.88411945e-01 -1.21778339e-01 -2.56387284e-03 3.20566565e-01
1.85805522e-02 -7.87089705e-01 -1.07233918e+00 5.13392806e-01
-5.14468551e-01 1.96831584e-01 7.82229245e-01 8.87588918e-01
8.92308772e-01 -1.71912722e-02 7.82671213e-01 -6.06825590e-01
3.00085872e-01 -2.28466131e-02 -6.32202446e-01 1.82634786e-01
-8.13753009e-01 1.33706003e-01 7.89153039e-01 -2.99213499e-01
-1.34726703e+00 3.70853812e-01 3.17819506e-01 -8.40000331e-01
-1.34121001e-01 1.27977356e-01 1.88754335e-01 -4.15525585e-01
5.00271976e-01 6.52660549e-01 -3.54667567e-02 -5.31792045e-01
3.23952556e-01 5.85526764e-01 9.91586089e-01 -1.67463049e-01
1.08096957e+00 6.42835557e-01 2.34416410e-01 -5.90392351e-01
-8.57667625e-01 -7.45519400e-01 -9.09881532e-01 -2.25133106e-01
6.41985595e-01 -1.12707400e+00 -6.48193359e-01 2.32872348e-02
-9.69537318e-01 4.41100776e-01 -5.91836534e-02 6.22276545e-01
-5.69432199e-01 7.21965909e-01 -4.37059015e-01 -8.51168990e-01
-6.48765743e-01 -1.07100427e+00 8.59341383e-01 4.61653143e-01
9.22468454e-02 -6.73240244e-01 -2.45882824e-01 5.43019712e-01
6.45082355e-01 2.68652350e-01 2.31719360e-01 -3.50757778e-01
-1.00904465e+00 3.03807463e-02 -6.91972077e-01 1.60907984e-01
3.22722435e-01 -3.67166817e-01 -7.42206514e-01 -5.33967555e-01
-7.33199939e-02 -4.04055230e-02 7.97973096e-01 8.97010192e-02
8.75410318e-01 -2.65017778e-01 -1.48585320e-01 7.08019733e-01
1.84387290e+00 1.20161563e-01 4.39639091e-01 5.65454841e-01
6.87274277e-01 4.69508082e-01 9.67907608e-01 3.56063783e-01
3.06618690e-01 9.48757768e-01 4.07569259e-01 -3.07679445e-01
-5.21573067e-01 -1.66399345e-01 3.38819176e-01 5.98107696e-01
-3.73764366e-01 3.16149145e-01 -4.60284919e-01 6.07823074e-01
-1.81817245e+00 -1.19509971e+00 1.46760702e-01 2.36759520e+00
6.80877268e-01 -1.33719519e-01 -1.35878369e-01 2.69062314e-02
1.03639007e+00 3.39173138e-01 -6.23653233e-01 -1.01557132e-02
-6.63994029e-02 1.83336854e-01 3.75020593e-01 4.12702620e-01
-9.11043346e-01 8.53752673e-01 5.16336679e+00 8.85112822e-01
-1.22631621e+00 1.33654684e-01 1.85027763e-01 -1.48155034e-01
1.33076623e-01 -6.49847984e-02 -6.97175384e-01 3.29740554e-01
5.65607846e-01 -3.06992948e-01 6.17428720e-01 6.82437479e-01
5.38297713e-01 -2.44808510e-01 -9.30203915e-01 1.24023700e+00
2.65756309e-01 -1.39006341e+00 2.50144713e-02 -3.66934359e-01
6.14378095e-01 -7.33597353e-02 -2.90679168e-02 -2.13425636e-01
-8.73171259e-03 -7.66510010e-01 2.94780225e-01 5.23840666e-01
8.59464586e-01 -9.07431066e-01 7.06277132e-01 1.93089798e-01
-1.42602229e+00 -1.41664833e-01 -4.08741713e-01 2.15584368e-01
8.21404308e-02 3.99854571e-01 -5.66051662e-01 1.00067008e+00
7.14587390e-01 8.37248385e-01 -6.88697398e-01 1.00728834e+00
4.07300293e-02 1.77877843e-01 -1.75954551e-01 4.38373208e-01
9.34001729e-02 -2.28312656e-01 8.23299468e-01 8.46884251e-01
3.14053088e-01 1.46518767e-01 2.15792924e-01 8.46206248e-01
3.88871841e-02 2.05183387e-01 -4.12304193e-01 3.11958462e-01
6.74857736e-01 1.43125582e+00 -6.28785074e-01 -2.42426753e-01
-4.66954321e-01 1.12823904e+00 3.15747619e-01 4.75991964e-01
-5.17115593e-01 -5.51632047e-01 1.67754427e-01 2.10897982e-01
5.52148342e-01 -3.47732641e-02 6.84967712e-02 -1.31157410e+00
2.43183509e-01 -6.86567962e-01 2.95641780e-01 -9.55244303e-01
-1.03403139e+00 9.05275524e-01 -2.05493286e-01 -1.70591879e+00
-3.16221774e-01 4.24325317e-02 -4.36052918e-01 9.53015506e-01
-2.10495758e+00 -1.16545582e+00 -6.28876925e-01 6.82393372e-01
6.16898715e-01 -8.85417387e-02 4.54538912e-01 3.01763356e-01
-5.49948156e-01 5.71744263e-01 1.01301864e-01 1.80357933e-01
7.89862275e-01 -8.54071081e-01 1.64065167e-01 1.30993176e+00
-4.87062335e-02 7.46584535e-01 5.95888734e-01 -4.88054872e-01
-1.14458537e+00 -1.13736296e+00 7.45292306e-01 1.10724621e-01
4.91223127e-01 1.48909329e-03 -9.89578843e-01 2.19428286e-01
-5.15837632e-02 4.40081656e-01 1.44147277e-01 -3.08885902e-01
-4.13628221e-01 -4.47793126e-01 -1.13174808e+00 4.22987223e-01
1.23333263e+00 -4.55122501e-01 -4.93937612e-01 -5.57690486e-02
7.03629792e-01 -4.56345707e-01 -9.53311861e-01 5.01867592e-01
4.84982759e-01 -1.14870799e+00 1.20013773e+00 -7.04966486e-02
4.60375279e-01 -6.35313272e-01 -5.16532399e-02 -9.97607827e-01
-4.97270525e-01 -6.71190262e-01 -2.27770641e-01 1.25264370e+00
-1.03360079e-02 -5.53007662e-01 5.84224463e-01 4.03821707e-01
-1.25358365e-02 -7.23626018e-01 -9.16129649e-01 -8.83141696e-01
-5.08311987e-01 1.58690959e-01 5.05322456e-01 9.58187521e-01
-4.30524915e-01 2.42478505e-01 -5.50111473e-01 4.96752828e-01
1.09571302e+00 6.61947548e-01 5.36700726e-01 -1.11721027e+00
-2.81958282e-01 -2.00369284e-01 -5.16926289e-01 -9.76901650e-01
-2.53245104e-02 -8.87421310e-01 6.98203640e-03 -1.48141003e+00
2.74021238e-01 -3.46110106e-01 -6.38771713e-01 1.51833743e-01
-2.41855010e-01 4.34078187e-01 4.01540339e-01 5.29557884e-01
-7.80310154e-01 3.95426720e-01 1.59261715e+00 2.73564994e-01
-2.73820460e-01 -8.65279883e-02 -5.81786573e-01 5.40817201e-01
6.28749669e-01 -2.61894196e-01 -4.22571599e-01 -2.52065957e-01
-2.23331600e-01 3.46543044e-01 6.19056165e-01 -1.07845461e+00
4.47041303e-01 -1.52070746e-01 4.58218038e-01 -7.28372157e-01
2.74861902e-01 -8.27845395e-01 8.19196627e-02 2.37618804e-01
-4.75007504e-01 9.78134573e-02 -1.72461390e-01 6.24425650e-01
-3.41005713e-01 -2.22229674e-01 1.10967350e+00 -1.31087586e-01
-9.62664187e-01 3.42263609e-01 1.95655227e-01 -1.07393585e-01
8.40235472e-01 -5.30607760e-01 -3.57769281e-01 -3.13637674e-01
-6.19224906e-01 1.30732104e-01 5.21555126e-01 4.54741240e-01
9.49702442e-01 -1.12501550e+00 -7.11182237e-01 -2.69353017e-02
-4.15333025e-02 2.77420487e-02 5.14076233e-01 8.35957527e-01
-1.50328457e-01 3.36191356e-01 -3.88006985e-01 -4.41330552e-01
-1.30846369e+00 8.30676138e-01 3.33323210e-01 -2.76951820e-01
-1.03653121e+00 3.72989833e-01 1.81253612e-01 1.95370257e-01
2.04416528e-01 2.17656091e-01 -4.94516075e-01 -2.83878595e-01
7.85969138e-01 5.85403919e-01 -1.11627385e-01 -9.63990271e-01
-4.07861531e-01 9.79558468e-01 -3.70892078e-01 1.47047952e-01
1.31657016e+00 -6.29646540e-01 -3.34186479e-02 1.15679018e-01
9.93503273e-01 -8.74694958e-02 -1.27642703e+00 -7.41261959e-01
-2.04422802e-01 -8.47876430e-01 4.02588785e-01 -7.11432576e-01
-1.35983288e+00 6.50827169e-01 8.77026856e-01 -4.53339577e-01
1.39832902e+00 -3.05887222e-01 9.43684399e-01 2.65977472e-01
4.59033310e-01 -1.02026522e+00 5.09271249e-02 -1.42228864e-02
8.53050768e-01 -1.36988842e+00 2.11479634e-01 -5.18194497e-01
-4.43767160e-01 1.07501340e+00 8.11223805e-01 -3.74632239e-01
3.16468537e-01 -2.34268028e-02 1.15919132e-02 1.27171174e-01
-7.24443972e-01 -3.71876150e-01 2.75734812e-01 6.80259645e-01
2.01488361e-01 -5.43642700e-01 -4.68149483e-01 1.25148520e-01
1.88363306e-02 3.13617766e-01 7.32033789e-01 7.33209133e-01
-3.91560018e-01 -7.89114833e-01 -1.98116139e-01 1.69536173e-01
-5.09995222e-01 -7.97618404e-02 1.50620729e-01 6.81090891e-01
-1.82961211e-01 8.73741984e-01 5.45948148e-02 -3.18301201e-01
2.86017627e-01 -3.12703013e-01 4.42904770e-01 -3.41229290e-01
-3.94650340e-01 4.71775942e-02 -4.24679555e-02 -9.45534468e-01
-9.24426258e-01 -4.55856442e-01 -1.34299755e+00 1.21721597e-02
-4.57268238e-01 3.13839950e-02 4.26442325e-01 7.80462742e-01
7.04144120e-01 4.22256649e-01 1.05937195e+00 -8.23163927e-01
-5.81439793e-01 -6.83619201e-01 -2.40652144e-01 4.78852868e-01
5.79187810e-01 -7.70461798e-01 -3.44049513e-01 1.75332457e-01]
|
[10.94304084777832, -1.956843376159668]
|
05d79aff-1bda-418a-b5e7-9c22bfe32486
|
residue-based-label-protection-mechanisms-in
|
2205.04166
| null |
https://arxiv.org/abs/2205.04166v1
|
https://arxiv.org/pdf/2205.04166v1.pdf
|
Residue-based Label Protection Mechanisms in Vertical Logistic Regression
|
Federated learning (FL) enables distributed participants to collaboratively learn a global model without revealing their private data to each other. Recently, vertical FL, where the participants hold the same set of samples but with different features, has received increased attention. This paper first presents one label inference attack method to investigate the potential privacy leakages of the vertical logistic regression model. Specifically, we discover that the attacker can utilize the residue variables, which are calculated by solving the system of linear equations constructed by local dataset and the received decrypted gradients, to infer the privately owned labels. To deal with this, we then propose three protection mechanisms, e.g., additive noise mechanism, multiplicative noise mechanism, and hybrid mechanism which leverages local differential privacy and homomorphic encryption techniques, to prevent the attack and improve the robustness of the vertical logistic regression. model. Experimental results show that both the additive noise mechanism and the multiplicative noise mechanism can achieve efficient label protection with only a slight drop in model testing accuracy, furthermore, the hybrid mechanism can achieve label protection without any testing accuracy degradation, which demonstrates the effectiveness and efficiency of our protection techniques
|
['Ye Wu', 'Anran Li', 'Yang Liu', 'Lan Zhang', 'Juntao Tan']
|
2022-05-09
| null | null | null | null |
['inference-attack']
|
['adversarial']
|
[ 9.64499563e-02 -2.29010701e-01 -2.15983927e-01 -4.83077049e-01
-8.97231460e-01 -1.21882796e+00 1.88960657e-01 1.57276914e-01
-3.37629735e-01 7.58875966e-01 -4.64649387e-02 -3.41553003e-01
1.74674224e-02 -9.10509944e-01 -6.03953779e-01 -1.25663424e+00
8.23472813e-03 -4.34040159e-01 -1.51477098e-01 3.55803907e-01
1.70469224e-01 4.13353533e-01 -1.17946196e+00 4.12839323e-01
8.23029995e-01 1.24557292e+00 -5.99326789e-01 4.18770850e-01
6.94424659e-02 1.06970024e+00 -6.13075912e-01 -8.03476989e-01
8.34537983e-01 -3.14284265e-01 -5.50011754e-01 -5.76550603e-01
2.77962893e-01 -8.89597297e-01 -4.56458688e-01 1.36801231e+00
5.60042202e-01 -2.17990920e-01 3.12692374e-01 -1.64502144e+00
-5.69436729e-01 6.87704563e-01 -6.32868290e-01 -2.06920922e-01
1.74358353e-01 2.81641304e-01 7.01401114e-01 -5.78397930e-01
5.28521121e-01 1.06768358e+00 8.19969118e-01 5.91435373e-01
-1.24657762e+00 -1.37922704e+00 9.42222495e-03 3.99359576e-02
-1.72426176e+00 -2.88264275e-01 6.07542038e-01 -2.10401937e-01
2.02583179e-01 6.92616999e-01 7.10352510e-02 9.50187504e-01
3.06080818e-01 5.57757974e-01 1.69474983e+00 -6.38095364e-02
2.36113697e-01 7.07255304e-01 2.34930739e-01 6.66074574e-01
5.80752730e-01 2.96406031e-01 -7.53005564e-01 -9.70771432e-01
2.94253588e-01 3.75260085e-01 -6.62027121e-01 -5.04845917e-01
-9.06913340e-01 7.19324529e-01 3.40349108e-01 -2.19494388e-01
3.65248509e-02 8.07035193e-02 6.16147459e-01 7.06771493e-01
2.97891259e-01 -1.94030508e-01 -7.88860440e-01 4.06483650e-01
-3.39201033e-01 1.73939019e-01 9.57205296e-01 7.93771505e-01
9.39091623e-01 -3.68619263e-01 -1.54503301e-01 1.20409369e-01
5.52286208e-01 6.27167106e-01 4.23419654e-01 -6.39363945e-01
7.48647571e-01 5.37646472e-01 7.63326958e-02 -1.22029305e+00
1.85938567e-01 -1.44691095e-01 -8.34258556e-01 1.23825222e-01
4.15204078e-01 -7.40597248e-01 -1.62910461e-01 2.15983796e+00
7.72171915e-01 2.85882890e-01 5.66652417e-01 5.96615672e-01
3.28339696e-01 3.50846708e-01 4.19409983e-02 -1.74684212e-01
1.25800383e+00 -7.91859746e-01 -6.87598825e-01 4.16661263e-01
1.02993989e+00 -3.88742328e-01 5.57501614e-01 1.31568536e-01
-7.73302674e-01 -7.85091519e-03 -1.38010764e+00 -9.97024626e-02
-5.31179130e-01 -1.87998429e-01 6.58487201e-01 1.37108886e+00
-6.98579609e-01 5.45075715e-01 -7.68834293e-01 3.08632135e-01
7.21413851e-01 7.50384331e-01 -7.40902007e-01 -8.77941623e-02
-1.35003603e+00 6.93609789e-02 6.16948642e-02 -6.77332431e-02
-8.88337612e-01 -1.05811501e+00 -7.42473602e-01 2.03273654e-01
1.66911274e-01 -4.77275968e-01 9.16863978e-01 -6.47056282e-01
-1.23246896e+00 6.13168657e-01 1.20838247e-01 -5.45772374e-01
8.79859149e-01 2.12464049e-01 -4.01196212e-01 5.36010750e-02
-1.93725795e-01 8.78159255e-02 6.56385601e-01 -1.21698725e+00
-7.49613881e-01 -9.05899942e-01 -1.33379549e-01 -1.91627257e-02
-7.74738491e-01 -4.55287695e-02 2.02821672e-01 -5.03832817e-01
-1.38544803e-02 -8.52167606e-01 -5.55869453e-02 3.43691438e-01
-4.60077703e-01 2.12250412e-01 1.22580695e+00 -6.87663019e-01
1.20414078e+00 -2.55028009e+00 -4.94377106e-01 7.37251878e-01
3.46741796e-01 -1.64652821e-02 1.75324053e-01 4.33730394e-01
1.96722999e-01 4.95312870e-01 -2.13555261e-01 -4.27892864e-01
1.41590074e-01 -8.91329646e-02 -7.45251358e-01 8.35606098e-01
-4.18583810e-01 8.47857356e-01 -6.37697577e-01 -3.14257115e-01
-3.58827412e-01 6.26019239e-01 -2.94759214e-01 8.69654343e-02
2.27158532e-01 4.42387372e-01 -7.33720362e-01 5.65249205e-01
1.40926659e+00 -2.15777159e-01 7.05496311e-01 -1.61364347e-01
1.45873725e-01 -6.45429222e-03 -1.32518709e+00 1.13705742e+00
-1.41212821e-01 3.76390480e-02 3.95817101e-01 -2.97172993e-01
8.57333958e-01 4.82144475e-01 2.59787768e-01 -4.94286150e-01
2.26453081e-01 2.90482432e-01 -4.54319119e-01 -2.47291192e-01
-7.70153403e-02 8.49033147e-02 -2.55293965e-01 1.03876603e+00
-3.70931149e-01 7.50177383e-01 -8.95422518e-01 7.70122632e-02
9.98530030e-01 -2.00117096e-01 -2.51654163e-02 -2.19595626e-01
9.14332211e-01 -4.38511014e-01 1.02088249e+00 8.73025119e-01
-4.55829144e-01 7.50389248e-02 5.82346797e-01 -4.25269097e-01
-4.79857802e-01 -6.98690712e-01 4.20329906e-02 9.76265669e-01
5.78555048e-01 -5.40462375e-01 -7.92262673e-01 -1.29917979e+00
5.29816926e-01 8.47604647e-02 -6.41233981e-01 -3.97230506e-01
-3.38339388e-01 -5.04023790e-01 1.13035667e+00 2.94341564e-01
1.22932816e+00 -3.99969786e-01 -3.03699523e-01 -3.50395232e-01
1.72447395e-02 -6.37347460e-01 -7.27008462e-01 4.06428846e-03
-6.76732719e-01 -1.18599629e+00 -1.37922704e-01 -7.16079891e-01
8.10694754e-01 4.60210174e-01 2.38195375e-01 2.80681908e-01
-9.37841237e-02 3.04252684e-01 1.42597079e-01 -2.92804837e-01
-2.20854133e-01 -1.05970815e-01 -2.88510974e-02 7.20952094e-01
4.93471920e-01 -5.55505276e-01 -9.30532336e-01 3.20718288e-01
-1.01116371e+00 -3.34591836e-01 1.75266296e-01 7.36160874e-01
5.47240794e-01 2.05513164e-01 4.70111340e-01 -1.36643863e+00
6.41855359e-01 -6.11635029e-01 -6.03488207e-01 5.59408724e-01
-8.80471230e-01 1.15322128e-01 9.08505499e-01 -5.23672640e-01
-1.13029623e+00 1.20319128e-01 1.94277152e-01 -3.87179881e-01
8.39848816e-02 1.50027886e-01 -6.59594715e-01 -7.21743584e-01
4.11361635e-01 3.90747219e-01 1.00008160e-01 -5.26287973e-01
3.49690229e-01 9.36746955e-01 3.05014968e-01 -5.63769221e-01
9.06059504e-01 8.23191226e-01 2.44707778e-01 1.26217946e-01
-4.25511092e-01 -4.86326870e-03 -9.64739993e-02 1.96331650e-01
6.73219442e-01 -1.06231761e+00 -1.38937390e+00 1.01151109e+00
-7.43157864e-01 -1.47644682e-02 -7.24252239e-02 2.38551974e-01
-1.29920751e-01 7.00745523e-01 -9.45289135e-01 -7.20676661e-01
-7.43210852e-01 -1.05938900e+00 6.62846386e-01 3.22049558e-01
4.30553913e-01 -9.15305376e-01 -1.52214497e-01 3.97759348e-01
4.85498101e-01 5.92056096e-01 9.48615193e-01 -7.87173927e-01
-7.65556276e-01 -4.65842187e-01 -1.44789204e-01 3.47992390e-01
2.09697366e-01 -5.66249609e-01 -1.22195733e+00 -8.37613523e-01
4.88562435e-01 -3.30351025e-01 4.39768106e-01 -5.30330300e-01
1.44064617e+00 -7.97070086e-01 -2.95430124e-01 9.64611292e-01
1.59589076e+00 2.20446400e-02 4.09613073e-01 7.00330883e-02
6.17695034e-01 5.36456525e-01 3.71985286e-01 6.86344206e-01
4.51068550e-01 1.86902553e-01 2.82341838e-01 7.12233856e-02
4.12589192e-01 -5.67181230e-01 4.39254940e-01 6.02496862e-01
6.17114067e-01 6.81392029e-02 -3.73178720e-01 -3.03770564e-02
-1.86429763e+00 -7.01726854e-01 2.51545589e-02 2.56357503e+00
8.79346550e-01 -4.65098977e-01 -3.70671690e-01 -5.09150065e-02
7.78370440e-01 5.54502457e-02 -1.00273836e+00 -3.26485991e-01
-3.29435050e-01 2.37157699e-02 1.08326757e+00 2.63536721e-01
-9.33955252e-01 6.11698508e-01 5.60849285e+00 7.60839641e-01
-1.23142648e+00 4.37865525e-01 8.92036140e-01 7.69267902e-02
-4.78958666e-01 2.54055977e-01 -7.17403054e-01 6.83479965e-01
8.56060386e-01 -4.47781533e-01 4.10890698e-01 8.36416185e-01
-3.09353679e-01 3.89069170e-01 -9.96923506e-01 9.74043548e-01
-3.23909745e-02 -1.25905132e+00 -1.31291270e-01 4.98971403e-01
7.93141305e-01 -3.99275720e-01 4.56607968e-01 1.58148646e-01
5.91995716e-01 -7.43184388e-01 4.28219646e-01 4.99948680e-01
7.99406052e-01 -1.15939629e+00 6.24260843e-01 4.73592222e-01
-1.08687127e+00 -4.90391403e-01 -4.09603566e-01 1.82279512e-01
-4.20394629e-01 1.28468812e-01 -2.40528688e-01 8.19609940e-01
6.52315915e-01 1.67342246e-01 -5.53591073e-01 5.06306529e-01
-2.71277815e-01 5.15787959e-01 -3.07247460e-01 1.11479737e-01
-5.44811301e-02 -3.00676972e-01 1.97614327e-01 7.39030421e-01
1.44113660e-01 4.20138128e-02 1.88825279e-01 7.45244384e-01
-5.94158411e-01 3.78511369e-01 -6.24908984e-01 3.65396917e-01
9.24738169e-01 1.18599164e+00 -1.12457022e-01 -1.03452988e-01
-3.66865188e-01 1.02799547e+00 3.74635011e-01 4.17422324e-01
-6.39312625e-01 -5.55917799e-01 1.00182605e+00 -4.59382772e-01
2.19355509e-01 -4.42336872e-02 -5.14626443e-01 -1.32213211e+00
1.83913350e-01 -1.05781615e+00 7.28747666e-01 -8.80456269e-02
-1.47067976e+00 1.99339896e-01 -5.21373212e-01 -1.00676548e+00
1.93053901e-01 6.99535431e-03 -4.57741350e-01 1.15361953e+00
-1.42160785e+00 -1.23533475e+00 -1.48452088e-01 1.24503005e+00
-4.58978057e-01 -1.91775113e-01 9.57343280e-01 1.63434237e-01
-6.29189134e-01 1.43130958e+00 6.16013825e-01 3.17821980e-01
8.44902217e-01 -8.40434790e-01 3.42519954e-03 8.32893014e-01
-1.76453173e-01 1.04130471e+00 4.01501097e-02 -6.79591000e-01
-1.95297527e+00 -1.26960790e+00 7.45239794e-01 -2.17992425e-01
3.42084229e-01 -8.22576821e-01 -9.36942339e-01 7.98971713e-01
-1.02497868e-01 3.96707833e-01 1.33902442e+00 -4.22535509e-01
-1.07262349e+00 -4.39775497e-01 -2.11973143e+00 5.53555667e-01
6.21911108e-01 -9.89255309e-01 2.47655064e-01 2.95125157e-01
1.02567995e+00 -2.49352068e-01 -1.10022044e+00 1.30063519e-01
7.66943455e-01 -8.95371377e-01 8.26767564e-01 -6.95820332e-01
-1.86599359e-01 -4.08058763e-01 -3.88971806e-01 -5.16494453e-01
-1.26134772e-02 -8.07902217e-01 -4.25018787e-01 1.66137183e+00
1.90380961e-01 -1.43327701e+00 1.03421307e+00 1.18839943e+00
8.62235963e-01 -5.35608292e-01 -1.08148003e+00 -5.31394124e-01
2.21504211e-01 1.03768548e-02 1.46172559e+00 1.27462912e+00
-1.45377278e-01 -4.39928949e-01 -5.91643751e-01 5.92034638e-01
9.19963658e-01 2.08164304e-01 9.45625186e-01 -9.37350392e-01
-2.69956350e-01 1.75548062e-01 -3.33993733e-01 -7.60530233e-01
4.06752408e-01 -1.04329729e+00 -5.10852039e-01 -5.19405723e-01
2.54566252e-01 -7.12157249e-01 -8.15108657e-01 8.13965499e-01
-1.51411355e-01 2.85540581e-01 2.59834915e-01 3.98310751e-01
-3.20153683e-01 3.03449929e-01 8.07430387e-01 -1.82757989e-01
-1.86636135e-01 3.14417295e-02 -1.25773442e+00 6.68832436e-02
8.30386639e-01 -7.72506773e-01 -5.77674568e-01 -2.46287107e-01
2.36940995e-01 -8.59780684e-02 2.89425880e-01 -7.04923391e-01
6.74129784e-01 -6.39232397e-02 4.17773783e-01 -1.88163862e-01
-1.93907008e-01 -1.20468712e+00 5.02125144e-01 6.80932939e-01
-4.91921097e-01 -3.34984884e-02 -1.42628208e-01 8.95191193e-01
-1.60351619e-02 7.60482103e-02 6.24207377e-01 1.49694502e-01
-1.49424881e-01 5.64199686e-01 2.86526494e-02 -2.92406261e-01
1.35952508e+00 -1.56690255e-01 -6.38433456e-01 -1.46034077e-01
-4.79320377e-01 5.53156376e-01 7.63129413e-01 1.21665314e-01
3.18522245e-01 -1.35104990e+00 -6.12092078e-01 6.54235005e-01
-4.54232022e-02 -1.89823031e-01 6.01085246e-01 7.63010502e-01
-3.28524381e-01 5.50754517e-02 -1.27308676e-02 -9.47297364e-02
-1.64023077e+00 8.05864930e-01 4.94016469e-01 -3.37543666e-01
-5.69022298e-01 7.30179429e-01 2.57104158e-01 -6.72373652e-01
4.25690264e-01 1.96716651e-01 2.62187064e-01 -1.49416476e-01
8.55952859e-01 5.81677020e-01 -3.26088257e-02 -5.08135259e-01
-3.93252820e-01 3.78737807e-01 -3.48981977e-01 1.26035050e-01
8.90242517e-01 -4.40856993e-01 -4.22684401e-01 -1.71949133e-01
1.67545199e+00 4.33638304e-01 -1.27512085e+00 -4.20080781e-01
-2.63515115e-01 -8.16224039e-01 -1.53246745e-02 -7.70458519e-01
-1.35524118e+00 5.83809078e-01 9.09851193e-01 2.01487318e-02
1.17363691e+00 -6.63379431e-01 1.06646955e+00 1.04136296e-01
7.03095675e-01 -6.67436481e-01 -5.61049223e-01 -1.27367735e-01
1.14017196e-01 -9.38788712e-01 -4.37391326e-02 -5.82464516e-01
-5.77788413e-01 8.86564374e-01 4.52949882e-01 1.22591853e-01
7.63127267e-01 3.25524241e-01 2.87541330e-01 2.42803618e-01
-7.88426399e-01 7.09525764e-01 -2.82060534e-01 6.71004117e-01
-3.55661273e-01 8.80256519e-02 -2.91927993e-01 1.02122867e+00
8.87423158e-02 -1.54675052e-01 3.29481900e-01 1.20467484e+00
1.58307537e-01 -1.52759540e+00 -4.79532868e-01 2.97512442e-01
-7.94479549e-01 1.28810912e-01 -5.89727342e-01 3.09380352e-01
2.52418250e-01 9.57342744e-01 -4.10697907e-01 -5.73346734e-01
-1.63210183e-02 2.92200714e-01 -3.71175930e-02 -7.33666169e-03
-1.21514142e+00 -3.87919158e-01 -4.27648365e-01 -5.32424629e-01
8.56141597e-02 -4.78313893e-01 -1.19128418e+00 -7.24741518e-01
-4.97574270e-01 2.85661668e-01 7.06242383e-01 5.07262349e-01
8.13404858e-01 -2.11221293e-01 1.45378506e+00 1.01115527e-02
-1.21498764e+00 -2.18588069e-01 -9.36153412e-01 4.78754699e-01
4.68923628e-01 -3.60969407e-03 -6.93208933e-01 -3.30271602e-01]
|
[5.837385177612305, 6.760833263397217]
|
6401b7a3-1ee9-4b5c-90c6-7ccf932575b2
|
rstt-real-time-spatial-temporal-transformer
|
2203.14186
| null |
https://arxiv.org/abs/2203.14186v1
|
https://arxiv.org/pdf/2203.14186v1.pdf
|
RSTT: Real-time Spatial Temporal Transformer for Space-Time Video Super-Resolution
|
Space-time video super-resolution (STVSR) is the task of interpolating videos with both Low Frame Rate (LFR) and Low Resolution (LR) to produce High-Frame-Rate (HFR) and also High-Resolution (HR) counterparts. The existing methods based on Convolutional Neural Network~(CNN) succeed in achieving visually satisfied results while suffer from slow inference speed due to their heavy architectures. We propose to resolve this issue by using a spatial-temporal transformer that naturally incorporates the spatial and temporal super resolution modules into a single model. Unlike CNN-based methods, we do not explicitly use separated building blocks for temporal interpolations and spatial super-resolutions; instead, we only use a single end-to-end transformer architecture. Specifically, a reusable dictionary is built by encoders based on the input LFR and LR frames, which is then utilized in the decoder part to synthesize the HFR and HR frames. Compared with the state-of-the-art TMNet \cite{xu2021temporal}, our network is $60\%$ smaller (4.5M vs 12.3M parameters) and $80\%$ faster (26.2fps vs 14.3fps on $720\times576$ frames) without sacrificing much performance. The source code is available at https://github.com/llmpass/RSTT.
|
['Ilya Zharkov', 'Tianyu Ding', 'Luming Liang', 'Zhicheng Geng']
|
2022-03-27
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Geng_RSTT_Real-Time_Spatial_Temporal_Transformer_for_Space-Time_Video_Super-Resolution_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Geng_RSTT_Real-Time_Spatial_Temporal_Transformer_for_Space-Time_Video_Super-Resolution_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['space-time-video-super-resolution', 'video-super-resolution']
|
['computer-vision', 'computer-vision']
|
[ 1.21934824e-01 -1.16945788e-01 -8.92494544e-02 -2.64934868e-01
-8.78366888e-01 -3.97491530e-02 3.01618218e-01 -5.70614457e-01
-3.86020452e-01 9.36322987e-01 1.68985110e-02 -2.97482222e-01
2.09690362e-01 -8.41972589e-01 -8.80388677e-01 -4.03637409e-01
4.53130119e-02 -2.45247573e-01 5.10865331e-01 -2.47693852e-01
-1.33889422e-01 2.13175133e-01 -1.62911034e+00 5.80381334e-01
9.30215299e-01 1.13020742e+00 5.16149223e-01 7.00113714e-01
6.55214861e-02 1.10130537e+00 -3.49558145e-01 -2.64464289e-01
2.80947179e-01 -5.49492955e-01 -6.61973536e-01 -1.52412161e-01
5.39548159e-01 -8.00594091e-01 -6.83940947e-01 9.86490548e-01
4.51453626e-01 1.25633508e-01 6.49831370e-02 -7.85301208e-01
-7.51600385e-01 5.18644869e-01 -9.28389251e-01 5.15590608e-01
3.00836504e-01 2.45247722e-01 6.42552495e-01 -1.01742244e+00
5.39899409e-01 1.18744767e+00 6.42199278e-01 6.46999359e-01
-1.13859689e+00 -1.05631685e+00 -5.06943911e-02 2.35056266e-01
-1.57264841e+00 -6.88347816e-01 4.81718928e-01 -1.24200195e-01
8.58393312e-01 7.73083493e-02 4.71955925e-01 1.02351737e+00
1.27623186e-01 3.50995958e-01 1.07872868e+00 -9.53118503e-02
-1.57947466e-01 -3.22526485e-01 -2.89997220e-01 6.95457876e-01
-1.23058617e-01 2.81581521e-01 -7.13654935e-01 4.19663221e-01
1.65069818e+00 4.74952022e-03 -4.52053547e-01 4.66512918e-01
-1.21324408e+00 5.01662791e-01 6.81389809e-01 2.80257046e-01
-3.50076526e-01 4.42218155e-01 2.18409568e-01 1.67760089e-01
4.92485791e-01 -9.21353474e-02 -2.10977614e-01 -7.75786191e-02
-1.29972243e+00 1.82774022e-01 1.13578744e-01 1.23825693e+00
8.15012634e-01 4.99169141e-01 -1.13055371e-01 8.86561275e-01
1.57533392e-01 3.08176905e-01 3.18695188e-01 -1.47086847e+00
5.12181342e-01 1.70292795e-01 3.42022896e-01 -7.93779731e-01
-8.87164325e-02 -4.55620438e-01 -1.21919000e+00 2.26894408e-01
3.12241346e-01 -1.17313519e-01 -9.68658686e-01 1.63250256e+00
9.39523950e-02 5.66986740e-01 -8.01800266e-02 1.21259439e+00
8.90466452e-01 9.48689699e-01 1.05510175e-01 -2.94305921e-01
1.36055434e+00 -1.07443857e+00 -7.35940814e-01 -1.00294597e-01
1.20119564e-01 -8.50843847e-01 9.58242774e-01 2.23907739e-01
-1.68876994e+00 -1.05508947e+00 -1.10257709e+00 -5.73700905e-01
1.30131438e-01 1.17810801e-01 2.75696456e-01 1.66655891e-02
-1.38894868e+00 7.93896735e-01 -7.84587979e-01 6.98413923e-02
4.60126489e-01 2.29025275e-01 -6.17407970e-02 -1.32459909e-01
-1.34620583e+00 6.88621223e-01 2.21803874e-01 1.27520338e-01
-8.78422022e-01 -8.81067872e-01 -8.08322549e-01 -8.63647237e-02
1.76061839e-01 -7.34080195e-01 1.20001733e+00 -1.03259134e+00
-1.59769440e+00 5.19175708e-01 -4.35375214e-01 -6.33199513e-01
5.76270640e-01 -3.39517474e-01 -5.69512486e-01 3.01922590e-01
1.09989997e-02 8.61158073e-01 8.48395884e-01 -1.10475802e+00
-1.01607299e+00 3.31385732e-02 1.43317074e-01 1.19505450e-01
2.15593129e-02 2.86927819e-01 -6.23000681e-01 -8.32223773e-01
3.67543995e-02 -5.37060559e-01 -2.25618124e-01 1.01001285e-01
-4.54924032e-02 2.83679217e-01 8.52555752e-01 -9.47205007e-01
1.35408878e+00 -2.14509940e+00 1.07749104e-01 -4.40000087e-01
3.47496241e-01 5.30290067e-01 -7.97277614e-02 -9.89611819e-02
-2.33682822e-02 7.06650643e-03 -9.38953385e-02 -4.57340926e-01
-3.67353976e-01 -6.62978068e-02 -2.88467288e-01 3.83735329e-01
1.33426532e-01 8.43791485e-01 -9.16026413e-01 -5.66241384e-01
5.91124117e-01 1.15159798e+00 -3.90414000e-01 1.85077026e-01
1.19816149e-02 7.56231546e-01 -1.52725056e-01 5.71672440e-01
7.38962352e-01 -3.59665334e-01 7.77683482e-02 -5.72861016e-01
-5.72884679e-01 3.02898347e-01 -1.16432059e+00 1.90224552e+00
-7.30558097e-01 7.38257468e-01 1.32904172e-01 -5.76981902e-01
9.11872566e-01 4.87618536e-01 4.73250180e-01 -1.07610202e+00
1.65525228e-01 4.10328954e-01 -4.18333590e-01 -2.25350887e-01
6.26794755e-01 3.29115503e-02 2.06130147e-01 -9.26199332e-02
9.50943381e-02 3.53998780e-01 1.55607909e-01 -7.29411915e-02
9.31988776e-01 5.36922932e-01 1.39508937e-02 -7.33468607e-02
5.67614436e-01 -2.85861224e-01 7.70000815e-01 1.79146111e-01
-6.14296645e-02 8.86492372e-01 4.69487011e-02 -5.67072213e-01
-1.50098848e+00 -1.25999141e+00 -1.20177723e-01 9.06953573e-01
3.80154401e-01 -4.47776437e-01 -5.87799728e-01 3.79855512e-04
-4.73864198e-01 5.01215816e-01 -2.65961230e-01 1.86155781e-01
-9.42716181e-01 -4.34192121e-01 4.54183877e-01 6.90969825e-01
1.03679001e+00 -9.04106379e-01 -8.13846946e-01 3.71098399e-01
-4.94740427e-01 -1.45665669e+00 -6.21569276e-01 -2.16671646e-01
-9.55773354e-01 -6.66737795e-01 -8.88166249e-01 -6.59078836e-01
3.64973783e-01 4.47489500e-01 1.24002182e+00 -7.67446011e-02
-1.91449881e-01 -3.12033832e-01 -3.95626724e-01 1.71698302e-01
-2.29692325e-01 -2.26476640e-01 -1.11193277e-01 -2.33360976e-02
2.65243440e-03 -8.81634533e-01 -1.09647822e+00 3.21601391e-01
-8.73114467e-01 7.91943789e-01 6.51053071e-01 7.41829157e-01
8.73134732e-01 5.86299226e-02 4.23114032e-01 -5.55588484e-01
-1.67987142e-02 -2.71376312e-01 -7.03546703e-01 -9.31326747e-02
-3.03744406e-01 -2.28690267e-01 8.62975299e-01 -3.73287767e-01
-1.22271264e+00 4.39610891e-03 -2.31836215e-01 -8.87519002e-01
-3.31511870e-02 7.68356351e-03 1.20775141e-01 5.24268746e-02
4.87229437e-01 3.23368609e-01 -1.56324655e-01 -6.09952927e-01
3.35870028e-01 5.66209257e-01 8.62271369e-01 -2.95838356e-01
8.25618446e-01 6.56046212e-01 -2.40223303e-01 -6.96048975e-01
-6.97469175e-01 -1.14937298e-01 -4.43223715e-01 -3.42927128e-01
1.08108902e+00 -1.48610210e+00 -5.49110532e-01 3.55845511e-01
-1.04550552e+00 -5.74914098e-01 -2.45942831e-01 5.24230599e-01
-6.38819873e-01 1.57076329e-01 -1.14419973e+00 -5.75685501e-01
-5.57045281e-01 -1.12901616e+00 1.09265852e+00 4.47535485e-01
7.57714361e-02 -4.21361059e-01 -4.90216196e-01 4.47060585e-01
8.25387537e-01 3.34722251e-01 3.75184089e-01 4.73344415e-01
-9.67947841e-01 2.94909298e-01 -7.25370228e-01 2.81584918e-01
2.52681822e-02 -3.92116606e-02 -9.47210848e-01 -4.03307289e-01
-5.62866628e-02 -3.04051824e-02 8.71532679e-01 5.71190298e-01
1.26738858e+00 -3.01308304e-01 3.05300225e-02 1.10137725e+00
1.78218269e+00 3.94160241e-01 1.13697004e+00 1.35387212e-01
8.45257521e-01 8.79225880e-02 6.33153439e-01 5.22175729e-01
4.75377649e-01 8.69232833e-01 3.96257430e-01 -3.93604845e-01
-6.28099978e-01 -2.61226982e-01 5.29634476e-01 5.49147427e-01
-6.43114507e-01 -2.56374087e-02 -5.96652389e-01 5.06500721e-01
-1.81487036e+00 -1.12354898e+00 -2.34823912e-01 2.17419887e+00
1.05356383e+00 5.68913631e-02 1.17779471e-01 -6.23114291e-04
8.20709229e-01 3.17349374e-01 -4.29648519e-01 -1.89544305e-01
-9.81919318e-02 5.29512882e-01 5.80009282e-01 6.82954431e-01
-9.63898778e-01 9.28545415e-01 4.74775028e+00 9.79108632e-01
-1.25973010e+00 2.73832917e-01 8.66228402e-01 -3.58345300e-01
-1.18160471e-01 -1.83527380e-01 -8.51783156e-01 5.60605705e-01
1.21016467e+00 3.71464044e-02 7.28491247e-01 4.62302595e-01
5.89138210e-01 -2.92205606e-02 -8.28994215e-01 1.18189108e+00
-3.55814755e-01 -1.64878333e+00 -1.86731559e-04 -1.37807310e-01
5.73646963e-01 -5.67649491e-02 1.51909649e-01 1.57373950e-01
1.79906666e-01 -1.21353567e+00 1.00443137e+00 4.86581147e-01
1.61644447e+00 -7.93920338e-01 4.07467276e-01 1.00893155e-01
-1.74886954e+00 -4.05494682e-02 -4.13852900e-01 6.32713037e-03
3.77572626e-01 5.16687393e-01 -1.45858169e-01 6.43022954e-01
1.22385538e+00 8.59574914e-01 -2.24127203e-01 5.54343164e-01
-9.15576443e-02 2.13359088e-01 -1.45383373e-01 7.14943886e-01
-1.30102793e-02 2.45923959e-02 3.56834739e-01 1.20590651e+00
5.12545347e-01 3.66326749e-01 -7.59254992e-02 1.00150025e+00
-1.16108127e-01 -3.24573159e-01 -1.98699892e-01 4.24757183e-01
6.35866165e-01 1.15465510e+00 -4.03000683e-01 -3.63093555e-01
-7.15645492e-01 1.16990304e+00 1.09567419e-01 4.29015547e-01
-1.44155324e+00 -3.48917961e-01 6.91635072e-01 4.02900338e-01
4.74451780e-01 -3.48838598e-01 -2.03170359e-01 -1.22792172e+00
9.30167064e-02 -7.96008229e-01 1.48114994e-01 -9.64443028e-01
-9.07245994e-01 9.50190723e-01 -1.79619744e-01 -1.55558765e+00
-1.50498167e-01 -9.42266881e-02 -1.94996580e-01 1.18415821e+00
-1.90026128e+00 -1.18493342e+00 -6.48974061e-01 7.87349284e-01
9.05650914e-01 2.63243407e-01 5.39799929e-01 7.84990251e-01
-4.87078846e-01 5.60137391e-01 -1.60410807e-01 1.57734394e-01
6.96150780e-01 -9.54167426e-01 5.07523060e-01 1.07487953e+00
-3.75495434e-01 3.84483516e-01 6.15146756e-01 -3.86052340e-01
-1.21470976e+00 -1.51053691e+00 8.28641653e-01 -6.47915751e-02
3.61000687e-01 -1.08088747e-01 -9.76192951e-01 5.74969351e-01
2.45112121e-01 4.22363132e-01 1.34162560e-01 -5.69263816e-01
-4.24791574e-01 -3.78198057e-01 -1.11201787e+00 6.51035964e-01
1.16013873e+00 -4.59562004e-01 -6.89309686e-02 -3.43670160e-01
1.09619725e+00 -7.02926219e-01 -1.18253279e+00 5.21539688e-01
5.62482834e-01 -1.33620632e+00 1.26102424e+00 7.22956210e-02
9.16469753e-01 -7.12519646e-01 -2.95142651e-01 -6.41222835e-01
-5.89757919e-01 -5.60119629e-01 -4.08396095e-01 1.02536106e+00
2.40196913e-01 -2.98259228e-01 5.09915590e-01 3.62682968e-01
-3.45068008e-01 -9.15325761e-01 -9.41985548e-01 -6.06829822e-01
-1.29689649e-01 -2.74204463e-01 5.09864151e-01 8.30738425e-01
-3.35158676e-01 3.20594162e-01 -6.65028870e-01 3.12634408e-01
8.63581359e-01 1.12825043e-01 4.24709350e-01 -6.73302472e-01
-2.37745106e-01 -2.65108168e-01 3.07531888e-03 -1.19214511e+00
-2.98498183e-01 -2.88985670e-01 3.33833955e-02 -1.42892575e+00
3.17568369e-02 -4.34805512e-01 -3.32217693e-01 3.28245133e-01
2.19998951e-03 6.94547832e-01 3.49272728e-01 2.41090074e-01
-5.36733806e-01 3.29174995e-01 1.36912143e+00 3.04415971e-01
-3.20074677e-01 -3.85001928e-01 -4.93907452e-01 7.10287452e-01
8.92133713e-01 9.47099999e-02 -3.63124490e-01 -8.67590427e-01
-1.43487796e-01 7.46188104e-01 6.71738446e-01 -1.25111949e+00
1.32813573e-01 -1.22885875e-01 6.62444651e-01 -5.98574579e-01
5.96774697e-01 -5.99283516e-01 5.76404631e-01 3.15589577e-01
-2.54859298e-01 1.55264810e-01 1.74645409e-01 3.51193756e-01
-2.40966409e-01 4.12551194e-01 1.31989324e+00 -2.73581982e-01
-9.16495979e-01 6.42668843e-01 -1.19986393e-01 -1.34341806e-01
8.75615180e-01 -3.04542512e-01 -4.61550772e-01 -2.53383428e-01
-6.80499017e-01 7.44750723e-03 6.18596554e-01 3.99627656e-01
9.29218650e-01 -1.30379689e+00 -9.48149145e-01 4.33680527e-02
-4.05203104e-01 3.90609592e-01 7.86520660e-01 7.61554897e-01
-7.16828644e-01 2.88083673e-01 -4.38270301e-01 -5.00266194e-01
-1.00946128e+00 4.87134635e-01 4.75930721e-01 -1.00493111e-01
-9.48562086e-01 8.40943038e-01 2.33587846e-01 4.34382647e-01
1.31782457e-01 -2.50950694e-01 -1.33753762e-01 -2.89468288e-01
8.01565588e-01 5.13890088e-01 -2.64827847e-01 -7.83438265e-01
-1.68736205e-01 5.91160178e-01 4.16031741e-02 -1.20929271e-01
1.36929655e+00 -3.61376166e-01 1.13317177e-01 1.01228073e-01
1.16233099e+00 -2.59306163e-01 -1.69778395e+00 -4.12573487e-01
-4.64194000e-01 -7.14207768e-01 1.10403389e-01 -4.79721606e-01
-1.37878001e+00 8.40656519e-01 6.40307307e-01 -1.11259170e-01
1.57780921e+00 -1.34146601e-01 1.24836600e+00 -5.13578355e-01
6.13936841e-01 -8.81220460e-01 -1.23820482e-02 4.63072360e-01
8.10817122e-01 -1.05313337e+00 -8.64756107e-03 -5.14462471e-01
-5.08258939e-01 1.04019201e+00 8.15743148e-01 -3.29351097e-01
3.61250341e-01 4.36490357e-01 -1.65843278e-01 1.65242866e-01
-8.47553372e-01 -2.23222390e-01 7.67768845e-02 4.36649561e-01
7.57497430e-01 -7.86969811e-02 -2.76535392e-01 4.43240762e-01
-1.26149431e-01 3.61217439e-01 4.83684808e-01 6.35432899e-01
-3.64477336e-01 -8.32273304e-01 -1.26727328e-01 1.79741934e-01
-6.49247766e-01 -2.70290732e-01 5.99676907e-01 5.30159652e-01
4.57643956e-01 8.88970077e-01 2.05815554e-01 -6.22592986e-01
1.99547067e-01 -3.61980855e-01 3.91963422e-01 -3.09806168e-01
-3.85767579e-01 3.08263332e-01 -1.38761569e-02 -1.01832259e+00
-6.11234963e-01 -3.60333532e-01 -1.33058453e+00 -7.86796868e-01
1.18493013e-01 -2.72083908e-01 2.07462236e-01 4.15410340e-01
4.49491054e-01 8.78646433e-01 5.82497597e-01 -1.08551884e+00
2.90176086e-02 -7.39038348e-01 -4.11828607e-01 1.76623002e-01
5.35164714e-01 -2.75926828e-01 -7.63772205e-02 5.22861421e-01]
|
[11.015734672546387, -1.8536139726638794]
|
ed4d9d9f-1cec-4fc3-9dc8-82bbc5f5bd87
|
selecting-robust-features-for-machine
|
2304.05294
| null |
https://arxiv.org/abs/2304.05294v5
|
https://arxiv.org/pdf/2304.05294v5.pdf
|
Selecting Robust Features for Machine Learning Applications using Multidata Causal Discovery
|
Robust feature selection is vital for creating reliable and interpretable Machine Learning (ML) models. When designing statistical prediction models in cases where domain knowledge is limited and underlying interactions are unknown, choosing the optimal set of features is often difficult. To mitigate this issue, we introduce a Multidata (M) causal feature selection approach that simultaneously processes an ensemble of time series datasets and produces a single set of causal drivers. This approach uses the causal discovery algorithms PC1 or PCMCI that are implemented in the Tigramite Python package. These algorithms utilize conditional independence tests to infer parts of the causal graph. Our causal feature selection approach filters out causally-spurious links before passing the remaining causal features as inputs to ML models (Multiple linear regression, Random Forest) that predict the targets. We apply our framework to the statistical intensity prediction of Western Pacific Tropical Cyclones (TC), for which it is often difficult to accurately choose drivers and their dimensionality reduction (time lags, vertical levels, and area-averaging). Using more stringent significance thresholds in the conditional independence tests helps eliminate spurious causal relationships, thus helping the ML model generalize better to unseen TC cases. M-PC1 with a reduced number of features outperforms M-PCMCI, non-causal ML, and other feature selection methods (lagged correlation, random), even slightly outperforming feature selection based on eXplainable Artificial Intelligence. The optimal causal drivers obtained from our causal feature selection help improve our understanding of underlying relationships and suggest new potential drivers of TC intensification.
|
['Andreas Gerhardus', 'Jakob Runge', 'Milton S. Gomez', 'Frederick Iat-Hin Tam', 'Tom Beucler', 'Saranya Ganesh S.']
|
2023-04-11
| null | null | null | null |
['causal-discovery', 'interpretable-machine-learning']
|
['knowledge-base', 'methodology']
|
[ 2.82519311e-01 -3.83247137e-01 -4.67706114e-01 -3.86764914e-01
-1.49869740e-01 -5.93591094e-01 9.61737692e-01 2.66005874e-01
2.76507456e-02 1.02206028e+00 4.98848259e-01 -8.14812541e-01
-9.51892376e-01 -1.04914856e+00 -5.53470910e-01 -7.78005719e-01
-7.75951385e-01 3.08385789e-01 7.20850304e-02 -8.01888853e-02
3.98422629e-01 4.94287848e-01 -1.74959123e+00 1.67200118e-01
9.86572742e-01 4.47383881e-01 9.06511992e-02 6.75682127e-01
8.26335996e-02 6.62968338e-01 -1.61071688e-01 1.73503041e-01
7.00203627e-02 -4.51960504e-01 -3.14361483e-01 -6.32172346e-01
6.84588552e-02 -9.49794278e-02 5.44155091e-02 3.34289104e-01
5.51667288e-02 -4.60939668e-02 9.38371658e-01 -1.50544834e+00
-3.10556173e-01 9.15823698e-01 -7.64716804e-01 5.32395661e-01
-3.18614356e-02 2.05771700e-01 1.27320147e+00 -8.93938541e-01
3.85208994e-01 1.63146877e+00 7.30568111e-01 -1.57296628e-01
-1.69617260e+00 -9.86949146e-01 3.61317188e-01 2.19670102e-01
-1.32913303e+00 -2.14281887e-01 4.01275277e-01 -8.71085346e-01
1.05793536e+00 6.97679043e-01 4.65081960e-01 8.99579525e-01
7.33469307e-01 2.94387877e-01 1.12941420e+00 -2.69499242e-01
2.28911728e-01 -1.00078762e-01 3.08663815e-01 3.78358930e-01
4.83236104e-01 9.53362107e-01 -8.51826847e-01 -7.27362454e-01
6.17283225e-01 1.81716144e-01 -8.39057714e-02 1.93748280e-01
-1.34039271e+00 1.19343543e+00 2.17371568e-01 1.36265963e-01
-5.69138706e-01 1.73032269e-01 3.29583623e-02 3.48082304e-01
5.79644322e-01 7.20118344e-01 -1.12864947e+00 1.74299970e-01
-8.48292291e-01 4.88981903e-01 6.05234265e-01 2.91310102e-01
8.46322596e-01 -1.00266300e-01 -1.43533602e-01 4.28047389e-01
3.62740815e-01 7.65115678e-01 1.35190368e-01 -4.81529593e-01
6.02957085e-02 6.45791113e-01 5.12184016e-02 -1.16684437e+00
-7.05439031e-01 -4.37981039e-01 -5.99321067e-01 1.18458271e-01
1.80738121e-01 -5.11056185e-01 -7.49184668e-01 1.67874491e+00
3.19529384e-01 4.61810976e-01 -3.17195535e-01 7.93174863e-01
2.94984132e-01 6.96115613e-01 6.11235440e-01 -5.49792111e-01
1.04759312e+00 8.64066277e-03 -5.97095728e-01 -1.09482490e-01
3.62470239e-01 -5.50374210e-01 8.66436422e-01 1.74483344e-01
-9.75008011e-02 -3.72698724e-01 -7.21598446e-01 6.37516201e-01
-5.20692527e-01 -1.76124394e-01 1.31445181e+00 1.99561045e-01
-6.92237973e-01 8.58868420e-01 -8.60839844e-01 -2.38401055e-01
6.67516440e-02 5.15871406e-01 -2.23217294e-01 2.24890038e-01
-1.55324090e+00 8.27840209e-01 1.72600076e-01 1.82177441e-03
-7.52702713e-01 -1.21381140e+00 -5.60877442e-01 3.51245366e-02
2.38762185e-01 -7.41912186e-01 5.96976459e-01 -8.63917530e-01
-7.81706989e-01 1.33270755e-01 -3.26105565e-01 -3.11946273e-01
1.18607596e-01 -4.72847283e-01 -5.01149654e-01 -3.39262664e-01
3.49003404e-01 3.59973222e-01 9.46400464e-01 -1.13105643e+00
-9.90193427e-01 -4.74074423e-01 -3.96587461e-01 -9.66631435e-03
-1.17171690e-01 9.27007943e-02 1.60633981e-01 -6.88717902e-01
5.72418533e-02 -8.78739417e-01 -5.32565475e-01 -5.98775089e-01
-5.10690570e-01 -1.38166457e-01 9.51764345e-01 -4.40727621e-01
1.55610502e+00 -1.83403134e+00 3.74622732e-01 4.73696053e-01
2.74466723e-01 -4.90477204e-01 -6.32582679e-02 5.46572149e-01
-3.54134321e-01 4.78656471e-01 -2.33754963e-01 3.40231091e-01
-5.05710602e-01 3.09890389e-01 -5.40416896e-01 4.16850597e-01
6.78886592e-01 5.11668622e-01 -7.86130071e-01 -2.77500659e-01
4.63874876e-01 2.70991147e-01 -6.48913145e-01 2.20634758e-01
-1.37187123e-01 6.16629064e-01 -6.24487579e-01 5.14439285e-01
4.34421927e-01 -1.10656559e-01 2.44304553e-01 1.74453616e-01
-7.95024395e-01 3.19502443e-01 -1.14830685e+00 6.12413526e-01
-4.40567404e-01 6.07397199e-01 -3.66212636e-01 -6.90188050e-01
9.09020782e-01 1.19427584e-01 7.16817498e-01 -2.89544046e-01
-3.18982124e-01 4.30035107e-02 2.88588434e-01 -4.94699717e-01
-7.60610565e-04 -2.38024428e-01 -1.49298087e-01 4.55203503e-01
-2.20736459e-01 1.57288313e-01 -5.01807034e-02 -4.59606713e-03
1.35377741e+00 1.20893955e-01 5.61685383e-01 -6.82114482e-01
7.85663277e-02 4.46450174e-01 9.90524113e-01 9.01556611e-01
2.41252750e-01 2.39962026e-01 7.29494393e-01 -6.26833558e-01
-6.43026471e-01 -1.16215038e+00 -6.03203058e-01 1.13152134e+00
-2.58541554e-01 -4.05498654e-01 1.75116569e-01 -4.92885500e-01
6.48716271e-01 1.08151364e+00 -1.00671244e+00 -2.47972935e-01
-3.82522732e-01 -1.56727719e+00 2.91892022e-01 2.65008986e-01
-2.57669091e-01 -6.38783157e-01 -4.37341839e-01 4.32966530e-01
4.46202159e-02 -3.96020681e-01 2.55871654e-01 6.92010999e-01
-9.69423771e-01 -1.28173542e+00 2.97258377e-01 3.29712123e-01
3.69656086e-01 2.40564495e-01 9.77221847e-01 -8.81494135e-02
-2.63361335e-01 -1.35982350e-01 -2.26515621e-01 -6.31131291e-01
-1.76178709e-01 -9.88991782e-02 2.50866294e-01 -6.62390962e-02
6.22230947e-01 -8.54764640e-01 -4.53060597e-01 3.65365475e-01
-6.95111752e-01 9.09983963e-02 5.63851535e-01 8.26084137e-01
4.63826925e-01 2.94972628e-01 7.19574034e-01 -7.07305014e-01
3.31323534e-01 -1.20591652e+00 -6.89447343e-01 5.31851985e-02
-8.93334031e-01 1.86387330e-01 4.96313900e-01 -4.89149392e-01
-1.07601380e+00 8.98521468e-02 4.47135538e-01 -3.53023529e-01
-1.98891625e-01 9.72841442e-01 1.06268413e-02 2.98941374e-01
7.42158890e-01 -2.42282495e-01 -2.16904044e-01 -4.22665238e-01
1.30285189e-01 2.28316113e-01 6.43232390e-02 -3.26998502e-01
9.48605776e-01 3.55636358e-01 3.78207147e-01 -8.46986175e-01
-5.52075088e-01 -3.75965059e-01 -9.29694951e-01 -2.86414146e-01
6.62375987e-01 -1.00152922e+00 -3.39946270e-01 -9.55110192e-02
-8.32143605e-01 -2.38998443e-01 1.19370334e-01 9.27002132e-01
-1.93395063e-01 -5.49483657e-01 -6.60481602e-02 -9.18877542e-01
-4.59676981e-02 -8.89862478e-01 9.01395917e-01 9.34492424e-02
-7.41445363e-01 -1.14898837e+00 4.28367615e-01 -1.72336385e-01
3.65572184e-01 5.62402308e-01 1.28628588e+00 -5.84853411e-01
-4.56077814e-01 -8.00297828e-04 -4.73292060e-02 -4.31098431e-01
5.03495157e-01 8.76153827e-01 -1.02325404e+00 2.17688128e-01
-4.98204887e-01 3.91928434e-01 1.18303072e+00 1.01339328e+00
1.00913191e+00 -3.68984729e-01 -7.63927758e-01 5.32794237e-01
1.30205846e+00 2.81554222e-01 3.28638822e-01 2.28862911e-01
6.36474311e-01 1.00009155e+00 7.84684360e-01 7.43263245e-01
3.70554209e-01 5.27091861e-01 3.47501814e-01 -3.63567501e-01
3.51616770e-01 -2.41317168e-01 3.24424982e-01 2.13146016e-01
-3.18625152e-01 1.03073850e-01 -1.19619071e+00 4.29717809e-01
-2.00162864e+00 -1.29453886e+00 -9.34101105e-01 2.35068345e+00
8.06521595e-01 2.36943085e-02 9.39797834e-02 2.81657446e-02
5.60427487e-01 -1.21711001e-01 -4.93370265e-01 -2.64338285e-01
-3.76544982e-01 1.78103909e-01 8.76297891e-01 6.20618045e-01
-1.36908793e+00 7.88688779e-01 6.63637352e+00 3.45191985e-01
-1.11356461e+00 -2.47098491e-01 7.70984530e-01 -1.92728594e-01
-5.09769619e-01 5.15459180e-01 -1.00595856e+00 1.23778433e-01
1.29937708e+00 -3.52729887e-01 3.31412733e-01 5.96320689e-01
1.16649389e+00 -2.80655354e-01 -1.06029510e+00 2.88306743e-01
-7.48767674e-01 -1.41819537e+00 -4.44550952e-03 2.63968319e-01
6.71806753e-01 1.83965936e-01 -1.79959089e-01 1.86174810e-01
8.57669413e-01 -1.06106412e+00 3.36599261e-01 7.92990923e-01
5.12283981e-01 -7.91553199e-01 4.65117365e-01 1.27720386e-01
-1.17250097e+00 -4.98704195e-01 -2.43725702e-01 -5.99960566e-01
-2.47839876e-02 1.25438917e+00 -8.32933009e-01 4.02270317e-01
8.82205248e-01 9.40951526e-01 -6.49525583e-01 7.85470068e-01
-1.99658498e-01 1.15648580e+00 -5.25991201e-01 6.14613593e-02
-8.61882195e-02 -3.93042825e-02 7.87621140e-01 1.05461216e+00
2.62863904e-01 2.72567272e-01 -3.17803584e-02 8.71465087e-01
6.52381003e-01 3.40214074e-02 -8.57860446e-01 1.38546288e-01
5.99879205e-01 1.16527653e+00 -6.06424034e-01 -2.69438401e-02
-4.65227425e-01 2.04993680e-01 -1.86668690e-02 4.02168900e-01
-7.42375970e-01 5.11921803e-03 1.16084754e+00 2.15159774e-01
-4.23547775e-02 -2.20725387e-01 -7.18524754e-01 -1.05985391e+00
-7.00632572e-01 -6.89378560e-01 6.72008991e-01 -2.84879357e-01
-1.33332920e+00 9.64314565e-02 4.56977457e-01 -1.14392257e+00
-3.77013952e-01 -2.71754980e-01 -1.01118863e+00 1.06440735e+00
-1.39552844e+00 -9.61568058e-01 2.91038789e-02 4.13257152e-01
3.57406527e-01 2.10486963e-01 8.67184818e-01 -9.51256603e-02
-7.24721313e-01 -2.25660369e-01 6.94729760e-02 -4.90313679e-01
7.64977098e-01 -1.24420440e+00 1.92794502e-01 7.93832183e-01
-2.27693751e-01 8.92409563e-01 9.20221031e-01 -1.21368086e+00
-1.51049829e+00 -1.39778185e+00 1.04189336e+00 -3.55782121e-01
1.09830439e+00 -2.19957352e-01 -9.27145898e-01 5.57991743e-01
-2.28615761e-01 -4.02054042e-01 1.11424720e+00 7.02962399e-01
-1.16228528e-01 -1.68617576e-01 -7.76274741e-01 5.68541884e-01
8.43694389e-01 8.01476389e-02 -5.33343077e-01 2.50057727e-01
6.09638333e-01 5.48142135e-01 -9.96276438e-01 5.99803925e-01
7.27726281e-01 -6.55040443e-01 8.76207232e-01 -7.63346851e-01
6.19127214e-01 -5.07474601e-01 -1.01425700e-01 -1.51582384e+00
-7.48779416e-01 -3.31823409e-01 2.83969909e-01 1.06751776e+00
8.90816331e-01 -6.02369726e-01 1.62066266e-01 5.79129994e-01
2.31422976e-01 -2.24521592e-01 -8.21635723e-01 -5.44539630e-01
9.14973542e-02 -7.52639294e-01 7.96381235e-01 1.39200544e+00
-6.35819063e-02 3.28852415e-01 -4.27767813e-01 6.30331576e-01
8.92761648e-01 5.02722681e-01 6.86530590e-01 -1.92795324e+00
-3.20727110e-01 -3.88440609e-01 -6.83861747e-02 2.49739978e-02
3.80478762e-02 -5.38571119e-01 -2.84192890e-01 -1.05317891e+00
1.40138939e-01 -8.53601754e-01 -2.82565802e-01 9.45925891e-01
-5.75975597e-01 -3.30234587e-01 -3.64387214e-01 3.37340921e-01
3.70242715e-01 4.69595551e-01 7.74924815e-01 6.82103038e-02
-8.07893991e-01 2.27267712e-01 -5.68860650e-01 6.85318589e-01
8.19506884e-01 -8.22668135e-01 -3.96112472e-01 -5.67885078e-02
3.20452154e-01 2.29511961e-01 6.32969022e-01 -4.83156919e-01
-1.40862197e-01 -1.16047442e+00 7.44125962e-01 -7.04176009e-01
-3.42897117e-01 -6.44023180e-01 7.62967587e-01 5.64942837e-01
-2.89523363e-01 1.11136533e-01 5.05495250e-01 4.76062477e-01
-8.99104923e-02 4.57788914e-01 3.30949992e-01 8.14769417e-02
-6.06682420e-01 1.94647908e-01 -7.03084767e-01 -5.23184240e-01
9.11921024e-01 1.29837632e-01 -2.61464030e-01 -1.37441158e-01
-7.50275433e-01 4.78687197e-01 2.22831946e-02 7.63336837e-01
5.94602585e-01 -1.02709544e+00 -1.06109202e+00 1.68606475e-01
1.41517520e-01 -5.38674533e-01 9.25662294e-02 1.02902579e+00
8.00929219e-03 5.33082366e-01 -1.50800850e-02 -6.66414082e-01
-1.10936511e+00 3.62812340e-01 1.18212588e-01 -1.76587969e-01
-3.64017904e-01 5.75495541e-01 4.19370651e-01 -3.25086236e-01
-5.41428089e-01 -2.62700021e-01 -4.93855923e-01 4.51128960e-01
5.85497916e-01 5.07221818e-01 -1.02461815e-01 -3.12720329e-01
-6.23459399e-01 3.40258718e-01 3.13303381e-01 -1.15235388e-01
1.84425640e+00 -1.03546225e-01 -2.97603905e-01 7.18493283e-01
8.13513994e-01 5.79477996e-02 -1.19809687e+00 2.42774665e-01
3.63246828e-01 -5.01523674e-01 5.24354875e-01 -8.00305963e-01
-8.39128971e-01 5.40550709e-01 4.52879906e-01 2.03647971e-01
1.10969102e+00 -1.11942068e-01 -2.29465455e-01 -6.68063536e-02
1.21801428e-01 -5.57773829e-01 -6.76685512e-01 2.73045480e-01
1.15358698e+00 -1.22992158e+00 3.28578174e-01 -2.68458933e-01
-3.08723152e-01 1.26630116e+00 5.63092172e-01 -3.44860666e-02
1.06255984e+00 5.07058322e-01 -2.24443957e-01 -3.86684865e-01
-1.50237632e+00 -3.23618591e-01 3.59715104e-01 3.35364074e-01
4.81345743e-01 5.49790323e-01 -5.21282971e-01 3.89315099e-01
-2.29409173e-01 -2.33008027e-01 3.45081508e-01 5.84413528e-01
-3.56970221e-01 -9.53730524e-01 -7.07724333e-01 9.94339406e-01
-3.93307924e-01 -5.22214770e-01 -4.53749448e-01 1.05264962e+00
1.07410885e-01 1.22742009e+00 3.10975015e-01 -6.92515850e-01
1.65805042e-01 -9.75996926e-02 -3.08107525e-01 -5.07952332e-01
-6.30613387e-01 3.40817302e-01 2.19035551e-01 -6.12937152e-01
-3.38702917e-01 -1.21542013e+00 -1.12057686e+00 -4.37908977e-01
-5.33497989e-01 1.20855860e-01 6.06043696e-01 9.15514290e-01
4.18513894e-01 5.45408666e-01 1.03256345e+00 -7.36229897e-01
-1.62639879e-02 -9.30970430e-01 -3.62773687e-01 4.01832350e-02
4.22535837e-01 -1.12174928e+00 -7.91888773e-01 1.09081745e-01]
|
[7.698473930358887, 5.106581211090088]
|
35a0cc49-02da-45b9-9e34-1864c7c76c63
|
quick-and-not-so-dirty-unsupervised-selection-1
|
1911.07176
| null |
https://arxiv.org/abs/1911.07176v2
|
https://arxiv.org/pdf/1911.07176v2.pdf
|
Quick and (not so) Dirty: Unsupervised Selection of Justification Sentences for Multi-hop Question Answering
|
We propose an unsupervised strategy for the selection of justification sentences for multi-hop question answering (QA) that (a) maximizes the relevance of the selected sentences, (b) minimizes the overlap between the selected facts, and (c) maximizes the coverage of both question and answer. This unsupervised sentence selection method can be coupled with any supervised QA approach. We show that the sentences selected by our method improve the performance of a state-of-the-art supervised QA model on two multi-hop QA datasets: AI2's Reasoning Challenge (ARC) and Multi-Sentence Reading Comprehension (MultiRC). We obtain new state-of-the-art performance on both datasets among approaches that do not use external resources for training the QA system: 56.82% F1 on ARC (41.24% on Challenge and 64.49% on Easy) and 26.1% EM0 on MultiRC. Our justification sentences have higher quality than the justifications selected by a strong information retrieval baseline, e.g., by 5.4% F1 in MultiRC. We also show that our unsupervised selection of justification sentences is more stable across domains than a state-of-the-art supervised sentence selection method.
|
['Mihai Surdeanu', 'Vikas Yadav', 'Steven Bethard']
|
2019-11-17
|
quick-and-not-so-dirty-unsupervised-selection
|
https://aclanthology.org/D19-1260
|
https://aclanthology.org/D19-1260.pdf
|
ijcnlp-2019-11
|
['multi-hop-question-answering']
|
['knowledge-base']
|
[ 3.81750852e-01 8.55830908e-01 1.16667688e-01 -7.42480278e-01
-2.01644373e+00 -7.11914062e-01 5.66356719e-01 6.49396896e-01
-4.89042521e-01 9.50402439e-01 5.76281428e-01 -4.58433837e-01
-4.36308891e-01 -8.95045877e-01 -8.35487783e-01 -1.38674229e-01
5.12971461e-01 1.03603375e+00 6.71585023e-01 -8.12214553e-01
4.43386883e-01 -3.75628889e-01 -1.40344286e+00 7.96984673e-01
1.51561511e+00 7.26720750e-01 1.92776173e-01 8.93119514e-01
-2.00037971e-01 1.37667143e+00 -8.04006696e-01 -6.45918250e-01
-3.76776248e-01 -6.39960885e-01 -1.90509439e+00 -3.58593434e-01
9.91954744e-01 -3.44836339e-02 4.72565144e-02 8.70136797e-01
4.91119087e-01 1.43566996e-01 6.15389407e-01 -6.50256932e-01
-5.52717805e-01 8.36266816e-01 -8.07470158e-02 3.27714920e-01
1.01629162e+00 -6.17437400e-02 1.52876806e+00 -6.60327852e-01
8.57674956e-01 1.23991942e+00 3.49969864e-01 5.80048501e-01
-1.04534638e+00 3.02007869e-02 -1.11152545e-01 5.75244725e-01
-9.20934319e-01 -6.28867626e-01 4.42487717e-01 -8.22951794e-02
1.24008942e+00 5.21283209e-01 3.75257246e-02 7.62022972e-01
1.53684244e-01 8.53067279e-01 1.09647238e+00 -9.09589052e-01
3.44116092e-01 2.71122251e-02 6.20615840e-01 8.18754911e-01
-1.85816392e-01 -7.12088227e-01 -6.22537255e-01 -4.25205082e-01
-1.64307401e-01 -8.53849113e-01 -2.67877609e-01 1.74082562e-01
-1.08613682e+00 1.11080086e+00 1.72030061e-01 2.22624898e-01
-3.67302686e-01 -9.25685316e-02 1.47493169e-01 6.45813584e-01
1.67858914e-01 1.01718771e+00 -7.28635609e-01 -2.03530490e-01
-6.71280503e-01 5.32343507e-01 1.03830230e+00 8.56575489e-01
6.47669315e-01 -6.25705659e-01 -4.99987066e-01 1.09459293e+00
1.71177104e-01 9.40887570e-01 2.92905450e-01 -1.54305160e+00
1.01425982e+00 6.93497598e-01 1.18919864e-01 -8.02524567e-01
-4.16209996e-01 -2.73752004e-01 -3.38411391e-01 -5.12881935e-01
6.11270607e-01 -2.07510039e-01 -6.11679375e-01 1.88217342e+00
1.27077043e-01 -7.69209802e-01 5.50773025e-01 6.38993084e-01
1.28026307e+00 7.47692049e-01 1.44627005e-01 -1.03765361e-01
1.52797341e+00 -1.23527801e+00 -7.76667297e-01 -4.04050946e-01
1.06754863e+00 -7.67583907e-01 1.43061495e+00 2.12204188e-01
-1.46092403e+00 -3.21983576e-01 -7.86272168e-01 -3.86210531e-01
1.80103220e-02 -8.03096965e-02 4.22596872e-01 4.77003276e-01
-1.08662736e+00 2.25940242e-01 -2.22589582e-01 -3.26120794e-01
1.42940460e-02 1.34651005e-01 -8.38862807e-02 -4.68990773e-01
-1.72019994e+00 1.35991645e+00 1.27564460e-01 -5.34303844e-01
-8.87736022e-01 -5.56231976e-01 -8.32264841e-01 2.45207608e-01
6.39730692e-01 -9.93301630e-01 1.70552647e+00 -6.46486104e-01
-1.39639127e+00 8.32482517e-01 -4.61838692e-01 -5.54630876e-01
1.81829646e-01 -5.29216468e-01 -3.51215482e-01 8.37747514e-01
7.81628966e-01 8.04467380e-01 3.28215688e-01 -1.00010085e+00
-5.41027486e-01 -9.16939927e-04 5.34934759e-01 5.98061442e-01
-1.04611762e-01 6.68413490e-02 -2.61154771e-01 1.97716504e-02
2.66675558e-02 -5.82480192e-01 -3.80371436e-02 -6.01224840e-01
-5.24111927e-01 -6.86143994e-01 2.44273514e-01 -8.27128351e-01
1.14216244e+00 -1.52385962e+00 3.35558981e-01 -9.01582092e-03
2.06655096e-02 -6.97251856e-02 -4.16999608e-01 8.08906257e-01
1.77553073e-01 3.22562642e-02 -4.93743598e-01 -1.71091691e-01
6.95861131e-02 1.15546383e-01 -4.80661780e-01 -4.69612032e-02
2.60073990e-01 8.75013113e-01 -1.15257716e+00 -9.21063840e-01
-2.85309047e-01 -2.08374515e-01 -6.27765357e-01 1.95746362e-01
-7.70250559e-01 2.64322430e-01 -6.32295132e-01 5.55294216e-01
1.75239041e-01 -7.25887537e-01 2.30348676e-01 3.93483452e-02
3.72909367e-01 1.14249921e+00 -5.19333720e-01 1.86808574e+00
-4.33431953e-01 5.31816423e-01 -3.96451354e-01 -6.27604008e-01
7.04293847e-01 4.23194796e-01 9.88489613e-02 -1.07262182e+00
-1.74468875e-01 4.71423477e-01 1.40508607e-01 -8.08598161e-01
5.71902156e-01 4.09690887e-02 -3.33502024e-01 6.44869924e-01
1.05556600e-01 -6.18648529e-01 6.38090789e-01 1.04451907e+00
1.47806621e+00 -3.75067681e-01 1.14983201e-01 -5.51893711e-01
9.15938199e-01 4.98457581e-01 9.83566628e-04 1.16079009e+00
1.41755417e-01 5.17464876e-01 7.02524424e-01 5.87143935e-02
-6.34829104e-01 -9.93355751e-01 -1.21097319e-01 1.12713802e+00
1.41587988e-01 -5.58303118e-01 -9.36477482e-01 -1.04885483e+00
-2.45371625e-01 1.41292024e+00 -3.84143174e-01 -6.71229718e-05
-7.87476540e-01 -3.02787185e-01 6.94609344e-01 1.97592318e-01
7.17184901e-01 -1.06773579e+00 -5.74595988e-01 6.09922297e-02
-1.07534730e+00 -1.18476391e+00 -2.12944135e-01 -1.20920494e-01
-6.71180964e-01 -1.19225347e+00 -2.57009745e-01 -7.96187341e-01
4.94517326e-01 -2.77844295e-02 1.79553270e+00 4.16603893e-01
5.21378338e-01 7.43613243e-01 -7.87530959e-01 -1.65187210e-01
-6.28172040e-01 4.21026200e-01 -3.23938936e-01 -5.29104292e-01
3.14501852e-01 -4.22873870e-02 -3.23490739e-01 2.03373462e-01
-7.09961772e-01 6.73361793e-02 4.72798347e-01 9.13008630e-01
4.50398564e-01 -1.55603126e-01 1.07484531e+00 -1.10581172e+00
1.01506400e+00 -4.73672718e-01 -1.12034492e-01 7.67241299e-01
-6.08647823e-01 2.90159613e-01 5.50576508e-01 1.00526847e-01
-1.28143406e+00 -4.68798161e-01 -5.36245525e-01 6.42456055e-01
-9.03238431e-02 8.40598583e-01 3.65301408e-03 2.20937535e-01
9.51084316e-01 -1.02203846e-01 -2.61108577e-01 -3.33509028e-01
4.99296844e-01 8.13338935e-01 5.09380579e-01 -7.92548716e-01
5.19176304e-01 1.18385792e-01 -2.35024482e-01 -6.41923308e-01
-1.67337418e+00 -4.47293580e-01 -3.66557628e-01 -1.50124520e-01
1.02210414e+00 -6.97659433e-01 -5.15036821e-01 -1.32147763e-02
-1.41422522e+00 -3.02653849e-01 -2.88406283e-01 2.27603629e-01
-5.58709681e-01 5.82993090e-01 -6.40997231e-01 -6.46468103e-01
-6.31982267e-01 -9.12245929e-01 1.13507426e+00 2.29739338e-01
-6.75965846e-01 -9.53674197e-01 1.18610814e-01 1.30040777e+00
3.42684984e-01 -1.80150509e-01 1.31405699e+00 -8.84172857e-01
-4.13822800e-01 -2.89383288e-02 -1.30251899e-01 1.41022220e-01
-1.26637936e-01 -4.24162030e-01 -8.04701865e-01 -1.58841088e-02
1.87279418e-01 -1.12060821e+00 9.13671434e-01 2.93246478e-01
7.66747832e-01 -2.66646177e-01 2.67931074e-02 -3.32888424e-01
1.20363545e+00 -1.42013237e-01 7.95110583e-01 5.75205386e-01
2.02685311e-01 8.78385663e-01 9.94415045e-01 -1.46002829e-01
9.00083959e-01 6.30915403e-01 2.95587659e-01 2.72011101e-01
-1.14046209e-01 -1.69665605e-01 3.84514004e-01 1.00484312e+00
2.93983728e-01 -4.26447868e-01 -1.20507300e+00 9.35466826e-01
-1.90806532e+00 -8.90801370e-01 -5.08577228e-01 1.77369905e+00
1.29675245e+00 2.63360202e-01 -1.33820683e-01 -8.30955952e-02
3.92246068e-01 1.85171533e-02 -2.47643173e-01 -5.11427641e-01
-5.86705387e-01 4.51955706e-01 -2.22849995e-02 1.04510617e+00
-9.30664122e-01 9.36788142e-01 6.13035011e+00 9.20188427e-01
-1.08797148e-01 2.81095952e-01 5.87050974e-01 1.74294531e-01
-8.67056966e-01 1.85906321e-01 -7.25226820e-01 6.67669326e-02
1.25799072e+00 -3.57209966e-02 1.41734509e-02 5.48697710e-01
-2.11109936e-01 -6.44411087e-01 -1.06831646e+00 2.38025531e-01
3.39546502e-01 -1.51547325e+00 1.61015734e-01 -4.59074199e-01
8.62905920e-01 1.05022518e-02 -3.42423052e-01 5.23242831e-01
4.91366506e-01 -1.12086535e+00 5.59711397e-01 4.41951692e-01
5.23953259e-01 -5.03399670e-01 1.21726918e+00 6.42818213e-01
-5.71627796e-01 -4.98857982e-02 -3.77179503e-01 -1.13451265e-01
5.69395304e-01 5.98519444e-01 -5.74393690e-01 7.81814098e-01
5.86023688e-01 1.71590477e-01 -9.21452761e-01 5.79478204e-01
-8.37494493e-01 9.28771317e-01 -2.16585860e-01 -4.61998880e-01
4.20516670e-01 1.97778970e-01 6.17320597e-01 1.11841166e+00
-2.39677399e-01 2.22489089e-01 -4.15353440e-02 5.15689373e-01
-3.58094126e-01 2.96388298e-01 -1.39116630e-01 1.50525540e-01
6.05629325e-01 7.84457922e-01 -9.19796079e-02 -6.21028781e-01
-2.51421362e-01 6.51779175e-01 5.75532615e-01 1.05836213e-01
-6.03867114e-01 -6.37822866e-01 -1.78967178e-01 -2.83278793e-01
9.73200202e-02 1.27638057e-01 -2.39964306e-01 -1.20352650e+00
2.81191915e-01 -1.24753559e+00 8.72014940e-01 -1.06778681e+00
-1.44874370e+00 8.09780896e-01 1.19338773e-01 -8.07035029e-01
-5.72860420e-01 -2.54309326e-01 -4.59368974e-01 8.42491448e-01
-1.71909976e+00 -7.38739252e-01 -1.26858249e-01 4.10526484e-01
6.55790329e-01 -1.03734240e-01 1.21992338e+00 1.18779078e-01
4.95388079e-03 3.72829676e-01 -2.12512583e-01 -6.11423664e-02
8.34326386e-01 -1.51340413e+00 1.34106144e-01 7.58054674e-01
1.44672260e-01 7.19194710e-01 7.09347129e-01 -7.03084111e-01
-1.30005848e+00 -6.37280166e-01 1.76433098e+00 -9.13876355e-01
6.95421100e-01 1.47973269e-01 -9.28588390e-01 5.56835175e-01
8.45020950e-01 -9.28292155e-01 6.47993326e-01 3.60382825e-01
-3.42756778e-01 7.33715668e-02 -1.23823369e+00 5.83613455e-01
5.63453376e-01 -5.56842744e-01 -1.40288055e+00 1.02640796e+00
1.07161927e+00 -4.92504239e-01 -9.09039974e-01 4.34910685e-01
5.44395596e-02 -7.83310831e-01 9.39505458e-01 -8.32767606e-01
1.03059947e+00 -1.09425738e-01 -4.23551589e-01 -1.17649472e+00
-6.50186464e-02 -3.99016649e-01 -4.33345065e-02 1.03647339e+00
1.13879228e+00 -4.83140767e-01 4.03749228e-01 6.63280725e-01
-3.02947611e-01 -8.69216442e-01 -1.02275372e+00 -5.04197657e-01
3.46743166e-01 -2.20413208e-01 2.72435665e-01 7.79006541e-01
2.48322695e-01 1.08230853e+00 5.13180271e-02 2.15435550e-01
4.80336398e-01 3.87257546e-01 4.63766187e-01 -9.38396096e-01
-3.68350565e-01 -1.43817842e-01 2.81386584e-01 -1.35494077e+00
1.65906459e-01 -8.26634586e-01 1.27292812e-01 -2.27336669e+00
3.30729216e-01 -2.57649988e-01 1.11221470e-01 3.81212115e-01
-4.48646396e-01 -2.13417634e-01 7.57373348e-02 1.80382147e-01
-1.22012460e+00 6.27874196e-01 1.28085721e+00 -2.28471443e-01
1.03141285e-01 -3.33972692e-01 -9.92828310e-01 6.00862265e-01
8.11335444e-01 -5.71382225e-01 -5.45722067e-01 -9.44349408e-01
6.23043418e-01 4.35434312e-01 1.04528517e-01 -7.08119571e-01
4.57538754e-01 -1.26491085e-01 -1.83571041e-01 -5.59883177e-01
4.00926948e-01 -1.43760443e-01 -5.45273542e-01 3.92868280e-01
-8.32934916e-01 9.08717513e-02 2.18581297e-02 2.85347313e-01
-3.86164218e-01 -8.95956218e-01 4.61374432e-01 -2.70042032e-01
-3.89632016e-01 -5.76174557e-01 -4.57096875e-01 8.90275419e-01
2.61359394e-01 3.38615924e-01 -1.09302437e+00 -8.69970620e-01
-3.67477059e-01 7.18257666e-01 -2.64734980e-02 1.21037729e-01
6.72797561e-01 -8.94079626e-01 -1.17962837e+00 -6.04441643e-01
1.59886062e-01 3.92623320e-02 2.36976147e-01 9.73887622e-01
-4.61036235e-01 8.50616157e-01 2.53916919e-01 -5.24055898e-01
-1.17438483e+00 -2.36911178e-01 1.57921731e-01 -7.82263160e-01
-3.72691631e-01 1.01502013e+00 -3.36738586e-01 -7.07510948e-01
-9.74581111e-03 1.60782412e-01 -5.68289757e-01 -3.07675600e-02
4.88453031e-01 2.75698423e-01 2.00692862e-01 -4.09791052e-01
-4.66075599e-01 5.28091371e-01 -1.78198516e-01 -4.15917367e-01
1.30998611e+00 -2.50710189e-01 -4.52352047e-01 2.36700594e-01
8.96772325e-01 3.78023326e-01 -3.24889123e-01 -1.68828547e-01
2.19616354e-01 -1.56742051e-01 -1.33311659e-01 -1.49698496e+00
-3.09229106e-01 6.20191276e-01 -8.98523480e-02 5.17778397e-01
9.88149703e-01 4.28629011e-01 9.57917511e-01 1.01265264e+00
3.53582829e-01 -1.08713937e+00 3.66819739e-01 9.44308579e-01
1.14199901e+00 -1.21711075e+00 3.51303257e-02 -6.29871070e-01
-1.06329572e+00 9.24164355e-01 5.92891157e-01 1.90266103e-01
-4.70917876e-04 -3.32486868e-01 2.22888783e-01 -8.08288574e-01
-1.19388735e+00 -2.72538662e-01 4.69184160e-01 2.37084746e-01
4.59067643e-01 -1.23731554e-01 -5.31372547e-01 6.23094559e-01
-5.24665236e-01 -4.54588890e-01 6.77006125e-01 9.67326820e-01
-6.90782547e-01 -1.00966406e+00 -1.22584715e-01 5.19834161e-01
-3.88391823e-01 -4.29697931e-01 -7.48197556e-01 6.51075482e-01
-4.33045238e-01 1.78578210e+00 -3.89528036e-01 -1.10688783e-01
4.40811098e-01 3.51093620e-01 5.51098168e-01 -8.53602409e-01
-8.66225719e-01 -3.78285974e-01 1.19984138e+00 -2.78357685e-01
-6.21651649e-01 -7.56679416e-01 -1.32016599e+00 -1.04820088e-01
-5.60944557e-01 6.89126015e-01 2.92774290e-01 1.40143311e+00
4.19867754e-01 4.14606422e-01 2.18170941e-01 2.69944280e-01
-8.94429505e-01 -1.17720103e+00 3.69725190e-02 5.08506119e-01
1.66425854e-02 -3.41388583e-01 -3.94403249e-01 -6.25078529e-02]
|
[11.260625839233398, 7.961385726928711]
|
431b4f2c-b8b7-440b-aff7-a115837fcd60
|
safer-data-efficient-and-safe-reinforcement
| null | null |
https://openreview.net/forum?id=xwAw8QZkpWZ
|
https://openreview.net/pdf?id=xwAw8QZkpWZ
|
SAFER: Data-Efficient and Safe Reinforcement Learning Through Skill Acquisition
|
Though many reinforcement learning (RL) problems involve learning policies in settings that are difficult to specify safety constraints and sparse rewards, current methods struggle to rapidly and safely acquire successful policies. Behavioral priors, which extract useful policy primitives for learning from offline datasets, have recently shown considerable promise at accelerating RL in more complex problems. However, we discover that current behavioral priors may not be well-equipped for safe policy learning, and in some settings, may promote unsafe behavior, due to their tendency to ignore data from undesirable behaviors. To overcome these issues, we propose SAFEty skill pRiors (SAFER), a behavioral prior learning algorithm that accelerates policy learning on complex control tasks, under safety constraints. Through principled contrastive training on safe and unsafe data, SAFER learns to extract a safety variable from offline data that encodes safety requirements, as well as the safe primitive skills over abstract actions in different scenarios. In the inference stage, SAFER composes a safe and successful policy from the safety skills according to the inferred safety variable and abstract action. We demonstrate its effectiveness on several complex safety-critical robotic grasping tasks inspired by the game Operation, in which SAFER not only out-performs baseline methods in learning successful policies but also enforces safety more effectively.
|
['Nevan Wichers', 'Bo Dai', 'Yinlam Chow', 'Dylan Z Slack']
|
2021-09-29
| null | null | null | null |
['robotic-grasping']
|
['robots']
|
[ 2.97410935e-01 1.97013125e-01 -6.59187734e-01 -2.30016783e-01
-7.46857524e-01 -6.90574348e-01 6.12870038e-01 2.09711000e-01
-7.24148750e-01 1.00453150e+00 3.22132617e-01 -3.72082084e-01
-3.85653734e-01 -6.16455674e-01 -1.01384342e+00 -6.19054079e-01
-3.55948031e-01 5.54544568e-01 1.39847383e-01 -3.16228509e-01
3.07077944e-01 3.00896049e-01 -1.53371871e+00 2.09701117e-02
1.02898240e+00 8.52460086e-01 2.34669313e-01 3.73404264e-01
4.99833256e-01 1.14646089e+00 -3.46396804e-01 3.18609960e-02
5.00016749e-01 -6.13695383e-02 -8.42635155e-01 -5.04986644e-01
1.76275730e-01 -9.47721124e-01 -2.64881849e-01 1.01535618e+00
1.42038256e-01 5.61489999e-01 4.79451150e-01 -1.44956100e+00
-4.11148250e-01 9.65653002e-01 -1.71055481e-01 -1.10019810e-01
3.17404985e-01 9.03116345e-01 1.12379277e+00 -6.53675273e-02
3.57234001e-01 1.46330428e+00 3.00291121e-01 1.03713179e+00
-1.37245488e+00 -7.20852435e-01 5.69902658e-01 2.40764621e-05
-5.70887446e-01 -1.35195285e-01 5.74207544e-01 -4.02662635e-01
1.00451410e+00 -8.69659428e-03 8.53302062e-01 1.97232831e+00
3.99955809e-01 1.13620234e+00 1.23951674e+00 9.73951444e-02
5.04107058e-01 -2.67734319e-01 -1.26117155e-01 5.42954266e-01
1.47169858e-01 1.13017678e+00 -4.06937629e-01 -3.32647473e-01
5.72058916e-01 -4.92531024e-02 -1.79938883e-01 -4.97172922e-01
-1.14347899e+00 7.93505847e-01 2.17915192e-01 -3.23250264e-01
-5.31397879e-01 4.82870698e-01 4.16319698e-01 3.30022097e-01
-8.38094205e-02 9.93927896e-01 -7.97613561e-01 -5.56421280e-01
-2.09876403e-01 8.02210927e-01 8.28947484e-01 1.04963601e+00
4.43487763e-01 2.87193626e-01 -5.65732658e-01 5.00266373e-01
1.29160568e-01 6.12816751e-01 3.53755839e-02 -1.42453420e+00
3.94328058e-01 2.63474196e-01 5.12948394e-01 -3.96015257e-01
-3.68494391e-01 -8.95080119e-02 -3.46920788e-01 7.02991188e-01
4.83406007e-01 -3.07000160e-01 -9.02175426e-01 2.20364380e+00
2.29823023e-01 2.23543584e-01 1.37799472e-01 7.93431401e-01
-5.20682312e-04 4.95647043e-01 5.83655775e-01 -1.85230240e-01
9.20279324e-01 -6.64110899e-01 -4.23147947e-01 -4.71780956e-01
4.47059572e-01 -4.94053662e-02 1.48275876e+00 9.28113997e-01
-1.14551127e+00 -3.24543655e-01 -8.26930404e-01 2.00443670e-01
-1.36905797e-02 -3.11060369e-01 1.00915241e+00 1.21558733e-01
-7.82361031e-01 1.05696213e+00 -1.07938457e+00 -3.50984894e-02
6.74508989e-01 5.63854396e-01 -1.66035086e-01 7.28194043e-02
-1.19942248e+00 1.28591132e+00 7.07741618e-01 -2.82795042e-01
-2.02529001e+00 -9.20958161e-01 -1.09018874e+00 9.86967012e-02
1.20425808e+00 -3.99818420e-01 1.62138629e+00 -6.66999996e-01
-1.96232986e+00 2.24041954e-01 7.15053737e-01 -8.01145613e-01
5.57016551e-01 -6.25361085e-01 5.03095624e-04 1.28237559e-02
-1.22438937e-01 9.43822980e-01 1.22179008e+00 -1.20369673e+00
-8.37048173e-01 -7.39573538e-02 7.18689978e-01 2.40365446e-01
-1.02278687e-01 -3.18662465e-01 5.85454702e-02 -5.57684302e-01
-8.07827234e-01 -1.04493356e+00 -4.52021718e-01 8.50662123e-03
-5.50606959e-02 -4.63190436e-01 5.72264791e-01 -4.48350281e-01
8.23595881e-01 -1.92806268e+00 4.59696382e-01 1.89603284e-01
-2.18629129e-02 2.95591384e-01 -3.92198354e-01 1.34516507e-01
2.16524318e-01 -2.11736098e-01 -4.16695088e-01 1.76871285e-01
2.74478495e-01 8.19363534e-01 -8.38896632e-01 4.21415210e-01
3.24958682e-01 7.91195929e-01 -1.61904097e+00 -2.88055956e-01
3.55994105e-01 -5.00124805e-02 -1.04758930e+00 6.89743698e-01
-9.02103782e-01 9.73192275e-01 -7.47293353e-01 5.13816655e-01
1.64754987e-01 3.49652350e-01 2.54365355e-01 3.68806630e-01
-3.09359562e-02 3.60704333e-01 -7.18744278e-01 1.69203615e+00
-5.18415987e-01 -9.06951129e-02 1.71674192e-01 -1.10924757e+00
6.14235342e-01 2.01296329e-01 7.59955227e-01 -9.51769471e-01
1.96030825e-01 -1.36718405e-02 1.01648517e-01 -7.12682128e-01
2.11745724e-01 -3.39444131e-01 -4.19079632e-01 3.75339657e-01
2.04647645e-01 -6.20915949e-01 1.80784479e-01 -2.18822137e-02
1.32144451e+00 9.25310194e-01 2.89872825e-01 -2.76664972e-01
3.54216844e-01 9.45964381e-02 9.22547460e-01 1.05648339e+00
-4.33168530e-01 -2.87661225e-01 8.49374771e-01 -4.00789142e-01
-9.12914038e-01 -1.30321205e+00 3.23291689e-01 1.39893866e+00
1.09065108e-01 -3.08204383e-01 -4.55916822e-01 -1.13555598e+00
3.75120312e-01 1.23555636e+00 -5.71942151e-01 -8.13162088e-01
-7.99104810e-01 -1.11483008e-01 5.75910211e-01 7.11165845e-01
5.33861369e-02 -1.57245719e+00 -1.15007460e+00 2.92041868e-01
1.70796320e-01 -8.67550552e-01 -3.54031324e-01 4.80390489e-01
-7.80527353e-01 -1.24132812e+00 -1.66664004e-01 -3.47445786e-01
6.26393080e-01 -1.48175225e-01 1.00899243e+00 4.96051135e-03
-1.23697020e-01 5.98321617e-01 -2.16206953e-01 -4.76495087e-01
-5.89147687e-01 -2.94629514e-01 4.85409886e-01 -6.05135858e-01
6.50149724e-03 -5.78213513e-01 -3.66286844e-01 1.27724856e-01
-8.32460403e-01 -4.56965379e-02 5.09222209e-01 1.05355811e+00
4.49103445e-01 -1.74063304e-03 4.14902151e-01 -7.16214240e-01
9.17210937e-01 -5.10174870e-01 -1.16131890e+00 2.36248240e-01
-7.24507093e-01 6.27189159e-01 1.17095041e+00 -7.63879299e-01
-1.19726217e+00 8.30499902e-02 -5.97014390e-02 -5.29716074e-01
-2.01411068e-01 2.17728794e-01 2.00084709e-02 2.47651041e-01
8.15239251e-01 2.64588147e-02 2.18914062e-01 -3.17692697e-01
4.18126822e-01 1.98808923e-01 5.26423275e-01 -1.82924795e+00
7.92985976e-01 2.33356386e-01 -1.34897744e-02 -1.45685688e-01
-1.02619338e+00 -2.48432737e-02 -8.33838284e-02 -1.01440020e-01
7.43484139e-01 -6.82038069e-01 -1.44357133e+00 2.10705295e-01
-6.36977494e-01 -1.11207104e+00 -5.78684330e-01 5.24650991e-01
-1.28334403e+00 1.67583510e-01 -5.77076852e-01 -9.77282584e-01
-9.86578837e-02 -1.36595547e+00 1.03669655e+00 1.03798270e-01
-3.69914293e-01 -6.83673799e-01 -4.80195209e-02 -1.40247300e-01
2.85192519e-01 2.29565099e-01 1.24282265e+00 -3.73302788e-01
-3.72594059e-01 3.80088776e-01 3.20510924e-01 6.21685445e-01
4.47029881e-02 -1.59364641e-01 -4.95352954e-01 -7.00619817e-01
-1.07030019e-01 -1.06919801e+00 5.69538474e-01 2.20698953e-01
1.60784233e+00 -6.89391911e-01 -1.25296459e-01 5.97056687e-01
1.12877464e+00 4.94671017e-01 3.85178924e-01 1.39993742e-01
3.25746626e-01 7.48653352e-01 1.02293229e+00 7.12125361e-01
1.30455196e-01 3.89490426e-01 8.00524533e-01 4.90260899e-01
3.72030914e-01 -7.06727028e-01 7.97035754e-01 -5.24444468e-02
-2.37059236e-01 8.31632987e-02 -7.49113202e-01 2.40043059e-01
-2.08756042e+00 -9.80068922e-01 6.56399667e-01 2.27697611e+00
1.28527820e+00 5.17250180e-01 2.83004671e-01 -3.05785358e-01
4.60704416e-02 -3.53615843e-02 -1.12761104e+00 -5.01171231e-01
4.63078946e-01 5.65730929e-01 6.04828596e-01 6.47981107e-01
-1.03872144e+00 1.27865255e+00 6.25873137e+00 7.12092876e-01
-9.22962666e-01 -2.02746704e-01 2.50294954e-01 -1.40356824e-01
-4.49521154e-01 1.84147567e-01 -5.98075807e-01 3.73958737e-01
6.85541987e-01 -1.17420286e-01 1.09145772e+00 1.21379578e+00
1.69507474e-01 -2.67876863e-01 -1.58358598e+00 4.39550161e-01
-5.28731465e-01 -8.61506343e-01 -3.80467698e-02 -2.51513124e-01
5.28640032e-01 -1.48156762e-01 2.68647790e-01 1.15699565e+00
1.20989966e+00 -1.20649898e+00 9.92231309e-01 3.36150229e-01
7.20142066e-01 -8.04770052e-01 2.05914110e-01 7.86114395e-01
-6.51676595e-01 -7.14873970e-01 -2.70182371e-01 -3.47320199e-01
-6.02088608e-02 -2.59844542e-01 -6.74802363e-01 1.52071357e-01
6.69683814e-01 9.04826581e-01 9.61992145e-03 5.39322019e-01
-8.75520647e-01 4.71084058e-01 -2.07530126e-01 -1.80683613e-01
5.82002580e-01 -2.03682855e-01 5.47558963e-01 5.88398159e-01
3.43491584e-02 1.55077383e-01 5.52636981e-01 1.01062191e+00
2.18161255e-01 -4.71682727e-01 -8.93365443e-01 -2.98867732e-01
2.80469805e-01 9.31588709e-01 -1.72584504e-01 -1.97487652e-01
-1.46595150e-01 4.07856315e-01 6.69350028e-01 5.06888449e-01
-9.96908665e-01 1.04529768e-01 1.09598136e+00 -2.94527173e-01
-3.13435420e-02 -4.15356249e-01 -1.46965623e-01 -9.53798950e-01
-2.46085390e-01 -1.37259316e+00 5.62417090e-01 -6.41991794e-01
-1.23692238e+00 2.29775775e-02 5.72823167e-01 -1.18962812e+00
-4.38598990e-01 -8.36105347e-01 -4.06578898e-01 4.58632916e-01
-1.31959677e+00 -7.87905514e-01 1.48813143e-01 7.73324370e-01
5.76285958e-01 -1.94683164e-01 7.70836651e-01 -3.09404075e-01
-4.06489372e-01 3.47008675e-01 -3.36791515e-01 -3.77278179e-01
6.96675777e-01 -1.35542643e+00 -1.41997505e-02 5.55318177e-01
-3.92029077e-01 7.80780137e-01 8.21216524e-01 -8.58449876e-01
-1.66493785e+00 -8.63475680e-01 -2.15933695e-01 -6.08502865e-01
8.35307360e-01 -3.35529059e-01 -7.26419866e-01 8.82310510e-01
-5.46524860e-03 -1.41348630e-01 1.52968898e-01 1.67667598e-01
-4.55360174e-01 -1.31588101e-01 -1.21354389e+00 1.14711618e+00
1.38388276e+00 -3.07135850e-01 -1.02945185e+00 2.22875014e-01
9.19341564e-01 -6.08224750e-01 -7.27179825e-01 6.57010853e-01
4.90542918e-01 -5.79716682e-01 1.02545559e+00 -1.22863483e+00
4.79515254e-01 -9.19857174e-02 6.34521917e-02 -1.54066980e+00
-2.96365023e-01 -9.22944427e-01 -5.10150492e-01 5.55064201e-01
-2.71186628e-03 -4.80210930e-01 6.02817655e-01 6.55341089e-01
-3.55999202e-01 -8.72554660e-01 -7.80611038e-01 -1.02473366e+00
3.91926467e-01 -5.90959489e-01 6.59080744e-01 6.56906247e-01
2.97225147e-01 -1.43066877e-02 -7.12549746e-01 7.14274719e-02
6.59178495e-01 1.52451051e-02 5.82018197e-01 -1.11289847e+00
-4.61469680e-01 -5.18559694e-01 4.62588698e-01 -1.02994692e+00
8.80669713e-01 -7.49593258e-01 7.34322727e-01 -1.10839260e+00
1.39125343e-02 -6.94285393e-01 -4.34119105e-01 1.00269949e+00
-1.91584811e-01 -7.98092425e-01 7.71652833e-02 -2.02963352e-01
-6.70973182e-01 8.26649606e-01 1.64615178e+00 -4.05934542e-01
-2.10547894e-01 4.52621914e-02 -7.68626928e-01 7.12273419e-01
8.60337377e-01 -3.85803461e-01 -8.93567204e-01 -2.71810025e-01
3.18189204e-01 3.20018232e-01 3.01123530e-01 -7.90278018e-01
-1.44265592e-01 -1.06664038e+00 2.61957571e-02 -1.32412508e-01
1.53182551e-01 -9.99746740e-01 -2.69279301e-01 9.12180483e-01
-7.44278669e-01 -1.13891542e-01 3.41597110e-01 8.01979482e-01
2.31620058e-01 -7.06821233e-02 7.39545643e-01 -2.71027952e-01
-9.98505473e-01 6.81022942e-01 -4.32743043e-01 3.14106494e-01
1.22620976e+00 2.15540051e-01 -3.43653083e-01 -2.43934721e-01
-5.52801490e-01 8.65376413e-01 3.43290478e-01 7.34417260e-01
6.79503441e-01 -9.84131515e-01 -3.26864570e-01 3.21581274e-01
2.18063787e-01 2.93579996e-01 1.48573846e-01 6.49832487e-01
-1.76305890e-01 1.60218015e-01 -6.90791070e-01 -2.74216175e-01
-6.51197493e-01 9.73901391e-01 1.39885828e-01 -4.52959299e-01
-7.42833257e-01 8.20730805e-01 2.04203576e-01 -5.08926034e-01
6.98644519e-01 -6.81195498e-01 -1.48669928e-01 -4.58129376e-01
2.65519887e-01 1.69245854e-01 -4.87851620e-01 1.02639161e-01
-1.65315345e-01 7.70686418e-02 -1.08128764e-01 -1.53625429e-01
1.36334383e+00 5.17657042e-01 2.12386206e-01 -2.70641856e-02
5.43206394e-01 -3.75723511e-01 -2.26102304e+00 8.58684853e-02
2.45429531e-01 -5.18949628e-01 -3.01758796e-01 -9.44111526e-01
-5.74863493e-01 7.17291355e-01 1.64007589e-01 -1.97370604e-01
1.03401875e+00 -2.52634853e-01 6.87728763e-01 7.55072773e-01
1.00794280e+00 -1.54368210e+00 3.85994822e-01 8.91709507e-01
9.49807584e-01 -1.32743609e+00 -1.07028745e-01 1.21440262e-01
-8.73973489e-01 8.60462904e-01 1.23091793e+00 -2.77622789e-01
3.02815288e-01 4.47804362e-01 -3.18901002e-01 1.46032006e-01
-1.08313620e+00 -1.42269775e-01 -4.82476428e-02 9.15372908e-01
-2.27777570e-01 1.70134917e-01 -1.13356024e-01 4.90769327e-01
-2.55414620e-02 -1.50467677e-03 2.47254282e-01 1.23378801e+00
-3.50690633e-01 -1.24557710e+00 -4.93115410e-02 3.94981444e-01
-3.31397712e-01 2.17258304e-01 -1.17078170e-01 8.41781139e-01
1.61754459e-01 6.56441212e-01 -1.70362264e-01 -4.82397467e-01
4.00841564e-01 -1.97923645e-01 9.43310440e-01 -7.95726418e-01
-7.01215446e-01 -3.24256301e-01 2.06551120e-01 -1.20341265e+00
-1.14714727e-02 -5.15668035e-01 -1.58238018e+00 -1.29190594e-01
3.66471976e-01 1.36610463e-01 2.12348595e-01 1.01785541e+00
3.60397287e-02 5.90084851e-01 4.46809798e-01 -7.69290745e-01
-1.55210495e+00 -6.57895684e-01 -4.19026434e-01 6.56980455e-01
4.88133550e-01 -1.09648931e+00 -1.38521403e-01 -3.52486938e-01]
|
[4.207458019256592, 1.6722207069396973]
|
4c3b8795-12f9-4f95-81eb-18d5ca42d339
|
pooled-motion-features-for-first-person
|
1412.6505
| null |
http://arxiv.org/abs/1412.6505v2
|
http://arxiv.org/pdf/1412.6505v2.pdf
|
Pooled Motion Features for First-Person Videos
|
In this paper, we present a new feature representation for first-person
videos. In first-person video understanding (e.g., activity recognition), it is
very important to capture both entire scene dynamics (i.e., egomotion) and
salient local motion observed in videos. We describe a representation framework
based on time series pooling, which is designed to abstract
short-term/long-term changes in feature descriptor elements. The idea is to
keep track of how descriptor values are changing over time and summarize them
to represent motion in the activity video. The framework is general, handling
any types of per-frame feature descriptors including conventional motion
descriptors like histogram of optical flows (HOF) as well as appearance
descriptors from more recent convolutional neural networks (CNN). We
experimentally confirm that our approach clearly outperforms previous feature
representations including bag-of-visual-words and improved Fisher vector (IFV)
when using identical underlying feature descriptors. We also confirm that our
feature representation has superior performance to existing state-of-the-art
features like local spatio-temporal features and Improved Trajectory Features
(originally developed for 3rd-person videos) when handling first-person videos.
Multiple first-person activity datasets were tested under various settings to
confirm these findings.
|
['Brandon Rothrock', 'M. S. Ryoo', 'Larry Matthies']
|
2014-12-19
|
pooled-motion-features-for-first-person-1
|
http://openaccess.thecvf.com/content_cvpr_2015/html/Ryoo_Pooled_Motion_Features_2015_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2015/papers/Ryoo_Pooled_Motion_Features_2015_CVPR_paper.pdf
|
cvpr-2015-6
|
['activity-recognition-in-videos']
|
['computer-vision']
|
[-7.52354935e-02 -8.03905249e-01 -3.01438749e-01 -2.48957440e-01
-2.64614135e-01 -6.19595885e-01 9.91657197e-01 1.32781804e-01
-4.83742326e-01 4.49218124e-01 7.33004391e-01 5.33963323e-01
-2.46125117e-01 -5.40487409e-01 -5.56993902e-01 -6.87010348e-01
-7.70173550e-01 -4.22526717e-01 3.41643929e-01 -1.33761898e-01
3.63819480e-01 8.21179271e-01 -1.86384237e+00 4.05903578e-01
2.52396949e-02 9.30671513e-01 -2.19976783e-01 9.62684035e-01
2.77717620e-01 1.16164887e+00 -5.96728444e-01 6.24921825e-03
9.85514373e-02 -3.66774857e-01 -7.72432745e-01 3.04525346e-01
8.82460415e-01 -5.53786933e-01 -1.06270242e+00 6.16312325e-01
2.42807209e-01 6.76106155e-01 7.28328288e-01 -1.35074174e+00
-6.14811182e-01 -2.31361389e-01 -2.14648858e-01 1.00817263e+00
1.08239615e+00 2.48023570e-01 9.55292702e-01 -9.53638375e-01
9.02732551e-01 1.33580446e+00 7.03059256e-01 2.54850954e-01
-8.13396692e-01 -7.08994865e-02 2.73561776e-01 7.44058132e-01
-1.37652779e+00 -4.09591734e-01 5.84621608e-01 -7.91792810e-01
1.32720637e+00 2.01444566e-01 1.05889630e+00 1.38886750e+00
6.76940322e-01 1.03928459e+00 5.09463727e-01 -1.66693658e-01
-1.11042656e-01 -2.83054233e-01 1.38878152e-01 8.08417976e-01
1.34947881e-01 6.06084391e-02 -8.83121848e-01 -2.25602105e-01
9.29212272e-01 5.34739733e-01 -4.22596425e-01 -5.73064387e-01
-1.71387959e+00 7.61883020e-01 1.20911613e-01 5.79299331e-01
-6.29678130e-01 5.57973564e-01 7.31721163e-01 3.09798896e-01
3.26134890e-01 1.26799196e-03 -1.68892503e-01 -8.34501743e-01
-9.35854197e-01 5.24708509e-01 7.49613047e-01 7.35714376e-01
7.47430980e-01 9.24874991e-02 -6.93323195e-01 5.31707585e-01
-1.65862799e-01 2.09241316e-01 7.79784024e-01 -1.14013267e+00
1.92590550e-01 4.23198521e-01 1.51018858e-01 -1.40964162e+00
-3.73916835e-01 -7.63080418e-02 -7.43068397e-01 -4.98516224e-02
3.91809940e-01 2.40726694e-01 -6.29375339e-01 1.56661057e+00
5.38632832e-02 4.99324381e-01 -1.42536789e-01 9.36729252e-01
7.79526830e-01 7.44275510e-01 -6.87856376e-02 -2.75445253e-01
1.42642856e+00 -1.00036645e+00 -8.21443319e-01 1.98311895e-01
6.02526844e-01 -4.79313910e-01 4.87699300e-01 2.00303987e-01
-1.03839874e+00 -1.09233582e+00 -7.88067877e-01 -3.02934498e-02
-6.43817544e-01 3.42234671e-02 6.68893576e-01 4.40790832e-01
-1.14083707e+00 9.80496824e-01 -8.60874355e-01 -8.95911098e-01
2.56774366e-01 2.80426323e-01 -8.67614806e-01 -2.76389066e-02
-1.10826433e+00 4.79737818e-01 1.89234316e-01 -1.33832067e-01
-1.05495632e+00 -4.86913085e-01 -1.20543206e+00 1.06745400e-01
1.30025059e-01 -6.67756557e-01 9.98965681e-01 -9.75076437e-01
-1.27077639e+00 6.76805794e-01 -6.09076202e-01 -3.89232457e-01
2.86391497e-01 -4.34301645e-01 -4.57071006e-01 6.92261338e-01
1.70897860e-02 5.28878391e-01 1.00931156e+00 -4.68696088e-01
-6.00781620e-01 -8.61652195e-02 3.38954091e-01 1.49221331e-01
-4.26063478e-01 2.43668735e-01 -2.86112398e-01 -8.85162711e-01
-3.52485210e-01 -8.93528521e-01 1.50282159e-01 2.87916958e-01
1.97967738e-01 -4.47306544e-01 1.17688179e+00 -5.07937491e-01
1.24735069e+00 -2.10307550e+00 2.42173299e-01 -9.31131840e-02
8.04015473e-02 2.98948884e-01 -1.34592652e-01 7.31267035e-01
-2.52900243e-01 -9.76414233e-02 1.60218194e-01 -1.28536940e-01
-2.08865196e-01 2.35979334e-01 -8.12346339e-02 9.49741244e-01
3.36678386e-01 1.02177417e+00 -1.02539062e+00 -4.49130356e-01
5.49346089e-01 7.79991806e-01 -5.05448639e-01 1.01749897e-01
4.75784898e-01 4.66652304e-01 -4.47375625e-01 6.81658864e-01
2.48963073e-01 -1.56512901e-01 -2.16158658e-01 -3.45567971e-01
-2.71445841e-01 -9.47328210e-02 -1.24692416e+00 1.73579586e+00
-1.03010453e-01 1.24762344e+00 -5.21131098e-01 -1.01802981e+00
5.19328952e-01 4.05426294e-01 1.01593101e+00 -3.86975557e-01
-1.77252162e-02 -3.10304075e-01 -3.17076683e-01 -7.65765786e-01
6.97990835e-01 4.59217817e-01 -7.92375058e-02 1.87271357e-01
4.79552567e-01 6.06616497e-01 5.90439737e-01 1.31966352e-01
1.26263130e+00 2.86282122e-01 6.75221920e-01 -2.69288957e-01
9.45086598e-01 -3.30204159e-01 4.24420029e-01 7.53893912e-01
-7.06271172e-01 6.69611990e-01 3.69247526e-01 -8.82903337e-01
-8.80236983e-01 -9.54720438e-01 1.77924722e-01 1.09641409e+00
9.80569720e-02 -8.39989364e-01 -4.27759767e-01 -5.00849426e-01
4.81427684e-02 -5.46326227e-02 -8.39156926e-01 -2.04741269e-01
-8.09167802e-01 -2.60191053e-01 3.79708856e-01 8.11692715e-01
5.12014389e-01 -1.13900232e+00 -9.32229042e-01 3.84001791e-01
-2.36531049e-01 -1.20272171e+00 -7.75804341e-01 -4.42329288e-01
-6.63523316e-01 -1.23061061e+00 -9.82575655e-01 -5.09704530e-01
2.07250938e-01 6.47940278e-01 8.96237969e-01 -9.44555402e-02
-7.05308616e-01 1.34630477e+00 -5.87845743e-01 2.52538919e-01
2.08209917e-01 -3.98923725e-01 3.47418010e-01 4.47939664e-01
5.50796330e-01 -3.76942754e-01 -9.25412655e-01 2.23465577e-01
-9.31959331e-01 -5.94734490e-01 1.25251725e-01 7.67319202e-01
3.69624197e-01 -1.70645583e-02 -1.34899437e-01 7.77365267e-03
5.56485355e-01 -4.03531522e-01 4.28607687e-02 2.37083972e-01
2.13208914e-01 -2.49638692e-01 5.52723765e-01 -7.47205913e-01
-7.05237985e-01 -1.03489824e-01 3.26926380e-01 -7.75091350e-01
-3.98112237e-01 2.26912424e-01 1.97482750e-01 -2.10586190e-01
3.15620512e-01 6.60689473e-01 -2.37780437e-01 -2.81923383e-01
2.80852109e-01 3.98534685e-01 5.29252529e-01 -2.94048339e-01
5.83216965e-01 9.95147049e-01 1.72217578e-01 -1.34722173e+00
-4.53532457e-01 -1.18648279e+00 -1.14782262e+00 -5.50264359e-01
1.14099324e+00 -1.06453896e+00 -1.01006079e+00 7.49215364e-01
-1.12805820e+00 1.72538429e-01 -3.83052409e-01 8.67441118e-01
-9.89238024e-01 7.21974552e-01 -6.63737178e-01 -7.46203959e-01
1.46880765e-02 -8.58561993e-01 1.24842250e+00 3.06162477e-01
-3.41867983e-01 -1.30277467e+00 4.00476694e-01 -4.82024476e-02
4.22955036e-01 5.53070426e-01 2.10351288e-01 -4.32200283e-01
-5.07565022e-01 -3.31253886e-01 3.69502441e-03 3.06141794e-01
3.51586044e-01 2.71262020e-01 -8.35429549e-01 -6.59446776e-01
-2.28004336e-01 -5.20872436e-02 1.08212078e+00 7.08539665e-01
1.16398120e+00 -2.67615408e-01 -4.13937122e-01 6.63399518e-01
1.19928312e+00 1.61278248e-01 7.01983869e-01 3.50021333e-01
7.16167569e-01 4.70802337e-01 5.44286907e-01 7.92292356e-01
2.40180999e-01 9.57441509e-01 1.40429556e-01 2.02568814e-01
-1.05795041e-01 -3.56149748e-02 8.55839372e-01 4.93997455e-01
-6.62833393e-01 -2.60968775e-01 -4.42146868e-01 7.27075517e-01
-2.03637314e+00 -1.87606263e+00 1.15238763e-02 1.97574270e+00
3.86390574e-02 -2.33867273e-01 5.37881732e-01 2.73959991e-02
6.86749101e-01 7.11254776e-01 -2.30678245e-01 -4.46769089e-01
-1.77491888e-01 -6.11348040e-02 2.10990742e-01 1.06510930e-01
-1.66823530e+00 6.77264214e-01 6.59058666e+00 4.69647914e-01
-9.53928351e-01 8.60932842e-02 5.16272746e-02 -1.88512355e-01
4.05231446e-01 -1.51560172e-01 -6.43387794e-01 3.43272507e-01
1.02488327e+00 -2.59302258e-01 1.67108133e-01 8.22572052e-01
3.69969606e-01 -2.44380590e-02 -1.23250437e+00 1.29749405e+00
5.14426589e-01 -1.34742534e+00 2.99188673e-01 1.00663602e-01
5.48220932e-01 -2.78938562e-01 -1.17341101e-01 2.81543165e-01
-4.17225003e-01 -8.88926387e-01 7.00906396e-01 1.06392360e+00
4.94649053e-01 -6.68107688e-01 6.88790023e-01 -6.39349744e-02
-1.93155718e+00 -2.79584736e-01 -4.20130640e-01 -2.12827116e-01
3.26607108e-01 2.23672420e-01 -1.70931324e-01 6.43239975e-01
9.28532481e-01 1.60113275e+00 -6.22791767e-01 1.22127211e+00
3.16512167e-01 1.98353529e-01 1.00213792e-02 8.37238207e-02
5.94981313e-01 -1.68924145e-02 7.19132602e-01 1.79586685e+00
4.97793466e-01 5.86903095e-02 2.88217187e-01 4.72968966e-01
3.83014888e-01 -2.53008418e-02 -9.19458210e-01 -2.53867835e-01
-9.56331268e-02 1.11740768e+00 -4.94333118e-01 -6.08655691e-01
-9.14814770e-01 1.43693864e+00 1.89717356e-02 4.42487031e-01
-8.21437418e-01 -4.47477609e-01 1.27752054e+00 1.52518317e-01
7.36115694e-01 -6.16951704e-01 9.97603714e-01 -1.51179934e+00
8.71473998e-02 -4.00504708e-01 5.49189031e-01 -7.35550761e-01
-1.21715367e+00 3.06513965e-01 3.72605026e-01 -1.66840184e+00
-5.80701530e-01 -8.42217743e-01 -7.96682298e-01 3.78523260e-01
-1.42877471e+00 -1.13792741e+00 -6.14668489e-01 1.09874618e+00
8.09145331e-01 -2.55575866e-01 5.65221429e-01 2.98055708e-01
-3.56978118e-01 3.82911384e-01 7.80630633e-02 4.43529129e-01
5.89099884e-01 -9.79225397e-01 3.13767731e-01 8.07783663e-01
3.10209006e-01 8.08188200e-01 3.88010979e-01 -4.42707956e-01
-1.55740464e+00 -9.60933030e-01 7.87115932e-01 -4.82200652e-01
7.09388077e-01 -1.63479701e-01 -7.57128298e-01 7.82341778e-01
1.11080006e-01 4.52714473e-01 7.25519359e-01 -3.40499222e-01
-3.10306907e-01 6.34413213e-02 -8.35173965e-01 3.76732737e-01
1.37050319e+00 -7.16647685e-01 -6.92246974e-01 2.06076980e-01
3.08392048e-01 -1.14107132e-02 -9.42654431e-01 1.74747661e-01
8.77602875e-01 -1.19486237e+00 1.30064988e+00 -9.21761394e-01
2.26373877e-03 -3.29034418e-01 -3.10369760e-01 -9.78816092e-01
-8.50438476e-01 -8.07913721e-01 -7.06321239e-01 7.60167122e-01
-6.09579802e-01 -3.76038492e-01 4.53068882e-01 1.26492769e-01
-2.37952266e-02 -4.02250022e-01 -1.13368559e+00 -1.16735840e+00
-3.71794194e-01 -2.01583415e-01 1.55619964e-01 8.43493223e-01
5.49314357e-02 -2.36721754e-01 -6.41161919e-01 -1.10970862e-01
4.79493499e-01 -6.75591528e-02 8.25112939e-01 -1.13182008e+00
7.35471584e-03 -4.16723430e-01 -1.48218131e+00 -1.12232196e+00
2.83296406e-01 -5.95849872e-01 -4.49586391e-01 -1.31750226e+00
5.15296400e-01 6.86355650e-01 -5.46359181e-01 2.60156095e-01
2.21276898e-02 2.00413093e-01 2.30322614e-01 2.37044498e-01
-9.03037786e-01 6.75647318e-01 1.14120626e+00 -1.29934400e-01
-6.38448521e-02 -1.13486707e-01 6.94593601e-03 6.81275725e-01
3.98302913e-01 -1.57292753e-01 -2.20631167e-01 7.97099397e-02
-2.61047333e-01 -9.29821283e-02 1.04095101e+00 -1.52478850e+00
2.48470381e-01 -2.15537086e-01 7.37134218e-01 -6.38525665e-01
6.79383039e-01 -6.23829484e-01 -1.02954833e-02 5.60504794e-01
-1.48849189e-01 3.47203225e-01 -1.92410871e-02 8.64426434e-01
-5.70303857e-01 5.81583567e-02 5.69282889e-01 -3.48743528e-01
-1.35488665e+00 5.45213640e-01 -7.68619895e-01 -2.08586082e-01
1.30639863e+00 -7.55668402e-01 -2.62152672e-01 -7.79026330e-01
-8.20500910e-01 -1.89659968e-01 3.89587969e-01 8.46532404e-01
8.49897206e-01 -1.73122466e+00 -5.99585295e-01 3.03071827e-01
3.72689098e-01 -9.44994032e-01 6.07944071e-01 9.49948072e-01
-4.51108485e-01 9.02474999e-01 -5.22602439e-01 -8.07911575e-01
-1.46302104e+00 6.46713018e-01 4.10435677e-01 -1.22561537e-01
-8.90672386e-01 5.58182538e-01 2.22606435e-01 2.77364016e-01
1.52316973e-01 -4.66964662e-01 -4.74847764e-01 3.99970144e-01
8.97046089e-01 6.67199612e-01 -3.37509423e-01 -1.30171895e+00
-6.86472535e-01 1.10856044e+00 7.45239779e-02 1.77881569e-01
1.27795935e+00 -1.59814015e-01 6.75271079e-02 5.33708751e-01
1.78369248e+00 -3.10026497e-01 -1.40858436e+00 -1.47191808e-01
-1.83504909e-01 -8.77338767e-01 -1.41105980e-01 5.60069643e-02
-8.68622363e-01 8.96562755e-01 7.75013089e-01 1.87161714e-01
9.35770392e-01 -9.35797673e-03 6.99566901e-01 5.33554137e-01
4.40371722e-01 -9.61909831e-01 5.85612893e-01 5.26771247e-01
9.35818017e-01 -1.09022987e+00 2.48267218e-01 8.60348623e-03
-6.52407050e-01 1.62812185e+00 2.81269461e-01 -3.90455008e-01
7.89681792e-01 -4.32613403e-01 -3.75728995e-01 -3.68491024e-01
-6.90194011e-01 -4.86053944e-01 7.55747080e-01 6.53289437e-01
4.16078538e-01 -2.37517223e-01 -2.51910649e-02 1.71436984e-02
2.11967856e-01 3.99848968e-02 6.44290149e-01 1.19573390e+00
-4.18047518e-01 -6.83246613e-01 -2.69207627e-01 1.96943998e-01
-6.00355327e-01 4.15158629e-01 -2.04224691e-01 1.07123780e+00
2.13328362e-01 8.49868000e-01 4.00010228e-01 -2.85272598e-01
3.69107902e-01 5.61712757e-02 6.80805564e-01 -3.72960180e-01
-4.57920462e-01 -1.33537531e-01 -7.49272183e-02 -1.29780674e+00
-1.16524160e+00 -9.63575006e-01 -7.41143286e-01 -3.59783530e-01
1.81407899e-01 -2.59453416e-01 2.59282500e-01 8.22117507e-01
2.62133479e-01 3.85797948e-01 4.93121356e-01 -1.34101593e+00
-2.24582764e-04 -8.25693488e-01 -7.25425482e-01 9.39068437e-01
7.51938641e-01 -1.05098069e+00 -4.07308221e-01 4.20151889e-01]
|
[8.132308006286621, 0.37899357080459595]
|
c13bc75d-9636-43fa-9605-2026b49f4466
|
gesture-to-gesture-translation-in-the-wild
|
1907.05916
| null |
https://arxiv.org/abs/1907.05916v3
|
https://arxiv.org/pdf/1907.05916v3.pdf
|
Gesture-to-Gesture Translation in the Wild via Category-Independent Conditional Maps
|
Recent works have shown Generative Adversarial Networks (GANs) to be particularly effective in image-to-image translations. However, in tasks such as body pose and hand gesture translation, existing methods usually require precise annotations, e.g. key-points or skeletons, which are time-consuming to draw. In this work, we propose a novel GAN architecture that decouples the required annotations into a category label - that specifies the gesture type - and a simple-to-draw category-independent conditional map - that expresses the location, rotation and size of the hand gesture. Our architecture synthesizes the target gesture while preserving the background context, thus effectively dealing with gesture translation in the wild. To this aim, we use an attention module and a rolling guidance approach, which loops the generated images back into the network and produces higher quality images compared to competing works. Thus, our GAN learns to generate new images from simple annotations without requiring key-points or skeleton labels. Results on two public datasets show that our method outperforms state of the art approaches both quantitatively and qualitatively. To the best of our knowledge, no work so far has addressed the gesture-to-gesture translation in the wild by requiring user-friendly annotations.
|
['Yahui Liu', 'Bruno Lepri', 'Marco De Nadai', 'Gloria Zen', 'Nicu Sebe']
|
2019-07-12
| null | null | null | null |
['gesture-to-gesture-translation']
|
['computer-vision']
|
[ 4.37518239e-01 2.14714855e-01 6.83401302e-02 -4.01603699e-01
-7.83784568e-01 -6.77334070e-01 7.06462741e-01 -6.98508978e-01
-3.20900321e-01 5.50333321e-01 8.75462294e-02 -1.16614841e-01
5.31391799e-01 -7.48191655e-01 -1.04623461e+00 -7.96622038e-01
5.61584532e-01 6.75498009e-01 2.86993206e-01 -1.65292040e-01
3.65790091e-02 4.15332764e-01 -1.25991523e+00 8.42189193e-02
6.16172075e-01 7.46953547e-01 -9.09752101e-02 6.94103479e-01
-3.28183062e-02 5.84633887e-01 -7.34580278e-01 -7.37302840e-01
5.42034388e-01 -9.34460580e-01 -6.26108289e-01 2.22381040e-01
6.98944569e-01 -6.67348087e-01 -2.73910165e-01 8.63845229e-01
7.83096671e-01 -1.59388278e-02 7.36018956e-01 -1.18965256e+00
-6.32687688e-01 4.87255871e-01 -7.60405183e-01 -4.68209952e-01
2.65943825e-01 6.01735055e-01 6.71726763e-01 -7.07742572e-01
9.32905853e-01 1.21783864e+00 5.83029747e-01 1.03810525e+00
-1.24834168e+00 -8.62202764e-01 7.35582113e-02 -2.56475955e-01
-1.09425616e+00 -3.22786301e-01 8.25012088e-01 -4.85530645e-01
4.07485485e-01 3.96434814e-01 7.63464808e-01 1.67891312e+00
-6.53715506e-02 8.18419933e-01 1.15259039e+00 -4.00916129e-01
2.11267963e-01 -3.79175514e-01 -6.76684320e-01 5.80260336e-01
-2.45848626e-01 1.16033860e-01 -4.31520253e-01 2.84392804e-01
1.32925022e+00 5.64916991e-02 -1.77625477e-01 -6.24503136e-01
-1.30028582e+00 6.20291829e-01 6.11874342e-01 1.63268328e-01
-2.93497562e-01 5.94676077e-01 2.50426501e-01 -8.15596431e-03
3.05996746e-01 1.76302284e-01 -2.91637063e-01 -1.15258098e-01
-9.89398181e-01 3.97367388e-01 6.41737044e-01 9.93078709e-01
5.08085608e-01 9.85069126e-02 -5.01810014e-01 5.11268973e-01
2.02048972e-01 7.63464928e-01 3.07851344e-01 -7.72383273e-01
6.31778955e-01 5.73137999e-01 1.45421997e-01 -7.99901247e-01
-2.15583488e-01 -2.76775986e-01 -9.58399475e-01 5.43637753e-01
7.41676986e-01 -3.13827872e-01 -1.33578098e+00 1.87548542e+00
4.47925031e-01 2.44613439e-01 -1.59378022e-01 1.33289039e+00
7.72822976e-01 3.93103778e-01 1.53351739e-01 3.34291935e-01
1.37990463e+00 -1.22219896e+00 -6.81244791e-01 -3.49980414e-01
4.50566076e-02 -8.44446003e-01 1.44893515e+00 1.52469039e-01
-1.21880352e+00 -5.01027763e-01 -8.43364418e-01 -1.97336867e-01
-6.86397851e-02 3.47428292e-01 5.24460793e-01 6.35528445e-01
-9.12201464e-01 3.39609414e-01 -1.07029831e+00 -2.97446370e-01
4.43472207e-01 3.17101777e-01 -2.66884208e-01 1.24375045e-01
-8.02963316e-01 6.86182916e-01 1.02844656e-01 2.44178504e-01
-9.43022192e-01 -5.81014693e-01 -7.63333976e-01 -1.20611064e-01
5.37684202e-01 -1.13498008e+00 1.39647353e+00 -1.23205066e+00
-2.02696753e+00 8.55403304e-01 9.48064849e-02 -2.76992351e-01
1.26150525e+00 -3.22958291e-01 2.22522527e-01 2.36191438e-03
-1.38789743e-01 1.02374005e+00 1.21855748e+00 -1.32822275e+00
-3.05138856e-01 -3.84766817e-01 2.41127953e-01 2.54157215e-01
-4.33237106e-02 4.87085339e-03 -9.56016243e-01 -1.09933329e+00
-1.20597882e-02 -1.18184578e+00 -1.50869980e-01 1.93614736e-01
-6.24256432e-01 3.79140526e-02 8.59803200e-01 -8.78467977e-01
8.46421540e-01 -1.96136403e+00 4.97246116e-01 1.21191956e-01
1.31704107e-01 1.95486680e-01 -1.21730208e-01 1.80809528e-01
1.77241519e-01 5.79446703e-02 -3.94207954e-01 -6.14224434e-01
1.23092651e-01 2.68212736e-01 -1.95189312e-01 3.56844634e-01
2.59459168e-01 1.47122276e+00 -8.64334702e-01 -4.83342052e-01
3.92957002e-01 7.87014842e-01 -5.96281946e-01 4.35663104e-01
-5.93768358e-01 1.05471611e+00 -3.32271785e-01 5.88985503e-01
5.45166135e-01 -6.26450330e-02 2.16784045e-01 -2.65621483e-01
1.97862387e-01 2.72041485e-02 -8.74063671e-01 2.15551758e+00
-5.83880782e-01 4.20742750e-01 9.67226550e-02 -6.18838966e-01
9.36022282e-01 2.85643131e-01 2.47198835e-01 -5.15932381e-01
2.93262452e-01 1.64213821e-01 -8.75337645e-02 -4.29443687e-01
1.21797584e-01 -7.06301332e-02 -7.52433687e-02 5.24394810e-01
-1.01662271e-01 -4.40948933e-01 -8.88083726e-02 -1.45351291e-01
9.06135321e-01 8.28469157e-01 -6.44930080e-02 2.34297097e-01
2.78458834e-01 -1.59268424e-01 2.98303545e-01 4.35031474e-01
2.21117958e-01 1.15285730e+00 4.22103584e-01 -4.76026952e-01
-1.11223710e+00 -9.73790050e-01 4.93716419e-01 1.13679016e+00
1.15913771e-01 -1.22949712e-01 -1.25773811e+00 -8.64756644e-01
-3.11861247e-01 4.07938689e-01 -8.08691740e-01 1.47544429e-01
-8.84583771e-01 -3.92656475e-01 6.95470750e-01 8.15833390e-01
5.83531797e-01 -1.34939384e+00 -9.17702913e-01 5.53658567e-02
-2.60429204e-01 -9.77580249e-01 -9.30859447e-01 -2.01805323e-01
-7.77891994e-01 -8.31860781e-01 -1.27419055e+00 -7.59402514e-01
9.68305647e-01 -2.91309386e-01 1.04416060e+00 3.50594521e-03
-2.93488055e-01 1.21438995e-01 -3.95392776e-01 -3.59940380e-01
-4.21627522e-01 1.63109794e-01 -3.67783964e-01 2.03254655e-01
-3.49384509e-02 -7.03550518e-01 -9.42083657e-01 5.42081296e-01
-1.00373816e+00 4.51424688e-01 8.78160655e-01 8.74630451e-01
6.62231147e-01 -5.44640779e-01 1.22583099e-01 -7.90158391e-01
3.91537488e-01 1.15228765e-01 -4.62436080e-01 1.88333839e-01
-2.19584599e-01 1.46179572e-01 5.28603494e-01 -5.96354544e-01
-1.00568318e+00 5.59000432e-01 -2.32505441e-01 -5.59555233e-01
-2.86532342e-01 -7.48150870e-02 -4.65178728e-01 -6.90936595e-02
6.73656523e-01 2.77547002e-01 3.37985950e-03 -5.26343405e-01
6.01832688e-01 4.49022949e-01 8.94595742e-01 -6.45698369e-01
1.06308699e+00 4.53046590e-01 4.26177308e-02 -1.90859944e-01
-6.09213829e-01 -7.13497698e-02 -9.15902734e-01 -1.68007821e-01
1.12783766e+00 -7.33388126e-01 -5.37861049e-01 7.32100904e-01
-1.20730388e+00 -7.87387252e-01 -3.48063380e-01 1.40229970e-01
-8.48023891e-01 1.48397952e-01 -5.48993886e-01 -4.73805755e-01
-5.70236802e-01 -1.23544931e+00 1.52569509e+00 2.42592201e-01
-1.82704747e-01 -6.76941276e-01 -6.02236204e-02 3.88249904e-01
5.43882549e-01 7.75742114e-01 3.66095304e-01 -1.82468534e-01
-7.47355640e-01 -1.19903781e-01 -1.59888074e-01 1.43393651e-01
1.30624801e-01 -1.68294668e-01 -9.84026432e-01 -2.30507955e-01
-3.94884288e-01 -3.13315630e-01 6.59064591e-01 2.37221390e-01
1.26285386e+00 -4.63325948e-01 -1.65603012e-01 8.81290793e-01
1.01451254e+00 9.83383507e-02 8.82427871e-01 1.55849755e-01
1.06976342e+00 4.31922793e-01 5.49583375e-01 2.36739054e-01
2.27605522e-01 1.17111409e+00 5.76345623e-01 -4.79631692e-01
-6.35942340e-01 -6.51826918e-01 3.25444281e-01 2.85405457e-01
-5.55909753e-01 -3.04388702e-01 -6.97669804e-01 3.50858450e-01
-1.82740605e+00 -6.90146744e-01 1.28123522e-01 1.96295393e+00
9.94345963e-01 -1.25749871e-01 2.01848373e-01 -9.80727971e-02
6.48134291e-01 5.11788540e-02 -7.67339528e-01 -6.71616346e-02
1.48314521e-01 5.20242333e-01 3.91473353e-01 3.31006259e-01
-1.07426834e+00 1.20000207e+00 5.16527748e+00 7.25512505e-01
-1.36955726e+00 2.40776435e-01 4.98210371e-01 -2.26627067e-01
-1.77920550e-01 -1.22696288e-01 -3.79421085e-01 3.53014320e-01
2.33812720e-01 4.32610244e-01 4.59957987e-01 7.49193788e-01
1.48794070e-01 2.70428926e-01 -1.09768760e+00 1.13178062e+00
1.92043260e-01 -1.03223252e+00 1.91179574e-01 1.08632080e-01
8.16402256e-01 -2.95117915e-01 1.01791568e-01 2.67804693e-02
3.47116262e-01 -1.11803937e+00 1.16552389e+00 4.83690411e-01
1.23846161e+00 -5.89114487e-01 6.95886791e-01 2.10637897e-01
-9.63198423e-01 4.18081880e-01 1.06456541e-01 2.01350138e-01
3.06033134e-01 1.22306362e-01 -7.71598697e-01 3.74104023e-01
5.73614001e-01 3.58014524e-01 -5.13724148e-01 6.77418292e-01
-7.82726765e-01 5.09105980e-01 -2.52015322e-01 1.62365437e-01
2.41350278e-01 -1.40483201e-01 3.49246085e-01 1.16072023e+00
3.75258833e-01 -9.12749469e-02 8.70530680e-02 9.82236147e-01
-2.74436027e-01 6.08456023e-02 -3.07573140e-01 8.91916379e-02
1.35720417e-01 1.15182829e+00 -9.12457585e-01 -3.00781876e-01
1.21484390e-02 1.57565403e+00 4.41025235e-02 3.32747668e-01
-9.83674228e-01 -2.81518757e-01 4.47632283e-01 3.54258657e-01
4.87989962e-01 -2.12673932e-01 -4.09270614e-01 -1.15952456e+00
2.82459795e-01 -9.45800364e-01 2.68082116e-02 -7.79150367e-01
-1.01554477e+00 6.55462742e-01 -1.52159333e-01 -1.14850378e+00
-6.25120044e-01 -4.29868132e-01 -5.18427551e-01 8.17689359e-01
-1.07176840e+00 -1.70656860e+00 -6.78085208e-01 6.85408533e-01
6.01805627e-01 1.19453996e-01 7.86079943e-01 2.80195832e-01
-3.67543429e-01 9.12776053e-01 -3.04986835e-01 5.39555490e-01
9.28823590e-01 -1.18340838e+00 7.98656285e-01 7.60654747e-01
3.23216259e-01 5.36769986e-01 5.80977798e-01 -6.31544173e-01
-1.34023190e+00 -1.01157165e+00 4.69572365e-01 -5.74837387e-01
2.94054478e-01 -5.78108609e-01 -5.31896830e-01 6.87425196e-01
1.83659270e-01 8.54093507e-02 1.77187413e-01 -3.44555438e-01
-3.15362215e-01 -3.28667872e-02 -1.12354755e+00 7.90794015e-01
1.34566939e+00 -2.41072983e-01 -3.09976190e-01 1.54268682e-01
4.50191766e-01 -1.05976653e+00 -5.64376533e-01 3.52554500e-01
8.84679556e-01 -9.88116920e-01 1.09207201e+00 -4.56668526e-01
6.85085237e-01 -5.70715189e-01 1.61467746e-01 -1.29041147e+00
9.79068428e-02 -7.96918690e-01 -5.12180440e-02 1.22299373e+00
1.99636474e-01 -3.18344057e-01 9.60261762e-01 6.07767999e-01
-6.16497826e-03 -6.05010390e-01 -7.90658236e-01 -5.63224733e-01
-4.26918678e-02 -2.25190759e-01 8.78526390e-01 7.02213168e-01
-6.05933309e-01 1.92278922e-01 -7.55386591e-01 -8.43569934e-02
4.60390240e-01 2.49032483e-01 1.41712296e+00 -7.98936188e-01
-4.86442804e-01 -4.45912480e-01 -3.22530299e-01 -1.22878325e+00
6.50182413e-03 -6.60536587e-01 1.74294755e-01 -1.47672594e+00
2.79521174e-03 -4.81428564e-01 3.27496648e-01 7.98985124e-01
-1.83776125e-01 8.20915878e-01 4.13215637e-01 1.75699845e-01
-3.38307887e-01 4.46122766e-01 1.73557675e+00 -2.36210033e-01
-2.76938021e-01 9.61604565e-02 -5.07158518e-01 7.75255799e-01
9.68507648e-01 -4.05370057e-01 -2.63509572e-01 -6.76402926e-01
-4.98417718e-03 2.16125958e-02 6.91033602e-01 -8.73238802e-01
-2.42740940e-02 -1.98805854e-01 4.88140017e-01 -3.56912524e-01
4.15353477e-01 -7.99024284e-01 4.67024565e-01 5.33429027e-01
-2.44642839e-01 -7.06834421e-02 -6.11168630e-02 4.79173273e-01
-4.32200953e-02 -1.56127550e-02 6.04795992e-01 -2.33771861e-01
-3.32565516e-01 3.74309391e-01 2.24858612e-01 -4.91428226e-02
1.05598760e+00 -1.16704628e-01 9.73660406e-03 -6.00439072e-01
-6.91561222e-01 -1.29715681e-01 7.37378538e-01 6.12333894e-01
4.09619093e-01 -1.32027459e+00 -8.11388135e-01 2.57459491e-01
-4.89800461e-02 4.91133809e-01 6.57304749e-02 7.23919749e-01
-7.98747778e-01 1.22681476e-01 -3.21168840e-01 -7.34166384e-01
-1.27878296e+00 3.93201500e-01 2.80994385e-01 -2.75238782e-01
-8.49574149e-01 7.24650621e-01 3.60546052e-01 -4.90557790e-01
3.35032374e-01 -4.46526140e-01 2.85804510e-01 -1.46234259e-01
4.35367048e-01 1.32613495e-01 -1.03868842e-01 -6.15471542e-01
-1.69866666e-01 1.05025244e+00 2.99771249e-01 -4.05840248e-01
1.17007685e+00 9.55229849e-02 8.73891190e-02 7.62432963e-02
7.43495047e-01 1.72321334e-01 -1.81008816e+00 5.18481014e-03
-4.97011006e-01 -6.49494112e-01 -3.68264556e-01 -1.11836851e+00
-1.49890184e+00 9.87553716e-01 5.49495518e-01 -3.54202598e-01
1.17560363e+00 -4.71841311e-03 9.36161458e-01 -1.41465468e-02
5.39025366e-01 -6.95440829e-01 2.43790507e-01 2.19326168e-01
1.17669332e+00 -1.02629876e+00 -3.87289613e-01 -4.04683411e-01
-6.65359557e-01 9.64915276e-01 6.65686071e-01 -1.55539989e-01
8.48351866e-02 4.85898137e-01 3.99067461e-01 7.47899190e-02
-8.14832523e-02 -1.51460513e-01 5.13399184e-01 7.39053309e-01
2.95298249e-01 2.04866767e-01 -3.21457595e-01 3.74583930e-01
-5.03378510e-01 1.66002721e-01 3.05709224e-02 7.15260625e-01
2.82889545e-01 -1.43964636e+00 -5.53166449e-01 4.35503051e-02
-4.27856356e-01 -1.07964110e-02 -7.36126244e-01 7.86357760e-01
2.66663760e-01 4.98591363e-01 -1.36118405e-03 -4.18153286e-01
3.99090827e-01 1.65567189e-01 7.74038911e-01 -4.98323470e-01
-7.85207093e-01 2.32183710e-01 -2.20999941e-01 -6.88779175e-01
-6.43222511e-01 -5.52659452e-01 -1.18042886e+00 -1.74566999e-01
-1.98436275e-01 -3.95508468e-01 7.64964640e-01 7.03372300e-01
2.55896568e-01 6.20812893e-01 1.70370787e-01 -1.35240483e+00
-4.69949424e-01 -1.18123639e+00 -1.90768078e-01 7.68169284e-01
1.15679234e-01 -5.57236314e-01 -1.11685894e-01 4.51378971e-01]
|
[11.741859436035156, -0.6981605887413025]
|
b94a5630-af83-432e-8b71-7e7370692aff
|
domain-adaptation-for-time-series-under
|
2302.03133
| null |
https://arxiv.org/abs/2302.03133v2
|
https://arxiv.org/pdf/2302.03133v2.pdf
|
Domain Adaptation for Time Series Under Feature and Label Shifts
|
Unsupervised domain adaptation (UDA) enables the transfer of models trained on source domains to unlabeled target domains. However, transferring complex time series models presents challenges due to the dynamic temporal structure variations across domains. This leads to feature shifts in the time and frequency representations. Additionally, the label distributions of tasks in the source and target domains can differ significantly, posing difficulties in addressing label shifts and recognizing labels unique to the target domain. Effectively transferring complex time series models remains a formidable problem. We present Raincoat, the first model for both closed-set and universal domain adaptation on complex time series. Raincoat addresses feature and label shifts by considering both temporal and frequency features, aligning them across domains, and correcting for misalignments to facilitate the detection of private labels. Additionally, Raincoat improves transferability by identifying label shifts in target domains. Our experiments with 5 datasets and 13 state-of-the-art UDA methods demonstrate that Raincoat can improve transfer learning performance by up to 16.33% and can handle both closed-set and universal domain adaptation.
|
['Marinka Zitnik', 'Theodoros Tsiligkaridis', 'Consuelo Cuevas', 'Teddy Koker', 'Owen Queen', 'Huan He']
|
2023-02-06
| null | null | null | null |
['universal-domain-adaptation']
|
['computer-vision']
|
[ 3.98824275e-01 -5.12234628e-01 -2.73736626e-01 -5.34114003e-01
-8.79637539e-01 -1.24474907e+00 4.41719323e-01 1.32566705e-01
-1.90478176e-01 7.16993809e-01 -1.03263870e-01 -1.80770665e-01
2.79388460e-03 -6.10919237e-01 -6.05374813e-01 -6.63809359e-01
-2.63562232e-01 5.43722332e-01 1.94716856e-01 -5.67971878e-02
-1.29040524e-01 4.34764296e-01 -1.37061298e+00 3.87921989e-01
9.21359420e-01 8.61749589e-01 5.04480265e-02 2.62251943e-01
-3.33779633e-01 2.34385476e-01 -8.93377423e-01 3.70163359e-02
2.29084089e-01 -4.52765971e-01 -8.00624669e-01 2.26297993e-02
5.51966190e-01 -1.09477110e-01 -1.63845927e-01 7.03661680e-01
3.77950013e-01 3.03088635e-01 1.07031751e+00 -1.52239501e+00
-8.64368081e-01 9.68327448e-02 -5.47086060e-01 5.30926347e-01
1.65349782e-01 -1.93525419e-01 6.69825256e-01 -8.45406294e-01
7.11318970e-01 1.13953376e+00 9.88576174e-01 6.24961019e-01
-1.43214643e+00 -1.22962201e+00 3.90893430e-01 3.20745856e-01
-1.11364830e+00 -2.11054340e-01 6.12879932e-01 -7.36375749e-01
1.05082083e+00 -1.21232308e-01 2.35587791e-01 1.48405528e+00
-6.67720586e-02 6.04843974e-01 1.15359533e+00 -3.13482761e-01
3.05488378e-01 1.42224422e-02 2.04572991e-01 7.59855956e-02
-1.93917572e-01 8.90392512e-02 -5.79818428e-01 -2.31932670e-01
7.48761654e-01 2.98625141e-01 -1.71906836e-02 -5.32753587e-01
-1.50089693e+00 7.26768434e-01 2.15251580e-01 2.73844838e-01
-2.72770613e-01 -3.89642626e-01 7.04911411e-01 7.90701151e-01
7.51263261e-01 3.88293207e-01 -9.79305744e-01 -1.68637410e-01
-5.94965696e-01 1.76267810e-02 6.12278879e-01 1.23549175e+00
8.64335775e-01 9.75489616e-02 -1.64933771e-01 1.15276396e+00
-1.41894132e-01 8.78684700e-01 8.38500142e-01 -7.25560784e-01
5.43254852e-01 3.50861996e-01 1.27162918e-01 -5.97352743e-01
-2.91027069e-01 -3.48427445e-01 -4.41713095e-01 -1.22990005e-01
7.23692834e-01 -3.41615647e-01 -1.13212264e+00 2.04984021e+00
4.99468148e-01 6.13146007e-01 1.47249877e-01 6.11056924e-01
5.24991870e-01 8.03717852e-01 2.77221262e-01 -3.08177531e-01
1.26862085e+00 -6.83480561e-01 -7.59109378e-01 -2.88656890e-01
7.60174274e-01 -8.46648812e-01 1.07464540e+00 2.87017822e-02
-3.92055511e-01 -6.69107199e-01 -7.85969675e-01 1.15651719e-01
-7.05925405e-01 -1.29778683e-01 1.63940817e-01 3.04666728e-01
-5.25529504e-01 5.88278115e-01 -8.15725505e-01 -7.77102947e-01
3.06052327e-01 2.66757905e-01 -5.21583974e-01 -3.28063905e-01
-1.39421201e+00 8.22862089e-01 2.63793558e-01 -6.67273700e-01
-6.84473574e-01 -1.19307971e+00 -9.41166401e-01 -1.26126617e-01
-1.11741967e-01 -1.61354035e-01 1.49124038e+00 -9.13900733e-01
-1.30586851e+00 9.64975715e-01 -1.70842677e-01 -3.94766003e-01
1.77481905e-01 -1.98492706e-01 -9.94464099e-01 -5.59795424e-02
4.23284918e-01 4.14436072e-01 1.01820779e+00 -7.13993788e-01
-8.19531024e-01 -3.83188963e-01 -6.75822377e-01 1.21986724e-01
-7.22999692e-01 -1.23241216e-01 1.94720353e-03 -1.04760408e+00
-5.42501407e-03 -1.03182518e+00 3.62137318e-01 -1.26467213e-01
2.11951509e-01 -3.34178299e-01 1.40250552e+00 -7.33633816e-01
1.13341510e+00 -2.53328848e+00 -9.50081572e-02 9.72319022e-02
-1.70682251e-01 1.50532737e-01 -4.11429346e-01 4.95511770e-01
-4.32339966e-01 -3.57815206e-01 -2.66261250e-01 -1.27734095e-01
5.73918708e-02 4.61311340e-01 -7.85534084e-01 4.04596239e-01
5.09190857e-01 4.63461280e-01 -1.06867504e+00 -1.98654100e-01
1.05822012e-02 2.07450509e-01 -2.29369417e-01 3.06351304e-01
3.34321037e-02 7.36875772e-01 -2.03846812e-01 6.11482918e-01
6.71067178e-01 -3.50268334e-01 2.21259937e-01 6.16540648e-02
8.08600113e-02 5.21334648e-01 -9.24244821e-01 1.60164630e+00
-4.46544886e-01 7.56897032e-01 -3.35640222e-01 -1.24477017e+00
1.24099910e+00 3.94848138e-01 7.44589031e-01 -8.88404489e-01
-2.26547569e-01 4.17378336e-01 -2.62265652e-01 -3.79035950e-01
2.41875604e-01 -4.29348409e-01 -3.93305123e-01 6.92682028e-01
4.11297888e-01 -2.62852199e-02 -9.82840918e-03 -2.75455505e-01
1.17283821e+00 1.89772099e-01 1.70059428e-01 -4.69368771e-02
9.88967046e-02 1.29378617e-01 8.56839538e-01 4.12231922e-01
-4.20684814e-01 6.10158205e-01 -2.83970963e-02 -4.04488713e-01
-9.15787876e-01 -1.41115057e+00 -3.54478925e-01 1.75394917e+00
-6.68056235e-02 -4.64994460e-03 -2.91057885e-01 -9.85167325e-01
4.86122072e-01 6.62174702e-01 -7.02203393e-01 -5.46909511e-01
-5.50634444e-01 -4.84344840e-01 4.77197379e-01 7.53482878e-01
2.38715261e-01 -9.33782041e-01 -3.63425910e-01 5.47798634e-01
-4.41885680e-01 -1.16914701e+00 -8.64469767e-01 5.62791348e-01
-1.14201808e+00 -8.68439853e-01 -9.46430981e-01 -1.04527605e+00
4.66806322e-01 4.36409414e-01 1.16962421e+00 -8.58526766e-01
-1.59625083e-01 5.53837597e-01 -5.29343545e-01 -5.93923748e-01
-5.51648319e-01 2.55744427e-01 3.37535858e-01 3.86013128e-02
8.97174895e-01 -7.58430302e-01 -3.27305287e-01 7.40264952e-01
-7.88698554e-01 -5.87563813e-01 2.14032218e-01 8.77963305e-01
5.35027027e-01 -8.09538141e-02 1.00230646e+00 -9.99888897e-01
4.92998391e-01 -1.00630903e+00 -3.73698592e-01 2.80721724e-01
-7.15450466e-01 -8.97963122e-02 5.71706235e-01 -1.22212279e+00
-1.08642483e+00 -2.63563022e-02 5.05048096e-01 -5.84129214e-01
-4.06762064e-01 3.55267376e-01 2.73636341e-01 3.13266486e-01
1.14904845e+00 1.66044667e-01 2.37728253e-01 -6.75624549e-01
2.27379054e-01 7.17115223e-01 6.55665755e-01 -7.71124542e-01
8.55498731e-01 3.80292803e-01 -4.35002238e-01 -5.60380459e-01
-8.75241578e-01 -8.13062489e-01 -8.96644413e-01 -1.21612065e-02
4.96415406e-01 -1.20191181e+00 1.38700739e-01 6.40750825e-01
-8.88708472e-01 -5.26338398e-01 -4.89612669e-01 5.99642336e-01
-4.86711085e-01 1.50353864e-01 -4.44227159e-01 -2.13185355e-01
-1.38188794e-01 -5.44718802e-01 9.75964963e-01 4.84793410e-02
-6.30081713e-01 -1.34177029e+00 3.63287359e-01 -8.21551606e-02
4.08806324e-01 2.48593673e-01 1.06039476e+00 -9.69891906e-01
1.76856369e-01 -3.68108088e-03 -7.32345209e-02 2.99423784e-01
7.57335186e-01 -3.43843162e-01 -1.07725334e+00 -6.07021272e-01
-1.19487271e-01 -3.14898998e-01 5.08631229e-01 2.71137446e-01
6.86849236e-01 6.46948814e-02 -4.06052738e-01 4.76001024e-01
9.15980399e-01 5.02222955e-01 1.70618966e-01 4.60395962e-01
4.22762156e-01 6.52042568e-01 1.01618099e+00 6.17718101e-01
2.78993666e-01 5.92331469e-01 -9.36257616e-02 -3.00183203e-02
-1.43127292e-01 -2.60350466e-01 6.11567914e-01 9.52698469e-01
4.26265359e-01 1.76695839e-01 -1.14232874e+00 1.15070701e+00
-1.73957765e+00 -8.23877394e-01 1.23669662e-01 2.20415115e+00
1.21663749e+00 -2.56668389e-01 3.34038138e-01 2.41302550e-02
1.04705989e+00 -2.25821242e-01 -1.09769976e+00 -1.75027296e-01
6.03481457e-02 3.83789033e-01 4.25811350e-01 -1.71795443e-01
-1.36277401e+00 7.00367272e-01 6.71342897e+00 6.18589938e-01
-1.39919806e+00 2.87721723e-01 2.95467023e-02 -2.98271570e-02
8.89525339e-02 -2.72309482e-01 -6.58585191e-01 5.43512881e-01
1.17607117e+00 -4.97854412e-01 3.09359312e-01 8.56941760e-01
-2.76370347e-01 5.48197687e-01 -1.40763271e+00 9.96463776e-01
-2.42924675e-01 -7.97890842e-01 -1.77135706e-01 -5.99982515e-02
1.04179060e+00 2.39451155e-01 2.80245692e-01 7.99837768e-01
4.18824762e-01 -6.77756429e-01 5.16702414e-01 1.01281144e-01
1.17525160e+00 -5.35857558e-01 5.09753168e-01 1.50011659e-01
-1.31692612e+00 -1.00555070e-01 -2.45597601e-01 -1.14651732e-02
-1.80753157e-01 4.64410603e-01 -1.20010769e+00 3.87025446e-01
7.69265056e-01 1.30939138e+00 -2.90230781e-01 8.38957608e-01
1.84441537e-01 8.73907626e-01 -3.83136272e-01 3.97224694e-01
-7.17029721e-02 -3.34234945e-02 3.56383294e-01 1.30043364e+00
7.86913574e-01 -1.15774505e-01 2.57703096e-01 5.08096337e-01
-7.68817728e-03 -2.01864671e-02 -7.71474242e-01 -2.39539817e-01
1.02987993e+00 6.66762590e-01 -4.52226818e-01 -2.72901148e-01
-4.87149119e-01 1.11051846e+00 2.24283710e-01 6.56167448e-01
-8.95541966e-01 -5.43864667e-01 1.17023718e+00 4.97419909e-02
4.70762044e-01 -2.22767696e-01 -9.40946117e-02 -1.29881597e+00
-7.43957534e-02 -8.76233339e-01 9.87706125e-01 -4.25641388e-01
-1.94627655e+00 3.07715893e-01 1.74584165e-01 -1.84750271e+00
-4.52465117e-01 -4.92102534e-01 -3.56965423e-01 9.57618415e-01
-1.61067617e+00 -1.01302624e+00 -2.55313843e-01 1.08539808e+00
6.61757052e-01 -3.64508897e-01 1.15718091e+00 5.64435720e-01
-1.91586137e-01 8.30627561e-01 7.46782839e-01 1.22248590e-01
1.66254389e+00 -1.28588092e+00 5.80147743e-01 5.36351621e-01
-1.33015171e-01 4.59370673e-01 4.95393395e-01 -3.76356721e-01
-9.88439083e-01 -1.54465199e+00 8.41640115e-01 -5.38403034e-01
9.07876670e-01 -3.97641838e-01 -1.55622411e+00 8.60477567e-01
-1.41847059e-01 7.28282034e-02 1.12878454e+00 3.62331331e-01
-1.00757003e+00 -2.84532905e-01 -1.21902132e+00 2.08254591e-01
1.00133944e+00 -9.64345455e-01 -8.11453521e-01 4.12274867e-01
6.33642972e-01 -3.11203271e-01 -1.25468385e+00 2.38601282e-01
3.27428222e-01 -3.47486138e-01 9.89893854e-01 -6.95559740e-01
2.78355945e-02 -1.41749203e-01 -3.56510542e-02 -1.66553271e+00
-4.80997592e-01 -3.45841557e-01 -2.21938357e-01 1.44404674e+00
1.59666687e-01 -9.20015991e-01 3.89443249e-01 3.75669181e-01
9.38111022e-02 1.79782689e-01 -8.41050744e-01 -1.36128235e+00
5.24860024e-01 -2.58826822e-01 7.44594693e-01 1.72195184e+00
1.83619455e-01 3.41386825e-01 -9.98677313e-02 1.66541874e-01
4.88948524e-01 3.75483632e-01 5.56013167e-01 -1.64075410e+00
-1.14339786e-02 -1.09607942e-01 -1.95112586e-01 -9.72894549e-01
3.95898372e-01 -8.83203983e-01 7.41625950e-02 -1.05763662e+00
-2.73352116e-01 -6.21716440e-01 -6.50535882e-01 8.05257618e-01
-1.20456610e-02 1.09945297e-01 1.94577873e-02 5.85619092e-01
-3.06202710e-01 5.07510662e-01 1.08141661e+00 -2.41692632e-01
-3.13687682e-01 3.99347842e-02 -3.48415881e-01 3.12283993e-01
9.51543093e-01 -6.46201670e-01 -6.66605890e-01 -4.41824347e-01
-4.11648780e-01 -9.74764898e-02 1.05586000e-01 -1.01149952e+00
-6.78180829e-02 -3.29798460e-01 3.39602023e-01 -5.06942868e-01
1.10118970e-01 -9.78462756e-01 1.07715540e-01 2.79796779e-01
-2.55713820e-01 9.91638675e-02 7.08395183e-01 7.93791592e-01
-3.85844618e-01 1.34798780e-01 9.61462259e-01 2.03549251e-01
-1.18125832e+00 1.55917883e-01 -3.78470272e-01 3.70067894e-01
1.05069876e+00 -2.16290265e-01 -3.96535158e-01 -2.03068167e-01
-8.23139668e-01 2.50316650e-01 2.99732149e-01 8.07924926e-01
2.10106611e-01 -1.71404636e+00 -6.19003713e-01 5.37385762e-01
5.87028027e-01 -5.41851334e-02 3.48526418e-01 4.96137261e-01
1.15133651e-01 2.00501040e-01 -5.71453989e-01 -8.09079111e-01
-1.31834912e+00 5.53562522e-01 1.55473500e-01 -2.26591557e-01
-5.16970575e-01 6.70246601e-01 4.23848391e-01 -9.26606119e-01
2.05678552e-01 -3.76427531e-01 -1.30254731e-01 3.63788605e-01
3.92735034e-01 3.33268672e-01 1.25883952e-01 -4.50195372e-01
-4.44094896e-01 7.65104651e-01 -2.88370341e-01 1.06741734e-01
1.28332686e+00 -2.35616758e-01 2.63151020e-01 8.08055520e-01
1.44368005e+00 -4.78889048e-01 -1.45783722e+00 -9.75320697e-01
9.72760543e-02 -1.72741741e-01 -4.76951033e-01 -9.12730575e-01
-5.50575435e-01 8.92009556e-01 8.74314070e-01 2.48152584e-01
1.35333145e+00 -9.69272330e-02 9.47580040e-01 2.76284158e-01
4.45319504e-01 -1.23278058e+00 3.25229645e-01 9.44943666e-01
6.05510592e-01 -1.33960962e+00 -4.11919475e-01 -2.01758638e-01
-5.69864810e-01 9.66276407e-01 6.81186497e-01 1.65419385e-01
6.74285710e-01 9.20843240e-03 4.05588627e-01 1.93288460e-01
-7.66486704e-01 -1.58470571e-02 2.56268948e-01 1.11105740e+00
3.38039607e-01 1.48300514e-01 1.27870366e-01 4.59100485e-01
5.94418868e-02 -5.10892943e-02 1.04885884e-01 1.15979040e+00
-1.68540955e-01 -1.27355099e+00 -7.18567669e-01 2.20585570e-01
-2.81700641e-01 3.13601315e-01 -1.17650203e-01 7.88365006e-01
1.36281699e-02 8.49688470e-01 3.98553818e-01 -3.71220022e-01
5.87718785e-01 6.78212166e-01 1.52565673e-01 -9.37658489e-01
-4.80397671e-01 2.25738324e-02 -1.37583300e-01 -1.71197176e-01
-4.05614793e-01 -9.82665062e-01 -1.40453231e+00 -1.36069387e-01
6.91185985e-03 5.11635058e-02 5.04716098e-01 7.80878127e-01
7.94314742e-01 5.91133595e-01 8.00417960e-01 -6.26368999e-01
-8.05956364e-01 -1.17868984e+00 -7.91283667e-01 1.00005734e+00
5.74238896e-01 -9.88407254e-01 -3.07676613e-01 6.14459097e-01]
|
[10.301176071166992, 3.0330746173858643]
|
30de02c5-7c1e-4f38-b1c8-e7e60d3999e3
|
duck-rumour-detection-on-social-media-by
| null | null |
https://openreview.net/forum?id=VxlfmC-73ww
|
https://openreview.net/pdf?id=VxlfmC-73ww
|
DUCK: Rumour Detection on Social Media by Modelling User and Comment Propagation Networks
|
Social media rumours, a form of misinformation, can mislead the public and cause significant economic and social disruption. Motivated by the observation that the user network --- which captures $\textit{who}$ engage with a story --- and the comment network --- which captures $\textit{how}$ they react to it --- provide complementary signals for rumour detection, in this paper, we propose DUCK (rumour $\underline{d}$etection with $\underline{u}$ser and $\underline{c}$omment networ$\underline{k}$s) for rumour detection on social media. We study how to leverage transformers and graph attention networks to jointly model the contents and structure of social media conversations, as well as the network of users who engaged in these conversations. Over four widely used benchmark rumour datasets in English and Chinese, we show that DUCK produces superior performance for detecting rumours, creating a new state-of-the-art. Source code for DUCK is available at: ANONYMISED.
|
['Anonymous']
|
2022-01-16
| null | null | null |
acl-arr-january-2022-1
|
['rumour-detection']
|
['natural-language-processing']
|
[-3.05108905e-01 6.07562244e-01 -3.76064956e-01 -8.41822103e-02
-3.55607450e-01 -4.45247054e-01 6.24676824e-01 4.07244414e-01
1.93538725e-01 8.09472978e-01 6.20885015e-01 -3.68870676e-01
6.13218360e-03 -9.06351686e-01 -4.49186862e-01 -1.90245122e-01
-6.51561916e-01 3.35116267e-01 3.14861746e-03 -7.33276725e-01
4.42938864e-01 1.31777391e-01 -8.00451398e-01 3.70740563e-01
5.89478076e-01 9.13206637e-01 -4.46719855e-01 6.48088872e-01
-2.03793019e-01 1.84924793e+00 -8.35058689e-01 -7.42582440e-01
-2.44935095e-01 -9.10162807e-01 -1.08139408e+00 2.05742940e-01
2.05484480e-01 -6.60924196e-01 -1.01121306e+00 1.09950578e+00
1.16852872e-01 1.41590089e-01 7.69166708e-01 -1.32114387e+00
-9.82428849e-01 1.31216002e+00 -9.58333969e-01 8.75896215e-01
4.76451159e-01 -1.07974939e-01 1.29188049e+00 -6.88531160e-01
7.79209375e-01 1.23627710e+00 9.70417440e-01 2.76436239e-01
-8.45016718e-01 -1.00049770e+00 4.09495980e-02 -4.33052555e-02
-1.16977870e+00 -5.38033307e-01 1.09876597e+00 -4.24146146e-01
6.64015174e-01 6.09184444e-01 5.11966467e-01 1.30864584e+00
4.83987220e-02 9.51488495e-01 8.37636173e-01 2.70064414e-01
-2.79911578e-01 1.85647205e-01 3.43549848e-01 9.42689240e-01
-7.20143318e-04 -1.51810035e-01 -7.72339642e-01 -6.29234195e-01
7.17976153e-01 1.25406608e-01 -3.56383681e-01 4.95388627e-01
-9.70881760e-01 1.28512800e+00 7.82840848e-01 1.39962301e-01
-4.87724185e-01 1.71901345e-01 6.00269556e-01 7.56701767e-01
1.07270217e+00 6.06286407e-01 3.24637115e-01 -1.33905038e-01
-1.04946351e+00 9.78863910e-02 1.15350080e+00 1.02272868e+00
6.52281821e-01 1.36177868e-01 9.65252593e-02 1.19150329e+00
2.04355996e-02 3.22130144e-01 1.20217718e-01 -7.67005444e-01
6.58468366e-01 6.26532495e-01 2.60261148e-01 -1.74373388e+00
-7.77378082e-01 -4.64487255e-01 -1.24792826e+00 -5.17727435e-01
3.17168474e-01 -4.85323399e-01 -2.06400186e-01 1.35088193e+00
1.26729742e-01 2.70617068e-01 -4.34936553e-01 8.16268444e-01
1.07439089e+00 6.73822582e-01 -5.67419648e-01 -3.84814233e-01
1.19992983e+00 -1.02849734e+00 -8.47184658e-01 -2.15796247e-01
4.62241918e-01 -7.16600776e-01 5.93942106e-01 6.98682740e-02
-1.38989127e+00 1.71717286e-01 -8.33605945e-01 1.56175464e-01
-2.20766768e-01 -6.52257144e-01 4.11389768e-01 5.45388401e-01
-1.04579651e+00 8.11391771e-01 -2.93105538e-03 -3.34092081e-01
7.39139318e-01 -5.76105043e-02 4.87425886e-02 5.37359491e-02
-1.55090249e+00 4.13724840e-01 -2.33428791e-01 -4.25098874e-02
-1.25648761e+00 -2.39294022e-01 -6.83141351e-01 -2.11546928e-01
6.03608072e-01 -1.80473343e-01 1.11887431e+00 -6.39502704e-01
-1.07481873e+00 9.10740614e-01 1.07263930e-01 -5.89913845e-01
7.14883745e-01 3.54771465e-02 -7.03134716e-01 3.35269481e-01
2.13286117e-01 -1.58307299e-01 1.12422681e+00 -1.15235007e+00
-3.66701037e-01 -3.39860201e-01 1.55070141e-01 -2.47233436e-02
-3.84994745e-01 7.07160056e-01 8.70508477e-02 -6.06708884e-01
1.68256491e-01 -7.58786023e-01 2.04076916e-01 -5.64802647e-01
-1.13900912e+00 -2.36754864e-02 9.38408315e-01 -9.06633198e-01
1.61344397e+00 -1.78898466e+00 -1.26667976e-01 2.35796288e-01
1.17678833e+00 8.58781114e-02 3.38286370e-01 8.42260599e-01
3.18561882e-01 4.77958590e-01 1.85210742e-02 -5.44433594e-01
-6.32182211e-02 -2.81196833e-02 -3.95164430e-01 1.01093078e+00
-2.28158832e-01 7.59586990e-01 -1.29654777e+00 -1.36037290e-01
-3.17724884e-01 1.20508105e-01 -3.92284185e-01 3.72717172e-01
-2.07800701e-01 2.87468135e-01 -5.54502547e-01 4.51733828e-01
5.62375188e-01 -7.09723890e-01 -3.41196433e-02 6.47995710e-01
1.20263107e-01 7.04463243e-01 -5.13392389e-01 6.27255797e-01
-2.59810444e-02 8.01352322e-01 7.90367305e-01 -6.09308302e-01
1.18114889e+00 3.43901426e-01 4.11722720e-01 -2.48806059e-01
5.22180021e-01 7.67758563e-02 -3.66593450e-01 -4.98723119e-01
9.54256237e-01 -1.13628604e-01 -3.61606598e-01 1.13811386e+00
-3.77140701e-01 -1.09922424e-01 -1.89736500e-01 7.91263700e-01
1.24560833e+00 -8.87536108e-01 4.49088931e-01 -3.36941361e-01
1.30346417e-01 -1.27430633e-01 1.49092451e-01 1.10612011e+00
-4.47388738e-01 3.55999380e-01 1.35898006e+00 -2.90269643e-01
-1.03992867e+00 -6.45888746e-01 3.62925559e-01 1.47256446e+00
1.80929616e-01 -2.52240866e-01 -8.25731218e-01 -9.01846111e-01
1.93579242e-01 7.57490993e-01 -8.71305645e-01 1.58452258e-01
-4.68176007e-01 -1.03146517e+00 1.08761704e+00 1.56936906e-02
6.03570104e-01 -1.30812430e+00 2.88352042e-01 3.48121762e-01
-7.15730786e-01 -1.01277637e+00 -1.09729970e+00 -4.20772284e-01
-3.10489386e-01 -1.08954525e+00 -6.32700682e-01 -5.50710261e-01
3.84512573e-01 4.76903975e-01 1.31431866e+00 5.47628462e-01
5.46162464e-02 7.06403852e-02 -4.41152215e-01 -3.49490196e-01
-8.77274752e-01 -3.65930088e-02 -8.40963498e-02 3.36260080e-01
1.25730246e-01 -8.99876237e-01 -5.46852946e-01 6.84301972e-01
-6.70481563e-01 -1.66864201e-01 -1.45037875e-01 5.20095289e-01
-3.53651762e-01 1.15330685e-02 1.08229756e+00 -1.32420146e+00
1.03000975e+00 -1.35204875e+00 1.50314607e-02 -5.25822401e-01
-4.75682378e-01 -6.61180675e-01 6.54123366e-01 3.76206031e-03
-5.50279677e-01 -1.22403347e+00 6.85319826e-02 -2.64576823e-01
8.42122734e-02 6.46125436e-01 5.11736989e-01 2.72221297e-01
8.75970900e-01 2.43982762e-01 4.08957042e-02 -5.08623660e-01
2.48005047e-01 1.15898848e+00 4.36288565e-01 8.67749155e-02
3.93517911e-01 8.10862660e-01 -3.97846103e-01 -1.23087835e+00
-1.03003263e+00 -8.45203400e-01 2.91099995e-02 -4.59863216e-01
4.88761246e-01 -7.86618292e-01 -9.17626500e-01 8.96181047e-01
-1.34482574e+00 -2.65379667e-01 2.25891903e-01 5.31177483e-02
-3.25324565e-01 5.48924804e-01 -1.60759473e+00 -1.09637201e+00
-5.77234089e-01 -4.53319639e-01 2.26993427e-01 -2.54267454e-01
-5.29994130e-01 -1.13260937e+00 -2.87640601e-01 7.51991034e-01
6.12324476e-01 2.90699989e-01 4.83459711e-01 -1.02910900e+00
-4.59456623e-01 -3.86704147e-01 -1.01076865e+00 1.77059859e-01
9.89170223e-02 -4.57108468e-01 -8.38118434e-01 -3.32845956e-01
-9.91219073e-04 -3.54462862e-01 8.58118415e-01 3.68436873e-02
7.89705336e-01 -1.12783861e+00 -2.20431462e-01 1.95753574e-01
7.40175068e-01 -5.01348674e-01 6.64631248e-01 -1.03127316e-01
8.40055048e-01 6.11108840e-01 -1.13771111e-01 1.03279221e+00
6.86701953e-01 1.75425738e-01 9.52233911e-01 9.04483944e-02
-9.78297070e-02 -6.56257153e-01 6.17686808e-01 1.04917717e+00
-1.42323107e-01 -7.82006502e-01 -6.80034518e-01 8.61506104e-01
-1.64674556e+00 -1.28546929e+00 -5.72036743e-01 1.66696560e+00
6.66331172e-01 1.29234016e-01 7.99787879e-01 -5.50293997e-02
1.03662503e+00 1.05346203e+00 -2.26335704e-01 -5.03282726e-01
-1.31421715e-01 -5.83956599e-01 6.00146711e-01 8.61222863e-01
-7.02480257e-01 6.34864926e-01 5.44775963e+00 4.52173561e-01
-7.99526095e-01 3.14546973e-01 6.76801085e-01 -2.53440350e-01
-3.35580260e-01 -3.35118026e-01 -5.71224272e-01 7.62354910e-01
1.03190041e+00 -1.60863146e-01 9.41033065e-01 5.56354463e-01
5.70635200e-01 1.78005740e-01 -6.68224752e-01 6.13201022e-01
4.28470492e-01 -1.50608826e+00 -1.16060466e-01 3.63762677e-01
9.28007543e-01 5.07282317e-01 5.74990138e-02 7.98566788e-02
8.33327174e-01 -1.07949269e+00 6.77784204e-01 2.37682998e-01
7.17074037e-01 -5.27963221e-01 7.15739310e-01 6.77062809e-01
-8.75923693e-01 -1.12319812e-01 -2.99829870e-01 -3.66941512e-01
3.50396723e-01 8.47611785e-01 -1.28626835e+00 4.85216677e-01
5.32188833e-01 9.84896064e-01 8.17682594e-02 5.64002872e-01
-2.25134432e-01 1.01498580e+00 1.66474329e-03 -3.86526376e-01
5.99584997e-01 -9.41427574e-02 1.24771273e+00 1.47452748e+00
9.85721499e-02 1.46686658e-01 2.83540130e-01 1.10170579e+00
-9.77539539e-01 4.24089432e-02 -6.10036969e-01 -3.40972930e-01
4.96442080e-01 7.69965708e-01 -5.65635264e-01 -3.71006668e-01
-1.07591085e-01 1.00933230e+00 5.61589420e-01 3.25973123e-01
-7.67295599e-01 -3.82467389e-01 6.00038767e-01 7.80200720e-01
1.54265895e-01 4.26302366e-02 -7.64722154e-02 -9.98299658e-01
-3.26867998e-01 -7.91383803e-01 5.77952027e-01 -5.87927401e-01
-1.63517499e+00 5.79120517e-01 -3.87197763e-01 -6.27498567e-01
-4.44874652e-02 1.78841427e-01 -8.23093593e-01 7.47284830e-01
-1.48993397e+00 -8.81494582e-01 -2.65854567e-01 5.71589530e-01
3.21215600e-01 -1.26546353e-01 4.14454043e-01 2.20452651e-01
-4.94146138e-01 5.18083334e-01 3.40872496e-01 7.44481504e-01
3.01865876e-01 -1.08545399e+00 6.64042652e-01 2.99640864e-01
-1.89690262e-01 2.95457870e-01 8.96691144e-01 -9.34029937e-01
-8.97119343e-01 -1.27209902e+00 1.17929244e+00 -4.29067910e-01
1.19468975e+00 -2.98274279e-01 -1.04730535e+00 9.97162640e-01
3.55051219e-01 -5.09497300e-02 4.63731050e-01 1.96226016e-01
-4.80878174e-01 4.68711525e-01 -1.34792781e+00 4.35582787e-01
1.23995101e+00 -8.47044706e-01 -4.44181204e-01 6.81128800e-01
7.00239837e-01 -2.52146542e-01 -8.12638581e-01 -4.69525069e-01
2.14776769e-01 -1.44039702e+00 6.36794627e-01 -7.96855032e-01
8.11810315e-01 4.11177933e-01 1.63852155e-01 -1.55910337e+00
-4.29319978e-01 -1.45307577e+00 -3.47087562e-01 1.08911085e+00
5.17906547e-01 -9.14981186e-01 9.00420487e-01 4.54529114e-02
-4.04289663e-02 -4.72960025e-01 -7.76642323e-01 -4.97940183e-01
2.47136895e-02 -2.56063282e-01 8.47026825e-01 1.52821851e+00
6.85111761e-01 3.99715543e-01 -1.18012953e+00 4.41932771e-03
6.78406000e-01 -1.50684938e-01 6.35257661e-01 -1.17413676e+00
-9.94515121e-02 -5.17688692e-01 2.04923507e-02 -1.27610445e+00
4.97311242e-02 -9.59619164e-01 -1.96912661e-01 -1.30780673e+00
4.13272202e-01 -3.67306381e-01 -2.27638394e-01 7.43370056e-02
1.45565137e-01 3.12337369e-01 2.06501819e-02 5.20169377e-01
-8.69022846e-01 3.04412246e-01 1.42971349e+00 -5.14746048e-02
-2.01158911e-01 1.93271503e-01 -9.53404963e-01 8.17127645e-01
5.54582357e-01 -7.24275947e-01 2.62980210e-03 1.90934725e-02
7.69336343e-01 8.84662747e-01 5.02936244e-01 -1.53350934e-01
3.38998511e-02 -4.99821408e-03 -4.33888942e-01 -4.78218079e-01
5.21004319e-01 -1.22135632e-01 -3.96374404e-01 7.49391541e-02
-5.94395578e-01 1.28040656e-01 -2.59943277e-01 1.13665283e+00
7.85849988e-02 -1.54236302e-01 7.05941081e-01 -4.66373324e-01
5.04439464e-03 3.79957914e-01 -6.72311962e-01 5.86267412e-01
4.41803843e-01 2.29242548e-01 -9.16416287e-01 -1.59460366e+00
-6.47672594e-01 2.78474212e-01 4.36597884e-01 3.19223613e-01
4.99962538e-01 -8.68137121e-01 -1.37712574e+00 -1.12536065e-01
-1.51397184e-01 -4.33105379e-01 3.15243781e-01 1.20833838e+00
-1.74504846e-01 1.42568782e-01 1.46807045e-01 6.53563887e-02
-1.05598938e+00 4.89674479e-01 4.36164558e-01 -3.50664467e-01
-5.44063985e-01 1.14365625e+00 -1.48078561e-01 -3.70986104e-01
6.74666241e-02 1.21150948e-01 -2.23093137e-01 3.82510304e-01
7.12620437e-01 1.08778811e+00 -2.58586705e-01 -9.78074491e-01
-8.44242796e-02 -6.07808053e-01 -3.02993357e-01 2.72204995e-01
1.27670419e+00 -7.19826281e-01 -7.97824919e-01 4.58472043e-01
1.49907267e+00 2.54973203e-01 -6.60469830e-01 -6.23542368e-01
-2.25773126e-01 -5.41065931e-01 2.18985990e-01 -4.86934066e-01
-9.13896620e-01 4.29479420e-01 -6.85950458e-01 1.28066289e+00
1.81560606e-01 3.48271877e-01 1.28168237e+00 4.92838398e-02
3.16296279e-01 -9.76537704e-01 3.47748369e-01 7.21112847e-01
1.03425968e+00 -1.08160520e+00 6.02262132e-02 -6.80203199e-01
-6.49638057e-01 6.66499555e-01 -2.00009476e-02 -1.81227088e-01
7.88791537e-01 -2.72459298e-01 -1.89843506e-01 -8.25141251e-01
-4.98507559e-01 3.63588989e-01 -1.91624254e-01 7.61483982e-02
2.81523138e-01 3.59243602e-01 4.14876677e-02 6.04430914e-01
-4.52034444e-01 -2.44346946e-01 1.23942351e+00 5.21860838e-01
-6.45954788e-01 -2.20028564e-01 -4.80906725e-01 1.09729218e+00
-8.36996138e-01 -1.39247999e-01 -8.74135911e-01 5.46415925e-01
-4.14821416e-01 1.53083777e+00 -1.80513918e-01 -5.33201218e-01
-1.40517056e-01 -1.16798937e-01 -2.03393251e-01 -5.33156693e-01
-9.94001269e-01 -2.11934879e-01 9.57052350e-01 -3.36456180e-01
-7.17144692e-03 -8.04356039e-01 -9.91934955e-01 -1.43210256e+00
-5.07023990e-01 3.92141461e-01 1.09214060e-01 7.90895104e-01
1.83066890e-01 2.90773004e-01 1.13961875e+00 -4.97812837e-01
-7.01176763e-01 -1.28786767e+00 -1.43529725e+00 5.60504556e-01
9.11122024e-01 -3.20097834e-01 -9.30626571e-01 -5.06424725e-01]
|
[8.185026168823242, 10.142399787902832]
|
dc452556-70d1-4916-a4b6-a88a140b623c
|
colo-scrl-self-supervised-contrastive
|
2303.15671
| null |
https://arxiv.org/abs/2303.15671v1
|
https://arxiv.org/pdf/2303.15671v1.pdf
|
Colo-SCRL: Self-Supervised Contrastive Representation Learning for Colonoscopic Video Retrieval
|
Colonoscopic video retrieval, which is a critical part of polyp treatment, has great clinical significance for the prevention and treatment of colorectal cancer. However, retrieval models trained on action recognition datasets usually produce unsatisfactory retrieval results on colonoscopic datasets due to the large domain gap between them. To seek a solution to this problem, we construct a large-scale colonoscopic dataset named Colo-Pair for medical practice. Based on this dataset, a simple yet effective training method called Colo-SCRL is proposed for more robust representation learning. It aims to refine general knowledge from colonoscopies through masked autoencoder-based reconstruction and momentum contrast to improve retrieval performance. To the best of our knowledge, this is the first attempt to employ the contrastive learning paradigm for medical video retrieval. Empirical results show that our method significantly outperforms current state-of-the-art methods in the colonoscopic video retrieval task.
|
['Suncheng Xiang', 'Dahong Qian', 'Zefang Yu', 'Crystal Cai', 'Shilun Cai', 'Qingzhong Chen']
|
2023-03-28
| null | null | null | null |
['video-retrieval', 'general-knowledge']
|
['computer-vision', 'miscellaneous']
|
[ 2.67814845e-01 -2.24706203e-01 -6.90339327e-01 4.61402647e-02
-9.69847620e-01 -2.08755657e-01 4.55297828e-01 3.97050619e-01
-4.80937809e-01 3.48797649e-01 3.08325559e-01 -3.78679037e-01
-3.62564832e-01 -6.05214059e-01 -6.84251726e-01 -5.30236185e-01
-1.14007063e-01 1.89732835e-01 -2.97089084e-03 -1.29860133e-01
2.99430907e-01 3.31428647e-02 -1.28556311e+00 8.11231434e-01
8.86286974e-01 8.89497161e-01 4.02434498e-01 7.84185052e-01
3.21844935e-01 1.06070650e+00 -4.42831486e-01 -1.99871942e-01
2.19173789e-01 -5.38070858e-01 -8.31084549e-01 2.87875887e-02
2.98232287e-01 -5.85168123e-01 -8.52586329e-01 1.23616123e+00
6.41194761e-01 1.40853673e-01 7.08779097e-01 -3.39877754e-01
-9.42160845e-01 5.66171110e-01 -1.80802688e-01 5.13803840e-01
6.40125930e-01 -1.04653284e-01 9.42142844e-01 -5.76005042e-01
8.05811822e-01 8.25004697e-01 3.78741980e-01 5.39747238e-01
-6.20687544e-01 -2.10860386e-01 1.34105742e-01 2.06507400e-01
-1.12904513e+00 -1.15253612e-01 5.25062501e-01 -3.48247826e-01
7.98439562e-01 4.78622466e-01 9.65874672e-01 1.20559263e+00
3.77060115e-01 1.34327793e+00 7.48849511e-01 -5.58165193e-01
1.77102499e-02 1.26933590e-01 1.45996278e-02 8.35220397e-01
4.23817486e-01 3.05846989e-01 -3.30638289e-01 -2.06253320e-01
8.06619108e-01 4.06406581e-01 -8.21737170e-01 -3.00396502e-01
-1.44871449e+00 8.29359949e-01 7.07554221e-01 4.56994981e-01
-3.88375133e-01 -6.86074467e-03 4.42842156e-01 6.59079909e-01
3.21831346e-01 1.00354362e+00 -2.98820913e-01 -1.58276543e-01
-9.35875416e-01 -3.10409311e-02 9.54820991e-01 6.74041688e-01
-5.76422811e-02 -2.72394121e-01 -1.62146181e-01 7.28749335e-01
3.38533342e-01 2.30657533e-01 1.31912422e+00 -4.61316049e-01
3.16183537e-01 6.03726387e-01 4.87582460e-02 -1.04449201e+00
-5.07900156e-02 -3.60128760e-01 -7.39896953e-01 -3.55311036e-01
8.77741650e-02 2.38884017e-01 -1.07689333e+00 1.02118909e+00
6.25585066e-03 4.27046090e-01 4.19717133e-01 1.13730359e+00
1.38751900e+00 3.55347216e-01 3.34803923e-03 -5.66726699e-02
1.31451547e+00 -1.44450366e+00 -6.75932467e-01 -4.76202741e-02
7.62457252e-01 -8.37863803e-01 5.48571825e-01 4.14529353e-01
-1.05325687e+00 -5.59611917e-01 -1.11625934e+00 5.65132648e-02
-2.11953089e-01 6.16230845e-01 9.09886420e-01 4.94322479e-01
-6.93545759e-01 5.58700562e-01 -1.11642325e+00 -3.59505981e-01
1.40805900e-01 2.11083516e-01 -3.24828476e-01 -4.25562412e-01
-1.29427433e+00 8.49488497e-01 5.24539053e-01 1.17098324e-01
-8.75442863e-01 -5.53188920e-01 -9.35794294e-01 7.39982873e-02
5.21614015e-01 -8.15113485e-01 1.37847078e+00 -1.13065314e+00
-1.51444435e+00 1.02876472e+00 2.82866031e-01 -8.23969781e-01
3.31455708e-01 -5.02148628e-01 -5.51677704e-01 6.89614475e-01
-3.27060938e-01 6.06253684e-01 1.19411421e+00 -7.53825545e-01
-4.68761742e-01 -4.32724841e-02 1.53031439e-01 4.72034305e-01
-3.30543816e-01 -3.17722946e-01 -9.18274105e-01 -9.49582100e-01
-4.97859828e-02 -1.21335685e+00 -5.83048522e-01 -6.45482093e-02
-2.67477352e-02 1.25685073e-02 1.43800095e-01 -8.34950924e-01
1.49907327e+00 -2.28621101e+00 3.84064376e-01 -1.84448343e-02
2.45600671e-01 7.97254384e-01 -2.21395150e-01 3.87689352e-01
-4.90627587e-02 7.48405159e-02 9.32760909e-02 1.58597201e-01
-6.08160973e-01 -5.43564744e-02 -1.43426746e-01 5.32498538e-01
1.10026903e-01 9.84612048e-01 -1.27509487e+00 -5.74534953e-01
3.44977736e-01 2.95655578e-01 -6.49323761e-01 3.30382437e-01
-2.89142668e-01 2.34932140e-01 -8.21885645e-01 9.93073404e-01
2.49995038e-01 -8.16405594e-01 3.51153672e-01 -1.91038325e-01
2.76258618e-01 9.49733704e-02 -6.17896795e-01 2.16066027e+00
-1.66244388e-01 5.71195245e-01 -3.91221106e-01 -1.24382544e+00
4.12741154e-01 5.59603512e-01 6.56700313e-01 -6.53471649e-01
3.16052377e-01 3.77424240e-01 2.22715497e-01 -1.05002379e+00
6.17281616e-01 3.45225155e-01 1.32290244e-01 -1.47867948e-02
-7.11638667e-03 -7.47856172e-03 2.88222581e-01 8.83435979e-02
1.23281312e+00 -3.35935727e-02 8.52254927e-01 9.87484008e-02
6.78235948e-01 3.42571080e-01 9.34103876e-02 9.59255219e-01
-3.50042254e-01 7.54705071e-01 -4.32307459e-02 -5.84363341e-01
-6.67217970e-01 -6.28858447e-01 -1.89871222e-01 4.48157609e-01
4.58042085e-01 -4.13875401e-01 -3.62181067e-01 -6.37928009e-01
6.31268471e-02 -3.84068526e-02 -7.50365198e-01 -5.16977608e-01
-6.44182742e-01 -7.70858407e-01 4.16923821e-01 4.08516735e-01
4.05721366e-01 -1.14397156e+00 -6.97013021e-01 3.47901374e-01
-2.75719196e-01 -8.46360564e-01 -5.01639009e-01 -4.77379970e-02
-1.07245827e+00 -1.59148431e+00 -1.36010253e+00 -1.08900845e+00
7.28831530e-01 7.02547669e-01 1.09173262e+00 5.03060520e-01
-7.65254617e-01 7.61187375e-01 -7.36510754e-01 -2.46610627e-01
-5.08771539e-01 1.89003050e-01 -3.38022172e-01 -2.95515716e-01
3.37327063e-01 1.75025225e-01 -1.09642541e+00 4.31594811e-02
-1.15736532e+00 -1.29322052e-01 1.10752940e+00 1.35514307e+00
8.44426870e-01 -2.08504677e-01 9.31686684e-02 -8.45101416e-01
8.44191432e-01 -7.26604104e-01 -4.39657003e-01 4.52018946e-01
-5.33386707e-01 -2.26830412e-02 3.84191036e-01 -9.03099179e-01
-7.63141513e-01 -1.37034595e-01 7.88063332e-02 -7.29752898e-01
1.46412581e-01 1.34272563e+00 8.30177188e-01 -2.06600189e-01
8.73194814e-01 4.34796453e-01 1.68219522e-01 -3.94713819e-01
3.04580443e-02 7.11090684e-01 3.52281541e-01 -1.07505331e-02
2.03736827e-01 4.25644100e-01 -2.09849954e-01 -7.51637101e-01
-8.37102354e-01 -1.04388249e+00 -1.19609721e-01 5.97439297e-02
6.73641086e-01 -1.37511027e+00 -3.47350210e-01 1.59803614e-01
-6.26964390e-01 -6.92161098e-02 -1.03295902e-02 1.06016326e+00
-3.98635030e-01 6.59010649e-01 -9.08031702e-01 -4.09624696e-01
-5.74744225e-01 -1.23941970e+00 1.00062597e+00 1.56277388e-01
1.35395274e-01 -8.14289033e-01 4.15323973e-01 3.33986461e-01
3.64436299e-01 3.98615077e-02 4.88057196e-01 -7.78014123e-01
-8.84423077e-01 -6.17617607e-01 -9.13927034e-02 4.06576604e-01
1.59524038e-01 -1.12876289e-01 -3.10229570e-01 -6.40848279e-01
8.71756598e-02 -4.62911636e-01 1.27915359e+00 4.93775785e-01
1.44296849e+00 -8.12396989e-04 -6.76351905e-01 5.69934905e-01
1.33913004e+00 3.45454782e-01 6.05321765e-01 5.22478282e-01
2.54663318e-01 7.15877935e-02 9.49084997e-01 2.42411986e-01
3.43846753e-02 3.37499142e-01 3.55462134e-01 1.71448626e-02
-3.29846919e-01 -1.91725478e-01 9.03626606e-02 1.13897991e+00
7.91666843e-03 -5.42449713e-01 -7.25368619e-01 7.06104040e-01
-1.89698100e+00 -8.29929471e-01 3.70176107e-01 2.06527853e+00
6.96421206e-01 -2.02075273e-01 -3.31324190e-01 -1.11732945e-01
4.15727705e-01 2.23662794e-01 -3.80106270e-01 8.65141377e-02
3.29253703e-01 9.16044712e-02 2.90036201e-01 8.32596943e-02
-1.44759083e+00 5.94667494e-01 6.41748381e+00 9.11797464e-01
-1.41363120e+00 -2.20551908e-01 3.62191796e-01 9.49236900e-02
1.06282093e-01 -4.05528158e-01 -4.99264151e-01 4.09647763e-01
7.18797624e-01 -3.90404984e-02 3.63600552e-01 1.13680387e+00
-4.15612549e-01 -1.23194754e-01 -1.17816734e+00 1.17088389e+00
5.14799535e-01 -1.50966227e+00 8.38784501e-02 -1.59073725e-01
8.68390083e-01 1.28393412e-01 3.71800333e-01 5.01463115e-01
-1.27578825e-01 -1.03223813e+00 -1.05805621e-02 6.13488793e-01
7.71825850e-01 -2.06381962e-01 8.63709509e-01 2.18934923e-01
-9.30952907e-01 -5.34457155e-02 -3.69239271e-01 3.39484662e-01
-2.25020319e-01 2.54226983e-01 -9.51604486e-01 7.97377229e-01
5.10852933e-01 1.24715173e+00 -5.63812613e-01 1.67357063e+00
-2.90336292e-02 5.41903138e-01 -4.56331745e-02 -2.94010967e-01
3.50677669e-01 1.97712675e-01 5.86155295e-01 1.15897954e+00
5.01491129e-01 1.78244722e-03 3.15627545e-01 2.48927221e-01
-2.89990872e-01 2.98432350e-01 -6.80876255e-01 -5.92181742e-01
1.66701875e-03 9.37698007e-01 -4.91082042e-01 -5.78381538e-01
-5.93067944e-01 1.06313360e+00 1.26263812e-01 3.15384418e-01
-5.88060319e-01 -1.96684852e-01 6.08006082e-02 -2.68572867e-01
3.59245718e-01 3.28817844e-01 5.09807944e-01 -1.69677985e+00
3.50474231e-02 -1.46438694e+00 6.64000869e-01 -4.38763440e-01
-1.19848800e+00 6.89532101e-01 -1.02035999e-01 -1.95173669e+00
-4.30869609e-01 -7.08243966e-01 -6.25268444e-02 2.68836886e-01
-1.93304181e+00 -9.55232203e-01 -3.77332687e-01 6.62214518e-01
7.40193546e-01 -4.81019408e-01 1.14075053e+00 2.80867457e-01
-2.20266223e-01 4.46674794e-01 4.63302523e-01 2.00620309e-01
9.25266206e-01 -1.08487105e+00 -5.67030683e-02 4.95808154e-01
3.41732502e-01 8.18534672e-01 3.16316247e-01 -7.28200674e-01
-1.63471532e+00 -9.68770027e-01 4.55332994e-01 -2.89762080e-01
5.44659138e-01 5.53019702e-01 -8.54878902e-01 6.16422832e-01
7.13658854e-02 1.78490877e-01 8.53268325e-01 -2.53772974e-01
-1.16392709e-01 2.45078683e-01 -8.31016243e-01 7.78572798e-01
6.34564281e-01 -6.34100080e-01 -9.09726977e-01 6.24814928e-01
6.05631948e-01 -9.61026192e-01 -1.26480305e+00 7.26430535e-01
6.57317936e-01 -4.39882129e-01 1.27061868e+00 -6.00916326e-01
8.57402444e-01 -4.00588326e-02 1.11412801e-01 -1.17417049e+00
-1.17686369e-01 -4.82642591e-01 -4.68463540e-01 1.37781769e-01
2.62898475e-01 -3.32935810e-01 7.57477641e-01 1.19531162e-01
-1.35657519e-01 -9.18071210e-01 -5.62400937e-01 -4.82564539e-01
-1.16004340e-01 3.12390327e-02 1.08975507e-01 9.63433743e-01
1.85369641e-01 -3.09136331e-01 -5.19929826e-01 2.20171977e-02
1.35329932e-01 4.41359639e-01 5.51515222e-01 -9.11419332e-01
-7.33371735e-01 -3.48573416e-01 -3.39943320e-01 -1.35521567e+00
-9.23796296e-02 -1.04082942e+00 -1.52230011e-02 -1.42158771e+00
3.50744575e-01 -5.02611957e-02 -6.36997640e-01 -8.79603699e-02
-5.41509867e-01 1.35881558e-01 9.37558189e-02 5.08670270e-01
-8.89004290e-01 2.49907568e-01 1.74512053e+00 -4.76318598e-01
-2.61021048e-01 2.26593003e-01 -6.31533623e-01 6.90261185e-01
6.05203986e-01 -4.37858462e-01 -3.81626040e-01 -3.85555297e-01
2.49033526e-01 5.25407910e-01 8.43435079e-02 -8.19037080e-01
4.10295486e-01 7.36655965e-02 3.90755653e-01 -5.64197540e-01
2.80033588e-01 -7.86240518e-01 1.63797718e-02 9.75214362e-01
-6.22119606e-01 -1.21854290e-01 8.00804198e-02 9.91051316e-01
-9.04297531e-01 -5.57591081e-01 3.48476887e-01 -7.35062718e-01
-7.81991243e-01 3.24842453e-01 -5.86061895e-01 -1.47724584e-01
8.63204181e-01 1.10228471e-01 -3.77801418e-01 -2.19737113e-01
-6.63187385e-01 1.72681794e-01 2.49185577e-01 6.23885870e-01
9.36296999e-01 -1.14879012e+00 -6.14833295e-01 7.14876279e-02
3.97666752e-01 -9.64459479e-02 2.80834854e-01 8.83572757e-01
-8.88146877e-01 8.51478457e-01 8.57124031e-02 -5.40902495e-01
-1.48800075e+00 9.58024740e-01 3.12308043e-01 -9.45598006e-01
-9.06904697e-01 8.79192472e-01 -3.34714167e-02 -4.17440161e-02
4.87604767e-01 -5.60714543e-01 -5.24885952e-01 -2.07667038e-01
7.55053401e-01 -8.75568911e-02 1.33279292e-02 -8.43234435e-02
-3.20129730e-02 3.39955926e-01 -6.49928272e-01 2.75348425e-01
1.11062622e+00 1.28450155e-01 1.32856116e-01 -1.79547034e-02
1.15750539e+00 -1.66599363e-01 -6.66069210e-01 -2.57589757e-01
-1.33328333e-01 -6.92605674e-01 2.40338266e-01 -8.83609772e-01
-9.91545558e-01 5.25548995e-01 7.75731087e-01 2.79226042e-02
1.01077175e+00 -2.05228299e-01 8.76759529e-01 1.00789452e+00
1.87191486e-01 -9.22477663e-01 5.52975833e-01 2.04622030e-01
9.48098421e-01 -1.79986691e+00 3.55410963e-01 -3.53771269e-01
-5.49096286e-01 1.16815472e+00 3.92324656e-01 -5.17058432e-01
7.24187970e-01 -2.81988621e-01 1.52599111e-01 -2.48724103e-01
-6.19455993e-01 -9.76964161e-02 7.66883373e-01 2.50843436e-01
5.08317947e-01 -2.18121074e-02 -6.19183421e-01 3.36122721e-01
3.26192081e-01 4.27839011e-01 1.82988524e-01 1.24521124e+00
-1.95624888e-01 -1.09950840e+00 -2.02598065e-01 7.30347693e-01
-1.01239312e+00 -3.49980265e-01 -8.07213262e-02 9.11514044e-01
-4.89779711e-01 7.05439746e-01 -3.65086257e-01 -2.78829802e-02
1.96572114e-02 -2.92688876e-01 7.42932320e-01 -5.65270007e-01
-8.22175741e-01 3.37985903e-01 3.77248116e-02 -4.91525441e-01
-8.68087769e-01 -3.82646531e-01 -6.65784895e-01 3.61828297e-01
-5.89522064e-01 2.08010182e-01 4.51716095e-01 5.74558258e-01
2.81215578e-01 7.75035739e-01 3.64759803e-01 -6.66313767e-01
-8.14132094e-01 -8.51902187e-01 -3.33597094e-01 7.29268909e-01
5.13126731e-01 -3.94713342e-01 -3.39248240e-01 2.13947184e-02]
|
[14.302314758300781, -3.066619873046875]
|
8600b667-fa57-43e3-9fe8-d4ab48099301
|
combating-the-elsagate-phenomenon-deep
|
1904.08910
| null |
http://arxiv.org/abs/1904.08910v1
|
http://arxiv.org/pdf/1904.08910v1.pdf
|
Combating the Elsagate phenomenon: Deep learning architectures for disturbing cartoons
|
Watching cartoons can be useful for children's intellectual, social and
emotional development. However, the most popular video sharing platform today
provides many videos with Elsagate content. Elsagate is a phenomenon that
depicts childhood characters in disturbing circumstances (e.g., gore, toilet
humor, drinking urine, stealing). Even with this threat easily available for
children, there is no work in the literature addressing the problem. As the
first to explore disturbing content in cartoons, we proceed from the most
recent pornography detection literature applying deep convolutional neural
networks combined with static and motion information of the video. Our solution
is compatible with mobile platforms and achieved 92.6% of accuracy. Our goal is
not only to introduce the first solution but also to bring up the discussion
around Elsagate.
|
['Sandra Avila', 'Edson Bollis', 'Akari Ishikawa']
|
2019-04-18
| null | null | null | null |
['pornography-detection']
|
['computer-vision']
|
[-2.17096344e-01 1.05093792e-01 -2.22912833e-01 2.20337719e-01
8.94914716e-02 -6.27405822e-01 2.74090201e-01 3.10218811e-01
-5.98636977e-02 4.66119111e-01 4.84820634e-01 1.07654044e-02
3.36725026e-01 -8.74230027e-01 -6.60511553e-01 -3.60351533e-01
-1.48391277e-01 -4.09290165e-01 3.84448349e-01 -2.68574834e-01
5.00592828e-01 4.48018134e-01 -1.86080277e+00 5.63581526e-01
5.67622840e-01 5.44424593e-01 -3.12746555e-01 9.60870385e-01
-1.26215175e-01 1.50228190e+00 -7.67070115e-01 -1.01469302e+00
-4.05970067e-01 -2.29503438e-01 -5.21054864e-01 -1.79191038e-01
1.03732455e+00 -1.18099761e+00 -7.62601793e-01 1.20246077e+00
7.31559575e-01 -9.66450647e-02 1.54664502e-01 -1.23586488e+00
-8.97023141e-01 6.24226570e-01 -6.26912534e-01 7.50765622e-01
1.02173281e+00 -2.20867828e-01 3.06235194e-01 -5.05095661e-01
9.47265863e-01 1.09007239e+00 6.67837381e-01 7.28756905e-01
-4.37088639e-01 -1.04337740e+00 -1.85212806e-01 5.33058941e-01
-9.19783056e-01 -2.57555157e-01 9.70499694e-01 -5.03525019e-01
8.71602833e-01 3.01739395e-01 1.48826647e+00 1.90507746e+00
5.90506718e-02 1.35646772e+00 4.44856912e-01 -3.37020785e-01
-1.58003762e-01 -1.23675719e-01 -2.15732589e-01 7.11572647e-01
4.58366185e-01 -3.86903197e-01 -8.90912890e-01 3.24271917e-02
4.64100659e-01 -2.58766294e-01 -4.44292277e-02 -1.69059947e-01
-4.73705530e-01 7.11672604e-01 -2.43018791e-01 6.89836085e-01
3.26356553e-02 5.05962729e-01 7.36806571e-01 4.19662118e-01
7.39702463e-01 1.72046602e-01 3.08376342e-01 -1.08098316e+00
-1.12152946e+00 3.99843276e-01 8.38689983e-01 7.18824923e-01
-1.18771717e-01 1.54967815e-01 2.37370670e-01 9.63976681e-01
1.98313579e-01 2.97821730e-01 3.76314044e-01 -9.12784338e-01
4.89746898e-01 7.30642527e-02 -1.03759974e-01 -1.70642269e+00
-5.13680875e-01 9.68977734e-02 -2.80691952e-01 7.00153112e-02
5.69363177e-01 -4.38849390e-01 -4.94631350e-01 1.34303486e+00
3.82391095e-01 5.60833693e-01 -5.81364334e-01 7.07933545e-01
1.62015057e+00 2.77684271e-01 -1.38769016e-01 -1.88612137e-02
1.11728096e+00 -9.15223241e-01 -1.09787822e+00 1.68437123e-01
6.22968554e-01 -1.02906251e+00 4.09248322e-01 1.07825112e+00
-1.56899512e+00 9.56890285e-02 -1.36681318e+00 -3.16070944e-01
-6.72648132e-01 -4.98466492e-01 9.48157072e-01 1.51082790e+00
-1.13893259e+00 8.71699750e-01 -8.54696691e-01 -6.92299843e-01
7.85063803e-01 8.14145058e-02 -5.44528902e-01 -1.99829731e-02
-1.14164829e+00 7.71921694e-01 8.68868902e-02 -4.04687852e-01
-6.39287531e-01 -7.62279689e-01 -7.39294231e-01 -1.60773203e-01
3.78792472e-02 3.05396587e-01 1.26232040e+00 -1.47553813e+00
-1.52506149e+00 1.19344938e+00 5.30238271e-01 -5.30583523e-02
7.21839845e-01 -6.01866543e-01 -7.88561702e-01 8.20031881e-01
-1.42865377e-02 6.18433595e-01 1.04529381e+00 -4.36642438e-01
-4.76047456e-01 1.44474134e-01 2.23794639e-01 -8.67384225e-02
-8.46617162e-01 6.54020905e-01 -5.99197388e-01 -7.89533377e-01
4.78188246e-02 -5.78115165e-01 6.44466579e-01 9.37972963e-02
-2.89184242e-01 -1.10726366e-02 1.01591313e+00 -1.05436611e+00
1.56315351e+00 -2.12715197e+00 -5.09362459e-01 -2.61553168e-01
3.94832581e-01 6.35431767e-01 1.03366219e-01 8.98393393e-01
-2.80269563e-01 1.75548226e-01 7.12550402e-01 9.05280840e-03
-2.29893420e-02 -3.34218681e-01 2.05307543e-01 8.55398774e-01
-2.14909136e-01 7.10456252e-01 -1.43251598e+00 -2.65363842e-01
3.43145460e-01 8.19276869e-01 -5.80336392e-01 -2.21653432e-01
3.28686327e-01 5.67086488e-02 -2.50397354e-01 1.08219552e+00
8.94521594e-01 2.40576848e-01 5.28952666e-02 3.49259079e-01
-3.54278088e-01 1.99497908e-01 -8.61707985e-01 1.32776892e+00
2.57990044e-02 1.56472862e+00 -1.35510217e-03 -6.38310611e-01
3.85978580e-01 5.85081100e-01 6.15290165e-01 -8.37029338e-01
3.75840753e-01 1.85137540e-01 -1.28678381e-01 -1.28209960e+00
6.33816302e-01 3.69971246e-01 2.53484637e-01 3.18992436e-01
2.53031515e-02 3.35662544e-01 1.47109672e-01 3.87273699e-01
1.33164859e+00 2.16302842e-01 4.14918244e-01 5.56104556e-02
7.17126653e-02 -5.82873642e-01 -1.03225544e-01 6.69328332e-01
-5.72956741e-01 1.00098753e+00 9.99627173e-01 -8.23394716e-01
-9.86296654e-01 -7.65353739e-01 2.24309284e-02 1.17478824e+00
-1.06760696e-01 -4.08686489e-01 -1.02695525e+00 -4.94658947e-01
5.28590679e-02 5.05179703e-01 -7.36204803e-01 7.18720183e-02
-5.63137233e-01 -3.94028187e-01 8.67185175e-01 2.40160808e-01
6.07775629e-01 -1.19364238e+00 -1.06715131e+00 3.36449862e-01
-2.79707052e-02 -9.14239824e-01 -3.16849142e-01 -5.11894763e-01
-4.69018221e-01 -1.02290702e+00 -9.14430559e-01 -7.47395337e-01
2.25067228e-01 4.71777409e-01 1.05020463e+00 4.25871789e-01
-2.98848242e-01 6.76052988e-01 -6.63281739e-01 -3.83393675e-01
-2.57585704e-01 -3.25097114e-01 -5.86179979e-02 -3.60874385e-01
8.34840834e-01 -1.24766922e+00 -8.11888039e-01 -3.29014748e-01
-1.04055679e+00 1.56838775e-01 -2.55791008e-01 3.04271489e-01
-5.67592978e-01 -2.56187737e-01 2.68780500e-01 -7.23921239e-01
3.63356799e-01 -1.09198558e+00 -2.01923683e-01 -4.13248152e-01
1.06543332e-01 -9.05547678e-01 2.83716351e-01 -8.52665603e-01
-7.01684356e-01 -5.04005671e-01 -4.27364856e-01 -6.22767687e-01
-6.63441896e-01 -8.16283897e-02 2.61455536e-01 -2.35560864e-01
3.93930882e-01 -2.58459419e-01 -5.58124840e-01 -5.64056635e-01
5.78767136e-02 6.37103319e-01 6.27171099e-01 -6.59731627e-02
4.05907512e-01 6.94238842e-01 -1.22146405e-01 -1.11306083e+00
-4.90169197e-01 -4.43477690e-01 -4.49285775e-01 -9.29904699e-01
1.00108683e+00 -6.87775791e-01 -1.16104662e+00 1.03261399e+00
-1.06274188e+00 4.52123061e-02 4.71536636e-01 3.18361104e-01
-2.41501451e-01 7.22627580e-01 -9.24404502e-01 -8.67005706e-01
-1.63421854e-01 -5.68279803e-01 3.20638925e-01 5.04875898e-01
-3.09803933e-01 -9.23967063e-01 7.39617050e-02 5.86438179e-01
1.57471642e-01 8.82138252e-01 3.76000196e-01 -6.97989762e-01
-2.67936170e-01 -3.63006294e-01 -1.69396684e-01 1.71138551e-02
-6.69880807e-02 5.12045264e-01 -1.23809671e+00 -2.82975972e-01
-2.41430625e-01 -5.49339354e-01 2.76920736e-01 1.72461882e-01
1.08874142e+00 -3.20050508e-01 4.56818799e-03 5.08561671e-01
1.38883281e+00 7.51672029e-01 9.15789247e-01 8.01245272e-01
8.14958990e-01 6.83842361e-01 2.29460731e-01 6.59488618e-01
2.00170889e-01 4.16429311e-01 7.43028700e-01 3.59431833e-01
-3.77087682e-01 -7.62833059e-01 4.44957703e-01 8.91498566e-01
-9.42216814e-02 -4.20446843e-01 -1.05610752e+00 1.05291545e+00
-1.50794315e+00 -1.29815269e+00 -2.36432821e-01 1.76912379e+00
2.00687513e-01 1.46449730e-01 3.56615841e-01 2.16918141e-01
7.59882331e-01 3.73699844e-01 -1.27758756e-01 -9.51570928e-01
1.99688494e-01 2.07617998e-01 2.89684623e-01 -5.09265773e-02
-1.35246336e+00 7.68854141e-01 6.45908022e+00 1.01231766e+00
-1.22137439e+00 3.25156093e-01 4.90125954e-01 -8.12756300e-01
-2.31001824e-02 -7.24474251e-01 -2.25449547e-01 1.09849107e+00
7.19110608e-01 3.88433635e-01 4.81792688e-01 8.21386576e-01
2.73324877e-01 -3.58457893e-01 -1.00373101e+00 1.24999607e+00
6.11103773e-01 -1.26338637e+00 -6.51574612e-01 -3.06453377e-01
8.69686663e-01 -9.51472204e-03 1.72927052e-01 7.66693801e-02
-1.06203705e-01 -1.16886103e+00 8.87596428e-01 6.42800704e-02
7.05683708e-01 -1.04354203e+00 3.42236102e-01 -7.46329576e-02
-6.62335575e-01 1.74422711e-01 -9.28376168e-02 -4.36478376e-01
-3.12666267e-01 2.79979110e-01 -5.33924460e-01 -8.39596018e-02
1.22792327e+00 9.41889286e-01 -4.43933398e-01 1.42088544e+00
-2.53178149e-01 8.43536377e-01 -2.26773381e-01 -4.69899595e-01
4.55526173e-01 1.05948355e-02 9.41216528e-01 1.61028159e+00
4.73261684e-01 1.17104277e-01 -8.02186131e-01 2.03373685e-01
-8.51527750e-02 3.41936350e-01 -1.09391022e+00 -2.40182534e-01
1.83188900e-01 1.03730333e+00 -1.11584055e+00 8.48562717e-02
-1.07267082e+00 9.63899851e-01 3.03295881e-01 -4.58066128e-02
-6.79041803e-01 -4.05658305e-01 6.37548685e-01 2.77384669e-01
4.00608629e-01 1.00872710e-01 -8.05557519e-02 -1.18387628e+00
-1.19568743e-01 -9.42070663e-01 1.97580263e-01 -1.01374197e+00
-7.93867171e-01 -2.95689851e-01 4.06690016e-02 -1.10315752e+00
-8.17588493e-02 -3.89913917e-01 -1.03160143e+00 -1.55196905e-01
-8.62402320e-01 -9.74674881e-01 -2.39686996e-01 2.27546737e-01
6.14393532e-01 -1.86000429e-02 2.53759772e-01 7.88253069e-01
-6.55859947e-01 8.25594723e-01 1.56870753e-01 2.59334743e-01
5.89238584e-01 -1.16309214e+00 3.64084214e-01 8.68496358e-01
-1.77338898e-01 1.94226131e-01 1.20222986e+00 -8.05824876e-01
-1.24457765e+00 -3.12407374e-01 9.12252426e-01 -3.51631135e-01
1.09494030e+00 -3.47485632e-01 -6.30913556e-01 5.69686294e-01
7.41460502e-01 -3.86265993e-01 8.19909990e-01 -1.02492169e-01
-2.64643133e-01 3.26073110e-01 -1.34620357e+00 7.67815471e-01
1.09248805e+00 -2.25612924e-01 -4.20613915e-01 2.88932085e-01
2.04638883e-01 -9.44609165e-01 -7.50280619e-01 -2.28718877e-01
1.25190389e+00 -1.50983202e+00 8.04311752e-01 -5.65441668e-01
1.29343593e+00 4.66191798e-01 1.87635362e-01 -9.55189109e-01
-7.16417432e-02 -8.94703269e-01 -3.34765345e-01 1.39100027e+00
-2.06220120e-01 -2.48343974e-01 1.26667750e+00 5.97172320e-01
-7.75882006e-02 -6.28125846e-01 -1.05123079e+00 -3.01016748e-01
2.03913659e-01 -6.94377244e-01 4.44035977e-01 1.36405218e+00
5.02936006e-01 -6.22244656e-01 -1.01857531e+00 -1.59240514e-02
3.75308037e-01 -4.84942257e-01 4.61336195e-01 -6.86892688e-01
-6.94839805e-02 -5.39237142e-01 -8.32961798e-01 -4.22036588e-01
-2.83059895e-01 -3.19400221e-01 -7.28453755e-01 -1.25583172e+00
3.42446029e-01 4.64080006e-01 -6.15222007e-02 2.80213028e-01
3.31230253e-01 7.16432691e-01 4.23185647e-01 -3.65081310e-01
-7.95735478e-01 -3.00743908e-01 1.31254685e+00 4.63010781e-02
5.19641526e-02 -3.33449304e-01 -5.35732508e-01 1.25922430e+00
6.92958295e-01 -2.65864730e-01 -3.73767823e-01 -3.03830415e-01
8.99339437e-01 6.37787879e-02 2.35927045e-01 -1.15522563e+00
-4.82294187e-02 1.45260379e-01 6.56999648e-01 -9.85647380e-01
3.74450505e-01 -5.32226980e-01 1.73232585e-01 3.94749016e-01
4.47580591e-02 -1.90133706e-01 2.74023771e-01 9.55630541e-02
2.36605909e-02 -7.02937245e-01 8.78357112e-01 -3.88783552e-02
-7.70580292e-01 -1.69560730e-01 -1.12127078e+00 -3.45894811e-03
1.05691385e+00 -4.23235178e-01 -7.07929134e-01 -9.99936581e-01
-7.14299679e-01 1.43117178e-02 5.81201136e-01 7.21625030e-01
5.85110605e-01 -1.39626539e+00 -2.94845551e-01 1.67424291e-01
-2.43699908e-01 -7.13307798e-01 5.08856535e-01 5.93986392e-01
-1.27427173e+00 4.09323901e-01 -7.62444377e-01 -4.61258590e-02
-1.71126413e+00 5.85470557e-01 -9.37146228e-03 3.73579562e-01
-9.56364155e-01 8.97040665e-01 -2.60234326e-01 3.50236177e-01
4.66566741e-01 2.08469126e-02 -8.16041112e-01 6.38630807e-01
9.48840559e-01 1.17243874e+00 -4.87710945e-02 -6.09306216e-01
-2.82548130e-01 9.34685469e-02 -1.59697279e-01 1.71859935e-01
1.27721977e+00 -1.70747265e-01 -3.58741395e-02 3.31928730e-01
1.30097520e+00 4.90845650e-01 -8.88336360e-01 6.43327773e-01
-3.22245032e-01 -9.93927538e-01 -2.34967675e-02 -4.71075773e-01
-1.36903393e+00 8.75424027e-01 6.90810621e-01 9.01225924e-01
9.73380089e-01 -2.25654423e-01 1.02910733e+00 -4.14173044e-02
2.77684808e-01 -1.26877022e+00 5.06815612e-01 1.47345141e-01
6.71406269e-01 -1.07947874e+00 1.80777147e-01 -3.83993089e-01
-1.70953125e-01 1.38240099e+00 7.78647244e-01 -2.29170442e-01
4.39131171e-01 2.37331584e-01 -1.74694750e-02 -3.32893163e-01
-6.36334300e-01 3.72475147e-01 5.12587801e-02 8.79311740e-01
6.04124010e-01 1.40066510e-02 -6.98407531e-01 4.49604601e-01
-4.43288445e-01 1.31282523e-01 1.32323217e+00 8.63252163e-01
-5.08109093e-01 -7.41534293e-01 -5.76153934e-01 5.54069817e-01
-1.46711957e+00 -9.76376049e-03 -4.07733589e-01 1.04143119e+00
6.00198328e-01 1.02838039e+00 6.96058869e-02 -2.14567974e-01
-1.60395250e-01 -9.73253325e-02 5.62177658e-01 -3.59741569e-01
-8.32176983e-01 7.17553943e-02 6.38363659e-01 -8.50309312e-01
-3.68912846e-01 -1.03862119e+00 -3.45205426e-01 -9.95259166e-01
2.04214647e-01 -5.38228810e-01 5.83455861e-01 5.66079378e-01
-3.99062216e-01 2.27789402e-01 2.51348317e-01 -7.80527771e-01
8.28882039e-01 -5.97773612e-01 -7.77232945e-01 3.36721927e-01
5.54332793e-01 -7.21567988e-01 -1.48625135e-01 -4.77726638e-01]
|
[12.453344345092773, 1.1874504089355469]
|
51939322-c0bc-4c76-948c-cc1f933825a8
|
automated-stance-detection-in-complex-topics
|
2305.13047
| null |
https://arxiv.org/abs/2305.13047v1
|
https://arxiv.org/pdf/2305.13047v1.pdf
|
Automated stance detection in complex topics and small languages: the challenging case of immigration in polarizing news media
|
Automated stance detection and related machine learning methods can provide useful insights for media monitoring and academic research. Many of these approaches require annotated training datasets, which limits their applicability for languages where these may not be readily available. This paper explores the applicability of large language models for automated stance detection in a challenging scenario, involving a morphologically complex, lower-resource language, and a socio-culturally complex topic, immigration. If the approach works in this case, it can be expected to perform as well or better in less demanding scenarios. We annotate a large set of pro and anti-immigration examples, and compare the performance of multiple language models as supervised learners. We also probe the usability of ChatGPT as an instructable zero-shot classifier for the same task. Supervised achieves acceptable performance, and ChatGPT yields similar accuracy. This is promising as a potentially simpler and cheaper alternative for text classification tasks, including in lower-resource languages. We further use the best-performing model to investigate diachronic trends over seven years in two corpora of Estonian mainstream and right-wing populist news sources, demonstrating the applicability of the approach for news analytics and media monitoring settings, and discuss correspondences between stance changes and real-world events.
|
['Maximilian Schich', 'Indrek Ibrus', 'Andres Karjus', 'Mark Mets']
|
2023-05-22
| null | null | null | null |
['stance-detection']
|
['natural-language-processing']
|
[-1.41812205e-01 2.42711473e-02 -6.78474069e-01 -1.27665430e-01
-1.24800098e+00 -6.91693485e-01 1.12184489e+00 8.87380183e-01
-8.73344421e-01 9.26334441e-01 7.68977106e-01 -5.06800711e-01
1.55771464e-01 -7.73892105e-01 -3.27212512e-01 -4.75801319e-01
1.35442078e-01 8.87842953e-01 3.68673086e-01 -6.35047436e-01
3.42833251e-01 -1.99293941e-02 -1.55320442e+00 5.49068689e-01
9.39566076e-01 3.89427423e-01 -6.64008036e-02 3.60907257e-01
-4.43219423e-01 7.84318030e-01 -8.96507919e-01 -8.37112963e-01
-1.49924800e-01 -1.43590927e-01 -1.00565696e+00 -2.43455276e-01
5.37187994e-01 3.38411778e-01 1.63556620e-01 7.05245376e-01
7.04604566e-01 -1.06935561e-01 6.72592580e-01 -5.74483931e-01
-1.44540936e-01 1.13914907e+00 -4.41755652e-01 6.94989502e-01
6.92080200e-01 -1.60798222e-01 1.15612674e+00 -4.50327039e-01
1.15653884e+00 1.41156256e+00 8.53977978e-01 1.98088866e-03
-1.21722865e+00 -5.97700536e-01 2.45161891e-01 1.21259458e-01
-6.86606228e-01 -5.14838398e-01 7.33490705e-01 -8.04077387e-01
7.16080070e-01 1.40601471e-01 7.94029415e-01 1.58449090e+00
8.93454403e-02 8.23347628e-01 1.34467196e+00 -6.27240956e-01
6.88956957e-03 4.92261976e-01 3.29734445e-01 3.80314976e-01
2.53232777e-01 -4.29257423e-01 -8.57534528e-01 -4.40155268e-01
-2.28304058e-01 -4.75230187e-01 -2.24088747e-02 2.13746905e-01
-1.39724529e+00 1.24324811e+00 -8.56489092e-02 8.85586023e-01
-1.80792391e-01 -4.53988045e-01 8.89223874e-01 6.58792794e-01
1.38107693e+00 6.26757979e-01 -4.76899981e-01 -6.16526127e-01
-1.19083261e+00 5.45645177e-01 1.06799197e+00 4.13047522e-01
-2.83185132e-02 -7.04446957e-02 -1.22061566e-01 1.12259889e+00
1.14802606e-02 3.50432605e-01 6.61978662e-01 -5.20907104e-01
8.07181835e-01 4.93532181e-01 -5.29152714e-02 -8.86161923e-01
-6.15831435e-01 -5.29102445e-01 -2.40422890e-01 -2.34510452e-02
1.04413724e+00 -1.44563988e-01 -2.14829415e-01 1.63997340e+00
5.52937984e-01 -5.25641263e-01 -4.54610065e-02 4.84910846e-01
9.60694492e-01 6.47223592e-01 3.24680179e-01 -6.36644125e-01
1.59183252e+00 -4.11602825e-01 -5.32446980e-01 -4.06716883e-01
1.06016660e+00 -1.18915141e+00 1.29190123e+00 5.13635576e-01
-1.18953836e+00 -1.20427929e-01 -7.62904525e-01 -5.26472330e-02
-6.62513137e-01 -1.08874686e-01 4.98441458e-01 7.21889138e-01
-5.09702563e-01 6.07750058e-01 -5.55206299e-01 -8.17872584e-01
3.23308527e-01 -1.42459646e-01 3.35946754e-02 2.74647981e-01
-1.32442915e+00 1.07745481e+00 3.05463225e-01 -5.09451807e-01
-1.58429176e-01 -7.42179096e-01 -7.09128737e-01 -4.05863911e-01
3.39354247e-01 6.78594736e-03 1.40764630e+00 -1.00786567e+00
-1.23586571e+00 1.57046509e+00 1.57891974e-01 -4.73589689e-01
8.59494448e-01 -1.97943375e-01 -4.78253037e-01 -2.62286335e-01
5.64676344e-01 -3.95383537e-02 5.27864337e-01 -5.22708476e-01
-6.54438794e-01 -3.92196953e-01 7.61454320e-03 1.62844658e-01
-6.37138247e-01 8.25297773e-01 2.68578202e-01 -1.02239335e+00
-1.70457408e-01 -7.02029586e-01 1.86007604e-01 -4.85995680e-01
-1.83656681e-02 -4.42561716e-01 5.66692829e-01 -9.70192552e-01
1.50724995e+00 -1.76644278e+00 -3.71409580e-02 2.95698717e-02
-9.64507461e-02 1.21433936e-01 3.16533118e-01 5.57316065e-01
2.26847038e-01 1.84959561e-01 1.93781465e-01 -1.20002814e-01
7.70053547e-03 1.42770618e-01 -1.79917619e-01 7.18023360e-01
-1.23852491e-01 8.04311693e-01 -9.80748773e-01 -6.12837434e-01
-3.26059341e-01 1.13580182e-01 -5.99519014e-01 -4.32293564e-01
-3.81041199e-01 4.69234824e-01 -1.27450094e-01 5.83557785e-01
-1.94152817e-01 7.54672661e-02 3.68095815e-01 2.16052264e-01
-6.28681421e-01 1.03783143e+00 -8.10380697e-01 1.25240946e+00
-4.83122587e-01 1.17533565e+00 -6.31375471e-03 -1.08999479e+00
9.06723559e-01 2.99225658e-01 3.36249053e-01 -1.01706314e+00
3.88294846e-01 4.84393448e-01 3.78086656e-01 -6.77189529e-01
4.28752840e-01 -3.69448990e-01 -4.94368345e-01 6.95376873e-01
-1.52321443e-01 -8.30314457e-02 5.81635654e-01 8.98754820e-02
7.30125308e-01 1.38208522e-02 6.91625297e-01 -7.69773722e-01
4.02491868e-01 2.40017220e-01 4.62242067e-01 5.49117923e-01
-1.57595381e-01 9.16110650e-02 6.06805921e-01 -4.92007792e-01
-1.01595533e+00 -4.64663833e-01 -4.87227350e-01 1.86003518e+00
-5.31844497e-01 -5.71390092e-01 -4.94133860e-01 -5.54913402e-01
-1.08335234e-01 8.67684007e-01 -5.19303501e-01 2.84927070e-01
-8.30072224e-01 -1.09583628e+00 5.57642639e-01 3.65324765e-02
1.96889848e-01 -1.05223989e+00 -5.83012938e-01 4.91236806e-01
-5.16216040e-01 -9.40102518e-01 1.84765365e-03 1.55624673e-01
-5.11018336e-01 -1.03125286e+00 -7.28021801e-01 -8.55250180e-01
-2.75210351e-01 -2.77906865e-01 1.39242554e+00 -1.60299346e-01
-2.03089360e-02 6.17147721e-02 -4.21332806e-01 -8.86112094e-01
-9.12020564e-01 5.39630353e-01 2.24747628e-01 -1.85133949e-01
4.46048886e-01 -4.31117266e-01 5.46700433e-02 -5.99127151e-02
-4.16769832e-01 -2.21335396e-01 -4.76903878e-02 7.92820036e-01
-3.58712897e-02 -3.80359441e-01 7.28537619e-01 -1.22764528e+00
8.15104425e-01 -8.27528059e-01 -3.33380818e-01 5.02186045e-02
-4.85304624e-01 -2.55095989e-01 4.93766040e-01 -7.11962163e-01
-1.04167175e+00 -6.97910905e-01 -3.03127974e-01 4.72593427e-01
-6.48589507e-02 9.18445885e-01 1.56742066e-01 4.01407063e-01
1.17763960e+00 -2.98224270e-01 -2.49702379e-01 -6.95553958e-01
8.73987284e-03 9.28410709e-01 1.19207397e-01 -6.63859069e-01
6.51541293e-01 3.78895760e-01 -4.55787838e-01 -1.27625430e+00
-1.30773807e+00 -5.88883936e-01 -6.67934954e-01 -3.51366043e-01
7.75321960e-01 -9.02372897e-01 -4.48481411e-01 3.27182263e-01
-1.00965548e+00 -5.06646037e-01 -2.20941484e-01 4.23451155e-01
-4.11926299e-01 1.72769085e-01 -7.51475275e-01 -7.33358085e-01
-3.74386013e-01 -6.44048512e-01 8.66247594e-01 -1.24139525e-01
-8.79529774e-01 -1.41300869e+00 4.39231485e-01 5.70266128e-01
1.22163542e-01 6.15008056e-01 1.23186457e+00 -1.20880449e+00
3.61434698e-01 -9.28819552e-02 3.27547193e-01 -2.91263759e-01
-1.68674394e-01 5.78248799e-02 -9.10523295e-01 -2.93867558e-01
-1.32489353e-01 -6.76937342e-01 5.97099900e-01 1.20815292e-01
3.25023383e-01 -5.56510925e-01 -1.15141541e-01 -1.00979634e-01
8.96925151e-01 -8.91670138e-02 1.30478442e-01 9.15242612e-01
1.48101076e-01 7.60277629e-01 7.52326846e-01 4.23119336e-01
4.59651262e-01 8.32773268e-01 -3.15069139e-01 1.25496477e-01
-7.01853409e-02 -1.11045331e-01 6.79144681e-01 1.26495898e+00
-1.92417711e-01 -1.09131321e-01 -1.40004337e+00 6.42444134e-01
-1.75344205e+00 -1.44224644e+00 -3.83581132e-01 1.97972274e+00
1.21360803e+00 7.45586395e-01 7.47390449e-01 3.52914363e-01
3.12310427e-01 3.20725083e-01 1.55760823e-02 -6.41301751e-01
-3.93908173e-01 3.62964988e-01 -2.34392611e-03 4.38349694e-01
-1.20656872e+00 8.59593451e-01 6.34260082e+00 8.20681632e-01
-1.12789714e+00 6.44930780e-01 5.04343450e-01 -3.55177581e-01
-1.51857555e-01 -2.69364715e-01 -8.35126519e-01 6.80842340e-01
1.30612898e+00 -3.06364417e-01 -9.33489501e-02 7.61365712e-01
4.75411028e-01 -3.44877988e-01 -9.11420166e-01 7.29858935e-01
3.03886145e-01 -1.42897964e+00 -2.98985600e-01 7.46114552e-03
6.60898387e-01 2.62010753e-01 -1.47771060e-01 6.47866488e-01
2.34396353e-01 -5.76406419e-01 1.14026737e+00 1.37564048e-01
6.55785739e-01 -5.05871594e-01 6.01362586e-01 6.51434124e-01
-8.00693870e-01 -2.09187299e-01 -6.92382753e-02 -5.01550615e-01
1.37648240e-01 4.12463784e-01 -8.62919450e-01 1.55606225e-01
7.63091207e-01 8.23708475e-01 -7.60990024e-01 8.96321774e-01
-7.34186172e-02 1.03352499e+00 -3.58881027e-01 -4.40771401e-01
2.81005532e-01 -5.09825982e-02 8.68356466e-01 1.76180983e+00
1.27959177e-01 -5.73332123e-02 4.49350774e-01 1.73560679e-01
-7.19353091e-03 8.18670750e-01 -6.34864986e-01 -1.59022748e-01
3.15041661e-01 1.04752374e+00 -9.03578877e-01 -4.80981141e-01
-5.75529456e-01 3.85489017e-01 5.75010121e-01 -9.07021239e-02
-6.36235356e-01 -6.95562810e-02 4.22864199e-01 6.49111688e-01
-1.39494091e-01 -6.63549230e-02 -4.11369681e-01 -1.27110040e+00
-2.51530856e-01 -1.24529684e+00 8.09604883e-01 -2.72583991e-01
-1.31582510e+00 2.92581201e-01 1.56134948e-01 -1.11458838e+00
-4.43206728e-01 -4.77821678e-01 -6.19508088e-01 5.15062273e-01
-1.25011981e+00 -1.29186594e+00 1.44591942e-01 3.93548608e-01
9.41016316e-01 -3.11644554e-01 6.52005553e-01 4.76867884e-01
-4.50204104e-01 4.58529621e-01 2.04268485e-01 8.97441059e-02
9.51340735e-01 -1.03399944e+00 1.19166195e-01 6.40801609e-01
3.26337188e-01 1.92840382e-01 1.05937421e+00 -8.26791048e-01
-6.87857568e-01 -9.32510436e-01 1.41118944e+00 -6.15273416e-01
1.28847480e+00 -6.50658727e-01 -7.94394612e-01 5.37591219e-01
3.84297758e-01 -9.18944478e-01 9.37121749e-01 7.22105801e-01
-4.42943335e-01 3.76635045e-01 -8.30006361e-01 7.82872617e-01
1.03384578e+00 -4.72855717e-01 -1.01011598e+00 9.04204309e-01
3.63521188e-01 -3.10631186e-01 -9.70024943e-01 -4.62877862e-02
5.73433280e-01 -7.50679612e-01 6.56800926e-01 -1.02839589e+00
6.11032546e-01 3.67139310e-01 -4.49111238e-02 -1.33163786e+00
-3.00729334e-01 -6.62550092e-01 1.11196332e-01 1.54318964e+00
7.35518932e-01 -4.83925968e-01 4.51504588e-01 1.13587290e-01
-8.00843239e-02 -5.69590509e-01 -1.04377532e+00 -7.37045288e-01
6.82168245e-01 -6.05700910e-01 6.70326203e-02 1.42967200e+00
4.28044766e-01 8.06263864e-01 -1.59967586e-01 -4.20106113e-01
2.09895164e-01 2.36531913e-01 6.48177922e-01 -1.94770145e+00
-2.27968976e-01 -8.12964916e-01 -3.15811217e-01 -2.28840142e-01
4.30216879e-01 -1.12559211e+00 -4.14927125e-01 -1.23133409e+00
1.73793808e-01 -4.65179205e-01 3.57752174e-01 3.48405331e-01
9.12848115e-02 4.61555541e-01 1.02094620e-01 3.40304285e-01
-4.18911338e-01 9.65735763e-02 7.74196208e-01 -2.25188807e-01
-5.10430157e-01 2.58061945e-01 -5.71209311e-01 1.07322371e+00
8.03921759e-01 -5.41591167e-01 1.73596200e-02 -4.68472429e-02
5.68467200e-01 -4.00412112e-01 4.56583127e-03 -6.76963806e-01
-6.71396330e-02 -2.03982949e-01 2.04895228e-01 -2.61908263e-01
1.85954869e-01 -2.83740848e-01 -1.95862025e-01 4.79141235e-01
-4.03152913e-01 2.58897215e-01 1.58105329e-01 1.56475618e-01
-1.20081373e-01 -2.77089059e-01 8.75675082e-01 -3.25059325e-01
-2.43840471e-01 -9.35348645e-02 -8.68143499e-01 8.12263131e-01
9.46526051e-01 -1.61370888e-01 -4.76341426e-01 -3.78008902e-01
-7.85591841e-01 -9.96099636e-02 3.33901048e-01 5.69791615e-01
-8.85661319e-02 -1.08652663e+00 -1.14179063e+00 -1.41909510e-01
2.03640655e-01 -7.04689085e-01 -4.19330388e-01 1.17091966e+00
-3.92264634e-01 3.20056617e-01 5.65240458e-02 -4.42895740e-01
-1.56895947e+00 3.86258364e-01 7.54221007e-02 -4.33323532e-01
-5.43172300e-01 5.60505509e-01 -3.71339202e-01 -5.19380748e-01
1.93652198e-01 -5.54644838e-02 -6.58044875e-01 1.15139008e+00
5.02945483e-01 3.22748065e-01 2.40038723e-01 -8.43990564e-01
-1.44922540e-01 1.85472891e-01 -1.32106021e-01 -4.05375272e-01
1.51649559e+00 -7.42548034e-02 -2.11182609e-02 1.14524400e+00
8.62809360e-01 4.89540309e-01 -3.83307636e-01 -4.31330234e-01
6.47010803e-01 -1.62974149e-01 -6.67640939e-02 -6.35370433e-01
-3.05433005e-01 7.26287305e-01 9.51638818e-02 5.48776686e-01
5.06811857e-01 3.51085782e-01 4.59236830e-01 2.06187651e-01
2.54000098e-01 -1.54144740e+00 -2.44579706e-02 8.57581496e-01
8.54219496e-01 -1.21032989e+00 2.77014703e-01 -7.57512227e-02
-4.48119193e-01 1.03402925e+00 2.12239474e-01 2.95196176e-01
6.53640449e-01 1.77494511e-01 1.15347035e-01 -3.31834286e-01
-7.38013387e-01 -2.31076911e-01 2.06109077e-01 2.80978113e-01
9.44988847e-01 1.46932201e-02 -9.71288919e-01 4.38049197e-01
-9.04830158e-01 -5.38044572e-01 5.99090099e-01 9.01964724e-01
-5.83562315e-01 -1.13047123e+00 -4.66809720e-01 6.26905501e-01
-9.66586590e-01 -1.73910484e-01 -6.50257289e-01 1.18174052e+00
1.57253325e-01 7.97938883e-01 6.98286071e-02 4.89576645e-02
2.90677965e-01 5.62898934e-01 2.82114267e-01 -8.44600916e-01
-1.11629033e+00 3.70267063e-01 8.84274900e-01 -1.45289063e-01
-7.13352680e-01 -1.36474979e+00 -6.01523340e-01 -5.51043212e-01
-8.79639015e-02 1.98013052e-01 4.64172155e-01 1.18764794e+00
-2.14270994e-01 1.35903731e-01 2.18720227e-01 -6.70573175e-01
-4.07461345e-01 -1.14029443e+00 -2.97730893e-01 5.16404450e-01
6.52607977e-02 -6.54094815e-01 -2.04679102e-01 -2.81798523e-02]
|
[9.048691749572754, 10.001863479614258]
|
c607c910-7812-4a23-8d48-0312c8cb948c
|
bridging-the-gap-between-decision-and-logits
|
2306.08909
| null |
https://arxiv.org/abs/2306.08909v1
|
https://arxiv.org/pdf/2306.08909v1.pdf
|
Bridging the Gap between Decision and Logits in Decision-based Knowledge Distillation for Pre-trained Language Models
|
Conventional knowledge distillation (KD) methods require access to the internal information of teachers, e.g., logits. However, such information may not always be accessible for large pre-trained language models (PLMs). In this work, we focus on decision-based KD for PLMs, where only teacher decisions (i.e., top-1 labels) are accessible. Considering the information gap between logits and decisions, we propose a novel method to estimate logits from the decision distributions. Specifically, decision distributions can be both derived as a function of logits theoretically and estimated with test-time data augmentation empirically. By combining the theoretical and empirical estimations of the decision distributions together, the estimation of logits can be successfully reduced to a simple root-finding problem. Extensive experiments show that our method significantly outperforms strong baselines on both natural language understanding and machine reading comprehension datasets.
|
['Yang Liu', 'Peng Li', 'Zonghan Yang', 'Qinhong Zhou']
|
2023-06-15
| null | null | null | null |
['reading-comprehension', 'machine-reading-comprehension']
|
['natural-language-processing', 'natural-language-processing']
|
[ 8.44160616e-02 5.08383393e-01 -4.08458263e-01 -6.48091555e-01
-8.30161154e-01 -8.57523918e-01 3.08790267e-01 5.68951607e-01
-6.25747859e-01 6.96447492e-01 5.50618395e-02 -9.38306153e-01
4.82474305e-02 -7.52756417e-01 -1.06661522e+00 -3.64390880e-01
5.17991722e-01 6.29559457e-01 2.00803906e-01 3.07530705e-02
-4.52793464e-02 5.78543171e-02 -1.55744147e+00 -1.32988960e-01
1.63163638e+00 1.02187657e+00 4.15688932e-01 6.19593024e-01
-4.31031525e-01 8.85759115e-01 -6.33406401e-01 -5.51658273e-01
-2.41579235e-01 -2.03559920e-01 -9.65978324e-01 -2.72523195e-01
6.21697605e-01 -8.44956875e-01 -2.38820314e-01 1.06214380e+00
3.30584764e-01 2.43924767e-01 1.04387558e+00 -1.07759106e+00
-7.88255036e-01 1.11389089e+00 -6.80567622e-01 2.88142320e-02
2.93902636e-01 1.60187334e-01 9.90053535e-01 -8.22599292e-01
8.20624456e-02 1.43598795e+00 1.83003604e-01 1.98021054e-01
-1.28151810e+00 -6.86099470e-01 5.40429652e-01 5.84033549e-01
-1.37781692e+00 -1.75790548e-01 5.62683702e-01 -5.12120485e-01
6.70862734e-01 -1.87202662e-01 4.71597940e-01 9.88588154e-01
-4.61252570e-01 1.43185556e+00 1.39756513e+00 -8.95559669e-01
1.96594864e-01 3.23110074e-01 5.00217140e-01 7.15602279e-01
2.67242312e-01 -2.44478375e-01 -5.89933574e-01 1.79516241e-01
5.74474454e-01 -4.40032542e-01 -3.63502294e-01 -2.66286016e-01
-1.00573277e+00 7.95872450e-01 1.31618723e-01 -1.93191722e-01
-1.05341032e-01 -5.92637695e-02 4.42429632e-02 5.24477363e-01
3.95481020e-01 3.01793784e-01 -1.03033328e+00 -4.38798487e-01
-6.01574719e-01 6.41773343e-02 1.03262174e+00 1.08412623e+00
1.00218642e+00 -2.18576401e-01 -2.91655838e-01 1.01137149e+00
3.79024565e-01 7.70452142e-01 3.49115789e-01 -5.59217036e-01
6.74045086e-01 6.34150505e-01 9.52844471e-02 -7.18714535e-01
-3.92584428e-02 -2.39339054e-01 -6.29142404e-01 -4.15961355e-01
9.72341418e-01 -3.62063050e-01 -1.04517603e+00 2.08630848e+00
6.37872756e-01 3.92679572e-01 4.84637991e-02 4.33017999e-01
9.03805912e-01 5.45672715e-01 1.44234046e-01 -1.09596692e-01
1.45456445e+00 -8.44470978e-01 -6.98013365e-01 -3.33152741e-01
9.33065474e-01 -4.98820812e-01 1.49131501e+00 5.62707007e-01
-9.52859342e-01 -3.60823929e-01 -7.87182748e-01 -3.55190456e-01
-3.25097740e-01 4.00810033e-01 3.30962837e-01 5.17502308e-01
-9.00828063e-01 1.33428991e-01 -7.67832994e-01 1.63069308e-01
3.96902919e-01 3.19844246e-01 1.58909429e-02 -3.55258793e-01
-1.28757548e+00 6.81471944e-01 4.83027488e-01 -1.65611595e-01
-9.54592288e-01 -8.35570216e-01 -1.06265271e+00 3.56158495e-01
8.54208529e-01 -4.88777280e-01 1.91110516e+00 -3.45730960e-01
-1.81800616e+00 5.98874569e-01 -2.13757873e-01 -2.34919056e-01
3.47733438e-01 -5.17023444e-01 3.89838248e-01 2.26375628e-02
-1.22554917e-02 7.70764768e-01 6.27486825e-01 -8.97055149e-01
-6.83865309e-01 -1.19982816e-01 3.60104591e-01 4.06040698e-01
-6.44746900e-01 -3.27566087e-01 -5.42234242e-01 -4.65975076e-01
-2.31742319e-02 -7.05108702e-01 -1.11412823e-01 -9.46824774e-02
-7.26030350e-01 -1.02127302e+00 1.63783416e-01 -6.79177523e-01
1.50256348e+00 -1.76178193e+00 -1.10231921e-01 2.87531465e-01
4.03086632e-01 3.01119000e-01 -2.38456473e-01 -1.19588457e-01
3.08098704e-01 1.97588980e-01 6.57106787e-02 -2.26649091e-01
3.74331355e-01 2.35958934e-01 -2.86488384e-01 -6.88248351e-02
1.65633038e-01 1.00889778e+00 -1.12837982e+00 -7.10472882e-01
1.26939103e-01 -6.07874878e-02 -6.65981591e-01 6.25590324e-01
-6.14602983e-01 2.59532571e-01 -5.49539864e-01 5.38572371e-01
6.46137595e-01 -3.21805716e-01 2.28938028e-01 1.10300720e-01
1.35784373e-01 9.76674259e-01 -1.13198733e+00 1.50550687e+00
-7.70005047e-01 4.80788648e-01 -1.32879972e-01 -1.28231192e+00
6.25887752e-01 9.11764577e-02 -6.19131625e-02 -4.88605887e-01
3.52533311e-02 4.39466164e-02 1.76812410e-01 -3.56491804e-01
2.11760849e-01 1.28062829e-01 -8.23291764e-02 7.70101726e-01
3.30322474e-01 -9.58854407e-02 2.56543964e-01 2.55040705e-01
8.83406341e-01 2.27337331e-03 4.44734782e-01 -1.69373125e-01
3.50242943e-01 -1.82457194e-01 3.68942291e-01 1.03221381e+00
1.09515071e-01 -7.11394176e-02 8.65843654e-01 9.84357521e-02
-4.52003300e-01 -1.14841044e+00 -6.95953071e-02 1.48912311e+00
-7.81701431e-02 -3.94420326e-01 -9.09123480e-01 -9.99600828e-01
1.40304357e-01 9.25474048e-01 -3.60366851e-01 -2.32101843e-01
-4.40814257e-01 -3.38987678e-01 5.24151623e-01 7.00522065e-01
4.07082200e-01 -6.16254210e-01 -2.33810484e-01 2.86030322e-01
-2.07290292e-01 -1.35753965e+00 -4.00577307e-01 3.33852679e-01
-6.50763094e-01 -9.49501574e-01 -5.41343927e-01 -7.59281695e-01
7.80531108e-01 1.46198645e-01 1.45760572e+00 6.32695705e-02
1.52856529e-01 4.11438167e-01 -2.42026716e-01 -8.28123987e-01
-3.57346207e-01 3.44912350e-01 2.39350453e-01 -4.30056214e-01
8.83587420e-01 -6.44076586e-01 -4.40244585e-01 2.44484484e-01
-6.06724977e-01 4.34979886e-01 8.58049572e-01 7.59583831e-01
7.66807437e-01 1.12055115e-01 8.04116666e-01 -9.59383786e-01
7.67087996e-01 -4.56323892e-01 -8.79059613e-01 6.17194355e-01
-7.08201826e-01 5.00587225e-01 6.05149984e-01 -1.04558659e+00
-1.19650340e+00 -2.63989400e-02 -4.75208368e-03 -3.55391711e-01
-4.08161432e-01 8.19384038e-01 -3.67166579e-01 1.16766103e-01
3.75518799e-01 1.66679606e-01 -3.08225483e-01 -8.14857364e-01
7.18567073e-01 8.49276841e-01 5.99541306e-01 -1.30154216e+00
7.23263681e-01 -2.82612652e-01 -4.65123832e-01 -6.50815189e-01
-1.63160348e+00 -4.15528774e-01 -6.60151482e-01 1.34069398e-01
4.36321616e-01 -1.16224074e+00 -9.67567503e-01 6.46115482e-01
-1.07317328e+00 -6.69382513e-01 -2.62014180e-01 6.37915790e-01
-2.97729075e-01 2.05777422e-01 -6.74448907e-01 -7.99048066e-01
4.20785062e-02 -1.23249078e+00 7.33474493e-01 4.35411960e-01
-8.24677646e-02 -1.16345084e+00 -1.31587029e-01 5.33529341e-01
1.79747865e-01 -4.71685201e-01 1.45331609e+00 -8.85849237e-01
-6.78270578e-01 1.53792828e-01 -2.50528485e-01 5.18312871e-01
1.19596245e-02 -1.61841556e-01 -1.12597692e+00 6.93034902e-02
-2.23574042e-01 -8.97225738e-01 8.70399654e-01 4.29545283e-01
1.58340073e+00 -6.49698377e-01 -2.31513470e-01 2.97621191e-01
8.57911646e-01 -2.45046392e-01 1.80674568e-02 6.53804168e-02
8.51317346e-01 5.18851817e-01 6.48925185e-01 4.31382805e-01
1.20497251e+00 4.50442255e-01 5.66410199e-02 1.63821667e-01
-9.39395279e-02 -8.59371483e-01 4.43814874e-01 1.29390526e+00
4.42207217e-01 -3.97069544e-01 -1.36070478e+00 7.32762516e-01
-1.50601828e+00 8.25199261e-02 3.28697413e-02 2.13364959e+00
1.79914236e+00 1.51897267e-01 -2.66241550e-01 -1.09447271e-01
3.97421092e-01 -3.35597187e-01 -1.00395763e+00 -2.92137533e-01
3.16882074e-01 3.88311803e-01 2.73045599e-01 6.78168297e-01
-8.64520490e-01 1.09991324e+00 5.72224283e+00 1.16882253e+00
-7.10357487e-01 8.98782816e-03 6.52845383e-01 1.89738169e-01
-5.56930244e-01 -4.56654392e-02 -1.23079658e+00 3.02969456e-01
1.01943147e+00 -2.50667095e-01 3.17576557e-01 7.26101995e-01
-1.55944809e-01 -3.83539915e-01 -1.38484979e+00 8.48423183e-01
-2.95385182e-01 -7.40073621e-01 1.26874208e-01 -9.61773749e-03
9.47224677e-01 -1.81606293e-01 1.33789644e-01 9.53880608e-01
8.83581102e-01 -1.05189443e+00 5.06992102e-01 2.86347717e-01
8.93955588e-01 -6.92833006e-01 6.56221747e-01 1.08986890e+00
-8.68103504e-01 4.82167564e-02 -4.49274033e-01 -1.80962116e-01
-3.47603381e-01 8.56170177e-01 -1.13548088e+00 9.92364734e-02
3.86329025e-01 5.33825576e-01 -5.98022282e-01 6.82320833e-01
-1.16851294e+00 1.29420495e+00 -7.81484783e-01 -2.57878155e-01
1.26913324e-01 -1.81359380e-01 -8.93478617e-02 1.03800678e+00
1.99806422e-01 4.54348296e-01 3.38287473e-01 1.07869613e+00
-5.63665688e-01 2.30806440e-01 -3.19670826e-01 -2.28817508e-01
9.27805424e-01 1.02103925e+00 -1.34172618e-01 -4.83138233e-01
-5.68248212e-01 5.57464004e-01 1.01524794e+00 5.33444166e-01
-4.73527908e-01 -2.14800179e-01 6.67335212e-01 -1.12259485e-01
2.22881630e-01 -2.23070592e-01 -2.00857729e-01 -1.33025575e+00
4.72953618e-02 -9.99765515e-01 3.45150292e-01 -5.98440945e-01
-1.31796086e+00 -1.32091999e-01 5.07001102e-01 -6.69712186e-01
-5.71576238e-01 -4.94264692e-01 -5.45979381e-01 9.31098163e-01
-1.95508552e+00 -7.02813327e-01 -3.39389704e-02 3.46365958e-01
3.95008117e-01 2.03547686e-01 5.58109879e-01 -1.15081053e-02
-7.94625044e-01 1.22432911e+00 1.51905999e-01 1.23538010e-01
7.42195308e-01 -1.66236222e+00 2.44725049e-01 5.42045414e-01
2.00874001e-01 5.39350092e-01 3.25201035e-01 -3.89289021e-01
-1.29467356e+00 -7.95887411e-01 8.65995705e-01 -5.12899160e-01
8.51730406e-01 -3.95212561e-01 -1.24083185e+00 6.85630262e-01
-9.70916972e-02 -3.53191942e-01 7.59928882e-01 4.77850795e-01
-4.48925108e-01 1.64337099e-01 -8.30104649e-01 5.95596611e-01
7.18187869e-01 -6.05206251e-01 -8.03635955e-01 2.81379163e-01
9.63395774e-01 -7.61038005e-01 -9.85742688e-01 2.71794587e-01
2.53311902e-01 -2.46073991e-01 9.07952309e-01 -6.89236164e-01
5.60342073e-01 -4.12471704e-02 2.66825229e-01 -1.58063924e+00
3.49512577e-01 -3.50427091e-01 -7.30328143e-01 1.36694992e+00
5.25436997e-01 -4.65019315e-01 6.24244809e-01 8.34151804e-01
1.07506588e-01 -1.11125302e+00 -7.56333709e-01 -6.05970979e-01
4.39940482e-01 -6.12106264e-01 7.19308674e-01 1.06406057e+00
-1.12777390e-01 5.18806815e-01 9.56796184e-02 4.27652746e-01
5.35287321e-01 2.65718609e-01 8.31054091e-01 -1.36148453e+00
-3.86988819e-01 -3.39346260e-01 7.13608935e-02 -2.12041450e+00
6.66817904e-01 -7.05447197e-01 3.23832661e-01 -1.44375098e+00
3.24420035e-01 -8.53000820e-01 -3.81876200e-01 8.79592955e-01
-7.74915755e-01 -5.48056126e-01 -1.55405194e-01 -2.51856685e-01
-6.69045269e-01 6.65061176e-01 1.55444074e+00 -8.03199485e-02
-1.63250089e-01 1.28115732e-02 -7.54925013e-01 1.11981380e+00
8.76032531e-01 -5.66324174e-01 -8.82926226e-01 -5.84912717e-01
2.94758856e-01 1.26684949e-01 -1.19074754e-01 -4.18630600e-01
4.06417400e-01 -5.36501884e-01 7.89387152e-02 -7.30027676e-01
8.19379017e-02 -4.26081419e-01 -9.71807778e-01 -4.32422385e-02
-7.59861827e-01 -3.92889380e-01 2.42584422e-01 7.21873045e-01
-1.95332766e-01 -4.78556812e-01 3.24614227e-01 1.26640961e-01
-6.28333151e-01 2.53526151e-01 -2.25972742e-01 5.37368000e-01
6.15359783e-01 2.33301669e-01 -5.22017181e-01 -6.16265237e-01
-4.97754455e-01 9.26955283e-01 4.83204871e-02 2.09195375e-01
4.71617609e-01 -1.04340231e+00 -6.75604284e-01 1.17107540e-01
1.04797445e-01 9.51529264e-01 -2.87839603e-02 7.88833678e-01
2.06917152e-02 5.59008718e-01 4.16373581e-01 -5.26638448e-01
-1.14729202e+00 3.36794585e-01 1.02461830e-01 -6.28888786e-01
-2.19358802e-01 1.23675811e+00 5.39213181e-01 -8.80526781e-01
6.77457631e-01 -9.41523135e-01 -3.30400258e-01 4.53873239e-02
7.21881986e-01 -1.28705492e-02 -1.79212540e-01 6.45212606e-02
1.12366825e-01 2.97412932e-01 -3.96522492e-01 -3.08831804e-03
1.07813156e+00 -2.69845039e-01 1.15695514e-01 4.66431260e-01
8.21927786e-01 -1.42354127e-02 -1.39998591e+00 -8.58935237e-01
2.59842843e-01 -2.43840948e-01 2.20375463e-01 -1.11812341e+00
-7.48818040e-01 1.35424054e+00 8.91980622e-03 -2.36128762e-01
1.17630589e+00 2.99769700e-01 6.93198204e-01 1.04096961e+00
2.41537958e-01 -1.02831328e+00 3.05742305e-02 7.38447666e-01
3.17555755e-01 -1.50515842e+00 -1.83828026e-01 -6.18528187e-01
-3.03773344e-01 9.99303877e-01 1.07679904e+00 6.05738759e-01
6.01284027e-01 2.22695336e-01 -1.03568465e-01 2.12599292e-01
-1.18903708e+00 -4.01455730e-01 5.84752321e-01 5.76459885e-01
5.45780361e-01 4.31043684e-01 4.47482243e-02 1.02327168e+00
-5.65286994e-01 -1.11659177e-01 6.04720473e-01 7.32314169e-01
-6.55694127e-01 -1.04757512e+00 -1.36140779e-01 6.75128520e-01
-2.04779282e-01 -5.70394158e-01 -3.46779883e-01 6.47038341e-01
-5.32055460e-02 1.05757236e+00 -1.61643714e-01 -1.55361414e-01
3.43955517e-01 3.00582588e-01 5.90771616e-01 -9.10563469e-01
-9.73796919e-02 -4.40457046e-01 5.79571538e-02 -1.80742443e-01
1.15127536e-02 -2.88950175e-01 -1.27186215e+00 -1.94494024e-01
-7.96009064e-01 1.28116623e-01 6.40372276e-01 1.29519033e+00
-7.37531334e-02 3.79481554e-01 3.43545735e-01 1.32724389e-01
-1.26492882e+00 -1.18990040e+00 -4.28441525e-01 -4.17852774e-02
3.08227450e-01 -6.98015809e-01 -3.51502895e-01 -2.53204852e-02]
|
[10.873868942260742, 8.237993240356445]
|
6eb25317-6805-417c-9377-376e660fb469
|
exploration-via-epistemic-value-estimation
|
2303.04012
| null |
https://arxiv.org/abs/2303.04012v1
|
https://arxiv.org/pdf/2303.04012v1.pdf
|
Exploration via Epistemic Value Estimation
|
How to efficiently explore in reinforcement learning is an open problem. Many exploration algorithms employ the epistemic uncertainty of their own value predictions -- for instance to compute an exploration bonus or upper confidence bound. Unfortunately the required uncertainty is difficult to estimate in general with function approximation. We propose epistemic value estimation (EVE): a recipe that is compatible with sequential decision making and with neural network function approximators. It equips agents with a tractable posterior over all their parameters from which epistemic value uncertainty can be computed efficiently. We use the recipe to derive an epistemic Q-Learning agent and observe competitive performance on a series of benchmarks. Experiments confirm that the EVE recipe facilitates efficient exploration in hard exploration tasks.
|
['Hado van Hasselt', 'John Shawe-Taylor', 'Simon Schmitt']
|
2023-03-07
| null | null | null | null |
['efficient-exploration']
|
['methodology']
|
[-2.23153636e-01 7.25342095e-01 -4.86054212e-01 -2.37278253e-01
-1.22794020e+00 -7.08422720e-01 4.92280453e-01 1.92313865e-02
-7.64905989e-01 1.56232572e+00 -2.78249718e-02 -6.05041802e-01
-5.66684306e-01 -9.23793077e-01 -1.08698034e+00 -8.63527536e-01
-4.65531051e-01 7.82049060e-01 1.23155542e-01 -1.00030333e-01
3.50813568e-01 -1.14363573e-01 -1.29311311e+00 -5.05657457e-02
1.01564193e+00 1.44259930e+00 -8.36052597e-02 7.21306682e-01
-3.26341093e-02 1.01602578e+00 -6.78633273e-01 -2.53994793e-01
5.78954697e-01 -3.57633233e-01 -9.01684940e-01 -4.64231074e-01
-5.04086673e-01 -7.75438130e-01 1.79687470e-01 1.30352581e+00
1.99371815e-01 3.73685777e-01 7.01199114e-01 -1.23181283e+00
-3.59054804e-01 1.39165032e+00 -3.64052206e-01 2.35960335e-02
1.75553977e-01 4.50710505e-01 1.09655845e+00 -1.87027574e-01
4.58564609e-01 1.12003314e+00 5.12222588e-01 5.91519833e-01
-1.21918023e+00 -3.17450166e-01 2.57280469e-01 1.62089050e-01
-8.74256730e-01 7.77494386e-02 4.01524663e-01 -1.15411460e-01
8.04715216e-01 -1.08922392e-01 8.98699701e-01 1.08425534e+00
3.17557395e-01 9.91852105e-01 1.60053849e+00 -2.56829768e-01
1.06241119e+00 1.48381338e-01 -2.91622013e-01 5.59448898e-01
2.39122450e-01 9.72467780e-01 -5.68626046e-01 -1.41905665e-01
6.77600145e-01 -4.16746378e-01 -5.50935306e-02 -5.99522769e-01
-1.05423641e+00 1.10936129e+00 3.79447401e-01 -3.86157840e-01
-4.58833516e-01 9.48638678e-01 3.91609877e-01 7.86266267e-01
1.85567260e-01 9.59512651e-01 -5.87251127e-01 -7.00587571e-01
-5.15141666e-01 4.97721493e-01 1.16557145e+00 5.48212409e-01
5.46247423e-01 1.44212350e-01 -1.82379812e-01 2.02491254e-01
3.16988140e-01 5.04930973e-01 2.70269901e-01 -1.66132569e+00
2.70397484e-01 -7.56386071e-02 8.40530217e-01 9.28052813e-02
-3.44363213e-01 -3.95472407e-01 -1.01451539e-01 1.04399657e+00
8.37155044e-01 -6.27853692e-01 -7.24883199e-01 1.92911994e+00
3.17735851e-01 2.05163866e-01 3.94674480e-01 7.33552933e-01
-8.35860241e-03 4.78039980e-01 1.11617379e-01 -2.71490842e-01
1.00804675e+00 -8.87272477e-01 -6.36775136e-01 -2.59694755e-02
6.13988340e-01 1.24051414e-01 9.72337961e-01 9.63723063e-01
-1.22652900e+00 4.18093149e-03 -1.18370748e+00 4.61980641e-01
-1.75097585e-01 -6.11304939e-01 1.03190672e+00 9.48794067e-01
-9.48092580e-01 1.15314102e+00 -1.14916730e+00 5.30887127e-01
5.89675963e-01 3.85876954e-01 9.58369449e-02 6.54216468e-01
-1.45694029e+00 1.29142630e+00 9.99239862e-01 -9.49063059e-03
-1.62123346e+00 -5.13957739e-01 -6.64396286e-01 8.45122412e-02
1.09839749e+00 -5.97711325e-01 2.00546694e+00 -8.84369910e-01
-2.36088061e+00 6.91741034e-02 5.89226365e-01 -1.15425515e+00
9.33349729e-01 -4.47302043e-01 1.68828487e-01 3.37730050e-02
-2.51548290e-01 6.91162765e-01 9.44304347e-01 -9.52909350e-01
-8.81428719e-01 4.02525021e-03 4.51650262e-01 3.82692784e-01
-2.20819619e-02 -4.71853614e-01 1.89810157e-01 -1.66582972e-01
-3.15644652e-01 -8.82587492e-01 -6.41009331e-01 -6.02165945e-02
-3.40341985e-01 -4.51394320e-01 -1.55606970e-01 -2.04671562e-01
7.79133677e-01 -1.58513236e+00 1.18455298e-01 5.50243020e-01
7.27044269e-02 -3.83346379e-01 1.28188297e-01 1.52337492e-01
3.52109164e-01 5.97085804e-02 -2.81070709e-01 -4.56996672e-02
9.05649722e-01 3.98698717e-01 -6.64801359e-01 4.50193673e-01
-1.21561490e-01 1.15310550e+00 -1.15921760e+00 -3.19223553e-01
8.87344107e-02 -6.08150735e-02 -6.83508992e-01 2.71304935e-01
-9.59228277e-01 1.75932020e-01 -6.42491102e-01 4.48303849e-01
2.67990023e-01 -2.58539617e-01 2.96462297e-01 5.79724550e-01
-4.27389145e-02 3.96773845e-01 -1.49492013e+00 1.55544364e+00
-3.59379232e-01 2.96950966e-01 -3.06577012e-02 -8.37971151e-01
7.68419266e-01 3.18442993e-02 2.45360255e-01 -4.76308584e-01
3.48258495e-01 4.38743591e-01 -8.98105726e-02 -1.31829679e-01
6.32206321e-01 -3.19049209e-01 -3.63207906e-01 8.70903075e-01
-1.85182448e-02 -4.22212511e-01 1.55503273e-01 -9.36854035e-02
1.14846694e+00 7.77239621e-01 3.11400115e-01 -5.72661042e-01
1.04473315e-01 1.79366767e-02 5.18749893e-01 1.31435335e+00
-3.83391112e-01 -1.52702197e-01 1.14647400e+00 -4.57612902e-01
-8.65427673e-01 -1.40981936e+00 -1.70574993e-01 1.32110703e+00
6.20915443e-02 -8.76738429e-02 -8.14772844e-01 -9.08076763e-01
3.07219863e-01 1.07177711e+00 -1.04514277e+00 -1.91440478e-01
-1.88034967e-01 -4.96601909e-01 2.90055215e-01 8.28573525e-01
3.03124696e-01 -1.21357238e+00 -1.19059193e+00 3.05055052e-01
3.12861383e-01 -3.71468276e-01 2.36494727e-02 9.59077299e-01
-7.01450884e-01 -9.07577753e-01 -4.99853849e-01 -7.13849440e-02
2.08129823e-01 -7.33003438e-01 1.19424117e+00 -2.48266518e-01
3.85407269e-01 4.27604169e-01 -9.08391178e-02 -6.08118296e-01
-3.82115483e-01 -4.31665592e-02 9.61045176e-02 -7.33538806e-01
2.85182208e-01 -4.51156288e-01 -4.91587132e-01 4.60644029e-02
-6.24922156e-01 -2.09008560e-01 3.82296741e-01 1.07618308e+00
6.17821574e-01 2.23060459e-01 7.81507552e-01 -7.77221799e-01
9.66031909e-01 -5.80342770e-01 -1.35919058e+00 3.24959606e-01
-8.35775971e-01 8.67529333e-01 5.63478470e-01 -3.54148328e-01
-1.25359786e+00 -6.54726997e-02 -2.18520965e-02 -1.29805744e-01
2.80287951e-01 6.34613156e-01 2.58751810e-01 -1.30591355e-02
8.02169681e-01 -4.57825325e-02 7.66516030e-02 -2.48295262e-01
5.54756641e-01 2.21898705e-01 6.88983023e-01 -1.35619605e+00
4.89181519e-01 2.14961395e-01 1.82762712e-01 6.88892826e-02
-1.15168262e+00 3.23297441e-01 -8.10055528e-03 -1.45423114e-01
4.88453239e-01 -8.48804295e-01 -1.48062408e+00 -4.98234220e-02
-6.41520441e-01 -9.98703301e-01 -1.03392732e+00 6.74210072e-01
-1.34833944e+00 9.93055031e-02 -4.12433535e-01 -1.35802495e+00
-1.51617780e-01 -1.30097687e+00 5.37373006e-01 5.40015936e-01
-3.36813480e-02 -9.12197292e-01 2.19061226e-01 -1.23350210e-01
3.84779662e-01 4.72982824e-01 5.63966334e-01 -5.30703068e-01
-7.75667369e-01 3.53146881e-01 8.99571404e-02 1.68397725e-01
-6.68670774e-01 -1.65381789e-01 -8.66920888e-01 -1.22267470e-01
7.54844630e-03 -9.19267476e-01 1.10075986e+00 6.48584843e-01
1.32963610e+00 -3.44293147e-01 -6.45424947e-02 6.47269011e-01
1.21779215e+00 1.49008170e-01 4.25498843e-01 8.36016059e-01
-2.09793568e-01 5.41552961e-01 9.22000945e-01 1.13131499e+00
2.81021923e-01 2.64524490e-01 8.69172752e-01 1.04283142e+00
6.70289934e-01 -4.19915110e-01 6.15731359e-01 8.02172571e-02
-1.25094235e-01 8.14181566e-02 -7.51777947e-01 4.26705539e-01
-2.04623723e+00 -8.55701804e-01 4.98070955e-01 2.32490182e+00
1.44283831e+00 7.08569109e-01 3.73079747e-01 -2.49700144e-01
2.04690188e-01 -2.07585827e-01 -1.25256586e+00 -7.02324450e-01
1.92516595e-01 2.72864103e-01 8.00515473e-01 8.73160064e-01
-9.26681519e-01 7.28801727e-01 6.86176682e+00 9.78846490e-01
-3.44465256e-01 2.25814879e-01 6.42773986e-01 -3.77545208e-01
-7.36469567e-01 1.47724956e-01 -8.61296773e-01 5.13856113e-01
1.29583561e+00 -3.31605911e-01 8.60041499e-01 1.36525202e+00
-3.15501720e-01 -5.76015413e-01 -1.22828102e+00 6.80600703e-01
-7.81623006e-01 -1.39018440e+00 -7.48632967e-01 1.33802414e-01
9.54144001e-01 7.83086866e-02 2.23001003e-01 9.16181445e-01
1.53099287e+00 -1.32408905e+00 1.11545527e+00 6.33213341e-01
6.00365758e-01 -1.29878390e+00 7.57089972e-01 4.42729920e-01
-5.93668878e-01 -5.66700280e-01 -5.55911660e-01 -1.42059922e-01
2.22528532e-01 2.91759044e-01 -8.26020896e-01 7.91357905e-02
5.27457654e-01 3.32451351e-02 -2.64992230e-02 1.00799072e+00
-9.18868363e-01 5.49304783e-01 -8.54507387e-01 -5.86945355e-01
7.41297662e-01 -3.42732817e-01 3.58167470e-01 6.70154631e-01
3.87393355e-01 -8.24298784e-02 5.99628016e-02 1.29532135e+00
-7.40888119e-02 -4.13418740e-01 -7.33361840e-02 -2.99332529e-01
6.19607747e-01 8.97824705e-01 -6.46828175e-01 -2.03254208e-01
2.09384084e-01 3.85199010e-01 6.77124858e-01 2.18085080e-01
-9.37526405e-01 -4.03536886e-01 5.57327032e-01 -6.16578341e-01
3.90731007e-01 -6.39763400e-02 -3.81291360e-01 -8.32538903e-01
-1.42971575e-01 -8.70036423e-01 5.94338834e-01 -5.43921769e-01
-1.07356215e+00 3.51837128e-01 2.48250082e-01 -8.56020987e-01
-1.08110988e+00 -9.46784139e-01 -4.00422961e-01 6.68051660e-01
-1.69515717e+00 -4.75707799e-01 2.09661379e-01 3.91027093e-01
9.79937017e-02 -8.73106495e-02 7.90734529e-01 -7.99524784e-01
-3.09282929e-01 4.82118845e-01 4.94798034e-01 -3.50779831e-01
2.23503888e-01 -2.10493398e+00 1.61382988e-01 2.53853887e-01
-6.29445165e-02 3.08394134e-01 1.04086637e+00 -4.92435932e-01
-1.53483045e+00 -3.09704691e-01 -1.08764321e-01 -7.12670386e-01
1.02347565e+00 6.10268489e-02 -5.80784678e-01 6.79152608e-01
3.03521067e-01 5.54892905e-02 4.86793399e-01 3.91623527e-01
-3.10697913e-01 1.16736665e-01 -1.08195579e+00 5.22293329e-01
6.98639393e-01 -3.43323022e-01 -8.88787806e-01 1.90602526e-01
8.00851643e-01 -7.79863894e-01 -1.12181616e+00 1.86418399e-01
7.04375327e-01 -1.08342361e+00 7.60728836e-01 -8.29178154e-01
3.81547511e-01 -1.50968611e-01 -1.11352652e-01 -1.64092529e+00
2.15072840e-01 -1.36576200e+00 -8.94602060e-01 4.98267382e-01
4.19327557e-01 -7.55413353e-01 9.23262894e-01 1.04613304e+00
5.06924428e-02 -1.04390085e+00 -1.22372115e+00 -1.08198416e+00
5.34240305e-01 -6.95574343e-01 1.00416946e+00 3.99739504e-01
4.31774586e-01 -3.01875383e-01 -3.44150364e-01 -1.15998819e-01
1.09524989e+00 2.84836888e-01 4.22611028e-01 -1.05811167e+00
-9.10757482e-01 -7.25646436e-01 3.19419533e-01 -9.78141487e-01
3.64637285e-01 -5.07035792e-01 2.80387998e-01 -1.17559326e+00
-5.85905388e-02 -5.17934501e-01 -6.27817512e-01 5.17324269e-01
8.88796449e-02 -3.30365539e-01 1.06488287e-01 -3.11701983e-01
-1.13428664e+00 8.28984261e-01 1.22714722e+00 1.08107395e-01
-3.14937651e-01 9.28628594e-02 -7.32982993e-01 9.07687128e-01
1.09225106e+00 -5.97850919e-01 -4.64485288e-01 -1.61922332e-02
1.10233676e+00 4.11108255e-01 3.02695572e-01 -7.95911729e-01
2.51624525e-01 -6.09272480e-01 3.63191217e-01 -4.47946638e-01
2.60962695e-01 -5.68774045e-01 3.83392945e-02 5.59873879e-01
-7.80236244e-01 -2.65472353e-01 7.01002702e-02 7.85492778e-01
2.42681444e-01 -8.90068889e-01 6.64239228e-01 -2.53884494e-01
-7.82227755e-01 -5.73436432e-02 -2.50855565e-01 3.66830111e-01
1.08976114e+00 -4.81136739e-02 -3.94903272e-01 -5.25174022e-01
-8.43301654e-01 6.18727148e-01 2.14245304e-01 -1.45978257e-01
4.19138312e-01 -1.28436100e+00 -5.42788744e-01 -7.90909678e-03
-1.73352927e-01 -3.55037563e-02 1.94376446e-02 6.37172520e-01
-3.02709281e-01 2.28891835e-01 -4.04942006e-01 -1.39160886e-01
-2.01442331e-01 5.58018446e-01 6.61790133e-01 -6.30797923e-01
-3.27270657e-01 1.06838429e+00 -3.44705641e-01 -3.76410246e-01
5.69124043e-01 -4.99718904e-01 7.37102702e-02 -6.35173544e-02
7.31218100e-01 4.36679572e-01 -5.91488361e-01 3.40693563e-01
8.87473971e-02 -8.56434032e-02 7.01934621e-02 -7.77173460e-01
1.39002478e+00 -9.41451713e-02 3.37407500e-01 4.25783306e-01
5.29497921e-01 -4.41972584e-01 -2.17877626e+00 -3.21220130e-01
2.85069048e-01 -4.22216266e-01 3.14558268e-01 -1.13047481e+00
-5.89561164e-01 6.92034185e-01 4.22174871e-01 4.12203521e-01
6.41987562e-01 -2.36068800e-01 4.47561026e-01 1.14382231e+00
8.46179187e-01 -1.89944160e+00 -6.48238063e-02 4.22728986e-01
7.86857724e-01 -1.23270249e+00 7.25167021e-02 5.96460342e-01
-8.59708965e-01 1.20741379e+00 5.86630702e-01 -2.06205353e-01
6.09938741e-01 5.54740191e-01 -3.21874797e-01 4.36742371e-03
-1.17118192e+00 -2.85186201e-01 -2.31793690e-02 4.68157232e-01
-3.13013762e-01 4.20745701e-01 -3.40858549e-02 9.18355465e-01
-4.53927159e-01 1.81478038e-01 5.36614776e-01 8.28052402e-01
-8.82780433e-01 -9.77997661e-01 -3.32238615e-01 5.67598343e-01
-4.88778979e-01 1.47757545e-01 2.00495467e-01 6.26594067e-01
-2.18668059e-01 7.37027764e-01 -9.11046788e-02 -6.13288023e-02
-2.82405853e-01 -1.39410433e-03 6.88554704e-01 -1.84849113e-01
-5.61595976e-01 -1.71147570e-01 2.78932154e-01 -9.93729472e-01
-6.20564818e-02 -6.23305857e-01 -1.41271913e+00 -3.81400257e-01
-2.54915237e-01 6.00650966e-01 6.87929690e-01 9.45258141e-01
-4.19807024e-02 3.88729751e-01 3.58659297e-01 -5.79866648e-01
-1.95532870e+00 -7.55875230e-01 -9.69365954e-01 -2.34106943e-01
2.66030312e-01 -9.46071804e-01 -5.81822515e-01 -5.04395187e-01]
|
[4.134439945220947, 2.407958507537842]
|
5898814e-c555-41e5-a917-3d95e7c3a44d
|
mlseg-image-and-video-segmentation-as-multi
|
2203.04187
| null |
https://arxiv.org/abs/2203.04187v2
|
https://arxiv.org/pdf/2203.04187v2.pdf
|
RankSeg: Adaptive Pixel Classification with Image Category Ranking for Segmentation
|
The segmentation task has traditionally been formulated as a complete-label pixel classification task to predict a class for each pixel from a fixed number of predefined semantic categories shared by all images or videos. Yet, following this formulation, standard architectures will inevitably encounter various challenges under more realistic settings where the scope of categories scales up (e.g., beyond the level of 1k). On the other hand, in a typical image or video, only a few categories, i.e., a small subset of the complete label are present. Motivated by this intuition, in this paper, we propose to decompose segmentation into two sub-problems: (i) image-level or video-level multi-label classification and (ii) pixel-level rank-adaptive selected-label classification. Given an input image or video, our framework first conducts multi-label classification over the complete label, then sorts the complete label and selects a small subset according to their class confidence scores. We then use a rank-adaptive pixel classifier to perform the pixel-wise classification over only the selected labels, which uses a set of rank-oriented learnable temperature parameters to adjust the pixel classifications scores. Our approach is conceptually general and can be used to improve various existing segmentation frameworks by simply using a lightweight multi-label classification head and rank-adaptive pixel classifier. We demonstrate the effectiveness of our framework with competitive experimental results across four tasks, including image semantic segmentation, image panoptic segmentation, video instance segmentation, and video semantic segmentation. Especially, with our RankSeg, Mask2Former gains +0.8%/+0.7%/+0.7% on ADE20K panoptic segmentation/YouTubeVIS 2019 video instance segmentation/VSPW video semantic segmentation benchmarks respectively.
|
['Han Hu', 'Xiangyu Yue', 'Yuhui Yuan', 'Haodi He']
|
2022-03-08
| null | null | null | null |
['video-instance-segmentation']
|
['computer-vision']
|
[ 8.71001065e-01 -2.08708003e-01 -6.02087438e-01 -6.16032600e-01
-1.09853625e+00 -6.65752828e-01 9.71421525e-02 -2.16253512e-02
-5.33312023e-01 4.17497426e-01 -5.87107658e-01 -1.90304175e-01
-7.98262060e-02 -5.68345249e-01 -8.36776257e-01 -9.95368659e-01
2.40646377e-01 3.28557521e-01 6.29865527e-01 2.81546891e-01
3.75068098e-01 1.23878062e-01 -1.69615090e+00 5.43243945e-01
7.66581416e-01 1.60819209e+00 3.46395016e-01 6.02533221e-01
-7.07559362e-02 6.51964664e-01 -3.80493432e-01 4.35358286e-03
3.96961451e-01 -2.53478289e-01 -1.04335189e+00 5.31786561e-01
7.82767355e-01 -2.51346946e-01 1.61157772e-01 1.33644187e+00
2.52546519e-01 2.29067400e-01 4.82529402e-01 -1.29386878e+00
-5.29669337e-02 5.76252401e-01 -9.16329443e-01 5.63544109e-02
1.58016533e-02 7.12756068e-02 1.08844006e+00 -6.27531528e-01
4.73799586e-01 1.05902934e+00 4.93929058e-01 6.23673022e-01
-1.17867076e+00 -8.09620142e-01 6.13886595e-01 1.80970043e-01
-1.29876208e+00 -1.06527358e-01 8.00465286e-01 -3.97544980e-01
4.53375041e-01 4.07378554e-01 4.58434284e-01 7.40828514e-01
5.89505136e-02 9.56637502e-01 1.61906672e+00 -7.07678795e-02
3.22244763e-01 4.65648510e-02 5.25053442e-01 6.67734027e-01
-1.87269688e-01 -3.21363121e-01 -3.77748162e-01 -3.04372497e-02
5.14449954e-01 1.68018460e-01 -3.65047008e-01 -1.97443128e-01
-1.19977415e+00 6.96921110e-01 3.05769563e-01 8.09166580e-02
-5.19294329e-02 4.52164680e-01 6.15903556e-01 4.10998575e-02
4.72418010e-01 1.56018943e-01 -8.60023320e-01 1.34285718e-01
-1.27759922e+00 6.50207745e-03 5.07107496e-01 9.11764622e-01
1.15951312e+00 -3.54645282e-01 -2.10653603e-01 1.17623127e+00
2.57766008e-01 4.34985578e-01 5.19119799e-01 -1.31859958e+00
2.91883916e-01 4.23219323e-01 -1.54668123e-01 -8.08545172e-01
-4.23689902e-01 -1.78049311e-01 -7.57307827e-01 1.52853385e-01
2.64936537e-01 -5.91893820e-03 -1.45831537e+00 1.71596885e+00
4.54820096e-01 6.12300217e-01 -1.43780634e-01 1.04530668e+00
7.77303636e-01 9.11099136e-01 3.07544082e-01 -4.74332869e-01
1.50603116e+00 -1.39625347e+00 -3.46982062e-01 -2.56597400e-01
5.31805873e-01 -5.77054977e-01 1.12174380e+00 5.50538838e-01
-9.07538474e-01 -7.81451225e-01 -9.35733855e-01 1.62474394e-01
-2.70805925e-01 6.20965064e-02 5.27218997e-01 7.39933491e-01
-1.07005227e+00 5.40483475e-01 -5.04549205e-01 -1.14094511e-01
4.97683227e-01 6.14029884e-01 2.10681260e-02 -1.85480908e-01
-1.08012784e+00 3.03372554e-02 7.04619646e-01 -1.35073379e-01
-9.47907627e-01 -7.18062222e-01 -6.59264743e-01 -2.64196157e-01
7.84833014e-01 -3.65179539e-01 1.08600640e+00 -1.45820272e+00
-1.30869365e+00 1.14738488e+00 -1.18622705e-01 -2.50148267e-01
3.98887902e-01 5.49868681e-02 -2.04399988e-01 4.43104982e-01
3.10928673e-01 1.19213474e+00 1.05277979e+00 -1.48224473e+00
-1.25770164e+00 -1.22996025e-01 3.50235462e-01 3.61234784e-01
-2.72233844e-01 -8.76407698e-02 -1.03080094e+00 -7.00073779e-01
3.97125512e-01 -1.32702434e+00 -3.45891267e-01 -1.10307708e-01
-5.13248503e-01 -2.76533842e-01 9.73694026e-01 -3.56126785e-01
1.15460443e+00 -2.18118358e+00 5.12595847e-02 2.59802163e-01
1.03819825e-01 2.32790355e-02 -1.52306944e-01 -4.91511196e-01
-6.00893423e-02 3.75887573e-01 -4.89393383e-01 -1.84496760e-01
-2.24453241e-01 2.05309957e-01 -4.72716689e-02 3.81610006e-01
-6.56382143e-02 6.33566916e-01 -8.45504224e-01 -8.07286024e-01
4.00260985e-01 1.18094824e-01 -5.34951091e-01 5.79485856e-02
-5.10014534e-01 2.91264802e-01 -6.76734388e-01 9.52180982e-01
7.51756430e-01 -4.42593485e-01 1.77943278e-02 -4.67981249e-01
8.09674859e-02 -2.47720197e-01 -1.44124603e+00 1.66770208e+00
-2.89405257e-01 2.93420911e-01 2.35383078e-01 -1.24028504e+00
4.74764913e-01 5.90760261e-02 9.53378856e-01 -4.79755312e-01
8.73213448e-03 3.33237022e-01 -4.28953975e-01 -2.46067509e-01
4.50089663e-01 3.92833427e-02 -3.51173133e-01 3.74885798e-01
-1.57696411e-01 -2.70088375e-01 3.49694490e-01 7.72259459e-02
8.83237839e-01 1.40897900e-01 1.65663827e-02 -3.08697790e-01
7.13991284e-01 1.27742171e-01 8.06980789e-01 8.15123975e-01
-5.65865636e-01 8.28595936e-01 3.81644577e-01 -3.17575365e-01
-6.66364849e-01 -7.82623053e-01 -3.91996503e-01 1.39783633e+00
7.60644674e-01 -2.43641481e-01 -1.04627573e+00 -9.49446499e-01
-2.05071911e-01 1.84577450e-01 -3.92302066e-01 3.58370952e-02
-6.54119909e-01 -1.08097196e+00 4.02237505e-01 3.35810274e-01
9.63020980e-01 -1.10841489e+00 -5.30272841e-01 5.49472086e-02
-1.92589432e-01 -1.26629448e+00 -7.76501060e-01 3.32879215e-01
-7.80902624e-01 -1.03858936e+00 -6.86849236e-01 -1.02041519e+00
5.92134237e-01 4.96808767e-01 8.65072310e-01 1.39050245e-01
-1.63859501e-02 3.87012929e-01 -3.99033844e-01 3.48260581e-01
-2.27821156e-01 5.51920235e-02 -1.43383682e-01 3.05910856e-01
-6.88516423e-02 -1.89196125e-01 -8.64715397e-01 6.72773182e-01
-1.09817374e+00 2.53589571e-01 5.78880906e-01 7.99234152e-01
1.21278250e+00 3.50358844e-01 3.96859467e-01 -1.23838770e+00
1.60098895e-02 -3.47160220e-01 -4.70981985e-01 3.79524708e-01
-6.51197731e-01 -3.24750870e-01 6.26396179e-01 -6.31428957e-01
-7.20449507e-01 4.54419136e-01 -7.37949694e-03 -5.53396642e-01
-2.52260149e-01 3.80707264e-01 -3.36338162e-01 -2.58126229e-01
7.51650855e-02 1.62914187e-01 -4.21972066e-01 -1.46245912e-01
4.44054097e-01 7.43396044e-01 4.43394244e-01 -6.54342949e-01
4.79028940e-01 3.83746713e-01 -4.18629572e-02 -5.63430727e-01
-9.67662573e-01 -7.65790999e-01 -4.85168457e-01 -5.39047480e-01
1.23946917e+00 -1.05093372e+00 -5.08257866e-01 8.39684725e-01
-6.48362815e-01 -7.09057748e-01 -1.05025917e-01 3.92079055e-02
-6.66887403e-01 5.42838037e-01 -8.33000302e-01 -2.91898429e-01
-4.17026728e-01 -1.76346505e+00 1.39888060e+00 3.96609366e-01
3.79309952e-02 -6.81761324e-01 -5.58008611e-01 8.80714715e-01
-6.07708935e-04 2.82390445e-01 9.66430485e-01 -4.83850986e-01
-8.60756814e-01 5.14331721e-02 -5.03105700e-01 6.32430553e-01
9.09083784e-02 -4.77921106e-02 -1.00541878e+00 -3.53575945e-01
-1.38794452e-01 -5.34810483e-01 1.11813200e+00 5.97736001e-01
1.84198308e+00 -6.75736293e-02 -5.20607591e-01 7.33665407e-01
1.56688702e+00 3.28483135e-01 2.26602867e-01 2.97324091e-01
1.09684658e+00 3.81943047e-01 1.08664155e+00 1.91796571e-01
2.57584721e-01 7.57755816e-01 4.22407389e-01 -1.16383247e-01
-1.89004466e-01 2.80507237e-01 2.82165051e-01 6.41944945e-01
2.72647113e-01 -3.21674794e-01 -8.01970840e-01 3.28158647e-01
-1.85491812e+00 -4.51944560e-01 -1.06386662e-01 1.99731517e+00
9.75789726e-01 3.11090469e-01 6.12298921e-02 1.35413736e-01
1.03373230e+00 4.58486438e-01 -1.06504381e+00 -9.32273567e-02
-9.22171324e-02 9.48992819e-02 8.86156499e-01 2.21720502e-01
-1.70466030e+00 1.08756137e+00 5.38773966e+00 1.43787205e+00
-1.15826321e+00 1.76179633e-01 1.36255562e+00 -8.91922787e-02
-6.86429590e-02 -6.19615838e-02 -8.46491814e-01 7.39582300e-01
7.83114791e-01 3.38658512e-01 4.02543604e-01 9.16047513e-01
9.40694064e-02 -3.07065606e-01 -9.83101070e-01 1.13592565e+00
8.92260671e-02 -1.06658065e+00 1.15565263e-01 -2.06976756e-01
1.07820809e+00 -6.44977689e-02 2.30818376e-01 2.33265013e-01
6.11622371e-02 -7.50889361e-01 9.53249454e-01 1.47238765e-02
1.07097769e+00 -4.84073937e-01 3.72762054e-01 1.05327219e-01
-1.34701192e+00 -2.01140285e-01 -7.72285387e-02 5.00308812e-01
6.56228140e-02 6.97437704e-01 -2.70649530e-02 2.02595130e-01
1.00763237e+00 9.31545913e-01 -4.14783388e-01 8.31036389e-01
-4.11444604e-02 7.15244651e-01 -3.26469958e-01 2.54955322e-01
6.36019647e-01 -2.61034548e-01 1.19389884e-01 1.26139736e+00
9.09869820e-02 2.97398359e-01 7.48216450e-01 2.13828802e-01
-3.24896812e-01 1.85777217e-01 1.78550243e-01 3.53704989e-01
3.83530736e-01 1.52131546e+00 -1.38905346e+00 -6.07058406e-01
-4.43779111e-01 1.24098837e+00 -1.31318018e-01 3.94286245e-01
-1.01075172e+00 -3.86666767e-02 6.41374826e-01 -1.29203498e-01
3.15876752e-01 1.75881162e-01 -4.05082971e-01 -9.41871464e-01
9.06991065e-02 -8.70129228e-01 6.20451152e-01 -7.08437383e-01
-8.75599265e-01 3.22749615e-01 2.57829716e-03 -1.12918258e+00
2.56753564e-01 -6.29096985e-01 -2.65050948e-01 4.06298995e-01
-1.67772055e+00 -1.10588884e+00 -4.47464406e-01 4.87117738e-01
1.08656430e+00 2.93666899e-01 3.25437993e-01 4.99654233e-01
-7.88715005e-01 4.55844730e-01 2.63433635e-01 5.04509080e-03
6.64253533e-01 -1.34082627e+00 -6.36397079e-02 5.77351272e-01
-7.01687410e-02 1.29386233e-02 3.12883258e-01 -4.42983687e-01
-9.82505798e-01 -1.50158823e+00 2.68648207e-01 -7.02971220e-02
4.73342896e-01 -3.01302105e-01 -7.81284034e-01 3.49014789e-01
-1.13190763e-01 3.19812685e-01 4.75275040e-01 -4.06309575e-01
-1.99347556e-01 -3.69030893e-01 -1.21253514e+00 5.78924716e-01
1.14958978e+00 -3.27383995e-01 1.73495021e-02 6.68501318e-01
1.00478327e+00 -4.86917704e-01 -9.99373913e-01 7.66108990e-01
4.84609127e-01 -7.43257642e-01 1.05714941e+00 -4.34830450e-02
4.61291909e-01 -4.88579422e-01 -2.77018130e-01 -9.22735035e-01
-1.63362816e-01 -3.49078000e-01 2.56512493e-01 1.22067440e+00
3.83641720e-01 -3.34728777e-01 1.04669142e+00 6.50863230e-01
-2.09071830e-01 -1.07415330e+00 -1.00574470e+00 -4.74914968e-01
-4.76931036e-02 -5.91388524e-01 3.15917075e-01 7.82119691e-01
-5.71267009e-01 -1.01344362e-02 -3.24751854e-01 2.19244823e-01
6.61851943e-01 4.74385828e-01 2.60624468e-01 -9.12630498e-01
-2.82592982e-01 -5.28099179e-01 -2.28250027e-01 -1.37336540e+00
4.15530026e-01 -8.50792289e-01 3.86034399e-01 -1.28654063e+00
4.78127867e-01 -9.36971188e-01 -7.85221577e-01 5.71919858e-01
-3.36915314e-01 6.44686162e-01 2.67596930e-01 4.18041795e-01
-1.14544344e+00 -5.90851530e-02 1.21126068e+00 -4.92301404e-01
-1.67653129e-01 9.13289413e-02 -5.43112457e-01 7.07543671e-01
7.66769886e-01 -5.56557119e-01 -5.59069574e-01 -2.08605528e-01
-7.86396116e-02 -5.38282841e-02 2.20212102e-01 -1.00870788e+00
2.75246631e-02 -4.73964393e-01 4.10787873e-02 -4.87682790e-01
3.31187963e-01 -8.59978318e-01 3.53961699e-02 4.25923616e-01
-4.44999754e-01 -4.53488976e-01 3.29441018e-02 5.96234977e-01
-1.69464841e-01 -3.17630380e-01 1.05642295e+00 -2.21138433e-01
-1.46052372e+00 4.80771273e-01 -3.59949112e-01 9.90385562e-02
1.38052106e+00 -5.74643612e-01 -2.06555873e-01 7.90842548e-02
-7.14395404e-01 4.56226856e-01 5.57001889e-01 3.87040168e-01
5.08664191e-01 -1.13444984e+00 -3.12451005e-01 2.94493092e-03
1.28355816e-01 1.98772833e-01 4.87726897e-01 7.33539164e-01
-4.12277997e-01 2.45130047e-01 6.19990677e-02 -1.02148283e+00
-1.43034470e+00 4.49869871e-01 3.36639047e-01 -2.11255580e-01
-3.66526961e-01 9.99108434e-01 5.29620886e-01 -2.91510284e-01
2.06859514e-01 -4.31397974e-01 -1.86007753e-01 1.59588590e-01
2.30481744e-01 2.73082823e-01 -9.70623866e-02 -9.11351144e-01
-4.00024205e-01 1.17113185e+00 -9.01623592e-02 1.86015368e-01
8.93282652e-01 -2.88933277e-01 -1.45097375e-01 4.73807871e-01
1.48468578e+00 -5.09278953e-01 -1.61739886e+00 -1.67190671e-01
-7.31462194e-03 -2.17298687e-01 1.72896639e-01 -7.01985896e-01
-1.59912539e+00 5.48038244e-01 8.34174156e-01 4.56777513e-02
1.61840379e+00 8.08248743e-02 1.07888710e+00 6.52827099e-02
4.30169880e-01 -1.47669828e+00 1.33096144e-01 1.76343516e-01
1.22987494e-01 -1.45426333e+00 -3.17264423e-02 -8.07852685e-01
-5.81150949e-01 8.60135555e-01 6.38756931e-01 5.34462258e-02
6.83552563e-01 5.39479740e-02 1.57615319e-01 3.27727571e-02
-4.75766212e-01 -2.21141800e-01 2.74974793e-01 1.72419176e-01
1.11595899e-01 3.23367774e-01 -3.53667647e-01 3.91329139e-01
3.59566152e-01 -1.89056888e-01 3.29690576e-01 7.26127207e-01
-7.46022105e-01 -9.71368194e-01 -3.54891390e-01 8.12799096e-01
-6.01738632e-01 -1.19346455e-01 7.49647766e-02 4.23871636e-01
5.07520854e-01 1.11520576e+00 3.06154080e-02 -5.00349164e-01
2.86518764e-02 -1.98151078e-02 8.84843394e-02 -6.61575973e-01
-6.05174601e-01 2.58733720e-01 -1.29637688e-01 -8.40085149e-01
-8.18987846e-01 -7.65073657e-01 -1.41533816e+00 8.45669657e-02
-3.30051750e-01 -1.98817234e-02 6.21236742e-01 8.84563565e-01
-1.24453548e-02 4.69105184e-01 6.82556987e-01 -8.77153397e-01
-2.04402760e-01 -5.11458814e-01 -6.58248842e-01 6.76688552e-01
1.42001823e-01 -4.45481211e-01 -4.48635250e-01 3.47245753e-01]
|
[9.42280101776123, 0.34918877482414246]
|
f564bc06-e997-4e71-b798-1dc21f75c004
|
human-evaluation-of-conversations-is-an-open
|
2201.04723
| null |
https://arxiv.org/abs/2201.04723v1
|
https://arxiv.org/pdf/2201.04723v1.pdf
|
Human Evaluation of Conversations is an Open Problem: comparing the sensitivity of various methods for evaluating dialogue agents
|
At the heart of improving conversational AI is the open problem of how to evaluate conversations. Issues with automatic metrics are well known (Liu et al., 2016, arXiv:1603.08023), with human evaluations still considered the gold standard. Unfortunately, how to perform human evaluations is also an open problem: differing data collection methods have varying levels of human agreement and statistical sensitivity, resulting in differing amounts of human annotation hours and labor costs. In this work we compare five different crowdworker-based human evaluation methods and find that different methods are best depending on the types of models compared, with no clear winner across the board. While this highlights the open problems in the area, our analysis leads to advice of when to use which one, and possible future directions.
|
['Jason Weston', 'Y-Lan Boureau', 'Stephen Roller', 'Rebecca Qian', 'Orion Hsu', 'Eric Michael Smith']
|
2022-01-12
| null |
https://aclanthology.org/2022.nlp4convai-1.8
|
https://aclanthology.org/2022.nlp4convai-1.8.pdf
|
nlp4convai-acl-2022-5
|
['dialogue-evaluation']
|
['natural-language-processing']
|
[ 3.24583962e-03 2.67750740e-01 -9.34607983e-02 -4.01205897e-01
-7.33256340e-01 -1.02062821e+00 8.12711298e-01 3.24956387e-01
-6.41379714e-01 9.40129161e-01 6.20598972e-01 -2.90239692e-01
-2.35964909e-01 -3.83026332e-01 -1.44406229e-01 -5.50929308e-01
3.39932382e-01 7.73746252e-01 2.52327681e-01 -4.31290895e-01
5.71078122e-01 -5.54368570e-02 -1.67407072e+00 1.80184543e-01
6.58235669e-01 6.07313097e-01 -3.94971520e-01 9.24291313e-01
-3.10990542e-01 1.02561605e+00 -1.10486841e+00 -8.68839145e-01
1.23183578e-01 -5.44796407e-01 -1.32901251e+00 5.48369531e-03
5.14366329e-01 3.86297181e-02 1.65218085e-01 9.12139118e-01
7.73143351e-01 -2.89458558e-02 4.50518996e-01 -1.45364189e+00
-5.54076374e-01 7.45664597e-01 -1.02596015e-01 2.47010842e-01
5.12951910e-01 2.89538622e-01 1.27391362e+00 -4.86813217e-01
6.94262147e-01 1.24831569e+00 6.85079396e-01 5.40689468e-01
-9.70387876e-01 -5.20807505e-01 -1.66878477e-01 2.02091768e-01
-1.08541846e+00 -5.98821104e-01 2.83695400e-01 -8.83009493e-01
6.32629991e-01 4.94003862e-01 5.39427042e-01 1.19600034e+00
-3.84880543e-01 5.31235099e-01 1.63945901e+00 -3.83996904e-01
2.19056576e-01 5.14456928e-01 4.43033338e-01 3.82608473e-01
5.01300931e-01 -3.10112596e-01 -6.58394098e-01 -3.28199178e-01
3.23612452e-01 -5.67124784e-01 -3.77719641e-01 -4.46524657e-02
-1.28970063e+00 8.84786844e-01 -3.86086367e-02 7.12108970e-01
-1.18995436e-01 -2.99063176e-01 6.79843485e-01 4.10555303e-01
4.98794079e-01 1.09147000e+00 -2.22408965e-01 -8.20552289e-01
-6.65607870e-01 7.09004164e-01 1.34979153e+00 6.35431409e-01
6.38163149e-01 -5.96212387e-01 -3.56035739e-01 9.14138973e-01
-5.60660958e-02 1.10337004e-01 2.10906506e-01 -1.39031649e+00
4.11204517e-01 7.22852707e-01 4.73614454e-01 -1.21320295e+00
-4.27181780e-01 -1.82698548e-01 -3.84055287e-01 2.49154150e-01
1.06725013e+00 -4.42168415e-01 -1.17485650e-01 1.62846315e+00
1.74047470e-01 -3.68546218e-01 -1.86248019e-01 1.09995103e+00
8.58219981e-01 3.16848278e-01 8.27710479e-02 -7.92087149e-03
1.26355970e+00 -1.06746805e+00 -7.53746390e-01 -2.83103108e-01
6.44866824e-01 -1.04145968e+00 1.10399985e+00 3.94512802e-01
-1.13250959e+00 -2.43815124e-01 -8.72408211e-01 -1.66495427e-01
-3.79936159e-01 -3.26595455e-01 6.57851219e-01 8.22821975e-01
-1.10801697e+00 5.82574785e-01 -3.13224226e-01 -7.14371622e-01
-2.87730899e-02 1.24099948e-01 -1.88768968e-01 7.84245357e-02
-1.30661643e+00 1.43482339e+00 4.37762635e-03 -7.09099323e-02
-3.53322387e-01 -4.07910109e-01 -5.33152401e-01 -2.84337580e-01
4.00890321e-01 -3.64383072e-01 1.54750252e+00 -1.01165998e+00
-1.55354524e+00 1.29323161e+00 -7.18314946e-02 -2.66675889e-01
8.61664772e-01 -1.43129174e-02 -8.59134197e-02 -2.76672125e-01
1.56982064e-01 5.04388511e-01 3.03370804e-02 -1.10900366e+00
-7.19177604e-01 -2.12143376e-01 5.43608606e-01 3.16345572e-01
-2.54328012e-01 4.70494002e-01 -1.99384734e-01 -2.68479228e-01
-3.58303070e-01 -1.10843587e+00 -8.75880867e-02 -3.06175679e-01
-5.16841263e-02 -7.54182458e-01 2.54879177e-01 -6.51293874e-01
1.29530740e+00 -1.76511526e+00 2.31193393e-01 -1.80490837e-01
4.48690861e-01 2.20643446e-01 1.37804374e-01 6.16600513e-01
4.24382091e-01 5.06743312e-01 -1.86402768e-01 -4.16696280e-01
2.97141939e-01 -8.57966952e-03 1.14836255e-02 4.52430964e-01
-8.70691165e-02 7.92648971e-01 -1.12612045e+00 -5.80807149e-01
1.29596755e-01 3.61249954e-01 -2.06077725e-01 1.66264445e-01
-5.12578059e-03 5.08434653e-01 -2.27223575e-01 4.02457267e-01
1.72010094e-01 -2.87398517e-01 1.51701406e-01 1.45731911e-01
-4.35059547e-01 6.64456546e-01 -1.15367544e+00 1.46760833e+00
-2.21164618e-03 1.17655408e+00 4.16374147e-01 -7.36045182e-01
8.47476840e-01 4.20641631e-01 2.13072911e-01 -5.46803296e-01
3.08555782e-01 3.54106992e-01 5.43501377e-01 -5.93125165e-01
7.21753418e-01 -7.17105344e-02 -1.94313914e-01 8.22155416e-01
-1.10580616e-01 -3.86872649e-01 5.43461978e-01 3.95784229e-02
1.37118316e+00 -4.24771421e-02 2.08783314e-01 -4.93080288e-01
3.25032324e-01 3.99721384e-01 4.26369786e-01 8.10410023e-01
-6.31118000e-01 5.61748981e-01 8.09730887e-01 -4.41896856e-01
-1.13832200e+00 -3.69217873e-01 -2.81336129e-01 1.29088378e+00
1.28992693e-02 -5.04007578e-01 -1.10233331e+00 -5.21402478e-01
-2.09742010e-01 5.06555319e-01 -7.91248620e-01 4.04437095e-01
-4.77864474e-01 -6.22237086e-01 6.39789701e-01 2.97198594e-02
1.67080402e-01 -1.27373779e+00 -9.42120612e-01 7.17327893e-02
-4.59168106e-01 -1.17042780e+00 -8.72906446e-02 -6.62074760e-02
-3.61938536e-01 -1.12977147e+00 -8.30354691e-01 -2.29970619e-01
2.61734724e-01 3.61428469e-01 1.59199929e+00 3.50950807e-01
-3.29114832e-02 4.72658604e-01 -4.99655038e-01 -7.44421244e-01
-7.03230083e-01 3.57725739e-01 -2.23450094e-01 -3.26600254e-01
6.56437695e-01 -2.80064642e-01 -4.63904500e-01 5.35982370e-01
-5.85186660e-01 -9.28819329e-02 2.80988574e-01 5.08914113e-01
-1.92447186e-01 -2.85345525e-01 3.97735864e-01 -1.08324671e+00
1.14067852e+00 -4.53727275e-01 -2.36848786e-01 2.44814143e-01
-6.59146488e-01 -1.18045710e-01 3.25633407e-01 -1.39771521e-01
-6.95080638e-01 -6.77468717e-01 1.67272151e-01 1.79795891e-01
-3.46080542e-01 2.90807277e-01 1.11896887e-01 1.23723403e-01
9.01984692e-01 -5.59721589e-01 9.99604315e-02 -2.48073235e-01
4.80888672e-02 7.67136574e-01 1.67641252e-01 -6.92377985e-01
4.04412806e-01 2.13839903e-01 -4.90100026e-01 -7.38302410e-01
-1.26609552e+00 -5.24447203e-01 -5.13052285e-01 -5.26381850e-01
1.01592863e+00 -6.60420477e-01 -6.07200086e-01 1.43612564e-01
-1.35501230e+00 -5.80633461e-01 -1.82649642e-01 3.60161692e-01
-3.46825212e-01 3.37907463e-01 -6.72475517e-01 -9.42638814e-01
-2.64755636e-01 -1.29717755e+00 8.90941978e-01 2.56717592e-01
-1.14941132e+00 -9.48480129e-01 4.07321215e-01 1.07428670e+00
6.60839617e-01 2.74507016e-01 4.59261149e-01 -7.28422344e-01
-2.18423069e-01 -3.43693495e-01 -1.11186199e-01 3.22783023e-01
-2.58740410e-02 1.63522914e-01 -1.14905775e+00 1.21243514e-01
-2.17417851e-01 -5.30059040e-01 4.56495881e-01 1.30384415e-01
7.51444757e-01 -8.21831003e-02 -1.37359068e-01 -2.21801832e-01
8.89600754e-01 -1.58480033e-01 6.83195055e-01 6.14294052e-01
5.14946342e-01 1.42552984e+00 5.97139776e-01 2.20146924e-01
6.21039212e-01 6.62680924e-01 -8.96544456e-02 6.63779154e-02
-2.88071446e-02 1.42574400e-01 4.18311357e-01 9.16848361e-01
-3.21693271e-01 -4.50665921e-01 -1.18569791e+00 5.40216327e-01
-2.00684714e+00 -1.22075295e+00 -4.26766992e-01 2.18013573e+00
8.60687613e-01 2.82766372e-01 5.96574724e-01 1.98911414e-01
8.10919464e-01 2.17885271e-01 4.44902591e-02 -5.68853974e-01
-2.16681033e-01 8.37618858e-02 1.98844865e-01 6.56084299e-01
-7.56362259e-01 6.05803072e-01 6.59015274e+00 3.63422781e-01
-9.13985848e-01 3.12831104e-01 5.91233492e-01 -5.08993007e-02
-1.68520063e-01 2.57342488e-01 -5.87175488e-01 5.13568521e-01
1.03418648e+00 -2.12449640e-01 5.06306708e-01 5.33190489e-01
6.63995445e-02 -2.16134861e-01 -1.04823482e+00 8.61618400e-01
3.44065905e-01 -9.55646574e-01 -6.20713055e-01 1.17789574e-01
7.01058447e-01 1.09787107e-01 -5.31963229e-01 4.13233846e-01
6.11586630e-01 -1.20971835e+00 6.86665952e-01 4.23866659e-01
8.58024061e-02 -4.22345787e-01 1.01838183e+00 2.86652595e-01
-5.56276262e-01 1.82561249e-01 -4.40256968e-02 -5.38859963e-01
1.83520377e-01 5.76384246e-01 -7.86101401e-01 1.76231429e-01
9.79149640e-01 2.41851538e-01 -7.57440329e-01 8.23323905e-01
-3.18293750e-01 7.24899113e-01 -7.64355212e-02 -6.15261734e-01
7.35985562e-02 -2.04869643e-01 5.76246500e-01 1.45677567e+00
5.79042621e-02 1.83046777e-02 2.00193629e-01 7.32519150e-01
4.70895097e-02 1.95129231e-01 -6.63380802e-01 -2.11335286e-01
5.59003294e-01 1.54302585e+00 -8.12992334e-01 -3.08683068e-01
-3.05386275e-01 5.60003042e-01 5.53956330e-01 5.46284765e-02
-5.03467023e-01 -9.08278972e-02 5.47744632e-01 1.42609298e-01
-2.85958767e-01 -1.66037261e-01 -5.32819390e-01 -7.88848877e-01
2.04797283e-01 -1.36345625e+00 3.95163476e-01 -5.25524199e-01
-1.38326395e+00 5.07614255e-01 -3.20267235e-03 -9.58613336e-01
-2.13281989e-01 -6.84512198e-01 -5.79710245e-01 6.28530562e-01
-1.02921963e+00 -6.21291339e-01 -7.14527369e-01 -1.37015820e-01
3.99821848e-01 1.09360620e-01 8.33485961e-01 2.73749143e-01
-5.15429795e-01 3.87084901e-01 -3.91116470e-01 1.81189284e-01
1.07916629e+00 -1.33866096e+00 3.06262016e-01 2.76469111e-01
9.82933193e-02 6.10527515e-01 1.14876795e+00 -2.52471834e-01
-1.09048307e+00 -4.11374450e-01 1.16982234e+00 -1.18890512e+00
7.72981405e-01 -3.42978358e-01 -9.50060487e-01 3.85697961e-01
8.36945713e-01 -5.43256342e-01 9.03879166e-01 5.24530351e-01
-2.45047316e-01 4.37897414e-01 -8.96658123e-01 6.66201234e-01
9.62095618e-01 -4.54284966e-01 -4.87972111e-01 3.96865666e-01
5.55715322e-01 -3.67935240e-01 -9.28197443e-01 2.85166353e-01
6.86869323e-01 -1.32916737e+00 3.61655235e-01 -3.17978472e-01
5.02040446e-01 -1.54695153e-01 -8.22168291e-02 -1.30267990e+00
-1.36710703e-01 -6.56010628e-01 5.59721470e-01 1.62437379e+00
6.73051238e-01 -5.76344907e-01 6.13752842e-01 1.06545651e+00
5.99879436e-02 -6.54148281e-01 -4.36484605e-01 -4.15377557e-01
1.59021854e-01 -4.51565325e-01 3.71515155e-01 1.28965104e+00
3.39114964e-01 8.04186165e-01 -1.82699591e-01 -4.00633752e-01
3.56133372e-01 -1.70164436e-01 1.30751896e+00 -1.54381466e+00
-1.27004862e-01 -9.96913493e-01 -3.53850693e-01 -4.50089067e-01
2.25528166e-01 -6.55132949e-01 1.52162209e-01 -1.62169623e+00
3.07493329e-01 -3.66961598e-01 -2.81499401e-02 1.18625857e-01
-2.92275041e-01 2.42547408e-01 3.33850026e-01 1.39532104e-01
-9.39523995e-01 9.17722806e-02 1.05591762e+00 1.09474741e-01
-6.82734102e-02 -2.07776830e-01 -9.61898685e-01 7.61560082e-01
8.52762759e-01 -3.20521832e-01 -5.09583615e-02 -4.95525837e-01
3.91414076e-01 -3.45397145e-01 3.73373091e-01 -9.95774090e-01
3.32289487e-01 -2.67170578e-01 -1.50113508e-01 -5.04848585e-02
3.20529550e-01 -3.82376552e-01 -4.63182516e-02 1.05403759e-01
-6.64180934e-01 3.91280413e-01 6.03623688e-02 1.45734787e-01
-2.17893183e-01 -5.91035068e-01 5.22642195e-01 -4.07053024e-01
-3.96336555e-01 -1.30864322e-01 -1.74308032e-01 5.62604904e-01
8.33734453e-01 -9.80557352e-02 -5.06186903e-01 -6.74260616e-01
-2.16289401e-01 2.95049399e-01 6.31353378e-01 5.84816933e-01
-2.62274474e-01 -9.78438973e-01 -1.08719587e+00 -6.94831491e-01
3.01723093e-01 -1.06913134e-01 -8.15017223e-02 9.50775325e-01
-3.81822139e-01 3.68287504e-01 4.10633199e-02 -5.65257132e-01
-1.16408968e+00 3.96557003e-02 2.32000813e-01 -2.90540963e-01
-8.65956694e-02 7.18522787e-01 -2.31007293e-01 -7.20434189e-01
4.58751589e-01 1.19599067e-01 -3.32841933e-01 4.08981800e-01
5.50971210e-01 7.38123775e-01 1.73215359e-01 -7.50223219e-01
-2.87585348e-01 3.16750228e-01 1.31926566e-01 -5.15630901e-01
1.14692867e+00 -3.78936976e-02 -3.31255466e-01 9.89714205e-01
9.77738798e-01 1.73256323e-01 -6.38419986e-01 -9.30195451e-02
3.90231758e-01 -3.98071855e-01 -3.20089340e-01 -7.61869013e-01
-4.65477824e-01 8.44242454e-01 3.62088352e-01 7.37445116e-01
5.05984604e-01 7.23192692e-02 3.94766688e-01 3.13148558e-01
4.11149710e-01 -1.48230886e+00 -1.81553021e-01 5.67233145e-01
1.04087770e+00 -1.47374308e+00 2.10269928e-01 -2.80335546e-01
-1.00699246e+00 8.85297775e-01 7.42401719e-01 1.26841411e-01
1.56370595e-01 2.00984836e-01 4.73048866e-01 -5.52190542e-01
-7.95126140e-01 -3.19847107e-01 1.54238284e-01 2.14950204e-01
1.03231382e+00 2.57552385e-01 -8.04638505e-01 2.15801388e-01
-8.30416501e-01 -2.23149195e-01 4.31701183e-01 7.85388827e-01
-3.05960894e-01 -1.21059394e+00 -5.41704714e-01 3.54320854e-01
-5.85080564e-01 2.94070780e-01 -1.15875065e+00 8.67226124e-01
1.51654452e-01 1.51024413e+00 -5.74224405e-02 -3.24957222e-01
4.02712315e-01 2.64232457e-01 4.12984431e-01 -6.81397617e-01
-1.02306259e+00 -4.45617288e-01 6.85864866e-01 -2.02601150e-01
-7.71404624e-01 -9.20870364e-01 -7.59981275e-01 -5.74689150e-01
-4.68572676e-01 5.54104984e-01 6.42513514e-01 1.11886573e+00
3.30576003e-01 1.22810148e-01 3.98726463e-01 -7.43840694e-01
-5.14544189e-01 -1.31767488e+00 -1.17451571e-01 7.19877422e-01
1.21390514e-01 -6.84781492e-01 -6.32030606e-01 -2.10773751e-01]
|
[12.493019104003906, 8.181510925292969]
|
ad7faf7e-df65-4c53-9b53-b95a9f1a93a2
|
superpoint-transformer-for-3d-scene-instance
|
2211.15766
| null |
https://arxiv.org/abs/2211.15766v1
|
https://arxiv.org/pdf/2211.15766v1.pdf
|
Superpoint Transformer for 3D Scene Instance Segmentation
|
Most existing methods realize 3D instance segmentation by extending those models used for 3D object detection or 3D semantic segmentation. However, these non-straightforward methods suffer from two drawbacks: 1) Imprecise bounding boxes or unsatisfactory semantic predictions limit the performance of the overall 3D instance segmentation framework. 2) Existing method requires a time-consuming intermediate step of aggregation. To address these issues, this paper proposes a novel end-to-end 3D instance segmentation method based on Superpoint Transformer, named as SPFormer. It groups potential features from point clouds into superpoints, and directly predicts instances through query vectors without relying on the results of object detection or semantic segmentation. The key step in this framework is a novel query decoder with transformers that can capture the instance information through the superpoint cross-attention mechanism and generate the superpoint masks of the instances. Through bipartite matching based on superpoint masks, SPFormer can implement the network training without the intermediate aggregation step, which accelerates the network. Extensive experiments on ScanNetv2 and S3DIS benchmarks verify that our method is concise yet efficient. Notably, SPFormer exceeds compared state-of-the-art methods by 4.3% on ScanNetv2 hidden test set in terms of mAP and keeps fast inference speed (247ms per frame) simultaneously. Code is available at https://github.com/sunjiahao1999/SPFormer.
|
['Xiangmin Xu', 'Junpeng Tan', 'Chunmei Qing', 'Jiahao Sun']
|
2022-11-28
| null | null | null | null |
['3d-instance-segmentation-1']
|
['computer-vision']
|
[ 9.13549587e-02 3.12595695e-01 -3.32325757e-01 -5.24362266e-01
-9.67421830e-01 -4.11407828e-01 2.89026856e-01 -1.61489993e-01
-2.83382744e-01 1.80249155e-01 -4.34275120e-01 -4.03277308e-01
8.32540840e-02 -9.49926138e-01 -1.04876482e+00 -4.77457404e-01
2.28297189e-01 7.86087334e-01 9.41956818e-01 8.23041424e-02
2.56827593e-01 5.47542512e-01 -1.55431366e+00 2.44786069e-01
8.76009822e-01 1.44088364e+00 4.54681963e-01 3.51332724e-01
-6.50453448e-01 8.89278427e-02 -5.13310730e-01 -2.46796966e-01
5.13125539e-01 4.43310700e-02 -7.25840151e-01 -2.44325213e-02
5.29717684e-01 -4.81645346e-01 -1.47185773e-01 1.07952380e+00
5.71503103e-01 -9.84956510e-03 4.49801534e-01 -1.30795789e+00
-3.73951465e-01 5.05884945e-01 -8.15763593e-01 1.54061168e-01
4.84795272e-02 2.27450401e-01 8.57037902e-01 -1.20219374e+00
3.15461874e-01 1.36092043e+00 7.42296576e-01 5.39813638e-01
-9.45920527e-01 -1.04480696e+00 3.89548391e-01 7.69631267e-02
-1.63200748e+00 -2.32724294e-01 6.43647790e-01 -1.82394996e-01
1.10156345e+00 3.18225175e-01 7.44971216e-01 5.41136742e-01
-2.65218437e-01 1.19071376e+00 6.05739772e-01 2.37422973e-01
6.82474822e-02 -6.77972734e-02 2.56250948e-01 7.98130333e-01
1.05099842e-01 6.60355389e-02 -1.90857962e-01 5.25334887e-02
1.10707903e+00 2.62769144e-02 -5.57569712e-02 -3.35910052e-01
-1.23434496e+00 6.58029199e-01 9.30121064e-01 -4.14744951e-02
-3.30644995e-01 3.09716791e-01 2.17661485e-01 -8.27866234e-03
6.00086689e-01 -3.41348536e-02 -6.26824975e-01 2.85014182e-01
-1.09979546e+00 2.33315751e-01 5.49982369e-01 1.53875518e+00
8.99584949e-01 -1.42555654e-01 -1.98021144e-01 5.50052643e-01
5.56907296e-01 8.68430614e-01 -5.66826668e-03 -7.33704031e-01
7.13833153e-01 9.51896727e-01 -1.27685726e-01 -8.87777627e-01
-4.21481967e-01 -6.00988269e-01 -6.16100490e-01 9.27750021e-02
2.30784744e-01 2.10180879e-01 -1.42648339e+00 1.14965165e+00
8.75809133e-01 5.99830568e-01 -1.98971629e-01 1.31765532e+00
1.30506778e+00 8.44418943e-01 1.48145765e-01 3.51764202e-01
1.35243428e+00 -1.04093480e+00 -1.31187648e-01 -3.14972192e-01
4.95766640e-01 -6.22774482e-01 9.46434915e-01 8.16646740e-02
-1.27953386e+00 -7.69665718e-01 -9.53385353e-01 -3.73045504e-01
-3.18417311e-01 8.71353745e-02 6.05690181e-01 3.74191791e-01
-9.56832886e-01 1.81358889e-01 -9.14030731e-01 -4.58590426e-02
8.50107729e-01 7.69303501e-01 -9.03777480e-02 9.88075808e-02
-9.24723983e-01 4.00817305e-01 6.15273416e-01 2.58104950e-01
-8.13124478e-01 -9.96497512e-01 -8.04236352e-01 1.11942239e-01
6.05168760e-01 -7.88277328e-01 1.42097962e+00 -7.02552021e-01
-1.29199827e+00 8.90908420e-01 -2.17572525e-01 -3.57421815e-01
5.17022729e-01 -2.18351513e-01 9.49989483e-02 1.53584704e-01
3.37256163e-01 1.14182842e+00 6.94761634e-01 -1.07911777e+00
-1.11550128e+00 -6.13983810e-01 2.65092645e-02 2.55467862e-01
3.73675942e-01 -3.62222284e-01 -1.23848724e+00 -2.27653295e-01
7.67361164e-01 -6.88824177e-01 -4.08411443e-01 1.44370794e-01
-7.22250402e-01 -5.85689127e-01 9.22669947e-01 -2.76437908e-01
7.82105625e-01 -2.22487164e+00 -1.72711775e-01 3.43377650e-01
3.03620160e-01 2.70682245e-01 -6.46858588e-02 -1.41838804e-01
2.47608155e-01 2.21920252e-01 -4.29491043e-01 -4.36513901e-01
1.76338717e-01 1.02318181e-02 -3.70306283e-01 3.84432018e-01
3.89786601e-01 1.20922089e+00 -7.30799973e-01 -7.78431237e-01
5.46804547e-01 5.70234179e-01 -5.81019998e-01 5.70009425e-02
-4.76297587e-01 2.76861906e-01 -8.65889132e-01 1.04259431e+00
1.10837448e+00 -4.19905603e-01 -4.63606209e-01 -4.57903206e-01
-1.27329886e-01 3.33176762e-01 -1.25102985e+00 1.88919377e+00
-1.52677134e-01 1.81362420e-01 -7.95672759e-02 -1.06232798e+00
8.91853929e-01 5.08792512e-02 3.13985080e-01 -8.84991586e-01
1.64943308e-01 3.90092045e-01 -5.51209748e-01 -3.21302325e-01
1.97946221e-01 7.30780512e-02 -1.80644929e-01 -4.15961444e-02
-5.96205480e-02 -2.89577991e-01 -1.64385512e-01 7.82424137e-02
7.22909987e-01 2.39747405e-01 -8.25704038e-02 4.21682047e-03
5.69121957e-01 3.65581185e-01 6.16214573e-01 8.46449137e-01
-1.03684790e-01 7.07785428e-01 3.11222106e-01 -3.75506967e-01
-9.01008785e-01 -1.39533353e+00 -2.74773389e-01 7.07545519e-01
8.80565166e-01 -5.16191870e-02 -6.39873087e-01 -8.81325841e-01
1.22709759e-01 5.69092751e-01 -3.10671955e-01 1.31041005e-01
-5.75300276e-01 -5.14195085e-01 6.11542344e-01 7.43833542e-01
7.85607040e-01 -8.61709058e-01 -6.25554979e-01 2.09188864e-01
-1.08372293e-01 -1.30165339e+00 -3.85653615e-01 8.66397284e-03
-1.17926335e+00 -1.00049376e+00 -6.98719859e-01 -8.10913086e-01
7.88821101e-01 5.11044323e-01 1.13766086e+00 2.19076797e-01
-1.90650329e-01 4.15567867e-02 -2.35533491e-01 -4.24740285e-01
2.09146157e-01 3.56352657e-01 -5.06573498e-01 -3.26492757e-01
6.64057314e-01 -3.28018397e-01 -8.85884762e-01 4.89460796e-01
-7.10243464e-01 3.86748165e-01 8.85791838e-01 5.46612978e-01
1.29884410e+00 -1.99694291e-01 3.23308229e-01 -7.32914150e-01
-8.50721672e-02 -4.76941228e-01 -1.05292869e+00 -6.74629584e-02
-3.41331512e-01 -1.19294330e-01 1.81968749e-01 -1.96450457e-01
-9.20306385e-01 3.52666408e-01 -3.68071496e-01 -6.03446901e-01
-3.08549076e-01 8.10166895e-02 -2.92946935e-01 -1.98605750e-02
2.01441705e-01 3.72413009e-01 -2.00661972e-01 -5.09081364e-01
3.27537864e-01 5.58423936e-01 4.89864260e-01 -3.19326431e-01
9.63066578e-01 6.86108530e-01 -9.20818448e-02 -6.04869902e-01
-8.80859435e-01 -6.89402521e-01 -5.19676268e-01 -5.50897010e-02
1.09595251e+00 -1.03965664e+00 -9.27994072e-01 3.20635051e-01
-1.34120595e+00 -2.62988865e-01 -5.51098846e-02 4.26740885e-01
-4.51604903e-01 7.07511827e-02 -5.10196686e-01 -6.59717858e-01
-5.49570441e-01 -1.33471489e+00 1.63845050e+00 2.23970369e-01
3.78809094e-01 -4.70300913e-01 -5.66612065e-01 5.06611884e-01
6.41130358e-02 1.19851030e-01 6.13087296e-01 -5.66088617e-01
-1.44554543e+00 -2.40634844e-01 -7.38551319e-01 1.21893220e-01
-4.23265010e-01 -2.22458988e-01 -8.87468517e-01 1.08672895e-01
-1.56552091e-01 7.56224245e-02 8.51116776e-01 5.12963951e-01
1.40243125e+00 4.64201579e-03 -6.62053168e-01 9.57304537e-01
1.49700260e+00 2.36404032e-01 4.61381257e-01 1.25371322e-01
9.15659189e-01 3.02091926e-01 8.55988860e-01 2.37039655e-01
6.19321942e-01 5.32612026e-01 8.40069711e-01 -3.10303390e-01
-1.21754572e-01 -3.64394963e-01 -8.69538635e-02 5.13477206e-01
1.87142685e-01 -3.06958020e-01 -1.00215435e+00 6.40083969e-01
-1.81354260e+00 -5.98614693e-01 -5.11890292e-01 1.91703808e+00
4.86780316e-01 4.62979138e-01 -9.49616134e-02 -1.92851163e-02
7.64588058e-01 8.81678984e-02 -7.76159704e-01 7.25902915e-02
1.71743393e-01 2.24633291e-01 8.59896958e-01 4.09864873e-01
-1.12421036e+00 1.29181862e+00 4.56003618e+00 1.03226089e+00
-9.54147577e-01 2.29189068e-01 6.75628960e-01 -2.14267015e-01
-1.58752233e-01 -1.62174091e-01 -1.20420134e+00 4.21755493e-01
4.36087400e-01 2.87244558e-01 -8.91759843e-02 1.05745065e+00
1.79592222e-02 -1.34461652e-02 -9.94466424e-01 1.07730234e+00
-1.91881716e-01 -1.43879867e+00 1.20643221e-01 -2.08249450e-01
4.03380513e-01 5.01317084e-01 7.13127712e-03 3.63762259e-01
-6.67270347e-02 -8.44237208e-01 9.30225670e-01 3.30983669e-01
7.14667082e-01 -8.94077599e-01 7.51920104e-01 5.28511286e-01
-1.41946232e+00 1.65111646e-01 -5.64283788e-01 3.41483951e-01
3.04547191e-01 6.81407332e-01 -1.01483703e+00 4.87392545e-01
9.25056398e-01 5.34500360e-01 -2.98189789e-01 1.31224871e+00
-2.73021787e-01 5.15357256e-01 -7.74574041e-01 -4.83880863e-02
6.93194151e-01 4.08696271e-02 6.32543266e-01 1.08694720e+00
4.08418000e-01 3.35643768e-01 2.34969258e-01 1.29655445e+00
-1.11859581e-02 -3.61925326e-02 -2.73320973e-01 3.45312387e-01
6.85388744e-01 1.01148200e+00 -1.13619709e+00 -4.79120463e-01
-4.18517530e-01 8.75683308e-01 -2.16539819e-02 2.64044404e-01
-1.28328490e+00 -4.72208023e-01 5.38191676e-01 1.80010825e-01
6.37644827e-01 -1.32600963e-01 -6.36306822e-01 -8.59881878e-01
2.29735702e-01 -3.71572942e-01 1.88986257e-01 -7.61385083e-01
-1.05551350e+00 6.26101673e-01 2.83698142e-02 -1.16394877e+00
3.25908542e-01 -4.69656050e-01 -3.70623827e-01 8.28141153e-01
-1.64897978e+00 -1.30703712e+00 -6.04146838e-01 5.98153412e-01
8.26544583e-01 2.95400649e-01 2.11925700e-01 5.50728381e-01
-4.80270118e-01 4.97369379e-01 -4.66850132e-01 1.01825029e-01
1.30448699e-01 -1.12194777e+00 9.57539499e-01 5.63146830e-01
2.20369939e-02 1.73450887e-01 2.45929375e-01 -6.92674279e-01
-1.28142214e+00 -1.24319601e+00 7.91967034e-01 -4.90732759e-01
2.18661234e-01 -5.15256345e-01 -8.62205684e-01 5.08447945e-01
-4.16989118e-01 1.12666212e-01 2.80810833e-01 -2.90917814e-01
-1.59676358e-01 -1.76919639e-01 -1.08146262e+00 4.36841488e-01
1.49086475e+00 -2.34624356e-01 -5.61073542e-01 3.86711240e-01
1.32247818e+00 -8.82541180e-01 -6.02124393e-01 7.36295283e-01
3.23432803e-01 -9.28548038e-01 1.29989100e+00 -1.44831553e-01
2.31149420e-01 -6.39193833e-01 -1.22377619e-01 -4.84491289e-01
-2.33850077e-01 -1.75380543e-01 -8.89533758e-02 9.36185956e-01
6.10535800e-01 -6.68313682e-01 1.16066349e+00 5.35524786e-01
-6.05143368e-01 -9.69010353e-01 -1.10483599e+00 -6.37497783e-01
-2.72049636e-01 -9.33676243e-01 1.10011637e+00 4.93608028e-01
-7.35089839e-01 1.80724770e-01 2.35250562e-01 7.42842138e-01
7.53798127e-01 3.59423101e-01 9.03272212e-01 -1.05396640e+00
2.13380288e-02 -4.91014034e-01 -5.82475841e-01 -1.80231786e+00
-5.25583588e-02 -1.00895596e+00 1.47368312e-01 -1.74936843e+00
-1.57994896e-01 -8.43151033e-01 -2.12221742e-01 5.34487307e-01
-2.09859386e-02 4.44018871e-01 2.43607149e-01 2.14950457e-01
-8.02704096e-01 4.97335076e-01 1.48759043e+00 -2.32069135e-01
-3.39403629e-01 2.35615388e-01 -4.24061269e-01 7.02037394e-01
7.48804331e-01 -5.58522403e-01 -5.46271741e-01 -8.41385365e-01
2.33316794e-02 1.54810339e-01 7.82787561e-01 -1.08145547e+00
4.45318609e-01 3.09741404e-03 4.80569661e-01 -1.50028896e+00
4.64665264e-01 -1.03594637e+00 2.19331738e-02 3.57545763e-01
4.08163033e-02 -1.53776169e-01 2.98796356e-01 5.84503412e-01
-1.41939491e-01 -5.34237437e-02 5.30615568e-01 -2.04547584e-01
-1.07345891e+00 8.08345497e-01 2.54872113e-01 7.04662576e-02
1.16972709e+00 -6.46219492e-01 -1.05160743e-01 2.85035223e-01
-6.31798029e-01 6.02366805e-01 1.74356237e-01 4.18352276e-01
8.56558204e-01 -1.16086316e+00 -5.02832532e-01 4.18480814e-01
-1.96258649e-02 9.52831745e-01 4.24511105e-01 9.18834209e-01
-6.78309977e-01 5.98440230e-01 2.60100335e-01 -1.21773493e+00
-1.03347337e+00 3.35193068e-01 4.22132850e-01 2.09024996e-01
-7.93012381e-01 1.27305758e+00 6.22792721e-01 -5.44652343e-01
3.39697778e-01 -5.76519012e-01 1.62210137e-01 -1.42459422e-01
2.23971501e-01 2.64563590e-01 8.42546020e-03 -5.64133108e-01
-5.68330884e-01 7.79091775e-01 -1.93424318e-02 1.64738476e-01
1.15679789e+00 -3.81490700e-02 1.16057217e-01 1.28485382e-01
1.21503174e+00 -5.11495471e-01 -1.40651333e+00 -1.93031624e-01
-1.22983761e-01 -5.29787362e-01 1.68005988e-01 -6.54030740e-01
-1.38086164e+00 9.60436821e-01 6.43029749e-01 4.64788824e-02
1.07992113e+00 4.36974823e-01 1.25490987e+00 2.75861859e-01
4.28491205e-01 -7.80383170e-01 -3.25053692e-01 4.44809645e-01
5.79422474e-01 -1.30328095e+00 -2.22358987e-01 -7.68846333e-01
-2.77702481e-01 7.81910837e-01 9.53218877e-01 -2.56547481e-01
6.96339309e-01 7.04076588e-02 3.67661454e-02 -4.74580318e-01
-4.24128920e-01 -2.52876550e-01 5.01806676e-01 4.23376054e-01
-7.19666556e-02 1.56748376e-03 -1.07593536e-01 7.10418463e-01
-5.94976693e-02 -1.57988548e-01 -1.59599885e-01 6.39070153e-01
-5.94281554e-01 -6.23447716e-01 -2.81622171e-01 4.55316693e-01
-2.66280502e-01 -1.76265642e-01 -6.52865320e-02 1.06058645e+00
4.03456807e-01 6.06279969e-01 3.89270425e-01 -3.35426718e-01
5.99414110e-01 -2.13250175e-01 1.65416569e-01 -6.19129360e-01
-4.92701024e-01 2.13881642e-01 -2.69953132e-01 -9.76564050e-01
-3.44800234e-01 -3.75819713e-01 -1.77624047e+00 -1.02214523e-01
-6.35782123e-01 2.24316325e-02 7.88875699e-01 7.29326189e-01
6.47565484e-01 4.42123741e-01 3.51883829e-01 -1.01969755e+00
-3.29245180e-01 -5.50413489e-01 -1.96953639e-01 1.04237527e-01
2.37209871e-01 -6.37650907e-01 -2.01568082e-01 -2.31026754e-01]
|
[8.02025318145752, -3.1766180992126465]
|
1c346adb-f137-40e6-b035-827403fa768b
|
security-aware-virtual-network-embedding
|
2202.02452
| null |
https://arxiv.org/abs/2202.02452v1
|
https://arxiv.org/pdf/2202.02452v1.pdf
|
Security-Aware Virtual Network Embedding Algorithm based on Reinforcement Learning
|
Virtual network embedding (VNE) algorithm is always the key problem in network virtualization (NV) technology. At present, the research in this field still has the following problems. The traditional way to solve VNE problem is to use heuristic algorithm. However, this method relies on manual embedding rules, which does not accord with the actual situation of VNE. In addition, as the use of intelligent learning algorithm to solve the problem of VNE has become a trend, this method is gradually outdated. At the same time, there are some security problems in VNE. However, there is no intelligent algorithm to solve the security problem of VNE. For this reason, this paper proposes a security-aware VNE algorithm based on reinforcement learning (RL). In the training phase, we use a policy network as a learning agent and take the extracted attributes of the substrate nodes to form a feature matrix as input. The learning agent is trained in this environment to get the mapping probability of each substrate node. In the test phase, we map nodes according to the mapping probability and use the breadth-first strategy (BFS) to map links. For the security problem, we add security requirements level constraint for each virtual node and security level constraint for each substrate node. Virtual nodes can only be embedded on substrate nodes that are not lower than the level of security requirements. Experimental results show that the proposed algorithm is superior to other typical algorithms in terms of long-term average return, long-term revenue consumption ratio and virtual network request (VNR) acceptance rate.
|
['Abderrahim Benslimane', 'Chunxiao Jiang', 'Chao Wang', 'Peiying Zhang']
|
2022-02-03
| null | null | null | null |
['network-embedding']
|
['methodology']
|
[-2.07639366e-01 -1.76052287e-01 -4.55019116e-01 1.32133111e-01
4.92339641e-01 -4.44768310e-01 1.28732160e-01 -1.00401700e-01
-5.53361118e-01 8.33943129e-01 -5.50516903e-01 -6.67328179e-01
-3.97669464e-01 -1.31634736e+00 -2.51496136e-01 -4.58837986e-01
-2.63742227e-02 4.93017763e-01 5.20782053e-01 -3.29144239e-01
3.47072572e-01 6.96379840e-01 -1.05312228e+00 -2.73357779e-01
7.36044526e-01 8.60288799e-01 4.11946863e-01 1.62532404e-01
-6.91367388e-01 3.41754466e-01 -7.39818394e-01 -1.31024018e-01
3.83739740e-01 -3.59791547e-01 -6.84338152e-01 -5.84671609e-02
-7.27887034e-01 -3.50241393e-01 -2.99776703e-01 1.05036640e+00
2.79174209e-01 -1.62634537e-01 4.43748802e-01 -1.79043818e+00
-1.68100163e-01 5.14704525e-01 -6.34810746e-01 2.72849947e-01
6.22666739e-02 -7.90799931e-02 9.55890000e-01 -4.23718840e-01
7.01304436e-01 8.95314813e-01 1.07468061e-01 7.17958510e-01
-9.87068772e-01 -8.38472605e-01 4.71965432e-01 4.31375951e-01
-1.36459374e+00 1.20562889e-01 9.55771208e-01 -9.17469040e-02
6.53393984e-01 2.61388153e-01 9.16337252e-01 6.23473942e-01
1.63808048e-01 4.37678367e-01 9.84570324e-01 -3.69283050e-01
4.44264501e-01 7.42557585e-01 -1.81414723e-01 5.62966168e-01
4.38529462e-01 2.13331088e-01 1.26909703e-01 -4.80777621e-02
7.67449439e-01 1.59981847e-01 -1.50236532e-01 -5.89890182e-01
-7.08016992e-01 9.06134605e-01 4.38155740e-01 3.35141659e-01
-4.30434555e-01 -1.44292593e-01 5.94047666e-01 5.12076318e-01
-2.08896667e-01 1.90911964e-01 -6.11309886e-01 5.50547466e-02
-6.57242715e-01 -2.46716425e-01 9.64386463e-01 5.67896783e-01
7.44978845e-01 2.91496158e-01 4.68124688e-01 4.35038000e-01
4.37986344e-01 3.49166423e-01 1.04057215e-01 -4.59740967e-01
2.69289076e-01 7.22029328e-01 -6.07355982e-02 -1.22837603e+00
-1.72401726e-01 -4.87560213e-01 -8.67582858e-01 6.03377104e-01
-1.68189600e-01 -3.54786098e-01 -5.38152277e-01 1.66763747e+00
4.96766299e-01 5.24794221e-01 2.34982833e-01 7.95087159e-01
4.26966310e-01 1.03216183e+00 -1.15256876e-01 -7.98295379e-01
1.04934251e+00 -7.84892261e-01 -6.89558804e-01 2.54592121e-01
3.99151772e-01 -6.94581747e-01 8.19924116e-01 4.30865616e-01
-4.53429341e-01 -2.60841191e-01 -1.31082761e+00 1.09813344e+00
-6.44221365e-01 -1.28393456e-01 7.50601470e-01 9.37940598e-01
-9.32777226e-01 2.63136268e-01 -4.03811932e-01 -4.19078618e-01
2.68711597e-02 6.37192309e-01 -1.96954921e-01 1.42824784e-01
-1.45415950e+00 6.57136798e-01 8.25075567e-01 2.53054593e-03
-8.64761591e-01 -2.91690081e-01 -4.39597875e-01 1.57252908e-01
8.08933437e-01 -2.60775685e-01 5.02580702e-01 -9.84770179e-01
-1.55570781e+00 1.22745670e-01 4.60645825e-01 -2.87583798e-01
4.18097556e-01 5.67178249e-01 -8.74596119e-01 -4.66982163e-02
-4.06240135e-01 2.46994451e-01 5.92182457e-01 -1.37330294e+00
-7.92185009e-01 1.57873988e-01 4.26139116e-01 2.17033550e-01
-7.87442565e-01 9.16427467e-03 -3.81610990e-01 -7.93928206e-02
-6.47024140e-02 -8.25213909e-01 -3.18089992e-01 -2.34303936e-01
-2.32632786e-01 -2.03095809e-01 1.21462810e+00 -4.28575546e-01
1.55097270e+00 -2.01893711e+00 5.29738888e-02 9.66841817e-01
2.07239956e-01 6.16630137e-01 -6.17228784e-02 4.01179075e-01
1.34933800e-01 4.24736679e-01 3.05716664e-01 4.49301869e-01
-1.37706891e-01 2.69990683e-01 -1.01694636e-01 -5.14999330e-02
-2.06425115e-01 2.15309888e-01 -8.15604627e-01 -7.56262183e-01
2.94119179e-01 2.47108534e-01 -5.21458447e-01 2.72147089e-01
-1.94472149e-01 1.79994330e-01 -6.95924520e-01 4.93625790e-01
9.01926994e-01 -2.18836367e-01 7.57176399e-01 -1.27723217e-01
-9.34935585e-02 -2.87748367e-01 -1.58717453e+00 9.07351136e-01
-6.20920241e-01 6.84137046e-02 2.77342916e-01 -1.02524912e+00
1.08950782e+00 3.45484167e-01 7.16841996e-01 -6.08551979e-01
1.81793109e-01 3.80180538e-01 1.64955765e-01 -4.94250596e-01
2.26182751e-02 -1.28506040e-02 2.26861283e-01 6.19122744e-01
-2.24175155e-01 5.31395853e-01 8.14186633e-02 2.56162852e-01
1.09690940e+00 -1.38780609e-01 1.30763754e-01 7.27673918e-02
9.22553837e-01 -1.72834978e-01 9.17061388e-01 2.35215589e-01
-1.80446520e-01 -4.99016702e-01 7.50640273e-01 -5.02403915e-01
-1.02387357e+00 -9.41284001e-01 1.90597817e-01 4.56390709e-01
4.73715842e-01 -4.23075944e-01 -5.58124423e-01 -1.03516734e+00
-1.16894446e-01 3.98830980e-01 -2.60751158e-01 -2.85434455e-01
-3.41782212e-01 -5.08598804e-01 2.17755251e-02 2.84651853e-02
6.03190958e-01 -1.41254985e+00 -5.11724412e-01 3.78720999e-01
1.43718138e-01 -9.50860858e-01 -9.55842733e-02 -2.90028155e-02
-7.66406953e-01 -1.00769067e+00 -9.78658274e-02 -8.38986337e-01
8.43682647e-01 3.08616430e-01 4.98842865e-01 6.04188979e-01
-1.58099398e-01 -1.03121586e-01 -4.56575245e-01 -1.04323417e-01
-3.86097819e-01 3.63336533e-01 2.41523236e-01 2.67773103e-02
1.75670221e-01 -6.15820646e-01 -4.03100878e-01 4.83607948e-01
-8.74430358e-01 -9.44224671e-02 8.38950992e-01 8.60525191e-01
4.10771728e-01 8.85267675e-01 7.84573436e-01 -7.54503191e-01
7.00504184e-01 -5.43982685e-01 -1.00469768e+00 5.14785290e-01
-1.07763171e+00 -6.32260591e-02 1.06161833e+00 -5.60237944e-01
-5.59484661e-01 -2.28557333e-01 6.72825500e-02 -5.49880326e-01
2.83173919e-01 5.12707114e-01 -6.09213352e-01 -2.69501954e-01
4.03684797e-03 2.31100723e-01 2.13150918e-01 -1.77691728e-01
-1.93957731e-01 7.58612752e-01 -3.52017999e-01 -3.28457355e-01
1.18695712e+00 6.63196966e-02 3.31819236e-01 -4.20497775e-01
-3.13800685e-02 9.48104709e-02 -4.36986890e-03 -4.00806278e-01
5.52597106e-01 -3.55644584e-01 -1.24371302e+00 -1.26397029e-01
-7.94028342e-01 -4.34008539e-02 2.35729545e-01 5.39796114e-01
-1.03913859e-01 3.11226785e-01 -3.25470299e-01 -1.03506458e+00
-3.74241531e-01 -1.21192837e+00 -1.06071509e-01 5.86598933e-01
3.86520863e-01 -9.39289391e-01 -2.10052684e-01 -8.62963498e-02
5.36460340e-01 2.26443291e-01 1.20763099e+00 -5.59066176e-01
-8.67951393e-01 -1.21377997e-01 -2.88626939e-01 4.15014267e-01
1.21372335e-01 3.70834678e-01 -1.11298092e-01 -5.27669072e-01
-1.78930134e-01 -9.11542401e-02 2.16771126e-01 -1.45671871e-02
1.07004404e+00 -3.84515077e-01 -5.33065200e-01 6.10557139e-01
1.99225628e+00 6.91879988e-01 6.31558061e-01 7.20094383e-01
4.69429195e-01 5.49938440e-01 8.78749132e-01 5.43839037e-01
7.02454075e-02 5.96709609e-01 8.65139306e-01 -6.21012971e-02
3.88397068e-01 -4.12616402e-01 2.98153400e-01 8.31175804e-01
8.20638984e-02 -3.79127800e-01 -6.79770112e-01 1.07946925e-01
-1.67879653e+00 -9.67727780e-01 4.20247793e-01 2.29835200e+00
1.69444248e-01 6.97123230e-01 1.26036778e-01 4.96651590e-01
8.74000967e-01 4.84753065e-02 -4.81525809e-01 -7.53763855e-01
1.64451152e-01 -2.00900689e-01 5.37700474e-01 3.54944736e-01
-5.57636976e-01 9.58328485e-01 5.09183645e+00 1.00338387e+00
-1.44923961e+00 2.30794828e-02 3.74751061e-01 1.83263734e-01
-4.17264521e-01 4.09435123e-01 -6.28769815e-01 7.00418293e-01
7.07373857e-01 -1.81462407e-01 9.95002687e-01 7.73022354e-01
1.53509930e-01 1.38067380e-01 -4.82768565e-01 9.48698759e-01
-2.14625910e-01 -1.19710827e+00 2.83222273e-03 4.13564235e-01
3.83240491e-01 -3.90301228e-01 -6.74280152e-03 3.93402576e-01
7.71214291e-02 -7.71480799e-01 1.13756172e-01 1.73488513e-01
7.55673826e-01 -1.23075128e+00 1.00065780e+00 4.84247297e-01
-1.29266024e+00 -4.18871671e-01 -5.63530385e-01 2.10193083e-01
8.40816051e-02 2.99457073e-01 -8.70268881e-01 8.65269065e-01
4.75336522e-01 1.06340028e-01 -2.16952130e-01 9.80088472e-01
-6.06714115e-02 2.75835156e-01 -4.13050383e-01 -4.14538085e-01
2.25937769e-01 -5.93634546e-01 4.35210794e-01 4.62710857e-01
4.43858445e-01 7.55759552e-02 3.93566638e-01 4.63142812e-01
-8.91738757e-02 5.98335207e-01 -6.86942279e-01 -1.32388622e-01
9.32895422e-01 1.50822461e+00 -1.01064312e+00 -2.03788728e-01
-3.48534942e-01 7.62912929e-01 2.16586888e-01 2.66444266e-01
-1.02726436e+00 -6.71172380e-01 4.26495373e-01 2.03478768e-01
2.62057930e-01 9.18839127e-03 9.49658379e-02 -6.95602000e-01
1.21860504e-02 -9.14106667e-01 3.39124799e-01 -3.94766808e-01
-9.05813456e-01 7.78613985e-01 -2.00917035e-01 -1.33679330e+00
1.76595509e-01 -5.00453472e-01 -8.35681081e-01 6.67305529e-01
-1.51446092e+00 -7.67494023e-01 -1.53393939e-01 6.43536091e-01
1.24462478e-01 -7.12041378e-01 9.56120789e-01 4.98163223e-01
-9.32327390e-01 6.35987222e-01 1.50339842e-01 7.67158270e-02
3.07092965e-01 -6.52949452e-01 -3.08206797e-01 7.89586842e-01
-1.02359883e-01 5.82467079e-01 6.38838172e-01 -7.65061080e-01
-1.32683897e+00 -6.15068018e-01 5.47052145e-01 2.99757063e-01
6.01549745e-01 -3.61920714e-01 -5.58762908e-01 4.27250355e-01
1.85362533e-01 1.61374867e-01 6.26739025e-01 -1.40196178e-03
-4.68168184e-02 -6.31395459e-01 -1.45770288e+00 7.39333332e-01
7.92216659e-01 -1.46276981e-01 9.31619704e-02 1.52443573e-01
8.26362133e-01 9.48082730e-02 -6.71454728e-01 4.71206129e-01
3.83092642e-01 -7.02115536e-01 7.59555101e-01 -6.80812359e-01
-2.04375014e-01 -6.65136516e-01 -7.14315027e-02 -1.12005091e+00
-3.46736878e-01 -4.32039291e-01 -1.87817171e-01 1.37011659e+00
4.32878673e-01 -9.54410434e-01 1.16154373e+00 1.22572958e-01
4.05612767e-01 -1.07267344e+00 -8.45788002e-01 -9.28481638e-01
-4.60948646e-01 1.81554928e-01 1.08436990e+00 9.53263164e-01
-4.62656580e-02 3.40641975e-01 -4.33800161e-01 2.51206130e-01
5.58784008e-01 1.64547384e-01 6.04446054e-01 -1.42895901e+00
-4.21087116e-01 -3.75054002e-01 -4.92698252e-01 -4.03140336e-01
3.23059350e-01 -7.45126605e-01 -7.02473760e-01 -1.65400743e+00
4.79601743e-03 -1.10862446e+00 -8.69569123e-01 2.86209524e-01
2.26074561e-01 -3.54090035e-01 3.05163980e-01 1.13704123e-01
-5.19298077e-01 2.92138249e-01 1.08239353e+00 -2.21946137e-03
-4.51304793e-01 3.56912203e-02 -4.21167850e-01 3.29529196e-01
1.29366231e+00 -3.62940311e-01 -9.01113451e-01 1.58072654e-02
3.25678110e-01 4.53898877e-01 -1.40969709e-01 -9.38142121e-01
1.89599097e-01 -6.56413913e-01 2.82561034e-01 -5.47942221e-01
7.50467330e-02 -1.45706534e+00 4.62212652e-01 9.96211350e-01
2.08998352e-01 5.68394959e-01 -8.89124423e-02 5.74975491e-01
7.04340264e-02 -2.61111170e-01 5.93206227e-01 5.72560988e-02
-8.19831848e-01 5.38293123e-01 -3.54832172e-01 -2.63763100e-01
1.47055781e+00 -3.69055659e-01 -1.27405882e-01 -1.23413496e-01
-4.59866196e-01 5.39417803e-01 5.38230479e-01 3.51440400e-01
8.75757635e-01 -1.22963083e+00 -3.71744692e-01 2.66851276e-01
-6.67764917e-02 -5.64659894e-01 3.92863572e-01 5.98355651e-01
-7.19885051e-01 1.92730188e-01 -7.27417290e-01 -5.09227812e-02
-1.30157495e+00 1.11046946e+00 2.26982072e-01 -5.82613528e-01
-2.57643402e-01 4.68932331e-01 -3.92162740e-01 -3.65480274e-01
2.67353505e-01 6.56981468e-01 -6.90804183e-01 -1.75318688e-01
2.88544923e-01 2.04969332e-01 -3.57119709e-01 -2.87789375e-01
-5.09016812e-01 3.80000800e-01 -2.57692337e-01 -2.29294643e-01
1.15428066e+00 -2.60282420e-02 -3.63372535e-01 -1.10300370e-01
9.60702002e-01 1.59832031e-01 -5.62061310e-01 -7.68363029e-02
-1.59858316e-01 -6.82823181e-01 8.14151391e-02 -7.52229393e-01
-1.48964548e+00 5.62199473e-01 8.40774775e-01 3.88246387e-01
1.14646184e+00 -6.17816269e-01 7.80436575e-01 1.36063054e-01
7.75497079e-01 -1.34598374e+00 5.96447848e-02 1.66931883e-01
8.90513062e-02 -9.49359596e-01 -2.90631633e-02 -6.58779085e-01
-4.62753445e-01 1.06235957e+00 1.28113186e+00 -2.02207323e-02
8.92463803e-01 1.74821600e-01 -7.93408602e-02 7.59204775e-02
-7.32185066e-01 2.72085108e-02 -6.77280426e-01 6.07125819e-01
-2.22508162e-01 1.71557933e-01 -7.67209589e-01 2.59485066e-01
1.10680863e-01 -1.65587544e-01 7.68687129e-01 6.80702984e-01
-5.37328064e-01 -1.79091394e+00 -2.32469201e-01 4.23892945e-01
-3.01908106e-01 1.82400092e-01 1.58395921e-03 8.10046554e-01
3.57205540e-01 8.88352573e-01 -1.68608978e-01 -9.30596709e-01
9.37018096e-02 -2.24765494e-01 1.93521827e-01 -3.56580555e-01
-3.21758777e-01 -1.15734041e-01 2.53540501e-02 -1.78635255e-01
1.87256858e-01 -2.52577275e-01 -1.54707825e+00 -7.74801075e-01
-5.33551216e-01 7.68746436e-01 9.16480184e-01 4.32417601e-01
2.33940706e-01 6.98934615e-01 1.34545839e+00 9.08140186e-03
-4.60703731e-01 -3.46680403e-01 -6.98217690e-01 2.81484928e-02
-2.30888128e-01 -8.13543200e-01 -3.94856483e-01 -8.63251209e-01]
|
[5.891761302947998, 1.7314261198043823]
|
716c24b1-7c85-4d2e-816f-53e596c64e61
|
moderately-balanced-representation-learning
|
2209.01956
| null |
https://arxiv.org/abs/2209.01956v1
|
https://arxiv.org/pdf/2209.01956v1.pdf
|
Moderately-Balanced Representation Learning for Treatment Effects with Orthogonality Information
|
Estimating the average treatment effect (ATE) from observational data is challenging due to selection bias. Existing works mainly tackle this challenge in two ways. Some researchers propose constructing a score function that satisfies the orthogonal condition, which guarantees that the established ATE estimator is "orthogonal" to be more robust. The others explore representation learning models to achieve a balanced representation between the treated and the controlled groups. However, existing studies fail to 1) discriminate treated units from controlled ones in the representation space to avoid the over-balanced issue; 2) fully utilize the "orthogonality information". In this paper, we propose a moderately-balanced representation learning (MBRL) framework based on recent covariates balanced representation learning methods and orthogonal machine learning theory. This framework protects the representation from being over-balanced via multi-task learning. Simultaneously, MBRL incorporates the noise orthogonality information in the training and validation stages to achieve a better ATE estimation. The comprehensive experiments on benchmark and simulated datasets show the superiority and robustness of our method on treatment effect estimations compared with existing state-of-the-art methods.
|
['Zhixiang Huang', 'Dongdong Wang', 'Qi Wu', 'Shumin Ma', 'Cheuk Hang Leung', 'Yiyan Huang']
|
2022-09-05
| null | null | null | null |
['selection-bias']
|
['natural-language-processing']
|
[ 2.87040859e-01 -2.73018211e-01 -9.19133365e-01 -3.43472540e-01
-1.05026150e+00 -6.68700859e-02 2.89444625e-01 5.95787428e-02
4.30793501e-02 6.43680990e-01 7.34480619e-01 -2.48527363e-01
-5.10383308e-01 -6.10444248e-01 -5.25434673e-01 -1.03440440e+00
1.72685534e-01 -3.96340638e-02 -4.51380461e-01 1.67863101e-01
2.90631175e-01 -4.77624014e-02 -1.23509896e+00 1.85078427e-01
1.18167174e+00 9.50950503e-01 -6.74551427e-02 -1.27214700e-01
1.50898695e-01 7.58012474e-01 -1.38505161e-01 -4.83310334e-02
2.52204329e-01 -4.15993273e-01 -2.75353402e-01 -7.20409080e-02
1.76088318e-01 -7.12092593e-02 -4.38816577e-01 1.12031603e+00
8.36572468e-01 -3.85139138e-02 1.15565813e+00 -1.21592331e+00
-8.25306833e-01 6.66896999e-01 -1.21501255e+00 -6.85294196e-02
3.63850929e-02 -1.52430668e-01 1.03860974e+00 -7.38898635e-01
1.60948694e-01 1.39747584e+00 7.50278056e-01 1.42852172e-01
-1.43965220e+00 -1.30395949e+00 2.63073474e-01 5.92090189e-02
-1.39591110e+00 -4.29847389e-01 8.58336985e-01 -6.65127516e-01
1.25016764e-01 1.08166225e-01 1.45512134e-01 1.33571577e+00
3.26353073e-01 8.57425869e-01 1.25113392e+00 -3.00893694e-01
1.02410577e-01 1.45419445e-02 3.70406926e-01 5.44298530e-01
5.39482057e-01 4.40682203e-01 -1.67468250e-01 -3.71577889e-01
8.43661249e-01 2.53627926e-01 -5.00475705e-01 -7.12399304e-01
-1.24434125e+00 1.16334331e+00 3.59191030e-01 1.22169249e-01
-6.36701524e-01 -1.27139390e-01 8.44054878e-01 1.44371554e-01
5.85796595e-01 6.91516474e-02 -3.28113228e-01 3.37175161e-01
-6.21691167e-01 1.50021955e-01 2.26979077e-01 8.17230284e-01
4.56637889e-01 2.19535246e-01 -6.90382719e-01 1.04332995e+00
3.77345502e-01 5.02408564e-01 8.01569223e-01 -6.72451258e-01
7.71669030e-01 6.14545226e-01 -2.37264484e-02 -1.35171330e+00
-4.54988182e-01 -9.07434642e-01 -1.51301384e+00 -1.86166480e-01
2.05818921e-01 -2.85531044e-01 -5.75582743e-01 2.06875181e+00
3.81167024e-01 4.27972019e-01 -5.60477423e-03 7.58463919e-01
8.90466571e-01 5.63304007e-01 2.74927020e-01 -5.36809981e-01
1.45627606e+00 -6.42227113e-01 -1.12307596e+00 -1.62781954e-01
8.79112184e-01 -6.22084081e-01 8.14507663e-01 4.17756379e-01
-7.77011454e-01 -7.04593658e-01 -1.14242494e+00 2.66716123e-01
6.90663904e-02 6.90134943e-01 9.22377288e-01 9.32075858e-01
-1.20905392e-01 3.58988822e-01 -3.98511380e-01 2.02678174e-01
7.28505552e-01 2.82213748e-01 -4.76727456e-01 -2.27145523e-01
-1.19968033e+00 5.53170383e-01 1.86350033e-01 1.38674811e-01
-8.68937612e-01 -1.10300028e+00 -1.00928390e+00 9.77789760e-02
4.07819957e-01 -8.66186619e-01 9.35296237e-01 -7.58193612e-01
-1.20395732e+00 5.28061390e-01 -1.48539841e-01 -2.25389767e-02
3.60230058e-01 -7.17746243e-02 -3.19228351e-01 -2.81559616e-01
3.01854908e-01 -4.03501801e-02 7.68004894e-01 -1.13422561e+00
-5.63561797e-01 -7.99160063e-01 -4.05670673e-01 1.47768721e-01
-5.10100543e-01 -1.06945105e-01 -3.15604880e-02 -1.16902959e+00
4.00901437e-01 -7.47577488e-01 -3.37765366e-01 -2.78853357e-01
-3.44089687e-01 -2.88372874e-01 5.19672573e-01 -6.04428172e-01
1.63197398e+00 -2.25004268e+00 2.26613030e-01 7.45833516e-02
3.36851776e-01 1.04789741e-01 -2.01954871e-01 2.31614694e-01
-7.14985371e-01 6.33002296e-02 1.36031397e-02 -2.77599931e-01
-9.28041711e-02 -4.59372029e-02 -2.79643595e-01 9.74395812e-01
-1.17382798e-02 5.71435690e-01 -7.65122235e-01 -5.63108563e-01
1.56924456e-01 2.58239388e-01 -6.61009073e-01 3.78624380e-01
4.32457417e-01 5.73761821e-01 -8.42812657e-01 5.50053000e-01
1.06687486e+00 -5.14344424e-02 1.89821407e-01 -5.43792427e-01
-4.86587472e-02 1.70133218e-01 -1.66900945e+00 1.53106606e+00
-5.52894473e-01 -3.15254480e-02 -3.89824621e-02 -1.68055189e+00
1.10426545e+00 6.36656165e-01 7.30313659e-01 -6.66576326e-01
2.72481680e-01 2.89138734e-01 -1.38138821e-02 -6.91039622e-01
-4.77455743e-02 -5.04463375e-01 -2.38272458e-01 1.62380651e-01
-2.01826960e-01 4.28808302e-01 -2.68153220e-01 -2.20274746e-01
5.52062571e-01 1.72512233e-02 7.62570262e-01 -3.73810858e-01
5.26230156e-01 -6.74038231e-01 1.29488802e+00 7.81008303e-01
-1.42676935e-01 5.44160545e-01 6.55870140e-01 -2.15663850e-01
-5.53890526e-01 -8.77922595e-01 -5.05838752e-01 8.88737857e-01
-1.51306242e-01 -1.57534689e-01 -3.37484181e-01 -6.65991426e-01
1.30194619e-01 6.68019772e-01 -8.92969906e-01 -6.31429732e-01
-2.94829130e-01 -1.15526319e+00 4.79271263e-01 8.06894898e-01
4.33093935e-01 -5.06610274e-01 1.22422939e-02 -1.94955878e-02
-2.74074912e-01 -4.95500565e-01 -5.67224145e-01 9.59550589e-02
-9.01652277e-01 -1.00433803e+00 -1.00048327e+00 -6.39202654e-01
5.52405238e-01 5.23118615e-01 5.60924470e-01 -3.84328961e-01
1.56874105e-01 -1.78583130e-01 -3.83173406e-01 -5.03306806e-01
-1.20969199e-01 4.56205904e-02 1.46293089e-01 2.89275408e-01
4.15924191e-01 -4.95617330e-01 -6.16006851e-01 4.33440328e-01
-7.37858355e-01 -1.31384954e-01 8.06082308e-01 1.22940946e+00
5.32443881e-01 5.95825911e-02 1.24440217e+00 -9.30588663e-01
5.34925759e-01 -9.58675802e-01 -4.12709236e-01 2.88868457e-01
-7.97387600e-01 1.03245094e-01 5.00545740e-01 -6.51609242e-01
-1.18100238e+00 -1.76481411e-01 2.49708995e-01 -4.66987193e-01
1.33862197e-01 7.71680415e-01 -5.21939933e-01 2.51639754e-01
5.47828496e-01 9.19306055e-02 3.90938632e-02 -6.25259340e-01
1.95813999e-01 8.93160045e-01 1.71171933e-01 -7.97066987e-01
3.57325763e-01 3.33476305e-01 2.71704018e-01 -3.23554963e-01
-1.01462972e+00 -6.02670014e-01 -2.54998595e-01 2.52389520e-01
5.74433744e-01 -1.17885602e+00 -8.33780468e-01 1.91757306e-01
-7.85300612e-01 1.97498128e-01 -6.33259416e-02 1.12331939e+00
-4.98238981e-01 5.80664098e-01 -2.33838409e-01 -1.00669563e+00
-2.82338828e-01 -1.34637952e+00 1.13461113e+00 5.95263503e-02
-6.18614815e-02 -8.21985722e-01 2.84628183e-01 3.74699950e-01
1.72170192e-01 2.34190539e-01 1.05083847e+00 -5.58493912e-01
1.05186477e-01 -3.85631502e-01 -4.86595780e-01 2.97971100e-01
4.42795813e-01 -2.72010207e-01 -8.44719827e-01 -4.25158113e-01
2.70498544e-01 -2.29886532e-01 1.00736892e+00 1.01504230e+00
1.49432433e+00 -3.29196095e-01 -5.29324412e-01 6.32344067e-01
1.26656139e+00 1.99770898e-01 6.76947415e-01 8.81392509e-02
6.22673094e-01 7.33101964e-01 6.63417995e-01 7.00212777e-01
3.77286315e-01 7.23831594e-01 1.76095560e-01 -2.97983170e-01
1.53890520e-01 -4.42897558e-01 6.65156320e-02 7.43144155e-01
7.11667538e-02 1.70970820e-02 -4.96786922e-01 4.76458609e-01
-2.09429026e+00 -8.94544065e-01 -4.23904479e-01 2.54554510e+00
6.81438804e-01 -9.67282355e-02 1.41867846e-01 1.99675605e-01
8.90011251e-01 2.09285125e-01 -3.28515917e-01 -1.54830441e-01
-1.25464216e-01 4.51285951e-02 6.26360893e-01 -1.46495849e-02
-1.32921636e+00 2.48245493e-01 5.99503374e+00 1.21098256e+00
-8.64887297e-01 2.23021656e-01 7.38341868e-01 2.68774718e-01
-2.90787518e-01 5.92934713e-02 -6.03118002e-01 4.91834521e-01
5.72220325e-01 -2.52446711e-01 -9.76919830e-02 6.40865088e-01
5.86623728e-01 3.02336276e-01 -9.74732578e-01 1.13582253e+00
-1.04743931e-02 -8.85284007e-01 4.95402031e-02 2.52622545e-01
8.59786332e-01 -3.50802362e-01 2.85388470e-01 6.75358355e-01
9.10855830e-02 -1.10151649e+00 4.23814982e-01 6.91144943e-01
7.24215269e-01 -6.32729113e-01 9.73172009e-01 3.32848638e-01
-1.29811704e+00 -4.87779826e-01 -6.66546166e-01 -1.32398635e-01
-1.18316561e-01 8.26521993e-01 -1.25517368e-01 1.24582052e+00
3.06356758e-01 1.08222532e+00 -4.26982969e-01 1.11566949e+00
2.87531700e-04 6.54585063e-01 1.61200702e-01 2.73102969e-01
7.92232603e-02 -5.45741618e-01 3.17331344e-01 9.45900500e-01
4.94815022e-01 -1.05051389e-02 4.85161573e-01 6.34175301e-01
-5.60129136e-02 5.07222831e-01 -7.76100457e-01 1.46851301e-01
4.15508956e-01 8.15400183e-01 -3.71431224e-02 -1.87038600e-01
-5.92516065e-01 3.54254097e-01 1.55528441e-01 4.18259174e-01
-8.17265689e-01 -1.97744817e-01 3.78393739e-01 -1.78654477e-01
2.16575265e-01 2.16461897e-01 -6.09909952e-01 -1.26185668e+00
-8.31046477e-02 -1.31412578e+00 8.50143909e-01 -3.53575051e-01
-1.59397840e+00 -6.04266636e-02 1.15596049e-01 -1.58931530e+00
1.37797609e-01 -4.54343885e-01 -4.45976138e-01 9.13197041e-01
-1.37922072e+00 -1.31628191e+00 -2.48892121e-02 3.49988163e-01
4.30476338e-01 -2.36178681e-01 7.64285326e-01 7.07784951e-01
-1.21339440e+00 7.89089084e-01 3.70728046e-01 -1.31224543e-02
1.06791270e+00 -9.64039862e-01 -5.71011841e-01 3.74447435e-01
-3.91950041e-01 7.95563102e-01 4.82758969e-01 -6.06434107e-01
-1.24121296e+00 -1.00091803e+00 5.55525303e-01 -1.35034084e-01
4.14264649e-01 -2.90633477e-02 -9.39587176e-01 7.12345302e-01
-1.18812650e-01 -1.67319160e-02 1.01652753e+00 6.28605604e-01
-7.18233049e-01 -4.76665586e-01 -8.31302106e-01 5.55679679e-01
7.88080990e-01 -2.36175060e-01 -6.87368095e-01 2.05625191e-01
2.98411995e-01 -6.80042133e-02 -9.27097797e-01 1.00316119e+00
7.68824399e-01 -6.52154922e-01 1.10077822e+00 -9.93416190e-01
5.24706483e-01 -2.27263913e-01 -2.96266973e-01 -1.34164166e+00
-6.42012239e-01 -2.71906912e-01 4.66378480e-02 1.31794536e+00
6.70044348e-02 -7.87020326e-01 3.31473887e-01 8.95618871e-02
1.96974166e-02 -7.68574357e-01 -9.78347123e-01 -7.90359437e-01
3.73743325e-01 -2.51751810e-01 6.47783816e-01 1.21709359e+00
-2.89338417e-02 6.27867579e-01 -7.90328205e-01 2.90588945e-01
6.97477758e-01 2.91199446e-01 7.76106179e-01 -1.48319590e+00
-1.76819652e-01 -6.41350448e-01 -2.88346171e-01 -8.72950554e-01
3.43242526e-01 -9.75503027e-01 -2.26286605e-01 -1.36468089e+00
6.99265838e-01 -5.36204517e-01 -6.86256707e-01 2.68444091e-01
-5.95495999e-01 -2.68583417e-01 -2.91757017e-01 1.74892232e-01
-2.21074089e-01 1.04902816e+00 1.20149636e+00 -2.38154605e-01
-1.50691390e-01 1.40370235e-01 -1.25991356e+00 6.37907743e-01
6.20340228e-01 -6.88682377e-01 -5.73587894e-01 -1.01763487e-01
-6.26185313e-02 2.09131837e-01 2.27419168e-01 -6.73535526e-01
-1.86833397e-01 -2.84932822e-01 3.53905171e-01 -3.73244047e-01
-2.14047208e-02 -7.35865116e-01 1.25118390e-01 5.44869900e-01
-5.68005145e-01 -1.21443085e-01 -4.53925543e-02 8.38869452e-01
-6.46810681e-02 -2.72452563e-01 6.92698419e-01 1.57481194e-01
-6.92380071e-02 3.75769526e-01 9.75790154e-03 1.30973924e-02
8.33661258e-01 1.58546716e-01 -2.46429458e-01 -2.39720047e-01
-2.49553010e-01 3.71171772e-01 -1.54061630e-01 4.49148178e-01
2.91168511e-01 -1.66603458e+00 -1.12523592e+00 2.31428713e-01
4.04932857e-01 -3.76529962e-01 6.23127759e-01 1.30669641e+00
2.20406428e-01 4.16267604e-01 9.68305171e-02 -4.92473751e-01
-1.03997016e+00 9.90650773e-01 1.98590130e-01 -5.13530314e-01
-4.83672291e-01 2.15945929e-01 8.58153462e-01 -5.19804835e-01
2.14988738e-01 -9.64596495e-02 -5.64548969e-01 3.31690669e-01
5.48066616e-01 5.87998629e-01 -1.28754571e-01 -5.64800024e-01
-1.62009180e-01 6.30281508e-01 -4.02757488e-02 2.32229993e-01
1.25166929e+00 -4.50061932e-02 1.69942807e-02 5.77519178e-01
1.48570895e+00 6.36029914e-02 -7.97108889e-01 -4.80658561e-01
-1.51769787e-01 -7.83043385e-01 4.37311351e-01 -2.79516220e-01
-1.03675354e+00 8.41660976e-01 7.62459159e-01 6.56529441e-02
1.13939321e+00 -4.56503868e-01 3.64033490e-01 3.27243842e-02
5.76470755e-02 -9.66213644e-01 3.46020274e-02 -1.23554409e-01
9.90835965e-01 -1.32421649e+00 3.08860898e-01 -5.35981357e-01
-6.48513913e-01 7.73003876e-01 3.80104244e-01 -1.90210357e-01
7.93380260e-01 -1.61371842e-01 -1.18761830e-01 -1.69114143e-01
-3.92058313e-01 -4.02418897e-03 4.83890444e-01 5.47469676e-01
7.27144003e-01 2.52806753e-01 -9.00988221e-01 1.12623250e+00
3.58776182e-01 4.75755557e-02 1.20808385e-01 4.35347676e-01
-2.31686849e-02 -1.05145645e+00 -5.99775016e-01 6.19853854e-01
-4.65396076e-01 6.09490089e-02 1.68883309e-01 7.06069887e-01
8.77517238e-02 1.06232488e+00 -2.99509019e-01 -1.33744478e-01
5.61240971e-01 -1.26221076e-01 1.83806062e-01 -4.77650106e-01
-2.88277775e-01 3.79878581e-01 -1.06772065e-01 -2.57950038e-01
-6.13317132e-01 -7.10805416e-01 -6.69189155e-01 2.96274461e-02
-6.91626966e-01 1.87238172e-01 2.13557482e-01 8.42332840e-01
2.17906892e-01 9.25083280e-01 1.20535803e+00 -4.33192015e-01
-1.15033567e+00 -1.16759384e+00 -7.93633401e-01 4.49817628e-01
2.70838559e-01 -1.13779700e+00 -3.94676477e-01 -4.07242924e-01]
|
[8.099390029907227, 5.394636154174805]
|
260abffd-54c1-4e07-a990-5bd7aa33a1ab
|
semantic-motion-segmentation-using-dense-crf
|
1504.06587
| null |
http://arxiv.org/abs/1504.06587v1
|
http://arxiv.org/pdf/1504.06587v1.pdf
|
Semantic Motion Segmentation Using Dense CRF Formulation
|
While the literature has been fairly dense in the areas of scene
understanding and semantic labeling there have been few works that make use of
motion cues to embellish semantic performance and vice versa. In this paper, we
address the problem of semantic motion segmentation, and show how semantic and
motion priors augments performance. We pro- pose an algorithm that jointly
infers the semantic class and motion labels of an object. Integrating semantic,
geometric and optical ow based constraints into a dense CRF-model we infer both
the object class as well as motion class, for each pixel. We found improvement
in performance using a fully connected CRF as compared to a standard
clique-based CRFs. For inference, we use a Mean Field approximation based
algorithm. Our method outperforms recently pro- posed motion detection
algorithms and also improves the semantic labeling compared to the
state-of-the-art Automatic Labeling Environment algorithm on the challenging
KITTI dataset especially for object classes such as pedestrians and cars that
are critical to an outdoor robotic navigation scenario.
|
['Prateek Singhal', 'N. Dinesh Reddy', 'K. Madhava Krishna']
|
2015-04-24
| null | null | null | null |
['motion-detection']
|
['computer-vision']
|
[ 2.34781399e-01 9.37682167e-02 -2.37751067e-01 -6.40846133e-01
-5.58617830e-01 -7.62824476e-01 7.49996841e-01 -1.17456866e-02
-7.81457961e-01 6.48884654e-01 -5.21024615e-02 -2.46795088e-01
1.79561704e-01 -6.49879575e-01 -6.15823328e-01 -6.31865203e-01
1.36597157e-01 1.02011013e+00 8.76563609e-01 8.71467292e-02
3.91965747e-01 4.28384930e-01 -1.46223950e+00 -7.32848942e-02
5.43128550e-01 4.62666094e-01 5.72742462e-01 9.30803120e-01
-1.40718043e-01 7.56675541e-01 -2.12317213e-01 1.05572514e-01
1.39363026e-02 -3.01308721e-01 -1.26795137e+00 4.81055230e-01
8.05899024e-01 -1.56046271e-01 -2.81501174e-01 8.94428551e-01
-3.79988644e-03 6.21579528e-01 7.76197672e-01 -1.31585646e+00
1.55498490e-01 1.30341381e-01 -6.80501997e-01 2.16269448e-01
2.55991995e-01 4.32273410e-02 7.80794203e-01 -5.83472490e-01
1.17380941e+00 1.39571321e+00 5.69628835e-01 4.41065222e-01
-1.22205091e+00 -2.69103676e-01 4.51443851e-01 3.57345611e-01
-1.06488299e+00 -4.20336127e-01 6.16544425e-01 -7.68638730e-01
1.00362825e+00 -2.14413509e-01 6.48579836e-01 6.79151475e-01
-1.42641455e-01 1.10184443e+00 9.78692889e-01 -3.89196247e-01
4.88948047e-01 -1.02516726e-01 3.32977623e-01 8.22387159e-01
1.95542619e-01 3.99611183e-02 -1.92825601e-01 2.87271351e-01
6.54628515e-01 -4.24599230e-01 2.42484331e-01 -7.32230186e-01
-1.33865047e+00 7.71759331e-01 3.29132259e-01 2.23495752e-01
-1.00748107e-01 7.53201187e-01 1.26315624e-01 -6.08265162e-01
3.76777381e-01 -3.69914509e-02 -5.51940143e-01 -2.56047640e-02
-1.35723341e+00 3.79713386e-01 9.71104980e-01 9.25408959e-01
9.72657502e-01 -8.96721482e-02 9.60770994e-02 6.11501813e-01
7.13500321e-01 4.68127251e-01 -7.71514326e-02 -1.80378044e+00
3.07792276e-01 1.69449225e-01 3.90517354e-01 -9.54499960e-01
-5.52615821e-01 6.18700683e-03 -9.73803103e-02 5.07487953e-01
8.80276740e-01 -1.02092788e-01 -1.43521762e+00 1.65412295e+00
6.61693871e-01 5.67268908e-01 1.33648127e-01 1.02241588e+00
6.40993237e-01 4.30003732e-01 5.17261624e-01 1.70342848e-01
1.41613555e+00 -1.40263808e+00 -5.41630566e-01 -7.60931909e-01
8.26060951e-01 -8.52145553e-01 2.13921189e-01 2.17335403e-01
-8.40241253e-01 -6.29707754e-01 -1.10016179e+00 -2.76110202e-01
-4.57275391e-01 1.15371995e-01 9.79652822e-01 7.06474185e-01
-1.10220861e+00 6.13774717e-01 -1.32623518e+00 -7.74061739e-01
5.99945307e-01 3.69327873e-01 -3.36842358e-01 -3.18531275e-01
-5.77338159e-01 1.19378531e+00 6.45598173e-01 -8.07011053e-02
-1.01129735e+00 -1.38067216e-01 -1.09776556e+00 -5.12028456e-01
3.88883471e-01 -8.96381080e-01 1.40621150e+00 -7.96243429e-01
-1.31528497e+00 1.03720415e+00 -4.96511787e-01 -3.59590650e-01
4.56248403e-01 -1.69516265e-01 -4.28830571e-02 3.95194441e-01
4.16257173e-01 1.73571014e+00 4.84433353e-01 -1.42633593e+00
-9.95069683e-01 -2.04542354e-01 2.43850976e-01 4.23012912e-01
5.66822588e-01 -1.55962169e-01 -6.19358540e-01 -2.56937861e-01
2.98408270e-01 -1.27478933e+00 -7.00925648e-01 7.09220096e-02
-4.68868852e-01 -1.15841448e-01 1.22180533e+00 -6.46582663e-01
5.22889733e-01 -1.77787304e+00 3.13392207e-02 -5.61643317e-02
-1.20404296e-01 4.97830771e-02 1.12864248e-01 -3.98092456e-02
2.13874251e-01 -1.15844570e-01 -6.56009912e-01 -5.87385595e-01
-5.70112057e-02 8.13705146e-01 1.60098404e-01 7.92641461e-01
1.26930609e-01 8.64832938e-01 -1.21695364e+00 -8.25001597e-01
6.57728374e-01 4.39894021e-01 -6.13916874e-01 -2.00317889e-01
-6.57350421e-01 6.85982525e-01 -5.41605353e-01 4.96460348e-01
7.22643197e-01 -1.40833259e-01 1.41874269e-01 1.20816886e-01
-2.79152364e-01 1.16117544e-01 -1.41422093e+00 2.21599340e+00
-7.15287477e-02 9.97336268e-01 2.60545403e-01 -9.78245914e-01
5.20042896e-01 9.06802043e-02 6.07286572e-01 -1.92123115e-01
1.99141830e-01 7.98268989e-02 -2.74759024e-01 -4.95396495e-01
7.80546010e-01 -1.65578827e-01 6.43865913e-02 1.65186763e-01
8.36096704e-02 -3.70850652e-01 3.90474737e-01 3.52509677e-01
7.19937861e-01 1.07768261e+00 1.24833837e-01 -5.60665011e-01
4.07705814e-01 8.59742463e-01 4.39556718e-01 7.36292183e-01
-3.68586570e-01 7.11474955e-01 4.84588556e-02 -1.30332604e-01
-1.04083908e+00 -7.28828549e-01 -1.66679308e-01 9.02329624e-01
7.23440528e-01 -3.96470539e-02 -9.95535254e-01 -5.86654305e-01
-9.46576521e-02 9.09402609e-01 -3.97081584e-01 4.79352802e-01
-8.11849535e-01 -7.94444621e-01 4.50689465e-01 8.02204907e-01
6.19360507e-01 -9.85075116e-01 -9.81764913e-01 4.67251867e-01
-4.57058609e-01 -1.64263856e+00 -3.28047365e-01 1.96848974e-01
-8.95185113e-01 -1.05372214e+00 -3.75716865e-01 -1.00195479e+00
5.63917875e-01 3.26055586e-01 1.02390981e+00 -2.14096263e-01
-4.56943959e-01 5.58257878e-01 -2.72698343e-01 -1.47626117e-01
-2.64210701e-01 -9.54459608e-02 -3.11354876e-01 -2.93801814e-01
4.57889646e-01 -1.03696942e-01 -7.59255528e-01 4.47278708e-01
-5.71963608e-01 2.36845940e-01 2.81974524e-01 3.31141293e-01
5.47572076e-01 -1.19882666e-01 -2.39944514e-02 -9.76572096e-01
-4.14490998e-01 -3.73437077e-01 -6.58385992e-01 -6.12003542e-02
-3.67859274e-01 1.43154398e-01 -2.86528170e-01 -7.35706016e-02
-1.33469903e+00 1.01811528e+00 -1.25327855e-01 -1.12249888e-02
-8.36139560e-01 2.99586467e-02 -4.61704135e-02 -8.62749442e-02
3.61334205e-01 -3.08872670e-01 -3.02774578e-01 -4.22272623e-01
9.56292748e-01 2.06396341e-01 9.24612403e-01 -6.41259551e-01
5.43049574e-01 1.05987513e+00 2.55015016e-01 -9.67853546e-01
-6.79782212e-01 -1.12517953e+00 -1.23329961e+00 -2.31356353e-01
1.73035264e+00 -9.59957540e-01 -2.91533262e-01 3.46883833e-01
-1.42572117e+00 -6.21017873e-01 -7.69402087e-02 6.46135926e-01
-1.04288363e+00 6.19216561e-01 -5.77799678e-01 -7.92275131e-01
3.99949223e-01 -1.25209057e+00 1.35879958e+00 2.89376706e-01
-3.46043557e-01 -1.32186902e+00 1.09630793e-01 6.61168158e-01
-1.51066920e-02 4.21514064e-01 3.83983016e-01 -2.75502652e-01
-1.07218194e+00 8.32178816e-02 -3.40029866e-01 -1.95258379e-01
-4.21777815e-01 -1.83365315e-01 -1.25948966e+00 1.06798314e-01
-3.96881640e-01 5.68021089e-04 1.16003704e+00 8.51414263e-01
5.18514037e-01 2.33573794e-01 -8.33108485e-01 4.87317085e-01
1.55695510e+00 2.72463232e-01 7.04935610e-01 4.56241071e-01
9.48188603e-01 8.46326172e-01 8.97376120e-01 -4.71839309e-03
5.19363344e-01 7.94525743e-01 3.63684893e-01 -1.58725753e-01
-4.06416357e-01 -2.17544109e-01 1.31054848e-01 2.34474108e-01
-5.39764529e-03 -4.67878819e-01 -9.89294350e-01 7.57813871e-01
-2.08635592e+00 -9.00126934e-01 -6.84304953e-01 1.68636429e+00
3.59610051e-01 1.42658457e-01 1.25978872e-01 -1.17392138e-01
6.81276143e-01 7.53979906e-02 -3.18239123e-01 2.27986835e-02
8.57053138e-03 -1.09506637e-01 1.02536869e+00 1.10629261e+00
-1.47626662e+00 1.53300488e+00 6.66205215e+00 5.46198487e-01
-5.29698730e-01 3.35172087e-01 4.42918718e-01 4.20649350e-01
1.42939361e-02 5.25530159e-01 -1.05157113e+00 -7.84404799e-02
7.06858516e-01 4.79305893e-01 6.94838539e-02 8.65730524e-01
3.85549814e-01 -9.07768130e-01 -1.15987134e+00 6.79514706e-01
2.14089528e-01 -1.10560751e+00 -5.23052216e-01 1.79085881e-01
9.09214854e-01 3.09261978e-01 -4.47005540e-01 -2.89760828e-02
7.83639371e-01 -9.84550595e-01 1.01178670e+00 4.73727912e-01
2.39085659e-01 -4.14929241e-01 6.10655546e-01 3.92557502e-01
-1.43932879e+00 4.30961043e-01 -9.95210633e-02 -1.16617344e-01
6.31636679e-01 2.12734088e-01 -9.54134464e-01 4.03041363e-01
5.02707720e-01 9.27015662e-01 -5.61383009e-01 1.22041333e+00
-2.60713637e-01 4.58731771e-01 -4.44783211e-01 1.63751438e-01
5.70922494e-01 -4.09220904e-01 5.29881835e-01 1.57255256e+00
-1.41561702e-01 1.84886366e-01 8.20105851e-01 7.17478096e-01
5.25351584e-01 -3.19144189e-01 -3.88584107e-01 1.38431668e-01
1.22189678e-01 1.23513222e+00 -1.74439764e+00 -6.49161100e-01
-2.38396630e-01 1.17524076e+00 7.72672221e-02 5.71964443e-01
-6.79670036e-01 -3.98001894e-02 5.16138136e-01 -1.91658118e-03
6.50309622e-01 -8.14458251e-01 -5.04146993e-01 -9.40761745e-01
-4.01062846e-01 4.54388969e-02 2.36658335e-01 -9.71197426e-01
-7.13467002e-01 8.68871137e-02 4.03512686e-01 -8.00788522e-01
-4.11255777e-01 -7.99539924e-01 -3.00788969e-01 6.49175107e-01
-1.47615504e+00 -1.21967900e+00 -3.89369696e-01 2.61204600e-01
8.70008111e-01 4.43616956e-01 5.17295718e-01 2.03044459e-01
-2.20074475e-01 -3.63597184e-01 -6.95506260e-02 1.76250696e-01
3.20053339e-01 -1.36206269e+00 2.82368749e-01 1.06645262e+00
2.76085883e-01 2.58658051e-01 9.35803413e-01 -8.69374514e-01
-9.35024321e-01 -1.07063901e+00 7.82962501e-01 -1.05678749e+00
4.39200103e-01 -1.65599868e-01 -5.04768491e-01 8.31332862e-01
1.23049207e-01 7.25948140e-02 1.63012221e-01 -2.46805727e-01
5.14308400e-02 5.93067050e-01 -1.03072643e+00 3.51526499e-01
1.29487169e+00 -3.21529567e-01 -5.39553285e-01 4.12350804e-01
5.60649574e-01 -5.34173250e-01 -3.40418637e-01 2.61549145e-01
3.63803774e-01 -7.07794845e-01 1.25716662e+00 -3.81164938e-01
1.01229340e-01 -8.09284210e-01 -3.99286181e-01 -7.61569619e-01
-1.25501752e-01 -5.11415973e-02 2.25866914e-01 9.65104401e-01
2.93371052e-01 -2.66735666e-02 1.30860424e+00 6.42186046e-01
-2.94304669e-01 -5.46078384e-02 -7.62829542e-01 -6.48883402e-01
1.33803468e-02 -7.54050434e-01 -2.52683580e-01 8.77650201e-01
-3.07194620e-01 3.71572226e-01 -1.40372649e-01 2.83739507e-01
9.25911009e-01 -3.71759161e-02 7.10387588e-01 -1.20460021e+00
-2.87797004e-01 -4.13415760e-01 -7.46207178e-01 -1.42524517e+00
4.44803029e-01 -9.11415756e-01 6.46938086e-01 -2.07134986e+00
1.82155892e-01 -5.12552619e-01 3.36591452e-01 4.92268145e-01
-4.07652557e-02 4.52144980e-01 1.01590328e-01 1.67047456e-01
-9.85959888e-01 3.66456322e-02 1.09806526e+00 -2.28210494e-01
-1.57856762e-01 -2.89824218e-01 -1.03989288e-01 1.04663086e+00
5.90820074e-01 -6.40574992e-01 -4.66348857e-01 -4.02048320e-01
-2.43580669e-01 1.98519811e-01 7.22625494e-01 -1.04604268e+00
4.93081838e-01 -2.66470551e-01 3.10049742e-01 -1.01111531e+00
5.94000757e-01 -6.91639245e-01 9.94248316e-02 5.78408659e-01
-1.40141562e-01 -2.47138038e-01 1.95050284e-01 9.25951004e-01
-5.77261820e-02 -4.70065266e-01 9.53335166e-01 -4.55158859e-01
-1.63988376e+00 1.46075860e-01 -6.94839537e-01 2.36959562e-01
1.16181982e+00 -5.75626373e-01 -1.08115099e-01 -2.40629688e-01
-1.10891652e+00 3.79392505e-01 5.65289140e-01 3.89065444e-01
3.29207212e-01 -8.24709594e-01 -4.38866496e-01 -2.26908028e-01
-8.94324332e-02 2.63712674e-01 5.43419365e-03 8.48517895e-01
-1.07258272e+00 6.43369794e-01 -5.45544289e-02 -1.06057274e+00
-1.23972213e+00 4.46568489e-01 3.16626787e-01 2.08409593e-01
-7.48377025e-01 8.22595358e-01 2.67546266e-01 -3.33188891e-01
1.50361389e-01 -3.67772371e-01 -3.29138607e-01 -6.40290156e-02
1.01897232e-01 5.83480477e-01 -3.13874066e-01 -1.17174208e+00
-6.19516730e-01 8.98193002e-01 3.46386015e-01 -6.87801778e-01
9.00133133e-01 -4.80761141e-01 1.59229666e-01 4.39338386e-01
1.06319594e+00 -3.53780508e-01 -1.70652568e+00 1.15159586e-01
3.89661491e-01 -4.05677050e-01 1.27349272e-01 -6.37291729e-01
-8.10110629e-01 9.71404672e-01 5.51975131e-01 -3.10368687e-01
4.59682584e-01 2.86719829e-01 5.15517533e-01 2.15895697e-01
5.51568627e-01 -1.26590347e+00 4.17209137e-03 6.61273658e-01
-3.06625962e-02 -1.31837654e+00 1.77194402e-01 -9.20423210e-01
-6.39444411e-01 9.65543330e-01 5.33138871e-01 -3.03636819e-01
6.13675952e-01 2.20989853e-01 1.16381228e-01 -1.36190310e-01
-1.88609049e-01 -6.49030268e-01 2.51539499e-01 8.49219084e-01
2.55950153e-01 7.57036731e-02 -2.58122943e-02 -3.01370740e-01
-5.43599352e-02 1.13616303e-01 4.84527677e-01 1.11601877e+00
-8.46432984e-01 -1.02967238e+00 -3.83754939e-01 -1.14825659e-01
-1.87778160e-01 1.60047293e-01 -2.78963953e-01 1.04107332e+00
5.41747153e-01 1.29390442e+00 1.63572267e-01 8.02940782e-03
-1.69241473e-01 1.09181650e-01 7.35471010e-01 -6.97520316e-01
7.65527189e-02 2.79550821e-01 5.41193724e-01 -6.31822944e-01
-1.18110967e+00 -1.06064534e+00 -1.85619211e+00 7.61874169e-02
-4.07402009e-01 -7.82933459e-02 1.06272304e+00 1.43233097e+00
-2.31594726e-01 4.50467795e-01 -2.58581817e-01 -1.31549942e+00
2.60828406e-01 -6.13279283e-01 -3.69293392e-01 4.61779028e-01
1.06354095e-01 -8.36090326e-01 -1.58016294e-01 5.98987460e-01]
|
[8.546489715576172, -1.59078848361969]
|
5b5f962d-0723-46ba-95b5-ecd3e26fa7a7
|
learning-to-transfer-transferring-latent-task
| null | null |
http://openaccess.thecvf.com/content_iccv_2015/html/Almaev_Learning_to_Transfer_ICCV_2015_paper.html
|
http://openaccess.thecvf.com/content_iccv_2015/papers/Almaev_Learning_to_Transfer_ICCV_2015_paper.pdf
|
Learning to Transfer: Transferring Latent Task Structures and Its Application to Person-Specific Facial Action Unit Detection
|
In this article we explore the problem of constructing person-specific models for the detection of facial Action Units (AUs), addressing the problem from the point of view of Transfer Learning and Multi-Task Learning. Our starting point is the fact that some expressions, such as smiles, are very easily elicited, annotated, and automatically detected, while others are much harder to elicit and to annotate. We thus consider a novel problem: all AU models for the target subject are to be learnt using person-specific annotated data for a reference AU (AU12 in our case), and no data or little data regarding the target AU. In order to design such a model, we propose a novel Multi-Task Learning and the associated Transfer Learning framework, in which we consider both relations across subjects and AUs. That is to say, we consider a tensor structure among the tasks. Our approach hinges on learning the latent relations among tasks using one single reference AU, and then transferring these latent relations to other AUs. We show that we are able to effectively make use of the annotated data for AU12 when learning other person-specific AU models, even in the absence of data for the target task. Finally, we show the excellent performance of our method when small amounts of annotated data for the target tasks are made available.
|
['Brais Martinez', 'Michel Valstar', 'Timur Almaev']
|
2015-12-01
| null | null | null |
iccv-2015-12
|
['action-unit-detection', 'facial-action-unit-detection']
|
['computer-vision', 'computer-vision']
|
[ 3.36331010e-01 3.69869024e-01 1.19204573e-01 -3.32618386e-01
-8.35698545e-01 -4.69434589e-01 6.27330005e-01 -1.30750522e-01
-3.80731463e-01 5.97156703e-01 2.14691222e-01 4.83601987e-01
7.93376490e-02 -4.60545540e-01 -7.13142574e-01 -6.95936322e-01
-1.14441449e-02 6.22439265e-01 -2.99147759e-02 -3.30017805e-01
-2.71868169e-01 2.15211868e-01 -1.35639060e+00 4.95601445e-01
5.69490016e-01 1.08237815e+00 -1.39020085e-01 5.37185073e-01
3.48522693e-01 7.76310384e-01 -4.97487426e-01 -7.17993021e-01
3.36508960e-01 -3.85819733e-01 -1.20846403e+00 3.94854963e-01
7.39933729e-01 -2.19787493e-01 5.65043576e-02 9.17058527e-01
3.74355644e-01 2.29424581e-01 9.63194489e-01 -1.45849562e+00
-4.52228844e-01 2.75664985e-01 -4.71653253e-01 -3.16935748e-01
4.01456565e-01 -4.65563945e-02 1.29319429e+00 -8.01862419e-01
7.29278564e-01 1.22990513e+00 4.15338010e-01 9.47210371e-01
-1.44832790e+00 -4.24655855e-01 6.60018474e-02 2.21286401e-01
-1.21233690e+00 -4.96557027e-01 8.38128686e-01 -9.37011957e-01
3.45472217e-01 7.90007189e-02 2.90842742e-01 1.59981108e+00
-2.95572817e-01 9.60405052e-01 1.29811823e+00 -4.22438294e-01
4.20160331e-02 2.97810793e-01 3.12831849e-01 8.81379306e-01
-1.35869324e-01 -5.69610670e-02 -3.90712827e-01 -9.98857319e-02
6.69538856e-01 -2.43580312e-01 -2.33971760e-01 -4.44410324e-01
-1.21480739e+00 8.29668820e-01 2.78673738e-01 5.00864983e-01
-3.28476578e-01 3.87276784e-02 6.06885552e-01 4.15213078e-01
7.42611766e-01 2.97322780e-01 -4.18549150e-01 1.12807013e-01
-5.42573631e-01 3.07283729e-01 8.19378734e-01 8.03336322e-01
1.09808648e+00 -3.93346876e-01 -2.90531516e-01 7.55907059e-01
-1.05907880e-01 2.33489186e-01 2.43966386e-01 -8.57651472e-01
5.72147310e-01 4.35103953e-01 3.93395483e-01 -8.94192934e-01
-4.14789855e-01 -8.38694125e-02 -6.99111879e-01 1.18504144e-01
8.23804379e-01 -5.41963756e-01 -3.88588220e-01 2.20002294e+00
2.56692559e-01 2.09335044e-01 1.57878190e-01 7.82469690e-01
4.76944506e-01 4.71866310e-01 7.13894367e-02 -2.68429786e-01
1.37324429e+00 -8.63805711e-01 -5.46498477e-01 -1.82986423e-01
1.01797616e+00 -4.23581958e-01 9.55925167e-01 4.69125539e-01
-8.41709554e-01 -7.45132744e-01 -6.54719770e-01 -1.92595258e-01
-2.15965346e-01 5.96148014e-01 5.20426035e-01 2.57689059e-01
-8.84777486e-01 5.82052648e-01 -5.68836629e-01 -5.60894310e-01
1.45038605e-01 3.64332080e-01 -6.91817760e-01 -4.03267071e-02
-1.13298488e+00 1.18998122e+00 3.78485858e-01 1.29110113e-01
-9.20468688e-01 -2.95744270e-01 -9.28717792e-01 -7.97293633e-02
6.68819070e-01 -6.85569465e-01 1.25728559e+00 -1.52418947e+00
-1.30564964e+00 1.18900597e+00 -5.97557276e-02 -1.01075806e-01
6.19843543e-01 -2.66879976e-01 -1.50627464e-01 2.10311398e-01
-1.68926490e-03 4.49377596e-01 1.11216760e+00 -1.33527827e+00
-4.06824112e-01 -5.96087575e-01 4.43102032e-01 5.50560541e-02
-5.46818137e-01 2.86219239e-01 -8.93469304e-02 -4.08879250e-01
-5.14129519e-01 -1.18365240e+00 2.03636638e-03 8.96058828e-02
-3.62981677e-01 -6.83535278e-01 6.63652658e-01 -7.31783807e-01
7.11158812e-01 -2.26432920e+00 7.82726467e-01 1.45458609e-01
4.50488418e-01 1.49608403e-01 -3.37804437e-01 2.28561983e-01
-2.20039859e-01 -1.79450437e-01 -1.40210271e-01 -6.07811809e-01
6.33574054e-02 2.76770324e-01 7.94763267e-02 3.82307172e-01
3.88253689e-01 8.59124064e-01 -8.97615314e-01 -5.63028097e-01
-8.49683136e-02 2.17688382e-01 -4.12175238e-01 5.21161556e-01
-2.64332652e-01 7.90498316e-01 -6.00418270e-01 2.44125590e-01
2.06799984e-01 -2.42934719e-01 1.68070421e-01 -5.19519806e-01
1.71879515e-01 -1.05725519e-01 -1.12336516e+00 1.49508619e+00
-7.34766364e-01 3.65383714e-01 2.98748851e-01 -1.34299886e+00
8.94643426e-01 6.61481619e-01 7.15954065e-01 -1.18265867e-01
2.31369689e-01 2.17376068e-01 8.61814097e-02 -7.65939474e-01
-5.31217232e-02 -3.99139822e-01 -1.96724594e-01 6.38588488e-01
4.96531993e-01 1.84164438e-02 2.60109156e-01 -2.01490689e-02
9.17574465e-01 1.55169696e-01 4.52009350e-01 -6.96027055e-02
6.83789849e-01 -3.50989014e-01 3.89792383e-01 2.92760700e-01
-1.51537940e-01 1.67044610e-01 7.46938586e-01 -5.30390084e-01
-1.00813270e+00 -7.41912484e-01 9.71856341e-02 1.36813092e+00
-4.22775775e-01 -4.47900504e-01 -8.03267896e-01 -1.01710987e+00
-1.45637199e-01 1.47739202e-01 -1.02129364e+00 7.24207461e-02
-5.57521164e-01 -5.23856938e-01 3.67505580e-01 3.65063995e-01
3.85481685e-01 -1.03536701e+00 -4.33798641e-01 1.61460298e-03
-5.11364102e-01 -1.47728920e+00 -4.37655538e-01 -1.31749392e-01
-4.56793547e-01 -1.24867702e+00 -8.88128400e-01 -6.07003152e-01
7.07693636e-01 -2.60207295e-01 9.51239765e-01 -1.92922488e-01
-4.07960527e-02 6.66134477e-01 -3.27944130e-01 -3.38677526e-01
-4.10923809e-01 -1.15493745e-01 1.77483127e-01 8.76026690e-01
2.70865828e-01 -5.68975627e-01 -1.49362817e-01 3.13510239e-01
-7.14695752e-01 1.12239577e-01 6.01853848e-01 9.03644383e-01
1.49461925e-02 -4.97923523e-01 3.45955551e-01 -9.59912658e-01
5.14871478e-01 -3.26820403e-01 -2.73165733e-01 4.82385993e-01
9.58178788e-02 1.16037048e-01 5.34648180e-01 -7.63304830e-01
-9.04542983e-01 3.99744809e-01 1.95202865e-02 -5.97348452e-01
-3.18251729e-01 3.47176135e-01 -2.52249539e-01 -2.61370510e-01
7.83663094e-01 -1.87793821e-01 1.83728173e-01 -4.82849807e-01
3.46709728e-01 6.17315114e-01 3.72851223e-01 -1.00405180e+00
7.58497238e-01 3.34781557e-01 2.42672071e-01 -9.51716781e-01
-1.32516718e+00 -4.57014322e-01 -1.20398438e+00 -4.06030804e-01
1.14857626e+00 -8.57014298e-01 -9.16589975e-01 3.89651418e-01
-1.46137321e+00 -5.22251070e-01 -9.60592702e-02 3.38777453e-01
-8.77437532e-01 4.49608594e-01 -6.52625442e-01 -8.69083047e-01
-4.68400791e-02 -1.03514099e+00 1.29309702e+00 -3.84251922e-01
-4.06663567e-01 -1.13452327e+00 1.13711730e-01 5.38584232e-01
2.51282323e-02 4.37953055e-01 9.05429125e-01 -8.28716755e-01
-1.51166975e-01 -6.74199387e-02 -2.18181312e-01 6.81027234e-01
2.91489393e-01 -3.37089092e-01 -1.10477281e+00 -2.70992190e-01
1.97194144e-01 -9.21864629e-01 6.51760459e-01 -5.47695858e-03
1.11174464e+00 -5.50076246e-01 -2.48372927e-01 2.33001500e-01
1.06022596e+00 -2.83843637e-01 2.38212243e-01 -2.45588630e-01
9.89755750e-01 1.33007073e+00 5.23392022e-01 2.43265986e-01
4.25387710e-01 1.22092998e+00 4.71027978e-02 -6.22187555e-02
1.90563396e-01 1.71976408e-03 5.99217236e-01 5.49153268e-01
-6.37412786e-01 1.44170046e-01 -7.68185437e-01 3.96446556e-01
-1.94630408e+00 -9.54443097e-01 -8.28373339e-03 2.08226562e+00
9.41746950e-01 -2.16199681e-01 5.28183162e-01 -7.33035430e-02
6.30246401e-01 8.16805474e-03 -2.15496987e-01 -1.72159523e-01
1.80784613e-01 2.58352041e-01 -2.18732655e-02 5.56389868e-01
-1.38939238e+00 8.52899075e-01 6.01067066e+00 4.63673592e-01
-1.09415281e+00 4.76505309e-01 4.36652541e-01 1.32652327e-01
1.12271965e-01 -2.02861920e-01 -6.83053851e-01 2.13622674e-01
8.87007833e-01 3.67408507e-02 2.58310705e-01 8.35990965e-01
1.26929238e-01 7.23785609e-02 -1.71079409e+00 1.13000488e+00
4.10601854e-01 -8.06543529e-01 2.91110128e-02 -8.68304726e-03
4.99767274e-01 -4.42212820e-01 -2.37375110e-01 3.48534316e-01
2.19250664e-01 -8.62363040e-01 5.10614574e-01 5.45065582e-01
7.84539580e-01 -2.64609694e-01 6.43901467e-01 4.67837691e-01
-9.13414061e-01 -1.18850984e-01 -1.43670872e-01 -1.82164863e-01
2.68694498e-02 3.08118314e-01 -6.06887817e-01 5.39479613e-01
3.05440068e-01 8.00995350e-01 -4.83346969e-01 5.50175011e-01
-3.31170559e-01 2.88917124e-01 -1.21181965e-01 -2.86745150e-02
1.33306623e-01 -3.89463753e-01 4.22341287e-01 1.07375216e+00
2.55498350e-01 7.18254372e-02 5.25838017e-01 8.40546131e-01
-6.71607554e-02 3.38050514e-01 -9.19747114e-01 1.82461530e-01
-4.09931093e-01 1.50210834e+00 -3.75064723e-02 -4.08124864e-01
-6.78321302e-01 1.24610198e+00 7.81333148e-01 3.98489267e-01
-5.27384520e-01 9.44104046e-02 5.64449430e-01 -7.13275671e-02
2.11922992e-02 -2.99914658e-01 1.31694719e-01 -1.46617782e+00
1.91952974e-01 -8.37184906e-01 4.79632944e-01 -7.00018883e-01
-1.63095689e+00 5.55380821e-01 2.46223196e-01 -1.13230371e+00
-5.75500607e-01 -1.02488208e+00 -4.88468915e-01 8.26140106e-01
-1.12933969e+00 -1.73145115e+00 -2.66148418e-01 1.00017643e+00
3.00725520e-01 -1.37652678e-03 9.27480817e-01 2.69781083e-01
-4.91125256e-01 4.51290965e-01 -3.94981802e-01 4.51529860e-01
9.47235286e-01 -1.24692595e+00 -2.54615933e-01 6.68806612e-01
3.02971035e-01 4.00382400e-01 6.58784807e-01 -3.10120016e-01
-1.10185659e+00 -1.09980333e+00 9.77609634e-01 -6.75899029e-01
1.02013707e+00 -8.56965065e-01 -8.80986273e-01 1.13790929e+00
-5.86605296e-02 1.76783711e-01 7.07309723e-01 5.87019205e-01
-4.90066946e-01 -2.99748071e-02 -8.92584324e-01 4.05698866e-01
9.93236244e-01 -7.71936119e-01 -6.06410921e-01 6.64007127e-01
2.90187746e-01 -2.20317394e-01 -9.72827137e-01 2.45429739e-01
4.49922651e-01 -7.01825321e-01 7.11320996e-01 -9.45045531e-01
4.68061835e-01 5.25772162e-02 -2.00772230e-02 -1.53414583e+00
-7.83464089e-02 -5.73444247e-01 -7.15728626e-02 1.18441463e+00
4.12932754e-01 -4.25469726e-01 5.78625977e-01 7.23760009e-01
8.16500708e-02 -4.94648725e-01 -9.46748793e-01 -7.58659005e-01
8.92344043e-02 -3.88675332e-01 1.00112827e-02 1.15432978e+00
2.02750996e-01 7.82886088e-01 -8.65634143e-01 -3.48943025e-02
6.13397419e-01 1.38855040e-01 9.83031213e-01 -1.42664349e+00
-5.39537072e-01 -2.65477598e-02 -4.46091801e-01 -8.45440686e-01
7.03054726e-01 -8.99246275e-01 1.96116939e-02 -1.03590238e+00
3.58798772e-01 -2.62930572e-01 4.96661440e-02 8.55406940e-01
-1.59320131e-01 2.63037175e-01 3.54691267e-01 2.45935231e-01
-5.52643538e-01 5.46789706e-01 1.41560197e+00 -1.91486925e-01
1.05316952e-01 3.05006027e-01 -5.37583232e-01 7.78201580e-01
4.72850204e-01 -2.25414082e-01 -1.69164479e-01 -2.18068942e-01
9.05057788e-02 1.24349095e-01 7.76336312e-01 -6.89604700e-01
-3.87871563e-02 -6.85040057e-02 1.03266478e-01 8.99237022e-02
6.10955119e-01 -7.98415244e-01 -2.39049688e-01 1.78193673e-01
-3.70972306e-01 -3.26603442e-01 -8.93690139e-02 2.86308140e-01
-2.36608684e-01 -3.02175432e-01 7.50134766e-01 -2.32674673e-01
-7.14219451e-01 4.83138978e-01 -1.05876148e-01 4.16059326e-03
1.12529743e+00 1.61082283e-01 -9.80560184e-02 -4.62620586e-01
-1.36610246e+00 1.74957529e-01 1.47594661e-01 3.84454876e-01
2.50864059e-01 -1.39270663e+00 -9.14184391e-01 -3.15409787e-02
5.26326358e-01 -2.47664288e-01 1.97883740e-01 1.04329526e+00
2.23432064e-01 2.31599942e-01 -5.01082182e-01 -4.66263235e-01
-1.37565589e+00 6.94050133e-01 3.91008586e-01 -3.28267097e-01
-3.65842819e-01 5.93913198e-01 6.16659522e-01 -3.66882741e-01
2.66488064e-02 -1.56741127e-01 -3.60416353e-01 3.77341092e-01
4.78683740e-01 1.70678884e-01 -1.26150951e-01 -9.29724753e-01
-3.66299152e-02 7.09207535e-01 9.35798213e-02 -1.66256681e-01
1.22944200e+00 7.18666464e-02 -2.63782024e-01 7.72102177e-01
1.38889754e+00 -1.43024445e-01 -1.12699127e+00 -5.11577666e-01
1.85611099e-01 -2.76528805e-01 -4.89181072e-01 -4.58509594e-01
-8.07994783e-01 1.09357977e+00 2.91464806e-01 9.30787027e-02
1.05445421e+00 4.32009876e-01 5.16883016e-01 4.62708324e-01
4.50095356e-01 -9.67554748e-01 4.17020798e-01 4.63120818e-01
1.17091691e+00 -1.39714170e+00 -4.10273433e-01 -5.59606850e-01
-8.15401316e-01 1.15994740e+00 7.68250704e-01 -1.36171341e-01
3.92241150e-01 -8.78826454e-02 -5.79800270e-02 -3.51891845e-01
-6.73984468e-01 -5.66848934e-01 5.75756490e-01 7.41489708e-01
4.18822378e-01 9.37593430e-02 -2.16829032e-01 4.60201681e-01
-7.11380783e-03 -3.78833711e-02 4.01940733e-01 4.69334215e-01
-3.96218635e-02 -1.35031247e+00 -2.45414883e-01 3.79834652e-01
-4.68323737e-01 1.53925344e-01 -6.68061674e-01 8.03863704e-01
3.29165637e-01 6.16266251e-01 -6.53525293e-02 -3.35039854e-01
3.39700729e-01 3.81082177e-01 7.93195009e-01 -9.43886817e-01
-3.77682567e-01 -1.02903627e-01 4.25892681e-01 -6.93328142e-01
-8.10623467e-01 -8.64719510e-01 -5.13731897e-01 8.68119150e-02
-9.62079270e-04 1.34025246e-01 3.61414433e-01 1.38560641e+00
-1.27499878e-01 8.25785995e-02 6.94153845e-01 -1.07386708e+00
-5.63053429e-01 -1.17922616e+00 -6.21892393e-01 9.37677324e-01
4.73209828e-01 -8.92806053e-01 -2.94876605e-01 2.81521827e-01]
|
[13.586904525756836, 1.6718381643295288]
|
368f3b12-fe54-494a-85cc-60f4ec234852
|
learning-symbolic-rules-over-abstract-meaning
|
2307.02689
| null |
https://arxiv.org/abs/2307.02689v1
|
https://arxiv.org/pdf/2307.02689v1.pdf
|
Learning Symbolic Rules over Abstract Meaning Representations for Textual Reinforcement Learning
|
Text-based reinforcement learning agents have predominantly been neural network-based models with embeddings-based representation, learning uninterpretable policies that often do not generalize well to unseen games. On the other hand, neuro-symbolic methods, specifically those that leverage an intermediate formal representation, are gaining significant attention in language understanding tasks. This is because of their advantages ranging from inherent interpretability, the lesser requirement of training data, and being generalizable in scenarios with unseen data. Therefore, in this paper, we propose a modular, NEuro-Symbolic Textual Agent (NESTA) that combines a generic semantic parser with a rule induction system to learn abstract interpretable rules as policies. Our experiments on established text-based game benchmarks show that the proposed NESTA method outperforms deep reinforcement learning-based techniques by achieving better generalization to unseen test games and learning from fewer training interactions.
|
['Alexander Gray', 'Asim Munawar', 'Pavan Kapanipathi', 'Achille Fokoue', 'Michiaki Tatsubori', 'Rosario Uceda-Sosa', 'Keerthiram Murugesan', 'Prithviraj Sen', 'Daiki Kimura', 'Sarathkrishna Swaminathan', 'Subhajit Chaudhury']
|
2023-07-05
| null | null | null | null |
['representation-learning']
|
['methodology']
|
[ 1.16184622e-01 4.70309347e-01 -2.31516585e-01 -2.65275657e-01
1.12750614e-02 -5.92145145e-01 7.82283962e-01 3.02957505e-01
-6.45528316e-01 9.70057547e-01 -8.30389187e-02 -5.74058414e-01
-4.39723879e-01 -1.24200010e+00 -7.40796864e-01 -3.30369502e-01
-7.34900683e-02 8.49991262e-01 2.76801169e-01 -6.95194483e-01
-6.57791048e-02 1.62994951e-01 -1.65936279e+00 1.45296142e-01
1.05748606e+00 9.87435758e-01 -2.97958478e-02 3.45711112e-01
-2.67192841e-01 1.27391732e+00 -6.17918491e-01 -5.39925337e-01
6.35922477e-02 -2.87942797e-01 -7.63886511e-01 -4.22924250e-01
-6.53259978e-02 -4.52961743e-01 -3.81848902e-01 1.04398835e+00
-3.48330624e-02 3.74746919e-01 4.38470483e-01 -1.31353223e+00
-8.37507427e-01 9.33801651e-01 -3.59510742e-02 -3.60004455e-02
3.38290900e-01 3.13243598e-01 1.11819649e+00 -3.54038835e-01
5.29413939e-01 1.52287436e+00 5.19283235e-01 1.14422715e+00
-1.29957616e+00 -5.55050790e-01 3.50461572e-01 2.95292556e-01
-8.74265134e-01 -2.87236348e-02 7.85089135e-01 -2.80982465e-01
1.37079120e+00 6.41623698e-03 7.49314368e-01 1.65967667e+00
1.33143574e-01 8.40015948e-01 1.13101339e+00 -4.97887880e-01
6.97288513e-01 5.34471050e-02 8.84151757e-02 9.65228498e-01
6.88291132e-01 6.17173195e-01 -3.64265054e-01 -2.12271921e-02
6.45013988e-01 1.45341724e-01 4.63349968e-02 -7.32213080e-01
-5.95183969e-01 1.24653375e+00 4.04339015e-01 2.09260181e-01
-2.12541774e-01 4.90524262e-01 8.40356231e-01 3.38168561e-01
2.52369970e-01 8.24411333e-01 -5.88857889e-01 -3.81934345e-01
-4.60955173e-01 4.99102414e-01 9.48737323e-01 7.16178477e-01
6.38258517e-01 6.55845642e-01 7.14900047e-02 6.32404745e-01
2.62469292e-01 3.87295723e-01 1.12663317e+00 -6.47741079e-01
2.61784524e-01 9.84358549e-01 -3.17199796e-01 -9.09577608e-01
-4.47878331e-01 -4.50250924e-01 -4.12563056e-01 5.86925328e-01
2.36148849e-01 -3.54250222e-01 -1.05498791e+00 2.08266830e+00
-1.13497473e-01 2.43084133e-01 5.80844283e-01 6.34035289e-01
6.31241083e-01 4.10674781e-01 5.38932323e-01 3.12385052e-01
1.12289691e+00 -7.76179135e-01 -3.68589699e-01 -6.45131171e-01
7.50724077e-01 3.75070572e-01 1.09519279e+00 4.59970981e-01
-7.08453953e-01 -4.54910964e-01 -1.17355263e+00 4.94408160e-02
-9.87895072e-01 -3.17936361e-01 8.97256255e-01 6.53211653e-01
-8.67676377e-01 6.22536063e-01 -8.36597264e-01 -4.68891263e-01
5.23764372e-01 6.45839036e-01 -1.52920172e-01 3.01795173e-02
-1.35281968e+00 9.09941733e-01 1.29483163e+00 -2.33518422e-01
-9.80537832e-01 -4.41663742e-01 -1.30447447e+00 3.80335212e-01
8.47616255e-01 -7.48740137e-01 1.41516387e+00 -1.07299113e+00
-2.00412655e+00 4.66874987e-01 3.73642951e-01 -1.13015342e+00
3.52028638e-01 -4.24255058e-02 -2.79285938e-01 5.29002696e-02
-2.47322306e-01 6.91549420e-01 8.10897291e-01 -1.07122922e+00
-5.87941945e-01 -3.95507455e-01 7.63441086e-01 2.13072728e-02
-5.05266249e-01 -5.88879645e-01 2.82879293e-01 -6.04351759e-01
-3.27227145e-01 -8.18311155e-01 -2.52247751e-01 -5.23410201e-01
-1.26934201e-01 -5.46854377e-01 8.11507285e-01 -1.53613359e-01
1.00372422e+00 -1.84828949e+00 2.66266286e-01 9.52761397e-02
3.75145912e-01 5.74376881e-01 -1.44405231e-01 2.63924986e-01
-1.39100999e-02 3.57499003e-01 1.39644325e-01 5.50124198e-02
3.16014558e-01 8.66262019e-01 -4.38141912e-01 -3.63665581e-01
5.49943209e-01 1.16251564e+00 -1.23622859e+00 -1.40453562e-01
4.21521097e-01 -1.78904355e-01 -8.24173570e-01 1.22099139e-01
-7.64452994e-01 1.56443685e-01 -7.65268147e-01 3.51726741e-01
2.18790043e-02 2.17606835e-02 4.54892844e-01 3.58174801e-01
2.18936190e-01 4.02915403e-02 -8.19706857e-01 1.68034947e+00
-6.63285792e-01 4.88361537e-01 -3.13911080e-01 -1.35456669e+00
9.62147772e-01 2.57303447e-01 2.89345067e-03 -8.09480667e-01
3.83389950e-01 1.98046923e-01 3.05704266e-01 -5.84448576e-01
3.87163550e-01 -3.55216414e-01 -2.63206989e-01 5.87963611e-02
4.90825474e-01 -8.52614567e-02 4.46290672e-01 2.87349038e-02
1.25745583e+00 6.38199806e-01 5.37733257e-01 -6.35103136e-02
5.75536311e-01 2.46911436e-01 6.97024405e-01 1.08102405e+00
-6.17198013e-02 -3.23539414e-02 6.03819788e-01 -8.30596745e-01
-7.29360759e-01 -8.95919442e-01 2.98920572e-01 1.18285906e+00
2.52188258e-02 -5.87380767e-01 -7.57506132e-01 -9.12852466e-01
7.94083700e-02 9.44043159e-01 -7.33982980e-01 -6.95315599e-01
-5.13835609e-01 -1.87125117e-01 8.36644530e-01 6.97783470e-01
4.18928444e-01 -1.37102175e+00 -8.83405328e-01 4.49192226e-01
3.51523131e-01 -1.08165121e+00 4.42307472e-01 6.47375226e-01
-9.48626280e-01 -1.10343969e+00 -3.04375645e-02 -5.89416683e-01
4.94668186e-01 -2.54959226e-01 1.06736231e+00 1.42216191e-01
-7.91255012e-02 4.06594485e-01 -6.07758582e-01 -8.18298757e-01
-6.00048840e-01 2.28826821e-01 2.43160531e-01 -3.25880319e-01
6.63746238e-01 -6.38034284e-01 4.52035591e-02 -1.44784143e-02
-1.12389135e+00 1.73213437e-01 4.53510940e-01 1.26725602e+00
5.55377640e-02 1.03051692e-01 7.25151777e-01 -1.13599491e+00
1.00590491e+00 -5.43474793e-01 -7.01333463e-01 3.41564894e-01
-7.72464752e-01 7.00323880e-01 1.30715561e+00 -5.14275372e-01
-1.28705215e+00 -2.49238580e-01 1.27370358e-01 -3.80308419e-01
-6.50640070e-01 7.40175068e-01 -3.28194164e-02 3.79662476e-02
1.08380401e+00 2.78832614e-01 1.84009410e-02 -1.50798365e-01
3.32110882e-01 3.13522995e-01 4.21518028e-01 -1.29198325e+00
1.00090969e+00 1.04185514e-01 1.46384895e-01 -4.37712669e-01
-7.48536646e-01 1.13030411e-01 -2.40123942e-01 8.61507580e-02
8.47198844e-01 -5.78883231e-01 -9.36338186e-01 1.09846257e-01
-8.86700809e-01 -6.05357409e-01 -5.45389533e-01 3.87155712e-01
-8.85611475e-01 -2.98241526e-03 -4.76621628e-01 -8.04641545e-01
-1.46784604e-01 -1.23228478e+00 5.84115624e-01 6.21100843e-01
-4.39293355e-01 -1.24423373e+00 5.18358946e-02 2.48040065e-01
3.96969110e-01 3.74672115e-01 1.40001535e+00 -1.31878066e+00
-2.66412705e-01 -1.02857441e-01 3.79175767e-02 1.74849093e-01
2.59942245e-02 -3.83353233e-01 -1.00220716e+00 -5.78980558e-02
-3.47347230e-01 -9.21782851e-01 6.04712427e-01 1.31388605e-01
1.19814193e+00 -3.11026603e-01 -9.42239463e-02 4.73983079e-01
1.36347187e+00 5.61966479e-01 3.37331414e-01 7.69241750e-01
3.62641543e-01 3.72725993e-01 3.77236009e-01 3.74282807e-01
1.70198277e-01 4.91735846e-01 7.22843230e-01 3.44652534e-01
1.02549046e-01 -8.03162158e-01 4.87071991e-01 3.03812236e-01
-3.01958054e-01 -2.73778409e-01 -9.06520844e-01 3.08954626e-01
-2.16774583e+00 -9.29996312e-01 5.92300773e-01 1.72927511e+00
8.04463029e-01 4.58392024e-01 8.84027705e-02 4.86837998e-02
3.39821070e-01 -3.54980864e-02 -8.01390409e-01 -9.19646502e-01
1.37601523e-02 8.34561288e-01 2.55393356e-01 3.00354600e-01
-8.25447023e-01 1.35264957e+00 5.02833557e+00 9.23196793e-01
-1.01048470e+00 -1.28628746e-01 3.15357298e-01 2.23750263e-01
-2.38543421e-01 -1.02846414e-01 -5.35089612e-01 3.85644026e-02
1.06284034e+00 -2.23886311e-01 7.16443062e-01 1.13066828e+00
6.63618967e-02 1.78591266e-01 -1.32306600e+00 6.96517050e-01
-1.33768037e-01 -1.28947985e+00 4.22144920e-01 6.70286070e-04
4.02393132e-01 -1.92679182e-01 3.23951505e-02 1.16857767e+00
7.86697805e-01 -1.30065465e+00 8.50882471e-01 1.71109602e-01
3.40776056e-01 -7.10971773e-01 7.61034429e-01 4.72482681e-01
-7.39523828e-01 -5.44997811e-01 -4.02686596e-01 -5.72978616e-01
-3.97701085e-01 -2.38528505e-01 -8.50336790e-01 7.78936982e-01
4.37167257e-01 7.36740291e-01 -5.81731379e-01 5.97502768e-01
-6.74376845e-01 5.87097585e-01 -1.74355179e-01 -4.95127559e-01
5.42880058e-01 -1.43627658e-01 3.12167615e-01 7.59508729e-01
1.60679534e-01 2.37229452e-01 3.11420918e-01 1.00719643e+00
-5.71868382e-02 -2.16446295e-01 -9.40542042e-01 -3.20747286e-01
1.68567762e-01 9.85349119e-01 -6.02754951e-01 -1.32443547e-01
-4.52378392e-01 5.98786414e-01 6.31602824e-01 4.57906723e-01
-8.21922243e-01 -3.54031324e-01 6.84345782e-01 -1.73784107e-01
3.66987824e-01 -2.74928182e-01 -1.26467124e-01 -1.23354578e+00
-3.03941369e-01 -1.04084742e+00 4.24275309e-01 -6.81592703e-01
-1.05342150e+00 7.09466338e-01 1.93447813e-01 -8.13981831e-01
-8.68421912e-01 -1.30683315e+00 -4.79264706e-01 3.86636734e-01
-1.42589962e+00 -1.14358151e+00 -5.29629961e-02 5.17651498e-01
6.54300153e-01 -7.42972732e-01 1.19114566e+00 -2.38996074e-01
-6.66150510e-01 5.32903671e-01 -1.90236047e-01 4.04317945e-01
-9.10200775e-02 -1.43400919e+00 3.19319144e-02 5.64660609e-01
2.57000834e-01 6.23113394e-01 7.06857622e-01 -3.11907977e-01
-1.36927330e+00 -9.51403618e-01 1.88218057e-01 -2.41650566e-01
8.84646475e-01 -3.69044751e-01 -6.95486844e-01 8.50177050e-01
2.61178762e-01 -7.22242668e-02 5.07041991e-01 3.26841623e-01
-4.67254251e-01 -8.76041576e-02 -1.07503974e+00 9.02976990e-01
1.10349071e+00 -4.61130649e-01 -1.06741452e+00 -1.38093770e-01
6.14164948e-01 -4.00272280e-01 -6.23679340e-01 2.11583093e-01
3.21665794e-01 -8.19298148e-01 8.82629991e-01 -1.23871660e+00
6.13498688e-01 -1.62740946e-01 1.03321619e-01 -1.68018305e+00
-2.29269072e-01 -3.46689761e-01 -2.27748334e-01 9.00188863e-01
4.06784534e-01 -8.39678168e-01 9.52306926e-01 6.76861823e-01
-6.43632933e-02 -7.77065873e-01 -8.79150331e-01 -1.05548334e+00
2.76869088e-01 -5.32302499e-01 6.52873576e-01 8.36362779e-01
4.52199578e-01 4.32266504e-01 -1.26375422e-01 -1.73843443e-01
3.63502979e-01 -7.12424293e-02 6.00107849e-01 -1.55221415e+00
-4.81284410e-01 -4.98259634e-01 -7.14200795e-01 -7.03670323e-01
9.08775866e-01 -1.05295479e+00 -4.73447703e-02 -1.11949360e+00
-9.81961787e-02 -2.98645616e-01 -3.37968081e-01 1.01887178e+00
1.82129085e-01 -2.27530733e-01 9.06641707e-02 -3.36630136e-01
-5.76485872e-01 7.26296842e-01 1.00727618e+00 -3.71654242e-01
-1.10368185e-01 -3.13881516e-01 -6.52394712e-01 9.78293657e-01
1.07405615e+00 -6.30261481e-01 -8.89916837e-01 -5.97872138e-01
3.63481253e-01 -2.48126552e-01 4.66943562e-01 -1.21474349e+00
1.69799894e-01 -4.62849021e-01 1.36230290e-01 4.45563316e-01
3.24312449e-01 -9.14939523e-01 -3.09200257e-01 6.70588493e-01
-4.99975652e-01 1.79284140e-02 5.68629265e-01 8.03245127e-01
-3.23765486e-01 -7.27403641e-01 3.72264862e-01 -2.40076289e-01
-1.05093682e+00 1.33092508e-01 -6.03544295e-01 2.94646651e-01
1.03685963e+00 -4.68798339e-01 -3.11479658e-01 -4.60665137e-01
-8.05508196e-01 8.46427605e-02 1.51922733e-01 6.06909335e-01
6.63158596e-01 -1.03349769e+00 -2.78015316e-01 2.30296522e-01
2.27912366e-01 5.00137247e-02 -1.61242768e-01 1.62274659e-01
-5.42448282e-01 5.39924443e-01 -4.64227498e-01 -2.19822317e-01
-8.18339109e-01 5.29332221e-01 4.98393536e-01 -6.19884074e-01
-4.69509631e-01 4.62117106e-01 1.60862178e-01 -7.60505676e-01
1.87166303e-01 -7.62056172e-01 -2.90854990e-01 -3.70187730e-01
2.84995914e-01 1.64094400e-02 9.58091542e-02 -1.60798356e-01
-1.04213975e-01 1.01388209e-01 -7.45928884e-02 1.09496387e-02
1.52010727e+00 5.15397847e-01 3.46102208e-01 1.39248356e-01
5.73267400e-01 -4.52752322e-01 -9.96235907e-01 -3.25290084e-01
4.23232585e-01 -2.87534762e-02 -1.43998442e-02 -9.61962640e-01
-9.27272856e-01 8.51904452e-01 3.15162241e-01 5.21945894e-01
9.38652575e-01 -1.98050886e-01 3.94276977e-01 9.70022380e-01
6.46048188e-01 -1.02530849e+00 1.39155731e-01 9.81158793e-01
5.47649443e-01 -1.08876801e+00 -3.02738786e-01 -2.24472191e-02
-4.28510785e-01 1.55068004e+00 1.06438780e+00 -2.65230387e-01
1.29744545e-01 1.48447156e-01 -2.50051230e-01 -1.83459938e-01
-9.38507259e-01 -3.86058778e-01 -1.58099771e-01 1.00116420e+00
1.04333900e-01 3.44855934e-02 -4.01535891e-02 1.08167148e+00
-3.57296467e-01 3.07700902e-01 4.43674266e-01 1.02070236e+00
-3.73628020e-01 -1.41168547e+00 -4.67972048e-02 2.69991755e-01
-2.70241022e-01 -8.55533108e-02 -3.26579690e-01 1.20282400e+00
2.07662538e-01 8.97808313e-01 -1.20747134e-01 -4.38265622e-01
4.23762679e-01 3.26117605e-01 6.69666111e-01 -8.73953760e-01
-7.15407431e-01 -5.94234288e-01 2.76554674e-01 -5.75023592e-01
-1.84573844e-01 -3.33543032e-01 -1.61087406e+00 -2.27253869e-01
-6.87469915e-02 2.96055913e-01 3.73856544e-01 1.39244318e+00
3.09495121e-01 7.19367564e-01 6.05922006e-02 -5.24114966e-01
-8.78321230e-01 -7.50787735e-01 -3.44514340e-01 4.51755345e-01
1.57395862e-02 -9.01980340e-01 -2.21588667e-02 -2.40821987e-01]
|
[3.9194955825805664, 1.3945703506469727]
|
15b02f2b-4a31-49ab-8b93-1618e6805f93
|
global-and-local-feature-learning-for-ego
|
2002.06685
| null |
https://arxiv.org/abs/2002.06685v1
|
https://arxiv.org/pdf/2002.06685v1.pdf
|
Global and Local Feature Learning for Ego-Network Analysis
|
In an ego-network, an individual (ego) organizes its friends (alters) in different groups (social circles). This social network can be efficiently analyzed after learning representations of the ego and its alters in a low-dimensional, real vector space. These representations are then easily exploited via statistical models for tasks such as social circle detection and prediction. Recent advances in language modeling via deep learning have inspired new methods for learning network representations. These methods can capture the global structure of networks. In this paper, we evolve these techniques to also encode the local structure of neighborhoods. Therefore, our local representations capture network features that are hidden in the global representation of large networks. We show that the task of social circle prediction benefits from a combination of global and local features generated by our technique.
|
['Konstantin Ziegler', 'Fatemeh Salehi Rizi', 'Michael Granitzer']
|
2020-02-16
| null | null | null | null |
['learning-network-representations']
|
['methodology']
|
[-8.20883960e-02 5.03663659e-01 -4.60855842e-01 -5.01507878e-01
4.03289974e-01 -4.91833806e-01 1.00956547e+00 5.51966786e-01
9.41627622e-02 2.22975984e-01 7.94085622e-01 4.30865027e-02
-2.87957430e-01 -1.52500677e+00 -5.52218974e-01 -3.46480846e-01
-5.64501762e-01 5.34780085e-01 -1.50285259e-01 -4.45897505e-02
8.85364935e-02 7.66314924e-01 -1.33839667e+00 2.80836225e-01
5.43729782e-01 5.02641082e-01 -2.61346310e-01 7.13351965e-01
-2.79007733e-01 1.20502913e+00 -3.16145390e-01 -2.34926239e-01
2.00416315e-02 -4.47378635e-01 -8.77594173e-01 7.33288303e-02
1.28846973e-01 -1.76093072e-01 -1.08281231e+00 7.11309493e-01
1.19560286e-01 1.63786024e-01 9.74166989e-01 -1.16778898e+00
-6.89775288e-01 1.15066302e+00 -5.02849817e-01 1.23900466e-01
2.77686894e-01 -3.67525220e-01 1.35016429e+00 -4.84399766e-01
8.39113355e-01 1.65336251e+00 6.54935122e-01 1.03842586e-01
-1.38447618e+00 -4.19459015e-01 1.65245265e-01 1.22833073e-01
-1.37202430e+00 -4.29371297e-01 9.72688019e-01 -6.60590291e-01
8.21463943e-01 7.13482648e-02 9.00256753e-01 7.92589605e-01
4.41242605e-02 8.89465988e-01 5.19789577e-01 -1.81235626e-01
-1.83508322e-01 4.36831545e-03 2.34025776e-01 9.55976665e-01
2.27528140e-01 1.68841958e-01 -3.62155527e-01 -1.42540261e-01
1.09916890e+00 3.89298588e-01 1.55224800e-01 -6.78461850e-01
-1.14362383e+00 1.26072812e+00 1.10466242e+00 5.64839184e-01
-3.96392524e-01 4.73934144e-01 1.78564295e-01 5.16712666e-01
7.26196408e-01 3.75782937e-01 5.47894537e-02 1.32188916e-01
-7.60282576e-01 -3.48181635e-01 9.00601923e-01 6.21959507e-01
1.22474277e+00 -4.30582017e-02 3.49523500e-02 8.88403475e-01
3.69001031e-01 3.14547606e-02 4.14480537e-01 -6.73284650e-01
7.93417841e-02 9.74382520e-01 -3.42428148e-01 -1.66870737e+00
-8.04986298e-01 -5.85201025e-01 -9.85806823e-01 -1.93633810e-01
1.78877518e-01 -2.46800542e-01 -3.00839931e-01 1.65975678e+00
3.28754365e-01 4.34868783e-01 -1.71654269e-01 2.33746201e-01
9.46337998e-01 4.36129361e-01 -2.52011389e-01 3.18784654e-01
7.93717265e-01 -9.89201963e-01 -3.83110106e-01 -2.05156043e-01
1.06384039e+00 -2.26400986e-01 1.81983083e-01 -4.45285350e-01
-8.60691369e-01 -4.50731188e-01 -8.91750693e-01 -6.61486702e-04
-5.51716328e-01 3.75100486e-02 1.05332720e+00 4.29499120e-01
-1.29281437e+00 1.01463461e+00 -8.14051092e-01 -8.35116506e-01
8.17682505e-01 5.06286263e-01 -5.05912483e-01 2.10872125e-02
-9.53137398e-01 4.47333813e-01 7.51110166e-02 -1.85348347e-01
-5.96573651e-01 -2.38433987e-01 -1.19881320e+00 3.69648039e-01
-8.03965479e-02 -6.49278760e-01 5.24461091e-01 -1.03758168e+00
-1.07179022e+00 1.01899803e+00 -3.22049230e-01 -5.66658556e-01
1.38021976e-01 4.30689640e-02 -3.26622039e-01 2.95584679e-01
-8.37684497e-02 5.22318542e-01 8.93163502e-01 -9.89552557e-01
-2.36353695e-01 -4.49843407e-01 2.03043401e-01 -2.85512488e-02
-7.16837227e-01 1.66522697e-01 -1.37668744e-01 -6.43196225e-01
1.37029022e-01 -8.93895566e-01 -2.64067113e-01 6.97922185e-02
-6.85351610e-01 -5.17853737e-01 4.56605583e-01 -8.51342157e-02
1.26799941e+00 -2.04422140e+00 1.09017439e-01 6.70058906e-01
1.04937220e+00 7.47485012e-02 -5.31293392e-01 7.15635478e-01
-2.79354274e-01 5.15391827e-01 4.06222165e-01 -4.76260215e-01
-1.55720994e-01 1.69836581e-01 -8.72496963e-02 8.10349584e-01
1.48264512e-01 1.19378066e+00 -1.11509883e+00 -4.93601084e-01
2.51821637e-01 5.80185890e-01 -6.27108276e-01 2.17398807e-01
3.59233975e-01 1.29933357e-01 -6.91782951e-01 1.20025158e-01
3.31558645e-01 -7.18269169e-01 4.88019824e-01 8.59381482e-02
2.87510216e-01 4.03757602e-01 -8.42661440e-01 1.09292877e+00
-3.89082462e-01 1.10442150e+00 5.11268415e-02 -1.34557486e+00
1.16794980e+00 1.15543760e-01 6.94323540e-01 -1.50779083e-01
3.42984408e-01 -4.41696972e-01 6.24690577e-02 -8.40316862e-02
-1.40370307e-02 2.03251243e-01 2.57619888e-01 1.03516722e+00
7.64821395e-02 3.35200697e-01 6.45947009e-02 6.46665454e-01
1.24073660e+00 -5.03712356e-01 3.54270220e-01 -2.39901587e-01
2.66794056e-01 -5.09564757e-01 3.03461581e-01 8.05645645e-01
-6.43795803e-02 3.93057346e-01 9.56408203e-01 -5.65438449e-01
-9.72746849e-01 -9.76185322e-01 3.09085194e-02 1.33624482e+00
-1.20843783e-01 -7.30199099e-01 -5.53158283e-01 -6.11322880e-01
3.91420633e-01 7.66719356e-02 -1.05333269e+00 -4.00903851e-01
-5.78613162e-01 -3.82325709e-01 4.23177123e-01 5.12751102e-01
-2.08475992e-01 -9.91033792e-01 2.27745205e-01 2.01370835e-01
2.75479525e-01 -1.02036774e+00 -4.64253753e-01 -3.13939363e-01
-9.94811416e-01 -1.46655297e+00 -3.07660043e-01 -1.00288260e+00
8.00741017e-01 5.83949447e-01 1.18299353e+00 4.28254724e-01
-1.98512271e-01 7.00887263e-01 -1.62282988e-01 -6.11787885e-02
-4.28692788e-01 1.97764769e-01 2.97031462e-01 6.58780336e-01
4.66696113e-01 -1.03689849e+00 -4.22920078e-01 2.06364900e-01
-6.30843163e-01 -3.64781916e-02 3.92308474e-01 6.06279492e-01
1.08934246e-01 -5.20414189e-02 7.94638395e-01 -1.29518759e+00
7.05616951e-01 -9.51640189e-01 -2.51836896e-01 1.13254584e-01
-2.52781421e-01 1.70838594e-01 4.56270993e-01 -4.71038103e-01
-4.32704240e-01 -1.49384454e-01 3.81511539e-01 -2.90710717e-01
1.05252527e-01 7.08483815e-01 1.62260577e-01 -1.49103701e-01
6.28487468e-01 3.01145520e-02 2.56683469e-01 -4.49249446e-01
6.11531019e-01 5.96446216e-01 -6.21943809e-02 -2.10268989e-01
9.49999154e-01 6.16532028e-01 1.27298415e-01 -1.23024344e+00
-7.51767159e-01 -5.05365908e-01 -1.02590573e+00 -1.35878459e-01
4.52672571e-01 -1.08328760e+00 -8.23268950e-01 2.26816669e-01
-8.37561429e-01 -2.41461992e-01 -3.73169422e-01 4.16530907e-01
-4.84023184e-01 4.97110337e-01 -6.60344183e-01 -6.50944829e-01
-8.03900436e-02 -3.84059101e-01 6.12653196e-01 2.54630268e-01
-4.66571152e-01 -1.64004302e+00 2.27818355e-01 6.90637082e-02
4.55120683e-01 4.71099287e-01 8.50715876e-01 -8.77334118e-01
-4.94763345e-01 -4.26055044e-01 -5.64525366e-01 8.80734157e-03
3.07712585e-01 1.89053759e-01 -6.55634940e-01 -3.74012440e-01
-7.48697162e-01 -1.66344985e-01 1.17587948e+00 4.61029440e-01
1.13849080e+00 -5.55252671e-01 -8.66048396e-01 7.93840706e-01
9.74141181e-01 -6.24108732e-01 3.55510473e-01 -7.68251717e-02
1.02005577e+00 8.51541758e-01 -1.42475203e-01 5.74557126e-01
5.96520841e-01 3.02813917e-01 4.52902198e-01 -1.51547655e-01
-2.60037512e-01 -6.09265566e-01 3.13637823e-01 7.28991032e-01
-8.39504302e-02 -2.57424805e-02 -9.72126722e-01 8.95555913e-01
-1.93583786e+00 -1.29006672e+00 1.20093049e-02 1.77705407e+00
5.43110907e-01 -1.03698689e-02 3.88460100e-01 -1.99774206e-01
1.13368750e+00 6.41000926e-01 -5.63560545e-01 -3.99213344e-01
-4.91774827e-01 5.43789417e-02 2.11323172e-01 6.22754216e-01
-1.25399184e+00 1.04953849e+00 7.24613237e+00 4.15302485e-01
-8.43145013e-01 -2.75283724e-01 6.94959223e-01 -4.96612974e-02
-5.91740489e-01 3.42556499e-02 -5.32959819e-01 8.42097402e-02
1.04940355e+00 -5.25901377e-01 6.12115204e-01 8.06367874e-01
1.27785087e-01 5.55699885e-01 -1.39948142e+00 9.40277219e-01
3.42302397e-02 -1.82024074e+00 9.85856578e-02 4.92078662e-01
1.00702703e+00 6.22803092e-01 5.51900677e-02 1.42193094e-01
7.00527668e-01 -1.52417660e+00 -1.39517384e-02 7.97141612e-01
7.53037274e-01 -7.87218034e-01 4.32812542e-01 2.68660694e-01
-1.28734052e+00 -4.34572995e-01 -6.57568038e-01 -3.37367713e-01
-2.50545353e-01 6.41011536e-01 -1.18220723e+00 5.16135618e-02
2.03814715e-01 1.56547701e+00 -7.63098538e-01 8.83005559e-01
-1.31067693e-01 6.47633076e-01 -3.99980187e-01 -2.04923779e-01
2.71025926e-01 -5.26078641e-01 4.75928336e-01 1.05955732e+00
5.90272583e-02 -1.24494530e-01 1.22544892e-01 1.07696819e+00
-6.09042883e-01 2.22735226e-01 -1.11198246e+00 -7.33851373e-01
4.77310717e-01 1.28232229e+00 -5.85079014e-01 -1.25833392e-01
-5.33691585e-01 5.43444753e-01 9.74619925e-01 4.33626473e-01
-1.80224761e-01 -5.16081452e-01 1.11816263e+00 7.00468361e-01
2.57381499e-01 -3.45671058e-01 -3.71441152e-03 -1.22259545e+00
-5.82017779e-01 -2.64189690e-01 3.17772090e-01 -4.55189109e-01
-1.59480655e+00 5.01210809e-01 -5.74734569e-01 -8.19478333e-01
-4.19348508e-01 -3.43946576e-01 -1.26157117e+00 4.95158345e-01
-1.31400979e+00 -1.37397027e+00 -1.15914881e-01 8.20663452e-01
-5.82603365e-02 -4.30368572e-01 8.18800807e-01 -1.83771491e-01
-5.45306146e-01 6.04518414e-01 6.20070457e-01 9.65834200e-01
3.39266568e-01 -1.24109638e+00 5.03956079e-01 3.83836567e-01
5.87578714e-01 7.80431330e-01 1.42120987e-01 -5.33536255e-01
-8.68236244e-01 -1.34792268e+00 1.29630053e+00 -3.14605623e-01
1.23725104e+00 -4.93745536e-01 -6.29323721e-01 1.12832296e+00
-4.30227891e-02 2.99152732e-01 1.05645347e+00 8.33825052e-01
-5.77195823e-01 -7.46049732e-02 -9.40048873e-01 5.94726920e-01
1.28440619e+00 -1.05645061e+00 -2.70264596e-01 7.20244110e-01
6.31569147e-01 5.33459425e-01 -1.08066535e+00 -1.75162643e-01
4.83864516e-01 -9.33779061e-01 1.13225806e+00 -1.08059001e+00
4.27642971e-01 3.03047270e-01 3.18116963e-01 -1.43251944e+00
-5.98254859e-01 -7.68484533e-01 -5.20105839e-01 1.00003052e+00
1.38237372e-01 -7.42583871e-01 1.16036916e+00 2.25114256e-01
4.10287738e-01 -7.01718330e-01 -6.84806764e-01 -1.83632836e-01
1.48210153e-01 7.27576464e-02 8.33279610e-01 1.25550985e+00
3.48468810e-01 6.46386504e-01 -3.25839788e-01 -1.10365875e-01
6.47522092e-01 4.03515309e-01 7.65959978e-01 -1.95572567e+00
1.32116340e-02 -7.28372037e-01 -9.77083862e-01 -1.24636328e+00
5.22231817e-01 -1.28504717e+00 -8.43045235e-01 -1.69288075e+00
3.34402561e-01 -4.97927696e-01 -3.08286071e-01 2.62408048e-01
1.39183074e-01 1.94370359e-01 1.61979362e-01 2.39026099e-01
-7.38766611e-01 6.96731925e-01 1.12839222e+00 -2.82940209e-01
-2.22402290e-01 3.09007794e-01 -8.27812612e-01 8.92773688e-01
9.21394289e-01 -3.75609756e-01 -1.70128554e-01 -3.04848582e-01
5.74590743e-01 -7.14886114e-02 1.99225798e-01 -9.11912858e-01
3.68026912e-01 8.88539106e-02 5.31898856e-01 -4.50682253e-01
4.62744743e-01 -4.30503935e-01 -3.64539832e-01 3.14394355e-01
-6.40678346e-01 -3.00778270e-01 -5.17138660e-01 1.00601590e+00
-2.08016336e-01 1.08097546e-01 7.93279111e-01 -6.75510466e-02
-1.40476108e-01 6.12216771e-01 -4.72314507e-01 -1.65902555e-01
9.48827207e-01 -2.47220978e-01 -2.45857865e-01 -1.12031186e+00
-1.22048903e+00 3.87059391e-01 4.85117882e-01 5.11154354e-01
7.13636160e-01 -1.47912693e+00 -8.57748926e-01 2.77424455e-01
1.05640486e-01 -3.50485027e-01 8.56047347e-02 6.14903986e-01
-4.35014725e-01 3.11535805e-01 -2.38098338e-01 -4.79322702e-01
-1.18610466e+00 6.71720207e-01 3.74991745e-01 -3.45367819e-01
-5.72675347e-01 9.29941833e-01 4.22318280e-01 -7.90911853e-01
1.18843101e-01 3.84943783e-02 -9.76826012e-01 4.19072658e-01
5.37609458e-01 3.60299021e-01 -6.32952631e-01 -1.13669538e+00
-1.43379733e-01 5.21625340e-01 -8.58035162e-02 5.93598783e-01
1.81671798e+00 -2.98356295e-01 -5.16438901e-01 3.71286273e-01
1.66539681e+00 1.24656141e-01 -7.75299430e-01 -8.65822196e-01
1.81014031e-01 -4.41771299e-01 1.84236448e-02 6.95506558e-02
-1.21174204e+00 1.04865634e+00 1.36429090e-02 5.74099958e-01
4.55015272e-01 5.25556743e-01 5.60696661e-01 6.60015106e-01
3.53318192e-02 -9.59312975e-01 3.17597151e-01 6.66982472e-01
7.18714952e-01 -1.24616003e+00 5.48755489e-02 -4.74135965e-01
-3.48134995e-01 1.35020041e+00 1.09565593e-01 -6.76233947e-01
1.33521485e+00 -2.82310724e-01 -4.54648197e-01 -4.63125318e-01
-7.19891787e-01 -4.21377122e-01 3.09041232e-01 9.44899261e-01
4.15391266e-01 3.25674534e-01 3.66783112e-01 5.06288648e-01
-1.82694823e-01 -6.32636905e-01 7.83195257e-01 3.61123770e-01
-7.74248540e-01 -1.03597999e+00 4.98152226e-02 8.17739785e-01
-1.70743346e-01 -1.12813808e-01 -8.28177989e-01 4.63932484e-01
-2.87374020e-01 1.10558856e+00 6.17707729e-01 -5.46328247e-01
-2.82027692e-01 -2.68818945e-01 9.52476561e-02 -8.67491305e-01
-4.17329431e-01 -1.68055430e-01 8.94305110e-02 -6.49005890e-01
-4.82698441e-01 -6.96256578e-01 -1.05867028e+00 -8.83915186e-01
-8.28054026e-02 3.26058090e-01 2.91861743e-01 6.86054170e-01
6.25099063e-01 4.07451987e-01 1.14922345e+00 -7.23125637e-01
-2.20089033e-01 -9.66650069e-01 -9.69506025e-01 4.66594666e-01
4.18006092e-01 -5.37694573e-01 -4.35586035e-01 -3.90389740e-01]
|
[7.167958736419678, 6.186869144439697]
|
238e16dd-2545-4dcc-814a-f6fe87d8eacc
|
safe-reinforcement-learning-of-dynamic-high
|
2209.13308
| null |
https://arxiv.org/abs/2209.13308v2
|
https://arxiv.org/pdf/2209.13308v2.pdf
|
Safe Reinforcement Learning of Dynamic High-Dimensional Robotic Tasks: Navigation, Manipulation, Interaction
|
Safety is a crucial property of every robotic platform: any control policy should always comply with actuator limits and avoid collisions with the environment and humans. In reinforcement learning, safety is even more fundamental for exploring an environment without causing any damage. While there are many proposed solutions to the safe exploration problem, only a few of them can deal with the complexity of the real world. This paper introduces a new formulation of safe exploration for reinforcement learning of various robotic tasks. Our approach applies to a wide class of robotic platforms and enforces safety even under complex collision constraints learned from data by exploring the tangent space of the constraint manifold. Our proposed approach achieves state-of-the-art performance in simulated high-dimensional and dynamic tasks while avoiding collisions with the environment. We show safe real-world deployment of our learned controller on a TIAGo++ robot, achieving remarkable performance in manipulation and human-robot interaction tasks.
|
['Georgia Chalvatzaki', 'Jan Peters', 'Zhiyuan Hu', 'Snehal Jauhri', 'Davide Tateo', 'Kuo Zhang', 'Puze Liu']
|
2022-09-27
| null | null | null | null |
['safe-exploration']
|
['robots']
|
[-1.21281862e-01 4.32158262e-01 -4.67019230e-01 1.80882961e-01
-1.68724701e-01 -5.71264505e-01 3.65030229e-01 -9.18063521e-02
-7.28537619e-01 1.05576098e+00 -4.63340789e-01 -2.45383844e-01
-6.73138797e-01 -3.79904598e-01 -8.17107320e-01 -7.66447604e-01
-7.15169728e-01 5.69985926e-01 4.53964472e-01 -6.56419337e-01
2.21146047e-01 7.84641206e-01 -1.45271587e+00 -5.18293917e-01
6.37309968e-01 8.70513737e-01 5.02586365e-01 3.77340138e-01
7.19999731e-01 4.78250265e-01 -2.83502221e-01 4.70296830e-01
6.45127952e-01 2.32447594e-01 -6.64010346e-01 1.19902641e-02
-3.01764488e-01 -2.73360580e-01 -5.11818528e-01 1.05913746e+00
1.29585922e-01 4.69051659e-01 5.02545178e-01 -1.81447089e+00
5.79814389e-02 4.51068252e-01 -4.66360688e-01 -2.74569720e-01
1.01631217e-01 5.28529882e-01 3.63771886e-01 -1.68355852e-01
6.72901392e-01 1.16783524e+00 2.67119825e-01 6.20627940e-01
-9.21460509e-01 -5.76806903e-01 2.30109200e-01 6.49094358e-02
-1.22463357e+00 -5.85693568e-02 4.35337782e-01 -5.46415329e-01
9.55192089e-01 -1.66369706e-01 5.60200870e-01 1.24079883e+00
9.08704400e-01 3.40757102e-01 7.37460971e-01 -1.65268987e-01
3.77558440e-01 -1.02593698e-01 -1.86983973e-01 6.64772868e-01
4.39409852e-01 8.04171562e-01 -1.06404401e-01 -1.78634286e-01
6.71999335e-01 -1.68461040e-01 -1.91581115e-01 -1.23141038e+00
-1.55304778e+00 6.53335750e-01 3.87299925e-01 -1.05013311e-01
-2.30579376e-01 4.17973399e-01 4.82068032e-01 4.09729302e-01
-5.55115521e-01 8.32602680e-01 -5.73948801e-01 -2.18575612e-01
2.72736419e-02 7.52467394e-01 9.27312434e-01 1.48002088e+00
3.75669122e-01 1.77331567e-01 3.91001850e-01 2.67668068e-01
2.13747576e-01 3.67702097e-01 -1.09028881e-02 -1.37098205e+00
4.32680488e-01 2.19079629e-01 6.69169188e-01 -7.06857026e-01
-8.28220904e-01 -1.44483149e-01 -6.04732454e-01 1.21369052e+00
3.34734291e-01 -4.51019883e-01 -2.61143416e-01 1.69974983e+00
5.35209358e-01 -3.85747373e-01 2.37984225e-01 9.82612193e-01
-3.52315784e-01 3.71763378e-01 -5.40048890e-02 -1.80157930e-01
9.91237283e-01 -6.83025420e-01 -7.31571913e-01 -3.90265524e-01
4.33892637e-01 -2.57424116e-01 8.73622835e-01 9.56318378e-01
-8.26399863e-01 -3.52347404e-01 -1.47735190e+00 2.90838867e-01
-1.25384882e-01 -1.88029677e-01 4.02666658e-01 3.17925103e-02
-4.86936301e-01 7.75908232e-01 -1.11225843e+00 -3.86167914e-01
1.49319193e-03 6.50940895e-01 -5.58603048e-01 4.12220448e-01
-8.86580586e-01 1.62436068e+00 7.33111024e-01 2.50799935e-02
-1.49688220e+00 -3.91960949e-01 -8.44318151e-01 -3.69507015e-01
1.05037868e+00 -2.79526651e-01 1.43020928e+00 3.27523910e-02
-1.69427371e+00 2.90149331e-01 7.53351212e-01 -5.07711709e-01
7.98176169e-01 -9.34191048e-01 -4.02365960e-02 -9.13731083e-02
-9.28768814e-02 5.74428082e-01 8.38581443e-01 -1.13840508e+00
-5.60153484e-01 -1.91249758e-01 1.54661238e-01 2.71860301e-01
-2.12315992e-01 -4.81729686e-01 1.12709902e-01 -3.60289097e-01
-1.87180042e-01 -1.65608382e+00 -6.37586057e-01 4.74409252e-01
-2.65881360e-01 -3.43403995e-01 1.19246364e+00 3.37281786e-02
5.64767122e-01 -2.18195462e+00 5.85423887e-01 1.20454602e-01
-1.92434296e-01 8.58076289e-02 -6.05102256e-02 5.67685425e-01
1.75300300e-01 -2.68874705e-01 -1.41917720e-01 2.24020734e-01
2.67434746e-01 4.95664746e-01 -7.58030415e-01 9.10817742e-01
-1.05189504e-02 4.47180361e-01 -1.13240433e+00 -3.01351368e-01
2.86818683e-01 2.36925274e-01 -5.53847790e-01 3.71648550e-01
-5.02509177e-01 7.61950791e-01 -8.42961550e-01 2.33826354e-01
2.44485706e-01 5.46295822e-01 2.12979913e-01 4.21202064e-01
-4.07562137e-01 -1.15977950e-01 -1.26403344e+00 1.77130818e+00
-2.42600307e-01 3.12995821e-01 6.30572200e-01 -9.13886428e-01
1.02749026e+00 1.92626193e-01 7.54542589e-01 -4.54582512e-01
4.90469038e-01 3.01579475e-01 1.70512378e-01 -6.04217410e-01
5.46832561e-01 -5.67837059e-02 -5.79102933e-01 2.08068699e-01
-2.52128951e-02 -7.79909909e-01 -5.36821485e-02 -2.28400305e-01
1.12487030e+00 6.37198746e-01 4.68776733e-01 -7.88001239e-01
3.54894757e-01 4.19701785e-01 6.09044313e-01 6.28404319e-01
-5.61763287e-01 -2.04118252e-01 5.45146167e-01 -3.12904656e-01
-9.87597287e-01 -1.22069526e+00 -2.93406006e-02 8.34703565e-01
7.13711739e-01 -2.85639446e-02 -6.78949773e-01 -4.90551382e-01
3.55556607e-01 7.48917460e-01 -4.61205840e-01 -5.99755585e-01
-8.62511039e-01 1.17208928e-01 2.85823226e-01 2.93784171e-01
2.36164182e-01 -1.11788881e+00 -1.75953829e+00 1.81775361e-01
2.30353743e-01 -1.31316555e+00 -2.10846469e-01 4.67265993e-01
-7.15210199e-01 -1.54088533e+00 6.37466535e-02 -8.46721113e-01
3.73329133e-01 1.13240920e-01 4.48976964e-01 -1.55115277e-01
-5.92490137e-01 3.81026298e-01 -1.40457943e-01 -5.74716032e-01
-5.14398515e-01 -1.13181226e-01 8.41241360e-01 -6.95541203e-01
-2.61711299e-01 -3.90376240e-01 -2.11479947e-01 6.79151833e-01
-5.74986339e-01 -3.26047748e-01 3.09393167e-01 6.19006634e-01
6.20471418e-01 4.90589827e-01 3.58222455e-01 2.34630611e-02
6.49918020e-01 -3.27522457e-01 -1.11998904e+00 -1.55919969e-01
-1.94860518e-01 2.44757146e-01 6.47764027e-01 -8.55306447e-01
-8.12077045e-01 5.19690335e-01 3.52770299e-01 -4.10383701e-01
-7.20582083e-02 1.23753317e-01 -1.35283172e-01 -1.87259942e-01
7.12046862e-01 -2.57084370e-01 4.98052031e-01 -2.93439865e-01
3.16584498e-01 3.30233395e-01 7.77671754e-01 -9.64230359e-01
9.72609878e-01 4.76717263e-01 7.14529335e-01 -6.95113897e-01
-3.26251805e-01 -2.39256863e-02 -8.52783561e-01 -2.60027885e-01
7.51119196e-01 -5.89156866e-01 -1.41837668e+00 1.36376500e-01
-9.19847786e-01 -7.41143048e-01 -3.40707183e-01 5.84128916e-01
-1.35875702e+00 1.92127913e-01 -1.77273378e-01 -1.01840031e+00
1.08715463e-02 -1.38830817e+00 8.40897679e-01 8.86264145e-02
-2.92598605e-01 -3.85199070e-01 1.51751429e-01 -4.51524258e-01
3.06164175e-01 7.73422241e-01 7.70841181e-01 -2.11695939e-01
-4.66751844e-01 -1.35024279e-01 5.31236768e-01 -2.81501804e-02
1.16545729e-01 -2.95154482e-01 -3.14786732e-01 -8.13114047e-01
3.39095473e-01 -6.38296783e-01 1.86496735e-01 5.97205646e-02
1.02438903e+00 -3.18338513e-01 -6.85987949e-01 1.82762355e-01
1.22387516e+00 3.95446956e-01 3.55856508e-01 6.84590876e-01
3.27251554e-01 9.93599534e-01 1.51319838e+00 6.20137393e-01
-2.88834833e-02 9.21158850e-01 1.25081646e+00 6.15312278e-01
5.29080212e-01 -7.45710731e-02 5.59784174e-01 -7.31526548e-03
2.89899886e-01 9.33431983e-02 -8.83891523e-01 5.33842802e-01
-2.21541572e+00 -5.05329192e-01 1.16651051e-01 2.29735589e+00
6.23357177e-01 4.72144246e-01 1.22237340e-01 5.76632060e-02
4.66123730e-01 -3.58462751e-01 -1.01220381e+00 -4.25081372e-01
3.22870135e-01 -3.27250123e-01 7.25503445e-01 5.69294989e-01
-1.27922153e+00 8.26921403e-01 6.19541025e+00 3.68860602e-01
-1.08641541e+00 -2.67155647e-01 -3.48630339e-01 -4.09052044e-01
6.03263855e-01 -1.55411735e-01 -6.66577756e-01 1.44912153e-01
6.06058955e-01 -3.30740988e-01 4.93509233e-01 1.46310091e+00
1.18291207e-01 -4.19807255e-01 -1.39518785e+00 6.52358353e-01
-4.73753035e-01 -7.15642750e-01 -6.32664204e-01 8.61359090e-02
5.17380595e-01 1.49765179e-01 9.91634429e-02 3.66915703e-01
5.72982192e-01 -9.79353666e-01 1.13290203e+00 3.05901110e-01
5.61117589e-01 -9.91449356e-01 2.35575303e-01 9.78478909e-01
-9.13418591e-01 -6.53043687e-01 -2.98577100e-01 -2.80918211e-01
4.54927057e-01 2.81742066e-02 -6.93578601e-01 2.41078392e-01
7.22984731e-01 4.71474946e-01 -6.75921515e-02 1.03307414e+00
-2.48657674e-01 -2.94608921e-01 -4.62102115e-01 -2.79645503e-01
4.26147372e-01 -1.93725228e-01 9.47287440e-01 6.61029518e-01
2.62591720e-01 5.77200390e-03 5.98345339e-01 6.35864735e-01
5.95648348e-01 -5.61709702e-01 -1.22363663e+00 1.96846843e-01
2.55329430e-01 1.07894254e+00 -3.99554372e-01 2.04390466e-01
1.98713616e-01 4.60280120e-01 4.29469526e-01 8.45942125e-02
-1.01824057e+00 -6.29574001e-01 1.12384701e+00 -4.99576852e-02
7.61536211e-02 -9.51357305e-01 -1.02342561e-01 -5.15939295e-01
2.24050298e-01 -7.68681943e-01 1.75647810e-01 -5.11940956e-01
-8.05938780e-01 3.84543031e-01 4.43338037e-01 -1.70634913e+00
-3.81014049e-01 -1.00171530e+00 -2.37003028e-01 3.58456790e-01
-1.38886976e+00 -4.50369000e-01 -2.28325725e-01 6.03098869e-01
4.81436372e-01 -2.77602464e-01 8.10355484e-01 -2.78366238e-01
-2.24535465e-01 -1.51548004e-02 -5.59900664e-02 -3.86751622e-01
7.66018212e-01 -8.64741743e-01 4.41377647e-02 4.78141755e-01
-6.96588159e-01 5.98340452e-01 1.10651040e+00 -7.30795801e-01
-1.92741692e+00 -1.10448408e+00 1.29897017e-02 -4.37545031e-01
9.42075133e-01 -4.50820923e-01 -7.95473814e-01 7.28361547e-01
1.29457280e-01 -3.02024395e-03 -2.18645364e-01 -4.11321193e-01
-6.95573688e-02 5.90374172e-02 -1.25186026e+00 9.53468144e-01
1.16893184e+00 3.53443958e-02 -8.07236612e-01 4.26575154e-01
9.68168795e-01 -7.43764997e-01 -8.05613160e-01 8.03599417e-01
6.18914008e-01 -3.52583915e-01 7.85197139e-01 -7.27939129e-01
5.04505709e-02 -6.15933120e-01 -1.90300252e-02 -1.37831545e+00
-1.98814973e-01 -1.19849753e+00 -3.56182933e-01 3.68582249e-01
6.62877783e-03 -4.92960513e-01 6.47492468e-01 3.40955764e-01
-4.74159658e-01 -6.75880671e-01 -1.31853056e+00 -1.44615388e+00
3.36508751e-01 -2.54113019e-01 3.23229492e-01 5.92566967e-01
7.78765857e-01 -1.88801989e-01 -6.08769536e-01 5.02035141e-01
9.24417019e-01 -2.70830959e-01 9.56685185e-01 -1.03516161e+00
-1.59087405e-01 -3.53534311e-01 -2.42247075e-01 -6.10000849e-01
5.92271388e-01 -5.45818925e-01 8.09262395e-01 -1.16008854e+00
-2.87058711e-01 -7.00526297e-01 1.28951594e-02 4.92990166e-01
5.04661679e-01 -2.08239585e-01 2.01004744e-01 9.16428193e-02
-6.35441124e-01 7.66609251e-01 1.40753448e+00 -1.65886492e-01
-3.39947581e-01 -7.45492950e-02 -1.38500512e-01 6.86664462e-01
1.13709903e+00 -4.05982792e-01 -6.63312912e-01 -4.94025946e-01
9.16870981e-02 2.42820114e-01 3.10957640e-01 -1.39945579e+00
2.86192894e-01 -8.29669833e-01 -2.71876156e-01 -3.76495391e-01
5.22538424e-01 -1.41365266e+00 3.38999003e-01 1.25772917e+00
-2.62977272e-01 1.02984086e-01 4.32470053e-01 8.11109602e-01
2.25519374e-01 -6.79877549e-02 1.06415486e+00 1.14408910e-01
-7.91363418e-01 1.60877779e-01 -4.94840294e-01 -8.84003267e-02
1.77071476e+00 1.17430769e-01 -4.62107599e-01 9.21144560e-02
-6.72184467e-01 8.91230643e-01 6.29373074e-01 1.00350869e+00
5.12484729e-01 -1.03640652e+00 -1.81732342e-01 1.16681673e-01
1.87045932e-01 6.44866154e-02 -1.18109420e-01 5.59258163e-01
-3.67039829e-01 5.04327297e-01 -7.97712028e-01 -6.52884960e-01
-9.69035983e-01 9.80061114e-01 3.24481428e-01 1.08618662e-02
-9.92512643e-01 2.78336018e-01 1.90992832e-01 -5.83372533e-01
6.03818178e-01 -4.01673794e-01 5.70790330e-03 -6.45189166e-01
2.61227340e-01 3.95374477e-01 -2.68733084e-01 -4.71614927e-01
-4.60217148e-01 5.57534039e-01 1.67356879e-01 -1.60676301e-01
1.27610517e+00 5.71612269e-02 9.85715464e-02 2.29309499e-01
6.82641804e-01 -4.81922328e-01 -1.90486395e+00 1.39503449e-01
2.95268983e-01 -2.82160014e-01 -3.16883355e-01 -5.29334128e-01
-5.58450639e-01 6.90151870e-01 4.93729591e-01 -1.44831344e-01
6.20618522e-01 -1.58244371e-01 4.24874783e-01 9.58459556e-01
1.30699062e+00 -1.48007500e+00 5.18758893e-01 8.96457732e-01
1.41001070e+00 -1.04961324e+00 5.91929108e-02 -3.53748649e-01
-8.09716046e-01 9.57752764e-01 1.07894850e+00 -6.11193895e-01
7.96781123e-01 5.96757829e-01 -2.59228766e-01 8.77146199e-02
-6.65155888e-01 1.97169900e-01 -1.38449579e-01 1.00636935e+00
-4.21249896e-01 4.90933843e-02 -2.47375712e-01 3.58789057e-01
-1.73625320e-01 -1.88931733e-01 7.61353374e-01 1.51230073e+00
-9.61818814e-01 -9.01694834e-01 -4.40712988e-01 -1.72437519e-01
6.63354760e-03 8.46289515e-01 3.72982919e-02 1.29640186e+00
-5.57162948e-02 8.90591443e-01 -6.32451549e-02 -3.25949252e-01
7.25703478e-01 -2.08337218e-01 7.30799735e-01 -6.72285795e-01
-2.56901205e-01 -1.05772123e-01 3.35054249e-02 -1.16494167e+00
-7.35971751e-03 -7.21753657e-01 -1.64179134e+00 5.57744950e-02
-1.31660968e-01 1.08859763e-01 7.97333181e-01 7.15602517e-01
2.46126577e-01 5.46715677e-01 4.97634679e-01 -1.21323967e+00
-1.33637512e+00 -6.59876645e-01 -7.14100122e-01 6.07550852e-02
8.39706421e-01 -1.34789181e+00 -7.91766793e-02 -2.79453874e-01]
|
[4.699005126953125, 1.5864927768707275]
|
3a778069-d27e-4bab-ac11-b71f74dbd7b3
|
probabilistic-radiomics-ambiguous-diagnosis
|
1910.08878
| null |
https://arxiv.org/abs/1910.08878v1
|
https://arxiv.org/pdf/1910.08878v1.pdf
|
Probabilistic Radiomics: Ambiguous Diagnosis with Controllable Shape Analysis
|
Radiomics analysis has achieved great success in recent years. However, conventional Radiomics analysis suffers from insufficiently expressive hand-crafted features. Recently, emerging deep learning techniques, e.g., convolutional neural networks (CNNs), dominate recent research in Computer-Aided Diagnosis (CADx). Unfortunately, as black-box predictors, we argue that CNNs are "diagnosing" voxels (or pixels), rather than lesions; in other words, visual saliency from a trained CNN is not necessarily concentrated on the lesions. On the other hand, classification in clinical applications suffers from inherent ambiguities: radiologists may produce diverse diagnosis on challenging cases. To this end, we propose a controllable and explainable {\em Probabilistic Radiomics} framework, by combining the Radiomics analysis and probabilistic deep learning. In our framework, 3D CNN feature is extracted upon lesion region only, then encoded into lesion representation, by a controllable Non-local Shape Analysis Module (NSAM) based on self-attention. Inspired from variational auto-encoders (VAEs), an Ambiguity PriorNet is used to approximate the ambiguity distribution over human experts. The final diagnosis is obtained by combining the ambiguity prior sample and lesion representation, and the whole network named $DenseSharp^{+}$ is end-to-end trainable. We apply the proposed method on lung nodule diagnosis on LIDC-IDRI database to validate its effectiveness.
|
['Rongyao Fang', 'Jiancheng Yang', 'Bingbing Ni', 'Yi Xu', 'Linguo Li', 'Yamin Li']
|
2019-10-20
| null | null | null | null |
['probabilistic-deep-learning']
|
['computer-vision']
|
[ 2.73307174e-01 5.50393581e-01 -3.66036087e-01 -2.53433794e-01
-1.12108636e+00 -8.91942754e-02 2.64481068e-01 -1.49989530e-01
-2.65779272e-02 6.51269257e-01 3.27515513e-01 -2.17889979e-01
-3.08829010e-01 -5.88321447e-01 -7.30666697e-01 -9.34658349e-01
5.06733596e-01 5.14033735e-01 1.51467070e-01 2.44655952e-01
-1.89527705e-01 3.91894728e-01 -1.33288443e+00 4.28254843e-01
9.62059379e-01 1.26761842e+00 6.06092155e-01 2.69412011e-01
-1.95994809e-01 1.10829937e+00 -1.78206012e-01 -9.44078043e-02
-2.18146235e-01 -5.28501272e-01 -6.93495393e-01 2.71959484e-01
-3.48462649e-02 -2.83686429e-01 -4.26765352e-01 1.18103015e+00
3.40938240e-01 -3.10312539e-01 1.17004323e+00 -9.23741341e-01
-9.97633755e-01 5.20689368e-01 -7.36408651e-01 2.33866736e-01
-1.32313445e-01 3.21209252e-01 8.19393575e-01 -9.00561690e-01
4.31522489e-01 1.14674509e+00 6.04296744e-01 4.82820690e-01
-7.77773857e-01 -3.81307691e-01 1.23437457e-02 1.93984285e-01
-1.44855583e+00 -8.73308405e-02 1.02650094e+00 -4.50802028e-01
3.70802701e-01 3.12728435e-01 5.60349643e-01 1.32503426e+00
6.44722939e-01 1.09103203e+00 9.49634552e-01 -4.85308878e-02
1.30691245e-01 2.70312399e-01 -2.97133267e-01 9.84199882e-01
1.06595516e-01 1.78699642e-01 5.72656244e-02 -3.94273102e-02
1.03852403e+00 3.90591890e-01 -3.73521864e-01 -4.47952121e-01
-1.22050273e+00 7.78236926e-01 1.01439142e+00 2.52912730e-01
-6.45398796e-01 3.14491987e-01 2.03251660e-01 -3.93947005e-01
3.25359643e-01 5.06687053e-02 -2.53237784e-01 3.65183800e-01
-8.49005640e-01 3.65759619e-02 3.04849178e-01 6.25149190e-01
4.52810436e-01 1.22534268e-01 -4.69467878e-01 8.01353812e-01
5.58697939e-01 6.80107296e-01 8.01421165e-01 -5.10392785e-01
3.65729891e-02 5.98519385e-01 -1.49565339e-01 -9.72387612e-01
-5.13067722e-01 -1.00448906e+00 -1.41172898e+00 1.30885497e-01
4.28837426e-02 7.63810799e-02 -1.28865921e+00 1.58483660e+00
3.43692720e-01 2.31876031e-01 -5.02841137e-02 1.20068622e+00
8.71683955e-01 4.44550186e-01 2.80290008e-01 -3.05903494e-01
1.65019178e+00 -9.00575995e-01 -6.48560405e-01 -6.63819686e-02
4.72459763e-01 -5.87710917e-01 9.62789536e-01 2.24316835e-01
-1.02609563e+00 -3.81227106e-01 -8.39561641e-01 -1.08816251e-01
2.06179302e-02 3.93112302e-01 5.06322026e-01 3.33605617e-01
-8.03271353e-01 2.95329750e-01 -9.69940722e-01 -2.37574633e-02
9.10715342e-01 2.59274900e-01 -2.39669811e-02 -1.89510137e-01
-1.04268909e+00 9.57414150e-01 2.17881799e-01 2.47789711e-01
-1.24403536e+00 -7.32108295e-01 -8.65596592e-01 1.12375937e-01
5.22165716e-01 -9.42364335e-01 1.31855094e+00 -1.14638484e+00
-1.42007804e+00 7.76526749e-01 -6.37652702e-04 -4.22781050e-01
5.25317132e-01 2.52757430e-01 -2.93634921e-01 1.32451355e-01
2.15461075e-01 7.04200864e-01 1.04848635e+00 -1.18935072e+00
-5.52000523e-01 -2.79228985e-01 -2.65212864e-01 2.33095571e-01
-2.75584832e-02 -3.69224131e-01 -2.90728629e-01 -8.69218588e-01
3.05378973e-01 -8.21076870e-01 -4.02361512e-01 2.56413847e-01
-7.24202275e-01 -2.16832370e-01 7.36297965e-01 -6.37410522e-01
9.61732805e-01 -2.08936906e+00 3.05993676e-01 1.64606035e-01
4.88857865e-01 1.14137763e-02 2.65511125e-01 -5.29055953e-01
-2.11750418e-01 5.30375801e-02 -7.02796102e-01 -5.00027612e-02
-1.56903520e-01 1.22099608e-01 -8.35496038e-02 6.33282602e-01
4.83692616e-01 1.18474674e+00 -9.27090585e-01 -1.08744061e+00
6.42534137e-01 5.51612675e-01 -4.44929361e-01 8.09926987e-02
-4.30931658e-01 6.29928887e-01 -9.10642862e-01 1.03188193e+00
6.45823658e-01 -7.10509121e-01 -1.30981132e-01 -3.84294510e-01
1.30990535e-01 -3.82519841e-01 -7.94058681e-01 1.78434074e+00
-8.22630703e-01 3.00580144e-01 1.79612692e-02 -1.00272202e+00
6.80127501e-01 4.92393702e-01 6.29987776e-01 -5.46023428e-01
3.21692228e-01 4.81042981e-01 5.65453619e-02 -7.25182831e-01
-1.96500853e-01 -1.65185183e-01 1.35919347e-01 1.32082596e-01
-1.90009642e-02 -1.98408052e-01 -5.25891125e-01 -7.34856799e-02
1.09535372e+00 -3.01483013e-02 5.13082981e-01 -1.28415510e-01
6.98357522e-01 1.07277632e-01 4.36791033e-01 3.35587144e-01
-2.17036217e-01 1.04139721e+00 4.97160405e-01 -2.60682076e-01
-9.88870740e-01 -1.31865454e+00 -4.92193431e-01 4.64876711e-01
2.29739547e-01 2.48535722e-01 -5.96147060e-01 -9.40007448e-01
-1.65174782e-01 5.28992414e-01 -8.26079190e-01 -3.45227599e-01
-4.35234666e-01 -6.64907873e-01 2.34543055e-01 6.83353961e-01
4.21279341e-01 -1.44773519e+00 -4.76609200e-01 3.56576055e-01
-9.81757790e-02 -8.23625863e-01 -3.21754575e-01 1.29430369e-01
-7.83634663e-01 -1.03387058e+00 -1.26346731e+00 -8.40137899e-01
7.30804741e-01 1.56248167e-01 1.07796669e+00 -1.89310983e-02
-7.19253421e-01 1.95811570e-01 -2.14687943e-01 -3.74946356e-01
-3.18608731e-01 8.19672793e-02 -3.00268888e-01 1.28595546e-01
1.64451540e-01 -5.75204432e-01 -9.73744631e-01 2.20753700e-01
-9.26068544e-01 2.68804580e-01 1.45598507e+00 1.10949826e+00
1.12593257e+00 2.87966896e-02 6.38149083e-01 -7.95667529e-01
1.37836009e-01 -8.25388253e-01 -3.28016847e-01 3.48207176e-01
-2.41525918e-01 1.97264716e-01 6.52871549e-01 -2.37297207e-01
-1.07704318e+00 3.51601750e-01 -3.19332331e-01 -9.83922124e-01
-2.49132946e-01 5.97284973e-01 -1.81291878e-01 1.72094822e-01
7.01037347e-01 4.36477363e-01 1.68873891e-01 -1.92212865e-01
3.42793584e-01 6.82704151e-01 6.91412866e-01 -3.19470108e-01
6.80258691e-01 5.90964139e-01 1.22634627e-01 -4.84017730e-01
-1.11579859e+00 -8.28244910e-02 -1.59347698e-01 -2.28265107e-01
1.24136674e+00 -8.58837545e-01 -6.56231999e-01 1.74042642e-01
-1.10730529e+00 8.74297991e-02 -3.79037708e-01 7.11825013e-01
-6.22010291e-01 -2.34341659e-02 -3.36226285e-01 -6.70614600e-01
-4.29360121e-01 -1.61316717e+00 1.36639524e+00 3.85428697e-01
9.25562009e-02 -7.38857865e-01 -2.58947343e-01 2.36091614e-01
3.47733229e-01 4.32562172e-01 9.61231470e-01 -3.62615198e-01
-8.50803018e-01 -1.76875815e-01 -6.63731396e-01 2.81128347e-01
3.25784445e-01 -2.36808419e-01 -1.05928266e+00 1.67753194e-02
1.95693627e-01 -3.51912826e-01 9.80998039e-01 1.04545200e+00
1.87770522e+00 -9.89166424e-02 -6.44288123e-01 6.26772761e-01
1.33663380e+00 -6.45204037e-02 4.18392003e-01 -1.17710426e-01
9.89970922e-01 3.89407516e-01 3.35433692e-01 4.00187343e-01
2.76828259e-01 3.46031547e-01 9.13530111e-01 -1.86190516e-01
-4.72303689e-01 -2.40304962e-01 -6.63393363e-02 6.09949291e-01
1.68735862e-01 -2.44866431e-01 -8.83111238e-01 6.55354738e-01
-1.69387698e+00 -5.67566037e-01 3.84555236e-02 1.69233906e+00
7.11739182e-01 4.73321276e-03 -4.37249064e-01 -9.28690284e-02
8.85356665e-01 1.71517894e-01 -8.00485134e-01 2.46201083e-01
-6.97104959e-03 1.94548205e-01 3.79199862e-01 2.87629277e-01
-1.17763364e+00 6.40117109e-01 4.41289806e+00 1.37603962e+00
-1.30863476e+00 3.52816641e-01 1.05751622e+00 2.08412148e-02
-5.33434272e-01 -2.77623415e-01 -3.69758964e-01 5.96760273e-01
3.60618681e-01 1.30483583e-01 3.34152766e-02 1.14817679e+00
1.72762319e-01 9.22476724e-02 -9.42734718e-01 1.14740217e+00
2.64399499e-02 -1.55141497e+00 3.96288708e-02 6.08147979e-02
6.43622100e-01 -1.55721635e-01 6.16701722e-01 3.06713521e-01
1.31900370e-01 -1.43118656e+00 5.62558651e-01 8.32756341e-01
1.11629188e+00 -7.23135948e-01 9.94976640e-01 2.45000288e-01
-1.06933570e+00 5.15705207e-04 -4.54498500e-01 6.29782438e-01
1.21661142e-01 7.75267899e-01 -9.73192930e-01 6.75765812e-01
5.37605107e-01 7.99850583e-01 -4.64045197e-01 9.40102279e-01
-1.11341670e-01 5.35821617e-01 -1.91862464e-01 -2.97758374e-02
3.44074219e-01 1.99279934e-01 5.00991940e-01 8.43649685e-01
4.62617785e-01 2.52464086e-01 8.88877884e-02 1.26135683e+00
-4.23257761e-02 -2.27198321e-02 -4.08748657e-01 1.53014287e-01
1.97886124e-01 1.48222053e+00 -7.98299253e-01 -2.49614894e-01
-1.41543284e-01 8.04639459e-01 1.89940453e-01 2.08112270e-01
-1.11622095e+00 -3.06250583e-02 5.07698730e-02 9.30639729e-02
2.06870675e-01 4.89029557e-01 -4.39401478e-01 -1.09456038e+00
-1.30697461e-02 -7.05254614e-01 2.18188748e-01 -9.77370262e-01
-1.58031940e+00 7.22921133e-01 -3.13588560e-01 -1.51766777e+00
-1.01869725e-01 -6.51860952e-01 -6.04291856e-01 8.76263976e-01
-1.43622363e+00 -1.35298884e+00 -5.18657267e-01 5.24427235e-01
6.84678972e-01 -1.14011288e-01 6.46493375e-01 1.82221919e-01
-5.28239012e-01 5.74356198e-01 -1.83831360e-02 1.35155469e-02
4.13726747e-01 -1.27481258e+00 -2.79850364e-01 3.40236127e-01
-9.22800452e-02 2.85065025e-02 3.61095309e-01 -6.04055822e-01
-1.12900138e+00 -1.55816257e+00 3.34246874e-01 -3.72241408e-01
4.99138027e-01 3.09881475e-02 -9.45775151e-01 4.77769822e-01
5.00226915e-02 5.53969622e-01 1.46203101e-01 -4.06606138e-01
9.52267274e-02 -1.57783814e-02 -1.26285982e+00 5.60467064e-01
8.39157820e-01 -3.94432038e-01 -4.59450245e-01 6.15467668e-01
1.00748432e+00 -6.07048988e-01 -7.80691564e-01 7.07154453e-01
2.63844192e-01 -7.58472860e-01 1.07766056e+00 -3.90397161e-01
7.30677247e-01 -3.17777842e-01 -1.80968449e-01 -1.10158503e+00
-4.13407475e-01 2.54461542e-02 -1.45964980e-01 7.63521850e-01
3.89022976e-01 -3.08438003e-01 1.00412548e+00 1.59056082e-01
-6.78652406e-01 -1.37589586e+00 -9.86873925e-01 -2.97895879e-01
9.14000124e-02 -4.90591496e-01 5.12612402e-01 7.81659365e-01
-5.04449964e-01 1.19369328e-01 -2.77956307e-01 4.39131349e-01
6.55362070e-01 -5.44519629e-03 3.21000209e-03 -1.05801797e+00
-4.92804915e-01 -6.39083266e-01 -3.45356792e-01 -8.89168024e-01
1.82069719e-01 -1.02867901e+00 1.14846013e-01 -1.58183599e+00
3.57212782e-01 -6.15267754e-01 -5.82875133e-01 2.78080016e-01
-2.40525067e-01 1.84435129e-01 -2.32403189e-01 2.16134623e-01
-4.42703873e-01 8.20214689e-01 1.74216962e+00 -4.45404351e-01
2.84281485e-02 1.20495364e-01 -6.48846090e-01 9.62732017e-01
5.09291232e-01 -4.51067239e-01 -4.87610787e-01 -2.20950171e-01
-4.97292839e-02 5.13698578e-01 8.64855587e-01 -1.01841712e+00
2.10649043e-01 -1.49700373e-01 7.67923474e-01 -9.28852320e-01
3.29046011e-01 -8.80488157e-01 3.07126524e-04 5.22188246e-01
-2.40856633e-01 -4.28617865e-01 -1.86351538e-01 7.91531622e-01
-3.40821624e-01 -5.74463829e-02 9.09482181e-01 -3.57679546e-01
-3.53166282e-01 8.11823726e-01 -1.69683456e-01 2.22811010e-02
9.65914786e-01 -1.41176898e-02 -2.31083948e-02 2.31205225e-02
-9.01291430e-01 6.49541467e-02 1.25163123e-01 1.99787900e-01
7.65779078e-01 -1.51569700e+00 -7.21026063e-01 -1.59903895e-02
1.71569973e-01 6.74903333e-01 8.82009506e-01 1.16430116e+00
-5.75273275e-01 4.74043280e-01 1.18683405e-01 -9.79732275e-01
-6.11380279e-01 7.29470313e-01 6.42592669e-01 -4.08093601e-01
-6.21177137e-01 9.34048414e-01 7.45197177e-01 -2.35286951e-01
1.86818987e-01 -5.58828175e-01 -2.61710644e-01 -2.39457652e-01
1.96453616e-01 -1.72188833e-01 -8.17975253e-02 -4.89220589e-01
-4.12339985e-01 6.00580096e-01 -2.27486804e-01 -2.78722290e-02
1.03820574e+00 9.35297310e-02 9.17045549e-02 2.54215389e-01
1.29335451e+00 -3.82284820e-01 -1.17671907e+00 -4.13014174e-01
-3.43866348e-01 -1.04144722e-01 3.35762322e-01 -7.77500570e-01
-1.36023510e+00 9.26578760e-01 9.37066555e-01 -9.84600633e-02
1.02896821e+00 3.51099074e-01 7.79989004e-01 9.73971412e-02
5.41776605e-02 -6.06154919e-01 1.92106545e-01 -2.98053529e-02
9.83946741e-01 -1.52254069e+00 -8.54488984e-02 -3.63404542e-01
-8.98010552e-01 9.35754776e-01 7.32450128e-01 -3.80525023e-01
9.96361375e-01 -7.63907060e-02 -1.23348109e-01 -2.99042672e-01
-5.32123208e-01 -1.53577715e-01 3.28782767e-01 4.94873345e-01
2.65973657e-01 3.88103396e-01 4.17203642e-02 1.03902304e+00
-3.19134407e-02 1.11862600e-01 2.43411034e-01 6.22733533e-01
-4.80834752e-01 -5.08512437e-01 -4.89357263e-01 7.79183149e-01
-4.24918711e-01 -1.61330551e-01 1.03578232e-01 7.85726249e-01
4.36608136e-01 3.71244371e-01 1.07432522e-01 -1.70464486e-01
-4.64684283e-03 -2.85603374e-01 3.02023500e-01 -6.62474215e-01
-1.26555026e-01 2.51188010e-01 -4.55002755e-01 -3.93999130e-01
-7.20366538e-02 -5.99520385e-01 -1.33925211e+00 1.62556246e-01
-2.89765328e-01 -5.61172515e-02 4.12400901e-01 9.70374882e-01
1.44857720e-01 1.16281974e+00 7.51553893e-01 -5.96230328e-01
-6.09167218e-01 -9.54514682e-01 -6.31153941e-01 3.49780843e-02
4.50034201e-01 -9.48297203e-01 -1.93951607e-01 -1.20539740e-01]
|
[14.892059326171875, -2.247610569000244]
|
80900f28-dd2c-4f01-88f7-a9027e58a855
|
multi-scale-self-contrastive-learning-with
|
2203.03838
| null |
https://arxiv.org/abs/2203.03838v1
|
https://arxiv.org/pdf/2203.03838v1.pdf
|
Multi-Scale Self-Contrastive Learning with Hard Negative Mining for Weakly-Supervised Query-based Video Grounding
|
Query-based video grounding is an important yet challenging task in video understanding, which aims to localize the target segment in an untrimmed video according to a sentence query. Most previous works achieve significant progress by addressing this task in a fully-supervised manner with segment-level labels, which require high labeling cost. Although some recent efforts develop weakly-supervised methods that only need the video-level knowledge, they generally match multiple pre-defined segment proposals with query and select the best one, which lacks fine-grained frame-level details for distinguishing frames with high repeatability and similarity within the entire video. To alleviate the above limitations, we propose a self-contrastive learning framework to address the query-based video grounding task under a weakly-supervised setting. Firstly, instead of utilizing redundant segment proposals, we propose a new grounding scheme that learns frame-wise matching scores referring to the query semantic to predict the possible foreground frames by only using the video-level annotations. Secondly, since some predicted frames (i.e., boundary frames) are relatively coarse and exhibit similar appearance to their adjacent frames, we propose a coarse-to-fine contrastive learning paradigm to learn more discriminative frame-wise representations for distinguishing the false positive frames. In particular, we iteratively explore multi-scale hard negative samples that are close to positive samples in the representation space for distinguishing fine-grained frame-wise details, thus enforcing more accurate segment grounding. Extensive experiments on two challenging benchmarks demonstrate the superiority of our proposed method compared with the state-of-the-art methods.
|
['Wei Hu', 'Daizong Liu', 'Shentong Mo']
|
2022-03-08
| null | null | null | null |
['video-grounding']
|
['computer-vision']
|
[ 4.22073424e-01 -7.93412700e-02 -6.94104493e-01 -4.12395954e-01
-1.05188584e+00 -4.57202405e-01 2.61953175e-01 1.43041490e-02
-2.75461167e-01 5.58328390e-01 -1.49714937e-02 1.17369860e-01
1.99244544e-01 -7.39926398e-01 -1.11084116e+00 -6.89349711e-01
1.50385737e-01 1.98201835e-01 8.53497267e-01 9.50826611e-03
2.00279325e-01 1.33329928e-01 -1.68140531e+00 7.30901659e-01
7.88823724e-01 1.12183905e+00 4.88064796e-01 2.98338085e-01
-1.07648291e-01 1.01652277e+00 -4.31314796e-01 -2.00440004e-01
2.00529233e-01 -6.82743728e-01 -9.81932044e-01 6.79931164e-01
7.88841009e-01 -3.31387460e-01 -3.60175341e-01 1.36545253e+00
-2.51265801e-02 1.98360354e-01 2.55312830e-01 -1.29733169e+00
-4.47035581e-01 2.26379707e-01 -6.92284048e-01 4.91957515e-01
7.08800733e-01 1.04082517e-01 1.16524887e+00 -9.50231254e-01
6.57476008e-01 1.21195734e+00 3.07506293e-01 4.25130576e-01
-1.09078515e+00 -7.12394118e-01 7.06183851e-01 5.59255481e-01
-1.71823001e+00 -3.94972980e-01 1.08060277e+00 -4.52233434e-01
4.59407121e-01 3.29027325e-01 6.15286291e-01 8.96018267e-01
-1.86391100e-01 1.09853315e+00 9.22480166e-01 -1.59024701e-01
1.67260513e-01 -1.78802103e-01 -5.44953793e-02 1.10589802e+00
-7.55688595e-03 -1.64561883e-01 -5.98277807e-01 -2.26947460e-02
9.11996365e-01 2.68390685e-01 -6.17139161e-01 -5.14288247e-01
-1.48606634e+00 6.59713387e-01 3.55731696e-01 4.40639347e-01
-3.72390687e-01 -1.29036736e-02 3.52940589e-01 5.62690943e-02
6.17659628e-01 4.22968417e-02 -5.70669830e-01 1.81051210e-01
-1.23423016e+00 7.27952197e-02 4.49476808e-01 9.39990580e-01
1.35023546e+00 -1.56004161e-01 -4.66743499e-01 6.36428475e-01
2.59074926e-01 2.26759255e-01 3.87540728e-01 -9.61162865e-01
5.61402559e-01 6.96398199e-01 1.72863811e-01 -1.32367158e+00
-3.04540470e-02 -4.08022016e-01 -7.04888880e-01 -1.74404830e-01
4.68875796e-01 2.40529597e-01 -9.12269533e-01 1.64973283e+00
5.04918396e-01 9.33602154e-01 -7.34198764e-02 1.21766281e+00
8.34918439e-01 7.27809012e-01 1.46846280e-01 -4.44178194e-01
1.29806507e+00 -1.29510140e+00 -4.47503120e-01 -2.63753146e-01
5.32563865e-01 -6.71799123e-01 1.21318328e+00 1.82608217e-01
-9.39442396e-01 -7.76709676e-01 -8.22083950e-01 6.75724894e-02
1.96745526e-03 2.02847704e-01 2.84324229e-01 3.11807811e-01
-8.29740822e-01 3.22337329e-01 -7.82422066e-01 -1.53691277e-01
4.82392699e-01 4.00835760e-02 -3.81490886e-01 -3.76225948e-01
-1.07595170e+00 2.73795873e-01 7.09015250e-01 1.02867700e-01
-1.11591756e+00 -5.29761255e-01 -1.05582404e+00 -3.92842740e-02
8.70592058e-01 -5.35841405e-01 8.45169127e-01 -1.45868134e+00
-1.25249326e+00 1.04789305e+00 -5.51911712e-01 -2.38426358e-01
5.61809599e-01 -1.23577468e-01 -3.09945077e-01 6.12831354e-01
4.74400431e-01 7.66015530e-01 1.03257704e+00 -1.32846439e+00
-1.04922378e+00 -8.33684877e-02 5.49631178e-01 1.97252318e-01
-1.60033822e-01 -6.29722700e-02 -1.06673539e+00 -8.92107069e-01
4.34061468e-01 -7.61251152e-01 -9.76989865e-02 1.40214428e-01
-3.03010195e-01 -4.33897823e-01 1.03271234e+00 -5.38834989e-01
1.16547382e+00 -2.12733030e+00 1.23818256e-01 -7.15219602e-02
1.94776416e-01 1.67182013e-01 -2.06353575e-01 -1.04625955e-01
7.88453668e-02 -1.62976179e-02 -1.69916809e-01 -1.14327706e-01
-3.88928443e-01 3.28864008e-01 -1.51460052e-01 5.29475570e-01
4.84496057e-01 8.66106391e-01 -1.32690096e+00 -9.54492629e-01
3.43435019e-01 1.61842108e-01 -5.53522348e-01 4.35750753e-01
-4.52411890e-01 6.08053863e-01 -7.14112878e-01 8.60885561e-01
5.30412436e-01 -4.96926337e-01 -6.93252608e-02 -6.55345559e-01
1.46081045e-01 -1.00140236e-01 -1.21466672e+00 1.85881460e+00
-1.89964280e-01 4.35076773e-01 3.69335599e-02 -1.51722181e+00
7.48993814e-01 1.71775907e-01 5.81143677e-01 -5.73332429e-01
-1.85615987e-01 1.52260885e-01 -3.32832754e-01 -7.41020560e-01
3.10474128e-01 -7.86452517e-02 3.67411482e-03 1.02779865e-01
6.86719045e-02 2.23735034e-01 2.84590691e-01 1.12192363e-01
8.63925934e-01 4.97194022e-01 2.70262092e-01 -1.57004610e-01
1.01962018e+00 -7.12207481e-02 1.04978883e+00 6.27723634e-01
-5.18383861e-01 9.18264091e-01 3.20811987e-01 -4.63021010e-01
-6.46203756e-01 -7.74940014e-01 1.39208987e-01 1.40141928e+00
9.09042954e-01 -4.51137245e-01 -8.05625618e-01 -1.03810930e+00
-3.67589474e-01 1.37670904e-01 -5.43435693e-01 2.23115925e-02
-7.15815604e-01 -3.44069958e-01 2.01684982e-01 4.26056385e-01
6.20663643e-01 -9.25422013e-01 -4.30855572e-01 1.59067422e-01
-6.95456266e-01 -1.43692040e+00 -7.35815942e-01 -1.89866066e-01
-6.34397984e-01 -1.31860662e+00 -9.18621302e-01 -1.18540502e+00
7.98063278e-01 7.04594433e-01 1.18497074e+00 4.11997408e-01
5.36853820e-02 2.77396291e-01 -5.70340216e-01 3.32272410e-01
-1.49550050e-01 -1.85216904e-01 -1.94392353e-01 5.02728760e-01
2.83262491e-01 1.41794924e-02 -8.28848541e-01 7.08613932e-01
-1.02319169e+00 2.91659236e-01 5.61039865e-01 8.72393847e-01
1.19124746e+00 1.98428780e-01 5.04636109e-01 -8.18431139e-01
-9.38581675e-02 -3.77108246e-01 -3.78238231e-01 5.71256936e-01
-4.33676057e-02 -1.27874389e-01 6.45942688e-01 -4.17507440e-01
-8.52230370e-01 2.28292525e-01 8.61177295e-02 -8.65741909e-01
-3.66437793e-01 1.92557186e-01 -3.34067136e-01 -1.37561277e-01
2.27293730e-01 4.80388105e-01 -2.90586889e-01 -1.76462263e-01
2.93313593e-01 3.63929212e-01 6.49365366e-01 -7.71256924e-01
7.92749166e-01 6.13867879e-01 -2.93379575e-01 -6.29997849e-01
-1.25885856e+00 -8.08302343e-01 -6.87713623e-01 -3.20611507e-01
1.17031884e+00 -1.18922925e+00 -3.16380084e-01 2.07693309e-01
-9.95653510e-01 -2.43898898e-01 -1.46007702e-01 4.35490519e-01
-7.18367577e-01 7.26611197e-01 -5.17319679e-01 -5.28579712e-01
-2.36601047e-02 -1.36807764e+00 1.58558977e+00 2.21784964e-01
-3.97913679e-02 -7.16932118e-01 -2.99479425e-01 5.69829404e-01
-1.66070998e-01 2.80338109e-01 6.04016602e-01 -4.76149619e-01
-9.55109954e-01 7.03397766e-02 -4.33319509e-01 2.94968933e-01
1.86155677e-01 -1.00082316e-01 -7.76556015e-01 -3.26354176e-01
-2.89397519e-02 -3.13893408e-01 9.53954041e-01 2.50147313e-01
1.45956469e+00 -2.49181688e-01 -3.71291250e-01 6.63629055e-01
1.22298634e+00 1.31113585e-02 3.92266542e-01 2.51538217e-01
1.08577216e+00 5.70339501e-01 1.24986327e+00 2.68637568e-01
4.27311778e-01 8.17527533e-01 4.57910478e-01 -3.05461466e-01
-1.19647175e-01 -3.77701223e-01 4.10427660e-01 5.42108119e-01
2.17985064e-02 -1.06596954e-01 -4.46981579e-01 5.47057867e-01
-2.09391809e+00 -1.16574657e+00 3.45041789e-02 2.10793471e+00
7.92851865e-01 1.90030411e-01 2.54771799e-01 5.07853739e-02
1.14727592e+00 5.57005107e-01 -6.59147203e-01 4.48586345e-01
-1.92743883e-01 -1.93352133e-01 2.65792310e-01 2.88518935e-01
-1.48650992e+00 1.13066649e+00 4.97643185e+00 1.16630924e+00
-1.18623757e+00 1.22456640e-01 1.00363779e+00 2.11109310e-01
-3.22818190e-01 4.56156582e-02 -6.82051122e-01 7.01415420e-01
3.48674953e-01 9.88056362e-02 2.33440340e-01 8.43538940e-01
3.96732777e-01 -3.17338705e-02 -1.16018462e+00 1.15738034e+00
3.59501332e-01 -1.45414174e+00 1.10320494e-01 -4.03273731e-01
9.62963700e-01 -1.45215347e-01 -3.31140697e-01 1.84946179e-01
-3.11285049e-01 -6.62589729e-01 1.07272613e+00 3.88734132e-01
7.54564464e-01 -5.36071301e-01 6.42532051e-01 2.90724069e-01
-1.69449818e+00 5.13122939e-02 -4.48998332e-01 7.12496862e-02
2.08070278e-01 4.64442611e-01 -3.30075413e-01 5.98302066e-01
8.94980013e-01 1.24148345e+00 -5.76366007e-01 8.54344308e-01
-2.10141703e-01 6.85153186e-01 -7.46984705e-02 3.67722720e-01
3.05200666e-01 -1.79658845e-01 3.06245148e-01 1.16641343e+00
1.87899709e-01 2.03138724e-01 8.87011588e-01 7.38295317e-01
-6.02915362e-02 1.69340193e-01 -2.26906329e-01 2.44195104e-01
3.62588942e-01 1.16716003e+00 -1.12184274e+00 -6.09628916e-01
-7.37780035e-01 1.22121680e+00 1.63586691e-01 4.97879207e-01
-9.21576917e-01 -5.60911335e-02 5.30618548e-01 2.48616710e-01
5.42247236e-01 -5.12788305e-04 2.14673936e-01 -1.56996346e+00
3.03536594e-01 -8.89629483e-01 6.07563376e-01 -6.85820639e-01
-1.30325079e+00 5.27942419e-01 -1.24000415e-01 -1.66978991e+00
-1.74212590e-01 -2.87327290e-01 -4.97697085e-01 3.94574642e-01
-1.76555192e+00 -1.20792353e+00 -5.83392560e-01 9.22545612e-01
1.08506823e+00 1.71619684e-01 3.99436712e-01 4.01981235e-01
-5.49703062e-01 5.31547308e-01 -2.15126440e-01 2.85895258e-01
7.35496759e-01 -9.37747777e-01 -3.55943702e-02 1.07454479e+00
3.80455911e-01 2.83501357e-01 4.19338316e-01 -5.63882589e-01
-1.20092452e+00 -1.48417330e+00 6.03943765e-01 -2.16650531e-01
5.13915658e-01 -2.54688799e-01 -1.19075370e+00 4.65661496e-01
-1.31915227e-01 6.04619443e-01 4.79239911e-01 -3.36687803e-01
-3.08856487e-01 -1.91956431e-01 -9.32359874e-01 2.61067539e-01
1.20130885e+00 -8.12390387e-01 -5.68744242e-01 4.19557393e-01
7.89754331e-01 -3.98181319e-01 -6.38244808e-01 6.09457493e-01
3.53101760e-01 -1.05036855e+00 1.08066440e+00 -3.79511297e-01
3.46114337e-01 -9.06435430e-01 -2.33448729e-01 -7.15124965e-01
-2.18980387e-01 -5.07029533e-01 -2.21076176e-01 1.36302376e+00
-1.25839323e-01 -1.70994297e-01 1.10160053e+00 2.49626860e-01
-1.37726143e-01 -8.48066688e-01 -8.26551080e-01 -7.41511166e-01
-4.31327939e-01 -3.69343787e-01 3.71434838e-01 1.01338780e+00
-3.44240099e-01 1.94442600e-01 -5.54788113e-01 3.17700386e-01
5.47399521e-01 6.12334251e-01 7.39309371e-01 -1.02913713e+00
-3.56177062e-01 -1.85142890e-01 -7.98548520e-01 -1.49543941e+00
5.27584076e-01 -7.20755696e-01 4.27923560e-01 -1.31631708e+00
5.20093739e-01 -3.26386005e-01 -5.65822423e-01 3.61328304e-01
-6.59061015e-01 6.34582400e-01 1.93562642e-01 4.67255592e-01
-1.30658329e+00 3.88648719e-01 1.22361219e+00 -3.81384283e-01
-4.24292833e-02 -1.33783907e-01 -4.17053550e-01 8.97390723e-01
4.01100665e-01 -3.54335964e-01 -5.20286798e-01 -4.03703123e-01
-2.88853675e-01 2.28171363e-01 5.84271371e-01 -9.39105868e-01
1.01331577e-01 -4.27214324e-01 3.74454409e-01 -6.17319107e-01
1.16438314e-01 -7.23256946e-01 -3.23203690e-02 3.49036872e-01
-3.85412008e-01 -1.52455673e-01 -2.54789114e-01 1.00750232e+00
-6.28168404e-01 -8.87057036e-02 8.07401359e-01 -1.36177018e-01
-1.36465538e+00 6.64267123e-01 -5.54397367e-02 3.05602372e-01
1.29258895e+00 -3.78527135e-01 -1.66420713e-01 -3.17889929e-01
-6.25991404e-01 1.83545753e-01 7.00315058e-01 5.38922548e-01
7.45217562e-01 -1.16251481e+00 -6.27544522e-01 1.84163496e-01
3.80040109e-01 1.26900658e-01 4.35439914e-01 7.11082220e-01
-3.95923316e-01 1.99775219e-01 -3.27807143e-02 -1.04269516e+00
-1.23798704e+00 7.69633532e-01 1.98899642e-01 -1.34817781e-02
-6.32743716e-01 9.44983721e-01 8.75155449e-01 1.77005038e-01
1.56361446e-01 -4.59103137e-01 -1.41600251e-01 4.15906729e-03
5.89303732e-01 9.14780870e-02 -1.60686120e-01 -1.20765638e+00
-5.35349429e-01 8.95015538e-01 5.17703556e-02 4.10420090e-01
9.15032446e-01 -4.24585909e-01 6.05756603e-02 3.53109896e-01
1.47096837e+00 -1.94813311e-01 -1.68231487e+00 -5.13303638e-01
7.33507518e-03 -8.23717654e-01 -1.17533982e-01 -1.97617754e-01
-1.35864842e+00 7.24798083e-01 5.60815454e-01 1.54086240e-02
1.31189382e+00 3.52808595e-01 8.99951816e-01 1.47308201e-01
5.15112638e-01 -8.77891243e-01 4.08960789e-01 1.56171143e-01
4.39163595e-01 -1.60500753e+00 -4.89397570e-02 -7.16493130e-01
-5.43774843e-01 9.91646588e-01 8.00119460e-01 -1.68411851e-01
2.93510020e-01 -2.47869030e-01 -5.40503813e-03 -1.26755070e-02
-4.84153539e-01 -5.29160619e-01 5.03811955e-01 5.13192892e-01
3.74396443e-01 -1.25708863e-01 -2.84971416e-01 3.67479295e-01
4.56847191e-01 2.32040315e-04 2.16958597e-01 6.93287909e-01
-5.83813488e-01 -9.60254550e-01 -3.59865248e-01 3.87439758e-01
-5.24319828e-01 5.47082722e-02 -1.07784666e-01 5.52493572e-01
3.27758938e-01 1.11764634e+00 3.73528711e-02 -3.45630258e-01
1.20294563e-01 -2.37241477e-01 2.97191381e-01 -6.73737049e-01
-1.36081338e-01 4.16136026e-01 -1.80922672e-01 -8.96225691e-01
-9.57877636e-01 -6.00507319e-01 -1.40905261e+00 2.57631898e-01
-3.78154635e-01 2.28038520e-01 -1.36774540e-01 1.15572906e+00
3.09714317e-01 4.20688540e-01 6.74305797e-01 -1.09483349e+00
-1.42039862e-02 -4.75263029e-01 -4.43945706e-01 8.90801013e-01
4.14797515e-01 -8.18755984e-01 -2.41244376e-01 4.64774102e-01]
|
[9.29771900177002, 0.29803866147994995]
|
83c45802-a6ef-48be-bacd-2d7e949c22b5
|
mnemosyne-learning-to-train-transformers-with
|
2302.01128
| null |
https://arxiv.org/abs/2302.01128v3
|
https://arxiv.org/pdf/2302.01128v3.pdf
|
Mnemosyne: Learning to Train Transformers with Transformers
|
In this work, we propose a new class of learnable optimizers, called \textit{Mnemosyne}. It is based on the novel spatio-temporal low-rank implicit attention Transformers that can learn to train entire neural network architectures, including other Transformers, without any task-specific optimizer tuning. We show that Mnemosyne: (a) outperforms popular LSTM optimizers (also with new feature engineering to mitigate catastrophic forgetting of LSTMs), (b) can successfully train Transformers while using simple meta-training strategies that require minimal computational resources, (c) matches accuracy-wise SOTA hand-designed optimizers with carefully tuned hyper-parameters (often producing top performing models). Furthermore, Mnemosyne provides space complexity comparable to that of its hand-designed first-order counterparts, which allows it to scale to training larger sets of parameters. We conduct an extensive empirical evaluation of Mnemosyne on: (a) fine-tuning a wide range of Vision Transformers (ViTs) from medium-size architectures to massive ViT-Hs (36 layers, 16 heads), (b) pre-training BERT models and (c) soft prompt-tuning large 11B+ T5XXL models. We complement our results with a comprehensive theoretical analysis of the compact associative memory used by Mnemosyne which we believe was never done before.
|
['Avinava Dubey', 'Jie Tan', 'Tingnan Zhang', 'Vikas Sindhwani', 'Sumeet Singh', 'Krzysztof Marcin Choromanski', 'Deepali Jain']
|
2023-02-02
| null | null | null | null |
['feature-engineering']
|
['methodology']
|
[-1.25441132e-02 1.72829881e-01 2.38811374e-01 -1.58189908e-01
-7.19424725e-01 -3.50593656e-01 4.70097780e-01 -2.41369054e-01
-7.33059227e-01 5.99444091e-01 2.27828711e-01 -3.77670586e-01
-4.19562221e-01 -4.63735372e-01 -9.70766306e-01 -7.09682703e-01
5.34795551e-03 9.92961347e-01 5.23514807e-01 -3.77061099e-01
1.81933954e-01 4.36892033e-01 -1.30012238e+00 3.21924925e-01
8.34826291e-01 1.03141725e+00 6.29571021e-01 9.79686499e-01
1.81186378e-01 1.19468510e+00 -4.27392602e-01 -6.98878646e-01
2.02798918e-02 5.24638779e-02 -1.05971026e+00 -4.16625053e-01
7.67998993e-01 -2.76297659e-01 -4.93664742e-01 6.41213298e-01
5.26200593e-01 3.67431074e-01 4.78549927e-01 -7.56321669e-01
-1.21239078e+00 7.74940014e-01 -7.56581798e-02 6.46339953e-01
-5.52170813e-01 5.69512010e-01 1.25515568e+00 -1.18580925e+00
2.55541742e-01 1.13031650e+00 1.00792384e+00 6.75330400e-01
-1.27425814e+00 -4.07768697e-01 4.08683002e-01 4.46526289e-01
-1.27098274e+00 -7.58467078e-01 2.03090549e-01 5.56684397e-02
1.90820527e+00 2.38436803e-01 7.71827340e-01 1.20289266e+00
3.52122426e-01 8.16808879e-01 9.27427471e-01 -1.93910316e-01
1.87911436e-01 -8.95183627e-03 2.77265906e-01 1.02992058e+00
-3.04433741e-02 6.87806681e-02 -7.57168531e-01 1.51613057e-01
6.56972408e-01 -3.61790173e-02 -1.35301456e-01 -1.48112610e-01
-1.27084553e+00 6.78349793e-01 8.33394229e-01 3.80046755e-01
-3.66030902e-01 8.31952572e-01 2.48027086e-01 4.05737281e-01
3.29810083e-01 9.60929930e-01 -8.43479991e-01 -5.53705953e-02
-9.88771796e-01 -1.16298340e-01 6.85120165e-01 8.09153259e-01
6.93747997e-01 5.70713341e-01 -3.55954647e-01 8.05549622e-01
1.41339302e-01 5.23418009e-01 8.04786503e-01 -1.04146898e+00
3.83092433e-01 1.40194714e-01 -2.98577063e-02 -5.25026560e-01
-5.02628803e-01 -6.97575629e-01 -6.23297453e-01 -2.41955332e-02
1.21527679e-01 1.92565635e-01 -1.29902542e+00 1.73882794e+00
-2.95052141e-01 1.26645699e-01 6.49539977e-02 7.88977861e-01
7.59125054e-01 7.50789464e-01 9.71354544e-02 1.25105336e-01
1.17199874e+00 -1.55065680e+00 -4.08022165e-01 -4.99327898e-01
5.28321266e-01 -4.25445259e-01 1.63993728e+00 5.65319419e-01
-1.36713016e+00 -3.81103754e-01 -1.08372390e+00 -5.14235973e-01
-4.56489652e-01 1.40477449e-01 8.89365077e-01 3.80240589e-01
-1.98648989e+00 8.52736354e-01 -1.23340619e+00 -3.77258241e-01
4.87874538e-01 9.44338441e-01 -1.77770928e-01 2.53978848e-01
-1.05672383e+00 1.35837877e+00 4.99068290e-01 2.35517040e-01
-1.52446628e+00 -7.61334121e-01 -5.93473256e-01 4.01275605e-01
3.40426058e-01 -1.24733078e+00 1.43637431e+00 -9.58818316e-01
-1.96433342e+00 6.70857251e-01 -2.78728992e-01 -9.74392176e-01
-4.98966239e-02 -5.99096537e-01 -3.07757650e-02 8.35164189e-02
-2.97953933e-01 8.87778640e-01 9.33780849e-01 -8.78265500e-01
-2.75451750e-01 -2.15433300e-01 1.66718975e-01 4.28649902e-01
-8.28113317e-01 -1.59379646e-01 -6.62813306e-01 -7.55568445e-01
-8.53300747e-03 -8.40740740e-01 -2.82897681e-01 -2.07544416e-01
-5.31797767e-01 -5.97584918e-02 5.76622307e-01 -5.03323913e-01
1.00711119e+00 -1.71862328e+00 5.58368385e-01 -7.43292645e-02
6.52926266e-01 6.44475758e-01 -4.21666026e-01 1.90749347e-01
-2.98526417e-03 9.55281779e-02 -1.83066100e-01 -9.50110435e-01
8.66079144e-03 5.90316951e-01 -5.85949659e-01 2.05953568e-01
1.06711961e-01 1.44670439e+00 -6.88529968e-01 -4.08135504e-01
2.51253635e-01 4.80596811e-01 -6.65406346e-01 1.24069070e-02
-3.74823749e-01 -7.19017759e-02 -2.91364878e-01 5.00010967e-01
-3.50011066e-02 -6.42385483e-01 -1.16541296e-01 -4.47973579e-01
-7.09949881e-02 6.14432633e-01 -4.67959732e-01 1.65177286e+00
-7.43010044e-01 8.31677318e-01 -1.30502701e-01 -1.02811229e+00
5.16620517e-01 2.34012887e-01 1.41151339e-01 -8.37124765e-01
3.47841233e-02 2.17611864e-01 -2.08796680e-01 -2.83510894e-01
4.64518189e-01 -2.48123538e-02 2.67734975e-01 3.44651341e-01
7.85214067e-01 5.88382669e-02 -2.31169779e-02 3.24410826e-01
1.26949108e+00 -8.37371945e-02 -1.28626764e-01 -4.48964506e-01
1.54438198e-01 -2.26146474e-01 1.69537574e-01 1.22186470e+00
1.61693227e-02 5.65295458e-01 9.44462493e-02 -6.71034813e-01
-1.17156720e+00 -1.23002040e+00 6.35884404e-02 1.53497946e+00
-1.76642939e-01 -4.52706158e-01 -7.11470842e-01 -3.23145092e-01
-2.55200446e-01 7.83032298e-01 -8.48481417e-01 -4.35736656e-01
-6.08405530e-01 -8.68726611e-01 9.03396368e-01 6.36294365e-01
6.02058351e-01 -1.25542736e+00 -5.41411877e-01 1.54557511e-01
6.82679191e-02 -8.85098517e-01 -6.36451662e-01 7.46684492e-01
-1.24245405e+00 -7.35126436e-01 -5.95880568e-01 -7.34541595e-01
4.13597524e-01 -3.49200368e-02 1.48117828e+00 1.42941341e-01
-2.16985662e-02 4.55839455e-01 1.39869396e-02 -2.14578226e-01
5.77910207e-02 7.39657223e-01 2.62526810e-01 -2.15520471e-01
-2.92187855e-02 -7.77820289e-01 -4.55118030e-01 1.44789219e-01
-5.81526518e-01 1.68231219e-01 1.00075209e+00 1.08431196e+00
7.14512169e-01 -3.10122430e-01 2.31615752e-01 -9.80097055e-01
4.18301255e-01 -8.62875208e-02 -6.19077921e-01 4.82332468e-01
-8.80771518e-01 6.68574810e-01 7.90604353e-01 -5.67264080e-01
-8.94470990e-01 -2.09388286e-01 -2.44810939e-01 -1.00455093e+00
4.44163471e-01 3.79791856e-01 2.59897858e-01 -4.46993351e-01
8.02434504e-01 5.03916919e-01 -2.37383023e-01 -6.35953605e-01
5.87049305e-01 -6.48093745e-02 7.96366513e-01 -5.11313200e-01
8.47116947e-01 3.08292359e-01 -2.84263641e-01 -7.23643959e-01
-1.22118175e+00 -6.10149349e-04 -4.33890074e-01 2.52241462e-01
8.65329683e-01 -7.65019476e-01 -8.37550282e-01 5.20429194e-01
-1.04598129e+00 -1.04726541e+00 -3.89638722e-01 3.61444354e-01
-6.49782240e-01 -1.79830611e-01 -1.10767221e+00 -5.23566365e-01
-7.53299534e-01 -1.23660290e+00 9.00886714e-01 5.47893047e-02
-9.11073163e-02 -1.33738959e+00 1.17002994e-01 3.53203803e-01
8.58526111e-01 -3.20351541e-01 9.64986861e-01 -6.06179476e-01
-8.75477195e-01 4.94886041e-01 -1.21850841e-01 2.77523994e-01
-2.82581925e-01 -1.22915626e-01 -1.19502532e+00 -3.94782215e-01
4.73105349e-02 -7.56398916e-01 1.53245986e+00 5.87960601e-01
1.24521267e+00 -8.09690714e-01 -3.01363856e-01 1.17493355e+00
1.18445790e+00 -1.66580528e-01 6.61281645e-01 5.07937372e-01
8.86260211e-01 -4.64762673e-02 1.51485028e-02 2.68717147e-02
6.09533548e-01 7.93606579e-01 6.56360865e-01 -3.11111242e-01
-2.53836811e-01 -2.14372694e-01 5.49238980e-01 1.21951890e+00
-2.07205951e-01 -1.91159546e-01 -7.21728444e-01 6.45266473e-01
-1.84195220e+00 -7.93777883e-01 2.13502780e-01 1.90338302e+00
1.01461887e+00 1.97558582e-01 -2.07988083e-01 -2.06628725e-01
2.11293846e-01 2.29122639e-01 -9.33646619e-01 -5.28103769e-01
-2.60087758e-01 9.35829401e-01 8.15488219e-01 7.47433424e-01
-1.03839839e+00 1.50059986e+00 7.16182804e+00 9.64814186e-01
-1.15703857e+00 5.15773654e-01 6.30899191e-01 -6.85130119e-01
-3.84749979e-01 9.70679335e-03 -9.17733014e-01 1.05970986e-01
1.36385262e+00 2.12343365e-01 1.01487494e+00 8.25324118e-01
-2.43097633e-01 3.73185724e-01 -1.11515319e+00 1.06070852e+00
3.36921774e-02 -1.66328180e+00 2.34764561e-01 -2.05287278e-01
7.17421174e-01 7.46623158e-01 6.94893837e-01 7.18264818e-01
6.71728969e-01 -1.48942995e+00 9.32531357e-01 5.85324228e-01
6.59414709e-01 -5.71002841e-01 4.13468748e-01 1.08144000e-01
-9.04322028e-01 -3.85449529e-01 -7.06934094e-01 1.62210241e-01
-6.37797043e-02 4.42243934e-01 -7.48480082e-01 -2.56760661e-02
1.14490044e+00 7.58905172e-01 -9.93183374e-01 8.83929551e-01
-2.65175700e-01 8.30587745e-01 -5.70133090e-01 -1.46787297e-02
5.73906779e-01 2.29147226e-01 5.76440573e-01 1.06640518e+00
1.99247032e-01 1.26796197e-02 -3.86441410e-01 8.51451814e-01
-3.29581916e-01 -3.41362953e-01 -2.29790628e-01 -1.37312070e-01
4.37927127e-01 1.14423811e+00 -4.27671820e-01 -2.89678365e-01
1.25001535e-01 1.18181551e+00 9.97943759e-01 6.24072850e-01
-8.79491925e-01 -2.02374861e-01 6.08390033e-01 -8.48763064e-02
6.96198940e-01 -2.11962789e-01 -3.37568730e-01 -1.19520032e+00
-2.49278679e-01 -8.01546872e-01 2.91881055e-01 -1.06692970e+00
-9.97065663e-01 1.05685842e+00 -3.78188819e-01 -3.10531199e-01
-8.94598663e-02 -8.41654003e-01 -5.89190722e-01 5.71779370e-01
-1.41345620e+00 -1.25827038e+00 4.12687510e-02 7.56486654e-01
6.17986798e-01 -4.17104065e-01 8.39283407e-01 2.59961516e-01
-6.86383843e-01 9.84529197e-01 8.98959413e-02 -2.36193761e-01
4.80748266e-01 -1.36403513e+00 6.68834567e-01 7.32231617e-01
5.63059151e-01 1.01613045e+00 8.03380549e-01 -1.11781888e-01
-1.49655139e+00 -1.17872846e+00 8.23572397e-01 -9.03580070e-01
7.30557740e-01 -5.09133697e-01 -9.55904603e-01 1.27565563e+00
3.24891925e-01 4.81292652e-03 1.88563645e-01 4.87799227e-01
-3.06210577e-01 -3.99685532e-01 -6.45804465e-01 7.15320289e-01
1.10318780e+00 -6.72410965e-01 -7.19116986e-01 4.87118244e-01
1.12184191e+00 -5.27376115e-01 -7.53324628e-01 3.26888233e-01
3.48509550e-01 -1.06770635e+00 1.24985814e+00 -7.17833161e-01
9.74702165e-02 -6.61730841e-02 -5.42672612e-02 -1.29177940e+00
-5.97886026e-01 -6.35816097e-01 -6.84028864e-01 6.91800177e-01
7.00146198e-01 -7.81628847e-01 6.17220759e-01 4.60380644e-01
-7.27985919e-01 -1.30644631e+00 -9.56259131e-01 -6.36088550e-01
-6.61173984e-02 -4.34335828e-01 3.16440433e-01 5.78991652e-01
-5.93623698e-01 4.60159361e-01 -5.23525298e-01 1.77566171e-01
3.79288465e-01 -2.85958022e-01 2.90887415e-01 -8.74876380e-01
-8.42447639e-01 -7.21866369e-01 -6.45395145e-02 -1.29922664e+00
2.68989168e-02 -8.14595878e-01 -1.12512425e-01 -1.46235394e+00
2.06776440e-01 -6.27327681e-01 -7.30976939e-01 1.07662416e+00
-8.31743628e-02 4.77484077e-01 7.70268217e-02 4.27835733e-01
-9.95903969e-01 9.33193326e-01 9.74649727e-01 -2.94441134e-01
-5.19621000e-02 -2.06855401e-01 -8.00271094e-01 6.19975090e-01
5.28693557e-01 -5.65008044e-01 -7.21833467e-01 -1.02402008e+00
3.76345426e-01 -3.50762397e-01 4.44998294e-01 -1.18087971e+00
3.46414506e-01 1.02529868e-01 3.93873841e-01 -2.84701049e-01
6.61939442e-01 -2.95078516e-01 2.94365119e-02 4.50244755e-01
-4.95339215e-01 4.53602701e-01 4.71076548e-01 2.77412176e-01
1.08742483e-01 -1.78977266e-01 9.12816226e-01 -3.40935171e-01
-8.16834927e-01 5.80204546e-01 -2.98722982e-01 1.86320126e-01
5.13854444e-01 -9.80569273e-02 -4.65897143e-01 -2.17513829e-01
-9.75448906e-01 2.52063990e-01 4.19295669e-01 2.06091821e-01
7.98372865e-01 -1.03312123e+00 -3.60636413e-01 -1.74222782e-01
-7.40459487e-02 -6.41429946e-02 1.03671305e-01 9.77138996e-01
-5.66317260e-01 6.73990726e-01 2.25324128e-02 -5.24637699e-01
-8.51063073e-01 5.64798117e-01 7.52080560e-01 -6.46372437e-01
-7.16710091e-01 1.43726826e+00 5.01280665e-01 -4.51253682e-01
2.66300410e-01 -4.67091709e-01 -1.03713669e-01 -3.12647074e-01
4.13662791e-01 -3.93418148e-02 3.07955891e-01 -3.14732343e-01
-4.21096712e-01 2.73232251e-01 -3.97291154e-01 -2.30634332e-01
1.57043862e+00 6.96273744e-02 -1.14639334e-01 4.87927973e-01
1.06451499e+00 -4.95456934e-01 -1.41387177e+00 -2.51842380e-01
-9.68177170e-02 9.97643992e-02 4.04198974e-01 -8.96370888e-01
-1.40003288e+00 1.04156792e+00 5.58662057e-01 -3.85450721e-01
1.05294728e+00 -3.86626571e-02 8.81657839e-01 1.11949778e+00
2.70819575e-01 -9.57243264e-01 6.63441896e-01 9.59875226e-01
8.04952741e-01 -7.15719700e-01 -2.46746555e-01 5.96374333e-01
-8.24118555e-01 9.13178802e-01 7.01682150e-01 -1.75002545e-01
3.71959418e-01 2.09520802e-01 -1.78120703e-01 -5.33903182e-01
-1.38599908e+00 -1.62372395e-01 3.92419189e-01 5.03832817e-01
9.53336209e-02 -3.31558347e-01 6.10038996e-01 5.31044900e-01
-4.26499337e-01 -1.26268685e-01 9.04968828e-02 3.88906151e-01
-5.06543934e-01 -7.21449196e-01 7.57258087e-02 5.77862680e-01
-3.26565921e-01 -8.07702780e-01 -1.57952055e-01 6.35440290e-01
-1.45681530e-01 4.98708457e-01 -3.36366659e-03 -4.28410172e-01
1.61608338e-01 -1.09002599e-02 6.08293295e-01 -4.53440309e-01
-8.98282766e-01 -4.88150775e-01 5.89784048e-03 -7.30777860e-01
6.70590624e-02 -2.14251250e-01 -1.04423976e+00 -3.85369837e-01
-3.17112297e-01 -3.00343782e-02 4.31735784e-01 1.19191611e+00
5.96475184e-01 7.26444423e-01 5.94869815e-02 -1.07866132e+00
-7.25887001e-01 -9.55924630e-01 -1.58643395e-01 9.06025246e-02
5.45155346e-01 -7.45274961e-01 -2.57604271e-01 -9.46510807e-02]
|
[8.926549911499023, 2.850337266921997]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.