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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
c6e0a0f9-2504-4f79-bbc0-4edf2babd547
|
learning-using-privileged-information-for
|
2206.08632
| null |
https://arxiv.org/abs/2206.08632v2
|
https://arxiv.org/pdf/2206.08632v2.pdf
|
Learning Using Privileged Information for Zero-Shot Action Recognition
|
Zero-Shot Action Recognition (ZSAR) aims to recognize video actions that have never been seen during training. Most existing methods assume a shared semantic space between seen and unseen actions and intend to directly learn a mapping from a visual space to the semantic space. This approach has been challenged by the semantic gap between the visual space and semantic space. This paper presents a novel method that uses object semantics as privileged information to narrow the semantic gap and, hence, effectively, assist the learning. In particular, a simple hallucination network is proposed to implicitly extract object semantics during testing without explicitly extracting objects and a cross-attention module is developed to augment visual feature with the object semantics. Experiments on the Olympic Sports, HMDB51 and UCF101 datasets have shown that the proposed method outperforms the state-of-the-art methods by a large margin.
|
['Zihui Guo', 'Yonghong Hou', 'Bin Yu', 'Wanqing Li', 'Zhiyi Gao']
|
2022-06-17
| null | null | null | null |
['zero-shot-action-recognition']
|
['computer-vision']
|
[ 4.29904938e-01 2.15377584e-01 -3.07351679e-01 -4.21606392e-01
-3.67020935e-01 -1.42246574e-01 7.86455274e-01 -2.43167222e-01
-5.86387277e-01 7.66076624e-01 4.01021421e-01 1.98386922e-01
1.87412016e-02 -4.93658096e-01 -7.45706499e-01 -5.42607784e-01
1.58308204e-02 2.03604504e-01 6.55605376e-01 -1.62389517e-01
2.95027822e-01 1.67057052e-01 -1.91071832e+00 5.03212392e-01
7.26288557e-01 8.56118321e-01 3.62761587e-01 3.30873638e-01
-1.87041625e-01 1.22329593e+00 -7.00601280e-01 -1.87940553e-01
2.38292828e-01 -7.04516351e-01 -9.06572282e-01 5.89345634e-01
3.14796478e-01 -4.64277983e-01 -6.90142214e-01 1.09060788e+00
1.60529763e-01 6.08177125e-01 3.95690829e-01 -1.61737835e+00
-8.58081520e-01 1.81813553e-01 -3.02197456e-01 3.39532793e-01
4.70042080e-01 2.44243532e-01 7.59395182e-01 -8.83946538e-01
7.52864599e-01 1.22979867e+00 -3.61011690e-03 8.44438732e-01
-8.22049201e-01 -4.36965227e-01 4.31134969e-01 6.99904919e-01
-1.38916922e+00 -2.97234058e-01 8.73482347e-01 -4.14843768e-01
1.01954925e+00 1.90418988e-01 8.32389593e-01 1.43089902e+00
-7.45260194e-02 1.23040760e+00 9.59459484e-01 -2.98150986e-01
4.62775320e-01 1.89453512e-01 5.73830344e-02 6.08537436e-01
-2.60493513e-02 9.63737443e-02 -8.30275416e-01 1.30631745e-01
6.86360657e-01 3.66300762e-01 -4.16560650e-01 -9.63537991e-01
-1.11117625e+00 7.41740882e-01 4.40314144e-01 2.51939088e-01
-3.55521023e-01 2.37978194e-02 5.04938483e-01 1.85112178e-01
1.35446623e-01 4.48684871e-01 -2.37632230e-01 -1.34642497e-01
-5.61285317e-01 1.64182705e-03 3.87057602e-01 8.55581701e-01
5.81504703e-01 1.94418505e-01 -3.46407443e-01 6.25620723e-01
1.74405366e-01 4.35965955e-01 8.15405726e-01 -7.77209759e-01
4.57095444e-01 8.01167607e-01 1.78636864e-01 -8.33776355e-01
-1.54106151e-02 -7.76131377e-02 -3.07403535e-01 5.41409671e-01
2.66711444e-01 3.62547785e-01 -1.23643875e+00 1.78462172e+00
2.99074769e-01 6.52785003e-01 5.39757371e-01 1.29421782e+00
7.97612071e-01 3.97683322e-01 3.11494112e-01 2.16885321e-02
1.21342289e+00 -1.20391798e+00 -7.00901806e-01 -5.91010869e-01
4.93337482e-01 -1.67955235e-01 1.34194255e+00 2.81329691e-01
-7.41853058e-01 -8.07749808e-01 -1.33402634e+00 -8.18434209e-02
-5.36703348e-01 6.58511892e-02 4.83674496e-01 3.32962751e-01
-5.94952464e-01 5.61559796e-01 -9.88015950e-01 -5.80128789e-01
6.88211739e-01 9.81027111e-02 -7.75113046e-01 -9.35135558e-02
-1.24516404e+00 9.39593494e-01 9.25034106e-01 -1.94677860e-01
-1.29343569e+00 -2.68103272e-01 -1.12121236e+00 1.41293183e-01
8.84718120e-01 -3.66923481e-01 9.67290044e-01 -1.45337129e+00
-1.44517398e+00 7.80241191e-01 1.48053899e-01 -6.83482528e-01
4.36777413e-01 -5.20012617e-01 -5.88281155e-01 5.08582652e-01
1.33628130e-01 7.10006475e-01 9.70558405e-01 -1.11177874e+00
-6.51914120e-01 -5.22416592e-01 2.67070800e-01 6.55960083e-01
-4.79414672e-01 -2.54115701e-01 -4.97350901e-01 -6.78255916e-01
6.25879318e-02 -7.83243299e-01 1.06951152e-03 -2.06903182e-02
-2.83769280e-01 -1.89516708e-01 1.15348601e+00 -4.75131720e-01
8.62174213e-01 -2.26745296e+00 3.66428614e-01 -5.14409952e-02
-3.37446444e-02 4.17276502e-01 -2.08524242e-01 2.92639822e-01
-2.16979235e-01 -4.26694065e-01 -1.41959026e-01 -5.91004027e-05
-1.63039699e-01 4.60140228e-01 -4.10530269e-01 5.56916893e-01
6.58401847e-02 8.40306222e-01 -1.12475920e+00 -4.43067968e-01
4.88506019e-01 3.99580687e-01 -3.65726262e-01 3.98440391e-01
-2.04445899e-01 5.04869401e-01 -4.74043101e-01 5.63049793e-01
2.47564122e-01 -2.55852610e-01 2.43568420e-01 7.42090568e-02
2.16729268e-01 -4.29397076e-02 -1.18584943e+00 2.08919501e+00
-4.78685647e-02 4.78654623e-01 -3.95597309e-01 -1.15542316e+00
7.95705676e-01 2.70971984e-01 3.59268725e-01 -8.09536994e-01
9.01338235e-02 -1.45017162e-01 -1.19150624e-01 -8.27113628e-01
3.51479143e-01 -3.50342244e-01 -7.80065656e-02 1.94390193e-01
4.04712647e-01 3.81976426e-01 -4.92628030e-02 1.07629120e-01
9.47323680e-01 5.34982443e-01 5.01092076e-01 1.99879035e-02
7.17143118e-01 4.05844599e-02 6.52556956e-01 7.28427291e-01
-4.23860490e-01 4.84405696e-01 3.31349999e-01 -4.40326482e-01
-7.12171614e-01 -1.29875135e+00 3.07879955e-01 1.16887581e+00
5.46844840e-01 -3.29200655e-01 -7.32921243e-01 -1.03561687e+00
-1.26838028e-01 1.06001604e+00 -7.87086070e-01 -6.43344939e-01
-3.70135516e-01 1.79471020e-02 3.38760346e-01 8.64088893e-01
7.09816337e-01 -1.42693937e+00 -1.15122068e+00 -1.13489129e-01
-1.39462441e-01 -1.30885601e+00 -3.13025385e-01 -1.78500175e-01
-6.94691360e-01 -1.37621343e+00 -6.51751220e-01 -5.20764768e-01
6.98301136e-01 4.67973739e-01 7.25626707e-01 -7.55142421e-02
-3.70556265e-01 6.49419069e-01 -5.99602342e-01 -3.12407166e-01
-1.40339375e-01 -3.38857055e-01 6.85769171e-02 4.50783819e-01
8.56630981e-01 -2.57984251e-01 -4.68147367e-01 2.08021536e-01
-9.92788017e-01 2.20071569e-01 6.22961700e-01 8.78384829e-01
4.14729446e-01 -1.03982367e-01 5.91709197e-01 -6.86064422e-01
1.25206083e-01 -4.41961914e-01 -3.39804322e-01 4.39532399e-01
-3.14312547e-01 1.08459495e-01 5.30801952e-01 -3.88241678e-01
-1.33223593e+00 1.73980981e-01 4.01739091e-01 -9.66609120e-01
-3.98513168e-01 3.12627912e-01 -6.99555695e-01 3.46609771e-01
4.29440916e-01 5.11587918e-01 1.58060730e-01 -5.42415380e-01
3.75230819e-01 5.99318385e-01 6.70292914e-01 -1.39350668e-01
5.46991467e-01 8.64251316e-01 -1.82093740e-01 -7.96377242e-01
-1.01855981e+00 -6.83026433e-01 -8.47673416e-01 -3.73379111e-01
1.32214344e+00 -1.00555170e+00 -3.95273656e-01 4.05931324e-01
-7.51615345e-01 -1.77803114e-01 -6.05733395e-01 8.22279811e-01
-9.16754007e-01 4.27553415e-01 -1.44183964e-01 -7.26772368e-01
1.08823769e-01 -9.78960574e-01 9.88041341e-01 2.47707590e-01
-1.73489377e-01 -7.06372797e-01 -9.57675278e-02 6.32978380e-01
-1.16052121e-01 2.37740785e-01 5.99433959e-01 -9.87844944e-01
-6.12480521e-01 -1.27980039e-01 -7.05639273e-02 4.43360955e-01
2.28788882e-01 -5.59689164e-01 -9.23917055e-01 -2.94541597e-01
1.67019323e-01 -6.03476167e-01 9.88128722e-01 5.33175729e-02
1.25218308e+00 -2.38070548e-01 -2.11094558e-01 4.21668828e-01
1.31969404e+00 3.82209063e-01 1.03262937e+00 4.05696124e-01
7.31671214e-01 5.37911415e-01 1.01103139e+00 4.01056796e-01
6.04412593e-02 5.75540125e-01 5.47450602e-01 2.32338440e-02
-8.68312258e-04 -5.65837443e-01 4.31884140e-01 1.36662975e-01
-3.96276936e-02 -2.14826643e-01 -5.63732564e-01 4.50446427e-01
-2.18030381e+00 -1.28740573e+00 3.94385368e-01 2.28811646e+00
4.38841075e-01 2.79700369e-01 -2.91466545e-02 9.56983306e-03
6.84309542e-01 3.90758306e-01 -7.34117746e-01 -3.20189372e-02
-8.06131307e-03 3.77536416e-02 3.31787467e-01 2.01951742e-01
-1.33749962e+00 1.20914006e+00 5.55054855e+00 6.75943494e-01
-8.92645121e-01 1.03717826e-01 7.90088400e-02 -1.89413011e-01
1.88377291e-01 -1.11972317e-01 -3.59468788e-01 3.75653774e-01
5.69714963e-01 -1.59171730e-01 1.05506890e-01 1.23040283e+00
-1.73400976e-02 -9.10404995e-02 -1.23376882e+00 1.01165819e+00
6.03136063e-01 -1.05325437e+00 2.66949862e-01 -1.07980080e-01
5.45337617e-01 -2.59527624e-01 -8.14371333e-02 6.03946030e-01
1.04241155e-01 -1.06423748e+00 5.38670361e-01 8.42230797e-01
6.96832836e-01 -7.14614153e-01 5.64398408e-01 2.84876555e-01
-9.49620187e-01 -3.80709976e-01 -4.09162134e-01 -1.92751780e-01
5.19053452e-02 -3.98801655e-01 -6.28489256e-01 3.90605628e-01
7.21792996e-01 1.01954532e+00 -6.51113570e-01 9.50574517e-01
-4.44841504e-01 2.15028718e-01 2.99422950e-01 1.94382936e-01
4.50007558e-01 -1.06516249e-01 7.49049902e-01 6.03959680e-01
2.73340583e-01 2.20408246e-01 4.23367560e-01 6.53122187e-01
1.76689655e-01 6.62150048e-03 -8.81890476e-01 -1.08265363e-01
2.24063449e-04 7.10657656e-01 -6.46313787e-01 -6.79181695e-01
-7.61040449e-01 1.44148719e+00 2.85176307e-01 5.08112669e-01
-9.82079327e-01 -2.91319937e-01 8.82966936e-01 1.05850384e-01
6.27041698e-01 2.88212877e-02 2.65264481e-01 -1.39268315e+00
-9.67774466e-02 -7.97587752e-01 6.03429496e-01 -1.00847566e+00
-9.55510974e-01 2.99917728e-01 1.82789162e-01 -1.69758344e+00
-2.37163976e-01 -6.16704404e-01 -4.20854688e-01 4.88227546e-01
-1.28598845e+00 -9.87764657e-01 -3.79671633e-01 8.38220179e-01
9.60080087e-01 -5.97204089e-01 8.15150380e-01 -3.16631235e-02
-2.07600370e-01 2.88465410e-01 -4.30346951e-02 3.92141491e-01
4.16075170e-01 -1.02194571e+00 -5.96166141e-02 7.81428397e-01
3.91124606e-01 4.18682665e-01 7.50761867e-01 -8.02840471e-01
-1.22906053e+00 -1.07692349e+00 4.10720825e-01 -5.65095127e-01
4.39900190e-01 -3.05032879e-01 -1.10409892e+00 8.71759772e-01
2.12691471e-01 2.32852057e-01 8.03537309e-01 -2.24315628e-01
-4.60673094e-01 5.50908819e-02 -9.85361993e-01 6.61990345e-01
1.28482509e+00 -6.30891562e-01 -1.24425244e+00 1.16115019e-01
5.53997278e-01 -2.31825292e-01 -3.98750663e-01 3.26928914e-01
4.74553108e-01 -9.87561345e-01 1.00237727e+00 -1.40447211e+00
2.70556867e-01 -4.38081622e-01 -2.63304442e-01 -1.29712129e+00
-4.54903431e-02 -7.05341697e-02 -3.24926466e-01 7.89788246e-01
-5.41793220e-02 -2.92589903e-01 8.76289904e-01 5.70356011e-01
-1.58652246e-01 -3.89920592e-01 -9.84264314e-01 -1.14117515e+00
-3.94318491e-01 -2.51127094e-01 4.87464070e-01 9.82781529e-01
2.22912535e-01 4.26680148e-01 -6.94205284e-01 5.68937100e-02
6.54289305e-01 1.10426605e-01 8.12558234e-01 -1.07626355e+00
-3.57202560e-01 -6.47738650e-02 -1.04927433e+00 -9.64807987e-01
4.79246348e-01 -7.66960740e-01 6.25637397e-02 -1.39094913e+00
4.27865356e-01 3.34144145e-01 -6.64268553e-01 5.95324695e-01
-2.50333816e-01 1.80189222e-01 3.31597805e-01 7.57976547e-02
-1.08096600e+00 1.05077326e+00 1.28206718e+00 -1.68275267e-01
-1.08144902e-01 -1.75493598e-01 -6.00968421e-01 9.63172138e-01
6.99185848e-01 -2.61027843e-01 -1.05081594e+00 -2.63023496e-01
-4.41247880e-01 1.13026828e-01 7.94482887e-01 -1.20382845e+00
3.44035141e-02 -3.56890887e-01 4.90578711e-01 -4.40263897e-01
6.59532666e-01 -1.03473115e+00 -1.91896901e-01 5.24165332e-01
-6.04276776e-01 -3.30401599e-01 -8.58659670e-02 1.06806540e+00
-4.13684964e-01 -1.87544599e-01 6.73528731e-01 -3.01900625e-01
-1.51667070e+00 1.60087898e-01 -1.83641165e-01 2.32090384e-01
1.59371054e+00 -5.87625921e-01 -2.91557223e-01 -1.94529340e-01
-1.18399715e+00 3.02559316e-01 4.56877619e-01 9.75820541e-01
1.06487250e+00 -1.50729656e+00 -3.71916503e-01 4.49037224e-01
5.55012226e-01 -5.30686855e-01 6.00517511e-01 6.05560303e-01
-1.25812441e-01 3.59495372e-01 -6.89999282e-01 -4.47749168e-01
-1.47671616e+00 9.47859168e-01 1.86314017e-01 1.36891961e-01
-1.12431479e+00 7.14206219e-01 4.87619936e-01 6.35063509e-03
4.03696686e-01 2.69023106e-02 -4.43726063e-01 -1.59532785e-01
8.43930483e-01 3.36257517e-01 -3.55860293e-01 -1.02142978e+00
-4.27374274e-01 1.75851941e-01 -3.32362168e-02 -5.44536971e-02
1.18384802e+00 -1.02138117e-01 4.47276950e-01 6.72381103e-01
1.25685275e+00 -6.25320137e-01 -1.63965142e+00 -4.34140950e-01
1.41424313e-01 -1.10931087e+00 -1.99638918e-01 -6.43243372e-01
-9.88227606e-01 8.91489863e-01 7.40310133e-01 -6.73676506e-02
9.59481537e-01 2.74141431e-01 4.97807473e-01 4.78242725e-01
3.31354856e-01 -1.52288401e+00 7.62931108e-01 2.32649490e-01
9.59725142e-01 -1.50210500e+00 -1.96237803e-01 -2.31256589e-01
-1.31805038e+00 9.27854180e-01 1.18329465e+00 -1.32213786e-01
3.75483245e-01 -3.38850796e-01 -9.78737101e-02 -3.84238839e-01
-6.36526287e-01 -5.00869274e-01 3.99750292e-01 7.19291270e-01
-8.21014345e-02 -1.58494681e-01 -9.63726256e-04 6.22714937e-01
3.01730067e-01 1.45556569e-01 4.86121178e-01 1.15560627e+00
-6.69906080e-01 -7.40563750e-01 -2.28798673e-01 2.59914607e-01
-3.07315856e-01 2.11969361e-01 -4.48366553e-01 1.03959894e+00
8.74882936e-02 7.57768214e-01 2.33685374e-01 -2.90624112e-01
3.47589791e-01 3.63927543e-01 4.61261719e-01 -8.10932040e-01
2.47806795e-02 -1.29074231e-01 -6.06738776e-02 -9.31761146e-01
-5.31611919e-01 -6.55964911e-01 -1.46515989e+00 4.53720182e-01
3.51795293e-02 6.93989098e-02 2.03545913e-01 1.03518379e+00
2.48079285e-01 5.92733741e-01 3.23792726e-01 -4.37768131e-01
-6.61436141e-01 -7.74407923e-01 -7.33261526e-01 8.94647717e-01
2.04643950e-01 -1.08730853e+00 -2.16118008e-01 1.13968156e-01]
|
[8.555285453796387, 0.9188034534454346]
|
f0fbbc07-f7f0-46e2-b422-4c4d5f66b129
|
drug-repurposing-targeting-covid-19-3cl
|
2305.18088
| null |
https://arxiv.org/abs/2305.18088v3
|
https://arxiv.org/pdf/2305.18088v3.pdf
|
Drug Repurposing Targeting COVID-19 3CL Protease using Molecular Docking and Machine Learning Regression Approach
|
The COVID-19 pandemic has created a global health crisis, driving the need for the rapid identification of potential therapeutics. To meet this challenge, drug repurposing is the only solution with saving cost and time. In this study, we used the Zinc database to screen the world-approved including FDA-approved 5903 drugs for repurposing as potential COVID-19 treatments targeting the main protease 3CL of SARS-CoV-2. We performed molecular docking using Autodock-Vina to check the efficacy of drug molecules. To enhance the efficiency of drug repurposing approach, we modeled the binding affinities using several machine learning regression approaches for QSAR modeling such as decision tree, extra trees, MLP, KNN, XGBoost, and gradient boosting. The computational results demonstrated that Decision Tree Regression (DTR) model has improved statistical measures of R2 and RMSE. These simulated results helped to identify drugs with high binding affinity and favorable binding energies. From the statistical analysis, we shortlisted six promising drugs with their respective Zinc IDs (ZINC000003873365, ZINC000085432544, ZINC000203757351, ZINC000085536956, ZINC000008214470 and ZINC000261494640) within the range of -15.1 kcal/mol to -13.6 kcal/mol. All are novel compounds except ZINC000203757351 antiviral compound that was already identified against COVID-19 in other studies. Further, we analyzed the physiochemical and pharmacokinetic properties of these selected drugs with respect to their best binding interaction to specific target protease 3CLpro. Our study has provided an efficient framework for drug repurposing against COVID-19. This highlights the potential of combining molecular docking with machine learning regression approaches to accelerate the identification of potential therapeutic candidates.
|
['Abdul Majid', 'Imra Aqeel']
|
2023-05-25
| null | null | null | null |
['molecular-docking']
|
['medical']
|
[-1.35079375e-03 -5.50153971e-01 -2.09011614e-01 5.31601347e-02
-5.16749084e-01 -6.46464825e-01 9.55685135e-03 5.20724773e-01
-3.10417265e-01 1.46133912e+00 -9.99874026e-02 -6.84867263e-01
-2.18573257e-01 -5.60563922e-01 -5.71006596e-01 -8.09595585e-01
-2.11031348e-01 5.70220709e-01 -2.12445050e-01 -2.53960460e-01
3.79655421e-01 8.09673667e-01 -7.30365217e-01 1.50137395e-01
1.56186521e+00 2.84010291e-01 4.04493421e-01 3.52733076e-01
3.45430434e-01 1.51948214e-01 -5.10052741e-01 -1.29111469e-01
1.58546809e-02 -4.42993134e-01 -4.30487931e-01 -8.13821852e-01
-2.80905277e-01 -2.08649009e-01 5.06626308e-01 6.07861996e-01
6.81845784e-01 -7.69057795e-02 9.86308694e-01 -5.80645382e-01
-3.41194391e-01 -2.93342590e-01 -4.78118330e-01 1.43648490e-01
6.54457927e-01 1.47308558e-01 6.68398678e-01 -1.15999269e+00
7.28176296e-01 7.63794124e-01 6.38473749e-01 5.64831793e-01
-1.16228294e+00 -9.16625798e-01 -4.78993654e-01 5.36888063e-01
-1.50067997e+00 1.58545494e-01 7.77494907e-02 -8.72324526e-01
1.62292218e+00 2.21540019e-01 5.57379127e-01 6.88869298e-01
6.70570970e-01 -1.94236357e-02 1.01867270e+00 -7.66878799e-02
3.58806580e-01 6.36304542e-02 6.59775138e-02 5.98350704e-01
5.50845325e-01 -1.77637432e-02 -4.14931685e-01 -9.38373089e-01
4.33351666e-01 3.55822980e-01 -3.56904298e-01 -6.44205436e-02
-6.44930124e-01 1.03215933e+00 3.33901286e-01 -8.93629715e-02
-6.30359590e-01 -4.84968603e-01 2.74235487e-01 -1.13463394e-01
2.41621137e-01 5.97835660e-01 -1.03636289e+00 -1.39673632e-02
-2.64979571e-01 3.75940740e-01 3.78344536e-01 1.47740215e-01
5.97477913e-01 -1.17992312e-01 2.23983288e-01 5.11794865e-01
6.35839403e-01 5.88026464e-01 2.57437341e-02 7.84456059e-02
2.51077890e-01 6.08485460e-01 4.85485286e-01 -7.51465082e-01
-6.63503528e-01 -1.42825603e-01 -6.56781495e-01 6.01525828e-02
1.76915035e-01 -3.47464740e-01 -6.22782946e-01 1.62282026e+00
3.32525164e-01 3.65381725e-02 3.03433746e-01 7.66266704e-01
6.90552413e-01 9.83898461e-01 6.09776080e-01 -8.35328937e-01
1.39396024e+00 -3.40565383e-01 -2.29211003e-01 6.88611746e-01
5.40263832e-01 -8.46748292e-01 6.24708593e-01 6.16330624e-01
-5.42554080e-01 -1.50868535e-01 -1.16566467e+00 7.44492292e-01
-1.71726525e-01 -1.62044689e-02 6.17028832e-01 8.00693154e-01
-6.33072078e-01 6.97202444e-01 -8.85084033e-01 -2.70934910e-01
1.27245709e-01 8.69780183e-01 -4.70669448e-01 1.04380913e-01
-1.31166375e+00 1.18423843e+00 4.09462154e-01 -1.13440126e-01
-8.11305702e-01 -7.50763595e-01 -3.01831216e-01 -1.57773480e-01
-1.08572580e-01 -7.39738524e-01 4.15891409e-01 -2.29128107e-01
-1.51209223e+00 3.77809107e-01 -3.96559507e-01 -3.35371405e-01
-1.09790258e-01 -2.09605023e-01 -3.55003029e-01 1.01679057e-01
2.16237288e-02 2.87337631e-01 4.11287043e-03 -7.38070190e-01
-2.58349210e-01 -8.41515422e-01 -3.74772429e-01 2.57719725e-01
1.29075617e-01 4.98694092e-01 2.42495224e-01 -4.78025734e-01
-1.02127209e-01 -8.41072500e-01 -2.82400757e-01 -9.00611162e-01
-3.33853036e-01 -1.97941497e-01 3.01888049e-01 -6.69856489e-01
9.42818403e-01 -1.48544741e+00 1.64485455e-03 6.51316822e-01
-3.60024497e-02 6.00868702e-01 1.83125183e-01 1.00842237e+00
-2.61410505e-01 2.99378753e-01 2.10184097e-01 7.60534883e-01
-7.79840469e-01 -4.32899296e-01 -1.27592459e-01 6.71879590e-01
8.33794475e-02 7.24273086e-01 -6.62866175e-01 6.12813495e-02
7.53683448e-02 7.82025993e-01 -6.96998000e-01 2.58134007e-01
-4.34303939e-01 7.60850787e-01 -8.87861013e-01 9.06727850e-01
8.59184206e-01 -1.26723230e-01 6.54459119e-01 -1.02981158e-01
-2.01434478e-01 2.79027641e-01 -5.89022577e-01 9.52342510e-01
3.95017453e-02 -9.86477658e-02 -6.49478078e-01 -4.58512008e-01
1.06496239e+00 5.30328512e-01 6.93694592e-01 -5.88918567e-01
6.36225343e-02 3.41978490e-01 1.32063061e-01 -3.51614952e-01
-3.57604586e-02 -2.99390376e-01 4.32878315e-01 -1.16379790e-01
-2.43673548e-01 4.81875300e-01 -7.74788260e-02 -1.94455922e-01
8.73866498e-01 3.04672539e-01 6.59772754e-01 -4.91472214e-01
7.24292576e-01 3.12805951e-01 6.97813869e-01 1.25507906e-01
6.89346269e-02 4.17680509e-04 3.13096315e-01 -4.46345657e-01
-9.41252649e-01 -7.17801094e-01 -6.67207778e-01 7.98069954e-01
-2.89302412e-02 -4.75899905e-01 -7.23455846e-01 -1.93305686e-01
-1.83026269e-01 2.84597367e-01 -2.77159035e-01 1.33953094e-01
-3.57386380e-01 -1.13907766e+00 4.41076040e-01 -1.39096901e-01
8.85626897e-02 -8.45484316e-01 -2.47047439e-01 5.17075837e-01
1.85329720e-01 -3.71022284e-01 7.93183967e-03 3.93845439e-01
-9.23754454e-01 -1.39217234e+00 -5.75612307e-01 -5.56967318e-01
3.90067399e-01 -4.03530970e-02 3.38443339e-01 4.08635139e-02
-2.88318813e-01 -5.11904955e-01 -3.29938471e-01 -6.75958395e-01
-2.89529592e-01 -1.64354920e-01 5.43518960e-01 -6.60779655e-01
5.71862459e-01 -4.68731940e-01 -8.54969323e-01 4.42556232e-01
-2.78571814e-01 -1.37944177e-01 3.47025514e-01 8.08966875e-01
6.84038162e-01 -3.00795972e-01 8.35275233e-01 -7.89942920e-01
7.79710412e-01 -8.11457336e-01 -6.97634280e-01 2.48997718e-01
-8.76214325e-01 -1.34978861e-01 7.17979014e-01 -8.38449523e-02
-9.09392655e-01 1.15592197e-01 -5.09129047e-01 2.38566637e-01
4.77134250e-02 7.69859731e-01 -3.67966384e-01 -3.16606760e-02
7.66967773e-01 9.71194729e-02 8.11603218e-02 -5.95590472e-01
-3.02611142e-01 6.10551715e-01 1.60843320e-02 -5.44760644e-01
4.43463266e-01 -2.89947037e-02 9.64116752e-02 -8.16111684e-01
-4.14272211e-02 -4.86519694e-01 -1.31621761e-02 1.43372327e-01
1.11991310e+00 -9.02688205e-01 -1.51706851e+00 3.71588767e-01
-1.14603651e+00 1.08386181e-01 9.01924074e-01 8.49444985e-01
-2.61321902e-01 4.75874543e-01 -6.32789671e-01 -5.44679046e-01
-9.72471893e-01 -1.36220920e+00 3.83451492e-01 2.19659925e-01
-3.60546976e-01 -6.22413874e-01 6.65065825e-01 5.44678450e-01
2.16506943e-01 5.85974574e-01 1.40176594e+00 -1.16764677e+00
-5.12934506e-01 8.72500092e-02 6.07133768e-02 -4.25077640e-02
1.65330112e-01 2.37042889e-01 -4.71744090e-01 -2.45880336e-01
-2.68232495e-01 -2.46130869e-01 4.39821810e-01 5.79632461e-01
5.10183036e-01 -2.76263565e-01 -4.69069988e-01 5.77597737e-01
1.58906829e+00 1.32857919e+00 8.73333037e-01 6.84712052e-01
4.76058632e-01 1.06551446e-01 9.12203789e-01 7.66877234e-01
-2.89066625e-03 9.13094401e-01 5.96307278e-01 -6.54070899e-02
8.11647296e-01 -1.29277976e-02 2.43416160e-01 -1.13133751e-01
-7.87637174e-01 -2.49329060e-01 -1.08429039e+00 -7.62817189e-02
-1.44986975e+00 -8.65022898e-01 -4.19269890e-01 2.36172199e+00
1.18584907e+00 -2.61261731e-01 1.50274307e-01 -3.60389948e-01
3.97292316e-01 -5.76124132e-01 -5.27735710e-01 -7.36705482e-01
-3.02687827e-02 5.72778821e-01 3.44778955e-01 4.97922987e-01
-8.02555442e-01 9.16443586e-01 5.17976665e+00 7.52077520e-01
-1.07284057e+00 -3.25705856e-01 5.96390784e-01 2.20726997e-01
-1.91437081e-02 1.62217617e-01 -9.35825884e-01 4.10629362e-01
9.90051925e-01 -1.57814905e-01 2.82729775e-01 6.75976396e-01
7.86946535e-01 -2.21950710e-01 -7.08226502e-01 6.30287349e-01
-3.80010307e-01 -1.63265073e+00 8.14865455e-02 3.61708850e-01
6.73213601e-01 8.45460966e-02 -1.31645977e-01 -2.44748652e-01
1.03963658e-01 -1.33423233e+00 -1.57541499e-01 4.40899998e-01
6.43570006e-01 -1.16584694e+00 9.04428780e-01 1.62274286e-01
-6.77639365e-01 1.88839614e-01 -3.85430992e-01 1.22620963e-01
-8.30198452e-02 3.30712676e-01 -1.47164786e+00 5.10497570e-01
3.96799475e-01 2.08092451e-01 -7.22763389e-02 1.02713609e+00
-9.13175941e-02 3.54767919e-01 -9.24179703e-02 -2.98478514e-01
-7.07151890e-02 -4.85441148e-01 2.66748041e-01 9.01466429e-01
5.46712056e-02 4.92733181e-01 2.16772705e-02 4.65323627e-01
2.75454819e-01 7.83868015e-01 -1.97740376e-01 8.57952386e-02
5.18170297e-01 7.14144647e-01 -4.28510278e-01 -6.62531406e-02
-2.28687197e-01 5.36955237e-01 -1.08641669e-01 3.79283071e-01
-8.16373587e-01 -1.67191163e-01 1.16209340e+00 1.93444505e-01
3.88563946e-02 -9.50667784e-02 -9.48098674e-02 -7.17200339e-01
-6.35791957e-01 -1.10623288e+00 3.90936583e-01 -5.75442553e-01
-9.54880834e-01 6.39538765e-01 4.77854609e-02 -9.46827888e-01
-5.85871376e-02 -6.09039724e-01 -4.83423799e-01 1.40177405e+00
-9.74706531e-01 -7.82856584e-01 1.65075824e-01 4.10028189e-01
2.02254802e-01 -3.69328737e-01 1.23216856e+00 5.95111623e-02
-4.75768358e-01 1.71645150e-01 6.83846653e-01 -6.81963742e-01
6.64913058e-01 -7.27411568e-01 -8.42003897e-02 1.45849675e-01
-6.48867726e-01 1.27645314e+00 8.35175931e-01 -1.14835322e+00
-1.32585418e+00 -7.17671573e-01 8.62596810e-01 -9.19724330e-02
4.77906913e-01 -6.05974160e-02 -7.88301528e-01 1.57401830e-01
5.47951683e-02 -8.57806206e-01 1.26833010e+00 -1.56565800e-01
-1.57771796e-01 2.57105470e-01 -1.17094147e+00 6.25758767e-01
3.02177668e-01 -2.01934561e-01 -1.34362519e-01 5.50688565e-01
4.06719446e-01 -2.25497171e-01 -1.01029015e+00 7.72239447e-01
6.54819369e-01 -8.51700723e-01 1.06654477e+00 -9.86114681e-01
2.92245727e-02 -5.83227038e-01 1.39665818e-02 -9.09681320e-01
-4.08369005e-01 -4.06133503e-01 4.65706646e-01 6.11567438e-01
6.79917455e-01 -6.88675404e-01 8.10432851e-01 3.64834487e-01
1.97824299e-01 -1.22774386e+00 -8.72749567e-01 -5.71156681e-01
1.54681996e-01 -1.26073696e-02 3.66265833e-01 8.05321991e-01
2.58427620e-01 4.00354952e-01 -5.05257785e-01 3.18225585e-02
2.29975343e-01 2.84635760e-02 6.07968569e-01 -1.21227980e+00
-3.13926250e-01 -1.57208666e-01 -2.15020746e-01 -4.65471238e-01
-2.67874211e-01 -6.89261794e-01 -7.74766505e-01 -1.49167991e+00
6.79760277e-01 -3.79439473e-01 -2.24184468e-01 5.49720228e-01
-8.43530446e-02 -6.44525960e-02 -2.54780501e-01 3.10103327e-01
1.88411754e-02 2.10232034e-01 8.84407699e-01 1.18240565e-01
-7.20444739e-01 3.04349720e-01 -5.90014935e-01 4.08298165e-01
1.05213785e+00 -4.61773694e-01 -4.51780021e-01 5.65813541e-01
5.19641936e-01 1.71786368e-01 -5.09741902e-02 -1.67848960e-01
-2.35447302e-01 -7.63179004e-01 5.72843015e-01 -8.94760132e-01
2.21780866e-01 -5.89373708e-01 8.72968554e-01 8.99435699e-01
4.16298687e-01 5.47289811e-02 3.51449221e-01 2.82267869e-01
2.46254310e-01 -2.14507446e-01 6.55636013e-01 1.70842454e-01
-3.35585326e-01 1.64248705e-01 -7.86685765e-01 -3.97187382e-01
1.05191779e+00 -6.97172403e-01 -2.37556234e-01 1.18440002e-01
-8.72246265e-01 -1.07532881e-01 5.02926588e-01 1.26007795e-01
6.25234544e-01 -6.55802488e-01 -6.92727506e-01 8.55658855e-03
1.70299470e-01 -6.23054564e-01 3.17575067e-01 7.27360129e-01
-1.18741393e+00 8.61629665e-01 -4.64703798e-01 -3.90364110e-01
-1.76028800e+00 6.48726165e-01 4.44312394e-01 -2.23125726e-01
-8.55500996e-02 6.22391999e-01 2.57103562e-01 1.83366507e-03
3.20073850e-02 4.73037586e-02 -5.75057089e-01 -1.04199909e-01
2.88769513e-01 5.34994483e-01 3.15414876e-01 -5.24854124e-01
-1.05546975e+00 6.60571396e-01 -1.04422159e-01 5.72455347e-01
1.43860042e+00 4.70183432e-01 -2.80430287e-01 -3.91964644e-01
1.08201075e+00 2.08067641e-01 -5.83289087e-01 2.71298230e-01
8.81013125e-02 -3.58211398e-01 -4.91468847e-01 -1.22507775e+00
-1.01046205e-01 2.15405658e-01 8.00817609e-01 -6.14289641e-01
9.34635878e-01 -1.03854224e-01 2.85288870e-01 3.08083594e-01
5.88370860e-01 -5.87594450e-01 -1.05451882e-01 5.43780386e-01
8.09784591e-01 -9.38503802e-01 1.26321018e-01 -2.69111007e-01
-5.65763175e-01 1.07020319e+00 5.49501240e-01 3.28743547e-01
3.94434541e-01 -1.84524208e-02 -9.80380252e-02 -3.89051855e-01
-8.50509644e-01 3.53793800e-01 2.81899273e-01 5.62083364e-01
8.03773105e-01 2.19481617e-01 -1.18752599e+00 6.80811107e-01
9.43474099e-02 -1.46001847e-02 3.34121704e-01 6.34833813e-01
-6.24477744e-01 -1.61596847e+00 -5.58521330e-01 8.64607915e-02
-6.79622769e-01 -5.52158296e-01 -5.99068940e-01 6.49140775e-01
2.09529400e-01 9.25170422e-01 -5.20335972e-01 -2.51606226e-01
2.43581399e-01 5.91283813e-02 3.81435364e-01 -4.89432305e-01
-7.39495456e-01 6.08448625e-01 -1.52354725e-02 5.24866134e-02
-1.30555257e-01 -2.74323553e-01 -1.74369717e+00 -3.93083394e-01
-8.01841557e-01 5.89365482e-01 9.80344772e-01 7.05661893e-01
6.20601475e-01 -4.47914973e-02 7.39323795e-01 -3.25480074e-01
-1.59066081e-01 -8.58850121e-01 -5.85337758e-01 -2.14132637e-01
-7.24108517e-02 -5.25253534e-01 1.98475435e-01 -6.74135461e-02]
|
[4.727728366851807, 5.2632246017456055]
|
9e77b6bb-f615-4012-b9c6-b1bf6520dbb0
|
an-empirical-study-of-end-to-end-temporal
|
2204.02932
| null |
https://arxiv.org/abs/2204.02932v1
|
https://arxiv.org/pdf/2204.02932v1.pdf
|
An Empirical Study of End-to-End Temporal Action Detection
|
Temporal action detection (TAD) is an important yet challenging task in video understanding. It aims to simultaneously predict the semantic label and the temporal interval of every action instance in an untrimmed video. Rather than end-to-end learning, most existing methods adopt a head-only learning paradigm, where the video encoder is pre-trained for action classification, and only the detection head upon the encoder is optimized for TAD. The effect of end-to-end learning is not systematically evaluated. Besides, there lacks an in-depth study on the efficiency-accuracy trade-off in end-to-end TAD. In this paper, we present an empirical study of end-to-end temporal action detection. We validate the advantage of end-to-end learning over head-only learning and observe up to 11\% performance improvement. Besides, we study the effects of multiple design choices that affect the TAD performance and speed, including detection head, video encoder, and resolution of input videos. Based on the findings, we build a mid-resolution baseline detector, which achieves the state-of-the-art performance of end-to-end methods while running more than 4$\times$ faster. We hope that this paper can serve as a guide for end-to-end learning and inspire future research in this field. Code and models are available at \url{https://github.com/xlliu7/E2E-TAD}.
|
['Xiang Bai', 'Song Bai', 'Xiaolong Liu']
|
2022-04-06
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Liu_An_Empirical_Study_of_End-to-End_Temporal_Action_Detection_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Liu_An_Empirical_Study_of_End-to-End_Temporal_Action_Detection_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['action-classification']
|
['computer-vision']
|
[ 1.07227489e-01 -2.30221704e-01 -6.40673220e-01 -4.02131855e-01
-1.00980341e+00 -4.08616334e-01 3.46598327e-01 -2.44881451e-01
-5.54637372e-01 1.43448904e-01 2.92683929e-01 -2.84731597e-01
2.01214537e-01 -1.97071716e-01 -7.56519973e-01 -3.86626393e-01
-4.55171943e-01 -1.21805586e-01 5.44302940e-01 2.79012501e-01
1.52987778e-01 1.39046714e-01 -1.44028664e+00 6.64293945e-01
3.20763767e-01 1.19841504e+00 2.55338162e-01 8.70137513e-01
4.45486605e-01 1.24690044e+00 -2.67121404e-01 -3.43756348e-01
2.76963443e-01 -6.31135821e-01 -6.59740329e-01 2.19940707e-01
5.11334360e-01 -8.10285628e-01 -8.74635398e-01 7.83625245e-01
6.95975661e-01 7.03439116e-02 4.11049932e-01 -1.33507979e+00
-3.23019922e-01 2.75350928e-01 -7.80823231e-01 6.98745251e-01
4.46624547e-01 4.28117007e-01 1.08349371e+00 -9.53997016e-01
6.05610192e-01 1.00166690e+00 5.18821836e-01 5.74966073e-01
-8.06769788e-01 -7.54423022e-01 3.12217712e-01 6.40488625e-01
-1.32177258e+00 -8.15232396e-01 6.14149332e-01 -5.34067571e-01
9.88615453e-01 -2.07678542e-01 3.59932423e-01 1.11961532e+00
2.78254896e-01 1.26179945e+00 7.77263880e-01 -3.47807795e-01
1.99464813e-01 -3.20112437e-01 -1.14035331e-01 9.05991256e-01
-2.20915601e-01 2.44391173e-01 -8.52171838e-01 2.11883143e-01
8.60334158e-01 -6.90224171e-02 -3.46488394e-02 -2.58257926e-01
-1.05765724e+00 6.06519282e-01 1.71650246e-01 -1.15661006e-02
-3.44466716e-01 3.50196123e-01 8.15622330e-01 2.26216078e-01
4.98139530e-01 6.26040474e-02 -6.01089835e-01 -8.89408886e-01
-1.09625924e+00 1.37163788e-01 3.59062612e-01 1.04525983e+00
3.09729844e-01 -1.13240905e-01 -5.06010771e-01 6.43452644e-01
1.49873897e-01 2.48797432e-01 2.66638547e-01 -1.42670763e+00
5.21426439e-01 3.01459312e-01 1.51677966e-01 -5.88039815e-01
-3.81687045e-01 -3.08294117e-01 -3.43580723e-01 3.67778808e-01
5.99516809e-01 -4.13517565e-01 -7.69751668e-01 1.87509108e+00
2.15079337e-01 5.20515442e-01 -2.40482062e-01 1.19366980e+00
5.76985061e-01 5.85546732e-01 2.15429381e-01 -3.20395023e-01
1.46199715e+00 -1.36228442e+00 -6.51473165e-01 -3.62268507e-01
9.80449915e-01 -7.10632026e-01 1.12370360e+00 3.86818349e-01
-1.27832496e+00 -6.78205431e-01 -8.10319722e-01 -2.16268152e-01
1.50109470e-01 6.07339442e-01 3.77418637e-01 1.24061182e-01
-7.13487089e-01 4.38146114e-01 -1.31690860e+00 -3.84469509e-01
6.47830307e-01 1.51987851e-01 -2.07691550e-01 -1.56876430e-01
-9.98126447e-01 6.56935275e-01 1.48183152e-01 -2.81308293e-01
-1.16550791e+00 -5.82008064e-01 -5.77370346e-01 -2.45986164e-01
7.82573760e-01 -4.64249372e-01 1.81022012e+00 -1.02366328e+00
-1.39385331e+00 8.83461952e-01 -4.70076799e-01 -6.53629363e-01
6.82329595e-01 -5.75697422e-01 -3.30907792e-01 4.53072757e-01
1.61602587e-01 6.60292506e-01 7.52103508e-01 -6.12385988e-01
-1.00304914e+00 -3.33903432e-01 2.78164983e-01 3.00068289e-01
-1.99596033e-01 2.64672756e-01 -1.08527946e+00 -7.17668176e-01
-1.66814774e-01 -9.88653064e-01 8.68167207e-02 3.22384119e-01
2.80681830e-02 -3.96499455e-01 9.41507816e-01 -7.41691411e-01
1.50141728e+00 -2.51462674e+00 -1.44904524e-01 -5.39331257e-01
9.05184746e-02 3.29240292e-01 -2.18019024e-01 2.83377081e-01
-1.45204842e-01 -1.28619596e-01 2.15837821e-01 -4.79704916e-01
5.27456542e-03 -2.23783553e-01 -6.33231252e-02 5.96400023e-01
1.13896504e-01 7.93483853e-01 -9.58846807e-01 -6.27889335e-01
3.84050608e-01 3.41856003e-01 -5.83694279e-01 3.93858194e-01
-2.53613383e-01 4.20979023e-01 -4.22377884e-01 6.22014701e-01
1.36265486e-01 -4.37431157e-01 -2.71643023e-03 -2.63805091e-01
-4.98662964e-02 3.66718620e-01 -1.02782726e+00 1.86978602e+00
-2.80060768e-01 8.80426705e-01 -3.10880262e-02 -8.11476529e-01
3.89023066e-01 3.93382698e-01 6.89843178e-01 -9.93002355e-01
7.63619617e-02 -6.35693967e-02 -5.47072366e-02 -7.42768824e-01
1.67360157e-01 6.89915717e-02 4.65495363e-02 3.24791789e-01
-5.83596267e-02 6.50586784e-01 4.13977414e-01 1.94018990e-01
1.34042907e+00 5.38654149e-01 3.38056892e-01 2.11426362e-01
1.54114172e-01 -8.49827901e-02 6.09860241e-01 6.57357752e-01
-6.13694608e-01 5.31667113e-01 6.94467127e-01 -4.45035040e-01
-9.92708087e-01 -9.03014243e-01 1.17857896e-01 1.58012295e+00
1.34560019e-01 -5.83664715e-01 -7.86887646e-01 -8.52272213e-01
-2.43146345e-01 7.23234296e-01 -4.98169243e-01 -1.02443099e-01
-5.66786468e-01 -2.36494526e-01 6.53144896e-01 8.80696654e-01
6.49829030e-01 -9.17389810e-01 -8.26177776e-01 4.02741916e-02
-3.22437465e-01 -1.49187458e+00 -8.50404441e-01 -8.29863101e-02
-1.02812815e+00 -1.06751525e+00 -4.68850315e-01 -6.20364845e-01
3.40196341e-01 4.38657969e-01 1.04931498e+00 -1.56443357e-01
-7.20005929e-02 4.59649444e-01 -5.76682448e-01 -2.18431264e-01
-1.80556819e-01 1.09672518e-02 2.51895320e-02 -1.10635839e-01
6.51228786e-01 -3.59257370e-01 -9.12966967e-01 5.42115927e-01
-6.25792384e-01 1.89262599e-01 6.93414927e-01 3.58503848e-01
5.83438575e-01 1.12987243e-01 3.41922611e-01 -5.18758297e-01
9.89200249e-02 -3.57020736e-01 -4.84315068e-01 3.23562436e-02
-4.91476148e-01 -3.98442475e-03 4.91950661e-01 -3.33245516e-01
-8.31989527e-01 3.16579431e-01 -1.24793455e-01 -8.37576449e-01
-2.75549352e-01 3.23966295e-01 -8.46639723e-02 3.74771297e-01
4.37858671e-01 2.70111620e-01 1.07495980e-02 -3.65523905e-01
1.93138957e-01 6.18050575e-01 4.93610829e-01 -2.37595350e-01
3.65424335e-01 4.09094751e-01 -2.26453692e-01 -7.30580330e-01
-1.07166672e+00 -6.55880690e-01 -6.24885142e-01 -4.49643701e-01
1.13324130e+00 -1.42958498e+00 -6.63332701e-01 5.29192924e-01
-8.30996931e-01 -7.69771576e-01 -3.36716510e-02 7.40959764e-01
-8.99112880e-01 3.21560651e-01 -7.66263366e-01 -7.59797215e-01
-1.68501705e-01 -1.05873919e+00 1.21524298e+00 1.29733786e-01
-3.53903055e-01 -7.63856351e-01 -2.31139913e-01 7.01353192e-01
-7.97892436e-02 3.33603472e-02 4.57803935e-01 -3.04557890e-01
-6.97152317e-01 -2.42865548e-01 -2.05143794e-01 3.72467279e-01
-1.10892884e-01 -4.78220768e-02 -9.04055357e-01 -5.13804972e-01
-1.65525198e-01 -4.61939186e-01 1.00370383e+00 7.33068228e-01
1.29253662e+00 -1.76176533e-01 -2.42440045e-01 4.88361806e-01
1.23706138e+00 2.58543193e-01 7.65981734e-01 2.72128969e-01
6.22301459e-01 3.33470672e-01 1.16094029e+00 7.69405365e-01
3.54673117e-01 1.04223669e+00 3.60250920e-01 5.92029504e-02
-4.41626966e-01 -3.44093412e-01 8.30558240e-01 1.85613155e-01
-2.29700804e-01 -4.02393937e-01 -7.39555955e-01 4.79735315e-01
-1.88889325e+00 -1.24812376e+00 1.96529791e-01 2.34946918e+00
6.95815504e-01 4.95179325e-01 6.96242094e-01 7.74436072e-02
6.43746018e-01 4.25471634e-01 -7.13490486e-01 -8.09551403e-02
3.24908704e-01 -2.12298840e-01 6.00994706e-01 4.07254785e-01
-1.49651229e+00 1.07016182e+00 5.88030577e+00 9.47972417e-01
-1.14150250e+00 1.36809707e-01 6.59459352e-01 -5.66948593e-01
6.27597809e-01 -5.97255528e-02 -9.59978938e-01 5.79247952e-01
1.08548784e+00 1.12186491e-01 2.44470835e-01 9.16427433e-01
8.88844550e-01 -2.03679964e-01 -1.33694887e+00 1.18211269e+00
-8.14745110e-03 -1.06465709e+00 -3.92909914e-01 4.38545868e-02
3.62184763e-01 1.54146567e-01 -2.54291948e-02 4.04214114e-01
-1.30285442e-01 -7.00120568e-01 7.86498845e-01 2.58832783e-01
8.95454705e-01 -5.65925896e-01 4.67229068e-01 3.59165907e-01
-1.43153536e+00 -4.02081102e-01 1.23271355e-02 -4.17007267e-01
3.29551518e-01 3.47877413e-01 -6.25133157e-01 1.64793372e-01
7.43680418e-01 1.07102859e+00 -4.80618984e-01 9.43532467e-01
-4.07083035e-01 9.85826850e-01 7.05654547e-02 1.57695383e-01
3.46611351e-01 1.57990694e-01 4.28544343e-01 1.39892209e+00
2.44838595e-01 3.90239239e-01 3.93089533e-01 3.08371365e-01
-8.17807689e-02 -1.88826293e-01 -4.23321277e-01 -2.23033473e-01
4.55727696e-01 8.29760969e-01 -6.97941780e-01 -3.57236207e-01
-7.83719003e-01 1.19907892e+00 1.68585286e-01 4.23150629e-01
-1.37283492e+00 -3.57945450e-02 8.21864605e-01 5.16867995e-01
7.48035491e-01 -3.86474699e-01 -2.10793346e-01 -9.37991381e-01
2.34345362e-01 -9.78990912e-01 5.97223163e-01 -7.32898653e-01
-1.01255035e+00 1.16346017e-01 2.94375122e-02 -1.55674064e+00
-2.57243067e-01 -5.95529437e-01 -3.98841918e-01 2.43955895e-01
-1.10279346e+00 -8.25280488e-01 -1.37258142e-01 4.68268722e-01
1.00928080e+00 7.10691698e-03 4.50370610e-01 4.97382283e-01
-7.35885203e-01 8.73787284e-01 2.90218033e-02 5.76126575e-01
7.89951980e-01 -8.87228072e-01 2.90726274e-01 1.04590154e+00
1.32824436e-01 -2.64367592e-02 7.02972889e-01 -5.74292183e-01
-1.44019294e+00 -1.21288145e+00 7.26440549e-01 -5.17381132e-01
5.33998370e-01 -2.77266264e-01 -4.89143252e-01 8.01251531e-01
1.02650277e-01 1.80951953e-01 5.63244045e-01 1.40297249e-01
-2.69983977e-01 -2.59023964e-01 -6.65757596e-01 7.00516105e-01
1.40104735e+00 -6.70897543e-01 -2.93046981e-01 2.30401754e-01
5.88022351e-01 -5.82011819e-01 -8.12427580e-01 2.80588090e-01
8.08255434e-01 -1.20818889e+00 9.47163939e-01 -5.85863352e-01
7.57555842e-01 -2.35786706e-01 -1.16875105e-01 -6.84212029e-01
-6.01959229e-01 -5.47158062e-01 -5.11351943e-01 8.62332106e-01
3.43566000e-01 -9.56751779e-03 9.75130856e-01 6.16471589e-01
-1.62418798e-01 -1.06477571e+00 -8.92036736e-01 -9.63221192e-01
-2.09713742e-01 -6.49186611e-01 -1.60146266e-01 6.43673480e-01
-6.38667047e-02 5.23626626e-01 -6.82321668e-01 3.04437727e-01
4.52482939e-01 1.81230269e-02 7.75515974e-01 -4.25512493e-01
-4.96568710e-01 -3.79281968e-01 -5.26505291e-01 -1.59001052e+00
-3.21149714e-02 -4.52836543e-01 1.55722825e-02 -1.34679413e+00
3.12140107e-01 1.38788551e-01 -3.49138558e-01 6.06576145e-01
-2.26189584e-01 3.17152083e-01 3.25176656e-01 2.23353356e-01
-1.25499666e+00 4.04712558e-01 1.18084371e+00 2.17480823e-01
-1.92439422e-01 1.18631683e-01 -3.53765666e-01 8.07736099e-01
8.22133243e-01 -4.97462541e-01 -6.70298100e-01 -5.91075540e-01
-4.30116653e-01 3.53058994e-01 3.74736786e-01 -1.24715757e+00
2.64835507e-01 -1.77741110e-01 3.74408960e-01 -6.32736206e-01
5.32772064e-01 -5.95076382e-01 -3.75406742e-01 4.52017069e-01
-5.13015509e-01 1.22209452e-01 1.38247937e-01 6.50096357e-01
-1.53872356e-01 6.72227656e-03 8.94075811e-01 -6.00349419e-02
-1.18738317e+00 4.56357419e-01 -4.30945784e-01 3.29077452e-01
1.31875837e+00 -4.12909627e-01 -1.35112703e-01 -6.32636607e-01
-6.22769117e-01 2.41384313e-01 5.09086430e-01 3.90577376e-01
5.64050376e-01 -1.20836258e+00 -7.94444323e-01 -9.88849998e-02
1.17146425e-01 -3.44436377e-01 4.86235559e-01 1.03015172e+00
-3.05221260e-01 3.82511586e-01 -4.05232459e-02 -6.52899861e-01
-1.60863054e+00 6.55984640e-01 3.09274554e-01 -1.69193104e-01
-6.17684007e-01 9.04291451e-01 2.23274902e-01 2.79259324e-01
7.44661510e-01 -9.87963974e-02 1.38566196e-01 -5.79848737e-02
6.70976579e-01 5.00143707e-01 -2.81189501e-01 -4.76486355e-01
-4.88372296e-01 4.39750135e-01 -1.54668272e-01 -1.47306889e-01
1.14586639e+00 -1.73695192e-01 5.02768219e-01 4.22776610e-01
1.29351079e+00 -3.76378030e-01 -1.90436995e+00 -2.14820936e-01
-1.73640728e-01 -6.68968916e-01 1.24283835e-01 -7.56932676e-01
-1.17937219e+00 8.24909568e-01 9.25909221e-01 -2.27040380e-01
1.42397916e+00 1.41184196e-01 8.40836048e-01 2.21160531e-01
2.26650283e-01 -1.23482955e+00 5.76068819e-01 5.35065174e-01
6.80878758e-01 -1.44863737e+00 1.29053786e-01 -2.18887925e-01
-8.71583104e-01 8.26655030e-01 8.25198531e-01 -5.28670885e-02
4.92217779e-01 2.96464562e-01 3.62453936e-03 -1.13804378e-01
-1.02121031e+00 -2.42475912e-01 1.40961245e-01 2.54536361e-01
6.45143688e-01 -6.94415867e-02 -2.56181449e-01 2.88200617e-01
2.34120131e-01 4.87894237e-01 2.67877638e-01 1.10657036e+00
-3.53203833e-01 -9.53373015e-01 -5.29183261e-02 3.98514986e-01
-5.65425694e-01 2.53193993e-02 -1.56357720e-01 6.39010549e-01
1.89859673e-01 1.11555171e+00 1.15001956e-02 -6.54430926e-01
3.62098604e-01 1.59569517e-01 4.76688772e-01 -4.02789265e-01
-8.28612447e-02 1.48571029e-01 3.32098871e-01 -1.02003598e+00
-4.90671128e-01 -9.02560174e-01 -1.42454088e+00 -4.04200912e-01
-1.35266811e-01 -2.26041019e-01 1.33174881e-01 8.66389036e-01
6.27049863e-01 4.48743284e-01 6.57055140e-01 -6.77852869e-01
-5.15497267e-01 -8.91891956e-01 -3.49719077e-01 4.05453295e-01
2.68812835e-01 -6.83086753e-01 -2.01732293e-01 3.70614558e-01]
|
[8.473756790161133, 0.37611404061317444]
|
ebabe482-2865-4ff2-b279-92a18c52679d
|
paracrawl-web-scale-acquisition-of-parallel
| null | null |
https://aclanthology.org/2020.acl-main.417
|
https://aclanthology.org/2020.acl-main.417.pdf
|
ParaCrawl: Web-Scale Acquisition of Parallel Corpora
|
We report on methods to create the largest publicly available parallel corpora by crawling the web, using open source software. We empirically compare alternative methods and publish benchmark data sets for sentence alignment and sentence pair filtering. We also describe the parallel corpora released and evaluate their quality and their usefulness to create machine translation systems.
|
['Pin-zhen Chen', 'Brian Thompson', "Elsa Sarr{\\'\\i}as", "Gema Ram{\\'\\i}rez-S{\\'a}nchez", 'Miquel Espl{\\`a}-Gomis', "Marta Ba{\\~n}{\\'o}n", 'William Waites', 'Mikel L. Forcada', 'Jaume Zaragoza', 'Hieu Hoang', 'Barry Haddow', 'Kenneth Heafield', 'Sergio Ortiz Rojas', 'Philipp Koehn', 'Marek Strelec', 'Leopoldo Pla Sempere', 'Faheem Kirefu', 'Dion Wiggins', 'Amir Kamran']
|
2020-07-01
| null | null | null |
acl-2020-6
|
['parallel-corpus-mining']
|
['natural-language-processing']
|
[ 2.36818954e-01 -2.60719627e-01 -2.81441659e-01 -4.56317127e-01
-1.69633019e+00 -1.06758428e+00 8.02431762e-01 2.48444989e-01
-5.85101902e-01 1.15521443e+00 5.02805769e-01 -5.38649440e-01
2.16272563e-01 -4.68172997e-01 -5.27759850e-01 -1.29534587e-01
6.77836835e-02 1.45202470e+00 3.90600920e-01 -9.34880376e-01
7.01829672e-01 1.85402691e-01 -1.09043336e+00 5.95953226e-01
1.14009666e+00 -1.19651332e-01 4.15270895e-01 9.65138376e-01
-3.47175360e-01 -1.61219314e-01 -9.93963420e-01 -1.13455856e+00
5.57477832e-01 -5.06477892e-01 -1.51509190e+00 -3.06622684e-01
8.24601591e-01 2.61780381e-01 -1.09627418e-01 1.11913621e+00
1.00601745e+00 -1.81739867e-01 1.25791967e-01 -7.84372449e-01
-4.49395150e-01 9.88716483e-01 -2.70330966e-01 9.42610621e-01
1.09625673e+00 1.61155179e-01 1.30736482e+00 -5.98142862e-01
1.22347665e+00 9.05071735e-01 7.31208026e-01 4.46790487e-01
-1.38375032e+00 -2.77047753e-01 -6.63892329e-01 4.56528157e-01
-1.15512979e+00 -7.64219761e-01 3.66401106e-01 -1.34139448e-01
1.70309961e+00 9.07671988e-01 6.87268734e-01 1.35496235e+00
5.99789739e-01 2.40103826e-01 1.28321409e+00 -9.59758937e-01
-3.58869970e-01 -2.22198486e-01 4.25837576e-01 2.56188929e-01
5.95778190e-02 -3.71678591e-01 -6.01959705e-01 -7.83977032e-01
1.17863193e-01 -8.73807967e-01 5.07605374e-02 4.07389402e-01
-1.58380878e+00 5.11952698e-01 -3.35112095e-01 7.02898800e-01
-3.47923249e-01 -8.05639505e-01 8.00290048e-01 1.03614259e+00
5.88309646e-01 9.74677086e-01 -4.34400082e-01 -4.33765948e-01
-7.65885234e-01 3.60351861e-01 1.16547799e+00 1.21318102e+00
4.32230085e-01 -7.30007648e-01 -5.60028404e-02 1.29497302e+00
-4.48987305e-01 6.01788521e-01 6.38110876e-01 -6.55180216e-01
1.13193667e+00 5.98775633e-02 -5.17133623e-02 -6.54274702e-01
-2.61584789e-01 9.72895399e-02 -2.27125406e-01 -7.39945352e-01
-1.88535638e-03 -2.55820602e-02 -1.08248375e-01 1.08977270e+00
2.36884058e-01 -4.12798673e-01 3.43744725e-01 5.40616751e-01
8.46144557e-01 5.69376409e-01 -3.14958841e-01 -5.29414058e-01
1.16270435e+00 -1.23246181e+00 -6.52921081e-01 -1.65007904e-01
8.63664508e-01 -1.98961627e+00 8.51275682e-01 1.29438758e-01
-1.49654686e+00 -4.91119325e-01 -1.07243180e+00 -9.89130139e-02
-4.76615019e-02 -3.34596038e-01 3.88989717e-01 3.47404838e-01
-1.35814488e+00 1.10113430e+00 -6.26258254e-01 -8.66009653e-01
-2.70990640e-01 5.01817644e-01 -5.88376522e-01 2.50028312e-01
-1.37351835e+00 1.53590441e+00 2.31267720e-01 -5.87857664e-01
8.54509845e-02 6.04150407e-02 -2.82750010e-01 -4.05167848e-01
-3.75711620e-01 -9.45040524e-01 1.40446842e+00 -8.25196147e-01
-1.33611071e+00 1.28635812e+00 -3.34256589e-01 -5.69358349e-01
2.12439165e-01 -1.52315237e-02 -9.83503580e-01 1.65500611e-01
5.35333395e-01 4.08287346e-01 -1.59828588e-02 -5.53073943e-01
-6.94002807e-01 -8.84585902e-02 -3.02429438e-01 3.78417730e-01
-4.01684076e-01 9.40763116e-01 -3.12446535e-01 -4.34079111e-01
-1.31125972e-01 -8.17864656e-01 -5.71041167e-01 -1.05749953e+00
-4.83688056e-01 -3.75571638e-01 -1.61486402e-01 -1.24338973e+00
1.25255108e+00 -1.25332797e+00 2.59814352e-01 1.27924904e-01
-1.17815234e-01 1.88373327e-01 -9.04266596e-01 1.38987494e+00
-7.49972239e-02 1.40774861e-01 3.98190804e-02 -2.26528242e-01
-1.98613897e-01 4.50930327e-01 -2.76895314e-01 3.73738974e-01
-6.43719509e-02 8.70478511e-01 -9.52559352e-01 -1.00245130e+00
-1.06701352e-01 -2.06283942e-01 -2.34721228e-01 1.15530320e-01
7.73385093e-02 4.57099974e-01 -2.66164720e-01 3.69680285e-01
4.07157183e-01 2.66425222e-01 5.04658520e-01 3.28263581e-01
-4.05335903e-01 1.38983774e+00 -4.53844965e-01 1.94528854e+00
-3.64618748e-01 8.22212100e-01 -1.13056794e-01 -6.50272131e-01
1.21855724e+00 4.59218949e-01 4.31394935e-01 -6.89686716e-01
-3.41372229e-02 5.89787066e-01 2.09339231e-01 -6.46057487e-01
9.19360697e-01 2.48938426e-01 -3.13508250e-02 6.17222428e-01
2.73355395e-01 -4.76158440e-01 9.83361483e-01 3.63653988e-01
1.27913296e+00 -1.38804898e-01 4.83749628e-01 -6.69200003e-01
6.52251601e-01 8.63818586e-01 4.42247689e-01 5.30692697e-01
-8.31319913e-02 5.38422465e-01 7.52390623e-02 -6.53239369e-01
-1.79983449e+00 -1.01830649e+00 -4.24040884e-01 8.55043113e-01
-2.81510770e-01 -7.35302150e-01 -9.18538511e-01 -5.79825103e-01
-5.31389117e-01 3.82570475e-01 8.74175280e-02 5.06328404e-01
-1.10562122e+00 -7.81807482e-01 7.62009144e-01 -1.70898251e-03
-1.13672696e-01 -9.12677586e-01 3.93418014e-01 2.91287899e-01
-1.04353154e+00 -1.05035460e+00 -8.67455125e-01 -9.60959196e-02
-8.40494692e-01 -1.09826994e+00 -2.87610114e-01 -1.33963120e+00
5.04232585e-01 1.53034478e-01 1.94742477e+00 6.93427697e-02
-1.38056785e-01 -3.08721717e-02 -5.70033252e-01 -1.39050990e-01
-1.15520501e+00 7.09934533e-01 3.81509632e-01 -1.00602603e+00
1.09919822e+00 -7.77029276e-01 -1.39023429e-02 4.01816100e-01
-2.60326535e-01 -8.41640756e-02 3.65013808e-01 9.30263698e-01
3.64373714e-01 -7.70800769e-01 5.01872241e-01 -7.69718349e-01
1.20317471e+00 -2.39107952e-01 -5.27832925e-01 4.78925616e-01
-5.50928652e-01 -4.75075934e-03 5.34722447e-01 -1.11455388e-01
-5.60870409e-01 -2.43867338e-02 -1.06190562e+00 5.42442322e-01
-8.42780471e-02 1.85952723e-01 2.70753533e-01 -1.81292906e-01
1.04835260e+00 2.73791194e-01 1.58315912e-01 -6.51026905e-01
1.59941033e-01 1.07008708e+00 5.29490530e-01 -6.59307182e-01
8.57734442e-01 -1.52133754e-03 -3.65548760e-01 -6.12500846e-01
-2.11655542e-01 -7.75496006e-01 -1.09830785e+00 1.66319430e-01
5.04202843e-01 -6.95091248e-01 -1.66561484e-01 -3.11015874e-01
-1.62803340e+00 3.04133475e-01 5.27636595e-02 4.92404521e-01
-6.22067094e-01 8.06723773e-01 -1.04650569e+00 2.04312615e-02
-1.14619243e+00 -1.11631703e+00 8.81489098e-01 -3.55311066e-01
-9.77849126e-01 -9.75069582e-01 1.08448052e+00 6.83745325e-01
1.46162719e-01 -4.15568739e-01 6.32266462e-01 -1.18829608e+00
2.18969584e-01 -1.25797957e-01 4.27189410e-01 1.26211196e-01
-3.85931547e-04 1.99432045e-01 -3.11151892e-01 -2.58579224e-01
-1.16542704e-01 -1.22720942e-01 2.14691937e-01 -1.03428297e-01
2.22080782e-01 -5.39796770e-01 -5.12641132e-01 7.20730722e-02
1.02978754e+00 -2.77791955e-02 5.61093867e-01 8.75218511e-01
1.91897333e-01 9.05916274e-01 6.19065166e-01 5.75331599e-02
2.75635302e-01 9.20909822e-01 -4.69908893e-01 4.94359955e-02
2.62933578e-02 5.63706271e-03 3.25153947e-01 2.29764915e+00
-1.83392689e-01 7.33680427e-02 -1.07185006e+00 5.05593896e-01
-1.68665040e+00 -1.20929992e+00 -6.90138280e-01 1.89061272e+00
1.44053769e+00 3.84402665e-04 4.05665189e-01 -2.22220272e-01
1.10246778e+00 -1.12430647e-01 3.85911435e-01 -1.10710800e+00
-6.57673776e-01 7.35062540e-01 6.01088464e-01 7.59155214e-01
-8.85447502e-01 1.22967589e+00 8.80445290e+00 6.30581200e-01
-5.75686693e-01 3.33801150e-01 2.65112638e-01 -5.55192344e-02
-4.07272190e-01 1.12358399e-01 -8.99215817e-01 2.06114694e-01
1.35370803e+00 -6.99545264e-01 4.00000632e-01 5.76955736e-01
1.51726082e-01 3.51725519e-01 -1.04552996e+00 5.51812828e-01
1.31818399e-01 -1.53533590e+00 1.63051069e-01 -1.34741023e-01
1.05131137e+00 8.79405797e-01 -5.49856961e-01 -4.35501412e-02
6.37419820e-01 -3.85032922e-01 1.80653825e-01 3.07735931e-02
4.67383802e-01 -6.39939368e-01 8.45175028e-01 6.59856498e-02
-8.39909315e-01 5.25543451e-01 -8.64773989e-01 -4.16245073e-01
4.20763731e-01 4.16711062e-01 -6.86763585e-01 9.41561520e-01
4.98272151e-01 7.36561537e-01 -7.91162252e-01 9.79974508e-01
-1.02147348e-01 3.99721652e-01 -2.53226548e-01 -3.38645548e-01
4.26083654e-02 -5.50886869e-01 8.95950317e-01 1.41971326e+00
1.61633432e-01 -2.07513407e-01 2.43139535e-01 1.07619911e-02
-2.40444764e-03 9.54847217e-01 -6.89681351e-01 4.17475067e-02
6.83035493e-01 1.15639627e+00 -4.21074986e-01 -6.36068940e-01
-5.40797532e-01 1.20850039e+00 4.65211779e-01 -6.32628128e-02
-3.90477687e-01 -6.17713273e-01 7.71294951e-01 -2.86785513e-01
-2.19336137e-01 -1.89509794e-01 -4.54721957e-01 -1.38895166e+00
4.27531958e-01 -1.59774470e+00 5.26905239e-01 -2.64629424e-01
-1.81786823e+00 1.22776651e+00 -1.39541477e-01 -1.37657678e+00
-5.34636438e-01 -3.45898062e-01 -8.04231822e-01 1.23223841e+00
-6.84213042e-01 -6.28105402e-01 5.93568444e-01 2.98473895e-01
8.23990405e-01 -4.01746422e-01 1.20733690e+00 5.76027513e-01
-4.07914251e-01 5.89200675e-01 4.81468052e-01 7.10583874e-04
1.19224918e+00 -1.04573250e+00 1.70905733e+00 8.52837324e-01
4.69632983e-01 7.84197032e-01 9.91250634e-01 -8.53392899e-01
-1.42642260e+00 -5.52200913e-01 2.00829625e+00 -8.64290893e-01
1.08495617e+00 -4.98761982e-01 -4.86231178e-01 6.00063503e-01
1.03049278e+00 -8.34284902e-01 9.08717573e-01 4.77526784e-01
-1.75080836e-01 1.10291965e-01 -9.25097227e-01 8.03194761e-01
1.36127996e+00 -5.03308356e-01 -1.30284536e+00 1.14754260e+00
6.01087987e-01 -2.94622332e-01 -1.20766521e+00 2.31004983e-01
3.80721658e-01 -7.29495406e-01 9.46380854e-01 -8.64070356e-01
8.15629601e-01 2.41405398e-01 1.74353287e-01 -1.63481939e+00
-3.84839714e-01 -9.81183112e-01 8.46508086e-01 1.22068322e+00
9.54012632e-01 -8.70539904e-01 5.56283176e-01 5.40800951e-02
-5.15322089e-01 -3.88959646e-01 -1.11391997e+00 -9.16295230e-01
5.44529617e-01 8.13178793e-02 8.06041956e-01 1.19256461e+00
9.48679268e-01 8.38684618e-01 -2.62740076e-01 -1.54913679e-01
2.52169162e-01 2.97833890e-01 9.54681516e-01 -8.55499625e-01
-5.96104085e-01 -6.48205340e-01 -4.53222722e-01 -9.45488930e-01
2.26615787e-01 -1.22831702e+00 -1.34243118e-02 -1.63928449e+00
4.73103642e-01 -1.21270269e-01 2.16970488e-01 -6.69921562e-03
-4.79224265e-01 4.41438735e-01 -2.18036294e-01 7.28155255e-01
-4.53453332e-01 -1.28655015e-02 9.94037926e-01 -2.64916509e-01
5.09846583e-02 -3.06240320e-01 -2.16318965e-01 1.20052472e-01
1.14051211e+00 -7.38787234e-01 2.70640552e-01 -8.42612147e-01
1.27686068e-01 4.26522782e-03 -6.37342691e-01 -7.39107549e-01
1.07870586e-01 -1.23205565e-01 2.09659878e-02 -7.90412068e-01
-5.63938320e-02 -1.84588253e-01 1.34260908e-01 6.23144388e-01
-4.63138223e-01 1.29535437e+00 7.15520680e-02 -2.63088048e-01
-4.34429884e-01 -4.42251891e-01 3.39998960e-01 -2.49417976e-01
-1.39892906e-01 -1.65995240e-01 -6.81780815e-01 8.14509019e-02
6.70906544e-01 3.10078412e-01 -4.60810483e-01 3.41089107e-02
-2.87443906e-01 3.75138670e-01 7.57284522e-01 7.77786732e-01
3.68076414e-02 -1.11893094e+00 -1.37243390e+00 -4.18886356e-02
2.45500714e-01 -1.12012565e+00 -2.46884510e-01 7.38020122e-01
-1.09445870e+00 6.38254285e-01 -7.29834676e-01 -4.37539369e-01
-1.88842845e+00 5.68239391e-01 -1.22435309e-01 -4.70664859e-01
-5.54927528e-01 5.58131754e-01 -8.65548551e-01 -8.83637011e-01
-3.56986880e-01 5.57491839e-01 -3.05188447e-01 -4.36537921e-01
5.67267418e-01 3.16803128e-01 5.64503253e-01 -7.15464294e-01
-3.29897493e-01 2.35294923e-01 -1.35935351e-01 -3.36970985e-01
1.04617548e+00 -2.65000224e-01 -1.00063562e+00 2.92380422e-01
1.16476631e+00 2.41664603e-01 2.50906229e-01 -8.61688778e-02
4.99701202e-01 -5.02588511e-01 -9.26767766e-01 -3.84250790e-01
-2.07686752e-01 2.68308431e-01 5.06332442e-02 1.62934944e-01
8.46821368e-01 1.61333997e-02 1.17987669e+00 8.26691866e-01
5.47566473e-01 -1.36707985e+00 -4.91369635e-01 8.84881794e-01
6.43621564e-01 -1.10076571e+00 7.79566765e-02 -7.63078451e-01
-2.90883541e-01 1.29135334e+00 2.80105829e-01 -2.43319541e-01
-8.24550018e-02 5.37698150e-01 6.21378243e-01 1.01059668e-01
-1.01636159e+00 -5.32730035e-02 1.99818462e-01 6.39431775e-01
9.70381558e-01 4.84258793e-02 -1.75495446e+00 7.09281936e-02
-9.78412449e-01 -6.80210650e-01 3.99475992e-01 9.06131029e-01
-2.05123335e-01 -2.36051965e+00 -4.55221504e-01 6.22188270e-01
-8.17467093e-01 -6.99954093e-01 -1.17020166e+00 4.29273933e-01
-5.08110523e-01 1.32009292e+00 1.19825259e-01 -6.17508769e-01
2.28878632e-01 3.33172023e-01 7.35764921e-01 -6.63413227e-01
-1.31569815e+00 -1.35880560e-01 1.16593373e+00 -2.15067521e-01
-2.93150544e-01 -9.05676544e-01 -5.80472052e-01 -6.73772693e-01
-3.80666047e-01 9.43355739e-01 6.67599499e-01 8.78564358e-01
3.96836251e-01 -1.50672600e-01 9.09784853e-01 -7.55292416e-01
-7.05336034e-01 -1.31552207e+00 2.99609512e-01 4.95093971e-01
-5.16177297e-01 2.59410650e-01 -6.27470762e-02 2.07964852e-02]
|
[11.466346740722656, 10.377930641174316]
|
e90728a9-d7d6-4412-bbf4-2e78c2ddf1e8
|
diachronic-parsing-of-pre-standard-irish
| null | null |
https://aclanthology.org/2022.cltw-1.2
|
https://aclanthology.org/2022.cltw-1.2.pdf
|
Diachronic Parsing of Pre-Standard Irish
|
Irish underwent a major spelling standardization in the 1940’s and 1950’s, and as a result it can be challenging to apply language technologies designed for the modern language to older, “pre-standard” texts. Lemmatization, tagging, and parsing of these pre-standard texts play an important role in a number of applications, including the lexicographical work on Foclóir Stairiúil na Gaeilge, a historical dictionary of Irish covering the period from 1600 to the present. We have two main goals in this paper. First, we introduce a small benchmark corpus containing just over 3800 words, annotated according to the Universal Dependencies guidelines and covering a range of dialects and time periods since 1600. Second, we establish baselines for lemmatization, tagging, and dependency parsing on this corpus by experimenting with a variety of machine learning approaches.
|
['Kevin Scannell']
| null | null | null | null |
cltw-lrec-2022-6
|
['lemmatization']
|
['natural-language-processing']
|
[ 1.00626417e-01 -1.01629332e-01 -2.21508726e-01 -4.14195746e-01
-8.20296049e-01 -1.08114445e+00 6.77945554e-01 4.71344590e-01
-9.45148528e-01 7.69870043e-01 7.22025514e-01 -5.92160583e-01
8.37972835e-02 -4.48242873e-01 -5.21392263e-02 -2.53850013e-01
5.81857003e-03 5.53232372e-01 8.56967643e-02 -5.13448596e-01
3.76237690e-01 3.94877881e-01 -8.89083505e-01 2.21161321e-01
5.29745162e-01 3.01557541e-01 2.75944352e-01 5.10729849e-01
-2.21770436e-01 2.64243215e-01 -8.30751061e-01 -8.44336569e-01
-1.02187833e-02 -3.48435283e-01 -1.20482683e+00 -2.13326305e-01
9.55698937e-02 1.69628680e-01 -3.29843074e-01 9.14730310e-01
4.27674741e-01 4.43801321e-02 4.54803735e-01 -4.61968988e-01
-2.89315611e-01 1.29351652e+00 -2.75303602e-01 4.35077637e-01
3.40438545e-01 -2.66559452e-01 1.22951937e+00 -4.76403147e-01
9.64423597e-01 1.04497707e+00 7.29125023e-01 4.64549631e-01
-9.62594628e-01 -6.72511697e-01 1.95241630e-01 1.72122885e-02
-1.24952936e+00 -4.91905391e-01 4.59605396e-01 -3.79300237e-01
1.13990831e+00 -3.53924595e-02 6.52142227e-01 8.93943071e-01
2.27909565e-01 7.15666234e-01 1.00906110e+00 -9.23049748e-01
-1.18493527e-01 -3.68938595e-02 2.04434350e-01 1.29512489e-01
3.67877513e-01 -3.09941977e-01 -3.33606452e-01 -5.23055159e-02
3.87085468e-01 -5.28420031e-01 -1.44153073e-01 1.96652323e-01
-1.22017264e+00 8.65327120e-01 -2.82753110e-01 8.33136141e-01
3.79750989e-02 -2.09553197e-01 9.10257936e-01 2.54361928e-01
2.89634019e-01 6.63492203e-01 -9.20899749e-01 -4.78675157e-01
-8.19942474e-01 9.44187865e-02 8.97196770e-01 8.87638807e-01
3.67164701e-01 -2.32242957e-01 5.82630277e-01 9.29367602e-01
2.09450722e-01 3.29997271e-01 5.99308372e-01 -8.13534677e-01
8.33514392e-01 1.01085685e-01 -2.12667207e-03 -4.78245229e-01
-6.09743237e-01 -1.20624594e-01 -3.97973567e-01 -3.16442788e-01
7.34046340e-01 -3.53497237e-01 -9.10279453e-01 1.72206640e+00
1.08962104e-01 -7.27066159e-01 4.47146416e-01 3.45726877e-01
4.94284362e-01 5.93510687e-01 4.59469616e-01 -4.05883580e-01
1.59280396e+00 -5.36820412e-01 -4.88559365e-01 -6.75733209e-01
1.00949931e+00 -1.19379461e+00 7.97918558e-01 4.80570436e-01
-1.12649381e+00 -3.01160783e-01 -9.16767538e-01 -1.13778777e-01
-2.97558010e-01 -2.05527216e-01 8.10063004e-01 6.74265146e-01
-8.76585603e-01 5.86326003e-01 -9.22001421e-01 -7.74668455e-01
-3.85614000e-02 1.59952462e-01 -4.97594059e-01 4.13416289e-02
-1.31834161e+00 9.03537571e-01 8.16080987e-01 -9.62037370e-02
-1.35358974e-01 -3.27884585e-01 -1.08462381e+00 -2.24036515e-01
1.74978554e-01 1.34374291e-01 1.44617665e+00 -1.02982593e+00
-1.03109252e+00 1.35237598e+00 1.20161861e-01 -2.97926605e-01
1.65522575e-01 -3.82714450e-01 -8.05670500e-01 -4.39645052e-02
3.18275511e-01 2.42014900e-01 5.79822995e-02 -6.38309240e-01
-1.21986473e+00 -3.05723995e-01 5.90157509e-02 2.25451887e-01
-1.52428314e-01 6.56759143e-01 -6.97664082e-01 -1.00045621e+00
1.35698780e-01 -9.23880398e-01 -2.70103544e-01 -1.00004208e+00
3.20758931e-02 -1.96341172e-01 1.53019175e-01 -1.09848464e+00
1.51822138e+00 -2.28822947e+00 6.03622571e-02 7.11917430e-02
-3.00226897e-01 3.97053331e-01 -1.12326950e-01 8.43757629e-01
-1.48293123e-01 3.12220305e-01 -2.31041342e-01 -2.35781670e-01
-5.84515482e-02 6.60211265e-01 -7.08233118e-02 5.02528012e-01
1.08802048e-02 7.24625409e-01 -1.05943716e+00 -5.92803299e-01
-1.13016933e-01 2.22530618e-01 -3.83064628e-01 -3.47315967e-01
1.01957373e-01 3.59780312e-01 -2.05019087e-01 6.37745321e-01
2.63088495e-01 4.30300206e-01 9.07022059e-01 7.96249807e-02
-5.96611559e-01 1.04891086e+00 -8.86088133e-01 1.99255872e+00
-5.30658782e-01 5.54368854e-01 1.19262606e-01 -8.57273877e-01
5.76642573e-01 4.05658722e-01 3.96942496e-01 -6.85500681e-01
3.80888432e-01 4.87346679e-01 2.08494335e-01 -2.43655547e-01
8.96659672e-01 -4.30355340e-01 -9.29181874e-01 4.70834136e-01
1.20234057e-01 -2.56495625e-01 8.93064857e-01 2.20653310e-01
1.15289307e+00 2.62868017e-01 8.18854094e-01 -6.34637773e-01
5.44808567e-01 3.62901121e-01 9.57836688e-01 1.62320316e-01
-7.61492997e-02 4.70538706e-01 4.88210797e-01 -3.41583759e-01
-1.15061235e+00 -8.05340350e-01 -6.08639956e-01 1.07088518e+00
-2.42807433e-01 -5.54310858e-01 -6.21430159e-01 -7.08310604e-01
-3.42998713e-01 8.65870416e-01 -3.77796292e-01 1.90565675e-01
-1.29014373e+00 -6.13527834e-01 9.40242708e-01 6.36443555e-01
2.25747153e-01 -1.33784211e+00 -4.57744092e-01 6.49878681e-01
-1.86661348e-01 -1.29049349e+00 -3.97192508e-01 4.95965481e-01
-6.09528542e-01 -9.57536280e-01 -5.17293632e-01 -1.14396572e+00
2.07159176e-01 -3.36762309e-01 1.39772975e+00 -6.73948973e-02
8.40442628e-02 6.79764673e-02 -8.02163064e-01 -4.17863876e-01
-7.90864170e-01 5.74386775e-01 -1.11448385e-01 -7.56326675e-01
6.09203875e-01 -3.48635465e-01 1.19459247e-02 -1.41378422e-03
-8.94601583e-01 -4.11186397e-01 5.04377365e-01 6.37483001e-01
3.00211936e-01 -1.90861404e-01 4.22152489e-01 -1.19463277e+00
1.69430003e-01 -3.94389987e-01 -4.56129253e-01 1.55767128e-01
-3.60895008e-01 3.53882276e-02 6.61216319e-01 -1.50197983e-01
-1.18923819e+00 -2.14578882e-01 -7.44927824e-01 8.72659147e-01
-1.85380712e-01 6.66106045e-01 -4.98227596e-01 4.39324558e-01
5.76493859e-01 -1.59263298e-01 -5.54459631e-01 -8.32262397e-01
3.16120505e-01 1.04550278e+00 9.44144666e-01 -7.45681763e-01
6.06328070e-01 2.32397690e-01 -3.32832843e-01 -7.88102925e-01
-8.92380774e-01 -7.21597791e-01 -1.04216659e+00 2.76473165e-01
8.44712853e-01 -8.97322059e-01 -8.28909129e-02 5.64849377e-01
-1.03425360e+00 -5.56108475e-01 -3.49737436e-01 6.40800297e-01
-2.80903667e-01 6.85367048e-01 -1.10013723e+00 -3.04418743e-01
-2.66716570e-01 -6.04055345e-01 5.88458836e-01 1.41100273e-01
-6.97108984e-01 -1.37264240e+00 4.36035603e-01 1.16513867e-03
-1.86430022e-01 1.87329948e-01 1.13494647e+00 -7.32871652e-01
3.34836125e-01 -1.31988311e-02 3.53272818e-02 4.51594800e-01
2.62552321e-01 -1.50414392e-01 -4.10674274e-01 -3.27712804e-01
-1.51590794e-01 -1.31022364e-01 6.18592918e-01 -6.30943663e-03
8.75625610e-02 1.05482720e-01 -2.08555475e-01 5.83849192e-01
1.56535673e+00 5.10423183e-01 8.52127612e-01 8.68537128e-01
4.52719688e-01 4.68180299e-01 7.62718737e-01 2.06635430e-01
6.44819856e-01 1.95093006e-01 -2.62574464e-01 1.56188160e-01
-1.01373024e-01 -1.54965937e-01 3.98957908e-01 1.45362580e+00
-3.76321636e-02 -2.18211398e-01 -1.34261370e+00 1.23377573e+00
-1.43223667e+00 -5.69449723e-01 -1.72575414e-01 2.13402128e+00
1.19474638e+00 4.12196606e-01 -3.31215374e-02 3.14833164e-01
6.74848795e-01 1.56124130e-01 1.23599194e-01 -8.34431410e-01
-2.55003959e-01 7.57752776e-01 7.95531511e-01 4.35058087e-01
-1.10834324e+00 1.26372707e+00 6.84839678e+00 5.57470024e-01
-9.76457059e-01 4.96468656e-02 1.91786990e-01 2.74995834e-01
-2.90579885e-01 3.57365161e-01 -1.16907847e+00 2.29317918e-01
1.29270506e+00 -1.98134243e-01 8.93503129e-02 3.99803162e-01
-1.00733325e-01 -2.56836325e-01 -8.44332159e-01 6.49871886e-01
1.27071902e-01 -8.96792471e-01 -3.72444123e-01 -1.52111379e-02
7.31163800e-01 3.88849437e-01 -4.45583701e-01 4.69737023e-01
5.61666489e-01 -6.77716494e-01 1.07826102e+00 -1.28692806e-01
1.12910354e+00 -9.79688883e-01 8.02792311e-01 9.41868052e-02
-1.29847562e+00 1.50909767e-01 -3.65124881e-01 -2.61105180e-01
6.49924755e-01 3.99192631e-01 -3.80322248e-01 5.74263036e-01
4.53874946e-01 7.12731242e-01 -3.72052461e-01 7.44357467e-01
-6.31019831e-01 8.87455046e-01 -2.80447841e-01 1.13789015e-01
5.71456194e-01 -2.14762673e-01 3.86616200e-01 1.67065895e+00
1.38889924e-01 2.16672167e-01 -1.47106536e-02 -3.66560578e-01
-4.20447141e-02 4.96480942e-01 -1.53343985e-02 -2.36034587e-01
3.59533399e-01 1.05927157e+00 -1.07459724e+00 -3.29486370e-01
-7.38477886e-01 8.19549620e-01 3.04784030e-01 1.22529142e-01
-5.89795172e-01 -8.99480581e-01 6.65365398e-01 2.87156671e-01
4.05283332e-01 -6.13295972e-01 -1.65107459e-01 -1.13913548e+00
-5.89640178e-02 -1.03008461e+00 7.08211839e-01 -7.37962052e-02
-1.02610767e+00 6.91704512e-01 1.61167428e-01 -5.79918742e-01
-3.07365716e-01 -6.66520059e-01 -2.12650955e-01 8.16873729e-01
-1.44578624e+00 -9.57445145e-01 4.04901236e-01 9.75585952e-02
5.61834812e-01 1.16260596e-01 9.16349769e-01 5.74563801e-01
-3.99861932e-01 5.23822427e-01 3.68608117e-01 6.17338240e-01
9.25094962e-01 -1.28424001e+00 1.18825805e+00 9.56671953e-01
3.12567741e-01 5.77468157e-01 5.84606588e-01 -7.84568250e-01
-8.89327824e-01 -8.65406215e-01 1.57108510e+00 -2.55133927e-01
1.08137083e+00 -3.92794818e-01 -7.09748745e-01 1.08320415e+00
3.26804042e-01 -7.44998574e-01 7.14602351e-01 3.48563612e-01
-2.45687053e-01 1.41208306e-01 -8.20028245e-01 5.23452580e-01
1.18658876e+00 -4.08907443e-01 -9.55487013e-01 7.65863061e-02
2.60522604e-01 -3.37388009e-01 -1.18023312e+00 1.98722929e-01
7.83489406e-01 -5.42956412e-01 3.86762023e-01 -4.54631209e-01
1.09133676e-01 1.29085600e-01 -1.47383347e-01 -1.35766613e+00
-4.01385665e-01 -1.09012854e+00 8.19687128e-01 1.77499020e+00
6.52730763e-01 -5.90441167e-01 4.22560781e-01 3.98587972e-01
-4.53087360e-01 -2.28100598e-01 -8.92921448e-01 -6.36858523e-01
5.79599619e-01 -6.03702307e-01 5.03789365e-01 9.85240996e-01
3.21719527e-01 4.48323548e-01 1.08798310e-01 -2.85967141e-01
3.83559138e-01 -9.61702093e-02 4.78870422e-01 -1.34499097e+00
-2.76949435e-01 -3.27007473e-01 -5.03107131e-01 -1.03220367e+00
2.57866085e-01 -9.34208691e-01 2.21678913e-01 -1.56630707e+00
-1.50420561e-01 -6.69027030e-01 -1.33891299e-01 5.01167119e-01
1.79433897e-02 4.00488406e-01 1.25091255e-01 2.37819970e-01
-3.25230509e-01 -1.83746621e-01 7.31314242e-01 1.95234299e-01
-2.46391594e-01 -3.03576648e-01 -8.31015170e-01 9.81947839e-01
9.30105448e-01 -6.94966078e-01 6.82950318e-02 -7.85328925e-01
6.38700008e-01 -1.21884458e-01 -6.48409963e-01 -8.51883650e-01
-1.55194849e-01 -1.95142508e-01 8.63887370e-02 -4.82230097e-01
-7.66841844e-02 -3.07226866e-01 1.23413488e-01 2.48374328e-01
6.53436184e-02 4.38442707e-01 4.33351904e-01 -6.61854297e-02
-2.68767148e-01 -6.43244684e-01 8.92253280e-01 -4.12731379e-01
-9.82364297e-01 -6.88634589e-02 -7.49850869e-01 8.55231643e-01
8.22065294e-01 -9.12275761e-02 1.25631569e-02 1.45894721e-01
-6.45428836e-01 1.03869438e-01 9.04650450e-01 2.48295009e-01
-1.75937191e-01 -9.37985420e-01 -8.51057947e-01 1.35110319e-01
3.70097645e-02 -2.39854425e-01 -1.82951912e-01 3.97270203e-01
-1.02079511e+00 4.53325450e-01 -2.23253593e-01 -6.83084056e-02
-1.13892388e+00 4.05105948e-01 -9.78007540e-02 -6.23466909e-01
-6.90514386e-01 6.06517732e-01 -9.19427946e-02 -2.57398427e-01
-1.61062047e-01 -2.40597904e-01 -3.48328322e-01 3.03625762e-01
3.99321735e-01 1.31950870e-01 9.43252817e-02 -1.14696264e+00
-5.43094873e-01 5.91424942e-01 -1.86924845e-01 -5.00761211e-01
1.33499002e+00 -3.38586479e-01 -1.34283200e-01 6.29219413e-01
1.07669187e+00 6.71414912e-01 -6.65617406e-01 -9.06904414e-02
5.42354167e-01 -2.12825000e-01 -2.94336408e-01 -5.28315961e-01
-6.38505876e-01 5.64764440e-01 -9.65744704e-02 1.71448603e-01
9.99165714e-01 2.16137111e-01 1.16245580e+00 1.98092476e-01
3.33440870e-01 -1.36397696e+00 -7.57613957e-01 1.03163099e+00
2.95945853e-01 -6.85523093e-01 5.51338047e-02 -2.83687234e-01
-6.73763037e-01 1.01973963e+00 -1.86756626e-01 -6.33634068e-03
5.82031369e-01 5.70690751e-01 5.94496369e-01 -4.52356189e-02
-3.21664631e-01 -4.53235120e-01 -7.55665302e-02 4.52864856e-01
8.38615239e-01 8.34678039e-02 -1.12676895e+00 5.63297272e-01
-6.03800833e-01 -3.61082584e-01 6.02449954e-01 1.12394059e+00
-3.69659543e-01 -1.85978973e+00 -3.16539019e-01 1.19317055e-01
-1.15385437e+00 -4.38556015e-01 -1.11709185e-01 1.24146974e+00
2.46121496e-01 8.94232333e-01 2.50778645e-01 -6.27318993e-02
3.23203355e-01 3.25881094e-01 5.30425310e-01 -9.57765520e-01
-8.94892752e-01 5.02458394e-01 6.87502921e-01 -3.43173780e-02
-5.45533419e-01 -1.32623672e+00 -1.46067774e+00 -2.89806098e-01
-3.06210458e-01 6.23234510e-01 6.75627887e-01 1.10648298e+00
-3.95506591e-01 1.96039632e-01 1.90033004e-01 -6.50301754e-01
-2.02838331e-01 -1.05986679e+00 -9.18315530e-01 3.67667854e-01
-3.01849931e-01 -2.33001411e-01 -4.06442821e-01 2.61857659e-01]
|
[10.402688980102539, 10.124384880065918]
|
b669fd9c-50bd-46fa-8e9b-f184cc5dbd3e
|
efficient-multi-grained-knowledge-reuse-for
|
2306.02027
| null |
https://arxiv.org/abs/2306.02027v1
|
https://arxiv.org/pdf/2306.02027v1.pdf
|
Efficient Multi-Grained Knowledge Reuse for Class Incremental Segmentation
|
Class Incremental Semantic Segmentation (CISS) has been a trend recently due to its great significance in real-world applications. Although the existing CISS methods demonstrate remarkable performance, they either leverage the high-level knowledge (feature) only while neglecting the rich and diverse knowledge in the low-level features, leading to poor old knowledge preservation and weak new knowledge exploration; or use multi-level features for knowledge distillation by retraining a heavy backbone, which is computationally intensive. In this paper, we for the first time propose to efficiently reuse the multi-grained knowledge for CISS by fusing multi-level features with the frozen backbone and show a simple aggregation of varying-level features, i.e., naive feature pyramid, can boost the performance significantly. We further introduce a novel densely-interactive feature pyramid (DEFY) module that enhances the fusion of high- and low-level features by enabling their dense interaction. Specifically, DEFY establishes a per-pixel relationship between pairs of feature maps, allowing for multi-pair outputs to be aggregated. This results in improved semantic segmentation by leveraging the complementary information from multi-level features. We show that DEFY can be effortlessly integrated into three representative methods for performance enhancement. Our method yields a new state-of-the-art performance when combined with the current SOTA by notably averaged mIoU gains on two widely used benchmarks, i.e., 2.5% on PASCAL VOC 2012 and 2.3% on ADE20K.
|
['Xinchao Wang', 'Shuicheng Yan', 'Zhihe Lu']
|
2023-06-03
| null | null | null | null |
['class-incremental-semantic-segmentation']
|
['computer-vision']
|
[ 3.80786061e-01 -2.98681529e-03 -1.14691675e-01 -3.70105535e-01
-8.29366148e-01 -4.89626586e-01 4.75339860e-01 1.94002420e-01
-6.49609447e-01 5.22825778e-01 -9.36734006e-02 5.35421148e-02
-1.61094218e-01 -8.79977047e-01 -8.76606882e-01 -6.29207194e-01
1.27631247e-01 5.14958845e-03 1.07230914e+00 -2.84158528e-01
8.35509896e-02 3.73473883e-01 -1.94063437e+00 3.43509406e-01
1.27040017e+00 1.34386885e+00 4.19823200e-01 2.61905849e-01
-2.27078363e-01 4.05751228e-01 -3.86930257e-01 -2.57634968e-01
2.71214604e-01 -3.93036269e-02 -8.48490000e-01 -9.50364396e-02
4.74443406e-01 -3.31301689e-01 -1.18127398e-01 1.01906514e+00
2.81736404e-01 8.34361240e-02 1.36533275e-01 -1.01061678e+00
-2.80087382e-01 4.12264675e-01 -8.18529367e-01 1.19400527e-02
1.07494339e-01 2.02953845e-01 9.83159602e-01 -8.27341199e-01
6.06479228e-01 1.12117815e+00 6.63740218e-01 1.40151680e-01
-1.11500764e+00 -6.57297969e-01 3.91415417e-01 6.02480352e-01
-1.42098331e+00 -1.46335021e-01 7.72669971e-01 -2.23750234e-01
9.37971890e-01 2.46637776e-01 6.91689372e-01 5.60258210e-01
-1.04498826e-01 9.75788176e-01 1.40269661e+00 -3.57027382e-01
1.69583514e-01 5.11928797e-02 4.34410810e-01 9.09731865e-01
3.12616676e-01 -6.08180389e-02 -7.94216216e-01 2.09153995e-01
5.37512124e-01 1.40782595e-01 -2.50658095e-01 -3.90997082e-01
-1.12409627e+00 7.29574084e-01 9.21507120e-01 3.48837286e-01
-4.42650050e-01 4.30983566e-02 2.90814012e-01 7.25329742e-02
2.51414925e-01 3.74865711e-01 -7.10409641e-01 -2.09657043e-01
-1.18140447e+00 1.50324225e-01 5.43309748e-01 7.36881793e-01
1.32847595e+00 -2.36358225e-01 -3.84044498e-01 9.07977581e-01
-6.10587746e-02 3.27158213e-01 3.82623285e-01 -1.07821190e+00
3.19661289e-01 1.01460421e+00 -3.02846972e-02 -7.64758110e-01
-4.41288084e-01 -5.83770216e-01 -6.53284907e-01 1.99119419e-01
3.20904583e-01 7.56511614e-02 -1.36851847e+00 1.67917943e+00
5.52823484e-01 3.28830093e-01 1.33298179e-02 6.50494397e-01
7.42798686e-01 5.09920359e-01 -7.52835348e-02 4.12367061e-02
1.46811271e+00 -1.37147951e+00 -3.90541881e-01 -1.31631270e-01
4.69199032e-01 -4.98787940e-01 1.18094814e+00 3.67827892e-01
-7.92230427e-01 -6.92279816e-01 -1.17250025e+00 -1.57008722e-01
-6.93343759e-01 -9.60031375e-02 8.28301489e-01 5.64029872e-01
-1.02846575e+00 8.22511494e-01 -9.82808232e-01 -2.15261251e-01
7.66629219e-01 6.38436913e-01 -3.47975403e-01 -2.94910818e-01
-1.19327676e+00 5.12383521e-01 5.99179864e-01 4.29328792e-02
-4.70225006e-01 -1.04392695e+00 -8.22851300e-01 1.49837479e-01
9.64061975e-01 -6.52698815e-01 1.03327274e+00 -8.32378864e-01
-1.54144335e+00 4.21230584e-01 -1.44594133e-01 -3.36636096e-01
3.76845419e-01 -5.64656556e-01 -2.22876891e-01 3.69318008e-01
1.57807589e-01 1.12580466e+00 7.78320491e-01 -1.28478909e+00
-9.37202334e-01 -4.24373001e-01 2.70897835e-01 2.70175725e-01
-7.46668577e-01 -3.22139531e-01 -1.02342665e+00 -6.10871315e-01
2.29208618e-01 -9.03257668e-01 -2.43253350e-01 -1.71396732e-02
-3.59969258e-01 -4.68360782e-02 1.01210570e+00 -5.35108864e-01
1.24993396e+00 -2.21388578e+00 5.24910986e-02 1.48165703e-01
2.35476524e-01 6.34940267e-01 -4.06277962e-02 -2.08457536e-03
3.28928173e-01 -2.79222578e-02 -5.30997217e-01 -1.91521659e-01
-8.55738968e-02 3.10377598e-01 7.47327134e-02 1.70335686e-03
3.86042982e-01 1.19682729e+00 -9.86359954e-01 -4.61018473e-01
4.27823275e-01 5.19868433e-01 -7.94587135e-01 -1.70009658e-01
-1.17537849e-01 3.75039667e-01 -5.90190291e-01 7.72817492e-01
7.78450549e-01 -3.03255588e-01 -1.77396014e-01 -5.01428127e-01
-2.31629923e-01 -8.17530379e-02 -1.24679196e+00 1.90402055e+00
-3.53086144e-01 1.60724714e-01 -1.25372140e-02 -7.20848739e-01
6.57133758e-01 -2.49254152e-01 3.78152311e-01 -8.05579543e-01
-4.94705066e-02 3.48615229e-01 -1.16904840e-01 -7.49263465e-02
6.78464413e-01 1.41116440e-01 -1.83301285e-01 -8.19631293e-02
2.62156606e-01 -1.91081032e-01 2.40384340e-01 2.36732110e-01
1.07458770e+00 3.29618186e-01 1.72960550e-01 -3.42383921e-01
5.86484969e-01 -5.23491390e-02 7.29164124e-01 8.01314056e-01
-1.35103494e-01 4.90010887e-01 3.30026388e-01 -1.53522030e-01
-5.30101061e-01 -1.13293529e+00 -1.71871230e-01 1.04813278e+00
4.99121994e-01 -5.39621770e-01 -8.60981882e-01 -8.90809715e-01
2.15731874e-01 5.23299754e-01 -6.31179392e-01 -2.58841962e-01
-5.07624865e-01 -7.92899072e-01 3.42028916e-01 8.15867305e-01
1.05977595e+00 -8.55075479e-01 -8.16549778e-01 2.12215602e-01
1.20635517e-01 -1.29410553e+00 -2.95395464e-01 3.99851054e-01
-8.33810568e-01 -7.86790133e-01 -7.98444629e-01 -4.44389075e-01
4.71677303e-01 2.69523412e-01 8.90284657e-01 -1.28307911e-02
-4.07020360e-01 2.17284396e-01 -5.88863611e-01 -2.21414536e-01
1.37029380e-01 3.84675205e-01 -2.94616103e-01 1.18570127e-01
1.05529204e-01 -6.65162921e-01 -8.05721998e-01 2.97502786e-01
-9.40768957e-01 3.48605037e-01 8.20672214e-01 1.01774764e+00
7.40305960e-01 7.75350779e-02 5.79632521e-01 -1.03770566e+00
-5.87418638e-02 -2.23140284e-01 -4.70439434e-01 1.78012028e-01
-6.92839801e-01 2.18002319e-01 5.62882006e-01 -1.01135127e-01
-1.23141587e+00 2.20121309e-01 -1.74169898e-01 -3.55910063e-01
-1.10171512e-02 3.24758351e-01 -2.81916291e-01 -3.35661113e-01
3.85910422e-01 2.52911955e-01 -2.37242967e-01 -6.26013458e-01
6.30121350e-01 5.25893807e-01 7.84107208e-01 -7.09282637e-01
7.06275821e-01 5.77727079e-01 -4.20884527e-02 -6.26112401e-01
-9.78163660e-01 -6.47245288e-01 -7.44790077e-01 2.99009448e-03
7.66519189e-01 -1.01464665e+00 -5.04258573e-01 9.47956145e-01
-8.36521804e-01 -2.73403376e-01 -5.61603069e-01 1.47304371e-01
-2.97998369e-01 4.13192540e-01 -5.31199634e-01 -3.32722872e-01
-2.93691933e-01 -1.31290376e+00 1.24709451e+00 5.96995234e-01
1.44555762e-01 -5.60580134e-01 -4.05167013e-01 4.96891558e-01
4.23157781e-01 3.59242558e-01 8.21495950e-01 -3.92885894e-01
-7.12507784e-01 2.40731519e-02 -7.26639509e-01 5.59284151e-01
1.74193189e-01 -2.34253377e-01 -1.23373687e+00 -2.06858873e-01
-3.79225999e-01 -3.30720127e-01 1.41594970e+00 1.66144490e-01
1.35185051e+00 2.76192576e-01 -4.51183826e-01 6.53046608e-01
1.48733664e+00 9.02535766e-02 4.80511457e-01 2.93132067e-01
1.04220903e+00 4.42009628e-01 7.11476147e-01 4.21031624e-01
5.57313740e-01 7.35391080e-01 3.69310707e-01 -3.75630409e-01
-3.90988320e-01 -8.64821579e-03 1.78407341e-01 7.95182586e-01
-6.80609792e-02 1.89579412e-01 -7.69238710e-01 6.19245052e-01
-1.99674714e+00 -4.47373509e-01 2.77434886e-02 2.06931520e+00
1.04793060e+00 3.95188779e-01 -7.83064216e-02 2.74013937e-01
6.01563990e-01 2.26075202e-01 -8.79896224e-01 -1.46318421e-01
-1.68217108e-01 6.59558952e-01 7.15554237e-01 3.85006100e-01
-1.15441394e+00 1.19516563e+00 5.36824560e+00 1.24470806e+00
-1.00178587e+00 1.81119949e-01 5.29794335e-01 -5.72840199e-02
-2.97130644e-01 3.37390751e-02 -9.12746787e-01 4.27178830e-01
5.37078083e-01 -5.07357344e-02 1.78340778e-01 7.20322192e-01
-2.25200519e-01 -5.72200358e-01 -8.44494104e-01 8.43391895e-01
-2.02162966e-01 -1.32344115e+00 8.77814218e-02 -1.08009391e-01
9.84655857e-01 4.23090868e-02 -5.25680333e-02 4.81205553e-01
2.11483091e-01 -7.06799388e-01 8.30537975e-01 4.19015586e-01
8.94280016e-01 -8.78650844e-01 7.39000320e-01 1.00460917e-01
-1.50905609e+00 -1.84518531e-01 9.43883415e-03 4.95286584e-02
1.59187391e-01 9.15808022e-01 -3.09230924e-01 1.08640516e+00
9.66788948e-01 8.15995395e-01 -8.24204504e-01 9.31931973e-01
-6.73353225e-02 6.23852730e-01 -5.86353302e-01 3.52746964e-01
6.04944527e-01 8.52560997e-02 2.82315493e-01 1.26731873e+00
1.62354857e-01 -5.67598790e-02 2.72021055e-01 6.47587538e-01
-9.89125594e-02 3.69394161e-02 -1.38087934e-02 2.56775349e-01
4.06517953e-01 1.44380140e+00 -9.15781379e-01 -6.11068785e-01
-5.68122447e-01 1.24215055e+00 3.20402980e-01 1.79649964e-01
-7.77144194e-01 -6.30849719e-01 6.34008110e-01 3.72694209e-02
7.62945116e-01 -5.85236549e-02 -4.01244044e-01 -1.02709115e+00
3.24818581e-01 -4.28882778e-01 4.29356903e-01 -4.48755562e-01
-9.47833538e-01 5.71530819e-01 9.01653171e-02 -7.68840790e-01
8.08273479e-02 -5.52223265e-01 -2.59096563e-01 7.90092349e-01
-2.16447616e+00 -1.32378685e+00 -6.00512862e-01 7.26873875e-01
4.88741934e-01 2.33090714e-01 5.95180571e-01 4.35959995e-01
-5.99766731e-01 7.84046948e-01 1.43782571e-01 -2.60560453e-01
4.81437504e-01 -1.40052223e+00 3.44670355e-01 7.08888590e-01
7.75297061e-02 3.57588649e-01 2.23249346e-01 -4.54927713e-01
-1.21349716e+00 -1.22560573e+00 5.18516600e-01 -1.07695870e-01
5.07823288e-01 -2.94749737e-01 -1.03627014e+00 1.84989914e-01
-1.79152951e-01 4.21918213e-01 2.80298829e-01 3.95870768e-02
-4.73873287e-01 -4.17798787e-01 -1.20365298e+00 4.97918576e-01
1.28891480e+00 -3.21322799e-01 -4.82507050e-01 -1.10788353e-01
1.02660918e+00 -3.44136268e-01 -8.37004900e-01 8.79155934e-01
5.67312181e-01 -1.14985597e+00 9.43661809e-01 -2.96441074e-02
2.00051472e-01 -5.76666296e-01 -2.88666427e-01 -1.01890409e+00
-3.31894368e-01 -5.52502453e-01 -1.43363059e-01 1.36945927e+00
2.38160223e-01 -7.66666532e-01 8.02860498e-01 3.19803238e-01
-3.74429196e-01 -1.08324897e+00 -1.07450736e+00 -9.31050241e-01
-6.76242858e-02 -4.10596132e-01 5.60868442e-01 6.76983237e-01
-3.00190598e-01 1.78407535e-01 -6.66028708e-02 1.07563339e-01
4.25245702e-01 3.19631845e-01 5.41924834e-01 -1.26808739e+00
-3.65715444e-01 -4.66901600e-01 -6.15492463e-01 -1.17300189e+00
6.59037894e-03 -8.48740160e-01 3.55758928e-02 -1.43193412e+00
4.08373892e-01 -5.86300075e-01 -6.93411350e-01 8.63266528e-01
-5.88858962e-01 5.65747917e-01 4.76539046e-01 -2.31816601e-02
-8.19931030e-01 6.29889429e-01 1.31131172e+00 1.44617306e-03
-3.82892311e-01 -4.73875374e-01 -8.28258634e-01 8.00136805e-01
6.05475247e-01 -2.72683859e-01 -3.61889929e-01 -2.81905293e-01
-2.56560713e-01 -4.27767009e-01 4.63399440e-01 -1.34776890e+00
2.80500978e-01 1.40209431e-02 3.19784075e-01 -5.75735569e-01
4.02144164e-01 -7.53509939e-01 -8.50268379e-02 4.01777983e-01
2.09480166e-01 -3.94028217e-01 5.51805794e-01 6.09859705e-01
-4.42612916e-01 -1.42882122e-02 8.68294239e-01 -2.01927181e-02
-1.39770103e+00 1.97145998e-01 2.53529072e-01 2.74153687e-02
1.12594879e+00 -5.18116474e-01 -1.15410194e-01 2.15311438e-01
-5.44119477e-01 3.45772177e-01 4.50153649e-01 3.68568152e-01
3.85105520e-01 -1.13572574e+00 -4.75349218e-01 1.95366651e-01
2.08399996e-01 3.52565974e-01 5.78958035e-01 8.94803822e-01
-2.38704503e-01 1.94557637e-01 -2.83267975e-01 -8.73720527e-01
-1.14793301e+00 3.70211273e-01 -2.36963369e-02 -5.52483201e-01
-6.61143363e-01 9.68621731e-01 4.97780621e-01 -1.88288361e-01
-3.64162475e-02 -4.30868298e-01 -3.75654288e-02 2.39969686e-01
4.19929475e-01 4.73910093e-01 2.97450304e-01 -5.12696683e-01
-5.43044329e-01 9.20361876e-01 -4.73614007e-01 1.74165010e-01
1.38060069e+00 -9.79584977e-02 3.94537300e-02 3.11516464e-01
1.25030398e+00 -1.34690523e-01 -1.64693236e+00 -5.21252334e-01
6.27923608e-02 -4.16378886e-01 3.09622854e-01 -1.06398737e+00
-1.33856928e+00 7.24540710e-01 5.65524995e-01 -2.54327416e-01
1.55894470e+00 1.20744035e-01 1.08286679e+00 2.51854718e-01
5.20497501e-01 -1.19344139e+00 9.02000815e-02 3.89356941e-01
5.20455837e-01 -1.16706765e+00 5.67081794e-02 -9.63474393e-01
-5.52160382e-01 8.24103117e-01 5.28471112e-01 -3.63568813e-02
6.30944133e-01 3.95517200e-01 -1.52745098e-01 -1.39439747e-01
-3.65342259e-01 -5.35704553e-01 3.63965034e-01 3.83189082e-01
-5.12000769e-02 7.23571479e-02 -2.63061881e-01 8.88224304e-01
-2.06689304e-03 1.93924248e-01 4.04354818e-02 9.98467386e-01
-5.52378833e-01 -1.12822425e+00 -1.81712568e-01 6.70972109e-01
-2.45400742e-01 -2.55711913e-01 2.78995037e-02 7.46985793e-01
5.68107903e-01 7.45741010e-01 -1.96104981e-02 -5.23900449e-01
5.20632446e-01 2.05016024e-02 4.29486573e-01 -4.51880604e-01
-6.35510504e-01 -1.00727282e-01 -1.74067348e-01 -9.48970377e-01
-4.17723387e-01 -7.51031458e-01 -1.50329709e+00 -7.19918609e-02
-4.06235456e-01 -1.51389092e-01 4.23447043e-01 9.68093932e-01
7.07198441e-01 7.89333880e-01 4.37428057e-01 -9.21109378e-01
-3.12569320e-01 -6.37537599e-01 -5.30991137e-01 3.84850770e-01
1.99519575e-01 -1.03630126e+00 -9.73906368e-02 2.81777885e-02]
|
[9.642404556274414, 0.17618820071220398]
|
d47905d7-e92b-4396-819e-53e3d124fa92
|
algebraic-learning-towards-interpretable
|
2203.06690
| null |
https://arxiv.org/abs/2203.06690v1
|
https://arxiv.org/pdf/2203.06690v1.pdf
|
Algebraic Learning: Towards Interpretable Information Modeling
|
Along with the proliferation of digital data collected using sensor technologies and a boost of computing power, Deep Learning (DL) based approaches have drawn enormous attention in the past decade due to their impressive performance in extracting complex relations from raw data and representing valuable information. Meanwhile, though, rooted in its notorious black-box nature, the appreciation of DL has been highly debated due to the lack of interpretability. On the one hand, DL only utilizes statistical features contained in raw data while ignoring human knowledge of the underlying system, which results in both data inefficiency and trust issues; on the other hand, a trained DL model does not provide to researchers any extra insight about the underlying system beyond its output, which, however, is the essence of most fields of science, e.g. physics and economics. This thesis addresses the issue of interpretability in general information modeling and endeavors to ease the problem from two scopes. Firstly, a problem-oriented perspective is applied to incorporate knowledge into modeling practice, where interesting mathematical properties emerge naturally which cast constraints on modeling. Secondly, given a trained model, various methods could be applied to extract further insights about the underlying system. These two pathways are termed as guided model design and secondary measurements. Remarkably, a novel scheme emerges for the modeling practice in statistical learning: Algebraic Learning (AgLr). Instead of being restricted to the discussion of any specific model, AgLr starts from idiosyncrasies of a learning task itself and studies the structure of a legitimate model class. This novel scheme demonstrates the noteworthy value of abstract algebra for general AI, which has been overlooked in recent progress, and could shed further light on interpretable information modeling.
|
['Tong Owen Yang']
|
2022-03-13
| null | null | null | null |
['abstract-algebra']
|
['reasoning']
|
[ 2.69016683e-01 6.01133704e-01 -3.44257504e-01 -2.66501009e-01
-2.11812973e-01 -3.90217096e-01 6.72161639e-01 4.18396413e-01
-6.25367556e-03 7.18427360e-01 -1.21045321e-01 -5.09532213e-01
-6.63551927e-01 -7.31560171e-01 -5.81496179e-01 -7.87161946e-01
-1.27370238e-01 1.68952733e-01 -3.61021876e-01 -2.56400019e-01
1.54831767e-01 7.29330957e-01 -1.36584783e+00 -7.15431497e-02
7.14890361e-01 1.07981288e+00 -4.51304093e-02 2.46807307e-01
-8.97917375e-02 9.11686659e-01 -3.27250421e-01 -5.03898561e-01
2.14650467e-01 -4.13468540e-01 -7.20999360e-01 1.25694215e-01
-1.64851800e-01 -6.76018968e-02 -3.80745232e-01 9.14604247e-01
3.34717445e-02 -3.23472381e-01 5.47098815e-01 -1.45555294e+00
-4.28331852e-01 4.82289284e-01 -3.21564347e-01 -1.55567288e-01
2.04096213e-01 7.62101114e-02 1.23618078e+00 -6.33143485e-01
2.13985071e-01 1.06951261e+00 4.77849364e-01 1.65500581e-01
-1.43337739e+00 -4.33358520e-01 1.81685507e-01 1.27451956e-01
-1.24307966e+00 -3.98143679e-01 9.66075063e-01 -5.31259179e-01
5.20835161e-01 3.49999428e-01 7.52010584e-01 1.02285314e+00
3.19192231e-01 8.85900974e-01 1.22064447e+00 -5.99535584e-01
4.16380525e-01 5.13434231e-01 4.00102139e-01 6.95336521e-01
5.91900826e-01 2.15005785e-01 -5.86902738e-01 6.67679235e-02
7.48133361e-01 -1.08810579e-02 -1.68651968e-01 -5.05499363e-01
-9.66543913e-01 7.50203431e-01 3.07466626e-01 4.14663523e-01
-2.94389576e-01 -7.01413304e-02 3.29594582e-01 4.90745336e-01
2.53111720e-01 8.68639767e-01 -6.29472852e-01 -1.08243994e-01
-6.34256124e-01 1.64647117e-01 1.07776296e+00 6.17906809e-01
8.75792384e-01 1.60199612e-01 4.61258113e-01 2.13442639e-01
2.49072403e-01 2.19257280e-01 8.19484890e-02 -9.51220334e-01
2.55138874e-01 9.71888602e-01 -8.49428624e-02 -1.19270170e+00
-4.26522255e-01 -7.78450847e-01 -1.09887791e+00 3.19950491e-01
5.89842319e-01 -1.78930476e-01 -4.21186090e-01 1.78331769e+00
1.43333524e-01 -3.02213222e-01 7.68032819e-02 8.15784276e-01
4.96149570e-01 6.26886129e-01 -1.84112027e-01 -3.53434294e-01
1.07008493e+00 -4.20260042e-01 -6.45735919e-01 -1.01908229e-01
6.35466158e-01 -2.65213698e-01 9.69619274e-01 8.24215889e-01
-1.04434478e+00 -5.11214018e-01 -1.15021145e+00 -6.86817542e-02
-4.75483775e-01 -1.52604654e-01 1.17043424e+00 5.97616196e-01
-7.89487064e-01 5.83543956e-01 -9.40640867e-01 -3.21198285e-01
4.43928003e-01 5.21536350e-01 -2.37647757e-01 8.34100619e-02
-1.12006021e+00 1.10296357e+00 4.32659775e-01 4.21228945e-01
-5.44045508e-01 -7.46102393e-01 -7.93597639e-01 2.17568934e-01
7.95573473e-01 -6.73309267e-01 9.25635219e-01 -1.09558678e+00
-1.42376614e+00 4.47512656e-01 -3.38756174e-01 -4.45440918e-01
5.12310505e-01 -1.60733894e-01 -4.26146060e-01 -1.56637281e-01
-3.43641132e-01 4.69787046e-02 6.98730469e-01 -1.48311007e+00
-2.30124667e-01 -4.75753725e-01 5.14373541e-01 -1.92277372e-01
-2.82402515e-01 -3.50756139e-01 6.38924241e-02 -3.94548357e-01
3.25848728e-01 -6.54710770e-01 -4.98802572e-01 2.47102361e-02
-4.98266160e-01 -1.18041947e-01 6.80336475e-01 -3.36950213e-01
1.47397017e+00 -1.94979167e+00 2.47871563e-01 4.72858340e-01
5.90145230e-01 2.61333019e-01 2.91927874e-01 8.63975763e-01
-3.57874423e-01 3.04707676e-01 -2.16590837e-01 2.00201161e-02
1.68786496e-01 3.11621219e-01 -4.67319727e-01 2.13629171e-01
4.81165558e-01 9.97398615e-01 -8.47927034e-01 -3.02558511e-01
4.03221965e-01 2.27561012e-01 -3.20803761e-01 1.68885544e-01
-1.65225968e-01 5.21125555e-01 -6.97097719e-01 5.66252291e-01
3.75293672e-01 -4.13139194e-01 3.49004745e-01 -3.02540451e-01
-1.11078985e-01 1.59343272e-01 -1.35595179e+00 1.36230350e+00
-4.02563989e-01 6.48231983e-01 -8.93673748e-02 -1.67080998e+00
8.53856087e-01 3.24207336e-01 5.52798271e-01 -6.24911189e-01
1.45387292e-01 9.06847939e-02 2.63328880e-01 -6.99707329e-01
3.14134657e-01 -2.39240214e-01 -4.76743095e-02 5.17930448e-01
-1.62717164e-01 -1.30202740e-01 -9.83197168e-02 6.82421997e-02
9.10192966e-01 1.95639774e-01 7.08437681e-01 -3.56442302e-01
5.38643718e-01 -9.48336199e-02 4.65373427e-01 7.25920379e-01
1.09474055e-01 6.10845126e-02 6.59905970e-01 -6.48993134e-01
-1.03371644e+00 -1.11833024e+00 -4.62810904e-01 6.35918617e-01
4.87204529e-02 -5.35348296e-01 -5.98119974e-01 -3.20224047e-01
-4.13081497e-02 6.66886210e-01 -5.58201253e-01 -2.02771321e-01
-4.31524992e-01 -7.87501514e-01 2.72057652e-01 4.73760337e-01
4.76351529e-01 -7.26920784e-01 -6.39424801e-01 2.08333239e-01
3.67812961e-02 -9.74497497e-01 4.67960596e-01 4.20012921e-01
-1.09133267e+00 -1.09794831e+00 -1.42316401e-01 -2.11249456e-01
6.21744275e-01 5.27138673e-02 1.11178815e+00 1.67623803e-01
-1.80993825e-01 3.94955158e-01 -1.15154110e-01 -7.77467072e-01
-5.12421846e-01 3.08080036e-02 2.03221351e-01 1.12437727e-02
5.16199052e-01 -8.04820180e-01 -1.83323458e-01 1.12696635e-02
-1.02050662e+00 1.99552581e-01 8.71203959e-01 7.33655632e-01
4.28052880e-02 3.89940470e-01 7.47030139e-01 -7.84111440e-01
5.93918443e-01 -5.53410828e-01 -4.56232041e-01 4.87395167e-01
-9.18398976e-01 3.96242112e-01 7.83643901e-01 -1.97412819e-01
-9.40259516e-01 -2.24652469e-01 1.40591606e-01 4.55208533e-02
-2.56019562e-01 8.76621306e-01 -5.10464728e-01 2.02970281e-01
4.56326067e-01 1.77492365e-01 2.01222420e-01 -4.05272543e-01
2.18863830e-01 5.85819900e-01 2.83281624e-01 -8.73217225e-01
9.25049722e-01 3.60213608e-01 4.90572453e-01 -1.11877227e+00
-1.01624620e+00 -4.75288332e-02 -9.92563069e-01 -5.02630696e-02
5.33377588e-01 -5.95986903e-01 -9.34813797e-01 1.03424773e-01
-1.10085964e+00 1.19718499e-01 -5.44580936e-01 4.46431220e-01
-4.72397208e-01 3.35533023e-01 -3.30932349e-01 -1.18164885e+00
1.07805818e-01 -1.07489371e+00 7.40491331e-01 1.07983477e-01
-5.60280979e-01 -1.32202017e+00 -3.21872324e-01 3.51765275e-01
1.28449708e-01 4.43158090e-01 1.41736257e+00 -6.60579383e-01
-8.89006019e-01 -4.46959645e-01 -2.18299508e-01 5.08006394e-01
1.61245555e-01 -1.49784356e-01 -1.21919453e+00 -1.46166191e-01
3.60569954e-01 -3.11958730e-01 3.30362827e-01 1.59963012e-01
1.24378562e+00 -3.91083658e-01 -1.99761927e-01 2.51032084e-01
1.38979757e+00 2.11934268e-01 5.20169973e-01 1.63645267e-01
6.09594405e-01 8.62186193e-01 2.31688783e-01 4.08165187e-01
2.90591091e-01 5.78219771e-01 5.56658745e-01 -2.84880370e-01
3.31180900e-01 -3.04155499e-01 1.04518138e-01 8.96798491e-01
-4.46909666e-01 -8.33696499e-03 -7.32154965e-01 1.26919135e-01
-1.86731315e+00 -9.08436298e-01 -2.30867341e-01 2.40997934e+00
7.07952201e-01 4.24728811e-01 -5.79470247e-02 6.35637462e-01
2.25658968e-01 6.09904947e-03 -7.67772377e-01 -5.13273537e-01
-1.42238960e-01 1.99846163e-01 1.37572259e-01 3.79099697e-01
-6.31036520e-01 4.58999902e-01 6.26680660e+00 7.63317823e-01
-1.01280546e+00 -2.69820780e-01 6.85782015e-01 4.52382803e-01
-3.82298082e-01 2.48371184e-01 -6.49773180e-01 2.03837305e-01
9.60391402e-01 -3.74376804e-01 4.76337075e-01 7.40310371e-01
6.09999478e-01 -2.40193635e-01 -1.67571175e+00 8.19375575e-01
-2.67839193e-01 -1.07848561e+00 9.07400250e-02 5.14303267e-01
4.10033017e-01 -4.80991453e-01 -4.54209074e-02 1.85588777e-01
8.33335668e-02 -1.32207954e+00 6.66914523e-01 7.44190693e-01
3.15125972e-01 -5.44628918e-01 7.54014552e-01 7.20196426e-01
-8.50036085e-01 -1.62346005e-01 -2.43428051e-01 -7.74852812e-01
-1.27164260e-01 8.98765922e-01 -7.26477444e-01 8.33097458e-01
2.33833358e-01 5.97873747e-01 -5.11722088e-01 6.98513329e-01
-6.86792433e-02 7.59305954e-01 -2.71475583e-01 -3.76434103e-02
2.07270399e-01 -3.23397756e-01 4.22044069e-01 9.44057882e-01
-6.46229759e-02 1.53667912e-01 -1.35115199e-02 1.22926664e+00
2.91100830e-01 -1.78916425e-01 -8.27754557e-01 -2.19937667e-01
2.17979729e-01 1.18660355e+00 -4.06782925e-01 -9.47589725e-02
-5.60156286e-01 2.73835719e-01 2.23586768e-01 4.02929068e-01
-5.61819851e-01 3.62298898e-02 4.68057454e-01 2.44017333e-01
-2.57419646e-01 -5.25092721e-01 -8.24754894e-01 -1.25491393e+00
2.16867730e-01 -1.02205467e+00 -7.09599536e-03 -4.24498022e-01
-1.24025261e+00 1.70858353e-01 2.56799400e-01 -1.06454849e+00
-4.67544973e-01 -7.53674686e-01 -3.95395607e-01 8.78163457e-01
-1.25897193e+00 -1.05619264e+00 -1.81148365e-01 3.00803125e-01
3.55849296e-01 3.72499935e-02 9.83025312e-01 -7.88443834e-02
-6.04826987e-01 3.90835196e-01 2.03930408e-01 3.03553529e-02
8.56516659e-02 -1.27421260e+00 1.21385761e-01 6.34380758e-01
2.06895679e-01 8.62581968e-01 8.85819256e-01 -7.12505430e-02
-1.79738629e+00 -5.97310722e-01 1.03000140e+00 -7.09595978e-01
9.66150701e-01 -5.44582427e-01 -1.00767469e+00 5.51382840e-01
-1.83960870e-02 -2.70386606e-01 6.85573995e-01 3.87022674e-01
-2.09576964e-01 -3.02593768e-01 -8.88291061e-01 6.30389690e-01
8.01364064e-01 -5.01785696e-01 -6.28539205e-01 1.21689811e-01
4.77755576e-01 -2.07760790e-03 -1.07191455e+00 3.81237477e-01
7.19093263e-01 -1.05371940e+00 8.34172368e-01 -9.14816678e-01
4.88431215e-01 -1.20186098e-01 -2.63466388e-01 -9.74774480e-01
-1.21333510e-01 -6.19364858e-01 -3.48033905e-01 1.30616546e+00
4.05234516e-01 -7.61868179e-01 6.74041152e-01 1.11963511e+00
-3.32065225e-02 -1.12977839e+00 -5.21188974e-01 -7.41473556e-01
2.03604817e-01 -7.95043051e-01 4.27379251e-01 1.02028179e+00
2.18860388e-01 3.81994635e-01 -2.23910868e-01 1.19381770e-01
6.17282033e-01 2.09760621e-01 8.71995866e-01 -1.61506760e+00
-4.68848139e-01 -3.73051822e-01 -4.51146930e-01 -1.04336405e+00
5.97503446e-02 -6.95333540e-01 -3.49267691e-01 -1.15341294e+00
1.15011506e-01 -4.80397373e-01 -2.68424481e-01 3.88325185e-01
2.03009799e-01 -1.35209113e-01 2.92496979e-01 2.99861997e-01
-1.83524147e-01 3.57150376e-01 1.14930606e+00 -3.28300558e-02
-1.26956880e-01 2.64142811e-01 -1.05696476e+00 1.00058627e+00
7.34269381e-01 -9.18271765e-02 -6.14935219e-01 -1.71238407e-01
5.79872310e-01 1.56865507e-01 7.94557989e-01 -7.86516607e-01
1.24411434e-01 -2.89834589e-01 3.42861235e-01 -2.29284912e-01
2.91583806e-01 -1.18636668e+00 2.63940454e-01 4.03167099e-01
-4.31708902e-01 -1.72925845e-01 -2.94314083e-02 4.37047541e-01
-2.09595352e-01 -3.03664595e-01 4.77812022e-01 -1.62742779e-01
-6.34218931e-01 4.18531001e-02 -2.79748052e-01 -6.82956651e-02
8.22298408e-01 -4.42854583e-01 -1.23570533e-02 -4.78214771e-01
-7.92441428e-01 -6.50503160e-03 1.97151348e-01 2.36774743e-01
2.39839360e-01 -9.58613932e-01 -3.57116610e-01 2.79130995e-01
1.80282071e-02 2.34833747e-01 -1.01408914e-01 9.04670119e-01
-8.43764693e-02 7.57890821e-01 1.06907986e-01 -6.14636779e-01
-6.82872415e-01 5.70868313e-01 1.48732349e-01 -2.56631047e-01
-5.69723487e-01 2.38220349e-01 3.75398129e-01 -1.14751384e-01
2.36068457e-01 -4.88433570e-01 6.62392601e-02 -1.03251757e-02
3.07902098e-01 4.20027107e-01 4.69059087e-02 -3.28999192e-01
-8.54019448e-02 3.99553180e-01 -2.60124821e-02 3.35729450e-01
1.40183878e+00 -2.71137029e-01 -1.46923721e-01 9.13666129e-01
1.00031328e+00 -2.42518947e-01 -1.15012002e+00 -3.03452045e-01
4.31542605e-01 -1.74325764e-01 8.65249634e-02 -8.83219242e-01
-8.14530075e-01 1.04657185e+00 1.45849228e-01 6.87272608e-01
1.14124322e+00 -1.17091546e-02 3.43625784e-01 7.69626677e-01
5.59414804e-01 -9.37497497e-01 -2.70646829e-02 2.62947589e-01
7.47938514e-01 -1.31580889e+00 2.14581206e-01 -3.50143343e-01
-3.72435868e-01 1.43808305e+00 3.59548956e-01 1.25875801e-01
7.55106628e-01 4.30331081e-01 -2.24372387e-01 -2.24573165e-01
-5.39674938e-01 -8.90708864e-02 1.97091550e-01 4.61497903e-01
5.08189797e-01 -1.35865495e-01 -2.43342489e-01 7.24950790e-01
-4.02472794e-01 2.67818272e-01 5.36946893e-01 8.19108009e-01
-3.39304656e-01 -1.06773782e+00 -2.48201281e-01 4.18197542e-01
-3.30501735e-01 5.38065620e-02 -3.24006736e-01 1.21230674e+00
2.49680787e-01 9.69470620e-01 -2.90471852e-01 -3.02359521e-01
1.40053332e-01 1.43952416e-02 4.02319998e-01 -4.32252944e-01
-2.72447288e-01 -1.39635727e-01 -8.60702433e-03 -4.18291479e-01
-5.71330905e-01 -6.66524887e-01 -1.03538597e+00 -4.25081015e-01
-4.03202027e-01 2.90019423e-01 7.04370081e-01 1.51729155e+00
1.21378236e-01 4.24250722e-01 6.67957783e-01 -7.42628038e-01
-7.49227524e-01 -6.97988868e-01 -6.89638555e-01 6.24113679e-02
4.19562250e-01 -6.24505103e-01 -3.76614541e-01 5.97588755e-02]
|
[8.567501068115234, 5.163493633270264]
|
fb0b2c87-9a22-4730-813d-9650ea4d2a8b
|
interpretability-and-transparency-driven
|
2307.01225
| null |
https://arxiv.org/abs/2307.01225v1
|
https://arxiv.org/pdf/2307.01225v1.pdf
|
Interpretability and Transparency-Driven Detection and Transformation of Textual Adversarial Examples (IT-DT)
|
Transformer-based text classifiers like BERT, Roberta, T5, and GPT-3 have shown impressive performance in NLP. However, their vulnerability to adversarial examples poses a security risk. Existing defense methods lack interpretability, making it hard to understand adversarial classifications and identify model vulnerabilities. To address this, we propose the Interpretability and Transparency-Driven Detection and Transformation (IT-DT) framework. It focuses on interpretability and transparency in detecting and transforming textual adversarial examples. IT-DT utilizes techniques like attention maps, integrated gradients, and model feedback for interpretability during detection. This helps identify salient features and perturbed words contributing to adversarial classifications. In the transformation phase, IT-DT uses pre-trained embeddings and model feedback to generate optimal replacements for perturbed words. By finding suitable substitutions, we aim to convert adversarial examples into non-adversarial counterparts that align with the model's intended behavior while preserving the text's meaning. Transparency is emphasized through human expert involvement. Experts review and provide feedback on detection and transformation results, enhancing decision-making, especially in complex scenarios. The framework generates insights and threat intelligence empowering analysts to identify vulnerabilities and improve model robustness. Comprehensive experiments demonstrate the effectiveness of IT-DT in detecting and transforming adversarial examples. The approach enhances interpretability, provides transparency, and enables accurate identification and successful transformation of adversarial inputs. By combining technical analysis and human expertise, IT-DT significantly improves the resilience and trustworthiness of transformer-based text classifiers against adversarial attacks.
|
['Sharif Abuadbba', 'M. Ali Babar', 'Bushra Sabir']
|
2023-07-03
| null | null | null | null |
['decision-making']
|
['reasoning']
|
[ 3.03049207e-01 9.28495377e-02 -1.63477138e-02 -1.68468550e-01
-7.67344117e-01 -1.39570415e+00 7.97716379e-01 3.59691232e-01
6.12811297e-02 2.30622426e-01 4.27460790e-01 -7.35776246e-01
5.69892339e-02 -7.65340209e-01 -5.48602581e-01 -3.11400682e-01
8.18032175e-02 2.88881898e-01 -2.82876104e-01 -4.05016243e-01
2.99380451e-01 6.65419459e-01 -7.15064466e-01 7.38762081e-01
1.23881626e+00 8.49408150e-01 -5.58120608e-01 7.23900259e-01
-6.42291233e-02 8.41864407e-01 -9.33163941e-01 -1.08239663e+00
5.18871546e-01 -9.11355987e-02 -7.33527124e-01 -3.97617757e-01
2.63343364e-01 -4.11362857e-01 -2.29200527e-01 1.24568677e+00
3.32553953e-01 -2.18743891e-01 7.65423894e-01 -1.50473630e+00
-1.29267967e+00 7.94367790e-01 -2.63295233e-01 2.87275493e-01
5.12086153e-01 5.92420995e-01 1.03171706e+00 -8.94711375e-01
4.10887748e-01 1.56910217e+00 8.68519068e-01 8.27115476e-01
-1.22714639e+00 -8.06199372e-01 4.59331572e-01 5.39009385e-02
-9.68255699e-01 -3.58798206e-01 8.38805676e-01 -5.50860465e-01
9.79389310e-01 8.66345584e-01 4.69586223e-01 1.67222786e+00
4.36317593e-01 7.79156506e-01 7.95406997e-01 -3.34893316e-01
1.59434423e-01 5.76383412e-01 -6.57877401e-02 5.81115127e-01
2.71721810e-01 3.73311728e-01 -4.33421016e-01 -5.68759084e-01
4.83730137e-02 1.03682607e-01 -3.25187922e-01 1.73792094e-01
-8.76575708e-01 1.02825928e+00 6.01678550e-01 2.22858623e-01
-3.29039395e-01 -8.74492675e-02 6.38982356e-01 3.89399320e-01
6.58147097e-01 1.15027785e+00 -2.04050168e-01 1.11447603e-01
-4.58093762e-01 1.42597631e-01 6.64255261e-01 7.51057804e-01
1.29585415e-01 4.21229213e-01 -4.80311334e-01 2.42370263e-01
1.14919327e-01 9.28724527e-01 1.79225385e-01 -3.35336983e-01
8.51685941e-01 8.88046265e-01 8.81336723e-03 -1.33502758e+00
-2.11769808e-02 -5.00236988e-01 -6.50831878e-01 5.13969779e-01
1.54523760e-01 -1.81542501e-01 -8.75202179e-01 1.49232519e+00
1.88965484e-01 -3.86598170e-01 3.96169454e-01 7.14510322e-01
3.16873848e-01 6.56565309e-01 2.99698085e-01 3.89779061e-01
1.42526186e+00 -5.45340836e-01 -7.18150496e-01 -5.95782936e-01
6.49821877e-01 -7.17347980e-01 1.33947992e+00 3.62976193e-01
-6.82974875e-01 -9.54709873e-02 -1.00972366e+00 1.92636222e-01
-5.39645493e-01 -1.09683104e-01 2.80833036e-01 1.01154900e+00
-6.22097313e-01 5.03434181e-01 -4.35854793e-01 4.84390929e-02
6.63328528e-01 1.98829874e-01 -4.05334979e-01 1.56647757e-01
-1.78184259e+00 1.16065454e+00 6.68006912e-02 4.95806225e-02
-1.02184021e+00 -9.66579914e-01 -8.08956087e-01 2.22242519e-01
1.51471004e-01 -5.51678240e-01 1.19167149e+00 -1.11015379e+00
-1.16835880e+00 4.34542358e-01 2.59423442e-02 -6.25163794e-01
8.72196496e-01 -3.73844385e-01 -6.30986094e-01 1.52909428e-01
7.32644126e-02 1.40232176e-01 1.39679158e+00 -1.28879714e+00
-3.29336494e-01 -2.96504229e-01 1.28355950e-01 9.41228494e-02
-9.66307223e-01 2.23562747e-01 1.95526436e-01 -1.05102026e+00
-5.26015520e-01 -6.06780112e-01 -1.26286864e-01 2.33852603e-02
-7.88630009e-01 2.51871020e-01 1.20664060e+00 -1.11768377e+00
1.41500342e+00 -2.11894536e+00 -3.59106585e-02 4.95858669e-01
5.71188569e-01 6.50024712e-01 -1.82364240e-01 5.68974912e-01
-2.56794274e-01 7.58181751e-01 -7.11680502e-02 -1.39898583e-01
3.11636358e-01 -1.78355426e-01 -1.19265199e+00 1.63768455e-01
7.38003612e-01 1.05678892e+00 -8.30482364e-01 -8.54637250e-02
1.88981697e-01 3.01182657e-01 -5.89335978e-01 1.93248287e-01
-2.59821743e-01 2.83502936e-01 -5.84015965e-01 7.64779866e-01
5.04433930e-01 1.35423452e-01 -1.00544065e-01 -2.28754848e-01
2.71868557e-01 2.15247691e-01 -6.32376492e-01 6.17426872e-01
-6.75587833e-01 8.34904611e-01 -1.00344360e-01 -6.01202190e-01
9.33070421e-01 3.50803852e-01 -1.49427503e-01 -6.45688534e-01
1.98477730e-01 1.48222120e-02 -1.41579553e-01 -4.48990315e-01
3.72481734e-01 4.71654870e-02 -4.37366903e-01 7.11374700e-01
-4.97396111e-01 -2.31340021e-01 -2.35521972e-01 4.73243088e-01
1.17566299e+00 -5.06191432e-01 2.66090810e-01 9.05432031e-02
4.85214323e-01 4.18656319e-02 3.77488464e-01 7.68151402e-01
-9.98684838e-02 1.73461318e-01 6.37549102e-01 -6.72884226e-01
-1.08432925e+00 -1.21924984e+00 2.27916896e-01 9.46862400e-01
-1.24523111e-01 -4.45375800e-01 -9.82230484e-01 -1.17690957e+00
1.97078884e-01 1.34072530e+00 -1.00510406e+00 -1.01286614e+00
-4.09349114e-01 -3.41159701e-01 9.15796220e-01 5.59952199e-01
2.72211522e-01 -9.62434232e-01 -4.06578988e-01 6.33352846e-02
-2.35711619e-01 -9.23443913e-01 -8.25512707e-01 6.50516674e-02
-5.56540430e-01 -9.88125920e-01 4.23318073e-02 -3.13329339e-01
1.05770957e+00 -3.21831815e-02 7.49553263e-01 3.40988897e-02
-4.28023458e-01 2.24671438e-01 -4.79452014e-01 -8.60367000e-01
-1.09894156e+00 -8.49151388e-02 1.72706038e-01 5.63354194e-02
1.02553040e-01 -1.85755491e-01 -1.92826882e-01 2.95811772e-01
-1.07641447e+00 -2.19447717e-01 4.79264289e-01 9.30758595e-01
-6.80264533e-02 1.00078113e-01 3.77084643e-01 -1.05668354e+00
1.30161440e+00 -3.67099822e-01 -5.30305445e-01 5.87948978e-01
-9.30836141e-01 1.23524435e-01 1.39359307e+00 -7.85799861e-01
-9.85256493e-01 -2.68549293e-01 4.46596593e-02 -6.31126463e-01
2.73452848e-01 3.69950920e-01 -3.82905126e-01 -1.99760169e-01
1.37304592e+00 4.88876142e-02 -2.70591885e-01 -5.29105105e-02
5.01041412e-01 7.60496438e-01 4.49303508e-01 -4.18645263e-01
1.63566613e+00 2.35665500e-01 -6.18950844e-01 -1.93166763e-01
-6.80010319e-01 2.16418579e-01 -3.40213031e-01 -2.32166588e-01
5.05319238e-01 -5.31854331e-01 -6.71481252e-01 2.84378856e-01
-1.36898816e+00 -6.32522777e-02 -3.45021099e-01 -1.21231422e-01
7.77532384e-02 3.03431302e-01 -4.38993096e-01 -9.89753306e-01
-9.23257828e-01 -1.23841345e+00 9.57268894e-01 -4.69996966e-02
-7.19613075e-01 -1.21744478e+00 -2.18320683e-01 4.72701043e-01
4.91829157e-01 5.06627142e-01 1.29021001e+00 -1.14668250e+00
-2.88466573e-01 -8.29006195e-01 8.33782181e-02 4.15025175e-01
2.60373622e-01 2.29478776e-01 -1.15987670e+00 -2.59680361e-01
-4.94365916e-02 -1.01558894e-01 4.47633892e-01 -2.36506134e-01
1.02810049e+00 -1.20824659e+00 -2.57571101e-01 4.82132584e-01
7.82382786e-01 3.70536327e-01 4.53746080e-01 2.84036219e-01
7.96568811e-01 6.43391490e-01 6.07000709e-01 4.01772827e-01
-1.15396820e-01 4.52627957e-01 6.82382822e-01 -4.18926120e-01
3.67372781e-01 -6.25535309e-01 7.12493718e-01 1.39826626e-01
4.58218813e-01 -5.34117401e-01 -1.06832635e+00 3.10354799e-01
-1.45187414e+00 -1.15403152e+00 1.72568709e-01 1.85591257e+00
8.39928389e-01 4.98448372e-01 -1.41909465e-01 4.21704680e-01
6.70943618e-01 5.62691465e-02 -7.52795696e-01 -9.74964559e-01
1.02478571e-01 4.44070000e-04 4.18970823e-01 6.95177257e-01
-1.03108418e+00 1.09423745e+00 6.13471031e+00 8.29490006e-01
-1.20699620e+00 3.26503813e-02 7.42279232e-01 -1.36277676e-01
-9.74753678e-01 -1.90839946e-01 -4.63952780e-01 3.41807216e-01
7.97814131e-01 -5.93622923e-01 5.83461165e-01 1.00967896e+00
3.86491358e-01 6.54052734e-01 -1.13888741e+00 4.15745139e-01
7.56037310e-02 -1.45147336e+00 5.33845663e-01 -1.75887868e-01
5.16519785e-01 -7.36657023e-01 7.22705305e-01 1.39052093e-01
7.32269168e-01 -1.28688419e+00 1.10115051e+00 2.51113951e-01
7.79566586e-01 -8.86423588e-01 7.11610496e-01 8.41559842e-02
-1.05317163e+00 -6.17975473e-01 -4.93313633e-02 1.78865150e-01
-1.30886391e-01 4.68712777e-01 -1.29511762e+00 4.65032578e-01
4.87127304e-01 4.50723082e-01 -6.72886133e-01 1.31262854e-01
-6.34602726e-01 6.92340970e-01 5.00167869e-02 -9.79033709e-02
2.93481767e-01 1.76240459e-01 9.11469877e-01 1.19118357e+00
3.24412175e-02 -1.87105581e-01 -3.85930836e-02 1.17443061e+00
-2.33042002e-01 -2.53660660e-02 -9.07680094e-01 -5.86691916e-01
8.15563202e-01 1.16484380e+00 -3.63451630e-01 -1.13359168e-01
2.59986371e-01 9.24352944e-01 2.14233696e-01 4.35322583e-01
-9.27555084e-01 -5.75729549e-01 9.64537799e-01 -1.83914939e-03
-7.57765099e-02 2.59021550e-01 -8.02611589e-01 -9.11082208e-01
2.96808690e-01 -1.53617966e+00 5.28822303e-01 -5.83894908e-01
-1.40892410e+00 9.77623999e-01 -1.10938743e-01 -1.30161333e+00
-1.76931366e-01 -5.70641279e-01 -1.13656402e+00 9.44889367e-01
-9.92911875e-01 -1.46288192e+00 -1.00392085e-02 4.98783916e-01
4.54557806e-01 -4.59534973e-01 8.76347303e-01 -1.65740907e-01
-7.70487070e-01 1.17656624e+00 -5.27672432e-02 4.23910677e-01
5.60181677e-01 -1.24518895e+00 1.27668476e+00 1.32659280e+00
5.40637672e-02 8.69157434e-01 8.97437096e-01 -9.07138169e-01
-1.28472865e+00 -1.47701597e+00 7.47074544e-01 -9.09641385e-01
9.66673613e-01 -8.74017119e-01 -9.94555473e-01 6.08874738e-01
-1.35759905e-01 -3.49840313e-01 7.03343749e-01 -1.97965980e-01
-1.04133224e+00 -1.13788866e-01 -1.45304847e+00 1.13430250e+00
7.27995276e-01 -1.04277754e+00 -6.68016374e-01 5.46342015e-01
1.00591028e+00 -3.00246537e-01 -6.44185722e-01 1.07380614e-01
2.57438421e-01 -5.02285063e-01 1.07935441e+00 -9.80812311e-01
4.36993301e-01 -9.70714837e-02 2.72168845e-01 -1.48845959e+00
-3.72063041e-01 -1.02379465e+00 7.89694414e-02 1.28565335e+00
7.45817840e-01 -9.64030027e-01 3.64655674e-01 9.52293754e-01
-7.89690688e-02 -6.29028082e-01 -7.68280387e-01 -6.67821288e-01
2.33045861e-01 -5.37051976e-01 8.70257556e-01 1.19509339e+00
2.59758919e-01 -1.45979747e-02 -3.95658672e-01 5.62654436e-01
4.18381572e-01 -2.10389912e-01 6.01291358e-01 -8.85480344e-01
5.88159673e-02 -5.04817426e-01 -2.65612751e-01 -1.35172352e-01
3.84652078e-01 -9.20103192e-01 -2.63488412e-01 -1.16133797e+00
-1.98048174e-01 -2.51380920e-01 1.74226001e-01 1.00312924e+00
-6.50155187e-01 1.03437893e-01 3.42379212e-01 2.14094251e-01
-2.32817959e-02 4.04523671e-01 8.41343820e-01 -6.36519134e-01
-3.03908680e-02 -5.21451645e-02 -1.25589192e+00 5.19127607e-01
1.04444993e+00 -7.85358131e-01 -3.89125943e-01 -4.67831016e-01
1.71318129e-01 -4.23559457e-01 5.50976932e-01 -5.97164512e-01
4.25490141e-02 -3.15827042e-01 5.58921278e-01 3.21990289e-02
2.09555235e-02 -9.11100328e-01 1.22979082e-01 8.84803653e-01
-6.93466961e-01 4.30906147e-01 5.06594181e-01 6.24601901e-01
-3.02552860e-02 -1.36393473e-01 5.89353383e-01 1.42395034e-01
-2.14642018e-01 2.49792576e-01 -6.58090234e-01 1.04549386e-01
1.08522403e+00 -1.85892209e-01 -5.98911822e-01 -5.75137317e-01
-3.89276028e-01 3.11531574e-01 4.58897650e-01 6.62102103e-01
8.05421889e-01 -1.24997830e+00 -8.00884247e-01 4.52191472e-01
2.40081489e-01 -5.15709043e-01 1.15429319e-01 3.40614855e-01
-4.19617742e-01 2.13573769e-01 -8.20216015e-02 -1.25707164e-01
-1.53990400e+00 6.87960207e-01 4.60445285e-01 -5.45290589e-01
-3.53771716e-01 7.63296783e-01 3.55635643e-01 -4.08246726e-01
2.00152263e-01 -2.01131701e-01 -1.34760499e-01 -2.08776787e-01
7.21179008e-01 1.52741879e-01 3.31519656e-02 -2.41234958e-01
-4.29787308e-01 1.60672188e-01 -5.13360679e-01 -1.68968495e-02
9.81999278e-01 3.54477793e-01 1.36970088e-01 -1.99932754e-01
8.89881074e-01 3.43588561e-01 -9.97547150e-01 -6.18884228e-02
6.03011772e-02 -6.76089585e-01 -2.17646196e-01 -1.25389397e+00
-8.26897025e-01 1.13659787e+00 3.01138461e-01 3.86899829e-01
1.13509655e+00 -4.24147725e-01 7.69803166e-01 3.77004802e-01
-1.10063605e-01 -6.81846857e-01 4.20938790e-01 4.94608045e-01
1.33752465e+00 -9.08853650e-01 -1.03243597e-01 -3.69790941e-01
-1.10096645e+00 1.19497311e+00 7.33717799e-01 3.49809915e-01
3.51713784e-02 3.76518339e-01 4.05160278e-01 -2.20932867e-02
-7.67486513e-01 6.10060871e-01 5.65133989e-01 8.91903460e-01
-1.74346015e-01 9.72185750e-03 4.86315340e-01 6.68201745e-01
-5.61754525e-01 -7.88020372e-01 2.22256228e-01 8.09992790e-01
-1.33775115e-01 -1.01505470e+00 -8.88666570e-01 4.05542612e-01
-3.70589614e-01 -4.33768719e-01 -1.23340476e+00 5.61735213e-01
-2.54639745e-01 1.21588790e+00 -4.08040494e-01 -1.00186956e+00
5.74382782e-01 1.72754675e-02 -1.56966895e-01 -4.37661141e-01
-1.51071715e+00 -4.54606980e-01 2.87708968e-01 -3.86426240e-01
7.33848631e-01 -4.46729094e-01 -1.01709890e+00 -7.09618449e-01
-4.28298712e-01 3.57003301e-01 4.69101936e-01 9.37778473e-01
5.78646541e-01 5.42395532e-01 9.78051782e-01 -4.19551909e-01
-1.17571509e+00 -6.54144168e-01 8.64919052e-02 7.00360298e-01
3.75821680e-01 -3.10282856e-01 -6.97728276e-01 3.37764546e-02]
|
[5.994237422943115, 8.067666053771973]
|
1bb632fb-6a00-4b12-8fc8-62f12f952e2f
|
how-far-are-we-from-solving-the-2d-3d-face
|
1703.07332
| null |
http://arxiv.org/abs/1703.07332v3
|
http://arxiv.org/pdf/1703.07332v3.pdf
|
How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)
|
This paper investigates how far a very deep neural network is from attaining
close to saturating performance on existing 2D and 3D face alignment datasets.
To this end, we make the following 5 contributions: (a) we construct, for the
first time, a very strong baseline by combining a state-of-the-art architecture
for landmark localization with a state-of-the-art residual block, train it on a
very large yet synthetically expanded 2D facial landmark dataset and finally
evaluate it on all other 2D facial landmark datasets. (b) We create a guided by
2D landmarks network which converts 2D landmark annotations to 3D and unifies
all existing datasets, leading to the creation of LS3D-W, the largest and most
challenging 3D facial landmark dataset to date ~230,000 images. (c) Following
that, we train a neural network for 3D face alignment and evaluate it on the
newly introduced LS3D-W. (d) We further look into the effect of all
"traditional" factors affecting face alignment performance like large pose,
initialization and resolution, and introduce a "new" one, namely the size of
the network. (e) We show that both 2D and 3D face alignment networks achieve
performance of remarkable accuracy which is probably close to saturating the
datasets used. Training and testing code as well as the dataset can be
downloaded from https://www.adrianbulat.com/face-alignment/
|
['Georgios Tzimiropoulos', 'Adrian Bulat']
|
2017-03-21
|
how-far-are-we-from-solving-the-2d-3d-face-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Bulat_How_Far_Are_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Bulat_How_Far_Are_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['head-pose-estimation']
|
['computer-vision']
|
[-2.41128072e-01 1.95746347e-01 -5.48261357e-03 -6.67242169e-01
-7.58470297e-01 -3.86748523e-01 7.76393950e-01 -3.85704011e-01
-3.84594470e-01 2.01251864e-01 2.66465902e-01 -1.77999541e-01
6.83445781e-02 -3.25013906e-01 -6.90071583e-01 -3.96959841e-01
-2.19422296e-01 8.17820847e-01 -1.89931557e-01 -3.89720410e-01
-7.59515399e-03 1.17710865e+00 -1.40630114e+00 -3.56067777e-01
8.03579763e-02 1.04371607e+00 -4.28560317e-01 3.65406185e-01
9.92049351e-02 -6.10425211e-02 -3.98658425e-01 -5.70473373e-01
7.15774715e-01 -3.22843879e-01 -7.83524811e-01 2.27885153e-02
1.25240600e+00 -4.47111219e-01 -1.65939495e-01 7.64218330e-01
1.04623091e+00 -7.62095824e-02 4.44876015e-01 -1.39020002e+00
-5.68733394e-01 2.42365271e-01 -6.74793243e-01 -8.50923173e-03
4.52851444e-01 1.23021662e-01 5.03989518e-01 -1.27261257e+00
8.10128927e-01 1.28536427e+00 1.17157054e+00 9.71059203e-01
-1.15645301e+00 -7.86748946e-01 -1.26067162e-01 -2.68633544e-01
-1.75169265e+00 -1.08529663e+00 9.56004739e-01 -3.13933015e-01
1.02517700e+00 -3.15485224e-02 4.80395287e-01 1.22655022e+00
-3.00966859e-01 3.42077404e-01 9.34416711e-01 -6.19487226e-01
-1.43895730e-01 -3.30829382e-01 -2.02148065e-01 1.06371105e+00
-9.12913904e-02 1.53728619e-01 -3.35342646e-01 -2.70729754e-02
7.43431151e-01 -4.88817632e-01 -2.08984271e-01 -5.69794953e-01
-9.07815099e-01 4.63179588e-01 4.59149480e-01 4.10353869e-01
-6.84136301e-02 9.63041037e-02 3.50378215e-01 2.58787900e-01
6.80352330e-01 2.91138142e-01 -5.08696020e-01 5.93826286e-02
-1.04877865e+00 2.70895541e-01 6.22208118e-01 8.37177455e-01
7.30045915e-01 9.94157717e-02 1.48346037e-01 9.58806753e-01
3.35268855e-01 5.32039940e-01 3.72951329e-01 -1.05791450e+00
2.32080072e-01 3.01313192e-01 -4.65379544e-02 -1.05521679e+00
-8.49978924e-01 -2.15286717e-01 -7.18139946e-01 5.21377265e-01
6.83823526e-01 -2.25026771e-01 -1.01060522e+00 2.16755295e+00
3.87767911e-01 3.69297087e-01 -1.85042188e-01 7.49959469e-01
1.12864792e+00 1.91621650e-02 -2.04447195e-01 1.07440263e-01
1.22860014e+00 -8.88196707e-01 -3.31070811e-01 -1.40388131e-01
4.86153394e-01 -9.03575540e-01 9.09630775e-01 -1.36915147e-01
-1.21801364e+00 -8.50203753e-01 -9.97960985e-01 -1.65462419e-01
-4.96568888e-01 1.84929281e-01 4.89538193e-01 7.30665267e-01
-1.81161332e+00 4.86673355e-01 -6.75204575e-01 -7.70541251e-01
6.12465560e-01 7.19212592e-01 -1.09943616e+00 1.30309269e-01
-8.13528180e-01 1.14353406e+00 7.15649724e-02 3.47373247e-01
-8.78435373e-01 -8.05373013e-01 -9.92422163e-01 -2.32934669e-01
2.67649770e-01 -6.93458855e-01 1.13773131e+00 -8.12738001e-01
-1.53636539e+00 1.62557888e+00 -2.20616341e-01 -1.02279591e-03
4.77803111e-01 -1.14747912e-01 -4.14739937e-01 -7.05321208e-02
3.92408296e-02 1.19535601e+00 7.24911869e-01 -1.32321620e+00
-1.18086077e-01 -7.06984997e-01 -1.99288204e-01 -9.63952765e-02
-1.34191573e-01 1.24705292e-01 -7.84058452e-01 -3.87459248e-01
6.66094129e-04 -8.89509976e-01 7.54925832e-02 2.38036327e-02
-3.93925726e-01 -2.92004108e-01 6.00360870e-01 -5.48143804e-01
6.20523036e-01 -2.19085121e+00 1.47074968e-01 2.74752289e-01
1.95238560e-01 4.13261503e-01 -7.56309807e-01 2.42703527e-01
-6.42307937e-01 2.03751266e-01 -1.29591385e-02 -1.02488327e+00
2.30659530e-01 -4.09574658e-02 9.22638625e-02 6.31249607e-01
3.90009910e-01 1.06357396e+00 -5.26986718e-01 -3.29796910e-01
3.77734751e-02 7.49170184e-01 -5.22388935e-01 6.44428879e-02
1.57210335e-01 4.84519720e-01 -3.96581367e-02 8.10815156e-01
9.22114491e-01 1.01449862e-01 -3.90492007e-02 -4.10223484e-01
-9.06612575e-02 -5.57993688e-02 -9.58799541e-01 1.97175300e+00
-2.42623895e-01 5.35289884e-01 1.73775092e-01 -6.80868089e-01
1.18116128e+00 3.59914780e-01 7.51624584e-01 -7.29592264e-01
2.79192895e-01 3.01517397e-01 -2.46856108e-01 -1.82958394e-01
2.29346097e-01 -1.08225886e-02 2.21991837e-01 4.47909832e-01
5.47318637e-01 -1.62657544e-01 -4.19300571e-02 -1.87510759e-01
8.53393734e-01 4.73156869e-01 8.02180916e-02 -3.76192003e-01
5.50466239e-01 -3.83545995e-01 2.28786752e-01 3.18730474e-01
-4.93885577e-01 8.53786469e-01 6.68121815e-01 -7.80442476e-01
-1.26324952e+00 -8.76372457e-01 -3.24064434e-01 8.53954077e-01
-2.90754139e-01 -4.08290684e-01 -9.52425301e-01 -8.85026932e-01
1.44282281e-01 5.03246114e-02 -9.72787201e-01 5.82368597e-02
-7.22047925e-01 -5.85082591e-01 9.75973368e-01 4.15325493e-01
4.88607734e-01 -7.89405346e-01 -4.45713960e-02 -3.44183803e-01
1.51503712e-01 -1.09479260e+00 -5.86858809e-01 1.35588171e-02
-5.18140852e-01 -1.09782112e+00 -6.33189142e-01 -8.84918392e-01
8.56991589e-01 -3.83523293e-02 1.39322436e+00 3.38047057e-01
-1.01705767e-01 5.15331447e-01 -1.55953113e-02 -4.41648155e-01
-2.59417951e-01 2.38787919e-01 4.22252208e-01 -1.74013674e-01
5.60559690e-01 -5.92165649e-01 -4.20278370e-01 3.68552595e-01
-4.00420815e-01 -3.09071958e-01 4.43494290e-01 5.71450412e-01
5.05931020e-01 -3.40997368e-01 4.13091719e-01 -5.05636156e-01
4.08263922e-01 -4.13295776e-02 -6.68325782e-01 6.40629306e-02
-3.62340391e-01 -1.29473731e-01 2.80326337e-01 -7.74649531e-02
-6.50865078e-01 2.56062001e-01 -8.36051762e-01 -5.06076455e-01
-4.41092074e-01 -4.14405167e-02 -3.00150543e-01 -5.03478527e-01
6.85390770e-01 -2.50970930e-01 5.11276126e-01 -6.07932508e-01
4.04745936e-01 3.39474767e-01 6.93494499e-01 -6.63147748e-01
1.03070700e+00 4.70956862e-01 3.67454112e-01 -6.46003723e-01
-6.71136320e-01 -9.58839282e-02 -1.28991032e+00 -1.48872331e-01
6.93388700e-01 -7.76033342e-01 -8.34153354e-01 6.45949721e-01
-1.00681233e+00 -6.20909452e-01 -2.34939218e-01 2.58827388e-01
-6.04851842e-01 1.55030638e-01 -4.33812231e-01 -3.07875425e-01
-2.43370265e-01 -1.22693729e+00 1.43291080e+00 1.14047512e-01
-2.95464993e-01 -1.05263007e+00 2.72400796e-01 1.57122359e-01
4.83526677e-01 5.66966593e-01 6.35325730e-01 -7.58215070e-01
-1.19804859e-01 -3.23760569e-01 -2.40374237e-01 2.85815686e-01
7.08957091e-02 1.53175473e-01 -1.16047156e+00 -4.20363426e-01
-9.21349600e-02 -4.57842797e-01 5.28088689e-01 2.25919098e-01
8.04371715e-01 7.12753311e-02 -2.45481625e-01 9.33463573e-01
1.13114417e+00 -9.43152457e-02 5.90580523e-01 3.37748677e-01
6.86233222e-01 6.46835506e-01 2.07809284e-01 6.89131692e-02
3.29566658e-01 1.19501746e+00 4.20670837e-01 -5.16538978e-01
-4.30282593e-01 -2.75576711e-01 5.68695590e-02 5.99060655e-01
-3.17257613e-01 1.97626159e-01 -1.03239429e+00 3.38663816e-01
-1.35335457e+00 -7.50080347e-01 2.83947200e-01 2.23349810e+00
7.78190851e-01 -1.04134709e-01 4.35470879e-01 -1.18820570e-01
4.93405282e-01 2.24740669e-01 -2.61915028e-01 -2.75716394e-01
1.51185514e-02 5.86962402e-01 2.56336629e-01 7.76930809e-01
-1.27501345e+00 1.22111464e+00 6.63277674e+00 6.14837825e-01
-1.30644619e+00 1.10159971e-01 7.75332153e-01 -9.72274244e-02
1.23216942e-01 -4.55360770e-01 -1.19406295e+00 1.86578244e-01
7.63969243e-01 2.57800519e-01 1.91839740e-01 8.00634146e-01
6.54407032e-03 2.90399641e-01 -1.42714167e+00 1.13228714e+00
5.46493948e-01 -1.00228047e+00 1.03528194e-01 2.03136176e-01
6.08240783e-01 1.43657103e-01 9.52778235e-02 1.89141452e-01
9.57549512e-02 -1.50696409e+00 5.16706884e-01 3.85354757e-01
1.12662768e+00 -7.86391079e-01 8.43275428e-01 -2.10285261e-01
-1.05306816e+00 2.43234396e-01 -3.68896931e-01 1.99645787e-01
3.33268829e-02 1.75033465e-01 -7.38277495e-01 5.24207890e-01
6.65953577e-01 5.28095484e-01 -9.06131089e-01 8.55047584e-01
-1.00505061e-01 1.03464782e-01 -5.13039529e-01 4.83943969e-01
3.39324206e-01 -9.51132253e-02 3.24857652e-01 1.02920461e+00
5.03346384e-01 -1.24502242e-01 -1.02090962e-01 5.76278269e-01
-4.94226128e-01 6.35348400e-03 -8.13714147e-01 3.59474391e-01
5.52260876e-01 1.36260426e+00 -6.36117995e-01 1.14692032e-01
-3.83140564e-01 7.53498614e-01 4.86834079e-01 2.67717242e-01
-8.02474916e-01 -8.23918283e-02 8.45068216e-01 3.45940068e-02
1.42303407e-01 -2.24560216e-01 5.11412248e-02 -7.80538619e-01
2.13224329e-02 -9.11150992e-01 8.67910683e-02 -7.12458432e-01
-1.27254760e+00 9.59572315e-01 5.20027168e-02 -7.47846365e-01
-2.57634044e-01 -8.54271889e-01 -5.12515485e-01 8.90355408e-01
-1.40711570e+00 -1.43471646e+00 -4.78376418e-01 7.20966697e-01
3.89361680e-02 -5.36686182e-01 1.03113902e+00 5.83207250e-01
-6.32168174e-01 1.16259682e+00 -2.71657586e-01 6.32526338e-01
1.08017254e+00 -1.06592560e+00 1.01298594e+00 5.92988193e-01
4.16568130e-01 5.35672903e-01 1.95715517e-01 -3.02365303e-01
-1.24572229e+00 -8.77168894e-01 1.02503633e+00 -1.03713405e+00
3.84905219e-01 -5.78664124e-01 -5.83522975e-01 1.02753878e+00
2.59942841e-02 2.51877010e-01 6.04240000e-01 3.02179188e-01
-4.56126332e-01 -3.32126200e-01 -1.28993320e+00 7.20956147e-01
1.42361414e+00 -6.16718233e-01 -4.87729549e-01 3.48544478e-01
3.92181039e-01 -6.72850490e-01 -9.20732737e-01 6.04084015e-01
7.68693924e-01 -1.13089228e+00 1.30282795e+00 -5.50123632e-01
2.40199760e-01 -1.82507008e-01 -1.14755556e-01 -1.22077858e+00
-1.64087266e-01 -6.84711277e-01 2.88245380e-01 1.35879171e+00
3.75459641e-01 -6.69182301e-01 9.91296828e-01 4.77029413e-01
-2.14289561e-01 -1.03213000e+00 -1.27067649e+00 -6.04216814e-01
2.49993965e-01 -2.74434030e-01 9.77034807e-01 1.14181399e+00
-5.03910303e-01 2.16858461e-01 -2.61368722e-01 3.56758274e-02
5.09041548e-01 -2.55590439e-01 1.21061003e+00 -1.15644455e+00
2.12464586e-01 -7.23118186e-01 -6.53197467e-01 -9.79471266e-01
4.96841639e-01 -1.03598022e+00 -2.27531537e-01 -9.15227294e-01
-5.25878929e-02 -5.33830643e-01 3.22165643e-03 8.46651316e-01
1.20241858e-01 1.03864193e+00 9.74133238e-02 5.09512872e-02
-1.62076831e-01 3.47351879e-01 1.09205317e+00 1.96786538e-01
-4.05641682e-02 -2.24365592e-01 -6.16910338e-01 7.28777647e-01
6.69999659e-01 -9.11590606e-02 -1.16405144e-01 -5.64183235e-01
-1.96202725e-01 -2.36173034e-01 2.98338443e-01 -9.67835486e-01
5.17239049e-02 4.28598225e-01 8.04979563e-01 -3.00444543e-01
4.85536754e-01 -7.85941243e-01 1.46583781e-01 8.72765556e-02
-5.97946793e-02 3.21088165e-01 6.05860710e-01 -1.49559468e-01
-6.99936226e-02 -7.07680285e-02 9.38768089e-01 -7.46111721e-02
-6.32091165e-01 6.12561584e-01 3.87389123e-01 -1.22817401e-02
9.65307772e-01 -2.55300492e-01 -3.90372485e-01 -1.75408870e-01
-7.14013278e-01 2.27614734e-02 7.31754720e-01 5.37671328e-01
3.68290365e-01 -1.57491112e+00 -8.33926797e-01 6.71158731e-01
-1.55994192e-01 -8.65391120e-02 -1.03540689e-01 7.62853980e-01
-7.14379132e-01 4.38324898e-01 -5.93294680e-01 -6.73845589e-01
-1.45000875e+00 4.22531605e-01 7.20212519e-01 1.03977948e-01
-3.76355797e-01 1.14144254e+00 -1.39511064e-01 -9.02015746e-01
2.88415968e-01 1.30882755e-01 -1.35268986e-01 1.58208996e-01
3.98057938e-01 3.05552948e-02 2.52274126e-01 -1.15985394e+00
-6.27515614e-01 1.26495647e+00 1.08132921e-01 1.08419269e-01
1.40759182e+00 8.41060504e-02 -2.72457033e-01 -1.10560358e-01
1.45758867e+00 1.78137943e-01 -1.12918234e+00 9.75003019e-02
-8.32604766e-02 -4.78795379e-01 -3.37682337e-01 -6.78796947e-01
-1.35823214e+00 8.15964580e-01 8.73107195e-01 -1.72543421e-01
1.05143106e+00 5.01510054e-02 5.59052467e-01 2.66067415e-01
1.90021440e-01 -5.99889278e-01 1.47865891e-01 5.40345669e-01
1.03901243e+00 -1.29774821e+00 1.41335845e-01 -9.98242497e-02
-1.57898337e-01 1.04533815e+00 7.72704899e-01 -1.24582149e-01
7.14686930e-01 3.95690322e-01 5.09907722e-01 -4.36380655e-01
-1.87426865e-01 -2.24140704e-01 4.12336171e-01 7.05618978e-01
5.36655247e-01 -3.79718244e-01 2.08868653e-01 8.22814927e-02
-5.68283260e-01 -6.48321956e-02 -1.60905883e-01 5.41160345e-01
1.51678517e-01 -1.41038132e+00 -4.03057277e-01 -9.72045213e-02
-4.92877930e-01 1.65767372e-02 -6.32814169e-01 1.34406638e+00
2.79327959e-01 4.95413154e-01 2.13687211e-01 -3.79501373e-01
5.43979764e-01 4.02700186e-01 7.28475571e-01 -3.44505966e-01
-4.12066996e-01 -1.70784071e-01 2.69048642e-02 -7.85903990e-01
-4.32601511e-01 -4.34462488e-01 -7.73741245e-01 -5.47741115e-01
9.51913465e-03 -9.33942385e-03 7.80427754e-01 7.54405439e-01
5.52529216e-01 1.51086673e-01 4.92793709e-01 -1.59038663e+00
-3.52395326e-01 -9.86364186e-01 -2.48236924e-01 4.93548870e-01
2.69581735e-01 -8.35521519e-01 -3.78222853e-01 -2.93411463e-01]
|
[13.440896034240723, 0.36690250039100647]
|
56c443ac-8358-4258-bad1-7a10a8e0c58c
|
improving-spoken-language-identification-with
|
2302.08229
| null |
https://arxiv.org/abs/2302.08229v1
|
https://arxiv.org/pdf/2302.08229v1.pdf
|
Improving Spoken Language Identification with Map-Mix
|
The pre-trained multi-lingual XLSR model generalizes well for language identification after fine-tuning on unseen languages. However, the performance significantly degrades when the languages are not very distinct from each other, for example, in the case of dialects. Low resource dialect classification remains a challenging problem to solve. We present a new data augmentation method that leverages model training dynamics of individual data points to improve sampling for latent mixup. The method works well in low-resource settings where generalization is paramount. Our datamaps-based mixup technique, which we call Map-Mix improves weighted F1 scores by 2% compared to the random mixup baseline and results in a significantly well-calibrated model. The code for our method is open sourced on https://github.com/skit-ai/Map-Mix.
|
['Eng Siong Chng', 'Tarun Gupta', 'Swaraj Dalmia', 'Kriti Anandan', 'Shangeth Rajaa']
|
2023-02-16
| null | null | null | null |
['spoken-language-identification']
|
['speech']
|
[-1.27551690e-01 -4.10782546e-01 -5.93336821e-01 -5.81491232e-01
-1.25448406e+00 -1.05103648e+00 7.16706991e-01 -7.56636932e-02
-4.77864772e-01 5.53090632e-01 4.68498617e-01 -5.39034069e-01
3.08097184e-01 -5.99503577e-01 -7.60096788e-01 -4.94441241e-01
3.35594267e-01 6.65906906e-01 -2.67369777e-01 -2.32777253e-01
-2.05998898e-01 1.81549296e-01 -1.18546140e+00 2.44038194e-01
1.02366078e+00 2.97957629e-01 1.11189507e-01 5.18674254e-01
-2.18841597e-01 2.78515041e-01 -1.98627368e-01 -4.63876516e-01
5.00986993e-01 -2.39632279e-01 -6.20003939e-01 -3.37691009e-01
8.97037685e-01 -2.61466682e-01 -3.36787581e-01 1.04729569e+00
4.64880943e-01 9.72779095e-03 4.72792357e-01 -9.54436660e-01
-7.39448726e-01 1.00006616e+00 -5.24606287e-01 1.48291811e-01
-2.06215773e-02 1.38491675e-01 1.08762276e+00 -8.77892256e-01
4.24697846e-01 1.39175236e+00 9.39964652e-01 7.28189826e-01
-1.75116718e+00 -1.13070476e+00 4.04622674e-01 -1.90038696e-01
-1.50552380e+00 -6.86141670e-01 5.68850935e-01 -5.05502343e-01
7.46748745e-01 1.73547417e-01 1.39629498e-01 1.35186017e+00
-4.06791806e-01 1.01189339e+00 1.38609302e+00 -3.57134223e-01
-1.66400149e-01 4.45076257e-01 4.58731472e-01 4.91231740e-01
2.18945280e-01 1.80539921e-01 -5.85288107e-01 -1.19051598e-01
5.76972544e-01 1.54359974e-02 -1.11540668e-02 -1.86994180e-01
-1.22026050e+00 9.21906888e-01 3.23374242e-01 3.58035862e-01
-9.51046199e-02 -4.88918684e-02 2.89255202e-01 4.12209213e-01
7.56230712e-01 5.95947981e-01 -8.02045822e-01 -1.47206172e-01
-1.05577886e+00 2.31788412e-01 4.03890103e-01 7.44761527e-01
7.52733588e-01 2.35452905e-01 -3.67293693e-02 1.37103021e+00
-1.35370847e-02 6.30345881e-01 7.02862799e-01 -5.98523915e-01
5.93265712e-01 6.52026176e-01 -2.02972397e-01 -2.16936305e-01
-2.90245920e-01 -5.98817110e-01 -7.10561812e-01 -1.05679467e-01
5.91091573e-01 -3.20106298e-01 -1.09786785e+00 2.30581284e+00
1.70940772e-01 1.49117157e-01 6.60764500e-02 5.32519877e-01
7.29324698e-01 6.97739542e-01 1.61803424e-01 1.91056028e-01
1.20398915e+00 -9.32240546e-01 -5.06352663e-01 -4.93826866e-01
8.59311283e-01 -6.99463427e-01 1.64275706e+00 2.27338925e-01
-7.32339144e-01 -5.22614062e-01 -9.09248888e-01 -7.57583827e-02
-6.25633836e-01 1.22586012e-01 8.04052174e-01 7.99315691e-01
-9.10386324e-01 1.88136339e-01 -8.24905634e-01 -4.26467896e-01
3.35275471e-01 4.72136140e-01 -5.82563937e-01 -2.40590572e-01
-1.20044911e+00 6.15527570e-01 3.94862771e-01 -3.36837262e-01
-9.04261649e-01 -1.13080990e+00 -8.62021208e-01 -2.16539472e-01
2.46543661e-01 -2.91675746e-01 1.34394503e+00 -8.96631598e-01
-1.26647460e+00 1.06682241e+00 -3.48375171e-01 -3.92557234e-01
5.72719038e-01 -4.28909987e-01 -5.11558771e-01 -5.93832552e-01
2.03622296e-01 6.86312139e-01 4.43674475e-01 -1.17020404e+00
-4.70043659e-01 -2.94753283e-01 -2.00675264e-01 2.13490158e-01
-5.66107571e-01 1.75451845e-01 -5.06699920e-01 -9.23671186e-01
-1.21359378e-01 -1.01834249e+00 -1.28014624e-01 -5.52401602e-01
-4.67392623e-01 -3.10515109e-02 3.84169579e-01 -8.24983537e-01
1.33825719e+00 -2.09911013e+00 -5.53060672e-04 8.89752209e-02
-2.49940553e-04 4.56308722e-01 -2.76848286e-01 3.11684340e-01
-2.56728768e-01 2.83184081e-01 -3.73966217e-01 -5.91445386e-01
-7.05655515e-02 6.04077317e-02 -4.16876763e-01 2.57010549e-01
8.42738710e-03 7.97219038e-01 -6.99498177e-01 1.57522748e-03
1.81741402e-01 3.86992067e-01 -6.86796963e-01 1.18945055e-01
-1.90483108e-01 5.79848289e-01 -7.83400517e-03 7.80877709e-01
8.12956750e-01 -8.77799541e-02 2.16407925e-01 6.49318919e-02
-1.85784250e-01 5.17866373e-01 -1.01289451e+00 1.64507091e+00
-5.88908494e-01 6.10927999e-01 -9.40109044e-02 -4.88199681e-01
9.27164257e-01 5.21756411e-02 2.58665055e-01 -5.57291687e-01
-1.88555852e-01 3.26832980e-01 -6.16429932e-02 9.70757455e-02
5.29750109e-01 -1.77422866e-01 -2.84445345e-01 6.46420300e-01
1.40247598e-01 1.81035653e-01 -4.75684032e-02 6.55065626e-02
7.20361471e-01 3.71513218e-02 2.24550024e-01 -4.86047477e-01
3.52370054e-01 5.93020171e-02 6.71488106e-01 9.00478363e-01
6.49087355e-02 5.91154516e-01 1.07233912e-01 -3.60572457e-01
-1.14896917e+00 -1.30404592e+00 -4.88426983e-01 1.42044342e+00
-2.69048631e-01 -5.41136444e-01 -7.17217803e-01 -7.20009387e-01
2.96015233e-01 7.06808448e-01 -4.43522811e-01 -2.11144015e-01
-6.73105597e-01 -1.28278267e+00 7.83543944e-01 4.92814660e-01
4.16056424e-01 -7.80218840e-01 5.65226138e-01 1.03742316e-01
-2.38434970e-01 -1.12648535e+00 -7.59588778e-01 2.85484701e-01
-5.40555954e-01 -6.43605590e-01 -7.30295539e-01 -7.87375510e-01
5.07211983e-01 1.13614842e-01 1.14625692e+00 -1.60524786e-01
1.83859672e-02 4.85581718e-02 -7.05340803e-02 -2.98023701e-01
-6.93432391e-01 7.07426012e-01 5.41567385e-01 1.32302716e-01
8.46756518e-01 -3.69756252e-01 -2.71579176e-01 3.18948388e-01
-6.19697332e-01 1.00138001e-01 3.61336410e-01 7.79099822e-01
5.67936540e-01 -2.33116493e-01 6.54365301e-01 -1.23111320e+00
5.27128398e-01 -6.57179117e-01 -7.26723075e-01 3.00527126e-01
-6.92674279e-01 2.59161800e-01 5.88881016e-01 -7.50688732e-01
-8.08364689e-01 1.46399423e-01 -2.50294000e-01 -2.40071192e-01
-2.07976475e-01 2.67134875e-01 -2.05898657e-01 5.42009100e-02
7.71747410e-01 1.04695298e-01 -8.01408216e-02 -1.02194905e+00
4.13444608e-01 9.46203232e-01 6.09785616e-01 -6.86093509e-01
9.97526407e-01 2.16105282e-01 -5.03849566e-01 -7.30799615e-01
-8.09591651e-01 -5.10604143e-01 -8.17562819e-01 2.59725690e-01
5.40782034e-01 -1.36506462e+00 -3.26492965e-01 6.07403159e-01
-5.47010183e-01 -9.13144052e-01 -1.06816620e-01 5.68177819e-01
-2.38333285e-01 5.82709946e-02 -6.29393637e-01 -6.58732235e-01
-3.52332294e-01 -1.07220352e+00 8.34348500e-01 3.83714661e-02
-2.72343516e-01 -1.19546711e+00 2.68009812e-01 2.96502352e-01
4.19350117e-01 -1.54404286e-02 8.83050025e-01 -9.19655740e-01
-4.68945593e-01 -3.08936238e-02 -5.51755540e-02 5.15220046e-01
4.19460297e-01 -1.04776144e-01 -1.20896816e+00 -4.38104540e-01
-4.92453575e-01 -4.18946356e-01 1.12261891e+00 4.11510468e-01
9.74300385e-01 -3.54988545e-01 -2.54332989e-01 9.98998106e-01
1.25559199e+00 -2.68461853e-01 -9.78002623e-02 3.49436671e-01
9.49427843e-01 5.77712238e-01 3.64966840e-01 6.98835263e-03
5.93407214e-01 8.97805631e-01 -1.41149208e-01 -3.68842721e-01
-2.26136163e-01 -6.30492568e-01 5.75844467e-01 9.85790670e-01
3.26108217e-01 -8.53056684e-02 -1.30485713e+00 7.53390670e-01
-1.62841558e+00 -7.62498617e-01 -5.21451011e-02 2.53604960e+00
1.19732785e+00 1.08305067e-01 4.63063538e-01 -1.50130346e-01
7.26824880e-01 -1.11388844e-02 -6.39083326e-01 -3.38837773e-01
-6.25777960e-01 1.38497725e-01 8.28902662e-01 1.03881276e+00
-1.02537930e+00 1.56809890e+00 6.36166668e+00 7.65599012e-01
-1.21301639e+00 2.71463156e-01 6.33010685e-01 -2.54488915e-01
-4.28270340e-01 -2.10349020e-02 -1.37497938e+00 4.41458017e-01
1.04361844e+00 -2.56129444e-01 7.97711194e-01 6.58653200e-01
4.71588261e-02 2.75175691e-01 -1.03999388e+00 1.04514289e+00
-3.30203287e-02 -9.90257978e-01 7.63921812e-02 3.47450942e-01
7.24391878e-01 6.96909666e-01 5.02126515e-01 4.80424374e-01
9.31331813e-01 -9.16961610e-01 5.14805079e-01 -5.95769137e-02
1.18741751e+00 -7.79993594e-01 2.98716426e-01 2.88672179e-01
-9.95185316e-01 2.62992606e-02 -2.16500238e-01 3.37574393e-01
2.43430566e-02 3.77394289e-01 -9.72191572e-01 2.47741073e-01
5.55124342e-01 6.70749128e-01 -9.84573781e-01 6.79975748e-01
-1.05464526e-01 1.04903293e+00 -5.25325596e-01 4.64139938e-01
1.57211602e-01 -1.34478837e-01 5.29012620e-01 1.42803800e+00
1.90381050e-01 -5.07520676e-01 6.55924976e-02 6.92856729e-01
-3.62303823e-01 1.54194117e-01 -7.02114463e-01 -1.63476050e-01
5.54676294e-01 1.15419996e+00 -1.25110507e-01 -3.33708793e-01
-4.52470958e-01 8.85961890e-01 5.30427992e-01 4.37085599e-01
-4.31999505e-01 3.79772224e-02 1.11243558e+00 2.27208957e-01
-7.93856755e-02 -3.17231059e-01 -3.14674824e-01 -1.55580175e+00
-2.12270632e-01 -1.21724486e+00 7.54671872e-01 -2.80249119e-01
-1.46077919e+00 7.13445544e-01 2.85553113e-02 -9.34029043e-01
-5.17834604e-01 -7.80641735e-01 -2.57446110e-01 1.21055615e+00
-1.52173471e+00 -1.47860682e+00 -1.40810266e-01 5.94142497e-01
5.94169021e-01 -5.88664353e-01 9.49790239e-01 4.96311277e-01
-7.26987779e-01 1.20321333e+00 4.70031649e-01 2.77281195e-01
1.08623457e+00 -1.44505620e+00 9.53599334e-01 9.73503649e-01
4.24287885e-01 8.88047636e-01 4.33534980e-01 -6.20111048e-01
-1.07990170e+00 -1.09955144e+00 9.93297398e-01 -7.63593674e-01
7.01125085e-01 -1.07179952e+00 -1.10801589e+00 1.09702504e+00
6.89566508e-02 -2.04475492e-01 1.00998807e+00 7.44140863e-01
-8.89650404e-01 -2.31394112e-01 -9.13734078e-01 7.57423639e-01
9.65135038e-01 -8.41011584e-01 -2.93022037e-01 4.61044312e-01
6.62780702e-01 -1.98639810e-01 -7.96204925e-01 4.01166320e-01
5.75375795e-01 -5.48542440e-01 1.09951866e+00 -7.47579396e-01
-6.12768158e-02 -1.30057588e-01 -4.66261357e-01 -1.46161163e+00
-3.42305630e-01 -6.79606378e-01 1.13083713e-01 1.54742324e+00
7.19461381e-01 -8.26658070e-01 6.67815149e-01 3.74577701e-01
1.35566220e-01 -2.07990170e-01 -6.15740538e-01 -9.99193549e-01
4.98871595e-01 -4.44377899e-01 8.14774752e-01 1.39860451e+00
-3.64215493e-01 4.16570842e-01 -5.73656857e-01 2.13054657e-01
6.73124552e-01 -2.46282872e-02 1.01572800e+00 -1.09700775e+00
-4.33638573e-01 -5.51628530e-01 3.21715288e-02 -8.99712801e-01
4.20610428e-01 -1.45082295e+00 -2.99148917e-01 -1.00614023e+00
1.83878943e-01 -9.00262654e-01 -4.64264005e-01 6.93366110e-01
-4.75472629e-01 4.11703765e-01 2.32053906e-01 2.89261878e-01
-7.08323270e-02 2.81400770e-01 6.51414096e-01 -6.54071644e-02
-5.50412953e-01 1.66279510e-01 -9.94622409e-01 4.30673957e-01
1.18744838e+00 -4.54041481e-01 -2.67137349e-01 -6.13011837e-01
-6.27720430e-02 -5.05681574e-01 -1.03156455e-01 -8.53664756e-01
-6.40638694e-02 -3.83594967e-02 2.68718630e-01 -5.01264155e-01
4.52965915e-01 -4.90550727e-01 1.56307593e-01 2.82411397e-01
-4.73373026e-01 3.20657492e-01 6.23891294e-01 4.49328683e-02
-3.88089083e-02 1.78660136e-02 9.24319327e-01 -8.61753896e-02
-5.58382750e-01 4.74048018e-01 -1.47616148e-01 3.52763385e-01
5.27556419e-01 1.51621535e-01 -4.00539577e-01 -3.43634695e-01
-4.74688619e-01 2.10435480e-01 6.14313126e-01 8.60019743e-01
-6.66496679e-02 -1.52887881e+00 -1.11811280e+00 4.94978994e-01
4.12925988e-01 -2.17953190e-01 3.32144462e-02 4.95126635e-01
-2.19387725e-01 4.46913540e-01 -2.61699129e-02 -5.93862534e-01
-1.38004267e+00 5.24536908e-01 5.11513531e-01 -2.45664597e-01
-3.74580026e-01 8.56474161e-01 7.28372216e-01 -1.13007414e+00
1.17507845e-01 -2.22342253e-01 7.40851387e-02 1.52588368e-03
6.57286286e-01 9.70207527e-02 -2.66210567e-02 -8.73037159e-01
-3.92992467e-01 6.26141548e-01 -5.46760261e-01 -3.41520518e-01
1.31763589e+00 -8.75262022e-02 1.23975225e-01 7.86240339e-01
1.28160393e+00 4.39393342e-01 -1.13005078e+00 -7.72073984e-01
8.41531456e-02 -3.49880993e-01 -1.45586785e-02 -1.05440211e+00
-7.39744127e-01 7.73644090e-01 6.37590587e-01 -1.02707461e-01
8.89781415e-01 3.30262333e-02 5.62758744e-01 6.59935921e-02
6.10366091e-02 -9.13582861e-01 -3.32099915e-01 5.36311686e-01
6.79983735e-01 -1.55482864e+00 -2.19967142e-01 -1.46990672e-01
-7.41292238e-01 5.97809255e-01 7.16794252e-01 1.46625564e-01
5.35183966e-01 3.79400373e-01 4.52629477e-01 2.66395777e-01
-6.37714565e-01 -2.67466635e-01 4.30596381e-01 4.68302995e-01
5.97003877e-01 4.64177400e-01 2.58138448e-01 5.95179975e-01
-6.36106730e-01 -3.45636934e-01 7.32822046e-02 4.96613950e-01
-8.15479308e-02 -1.51096547e+00 -3.32918316e-01 4.74686891e-01
-5.23544133e-01 -5.68026543e-01 -5.16183317e-01 7.36898184e-01
-6.52020648e-02 6.00849569e-01 1.66173473e-01 -6.11227930e-01
1.52566046e-01 4.70694423e-01 2.38625377e-01 -6.86084807e-01
-8.22056055e-01 8.59018639e-02 1.29101694e-01 -2.37558842e-01
-1.82524994e-02 -9.70665336e-01 -8.46089542e-01 -6.02716684e-01
-1.03336629e-02 -1.60408989e-01 7.46360958e-01 5.22081912e-01
3.35285991e-01 1.10677652e-01 7.10994720e-01 -2.27484658e-01
-3.54215592e-01 -1.11328435e+00 -2.42752984e-01 4.68964875e-01
6.42710268e-01 -4.37429935e-01 -3.10323685e-01 -1.20064884e-01]
|
[10.99407958984375, 9.96546459197998]
|
835c0e5f-d257-4d83-80c3-db12ece8d66c
|
one-general-teacher-for-multi-data-multi-task
| null | null |
https://openreview.net/forum?id=z5IgrlFV_e
|
https://openreview.net/pdf?id=z5IgrlFV_e
|
One General Teacher for Multi-Data Multi-Task: A New Knowledge Distillation Framework for Discourse Relation Analysis
|
Automatically identifying the discourse relations can help many downstream NLP tasks such as reading comprehension. It can be categorized into explicit and implicit discourse relation recognition (EDRR and IDRR). Due to the lack of connectives, IDRR remains to be a big challenge. A good number of methods have been developed to combine explicit data with implicit ones under the multi-task learning framework. However, the difference in linguistic property and class distribution makes it hard to directly optimize EDRR and IDRR with multi-task learning.
In this paper, we take the first step to exploit the knowledge distillation (KD) technique for discourse relation analysis. Our target is to train a focused single-data single-task student with the help of a general multi-data multi-task teacher. Specifically, we first train one teacher for both the top and second level relation classification tasks with explicit and implicit data. We then transfer the feature embeddings and soft labels from the teacher network to the student network. Extensive experimental results on the popular PDTB dataset proves that our model achieves a new state-of-the-art performance. We also show the effectiveness of our proposed KD architecture through detailed analysis.
|
['Anonymous']
|
2021-11-16
| null | null | null |
acl-arr-november-2021-11
|
['relation-classification']
|
['natural-language-processing']
|
[ 2.41479948e-01 7.52018631e-01 -3.62713456e-01 -5.21588743e-01
-7.82896161e-01 -4.61140722e-01 7.17022955e-01 2.06148952e-01
-3.65703017e-01 8.33618939e-01 3.33062947e-01 -6.24606669e-01
-2.94495344e-01 -6.55884564e-01 -5.18001199e-01 -7.90278316e-01
2.06947237e-01 5.40920973e-01 2.85561889e-01 -2.05782309e-01
-7.77562857e-02 -1.88970324e-02 -1.18102443e+00 5.87476313e-01
1.14819300e+00 9.66979861e-01 2.69467324e-01 5.53625822e-01
-2.67563730e-01 1.41450822e+00 -5.66711962e-01 -4.55198109e-01
-3.94004315e-01 -2.48634741e-01 -1.51213479e+00 -3.47330809e-01
7.87838455e-03 -1.89776972e-01 -8.17141235e-02 7.24366903e-01
5.84746957e-01 3.74691874e-01 7.65784323e-01 -9.49941278e-01
-1.01330209e+00 1.00538659e+00 -4.64178532e-01 2.52139330e-01
2.99466491e-01 -3.14356118e-01 1.30505931e+00 -8.37135971e-01
4.09341514e-01 1.43172920e+00 3.04607958e-01 4.58816707e-01
-1.03677630e+00 -4.65531200e-01 4.13853616e-01 7.14640975e-01
-1.12368238e+00 -3.71443540e-01 9.31626916e-01 -5.82963347e-01
1.08084393e+00 2.03584462e-01 5.88417388e-02 1.02666771e+00
-2.87604779e-01 1.03216064e+00 1.11849678e+00 -5.83833575e-01
-2.06094816e-01 2.23538831e-01 6.96783781e-01 5.84300041e-01
-1.81388006e-01 -3.35110128e-01 -5.79387128e-01 1.56723231e-01
2.03328937e-01 -4.35171723e-01 -5.26371419e-01 7.36939348e-03
-1.07980371e+00 7.81277239e-01 4.18374687e-01 4.92416054e-01
5.54709183e-03 -2.03447804e-01 4.84180897e-01 5.47236443e-01
7.39842236e-01 3.46108258e-01 -6.81890905e-01 -1.49054125e-01
-1.34286627e-01 -3.77663374e-02 8.99397790e-01 7.06924736e-01
7.13883579e-01 -3.63609165e-01 -3.55434269e-01 1.17746794e+00
2.13483587e-01 7.80885667e-03 4.98810202e-01 -6.18103385e-01
8.33292007e-01 8.28377485e-01 -3.85932535e-01 -1.10442328e+00
-3.27077746e-01 -1.07726730e-01 -9.61635828e-01 -5.70648722e-02
4.38839078e-01 -3.01027536e-01 -3.15627545e-01 1.67258143e+00
5.63016415e-01 1.75598651e-01 4.94507074e-01 8.14844549e-01
1.29726875e+00 7.30021894e-01 1.56919241e-01 -3.68989646e-01
1.56446743e+00 -1.46664906e+00 -1.04291034e+00 -1.62848860e-01
1.08265054e+00 -6.19081616e-01 9.25597191e-01 2.28010729e-01
-8.13181221e-01 -5.14064074e-01 -8.90388668e-01 -6.16234541e-01
-3.56082231e-01 2.65364826e-01 6.23943806e-01 1.11880668e-01
-5.49920321e-01 2.68067718e-01 -5.21839023e-01 3.77480537e-02
6.33244216e-01 2.44843259e-01 -4.07037377e-01 4.73132469e-02
-1.44111824e+00 1.23764527e+00 7.43113935e-01 2.52000362e-01
-5.14322937e-01 -6.80362165e-01 -9.19667363e-01 1.96093053e-01
7.38504410e-01 -3.58547688e-01 1.22595680e+00 -6.83375537e-01
-1.89079285e+00 9.78674114e-01 -9.23158154e-02 -2.41153553e-01
2.56287813e-01 -5.71447372e-01 5.91881983e-02 -6.34911470e-03
-1.86337441e-01 7.72234946e-02 3.78520161e-01 -1.09122896e+00
-6.46638811e-01 -1.84773043e-01 3.29037249e-01 5.49823403e-01
-4.44609314e-01 1.72835514e-01 1.71517059e-01 -4.17584062e-01
-3.68245319e-02 -5.12743354e-01 3.10892165e-01 -4.01232839e-01
-5.52503467e-01 -1.24291873e+00 9.39860761e-01 -6.36471868e-01
1.29776824e+00 -1.94822907e+00 6.20595753e-01 -3.94736022e-01
4.09500897e-01 7.11332381e-01 -1.92447826e-02 1.77004755e-01
-3.18023443e-01 1.20623544e-01 1.44316435e-01 -3.86204809e-01
-9.37521979e-02 1.53529197e-01 -3.48659486e-01 1.66593313e-01
5.85060418e-01 9.86926258e-01 -1.06786501e+00 -6.32797062e-01
3.04852091e-02 1.61558717e-01 -1.38058901e-01 8.07303905e-01
-3.86917502e-01 5.54651320e-01 -6.11467004e-01 2.66905963e-01
3.78215879e-01 -4.35841411e-01 4.66216266e-01 -3.43645632e-01
-5.64183779e-02 9.71095264e-01 -7.83823431e-01 1.54965878e+00
-5.23791254e-01 7.38168836e-01 -1.05576897e-02 -1.66108227e+00
9.38597918e-01 6.68075025e-01 1.16931915e-01 -4.91228729e-01
2.85750002e-01 2.13951617e-01 3.99833202e-01 -8.68304491e-01
2.01162472e-01 -1.91522524e-01 -8.23525637e-02 6.24768913e-01
1.90304667e-01 2.78482884e-01 -1.71693236e-01 3.28105949e-02
1.04676783e+00 1.34731218e-01 4.84048575e-01 -1.49515107e-01
8.10708463e-01 -2.01285958e-01 4.98915523e-01 1.60054252e-01
3.61729078e-02 1.58017009e-01 8.23667109e-01 -3.45570743e-01
-4.40980852e-01 -6.99941218e-01 -1.34325111e-02 1.59547400e+00
1.18497647e-01 -1.82821214e-01 -3.36192608e-01 -8.81164253e-01
-1.08790085e-01 5.94736993e-01 -5.90943158e-01 -1.19090900e-01
-1.04910529e+00 -6.77170932e-01 5.42595148e-01 4.55059409e-01
7.60190547e-01 -1.01949573e+00 -3.26935083e-01 1.92067087e-01
-2.54174888e-01 -1.37633097e+00 -1.80995651e-02 4.78344440e-01
-2.93638736e-01 -1.15413535e+00 -2.46344239e-01 -1.25571656e+00
3.11799318e-01 2.47796223e-01 9.74908531e-01 1.00208685e-01
2.75260925e-01 -1.56761006e-01 -6.40284956e-01 -1.39126763e-01
-3.05338293e-01 5.05096018e-01 -2.49936268e-01 1.18898344e-03
3.93254697e-01 -6.29575670e-01 -1.13158487e-01 4.93575893e-02
-4.20866758e-01 5.01119554e-01 4.90302533e-01 9.97019529e-01
3.58104676e-01 -2.87004579e-02 9.81112659e-01 -1.10458255e+00
9.30257440e-01 -4.83617187e-01 -2.08552644e-01 7.70855308e-01
-3.03754151e-01 2.57893324e-01 5.15950561e-01 -7.25361645e-01
-1.35351205e+00 -4.11555499e-01 5.75740784e-02 -2.09373698e-01
-1.60512496e-02 8.90134037e-01 -5.57944953e-01 3.24477643e-01
3.30293626e-01 -2.51489103e-01 -1.32812515e-01 -5.62057376e-01
6.82293236e-01 9.96751904e-01 3.84506941e-01 -8.89384031e-01
5.34248412e-01 -2.23770902e-01 -2.48795524e-01 -7.79680669e-01
-1.57498300e+00 -2.24097118e-01 -8.35304976e-01 -2.00640932e-02
1.27342880e+00 -8.45626831e-01 -1.13959873e+00 4.76228088e-01
-1.52062011e+00 -7.07721055e-01 -9.04414505e-02 2.44154349e-01
-2.63959289e-01 9.50015038e-02 -4.58726525e-01 -8.19373786e-01
-3.35966796e-01 -1.08571160e+00 7.01377928e-01 3.41975182e-01
-1.93117291e-01 -1.32223248e+00 3.75380814e-02 6.18001580e-01
1.92412883e-01 1.06241770e-01 1.41655290e+00 -1.16406941e+00
-5.42747557e-01 4.67120081e-01 -5.37978053e-01 3.02118063e-01
4.33077544e-01 -2.98671544e-01 -1.21776640e+00 2.03222349e-01
9.13217217e-02 -1.00158513e+00 8.78749430e-01 -3.44416965e-03
1.24987471e+00 -3.65072370e-01 -4.17505473e-01 3.76008630e-01
8.82341504e-01 2.59207450e-02 3.12956214e-01 2.81366676e-01
1.16892886e+00 9.13847387e-01 7.15851188e-01 -1.94597691e-01
9.30468202e-01 7.70787716e-01 4.24409658e-02 9.25428271e-02
-3.80598634e-01 1.52316401e-02 3.08312386e-01 1.40588176e+00
-4.96370941e-02 -2.21857935e-01 -1.19806242e+00 6.21756017e-01
-2.14982080e+00 -8.65317583e-01 -1.28259838e-01 1.48485816e+00
1.63726974e+00 7.26612732e-02 -2.04510435e-01 1.05097853e-01
5.92587769e-01 3.23061615e-01 -2.29716286e-01 -5.81844389e-01
-5.14419302e-02 2.85435468e-01 -2.23213896e-01 7.92257845e-01
-1.30967593e+00 1.13127708e+00 4.30365038e+00 9.65801656e-01
-1.10118520e+00 2.82322079e-01 6.48106813e-01 2.55635232e-01
-9.86228287e-02 -6.82849586e-02 -9.76785302e-01 5.82418777e-02
9.30676877e-01 -2.47867256e-01 1.01611741e-01 6.40508831e-01
-2.73819357e-01 -1.18274704e-01 -1.37011695e+00 7.43741095e-01
4.78727520e-02 -1.25534570e+00 -1.95990130e-01 -2.01123342e-01
5.00631571e-01 -3.09776694e-01 -2.13998780e-01 7.25774944e-01
3.29885095e-01 -1.32036793e+00 2.54924119e-01 3.39308023e-01
7.95869827e-01 -4.25539106e-01 8.33498001e-01 6.37690544e-01
-1.18384111e+00 7.02810884e-02 -1.71535268e-01 -5.29726982e-01
7.63037875e-02 4.95891541e-01 -1.01357353e+00 8.39381933e-01
2.21777543e-01 7.79937804e-01 -3.06268930e-01 3.56413901e-01
-7.57401466e-01 6.16108298e-01 -1.40973285e-01 -3.08532745e-01
-6.37544468e-02 -1.41029194e-01 2.62050986e-01 1.08979011e+00
-2.00013489e-01 5.26309788e-01 3.83731961e-01 6.15720212e-01
-3.35073024e-01 -2.37662327e-02 -3.82866234e-01 -1.90934055e-02
6.66430831e-01 1.24365258e+00 -1.44399419e-01 -3.24486673e-01
-3.93066853e-01 5.06706595e-01 1.22145629e+00 1.85548216e-01
-6.06215656e-01 -3.52400780e-01 6.25347376e-01 -4.44653869e-01
2.45863676e-01 -1.50064647e-01 -3.44696879e-01 -1.16652477e+00
-2.40633991e-02 -8.43235910e-01 3.94792914e-01 -3.83505225e-01
-1.47639525e+00 6.08052850e-01 2.05229521e-01 -6.80341542e-01
-2.20795140e-01 -6.39015377e-01 -7.48441935e-01 1.04958272e+00
-1.89510536e+00 -1.31580794e+00 -3.17288548e-01 2.59930670e-01
5.96063316e-01 -2.19511554e-01 1.06055033e+00 2.75356472e-01
-8.69073331e-01 6.81618035e-01 -2.83356726e-01 5.48220396e-01
7.19108343e-01 -1.43659031e+00 -1.86620161e-01 3.72619241e-01
-1.07045516e-01 4.65692341e-01 3.67533118e-01 -4.08717364e-01
-1.17613173e+00 -9.64768767e-01 1.11844909e+00 -3.67986202e-01
9.89156544e-01 -3.28354448e-01 -1.36821640e+00 8.75725746e-01
5.85749865e-01 -1.09429814e-01 8.84715378e-01 7.14184523e-01
-3.45010251e-01 -5.49265333e-02 -6.25620782e-01 3.59322280e-01
8.29687834e-01 -6.60239518e-01 -1.22714579e+00 4.19850528e-01
1.12589371e+00 -6.74668729e-01 -1.12778533e+00 5.18097103e-01
2.29297653e-01 -4.53004450e-01 7.42759883e-01 -8.43448877e-01
8.60506713e-01 1.19476179e-02 1.99819300e-02 -1.33417261e+00
-5.10740615e-02 -4.42942023e-01 -3.95118684e-01 1.81275237e+00
5.21427512e-01 -6.36367679e-01 3.11677426e-01 4.72241133e-01
-2.48529822e-01 -1.29124033e+00 -9.22534585e-01 -4.75710630e-01
4.69791263e-01 -8.36074129e-02 2.49686211e-01 1.24993110e+00
4.34711754e-01 1.24970484e+00 -2.30461042e-02 1.35258228e-01
1.76294073e-01 4.97342050e-01 6.92872822e-01 -1.28692675e+00
-2.87980884e-01 -4.32357788e-01 2.50967685e-02 -1.39085174e+00
7.36052155e-01 -1.14603901e+00 -9.80181526e-03 -1.58775401e+00
1.30207822e-01 -8.24894428e-01 -7.60619044e-02 7.07182765e-01
-6.96093202e-01 -4.49256629e-01 -1.15946442e-01 6.81850165e-02
-7.19046175e-01 9.27868307e-01 1.34011173e+00 -4.65062410e-02
-1.82743162e-01 -1.21346660e-01 -6.59330904e-01 7.81845152e-01
7.38779902e-01 -3.74084085e-01 -6.55269265e-01 -8.18508208e-01
2.35785469e-01 2.00922824e-02 2.42037401e-01 -4.23893660e-01
3.19917440e-01 -2.17337698e-01 -1.53615654e-01 -4.66200829e-01
3.33680540e-01 -4.89875734e-01 -5.36065936e-01 -1.26130534e-02
-6.40289783e-01 -3.81631732e-01 1.05694309e-02 2.95124739e-01
-3.99591655e-01 -2.80653119e-01 5.20087242e-01 1.73940584e-01
-4.62038100e-01 -2.01652139e-01 1.03986681e-01 4.57517356e-01
1.05069268e+00 4.42961931e-01 -1.04250729e+00 -1.32973036e-02
-7.08528161e-01 6.99015498e-01 -2.81363279e-01 3.51338446e-01
4.67557281e-01 -1.30380118e+00 -8.57266843e-01 -1.04881048e-01
-1.27080515e-01 8.53183329e-01 -7.50357751e-03 9.95719910e-01
-3.08190994e-02 4.28224802e-01 1.41668439e-01 -5.20571291e-01
-1.60928273e+00 3.65967661e-01 2.38495708e-01 -6.98285162e-01
-5.82439005e-01 1.21201730e+00 2.51643807e-01 -5.39923072e-01
3.50949526e-01 -3.61911297e-01 -6.97488606e-01 5.09553194e-01
4.89240021e-01 7.59314373e-02 -1.31157860e-01 -5.11613071e-01
-2.52534211e-01 4.46114093e-01 -3.60094607e-01 2.83640236e-01
1.53751075e+00 -8.82804841e-02 -3.41117859e-01 7.33915567e-01
1.25244308e+00 -2.00274870e-01 -8.32322776e-01 -5.30405045e-01
4.44280148e-01 1.57905277e-02 -3.77555639e-02 -6.77590787e-01
-7.00945556e-01 1.30624175e+00 -1.97863221e-01 4.48086321e-01
9.00202632e-01 3.49283159e-01 6.72256470e-01 5.07235825e-01
-2.69212131e-03 -8.08403075e-01 2.92696923e-01 9.28695321e-01
8.29557538e-01 -1.31726718e+00 6.47059157e-02 -7.78364003e-01
-6.85813963e-01 1.07052016e+00 9.52038288e-01 5.98270223e-02
7.00655520e-01 2.28067577e-01 1.45312980e-01 -2.75129467e-01
-1.16530812e+00 -2.38454387e-01 4.61941749e-01 5.41756690e-01
8.89958441e-01 9.16439220e-02 -3.30282837e-01 8.26079190e-01
-2.25244701e-01 -1.71256199e-01 2.16364205e-01 7.00007260e-01
-4.55956280e-01 -1.35810757e+00 1.08580120e-01 2.07610846e-01
-2.92617351e-01 2.16073189e-02 -3.55238587e-01 5.98626137e-01
6.99765459e-02 1.00249958e+00 -1.15806475e-01 -4.10689116e-01
2.77970403e-01 3.58871967e-01 5.03496945e-01 -9.48086739e-01
-7.33885825e-01 -3.79258871e-01 6.74916506e-01 -1.77946806e-01
-7.66703606e-01 -4.07940328e-01 -1.40677905e+00 -5.41696697e-02
-5.95044076e-01 2.48808563e-01 1.22364275e-01 1.48943126e+00
1.06568322e-01 1.05708027e+00 3.76926392e-01 -4.39444274e-01
-6.63061976e-01 -1.36160910e+00 -7.58490432e-03 3.19709122e-01
4.28738147e-01 -8.76848102e-01 -2.85908490e-01 9.62036848e-02]
|
[10.689582824707031, 9.282344818115234]
|
574904dc-2d57-418d-8491-40293b319e98
|
robust-pose-transfer-with-dynamic-details
|
2106.14132
| null |
https://arxiv.org/abs/2106.14132v3
|
https://arxiv.org/pdf/2106.14132v3.pdf
|
Robust Pose Transfer with Dynamic Details using Neural Video Rendering
|
Pose transfer of human videos aims to generate a high fidelity video of a target person imitating actions of a source person. A few studies have made great progress either through image translation with deep latent features or neural rendering with explicit 3D features. However, both of them rely on large amounts of training data to generate realistic results, and the performance degrades on more accessible internet videos due to insufficient training frames. In this paper, we demonstrate that the dynamic details can be preserved even trained from short monocular videos. Overall, we propose a neural video rendering framework coupled with an image-translation-based dynamic details generation network (D2G-Net), which fully utilizes both the stability of explicit 3D features and the capacity of learning components. To be specific, a novel texture representation is presented to encode both the static and pose-varying appearance characteristics, which is then mapped to the image space and rendered as a detail-rich frame in the neural rendering stage. Moreover, we introduce a concise temporal loss in the training stage to suppress the detail flickering that is made more visible due to high-quality dynamic details generated by our method. Through extensive comparisons, we demonstrate that our neural human video renderer is capable of achieving both clearer dynamic details and more robust performance even on accessible short videos with only 2k - 4k frames.
|
['Lin Gao', 'Wei Liu', 'Yu-Kun Lai', 'Xuan Wang', 'Hao-Zhi Huang', 'Yang-tian Sun']
|
2021-06-27
| null | null | null | null |
['pose-transfer']
|
['computer-vision']
|
[ 3.66398335e-01 -1.47123873e-01 2.03123704e-01 -1.34385183e-01
-5.31746447e-01 -3.53089154e-01 7.87357390e-01 -6.59817815e-01
-9.60651636e-02 7.55212367e-01 2.34137416e-01 2.36658975e-01
3.73799652e-01 -6.35390162e-01 -1.03081179e+00 -7.85550833e-01
-4.35280707e-03 -7.08408579e-02 1.53056264e-01 -1.63259447e-01
-1.22627147e-01 4.08724189e-01 -1.76166499e+00 2.89967239e-01
6.76950693e-01 1.16940689e+00 2.87320428e-02 7.04353094e-01
1.93534315e-01 8.26168358e-01 -6.31800354e-01 -4.14738864e-01
6.22380257e-01 -5.12342036e-01 -2.66934186e-01 4.14247513e-01
9.53606844e-01 -9.41687822e-01 -9.13443983e-01 7.98055530e-01
5.24376810e-01 2.73100913e-01 3.92239213e-01 -1.19450843e+00
-7.48054206e-01 -1.56537220e-01 -3.90560269e-01 -6.49785921e-02
6.50040865e-01 6.55313492e-01 4.74016577e-01 -7.98296869e-01
7.06138551e-01 1.40626037e+00 4.70406264e-01 8.40022087e-01
-1.16885555e+00 -7.27391779e-01 2.72704214e-01 2.13164389e-01
-1.13474655e+00 -4.94666576e-01 9.48490262e-01 -3.37429285e-01
6.91185355e-01 4.12456304e-01 9.92400229e-01 1.55695760e+00
2.45884717e-01 6.06707036e-01 1.11974239e+00 -2.00768709e-01
-5.62272966e-02 -1.58483133e-01 -4.99564350e-01 7.26244867e-01
2.70427600e-03 6.10611022e-01 -6.62738800e-01 1.21639065e-01
1.44795692e+00 2.43207544e-01 -7.21076906e-01 -2.39844203e-01
-1.23901761e+00 5.32340705e-01 4.71985489e-01 -1.07278474e-01
-3.53804529e-01 2.76632547e-01 3.58433694e-01 2.16804475e-01
6.64187312e-01 1.33455560e-01 -1.26850279e-02 -1.09307952e-01
-9.22984362e-01 3.80694985e-01 4.27139103e-01 9.96655107e-01
6.49770379e-01 4.68594700e-01 -5.16737938e-01 6.30607724e-01
-4.31844629e-02 6.52963936e-01 3.43388349e-01 -1.14583397e+00
5.09346008e-01 3.74778450e-01 2.58562416e-01 -1.36444557e+00
1.14394212e-03 -4.19068396e-01 -1.16095412e+00 4.19701189e-01
2.37255424e-01 -3.53399664e-02 -8.83276641e-01 1.85837305e+00
3.08977902e-01 4.57646489e-01 -1.46306098e-01 1.28029108e+00
6.80317581e-01 8.29770029e-01 -1.51607782e-01 -1.37319714e-01
1.15537643e+00 -1.15257025e+00 -8.42266798e-01 -1.06527604e-01
-8.64555463e-02 -7.09123552e-01 1.14073050e+00 2.55049914e-01
-1.29534197e+00 -1.02935433e+00 -1.07531071e+00 -3.67131889e-01
9.48655084e-02 2.03097910e-02 4.92748678e-01 3.91062021e-01
-1.13710201e+00 5.55864632e-01 -7.09151983e-01 -2.31622178e-02
2.52533972e-01 1.07700534e-01 -5.11932611e-01 -2.68879861e-01
-1.28764641e+00 4.42351699e-01 7.25925788e-02 3.68339747e-01
-1.06696081e+00 -6.68149531e-01 -1.00130987e+00 -6.59300983e-02
3.43907744e-01 -1.07617164e+00 7.79171884e-01 -1.39309549e+00
-2.00432253e+00 6.03119791e-01 8.16567522e-03 -1.76520318e-01
1.05387449e+00 -4.36262667e-01 -3.38946909e-01 5.65410733e-01
-1.23653822e-01 8.22754741e-01 1.34889805e+00 -1.34712875e+00
-3.98477674e-01 -9.49387848e-02 3.75412077e-01 4.46573317e-01
-3.83014083e-01 -8.57941434e-02 -7.74776518e-01 -1.04211950e+00
-2.88116127e-01 -9.65319097e-01 -5.66325001e-02 4.85662401e-01
-2.22946003e-01 3.28336537e-01 8.25025022e-01 -1.04921663e+00
8.80217731e-01 -2.11982012e+00 4.13682789e-01 -1.62152305e-01
3.67816180e-01 3.87922049e-01 -3.00305665e-01 2.19250396e-01
-4.88907397e-02 -2.26234064e-01 4.38567530e-03 -4.88365144e-01
-7.88182393e-02 2.21548676e-02 -4.02346164e-01 3.62466842e-01
3.71881068e-01 1.07510722e+00 -8.54967773e-01 -1.76226169e-01
4.43179548e-01 1.06531596e+00 -7.27300227e-01 6.45458341e-01
-1.02793999e-01 7.86516011e-01 -3.29141438e-01 4.22965169e-01
7.43053734e-01 -2.86805779e-01 -3.17581487e-03 -3.82473826e-01
-5.38796047e-03 -1.88354589e-02 -9.14415181e-01 1.79556537e+00
-4.56797719e-01 6.58712387e-01 1.43899202e-01 -5.77557981e-01
7.12019026e-01 2.47442231e-01 3.30813468e-01 -8.36348891e-01
2.49845311e-02 -9.09798592e-02 -4.09985095e-01 -4.00678545e-01
5.65890133e-01 -2.02318076e-02 1.69970900e-01 8.53479579e-02
-1.98423788e-01 6.13656752e-02 -7.19224364e-02 1.12438425e-01
9.21363890e-01 4.37274039e-01 -2.99436182e-01 7.50991404e-02
5.25875986e-01 -4.74000216e-01 5.15799344e-01 4.43321109e-01
-6.22534640e-02 1.06506884e+00 2.16841102e-01 -7.16507912e-01
-1.34381974e+00 -1.18571770e+00 2.15956762e-01 6.92811966e-01
4.07736123e-01 -4.42561597e-01 -7.27566540e-01 -3.93649697e-01
-2.55666584e-01 2.40074068e-01 -6.23834074e-01 -2.89271384e-01
-8.62816870e-01 -2.98556060e-01 3.05031121e-01 3.51334542e-01
8.95814776e-01 -9.75968957e-01 -6.10554397e-01 1.97367758e-01
-4.41109806e-01 -1.37042391e+00 -8.41163516e-01 -5.69172919e-01
-7.07488060e-01 -8.82027745e-01 -1.18592417e+00 -6.57996058e-01
7.44053304e-01 4.53720778e-01 8.83115351e-01 1.84866175e-01
-4.07718599e-01 3.37838113e-01 -2.10969701e-01 2.77873516e-01
-3.25072438e-01 -4.29421812e-01 1.79255590e-01 3.38258326e-01
-3.62794876e-01 -7.74848521e-01 -1.02329385e+00 3.43085110e-01
-1.01702499e+00 7.52464771e-01 4.79745418e-01 9.68295872e-01
3.58452916e-01 -7.41376076e-03 6.87216148e-02 -4.46140319e-01
2.48351708e-01 1.81750320e-02 -4.07349586e-01 7.01133981e-02
-7.94601440e-02 -2.23787889e-01 8.75966728e-01 -6.35897100e-01
-1.17853439e+00 -2.03965619e-01 -3.92639637e-02 -9.90091681e-01
1.17099453e-02 -9.58898738e-02 -3.30240190e-01 -2.20380560e-01
5.07826686e-01 4.81699675e-01 1.86583489e-01 -3.39396358e-01
3.20804179e-01 2.82493502e-01 7.62032211e-01 -6.84550941e-01
1.20196688e+00 5.02502561e-01 -5.98694347e-02 -6.81399405e-01
-6.78055286e-01 1.08276248e-01 -3.70140225e-01 -5.32820940e-01
7.84396827e-01 -1.31427133e+00 -7.21296728e-01 8.50265086e-01
-1.06666660e+00 -5.89686811e-01 -2.07031652e-01 3.45472723e-01
-7.36933649e-01 6.50939703e-01 -1.01089191e+00 -4.79751885e-01
-4.02644277e-01 -1.21380377e+00 1.32285857e+00 1.21672109e-01
1.87871903e-01 -5.89361608e-01 -1.55127779e-01 3.81990701e-01
5.87646365e-01 6.33990705e-01 6.11725390e-01 3.93562406e-01
-1.04079378e+00 -4.93065380e-02 -4.37208265e-01 5.50862610e-01
2.49199763e-01 -9.30422992e-02 -9.10260916e-01 -6.84757829e-01
3.73913720e-02 -3.74060541e-01 6.95441008e-01 2.08242938e-01
1.18042266e+00 -6.01393700e-01 2.32327580e-02 1.01206934e+00
1.26851296e+00 -9.25255660e-03 9.08201337e-01 1.42721832e-01
1.14841294e+00 4.96361703e-01 4.08640981e-01 2.73286670e-01
1.77899152e-01 1.07563782e+00 2.62085289e-01 -3.69591981e-01
-5.95089197e-01 -4.89642680e-01 6.25598669e-01 7.57141173e-01
-5.37076533e-01 -1.81108326e-01 -3.31277549e-01 5.50431833e-02
-1.73386288e+00 -1.07106113e+00 2.75897652e-01 2.30024433e+00
7.84546435e-01 1.31750301e-01 1.39398202e-01 -4.08482812e-02
7.61755466e-01 3.45832735e-01 -5.33042848e-01 1.35548055e-01
-2.46154055e-01 -5.23215830e-02 8.98390114e-02 4.44918275e-01
-9.19049144e-01 7.87718534e-01 5.73788643e+00 8.24167609e-01
-1.34092176e+00 -5.88964745e-02 8.30706179e-01 -4.30727035e-01
-2.63208419e-01 -3.03355813e-01 -4.52152878e-01 7.21112967e-01
7.25479424e-01 5.76725136e-03 5.48200190e-01 6.08414054e-01
4.38504636e-01 2.87971467e-01 -9.58617032e-01 1.36806309e+00
2.55656123e-01 -1.40268052e+00 5.55171609e-01 1.70599282e-01
9.05924499e-01 -5.38519561e-01 4.03271794e-01 1.43751785e-01
-1.96016267e-01 -1.02009690e+00 1.03047454e+00 6.97751880e-01
1.34587455e+00 -7.40463555e-01 3.52069318e-01 8.29416364e-02
-1.17063200e+00 -8.98855999e-02 -3.87094647e-01 -1.68196693e-01
3.75860542e-01 5.28841853e-01 -1.28692865e-01 7.68245041e-01
6.72521412e-01 8.70460272e-01 -4.56189454e-01 7.95309901e-01
-2.16607273e-01 2.14765951e-01 -9.90616456e-02 4.44058776e-01
2.01005384e-01 -2.18418613e-01 5.93034267e-01 9.47190285e-01
4.42736328e-01 2.21342444e-01 3.49695146e-01 8.70457232e-01
-4.56800982e-02 -5.61229326e-02 -5.14620841e-01 1.53122187e-01
-4.76262830e-02 1.10952806e+00 -3.20609570e-01 -4.84865785e-01
-4.32127148e-01 1.53470159e+00 2.97867268e-01 5.61604798e-01
-1.00742221e+00 -2.71668304e-02 7.51598716e-01 3.22707742e-01
2.06474498e-01 -3.33229154e-01 3.04834485e-01 -1.68452895e+00
3.28145683e-01 -1.08949673e+00 -1.59207642e-01 -9.37234461e-01
-1.21722591e+00 1.00824690e+00 -9.07893404e-02 -1.45188272e+00
-4.03968483e-01 -5.38208723e-01 -4.97057259e-01 8.74505043e-01
-1.56259215e+00 -1.24323142e+00 -7.39074349e-01 8.37530196e-01
6.33385062e-01 -1.59369379e-01 5.20142019e-01 6.01541698e-01
-5.43970704e-01 7.45184243e-01 -9.04979929e-02 7.37183616e-02
8.37653875e-01 -8.92766535e-01 5.90187073e-01 8.99573684e-01
-1.61116928e-01 4.47699606e-01 6.31077528e-01 -6.37869358e-01
-1.38461542e+00 -1.25273585e+00 3.65197152e-01 -3.94452780e-01
8.83956179e-02 -5.04996300e-01 -8.74614775e-01 4.75463569e-01
1.05368450e-01 2.12548628e-01 1.85084552e-01 -4.93663788e-01
-3.87715906e-01 -4.02674265e-02 -8.67792547e-01 8.30797672e-01
1.39338243e+00 -6.40169561e-01 -2.00529680e-01 1.60961524e-01
7.64892757e-01 -6.21803343e-01 -7.00482488e-01 3.77216548e-01
7.29323983e-01 -1.22932315e+00 1.04417837e+00 -2.78842330e-01
7.38905966e-01 -4.57507938e-01 4.75308709e-02 -1.08623135e+00
-4.78284121e-01 -9.92601335e-01 -1.97206482e-01 1.07366765e+00
-3.14415306e-01 -4.10045117e-01 7.48390317e-01 6.37882650e-01
-5.51413521e-02 -5.84361911e-01 -7.96376944e-01 -6.67104483e-01
-2.55932450e-01 -8.97003785e-02 4.90718693e-01 7.07219124e-01
-4.98064905e-01 1.08404674e-01 -1.06851757e+00 6.06387407e-02
7.14612007e-01 1.95508122e-01 1.10242534e+00 -7.18905866e-01
-4.88096565e-01 -5.78537770e-02 -5.07327914e-01 -1.47187531e+00
8.22136998e-02 -4.52947468e-01 -4.88577001e-02 -1.04630053e+00
2.68632233e-01 -1.90395877e-01 8.97188392e-03 1.33925438e-01
-3.35933596e-01 6.39940798e-01 4.94687110e-01 3.00478339e-01
-4.40093458e-01 9.47893083e-01 1.81515765e+00 4.02737148e-02
-6.09285086e-02 -1.69313177e-01 -3.66222888e-01 5.72702944e-01
3.48550051e-01 7.61928931e-02 -5.47402203e-01 -5.26934505e-01
-5.48301488e-02 2.77812570e-01 9.11583900e-01 -1.16698468e+00
-9.62775573e-02 -5.59542961e-02 9.46798205e-01 -3.20745081e-01
7.35440195e-01 -7.22163439e-01 5.60545385e-01 3.61684889e-01
-2.65226752e-01 1.04082257e-01 1.21093228e-01 7.80066967e-01
-1.56930402e-01 6.06238365e-01 7.74736404e-01 -1.01884581e-01
-6.04181767e-01 6.93181515e-01 2.84605809e-02 -1.99871376e-01
8.04696560e-01 -3.02199870e-01 -2.26327121e-01 -7.18197644e-01
-5.37486970e-01 -1.44479170e-01 7.95319974e-01 5.92328072e-01
8.10825944e-01 -1.67090833e+00 -7.69233286e-01 6.35428190e-01
-1.25242621e-01 -6.27432466e-02 7.89873242e-01 4.78461862e-01
-7.87771106e-01 2.32540101e-01 -6.15040720e-01 -5.94534397e-01
-1.01066768e+00 6.87585175e-01 2.62991279e-01 -1.59122020e-01
-1.25958395e+00 5.55479348e-01 8.03049862e-01 2.26151291e-02
2.80342698e-01 -1.56755596e-01 1.17234141e-01 -3.53688568e-01
8.60868573e-01 1.31252959e-01 -3.19991559e-01 -7.98092484e-01
5.72840385e-02 7.16600835e-01 -5.09250984e-02 -2.94557912e-03
1.28599608e+00 -3.02407980e-01 2.61056960e-01 1.00945324e-01
1.25756609e+00 -5.30416407e-02 -2.30415750e+00 -1.79749042e-01
-8.77885580e-01 -1.01654375e+00 -1.92055255e-01 -5.42993844e-01
-1.35794973e+00 7.25232244e-01 6.56259537e-01 -1.86564669e-01
1.28548682e+00 -5.04457712e-01 1.12259793e+00 2.26563178e-02
5.02657413e-01 -7.43118227e-01 5.10499179e-01 1.58694163e-01
1.21100485e+00 -9.88347948e-01 -9.55748260e-02 -5.89346290e-01
-5.22990763e-01 1.13589263e+00 7.43351161e-01 -3.27254444e-01
2.17393637e-01 1.87838711e-02 1.05564162e-01 7.94267356e-02
-6.96488857e-01 1.60369888e-01 4.83568996e-01 6.34999037e-01
2.00592075e-02 -1.68598384e-01 5.76611049e-02 1.98407054e-01
-1.80500031e-01 2.47830041e-02 4.00125265e-01 4.95622039e-01
-1.38728712e-02 -8.43458414e-01 -3.33889931e-01 8.56754035e-02
-2.93228477e-01 -3.62273231e-02 -7.26875570e-03 8.05453598e-01
6.79195672e-02 6.57328188e-01 1.57151911e-02 -3.43700141e-01
3.14027756e-01 -3.21327865e-01 6.94922566e-01 -3.27657968e-01
-2.71784544e-01 2.51177728e-01 -5.54996021e-02 -9.20850873e-01
-4.84589696e-01 -3.39314431e-01 -6.42379165e-01 -6.22131288e-01
1.25203148e-01 -2.10114971e-01 1.93641976e-01 6.17392302e-01
4.15606469e-01 6.80621207e-01 7.06515968e-01 -1.58639741e+00
-2.37865001e-01 -5.74433982e-01 -5.67414284e-01 7.90563524e-01
5.09468973e-01 -7.65989602e-01 -4.26500171e-01 3.05028528e-01]
|
[11.313522338867188, -1.1458417177200317]
|
0b54367b-e0f2-42ac-8cab-4d7708ff33b7
|
spac-net-synthetic-pose-aware-animal
|
2305.17845
| null |
https://arxiv.org/abs/2305.17845v2
|
https://arxiv.org/pdf/2305.17845v2.pdf
|
SPAC-Net: Synthetic Pose-aware Animal ControlNet for Enhanced Pose Estimation
|
Animal pose estimation has become a crucial area of research, but the scarcity of annotated data is a significant challenge in developing accurate models. Synthetic data has emerged as a promising alternative, but it frequently exhibits domain discrepancies with real data. Style transfer algorithms have been proposed to address this issue, but they suffer from insufficient spatial correspondence, leading to the loss of label information. In this work, we present a new approach called Synthetic Pose-aware Animal ControlNet (SPAC-Net), which incorporates ControlNet into the previously proposed Prior-Aware Synthetic animal data generation (PASyn) pipeline. We leverage the plausible pose data generated by the Variational Auto-Encoder (VAE)-based data generation pipeline as input for the ControlNet Holistically-nested Edge Detection (HED) boundary task model to generate synthetic data with pose labels that are closer to real data, making it possible to train a high-precision pose estimation network without the need for real data. In addition, we propose the Bi-ControlNet structure to separately detect the HED boundary of animals and backgrounds, improving the precision and stability of the generated data. Using the SPAC-Net pipeline, we generate synthetic zebra and rhino images and test them on the AP10K real dataset, demonstrating superior performance compared to using only real images or synthetic data generated by other methods. Our work demonstrates the potential for synthetic data to overcome the challenge of limited annotated data in animal pose estimation.
|
['Sarah Ostadabbas', 'Le Jiang']
|
2023-05-29
| null | null | null | null |
['style-transfer', 'pose-estimation', 'edge-detection', 'animal-pose-estimation']
|
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
|
[ 1.93200961e-01 1.24702707e-01 3.93379778e-01 -3.32930684e-01
-7.45962679e-01 -5.61089575e-01 5.68225622e-01 -1.62294880e-02
-6.08393610e-01 7.52368569e-01 -2.22180095e-02 3.62063348e-01
2.16184556e-01 -7.07736552e-01 -1.23475087e+00 -3.90858322e-01
3.89281064e-02 5.97139418e-01 5.39027631e-01 -3.60648572e-01
-7.97923133e-02 4.78014112e-01 -1.69527626e+00 1.09923460e-01
7.37746656e-01 7.93825507e-01 3.66053641e-01 5.30318260e-01
3.71410221e-01 4.17058349e-01 -6.04460895e-01 -4.97821599e-01
4.95162338e-01 -4.72999036e-01 -3.62212002e-01 -1.01637624e-01
7.80778885e-01 -5.44636488e-01 -1.07383929e-01 7.23725855e-01
5.51679373e-01 -5.38639575e-02 6.15023375e-01 -1.40329766e+00
-2.32906953e-01 2.81208724e-01 -7.38569319e-01 -1.97534025e-01
1.16567880e-01 4.44954753e-01 7.71956801e-01 -7.46541977e-01
1.01899099e+00 1.24581110e+00 1.02981424e+00 7.21808493e-01
-1.86432314e+00 -7.30763555e-01 -1.31098554e-01 1.00927547e-01
-1.35927606e+00 -2.54443556e-01 7.59001076e-01 -7.16755331e-01
5.51571488e-01 -3.81028056e-02 9.36614752e-01 1.56485343e+00
2.02859119e-02 6.97345257e-01 1.02932096e+00 -9.40643847e-02
1.97430670e-01 -1.82619989e-01 -5.42495489e-01 5.96708536e-01
2.10622713e-01 5.03994644e-01 -6.14847422e-01 1.40024461e-02
9.37856257e-01 -3.27576995e-01 -3.36521834e-01 -9.96115148e-01
-1.53391981e+00 6.81012869e-01 6.81864262e-01 -3.59778166e-01
-4.41508710e-01 3.39803070e-01 3.08973908e-01 -1.05119392e-01
5.02805650e-01 7.79115140e-01 -5.50134122e-01 2.39302367e-01
-1.06358850e+00 8.12752128e-01 5.01276553e-01 1.05334520e+00
6.04208291e-01 1.25893414e-01 -3.14489156e-01 8.08736682e-01
3.70965630e-01 5.87590516e-01 9.76464972e-02 -1.01462936e+00
4.05528247e-01 4.94676381e-01 3.06396365e-01 -1.05336630e+00
-5.53365052e-01 -5.84928691e-01 -5.59706569e-01 4.66725230e-01
6.57898784e-01 -1.61504999e-01 -1.07856512e+00 1.98830497e+00
6.89676285e-01 2.54714340e-01 -7.39012659e-02 1.08942938e+00
8.06803465e-01 5.78607321e-01 2.54080981e-01 4.35908198e-01
1.20906281e+00 -9.74891245e-01 -2.02308401e-01 -4.49475765e-01
4.09146935e-01 -5.87840855e-01 8.47450495e-01 1.65581182e-01
-7.44015574e-01 -5.37091970e-01 -1.16763127e+00 -7.68102929e-02
-2.60047525e-01 2.55534440e-01 4.10396934e-01 1.86534792e-01
-8.74096274e-01 6.45809889e-01 -9.93170023e-01 -3.46339345e-01
4.94590133e-01 2.12070674e-01 -5.83490729e-01 1.29926622e-01
-1.00355601e+00 8.78384352e-01 4.88953173e-01 4.38943326e-01
-1.02845979e+00 -1.14102125e+00 -1.28832436e+00 -2.91469902e-01
3.90615761e-01 -7.02618122e-01 1.14939630e+00 -6.76851451e-01
-1.29335093e+00 8.29235017e-01 4.45559829e-01 -6.47128880e-01
1.06658959e+00 -3.35656047e-01 2.12626648e-03 9.69747305e-02
2.42255211e-01 1.54262495e+00 8.55233431e-01 -1.41798162e+00
-5.34303129e-01 -3.30206603e-01 -1.55353740e-01 -1.29571389e-02
3.79019618e-01 -2.85494000e-01 -4.66735989e-01 -9.36377645e-01
-1.51369229e-01 -1.16376030e+00 -3.38607073e-01 5.22975802e-01
-1.45696908e-01 2.74574578e-01 7.62168169e-01 -9.46691930e-01
7.05263674e-01 -2.02371740e+00 3.27684462e-01 -1.06072821e-01
4.97995503e-02 4.21582490e-01 -4.45924133e-01 2.60366321e-01
-2.07520928e-02 -2.05839097e-01 -6.25650704e-01 -3.59694362e-01
-8.21168572e-02 2.72482097e-01 -5.91741465e-02 4.42017496e-01
7.08558798e-01 9.74204481e-01 -9.15234447e-01 -5.77967167e-01
3.31492364e-01 7.23599195e-01 -1.11989152e+00 3.56567234e-01
-6.41842604e-01 9.19435203e-01 -6.61354735e-02 4.28311199e-01
7.95731843e-01 1.70272991e-01 -4.33257967e-02 -4.49227452e-01
-7.59263337e-02 -1.23927392e-01 -1.10750175e+00 1.74118018e+00
-3.84821534e-01 4.96442795e-01 1.51598811e-01 -4.93538857e-01
6.77519917e-01 5.95168918e-02 4.38804507e-01 -6.04902446e-01
2.69993814e-03 2.05907285e-01 5.28053194e-02 -3.71597171e-01
4.76582319e-01 -1.43934548e-01 -2.07505122e-01 -2.43688181e-01
2.28804663e-01 -5.20909071e-01 3.71397555e-01 -3.96941230e-02
8.38387072e-01 1.15774703e+00 5.59154339e-03 -1.42410338e-01
1.58122942e-01 4.57633227e-01 8.00928891e-01 2.05261737e-01
-1.38286769e-01 1.08184624e+00 4.70095009e-01 -3.25123489e-01
-1.32365716e+00 -1.11180842e+00 -2.20154822e-01 7.05837965e-01
2.10204348e-01 -3.05927217e-01 -9.65847254e-01 -6.90905213e-01
2.30216399e-01 6.42098665e-01 -7.76753008e-01 -1.18269950e-01
-8.12985301e-01 -6.48211062e-01 6.52454615e-01 5.52466571e-01
6.63275301e-01 -1.13435030e+00 -1.04147422e+00 3.17073941e-01
-4.15494382e-01 -1.33491564e+00 -3.15017641e-01 4.36689556e-02
-5.00227809e-01 -9.88147497e-01 -8.73763740e-01 -4.91090983e-01
6.62019968e-01 -2.21373990e-01 1.03247619e+00 -3.18712294e-01
-5.77541471e-01 -1.01924419e-01 -2.30654642e-01 -3.20378542e-01
-5.36780775e-01 -1.20123485e-02 8.48250557e-03 -1.69216126e-01
-1.31700814e-01 -4.08577234e-01 -9.04070735e-01 6.06646955e-01
-1.02551270e+00 4.02995050e-01 5.42045236e-01 9.95929003e-01
8.14025939e-01 -6.13567591e-01 5.51627278e-01 -5.96976697e-01
2.24639382e-02 -4.01793689e-01 -1.01584852e+00 -8.52690935e-02
-1.28374994e-01 2.66809911e-01 6.45565689e-01 -4.00152743e-01
-1.01040900e+00 4.65355754e-01 -3.66480231e-01 -3.84881526e-01
-3.01117420e-01 3.17267090e-01 -1.63099766e-01 -4.03454378e-02
8.75983536e-01 -1.84875086e-01 2.45806113e-01 -5.59668541e-01
2.37807646e-01 2.73771644e-01 8.45207930e-01 -4.41672206e-01
8.43681693e-01 3.99679244e-01 1.49796903e-01 -8.04734051e-01
-7.31510401e-01 -1.22778110e-01 -7.35831141e-01 -3.12907904e-01
1.10978544e+00 -1.04062569e+00 -3.87252688e-01 6.21987581e-01
-1.13539124e+00 -7.26261854e-01 -3.31345022e-01 5.81019521e-01
-7.09825695e-01 2.49239072e-01 -4.62750494e-01 -3.92972112e-01
-3.54721472e-02 -1.21167135e+00 1.55615342e+00 -3.88207287e-02
-3.65254372e-01 -4.66808647e-01 7.37346709e-02 4.86887127e-01
2.30129883e-01 9.49064910e-01 4.00288045e-01 -3.33736211e-01
-6.73861623e-01 -2.31008902e-01 -2.08643094e-01 2.76296824e-01
-2.12628111e-01 1.09919041e-01 -7.61098206e-01 -3.12267095e-01
-4.07363266e-01 -5.53567231e-01 6.64492369e-01 2.63673753e-01
9.02078629e-01 -1.57361496e-02 -1.89147070e-01 8.17391813e-01
1.29079676e+00 -3.13495964e-01 5.24816513e-01 2.00963050e-01
7.55102575e-01 8.74169409e-01 8.59059155e-01 2.02979594e-01
4.27311778e-01 1.11451185e+00 6.68480098e-01 -2.90671468e-01
-5.44303358e-01 -7.70756960e-01 2.16972858e-01 2.50093728e-01
3.86801623e-02 -1.90151840e-01 -8.99115145e-01 6.64386451e-01
-1.72427535e+00 -8.60538363e-01 -3.37858528e-01 2.17438412e+00
7.47561097e-01 -2.89199818e-02 2.86359727e-01 -2.04781026e-01
5.41054308e-01 -1.73470471e-02 -5.32271087e-01 1.58459723e-01
-7.39601254e-03 9.84282643e-02 8.32586527e-01 2.26868600e-01
-1.17147827e+00 1.07356465e+00 5.62804842e+00 7.27394402e-01
-1.21541119e+00 -1.27722636e-01 2.41758168e-01 -1.76555216e-02
-1.26183638e-02 -1.75988123e-01 -7.76943445e-01 4.64344919e-01
6.81478262e-01 3.79800171e-01 9.77170765e-02 9.21064258e-01
2.18403786e-01 -3.32789540e-01 -1.23191071e+00 8.47796738e-01
6.04318827e-02 -1.08146203e+00 -3.96600813e-02 7.99210966e-02
6.19043410e-01 1.64839208e-01 -1.51807964e-01 2.05740511e-01
2.86197096e-01 -8.37661326e-01 1.10540581e+00 3.29942733e-01
8.55652392e-01 -6.07967138e-01 7.54447222e-01 4.53344673e-01
-1.11904907e+00 3.04029137e-01 -1.93329468e-01 2.96687394e-01
3.43635023e-01 1.54492885e-01 -1.02835262e+00 5.54862201e-01
7.64860392e-01 4.77691770e-01 -9.00870323e-01 1.31453633e+00
-1.88711792e-01 4.27073956e-01 -6.56759262e-01 2.26612031e-01
3.31359804e-02 -1.96955174e-01 6.20755434e-01 9.80093718e-01
3.06413352e-01 -2.93234289e-01 -3.26568298e-02 1.16796649e+00
-9.23625901e-02 -4.67880964e-02 -6.26837909e-01 1.40532762e-01
1.57595307e-01 1.17180431e+00 -8.28545749e-01 3.70551795e-02
-1.55042544e-01 1.10261464e+00 3.02302599e-01 5.80431409e-02
-1.12742484e+00 -1.24730878e-01 7.25486994e-01 3.50597233e-01
4.24104840e-01 -2.84472138e-01 -2.26655174e-02 -1.03679645e+00
8.41022879e-02 -8.87408495e-01 2.47017458e-01 -8.05837691e-01
-1.11022997e+00 6.35123432e-01 5.30368447e-01 -1.60286212e+00
-3.52644801e-01 -5.43435872e-01 -6.71681687e-02 6.99457586e-01
-1.28039348e+00 -1.58808088e+00 -4.44777638e-01 3.19808512e-03
5.97997427e-01 2.26374090e-01 5.70079625e-01 3.69168222e-01
-3.19040328e-01 5.00478208e-01 -2.31704175e-01 1.41202174e-02
8.29247177e-01 -1.07964957e+00 7.21773684e-01 8.48912477e-01
4.89726178e-02 1.28401577e-01 9.53824461e-01 -8.23974371e-01
-9.55444038e-01 -1.44610822e+00 3.90060961e-01 -6.07399583e-01
3.28063220e-01 -5.75247467e-01 -8.39220703e-01 5.65155745e-01
-1.89594328e-01 1.78547829e-01 2.88165629e-01 -6.45974398e-01
-1.72322482e-01 3.66099253e-02 -1.25516737e+00 8.82326841e-01
1.21361411e+00 -1.07597120e-01 -4.83726084e-01 -2.48388648e-02
7.80920744e-01 -6.77747190e-01 -6.60303652e-01 8.80510569e-01
5.84634960e-01 -9.02058482e-01 1.13089263e+00 -1.63102940e-01
7.76583672e-01 -8.28604579e-01 7.71655664e-02 -1.63836181e+00
-1.04812175e-01 -1.96503237e-01 2.40060642e-01 1.18995535e+00
5.36732435e-01 -3.51564825e-01 8.38593483e-01 3.53814542e-01
-1.48709476e-01 -5.17477155e-01 -8.25146914e-01 -7.71583438e-01
2.14475617e-02 -4.16878641e-01 5.83351195e-01 7.10578084e-01
-5.93345404e-01 2.18091145e-01 -5.52668273e-01 2.28128374e-01
7.97183573e-01 -6.52386695e-02 1.21895754e+00 -1.20294654e+00
-3.35294724e-01 -8.98351744e-02 -7.64616311e-01 -1.12166023e+00
8.26352537e-02 -7.71588385e-01 4.95625496e-01 -1.59764159e+00
-2.90360570e-01 -2.70245910e-01 4.10220802e-01 4.17591631e-01
-2.60602366e-02 8.04622352e-01 6.30938187e-02 -1.25679791e-01
-2.16123134e-01 8.72394562e-01 1.43288410e+00 1.39203206e-01
-7.15846494e-02 -2.44887561e-01 -1.58128440e-01 8.92538488e-01
4.78887022e-01 -5.24781108e-01 -3.16192687e-01 -4.42205787e-01
1.76897228e-01 4.12519909e-02 8.82872999e-01 -1.33353746e+00
-8.33105147e-02 5.82830496e-02 5.16579211e-01 -7.18825102e-01
4.12027836e-01 -9.80036497e-01 5.82866430e-01 3.78261894e-01
-1.97987333e-01 -9.79646072e-02 4.02421087e-01 7.03097820e-01
-5.14154844e-02 1.01353236e-01 9.80877995e-01 1.24150738e-01
-7.12971270e-01 2.64052600e-01 -1.56431627e-02 2.45200798e-01
1.21260941e+00 -2.08827972e-01 2.60693915e-02 -7.54032433e-02
-5.10381281e-01 5.38246989e-01 7.85765231e-01 6.27660990e-01
4.51984406e-01 -1.21750844e+00 -9.11121488e-01 4.20886785e-01
5.59788704e-01 6.19355321e-01 3.28626692e-01 7.68200755e-01
-1.31552160e+00 -9.22057107e-02 -4.54190880e-01 -8.92706752e-01
-1.16274488e+00 3.65983218e-01 4.25208688e-01 -6.49242029e-02
-8.18888783e-01 7.59742022e-01 3.91918451e-01 -8.18819523e-01
2.66933609e-02 -4.28381383e-01 -7.05852211e-02 4.44566943e-02
1.74116433e-01 2.25395381e-01 3.19648162e-02 -9.41040456e-01
-2.15379924e-01 6.03334963e-01 2.41848096e-01 -2.54180789e-01
1.52445686e+00 1.20145358e-01 3.17902446e-01 7.42034540e-02
8.76604021e-01 -1.94645058e-02 -1.90292227e+00 5.37324734e-02
1.31628104e-02 -4.85527039e-01 -2.70614356e-01 -6.48444295e-01
-9.34129059e-01 8.09821844e-01 6.50722086e-01 -3.90681386e-01
6.24777198e-01 -6.39924705e-02 8.50427151e-01 -9.95398089e-02
6.70402586e-01 -1.21878648e+00 1.02499135e-01 2.17787161e-01
1.27901232e+00 -1.32983637e+00 -1.95778236e-01 -6.26899481e-01
-7.51424968e-01 6.81917071e-01 1.06441009e+00 -1.27588272e-01
3.07804912e-01 2.76941568e-01 1.16501011e-01 -1.94031879e-01
-4.13276345e-01 -3.11377198e-01 3.23972553e-01 8.26028943e-01
2.67086387e-01 -1.17918439e-01 -1.50265038e-01 3.39575827e-01
-4.39928383e-01 1.19952679e-01 2.79524922e-01 8.75458241e-01
1.28675133e-01 -1.06511736e+00 -4.20051515e-01 8.37668851e-02
-2.72951484e-01 4.56619672e-02 -3.27828050e-01 8.87814999e-01
4.43932265e-01 5.15493810e-01 1.16621861e-02 -2.89183855e-01
6.83626890e-01 -4.29798104e-02 4.98659551e-01 -6.96238399e-01
-4.42681193e-01 4.48736213e-02 1.51112020e-01 -5.95276594e-01
-3.51551801e-01 -4.20809835e-01 -1.02617276e+00 3.11713349e-02
-1.86260194e-01 -5.63276745e-02 6.29811525e-01 6.18166983e-01
4.75033134e-01 6.98612213e-01 7.80132115e-02 -1.44728827e+00
-4.10055578e-01 -9.57008660e-01 -2.04672977e-01 5.88223994e-01
3.01878452e-01 -9.04720068e-01 -2.31688559e-01 3.09760034e-01]
|
[7.592082977294922, -1.0702061653137207]
|
5f5e8fd4-df13-4606-b04f-616c951f2a1e
|
multivariate-time-series-anomaly-detection
|
2009.02040
| null |
https://arxiv.org/abs/2009.02040v1
|
https://arxiv.org/pdf/2009.02040v1.pdf
|
Multivariate Time-series Anomaly Detection via Graph Attention Network
|
Anomaly detection on multivariate time-series is of great importance in both data mining research and industrial applications. Recent approaches have achieved significant progress in this topic, but there is remaining limitations. One major limitation is that they do not capture the relationships between different time-series explicitly, resulting in inevitable false alarms. In this paper, we propose a novel self-supervised framework for multivariate time-series anomaly detection to address this issue. Our framework considers each univariate time-series as an individual feature and includes two graph attention layers in parallel to learn the complex dependencies of multivariate time-series in both temporal and feature dimensions. In addition, our approach jointly optimizes a forecasting-based model and are construction-based model, obtaining better time-series representations through a combination of single-timestamp prediction and reconstruction of the entire time-series. We demonstrate the efficacy of our model through extensive experiments. The proposed method outperforms other state-of-the-art models on three real-world datasets. Further analysis shows that our method has good interpretability and is useful for anomaly diagnosis.
|
['Yujing Wang', 'Yunhai Tong', 'Congrui Huang', 'Hang Zhao', 'Jing Bai', 'Defu Cao', 'Bixiong Xu', 'Qi Zhang', 'Juanyong Duan', 'Jie Tong']
|
2020-09-04
| null | null | null | null |
['spatio-temporal-forecasting']
|
['time-series']
|
[ 6.94195703e-02 -2.43158624e-01 -8.12604465e-03 -3.86518806e-01
-4.63834871e-03 -1.42767340e-01 3.54503393e-01 7.66976237e-01
1.39552634e-02 1.28139034e-01 -1.02710359e-01 -4.47762221e-01
-3.49335670e-01 -7.08608329e-01 -4.37685490e-01 -4.89733726e-01
-6.71626747e-01 2.89869934e-01 2.79972553e-01 -2.02192321e-01
1.77645639e-01 6.01201653e-01 -1.33448434e+00 -1.24115616e-01
1.14364243e+00 1.29189670e+00 -3.38185817e-01 3.89655411e-01
-5.58259748e-02 9.00414288e-01 -6.37151539e-01 -2.84919254e-02
9.10237730e-02 -3.91694456e-01 -3.59586835e-01 3.26257259e-01
-5.80528043e-02 -1.22682400e-01 -5.47907650e-01 9.78922427e-01
6.57133684e-02 2.97541261e-01 3.88147503e-01 -1.74915946e+00
-6.24750853e-01 4.09722298e-01 -8.10701191e-01 6.73958480e-01
-4.28455919e-02 1.00917540e-01 1.08409750e+00 -4.37135398e-01
-1.13836057e-01 1.05343974e+00 5.43213964e-01 6.59756809e-02
-1.12393284e+00 -6.34965122e-01 7.44816840e-01 7.53142715e-01
-1.03179336e+00 -3.98166589e-02 1.42730117e+00 -3.12873125e-01
1.35756791e+00 1.87535942e-01 7.86595583e-01 8.83319616e-01
5.20547450e-01 7.67044723e-01 5.52534163e-01 -2.77142495e-01
3.44573557e-01 -5.45305490e-01 4.29501623e-01 5.17725825e-01
2.55268127e-01 -3.48335877e-02 -2.09633484e-01 -3.02893519e-01
5.99721372e-01 7.15545654e-01 6.41780943e-02 -2.37265646e-01
-1.16740584e+00 6.65325284e-01 3.34848642e-01 5.06913424e-01
-6.71276748e-01 6.54735118e-02 6.85204148e-01 6.73956752e-01
9.18952167e-01 3.57149065e-01 -4.22475219e-01 -9.71151665e-02
-5.30984879e-01 -9.38769355e-02 7.22114205e-01 6.40247107e-01
3.27693194e-01 5.89686871e-01 1.45171002e-01 5.45487046e-01
2.74200767e-01 1.74835533e-01 6.75912142e-01 -3.14443648e-01
4.60716844e-01 8.68134141e-01 -2.54065603e-01 -1.51708937e+00
-6.73460186e-01 -6.66996598e-01 -1.15271890e+00 2.18335334e-02
2.61351764e-01 -1.11572899e-01 -7.61897743e-01 1.62468255e+00
3.32490057e-01 9.15175796e-01 -2.09777921e-01 6.88603222e-01
1.48614913e-01 6.43881917e-01 8.57690200e-02 -4.65075791e-01
9.30930316e-01 -1.00484014e+00 -1.12259567e+00 -3.12754422e-01
6.38563693e-01 -3.66324246e-01 8.49753141e-01 3.79225910e-01
-6.08528376e-01 -4.39923733e-01 -1.06916916e+00 3.33264381e-01
-4.47418153e-01 2.37204880e-02 6.66258752e-01 1.40612781e-01
-5.50134301e-01 8.12393844e-01 -1.36328387e+00 -3.94913375e-01
2.43796572e-01 3.04296702e-01 -1.31045058e-01 2.23193303e-01
-1.14426637e+00 5.67850769e-01 3.65257204e-01 3.46711904e-01
-5.29657722e-01 -4.07434672e-01 -9.41319644e-01 1.19181730e-01
6.39124215e-01 -3.56818020e-01 9.94593322e-01 -8.68275583e-01
-1.16660500e+00 2.46122628e-01 -2.46401727e-01 -6.82806194e-01
1.73011810e-01 -4.63998675e-01 -1.26961541e+00 -1.46407902e-01
1.70092117e-02 -5.25280297e-01 1.05261099e+00 -6.96268380e-01
-5.87942839e-01 -6.96023405e-01 -2.12226480e-01 -2.55014360e-01
-6.93082154e-01 -1.13017142e-01 -2.07950786e-01 -1.07587755e+00
3.60395491e-01 -6.90190315e-01 -5.38207650e-01 -2.42216200e-01
-5.49647868e-01 -4.52492476e-01 1.21189129e+00 -5.67166507e-01
1.79891682e+00 -2.28058052e+00 7.00525120e-02 4.18814301e-01
3.65695447e-01 1.81106672e-01 -4.31742184e-02 7.58105040e-01
-5.22513151e-01 -5.31916916e-02 -4.27316397e-01 -4.31737244e-01
-1.56445399e-01 4.40744340e-01 -6.39356256e-01 6.73676848e-01
5.17234921e-01 7.10286081e-01 -1.06291807e+00 -1.58335835e-01
3.49783868e-01 2.22802237e-01 -2.70220369e-01 3.57660532e-01
-1.96474805e-01 6.81178987e-01 -7.39118814e-01 7.71120369e-01
2.56633997e-01 -4.68358546e-01 1.24108698e-02 -9.57895257e-03
3.36810611e-02 6.32256940e-02 -1.00824654e+00 1.54494345e+00
-1.85647517e-01 4.39212203e-01 -4.64562446e-01 -1.47875774e+00
9.23595309e-01 2.70479321e-01 9.57098544e-01 -6.98720515e-01
6.65969253e-02 1.31418765e-01 2.32613549e-01 -7.32983410e-01
2.77339548e-01 2.18813777e-01 -4.40297201e-02 6.38370812e-01
-8.59276950e-02 2.58988529e-01 2.42958948e-01 1.52841443e-02
1.33971310e+00 -9.82524827e-02 4.44559962e-01 6.01486042e-02
7.98198581e-01 -2.33853713e-01 7.81287014e-01 4.44831848e-01
-2.49376640e-01 2.61585981e-01 7.33812571e-01 -7.86091030e-01
-8.57696116e-01 -7.91463315e-01 2.00930089e-01 8.54133129e-01
2.96074748e-02 -6.26623869e-01 -2.37121731e-01 -1.08972156e+00
1.54139057e-01 1.02402651e+00 -5.40385067e-01 -4.46388751e-01
-6.91890955e-01 -8.54698658e-01 3.40824157e-01 6.81173444e-01
4.57815193e-02 -1.18140864e+00 -3.70412499e-01 4.66670752e-01
-1.43127322e-01 -1.07456172e+00 -4.07660872e-01 1.50306508e-01
-1.37715292e+00 -1.14989960e+00 -3.43661495e-02 -4.27162379e-01
6.73084557e-01 1.97559163e-01 9.78285730e-01 9.95091796e-02
-3.27714086e-01 4.53365028e-01 -5.07835507e-01 -7.18042731e-01
-1.90015242e-01 -6.68001920e-02 1.61634505e-01 6.16924524e-01
6.86469078e-01 -1.08088386e+00 -4.83068973e-01 2.11071625e-01
-1.13290942e+00 -5.89733720e-01 3.63407969e-01 7.61931658e-01
7.52748132e-01 4.43980932e-01 9.32475150e-01 -7.67262399e-01
7.62000561e-01 -1.05422127e+00 -5.93936503e-01 1.14612430e-02
-9.21368122e-01 -1.34782687e-01 1.11003101e+00 -4.33965296e-01
-6.08015239e-01 -9.64895487e-02 5.04207648e-02 -9.07548845e-01
-3.16725761e-01 7.59318352e-01 1.21773534e-01 2.88609892e-01
3.73109519e-01 3.32047254e-01 1.27767414e-01 -7.21182704e-01
-6.47578090e-02 3.19013029e-01 4.60107774e-01 -3.87117326e-01
9.21688199e-01 5.30587614e-01 2.17141166e-01 -9.08250809e-01
-6.87256992e-01 -6.13684535e-01 -6.99133992e-01 -1.38988167e-01
4.51291621e-01 -7.07574129e-01 -5.56821585e-01 6.44761860e-01
-9.79444683e-01 1.09907955e-01 -3.73023927e-01 5.48046708e-01
-2.33624488e-01 7.38312185e-01 -5.32739997e-01 -1.00064278e+00
-2.36208588e-01 -7.33636856e-01 9.19253886e-01 -8.65079239e-02
-2.72458017e-01 -1.36723709e+00 4.63238135e-02 -2.34422922e-01
3.76374424e-01 6.37224019e-01 1.08474755e+00 -1.25766122e+00
-3.35795641e-01 -6.82906091e-01 -3.32979523e-02 2.51318306e-01
5.71309388e-01 -9.03639719e-02 -6.99154556e-01 -4.18825746e-01
3.22801620e-01 1.03772886e-01 7.21943438e-01 1.43871471e-01
1.79021680e+00 -2.91986853e-01 -2.96421766e-01 4.85544235e-01
1.12544000e+00 3.95598203e-01 3.88685435e-01 3.60221297e-01
1.04907620e+00 5.38655519e-01 7.01403260e-01 7.26760268e-01
4.47442383e-01 3.62672865e-01 8.85936856e-01 1.28053762e-02
4.95987505e-01 -7.59379640e-02 4.21836883e-01 1.13135386e+00
-2.51698941e-01 -3.93522292e-01 -9.64829326e-01 7.35631049e-01
-2.27450752e+00 -1.05119598e+00 -3.92029494e-01 2.18382788e+00
1.28697678e-01 2.17939049e-01 2.70422459e-01 6.72794402e-01
4.95774895e-01 4.28018600e-01 -9.73848403e-01 -2.18797073e-01
2.84811039e-03 3.05807553e-02 6.66995645e-02 5.74029498e-02
-1.42011809e+00 4.79200989e-01 6.03333473e+00 2.98665613e-01
-1.22246540e+00 -1.98056743e-01 4.40013140e-01 -3.95142622e-02
-6.56305999e-02 -1.86648995e-01 -8.81367102e-02 5.96186519e-01
1.04715037e+00 -4.10005599e-01 3.19277614e-01 8.28460932e-01
2.48634264e-01 5.53651869e-01 -1.34257174e+00 8.68160427e-01
8.07869360e-02 -8.11509252e-01 6.84346333e-02 7.18766302e-02
3.72753859e-01 3.96434851e-02 -6.21826574e-02 1.91970184e-01
-1.03674382e-01 -7.40231097e-01 3.83978873e-01 5.37870169e-01
1.33455455e-01 -7.40417242e-01 8.02929044e-01 2.68127531e-01
-1.36911356e+00 -3.98870438e-01 -2.64834743e-02 -3.41335416e-01
1.82622537e-01 9.70698655e-01 -5.55726409e-01 9.68793631e-01
7.07660854e-01 1.42233217e+00 -5.90172231e-01 1.00457931e+00
-1.35204215e-02 8.83682251e-01 -4.04466510e-01 3.00685287e-01
1.64030418e-01 -4.50737894e-01 9.76763844e-01 8.69641900e-01
4.02152061e-01 -1.06787547e-01 5.05146027e-01 6.92557871e-01
4.24550503e-01 1.11945644e-01 -7.42741883e-01 -4.77666348e-01
2.94603318e-01 9.94136691e-01 -6.24151349e-01 -1.45665303e-01
-7.63169169e-01 8.60267043e-01 2.74321824e-01 4.81547385e-01
-8.55620742e-01 -5.14771283e-01 7.23138809e-01 4.66767661e-02
1.97582081e-01 -3.88348728e-01 -1.49989083e-01 -1.50464547e+00
4.10935521e-01 -8.97731483e-01 1.01398897e+00 -2.18038455e-01
-1.86508036e+00 6.87425017e-01 -1.43989787e-01 -1.79025126e+00
-6.14728034e-01 -4.05577540e-01 -9.73061979e-01 5.24616003e-01
-1.50891316e+00 -1.07918048e+00 -2.97068894e-01 8.75431240e-01
6.08871162e-01 -2.87255287e-01 7.99338639e-01 4.45709646e-01
-1.09124804e+00 4.63480443e-01 5.75803779e-02 1.89936176e-01
4.22209412e-01 -1.34152997e+00 8.41190755e-01 1.24646425e+00
1.99771315e-01 4.72197235e-01 6.60858154e-01 -7.34894097e-01
-1.48482251e+00 -1.38853288e+00 7.94941366e-01 -1.94930643e-01
1.16727543e+00 -2.04715878e-03 -1.41306400e+00 9.75329697e-01
-5.28680496e-02 2.92608947e-01 7.51624584e-01 2.79274583e-01
-2.26579934e-01 -3.73452723e-01 -7.81272411e-01 5.21531522e-01
1.06048191e+00 -4.58090454e-01 -6.69890761e-01 3.32770556e-01
7.92569041e-01 -2.33923718e-01 -8.34096849e-01 4.87059474e-01
1.48663044e-01 -9.30539668e-01 7.89654016e-01 -8.29901993e-01
2.50741631e-01 -4.10144597e-01 9.85613167e-02 -1.54549086e+00
-4.15260732e-01 -7.19446182e-01 -9.95078325e-01 9.81807411e-01
8.53138342e-02 -1.21194065e+00 4.40678537e-01 3.15073967e-01
-4.20542419e-01 -8.09556961e-01 -8.51155221e-01 -9.86831963e-01
-5.28950572e-01 -7.78369248e-01 8.38114381e-01 1.21586692e+00
1.84062600e-01 3.37959647e-01 -4.85039443e-01 5.36111772e-01
7.12689281e-01 3.26734930e-01 6.05095506e-01 -1.54121208e+00
-2.33676374e-01 -4.33191359e-01 -7.72520840e-01 -7.57910013e-01
2.55095273e-01 -6.44842863e-01 -3.32122862e-01 -1.12449801e+00
-4.93016571e-01 -1.87019289e-01 -1.07399344e+00 5.55532217e-01
-2.65644789e-01 -2.99881041e-01 -2.59779841e-01 2.84144402e-01
-6.04619563e-01 7.88353026e-01 9.04385746e-01 -1.08405478e-01
-3.82666886e-01 2.05691576e-01 -3.08394790e-01 8.47295403e-01
9.33928251e-01 -4.63943273e-01 -6.83471799e-01 -3.85255903e-01
-1.67128161e-01 1.10479541e-01 4.29861099e-01 -1.13519561e+00
2.81939477e-01 -2.09698841e-01 2.86037296e-01 -7.73163795e-01
1.55367047e-01 -1.29989970e+00 -9.70173776e-02 4.38169390e-01
-8.10736790e-02 7.71554351e-01 2.78963745e-01 1.28995466e+00
-5.26971459e-01 3.85768563e-01 1.89036399e-01 2.63063401e-01
-9.25426900e-01 6.59161031e-01 -1.56646281e-01 -1.00189045e-01
1.14174747e+00 9.69746057e-03 -2.70625595e-02 -5.24720371e-01
-6.67887509e-01 4.84447181e-01 5.76912351e-02 8.73727679e-01
6.31898344e-01 -1.50462484e+00 -5.38114846e-01 6.62698328e-01
4.18239713e-01 2.09122691e-02 2.80343026e-01 9.94117677e-01
-2.26912200e-01 2.65476823e-01 -1.46438256e-01 -7.60378063e-01
-8.74128819e-01 9.75024521e-01 1.35805845e-01 -4.45486456e-01
-9.51959372e-01 1.33807868e-01 -7.35032782e-02 -1.05475023e-01
3.64104360e-01 -4.97443438e-01 -3.84376705e-01 -1.72905460e-01
4.83575404e-01 6.03154182e-01 9.48155820e-02 -5.27940154e-01
-3.69158924e-01 5.02174139e-01 -1.06135897e-01 4.26378906e-01
1.44387865e+00 -1.47398368e-01 -3.28172773e-01 1.05360043e+00
9.01420534e-01 -2.11547941e-01 -9.10924017e-01 -4.45054948e-01
4.04647082e-01 -5.21072507e-01 -6.43957555e-02 -3.65171701e-01
-1.31923997e+00 7.73632705e-01 3.57894838e-01 8.49863946e-01
1.55159807e+00 -3.42681944e-01 9.51385200e-01 4.81694102e-01
1.65001914e-01 -9.34357464e-01 1.88127786e-01 5.25522232e-01
8.45547676e-01 -1.35153830e+00 -4.42943163e-02 -3.86045307e-01
-4.87324148e-01 1.17121065e+00 8.48674476e-01 -3.73465419e-01
9.01822269e-01 -1.63172528e-01 -5.55426516e-02 -3.34201962e-01
-9.04991746e-01 -9.93028060e-02 6.25772476e-01 4.61416095e-01
5.02987146e-01 -1.82685964e-02 -1.29828349e-01 6.47425413e-01
1.79384857e-01 -2.29876563e-01 2.39917472e-01 9.39507604e-01
3.44751440e-02 -1.05297852e+00 -1.72207952e-01 6.90290451e-01
-6.53067470e-01 3.16339940e-01 -1.81616709e-01 7.66286135e-01
-3.03309798e-01 1.25539339e+00 3.95452201e-01 -5.66931725e-01
6.33936644e-01 2.30852857e-01 6.14385994e-04 -4.46868032e-01
-3.97052526e-01 1.75772876e-01 -1.60761580e-01 -8.79231989e-01
-1.98761478e-01 -5.92332900e-01 -1.23559809e+00 -2.12224513e-01
-1.96673095e-01 1.55167535e-01 3.09431851e-01 9.52689648e-01
7.68109977e-01 1.04978919e+00 9.45295274e-01 -6.07680857e-01
-6.66281700e-01 -8.69494021e-01 -7.08774686e-01 6.68735683e-01
8.10226738e-01 -5.00948370e-01 -4.86209750e-01 -1.60207778e-01]
|
[7.24104642868042, 2.772202730178833]
|
616e63c7-13ca-4aa7-8f00-78ce8c7039b8
|
assignment-space-based-multi-object-tracking
| null | null |
http://openaccess.thecvf.com//content/ICCV2021/html/Choudhuri_Assignment-Space-Based_Multi-Object_Tracking_and_Segmentation_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Choudhuri_Assignment-Space-Based_Multi-Object_Tracking_and_Segmentation_ICCV_2021_paper.pdf
|
Assignment-Space-Based Multi-Object Tracking and Segmentation
|
Multi-object tracking and segmentation (MOTS) is important for understanding dynamic scenes in video data. Existing methods perform well on multi-object detection and segmentation for independent video frames, but tracking of objects over time remains a challenge. MOTS methods formulate tracking locally, i.e., frame-by-frame, leading to sub-optimal results. Classical global methods on tracking operate directly on object detections, which leads to a combinatorial growth in the detection space. In contrast, we formulate a global method for MOTS over the space of assignments rather than detections: First, we find all top-k assignments of objects detected and segmented between any two consecutive frames and develop a structured prediction formulation to score assignment sequences across any number of consecutive frames. We use dynamic programming to find the global optimizer of this formulation in polynomial time. Second, we connect objects which reappear after having been out of view for some time. For this we formulate an assignment problem. On the challenging KITTI-MOTS and MOTSChallenge datasets, this achieves state-of-the-art results among methods which don't use depth data.
|
['Alexander G. Schwing', 'Girish Chowdhary', 'Anwesa Choudhuri']
|
2021-01-01
| null | null | null |
iccv-2021-1
|
['multi-object-tracking-and-segmentation']
|
['computer-vision']
|
[ 2.43631572e-01 -3.83050263e-01 -1.28547713e-01 3.80232073e-02
-8.89896333e-01 -8.32762063e-01 -8.69552791e-02 1.67354494e-01
-5.47430933e-01 3.89833391e-01 -5.23581505e-01 5.16652539e-02
-1.01826623e-01 -4.18394983e-01 -1.09670210e+00 -7.70564377e-01
-1.76365852e-01 6.93035662e-01 1.28287244e+00 1.70382902e-01
4.17495742e-02 3.62714767e-01 -1.35008490e+00 1.56156093e-01
3.24485749e-01 8.45129132e-01 4.93348300e-01 1.11502349e+00
-4.38522808e-02 7.07772255e-01 -4.17092055e-01 -3.08289737e-01
5.29929280e-01 -4.06542569e-01 -8.33501399e-01 5.07399857e-01
9.69042122e-01 -3.40809673e-01 -1.63450494e-01 1.31722295e+00
1.68536857e-01 2.07547501e-01 3.28580141e-01 -1.56484413e+00
-1.64614379e-01 3.17671210e-01 -1.01744533e+00 5.70464969e-01
1.92446411e-01 1.25525236e-01 1.12073779e+00 -6.73641205e-01
8.81565571e-01 1.17295146e+00 7.88563132e-01 4.85774666e-01
-1.29498196e+00 -3.19794357e-01 5.06488204e-01 2.84994006e-01
-1.31400013e+00 -2.19042882e-01 2.83923209e-01 -8.91990781e-01
4.96954173e-01 4.88750398e-01 1.00610316e+00 2.71328509e-01
1.79868028e-01 1.16733003e+00 6.17346823e-01 -2.03499421e-01
-4.55128364e-02 -1.70647085e-01 1.52205110e-01 9.51628029e-01
2.47744933e-01 -7.35756848e-03 -6.89902425e-01 1.28698379e-01
6.28543317e-01 3.19539398e-01 -1.08747780e-01 -6.19673908e-01
-1.42982173e+00 6.40192866e-01 2.51826853e-01 -4.98425700e-02
-3.03697914e-01 5.24048090e-01 2.48950645e-01 1.22060798e-01
2.71905243e-01 8.98039863e-02 -8.04142594e-01 -1.46456854e-02
-1.17562389e+00 3.96474481e-01 4.95257556e-01 1.04295325e+00
9.73389268e-01 -3.50297660e-01 -2.65216798e-01 4.71334815e-01
3.03808898e-01 5.94236672e-01 -5.17385639e-02 -1.13325751e+00
3.31084490e-01 5.33598125e-01 2.72067189e-01 -1.07936203e+00
-4.40511972e-01 -1.66692495e-01 -2.90438503e-01 2.66126782e-01
6.78034663e-01 -1.71948805e-01 -9.47753668e-01 1.73380828e+00
8.91563892e-01 6.29960775e-01 -3.17812502e-01 1.07931435e+00
3.89773935e-01 6.19961679e-01 -1.76379129e-01 -4.44277287e-01
1.36175442e+00 -1.39565229e+00 -4.94501859e-01 -3.61982942e-01
5.70694804e-01 -6.55350089e-01 3.68397236e-01 2.86726296e-01
-1.25334561e+00 -4.62061763e-01 -6.47238314e-01 1.76528841e-01
-2.14303032e-01 -5.60974777e-02 3.48931372e-01 5.25080681e-01
-1.18251681e+00 4.86051142e-01 -1.24102199e+00 -3.88405263e-01
4.31603163e-01 6.15880072e-01 -6.64450526e-02 -7.44829699e-02
-5.43130338e-01 6.63916111e-01 6.38075113e-01 9.46603119e-02
-1.08335900e+00 -7.64791369e-01 -6.24195576e-01 -2.13278472e-01
9.92566884e-01 -6.84726715e-01 1.34285033e+00 -9.38470542e-01
-1.06212020e+00 8.02259147e-01 -5.01166224e-01 -5.78752875e-01
7.07432032e-01 -3.20764750e-01 4.95377881e-03 2.46244952e-01
3.91426057e-01 9.10835326e-01 8.03567231e-01 -1.07046306e+00
-1.40043533e+00 -2.59622753e-01 3.44984621e-01 1.80479631e-01
-1.83683425e-01 3.51430625e-01 -1.28591990e+00 -2.95520544e-01
2.36676291e-01 -1.19228899e+00 -4.97722387e-01 2.95880079e-01
-3.87983680e-01 -2.16321453e-01 1.13856804e+00 -4.89817590e-01
1.06283891e+00 -1.84468722e+00 5.54683447e-01 -1.57543898e-01
3.99571091e-01 1.60233870e-01 2.45016292e-02 -1.67744130e-01
3.74159962e-01 -1.75123245e-01 1.40093500e-02 -7.07715690e-01
-1.87414065e-01 3.30565870e-01 1.05852902e-01 6.46052241e-01
7.88803250e-02 1.01011217e+00 -1.13063979e+00 -9.13294196e-01
3.83824348e-01 -9.50156525e-02 -7.95287013e-01 -2.61599496e-02
-5.82947195e-01 5.56420505e-01 -3.80434304e-01 8.48289192e-01
7.34067857e-01 -5.62817574e-01 -7.05243945e-02 -3.45234163e-02
-4.39440191e-01 -3.10838759e-01 -1.50288844e+00 1.61257493e+00
2.30573297e-01 7.29466200e-01 1.78117126e-01 -8.89612615e-01
1.96193099e-01 5.00336066e-02 1.17937970e+00 -1.80703431e-01
4.16969657e-02 1.72549918e-01 -2.12174833e-01 -3.42780769e-01
6.55158758e-01 4.70504910e-02 2.63178889e-02 1.74981095e-02
-9.84702036e-02 4.32808727e-01 7.90693462e-01 3.29458535e-01
1.27361381e+00 2.10453615e-01 -2.00244598e-02 -1.00649195e-02
4.30099368e-01 4.49953884e-01 8.69547784e-01 1.02878165e+00
-3.97480100e-01 4.73207414e-01 3.43311846e-01 -4.34620470e-01
-7.77103841e-01 -7.88578629e-01 1.71429947e-01 1.32523525e+00
7.27967918e-01 -4.13221985e-01 -7.23708093e-01 -6.15806520e-01
-6.12485558e-02 -1.10386387e-01 -5.13330281e-01 3.07995409e-01
-7.68714011e-01 -5.31074822e-01 1.20186187e-01 5.06040394e-01
-9.81812272e-03 -9.35600758e-01 -1.21472192e+00 5.23270488e-01
-2.98351020e-01 -1.54103041e+00 -8.72422755e-01 2.01630130e-01
-8.13428760e-01 -1.25762951e+00 -7.50067115e-01 -6.35808170e-01
6.18443489e-01 6.14645422e-01 1.05090165e+00 2.29596660e-01
-4.79206055e-01 4.44730431e-01 -2.09760696e-01 -2.18682036e-01
5.07476181e-02 -2.32592002e-02 -1.55755887e-02 4.04179096e-02
1.03583843e-01 5.48375025e-02 -6.02230787e-01 6.13736093e-01
-6.15599155e-01 2.01424003e-01 3.23202819e-01 4.39922869e-01
1.16002536e+00 2.63021095e-03 -4.68654484e-02 -6.91430032e-01
-3.80717069e-01 -3.52021486e-01 -1.22176671e+00 3.95143539e-01
-4.49780226e-02 -2.10772425e-01 1.53738648e-01 -5.93116820e-01
-4.22002286e-01 7.02247798e-01 3.23879480e-01 -1.02160072e+00
1.91156030e-01 1.02867924e-01 1.27043217e-01 -3.76760483e-01
1.82757199e-01 8.14204291e-02 -2.79958367e-01 -2.79556334e-01
1.30892321e-01 -5.39932847e-02 7.20785677e-01 -5.04812300e-01
8.95128429e-01 7.14662910e-01 1.51940748e-01 -7.24083960e-01
-1.17406857e+00 -1.19692409e+00 -8.14866364e-01 -7.78758585e-01
1.26787436e+00 -8.40151370e-01 -9.42297161e-01 3.09311777e-01
-1.22497845e+00 -3.45862210e-01 -2.24385351e-01 3.31825286e-01
-5.71657658e-01 3.21472824e-01 -3.90208453e-01 -9.62092221e-01
8.88909027e-02 -1.34067059e+00 1.36284196e+00 2.98146576e-01
9.43362489e-02 -8.56391847e-01 5.77948056e-02 3.03205311e-01
-4.52526323e-02 4.82046306e-01 -1.74373365e-03 -1.72482550e-01
-1.43946290e+00 -1.16655469e-01 -9.37091112e-02 -1.56683177e-01
-1.16177753e-01 4.55769673e-02 -3.33953828e-01 -4.66841847e-01
-3.59404385e-01 3.27125937e-02 9.67556357e-01 8.73240292e-01
1.17382646e+00 -1.75691426e-01 -8.26666534e-01 7.19531476e-01
1.62093866e+00 2.08830968e-01 2.26311982e-01 4.01398867e-01
1.02145171e+00 3.22502345e-01 9.43723500e-01 3.28029543e-01
3.84674907e-01 1.05137682e+00 5.60258210e-01 -2.22025719e-02
-1.75460532e-01 1.02393948e-01 4.44522530e-01 4.55840528e-01
-4.20490233e-03 -3.48329604e-01 -8.77562523e-01 9.22343612e-01
-2.40201974e+00 -9.60788071e-01 -6.63504243e-01 2.15655828e+00
4.90373433e-01 2.03874961e-01 5.22628844e-01 -2.03844264e-01
1.06273353e+00 -5.13141938e-02 -7.73707926e-01 2.70442933e-01
-4.77690957e-02 -2.36468226e-01 1.00980484e+00 6.05680764e-01
-1.40189576e+00 1.10031092e+00 5.98591661e+00 6.89702809e-01
-7.95660257e-01 3.77930403e-01 2.10961908e-01 -4.35858995e-01
2.98665255e-01 2.66735494e-01 -1.36980474e+00 6.06838346e-01
6.16670132e-01 1.77536942e-02 3.65559459e-01 7.36714005e-01
7.45287985e-02 -2.52103448e-01 -1.21498036e+00 8.70008171e-01
-1.06180638e-01 -1.40267563e+00 -3.87634963e-01 2.79739369e-02
1.00505590e+00 1.55127242e-01 -1.73110604e-01 1.46167967e-02
3.72660637e-01 -3.02187294e-01 1.23356199e+00 3.28909069e-01
3.76897454e-01 -3.46378803e-01 3.91686827e-01 4.08834487e-01
-1.76234210e+00 -1.01543546e-01 -3.34748298e-01 2.90314257e-01
6.91857696e-01 2.63983965e-01 -5.42195916e-01 4.51978266e-01
8.67568910e-01 7.94090748e-01 -5.68658590e-01 1.57577550e+00
1.41416386e-01 2.73140758e-01 -7.52299905e-01 4.90225814e-02
2.14117005e-01 -1.57505766e-01 8.80591571e-01 1.21550930e+00
2.95250505e-01 9.66893956e-02 1.02028632e+00 5.31423390e-01
3.41951512e-02 -2.56465465e-01 -8.23204890e-02 8.82612616e-02
3.49415988e-01 1.34594834e+00 -1.19384527e+00 -4.48817074e-01
-5.72994471e-01 1.01331806e+00 2.76571602e-01 3.27630714e-02
-1.16571975e+00 1.32197633e-01 5.77386618e-01 1.66241854e-01
8.53565693e-01 -2.73216009e-01 5.14474921e-02 -1.01190722e+00
1.02508493e-01 -4.57767367e-01 7.14404464e-01 -5.60329676e-01
-8.56474996e-01 8.95905122e-02 -2.91765761e-02 -1.23869050e+00
3.55047621e-02 -5.73035717e-01 -3.38637769e-01 2.91164011e-01
-1.49098921e+00 -1.12492168e+00 -3.38545561e-01 7.16098368e-01
9.30571973e-01 3.42524976e-01 1.45417780e-01 7.54448295e-01
-5.54461718e-01 3.29697311e-01 6.51114210e-02 1.89977631e-01
4.62018907e-01 -1.31758690e+00 5.89339510e-02 1.26131952e+00
2.70652205e-01 3.01084578e-01 8.20454180e-01 -7.01594830e-01
-1.45914435e+00 -1.30093515e+00 5.34454286e-01 -7.71091759e-01
7.41218448e-01 -1.21683329e-01 -6.48990393e-01 8.83785188e-01
-6.00728691e-02 3.51375967e-01 2.78051019e-01 -9.51024890e-02
1.08877823e-01 1.43379122e-01 -7.47024179e-01 4.11879838e-01
1.22185695e+00 5.27999364e-02 -1.37780949e-01 6.75951123e-01
8.58857155e-01 -9.93249297e-01 -5.22240818e-01 1.79325536e-01
4.44348216e-01 -6.39009297e-01 1.09975612e+00 -5.37435412e-01
1.88844845e-01 -9.87109303e-01 -1.89119712e-01 -6.61572754e-01
-2.14352533e-01 -8.97901535e-01 -6.18012547e-01 9.64663088e-01
2.62207240e-01 -1.78679302e-01 1.04312694e+00 3.21896583e-01
-2.41041288e-01 -7.54906118e-01 -9.24703777e-01 -1.06694877e+00
-4.12856400e-01 -3.41412902e-01 8.47745091e-02 5.72688699e-01
-6.21292949e-01 1.41359061e-01 -5.60790777e-01 6.76812232e-01
1.04153085e+00 3.58914286e-01 9.96269166e-01 -1.22911990e+00
-5.00881255e-01 -3.51318836e-01 -5.54803610e-01 -1.56749296e+00
-1.36434197e-01 -6.61487520e-01 5.03702164e-01 -1.50961530e+00
6.50728583e-01 -4.42267150e-01 -2.27222785e-01 5.75794160e-01
-4.56298083e-01 3.99572462e-01 6.09552801e-01 3.48780245e-01
-1.57515955e+00 -4.08153422e-02 1.22460318e+00 -1.70365959e-01
-2.73690462e-01 8.46281722e-02 -2.53661275e-01 8.12732577e-01
3.60967934e-01 -1.00444829e+00 3.97442691e-02 -6.17378354e-01
1.40132844e-01 3.23650539e-01 5.70587397e-01 -1.12449300e+00
7.64612675e-01 -4.03536439e-01 2.15716437e-01 -1.22685766e+00
6.17194474e-01 -8.90978277e-01 3.18119019e-01 7.29160190e-01
-4.76914123e-02 1.91215575e-01 2.16542050e-01 9.51271355e-01
3.02722398e-02 -3.52495492e-01 9.60210204e-01 -9.22049582e-02
-1.10381019e+00 8.29681575e-01 -2.28047445e-01 1.71414703e-01
1.48376310e+00 -5.43245614e-01 -8.88999924e-02 1.68040738e-01
-9.10695910e-01 8.48024189e-01 4.50085253e-01 4.05715466e-01
3.91869932e-01 -1.15361989e+00 -5.91616273e-01 -3.30054224e-01
-1.00084886e-01 3.78576785e-01 4.13694650e-01 1.23396015e+00
-4.31779593e-01 3.03543299e-01 7.80162066e-02 -1.35740745e+00
-1.86367261e+00 6.53469324e-01 2.91602910e-01 -2.87874877e-01
-7.51271307e-01 1.19870412e+00 2.64345318e-01 -8.70660692e-03
2.33157679e-01 -1.71830878e-01 5.58322109e-02 2.14290217e-01
3.42241883e-01 4.61191148e-01 -1.99197054e-01 -6.89008415e-01
-6.00656331e-01 9.85675931e-01 -2.74708480e-01 -1.83874249e-01
1.11337328e+00 -2.94181228e-01 1.43487137e-02 4.45142329e-01
1.04842997e+00 -9.95856896e-02 -1.71489894e+00 -2.96509057e-01
1.33852497e-01 -8.35586369e-01 -2.02397496e-01 -3.21871638e-01
-1.30718827e+00 5.12507200e-01 7.18727231e-01 3.31576139e-01
9.38292027e-01 3.90554219e-01 8.94736528e-01 2.70657271e-01
3.89132798e-01 -9.94239628e-01 3.12760979e-01 6.17843091e-01
1.96470752e-01 -1.20368302e+00 -4.77240719e-02 -4.25153732e-01
-4.68372583e-01 9.49505448e-01 7.13855982e-01 1.24369189e-01
3.01002860e-01 3.30392748e-01 -2.63145179e-01 -2.76036888e-01
-5.96777201e-01 -5.39899886e-01 2.49109849e-01 3.39046806e-01
-1.63275912e-01 4.00626883e-02 6.48675263e-02 3.23775620e-03
1.61313415e-01 -4.75777909e-02 3.79409671e-01 1.07584751e+00
-8.48575592e-01 -9.39195514e-01 -6.15105808e-01 3.24564129e-01
-5.86454690e-01 2.05211952e-01 -1.51552245e-01 5.69233537e-01
3.48928452e-01 8.90014291e-01 -7.19080344e-02 -1.66484118e-01
1.75776854e-01 -2.29050085e-01 6.13843024e-01 -6.87937438e-01
-2.03818619e-01 3.24834287e-01 -3.06650341e-01 -6.19896829e-01
-8.18314850e-01 -1.23807502e+00 -1.36207151e+00 -2.12281689e-01
-8.93103838e-01 -9.61674824e-02 4.55309033e-01 1.05281782e+00
5.33143915e-02 8.41242135e-01 2.97629207e-01 -1.05705810e+00
-2.61766389e-02 -3.82168591e-01 -2.63501614e-01 2.63707280e-01
7.42027640e-01 -7.47261465e-01 -9.02380347e-02 4.75123644e-01]
|
[6.437530517578125, -2.031482219696045]
|
43d8e0a7-9bd2-4137-b23c-45c04c01e615
|
dense-3d-point-cloud-reconstruction-using-a
|
1901.08906
| null |
http://arxiv.org/abs/1901.08906v1
|
http://arxiv.org/pdf/1901.08906v1.pdf
|
Dense 3D Point Cloud Reconstruction Using a Deep Pyramid Network
|
Reconstructing a high-resolution 3D model of an object is a challenging task
in computer vision. Designing scalable and light-weight architectures is
crucial while addressing this problem. Existing point-cloud based
reconstruction approaches directly predict the entire point cloud in a single
stage. Although this technique can handle low-resolution point clouds, it is
not a viable solution for generating dense, high-resolution outputs. In this
work, we introduce DensePCR, a deep pyramidal network for point cloud
reconstruction that hierarchically predicts point clouds of increasing
resolution. Towards this end, we propose an architecture that first predicts a
low-resolution point cloud, and then hierarchically increases the resolution by
aggregating local and global point features to deform a grid. Our method
generates point clouds that are accurate, uniform and dense. Through extensive
quantitative and qualitative evaluation on synthetic and real datasets, we
demonstrate that DensePCR outperforms the existing state-of-the-art point cloud
reconstruction works, while also providing a light-weight and scalable
architecture for predicting high-resolution outputs.
|
['R. Venkatesh Babu', 'Priyanka Mandikal']
|
2019-01-25
| null | null | null | null |
['3d-point-cloud-reconstruction', 'point-cloud-reconstruction']
|
['computer-vision', 'computer-vision']
|
[-1.05284609e-01 -1.52525976e-01 2.13396266e-01 -3.67030531e-01
-8.68520319e-01 -3.53831798e-01 5.36710560e-01 4.09908332e-02
1.43500492e-01 2.66961902e-01 -2.08243549e-01 -2.82971971e-02
2.17763111e-01 -1.21947050e+00 -1.15505993e+00 -2.18825117e-01
6.67058676e-02 1.08117366e+00 6.44418836e-01 -1.19974725e-01
2.16267392e-01 1.35803556e+00 -1.82458246e+00 4.88443345e-01
6.85191154e-01 1.07906163e+00 5.17939210e-01 4.52170849e-01
-3.08704168e-01 4.00895625e-01 3.07541993e-02 -2.74076879e-01
4.62409139e-01 3.31642359e-01 -6.06553733e-01 1.76717669e-01
9.29384887e-01 -5.15975296e-01 3.77464555e-02 7.39218175e-01
2.29557917e-01 -2.53323793e-01 5.96638024e-01 -9.41940248e-01
-6.12145007e-01 2.09855959e-02 -7.60085762e-01 -3.20607752e-01
1.11841895e-01 1.35350615e-01 7.75130272e-01 -1.49571252e+00
6.00980222e-01 1.38272977e+00 1.13765001e+00 3.84420276e-01
-1.39686596e+00 -7.94422328e-01 1.60746098e-01 -2.98764557e-01
-1.62148619e+00 -1.56932265e-01 1.05225432e+00 -5.30203164e-01
1.14498031e+00 2.04733223e-01 9.58777964e-01 8.53703320e-01
1.09064840e-01 4.22235101e-01 7.28775978e-01 -4.86794710e-02
3.28028262e-01 -2.54128098e-01 -3.48173678e-01 5.22810519e-01
2.60461032e-01 1.92007318e-01 -4.75944668e-01 -4.49760556e-01
1.56019187e+00 4.52090144e-01 -1.27256205e-02 -6.40227079e-01
-1.24293756e+00 6.35507941e-01 9.43391979e-01 -5.85578848e-03
-9.46597695e-01 5.16826749e-01 -1.16417088e-01 -1.54493541e-01
8.09684217e-01 2.91087240e-01 -3.72085303e-01 1.39892027e-01
-1.28230298e+00 5.33664525e-01 4.25988317e-01 1.17614877e+00
1.02739370e+00 3.26446109e-02 3.23100239e-01 5.84786475e-01
5.18237650e-01 5.70104361e-01 -2.85324365e-01 -1.35712028e+00
4.08312291e-01 9.41723824e-01 3.85721922e-01 -1.15808606e+00
-2.51662999e-01 -5.32350302e-01 -1.00182092e+00 8.04865122e-01
-3.35447490e-01 3.58231366e-01 -9.65242624e-01 1.02336323e+00
5.91616213e-01 5.36893010e-01 -1.21946990e-01 1.15861905e+00
8.48636627e-01 9.64409411e-01 -6.04586825e-02 1.35781363e-01
1.05165613e+00 -8.01921666e-01 -1.03746794e-01 -1.95786074e-01
-1.49882957e-01 -5.17248631e-01 9.05197024e-01 3.66793931e-01
-1.38229597e+00 -6.11614347e-01 -7.84272373e-01 -5.10456860e-01
1.73936859e-01 -6.71775490e-02 4.55529809e-01 -1.45590901e-01
-1.20867181e+00 8.18673551e-01 -1.16652191e+00 9.13216993e-02
7.80864835e-01 2.58572787e-01 -5.13087273e-01 -2.46550009e-01
-3.26255441e-01 6.96406782e-01 5.93934506e-02 2.44667567e-02
-8.19790542e-01 -1.15191948e+00 -6.54556036e-01 2.30524182e-01
-2.05867931e-01 -1.18633044e+00 1.19787097e+00 -6.02537751e-01
-1.25129616e+00 8.64558280e-01 -2.53999382e-01 -4.04305935e-01
5.45198619e-01 -1.03714325e-01 3.14459920e-01 1.68580323e-01
3.04348059e-02 1.04172587e+00 8.48678470e-01 -1.92483723e+00
-6.43262804e-01 -6.39032781e-01 -2.13183373e-01 2.16490269e-01
1.78696513e-01 -2.25725397e-01 -4.94054884e-01 -2.10950017e-01
5.72725296e-01 -6.10422492e-01 -6.40142322e-01 5.75025797e-01
-1.77760869e-01 -2.29397148e-01 9.10867393e-01 -3.96358043e-01
3.74479175e-01 -1.92468441e+00 5.29635549e-02 7.75129870e-02
5.45103073e-01 1.01888534e-02 -1.74580216e-02 2.67309785e-01
2.33964562e-01 1.40104860e-01 -3.76669466e-01 -9.50971544e-01
-1.31498992e-01 2.92354137e-01 -6.41244650e-01 3.66701424e-01
4.94009256e-01 9.64850366e-01 -6.64411783e-01 -4.68477339e-01
5.65458834e-01 1.28155434e+00 -7.34823823e-01 1.61497518e-01
-6.15273178e-01 5.36207438e-01 -4.63230520e-01 8.07152569e-01
1.03438377e+00 -6.23716772e-01 -3.03033292e-01 -3.38349730e-01
-4.93630677e-01 -1.27808598e-03 -1.03434932e+00 1.99883413e+00
-4.86693889e-01 4.11687016e-01 1.48264468e-01 -3.04457664e-01
1.30079222e+00 2.22381830e-01 7.27871656e-01 -4.82647747e-01
-3.76230702e-02 2.80762166e-01 -6.09006345e-01 1.41583085e-01
7.51736820e-01 -3.29218328e-01 2.26955324e-01 3.80307399e-02
-1.84954211e-01 -6.78873122e-01 -6.26760006e-01 -5.62316328e-02
9.32640791e-01 4.21795100e-01 -2.11632587e-02 1.23959027e-01
1.76882535e-01 5.13637304e-01 4.95134622e-01 3.12872857e-01
4.26531166e-01 1.37563026e+00 9.80667211e-03 -7.84620523e-01
-1.65310919e+00 -1.25381732e+00 -3.11078727e-01 3.68118584e-01
2.80798942e-01 -3.99237216e-01 -4.03404713e-01 -1.11316651e-01
2.30068102e-01 3.40044856e-01 -2.32951835e-01 3.59531850e-01
-6.81000173e-01 -4.77433950e-02 -4.69142720e-02 6.68962419e-01
3.67091388e-01 -1.12768543e+00 -7.32215762e-01 2.53708422e-01
-1.34834319e-01 -1.18610406e+00 1.66066781e-01 -1.72015443e-01
-1.38390386e+00 -7.98903883e-01 -5.49638152e-01 -7.78918624e-01
8.03634107e-01 4.89584595e-01 1.57269871e+00 2.04623237e-01
-8.08843374e-02 1.81932122e-01 -1.78490177e-01 -4.54107612e-01
-2.51353741e-01 9.53512825e-03 -2.36780476e-02 -2.91184276e-01
1.66814014e-01 -1.03793943e+00 -6.99336827e-01 1.15235612e-01
-7.81497240e-01 4.51519668e-01 6.43212795e-01 3.34530771e-01
1.62740791e+00 -1.24181230e-02 7.13656396e-02 -5.88968754e-01
2.11555108e-01 -4.28337306e-01 -9.91991520e-01 -1.88350886e-01
-2.85858333e-01 -2.17183262e-01 7.89713562e-01 -7.62869194e-02
-6.45290077e-01 7.08954394e-01 -5.67150533e-01 -1.32743490e+00
-2.20170647e-01 1.79868892e-01 2.18822643e-01 -3.93340617e-01
7.30315745e-01 3.04568082e-01 -1.01685420e-01 -8.45590591e-01
3.38681936e-01 3.40633839e-01 6.42940640e-01 -4.51661259e-01
1.06448793e+00 1.00902879e+00 2.71171153e-01 -7.22729087e-01
-6.37565732e-01 -4.78516906e-01 -1.02446032e+00 -1.60032943e-01
7.42696762e-01 -1.30456769e+00 -6.46575630e-01 1.53175935e-01
-1.55785632e+00 -2.66407609e-01 -5.09126425e-01 7.07654878e-02
-8.29513013e-01 1.05877668e-01 -4.74680126e-01 -6.82527721e-01
-6.09132767e-01 -1.06572163e+00 1.77245617e+00 -1.10014267e-01
7.07525238e-02 -5.30076861e-01 1.63850740e-01 2.72801608e-01
3.12931389e-01 6.65635765e-01 4.24168676e-01 2.38045231e-01
-1.42598283e+00 -2.04131201e-01 -3.97679061e-01 1.15483917e-01
-1.87317133e-01 1.95487455e-01 -7.90620446e-01 -1.47072777e-01
-7.78561831e-02 -3.09630126e-01 5.89512229e-01 3.75926942e-01
1.49847329e+00 -2.58364767e-01 -4.19554710e-01 1.00922883e+00
1.83773232e+00 -5.04230618e-01 6.78210557e-01 2.80577928e-01
9.89570200e-01 2.34472647e-01 5.77688396e-01 4.50248897e-01
6.45682812e-01 6.75360322e-01 1.10659778e+00 -2.59551764e-01
-9.13800299e-02 -5.84535480e-01 -1.80420771e-01 6.96233630e-01
-3.90041918e-01 2.33122945e-01 -1.23297191e+00 5.74553967e-01
-1.84771097e+00 -8.87602270e-01 -5.31373560e-01 1.94199705e+00
6.71072721e-01 -1.04503617e-01 -7.01599866e-02 -5.51458038e-02
3.70492220e-01 -2.76826434e-02 -6.28306448e-01 -2.94734836e-02
1.10863633e-01 4.45255220e-01 2.93932647e-01 4.48060155e-01
-8.28131914e-01 1.13585424e+00 6.26222610e+00 4.49257374e-01
-1.33022118e+00 1.61987945e-01 3.46241176e-01 -2.42437571e-01
-4.64719027e-01 -7.87184685e-02 -8.23102057e-01 1.97175160e-01
7.48930395e-01 1.40888005e-01 1.16983756e-01 1.24509060e+00
1.57886654e-01 3.77558827e-01 -1.07486951e+00 1.31858671e+00
-2.25592151e-01 -2.06947184e+00 4.33764577e-01 2.77495772e-01
8.15464914e-01 6.98643625e-01 -1.71495359e-02 -6.53319210e-02
4.49440330e-01 -1.20319033e+00 9.49354112e-01 7.50593185e-01
1.00261247e+00 -8.52980912e-01 3.69574666e-01 8.45995903e-01
-1.29518652e+00 2.74125069e-01 -9.38969791e-01 -4.42946292e-02
2.76828170e-01 6.67006910e-01 -6.58796430e-01 4.75441962e-01
1.08783507e+00 8.57463181e-01 -3.76594067e-01 1.10791576e+00
5.77983670e-02 1.76133126e-01 -5.82090497e-01 4.56386954e-01
1.39789388e-01 -1.67541578e-01 4.06033933e-01 8.57929528e-01
5.36730945e-01 3.46840978e-01 2.66593814e-01 1.45965910e+00
-1.57481804e-01 -1.40009984e-01 -8.35062206e-01 4.78123546e-01
6.49927378e-01 1.36576390e+00 -4.99559492e-01 -1.22921355e-01
-2.54823923e-01 7.67195106e-01 8.54245126e-01 -6.16038330e-02
-5.45331419e-01 3.30648750e-01 8.30358148e-01 6.40040874e-01
4.24371034e-01 -6.44932270e-01 -7.46513188e-01 -9.89455521e-01
2.16896072e-01 -3.35269421e-01 -3.84273112e-01 -1.27433324e+00
-1.36034417e+00 8.99615109e-01 -2.31584147e-01 -1.71574354e+00
-1.98352590e-01 -3.71994227e-01 -3.68603915e-01 1.24935794e+00
-1.84834635e+00 -1.52841997e+00 -7.86459684e-01 6.04663730e-01
5.52405119e-01 1.29537463e-01 8.88930619e-01 8.67766514e-02
1.03287809e-01 -9.08432007e-02 -2.07494318e-01 -1.06275715e-01
-4.90624048e-02 -9.59435821e-01 9.38694656e-01 6.09170377e-01
1.15120374e-01 5.35283387e-01 4.94234860e-01 -7.65146613e-01
-1.58227277e+00 -1.60059416e+00 5.40539801e-01 -5.85116446e-01
1.03647083e-01 -4.37221736e-01 -1.30454218e+00 6.24110281e-01
-3.97723705e-01 6.02670372e-01 1.41053557e-01 -1.67597681e-01
-3.30321372e-01 -1.49862673e-02 -1.39120495e+00 4.41898137e-01
1.17961133e+00 -3.75735730e-01 -4.38886315e-01 3.20522368e-01
9.25253630e-01 -7.44161069e-01 -1.17987132e+00 5.54344237e-01
3.30665588e-01 -1.20928955e+00 1.41645527e+00 -1.22541890e-01
9.34617341e-01 -4.68140781e-01 -1.90451875e-01 -1.20386398e+00
-8.50096345e-01 -2.31018942e-02 -2.71829754e-01 6.68589175e-01
1.58559978e-01 -2.82270730e-01 1.29902256e+00 4.57515508e-01
-5.00169873e-01 -9.78392899e-01 -9.36160922e-01 -5.11543214e-01
3.40675771e-01 -5.17561555e-01 9.87491012e-01 8.74197841e-01
-6.90482259e-01 1.79474756e-01 -9.53952000e-02 7.17539251e-01
9.93298948e-01 5.16333163e-01 1.00874674e+00 -1.75134003e+00
2.62150109e-01 -2.41529435e-01 -6.18156731e-01 -1.10486281e+00
9.81604904e-02 -8.16287398e-01 6.90445080e-02 -2.16923332e+00
-5.53888753e-02 -9.04354215e-01 3.75401735e-01 3.87355745e-01
3.11924011e-01 6.34521425e-01 2.32026652e-01 7.51343071e-01
-2.84894973e-01 6.51049614e-01 1.30472410e+00 7.24247715e-04
-7.47500733e-02 -5.00854775e-02 -4.09644455e-01 7.89168119e-01
6.71068668e-01 -3.94260049e-01 -1.93287596e-01 -9.64584947e-01
2.48484582e-01 1.79244146e-01 7.31818020e-01 -1.32612443e+00
3.10696065e-01 -2.07243636e-01 7.42083073e-01 -1.43309224e+00
9.92181063e-01 -1.33811879e+00 5.96406698e-01 5.77356629e-02
1.82192504e-01 1.72317833e-01 1.99573413e-01 5.14645040e-01
-1.72672883e-01 2.08069950e-01 9.10456955e-01 -5.90521932e-01
-4.87486392e-01 1.02482367e+00 3.66298437e-01 -6.16183698e-01
9.68368888e-01 -4.41553682e-01 -9.79460850e-02 1.19781494e-01
-5.04482627e-01 2.16350213e-01 1.14410305e+00 4.20878232e-01
1.33052087e+00 -1.48909891e+00 -9.32060361e-01 4.11128610e-01
1.72009051e-01 1.23215771e+00 2.37678841e-01 1.58974692e-01
-1.04556704e+00 2.59477377e-01 -3.11557263e-01 -1.26399386e+00
-1.09359086e+00 3.83850276e-01 4.90606248e-01 6.30766526e-02
-1.34173357e+00 7.27469206e-01 1.48710996e-01 -7.74220943e-01
-2.91842744e-02 -5.32139122e-01 -4.15359735e-02 -5.38625240e-01
3.41890603e-01 8.85550678e-03 2.57164925e-01 -9.36289489e-01
-1.85567990e-01 9.95567739e-01 2.22702801e-01 4.32765745e-02
1.76073921e+00 1.46684900e-01 -2.90794790e-01 5.07726610e-01
9.83125329e-01 -2.86781341e-01 -1.65809584e+00 -2.59088963e-01
-3.73843640e-01 -8.00482988e-01 3.29537988e-01 -3.29713255e-01
-1.15891635e+00 9.48771000e-01 2.61023730e-01 1.03914693e-01
7.85231650e-01 1.75335392e-01 8.41341197e-01 2.19284311e-01
9.92189407e-01 -4.05002475e-01 -1.02616072e-01 5.43027878e-01
1.34329224e+00 -1.16469467e+00 2.40551338e-01 -7.25451291e-01
-3.15089852e-01 1.10327339e+00 7.08225250e-01 -6.82216167e-01
6.18340075e-01 4.14796174e-01 -2.83778459e-01 -6.44551098e-01
-9.62486982e-01 1.16246358e-01 1.72960028e-01 7.51635611e-01
1.36016801e-01 1.78677067e-01 4.21468168e-01 2.02524140e-01
-5.58231890e-01 2.03107044e-01 3.26219112e-01 7.48682439e-01
-6.13118112e-01 -7.99416959e-01 -6.95258677e-01 4.35416430e-01
3.50630432e-02 5.06743677e-02 -4.11960036e-01 4.82969433e-01
8.58873781e-03 3.58803868e-01 6.76261246e-01 -4.09000486e-01
5.47558129e-01 -4.54286009e-01 4.01725769e-01 -8.52475762e-01
-5.36868274e-01 -1.67851508e-01 -4.67812628e-01 -8.22669744e-01
-2.44038999e-01 -5.89564204e-01 -1.41109335e+00 -5.21825671e-01
4.75831702e-02 -2.00393200e-01 1.01572847e+00 4.07517523e-01
7.56872594e-01 3.74691129e-01 6.24312341e-01 -1.81513309e+00
-2.97233433e-01 -7.11270988e-01 -3.53387028e-01 2.06474140e-01
5.02245009e-01 -5.46786487e-01 -1.59302995e-01 -7.37437382e-02]
|
[8.406600952148438, -3.5586838722229004]
|
30ee8969-81bc-4bd3-8481-83e4e722d5bb
|
gred-graph-regularized-3d-shape
|
1309.4426
| null |
http://arxiv.org/abs/1309.4426v1
|
http://arxiv.org/pdf/1309.4426v1.pdf
|
GRED: Graph-Regularized 3D Shape Reconstruction from Highly Anisotropic and Noisy Images
|
Analysis of microscopy images can provide insight into many biological
processes. One particularly challenging problem is cell nuclear segmentation in
highly anisotropic and noisy 3D image data. Manually localizing and segmenting
each and every cell nuclei is very time consuming, which remains a bottleneck
in large scale biological experiments. In this work we present a tool for
automated segmentation of cell nuclei from 3D fluorescent microscopic data. Our
tool is based on state-of-the-art image processing and machine learning
techniques and supports a friendly graphical user interface (GUI). We show that
our tool is as accurate as manual annotation but greatly reduces the time for
the registration.
|
['Gunnar Rätsch', 'Xinghua Lou', 'Christian Widmer', 'Stephanie Heinrich', 'Philipp Drewe', 'Shefali Umrania']
|
2013-09-17
| null | null | null | null |
['nuclear-segmentation']
|
['medical']
|
[ 2.49452284e-03 -3.55867416e-01 4.18229669e-01 -2.61023670e-01
-5.47500253e-01 -7.21656501e-01 1.06246792e-01 6.57559752e-01
-1.07030308e+00 7.57703543e-01 -6.50868475e-01 -3.70836169e-01
2.89759487e-01 -5.42035639e-01 -2.30857536e-01 -1.03203464e+00
1.81425229e-01 1.09059870e+00 8.52738976e-01 1.96529716e-01
4.18355376e-01 1.23433459e+00 -9.17878389e-01 -4.40525735e-04
3.50459367e-01 3.40992242e-01 5.10073662e-01 1.22727859e+00
-3.29309583e-01 1.75745025e-01 -5.91054380e-01 -1.05589712e-02
-2.92730313e-02 -4.14827615e-01 -1.17517686e+00 2.33542994e-01
1.22401774e-01 7.44582340e-02 4.47450131e-01 1.04663563e+00
6.39605343e-01 -1.42762378e-01 8.15589845e-01 -8.59984815e-01
-7.22939521e-02 -9.15264934e-02 -5.71723342e-01 7.02886045e-01
7.07390532e-02 -2.26077600e-03 1.80960521e-01 -6.30593956e-01
1.10331666e+00 9.51244295e-01 6.25235021e-01 5.70676982e-01
-1.73882282e+00 -2.64747858e-01 -3.33385408e-01 -9.75847170e-02
-1.38656342e+00 -2.58798569e-01 4.03992712e-01 -8.62655282e-01
9.52183783e-01 2.26478636e-01 9.07948971e-01 2.15684190e-01
4.35478777e-01 3.30607384e-01 1.48920035e+00 -4.64987785e-01
4.03226823e-01 -8.66832137e-02 3.19441915e-01 6.98002875e-01
1.34657517e-01 -5.97406268e-01 2.22927183e-01 -1.38266310e-01
1.25094914e+00 8.77706241e-03 -1.59482822e-01 -2.80164272e-01
-1.61004090e+00 3.87874544e-01 -2.16484427e-01 9.13742840e-01
-1.12675436e-01 1.60212314e-03 2.70564467e-01 8.47733691e-02
3.36104453e-01 2.52072752e-01 -5.10380447e-01 -2.87659526e-01
-1.17926157e+00 1.04460590e-01 7.33770669e-01 4.48579222e-01
9.27108467e-01 -4.35958892e-01 2.85401195e-01 6.08120918e-01
2.00356275e-01 3.97807628e-01 3.80169362e-01 -1.23995793e+00
-4.23617303e-01 7.15365767e-01 1.80513654e-02 -7.63577104e-01
-8.33837986e-01 2.43769422e-01 -8.43825042e-01 7.16016233e-01
1.08671367e+00 3.91618237e-02 -1.01882648e+00 1.01861811e+00
6.91638231e-01 -2.66407281e-01 -2.57849723e-01 7.64732301e-01
6.83898091e-01 4.24155653e-01 9.86792799e-03 -5.41324675e-01
1.30492735e+00 -4.08718437e-01 -8.61984193e-01 3.38313788e-01
7.60992467e-01 -1.13088727e+00 7.11141229e-01 2.64233202e-01
-1.00902963e+00 -2.45732978e-01 -5.53420901e-01 -2.87683070e-01
-7.48236239e-01 1.38429785e-02 4.79907662e-01 5.10861278e-01
-1.19839740e+00 6.21914208e-01 -1.44928372e+00 -8.58203530e-01
6.08929276e-01 8.05610895e-01 -9.97538149e-01 3.52131367e-01
-6.19985536e-02 9.84444559e-01 3.59044582e-01 -5.36056198e-02
-3.82421345e-01 -4.72373158e-01 -4.44274068e-01 -4.09974456e-01
-6.18389845e-02 -6.11339867e-01 1.02625477e+00 -1.48624212e-01
-1.62089455e+00 1.66826439e+00 -3.90719116e-01 -7.40817785e-02
4.86993462e-01 3.78118545e-01 1.42639488e-01 3.31295341e-01
-1.03672057e-01 7.04980552e-01 1.93372384e-01 -1.33047104e+00
-4.92908269e-01 -7.72450268e-01 -6.14530206e-01 -2.70874113e-01
2.69267887e-01 4.19954360e-01 -4.43508834e-01 -3.94217223e-01
2.85846859e-01 -8.55128586e-01 -4.70070511e-01 2.17654616e-01
-2.72270411e-01 4.25423272e-02 1.11228192e+00 -6.82911813e-01
5.78288972e-01 -1.85326326e+00 2.24734962e-01 2.36791357e-01
3.49838793e-01 1.71146572e-01 3.99458468e-01 7.15349987e-03
8.69837403e-02 3.18170696e-01 -2.67285496e-01 -4.29880828e-01
-2.49853119e-01 4.22163486e-01 4.03767198e-01 9.72036183e-01
-3.55165988e-01 6.23607695e-01 -6.94645166e-01 -9.81439412e-01
4.63184059e-01 7.30135560e-01 -1.68623120e-01 1.21224307e-01
9.69366953e-02 1.07170570e+00 -3.48211586e-01 7.13568509e-01
7.21860349e-01 -3.57437581e-01 1.77822784e-01 -2.97645688e-01
-3.93730819e-01 -4.28093463e-01 -1.13560581e+00 1.68599498e+00
1.73996627e-01 4.59268272e-01 5.47052383e-01 -1.16663694e+00
8.82392049e-01 4.47412789e-01 5.50820529e-01 8.79953653e-02
5.30779541e-01 5.01227617e-01 -2.29421064e-01 -4.55484480e-01
-2.31006090e-02 -4.36996877e-01 3.17879021e-01 6.68315411e-01
2.83609271e-01 -7.42960632e-01 5.57788014e-01 -7.78697506e-02
9.35267627e-01 1.34350240e-01 5.96219361e-01 -6.14240468e-01
7.27164447e-01 1.19965591e-01 6.12559855e-01 2.51110494e-01
-5.46276212e-01 7.27062047e-01 6.08012557e-01 -6.10081196e-01
-1.35618389e+00 -7.10618854e-01 -2.51390576e-01 4.03955102e-01
1.85296219e-02 2.02720851e-01 -1.21029639e+00 -4.88355070e-01
-1.56142071e-01 -1.40109688e-01 -6.30928993e-01 7.31587231e-01
-5.47611773e-01 -1.05124974e+00 3.77017349e-01 1.35618255e-01
4.04033139e-02 -9.61742938e-01 -6.90489769e-01 7.70517364e-02
1.20670512e-01 -1.11980963e+00 -2.12814704e-01 3.19178551e-01
-1.16191971e+00 -1.15288222e+00 -8.77672851e-01 -1.07201254e+00
1.23972523e+00 1.94364130e-01 9.16567922e-01 3.78123134e-01
-6.85433090e-01 1.50514692e-01 -6.98870001e-03 -2.37011150e-01
-5.33902109e-01 1.59560945e-02 1.27674481e-02 -4.93517786e-01
4.07626152e-01 -8.07189882e-01 -3.79863888e-01 6.65278435e-01
-9.64039266e-01 -1.33514866e-01 1.29103333e-01 5.59219420e-01
1.44651878e+00 2.56496698e-01 8.40239634e-04 -1.09712458e+00
3.68537605e-01 1.99425101e-01 -1.06625760e+00 8.97310674e-02
-6.96429759e-02 -1.96643472e-01 5.93450010e-01 -1.04904287e-01
-6.62594914e-01 5.72478950e-01 -4.81863111e-01 -4.09334302e-02
-1.00584924e+00 7.01728091e-02 -6.66632056e-02 -6.57275438e-01
3.78070086e-01 1.10579677e-01 3.68667424e-01 -6.55401289e-01
-1.47760227e-01 4.56241518e-01 6.01620555e-01 -3.67076546e-01
6.43312037e-01 1.12278235e+00 4.38681751e-01 -1.08226991e+00
-3.11118245e-01 -8.21763515e-01 -1.40496778e+00 -2.76843607e-01
1.14325988e+00 -1.19400792e-01 -9.97759461e-01 6.52391553e-01
-1.18058169e+00 -5.62581837e-01 -1.81451485e-01 4.67806906e-01
-6.46752119e-01 6.30007327e-01 -7.97381043e-01 -4.07374829e-01
-2.63649851e-01 -1.32320952e+00 1.00571692e+00 4.60847080e-01
-2.18519941e-01 -1.34645021e+00 6.15885556e-01 4.18707609e-01
2.48359725e-01 4.97943670e-01 8.65507960e-01 -5.03049076e-01
-3.42712432e-01 -2.83108264e-01 -8.50031525e-02 -8.52360278e-02
2.74331808e-01 5.83696723e-01 -8.82901967e-01 -1.29759669e-01
-1.84637737e-02 -4.20455039e-02 5.95864952e-01 7.03272521e-01
1.08206928e+00 3.47907096e-01 -7.13723242e-01 5.61064124e-01
1.57558692e+00 3.17577809e-01 5.05100787e-01 3.41347873e-01
5.30573070e-01 7.20437586e-01 5.00173271e-01 3.71356332e-03
-1.51280925e-01 5.37541270e-01 8.41054469e-02 -4.81071681e-01
1.53518945e-01 6.63096607e-01 -3.53543848e-01 6.97905183e-01
-5.70941627e-01 1.22719169e-01 -1.19567358e+00 5.00450671e-01
-1.60208130e+00 -9.02063906e-01 -4.62451428e-01 1.87905550e+00
1.13518810e+00 -1.30524576e-01 2.79806346e-01 4.28236187e-01
7.28692532e-01 -5.90346396e-01 -7.15125129e-02 -3.09548616e-01
-8.37560967e-02 2.87283182e-01 4.43649232e-01 8.78847897e-01
-1.15176916e+00 8.05226147e-01 7.30812550e+00 6.97070241e-01
-1.05576491e+00 8.58961791e-02 8.00857425e-01 2.46359959e-01
2.71868199e-01 -1.98635012e-01 -9.04984593e-01 2.32841283e-01
4.53257024e-01 7.87724480e-02 2.80324429e-01 3.47179502e-01
5.91778934e-01 -5.95386684e-01 -9.36873913e-01 8.18600535e-01
-2.24271759e-01 -1.44778013e+00 -1.32286072e-01 4.39539701e-01
4.40879524e-01 -6.38001710e-02 -3.41584742e-01 -6.61544800e-01
2.27796689e-01 -1.01667631e+00 1.92675859e-01 5.99626005e-01
6.12883210e-01 -7.43155897e-01 1.16207659e+00 3.76847565e-01
-9.46709335e-01 6.35702074e-01 -5.28412938e-01 2.62400001e-01
3.83035809e-01 9.94277000e-01 -7.99068153e-01 2.38256715e-02
7.61813223e-01 3.23898017e-01 -6.32883012e-01 1.27932620e+00
3.04686219e-01 2.72444129e-01 -5.40259063e-01 1.84206605e-01
-5.67520522e-02 -5.88279307e-01 3.60281765e-01 1.65561831e+00
2.36716792e-01 1.76001966e-01 2.14915782e-01 6.99476123e-01
1.43538803e-01 1.58722386e-01 -4.69859749e-01 -1.51925907e-01
-6.04882352e-02 1.78190887e+00 -1.96552956e+00 -2.49275804e-01
-1.31857142e-01 9.26686466e-01 2.92269021e-01 1.26415506e-01
-3.40447605e-01 -5.44522762e-01 3.42939168e-01 2.76269317e-01
2.46958196e-01 -5.21667600e-01 -2.26970062e-01 -7.42439151e-01
-5.89682221e-01 -3.65972936e-01 2.92558849e-01 -6.78667188e-01
-1.12402785e+00 3.40595275e-01 -2.58569270e-01 -6.62160933e-01
1.18566781e-01 -1.00718975e+00 -4.67559993e-01 7.00360000e-01
-1.10788393e+00 -1.07376146e+00 -2.49036238e-01 5.49934268e-01
1.63819045e-01 9.81792510e-02 1.23749912e+00 2.92816937e-01
-5.52727461e-01 -2.23477095e-01 1.41038984e-01 2.48484239e-01
6.01857066e-01 -1.71284318e+00 -8.73845723e-03 6.12166584e-01
-8.67104623e-03 7.26754725e-01 8.33542168e-01 -4.18831050e-01
-1.06354427e+00 -7.16808438e-01 1.02418089e+00 -3.47122520e-01
5.90853691e-01 -1.54922143e-01 -9.42130625e-01 5.16596913e-01
2.36962497e-01 4.82032239e-01 1.15990663e+00 -4.45838749e-01
3.76769155e-01 1.36707440e-01 -1.50026321e+00 5.46369255e-01
3.94999295e-01 -2.80578911e-01 -3.10479552e-01 5.45461655e-01
1.06877461e-03 -3.84295732e-01 -1.22883976e+00 -2.35574748e-02
3.80386353e-01 -1.00899673e+00 8.41534078e-01 -1.34029746e-01
-2.89201915e-01 -9.24745619e-01 3.03491741e-01 -1.03709543e+00
-2.04948217e-01 -7.00262725e-01 3.94376546e-01 1.01576030e+00
3.29781801e-01 -3.62455606e-01 1.03222525e+00 4.10217285e-01
4.70847040e-02 -3.27558875e-01 -1.16109312e+00 -4.80854154e-01
-2.76263487e-02 3.63787971e-02 8.03824738e-02 8.99355948e-01
2.22243398e-01 1.64459497e-01 5.40151298e-01 -1.52584702e-01
8.57157528e-01 8.76039192e-02 8.18236947e-01 -1.50688851e+00
1.51964560e-01 -5.44616818e-01 -9.04387593e-01 -5.58258891e-01
1.64534952e-02 -6.00273788e-01 5.83515540e-02 -1.61093998e+00
4.52688664e-01 -1.43424109e-01 2.38345582e-02 5.20800591e-01
1.85926780e-01 7.74963915e-01 -3.58483315e-01 2.36447185e-01
-6.47126138e-01 -2.94016153e-01 1.47198057e+00 -1.27291121e-02
-9.52492505e-02 2.40579117e-02 -6.78452849e-02 9.35795963e-01
9.27013218e-01 -6.66311324e-01 1.93031386e-01 -3.97038981e-02
-6.36409000e-02 -3.40496093e-01 2.81412274e-01 -8.19950640e-01
3.58473957e-01 -1.04402810e-01 4.98972893e-01 -7.78966248e-01
1.78129956e-01 -1.12308180e+00 2.83244550e-01 5.49637437e-01
1.70390427e-01 3.51171643e-02 2.43590042e-01 3.07788044e-01
-1.63137987e-01 -3.65535229e-01 1.36926961e+00 -6.96281135e-01
-1.46176979e-01 9.59307179e-02 -1.01504946e+00 -2.16838524e-01
1.24782491e+00 -5.28719664e-01 -2.15557531e-01 2.05086991e-01
-1.16435015e+00 -1.01951078e-01 1.29606903e+00 -6.80783033e-01
3.39441031e-01 -9.11629617e-01 -2.86986142e-01 9.09712762e-02
-3.18687648e-01 3.18985581e-01 4.57332507e-02 1.17879117e+00
-1.45036519e+00 5.37350774e-01 -5.96977651e-01 -9.39380109e-01
-1.89589810e+00 4.41491663e-01 4.85307097e-01 -3.87008548e-01
-4.53216314e-01 6.85544610e-01 -2.39801362e-01 -5.17585516e-01
-2.45406643e-01 -6.78078353e-01 -6.31436527e-01 -1.82795733e-01
5.37818551e-01 6.15968227e-01 8.55039507e-02 -8.53859425e-01
-3.91210705e-01 1.18059027e+00 -1.97207313e-02 -1.09078877e-01
1.50141466e+00 -2.11141303e-01 -8.50298524e-01 6.28593504e-01
1.01795137e+00 -8.54427144e-02 -1.01843441e+00 3.72644067e-01
6.77502379e-02 -3.49864751e-01 1.81119338e-01 -3.99855286e-01
-9.97524321e-01 9.62306082e-01 6.34465396e-01 4.29306954e-01
8.19305718e-01 5.08858599e-02 5.31226218e-01 5.24368823e-01
3.93445253e-01 -1.20131505e+00 -4.47907060e-01 4.30868864e-01
2.04131156e-01 -1.10161102e+00 2.98612088e-01 -6.11624539e-01
-1.24839365e-01 1.22143495e+00 2.30382279e-01 1.39193218e-02
8.88234317e-01 8.87982130e-01 4.30738300e-01 -3.33512545e-01
-3.78912508e-01 -1.25665054e-01 -2.14030802e-01 8.60767782e-01
7.07765341e-01 -3.45801562e-01 -4.96664524e-01 1.94891497e-01
1.88980788e-01 2.01359123e-01 6.70210183e-01 1.29944599e+00
-6.21653259e-01 -1.27902567e+00 -7.52052963e-01 1.14240602e-01
-1.14402378e+00 4.34787154e-01 -4.90065008e-01 7.55247116e-01
1.63736388e-01 5.67798078e-01 1.68835521e-01 3.22719514e-01
-3.25996755e-03 2.56808940e-02 8.36280048e-01 -6.43678069e-01
-4.01820272e-01 6.28118455e-01 -5.02470553e-01 -3.32765043e-01
-9.79350209e-01 -6.53433383e-01 -1.89050794e+00 -4.52646285e-01
-2.53009707e-01 3.17794889e-01 8.76423359e-01 9.01423514e-01
9.36597139e-02 2.48272657e-01 3.61585729e-02 -1.37089252e+00
4.20436054e-01 -7.80005038e-01 -1.05485523e+00 2.06740707e-01
1.54318020e-01 -5.13506114e-01 -3.10083777e-01 9.47068095e-01]
|
[14.399277687072754, -3.1540346145629883]
|
292af0a0-ee56-4e88-bd57-0a3051123bfa
|
adversarial-attacks-on-knowledge-graph-1
|
2111.03120
| null |
https://arxiv.org/abs/2111.03120v1
|
https://arxiv.org/pdf/2111.03120v1.pdf
|
Adversarial Attacks on Knowledge Graph Embeddings via Instance Attribution Methods
|
Despite the widespread use of Knowledge Graph Embeddings (KGE), little is known about the security vulnerabilities that might disrupt their intended behaviour. We study data poisoning attacks against KGE models for link prediction. These attacks craft adversarial additions or deletions at training time to cause model failure at test time. To select adversarial deletions, we propose to use the model-agnostic instance attribution methods from Interpretable Machine Learning, which identify the training instances that are most influential to a neural model's predictions on test instances. We use these influential triples as adversarial deletions. We further propose a heuristic method to replace one of the two entities in each influential triple to generate adversarial additions. Our experiments show that the proposed strategies outperform the state-of-art data poisoning attacks on KGE models and improve the MRR degradation due to the attacks by up to 62% over the baselines.
|
["Declan O'Sullivan", 'Luca Costabello', 'John Kelleher', 'Peru Bhardwaj']
|
2021-11-04
|
adversarial-attacks-on-knowledge-graph
|
https://aclanthology.org/2021.emnlp-main.648
|
https://aclanthology.org/2021.emnlp-main.648.pdf
|
emnlp-2021-11
|
['knowledge-graph-embeddings', 'knowledge-graph-embeddings']
|
['graphs', 'methodology']
|
[ 5.26059158e-02 6.83309138e-01 -5.20210087e-01 -1.73670538e-02
-5.01774430e-01 -7.89963365e-01 6.97070837e-01 4.58722651e-01
-1.72830030e-01 8.18362951e-01 -5.62779233e-02 -6.50544226e-01
-1.67191505e-01 -1.32142806e+00 -1.23184395e+00 -4.37845021e-01
-4.51261610e-01 6.58252358e-01 5.36200404e-01 -1.29844651e-01
1.25541225e-01 6.74575329e-01 -6.47045612e-01 1.09153792e-01
8.75361443e-01 4.29665893e-01 -8.11789215e-01 5.28900146e-01
3.64963338e-02 1.03681171e+00 -8.04421723e-01 -1.40370750e+00
3.63677144e-01 -6.67154193e-02 -8.26057673e-01 -5.45623839e-01
4.33075070e-01 -4.20469999e-01 -9.66687500e-01 1.35609591e+00
4.44380939e-01 -2.67652184e-01 5.43010712e-01 -1.98832500e+00
-1.12498319e+00 1.17065418e+00 -4.21690106e-01 3.98454130e-01
1.63729087e-01 1.86620578e-01 8.43793511e-01 -6.01254702e-01
7.34202564e-01 1.17486548e+00 7.70459652e-01 9.30289805e-01
-1.26695633e+00 -9.38301146e-01 2.18467247e-02 4.11836892e-01
-1.38900685e+00 -4.48333055e-01 9.78448629e-01 -1.55356243e-01
7.68274248e-01 4.29433018e-01 2.11046278e-01 1.51438737e+00
1.14569530e-01 5.41088998e-01 6.18548810e-01 -1.48473412e-01
2.91863173e-01 4.52317894e-01 2.96327084e-01 9.10863936e-01
8.79531026e-01 2.36931950e-01 -5.57764173e-01 -8.23818207e-01
3.01743984e-01 -2.01213703e-01 -1.58735484e-01 -3.95761102e-01
-5.97573161e-01 1.13615692e+00 8.17272127e-01 -4.01046783e-01
-3.48095655e-01 5.07031858e-01 4.94333655e-01 3.65752667e-01
4.28897440e-01 7.61548281e-01 -6.75585866e-01 4.55766380e-01
-2.12208733e-01 2.96864480e-01 8.41955721e-01 8.49266231e-01
4.52258378e-01 1.01230264e-01 -2.40384966e-01 3.03049326e-01
5.30257463e-01 3.95988971e-01 -6.00724481e-02 -3.11700225e-01
7.44727552e-01 7.88424790e-01 4.31678183e-02 -1.23083687e+00
5.65391369e-02 -4.80224431e-01 -4.81288403e-01 7.18690008e-02
1.60992131e-01 -2.30588228e-01 -9.57092106e-01 1.63887346e+00
6.31086349e-01 5.90429544e-01 3.18603039e-01 4.45914119e-01
6.75320387e-01 3.23259950e-01 8.31216872e-01 2.33905584e-01
9.70216036e-01 -6.85985506e-01 -3.84755105e-01 -9.30835754e-02
8.97783220e-01 -1.10412553e-01 5.68717182e-01 -5.68191409e-02
-7.31441438e-01 2.17722133e-01 -1.13740849e+00 4.13565397e-01
-8.58672500e-01 -5.83519459e-01 6.51818991e-01 1.10451698e+00
-4.18208748e-01 6.77502036e-01 -8.59560728e-01 -1.70178674e-02
9.49248075e-01 4.88562047e-01 -5.66105783e-01 -7.98244588e-03
-1.77098954e+00 1.03024650e+00 5.85415781e-01 -2.14297697e-01
-1.32768047e+00 -1.05946553e+00 -8.38722885e-01 4.56936397e-02
5.56139648e-01 -6.06885493e-01 6.66292429e-01 -3.89720589e-01
-8.30539286e-01 4.50154275e-01 3.06583136e-01 -1.11386824e+00
4.66385156e-01 -2.34644905e-01 -7.95932770e-01 2.47036833e-02
-2.49465257e-01 4.00356919e-01 6.12192810e-01 -1.56494069e+00
-2.07491979e-01 -2.02181160e-01 3.96726072e-01 -2.72712596e-02
-8.58332157e-01 3.99823859e-02 -2.38947973e-01 -6.46955609e-01
-4.58560616e-01 -9.04754400e-01 -3.08771551e-01 -1.51460737e-01
-1.13762212e+00 -6.22292086e-02 1.19386506e+00 -7.59215891e-01
1.58064950e+00 -1.68254137e+00 -1.86299849e-02 5.28212428e-01
4.68656361e-01 7.96938777e-01 -1.19618826e-01 5.70946276e-01
-1.49902076e-01 8.17775130e-01 -4.22275066e-02 -1.91423837e-02
1.30519852e-01 1.48428008e-01 -6.70578957e-01 4.01520997e-01
1.85565084e-01 1.09602344e+00 -9.09548461e-01 -2.41925299e-01
-4.81724367e-02 3.06374669e-01 -3.91948879e-01 1.06678806e-01
-4.92957771e-01 4.18699831e-02 -7.08168685e-01 7.02787757e-01
7.26180792e-01 1.18079230e-01 1.22368768e-01 -2.32087821e-01
7.91992724e-01 3.93701553e-01 -8.71057630e-01 6.75489485e-01
2.84402352e-02 1.66030541e-01 -4.64665294e-01 -5.34067690e-01
8.83797705e-01 3.11127841e-01 -4.91784215e-02 -1.97801456e-01
-1.56057775e-01 -2.13390328e-02 1.23367101e-01 -3.76160443e-01
8.00134540e-02 1.70806542e-01 -1.66470140e-01 3.37545484e-01
-1.14007652e-01 5.15175402e-01 -1.76163688e-01 7.32088566e-01
1.64254117e+00 -2.97652692e-01 1.64272353e-01 1.83381438e-01
4.60099041e-01 9.52964649e-02 8.68692636e-01 9.52469051e-01
-1.21399105e-01 -1.03729375e-01 8.12978864e-01 -6.94180369e-01
-1.07869065e+00 -1.20534706e+00 2.32466534e-01 7.68910825e-01
1.22500964e-01 -5.89504600e-01 -6.91054106e-01 -1.79981351e+00
4.75040138e-01 1.18428266e+00 -7.38484800e-01 -9.19904947e-01
-4.18838233e-01 -7.19302893e-01 1.28244662e+00 4.41232800e-01
5.60870886e-01 -1.02604222e+00 2.49023333e-01 4.11057435e-02
2.56234795e-01 -8.71499658e-01 -2.33146667e-01 8.83338526e-02
-5.99840224e-01 -1.49225831e+00 2.12352589e-01 -4.79160517e-01
9.56824005e-01 -2.63857037e-01 9.97642696e-01 4.69375342e-01
3.13281007e-02 7.49045834e-02 -4.43439871e-01 -5.04942596e-01
-8.85504425e-01 2.59681791e-01 2.44025692e-01 -1.91027403e-01
6.73136711e-01 -4.68179315e-01 -3.00939023e-01 2.17603385e-01
-1.02489996e+00 -3.77802700e-01 3.43345255e-01 4.34010148e-01
2.97939688e-01 3.10260952e-01 1.02430844e+00 -1.74035776e+00
7.78891981e-01 -1.00809276e+00 -4.52962875e-01 6.18739009e-01
-8.06899846e-01 2.27146253e-01 1.04225588e+00 -6.23331130e-01
-4.89296943e-01 -1.95018053e-01 3.90364863e-02 -6.89528644e-01
-1.18464231e-04 5.25028527e-01 -5.51294625e-01 -5.28475881e-01
9.73422348e-01 -9.63527337e-02 -5.06160498e-01 -3.23128045e-01
5.66458285e-01 2.91496277e-01 3.51374507e-01 -5.00877380e-01
1.53910279e+00 -1.29267141e-01 3.33207883e-02 -2.17957348e-01
-5.76518416e-01 1.55202255e-01 -2.14717060e-01 1.42106399e-01
5.47823012e-01 -5.80551744e-01 -5.73428988e-01 3.89728040e-01
-1.15992379e+00 -1.08965874e-01 3.74978967e-02 -2.74769329e-02
9.25486386e-02 3.36210787e-01 -6.08772337e-01 -6.02891386e-01
-5.57597458e-01 -8.45491529e-01 2.20533073e-01 7.24579245e-02
-2.40076762e-02 -1.07406747e+00 -1.55746145e-02 3.39023381e-01
1.35487467e-01 6.10831618e-01 1.36831570e+00 -1.43977284e+00
-6.36013448e-01 -5.61194777e-01 1.30770490e-01 1.75240174e-01
-1.19363993e-01 -1.43704412e-03 -6.76729798e-01 -2.70313084e-01
-5.71080506e-01 -3.11971039e-01 7.43381262e-01 -4.03634220e-01
1.25266612e+00 -8.61525297e-01 -8.42626929e-01 5.68178058e-01
1.41058457e+00 1.41468838e-01 9.89930391e-01 4.51628804e-01
1.12071359e+00 1.94619164e-01 2.10054994e-01 5.11848703e-02
1.76753983e-01 5.54088116e-01 9.02006984e-01 1.54482394e-01
1.85783003e-02 -8.76937509e-01 2.78580278e-01 2.81457454e-02
2.16703683e-01 -6.39913738e-01 -8.92035902e-01 6.14092708e-01
-1.63632369e+00 -9.73147273e-01 -2.56727338e-01 2.25800967e+00
8.68283927e-01 4.97842878e-01 -5.73992580e-02 1.34234041e-01
8.59714329e-01 1.80970088e-01 -7.45169401e-01 -5.63162923e-01
1.77900359e-01 1.62676826e-01 1.17284060e+00 4.85334784e-01
-1.35649562e+00 1.29938293e+00 5.98831844e+00 7.45150268e-01
-5.11835039e-01 1.83794439e-01 4.77064162e-01 7.25544244e-02
-5.69242895e-01 9.59142074e-02 -9.46684957e-01 6.19564414e-01
1.23127341e+00 -2.57696122e-01 5.36135674e-01 9.21235681e-01
-1.91089213e-01 6.67912722e-01 -1.07777131e+00 1.27890810e-01
-5.91153428e-02 -1.60388064e+00 4.35753167e-01 8.17792341e-02
7.58910358e-01 -2.08766431e-01 1.17825486e-01 4.93971705e-01
1.16196239e+00 -1.18229568e+00 2.79394120e-01 4.26248252e-01
3.96394610e-01 -1.14099896e+00 6.27822101e-01 1.86895162e-01
-6.61284864e-01 -2.60940969e-01 -4.85507190e-01 4.66436505e-01
-1.11488916e-01 5.27310073e-01 -1.35774839e+00 6.48875475e-01
3.54369342e-01 1.94996178e-01 -9.18207467e-01 9.98046458e-01
-9.26911414e-01 1.22820079e+00 -1.82754353e-01 1.11073062e-01
2.97009088e-02 3.34593773e-01 7.46275783e-01 8.56009901e-01
-1.07994035e-01 -1.80428952e-01 -1.83037911e-02 8.16688478e-01
-8.18609715e-01 -3.84640366e-01 -1.07648849e+00 -2.16934621e-01
1.21169150e+00 1.04520845e+00 -2.70597160e-01 -2.07924247e-01
-2.28466287e-01 9.63277340e-01 6.90955281e-01 3.87852252e-01
-1.11474895e+00 -4.96735424e-01 7.72207975e-01 2.16269150e-01
3.68408531e-01 7.71367848e-02 -9.06075165e-02 -7.83022225e-01
1.04025740e-03 -8.88162076e-01 6.82905316e-01 -4.19948757e-01
-1.52712727e+00 4.26278591e-01 -1.13355510e-01 -7.61221290e-01
1.31674707e-01 -2.63268977e-01 -1.09754074e+00 7.17730522e-01
-1.32406557e+00 -1.42775095e+00 3.05737615e-01 5.35611212e-01
-1.40390038e-01 -4.04140800e-01 1.11262524e+00 2.94368565e-01
-9.52454031e-01 1.19762039e+00 -3.31759080e-02 7.44269669e-01
4.38007087e-01 -1.26333642e+00 9.23663676e-01 1.14495265e+00
3.00686210e-01 8.70314181e-01 7.07431257e-01 -1.30136573e+00
-1.26084375e+00 -1.68312061e+00 9.78478432e-01 -9.28947270e-01
8.83866191e-01 -4.90764201e-01 -1.14523470e+00 1.18866837e+00
-1.42425299e-01 3.97107661e-01 7.53919661e-01 3.05028167e-02
-1.02288711e+00 -2.21426599e-02 -1.69300008e+00 7.69027352e-01
9.89465475e-01 -4.97869879e-01 -6.39082789e-01 4.17933911e-01
1.05935967e+00 -8.99638310e-02 -9.63500619e-01 3.87834579e-01
8.72078836e-02 -4.17944491e-01 1.28190768e+00 -1.42297232e+00
3.53701860e-01 -4.71638918e-01 1.17249742e-01 -1.39516878e+00
-3.59765112e-01 -5.37151396e-01 -7.96777487e-01 1.46290553e+00
9.08955157e-01 -7.74290800e-01 1.01504791e+00 8.40996027e-01
1.42607793e-01 -6.33831263e-01 -7.56329417e-01 -7.53252983e-01
2.13356569e-01 -2.69546174e-02 9.53192413e-01 1.41427827e+00
-2.00194716e-01 8.94398168e-02 -4.87381399e-01 1.01021552e+00
9.28256452e-01 -5.53927898e-01 7.22938001e-01 -1.25648677e+00
-1.85755879e-01 3.63317877e-02 -7.66557395e-01 -8.19010511e-02
3.51579040e-01 -1.08327174e+00 -5.54179609e-01 -1.37484407e+00
5.27871400e-02 -6.11447930e-01 -5.35857916e-01 9.48897898e-01
-5.30595541e-01 3.12993288e-01 8.11199695e-02 3.11429631e-02
-5.75709224e-01 3.36519271e-01 6.34238541e-01 -3.70975763e-01
5.06578125e-02 -1.07251741e-01 -1.01570511e+00 6.17510080e-01
9.19969022e-01 -1.13267446e+00 -6.51065171e-01 -3.33899528e-01
3.87990564e-01 -2.20602751e-01 5.79601884e-01 -8.33525956e-01
3.34006488e-01 -1.06718734e-01 3.86855990e-01 -2.00409487e-01
-9.96224806e-02 -8.90702665e-01 4.38474268e-01 5.92437208e-01
-6.47915900e-01 -1.43070072e-01 1.63132519e-01 1.00271416e+00
3.06246877e-01 -3.74196410e-01 5.24671614e-01 2.33923763e-01
-6.15739286e-01 6.87208712e-01 5.86924329e-02 -3.50631587e-02
1.39502251e+00 2.63150334e-01 -1.03707206e+00 -8.70967060e-02
-8.51166129e-01 3.64336133e-01 4.07496125e-01 3.95831972e-01
6.26880884e-01 -1.51370919e+00 -7.21288621e-01 4.00660336e-02
2.56663948e-01 -3.51229757e-01 -1.24342896e-01 3.83614868e-01
-3.77750993e-01 9.80738774e-02 -6.43797740e-02 3.95499349e-01
-1.36561799e+00 9.72149670e-01 3.29455942e-01 -4.40862179e-01
-4.43064362e-01 8.66517961e-01 -3.14688772e-01 -5.69150627e-01
2.95240641e-01 4.34715182e-01 -2.15650856e-01 -6.47817552e-01
4.40830499e-01 4.84068215e-01 3.62182781e-02 -2.97485322e-01
-5.87439179e-01 -1.95640609e-01 -4.73912477e-01 4.03939277e-01
1.26555884e+00 5.15693426e-01 -5.12572713e-02 -3.27941179e-01
1.06373990e+00 2.62467802e-01 -9.58320081e-01 -1.42949641e-01
1.08914569e-01 -5.79056740e-01 -3.57550234e-02 -1.31774819e+00
-1.07993865e+00 3.43486309e-01 1.53070599e-01 5.06507814e-01
6.57956719e-01 8.67688060e-02 7.65661120e-01 4.55960333e-01
3.63456190e-01 -4.46114510e-01 -2.48655617e-01 1.26786724e-01
6.33861542e-01 -9.53483522e-01 1.81856900e-02 -7.73575783e-01
-6.49264693e-01 7.23128438e-01 9.43168998e-01 -3.40375066e-01
4.22893852e-01 6.19042255e-02 -1.94089249e-01 -3.55482072e-01
-8.39663684e-01 3.30898732e-01 1.26570910e-01 1.02696872e+00
-1.82399556e-01 1.53438926e-01 -1.29159793e-01 6.00730598e-01
1.16889216e-02 -2.68305600e-01 5.66232622e-01 6.06399357e-01
-2.10701227e-01 -1.32549393e+00 -2.97513247e-01 6.76224351e-01
-6.52984083e-01 -2.65849888e-01 -9.17987227e-01 9.08521891e-01
2.70910468e-02 9.65283871e-01 -5.06241798e-01 -8.96595240e-01
4.00137126e-01 4.41473812e-01 1.63676530e-01 -4.95682538e-01
-8.09222519e-01 -8.90759110e-01 5.59320986e-01 -3.24232370e-01
2.49996036e-01 -2.05578536e-01 -1.20455003e+00 -8.05971861e-01
-5.57613432e-01 3.69262606e-01 2.66800374e-01 5.82130015e-01
5.73924601e-01 4.64900583e-01 7.19086230e-01 4.94304746e-02
-7.48174608e-01 -7.42405474e-01 -4.36098427e-01 7.20651746e-01
-1.00418240e-01 -7.29962170e-01 -5.43886483e-01 -1.67233258e-01]
|
[6.199313163757324, 7.372837543487549]
|
f21a3644-14d0-46a4-a1e7-2f56057a2dbb
|
gp-unit-generative-prior-for-versatile
|
2306.04636
| null |
https://arxiv.org/abs/2306.04636v1
|
https://arxiv.org/pdf/2306.04636v1.pdf
|
GP-UNIT: Generative Prior for Versatile Unsupervised Image-to-Image Translation
|
Recent advances in deep learning have witnessed many successful unsupervised image-to-image translation models that learn correspondences between two visual domains without paired data. However, it is still a great challenge to build robust mappings between various domains especially for those with drastic visual discrepancies. In this paper, we introduce a novel versatile framework, Generative Prior-guided UNsupervised Image-to-image Translation (GP-UNIT), that improves the quality, applicability and controllability of the existing translation models. The key idea of GP-UNIT is to distill the generative prior from pre-trained class-conditional GANs to build coarse-level cross-domain correspondences, and to apply the learned prior to adversarial translations to excavate fine-level correspondences. With the learned multi-level content correspondences, GP-UNIT is able to perform valid translations between both close domains and distant domains. For close domains, GP-UNIT can be conditioned on a parameter to determine the intensity of the content correspondences during translation, allowing users to balance between content and style consistency. For distant domains, semi-supervised learning is explored to guide GP-UNIT to discover accurate semantic correspondences that are hard to learn solely from the appearance. We validate the superiority of GP-UNIT over state-of-the-art translation models in robust, high-quality and diversified translations between various domains through extensive experiments.
|
['Chen Change Loy', 'Ziwei Liu', 'Liming Jiang', 'Shuai Yang']
|
2023-06-07
| null | null | null | null |
['unsupervised-image-to-image-translation', 'image-to-image-translation', 'image-to-image-translation']
|
['computer-vision', 'computer-vision', 'miscellaneous']
|
[ 3.13891172e-01 7.99140707e-02 -1.56643018e-01 -3.26622128e-01
-8.93115520e-01 -6.54260159e-01 6.89168155e-01 -5.65285981e-01
1.20786160e-01 7.37203956e-01 4.91788005e-03 1.19792476e-01
2.35887557e-01 -8.55074704e-01 -9.26568210e-01 -6.84322476e-01
6.88489854e-01 6.19639635e-01 1.93387493e-01 -3.44424129e-01
-1.65827483e-01 2.82275885e-01 -1.14150155e+00 2.94907451e-01
1.22074556e+00 8.24548960e-01 3.16326201e-01 1.63130820e-01
-1.50755733e-01 1.74499005e-01 -4.33866739e-01 -5.75071573e-01
5.01934707e-01 -8.76500607e-01 -5.11957943e-01 4.07999426e-01
4.42880183e-01 -2.99781799e-01 -1.34416610e-01 1.20792818e+00
4.75680619e-01 -2.27076143e-01 7.95077324e-01 -1.31710207e+00
-1.14238334e+00 1.36053875e-01 -6.45642877e-01 -3.29797119e-01
3.89832199e-01 4.61458027e-01 8.10393155e-01 -8.13726187e-01
9.41628635e-01 1.24886549e+00 5.26424885e-01 5.83365738e-01
-1.56388938e+00 -8.35470557e-01 1.95332877e-02 -8.35074782e-02
-1.23257172e+00 -3.80546749e-01 1.01192677e+00 -5.45959592e-01
5.15244901e-01 -1.01527393e-01 6.17343962e-01 1.56173813e+00
2.26609230e-01 4.88567352e-01 1.35814965e+00 -5.25310099e-01
1.11466169e-01 2.18043894e-01 -9.79529977e-01 5.78556955e-01
-1.28642973e-02 3.50171357e-01 -5.77476263e-01 1.70970276e-01
1.36937284e+00 -1.25154734e-01 -2.65894055e-01 -7.60372043e-01
-1.32267189e+00 7.40965009e-01 6.00500584e-01 2.74535239e-01
-3.12510043e-01 -1.14569888e-01 9.30651873e-02 4.29477006e-01
4.89263773e-01 5.75985432e-01 -2.45811909e-01 8.00587237e-02
-9.88014102e-01 1.89606607e-01 3.76598597e-01 1.29846501e+00
1.01466978e+00 2.20150143e-01 -4.42515790e-01 9.90801752e-01
2.07999095e-01 7.93265522e-01 4.34670836e-01 -7.13757992e-01
4.89005268e-01 6.51483834e-01 8.40710625e-02 -1.01641679e+00
2.36566901e-01 -4.43400800e-01 -9.92456675e-01 3.09868038e-01
2.67548740e-01 1.97325498e-02 -1.10371983e+00 1.86447561e+00
3.17487627e-01 -2.87962407e-02 -6.93600848e-02 8.94630730e-01
6.01141810e-01 4.67309982e-01 -5.26163466e-02 3.75114419e-02
1.06681013e+00 -9.02271867e-01 -5.71067452e-01 -4.97179985e-01
1.73304796e-01 -1.00616205e+00 1.37515366e+00 -1.57170519e-01
-1.16567409e+00 -8.51718545e-01 -9.17171896e-01 -3.39400619e-02
-1.27119586e-01 2.38061368e-01 1.57146573e-01 2.18232378e-01
-1.12709939e+00 3.52666199e-01 -6.39598131e-01 -5.04596114e-01
5.88790298e-01 2.37611607e-01 -4.21872973e-01 -2.08103418e-01
-1.26057005e+00 8.05671453e-01 2.60288239e-01 -1.93256766e-01
-1.14478862e+00 -6.32639587e-01 -9.01347339e-01 -1.98106065e-01
1.36261240e-01 -1.14181292e+00 8.41846466e-01 -1.52805364e+00
-1.91271698e+00 1.26868773e+00 5.27608804e-02 -6.55183643e-02
8.89320374e-01 5.53284846e-02 -2.00328529e-01 -1.85872391e-02
4.34686035e-01 1.06038392e+00 1.32711494e+00 -1.53295350e+00
-2.56583095e-01 -1.25121504e-01 -1.27122477e-01 3.65366876e-01
-2.15429351e-01 -2.02631056e-01 -7.11814106e-01 -9.31411564e-01
2.27554068e-02 -1.06429446e+00 3.72550683e-03 3.90056103e-01
-4.08308566e-01 2.31580734e-01 8.18498969e-01 -6.55768871e-01
5.72134137e-01 -2.16302443e+00 5.61785996e-01 2.62350552e-02
-3.18738967e-02 2.31169671e-01 -4.81689781e-01 3.14066499e-01
5.50552225e-03 -2.36124039e-01 -3.89348745e-01 -4.11010116e-01
1.21197663e-02 3.84295076e-01 -4.10366297e-01 2.91919380e-01
5.31548440e-01 1.31292021e+00 -9.23593283e-01 -4.26808685e-01
2.25440979e-01 5.44483185e-01 -6.01008236e-01 6.86779559e-01
-4.67807621e-01 1.17929506e+00 -4.63985860e-01 6.94675982e-01
7.86570728e-01 -2.37067565e-01 1.90709699e-02 -3.35557073e-01
3.53712261e-01 4.04524105e-03 -8.20460498e-01 2.10048485e+00
-5.49964726e-01 5.34730673e-01 -8.48387256e-02 -9.68963146e-01
1.28810787e+00 8.99427086e-02 3.16546053e-01 -9.36929464e-01
1.76704943e-01 2.88247794e-01 -3.20187509e-01 -2.78933674e-01
1.07948624e-01 -4.57275003e-01 -1.04116552e-01 2.20232412e-01
1.80361465e-01 -7.17915058e-01 -1.68403059e-01 -5.45098148e-02
6.47335589e-01 6.35613024e-01 5.46210110e-02 -5.00487238e-02
5.45291603e-01 -1.36899337e-01 5.65147877e-01 3.20907086e-01
7.35657364e-02 1.15202451e+00 2.59780645e-01 -1.77577451e-01
-1.39134443e+00 -1.37564242e+00 7.45311230e-02 8.04421484e-01
3.14498782e-01 7.33892769e-02 -8.74017477e-01 -7.44368076e-01
-2.15121284e-01 4.38948095e-01 -5.94638884e-01 -3.55633438e-01
-3.70812744e-01 -3.14371884e-01 3.83579433e-01 3.76618356e-01
9.80613589e-01 -1.09461534e+00 -8.98714662e-02 1.83629319e-01
-4.47267205e-01 -1.40670502e+00 -6.69801235e-01 -1.50707841e-01
-8.13654184e-01 -7.13152170e-01 -1.05773032e+00 -1.07103527e+00
1.02641034e+00 -3.66505049e-02 1.29535151e+00 -1.62965387e-01
-9.01653469e-02 1.51100919e-01 -3.51771295e-01 -6.24369793e-02
-7.49987006e-01 2.34006792e-02 -1.34910315e-01 1.93570137e-01
2.14035995e-02 -9.12010849e-01 -6.13466620e-01 7.23040760e-01
-1.07211030e+00 3.48088503e-01 7.43183017e-01 1.14869416e+00
8.62769544e-01 -1.94090545e-01 4.06782597e-01 -7.17123508e-01
5.09808898e-01 -2.02320859e-01 -5.85544825e-01 3.11954260e-01
-4.58557546e-01 1.20969377e-01 6.87931359e-01 -8.28455389e-01
-1.03034258e+00 1.22530796e-01 1.83565095e-02 -8.42144608e-01
-9.47165862e-02 1.17700368e-01 -6.35690331e-01 -1.78289294e-01
7.31010854e-01 5.59729815e-01 1.96921509e-02 -2.18898207e-01
6.94515347e-01 3.97153974e-01 8.09682012e-01 -9.60941970e-01
1.22984612e+00 5.15020430e-01 -3.07333142e-01 -2.86452532e-01
-7.32203424e-01 8.97265598e-02 -9.19891238e-01 -3.03472839e-02
1.06259441e+00 -1.11155629e+00 1.74444705e-01 5.23048699e-01
-1.08173776e+00 -7.19730496e-01 -4.80118811e-01 5.43003157e-02
-9.09844458e-01 1.19665034e-01 -3.38980287e-01 -9.06235054e-02
-2.18742803e-01 -1.25309527e+00 1.42784369e+00 1.54133767e-01
-1.93781912e-01 -1.18843102e+00 1.85661018e-01 4.20983255e-01
5.03120780e-01 4.15786207e-01 8.11990261e-01 -8.27460736e-02
-6.76880658e-01 2.06957757e-03 -3.58462453e-01 5.52460194e-01
5.18593073e-01 -7.31964186e-02 -6.61345065e-01 -4.96767730e-01
-1.46536991e-01 -4.91460234e-01 4.44692850e-01 2.64878899e-01
9.15714502e-01 -2.22508401e-01 -3.61706793e-01 1.03121233e+00
1.31095970e+00 1.36900041e-03 1.01678824e+00 3.67958963e-01
8.51528764e-01 3.45780820e-01 6.69080138e-01 5.83827533e-02
2.12905571e-01 9.48993027e-01 2.35624567e-01 -5.31705499e-01
-4.82613802e-01 -7.97052085e-01 4.64990139e-01 5.76861322e-01
6.58524036e-02 -5.14508039e-02 -5.80114365e-01 5.89376926e-01
-1.70693386e+00 -6.75534308e-01 4.81601268e-01 2.09298968e+00
1.16261256e+00 1.45342097e-01 -7.15908557e-02 -3.99962097e-01
7.70315409e-01 1.09724728e-02 -5.79748273e-01 -1.18117839e-01
-1.70579672e-01 2.74692535e-01 1.64883509e-01 3.77289653e-01
-9.10820127e-01 1.34355950e+00 5.81191635e+00 1.00725448e+00
-1.33321500e+00 3.21337074e-01 6.61322892e-01 1.77920401e-01
-5.91553688e-01 1.73681695e-02 -4.88562524e-01 5.97315609e-01
2.25031868e-01 1.24791943e-01 4.97265697e-01 7.12791145e-01
8.18225518e-02 2.52468854e-01 -1.21086884e+00 9.72208798e-01
4.62905355e-02 -1.28449893e+00 3.89447182e-01 3.15189064e-02
1.41748714e+00 -1.77376494e-01 3.87928247e-01 1.85299382e-01
5.28067589e-01 -1.06087518e+00 6.73477948e-01 3.67973864e-01
1.37817287e+00 -5.63464582e-01 4.82862294e-01 1.91100657e-01
-9.80260849e-01 4.21695530e-01 -4.76899773e-01 3.52840424e-01
1.84858322e-01 5.04843652e-01 -6.58111393e-01 7.41311193e-01
5.55704117e-01 9.29573476e-01 -3.75078321e-01 4.78967279e-01
-6.71729088e-01 1.88347057e-01 -1.29386947e-01 5.57898700e-01
2.45392933e-01 -6.06302202e-01 4.81062055e-01 9.52129900e-01
5.44514418e-01 -1.96465194e-01 2.74698496e-01 1.45328295e+00
-2.00308442e-01 -9.62604061e-02 -7.84526408e-01 5.00581153e-02
4.67504889e-01 1.02639949e+00 -5.24628401e-01 -2.33812809e-01
-2.15272337e-01 1.60336149e+00 3.80805284e-01 5.73694646e-01
-1.00963187e+00 4.82932925e-02 7.89463699e-01 4.45872933e-01
3.92110884e-01 -2.00895667e-01 -2.92508632e-01 -1.24967039e+00
5.99326380e-02 -1.16112995e+00 -2.76611727e-02 -1.00525725e+00
-1.43028998e+00 8.32212567e-01 -1.07143231e-01 -1.68305838e+00
-2.41560772e-01 -2.90904284e-01 -6.87350988e-01 1.07579041e+00
-1.45090938e+00 -1.79160154e+00 -4.93352324e-01 9.21727479e-01
5.32850385e-01 -3.96499127e-01 7.15876341e-01 1.80757791e-01
-2.94677407e-01 8.93043697e-01 1.37466565e-01 2.44534656e-01
1.16140580e+00 -9.10485148e-01 4.43289638e-01 8.68796110e-01
1.91600069e-01 3.90792519e-01 5.96914530e-01 -6.93129241e-01
-1.05300879e+00 -1.33873475e+00 4.57835764e-01 -5.29385805e-01
3.84479403e-01 -4.02805448e-01 -8.58268857e-01 7.19208598e-01
3.18517238e-01 1.87451363e-01 3.18859547e-01 -3.49670380e-01
-5.41861236e-01 -2.09455416e-01 -1.21059728e+00 6.25982106e-01
1.11611736e+00 -6.91814661e-01 -4.33607817e-01 3.42078149e-01
6.04502916e-01 -6.15966201e-01 -8.34114611e-01 2.89452940e-01
3.11463445e-01 -9.44262087e-01 1.10930479e+00 -3.27583253e-01
8.06377590e-01 -4.00152594e-01 2.46250592e-02 -1.70073557e+00
-3.62337172e-01 -6.97005928e-01 3.59675169e-01 1.41576636e+00
2.93425769e-01 -6.00253403e-01 6.87931240e-01 2.29023114e-01
4.01257873e-02 -5.28549314e-01 -7.39475429e-01 -8.08253586e-01
2.86613524e-01 9.84810218e-02 6.56567574e-01 1.17555666e+00
-4.73034590e-01 3.93203467e-01 -5.82345784e-01 1.18381865e-01
6.19920075e-01 3.78639042e-01 1.14780402e+00 -8.18644047e-01
-5.47352135e-01 -3.59717399e-01 -4.29754913e-01 -1.27534389e+00
2.75687009e-01 -8.66935790e-01 1.06656596e-01 -1.54851198e+00
1.90539032e-01 -5.56514859e-01 9.07978881e-03 6.13767743e-01
-1.87933534e-01 5.40879965e-01 5.85034750e-02 4.18967426e-01
-2.76179194e-01 9.40610230e-01 2.03357387e+00 -3.01412910e-01
-1.82113558e-01 -2.31312469e-01 -6.81302786e-01 4.05111492e-01
5.30955493e-01 -4.90826994e-01 -6.46276295e-01 -6.63840353e-01
-3.32018584e-02 -7.26358593e-02 4.09535885e-01 -8.71766508e-01
-9.29729789e-02 -4.07705247e-01 6.07015014e-01 -1.49828836e-01
2.32464224e-01 -8.00296187e-01 4.48501855e-01 1.67586833e-01
-2.66258180e-01 -2.09642142e-01 1.28385976e-01 4.76425350e-01
-4.67628241e-01 2.52896905e-01 1.12214458e+00 -7.23402649e-02
-5.19859254e-01 5.06396472e-01 3.21588188e-01 2.84427702e-01
1.01123667e+00 -3.45206439e-01 9.82058644e-02 -5.63218236e-01
-6.90994918e-01 -1.36841163e-02 9.96209741e-01 5.97024858e-01
6.22199416e-01 -1.76121223e+00 -8.58414829e-01 5.90736449e-01
3.59021425e-01 2.68912762e-01 1.01475462e-01 6.04756236e-01
-4.03212070e-01 5.26571609e-02 -7.03534186e-01 -1.00873363e+00
-9.83931065e-01 4.93065119e-01 3.99785846e-01 -2.81761765e-01
-5.78729987e-01 8.64901364e-01 6.62472486e-01 -6.50525153e-01
-1.32113099e-01 -1.29205540e-01 4.22290415e-01 -3.50599557e-01
1.23671226e-01 -3.26785296e-01 -1.47842050e-01 -6.34615302e-01
-1.52145416e-01 9.77205157e-01 6.13428429e-02 -1.70362115e-01
1.15257704e+00 -2.05500931e-01 4.30023186e-02 -3.11452076e-02
1.18379736e+00 -2.07675070e-01 -1.92880070e+00 -4.31531399e-01
-6.43868864e-01 -7.19368219e-01 -2.00316578e-01 -9.02162969e-01
-1.29446197e+00 8.81059170e-01 6.78199112e-01 -4.27814245e-01
1.37485647e+00 5.43164909e-02 8.74986291e-01 -2.14060053e-01
5.60590506e-01 -8.62636089e-01 5.83205938e-01 2.61293083e-01
1.12728906e+00 -1.29891872e+00 -1.60904244e-01 -4.67631906e-01
-8.30343604e-01 8.46476912e-01 8.07392418e-01 -1.65484935e-01
2.66335487e-01 2.09699243e-01 2.24815831e-01 3.84137928e-02
-2.19120622e-01 -3.52702364e-02 5.65659940e-01 1.00855184e+00
1.45982131e-01 -2.64963135e-02 2.37822458e-02 1.93760455e-01
-2.12452665e-01 6.48479313e-02 1.17547717e-03 5.63783169e-01
-2.00124960e-02 -1.52218378e+00 -3.71037096e-01 -1.58927932e-01
2.24452037e-02 -1.89808756e-02 -5.06669521e-01 7.47205734e-01
3.50887358e-01 5.94473183e-01 6.50122166e-02 -4.18878496e-01
4.01959687e-01 -1.93801537e-01 7.75766492e-01 -6.22669876e-01
-2.73225099e-01 3.20303142e-01 -2.67805099e-01 -5.97981572e-01
-4.13400143e-01 -4.52785820e-01 -8.61900330e-01 -1.59218058e-01
-4.89696749e-02 -2.13516563e-01 3.81406963e-01 9.39463913e-01
4.95099694e-01 4.69940245e-01 7.07005143e-01 -9.48552728e-01
-4.76430833e-01 -8.42561781e-01 -2.94639975e-01 8.18730295e-01
1.15246335e-02 -8.51392090e-01 -1.27070293e-01 4.23483312e-01]
|
[11.728214263916016, -0.393639475107193]
|
fc7c575c-0863-42c4-b925-5793fd9faf05
|
an-evaluation-of-log-parsing-with-chatgpt
|
2306.01590
| null |
https://arxiv.org/abs/2306.01590v1
|
https://arxiv.org/pdf/2306.01590v1.pdf
|
An Evaluation of Log Parsing with ChatGPT
|
Software logs play an essential role in ensuring the reliability and maintainability of large-scale software systems, as they are often the sole source of runtime information. Log parsing, which converts raw log messages into structured data, is an important initial step towards downstream log analytics. In recent studies, ChatGPT, the current cutting-edge large language model (LLM), has been widely applied to a wide range of software engineering tasks. However, its performance in automated log parsing remains unclear. In this paper, we evaluate ChatGPT's ability to undertake log parsing by addressing two research questions. (1) Can ChatGPT effectively parse logs? (2) How does ChatGPT perform with different prompting methods? Our results show that ChatGPT can achieve promising results for log parsing with appropriate prompts, especially with few-shot prompting. Based on our findings, we outline several challenges and opportunities for ChatGPT-based log parsing.
|
['Hongyu Zhang', 'Van-Hoang Le']
|
2023-06-02
| null | null | null | null |
['log-parsing']
|
['computer-code']
|
[ 4.33990099e-02 2.40513142e-02 -3.09516072e-01 -2.47513637e-01
-1.07801723e+00 -6.84502304e-01 3.17420542e-01 6.26099646e-01
8.33486021e-02 1.14001736e-01 2.09201559e-01 -1.06624532e+00
1.98504031e-01 -5.82835257e-01 -3.91487479e-01 3.23740095e-01
-4.19696122e-01 3.25105727e-01 7.07882941e-01 -7.28250667e-02
4.45729613e-01 8.54384378e-02 -1.30421937e+00 5.35760343e-01
7.27636099e-01 5.45211136e-01 2.61440933e-01 9.77522850e-01
-6.42626703e-01 1.65045452e+00 -8.91737938e-01 -2.44050585e-02
-4.40799110e-02 -6.43737376e-01 -1.32685065e+00 -2.06256315e-01
-1.61553919e-01 -4.84923095e-01 -2.02042311e-02 7.62445152e-01
1.99618507e-02 -3.05271029e-01 -2.41751060e-01 -1.32261932e+00
-1.77933440e-01 1.15902972e+00 -5.90508938e-01 3.84495676e-01
7.47047007e-01 1.07095070e-01 9.82567191e-01 -4.79991257e-01
5.40365100e-01 1.07326484e+00 5.05358338e-01 1.14263475e-01
-1.10643172e+00 -5.46160638e-01 -1.36959925e-01 1.87989160e-01
-1.01321971e+00 -5.57143092e-01 5.19846261e-01 -5.19442737e-01
1.75825822e+00 3.34995627e-01 3.14940363e-01 3.68103802e-01
6.75070405e-01 5.91940761e-01 1.00833368e+00 -8.20915818e-01
1.59254640e-01 -8.67574662e-02 7.21549034e-01 6.48894191e-01
1.07131198e-01 -3.30319375e-01 -1.03181016e+00 -6.38918698e-01
3.49290043e-01 -1.56413391e-01 1.31569758e-01 4.80986655e-01
-8.52627933e-01 6.58529758e-01 -4.78890061e-01 4.74436998e-01
1.58016071e-01 3.54558378e-01 5.68046868e-01 8.08768392e-01
4.50532585e-01 4.01120156e-01 -3.86264950e-01 -1.09435999e+00
-6.31997347e-01 -2.44617656e-01 1.59190857e+00 1.36851430e+00
8.34966719e-01 -1.69625208e-01 1.06066346e-01 4.40432817e-01
6.27795339e-01 1.84060872e-01 2.41561413e-01 -8.98097098e-01
7.16884851e-01 1.02176702e+00 -1.93155870e-01 -5.28498948e-01
-1.71243012e-01 4.08411324e-01 -9.30421278e-02 1.54196620e-01
3.80112708e-01 1.24286748e-01 -2.37449437e-01 1.17668438e+00
-1.59869585e-02 5.28923422e-03 -2.83712894e-01 9.00591612e-02
4.26025748e-01 5.64633489e-01 1.48983181e-01 -3.93096894e-01
1.49405050e+00 -6.78263009e-01 -5.84070206e-01 -8.17474365e-01
1.08120859e+00 -1.09557056e+00 1.32167017e+00 2.44470149e-01
-8.44842434e-01 -1.05438359e-01 -8.81923795e-01 -2.20276229e-02
8.64578262e-02 4.42527886e-03 9.30071950e-01 7.34105170e-01
-1.29264414e+00 5.10238945e-01 -1.58231044e+00 -5.58192015e-01
-7.58870989e-02 1.67523846e-01 -1.65077135e-01 -8.73048231e-02
-5.49375892e-01 5.05033851e-01 2.81816304e-01 -2.44966358e-01
-6.74812198e-01 -5.48621714e-01 -8.72754395e-01 3.40448439e-01
7.84638882e-01 -2.21050624e-02 1.99023223e+00 8.68521258e-02
-1.44153535e+00 5.35262764e-01 -4.43256021e-01 -2.91178435e-01
1.44649118e-01 -5.39364159e-01 -1.92991018e-01 -1.50553584e-01
2.22744316e-01 -4.07043070e-01 3.63921255e-01 -5.30931175e-01
-6.60271108e-01 -1.72930241e-01 -9.31157917e-03 -3.73296171e-01
-4.13259894e-01 1.03787124e+00 -5.09092450e-01 2.55508963e-02
-2.87185490e-01 -6.77051425e-01 -1.22623980e-01 -6.11825645e-01
-2.28843212e-01 -4.82801378e-01 8.51772785e-01 -6.90641284e-01
2.03869557e+00 -2.14555144e+00 -5.41280329e-01 -1.63193196e-01
5.43356180e-01 1.05759114e-01 -7.10705295e-02 1.11172390e+00
3.05478990e-01 4.97779727e-01 2.45586745e-02 -3.88777733e-01
6.78271204e-02 5.19031323e-02 -4.16297495e-01 -3.83298248e-02
2.94436306e-01 1.14333510e+00 -8.70988131e-01 -8.02601516e-01
8.44800100e-02 -2.16329917e-01 -4.31568563e-01 6.63751304e-01
-4.41230625e-01 1.32487059e-01 -5.35532951e-01 6.62016571e-01
2.38389671e-01 -8.17383468e-01 5.73950469e-01 6.71541035e-01
-5.25557280e-01 8.19407463e-01 -8.12897801e-01 1.37763011e+00
-5.55830598e-01 8.56637657e-01 1.88650623e-01 -3.04109454e-01
8.36446285e-01 2.92870283e-01 2.53137559e-01 -7.20263064e-01
-3.54403518e-02 1.88247755e-01 -7.42886066e-02 -7.04410434e-01
4.74507272e-01 2.09890664e-01 -6.03832364e-01 1.29270864e+00
-2.11966991e-01 4.57081757e-02 4.82200712e-01 5.30748010e-01
2.07656193e+00 -8.50302503e-02 7.04722226e-01 1.23781301e-01
2.11617723e-01 4.58171278e-01 5.07758439e-01 6.32887065e-01
-1.29468441e-01 2.46555626e-01 1.20094669e+00 -4.92091253e-02
-6.50772393e-01 -5.76129079e-01 4.57984269e-01 1.43483543e+00
-1.08290412e-01 -1.10733044e+00 -7.99281478e-01 -8.40564907e-01
-2.92293191e-01 8.51317108e-01 1.00145796e-02 -1.07570022e-01
-1.12183022e+00 -5.46483517e-01 6.74240649e-01 7.86242843e-01
1.46760553e-01 -1.38853848e+00 -7.65580237e-01 6.66936874e-01
-2.97997117e-01 -1.35376489e+00 -7.84552932e-01 3.03962678e-01
-8.28503132e-01 -1.48830163e+00 3.65860254e-01 -8.36138785e-01
4.23177481e-01 4.72211629e-01 1.51548040e+00 5.44893801e-01
-3.06971103e-01 4.09067124e-01 -7.94966102e-01 -3.68777901e-01
-1.31559086e+00 1.98272139e-01 -4.93435979e-01 -6.36095703e-01
8.12199295e-01 -7.38646984e-01 -6.59748539e-02 3.93152595e-01
-8.02943289e-01 3.35638896e-02 7.25984573e-01 3.02285403e-01
9.13258418e-02 1.13174386e-01 5.29205799e-01 -1.48384535e+00
1.15953040e+00 -4.64858472e-01 -8.12202752e-01 3.96274656e-01
-9.77279007e-01 -4.51490097e-02 9.91824746e-01 -2.93095976e-01
-1.04841506e+00 -3.09028924e-01 -3.19765776e-01 2.31867939e-01
-1.74257517e-01 8.78131807e-01 -5.12158237e-02 -8.57673958e-02
5.77664196e-01 1.50569215e-01 4.37181480e-02 -6.88540161e-01
-9.77882072e-02 8.89947951e-01 2.15448156e-01 -7.02332795e-01
6.10563993e-01 1.11418121e-01 -6.12222314e-01 -6.00103319e-01
-3.66040230e-01 -8.75072002e-01 -3.92701715e-01 1.84588626e-01
5.00208676e-01 -3.36071223e-01 -8.14790726e-01 5.14543533e-01
-1.28736913e+00 -8.27702045e-01 -1.42613515e-01 9.54969153e-02
-4.21135545e-01 7.25453019e-01 -1.20546770e+00 -8.92426610e-01
-5.82680285e-01 -1.05538690e+00 1.12283933e+00 1.43688440e-01
-6.38624668e-01 -8.23978305e-01 3.42333138e-01 3.23682547e-01
6.37829363e-01 -1.35143474e-01 1.33388042e+00 -8.59685659e-01
-1.07739961e+00 -5.07366717e-01 -1.44930214e-01 1.00294739e-01
7.38693029e-02 4.98970412e-02 -6.37644470e-01 -3.26296657e-01
1.28008187e-01 -1.49941698e-01 -4.64809351e-02 -3.93468052e-01
6.21822596e-01 -2.57362694e-01 -2.37402886e-01 2.46313795e-01
1.26854849e+00 4.16572690e-01 4.52641517e-01 2.60591716e-01
4.86084849e-01 2.88112074e-01 8.59559655e-01 5.51719010e-01
3.59321326e-01 1.34687319e-01 1.21989928e-01 5.96949160e-01
-1.38159394e-01 -5.80838501e-01 8.51440966e-01 1.55641699e+00
4.47094440e-01 -2.57668942e-01 -1.30095136e+00 5.08347452e-01
-1.88485682e+00 -4.37402189e-01 -5.10130644e-01 2.01575685e+00
6.98044062e-01 4.23399478e-01 -1.35008335e-01 2.27045845e-02
2.17260376e-01 6.32463843e-02 -3.22660297e-01 -3.48424345e-01
6.07322872e-01 2.36562192e-01 1.68960646e-01 2.21279562e-01
-4.34276640e-01 9.60102439e-01 6.70044804e+00 5.71769476e-01
-8.40950131e-01 2.94684827e-01 -1.47117719e-01 4.42195535e-01
-1.83471799e-01 8.36672843e-01 -1.06967103e+00 4.29656476e-01
1.65017056e+00 -8.85320902e-01 4.86887693e-01 1.11379862e+00
4.01921600e-01 -5.15537500e-01 -1.47597873e+00 5.67116976e-01
-2.69307643e-01 -1.32393706e+00 -5.29179454e-01 2.22667590e-01
5.58602251e-02 4.22294848e-02 -6.92845523e-01 6.66025162e-01
6.44745171e-01 -9.83951986e-01 4.54788208e-01 -1.43410027e-01
9.09172356e-01 -2.68655479e-01 7.60913849e-01 7.76967585e-01
-1.76631892e+00 3.23187932e-02 2.73199379e-02 -3.78492057e-01
5.82609892e-01 3.14030051e-01 -1.41159511e+00 3.79598022e-01
7.58773804e-01 5.15139043e-01 -7.29748666e-01 9.22924519e-01
-3.59526008e-01 1.30332327e+00 -4.23024416e-01 -1.33549690e-01
-4.02697474e-01 3.48124094e-02 2.13199198e-01 1.33492637e+00
3.00803870e-01 -1.19960040e-01 6.37939751e-01 6.94167435e-01
1.95583515e-02 1.40873874e-02 -3.34408402e-01 -7.43682086e-01
8.99887621e-01 1.14545488e+00 -1.11700416e+00 -2.05842450e-01
-8.22526217e-01 4.58927691e-01 2.74428070e-01 8.26499909e-02
-4.34468538e-01 -6.77279353e-01 5.98395228e-01 5.73198736e-01
-8.02488327e-02 -6.48313224e-01 -2.76010454e-01 -9.17998493e-01
5.00523031e-01 -9.58269656e-01 5.18224835e-01 -4.59613591e-01
-9.47624624e-01 6.07390523e-01 9.07220468e-02 -1.06222737e+00
-7.05910087e-01 -3.45684588e-01 -1.06274247e+00 7.27501988e-01
-1.27176869e+00 -9.56462204e-01 -4.41237450e-01 1.33493528e-01
6.09010994e-01 2.11535424e-01 8.78911674e-01 3.45703363e-01
-3.08157593e-01 4.67956930e-01 -3.13368082e-01 3.31283808e-01
7.07393646e-01 -1.29712427e+00 1.18623221e+00 1.14079237e+00
-4.06820886e-03 1.03671575e+00 5.81763744e-01 -8.89879227e-01
-1.98977506e+00 -9.24849391e-01 1.03944182e+00 -8.97119284e-01
1.17373002e+00 -5.65649509e-01 -1.27792132e+00 1.11704302e+00
1.31941646e-01 -2.58179992e-01 7.20663369e-01 8.98100138e-02
-2.86353916e-01 2.54849553e-01 -6.09439075e-01 1.75801143e-01
9.54849303e-01 -9.16619182e-01 -4.01890695e-01 2.43241802e-01
1.11048448e+00 -4.96317744e-01 -1.16161311e+00 -2.10104883e-01
2.44530514e-01 -1.12430036e+00 3.41963261e-01 -5.92022121e-01
3.51136595e-01 -2.19372839e-01 1.85238570e-01 -7.91058183e-01
5.39876781e-02 -1.42116296e+00 -5.16443610e-01 1.56861699e+00
2.13604018e-01 -9.17700291e-01 8.46244633e-01 6.86536908e-01
-2.72967279e-01 -4.45489258e-01 -5.03389716e-01 -9.16836143e-01
-2.87363887e-01 -9.69957948e-01 6.21189594e-01 4.61025327e-01
7.25952387e-01 5.74511945e-01 -1.30683094e-01 -2.73261592e-02
2.93131530e-01 3.10605377e-01 1.03306973e+00 -1.33291638e+00
-6.67056382e-01 -3.62958796e-02 -8.18741843e-02 -1.27558732e+00
9.61088538e-02 -7.02093959e-01 2.90324211e-01 -1.60248363e+00
3.81719232e-01 -3.48106503e-01 2.88209379e-01 7.47514307e-01
-1.31212503e-01 -4.39111948e-01 2.95454800e-01 4.87590343e-01
-8.44581306e-01 -1.56644449e-01 5.35331905e-01 3.19320530e-01
-5.66660583e-01 4.21320915e-01 -8.82338285e-01 5.00479281e-01
6.38146758e-01 -8.92138779e-01 -3.93954396e-01 -3.00507337e-01
5.17840564e-01 7.47478127e-01 -2.10488066e-01 -6.20783448e-01
5.90831339e-01 -3.73862237e-01 -6.05186522e-01 -2.31275827e-01
-2.15916172e-01 -4.90414351e-01 1.47944823e-01 2.44759738e-01
-3.73917371e-02 5.90491414e-01 2.06505299e-01 5.96863329e-01
-5.11904895e-01 -3.47307801e-01 2.38378599e-01 -1.62909031e-01
-1.01249182e+00 6.97704926e-02 -9.06378746e-01 5.71615934e-01
1.02623677e+00 -1.00999326e-01 -7.01771140e-01 -2.96242535e-01
-1.95942625e-01 4.11547273e-01 9.22252417e-01 4.85130578e-01
4.28815484e-01 -6.49797440e-01 -4.81813550e-01 3.24361026e-01
3.02394599e-01 -4.89048883e-02 -2.20467091e-01 9.48472500e-01
-6.04666233e-01 4.62550551e-01 3.07587206e-01 -5.58921456e-01
-1.69762671e+00 2.85373867e-01 -2.15359062e-01 -8.55215371e-01
-8.79815340e-01 4.06622082e-01 -1.88989341e-01 -2.11954668e-01
-2.32850984e-02 -4.37949479e-01 1.90303162e-01 -3.09993774e-01
8.93486977e-01 4.57536578e-01 5.08634567e-01 1.81986272e-01
-4.73727584e-01 1.48151815e-01 -3.22627813e-01 4.62187007e-02
1.40296423e+00 -4.97345030e-02 -7.40246177e-01 6.46989822e-01
1.10886896e+00 2.99052835e-01 -8.25948775e-01 -1.14009894e-01
9.06300366e-01 -4.57181007e-01 -4.05567825e-01 -3.57738227e-01
-3.87867630e-01 8.42624545e-01 -1.91008821e-01 7.60220706e-01
1.03519309e+00 5.03874362e-01 1.07023263e+00 2.31757537e-01
1.19539630e+00 -7.42953122e-01 1.12733193e-01 7.28351295e-01
2.58868247e-01 -1.11358654e+00 -2.56632954e-01 -7.45317280e-01
-2.74571419e-01 1.30259812e+00 7.78446794e-01 3.30704361e-01
4.88163471e-01 1.01495230e+00 1.58335373e-01 -2.85948992e-01
-1.18142533e+00 1.60476267e-01 -3.88237208e-01 3.78835499e-01
8.31621945e-01 -1.02566004e-01 -2.76341796e-01 6.56437695e-01
-1.37146205e-01 2.42370367e-01 1.06678784e+00 1.93183327e+00
-6.51498556e-01 -1.65443802e+00 -3.64542723e-01 6.02820337e-01
-5.31878114e-01 -1.85352758e-01 -5.37680328e-01 7.80236363e-01
-7.82549262e-01 1.32091725e+00 -2.15339184e-01 -6.87707365e-01
3.27522755e-01 2.70442635e-01 9.07146856e-02 -1.41342318e+00
-8.84492874e-01 -8.50117430e-02 4.03279305e-01 -9.16371346e-01
4.24333066e-01 -5.36979139e-01 -1.50587904e+00 -6.33267105e-01
-3.63552421e-01 5.60226917e-01 3.24641198e-01 9.27900136e-01
6.49785221e-01 5.50282121e-01 3.68481249e-01 -1.91371471e-01
-8.79481912e-01 -9.54746842e-01 -3.73567969e-01 4.27429304e-02
1.00764938e-01 4.97481190e-02 -3.03858936e-01 4.50717032e-01]
|
[7.941205978393555, 7.009542465209961]
|
32398f74-f56b-4b24-8a34-f8c13af1affe
|
the-conditional-cauchy-schwarz-divergence
|
2301.08970
| null |
https://arxiv.org/abs/2301.08970v1
|
https://arxiv.org/pdf/2301.08970v1.pdf
|
The Conditional Cauchy-Schwarz Divergence with Applications to Time-Series Data and Sequential Decision Making
|
The Cauchy-Schwarz (CS) divergence was developed by Pr\'{i}ncipe et al. in 2000. In this paper, we extend the classic CS divergence to quantify the closeness between two conditional distributions and show that the developed conditional CS divergence can be simply estimated by a kernel density estimator from given samples. We illustrate the advantages (e.g., the rigorous faithfulness guarantee, the lower computational complexity, the higher statistical power, and the much more flexibility in a wide range of applications) of our conditional CS divergence over previous proposals, such as the conditional KL divergence and the conditional maximum mean discrepancy. We also demonstrate the compelling performance of conditional CS divergence in two machine learning tasks related to time series data and sequential inference, namely the time series clustering and the uncertainty-guided exploration for sequential decision making.
|
['José C. Príncipe', 'Robert Jenssen', 'Sigurd Løkse', 'Hongming Li', 'Shujian Yu']
|
2023-01-21
| null | null | null | null |
['time-series-clustering']
|
['time-series']
|
[-7.16443658e-02 -1.03970289e-01 -2.87309382e-02 -4.94991720e-01
-9.28354204e-01 -5.15854359e-01 4.87864941e-01 1.45083547e-01
-3.78756613e-01 1.09608734e+00 -3.61865819e-01 -3.28363866e-01
-5.80284595e-01 -4.20444995e-01 -4.27669287e-01 -9.36836898e-01
-6.11083865e-01 4.36432719e-01 1.51557073e-01 1.91310644e-01
4.87696320e-01 2.86702752e-01 -1.43739033e+00 -5.17539144e-01
1.34126425e+00 1.50778329e+00 1.43757641e-01 5.58677018e-01
2.49557361e-01 3.76771867e-01 -3.82072955e-01 -2.59458840e-01
6.55966923e-02 -6.06332481e-01 -5.32131672e-01 -2.79515445e-01
-2.40229324e-01 -1.18239507e-01 1.32165268e-01 1.36689818e+00
2.14248776e-01 4.47642863e-01 9.58034992e-01 -1.54613817e+00
-4.82224852e-01 4.15369093e-01 -7.42982328e-01 2.66904175e-01
2.49831274e-01 -2.16081962e-01 1.02512705e+00 -8.22845638e-01
2.71186501e-01 9.68974829e-01 9.28092062e-01 1.93702310e-01
-1.52673972e+00 -5.91736853e-01 -9.95889232e-02 3.13279442e-02
-1.59238446e+00 -2.20879823e-01 4.44041610e-01 -6.87393486e-01
3.67658973e-01 2.58824348e-01 4.20537740e-01 8.61812532e-01
2.41132393e-01 1.00280905e+00 1.23004794e+00 -4.83678788e-01
8.54369581e-01 1.02247253e-01 -2.31915563e-02 4.20335889e-01
1.19064316e-01 3.78954530e-01 -2.20649168e-01 -6.31551147e-01
8.03891659e-01 4.95628007e-02 -2.84942478e-01 -4.52013880e-01
-1.13405359e+00 1.02496278e+00 -9.68851671e-02 2.89842576e-01
-1.01364262e-01 1.35740995e-01 2.27511168e-01 3.30292672e-01
6.70892179e-01 1.60262227e-01 -3.61998409e-01 -3.20062697e-01
-1.12384307e+00 3.17274034e-01 9.72042680e-01 9.18352962e-01
4.19001281e-01 1.81074440e-01 -1.06576689e-01 3.94997984e-01
5.18306255e-01 8.12172830e-01 2.14680240e-01 -1.28336036e+00
2.92144679e-02 -3.16802889e-01 3.25310886e-01 -8.55582833e-01
-9.65074152e-02 -3.06061029e-01 -9.62443352e-01 8.47308114e-02
5.22372782e-01 -3.37698698e-01 -3.53756338e-01 2.12439394e+00
1.72213584e-01 4.25176859e-01 -1.79518178e-01 6.68077409e-01
-1.29585147e-01 5.01137257e-01 -2.37014562e-01 -6.83507085e-01
6.59743369e-01 -1.90876275e-01 -7.72079825e-01 2.45807320e-01
3.64122808e-01 -6.39732838e-01 7.25245178e-01 6.28622234e-01
-1.10132301e+00 -3.32782745e-01 -9.35529053e-01 4.67595756e-01
8.44902024e-02 -2.18999058e-01 6.17116928e-01 5.99663138e-01
-1.06855226e+00 1.04779673e+00 -1.20485163e+00 -2.92367071e-01
2.24495620e-01 -1.46582182e-02 5.23661748e-02 2.56952316e-01
-1.07694304e+00 5.00282586e-01 3.64025533e-01 -2.04424158e-01
-7.42652893e-01 -8.57911527e-01 -5.46745658e-01 7.47451410e-02
2.21743152e-01 -3.54248911e-01 1.38188982e+00 -8.41328442e-01
-1.65048695e+00 5.36298335e-01 6.99536428e-02 -6.40619218e-01
7.26193011e-01 -1.37091041e-01 -3.95384908e-01 1.16433844e-01
2.85730034e-01 1.52461335e-01 8.62420857e-01 -8.26237798e-01
-5.62409520e-01 -4.14212883e-01 -5.10832012e-01 -2.00119704e-01
3.83991376e-02 -2.33897746e-01 6.58837035e-02 -8.21955025e-01
1.36136621e-01 -8.94777775e-01 -3.78751606e-01 5.11193164e-02
-2.14727193e-01 -3.32747668e-01 2.24444419e-01 -6.09467447e-01
1.21596611e+00 -2.49419880e+00 1.17951520e-01 5.87805867e-01
2.38806568e-02 -3.67385119e-01 2.74127215e-01 4.59294230e-01
-1.02295585e-01 6.72851205e-02 -7.45909512e-01 -2.01799467e-01
2.16654897e-01 7.50243589e-02 -4.55005199e-01 9.82799768e-01
-1.58211272e-02 3.99207562e-01 -1.05116951e+00 -2.74481982e-01
-7.37388656e-02 1.41883388e-01 -5.22561967e-01 1.73997596e-01
-2.03541309e-01 5.84256291e-01 -1.36476144e-01 2.30032727e-01
8.12369168e-01 -4.06093746e-01 1.01177253e-01 2.46048301e-01
-1.97811216e-01 -2.06702240e-02 -1.33400857e+00 1.70356882e+00
-6.22276440e-02 6.92655563e-01 1.92897618e-01 -1.30619550e+00
8.48645389e-01 3.48702252e-01 6.53496325e-01 -4.29040104e-01
1.91687746e-03 3.32579702e-01 -2.53241271e-01 -2.68765360e-01
3.84322584e-01 -4.90835756e-01 -2.33629957e-01 5.32422006e-01
-4.46572937e-02 -2.84690201e-01 1.64221996e-03 1.01678431e-01
6.28099203e-01 2.30938077e-01 5.14051199e-01 -1.06606138e+00
1.32222563e-01 -3.42737764e-01 6.68216228e-01 7.03334928e-01
-2.73072809e-01 6.34412289e-01 5.79869807e-01 3.12007576e-01
-8.91148865e-01 -1.82602942e+00 -6.23648286e-01 8.54397535e-01
2.09098984e-03 -1.97419599e-01 -4.07076091e-01 -2.66267478e-01
1.90190718e-01 1.15735376e+00 -6.97910011e-01 -1.79829881e-01
-1.59324214e-01 -5.74484050e-01 5.33010840e-01 6.06809974e-01
2.82243460e-01 -2.78897494e-01 -6.47130311e-01 -3.24225053e-02
-2.43809417e-01 -7.50866115e-01 -5.46056092e-01 2.72417009e-01
-1.13238764e+00 -8.05200040e-01 -7.18710780e-01 -5.02125382e-01
2.51767457e-01 -1.57586768e-01 9.96091366e-01 -7.04216361e-01
-3.15634519e-01 6.03854418e-01 -6.32658824e-02 -3.59033942e-01
-2.72895306e-01 -6.69283986e-01 4.55138147e-01 -3.77191678e-02
2.96447814e-01 -8.54497790e-01 -4.77139771e-01 4.17217612e-01
-7.81257153e-01 -4.90781575e-01 1.98241264e-01 8.72469604e-01
6.80828631e-01 1.15099020e-01 6.91155791e-01 -4.33704972e-01
7.56897569e-01 -8.48581970e-01 -9.31554675e-01 2.31531814e-01
-8.64526033e-01 2.98703343e-01 3.61104429e-01 -4.54983532e-01
-1.18858361e+00 -4.35681731e-01 1.47065595e-01 -4.06350195e-01
9.47733596e-02 6.58542752e-01 4.62419063e-01 2.83307165e-01
5.21137476e-01 2.82369941e-01 2.20193744e-01 -4.35596704e-01
3.30351084e-01 4.89121348e-01 6.49877608e-01 -9.47695911e-01
3.32195938e-01 5.95815837e-01 1.55297354e-01 -7.57828057e-01
-6.18449271e-01 -3.43210340e-01 -4.78036731e-01 4.25257608e-02
1.00801826e+00 -7.04599202e-01 -8.48710775e-01 1.84049889e-01
-8.91476750e-01 -2.48945951e-01 -6.36381090e-01 8.93389940e-01
-1.07423627e+00 8.03637326e-01 -4.52390879e-01 -1.38014972e+00
8.59642178e-02 -6.73317432e-01 7.46906340e-01 1.38663530e-01
-2.84494847e-01 -1.50735247e+00 4.67138946e-01 -4.31142092e-01
2.85360575e-01 5.60805976e-01 7.97462821e-01 -6.73694909e-01
-7.99327567e-02 -1.39802136e-02 -1.55458406e-01 4.99100178e-01
1.40212551e-01 3.16082835e-01 -6.98957622e-01 -4.60586280e-01
4.10048783e-01 -2.02241212e-01 7.29986012e-01 8.60278189e-01
1.20773375e+00 -7.04062507e-02 -2.44381458e-01 5.62060177e-01
1.45738530e+00 3.66242111e-01 4.50446427e-01 -2.77182013e-01
-3.37716751e-02 6.10897779e-01 8.71245205e-01 9.64227259e-01
1.34291854e-02 1.67637691e-01 -1.01517467e-02 4.25532877e-01
5.93992949e-01 -1.90706879e-01 4.44811821e-01 1.02538764e+00
5.05047292e-02 4.64483015e-02 -8.34580660e-01 6.52654469e-01
-2.16837859e+00 -1.06924379e+00 6.18789271e-02 2.62767005e+00
9.56213593e-01 4.11493070e-02 2.29672760e-01 -8.80560949e-02
8.57370496e-01 -2.53671259e-01 -6.87963188e-01 -2.50470459e-01
-1.03683367e-01 3.24202031e-01 3.10149282e-01 5.09296536e-01
-8.94359648e-01 2.90210694e-01 7.49154043e+00 1.19740939e+00
-6.18012369e-01 4.33000997e-02 6.50379837e-01 -4.36556153e-02
-4.18858171e-01 7.28856176e-02 -2.85400659e-01 6.70351505e-01
1.04017735e+00 -6.13064110e-01 3.22327197e-01 8.21423173e-01
-9.49611068e-02 -4.84861076e-01 -1.40923500e+00 1.25131118e+00
-8.89850780e-02 -1.03588271e+00 -7.07237720e-01 1.02120757e-01
8.56564820e-01 3.41943763e-02 2.97484756e-01 3.41555141e-02
6.71152413e-01 -1.08371031e+00 7.26392448e-01 7.53432512e-01
7.87241399e-01 -8.77219677e-01 4.93649185e-01 3.86752933e-01
-1.14065647e+00 2.66688652e-02 -2.40286514e-01 -9.34421346e-02
2.10028410e-01 1.22036672e+00 -3.31851006e-01 6.69290841e-01
8.30036581e-01 6.98757648e-01 -1.02415727e-02 1.06540298e+00
2.17597440e-01 6.22328401e-01 -6.08685017e-01 -8.09908584e-02
9.05025974e-02 -8.31784189e-01 7.81945407e-01 1.14497209e+00
6.10952914e-01 -3.16471457e-02 1.25102669e-01 1.31009614e+00
5.15246451e-01 -2.70056218e-01 -4.12174821e-01 -1.93292201e-01
5.56079388e-01 6.66696012e-01 -7.17371941e-01 -1.57355011e-01
-2.01241463e-01 8.91834199e-01 5.86415939e-02 5.59353709e-01
-9.01302814e-01 -5.99880219e-01 7.43763626e-01 -3.00817549e-01
4.75131214e-01 -4.23622727e-01 -4.84236211e-01 -1.19333267e+00
9.46174264e-02 -3.72812033e-01 7.32197285e-01 -2.99275786e-01
-1.83513868e+00 1.06869586e-01 3.95871162e-01 -1.20581532e+00
-5.45248210e-01 -5.25223911e-01 -6.64000213e-01 9.12926733e-01
-8.85987818e-01 -1.39126688e-01 2.57028013e-01 6.77226901e-01
8.90512317e-02 1.77322160e-02 5.18010080e-01 6.33315966e-02
-2.62515813e-01 4.09550875e-01 8.60514998e-01 -1.22400917e-01
7.16257751e-01 -1.60567391e+00 8.47377852e-02 7.80346394e-01
-2.33834788e-01 5.77977896e-01 9.62925196e-01 -5.39398909e-01
-1.01027632e+00 -5.49711287e-01 5.29771566e-01 -3.60191584e-01
1.05945396e+00 -2.63841689e-01 -6.95649385e-01 6.37907386e-01
-2.21211165e-02 -3.09145719e-01 9.62907076e-01 9.98088345e-02
-2.24797815e-01 2.85776444e-02 -1.39545226e+00 4.24065918e-01
6.91515863e-01 -5.05741119e-01 -6.03993177e-01 1.71660900e-01
5.06006360e-01 1.08737759e-01 -1.16813684e+00 3.85457546e-01
6.24550164e-01 -1.36382425e+00 7.96321809e-01 -4.63258386e-01
7.91477785e-02 -1.20024189e-01 -7.14206636e-01 -1.15122294e+00
-2.62418896e-01 -8.97351801e-01 -3.86740267e-02 1.12292552e+00
2.25165084e-01 -8.51974666e-01 2.61174262e-01 6.53137863e-01
1.21000797e-01 -6.40212357e-01 -1.27614510e+00 -1.19610572e+00
4.87698108e-01 -7.14380562e-01 3.31378669e-01 1.01165688e+00
4.63410109e-01 -1.44400299e-01 -2.28921562e-01 -4.22017127e-02
1.00785351e+00 3.55227381e-01 1.53789490e-01 -1.70518994e+00
-7.92736828e-01 -5.75144231e-01 -3.17739964e-01 -1.29003131e+00
-1.53698521e-02 -8.42726350e-01 3.38879287e-01 -8.57184887e-01
2.26128951e-01 -5.91114283e-01 -4.17289168e-01 -2.54416436e-01
-6.59753606e-02 -3.54539096e-01 -1.47048995e-01 4.09395158e-01
-6.60598874e-01 7.94452548e-01 8.97964299e-01 4.13387716e-01
-8.77654850e-02 2.49348491e-01 -4.34754759e-01 8.72802079e-01
6.25057280e-01 -4.58237380e-01 -5.54459095e-01 2.62261853e-02
2.14315251e-01 5.35275996e-01 3.76531363e-01 -8.45245540e-01
2.25719616e-01 -3.98847044e-01 2.22855076e-01 -4.73441422e-01
8.59297141e-02 -6.83745682e-01 6.50239438e-02 6.83587134e-01
-4.23649251e-01 7.46977255e-02 4.83391471e-02 1.18343842e+00
-2.33645499e-01 -2.17730284e-01 1.06484079e+00 8.31783041e-02
-4.24673319e-01 2.60466963e-01 -3.59726161e-01 4.14306074e-01
1.17267954e+00 -1.25149205e-01 1.94725126e-01 -6.34954691e-01
-8.25157046e-01 2.21574351e-01 3.11940998e-01 -8.70556235e-02
6.46780312e-01 -1.42264652e+00 -6.79993331e-01 1.40584812e-01
-1.95288464e-01 -1.45959690e-01 -9.42753628e-03 1.37825596e+00
-1.69732586e-01 1.98087722e-01 -1.74565222e-02 -1.01764822e+00
-6.13939464e-01 4.76181179e-01 1.75664812e-01 -1.57674961e-02
-4.24736977e-01 7.70979226e-01 2.17140272e-01 -3.66661727e-01
3.19445610e-01 -5.31962693e-01 5.34746349e-01 -1.40227184e-01
4.56291944e-01 9.12996769e-01 -4.43896323e-01 1.53837100e-01
-5.28091431e-01 4.67647702e-01 3.97399902e-01 -6.73860967e-01
1.10434520e+00 -3.05603892e-01 -3.07272494e-01 1.16234684e+00
1.19975150e+00 -9.76749361e-02 -1.48691928e+00 -3.46611083e-01
3.95045787e-01 -5.12454748e-01 -1.45253837e-01 -5.26211083e-01
-5.95595300e-01 8.63571882e-01 6.46597445e-01 6.12236381e-01
1.00716257e+00 5.69086298e-02 4.57808882e-01 2.53233999e-01
4.67958421e-01 -1.30560303e+00 -2.58956794e-02 2.45957807e-01
6.49335325e-01 -1.23672938e+00 -7.23003149e-02 -1.68490559e-01
-6.63944840e-01 1.20117235e+00 2.97769327e-02 -2.70477891e-01
1.16074336e+00 4.97411370e-01 -4.69864249e-01 2.54367471e-01
-5.83387733e-01 -5.96583821e-02 2.41266280e-01 8.17972660e-01
3.92734885e-01 3.26346368e-01 -2.27958083e-01 6.72361791e-01
-9.29411948e-02 6.49967715e-02 2.82415509e-01 7.35866070e-01
-5.44637680e-01 -5.68950593e-01 -8.12624618e-02 3.66791904e-01
-3.34330469e-01 8.42036679e-02 1.39116850e-02 5.38916051e-01
-2.33521983e-01 1.01506376e+00 3.00220460e-01 -1.57844424e-01
-3.01763922e-01 1.12251699e-01 5.39065361e-01 -1.45386755e-01
2.25916669e-01 2.76647240e-01 -2.20099851e-01 -6.07639015e-01
-6.13752663e-01 -1.16671658e+00 -1.09562135e+00 -5.15188992e-01
-3.70818287e-01 4.40660089e-01 5.80672026e-01 1.00470591e+00
3.13131630e-01 7.88366348e-02 7.18847275e-01 -4.19908762e-01
-1.26357043e+00 -8.33466709e-01 -1.19348407e+00 2.07898587e-01
3.84394079e-01 -5.65251231e-01 -8.79451513e-01 -7.75214657e-02]
|
[7.221261978149414, 4.088985919952393]
|
e1dcc2c7-6d5d-4632-b706-6bcd30a20f32
|
enhancement-of-noisy-speech-with-low-speech
|
1802.05125
| null |
http://arxiv.org/abs/1802.05125v1
|
http://arxiv.org/pdf/1802.05125v1.pdf
|
Enhancement of Noisy Speech with Low Speech Distortion Based on Probabilistic Geometric Spectral Subtraction
|
A speech enhancement method based on probabilistic geometric approach to
spectral subtraction (PGA) performed on short time magnitude spectrum is
presented in this paper. A confidence parameter of noise estimation is
introduced in the gain function of the proposed method to prevent subtraction
of the overestimated and underestimated noise, which not only removes the noise
efficiently but also prevents the speech distortion. The noise compensated
magnitude spectrum is then recombined with the unchanged phase spectrum to
produce a modified complex spectrum prior to synthesize an enhanced frame.
Extensive simulations are carried out using the speech files available in the
NOIZEUS database in order to evaluate the performance of the proposed method.
|
[]
|
2018-02-13
| null | null | null | null |
['noise-estimation']
|
['medical']
|
[ 6.36069894e-01 -8.45174417e-02 5.41340888e-01 -1.33666426e-01
-6.64546967e-01 -4.15848911e-01 2.49675199e-01 8.71121585e-02
-3.87434304e-01 7.95824289e-01 4.69997108e-01 -1.07528962e-01
-2.60483772e-01 -5.31405747e-01 -5.91643006e-02 -1.04166532e+00
2.24760354e-01 -5.22438705e-01 3.85898769e-01 -1.71742067e-01
8.42405260e-02 2.43835896e-01 -1.50388312e+00 -1.06634058e-01
9.33239639e-01 8.51679921e-01 7.02505589e-01 8.93437266e-01
1.68362454e-01 4.20876533e-01 -1.12616777e+00 -6.33290783e-03
5.50193071e-01 -5.45362413e-01 -2.02145293e-01 1.38643056e-01
-2.11967856e-01 -2.52539039e-01 -2.22602785e-01 1.61560011e+00
9.39070821e-01 4.28971201e-01 5.48717618e-01 -5.73144257e-01
9.29555297e-02 5.14057696e-01 -4.94242996e-01 4.20987576e-01
3.37429672e-01 -1.02648012e-01 3.96336704e-01 -7.36474752e-01
3.29085380e-01 1.05267894e+00 5.95031917e-01 7.49458745e-02
-9.53866541e-01 -6.66606903e-01 -5.54302454e-01 3.05957168e-01
-1.49509990e+00 -6.47569478e-01 1.16269910e+00 -1.11528195e-01
7.19019473e-01 3.79221529e-01 7.37139285e-01 1.73554629e-01
4.80842516e-02 1.86297074e-01 1.46059811e+00 -1.09937704e+00
5.00063151e-02 2.89801750e-02 1.07206613e-01 3.24893653e-01
1.84309438e-01 4.96632695e-01 -3.10118109e-01 -1.53752908e-01
3.46091121e-01 -7.28261411e-01 -7.53188252e-01 1.07395001e-01
-7.60343850e-01 3.63319218e-01 -4.92772684e-02 4.40818727e-01
-5.38128614e-01 -3.03346217e-01 1.88538447e-01 1.71996385e-01
4.99693096e-01 -4.83367778e-02 -9.15107056e-02 -1.49211571e-01
-1.06334364e+00 1.91647097e-01 6.10160232e-01 6.11299694e-01
3.99735779e-01 7.62226522e-01 1.94593146e-01 9.74562228e-01
4.73708481e-01 9.11819637e-01 2.94007570e-01 -8.12896132e-01
3.87146801e-01 -5.26097864e-02 1.55426085e-01 -9.06148016e-01
-1.95857063e-01 -5.28110802e-01 -6.37538254e-01 3.98963332e-01
3.37502837e-01 -5.47298610e-01 -9.27984953e-01 1.44892740e+00
4.30675268e-01 1.21774688e-01 4.03545201e-01 7.58157313e-01
3.38431001e-01 1.18449867e+00 -2.11727321e-01 -7.08485305e-01
1.12445951e+00 -3.67650330e-01 -1.40924931e+00 7.87332803e-02
1.30579993e-02 -1.48450041e+00 3.50694597e-01 5.36694586e-01
-1.12964427e+00 -7.55204856e-01 -1.61614025e+00 3.39354783e-01
-4.10338007e-02 1.96914211e-01 -2.43953153e-01 1.23069036e+00
-9.64275062e-01 4.11989093e-01 -4.80583400e-01 -5.46053685e-02
-1.63928211e-01 2.00676054e-01 -6.98766485e-03 4.01906580e-01
-1.32961178e+00 9.04913962e-01 6.63075387e-01 4.87143904e-01
-2.03802750e-01 -4.92499441e-01 -7.69062400e-01 9.73521098e-02
1.61276162e-01 -1.02265731e-01 1.24758744e+00 -1.06784499e+00
-1.82583177e+00 1.33286163e-01 -2.21540049e-01 -5.55926919e-01
1.78526834e-01 -2.71333884e-02 -1.00108397e+00 2.10982963e-01
-2.81936020e-01 5.10730594e-02 1.09618938e+00 -9.41192269e-01
-6.08713329e-01 6.49768114e-03 -4.38887745e-01 5.02185166e-01
3.43410343e-01 7.78584555e-02 5.58990352e-02 -1.06121361e+00
5.46618819e-01 -4.86778587e-01 2.83330530e-02 -4.75004554e-01
-1.05390258e-01 5.39926529e-01 1.00602853e+00 -1.34715259e+00
1.16362429e+00 -2.46361709e+00 -2.70975977e-01 5.58534086e-01
-3.82949859e-01 6.38414204e-01 1.73751727e-01 3.54593217e-01
-3.63173217e-01 -4.31464493e-01 -5.22400677e-01 1.17182024e-01
-4.04292196e-01 -1.75859541e-01 -1.39718413e-01 5.95646083e-01
-5.23841940e-02 -2.67441660e-01 -6.83562100e-01 -3.22758287e-01
5.73875844e-01 5.87463737e-01 -2.31440485e-01 1.29621863e-01
2.98613697e-01 1.62350342e-01 4.13722731e-02 3.14659685e-01
1.05514455e+00 7.98373818e-01 1.68901145e-01 -5.04113317e-01
-4.41108316e-01 4.32183832e-01 -1.82784295e+00 9.94424880e-01
-2.94549435e-01 7.54735649e-01 5.63403845e-01 -7.50025332e-01
1.06128895e+00 7.17161715e-01 2.41961673e-01 -5.83996952e-01
3.70635599e-01 2.04542682e-01 3.89241666e-01 -2.90546864e-01
7.24018633e-01 -3.03924173e-01 4.99224275e-01 1.16973296e-01
6.05481640e-02 -5.17575681e-01 4.01021652e-02 -2.39300989e-02
3.41108203e-01 7.66993836e-02 7.32505560e-01 -4.63123888e-01
1.05333066e+00 -3.04191411e-01 7.84094751e-01 1.51538178e-01
-6.15513921e-01 4.78714138e-01 -1.63985044e-02 3.88176709e-01
-1.28676772e+00 -1.14899492e+00 -6.05893061e-02 4.03779179e-01
2.02423126e-01 -2.54350267e-02 -1.00563347e+00 8.13061371e-02
-4.32353199e-01 8.91089678e-01 2.63548940e-01 4.45412211e-02
-5.21902978e-01 -7.23774314e-01 5.24264693e-01 1.62979499e-01
6.94401503e-01 -6.60000861e-01 -2.64505237e-01 4.32801336e-01
-5.12498319e-01 -1.01730812e+00 -4.62171912e-01 1.52645141e-01
-5.47842324e-01 -8.24406087e-01 -5.23747981e-01 -6.51286721e-01
4.41777766e-01 4.78033125e-01 7.90606961e-02 -1.57300338e-01
-1.01611987e-01 1.78234771e-01 -5.24126947e-01 -5.95969439e-01
-1.01968503e+00 -6.59898996e-01 9.53628495e-02 1.09247290e-01
-5.37532978e-02 -4.60897297e-01 -3.32137764e-01 2.01885343e-01
-7.02384770e-01 -1.88167199e-01 3.36479902e-01 7.04491377e-01
3.39834183e-01 8.93408716e-01 7.41800606e-01 -2.65074760e-01
7.98649311e-01 -5.00103366e-03 -1.15017164e+00 -9.34367068e-03
-4.39636111e-01 -2.07966715e-01 4.25243318e-01 -2.76517600e-01
-1.82191718e+00 -4.94847680e-03 -4.21017498e-01 -5.17982282e-02
1.43956795e-01 3.17669898e-01 -4.22337860e-01 -1.42253846e-01
5.17226040e-01 3.61701548e-01 1.33818969e-01 -3.51061493e-01
1.77783191e-01 1.10280812e+00 1.07637560e+00 -1.14753477e-01
9.75961268e-01 2.73379475e-01 -5.84277622e-02 -1.40629339e+00
-1.04665466e-01 -5.88735878e-01 -2.99878359e-01 -3.94400865e-01
7.52714217e-01 -8.51919115e-01 -2.17931628e-01 9.02312636e-01
-1.09215987e+00 2.29934156e-01 -2.94391457e-02 1.01593792e+00
-3.88948679e-01 7.33748257e-01 -4.23127830e-01 -1.52719951e+00
-4.70181495e-01 -1.00477135e+00 3.33021104e-01 6.02656305e-01
-1.12799980e-01 -6.26817167e-01 -8.76393020e-02 2.94115245e-01
3.73824775e-01 -2.07956135e-01 6.26276910e-01 -3.96231562e-01
-4.86688554e-01 -1.79733872e-01 5.85576221e-02 8.67172480e-01
3.80644232e-01 2.09143162e-01 -1.14364684e+00 -5.19920029e-02
7.46406913e-01 4.04416412e-01 4.54011351e-01 5.83667755e-01
2.49412537e-01 1.91692725e-01 2.12595388e-02 3.03216189e-01
1.34558344e+00 9.12331998e-01 9.14247572e-01 7.24465474e-02
9.65815410e-03 4.23589349e-01 9.10290241e-01 3.05458665e-01
-2.66267031e-01 3.85775566e-01 -8.03430155e-02 -8.78475700e-03
-4.36938375e-01 1.96155962e-02 2.64046520e-01 1.29609168e+00
7.09086284e-02 -3.22552890e-01 -3.53866398e-01 5.19808531e-01
-1.21927750e+00 -1.28490388e+00 -3.58389199e-01 2.25279808e+00
8.38873446e-01 2.47242421e-01 -2.85276234e-01 9.68010485e-01
1.13136852e+00 4.40636307e-01 1.23222113e-01 -6.02367043e-01
-3.83187175e-01 5.97667992e-01 9.00388956e-01 1.02535677e+00
-1.05414951e+00 4.27704871e-01 6.51946211e+00 1.07983685e+00
-1.13425565e+00 1.55303599e-02 4.24960554e-02 2.45278418e-01
1.99338309e-02 4.94808927e-02 -3.71622503e-01 4.19215292e-01
1.01415801e+00 -5.64407349e-01 5.06313801e-01 4.69451904e-01
7.39690840e-01 -7.60479987e-01 6.22760095e-02 8.29087198e-01
-9.86776203e-02 -8.07581306e-01 -3.18655223e-01 -2.61097699e-01
5.15705228e-01 -4.60438132e-01 -6.70009404e-02 -1.52544275e-01
-9.95799452e-02 -3.45347732e-01 9.38073575e-01 5.10426044e-01
3.81721467e-01 -1.06439292e+00 8.40034842e-01 3.69517475e-01
-1.25400651e+00 7.77001008e-02 -2.67557174e-01 1.42364204e-03
4.78027344e-01 7.48708904e-01 -1.32431209e+00 9.28514600e-01
2.34332755e-01 -1.27282754e-01 1.17716283e-01 1.21853173e+00
-2.54338622e-01 9.83491540e-01 -5.74155271e-01 2.40115568e-01
-1.44322544e-01 -5.34219325e-01 1.07667875e+00 1.04877937e+00
6.18829846e-01 4.78055090e-01 -4.57620203e-01 6.17609203e-01
3.49052668e-01 3.03152829e-01 -1.91702381e-01 -8.65975469e-02
8.10081899e-01 8.98622096e-01 -4.84828770e-01 -3.27170908e-01
-4.77158189e-01 5.93949676e-01 -5.67246497e-01 4.36013550e-01
-6.94116712e-01 -8.03286672e-01 4.20204043e-01 -8.70089792e-03
3.13126028e-01 -2.81776667e-01 -1.66276932e-01 -4.75779235e-01
1.58802271e-02 -7.90540159e-01 1.05724178e-01 -1.07630825e+00
-5.96255422e-01 4.09662575e-01 9.68088210e-02 -1.40400970e+00
-4.01307076e-01 -1.76797330e-01 -4.86788005e-01 1.58436704e+00
-1.13260770e+00 -4.78967905e-01 4.64924164e-02 4.14746672e-01
5.40388584e-01 -2.20318139e-01 5.69060147e-01 4.61476475e-01
-2.25922003e-01 2.91990638e-01 3.27552795e-01 -1.82454914e-01
6.03780091e-01 -9.12446558e-01 1.81551591e-01 1.49570477e+00
-3.23650539e-01 3.10360789e-01 1.28122711e+00 -9.85324681e-01
-5.74946523e-01 -6.87163949e-01 8.60142767e-01 4.83208418e-01
3.19574416e-01 -6.57917559e-02 -1.01175892e+00 9.89061445e-02
4.20288026e-01 -4.25093055e-01 3.95451844e-01 -7.67235816e-01
8.13751593e-02 -1.61510184e-01 -1.44028676e+00 5.49669981e-01
2.25089177e-01 -6.42490327e-01 -9.39649165e-01 -3.14471960e-01
7.11862266e-01 -5.21225572e-01 -5.57364404e-01 4.81219947e-01
3.13278496e-01 -1.09237790e+00 8.52735102e-01 5.41505873e-01
-4.89414901e-01 -9.22682405e-01 -4.84594285e-01 -1.52200317e+00
8.51420611e-02 -1.09518504e+00 4.44722146e-01 1.34947157e+00
3.14150840e-01 -8.27456295e-01 3.68240744e-01 -1.45075381e-01
-1.56170473e-01 1.57048702e-02 -1.00914431e+00 -6.98317647e-01
-6.10539615e-01 -4.87416148e-01 3.33252132e-01 5.46971262e-01
-5.27476743e-02 -9.85435545e-02 -4.53159034e-01 7.36697674e-01
8.76768768e-01 -4.95099872e-01 3.29390496e-01 -8.17020774e-01
-2.25145847e-01 -9.43446532e-02 -5.04275620e-01 -5.33673763e-01
-2.78665990e-01 -2.03847274e-01 3.51401359e-01 -1.22398424e+00
-2.88170815e-01 1.63095519e-01 -1.89721808e-01 -4.24194455e-01
-3.49064618e-01 -6.29412681e-02 1.02936283e-01 -2.90096879e-01
5.04719257e-01 5.06197929e-01 8.74727070e-01 9.37457532e-02
-2.74049073e-01 4.18188781e-01 -1.82834283e-01 9.55194354e-01
7.30517089e-01 -5.50829351e-01 -7.47369468e-01 2.55428135e-01
-5.18794179e-01 4.49379653e-01 2.26601988e-01 -1.28143239e+00
7.98031464e-02 1.23185158e-01 1.75181448e-01 -1.02166677e+00
5.73690236e-01 -1.00163615e+00 3.55733305e-01 4.62503076e-01
1.41222894e-01 -9.02898684e-02 2.67128497e-01 5.78028798e-01
-4.18762326e-01 -5.79810262e-01 1.25205600e+00 3.40277642e-01
-5.86943328e-01 -5.43226719e-01 -6.82797313e-01 -3.50250512e-01
8.67920160e-01 -5.59201300e-01 -1.33301213e-01 -6.77008212e-01
-7.54512072e-01 -4.34318662e-01 7.42649063e-02 -6.26602694e-02
3.76008838e-01 -9.67504978e-01 -5.94724834e-01 2.74992973e-01
-4.62155104e-01 -5.27117968e-01 7.72528410e-01 5.39035201e-01
-8.51241589e-01 2.69525886e-01 -2.57026434e-01 -1.80818960e-01
-1.77234066e+00 3.49875182e-01 6.44312918e-01 1.04197294e-01
-3.28690767e-01 5.68745136e-01 -3.96901459e-01 3.21690179e-02
2.38640264e-01 -2.14359328e-01 -2.16868699e-01 -2.08070233e-01
6.83122218e-01 7.26732910e-01 2.26202741e-01 -8.88331532e-01
-2.19469935e-01 3.89933646e-01 4.28888053e-01 -8.91529799e-01
8.17248464e-01 -5.81441760e-01 -7.10132420e-02 1.78626865e-01
9.66213703e-01 8.44268143e-01 -8.88188899e-01 -1.60841882e-01
-1.12777077e-01 -5.18237591e-01 3.02994132e-01 -9.78974521e-01
-6.69937372e-01 4.78265971e-01 9.67102528e-01 2.56360888e-01
1.60948920e+00 -9.22471583e-01 6.26763046e-01 -3.91124636e-02
1.00041702e-01 -1.41689968e+00 -4.59834099e-01 3.78592700e-01
6.05344832e-01 -5.31555235e-01 2.09422857e-01 -7.19982147e-01
-5.20253360e-01 1.16260731e+00 6.46042153e-02 -4.65701930e-02
7.63085127e-01 6.73945785e-01 3.85236293e-01 5.56241810e-01
-2.32499123e-01 -4.97454971e-01 3.87091492e-03 1.02983224e+00
3.19495380e-01 -4.26895507e-02 -7.75369108e-01 4.45363164e-01
-2.63758123e-01 -1.80357590e-01 1.02776921e+00 9.97606575e-01
-9.05560195e-01 -1.02919841e+00 -1.17719471e+00 -3.23527865e-02
-6.79872453e-01 -1.77445903e-01 2.54140366e-02 6.71465576e-01
1.43921480e-01 1.37760043e+00 -1.01266786e-01 -3.03715110e-01
5.35349965e-01 4.35694158e-01 3.99978578e-01 -6.85786158e-02
-4.49849993e-01 7.92235434e-01 4.25656259e-01 2.77328975e-02
-4.34488058e-01 -7.14154720e-01 -1.45509338e+00 -1.59200758e-01
-6.09294057e-01 3.71761233e-01 9.67825472e-01 7.58930206e-01
-1.38855442e-01 7.40363657e-01 6.58534169e-01 -4.94822294e-01
-6.74587727e-01 -1.08758676e+00 -7.19740212e-01 -2.06112191e-02
3.92410010e-01 -5.02972007e-01 -6.59016490e-01 2.39230007e-01]
|
[15.00145435333252, 5.779177188873291]
|
10f99de3-5069-415b-8a40-ebde9d08a7b1
|
real-time-joint-semantic-segmentation-and
|
1809.04766
| null |
http://arxiv.org/abs/1809.04766v2
|
http://arxiv.org/pdf/1809.04766v2.pdf
|
Real-Time Joint Semantic Segmentation and Depth Estimation Using Asymmetric Annotations
|
Deployment of deep learning models in robotics as sensory information
extractors can be a daunting task to handle, even using generic GPU cards.
Here, we address three of its most prominent hurdles, namely, i) the adaptation
of a single model to perform multiple tasks at once (in this work, we consider
depth estimation and semantic segmentation crucial for acquiring geometric and
semantic understanding of the scene), while ii) doing it in real-time, and iii)
using asymmetric datasets with uneven numbers of annotations per each modality.
To overcome the first two issues, we adapt a recently proposed real-time
semantic segmentation network, making changes to further reduce the number of
floating point operations. To approach the third issue, we embrace a simple
solution based on hard knowledge distillation under the assumption of having
access to a powerful `teacher' network. We showcase how our system can be
easily extended to handle more tasks, and more datasets, all at once,
performing depth estimation and segmentation both indoors and outdoors with a
single model. Quantitatively, we achieve results equivalent to (or better than)
current state-of-the-art approaches with one forward pass costing just 13ms and
6.5 GFLOPs on 640x480 inputs. This efficiency allows us to directly incorporate
the raw predictions of our network into the SemanticFusion framework for dense
3D semantic reconstruction of the scene.
|
['Vladimir Nekrasov', 'Tom Drummond', 'Chunhua Shen', 'Andrew Spek', 'Thanuja Dharmasiri', 'Ian Reid']
|
2018-09-13
| null | null | null | null |
['surface-normals-estimation']
|
['computer-vision']
|
[ 2.47402415e-01 3.68308365e-01 3.94950271e-01 -2.67440677e-01
-7.56942749e-01 -6.99289501e-01 3.01056623e-01 2.59694427e-01
-8.81603301e-01 6.11600757e-01 -2.97098964e-01 -3.87731284e-01
6.55905753e-02 -7.97637820e-01 -1.01293576e+00 -6.81546926e-01
9.66619998e-02 7.42806613e-01 7.87031114e-01 -2.99480353e-02
2.65978813e-01 6.89827085e-01 -1.92017245e+00 6.45579547e-02
5.59709966e-01 1.29190063e+00 5.06551981e-01 8.63093674e-01
1.73937958e-02 6.72248960e-01 -4.45831627e-01 -2.96545923e-01
4.03449535e-01 2.51177162e-01 -9.25765693e-01 -8.22192430e-03
3.69773358e-01 -4.40547138e-01 -1.31122664e-01 8.13387811e-01
5.59455276e-01 9.00897235e-02 2.56787807e-01 -1.12999034e+00
2.67562270e-01 1.02762058e-01 -4.85628724e-01 1.61179074e-03
1.60215572e-01 2.46905357e-01 4.23128814e-01 -4.98569340e-01
4.18843120e-01 1.08117533e+00 6.45355701e-01 4.39963847e-01
-1.08596992e+00 -4.25508708e-01 1.32012144e-01 1.72566295e-01
-1.16298223e+00 -5.72638571e-01 3.94779027e-01 -3.60591114e-01
1.21863663e+00 4.75953743e-02 7.68456876e-01 7.66472161e-01
-2.66783647e-02 7.19118178e-01 1.07817054e+00 -2.57722944e-01
5.63206792e-01 -8.55066627e-03 1.19544037e-01 8.11109960e-01
2.24702060e-01 -1.04965493e-01 -3.91816258e-01 7.62162134e-02
8.85149896e-01 -4.85361777e-02 -6.61804974e-02 -5.01024485e-01
-1.15239060e+00 6.85187399e-01 5.13813198e-01 -1.49011821e-01
-2.27266565e-01 4.30200100e-01 5.30706525e-01 1.40198022e-01
3.32043380e-01 2.26039484e-01 -7.64537215e-01 -2.69649446e-01
-7.67021477e-01 3.19211066e-01 9.84055102e-01 9.73491371e-01
1.17030323e+00 -3.44633669e-01 5.25335014e-01 3.58220160e-01
2.64684170e-01 3.96798372e-01 3.04031253e-01 -1.18488526e+00
4.42217231e-01 3.61459553e-01 1.27869487e-01 -6.88831627e-01
-7.01276064e-01 -1.76156729e-01 -5.03698766e-01 5.74430406e-01
8.59526813e-01 -3.06172043e-01 -1.15866685e+00 1.40903711e+00
5.55901408e-01 1.66058809e-01 6.62575811e-02 9.51178133e-01
5.33357859e-01 2.88159579e-01 7.15261996e-02 4.01435852e-01
1.52985013e+00 -1.15539312e+00 -1.80057567e-02 -5.22210836e-01
7.01331496e-01 -6.95945561e-01 9.62248445e-01 6.65993631e-01
-9.43312943e-01 -3.75080884e-01 -1.09342134e+00 -5.27451515e-01
-5.80118358e-01 1.00594379e-01 1.03239620e+00 4.99943793e-01
-1.26684856e+00 7.45626867e-01 -1.30871177e+00 -4.73663002e-01
5.56179702e-01 7.95122921e-01 -4.14743006e-01 -2.06995457e-01
-5.18844783e-01 8.39641273e-01 5.05119860e-01 2.89373472e-02
-8.19175005e-01 -6.33025527e-01 -7.22969770e-01 3.48410159e-02
5.40860772e-01 -1.01270831e+00 1.26983464e+00 -8.09214473e-01
-1.64293039e+00 8.15055907e-01 -1.29401445e-01 -4.46495205e-01
6.51468456e-01 -4.75302309e-01 3.46539915e-01 2.18709245e-01
-4.20495905e-02 8.92334819e-01 4.76538479e-01 -1.24899924e+00
-5.92014492e-01 -7.97187507e-01 4.51768994e-01 3.74585688e-01
2.52760276e-02 -2.50057578e-01 -7.23497093e-01 -1.52932376e-01
3.55254531e-01 -9.99180377e-01 -5.52838802e-01 3.20356250e-01
-2.61028379e-01 5.65209202e-02 7.75256872e-01 -5.29303491e-01
3.43111724e-01 -2.20809960e+00 1.28697440e-01 6.32774681e-02
1.76963538e-01 3.43518615e-01 1.70109868e-01 5.26273362e-02
3.48816216e-01 -3.66390973e-01 -3.00577939e-01 -8.11927259e-01
-6.46028891e-02 5.47844827e-01 -1.15466744e-01 4.47827488e-01
-3.24502215e-02 8.27427924e-01 -8.05401444e-01 -4.82437521e-01
4.61376429e-01 5.87631047e-01 -7.44437039e-01 -1.93189569e-02
-2.18314752e-01 6.77399457e-01 -3.47401768e-01 4.37368363e-01
7.89204061e-01 -1.73904181e-01 1.02011330e-01 -7.11426288e-02
-1.54023185e-01 3.64695549e-01 -1.41645646e+00 2.21627402e+00
-7.34558940e-01 4.20519143e-01 3.85212362e-01 -1.13823104e+00
5.50204158e-01 -1.10713160e-02 3.90962541e-01 -5.57913840e-01
4.33125824e-01 4.88619477e-01 -4.86451536e-01 -3.84245723e-01
3.90138090e-01 -2.20636711e-01 -1.27850726e-01 3.72517258e-01
3.82494837e-01 -3.22467178e-01 -1.44329444e-01 -1.37200192e-01
1.16097701e+00 4.24241036e-01 2.92853899e-02 -1.91891015e-01
1.29993364e-01 2.25348681e-01 2.56832242e-01 6.81130588e-01
-1.64614916e-01 5.90255499e-01 3.51611614e-01 -5.73989332e-01
-9.85974789e-01 -1.06420588e+00 7.41348565e-02 1.09887755e+00
4.19697791e-01 -1.44287676e-01 -9.27846611e-01 -4.65729505e-01
-5.07858507e-02 2.79949516e-01 -4.31117237e-01 2.67206132e-01
-6.19859755e-01 -6.95781946e-01 6.06479228e-01 7.13922918e-01
6.32977724e-01 -8.30736756e-01 -1.42977607e+00 1.64981037e-01
7.29225529e-03 -1.38431668e+00 4.05820936e-01 6.99410379e-01
-9.49895740e-01 -1.03915977e+00 -5.21899819e-01 -5.84604383e-01
6.03069901e-01 4.47201461e-01 9.14085448e-01 -9.68618598e-03
-4.25570399e-01 1.99057534e-01 -1.85229495e-01 -4.89076436e-01
2.05743968e-01 2.71680087e-01 -2.45943382e-01 -5.12010872e-01
1.95306465e-01 -8.56963634e-01 -7.09641814e-01 6.91560507e-02
-9.56095040e-01 3.37331653e-01 5.85779428e-01 4.64848071e-01
6.92889690e-01 2.02200580e-02 4.74518500e-02 -7.83915937e-01
-9.65453088e-02 -3.08247626e-01 -8.50400031e-01 -1.25166252e-01
-2.57808775e-01 1.03057750e-01 7.48992085e-01 -1.77567199e-01
-8.82454932e-01 5.41283488e-01 -4.48889881e-01 -2.64721304e-01
-5.06628275e-01 8.98919255e-02 -1.10113122e-01 -4.17622387e-01
3.44867796e-01 2.39559505e-02 7.24787563e-02 -6.19134784e-01
4.44354177e-01 5.03693104e-01 6.99802399e-01 -6.77539170e-01
4.65373576e-01 9.14657056e-01 1.79991752e-01 -7.54816294e-01
-9.04763341e-01 -6.44385993e-01 -9.10901308e-01 -6.16618097e-02
1.01807940e+00 -1.12032080e+00 -1.09653723e+00 7.73694813e-01
-1.23393869e+00 -7.94479251e-01 -3.54750901e-01 4.64905709e-01
-8.63263607e-01 3.16310197e-01 -6.45604372e-01 -5.22705317e-01
-1.51046991e-01 -1.22227943e+00 1.42584145e+00 2.55551487e-01
1.07982598e-01 -8.50655735e-01 -1.64931431e-01 5.84603369e-01
3.82748604e-01 2.87719995e-01 5.88966191e-01 -3.89080644e-01
-8.47582638e-01 -6.90933242e-02 -6.58255816e-01 1.28664821e-01
-3.39553982e-01 -4.76669312e-01 -1.42285335e+00 -1.73852742e-01
1.84639305e-01 -5.01831412e-01 8.02404404e-01 9.61981043e-02
1.27057755e+00 -1.94899272e-02 -2.90612131e-01 8.77777100e-01
1.76883245e+00 -1.21030577e-01 6.27831638e-01 5.36381304e-01
8.34641278e-01 5.94990134e-01 5.35485327e-01 4.51342106e-01
8.27946901e-01 7.16914594e-01 6.82155073e-01 -1.95763096e-01
-7.88878053e-02 -4.96684760e-02 1.63368806e-02 5.90053916e-01
-2.62117416e-01 -8.28826651e-02 -1.06029546e+00 4.66581523e-01
-1.97938013e+00 -4.00213480e-01 -6.90014362e-02 2.24118257e+00
3.36627066e-01 2.34917253e-01 7.67403916e-02 2.44459972e-01
1.48230016e-01 -1.18594676e-01 -5.95584214e-01 -3.69133413e-01
2.40322709e-01 5.35987556e-01 9.65535700e-01 5.62872708e-01
-1.13462961e+00 1.14618444e+00 5.84156084e+00 5.40875137e-01
-1.39589679e+00 3.01619172e-01 3.93404752e-01 -2.72354186e-01
6.83750138e-02 7.92850107e-02 -7.00943589e-01 2.95190096e-01
9.79018092e-01 4.54661816e-01 6.64984047e-01 9.84653771e-01
-1.18244931e-01 -6.68037236e-01 -9.50624287e-01 9.87231851e-01
-5.11890836e-02 -9.69201624e-01 -2.42268875e-01 1.43110603e-01
3.47168118e-01 5.71795762e-01 -3.92492950e-01 1.74125299e-01
4.12051827e-01 -9.50864434e-01 8.17516029e-01 2.17927888e-01
5.06957352e-01 -6.60381556e-01 7.78966367e-01 7.07429409e-01
-1.03790760e+00 -1.35326609e-01 -4.21514571e-01 -3.55472654e-01
1.62341863e-01 7.04614103e-01 -7.76253104e-01 5.92604637e-01
9.41315114e-01 2.46187255e-01 -5.03396094e-01 1.00917137e+00
-1.17080972e-01 8.97421315e-02 -9.07140195e-01 8.79032612e-02
3.59865636e-01 1.60722762e-01 1.39543742e-01 1.07759404e+00
3.21835518e-01 2.96293888e-02 1.15492314e-01 5.34902155e-01
2.27553427e-01 -2.38981560e-01 -4.07886058e-01 4.76677537e-01
2.25522622e-01 1.13061845e+00 -1.28475487e+00 -3.48532826e-01
-2.50648886e-01 1.22510970e+00 5.76831162e-01 1.04417130e-01
-7.91545987e-01 -3.52436364e-01 5.98460615e-01 7.70725608e-02
5.81549644e-01 -6.14341736e-01 -7.19776452e-01 -1.16005683e+00
1.65332764e-01 -3.67183656e-01 5.59121892e-02 -7.60642946e-01
-6.76464319e-01 3.17818224e-01 -1.25420570e-01 -7.35823572e-01
-5.14986552e-02 -9.23853636e-01 -1.64661169e-01 6.74345851e-01
-1.89136064e+00 -1.14636004e+00 -5.94005525e-01 7.50634253e-01
3.42518866e-01 5.57894170e-01 9.73240614e-01 4.77366954e-01
-2.36812785e-01 2.65595257e-01 -1.74617201e-01 -1.40106753e-01
3.27379465e-01 -1.27499330e+00 5.41928709e-01 6.95760548e-01
-8.79648253e-02 3.70524734e-01 5.92302740e-01 -2.73846835e-01
-1.60388219e+00 -9.69410777e-01 6.58072710e-01 -4.43658382e-01
4.97105688e-01 -6.72864556e-01 -6.32580400e-01 7.02808738e-01
-3.22055042e-01 1.28743842e-01 4.18194681e-01 1.23791816e-02
-1.66430533e-01 4.84478325e-02 -1.24492633e+00 3.64897579e-01
1.17350590e+00 -4.72321689e-01 -2.49920920e-01 3.20449710e-01
7.13877499e-01 -7.93251574e-01 -7.68331110e-01 4.20750111e-01
5.66385388e-01 -1.18353450e+00 1.01425600e+00 -8.18000585e-02
2.28133500e-01 -4.16025788e-01 -2.73125082e-01 -9.07767951e-01
2.13656455e-01 -4.51512903e-01 -6.82209730e-02 7.25488842e-01
1.02196448e-01 -7.34111845e-01 1.09569097e+00 5.72977662e-01
-2.93441206e-01 -8.09948266e-01 -1.14070785e+00 -5.57633638e-01
-1.22024119e-01 -7.59299338e-01 5.89227855e-01 5.21715522e-01
-2.28578225e-01 2.40145639e-01 5.13399765e-02 3.95477831e-01
5.17409384e-01 2.89833192e-02 1.05185831e+00 -1.23690116e+00
-5.75911880e-01 -1.46421015e-01 -6.79304540e-01 -1.39238596e+00
-8.41446295e-02 -5.41398227e-01 3.53308707e-01 -1.58361816e+00
-7.32640699e-02 -7.42188990e-01 1.90839195e-03 7.53223538e-01
1.66587740e-01 6.01318836e-01 2.47892261e-01 1.46431625e-01
-6.24651968e-01 2.78165281e-01 1.16254258e+00 3.50492060e-01
-2.31021475e-02 -3.91603783e-02 -6.28913164e-01 9.71254408e-01
6.99540377e-01 -3.30320656e-01 -4.52413797e-01 -9.65543807e-01
2.35230297e-01 3.72099914e-02 8.18624496e-01 -1.52739048e+00
4.01089817e-01 1.62646383e-01 1.98641494e-01 -3.59492511e-01
6.85717404e-01 -9.53669786e-01 1.42437348e-03 3.89799744e-01
2.59683907e-01 -1.76116914e-01 5.23607194e-01 4.72490788e-01
-2.75832135e-02 -1.94778800e-01 7.69913554e-01 -3.86996180e-01
-9.41585660e-01 5.51277846e-02 -1.84401557e-01 -1.96963623e-01
1.15042174e+00 -4.05384511e-01 -3.47490549e-01 1.02623619e-01
-8.42959285e-01 9.19300038e-03 7.66024947e-01 8.96186456e-02
3.57430935e-01 -6.26957238e-01 -6.21218309e-02 2.49669150e-01
-1.15051471e-01 8.81473958e-01 3.33257467e-01 8.35800529e-01
-9.78236854e-01 3.72378528e-01 -2.43067890e-01 -6.90637767e-01
-9.48690116e-01 4.28062171e-01 2.03014836e-01 -2.65596390e-01
-8.03342640e-01 9.77802455e-01 1.89508319e-01 -5.86432517e-01
3.00961763e-01 -4.30406302e-01 1.36511847e-01 -1.11712642e-01
3.82493138e-01 3.77658695e-01 5.12690425e-01 -3.44994903e-01
-3.99941295e-01 7.00942159e-01 2.19333991e-01 -1.30768195e-01
1.52528536e+00 -2.28081048e-01 -7.38521889e-02 2.63781160e-01
1.16009772e+00 -2.51839876e-01 -1.61583126e+00 6.44635409e-02
-1.43783599e-01 -2.45289937e-01 1.98114023e-01 -8.00152421e-01
-9.76603210e-01 1.11059785e+00 6.39525712e-01 -5.57102971e-02
1.13033974e+00 2.71465350e-02 1.05059409e+00 4.90414977e-01
9.09769893e-01 -1.07392550e+00 -1.67824432e-01 5.93914390e-01
1.69910729e-01 -1.26834989e+00 8.78487900e-03 -6.45838857e-01
-3.00207913e-01 1.14246368e+00 3.82204801e-01 -3.14788401e-01
5.96430659e-01 5.12624264e-01 2.30024382e-01 -3.65131527e-01
-3.95771772e-01 -2.91090697e-01 -2.32706919e-01 5.62658727e-01
-9.78921633e-03 2.49193478e-02 2.47202963e-01 2.09798843e-01
-3.91993165e-01 3.03482205e-01 3.58512282e-01 1.22869754e+00
-5.43838978e-01 -9.11765039e-01 -1.33272424e-01 2.26629212e-01
-3.73771220e-01 1.35740116e-02 1.12416893e-02 6.90577686e-01
3.81109834e-01 7.56717980e-01 1.62102342e-01 -2.83957124e-01
1.91869318e-01 -9.79975760e-02 6.54786766e-01 -7.30697513e-01
-4.76855993e-01 -1.09546341e-01 6.01708982e-03 -9.21011865e-01
-4.42696363e-01 -5.20289540e-01 -1.43418419e+00 -3.31313819e-01
-1.36031404e-01 -3.57203484e-01 1.28504324e+00 1.23450005e+00
5.10532439e-01 4.93069470e-01 8.98350105e-02 -1.50906289e+00
-3.58922333e-01 -5.95442355e-01 -4.37135696e-01 5.91555387e-02
2.50935107e-01 -7.21139491e-01 -2.48344690e-01 -8.18483084e-02]
|
[8.484028816223145, -2.3378190994262695]
|
9c76825d-f7f0-434e-8bf4-2032eaf3c271
|
butknot-at-semeval-2016-task-5-supervised
| null | null |
https://aclanthology.org/S16-1048
|
https://aclanthology.org/S16-1048.pdf
|
BUTknot at SemEval-2016 Task 5: Supervised Machine Learning with Term Substitution Approach in Aspect Category Detection
| null |
["Jakub Mach{\\'a}{\\v{c}}ek"]
|
2016-06-01
| null | null | null |
semeval-2016-6
|
['aspect-category-detection']
|
['natural-language-processing']
|
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
|
[-7.350883960723877, 3.7655811309814453]
|
109f428d-f691-4f67-8c41-d0a132d4e401
|
supervised-learning-in-the-presence-of-noise
|
2103.07808
| null |
https://arxiv.org/abs/2103.07808v1
|
https://arxiv.org/pdf/2103.07808v1.pdf
|
Supervised Learning in the Presence of Noise: Application in ICD-10 Code Classification
|
ICD coding is the international standard for capturing and reporting health conditions and diagnosis for revenue cycle management in healthcare. Manually assigning ICD codes is prone to human error due to the large code vocabulary and the similarities between codes. Since machine learning based approaches require ground truth training data, the inconsistency among human coders is manifested as noise in labeling, which makes the training and evaluation of ICD classifiers difficult in presence of such noise. This paper investigates the characteristics of such noise in manually-assigned ICD-10 codes and furthermore, proposes a method to train robust ICD-10 classifiers in the presence of labeling noise. Our research concluded that the nature of such noise is systematic. Most of the existing methods for handling label noise assume that the noise is completely random and independent of features or labels, which is not the case for ICD data. Therefore, we develop a new method for training robust classifiers in the presence of systematic noise. We first identify ICD-10 codes that human coders tend to misuse or confuse, based on the codes' locations in the ICD-10 hierarchy, the types of the codes, and baseline classifier's prediction behaviors; we then develop a novel training strategy that accounts for such noise. We compared our method with the baseline that does not handle label noise and the baseline methods that assume random noise, and demonstrated that our proposed method outperforms all baselines when evaluated on expert validated labels.
|
['Javed Aslam', 'Amir Tahmasebi', 'Bingyang Ye', 'Cheng Li', 'Youngwoo Kim']
|
2021-03-13
| null | null | null | null |
['code-classification']
|
['computer-code']
|
[ 6.84114471e-02 8.42672884e-02 -4.72251862e-01 -5.03394842e-01
-7.99745321e-01 -7.05269992e-01 -8.91227217e-04 6.85154796e-01
-1.61029339e-01 4.97573644e-01 3.61488551e-01 -6.19164288e-01
-3.28975946e-01 -7.14937508e-01 -3.75590533e-01 -2.66376883e-01
1.81965694e-01 5.82926869e-01 -2.64095873e-01 2.42371067e-01
8.83825868e-02 9.26899761e-02 -1.52393174e+00 5.84083796e-01
9.83015060e-01 1.02980924e+00 5.90129420e-02 2.35130221e-01
-1.71617702e-01 1.36578631e+00 -6.76741838e-01 -4.02760237e-01
3.22892666e-01 -4.97587860e-01 -8.65652859e-01 3.69197614e-02
1.00889720e-01 -8.46577585e-02 4.01885621e-03 1.23916829e+00
2.63115346e-01 -3.54675949e-01 1.10376525e+00 -1.14206696e+00
-6.06874645e-01 5.64490736e-01 -3.57297547e-02 2.39700060e-02
6.48498952e-01 -3.63837987e-01 6.20660901e-01 -4.67630923e-01
5.54165065e-01 9.30553019e-01 1.37388122e+00 4.99346763e-01
-1.18080997e+00 -9.88028049e-01 -2.84978095e-02 -1.61850289e-01
-1.65126514e+00 -1.33351296e-01 4.44936991e-01 -1.09432352e+00
7.39249289e-01 2.51582384e-01 4.29951817e-01 1.04427028e+00
3.24627578e-01 6.41379431e-02 1.14204979e+00 -7.17221737e-01
4.05020654e-01 5.19639015e-01 3.91082227e-01 5.16421616e-01
6.42200768e-01 1.27464235e-01 2.60470480e-01 -6.82744801e-01
4.44769681e-01 4.89672542e-01 -9.72225890e-03 -1.53255597e-01
-1.07496929e+00 9.44937110e-01 1.57876790e-01 3.59815180e-01
-5.95413804e-01 -3.07694137e-01 5.90303898e-01 2.57219374e-01
3.05382490e-01 4.79276478e-01 -7.18682289e-01 -9.39704329e-02
-1.05613899e+00 2.23449707e-01 7.48227119e-01 1.20283639e+00
4.56946522e-01 -3.68989736e-01 -2.30146080e-01 1.03359401e+00
2.47305676e-01 4.36255425e-01 7.24226236e-01 -9.97079909e-01
4.90268052e-01 9.97051299e-01 2.20353767e-01 -1.44567657e+00
-6.80226743e-01 -5.09001255e-01 -1.07461405e+00 -2.14491710e-01
1.70318693e-01 -1.88831419e-01 -9.59688604e-01 1.47946525e+00
-4.25465368e-02 -2.51549482e-01 2.26672158e-01 3.70720863e-01
1.01780021e+00 9.73447114e-02 2.91152596e-01 -5.88579059e-01
1.42983580e+00 -4.10827279e-01 -1.14955580e+00 5.66155091e-02
1.17347717e+00 -6.02759898e-01 7.61399329e-01 4.02105302e-01
-6.10608041e-01 -5.95976233e-01 -4.49919701e-01 4.23451960e-01
-3.90801877e-01 2.06833780e-01 5.61450481e-01 9.30227101e-01
-5.90994358e-01 2.44836673e-01 -3.42883915e-01 -3.09840173e-01
4.74665374e-01 1.60701662e-01 -2.36259073e-01 -9.26627740e-02
-1.24861717e+00 8.32166612e-01 4.67294782e-01 -3.60297441e-01
-4.93497074e-01 -5.88859141e-01 -1.03079152e+00 -5.51070422e-02
1.22299634e-01 -3.69726866e-01 1.39394188e+00 -1.05288601e+00
-4.50539440e-01 9.99957979e-01 -1.02729350e-01 -3.21889281e-01
4.35910732e-01 2.33564377e-01 -9.08469498e-01 -1.75439715e-01
6.50333405e-01 1.93486646e-01 2.83820301e-01 -1.43038201e+00
-7.61830986e-01 -1.55129597e-01 -7.68068507e-02 -8.40224698e-02
-1.02059357e-01 7.09644705e-02 2.31438372e-02 -9.74540591e-01
2.93036342e-01 -1.01100028e+00 -3.40272516e-01 -5.26647985e-01
-1.39231369e-01 -3.58925104e-01 7.38973841e-02 -4.73721296e-01
1.83446932e+00 -2.22500253e+00 -7.24149287e-01 6.28665388e-01
2.45035753e-01 9.71623585e-02 4.73986119e-01 3.74253362e-01
-4.19030786e-01 3.01693559e-01 -3.21792126e-01 9.15428624e-02
-2.59671569e-01 5.11040151e-01 -2.31779143e-01 1.75362512e-01
-3.53928983e-01 4.26401705e-01 -9.91850078e-01 -7.31738687e-01
2.51380391e-02 6.54931739e-03 -6.32714629e-01 1.08587682e-01
3.51092458e-01 2.43679687e-01 -1.17655888e-01 7.88040876e-01
6.07502401e-01 -4.26190734e-01 3.60925019e-01 -2.65203446e-01
1.66085035e-01 2.42334068e-01 -1.44721687e+00 1.16011679e+00
-2.60625362e-01 3.21601257e-02 -4.81041580e-01 -1.17585218e+00
9.89909947e-01 6.69424534e-01 7.46241808e-01 -3.19522828e-01
1.45910397e-01 4.97997314e-01 1.15550663e-02 -8.56465101e-01
-8.69460702e-02 -2.71980435e-01 -3.72611433e-01 2.00430200e-01
-2.61931151e-01 3.60746682e-02 2.61237286e-03 1.58849489e-02
1.32443798e+00 -6.06819212e-01 7.59994149e-01 -2.77421743e-01
2.86926091e-01 2.24403024e-01 8.72484863e-01 1.09464359e+00
-2.22639978e-01 7.70625949e-01 3.07827353e-01 -7.65277684e-01
-7.09774673e-01 -8.05464029e-01 -6.58437490e-01 7.65285313e-01
-1.76118314e-01 -5.09070277e-01 -6.69281483e-01 -1.11169112e+00
2.54203111e-01 6.30031884e-01 -7.34407365e-01 -1.85274959e-01
-3.23237628e-02 -7.60881841e-01 7.21755266e-01 6.51188076e-01
1.63538724e-01 -9.19907868e-01 -5.72571933e-01 3.65218669e-01
-5.27049065e-01 -9.68873382e-01 -2.73853034e-01 5.66304862e-01
-7.64455438e-01 -1.58115804e+00 -2.68835366e-01 -1.00132096e+00
1.01619339e+00 -9.52755809e-02 1.34115636e+00 3.31413686e-01
-1.20558403e-01 2.14155793e-01 -6.42108321e-01 -8.06096435e-01
-7.98024833e-01 -1.11553133e-01 1.98353723e-01 -2.60598689e-01
8.68011594e-01 -2.44317986e-02 -3.88650149e-01 3.04486066e-01
-8.52822244e-01 -2.62175322e-01 3.47604662e-01 7.99044847e-01
5.81641495e-01 6.41607583e-01 7.76942015e-01 -1.42715073e+00
8.82718861e-01 -8.00902069e-01 -1.33493707e-01 1.83413759e-01
-1.07391560e+00 -1.84197173e-01 4.79669929e-01 -2.99739540e-01
-6.85224771e-01 2.28007630e-01 -2.92364299e-01 -9.46354643e-02
-5.25899708e-01 6.39720976e-01 1.01636931e-01 1.74726337e-01
9.80282784e-01 -2.32743949e-01 -2.64956295e-01 -5.39027154e-01
-2.90083855e-01 1.40989661e+00 1.93764836e-01 -3.52619886e-01
2.74700761e-01 2.67081976e-01 -3.55936646e-01 8.60431492e-02
-1.13676083e+00 -8.03723454e-01 -3.38180721e-01 5.31722382e-02
7.69424081e-01 -1.03858232e+00 -2.78822124e-01 2.44240597e-01
-1.01283038e+00 1.84965312e-01 -3.32737029e-01 6.92987502e-01
-4.05577540e-01 1.69780180e-01 -5.75399935e-01 -8.51362944e-01
-8.25699270e-02 -1.22223759e+00 7.12792516e-01 -4.17633653e-01
-9.01121378e-01 -9.90771651e-01 -1.19008850e-02 3.58713388e-01
1.28646940e-01 3.58184218e-01 1.23043370e+00 -8.34803760e-01
4.47075069e-01 -4.56459463e-01 -2.08602354e-01 4.82807010e-01
6.78027630e-01 -3.87594223e-01 -6.60611570e-01 -6.12162650e-02
2.66163230e-01 -1.50180638e-01 5.31499267e-01 4.74014461e-01
1.49686337e+00 -5.29715538e-01 -5.68751872e-01 3.46394926e-01
1.43175066e+00 7.23822892e-01 4.66608196e-01 2.86085874e-01
5.85673213e-01 5.35833240e-01 5.53562939e-01 4.16652888e-01
5.72787583e-01 5.85238457e-01 1.03823520e-01 -4.16852385e-01
4.62143540e-01 -1.92521572e-01 -2.24827036e-01 8.45856607e-01
-1.07305191e-01 2.27018535e-01 -1.38004267e+00 6.32976890e-01
-1.75299764e+00 -7.86394536e-01 -3.32556009e-01 2.08474660e+00
1.17609251e+00 2.09359631e-01 9.05001685e-02 6.51108801e-01
7.83052206e-01 -7.62424469e-01 -1.47749543e-01 -5.91097593e-01
2.39102647e-01 -4.75023426e-02 8.60319376e-01 1.03412174e-01
-1.17994618e+00 1.65403008e-01 7.48954296e+00 4.42923844e-01
-5.15664637e-01 2.84594297e-01 7.29820013e-01 3.35115224e-01
-1.97875589e-01 -3.89072061e-01 -5.55847585e-01 8.86092663e-01
9.14826930e-01 3.44907522e-01 3.46768685e-02 1.01568985e+00
1.54272780e-01 -1.10167069e-02 -1.10322821e+00 1.24314439e+00
2.31333952e-02 -1.05237794e+00 1.16286054e-01 1.47654414e-01
1.12687242e+00 -3.62753689e-01 -3.04677039e-01 4.11534458e-01
6.00218832e-01 -1.08268476e+00 7.44214535e-01 5.46710134e-01
9.84285295e-01 -6.85629547e-01 1.43932462e+00 4.44803298e-01
-9.50665414e-01 -5.89379728e-01 -2.46092349e-01 -3.46701622e-01
-1.82069272e-01 8.81710768e-01 -5.61193585e-01 3.84601116e-01
8.78348827e-01 6.90012813e-01 -5.50809622e-01 9.64833796e-01
2.11006328e-01 8.44476759e-01 9.89057720e-02 6.42628133e-01
-2.16517765e-02 3.10857028e-01 -1.82317376e-01 1.44850254e+00
5.38514316e-01 1.43912852e-01 5.02232075e-01 4.16093439e-01
-1.08803093e-01 2.85119772e-01 -8.28852534e-01 2.80486047e-01
7.93991446e-01 6.73297167e-01 -8.35673213e-01 -6.21717155e-01
-8.06670129e-01 5.05411208e-01 1.56087102e-02 2.19506606e-01
-8.48084331e-01 -2.62364864e-01 4.06545639e-01 4.58038002e-01
2.82214545e-02 4.78959322e-01 -6.65748477e-01 -9.55069363e-01
5.98814227e-02 -1.24734461e+00 8.35291326e-01 -3.98623616e-01
-1.57126963e+00 5.45888007e-01 9.64277834e-02 -1.58988488e+00
-2.35996038e-01 -3.71641815e-01 1.09135754e-01 4.97877061e-01
-1.08637798e+00 -6.94993377e-01 -3.80640268e-01 5.81283808e-01
1.96463674e-01 -2.76037782e-01 1.23332858e+00 6.68758333e-01
-1.17788717e-01 8.27271283e-01 1.89396381e-01 5.50814688e-01
7.65333235e-01 -1.12398744e+00 -1.19110927e-01 1.89279184e-01
-8.79676193e-02 7.60139883e-01 4.83762890e-01 -9.25751090e-01
-4.15433943e-01 -1.42172039e+00 1.32363749e+00 -6.09900713e-01
1.71101838e-01 -2.35093802e-01 -9.01779652e-01 7.00620234e-01
-4.01572168e-01 6.92113563e-02 1.20615721e+00 9.96920466e-02
-4.07484710e-01 -2.13130444e-01 -1.61166215e+00 -1.17822476e-01
1.15303326e+00 -5.34272790e-01 -9.02764440e-01 5.27482688e-01
4.56932247e-01 -3.29403132e-01 -9.60781276e-01 6.31640017e-01
4.87853676e-01 -6.38743222e-01 7.54661500e-01 -5.54137647e-01
2.71350920e-01 -2.49375477e-01 -2.52967864e-01 -1.10081232e+00
-6.82563901e-01 1.09204233e-01 2.82267898e-01 1.14824831e+00
4.90009964e-01 -4.93694305e-01 2.74803311e-01 7.56204486e-01
-3.42077725e-02 -4.56190765e-01 -6.98067248e-01 -8.93078327e-01
-1.81767102e-02 -7.12504089e-01 6.83097482e-01 1.53086185e+00
3.11689913e-01 -2.79030144e-01 -3.37082118e-01 2.68784091e-02
2.01949820e-01 -1.92070797e-01 2.47324795e-01 -1.68221951e+00
2.15047020e-02 -7.27330595e-02 -6.51151538e-01 -2.58832753e-01
5.98785281e-03 -9.29303586e-01 1.97265878e-01 -1.71859515e+00
4.39615071e-01 -9.06382680e-01 -5.12106061e-01 6.25757396e-01
-1.85420379e-01 3.23191583e-01 -1.31335467e-01 6.47429228e-01
-6.34229958e-01 -3.90158266e-01 6.20800734e-01 -3.08117960e-02
-1.86168432e-01 1.36333406e-01 -1.05440974e+00 1.08105433e+00
7.28042006e-01 -1.12153590e+00 -3.23522031e-01 -7.01126903e-02
5.15034795e-01 -1.57037988e-01 1.88947126e-01 -1.23960876e+00
1.55030340e-01 -2.81147897e-01 3.79947901e-01 -3.79396498e-01
-4.84339058e-01 -1.49000275e+00 5.64515471e-01 6.18671060e-01
-6.42657995e-01 2.80247211e-01 -1.56916417e-02 4.08450484e-01
-1.84724554e-01 -5.43484390e-01 5.64609766e-01 -4.35866475e-01
-1.74571320e-01 -2.40092978e-01 -8.68902087e-01 3.95989597e-01
1.04805350e+00 -2.60734379e-01 2.17189968e-01 -2.22676486e-01
-9.06183839e-01 -1.02807030e-01 4.03438807e-01 3.11479479e-01
3.61564189e-01 -1.53311980e+00 -5.38877130e-01 3.92340779e-01
5.56684673e-01 -1.88093826e-01 -1.79759502e-01 5.46413839e-01
-4.04669702e-01 5.01786292e-01 1.10734314e-01 -4.38756019e-01
-1.20648277e+00 7.97621965e-01 4.72759128e-01 -3.83929551e-01
-3.28889430e-01 3.37551326e-01 1.71523288e-01 -5.67561030e-01
5.97788930e-01 -7.91108251e-01 -5.21774888e-01 1.41036332e-01
3.55654478e-01 3.01421046e-01 2.99760401e-01 -7.26771772e-01
-4.73374337e-01 5.27599275e-01 4.07849662e-02 5.44384837e-01
8.94423604e-01 -2.39200413e-01 -2.22057812e-02 6.09449744e-01
9.84334767e-01 -1.35504559e-01 -3.39394569e-01 -2.44171634e-01
2.64662355e-01 -3.64455789e-01 -7.02346787e-02 -1.06342757e+00
-8.94619286e-01 2.56913394e-01 9.29104030e-01 4.48551893e-01
1.17269695e+00 1.33160744e-02 5.00510573e-01 2.07413658e-01
4.84058768e-01 -1.33408439e+00 -3.50897014e-01 3.52441728e-01
4.40134525e-01 -1.52849483e+00 -2.10395366e-01 -5.17690003e-01
-7.39024937e-01 7.70492435e-01 3.21715385e-01 2.12798968e-01
9.77835298e-01 3.84380341e-01 4.76896703e-01 -2.86056519e-01
-4.28676933e-01 -5.68710566e-02 2.73899198e-01 7.78578103e-01
5.59840500e-01 4.62136328e-01 -7.19051361e-01 1.18916345e+00
-2.08440885e-01 4.90766853e-01 3.35753471e-01 9.26521957e-01
-1.86579108e-01 -9.70013261e-01 -6.47774458e-01 1.09003842e+00
-9.06626165e-01 -2.10905075e-01 -1.35795563e-01 3.98494929e-01
1.18912876e+00 1.28175008e+00 1.12367444e-01 -3.82851779e-01
7.07873046e-01 2.45735109e-01 -1.38105065e-01 -9.44868863e-01
-8.12342405e-01 -2.73722112e-01 4.61450256e-02 -2.74757296e-01
-6.65194690e-01 -7.46854484e-01 -1.28906500e+00 -1.14083476e-02
-3.00155520e-01 4.74936813e-01 2.24773303e-01 9.68339622e-01
3.24243188e-01 5.93414664e-01 5.09445786e-01 1.81663617e-01
-5.15852392e-01 -9.32769001e-01 -5.92574120e-01 1.12022829e+00
4.20318007e-01 -9.70818400e-01 -5.55804372e-01 4.00495082e-01]
|
[8.01131534576416, 6.8083014488220215]
|
bf6ceb8e-83c6-4805-a7c0-330907b80aea
|
blind-image-deblurring-with-unknown-kernel
|
2208.09483
| null |
https://arxiv.org/abs/2208.09483v1
|
https://arxiv.org/pdf/2208.09483v1.pdf
|
Blind Image Deblurring with Unknown Kernel Size and Substantial Noise
|
Blind image deblurring (BID) has been extensively studied in computer vision and adjacent fields. Modern methods for BID can be grouped into two categories: single-instance methods that deal with individual instances using statistical inference and numerical optimization, and data-driven methods that train deep-learning models to deblur future instances directly. Data-driven methods can be free from the difficulty in deriving accurate blur models, but are fundamentally limited by the diversity and quality of the training data -- collecting sufficiently expressive and realistic training data is a standing challenge. In this paper, we focus on single-instance methods that remain competitive and indispensable. However, most such methods do not prescribe how to deal with unknown kernel size and substantial noise, precluding practical deployment. Indeed, we show that several state-of-the-art (SOTA) single-instance methods are unstable when the kernel size is overspecified, and/or the noise level is high. On the positive side, we propose a practical BID method that is stable against both, the first of its kind. Our method builds on the recent ideas of solving inverse problems by integrating the physical models and structured deep neural networks, without extra training data. We introduce several crucial modifications to achieve the desired stability. Extensive empirical tests on standard synthetic datasets, as well as real-world NTIRE2020 and RealBlur datasets, show the superior effectiveness and practicality of our BID method compared to SOTA single-instance as well as data-driven methods. The code of our method is available at: \url{https://github.com/sun-umn/Blind-Image-Deblurring}.
|
['Ju Sun', 'Hengkang Wang', 'Taihui Li', 'Zhong Zhuang']
|
2022-08-18
| null | null | null | null |
['blind-image-deblurring']
|
['computer-vision']
|
[-6.87024817e-02 -4.12469268e-01 2.52797380e-02 -1.54212657e-02
-7.65897036e-01 -4.16438311e-01 5.23683906e-01 -7.26742744e-01
-1.98461354e-01 9.22749162e-01 1.04135640e-01 -3.55626494e-01
-4.70884442e-01 -3.22445899e-01 -7.03176498e-01 -1.06855059e+00
8.84753540e-02 1.23303838e-01 1.12070166e-01 -1.21914662e-01
2.67692238e-01 3.97803694e-01 -1.34207249e+00 -1.77172825e-01
1.21562922e+00 1.09991384e+00 1.45218775e-01 6.50761664e-01
3.33420038e-01 7.78510928e-01 -3.35643709e-01 -2.01489151e-01
4.01048779e-01 -5.67100167e-01 -6.01467609e-01 7.96938464e-02
4.41351444e-01 -5.66969991e-01 -6.50042951e-01 1.20604587e+00
7.20451474e-01 -6.15413375e-02 5.35359740e-01 -1.06786525e+00
-9.89894450e-01 1.56030521e-01 -7.23445833e-01 2.95000404e-01
-1.09464183e-01 3.28833997e-01 4.92918402e-01 -9.94313538e-01
3.16124558e-01 7.08440006e-01 1.01273060e+00 5.67996562e-01
-1.19014204e+00 -4.91843522e-01 -4.95987684e-02 5.52399397e-01
-1.35656214e+00 -6.79734826e-01 7.75627673e-01 -5.57988763e-01
4.60363925e-01 2.44576558e-01 3.63246471e-01 1.13128412e+00
-4.57001776e-02 5.91857970e-01 1.51872575e+00 -4.68780071e-01
4.37729567e-01 2.76922155e-03 2.44663134e-01 4.78861600e-01
3.11800867e-01 4.64966893e-01 -4.63478774e-01 -2.28021890e-01
9.41861391e-01 -1.48764729e-01 -9.88490999e-01 -4.75906223e-01
-1.29928577e+00 5.63763857e-01 4.86772656e-01 2.24186331e-01
-3.42369407e-01 2.43827060e-01 1.55000404e-01 2.08873183e-01
6.24543250e-01 3.21558535e-01 -5.19125402e-01 -7.78046921e-02
-1.27106345e+00 3.11089933e-01 7.20959187e-01 8.31685364e-01
6.82057083e-01 3.77369560e-02 -1.09526768e-01 8.92463744e-01
1.15720734e-01 5.20598531e-01 5.36587775e-01 -1.13673687e+00
1.06788076e-01 -5.09947389e-02 5.47036886e-01 -7.25532293e-01
-1.81272671e-01 -5.46294153e-01 -1.30131555e+00 3.63780946e-01
6.61841810e-01 -2.66510040e-01 -9.72865283e-01 1.65013218e+00
1.57895461e-01 5.95558465e-01 -1.11482792e-01 1.41120231e+00
5.26060224e-01 4.95929122e-01 -6.24148726e-01 -4.37144309e-01
1.19025981e+00 -9.96413589e-01 -6.57248020e-01 -2.42990002e-01
2.13838145e-01 -7.99278259e-01 9.42514539e-01 6.93084717e-01
-1.03506207e+00 -3.83421630e-01 -1.06150579e+00 2.33453047e-02
-1.23983179e-03 3.22128713e-01 3.50601315e-01 6.43804967e-01
-1.27028716e+00 7.43799806e-01 -7.49957383e-01 -2.05684468e-01
3.80873859e-01 1.25210091e-01 -1.79943405e-02 -1.12848096e-01
-1.22775924e+00 8.49533379e-01 -6.81447983e-02 5.43982744e-01
-7.85995901e-01 -8.00093591e-01 -5.03764868e-01 -1.65871456e-01
3.92946392e-01 -8.26006114e-01 1.36177647e+00 -9.82170045e-01
-1.53228438e+00 6.07671142e-01 -2.47420058e-01 -5.85193753e-01
8.85490894e-01 -5.41373789e-01 -1.80778623e-01 7.58857131e-02
-1.68848097e-01 1.44440532e-01 1.36798882e+00 -1.43647087e+00
-5.12741059e-02 -2.82714516e-01 5.06004766e-02 -1.17551140e-01
-3.16111028e-01 -1.91243768e-01 -3.96444052e-01 -7.88910925e-01
8.47351849e-02 -9.43265557e-01 -2.10961238e-01 2.39468843e-01
-4.28759843e-01 2.78983206e-01 7.52435684e-01 -8.26204658e-01
1.14608920e+00 -2.18014669e+00 2.29739457e-01 -2.89987594e-01
2.37914190e-01 4.79091793e-01 8.31399634e-02 1.83990493e-01
-1.91897243e-01 -9.21087638e-02 -5.15531659e-01 -3.59172404e-01
4.19656187e-02 1.41856940e-02 -5.46047151e-01 8.16406608e-01
3.34361047e-02 8.41723919e-01 -6.84597850e-01 -1.46729887e-01
2.31665418e-01 6.00041807e-01 -3.94057274e-01 2.60799468e-01
-1.03341356e-01 5.20598173e-01 -2.47580335e-01 4.07509178e-01
9.91736114e-01 -4.56160635e-01 -1.42094776e-01 -4.97305781e-01
-1.29165083e-01 2.74828803e-02 -1.30818892e+00 1.46039426e+00
-4.52159226e-01 8.08342099e-01 4.08976585e-01 -1.15748227e+00
4.49422896e-01 3.68218511e-01 2.36297518e-01 -4.77969319e-01
4.35044058e-02 5.37661791e-01 -2.19217971e-01 -7.36646295e-01
3.69428158e-01 -2.43966252e-01 3.39839101e-01 4.82292205e-01
-2.20391512e-01 -1.52657405e-01 -6.12454452e-02 4.21199873e-02
9.65731025e-01 2.68222898e-01 1.74342409e-01 -4.52882379e-01
4.60596085e-01 -1.51064694e-01 5.15255570e-01 8.64073813e-01
-3.26747149e-01 1.01883984e+00 3.69764268e-01 -3.42748791e-01
-1.09169316e+00 -9.97386217e-01 -3.36245775e-01 4.80846643e-01
4.99596566e-01 3.57750542e-02 -1.02555799e+00 -4.28532571e-01
-2.02356517e-01 5.57036877e-01 -5.43910742e-01 -3.04678697e-02
-5.11087298e-01 -1.31479406e+00 3.78047854e-01 2.87336618e-01
6.40123546e-01 -6.79197371e-01 -4.89486635e-01 -9.53252986e-03
-2.98317522e-01 -1.16703939e+00 -5.51574528e-01 -7.47275278e-02
-8.02671015e-01 -1.06730759e+00 -1.15983248e+00 -5.27600706e-01
6.24344110e-01 5.01206160e-01 9.78282392e-01 8.37367922e-02
-2.61566222e-01 2.64246702e-01 -1.87400192e-01 -4.50814776e-02
-3.99265498e-01 -2.69764364e-01 2.29612529e-01 2.52451688e-01
3.39548662e-02 -7.91634560e-01 -9.47598875e-01 4.38021272e-01
-9.17534530e-01 1.84040442e-01 6.74222589e-01 1.09331429e+00
1.88663200e-01 1.06843963e-01 5.63949883e-01 -4.37472075e-01
5.89641333e-01 -3.62048298e-01 -7.44442582e-01 1.62860125e-01
-7.08824635e-01 1.82786211e-01 5.66962183e-01 -6.27686560e-01
-1.08850324e+00 -1.42832965e-01 1.13549218e-01 -6.48371935e-01
-1.64039865e-01 3.99369955e-01 8.78582150e-03 -2.05811664e-01
9.08572674e-01 4.75428998e-01 2.33870428e-02 -6.67747438e-01
3.62197787e-01 8.34759235e-01 7.48030066e-01 -6.55786037e-01
9.06224132e-01 5.90592086e-01 -1.46281421e-01 -8.19259524e-01
-8.21297646e-01 -2.98099518e-01 -4.57840741e-01 -2.62084544e-01
5.00336468e-01 -8.57055128e-01 -4.11552310e-01 1.21920323e+00
-1.24295616e+00 -6.28898740e-01 -8.64115730e-02 4.33725476e-01
-6.59213185e-01 6.79234147e-01 -6.87258184e-01 -7.35824585e-01
-3.16575676e-01 -1.31172538e+00 8.27771485e-01 2.49773070e-01
2.46080279e-01 -9.10281181e-01 -4.00837250e-02 4.69460547e-01
7.86027730e-01 -8.85756463e-02 5.92825770e-01 -3.58463190e-02
-7.75426269e-01 -1.67690236e-02 -6.72690630e-01 5.54388523e-01
2.01489851e-01 -1.63399398e-01 -1.11205959e+00 -4.30621415e-01
4.59379137e-01 -2.91863352e-01 9.46680844e-01 6.70640051e-01
1.31640184e+00 -4.39606160e-01 -6.39474913e-02 8.19281697e-01
1.62320459e+00 -3.40041876e-01 7.59470284e-01 3.23155165e-01
4.83221203e-01 3.22378606e-01 2.03793749e-01 3.35583985e-01
2.13833496e-01 8.90391409e-01 4.99116927e-01 -1.89124048e-01
-2.99208999e-01 2.84350127e-01 3.47175211e-01 5.67635179e-01
-3.25469553e-01 -1.32234707e-01 -7.49420524e-01 5.97494781e-01
-1.96472561e+00 -9.20275450e-01 -4.65429813e-01 2.34416890e+00
1.10933995e+00 -4.83404249e-02 -6.11645766e-02 1.76819891e-01
8.30815732e-01 1.13120094e-01 -7.44519830e-01 1.76067084e-01
-8.65813866e-02 -1.68893598e-02 4.66654748e-01 6.08498931e-01
-1.10743737e+00 5.98148406e-01 5.21004057e+00 8.20806980e-01
-1.22186458e+00 3.70717496e-01 6.65755093e-01 -1.76047713e-01
8.92288163e-02 -6.85873535e-03 -4.46236163e-01 6.60779774e-01
7.16544926e-01 -2.58839522e-02 8.34576488e-01 5.45502722e-01
3.80325079e-01 -2.27962956e-01 -8.80900443e-01 1.14960015e+00
1.76673308e-02 -1.38146305e+00 -4.65596676e-01 4.20432836e-02
8.24101508e-01 2.16853261e-01 1.00075811e-01 -4.07817774e-02
-4.52296697e-02 -8.04201841e-01 8.87983501e-01 7.10714817e-01
8.55949700e-01 -1.47424519e-01 6.93498969e-01 4.48272794e-01
-5.72759211e-01 3.75444070e-02 -2.31018379e-01 -1.12751447e-01
3.12360883e-01 1.21583974e+00 -2.95069963e-02 7.03364909e-01
9.45572555e-01 7.66930699e-01 -3.99114609e-01 1.38431776e+00
-1.79885998e-01 7.87967920e-01 -2.40354657e-01 2.84759790e-01
-2.32169013e-02 -3.26714873e-01 7.47778237e-01 9.95965838e-01
5.46203256e-01 3.39780748e-02 -3.82772416e-01 1.12624133e+00
1.17494121e-01 -4.32193786e-01 -2.57939637e-01 2.89978087e-01
2.40577653e-01 1.02662730e+00 -3.81127149e-01 -2.14327052e-01
-4.66531962e-01 1.24906659e+00 1.98728770e-01 8.03041518e-01
-1.05098391e+00 -2.30073854e-01 7.88714111e-01 1.47842064e-01
3.46805751e-01 -5.19481242e-01 -3.89975995e-01 -1.51680100e+00
3.64649534e-01 -9.30732787e-01 -6.08140007e-02 -1.02404451e+00
-1.63022578e+00 5.77116311e-01 -1.29790410e-01 -1.31672013e+00
-3.57397133e-03 -8.22720945e-01 -4.55447376e-01 9.24221694e-01
-1.73299086e+00 -8.93731892e-01 -6.12119138e-01 5.67616105e-01
3.97239536e-01 1.85682982e-01 4.68722492e-01 3.84191841e-01
-6.87390566e-01 3.28279883e-01 4.84239966e-01 4.02744077e-02
9.62842643e-01 -1.14352429e+00 2.72387922e-01 1.12490213e+00
-1.25858719e-02 6.52026594e-01 9.54071760e-01 -2.53121197e-01
-1.45617211e+00 -8.90860558e-01 5.31726718e-01 -3.74556631e-01
9.49464679e-01 -2.91071177e-01 -1.13808906e+00 3.30210835e-01
1.79274365e-01 4.62528765e-01 5.07234819e-02 -1.76145807e-01
-3.17845762e-01 -1.45453468e-01 -8.87665451e-01 4.74215508e-01
1.00403178e+00 -4.54670846e-01 -4.64348227e-01 4.09885079e-01
4.19766784e-01 -5.42587996e-01 -5.01686335e-01 5.35562813e-01
3.79263371e-01 -1.46269107e+00 1.19553673e+00 -3.23393315e-01
5.52880108e-01 -3.66334945e-01 4.42347229e-02 -1.38543713e+00
-2.00574771e-01 -8.59842539e-01 -5.56230307e-01 9.71113622e-01
8.33031237e-02 -9.16058242e-01 5.09754419e-01 4.63799715e-01
-6.58891648e-02 -8.08768630e-01 -9.82695699e-01 -1.08613241e+00
2.61433214e-01 -4.86600012e-01 3.76167536e-01 9.18252468e-01
-3.79657567e-01 1.03687070e-01 -6.99140608e-01 5.27317643e-01
1.02736700e+00 2.21934304e-01 5.55338025e-01 -9.63902295e-01
-5.36675274e-01 -5.04548371e-01 -1.56355649e-01 -1.37870145e+00
-1.04401305e-01 -3.45862597e-01 2.89514661e-01 -1.43022835e+00
1.95316613e-01 -5.86915493e-01 -4.60293628e-02 2.62961626e-01
-3.41100037e-01 3.65466684e-01 -5.09327687e-02 5.31117558e-01
-1.86179221e-01 6.37378931e-01 1.17370260e+00 2.31598262e-02
-9.34170112e-02 1.05048686e-01 -5.39507091e-01 6.01126373e-01
8.13686311e-01 -2.35929936e-01 -3.15328628e-01 -5.94000161e-01
2.44252935e-01 5.55711575e-02 1.11694860e+00 -1.06263900e+00
3.63840759e-01 -2.00565994e-01 8.77961963e-02 -1.13125809e-01
2.99948066e-01 -6.10273778e-01 1.59537122e-01 3.10636193e-01
5.96129224e-02 -4.19397831e-01 1.15226455e-01 5.60542285e-01
-4.60974611e-02 -2.78189570e-01 1.07744312e+00 2.60902867e-02
-6.15357280e-01 2.99893051e-01 -1.93030640e-01 2.31782168e-01
6.27846718e-01 -7.75266960e-02 -7.14289308e-01 -6.10627353e-01
-5.82356274e-01 -1.00239910e-01 6.51919842e-01 2.93477297e-01
3.61632973e-01 -1.11212909e+00 -8.01722646e-01 1.92659661e-01
-1.55669376e-01 -2.37842239e-02 3.89072984e-01 1.42505920e+00
-4.31041539e-01 1.93315104e-01 1.63293496e-01 -5.72167695e-01
-8.12645674e-01 7.10373819e-01 7.14865029e-01 1.00415029e-01
-6.99300170e-01 7.59179235e-01 2.69331902e-01 -1.85793594e-01
1.12421200e-01 -1.85311645e-01 4.05140996e-01 -3.10539484e-01
6.06364489e-01 3.51997346e-01 1.66158989e-01 -4.01576519e-01
-2.32242018e-01 6.41070247e-01 9.81726646e-02 -1.80885270e-02
1.31942368e+00 -3.32077116e-01 -3.48151922e-01 1.29921913e-01
1.02966940e+00 -5.42918071e-02 -1.54053509e+00 -4.04210657e-01
-1.92920670e-01 -4.18026745e-01 3.91852587e-01 -8.49887192e-01
-1.20915771e+00 9.25555468e-01 6.98009610e-01 3.95605356e-01
1.35361814e+00 -3.60652246e-02 8.44749391e-01 8.13200250e-02
3.84718716e-01 -7.66800880e-01 7.17825592e-02 3.74684781e-01
1.12812626e+00 -1.19894838e+00 6.79057539e-02 -2.97773421e-01
-2.34415740e-01 9.97188985e-01 4.68033880e-01 -1.54005483e-01
8.49103630e-01 3.06509823e-01 2.21424624e-01 -8.21743719e-03
-3.73252451e-01 -1.38602208e-03 3.23403180e-01 3.27753961e-01
1.48335159e-01 -3.35129589e-01 -2.13781789e-01 5.93997657e-01
8.74955729e-02 4.58766729e-01 6.19870603e-01 6.44344211e-01
-2.11506933e-01 -8.57425034e-01 -7.08869517e-01 2.64707983e-01
-2.90151566e-01 -3.53771627e-01 -9.97329578e-02 4.96755421e-01
1.06484918e-02 7.81902850e-01 -3.17780018e-01 -4.75260951e-02
5.45479767e-02 -1.11617625e-01 5.46744823e-01 -7.55090043e-02
3.67539586e-03 -6.26480579e-02 -1.52656525e-01 -4.62407351e-01
-6.17014110e-01 -6.25773311e-01 -7.68352270e-01 -3.49383056e-01
-5.53466022e-01 -4.01396304e-02 4.77147520e-01 1.07605314e+00
4.67785060e-01 2.50735551e-01 5.10296881e-01 -1.28795683e+00
-8.73100281e-01 -9.68734980e-01 -4.63648826e-01 1.02379404e-01
9.93354321e-01 -7.11533606e-01 -8.80529523e-01 3.36842924e-01]
|
[11.606806755065918, -2.640920639038086]
|
cac3bc1c-f227-4b8c-80fd-263594955cb5
|
zerokbc-a-comprehensive-benchmark-for-zero
|
2212.03091
| null |
https://arxiv.org/abs/2212.03091v1
|
https://arxiv.org/pdf/2212.03091v1.pdf
|
ZeroKBC: A Comprehensive Benchmark for Zero-Shot Knowledge Base Completion
|
Knowledge base completion (KBC) aims to predict the missing links in knowledge graphs. Previous KBC tasks and approaches mainly focus on the setting where all test entities and relations have appeared in the training set. However, there has been limited research on the zero-shot KBC settings, where we need to deal with unseen entities and relations that emerge in a constantly growing knowledge base. In this work, we systematically examine different possible scenarios of zero-shot KBC and develop a comprehensive benchmark, ZeroKBC, that covers these scenarios with diverse types of knowledge sources. Our systematic analysis reveals several missing yet important zero-shot KBC settings. Experimental results show that canonical and state-of-the-art KBC systems cannot achieve satisfactory performance on this challenging benchmark. By analyzing the strength and weaknesses of these systems on solving ZeroKBC, we further present several important observations and promising future directions.
|
['Jianshu Chen', 'Dong Yu', 'Dian Yu', 'Xiaoman Pan', 'Hongming Zhang', 'Wenlin Yao', 'Pei Chen']
|
2022-12-06
| null | null | null | null |
['knowledge-base-completion', 'knowledge-base-completion']
|
['graphs', 'knowledge-base']
|
[-1.32525638e-01 4.99123961e-01 -5.93484044e-01 -2.13932265e-02
-5.10098994e-01 -3.93654495e-01 5.38112104e-01 2.06718609e-01
-2.33746752e-01 1.21613169e+00 9.06831324e-02 -3.34370762e-01
-5.21297157e-01 -9.43815947e-01 -7.43311107e-01 -2.56975085e-01
-4.20486152e-01 9.09351587e-01 7.31699765e-01 -6.51019573e-01
-1.42833188e-01 -4.63123545e-02 -1.28081143e+00 3.40124577e-01
9.42485929e-01 3.36483061e-01 -1.37344614e-01 4.82371897e-01
-1.57612324e-01 1.05653107e+00 -5.78352034e-01 -1.22147954e+00
-2.03108806e-02 -1.18463948e-01 -1.36461151e+00 -5.19424677e-01
3.35091084e-01 -1.16331324e-01 -6.13315046e-01 1.15386498e+00
4.57675606e-01 1.91350147e-01 6.38005257e-01 -1.42880285e+00
-1.05162132e+00 1.16012776e+00 -2.56266087e-01 5.60177505e-01
5.23757100e-01 -2.19780430e-01 1.33199859e+00 -1.08001101e+00
1.11977911e+00 8.86832833e-01 8.01917434e-01 7.15088069e-01
-9.30427611e-01 -4.52609897e-01 2.20119730e-01 1.00753117e+00
-1.74855220e+00 -4.21311229e-01 4.32083577e-01 -2.04079181e-01
1.41325259e+00 1.02604814e-01 5.89020371e-01 1.21067536e+00
-4.96058106e-01 6.77820683e-01 5.41007757e-01 -7.42591679e-01
2.24447355e-01 1.78684890e-01 7.65868306e-01 6.61397755e-01
8.56732190e-01 -4.24449593e-01 -8.87699902e-01 -2.51072317e-01
3.17648500e-01 -5.00174403e-01 -6.67696595e-01 -3.66986632e-01
-1.08493006e+00 5.76902211e-01 1.91432491e-01 1.90841794e-01
-1.50006488e-01 -3.94478999e-03 9.44319591e-02 2.69483179e-01
3.03251147e-01 6.17093325e-01 -7.92920351e-01 -6.22788779e-02
-4.45829540e-01 1.86471418e-01 1.51386988e+00 1.77879333e+00
8.72641921e-01 -3.71703595e-01 -2.51757681e-01 7.19312906e-01
-1.38362810e-01 3.01752955e-01 1.57686636e-01 -6.19210303e-01
1.00779295e+00 6.61804438e-01 2.97731429e-01 -1.07543075e+00
-2.99353004e-01 -3.58524442e-01 -5.46804547e-01 -6.56622231e-01
1.56744197e-01 -2.44403049e-01 -8.78650010e-01 1.44349015e+00
4.44127172e-01 7.62936771e-01 6.10944867e-01 7.36153662e-01
1.32219398e+00 3.72500539e-01 9.21127722e-02 -1.91202730e-01
1.26491404e+00 -1.10953498e+00 -8.61623883e-01 -2.36038625e-01
8.75647426e-01 -1.45615518e-01 9.58589017e-01 -1.33900866e-01
-8.15598786e-01 1.04446076e-01 -1.08284533e+00 -1.13691039e-01
-1.03255844e+00 -3.29481602e-01 9.16040182e-01 6.14153445e-01
-9.89885032e-01 4.66039687e-01 -4.38684821e-01 -5.62315762e-01
1.31919369e-01 1.38284385e-01 -3.79787952e-01 -5.97932160e-01
-2.06683993e+00 1.36679864e+00 8.18028688e-01 1.11330323e-01
-8.14023554e-01 -9.93740141e-01 -8.81667256e-01 4.28952813e-01
1.36342967e+00 -6.67008638e-01 1.07020867e+00 -2.22195890e-02
-8.16848993e-01 4.68638122e-01 -8.47039297e-02 -4.31444973e-01
-5.37822442e-03 -2.91431755e-01 -8.51162612e-01 8.23786557e-02
2.48248845e-01 2.49092475e-01 2.24823937e-01 -1.38433337e+00
-5.71583986e-01 -1.30838275e-01 3.33166301e-01 2.56709844e-01
-4.04345840e-01 -2.22363338e-01 -1.25732911e+00 -2.93524235e-01
-3.77657533e-01 -7.04339921e-01 7.21725821e-02 -7.38005757e-01
-6.00109160e-01 -3.00369084e-01 6.64758980e-01 -4.32623476e-01
1.63648832e+00 -1.57835436e+00 1.94160059e-01 3.20635796e-01
2.01039538e-01 5.03134251e-01 -1.79286122e-01 8.09999883e-01
3.20411138e-02 3.66150230e-01 4.21916731e-02 -9.69623327e-02
-3.93234305e-02 5.59852362e-01 -3.35584193e-01 -6.28232360e-02
2.26397440e-01 1.43601322e+00 -1.19947267e+00 -7.84046531e-01
-3.29554081e-01 1.96841836e-01 -2.58821517e-01 -3.98979366e-01
-2.81942248e-01 -2.79031873e-01 -5.02604544e-01 1.11568069e+00
3.76298666e-01 -8.08167279e-01 6.61937833e-01 -5.25199100e-02
4.79206115e-01 -3.18791717e-02 -1.08554709e+00 1.42827594e+00
1.37052506e-01 4.18435395e-01 -3.83218914e-01 -1.09279406e+00
3.41475695e-01 4.19887334e-01 1.20586313e-01 -4.19571936e-01
-3.20482135e-01 1.32747427e-01 1.77030510e-03 -5.37895858e-01
7.30767906e-01 -1.04415476e-01 -4.90705892e-02 2.25558709e-02
6.73989415e-01 2.45828986e-01 5.97692370e-01 9.21666265e-01
1.52396905e+00 -3.22474271e-01 6.17519081e-01 -3.73520292e-02
2.64686406e-01 4.06900287e-01 6.42357886e-01 1.10421145e+00
-1.66302100e-01 2.01286688e-01 5.99427104e-01 -1.52296513e-01
-6.33132875e-01 -7.89409399e-01 2.01766506e-01 8.74732375e-01
5.33633113e-01 -1.08615839e+00 -3.40174496e-01 -9.74088550e-01
1.85779631e-01 9.06438112e-01 -8.67371500e-01 -3.21118087e-01
-2.42883474e-01 -7.47859299e-01 6.25196874e-01 4.85756606e-01
1.41355574e-01 -9.03804481e-01 2.27565959e-01 1.93277404e-01
-4.90181804e-01 -1.39915156e+00 8.33720416e-02 7.07805203e-03
-4.05888289e-01 -1.65047717e+00 -4.62158531e-01 -7.32520163e-01
5.00073373e-01 5.06521463e-01 1.54965580e+00 1.57302037e-01
-4.10318106e-01 8.53999197e-01 -7.85208225e-01 -4.93667573e-01
3.70823517e-02 3.65587801e-01 -6.06514141e-02 -4.89122331e-01
6.24325573e-01 -4.79392081e-01 -4.18047965e-01 1.50244102e-01
-4.88694757e-01 -3.21399942e-02 4.02198166e-01 7.41501272e-01
3.43137562e-01 4.74267811e-01 9.47702289e-01 -1.37961936e+00
7.85201371e-01 -8.70208561e-01 -2.36123845e-01 1.06920528e+00
-1.06642365e+00 -4.47525494e-02 3.45728129e-01 -3.12045723e-01
-1.27855456e+00 -4.35242236e-01 3.67831379e-01 -3.27749431e-01
4.59114254e-01 1.14318562e+00 -6.30726144e-02 -2.87426803e-02
6.90300345e-01 7.29647726e-02 -7.08985388e-01 -3.34027588e-01
7.94846237e-01 2.18188569e-01 6.17630541e-01 -7.90510178e-01
7.29877591e-01 3.50172400e-01 -1.01626903e-01 -6.51019990e-01
-1.37831557e+00 -9.79091883e-01 -6.94774270e-01 -2.43428826e-01
5.07618487e-01 -1.12490833e+00 -5.32332122e-01 1.61441296e-01
-1.10474932e+00 -2.86922216e-01 -4.30870056e-01 2.90953428e-01
-2.39439905e-01 3.85636806e-01 -5.73152244e-01 -6.46825254e-01
-3.97172540e-01 -4.16743100e-01 7.16830015e-01 2.15931594e-01
5.88980503e-03 -1.20547259e+00 2.86981046e-01 3.67039114e-01
1.45635203e-01 -8.37871134e-02 1.25449777e+00 -7.78765142e-01
-7.47927964e-01 -2.00875580e-01 -1.15705743e-01 -1.19337797e-01
-5.90291023e-02 -1.29104137e-01 -6.38879657e-01 -7.11053461e-02
-8.58904243e-01 -5.99706113e-01 1.17454314e+00 -2.20415056e-01
7.38599837e-01 -1.54321149e-01 -7.68186271e-01 2.87239075e-01
1.55830312e+00 -9.49122161e-02 7.76644349e-01 2.37788990e-01
6.63488865e-01 4.59565133e-01 6.40621364e-01 2.83634335e-01
8.58610213e-01 4.40911978e-01 3.18318665e-01 2.82122135e-01
-1.91160932e-01 -3.94873261e-01 -2.09748000e-01 9.23122525e-01
-6.27123237e-01 -5.08037627e-01 -1.30260456e+00 9.43293035e-01
-2.34747171e+00 -1.04883814e+00 -2.81161875e-01 1.85484564e+00
1.28745329e+00 1.49359196e-01 -5.67323983e-01 -2.19713673e-01
9.02392864e-01 -1.11172162e-01 -4.88756448e-01 4.56240565e-01
-4.33137506e-01 3.02864760e-01 4.58185315e-01 3.96143317e-01
-9.54791248e-01 1.42537308e+00 6.93733358e+00 8.65298092e-01
-1.76884979e-01 1.05863184e-01 -3.03005725e-02 -5.74534498e-02
-2.59643137e-01 3.79825801e-01 -1.26977444e+00 8.28148574e-02
8.28142107e-01 -5.56317985e-01 3.65642518e-01 8.98780346e-01
-6.88125134e-01 -1.44344345e-01 -1.12034857e+00 8.24250102e-01
2.64215142e-01 -1.60979724e+00 2.03848496e-01 -1.89073980e-01
1.05427980e+00 8.51324666e-03 -4.12827909e-01 1.18535388e+00
8.05910707e-01 -7.95520663e-01 1.34515300e-01 8.28425229e-01
7.47832894e-01 -4.18668598e-01 8.09373021e-01 3.75008970e-01
-1.09315741e+00 5.23840524e-02 -5.90898693e-01 -6.57105818e-02
2.53863871e-01 5.09043157e-01 -1.08267510e+00 1.10792196e+00
8.38349223e-01 7.78301120e-01 -8.30192506e-01 8.39048266e-01
-4.75147963e-01 6.89606190e-01 -4.81698178e-02 -1.20433129e-01
-6.64323270e-02 2.06104994e-01 2.48142660e-01 1.22917724e+00
9.17154327e-02 5.00632763e-01 -8.21976662e-02 5.13977110e-01
-4.34119344e-01 -3.53968740e-02 -6.43099010e-01 -1.08536728e-01
9.17755723e-01 1.23878682e+00 -4.90507185e-01 -7.60416448e-01
-9.02599692e-01 8.31210554e-01 1.13907146e+00 7.67792940e-01
-8.51522326e-01 -4.21637982e-01 4.13208842e-01 -2.64558852e-01
4.08461690e-01 2.07066163e-02 2.22825065e-01 -1.61240208e+00
1.15915939e-01 -3.12250286e-01 7.82712936e-01 -7.95876741e-01
-1.71525896e+00 6.20225444e-02 3.18672359e-01 -6.70386791e-01
-7.26530254e-02 -4.79125082e-01 -4.85078752e-01 5.76977015e-01
-1.78807533e+00 -1.09136665e+00 -4.50909376e-01 8.49642098e-01
1.79079115e-01 -1.07026182e-01 1.10699069e+00 3.25896859e-01
-7.50391722e-01 3.85138571e-01 2.20907211e-01 2.12468550e-01
8.92642260e-01 -1.42083979e+00 4.72567886e-01 8.07743132e-01
1.28063038e-01 9.68605638e-01 4.74192530e-01 -1.14904964e+00
-1.55819261e+00 -1.18159509e+00 9.84385312e-01 -7.80519903e-01
1.02895546e+00 -3.05526644e-01 -1.12693799e+00 1.08576190e+00
-2.38473136e-02 3.33549738e-01 9.19375300e-01 9.31725562e-01
-5.69753766e-01 8.51091444e-02 -5.40303051e-01 6.97865665e-01
1.40348828e+00 -5.44275045e-01 -8.62737298e-01 3.38569611e-01
9.51561391e-01 -1.93695962e-01 -1.06824517e+00 6.41289830e-01
3.64329696e-01 -3.55440170e-01 1.11254501e+00 -1.31077731e+00
3.04344893e-01 -2.26516888e-01 7.91911501e-03 -1.38157868e+00
-5.14420986e-01 -3.10074538e-01 -1.00157583e+00 1.30819452e+00
8.99752080e-01 -3.41743141e-01 9.46298718e-01 9.25402641e-01
-1.33333923e-02 -7.39535987e-01 -7.01372027e-01 -9.84038472e-01
-2.05547959e-01 -2.99924463e-01 5.06003976e-01 1.55448639e+00
6.62077367e-01 8.57343495e-01 -6.10873044e-01 2.63716161e-01
4.92454290e-01 5.46108708e-02 7.99668193e-01 -1.39634573e+00
-1.95697948e-01 1.71504140e-01 -2.28733584e-01 -5.95763862e-01
2.91067868e-01 -9.75219965e-01 -2.26298511e-01 -1.99583483e+00
7.02410758e-01 -3.79376769e-01 -3.25736463e-01 8.08929682e-01
-6.95087075e-01 -1.17344469e-01 5.36672352e-03 3.33717316e-01
-1.23221612e+00 5.38224936e-01 1.14895785e+00 -3.75796884e-01
-2.49246769e-02 -3.85145515e-01 -8.82921338e-01 4.28893298e-01
5.99452436e-01 -2.08408877e-01 -8.92776132e-01 -2.21803397e-01
7.77076900e-01 2.05392130e-02 2.54934549e-01 -5.35809278e-01
8.48600805e-01 -2.29275435e-01 1.58277094e-01 -5.42400599e-01
4.03928816e-01 -5.54316640e-01 1.61103368e-01 -1.15098365e-01
-1.56552773e-02 -4.72224653e-01 -9.15005654e-02 1.11203611e+00
-2.46941119e-01 -5.06801844e-01 2.19990127e-02 -3.39924663e-01
-1.49119985e+00 3.05950671e-01 7.26756454e-02 5.62777936e-01
1.11833858e+00 -1.10864770e-02 -1.18820679e+00 -3.83405089e-01
-9.87046123e-01 7.75296986e-01 5.26396036e-02 3.83670807e-01
7.08231330e-01 -1.35214889e+00 -5.84406555e-01 -4.48261291e-01
7.47651398e-01 2.08828747e-01 3.31095308e-01 7.57944286e-01
-9.14937332e-02 6.57906592e-01 1.80782482e-01 2.10103974e-01
-1.10547280e+00 9.63897109e-01 -6.60445215e-03 -5.41105330e-01
-6.30029976e-01 8.39956582e-01 -2.07868472e-01 -2.67863095e-01
2.94031262e-01 2.13513486e-02 -4.12130177e-01 7.76869059e-02
4.80803847e-01 5.22756815e-01 3.20441514e-01 -1.84210584e-01
-5.83706081e-01 7.07445666e-02 -4.04351860e-01 2.20734313e-01
1.27021086e+00 -5.93749359e-02 -1.90541372e-02 3.34798753e-01
6.36747837e-01 -1.75086111e-01 -6.66325986e-01 -5.74331760e-01
4.89182770e-01 -1.43124521e-01 -3.71150434e-01 -1.03385258e+00
-7.65625060e-01 3.59053850e-01 -4.21956658e-01 1.67313308e-01
6.69168472e-01 4.92075086e-01 3.67392153e-01 9.97596323e-01
8.04377615e-01 -1.31352842e+00 -9.96203795e-02 9.29951787e-01
7.28255033e-01 -1.24777615e+00 2.59769112e-01 -9.53883886e-01
-8.57293665e-01 7.09610403e-01 9.88111079e-01 4.23765510e-01
9.27294075e-01 -1.39935566e-02 -5.40293872e-01 -7.25614429e-01
-1.24869597e+00 -7.40710139e-01 2.40124330e-01 9.01914001e-01
2.89682038e-02 6.94659576e-02 -3.15498233e-01 1.06449485e+00
-9.91503224e-02 -3.58575955e-02 6.04862869e-01 1.19818211e+00
-3.40892881e-01 -9.34262097e-01 9.51567516e-02 5.36050379e-01
-1.99831322e-01 -5.32377124e-01 -8.55207860e-01 1.16871428e+00
-5.83829358e-02 9.40086782e-01 -4.95852321e-01 -4.05583769e-01
5.26894391e-01 4.73934561e-01 5.02151489e-01 -8.55870843e-01
5.62739857e-02 -5.89153171e-01 6.91752434e-01 -3.21047604e-01
-4.27336484e-01 -1.90514192e-01 -1.16305208e+00 -2.47271672e-01
-8.71264338e-01 3.46654743e-01 1.36325046e-01 7.67568409e-01
4.93518919e-01 4.81150866e-01 -3.14365000e-01 -8.96130577e-02
-3.26441407e-01 -1.06055057e+00 -9.28030074e-01 4.74401593e-01
-7.57218972e-02 -1.02771866e+00 -1.80753060e-02 1.67195871e-01]
|
[8.930124282836914, 8.08023452758789]
|
fd70e0d9-a374-41d9-b6f8-f6d5c0afef82
|
joint-chinese-word-segmentation-and-span
|
2211.01638
| null |
https://arxiv.org/abs/2211.01638v2
|
https://arxiv.org/pdf/2211.01638v2.pdf
|
Joint Chinese Word Segmentation and Span-based Constituency Parsing
|
In constituency parsing, span-based decoding is an important direction. However, for Chinese sentences, because of their linguistic characteristics, it is necessary to utilize other models to perform word segmentation first, which introduces a series of uncertainties and generally leads to errors in the computation of the constituency tree afterward. This work proposes a method for joint Chinese word segmentation and Span-based Constituency Parsing by adding extra labels to individual Chinese characters on the parse trees. Through experiments, the proposed algorithm outperforms the recent models for joint segmentation and constituency parsing on CTB 5.1.
|
['Cong Liu', 'Tianyu Shi', 'Zhicheng Wang']
|
2022-11-03
| null | null | null | null |
['constituency-parsing', 'chinese-word-segmentation']
|
['natural-language-processing', 'natural-language-processing']
|
[-2.55037230e-02 -1.02734543e-01 -9.00404751e-02 -7.37435579e-01
-9.59896386e-01 -7.36348212e-01 -6.33255094e-02 4.51764494e-01
-4.97229755e-01 9.21206892e-01 3.61892432e-01 -9.14827526e-01
5.27753532e-01 -9.14701998e-01 -2.62010098e-01 -3.98473918e-01
3.79559904e-01 2.93348908e-01 4.80853945e-01 -1.55737147e-01
4.04432952e-01 1.41583726e-01 -5.85816622e-01 8.28094482e-02
1.01558328e+00 4.69059289e-01 5.14000058e-01 6.96856916e-01
-8.83771062e-01 2.26465493e-01 -8.27509820e-01 -4.21757102e-01
5.96679701e-03 -5.83401144e-01 -8.95843625e-01 1.25646591e-04
-3.12883127e-03 -3.30886006e-01 2.89366692e-01 1.28082561e+00
9.63658094e-02 -1.84761271e-01 2.51762301e-01 -4.08974290e-01
-3.87750298e-01 1.30448735e+00 -4.12597567e-01 3.44212770e-01
1.04767293e-01 -3.65651667e-01 1.16143572e+00 -4.86634016e-01
3.68414789e-01 1.32147813e+00 4.45543915e-01 6.05152249e-01
-7.43929863e-01 -6.58712029e-01 7.02698290e-01 -2.67159581e-01
-1.02147853e+00 -1.21865265e-01 8.10743570e-01 4.05728742e-02
1.00498414e+00 2.75675833e-01 5.85697651e-01 5.28708458e-01
3.34945947e-01 7.92290568e-01 1.35830748e+00 -6.93976223e-01
1.78418189e-01 -1.94930121e-01 9.11229193e-01 4.51214612e-01
4.08425868e-01 -2.95301914e-01 6.33836091e-02 8.04799646e-02
5.91194987e-01 -3.56818765e-01 4.80525978e-02 7.31883943e-01
-8.01802695e-01 7.88372397e-01 -6.54452816e-02 6.31100237e-01
-2.70099305e-02 1.78423107e-01 3.73430103e-01 -2.14372501e-01
5.25309563e-01 1.26077235e-01 -5.35954118e-01 -2.45840028e-01
-1.03831184e+00 2.41999961e-02 6.52524769e-01 1.20203829e+00
6.43443823e-01 -3.33344005e-02 -5.70546389e-02 5.41306496e-01
4.32337701e-01 2.28029981e-01 1.80501208e-01 -4.91222322e-01
8.54666471e-01 5.26760340e-01 9.03202919e-04 -5.19001961e-01
-3.58181745e-01 -2.43098229e-01 -3.41793686e-01 -3.08048636e-01
5.54389238e-01 -4.39393252e-01 -1.22727978e+00 1.60597837e+00
3.06227654e-01 -1.22292243e-01 2.26932064e-01 7.29204118e-01
7.67204344e-01 1.06528342e+00 6.90476537e-01 -4.72037256e-01
1.59256434e+00 -6.56433225e-01 -1.03662503e+00 -4.10472512e-01
5.74862540e-01 -1.17762399e+00 8.73429596e-01 2.49883339e-01
-1.05311823e+00 -4.36766148e-01 -8.61448109e-01 -2.15928182e-01
-1.38738334e-01 2.18342230e-01 6.26277685e-01 1.03843856e+00
-8.01057041e-01 1.84049934e-01 -1.05803335e+00 -1.96524691e-02
-1.40204668e-01 1.79704592e-01 5.75813651e-02 3.51657416e-03
-1.36942673e+00 8.26477706e-01 8.07424188e-01 3.93543661e-01
-3.59548688e-01 2.84095034e-02 -7.61509717e-01 1.29599363e-01
2.04746395e-01 1.67043239e-01 1.23751414e+00 -4.74882722e-01
-1.35104764e+00 5.50410330e-01 -3.73517811e-01 -2.94710416e-02
1.70622438e-01 -3.14280152e-01 -2.74409175e-01 -2.38513481e-02
1.63117871e-01 4.59214509e-01 3.72406900e-01 -1.12934840e+00
-9.77834463e-01 -3.58595610e-01 8.02573189e-02 2.65489817e-01
4.97477613e-02 4.59283859e-01 -7.99441636e-01 -8.37285638e-01
6.57759190e-01 -7.17717350e-01 -6.78670168e-01 -1.05015755e+00
-4.50052023e-01 -5.34160793e-01 5.30863047e-01 -1.27953243e+00
1.73095727e+00 -2.04811835e+00 -3.05927396e-01 4.99140918e-02
-1.32595867e-01 1.90949231e-01 3.56612206e-02 1.92098692e-01
1.05123311e-01 5.88491261e-01 -3.64611238e-01 -4.84996766e-01
-1.82920903e-01 4.85821962e-01 -5.20931892e-02 1.28649771e-01
2.62013972e-01 7.54867435e-01 -6.94514751e-01 -9.44725573e-01
1.68015610e-03 3.65712345e-02 -2.96558917e-01 1.69850305e-01
-3.23414177e-01 4.66068178e-01 -7.92147517e-01 8.56654227e-01
1.03275526e+00 4.74722773e-01 5.80458939e-01 6.67984486e-02
-6.26471579e-01 8.88282359e-01 -1.08435011e+00 1.44747412e+00
-3.41198444e-01 1.72536373e-01 1.51038364e-01 -9.75986779e-01
1.03447664e+00 3.62908602e-01 -2.43621990e-02 -2.20886737e-01
5.46955526e-01 3.93203139e-01 1.45121455e-01 -2.21355706e-01
9.82621312e-01 -3.04546326e-01 -8.22753727e-01 2.98235387e-01
-2.04585969e-01 -4.01801050e-01 4.56980765e-01 4.02145181e-03
6.78375483e-01 1.04809426e-01 3.27869445e-01 -4.39076602e-01
4.01615292e-01 3.38373363e-01 1.25692165e+00 2.96825051e-01
-4.18322384e-01 4.95904088e-01 5.80800533e-01 -1.93178654e-01
-9.03937697e-01 -9.89673436e-01 -2.67921150e-01 8.73455703e-01
2.58957893e-01 -3.34801823e-01 -1.00651193e+00 -8.20325792e-01
-7.40253806e-01 1.08509195e+00 -8.00729468e-02 3.36662501e-01
-1.41118336e+00 -9.26826894e-01 5.72588265e-01 7.15963125e-01
5.41036487e-01 -1.01163399e+00 -3.60293984e-01 8.83688271e-01
-5.61021984e-01 -1.36352968e+00 -5.57892621e-01 4.45575505e-01
-1.23861456e+00 -8.30553114e-01 -4.99562562e-01 -1.18225849e+00
5.57734013e-01 3.18928882e-02 9.04845655e-01 3.90537232e-01
9.34972838e-02 -3.72922778e-01 -8.01030040e-01 -3.01820248e-01
-6.34628952e-01 3.17221582e-01 -4.68327969e-01 -6.48580670e-01
4.87951159e-01 -1.76920384e-01 -3.16465646e-01 1.02069542e-01
-6.50933027e-01 5.86361922e-02 4.96777266e-01 6.04254305e-01
6.40624881e-01 1.17022544e-01 3.77615243e-01 -1.12530553e+00
6.41397655e-01 -9.60494652e-02 -1.01726341e+00 6.32702231e-01
-4.44577307e-01 -7.15447143e-02 6.78152561e-01 -1.41646132e-01
-1.64643848e+00 1.32748038e-01 -7.36755133e-01 6.48816586e-01
-2.87337840e-01 4.97650534e-01 -6.57318890e-01 3.06254268e-01
-6.49515986e-02 1.97766557e-01 -6.72513127e-01 -7.54388034e-01
3.36421728e-01 8.04889917e-01 4.14499462e-01 -7.37702906e-01
3.48073989e-01 -7.63180256e-02 -2.51140207e-01 -7.12124825e-01
-7.00225532e-01 -3.82025033e-01 -7.40337610e-01 1.00031883e-01
1.34436786e+00 -8.05411816e-01 -1.48263618e-01 5.91537476e-01
-1.70332277e+00 9.28063598e-03 2.96506971e-01 7.61301696e-01
-1.26313632e-02 7.81452775e-01 -9.83437955e-01 -8.51247370e-01
-4.29317892e-01 -1.23462045e+00 7.92892039e-01 5.05269051e-01
-1.86253905e-01 -8.65813315e-01 -1.83119029e-01 2.48304218e-01
-8.04644823e-02 -6.53850958e-02 1.12383723e+00 -4.62719589e-01
-5.27502298e-01 -3.41398746e-01 -1.76664472e-01 3.91359746e-01
2.00531274e-01 2.03949168e-01 -4.43444610e-01 2.15176880e-01
1.56219304e-01 3.47670987e-02 7.94512868e-01 5.53457201e-01
5.71065903e-01 6.03915472e-03 -2.31024489e-01 4.16407049e-01
1.49498415e+00 6.37726903e-01 5.82320571e-01 6.55297488e-02
5.24889290e-01 6.95627451e-01 1.09640563e+00 2.44345412e-01
6.73017561e-01 -2.20785048e-02 2.64779061e-01 4.10944074e-02
1.23601533e-01 -7.08976462e-02 3.33889693e-01 1.36251044e+00
3.28080878e-02 -4.96332765e-01 -7.90233314e-01 7.33812273e-01
-1.41610730e+00 -3.03672075e-01 -6.34302795e-01 1.73357844e+00
1.07949662e+00 3.85779172e-01 -2.47379109e-01 1.97123721e-01
1.15772438e+00 2.38730505e-01 -2.01488547e-02 -8.45791936e-01
-1.42908454e-01 3.42028439e-01 6.90260172e-01 7.12771833e-01
-1.07401061e+00 1.70954251e+00 6.89447069e+00 8.06774974e-01
-9.64959979e-01 2.04786465e-01 8.83323133e-01 6.45680249e-01
-6.07476711e-01 5.07987261e-01 -1.50907636e+00 3.42093319e-01
4.87178117e-01 3.11855674e-01 -6.92396387e-02 6.58189893e-01
1.42257035e-01 -6.54572189e-01 -4.70355451e-01 4.79525268e-01
-3.72420251e-01 -9.01580691e-01 -2.47895911e-01 -3.67677212e-01
6.11322641e-01 -3.39060158e-01 -4.46353763e-01 2.67649859e-01
4.00034130e-01 -5.22416234e-01 8.31046224e-01 -2.05007598e-01
5.98984480e-01 -8.08705986e-01 8.72624874e-01 4.17840809e-01
-1.50146127e+00 3.52798402e-01 -6.62148297e-01 -2.32115433e-01
8.69027972e-01 6.88082039e-01 -5.99532187e-01 3.81759584e-01
4.64736879e-01 -7.22264573e-02 -2.88653761e-01 8.15445364e-01
-8.22354555e-01 1.18584013e+00 -4.01729256e-01 -3.64229321e-01
5.34595788e-01 -5.04910648e-01 2.18863651e-01 1.49187255e+00
3.89851242e-01 6.65387452e-01 3.89778405e-01 5.11530876e-01
2.20038459e-01 3.54209840e-01 1.37747884e-01 -1.39944777e-01
7.54124761e-01 1.20547664e+00 -1.48587060e+00 -5.96558273e-01
-3.55909765e-01 8.49110126e-01 2.76803434e-01 1.79171309e-01
-8.06929886e-01 -2.91026115e-01 2.62636513e-01 -3.50139230e-01
2.60374635e-01 -8.92926812e-01 -8.66476595e-01 -8.94524097e-01
-4.60558496e-02 -6.40212893e-01 4.47024733e-01 -7.22199604e-02
-7.41735518e-01 7.89448202e-01 8.58485401e-02 -7.23173082e-01
7.85771310e-02 -5.15987813e-01 -1.02361083e+00 1.14432609e+00
-1.57838321e+00 -1.09161747e+00 2.85148561e-01 9.48358178e-02
7.25732386e-01 3.25761825e-01 7.24569619e-01 1.75239891e-01
-8.07798266e-01 5.21474600e-01 -1.70239925e-01 4.28099662e-01
3.10806990e-01 -1.30073905e+00 8.93805504e-01 1.41318214e+00
-1.63488311e-03 6.61614776e-01 7.14356899e-01 -1.19234157e+00
-8.16846251e-01 -8.43978763e-01 1.46771383e+00 -4.80642505e-02
3.56773913e-01 -4.41712141e-01 -8.24144006e-01 6.86983407e-01
2.83481866e-01 -4.26802218e-01 7.36004710e-01 1.09668322e-01
2.82486498e-01 1.75270990e-01 -1.00632286e+00 4.93211925e-01
8.60098541e-01 -8.72406587e-02 -1.03653467e+00 -5.09268343e-02
1.10364711e+00 -5.56507468e-01 -5.96793950e-01 2.39321128e-01
3.95820946e-01 -6.82474136e-01 3.86649489e-01 -1.94517881e-01
4.79080528e-02 -3.29991519e-01 -2.87359267e-01 -1.01118505e+00
-2.14819551e-01 -3.89726222e-01 6.66071653e-01 1.72832179e+00
7.43831694e-01 -3.77003819e-01 8.95307839e-01 8.71127784e-01
-4.37595516e-01 -4.56399322e-01 -9.56044137e-01 -6.31195962e-01
3.09596717e-01 -7.77633667e-01 8.10139656e-01 4.80954379e-01
1.01672113e-02 3.10059428e-01 -2.34543920e-01 2.51189768e-01
4.57898706e-01 4.16414648e-01 6.98886141e-02 -7.01624572e-01
-1.10153787e-01 -3.20374519e-01 2.51929253e-01 -1.38129890e+00
3.73434424e-02 -4.42479432e-01 7.14155912e-01 -1.67072237e+00
-3.15613955e-01 -1.02559543e+00 -1.02668248e-01 3.45312893e-01
-7.24517107e-01 2.09022481e-02 1.35764614e-01 -1.96786433e-01
-3.86046290e-01 3.58864546e-01 1.18807352e+00 1.45318881e-01
-3.76806468e-01 2.62676865e-01 -5.98405600e-01 8.29996109e-01
1.10854053e+00 -7.10738957e-01 1.27106041e-01 -6.82746589e-01
1.14600383e-01 3.62517953e-01 -5.80367267e-01 -6.03253067e-01
5.66717125e-02 -3.42776626e-01 4.81848828e-02 -1.15251124e+00
-2.86822468e-02 -7.08490133e-01 -3.26941282e-01 6.56040370e-01
1.07801385e-01 4.40328181e-01 1.59841374e-01 2.35330209e-01
-2.77712047e-01 -8.69110405e-01 5.81367552e-01 -4.41746324e-01
-7.61185884e-01 -4.04027570e-03 -7.94998050e-01 5.54305017e-02
7.69671321e-01 -2.85823554e-01 5.08702500e-03 1.35539100e-01
-5.98337948e-01 2.51793146e-01 7.71430880e-02 3.04219127e-02
4.76356208e-01 -8.46209884e-01 -5.85181057e-01 -1.61196031e-02
-3.58118504e-01 2.47608975e-01 1.48008168e-01 3.53558540e-01
-9.26128447e-01 4.18032795e-01 4.27303724e-02 -2.01083630e-01
-1.35447240e+00 2.74145603e-01 -2.74457455e-01 -6.18695617e-01
-2.40130112e-01 1.03710866e+00 -2.76084635e-02 -1.87710002e-01
1.76075175e-02 -7.70771384e-01 -4.61079985e-01 -5.80600761e-02
1.49508193e-01 2.67555118e-01 -1.30480036e-01 -7.54868746e-01
-4.75284874e-01 7.75047779e-01 -4.61389683e-02 -3.01047504e-01
8.35165441e-01 -4.37589049e-01 -4.58504111e-01 2.21768111e-01
7.23789096e-01 4.24339235e-01 -9.27502632e-01 2.44694147e-02
5.36964893e-01 -2.92848706e-01 3.83928604e-03 -6.33878052e-01
-6.75023139e-01 1.02411330e+00 6.87773302e-02 6.68379590e-02
1.27641833e+00 -7.73846880e-02 1.36722040e+00 1.29135281e-01
4.05702084e-01 -1.51881313e+00 -6.26750588e-01 9.82602775e-01
3.43515515e-01 -1.19880950e+00 -1.66137114e-01 -9.78603184e-01
-4.62339669e-01 1.29856956e+00 6.73777580e-01 -7.56796747e-02
7.19631851e-01 4.61168736e-01 3.55365306e-01 2.32967496e-01
-3.55403781e-01 -2.01379493e-01 -4.25508618e-02 2.16817021e-01
8.71819973e-01 4.87900496e-01 -1.49345303e+00 1.03389692e+00
-3.51370424e-01 -5.39894998e-01 4.75366384e-01 1.04186964e+00
-7.83217072e-01 -1.91654432e+00 -5.10406613e-01 2.27230117e-02
-1.06558144e+00 -5.47456980e-01 -1.89792246e-01 6.71147645e-01
2.76917458e-01 1.30734098e+00 6.89471737e-02 -2.36145720e-01
-4.55015944e-03 3.11304200e-02 3.91977876e-01 -9.59494472e-01
-6.70305669e-01 6.84025586e-01 5.36928356e-01 -4.41581905e-02
-3.00702721e-01 -6.26348495e-01 -1.82227290e+00 -3.63036394e-02
-8.82881224e-01 5.16128778e-01 8.42524886e-01 1.09253800e+00
-3.32501978e-01 3.73912960e-01 5.46985328e-01 -2.45667338e-01
-4.19674724e-01 -1.04048944e+00 -6.85580373e-01 -1.98917150e-01
-3.15525770e-01 -8.74176547e-02 -3.94248143e-02 -1.21795893e-01]
|
[10.111282348632812, 10.101997375488281]
|
e80729ad-5ffd-4515-a1b5-8725ed7a37d2
|
hierarchical-multi-resolution-mesh-networks
|
1607.07695
| null |
http://arxiv.org/abs/1607.07695v2
|
http://arxiv.org/pdf/1607.07695v2.pdf
|
Hierarchical Multi-resolution Mesh Networks for Brain Decoding
|
We propose a new framework, called Hierarchical Multi-resolution Mesh
Networks (HMMNs), which establishes a set of brain networks at multiple time
resolutions of fMRI signal to represent the underlying cognitive process. The
suggested framework, first, decomposes the fMRI signal into various frequency
subbands using wavelet transforms. Then, a brain network, called mesh network,
is formed at each subband by ensembling a set of local meshes. The locality
around each anatomic region is defined with respect to a neighborhood system
based on functional connectivity. The arc weights of a mesh are estimated by
ridge regression formed among the average region time series. In the final
step, the adjacency matrices of mesh networks obtained at different subbands
are ensembled for brain decoding under a hierarchical learning architecture,
called, fuzzy stacked generalization (FSG). Our results on Human Connectome
Project task-fMRI dataset reflect that the suggested HMMN model can
successfully discriminate tasks by extracting complementary information
obtained from mesh arc weights of multiple subbands. We study the topological
properties of the mesh networks at different resolutions using the network
measures, namely, node degree, node strength, betweenness centrality and global
efficiency; and investigate the connectivity of anatomic regions, during a
cognitive task. We observe significant variations among the network topologies
obtained for different subbands. We, also, analyze the diversity properties of
classifier ensemble, trained by the mesh networks in multiple subbands and
observe that the classifiers in the ensemble collaborate with each other to
fuse the complementary information freed at each subband. We conclude that the
fMRI data, recorded during a cognitive task, embed diverse information across
the anatomic regions at each resolution.
|
['Mete Ozay', 'Fatos Tunay Yarman Vural', 'Itir Onal Ertugrul']
|
2016-07-12
| null | null | null | null |
['brain-decoding', 'brain-decoding']
|
['medical', 'miscellaneous']
|
[ 4.18663211e-02 4.00019661e-02 3.39987833e-04 -2.21972585e-01
9.22644958e-02 -5.69303513e-01 3.65629613e-01 1.60896063e-01
-5.55603728e-02 5.76371014e-01 3.18892211e-01 1.04322545e-01
-1.04444170e+00 -1.04724526e+00 -5.30833662e-01 -6.79449916e-01
-7.25714207e-01 3.54412287e-01 2.59200990e-01 -2.52406090e-01
2.28271127e-01 7.72955418e-01 -1.27199447e+00 5.97107053e-01
8.41233492e-01 1.01521301e+00 4.24270742e-02 2.63514578e-01
-1.11832600e-02 4.18061525e-01 -2.16346219e-01 -5.07482141e-02
1.94038838e-01 -3.33422035e-01 -8.45902681e-01 -1.59028158e-01
2.98256785e-01 3.94964784e-01 -3.42882514e-01 1.36105072e+00
1.46003038e-01 2.93298662e-01 8.96099091e-01 -7.15491951e-01
-4.39661950e-01 9.74829853e-01 -6.69750869e-01 9.38727856e-01
1.48664385e-01 -1.72465771e-01 9.87432778e-01 -8.02670896e-01
7.83226252e-01 1.26887596e+00 7.01756954e-01 -4.18071374e-02
-1.64854074e+00 -6.56369746e-01 1.22118294e-01 2.20785752e-01
-1.56448925e+00 -1.82253435e-01 9.90806580e-01 -8.57209206e-01
5.31322479e-01 1.54130787e-01 8.90741348e-01 1.23168898e+00
1.04354906e+00 -3.33663464e-01 1.63582075e+00 8.65319967e-02
8.77161026e-02 -3.95341665e-01 4.57045376e-01 1.04957628e+00
2.76632458e-01 -5.92191741e-02 -5.42140484e-01 -3.22932690e-01
1.16335881e+00 -1.33321173e-02 -4.35102552e-01 -1.52404919e-01
-1.48366117e+00 6.77206814e-01 9.80085611e-01 1.08171165e+00
-6.92727029e-01 -2.90317327e-01 3.65037948e-01 4.51984674e-01
4.91855741e-01 3.62716973e-01 -1.06652714e-01 7.87947357e-01
-1.05695891e+00 -3.33784729e-01 5.16569555e-01 3.52749795e-01
7.10432410e-01 1.84807815e-02 -9.74536836e-02 6.87473416e-01
1.36119828e-01 -9.26906168e-02 6.15624249e-01 -9.37860012e-01
2.63426423e-01 7.91240692e-01 -6.90875232e-01 -1.63020015e+00
-9.58313942e-01 -7.60931313e-01 -1.74396861e+00 5.33347502e-02
3.03126514e-01 -8.18555281e-02 -3.27173412e-01 1.83935249e+00
2.86730081e-01 2.18999311e-01 -5.13511658e-01 7.43223131e-01
6.68231368e-01 1.60258114e-01 8.95981863e-02 -3.91889274e-01
1.56802607e+00 -4.37626868e-01 -6.58279002e-01 2.13583186e-01
6.41164631e-02 -6.06355704e-02 5.41971266e-01 3.04620713e-01
-1.19558477e+00 -1.06860256e+00 -1.01734614e+00 4.61057276e-01
-5.41800737e-01 -1.63044006e-01 3.21813405e-01 1.36396751e-01
-1.45367599e+00 9.58827853e-01 -6.89827800e-01 -2.37149045e-01
2.63348311e-01 4.34354633e-01 -7.88213193e-01 2.39599004e-01
-1.30661654e+00 8.55126381e-01 4.95141745e-01 3.72991204e-01
-7.21178234e-01 -6.81759775e-01 -4.80269253e-01 1.35367855e-01
-1.76633626e-01 -8.02854478e-01 -1.09864406e-01 -9.22620773e-01
-9.33150887e-01 5.80261827e-01 -1.34932473e-02 -2.42173463e-01
2.34637097e-01 8.28199923e-01 -7.03704298e-01 5.99542916e-01
2.44608689e-02 4.20470059e-01 9.14141476e-01 -1.21545076e+00
1.54719085e-01 -7.27756143e-01 -1.85404345e-01 -9.39166769e-02
-3.22540849e-01 -3.01920086e-01 3.88277709e-01 -6.64797843e-01
6.73790157e-01 -6.07128918e-01 -1.16275482e-01 -2.09831655e-01
-4.68449891e-01 -2.36276850e-01 4.40738559e-01 -8.64499331e-01
1.21620047e+00 -2.12948251e+00 8.27048540e-01 1.08816147e+00
1.02673638e+00 -5.59763312e-01 -2.81587243e-01 7.19326884e-02
-4.99966532e-01 3.27297986e-01 -3.64482701e-01 4.59289312e-01
-4.56607163e-01 -7.66643137e-02 3.73584867e-01 7.30491877e-01
-1.83715560e-02 7.32459605e-01 -7.97743201e-01 -7.57663786e-01
-4.08672206e-02 5.77562928e-01 -2.93697000e-01 -1.86858714e-01
4.60940033e-01 8.70919406e-01 -4.36716557e-01 2.26317555e-01
6.52486920e-01 -4.38327223e-01 6.67622149e-01 -8.55840862e-01
-6.52278960e-02 -3.56151223e-01 -8.54397714e-01 1.60965896e+00
-1.50733516e-01 5.10335743e-01 5.48335910e-01 -1.66148365e+00
1.01931214e+00 4.15363193e-01 8.32588077e-01 -6.20302677e-01
2.63759583e-01 4.65606898e-03 6.50211751e-01 -5.37610352e-01
-2.15053946e-01 1.60183698e-01 8.44247490e-02 2.88421243e-01
5.72807074e-01 5.26333272e-01 1.41649187e-01 -1.32212518e-02
1.34740531e+00 -4.65108931e-01 2.78070718e-01 -1.05421245e+00
8.14382315e-01 -4.14479524e-01 1.03864625e-01 5.44433415e-01
-3.19670171e-01 5.35259908e-03 6.79131806e-01 -5.13443112e-01
-9.52234924e-01 -1.46888375e+00 -6.13206625e-01 8.66014302e-01
-1.16393737e-01 5.88434301e-02 -8.55540514e-01 -3.67196590e-01
8.92304182e-02 -1.70970783e-01 -1.05897987e+00 -1.93738163e-01
-6.92528605e-01 -7.64869094e-01 5.95256925e-01 1.32984072e-01
7.18218625e-01 -1.00434268e+00 -3.16010058e-01 1.02391101e-01
-3.92954439e-01 -1.05414629e+00 -2.79752374e-01 -8.56141280e-03
-1.34182191e+00 -1.19799030e+00 -3.46284240e-01 -8.33190739e-01
8.37063193e-01 3.13298660e-03 1.04826689e+00 2.66048998e-01
-3.29915166e-01 3.44659716e-01 -4.91380692e-02 6.49509966e-01
-1.57083049e-01 7.95173943e-02 3.17040503e-01 4.34277624e-01
-3.00148398e-01 -1.38745117e+00 -6.56040132e-01 3.58039975e-01
-7.24146426e-01 -2.10120186e-01 7.14242160e-01 7.23220706e-01
7.41079092e-01 5.11340857e-01 7.49829292e-01 -6.02617264e-01
8.65639567e-01 -8.99704397e-01 -1.10949613e-01 3.26154739e-01
-2.22265244e-01 2.01694027e-01 6.66140199e-01 -4.01411593e-01
-8.88435423e-01 -2.72545785e-01 3.69384378e-01 -2.74663180e-01
-1.37944803e-01 7.76863754e-01 7.75334314e-02 -6.03390872e-01
7.88645744e-01 2.15991870e-01 2.40163669e-01 -3.00711840e-01
3.01842272e-01 2.65922666e-01 4.79014128e-01 -7.18648136e-01
4.71564651e-01 4.96990114e-01 5.35866439e-01 -8.98276448e-01
-3.82319987e-01 2.02161465e-02 -1.28653610e+00 -7.20725358e-01
1.09717166e+00 -6.54937744e-01 -1.12554204e+00 -3.97875085e-02
-1.00196850e+00 1.13421023e-01 1.49768308e-01 5.04858613e-01
-1.71064943e-01 3.63823295e-01 -8.59648645e-01 -4.70311284e-01
-2.82950938e-01 -8.32094908e-01 4.56775367e-01 -9.13226604e-02
-2.77852327e-01 -1.39477384e+00 -5.31500839e-02 1.99095413e-01
4.82461691e-01 7.65397370e-01 1.51505709e+00 -4.71993834e-01
-2.26057693e-01 2.08952382e-01 -2.49685079e-01 6.01270096e-03
7.28246644e-02 -8.92282426e-02 -4.94818270e-01 -2.54081041e-01
1.93142653e-01 1.30187586e-01 9.15046930e-01 6.63543820e-01
1.38220549e+00 -2.96269804e-01 -4.01094973e-01 3.06499362e-01
1.38320482e+00 7.58858910e-03 2.09599420e-01 -2.41202861e-01
5.42898536e-01 9.08578455e-01 -4.82001543e-01 1.43777756e-02
1.52690783e-01 2.21651942e-01 2.42742285e-01 -6.55668154e-02
-1.09636888e-01 2.64320552e-01 2.20566437e-01 1.43897271e+00
-8.96684170e-01 3.73312056e-01 -8.54335070e-01 2.42166460e-01
-1.57092893e+00 -1.15657449e+00 -2.68069714e-01 1.91667402e+00
5.30652046e-01 1.27028838e-01 1.93425030e-01 2.48045544e-03
1.08737040e+00 1.93407923e-01 -4.60983694e-01 1.78298505e-03
-3.18268031e-01 4.98941034e-01 1.09567657e-01 5.68615675e-01
-7.54053473e-01 3.18040252e-01 6.49844360e+00 5.19862711e-01
-8.09076309e-01 5.45119464e-01 6.48576319e-01 7.15739280e-02
-1.85811058e-01 -4.15806592e-01 1.43940449e-01 3.00895095e-01
1.09402096e+00 -9.22885984e-02 8.71919155e-01 2.01928407e-01
3.19728583e-01 9.57510844e-02 -8.63343060e-01 6.74081862e-01
-1.76026776e-01 -1.30169439e+00 2.20201746e-01 1.78171337e-01
5.00835240e-01 5.37463501e-02 -6.97058961e-02 -1.87153071e-01
1.82080060e-01 -1.10029221e+00 3.48052323e-01 1.24478173e+00
7.96834230e-01 -5.91286421e-01 4.76239055e-01 2.93421268e-01
-1.88431954e+00 -2.23260164e-01 -4.23487633e-01 -2.20791157e-02
-2.05525652e-01 9.08760607e-01 1.42733194e-02 8.45707536e-01
7.04494655e-01 7.19266832e-01 -6.74778461e-01 7.56056547e-01
2.18927726e-01 2.67719179e-01 1.37917697e-01 2.78767109e-01
-1.68683097e-01 -6.46841645e-01 3.95783126e-01 1.04721618e+00
4.58208293e-01 2.35248834e-01 1.68558136e-01 1.26139784e+00
-8.11057314e-02 1.23921253e-01 -7.24516809e-01 1.27519459e-01
5.37361145e-01 1.76051366e+00 -1.19595337e+00 -3.02975595e-01
-1.83617830e-01 5.59428573e-01 7.05507874e-01 5.31934381e-01
-6.79636061e-01 -2.42587298e-01 4.12921786e-01 1.23856463e-01
-2.29299575e-01 -5.41346788e-01 -3.78519058e-01 -1.08161831e+00
-1.15909368e-01 -4.52322155e-01 4.13233459e-01 -6.56106055e-01
-1.77209795e+00 1.00944984e+00 1.17293149e-01 -8.72314632e-01
3.60130817e-01 -6.00207925e-01 -5.34665227e-01 1.07314312e+00
-9.30577219e-01 -5.50490618e-01 -3.80537331e-01 1.03773081e+00
7.60293007e-02 -6.27053827e-02 7.65364468e-01 3.57317030e-01
-6.02031410e-01 1.21189058e-01 -1.39172405e-01 3.36607575e-01
1.38871059e-01 -8.93163800e-01 -1.81362703e-01 4.16516632e-01
1.38369370e-02 8.95616710e-01 3.45798805e-02 -9.23082292e-01
-1.03509462e+00 -9.66604173e-01 5.81489086e-01 -1.33499280e-01
9.50083733e-01 -4.69863385e-01 -1.02454567e+00 4.20108318e-01
3.13735962e-01 3.33533704e-01 5.74355483e-01 7.32987151e-02
-3.69408756e-01 -1.48585975e-01 -1.35821426e+00 2.00122610e-01
1.55275226e+00 -8.08824837e-01 -6.38984442e-01 4.82463509e-01
4.74691927e-01 1.66955993e-01 -1.82884502e+00 4.10147518e-01
6.23646975e-01 -1.09828365e+00 1.18682075e+00 -5.45665562e-01
3.49789143e-01 -1.67553619e-01 8.81963149e-02 -1.55887938e+00
-1.27946758e+00 7.73824677e-02 -2.06690133e-02 7.35037565e-01
4.17176366e-01 -8.82373214e-01 1.97088599e-01 -6.44236431e-02
-8.97809863e-02 -5.34379780e-01 -1.32083023e+00 -5.79154611e-01
1.14530943e-01 8.57404619e-02 4.20522451e-01 1.39663804e+00
2.33908251e-01 4.37677294e-01 1.41802713e-01 1.88574880e-01
9.17538524e-01 -1.60342380e-01 -4.59226161e-01 -1.99389005e+00
-3.53678577e-02 -7.46099830e-01 -4.08888042e-01 -1.43987060e-01
5.62408984e-01 -1.45034182e+00 -5.31465411e-01 -1.35998201e+00
1.75729468e-01 -1.00309163e-01 -6.54709280e-01 2.05732748e-01
3.70238304e-01 2.87930518e-01 2.40183179e-03 5.02438128e-01
-1.05352804e-01 1.13971934e-01 1.72820842e+00 -2.46637359e-01
-3.39618921e-02 -5.16709566e-01 -4.16911900e-01 8.13404143e-01
6.42839730e-01 -8.24879184e-02 -3.26629251e-01 -1.24381505e-01
1.57352060e-01 5.76423287e-01 5.33756614e-01 -1.46228850e+00
3.20340395e-01 1.62211657e-01 9.92118657e-01 -2.36772001e-02
7.68554583e-02 -9.58666027e-01 5.72158039e-01 6.71757519e-01
-4.03544366e-01 4.59357262e-01 -1.05981819e-01 5.81862330e-01
-2.14987844e-01 3.42677146e-01 7.97371447e-01 -3.13894212e-01
-1.87037572e-01 3.23855966e-01 -4.98796850e-01 -1.53738186e-01
7.61044800e-01 -1.89100742e-01 -3.60466748e-01 1.75593287e-01
-1.55831182e+00 -2.01309070e-01 -2.69113630e-01 4.25263084e-02
6.28298342e-01 -1.62767315e+00 -6.81216180e-01 2.99482375e-01
-4.65241313e-01 -6.88815057e-01 7.58654714e-01 1.44125295e+00
1.46573884e-02 1.76511094e-01 -1.03382516e+00 -5.57833076e-01
-9.78977680e-01 5.10895789e-01 6.61885142e-01 -4.40763414e-01
-5.44076979e-01 5.38582981e-01 5.29702865e-02 -2.81972766e-01
-2.72592098e-01 -4.39108193e-01 -9.01962996e-01 5.71742594e-01
3.06802273e-01 5.59442222e-01 1.39316702e-02 -1.09902453e+00
-4.02176619e-01 1.01345325e+00 5.21658063e-01 2.42944416e-02
1.32452261e+00 -1.64798081e-01 -9.78389263e-01 5.22565007e-01
1.17324054e+00 -1.60453409e-01 -6.52126670e-01 -2.48117223e-01
1.38350902e-02 7.25936145e-02 4.24666964e-02 -5.13794482e-01
-1.48109818e+00 7.14424133e-01 6.53814554e-01 7.23761201e-01
1.28197551e+00 -2.91385297e-02 1.59567773e-01 2.15344340e-01
6.30921543e-01 -7.70883441e-01 1.06399886e-01 4.41851228e-01
1.02956092e+00 -5.80934763e-01 -2.57419854e-01 -4.05756563e-01
-6.27238080e-02 1.40548432e+00 4.46997911e-01 -5.57181239e-01
1.16446078e+00 8.69806632e-02 -6.35294855e-01 -7.76970148e-01
-3.93118322e-01 1.18531972e-01 7.36469269e-01 4.25182998e-01
5.07589638e-01 3.70028228e-01 -4.35063839e-01 6.69949889e-01
-2.85102397e-01 -3.38355064e-01 1.92215830e-01 1.77236989e-01
-6.83956206e-01 -6.36758447e-01 -4.45942670e-01 8.19983482e-01
-3.33457366e-02 -1.05802424e-01 -4.32295203e-01 6.64846659e-01
4.66138929e-01 7.33238280e-01 2.90146232e-01 -7.89739728e-01
2.44875208e-01 3.50184083e-01 7.56306827e-01 -2.83741176e-01
-7.60043681e-01 -9.79554579e-02 -1.11988552e-01 -7.20422924e-01
-6.74777806e-01 -4.35248405e-01 -1.18489802e+00 -5.19982934e-01
1.55931413e-01 2.94909239e-01 1.78811148e-01 9.20894206e-01
4.46373940e-01 9.56595182e-01 5.59749961e-01 -1.17447603e+00
6.98775798e-02 -1.09112525e+00 -9.51502144e-01 3.32756013e-01
2.41694808e-01 -9.92500842e-01 -4.79764134e-01 -2.05419213e-01]
|
[12.479047775268555, 3.389512062072754]
|
c232e78e-405f-46c9-b265-d2bfa364fbf3
|
placing-historical-events-on-a-timeline-a
| null | null |
https://openreview.net/forum?id=zMLzqW4AC20
|
https://openreview.net/pdf?id=zMLzqW4AC20
|
Placing (Historical) Events on a Timeline: A Classification cum Co-ref Resolution Approach
|
The event timeline provides one of the most effective ways to visualize the important historical events that occurred over a period of time, presenting the insights that may not be so apparent from reading the equivalent information in textual form. By leveraging generative adversarial learning for important event classification and by assimilating knowledge based tags for improving the performance of event coreference resolution we introduce a two staged system for event timeline generation from multiple (historical) text documents. In addition, we propose a vis-timeline based visualization technique to portray the event timeline. We demonstrate our results on two very well known historical documents -- the Collected Works of Mahatma Gandhi (CWMG) and the Collected Works of Abraham Lincoln (CWAL). Our results can be extremely helpful for historians, in advancing research in history and in understanding the socio-political landscape of a country as reflected in the writings of political leaders/scholars. Our work has some parallels with timeline summarization (TLS) tasks and therefore we use these as baselines. Rigorous experiments demonstrate that prior event detection which was hitherto absent in the TLS methods can improve summarization performance. In order to show that our methods are very generic we reuse our method to visualize the evolution of coronavirus related events in India from a collection of various COVID-19 articles.
|
['Anonymous']
|
2021-11-16
| null |
https://openreview.net/forum?id=Y5tTolyfhoP
|
https://openreview.net/pdf?id=Y5tTolyfhoP
|
acl-arr-september-2021-9
|
['timeline-summarization']
|
['natural-language-processing']
|
[ 5.29423177e-01 2.13621736e-01 1.06668629e-01 -1.28605410e-01
-1.00652075e+00 -8.31782997e-01 1.41118741e+00 7.21328437e-01
-2.24448875e-01 1.01440847e+00 1.15225232e+00 -5.29109776e-01
-1.73830971e-01 -8.02555084e-01 -5.14003158e-01 -5.45437515e-01
-3.32938850e-01 4.51952010e-01 -4.03513424e-02 -4.47760552e-01
4.64171469e-01 6.45172179e-01 -9.46657062e-01 3.86206597e-01
7.96482921e-01 1.63364828e-01 -5.94831556e-02 7.99112558e-01
-2.19091013e-01 9.11524534e-01 -1.38775229e+00 -6.11306012e-01
-9.75620374e-02 -7.62865424e-01 -7.26494491e-01 -4.40488458e-01
4.77067977e-01 1.65544093e-01 -3.54865581e-01 6.32290244e-01
5.83300531e-01 -1.13199390e-02 8.43090475e-01 -1.04291749e+00
-3.66650969e-01 1.07911611e+00 -6.91388845e-01 7.55576611e-01
5.24735749e-01 -1.39472261e-01 7.11512566e-01 -3.74732554e-01
1.43300688e+00 1.47933257e+00 8.04513156e-01 -4.85168799e-04
-1.12048745e+00 -5.03842652e-01 -3.39207873e-02 3.17407072e-01
-7.36357093e-01 -2.20234647e-01 9.96575713e-01 -6.17566943e-01
6.34459019e-01 5.96936345e-01 7.67346859e-01 1.45815086e+00
4.88923997e-01 5.71320772e-01 1.03076971e+00 -4.25265163e-01
9.49126631e-02 -2.62641937e-01 1.59249574e-01 3.65563631e-01
3.02912761e-02 -9.42493752e-02 -5.55818439e-01 -5.48615158e-01
4.22955155e-01 1.47931233e-01 -3.09549659e-01 5.12323797e-01
-1.50662351e+00 1.06514156e+00 3.66758764e-01 6.43186569e-01
-5.76243997e-01 6.47055358e-02 8.15053701e-01 2.77240545e-01
7.51610577e-01 5.68465889e-01 1.89694658e-01 -1.88400000e-01
-1.44942987e+00 4.69931632e-01 8.19927752e-01 3.12242240e-01
2.02173576e-01 1.14734255e-01 -5.58748782e-01 2.77972162e-01
-1.89144865e-01 4.22915608e-01 1.90326363e-01 -7.81199515e-01
4.89116311e-01 4.39023674e-01 1.74189866e-01 -1.33970547e+00
-3.64605606e-01 -2.68867999e-01 -8.00249577e-01 7.59914145e-02
4.10272986e-01 -2.86462516e-01 -9.13610518e-01 1.67202461e+00
1.75602674e-01 9.47925970e-02 1.40805915e-01 3.78375053e-01
9.80927050e-01 1.19293845e+00 1.50637880e-01 -7.05525339e-01
1.58719516e+00 -3.51182133e-01 -9.63200152e-01 5.18455245e-02
1.68878600e-01 -8.48758340e-01 8.30579221e-01 6.19666763e-02
-9.54173803e-01 -2.45866731e-01 -1.23823893e+00 2.37769727e-02
-5.94710827e-01 -2.65188307e-01 4.59350765e-01 2.64726132e-01
-6.10168219e-01 7.81266987e-01 -8.80728662e-01 -8.20178330e-01
5.06882429e-01 -4.23927933e-01 -2.39997417e-01 5.33019543e-01
-1.34623158e+00 9.53816056e-01 4.99041319e-01 -2.77299047e-01
-6.63754761e-01 -9.24413264e-01 -6.30941808e-01 -6.26285598e-02
3.20548385e-01 -4.55143511e-01 1.11667967e+00 -4.07572210e-01
-7.39001215e-01 7.68406212e-01 -1.49263978e-01 -7.54227459e-01
7.47798681e-01 -1.52407303e-01 -6.48565769e-01 4.91816133e-01
3.91500920e-01 1.15704171e-01 5.23429155e-01 -1.07245719e+00
-4.83595580e-01 -3.56748194e-01 -8.02658871e-02 -8.49420130e-02
-1.24475583e-01 4.88298893e-01 4.52749506e-02 -1.34408534e+00
-2.54545659e-01 -5.79856515e-01 -2.77460873e-01 -4.50379461e-01
-7.63691604e-01 -1.40519127e-01 1.17483926e+00 -1.28953528e+00
1.51227021e+00 -1.88889360e+00 2.42672890e-01 5.56683950e-02
1.57953233e-01 1.82613893e-03 3.11674982e-01 1.20551097e+00
-9.65918973e-02 3.33855212e-01 -5.63150346e-01 -1.86997905e-01
-2.05304861e-01 1.38275906e-01 -1.02039635e+00 3.69504213e-01
1.95693653e-02 1.04643309e+00 -1.14949512e+00 -6.15224302e-01
1.25288203e-01 5.55284262e-01 1.60307810e-01 -7.87894651e-02
-1.60835013e-01 5.78430295e-01 -2.03868598e-01 2.46000290e-01
1.27447158e-01 -1.54201046e-01 1.63429484e-01 -2.31212288e-01
-3.09097022e-01 3.02978843e-01 -7.42825150e-01 1.59271693e+00
-2.59341504e-02 1.29321647e+00 -4.97578561e-01 -6.40020609e-01
7.22051978e-01 4.30059731e-01 2.01699853e-01 -6.13195717e-01
-9.78955925e-02 -2.55918980e-01 -2.78475076e-01 -3.76414478e-01
8.71110082e-01 -2.07635775e-01 -5.45156479e-01 9.34686720e-01
-2.08537266e-01 -2.35137828e-02 5.67472398e-01 9.05614734e-01
9.51202750e-01 -2.62240157e-03 6.85452342e-01 -1.31065696e-01
5.90276532e-02 4.46514457e-01 3.44630986e-01 9.55931425e-01
2.33386844e-01 7.36903012e-01 8.16752434e-01 -5.09112895e-01
-1.24807036e+00 -1.13502562e+00 -1.23826459e-01 7.97016084e-01
-2.58657515e-01 -8.87760878e-01 -4.75641102e-01 -6.34049892e-01
-2.99003810e-01 1.13699615e+00 -1.14275253e+00 1.80025473e-01
-1.06579351e+00 -8.60522270e-01 7.92876959e-01 4.19321537e-01
1.37700886e-01 -1.30646157e+00 -1.10839152e+00 4.19364333e-01
-3.54343086e-01 -6.65581524e-01 -2.31646642e-01 -4.90377657e-02
-5.48790038e-01 -9.90891993e-01 -8.94880712e-01 -3.66043478e-01
2.89096415e-01 -2.25739583e-01 1.13544214e+00 -5.33520520e-01
-5.27429938e-01 2.23243773e-01 -2.60380358e-01 -1.02995300e+00
-9.90478754e-01 -3.96321714e-02 -3.51348549e-01 -4.43471521e-01
-6.61047399e-02 -8.48631978e-01 -4.39290881e-01 -3.28888804e-01
-1.18634784e+00 2.54219353e-01 8.42227787e-02 5.15098631e-01
1.19434036e-01 -1.61816984e-01 6.47690892e-01 -1.27970529e+00
8.76247644e-01 -6.92590714e-01 -9.48771760e-02 4.39260751e-01
-1.75201505e-01 5.81880054e-03 4.75283533e-01 -3.86812121e-01
-1.32447982e+00 -6.30483747e-01 1.30750373e-01 1.35247692e-01
7.14275837e-02 7.95816123e-01 4.21817392e-01 7.78815627e-01
9.04240608e-01 6.29303306e-02 -3.87132466e-01 -4.50835824e-01
6.51898384e-01 5.38179338e-01 1.13254666e+00 -2.49452353e-01
8.00282121e-01 8.29157114e-01 -1.25462338e-01 -8.06443155e-01
-6.80141926e-01 -2.79884279e-01 -4.29198235e-01 -4.14819777e-01
7.26460159e-01 -6.94479764e-01 -3.95438194e-01 1.56960279e-01
-1.29984820e+00 1.08243644e-01 -6.33468032e-01 3.93512361e-02
-3.12353313e-01 4.20261234e-01 -5.66902757e-01 -7.41325080e-01
-6.23131454e-01 -4.69071031e-01 9.70968664e-01 4.75018710e-01
-8.01723897e-01 -1.22180557e+00 5.39298475e-01 -1.18019409e-01
3.46376568e-01 1.44261646e+00 1.08472896e+00 -9.38368917e-01
2.29270551e-02 6.54418468e-02 5.47887497e-02 -4.88868415e-01
1.94083720e-01 1.91975251e-01 -8.33135724e-01 -1.74437091e-01
-4.41056974e-02 5.50039038e-02 9.58404422e-01 3.13678950e-01
4.63010103e-01 -7.73965299e-01 -5.94039977e-01 8.87473449e-02
1.23457849e+00 4.37717825e-01 7.11591780e-01 4.99880046e-01
6.18754625e-01 6.88958049e-01 3.71183634e-01 5.80160439e-01
2.24917859e-01 5.61692834e-01 -4.55491208e-02 -2.22679943e-01
-2.37970412e-01 -4.37395722e-01 4.30028230e-01 5.86702526e-01
-2.13197500e-01 -4.50862080e-01 -9.83837783e-01 8.73953879e-01
-2.05332375e+00 -1.69868028e+00 -2.97322482e-01 1.89079857e+00
8.93407404e-01 9.74540412e-02 3.02416503e-01 1.69799969e-01
6.74860835e-01 7.77264953e-01 -1.72366425e-01 -5.82159936e-01
-4.18364078e-01 2.62639314e-01 1.34605542e-01 4.20087874e-01
-1.06007338e+00 5.49314618e-01 6.39254999e+00 7.59879827e-01
-1.10659730e+00 8.70409235e-02 4.50500488e-01 -2.27939919e-01
-5.69598556e-01 1.48804218e-01 -3.31224203e-01 5.36366284e-01
1.17742360e+00 -7.58072197e-01 -2.05544040e-01 2.93594629e-01
4.86096084e-01 -1.09637626e-01 -9.53632355e-01 7.14182973e-01
8.15111920e-02 -2.09246993e+00 1.91955671e-01 -9.68884602e-02
7.76954651e-01 -2.50047952e-01 -2.40096882e-01 1.75850149e-02
5.27945399e-01 -8.68784428e-01 8.48124802e-01 5.89315832e-01
6.74000025e-01 -8.44413519e-01 4.79127258e-01 2.02699378e-02
-8.84375215e-01 1.92909822e-01 1.16284668e-01 9.90616977e-02
8.48911524e-01 6.12566888e-01 -1.16795123e+00 1.00593972e+00
4.77868766e-01 7.65413582e-01 -7.06478953e-01 8.73301506e-01
-3.03255767e-01 9.45485175e-01 -3.41734290e-01 1.51957676e-01
2.67446965e-01 6.01074898e-05 1.17408597e+00 1.75292611e+00
4.05787528e-01 1.65756315e-01 -1.51914567e-01 6.60640895e-01
-1.23546477e-02 9.82155204e-02 -7.70579696e-01 -3.77840340e-01
5.12679636e-01 1.20171261e+00 -1.07901394e+00 -5.34198225e-01
4.98544537e-02 8.88702154e-01 -5.31141534e-02 2.16413096e-01
-7.91422248e-01 -5.63878477e-01 6.56436011e-02 2.45986611e-01
1.49824545e-01 -5.29923514e-02 -3.76060642e-02 -9.27259922e-01
-1.97989196e-01 -7.25256145e-01 9.76687670e-01 -9.54780817e-01
-1.11347497e+00 7.12043762e-01 4.78260159e-01 -8.62420917e-01
-5.78296602e-01 2.07849264e-01 -1.23627114e+00 6.51801586e-01
-8.37642074e-01 -1.37934482e+00 1.48200449e-02 2.00043723e-01
6.38633847e-01 4.28038128e-02 6.59028530e-01 -2.45908350e-02
-4.04775798e-01 6.08456694e-02 2.80697793e-01 2.54143894e-01
7.37062931e-01 -1.49847198e+00 7.99843848e-01 1.20420325e+00
4.79188442e-01 6.45603001e-01 1.30015600e+00 -1.04281640e+00
-7.44199812e-01 -9.20371592e-01 1.16909993e+00 -5.31885386e-01
7.69056737e-01 -3.07925075e-01 -9.16742265e-01 7.75995195e-01
8.41577768e-01 -1.03463781e+00 6.87239230e-01 -1.31250806e-02
-2.24828109e-01 3.63283366e-01 -9.76298392e-01 8.81339550e-01
7.85929203e-01 -4.29677159e-01 -1.36232913e+00 5.69318414e-01
7.37599075e-01 -1.59579605e-01 -6.90565050e-01 1.60093322e-01
3.73636037e-01 -6.55036688e-01 9.07357097e-01 -7.27944672e-01
6.37727022e-01 -3.88577253e-01 1.84351772e-01 -1.30173159e+00
-1.06263503e-01 -1.13046026e+00 1.87658928e-02 1.67241383e+00
2.53927886e-01 -4.49965835e-01 2.29247198e-01 -5.96016198e-02
-4.83547226e-02 -1.42435864e-01 -7.96323061e-01 -4.03272957e-01
-1.25356123e-01 -2.12644830e-01 3.15300286e-01 1.22243857e+00
2.05200026e-03 5.26185513e-01 -4.57711399e-01 -1.01856083e-01
6.42247558e-01 4.25319582e-01 5.54941058e-01 -1.33245826e+00
3.43296304e-02 -4.16985154e-01 -2.12936357e-01 2.86548678e-02
-3.22671741e-01 -8.00811172e-01 -3.87897730e-01 -1.82868361e+00
4.92507607e-01 2.26530451e-02 -1.12641854e-02 4.58538055e-01
-2.79869318e-01 3.45340639e-01 4.55393940e-01 4.97520924e-01
-2.83084601e-01 1.39287546e-01 8.37337196e-01 -1.44369900e-01
-2.70968884e-01 -4.11957920e-01 -8.31727505e-01 7.45235622e-01
6.25661194e-01 -6.83475912e-01 -2.83543348e-01 1.92373797e-01
4.20170307e-01 2.67812401e-01 4.36795801e-01 -9.06761944e-01
2.16148511e-01 -6.36903122e-02 5.29673278e-01 -1.08108938e+00
1.91266055e-03 -1.71197265e-01 7.72555053e-01 5.57422101e-01
-5.18376589e-01 3.28856170e-01 4.18737620e-01 7.12746024e-01
-1.52233720e-01 2.67075356e-02 3.24284106e-01 -2.66063094e-01
-5.67190528e-01 -2.33972728e-01 -5.03305435e-01 2.33647123e-01
9.68860745e-01 -7.78017491e-02 -8.93257499e-01 -7.05727696e-01
-7.43184328e-01 1.22606590e-01 4.13212001e-01 4.12064284e-01
2.94480234e-01 -1.21980274e+00 -1.24465621e+00 -4.24666464e-01
-9.62518454e-02 -4.18686688e-01 2.10925817e-01 6.10816896e-01
-6.91029727e-01 2.13105455e-01 -4.72808063e-01 -3.17914218e-01
-1.41966701e+00 3.60651284e-01 -3.47344905e-01 -5.74077785e-01
-1.01303220e+00 4.39316690e-01 1.19335532e-01 2.24557891e-01
3.83902714e-03 1.00822203e-01 -6.33306980e-01 8.30234766e-01
9.85722423e-01 5.53275585e-01 -2.22500995e-01 -6.04268491e-01
-2.98907965e-01 2.52291024e-01 -2.45571241e-01 -6.14028513e-01
1.78339469e+00 -5.04796691e-02 7.30990544e-02 8.52342546e-01
8.68065655e-01 5.89147449e-01 -8.48878980e-01 -4.03985120e-02
2.19326422e-01 -5.26756644e-02 -3.05033654e-01 -1.14798212e+00
-4.35519785e-01 7.36926675e-01 2.89783150e-01 6.65389299e-01
9.92167652e-01 3.51500928e-01 6.72338843e-01 1.96111407e-02
-7.41994902e-02 -6.87144935e-01 -2.22973213e-01 1.77170992e-01
1.33138359e+00 -8.48294795e-01 4.54539627e-01 -7.86992833e-02
-6.94108605e-01 1.15373027e+00 -2.32854053e-01 1.05164573e-01
4.07126881e-02 2.60143369e-01 1.27570570e-01 -5.89964628e-01
-5.72387040e-01 1.40910391e-02 3.21179301e-01 3.94188583e-01
2.80031949e-01 -7.20371306e-02 -6.86504126e-01 3.66808534e-01
-5.18204927e-01 -3.68115634e-01 7.46005535e-01 1.08493996e+00
-1.25110924e-01 -9.07069623e-01 -6.21466994e-01 3.96164387e-01
-8.96817207e-01 -1.36966154e-01 -7.16080785e-01 1.07963538e+00
-1.67263359e-01 5.67763805e-01 3.11119914e-01 3.12280823e-02
4.52343822e-02 2.21848026e-01 3.24973851e-01 -3.60336304e-01
-9.76612449e-01 1.38708577e-01 3.95319819e-01 -1.78476483e-01
-6.32113874e-01 -8.93625736e-01 -1.15381885e+00 -4.81460065e-01
1.97424397e-01 3.49515855e-01 5.53016484e-01 7.46869445e-01
3.04752320e-01 6.60210073e-01 2.28438586e-01 -7.90854692e-01
1.40288606e-01 -1.08083212e+00 -1.29346445e-01 6.39183104e-01
3.44393373e-01 -2.14668721e-01 -1.73528865e-01 4.37325567e-01]
|
[11.265583992004395, 9.0396146774292]
|
21a0340a-0cb0-4ac2-80ff-4f46ac54ad91
|
initialization-and-regularization-of-1
|
2105.01029
| null |
https://arxiv.org/abs/2105.01029v2
|
https://arxiv.org/pdf/2105.01029v2.pdf
|
Initialization and Regularization of Factorized Neural Layers
|
Factorized layers--operations parameterized by products of two or more matrices--occur in a variety of deep learning contexts, including compressed model training, certain types of knowledge distillation, and multi-head self-attention architectures. We study how to initialize and regularize deep nets containing such layers, examining two simple, understudied schemes, spectral initialization and Frobenius decay, for improving their performance. The guiding insight is to design optimization routines for these networks that are as close as possible to that of their well-tuned, non-decomposed counterparts; we back this intuition with an analysis of how the initialization and regularization schemes impact training with gradient descent, drawing on modern attempts to understand the interplay of weight-decay and batch-normalization. Empirically, we highlight the benefits of spectral initialization and Frobenius decay across a variety of settings. In model compression, we show that they enable low-rank methods to significantly outperform both unstructured sparsity and tensor methods on the task of training low-memory residual networks; analogs of the schemes also improve the performance of tensor decomposition techniques. For knowledge distillation, Frobenius decay enables a simple, overcomplete baseline that yields a compact model from over-parameterized training without requiring retraining with or pruning a teacher network. Finally, we show how both schemes applied to multi-head attention lead to improved performance on both translation and unsupervised pre-training.
|
['Nicolò Fusi', 'Lester Mackey', 'Neil Tenenholtz', 'Mikhail Khodak']
|
2021-05-03
|
initialization-and-regularization-of
|
https://openreview.net/forum?id=KTlJT1nof6d
|
https://openreview.net/pdf?id=KTlJT1nof6d
|
iclr-2021-1
|
['unsupervised-pre-training']
|
['methodology']
|
[ 3.54382336e-01 2.17792317e-01 -3.78001541e-01 -3.15411925e-01
-6.39118791e-01 -4.86088306e-01 7.50493884e-01 -7.87085891e-02
-6.23326421e-01 3.34717929e-01 7.78171837e-01 -4.80807036e-01
-2.69123226e-01 -3.98873955e-01 -9.69423771e-01 -7.28472054e-01
-1.91544235e-01 4.34437573e-01 -3.32061112e-01 -2.90951997e-01
-3.99378501e-02 5.04094005e-01 -1.19473839e+00 3.85551006e-01
4.14844662e-01 8.38292956e-01 -1.53701127e-01 5.96983016e-01
2.70611972e-01 8.49632621e-01 -2.04952344e-01 -6.39140129e-01
4.46300060e-01 -5.73033132e-02 -1.04077089e+00 1.46991611e-01
6.82029188e-01 -4.18855965e-01 -5.88742197e-01 7.89275169e-01
3.06430906e-01 4.61535126e-01 6.41280234e-01 -6.35060966e-01
-8.70959461e-01 1.08913946e+00 -4.12402332e-01 4.37527269e-01
-1.74213052e-01 2.10565552e-01 1.30490506e+00 -9.53377128e-01
4.32386518e-01 1.20406806e+00 9.83827949e-01 3.21960598e-01
-1.78436828e+00 -3.70093733e-01 1.12081468e-01 -1.75429750e-02
-1.24815774e+00 -8.09091151e-01 5.13751626e-01 -5.92253029e-01
1.11247551e+00 1.25077531e-01 4.46721941e-01 1.15570927e+00
-1.50113136e-01 7.16106236e-01 6.23777211e-01 -4.87597197e-01
-1.49931148e-01 -6.08049566e-03 3.89346063e-01 7.88453102e-01
5.35403311e-01 -3.69540565e-02 -6.12474501e-01 -2.54904896e-01
8.20547223e-01 6.74372166e-02 -5.01535177e-01 -4.82756704e-01
-1.40180635e+00 8.80149662e-01 2.96030492e-01 3.48471195e-01
-3.06065351e-01 4.79175210e-01 5.91000795e-01 3.25021058e-01
5.28156936e-01 8.24706376e-01 -7.09869921e-01 2.33755838e-02
-1.14379084e+00 1.39990702e-01 8.93539131e-01 7.24972546e-01
8.56891692e-01 4.34022069e-01 -2.03692392e-01 8.26757193e-01
-1.08901680e-01 1.21479221e-01 6.52426302e-01 -1.07853961e+00
5.15979648e-01 2.00876892e-01 -2.00768694e-01 -9.00887787e-01
-4.64139670e-01 -1.02326083e+00 -1.02362967e+00 -3.22552294e-01
4.54378068e-01 -1.45226315e-01 -9.44102824e-01 1.97150087e+00
-2.39198487e-02 8.91405046e-02 -2.65178848e-02 7.73055911e-01
2.72926182e-01 4.08876121e-01 -8.58873129e-02 7.55314948e-03
1.35538423e+00 -1.06283569e+00 -4.51867253e-01 -2.45055452e-01
1.00298822e+00 -6.80837989e-01 1.28222835e+00 4.69555140e-01
-1.22378135e+00 -2.95910567e-01 -1.01822293e+00 -6.37196839e-01
-1.96008071e-01 2.32965454e-01 1.08521247e+00 4.32006389e-01
-1.13117146e+00 1.14081335e+00 -1.14358997e+00 -2.98501194e-01
4.42879856e-01 3.67046982e-01 -4.56227511e-01 -2.71445632e-01
-9.90491390e-01 9.88012016e-01 3.31845343e-01 2.39960879e-01
-8.47299695e-01 -9.11710918e-01 -7.51318097e-01 2.64730424e-01
2.84073532e-01 -1.10242963e+00 1.26137292e+00 -8.26730728e-01
-1.32924545e+00 6.77730680e-01 -2.94800680e-02 -7.30366290e-01
7.20314533e-02 -5.40001869e-01 -7.36690611e-02 4.55558151e-02
-1.07338831e-01 4.04319912e-01 1.06883037e+00 -8.06537211e-01
7.72336796e-02 -2.99625307e-01 2.83617288e-01 1.66395247e-01
-7.28995085e-01 -1.76248237e-01 -5.39169848e-01 -8.96090567e-01
1.53918013e-01 -9.13360476e-01 -4.37363952e-01 -3.07230473e-01
-4.61353809e-01 1.54309735e-01 4.88936990e-01 -7.86536694e-01
1.19379807e+00 -2.09403300e+00 5.18146813e-01 2.83629805e-01
4.99653369e-01 1.80766732e-01 -4.92448360e-01 4.33672637e-01
-3.74626547e-01 1.23458743e-01 -2.80200720e-01 -6.43014789e-01
5.29754721e-02 4.42965895e-01 -4.59159255e-01 5.08403897e-01
3.26397091e-01 8.26332152e-01 -7.03540206e-01 -2.70496793e-02
-1.75693884e-01 7.79376984e-01 -1.05799246e+00 -9.37872380e-02
-1.83781758e-01 2.25333542e-01 -7.11803511e-02 2.35580444e-01
2.35764474e-01 -5.96915841e-01 2.74245411e-01 -6.68294370e-01
1.60385802e-01 7.20425665e-01 -1.08072019e+00 1.79925525e+00
-4.93955255e-01 6.48955345e-01 3.17789346e-01 -1.09285498e+00
2.58709699e-01 1.73231274e-01 3.60217631e-01 -8.06678087e-02
2.03701317e-01 1.81631371e-01 9.08581465e-02 -4.00126368e-01
6.66979015e-01 -2.49213800e-01 4.31078225e-01 6.23713672e-01
3.42142403e-01 2.38995906e-02 3.29181433e-01 5.33241153e-01
1.19289291e+00 1.00276567e-01 -1.61309078e-01 -3.18119735e-01
9.63620692e-02 -2.37292945e-01 4.41229314e-01 8.27116966e-01
3.85116726e-01 5.49049318e-01 6.32727087e-01 -4.31677729e-01
-1.25594592e+00 -6.17632389e-01 -1.60088688e-01 1.55926514e+00
-6.53280675e-01 -8.79597902e-01 -6.32636487e-01 -1.76915601e-01
1.35915041e-01 5.66438079e-01 -7.21018374e-01 -4.40913528e-01
-7.20530391e-01 -1.11483359e+00 7.20364273e-01 6.83522463e-01
1.52873620e-01 -4.49937761e-01 -1.31153733e-01 8.38064849e-02
6.33651298e-03 -1.18630159e+00 -6.62032664e-01 6.93563044e-01
-1.23026824e+00 -8.25935781e-01 -6.10493243e-01 -5.27575016e-01
6.74980760e-01 2.95480818e-01 1.20796800e+00 2.01975718e-01
1.72642857e-01 4.54126298e-01 -7.53195137e-02 2.55035013e-01
-2.01735094e-01 5.14695644e-01 2.01623112e-01 1.32439926e-01
8.94508362e-02 -1.06548190e+00 -3.88739020e-01 -4.41218987e-02
-1.25462127e+00 1.21404253e-01 8.45496893e-01 1.08725846e+00
2.91323453e-01 -1.61398187e-01 -1.91415831e-01 -1.07424521e+00
7.86354005e-01 -3.27676326e-01 -3.51549745e-01 1.21666700e-01
-6.35560274e-01 8.87638092e-01 6.09052002e-01 -6.37053490e-01
-5.42597115e-01 -1.68564707e-01 1.07719824e-01 -8.09601724e-01
4.52997804e-01 8.21591914e-01 2.05849871e-01 -2.12221891e-01
9.06613171e-01 1.00187555e-01 3.13156769e-02 -8.43825102e-01
9.43365514e-01 4.92136739e-02 6.85891926e-01 -1.12161899e+00
1.04694855e+00 4.17585611e-01 2.11298503e-02 -6.80773795e-01
-1.21938026e+00 -1.88751951e-01 -5.19894242e-01 6.16117358e-01
5.76101124e-01 -1.07945776e+00 -5.60986578e-01 1.11159138e-01
-1.03700590e+00 -5.08397222e-01 -5.07665455e-01 6.67278111e-01
-3.30880672e-01 4.40093726e-01 -1.05574155e+00 -2.21833184e-01
-2.72230953e-01 -1.15003693e+00 8.66933823e-01 -3.48482132e-01
-2.72871181e-02 -1.04403162e+00 -3.44534628e-02 5.13321579e-01
8.46188068e-01 -2.22465798e-01 1.20971346e+00 -7.48140395e-01
-5.99327266e-01 4.22195531e-02 -1.91449136e-01 6.61561847e-01
-3.40484947e-01 -1.68015584e-01 -9.43371475e-01 -4.77441549e-01
3.96479741e-02 -4.65686351e-01 1.43050408e+00 2.02186197e-01
1.12862277e+00 -7.82259107e-01 -1.17716286e-02 1.28076231e+00
1.14360571e+00 -5.94879150e-01 2.25000769e-01 2.06729501e-01
1.10112178e+00 3.53507668e-01 -4.27589476e-01 2.75152326e-01
3.63201231e-01 4.78928536e-01 1.62190154e-01 -1.57753248e-02
-6.60555884e-02 -1.65097743e-01 2.97370166e-01 1.25385523e+00
-4.24376190e-01 3.74622315e-01 -9.40984607e-01 4.09335345e-01
-1.67836511e+00 -9.12343562e-01 2.16814548e-01 2.28238368e+00
1.26266778e+00 1.73989147e-01 1.51777297e-01 9.73269343e-03
3.22652310e-01 4.52524930e-01 -5.41644037e-01 -2.62040555e-01
-3.13988686e-01 4.53018636e-01 8.66378546e-01 7.77980268e-01
-1.09345114e+00 1.01603925e+00 6.40391684e+00 7.01324701e-01
-1.19442773e+00 5.25433302e-01 6.26401126e-01 -3.39946300e-01
-5.28486729e-01 1.76371709e-01 -6.96881354e-01 -1.47803530e-01
1.21011293e+00 1.33348048e-01 9.98692095e-01 8.19005668e-01
5.28478697e-02 3.71811301e-01 -1.54851019e+00 8.39747846e-01
6.98300749e-02 -1.59403336e+00 1.82891309e-01 4.12709340e-02
8.00639212e-01 4.28399980e-01 2.13221118e-01 5.07521749e-01
6.77125096e-01 -1.04567087e+00 6.48980021e-01 2.67842382e-01
7.57173240e-01 -3.70037585e-01 2.60481715e-01 3.32146399e-02
-8.59278262e-01 -3.02666098e-01 -4.36784625e-01 -1.43101230e-01
2.58111022e-02 8.40687096e-01 -5.56907356e-01 2.29242370e-01
3.09894651e-01 6.03957713e-01 -5.76018989e-01 6.44972801e-01
-9.10726786e-02 7.97029138e-01 -5.56629241e-01 5.08007169e-01
4.43662643e-01 -1.90596908e-01 4.27368045e-01 1.38980949e+00
-7.11742640e-02 -2.77072787e-02 2.74396874e-03 8.13203573e-01
-2.95765579e-01 -8.75982568e-02 -3.41590583e-01 -3.83651346e-01
1.76869392e-01 1.07336211e+00 -2.55386621e-01 -4.02324587e-01
-4.74187374e-01 7.48474538e-01 7.97570884e-01 7.80053020e-01
-6.13583922e-01 -1.43970817e-01 8.12291205e-01 3.70927989e-01
5.94130695e-01 -6.62562072e-01 -3.57255369e-01 -1.74590886e+00
-2.90672127e-02 -1.09853995e+00 2.98804760e-01 -5.94468772e-01
-1.14833307e+00 5.36875427e-01 1.11362904e-01 -6.44863009e-01
-1.81005746e-01 -6.61527693e-01 -3.37078124e-01 7.82642961e-01
-1.57424283e+00 -1.21764898e+00 9.03076082e-02 6.81032658e-01
1.62602574e-01 -5.73872626e-02 6.07519805e-01 5.58928370e-01
-8.85366201e-01 7.07980394e-01 3.10863405e-01 3.14649642e-01
5.51786304e-01 -1.00061989e+00 3.42603981e-01 1.03236270e+00
3.93142551e-01 1.31406140e+00 6.78115427e-01 -2.18799651e-01
-1.73175585e+00 -1.08803201e+00 6.90513372e-01 -3.83580536e-01
1.09773183e+00 -3.35282117e-01 -8.88009250e-01 1.22142613e+00
-5.34113438e-04 1.35238722e-01 4.93897617e-01 7.30423629e-01
-9.38501239e-01 -1.61152527e-01 -5.17209172e-01 6.81082785e-01
1.02181208e+00 -9.10655141e-01 -5.45535088e-01 7.52013743e-01
9.82728779e-01 -3.72150272e-01 -9.98315454e-01 2.61046439e-01
5.17979860e-01 -7.37931490e-01 1.11221230e+00 -1.19355202e+00
5.09208739e-01 4.74251769e-02 -3.87881458e-01 -1.18704081e+00
-7.62743354e-01 -1.06038153e+00 -4.14225966e-01 7.90098131e-01
5.44742763e-01 -6.30405068e-01 8.08550775e-01 7.74425089e-01
-3.91100019e-01 -9.35213566e-01 -5.72703063e-01 -6.13250077e-01
2.71967977e-01 -4.82095420e-01 3.46046448e-01 1.19652402e+00
-1.59721389e-01 6.72903419e-01 -4.93016243e-01 1.62355416e-02
5.40574431e-01 -2.68165797e-01 7.61554360e-01 -9.63776529e-01
-8.94165754e-01 -5.59569359e-01 -1.46507636e-01 -1.36912942e+00
1.18353985e-01 -1.26913857e+00 -5.37593305e-01 -1.13791037e+00
2.30022565e-01 -2.89992273e-01 -3.41889769e-01 8.69898558e-01
3.38646770e-02 1.63189188e-01 1.92189604e-01 5.16383708e-01
-2.63286620e-01 4.47623342e-01 9.81358111e-01 -1.61250398e-01
-1.21860690e-01 -2.81148821e-01 -1.04354846e+00 6.36952341e-01
4.47602212e-01 -4.32463169e-01 -3.32978159e-01 -9.88541305e-01
5.24738908e-01 -1.22440450e-01 3.59481335e-01 -7.73747742e-01
2.86787391e-01 6.34661838e-02 1.19523488e-01 -1.60328865e-01
4.96075690e-01 -5.54642498e-01 6.50737733e-02 2.60319859e-01
-6.50208235e-01 2.12762371e-01 1.11888535e-01 4.21261966e-01
7.99702480e-02 -2.18092352e-01 7.17483103e-01 -2.77245879e-01
-1.55637518e-01 3.45029414e-01 -8.26700255e-02 2.31555134e-01
9.89488214e-02 -1.01081297e-01 -2.96196878e-01 -3.91064733e-01
-9.68516469e-01 -4.13351133e-02 2.17306361e-01 6.53381944e-02
1.84975639e-01 -1.17307281e+00 -6.34590149e-01 3.73366565e-01
-3.58735234e-01 1.12847656e-01 -2.14084946e-02 1.32359624e+00
-3.85628700e-01 4.89940792e-01 1.49439991e-01 -4.02862698e-01
-7.32084095e-01 5.29360950e-01 3.90609592e-01 -6.05637252e-01
-5.49642444e-01 1.04499912e+00 2.76605815e-01 -4.94012386e-01
5.09858191e-01 -7.69619644e-01 3.55152071e-01 3.57707264e-03
1.94988906e-01 2.76006997e-01 3.65542948e-01 -4.37438995e-01
-1.57042835e-02 3.99392188e-01 -4.20250088e-01 -5.76060563e-02
1.58118165e+00 1.67065173e-01 -2.76951104e-01 3.12312245e-01
1.41529167e+00 -2.36055162e-02 -1.12330568e+00 -6.85217261e-01
-9.47615877e-02 -8.56992006e-02 3.52127463e-01 -4.82898206e-01
-1.29047596e+00 8.97180438e-01 4.08622473e-02 3.40557694e-02
8.76227856e-01 -2.10156322e-01 7.12665439e-01 9.95093942e-01
-7.34007582e-02 -7.50511646e-01 1.28928572e-01 8.14987481e-01
7.98011482e-01 -9.40589845e-01 3.87651145e-01 -2.35479921e-02
-2.80025452e-01 1.04903924e+00 2.19528645e-01 -1.25760362e-01
8.12858939e-01 3.04223716e-01 -3.60455424e-01 -2.94025362e-01
-9.72074270e-01 -8.16622227e-02 3.43103439e-01 2.02209875e-01
4.98157591e-01 -1.66455731e-01 8.25223234e-03 5.25116682e-01
-4.51473832e-01 -1.81453004e-01 4.58507299e-01 6.69285953e-01
-2.88784176e-01 -9.86876130e-01 -3.87517065e-01 6.91582322e-01
-7.14584827e-01 -6.49961770e-01 3.00613977e-02 6.87890589e-01
-1.39442846e-01 5.19205034e-01 -8.62696096e-02 -5.19550502e-01
3.76680195e-02 3.04557741e-01 5.97598016e-01 -8.16830039e-01
-7.73448229e-01 1.96828559e-01 2.86082774e-01 -6.55761182e-01
-3.13726664e-01 -5.13801277e-01 -7.63041019e-01 -5.49446702e-01
-4.93265778e-01 9.27110463e-02 6.08558178e-01 1.00928080e+00
5.55136085e-01 3.34589750e-01 8.37593898e-02 -1.26050651e+00
-1.11910963e+00 -1.02980721e+00 -4.54906225e-01 4.87488002e-01
5.44961929e-01 -5.06341696e-01 -6.37686014e-01 1.27264559e-01]
|
[8.401971817016602, 3.5988972187042236]
|
802ccce4-a4dc-427d-9a4e-4d06d5701b98
|
tax2vec-constructing-interpretable-features
|
1902.00438
| null |
https://arxiv.org/abs/1902.00438v3
|
https://arxiv.org/pdf/1902.00438v3.pdf
|
tax2vec: Constructing Interpretable Features from Taxonomies for Short Text Classification
|
The use of background knowledge is largely unexploited in text classification tasks. This paper explores word taxonomies as means for constructing new semantic features, which may improve the performance and robustness of the learned classifiers. We propose tax2vec, a parallel algorithm for constructing taxonomy-based features, and demonstrate its use on six short text classification problems: prediction of gender, personality type, age, news topics, drug side effects and drug effectiveness. The constructed semantic features, in combination with fast linear classifiers, tested against strong baselines such as hierarchical attention neural networks, achieves comparable classification results on short text documents. The algorithm's performance is also tested in a few-shot learning setting, indicating that the inclusion of semantic features can improve the performance in data-scarce situations. The tax2vec capability to extract corpus-specific semantic keywords is also demonstrated. Finally, we investigate the semantic space of potential features, where we observe a similarity with the well known Zipf's law.
|
['Senja Pollak', 'Nada Lavrač', 'Jan Kralj', 'Matej Martinc', 'Blaž Škrlj']
|
2019-02-01
| null | null | null | null |
['type-prediction']
|
['computer-code']
|
[ 1.03616714e-01 1.94337904e-01 -5.76610744e-01 -4.57706034e-01
-4.74143296e-01 -3.02824706e-01 1.09691000e+00 7.54215896e-01
-7.55856454e-01 8.50382328e-01 6.36228681e-01 -1.63145244e-01
-3.66504550e-01 -7.74394691e-01 -9.13084745e-02 -6.21534586e-01
-1.06717581e-02 5.48171222e-01 -3.92680205e-02 -4.54956353e-01
6.30074799e-01 1.46667704e-01 -1.89601135e+00 2.53884256e-01
7.41845787e-01 1.01836014e+00 1.30915716e-01 5.03527403e-01
-5.24516344e-01 5.98259926e-01 -5.93891263e-01 -7.40162849e-01
-1.60979152e-01 -5.43744750e-02 -1.05515838e+00 -1.31492630e-01
2.65864968e-01 -5.72171547e-02 -1.97921470e-01 7.56818056e-01
5.00606418e-01 2.79102176e-01 9.91484225e-01 -1.10401034e+00
-6.24184668e-01 6.75249338e-01 -2.13841587e-01 3.49375397e-01
4.35845047e-01 -2.58938134e-01 1.42913663e+00 -8.04248512e-01
8.61273408e-01 1.39905083e+00 7.63220370e-01 4.29132760e-01
-8.60329986e-01 -4.97435331e-01 8.00917298e-02 4.61969167e-01
-1.23290384e+00 -2.85288215e-01 5.46103179e-01 -5.61285675e-01
1.03416967e+00 1.62557662e-01 1.93803653e-01 1.49912906e+00
4.02044058e-01 5.97118974e-01 1.06214905e+00 -5.01918674e-01
3.41579676e-01 5.89149654e-01 8.19764256e-01 6.92621529e-01
2.93310702e-01 -1.21381968e-01 -6.73895001e-01 -4.23845977e-01
9.68375430e-02 7.12073967e-02 -4.70092185e-02 -1.12454839e-01
-8.67523789e-01 1.26596284e+00 2.75560021e-01 6.92909956e-01
-1.65904596e-01 -1.42264217e-01 8.64915848e-01 2.54520446e-01
1.00396132e+00 7.98130989e-01 -7.58543849e-01 -1.85387895e-01
-7.03779638e-01 1.90347880e-01 8.85539889e-01 6.78395987e-01
6.38355553e-01 -1.56476036e-01 -5.05908370e-01 1.03935313e+00
-3.39247994e-02 1.60571873e-01 1.09063864e+00 -3.37190658e-01
3.65250945e-01 6.98921561e-01 -3.32126208e-02 -1.08716118e+00
-9.69465792e-01 -4.43184048e-01 -4.92621422e-01 -3.71514678e-01
3.00218284e-01 -1.58301339e-01 -7.30065405e-01 1.52089345e+00
3.07314813e-01 8.56815949e-02 1.20597325e-01 5.34259260e-01
9.93366301e-01 3.97731870e-01 3.38978201e-01 -3.27337325e-01
1.72173548e+00 -6.60801828e-01 -7.83247888e-01 -3.00042704e-02
8.75552118e-01 -4.63328004e-01 9.16735709e-01 2.05739379e-01
-5.97283959e-01 -5.10994136e-01 -7.54350305e-01 -2.85475373e-01
-1.09575057e+00 -2.64492799e-02 9.89623308e-01 8.26496542e-01
-6.51730239e-01 9.47432637e-01 -4.56221133e-01 -7.24032164e-01
5.55222809e-01 4.06471103e-01 -3.29552472e-01 -1.01523697e-01
-1.46025085e+00 1.02059746e+00 4.15911973e-01 -6.45938158e-01
-4.93529588e-01 -7.07454562e-01 -8.86760354e-01 4.38484341e-01
2.15062484e-01 -7.80049026e-01 8.79171789e-01 -7.11346865e-01
-1.26831830e+00 8.10652375e-01 -1.66267052e-01 -5.65423727e-01
2.52293169e-01 1.04566827e-01 -4.25564378e-01 1.34138480e-01
2.81045765e-01 5.09357810e-01 8.65826547e-01 -6.99670196e-01
-7.93112338e-01 -7.65894711e-01 -9.71923321e-02 2.31501758e-01
-1.18320954e+00 -3.71302180e-02 -1.73343346e-01 -5.72745562e-01
-4.74346936e-01 -6.39320076e-01 -2.96071649e-01 -5.08935034e-01
-2.58925974e-01 -8.25720608e-01 4.10057068e-01 -5.35645247e-01
1.22935486e+00 -1.90177345e+00 3.21143791e-02 2.45565683e-01
1.44549429e-01 1.20590096e-02 -2.18234777e-01 6.30980134e-01
6.98162094e-02 1.68493077e-01 9.56530198e-02 -1.90128714e-01
5.89984730e-02 6.33594170e-02 -9.66012925e-02 3.31697524e-01
7.53290951e-02 1.05829644e+00 -7.63732612e-01 -2.79547304e-01
-6.96316035e-03 3.04287523e-01 -5.42593002e-01 -1.33784622e-01
-1.02103241e-01 -9.04518440e-02 -6.35043740e-01 4.46853936e-01
1.93716139e-01 -1.91623360e-01 1.80325866e-01 1.00792162e-01
9.57217067e-02 3.08665961e-01 -6.95024073e-01 1.51446748e+00
-6.06237531e-01 5.38222551e-01 -4.45827007e-01 -1.28400064e+00
9.71582294e-01 2.55467802e-01 4.45462227e-01 -7.89131463e-01
4.81930256e-01 -1.68492153e-01 -3.32214870e-02 -6.42488420e-01
6.21133029e-01 -9.85850543e-02 -1.68428048e-01 3.72064412e-01
4.94217157e-01 1.12840518e-01 2.48400345e-01 1.77696645e-01
1.01386595e+00 -2.92161316e-01 5.83367884e-01 -7.87552655e-01
4.69583094e-01 1.00603392e-02 1.56237751e-01 9.42557454e-01
2.38255248e-03 2.13145494e-01 6.37867451e-01 -3.37157667e-01
-9.89731312e-01 -3.70106697e-01 -6.73927307e-01 1.71826768e+00
-1.12487599e-02 -8.05306017e-01 -6.62224829e-01 -9.12183344e-01
2.33578607e-01 8.37115049e-01 -1.10280359e+00 -2.83919364e-01
9.43540111e-02 -1.06123424e+00 3.73106152e-01 5.61685801e-01
-7.68270567e-02 -8.58497977e-01 -3.97727072e-01 1.95897534e-01
2.10851490e-01 -1.02002859e+00 -4.97490503e-02 4.63187933e-01
-8.03034604e-01 -1.12906218e+00 -6.94176078e-01 -5.93585610e-01
2.78661937e-01 1.98443770e-01 9.59688067e-01 1.92646869e-02
-4.52362835e-01 6.04662716e-01 -6.80942476e-01 -7.05172539e-01
-1.43908948e-01 4.87750500e-01 1.78959101e-01 -1.36359438e-01
7.87776113e-01 -2.85109460e-01 -3.54472637e-01 -4.21218164e-02
-7.18892813e-01 -3.02679509e-01 1.04433581e-01 1.27759588e+00
-1.22378230e-01 1.21324264e-01 9.63143110e-01 -1.27615321e+00
9.16550934e-01 -7.09281385e-01 -9.62552894e-03 3.21889482e-02
-9.49427307e-01 1.91501409e-01 6.86893344e-01 -4.01124507e-01
-1.02679443e+00 -3.94681484e-01 -3.18299174e-01 1.83879197e-01
-3.10307831e-01 6.20102286e-01 1.28951028e-01 1.12810113e-01
8.54745209e-01 1.23943791e-01 -2.68859025e-02 -6.18895054e-01
3.80553633e-01 1.04770362e+00 1.78373735e-02 -4.50453430e-01
4.03601140e-01 3.14662367e-01 -3.49934734e-02 -1.16770875e+00
-1.08447456e+00 -9.05541003e-01 -6.25851750e-01 3.84803295e-01
7.83264637e-01 -8.44492376e-01 -6.45985067e-01 4.80994396e-02
-9.33880150e-01 6.52493015e-02 -7.33156577e-02 4.85352844e-01
-4.55743492e-01 2.76746601e-01 -6.84223354e-01 -5.97075820e-01
-5.31797588e-01 -8.74490440e-01 1.10189199e+00 1.15262292e-01
-4.13355052e-01 -1.32011139e+00 -8.40463862e-02 5.07420599e-01
2.17025340e-01 -2.67703712e-01 1.34022009e+00 -1.54410684e+00
3.24931145e-01 -2.77673572e-01 -2.14089558e-01 3.74456942e-02
-2.79392600e-02 -3.78022552e-01 -1.17087722e+00 -2.62399435e-01
-1.98301598e-01 -4.01998222e-01 1.14781046e+00 4.43241805e-01
1.31187749e+00 -2.80171096e-01 -4.42474216e-01 3.62375081e-01
1.14140224e+00 1.64013878e-01 3.41135174e-01 3.38017702e-01
5.89449048e-01 8.76457155e-01 5.55544913e-01 9.49587882e-01
3.59400421e-01 6.16982400e-01 4.10170592e-02 1.17858183e-02
1.56002328e-01 -1.01501979e-01 5.78517057e-02 6.51687264e-01
-1.65364176e-01 -7.62432143e-02 -8.41231227e-01 3.18408787e-01
-1.67879450e+00 -8.50287378e-01 -5.34850918e-02 1.94927025e+00
7.10465252e-01 8.21479261e-02 1.62700459e-01 1.82224527e-01
5.89764833e-01 2.08535716e-02 -3.03648621e-01 -6.90431237e-01
2.31461637e-02 5.75939476e-01 5.48513710e-01 2.91502059e-01
-1.21378648e+00 1.13604355e+00 6.37667990e+00 1.19518316e+00
-9.78441119e-01 2.01620042e-01 7.62212455e-01 2.85237372e-01
-2.65019298e-01 -4.57253069e-01 -1.01628077e+00 3.76957506e-01
1.14349532e+00 -5.64228475e-01 7.28231743e-02 9.45090950e-01
6.93207458e-02 9.94585007e-02 -1.08156824e+00 8.53566110e-01
5.34302711e-01 -1.33906043e+00 8.22206959e-02 5.82922921e-02
6.63367927e-01 -6.83631226e-02 1.09301560e-01 6.72856987e-01
2.58259088e-01 -1.06523013e+00 1.09283976e-01 1.22233436e-01
6.93482816e-01 -9.00253236e-01 9.72153544e-01 9.10445899e-02
-7.70456016e-01 -4.75628406e-01 -7.54626215e-01 -2.45489582e-01
-3.94319832e-01 4.20318186e-01 -1.20707631e+00 6.22942090e-01
4.84075636e-01 1.00348282e+00 -1.04063010e+00 8.53044629e-01
1.11281328e-01 6.08518600e-01 1.29278690e-01 -6.01459622e-01
4.88025844e-01 9.98547673e-02 1.83765292e-01 1.42523098e+00
2.60980695e-01 -1.68421976e-02 1.25735402e-01 3.71488243e-01
-9.81345400e-02 9.95035410e-01 -7.54490554e-01 -2.65687406e-01
2.22160995e-01 1.28624809e+00 -9.36972558e-01 -4.83354747e-01
-5.64853847e-01 8.78509343e-01 1.96428999e-01 3.07417035e-01
-5.45655370e-01 -5.43764055e-01 6.08556926e-01 -1.31786138e-01
2.30102465e-01 2.61796117e-01 -5.38712084e-01 -1.22852457e+00
-5.33742964e-01 -7.66438186e-01 6.51289821e-01 -3.90887856e-01
-1.50789893e+00 3.68827730e-01 -1.47671148e-01 -8.92422318e-01
-1.27733469e-01 -8.72707427e-01 -5.56887746e-01 5.12126684e-01
-1.26320779e+00 -1.09267855e+00 -8.13059285e-02 5.78259885e-01
9.47608113e-01 -6.07395589e-01 1.08176434e+00 1.58989951e-02
-4.95007187e-01 6.91065013e-01 6.03638649e-01 -5.78386374e-02
7.27653861e-01 -1.36133301e+00 1.45797983e-01 5.99105768e-02
2.19911769e-01 4.47806627e-01 7.01441467e-01 -5.81752837e-01
-1.13721347e+00 -1.02536285e+00 1.07566273e+00 -4.46931094e-01
8.90242934e-01 -5.83866000e-01 -8.67055714e-01 3.07454169e-01
1.09260827e-01 -4.87862796e-01 1.08124816e+00 8.99217308e-01
-4.83245552e-01 1.28690958e-01 -1.05606771e+00 5.29571474e-01
9.13860440e-01 -4.20396954e-01 -7.71726668e-01 6.43611312e-01
7.39727139e-01 2.11684808e-01 -8.94844770e-01 1.71417147e-01
6.18691981e-01 -6.22409642e-01 1.06352127e+00 -1.22903287e+00
8.04127991e-01 6.97013557e-01 -3.62248905e-02 -1.51444125e+00
-5.34314573e-01 -9.77380350e-02 2.21901402e-01 1.27610946e+00
3.75357687e-01 -6.20283127e-01 6.73681617e-01 5.46470523e-01
5.79608344e-02 -7.46593058e-01 -8.98600578e-01 -7.55557418e-01
5.91725290e-01 -2.75758415e-01 4.19273198e-01 1.31313038e+00
5.61490238e-01 8.14834118e-01 -3.21860343e-01 -4.92343515e-01
2.18475342e-01 -2.75657643e-02 4.91372824e-01 -1.73745430e+00
-7.75965899e-02 -6.19993269e-01 -6.88920975e-01 -3.38532567e-01
6.44057274e-01 -1.31583655e+00 -3.61563712e-01 -1.24818778e+00
5.19345045e-01 -2.09281743e-01 -4.75347161e-01 3.86075109e-01
-4.08272445e-01 -2.51018982e-02 1.41876623e-01 -3.31131279e-01
-5.23035824e-01 7.09007084e-01 1.00770402e+00 -3.59164685e-01
-1.20523714e-01 1.73388775e-02 -9.44034934e-01 7.40467131e-01
7.60040760e-01 -4.71473396e-01 -3.71057838e-01 2.16787755e-01
-3.52514647e-02 -7.14401007e-02 -1.32787913e-01 -7.56304741e-01
9.89032164e-02 -1.69184864e-01 5.16293406e-01 -5.59665896e-02
2.35233456e-01 -5.80078304e-01 -5.99799514e-01 4.65306371e-01
-7.73581922e-01 -1.03642225e-01 1.40908867e-01 6.18057609e-01
-5.44507429e-02 -6.09635651e-01 4.36081231e-01 -1.04162179e-01
-7.17261910e-01 3.11969668e-01 -5.10085702e-01 1.38774291e-01
8.21364760e-01 -6.88588172e-02 -2.80455410e-01 -2.69074261e-01
-6.92054272e-01 1.37619421e-01 6.08453378e-02 7.40138352e-01
2.84128904e-01 -1.04524231e+00 -6.49918556e-01 -3.85324582e-02
3.80757928e-01 -8.60365689e-01 2.40712270e-01 6.27986789e-01
-6.79016858e-02 7.55005836e-01 -1.74279109e-01 -2.89290220e-01
-1.49032128e+00 9.22315180e-01 -1.80347979e-01 -3.51356119e-01
-5.52025914e-01 7.79009283e-01 3.84239823e-01 -2.28527576e-01
3.83805782e-01 -7.22703263e-02 -8.48109543e-01 7.66478240e-01
7.21545219e-01 3.90159518e-01 1.60500363e-01 -4.63927001e-01
-2.42000818e-01 3.55734944e-01 -3.39444280e-01 1.16147593e-01
1.45573437e+00 -1.89154789e-01 1.35924980e-01 4.63676691e-01
1.38726282e+00 -1.72950670e-01 -4.46751058e-01 -2.75590748e-01
5.48742294e-01 -3.99167418e-01 2.15677395e-01 -7.46521175e-01
-6.25251055e-01 9.78888929e-01 5.43684900e-01 3.00983012e-01
6.25650585e-01 3.85794863e-02 5.37295043e-01 5.76679409e-01
2.19350904e-01 -1.38626802e+00 1.57017969e-02 7.38501668e-01
5.02847135e-01 -1.43088174e+00 -3.12600657e-02 -2.37721652e-01
-8.37450027e-01 1.33175647e+00 4.70932484e-01 2.32651278e-01
6.70325875e-01 -4.96620312e-02 -1.92300349e-01 -2.66966492e-01
-9.97199059e-01 -6.15397155e-01 4.85915601e-01 5.74148834e-01
6.70921564e-01 8.47018138e-02 -8.46839070e-01 9.50693309e-01
-3.39443564e-01 -2.27188379e-01 5.45270920e-01 5.02740860e-01
-5.75374305e-01 -1.01102531e+00 -5.56127764e-02 1.03038776e+00
-8.07327807e-01 -3.76212448e-01 -4.08838093e-01 6.61298275e-01
2.15468004e-01 8.64571333e-01 1.44152671e-01 -2.68031389e-01
-1.69146191e-02 6.41237319e-01 2.30916277e-01 -9.26829576e-01
-8.03464413e-01 -7.56527260e-02 3.51062804e-01 -3.97849500e-01
-2.07156286e-01 -5.81808388e-01 -8.28858018e-01 -1.67835832e-01
-2.83804417e-01 3.49856287e-01 7.79842675e-01 1.20516419e+00
2.78005481e-01 4.51675713e-01 6.04977667e-01 -5.37884593e-01
-6.23135507e-01 -1.07590699e+00 -7.83016622e-01 9.40084577e-01
-5.00037372e-02 -9.16076601e-01 -2.24746197e-01 -9.96698588e-02]
|
[10.545265197753906, 7.829506874084473]
|
9c28e550-f2f8-4d61-b372-523bc4afcbf6
|
docformerv2-local-features-for-document
|
2306.01733
| null |
https://arxiv.org/abs/2306.01733v1
|
https://arxiv.org/pdf/2306.01733v1.pdf
|
DocFormerv2: Local Features for Document Understanding
|
We propose DocFormerv2, a multi-modal transformer for Visual Document Understanding (VDU). The VDU domain entails understanding documents (beyond mere OCR predictions) e.g., extracting information from a form, VQA for documents and other tasks. VDU is challenging as it needs a model to make sense of multiple modalities (visual, language and spatial) to make a prediction. Our approach, termed DocFormerv2 is an encoder-decoder transformer which takes as input - vision, language and spatial features. DocFormerv2 is pre-trained with unsupervised tasks employed asymmetrically i.e., two novel document tasks on encoder and one on the auto-regressive decoder. The unsupervised tasks have been carefully designed to ensure that the pre-training encourages local-feature alignment between multiple modalities. DocFormerv2 when evaluated on nine datasets shows state-of-the-art performance over strong baselines e.g. TabFact (4.3%), InfoVQA (1.4%), FUNSD (1%). Furthermore, to show generalization capabilities, on three VQA tasks involving scene-text, Doc- Formerv2 outperforms previous comparably-sized models and even does better than much larger models (such as GIT2, PaLi and Flamingo) on some tasks. Extensive ablations show that due to its pre-training, DocFormerv2 understands multiple modalities better than prior-art in VDU.
|
['R. Manmatha', 'Yichu Zhou', 'Nishant Sankaran', 'Qi Dong', 'Peng Tang', 'Srikar Appalaraju']
|
2023-06-02
| null | null | null | null |
['optical-character-recognition']
|
['computer-vision']
|
[ 3.19357038e-01 2.52545416e-01 4.78388220e-02 -4.28783268e-01
-1.07516265e+00 -1.01032627e+00 1.21493518e+00 4.08257544e-02
-1.46776959e-01 2.35089034e-01 6.93482637e-01 -6.33542001e-01
2.20275670e-01 -4.69677001e-01 -1.07520616e+00 -2.32651204e-01
3.25958788e-01 9.05665219e-01 1.48109183e-01 -1.56467631e-01
2.97564507e-01 9.39784944e-03 -1.25685513e+00 1.15189898e+00
6.03340805e-01 1.06034458e+00 6.02463901e-01 1.11657512e+00
-4.59112704e-01 9.78680789e-01 -3.97343934e-01 -5.28949082e-01
-1.20562300e-01 -2.74357110e-01 -1.17850804e+00 1.56406403e-01
1.04486096e+00 -5.72012782e-01 -3.40581387e-01 4.55330372e-01
2.24763453e-01 -3.02931969e-03 1.25228429e+00 -1.03450501e+00
-1.26980722e+00 5.53613245e-01 -5.95037878e-01 1.12832628e-01
5.34412622e-01 3.70771617e-01 1.36619842e+00 -1.19662356e+00
9.78299677e-01 1.73497856e+00 4.78615880e-01 6.52868092e-01
-1.33939219e+00 -1.63407311e-01 2.32996747e-01 2.01109216e-01
-1.06845689e+00 -5.38830936e-01 1.86081231e-01 -5.43934941e-01
1.58247566e+00 2.83975363e-01 1.03813201e-01 1.56853485e+00
3.47277671e-01 1.43941987e+00 1.11337173e+00 -4.62650657e-01
-1.49081901e-01 1.68573678e-01 1.96392745e-01 5.69583178e-01
-1.43450364e-01 -1.13722898e-01 -7.47779548e-01 3.23245049e-01
5.13945937e-01 -2.99033880e-01 -3.42124015e-01 -2.42398858e-01
-1.25855780e+00 5.86057127e-01 5.96543729e-01 2.09875688e-01
-5.64507209e-02 3.43472511e-01 4.65498120e-01 2.70642608e-01
3.99515837e-01 4.32179362e-01 -5.09236336e-01 -2.26405025e-01
-9.89278555e-01 1.80123374e-02 3.65895629e-01 1.07696700e+00
5.93578279e-01 6.83901459e-02 -5.36782920e-01 8.13762486e-01
5.73129177e-01 8.11724365e-01 3.25965047e-01 -7.18935013e-01
1.02188277e+00 4.70528573e-01 -1.35061909e-02 -5.07090569e-01
-3.15927804e-01 -2.54204988e-01 -5.39184153e-01 4.04672623e-02
4.97507423e-01 3.00313920e-01 -1.75455749e+00 1.42448926e+00
-3.20061147e-01 -2.31065407e-01 3.45977783e-01 1.01120627e+00
1.17944157e+00 1.17867672e+00 2.68820316e-01 2.45507196e-01
1.38865817e+00 -1.05720448e+00 -6.36728883e-01 -8.41017246e-01
6.66674018e-01 -9.20883536e-01 1.51032186e+00 4.06749010e-01
-1.15274334e+00 -7.50043809e-01 -9.06947255e-01 -7.42480993e-01
-6.35117352e-01 3.40998799e-01 3.31442207e-01 3.97860110e-01
-1.32951915e+00 5.28968088e-02 -5.57967365e-01 -5.70896864e-01
4.62110877e-01 4.95347939e-03 -5.72074354e-01 -4.40210015e-01
-9.14067745e-01 1.04114151e+00 4.59959835e-01 -2.65378654e-01
-1.44534171e+00 -6.32951319e-01 -1.08594370e+00 2.52159357e-01
2.50340164e-01 -7.86928356e-01 1.42463386e+00 -8.95423830e-01
-1.08461130e+00 9.36311245e-01 -5.55277824e-01 -4.77142483e-01
4.83184814e-01 -4.16190207e-01 -2.83885181e-01 2.90804297e-01
1.40189454e-01 1.04928482e+00 9.47439075e-01 -1.68260252e+00
-4.23133731e-01 -4.26220536e-01 6.52610511e-02 3.28269184e-01
-1.64861172e-01 -3.70935559e-01 -8.95811200e-01 -4.61751014e-01
-2.64267018e-03 -6.28905356e-01 5.26678920e-01 -2.22371414e-01
-7.38830507e-01 -2.13059530e-01 1.14627290e+00 -9.73982275e-01
8.00615668e-01 -2.02619910e+00 3.51229966e-01 1.08242951e-01
2.21132189e-01 2.20666945e-01 -7.07706451e-01 6.28508687e-01
4.74074185e-02 2.66477168e-01 -8.86621475e-02 -8.82141948e-01
8.35847184e-02 2.57797390e-01 -6.44516051e-01 2.04064190e-01
1.24513254e-01 1.31614602e+00 -5.82715750e-01 -4.23119962e-01
3.20338547e-01 5.50004125e-01 -4.20939535e-01 2.08764002e-01
-7.29349792e-01 1.61255985e-01 -1.69377074e-01 6.98449552e-01
5.76176882e-01 -5.89690745e-01 8.56565237e-02 -3.87899697e-01
6.98983744e-02 3.18575084e-01 -6.56709254e-01 1.92007256e+00
-8.25370371e-01 1.10451853e+00 -1.58390537e-01 -6.53857827e-01
5.83787143e-01 2.87314981e-01 -2.25510329e-01 -1.11856711e+00
2.05617368e-01 4.23121303e-02 -2.58005768e-01 -4.16043699e-01
6.50184810e-01 2.48741105e-01 1.81443349e-01 4.58166301e-01
5.64418852e-01 -2.18313545e-01 1.69031918e-01 1.04357290e+00
9.65337694e-01 2.40322351e-01 4.43785340e-02 -7.49325901e-02
3.22450042e-01 9.40350518e-02 -2.88927078e-01 1.05616236e+00
1.17493279e-01 9.07191932e-01 5.38986623e-01 5.84767908e-02
-1.16949844e+00 -1.23258567e+00 5.47140278e-02 1.31177890e+00
1.23537123e-01 -7.12273121e-01 -5.40937603e-01 -8.09323847e-01
2.11127669e-01 1.27930093e+00 -8.60844672e-01 1.20212706e-02
-4.80438024e-02 -5.22276424e-02 5.90592563e-01 9.38942373e-01
4.16586697e-01 -1.00355136e+00 -2.32803792e-01 -9.29399729e-02
-1.78537145e-01 -1.38161218e+00 -4.51133907e-01 3.49423811e-02
-5.88211536e-01 -9.32215810e-01 -6.36863947e-01 -6.09475791e-01
4.49697375e-01 4.44054425e-01 1.37570965e+00 -9.37299430e-02
-8.03421214e-02 1.02471566e+00 -5.49046695e-01 -4.80775028e-01
-5.40701509e-01 4.18671742e-02 -3.13876927e-01 -1.73198327e-01
2.49080777e-01 -7.31606334e-02 -3.65975320e-01 1.39791876e-01
-9.15656269e-01 4.99103010e-01 7.73616552e-01 9.01997864e-01
6.02584898e-01 -6.26311302e-01 7.47939497e-02 -9.91927266e-01
4.97844338e-01 -3.50623101e-01 -3.49104494e-01 7.42987633e-01
-3.69013429e-01 1.02251917e-01 5.91491461e-01 -1.72609687e-01
-1.39993560e+00 -2.86383033e-01 1.06050149e-01 -4.34684664e-01
-1.61604881e-01 5.45453072e-01 -3.11510324e-01 3.87986034e-01
8.63135397e-01 5.75726211e-01 -1.14192814e-01 -5.32378018e-01
8.56488645e-01 7.87156403e-01 6.74753249e-01 -3.38728845e-01
7.21390188e-01 4.93087530e-01 -4.40584451e-01 -9.17163134e-01
-9.12141025e-01 -4.41539705e-01 -7.34206975e-01 -1.23084493e-01
1.16671741e+00 -9.99757051e-01 -6.85132205e-01 3.44118863e-01
-1.54064333e+00 -5.72016239e-01 1.55213863e-01 1.32031720e-02
-4.61902380e-01 1.98913842e-01 -5.26752234e-01 -8.29059064e-01
-3.44185621e-01 -1.12079036e+00 1.46823347e+00 3.99412140e-02
-2.11236328e-01 -1.16771448e+00 -8.81920010e-02 8.96896601e-01
2.64645815e-01 -1.46151349e-01 9.79131639e-01 -7.32714236e-01
-6.89157248e-01 2.70920396e-01 -8.33832324e-01 2.66505748e-01
-7.82930702e-02 -5.34639880e-02 -1.35954988e+00 -4.21930850e-01
-6.22956336e-01 -9.45892215e-01 1.44324613e+00 2.88621366e-01
1.24499536e+00 -2.04694137e-01 -3.94295961e-01 5.96147120e-01
1.36613834e+00 3.06947172e-01 7.25963235e-01 3.55035067e-01
1.03821588e+00 4.60202724e-01 4.96355861e-01 1.82797045e-01
6.82977021e-01 4.92079139e-01 7.56518245e-01 -1.70563146e-01
-4.86245483e-01 -4.03262764e-01 5.32752991e-01 4.11405951e-01
5.77659421e-02 -9.35347438e-01 -1.32398844e+00 7.62715161e-01
-1.76491237e+00 -8.48576427e-01 -1.54541284e-01 1.62045574e+00
5.97009778e-01 1.40006125e-01 -2.22101420e-01 -2.80454725e-01
-6.36175694e-03 4.80724007e-01 -5.23811638e-01 -7.32222259e-01
-4.42144066e-01 -4.73234393e-02 2.68187106e-01 7.17743754e-01
-1.08620381e+00 1.27614331e+00 6.14533472e+00 7.01526046e-01
-9.90337193e-01 1.25294119e-01 6.61026537e-01 -9.07938331e-02
-7.52253592e-01 -6.52496815e-02 -8.02274287e-01 1.19414642e-01
8.05115104e-01 3.66349190e-01 4.80759829e-01 6.03442311e-01
2.55037062e-02 -2.05310941e-01 -1.30009592e+00 9.69649196e-01
5.50251901e-01 -1.61285520e+00 5.97014666e-01 -1.43363252e-01
6.38600886e-01 3.11151952e-01 3.38159859e-01 3.51560086e-01
3.59542638e-01 -1.57338154e+00 1.00626469e+00 4.18249041e-01
1.10603786e+00 -4.60187852e-01 5.47585070e-01 2.48131976e-01
-9.83175635e-01 8.51519480e-02 -3.30892444e-01 2.24094763e-01
2.65160024e-01 1.93477094e-01 -1.06566560e+00 5.41542053e-01
7.52039015e-01 8.98740649e-01 -9.52437520e-01 4.77400869e-01
-3.96774799e-01 7.16272652e-01 5.02725085e-03 1.78682469e-02
6.60851181e-01 1.61839336e-01 4.50768262e-01 1.47961128e+00
8.89570042e-02 6.37284517e-02 -1.28330067e-01 7.90825665e-01
-1.65484041e-01 1.89299025e-02 -6.63135469e-01 -2.50611395e-01
2.34023944e-01 9.09935057e-01 -3.21676165e-01 -4.30558294e-01
-6.72605038e-01 1.30429041e+00 4.23533410e-01 6.07115805e-01
-6.31846547e-01 -8.67577568e-02 3.10963809e-01 9.45858061e-02
5.77103019e-01 -2.43375868e-01 -3.48638415e-01 -1.15857923e+00
-1.10582069e-01 -1.04112160e+00 5.36980569e-01 -1.53174436e+00
-1.05558431e+00 7.73399234e-01 -4.59062196e-02 -8.53401840e-01
-5.19501328e-01 -1.14461184e+00 -4.09813106e-01 9.56988156e-01
-1.46273208e+00 -1.83008862e+00 -3.04470748e-01 7.73950815e-01
9.67415154e-01 -3.20542246e-01 6.55519545e-01 -2.80157149e-01
-1.77098572e-01 4.91136402e-01 2.14290246e-01 4.89211492e-02
9.28172469e-01 -1.59555936e+00 5.60928106e-01 9.23087656e-01
6.94023311e-01 5.31826615e-01 6.78483546e-01 -7.41751492e-01
-1.52563059e+00 -8.66502523e-01 8.69057536e-01 -1.08476400e+00
5.79371274e-01 -7.10356414e-01 -9.68918025e-01 1.19809115e+00
8.48924279e-01 -2.69120842e-01 4.78208244e-01 2.38859251e-01
-8.69814634e-01 2.41687238e-01 -7.24324584e-01 4.59678024e-01
9.95194733e-01 -9.17344809e-01 -8.51782978e-01 3.93685967e-01
6.36704981e-01 -5.82569838e-01 -5.27055323e-01 -1.32788885e-02
5.58011591e-01 -8.94031465e-01 1.03967226e+00 -9.04463649e-01
8.70862603e-01 -1.75878033e-01 -3.42883617e-01 -1.44808304e+00
-2.10761622e-01 -1.46926627e-01 -3.19623321e-01 1.26249015e+00
6.94087505e-01 -2.12014779e-01 3.38398516e-01 1.86961755e-01
-1.83151424e-01 -4.62692499e-01 -6.64668977e-01 -7.00036049e-01
4.20773663e-02 -8.16345215e-01 1.26431525e-01 7.99291432e-01
-3.02317590e-01 8.53118002e-01 -4.76154953e-01 7.25030079e-02
3.40475917e-01 3.00981123e-02 9.34270561e-01 -7.79227436e-01
-3.33234668e-01 -2.65496105e-01 -6.43913150e-02 -1.47559214e+00
6.32941276e-02 -9.81419742e-01 -9.72965732e-02 -2.23418665e+00
4.07080621e-01 1.05418906e-01 -2.64180377e-02 9.01186287e-01
-5.05508147e-02 2.18653217e-01 5.73320627e-01 3.49096060e-01
-7.03962326e-01 5.72781980e-01 1.42283070e+00 -7.61334658e-01
-5.69508858e-02 -4.52977955e-01 -6.64636314e-01 5.25044560e-01
3.82311642e-01 1.79669529e-01 -8.24433208e-01 -1.11705613e+00
1.56536147e-01 2.90071350e-02 4.07300413e-01 -6.65874302e-01
3.50731947e-02 3.69442217e-02 8.36483359e-01 -8.98028374e-01
5.72367072e-01 -6.09567344e-01 -4.69317734e-01 6.89020157e-02
-3.39713961e-01 1.81325451e-01 5.35569489e-01 4.34281409e-01
-2.95966834e-01 -3.03684082e-02 3.96067798e-01 -8.82291887e-03
-1.32326961e+00 4.22357349e-03 -3.58528018e-01 2.25113094e-01
5.95516086e-01 -3.17651212e-01 -1.02486742e+00 -8.79793465e-01
-7.30371475e-01 5.11112452e-01 3.42183381e-01 7.26877928e-01
9.74724770e-01 -9.42422569e-01 -8.07275057e-01 -1.19809091e-01
5.30335605e-01 -1.04949750e-01 3.61374289e-01 6.48607016e-01
-5.57471931e-01 9.78103697e-01 -1.12202898e-01 -8.87693584e-01
-1.40528131e+00 4.65931267e-01 1.45157039e-01 -4.14050907e-01
-4.51236576e-01 1.09315801e+00 6.79603159e-01 -3.98810446e-01
1.39274850e-01 -2.67079324e-01 -2.94160485e-01 1.15391016e-01
6.99552715e-01 -5.87462448e-02 5.76986670e-02 -8.46407115e-01
-6.07562125e-01 5.03696144e-01 -4.54382062e-01 -4.65751112e-01
1.10461330e+00 -2.39390254e-01 2.05676943e-01 3.63883764e-01
1.36659753e+00 -1.32406801e-01 -1.32022786e+00 -1.51361302e-01
-1.85984761e-01 -3.80335867e-01 1.73924357e-01 -1.47936070e+00
-7.64891684e-01 1.27194440e+00 4.43570673e-01 -3.00461482e-02
9.60892797e-01 3.41719389e-01 4.26348746e-01 4.84598100e-01
-8.55534896e-02 -9.04368579e-01 4.64193404e-01 9.02194977e-01
1.28586888e+00 -1.46363568e+00 -6.13194257e-02 -1.84865355e-01
-1.23832870e+00 1.01989686e+00 9.62418079e-01 3.61897945e-01
1.73689052e-01 -9.77407023e-02 3.37489605e-01 -3.60815525e-01
-9.61026847e-01 -2.53810465e-01 9.46186721e-01 6.23813570e-01
5.07061541e-01 -1.31812289e-01 3.36548686e-01 3.63070339e-01
-2.05721766e-01 -4.72596765e-01 3.64079326e-01 7.01796234e-01
-2.75253534e-01 -7.86860406e-01 -4.72023338e-01 5.33094704e-01
-6.51862919e-02 -2.91698247e-01 -7.39570618e-01 1.12213397e+00
-1.39905587e-01 7.99215555e-01 3.64361852e-01 -2.38786772e-01
2.62596011e-01 1.01213783e-01 5.49555123e-01 -5.43482125e-01
-3.65200996e-01 1.32822558e-01 3.08743685e-01 -7.22739697e-01
-2.08353147e-01 -4.57350224e-01 -1.20443559e+00 -4.70575504e-03
-3.99245396e-02 -2.95452923e-01 6.13681793e-01 1.32879043e+00
3.51829797e-01 6.88561559e-01 -1.55309392e-02 -5.91413617e-01
-1.57269344e-01 -1.16911888e+00 -3.05380732e-01 5.76963663e-01
4.12835360e-01 -5.40489733e-01 -1.42425820e-01 2.42645770e-01]
|
[11.238581657409668, 1.9897083044052124]
|
76b06b18-5801-496f-83cf-17b8f4c394e1
|
improving-candidate-generation-for-low
|
2003.01343
| null |
https://arxiv.org/abs/2003.01343v1
|
https://arxiv.org/pdf/2003.01343v1.pdf
|
Improving Candidate Generation for Low-resource Cross-lingual Entity Linking
|
Cross-lingual entity linking (XEL) is the task of finding referents in a target-language knowledge base (KB) for mentions extracted from source-language texts. The first step of (X)EL is candidate generation, which retrieves a list of plausible candidate entities from the target-language KB for each mention. Approaches based on resources from Wikipedia have proven successful in the realm of relatively high-resource languages (HRL), but these do not extend well to low-resource languages (LRL) with few, if any, Wikipedia pages. Recently, transfer learning methods have been shown to reduce the demand for resources in the LRL by utilizing resources in closely-related languages, but the performance still lags far behind their high-resource counterparts. In this paper, we first assess the problems faced by current entity candidate generation methods for low-resource XEL, then propose three improvements that (1) reduce the disconnect between entity mentions and KB entries, and (2) improve the robustness of the model to low-resource scenarios. The methods are simple, but effective: we experiment with our approach on seven XEL datasets and find that they yield an average gain of 16.9% in Top-30 gold candidate recall, compared to state-of-the-art baselines. Our improved model also yields an average gain of 7.9% in in-KB accuracy of end-to-end XEL.
|
['Shruti Rijhawani', 'Jaime Carbonell', 'Shuyan Zhou', 'John Wieting', 'Graham Neubig']
|
2020-03-03
|
improving-candidate-generation-for-low-1
|
https://aclanthology.org/2020.tacl-1.8
|
https://aclanthology.org/2020.tacl-1.8.pdf
|
tacl-2020-1
|
['cross-lingual-entity-linking']
|
['natural-language-processing']
|
[-4.86599624e-01 5.74789643e-01 -4.99554217e-01 -5.97084314e-02
-1.87525809e+00 -7.97532976e-01 6.49645567e-01 3.82644981e-01
-7.79745698e-01 1.34054756e+00 4.95471925e-01 -2.46312976e-01
3.91119681e-02 -7.65941501e-01 -1.04937375e+00 4.41565886e-02
-3.78456600e-02 9.56147909e-01 6.21710658e-01 -4.94247347e-01
-2.55774204e-02 3.82966883e-02 -9.20898020e-01 4.12397832e-01
1.19236481e+00 4.43801820e-01 2.19307169e-02 8.91531929e-02
-3.93557549e-01 5.72793841e-01 -4.41448808e-01 -9.06922936e-01
3.96128651e-03 -1.09055966e-01 -1.20514524e+00 -7.25556612e-01
5.98540246e-01 2.48172760e-01 -2.65743166e-01 8.82199883e-01
4.85437095e-01 -8.83988589e-02 7.47508705e-01 -9.27918494e-01
-8.81429851e-01 1.20606935e+00 -5.15959740e-01 1.92977250e-01
5.73990941e-01 -2.11989805e-01 1.26547742e+00 -1.47622526e+00
1.15127587e+00 1.29959929e+00 9.25224900e-01 5.35552979e-01
-1.12673187e+00 -7.30333209e-01 -4.96217944e-02 1.06784508e-01
-1.81348419e+00 -7.29540765e-01 9.71342698e-02 -1.58142656e-01
1.61637282e+00 6.80533145e-03 1.18010439e-01 8.06820273e-01
-7.76848271e-02 7.22555578e-01 9.37695265e-01 -8.44785333e-01
-1.63731486e-01 5.04994631e-01 1.17904417e-01 6.00760102e-01
8.36231351e-01 -3.39154065e-01 -7.67833412e-01 -3.78560066e-01
2.28108495e-01 -1.07522869e+00 -3.74952704e-01 -6.39472753e-02
-1.25733542e+00 7.55911350e-01 3.45407397e-01 4.26685721e-01
-2.74709940e-01 -3.47978882e-02 5.56380808e-01 1.25900269e-01
7.39117742e-01 8.97727370e-01 -9.94391203e-01 2.23828018e-01
-9.11822855e-01 4.67502207e-01 1.10399234e+00 1.33644569e+00
8.00262749e-01 -4.08513486e-01 -9.58146080e-02 7.96313763e-01
3.87231171e-01 5.66706419e-01 3.17212105e-01 -4.49785084e-01
1.19956195e+00 5.63510418e-01 4.87537503e-01 -7.78693020e-01
-3.67729038e-01 -2.60627747e-01 -2.56512672e-01 -4.55968380e-01
4.81247991e-01 -3.25820297e-01 -6.61149859e-01 1.88533378e+00
3.80413920e-01 8.28084424e-02 5.38464725e-01 3.72506618e-01
1.05348563e+00 7.93446481e-01 6.51182652e-01 -1.06353343e-01
1.38802898e+00 -9.03718412e-01 -4.06018168e-01 -5.22207618e-01
1.04199219e+00 -7.58833885e-01 1.04515362e+00 -3.02138925e-01
-1.04071546e+00 -2.86828935e-01 -8.76256227e-01 -3.34275782e-01
-6.63628817e-01 4.06957656e-01 5.12731910e-01 5.79513967e-01
-1.10599113e+00 3.10281962e-01 -5.86695790e-01 -5.95217884e-01
-6.45356178e-02 2.71357059e-01 -5.49509883e-01 -1.15081809e-01
-1.87348819e+00 1.39003789e+00 9.89484012e-01 -1.46006182e-01
-2.85842448e-01 -1.08607578e+00 -9.20660496e-01 6.73052222e-02
5.92992902e-01 -7.50394046e-01 1.11105740e+00 -5.04300475e-01
-8.48724902e-01 9.21907365e-01 -1.16529882e-01 -3.73185545e-01
2.15226948e-01 -5.89462578e-01 -6.67681456e-01 -1.81075603e-01
7.51362145e-01 8.10356855e-01 -6.66730031e-02 -1.25928891e+00
-9.74111140e-01 1.00182503e-01 1.28236189e-01 3.90833169e-01
-1.68354109e-01 3.23560655e-01 -8.91693234e-01 -5.09657562e-01
-2.92269260e-01 -9.66552317e-01 -9.12144184e-02 -5.69349766e-01
-5.05340695e-01 -6.62528574e-01 1.85530186e-01 -1.05186534e+00
1.44552755e+00 -1.65996921e+00 -1.69208348e-01 1.60789967e-01
-2.43067116e-01 4.74564016e-01 -1.80797681e-01 6.74455106e-01
1.18039735e-01 5.53467870e-01 1.11093633e-02 -7.03438967e-02
1.22037521e-02 -1.45810649e-01 -4.13238764e-01 -5.75926900e-02
3.41412663e-01 1.09474254e+00 -1.19515920e+00 -7.68610775e-01
-4.13849682e-01 3.18034977e-01 -3.33839357e-01 -1.45704255e-01
-3.39517802e-01 -1.27529250e-02 -4.86033499e-01 5.57845652e-01
1.35202155e-01 -1.94452226e-01 4.87019092e-01 -4.56229597e-01
-1.19756922e-01 1.06033683e+00 -1.17613578e+00 1.52491224e+00
-6.09712362e-01 3.49006087e-01 -3.37436795e-01 -1.13727145e-01
6.04639947e-01 4.63334799e-01 5.30079938e-02 -4.07008439e-01
-4.46016639e-01 8.06525409e-01 -1.67640030e-01 -2.70203322e-01
8.12645137e-01 1.88019853e-02 -5.76816142e-01 2.46585146e-01
2.60859489e-01 1.57004058e-01 5.17158031e-01 5.85605502e-01
1.06132936e+00 3.18990648e-01 7.21591234e-01 -3.85250092e-01
5.09844780e-01 4.48832661e-01 6.24737561e-01 6.55027092e-01
2.26999715e-01 3.00745144e-02 1.45143032e-01 2.25852290e-03
-1.08665371e+00 -1.00524676e+00 -8.93057212e-02 9.22777951e-01
1.63034528e-01 -6.71630442e-01 -5.65424204e-01 -1.04704714e+00
1.70419022e-01 1.00517106e+00 -3.01423639e-01 3.43603194e-02
-9.48325753e-01 -8.42544734e-01 1.08960831e+00 5.30074358e-01
3.77883196e-01 -1.11090291e+00 -8.00711736e-02 5.24186790e-01
-5.80188572e-01 -1.38857222e+00 -3.85542065e-01 -3.59324515e-02
-4.40622956e-01 -8.94761503e-01 -4.94411975e-01 -8.22479963e-01
4.39562380e-01 -2.22184479e-01 1.77051318e+00 -1.43996909e-01
2.89404988e-02 2.68405885e-01 -2.78688043e-01 -2.53783494e-01
-5.23322701e-01 6.75260425e-01 3.54997039e-01 -6.23332560e-01
7.14828908e-01 7.31435791e-02 -1.21533170e-01 -3.18823103e-03
-3.84663492e-01 -1.39622673e-01 6.77964032e-01 7.61981189e-01
7.35950410e-01 -3.15806061e-01 1.15434778e+00 -1.31082976e+00
5.97917974e-01 -7.01227069e-01 -4.04264241e-01 8.93489480e-01
-9.14769411e-01 2.11340666e-01 3.92299026e-01 -9.68304649e-02
-1.41852260e+00 2.00667346e-04 -1.88279420e-01 3.22917640e-01
1.76428214e-01 1.02369523e+00 -3.39756161e-01 7.81231076e-02
8.57183039e-01 -1.66850314e-01 -8.58072698e-01 -6.19142354e-01
7.89376318e-01 5.62805176e-01 6.50920391e-01 -1.01383841e+00
8.40037167e-01 -1.99846119e-01 -4.70986605e-01 -3.91164958e-01
-1.18645060e+00 -5.88125288e-01 -8.27187181e-01 1.45184919e-01
6.32483006e-01 -1.52010572e+00 -8.05169940e-02 -5.95117286e-02
-1.14138710e+00 -1.97676882e-01 -1.84201047e-01 4.57704306e-01
-2.12288544e-01 2.36992016e-01 -9.43507493e-01 -3.50356966e-01
-6.73688948e-01 -6.05281949e-01 1.05207729e+00 2.11338475e-01
-3.71080369e-01 -1.05752611e+00 4.76709783e-01 1.56625256e-01
1.86941251e-01 -1.28752828e-01 1.13597548e+00 -8.79695773e-01
-6.23987138e-01 -1.45356208e-01 -4.08960313e-01 -1.77457795e-01
-3.40386368e-02 -2.75359750e-01 -5.44372439e-01 -3.22210819e-01
-8.85709047e-01 -6.51799858e-01 8.75484765e-01 -8.57065693e-02
1.25453636e-01 -4.33043242e-01 -8.89930964e-01 1.28948867e-01
1.67713404e+00 -1.75270960e-01 5.01691520e-01 6.61939979e-01
6.40378356e-01 6.74983799e-01 1.02465880e+00 -1.23843223e-01
8.17552388e-01 8.72218311e-01 -3.30947369e-01 -5.86270690e-02
-4.34232175e-01 -6.45371497e-01 5.20377338e-01 1.03964353e+00
-9.89613682e-03 -4.17968124e-01 -1.33332145e+00 1.05151975e+00
-1.79189134e+00 -8.72464120e-01 -1.97448730e-01 2.02879167e+00
1.50109410e+00 2.31518522e-01 -6.90497085e-02 -6.20709956e-01
6.52797461e-01 -2.99150765e-01 -3.49016875e-01 7.35187680e-02
-2.26146206e-01 2.65392601e-01 8.05258811e-01 6.98827744e-01
-1.22873056e+00 1.65760207e+00 6.02463722e+00 9.55993652e-01
-6.81260645e-01 2.72258013e-01 3.82591337e-01 1.47754952e-01
-4.28036839e-01 2.23320633e-01 -1.79364300e+00 2.49118209e-01
1.20967424e+00 -5.98115742e-01 -1.52547002e-01 9.78232026e-01
-3.45698714e-01 -9.39221587e-04 -1.02453244e+00 4.52894241e-01
2.12318465e-01 -1.47900999e+00 6.90548271e-02 -2.64575005e-01
7.88232684e-01 4.55655992e-01 -3.37555975e-01 8.86598527e-01
7.49435127e-01 -8.28493178e-01 6.94790959e-01 3.79252911e-01
1.13081002e+00 -8.55437636e-01 8.15358877e-01 3.09548885e-01
-1.51581645e+00 4.33153808e-01 -5.77472806e-01 5.79955578e-01
4.83947396e-01 6.12740874e-01 -1.18354642e+00 7.67852783e-01
7.31995642e-01 3.90582740e-01 -7.55327642e-01 1.02133453e+00
-5.13890803e-01 6.15181625e-01 -3.23468089e-01 -1.30314127e-01
3.17312837e-01 3.43341649e-01 5.09347618e-01 1.80646336e+00
3.54646385e-01 7.11073950e-02 1.97207332e-01 7.33109891e-01
-6.06854498e-01 5.70601404e-01 -5.88951886e-01 8.16542357e-02
1.02408338e+00 1.17064452e+00 -4.53698546e-01 -4.36139733e-01
-5.57757318e-01 7.91367292e-01 9.40047622e-01 1.50830433e-01
-7.39606977e-01 -5.65806508e-01 3.58897567e-01 -7.34343082e-02
1.97324768e-01 -6.88000917e-02 1.11909166e-01 -1.18619227e+00
1.60931721e-01 -7.98636079e-01 5.94464540e-01 -4.84169275e-01
-1.63351130e+00 7.21022010e-01 2.20940888e-01 -8.92001092e-01
-5.05335152e-01 -3.93337518e-01 6.19784277e-03 1.09322703e+00
-1.77913666e+00 -1.57438731e+00 3.31019729e-01 3.10058922e-01
5.82760513e-01 -1.51748106e-01 9.48567092e-01 6.47094905e-01
-3.66288543e-01 8.45812976e-01 2.15265527e-01 3.77696961e-01
1.24409151e+00 -1.43549454e+00 8.48349452e-01 1.04949629e+00
4.39110845e-01 9.95840967e-01 4.57630396e-01 -1.19460237e+00
-9.53045070e-01 -1.41929913e+00 1.90829909e+00 -9.32002664e-01
8.27811837e-01 -3.38169843e-01 -9.80075419e-01 1.01453948e+00
2.06445962e-01 -1.25489637e-01 5.81874847e-01 6.19043648e-01
-6.42952740e-01 2.35411510e-01 -9.65706050e-01 5.87965488e-01
8.56071651e-01 -6.49374187e-01 -8.42244387e-01 3.65618229e-01
8.00548136e-01 -5.77823997e-01 -1.10790431e+00 4.81016189e-01
3.17089945e-01 -1.92777529e-01 1.11896813e+00 -7.59460688e-01
2.28006735e-01 -5.41519225e-01 -7.15004355e-02 -1.40977919e+00
-3.75753611e-01 -2.11539030e-01 -1.67287186e-01 1.77344120e+00
1.40263689e+00 -3.01828444e-01 5.32369852e-01 8.23951066e-01
-1.69079736e-01 -6.12473130e-01 -9.16424990e-01 -9.70731139e-01
3.83869976e-01 -1.36842087e-01 3.29077989e-01 1.09868515e+00
2.46245295e-01 7.99807727e-01 -2.41983950e-01 3.78218889e-01
4.35645163e-01 -8.41963738e-02 5.82533777e-01 -1.14539754e+00
-2.22055241e-01 4.46661375e-03 6.62431642e-02 -8.57056022e-01
5.31967700e-01 -1.15496027e+00 2.06836745e-01 -1.76913488e+00
4.10235643e-01 -1.05538881e+00 -1.37292743e-01 8.88985574e-01
-6.22989058e-01 2.47449964e-01 6.64566383e-02 3.45598191e-01
-7.45558441e-01 2.26213887e-01 3.69037747e-01 3.67448889e-02
-2.37886578e-01 -3.72117549e-01 -7.86687791e-01 6.51870549e-01
4.85427856e-01 -7.33996987e-01 -2.70680636e-01 -5.41397512e-01
6.34220660e-01 -1.29211038e-01 -1.31353214e-01 -7.38497794e-01
3.89563382e-01 1.00816749e-01 2.05458999e-01 -4.79159027e-01
6.08964451e-03 -2.98556060e-01 1.60523713e-01 1.53529853e-01
-3.81355554e-01 2.05416739e-01 3.82919937e-01 4.10005569e-01
-2.25291446e-01 -4.31616277e-01 4.93259043e-01 -2.61465222e-01
-1.00834835e+00 -2.98385322e-02 -1.05776303e-01 7.70361483e-01
7.83830166e-01 3.29251856e-01 -6.18558884e-01 -1.34278703e-02
-3.72430354e-01 1.81861281e-01 3.98208439e-01 5.30133605e-01
1.67631999e-01 -1.32660055e+00 -1.04277277e+00 -4.79304522e-01
4.31328803e-01 -1.39720619e-01 -1.61445096e-01 5.22326648e-01
-3.28246474e-01 8.43758941e-01 7.54468590e-02 6.19779453e-02
-1.09002614e+00 4.03338492e-01 1.25999302e-01 -9.38497961e-01
-4.57523316e-01 1.00590467e+00 -1.67074323e-01 -6.43336236e-01
6.12960709e-03 9.69226584e-02 -3.01280767e-01 4.37529832e-02
4.61346179e-01 2.46102780e-01 2.72281438e-01 -9.05498505e-01
-6.81304634e-01 4.06546503e-01 -4.38016415e-01 -2.50508338e-01
1.20969057e+00 -1.97320297e-01 -8.99886191e-02 3.18496138e-01
8.18518996e-01 6.37228429e-01 -4.00213152e-01 -4.66259539e-01
9.03138459e-01 -1.27931219e-03 -2.57892877e-01 -1.15625560e+00
-5.77189505e-01 2.71310568e-01 6.11347333e-02 -2.84270614e-01
6.62558019e-01 5.08029819e-01 1.00011754e+00 6.08182430e-01
8.79540086e-01 -1.03861153e+00 -3.99902284e-01 7.50972331e-01
7.31559336e-01 -1.08740532e+00 8.61424059e-02 -7.69779265e-01
-6.49778783e-01 7.57974088e-01 9.46335554e-01 1.63089693e-01
4.62700367e-01 2.66507179e-01 -4.25697863e-02 -2.75236934e-01
-8.80344212e-01 -3.32257837e-01 5.68260670e-01 5.31876147e-01
8.19532633e-01 -4.50522602e-02 -5.12129307e-01 6.98384643e-01
-9.73272920e-02 -2.34257624e-01 2.59010226e-01 5.60276866e-01
-4.36233729e-01 -1.33905065e+00 2.53323596e-02 3.63952070e-01
-6.96523249e-01 -7.11437702e-01 -4.16767210e-01 1.26746833e+00
1.65164173e-01 6.86413705e-01 -3.91817689e-01 1.78083181e-02
5.63210607e-01 2.91014999e-01 3.36951017e-01 -9.54158366e-01
-6.72867894e-01 -2.01969698e-01 9.02075946e-01 -3.12912166e-01
-4.21389431e-01 -7.22433209e-01 -1.36241782e+00 -4.15847488e-02
-6.04637623e-01 4.30940419e-01 2.74406910e-01 8.40429902e-01
5.12633741e-01 -3.97074223e-02 -6.57590851e-02 -3.28731298e-01
-3.72739106e-01 -9.34561789e-01 -2.99342275e-01 1.69451624e-01
-2.83851683e-01 -6.08882129e-01 -1.01935817e-02 4.15283628e-02]
|
[9.550453186035156, 8.956287384033203]
|
59fb53a2-74a7-40f4-a5f7-d0d510c47dba
|
enhancing-underexposed-photos-using
|
1907.10992
| null |
https://arxiv.org/abs/1907.10992v3
|
https://arxiv.org/pdf/1907.10992v3.pdf
|
Enhancing Underexposed Photos using Perceptually Bidirectional Similarity
|
Although remarkable progress has been made, existing methods for enhancing underexposed photos tend to produce visually unpleasing results due to the existence of visual artifacts (e.g., color distortion, loss of details and uneven exposure). We observed that this is because they fail to ensure the perceptual consistency of visual information between the source underexposed image and its enhanced output. To obtain high-quality results free of these artifacts, we present a novel underexposed photo enhancement approach that is able to maintain the perceptual consistency. We achieve this by proposing an effective criterion, referred to as perceptually bidirectional similarity, which explicitly describes how to ensure the perceptual consistency. Particularly, we adopt the Retinex theory and cast the enhancement problem as a constrained illumination estimation optimization, where we formulate perceptually bidirectional similarity as constraints on illumination and solve for the illumination which can recover the desired artifact-free enhancement results. In addition, we describe a video enhancement framework that adopts the presented illumination estimation for handling underexposed videos. To this end, a probabilistic approach is introduced to propagate illuminations of sampled keyframes to the entire video by tackling a Bayesian Maximum A Posteriori problem. Extensive experiments demonstrate the superiority of our method over the state-of-the-art methods.
|
['Wei-Shi Zheng', 'Yongwei Nie', 'Chunxia Xiao', 'Lei Zhu', 'Qing Zhang']
|
2019-07-25
| null | null | null | null |
['video-enhancement']
|
['computer-vision']
|
[ 6.93618000e-01 -2.70624340e-01 3.48785132e-01 -2.56337792e-01
-7.17953682e-01 -2.61560023e-01 4.46288824e-01 -2.70680726e-01
-1.60600901e-01 6.65481985e-01 7.74726868e-02 1.66881979e-01
-2.39847302e-01 -5.61254978e-01 -6.97819948e-01 -9.76612031e-01
2.88194418e-01 -6.24501765e-01 1.90010220e-01 -7.84194618e-02
3.56325686e-01 3.64466608e-01 -1.84890926e+00 1.16794422e-01
1.03152955e+00 9.71417069e-01 3.71425897e-01 5.29811263e-01
3.14713538e-01 5.47626495e-01 -4.79019433e-01 -3.28100115e-01
3.45833033e-01 -5.96619308e-01 -5.72069764e-01 7.33952165e-01
5.71584225e-01 -5.51337063e-01 -2.67152697e-01 1.54042506e+00
4.38245177e-01 3.11922222e-01 6.42326713e-01 -1.21957016e+00
-7.65617430e-01 -1.06726408e-01 -8.28207254e-01 7.73686841e-02
4.16210562e-01 -1.48495167e-01 4.81671751e-01 -1.02807605e+00
4.72355783e-01 1.05766690e+00 3.17216367e-01 2.77956069e-01
-1.19398761e+00 -2.16457397e-01 1.56379372e-01 4.58651304e-01
-1.50540507e+00 -6.92448437e-01 1.11084592e+00 -2.44936392e-01
2.97057807e-01 4.46709633e-01 5.45836449e-01 7.34469116e-01
3.47916991e-01 5.72682917e-01 1.47999001e+00 -6.49494588e-01
2.48033971e-01 3.57296735e-01 -5.80897667e-02 5.66006005e-01
1.84111133e-01 2.26308942e-01 -4.80240166e-01 9.14204866e-03
6.56788111e-01 -2.69357339e-02 -8.09258640e-01 -3.64647686e-01
-9.06078696e-01 1.12924173e-01 2.32316747e-01 3.21397603e-01
-6.05438232e-01 -1.32852793e-01 -5.75562287e-03 -3.41487713e-02
5.45640588e-01 7.11577460e-02 -2.22309679e-02 2.58172840e-01
-8.63176703e-01 3.19065750e-02 3.23378205e-01 9.34090555e-01
6.87702894e-01 6.64991364e-02 -2.40573928e-01 7.25134254e-01
5.51798761e-01 5.37450016e-01 -4.23743501e-02 -1.09039700e+00
2.80183017e-01 1.09977163e-01 4.20752108e-01 -1.38203740e+00
1.72108963e-01 -3.78214538e-01 -9.50946808e-01 6.15887105e-01
2.32142568e-01 1.09849922e-01 -6.45022929e-01 1.72174704e+00
4.68698025e-01 1.47823587e-01 2.00610608e-01 1.06146073e+00
4.58457530e-01 9.11169350e-01 -1.13632783e-01 -8.12668383e-01
1.28730857e+00 -7.05775738e-01 -1.45691454e+00 4.81477827e-02
-3.96391213e-01 -1.16672254e+00 7.36695051e-01 7.85601735e-01
-1.54321325e+00 -8.50787520e-01 -1.31963992e+00 4.10370342e-02
-9.54310745e-02 5.91648445e-02 -3.03194970e-02 7.93975532e-01
-1.18775976e+00 5.49163699e-01 -5.53751528e-01 -1.68717712e-01
-3.25477831e-02 -5.89607330e-03 -1.95203334e-01 -2.74414212e-01
-1.02780199e+00 1.04848421e+00 2.57585287e-01 5.54941118e-01
-8.75387251e-01 -4.31788564e-01 -8.44059169e-01 1.09830908e-02
5.14215469e-01 -6.57725692e-01 8.63973081e-01 -1.28983188e+00
-1.70179713e+00 6.01553619e-01 -3.34410697e-01 1.79842308e-01
4.90967363e-01 -2.51338691e-01 -5.95086038e-01 3.91191959e-01
-3.19068104e-01 1.93710640e-01 1.32008731e+00 -1.94278634e+00
-4.91107851e-01 -3.98284733e-01 -4.79942858e-02 4.03932393e-01
-5.73688090e-01 1.57427341e-01 -8.62430453e-01 -7.81776905e-01
3.89757752e-01 -6.62766993e-01 3.21689658e-02 2.44297922e-01
-3.66715252e-01 2.63520479e-01 8.03982556e-01 -1.01149380e+00
1.22676384e+00 -2.31535816e+00 2.48562887e-01 2.06776410e-01
8.68664384e-02 1.91759288e-01 -4.24756892e-02 3.09526354e-01
5.68987569e-03 -2.25290090e-01 -4.64771390e-01 -6.07458770e-01
-7.75171146e-02 -4.86922538e-04 -1.12228721e-01 8.53663743e-01
5.55083342e-02 1.73467919e-01 -8.80817115e-01 -7.40938663e-01
5.70094585e-01 1.06919026e+00 -3.09651226e-01 4.85024005e-01
2.38316059e-01 4.05031562e-01 -1.74687400e-01 6.20454729e-01
1.21599972e+00 1.36038572e-01 5.86915538e-02 -7.36232877e-01
-3.36006463e-01 -4.63829309e-01 -1.54303098e+00 1.49102247e+00
-2.47943267e-01 5.23553133e-01 4.34713006e-01 -6.10384047e-01
9.19239104e-01 3.63963157e-01 3.95419955e-01 -4.49883997e-01
1.88782781e-01 1.03629693e-01 -5.19837797e-01 -7.83508897e-01
7.29835570e-01 -3.21975350e-01 5.53178906e-01 1.95346773e-03
-2.15829313e-01 -7.80955330e-02 1.01224547e-02 4.57167774e-02
4.82592225e-01 3.44411016e-01 3.39889169e-01 -1.85968027e-01
9.05807257e-01 -5.63014030e-01 6.35023236e-01 5.06847024e-01
-3.11369777e-01 7.83954144e-01 -4.33273427e-02 1.34047046e-01
-1.05149627e+00 -1.14121318e+00 -2.54397571e-01 6.34175599e-01
7.04249740e-01 -1.56969890e-01 -8.58613551e-01 -2.43547246e-01
-5.70414662e-01 5.17038822e-01 -3.89310539e-01 -9.71929580e-02
-2.95329660e-01 -6.60947919e-01 1.03467919e-05 1.49486318e-01
8.83128703e-01 -6.11096144e-01 -3.94095451e-01 1.23518825e-01
-5.20691693e-01 -1.14124715e+00 -4.65843409e-01 -3.31339777e-01
-6.43295467e-01 -9.99213636e-01 -1.14139605e+00 -7.72280514e-01
9.24869120e-01 6.71769679e-01 7.77484834e-01 7.78824985e-02
-2.28346661e-01 4.86780614e-01 -3.46490890e-01 -6.81291744e-02
-4.57883596e-01 -9.00327563e-01 -2.18428802e-02 6.77322209e-01
-1.66427955e-01 -4.34902698e-01 -8.36427033e-01 4.51082647e-01
-1.33280730e+00 9.62291211e-02 5.60300469e-01 7.34454930e-01
7.73808360e-01 6.44801319e-01 2.64899939e-01 -4.16709989e-01
4.27016646e-01 -1.61667943e-01 -6.69487119e-01 4.45700735e-01
-6.20787382e-01 -1.84132859e-01 6.17623448e-01 -2.02110246e-01
-1.72065878e+00 -3.81782725e-02 2.82948595e-02 -5.33847153e-01
-2.99730361e-01 1.28408656e-01 -5.30665874e-01 -3.29874754e-01
3.33527833e-01 5.72606146e-01 -5.08465730e-02 -4.56104785e-01
3.12261671e-01 6.85967267e-01 7.94032931e-01 -4.29233164e-01
8.83264899e-01 7.69296169e-01 2.36082047e-01 -1.00475204e+00
-6.08042896e-01 -4.64633375e-01 -3.31970036e-01 -7.74880171e-01
9.00601566e-01 -6.72267079e-01 -6.28260255e-01 6.62250817e-01
-1.17324126e+00 1.04057200e-01 9.90530699e-02 5.99642098e-01
-6.98150158e-01 1.01056612e+00 -4.51893002e-01 -1.20971715e+00
-6.52553514e-02 -1.26820540e+00 9.23510909e-01 2.50896513e-01
3.05520594e-01 -9.04080927e-01 -9.68521554e-03 2.50509620e-01
5.01170576e-01 3.50318074e-01 6.33314192e-01 4.17362124e-01
-7.22157776e-01 9.97801051e-02 -3.96070033e-01 8.55687320e-01
2.31796578e-01 1.92851022e-01 -1.13049746e+00 -3.85997593e-01
5.07232726e-01 1.70356810e-01 5.76847732e-01 4.59809303e-01
1.19687784e+00 -3.40768069e-01 -1.12000078e-01 6.83743298e-01
1.91076672e+00 2.29349643e-01 9.62855816e-01 3.00903529e-01
2.96174079e-01 8.66113007e-01 8.76554549e-01 4.89729434e-01
1.80725083e-02 8.59046280e-01 6.29899740e-01 -4.54275370e-01
-3.20532203e-01 -9.25544351e-02 3.82463455e-01 6.84631646e-01
-3.01129669e-01 -4.54691350e-01 -2.92477459e-01 4.46456134e-01
-1.70738769e+00 -9.50878441e-01 -2.67262012e-01 2.47974849e+00
7.87760258e-01 -2.18172029e-01 -3.25483710e-01 4.80885834e-01
1.03187907e+00 2.38342106e-01 -1.24298282e-01 -1.87913209e-01
-1.93216667e-01 -2.45148107e-01 2.78041422e-01 6.96919501e-01
-9.81929064e-01 3.72497976e-01 6.22133923e+00 8.29035878e-01
-7.63956010e-01 4.20936830e-02 4.79350150e-01 3.27055991e-01
-3.23582590e-01 -2.13347897e-02 -4.33319509e-01 6.04027092e-01
3.05573761e-01 -3.64935435e-02 4.31671023e-01 3.38445216e-01
5.66467583e-01 -3.73561949e-01 -8.02293837e-01 1.12815475e+00
5.46525717e-01 -6.99138939e-01 -2.49761268e-02 1.27011230e-02
9.11329985e-01 -8.83435011e-01 2.99814612e-01 -3.00816000e-01
-3.67806703e-01 -5.64769745e-01 7.63235569e-01 9.25925612e-01
6.47475779e-01 -7.77087390e-01 6.03397310e-01 1.90888315e-01
-1.00749445e+00 1.33373216e-01 -3.63035709e-01 1.25624388e-01
5.02873123e-01 6.99422181e-01 -2.11728901e-01 9.24202561e-01
6.77561760e-01 7.36292601e-01 -3.52621228e-01 1.32001340e+00
-4.47270840e-01 5.38400933e-02 3.59364115e-02 4.13105428e-01
-9.68794078e-02 -4.74078864e-01 9.02149975e-01 9.89770591e-01
5.59303880e-01 3.29124838e-01 -4.81525511e-02 9.52722609e-01
1.82736099e-01 1.52104869e-01 -5.15298843e-01 5.62444687e-01
1.11408331e-01 1.23512805e+00 -5.56955993e-01 -3.35574865e-01
-3.77084136e-01 1.32802892e+00 -3.32594424e-01 7.29601920e-01
-9.18535948e-01 -4.06060010e-01 3.49525303e-01 -1.01733491e-01
8.65219086e-02 -5.09938970e-02 -6.68447837e-02 -1.10169029e+00
2.57933468e-01 -8.88918579e-01 8.38252530e-02 -1.20152342e+00
-1.18453133e+00 5.72400689e-01 6.78097978e-02 -1.49583960e+00
1.88800976e-01 -4.04311329e-01 -4.76158470e-01 9.10806060e-01
-2.00807738e+00 -9.36327457e-01 -5.79266429e-01 8.93094897e-01
5.97188115e-01 3.76145780e-01 3.76666695e-01 5.33473969e-01
-4.78972226e-01 3.10988128e-01 2.56201357e-01 -3.88122499e-01
1.00108612e+00 -1.05612242e+00 -4.80498910e-01 1.44389129e+00
-2.77245462e-01 4.07765925e-01 1.09495997e+00 -5.19835234e-01
-1.41575134e+00 -9.09653842e-01 7.91443706e-01 1.34314597e-01
2.47894764e-01 1.32933661e-01 -9.25360858e-01 3.40248257e-01
5.44569373e-01 -1.78924248e-01 4.90890652e-01 -4.03776675e-01
-6.92042336e-02 -2.34427825e-01 -1.21857870e+00 5.49394727e-01
7.35061824e-01 -3.89528275e-01 -6.13355398e-01 6.90145269e-02
3.56289178e-01 -1.41495705e-01 -8.89680684e-01 5.14726758e-01
4.97255623e-01 -1.24726212e+00 1.03742778e+00 6.87047020e-02
4.66850191e-01 -7.55670667e-01 -3.36832613e-01 -1.18960750e+00
-2.74185747e-01 -7.15683639e-01 -8.87991264e-02 1.56624341e+00
-6.18490428e-02 -5.20349681e-01 2.18914449e-01 6.01101041e-01
-1.29675716e-01 -3.44102472e-01 -5.05408287e-01 -8.46657515e-01
-6.19046748e-01 -2.33237937e-01 2.76942432e-01 8.97702515e-01
1.91345718e-02 -1.91230312e-01 -8.41213584e-01 6.98518634e-01
1.31608570e+00 3.64563265e-03 4.53089029e-01 -6.72861814e-01
-3.45184177e-01 -1.71210304e-01 -2.72627234e-01 -1.00662756e+00
-9.85258147e-02 -2.75307447e-01 5.42787075e-01 -1.44557393e+00
4.84371006e-01 -1.24139890e-01 -4.45523828e-01 -1.05615444e-01
-4.37697589e-01 5.27081966e-01 2.38001749e-01 9.82290059e-02
-5.33617020e-01 6.82790577e-01 1.50609100e+00 -7.19925612e-02
-4.13018242e-02 -1.12479880e-01 -5.55732906e-01 8.55760217e-01
4.73120570e-01 -2.60694087e-01 -3.32904935e-01 -5.03057957e-01
1.33727238e-01 2.30706766e-01 6.09224737e-01 -1.02425659e+00
3.28403145e-01 -1.12193726e-01 4.27122712e-01 -5.97303152e-01
5.75150430e-01 -1.12935531e+00 1.94513738e-01 1.22480430e-01
-3.14195454e-01 -2.32754126e-01 -2.28766911e-02 8.16197157e-01
-5.09698749e-01 -3.61668795e-01 1.11771548e+00 1.81288540e-01
-6.67427540e-01 9.19030011e-02 -4.88058120e-01 -4.54327703e-01
1.03169823e+00 -3.87020797e-01 -1.88552573e-01 -4.04051691e-01
-7.20072567e-01 -1.31988049e-01 6.22739017e-01 7.83347264e-02
8.87641788e-01 -1.15249503e+00 -7.66510129e-01 2.07750931e-01
-5.80878593e-02 -4.72410589e-01 7.48616636e-01 9.64124203e-01
-3.23293000e-01 -6.42219186e-02 -2.32228383e-01 -5.61997652e-01
-1.59326005e+00 8.40094388e-01 2.71161944e-01 5.97570203e-02
-4.46302116e-01 4.63791788e-01 2.34072253e-01 3.80557209e-01
3.78615856e-01 -1.04622446e-01 -3.06385368e-01 -2.88611710e-01
8.01052570e-01 7.05489337e-01 -3.04290075e-02 -7.16656387e-01
-1.88362777e-01 8.87252927e-01 2.53069580e-01 -2.24783599e-01
1.04783535e+00 -8.33578110e-01 -1.55674785e-01 1.31372005e-01
1.20353532e+00 1.47169024e-01 -1.52411509e+00 -1.36985391e-01
-5.88937163e-01 -1.11301303e+00 3.95871103e-01 -6.74925506e-01
-9.84000206e-01 7.76542425e-01 8.39343071e-01 3.39984030e-01
1.67341554e+00 -4.14821863e-01 5.70210874e-01 -6.58723935e-02
2.06808716e-01 -1.12938154e+00 7.04596788e-02 -7.43607059e-02
9.01244402e-01 -1.18823016e+00 2.02260181e-01 -8.04608285e-01
-3.42351466e-01 1.17214441e+00 3.24738175e-01 3.45657244e-02
5.04377306e-01 1.64368808e-01 -8.92055780e-02 7.80468285e-02
-3.62241507e-01 -2.24356905e-01 4.12387341e-01 5.88964403e-01
1.69071153e-01 -2.87180722e-01 -4.22835499e-01 1.76675484e-01
4.99950558e-01 1.11805409e-01 5.52199543e-01 7.46670604e-01
-4.92417365e-01 -8.10095370e-01 -9.81090844e-01 -3.75169426e-01
-5.91462612e-01 3.30340751e-02 1.10104149e-02 5.59909940e-01
1.96328551e-01 1.38032651e+00 -2.60042220e-01 -5.18958159e-02
3.56367350e-01 -3.05683941e-01 7.08489001e-01 -4.76810411e-02
-1.55903816e-01 5.55649519e-01 -1.68445617e-01 -5.87108374e-01
-1.02006471e+00 -4.02445257e-01 -6.62854195e-01 -1.56493649e-01
-3.73125404e-01 -3.82165536e-02 7.97803283e-01 6.88358784e-01
1.47052864e-02 6.65721893e-01 9.49554801e-01 -1.05066407e+00
-4.26989466e-01 -6.45812333e-01 -9.24120605e-01 6.63183510e-01
4.73331928e-01 -5.39543569e-01 -6.90203547e-01 4.63271737e-01]
|
[10.802980422973633, -2.5109128952026367]
|
0a25e256-d443-477f-8642-db02206504d2
|
multi-stage-feature-selection-based
|
1708.08750
| null |
http://arxiv.org/abs/1708.08750v1
|
http://arxiv.org/pdf/1708.08750v1.pdf
|
Multi-Stage Feature Selection Based Intelligent Classifier for Classification of Incipient Stage Fire in Building
|
In this study, an early fire detection algorithm has been proposed based on
low cost array sensing system, utilizing gas sensors, dust particles and
ambient sensors such as temperature and humidity sensor. The odor or
smell-print emanated from various fire sources and building construction
materials at early stage are measured. For this purpose, odor profile data from
five common fire sources and three common building construction materials were
used to develop the classification model. Normalized feature extractions of the
smell print data were performed before subjected to prediction classifier.
These features represent the odor signals in the time domain. The obtained
features undergo the proposed multi-stage feature selection technique and
lastly, further reduced by Principal Component Analysis (PCA), a dimension
reduction technique. The hybrid PCA-PNN based approach has been applied on
different datasets from in-house developed system and the portable electronic
nose unit. Experimental classification results show that the dimension
reduction process performed by PCA has improved the classification accuracy and
provided high reliability, regardless of ambient temperature and humidity
variation, baseline sensor drift, the different gas concentration level and
exposure towards different heating temperature range.
|
['Shaharil Mad Saad', 'Allan Melvin Andrew', 'Ammar Zakaria', 'Ali Yeon Md Shakaff']
|
2017-08-12
| null | null | null | null |
['fire-detection']
|
['time-series']
|
[ 5.42338014e-01 -9.29337263e-01 3.83838385e-01 -5.94932400e-02
1.42540798e-01 -6.40704632e-01 2.54257381e-01 3.93854856e-01
-4.38653916e-01 6.08498216e-01 -1.10500038e-01 2.13402733e-01
-7.66287863e-01 -1.00958562e+00 2.03040298e-02 -9.24160480e-01
-1.90155208e-01 3.80888492e-01 -2.83374876e-01 2.98949098e-03
2.01997027e-01 7.70150661e-01 -2.17408299e+00 1.13556154e-01
6.14897788e-01 1.07895339e+00 3.42060924e-01 8.27879190e-01
1.63828239e-01 -5.75249717e-02 -6.38578892e-01 6.67002738e-01
5.70898592e-01 -2.41966322e-01 5.77335805e-02 6.60934076e-02
-5.17493784e-01 2.19473079e-01 6.40128672e-01 7.81498551e-01
5.81890583e-01 6.60281241e-01 1.00970447e+00 -1.08178961e+00
-5.07281482e-01 8.53105560e-02 -2.72154361e-01 2.04377323e-01
3.32095206e-01 1.65614724e-01 2.50392258e-01 -8.64528358e-01
9.49392468e-02 1.11072648e+00 5.37939429e-01 4.83633906e-01
-1.10427272e+00 -7.47276127e-01 -4.20793921e-01 -2.90394239e-02
-1.45031822e+00 -9.32915807e-02 9.09745216e-01 -4.81229365e-01
1.28172624e+00 8.71936083e-01 8.83361936e-01 6.05768025e-01
5.01176238e-01 -1.98079154e-01 1.54132271e+00 -4.46113318e-01
4.67014253e-01 5.39376676e-01 9.29386169e-02 3.54031086e-01
6.55083239e-01 4.26743001e-01 -1.23367302e-01 -1.66949078e-01
2.13000923e-02 5.58425188e-01 -1.22048907e-01 3.67841423e-01
-3.59570384e-01 5.52852929e-01 1.12737902e-01 6.37902558e-01
-8.97936523e-01 -4.24469978e-01 1.57015026e-01 1.75099120e-01
3.90945733e-01 5.29769003e-01 -6.49396956e-01 -7.78167024e-02
-6.92376137e-01 -5.57186939e-02 1.07824230e+00 5.60032010e-01
4.92864519e-01 1.24838594e-02 -1.40069248e-02 7.64686763e-01
7.21139371e-01 1.06324780e+00 5.77764332e-01 -2.11022839e-01
5.11266813e-02 8.41190398e-01 1.35284647e-01 -1.17119265e+00
-4.30197150e-01 -2.42628660e-02 -6.62118793e-01 2.83636451e-01
-2.75450259e-01 -5.02822399e-01 -1.27091324e+00 1.07109046e+00
3.45267177e-01 -1.41211748e-01 3.42344254e-01 7.92556763e-01
6.20040059e-01 8.69354129e-01 2.78214604e-01 -4.93806481e-01
1.50867558e+00 -3.44619066e-01 -9.32463646e-01 8.57004449e-02
-4.01329815e-01 -7.58515000e-01 6.33525431e-01 6.97603941e-01
-2.21816212e-01 -6.05688751e-01 -1.43460476e+00 8.18166673e-01
-1.08518147e+00 2.60241088e-02 5.54751575e-01 1.19845831e+00
-4.64073539e-01 1.00244808e+00 -1.00046515e+00 -7.82856584e-01
1.45296812e-01 7.27635145e-01 -1.78010985e-01 -3.95302325e-02
-8.39763999e-01 7.75758803e-01 1.79292053e-01 4.83840883e-01
-5.97564876e-01 -3.52402180e-01 -3.94836903e-01 -2.85607368e-01
-2.87082076e-01 -4.84175593e-01 6.10593200e-01 -4.35447961e-01
-1.48602569e+00 3.86297405e-01 -3.93532872e-01 -4.53535691e-02
5.51981805e-03 -2.15298429e-01 -9.84768093e-01 4.35851254e-02
-3.84556130e-02 -2.00119749e-01 8.39316010e-01 -1.03162646e+00
-3.82915348e-01 -9.76421535e-01 -8.53462160e-01 1.47124812e-01
-4.03895438e-01 7.26291984e-02 5.54640830e-01 5.15327463e-03
3.28470141e-01 -8.79324198e-01 -3.04215491e-01 -9.30619836e-01
-2.81965196e-01 -7.98847377e-02 1.08062625e+00 -6.16049409e-01
1.20339298e+00 -2.09265375e+00 -1.99909866e-01 1.01881135e+00
-4.58644152e-01 1.70812935e-01 3.32328558e-01 4.99776781e-01
-2.27180988e-01 2.12346330e-01 -4.30198371e-01 2.22698867e-01
-1.17445566e-01 2.66194820e-01 4.32983041e-01 4.42723155e-01
3.65344226e-01 2.48560794e-02 -8.10183287e-01 -3.39554995e-02
6.83661520e-01 9.96509373e-01 1.31602257e-01 2.48219267e-01
1.15086824e-01 4.21660542e-01 -3.98699731e-01 1.05129886e+00
7.62773395e-01 6.99957252e-01 -2.72955745e-01 -4.31040287e-01
-3.34047556e-01 -2.90338427e-01 -1.55313420e+00 1.24741209e+00
-3.75178456e-01 4.68767062e-02 2.95640975e-01 -4.28449482e-01
1.59282565e+00 7.18739331e-01 7.43325233e-01 -5.54311931e-01
5.19360065e-01 3.76043558e-01 3.88166271e-02 -1.05643737e+00
8.03728774e-02 -4.76097763e-01 6.70191705e-01 1.38844147e-01
-1.85972929e-01 -4.08783220e-02 3.91615063e-01 -9.34261680e-01
7.45053768e-01 -1.06988579e-01 1.75599188e-01 -3.32968414e-01
7.76596367e-01 3.15790214e-02 3.42160195e-01 2.35671744e-01
-1.21753469e-01 6.44370839e-02 -4.45861638e-01 -1.30482405e-01
-7.88378775e-01 -6.77667260e-01 -4.92392510e-01 7.76566863e-01
-1.76838130e-01 -1.09195158e-01 -4.31689411e-01 1.03072576e-01
1.15499668e-01 5.76931119e-01 -4.01255429e-01 -1.73326105e-01
-1.02346003e-01 -1.22412038e+00 1.33078292e-01 2.24750444e-01
3.95191342e-01 -1.18009603e+00 -1.03587794e+00 5.84156394e-01
7.65624464e-01 -3.95757705e-01 5.05846798e-01 8.16818774e-01
-1.05265188e+00 -1.28827393e+00 -1.32180467e-01 -3.64875793e-01
3.76626581e-01 -6.74871951e-02 4.62492615e-01 -3.44497502e-01
-9.46985841e-01 5.55373967e-01 -4.54739511e-01 -1.34711659e+00
-1.07277371e-01 -3.43001127e-01 2.36127898e-01 4.80759256e-02
1.12448287e+00 -1.04742324e+00 -6.90518379e-01 -5.79243228e-02
-7.98023045e-01 -7.90548503e-01 5.96474826e-01 3.56135666e-01
5.81460655e-01 9.49270785e-01 2.96308368e-01 -9.44968224e-01
1.10859239e+00 -6.21227145e-01 -1.71386167e-01 -1.04668979e-02
-1.17038286e+00 3.61684002e-02 5.17806113e-01 -1.13070488e-01
-9.89895344e-01 8.33161473e-02 -2.27039121e-02 1.08852389e-03
-8.15122008e-01 3.92158926e-01 -4.30588394e-01 2.16002241e-01
9.79370892e-01 7.36808926e-02 -3.10843475e-02 -6.66848838e-01
-3.71244788e-01 9.50400352e-01 2.79230833e-01 -4.87066239e-01
9.54387128e-01 2.83254564e-01 2.96398610e-01 -1.15824902e+00
-6.42490461e-02 -7.93369114e-01 -6.04852438e-01 -3.29166591e-01
1.12244129e+00 -5.90467215e-01 -6.80621743e-01 3.17722023e-01
-9.13767934e-01 4.84844774e-01 -1.14909276e-01 7.61896431e-01
1.62882656e-01 -2.49052778e-01 -2.90054709e-01 -1.86067164e+00
-1.14337432e+00 -6.91679537e-01 3.51974636e-01 6.06172860e-01
-2.94551343e-01 -5.46308577e-01 1.71520427e-01 -3.96440402e-02
5.81856608e-01 9.72841561e-01 7.27586627e-01 -6.41530931e-01
4.49743867e-02 -5.75054288e-01 4.50922817e-01 6.38706565e-01
8.80870461e-01 3.28822643e-01 -1.21675026e+00 -9.61179882e-02
5.39008856e-01 4.97623026e-01 6.48331106e-01 1.62359089e-01
5.59453547e-01 -1.32463202e-01 -2.69808888e-01 6.79849684e-01
2.09405017e+00 1.30879390e+00 1.29483655e-01 4.34767157e-01
6.00984633e-01 4.30394679e-01 7.09856212e-01 5.82265198e-01
-6.27361238e-01 -2.64227092e-01 4.63284433e-01 1.27957672e-01
3.19270253e-01 1.65449321e-01 3.69794846e-01 6.29004598e-01
-7.64112592e-01 -1.68898553e-01 -6.57328367e-01 1.61913544e-01
-1.30914223e+00 -1.04214859e+00 -3.95620555e-01 1.99033213e+00
3.79256099e-01 -1.68221980e-01 1.59770682e-01 8.96928132e-01
6.46648824e-01 -1.91064551e-01 -4.64221358e-01 -7.92084455e-01
3.14671127e-03 1.11875629e+00 6.73535824e-01 2.11223394e-01
-1.14038289e+00 4.10690233e-02 5.67766047e+00 -1.50838673e-01
-1.21257031e+00 -8.69531259e-02 -8.84111598e-02 -1.49162397e-01
3.91228758e-02 -2.72576392e-01 -5.05218863e-01 3.54651392e-01
1.13299668e+00 -6.56135753e-02 7.62241364e-01 9.14126098e-01
5.12168169e-01 -3.01213712e-01 -9.93544579e-01 9.86633420e-01
-4.11379663e-03 -3.21276963e-01 -1.79080918e-01 5.70689002e-03
4.88715559e-01 -1.64262727e-01 1.66864824e-02 -2.16077849e-01
-4.06928122e-01 -9.61577237e-01 -1.57939624e-02 8.56928349e-01
3.50727916e-01 -1.04405200e+00 7.76185036e-01 -3.59358974e-02
-1.48284674e+00 -5.62057793e-01 -3.07027668e-01 -3.12169045e-01
-1.88446090e-01 6.84257805e-01 -1.13587606e+00 6.77474916e-01
5.97002745e-01 5.28956711e-01 -5.50764382e-01 9.11765695e-01
7.94453695e-02 5.07618427e-01 -8.70051503e-01 -3.97184610e-01
-1.24807633e-01 -7.03095317e-01 6.53642952e-01 1.33035123e+00
7.75036871e-01 1.70030802e-01 -3.36513609e-01 1.06248796e+00
7.67626107e-01 3.60854626e-01 -8.67167950e-01 -4.50394094e-01
7.36790597e-01 1.33116555e+00 -8.14158857e-01 2.73484662e-02
-1.46157034e-02 6.44829094e-01 -8.74886930e-01 3.74536335e-01
-3.66181880e-01 -6.85057342e-01 6.41757905e-01 1.22111283e-01
8.03397074e-02 -1.83268845e-01 -2.25400463e-01 -5.21696627e-01
-3.45830694e-02 -4.64757711e-01 3.31029236e-01 -3.98788214e-01
-1.30751681e+00 3.17782104e-01 -1.70611236e-02 -8.09778690e-01
4.00149934e-02 -9.59619284e-01 -9.50808465e-01 1.18120944e+00
-9.19304013e-01 -7.16857910e-01 -4.84663218e-01 7.51794994e-01
2.93443233e-01 -3.90570551e-01 1.43933618e+00 4.11144733e-01
-8.85170877e-01 -2.62533640e-03 2.13580012e-01 -2.15155691e-01
2.13183939e-01 -1.15973985e+00 -6.76485658e-01 9.96823311e-01
-1.93428338e-01 7.85157442e-01 7.60686576e-01 -8.08773637e-01
-1.66596735e+00 -1.11842835e+00 3.87854546e-01 -9.02958065e-02
4.40068208e-02 -3.15044910e-01 -4.30033386e-01 5.22389747e-02
9.63360891e-02 -4.75990981e-01 1.22773123e+00 -1.44502029e-01
3.60021830e-01 -5.16452074e-01 -1.65996575e+00 -6.38900772e-02
4.39428359e-01 -4.09802288e-01 -4.98594999e-01 2.91343123e-01
1.66191131e-01 1.70746982e-01 -1.33166146e+00 2.66378760e-01
7.85163879e-01 -8.17325294e-01 8.27305615e-01 -5.17162979e-01
-4.13112193e-02 -5.60952067e-01 -3.02275002e-01 -8.62811267e-01
-4.83496219e-01 -2.66886979e-01 -2.00268440e-02 1.69838834e+00
4.85283732e-01 -5.79418242e-01 5.25198221e-01 1.08791506e+00
6.28835410e-02 -3.41465861e-01 -5.52376807e-01 -5.83879769e-01
-8.49650443e-01 -3.62506002e-01 7.74574399e-01 6.32125735e-01
-6.51910231e-02 4.04767632e-01 8.56450107e-03 3.81493509e-01
6.00661695e-01 -3.16888839e-01 1.94814235e-01 -1.76193178e+00
-6.16932474e-02 1.28602743e-01 -4.63752687e-01 5.21398187e-01
-5.15653849e-01 -8.48871529e-01 -2.84330677e-02 -1.58423960e+00
-3.38012695e-01 -1.99193925e-01 -8.11532438e-01 1.53286919e-01
-5.80446608e-02 -1.33311331e-01 6.35647327e-02 -3.71799134e-02
4.07667071e-01 1.71316266e-01 7.83551872e-01 -1.91466227e-01
-8.05463076e-01 2.39002123e-01 -4.16603506e-01 5.10031044e-01
9.40299869e-01 -7.62175679e-01 -6.10966086e-01 -4.99728601e-03
9.27375630e-02 -4.37114209e-01 9.65779424e-02 -1.57171965e+00
6.78681433e-02 -1.04757614e-01 1.08762217e+00 -6.99871659e-01
3.87471765e-01 -1.67696977e+00 8.39204848e-01 9.10232663e-01
4.37615067e-01 2.44809136e-01 3.18220735e-01 6.76289797e-01
-5.82429469e-02 -1.84188947e-01 4.00550216e-01 -3.00579697e-01
-6.72878325e-01 2.52590209e-01 -5.27557731e-01 -8.87178242e-01
1.19514894e+00 -7.78868198e-01 9.80884582e-02 5.21475434e-01
-1.11607420e+00 -2.16291457e-01 6.09457726e-03 7.74140283e-02
6.58541083e-01 -1.22845984e+00 -2.91482419e-01 6.73568606e-01
-1.53778195e-01 3.27110030e-02 9.93621498e-02 7.38223016e-01
-4.66667384e-01 3.44145626e-01 -5.95647335e-01 -4.80008692e-01
-1.49894774e+00 6.04681492e-01 2.41029218e-01 2.71166474e-01
-9.15596262e-02 7.44840026e-01 -9.99549866e-01 -1.64051905e-01
-1.43002555e-01 -5.56331515e-01 -7.72650898e-01 3.46498281e-01
2.75559872e-01 8.70406508e-01 3.73596996e-01 -2.67042816e-01
-5.87516069e-01 7.85551369e-01 6.21526837e-01 -1.33751228e-01
1.84238505e+00 1.78423792e-01 -3.88847411e-01 9.22633111e-01
1.19803989e+00 -9.77197438e-02 -5.24684191e-01 4.62275475e-01
1.18322887e-01 -2.40170136e-01 -2.12284476e-01 -9.41645145e-01
-7.83422351e-01 5.11075497e-01 1.45512486e+00 3.56559068e-01
1.52690899e+00 -8.26884449e-01 4.20354396e-01 5.60943425e-01
-9.04304832e-02 -1.33574474e+00 -6.66501701e-01 1.68588609e-01
6.45712614e-01 -1.00263524e+00 1.05998904e-01 -1.88934043e-01
-1.23201944e-01 1.37334085e+00 3.89663070e-01 -3.85736644e-01
9.96637166e-01 4.54368830e-01 -6.80985227e-02 -5.45959771e-01
-5.51791728e-01 -1.43172681e-01 1.60525724e-01 1.16628802e+00
6.35043085e-01 4.16480124e-01 -5.10110080e-01 9.01809633e-01
-4.13245738e-01 1.57218382e-01 -4.11805287e-02 1.17950666e+00
-6.80169344e-01 -1.03866577e+00 -9.55425203e-01 8.60477984e-01
-3.85613292e-01 1.78005174e-01 -4.61378396e-01 8.00824940e-01
7.87015021e-01 1.52194989e+00 -1.08230010e-01 -9.72661734e-01
7.52309740e-01 5.16694725e-01 3.45152944e-01 -5.59388399e-01
-7.77065635e-01 1.65571809e-01 5.12370542e-02 -2.96778023e-01
-7.96112657e-01 -6.92090511e-01 -1.15108228e+00 4.32011709e-02
-3.35664988e-01 4.81508762e-01 1.41933513e+00 7.96831727e-01
5.52666970e-02 7.35087872e-01 8.13421190e-01 -7.09542096e-01
-1.82429910e-01 -1.31261098e+00 -1.12105656e+00 4.74574685e-01
2.49818146e-01 -5.86443901e-01 -7.16552615e-01 -4.87219915e-02]
|
[9.884026527404785, -1.5238770246505737]
|
312d1f31-771e-4ed3-9c28-72743bf9ece4
|
cov-ti-net-transferred-initialization-with
|
2209.09556
| null |
https://arxiv.org/abs/2209.09556v1
|
https://arxiv.org/pdf/2209.09556v1.pdf
|
CoV-TI-Net: Transferred Initialization with Modified End Layer for COVID-19 Diagnosis
|
This paper proposes transferred initialization with modified fully connected layers for COVID-19 diagnosis. Convolutional neural networks (CNN) achieved a remarkable result in image classification. However, training a high-performing model is a very complicated and time-consuming process because of the complexity of image recognition applications. On the other hand, transfer learning is a relatively new learning method that has been employed in many sectors to achieve good performance with fewer computations. In this research, the PyTorch pre-trained models (VGG19\_bn and WideResNet -101) are applied in the MNIST dataset for the first time as initialization and with modified fully connected layers. The employed PyTorch pre-trained models were previously trained in ImageNet. The proposed model is developed and verified in the Kaggle notebook, and it reached the outstanding accuracy of 99.77% without taking a huge computational time during the training process of the network. We also applied the same methodology to the SIIM-FISABIO-RSNA COVID-19 Detection dataset and achieved 80.01% accuracy. In contrast, the previous methods need a huge compactional time during the training process to reach a high-performing model. Codes are available at the following link: github.com/dipuk0506/SpinalNet
|
['Saeid Nahavandi', 'Abbas Khorsavi', 'Shady Mohamed', 'Farzin Tabarsinezhad', 'Keshav Kumar', 'Houshyar Asadi', 'Abadhan S. Sabyasachi', 'H M Dipu Kabir', 'Subrota Kumar Mondal', 'Mohammad Reza Chalak Qazani', 'Sadia Khanam']
|
2022-09-20
| null | null | null | null |
['covid-19-detection']
|
['medical']
|
[ 1.49362564e-01 1.21869184e-01 -5.86419329e-02 -1.70871377e-01
-3.41940284e-01 -1.34596294e-02 3.34422916e-01 -2.01459378e-02
-1.06461966e+00 9.10490692e-01 -4.50654060e-01 -4.14318621e-01
-2.79081523e-01 -8.27222168e-01 -6.57105863e-01 -7.91154027e-01
1.78045645e-01 7.09172010e-01 3.54856104e-01 -2.67470896e-01
1.55800641e-01 7.93089449e-01 -1.47532177e+00 2.27173641e-01
8.74544561e-01 8.65632296e-01 7.23378241e-01 5.14187396e-01
1.59224957e-01 7.55910814e-01 -5.15107572e-01 -2.98738420e-01
3.11512202e-02 -2.73857772e-01 -7.83108115e-01 -2.32795790e-01
4.66620177e-02 -2.35469148e-01 -2.62968749e-01 1.01521158e+00
5.85039616e-01 -8.38026404e-02 7.60469019e-01 -8.05900931e-01
-6.72700033e-02 5.99262655e-01 -3.17426920e-01 3.96876007e-01
-4.69027758e-01 -1.26188071e-02 3.04153621e-01 -7.81460166e-01
5.60448706e-01 7.10522115e-01 6.04936182e-01 5.26803672e-01
-5.23828983e-01 -1.02526319e+00 -4.14206117e-01 5.50845325e-01
-1.46585369e+00 7.15402365e-02 3.83439392e-01 -5.39401591e-01
8.31150949e-01 4.39268127e-02 7.36817241e-01 9.29295719e-01
3.42621624e-01 5.04320204e-01 1.19191206e+00 -3.20140243e-01
-3.69959921e-02 3.00122947e-01 2.71157265e-01 8.86771441e-01
5.98135471e-01 -8.27737376e-02 -3.86147201e-02 4.15406317e-01
8.81234229e-01 1.36982230e-02 -1.35628656e-01 1.66214615e-01
-9.90223527e-01 9.11265135e-01 8.75978410e-01 8.63704264e-01
-3.34278464e-01 -6.64734468e-02 5.25186777e-01 2.55764332e-02
1.74815312e-01 3.01128507e-01 -4.21056509e-01 -5.32878339e-02
-9.22853768e-01 -2.25605980e-01 4.18858320e-01 7.50335813e-01
4.09364998e-01 1.57361791e-01 2.72206753e-01 8.21239829e-01
-4.60841693e-02 2.66301841e-01 8.39915633e-01 -2.30890825e-01
2.37098947e-01 5.99488914e-01 -4.30322558e-01 -8.79870117e-01
-7.27425516e-01 -9.98040795e-01 -1.14725292e+00 3.39487314e-01
4.63370055e-01 -1.18212238e-01 -1.19979966e+00 1.18470967e+00
1.27174612e-02 8.23018327e-02 4.19717729e-02 7.48566270e-01
1.01947987e+00 4.94627059e-01 5.39236218e-02 1.23553567e-01
1.28413618e+00 -1.00552952e+00 -5.50688744e-01 1.05621181e-01
7.81140387e-01 -7.18323171e-01 5.89149535e-01 6.81030035e-01
-9.58817065e-01 -7.22357154e-01 -1.21446693e+00 1.92373127e-01
-7.48700321e-01 7.71157205e-01 7.61756063e-01 6.16831243e-01
-1.07225966e+00 6.37125134e-01 -8.86636972e-01 -6.06174409e-01
6.05263054e-01 6.65733159e-01 -4.17234838e-01 1.58070889e-03
-1.13182759e+00 1.23313034e+00 9.12642419e-01 2.73748964e-01
-9.29522455e-01 -2.47258112e-01 -3.65221471e-01 -1.11232758e-01
2.80363411e-01 -4.29779530e-01 8.49620223e-01 -9.46397424e-01
-1.33996344e+00 9.76338387e-01 3.52489322e-01 -6.86597347e-01
6.60400331e-01 3.37577686e-02 -4.62125033e-01 3.29532593e-01
-1.90228477e-01 7.49077916e-01 5.81774950e-01 -7.24338770e-01
-6.45484924e-01 -2.14839861e-01 -9.38027874e-02 1.53168589e-01
-4.31068450e-01 -5.68213407e-03 -4.90240186e-01 -5.92009425e-01
1.29848182e-01 -9.58810866e-01 -1.07404292e-01 -3.20120066e-01
-2.86295652e-01 -9.66087058e-02 6.58272326e-01 -8.10190201e-01
9.06682193e-01 -2.05987215e+00 -1.14682458e-01 3.64385575e-01
7.84992576e-02 8.01461041e-01 2.97753632e-01 -1.42413564e-02
-3.81818324e-01 -9.76361409e-02 -2.19160542e-01 1.91412084e-02
-5.88955522e-01 2.43336573e-01 3.93635482e-01 4.94279563e-01
-2.12703273e-02 7.08536565e-01 -6.44845724e-01 -7.59873629e-01
4.11334366e-01 4.65606362e-01 -5.32156408e-01 2.84084454e-02
3.12215149e-01 5.95245361e-01 -2.08251595e-01 5.25214314e-01
6.55121684e-01 -2.68032193e-01 1.21013429e-02 -3.50494832e-01
-2.70157997e-02 -2.55181402e-01 -9.01801825e-01 1.50140834e+00
-3.58949929e-01 3.80475521e-01 -1.38757408e-01 -1.52066255e+00
1.03404403e+00 3.36007684e-01 4.38073128e-01 -6.30709767e-01
7.54173636e-01 5.44145644e-01 4.73414868e-01 -5.65548062e-01
3.99316341e-01 -1.19361304e-01 2.00793192e-01 -9.89733711e-02
4.10816193e-01 3.57146002e-02 3.35183710e-01 -6.79837540e-02
6.79972351e-01 -2.91070249e-02 2.71045268e-01 -3.52044076e-01
8.31335247e-01 3.03831190e-01 4.29296702e-01 5.56467235e-01
-6.03406802e-02 4.63647395e-01 2.10391268e-01 -4.40224588e-01
-1.04633188e+00 -7.87870467e-01 -5.63529968e-01 5.10351121e-01
-9.30481851e-02 4.59042452e-02 -8.57918978e-01 -4.70219493e-01
-3.98154736e-01 3.98199886e-01 -6.74020469e-01 -1.67528823e-01
-4.97749537e-01 -9.06820774e-01 8.49113643e-01 5.31724632e-01
8.79371643e-01 -1.32197928e+00 -4.56681728e-01 2.38725692e-01
1.61859185e-01 -1.02834749e+00 2.16132373e-01 3.72798949e-01
-1.05940509e+00 -1.18960452e+00 -1.12706757e+00 -1.10576940e+00
9.64262962e-01 -3.63209188e-01 4.76341724e-01 3.20554763e-01
-6.17778480e-01 -2.42942601e-01 -2.68846214e-01 -6.26018405e-01
-3.38544905e-01 4.51273292e-01 -2.51602493e-02 -1.10232264e-01
2.06568360e-01 -3.04266930e-01 -4.47466075e-01 1.96441561e-01
-8.23036909e-01 3.73532861e-01 1.06793380e+00 1.06534767e+00
5.35777211e-01 1.27038896e-01 6.97234273e-01 -9.66381848e-01
3.16629440e-01 -3.27556938e-01 -6.75938785e-01 -6.84429482e-02
-5.74046195e-01 -4.15065587e-01 6.74495578e-01 -3.09274584e-01
-9.85322177e-01 3.56145464e-02 -6.28116846e-01 -4.40463156e-01
-3.34543347e-01 5.13013482e-01 1.24043822e-01 -1.69396833e-01
5.28727055e-01 2.36128077e-01 2.00430155e-01 -4.56215382e-01
-1.20993674e-01 6.60688281e-01 6.76127136e-01 -1.56965166e-01
5.33910573e-01 4.69304949e-01 1.26879051e-01 -8.72025073e-01
-5.53305745e-01 -3.61024529e-01 -8.28194797e-01 -2.40180343e-01
1.16292202e+00 -9.11840796e-01 -4.79920864e-01 9.38871086e-01
-7.60391057e-01 -2.13210225e-01 4.02883776e-02 1.03316164e+00
-2.64385194e-01 2.46944893e-02 -7.19551623e-01 -2.51510173e-01
-6.45795047e-01 -1.36176383e+00 2.11366504e-01 4.11818773e-01
1.77533686e-01 -9.38069165e-01 -2.85165131e-01 3.52727085e-01
5.25779963e-01 2.96969920e-01 8.60877335e-01 -7.83991635e-01
-1.56724721e-01 -4.33595866e-01 -3.30213219e-01 6.43175185e-01
-9.28336009e-03 -6.29226640e-02 -9.13632691e-01 -2.55618066e-01
-1.54069766e-01 -3.03339243e-01 8.89716864e-01 3.78517509e-01
1.34215462e+00 1.63034454e-01 -3.83544803e-01 7.53336787e-01
1.57785797e+00 6.54051483e-01 6.95434391e-01 6.93358779e-01
6.36351883e-01 1.00490324e-01 5.71070969e-01 1.69909358e-01
-6.04408048e-02 4.56102759e-01 4.80427027e-01 -4.48070943e-01
-1.90679505e-01 8.05820301e-02 -2.22016498e-01 1.00037754e+00
-5.87887645e-01 -3.42270732e-02 -9.99605238e-01 5.10485291e-01
-1.42032206e+00 -8.04331839e-01 -4.08039391e-01 1.85851848e+00
6.34014487e-01 5.00630915e-01 -7.69419521e-02 4.41659272e-01
7.31988847e-01 -3.19870502e-01 -2.20912665e-01 -5.28955698e-01
-1.60623770e-02 6.88532531e-01 5.79204559e-01 1.75242797e-01
-1.03630149e+00 8.76867592e-01 4.94347954e+00 1.02871919e+00
-1.53390431e+00 3.06096256e-01 5.44870496e-01 -2.10142452e-02
5.55100858e-01 -4.38828498e-01 -9.68911111e-01 4.98370260e-01
1.04425776e+00 2.85796821e-01 5.48992939e-02 8.36421967e-01
6.31440878e-02 -2.63887584e-01 -4.41454858e-01 1.01390934e+00
8.38296115e-02 -1.35845506e+00 -2.16381950e-03 -1.72710568e-02
5.60295105e-01 2.51767218e-01 2.86316928e-02 3.17862272e-01
-2.23089606e-01 -1.08131397e+00 3.64819199e-01 5.62369883e-01
9.54104304e-01 -1.10258532e+00 1.30551171e+00 4.36394930e-01
-8.59161019e-01 -6.52543232e-02 -6.37462437e-01 7.96793997e-02
-1.20788217e-01 6.25621974e-01 -1.15855145e+00 7.33386219e-01
8.87329340e-01 5.50255418e-01 -6.49708211e-01 1.25696826e+00
-2.54921615e-01 8.22191834e-01 -2.10143581e-01 -2.12765843e-01
4.63776559e-01 -2.35052049e-01 1.11030638e-01 1.17990494e+00
3.47314060e-01 -1.53929740e-01 -3.12143981e-01 5.51197886e-01
-1.82654727e-02 3.61258924e-01 -4.46349859e-01 7.25063980e-02
-9.16176960e-02 1.52254772e+00 -1.04057062e+00 -5.90267181e-01
-2.44045153e-01 6.66108787e-01 1.39179990e-01 -8.82458836e-02
-1.12259078e+00 -6.58395469e-01 3.97353759e-03 5.87233119e-02
3.05227369e-01 -1.92068163e-02 -2.75601596e-01 -8.74545276e-01
-3.93709540e-01 -6.82827234e-01 3.75382155e-01 -8.00180912e-01
-8.13500047e-01 1.01961398e+00 2.99984347e-02 -1.26289499e+00
-8.19893479e-02 -9.94860351e-01 -5.50006509e-01 9.65335608e-01
-1.31393099e+00 -1.04834223e+00 -5.77945471e-01 8.23695004e-01
3.29619735e-01 -3.59422445e-01 8.68115723e-01 5.68285108e-01
-7.07414269e-01 6.21924698e-01 2.40184486e-01 4.03270185e-01
5.34036458e-01 -9.16735709e-01 -2.27137089e-01 6.28985226e-01
-3.30325603e-01 4.85160977e-01 2.12211266e-01 -5.93412936e-01
-7.48778701e-01 -1.04618406e+00 7.02896774e-01 2.60082752e-01
4.94470507e-01 -1.92946598e-01 -8.05609703e-01 6.47209406e-01
3.78289551e-01 -4.47217822e-02 4.44759786e-01 -4.47187215e-01
3.56996804e-01 -6.28295913e-02 -1.23936141e+00 3.62256020e-01
6.87870443e-01 -1.48136035e-01 -6.51144266e-01 3.97624969e-01
3.11642557e-01 -5.78989267e-01 -9.71197605e-01 5.98549128e-01
4.07966226e-01 -6.49747550e-01 7.95341611e-01 -2.73290157e-01
3.27305228e-01 -2.52032816e-01 2.68653899e-01 -1.06349289e+00
-1.11525640e-01 -1.60404965e-02 4.92886305e-01 9.91848946e-01
5.14442861e-01 -8.94560158e-01 7.34296620e-01 -3.12062930e-02
-3.34335417e-01 -1.09625661e+00 -7.32385457e-01 -7.16431260e-01
8.88917148e-02 -2.50862151e-01 1.52411252e-01 9.33048010e-01
-2.13886529e-01 1.28627062e-01 -5.31220622e-02 -1.83078796e-02
4.58949208e-01 -3.22956294e-01 4.58693504e-01 -1.16459262e+00
-2.29961932e-01 -2.51156092e-01 -5.71783423e-01 -3.52639884e-01
-2.61768937e-01 -1.15057504e+00 -3.01688433e-01 -1.46528780e+00
5.86180985e-02 -6.15473151e-01 -4.43062603e-01 5.19610763e-01
2.05806226e-01 4.68155116e-01 1.12206548e-01 1.89672485e-01
-8.26116577e-02 9.42741558e-02 1.38783395e+00 2.72483341e-02
1.20948069e-01 1.73401177e-01 -2.87230045e-01 9.38354254e-01
1.17438436e+00 -4.38295603e-01 -4.83218491e-01 -6.43203557e-02
-6.31641671e-02 -1.99731305e-01 2.30367497e-01 -1.44966018e+00
3.99568170e-01 3.96948755e-01 8.13219726e-01 -1.02869570e+00
3.60293686e-01 -8.87686670e-01 3.42814565e-01 1.09075081e+00
4.82848026e-02 9.84496325e-02 4.14728165e-01 -8.14344808e-02
-2.91823059e-01 -7.06644893e-01 1.04049301e+00 -3.59168112e-01
-9.26422477e-01 3.36388588e-01 -5.40543318e-01 -3.55078399e-01
1.19957364e+00 -1.24988109e-01 -2.84505695e-01 6.83492348e-02
-9.67280805e-01 -1.88390419e-01 1.27188623e-01 1.50133744e-01
5.37514150e-01 -8.60523701e-01 -5.28348267e-01 2.63391525e-01
-1.44304216e-01 3.73429209e-02 4.89701599e-01 1.22846210e+00
-1.06816685e+00 5.43685079e-01 -8.55856776e-01 -6.68797433e-01
-1.23784077e+00 4.63504821e-01 5.20823598e-01 -4.27153409e-01
-6.22788429e-01 8.10413003e-01 3.69816311e-02 -4.25454110e-01
1.48201913e-01 -3.09791774e-01 -7.63798296e-01 1.06050253e-01
2.32894748e-01 2.49069929e-01 5.37653029e-01 -6.26521230e-01
-3.50421399e-01 5.54727435e-01 -2.48149633e-01 2.56134480e-01
1.47006238e+00 4.22521412e-01 4.72070929e-03 1.86677799e-02
1.26384902e+00 -4.58942890e-01 -6.89957678e-01 4.40212479e-03
-1.15984991e-01 -8.21277685e-03 1.93248078e-01 -8.31255078e-01
-1.34263778e+00 1.00730741e+00 9.34105873e-01 -4.26271319e-01
1.14753067e+00 -2.30188653e-01 5.78069925e-01 3.74076277e-01
4.69419360e-01 -1.13601720e+00 -9.07118022e-02 5.08355677e-01
7.05584586e-01 -1.06568694e+00 4.37388793e-02 -2.66094476e-01
-5.47462821e-01 1.21902680e+00 8.04660320e-01 -3.62760425e-01
7.18218744e-01 2.52727211e-01 -7.18469499e-03 -2.27673918e-01
-2.48789772e-01 -2.01684281e-01 1.94678053e-01 4.38487321e-01
3.31998438e-01 4.95538190e-02 -8.28884006e-01 4.86992598e-01
-3.09305519e-01 3.55586708e-01 4.61505383e-01 9.19543266e-01
-2.25790262e-01 -8.59310865e-01 -1.98819935e-01 7.22487211e-01
-7.24508524e-01 -6.97063878e-02 1.21140234e-01 1.30056596e+00
4.33083266e-01 5.09026289e-01 3.77065241e-02 -4.64350522e-01
1.56556383e-01 -1.01131082e-01 3.89202714e-01 -4.32084411e-01
-7.40139246e-01 2.67556664e-02 -7.11714989e-03 -1.93137899e-01
-4.63987887e-01 -3.74780446e-01 -1.54956901e+00 -2.11302653e-01
-3.37057739e-01 1.72389254e-01 1.02740359e+00 8.97520244e-01
-1.29360452e-01 9.81637836e-01 2.32962117e-01 -7.02977240e-01
-2.19723865e-01 -1.27921879e+00 -6.65874541e-01 1.60331756e-01
-1.67805135e-01 -7.61426270e-01 -1.53168961e-01 -6.71782866e-02]
|
[14.924219131469727, -2.5939557552337646]
|
61847e5b-33df-4591-9463-24c3222ad999
|
resource-constrained-neural-networks-for-5g
|
2107.11070
| null |
https://arxiv.org/abs/2107.11070v1
|
https://arxiv.org/pdf/2107.11070v1.pdf
|
Resource Constrained Neural Networks for 5G Direction-of-Arrival Estimation in Micro-controllers
|
With the introduction of shared spectrum sensing and beam-forming based multi-antenna transceivers, 5G networks demand spectrum sensing to identify opportunities in time, frequency, and spatial domains. Narrow beam-forming makes it difficult to have spatial sensing (direction-of-arrival, DoA, estimation) in a centralized manner, and with the evolution of paradigms such as artificial intelligence of Things (AIOT), ultra-reliable low latency communication (URLLC) services and distributed networks, intelligence for edge devices (Edge-AI) is highly desirable. It helps to reduce the data-communication overhead compared to cloud-AI-centric networks and is more secure and free from scalability limitations. However, achieving desired functional accuracy is a challenge on edge devices such as microcontroller units (MCU) due to area, memory, and power constraints. In this work, we propose low complexity neural network-based algorithm for accurate DoA estimation and its efficient mapping on the off-the-self MCUs. An ad-hoc graphical-user interface (GUI) is developed to configure the STM32 NUCLEO-H743ZI2 MCU with the proposed algorithm and to validate its functionality. The performance of the proposed algorithm is analyzed for different signal-to-noise ratios (SNR), word-length, the number of antennas, and DoA resolution. In-depth experimental results show that it outperforms the conventional statistical spatial sensing approach.
|
['Hem-Dutt Dabral', 'Danilo Pau', 'S. J. Darak', 'Shivam Chandhok', 'Romesh Rajoria', 'Piyush Sahoo']
|
2021-07-23
| null | null | null | null |
['direction-of-arrival-estimation']
|
['audio']
|
[ 7.94275627e-02 -3.45990777e-01 -1.54391870e-01 -6.75110966e-02
-2.21018210e-01 -3.98405492e-01 -2.86741555e-02 -3.20474535e-01
-1.30761817e-01 8.75086188e-01 -5.27494699e-02 -6.96114779e-01
-6.52092934e-01 -8.68987918e-01 2.03381255e-01 -6.86729550e-01
-3.89945865e-01 1.45777306e-02 8.94431025e-02 7.56443888e-02
-2.47793153e-01 6.74068511e-01 -1.03934193e+00 -3.46487254e-01
6.26352012e-01 1.85159504e+00 4.47647840e-01 6.06975675e-01
2.68724114e-01 5.19780159e-01 -8.87796462e-01 2.31747374e-01
4.71871883e-01 -5.43136537e-01 5.84391765e-02 -1.81347117e-01
-3.58298510e-01 -2.02435821e-01 -3.44857395e-01 1.05487514e+00
1.08353055e+00 -2.67520428e-01 3.31042737e-01 -1.18899667e+00
-3.24196458e-01 6.39055729e-01 -4.80443299e-01 3.05180848e-01
4.44181589e-03 -1.15311451e-01 4.84523267e-01 -4.04600829e-01
4.11386639e-01 4.94588882e-01 8.57985973e-01 1.08577311e-01
-4.29961354e-01 -8.21544290e-01 -6.61034107e-01 4.66042161e-01
-1.71805394e+00 -3.42235059e-01 8.16634417e-01 1.54894471e-01
7.44154572e-01 4.23996121e-01 7.48246193e-01 4.98897463e-01
2.80297101e-01 -4.28918973e-02 9.74719644e-01 -8.09272170e-01
7.62341142e-01 -1.38332754e-01 -3.35216522e-01 6.16460443e-01
6.88820183e-01 1.86069995e-01 -4.56376076e-01 3.76591533e-02
8.57026041e-01 4.73108441e-02 -1.64070755e-01 -4.68390435e-02
-1.17017508e+00 2.92714059e-01 3.83274794e-01 8.56252849e-01
-7.98249006e-01 4.53236192e-01 -1.99403420e-01 3.87652665e-01
1.68871898e-02 5.13781011e-02 -3.01324546e-01 -3.43258709e-01
-9.99765813e-01 -6.04862332e-01 6.72806144e-01 1.01548243e+00
4.30557311e-01 6.73372030e-01 1.00806698e-01 4.66201842e-01
4.99434739e-01 9.68341947e-01 4.58766460e-01 -6.75468922e-01
1.22909872e-02 1.42594442e-01 1.02476247e-01 -1.05742502e+00
-1.09563971e+00 -1.49550939e+00 -1.32914197e+00 -3.51621322e-02
8.22864398e-02 -7.84289300e-01 -5.64128816e-01 1.47212005e+00
1.36239737e-01 3.04980993e-01 1.33184418e-01 8.06194186e-01
4.36319858e-01 5.63013494e-01 -3.69729251e-01 -6.65212929e-01
1.12290275e+00 -4.95442957e-01 -9.04814422e-01 -2.66035348e-01
3.69559586e-01 -6.06102347e-01 1.24157377e-01 4.72318977e-01
-6.68918729e-01 -5.76049149e-01 -1.53618371e+00 7.69776762e-01
-1.60609946e-01 3.48043203e-01 5.48640490e-01 1.56637633e+00
-1.27150893e+00 -6.23277621e-03 -5.50062954e-01 -4.61778998e-01
2.95311093e-01 3.79549503e-01 3.24239701e-01 8.22280571e-02
-1.04314733e+00 5.70222735e-01 1.65330529e-01 -3.75121683e-02
-2.48504907e-01 -6.10670924e-01 -2.97365695e-01 4.26174641e-01
1.61663413e-01 -7.18891203e-01 7.96630144e-01 -7.32391894e-01
-1.73143470e+00 -1.38181522e-01 1.46343529e-01 -7.78994024e-01
2.34334525e-02 4.89340186e-01 -1.32655799e+00 6.93130195e-02
-1.68595120e-01 1.20909907e-01 5.49599349e-01 -6.13057911e-01
-7.70948470e-01 -4.68433857e-01 -3.13624918e-01 -1.70895085e-01
-3.52719307e-01 -1.50014579e-01 1.35223895e-01 -4.93544966e-01
6.38688445e-01 -8.69990051e-01 -3.56954306e-01 -9.53743979e-02
-1.83747768e-01 3.74400198e-01 1.05561638e+00 -2.17131868e-01
1.32972312e+00 -2.00961304e+00 -6.15720510e-01 7.03689218e-01
2.01450422e-01 2.76425749e-01 3.45031172e-01 4.21757668e-01
1.58533975e-01 -1.81574866e-01 3.50085914e-01 5.24766482e-02
-1.98120371e-01 -9.67109054e-02 2.19105303e-01 4.48252738e-01
-6.70289695e-01 5.86996496e-01 -3.51647824e-01 -7.14059034e-03
5.08526444e-01 2.77984679e-01 -2.41758689e-01 -4.09711629e-01
6.62335604e-02 3.70049447e-01 -6.02745175e-01 9.60711300e-01
6.19649708e-01 -4.56059217e-01 4.25228447e-01 -4.23256367e-01
-3.56962234e-01 -1.90828577e-01 -1.66775608e+00 1.62358832e+00
-9.62928414e-01 7.05232620e-01 3.10243696e-01 -8.90133739e-01
1.07682824e+00 6.12550557e-01 6.67126715e-01 -1.01617062e+00
5.33280015e-01 4.96895283e-01 2.61753649e-01 -1.12560473e-01
-1.10678323e-01 -2.50178706e-02 -3.45014371e-02 6.60603881e-01
-1.57112684e-02 3.14801902e-01 -1.56836480e-01 -2.25355312e-01
1.43200707e+00 -4.10654813e-01 8.27600479e-01 -5.02932072e-01
5.14064312e-01 -3.01719695e-01 7.48824298e-01 7.52968729e-01
-4.30812478e-01 -3.03294044e-02 -3.37555140e-01 -3.68296176e-01
-5.99734902e-01 -8.89265835e-01 -1.66729707e-02 4.91042078e-01
3.83704334e-01 -1.81549251e-01 -3.53758872e-01 8.90498385e-02
-1.62557021e-01 7.61313558e-01 3.31929415e-01 1.62378579e-01
-5.83193488e-02 -7.57389724e-01 5.13339639e-01 2.57456958e-01
1.04374409e+00 -4.75787848e-01 -9.44065392e-01 5.04962444e-01
3.85918021e-01 -1.37253499e+00 1.07698880e-01 1.71334788e-01
-4.32630956e-01 -3.91918838e-01 -3.55518907e-01 -4.13103521e-01
1.52644366e-01 5.58394670e-01 7.50042796e-01 -2.59613603e-01
-2.08805516e-01 5.66134214e-01 -2.55176783e-01 -7.26274133e-01
-1.38152195e-02 -3.44073586e-02 5.29087543e-01 2.04981223e-01
3.43058258e-01 -1.40530491e+00 -8.40341568e-01 5.48399985e-01
-3.08899045e-01 -1.28094941e-01 8.71198714e-01 1.26499951e-01
1.99062169e-01 5.09084165e-01 1.06132996e+00 -1.69066563e-01
4.54379588e-01 -4.69989389e-01 -9.42734003e-01 1.47930503e-01
-7.93598652e-01 -4.77345556e-01 8.42624903e-01 -7.62721011e-03
-5.87420166e-01 -7.58367591e-03 -6.38233125e-02 1.59430560e-02
8.94703791e-02 5.55631697e-01 -4.07029003e-01 -5.88294327e-01
8.91329527e-01 2.37012818e-01 -1.50478393e-01 8.17241967e-02
7.57052526e-02 1.43615365e+00 3.59891146e-01 1.09265141e-01
9.58980381e-01 5.15509903e-01 5.29639840e-01 -1.22777498e+00
-2.53796965e-01 -3.09859872e-01 5.02394140e-03 -6.33356035e-01
5.19732773e-01 -1.08544946e+00 -9.42983687e-01 1.43951297e-01
-8.64476264e-01 4.00396399e-02 1.55595452e-01 8.29716980e-01
-3.03383708e-01 -2.54158936e-02 1.29831303e-02 -9.67381477e-01
-6.74225509e-01 -8.91638100e-01 3.11403573e-01 5.53993523e-01
-2.44488701e-01 -6.81902587e-01 -3.73985171e-01 2.23604307e-01
1.39361191e+00 1.07630759e-01 4.84258413e-01 -3.50292742e-01
-9.97639716e-01 -3.79490167e-01 -1.10210560e-01 -1.46629319e-01
2.84567028e-01 -7.18746424e-01 -7.98239827e-01 -9.45292339e-02
2.20785499e-01 4.74999666e-01 -7.21038319e-03 1.04114091e+00
9.31948125e-01 -1.13141919e-02 -6.62962317e-01 6.29819870e-01
1.79691684e+00 6.11136019e-01 6.67821884e-01 9.16968212e-02
9.20685530e-02 -4.87355888e-01 2.87665606e-01 7.73525476e-01
3.02054789e-02 6.07391894e-01 5.73081553e-01 1.07923411e-01
-2.82263398e-01 4.37084556e-01 -7.52336681e-02 5.83741188e-01
1.06164664e-01 -6.96419179e-01 -6.33271515e-01 7.61939511e-02
-1.49900258e+00 -9.57357943e-01 -2.70130001e-02 1.98590302e+00
1.27454355e-01 2.98689723e-01 1.26819402e-01 4.73021090e-01
6.92250311e-01 2.92814784e-02 -4.96411502e-01 -2.00116679e-01
-1.43905565e-01 2.28861615e-01 9.96318877e-01 3.65756452e-01
-6.22755706e-01 2.06960186e-01 5.04651356e+00 9.88162577e-01
-1.44367301e+00 6.56728566e-01 3.42987329e-01 -9.35723335e-02
-1.16077200e-01 -3.39091241e-01 -3.87787610e-01 5.79322338e-01
1.28346753e+00 -5.96870221e-02 4.75990802e-01 9.13747191e-01
5.13888240e-01 -1.91870764e-01 -5.60742617e-01 1.51463580e+00
-2.37997219e-01 -1.63606691e+00 -7.04502046e-01 1.14490546e-01
5.93383372e-01 1.31790385e-01 -1.26459628e-01 -1.13597490e-01
-1.73434854e-01 -6.63111925e-01 5.20411849e-01 3.93585026e-01
1.01126242e+00 -6.65712893e-01 9.82344389e-01 4.24073786e-01
-1.38146663e+00 -5.88216603e-01 -7.00218603e-02 -4.19192106e-01
2.44252667e-01 1.42011285e+00 -7.89997101e-01 6.50416315e-01
4.14361715e-01 5.75534515e-02 1.13151737e-01 1.09573984e+00
1.03226170e-01 6.67746902e-01 -8.21742237e-01 -6.89559162e-01
-1.33278808e-02 -1.80918440e-01 7.36040592e-01 6.76198125e-01
1.06269324e+00 2.35289156e-01 -2.61870623e-02 5.01782358e-01
6.57611862e-02 -1.96994901e-01 -4.38202709e-01 2.79220730e-01
1.08233440e+00 1.35887003e+00 -9.09507036e-01 8.60272720e-02
-6.49919391e-01 4.63989884e-01 -7.51332402e-01 2.07325071e-01
-6.45738542e-01 -4.62848276e-01 5.17102420e-01 2.67793179e-01
2.73258626e-01 -8.11476886e-01 -7.79388726e-01 -2.77004510e-01
-8.72583911e-02 -5.09868562e-01 2.05789804e-02 -8.02436173e-01
-5.94342649e-01 5.57914138e-01 -6.18219316e-01 -1.63325095e+00
-3.21778268e-01 -2.94520348e-01 -3.98474753e-01 6.49466157e-01
-1.09960413e+00 -8.92014861e-01 -5.59846759e-01 4.88402903e-01
2.28191942e-01 -7.14674771e-01 9.59762335e-01 6.23693645e-01
-1.77308351e-01 4.60195303e-01 3.26330692e-01 -1.56009242e-01
-3.17942463e-02 -7.69764423e-01 -1.84890538e-01 1.09916949e+00
2.95620039e-02 1.58391356e-01 7.63696373e-01 -5.73180079e-01
-1.73015106e+00 -7.38314450e-01 4.09222811e-01 5.32760262e-01
5.31447649e-01 -3.28666270e-01 3.01590323e-01 1.82712197e-01
3.19162548e-01 4.27917778e-01 8.53334546e-01 -1.26752466e-01
3.82497728e-01 -8.57555866e-01 -1.34955442e+00 7.26335883e-01
1.11932969e+00 -2.31855080e-01 4.71270353e-01 2.77906209e-01
1.93990469e-01 -9.74767953e-02 -5.66578925e-01 2.27768108e-01
5.47358513e-01 -1.17832804e+00 8.35356176e-01 3.43191624e-01
-7.19309628e-01 -6.81419492e-01 -6.73313320e-01 -1.19412947e+00
-3.99004906e-01 -1.04393649e+00 -2.78491616e-01 7.94222713e-01
4.04775202e-01 -9.21653569e-01 1.01426387e+00 -1.25376448e-01
-1.49826258e-01 -5.23173332e-01 -1.57514143e+00 -9.68066871e-01
-8.86939466e-01 -7.03377545e-01 7.00832546e-01 5.30688345e-01
1.92730159e-01 5.12325108e-01 -2.80437648e-01 7.55224466e-01
6.60155416e-01 -6.40906468e-02 3.34128827e-01 -1.51563716e+00
-4.59261090e-01 -2.33772039e-01 -9.91056323e-01 -9.64046061e-01
-6.15257919e-01 -4.89795685e-01 -5.01471937e-01 -1.45790350e+00
-5.64177632e-01 -7.30770528e-01 -2.39024937e-01 1.45888701e-01
6.50644183e-01 3.01966310e-01 3.21367010e-02 -1.54260978e-01
-7.13313222e-01 1.04061544e-01 6.28845930e-01 7.33907744e-02
-2.41535470e-01 5.53508997e-01 -4.22823012e-01 5.18999338e-01
1.04449654e+00 -2.74166375e-01 -5.73084950e-01 -1.07321098e-01
3.56171995e-01 7.04181790e-01 3.85326184e-02 -2.06112003e+00
8.08054507e-01 2.41068915e-01 4.84825790e-01 -5.22975326e-01
4.60616648e-01 -1.55239856e+00 6.15577340e-01 7.51952767e-01
6.10505760e-01 -1.13549873e-01 -1.26384556e-01 3.89561623e-01
1.86895430e-01 1.02624319e-01 5.95026433e-01 1.99491397e-01
-6.58588350e-01 1.27042785e-01 -7.37564862e-01 -5.18992782e-01
1.23642027e+00 -6.69749022e-01 -2.90391982e-01 -8.59800220e-01
-4.61369544e-01 -8.83597583e-02 -1.59855828e-01 -1.26775980e-01
3.83909911e-01 -1.27782059e+00 -3.75554979e-01 2.47324586e-01
-3.77606839e-01 -4.08612281e-01 4.97186482e-01 8.81736338e-01
-5.49391389e-01 8.19690108e-01 -2.18706086e-01 -6.43698812e-01
-1.01421440e+00 1.07987784e-01 6.72670901e-01 1.51639134e-01
-1.75668254e-01 5.02393544e-01 -8.24241638e-01 1.32228643e-01
9.54902098e-02 -1.17876060e-01 1.40700102e-01 -3.01090598e-01
6.44789696e-01 6.52873635e-01 3.56460392e-01 5.59688807e-02
-4.79656607e-01 6.07342303e-01 6.62258804e-01 -1.19952954e-01
1.11032331e+00 -4.67721164e-01 1.40742838e-01 -1.34932652e-01
7.37153172e-01 4.11715388e-01 -4.47920054e-01 -2.54478842e-01
-1.94688797e-01 -2.61949033e-01 6.85019195e-01 -1.01205945e+00
-1.06100619e+00 2.83127993e-01 1.29215205e+00 7.25544453e-01
1.49327576e+00 -2.23302156e-01 7.80870497e-01 4.13205832e-01
1.08048904e+00 -1.00891793e+00 -2.65831828e-01 1.61081791e-01
2.80077994e-01 -6.40903592e-01 1.56168357e-01 -4.20251459e-01
1.40679568e-01 1.19547105e+00 6.50770739e-02 4.46143538e-01
1.19692373e+00 6.74188852e-01 2.54302293e-01 -9.53296572e-03
-2.65248924e-01 -3.30638587e-01 -2.42735028e-01 9.84706700e-01
-7.70673156e-02 3.14662457e-01 -2.41893128e-01 8.21793377e-01
-3.75659913e-01 2.61153787e-01 6.30380452e-01 7.64011919e-01
-8.10336709e-01 -8.78830791e-01 -5.83079100e-01 6.29757583e-01
-3.35557312e-01 2.88415272e-02 2.34288231e-01 4.89259511e-01
3.02125871e-01 1.36728525e+00 2.01147392e-01 -6.97080970e-01
-2.56232899e-02 -3.29257429e-01 1.25677884e-01 -9.85079184e-02
4.06176038e-02 4.76887003e-02 2.53950506e-01 -4.68632579e-01
-2.23746359e-01 -4.52207893e-01 -1.23135364e+00 -3.56628567e-01
-3.01230609e-01 6.87323464e-03 1.11825049e+00 1.10393572e+00
8.29232037e-01 8.87210608e-01 9.31383908e-01 -4.00147974e-01
-2.39837512e-01 -9.40239727e-01 -7.84673929e-01 -7.76976168e-01
1.04386270e-01 -5.22601604e-01 -1.10507995e-01 -6.64484978e-01]
|
[6.333780765533447, 1.2005078792572021]
|
a9de4016-1d8f-4a58-a576-29a50252b2cb
|
segment-anything-model-sam-for-digital
|
2304.04155
| null |
https://arxiv.org/abs/2304.04155v1
|
https://arxiv.org/pdf/2304.04155v1.pdf
|
Segment Anything Model (SAM) for Digital Pathology: Assess Zero-shot Segmentation on Whole Slide Imaging
|
The segment anything model (SAM) was released as a foundation model for image segmentation. The promptable segmentation model was trained by over 1 billion masks on 11M licensed and privacy-respecting images. The model supports zero-shot image segmentation with various segmentation prompts (e.g., points, boxes, masks). It makes the SAM attractive for medical image analysis, especially for digital pathology where the training data are rare. In this study, we evaluate the zero-shot segmentation performance of SAM model on representative segmentation tasks on whole slide imaging (WSI), including (1) tumor segmentation, (2) non-tumor tissue segmentation, (3) cell nuclei segmentation. Core Results: The results suggest that the zero-shot SAM model achieves remarkable segmentation performance for large connected objects. However, it does not consistently achieve satisfying performance for dense instance object segmentation, even with 20 prompts (clicks/boxes) on each image. We also summarized the identified limitations for digital pathology: (1) image resolution, (2) multiple scales, (3) prompt selection, and (4) model fine-tuning. In the future, the few-shot fine-tuning with images from downstream pathological segmentation tasks might help the model to achieve better performance in dense object segmentation.
|
['Yuankai Huo', 'Yucheng Tang', 'Haichun Yang', 'Agnes B. Fogo', 'Shilin Zhao', 'Yaohong Wang', 'Keith T. Wilson', 'Lori A. Coburn', 'Lee E. Wheless', 'Bennett A. Landman', 'Shunxing Bao', 'Lucas W. Remedios', 'Tianyuan Yao', 'Quan Liu', 'Can Cui', 'Ruining Deng']
|
2023-04-09
| null | null | null | null |
['zero-shot-segmentation', 'tumor-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 5.16285121e-01 4.66771454e-01 -4.72455502e-01 -4.77204949e-01
-1.46532822e+00 -5.64907432e-01 2.85920501e-03 2.87259698e-01
-5.30932367e-01 3.38640958e-01 2.05578711e-02 -7.03247011e-01
1.78896058e-02 -4.01885301e-01 -3.68537605e-01 -8.25707972e-01
2.62282729e-01 5.23171842e-01 5.97591817e-01 2.84523696e-01
4.31797504e-02 4.91502553e-01 -9.10787284e-01 4.01548743e-01
8.09995770e-01 7.59031951e-01 3.99705321e-01 9.31811988e-01
-3.48738760e-01 4.35033739e-01 -5.79854965e-01 -3.36427122e-01
2.31969282e-01 -1.49097770e-01 -1.05392838e+00 4.07371491e-01
4.85918343e-01 -6.70536101e-01 -4.08566445e-02 1.09809375e+00
5.86757958e-01 -1.50761917e-01 4.69226301e-01 -1.05152142e+00
-4.69145387e-01 5.69172800e-01 -8.27663004e-01 5.64737201e-01
-1.02767237e-01 7.52602100e-01 7.56094933e-01 -5.34802854e-01
9.98010814e-01 8.02880883e-01 6.54862046e-01 8.17651629e-01
-1.28186667e+00 -6.39044583e-01 -8.20859224e-02 -1.04711764e-01
-1.39460218e+00 -3.17816287e-01 -9.62366089e-02 -4.61937726e-01
8.21623325e-01 8.04399133e-01 5.51334739e-01 7.66449869e-01
2.74247169e-01 8.97615850e-01 1.10326695e+00 -5.39692864e-02
3.63807201e-01 2.88064092e-01 6.90285802e-01 4.95484918e-01
2.14092568e-01 -4.89408135e-01 -2.96133608e-01 -3.87349695e-01
7.79242814e-01 5.29625639e-02 -1.90652609e-01 1.91304103e-01
-1.24928558e+00 6.15795434e-01 3.39212328e-01 4.58566338e-01
1.89189538e-02 -4.13010605e-02 4.10944939e-01 -3.67031172e-02
5.31148911e-01 3.98855716e-01 -1.54583111e-01 -3.03834882e-02
-1.48349440e+00 -1.32730693e-01 5.24081945e-01 1.06072450e+00
4.86969292e-01 -4.84127313e-01 -8.10551345e-01 7.03111708e-01
1.40109703e-01 6.41880482e-02 6.24995768e-01 -7.13552356e-01
9.50513408e-02 6.27789259e-01 -1.14609338e-01 -5.40711045e-01
-8.80297244e-01 -2.30355367e-01 -7.60256231e-01 -1.11897498e-01
6.47590935e-01 -1.48652181e-01 -1.59718013e+00 1.29290080e+00
3.33205909e-01 2.37142473e-01 -4.67175037e-01 9.24436986e-01
1.33965719e+00 2.39068836e-01 4.27752137e-01 -1.76473349e-01
1.89134443e+00 -8.48982394e-01 -9.87769961e-01 -1.88782126e-01
9.02239978e-01 -6.88138485e-01 1.27586722e+00 -3.09312306e-02
-1.03346634e+00 -1.82355821e-01 -6.86857402e-01 -3.38729799e-01
-5.56950867e-01 9.13862213e-02 6.02697015e-01 7.79473305e-01
-1.32026327e+00 1.54889092e-01 -1.07273674e+00 -7.38056839e-01
9.88391459e-01 5.58630049e-01 -2.60233343e-01 3.03279255e-02
-6.96221352e-01 4.01172370e-01 8.40181038e-02 -2.60370106e-01
-7.17286885e-01 -1.09770608e+00 -6.16897225e-01 -1.62046496e-02
3.96944761e-01 -4.90403384e-01 1.33967233e+00 -5.35071492e-01
-1.04555595e+00 1.58536685e+00 -2.54856139e-01 -4.50359404e-01
6.63603663e-01 3.40776920e-01 -4.87316512e-02 5.94256699e-01
3.90554130e-01 1.21445847e+00 4.80647147e-01 -9.22546387e-01
-6.40905380e-01 -4.08523142e-01 -3.25186878e-01 7.68349245e-02
-2.08395675e-01 2.27172941e-01 -7.50450015e-01 -4.73184407e-01
1.61611483e-01 -8.76997054e-01 -6.83660984e-01 3.79426032e-01
-9.12193060e-01 -3.16330977e-02 8.54081035e-01 -7.15636969e-01
1.15603983e+00 -2.51030302e+00 -6.44759297e-01 1.14859186e-01
3.71345341e-01 2.89558232e-01 1.77838039e-02 -1.64390326e-01
1.18217871e-01 6.84571743e-01 -1.95865169e-01 -4.21267927e-01
-2.16725454e-01 1.84603427e-02 1.62247866e-01 5.43706834e-01
-1.15879484e-01 1.37184298e+00 -7.33022153e-01 -1.15667951e+00
1.01424322e-01 -8.44178274e-02 -4.45610046e-01 3.77456546e-02
-1.11916721e-01 5.72300792e-01 -3.87595654e-01 1.05306089e+00
8.02349389e-01 -9.54631507e-01 2.21598335e-02 -1.88992843e-01
4.71582189e-02 -2.38033429e-01 -9.11602437e-01 1.71706164e+00
3.66844982e-01 5.43750942e-01 6.03670955e-01 -2.54159182e-01
3.95697147e-01 5.70553005e-01 8.23290229e-01 -1.64690509e-01
2.07625628e-01 1.07937828e-01 9.64026600e-02 -8.22805643e-01
3.94741535e-01 -4.57908772e-02 -1.49352476e-02 4.13416475e-01
-1.58403143e-02 -2.29915306e-01 3.04295331e-01 5.09689450e-01
1.32092345e+00 -7.47194350e-01 2.10362718e-01 -4.33773667e-01
9.09343064e-02 4.03166324e-01 6.13995254e-01 9.47924018e-01
-8.35624576e-01 1.01776612e+00 6.88963532e-01 -1.07917145e-01
-6.72879636e-01 -9.14270699e-01 -5.00981450e-01 1.03690541e+00
3.60501468e-01 -2.34900385e-01 -1.02807808e+00 -7.32305229e-01
-1.33236244e-01 4.52944428e-01 -7.03873456e-01 2.66432136e-01
-2.74262965e-01 -9.57833171e-01 7.69461751e-01 3.98653954e-01
4.36886698e-01 -8.01850259e-01 -7.43732989e-01 6.15536086e-02
-2.55564064e-01 -1.12434947e+00 -7.68702090e-01 1.45565689e-01
-8.61275256e-01 -1.09603500e+00 -1.02923870e+00 -9.13059056e-01
1.15548182e+00 4.12656575e-01 7.00686872e-01 2.13998094e-01
-9.64538276e-01 5.90031624e-01 -8.42049122e-02 -4.43725169e-01
-2.07478240e-01 1.35849491e-01 -5.17873108e-01 -1.07349135e-01
4.58505690e-01 1.38805836e-01 -9.96186852e-01 5.22946656e-01
-1.13961303e+00 3.03141177e-01 5.93619287e-01 8.29860866e-01
1.11822653e+00 -1.69721335e-01 3.19373488e-01 -1.46335745e+00
4.23864394e-01 -4.23544884e-01 -3.80145013e-01 3.54221433e-01
-3.95734519e-01 -9.16851759e-01 6.79154415e-03 -3.12235534e-01
-1.09261167e+00 1.03051022e-01 -2.44251370e-01 -2.43718714e-01
-4.31549013e-01 2.13221088e-01 2.99572051e-01 -2.53817886e-01
6.35835290e-01 9.19054088e-04 3.84375185e-01 -2.89104134e-01
1.11668825e-01 8.60263765e-01 4.89255756e-01 1.07199298e-02
3.66685629e-01 7.82546699e-01 -2.55293876e-01 -7.63727784e-01
-6.70434117e-01 -9.36947048e-01 -4.10048604e-01 -1.05288446e-01
1.23531234e+00 -7.29671776e-01 -6.18120074e-01 4.01555151e-01
-6.88801229e-01 -6.61405325e-01 -4.49731469e-01 2.34696820e-01
-4.19824570e-01 2.34995246e-01 -1.30157733e+00 -5.01967192e-01
-6.93046927e-01 -1.55292726e+00 1.32800746e+00 6.87518597e-01
-5.04700899e-01 -6.99064791e-01 -3.25962692e-01 7.11301923e-01
4.74789411e-01 2.10733995e-01 9.40051377e-01 -9.27496254e-01
-7.67498553e-01 -4.04065043e-01 -4.65098858e-01 -2.01796129e-01
1.75266042e-01 2.59359211e-01 -9.38943267e-01 -3.39488804e-01
-6.31018877e-02 -1.87621295e-01 7.25826442e-01 9.54426467e-01
1.74460018e+00 -8.29763412e-02 -8.94135118e-01 7.66577363e-01
1.13607883e+00 2.14009702e-01 6.82936132e-01 1.08886838e-01
3.60682160e-01 3.64433706e-01 9.48013544e-01 2.12178066e-01
4.69204746e-02 1.10974222e-01 2.91628063e-01 -7.59278119e-01
-2.14581504e-01 1.44344196e-01 -3.94710124e-01 2.76331544e-01
3.80036861e-01 -6.68583289e-02 -1.17862010e+00 7.86644220e-01
-1.56418419e+00 -4.64507699e-01 -2.42696449e-01 1.71564579e+00
8.79667699e-01 8.87649730e-02 8.68559852e-02 -4.32223976e-01
8.34270060e-01 2.54388247e-02 -8.33704889e-01 -8.70763361e-02
6.08123876e-02 2.92420238e-02 7.74791777e-01 1.79137662e-01
-1.07737350e+00 9.79193747e-01 7.46706486e+00 1.21951950e+00
-1.14740264e+00 4.43728060e-01 1.39988339e+00 -3.55300963e-01
-1.32043585e-01 -1.26385316e-01 -1.04360640e+00 5.48224449e-01
5.95069826e-01 -2.05420047e-01 -1.47139296e-01 7.78963387e-01
1.53042465e-01 -3.81164491e-01 -8.44001591e-01 8.87295842e-01
-2.82246292e-01 -1.86572206e+00 -1.79399341e-01 3.82366866e-01
7.12013662e-01 1.05027519e-01 2.67047286e-01 1.22136340e-01
1.35572881e-01 -1.16763675e+00 1.58996955e-01 3.31954837e-01
1.37982738e+00 -1.89442948e-01 8.80341887e-01 3.15681547e-01
-6.19480669e-01 1.43822491e-01 -3.07787150e-01 7.27315009e-01
1.79268703e-01 5.39267600e-01 -1.31911802e+00 1.90837115e-01
6.65772438e-01 3.51161480e-01 -8.02354097e-01 1.30057824e+00
2.66767889e-01 8.44127357e-01 -2.69737959e-01 4.30541020e-03
2.74963260e-01 7.56871328e-02 5.18781662e-01 1.48472726e+00
1.12576090e-01 5.77231705e-01 9.92004573e-02 7.42694199e-01
-3.28610539e-02 1.81065291e-01 -1.31475955e-01 -1.79987594e-01
4.16994035e-01 1.73500443e+00 -1.54333544e+00 -4.44253743e-01
-3.43995690e-01 5.87055862e-01 -3.09346095e-02 4.06858057e-01
-7.01577783e-01 -9.89653394e-02 4.79470462e-01 3.81745160e-01
-6.95654005e-02 1.44484207e-01 -7.66430438e-01 -5.59127212e-01
-5.95916688e-01 -7.48312593e-01 8.51829112e-01 -5.43302655e-01
-1.22322202e+00 2.38834620e-01 -1.82237715e-01 -1.05887079e+00
2.01932371e-01 -5.54494321e-01 -7.76219070e-01 5.35636425e-01
-1.16777110e+00 -1.13987982e+00 -3.95628065e-01 4.82141525e-01
5.49383521e-01 1.78781301e-01 8.21380258e-01 2.55622417e-01
-8.91228914e-01 9.14013982e-01 -1.12396225e-01 3.93001914e-01
8.58779728e-01 -1.29793024e+00 1.50090441e-01 5.53724289e-01
-3.79560173e-01 7.36139357e-01 4.64976430e-01 -7.36789405e-01
-1.12943351e+00 -1.06513834e+00 2.70651639e-01 -3.05925995e-01
4.08725441e-01 -1.49933636e-01 -8.91081452e-01 9.60730553e-01
1.12277493e-01 3.33353788e-01 1.47912157e+00 -1.30211815e-01
3.79861653e-01 1.09113585e-02 -1.81528807e+00 7.47117639e-01
5.64412057e-01 -1.23060890e-01 6.04299977e-02 7.23942816e-01
7.97025681e-01 -1.01861644e+00 -1.03255558e+00 2.22178116e-01
2.41652012e-01 -7.29344308e-01 6.48370028e-01 -4.66776878e-01
1.92813486e-01 -2.47453582e-02 3.30764949e-01 -6.24071240e-01
-5.08163691e-01 -6.83537424e-01 2.17527688e-01 1.13622081e+00
6.62930310e-01 -6.89705133e-01 1.17068040e+00 1.39092946e+00
-3.24298382e-01 -1.08564126e+00 -1.25557303e+00 -3.44058573e-01
-1.15952261e-01 -1.79041430e-01 6.29795909e-01 7.78577805e-01
1.65710568e-01 -2.82804370e-01 2.05881208e-01 -5.76582178e-02
4.12047058e-01 -9.63907912e-02 6.21480286e-01 -6.24719381e-01
-2.10833654e-01 -5.29609323e-01 -3.08794767e-01 -7.66383767e-01
-4.79955643e-01 -9.49220836e-01 -3.46478224e-02 -1.65887809e+00
5.79951525e-01 -5.31795561e-01 -2.75743723e-01 7.42820203e-01
-4.37924296e-01 4.17706698e-01 8.70994106e-02 3.63443732e-01
-7.11578846e-01 -3.29545230e-01 1.64406073e+00 -2.44321972e-01
4.53944737e-03 1.96690083e-01 -9.95398164e-01 6.66341841e-01
8.03450048e-01 -4.17025715e-01 -2.03636587e-01 -1.69984028e-01
-3.09703529e-01 9.15109366e-02 2.90875882e-01 -7.14856088e-01
6.86433196e-01 -2.76092350e-01 3.95687342e-01 -9.91074562e-01
2.90419132e-01 -5.40070772e-01 1.13368094e-01 6.13475442e-01
-3.68304610e-01 -7.31516182e-01 4.10932899e-01 4.57236022e-01
-4.70744539e-03 -3.25732946e-01 9.85399246e-01 -4.67876285e-01
-6.18200183e-01 5.14950216e-01 -5.97470582e-01 -6.35224581e-02
1.40927017e+00 -7.53401101e-01 -7.34447896e-01 -2.42704735e-03
-1.17508864e+00 6.82778955e-01 5.53082883e-01 -9.31774303e-02
4.06179100e-01 -7.98748791e-01 -5.99728763e-01 7.23272040e-02
1.12500601e-01 4.14103836e-01 7.36487865e-01 1.31712389e+00
-7.07467556e-01 4.60266531e-01 9.53879580e-02 -9.89196897e-01
-1.60612464e+00 3.70880008e-01 3.16433966e-01 -5.23651481e-01
-8.29774737e-01 1.32432342e+00 5.55163682e-01 -1.62655786e-01
2.68224567e-01 -4.71328795e-01 1.28212214e-01 7.57510832e-04
6.93006992e-01 3.96290123e-01 2.83551868e-02 -2.25674540e-01
-4.50953335e-01 2.01651469e-01 -7.37461984e-01 5.06411009e-02
9.00143504e-01 -1.45732492e-01 -1.05395466e-01 3.15616876e-01
9.12447572e-01 -2.82370359e-01 -1.18230438e+00 -4.13877033e-02
-6.94950372e-02 -5.01874566e-01 2.66837515e-03 -8.55943620e-01
-1.05683947e+00 5.50842285e-01 6.24518573e-01 2.77636170e-01
9.62754190e-01 2.19277948e-01 9.59730864e-01 -3.86551976e-01
2.59184629e-01 -1.26828170e+00 -1.36740878e-01 8.93979613e-03
3.61300319e-01 -1.32302010e+00 -8.09992256e-04 -6.88821197e-01
-7.30427027e-01 9.20321167e-01 7.30367124e-01 4.51945305e-01
6.73028708e-01 6.94077194e-01 3.47196341e-01 -5.27033389e-01
-6.32667303e-01 -1.73650160e-01 1.42374009e-01 4.72819775e-01
4.34572846e-01 3.76981735e-01 -4.12143648e-01 8.21364105e-01
-8.48314376e-04 1.05372094e-01 4.06207204e-01 9.61165845e-01
-5.01478314e-01 -5.77319682e-01 -4.73209620e-01 1.24483967e+00
-7.17904806e-01 1.18133046e-01 -4.02332962e-01 7.21854389e-01
1.02855228e-01 9.48947608e-01 2.29357228e-01 1.20534532e-01
6.54506758e-02 -2.16250435e-01 -6.19706810e-02 -1.05561876e+00
-8.89891148e-01 2.90704966e-01 -2.39454031e-01 -6.10931933e-01
2.21996143e-01 -6.64915800e-01 -1.50552106e+00 -1.49020836e-01
-3.11192542e-01 -2.32731581e-01 5.61021209e-01 6.04377389e-01
5.32841146e-01 6.68110669e-01 9.32610258e-02 -3.68900120e-01
-3.06491077e-01 -9.16308522e-01 -9.45720851e-01 2.87588328e-01
9.20891315e-02 -1.27215430e-01 -2.62323439e-01 1.24962986e-01]
|
[14.740694046020508, -2.2630302906036377]
|
36a0a1cd-ec43-40b2-b6c6-61071979456a
|
robust-deep-auc-maximization-a-new-surrogate
|
2012.03173
| null |
https://arxiv.org/abs/2012.03173v2
|
https://arxiv.org/pdf/2012.03173v2.pdf
|
Large-scale Robust Deep AUC Maximization: A New Surrogate Loss and Empirical Studies on Medical Image Classification
|
Deep AUC Maximization (DAM) is a new paradigm for learning a deep neural network by maximizing the AUC score of the model on a dataset. Most previous works of AUC maximization focus on the perspective of optimization by designing efficient stochastic algorithms, and studies on generalization performance of large-scale DAM on difficult tasks are missing. In this work, we aim to make DAM more practical for interesting real-world applications (e.g., medical image classification). First, we propose a new margin-based min-max surrogate loss function for the AUC score (named as AUC min-max-margin loss or simply AUC margin loss for short). It is more robust than the commonly used AUC square loss, while enjoying the same advantage in terms of large-scale stochastic optimization. Second, we conduct extensive empirical studies of our DAM method on four difficult medical image classification tasks, namely (i) classification of chest x-ray images for identifying many threatening diseases, (ii) classification of images of skin lesions for identifying melanoma, (iii) classification of mammogram for breast cancer screening, and (iv) classification of microscopic images for identifying tumor tissue. Our studies demonstrate that the proposed DAM method improves the performance of optimizing cross-entropy loss by a large margin, and also achieves better performance than optimizing the existing AUC square loss on these medical image classification tasks. Specifically, our DAM method has achieved the 1st place on Stanford CheXpert competition on Aug. 31, 2020. To the best of our knowledge, this is the first work that makes DAM succeed on large-scale medical image datasets. We also conduct extensive ablation studies to demonstrate the advantages of the new AUC margin loss over the AUC square loss on benchmark datasets. The proposed method is implemented in our open-sourced library LibAUC (www.libauc.org).
|
['Tianbao Yang', 'Milan Sonka', 'Yan Yan', 'Zhuoning Yuan']
|
2020-12-06
| null |
http://openaccess.thecvf.com//content/ICCV2021/html/Yuan_Large-Scale_Robust_Deep_AUC_Maximization_A_New_Surrogate_Loss_and_ICCV_2021_paper.html
|
http://openaccess.thecvf.com//content/ICCV2021/papers/Yuan_Large-Scale_Robust_Deep_AUC_Maximization_A_New_Surrogate_Loss_and_ICCV_2021_paper.pdf
|
iccv-2021-1
|
['graph-property-prediction']
|
['graphs']
|
[ 3.17987263e-01 1.52193204e-01 -3.22334528e-01 -6.51493251e-01
-1.23623145e+00 -1.53304756e-01 8.31387118e-02 4.46973711e-01
-7.97465742e-01 7.43516982e-01 -1.79974154e-01 -5.40154994e-01
-1.36254802e-01 -4.78116840e-01 -6.93992913e-01 -7.55929232e-01
-3.41337562e-01 2.07955137e-01 7.89221302e-02 2.31638879e-01
-9.92819965e-02 3.63727689e-01 -9.68012869e-01 1.34913549e-01
1.00974905e+00 1.55152833e+00 8.56275186e-02 4.32837993e-01
4.16019142e-01 7.27675676e-01 -3.50263476e-01 -7.50148892e-01
2.11019218e-01 -3.73181731e-01 -8.26142371e-01 -6.18690699e-02
3.31380278e-01 -3.20496857e-01 -1.08850524e-01 1.23162580e+00
6.55219376e-01 -9.77030620e-02 8.44170272e-01 -1.03612280e+00
-2.37872154e-01 5.77208638e-01 -7.43789375e-01 2.31214240e-01
-2.09603786e-01 9.27126706e-02 1.17339420e+00 -6.43157125e-01
2.52113819e-01 7.52887189e-01 9.38200235e-01 5.90401173e-01
-9.42908466e-01 -5.08910000e-01 -1.47996455e-01 -5.23761548e-02
-1.46281898e+00 -1.68777406e-01 4.56172049e-01 -4.06842887e-01
5.46132982e-01 4.41597074e-01 2.36832365e-01 9.30976987e-01
5.30919194e-01 1.10213065e+00 9.97876346e-01 -1.53390303e-01
1.33729160e-01 3.50081861e-01 2.49471486e-01 1.05687296e+00
2.15468779e-01 7.66441599e-02 9.93096363e-03 -2.69075722e-01
5.02844214e-01 -3.69376466e-02 -2.07001150e-01 -2.67510831e-01
-9.97693062e-01 8.32839251e-01 5.89300156e-01 3.12583297e-01
-3.93773675e-01 1.59117430e-01 3.88925761e-01 4.49250005e-02
6.35091543e-01 5.26654840e-01 -3.99427533e-01 1.03414260e-01
-9.94050980e-01 1.03474006e-01 7.44598866e-01 4.13365692e-01
2.69605279e-01 -2.63781756e-01 -4.83383387e-01 1.08784068e+00
3.13920557e-01 3.40854645e-01 6.72600627e-01 -8.20347250e-01
4.84400839e-01 4.08116937e-01 -2.57188171e-01 -6.65918469e-01
-7.74278283e-01 -7.59112239e-01 -1.18362951e+00 2.10371893e-03
4.49461818e-01 -4.36932623e-01 -7.41158366e-01 1.98085213e+00
1.43090218e-01 -1.52730510e-01 -8.96231621e-04 7.70255208e-01
7.69343257e-01 3.17609906e-01 1.71622306e-01 -2.93656498e-01
1.45142233e+00 -7.21482515e-01 -4.66628194e-01 -9.45732296e-02
8.15179288e-01 -4.43371117e-01 9.69597518e-01 3.86074781e-01
-1.04683316e+00 -2.23195598e-01 -1.13777018e+00 1.74673349e-01
-2.45600939e-02 4.32427704e-01 8.65992785e-01 7.81076670e-01
-8.06337833e-01 5.90825081e-01 -1.04414296e+00 -2.08823726e-01
8.48465741e-01 4.14240569e-01 -1.95643604e-01 8.63294229e-02
-1.27945447e+00 7.23855615e-01 2.11604252e-01 -2.59944219e-02
-1.01805747e+00 -9.47170675e-01 -9.12294865e-01 6.19240254e-02
3.05563956e-01 -7.19005227e-01 1.38881397e+00 -1.12173665e+00
-1.25421262e+00 1.10341454e+00 8.86360407e-02 -9.04710710e-01
8.06284845e-01 -1.81655422e-01 -1.32940024e-01 -4.27858308e-02
1.60823762e-02 5.80649793e-01 6.04821205e-01 -7.61930704e-01
-3.95297825e-01 -5.01071453e-01 -8.48218054e-03 5.30111231e-02
-8.95612419e-01 -2.01996312e-01 -3.40088069e-01 -6.85519874e-01
-2.29195759e-01 -7.50305235e-01 -6.76967084e-01 7.57532343e-02
-6.53234243e-01 -2.12044254e-01 3.01649213e-01 -6.77730262e-01
1.47343171e+00 -2.17031217e+00 -2.98877358e-01 1.15178213e-01
3.24553043e-01 4.07488108e-01 2.30968161e-03 -3.66866216e-02
-5.50492555e-02 1.99355349e-01 -7.11252630e-01 -4.36748683e-01
-2.17505068e-01 -2.60011166e-01 1.90808713e-01 5.67514539e-01
2.77234554e-01 1.00007522e+00 -7.26902425e-01 -7.72397399e-01
-2.53651470e-01 4.27136540e-01 -5.91955185e-01 -9.80904442e-04
-6.91774637e-02 1.30959049e-01 -4.67207253e-01 6.50806487e-01
6.14840984e-01 -7.84943342e-01 5.34675689e-03 -2.85654694e-01
3.74781042e-01 -1.84899926e-01 -7.66764462e-01 1.47573781e+00
-5.20364225e-01 5.40875137e-01 -1.98688656e-01 -1.06043863e+00
6.97734118e-01 2.24302545e-01 7.65940011e-01 -3.68828177e-01
1.77870050e-01 3.16063583e-01 -3.84358340e-03 -4.38385874e-01
7.21492246e-02 -3.99633646e-01 -7.01731518e-02 3.56257170e-01
-8.17251131e-02 3.19162682e-02 1.28807679e-01 1.48939878e-01
1.06696975e+00 -3.70315343e-01 6.19313002e-01 -3.74014705e-01
5.61752617e-01 -1.79124355e-01 4.00503099e-01 7.80447543e-01
-4.89729166e-01 7.07442105e-01 5.37961066e-01 -3.53257000e-01
-8.38213742e-01 -1.07236707e+00 -6.51117742e-01 6.30531609e-01
-2.48897057e-02 -4.43674624e-02 -8.58161330e-01 -1.02971745e+00
3.14252451e-02 4.67982829e-01 -6.33018672e-01 -2.61909008e-01
-3.41936886e-01 -1.53257096e+00 7.86270440e-01 4.90271091e-01
7.79458046e-01 -7.49858916e-01 -4.36652094e-01 1.82864722e-02
-2.36222148e-01 -1.03432214e+00 -7.81722248e-01 1.41101480e-01
-9.96824741e-01 -1.17142498e+00 -9.82238293e-01 -7.53481984e-01
7.20874965e-01 -2.74358928e-01 1.04512370e+00 -8.90111923e-02
-5.61439812e-01 1.58670932e-01 -1.62386317e-02 -7.00895131e-01
-2.85414189e-01 2.38040790e-01 -1.08901612e-01 2.51643509e-01
1.98629528e-01 -2.48655111e-01 -9.30401325e-01 4.12768781e-01
-1.02246308e+00 -8.48596096e-02 7.75004506e-01 1.08363867e+00
7.75004208e-01 1.01551786e-02 7.41418183e-01 -1.04720962e+00
7.82014489e-01 -4.12789941e-01 -5.54064035e-01 2.87832528e-01
-8.96504819e-01 4.28177305e-02 5.53969562e-01 -2.62189597e-01
-9.14776921e-01 6.01171367e-02 -4.19322222e-01 -1.36175588e-01
2.26978451e-01 6.75238550e-01 1.69953823e-01 8.67382959e-02
9.36357856e-01 1.59331441e-01 1.55872166e-01 -3.43351424e-01
-5.54817691e-02 7.71432102e-01 2.94036627e-01 -3.14352214e-01
2.96870291e-01 5.02828121e-01 3.52353424e-01 -7.43017673e-01
-1.34558058e+00 -4.12417501e-01 -2.11134121e-01 1.21324159e-01
9.48953331e-01 -7.95100451e-01 -7.13266373e-01 6.81712389e-01
-8.52698982e-01 -3.00992012e-01 -3.18497747e-01 6.86147869e-01
-5.31130433e-01 5.33179700e-01 -6.75611079e-01 -8.06319892e-01
-7.51052737e-01 -1.18993855e+00 1.06929827e+00 2.32277706e-01
-1.02875140e-02 -1.16173625e+00 -8.71158019e-03 3.33994389e-01
4.31173444e-01 4.23627168e-01 7.61629701e-01 -6.91000581e-01
-1.97339475e-01 -4.47327822e-01 -3.01560223e-01 9.96293664e-01
-3.27557214e-02 -2.48587132e-01 -9.83357012e-01 -4.39652324e-01
1.44242778e-01 -5.52081704e-01 1.22536469e+00 8.52882266e-01
1.87640417e+00 -5.30903935e-02 -3.14214200e-01 9.27930117e-01
1.51221633e+00 1.99995205e-01 5.12150049e-01 1.07885323e-01
3.48495156e-01 3.78152817e-01 5.85915864e-01 4.77798551e-01
2.25106239e-01 5.49999595e-01 5.36176026e-01 -3.36946428e-01
7.63490647e-02 -6.46970496e-02 3.09285253e-01 4.94837493e-01
2.77420748e-02 -2.50011981e-01 -8.41129780e-01 5.18112302e-01
-1.70151818e+00 -6.24976754e-01 1.42336696e-01 2.35520220e+00
9.35804069e-01 1.89799860e-01 -6.59210654e-03 -1.25259489e-01
6.08033895e-01 1.07862376e-01 -7.65013218e-01 -2.52180070e-01
-6.45177346e-03 1.15346380e-01 5.84202051e-01 4.59335327e-01
-1.58244312e+00 4.79643792e-01 5.79873753e+00 1.28899038e+00
-1.01896727e+00 2.50130236e-01 1.44952941e+00 -1.28683254e-01
-1.91450007e-02 -4.26972270e-01 -9.16550696e-01 4.39804077e-01
7.93779910e-01 -9.71751660e-02 -9.53026935e-02 9.18060541e-01
9.05323923e-02 -3.98704346e-04 -1.18695617e+00 9.41275477e-01
-2.18038019e-02 -1.26883686e+00 -1.34583518e-01 1.39267012e-01
8.70209336e-01 1.69124141e-01 3.62243146e-01 1.74719274e-01
8.82230178e-02 -1.24595559e+00 2.69465238e-01 4.38863873e-01
1.05394661e+00 -6.97077572e-01 1.09626305e+00 3.65066886e-01
-7.00395763e-01 -4.67685750e-03 -2.63675541e-01 5.04596055e-01
6.87636733e-02 1.23527181e+00 -8.14006031e-01 3.73219073e-01
4.52366471e-01 5.69251180e-01 -5.14549971e-01 1.28595603e+00
4.98092137e-02 7.59456754e-01 -3.64312232e-01 -3.34545881e-01
3.81225675e-01 1.03857584e-01 5.90123653e-01 1.23257184e+00
2.42625684e-01 -1.34541601e-01 -6.45692572e-02 8.93174887e-01
-4.52830255e-01 2.92984635e-01 -4.49299455e-01 6.42440170e-02
-1.13396227e-01 1.26527274e+00 -4.21523422e-01 -1.36240616e-01
-1.47932246e-01 8.19617808e-01 1.74272925e-01 1.48744076e-01
-9.98745203e-01 -6.25985026e-01 4.95345861e-01 3.71857248e-02
4.23300043e-02 2.63354897e-01 -5.02902389e-01 -9.99650180e-01
1.16876304e-01 -6.79381371e-01 7.04931259e-01 -3.07481140e-01
-1.40196550e+00 7.49432743e-01 -1.15501031e-01 -1.21011043e+00
-7.68680423e-02 -7.84473956e-01 -4.54717904e-01 6.62496746e-01
-1.53766239e+00 -7.43638515e-01 -1.74789548e-01 5.08803964e-01
2.99012661e-01 -2.20038846e-01 6.65126920e-01 4.48156625e-01
-6.58618450e-01 1.24718249e+00 4.11086917e-01 2.11216867e-01
5.41318059e-01 -1.29284275e+00 -6.85379580e-02 5.09110808e-01
-4.26191633e-04 3.04438233e-01 3.68970841e-01 -2.90367842e-01
-8.12562346e-01 -1.21821010e+00 7.68771529e-01 -3.66574049e-01
5.08405387e-01 -1.32515430e-01 -5.93734860e-01 4.19766307e-01
-1.25363320e-01 8.63376725e-03 7.48126268e-01 -7.16616139e-02
2.00559180e-02 -4.81707811e-01 -1.34107375e+00 4.39662606e-01
6.18685424e-01 -9.90376323e-02 -4.14468162e-03 8.64805639e-01
7.08613038e-01 -2.95252025e-01 -1.15169263e+00 8.71034920e-01
6.51801229e-01 -8.69098842e-01 9.22279596e-01 -7.31530845e-01
9.19238985e-01 2.26584122e-01 -2.29116306e-01 -1.18195057e+00
4.47630435e-02 -5.36206663e-01 4.48681451e-02 8.29355896e-01
9.28156853e-01 -6.91922426e-01 9.72021282e-01 3.68344575e-01
4.17111404e-02 -1.64218771e+00 -9.62173402e-01 -7.61816680e-01
2.91432709e-01 -5.28192937e-01 2.65214175e-01 6.38313532e-01
-1.46499857e-01 1.37868989e-02 -3.54634136e-01 -3.14826667e-02
8.59057426e-01 -7.23067969e-02 3.82355452e-01 -9.06760633e-01
-6.37521565e-01 -6.30186558e-01 -4.89365369e-01 -9.38599885e-01
-1.68999523e-01 -1.09764612e+00 -4.14694846e-02 -1.31479418e+00
7.41696000e-01 -4.57541436e-01 -6.90886438e-01 2.65167892e-01
-5.05014241e-01 1.85042217e-01 -8.62652669e-04 1.32290065e-01
-6.33674860e-01 5.90143383e-01 1.33204615e+00 -2.42612347e-01
6.79238513e-02 5.65435231e-01 -9.89253283e-01 7.53004134e-01
8.76415074e-01 -5.57977140e-01 -3.06428373e-01 -2.81149775e-01
6.86672470e-03 4.81435172e-02 3.06117266e-01 -9.68208015e-01
1.11031504e-02 2.90535484e-02 3.14169586e-01 -2.96853662e-01
1.44837663e-01 -5.30710042e-01 -3.01335871e-01 8.41339529e-01
-5.30394197e-01 -1.65935189e-01 1.43620297e-01 4.95373935e-01
-4.28898931e-01 -3.31678629e-01 1.26071441e+00 -2.02326745e-01
-3.58158529e-01 6.13174379e-01 1.98175833e-02 2.96763569e-01
1.19212389e+00 9.28679481e-02 -2.86866903e-01 -4.07040149e-01
-7.55360484e-01 4.17210966e-01 -1.36936400e-02 2.69299205e-02
6.79355443e-01 -1.25397766e+00 -1.00769651e+00 1.09050963e-02
9.67646018e-02 1.88907050e-03 1.91582844e-01 1.36147130e+00
-5.45741081e-01 5.11499226e-01 1.65256470e-01 -7.36413300e-01
-1.31993186e+00 2.31795982e-01 7.96956956e-01 -1.12944722e+00
-3.52848530e-01 1.24250090e+00 7.14387417e-01 -4.23787504e-01
4.20677871e-01 -2.62860477e-01 -7.73598999e-02 -1.93921492e-01
5.81935167e-01 2.18890727e-01 2.35802054e-01 -2.18499571e-01
-3.86493415e-01 3.90864372e-01 -2.50646472e-01 -2.75354064e-03
1.34770572e+00 2.45348677e-01 -6.24359660e-02 2.20310867e-01
1.69802499e+00 -3.21572274e-01 -1.11221707e+00 -3.53782922e-01
-9.93461311e-02 -7.78016895e-02 9.61482450e-02 -1.01109505e+00
-1.30284441e+00 9.86564398e-01 9.67462718e-01 -1.64971408e-02
1.38715613e+00 3.39335315e-02 8.60166073e-01 3.43746096e-01
1.28593430e-01 -9.46919858e-01 1.82259336e-01 2.33652413e-01
7.54918396e-01 -1.53272092e+00 1.28151968e-01 -1.64718360e-01
-9.06610072e-01 8.73307884e-01 5.70781589e-01 2.02333018e-01
9.35327232e-01 2.37823218e-01 8.82435590e-03 -2.93030351e-01
-6.61364317e-01 -2.97826566e-02 5.31622827e-01 2.49546394e-01
4.71363097e-01 3.18667442e-01 -6.64858520e-01 9.03548837e-01
-4.22829092e-02 5.79025336e-02 2.03965843e-01 5.55598795e-01
-3.49991709e-01 -8.25831115e-01 -4.56035584e-02 1.02484608e+00
-1.10752165e+00 -2.55728841e-01 -2.26636276e-01 6.39339566e-01
-1.31889760e-01 5.65104008e-01 -2.37748206e-01 -3.42807680e-01
1.79759830e-01 -8.90319720e-02 3.20517987e-01 -3.33629757e-01
-4.47044283e-01 -2.46553689e-01 5.04535586e-02 -4.77571785e-01
-2.25481272e-01 -7.59855449e-01 -9.43154991e-01 1.65075094e-01
-4.33604747e-01 1.97440371e-01 7.54636347e-01 6.74696326e-01
6.11897223e-02 4.12958115e-01 7.85681188e-01 -2.82621592e-01
-1.18214250e+00 -9.60216641e-01 -7.63557017e-01 4.20432538e-01
3.72917652e-01 -3.80229414e-01 -3.00811410e-01 -2.42729306e-01]
|
[14.984713554382324, -2.4747369289398193]
|
9010a684-68f0-469d-bf20-2e5fa2683136
|
image-generation-network-for-covert
|
2207.10292
| null |
https://arxiv.org/abs/2207.10292v1
|
https://arxiv.org/pdf/2207.10292v1.pdf
|
Image Generation Network for Covert Transmission in Online Social Network
|
Online social networks have stimulated communications over the Internet more than ever, making it possible for secret message transmission over such noisy channels. In this paper, we propose a Coverless Image Steganography Network, called CIS-Net, that synthesizes a high-quality image directly conditioned on the secret message to transfer. CIS-Net is composed of four modules, namely, the Generation, Adversarial, Extraction, and Noise Module. The receiver can extract the hidden message without any loss even the images have been distorted by JPEG compression attacks. To disguise the behaviour of steganography, we collected images in the context of profile photos and stickers and train our network accordingly. As such, the generated images are more inclined to escape from malicious detection and attack. The distinctions from previous image steganography methods are majorly the robustness and losslessness against diverse attacks. Experiments over diverse public datasets have manifested the superior ability of anti-steganalysis.
|
['Xinpeng Zhang', 'Zhenxing Qian', 'Sheng Li', 'Qichao Ying', 'Zhengxin You']
|
2022-07-21
| null | null | null | null |
['steganalysis', 'image-steganography']
|
['computer-vision', 'computer-vision']
|
[ 1.18195164e+00 7.95745850e-01 2.65585780e-01 2.61995941e-02
-1.36061370e-01 -5.86659968e-01 8.07610691e-01 -6.73486412e-01
-2.33105138e-01 7.78913200e-01 -8.94973278e-02 -4.29037720e-01
2.53247142e-01 -1.19269395e+00 -8.02090466e-01 -9.61600244e-01
-3.79796028e-01 -1.59068689e-01 7.36686811e-02 -5.68231642e-01
3.68467480e-01 1.07978798e-01 -1.23030829e+00 2.49173343e-01
6.15678966e-01 7.43096828e-01 -8.80849361e-02 9.31241214e-01
4.27053332e-01 8.48723829e-01 -7.37814128e-01 -7.01718867e-01
6.92691028e-01 -8.66078615e-01 -6.63818121e-01 5.02055466e-01
-2.43254706e-01 -5.63428462e-01 -8.41088593e-01 1.45397568e+00
3.39657456e-01 -6.51752770e-01 3.51248294e-01 -1.54114282e+00
-5.87655008e-01 8.70944083e-01 -2.40052179e-01 -3.38645101e-01
4.10479397e-01 6.68528140e-01 4.62501138e-01 -2.68101782e-01
8.64651859e-01 1.21841037e+00 4.99936193e-01 7.74004936e-01
-9.47940171e-01 -9.36215580e-01 -6.73563182e-01 -6.87892959e-02
-1.21751273e+00 -6.02679372e-01 6.48703635e-01 9.99147352e-03
3.51286560e-01 5.83668768e-01 8.41400623e-01 1.35535634e+00
4.89656806e-01 4.05824989e-01 1.36501348e+00 -5.64151764e-01
-1.35342315e-01 3.26249987e-01 -9.75373685e-01 7.03581512e-01
6.09447181e-01 5.47210038e-01 1.11298775e-02 -2.21860170e-01
8.33564818e-01 -1.55844212e-01 -6.18270457e-01 -1.62818223e-01
-1.34009373e+00 9.61898506e-01 3.57899666e-01 3.80832285e-01
-1.81530073e-01 4.02209729e-01 6.85870722e-02 1.02654588e+00
2.32336044e-01 3.23324412e-01 1.63087010e-01 4.25422907e-01
-6.40652299e-01 -1.26449794e-01 1.24171531e+00 9.53454554e-01
5.98801196e-01 1.75970912e-01 3.62449259e-01 -9.73643064e-02
5.36262751e-01 8.43153954e-01 4.19562668e-01 -8.51234972e-01
4.17027384e-01 3.83914798e-01 -2.15356708e-01 -1.72806323e+00
2.43303061e-01 -2.41509363e-01 -1.40723336e+00 3.93960387e-01
1.60919309e-01 -4.01967287e-01 -7.42975533e-01 1.50350201e+00
4.39228676e-02 1.59057185e-01 5.72655857e-01 5.78415692e-01
4.68646139e-01 7.77148902e-01 -4.11349297e-01 -1.11830384e-01
1.04447293e+00 -6.58050239e-01 -6.83728874e-01 -3.11262935e-01
5.62989771e-01 -8.00993621e-01 1.67009711e-01 2.12293968e-01
-1.15939713e+00 -2.91186243e-01 -1.48174822e+00 6.81407750e-01
-3.89771223e-01 -6.77064955e-01 4.06519651e-01 1.17728245e+00
-1.10612452e+00 7.22448111e-01 -2.00345948e-01 -6.07018955e-02
3.80350441e-01 6.06864989e-01 -6.38890922e-01 -7.92753771e-02
-1.73358822e+00 5.49401641e-01 8.57884347e-01 1.13405921e-02
-1.05551851e+00 1.10881805e-01 -9.07284975e-01 1.40100950e-02
1.85642675e-01 -7.00915992e-01 5.49798667e-01 -1.58811522e+00
-1.46964025e+00 1.04150522e+00 6.62019312e-01 -7.51314402e-01
9.53571558e-01 6.77440166e-01 -8.05841982e-01 5.03565252e-01
-2.02052489e-01 5.92629611e-01 1.59225714e+00 -1.23772418e+00
-3.66425365e-01 2.81545520e-03 -1.02812104e-01 -1.97804138e-01
-1.17640056e-01 -1.64079979e-01 -1.23851724e-01 -5.43383896e-01
1.57560542e-01 -1.33105290e+00 -2.22366214e-01 -1.16237722e-01
-7.74149001e-01 6.57864869e-01 1.15710318e+00 -5.62197626e-01
8.10839891e-01 -2.31369877e+00 -1.53307334e-01 7.63545930e-01
2.97429472e-01 6.00953758e-01 -3.13938111e-01 8.54408503e-01
-1.61532253e-01 6.29543483e-01 -3.95209402e-01 2.22368374e-01
-1.29357636e-01 2.60835171e-01 -2.36778095e-01 8.32241476e-01
1.45172654e-02 9.20383155e-01 -8.08651924e-01 -5.20590603e-01
3.58033217e-02 5.52614689e-01 -4.01609421e-01 2.48383284e-02
8.34535360e-02 5.88870227e-01 -4.16783959e-01 3.75888914e-01
9.08788979e-01 -4.07206208e-01 5.40387213e-01 1.55792072e-01
3.42854023e-01 -1.27946034e-01 -9.94963944e-01 9.48059738e-01
2.38678064e-02 7.65737236e-01 2.91446447e-01 -7.40937710e-01
7.80695438e-01 7.06687212e-01 2.22020924e-01 -4.85987335e-01
5.80795705e-01 3.59900355e-01 2.28981331e-01 -6.74255371e-01
3.65328521e-01 1.13837175e-01 -1.26177147e-01 6.60809934e-01
-9.99350548e-02 -1.52949110e-01 -1.03105269e-01 3.25009018e-01
1.09310579e+00 -3.04555088e-01 1.46723881e-01 1.10509954e-01
6.94362640e-01 -3.50524515e-01 6.58325329e-02 9.86950457e-01
4.33003297e-03 3.74292463e-01 4.88733381e-01 -7.90149719e-02
-1.44214916e+00 -6.20633781e-01 4.10543531e-01 2.33620062e-01
5.35540700e-01 -1.12191431e-01 -1.02511704e+00 -7.26630270e-01
-2.34436080e-01 1.65690765e-01 -3.37223858e-01 -4.58702564e-01
-5.89111626e-01 -5.58336437e-01 1.14405739e+00 -5.20584643e-01
1.29169190e+00 -1.27985156e+00 -2.78560489e-01 2.60818094e-01
-4.46160406e-01 -1.24569166e+00 -2.70985007e-01 -5.53708613e-01
-4.96448010e-01 -1.40366650e+00 -6.79460704e-01 -9.28049624e-01
9.61715043e-01 4.19780016e-01 7.11339295e-01 9.28569019e-01
-7.92997703e-02 1.52060196e-01 -3.70119333e-01 -3.34909767e-01
-1.46530819e+00 1.82752516e-02 -2.76506871e-01 4.11315441e-01
-1.79987594e-01 -6.34954393e-01 -8.23370934e-01 6.13739312e-01
-1.48048508e+00 2.19269082e-01 9.21088874e-01 7.17281997e-01
-2.42617249e-01 5.44963837e-01 3.76103550e-01 -1.15858519e+00
3.72540027e-01 -6.22538030e-01 -5.02116680e-01 2.05360763e-02
-6.82170451e-01 -7.62737915e-02 5.14764071e-01 -3.87162924e-01
-7.44765997e-01 -1.69045374e-01 -9.37943459e-02 2.03601047e-01
3.44139971e-02 1.09707765e-01 -2.82301575e-01 -9.11215127e-01
4.03170824e-01 7.87514031e-01 5.79635799e-01 2.13898316e-01
2.38871738e-01 9.69077885e-01 4.26595688e-01 2.49437019e-01
1.64701843e+00 7.41388142e-01 1.93945840e-01 -9.24192131e-01
-6.10877573e-02 1.73910379e-01 -1.84544828e-03 -3.86691928e-01
6.46506011e-01 -7.12027252e-01 -8.89596760e-01 1.09436190e+00
-1.10798931e+00 1.24106474e-01 1.62631556e-01 2.67093986e-01
-5.15010834e-01 7.61024535e-01 -5.89029491e-01 -5.79264700e-01
-4.35745150e-01 -9.30107296e-01 3.30545187e-01 6.02145121e-03
2.04806402e-01 -1.04108918e+00 -3.33970159e-01 3.96104813e-01
7.30657279e-01 7.99708784e-01 4.02711004e-01 -5.25067985e-01
-1.13130438e+00 -5.49641550e-01 -2.15115145e-01 6.62526786e-01
6.84221685e-02 -1.74746469e-01 -8.01369488e-01 -7.47987270e-01
2.44269803e-01 -1.56150579e-01 6.75064504e-01 -2.39669099e-01
8.98311436e-01 -1.12816596e+00 -2.47080639e-01 1.01469195e+00
1.62453592e+00 2.95721501e-01 1.49947941e+00 4.66290861e-01
3.62438530e-01 5.89079916e-01 5.22583015e-02 1.43400922e-01
-3.23916078e-02 -3.21407765e-02 9.40239608e-01 -2.48393700e-01
2.13032648e-01 -4.71213490e-01 5.21535456e-01 7.53623486e-01
-1.46206602e-01 -8.93885136e-01 -3.00564468e-01 2.43622229e-01
-1.26987839e+00 -1.30096447e+00 -1.44520879e-01 1.77617264e+00
7.55423725e-01 3.00187230e-01 -3.61830503e-01 4.57453161e-01
1.08227265e+00 5.58541775e-01 -1.18069850e-01 -2.68508196e-01
-3.14290553e-01 -8.43736529e-02 1.20685589e+00 3.44326615e-01
-8.97149086e-01 8.41941535e-01 6.13796568e+00 8.13143492e-01
-1.06839836e+00 -1.26240030e-01 6.41511738e-01 6.45516932e-01
-2.96883196e-01 3.31718266e-01 -3.35850716e-01 7.20986605e-01
8.57428491e-01 -4.04411368e-02 6.16393328e-01 3.12941641e-01
-2.64843255e-01 5.75354993e-02 -4.61997777e-01 6.71763241e-01
2.95298606e-01 -1.42438447e+00 1.52781293e-01 5.76260030e-01
8.28163803e-01 -4.11527395e-01 2.05617145e-01 -1.97795466e-01
3.20788592e-01 -1.07249129e+00 4.15265083e-01 2.49057308e-01
1.01255882e+00 -6.38895512e-01 7.82846868e-01 5.01017272e-01
-5.23382127e-01 3.75729948e-02 -3.40879112e-01 7.11236969e-02
1.47749081e-01 1.41702250e-01 -9.06142414e-01 6.34410799e-01
1.20111458e-01 4.24541712e-01 -2.79119164e-01 6.15442216e-01
-4.39275235e-01 7.77368367e-01 -4.60240245e-02 1.45331815e-01
3.94879580e-01 -1.99129894e-01 9.98808086e-01 9.62042511e-01
5.37020922e-01 1.51680514e-01 -2.59839505e-01 5.50293386e-01
-3.44585985e-01 -3.37721348e-01 -1.12204695e+00 -2.64316350e-01
2.93866843e-01 9.29858029e-01 -7.55181730e-01 -3.61823946e-01
-2.75777280e-02 1.26408923e+00 -7.30169296e-01 1.99885860e-01
-6.79995775e-01 -8.00957739e-01 1.40113056e-01 1.87089160e-01
3.99382442e-01 -2.14878097e-02 2.19851807e-01 -1.20968151e+00
-3.07819933e-01 -1.34893894e+00 -6.72341809e-02 -6.02158964e-01
-8.69783401e-01 6.05817795e-01 -4.82750803e-01 -1.39718306e+00
-2.21521631e-01 -1.14946656e-01 -4.95862842e-01 4.95197207e-01
-1.59318030e+00 -1.15020740e+00 -1.53555676e-01 8.35858524e-01
-4.30800319e-02 -5.04652083e-01 7.35464156e-01 5.30597754e-02
-7.24403560e-02 5.82877100e-01 5.02615869e-02 5.62583625e-01
3.86132717e-01 -5.65789878e-01 6.52282298e-01 9.54262435e-01
-2.75812000e-01 2.55197257e-01 8.56385052e-01 -7.67666817e-01
-1.39266884e+00 -1.06311476e+00 8.93648624e-01 4.74438965e-02
5.55099547e-01 -4.67286795e-01 -5.25560856e-01 7.22656310e-01
5.44271469e-01 -3.30674529e-01 2.81348914e-01 -1.29071534e+00
-2.34003350e-01 1.51051879e-01 -1.71306813e+00 5.72310627e-01
1.00771773e+00 -4.62972790e-01 -2.55617321e-01 2.42322445e-01
6.66323662e-01 -1.65476546e-01 -5.45267761e-01 2.48177741e-02
6.67166710e-01 -1.18175030e+00 1.05587161e+00 -5.53324223e-02
6.97338820e-01 -6.63656518e-02 6.18910678e-02 -1.13740921e+00
1.47209996e-02 -1.48018849e+00 2.01297760e-01 9.60239708e-01
2.39833966e-01 -1.15083921e+00 7.72339344e-01 -9.26668644e-02
5.23009956e-01 5.89877069e-02 -6.39846981e-01 -7.75438488e-01
-3.60512763e-01 7.74673074e-02 9.26666558e-01 1.16018796e+00
-2.25929305e-01 -1.63497329e-01 -1.05263567e+00 3.03669870e-01
1.09267330e+00 -4.66592550e-01 9.72785473e-01 -9.55515623e-01
-2.84427404e-01 9.87946987e-04 -6.78132296e-01 -8.18321407e-01
-9.71345305e-02 -8.62778842e-01 -1.66283190e-01 -8.82882118e-01
-1.59439012e-01 -3.34283501e-01 1.78104475e-01 1.11687116e-01
3.09007972e-01 8.14284384e-01 3.25624108e-01 4.02498484e-01
-2.67757773e-01 8.52880254e-02 1.85523760e+00 -1.39912203e-01
2.54805833e-01 1.68259799e-01 -7.94187307e-01 7.00997829e-01
1.07663393e+00 -1.00354540e+00 -3.54846716e-01 6.63865507e-02
5.72838843e-01 5.35318911e-01 6.19713604e-01 -1.04383957e+00
6.00179881e-02 -8.54472592e-02 2.28086948e-01 1.28747365e-02
6.44566640e-02 -1.16233790e+00 5.63920975e-01 1.23563433e+00
-3.73990625e-01 -4.23599273e-01 -5.55640161e-01 6.98895633e-01
-1.90653265e-01 -2.08366275e-01 8.36746633e-01 -5.00615716e-01
-4.15026665e-01 2.17557549e-01 -6.97714150e-01 -1.96582511e-01
1.11594760e+00 -5.87605298e-01 -4.66355652e-01 -8.76124859e-01
-5.66013277e-01 -1.77374437e-01 6.84506595e-01 1.84005633e-01
8.86148691e-01 -1.05368316e+00 -9.25852776e-01 8.90734494e-01
-2.07467854e-01 -4.69975144e-01 9.11874976e-03 4.37673092e-01
-8.89880657e-01 2.74864640e-02 -4.48594481e-01 -2.58214563e-01
-1.28053534e+00 4.48589921e-01 3.90521884e-01 -2.32648984e-01
-5.61929047e-01 3.47186714e-01 -1.72017038e-01 -1.51974067e-01
-4.47135456e-02 3.74518126e-01 -9.16218981e-02 -4.07907099e-01
5.79675853e-01 3.02037805e-01 -4.97220755e-01 -8.23513985e-01
1.67297468e-01 2.16376081e-01 1.28045425e-01 -1.21989310e-01
1.12987077e+00 -5.86707711e-01 -4.33197021e-01 -4.69356924e-01
1.36640108e+00 -5.43510467e-02 -9.56622124e-01 -1.11377515e-01
-3.04885298e-01 -7.93426752e-01 -1.59484878e-01 -4.48453516e-01
-1.30579412e+00 3.23952109e-01 3.75670224e-01 9.96223509e-01
9.39679563e-01 -5.41685164e-01 1.26836228e+00 2.90606558e-01
5.20058990e-01 -7.05454648e-01 7.84488395e-03 2.59793282e-01
5.11672735e-01 -1.19773209e+00 -2.01882482e-01 -5.80141842e-01
-3.33848566e-01 1.10484421e+00 -6.47361055e-02 -3.95354360e-01
6.57076299e-01 1.17832936e-01 -8.63786936e-02 -2.68832773e-01
-4.76652741e-01 2.77948618e-01 -1.78147256e-01 8.78720939e-01
-4.50710148e-01 -9.28082094e-02 -1.72852159e-01 -3.56552452e-01
-3.23252559e-01 8.94298330e-02 1.11216223e+00 1.04052305e+00
-5.87758720e-01 -1.10649419e+00 -6.18495822e-01 -2.73201223e-02
-9.20583904e-01 -4.47902046e-02 -4.38966841e-01 9.74837303e-01
1.60327733e-01 1.22982597e+00 -2.88094789e-01 -5.98602116e-01
-2.08532140e-01 -3.51167053e-01 1.48039490e-01 -5.76461852e-02
-6.67349815e-01 -2.06048325e-01 1.02865696e-01 -3.86227518e-01
-7.42105961e-01 -7.49801323e-02 -7.06760228e-01 -1.01259339e+00
-4.69333321e-01 2.61260539e-01 8.36434543e-01 7.32444584e-01
1.41441122e-01 2.02423885e-01 1.42747414e+00 -6.19419396e-01
-5.98795414e-01 -5.44397771e-01 -6.01574123e-01 5.97677231e-01
7.20329285e-01 3.46014887e-01 -1.00687218e+00 2.73514181e-01]
|
[4.325167655944824, 8.045863151550293]
|
60b4e0f5-18a4-437c-8a5b-2a4aa7dd0649
|
instance-dependent-noisy-label-learning-via
|
2209.00906
| null |
https://arxiv.org/abs/2209.00906v1
|
https://arxiv.org/pdf/2209.00906v1.pdf
|
Instance-Dependent Noisy Label Learning via Graphical Modelling
|
Noisy labels are unavoidable yet troublesome in the ecosystem of deep learning because models can easily overfit them. There are many types of label noise, such as symmetric, asymmetric and instance-dependent noise (IDN), with IDN being the only type that depends on image information. Such dependence on image information makes IDN a critical type of label noise to study, given that labelling mistakes are caused in large part by insufficient or ambiguous information about the visual classes present in images. Aiming to provide an effective technique to address IDN, we present a new graphical modelling approach called InstanceGM, that combines discriminative and generative models. The main contributions of InstanceGM are: i) the use of the continuous Bernoulli distribution to train the generative model, offering significant training advantages, and ii) the exploration of a state-of-the-art noisy-label discriminative classifier to generate clean labels from instance-dependent noisy-label samples. InstanceGM is competitive with current noisy-label learning approaches, particularly in IDN benchmarks using synthetic and real-world datasets, where our method shows better accuracy than the competitors in most experiments.
|
['Gustavo Carneiro', 'Thanh-Toan Do', 'Rafael Felix', 'Cuong Nguyen', 'Arpit Garg']
|
2022-09-02
| null | null | null | null |
['learning-with-noisy-labels', 'learning-with-noisy-labels']
|
['computer-vision', 'natural-language-processing']
|
[ 4.47688133e-01 -4.06389460e-02 2.54536215e-02 -6.04007065e-01
-1.02951872e+00 -6.34234548e-01 8.13344657e-01 -1.31218195e-01
-3.49013776e-01 7.61478782e-01 -1.10915944e-01 2.38473788e-02
-1.55870944e-01 -5.28891504e-01 -6.70027792e-01 -1.07480824e+00
3.23877752e-01 7.12667167e-01 2.10843105e-02 1.43158033e-01
3.00703421e-02 3.63049358e-01 -1.97744715e+00 2.38303632e-01
6.54870331e-01 1.19324648e+00 2.16243327e-01 2.98149884e-01
-4.00681347e-01 9.17354524e-01 -8.43454301e-01 -5.41669190e-01
2.07963467e-01 -5.94352305e-01 -7.91023016e-01 3.87920260e-01
3.95815462e-01 2.17158467e-01 1.62471846e-01 1.38203967e+00
7.37053692e-01 7.27649480e-02 1.14768100e+00 -1.33314145e+00
-5.59090495e-01 6.30243361e-01 -4.80504721e-01 -2.19618917e-01
-6.43354431e-02 4.65931594e-02 7.87902951e-01 -8.44825625e-01
8.09565663e-01 1.59310305e+00 7.94746697e-01 7.80638278e-01
-1.68038607e+00 -6.54353201e-01 1.11254573e-01 3.16489398e-01
-1.23939359e+00 -1.98610261e-01 7.21958399e-01 -5.08032739e-01
2.63016641e-01 2.66590983e-01 4.23422903e-01 1.87618566e+00
1.34215252e-02 7.17948318e-01 1.84075928e+00 -5.38538873e-01
5.32919705e-01 2.26343781e-01 8.71148109e-02 2.55003750e-01
2.12389886e-01 9.79217142e-02 -3.76132548e-01 -1.49578646e-01
3.01347226e-01 3.52520566e-03 -7.69946203e-02 -4.67840612e-01
-9.04055297e-01 9.61292267e-01 3.94322336e-01 2.69981474e-01
-7.60614350e-02 3.69088024e-01 4.45532948e-01 9.34268758e-02
7.29662120e-01 3.30839336e-01 -3.24107647e-01 -1.48831740e-01
-9.76112068e-01 1.08557194e-01 7.31471777e-01 8.59624088e-01
7.97016859e-01 -9.78123676e-03 -3.85348946e-01 1.08006489e+00
3.59823316e-01 6.59239769e-01 4.76762921e-01 -7.30301023e-01
4.55703326e-02 4.23615515e-01 -2.86133327e-02 -9.23861325e-01
-4.72169697e-01 -7.37944961e-01 -1.09453464e+00 2.98033446e-01
4.99401212e-01 1.32724777e-01 -1.49743080e+00 1.81257534e+00
2.99388319e-01 6.86737150e-02 -3.01644653e-01 7.55523205e-01
1.05763364e+00 3.84581804e-01 4.22918111e-01 -2.12549195e-01
1.17210412e+00 -8.70237231e-01 -9.43295240e-01 -3.44865501e-01
7.35688090e-01 -8.60681772e-01 9.68499660e-01 5.96871138e-01
-5.81382036e-01 -6.17656887e-01 -6.98111773e-01 1.25672609e-01
-5.80324531e-01 1.87061518e-01 5.18068612e-01 8.08532596e-01
-1.03941965e+00 5.43912053e-01 -4.53050107e-01 -1.89431816e-01
6.41749382e-01 3.46578695e-02 -3.60974222e-01 -5.28328896e-01
-1.02535224e+00 9.19725537e-01 4.48195279e-01 1.88145831e-01
-1.35544837e+00 -3.63359571e-01 -9.51245725e-01 -1.96325913e-01
6.25404954e-01 -3.79008859e-01 1.20591831e+00 -1.13464701e+00
-1.08518636e+00 1.24464452e+00 7.16802552e-02 -2.08242372e-01
8.08015704e-01 1.44741237e-01 -2.67390877e-01 -9.07327756e-02
1.77703202e-01 7.53746331e-01 1.14752316e+00 -1.93742645e+00
-2.76103914e-01 -2.63214618e-01 -3.33944321e-01 -2.19616309e-01
1.16010584e-01 -1.88724443e-01 -1.01009503e-01 -8.38706851e-01
2.64046401e-01 -1.05206394e+00 -1.48580432e-01 -3.68806869e-01
-6.26584589e-01 -5.28647006e-01 7.91896760e-01 -4.25666660e-01
9.60477114e-01 -2.08102751e+00 9.04606935e-03 1.59870759e-01
3.25084120e-01 4.43144739e-01 -1.55331656e-01 3.59047920e-01
-2.50649780e-01 3.50071073e-01 -2.98286021e-01 -7.31064439e-01
2.84042001e-01 5.81805229e-01 2.29180548e-02 4.20612097e-01
3.00029963e-01 9.93824542e-01 -1.10746491e+00 -5.96759140e-01
1.44839719e-01 5.64421475e-01 2.00298294e-01 1.62803754e-01
-3.99304628e-01 6.56164885e-01 -1.93031833e-01 6.58087969e-01
9.27898169e-01 -2.92032212e-01 1.18655987e-01 -1.22494273e-01
2.15763956e-01 3.26052420e-02 -1.44487357e+00 1.50849426e+00
-3.64908189e-01 3.13203454e-01 -9.28995907e-02 -1.06117582e+00
1.01855028e+00 2.48255193e-01 1.74682349e-01 -6.59460127e-01
3.82772982e-01 4.03020859e-01 -3.77188832e-01 -6.06576383e-01
-2.50742566e-02 -4.33876127e-01 -1.19037278e-01 2.65123755e-01
5.20853877e-01 -2.14344606e-01 3.62819642e-01 1.16435617e-01
9.53962266e-01 2.45186955e-01 1.34960204e-01 -3.28824759e-01
2.21222565e-01 -3.68418843e-01 6.69054925e-01 1.09532893e+00
-2.42713302e-01 9.59182024e-01 7.27109253e-01 -4.85826701e-01
-9.66926813e-01 -8.62402201e-01 -3.23352665e-01 9.76034939e-01
5.16258441e-02 -9.19384360e-02 -8.87033343e-01 -1.19087350e+00
-1.86796442e-01 5.44180393e-01 -9.00128245e-01 -5.34838401e-02
-1.19273737e-01 -9.70153511e-01 4.05920088e-01 2.91844040e-01
4.62264299e-01 -1.24828124e+00 1.57744940e-02 9.88724008e-02
-2.64368713e-01 -9.26616311e-01 -1.52995288e-01 8.67542446e-01
-5.14171660e-01 -1.17754936e+00 -6.55146122e-01 -7.10871637e-01
6.78588450e-01 1.16956301e-01 1.60341549e+00 2.84900120e-03
-4.31766927e-01 2.93071389e-01 -5.90789378e-01 -3.93101364e-01
-8.14745367e-01 5.64958304e-02 -2.79582351e-01 1.97432190e-01
3.39064270e-01 -4.46437627e-01 -4.37993795e-01 4.62718815e-01
-1.24866772e+00 -2.10221529e-01 7.23449171e-01 1.20068467e+00
8.38506758e-01 1.52962536e-01 5.69756925e-01 -1.28125191e+00
2.46307880e-01 -4.90152061e-01 -6.01311386e-01 1.62172034e-01
-7.14215636e-01 2.53447831e-01 4.43709522e-01 -6.06866658e-01
-9.92390513e-01 2.48910740e-01 -3.62725556e-01 -2.93532848e-01
-5.74870706e-01 1.58837676e-01 -5.76250196e-01 -3.10091712e-02
8.06443393e-01 -3.19675840e-02 -1.26991004e-01 -8.22335243e-01
3.98642004e-01 6.09243929e-01 3.09583277e-01 -5.38213670e-01
5.65053165e-01 5.04181504e-01 3.43425393e-01 -4.91744697e-01
-1.35809493e+00 -6.62625968e-01 -6.16420269e-01 -3.41530740e-01
9.13971484e-01 -6.78072274e-01 -3.86112124e-01 8.97310615e-01
-9.78107631e-01 -2.21273929e-01 -3.04058671e-01 1.28091648e-01
-3.84854436e-01 3.74242574e-01 -5.83253980e-01 -8.80175769e-01
-4.47991071e-03 -1.46325254e+00 1.26498008e+00 1.40521511e-01
1.45353407e-01 -1.03994679e+00 -1.49562694e-02 3.39430600e-01
3.34176540e-01 4.88626361e-01 7.20162570e-01 -8.99950981e-01
-3.39451939e-01 -2.93368638e-01 -3.62404823e-01 9.40226674e-01
-8.29803050e-02 -1.24441676e-01 -1.44707644e+00 -1.89798325e-01
1.20114632e-01 -8.09418023e-01 9.93126392e-01 3.29472274e-01
1.11058903e+00 1.47668989e-02 -3.62317204e-01 3.67861241e-01
1.65703714e+00 -1.56840354e-01 6.94617331e-01 2.37660676e-01
8.98461819e-01 5.55694520e-01 5.09634197e-01 2.44108647e-01
4.93139103e-02 7.99346507e-01 8.98866355e-01 -2.89730906e-01
-4.07855332e-01 -1.99114487e-01 -3.95190269e-02 6.39279664e-01
1.70298845e-01 -3.55891883e-01 -7.58769512e-01 3.93754959e-01
-1.78021312e+00 -7.01589286e-01 -6.02769017e-01 2.10382843e+00
1.00380433e+00 4.45503034e-02 -7.58343861e-02 3.44087958e-01
8.92475963e-01 9.45042260e-03 -2.93584704e-01 -1.44264042e-01
-2.74952829e-01 2.90007114e-01 4.82149124e-01 2.08938673e-01
-1.57991052e+00 7.01531112e-01 5.91958570e+00 1.54171026e+00
-8.60305905e-01 4.85263109e-01 9.42726851e-01 2.51744717e-01
-2.11397216e-01 -1.83026716e-01 -9.64088202e-01 9.14010882e-01
8.14116597e-01 6.70230269e-01 1.71279132e-01 1.01645005e+00
-8.30908045e-02 -2.30301738e-01 -9.64573801e-01 1.14984000e+00
2.24840507e-01 -6.71484709e-01 -1.81817383e-01 1.80838510e-01
1.12159026e+00 -1.09135114e-01 1.37808457e-01 3.34704638e-01
5.34820080e-01 -1.21323669e+00 9.18235362e-01 5.74298143e-01
6.72092497e-01 -9.10039842e-01 1.28008330e+00 3.82240921e-01
-6.63027465e-01 9.34478175e-03 -6.34578466e-01 2.11110428e-01
-5.14075756e-02 1.37876451e+00 -5.08244991e-01 4.66675073e-01
7.75557578e-01 5.83609700e-01 -8.64471555e-01 1.09023285e+00
-6.50249720e-01 8.50459814e-01 -2.22323149e-01 1.67956442e-01
3.66603136e-01 -2.24106744e-01 1.98117167e-01 1.31625247e+00
2.48812616e-01 -4.75718290e-01 2.75363624e-01 9.18073475e-01
-1.51581094e-01 5.79509847e-02 -6.62651718e-01 3.05411130e-01
2.01017305e-01 1.53722811e+00 -1.07042539e+00 -2.44680062e-01
-1.69884473e-01 8.48355889e-01 1.91435739e-01 2.09715292e-01
-7.86385834e-01 1.33559868e-01 2.47090682e-01 -1.04358494e-01
3.12103212e-01 1.26220539e-01 -2.76934654e-01 -8.22822988e-01
1.03139188e-02 -1.05524027e+00 2.08976299e-01 -6.33802533e-01
-1.73384500e+00 6.32692039e-01 -1.09731682e-01 -1.19587791e+00
-1.27750620e-01 -6.01525664e-01 4.47733700e-02 8.15728009e-01
-1.49777937e+00 -1.38990378e+00 -5.56670308e-01 2.72343934e-01
3.88427496e-01 2.36785442e-01 9.53223944e-01 3.80630344e-01
-2.58037329e-01 4.83353078e-01 4.86514390e-01 -4.26721871e-02
9.42404807e-01 -1.53039479e+00 2.69013464e-01 6.01832032e-01
4.50033069e-01 -2.11192411e-03 6.82776630e-01 -4.93283719e-01
-6.58137918e-01 -1.18459666e+00 8.21887672e-01 -7.47086704e-01
4.02858496e-01 -8.47439945e-01 -7.83882618e-01 3.00738633e-01
-5.00219539e-02 4.04957652e-01 6.53715849e-01 -1.50429204e-01
-5.86212575e-01 -9.59592778e-03 -1.22780311e+00 2.55922228e-01
1.23207104e+00 -2.75693148e-01 -2.06381708e-01 5.96310914e-01
4.54429001e-01 -2.57566452e-01 -5.52062809e-01 4.77446645e-01
1.11791372e-01 -1.22319055e+00 8.09391797e-01 -1.95457697e-01
3.80971014e-01 -2.68685728e-01 -2.88673062e-02 -1.51716459e+00
-3.02068532e-01 -2.73769975e-01 2.39034325e-01 1.60438573e+00
3.35278630e-01 -4.36261058e-01 5.59859037e-01 2.74283767e-01
-4.91696708e-02 -5.98890007e-01 -1.14240098e+00 -9.06637311e-01
-7.31794834e-02 -5.78789771e-01 3.56710315e-01 9.36329305e-01
-7.35876560e-01 2.00152084e-01 -6.82083249e-01 -3.12379032e-01
7.55329072e-01 -2.72490084e-01 5.67210615e-01 -1.53714836e+00
-1.57392263e-01 -2.39208490e-01 -6.99014187e-01 -7.23858535e-01
2.82557696e-01 -9.33528066e-01 4.22456890e-01 -1.52566195e+00
3.55822265e-01 -6.58753037e-01 -2.89987385e-01 5.45446098e-01
-3.19567442e-01 8.86333525e-01 1.27099797e-01 2.07222223e-01
-8.69292796e-01 5.11065006e-01 1.39616096e+00 -3.11902553e-01
5.18696368e-01 1.70307919e-01 -4.72489655e-01 6.66086733e-01
4.34519857e-01 -1.02518439e+00 -4.11720455e-01 -5.97612048e-03
3.85557711e-01 -5.39532840e-01 4.48162466e-01 -9.81123149e-01
-3.45094591e-01 1.62898511e-01 3.02815229e-01 -4.55066085e-01
2.67894596e-01 -1.02445066e+00 3.20363522e-01 1.47109479e-01
-2.82674372e-01 -3.65820855e-01 -2.23595142e-01 7.58706629e-01
-2.93567955e-01 -7.58200109e-01 1.06390488e+00 -4.16982293e-01
-5.66923082e-01 -4.79880162e-02 -2.45897993e-01 3.46807301e-01
8.38087976e-01 -9.76793584e-04 -2.97838867e-01 -3.61377239e-01
-9.29227710e-01 -1.47890404e-01 4.87906843e-01 3.04257244e-01
1.83722511e-01 -1.38233566e+00 -5.82447648e-01 1.30727962e-01
2.76172310e-01 1.39092028e-01 3.55583280e-01 7.85559475e-01
-2.23013654e-01 1.94215745e-01 1.66899353e-01 -8.54739845e-01
-1.10721231e+00 8.44010472e-01 3.61619085e-01 -6.58691049e-01
-4.69595075e-01 1.06605101e+00 3.41305435e-01 -5.59659600e-01
4.85022873e-01 -2.34042719e-01 -2.50086308e-01 3.26566041e-01
4.29188877e-01 3.30843002e-01 4.41311866e-01 -7.13386714e-01
-1.21548213e-01 6.05006695e-01 1.17299527e-01 2.29038417e-01
1.05670011e+00 -1.17488936e-01 -2.25917801e-01 6.51803195e-01
1.22482824e+00 -4.70574170e-01 -1.32087505e+00 -2.20977679e-01
1.66013673e-01 -4.17675823e-01 -2.45055440e-03 -1.11933470e+00
-1.08609545e+00 9.04855072e-01 9.74516273e-01 5.39019346e-01
1.01840675e+00 1.37599319e-01 3.97876412e-01 -5.17040491e-02
5.48114717e-01 -1.22938681e+00 2.34536514e-01 3.84957403e-01
6.15390062e-01 -1.58568001e+00 -2.69895017e-01 -4.86499906e-01
-5.96911550e-01 7.83165812e-01 1.58432871e-01 7.73919225e-02
6.93539917e-01 2.16692165e-01 4.86441225e-01 -1.63094044e-01
-3.37268233e-01 -4.44287509e-01 2.58378506e-01 9.28657413e-01
1.90335721e-01 3.84389497e-02 -5.60382247e-01 5.54653108e-01
1.63203880e-01 -1.31613687e-01 2.12551862e-01 7.18319952e-01
-1.92153022e-01 -1.43905461e+00 -5.90618253e-01 4.00794089e-01
-6.55877888e-01 -8.63445029e-02 -2.84249127e-01 7.04015911e-01
7.32393265e-01 1.11057127e+00 -2.95582801e-01 -2.23714903e-01
1.88125342e-01 4.48227108e-01 3.19055915e-01 -6.50084794e-01
-5.00940025e-01 1.68843240e-01 -1.68163446e-03 -5.28004408e-01
-6.29818976e-01 -4.83532608e-01 -7.69257724e-01 4.39593941e-02
-5.75434685e-01 4.75637540e-02 1.16119492e+00 1.16554701e+00
1.69254482e-01 5.43445408e-01 5.97260892e-01 -1.13598192e+00
-7.92303979e-01 -1.43800426e+00 -9.79306936e-01 1.04411411e+00
3.38839032e-02 -1.13921416e+00 -8.85901511e-01 1.78778976e-01]
|
[9.426602363586426, 3.868515729904175]
|
06982017-eaec-47b8-92bd-b98a5fbf4d16
|
a-framework-for-semi-automated-web-service
|
1311.6709
| null |
http://arxiv.org/abs/1311.6709v1
|
http://arxiv.org/pdf/1311.6709v1.pdf
|
A Framework for Semi-automated Web Service Composition in Semantic Web
|
Number of web services available on Internet and its usage are increasing
very fast. In many cases, one service is not enough to complete the business
requirement; composition of web services is carried out. Autonomous composition
of web services to achieve new functionality is generating considerable
attention in semantic web domain. Development time and effort for new
applications can be reduced with service composition. Various approaches to
carry out automated composition of web services are discussed in literature.
Web service composition using ontologies is one of the effective approaches. In
this paper we demonstrate how the ontology based composition can be made faster
for each customer. We propose a framework to provide precomposed web services
to fulfil user requirements. We detail how ontology merging can be used for
composition which expedites the whole process. We discuss how framework
provides customer specific ontology merging and repository. We also elaborate
on how merging of ontologies is carried out.
|
['Archana Chougule', 'Debajyoti Mukhopadhyay']
|
2013-11-26
| null | null | null | null |
['service-composition']
|
['miscellaneous']
|
[ 1.21235967e-01 -4.30745631e-02 1.22579597e-01 -7.23043442e-01
-2.66304493e-01 -8.59776199e-01 5.96657157e-01 -6.31318390e-02
-1.18827380e-01 4.63484406e-01 1.72459394e-01 -3.31090420e-01
-4.01311427e-01 -1.11570823e+00 -2.76260916e-02 -4.89210367e-01
2.62927234e-01 9.00280654e-01 7.49295533e-01 -6.74116015e-01
2.56370574e-01 5.86826384e-01 -2.20016503e+00 4.64512080e-01
8.55558455e-01 1.02702367e+00 4.67970043e-01 2.28457481e-01
-1.31037903e+00 1.12823069e-01 -2.64547884e-01 -3.84430200e-01
4.33224291e-01 -4.77051705e-01 -9.47282732e-01 7.58799493e-01
-5.19238949e-01 5.95041132e-03 6.27054691e-01 1.18292582e+00
3.26697409e-01 6.97623864e-02 4.76038784e-01 -1.68413639e+00
-3.28440875e-01 8.24990273e-01 2.59575397e-02 -3.58507454e-01
6.26302183e-01 -4.44077581e-01 9.23466504e-01 -4.04175639e-01
6.76194072e-01 9.42811370e-01 3.76228131e-02 5.98580658e-01
-6.36868179e-01 -4.43467855e-01 2.63585716e-01 6.16848469e-01
-1.22589290e+00 -2.47193322e-01 5.45587182e-01 7.07363337e-02
9.89493012e-01 5.41369915e-01 7.62992084e-01 2.14365020e-01
-2.24600211e-01 1.87296510e-01 9.17925417e-01 -6.57129943e-01
4.29035485e-01 3.62219572e-01 2.39646390e-01 1.51094317e-01
6.71427906e-01 -7.01667070e-01 1.17673315e-02 -1.25037670e-01
1.60281613e-01 3.07800323e-01 -1.69667900e-01 -3.62184584e-01
-4.66694146e-01 4.29372489e-01 -2.81345755e-01 1.01339197e+00
-3.68273228e-01 -2.67624676e-01 4.10552531e-01 4.40794677e-01
-2.39155993e-01 8.33494589e-02 -6.58010840e-01 -4.93217200e-01
-1.82753444e-01 4.43429321e-01 1.34781170e+00 1.28172040e+00
5.56873262e-01 1.25195757e-01 1.01570475e+00 1.02787530e+00
3.62925351e-01 2.47842133e-01 6.56897366e-01 -1.04912949e+00
-2.89514840e-01 1.28640819e+00 3.42344522e-01 -4.30404544e-01
-1.30886510e-01 1.94090735e-02 1.42296970e-01 3.54250789e-01
1.87805086e-01 1.73289165e-01 -7.10695088e-01 8.46713960e-01
6.00119770e-01 -2.57728964e-01 4.31051314e-01 7.68307030e-01
7.11952567e-01 5.05889177e-01 1.33485064e-01 -1.95311010e-01
1.83117592e+00 -8.94674122e-01 -1.03428650e+00 3.97629261e-01
2.57820904e-01 -8.99576366e-01 7.05383539e-01 4.98059690e-01
-9.55077231e-01 1.18903436e-01 -1.03071582e+00 2.96268821e-01
-8.61691415e-01 -9.12247419e-01 8.71241927e-01 8.13860238e-01
-8.27452838e-01 4.84046161e-01 -3.95345360e-01 -8.98824155e-01
7.93270953e-03 6.03135407e-01 -5.68976164e-01 -1.76354349e-01
-9.43546712e-01 9.74472404e-01 6.62189186e-01 -3.45925033e-01
-6.20703757e-01 1.58814400e-01 -5.55916965e-01 2.80946821e-01
6.68997228e-01 -3.60232174e-01 1.42540681e+00 -1.39802027e+00
-1.38340187e+00 6.47171736e-01 4.96870987e-02 -4.36048478e-01
2.64391750e-01 4.40259218e-01 -1.22914243e+00 1.07394382e-01
1.27486721e-01 -4.30653282e-02 2.75528044e-01 -1.32509494e+00
-1.76462376e+00 -3.01933885e-01 2.57911831e-01 2.62887001e-01
-4.21714067e-01 4.85840201e-01 -3.99592072e-01 2.35208496e-01
4.36612368e-01 -5.58293045e-01 -2.15755284e-01 -4.63707596e-01
5.29203057e-01 -3.68009418e-01 9.02730823e-01 -4.11389232e-01
1.06984663e+00 -1.68282294e+00 -3.47837597e-01 3.81596088e-01
-1.56370759e-01 -1.01987189e-02 6.56463876e-02 1.02300441e+00
2.11357579e-01 1.93444610e-01 -2.86534131e-01 5.68090677e-01
4.14732665e-01 6.39556706e-01 2.18626350e-01 -1.13125131e-01
-3.23393196e-01 6.67843297e-02 -6.28995895e-01 -3.85725915e-01
-6.20925287e-03 6.08010471e-01 -5.35097063e-01 4.53254506e-02
-3.53170037e-01 3.47278506e-01 -5.94270885e-01 1.06695569e+00
6.18171930e-01 1.32962704e-01 6.62419915e-01 1.32677183e-01
-3.27795982e-01 2.68988609e-01 -1.58268440e+00 1.48373270e+00
-6.17215812e-01 -2.58250743e-01 3.34468752e-01 -1.16899133e+00
1.07703674e+00 1.08514202e+00 9.64211464e-01 -5.54890275e-01
3.38139057e-01 8.78867507e-01 3.53844076e-01 -8.61123860e-01
-3.30873914e-02 -3.22751909e-01 1.24028668e-01 5.26000321e-01
-2.94323135e-02 -8.94621834e-02 7.63455033e-01 -2.93303579e-01
9.89332020e-01 6.04355514e-01 6.47503853e-01 -2.16070265e-01
1.09740353e+00 2.55083919e-01 6.77577496e-01 -1.26398704e-03
-1.35264099e-01 -1.28273845e-01 1.85491577e-01 -5.15875876e-01
-1.22133482e+00 -8.16750288e-01 4.71140668e-02 9.78168070e-01
1.96756303e-01 -4.90342587e-01 -6.63448274e-01 -4.34106410e-01
-1.24920309e-01 6.42768621e-01 2.96771586e-01 4.34155971e-01
-4.35328603e-01 -3.93466979e-01 -1.37701228e-01 2.15111650e-03
6.53817058e-01 -1.18405783e+00 -6.78955197e-01 6.86970949e-01
-1.19019538e-01 -1.39384866e+00 2.40019128e-01 -2.13971213e-01
-8.73712718e-01 -1.02022171e+00 -1.96296647e-01 -8.75143111e-01
6.63190305e-01 5.09811342e-01 6.95982814e-01 4.36562300e-01
-1.15647227e-01 3.17630857e-01 -9.90160346e-01 -5.76830745e-01
-4.79248673e-01 1.23378132e-02 -5.88365197e-02 1.54358119e-01
8.68788421e-01 -1.08393145e+00 -5.11276186e-01 4.94593740e-01
-1.21426010e+00 4.42900956e-02 1.29989743e-01 5.94538860e-02
1.41421169e-01 8.44981909e-01 8.16210270e-01 -8.63523722e-01
6.04147136e-01 -5.33342361e-01 -6.54818892e-01 2.98454296e-02
-9.29955661e-01 -9.61132124e-02 5.92790604e-01 1.86768807e-02
-1.26269150e+00 -3.46033424e-02 -5.63233674e-01 6.38803661e-01
-5.32669961e-01 3.12336415e-01 -6.95224106e-01 -7.81431422e-02
5.21554006e-03 2.02679113e-02 8.19000527e-02 -9.41129923e-01
1.11494742e-01 1.19893098e+00 -1.01382022e-04 -6.95708513e-01
6.54357016e-01 7.24245310e-01 -1.21086985e-01 -2.73956001e-01
7.34519735e-02 -7.07352102e-01 -1.45224333e-01 -2.79293031e-01
7.21019387e-01 -1.53419301e-01 -1.01392663e+00 -2.15273872e-01
-7.58932769e-01 2.77856797e-01 1.15657598e-01 1.88710153e-01
-5.17992854e-01 1.29304498e-01 -9.23594087e-03 -1.08263278e+00
-3.65501165e-01 -1.27926815e+00 3.91068965e-01 3.16268772e-01
-7.61888102e-02 -6.97111785e-01 -3.54107082e-01 4.82812762e-01
6.74460649e-01 4.48440313e-02 7.82254457e-01 -1.27133131e+00
-5.68232656e-01 -6.06093287e-01 -5.68626598e-02 2.53846765e-01
5.97302735e-01 -4.61082906e-02 -4.14988458e-01 2.63884723e-01
1.60186693e-01 7.31193900e-01 -2.24768281e-01 -3.01206440e-01
6.80544436e-01 -2.00793475e-01 -4.10938174e-01 9.70172212e-02
2.07805657e+00 1.09390140e+00 6.25223696e-01 9.05618846e-01
2.35354856e-01 1.21130288e+00 8.47266257e-01 4.15776402e-01
3.96240354e-01 8.30511689e-01 4.58894193e-01 4.78801191e-01
1.46015301e-01 1.83109716e-01 1.36408284e-01 9.61145759e-01
-5.96172273e-01 1.63489237e-01 -1.12187457e+00 5.03921986e-01
-2.10921478e+00 -1.02978551e+00 -2.12136433e-01 1.98491168e+00
5.58636546e-01 1.77039057e-02 2.52548099e-01 4.85285401e-01
7.52738714e-01 -6.44877255e-01 4.89683263e-02 -7.34952927e-01
2.01906353e-01 2.03453988e-01 3.19487065e-01 6.52697325e-01
-3.25370818e-01 8.94239068e-01 5.25251579e+00 2.48467147e-01
-7.36276805e-01 6.81931615e-01 -3.08782130e-01 2.68202037e-01
-5.45089483e-01 6.25723720e-01 -9.38294888e-01 5.11231124e-01
1.05933690e+00 -7.17754960e-01 7.26418614e-01 9.99564588e-01
5.06415009e-01 -8.12527537e-02 -5.99120975e-01 7.69395351e-01
-7.09366351e-02 -1.26924026e+00 1.90784842e-01 7.23642483e-02
6.32940352e-01 -2.97991455e-01 -9.13702965e-01 -7.65920058e-02
4.26889271e-01 -4.50686127e-01 3.91442209e-01 2.91922510e-01
1.71588048e-01 -9.44560528e-01 9.37386632e-01 2.44343951e-01
-1.35053205e+00 -4.27507639e-01 -1.09765962e-01 -8.38129520e-02
6.90647185e-01 1.73690110e-01 -6.30415261e-01 8.04165363e-01
1.00360441e+00 -8.42011645e-02 1.69390161e-02 1.27946389e+00
8.04834664e-02 -5.13001457e-02 -2.30265722e-01 -2.40017816e-01
1.59116238e-01 -9.33990240e-01 3.72190088e-01 7.97083199e-01
6.16377532e-01 1.23767719e-01 1.66396171e-01 2.31870651e-01
3.57230276e-01 8.95115495e-01 -4.60879803e-01 -2.17922211e-01
4.31801766e-01 1.26504898e+00 -8.70359123e-01 -6.60606921e-01
-7.78076112e-01 8.65026355e-01 -4.02605742e-01 2.34938696e-01
-6.93893611e-01 -6.11606181e-01 8.20954442e-01 5.81729949e-01
1.43325135e-01 -1.73544839e-01 9.97895673e-02 -9.60403621e-01
-1.68151841e-01 -1.23941827e+00 3.65691841e-01 -6.09461486e-01
-8.65278840e-01 7.16553330e-01 1.51752383e-02 -1.19178486e+00
-1.67313546e-01 -5.57724237e-01 -1.41548961e-01 5.53763807e-01
-1.46211171e+00 -1.07914639e+00 -4.24337268e-01 3.83175522e-01
7.67237723e-01 -4.15370375e-01 1.16385543e+00 7.28728831e-01
6.42707432e-03 -4.10144925e-01 -1.49375647e-01 -4.45162356e-01
4.49295342e-01 -9.31814969e-01 -1.69359490e-01 7.12929189e-01
-1.84771389e-01 5.83406925e-01 1.03119826e+00 -6.37919605e-01
-1.27779305e+00 -2.55466759e-01 1.27441597e+00 1.69770733e-01
8.11371207e-01 -5.62694930e-02 -6.86862648e-01 6.30755961e-01
3.94414306e-01 -4.68596458e-01 9.71495330e-01 -2.78177172e-01
-9.23083723e-02 -7.43644297e-01 -1.52280903e+00 9.14213538e-01
1.14778590e+00 6.46241307e-02 -4.52502012e-01 3.16946924e-01
6.52658701e-01 2.39158392e-01 -9.90215778e-01 2.58776546e-01
7.58291900e-01 -1.14398074e+00 4.86623526e-01 -4.91293103e-01
-8.70188773e-02 -8.06170166e-01 -4.85969365e-01 -6.32965863e-01
6.29312918e-02 -6.65648997e-01 3.76976967e-01 1.47374642e+00
3.06948096e-01 -1.24764717e+00 5.53225040e-01 9.30302143e-01
-2.20145375e-01 -1.71556339e-01 -6.21189237e-01 -7.76853383e-01
-6.42218649e-01 -5.91472208e-01 1.42577243e+00 7.63117671e-01
3.86530697e-01 8.05881768e-02 2.60262966e-01 -3.10204271e-02
5.16048133e-01 1.27523601e-01 6.10902727e-01 -1.51175857e+00
-2.85812169e-01 -6.30429089e-01 -5.91264963e-01 -2.53267400e-02
-2.14941174e-01 -8.30267787e-01 -3.69388610e-01 -2.26838779e+00
-2.48251855e-01 -2.75263309e-01 -1.64778396e-01 3.48261118e-01
5.97658813e-01 5.76804541e-02 1.70093060e-01 2.78138340e-01
-2.61153936e-01 -2.80805081e-01 1.07240021e+00 3.29699636e-01
2.09790077e-02 1.25024945e-01 -7.69107521e-01 7.04110086e-01
1.15347934e+00 -4.87215340e-01 -5.39981902e-01 4.26189713e-02
3.79393131e-01 -2.07472742e-01 -5.35601020e-01 -9.53718364e-01
1.55961812e-01 -5.64969659e-01 -5.16294897e-01 2.02129371e-02
1.31984651e-01 -2.00472713e+00 9.22564626e-01 2.99929500e-01
1.24257460e-01 5.49725518e-02 -3.67994368e-01 1.05940349e-01
-3.09567958e-01 -1.03040051e+00 5.52909136e-01 -6.69011474e-01
-1.05514038e+00 1.32505313e-01 -6.04912758e-01 -8.59964848e-01
1.31790078e+00 -5.05853474e-01 2.30530590e-01 -3.58455122e-01
-9.66759682e-01 1.46529600e-01 6.87826991e-01 5.38898587e-01
1.28720179e-01 -1.20060587e+00 -2.33657792e-01 -1.85147867e-01
3.31061006e-01 -6.53957784e-01 -1.39297932e-01 5.15188813e-01
-1.01581097e+00 3.69543016e-01 -7.19864130e-01 9.10479575e-02
-1.43090212e+00 6.88710511e-01 4.17270780e-01 1.69815615e-01
-6.51295125e-01 9.99816731e-02 -4.06027436e-01 -3.10559839e-01
1.30491570e-01 1.68671086e-01 -6.73433483e-01 -1.49054304e-01
7.60089934e-01 2.55254596e-01 4.69868332e-02 -7.35790312e-01
-4.80593354e-01 5.63334048e-01 2.32113302e-01 -6.81946516e-01
1.50108325e+00 -4.58673149e-01 -7.31809855e-01 2.19309449e-01
7.80735612e-01 -1.37845045e-02 -5.38116634e-01 2.04433069e-01
5.64058542e-01 -5.12578666e-01 -3.59042495e-01 -7.70187736e-01
-7.96966553e-01 3.21155429e-01 5.82239091e-01 9.14194405e-01
1.30327010e+00 -1.92383885e-01 8.17439973e-01 1.12193339e-01
9.36217368e-01 -1.58402014e+00 -6.37250900e-01 -2.16809530e-02
6.47078574e-01 -8.89697433e-01 -2.05062032e-01 -9.25866127e-01
-5.14282048e-01 1.39455950e+00 2.96864212e-01 -6.02940358e-02
7.70115674e-01 5.90748370e-01 1.83706716e-01 -2.79662877e-01
-7.00628638e-01 -8.07669699e-01 -3.56524855e-01 8.11553776e-01
4.83706236e-01 9.22744498e-02 -1.45707190e+00 7.88119793e-01
-5.57731800e-02 6.48996532e-02 5.59796810e-01 1.26239479e+00
-1.03639030e+00 -2.09794545e+00 -4.82038736e-01 1.04398571e-01
-7.37266541e-01 2.47409880e-01 7.29595404e-03 5.30484438e-01
4.82486278e-01 1.19018745e+00 -6.65067062e-02 2.93410812e-02
5.34342766e-01 6.16520762e-01 2.75602818e-01 -7.08192348e-01
-5.09277761e-01 3.72036994e-01 6.72538996e-01 -3.33727539e-01
-6.56939864e-01 -5.29041052e-01 -1.95939970e+00 -3.04677010e-01
-1.92969710e-01 6.80122495e-01 1.48978925e+00 9.89440024e-01
1.90677211e-01 4.99948174e-01 2.85963774e-01 -1.81364164e-01
-1.18856438e-01 -5.29580176e-01 -7.13497758e-01 6.35872960e-01
-6.08266175e-01 -6.11019731e-01 -3.47230226e-01 4.18865114e-01]
|
[8.689482688903809, 7.021985054016113]
|
1ebc3c55-317f-417b-ae69-a548c689c8ac
|
mugs-a-multi-granular-self-supervised
|
2203.14415
| null |
https://arxiv.org/abs/2203.14415v1
|
https://arxiv.org/pdf/2203.14415v1.pdf
|
Mugs: A Multi-Granular Self-Supervised Learning Framework
|
In self-supervised learning, multi-granular features are heavily desired though rarely investigated, as different downstream tasks (e.g., general and fine-grained classification) often require different or multi-granular features, e.g.~fine- or coarse-grained one or their mixture. In this work, for the first time, we propose an effective MUlti-Granular Self-supervised learning (Mugs) framework to explicitly learn multi-granular visual features. Mugs has three complementary granular supervisions: 1) an instance discrimination supervision (IDS), 2) a novel local-group discrimination supervision (LGDS), and 3) a group discrimination supervision (GDS). IDS distinguishes different instances to learn instance-level fine-grained features. LGDS aggregates features of an image and its neighbors into a local-group feature, and pulls local-group features from different crops of the same image together and push them away for others. It provides complementary instance supervision to IDS via an extra alignment on local neighbors, and scatters different local-groups separately to increase discriminability. Accordingly, it helps learn high-level fine-grained features at a local-group level. Finally, to prevent similar local-groups from being scattered randomly or far away, GDS brings similar samples close and thus pulls similar local-groups together, capturing coarse-grained features at a (semantic) group level. Consequently, Mugs can capture three granular features that often enjoy higher generality on diverse downstream tasks over single-granular features, e.g.~instance-level fine-grained features in contrastive learning. By only pretraining on ImageNet-1K, Mugs sets new SoTA linear probing accuracy 82.1$\%$ on ImageNet-1K and improves previous SoTA by $1.1\%$. It also surpasses SoTAs on other tasks, e.g. transfer learning, detection and segmentation.
|
['Shuicheng Yan', 'Teck Khim Ng', 'Weihao Yu', 'Chenyang Si', 'Yichen Zhou', 'Pan Zhou']
|
2022-03-27
| null | null | null | null |
['self-supervised-image-classification']
|
['computer-vision']
|
[ 2.31635794e-01 5.16548567e-02 -5.94311118e-01 -6.00131691e-01
-9.31080639e-01 -5.28904617e-01 4.34783548e-01 3.93079102e-01
-2.26555154e-01 6.55245125e-01 -2.26754010e-01 -2.44679347e-01
-1.88252911e-01 -1.19447744e+00 -9.00031090e-01 -9.01413321e-01
-1.06537797e-01 3.37221533e-01 5.14916003e-01 -7.50322491e-02
-7.74052972e-03 4.86141175e-01 -1.89670837e+00 6.53361380e-01
9.49262142e-01 1.30513835e+00 4.92663860e-01 3.81819397e-01
-3.11568797e-01 3.19528759e-01 -6.32780552e-01 1.26895055e-01
1.13098845e-01 -2.88039297e-01 -8.24193537e-01 3.05969387e-01
7.25472569e-01 -3.54353450e-02 3.01923066e-01 1.27829766e+00
1.00306943e-01 -2.08684534e-01 9.01990950e-01 -1.19285893e+00
-6.61166608e-01 5.73657155e-01 -9.87736344e-01 1.86120406e-01
-9.74957794e-02 2.44573668e-01 1.20691466e+00 -5.66185534e-01
2.35356554e-01 1.32021344e+00 5.67664564e-01 3.43918949e-01
-1.33231044e+00 -8.66202116e-01 6.74048543e-01 -1.48186147e-01
-1.16424239e+00 -4.39044163e-02 5.61525702e-01 -3.69614899e-01
7.04249978e-01 2.94296026e-01 4.20550346e-01 7.98746824e-01
1.31469995e-01 1.08008838e+00 1.60681248e+00 -3.46190453e-01
2.06379771e-01 -2.76164655e-02 4.08529311e-01 1.00686049e+00
2.65142113e-01 5.62112294e-02 -3.12189966e-01 7.58924857e-02
9.54643488e-01 3.41645509e-01 4.56261486e-02 -1.14019096e-01
-1.17776442e+00 9.35526192e-01 9.30586159e-01 5.74084401e-01
-1.53550759e-01 1.00951903e-01 2.10081965e-01 5.68433285e-01
5.64533174e-01 3.57286990e-01 -8.49281669e-01 2.50277638e-01
-7.92260468e-01 -7.19676539e-02 5.15426040e-01 9.93146420e-01
1.71703041e+00 -1.72608271e-01 -2.46756762e-01 9.32986736e-01
1.12255588e-01 5.53008258e-01 6.95374608e-01 -5.73348701e-01
5.17603338e-01 9.27669704e-01 -2.92039782e-01 -8.38053703e-01
-4.20274585e-01 -5.92570603e-01 -1.12007916e+00 2.61932969e-01
4.12125468e-01 -5.03801256e-02 -1.34705961e+00 1.89047873e+00
1.46574810e-01 6.23203330e-02 -2.62800694e-01 6.52502298e-01
9.03059483e-01 6.87571526e-01 2.32404962e-01 -4.83323410e-02
1.40598834e+00 -1.09532118e+00 -7.62941092e-02 -5.93707323e-01
7.06450284e-01 -4.00604337e-01 1.33055997e+00 1.09760597e-01
-8.08294356e-01 -9.25924540e-01 -9.32782948e-01 2.38628671e-01
-7.39856541e-01 2.47228935e-01 8.77942145e-01 4.19265866e-01
-1.07149327e+00 5.86482942e-01 -7.33797550e-01 -3.08661193e-01
5.51319540e-01 4.00432378e-01 -4.45338756e-01 -2.41973802e-01
-1.02740276e+00 2.85841674e-01 3.46764892e-01 -2.43361533e-01
-7.34819889e-01 -5.01201391e-01 -1.05149114e+00 1.51450634e-01
3.71867567e-01 -5.08819818e-01 8.37361097e-01 -9.46596801e-01
-1.35799909e+00 1.10463011e+00 -9.90385413e-02 -2.09849656e-01
-1.34951174e-02 1.48311421e-01 -3.60501856e-01 -5.45813842e-03
6.64331675e-01 9.04358149e-01 1.18015802e+00 -1.28703392e+00
-1.15418327e+00 -4.78159219e-01 7.85428584e-02 6.52724877e-02
-3.02597851e-01 -3.24261665e-01 -2.77442724e-01 -7.94618547e-01
3.78916174e-01 -7.50654340e-01 -1.99671984e-01 -2.62290746e-01
-4.81415659e-01 -4.02559817e-01 7.58623481e-01 -7.51710907e-02
1.21425712e+00 -2.31887650e+00 -1.80860963e-02 2.97455609e-01
4.06748086e-01 1.92093670e-01 -4.00847852e-01 4.17380258e-02
-1.19302519e-01 1.62521839e-01 -2.82217383e-01 -1.08413257e-01
-2.96785533e-02 3.26969206e-01 -2.92587653e-02 1.92060247e-01
6.33086920e-01 1.21837068e+00 -9.98374581e-01 -3.20123762e-01
7.25885034e-02 -2.00399652e-01 -4.19184357e-01 1.20406061e-01
-2.06721187e-01 1.95467904e-01 -7.75784075e-01 9.27889168e-01
5.51401675e-01 -6.36600375e-01 -8.14218149e-02 -2.92666852e-01
-1.00638658e-01 -7.67030343e-02 -1.08975518e+00 1.39400065e+00
-5.69211304e-01 1.12648882e-01 6.45476067e-03 -1.43389332e+00
1.06039000e+00 -1.33320346e-01 1.23103403e-01 -7.16207087e-01
-8.14718679e-02 2.35207886e-01 -1.55989408e-01 -7.51498863e-02
1.84293598e-01 -1.47557840e-01 -6.47381306e-01 3.43442470e-01
2.42362276e-01 -4.02350305e-03 -3.37916613e-02 -7.28286803e-02
1.15620255e+00 -9.75376591e-02 4.56137478e-01 -4.54618961e-01
2.65209138e-01 -1.45609230e-02 6.65800631e-01 1.01974607e+00
-2.12445229e-01 4.86338466e-01 4.14121181e-01 -3.30013335e-01
-4.95118320e-01 -1.13102412e+00 -2.05307633e-01 1.73796010e+00
5.47911763e-01 -1.91339493e-01 -6.75325513e-01 -1.13330460e+00
3.99979353e-01 -5.36912978e-02 -9.38363314e-01 -2.06921503e-01
-3.37108672e-01 -9.35772777e-01 4.82368708e-01 7.04552412e-01
7.59179294e-01 -1.22176659e+00 -2.66700536e-01 1.68413952e-01
1.01137280e-01 -6.75693870e-01 -2.90434629e-01 9.68472540e-01
-8.43269944e-01 -8.87858927e-01 -5.73549509e-01 -9.73511219e-01
7.65527308e-01 4.39020783e-01 1.33195543e+00 4.21080180e-02
-3.54631186e-01 -1.52375355e-01 -5.72432101e-01 -1.31853044e-01
-9.70875844e-02 3.96276265e-01 -1.11828201e-01 8.13265145e-02
3.42619210e-01 -4.68952328e-01 -4.74845529e-01 6.54899001e-01
-7.65083969e-01 -5.08668907e-02 1.06466961e+00 1.22093284e+00
8.95533562e-01 2.16991439e-01 7.92339623e-01 -1.17915142e+00
2.32764423e-01 -6.08267248e-01 -3.10815454e-01 3.02177131e-01
-4.10618991e-01 3.70170809e-02 8.30268085e-01 -5.56451559e-01
-7.42979527e-01 -2.41257146e-01 4.61277664e-02 -2.28498921e-01
-7.95049667e-01 4.05829102e-01 -4.85172391e-01 -2.17677094e-02
7.92910933e-01 1.16786644e-01 -3.09370279e-01 -5.43446779e-01
3.29725027e-01 6.56715751e-01 3.21919799e-01 -8.66889179e-01
8.47875595e-01 3.29931349e-01 -3.45127374e-01 -5.61295569e-01
-1.18124878e+00 -5.73879898e-01 -5.53144693e-01 3.10476243e-01
8.15278530e-01 -9.09897983e-01 -3.47765625e-01 8.48121762e-01
-3.37682188e-01 -6.48878038e-01 -5.40881693e-01 1.81121364e-01
-3.56692851e-01 -4.41958606e-02 -8.10012162e-01 -2.30558440e-01
5.59352376e-02 -1.19444323e+00 1.51896083e+00 4.96450752e-01
1.36542469e-01 -8.38028669e-01 -4.36006606e-01 1.34868443e-01
1.96694180e-01 3.07664186e-01 9.49563980e-01 -4.86894727e-01
-5.06319106e-01 -1.16669402e-01 -6.86649442e-01 4.18929875e-01
6.20327175e-01 -1.14267990e-01 -9.69295025e-01 -5.59405684e-01
-4.02092755e-01 -6.29735351e-01 1.21711946e+00 4.95909154e-01
1.42624390e+00 -1.18725166e-01 -6.88027978e-01 8.47038209e-01
1.28352857e+00 7.39329010e-02 1.47612035e-01 4.68675345e-01
7.70115852e-01 3.80704999e-01 1.02429736e+00 2.35015437e-01
2.74644077e-01 3.85061055e-01 5.19518971e-01 -4.77152467e-01
-1.10585049e-01 -1.05021715e-01 2.37514898e-01 5.95499098e-01
-3.25999595e-02 -2.10722778e-02 -6.14555895e-01 6.18997812e-01
-1.86192799e+00 -6.96973860e-01 1.93924099e-01 1.72246909e+00
1.03002667e+00 3.48976195e-01 1.95639938e-01 5.62218279e-02
9.29483235e-01 4.04077470e-01 -7.00566232e-01 -3.38584989e-01
-1.76900312e-01 4.86082435e-01 6.06285632e-01 1.96897119e-01
-1.41741562e+00 1.23918307e+00 4.78542709e+00 1.36061704e+00
-1.28577721e+00 -6.13243878e-02 1.01836264e+00 3.66939098e-01
-3.00640792e-01 -2.57899106e-01 -1.21945834e+00 5.91064811e-01
4.15405422e-01 2.85415202e-01 1.75587371e-01 1.00045848e+00
-3.85591269e-01 -1.41378328e-01 -1.07409060e+00 7.31856704e-01
-3.39057267e-01 -1.29363704e+00 7.41869658e-02 1.18029147e-01
8.00671816e-01 2.27624610e-01 1.69556215e-01 5.83545804e-01
6.63149953e-01 -1.07911074e+00 5.49800217e-01 1.11832395e-01
1.16416276e+00 -6.52134299e-01 6.59116268e-01 6.04801416e-01
-1.57265639e+00 -1.60274252e-01 -5.21975577e-01 -7.52136335e-02
-3.83991718e-01 8.82716298e-01 -4.35155362e-01 6.25970840e-01
1.04233325e+00 8.45697343e-01 -5.40142298e-01 5.53155243e-01
-1.75918773e-01 6.89622462e-01 -3.61528963e-01 1.54185176e-01
7.70652533e-01 -3.47251184e-02 -1.12396451e-02 1.37285149e+00
1.38250008e-01 4.56378870e-02 7.16923118e-01 5.68791389e-01
-4.59817611e-02 -2.64691740e-01 -2.92582959e-01 1.29455522e-01
4.20262754e-01 1.39374661e+00 -8.99163604e-01 -4.77406651e-01
-3.78080040e-01 9.94718552e-01 5.99385738e-01 3.39860260e-01
-6.29686832e-01 -5.18915176e-01 8.92184377e-01 6.02241084e-02
5.21890104e-01 1.48028269e-01 -1.95578784e-01 -1.19626892e+00
-1.48949921e-01 -7.63229847e-01 7.06194460e-01 -2.52270162e-01
-1.76365519e+00 8.56015086e-01 -1.07851624e-01 -1.33042610e+00
-3.43336314e-01 -6.93517387e-01 -5.04603446e-01 1.07118797e+00
-1.84946299e+00 -1.41325319e+00 -4.84746009e-01 8.73954952e-01
4.54550117e-01 -9.88967046e-02 8.74636173e-01 -3.08820773e-02
-5.52121222e-01 9.05162632e-01 5.65740876e-02 1.91386476e-01
8.12422514e-01 -1.56917048e+00 3.75562251e-01 5.21422446e-01
1.32472187e-01 5.76993525e-01 1.04405299e-01 -4.97531563e-01
-8.24593008e-01 -1.41486788e+00 5.49891651e-01 -1.10067271e-01
8.43981683e-01 -3.18432003e-01 -1.09143925e+00 6.03699923e-01
-4.45751578e-01 6.60009503e-01 5.63082397e-01 3.16592634e-01
-6.50820792e-01 -4.59464669e-01 -1.36943197e+00 1.07701711e-01
1.19369268e+00 -7.03020036e-01 -5.10979652e-01 1.63198814e-01
7.81492531e-01 -2.51607418e-01 -9.97072041e-01 6.06019258e-01
2.32848495e-01 -1.01743615e+00 8.19033325e-01 -4.58357066e-01
3.92813861e-01 -5.05841851e-01 -3.18733782e-01 -1.59102881e+00
-8.45419407e-01 -1.98235646e-01 -1.63287148e-02 1.17411327e+00
3.13288480e-01 -8.86904776e-01 8.33532929e-01 -1.99736312e-01
-3.04326832e-01 -9.96824622e-01 -6.33478880e-01 -9.16636229e-01
2.31189683e-01 -1.73190489e-01 7.95750976e-01 1.15213919e+00
-3.31569225e-01 2.25352854e-01 2.74846214e-03 4.05772537e-01
6.22556210e-01 7.86372185e-01 4.33098853e-01 -1.44587731e+00
-5.23588598e-01 -6.20444775e-01 -4.62492913e-01 -1.26198936e+00
1.27655536e-01 -8.63908350e-01 9.39647853e-02 -1.22980893e+00
1.95304826e-01 -1.10855532e+00 -7.09790766e-01 1.02289629e+00
-4.92376089e-01 6.95440173e-01 -1.65888462e-02 2.48558983e-01
-5.99171221e-01 9.77008566e-02 1.39785361e+00 -2.81212866e-01
-2.07622856e-01 1.71782061e-01 -1.02776015e+00 7.24574447e-01
6.97902739e-01 -2.62879312e-01 -3.14374596e-01 -2.48133093e-01
-2.23151132e-01 -1.15591273e-01 3.87275755e-01 -8.73933256e-01
-1.02762960e-01 -3.13034564e-01 6.38308287e-01 -4.60093290e-01
-8.86600167e-02 -5.27931333e-01 -2.93616533e-01 3.34955037e-01
-8.65443870e-02 -3.58840615e-01 1.91555217e-01 5.60078263e-01
-5.35597742e-01 -9.34238546e-03 9.87197459e-01 -3.79041433e-01
-1.15294647e+00 6.04152441e-01 2.14715842e-02 7.86413327e-02
9.02434289e-01 -5.68784595e-01 -5.69522142e-01 1.74045250e-01
-9.40062106e-01 1.90126687e-01 4.96325552e-01 3.79377633e-01
2.50626445e-01 -1.21675169e+00 -5.27698636e-01 5.59583604e-01
4.87441778e-01 4.72971976e-01 9.02251229e-02 5.58286905e-01
9.44713131e-02 1.60827413e-01 -3.66743118e-01 -1.02530301e+00
-1.05385792e+00 3.74995828e-01 8.05757865e-02 -5.63871026e-01
-5.21778524e-01 1.32491970e+00 9.70308423e-01 -6.09112322e-01
-2.23345533e-02 -7.51466393e-01 -1.53205007e-01 2.18764767e-01
4.65779990e-01 -7.30839521e-02 8.57564658e-02 -4.34535086e-01
-4.34339046e-01 8.63927245e-01 -2.52353162e-01 5.76545537e-01
1.39388454e+00 -7.66808093e-02 -1.26360804e-01 5.51955104e-01
1.29997909e+00 -1.21917062e-01 -1.60355306e+00 -4.40658063e-01
1.15919024e-01 -3.51275146e-01 -1.04950458e-01 -1.00733125e+00
-1.13395619e+00 7.91375220e-01 3.58782351e-01 5.96677840e-01
1.36352599e+00 5.62901616e-01 5.68583190e-01 1.38292894e-01
7.24807560e-01 -7.43066192e-01 2.13148803e-01 5.80429018e-01
3.82062584e-01 -1.40473008e+00 -2.99774200e-01 -6.12792253e-01
-4.08528328e-01 8.17661941e-01 8.32492828e-01 -2.59423971e-01
6.17001295e-01 4.42324400e-01 5.54947145e-02 -1.45174623e-01
-5.80365896e-01 -5.62671483e-01 4.67440486e-01 6.25330627e-01
3.14248860e-01 5.02180457e-01 2.21598983e-01 8.55748355e-01
1.86226964e-02 -1.78957909e-01 -1.33567095e-01 8.94548297e-01
-8.05217624e-01 -1.02702367e+00 -2.62202650e-01 9.43139136e-01
-1.18151560e-01 -1.99425295e-01 -9.33023077e-03 8.31355095e-01
6.49302959e-01 7.22357929e-01 3.55129331e-01 -4.28468436e-01
2.02911600e-01 -3.38520765e-01 4.54373628e-01 -1.02735901e+00
-6.56755149e-01 5.32311127e-02 -1.24446787e-01 -7.49928057e-01
-4.11550194e-01 -4.61330950e-01 -1.26880562e+00 -3.00411493e-01
-3.67738485e-01 2.10107327e-01 2.09275663e-01 1.01189899e+00
2.75590599e-01 5.96017122e-01 9.55716193e-01 -9.54548359e-01
-4.73663986e-01 -8.03662539e-01 -9.80325043e-01 4.34338152e-01
4.37730551e-01 -7.00293422e-01 -4.66953218e-01 -1.46119565e-01]
|
[9.501462936401367, 1.9890329837799072]
|
7f52fbdc-7020-4bb9-9951-07c0bfb10355
|
optimization-of-robot-trajectory-planning
|
2206.03651
| null |
https://arxiv.org/abs/2206.03651v1
|
https://arxiv.org/pdf/2206.03651v1.pdf
|
Optimization of Robot Trajectory Planning with Nature-Inspired and Hybrid Quantum Algorithms
|
We solve robot trajectory planning problems at industry-relevant scales. Our end-to-end solution integrates highly versatile random-key algorithms with model stacking and ensemble techniques, as well as path relinking for solution refinement. The core optimization module consists of a biased random-key genetic algorithm. Through a distinct separation of problem-independent and problem-dependent modules, we achieve an efficient problem representation, with a native encoding of constraints. We show that generalizations to alternative algorithmic paradigms such as simulated annealing are straightforward. We provide numerical benchmark results for industry-scale data sets. Our approach is found to consistently outperform greedy baseline results. To assess the capabilities of today's quantum hardware, we complement the classical approach with results obtained on quantum annealing hardware, using qbsolv on Amazon Braket. Finally, we show how the latter can be integrated into our larger pipeline, providing a quantum-ready hybrid solution to the problem.
|
['Helmut G. Katzgraber', 'Mauricio G. C. Resende', 'Andre Luckow', 'Philipp Ross', 'Johannes Klepsch', 'Yannick van Dijk', 'Henry Montagu', 'J. Kyle Brubaker', 'Martin J. A. Schuetz']
|
2022-06-08
| null | null | null | null |
['trajectory-planning']
|
['robots']
|
[ 2.62366086e-01 1.26371801e-01 -1.35768220e-01 -1.50324091e-01
-1.34587455e+00 -8.78836036e-01 5.89007318e-01 -2.81603099e-03
-5.66414058e-01 7.39733160e-01 -6.23587593e-02 -6.70674801e-01
-3.86603564e-01 -9.48454678e-01 -8.99397671e-01 -8.91490996e-01
-2.06430539e-01 1.03837061e+00 -1.73961371e-02 -9.38131034e-01
9.32539821e-01 3.34118366e-01 -1.41110790e+00 -2.00693771e-01
6.80528641e-01 6.31360769e-01 1.22686848e-01 8.19811404e-01
3.86693001e-01 2.51634836e-01 2.79242154e-02 -2.06279963e-01
6.67218924e-01 -1.55727595e-01 -1.14826715e+00 -1.98883548e-01
1.18212536e-01 1.56642571e-01 -5.89213729e-01 9.63947773e-01
4.91766244e-01 3.56132090e-01 3.23882878e-01 -1.11474252e+00
-1.16603337e-01 5.54397106e-01 -1.00804932e-01 -1.18604489e-01
3.78548980e-01 7.08790302e-01 1.06522334e+00 -4.06397492e-01
1.02666855e+00 1.10741925e+00 6.38922751e-01 3.32826614e-01
-1.52782679e+00 -4.38199520e-01 -3.90471727e-01 6.14605904e-01
-1.39892960e+00 -5.88532507e-01 2.23224103e-01 1.00676371e-02
1.58630848e+00 1.66883424e-01 5.85489690e-01 8.19020450e-01
6.79575264e-01 2.12608054e-01 1.01049888e+00 -3.23690236e-01
7.38291085e-01 -3.82607490e-01 3.69637758e-01 9.35546219e-01
7.81683624e-02 7.90131629e-01 -5.15136063e-01 -5.12585700e-01
2.46990815e-01 -5.63284576e-01 2.84282565e-01 -7.30842113e-01
-1.36212361e+00 8.48021567e-01 5.83394468e-01 -4.68385011e-01
-2.85947055e-01 9.71226454e-01 2.92691499e-01 2.88670987e-01
-3.68785977e-01 8.98436904e-01 -4.87850219e-01 -5.22118092e-01
-4.01135176e-01 1.00208318e+00 8.42265725e-01 1.35541630e+00
1.17553604e+00 -4.58572060e-01 1.73407570e-01 1.34118065e-01
2.08132878e-01 4.56865370e-01 -1.09359227e-01 -1.69175863e+00
3.76129031e-01 1.91400833e-02 5.05615890e-01 -4.66459632e-01
-9.31777477e-01 -1.90741420e-01 -3.06923330e-01 3.30674022e-01
3.01557004e-01 -1.93824455e-01 -8.39011908e-01 1.58752048e+00
5.24800837e-01 1.50356233e-01 3.46453875e-01 8.02626431e-01
1.45097032e-01 6.31308019e-01 -4.25741911e-01 -2.03210235e-01
1.19716311e+00 -1.35217631e+00 -3.30722064e-01 -2.51863360e-01
1.15082896e+00 -6.49903417e-01 5.94195426e-01 9.05641198e-01
-1.27828813e+00 -9.49419066e-02 -1.31270671e+00 -3.15066665e-01
-3.30913037e-01 -4.55010414e-01 1.22999954e+00 8.28898609e-01
-1.32186353e+00 1.09458184e+00 -9.19855595e-01 -3.87707859e-01
3.50671858e-02 7.70616174e-01 -2.58128226e-01 -1.12486012e-01
-8.89097750e-01 1.23555684e+00 5.87140918e-01 -6.12311102e-02
-7.75111377e-01 -4.07886297e-01 -5.77002943e-01 -3.74452502e-01
6.88107252e-01 -1.11582351e+00 1.72661161e+00 -3.48021947e-02
-2.07555795e+00 5.02998292e-01 -1.80813119e-01 -7.77136445e-01
1.89805269e-01 1.73121795e-01 2.37690862e-02 -6.89399540e-02
-1.33594021e-01 6.47545695e-01 4.50884014e-01 -8.96112442e-01
-6.43782079e-01 -2.16432735e-01 1.53954774e-01 2.77813017e-01
4.50924605e-01 -1.54709041e-01 -4.79144305e-01 2.49108627e-01
4.90746230e-01 -1.75909770e+00 -1.11393702e+00 -7.16401696e-01
-4.12893772e-01 -3.19024511e-02 7.53763318e-02 -1.37689114e-01
1.06892109e+00 -1.69445622e+00 8.35613608e-01 4.93414670e-01
-6.67457879e-02 -5.23672104e-01 -1.17502041e-01 9.79415715e-01
3.61549966e-02 -6.96017146e-02 -6.01121366e-01 -3.41432929e-01
4.53685075e-01 3.16757977e-01 -2.76299983e-01 6.89527035e-01
-2.54482199e-02 1.12124574e+00 -1.01572418e+00 -2.02964284e-02
3.17399800e-01 -4.06899720e-01 -1.17760408e+00 -4.75126863e-01
-5.40815055e-01 3.98657024e-01 -2.85753131e-01 4.42738682e-01
6.37667060e-01 -9.11208466e-02 3.70233834e-01 7.30432943e-02
-5.06448865e-01 6.62397742e-01 -1.24659693e+00 2.58182096e+00
-3.58254313e-01 1.40231535e-01 3.75686884e-01 -6.01293981e-01
3.60990137e-01 -3.22839588e-01 4.09054786e-01 -7.84340620e-01
2.51299739e-01 5.12876570e-01 1.34354845e-01 -1.57524437e-01
1.38873518e+00 1.78805396e-01 -6.82957649e-01 4.12375897e-01
-9.80038196e-02 -9.77925599e-01 2.22720683e-01 3.96071225e-01
1.52438223e+00 5.07027984e-01 4.05549496e-01 -5.76422215e-01
9.01284590e-02 8.14072788e-01 1.33855447e-01 1.07211924e+00
-1.98938847e-01 4.56585467e-01 2.62340963e-01 -3.89408439e-01
-1.33393466e+00 -8.96491766e-01 -2.56108284e-01 9.19748545e-01
7.90016234e-01 -1.17839444e+00 -7.66608119e-01 -1.79092094e-01
5.47097577e-03 1.05378032e+00 -4.61105108e-01 -2.34876484e-01
-7.96262860e-01 -1.20256817e+00 4.46723580e-01 -2.18358889e-01
7.91548789e-02 -6.80542469e-01 -6.94163799e-01 5.05422831e-01
1.81181431e-01 -8.67805898e-01 1.20935686e-01 7.33132958e-01
-7.04614937e-01 -9.44261611e-01 3.20118725e-01 -2.31597856e-01
1.96527496e-01 3.91447246e-01 9.84664977e-01 -1.25133609e-02
-4.37047482e-01 1.05473496e-01 -2.91134953e-01 3.10197398e-02
-5.81780612e-01 3.54557991e-01 3.06769460e-01 -9.23435509e-01
-1.12020284e-01 -6.24638081e-01 -6.71542168e-01 1.26397058e-01
-2.64223725e-01 1.29266065e-02 5.15527427e-01 8.28377485e-01
5.36425710e-01 -4.09704782e-02 -6.30246848e-02 -5.67603648e-01
4.21355993e-01 -6.48659706e-01 -9.25875127e-01 -1.30552351e-01
-9.36420918e-01 6.06982470e-01 4.08880413e-01 2.21593052e-01
-6.06164634e-01 3.75853091e-01 -3.24584723e-01 2.51168281e-01
-2.86450610e-02 4.04467553e-01 3.01441818e-01 -5.05998492e-01
9.52305019e-01 1.20448768e-01 2.82612666e-02 -1.37231294e-02
1.06691420e+00 3.96157205e-01 6.55367851e-01 -1.00341332e+00
6.49115443e-01 5.09599507e-01 5.83883524e-01 -6.06496751e-01
-1.74054012e-01 -2.67583221e-01 -2.48405188e-01 2.68181831e-01
8.43309462e-01 -5.76392233e-01 -1.14026976e+00 3.65334064e-01
-1.10362625e+00 -6.24463320e-01 -1.21677704e-01 3.70591372e-01
-1.25130713e+00 3.45218122e-01 -5.28288484e-01 -7.53182292e-01
-1.06309764e-01 -1.82099378e+00 1.26543498e+00 1.61088824e-01
-1.08326063e-01 -4.27154958e-01 4.16397005e-01 4.21439230e-01
2.84022331e-01 -1.06763728e-01 8.00445616e-01 -4.46973294e-01
-1.23220623e+00 5.94506739e-03 -5.22919325e-03 -6.54168665e-01
-6.74765050e-01 8.84931833e-02 -6.05574071e-01 -3.33027065e-01
-3.21155787e-01 -6.12145841e-01 8.27960372e-01 9.53261182e-02
8.48565280e-01 7.37462118e-02 -5.66442609e-01 7.99836755e-01
1.46726644e+00 -1.40819222e-01 8.29647422e-01 8.27246070e-01
3.87731403e-01 3.33200216e-01 7.81423450e-01 3.37344170e-01
7.58416474e-01 9.95952010e-01 8.14623594e-01 6.34987116e-01
3.60626012e-01 1.78862050e-01 3.89417678e-01 7.74339139e-01
-1.38529852e-01 -3.49972807e-02 -1.22785985e+00 1.03918150e-01
-2.15540528e+00 -9.13222909e-01 -5.01027524e-01 2.07992101e+00
4.52284932e-01 2.10394904e-01 1.25918582e-01 -2.95358956e-01
1.44745320e-01 -1.37407303e-01 -7.58959472e-01 -7.16118097e-01
1.70959264e-01 5.64146280e-01 1.40921545e+00 7.96126604e-01
-9.82844591e-01 1.43414652e+00 7.43690538e+00 1.17815256e+00
-5.12418509e-01 1.75685972e-01 7.79325739e-02 -4.32363212e-01
-3.44642818e-01 5.44298410e-01 -7.26924181e-01 1.87563434e-01
1.45747888e+00 -2.21873805e-01 1.23436511e+00 7.52323449e-01
-7.54921734e-02 -4.41694945e-01 -1.10857153e+00 1.08442652e+00
-5.29255867e-01 -1.83124864e+00 -7.15417862e-01 5.16634524e-01
7.89884686e-01 7.41958976e-01 1.06201708e-01 4.51491475e-01
1.06288028e+00 -1.13572991e+00 1.09713304e+00 3.32676560e-01
3.22641253e-01 -1.08040869e+00 3.22548687e-01 3.79793882e-01
-7.96508551e-01 -3.66145223e-01 -4.51263458e-01 -4.86960173e-01
5.43483019e-01 -1.27284989e-01 -6.58274889e-01 1.30955994e+00
4.81389076e-01 4.53678578e-01 -2.74151474e-01 9.60990965e-01
-4.08863612e-02 6.64698482e-02 -7.19017625e-01 -2.05746517e-01
6.23331845e-01 -8.06818008e-01 5.79577386e-01 8.59583378e-01
3.96326810e-01 3.57664198e-01 5.78872152e-02 7.37006426e-01
2.37341180e-01 -3.00154418e-01 -6.50371492e-01 1.39707834e-01
5.73103130e-01 1.35861778e+00 -7.49839067e-01 -2.62439679e-02
2.10895725e-02 1.00868893e+00 5.54266393e-01 7.38852322e-02
-9.59196866e-01 -1.51716486e-01 7.88839698e-01 -7.33005166e-01
2.37562850e-01 -9.21001792e-01 -6.35736585e-01 -8.50446939e-01
-3.73885870e-01 -1.00785804e+00 1.34617249e-02 -5.70956290e-01
-7.67439544e-01 7.47938156e-02 -4.94195968e-02 -6.91206872e-01
-4.50093180e-01 -7.78245807e-01 -3.88647437e-01 5.98134756e-01
-1.13657260e+00 -6.23857915e-01 1.07459731e-01 3.64788510e-02
2.15391904e-01 -3.28318700e-02 1.05480945e+00 -5.97057864e-02
-7.12220609e-01 2.94036955e-01 5.98930240e-01 -9.79023099e-01
3.95785749e-01 -1.26451504e+00 1.14211369e+00 7.13370025e-01
-2.97916420e-02 9.48438227e-01 1.20158911e+00 -5.73178768e-01
-2.62630963e+00 -5.78129530e-01 1.08853728e-01 -9.26470339e-01
1.03003979e+00 -4.33765948e-01 -1.10186443e-01 6.19311750e-01
3.26446116e-01 -4.92276132e-01 3.27491492e-01 4.29628015e-01
2.89976466e-02 3.61018270e-01 -9.62983906e-01 8.73566031e-01
1.59858716e+00 -3.06479067e-01 -3.28711241e-01 6.73386276e-01
9.06827211e-01 -8.77511621e-01 -5.45273423e-01 2.49768794e-01
5.75827777e-01 -1.03153992e+00 1.15495861e+00 -6.30604506e-01
3.38716596e-01 -2.89641678e-01 -5.72724879e-01 -1.27843511e+00
-4.99352336e-01 -1.37417734e+00 -4.47536707e-02 3.76418650e-01
6.39267981e-01 -7.09473908e-01 9.28319454e-01 7.43803620e-01
-6.36586487e-01 -6.68556392e-01 -1.25946784e+00 -7.65760839e-01
4.60954696e-01 -8.22224498e-01 6.99208319e-01 5.00084102e-01
3.74678701e-01 3.18507075e-01 -3.09555531e-01 4.13579643e-01
7.64558554e-01 1.90742448e-01 1.17387426e+00 -5.20777702e-01
-4.65524375e-01 -5.75305641e-01 -3.28091621e-01 -1.03815949e+00
1.36337802e-01 -1.15894306e+00 3.01043242e-01 -1.32865393e+00
8.52828100e-02 -7.96863854e-01 1.39849856e-02 -8.00850540e-02
2.03665420e-01 6.40022755e-02 1.32923186e-01 -4.14522998e-02
-9.16865170e-01 7.36059070e-01 8.81218016e-01 1.81049019e-01
-2.27454826e-01 -5.03681898e-01 -5.14835656e-01 5.04973829e-01
9.37416494e-01 -5.21470129e-01 -2.36128181e-01 -4.16533232e-01
8.05064678e-01 1.37250230e-01 2.26213440e-01 -1.04807556e+00
6.41002476e-01 -4.11922097e-01 -5.30956388e-01 -6.28109455e-01
5.83754838e-01 -3.94305974e-01 4.15397078e-01 6.76833630e-01
4.02377695e-02 3.19555074e-01 3.22167486e-01 5.44830143e-01
6.46126807e-01 -3.51651520e-01 5.27671337e-01 -2.03849837e-01
-9.48603630e-01 1.47361279e-01 -4.78983074e-01 -2.65142351e-01
1.23702645e+00 2.53124852e-02 -7.88350999e-01 -2.20800797e-03
-4.66796905e-01 5.99538565e-01 1.12023032e+00 6.75387830e-02
1.02455877e-01 -1.08631408e+00 -4.31943715e-01 -8.41719285e-02
3.26629132e-01 4.12835833e-03 1.79556176e-01 1.06523240e+00
-1.08111000e+00 6.41653776e-01 -1.25114754e-01 -4.98222679e-01
-7.19827116e-01 8.80465329e-01 2.83288270e-01 -8.42594281e-02
-4.28126127e-01 8.79350781e-01 -4.49296027e-01 -9.61754918e-01
-2.44579598e-01 -3.85405183e-01 7.24626780e-01 -3.75671566e-01
2.41078779e-01 7.06887126e-01 1.82580307e-01 -4.47751611e-01
-5.15953243e-01 4.56763685e-01 6.66397437e-02 -6.52485728e-01
1.21463192e+00 -2.64253318e-01 -2.62573749e-01 -3.65524106e-02
8.82375002e-01 -2.23646369e-02 -9.92388546e-01 2.82500267e-01
1.92814320e-01 1.37740389e-01 1.03923663e-01 -6.66288555e-01
-6.25960976e-02 4.55930561e-01 3.73270005e-01 9.44830179e-02
4.93961453e-01 1.90174452e-03 8.41362000e-01 1.25825763e+00
1.42951882e+00 -1.33561492e+00 -3.92670333e-01 8.58458161e-01
4.22954559e-01 -1.10371315e+00 2.67094731e-01 -4.26472902e-01
-2.61815399e-01 9.61165547e-01 1.14942335e-01 -4.45500255e-01
-3.39832809e-03 2.54552454e-01 -3.19825411e-01 -4.00715321e-01
-9.94218230e-01 -5.25467277e-01 -2.98343211e-01 3.58748645e-01
-3.30329955e-01 1.10491864e-01 -3.62582713e-01 3.50327492e-01
-8.31637800e-01 -2.57664412e-01 7.24063337e-01 1.26906347e+00
-5.30595303e-01 -1.70939755e+00 -3.38840753e-01 4.31790091e-02
1.66266337e-01 -1.78409964e-01 -2.31357992e-01 7.92191029e-01
-5.87809011e-02 9.93445635e-01 -2.16109037e-01 -7.70595670e-01
1.90463528e-01 -3.88449766e-02 1.10965765e+00 -5.56736469e-01
-6.63625598e-01 -3.65147829e-01 6.27136588e-01 -1.42304456e+00
1.54110551e-01 -8.71548951e-01 -1.76421452e+00 -7.80795932e-01
-1.93283856e-01 1.46777898e-01 9.83126640e-01 8.62143576e-01
6.88482702e-01 3.66575330e-01 4.27952588e-01 -1.58128238e+00
-9.96291101e-01 -3.92607421e-01 -1.65188804e-01 -1.79252580e-01
1.88894212e-01 -6.76009417e-01 -5.01850136e-02 -6.37278259e-01]
|
[5.593486309051514, 4.781510829925537]
|
46c2ec95-14df-44d3-95cf-3d13dfaf3c4d
|
explainable-artificial-intelligence-in-2
|
2212.07058
| null |
https://arxiv.org/abs/2212.07058v1
|
https://arxiv.org/pdf/2212.07058v1.pdf
|
Explainable Artificial Intelligence in Retinal Imaging for the detection of Systemic Diseases
|
Explainable Artificial Intelligence (AI) in the form of an interpretable and semiautomatic approach to stage grading ocular pathologies such as Diabetic retinopathy, Hypertensive retinopathy, and other retinopathies on the backdrop of major systemic diseases. The experimental study aims to evaluate an explainable staged grading process without using deep Convolutional Neural Networks (CNNs) directly. Many current CNN-based deep neural networks used for diagnosing retinal disorders might have appreciable performance but fail to pinpoint the basis driving their decisions. To improve these decisions' transparency, we have proposed a clinician-in-the-loop assisted intelligent workflow that performs a retinal vascular assessment on the fundus images to derive quantifiable and descriptive parameters. The retinal vessel parameters meta-data serve as hyper-parameters for better interpretation and explainability of decisions. The semiautomatic methodology aims to have a federated approach to AI in healthcare applications with more inputs and interpretations from clinicians. The baseline process involved in the machine learning pipeline through image processing techniques for optic disc detection, vessel segmentation, and arteriole/venule identification.
|
['Dr Prakash Kamaraj', 'Meghna Kulkarni', 'Rajkumar Vaghashiya', 'Ayushi Raj Bhatt']
|
2022-12-14
| null | null | null | null |
['optic-disc-detection']
|
['medical']
|
[ 2.54188031e-01 9.80116367e-01 1.41039446e-01 -8.04410398e-01
5.97256906e-02 -3.71565998e-01 3.32101911e-01 2.24253666e-02
-1.99447781e-01 5.39209366e-01 2.98037261e-01 -8.36693883e-01
-5.21637380e-01 -5.19117951e-01 -3.79536897e-02 -4.80442643e-01
3.06972861e-01 8.18457723e-01 -1.28963590e-01 6.84978664e-02
5.29949129e-01 9.66014385e-01 -1.52930117e+00 6.17887199e-01
1.34455788e+00 1.07351637e+00 -4.68162745e-01 1.32917035e+00
-2.75885016e-01 1.17362607e+00 -4.08499449e-01 -3.34387273e-01
5.74103832e-01 -6.63988888e-01 -1.06238353e+00 3.16341937e-01
8.55058968e-01 -5.28216839e-01 6.89608753e-02 9.79505658e-01
4.84965235e-01 -6.00494683e-01 7.72128105e-01 -7.89549232e-01
-7.34870791e-01 2.94116199e-01 -4.87986952e-02 2.68841565e-01
-2.60866523e-01 8.68881106e-01 2.99768507e-01 -2.08787724e-01
3.80848259e-01 1.07824552e+00 5.74098051e-01 6.05020404e-01
-8.05657566e-01 1.19759656e-01 -4.90632236e-01 1.25854209e-01
-6.72745705e-01 -4.86755162e-01 -9.91409346e-02 -1.00699699e+00
8.21901143e-01 4.69787240e-01 1.05948412e+00 3.71001214e-02
1.41646370e-01 6.62536025e-02 1.35838246e+00 -5.30277729e-01
1.29583299e-01 2.15527430e-01 3.41934621e-01 1.10537732e+00
6.48505747e-01 2.56127268e-01 2.92234328e-02 2.21491233e-01
9.81406569e-01 -1.97130367e-01 -3.98196541e-02 2.91354209e-01
-8.62550795e-01 5.98143399e-01 5.74861526e-01 -1.70471862e-01
-5.85229218e-01 3.36514339e-02 2.90677190e-01 2.55642682e-01
1.96892515e-01 1.08743131e+00 -4.87160504e-01 6.71940669e-02
-9.29416120e-01 -8.14569145e-02 5.94507992e-01 4.42048013e-01
4.68017340e-01 -4.46200073e-02 -5.44347644e-01 2.62897879e-01
4.77069736e-01 5.78312464e-02 3.00710559e-01 -1.41817570e+00
-2.90606618e-01 1.54975688e+00 1.13723740e-01 -4.26532179e-01
-7.27651298e-01 -4.11000758e-01 -8.75790834e-01 1.14506662e+00
8.68798792e-01 -4.08546060e-01 -1.53447878e+00 5.47242582e-01
2.76876688e-01 -3.51821691e-01 -2.15792313e-01 1.41921473e+00
1.03424573e+00 -1.85523510e-01 1.24173323e-02 7.59046078e-02
1.48205173e+00 -1.06040418e+00 -5.11285245e-01 2.23492309e-01
8.87950838e-01 -7.12725580e-01 8.25807393e-01 6.95941746e-01
-1.24129796e+00 -4.58641112e-01 -8.58970404e-01 -6.21310353e-01
-3.83384496e-01 5.88115215e-01 9.02159095e-01 6.38643861e-01
-1.38592649e+00 5.76540768e-01 -6.45920932e-01 -3.76168370e-01
1.04364765e+00 6.44744813e-01 -3.12752187e-01 2.91219324e-01
-4.32907760e-01 1.33148575e+00 1.04082920e-01 5.02104878e-01
-3.05994481e-01 -8.00307095e-01 -3.64706308e-01 3.71063240e-02
-1.80124477e-01 -1.49180102e+00 1.28351438e+00 -1.15431845e+00
-1.55109584e+00 1.37813509e+00 -3.00229162e-01 -8.45072746e-01
9.13333118e-01 -4.10895944e-02 -6.72573820e-02 4.99645472e-01
-3.43704730e-01 8.18675220e-01 6.70937657e-01 -6.67778671e-01
-9.92615521e-01 -4.31605250e-01 5.51028512e-02 -2.53804654e-01
1.60562277e-01 7.74830282e-02 1.04075447e-01 9.36105475e-02
-8.88166875e-02 -7.26527691e-01 -5.11800766e-01 8.13014388e-01
-7.29249835e-01 -1.41341537e-01 1.62724972e-01 -6.16377771e-01
1.03240263e+00 -1.45621169e+00 -3.44248265e-01 2.06074044e-01
1.28674626e+00 1.01564801e+00 2.33537510e-01 -3.84481728e-01
-2.33363226e-01 3.74920607e-01 1.21456161e-01 -2.69913524e-02
-2.35831112e-01 4.07946147e-02 3.90582561e-01 3.54327500e-01
6.90930963e-01 1.11088228e+00 -5.79295516e-01 -4.98318106e-01
4.96121556e-01 4.44930255e-01 -3.75147551e-01 7.60559365e-02
-2.06000447e-01 6.88864887e-01 -3.25381398e-01 5.59644461e-01
4.41177428e-01 -5.02547801e-01 -2.51485974e-01 -4.33342814e-01
-4.15660530e-01 1.48120478e-01 -6.23449266e-01 9.81060684e-01
-9.35043022e-02 1.27229893e+00 -2.48254329e-01 -5.25776386e-01
8.88780415e-01 3.10188472e-01 1.68831348e-01 -4.36513215e-01
7.02226460e-01 3.54434490e-01 7.08580494e-01 -1.08514321e+00
-1.60210431e-02 1.46333531e-01 1.07939303e+00 2.44069144e-01
-1.77348122e-01 5.29946536e-02 2.65040159e-01 -1.99963272e-01
9.59407032e-01 1.10353023e-01 4.88892555e-01 -1.09994642e-01
6.44049406e-01 5.78840256e-01 1.62241593e-01 6.25833273e-01
-5.49752951e-01 8.25968862e-01 9.54934895e-01 -1.37758982e+00
-1.45953763e+00 -4.97919649e-01 -2.73371905e-01 2.76377215e-03
-4.34010029e-01 2.90647417e-01 -1.07178152e+00 -4.38487828e-01
-1.17977358e-01 3.09512794e-01 -9.52266812e-01 7.11186379e-02
-3.67927402e-02 -5.85197866e-01 4.04077828e-01 2.21260428e-01
3.54923010e-01 -8.54290843e-01 -9.51570868e-01 1.10172443e-01
4.56995398e-01 -7.47560740e-01 1.37272060e-01 -2.23555595e-01
-9.41439450e-01 -1.54611838e+00 -5.85485995e-01 -4.67062056e-01
1.20952773e+00 -2.82822132e-01 9.50200677e-01 2.54742354e-01
-1.00171196e+00 -2.08922416e-01 -5.69324866e-02 -1.09215820e+00
-8.15075159e-01 -1.72296062e-01 -2.21290573e-01 1.79689944e-01
7.72327602e-01 -2.60028422e-01 -1.28487217e+00 1.13865942e-01
-4.72711295e-01 3.35835218e-01 1.04942286e+00 3.63151968e-01
6.81583881e-01 -3.99073392e-01 5.34290448e-04 -1.06336474e+00
7.99763143e-01 1.37351248e-02 -6.30070686e-01 3.06986332e-01
-1.14872265e+00 1.51261255e-01 4.47560817e-01 -4.98874336e-02
-6.45992339e-01 7.29823932e-02 2.75107056e-01 -2.64900714e-01
-8.76106024e-01 2.01162398e-01 4.35513645e-01 -4.20777768e-01
1.25476336e+00 -3.65861565e-01 6.91040635e-01 -3.12911034e-01
4.29233372e-01 1.07507229e+00 7.10366488e-01 1.61785945e-01
4.18531120e-01 7.50996053e-01 5.27957439e-01 -4.22307372e-01
-1.03108239e+00 -3.94473523e-01 -9.06334937e-01 -4.23558950e-01
1.00130880e+00 -3.97777528e-01 -1.14070809e+00 3.38800281e-01
-1.27293253e+00 -2.02297106e-01 -5.93703449e-01 5.25501132e-01
-4.32751000e-01 3.43601666e-02 -2.79602110e-01 -6.53345287e-01
-7.41687953e-01 -1.21900940e+00 6.74604177e-01 9.10404444e-01
-4.64099765e-01 -8.67202878e-01 -1.17169701e-01 7.45025873e-01
6.18760705e-01 4.72183257e-01 1.24631512e+00 -6.76105082e-01
-6.78507626e-01 -3.03682029e-01 -7.54272521e-01 5.65680504e-01
8.71336535e-02 7.41761029e-01 -1.12642360e+00 5.24311364e-01
-4.93295342e-01 1.78846195e-01 4.73579049e-01 9.89487350e-01
9.09801543e-01 -5.78667641e-01 1.46548271e-01 8.06412935e-01
1.22384799e+00 2.51038611e-01 9.93480206e-01 3.64565879e-01
4.92016733e-01 9.61951077e-01 4.51891273e-02 1.41057476e-01
2.79928237e-01 -4.22222028e-03 5.96784651e-01 -8.48484337e-01
-6.59014165e-01 5.41334569e-01 -4.85165924e-01 1.48635030e-01
-7.13955581e-01 8.16309005e-02 -1.17638016e+00 6.08564138e-01
-1.57237434e+00 -5.55604219e-01 -9.68587577e-01 1.82165504e+00
6.98668122e-01 1.12691529e-01 4.71846350e-02 -1.35433480e-01
5.47777414e-01 -1.04306018e+00 -7.51557410e-01 -1.06694257e+00
5.81220277e-02 3.70476186e-01 6.00952268e-01 4.04426575e-01
-7.58569956e-01 7.53760338e-01 6.08284807e+00 -2.40698054e-01
-1.41973543e+00 -4.18807477e-01 9.82155204e-01 -1.10565245e-01
1.51185691e-01 6.81545958e-02 -4.96398747e-01 2.43157685e-01
1.01520443e+00 1.66840985e-01 3.25926065e-01 4.08434927e-01
1.07834888e+00 -1.41255453e-01 -1.06989229e+00 6.61439002e-01
-3.83144081e-01 -1.89460969e+00 4.36917037e-01 2.42114469e-01
5.82651258e-01 5.72102591e-02 3.99026349e-02 -5.16972601e-01
1.37902334e-01 -1.49515402e+00 4.46622632e-02 1.51944745e+00
1.10326922e+00 -2.21477702e-01 1.00036454e+00 -3.95346016e-01
-1.34873524e-01 -1.16713159e-01 -3.09934020e-01 -2.90137351e-01
-9.64151174e-02 4.99097586e-01 -1.53389335e+00 5.25975414e-02
4.49062109e-01 4.44143772e-01 -8.82350087e-01 1.83740318e+00
-1.88011780e-01 6.03478134e-01 5.82885519e-02 2.59079598e-02
3.10746133e-01 -4.04729068e-01 5.95662951e-01 8.17859232e-01
6.49156701e-03 1.10672161e-01 -7.10892677e-01 1.25309181e+00
3.63963723e-01 1.47710592e-01 -1.43878549e-01 -2.40301475e-01
-1.97368395e-02 1.46606445e+00 -6.32405221e-01 -3.58470172e-01
-1.78834707e-01 3.16066742e-01 -2.11250752e-01 2.09766179e-01
-1.23951331e-01 -5.81852734e-01 7.86160350e-01 6.33243263e-01
-2.64687091e-01 3.00091177e-01 -1.16959727e+00 -5.63002646e-01
-1.86975390e-01 -8.01801324e-01 5.33068739e-02 -1.25785327e+00
-9.55592811e-01 7.58144081e-01 -6.36548936e-01 -1.23811531e+00
-2.39489153e-01 -1.27405298e+00 -8.76302063e-01 1.37786508e+00
-1.77780688e+00 -1.29324210e+00 -8.28576982e-01 1.73508063e-01
1.75487906e-01 -5.94417274e-01 7.12672770e-01 -1.23814076e-01
-6.17353201e-01 2.89906561e-01 -3.24010819e-01 3.47871929e-01
6.31696284e-01 -1.59530425e+00 1.49624810e-01 7.57453561e-01
-5.43561697e-01 6.98424101e-01 7.14614034e-01 -2.93285221e-01
-6.75815880e-01 -1.09778392e+00 9.31165278e-01 -5.76941669e-01
4.97696906e-01 7.41789818e-01 -6.11097574e-01 2.36659393e-01
2.47109443e-01 1.58125296e-01 7.72578299e-01 -2.11369410e-01
6.99644610e-02 -1.56388715e-01 -1.26567793e+00 6.73683643e-01
4.91852313e-01 -1.63908169e-01 -5.53751528e-01 5.37851334e-01
3.96808594e-01 -3.87296885e-01 -1.00830495e+00 2.24550083e-01
5.93016386e-01 -1.11250675e+00 4.38965529e-01 -1.40836036e+00
1.00264490e+00 -4.54976827e-01 7.83624828e-01 -8.34812820e-01
-1.56419456e-01 -1.13342190e+00 -4.19387780e-02 6.08528614e-01
6.70889735e-01 -9.03394461e-01 8.30133855e-01 9.96856987e-01
-3.74479055e-01 -9.90458548e-01 -4.64117020e-01 7.63931200e-02
-1.85739532e-01 8.59166607e-02 5.03439784e-01 5.24909496e-01
-3.83805096e-01 -5.89981228e-02 2.16692463e-01 3.15208733e-01
5.53674340e-01 -1.37477025e-01 7.28954792e-01 -1.73615360e+00
1.50192559e-01 -8.48927855e-01 -1.21910799e+00 -2.29552016e-01
-6.85249746e-01 -6.65333092e-01 -4.80610013e-01 -2.08948946e+00
-7.30250776e-02 -2.37307981e-01 -2.45716386e-02 5.88582993e-01
7.34987184e-02 1.47617131e-01 -2.26290166e-01 2.91789174e-01
-7.67624751e-02 -4.39885706e-01 1.74879754e+00 3.87726724e-02
-3.17192793e-01 1.33889571e-01 -1.14419818e+00 9.45457697e-01
7.80133426e-01 -2.15879858e-01 -3.75426471e-01 -4.74325448e-01
2.79566228e-01 -1.05489418e-01 7.52846420e-01 -8.77337039e-01
3.92507970e-01 -3.71808037e-02 4.19716984e-01 -1.75805852e-01
-3.90125811e-01 -4.30082172e-01 -1.74931958e-01 4.95835751e-01
-5.47306299e-01 -5.27963638e-01 -5.52332811e-02 8.30238033e-03
-1.63383678e-01 -9.68502983e-02 1.02271140e+00 -2.79041111e-01
-3.68608266e-01 3.98645699e-01 -3.99993777e-01 -1.10861056e-01
8.70347142e-01 -9.09571826e-01 -8.97369981e-01 -2.08190739e-01
-1.18614495e+00 1.52380526e-01 2.95299470e-01 -5.19801416e-02
4.55195695e-01 -3.48994821e-01 -1.03045070e+00 2.37445142e-02
5.82372062e-02 3.60652685e-01 1.36823803e-01 1.30690587e+00
-1.46105826e+00 8.49709928e-01 -6.46456540e-01 -5.66749692e-01
-1.35009623e+00 -7.29342178e-02 1.29841995e+00 1.97589800e-01
-4.74379033e-01 6.93505883e-01 -2.85992086e-01 -3.65337878e-02
1.01186387e-01 -7.54977942e-01 -7.63940692e-01 -1.03584968e-01
7.32032478e-01 5.61678290e-01 2.57859230e-01 -1.80343017e-01
2.64400512e-01 5.26264727e-01 -2.65069485e-01 4.52628314e-01
1.31979501e+00 -2.35337764e-01 -4.70476747e-01 -2.42834345e-01
5.42300045e-01 -4.73732471e-01 -1.23561776e+00 1.19719781e-01
-7.81345516e-02 -2.05147460e-01 4.53883350e-01 -1.61307657e+00
-9.83159781e-01 1.14126015e+00 1.12823403e+00 3.26106459e-01
1.05891156e+00 -3.30302715e-01 5.27708232e-01 2.32528955e-01
-3.45348865e-01 -8.71308386e-01 -7.05542803e-01 -5.50722778e-02
6.43368423e-01 -1.34952414e+00 8.52490738e-02 -4.07553703e-01
-6.27613366e-01 1.90119624e+00 5.65694988e-01 2.27176487e-01
4.76228654e-01 -1.01461401e-02 7.31199801e-01 -5.38857341e-01
-5.11459112e-01 -5.96436918e-01 8.90206397e-01 7.75005758e-01
4.29392427e-01 9.16370526e-02 -3.74516010e-01 3.73621076e-01
-2.74345011e-01 6.12568736e-01 9.07548130e-01 2.32301801e-01
-6.49066985e-01 -6.22092605e-01 -7.58419465e-03 1.06017542e+00
-3.87555331e-01 -1.97889537e-01 -7.97297060e-01 5.71094155e-01
5.78563511e-01 8.17827821e-01 2.00861081e-01 6.73119277e-02
1.89462140e-01 -7.40326643e-02 2.83276588e-01 -5.61463118e-01
-6.62187159e-01 -2.65371829e-01 3.18865746e-01 -5.88779449e-01
-4.33926523e-01 -2.95440316e-01 -1.20910048e+00 -7.89972395e-02
2.47837212e-02 -3.46640646e-01 8.08635056e-01 1.13512313e+00
7.14622557e-01 7.21851647e-01 2.28512928e-01 -2.31263861e-01
-1.27336279e-01 -9.27810788e-01 -2.37271890e-01 2.00196102e-01
7.20555544e-01 -1.51137590e-01 -2.45966300e-01 5.97180128e-01]
|
[15.832551956176758, -4.0009846687316895]
|
f051a52d-f7ac-4321-8cd9-1119f3228744
|
explainable-machine-learning-for-categorical
|
2305.18437
| null |
https://arxiv.org/abs/2305.18437v1
|
https://arxiv.org/pdf/2305.18437v1.pdf
|
Explainable Machine Learning for Categorical and Mixed Data with Lossless Visualization
|
Building accurate and interpretable Machine Learning (ML) models for heterogeneous/mixed data is a long-standing challenge for algorithms designed for numeric data. This work focuses on developing numeric coding schemes for non-numeric attributes for ML algorithms to support accurate and explainable ML models, methods for lossless visualization of n-D non-numeric categorical data with visual rule discovery in these visualizations, and accurate and explainable ML models for categorical data. This study proposes a classification of mixed data types and analyzes their important role in Machine Learning. It presents a toolkit for enforcing interpretability of all internal operations of ML algorithms on mixed data with a visual data exploration on mixed data. A new Sequential Rule Generation (SRG) algorithm for explainable rule generation with categorical data is proposed and successfully evaluated in multiple computational experiments. This work is one of the steps to the full scope ML algorithms for mixed data supported by lossless visualization of n-D data in General Line Coordinates beyond Parallel Coordinates.
|
['Elijah McCoy', 'Boris Kovalerchuk']
|
2023-05-29
| null | null | null | null |
['interpretable-machine-learning']
|
['methodology']
|
[-1.45491973e-01 5.65725029e-01 -3.59041780e-01 -8.03994358e-01
-1.73746850e-02 -6.28583968e-01 5.87715626e-01 5.42383850e-01
3.61808866e-01 7.63626695e-01 -6.48302585e-02 -1.04008007e+00
-8.07791948e-01 -6.41311169e-01 -2.30667844e-01 -3.59732181e-01
-5.23536742e-01 1.00492597e+00 -5.07757902e-01 5.65899834e-02
3.47083032e-01 1.38587725e+00 -2.05252218e+00 1.18028748e+00
8.23660493e-01 6.68326020e-01 -4.53075826e-01 1.00686228e+00
-6.25794590e-01 8.91096354e-01 -8.10992718e-01 -9.56351906e-02
2.00812489e-01 -1.14163354e-01 -6.85896873e-01 -2.51589835e-01
8.08159888e-01 2.83331752e-01 5.05671859e-01 3.55242908e-01
1.76390857e-01 -2.00239629e-01 1.00833333e+00 -2.44920230e+00
-6.38694584e-01 6.91489100e-01 -3.43224198e-01 -4.23986584e-01
3.72500032e-01 1.52204931e-01 5.94670355e-01 -1.17192137e+00
9.42889035e-01 1.80614185e+00 7.36907005e-01 2.82276064e-01
-1.62360787e+00 -5.09471238e-01 1.79583505e-01 3.02628338e-01
-1.24891996e+00 1.70615241e-01 5.95295370e-01 -7.31983066e-01
1.13385522e+00 1.29204261e+00 9.61543679e-01 4.91543621e-01
1.20929234e-01 5.46575010e-01 1.10622573e+00 -7.35446453e-01
5.14919400e-01 4.04600352e-01 6.52489007e-01 9.00524974e-01
5.66380382e-01 4.50698435e-02 -7.61972725e-01 -3.00575405e-01
7.32374251e-01 2.67385423e-01 3.37831676e-01 -7.21223593e-01
-1.16529262e+00 9.13108766e-01 4.41212445e-01 5.67784794e-02
4.87530045e-02 -1.65132508e-01 4.71689016e-01 5.97656488e-01
3.08141798e-01 6.88193560e-01 -6.78699076e-01 -1.09979883e-01
-8.96155596e-01 4.33517545e-01 7.82751203e-01 1.25090969e+00
6.38338387e-01 7.75559023e-02 -2.13049009e-01 4.79579508e-01
4.24393982e-01 2.66952872e-01 -5.24102570e-03 -7.70609081e-01
4.99648809e-01 1.53585589e+00 -2.84183137e-02 -1.03571129e+00
-1.03739095e+00 -7.12489188e-02 -1.14135814e+00 1.35086071e+00
4.37939584e-01 2.39427254e-01 -1.18742275e+00 1.12481070e+00
5.87081254e-01 -7.78022408e-01 1.62054434e-01 5.39358735e-01
1.23909616e+00 6.22767031e-01 3.73254865e-01 -2.64433950e-01
1.04715288e+00 -5.38049817e-01 -9.20142055e-01 5.70717812e-01
1.13960135e+00 -3.93984139e-01 1.56774068e+00 7.96901524e-01
-9.32300091e-01 -8.02533090e-01 -1.11854708e+00 -7.59293437e-01
-1.23965526e+00 5.20110190e-01 8.63550365e-01 6.48610413e-01
-7.53595233e-01 3.62168461e-01 -8.62037897e-01 -1.72461256e-01
5.91522694e-01 5.83118737e-01 -5.52079439e-01 2.93160886e-01
-7.38646090e-01 8.29505146e-01 5.74765921e-01 -9.83151980e-03
-1.52763620e-01 -1.29173517e+00 -7.91385710e-01 -3.77808586e-02
-1.25661969e-01 -6.02071404e-01 5.91729343e-01 -6.88604712e-01
-7.53026068e-01 9.09838259e-01 -1.92474246e-01 -2.79704541e-01
9.22931254e-01 -2.12370176e-02 -5.30811131e-01 -3.80660385e-01
-2.56747544e-01 8.77321243e-01 2.62509406e-01 -1.66033030e+00
-4.20671135e-01 -4.98891234e-01 -3.87025744e-01 1.05854206e-01
4.35176380e-02 -3.17757487e-01 2.81929314e-01 -9.03035939e-01
1.20834850e-01 -5.29520571e-01 -5.74760586e-02 6.87717617e-01
-8.18613946e-01 -3.41649443e-01 1.63107729e+00 -4.26867545e-01
1.46969223e+00 -1.97961915e+00 -1.02335922e-01 7.37368166e-01
5.38009584e-01 1.01595722e-01 3.95954043e-01 4.17088896e-01
-6.00793183e-01 5.82927048e-01 -2.47406200e-01 -3.11630040e-01
2.01368079e-01 4.30895478e-01 -3.74056101e-01 -8.58275667e-02
1.23193040e-01 7.46719658e-01 -4.04036194e-01 -8.37372184e-01
5.55866301e-01 4.29818243e-01 -2.92590618e-01 2.64341950e-01
-3.11407566e-01 1.46851376e-01 8.80273432e-03 6.59359097e-01
1.08566642e+00 -7.02531859e-02 3.59178305e-01 -3.00092995e-01
-6.34777904e-01 -2.66145438e-01 -1.68019855e+00 1.32974648e+00
-2.57853419e-01 8.63180578e-01 -1.53555438e-01 -5.24979770e-01
1.42833054e+00 -5.47656827e-02 2.96692789e-01 -1.54572576e-01
-4.00613219e-01 7.87640363e-02 -6.26856387e-02 -6.08417988e-01
3.79300117e-01 1.75938532e-01 1.86859399e-01 5.32449841e-01
-3.83522570e-01 -3.29361737e-01 2.29253292e-01 3.11603218e-01
5.53555489e-01 1.83508858e-01 6.33608282e-01 -3.58110875e-01
4.49586689e-01 6.77598596e-01 8.53739232e-02 6.62164807e-01
5.91195762e-01 5.62345207e-01 9.07194018e-01 -1.29526091e+00
-1.05273080e+00 -1.04403853e+00 -3.30825418e-01 9.62513268e-01
-2.68618107e-01 -7.84431100e-01 -2.73095936e-01 -7.04705000e-01
5.30796885e-01 1.22266448e+00 -1.08823752e+00 2.62511581e-01
-3.18776935e-01 -4.47994590e-01 3.00847858e-01 5.52472353e-01
2.00731397e-01 -1.06595910e+00 -8.17063928e-01 -2.05274194e-01
6.08399212e-01 -3.54103237e-01 4.74864304e-01 5.95484614e-01
-1.15298295e+00 -1.43353832e+00 2.24374384e-01 -3.28369051e-01
1.08025396e+00 -2.97010243e-01 1.22717309e+00 3.82343650e-01
-8.63942385e-01 1.17745720e-01 -9.30014327e-02 -1.05420065e+00
-6.27090871e-01 -1.49508357e-01 -1.38005510e-01 -3.93240958e-01
3.45683247e-01 -2.69433141e-01 -1.08164698e-01 3.55635434e-01
-9.70292628e-01 1.02061522e+00 2.48575017e-01 4.23652738e-01
9.28441286e-01 -3.53021950e-01 1.52486309e-01 -1.20245504e+00
6.53890848e-01 -4.56939399e-01 -4.53146607e-01 7.54647434e-01
-9.72580254e-01 4.69272733e-01 7.80173421e-01 -3.28671157e-01
-1.00594592e+00 2.29530349e-01 3.81660581e-01 -4.30006385e-01
-6.12669051e-01 2.72498190e-01 -3.19924504e-01 2.66553462e-01
1.00485635e+00 -3.91020000e-01 2.44123548e-01 -7.70646691e-01
7.73997545e-01 6.10697210e-01 2.17460170e-01 -4.21523482e-01
5.86973846e-01 5.25026321e-01 7.80515909e-01 -5.69712639e-01
-8.29590112e-02 2.11725146e-01 -1.31012952e+00 -1.80960476e-01
7.01911390e-01 -1.84340239e-01 -9.05236542e-01 -5.11960611e-02
-1.05626488e+00 -3.93849730e-01 -5.70008457e-01 1.20114222e-01
-5.16613483e-01 -2.35572934e-01 2.78122649e-02 -9.16643560e-01
-1.97704449e-01 -8.76773953e-01 8.58451009e-01 -1.18673578e-01
-8.85136843e-01 -1.35918283e+00 -1.19855613e-01 -1.30344227e-01
2.09877089e-01 9.89039540e-01 1.75732303e+00 -8.21099401e-01
-2.11691231e-01 1.17755719e-01 -3.57851386e-01 -3.78896356e-01
2.17313513e-01 9.60394263e-01 -8.11638594e-01 3.17361593e-01
-7.34919250e-01 -2.05713183e-01 2.81031579e-01 3.30542207e-01
1.83799493e+00 -7.70861208e-01 -6.14017069e-01 7.98881173e-01
1.17343390e+00 3.32416177e-01 2.30988741e-01 1.26046717e-01
9.22934115e-01 7.85960793e-01 1.03841329e+00 5.43346107e-01
2.92375267e-01 7.77575254e-01 5.39170325e-01 -9.06185269e-01
-2.74050444e-01 -2.17449576e-01 -3.88096958e-01 -1.58631861e-01
-1.74714059e-01 -4.93692383e-02 -1.31039703e+00 -1.46252051e-01
-1.94948697e+00 -9.64737535e-01 -1.18682551e+00 1.99620163e+00
7.17599988e-01 -1.44537807e-01 2.93891490e-01 8.73795152e-01
4.09355789e-01 -3.91249418e-01 -5.95519185e-01 -1.39529264e+00
-1.55428514e-01 -1.25801578e-01 -6.06578179e-02 8.12761426e-01
-9.09287632e-01 3.00173819e-01 6.55677032e+00 6.70101821e-01
-1.08030701e+00 -4.64237213e-01 1.05392182e+00 -4.07149434e-01
-5.75574875e-01 -2.46648654e-01 -8.23501885e-01 9.44690406e-02
7.33222783e-01 -2.22926825e-01 -7.83516280e-03 1.19184315e+00
5.66708982e-01 6.17906414e-02 -1.62138104e+00 1.14365196e+00
-3.07329953e-01 -1.91255152e+00 6.30148053e-01 -3.30716223e-02
5.77501297e-01 -8.56321394e-01 1.56420827e-01 2.35594194e-02
3.98310959e-01 -1.75524807e+00 8.57290089e-01 8.38604808e-01
1.27038825e+00 -9.59440827e-01 1.40504494e-01 1.19450629e-01
-1.00119722e+00 -1.36106282e-01 -1.30119935e-01 -2.09609672e-01
-4.72547859e-01 5.18828750e-01 -1.23468137e+00 6.31119907e-01
5.51529765e-01 8.92795444e-01 -1.15216494e+00 7.38725305e-01
1.29950926e-01 2.82109290e-01 -3.87608320e-01 7.61369662e-03
-2.27014914e-01 1.28655419e-01 2.40534499e-01 1.51211917e+00
2.37682074e-01 2.16422626e-03 -1.28177792e-01 1.00575268e+00
6.03859603e-01 2.16889814e-01 -9.50828493e-01 4.17471290e-01
5.39874971e-01 1.17333674e+00 -7.58978665e-01 -5.24708509e-01
5.77282198e-02 2.87730813e-01 2.51184970e-01 3.32200021e-01
-5.96043885e-01 -2.88994521e-01 4.53252643e-01 5.35110533e-01
-5.02134740e-01 -2.47846588e-01 -1.45768952e+00 -5.42536676e-01
-1.76486358e-01 -1.00003850e+00 1.23580527e+00 -1.27642882e+00
-9.62947488e-01 5.70475519e-01 7.21443534e-01 -1.56024337e+00
-4.79116738e-01 -9.73041892e-01 -4.68447983e-01 9.47981715e-01
-7.70251811e-01 -1.64672339e+00 -8.45426619e-01 8.36010396e-01
3.52528691e-01 -3.00450921e-01 1.23954630e+00 -3.04060042e-01
-2.92984605e-01 5.11932135e-01 -1.39649361e-02 -3.59091938e-01
2.24959165e-01 -1.67654169e+00 2.32588395e-01 3.50901097e-01
3.21420848e-01 7.17180014e-01 7.94656157e-01 -7.21015573e-01
-1.28464866e+00 -9.43336010e-01 7.32208848e-01 -5.93044698e-01
-3.21320519e-02 -7.95538306e-01 -1.09089756e+00 7.10093617e-01
4.35149809e-03 5.50768189e-02 9.49847341e-01 2.28709131e-01
-3.46774161e-01 -2.58327663e-01 -1.44960189e+00 6.14443123e-01
1.00981236e+00 1.31615132e-01 -3.45173389e-01 4.37316120e-01
3.00845861e-01 -5.56304336e-01 -8.69728029e-01 4.99770284e-01
5.28707564e-01 -1.13815498e+00 9.71284032e-01 -1.04577971e+00
3.29592735e-01 -6.57572031e-01 3.03300560e-01 -1.04456401e+00
6.98388442e-02 -5.53751111e-01 -4.05648232e-01 9.79688704e-01
6.94682598e-01 -3.80653650e-01 7.31064379e-01 1.11425865e+00
-1.74826667e-01 -7.55406380e-01 -7.74061859e-01 -4.53723639e-01
-8.51133540e-02 -7.54511178e-01 9.21987653e-01 1.19719887e+00
2.05730677e-01 -2.44104594e-01 -6.90045282e-02 -1.58024430e-01
4.79455799e-01 4.43547636e-01 1.16446698e+00 -1.60336041e+00
3.28564227e-01 -5.68731606e-01 -6.63233578e-01 -1.92661688e-01
-2.56019861e-01 -1.03740466e+00 -9.61542606e-01 -1.78171468e+00
-2.19543651e-01 -1.01623440e+00 2.91095376e-01 1.19901562e+00
2.68855602e-01 -1.73403341e-02 2.91553646e-01 2.81236678e-01
-1.21843711e-01 1.43590093e-01 1.11480153e+00 -4.59866151e-02
-5.68130195e-01 -8.97723660e-02 -3.03462833e-01 5.52540720e-01
6.19871914e-01 -6.00581765e-01 -9.13293540e-01 -2.22106073e-02
4.19332951e-01 -6.26349300e-02 6.95750833e-01 -9.18526530e-01
1.84680074e-02 -6.41192436e-01 1.02835226e+00 -1.17271984e+00
4.01515290e-02 -1.16441524e+00 7.79407084e-01 6.05938971e-01
-8.20830047e-01 4.80026156e-01 6.20635808e-01 2.87980810e-02
-1.07179843e-01 3.18825960e-01 5.61523795e-01 2.75186598e-01
-4.32094604e-01 -1.29767939e-01 -1.53497666e-01 -3.97919238e-01
1.37873912e+00 -6.92811608e-01 -6.47609353e-01 -1.65675089e-01
-1.39935815e+00 2.12754801e-01 2.81408250e-01 5.28449059e-01
7.77095139e-01 -1.58701491e+00 -6.10859513e-01 8.19787323e-01
5.81920892e-02 1.67365685e-01 -2.88881850e-03 4.27842885e-01
-1.06656289e+00 5.48281491e-01 -7.12947488e-01 -7.05682456e-01
-2.11614156e+00 7.03887939e-01 2.60556489e-01 -1.45148396e-01
-5.71290374e-01 3.99159819e-01 -1.67258948e-01 -6.13710165e-01
3.97400975e-01 -8.80149603e-01 -4.38546389e-01 1.44958422e-01
5.46790481e-01 8.37127686e-01 3.20341825e-01 -1.62570149e-01
-4.31209147e-01 3.65672886e-01 1.87327147e-01 1.64578065e-01
1.32313323e+00 2.81845719e-01 -1.26146600e-01 8.83283377e-01
9.65047181e-01 -2.42282093e-01 -1.04938996e+00 6.18613958e-01
7.64178997e-03 -4.37488824e-01 -6.20522261e-01 -1.45552361e+00
-6.53616488e-01 1.15219414e+00 7.39958048e-01 6.61516666e-01
1.00704300e+00 -1.51827782e-01 -7.28533506e-01 2.54223228e-01
-1.39221177e-01 -8.13467622e-01 -3.03120226e-01 4.96687787e-03
1.75452924e+00 -1.10051060e+00 5.08713126e-01 -6.60456240e-01
-8.85298014e-01 1.91809046e+00 6.39966309e-01 5.07276952e-01
8.10750782e-01 6.61541700e-01 4.43341643e-01 -5.99910259e-01
-1.02042007e+00 4.76827681e-01 6.79937124e-01 1.01409423e+00
5.68083823e-01 1.82073995e-01 -2.72994101e-01 3.71619552e-01
-5.73180676e-01 3.49890403e-02 1.25967965e-01 9.40644979e-01
-3.77773792e-02 -1.03938675e+00 -7.52779424e-01 6.29253805e-01
1.83498919e-01 1.58604145e-01 -8.81090879e-01 1.56256354e+00
5.46697378e-01 8.11520517e-01 3.99073064e-01 -2.13444144e-01
3.25677663e-01 2.73142189e-01 4.23242413e-02 -2.81033188e-01
-8.02750230e-01 -6.55967355e-01 1.01262957e-01 -4.97002482e-01
-7.60317594e-02 -7.11516619e-01 -1.86122382e+00 -5.11571765e-01
5.80658913e-01 1.93017945e-01 9.67965543e-01 3.22736830e-01
7.52135634e-01 3.05976063e-01 1.81427389e-01 -5.40971994e-01
6.73383027e-02 -7.10996866e-01 -4.21601027e-01 7.60542452e-01
2.94701308e-01 -6.88807189e-01 -2.94844091e-01 3.73966724e-01]
|
[8.0935697555542, 4.68079137802124]
|
bc9efeaa-0991-4e7d-bd45-f2abf1b89267
|
icanet-a-method-of-short-video-emotion
|
2208.11346
| null |
https://arxiv.org/abs/2208.11346v1
|
https://arxiv.org/pdf/2208.11346v1.pdf
|
ICANet: A Method of Short Video Emotion Recognition Driven by Multimodal Data
|
With the fast development of artificial intelligence and short videos, emotion recognition in short videos has become one of the most important research topics in human-computer interaction. At present, most emotion recognition methods still stay in a single modality. However, in daily life, human beings will usually disguise their real emotions, which leads to the problem that the accuracy of single modal emotion recognition is relatively terrible. Moreover, it is not easy to distinguish similar emotions. Therefore, we propose a new approach denoted as ICANet to achieve multimodal short video emotion recognition by employing three different modalities of audio, video and optical flow, making up for the lack of a single modality and then improving the accuracy of emotion recognition in short videos. ICANet has a better accuracy of 80.77% on the IEMOCAP benchmark, exceeding the SOTA methods by 15.89%.
|
['Lanhang Zhai', 'Mengmeng Tian', 'Xuecheng Wu']
|
2022-08-24
| null | null | null | null |
['video-emotion-recognition']
|
['computer-vision']
|
[-7.20389336e-02 -4.97722358e-01 5.32823429e-02 -1.38598457e-01
-1.35799065e-01 -3.32707494e-01 4.61095691e-01 -1.91979483e-01
-5.63542843e-01 6.92986131e-01 9.71223041e-02 2.41572276e-01
9.38649997e-02 -2.56757230e-01 -9.67601463e-02 -7.97263503e-01
2.51961142e-01 -1.46383092e-01 -2.42863402e-01 -1.38071001e-01
1.15908504e-01 3.52253705e-01 -1.85984087e+00 2.20018387e-01
7.99750566e-01 1.44925451e+00 -3.47664833e-01 5.38529277e-01
-2.36547247e-01 8.84954453e-01 -6.47109926e-01 -6.52570367e-01
-1.44510478e-01 -5.68677783e-01 -6.74214780e-01 1.74355865e-01
1.43382162e-01 -2.48847574e-01 -4.34758753e-01 1.20482659e+00
6.16587818e-01 2.02855468e-01 3.29427987e-01 -1.61081076e+00
-3.12380254e-01 2.69030154e-01 -5.13634503e-01 5.26130050e-02
8.23895335e-01 -1.33574769e-01 8.83575797e-01 -6.73999071e-01
6.60240650e-01 1.04070508e+00 4.42588359e-01 6.09443247e-01
-6.43049896e-01 -5.72629631e-01 1.01320401e-01 6.74435318e-01
-1.49102747e+00 -4.07254964e-01 8.67353320e-01 -2.39204183e-01
6.43310785e-01 3.28627706e-01 8.84639978e-01 1.20026636e+00
5.59187494e-02 9.32399333e-01 1.18038297e+00 9.18548671e-04
6.32306114e-02 1.61837384e-01 2.30683163e-02 4.03354436e-01
-2.25795433e-01 -1.34248212e-01 -5.38632333e-01 1.23560436e-01
3.97175312e-01 2.98351705e-01 -5.55158377e-01 -1.19175054e-01
-1.35635555e+00 4.77398694e-01 2.79585421e-01 7.11614430e-01
-4.38529968e-01 -2.54597723e-01 8.07282925e-01 6.21429145e-01
2.76168942e-01 1.38230100e-01 -2.17299148e-01 -9.13112402e-01
-7.46348739e-01 -1.02176204e-01 8.99387777e-01 4.37626153e-01
4.16611850e-01 6.71738982e-02 3.29351127e-01 7.25992024e-01
5.40150814e-02 4.39625412e-01 6.70756698e-01 -8.96028817e-01
-1.05958441e-02 6.77636683e-01 6.56333342e-02 -1.59035134e+00
-5.18794119e-01 -2.40926638e-01 -1.05453944e+00 -2.35779826e-02
5.79968393e-01 -2.58040249e-01 -4.02397335e-01 1.66651392e+00
1.70901641e-01 3.60674679e-01 2.42789477e-01 1.15395629e+00
9.20377195e-01 8.06324899e-01 2.75454335e-02 -5.46093404e-01
1.22019660e+00 -9.61985171e-01 -1.12677515e+00 1.58262447e-01
6.18494213e-01 -8.31275880e-01 8.41366053e-01 9.02787387e-01
-6.58359408e-01 -5.44903815e-01 -9.08952355e-01 2.83553720e-01
-5.28634310e-01 1.73351113e-02 7.58028746e-01 7.05118299e-01
-7.17565656e-01 3.60268682e-01 -5.21621704e-01 -4.88751560e-01
1.93138674e-01 2.40642965e-01 -7.11578190e-01 -2.08986431e-01
-1.33095515e+00 7.70518005e-01 3.61661881e-01 2.90937096e-01
-3.72119725e-01 -2.65713453e-01 -6.89953506e-01 1.23022189e-02
3.61896902e-01 -2.79439032e-01 8.23881805e-01 -1.73967910e+00
-1.69710183e+00 5.67412853e-01 -1.19983062e-01 -1.36619061e-01
4.49393183e-01 -2.75138319e-01 -9.80802357e-01 5.28182924e-01
-4.77091700e-01 4.74226624e-01 1.01357019e+00 -9.06615794e-01
-5.36875367e-01 -3.68988752e-01 1.56842887e-01 3.56192678e-01
-7.57033587e-01 3.39034736e-01 -2.78520077e-01 -3.14099371e-01
-2.88229454e-02 -9.47961748e-01 3.72034609e-02 -1.62827224e-01
-1.50788054e-01 -2.56250173e-01 1.10598743e+00 -6.30392671e-01
1.32874513e+00 -2.57289028e+00 3.23889732e-01 -1.00176387e-01
1.78949922e-01 4.19518918e-01 -5.33585325e-02 5.08668609e-02
-2.30396271e-01 -3.24967280e-02 -5.46944216e-02 -9.46374461e-02
-3.39105055e-02 1.06849164e-01 -1.03718877e-01 3.83382410e-01
4.29330580e-02 6.06795549e-01 -9.23286855e-01 -6.28678858e-01
4.63746637e-01 8.33297312e-01 -3.45151842e-01 2.31157213e-01
3.22340041e-01 6.15636468e-01 -3.95809382e-01 7.66712964e-01
6.44599736e-01 -1.06217875e-03 2.38365587e-02 -3.12476158e-01
-1.86033055e-01 -4.32461798e-01 -1.33305478e+00 1.82436574e+00
-2.97867507e-01 7.40693212e-01 -1.62241107e-03 -1.22788262e+00
7.87094355e-01 8.19199026e-01 9.99401033e-01 -6.30040228e-01
4.76365358e-01 2.08326161e-01 1.02543859e-02 -9.83701885e-01
3.52378815e-01 -2.88864404e-01 -1.48101479e-01 2.11328566e-01
6.05521873e-02 1.07849129e-01 3.62038642e-01 8.43687579e-02
8.60886514e-01 -6.43403362e-03 2.22160056e-01 4.00270760e-01
8.93377125e-01 -3.60687286e-01 4.87501591e-01 2.13293761e-01
-6.64886057e-01 6.13196254e-01 4.38553482e-01 -6.27857804e-01
-4.92164493e-01 -6.96542740e-01 -5.82160475e-03 8.11108708e-01
2.70944893e-01 -5.40909588e-01 -6.52252257e-01 -6.31108880e-01
-4.03065503e-01 3.29441205e-02 -3.12352449e-01 -2.94919193e-01
-1.68284491e-01 -7.00123310e-01 6.32440925e-01 1.38283923e-01
7.83332765e-01 -1.27808344e+00 -7.45834887e-01 1.48687318e-01
-7.01147854e-01 -1.56220233e+00 -6.30262122e-02 -2.92705536e-01
-6.43226624e-01 -1.05875933e+00 -8.71940255e-01 -3.82424712e-01
3.20993870e-01 1.75275937e-01 8.13300133e-01 1.52921334e-01
-1.93471029e-01 3.02757680e-01 -7.75745630e-01 -4.74732406e-02
-9.74462256e-02 -7.42001235e-02 2.70642906e-01 5.19578695e-01
4.80278045e-01 -5.11985004e-01 -4.86373037e-01 3.58511478e-01
-9.53865647e-01 -2.02457368e-01 4.89636481e-01 6.88024104e-01
1.08777672e-01 3.08223337e-01 6.29487395e-01 -1.98672652e-01
3.24691504e-01 -4.09904510e-01 -1.96227226e-02 7.91370571e-02
-2.55956262e-01 -4.17322367e-01 8.58913004e-01 -5.98055899e-01
-1.16048884e+00 1.19679771e-01 -3.16250622e-01 -5.38356364e-01
-6.28978670e-01 6.95221484e-01 -2.61350095e-01 -2.20857859e-01
1.50519982e-01 2.01864153e-01 3.72415259e-02 -1.89706862e-01
1.00113675e-01 8.89239430e-01 3.85557413e-01 -1.32672176e-01
1.91881076e-01 4.03150618e-01 -2.50244498e-01 -1.18214524e+00
-6.84570074e-01 -6.56949162e-01 -4.11392927e-01 -8.14022005e-01
9.04991210e-01 -7.54486263e-01 -1.15703833e+00 9.12128747e-01
-1.11175609e+00 4.19203013e-01 1.15584195e-01 7.79520512e-01
-3.97432953e-01 6.91848993e-01 -6.84973896e-01 -6.55223727e-01
-1.18850663e-01 -1.17469049e+00 6.60560787e-01 4.11881268e-01
-2.31839344e-01 -8.15427065e-01 -7.67151341e-02 4.55325127e-01
4.32206064e-01 3.22917938e-01 3.52487087e-01 -4.61849093e-01
-8.37231427e-02 -4.76421714e-01 -3.44907075e-01 4.39817369e-01
8.12889412e-02 2.94296741e-01 -1.01779282e+00 3.12763522e-03
1.87500089e-01 -6.17879450e-01 6.40234530e-01 1.22475430e-01
1.08114254e+00 8.90667364e-02 7.70782456e-02 5.08253932e-01
1.21385741e+00 4.31278050e-01 9.60678160e-01 2.34953806e-01
6.27413630e-01 6.54582739e-01 6.19247019e-01 6.42457366e-01
1.17828876e-01 4.87127990e-01 4.95602131e-01 1.38536572e-01
3.49480391e-01 1.62464991e-01 4.96339381e-01 1.15964556e+00
-4.07288998e-01 -2.78215170e-01 -6.23497009e-01 2.87898958e-01
-1.84316945e+00 -1.31576097e+00 1.26227373e-02 1.94837308e+00
4.41179812e-01 -8.89783502e-02 4.97474894e-02 5.76668739e-01
7.42007434e-01 4.16193217e-01 -3.24667364e-01 -4.84616458e-01
-2.97541380e-01 -1.38912678e-01 -2.44356066e-01 3.58621031e-02
-1.10541892e+00 6.49484277e-01 5.68700314e+00 6.64331079e-01
-1.47535157e+00 -4.77687456e-02 5.12653828e-01 4.52656895e-02
1.98994115e-01 -3.01031739e-01 -2.67881364e-01 5.98155558e-01
1.04188490e+00 -4.27493230e-02 4.66162592e-01 6.99527621e-01
1.63541064e-01 -1.87046096e-01 -7.56969154e-01 1.67683256e+00
4.66622442e-01 -7.11253643e-01 -4.84300638e-03 -2.08152816e-01
4.89368975e-01 -2.77451187e-01 -7.71202520e-02 3.61864060e-01
-6.54184639e-01 -1.04162824e+00 2.28521213e-01 5.77676415e-01
5.56353092e-01 -9.83854294e-01 1.00135171e+00 3.41663450e-01
-1.08426511e+00 -6.44785762e-02 -1.25178382e-01 -3.19920391e-01
3.65618169e-01 5.19325018e-01 4.67142090e-02 6.56036496e-01
7.53800333e-01 1.11896086e+00 -2.23933026e-01 1.08850908e+00
-6.58277748e-03 3.68282527e-01 -2.79966086e-01 -1.25210494e-01
1.40083581e-01 -3.19224268e-01 5.13125479e-01 1.06992722e+00
3.96704316e-01 1.88469961e-01 8.76483843e-02 8.98075327e-02
-8.69540349e-02 3.74648601e-01 -6.57437921e-01 -5.28189480e-01
-9.14578140e-02 1.42811203e+00 -6.89361155e-01 -3.89115244e-01
-7.22176433e-01 1.35192978e+00 -8.93304646e-02 2.80042499e-01
-9.73026693e-01 -5.91680348e-01 5.56725085e-01 -6.99934006e-01
-4.89555337e-02 -1.32464141e-01 2.25085467e-01 -1.54513001e+00
-2.00739969e-02 -1.03246915e+00 4.85803992e-01 -8.32930565e-01
-9.80663955e-01 7.93989241e-01 -4.55451518e-01 -1.38665307e+00
-1.15146518e-01 -7.14528918e-01 -4.02483582e-01 2.20202431e-01
-1.31731915e+00 -6.85953856e-01 -7.00419962e-01 9.04231548e-01
4.10486668e-01 7.63475895e-02 9.82225001e-01 7.05170453e-01
-7.73806691e-01 4.27376270e-01 2.27894615e-02 8.23114514e-02
1.00485599e+00 -7.55719066e-01 -5.57542384e-01 5.91476023e-01
3.55564058e-02 2.05842793e-01 7.91956067e-01 -1.62244648e-01
-1.46412802e+00 -3.89151037e-01 8.87459040e-01 7.43402401e-03
7.57560134e-01 4.10870090e-02 -7.89975643e-01 2.28893802e-01
4.68462795e-01 1.03602357e-01 8.39762449e-01 -1.73644740e-02
-2.37809941e-01 -2.90642172e-01 -1.05518019e+00 5.36320865e-01
7.30113029e-01 -6.86095834e-01 -4.44649041e-01 -4.36125994e-02
3.42779368e-01 -7.41168633e-02 -1.10287595e+00 5.30343950e-01
9.32081580e-01 -1.25959194e+00 8.06670249e-01 -4.36665773e-01
5.01510024e-01 -1.54096499e-01 -1.61120012e-01 -1.34441984e+00
2.17095271e-01 -5.05177796e-01 -1.24727942e-01 1.14482880e+00
2.85505597e-02 -6.08706534e-01 6.26080751e-01 5.28236866e-01
1.53502941e-01 -4.45844352e-01 -9.27023530e-01 -5.82304418e-01
-4.34085816e-01 -5.70915401e-01 4.44646806e-01 1.25337040e+00
5.53402066e-01 3.24508697e-01 -6.88770294e-01 -2.05686525e-01
4.36487585e-01 2.87745386e-01 5.83563626e-01 -1.26466775e+00
1.59228593e-01 -5.92756212e-01 -8.17461550e-01 -7.59718895e-01
3.93481106e-01 -3.71249050e-01 -1.62924662e-01 -1.07385898e+00
3.20599884e-01 4.59190719e-02 -7.11772859e-01 1.67987928e-01
-3.16554643e-02 6.45611882e-01 2.43496075e-01 -6.49044216e-02
-1.03926194e+00 7.29232550e-01 1.30211627e+00 -8.62694532e-02
-2.99921911e-02 -2.34666958e-01 -3.98887336e-01 8.83953214e-01
8.55364025e-01 -8.54457542e-02 -1.89137161e-01 -3.15100551e-02
3.85323137e-01 2.53447652e-01 1.85535997e-01 -1.21445489e+00
2.22072080e-01 -8.86631757e-02 1.67262882e-01 -3.77607226e-01
5.71693182e-01 -1.16017461e+00 2.06917644e-01 3.78950238e-01
2.42003463e-02 -1.80828720e-02 1.33472681e-02 3.28063756e-01
-9.01657283e-01 -2.09905091e-03 6.55811250e-01 -1.20311137e-02
-1.19551408e+00 3.42155963e-01 -6.06912732e-01 -5.50178103e-02
9.99742866e-01 -2.91133612e-01 -1.90581545e-01 -5.43413162e-01
-8.96162868e-01 8.78898874e-02 3.78094822e-01 6.21017456e-01
8.10025632e-01 -1.33161509e+00 -4.00821209e-01 8.41630101e-02
1.42526105e-01 -5.44419467e-01 7.26131022e-01 1.32777441e+00
-3.06060165e-01 2.88443148e-01 -5.99773645e-01 -4.79995161e-01
-1.45277619e+00 5.70663631e-01 2.47154787e-01 4.66322079e-02
-4.19861197e-01 5.21093249e-01 -5.67963906e-02 -7.33013153e-02
2.36698419e-01 1.83239490e-01 -5.97356796e-01 4.00135159e-01
8.50026011e-01 4.10288930e-01 -1.01419218e-01 -1.14810574e+00
-4.11001265e-01 7.16817319e-01 1.73409596e-01 1.17489912e-01
1.04399228e+00 -3.54324728e-01 -2.71140695e-01 6.70062900e-01
1.37007928e+00 -1.09987371e-01 -7.11805224e-01 1.37525782e-01
-2.42014036e-01 -5.71395874e-01 -4.61801328e-02 -6.88523889e-01
-1.24643159e+00 1.08884680e+00 6.64906263e-01 3.71078849e-01
1.47298050e+00 -4.50460851e-01 1.00889623e+00 5.42313457e-01
3.57507467e-01 -1.24772501e+00 1.91442028e-01 5.95177114e-01
6.61544740e-01 -1.42054307e+00 -2.78340966e-01 -2.78981358e-01
-8.56186986e-01 1.27648151e+00 5.18341064e-01 2.02334806e-01
6.66346431e-01 -2.66242959e-02 3.43089968e-01 -7.56488368e-02
-5.66142261e-01 -1.26694337e-01 1.28010944e-01 1.91172898e-01
5.10304332e-01 -2.51694713e-02 -4.77756292e-01 5.46968102e-01
1.22255623e-01 2.38574550e-01 4.42058921e-01 6.94041610e-01
-2.18295574e-01 -8.62060189e-01 -4.14861292e-01 2.64165044e-01
-9.50116456e-01 2.87381083e-01 -3.56567025e-01 5.86631358e-01
1.11931980e-01 1.19490063e+00 -9.73270908e-02 -6.11623406e-01
1.72841430e-01 3.23583603e-01 4.00842965e-01 1.26739278e-01
-3.61690164e-01 -3.20495032e-02 9.43640098e-02 -8.25953841e-01
-1.12140667e+00 -6.29420340e-01 -1.22270238e+00 -5.04676998e-01
-2.72748530e-01 4.29421753e-01 6.27444923e-01 1.17378151e+00
2.11183324e-01 2.48200849e-01 8.08292806e-01 -7.26168036e-01
-4.16194387e-02 -8.22554111e-01 -6.69582248e-01 8.84339452e-01
2.51750171e-01 -7.31231272e-01 -6.32122815e-01 4.35521342e-02]
|
[13.303299903869629, 4.85098123550415]
|
a8bcf8fe-5081-46d0-b848-fa3f167d12ac
|
look-into-person-self-supervised-structure
|
1703.05446
| null |
http://arxiv.org/abs/1703.05446v2
|
http://arxiv.org/pdf/1703.05446v2.pdf
|
Look into Person: Self-supervised Structure-sensitive Learning and A New Benchmark for Human Parsing
|
Human parsing has recently attracted a lot of research interests due to its
huge application potentials. However existing datasets have limited number of
images and annotations, and lack the variety of human appearances and the
coverage of challenging cases in unconstrained environment. In this paper, we
introduce a new benchmark "Look into Person (LIP)" that makes a significant
advance in terms of scalability, diversity and difficulty, a contribution that
we feel is crucial for future developments in human-centric analysis. This
comprehensive dataset contains over 50,000 elaborately annotated images with 19
semantic part labels, which are captured from a wider range of viewpoints,
occlusions and background complexity. Given these rich annotations we perform
detailed analyses of the leading human parsing approaches, gaining insights
into the success and failures of these methods. Furthermore, in contrast to the
existing efforts on improving the feature discriminative capability, we solve
human parsing by exploring a novel self-supervised structure-sensitive learning
approach, which imposes human pose structures into parsing results without
resorting to extra supervision (i.e., no need for specifically labeling human
joints in model training). Our self-supervised learning framework can be
injected into any advanced neural networks to help incorporate rich high-level
knowledge regarding human joints from a global perspective and improve the
parsing results. Extensive evaluations on our LIP and the public
PASCAL-Person-Part dataset demonstrate the superiority of our method.
|
['Xiaodan Liang', 'Liang Lin', 'Ke Gong', 'Xiaohui Shen', 'Dongyu Zhang']
|
2017-03-16
|
look-into-person-self-supervised-structure-1
|
http://openaccess.thecvf.com/content_cvpr_2017/html/Gong_Look_Into_Person_CVPR_2017_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2017/papers/Gong_Look_Into_Person_CVPR_2017_paper.pdf
|
cvpr-2017-7
|
['human-parsing']
|
['computer-vision']
|
[ 2.70162255e-01 2.24803746e-01 -2.37675861e-01 -5.58870971e-01
-8.70289028e-01 -5.43024480e-01 3.60542923e-01 -5.25784135e-01
-3.38203788e-01 4.12246764e-01 3.42898369e-01 2.35229939e-01
3.10918003e-01 -4.30153906e-01 -6.53163314e-01 -5.09775579e-01
1.26670897e-01 5.35586119e-01 5.71924865e-01 -2.70421892e-01
-2.62088776e-01 3.20495486e-01 -1.59171212e+00 2.33926162e-01
5.09017169e-01 9.59980667e-01 1.38456047e-01 6.24224961e-01
9.77948979e-02 5.27214408e-01 -4.23913062e-01 -9.36780334e-01
3.40677321e-01 -2.46708646e-01 -9.40189600e-01 2.85290390e-01
8.20098758e-01 -2.43746325e-01 -3.84481624e-02 1.01476800e+00
6.97466671e-01 -5.23534343e-02 2.38203853e-01 -1.26667535e+00
-4.15524393e-01 2.21691072e-01 -6.43504143e-01 -1.91848114e-01
6.18097544e-01 1.69942304e-01 1.10636437e+00 -7.48211145e-01
8.95725965e-01 1.45059741e+00 9.16819334e-01 9.41121995e-01
-8.48658025e-01 -4.92164850e-01 2.95401096e-01 1.27724901e-01
-8.36326301e-01 -4.75953579e-01 9.59976554e-01 -2.79981852e-01
6.66442096e-01 6.02631420e-02 7.30476439e-01 1.50631416e+00
-2.43382335e-01 1.29741895e+00 9.91824567e-01 -4.59723800e-01
-3.53131909e-03 -3.15187052e-02 2.24158466e-01 1.09316325e+00
2.57211745e-01 -1.26229957e-01 -5.84375620e-01 7.29980692e-02
6.19221628e-01 -2.44899541e-01 1.32371867e-02 -8.62719357e-01
-1.13779879e+00 6.74986601e-01 3.43534499e-01 -3.30768526e-02
-7.69702196e-02 1.11145079e-01 5.86625218e-01 -2.44706884e-01
2.49739274e-01 6.79528266e-02 -7.70885646e-01 -1.16816744e-01
-7.31710553e-01 2.59526283e-01 7.78869212e-01 1.01046908e+00
5.54114223e-01 -1.82525888e-01 1.36724692e-02 1.03956902e+00
2.11962044e-01 5.00470996e-01 2.64405072e-01 -1.33222461e+00
6.46883845e-01 6.90780580e-01 -1.34583130e-01 -8.43673766e-01
-7.04447627e-01 -2.29965970e-01 -6.76193237e-01 2.10025266e-01
8.19749475e-01 -7.19955266e-02 -9.75390732e-01 1.83163238e+00
7.19887078e-01 -3.44443709e-01 -8.60619992e-02 9.69517112e-01
9.94092584e-01 9.44432020e-02 3.18227708e-01 2.56653547e-01
1.85055113e+00 -1.53268361e+00 -6.07419252e-01 -4.62778687e-01
3.80687773e-01 -9.13913786e-01 1.21228135e+00 5.90006232e-01
-1.00531578e+00 -7.81633973e-01 -8.12358975e-01 -3.41959715e-01
-4.81906205e-01 3.20299000e-01 1.09089375e+00 9.22332048e-01
-8.73318434e-01 3.76677573e-01 -9.09619868e-01 -6.81373477e-01
5.56125224e-01 3.37031215e-01 -7.55241752e-01 -2.38995776e-01
-9.09627616e-01 7.10647643e-01 2.55016536e-01 2.50587970e-01
-3.66317421e-01 -2.15170532e-01 -1.13183594e+00 -2.42118314e-01
7.79449880e-01 -9.23624694e-01 1.26724207e+00 -8.58964145e-01
-1.55966985e+00 1.07664490e+00 -1.29926443e-01 -1.56636447e-01
8.55247855e-01 -5.08848250e-01 -2.92611793e-02 3.17563713e-01
2.60938883e-01 1.16877341e+00 5.85913599e-01 -1.18351912e+00
-6.93369508e-01 -5.05635738e-01 2.26682812e-01 1.25321135e-01
-2.27985293e-01 1.72014564e-01 -1.22032773e+00 -6.34272695e-01
2.24014536e-01 -1.26908863e+00 -2.84623146e-01 1.20362997e-01
-6.41773582e-01 -3.24022025e-01 6.82231605e-01 -7.55697191e-01
7.06867039e-01 -1.82626259e+00 3.07154179e-01 -1.92201853e-01
2.44715922e-02 3.01214933e-01 -1.40234038e-01 3.36278021e-01
2.20803306e-01 3.05338111e-02 -2.37895206e-01 -7.30816662e-01
2.02118590e-01 4.12003100e-01 2.47913644e-01 3.21754634e-01
3.04368049e-01 1.00958073e+00 -7.57338583e-01 -1.06980503e+00
2.80881792e-01 4.86687124e-01 -7.15685308e-01 3.00696313e-01
-6.76534548e-02 5.87898135e-01 -5.83214641e-01 1.28291011e+00
6.26672804e-01 -1.79037824e-01 7.72053152e-02 -4.63816285e-01
1.68538526e-01 -1.40146837e-01 -1.12450087e+00 2.07193899e+00
-2.51723588e-01 2.66411275e-01 1.85800910e-01 -8.04435670e-01
4.63273913e-01 1.91102475e-01 5.56248367e-01 -6.04361951e-01
1.62474558e-01 -1.28940552e-01 -1.95966303e-01 -6.24338448e-01
2.47459516e-01 1.74076363e-01 -3.35609823e-01 2.26534624e-02
3.17638457e-01 1.35328442e-01 4.45758522e-01 2.81416662e-02
8.81464720e-01 7.58536518e-01 3.27238917e-01 1.60503089e-02
5.70478201e-01 9.20218974e-02 8.44759643e-01 4.42871988e-01
-6.96035445e-01 9.49461460e-01 4.38168883e-01 -5.34777284e-01
-8.88475060e-01 -9.51655090e-01 -1.17855027e-01 1.45560765e+00
1.46064386e-01 -3.68917018e-01 -1.15226269e+00 -1.05110836e+00
-1.83293030e-01 -2.82656159e-02 -7.20604420e-01 2.42887810e-01
-9.04875815e-01 -8.05570185e-01 6.90388381e-01 9.54661429e-01
7.48836279e-01 -1.23627853e+00 -8.09167683e-01 -4.99677137e-02
-4.04425561e-01 -1.67075145e+00 -3.29916775e-01 3.64836119e-02
-6.81316972e-01 -1.37398052e+00 -9.13203359e-01 -9.95209932e-01
4.89863813e-01 -3.84556688e-03 1.25166082e+00 -1.54832289e-01
-6.74042761e-01 6.70137644e-01 -3.52622807e-01 -3.75681847e-01
-1.52012646e-01 1.88140765e-01 -1.71192989e-01 -2.29218930e-01
1.94910288e-01 -2.19993174e-01 -8.46653163e-01 5.43210864e-01
-4.33315039e-01 1.04069367e-01 8.01245689e-01 8.28807473e-01
4.73181874e-01 -2.28973687e-01 3.71249050e-01 -9.99677837e-01
1.27334982e-01 -3.88180390e-02 -3.07832897e-01 3.45494032e-01
-2.79587150e-01 -1.07814461e-01 4.00935948e-01 -2.45373219e-01
-1.34145975e+00 5.57551444e-01 -4.28971976e-01 -1.57670289e-01
-4.31664079e-01 -5.50881810e-02 -6.84888840e-01 -9.51442122e-02
2.44461909e-01 -1.04462862e-01 -5.84124066e-02 -7.29226291e-01
5.93861580e-01 2.60098130e-01 9.19599771e-01 -9.26879108e-01
6.80169582e-01 5.56437731e-01 9.13623273e-02 -5.78459382e-01
-1.04749131e+00 -7.46585310e-01 -1.13439679e+00 -2.20159262e-01
1.20950770e+00 -9.11360264e-01 -7.35310376e-01 5.83154202e-01
-1.10546386e+00 -1.48583457e-01 1.84325930e-02 2.73715913e-01
-6.41193390e-01 8.63086224e-01 -7.88075805e-01 -6.69215798e-01
-3.61720711e-01 -1.30687881e+00 1.45905697e+00 2.39093706e-01
-2.12377504e-01 -8.40387344e-01 9.84708313e-03 1.08656085e+00
1.08925523e-02 4.87561554e-01 6.91020310e-01 -5.97481310e-01
-4.84588414e-01 -3.38989109e-01 -3.63130033e-01 4.67454880e-01
-3.90707627e-02 -9.49082747e-02 -1.04616201e+00 -1.48982346e-01
-3.60482037e-01 -6.99196100e-01 7.58179069e-01 2.31648669e-01
9.83472705e-01 8.96745455e-03 -3.43500584e-01 5.59424400e-01
1.10094070e+00 -1.97637156e-01 3.74229789e-01 4.11393464e-01
1.01510406e+00 1.00139165e+00 7.57272899e-01 2.27192551e-01
5.58120370e-01 8.53792608e-01 5.16923368e-01 -4.68277633e-01
-6.09939873e-01 -3.81576687e-01 1.97834775e-01 6.18127465e-01
-4.54683095e-01 -1.42197341e-01 -8.68077636e-01 3.89333695e-01
-1.83201122e+00 -8.21748793e-01 -2.88188155e-03 1.86350000e+00
6.86229169e-01 1.98687211e-01 4.47059244e-01 -3.51275280e-02
7.70877719e-01 1.26561150e-01 -5.26018739e-01 -1.92585513e-02
-3.85494605e-02 1.83679461e-01 3.74625415e-01 6.70162216e-02
-1.54167664e+00 1.24482000e+00 6.39121723e+00 7.32505679e-01
-7.31205165e-01 8.29062536e-02 4.45920587e-01 2.62825012e-01
3.40270102e-01 -1.75406367e-01 -1.03029001e+00 1.85151994e-01
3.96167696e-01 6.45956099e-01 6.88346699e-02 1.25506544e+00
-8.17494094e-02 -1.80186946e-02 -1.09980261e+00 9.34984088e-01
2.64152050e-01 -8.20191562e-01 -1.82498172e-01 -6.88071409e-03
4.38945562e-01 -1.70725673e-01 -4.25626487e-02 4.74144459e-01
2.51361281e-01 -8.05066824e-01 7.49586940e-01 2.62554795e-01
4.79530752e-01 -5.66624582e-01 7.86625803e-01 2.94494987e-01
-1.37988126e+00 7.33505115e-02 -3.75319719e-01 1.01698354e-01
3.76190543e-01 2.14651749e-01 -6.75778508e-01 5.52186668e-01
1.06039357e+00 5.56865811e-01 -8.63859236e-01 7.37936378e-01
-3.23987693e-01 4.93264109e-01 -2.70298153e-01 2.30024740e-01
1.03461944e-01 -3.40084136e-02 2.46744707e-01 1.38636887e+00
-1.70848116e-01 -8.54495987e-02 4.52372402e-01 4.78830785e-01
4.61079553e-02 2.01941028e-01 -2.86845326e-01 1.97521418e-01
7.83541575e-02 1.52171242e+00 -9.83095706e-01 -3.20059448e-01
-6.62679553e-01 1.05022109e+00 4.40580517e-01 1.67128712e-01
-8.69277477e-01 1.08702341e-03 4.14430588e-01 4.27887626e-02
4.10745382e-01 -1.74995303e-01 -1.91297919e-01 -1.27288008e+00
2.55078226e-01 -9.59333420e-01 5.73107719e-01 -6.94207311e-01
-1.30511165e+00 5.61460614e-01 1.03643119e-01 -9.87183511e-01
-3.62765551e-01 -9.37917411e-01 -3.67863059e-01 3.24026376e-01
-1.46881306e+00 -1.90570533e+00 -4.68807995e-01 6.23368740e-01
8.55300307e-01 8.61341041e-03 7.57429540e-01 3.22724938e-01
-8.08645546e-01 7.70725667e-01 -5.74332535e-01 5.11416912e-01
8.84506702e-01 -1.30135190e+00 4.15590525e-01 8.21917295e-01
1.69079497e-01 6.00939155e-01 6.14390016e-01 -5.74866056e-01
-1.40234911e+00 -8.83794069e-01 7.77800858e-01 -6.98260307e-01
4.36356544e-01 -5.76630235e-01 -5.78148484e-01 6.92475021e-01
2.23882839e-01 3.12451452e-01 7.22775877e-01 3.38829339e-01
-4.83674467e-01 9.31529328e-02 -1.11190856e+00 5.74160516e-01
1.47232175e+00 -1.29923910e-01 -6.83511257e-01 3.18077564e-01
6.28134191e-01 -4.63523328e-01 -8.06086957e-01 8.28765035e-01
8.39294434e-01 -1.23574162e+00 1.32070923e+00 -6.89549983e-01
3.46371293e-01 -8.76520723e-02 -2.60597646e-01 -5.98037362e-01
-1.22064501e-01 -3.19031924e-01 3.17936600e-03 1.46209502e+00
1.09491035e-01 -2.81443834e-01 1.24432635e+00 7.79556334e-01
-1.25615641e-01 -8.88261437e-01 -8.89540792e-01 -6.24618173e-01
-3.86212878e-02 -4.09108520e-01 1.71166956e-01 6.13875031e-01
-2.38808811e-01 2.76530415e-01 -6.90359354e-01 9.28475708e-02
7.28254735e-01 6.37195781e-02 1.17269337e+00 -1.17737436e+00
-4.21100557e-01 -2.24178240e-01 -5.22920907e-01 -9.79085445e-01
2.69198477e-01 -6.15176320e-01 9.70380902e-02 -1.51728737e+00
4.80555356e-01 -1.60968959e-01 -1.02286562e-01 8.44670236e-01
-3.48331809e-01 6.17917001e-01 4.15651858e-01 1.99457705e-01
-1.02328539e+00 4.24839735e-01 1.37376213e+00 -1.98977366e-02
2.06715003e-01 1.99943513e-01 -4.29846257e-01 1.27711928e+00
4.25596088e-01 -2.87465960e-01 -2.38593534e-01 -3.93194973e-01
-6.41402975e-02 -5.81568629e-02 4.90872502e-01 -1.16042292e+00
1.66127801e-01 1.33055255e-01 5.70425034e-01 -6.63125992e-01
5.99716902e-01 -6.95754409e-01 -2.14706168e-01 3.58805507e-01
-4.59801145e-02 5.17128482e-02 -5.53036369e-02 6.29656494e-01
-2.71760792e-01 -1.47088751e-01 6.96358800e-01 -5.22401392e-01
-1.14645779e+00 3.28653485e-01 1.72260553e-01 3.42362732e-01
8.60466480e-01 -4.43779618e-01 4.46826331e-02 -9.42961648e-02
-8.39305103e-01 1.93277985e-01 5.38439035e-01 5.81680655e-01
2.14609385e-01 -1.04448938e+00 -3.67895752e-01 3.15223224e-02
2.37998500e-01 1.19206138e-01 3.56351048e-01 6.25025332e-01
-4.15406883e-01 3.74376476e-01 -3.53819519e-01 -8.21606457e-01
-1.44357550e+00 4.35193509e-01 9.29792691e-03 -4.53561008e-01
-7.25416720e-01 7.89502144e-01 3.50100845e-01 -7.44627655e-01
3.61395836e-01 -1.65204793e-01 -1.88325152e-01 9.33686718e-02
1.48023888e-01 3.87319624e-01 -4.95890677e-02 -8.72485340e-01
-5.30294716e-01 1.03593099e+00 4.06226888e-02 9.66503769e-02
1.11688673e+00 -8.92262980e-02 2.72756785e-01 1.12157568e-01
1.23463750e+00 1.89859048e-01 -1.53703773e+00 -5.03346547e-02
1.21181063e-01 -3.05050701e-01 -4.72224534e-01 -9.28899527e-01
-1.28534877e+00 9.84767020e-01 6.08463883e-01 -3.03745627e-01
9.84317482e-01 3.62739295e-01 9.97584045e-01 2.83198625e-01
6.01617992e-01 -1.23142612e+00 3.35771888e-01 2.90475190e-01
6.45154595e-01 -1.59665465e+00 4.97235730e-02 -9.11442876e-01
-8.51297081e-01 1.10070705e+00 7.83292115e-01 -4.69093956e-02
2.17799187e-01 2.40519941e-01 4.13687706e-01 -1.60491735e-01
-2.93664157e-01 -5.44323325e-01 4.03339982e-01 8.56631994e-01
4.64601487e-01 -1.58786938e-01 -3.74300063e-01 8.73168409e-01
-3.41666043e-01 -1.18550844e-01 3.80772203e-02 9.75529969e-01
-2.91871399e-01 -1.44054842e+00 -3.68374646e-01 -1.79512486e-01
-5.08278191e-01 2.74730474e-01 -4.60124969e-01 1.30002630e+00
3.64299357e-01 7.11603940e-01 -4.22653377e-01 -8.86623636e-02
5.51493764e-01 2.34337211e-01 5.35907447e-01 -4.89963800e-01
-3.10046524e-01 1.97853550e-01 2.85261005e-01 -9.26377952e-01
-6.98700964e-01 -8.24354053e-01 -1.20432782e+00 2.32242212e-01
-1.93835378e-01 -2.83071846e-01 7.40672648e-01 9.38470840e-01
1.41535297e-01 3.97763580e-01 4.66964729e-02 -1.16722155e+00
-4.91048336e-01 -7.53651798e-01 -1.85302928e-01 7.80233324e-01
-8.18860605e-02 -8.93185616e-01 -1.17924400e-01 2.28430599e-01]
|
[8.194456100463867, -0.2464849054813385]
|
92a69380-579f-44fd-9b65-d51abecc2b8a
|
prefix-projection-global-constraint-for
|
1504.07877
| null |
http://arxiv.org/abs/1504.07877v2
|
http://arxiv.org/pdf/1504.07877v2.pdf
|
Prefix-Projection Global Constraint for Sequential Pattern Mining
|
Sequential pattern mining under constraints is a challenging data mining
task. Many efficient ad hoc methods have been developed for mining sequential
patterns, but they are all suffering from a lack of genericity. Recent works
have investigated Constraint Programming (CP) methods, but they are not still
effective because of their encoding. In this paper, we propose a global
constraint based on the projected databases principle which remedies to this
drawback. Experiments show that our approach clearly outperforms CP approaches
and competes well with ad hoc methods on large datasets.
|
['Amina Kemmar', 'Yahia Lebbah', 'Thierry Charnois', 'Samir Loudni', 'Patrice Boizumault']
|
2015-04-29
| null | null | null | null |
['sequential-pattern-mining']
|
['natural-language-processing']
|
[ 2.22728521e-01 -2.75096595e-01 -4.81082797e-01 -4.07403678e-01
-2.11915281e-02 -2.13473693e-01 3.64926517e-01 8.68908837e-02
-3.19690555e-01 8.52451444e-01 -1.02804057e-01 -1.69650286e-01
-5.55790842e-01 -1.03101397e+00 -2.42654622e-01 -4.63752866e-01
-1.31314054e-01 6.86757147e-01 9.10887063e-01 -1.64342627e-01
4.94268775e-01 3.14743817e-01 -1.86906993e+00 5.37250876e-01
8.15894067e-01 9.53445494e-01 4.45068441e-02 -7.95492716e-03
-5.55738747e-01 4.04537261e-01 -4.92531002e-01 -1.87876612e-01
3.11948329e-01 -6.08317733e-01 -5.96549451e-01 2.31856719e-01
-3.90188098e-01 3.96083623e-01 1.04493447e-01 8.68583262e-01
5.04599661e-02 -1.28815159e-01 1.81509271e-01 -1.47902501e+00
-7.13737607e-02 5.80974519e-01 -7.62983382e-01 2.35888828e-02
6.84882700e-01 -4.74916250e-01 1.24885809e+00 -9.87204015e-01
7.92096734e-01 9.44155931e-01 5.93307853e-01 4.70231801e-01
-1.02105546e+00 -5.76719105e-01 4.45205569e-01 5.61824381e-01
-1.64040768e+00 -8.47284049e-02 7.69292533e-01 -1.33272305e-01
1.17863655e+00 5.72007596e-01 8.41350138e-01 6.22339129e-01
2.18685698e-02 8.14944625e-01 1.39048624e+00 -6.87096000e-01
5.11145234e-01 3.31458151e-01 5.13670623e-01 3.50108802e-01
6.41541839e-01 9.87977237e-02 -7.28717566e-01 -3.72690797e-01
4.40722778e-02 -5.06940521e-02 -2.94930041e-01 -5.67360997e-01
-7.81985939e-01 8.83957744e-01 -3.92729282e-01 6.23168945e-01
-1.57823175e-01 -6.23328924e-01 4.20071751e-01 6.19877934e-01
6.73885792e-02 3.61075699e-01 -7.59902358e-01 -3.93890403e-02
-7.98156142e-01 8.82423282e-01 1.21082437e+00 1.24973249e+00
4.38281327e-01 -2.38780528e-01 2.29102120e-01 6.00723743e-01
1.04960866e-01 7.27680624e-02 2.26901308e-01 -2.36746535e-01
2.84921199e-01 1.26801181e+00 5.04505299e-02 -1.39477229e+00
-6.16452754e-01 -2.92768896e-01 -4.99688894e-01 3.33074741e-02
1.19182386e-01 8.43102783e-02 -5.21492660e-01 1.23907471e+00
2.75805175e-01 -2.57029057e-01 7.13124573e-02 7.26565480e-01
3.35335582e-01 5.88916838e-01 2.57071815e-02 -9.52159762e-01
8.49994957e-01 -8.83076906e-01 -9.27730978e-01 -9.60766803e-03
3.33310455e-01 -5.58369994e-01 5.79960942e-01 1.26917410e+00
-9.82379496e-01 -2.22851604e-01 -1.17981899e+00 6.62554502e-01
-5.53371847e-01 -4.09263968e-01 8.43303025e-01 9.90732074e-01
-6.16105139e-01 3.65467608e-01 -6.86979592e-01 -5.11421144e-01
3.06971997e-01 6.25119507e-01 -2.68276960e-01 -1.44190967e-01
-8.23913455e-01 9.89492834e-01 8.15457880e-01 -3.44275981e-02
-1.40641838e-01 -4.10680562e-01 -5.25229633e-01 -2.91251559e-02
1.07262623e+00 -1.35302737e-01 8.50279212e-01 -6.44837618e-01
-1.19429958e+00 5.80921412e-01 -9.80633870e-02 -5.45338333e-01
4.23178196e-01 8.89331773e-02 -8.60933900e-01 -2.19647974e-01
-2.53603339e-01 2.28682868e-02 3.21907341e-01 -1.12807429e+00
-1.06534851e+00 -5.00495076e-01 -9.23641771e-02 -1.51124209e-01
-6.22723520e-01 4.43988919e-01 -4.83910054e-01 -4.26005363e-01
4.49993283e-01 -7.27230787e-01 -7.33530462e-01 -3.72582167e-01
-2.43873328e-01 -5.00184417e-01 9.72388029e-01 -2.70293970e-02
2.03036451e+00 -1.78462481e+00 1.60049543e-01 8.14277411e-01
-2.14738399e-01 4.10044968e-01 3.98707092e-01 6.95697904e-01
1.86351761e-01 3.05720806e-01 -6.14618182e-01 -6.65481016e-02
1.99990682e-02 6.47037089e-01 -1.71114996e-01 2.66094536e-01
7.69536793e-02 4.16017205e-01 -4.80030656e-01 -7.58396447e-01
-1.23100407e-01 -9.08496529e-02 -7.59141624e-01 -7.89741799e-03
-6.55286193e-01 -8.48088935e-02 -5.32827020e-01 1.00982833e+00
9.35721934e-01 2.49384224e-01 1.06165004e+00 4.28997546e-01
-4.65563774e-01 -5.79221658e-02 -1.81387150e+00 1.52596927e+00
2.62757033e-01 2.93517336e-02 1.58340726e-02 -1.51914191e+00
1.21395946e+00 3.22605640e-01 8.16054106e-01 -8.63221884e-01
-9.27266255e-02 5.82243860e-01 6.45726919e-02 -7.25691736e-01
2.48871580e-01 -3.43820870e-01 -1.09315597e-01 3.60273182e-01
-4.59558606e-01 1.65161759e-01 5.94845057e-01 -2.22969636e-01
1.21548474e+00 1.38084814e-01 7.63724387e-01 -5.34861922e-01
9.03217256e-01 6.09129310e-01 1.41091669e+00 5.79200983e-01
1.45215765e-01 4.75801975e-01 5.98841667e-01 -7.85816133e-01
-7.90296674e-01 -4.84616786e-01 -3.95029366e-01 3.49883288e-01
1.01852164e-01 -1.06073618e+00 -1.64578214e-01 -8.02568555e-01
1.89218268e-01 2.53804475e-01 -2.54955769e-01 3.79636973e-01
-7.91652262e-01 -9.61163759e-01 9.67311934e-02 3.90606940e-01
2.97170490e-01 -1.13979542e+00 -9.04787719e-01 7.22653508e-01
1.28585786e-01 -1.10786533e+00 1.96867645e-01 2.88318366e-01
-8.99580538e-01 -1.25281227e+00 -1.89330839e-02 -6.28472269e-01
6.13836825e-01 3.73930186e-01 9.66414094e-01 4.76742148e-01
-4.28326100e-01 -2.56971270e-01 -9.66701806e-01 -8.87529790e-01
-1.25107080e-01 6.89170435e-02 2.44318217e-01 4.58248220e-02
1.06290054e+00 -6.69885516e-01 -7.69305304e-02 7.83123910e-01
-9.35733557e-01 -3.58939499e-01 5.17397046e-01 6.99812353e-01
7.77257562e-01 7.31355965e-01 7.08024383e-01 -1.39040899e+00
7.63057649e-01 -5.43698728e-01 -8.42479944e-01 3.86922330e-01
-1.37105942e+00 -1.25711739e-01 5.35381556e-01 -1.06391720e-01
-1.03303146e+00 2.78106004e-01 6.81164190e-02 -8.21988732e-02
-2.81349957e-01 1.01741850e+00 -3.64357799e-01 -4.38379543e-03
4.61427391e-01 9.44564044e-02 -1.94787771e-01 -7.30497658e-01
-3.65858674e-01 6.36882663e-01 -2.91902525e-03 -7.29341507e-01
7.29320824e-01 5.21115422e-01 2.05110893e-01 -6.71934843e-01
-5.23179114e-01 -7.31284261e-01 -4.37081635e-01 1.08861603e-01
4.78790134e-01 -2.61016697e-01 -5.44958830e-01 1.95899665e-01
-8.45770717e-01 2.72339404e-01 -9.36649889e-02 2.92624652e-01
-4.24877822e-01 5.40361404e-01 1.54169984e-02 -1.11359286e+00
1.16674444e-02 -6.29748702e-01 5.64749986e-02 4.15754952e-02
-1.69223636e-01 -4.15479869e-01 3.20799410e-01 -2.31677983e-02
4.91589040e-01 5.81922591e-01 8.57938170e-01 -7.43716002e-01
-4.27025795e-01 -2.40925416e-01 -1.35572664e-02 1.41094074e-01
-8.40536430e-02 2.14617580e-01 -2.81793088e-01 -1.31783679e-01
1.71304554e-01 -6.90331683e-02 4.74046856e-01 2.05426496e-02
1.26095986e+00 -2.54957050e-01 -7.92699933e-01 1.73596457e-01
1.79036307e+00 7.70321608e-01 7.21711576e-01 5.55555224e-01
-2.68479809e-02 9.79539573e-01 1.25591922e+00 8.12072992e-01
1.66588619e-01 9.52725530e-01 2.81792998e-01 2.46011660e-01
3.62743318e-01 1.25269994e-01 2.25542095e-02 5.49020648e-01
-2.27158964e-01 -2.57310212e-01 -1.05851746e+00 6.50433838e-01
-2.37071633e+00 -1.20519805e+00 -6.43256247e-01 1.96202874e+00
6.92527175e-01 6.24701440e-01 4.73432600e-01 1.03721070e+00
3.93097878e-01 -2.22386584e-01 -9.10427943e-02 -8.57963443e-01
-4.10312772e-01 5.06040990e-01 1.07596606e-01 1.39811024e-01
-9.17427659e-01 6.04397416e-01 6.64104271e+00 4.10479099e-01
-5.27514219e-01 -5.83527163e-02 -2.28522196e-01 1.06657736e-01
-3.48026484e-01 2.97312796e-01 -9.68334198e-01 3.79336059e-01
7.03611374e-01 -3.29451054e-01 -9.51243788e-02 9.41615760e-01
-3.33651826e-02 -2.49576390e-01 -9.77116644e-01 8.01859319e-01
1.81760237e-01 -1.23933852e+00 -1.14769645e-01 2.69436151e-01
7.23494172e-01 -5.53643346e-01 -5.18308043e-01 1.95609644e-01
-2.95471489e-01 -9.53797996e-01 5.29737473e-01 4.00456399e-01
5.31129874e-02 -9.83692408e-01 8.75117064e-01 4.87956047e-01
-1.20121944e+00 -3.46406639e-01 -4.18040305e-01 -4.56799209e-01
2.05407515e-01 7.39304125e-01 -4.14275318e-01 9.13967371e-01
8.30912471e-01 5.36103547e-01 -2.92853624e-01 1.39846039e+00
-5.12329601e-02 3.27230960e-01 -5.42447329e-01 -2.61440307e-01
3.73490378e-02 -2.54022926e-01 5.47831953e-01 1.39563882e+00
2.07798481e-01 2.90559202e-01 4.14052606e-01 7.34438658e-01
5.74614286e-01 4.00728106e-01 -7.28391528e-01 -5.14465105e-03
3.73379260e-01 8.23704720e-01 -7.78491914e-01 6.54164106e-02
-9.33579028e-01 4.04240817e-01 -3.77728604e-02 -1.83117166e-01
-6.55353248e-01 -3.71557444e-01 4.65166301e-01 2.27157503e-01
4.01872039e-01 -5.14169812e-01 -5.53957164e-01 -1.05473924e+00
6.95860147e-01 -1.25568187e+00 9.69968200e-01 6.06075153e-02
-1.16913426e+00 5.39582491e-01 3.29344422e-01 -1.36891413e+00
2.56426428e-02 -6.58153415e-01 -3.73416901e-01 3.15681309e-01
-1.61126113e+00 -8.28621626e-01 -1.75826281e-01 8.49671006e-01
4.99250114e-01 -2.22739965e-01 9.54592705e-01 4.58787411e-01
-6.49765313e-01 4.80746090e-01 -3.72847438e-01 -4.30766553e-01
2.93940514e-01 -9.58068252e-01 -7.71970823e-02 1.07661819e+00
1.14066914e-01 5.81919909e-01 8.99689019e-01 -5.80823779e-01
-1.77063835e+00 -4.55815136e-01 1.23124957e+00 -1.98921949e-01
4.32186961e-01 -2.39151478e-01 -9.56621587e-01 4.18400496e-01
1.23711117e-02 -1.27600610e-01 8.44321012e-01 3.72836709e-01
-2.18338877e-01 -2.31804296e-01 -1.25809777e+00 3.00245345e-01
1.39928460e+00 2.47225106e-01 -8.01351249e-01 1.19299188e-01
3.97177994e-01 1.21160429e-02 -6.90898180e-01 8.04818988e-01
5.36248326e-01 -1.27056682e+00 4.46875215e-01 -6.34457469e-01
1.28048003e-01 -6.05627596e-01 -2.31920093e-01 -6.41892493e-01
-9.79416519e-02 -5.73997617e-01 -2.06700072e-01 1.17609811e+00
6.18530154e-01 -5.74515939e-01 1.10250759e+00 7.06101716e-01
-1.05511390e-01 -9.64742541e-01 -9.25582170e-01 -1.17449808e+00
-2.88539529e-01 -6.65395439e-01 7.93162107e-01 1.00155306e+00
7.33839035e-01 -1.13977753e-01 -4.81679618e-01 -6.63251057e-02
6.30173445e-01 6.44164026e-01 6.30798280e-01 -1.73271227e+00
-5.10635436e-01 -4.98570651e-01 -4.62127179e-01 -3.95533115e-01
-1.33807406e-01 -6.98202431e-01 -2.51022160e-01 -1.36969233e+00
2.62743741e-01 -8.93646836e-01 -3.75836164e-01 5.58349192e-01
3.22989762e-01 2.50106640e-02 -3.64318602e-02 1.79907769e-01
-7.13976979e-01 -1.13308979e-02 6.60086274e-01 2.86587439e-02
-5.34257650e-01 -1.64268110e-02 -6.11030400e-01 5.55894256e-01
9.92252290e-01 -8.12563956e-01 -3.56489509e-01 -6.87309057e-02
7.01907992e-01 2.65685073e-03 -3.67922813e-01 -1.14524484e+00
5.86134970e-01 -8.07340324e-01 -2.03362927e-01 -8.76009285e-01
-1.24218188e-01 -1.25926149e+00 6.53364837e-01 6.46118045e-01
7.66191185e-02 1.83068261e-01 -8.80323444e-03 6.46455884e-01
-5.59430718e-01 -3.68224144e-01 6.58987343e-01 -1.64370432e-01
-9.08498764e-01 2.52885461e-01 -4.17178959e-01 -2.90468454e-01
1.45074403e+00 -4.66591775e-01 9.15937424e-02 2.81809866e-01
-7.35835731e-01 4.31223661e-01 3.66694510e-01 4.29092348e-01
7.92070508e-01 -1.08694541e+00 -5.67195356e-01 2.95669019e-01
3.36756408e-01 9.72600281e-02 -2.41133854e-01 8.37070167e-01
-4.36048001e-01 8.32986414e-01 -4.87624943e-01 -3.34038764e-01
-1.28642511e+00 1.02125943e+00 -1.16655238e-01 -3.70909840e-01
-6.80907607e-01 5.28987229e-01 -7.35687673e-01 -1.23393618e-01
5.39223433e-01 -7.21576512e-02 -2.98265934e-01 1.43245757e-01
5.72354853e-01 4.50027019e-01 3.33279669e-01 -7.54460543e-02
-7.58685648e-01 7.04351544e-01 -8.25441703e-02 6.27677590e-02
1.61785865e+00 1.13415532e-01 -2.90429890e-01 6.01924807e-02
5.78516662e-01 1.29713744e-01 -3.56532097e-01 -3.94505322e-01
8.58423352e-01 -8.26848626e-01 -4.87565339e-01 -7.05732465e-01
-9.94797349e-01 2.83970237e-01 5.01627699e-02 6.18995309e-01
1.38332546e+00 -2.43655190e-01 6.63291514e-01 4.66825336e-01
1.01310420e+00 -1.46992612e+00 -2.89478093e-01 3.76218736e-01
7.28392661e-01 -1.07863343e+00 4.35405910e-01 -9.35125232e-01
-5.42849123e-01 1.30032468e+00 7.62878060e-01 -1.45710140e-01
9.40002799e-01 6.58338845e-01 -2.87664920e-01 -3.45837444e-01
-9.43467677e-01 -3.48657250e-01 -2.18600526e-01 5.65268934e-01
1.83007613e-01 4.59478721e-02 -1.56013513e+00 9.58921552e-01
6.29941374e-02 2.61630177e-01 4.93373871e-01 1.55274916e+00
-4.42256153e-01 -1.93352532e+00 -2.71023035e-01 1.81104437e-01
-4.40875798e-01 4.54055160e-01 -6.49148703e-01 1.05427003e+00
6.84553266e-01 1.20949197e+00 -3.93878996e-01 -5.50880849e-01
7.01133072e-01 6.28515482e-02 4.85881746e-01 -7.16950834e-01
-6.12011969e-01 -1.28276035e-01 3.91685694e-01 -6.47543311e-01
-7.51248777e-01 -9.44132745e-01 -1.18427229e+00 -2.15361565e-01
-3.95348132e-01 5.08938670e-01 6.23421550e-01 7.06682205e-01
-3.20829153e-02 -6.37904853e-02 7.92999804e-01 2.95660365e-02
-4.22788173e-01 -4.60572034e-01 -7.01866269e-01 2.41119966e-01
-3.44165534e-01 -7.86891401e-01 -2.50097543e-01 -2.74510324e-01]
|
[8.33487606048584, 6.305901050567627]
|
f3c14040-7a92-4628-bb53-ea789bf1ba74
|
search-to-capture-long-range-dependency-with
|
2302.08671
| null |
https://arxiv.org/abs/2302.08671v1
|
https://arxiv.org/pdf/2302.08671v1.pdf
|
Search to Capture Long-range Dependency with Stacking GNNs for Graph Classification
|
In recent years, Graph Neural Networks (GNNs) have been popular in the graph classification task. Currently, shallow GNNs are more common due to the well-known over-smoothing problem facing deeper GNNs. However, they are sub-optimal without utilizing the information from distant nodes, i.e., the long-range dependencies. The mainstream methods in the graph classification task can extract the long-range dependencies either by designing the pooling operations or incorporating the higher-order neighbors, while they have evident drawbacks by modifying the original graph structure, which may result in information loss in graph structure learning. In this paper, by justifying the smaller influence of the over-smoothing problem in the graph classification task, we evoke the importance of stacking-based GNNs and then employ them to capture the long-range dependencies without modifying the original graph structure. To achieve this, two design needs are given for stacking-based GNNs, i.e., sufficient model depth and adaptive skip-connection schemes. By transforming the two design needs into designing data-specific inter-layer connections, we propose a novel approach with the help of neural architecture search (NAS), which is dubbed LRGNN (Long-Range Graph Neural Networks). Extensive experiments on five datasets show that the proposed LRGNN can achieve the best performance, and obtained data-specific GNNs with different depth and skip-connection schemes, which can better capture the long-range dependencies.
|
['Quanming Yao', 'Huan Zhao', 'Zhiqiang He', 'Lanning Wei']
|
2023-02-17
| null | null | null | null |
['graph-structure-learning', 'graph-classification']
|
['graphs', 'graphs']
|
[-2.95074545e-02 5.27138561e-02 -2.18946338e-01 -3.58158618e-01
1.83428854e-01 -2.60739494e-02 2.34643593e-01 -3.25177051e-02
-3.90875340e-01 3.44046533e-01 1.87952116e-01 -2.67676711e-01
-4.43629265e-01 -1.09714639e+00 -5.08818030e-01 -8.66990030e-01
-2.31734440e-02 -1.62533686e-01 4.49423492e-01 -3.26076150e-01
2.55731374e-01 4.35714215e-01 -1.15954316e+00 4.13865745e-02
1.17676580e+00 1.02703476e+00 3.56253892e-01 4.28055003e-02
-6.17257655e-01 5.60415268e-01 -4.20415044e-01 -2.84282804e-01
1.76125959e-01 -4.73606348e-01 -3.24991286e-01 -1.71761170e-01
-3.61568108e-02 -2.59970397e-01 -6.34896457e-01 1.18596613e+00
4.89885747e-01 1.22828394e-01 3.87991339e-01 -1.09820950e+00
-8.99101913e-01 8.34979713e-01 -6.40741587e-01 2.79351473e-01
-2.05412433e-01 1.56084523e-01 1.16504467e+00 -5.79445779e-01
8.99190232e-02 1.35991454e+00 7.77927101e-01 4.17514831e-01
-8.00323069e-01 -8.14191699e-01 6.39895558e-01 3.04443032e-01
-1.45890462e+00 -5.21905795e-02 1.22542632e+00 -5.20602800e-02
8.91069829e-01 4.07036804e-02 6.28218710e-01 9.06872272e-01
1.87965348e-01 7.25147724e-01 6.52038395e-01 -2.10525081e-01
-6.90941960e-02 -2.05614835e-01 3.85163516e-01 8.55248511e-01
5.54886818e-01 -5.13671152e-02 -6.14679568e-02 1.43373668e-01
1.14312577e+00 3.65956068e-01 -3.94909531e-01 -2.72825092e-01
-8.19202960e-01 7.15733051e-01 1.26622093e+00 5.35877824e-01
-2.73983330e-01 7.56261200e-02 4.76204306e-01 2.62456089e-01
3.17208022e-01 2.19003618e-01 -4.24026012e-01 3.28006208e-01
-5.13139844e-01 -2.80008703e-01 4.40433949e-01 9.02341068e-01
9.25823689e-01 1.32437631e-01 -1.90095797e-01 8.59394550e-01
3.53864074e-01 -8.61762092e-02 7.28183210e-01 9.05198008e-02
8.33208084e-01 1.33435309e+00 -5.77149272e-01 -1.41001225e+00
-6.49069309e-01 -9.07278776e-01 -1.38646603e+00 -1.53782368e-01
1.65733323e-01 -2.62078524e-01 -1.22646129e+00 1.83461440e+00
4.00173031e-02 2.12732106e-01 5.03349537e-03 8.42726409e-01
1.10876834e+00 7.64239788e-01 1.11313678e-01 -1.24673784e-01
1.01742351e+00 -1.08567035e+00 -6.84620261e-01 -5.20472109e-01
6.86347961e-01 -2.80239612e-01 1.16568923e+00 8.09755176e-02
-6.63099527e-01 -7.51157224e-01 -1.24824893e+00 -9.13394615e-02
-6.41575038e-01 5.26960194e-03 8.43249977e-01 5.19129574e-01
-9.25367594e-01 8.23353767e-01 -5.41638076e-01 -2.01167792e-01
3.83743554e-01 4.02789891e-01 -2.13262826e-01 -8.71509239e-02
-1.42652392e+00 5.21865785e-01 8.05396736e-01 7.77776420e-01
-3.42984170e-01 -3.72075915e-01 -7.24229515e-01 4.34245825e-01
6.29193485e-01 -3.32475334e-01 5.93493521e-01 -1.00953925e+00
-1.20123470e+00 1.65626481e-01 1.50907978e-01 -8.84826183e-02
2.26464882e-01 1.00527532e-01 -5.65214813e-01 -1.53889611e-01
-4.42962945e-01 3.17809969e-01 6.79973483e-01 -8.32398415e-01
-3.62589329e-01 -4.83297586e-01 1.66785419e-01 3.16204339e-01
-8.28425050e-01 -2.97588706e-01 -6.27813697e-01 -8.10143828e-01
4.48970914e-01 -7.55475998e-01 -3.63538891e-01 -2.74173021e-01
-5.55649877e-01 -4.20164436e-01 7.90431380e-01 -6.77702665e-01
1.74508107e+00 -2.22987890e+00 1.62360072e-01 3.54894400e-01
4.48294073e-01 5.97657919e-01 -3.48870784e-01 3.05848718e-01
-1.73213899e-01 2.61532754e-01 -1.31638095e-01 5.86080961e-02
-1.91920593e-01 3.34519625e-01 1.02973886e-01 1.09699689e-01
2.84652203e-01 1.09542835e+00 -6.88674510e-01 -4.30237234e-01
7.03089908e-02 3.11452746e-01 -4.10092086e-01 1.03891812e-01
-2.46738017e-01 1.32007793e-01 -8.25608253e-01 3.06043983e-01
8.54417622e-01 -3.88901055e-01 7.44599104e-02 -5.14434636e-01
1.35635272e-01 1.74720377e-01 -1.24710774e+00 1.47561312e+00
-3.17972571e-01 2.07746595e-01 1.52205780e-01 -1.17084599e+00
1.26252437e+00 -7.94143379e-02 2.93085948e-02 -7.64888644e-01
2.82619745e-01 2.60582715e-01 3.64463627e-01 -3.92005742e-01
1.48683950e-01 1.00922100e-01 6.68321773e-02 5.26645109e-02
-1.15949072e-01 5.36572695e-01 -1.51223820e-02 1.77698970e-01
1.00289476e+00 -1.51338205e-01 5.96641116e-02 -4.14325953e-01
7.31635332e-01 -4.79381919e-01 7.33396471e-01 4.61871833e-01
5.59202246e-02 4.24888492e-01 6.12876892e-01 -7.18472362e-01
-5.03330231e-01 -5.89240193e-01 1.98840395e-01 9.35267270e-01
4.37350392e-01 -4.88359869e-01 -6.85044527e-01 -8.70549798e-01
-1.88448355e-01 2.27682948e-01 -5.63623428e-01 -6.16975486e-01
-7.92059898e-01 -9.08478320e-01 5.08459628e-01 6.33544743e-01
8.67826343e-01 -1.24677122e+00 -7.97971338e-02 3.05554628e-01
1.61565274e-01 -7.75740862e-01 -5.06209671e-01 2.29411066e-01
-1.04235053e+00 -8.96212876e-01 -8.72694433e-01 -9.78524745e-01
8.88759136e-01 3.44553828e-01 7.35803723e-01 7.49997973e-01
1.25602573e-01 -4.54982102e-01 -4.18247789e-01 -4.48947921e-02
1.68436050e-01 6.68940902e-01 -3.58096302e-01 1.47947297e-01
3.76025736e-01 -9.30675983e-01 -6.01192713e-01 3.73522401e-01
-9.32535172e-01 2.49551281e-01 1.10277236e+00 8.48352015e-01
3.78961951e-01 2.93048054e-01 8.36875081e-01 -9.55926776e-01
9.29179490e-01 -3.35939407e-01 -5.62234402e-01 3.62713844e-01
-7.45882809e-01 3.69185925e-01 1.17056489e+00 -4.29374844e-01
-8.86759341e-01 -3.82850856e-01 -3.31972182e-01 -3.89681339e-01
8.38560462e-02 9.42197740e-01 -6.61455691e-01 -2.43099794e-01
2.84402400e-01 1.99395895e-01 -1.11062109e-01 -6.96284115e-01
5.41409850e-02 4.84931916e-01 6.04746975e-02 -3.48620832e-01
6.58795536e-01 3.54575552e-03 1.94547102e-01 -6.17927849e-01
-7.49973595e-01 -1.23248190e-01 -3.68747652e-01 1.39865234e-01
8.03288698e-01 -5.94228804e-01 -4.88499790e-01 7.50701189e-01
-1.05899489e+00 -1.58555850e-01 2.16430560e-01 3.64809275e-01
2.48876959e-01 4.87481534e-01 -7.05245614e-01 -5.38782060e-01
-5.26857197e-01 -1.18607891e+00 5.36458492e-01 7.88445532e-01
4.36608315e-01 -9.95392978e-01 -4.54931587e-01 -2.02339128e-01
6.16395354e-01 8.94806236e-02 1.44434321e+00 -8.38975608e-01
-7.16753542e-01 -9.03128088e-02 -7.89667308e-01 3.95551145e-01
2.48689279e-01 -2.72525609e-01 -6.42040908e-01 -2.58663177e-01
-1.92480043e-01 4.60681431e-02 1.05028653e+00 3.44176918e-01
1.49329102e+00 -3.40818644e-01 -3.89960974e-01 9.11564171e-01
1.49287438e+00 1.94269761e-01 7.32793450e-01 1.89697728e-01
1.25363290e+00 6.23059332e-01 1.38967484e-01 2.72116940e-02
3.38558674e-01 3.35357815e-01 5.88550091e-01 -2.60410339e-01
-1.63799211e-01 -5.16014457e-01 1.76249612e-02 1.14246213e+00
-1.28464207e-01 -3.55912745e-01 -7.28531778e-01 2.70631313e-01
-1.90861499e+00 -3.86460453e-01 -1.37368932e-01 1.89906740e+00
4.98628676e-01 4.58345622e-01 -1.66011557e-01 -8.23691860e-03
9.65704620e-01 5.70864797e-01 -7.54274309e-01 -3.08706135e-01
-8.78139958e-02 5.84463589e-02 4.37007874e-01 9.97984409e-02
-7.58448064e-01 8.02230120e-01 4.58866262e+00 1.22888279e+00
-1.18975317e+00 -2.23428488e-01 5.76913357e-01 3.07626992e-01
-4.32311743e-01 7.20844939e-02 -9.19717491e-01 5.87431431e-01
5.41491807e-01 9.20293182e-02 5.27731359e-01 7.64613628e-01
-8.78811553e-02 3.54273081e-01 -7.39817739e-01 1.00678837e+00
-1.02965362e-01 -1.13238072e+00 2.46640235e-01 4.99692522e-02
4.89194930e-01 -2.48032082e-02 -3.75967532e-01 5.20166278e-01
1.05107717e-01 -1.07415140e+00 2.62193412e-01 3.96673501e-01
5.57521224e-01 -8.51571620e-01 1.02887464e+00 4.80311900e-01
-1.56834579e+00 -2.95217186e-01 -6.66759789e-01 -1.60552487e-01
-5.80082573e-02 7.56876647e-01 -1.48486480e-01 1.01953042e+00
6.28262818e-01 8.42011511e-01 -8.35495174e-01 1.10448003e+00
-3.60815734e-01 4.54192132e-01 -3.67553622e-01 -4.73254979e-01
4.95705873e-01 -6.16071343e-01 2.69010425e-01 9.42577243e-01
4.00884360e-01 7.98163489e-02 2.46538058e-01 8.86822760e-01
-2.35487968e-01 1.77006677e-01 -3.99756938e-01 -2.13147610e-01
3.87781024e-01 1.29920888e+00 -8.30350757e-01 -1.07596228e-02
-4.48691726e-01 6.44294143e-01 6.71925366e-01 5.75507522e-01
-6.65809095e-01 -9.63948965e-01 2.54860818e-01 1.02297999e-01
4.08608496e-01 -1.58312708e-01 -2.30265066e-01 -9.48078394e-01
3.57094496e-01 -7.17703760e-01 6.03891790e-01 -6.10496223e-01
-1.28214276e+00 8.94178987e-01 -9.29733589e-02 -8.10376287e-01
3.35376412e-01 -7.31261551e-01 -9.82384801e-01 1.06542790e+00
-1.64057243e+00 -1.36409068e+00 -6.09135091e-01 5.80690384e-01
1.75291240e-01 -7.40631658e-04 3.58827323e-01 4.48230177e-01
-9.96532798e-01 8.02496195e-01 -2.04301715e-01 3.14017385e-01
3.60566050e-01 -8.38715732e-01 4.39209402e-01 8.60578358e-01
-1.26482293e-01 8.59012663e-01 8.11662823e-02 -7.05586433e-01
-1.27825034e+00 -1.05516636e+00 5.54700077e-01 2.98016578e-01
5.08828402e-01 -5.22236347e-01 -1.28616226e+00 5.91142058e-01
-8.25760737e-02 1.34358644e-01 1.46701768e-01 3.72118384e-01
-2.95393109e-01 -4.88357633e-01 -8.55526984e-01 7.80367553e-01
1.40865946e+00 -3.49000424e-01 -5.38381457e-01 -4.67987023e-02
1.01957858e+00 -9.17540416e-02 -5.20027101e-01 6.74594760e-01
3.61196816e-01 -9.93488371e-01 8.52599561e-01 -4.00279969e-01
2.18331411e-01 -3.93952191e-01 2.77708918e-01 -1.32557142e+00
-5.46229601e-01 -4.32805538e-01 1.06657378e-01 1.37555420e+00
4.75440949e-01 -1.13950980e+00 9.51010048e-01 3.88451785e-01
-3.00307333e-01 -1.07510078e+00 -6.77819848e-01 -7.28986800e-01
-3.13562974e-02 -1.00033600e-02 9.26366746e-01 9.33550179e-01
-3.40238929e-01 5.62308609e-01 -3.40882480e-01 2.91973263e-01
1.42163590e-01 1.26244083e-01 4.80315238e-01 -1.39171803e+00
-1.33040771e-01 -6.52473807e-01 -4.58662659e-01 -1.39380395e+00
3.44620831e-02 -8.28996062e-01 -9.05812308e-02 -1.79873788e+00
7.04357214e-03 -7.47418106e-01 -6.57309473e-01 5.70309460e-01
-4.56034660e-01 -3.35485071e-01 -5.87063050e-03 1.39760047e-01
-3.91651720e-01 7.20464647e-01 1.49689698e+00 -2.94158943e-02
-2.73231119e-01 -9.35851634e-02 -8.60832334e-01 6.78253829e-01
8.23567629e-01 -3.08298171e-01 -7.78264046e-01 -7.79255033e-01
3.25576931e-01 -1.81060016e-01 2.23336786e-01 -1.03107333e+00
4.01851922e-01 -5.18862084e-02 3.73729140e-01 -5.76827347e-01
2.92047597e-02 -8.50800812e-01 1.34058088e-01 4.36029822e-01
-5.91595396e-02 2.08715439e-01 1.83394521e-01 7.39879191e-01
-2.73206592e-01 -3.69160891e-01 4.80714947e-01 -2.30531693e-01
-8.98458660e-01 5.95710039e-01 1.63538262e-01 -7.82683492e-02
6.69649482e-01 -4.02265131e-01 -4.52627867e-01 -2.07320422e-01
-3.39932293e-01 4.58507866e-01 2.58190513e-01 4.74695802e-01
5.38760424e-01 -1.29320288e+00 -4.92289215e-01 5.20658970e-01
9.32950061e-03 4.01203662e-01 5.18531203e-01 7.00187624e-01
-3.63968968e-01 2.06030071e-01 -1.41040757e-01 -2.98884273e-01
-8.47219884e-01 5.92396080e-01 2.67254442e-01 -6.39789581e-01
-6.35915101e-01 9.95016515e-01 5.32716513e-01 -5.30127943e-01
2.33469188e-01 -5.12892008e-01 -4.45527136e-01 -1.09134302e-01
3.35129678e-01 6.22248538e-02 1.40731186e-01 -2.49595538e-01
-2.54297376e-01 6.84702575e-01 -3.24112952e-01 6.67976797e-01
1.25225425e+00 -8.71712491e-02 -2.28539675e-01 1.03292771e-01
1.18845105e+00 -2.08891779e-01 -1.09651804e+00 -2.88204640e-01
4.38343873e-03 -2.06495285e-01 2.11187646e-01 -4.56866860e-01
-1.57026184e+00 1.03493595e+00 2.79902577e-01 5.22270620e-01
1.31205630e+00 -2.07301483e-01 1.04799581e+00 3.57888192e-01
2.58134305e-01 -7.50258267e-01 -3.57758962e-02 5.58422506e-01
6.16444528e-01 -8.59013855e-01 -3.93809862e-02 -6.67180538e-01
-2.03478470e-01 1.15349030e+00 9.58101928e-01 -3.05406541e-01
6.37396693e-01 -1.31453529e-01 -2.37554356e-01 -3.30415338e-01
-2.98186421e-01 -1.18751526e-01 3.16117823e-01 4.57846016e-01
1.76454604e-01 -8.41156691e-02 -3.89760077e-01 9.91650522e-01
1.11399643e-01 -2.23470360e-01 1.66168183e-01 7.12371945e-01
-3.79622579e-01 -1.08192873e+00 1.26973882e-01 7.68989801e-01
-1.47239447e-01 -3.62559885e-01 -3.53934795e-01 8.28103840e-01
8.71538371e-02 7.31591821e-01 -1.16360381e-01 -6.46338463e-01
4.45027679e-01 -2.06136435e-01 1.26585916e-01 -4.72377181e-01
-5.57663620e-01 -5.88042699e-02 5.24282679e-02 -4.02661979e-01
-1.54573455e-01 1.11312710e-01 -1.31708050e+00 -2.80474126e-01
-7.83131659e-01 2.66317129e-01 3.68117929e-01 9.69197989e-01
5.56220591e-01 8.17120194e-01 4.66807753e-01 -6.47782743e-01
-5.94775856e-01 -1.05372763e+00 -5.67327857e-01 2.89373487e-01
1.94856092e-01 -5.65527081e-01 -4.54987288e-01 -6.98347211e-01]
|
[7.267803192138672, 6.2573747634887695]
|
29405175-8711-4daa-8bb4-daac07c17930
|
lexicon-infused-phrase-embeddings-for-named
|
1404.5367
| null |
http://arxiv.org/abs/1404.5367v1
|
http://arxiv.org/pdf/1404.5367v1.pdf
|
Lexicon Infused Phrase Embeddings for Named Entity Resolution
|
Most state-of-the-art approaches for named-entity recognition (NER) use semi
supervised information in the form of word clusters and lexicons. Recently
neural network-based language models have been explored, as they as a byproduct
generate highly informative vector representations for words, known as word
embeddings. In this paper we present two contributions: a new form of learning
word embeddings that can leverage information from relevant lexicons to improve
the representations, and the first system to use neural word embeddings to
achieve state-of-the-art results on named-entity recognition in both CoNLL and
Ontonotes NER. Our system achieves an F1 score of 90.90 on the test set for
CoNLL 2003---significantly better than any previous system trained on public
data, and matching a system employing massive private industrial query-log
data.
|
['Andrew McCallum', 'Vineet Kumar', 'Alexandre Passos']
|
2014-04-22
|
lexicon-infused-phrase-embeddings-for-named-1
|
https://aclanthology.org/W14-1609
|
https://aclanthology.org/W14-1609.pdf
|
ws-2014-6
|
['learning-word-embeddings']
|
['methodology']
|
[-3.44841897e-01 9.19328630e-02 -3.54827821e-01 -3.37546587e-01
-1.18951762e+00 -7.49181390e-01 5.88068724e-01 3.74195635e-01
-1.10867500e+00 5.70813477e-01 7.09470809e-01 -2.65401751e-01
1.07990399e-01 -7.62745202e-01 -2.32220963e-01 -3.05884182e-01
3.68157998e-02 7.83696949e-01 -3.53699550e-02 -2.35684261e-01
2.20080614e-01 8.42079043e-01 -8.77548277e-01 7.67941400e-02
5.46218276e-01 5.45764089e-01 -1.72587380e-01 5.98785818e-01
-8.14063251e-01 5.53647995e-01 -8.40876818e-01 -5.03557086e-01
1.20777734e-01 9.31700990e-02 -1.06121981e+00 -2.52770215e-01
2.84157544e-01 1.09330386e-01 -6.09194636e-01 8.72159719e-01
8.77837598e-01 3.88267517e-01 6.46886647e-01 -6.63447917e-01
-1.25639927e+00 7.17024446e-01 2.12104786e-02 2.98951268e-01
1.72764495e-01 -1.70318887e-01 1.50766349e+00 -1.03189039e+00
9.23627734e-01 9.73001659e-01 8.17998946e-01 7.49427497e-01
-1.15520525e+00 -6.50093079e-01 -9.53149199e-02 1.33896042e-02
-1.52411962e+00 -3.80907208e-01 4.64574128e-01 -2.99046785e-01
1.66562724e+00 -2.01835647e-01 2.77572960e-01 1.05076671e+00
-5.68813793e-02 8.07234406e-01 8.10023963e-01 -7.88264036e-01
1.93760425e-01 3.14985663e-01 6.25102818e-01 5.95060229e-01
4.50102419e-01 9.54769775e-02 -4.87020940e-01 -5.04808903e-01
7.45459735e-01 -1.12020046e-01 -6.93090037e-02 -1.50843859e-01
-1.16766131e+00 1.07741129e+00 2.75897026e-01 8.18331540e-01
-6.72711134e-01 3.14466730e-02 4.45095181e-01 1.52448609e-01
7.88030088e-01 9.42544341e-01 -1.15984595e+00 -2.10454687e-01
-9.12098467e-01 -1.42160490e-01 1.27411723e+00 8.13121974e-01
6.30636394e-01 2.80786067e-01 -1.97124422e-01 1.12109530e+00
4.01325822e-01 2.54431307e-01 9.44727659e-01 -3.38726431e-01
3.79200995e-01 6.53897226e-01 5.89441974e-03 -3.65619868e-01
-3.92199099e-01 -2.49692634e-01 -4.19027239e-01 -1.30950242e-01
3.13826114e-01 -4.94137764e-01 -1.36293471e+00 1.46217251e+00
9.18049142e-02 1.04686528e-01 3.99174243e-01 3.97261202e-01
8.25840533e-01 6.72972083e-01 4.99975771e-01 2.55905986e-01
1.56930006e+00 -8.18054914e-01 -8.20840597e-01 -1.99316368e-01
1.08550918e+00 -8.38574708e-01 7.09485710e-01 -1.41251124e-02
-6.57168627e-01 -5.03874242e-01 -7.72351921e-01 -1.25033528e-01
-1.27123129e+00 2.03475371e-01 5.97277343e-01 1.03648472e+00
-1.00935626e+00 5.30635476e-01 -7.22847641e-01 -8.29551995e-01
3.50532800e-01 5.07905126e-01 -6.81808770e-01 -8.92154276e-02
-1.21108711e+00 1.01329470e+00 7.64413059e-01 -3.17587346e-01
-5.07519722e-01 -8.51074576e-01 -1.12865698e+00 2.78348327e-01
6.80865347e-02 -3.15499097e-01 1.18125653e+00 -1.61435649e-01
-1.25533259e+00 9.88896608e-01 -7.45350271e-02 -3.70718181e-01
-3.05838823e-01 -5.62782764e-01 -9.45576370e-01 -1.19393781e-01
1.56252325e-01 6.02034271e-01 1.14740683e-02 -1.01865125e+00
-6.05098784e-01 -2.53234416e-01 -2.91468292e-01 -1.84845641e-01
-8.74765337e-01 3.57999802e-01 -2.36165524e-01 -7.26351082e-01
-3.46775562e-01 -7.40328312e-01 -6.30121708e-01 -5.25086939e-01
-6.67408705e-01 -9.23119605e-01 2.92387903e-01 -7.19060183e-01
1.37003922e+00 -2.05437136e+00 -4.83260542e-01 2.32588634e-01
3.67612690e-02 7.25244999e-01 -4.76640433e-01 6.89667702e-01
-5.17913580e-01 6.25932515e-01 2.30054021e-01 -3.36319894e-01
3.28440338e-01 2.04925895e-01 -2.48196647e-01 2.35216826e-01
5.08996904e-01 1.06372404e+00 -8.36280704e-01 -4.19273168e-01
1.86815053e-01 6.42034054e-01 -2.98820108e-01 2.69670814e-01
2.65926749e-01 -3.59444946e-01 -4.56240833e-01 4.94694084e-01
2.35456809e-01 1.50043089e-02 3.18232566e-01 -1.76544428e-01
-2.85996109e-01 5.91656506e-01 -1.21376395e+00 1.71018493e+00
-6.13066316e-01 5.60041308e-01 -3.13331813e-01 -7.87776649e-01
1.10344672e+00 8.11531723e-01 4.88725722e-01 -2.13269621e-01
1.41259998e-01 1.94524392e-01 -5.32828987e-01 -2.78943390e-01
7.13802397e-01 -1.48733780e-01 -4.39754635e-01 2.86192119e-01
9.26756859e-01 2.07166567e-01 2.81220198e-01 1.09487891e-01
1.28809524e+00 -1.51514560e-01 5.73821962e-01 -2.15982810e-01
2.35281065e-01 3.16396892e-01 6.60728037e-01 6.68403864e-01
-9.41391662e-02 4.85209942e-01 4.27435488e-02 -5.81055760e-01
-1.07550037e+00 -1.05980372e+00 -1.00973248e-01 1.38871312e+00
-4.47293431e-01 -4.91451770e-01 -5.49245119e-01 -9.95222211e-01
-6.83881119e-02 1.15657163e+00 -3.34505945e-01 -2.03951336e-02
-7.50193298e-01 -6.22033119e-01 1.14993179e+00 1.02977586e+00
-3.83088142e-02 -1.38174582e+00 1.80941179e-01 5.26753366e-01
1.72575295e-01 -1.19121838e+00 -5.63496709e-01 8.48830938e-01
-8.21439862e-01 -9.77894545e-01 -8.64862740e-01 -1.37238467e+00
4.47706968e-01 -2.86624044e-01 1.41710401e+00 -3.88643771e-01
-5.10103166e-01 5.33770502e-01 -3.46765965e-01 -5.89927733e-01
-2.71915823e-01 5.15570462e-01 2.22637907e-01 -2.73956418e-01
1.17855954e+00 -2.81779498e-01 2.84462478e-02 -8.18349048e-02
-9.95091558e-01 -1.00622129e+00 8.16881835e-01 9.25399840e-01
4.05912340e-01 -2.73677051e-01 5.67739904e-01 -1.07587457e+00
8.79631102e-01 -4.29284781e-01 -3.91875863e-01 4.77208018e-01
-8.87033045e-01 4.38876331e-01 3.81151885e-01 -6.03294432e-01
-1.01154220e+00 3.31633270e-01 -5.41721582e-01 -1.95937246e-01
-7.66899586e-01 4.82257634e-01 -1.58855245e-01 2.23727420e-01
7.22330987e-01 1.33789167e-01 -7.77297795e-01 -1.02327347e+00
9.84904706e-01 1.11201060e+00 4.82805312e-01 -5.18710971e-01
8.91785622e-01 1.93347703e-04 -5.44875681e-01 -1.18311477e+00
-9.00086403e-01 -1.34429359e+00 -9.65505302e-01 4.39946353e-01
1.29879570e+00 -1.06494713e+00 -2.13503838e-01 4.93328739e-03
-1.48376286e+00 1.10472582e-01 -8.39473248e-01 8.26096356e-01
-6.77171648e-02 1.40563801e-01 -9.97211635e-01 -7.43608057e-01
-5.01768053e-01 -4.82497692e-01 9.18578207e-01 5.17552257e-01
-4.69091147e-01 -1.37885702e+00 6.69120312e-01 1.98953867e-01
3.65635455e-01 -1.84090927e-01 9.37570333e-01 -2.01219630e+00
-1.52820665e-02 -5.87320030e-01 -2.18610972e-01 5.47255576e-01
1.34262919e-01 -6.19102597e-01 -1.09250212e+00 -6.97264671e-02
-5.17713308e-01 -3.22085857e-01 8.51086378e-01 2.65520327e-02
4.94040012e-01 -6.81393519e-02 -4.25976217e-01 3.29150438e-01
1.52246916e+00 1.98601499e-01 5.97385943e-01 5.08124173e-01
6.85642064e-01 3.06468397e-01 1.27106577e-01 2.08854437e-01
2.24884033e-01 2.25915045e-01 -1.59183592e-01 -4.35050309e-01
-2.09555238e-01 -3.85827303e-01 2.63954788e-01 1.23435879e+00
2.19803322e-02 -5.04846036e-01 -1.11462438e+00 1.22741365e+00
-1.40903294e+00 -7.96236813e-01 -3.78912762e-02 1.73549759e+00
9.64323878e-01 -3.90670672e-02 -2.97454178e-01 -3.24087143e-01
8.93741012e-01 2.53394872e-01 -2.12045506e-01 -4.76238549e-01
-2.65567396e-02 1.03825474e+00 8.76552641e-01 2.62888044e-01
-1.28223407e+00 1.43531394e+00 7.11638212e+00 9.26797628e-01
-5.66595674e-01 4.16077554e-01 1.50107443e-01 2.80161679e-01
-9.18791592e-02 -1.18890971e-01 -1.32546210e+00 -1.95001215e-01
1.49381506e+00 -1.96961671e-01 -5.65569028e-02 1.18956113e+00
-3.26304287e-01 5.73321879e-01 -9.30385292e-01 9.19034481e-01
3.84455770e-01 -1.32692218e+00 1.04235664e-01 3.25670421e-01
7.88437486e-01 5.47915816e-01 -3.79074275e-01 7.42070973e-01
1.04147422e+00 -9.07594025e-01 1.26587108e-01 3.58056307e-01
8.07082355e-01 -8.40638995e-01 1.19031298e+00 1.03786746e-02
-1.30776083e+00 1.25490934e-01 -5.10627270e-01 4.33371246e-01
4.68563676e-01 6.09234929e-01 -9.50035930e-01 5.09900272e-01
4.05827165e-01 3.12710166e-01 -4.54542637e-01 1.10453153e+00
-3.79548907e-01 9.52925146e-01 -2.46435270e-01 -3.16030800e-01
3.45005780e-01 3.30098420e-01 2.94138670e-01 1.86947310e+00
1.15497366e-01 -1.75485298e-01 4.92930084e-01 6.24094427e-01
-5.52357316e-01 4.10645813e-01 -7.99034774e-01 -7.89750278e-01
4.49911118e-01 1.26868618e+00 -5.40295362e-01 -5.06871879e-01
-4.86475796e-01 1.10290051e+00 4.29175764e-01 2.93341935e-01
-3.59991372e-01 -1.21990454e+00 9.32368755e-01 -2.81975955e-01
6.14601195e-01 -6.74258173e-01 -1.94468588e-01 -1.30525863e+00
-4.00986910e-01 -4.31952626e-01 6.34695113e-01 -4.02701586e-01
-1.92466617e+00 9.65237558e-01 -2.70319998e-01 -7.54988194e-01
-2.70087093e-01 -1.17501676e+00 -5.51903009e-01 8.87697458e-01
-1.56173193e+00 -1.15233469e+00 3.05792958e-01 4.62259918e-01
6.04115665e-01 -5.75304151e-01 1.54427624e+00 5.17638028e-01
-4.22690690e-01 5.79011619e-01 4.01340425e-01 1.07015693e+00
9.68534231e-01 -1.59534407e+00 8.80761921e-01 6.11331046e-01
1.03686512e+00 9.74203289e-01 6.85830638e-02 -5.78589380e-01
-9.28491235e-01 -1.10335231e+00 1.79769683e+00 -8.05844128e-01
9.18110788e-01 -4.15265709e-01 -5.88531554e-01 7.82781243e-01
5.14014602e-01 1.10940747e-01 1.43532884e+00 6.62830532e-01
-5.80083728e-01 3.17001700e-01 -1.00399005e+00 3.41875702e-01
8.75086308e-01 -9.70348179e-01 -1.29161441e+00 5.36082327e-01
9.79649425e-01 2.51992732e-01 -1.23690951e+00 -1.12433225e-01
2.53619760e-01 -1.95194930e-01 1.12042212e+00 -1.22886181e+00
-2.37789392e-01 -8.54947418e-02 -2.52851576e-01 -1.44418800e+00
-5.19902885e-01 -4.19081450e-01 7.33652934e-02 1.78451979e+00
7.68743038e-01 -5.20982444e-01 8.07568133e-01 4.93790925e-01
-1.33641198e-01 -3.34569067e-01 -9.92025435e-01 -1.02014267e+00
3.01214308e-01 -5.14502525e-01 4.16599661e-01 1.21319962e+00
5.00735082e-02 6.43192768e-01 9.59981754e-02 1.59646004e-01
3.02049220e-01 -3.62987429e-01 2.71982998e-01 -1.50565195e+00
2.57807106e-01 -1.03571773e-01 -5.99093318e-01 -1.10031164e+00
6.29376590e-01 -1.02646184e+00 2.49044433e-01 -1.77777958e+00
-5.00880890e-02 -3.08634520e-01 -8.54274452e-01 8.68091583e-01
6.05833717e-02 3.71303141e-01 1.60698399e-01 -9.46825519e-02
-6.24985933e-01 3.39723587e-01 2.03543946e-01 -1.95070148e-01
-2.64865935e-01 -4.56160605e-01 -6.31109774e-01 6.82163417e-01
8.63962829e-01 -9.97692525e-01 2.96763778e-01 -4.78240401e-01
-7.01064616e-02 -5.01883566e-01 -1.41251430e-01 -8.76224101e-01
3.96506697e-01 3.06655541e-02 5.33281446e-01 -5.96403480e-01
2.66353279e-01 -6.12542868e-01 -5.07991970e-01 1.39236122e-01
-4.53691363e-01 3.89653653e-01 1.60847291e-01 6.99172795e-01
-3.35002035e-01 -5.09122074e-01 3.76237571e-01 -3.33120108e-01
-1.10782373e+00 1.00639164e-01 -6.06693029e-01 4.80758280e-01
6.18538499e-01 1.24135025e-01 -5.24713211e-02 2.35524122e-02
-7.89402306e-01 -1.27960026e-01 -4.15338837e-02 6.62962258e-01
2.68480659e-01 -1.54319680e+00 -6.66752517e-01 1.15325995e-01
3.01079541e-01 -5.60981333e-01 -3.20066929e-01 6.27109483e-02
-3.61558288e-01 7.72652984e-01 8.15767273e-02 6.38896674e-02
-9.95554328e-01 4.35026258e-01 -2.97444668e-02 -9.27539110e-01
-3.17882508e-01 9.05523896e-01 -2.76799262e-01 -1.10072601e+00
1.93412647e-01 -2.28491917e-01 -4.65503544e-01 3.26995820e-01
5.41297197e-01 2.98007488e-01 1.66755348e-01 -7.28855371e-01
-5.48232853e-01 4.78981644e-01 -1.26204804e-01 -2.76539326e-01
1.65992677e+00 2.95706272e-01 1.92369983e-01 4.91254628e-01
1.39221263e+00 3.09602499e-01 -1.96160942e-01 -3.81157041e-01
7.11284637e-01 6.11880012e-02 2.00434640e-01 -8.52308154e-01
-7.36172259e-01 8.07853937e-01 7.76259959e-01 1.42005265e-01
5.17007411e-01 2.24897355e-01 1.16986394e+00 1.05390036e+00
2.18932375e-01 -1.47404039e+00 -8.45122784e-02 9.85795021e-01
9.14327800e-02 -1.14625132e+00 -2.88467765e-01 -8.74219835e-02
-5.53821027e-01 1.40201855e+00 2.11171865e-01 -2.46448383e-01
8.94539714e-01 3.24639797e-01 5.71264625e-01 -3.13722968e-01
-3.77275825e-01 -6.94361389e-01 4.80604202e-01 7.98504055e-01
8.45766842e-01 -1.08027374e-02 -3.08773607e-01 9.50984240e-01
8.25141296e-02 -1.09877087e-01 1.64036050e-01 1.09509182e+00
-4.10560310e-01 -1.56249595e+00 -2.09865659e-01 3.55953842e-01
-9.86897707e-01 -5.42574346e-01 -5.88194489e-01 8.58547270e-01
5.52711226e-02 1.11565948e+00 -6.69555785e-03 -3.49827677e-01
6.93503022e-01 9.30195928e-01 -2.21550584e-01 -1.23561001e+00
-1.01603711e+00 -2.11825907e-01 5.34465790e-01 -5.13371170e-01
-2.23592296e-01 -2.50901133e-01 -1.37750185e+00 1.29494965e-01
-9.64049578e-01 5.95138788e-01 9.12387490e-01 9.59093392e-01
5.50904810e-01 5.17994046e-01 2.67133296e-01 -6.76423550e-01
-7.09860384e-01 -1.17249084e+00 -9.26952004e-01 5.52206397e-01
-2.12191716e-01 -1.77589774e-01 -2.95310855e-01 1.79646567e-01]
|
[9.744842529296875, 9.579429626464844]
|
9d6401f9-44ec-4bf1-a2b3-51da492e3a07
|
collective-knowledge-graph-completion-with
|
2305.15895
| null |
https://arxiv.org/abs/2305.15895v1
|
https://arxiv.org/pdf/2305.15895v1.pdf
|
Collective Knowledge Graph Completion with Mutual Knowledge Distillation
|
Knowledge graph completion (KGC), the task of predicting missing information based on the existing relational data inside a knowledge graph (KG), has drawn significant attention in recent years. However, the predictive power of KGC methods is often limited by the completeness of the existing knowledge graphs from different sources and languages. In monolingual and multilingual settings, KGs are potentially complementary to each other. In this paper, we study the problem of multi-KG completion, where we focus on maximizing the collective knowledge from different KGs to alleviate the incompleteness of individual KGs. Specifically, we propose a novel method called CKGC-CKD that uses relation-aware graph convolutional network encoder models on both individual KGs and a large fused KG in which seed alignments between KGs are regarded as edges for message propagation. An additional mutual knowledge distillation mechanism is also employed to maximize the knowledge transfer between the models of "global" fused KG and the "local" individual KGs. Experimental results on multilingual datasets have shown that our method outperforms all state-of-the-art models in the KGC task.
|
['Yi-Ke Guo', 'Jiahao Sun', 'Ovidiu Serban', 'Weihang Zhang']
|
2023-05-25
| null | null | null | null |
['knowledge-graph-completion']
|
['knowledge-base']
|
[-2.77781636e-01 7.70573199e-01 -5.50851822e-01 -2.08551392e-01
-5.93072116e-01 -4.70097423e-01 4.79785532e-01 4.40046817e-01
-8.82404745e-02 9.86168087e-01 3.14444304e-01 -2.72643447e-01
-3.12048614e-01 -1.11808014e+00 -1.23273206e+00 -3.30212235e-01
-1.95977598e-01 5.41189909e-01 7.30676726e-02 -2.50445843e-01
-3.56888056e-01 -7.69979600e-03 -8.96948993e-01 3.53460222e-01
1.24776959e+00 5.84917486e-01 5.32111406e-01 2.39890844e-01
-2.87122726e-01 1.43315268e+00 -2.03955352e-01 -1.15566242e+00
-7.34663084e-02 -8.27540010e-02 -1.21579480e+00 -1.97539270e-01
3.01117361e-01 2.99642533e-02 -7.56903410e-01 1.22449911e+00
1.71797395e-01 -3.49711329e-01 3.06302547e-01 -1.47064471e+00
-1.16605735e+00 1.59629345e+00 -5.01946926e-01 -3.38806480e-01
1.31069094e-01 -2.80084878e-01 1.29102361e+00 -6.59957886e-01
9.97137666e-01 1.19143462e+00 6.75540090e-01 3.04267239e-02
-9.68918741e-01 -6.34899557e-01 3.97815764e-01 7.09471285e-01
-1.70869851e+00 -1.84278637e-01 7.57970631e-01 -3.66822511e-01
1.12479031e+00 -3.71750087e-01 6.65563345e-01 5.89672029e-01
-2.58393250e-02 7.96905637e-01 8.01809371e-01 -4.87685233e-01
-3.87846857e-01 1.31548777e-01 6.68196976e-02 1.24565542e+00
6.16393089e-01 -3.98927420e-01 -8.68986547e-01 3.20691206e-02
5.56256592e-01 -4.28510010e-01 -5.71744800e-01 -5.76567292e-01
-1.19882286e+00 6.42019451e-01 7.04459548e-01 4.75036614e-02
-3.37911159e-01 3.05656999e-01 1.22334793e-01 3.77798200e-01
6.49370790e-01 2.07117036e-01 -8.17511439e-01 3.26272696e-01
-3.33466083e-01 5.64434044e-02 1.13230002e+00 1.45549917e+00
1.12944603e+00 -2.65299678e-01 2.55007416e-01 6.26690984e-01
3.78329605e-01 4.23999965e-01 -1.62277301e-03 -5.79328358e-01
1.17347562e+00 1.13082135e+00 -2.11221591e-01 -1.22048426e+00
-2.21319914e-01 -8.45753014e-01 -9.98906314e-01 -5.36577940e-01
1.45399928e-01 -8.23238268e-02 -7.47364104e-01 1.91357601e+00
5.25124907e-01 2.88524359e-01 5.11697650e-01 5.36501825e-01
1.13651514e+00 2.72955090e-01 1.36978000e-01 -1.32254483e-02
1.00672317e+00 -1.07678461e+00 -8.46000552e-01 -1.79445609e-01
1.11170411e+00 -5.95608711e-01 3.37895989e-01 -4.73752357e-02
-7.92466760e-01 -2.67317772e-01 -1.00413775e+00 -2.79222041e-01
-7.44694412e-01 1.57113403e-01 9.27720249e-01 1.65979370e-01
-1.22220516e+00 5.12566686e-01 -7.05301523e-01 -1.45923570e-01
3.57051194e-01 2.47680694e-01 -7.90785611e-01 -6.40048683e-01
-1.67721355e+00 1.03391063e+00 1.04635584e+00 3.91736180e-01
-8.09139013e-01 -8.51066232e-01 -1.14136040e+00 1.74690858e-01
9.01395559e-01 -8.49799395e-01 7.12792218e-01 -5.17588615e-01
-8.65237117e-01 6.17197275e-01 8.29929933e-02 -4.24588054e-01
2.22841904e-01 -2.26463079e-01 -5.99785686e-01 4.01236303e-02
3.71927828e-01 4.88618314e-01 3.69865030e-01 -1.37297440e+00
-6.35260701e-01 -3.76710206e-01 2.70985305e-01 5.29285431e-01
2.16666777e-02 -3.20499122e-01 -1.04172695e+00 -5.02613008e-01
1.62412688e-01 -7.40471482e-01 5.37404343e-02 -6.44357204e-01
-7.36495197e-01 -3.54860216e-01 6.12668574e-01 -1.12426901e+00
1.25495958e+00 -1.55856252e+00 4.80529875e-01 3.84589851e-01
5.80896556e-01 1.53934225e-01 -1.77572325e-01 8.12900841e-01
7.90313408e-02 -6.93035871e-03 -9.07256752e-02 -2.50209242e-01
-1.28653035e-01 6.46165311e-01 -1.86042953e-02 1.69331983e-01
1.58092782e-01 1.37744725e+00 -1.18486357e+00 -7.48673499e-01
-2.41188690e-01 4.37665612e-01 -2.92712718e-01 2.56560389e-02
-3.94435018e-01 3.78620207e-01 -3.95437717e-01 8.28754485e-01
7.15883374e-01 -7.47993410e-01 9.60322261e-01 -4.17823344e-01
2.84181952e-01 2.75711805e-01 -1.09257162e+00 2.05685520e+00
-3.30268294e-01 1.24848731e-01 5.47800958e-02 -1.14767909e+00
6.76066756e-01 2.85262167e-01 2.81274855e-01 -3.98638636e-01
-4.20211405e-01 3.35028023e-01 -3.66468206e-02 -2.28804156e-01
7.30567336e-01 1.94757044e-01 2.76972670e-02 2.68996000e-01
5.93346477e-01 1.72794461e-01 3.03968698e-01 1.04973066e+00
1.01702380e+00 2.63149977e-01 3.43137980e-01 -4.33718748e-02
4.44023579e-01 6.50679246e-02 7.21058071e-01 6.00281894e-01
1.42136201e-01 8.86302814e-03 7.15037405e-01 -4.11965586e-02
-6.35410428e-01 -7.44132638e-01 4.09482419e-01 6.16784513e-01
2.83992976e-01 -8.03827167e-01 -4.95520711e-01 -1.09326088e+00
4.33910549e-01 4.90514308e-01 -4.98071402e-01 -2.50808567e-01
-5.08535504e-01 -6.83592141e-01 7.03510344e-01 5.42475760e-01
5.59644043e-01 -6.87199712e-01 4.44384843e-01 3.06386590e-01
-6.49577081e-01 -1.69180775e+00 -2.96806633e-01 2.33821124e-02
-5.72865725e-01 -1.59450209e+00 -3.05316508e-01 -9.51411724e-01
7.95159221e-01 2.20765248e-01 1.37431097e+00 9.13168266e-02
8.28223675e-02 5.98662138e-01 -5.70455313e-01 -2.65824683e-02
-3.55600029e-01 4.39025551e-01 -1.81491494e-01 9.89275128e-02
1.53878093e-01 -5.82084715e-01 -2.71468192e-01 -3.30485702e-02
-7.36238182e-01 6.65638268e-01 9.17357385e-01 6.02947593e-01
6.67912483e-01 4.46254522e-01 8.88277173e-01 -1.36806238e+00
4.60839629e-01 -7.70817339e-01 -4.54418749e-01 9.61527765e-01
-9.75954294e-01 3.66709113e-01 4.36921626e-01 2.64107599e-03
-1.35758460e+00 -9.16224197e-02 3.58244091e-01 -3.25743943e-01
5.13958216e-01 1.49519992e+00 -5.10689616e-01 -2.51746178e-01
1.07646063e-01 1.69136286e-01 -2.42994875e-01 -6.66009843e-01
9.47219551e-01 2.88574785e-01 6.98826373e-01 -8.04014146e-01
6.66123748e-01 1.58501074e-01 3.85541059e-02 -4.21583921e-01
-9.44382608e-01 -4.63104039e-01 -8.41628551e-01 -2.49482974e-01
5.21246016e-01 -1.45402658e+00 -5.29938459e-01 6.22830808e-01
-1.37419856e+00 -2.49097958e-01 -8.72059390e-02 5.18158615e-01
-1.40117690e-01 5.14293253e-01 -6.69051766e-01 -4.44574714e-01
-3.83586943e-01 -8.72861743e-01 9.98644888e-01 -1.22869732e-02
3.98143917e-01 -1.31551433e+00 -6.39742315e-02 6.57803893e-01
3.06505170e-02 8.45276266e-02 1.27515602e+00 -5.55940628e-01
-1.11757207e+00 -7.46454000e-02 -5.08914649e-01 3.13766867e-01
2.52739787e-01 -4.65461224e-01 -4.45022196e-01 -2.32762992e-01
-8.49968135e-01 -5.33990383e-01 9.81619537e-01 -9.86086950e-02
6.82823062e-01 -5.61639428e-01 -6.64004982e-01 4.30091292e-01
1.70653379e+00 -3.32686096e-01 5.22630274e-01 5.43464348e-02
1.42361856e+00 7.17121542e-01 2.47414649e-01 -5.57382070e-02
1.44017279e+00 3.91312391e-01 5.35926163e-01 5.71786426e-02
-5.08730352e-01 -8.79288673e-01 2.65080720e-01 1.55054927e+00
-2.35256970e-01 -3.39437217e-01 -9.80305672e-01 9.53294694e-01
-2.21499419e+00 -4.82525080e-01 -3.82264435e-01 1.96183705e+00
1.13728178e+00 -3.03313315e-01 -7.22681046e-01 -4.13888425e-01
9.71224070e-01 9.15388837e-02 -3.27222079e-01 3.75240743e-01
-6.14705145e-01 4.48054336e-02 8.06713521e-01 7.64901280e-01
-9.55585778e-01 1.29604876e+00 5.05615997e+00 8.84479284e-01
-3.16606581e-01 1.79224163e-01 1.53343782e-01 3.58353078e-01
-4.61116672e-01 5.56101024e-01 -9.78491902e-01 1.92886651e-01
6.24604344e-01 -2.49848738e-01 4.50624049e-01 6.03519857e-01
-4.72560674e-01 -1.49844661e-01 -9.15809989e-01 6.28110051e-01
3.75546850e-02 -1.51360321e+00 2.88333297e-01 9.25999805e-02
1.14650691e+00 3.24856102e-01 -4.11644638e-01 6.37074232e-01
1.22371769e+00 -7.42232144e-01 4.13465708e-01 8.32672000e-01
8.34756196e-01 -7.34471858e-01 8.66947651e-01 3.98503482e-01
-1.53716230e+00 3.21533531e-01 -3.74088764e-01 2.67585814e-01
1.33131221e-01 8.02071929e-01 -9.65535641e-01 1.71184635e+00
5.06425083e-01 9.96124864e-01 -6.53735459e-01 5.00371218e-01
-7.67662942e-01 4.10009891e-01 -1.08025141e-01 3.45016807e-01
8.59428719e-02 -2.23397627e-01 1.86416462e-01 1.00105476e+00
9.86812711e-02 -1.19899595e-02 4.04566854e-01 8.35695267e-01
-7.22560287e-01 1.69701338e-01 -7.40179360e-01 -2.83193707e-01
5.22692919e-01 1.27996647e+00 -1.79876342e-01 -5.04658341e-01
-7.98857331e-01 8.56143713e-01 1.18815529e+00 5.70127368e-01
-4.62233692e-01 -3.23403567e-01 2.43065670e-01 -1.46101788e-01
3.12272191e-01 -1.66543856e-01 2.36264616e-01 -1.48171091e+00
2.06889346e-01 -4.72401589e-01 7.01014936e-01 -9.22134578e-01
-1.35199845e+00 1.58562332e-01 -3.29393260e-02 -5.95933080e-01
-6.54340535e-02 -1.91414788e-01 -2.38063931e-02 1.17899930e+00
-2.05912590e+00 -1.92057145e+00 -2.95239426e-02 8.52966309e-01
-3.08872074e-01 -1.32667333e-01 6.67190433e-01 4.08608705e-01
-2.54970849e-01 4.99949574e-01 7.61813596e-02 4.25778270e-01
5.78671217e-01 -1.30797911e+00 2.00109124e-01 9.49423313e-01
1.48481190e-01 6.21113300e-01 6.61214963e-02 -1.50725210e+00
-1.54247093e+00 -1.58307755e+00 1.51806748e+00 -2.36023515e-01
9.90528166e-01 -1.39878049e-01 -1.04689825e+00 1.29390335e+00
9.21510458e-02 1.06500380e-01 6.23113215e-01 5.69397032e-01
-7.14869678e-01 -1.57509670e-01 -5.83982050e-01 3.87411386e-01
1.19461811e+00 -7.55315900e-01 -3.11405033e-01 4.50572610e-01
1.13897419e+00 -4.94013190e-01 -1.39259839e+00 6.07943475e-01
2.99977750e-01 -3.94033939e-01 7.55341172e-01 -6.76494658e-01
4.58099455e-01 -3.52656275e-01 -1.95033312e-01 -1.58114421e+00
-1.64839551e-01 -2.54818708e-01 -6.11447453e-01 1.39470887e+00
5.09475350e-01 -6.65189862e-01 8.39502752e-01 2.78966993e-01
-3.10434490e-01 -5.12406528e-01 -7.32326746e-01 -6.87566459e-01
-1.01086229e-01 -3.49201053e-01 6.68598354e-01 1.47618353e+00
4.03800249e-01 6.78798139e-01 -6.65330410e-01 6.93691671e-01
9.34792697e-01 3.18598837e-01 6.87850595e-01 -1.23435521e+00
-3.40732634e-01 1.21699646e-01 -3.59686881e-01 -8.71805012e-01
4.89019036e-01 -1.62803972e+00 -2.87493378e-01 -1.98792863e+00
4.74169105e-01 -5.47894239e-01 -1.54625565e-01 9.98032749e-01
-4.46473241e-01 -4.09164220e-01 5.21242991e-03 3.52293670e-01
-8.73521030e-01 7.54411876e-01 1.43319738e+00 -3.68517756e-01
1.52025148e-01 -5.62516689e-01 -8.99733186e-01 4.03677702e-01
4.72009212e-01 -4.57147241e-01 -7.38421023e-01 -7.33174801e-01
8.11095595e-01 2.96363652e-01 3.89616966e-01 -4.11231965e-01
7.72348344e-01 8.27324614e-02 -3.72408777e-02 -5.98209500e-01
1.38548259e-02 -6.54234052e-01 5.15290856e-01 1.63305357e-01
-1.77062958e-01 -2.70562977e-01 -1.21376626e-01 1.13038087e+00
-5.20867348e-01 7.42782876e-02 4.14003618e-02 -4.34574515e-01
-8.99978280e-01 4.93406236e-01 3.38907719e-01 1.52815282e-01
6.71262205e-01 3.97073418e-01 -7.62699246e-01 -4.01819676e-01
-8.74171734e-01 8.41766298e-01 8.10958445e-02 4.11908478e-01
6.30333006e-01 -1.47542572e+00 -9.53242719e-01 -4.99107353e-02
4.64415610e-01 4.81600195e-01 3.91655892e-01 9.41483557e-01
-2.95731753e-01 7.18553722e-01 2.87051439e-01 -6.08365284e-03
-1.10426974e+00 5.09342968e-01 1.88913807e-01 -9.72532392e-01
-4.97684956e-01 7.95972884e-01 1.50907487e-01 -8.47289741e-01
1.57735676e-01 -1.18149921e-01 -2.60167301e-01 -1.39692843e-01
-1.01201773e-01 2.60759562e-01 2.36878425e-01 -7.71858275e-01
-2.13884860e-01 1.37912780e-01 -4.42259312e-01 2.73405135e-01
1.27913177e+00 -3.68615329e-01 -6.42848730e-01 1.11511283e-01
9.95869160e-01 -1.89520404e-01 -7.64976561e-01 -9.55089927e-01
1.79235771e-01 -1.02135859e-01 -8.58259574e-02 -1.08397293e+00
-1.46844125e+00 4.69458163e-01 -3.19794774e-01 -2.40425691e-01
8.37072968e-01 3.89293581e-01 6.61476672e-01 5.10067225e-01
7.34634757e-01 -1.02940822e+00 -3.39807481e-01 5.13772130e-01
8.04135799e-01 -1.26818490e+00 1.14001311e-01 -1.00737500e+00
-6.40705585e-01 6.66456819e-01 6.58944845e-01 4.89654064e-01
8.03253591e-01 -1.31272495e-01 -3.57775182e-01 -5.97376823e-01
-8.71382117e-01 -4.75648195e-01 2.97669470e-01 8.01958978e-01
9.34177935e-02 4.81607199e-01 -2.02249929e-01 7.84360766e-01
-8.50437656e-02 5.38363121e-02 2.83924937e-01 8.27614784e-01
-3.04257171e-03 -1.23394287e+00 1.48880944e-01 4.20022637e-01
-3.36875826e-01 -5.49012721e-01 -5.66364884e-01 9.73927259e-01
2.97160298e-01 9.51405466e-01 -5.18238842e-01 -5.39418936e-01
7.35549256e-02 1.01388171e-01 5.43543160e-01 -5.60062289e-01
-3.03903133e-01 -2.49922946e-01 5.90775013e-01 -2.35512018e-01
-6.26753986e-01 -2.60199070e-01 -1.13997984e+00 -4.59845185e-01
-5.07172942e-01 3.20991069e-01 4.08646315e-01 9.25051153e-01
5.89397371e-01 5.27657092e-01 1.11774862e-01 5.77879548e-02
-4.85501625e-02 -1.06869388e+00 -1.12550581e+00 3.72037143e-01
-1.59682736e-01 -7.11530745e-01 9.15664136e-02 1.83540076e-01]
|
[8.972822189331055, 8.060175895690918]
|
7a411484-4e99-4317-9aad-08716fa475c3
|
on-the-benefit-of-syntactic-supervision-for
| null | null |
https://aclanthology.org/2021.emnlp-main.503
|
https://aclanthology.org/2021.emnlp-main.503.pdf
|
On the Benefit of Syntactic Supervision for Cross-lingual Transfer in Semantic Role Labeling
|
Although recent developments in neural architectures and pre-trained representations have greatly increased state-of-the-art model performance on fully-supervised semantic role labeling (SRL), the task remains challenging for languages where supervised SRL training data are not abundant. Cross-lingual learning can improve performance in this setting by transferring knowledge from high-resource languages to low-resource ones. Moreover, we hypothesize that annotations of syntactic dependencies can be leveraged to further facilitate cross-lingual transfer. In this work, we perform an empirical exploration of the helpfulness of syntactic supervision for crosslingual SRL within a simple multitask learning scheme. With comprehensive evaluations across ten languages (in addition to English) and three SRL benchmark datasets, including both dependency- and span-based SRL, we show the effectiveness of syntactic supervision in low-resource scenarios.
|
['Eduard Hovy', 'Emma Strubell', 'Zhisong Zhang']
| null | null | null | null |
emnlp-2021-11
|
['semantic-role-labeling']
|
['natural-language-processing']
|
[ 4.33784366e-01 2.37848386e-01 -9.21636343e-01 -7.00077951e-01
-1.15202677e+00 -8.15122724e-01 7.66256332e-01 7.02925622e-02
-8.46620619e-01 9.01665032e-01 7.95392931e-01 -1.64823323e-01
1.34841248e-01 -2.45692343e-01 -7.72242725e-01 -1.57542676e-01
1.50520146e-01 5.69700122e-01 8.10067430e-02 -5.26013434e-01
-3.47448736e-01 1.15588224e-02 -1.19340515e+00 6.87241197e-01
7.75831997e-01 6.80980563e-01 3.24983090e-01 -6.53543621e-02
-2.60842413e-01 1.13027132e+00 -3.76248837e-01 -3.49956155e-01
1.00141853e-01 -7.25877881e-02 -1.12333834e+00 -2.90268570e-01
4.47282821e-01 1.99754521e-01 -6.30560741e-02 6.65342510e-01
4.60725963e-01 1.47404552e-01 3.97608846e-01 -9.48886395e-01
-9.12163794e-01 1.19050395e+00 -4.13260043e-01 2.59945124e-01
3.04827392e-01 -7.61966705e-02 1.75456285e+00 -7.36399770e-01
9.71129775e-01 1.62942493e+00 6.60840511e-01 7.89664447e-01
-1.35132062e+00 -8.03614974e-01 3.11193287e-01 3.99218239e-02
-1.05561030e+00 -6.82784975e-01 9.45565462e-01 -9.27360654e-02
1.29135454e+00 -3.38262051e-01 -2.62890421e-02 1.34036589e+00
-5.27637541e-01 1.15802813e+00 1.42594802e+00 -6.85458839e-01
-3.91664088e-01 1.34115815e-01 2.05191687e-01 5.24359584e-01
1.80942640e-01 -1.70764849e-01 -9.76923168e-01 1.65233016e-01
4.38085467e-01 -4.93453085e-01 -1.68676719e-01 -2.51928777e-01
-1.13290083e+00 9.47543979e-01 4.61282730e-01 6.82192564e-01
1.47270396e-01 3.19816053e-01 9.43735838e-01 4.61975127e-01
9.58597541e-01 8.19904685e-01 -1.13787436e+00 -1.30727841e-02
-4.83514816e-01 -1.37543648e-01 6.11160696e-01 7.93733358e-01
7.07286239e-01 8.58490393e-02 1.03347590e-02 1.47474825e+00
9.65595543e-02 2.66474187e-01 6.25529230e-01 -8.79843712e-01
9.03629005e-01 6.30871117e-01 -1.21799245e-01 -2.86134958e-01
-2.97736228e-01 -3.11405301e-01 -4.27179664e-01 -4.09041464e-01
4.48093534e-01 -1.30487951e-02 -4.77321327e-01 2.10428333e+00
4.46678661e-02 3.19672823e-02 4.81960416e-01 5.36449015e-01
7.62206078e-01 4.31820512e-01 6.01088762e-01 1.51224062e-01
1.43276680e+00 -1.09899306e+00 -5.98871231e-01 -6.84304714e-01
1.29018140e+00 -3.12082291e-01 1.69470024e+00 -6.59052432e-02
-7.82481790e-01 -5.70966661e-01 -8.95489097e-01 -5.80232680e-01
-5.25121748e-01 4.36230868e-01 7.88169205e-01 2.28628546e-01
-1.01396048e+00 3.43648225e-01 -5.76609910e-01 -3.95173192e-01
5.13754070e-01 4.43905368e-02 -5.71828127e-01 -6.15250528e-01
-1.79173255e+00 1.23575974e+00 7.59712875e-01 -3.23975295e-01
-9.86848891e-01 -1.14917147e+00 -1.26644158e+00 -1.21442236e-01
6.04659855e-01 -2.13383272e-01 1.17308033e+00 -1.16949141e+00
-1.11407387e+00 1.64609933e+00 -7.91611969e-02 -6.34434283e-01
2.18327910e-01 -4.59468424e-01 -2.66250432e-01 6.00081868e-02
5.77328444e-01 8.41219008e-01 4.67406392e-01 -1.24510276e+00
-5.96471012e-01 -3.95730615e-01 4.91233170e-01 6.47826552e-01
-3.80715072e-01 5.07424533e-01 -6.24230243e-02 -6.10825300e-01
-5.56537986e-01 -7.70317912e-01 -3.47017236e-02 -4.15995240e-01
1.60004392e-01 -9.82073307e-01 7.11424351e-01 -7.89821565e-01
7.00138688e-01 -2.21258521e+00 2.01526031e-01 -5.43828011e-01
-4.35703158e-01 2.17113301e-01 -2.49777928e-01 2.40155175e-01
-1.70232981e-01 3.33655953e-01 -2.83611178e-01 -7.77552724e-01
-1.91235825e-01 5.82109094e-01 -3.06541264e-01 2.94051558e-01
5.53059399e-01 9.71050799e-01 -1.05338025e+00 -3.72046411e-01
-1.04340881e-01 3.93390715e-01 -3.91785294e-01 1.22400314e-01
-4.25278902e-01 7.59501874e-01 -1.86918020e-01 5.63979566e-01
-7.21776634e-02 -3.22367907e-01 5.45229197e-01 -8.83416831e-02
2.49093529e-02 1.15178609e+00 -5.30270517e-01 2.31684160e+00
-1.30447745e+00 4.14016992e-01 6.93734437e-02 -1.20231307e+00
7.73451984e-01 4.05514151e-01 3.45079780e-01 -8.41820419e-01
-1.95324168e-01 4.09668326e-01 1.08414657e-01 -6.64500520e-02
3.23708206e-01 -4.11517292e-01 -4.19669539e-01 6.49906635e-01
4.77209330e-01 -8.43664929e-02 2.60658354e-01 8.26021805e-02
8.02922964e-01 5.14502347e-01 5.59456289e-01 -5.52449524e-01
6.94995940e-01 1.03400156e-01 4.73895043e-01 3.75386149e-01
-1.23537607e-01 1.17190875e-01 3.26570600e-01 -9.81251746e-02
-8.15158010e-01 -6.86486483e-01 -4.56289530e-01 1.89496589e+00
-1.10101245e-01 -3.74624133e-01 -3.11502546e-01 -1.17551303e+00
1.60185665e-01 9.03203607e-01 -5.44067025e-01 -1.72144212e-02
-9.51342285e-01 -7.50372589e-01 1.03142226e+00 7.16884255e-01
5.27932465e-01 -1.36431158e+00 -3.15706074e-01 2.46085897e-01
-1.26674011e-01 -1.78108633e+00 -1.59779474e-01 3.76737833e-01
-6.08772933e-01 -1.00497782e+00 -3.11470777e-01 -8.43595266e-01
3.23747486e-01 2.99953699e-01 1.69257462e+00 3.79783846e-02
-3.78415063e-02 3.85052800e-01 -4.83676583e-01 -2.32843503e-01
-4.80901390e-01 4.84316260e-01 6.49667531e-02 -2.42402628e-01
3.77866536e-01 -2.28138342e-01 -3.04576297e-05 1.92731202e-01
-5.67771256e-01 -1.68102998e-02 2.28279904e-01 8.16623211e-01
5.47922552e-01 -1.61114052e-01 8.95721436e-01 -1.40921879e+00
3.41163844e-01 -6.95035934e-01 -3.63021135e-01 2.52033323e-01
-4.01080281e-01 3.30419540e-01 7.40482390e-01 -1.73620924e-01
-1.63289607e+00 -2.17465267e-01 2.24580653e-02 -1.24638133e-01
-2.31179029e-01 7.31932580e-01 -3.40483487e-01 2.38298565e-01
6.58470035e-01 4.44561392e-02 -2.94059455e-01 -8.52329135e-01
7.87902474e-01 4.74843591e-01 3.32524240e-01 -1.17681074e+00
5.65318942e-01 4.74693954e-01 -2.81419665e-01 -7.19349325e-01
-1.88608003e+00 -5.73612750e-01 -9.72099185e-01 3.89284700e-01
9.81383383e-01 -1.60351610e+00 -3.09823211e-02 1.82168260e-01
-1.10368514e+00 -8.29769790e-01 -3.21140230e-01 9.57740396e-02
-4.18239534e-01 3.53785604e-03 -7.69802570e-01 -2.54276395e-01
-1.78195447e-01 -8.64598155e-01 1.31648099e+00 -2.47831583e-01
-1.97038457e-01 -1.66122389e+00 4.36934419e-02 8.20456982e-01
2.21473470e-01 -1.51368618e-01 1.15983522e+00 -1.16256797e+00
-1.95173502e-01 1.87616423e-01 -3.90052497e-01 5.80162168e-01
2.90521294e-01 -7.78020442e-01 -1.14555812e+00 -4.58613932e-01
-3.26849163e-01 -1.13238084e+00 9.92865741e-01 1.30329356e-01
1.05642152e+00 2.59568095e-02 -2.44750008e-01 5.00424981e-01
1.32300973e+00 -3.46587241e-01 7.53886402e-02 4.93128955e-01
1.11298656e+00 1.02522397e+00 9.79275227e-01 -1.60821378e-01
7.57132471e-01 6.80392206e-01 1.93069167e-02 -1.27208725e-01
-5.98816872e-01 -3.96235019e-01 6.37887657e-01 7.40413010e-01
7.44418753e-03 5.77715598e-02 -1.03901911e+00 7.51294255e-01
-1.66866708e+00 -5.53618312e-01 5.46006113e-02 1.83546793e+00
1.52028275e+00 1.04839854e-01 -3.36577594e-02 -1.88055441e-01
4.56026286e-01 6.14697516e-01 -5.47460020e-01 -1.02790609e-01
-3.83805543e-01 3.66566509e-01 5.09505987e-01 4.97583121e-01
-1.34207332e+00 1.77141845e+00 6.03530312e+00 8.43001366e-01
-9.95218813e-01 8.39387953e-01 6.02614045e-01 1.41384393e-01
-3.41267109e-01 -5.93919642e-02 -1.40373206e+00 2.20770687e-01
1.09071600e+00 -4.24296968e-02 1.92083612e-01 1.08667815e+00
-1.04865432e-01 9.70319659e-02 -1.28284979e+00 6.69918239e-01
3.20415169e-01 -1.21454930e+00 -8.44811201e-02 -2.51769096e-01
8.75650406e-01 6.76704049e-01 -1.84339136e-01 7.04998434e-01
8.08866024e-01 -1.17969072e+00 6.13784611e-01 -5.21571875e-01
1.32652307e+00 -5.06011069e-01 7.01228678e-01 2.36174256e-01
-1.31291354e+00 9.26460549e-02 -2.78831124e-01 -1.12406597e-01
1.68518722e-01 1.86825916e-01 -8.99933577e-01 5.29620290e-01
7.12377131e-01 1.43588591e+00 -7.33074069e-01 2.40282924e-03
-1.03739262e+00 8.08570981e-01 -5.07107824e-02 4.94813114e-01
3.79103333e-01 1.25408575e-01 1.81299657e-01 1.49744904e+00
-2.72582412e-01 -2.86023140e-01 4.68762398e-01 6.07601702e-01
-5.55390537e-01 2.97030926e-01 -1.03206491e+00 -1.22206613e-01
4.69934464e-01 1.09124053e+00 -3.17652494e-01 -3.16086411e-01
-7.56659389e-01 8.04487526e-01 9.67278600e-01 7.90132433e-02
-5.43253064e-01 3.18528861e-01 7.58010089e-01 1.73854113e-01
-7.49408826e-02 -4.14148062e-01 -3.53525996e-01 -1.33077586e+00
-2.38669232e-01 -9.03335512e-01 8.46297085e-01 -5.89489996e-01
-1.54814935e+00 3.33464950e-01 3.38965535e-01 -6.91211402e-01
-3.50675285e-01 -8.93862963e-01 -7.12866932e-02 7.73267090e-01
-2.28344226e+00 -1.89776659e+00 2.86906153e-01 6.92681909e-01
1.07527637e+00 -4.45317119e-01 8.10652733e-01 3.08021396e-01
-3.50619048e-01 6.04838252e-01 -2.49589860e-01 3.21191370e-01
1.01400304e+00 -1.30391252e+00 3.40209723e-01 8.29905450e-01
3.87966007e-01 5.44037938e-01 2.60188341e-01 -6.02354467e-01
-1.03772950e+00 -1.21084881e+00 9.53944206e-01 -7.29422092e-01
1.11261916e+00 -6.30299389e-01 -1.14737082e+00 1.12143743e+00
2.06613973e-01 2.91665614e-01 7.81503499e-01 7.70869672e-01
-8.66046727e-01 1.52500615e-01 -8.68556142e-01 3.30068499e-01
1.54786396e+00 -1.14796507e+00 -9.54699039e-01 4.04369831e-01
1.18077242e+00 -1.46925777e-01 -9.82472062e-01 4.04137790e-01
1.52513266e-01 -3.66546988e-01 1.01037991e+00 -9.32951987e-01
5.88371933e-01 -2.50807628e-02 -3.86142969e-01 -1.56422102e+00
2.27005593e-02 -1.35933280e-01 3.72542977e-01 1.47845292e+00
6.83945537e-01 -6.32583380e-01 3.55883121e-01 1.79141492e-01
-4.49180126e-01 -4.44339126e-01 -9.61737335e-01 -9.72018540e-01
3.95211935e-01 -4.23096776e-01 1.34512976e-01 1.53963494e+00
-2.12770149e-01 9.70625401e-01 -3.17742884e-01 -1.24124167e-02
5.55926681e-01 -5.05426787e-02 4.28976506e-01 -1.35326755e+00
-4.55715567e-01 -1.01942584e-01 1.67947412e-01 -6.74189031e-01
1.44037378e+00 -1.62433898e+00 2.74662059e-02 -1.41516674e+00
2.18555272e-01 -9.24689651e-01 -4.85404760e-01 1.13667202e+00
-2.67003834e-01 1.91137850e-01 1.03198066e-01 2.69524515e-01
-7.80743003e-01 5.73014081e-01 9.60266829e-01 2.60176584e-02
1.35663226e-01 -5.48890948e-01 -7.95591295e-01 8.59404743e-01
7.07376301e-01 -6.65664434e-01 -5.11307776e-01 -8.90298843e-01
2.99965560e-01 -2.39470661e-01 1.12819392e-02 -4.74829972e-01
-4.63154703e-01 -2.94171143e-02 -1.88763037e-01 4.55708466e-02
2.95052379e-01 -5.45128465e-01 -6.06984317e-01 1.53214365e-01
-8.21845055e-01 -2.70513713e-01 2.54839212e-01 4.29413974e-01
-4.76537287e-01 -2.86804080e-01 8.33436906e-01 -4.70459431e-01
-1.12358248e+00 1.11543633e-01 7.30474815e-02 8.42658579e-01
5.85711539e-01 3.54161024e-01 -5.16806602e-01 -8.44912231e-02
-4.74683434e-01 3.40699971e-01 5.42087972e-01 8.41724992e-01
1.77676287e-02 -1.28730118e+00 -8.28693569e-01 -3.70048955e-02
5.75551808e-01 1.26843408e-01 -9.56309438e-02 5.33675909e-01
3.47317755e-02 6.74325407e-01 -1.45806462e-01 -4.24445570e-01
-1.10062444e+00 2.59337187e-01 2.25999266e-01 -9.04576182e-01
-3.20762932e-01 1.00539911e+00 3.26445699e-01 -1.01925635e+00
-2.90327128e-02 1.19266631e-02 -4.82610375e-01 2.72576481e-01
1.59184456e-01 -1.55951664e-01 -1.35697825e-02 -8.89368653e-01
-4.82797444e-01 3.87106180e-01 -1.51393995e-01 -1.51581168e-01
1.46963954e+00 -5.02045676e-02 -1.71232536e-01 7.86016583e-01
1.22612500e+00 6.71190694e-02 -1.15116453e+00 -7.18125701e-01
6.56650782e-01 -1.68780193e-01 -7.20713586e-02 -9.84511077e-01
-7.30600119e-01 8.99004281e-01 -9.01000053e-02 -3.61351430e-01
6.89139545e-01 5.84920704e-01 5.57399273e-01 3.53604257e-01
7.03633308e-01 -1.18041253e+00 3.07681382e-01 9.84894574e-01
9.74881411e-01 -1.63928962e+00 -1.20943420e-01 -4.76246119e-01
-1.02006483e+00 5.03278255e-01 8.23836684e-01 -1.50791764e-01
4.95424300e-01 9.50644091e-02 2.02406826e-03 -1.40652761e-01
-9.16188657e-01 -4.52184647e-01 2.37210229e-01 5.29117942e-01
1.10218120e+00 1.50806561e-01 -1.85309723e-01 6.11954153e-01
-1.71081990e-01 -2.56936044e-01 1.87249288e-01 6.68902099e-01
-7.46166706e-02 -1.37496960e+00 1.97651401e-01 1.53311882e-02
-8.36271226e-01 -5.71905613e-01 -2.45341986e-01 9.55590546e-01
4.52541225e-02 6.79044664e-01 3.23089212e-02 4.11352247e-01
1.87897697e-01 4.82354581e-01 5.23434699e-01 -1.37880099e+00
-5.81604242e-01 -2.66597688e-01 9.63690281e-01 -6.61245644e-01
-9.22719598e-01 -6.99911237e-01 -1.38694811e+00 5.40320277e-01
1.97787419e-01 -1.39310569e-01 6.04629934e-01 1.26308966e+00
2.85648942e-01 4.19937402e-01 1.31936729e-01 -6.20321453e-01
-4.83810961e-01 -1.12013185e+00 -3.58800858e-01 7.32128799e-01
9.86867025e-03 -8.56601954e-01 -3.20935428e-01 1.24998149e-02]
|
[10.456666946411133, 9.5051908493042]
|
4a2cd9b5-dd20-4319-8cd5-b1c8dfdc3439
|
the-best-of-both-worlds-accurate-global-and
|
2301.08968
| null |
https://arxiv.org/abs/2301.08968v2
|
https://arxiv.org/pdf/2301.08968v2.pdf
|
The Best of Both Worlds: Accurate Global and Personalized Models through Federated Learning with Data-Free Hyper-Knowledge Distillation
|
Heterogeneity of data distributed across clients limits the performance of global models trained through federated learning, especially in the settings with highly imbalanced class distributions of local datasets. In recent years, personalized federated learning (pFL) has emerged as a potential solution to the challenges presented by heterogeneous data. However, existing pFL methods typically enhance performance of local models at the expense of the global model's accuracy. We propose FedHKD (Federated Hyper-Knowledge Distillation), a novel FL algorithm in which clients rely on knowledge distillation (KD) to train local models. In particular, each client extracts and sends to the server the means of local data representations and the corresponding soft predictions -- information that we refer to as ``hyper-knowledge". The server aggregates this information and broadcasts it to the clients in support of local training. Notably, unlike other KD-based pFL methods, FedHKD does not rely on a public dataset nor it deploys a generative model at the server. We analyze convergence of FedHKD and conduct extensive experiments on visual datasets in a variety of scenarios, demonstrating that FedHKD provides significant improvement in both personalized as well as global model performance compared to state-of-the-art FL methods designed for heterogeneous data settings.
|
['Haris Vikalo', 'Wang', 'Johnny', 'Huancheng Chen']
|
2023-01-21
| null | null | null | null |
['personalized-federated-learning']
|
['methodology']
|
[-5.02177298e-01 3.16663124e-02 -4.91756499e-01 -4.81120676e-01
-1.24670553e+00 -5.73672891e-01 3.69272798e-01 -1.30925432e-01
8.10683370e-02 8.61483216e-01 3.43179613e-01 -8.30763131e-02
-1.60853595e-01 -8.53670359e-01 -8.58081341e-01 -1.02818656e+00
1.55046538e-01 9.80840027e-01 7.09702000e-02 1.75547078e-01
-3.16148400e-01 4.28888708e-01 -1.51782453e+00 7.69192815e-01
5.87994933e-01 1.24727750e+00 -3.04174647e-02 4.62340713e-01
-3.85941923e-01 1.26316667e+00 -4.72537965e-01 -6.98645532e-01
3.20079476e-01 -7.53485337e-02 -9.29399133e-01 -2.42641479e-01
3.92173409e-01 -6.27783716e-01 -4.51787025e-01 5.51194191e-01
7.77777135e-01 -2.09252778e-02 2.41745532e-01 -1.50971115e+00
-7.08912909e-01 9.59203005e-01 -1.81335047e-01 1.61444321e-01
-6.54415861e-02 4.27641004e-01 9.45790350e-01 -9.10276830e-01
7.52908826e-01 1.05783629e+00 8.25328887e-01 4.48108107e-01
-1.36633778e+00 -5.94178081e-01 3.44501823e-01 3.85511607e-01
-1.28238583e+00 -5.10053635e-01 6.07581198e-01 -2.90216088e-01
8.99095356e-01 2.55014420e-01 1.94809586e-01 1.24363053e+00
-2.00755581e-01 1.15624535e+00 9.47401702e-01 -1.71290934e-01
5.89370966e-01 4.22663897e-01 1.63236380e-01 4.04768586e-01
4.57446575e-02 -2.05338851e-01 -1.08976316e+00 -7.77279854e-01
2.15968639e-01 3.60423446e-01 -2.49726519e-01 -5.95095396e-01
-7.45280027e-01 8.14757109e-01 3.55854869e-01 8.11870582e-03
-7.75550663e-01 5.79517186e-02 4.28299606e-01 2.93957353e-01
7.96266735e-01 -2.08891258e-01 -8.22305918e-01 -1.10898368e-01
-8.02472353e-01 3.49516541e-01 1.10944510e+00 8.10108304e-01
1.20150352e+00 -1.00407653e-01 -4.68062431e-01 8.33217561e-01
5.97721115e-02 4.41547126e-01 6.39240444e-01 -9.84554291e-01
6.20317936e-01 8.96808565e-01 -6.29241467e-02 -5.89842021e-01
-7.11365119e-02 -7.30113208e-01 -6.95087671e-01 -1.74026325e-01
-9.17651504e-03 -2.28572890e-01 -4.99747634e-01 1.73516679e+00
8.48004997e-01 3.45564365e-01 2.39958510e-01 7.05010593e-01
6.46827519e-01 4.98032331e-01 4.75859754e-02 1.88802381e-03
6.03361964e-01 -1.24055576e+00 -3.95716697e-01 1.36841998e-01
7.24773347e-01 -2.79563487e-01 7.68530548e-01 2.71668255e-01
-9.59695339e-01 -2.15485618e-02 -4.22347993e-01 2.78060436e-01
-3.37535441e-01 -3.45589757e-01 5.22225559e-01 4.55301285e-01
-1.15193641e+00 5.02220452e-01 -7.25812495e-01 -2.78593063e-01
8.24180841e-01 2.85614491e-01 -3.47185820e-01 -5.58994651e-01
-7.59952307e-01 5.58885157e-01 9.23984572e-02 -2.85262406e-01
-1.25110197e+00 -1.23915017e+00 -4.58218575e-01 2.79205650e-01
2.80487657e-01 -9.65250075e-01 1.63533998e+00 -8.96892130e-01
-1.30118930e+00 5.01675963e-01 -5.01770303e-02 -4.49526906e-01
7.37934113e-01 -1.49294108e-01 -2.98106104e-01 9.44625866e-03
-1.75061420e-01 4.26504668e-03 5.72437227e-01 -1.45489740e+00
-7.45794952e-01 -5.86716950e-01 -1.19764358e-01 -8.87194835e-03
-4.98491496e-01 -1.35311991e-01 -6.00799501e-01 -2.90648043e-01
-2.29221627e-01 -6.09072328e-01 -6.19447753e-02 -1.08589055e-02
-3.22073609e-01 -4.31219071e-01 1.27822948e+00 -5.43148458e-01
1.06160855e+00 -2.16246462e+00 -1.98092446e-01 3.59046042e-01
4.40131932e-01 3.14091295e-01 -1.93095565e-01 6.95354640e-01
4.40781057e-01 -1.24769554e-01 1.32228240e-01 -7.68625557e-01
4.08511251e-01 5.43691218e-01 -4.01754707e-01 3.97684306e-01
-1.25992835e-01 9.35031950e-01 -7.03884304e-01 -6.02095425e-01
-1.81619048e-01 6.90330386e-01 -7.45203555e-01 5.07791877e-01
-6.89122319e-01 2.77956754e-01 -5.53105831e-01 6.77471161e-01
8.71881545e-01 -7.74897218e-01 6.19580209e-01 -6.76266104e-02
3.15176785e-01 1.89604759e-01 -1.13568878e+00 1.52128303e+00
-5.36398470e-01 7.96435177e-02 2.01827794e-01 -8.37439597e-01
7.34058797e-01 4.36446130e-01 6.63288116e-01 -7.21938312e-01
-9.35692489e-02 2.07849279e-01 -8.05182159e-01 -3.46467584e-01
1.20153144e-01 7.75661543e-02 2.06144646e-01 9.39792097e-01
3.54738504e-01 7.63912082e-01 -3.17305863e-01 4.95633930e-01
1.43431199e+00 1.22967243e-01 -2.33417556e-01 1.05849274e-01
7.19862282e-02 -5.45126535e-02 7.47720301e-01 8.78784716e-01
-4.71506268e-02 3.62479687e-01 1.99446648e-01 -7.16217875e-01
-8.36179376e-01 -1.08115745e+00 2.19823614e-01 1.71672046e+00
-4.26324904e-02 -5.42753637e-01 -6.50879741e-01 -1.00268006e+00
6.44816101e-01 5.74631572e-01 -5.86632431e-01 -1.73527122e-01
-2.73380309e-01 -6.45022094e-01 4.36691821e-01 5.78244984e-01
5.04538596e-01 -8.06666732e-01 -3.08173537e-01 4.09245223e-01
-3.31560344e-01 -9.99294221e-01 -2.48578265e-01 2.37021998e-01
-6.55100107e-01 -1.07696509e+00 -4.53001022e-01 -3.34038794e-01
3.38973463e-01 3.66768897e-01 1.21494973e+00 -1.41599640e-01
-2.40056574e-01 6.69625461e-01 -4.26300764e-01 -2.03375012e-01
-3.04482818e-01 2.50684202e-01 -1.29508123e-01 5.33994019e-01
3.49303424e-01 -6.58524871e-01 -8.49789500e-01 3.52946252e-01
-8.79085779e-01 -3.68950255e-02 4.01131302e-01 7.18778133e-01
7.17889369e-01 -2.33656123e-01 8.01293135e-01 -1.14915240e+00
3.28851163e-01 -1.13387811e+00 -2.61261076e-01 6.05728805e-01
-8.61529827e-01 9.61549282e-02 8.49039793e-01 -3.39782834e-01
-1.36420643e+00 -2.33864039e-01 3.19220990e-01 -9.02558446e-01
4.29552644e-02 4.18994397e-01 -2.55083889e-01 -3.23332995e-02
7.10708916e-01 2.38346711e-01 -4.07092646e-02 -1.07237422e+00
5.45755148e-01 9.46908236e-01 5.49807549e-01 -8.56692731e-01
4.50936675e-01 4.29866046e-01 -3.81966412e-01 -4.77832295e-02
-8.74461234e-01 -3.97345722e-01 -1.44026801e-01 -1.08913876e-01
1.67066440e-01 -1.16580272e+00 -8.89183223e-01 5.46659291e-01
-7.76131868e-01 -6.77416623e-01 -5.47559798e-01 1.55822247e-01
-6.80709124e-01 -1.61019325e-01 -6.49097204e-01 -7.60251284e-01
-7.24723816e-01 -7.83185780e-01 8.43081594e-01 1.43982843e-01
2.08776757e-01 -9.63539422e-01 4.33151215e-01 6.81267440e-01
9.75615621e-01 1.31544709e-01 9.34616685e-01 -1.03793156e+00
-7.27764130e-01 -3.00956905e-01 -1.00642093e-01 3.19303304e-01
-7.87278563e-02 -2.02710956e-01 -1.43870914e+00 -3.77256304e-01
-4.53166306e-01 -7.52036750e-01 5.56303501e-01 -1.52609572e-01
1.42018533e+00 -8.60766947e-01 -4.28933948e-01 7.87527502e-01
1.67114854e+00 -4.21249539e-01 8.75262842e-02 3.25884253e-01
5.51497817e-01 2.78910726e-01 1.03270493e-01 1.11099851e+00
8.18936706e-01 7.06448495e-01 6.13168061e-01 4.88478616e-02
-3.24411094e-01 -4.51356888e-01 3.40847850e-01 5.20222783e-01
3.30099076e-01 -2.45598793e-01 -9.16050255e-01 7.01460719e-01
-2.38752413e+00 -9.06990826e-01 2.93600738e-01 2.14366198e+00
1.09047925e+00 -6.95865512e-01 2.61548996e-01 -2.67451197e-01
7.01714635e-01 -1.93700083e-02 -1.03440928e+00 -1.64616019e-01
-4.47106391e-01 1.09419622e-01 3.78083289e-01 6.17307536e-02
-7.22644866e-01 7.67326891e-01 5.79226828e+00 8.01774085e-01
-1.14042437e+00 6.33840919e-01 7.36780822e-01 -5.58027804e-01
-5.27952433e-01 -1.70372397e-01 -9.51316297e-01 6.37332857e-01
1.26704454e+00 -3.66431922e-01 6.07743979e-01 1.35721087e+00
-2.17000902e-01 1.93743706e-01 -1.11537230e+00 9.12194073e-01
9.65754408e-03 -1.68563986e+00 1.57156393e-01 8.85253251e-02
1.08427715e+00 7.26274431e-01 5.22686802e-02 3.70042503e-01
1.14492023e+00 -6.91635847e-01 5.93498111e-01 7.91128159e-01
5.42224646e-01 -8.20205510e-01 5.92835426e-01 5.90383410e-01
-8.27132165e-01 -4.60225165e-01 -2.77000636e-01 4.81753439e-01
-1.09735750e-01 8.07338297e-01 -1.11615396e+00 7.90637553e-01
1.06297648e+00 3.44096273e-01 -4.28282470e-01 1.06390750e+00
3.94050479e-01 9.80268717e-01 -5.54015279e-01 5.11907935e-01
9.15213209e-03 2.76804417e-01 -6.10254705e-02 1.05080986e+00
1.65395349e-01 -1.55426458e-01 2.57053167e-01 7.06063449e-01
-4.44169044e-01 2.58437634e-01 -4.10395712e-01 2.45779768e-01
7.20031857e-01 1.14804971e+00 1.99346468e-01 -5.40942907e-01
-3.75372589e-01 7.39586174e-01 8.44210982e-01 5.00229359e-01
-5.81902802e-01 1.81772634e-01 9.44764972e-01 3.52392048e-02
5.36405742e-01 3.24457735e-01 8.27608928e-02 -1.18817091e+00
9.94307250e-02 -9.87209558e-01 9.81613219e-01 -6.53596401e-01
-1.87870610e+00 6.11940503e-01 -3.41516256e-01 -7.43583739e-01
-4.51049507e-01 -2.24014162e-03 -4.51848000e-01 8.07497263e-01
-1.79180217e+00 -1.60386908e+00 -5.80692410e-01 1.27965772e+00
1.26896709e-01 -1.46232009e-01 1.00235474e+00 3.00924242e-01
-7.83713639e-01 9.81610537e-01 8.06149721e-01 -1.15833744e-01
8.76323938e-01 -8.51674855e-01 -2.25160760e-03 3.05045307e-01
-6.40386641e-02 2.28402779e-01 3.49892557e-01 -5.19593358e-01
-1.74045742e+00 -1.61160207e+00 7.08291292e-01 -3.37036908e-01
4.40378278e-01 -2.85248756e-01 -9.68109190e-01 1.00230527e+00
2.28737351e-02 7.63291597e-01 1.10996127e+00 5.31744994e-02
-8.11909258e-01 -6.80779934e-01 -1.47747695e+00 5.87667637e-02
9.29604948e-01 -5.21050870e-01 1.28541797e-01 6.49814367e-01
8.65187585e-01 -3.86432968e-02 -1.09412777e+00 2.46104777e-01
4.42524284e-01 -1.16382003e+00 6.41467154e-01 -1.09438717e+00
-5.74856289e-02 1.14779167e-01 -6.59106731e-01 -1.20874572e+00
-4.70421851e-01 -7.04576135e-01 -7.04366446e-01 1.60014904e+00
2.92248577e-01 -9.71907198e-01 9.81707335e-01 9.36807215e-01
8.97953138e-02 -8.17143261e-01 -9.13411021e-01 -7.61711061e-01
-6.92541152e-02 -3.92585278e-01 1.32934117e+00 8.99712801e-01
-9.62371081e-02 -3.41901898e-01 -1.98488459e-01 2.29862660e-01
8.90426040e-01 4.61665362e-01 1.15892804e+00 -1.09150779e+00
-5.48357368e-01 -1.48615073e-02 -1.52476773e-01 -5.88253021e-01
4.00414050e-01 -1.09818077e+00 -4.12223518e-01 -1.44118726e+00
5.27006388e-01 -8.66477013e-01 -6.00535393e-01 1.16465628e+00
-1.49225059e-03 1.00208014e-01 3.33150476e-01 6.25933886e-01
-1.10335720e+00 7.19798028e-01 6.47645056e-01 -5.09947538e-02
-4.35512289e-02 6.61092326e-02 -7.83719540e-01 1.38332754e-01
6.33991659e-01 -4.40838158e-01 -3.42675447e-01 -5.53490162e-01
1.02899805e-01 -4.37386222e-02 5.14312148e-01 -7.26332545e-01
4.79925513e-01 -3.10339272e-01 3.22197050e-01 -3.45633417e-01
1.82232782e-01 -8.53368521e-01 4.99639571e-01 7.97641184e-03
-2.49789417e-01 -1.32371530e-01 -6.92543909e-02 6.14833057e-01
-9.62013006e-02 4.55861598e-01 6.46619320e-01 -7.95320943e-02
-4.51216221e-01 5.91979623e-01 1.29374474e-01 8.46939161e-02
1.14946187e+00 2.21855000e-01 -9.63861525e-01 -4.54757094e-01
-6.19905591e-01 3.59360039e-01 5.48867881e-01 1.28044248e-01
2.74601012e-01 -1.30251861e+00 -6.86618745e-01 3.06064188e-01
2.30216369e-01 1.99651077e-01 7.06789196e-01 9.45985079e-01
-6.69480115e-02 2.70073920e-01 2.28360146e-01 -4.78932410e-01
-8.85249138e-01 7.12532341e-01 4.70505774e-01 -3.05974960e-01
-7.63161600e-01 6.40030205e-01 1.55962572e-01 -6.98860884e-01
5.86437821e-01 3.21871251e-01 4.42090750e-01 -1.02572180e-01
6.88360631e-01 6.38054013e-01 5.75192213e-01 -3.74383956e-01
-4.31644171e-01 -7.67310262e-02 -1.91395730e-01 2.94892490e-01
1.74911284e+00 -7.50797912e-02 -2.99133863e-02 2.09830672e-01
1.29882288e+00 -2.71430552e-01 -1.56909323e+00 -8.54204237e-01
-4.21174407e-01 -5.93364060e-01 3.89270931e-02 -1.32485008e+00
-1.55932164e+00 4.44886088e-01 4.66153681e-01 -2.10521594e-01
1.07886410e+00 3.15682709e-01 8.60625625e-01 2.43326262e-01
7.31687665e-01 -8.79541337e-01 -3.03382277e-02 1.57815859e-01
5.62179923e-01 -9.40365553e-01 -4.08102751e-01 1.44216552e-01
-7.76121497e-01 6.91799760e-01 6.86063170e-01 2.03987837e-01
7.07712829e-01 4.32037383e-01 2.56543040e-01 4.64996975e-03
-1.59235632e+00 1.39908820e-01 -2.42965296e-01 5.63076973e-01
-3.47701252e-01 7.77694676e-03 5.74065745e-01 9.87349212e-01
6.43996894e-02 3.69608253e-01 -4.31812182e-02 1.16901314e+00
-2.01241776e-01 -1.05123532e+00 -3.88223231e-01 4.68573272e-01
-3.68758619e-01 1.99391380e-01 -2.40641832e-01 1.70128167e-01
1.29572660e-01 9.17236686e-01 3.90600227e-02 -4.77867931e-01
1.43041417e-01 5.10784686e-01 1.18458800e-01 -3.09382737e-01
-9.58395302e-01 -1.63219199e-01 -1.65275246e-01 -8.99464667e-01
5.05692267e-04 -3.91041785e-01 -1.13722908e+00 -7.13752806e-01
-1.62233472e-01 3.45657110e-01 8.50287974e-01 8.14910173e-01
1.28284991e+00 3.36641483e-02 1.04089296e+00 -7.17808247e-01
-1.08276641e+00 -6.47180140e-01 -8.28106821e-01 5.24016261e-01
2.14461774e-01 -4.22553688e-01 -4.70301598e-01 -6.45467490e-02]
|
[5.835948944091797, 6.3117804527282715]
|
399104eb-d64f-4a34-b39e-a39450170eaf
|
towards-building-a-crowd-sourced-sky-map
|
1406.1528
| null |
http://arxiv.org/abs/1406.1528v1
|
http://arxiv.org/pdf/1406.1528v1.pdf
|
Towards building a Crowd-Sourced Sky Map
|
We describe a system that builds a high dynamic-range and wide-angle image of
the night sky by combining a large set of input images. The method makes use of
pixel-rank information in the individual input images to improve a "consensus"
pixel rank in the combined image. Because it only makes use of ranks and the
complexity of the algorithm is linear in the number of images, the method is
useful for large sets of uncalibrated images that might have undergone unknown
non-linear tone mapping transformations for visualization or aesthetic reasons.
We apply the method to images of the night sky (of unknown provenance)
discovered on the Web. The method permits discovery of astronomical objects or
features that are not visible in any of the input images taken individually.
More importantly, however, it permits scientific exploitation of a huge source
of astronomical images that would not be available to astronomical research
without our automatic system.
|
['Bernhard Scholkopf', 'Dustin Lang', 'David W. Hogg']
|
2014-06-05
| null | null | null | null |
['tone-mapping']
|
['computer-vision']
|
[ 3.72685552e-01 -8.88251662e-02 7.25907683e-01 -1.76349372e-01
-6.15256131e-01 -1.18090391e+00 6.58389688e-01 -1.07348815e-01
-3.94948691e-01 5.71667552e-01 -1.84768021e-01 -6.17726624e-01
-6.21797815e-02 -7.69934535e-01 -6.39813542e-01 -8.02792132e-01
-1.19610270e-02 2.72903740e-01 7.54871070e-01 -2.12947026e-01
3.02742243e-01 7.09759057e-01 -1.90679526e+00 4.01682198e-01
3.56046528e-01 7.59244382e-01 4.18449730e-01 9.75413561e-01
-9.24750790e-02 5.90424180e-01 -6.87656999e-01 -4.26663607e-02
8.08354020e-01 -4.89459544e-01 -5.54878533e-01 -8.44508111e-02
1.11538053e+00 -8.10160562e-02 -6.33277968e-02 1.37487185e+00
2.26913005e-01 1.69056132e-02 1.72803044e-01 -9.41910267e-01
-4.21509355e-01 -1.57260090e-01 -6.88006103e-01 4.69897360e-01
2.26945654e-01 1.28844216e-01 8.98628652e-01 -6.52358174e-01
8.78848553e-01 8.88012052e-01 4.38483745e-01 -1.75436825e-01
-1.43061090e+00 -5.06517053e-01 -3.98920894e-01 4.00204808e-01
-1.34059608e+00 -2.94523627e-01 4.99345034e-01 -4.16477710e-01
4.15001214e-01 9.22510207e-01 7.96731412e-01 2.87555814e-01
-1.49232388e-01 -1.72478671e-03 1.95351577e+00 -6.77737415e-01
1.41471609e-01 2.91076303e-01 -1.13324650e-01 6.50942564e-01
3.29939783e-01 1.75173283e-01 -5.14256954e-01 -3.28281283e-01
8.13216090e-01 -2.19607502e-01 -4.34011728e-01 -2.87822187e-01
-1.48349261e+00 3.36275667e-01 5.51669300e-01 3.48330438e-01
-2.60262806e-02 -8.04027617e-02 -1.40744224e-01 5.66753685e-01
3.22254300e-01 8.76218498e-01 -4.36296046e-01 5.06273694e-02
-1.25415540e+00 4.64280546e-02 6.07542217e-01 3.82077217e-01
1.08578753e+00 -2.60954142e-01 5.35357416e-01 5.38102925e-01
-1.49961039e-01 6.06455386e-01 2.28837103e-01 -1.17544210e+00
-1.05948530e-01 5.58872104e-01 3.03991944e-01 -1.03815317e+00
-3.73574734e-01 -2.70024361e-03 -4.75052327e-01 1.18810904e+00
7.12759554e-01 3.67313385e-01 -1.05927265e+00 1.09567034e+00
5.85884273e-01 -1.24010824e-01 -1.95973054e-01 1.08332872e+00
7.36288309e-01 5.30853510e-01 -4.56128895e-01 -2.33982783e-02
1.71222281e+00 -6.34992957e-01 -3.80125999e-01 -2.12847799e-01
1.48921013e-01 -1.16851985e+00 1.35962796e+00 6.50402784e-01
-9.09687221e-01 -5.45917869e-01 -1.28082120e+00 -1.88939705e-01
-7.33851552e-01 3.41973677e-02 2.43263856e-01 5.57628334e-01
-1.17034507e+00 6.47218466e-01 -2.49117509e-01 -4.33148712e-01
8.18668678e-03 9.82032046e-02 -3.95042181e-01 -7.57018337e-03
-9.03128266e-01 8.93628836e-01 3.79147142e-01 -1.13368474e-01
-3.06458831e-01 -4.72683489e-01 -2.51679301e-01 -3.81813906e-02
3.97184402e-01 -3.55907977e-01 9.07705426e-01 -1.25949526e+00
-1.09564126e+00 1.20815969e+00 8.62146616e-02 -6.42348372e-04
5.95727563e-01 1.56245492e-02 -7.08265424e-01 4.73138660e-01
2.41555758e-02 3.48067015e-01 9.67599034e-01 -1.35324526e+00
-7.85271406e-01 -3.19383055e-01 8.89443904e-02 2.34482259e-01
-1.20161712e-01 1.44436151e-01 -6.97606564e-01 -5.71436286e-01
4.34494257e-01 -1.17513919e+00 -1.50659278e-01 3.04286063e-01
-7.52451271e-02 5.36836088e-01 9.49744999e-01 -9.06415939e-01
9.29897010e-01 -2.36844659e+00 -2.97197163e-01 5.84015727e-01
1.99036583e-01 5.33040753e-03 2.02187933e-02 3.38936269e-01
-1.33736432e-01 1.98563486e-01 -2.11797372e-01 3.72032076e-01
-4.27918434e-01 2.07729176e-01 -3.52147013e-01 6.52285814e-01
-3.37916881e-01 4.58604425e-01 -8.81383359e-01 -5.44348836e-01
2.31679142e-01 1.14474617e-01 9.65360627e-02 1.03927344e-01
-2.09265143e-01 4.13571686e-01 -2.70431526e-02 5.27563691e-01
7.85914540e-01 -2.98133820e-01 1.52912855e-01 -3.45975816e-01
-6.12007141e-01 1.36825025e-01 -1.41042399e+00 1.45324481e+00
-3.69243808e-02 1.00631261e+00 1.66492648e-02 -4.43357863e-02
8.18341911e-01 6.29751571e-03 2.26317421e-01 -8.31383228e-01
-3.20780724e-01 1.93973646e-01 -1.94540575e-01 -4.11611855e-01
8.96144331e-01 -1.75294161e-01 2.16267705e-01 7.06811070e-01
-4.04522896e-01 -5.54593742e-01 3.51616532e-01 3.46055567e-01
1.05002761e+00 1.45204917e-01 1.77783191e-01 -3.24221104e-01
2.92880267e-01 3.98009390e-01 1.75197437e-01 7.60495245e-01
3.12330484e-01 1.10551655e+00 3.48311633e-01 -7.44582117e-01
-1.57696366e+00 -1.08055151e+00 -3.66360486e-01 1.05611050e+00
2.50098556e-01 -3.49857837e-01 -2.94507504e-01 -3.03798437e-01
-1.52008712e-01 4.01448816e-01 -7.44456947e-01 4.63298112e-01
-2.18497649e-01 -7.89226711e-01 3.16406161e-01 -6.06577359e-02
2.68707782e-01 -8.84243846e-01 -9.33222532e-01 -2.45450035e-01
-1.30845785e-01 -6.95965827e-01 -7.95494393e-02 2.27090511e-02
-6.54730260e-01 -1.37063503e+00 -4.29013133e-01 -3.58272463e-01
1.01936591e+00 7.12944269e-01 1.10663569e+00 2.04038352e-01
-7.98978031e-01 2.96244264e-01 -3.72909874e-01 -2.91491836e-01
-3.07037681e-01 -4.18070078e-01 -1.68607280e-01 -6.45717382e-02
1.05433628e-01 -4.66376603e-01 -6.28635168e-01 5.11245310e-01
-1.25692022e+00 2.42744163e-01 3.66652668e-01 7.02473640e-01
7.44859457e-01 4.34502214e-01 -1.43701121e-01 -9.80640352e-01
2.84463823e-01 -2.81121552e-01 -1.17628169e+00 3.98290068e-01
-7.33345747e-01 1.59839079e-01 4.30178583e-01 -1.62416875e-01
-1.13535881e+00 1.94935337e-01 6.47161901e-01 -3.74903321e-01
3.36547866e-02 3.25517416e-01 3.32856387e-01 -3.96595210e-01
1.07049465e+00 -5.11380099e-02 -2.03323811e-01 -7.66391039e-01
7.25464225e-01 5.26163995e-01 1.27113569e+00 -1.98275615e-02
1.05172932e+00 8.72981429e-01 2.36986317e-02 -6.80381775e-01
-5.17386734e-01 -7.43915439e-01 -7.54069507e-01 -3.09783280e-01
5.38791656e-01 -8.48815739e-01 -2.99446225e-01 1.10975698e-01
-7.31501698e-01 -4.15749997e-02 -4.47990865e-01 2.42033616e-01
-2.09449619e-01 6.10757709e-01 -1.31675890e-02 -6.19359493e-01
2.92734206e-02 -5.81193268e-01 6.55981302e-01 4.63070989e-01
-2.39738181e-01 -5.31454086e-01 2.63839155e-01 2.47147799e-01
1.78444251e-01 3.23610663e-01 7.20061898e-01 -1.94047555e-01
-8.03227067e-01 -4.00049508e-01 -5.19175231e-01 6.01860024e-02
1.62712693e-01 3.70031625e-01 -1.23529458e+00 -2.57585794e-01
-7.91229755e-02 -7.63831213e-02 7.76235104e-01 1.10411840e-02
1.01741672e+00 -1.78234592e-01 7.71380216e-02 4.74533856e-01
1.52762854e+00 -5.49862720e-03 9.30794656e-01 8.71988952e-01
5.88833392e-01 7.10741103e-01 5.18058419e-01 2.43910506e-01
-5.48005151e-03 6.78872645e-01 4.73897547e-01 -5.92208266e-01
-1.25834063e-01 1.20883919e-01 5.52532561e-02 3.86639416e-01
-2.97947228e-01 1.85968623e-01 -8.01541865e-01 6.72990859e-01
-1.68100727e+00 -1.18800628e+00 -6.54067814e-01 2.68459105e+00
7.37203062e-01 -1.00470982e-01 -6.27554730e-02 9.49311012e-04
6.75407171e-01 1.15529336e-01 -4.28281724e-01 -4.14682448e-01
-5.79071462e-01 2.43015990e-01 1.02433479e+00 1.46678045e-01
-9.57316160e-01 4.39150274e-01 7.28819370e+00 4.98417020e-01
-1.04748285e+00 -1.21027045e-01 5.03710926e-01 -3.66832048e-01
-5.01572371e-01 3.42264980e-01 -1.12782687e-01 4.26401436e-01
7.30519474e-01 -1.56697780e-01 6.63384736e-01 6.54805720e-01
1.83872223e-01 -8.59168410e-01 -6.50884390e-01 8.26329708e-01
-1.05980866e-01 -1.31554782e+00 -5.37982762e-01 1.24479324e-01
1.02026200e+00 1.88615978e-01 -4.61644605e-02 -6.11782968e-01
5.50665557e-01 -7.39250362e-01 6.72219157e-01 7.51006365e-01
1.05813289e+00 -5.14795125e-01 4.94830966e-01 1.66341066e-02
-7.49523103e-01 1.36265844e-01 -3.87268245e-01 -5.63299516e-03
8.29808041e-03 8.33536148e-01 -8.61412406e-01 3.33895892e-01
1.29240334e+00 5.61832413e-02 -1.17993486e+00 1.34215355e+00
-3.37650001e-01 3.09921175e-01 -7.05767870e-01 3.45009476e-01
-2.70166248e-02 -3.97177637e-01 4.01972979e-01 8.67450416e-01
5.86605012e-01 8.82978216e-02 -2.44109899e-01 5.61808705e-01
1.03652090e-01 1.03824541e-01 -6.60909832e-01 1.69224739e-02
3.81676704e-01 1.72656727e+00 -8.93587649e-01 -5.20032942e-01
-5.00932157e-01 9.47070301e-01 -6.66100904e-02 1.78058535e-01
-2.55741030e-01 -3.50043476e-01 2.55427808e-01 4.40166503e-01
3.70153368e-01 -2.82902926e-01 -4.88499135e-01 -9.97973561e-01
7.40613490e-02 -9.23902094e-01 6.25332594e-01 -1.62587202e+00
-1.04809141e+00 6.83119655e-01 5.49701089e-03 -1.60477543e+00
-1.81322604e-01 -3.79961073e-01 -5.63496947e-01 1.07801616e+00
-1.24875939e+00 -8.79430056e-01 -5.32631278e-01 3.61094236e-01
-1.37977213e-01 1.12617083e-01 8.38757575e-01 -5.04320599e-02
1.28863290e-01 -2.59396762e-01 7.00968504e-01 -4.62777615e-01
1.02440524e+00 -1.50247490e+00 5.75976312e-01 1.14057302e+00
4.76774901e-01 2.82442480e-01 1.02973068e+00 -5.26443839e-01
-1.00599635e+00 -7.08213091e-01 8.04414511e-01 -5.03705025e-01
8.27170610e-01 -4.32052612e-02 -1.06415856e+00 2.27132618e-01
3.10140967e-01 6.44029528e-02 5.50760508e-01 -8.31296667e-02
-5.61770916e-01 -5.97354621e-02 -1.05907917e+00 6.22143388e-01
5.49783468e-01 -7.05760658e-01 -5.41594744e-01 4.37460184e-01
1.41225860e-01 -3.09419036e-01 -6.32947862e-01 -3.91496113e-03
7.34506905e-01 -1.04309177e+00 1.07390976e+00 -2.16518283e-01
1.72819361e-01 -1.20160902e+00 2.77019590e-02 -1.08874476e+00
-3.83141249e-01 -4.40089822e-01 6.65015697e-01 7.22019255e-01
3.82522672e-01 -3.87104988e-01 3.22925001e-01 7.66003013e-01
2.00734094e-01 1.67217627e-01 -9.34219539e-01 -6.61930025e-01
-4.75190878e-01 -1.30027652e-01 4.96210188e-01 9.89043355e-01
-2.45516911e-01 -1.59352884e-01 -6.93865240e-01 3.52593631e-01
6.08783841e-01 6.70919120e-01 1.04825664e+00 -1.37305355e+00
-4.71170485e-01 -9.65632573e-02 -3.45940828e-01 -1.89184442e-01
-8.04467142e-01 -6.07730210e-01 1.56453207e-01 -1.32141197e+00
7.61315674e-02 -5.22788286e-01 -3.91194999e-01 7.27143228e-01
-1.60738900e-01 1.03169346e+00 3.16436142e-01 8.04982066e-01
-4.35905427e-01 -1.95879087e-01 1.01968896e+00 3.72609659e-03
-7.23385215e-02 -3.71160358e-01 -5.60697734e-01 8.38736236e-01
8.21220756e-01 -6.88297093e-01 -2.59361528e-02 -1.87104464e-01
5.04149318e-01 -2.81779349e-01 6.41539633e-01 -1.03238320e+00
3.50124501e-02 -3.46616477e-01 6.84684336e-01 -6.48715377e-01
2.13291869e-01 -8.13853443e-01 8.99833322e-01 9.13890004e-02
-2.21711546e-02 2.69995779e-01 2.10541636e-01 4.68178898e-01
-1.18629023e-01 -3.67729515e-01 8.96169543e-01 -5.43856204e-01
-1.04474306e+00 -3.77057463e-01 -2.68493801e-01 -4.00890231e-01
7.83786058e-01 -4.00146365e-01 -7.04221487e-01 -2.99034208e-01
-5.34349263e-01 -1.48968786e-01 1.37263715e+00 2.95876920e-01
1.98134065e-01 -1.18866503e+00 -4.98206854e-01 3.15205306e-01
4.05405611e-01 -1.21832550e-01 2.34093696e-01 6.08197391e-01
-1.09309649e+00 -1.89576492e-01 -6.61808491e-01 -3.91851693e-01
-1.91744816e+00 5.01474142e-01 -1.74957551e-02 -8.95574316e-03
-9.36769605e-01 5.84846020e-01 3.65842432e-01 -6.44326434e-02
-3.96884412e-01 1.92153215e-01 4.32231314e-02 1.13999963e-01
9.13087428e-01 2.17465878e-01 1.82862818e-01 -6.18919671e-01
-1.52100429e-01 5.40281951e-01 1.51980355e-01 -5.69276869e-01
1.40677440e+00 -3.95292968e-01 -4.57717001e-01 6.72447026e-01
7.92376220e-01 5.07673800e-01 -1.14827108e+00 -3.70811373e-01
-1.51667416e-01 -1.22512043e+00 2.13897660e-01 -1.10492814e+00
-8.56336176e-01 6.83005512e-01 8.95491838e-01 5.25250494e-01
1.44794345e+00 2.86999694e-03 1.59443483e-01 4.71504956e-01
2.18572840e-01 -1.30960178e+00 -2.06402436e-01 1.20639624e-02
8.21327865e-01 -1.16142094e+00 6.20841622e-01 -6.72578737e-02
-4.40843433e-01 1.50043774e+00 1.67986169e-01 2.54630893e-01
2.31417507e-01 1.33199111e-01 6.39936745e-01 -4.82981235e-01
-5.54593265e-01 -3.87760043e-01 5.18306017e-01 4.76084322e-01
5.19074947e-02 2.67642383e-02 -4.56757814e-01 -2.73077041e-01
-3.64852935e-01 -1.39187828e-01 9.47891414e-01 9.03936744e-01
-6.71653926e-01 -1.19251001e+00 -1.14655817e+00 4.46796834e-01
-2.55916983e-01 -3.21292251e-01 -6.95475757e-01 5.10801971e-01
2.86082387e-01 6.37648702e-01 1.65372372e-01 -2.25597605e-01
1.07955523e-01 6.72707930e-02 4.21690404e-01 -1.90219283e-01
-3.93029928e-01 2.12947667e-01 2.51361489e-01 -3.74757260e-01
-5.22922695e-01 -7.34337270e-01 -8.82762790e-01 -3.15672338e-01
-2.44670153e-01 5.67672960e-03 1.13260424e+00 4.26010489e-01
2.33169600e-01 9.23923776e-02 6.39878213e-01 -7.18369365e-01
1.95646435e-01 -5.47143281e-01 -8.33453834e-01 7.32975543e-01
2.42985412e-01 -3.12096506e-01 -4.23684478e-01 5.86413026e-01]
|
[10.511492729187012, -2.501208543777466]
|
4852b351-3a4e-44fa-9c17-34b28a9bd8a6
|
neurst-neural-speech-translation-toolkit
|
2012.10018
| null |
https://arxiv.org/abs/2012.10018v3
|
https://arxiv.org/pdf/2012.10018v3.pdf
|
NeurST: Neural Speech Translation Toolkit
|
NeurST is an open-source toolkit for neural speech translation. The toolkit mainly focuses on end-to-end speech translation, which is easy to use, modify, and extend to advanced speech translation research and products. NeurST aims at facilitating the speech translation research for NLP researchers and building reliable benchmarks for this field. It provides step-by-step recipes for feature extraction, data preprocessing, distributed training, and evaluation. In this paper, we will introduce the framework design of NeurST and show experimental results for different benchmark datasets, which can be regarded as reliable baselines for future research. The toolkit is publicly available at https://github.com/bytedance/neurst/ and we will continuously update the performance of NeurST with other counterparts and studies at https://st-benchmark.github.io/.
|
['Lei LI', 'Rong Ye', 'Qianqian Dong', 'Mingxuan Wang', 'Chengqi Zhao']
|
2020-12-18
| null |
https://aclanthology.org/2021.acl-demo.7
|
https://aclanthology.org/2021.acl-demo.7.pdf
|
acl-2021-5
|
['speech-to-text-translation']
|
['natural-language-processing']
|
[-4.96018827e-02 -1.36245981e-01 -4.61889237e-01 -4.21665162e-01
-1.31349194e+00 -6.22236788e-01 7.25558698e-01 -5.97212255e-01
-3.58612180e-01 6.84171915e-01 4.45904225e-01 -8.78572702e-01
6.44588947e-01 -4.17924345e-01 -7.76141942e-01 -6.48954332e-01
4.24925029e-01 9.01777089e-01 4.08699252e-02 -5.18130481e-01
-3.46465647e-01 -1.27015682e-02 -1.02314746e+00 5.81215739e-01
6.58452809e-01 5.31708777e-01 4.42343414e-01 5.29887974e-01
-1.35483267e-02 5.29699206e-01 -5.37777305e-01 -7.52606511e-01
6.70594051e-02 -6.82832897e-01 -1.05528450e+00 -3.55125815e-01
2.21678596e-02 -3.54736239e-01 -4.70236003e-01 8.16403329e-01
1.11350369e+00 2.48559341e-01 2.55526274e-01 -1.17814076e+00
-1.17820477e+00 1.17326677e+00 1.27487555e-01 2.90519685e-01
2.71257907e-01 2.74087071e-01 9.94790137e-01 -1.38012731e+00
6.36686862e-01 1.14709866e+00 3.41195226e-01 9.45324898e-01
-6.68588400e-01 -7.05974102e-01 -2.18627751e-01 2.92925179e-01
-1.26316035e+00 -1.26398170e+00 5.55522442e-01 -9.52823460e-02
1.24243307e+00 5.95121980e-01 3.35244983e-01 1.98379290e+00
-3.07836756e-02 1.34166539e+00 9.87964690e-01 -3.50894123e-01
8.39775652e-02 1.87470838e-02 -1.08665571e-01 4.47553933e-01
-4.27895486e-01 2.85023361e-01 -6.82671309e-01 -9.38150883e-02
7.42410421e-01 -4.67162699e-01 -3.33169699e-01 1.96984410e-01
-1.71667957e+00 7.58524060e-01 2.21381962e-01 5.90136588e-01
-2.52969503e-01 -1.73354130e-02 6.09743297e-01 7.05463231e-01
7.03889072e-01 1.26100942e-01 -6.68242037e-01 -7.35231757e-01
-6.79524541e-01 1.12007380e-01 8.85437012e-01 1.30852544e+00
2.94312328e-01 3.01365942e-01 -4.34073508e-01 1.29858792e+00
1.10838063e-01 9.65347528e-01 8.66378069e-01 -8.71778846e-01
9.10426915e-01 -1.56236701e-02 -2.02639073e-01 -3.17056894e-01
-3.77910472e-02 -4.56327736e-01 -8.30233812e-01 -6.45585775e-01
3.45215835e-02 -4.65138286e-01 -8.90674055e-01 1.43793702e+00
1.76985592e-01 1.65271312e-01 3.70147198e-01 1.11043334e+00
1.54369354e+00 1.03650308e+00 -2.34433666e-01 -1.39231130e-01
1.20457721e+00 -1.63720906e+00 -9.03610826e-01 -2.90178716e-01
8.31876099e-01 -1.22620237e+00 1.55768740e+00 -1.17351055e-01
-1.34338880e+00 -3.97079825e-01 -6.02934837e-01 -2.46840611e-01
-3.78291726e-01 5.74916959e-01 5.21544576e-01 5.01982212e-01
-1.53712285e+00 2.57030398e-01 -1.27374840e+00 -6.51813805e-01
8.38998854e-02 2.25515470e-01 -1.54257849e-01 2.36707821e-01
-1.63104630e+00 8.73499393e-01 1.94413260e-01 8.47302303e-02
-9.39425707e-01 -2.79213220e-01 -7.31477201e-01 -1.14294142e-01
2.77008176e-01 -8.79049718e-01 2.22759914e+00 -8.35074306e-01
-2.12895656e+00 6.38822854e-01 -4.91407573e-01 -3.77449930e-01
3.65459889e-01 -2.58508977e-02 -6.81407154e-01 -2.41471976e-01
1.58059478e-01 5.77651918e-01 5.56916535e-01 -6.90874279e-01
-1.67087018e-01 -5.85013777e-02 -4.33866173e-01 5.11307597e-01
-3.54393870e-01 6.69127226e-01 -8.31466079e-01 -7.71072924e-01
1.00034038e-02 -1.19716358e+00 -7.84964561e-02 -4.58156317e-01
-6.37054741e-01 -2.71855205e-01 6.14078045e-01 -7.93937922e-01
1.15478110e+00 -2.11069298e+00 2.78589666e-01 -4.05907035e-01
-1.84540868e-01 5.56483269e-01 -4.64287281e-01 9.11870420e-01
2.49973927e-02 7.11194531e-04 -2.49732494e-01 -8.12166035e-01
1.19779229e-01 1.00198969e-01 -4.63359565e-01 1.47612646e-01
1.07995130e-03 1.42937136e+00 -7.26488709e-01 -3.94115821e-02
1.76474139e-01 6.22903407e-01 3.24008218e-03 3.49348217e-01
-1.07824855e-01 5.73461831e-01 -4.88079518e-01 7.03218162e-01
3.84635627e-01 -2.64206022e-01 -1.77203804e-01 2.90309906e-01
-8.16339999e-02 1.20000601e+00 -3.58958393e-01 1.97204733e+00
-4.97792780e-01 6.69981778e-01 -1.44737065e-02 -6.81381762e-01
9.46290433e-01 7.83283293e-01 8.41881633e-02 -7.77575493e-01
3.98563057e-01 5.96495271e-01 -2.25969866e-01 -3.05986881e-01
5.31489670e-01 2.92387098e-01 -6.42133504e-02 5.83406925e-01
2.26034001e-01 -1.76680803e-01 3.66651900e-02 8.02474692e-02
8.61193538e-01 -4.27797176e-02 8.49910453e-02 -4.56762351e-02
4.29318458e-01 3.63299251e-02 3.49485159e-01 3.65288287e-01
-1.90871716e-01 5.21553755e-01 5.95778227e-02 -3.13912719e-01
-9.51566041e-01 -1.17469025e+00 9.65643451e-02 1.43249822e+00
-3.38489264e-01 -5.04596531e-01 -1.04234302e+00 -6.54241025e-01
-3.68973970e-01 8.36607695e-01 -2.48588115e-01 -8.84834304e-02
-5.66350639e-01 -6.61502421e-01 1.09607673e+00 4.51639295e-01
6.56950772e-01 -1.45207703e+00 3.80819380e-01 7.72329718e-02
-8.61129165e-01 -1.36728752e+00 -8.86273026e-01 -1.52220367e-03
-6.83265209e-01 -2.83638418e-01 -9.30819392e-01 -1.17609692e+00
2.93688893e-01 5.31006098e-01 1.20138812e+00 9.59872641e-03
4.49423850e-01 -5.44490218e-02 -6.75999761e-01 -2.70441145e-01
-8.46305728e-01 7.61420488e-01 1.46998912e-01 -3.31778049e-01
4.15186018e-01 -5.40086508e-01 -3.20127875e-01 6.26547754e-01
-4.83677924e-01 3.01179379e-01 7.10253000e-01 7.78018594e-01
4.86082941e-01 -6.91720009e-01 5.71104169e-01 -6.55765533e-01
1.00229084e+00 -4.39644605e-01 -4.27399039e-01 2.46963710e-01
-3.30587387e-01 -1.43711656e-01 5.92112720e-01 -2.94373721e-01
-8.49476814e-01 -1.95843041e-01 -8.45963120e-01 -2.76857495e-01
-1.68134019e-01 7.02915430e-01 -8.87237042e-02 2.14497492e-01
5.27778506e-01 7.29637861e-01 -7.54165836e-03 -6.19324505e-01
3.59108180e-01 1.33138859e+00 3.30355167e-01 -5.75263321e-01
6.02128923e-01 -1.00433886e-01 -8.77581954e-01 -6.76410198e-01
-4.98112470e-01 -3.64155024e-01 -4.18248147e-01 2.38454178e-01
5.96068144e-01 -1.08748186e+00 -1.68150350e-01 5.40095091e-01
-1.36223006e+00 -9.01619613e-01 1.55522600e-01 5.20135880e-01
-6.99874938e-01 8.08528513e-02 -9.74945247e-01 -2.91179389e-01
-9.16423202e-01 -1.61052656e+00 1.31367373e+00 -1.66178972e-01
-9.66576487e-02 -1.10565078e+00 2.56538004e-01 7.98195899e-01
6.50553346e-01 -6.25326395e-01 1.83879852e-01 -9.30348873e-01
-3.32480431e-01 2.30149895e-01 9.76852924e-02 5.20172238e-01
8.67894888e-02 3.40974471e-03 -9.95287001e-01 -3.72047544e-01
-1.50329307e-01 -4.31555957e-01 6.98328733e-01 1.89490929e-01
8.39854121e-01 -6.28011227e-01 -2.24927410e-01 7.04156935e-01
5.26510537e-01 1.75547287e-01 5.33634365e-01 3.30511481e-01
5.91078997e-01 3.51261377e-01 4.44001913e-01 -1.41945958e-01
7.07893848e-01 1.09840894e+00 -1.90502703e-01 -2.42907256e-01
-4.93147433e-01 -2.16252640e-01 9.91858006e-01 1.81613088e+00
1.69990975e-02 -5.41424274e-01 -1.21733463e+00 5.44223666e-01
-1.98321533e+00 -8.00102472e-01 -1.13293029e-01 1.82072854e+00
1.05687356e+00 -3.05277735e-01 2.64051795e-01 -3.47829789e-01
7.30527520e-01 1.29014701e-01 -3.58375788e-01 -5.63173175e-01
-2.02348456e-01 2.13648871e-01 1.79336354e-01 6.11210823e-01
-8.81887376e-01 1.80622160e+00 5.98952675e+00 1.08062315e+00
-1.38296616e+00 7.35911369e-01 7.16697633e-01 -2.41802260e-02
-1.82555780e-01 -1.32455930e-01 -9.38901901e-01 5.43020844e-01
1.56932557e+00 -4.53431517e-01 7.93192565e-01 7.64423072e-01
4.91011113e-01 7.13315010e-01 -8.57519090e-01 7.54769683e-01
-1.31397441e-01 -1.39383292e+00 1.99353322e-02 -1.08524330e-01
6.43670082e-01 1.11490250e+00 3.75400007e-01 5.82139075e-01
5.28885126e-01 -7.92397857e-01 5.27377486e-01 -2.99150854e-01
9.98315513e-01 -5.69496274e-01 8.67984951e-01 3.51777554e-01
-9.57752287e-01 2.65749216e-01 -3.73937398e-01 4.06583920e-02
2.98911959e-01 3.08729768e-01 -1.21473241e+00 6.15766406e-01
5.88222623e-01 8.72804284e-01 -2.60600388e-01 6.28158569e-01
-5.93002558e-01 1.05791211e+00 -2.03503385e-01 -2.54342526e-01
4.38257515e-01 -1.31177261e-01 5.02908587e-01 1.48044980e+00
5.17619550e-01 -2.04744309e-01 1.24363862e-01 6.53094590e-01
-4.05082673e-01 4.91946757e-01 -6.88506365e-01 -3.79511088e-01
8.74361813e-01 1.15342140e+00 -4.05046076e-01 -3.93729419e-01
-5.26286006e-01 1.22252727e+00 3.32570583e-01 6.38133049e-01
-9.23253536e-01 -2.09614992e-01 8.47323298e-01 -1.79877505e-01
2.25189745e-01 -4.49809432e-01 -2.09649086e-01 -1.44261563e+00
2.16920361e-01 -1.26369166e+00 4.03179862e-02 -7.42478907e-01
-1.04482162e+00 1.22806740e+00 -2.37110332e-01 -1.28873408e+00
-5.70089459e-01 -4.06156033e-01 -6.28879070e-01 9.08601642e-01
-1.30886829e+00 -1.33730447e+00 1.22764342e-01 7.52907991e-01
1.04723895e+00 -5.91779888e-01 1.04310787e+00 4.22173977e-01
-8.91466379e-01 1.09451759e+00 2.47786880e-01 5.65872490e-01
8.42219889e-01 -8.20287824e-01 1.52986860e+00 7.69076586e-01
4.12269503e-01 7.40281701e-01 5.46030104e-01 -5.19547582e-01
-1.60210907e+00 -1.31371152e+00 1.33095360e+00 -6.81529522e-01
1.06990063e+00 -7.55085588e-01 -6.50642395e-01 9.91678119e-01
7.02921569e-01 -3.10494989e-01 6.17922604e-01 1.73909724e-01
-1.80938661e-01 8.50777701e-02 -6.53264821e-01 8.76982808e-01
1.16304219e+00 -7.03944862e-01 -3.08914989e-01 7.28126109e-01
1.24176598e+00 -8.92397344e-01 -8.94779503e-01 2.00220212e-01
3.69864047e-01 -5.02141654e-01 7.29168892e-01 -3.86437982e-01
3.01188231e-01 -2.35895030e-02 -1.73472479e-01 -1.75637770e+00
-1.38923407e-01 -1.22730410e+00 -1.78826764e-01 1.04861450e+00
1.14001787e+00 -9.10607457e-01 4.02954966e-01 3.52570340e-02
-8.14027786e-01 -8.61705542e-01 -1.20566213e+00 -1.07439268e+00
4.28246349e-01 -4.44877803e-01 9.11462843e-01 1.14979947e+00
8.76648277e-02 7.19581485e-01 -3.35482895e-01 6.32783994e-02
6.99767619e-02 -1.02985144e-01 8.92910957e-01 -3.91938150e-01
-4.72549438e-01 -6.76602006e-01 1.84158850e-02 -1.42667603e+00
2.49708638e-01 -1.21701527e+00 1.84876118e-02 -1.62659931e+00
2.08087638e-02 -1.45081982e-01 -2.30165143e-02 9.60431993e-01
-9.24637392e-02 4.41896439e-01 3.54583748e-02 4.03067470e-01
-3.28996658e-01 9.04575586e-01 1.44204283e+00 -1.87179387e-01
-3.56482029e-01 3.26531857e-01 -7.07424283e-01 3.71514224e-02
1.60321963e+00 -4.66785312e-01 -4.87972528e-01 -1.06697869e+00
-1.65562555e-01 1.92091111e-02 8.34028274e-02 -7.25131452e-01
1.93846658e-01 -1.23700395e-01 -3.00128341e-01 -5.12767851e-01
5.00606358e-01 -3.48179847e-01 -5.55851124e-02 5.96179329e-02
-4.05673712e-01 5.23594379e-01 2.99087763e-01 -2.93239474e-01
-3.62384856e-01 1.90105766e-01 3.20044965e-01 -6.08487651e-02
-3.05857092e-01 4.65312719e-01 -3.53484422e-01 -1.40466476e-02
5.24354815e-01 3.28443378e-01 -6.22913539e-01 -7.23742664e-01
-6.86009884e-01 1.88454628e-01 2.67389774e-01 9.88861263e-01
6.20123625e-01 -1.57592273e+00 -1.21081841e+00 2.62384325e-01
2.64437020e-01 -3.13655257e-01 -1.25404477e-01 1.17840159e+00
-3.43526781e-01 7.74692178e-01 1.27631649e-01 -5.55100024e-01
-1.22100973e+00 1.93484813e-01 3.27776194e-01 6.53298646e-02
-5.67088425e-01 9.36642051e-01 -1.09404720e-01 -1.17991126e+00
1.00970447e-01 -4.08110619e-01 1.75688788e-01 -6.15502536e-01
6.74303472e-01 1.63964972e-01 5.38479090e-01 -8.45090210e-01
-3.15400720e-01 -1.32431807e-02 -2.91258305e-01 -4.79779333e-01
1.15218461e+00 -1.85496271e-01 -2.25196585e-01 5.93079805e-01
1.09860551e+00 -2.46651292e-01 -6.73992038e-01 -4.27156955e-01
-2.42502972e-01 -1.41342485e-03 1.18691675e-01 -1.13656700e+00
-1.16817403e+00 9.48033214e-01 1.77142426e-01 -9.45908278e-02
1.02193570e+00 3.11429292e-01 1.33181524e+00 7.54940569e-01
3.82043093e-01 -9.52982903e-01 -2.56471187e-01 1.07223213e+00
1.15847325e+00 -1.11867118e+00 -5.46315968e-01 -4.17378128e-01
-9.09471214e-01 7.72102475e-01 4.13495153e-01 3.84650290e-01
4.00219411e-01 4.00346011e-01 6.30993068e-01 2.33328372e-01
-1.09250426e+00 -1.37828812e-01 3.97163391e-01 5.47370136e-01
9.47692454e-01 3.81838948e-01 -2.90411919e-01 6.65342450e-01
-8.44101727e-01 -7.12912232e-02 2.78663218e-01 6.53062701e-01
-1.78960755e-01 -1.58602643e+00 -3.66713256e-02 2.49864738e-02
-2.90914446e-01 -7.87368357e-01 -9.15635169e-01 5.24838030e-01
-4.96589839e-01 1.15647531e+00 -2.94429868e-01 -7.66325712e-01
3.86824042e-01 2.13866800e-01 1.80686146e-01 -6.68115973e-01
-6.49031460e-01 2.04079121e-01 5.70931017e-01 -5.77618480e-01
-1.79911759e-02 -7.39620149e-01 -1.14165187e+00 -6.56021655e-01
-1.45769566e-01 5.20741343e-01 9.12887156e-01 7.24612594e-01
7.61035144e-01 2.71187484e-01 7.28840172e-01 -5.96908331e-01
-5.99998116e-01 -1.31599486e+00 2.12115139e-01 -1.46834284e-01
2.67754763e-01 -2.22763002e-01 -3.21237624e-01 -4.36435826e-02]
|
[14.475046157836914, 7.166932582855225]
|
920babf4-8a51-4cc7-9290-e5e308eabfc6
|
the-trajectory-of-voice-onset-time-with-vocal
|
1810.07030
| null |
http://arxiv.org/abs/1810.07030v1
|
http://arxiv.org/pdf/1810.07030v1.pdf
|
The Trajectory of Voice Onset Time with Vocal Aging
|
Vocal aging, a universal process of human aging, can largely affect one's
language use, possibly including some subtle acoustic features of one's
utterances like Voice Onset Time. To figure out the time effects, Queen
Elizabeth's Christmas speeches are documented and analyzed in the long-term
trend. We build statistical models of time dependence in Voice Onset Time,
controlling a wide range of other fixed factors, to present annual variations
and the simulated trajectory. It is revealed that the variation range of Voice
Onset Time has been narrowing over fifty years with a slight reduction in the
mean value, which, possibly, is an effect of diminishing exertion, resulting
from subdued muscle contraction, transcending other non-linguistic factors in
forming Voice Onset Time patterns over a long time.
|
['Jian Hu', 'Xuanda Chen', 'Ziyu Xiong']
|
2018-10-15
| null | null | null | null |
['human-aging']
|
['miscellaneous']
|
[-3.17112505e-01 -5.30800410e-02 -3.71378273e-01 -1.06546283e-02
-2.66470194e-01 -1.54233292e-01 5.69719374e-01 -1.39705002e-01
-5.73979676e-01 8.85039091e-01 9.29429054e-01 -4.37061697e-01
-1.53449342e-01 -3.50153893e-01 -4.64760453e-01 -6.99004769e-01
-3.51393640e-01 -2.04390064e-01 6.00041598e-02 -3.26714575e-01
1.01737648e-01 1.86684400e-01 -1.65333164e+00 -4.94276375e-01
6.10854685e-01 2.18634978e-01 1.26282752e-01 6.73412442e-01
-5.26547339e-03 2.06022263e-01 -1.03898740e+00 -1.59869432e-01
-4.43270653e-01 -2.85057902e-01 -2.48627976e-01 1.00952432e-01
2.49975286e-02 -1.38719171e-01 -5.33505440e-01 5.97861946e-01
6.54748440e-01 3.17286476e-02 8.27829480e-01 -8.67869258e-01
-6.19914651e-01 8.02264929e-01 -4.73826587e-01 3.83375138e-01
2.75875837e-01 4.99424666e-01 6.22834325e-01 -5.02384424e-01
5.16789675e-01 1.60134554e+00 8.42052579e-01 7.29431570e-01
-1.02473056e+00 -5.06425440e-01 1.84070632e-01 -3.82980816e-02
-1.29218721e+00 -5.20341754e-01 7.75378108e-01 -5.88728368e-01
6.71009660e-01 3.37120324e-01 1.06122994e+00 1.09222221e+00
9.01609957e-01 1.25850245e-01 1.02305329e+00 -5.14806688e-01
1.31013200e-01 -6.91310391e-02 1.62607670e-01 3.20574135e-01
1.40406653e-01 3.22251588e-01 -6.64502442e-01 -3.76050264e-01
3.79955143e-01 -3.78339142e-01 -2.98829824e-01 5.99378109e-01
-8.49513888e-01 1.98452517e-01 6.46554027e-03 6.84745133e-01
-3.52988511e-01 3.17637771e-01 4.92104262e-01 5.05107105e-01
7.01050282e-01 1.51149631e-01 -1.01558149e+00 -6.78191781e-01
-5.96144557e-01 2.64616609e-01 7.25295007e-01 3.12192947e-01
2.45561436e-01 3.31731021e-01 5.45382760e-02 1.24135220e+00
3.76712948e-01 8.65956604e-01 7.15193748e-01 -9.93198276e-01
-3.79220881e-02 2.09356025e-02 -2.78195173e-01 -3.19262534e-01
-5.27784467e-01 -4.46412802e-01 -2.05310851e-01 1.26946136e-01
8.40878427e-01 -4.34455037e-01 -8.27886224e-01 2.20786572e+00
1.59556530e-02 -2.63362765e-01 -2.47426495e-01 3.50365222e-01
2.40647927e-01 4.46166217e-01 5.99129915e-01 -1.07200873e+00
1.64047861e+00 -3.73036504e-01 -1.06066096e+00 -2.79175580e-01
1.62963182e-01 -7.14229345e-01 1.22374475e+00 2.80824721e-01
-8.75402570e-01 -5.51012456e-01 -6.80278003e-01 3.68198365e-01
-2.00892404e-01 -1.01375811e-01 3.47326964e-01 1.16958642e+00
-6.89320207e-01 9.02200937e-01 -8.88882875e-01 -6.99498773e-01
-3.02160561e-01 1.51440948e-01 -3.94967385e-02 5.04369795e-01
-1.33790994e+00 7.42970765e-01 -3.14270228e-01 2.00087845e-01
-4.14015263e-01 -9.18524742e-01 -5.61751902e-01 -3.40962499e-01
2.87200548e-02 -6.93786919e-01 1.45830488e+00 -5.39371014e-01
-1.65382719e+00 5.91467857e-01 -3.02670002e-01 1.48843937e-02
3.06692511e-01 -3.83583456e-01 -1.11435342e+00 -3.90161783e-01
4.65443218e-03 -4.88907332e-03 1.08839786e+00 -9.46314573e-01
-1.36235967e-01 -7.32166111e-01 -6.45387828e-01 3.92603278e-02
-4.25015777e-01 3.14829677e-01 -1.68432534e-01 -7.88995385e-01
-1.54903293e-01 -1.08525491e+00 -2.52552330e-01 -3.65017205e-01
-1.51781887e-01 -4.73521113e-01 4.83303010e-01 -1.23349047e+00
1.83538151e+00 -2.34804988e+00 -1.42544195e-01 -1.85558692e-01
5.19238831e-03 -3.74536425e-01 1.73058003e-01 4.77737039e-01
-5.98294064e-02 2.55224347e-01 -3.23724419e-01 -2.75808722e-01
-8.72607827e-02 4.10240859e-01 6.22647591e-02 6.41925156e-01
4.50230129e-02 2.70906508e-01 -6.34368598e-01 -6.29046112e-02
-1.25555113e-01 1.82579592e-01 -2.77652204e-01 -2.22868413e-01
9.67109874e-02 3.04925889e-01 -2.74369475e-02 8.07172537e-01
3.16546559e-01 7.60186851e-01 -4.46705818e-02 1.25665158e-01
-7.91648567e-01 2.91461915e-01 -4.39669579e-01 1.20836627e+00
-4.97993022e-01 5.43784142e-01 8.41316655e-02 -2.35493004e-01
7.90809512e-01 5.02116501e-01 5.02742589e-01 -5.86257696e-01
2.04220921e-01 5.39056242e-01 8.34556758e-01 -6.12720907e-01
4.04885232e-01 -4.08610910e-01 -1.70748949e-01 1.51805982e-01
-2.85167396e-01 -2.09650084e-01 7.00793415e-02 -2.55401999e-01
1.02891850e+00 -1.93820298e-01 2.03120187e-01 -5.98874211e-01
2.15333521e-01 -5.28886139e-01 6.82062864e-01 2.48381570e-01
-6.24205768e-01 1.29152402e-01 6.14590347e-01 2.71856368e-01
-1.03067529e+00 -1.15983951e+00 -4.80277538e-01 1.23706293e+00
-5.30669987e-01 -4.58069623e-01 -5.44927180e-01 3.56251597e-02
3.11676234e-01 9.89864469e-01 -6.66496694e-01 -7.00882912e-01
-7.00218201e-01 -9.06709254e-01 6.63161039e-01 4.23637629e-01
-1.43534303e-01 -1.22396088e+00 -3.02398831e-01 4.52409297e-01
1.58516884e-01 -6.35158598e-01 -9.72172976e-01 2.54121035e-01
-1.04410183e+00 -4.60393161e-01 -8.16999674e-01 -2.99673021e-01
1.16684131e-01 -2.48312041e-01 9.62366760e-01 -6.70829341e-02
-4.62772042e-01 5.84888637e-01 1.37529835e-01 -9.95762944e-01
-7.46750951e-01 1.23101138e-01 6.14781082e-01 -5.85628390e-01
-8.69554095e-03 -7.56679237e-01 -5.82502365e-01 8.90579298e-02
-5.16423941e-01 -1.02333748e+00 3.53477627e-01 4.88970131e-01
-2.01211795e-02 7.27207586e-02 1.15167081e+00 -4.61608976e-01
8.66987646e-01 -6.87999427e-01 2.17961967e-01 -2.58824080e-01
-9.30084407e-01 -8.35458413e-02 -5.88343441e-02 -8.48989844e-01
-1.26739621e+00 -2.41334185e-01 -2.25437656e-01 7.04611279e-03
-4.12175506e-02 3.45223010e-01 -2.85214752e-01 7.50447869e-01
5.45005381e-01 -1.75504126e-02 5.17860830e-01 -6.44317806e-01
2.01098144e-01 9.16186094e-01 7.98981190e-01 -6.56287849e-01
8.08720767e-01 1.38107643e-01 -3.85393023e-01 -1.45807898e+00
-2.48859435e-01 -3.24548572e-01 -5.09959221e-01 -7.51929283e-01
9.53379631e-01 -8.79737437e-01 -3.56858194e-01 9.06899273e-01
-7.38345683e-01 -6.55066729e-01 -1.00668937e-01 5.51401377e-01
-2.26921424e-01 2.83497851e-02 -6.85474455e-01 -1.29126191e+00
-2.70864874e-01 -5.44804871e-01 8.94444704e-01 3.59483719e-01
-8.93863559e-01 -1.06300616e+00 2.55780458e-01 2.35919505e-01
2.69358367e-01 -7.39679858e-02 1.19515359e+00 -1.97779611e-01
3.45156878e-01 1.81494832e-01 5.63896000e-01 5.20769954e-01
7.34946191e-01 7.47021139e-01 -8.99827540e-01 -8.48072618e-02
2.83569992e-01 4.18081194e-01 4.23777312e-01 8.95188749e-01
7.81354547e-01 -1.14629671e-01 -2.50928074e-01 -1.06582969e-01
8.06633949e-01 6.21148109e-01 5.98412454e-01 1.14265956e-01
4.04415369e-01 8.20577979e-01 4.95347440e-01 3.42133701e-01
-1.43425316e-02 4.54007179e-01 -2.80238427e-02 1.75716896e-02
-4.05965894e-01 -5.37909418e-02 8.66968393e-01 1.57260692e+00
-5.09920239e-01 1.77482843e-01 -5.87984622e-01 8.06149900e-01
-1.11334777e+00 -8.51858079e-01 -3.64514709e-01 2.37652397e+00
1.03912127e+00 4.01989847e-01 4.35004801e-01 3.41338813e-01
7.48627424e-01 3.31302136e-01 -6.64125800e-01 -6.34981394e-01
-5.15535881e-04 1.90398753e-01 7.02572167e-01 4.59595501e-01
-3.27853262e-01 5.55040240e-01 8.13607121e+00 4.41213310e-01
-1.31566131e+00 -5.29331751e-02 4.07534689e-01 -9.78032351e-02
-4.00543988e-01 -3.59415084e-01 -5.33815026e-01 5.85116923e-01
1.60679317e+00 -5.30915797e-01 2.83620775e-01 5.02586663e-01
9.21562612e-01 -1.77002877e-01 -5.24695754e-01 2.27610424e-01
-2.48683587e-01 -3.03407937e-01 -4.87598866e-01 5.69916427e-01
2.82287747e-01 -3.33643973e-01 -3.01327072e-02 6.05822444e-01
-2.45659068e-01 -6.08595014e-01 1.04286170e+00 7.04440296e-01
9.83384728e-01 -6.42807901e-01 1.69636562e-01 1.70095056e-01
-1.18019140e+00 -3.87167871e-01 1.07192136e-01 -4.61014569e-01
3.50994170e-01 8.42619896e-01 -7.35613585e-01 1.00811593e-01
5.54575622e-01 1.75619766e-01 -4.55305845e-01 6.56596303e-01
-5.02910875e-02 1.28633654e+00 -2.07806975e-01 -1.28734773e-02
-4.08115774e-01 -1.12004332e-01 1.02150822e+00 8.55776072e-01
5.16938269e-01 -1.76859960e-01 -6.90418959e-01 4.50808108e-01
1.80510357e-01 7.75587857e-02 -4.93853629e-01 -3.52728903e-01
9.15019393e-01 8.60151172e-01 -4.60017174e-01 1.05100498e-01
-5.97034574e-01 5.54927588e-01 -4.72648829e-01 3.26219290e-01
-8.47815096e-01 -2.08947256e-01 1.11274672e+00 6.46473527e-01
-2.59121001e-01 -4.20842111e-01 -3.13570678e-01 -3.00899953e-01
3.27113830e-02 -8.73396277e-01 1.08101562e-01 -5.06378233e-01
-9.53935862e-01 1.20392039e-01 4.00220640e-02 -7.26867974e-01
-3.11486572e-01 -3.39706928e-01 -9.83630240e-01 9.63074148e-01
-6.26581669e-01 -7.20414162e-01 2.98481464e-01 1.85043827e-01
7.55568862e-01 -1.55915633e-01 5.58099508e-01 5.17925918e-01
-8.37580979e-01 6.18382156e-01 -5.92402853e-02 -6.09268248e-01
9.70520914e-01 -1.29022348e+00 5.39284229e-01 5.99967837e-01
-3.80198956e-01 9.82971489e-01 1.15959990e+00 -9.70633268e-01
-1.06214142e+00 -6.13136172e-01 9.52401876e-01 -4.15532231e-01
1.08937645e+00 -1.23472400e-01 -9.40972865e-01 3.96110445e-01
2.49713898e-01 -8.89152229e-01 6.47716045e-01 3.98804545e-01
7.23250397e-03 -1.30142927e-01 -7.28990078e-01 9.04939115e-01
1.23763633e+00 -7.49834776e-01 -6.39181793e-01 -2.02529430e-01
1.15367699e+00 1.22537822e-01 -1.06754553e+00 3.68134141e-01
1.09771776e+00 -4.00790185e-01 7.25952148e-01 -4.05056596e-01
1.83122635e-01 2.65910616e-03 1.09590650e-01 -1.39044976e+00
-2.79033750e-01 -8.99770498e-01 2.47693911e-01 1.69770455e+00
5.62051713e-01 -8.38063240e-01 1.65478662e-01 5.52800715e-01
-5.56485772e-01 -5.80542266e-01 -1.28017497e+00 -8.97447824e-01
3.14563274e-01 -3.54641408e-01 4.58461076e-01 5.08570135e-01
2.17829049e-02 3.56943339e-01 -5.52242219e-01 -7.54789338e-02
9.20013860e-02 -8.84684861e-01 4.12303925e-01 -1.32880223e+00
-1.26537010e-01 -5.67356408e-01 -4.03005518e-02 -2.89645165e-01
5.42718954e-02 -1.74278706e-01 2.55947024e-01 -9.14521754e-01
-5.77849895e-03 -3.94686237e-02 -2.48152420e-01 1.64738186e-02
-4.10036713e-01 -4.68413234e-01 -1.67055596e-02 -2.83927959e-03
7.17771828e-01 6.49886429e-01 1.47573566e+00 1.02535978e-01
-6.12288952e-01 5.07465720e-01 -5.77594876e-01 9.28505898e-01
6.48082256e-01 -5.37835121e-01 -4.56099004e-01 2.29963258e-01
-2.26686329e-01 2.99700946e-01 -1.03610188e-01 -1.04164219e+00
-4.46222067e-01 -2.85605013e-01 8.35527387e-03 -2.52439976e-01
2.92252362e-01 -3.70392412e-01 4.01763737e-01 9.78920400e-01
-6.09082580e-02 2.02610239e-01 3.80533159e-01 6.03733182e-01
1.25605941e-01 -2.33321637e-01 5.71722567e-01 3.17806005e-01
-1.33677587e-01 7.78123438e-02 -1.25802565e+00 -3.73602122e-01
5.45169711e-01 -1.28606156e-01 -2.39182636e-01 -2.18561366e-01
-1.13314986e+00 -2.20069498e-01 3.00569743e-01 7.82920420e-01
-3.47080857e-01 -1.32353079e+00 -4.64539349e-01 -3.33500981e-01
-3.16141933e-01 -8.88269067e-01 3.96227062e-01 9.07129169e-01
-1.57604516e-01 1.04573369e-01 -4.66014557e-02 -1.07521787e-01
-1.47139227e+00 2.87786484e-01 4.35195625e-01 1.43464953e-01
-5.33329904e-01 6.84162617e-01 -1.42370127e-02 -2.02117264e-01
1.75350025e-01 -7.68370211e-01 -2.11435616e-01 5.59147954e-01
2.81265557e-01 9.14590478e-01 -2.00998604e-01 -4.88478720e-01
-4.48692381e-01 6.32339060e-01 3.25787365e-01 -3.08892906e-01
9.86196160e-01 -4.70949233e-01 -3.15694332e-01 1.43596721e+00
9.34388995e-01 8.38358164e-01 -9.08653080e-01 2.23346472e-01
1.93822458e-01 1.00275062e-01 -3.37954879e-01 -4.71386790e-01
-9.46967900e-01 1.81960508e-01 1.03468823e+00 3.15743923e-01
9.92522597e-01 3.17435324e-01 9.25860286e-01 -1.62595138e-01
1.78966988e-02 -1.22062933e+00 -3.36141661e-02 1.49931028e-01
1.31835473e+00 -6.20132744e-01 -2.01824624e-02 -6.03540540e-01
-2.47041360e-01 8.46047997e-01 4.93090749e-01 3.67173672e-01
1.15647876e+00 3.85677993e-01 3.74215394e-01 1.11339644e-01
-9.00285661e-01 -9.52312201e-02 1.22224174e-01 5.87720871e-01
9.21741307e-01 6.08476520e-01 -1.37388515e+00 7.34370708e-01
-8.15564990e-01 -3.83094549e-01 4.54979002e-01 4.07491535e-01
-4.33682710e-01 -1.37564600e+00 -5.26237428e-01 6.02200031e-01
-7.34268844e-01 8.37919861e-02 -1.68475226e-01 1.07393587e+00
3.75287682e-01 8.13872278e-01 3.88573080e-01 -3.65831286e-01
6.49097860e-01 7.06280887e-01 2.01693118e-01 -5.91683805e-01
-4.67020780e-01 4.31012779e-01 6.48435354e-01 -8.25410560e-02
-2.60914177e-01 -1.53677583e+00 -1.37012124e+00 -1.91839188e-01
-2.10183412e-01 -1.95348039e-01 7.49357164e-01 1.02843285e+00
-2.47250587e-01 9.65507686e-01 4.93918002e-01 -4.18508857e-01
-6.06281400e-01 -1.59295738e+00 -1.05623198e+00 8.85399207e-02
3.32402438e-01 -6.19202316e-01 -8.44091296e-01 6.64392626e-03]
|
[14.304320335388184, 6.170373439788818]
|
c59a7241-9b7a-4a26-b896-bb4a822c539a
|
a-bi-lstm-autoencoder-framework-for-anomaly
|
2303.09703
| null |
https://arxiv.org/abs/2303.09703v1
|
https://arxiv.org/pdf/2303.09703v1.pdf
|
A Bi-LSTM Autoencoder Framework for Anomaly Detection -- A Case Study of a Wind Power Dataset
|
Anomalies refer to data points or events that deviate from normal and homogeneous events, which can include fraudulent activities, network infiltrations, equipment malfunctions, process changes, or other significant but infrequent events. Prompt detection of such events can prevent potential losses in terms of finances, information, and human resources. With the advancement of computational capabilities and the availability of large datasets, anomaly detection has become a major area of research. Among these, anomaly detection in time series has gained more attention recently due to the added complexity imposed by the time dimension. This study presents a novel framework for time series anomaly detection using a combination of Bidirectional Long Short Term Memory (Bi-LSTM) architecture and Autoencoder. The Bi-LSTM network, which comprises two unidirectional LSTM networks, can analyze the time series data from both directions and thus effectively discover the long-term dependencies hidden in the sequential data. Meanwhile, the Autoencoder mechanism helps to establish the optimal threshold beyond which an event can be classified as an anomaly. To demonstrate the effectiveness of the proposed framework, it is applied to a real-world multivariate time series dataset collected from a wind farm. The Bi-LSTM Autoencoder model achieved a classification accuracy of 96.79% and outperformed more commonly used LSTM Autoencoder models.
|
['Imtiaz Ahmed', 'Ahmed Shoyeb Raihan']
|
2023-03-17
| null | null | null | null |
['time-series-anomaly-detection']
|
['time-series']
|
[-1.87151115e-02 -5.63451707e-01 3.51211905e-01 -1.85149163e-02
2.24300042e-01 -2.12826863e-01 3.47961068e-01 6.29727185e-01
-2.94910818e-01 4.21223223e-01 -1.74767613e-01 -4.60689515e-01
-2.59695411e-01 -9.81619239e-01 -4.21902567e-01 -8.03372204e-01
-5.19541025e-01 -5.39552271e-02 -3.51973460e-03 -1.77606180e-01
2.11252987e-01 7.66351879e-01 -1.53865004e+00 1.45051762e-01
7.46896029e-01 1.36529601e+00 -1.44774497e-01 3.64356279e-01
-4.20130342e-01 8.54197979e-01 -9.33658361e-01 1.35055065e-01
-2.04507597e-02 -1.76269531e-01 -3.42295766e-01 -1.14761382e-01
-3.71269494e-01 -4.65868711e-01 -3.49970132e-01 8.71295869e-01
2.81763881e-01 4.29064065e-01 1.78016901e-01 -1.29704881e+00
-5.30253232e-01 3.69590163e-01 -5.41257918e-01 9.91931856e-01
-1.09649152e-01 -1.05906181e-01 5.83821118e-01 -6.64205611e-01
-6.24066703e-02 7.86258161e-01 7.04682767e-01 -4.27100360e-02
-9.64075506e-01 -6.36426568e-01 3.57295394e-01 5.67745209e-01
-9.85457182e-01 8.70603174e-02 9.50789690e-01 -5.28251469e-01
1.38766360e+00 9.65668187e-02 7.54075289e-01 1.04001939e+00
7.76190221e-01 4.01769727e-01 5.50249815e-01 -2.84141928e-01
3.14822555e-01 -3.86062443e-01 2.31168941e-01 1.57146350e-01
2.56021768e-01 1.88596979e-01 -3.08009923e-01 -3.25050265e-01
5.52493870e-01 8.10813308e-01 -4.75031771e-02 2.81922579e-01
-1.01187766e+00 5.66458941e-01 3.16701412e-01 8.74145031e-01
-9.73302364e-01 -1.82606161e-01 8.42742205e-01 7.55471528e-01
7.48350918e-01 2.54240483e-01 -7.23240972e-01 -2.82418847e-01
-5.17672181e-01 -3.19269985e-01 5.07768989e-01 2.96624631e-01
2.22131953e-01 1.00724983e+00 2.49402389e-01 6.11354351e-01
-5.03543019e-02 4.75905329e-01 1.13917077e+00 -1.89234078e-01
3.02909315e-01 8.11291575e-01 -1.34820238e-01 -1.49534023e+00
-3.53307843e-01 -6.20871067e-01 -1.15416515e+00 -2.59596705e-02
4.69919853e-02 -2.04490453e-01 -7.22276688e-01 1.36746597e+00
1.73286498e-01 5.77062070e-01 2.06723705e-01 5.93921840e-01
2.80249596e-01 1.03902733e+00 1.25526875e-01 -3.66663724e-01
1.01572847e+00 -3.50218743e-01 -1.12680054e+00 -1.39821783e-01
6.59892440e-01 -6.10437453e-01 6.16438508e-01 2.82617837e-01
-5.53540528e-01 -4.17858481e-01 -1.02463615e+00 5.92228889e-01
-8.86512995e-01 3.07387090e-03 4.02403980e-01 -4.69778627e-02
-5.60152531e-01 7.64022768e-01 -1.25861132e+00 -5.13210535e-01
-2.77816541e-02 2.90114611e-01 -3.12462121e-01 4.13081080e-01
-1.34291077e+00 6.90931737e-01 8.37077916e-01 7.58501172e-01
-4.23953444e-01 -4.18266863e-01 -7.90706277e-01 2.26778179e-01
1.68618649e-01 -3.37234959e-02 9.87420261e-01 -1.10751939e+00
-1.04650271e+00 1.36605397e-01 -7.32774213e-02 -7.97578096e-01
-5.25931902e-02 -3.72703671e-01 -1.33056521e+00 -1.01866469e-01
-1.67825386e-01 -4.07871306e-01 8.92789423e-01 -2.81036705e-01
-7.39804983e-01 -6.44481957e-01 -4.82435495e-01 -1.81522176e-01
-8.96251321e-01 -6.06826972e-04 5.78765869e-01 -9.78483617e-01
2.73198724e-01 -7.37377346e-01 -4.38667648e-02 -6.00799203e-01
-1.45228788e-01 -2.40513325e-01 1.59717774e+00 -9.47198451e-01
1.63626611e+00 -2.35931015e+00 -2.60426909e-01 3.66028100e-01
-2.00448796e-01 4.47461903e-01 1.41003147e-01 7.36286998e-01
-5.48139870e-01 -1.38594046e-01 -1.92460090e-01 3.68984550e-01
-4.69315678e-01 5.57618916e-01 -7.89679229e-01 4.36125606e-01
2.79741883e-01 3.36294889e-01 -7.99400568e-01 8.75703916e-02
5.25638103e-01 3.14574003e-01 5.67849725e-02 3.19713980e-01
1.10362642e-01 3.47818464e-01 -5.74314594e-01 5.30050695e-01
2.85211146e-01 -1.45048514e-01 -2.38288417e-01 -9.31592949e-04
-3.82658184e-01 -9.50965583e-02 -8.92627776e-01 1.03422225e+00
-5.14127493e-01 8.26096654e-01 -4.19476658e-01 -1.37794507e+00
1.06959939e+00 7.53934562e-01 6.96267486e-01 -7.97283590e-01
8.09010640e-02 3.76155674e-01 1.46179572e-01 -6.85218632e-01
2.14374498e-01 3.71027626e-02 1.41705588e-01 4.77310181e-01
-1.02605477e-01 7.04563618e-01 3.87067162e-02 -3.92868668e-01
1.13504970e+00 -3.50165159e-01 4.20482039e-01 5.29753044e-02
6.74388289e-01 -1.30624294e-01 6.71275437e-01 2.22790867e-01
-1.16010001e-02 -2.41011977e-01 1.52656928e-01 -1.11135435e+00
-9.28693175e-01 -7.51010418e-01 1.48752509e-02 9.39065278e-01
-2.77971685e-01 1.24244820e-02 -8.01323280e-02 -4.74488705e-01
1.44498840e-01 8.42755079e-01 -5.83217084e-01 -5.59253514e-01
-7.29734898e-01 -8.37516487e-01 5.58737695e-01 7.39155412e-01
5.20805180e-01 -1.48678350e+00 -8.24012637e-01 6.02768838e-01
-7.01583624e-02 -8.82326305e-01 -4.77951244e-02 4.09949094e-01
-1.34059155e+00 -9.21905100e-01 -3.14795494e-01 -5.88039160e-01
4.86108810e-01 2.12236680e-02 7.20270097e-01 8.03099349e-02
-2.02749029e-01 6.53222352e-02 -3.56730014e-01 -6.64656699e-01
-2.13672310e-01 -1.24223396e-01 2.97375053e-01 2.49266237e-01
8.15419853e-01 -8.61852467e-01 -5.12538433e-01 1.84357971e-01
-1.12455678e+00 -6.43462002e-01 4.11858469e-01 8.68973911e-01
5.34591615e-01 6.91376626e-01 9.72632229e-01 -2.25744158e-01
7.50753045e-01 -1.01464915e+00 -6.64805114e-01 -7.45563582e-02
-7.10388660e-01 -1.11787446e-01 1.09579301e+00 -4.79695916e-01
-9.57872629e-01 -4.69259232e-01 -1.65232718e-02 -4.95606095e-01
-3.80290985e-01 9.59672809e-01 2.54297435e-01 2.44722083e-01
2.33480334e-01 5.38845301e-01 -1.00664973e-01 -6.03459656e-01
-4.30966407e-01 7.50144124e-01 4.84878063e-01 -4.82579097e-02
4.75536972e-01 4.54774588e-01 -1.91755787e-01 -1.20485759e+00
-5.27108848e-01 -4.89404827e-01 -3.89867991e-01 -2.51812220e-01
5.78873217e-01 -7.17167974e-01 -6.04112089e-01 8.26132894e-01
-1.09284317e+00 1.61997646e-01 -2.49026775e-01 7.27880776e-01
1.35116115e-01 2.54519790e-01 -7.92354465e-01 -9.09610748e-01
-4.82861370e-01 -5.08677423e-01 4.67766941e-01 2.08026931e-01
-2.62203723e-01 -1.36363471e+00 -4.11567688e-02 -3.74853224e-01
6.39388144e-01 5.50514817e-01 9.88277555e-01 -1.17161810e+00
-5.99725582e-02 -8.62704873e-01 2.36360416e-01 5.41641295e-01
5.19088566e-01 1.59920871e-01 -6.92138076e-01 -4.70815420e-01
4.25666273e-01 3.40316474e-01 4.05869067e-01 3.18038076e-01
1.38897419e+00 -5.84308267e-01 -1.88092723e-01 4.38255370e-01
1.17275262e+00 9.50802088e-01 4.63100940e-01 6.47939861e-01
5.73256254e-01 3.70495051e-01 3.87952566e-01 7.74478018e-01
-2.97471043e-02 2.23646194e-01 5.68939328e-01 -6.20473288e-02
7.41837204e-01 -7.34073371e-02 4.48425174e-01 1.35071635e+00
-1.88895538e-01 -2.96098173e-01 -8.94084513e-01 7.31498778e-01
-1.73315406e+00 -1.37188196e+00 -1.67464226e-01 2.17023158e+00
1.41330183e-01 1.70164421e-01 -1.73869759e-01 7.42561281e-01
8.06459963e-01 2.68234134e-01 -7.96768486e-01 -7.82515824e-01
-1.93669215e-01 2.86174845e-02 2.42846727e-01 -1.63842931e-01
-1.13762569e+00 3.90933424e-01 5.28453732e+00 3.89127910e-01
-1.59534109e+00 -1.62696958e-01 4.10366625e-01 3.08881477e-02
1.46907747e-01 -4.51515794e-01 -2.34353065e-01 7.82298625e-01
1.34527802e+00 -3.18415940e-01 1.34869948e-01 7.66639471e-01
4.88326728e-01 2.80357987e-01 -8.36715400e-01 7.33306408e-01
-1.69567376e-01 -7.26091146e-01 7.29682371e-02 4.07898352e-02
4.49003875e-01 8.15778151e-02 2.31697932e-02 2.87312031e-01
-1.36028498e-01 -7.76633739e-01 1.78061023e-01 4.57893103e-01
1.43846765e-01 -8.73529017e-01 1.18975341e+00 3.45172644e-01
-1.32562041e+00 -6.28802478e-01 -1.08214997e-01 -4.23162043e-01
1.18403912e-01 9.09694195e-01 -5.10528624e-01 5.36385775e-01
1.06224442e+00 9.06445086e-01 -1.03157997e-01 9.18123245e-01
7.39137828e-02 9.18695390e-01 -5.41744113e-01 2.31346563e-01
4.03733164e-01 -2.35491738e-01 7.54448235e-01 8.59337270e-01
7.64132798e-01 8.44189003e-02 4.83340397e-02 3.46367329e-01
3.71966779e-01 5.91972247e-02 -8.61255944e-01 -4.74958390e-01
4.57295358e-01 8.94273579e-01 -6.03897095e-01 -1.63243428e-01
-4.34629112e-01 8.73347521e-01 -1.04744419e-01 5.96887946e-01
-6.29396617e-01 -5.93178988e-01 6.47869110e-01 -2.42995992e-01
2.67018974e-01 -1.45176202e-01 -7.39139989e-02 -1.06681514e+00
3.84505868e-01 -6.37200654e-01 8.29336524e-01 -4.07653868e-01
-1.45596826e+00 8.26461494e-01 -2.77495652e-01 -1.49014354e+00
-6.77668154e-01 -4.10224974e-01 -9.98839259e-01 8.15476954e-01
-1.33261454e+00 -5.36617756e-01 -3.40425164e-01 6.63442194e-01
5.09939909e-01 -4.02566135e-01 1.00990045e+00 4.29066598e-01
-9.82219279e-01 1.37062043e-01 4.03023481e-01 3.31312209e-01
2.60232180e-01 -1.03790438e+00 3.85894150e-01 1.21047711e+00
-1.46093383e-01 4.91910070e-01 6.29263878e-01 -6.97538733e-01
-1.00313270e+00 -1.20081055e+00 9.03585076e-01 2.40234077e-01
9.43155050e-01 1.83935255e-01 -1.43748224e+00 8.60783339e-01
-7.09676296e-02 2.62222528e-01 7.80177236e-01 -1.25326440e-01
1.01223588e-01 -4.34186965e-01 -8.69734406e-01 3.65560621e-01
3.88078451e-01 -5.75645030e-01 -7.41548359e-01 -4.07680385e-02
4.52562094e-01 -1.87884957e-01 -1.03139734e+00 5.48713148e-01
4.28180546e-01 -7.93701887e-01 7.11967647e-01 -6.81355536e-01
2.56530821e-01 -2.62315005e-01 -6.98131770e-02 -1.56168032e+00
-2.77443200e-01 -2.45284751e-01 -6.41277254e-01 1.01552224e+00
2.27784812e-02 -1.08502221e+00 3.16237003e-01 2.60922074e-01
-2.66660899e-01 -7.56563008e-01 -9.33286250e-01 -9.25977588e-01
-4.96033698e-01 -4.31360066e-01 9.10309911e-01 1.11214113e+00
1.48778642e-02 -1.17584586e-01 -3.07190627e-01 4.89079118e-01
3.51632297e-01 2.15747938e-01 1.85208112e-01 -1.42829001e+00
1.43594354e-01 -3.66407543e-01 -7.71686971e-01 -4.41135943e-01
1.86769009e-01 -4.79036301e-01 -2.73890257e-01 -1.14078403e+00
-5.94774961e-01 -1.78574614e-04 -9.85015392e-01 6.66276276e-01
1.77509822e-02 -1.83249861e-01 -4.39819485e-01 2.80679286e-01
5.25163487e-02 8.05988193e-01 5.40776014e-01 -3.38319615e-02
-4.63759422e-01 2.61341453e-01 -7.80815408e-02 9.76963639e-01
1.09817171e+00 -2.90969968e-01 -2.89335370e-01 -6.17005229e-01
1.22330189e-01 2.81906188e-01 2.97485590e-01 -1.19502604e+00
2.95440108e-01 -1.39563382e-01 6.35954678e-01 -7.29112387e-01
-4.20364039e-03 -1.38212729e+00 2.44879335e-01 7.28290260e-01
-1.12233676e-01 8.54430139e-01 6.15221381e-01 8.21913123e-01
-7.97407866e-01 2.41984174e-01 1.81454942e-01 1.86804220e-01
-1.01251125e+00 3.32316756e-01 -6.69505358e-01 -3.92356336e-01
1.26254964e+00 -3.19405764e-01 -1.95423350e-01 -1.95085734e-01
-7.21581817e-01 2.15801984e-01 -1.67815879e-01 7.35245943e-01
7.32650340e-01 -1.34965324e+00 -4.54067767e-01 6.32488072e-01
1.04258895e-01 -1.90746576e-01 5.08716047e-01 8.17395210e-01
-4.36500698e-01 5.05133033e-01 -3.63020211e-01 -5.03756404e-01
-1.10853243e+00 4.51451421e-01 4.04009461e-01 -1.16880700e-01
-1.02388573e+00 2.49069273e-01 -2.04262599e-01 1.56728566e-01
1.52492329e-01 -4.50109124e-01 -4.72944677e-01 1.93830952e-01
5.97276568e-01 6.70365155e-01 4.45114255e-01 -5.42339385e-01
-4.36130315e-01 2.82150209e-01 -5.39190471e-02 4.02012289e-01
1.45977688e+00 -8.20980445e-02 -2.60083348e-01 1.04122770e+00
9.65987921e-01 -4.22887415e-01 -7.52630174e-01 -2.34502897e-01
3.13753098e-01 -1.65427968e-01 2.95576483e-01 -5.06449461e-01
-1.27640033e+00 8.79734874e-01 7.83419788e-01 8.18075836e-01
1.41671097e+00 -5.74946582e-01 1.24962223e+00 5.58488250e-01
2.58682556e-02 -1.11539125e+00 -7.90399611e-02 6.53053939e-01
6.83777988e-01 -9.99586463e-01 -5.04597187e-01 3.62694263e-01
-2.84757763e-01 1.40180123e+00 6.08923614e-01 -1.34948328e-01
9.25791085e-01 2.16104195e-01 1.09390140e-01 -1.87907040e-01
-7.27973759e-01 1.83511540e-01 1.90280035e-01 2.07315668e-01
3.34757507e-01 -4.09272201e-02 -1.78524163e-02 6.03672147e-01
-3.84731800e-03 1.57191344e-02 3.15977007e-01 1.02116954e+00
-2.10663274e-01 -5.19980073e-01 -3.81552637e-01 7.98941135e-01
-8.92092347e-01 1.49909452e-01 7.68756792e-02 5.84620178e-01
-1.37940375e-03 1.08682692e+00 6.67720973e-01 -4.21509951e-01
4.89238143e-01 3.23641539e-01 -3.47087324e-01 -8.90019238e-02
-5.48603594e-01 -2.19618589e-01 -3.68775576e-01 -4.97013330e-01
-2.98624784e-01 -4.54571277e-01 -1.29522562e+00 -4.40593183e-01
-4.05509561e-01 3.61160606e-01 6.29072428e-01 1.14378846e+00
5.28273761e-01 9.30850744e-01 8.51630151e-01 -3.91485870e-01
-4.27136153e-01 -1.09595704e+00 -6.23424351e-01 4.41771299e-01
7.18725920e-01 -4.66277927e-01 -6.69912100e-01 5.76508604e-02]
|
[7.08329963684082, 2.7227303981781006]
|
fe80e8b2-f56a-4d66-b454-fa1d833e839f
|
otre-where-optimal-transport-guided-unpaired
|
2302.03003
| null |
https://arxiv.org/abs/2302.03003v4
|
https://arxiv.org/pdf/2302.03003v4.pdf
|
OTRE: Where Optimal Transport Guided Unpaired Image-to-Image Translation Meets Regularization by Enhancing
|
Non-mydriatic retinal color fundus photography (CFP) is widely available due to the advantage of not requiring pupillary dilation, however, is prone to poor quality due to operators, systemic imperfections, or patient-related causes. Optimal retinal image quality is mandated for accurate medical diagnoses and automated analyses. Herein, we leveraged the Optimal Transport (OT) theory to propose an unpaired image-to-image translation scheme for mapping low-quality retinal CFPs to high-quality counterparts. Furthermore, to improve the flexibility, robustness, and applicability of our image enhancement pipeline in the clinical practice, we generalized a state-of-the-art model-based image reconstruction method, regularization by denoising, by plugging in priors learned by our OT-guided image-to-image translation network. We named it as regularization by enhancing (RE). We validated the integrated framework, OTRE, on three publicly available retinal image datasets by assessing the quality after enhancement and their performance on various downstream tasks, including diabetic retinopathy grading, vessel segmentation, and diabetic lesion segmentation. The experimental results demonstrated the superiority of our proposed framework over some state-of-the-art unsupervised competitors and a state-of-the-art supervised method.
|
['Jacob M. Sobczak', 'Yalin Wang', 'Keshav Nandakumar', 'Zhangsihao Yang', 'Mohammad Farazi', 'Oana M. Dumitrascu', 'Peijie Qiu', 'Wenhui Zhu']
|
2023-02-06
| null | null | null | null |
['diabetic-retinopathy-grading']
|
['medical']
|
[ 4.35768127e-01 5.46409003e-02 -8.86269007e-03 -5.22865832e-01
-7.66473651e-01 -1.92660391e-01 1.79026261e-01 -3.04256827e-01
-4.41185594e-01 6.47639096e-01 1.57952815e-01 -5.59693336e-01
-2.66456693e-01 -4.48446900e-01 -6.57509327e-01 -7.97284663e-01
2.62725651e-01 -1.47771046e-01 2.77745515e-01 2.22414523e-01
3.10972661e-01 3.95684808e-01 -1.44728982e+00 3.34795892e-01
1.65948558e+00 1.06463134e+00 1.07245915e-01 5.83202541e-01
1.01744674e-01 8.47597301e-01 -1.93000752e-02 -4.92657602e-01
7.05663145e-01 -7.11915672e-01 -6.23090565e-01 4.72618878e-01
8.58337164e-01 -5.45906305e-01 -2.47838065e-01 1.25288665e+00
7.04188645e-01 -1.08803473e-01 4.33147132e-01 -6.15049005e-01
-9.79093015e-01 -1.91654470e-02 -7.22527087e-01 3.82846296e-01
-2.80340344e-01 7.46603191e-01 4.71191645e-01 -5.84745526e-01
7.26516724e-01 8.12967479e-01 4.14995790e-01 3.72607827e-01
-1.31638575e+00 -2.96896458e-01 -6.01695441e-02 2.70893760e-02
-8.47752988e-01 -6.75814807e-01 1.84158504e-01 -8.15228879e-01
7.59892285e-01 -8.99920240e-02 7.60464072e-01 3.63182902e-01
2.04401284e-01 5.58507860e-01 1.78912759e+00 -3.61541569e-01
-1.23568758e-01 -1.33303717e-01 -4.95537482e-02 9.60010290e-01
2.69720733e-01 5.20234525e-01 -3.13342176e-02 2.48049706e-01
9.96319175e-01 -2.92680204e-01 -6.19382799e-01 -2.45735615e-01
-9.15144503e-01 4.72985089e-01 6.18322968e-01 -2.35504001e-01
-3.80174428e-01 -7.12467581e-02 1.82574019e-01 1.42219871e-01
5.58089614e-01 5.72026551e-01 -2.54152179e-01 3.60470787e-02
-7.84016311e-01 -8.15373659e-02 2.30426535e-01 5.07897019e-01
4.73837554e-01 -1.99909925e-01 -5.51674485e-01 9.21376050e-01
3.22773576e-01 1.17780037e-01 4.59611624e-01 -1.20650721e+00
2.65228331e-01 8.29323411e-01 2.01674640e-01 -2.35721081e-01
-5.26314199e-01 -7.70433247e-01 -9.26239371e-01 5.86213350e-01
5.63251913e-01 -1.56799406e-01 -1.36648917e+00 1.22319436e+00
3.15766543e-01 4.05858785e-01 -4.21177857e-02 1.45113349e+00
8.54194164e-01 9.15762559e-02 4.15373445e-02 -4.87951666e-01
1.17662358e+00 -1.25225496e+00 -5.58729589e-01 -3.44401598e-03
5.98043382e-01 -9.19473052e-01 1.00534344e+00 4.34902310e-01
-1.24760425e+00 -5.32745957e-01 -7.51896083e-01 -2.35265270e-01
2.30756894e-01 7.02760339e-01 7.19020307e-01 5.74697316e-01
-1.10893762e+00 5.41669786e-01 -1.03153050e+00 -2.77867645e-01
9.31462049e-01 2.34749436e-01 -4.77774948e-01 -2.02541292e-01
-4.68451917e-01 9.94292676e-01 -9.08496790e-03 3.75435859e-01
-5.54370642e-01 -1.00766456e+00 -6.83497727e-01 -3.77320707e-01
1.24271557e-01 -1.36822438e+00 9.53217268e-01 -6.91600263e-01
-1.80333745e+00 1.22470284e+00 -3.66049141e-01 -6.71384633e-01
6.94577813e-01 -2.98533410e-01 -4.11660403e-01 5.02884269e-01
-9.11743566e-02 5.74107707e-01 8.58803868e-01 -8.27847540e-01
-6.82871044e-01 -5.18105030e-01 -1.34760261e-01 1.13835752e-01
1.96418375e-01 2.46857971e-01 -6.16509318e-01 -2.99601376e-01
1.84704885e-01 -7.88110256e-01 -2.72102773e-01 4.60691303e-01
-4.66504216e-01 1.49282873e-01 1.76115170e-01 -8.78286958e-01
1.03781271e+00 -2.03015924e+00 -1.01236977e-01 8.96748621e-03
4.76752847e-01 7.75839329e-01 -3.24350864e-01 -3.01842511e-01
-1.11975610e-01 1.06695734e-01 -3.34057182e-01 -2.08082557e-01
-4.79473025e-01 -1.12751760e-01 1.32920384e-01 7.32491016e-01
5.34014225e-01 9.40229535e-01 -9.35846865e-01 -4.65608269e-01
5.08698106e-01 5.88550627e-01 -5.64395308e-01 1.96382195e-01
1.51145449e-02 8.76308560e-01 -2.51178354e-01 7.67773032e-01
6.60575867e-01 -5.80392063e-01 -6.93992302e-02 -6.53323114e-01
-3.58775288e-01 1.17575616e-01 -7.35209107e-01 1.58401418e+00
-2.92353541e-01 6.48635864e-01 -2.05772594e-01 -4.40384030e-01
7.09010780e-01 3.00881453e-02 4.62816864e-01 -6.92482471e-01
2.77165920e-01 4.43153441e-01 4.13718998e-01 -9.21630502e-01
-9.37017873e-02 7.36737475e-02 1.06396937e+00 3.40246633e-02
-3.08564864e-02 5.16109467e-02 3.18120569e-01 -1.26827896e-01
8.84142578e-01 4.46645349e-01 3.27892870e-01 2.13215381e-01
4.42603886e-01 -2.03887373e-01 5.60256481e-01 4.52672154e-01
-6.15776956e-01 1.07044566e+00 5.97676754e-01 -1.32566869e-01
-9.70252693e-01 -9.08552051e-01 -6.87930405e-01 2.45272651e-01
4.91751581e-02 -9.64981690e-02 -6.38401151e-01 -5.79055905e-01
-7.67472610e-02 1.27933830e-01 -5.13737857e-01 6.98034167e-02
-1.85940728e-01 -1.13850820e+00 3.14245909e-01 2.82448977e-01
8.33506286e-01 -4.88407195e-01 -4.51824248e-01 1.79721773e-01
-1.88526154e-01 -1.17892730e+00 -5.01001358e-01 -6.38840973e-01
-9.54899609e-01 -1.41407549e+00 -9.52287138e-01 -5.95308661e-01
8.70556831e-01 3.20631772e-01 9.07450497e-01 -6.78161392e-03
-7.42181957e-01 1.15596764e-01 -1.63420483e-01 -2.12652504e-01
-2.86672086e-01 -4.51446652e-01 -2.87456125e-01 5.45758963e-01
9.38600451e-02 -4.96013761e-01 -1.36752057e+00 2.76501685e-01
-7.72287786e-01 1.13067918e-01 1.17813063e+00 1.00748193e+00
9.96597826e-01 -3.54282185e-02 2.81498671e-01 -1.02111304e+00
4.50002521e-01 2.00735852e-01 -9.17162597e-01 3.17918330e-01
-8.90350521e-01 -1.70842797e-01 1.00648016e-01 -1.86192051e-01
-1.14741051e+00 4.62163053e-02 1.14180706e-01 -3.87318820e-01
-1.57183573e-01 3.37695092e-01 3.71575616e-02 -7.16980219e-01
8.49150956e-01 1.03534415e-01 5.07772326e-01 -3.66235316e-01
6.23675048e-01 7.34080970e-01 8.19487214e-01 -2.17553258e-01
3.69672477e-01 7.13669181e-01 1.74288541e-01 -4.42659646e-01
-1.19554496e+00 -6.88859105e-01 -4.39786673e-01 -2.24537626e-01
9.89575148e-01 -9.47992265e-01 -7.51141727e-01 9.09565091e-01
-9.69029725e-01 -3.15644145e-01 -1.62183911e-01 7.64874935e-01
-5.04059970e-01 6.44524455e-01 -6.12500787e-01 -5.41386783e-01
-3.77169043e-01 -1.45908320e+00 7.73646653e-01 5.70471525e-01
5.31482220e-01 -7.76560962e-01 -1.20834157e-01 6.87904000e-01
4.92211401e-01 1.42065778e-01 1.01482344e+00 4.28647771e-02
-8.10824692e-01 5.21323793e-02 -8.67496789e-01 9.31656539e-01
1.72288135e-01 3.11265796e-01 -1.00618255e+00 -1.44470558e-01
-3.69751036e-01 -2.37827837e-01 1.19661856e+00 1.03712153e+00
1.15865052e+00 -2.00505704e-02 -6.46557733e-02 1.17467368e+00
1.59610248e+00 -1.96678117e-01 1.26105082e+00 3.24178010e-01
4.14301962e-01 6.51796281e-01 6.12723768e-01 2.53095359e-01
3.58039141e-01 6.05133533e-01 4.50934827e-01 -7.70267308e-01
-8.00201714e-01 -3.49290520e-02 -5.71121611e-02 1.57323509e-01
-4.78579134e-01 3.40970717e-02 -6.10657454e-01 6.35494411e-01
-1.77813649e+00 -5.72870851e-01 -3.88900012e-01 2.18387675e+00
1.02344358e+00 -1.28097415e-01 -1.95438806e-02 -5.46607912e-01
6.20616317e-01 -3.14497024e-01 -7.39789546e-01 1.86086819e-01
-3.59140903e-01 2.44606659e-01 7.15515494e-01 2.35815793e-01
-1.12816823e+00 8.65824044e-01 5.68957138e+00 5.29331744e-01
-1.23398364e+00 1.08038537e-01 9.09133077e-01 -1.20010599e-01
1.85859263e-01 1.45965278e-01 -6.75749421e-01 3.75093967e-01
5.22947848e-01 2.51468122e-01 3.95122945e-01 1.65283173e-01
7.74024487e-01 -2.74720222e-01 -7.77875483e-01 9.23622847e-01
-1.60184652e-01 -1.49360335e+00 -1.98068097e-02 1.48766115e-01
9.74049032e-01 2.62890071e-01 2.87714273e-01 -3.32831144e-01
4.03634384e-02 -1.00867593e+00 -6.66478798e-02 9.81632769e-01
1.09684825e+00 -2.43004620e-01 7.69594967e-01 -2.81879723e-01
-5.19270539e-01 -2.62710657e-02 -4.42211866e-01 3.43213439e-01
2.08371818e-01 7.20037937e-01 -6.19209409e-01 5.39282262e-01
6.40634656e-01 1.09530997e+00 -8.01448703e-01 1.90230966e+00
-3.83257538e-01 5.78995645e-01 1.40367508e-01 7.74864674e-01
9.39470902e-02 -6.91395104e-01 6.75024629e-01 7.43128479e-01
2.07070455e-01 3.05670708e-01 -1.84079871e-01 1.14963937e+00
-1.42550946e-03 1.75004527e-01 -1.61252797e-01 5.69281131e-02
2.35597059e-01 1.49538743e+00 -3.56020093e-01 -4.95531736e-03
-6.78403735e-01 6.39133155e-01 -2.34816093e-02 7.80572832e-01
-4.75675195e-01 -3.35279822e-01 5.78515649e-01 4.45457041e-01
1.26349151e-01 1.91093937e-01 -3.93573403e-01 -1.05442035e+00
2.01330677e-01 -8.10777009e-01 2.22778231e-01 -1.07624125e+00
-1.39886963e+00 5.80826402e-01 -6.54530883e-01 -1.57051849e+00
2.81022817e-01 -6.99623704e-01 -4.64222848e-01 1.19104731e+00
-2.37975717e+00 -1.41583586e+00 -4.93975312e-01 4.93383735e-01
-8.82371441e-02 -3.02773416e-01 3.79804939e-01 3.79984796e-01
-7.67661452e-01 4.99301970e-01 3.43853720e-02 4.32301499e-02
1.18251395e+00 -1.12935841e+00 -3.55188362e-02 1.21568930e+00
-4.12595123e-01 6.72576427e-01 2.26849884e-01 -3.63790244e-01
-1.03877616e+00 -1.44379342e+00 5.36627412e-01 -1.38297915e-01
6.45890474e-01 5.83997428e-01 -7.78063059e-01 5.13319850e-01
5.96718304e-02 5.21256208e-01 5.88201165e-01 -2.42020801e-01
-1.29449606e-01 -3.68659884e-01 -1.08403552e+00 5.60650289e-01
8.81125271e-01 -2.36459136e-01 -3.24045122e-01 4.30795491e-01
6.19833946e-01 -7.14947939e-01 -1.04361475e+00 5.75803876e-01
4.05140758e-01 -1.24536240e+00 8.82963955e-01 -7.56083786e-01
7.02239871e-01 -5.08069336e-01 2.58105904e-01 -1.13006377e+00
-1.44463912e-01 -1.09693849e+00 1.61846250e-01 1.08902562e+00
3.21296364e-01 -1.11029410e+00 4.85238343e-01 6.84899688e-01
-6.01942360e-01 -8.45011294e-01 -7.06832409e-01 -3.93537313e-01
-1.28595218e-01 -1.26985252e-01 2.26433739e-01 4.71112460e-01
-5.92892945e-01 -4.36537899e-02 -2.90602356e-01 5.02418518e-01
6.01481736e-01 2.08213776e-01 7.61647403e-01 -9.90872443e-01
-3.66720259e-01 -5.46221197e-01 -7.14670122e-01 -1.08273005e+00
-1.18178308e-01 -8.98527265e-01 -2.54664868e-01 -1.81758332e+00
6.21488839e-02 -3.63437921e-01 -4.03280675e-01 4.46088344e-01
-3.69336158e-01 5.47122300e-01 -3.07687633e-02 3.63716394e-01
-4.40090567e-01 3.45793873e-01 2.06915545e+00 5.84740483e-05
-6.36184692e-01 1.83797494e-01 -1.01269972e+00 6.70313179e-01
4.11241025e-01 -2.41187494e-02 -3.34026635e-01 -5.33844113e-01
1.14214107e-01 1.54397070e-01 7.98930943e-01 -8.27527046e-01
2.78342158e-01 1.33102655e-01 1.30855352e-01 -2.58178264e-01
5.17271832e-02 -2.58197010e-01 -2.53419966e-01 1.60030827e-01
-1.84343383e-01 -7.67412603e-01 5.71910059e-03 4.59587693e-01
-4.49909449e-01 1.32731870e-01 1.33557999e+00 7.53270984e-02
-4.31879789e-01 5.42290807e-01 1.17612608e-01 2.59606391e-01
7.26113677e-01 -3.43503803e-01 -8.96097600e-01 -2.94616222e-02
-7.32808053e-01 1.04690269e-01 5.17604113e-01 -3.46992016e-02
8.67285013e-01 -6.15732849e-01 -1.14079714e+00 2.92911768e-01
3.18255603e-01 -4.55807969e-02 6.03677809e-01 1.85678315e+00
-6.23389959e-01 3.45510453e-01 -2.24533930e-01 -8.44561994e-01
-1.11403859e+00 9.43423957e-02 1.02208745e+00 6.02131300e-02
-1.00906265e+00 8.05120289e-01 2.32942790e-01 -6.15877360e-02
1.23528294e-01 -6.78191662e-01 -2.29104340e-01 -5.20061076e-01
6.07265830e-01 3.69707942e-01 3.76116991e-01 -2.28865787e-01
1.59191400e-01 9.63683546e-01 -3.06022376e-01 4.17780787e-01
1.15549815e+00 -4.48092192e-01 -3.97754937e-01 -1.88660100e-01
6.31516874e-01 -2.73899108e-01 -1.54226375e+00 -3.06279391e-01
-5.09530783e-01 -8.67033362e-01 6.48452699e-01 -1.32816482e+00
-1.33172333e+00 8.54951918e-01 1.10347188e+00 -3.97305340e-01
1.50676358e+00 -3.40652406e-01 7.60337710e-01 -5.42637408e-02
1.10034369e-01 -6.51838422e-01 -3.43167961e-01 -5.25823794e-02
5.69938004e-01 -1.59173751e+00 4.45234887e-02 -8.61157835e-01
-7.72430360e-01 1.01235712e+00 4.84634608e-01 1.02506511e-01
5.54836929e-01 -2.37631932e-01 5.69351912e-01 1.81040447e-02
-5.48494220e-01 -6.77543521e-01 8.38325799e-01 7.72403657e-01
4.77145970e-01 -2.54480630e-01 -4.04402584e-01 3.84187877e-01
1.75348103e-01 4.97711122e-01 8.01699877e-01 3.07008713e-01
-2.39013940e-01 -1.04263353e+00 1.64582059e-01 7.39524961e-01
-6.04766965e-01 -3.47492576e-01 5.76615520e-02 6.69663966e-01
2.80710012e-01 1.12399435e+00 -1.74925566e-01 2.34700456e-01
2.94177443e-01 -4.59977627e-01 7.12844014e-01 -5.77027559e-01
-4.45487887e-01 5.03473401e-01 4.42816578e-02 -7.37911999e-01
-7.95175493e-01 -3.95086080e-01 -9.62040722e-01 2.71043509e-01
-2.47845396e-01 -4.99910861e-01 4.59425777e-01 7.01135695e-01
7.86849499e-01 5.98060489e-01 5.43836772e-01 -3.50472420e-01
-4.74399775e-01 -9.27392721e-01 -7.63426661e-01 3.24952692e-01
5.16851127e-01 -5.82368910e-01 -2.12578073e-01 3.08919519e-01]
|
[15.769476890563965, -3.95688796043396]
|
fa2aabae-3257-4d9e-9ea3-716bef1724fe
|
mutual-information-alleviates-hallucinations
|
2210.13210
| null |
https://arxiv.org/abs/2210.13210v2
|
https://arxiv.org/pdf/2210.13210v2.pdf
|
Mutual Information Alleviates Hallucinations in Abstractive Summarization
|
Despite significant progress in the quality of language generated from abstractive summarization models, these models still exhibit the tendency to hallucinate, i.e., output content not supported by the source document. A number of works have tried to fix--or at least uncover the source of--the problem with limited success. In this paper, we identify a simple criterion under which models are significantly more likely to assign more probability to hallucinated content during generation: high model uncertainty. This finding offers a potential explanation for hallucinations: models default to favoring text with high marginal probability, i.e., high-frequency occurrences in the training set, when uncertain about a continuation. It also motivates possible routes for real-time intervention during decoding to prevent such hallucinations. We propose a decoding strategy that switches to optimizing for pointwise mutual information of the source and target token--rather than purely the probability of the target token--when the model exhibits uncertainty. Experiments on the XSum dataset show that our method decreases the probability of hallucinated tokens while maintaining the Rouge and BertS scores of top-performing decoding strategies.
|
['Clara Meister', 'Ryan Cotterell', 'Liam van der Poel']
|
2022-10-24
| null | null | null | null |
['abstractive-text-summarization']
|
['natural-language-processing']
|
[ 2.98793644e-01 8.25041831e-01 -3.05273265e-01 -9.55486372e-02
-1.10233092e+00 -4.68813092e-01 6.83198333e-01 5.19739985e-01
-1.15032412e-01 1.07204270e+00 9.59374726e-01 -2.75985509e-01
4.99690436e-02 -5.33950746e-01 -6.07241511e-01 -5.28927922e-01
2.22852901e-02 6.10662580e-01 -2.20569327e-01 -9.78873149e-02
6.93400502e-01 3.73740308e-02 -1.49144673e+00 3.61081988e-01
1.19356072e+00 2.26005390e-01 4.94575858e-01 5.99442124e-01
-1.14479773e-01 1.02642941e+00 -1.19115222e+00 -4.27311003e-01
-2.19353184e-01 -9.67236102e-01 -7.66225040e-01 3.13893527e-01
3.59779634e-02 -2.74468869e-01 -2.50766158e-01 1.13593328e+00
5.83591938e-01 -6.01626188e-02 8.81637692e-01 -1.09947407e+00
-5.80897689e-01 1.15683937e+00 -2.42278859e-01 3.00762981e-01
6.88579559e-01 2.23712653e-01 1.00130439e+00 -7.47332692e-01
7.11659789e-01 1.24252808e+00 4.45832074e-01 5.44620156e-01
-1.24260414e+00 -3.59947383e-01 -1.14583291e-01 4.94784489e-02
-1.31865704e+00 -9.33680832e-01 4.39521819e-01 -3.45198631e-01
1.12290609e+00 4.23028082e-01 6.20375276e-01 1.19408572e+00
5.14966309e-01 8.73371124e-01 8.39287877e-01 -4.93964404e-01
2.97607750e-01 3.79912078e-01 -1.94096133e-01 4.33223933e-01
5.41683674e-01 -3.41395617e-01 -1.01854813e+00 -4.52173650e-01
2.71739155e-01 -7.76733935e-01 -5.66421866e-01 3.03433150e-01
-1.26862073e+00 7.22625315e-01 -2.77603984e-01 3.43395174e-01
-5.82494617e-01 3.90660241e-02 9.39898118e-02 1.07726015e-01
6.41354978e-01 1.00794816e+00 -3.16840887e-01 -6.19218469e-01
-1.37873805e+00 5.71237624e-01 9.91129220e-01 9.84421015e-01
3.13653201e-01 1.43456325e-01 -4.62400705e-01 8.12516987e-01
2.61836082e-01 2.40926921e-01 7.03310907e-01 -9.51700687e-01
5.65402210e-01 2.36013561e-01 2.83555150e-01 -8.14151824e-01
-1.62856296e-01 -6.80865705e-01 -4.71665174e-01 -1.70428351e-01
2.73114890e-01 -2.75204152e-01 -6.48279130e-01 1.85802078e+00
-2.63371944e-01 -1.75882995e-01 4.45447028e-01 5.24172068e-01
3.19764316e-01 9.05616403e-01 -4.99038845e-02 -7.84470737e-01
1.04225063e+00 -6.20648742e-01 -1.09941971e+00 -3.68641794e-01
6.99866951e-01 -9.24835563e-01 9.74981129e-01 4.26962942e-01
-1.56913447e+00 2.40733251e-02 -1.13688135e+00 3.02221715e-01
2.00155601e-01 -1.17084436e-01 3.45633507e-01 5.32090604e-01
-1.00496960e+00 9.11745548e-01 -5.03889740e-01 -3.62499356e-01
1.31672531e-01 -2.02369709e-02 -5.64553291e-02 2.20536858e-01
-1.22174168e+00 1.21280932e+00 5.96314967e-01 -2.08777189e-01
-7.88864672e-01 -4.58036065e-01 -5.71209013e-01 2.57250726e-01
1.74613103e-01 -6.62595510e-01 1.44293749e+00 -8.49238873e-01
-1.20428431e+00 3.99627566e-01 -5.91200590e-01 -5.80031991e-01
6.78240359e-01 -1.61752895e-01 -2.96654969e-01 1.25554159e-01
3.34777832e-01 6.87981725e-01 7.95130193e-01 -1.53558755e+00
-4.18516815e-01 1.73985064e-02 -3.52467895e-01 6.26546860e-01
-2.97398388e-01 -7.98568055e-02 -1.42315045e-01 -6.49362445e-01
3.60965103e-01 -7.22304165e-01 -6.79557491e-03 -5.08568406e-01
-1.15695620e+00 -3.54033530e-01 2.16803148e-01 -6.54998779e-01
1.53750467e+00 -1.77744842e+00 1.64861754e-01 6.54057562e-02
7.52404034e-02 8.27412214e-03 9.78070423e-02 8.41616213e-01
-1.18012223e-02 5.50359309e-01 -1.95991948e-01 -3.96134287e-01
5.75223975e-02 2.47606505e-02 -6.24446332e-01 3.22224528e-01
1.97375447e-01 4.85444039e-01 -1.11980975e+00 -6.22595727e-01
-1.84361339e-01 9.00935009e-02 -5.11049390e-01 1.68135554e-01
-4.15530175e-01 1.78232521e-01 -1.59190282e-01 5.81643581e-01
4.15644348e-01 -2.70599931e-01 8.40552971e-02 2.40569666e-01
-1.30901724e-01 6.86479807e-01 -9.03994858e-01 1.31261086e+00
-1.00104891e-01 8.19310308e-01 -2.22230434e-01 -4.70688343e-01
7.21298754e-01 5.21177828e-01 1.24575533e-01 -2.18607306e-01
-5.99373057e-02 3.62787366e-01 3.37515026e-01 -6.67002559e-01
1.11022270e+00 -3.44366252e-01 5.20698503e-02 5.60545146e-01
-1.05798587e-01 -4.03058648e-01 2.66846895e-01 5.91829181e-01
1.03990304e+00 -2.82856107e-01 3.26794207e-01 -1.41562715e-01
-1.02497377e-01 1.96297660e-01 5.85696757e-01 1.14578629e+00
-1.89754833e-02 8.34398270e-01 9.50639725e-01 5.03141940e-01
-1.17798972e+00 -1.00893247e+00 -2.44730651e-01 4.88671750e-01
-1.88364033e-02 -8.42302859e-01 -9.77573931e-01 -1.89041108e-01
-2.28490070e-01 1.72986424e+00 -4.79501396e-01 -3.36301982e-01
-1.39208004e-01 -7.38876164e-01 8.01557362e-01 1.55785263e-01
4.11729924e-02 -1.08184397e+00 -6.40242755e-01 3.99239928e-01
-7.87987232e-01 -6.79379880e-01 -2.47320518e-01 2.97981918e-01
-9.41653848e-01 -5.34210801e-01 -6.85915232e-01 -2.88506866e-01
6.18705809e-01 -6.59485385e-02 1.03986633e+00 2.01931939e-01
2.86086649e-01 1.04337402e-01 -4.15546060e-01 -5.18105567e-01
-8.85837078e-01 -6.19504377e-02 8.32663178e-02 -4.10547882e-01
-5.34594171e-02 -3.37672472e-01 -2.16147438e-01 -3.29756252e-02
-8.87995720e-01 2.56685972e-01 4.72376227e-01 7.50565469e-01
2.26803645e-01 9.76359919e-02 8.66512835e-01 -6.69759572e-01
1.30937254e+00 -8.99907470e-01 1.83492705e-01 2.27845892e-01
-7.21500576e-01 1.30635321e-01 4.36109662e-01 -3.84343296e-01
-9.17710125e-01 -4.67117816e-01 -3.24851647e-02 -9.64199901e-02
9.47835855e-03 8.37430060e-01 -1.24741115e-01 8.00593197e-01
8.10424149e-01 5.62928975e-01 -9.30968821e-02 -2.31315851e-01
8.09253156e-02 7.57396162e-01 4.11841869e-01 -5.17428517e-01
3.91969830e-01 5.01581207e-02 -4.69312876e-01 -1.13921070e+00
-7.00425565e-01 -7.76864812e-02 -9.07358676e-02 -2.49014348e-01
5.10061681e-01 -9.69571590e-01 1.52650254e-03 3.10751230e-01
-1.47765207e+00 -9.42504257e-02 -2.81674206e-01 4.21090841e-01
-7.56713629e-01 4.28102851e-01 -3.70064557e-01 -1.17288768e+00
-1.60463959e-01 -1.02872276e+00 8.69187534e-01 2.82553196e-01
-1.07432961e+00 -7.62050867e-01 5.38204312e-02 2.25837588e-01
3.04480702e-01 -4.08680327e-02 1.19403934e+00 -9.56559181e-01
-3.29343706e-01 -2.00802997e-01 9.62955281e-02 7.87530690e-02
-3.63697857e-02 2.01440632e-01 -8.88406396e-01 -1.03638329e-01
1.47553030e-02 -3.06094438e-01 7.91132629e-01 4.31902349e-01
9.03779030e-01 -6.88698769e-01 -3.40693474e-01 1.31191954e-01
1.02787054e+00 1.79870263e-01 8.20515871e-01 1.91792533e-01
2.05405742e-01 7.79148102e-01 4.81065899e-01 9.76958275e-01
2.62301117e-01 4.89285499e-01 2.86273628e-01 4.51636374e-01
-7.05754338e-03 -6.88727260e-01 5.07423103e-01 9.41377878e-01
2.72392601e-01 -1.00354004e+00 -7.48490155e-01 8.41304839e-01
-1.65892315e+00 -1.24473441e+00 -5.27946800e-02 2.18127179e+00
1.09352469e+00 3.90288502e-01 -1.35093272e-01 1.36008650e-01
7.65461624e-01 3.47003043e-01 -4.91269737e-01 -4.40246552e-01
-3.68713289e-01 -3.11610132e-01 1.77497387e-01 7.99721777e-01
-3.53945196e-01 8.06999028e-01 6.85449123e+00 7.68941700e-01
-8.04072618e-01 -1.37258470e-01 5.65494061e-01 -3.53419513e-01
-8.90617013e-01 2.15619162e-01 -6.18926466e-01 7.07975745e-01
1.18249857e+00 -8.41861844e-01 2.32433841e-01 5.90812445e-01
5.72720468e-01 -5.53750634e-01 -1.03522098e+00 5.35871327e-01
4.17900592e-01 -1.31587827e+00 2.76558459e-01 1.11849375e-01
5.23945034e-01 -2.69686222e-01 -1.83763858e-02 1.48647919e-01
7.93627873e-02 -9.92016017e-01 1.34448469e+00 6.12442732e-01
5.60761869e-01 -7.28266299e-01 4.73723441e-01 7.96802342e-01
-2.31789887e-01 8.30039680e-02 -4.37766492e-01 -2.48583220e-02
3.17010015e-01 6.66509748e-01 -1.33616626e+00 1.88198909e-01
1.11775011e-01 3.24500352e-01 -5.70253193e-01 1.13730693e+00
-3.14111203e-01 9.08377409e-01 -1.86653242e-01 -2.71393239e-01
5.52805290e-02 1.81442991e-01 1.12917995e+00 1.33538854e+00
7.43330061e-01 1.82400402e-02 -2.04804644e-01 9.86988008e-01
6.18616790e-02 4.74105626e-02 -8.05216312e-01 -3.38250548e-01
8.36009204e-01 6.69748902e-01 -6.57665789e-01 -4.77711529e-01
2.97198445e-01 1.03196585e+00 1.61872953e-01 1.93187401e-01
-5.90332270e-01 -3.41666728e-01 2.36221015e-01 1.65226534e-01
-4.86990623e-02 -1.08581707e-02 -7.27918863e-01 -1.24271047e+00
2.90142726e-02 -1.02220678e+00 -4.18940969e-02 -1.08105505e+00
-7.85065591e-01 6.50999129e-01 1.76057443e-01 -1.02003849e+00
-7.64300823e-01 1.46209612e-01 -8.71984065e-01 1.04585123e+00
-1.05993950e+00 -3.12026978e-01 3.19523871e-01 1.87770501e-01
8.15249324e-01 -1.33671179e-01 8.21919382e-01 -3.45686406e-01
-4.41986889e-01 5.15823603e-01 1.54103011e-01 -3.78167152e-01
6.59657180e-01 -1.30224407e+00 1.85204372e-01 8.70379627e-01
2.16531828e-02 8.17482769e-01 1.60544682e+00 -1.10224998e+00
-1.14801741e+00 -7.17154145e-01 1.62871444e+00 -4.48127747e-01
4.20879036e-01 -1.06766960e-03 -1.03896689e+00 5.35606682e-01
4.55696344e-01 -1.01030397e+00 7.34907448e-01 8.03935975e-02
5.24153858e-02 6.03634357e-01 -1.16233706e+00 9.88504350e-01
6.24983072e-01 -3.53969723e-01 -9.73450303e-01 6.18218303e-01
6.41720712e-01 -2.21263871e-01 -4.34622675e-01 2.47537158e-02
3.30174178e-01 -1.07466006e+00 4.45358485e-01 -3.89198244e-01
8.91116321e-01 -1.03391282e-01 -2.30402440e-01 -1.82587361e+00
-1.40342608e-01 -8.66356969e-01 -1.58044666e-01 1.27141786e+00
8.68765831e-01 -1.11014560e-01 6.24414325e-01 6.60511553e-01
-4.01431292e-01 -7.32922494e-01 -9.01459396e-01 -6.31107032e-01
1.30007759e-01 -5.18654227e-01 2.97830850e-01 8.14347327e-01
6.06312275e-01 3.37820590e-01 -5.55166364e-01 7.16954097e-02
6.31126463e-01 -3.37025136e-01 3.69798481e-01 -8.39860678e-01
-1.99484363e-01 -5.02360344e-01 -4.64431383e-02 -9.36097801e-01
1.76176995e-01 -8.24815392e-01 4.16321188e-01 -1.82711649e+00
2.81310707e-01 -1.28388911e-01 1.41454637e-01 3.07122320e-01
-2.25250900e-01 -2.41483569e-01 2.79930741e-01 4.76775229e-01
-4.40360934e-01 5.88642895e-01 1.16375971e+00 8.31253678e-02
-3.38497311e-01 -8.55207220e-02 -1.19113564e+00 7.34618545e-01
9.01141107e-01 -7.07843244e-01 -5.31769037e-01 -2.92502552e-01
4.71127748e-01 6.24873638e-01 1.94847211e-01 -9.74212527e-01
2.60294676e-01 -1.37457848e-01 3.00573885e-01 -8.74072731e-01
3.96756321e-01 -2.25032955e-01 1.17381454e-01 4.49854493e-01
-8.45026016e-01 2.03572139e-01 3.57119441e-02 5.58078468e-01
-1.43470494e-02 -8.03242624e-01 4.97972190e-01 -1.80837408e-01
-1.12872213e-01 -4.46577609e-01 -1.20431697e+00 2.71866053e-01
7.33353555e-01 -2.96434015e-01 -4.85529482e-01 -9.67671692e-01
-5.34392357e-01 1.06787831e-01 6.32800400e-01 3.27370018e-01
6.93101764e-01 -9.31045175e-01 -1.08975697e+00 -1.09862134e-01
4.41752672e-02 -3.07416588e-01 -6.50068745e-02 6.28863275e-01
-2.50295073e-01 3.64171565e-01 4.26246710e-02 -3.39759380e-01
-9.77311075e-01 -2.45557483e-02 1.83226392e-01 -9.33070257e-02
-5.31400502e-01 7.21571147e-01 -3.30440670e-01 1.07955948e-01
2.98231781e-01 -1.18050039e-01 4.14953381e-03 3.92069429e-01
5.83643794e-01 3.98136467e-01 8.18854943e-02 -6.68642402e-01
-1.66688964e-01 -2.68542677e-01 -3.36272597e-01 -7.38864958e-01
1.11099446e+00 -1.67616844e-01 -1.67417347e-01 7.81083047e-01
8.17454517e-01 2.82452941e-01 -1.13406873e+00 1.00870542e-01
9.13456604e-02 -5.35194039e-01 1.35615198e-02 -1.07608306e+00
-2.68930465e-01 6.69812202e-01 -1.30725533e-01 5.25037110e-01
5.56180239e-01 1.27035424e-01 4.76936907e-01 4.95036691e-01
1.42030001e-01 -1.15059769e+00 9.36459526e-02 6.23929739e-01
1.13098443e+00 -7.06920922e-01 2.30091661e-01 -6.00524917e-02
-1.05869257e+00 1.02106869e+00 4.40600127e-01 2.07587093e-01
5.27832918e-02 1.51214257e-01 -5.66947721e-02 -4.61784676e-02
-1.33203924e+00 2.35483393e-01 -6.04539737e-02 3.74569893e-01
7.11009502e-01 1.53142467e-01 -4.79574203e-01 4.86613244e-01
-6.22570813e-01 -3.25800747e-01 1.22953975e+00 7.56773233e-01
-8.41984928e-01 -6.87331676e-01 -5.33561409e-01 8.32783103e-01
-6.17453337e-01 -4.11162645e-01 -7.08071232e-01 3.41610461e-01
-6.51373267e-02 1.41812396e+00 6.27106652e-02 -2.81500936e-01
5.83608709e-02 4.03979301e-01 3.42611045e-01 -7.31121063e-01
-5.62408447e-01 3.07526499e-01 4.00891513e-01 1.01427864e-02
-3.29483263e-02 -1.08257198e+00 -1.16411340e+00 -3.16539675e-01
-6.29209280e-01 4.57779855e-01 6.04727268e-01 1.03956854e+00
2.33783811e-01 4.40395445e-01 3.91192794e-01 -7.46036351e-01
-9.13206160e-01 -1.21320748e+00 -6.80655956e-01 1.78062275e-01
3.14555526e-01 -2.96764880e-01 -6.44843042e-01 -1.30502030e-01]
|
[11.880477905273438, 9.14864444732666]
|
bb6ebccb-cff2-4623-accf-2cd35d61de05
|
automatic-lesion-detection-system-alds-for
|
2003.06276
| null |
https://arxiv.org/abs/2003.06276v1
|
https://arxiv.org/pdf/2003.06276v1.pdf
|
Automatic Lesion Detection System (ALDS) for Skin Cancer Classification Using SVM and Neural Classifiers
|
Technology aided platforms provide reliable tools in almost every field these days. These tools being supported by computational power are significant for applications that need sensitive and precise data analysis. One such important application in the medical field is Automatic Lesion Detection System (ALDS) for skin cancer classification. Computer aided diagnosis helps physicians and dermatologists to obtain a second opinion for proper analysis and treatment of skin cancer. Precise segmentation of the cancerous mole along with surrounding area is essential for proper analysis and diagnosis. This paper is focused towards the development of improved ALDS framework based on probabilistic approach that initially utilizes active contours and watershed merged mask for segmenting out the mole and later SVM and Neural Classifier are applied for the classification of the segmented mole. After lesion segmentation, the selected features are classified to ascertain that whether the case under consideration is melanoma or non-melanoma. The approach is tested for varying datasets and comparative analysis is performed that reflects the effectiveness of the proposed system.
|
['Rana Hammad Raza', 'Muhammad Aatif Mobeen Azhar', 'Muhammad Ali Farooq']
|
2020-03-13
| null | null | null | null |
['skin-cancer-classification']
|
['medical']
|
[ 7.39446700e-01 1.60503238e-02 -1.54885799e-01 -2.59875923e-01
-6.77533448e-01 -5.62321782e-01 4.12879139e-01 8.62773776e-01
-5.76910794e-01 7.23708153e-01 -1.59010321e-01 -3.65985781e-01
-3.04099023e-01 -7.92214334e-01 2.00165972e-01 -9.26475286e-01
3.77425224e-01 6.60710037e-01 5.17929256e-01 1.73608333e-01
5.92803001e-01 1.00949275e+00 -1.10289288e+00 6.26931265e-02
1.07848799e+00 4.31259722e-01 3.87353182e-01 1.13970053e+00
-2.01844901e-01 3.77155900e-01 -2.11959109e-01 2.86552981e-02
2.90956408e-01 -3.85633737e-01 -5.77123702e-01 2.28734314e-01
-1.32413104e-01 -1.04732402e-01 4.60255921e-01 9.65431452e-01
2.32098103e-01 -2.64957324e-02 1.31085658e+00 -7.19285727e-01
3.54265362e-01 -6.24154173e-02 -1.01903653e+00 4.78070587e-01
1.22741260e-01 2.00343207e-01 2.79974997e-01 -5.39259553e-01
4.19454366e-01 6.29399240e-01 6.14622235e-01 1.57333076e-01
-7.52081931e-01 -3.15684557e-01 -5.53533018e-01 2.83765197e-01
-1.34196830e+00 -3.00066024e-01 3.61703962e-01 -5.80971837e-01
6.08315229e-01 5.33191323e-01 4.73560303e-01 -7.12075680e-02
5.64450860e-01 4.53510821e-01 1.48729908e+00 -8.04104447e-01
4.06965852e-01 6.68351591e-01 7.21337318e-01 8.35186422e-01
6.03870511e-01 -2.16932833e-01 -1.16681881e-01 -5.52064292e-02
3.86237770e-01 -6.79931194e-02 2.35635310e-01 2.67986469e-02
-4.98129189e-01 6.67176187e-01 1.26733631e-01 5.70991814e-01
-8.45304966e-01 -2.93783247e-01 3.06358427e-01 -4.54709917e-01
1.39647707e-01 -2.53339976e-01 4.89089377e-02 4.89494763e-02
-1.37976670e+00 -6.84226602e-02 3.93037736e-01 1.95655078e-01
3.31036061e-01 -2.19948232e-01 9.08164307e-02 4.43571001e-01
4.95041281e-01 3.41037720e-01 3.81508827e-01 -2.14706942e-01
-1.31002322e-01 1.06880617e+00 -1.36800800e-02 -5.57278097e-01
-3.67479026e-01 -1.80104613e-01 -7.03423738e-01 7.31506348e-01
3.93723696e-01 -2.13074964e-02 -1.48377633e+00 6.31082773e-01
7.79525757e-01 1.12641007e-01 3.05333793e-01 6.08877897e-01
5.40578127e-01 5.05964458e-01 7.17070997e-01 -3.43142077e-02
1.77656031e+00 -2.92800248e-01 -5.55411875e-01 2.79365867e-01
3.35372716e-01 -1.04605114e+00 2.22029522e-01 4.41800058e-01
-6.06464982e-01 -1.33825526e-01 -1.09728777e+00 2.17598960e-01
-4.20546681e-01 4.30981845e-01 5.67790210e-01 8.56148660e-01
-6.68599725e-01 2.26111680e-01 -1.23507059e+00 -1.14967692e+00
7.28574336e-01 6.97527409e-01 -4.68992442e-01 5.80180883e-02
-4.95532185e-01 1.00752401e+00 4.59180802e-01 4.03634161e-02
-4.92049009e-01 -3.27122897e-01 -6.36914015e-01 -3.64700466e-01
-3.36743146e-02 -4.46012169e-01 8.71962965e-01 -9.56560314e-01
-1.05241919e+00 9.69244301e-01 -4.39251393e-01 -6.66310608e-01
5.82906783e-01 5.33928394e-01 3.92174115e-03 3.97929370e-01
-6.77733645e-02 2.23811820e-01 5.76647878e-01 -7.47877598e-01
-1.12574804e+00 -8.86459172e-01 -5.54965913e-01 5.95984221e-01
1.28078818e-01 1.08858667e-01 -1.10613285e-02 -1.64377332e-01
2.71450937e-01 -5.92550397e-01 -1.82091802e-01 1.24623738e-01
-5.89996576e-01 2.76332479e-02 1.12365520e+00 -1.04268587e+00
8.55665028e-01 -1.70534897e+00 -3.96862268e-01 6.87277675e-01
-5.62264696e-02 4.28385526e-01 5.73166013e-01 4.31893528e-01
1.00501716e-01 4.77311574e-02 -4.99404907e-01 -7.45299086e-02
-4.64650691e-01 -1.60231054e-01 4.17416990e-01 7.60829985e-01
4.27315444e-01 2.01705277e-01 -2.48361394e-01 -1.01206589e+00
5.75502574e-01 8.05925369e-01 2.94198543e-01 1.43900171e-01
-2.91305669e-02 3.92186284e-01 -7.56247401e-01 1.06420612e+00
7.15483189e-01 1.21475831e-01 -6.29764348e-02 -4.19632286e-01
-2.93043405e-01 -5.35272181e-01 -1.19974732e+00 8.69615436e-01
-2.13220239e-01 5.88630199e-01 1.27370447e-01 -9.95012462e-01
7.52061367e-01 5.26657999e-01 3.86953086e-01 -1.03951447e-01
5.11711061e-01 1.87045231e-01 7.30962455e-02 -9.69278038e-01
2.00933173e-01 -3.10895890e-01 6.03057921e-01 3.02675724e-01
-3.78279239e-01 1.24407141e-02 3.16114187e-01 4.57863994e-02
8.31425548e-01 2.59062201e-01 1.10955620e+00 -2.29477778e-01
8.18122923e-01 7.84385681e-01 2.66008556e-01 -2.30395701e-02
-6.16977990e-01 1.11564405e-01 1.27935141e-01 1.77260876e-01
-7.07812011e-01 -8.85864437e-01 -5.99829316e-01 2.73932874e-01
9.73510519e-02 6.71450377e-01 -8.25902939e-01 -5.70961475e-01
-2.83561200e-02 4.95895714e-01 -6.35868371e-01 6.00788414e-01
-1.24022216e-01 -1.04379904e+00 7.17504770e-02 2.71351710e-02
7.84224927e-01 -7.33179212e-01 -9.51101363e-01 1.91270649e-01
6.75116062e-01 -5.46325862e-01 3.97690773e-01 -2.33640359e-03
-8.96067917e-01 -1.43409169e+00 -8.31439555e-01 -7.92981505e-01
1.02520764e+00 -1.02856874e-01 1.61040202e-01 1.09812163e-01
-1.21273887e+00 4.59493436e-02 -8.59879255e-02 -7.30281651e-01
-6.14224374e-01 -4.14460190e-02 -4.59115654e-01 -6.19023666e-02
6.07814610e-01 -3.30335587e-01 -8.06916416e-01 -1.42235130e-01
-7.36585915e-01 -6.75115585e-02 8.24515104e-01 3.49381208e-01
5.90182900e-01 6.19425535e-01 5.23000121e-01 -1.34648609e+00
5.74871659e-01 -6.32304549e-01 -6.21537566e-01 4.91354585e-01
-2.72737652e-01 -2.36949116e-01 4.02916610e-01 2.16352805e-01
-1.32843339e+00 4.74269897e-01 -4.29344364e-02 5.49491048e-01
-6.87728703e-01 4.41383958e-01 -1.61443744e-02 -3.06330353e-01
6.36924982e-01 1.89325839e-01 1.47516038e-02 -2.66938150e-01
-3.06602955e-01 1.06561148e+00 5.65377951e-01 6.76643476e-02
5.34417391e-01 6.92091763e-01 4.87946451e-01 -1.07636440e+00
-6.95542172e-02 -8.83983672e-01 -6.11498654e-01 -4.10827279e-01
1.05656672e+00 -2.77973324e-01 -3.97627652e-01 4.88364398e-01
-7.82825768e-01 3.17948759e-01 4.25434381e-01 5.07351875e-01
-3.93589698e-02 3.89489353e-01 -2.78024584e-01 -1.68886745e+00
-7.88655162e-01 -1.04041052e+00 6.18592024e-01 9.33516324e-01
-5.09974658e-01 -1.18883777e+00 3.99659052e-02 5.89772046e-01
3.74692202e-01 6.00577414e-01 9.81864393e-01 -6.10505462e-01
-4.28671777e-01 -9.41539824e-01 -1.86987653e-01 1.03819296e-02
5.35420001e-01 5.86737275e-01 -9.26847458e-01 1.90470085e-01
-7.19951913e-02 7.46078640e-02 6.56373918e-01 6.70506597e-01
4.13740993e-01 1.12354569e-01 -7.40162730e-01 1.88539103e-01
2.10832095e+00 6.43605709e-01 4.10722822e-01 4.61695120e-02
3.10153067e-01 7.55208731e-01 9.08318460e-01 2.63346672e-01
-2.14034710e-02 -1.96510572e-02 1.91597685e-01 -2.63024211e-01
-1.80418238e-01 2.01038778e-01 -9.42450911e-02 5.97925782e-02
-1.91556048e-02 -2.87999064e-01 -1.12158215e+00 8.37273955e-01
-1.20753539e+00 -7.00745702e-01 -4.07985479e-01 2.17037582e+00
7.48724043e-01 -4.09056582e-02 9.54342708e-02 8.73870671e-01
1.06000221e+00 -4.87445235e-01 -4.66107935e-01 -5.03086150e-01
3.48613352e-01 8.59795153e-01 6.36764467e-01 9.15122211e-01
-1.01771963e+00 5.78721881e-01 4.42161942e+00 9.55523431e-01
-1.40265918e+00 -1.50281027e-01 7.71440506e-01 3.55481923e-01
1.70253232e-01 -2.24785581e-02 -8.27547193e-01 4.40289080e-01
4.42308635e-01 -8.82223547e-02 -2.47153148e-01 3.92985731e-01
6.24065101e-01 -1.47200894e+00 -5.90470731e-01 3.62859070e-01
-2.43526459e-01 -1.16849232e+00 -5.65912612e-02 -9.29355621e-02
6.48617864e-01 -5.04492819e-01 -4.77228984e-02 -6.06146038e-01
-8.51625577e-02 -1.12441659e+00 -8.00318569e-02 8.98293912e-01
6.38277352e-01 -8.19908440e-01 9.81818676e-01 4.51724976e-01
-9.45350468e-01 4.12750632e-01 1.66643411e-01 3.10724109e-01
7.87910745e-02 3.80954653e-01 -1.86102033e+00 3.51514637e-01
-1.79307282e-01 1.18708722e-01 -4.60643500e-01 1.54464364e+00
2.24758714e-01 7.99814939e-01 -6.30314231e-01 -2.69503236e-01
1.83653623e-01 -2.16262534e-01 6.08768284e-01 1.22464538e+00
2.74877012e-01 2.18113631e-01 -4.36309040e-01 4.74357337e-01
6.19177878e-01 5.90325832e-01 -2.79519737e-01 -2.76874512e-01
4.95491803e-01 1.41357505e+00 -1.48510373e+00 -2.09708065e-01
1.24340117e-01 8.35328460e-01 -2.89964348e-01 -4.22229432e-03
-3.14944059e-01 -6.35041356e-01 -3.01878024e-02 6.99764848e-01
-2.33998984e-01 1.63059290e-02 -7.01673329e-01 -1.17732078e-01
-4.46497798e-01 -3.70605916e-01 5.76444983e-01 -2.29943573e-01
-6.76148117e-01 3.92550856e-01 -1.00685284e-01 -7.14103878e-01
-8.40690956e-02 -5.86460412e-01 -1.08193147e+00 1.17799020e+00
-1.42050803e+00 -1.40085614e+00 -5.47981799e-01 3.24991196e-01
6.64381504e-01 -2.49657154e-01 8.59827876e-01 -1.44083619e-01
-7.41120577e-01 1.78966701e-01 1.64686114e-01 -4.64858711e-02
4.23860520e-01 -1.18018711e+00 -5.25898159e-01 1.08033693e+00
-3.66220027e-01 3.14164609e-01 7.19615102e-01 -1.01280105e+00
-8.72549534e-01 -8.24805081e-01 7.10516989e-01 6.58408403e-02
3.23390871e-01 2.72884339e-01 -4.56882209e-01 4.75128479e-02
2.38148913e-01 -2.46389434e-01 8.37294996e-01 -7.32542992e-01
6.62064254e-01 -1.48190763e-02 -1.96245182e+00 3.12900096e-01
-2.70184249e-01 -7.07498044e-02 -1.50340751e-01 2.51409560e-01
-5.78903973e-01 -3.00817281e-01 -5.07473588e-01 3.02588344e-01
4.95121151e-01 -1.07904184e+00 3.88646573e-01 -3.63575757e-01
1.06143311e-01 -4.46804315e-01 1.04815163e-01 -6.60698891e-01
3.44920844e-01 -1.47155508e-01 6.52999699e-01 1.15937436e+00
3.43307853e-01 -4.25302446e-01 1.44835973e+00 7.65580535e-01
4.68506753e-01 -9.29592967e-01 -7.37158656e-01 1.18325725e-01
-3.80836457e-01 -8.90869945e-02 -8.39178562e-02 6.98588431e-01
-3.06077123e-01 -1.96775496e-01 4.62347656e-01 3.52416992e-01
1.05237401e+00 -3.33666235e-01 2.40639418e-01 -1.04546154e+00
7.21824467e-02 -4.60672081e-02 -1.01456821e+00 2.79573292e-01
-4.76954520e-01 -6.06396914e-01 -2.10098341e-01 -1.91016030e+00
3.33505571e-01 -6.40610158e-01 5.59152570e-03 3.56756985e-01
-3.30223411e-01 4.62803245e-01 -4.20510799e-01 2.43731607e-02
2.28384182e-01 -3.45999390e-01 8.02257419e-01 1.78654835e-01
-2.03303322e-01 4.09156680e-01 -5.05487084e-01 8.01804304e-01
9.08640802e-01 -2.51580358e-01 -4.94119108e-01 2.22484127e-01
-4.18589622e-01 7.08190054e-02 3.27987343e-01 -1.12265003e+00
4.60579515e-01 -4.02402431e-01 5.93854129e-01 -6.89777434e-01
2.31478438e-01 -1.04437804e+00 2.45819449e-01 6.37006879e-01
-1.22565642e-01 -5.03098905e-01 7.74056911e-02 5.59537053e-01
2.18690671e-02 -7.11943448e-01 1.20539963e+00 -9.96754915e-02
-7.36278474e-01 1.19691398e-02 -7.02981949e-01 -6.30240917e-01
1.90330887e+00 -8.65219176e-01 4.33766730e-02 -7.63073936e-02
-8.99316728e-01 3.53845581e-02 3.91007423e-01 -6.46481633e-01
3.46380532e-01 -5.79454780e-01 -8.96835029e-01 -9.98420715e-02
-9.28078517e-02 9.07284915e-02 4.75478590e-01 8.83821011e-01
-1.45250261e+00 4.78741467e-01 -3.47001076e-01 -3.57926130e-01
-2.18964720e+00 -2.74693817e-01 4.67091441e-01 -1.58648476e-01
-1.23797588e-01 8.99453878e-01 -4.22708094e-01 5.14434874e-01
5.13935499e-02 -2.25018948e-01 -7.10988283e-01 7.31619541e-03
2.99676746e-01 6.26679778e-01 9.44638997e-02 -6.87427580e-01
-4.40046966e-01 7.04023361e-01 -3.62650990e-01 -2.38669097e-01
1.07806981e+00 1.25881538e-01 -1.89378634e-01 1.36456937e-01
7.74443388e-01 2.21070588e-01 -7.86240399e-01 1.14731438e-01
1.67093784e-01 -3.96756589e-01 4.86431658e-01 -1.19962668e+00
-5.47594905e-01 7.27117062e-01 1.02987421e+00 -1.62759442e-02
1.13800800e+00 -5.12255728e-01 3.47935975e-01 2.58964244e-02
-7.72044733e-02 -1.04299986e+00 -7.54527032e-01 -4.16540444e-01
3.42616260e-01 -1.35673714e+00 5.89101255e-01 -8.43048632e-01
-8.04368496e-01 1.32077265e+00 2.85463661e-01 -3.99768621e-01
8.18967164e-01 5.66001058e-01 3.36323053e-01 -1.66663170e-01
-3.04006070e-01 -2.65087157e-01 1.64622784e-01 8.14962506e-01
6.07612550e-01 8.80623385e-02 -7.84631610e-01 3.77043396e-01
2.74145976e-03 3.53111267e-01 3.83247465e-01 1.25966740e+00
-8.06412518e-01 -9.64899540e-01 -8.30419540e-01 8.79023731e-01
-8.07223618e-01 1.26536861e-01 -6.06975079e-01 1.04852116e+00
4.52886164e-01 7.82436490e-01 3.40593755e-02 2.96709120e-01
-3.43318403e-01 2.11983435e-02 6.93617702e-01 -6.24495745e-01
-6.26506150e-01 -5.51324934e-02 1.74121439e-01 2.97087580e-01
-4.35857147e-01 -7.37345815e-01 -1.58615768e+00 3.82506363e-02
-4.26172435e-01 2.41594329e-01 1.41999805e+00 8.99945199e-01
-2.80272514e-01 1.45086959e-01 4.24452603e-01 -1.26429036e-01
-1.64949343e-01 -8.52995038e-01 -7.92919755e-01 -1.29842877e-01
1.75954372e-01 -3.58519584e-01 -1.11280501e-01 3.74104798e-01]
|
[15.466198921203613, -3.0252604484558105]
|
2ae891d3-19ad-44b1-90a8-c825c34de9da
|
soft-layer-selection-with-meta-learning-for
|
2107.09840
| null |
https://arxiv.org/abs/2107.09840v1
|
https://arxiv.org/pdf/2107.09840v1.pdf
|
Soft Layer Selection with Meta-Learning for Zero-Shot Cross-Lingual Transfer
|
Multilingual pre-trained contextual embedding models (Devlin et al., 2019) have achieved impressive performance on zero-shot cross-lingual transfer tasks. Finding the most effective fine-tuning strategy to fine-tune these models on high-resource languages so that it transfers well to the zero-shot languages is a non-trivial task. In this paper, we propose a novel meta-optimizer to soft-select which layers of the pre-trained model to freeze during fine-tuning. We train the meta-optimizer by simulating the zero-shot transfer scenario. Results on cross-lingual natural language inference show that our approach improves over the simple fine-tuning baseline and X-MAML (Nooralahzadeh et al., 2020).
|
['Saab Mansour', 'Jason Krone', 'Batool Haider', 'Weijia Xu']
|
2021-07-21
| null |
https://aclanthology.org/2021.metanlp-1.2
|
https://aclanthology.org/2021.metanlp-1.2.pdf
|
acl-metanlp-2021-8
|
['cross-lingual-natural-language-inference']
|
['natural-language-processing']
|
[-2.30044156e-01 4.18913290e-02 -5.05012989e-01 -6.43387794e-01
-1.35039091e+00 -5.96489608e-01 8.84077847e-01 -2.08393171e-01
-8.42064857e-01 9.73770022e-01 5.25950313e-01 -5.20832598e-01
2.62044221e-01 -6.78628922e-01 -1.06231356e+00 -3.63002777e-01
1.26279309e-01 7.50144958e-01 7.69885024e-03 -4.54317480e-01
-2.46432304e-01 8.48015863e-03 -1.15525889e+00 3.88520777e-01
7.44472146e-01 2.26295546e-01 3.85981977e-01 7.65325069e-01
-1.93510890e-01 3.28511178e-01 -1.35664985e-01 -7.12329626e-01
7.50271827e-02 -2.73086131e-01 -9.42162395e-01 -5.31542003e-01
7.56374896e-01 -2.22777873e-01 -1.99196428e-01 9.90125656e-01
6.24182045e-01 3.76196474e-01 8.07378769e-01 -7.06436932e-01
-9.40564275e-01 1.12243545e+00 -3.49054158e-01 3.02679241e-01
-2.19380885e-01 1.14702046e-01 1.18431640e+00 -1.06368256e+00
6.54569030e-01 1.46992135e+00 7.75860965e-01 1.02080202e+00
-1.50415468e+00 -7.73424447e-01 1.36840910e-01 1.89385742e-01
-1.44807458e+00 -6.58136964e-01 4.37223434e-01 -3.85183036e-01
1.38321018e+00 -1.07835218e-01 2.20625639e-01 1.46777284e+00
2.64469683e-01 6.76662445e-01 9.93038476e-01 -8.30235839e-01
-2.93285470e-03 3.64530981e-01 -2.96154507e-02 7.44976819e-01
-4.39760871e-02 1.85348958e-01 -5.08004308e-01 -1.21031180e-01
3.26014131e-01 -4.38215256e-01 7.92369023e-02 -1.55626461e-01
-1.03172290e+00 1.02924907e+00 3.46209586e-01 5.56346118e-01
-1.01454616e-01 4.18334663e-01 6.29506409e-01 4.38557446e-01
8.21277440e-01 6.27949417e-01 -9.29950118e-01 -2.05655485e-01
-1.00169468e+00 7.17284530e-02 6.07241035e-01 9.59144413e-01
9.19847667e-01 4.70708534e-02 -3.76810163e-01 1.20309293e+00
1.90986067e-01 1.84944034e-01 6.77915454e-01 -6.87062144e-01
6.86258137e-01 -3.75728980e-02 -8.98358375e-02 -9.85012203e-02
-2.87339985e-02 -3.13207179e-01 -6.36851370e-01 -1.29068000e-02
1.59424022e-01 -3.32664937e-01 -9.00725365e-01 2.09188223e+00
1.64268896e-01 2.01065391e-01 2.53073931e-01 4.64996248e-01
3.38940740e-01 1.03375423e+00 6.39984012e-01 1.42179936e-01
1.36904812e+00 -1.28752232e+00 -6.40824318e-01 -4.60591227e-01
8.93765926e-01 -7.61656821e-01 1.71530461e+00 -8.24401379e-02
-1.07379103e+00 -7.39458144e-01 -1.15721571e+00 -4.56208140e-01
-6.85314655e-01 8.72710794e-02 6.44989610e-01 4.46535408e-01
-9.69408691e-01 5.80956340e-01 -9.03175712e-01 -4.24297571e-01
2.02732444e-01 1.59234762e-01 -4.10972267e-01 -3.13667685e-01
-1.78502941e+00 1.34113407e+00 6.24375224e-01 -2.92139679e-01
-1.11934638e+00 -1.17952323e+00 -9.12017465e-01 -5.68054989e-03
1.95019051e-01 -8.17314506e-01 1.42923033e+00 -8.13741207e-01
-1.72103024e+00 1.09371543e+00 -2.49870867e-01 -4.12705272e-01
3.22466671e-01 -4.10099894e-01 -2.51356184e-01 -4.73280132e-01
1.87305883e-01 8.47183347e-01 6.06408238e-01 -9.56986248e-01
-5.66423416e-01 -4.45805714e-02 1.68412730e-01 3.31396967e-01
-4.42699522e-01 1.61220983e-01 -6.11320376e-01 -7.11006165e-01
-9.07826722e-01 -9.22283351e-01 -1.79691598e-01 -4.45171118e-01
-6.39442727e-02 -5.01803398e-01 3.86324197e-01 -5.88062584e-01
1.32669282e+00 -1.93848872e+00 3.09977055e-01 -3.30262452e-01
-3.80021393e-01 5.96273422e-01 -4.24673170e-01 5.40522814e-01
5.41284606e-02 1.85474947e-01 -3.14844251e-01 -6.69633806e-01
3.12975824e-01 5.18575490e-01 -3.50272208e-01 2.15899661e-01
2.00388953e-01 1.18777418e+00 -8.84919941e-01 -5.89278638e-01
3.15243304e-01 7.41572976e-01 -8.39538217e-01 3.52096796e-01
-4.15918648e-01 3.31372857e-01 -1.74328208e-01 2.23532125e-01
2.43318722e-01 8.40074569e-03 2.06233308e-01 -2.98470229e-01
-1.57255828e-01 5.66585422e-01 -7.22221851e-01 2.11923027e+00
-1.20623147e+00 5.05498707e-01 -2.00047493e-01 -6.04185224e-01
5.93150020e-01 4.70072567e-01 -4.31584194e-02 -6.70201898e-01
2.56523229e-02 1.26291901e-01 -1.12637140e-01 -4.81255859e-01
6.43911242e-01 -5.44830143e-01 -4.47900862e-01 6.77502573e-01
6.59438610e-01 -9.60356966e-02 1.78747326e-01 9.34559293e-03
6.80358350e-01 4.77592617e-01 4.47827727e-01 -7.11998641e-01
5.40419459e-01 -2.06435800e-01 2.85237014e-01 6.94322884e-01
-1.40351981e-01 2.77706236e-01 -8.80400371e-03 -1.95995867e-01
-1.24865472e+00 -1.14184213e+00 -1.61167160e-01 1.91497922e+00
-4.75863159e-01 -4.54511821e-01 -1.02404535e+00 -7.63700843e-01
-6.94070905e-02 1.18401933e+00 -9.48657155e-01 -2.93471396e-01
-7.80865312e-01 -6.64040446e-01 7.19653785e-01 3.70280683e-01
1.91123173e-01 -1.01111686e+00 -9.54370573e-02 4.78926361e-01
-1.91313118e-01 -8.36753666e-01 -9.03205693e-01 3.90813172e-01
-5.86230814e-01 -3.89369637e-01 -8.40517998e-01 -8.54452729e-01
4.73078072e-01 -2.97037780e-01 1.52578497e+00 -4.10228640e-01
-1.68550342e-01 7.16251954e-02 -8.43869708e-03 -2.07820609e-01
-5.74047685e-01 7.84786344e-01 2.74463236e-01 -2.69018680e-01
6.04382396e-01 -3.69184196e-01 -2.19990402e-01 -1.14954494e-01
-6.03693724e-01 -5.52114770e-02 3.10497284e-01 9.73395526e-01
7.60170937e-01 -3.94692987e-01 6.02472782e-01 -1.22586954e+00
7.86028385e-01 -3.67050737e-01 -5.80264330e-01 7.46927381e-01
-6.89293802e-01 6.42448545e-01 8.05073917e-01 -4.92891967e-01
-1.28056550e+00 -1.98212788e-01 -3.72775555e-01 -3.75111967e-01
5.31674549e-02 2.85898536e-01 -3.59572232e-01 1.48896009e-01
6.72329783e-01 -6.49882108e-02 -7.04352915e-01 -8.12525868e-01
1.10069335e+00 6.16638303e-01 4.81741995e-01 -1.16767049e+00
7.16950417e-01 3.07231732e-02 -5.50799966e-01 -6.16410732e-01
-1.36554611e+00 -2.40260929e-01 -8.02650809e-01 1.79022714e-01
1.09232736e+00 -1.15020180e+00 -1.45257115e-01 -7.19382986e-02
-1.19978273e+00 -8.80303979e-01 -2.64828682e-01 5.90301156e-01
-6.04527950e-01 -5.75163588e-02 -6.86289489e-01 -4.48155880e-01
-5.63112855e-01 -1.06298327e+00 9.31873679e-01 -1.40204979e-02
-3.45024347e-01 -1.55995452e+00 8.88481915e-01 3.01748421e-02
7.08601952e-01 -3.03307295e-01 1.13052225e+00 -3.31899345e-01
-2.80086190e-01 3.53280813e-01 -2.98273358e-02 3.64041328e-01
5.48335798e-02 -4.83625531e-02 -1.20380056e+00 -4.80562061e-01
-4.14452076e-01 -6.56148791e-01 1.00547838e+00 3.78266394e-01
1.03958583e+00 -3.73731911e-01 -2.37072319e-01 1.06706798e+00
1.51532745e+00 -2.02354625e-01 3.60907435e-01 4.29167181e-01
6.88716710e-01 3.76512617e-01 5.07456899e-01 -2.98880130e-01
7.38816381e-01 8.04593265e-01 -2.17331097e-01 3.08551341e-02
-3.42171133e-01 -6.04952812e-01 5.47973454e-01 1.39687085e+00
1.02047874e-02 -1.57881632e-01 -7.52020717e-01 7.69221187e-01
-1.66834605e+00 -8.77395034e-01 4.96645868e-01 1.96311986e+00
1.50067866e+00 5.71356788e-02 -1.71162412e-01 -6.98043883e-01
5.60583949e-01 1.96397990e-01 -3.34617704e-01 -8.76927078e-01
-1.19228050e-01 6.38294101e-01 4.73443806e-01 1.13069212e+00
-1.11972117e+00 1.76473975e+00 6.28426838e+00 9.87907767e-01
-1.03234839e+00 7.62195528e-01 3.80100578e-01 -9.83898342e-02
-4.76595342e-01 -5.33154979e-02 -1.37312627e+00 3.81136209e-01
1.49486291e+00 -3.86973053e-01 8.22167218e-01 6.96990192e-01
-2.15366751e-01 3.40265781e-01 -1.23786926e+00 5.81608474e-01
2.14746520e-01 -1.18634081e+00 -2.69332975e-02 -2.21110299e-01
1.14898264e+00 6.25968695e-01 8.03860575e-02 9.90554154e-01
8.55789006e-01 -9.49235320e-01 4.45088476e-01 3.45317274e-01
1.10293043e+00 -9.08253849e-01 3.91215026e-01 2.72496611e-01
-1.00765991e+00 3.60465765e-01 -5.38201511e-01 2.95873702e-01
3.77807349e-01 3.00378174e-01 -7.67544150e-01 1.93149343e-01
5.56572556e-01 4.11722332e-01 -4.75370705e-01 3.33364725e-01
-5.24513900e-01 9.07194734e-01 -1.12147354e-01 1.06083915e-01
4.62765366e-01 -1.38267446e-02 2.21915916e-01 1.78322065e+00
3.06792498e-01 -3.15789402e-01 -2.01368243e-01 7.61388540e-01
-4.42522526e-01 2.47895166e-01 -6.60208344e-01 -1.76222712e-01
3.49486470e-01 1.06513083e+00 3.71351615e-02 -5.62292397e-01
-5.74828565e-01 9.90703762e-01 1.07372642e+00 3.87417942e-01
-8.48566055e-01 -4.27216768e-01 9.76920247e-01 -1.00529112e-01
4.26812559e-01 -2.56455660e-01 -8.65676180e-02 -1.42492640e+00
-4.63346183e-01 -6.86894059e-01 4.51564401e-01 -5.69050848e-01
-1.52723932e+00 7.26919770e-01 1.61042616e-01 -7.47446895e-01
-6.33470416e-01 -6.48561060e-01 -4.14256394e-01 1.24372327e+00
-1.64738464e+00 -1.51153839e+00 4.68171775e-01 6.39692605e-01
6.81693554e-01 -2.86812156e-01 1.25045192e+00 6.03958786e-01
-4.69517052e-01 1.12191737e+00 2.03136519e-01 1.75984763e-02
1.05820298e+00 -1.26826608e+00 7.94854701e-01 7.42031991e-01
3.12168062e-01 8.83162022e-01 6.59668148e-01 -4.20069039e-01
-1.16853750e+00 -1.37348962e+00 1.40712237e+00 -5.82780302e-01
8.94536376e-01 -6.57167435e-01 -1.16020656e+00 1.09171438e+00
7.80726194e-01 -9.33377892e-02 7.31310904e-01 5.71322799e-01
-6.83879137e-01 -2.06988052e-01 -7.53215492e-01 8.27076256e-01
8.19226444e-01 -1.05894816e+00 -1.01520181e+00 2.74464756e-01
1.00177264e+00 -1.70095369e-01 -9.83196855e-01 1.42501563e-01
4.08733070e-01 -3.75905931e-01 9.62427318e-01 -1.06868684e+00
4.03959095e-01 6.02798387e-02 -3.27711761e-01 -1.80036068e+00
-4.63801324e-01 -6.55249774e-01 -4.76778969e-02 1.47146010e+00
6.88111186e-01 -5.05850613e-01 2.25071475e-01 2.32000753e-01
-1.25827819e-01 -4.89641488e-01 -9.59658802e-01 -7.69188523e-01
7.60065198e-01 -3.58912170e-01 5.54460108e-01 1.15999925e+00
-2.48216748e-01 8.54512095e-01 -6.06707692e-01 -1.40326932e-01
6.95126653e-01 -2.26586819e-01 7.91731119e-01 -6.97386980e-01
-7.19253421e-01 -3.52743477e-01 2.56814539e-01 -7.92115927e-01
7.62094140e-01 -1.43307948e+00 1.52694628e-01 -1.38242710e+00
3.25010449e-01 -4.91376400e-01 -7.51266718e-01 6.54255688e-01
-5.03133953e-01 1.96201995e-01 1.11730412e-01 -1.75208762e-01
-4.61148351e-01 5.63706994e-01 9.50054646e-01 -1.71220750e-01
-2.71513909e-02 -2.71604151e-01 -5.87194204e-01 4.35468137e-01
8.42466176e-01 -7.08176792e-01 -3.15630674e-01 -9.67228532e-01
3.08706820e-01 -3.44892412e-01 -1.43131509e-01 -4.66050535e-01
7.41276592e-02 -2.43001863e-01 -6.68675154e-02 -3.07082593e-01
4.28791523e-01 -4.40506697e-01 -4.84991446e-02 3.06078851e-01
-6.65976048e-01 1.07214198e-01 5.51260591e-01 2.33386919e-01
-7.31329918e-02 -3.42044830e-01 9.99071479e-01 -2.64977157e-01
-7.71151781e-01 4.24466133e-01 -1.45181835e-01 6.11790776e-01
7.41614878e-01 4.06513631e-01 -3.12267005e-01 -5.46540990e-02
-6.60515428e-01 1.53237611e-01 3.27393413e-01 6.90221667e-01
-5.96575923e-02 -1.58472157e+00 -9.18280005e-01 1.60964414e-01
2.64516234e-01 -3.54028523e-01 2.28034884e-01 4.55681622e-01
-9.91209596e-02 7.32714355e-01 -2.51298040e-01 -2.63648987e-01
-9.22316849e-01 5.61089039e-01 3.64483356e-01 -6.24591410e-01
-3.63450289e-01 1.17877662e+00 3.87082547e-01 -1.04035950e+00
5.42329177e-02 -9.98286009e-02 2.41031691e-01 8.43653455e-02
5.55868864e-01 9.36023593e-02 3.21626887e-02 -5.26601017e-01
-2.56092280e-01 7.48286307e-01 -3.28232437e-01 -5.01970232e-01
1.40769124e+00 -1.96359947e-01 -4.04713973e-02 9.51628685e-01
1.74918020e+00 3.89482491e-02 -1.20538354e+00 -5.76109767e-01
-8.04451481e-02 1.09671056e-03 1.25427574e-01 -8.37171912e-01
-6.95871234e-01 1.41709232e+00 3.88492256e-01 -3.89645994e-01
6.00008130e-01 8.82301182e-02 7.43275106e-01 4.41925615e-01
3.64247978e-01 -1.29172909e+00 -3.07020932e-01 8.81697536e-01
7.76727200e-01 -1.25429928e+00 -1.38008267e-01 4.20038134e-01
-6.48470759e-01 9.23839748e-01 7.06196904e-01 -6.81122541e-02
7.45390058e-01 3.42934310e-01 1.15193322e-01 1.79040521e-01
-1.16032815e+00 -1.96848661e-01 5.08117735e-01 3.36943269e-01
7.24849105e-01 5.01820803e-01 -1.76632062e-01 4.94866908e-01
-4.28530991e-01 9.57346112e-02 -7.83054903e-02 4.70589489e-01
-3.96274984e-01 -1.25877643e+00 -9.09794495e-03 9.54360738e-02
-4.94396627e-01 -5.68400025e-01 -9.34765786e-02 8.26591671e-01
1.43723816e-01 3.47599655e-01 9.84998494e-02 -9.27241147e-02
2.81791687e-01 6.21443927e-01 8.06121051e-01 -8.56517017e-01
-7.96069860e-01 -1.25168517e-01 2.08255246e-01 -4.36782122e-01
-1.05302930e-01 -4.68459457e-01 -8.70872378e-01 -2.11422443e-01
-5.86246736e-02 1.22860610e-01 7.71271527e-01 9.68498409e-01
2.99236685e-01 6.27898335e-01 3.04201305e-01 -7.65579224e-01
-7.56557703e-01 -1.18721271e+00 -1.15019277e-01 1.77206919e-01
3.47980678e-01 -5.02175331e-01 -2.77166128e-01 1.56679362e-01]
|
[10.995960235595703, 9.658278465270996]
|
082af2a5-ca03-455a-96ae-2c4f1554cc88
|
hybridformer-improving-squeezeformer-with
|
2303.08636
| null |
https://arxiv.org/abs/2303.08636v1
|
https://arxiv.org/pdf/2303.08636v1.pdf
|
HYBRIDFORMER: improving SqueezeFormer with hybrid attention and NSR mechanism
|
SqueezeFormer has recently shown impressive performance in automatic speech recognition (ASR). However, its inference speed suffers from the quadratic complexity of softmax-attention (SA). In addition, limited by the large convolution kernel size, the local modeling ability of SqueezeFormer is insufficient. In this paper, we propose a novel method HybridFormer to improve SqueezeFormer in a fast and efficient way. Specifically, we first incorporate linear attention (LA) and propose a hybrid LASA paradigm to increase the model's inference speed. Second, a hybrid neural architecture search (NAS) guided structural re-parameterization (SRep) mechanism, termed NSR, is proposed to enhance the ability of the model to extract local interactions. Extensive experiments conducted on the LibriSpeech dataset demonstrate that our proposed HybridFormer can achieve a 9.1% relative word error rate (WER) reduction over SqueezeFormer on the test-other dataset. Furthermore, when input speech is 30s, the HybridFormer can improve the model's inference speed up to 18%. Our source code is available online.
|
['Heng Lu', 'Lei Ma', 'Jiangyu Han', 'JingJing Yin', 'Yu Pan', 'Yuguang Yang']
|
2023-03-15
| null | null | null | null |
['architecture-search']
|
['methodology']
|
[ 2.22818866e-01 3.44161168e-02 5.80105707e-02 -4.65645850e-01
-9.45448816e-01 -1.55926362e-01 2.81192452e-01 -3.81250560e-01
-4.33573961e-01 4.04721111e-01 2.65472025e-01 -7.37508953e-01
2.53493004e-02 -2.67597467e-01 -6.20211244e-01 -7.37945199e-01
3.84843320e-01 2.48925705e-02 5.39960042e-02 -1.50885105e-01
7.90494904e-02 4.14044768e-01 -1.25369000e+00 1.41742364e-01
1.14444041e+00 8.53188396e-01 8.85838747e-01 7.24496663e-01
-4.67432439e-02 5.49330235e-01 -7.32298553e-01 -1.01163276e-01
-9.84822437e-02 -2.68814474e-01 -5.72532833e-01 -1.68542489e-01
2.59238124e-01 -4.07517076e-01 -6.39739096e-01 9.89146292e-01
8.14993143e-01 5.74605763e-01 6.56397343e-02 -6.84847593e-01
-4.77816492e-01 1.09988403e+00 -2.66957343e-01 5.58884203e-01
1.84723157e-02 2.64164388e-01 1.12609506e+00 -1.18829477e+00
-9.91353616e-02 1.39519858e+00 2.85993248e-01 6.92921221e-01
-1.02030063e+00 -1.05766535e+00 5.80518901e-01 3.15823257e-01
-1.65173280e+00 -1.12025976e+00 7.71785319e-01 1.74237445e-01
1.54582417e+00 6.40372813e-01 1.48932457e-01 1.06385934e+00
-4.22224969e-01 1.07661796e+00 7.33654320e-01 -3.71855974e-01
2.96067178e-01 -1.19469188e-01 2.04619989e-01 4.96312231e-01
-3.70208681e-01 7.75041059e-02 -6.96517408e-01 1.76297203e-01
8.95840466e-01 -3.16309303e-01 -4.74170119e-01 5.70900738e-01
-1.06663167e+00 3.93912137e-01 4.23887938e-01 4.29604560e-01
-3.45615238e-01 2.46662691e-01 2.01887354e-01 9.34207346e-03
3.24296027e-01 5.43076396e-01 -5.93930662e-01 -4.77374643e-01
-9.11430597e-01 -2.03267887e-01 4.43100154e-01 8.76371503e-01
1.86855704e-01 6.25689924e-01 -3.78813058e-01 1.37079704e+00
5.75887799e-01 4.94686574e-01 8.43746126e-01 -5.34482419e-01
9.06137228e-01 3.33334535e-01 -3.22017580e-01 -7.18598664e-01
-2.17261001e-01 -9.22105551e-01 -8.89685929e-01 -3.27651978e-01
-4.38885912e-02 -7.46830404e-02 -1.02574825e+00 1.87281036e+00
2.27371380e-02 4.58959579e-01 6.58529103e-02 9.70076203e-01
8.86526465e-01 1.03071952e+00 4.86304797e-02 -3.34062099e-01
1.10267985e+00 -1.29970753e+00 -1.00043869e+00 -5.53451896e-01
5.04294753e-01 -6.32240474e-01 1.14491427e+00 3.80611211e-01
-1.35108101e+00 -7.55368710e-01 -1.13642812e+00 -9.09668803e-02
-1.54429212e-01 4.17375654e-01 3.16720128e-01 6.06901765e-01
-9.51118767e-01 2.40067825e-01 -1.07534397e+00 -7.59679973e-02
2.39672840e-01 5.92828989e-01 -9.52310190e-02 1.60655588e-01
-1.21271515e+00 5.66075683e-01 4.93120223e-01 4.78375018e-01
-6.19276285e-01 -6.29643023e-01 -6.91407084e-01 6.16763115e-01
5.92207313e-01 -2.18663275e-01 1.41055191e+00 -5.41309655e-01
-2.15905929e+00 3.22387367e-02 -6.84193075e-01 -3.45500797e-01
-1.21503867e-01 -2.35217452e-01 -5.49065411e-01 -2.87467539e-01
-6.77588403e-01 4.63445753e-01 5.88082612e-01 -7.49123335e-01
-3.91354561e-01 -1.97230563e-01 -3.32118310e-02 3.27783197e-01
-6.54704630e-01 2.18986124e-01 -8.00788701e-01 -1.01441514e+00
2.72841662e-01 -7.44012356e-01 -2.84614623e-01 -6.44359529e-01
-5.00349939e-01 -5.02323985e-01 5.91682315e-01 -8.83327484e-01
2.04911327e+00 -2.46768713e+00 2.42555022e-01 2.94853300e-01
5.56692965e-02 8.82831514e-01 -3.65901738e-01 1.66564927e-01
-1.23398818e-01 2.40566909e-01 -2.34994903e-01 -6.12291217e-01
8.25987756e-02 1.25711128e-01 -2.45381221e-01 -1.66355120e-03
2.07490355e-01 9.40045536e-01 -5.59339881e-01 -1.01603225e-01
2.87003964e-01 4.95101273e-01 -7.49434948e-01 3.45627844e-01
-1.02478988e-01 1.10239305e-01 -3.48019391e-01 4.99209464e-01
6.41308248e-01 -1.94640562e-01 1.23808071e-01 -2.78414488e-01
-1.67625219e-01 8.29047084e-01 -9.73179042e-01 1.63938832e+00
-9.02009606e-01 5.98712325e-01 1.34710670e-01 -7.43245304e-01
9.86017227e-01 4.39498276e-01 -1.64158419e-01 -8.40255737e-01
1.56687692e-01 1.85098171e-01 3.81452560e-01 -2.60710627e-01
3.83614272e-01 3.52067918e-01 3.29093158e-01 2.78861851e-01
-5.83919398e-02 3.15831810e-01 -2.94531524e-01 -2.33274344e-02
1.07792258e+00 -2.52840251e-01 1.96264252e-01 -5.61215961e-03
7.42491245e-01 -7.79712081e-01 5.36929429e-01 6.31694734e-01
-1.31999314e-01 3.67305726e-01 3.27041484e-02 4.78297696e-02
-6.06991827e-01 -9.15656269e-01 -1.01795793e-01 1.23405063e+00
-9.29405168e-02 -6.16838813e-01 -8.40044260e-01 -4.44171637e-01
-4.01370347e-01 1.06479466e+00 -1.04895540e-01 -2.35842168e-01
-8.47384334e-01 -7.37097800e-01 6.89059079e-01 7.67866313e-01
6.52370989e-01 -1.03077519e+00 -7.25988001e-02 3.18582505e-01
-2.57128954e-01 -1.24641097e+00 -8.79979730e-01 8.57301056e-02
-6.51186824e-01 -3.37500811e-01 -4.97459799e-01 -5.78944266e-01
6.55153155e-01 3.01444530e-01 3.91766101e-01 8.66394565e-02
1.58127561e-01 -3.51835907e-01 -3.82147759e-01 1.50518632e-02
-2.65522212e-01 3.44172806e-01 1.62540928e-01 1.13910168e-01
2.99429446e-01 -5.22246480e-01 -3.87792617e-01 3.48205894e-01
-5.83312809e-01 2.42276832e-01 9.70548570e-01 1.04881358e+00
4.73083913e-01 5.90133183e-02 7.85206378e-01 -4.22653943e-01
6.90934300e-01 -3.46885294e-01 -5.13431966e-01 2.95107365e-01
-6.06770337e-01 1.32788777e-01 6.53934062e-01 -6.24845982e-01
-1.31356347e+00 -1.97285846e-01 -6.09387994e-01 -5.72480261e-01
-6.51220903e-02 8.28339636e-01 -4.72955018e-01 1.39751822e-01
1.68876499e-01 5.13676226e-01 -2.30990916e-01 -8.34154725e-01
1.43234268e-01 1.18987274e+00 4.07558411e-01 -2.56340653e-01
5.51924586e-01 -3.26815337e-01 -6.12395585e-01 -1.09541202e+00
-6.11750126e-01 -3.96816432e-01 -5.22504561e-02 1.30465319e-02
6.07647598e-01 -9.31836367e-01 -9.04536486e-01 5.49359798e-01
-1.15743709e+00 -4.68901932e-01 3.96582723e-01 7.80530095e-01
-8.18535835e-02 1.88704833e-01 -5.56057036e-01 -1.16088343e+00
-5.64856291e-01 -1.34490526e+00 8.47056985e-01 2.58012146e-01
-1.61723465e-01 -6.73639536e-01 -5.09734988e-01 6.82902098e-01
7.81863213e-01 -7.12549329e-01 7.08918810e-01 -8.56713474e-01
-6.98688805e-01 9.11395475e-02 -3.52149099e-01 5.92419446e-01
8.09657723e-02 -2.84331888e-01 -1.16108155e+00 -3.62270892e-01
-5.42266704e-02 1.31765470e-01 6.64197803e-01 3.64909708e-01
1.67324126e+00 -5.56041718e-01 -2.60734767e-01 7.00854659e-01
8.22778463e-01 6.76713705e-01 5.64627945e-01 -1.00871570e-01
7.05676973e-01 8.92645493e-02 3.18550140e-01 3.43211561e-01
2.01831967e-01 1.10433626e+00 1.08219840e-01 7.79116452e-02
-4.25457180e-01 -2.70739675e-01 6.86596215e-01 1.72736061e+00
-1.78472325e-02 -4.67295617e-01 -8.51143718e-01 3.06357950e-01
-1.79263616e+00 -7.50454903e-01 2.52498537e-01 2.07379174e+00
9.03723180e-01 3.40528220e-01 -4.20993835e-01 6.00710399e-02
5.81545830e-01 1.26380414e-01 -6.08232856e-01 -5.03903031e-01
1.72099117e-02 3.95762503e-01 3.12513679e-01 8.30993593e-01
-6.97136045e-01 1.14578807e+00 5.46300650e+00 1.38669062e+00
-1.21973240e+00 2.42480978e-01 4.76489991e-01 -2.71035433e-01
-1.92820281e-01 -3.25490922e-01 -1.05977499e+00 6.46837831e-01
1.28160334e+00 -8.95918012e-02 8.57213974e-01 7.04813600e-01
4.80829448e-01 3.55085611e-01 -8.02174807e-01 1.23958576e+00
9.39726606e-02 -1.10286236e+00 -5.92742637e-02 1.80701315e-02
2.30571523e-01 9.99841318e-02 1.30503088e-01 5.94310999e-01
4.28040847e-02 -1.01446915e+00 5.23101628e-01 3.94712687e-01
7.77078927e-01 -8.77628565e-01 6.37167454e-01 4.54521209e-01
-1.21239781e+00 -2.70087779e-01 -9.72939059e-02 4.17511761e-02
1.70103580e-01 2.37827152e-01 -9.93863583e-01 3.29708666e-01
4.20691788e-01 3.06452721e-01 -4.06805873e-01 9.86285508e-01
-3.32950205e-01 1.17397463e+00 -3.35306376e-01 -3.00723076e-01
2.47925520e-01 1.28588721e-01 7.20337451e-01 1.19306409e+00
3.77774775e-01 3.93578529e-01 -5.43856248e-02 7.95705616e-01
-2.67103255e-01 7.75484145e-02 1.56452745e-01 -4.16527122e-01
9.73426580e-01 8.78615558e-01 -1.55140728e-01 -2.02828228e-01
-2.11139888e-01 9.25627410e-01 4.39149559e-01 7.12211549e-01
-9.28369522e-01 -6.71871603e-01 7.05090463e-01 -2.50727296e-01
3.88834804e-01 -4.14246857e-01 -1.81490123e-01 -1.01403940e+00
1.86181348e-02 -1.00461566e+00 -1.69786140e-01 -9.89114344e-01
-7.75799692e-01 9.47857082e-01 -1.03637084e-01 -6.85704350e-01
-1.54581696e-01 -4.43897456e-01 -4.30239111e-01 1.18203557e+00
-1.27567887e+00 -1.00263989e+00 2.78865499e-03 2.84026146e-01
1.22176540e+00 -3.31522912e-01 7.85017908e-01 5.40089786e-01
-1.25426614e+00 1.19119477e+00 -3.44213620e-02 1.04242610e-02
3.06832254e-01 -9.04971302e-01 6.98794782e-01 1.09573078e+00
1.59894109e-01 9.95784461e-01 3.79752606e-01 -3.92285287e-01
-1.43069828e+00 -1.05437183e+00 9.87945676e-01 5.42464107e-02
6.07620716e-01 -4.69388038e-01 -1.16647673e+00 5.61010242e-01
1.45913333e-01 -2.38974482e-01 5.47344506e-01 3.51584911e-01
-4.48999137e-01 -2.39783704e-01 -5.93796551e-01 8.64483476e-01
1.03535283e+00 -7.06527114e-01 -4.83406454e-01 -5.50429113e-02
1.16040778e+00 -5.74131191e-01 -6.39407933e-01 4.00701135e-01
4.00562018e-01 -3.73099953e-01 8.52430940e-01 -2.69602925e-01
-6.19487278e-02 -3.01863313e-01 -3.96001011e-01 -1.42147636e+00
-4.91105288e-01 -8.83551657e-01 -6.98380172e-01 1.26218414e+00
6.69089258e-01 -8.25354695e-01 2.91712254e-01 5.73763907e-01
-6.92228556e-01 -9.71652448e-01 -1.07223129e+00 -8.37543428e-01
-4.26538020e-01 -7.02456355e-01 6.91955984e-01 5.77100873e-01
6.63346276e-02 4.49804604e-01 -1.78386733e-01 4.94394630e-01
1.58271760e-01 -4.03452486e-01 4.09655988e-01 -5.81390262e-01
-6.99055254e-01 -6.79804623e-01 1.85562540e-02 -1.72649467e+00
1.91260710e-01 -8.65600228e-01 2.86944121e-01 -1.15589678e+00
2.06462783e-03 -4.72942412e-01 -7.97546983e-01 6.87301457e-01
-2.67268002e-01 -3.15263718e-01 1.84370369e-01 -6.07662201e-02
-3.76454026e-01 8.19079638e-01 1.12786603e+00 -8.74026716e-02
-3.59838516e-01 1.22262076e-01 -5.45546710e-01 4.14254218e-01
8.95713508e-01 -1.28601894e-01 -5.03576696e-01 -6.89884126e-01
-1.75629333e-01 5.36891297e-02 5.31817637e-02 -1.00128484e+00
5.97862959e-01 3.03166881e-02 -8.51812288e-02 -6.34740829e-01
6.69714272e-01 -4.02797282e-01 -5.45134433e-02 2.55548000e-01
-5.16435027e-01 -1.49698183e-01 5.78230321e-01 4.12551045e-01
-2.70891666e-01 -9.73666757e-02 5.91413915e-01 3.42204958e-01
-5.30702412e-01 1.86829120e-01 -3.36667895e-01 -3.59801292e-01
4.39100176e-01 -2.61973999e-02 -1.94420815e-01 -3.05740207e-01
-5.46682715e-01 2.08689764e-01 -3.03212047e-01 5.76938093e-01
8.45899045e-01 -1.22583282e+00 -5.66033840e-01 5.82278132e-01
-8.56485441e-02 -9.26305354e-02 5.41681051e-01 1.01193058e+00
3.04800514e-02 8.02223563e-01 5.45432031e-01 -2.66115278e-01
-1.48845255e+00 2.02203184e-01 3.36840779e-01 2.67818719e-02
-4.30514067e-01 1.27056277e+00 2.02188581e-01 -2.32923195e-01
6.01675630e-01 -4.98640656e-01 -5.29186195e-03 -3.92774343e-01
7.77426839e-01 3.03476036e-01 1.87861487e-01 -4.81182814e-01
-4.40994114e-01 1.19272240e-01 -4.61348742e-01 -1.65045351e-01
1.12582982e+00 -2.37293020e-01 9.59808975e-02 1.85461059e-01
1.01951480e+00 1.75295472e-01 -1.11400592e+00 -4.17857826e-01
-1.91448703e-01 -3.33702505e-01 6.80040956e-01 -1.01392376e+00
-1.03164542e+00 1.00033593e+00 4.41271037e-01 -2.33666715e-03
1.19036698e+00 -6.35413900e-02 1.07548320e+00 5.80419004e-01
-6.19694218e-03 -1.10303354e+00 -1.61508828e-01 6.88596308e-01
1.12551868e+00 -1.02039063e+00 -5.54758132e-01 -3.40602160e-01
-3.22260112e-01 8.64144623e-01 8.75721633e-01 3.81793529e-01
5.94653130e-01 3.73527944e-01 -1.79741070e-01 2.52256036e-01
-1.00647914e+00 -1.41855791e-01 4.11607355e-01 -6.44211248e-02
4.69535112e-01 1.58041030e-01 -2.05041729e-02 9.91247118e-01
-1.60377160e-01 -3.11636716e-01 8.04461464e-02 4.64794725e-01
-5.51120579e-01 -9.26907301e-01 -2.12001845e-01 1.85775831e-01
-2.54524082e-01 -5.90945601e-01 -8.80720168e-02 2.82306194e-01
-2.10280657e-01 1.21014738e+00 1.30726114e-01 -7.22059846e-01
2.92511582e-01 1.16128311e-01 1.47584096e-01 -5.31867802e-01
-4.39232945e-01 4.94527221e-01 1.02484867e-01 -6.43190920e-01
5.35040013e-02 -3.13777804e-01 -1.30016315e+00 -4.02096324e-02
-7.44898677e-01 1.28210038e-01 8.53269577e-01 1.15297830e+00
6.00227118e-01 1.06136322e+00 6.26766026e-01 -3.94063681e-01
-5.99994481e-01 -1.37573600e+00 -1.47057459e-01 -2.30178460e-01
3.21395814e-01 -5.33729374e-01 -4.11468923e-01 -1.72261089e-01]
|
[14.565415382385254, 6.331838130950928]
|
c20642cc-3b31-4a83-a8ac-aefb685f1000
|
using-deep-cross-modal-hashing-and-error
|
1902.04139
| null |
http://arxiv.org/abs/1902.04139v1
|
http://arxiv.org/pdf/1902.04139v1.pdf
|
Using Deep Cross Modal Hashing and Error Correcting Codes for Improving the Efficiency of Attribute Guided Facial Image Retrieval
|
With benefits of fast query speed and low storage cost, hashing-based image
retrieval approaches have garnered considerable attention from the research
community. In this paper, we propose a novel Error-Corrected Deep Cross Modal
Hashing (CMH-ECC) method which uses a bitmap specifying the presence of certain
facial attributes as an input query to retrieve relevant face images from the
database. In this architecture, we generate compact hash codes using an
end-to-end deep learning module, which effectively captures the inherent
relationships between the face and attribute modality. We also integrate our
deep learning module with forward error correction codes to further reduce the
distance between different modalities of the same subject. Specifically, the
properties of deep hashing and forward error correction codes are exploited to
design a cross modal hashing framework with high retrieval performance.
Experimental results using two standard datasets with facial attributes-image
modalities indicate that our CMH-ECC face image retrieval model outperforms
most of the current attribute-based face image retrieval approaches.
|
['Matthew C. Valenti', 'Nasser M. Nasrabadi', 'Veeru Talreja', 'Fariborz Taherkhani']
|
2019-02-11
| null | null | null | null |
['face-image-retrieval']
|
['computer-vision']
|
[-7.89154544e-02 -2.53237933e-01 -1.77613467e-01 -7.56324053e-01
-1.15094090e+00 -2.17077285e-01 5.13783276e-01 2.62000620e-01
-2.46368214e-01 2.87337691e-01 1.60505250e-01 2.65614718e-01
-2.05339849e-01 -9.35430408e-01 -6.43814027e-01 -9.20485198e-01
-1.17588453e-01 3.92481625e-01 -1.87116444e-01 -2.38096267e-02
2.93550611e-01 4.28008378e-01 -1.89911103e+00 2.81114370e-01
1.48079365e-01 1.43051994e+00 -3.23669344e-01 1.14007793e-01
2.69465476e-01 3.85631025e-01 -1.60874754e-01 -7.79844880e-01
2.65038878e-01 -2.08797798e-01 -6.73822522e-01 -5.12933552e-01
7.01608717e-01 -8.23835731e-01 -8.82138014e-01 7.55932033e-01
1.09509015e+00 -2.66877055e-01 6.40626132e-01 -1.52144969e+00
-1.12620068e+00 3.56733322e-01 -5.64713776e-01 -1.00412928e-01
5.55841863e-01 -5.23025207e-02 1.02400696e+00 -1.20824516e+00
4.46465790e-01 1.51764321e+00 7.04037368e-01 7.16804802e-01
-9.93294120e-01 -9.77319539e-01 -7.98382580e-01 6.67752862e-01
-2.14488530e+00 -5.90791285e-01 7.37770796e-01 -3.00496127e-02
5.09926856e-01 -4.14779522e-02 4.86400306e-01 6.97434604e-01
2.29076564e-01 8.28628600e-01 7.20679104e-01 -1.81146696e-01
1.89413026e-01 -8.60592071e-03 -2.45045736e-01 8.58061016e-01
1.26604382e-02 1.75071329e-01 -9.17094111e-01 -5.37311971e-01
2.62372971e-01 4.07377481e-01 -3.80913205e-02 -5.16010404e-01
-7.86606073e-01 1.05753052e+00 7.74417341e-01 -5.90813495e-02
-4.57219630e-01 4.23742861e-01 5.56378663e-01 3.04055959e-01
6.45903349e-02 -3.73587281e-01 1.01807453e-01 2.91821063e-01
-9.99431849e-01 4.08614099e-01 5.69961190e-01 8.57953370e-01
1.01120782e+00 -5.44053733e-01 -2.68580854e-01 8.17256391e-01
6.45303726e-01 8.63883853e-01 4.01015311e-01 -8.60196054e-01
-4.27912511e-02 6.03112221e-01 -4.18581694e-01 -1.38089728e+00
1.67212784e-01 2.04472616e-01 -8.48809540e-01 -2.30504975e-01
-2.18251899e-01 5.83232820e-01 -6.34637117e-01 1.94853914e+00
5.69301248e-01 1.21810041e-01 3.02604120e-02 8.51072609e-01
8.27707767e-01 6.47482336e-01 2.46365234e-01 1.50808215e-01
1.51468408e+00 -4.50177103e-01 -6.15915239e-01 2.88851976e-01
5.59233367e-01 -7.67152190e-01 7.24245787e-01 -1.71314701e-01
-1.14872062e+00 -5.62482238e-01 -1.04764843e+00 -5.30285358e-01
-5.75552821e-01 5.03608435e-02 4.49813068e-01 6.94689512e-01
-1.35044253e+00 9.91576165e-02 -3.85375589e-01 -1.42948478e-01
5.83182812e-01 7.08616793e-01 -7.56705344e-01 -6.30867302e-01
-1.40267050e+00 5.04570663e-01 3.11307013e-01 -8.94839410e-04
-9.99999046e-01 -6.54572308e-01 -1.05606186e+00 3.00585359e-01
-3.05946916e-01 -4.48164970e-01 8.79187405e-01 -3.92964154e-01
-8.34682107e-01 1.19735885e+00 -2.38111809e-01 -2.17171103e-01
-6.62267581e-02 -8.48084465e-02 -4.04954463e-01 6.58139944e-01
4.94694598e-02 1.28150856e+00 1.13983417e+00 -9.46591914e-01
-2.48861998e-01 -9.21089768e-01 -3.64199251e-01 -2.06122007e-02
-7.17262566e-01 6.94992915e-02 -5.33997178e-01 -4.32604223e-01
-1.03598563e-02 -1.01113141e+00 6.26229465e-01 6.80704176e-01
-7.70603791e-02 -6.50414705e-01 1.28784120e+00 -4.76341426e-01
1.25152314e+00 -2.57981205e+00 5.32787666e-02 3.00562918e-01
1.06575049e-01 1.77874565e-01 -3.44017446e-01 6.14305198e-01
7.43367523e-02 -2.20962450e-01 -1.20830014e-01 -5.81685424e-01
2.75341749e-01 1.87853038e-01 -2.98466086e-01 6.76355064e-01
3.91414672e-01 9.48088825e-01 -5.47546864e-01 -9.80732262e-01
-1.44038185e-01 1.23872972e+00 -6.37402594e-01 3.44531059e-01
3.00413311e-01 -3.12140256e-01 -2.44500190e-01 1.10842943e+00
1.10456395e+00 -1.00533813e-01 -3.58349942e-02 -3.59671265e-01
4.42317009e-01 -3.76827223e-03 -8.34522247e-01 1.67228246e+00
-1.75562888e-01 4.70092237e-01 8.27537030e-02 -5.91780007e-01
9.68131542e-01 3.65160704e-01 4.83085901e-01 -9.78002250e-01
1.57764003e-01 3.68416607e-01 -5.85804105e-01 -2.22917601e-01
5.20463109e-01 -4.66602072e-02 -1.15126073e-01 6.01634443e-01
1.42627850e-01 2.46984839e-01 -2.67529815e-01 2.63062388e-01
7.33171940e-01 -3.30958843e-01 -1.19227722e-01 -1.55917099e-02
8.65777671e-01 -5.30872107e-01 3.13871086e-01 3.69019479e-01
-4.12805676e-01 5.92246056e-01 4.28126380e-02 -6.13618970e-01
-1.09513271e+00 -9.63665426e-01 -5.15923738e-01 1.21682620e+00
2.95172513e-01 -5.78550518e-01 -8.42610061e-01 -3.96788359e-01
4.39660937e-01 -3.05745244e-01 -8.38813424e-01 -7.33074665e-01
-4.89413470e-01 -3.16391766e-01 9.54240799e-01 4.74716991e-01
8.05345416e-01 -1.11930239e+00 -4.75654036e-01 -1.45295173e-01
-3.03237379e-01 -9.47164178e-01 -8.08420718e-01 -3.62997800e-01
-4.55493003e-01 -9.59789753e-01 -8.41107547e-01 -1.22689545e+00
3.88685733e-01 4.25613880e-01 8.12932909e-01 3.64652365e-01
-8.08138907e-01 7.11886466e-01 -2.20411718e-01 4.11191881e-02
-1.71276659e-01 9.60751474e-02 1.28504559e-01 3.11723202e-01
9.85576570e-01 -2.31003091e-01 -1.11607695e+00 3.16716075e-01
-1.32027376e+00 -6.13459408e-01 7.55242586e-01 8.81544411e-01
6.13680601e-01 -1.66366845e-01 4.95010734e-01 -1.11698247e-01
3.30543220e-01 -6.19720638e-01 -3.42316419e-01 4.46709424e-01
-5.74849308e-01 2.18916982e-01 -5.55433938e-03 -3.61157715e-01
-4.13949251e-01 1.72734872e-01 -1.56661406e-01 -5.09020567e-01
1.53577253e-01 2.63324142e-01 -3.09921890e-01 -4.83053595e-01
1.66848600e-01 6.55499935e-01 4.05201107e-01 -1.83016762e-01
2.42551073e-01 9.34368014e-01 5.85551977e-01 -5.06015003e-01
6.79899335e-01 5.63616991e-01 3.02531958e-01 -5.17028749e-01
-3.07287335e-01 -3.63704473e-01 -4.39026356e-01 -1.42161667e-01
7.84101903e-01 -1.23331463e+00 -1.17331350e+00 6.44607723e-01
-1.07087648e+00 4.10618514e-01 1.97171211e-01 2.24160865e-01
-5.77947795e-01 4.18188602e-01 -7.40513265e-01 -6.04639947e-01
-8.42986286e-01 -1.22865343e+00 1.77786243e+00 2.34694734e-01
2.17998132e-01 -4.44731027e-01 4.68583182e-02 3.75229061e-01
5.25049269e-01 -1.26315638e-01 1.08093715e+00 -4.45284992e-01
-8.07126760e-01 -3.88400733e-01 -5.71461499e-01 1.45271882e-01
-1.16804123e-01 -2.76165217e-01 -1.04237509e+00 -6.84462786e-01
-4.21998024e-01 -9.20226097e-01 7.71476805e-01 -7.99510702e-02
1.37369800e+00 -4.25764352e-01 -3.08630466e-01 6.14763081e-01
1.60475254e+00 -5.55379353e-02 9.29809332e-01 1.41035110e-01
4.02813494e-01 6.56942308e-01 5.25945425e-01 8.49289298e-01
7.17089474e-01 6.82737947e-01 4.95623291e-01 1.34835258e-01
-5.78090586e-02 -4.57079858e-01 2.10254520e-01 5.65082848e-01
6.37248278e-01 -2.51920410e-02 -6.50275826e-01 7.28664339e-01
-1.42909014e+00 -1.10512519e+00 5.71155190e-01 2.09346628e+00
1.09818625e+00 -6.18775547e-01 -6.71041664e-03 2.55232483e-01
7.88180113e-01 7.58178085e-02 -5.06549895e-01 -1.50142670e-01
-1.21121690e-01 2.87419438e-01 9.66711491e-02 1.72687814e-01
-1.12157011e+00 7.04754233e-01 5.88251877e+00 8.03569973e-01
-1.04691601e+00 -6.68928027e-02 5.76769590e-01 2.89847758e-02
-2.06880942e-01 -2.94950038e-01 -1.12689567e+00 5.74720204e-01
9.00711298e-01 -2.30210833e-02 4.03463483e-01 8.33277345e-01
-4.87208247e-01 2.50817806e-01 -1.25665474e+00 1.43951356e+00
5.05929768e-01 -1.00707948e+00 4.80668366e-01 4.14185524e-01
2.57209301e-01 -3.13999116e-01 6.35964811e-01 2.69816607e-01
-2.79355407e-01 -1.03217089e+00 4.60251927e-01 4.44216371e-01
1.02556276e+00 -9.26633894e-01 6.42437518e-01 -2.92172849e-01
-1.38343287e+00 -2.84788549e-01 -5.44521749e-01 6.86251462e-01
-2.93818921e-01 1.40845031e-01 -7.80555248e-01 2.17053548e-01
1.03666902e+00 5.16598880e-01 -7.08835959e-01 8.97228658e-01
4.52464312e-01 1.92943402e-02 -4.12375897e-01 3.09453785e-01
1.16964929e-01 2.59072274e-01 -5.97242229e-02 1.01804948e+00
5.46727180e-01 2.02344790e-01 -2.50134766e-01 6.57204747e-01
-5.22147417e-01 1.51345342e-01 -9.56951320e-01 -2.89684255e-03
9.83368397e-01 1.11886609e+00 6.22411445e-02 -1.02174245e-01
-3.50008667e-01 1.03102016e+00 2.20491678e-01 -5.62819615e-02
-7.65801549e-01 -6.06459379e-01 8.69917989e-01 8.69049206e-02
5.69824755e-01 5.75699136e-02 3.11095953e-01 -6.50326133e-01
8.46365169e-02 -7.96820164e-01 6.67248428e-01 -6.99262738e-01
-1.19734299e+00 4.72931325e-01 -1.16479777e-01 -9.87540305e-01
-1.68445379e-01 -3.10338557e-01 6.05834695e-03 7.50843644e-01
-1.93907189e+00 -1.54337287e+00 -4.47846413e-01 1.11024523e+00
-8.31670389e-02 -3.32461596e-01 1.01904881e+00 8.00983727e-01
-2.69793153e-01 1.28085268e+00 8.88645723e-02 3.45523417e-01
1.12945557e+00 -6.27556443e-01 2.21560821e-02 1.10233635e-01
-2.05133200e-01 9.65791345e-01 9.61166248e-02 -3.02006781e-01
-2.04839778e+00 -1.05784166e+00 1.14847767e+00 2.51969155e-02
1.70748383e-01 -2.92296618e-01 -1.09215128e+00 2.84568042e-01
2.98695236e-01 3.68879169e-01 1.15348387e+00 -5.27615666e-01
-1.06976414e+00 -7.56501615e-01 -1.40477645e+00 1.06266312e-01
6.14958286e-01 -1.28708148e+00 -2.26950437e-01 9.20260102e-02
5.45322299e-01 8.59567747e-02 -1.14362097e+00 5.58480382e-01
1.02649379e+00 -7.57402837e-01 1.30852556e+00 -2.22003892e-01
3.34644020e-01 -1.65424362e-01 -5.87391734e-01 -5.30252457e-01
-2.45019376e-01 -2.32783183e-01 -2.41306245e-01 1.27810919e+00
-9.63904262e-02 -3.34710687e-01 6.78162634e-01 6.20802879e-01
4.81546998e-01 -7.08319008e-01 -1.23958540e+00 -4.26459253e-01
-2.19549667e-02 3.02091032e-01 1.11016679e+00 7.33367443e-01
-1.82633027e-01 -3.64200510e-02 -4.49684829e-01 2.44457960e-01
9.99149740e-01 2.81457156e-01 4.98575479e-01 -1.29776216e+00
1.75263926e-01 -1.91596910e-01 -8.52832615e-01 -6.70984387e-01
4.62367952e-01 -8.76318455e-01 -3.86697985e-03 -7.88621902e-01
6.64430737e-01 -4.77639884e-01 -4.08162922e-01 7.42893636e-01
3.16815898e-02 8.51490617e-01 -5.17509058e-02 3.90565038e-01
-7.59341955e-01 1.04056203e+00 7.47616529e-01 -4.22120839e-01
4.60293353e-01 -6.45586789e-01 -5.51971614e-01 -1.46404028e-01
5.13365448e-01 -5.08834898e-01 -3.03787977e-01 -6.63236916e-01
1.31801128e-01 1.57790497e-01 6.63619816e-01 -1.02985930e+00
6.31040812e-01 4.62146670e-01 5.88136315e-01 -6.51514828e-01
7.27146685e-01 -1.10190082e+00 -5.49305752e-02 5.24773002e-01
-5.93651772e-01 4.55386668e-01 3.17981765e-02 5.47100067e-01
-5.11464179e-01 1.01163931e-01 1.00206864e+00 2.58248538e-01
-6.50346160e-01 7.86725044e-01 5.96449077e-02 -2.94164747e-01
1.07462776e+00 -1.39433116e-01 -2.77315080e-01 -3.83607864e-01
-3.17451358e-01 1.97162554e-01 4.01455790e-01 4.16724414e-01
1.29729867e+00 -1.86306381e+00 -6.87668383e-01 6.78214431e-01
6.29852831e-01 -5.12443602e-01 2.95770526e-01 2.58797377e-01
-3.55101764e-01 5.48112333e-01 -4.46639150e-01 -5.88272512e-01
-1.52432513e+00 6.55069172e-01 1.32750869e-01 3.45395863e-01
-1.76624119e-01 7.13849247e-01 -1.45515919e-01 -3.10646027e-01
4.83166367e-01 5.78212321e-01 1.36608690e-01 1.41605884e-01
9.15675759e-01 1.48672581e-01 1.08010560e-01 -1.05112314e+00
-5.32058060e-01 8.31665635e-01 -3.73144895e-01 3.48486304e-02
1.18030500e+00 -2.30504334e-01 -4.50197160e-01 -2.48848245e-01
2.06622672e+00 -6.27371311e-01 -6.93608165e-01 -5.79084873e-01
-2.16625661e-01 -7.01506197e-01 1.25505298e-01 -4.37817872e-01
-1.33008540e+00 1.11058831e+00 1.26497066e+00 -3.08413208e-01
1.44346344e+00 1.31414875e-01 1.34474421e+00 3.78585190e-01
4.14879501e-01 -8.42595756e-01 2.18579009e-01 1.30122751e-01
6.51710808e-01 -1.41909897e+00 -1.26584053e-01 -2.18551308e-02
-1.83975145e-01 9.29175496e-01 5.69201171e-01 1.48467690e-01
8.89444888e-01 -2.43112780e-02 -1.63553268e-01 -4.34354454e-01
-7.16601193e-01 5.57060130e-02 1.25416726e-01 4.69018191e-01
1.97181910e-01 -2.46109635e-01 -2.22392723e-01 -3.32482494e-02
9.87514779e-02 1.16932625e-02 -2.84687132e-01 1.17570841e+00
-4.14076865e-01 -1.24126041e+00 -2.86925495e-01 -5.87317571e-02
-4.40853924e-01 -1.16547771e-01 -3.33508730e-01 4.24118191e-01
-1.25479758e-01 5.92774630e-01 1.18758067e-01 -4.85229999e-01
-8.65592211e-02 2.44495258e-01 4.47553039e-01 -2.60234922e-02
-3.74503076e-01 -3.66973191e-01 -8.19980204e-01 -6.44963622e-01
-3.96335304e-01 -4.78347540e-01 -1.06429672e+00 -7.59657979e-01
-1.24077462e-01 1.93190768e-01 8.73197734e-01 4.02611047e-01
9.39248979e-01 -4.88324314e-01 9.76605237e-01 -5.52056313e-01
-7.12241411e-01 -5.68002820e-01 -5.82465768e-01 5.90619624e-01
6.68317616e-01 -6.19384944e-01 -1.88604817e-01 -5.30860620e-03]
|
[11.457079887390137, 0.9090246558189392]
|
c51c39b5-ccae-4a56-a4dd-d361d6894d2f
|
zero-shot-end-to-end-spoken-language
|
2305.12793
| null |
https://arxiv.org/abs/2305.12793v1
|
https://arxiv.org/pdf/2305.12793v1.pdf
|
Zero-Shot End-to-End Spoken Language Understanding via Cross-Modal Selective Self-Training
|
End-to-end (E2E) spoken language understanding (SLU) is constrained by the cost of collecting speech-semantics pairs, especially when label domains change. Hence, we explore \textit{zero-shot} E2E SLU, which learns E2E SLU without speech-semantics pairs, instead using only speech-text and text-semantics pairs. Previous work achieved zero-shot by pseudolabeling all speech-text transcripts with a natural language understanding (NLU) model learned on text-semantics corpora. However, this method requires the domains of speech-text and text-semantics to match, which often mismatch due to separate collections. Furthermore, using the entire speech-text corpus from any domains leads to \textit{imbalance} and \textit{noise} issues. To address these, we propose \textit{cross-modal selective self-training} (CMSST). CMSST tackles imbalance by clustering in a joint space of the three modalities (speech, text, and semantics) and handles label noise with a selection network. We also introduce two benchmarks for zero-shot E2E SLU, covering matched and found speech (mismatched) settings. Experiments show that CMSST improves performance in both two settings, with significantly reduced sample sizes and training time.
|
['Jinglun Cai', 'Haoqi Li', 'Kaisheng Yao', 'Julian Salazar', 'Jianfeng He']
|
2023-05-22
| null | null | null | null |
['spoken-language-understanding', 'spoken-language-understanding']
|
['natural-language-processing', 'speech']
|
[ 5.62041044e-01 3.70561153e-01 -1.94024235e-01 -6.70829773e-01
-1.58487368e+00 -5.07193387e-01 4.60087836e-01 2.61484161e-02
-4.39170688e-01 4.77502078e-01 5.74184120e-01 4.35826145e-02
2.15546833e-03 -2.08840206e-01 -7.18890429e-01 -4.44570273e-01
2.44159460e-01 9.16937470e-01 5.33198528e-02 -1.11894906e-01
-3.00266206e-01 -5.33729434e-01 -1.87029982e+00 5.42803824e-01
8.77032399e-01 1.11263502e+00 3.53586465e-01 5.56149304e-01
-5.86946607e-01 7.98309088e-01 -4.83222485e-01 -2.42065266e-01
1.55687809e-01 -8.40123057e-01 -9.35838342e-01 4.11636680e-01
5.21266043e-01 -7.79376253e-02 -2.41215765e-01 9.58558142e-01
8.52494001e-01 5.28417230e-01 6.95223808e-01 -1.59803045e+00
-2.27626011e-01 7.99334586e-01 -4.00676131e-01 -1.63053423e-01
5.18215775e-01 -4.40343134e-02 1.11201811e+00 -1.10438120e+00
8.75821710e-01 1.54144514e+00 6.09796822e-01 9.32532072e-01
-1.27440631e+00 -7.57701755e-01 1.22315645e-01 1.45079523e-01
-1.22462559e+00 -1.01971900e+00 3.50633144e-01 -1.20459616e-01
1.09246826e+00 3.62189323e-01 -8.16378221e-02 1.44272959e+00
-7.06161737e-01 1.35485756e+00 9.67605889e-01 -5.64783752e-01
5.94385624e-01 2.37343341e-01 4.30773377e-01 2.73951530e-01
-5.45557141e-01 -3.37784201e-01 -9.74766850e-01 -2.47377753e-02
-1.23402238e-01 -2.56291717e-01 -4.52276051e-01 -2.03416228e-01
-1.13519394e+00 6.50217056e-01 -2.04359293e-01 1.90892756e-01
-1.11102248e-02 -3.38184118e-01 8.51804554e-01 8.02403152e-01
7.07346976e-01 4.78399619e-02 -6.85638666e-01 -4.57375318e-01
-9.34149325e-01 -1.02314748e-01 9.73541379e-01 1.34040558e+00
8.88284206e-01 -1.63352802e-01 -1.93261042e-01 1.50441062e+00
1.22005790e-01 6.66375935e-01 8.56477141e-01 -8.45150292e-01
8.41221511e-01 2.90390432e-01 -3.01149487e-01 -2.33471379e-01
-4.23517525e-01 1.74700413e-02 -8.32833350e-01 -4.00917858e-01
3.16501498e-01 -2.84411460e-01 -1.06230724e+00 2.13262868e+00
4.03619915e-01 3.68557900e-01 5.94191134e-01 8.96060705e-01
1.19707620e+00 7.97536254e-01 2.61507668e-02 -5.50213695e-01
1.29889214e+00 -1.08522213e+00 -9.70706522e-01 -6.16221011e-01
9.62179303e-01 -5.40647089e-01 1.57304251e+00 5.65899760e-02
-8.81712973e-01 -5.61283380e-02 -6.78782880e-01 -7.06616342e-02
-2.47606412e-01 -6.47310987e-02 -1.94309831e-01 4.96669620e-01
-8.96198690e-01 2.19302222e-01 -4.90193576e-01 -6.68617249e-01
2.18985513e-01 1.31626859e-01 -3.73543471e-01 -3.28041792e-01
-1.60385334e+00 5.21733284e-01 4.55168277e-01 -4.57724661e-01
-8.22337627e-01 -7.28520811e-01 -1.11774361e+00 6.26543760e-02
1.01480377e+00 -3.28328788e-01 1.45949447e+00 -1.06347251e+00
-1.52114344e+00 1.00925994e+00 -3.99676353e-01 -3.10655385e-01
5.23155034e-01 1.73726454e-01 -4.21798408e-01 -8.53825659e-02
1.90544009e-01 7.09865868e-01 6.44473016e-01 -1.23903596e+00
-5.86373866e-01 -6.43973887e-01 -6.07834160e-01 5.32987535e-01
-4.68924373e-01 -3.55883241e-02 -6.47652805e-01 -5.34439564e-01
3.61085534e-01 -7.64061749e-01 3.06957662e-01 -4.19617325e-01
-5.19637883e-01 -5.94384313e-01 8.38554859e-01 -6.24376297e-01
1.23312330e+00 -2.43142533e+00 1.28232449e-01 -3.89567018e-03
6.25826716e-02 4.39199209e-02 -5.29483020e-01 5.23725212e-01
-2.39000507e-02 -4.23421413e-02 -4.44492519e-01 -1.01291239e+00
4.38575536e-01 5.57614684e-01 -3.53010625e-01 2.04951122e-01
-2.11521894e-01 6.19160116e-01 -8.73642027e-01 -5.23560107e-01
7.51623362e-02 -9.28232446e-02 -4.72674876e-01 3.26195806e-01
-4.63731140e-01 4.15101290e-01 -2.44971067e-01 6.31248176e-01
6.14277959e-01 -1.78551570e-01 4.40185040e-01 -4.22769934e-02
3.62420738e-01 4.59219098e-01 -1.20326483e+00 2.03121734e+00
-6.54011786e-01 4.14857984e-01 2.46315658e-01 -1.09108377e+00
7.41097569e-01 5.05910218e-01 4.86636579e-01 -8.71810973e-01
2.79520422e-01 4.70093071e-01 -3.01897943e-01 -5.96155882e-01
1.48565292e-01 -4.76363063e-01 -4.97941822e-01 8.24685752e-01
6.96879864e-01 -2.24117130e-01 8.22873339e-02 3.96407574e-01
1.17319047e+00 -3.55162621e-01 1.39679641e-01 4.94454801e-02
1.66822281e-02 -2.28035733e-01 6.11765742e-01 8.24544132e-01
-5.13771176e-01 8.01360905e-01 5.66275001e-01 2.22305894e-01
-9.34422493e-01 -1.02515018e+00 -7.02316612e-02 1.52687180e+00
1.01971470e-01 -4.15680319e-01 -1.04583943e+00 -8.47714126e-01
-2.75163561e-01 1.19429028e+00 -3.91615391e-01 -2.85409182e-01
-1.12432659e-01 -5.50899804e-01 7.63671219e-01 1.13212824e-01
2.39762709e-01 -1.01907837e+00 -1.53498426e-01 1.00751758e-01
-7.52773225e-01 -1.46542203e+00 -7.00815976e-01 5.32356024e-01
-3.60141814e-01 -7.07551003e-01 -6.24721169e-01 -6.69592977e-01
3.03527892e-01 2.07977086e-01 1.20268571e+00 -4.20687586e-01
-4.54747789e-02 5.40999293e-01 -8.45640182e-01 -3.73880565e-01
-8.21673453e-01 9.39094480e-06 2.99939096e-01 1.38519287e-01
6.63499534e-01 -4.59578395e-01 -4.55073677e-02 5.30722439e-01
-1.03335583e+00 8.96739736e-02 1.26245826e-01 1.03659725e+00
4.89802480e-01 -2.06510156e-01 9.50882077e-01 -6.65533245e-01
4.06594604e-01 -7.32226551e-01 6.24952721e-04 5.97163975e-01
-2.51655161e-01 -6.79267058e-03 5.47782421e-01 -3.58348012e-01
-9.93352473e-01 -1.52202845e-01 -1.79204240e-01 -7.80227423e-01
-4.12532866e-01 2.64430612e-01 -6.14122689e-01 6.38119102e-01
5.74830532e-01 3.53722543e-01 2.52686918e-01 -5.13944268e-01
4.18295830e-01 1.37711716e+00 4.92499709e-01 -4.34493721e-01
-1.11356862e-02 1.87371522e-01 -8.80742311e-01 -1.16799068e+00
-1.11818814e+00 -8.75885606e-01 -3.71265948e-01 -1.12869009e-01
6.49951220e-01 -9.98242438e-01 -4.48605269e-01 5.17839730e-01
-9.44606721e-01 -5.59307098e-01 -4.85063165e-01 4.81188476e-01
-8.93682659e-01 4.00159180e-01 -3.53312850e-01 -1.07142198e+00
-1.91458449e-01 -1.21380866e+00 1.52280903e+00 -2.66175777e-01
-4.88812298e-01 -6.64176702e-01 -1.79640129e-01 6.53566480e-01
-8.76794383e-02 -2.46158019e-01 7.88128674e-01 -1.36657989e+00
-3.59676182e-02 1.53385833e-01 -2.18921721e-01 4.09846723e-01
-2.15325668e-03 -7.24242687e-01 -1.40199649e+00 -4.16498363e-01
1.72085732e-01 -1.14180171e+00 9.02012706e-01 3.17115188e-01
8.96603346e-01 -4.15471196e-01 -1.16211474e-01 2.78202891e-01
9.91364300e-01 1.06371559e-01 2.94402063e-01 1.00008177e-03
5.35010874e-01 1.06319988e+00 6.49260700e-01 6.96341157e-01
7.35126257e-01 7.66335309e-01 4.23413999e-02 4.98824492e-02
-3.13483298e-01 -2.99015462e-01 6.13454700e-01 1.05900085e+00
9.23096061e-01 -7.23843813e-01 -1.09355927e+00 8.05227757e-01
-1.99340093e+00 -7.42393196e-01 2.79316336e-01 2.08999538e+00
1.10207260e+00 -3.17944676e-01 6.02261275e-02 1.42189220e-01
7.59974003e-01 9.15303826e-02 -9.02212858e-01 2.53958050e-02
-2.74553806e-01 -1.69932842e-01 2.17224747e-01 4.15942758e-01
-1.02023506e+00 1.11798620e+00 5.07842255e+00 1.41723824e+00
-7.60459006e-01 7.21641123e-01 7.35125661e-01 -5.41538119e-01
-5.81064641e-01 -3.11747551e-01 -6.68328524e-01 6.36614919e-01
1.04134285e+00 1.77670997e-02 4.66267705e-01 5.13792574e-01
1.38378898e-02 -2.67754108e-01 -1.26147532e+00 1.20759833e+00
3.57843816e-01 -7.68801332e-01 -1.74807623e-01 -3.44620645e-01
6.58060730e-01 4.00192946e-01 -1.48700595e-01 5.72192252e-01
5.05172372e-01 -6.87737226e-01 8.39582384e-01 5.42412736e-02
1.33396447e+00 -5.47392845e-01 5.70656359e-01 8.09273899e-01
-8.57146561e-01 -1.07065663e-01 -1.26033872e-01 4.05941516e-01
3.13076377e-01 5.54587781e-01 -9.25848067e-01 5.19821882e-01
8.45207751e-01 5.30447602e-01 1.54382214e-01 1.07451439e-01
2.60272294e-01 5.96340358e-01 -5.66213727e-01 -4.21920121e-02
4.26097572e-01 -5.34204580e-02 8.35675120e-01 1.18191624e+00
3.62468511e-01 1.37383059e-01 5.39390266e-01 5.82796693e-01
-3.11378062e-01 2.72245616e-01 -6.34883046e-01 -1.78066120e-01
8.00577283e-01 5.93789697e-01 -6.65231586e-01 -5.86428583e-01
-4.15811926e-01 1.22685218e+00 2.51364082e-01 3.45860720e-01
-4.74040717e-01 -1.04597129e-01 8.22851717e-01 -2.00637668e-01
-1.78736057e-02 2.67025977e-01 -2.05105305e-01 -1.44336641e+00
8.10696185e-02 -1.10647345e+00 5.99352062e-01 -5.24619222e-01
-1.50995553e+00 5.63976884e-01 -4.43035215e-02 -1.20443761e+00
-4.64127839e-01 -5.62595539e-02 -3.97925191e-02 4.84700471e-01
-1.23102605e+00 -9.21246052e-01 1.88714750e-02 6.25590742e-01
1.25101268e+00 -4.11656946e-01 8.26040983e-01 4.79475737e-01
-6.14177406e-01 9.03264165e-01 3.30291420e-01 -1.51813433e-01
1.08578634e+00 -1.12197733e+00 3.56842011e-01 4.74044055e-01
1.25624672e-01 9.65066776e-02 7.42391646e-01 -6.32946730e-01
-1.19809926e+00 -1.19484389e+00 1.18105042e+00 -4.27786708e-01
7.22569346e-01 -7.70639956e-01 -9.81329501e-01 5.87743819e-01
1.32767946e-01 -9.04980674e-02 8.87259007e-01 1.69784769e-01
-2.49201208e-01 1.01781599e-01 -1.11354673e+00 5.43905497e-01
1.31343114e+00 -9.12800848e-01 -6.03974998e-01 4.41969305e-01
1.18881512e+00 -3.46939147e-01 -6.35232449e-01 3.79743397e-01
1.27135888e-01 -8.98590207e-01 6.27102494e-01 -5.51524162e-01
1.99069768e-01 4.30157892e-02 -5.50840437e-01 -1.58616829e+00
5.36785960e-01 -5.35197914e-01 1.72591969e-01 1.39398062e+00
4.86162096e-01 -3.98434967e-01 4.07117933e-01 4.95299041e-01
-3.49394947e-01 -4.18692738e-01 -1.42836058e+00 -7.86399841e-01
-2.60810822e-01 -9.22384262e-01 5.13016224e-01 1.15489566e+00
3.64792824e-01 6.72270179e-01 -5.35408974e-01 -1.39648691e-01
5.46644747e-01 -1.96915999e-01 6.75361395e-01 -9.27200258e-01
-1.32664755e-01 -2.96872735e-01 -3.34452465e-02 -1.09645545e+00
8.06765199e-01 -1.21755195e+00 5.07262290e-01 -1.24265194e+00
2.28449762e-01 -3.48396927e-01 4.79515493e-02 7.19982624e-01
-1.48722708e-01 -7.18511790e-02 3.00100565e-01 2.78920740e-01
-9.63202596e-01 1.06625962e+00 8.98257494e-01 -1.41826183e-01
-2.23473206e-01 -1.34564757e-01 -3.76801789e-01 5.99777281e-01
4.39265311e-01 -4.33549434e-01 -7.23004818e-01 -2.45299563e-01
-2.02522501e-01 3.21059346e-01 -4.85219620e-02 -7.57310152e-01
3.46520066e-01 2.80034989e-01 -3.29124421e-01 -6.64144337e-01
3.91927570e-01 -8.18736434e-01 -3.04563671e-01 -1.20536208e-01
-6.17806852e-01 -5.39116561e-01 1.45667806e-01 7.09836364e-01
-5.16909480e-01 -3.92082632e-01 7.63314068e-01 -1.44603914e-02
-8.44148576e-01 1.39442220e-01 -3.34971905e-01 5.69854736e-01
8.77257407e-01 -1.65001348e-01 -1.58363000e-01 -7.23731279e-01
-1.03315687e+00 7.29531109e-01 1.87187269e-01 5.77807665e-01
4.31633264e-01 -1.17451310e+00 -6.63890421e-01 3.55869234e-01
5.79278708e-01 2.57184446e-01 8.22956324e-01 9.36319947e-01
4.07699764e-01 1.73436731e-01 4.00916487e-01 -8.93392503e-01
-1.25753748e+00 1.76466733e-01 2.72239745e-01 -1.62763312e-01
-4.01461333e-01 1.04207551e+00 4.90561903e-01 -1.38422561e+00
7.40414500e-01 -9.47410986e-02 7.09008276e-02 6.05578661e-01
4.77778882e-01 4.27830398e-01 1.97407380e-01 -6.03985846e-01
-2.99418420e-01 2.94439048e-01 1.39513463e-01 -4.83928680e-01
1.11679471e+00 -7.37769127e-01 8.42243358e-02 9.84919190e-01
1.47908437e+00 -4.79616344e-01 -1.15101182e+00 -8.13232601e-01
2.04635248e-01 -1.11002035e-01 4.72408421e-02 -9.15364444e-01
-6.86116815e-01 7.72789896e-01 5.60795963e-01 1.57492027e-01
1.06139100e+00 4.18451101e-01 1.23226917e+00 4.58714843e-01
2.79869884e-01 -1.54738212e+00 8.76723677e-02 9.14136350e-01
7.91784227e-01 -1.66914749e+00 -7.99823403e-01 -2.85480142e-01
-1.17116666e+00 5.71952522e-01 6.92969084e-01 7.08198249e-01
5.48209131e-01 2.86503792e-01 3.81933063e-01 -1.18991211e-01
-9.39191997e-01 -6.29716277e-01 9.78351161e-02 3.92272741e-01
2.04381630e-01 2.01958120e-01 3.92293558e-02 8.15784991e-01
-2.17298999e-01 -1.65928766e-01 1.06774360e-01 7.23348498e-01
-4.14439261e-01 -1.06159723e+00 -3.72641087e-01 5.50671101e-01
-6.27631024e-02 -2.15667605e-01 -4.71929520e-01 3.61600220e-01
-3.50433327e-02 1.28385115e+00 3.69458079e-01 -3.59531730e-01
5.06521106e-01 5.21303356e-01 -2.36171782e-02 -7.45503008e-01
-3.20840538e-01 5.76402545e-01 3.55130941e-01 -5.74467838e-01
-3.15683395e-01 -7.21568346e-01 -1.41170025e+00 4.88761999e-02
-3.42619419e-01 2.48944253e-01 6.12394392e-01 1.16031349e+00
5.08335412e-01 5.72221696e-01 8.20958674e-01 -5.11251807e-01
-8.76560509e-01 -1.29942453e+00 -8.82984757e-01 7.17592597e-01
3.99476975e-01 -5.62411845e-01 -5.58420599e-01 -4.82867099e-02]
|
[13.81321907043457, 7.035414695739746]
|
045fae45-396e-46a7-8dec-d2a5bf4c58a1
|
a-continual-development-methodology-for-large
|
2209.07326
| null |
https://arxiv.org/abs/2209.07326v3
|
https://arxiv.org/pdf/2209.07326v3.pdf
|
A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems
|
The traditional Machine Learning (ML) methodology requires to fragment the development and experimental process into disconnected iterations whose feedback is used to guide design or tuning choices. This methodology has multiple efficiency and scalability disadvantages, such as leading to spend significant resources into the creation of multiple trial models that do not contribute to the final solution.The presented work is based on the intuition that defining ML models as modular and extensible artefacts allows to introduce a novel ML development methodology enabling the integration of multiple design and evaluation iterations into the continuous enrichment of a single unbounded intelligent system. We define a novel method for the generation of dynamic multitask ML models as a sequence of extensions and generalizations. We first analyze the capabilities of the proposed method by using the standard ML empirical evaluation methodology. Finally, we propose a novel continuous development methodology that allows to dynamically extend a pre-existing multitask large-scale ML system while analyzing the properties of the proposed method extensions. This results in the generation of an ML model capable of jointly solving 124 image classification tasks achieving state of the art quality with improved size and compute cost.
|
['Andrea Gesmundo']
|
2022-09-15
| null | null | null | null |
['scene-classification', 'fine-grained-image-classification']
|
['computer-vision', 'computer-vision']
|
[ 2.95266300e-01 3.02847862e-01 1.34697482e-01 -3.96620363e-01
-6.86530530e-01 -5.06771922e-01 6.28941834e-01 4.57170635e-01
-4.96293098e-01 6.49214983e-01 -4.40347135e-01 -4.38828439e-01
-7.07371354e-01 -4.84142691e-01 -5.73319733e-01 -5.45076847e-01
-1.14821270e-01 9.64963555e-01 4.36618954e-01 6.57719895e-02
4.10547584e-01 4.54358220e-01 -1.88960779e+00 3.87457192e-01
1.04228997e+00 9.38495517e-01 5.84325433e-01 7.28052974e-01
-1.42946035e-01 6.75807238e-01 -6.61044121e-01 -1.73565075e-01
2.36050516e-01 -1.49878323e-01 -8.56130183e-01 3.46705586e-01
2.92066813e-01 1.48634747e-01 9.12747383e-01 6.40351057e-01
5.74141622e-01 -9.27807093e-02 4.04116958e-01 -1.42706156e+00
3.91105041e-02 6.87673509e-01 -2.56313115e-01 -2.54188150e-01
1.00424148e-01 -2.84530176e-03 7.60351539e-01 -6.27079248e-01
4.82428610e-01 9.92475986e-01 5.13383329e-01 8.50438327e-02
-1.42865825e+00 -3.55305672e-01 4.04060706e-02 1.72182620e-01
-1.30719721e+00 -5.70497811e-01 6.78893268e-01 -5.65384209e-01
1.09107029e+00 2.80885726e-01 6.24136806e-01 8.06575179e-01
-4.92181815e-02 7.41357803e-01 1.38461947e+00 -1.11666107e+00
6.81351066e-01 8.02225649e-01 3.80167931e-01 8.15781116e-01
5.25750101e-01 -2.34979287e-01 -1.63684189e-01 -2.56860822e-01
3.63662004e-01 -5.58235586e-01 2.38174096e-01 -6.53937638e-01
-9.63545680e-01 7.36798227e-01 -1.48591757e-01 7.35909283e-01
-2.40127489e-01 -7.21784309e-02 4.13086146e-01 4.86437201e-01
2.35831156e-01 6.63030148e-01 -5.51231503e-01 -1.24930516e-01
-1.13926911e+00 3.62357020e-01 1.00189590e+00 6.67765617e-01
9.16631877e-01 -9.69926789e-02 1.68181419e-01 5.67370236e-01
3.22153777e-01 1.81553006e-01 7.66220689e-01 -9.18343604e-01
1.54706240e-01 9.17644858e-01 3.48302662e-01 -7.95488060e-01
-7.25545049e-01 -8.97894919e-01 -2.44939134e-01 5.12711406e-01
3.28808039e-01 -2.06682727e-01 -3.49698752e-01 1.61545646e+00
4.50432330e-01 -2.70011753e-01 -3.06532979e-02 3.24788094e-01
2.67282605e-01 2.01042026e-01 4.71422896e-02 -5.07110894e-01
1.10175776e+00 -1.00510716e+00 -5.63624263e-01 6.85408190e-02
1.02454114e+00 -7.52941072e-01 1.23147428e+00 9.51159358e-01
-1.11765146e+00 -9.10918772e-01 -1.37241864e+00 3.56772572e-01
-5.36092937e-01 4.67558026e-01 6.19434118e-01 1.11284971e+00
-1.09448504e+00 4.17555630e-01 -6.92374706e-01 -3.38482976e-01
3.78728993e-02 4.41864878e-01 2.35709939e-02 4.12818223e-01
-6.57326758e-01 1.13622308e+00 5.52957058e-01 1.86008930e-01
-6.27749145e-01 -3.81508112e-01 -5.07369339e-01 -1.45779610e-01
5.42794645e-01 -7.19424665e-01 1.15946615e+00 -1.38054168e+00
-1.52626133e+00 6.44646883e-01 1.69648807e-02 -4.22153860e-01
8.86043489e-01 -1.77441418e-01 -2.09914669e-01 -2.25211993e-01
-3.58253368e-03 4.97847438e-01 7.45063841e-01 -1.36995196e+00
-7.32937753e-01 -4.16553110e-01 -6.06613122e-02 6.65657595e-03
-6.79402769e-01 -1.30877644e-01 -1.39475062e-01 -3.67750645e-01
-2.63517797e-01 -8.32622588e-01 -2.37153575e-01 -5.88055909e-01
-1.34552242e-02 -3.32599372e-01 5.47788620e-01 -1.62945837e-01
1.51457298e+00 -1.83828759e+00 3.42248678e-01 2.62578636e-01
1.58899754e-01 2.88044244e-01 -1.41576827e-01 4.14442509e-01
-1.40384883e-01 6.33183494e-02 -3.22108008e-02 -4.28043127e-01
1.88904479e-01 9.20250490e-02 1.66658163e-01 3.09254855e-01
-5.30104339e-02 7.15860069e-01 -5.02995312e-01 -7.18282640e-01
3.45557451e-01 9.28853229e-02 -4.19345707e-01 3.61302048e-02
-5.41960061e-01 1.21501304e-01 -5.50899029e-01 5.14769614e-01
4.74289656e-01 -1.41940534e-01 4.38111633e-01 -1.88184261e-01
-2.43350685e-01 -1.62633166e-01 -1.63704491e+00 1.75785732e+00
-7.79336810e-01 2.62944996e-01 -3.97417545e-02 -1.34300947e+00
1.10903883e+00 2.60389209e-01 5.65985024e-01 -5.38738191e-01
3.87320340e-01 4.98967618e-01 1.70392655e-02 -6.42848790e-01
3.06024164e-01 3.92656997e-02 -9.54539478e-02 7.44711816e-01
4.47117649e-02 8.79795011e-03 3.93098027e-01 -2.44814247e-01
7.72162616e-01 4.59811717e-01 3.70767772e-01 -3.82648408e-01
7.85079718e-01 1.38638765e-01 8.44798535e-02 8.26517522e-01
3.89956981e-02 -1.24917924e-01 4.65291798e-01 -4.86876011e-01
-1.06524467e+00 -7.49052405e-01 -2.11941034e-01 1.00969577e+00
-3.79153222e-01 -2.37759441e-01 -8.36248636e-01 -7.72039950e-01
-2.75034338e-01 7.03570545e-01 -6.36010587e-01 2.81602472e-01
-3.49605024e-01 -9.53774333e-01 4.93545711e-01 8.72727949e-03
3.53099316e-01 -1.11281693e+00 -1.27558565e+00 3.60997438e-01
6.97524995e-02 -1.02252805e+00 3.86611253e-01 3.85984898e-01
-1.09891415e+00 -8.91444623e-01 -2.15813071e-01 -6.67275786e-01
4.44763631e-01 -3.82347614e-01 1.07375121e+00 3.06888539e-02
-4.59503412e-01 2.14355946e-01 -4.53295261e-01 -4.19964254e-01
-7.71008968e-01 3.70173901e-01 -4.35885414e-02 5.24366684e-02
2.37541944e-01 -6.60130680e-01 -9.09038261e-02 2.40091190e-01
-9.92423892e-01 1.78313032e-01 8.12811196e-01 7.53981650e-01
3.64148378e-01 3.04774106e-01 1.00462496e+00 -1.07534575e+00
9.37768757e-01 -1.60430834e-01 -8.93705726e-01 6.90884829e-01
-1.32083714e+00 3.94024372e-01 6.37527466e-01 -4.98589605e-01
-1.09533453e+00 1.97483256e-01 9.74117666e-02 -4.56081554e-02
-3.83915454e-01 5.36770642e-01 -9.79053080e-02 -1.93589687e-01
8.60565901e-01 5.76958954e-02 3.40347923e-02 -6.40641749e-01
3.69737089e-01 6.86022222e-01 -7.97747523e-02 -8.24554324e-01
3.91724348e-01 -8.85011256e-02 2.20885798e-01 -6.69359803e-01
-2.81634420e-01 -2.38285184e-01 -8.55430126e-01 -4.51318711e-01
4.09291089e-01 -4.87133503e-01 -7.76906848e-01 3.05924654e-01
-9.40734386e-01 -4.48338419e-01 -2.91079760e-01 2.72975683e-01
-6.58814609e-01 2.98050225e-01 -4.16811965e-02 -8.48081589e-01
-1.31824523e-01 -1.41680980e+00 8.65379810e-01 -1.02582015e-01
-4.37357426e-01 -1.04933333e+00 2.60264188e-01 6.38517737e-01
2.63871431e-01 7.32136592e-02 1.01208711e+00 -8.15187693e-01
-4.02358323e-01 -3.36993545e-01 5.66694066e-02 3.79245311e-01
-1.30276993e-01 -2.49216873e-02 -9.49933589e-01 -2.60585904e-01
1.25666171e-01 -4.80826944e-01 3.08639020e-01 2.26863921e-01
6.88419163e-01 -6.70818612e-03 -2.22222045e-01 2.80088067e-01
1.72014987e+00 3.37282687e-01 4.35389519e-01 7.17396140e-01
2.43759513e-01 9.36851323e-01 7.26753294e-01 3.77389938e-01
7.94092640e-02 9.53474700e-01 1.33104324e-01 -3.01440507e-02
1.52783960e-01 2.18231216e-01 3.46307963e-01 6.97449923e-01
-5.97035065e-02 -3.29977162e-02 -1.00717103e+00 5.07057786e-01
-2.10198832e+00 -5.96930921e-01 -1.70019254e-01 2.35095572e+00
5.41963041e-01 3.76294911e-01 3.52084398e-01 3.83078635e-01
3.00604820e-01 -3.07359964e-01 -1.38378128e-01 -6.94692373e-01
2.85816848e-01 4.13852572e-01 2.77941465e-01 6.69462621e-01
-9.36646342e-01 6.92632198e-01 6.09902191e+00 8.52128208e-01
-1.02103198e+00 3.13661844e-01 2.70088553e-01 -1.08913980e-01
-2.05295086e-01 -4.92786756e-03 -7.14454114e-01 9.45592299e-02
1.11775494e+00 -2.19570145e-01 2.42440745e-01 8.90524209e-01
4.44102347e-01 -4.54483032e-01 -1.05454361e+00 5.80569744e-01
1.09304249e-01 -1.20235765e+00 1.40678287e-01 7.22375661e-02
5.37082732e-01 -2.51284003e-01 -5.56661822e-02 1.46256536e-01
7.92649910e-02 -8.47427726e-01 8.84144008e-01 4.19224799e-01
3.58550310e-01 -7.19509542e-01 7.88071871e-01 7.35089898e-01
-9.01410460e-01 -3.53587568e-01 -9.79619101e-02 -1.62114203e-01
-1.66002527e-01 4.10382569e-01 -1.09368944e+00 9.10135686e-01
3.17238092e-01 7.53723830e-02 -1.00284100e+00 9.87439513e-01
2.96254665e-01 3.37244600e-01 -2.50851750e-01 -2.16115370e-01
2.27563724e-01 -6.52640164e-02 4.11188275e-01 1.39267242e+00
2.16311783e-01 -7.52296507e-01 3.49599689e-01 8.58957291e-01
5.77589929e-01 5.52414417e-01 -4.74340677e-01 1.90034196e-01
2.61778504e-01 1.30414474e+00 -9.33796585e-01 -2.92483062e-01
-2.45269179e-01 5.17790496e-01 3.74492854e-01 1.03488535e-01
-8.20928931e-01 -2.36519605e-01 -1.39127582e-01 2.05615312e-01
2.74309188e-01 -2.25237906e-01 -6.70988858e-01 -8.62246633e-01
2.20757827e-01 -1.14076042e+00 2.26858482e-01 -4.75423843e-01
-7.23274529e-01 8.22618842e-01 3.32384944e-01 -9.94360209e-01
-5.47771215e-01 -3.52407694e-01 -6.88852668e-02 8.66745412e-01
-1.25196409e+00 -1.27759504e+00 -3.03144872e-01 3.41663957e-01
6.67660534e-01 -4.69692498e-01 9.76483285e-01 2.89125800e-01
-5.66867828e-01 5.27957261e-01 -1.20677672e-01 -6.46523297e-01
5.50267398e-01 -1.17827332e+00 -2.28074864e-01 8.58387768e-01
-9.27330181e-02 5.04508853e-01 7.21786022e-01 -5.01375198e-01
-1.07854271e+00 -7.85186887e-01 9.41157520e-01 -4.06150579e-01
6.30298018e-01 -3.87147784e-01 -5.18362403e-01 5.17556310e-01
6.22221380e-02 -5.95093668e-01 4.67956752e-01 2.79953241e-01
8.22854228e-03 -3.70137274e-01 -1.03047574e+00 3.58998626e-01
5.54008245e-01 -1.39020666e-01 -6.02131307e-01 8.79227519e-02
3.18404377e-01 2.49844566e-01 -7.38943577e-01 4.58897859e-01
6.84311748e-01 -1.17988706e+00 6.62383139e-01 -2.28889316e-01
1.28840178e-01 -2.51767993e-01 9.77490768e-02 -8.14028263e-01
-8.65472481e-02 -6.84155643e-01 5.43026961e-02 1.13374245e+00
7.08279371e-01 -6.12901568e-01 6.83743298e-01 5.43868721e-01
-1.61961615e-02 -7.96788275e-01 -8.47862303e-01 -7.47467637e-01
-1.75802186e-01 -6.79214716e-01 2.30429024e-01 6.97591186e-01
-1.56307649e-02 3.74001145e-01 -8.82197917e-02 -1.19999059e-01
7.35561788e-01 1.86923712e-01 9.07158852e-01 -1.46441221e+00
-7.63853252e-01 -6.02191031e-01 -2.51285642e-01 -5.51386476e-01
-1.05491273e-01 -7.33045876e-01 -3.95588502e-02 -1.25074601e+00
1.44723907e-01 -8.07980418e-01 -3.10439080e-01 4.69654590e-01
2.89672792e-01 -5.51476292e-02 1.30732164e-01 2.43612766e-01
-6.13611579e-01 4.33469117e-02 8.61756206e-01 2.12964669e-01
-4.58308905e-01 1.62670657e-01 -5.97746015e-01 6.94325745e-01
6.33219182e-01 -4.89903212e-01 -8.94922018e-01 -1.55306116e-01
5.54421425e-01 -5.85500933e-02 1.62378892e-01 -1.19974303e+00
1.43380150e-01 1.01436101e-01 1.56658188e-01 -1.91191316e-01
-4.54715267e-02 -1.07683516e+00 5.21586061e-01 5.09505391e-01
-3.84967744e-01 1.01393737e-01 2.21753389e-01 3.46542001e-01
-9.97465476e-02 -5.95181227e-01 6.55151248e-01 -2.09599063e-01
-7.13077009e-01 -4.37206119e-01 -4.08640832e-01 -4.37349468e-01
1.42440951e+00 -4.11494970e-01 -2.30142232e-02 2.27737352e-01
-1.05849051e+00 1.24624155e-01 3.57925117e-01 3.02135408e-01
1.52205124e-01 -6.65138662e-01 -4.78929043e-01 2.74133116e-01
-8.41010883e-02 -4.19056356e-01 9.35425311e-02 9.32522297e-01
-5.77112794e-01 6.15623951e-01 -3.43230218e-01 -6.30697370e-01
-1.56376088e+00 6.75880253e-01 3.20295542e-01 -7.38239825e-01
-3.85879576e-01 3.67611349e-01 -1.11418828e-01 -4.13742274e-01
4.01374698e-01 -2.99600929e-01 -3.94314349e-01 3.46270621e-01
1.63877800e-01 3.22271824e-01 4.47088122e-01 -4.57008108e-02
-1.55615151e-01 4.95544136e-01 1.82325259e-01 -3.58584225e-01
1.49676955e+00 -1.55765474e-01 -1.48647830e-01 6.51389778e-01
8.04332376e-01 4.89478884e-03 -7.68305719e-01 1.11280635e-01
4.30278838e-01 -3.63312028e-02 1.47790059e-01 -1.15028501e+00
-4.55240577e-01 5.74468613e-01 8.71103108e-01 4.41487402e-01
1.31950808e+00 -1.90739453e-01 1.30622178e-01 6.62465453e-01
6.04986787e-01 -1.27033150e+00 -9.15709417e-03 -3.05211381e-03
6.20362580e-01 -9.42488551e-01 9.57155451e-02 -2.48081222e-01
-4.75747228e-01 1.38718462e+00 4.71737593e-01 1.93927109e-01
6.42898798e-01 5.40857732e-01 -1.59880593e-01 -3.57086986e-01
-9.25311148e-01 -1.23899721e-01 8.68520662e-02 4.96598631e-01
4.29653198e-01 1.39642566e-01 -8.81555378e-01 7.33772218e-01
4.18019481e-02 5.18569946e-01 4.81369019e-01 1.08437955e+00
-5.65661967e-01 -1.65363348e+00 -3.13831210e-01 1.47022530e-01
-1.46225065e-01 1.36205822e-01 -3.02868754e-01 9.92649615e-01
6.86537325e-01 8.10196638e-01 -2.75830448e-01 -3.59933108e-01
2.98492044e-01 4.63499784e-01 8.01817954e-01 -6.11338913e-01
-8.52261126e-01 1.94223285e-01 4.48686987e-01 -2.70987332e-01
-6.40800178e-01 -6.93825960e-01 -6.96706355e-01 1.15907930e-01
-4.40757722e-01 2.75494307e-01 1.13104522e+00 1.13007128e+00
3.34082454e-01 7.01533556e-01 5.22541761e-01 -5.86010575e-01
-5.94990373e-01 -1.05867052e+00 -2.80780464e-01 9.02220160e-02
-2.33755596e-02 -8.31931889e-01 -1.82855576e-01 2.05986276e-01]
|
[8.346049308776855, 4.363644599914551]
|
2211d902-36b0-4e41-8fce-265dc74859c2
|
reference-based-autoencoder-for-surface
|
2211.10060
| null |
https://arxiv.org/abs/2211.10060v2
|
https://arxiv.org/pdf/2211.10060v2.pdf
|
Normal Reference Attention and Defective Feature Perception Network for Surface Defect Detection
|
Visual anomaly detection plays a significant role in the development of industrial automatic product quality inspection. As a result of the utmost imbalance in the amount of normal and abnormal data, growing attention has been given to unsupervised methods for defect detection. Although existing reconstruction-based methods have been widely studied recently, establishing a robust reconstruction model for various textured surface defect detection remains a challenging task due to homogeneous and nonregular surface textures. In this paper, we propose a novel unsupervised reconstruction-based method called the normal reference attention and defective feature perception network (NDP-Net) to accurately inspect a variety of textured defects. Unlike most reconstruction-based methods, our NDP-Net first employs an encoding module that extracts multi scale discriminative features of the surface textures, which is augmented with the defect discriminative ability by the proposed artificial defects and the novel pixel-level defect perception loss. Subsequently, a novel reference-based attention module (RBAM) is proposed to leverage the normal features of the fixed reference image to repair the defective features and restrain the reconstruction of the defects. Next, the repaired features are fed into a decoding module to reconstruct the normal textured background. Finally, the novel multi scale defect segmentation module (MSDSM) is introduced for precise defect detection and segmentation. In addition, a two-stage training strategy is utilized to enhance the inspection performance.
|
['Wenyong Yu', 'Haiming Yao', 'Wei Luo']
|
2022-11-18
| null | null | null | null |
['defect-detection']
|
['computer-vision']
|
[ 6.82249248e-01 9.73347500e-02 2.84390867e-01 -2.49396279e-01
-5.19069612e-01 1.67383865e-01 1.05376795e-01 2.83653319e-01
1.97527364e-01 1.03849664e-01 -3.53034675e-01 4.36257338e-03
-1.05795227e-01 -8.04859936e-01 -4.46261287e-01 -1.00655591e+00
2.99123824e-01 -2.04095361e-03 5.60473025e-01 -3.12003046e-01
5.40539265e-01 5.35816371e-01 -1.74215114e+00 3.34683955e-01
1.10799992e+00 1.38359141e+00 5.10404289e-01 4.02574688e-01
-8.60709995e-02 4.74419713e-01 -4.78596270e-01 7.60254860e-02
2.88020283e-01 -5.43540537e-01 -4.71329302e-01 8.12271774e-01
1.31133378e-01 -4.05577898e-01 -3.04922521e-01 1.31086195e+00
4.98188317e-01 3.49310100e-01 5.74085951e-01 -8.33182991e-01
-6.63377285e-01 -3.47664207e-01 -8.48263621e-01 3.28305602e-01
1.48275122e-01 3.55425805e-01 8.02561700e-01 -9.68036294e-01
3.54860157e-01 1.18351305e+00 3.05572331e-01 3.77388120e-01
-1.06820321e+00 -4.39808555e-02 2.08429903e-01 3.25794309e-01
-9.05669153e-01 -2.48368457e-01 1.39537978e+00 -3.86353135e-01
7.69289076e-01 2.34921888e-01 6.13650680e-01 6.30231202e-01
3.77238601e-01 8.37326705e-01 7.67825007e-01 -5.18692911e-01
1.26683757e-01 -1.49156660e-01 -9.45134833e-02 8.83882523e-01
3.37058753e-01 4.43854183e-02 -1.83777466e-01 4.55449849e-01
1.01901054e+00 4.79385257e-01 -3.70528549e-01 -3.87635916e-01
-6.98809624e-01 5.61199963e-01 3.64182413e-01 2.77493060e-01
-6.24969780e-01 -3.60412389e-01 5.48391104e-01 3.69604111e-01
6.58801615e-01 3.64867628e-01 -3.68055761e-01 2.57132709e-01
-6.83492959e-01 -1.59170300e-01 2.71543115e-01 3.20128113e-01
8.88191342e-01 3.86323661e-01 -1.73974246e-01 1.24491668e+00
4.77883667e-01 2.21516415e-01 5.28517663e-01 -7.40716636e-01
2.89341569e-01 1.09414935e+00 -1.30641043e-01 -1.61440456e+00
-4.70207445e-02 -3.83689761e-01 -1.06477594e+00 6.36542678e-01
4.56239544e-02 6.01146638e-01 -1.22659075e+00 9.56035733e-01
5.60669601e-01 -1.36108682e-01 -1.82110697e-01 9.16928947e-01
4.74569499e-01 5.22437751e-01 -3.17887068e-01 -8.99207667e-02
1.06345963e+00 -8.79545987e-01 -5.73521435e-01 -3.29518050e-01
4.02089208e-01 -7.23258495e-01 1.11000931e+00 6.36775970e-01
-9.73722577e-01 -7.27032006e-01 -1.18933070e+00 -1.21029399e-01
-1.57554671e-01 2.05019400e-01 3.81404907e-01 3.22939962e-01
-6.05750620e-01 4.99478877e-01 -8.87999356e-01 -2.28778228e-01
7.44813800e-01 1.97687194e-01 -3.65778118e-01 -6.84141755e-01
-6.13208652e-01 4.78509605e-01 2.37615153e-01 7.07377315e-01
-9.99453783e-01 -1.56666577e-01 -1.04507399e+00 -1.55824184e-01
3.74108940e-01 -1.68565050e-01 6.81799233e-01 -1.01890087e+00
-1.48053432e+00 8.82675171e-01 -1.21757842e-03 -3.67795080e-02
2.33110934e-01 -5.79414107e-02 -4.13313240e-01 3.98170859e-01
3.13644320e-01 1.69704333e-01 1.16327560e+00 -1.51602483e+00
-7.35400796e-01 -4.51889545e-01 -1.77125916e-01 1.62347645e-01
-1.47846356e-01 -2.56490290e-01 -5.69581211e-01 -9.41823006e-01
8.40405345e-01 -1.93155393e-01 -3.30398828e-01 8.42802897e-02
-7.27746129e-01 -1.56704336e-01 1.02591431e+00 -1.02179348e+00
1.01084840e+00 -2.59771466e+00 1.63939923e-01 3.54943365e-01
1.86281234e-01 2.89993793e-01 -2.88876325e-01 -4.35090512e-02
-1.58541247e-01 -2.11643159e-01 -7.33547151e-01 -4.06241983e-01
-1.58573791e-01 3.28865618e-01 -2.55079363e-02 6.64425850e-01
6.61064386e-01 5.09661257e-01 -6.56719148e-01 -4.43364263e-01
5.41074455e-01 1.08625432e-02 -6.47091508e-01 4.04234767e-01
-1.56036481e-01 8.11382711e-01 -5.17303109e-01 1.25868285e+00
8.95334423e-01 7.57187605e-02 -2.29287952e-01 -3.72543156e-01
-7.68152624e-02 -9.25858915e-02 -9.33067739e-01 1.60843921e+00
-2.95544803e-01 1.74094647e-01 3.81593466e-01 -1.59105659e+00
1.16696501e+00 1.08433127e-01 5.98461986e-01 -1.02639103e+00
4.20361459e-02 4.26013738e-01 -3.29221636e-02 -9.29729760e-01
2.47300699e-01 -1.32847846e-01 1.24371886e-01 7.09662810e-02
-6.41495511e-02 -2.52562732e-01 -9.74730216e-03 -1.45109236e-01
1.11167014e+00 9.02227536e-02 -6.01695105e-02 3.20023261e-02
7.22954273e-01 -1.44840062e-01 1.00071907e+00 3.90590757e-01
-3.18648189e-01 8.81511688e-01 3.58450025e-01 -5.14324784e-01
-9.46690917e-01 -9.37069178e-01 -6.94477484e-02 6.27262950e-01
4.74774539e-01 1.93346396e-01 -4.85716164e-01 -9.76658702e-01
9.90068093e-02 3.20818603e-01 -6.00404322e-01 -5.34115076e-01
-6.11405075e-01 -7.22442687e-01 -1.13699911e-03 2.46658906e-01
7.76816547e-01 -1.35972404e+00 -4.18519080e-01 1.69405580e-01
7.28858784e-02 -5.59026659e-01 -2.57527053e-01 2.49713421e-01
-9.51625407e-01 -1.34557724e+00 -5.54890335e-01 -1.06977546e+00
1.14948320e+00 3.92392129e-01 6.80347383e-01 4.32837814e-01
-7.20156848e-01 2.71231264e-01 -4.93285120e-01 -1.71771631e-01
-4.09907550e-01 -4.84325558e-01 -2.74151325e-01 4.80860472e-01
1.87633306e-01 -4.34792906e-01 -7.89667845e-01 2.59330690e-01
-1.27937293e+00 -1.20207459e-01 9.52841222e-01 1.20673323e+00
9.96042609e-01 6.76514626e-01 6.49950564e-01 -8.56013596e-01
2.48504251e-01 -5.34931839e-01 -4.16934431e-01 -1.85175054e-02
-5.21773696e-01 -1.91338658e-01 5.01113355e-01 -2.16331154e-01
-1.19552565e+00 -1.20829090e-01 -4.94870752e-01 -5.08784235e-01
-4.20708418e-01 4.34592187e-01 -5.79019904e-01 -1.90936457e-02
1.46044806e-01 5.00611007e-01 3.29893410e-01 -7.96017230e-01
-2.21607283e-01 6.36936486e-01 5.80794632e-01 -1.98447436e-01
8.36269975e-01 4.95704651e-01 -1.72349632e-01 -9.08197939e-01
-9.09357011e-01 -6.42764330e-01 -5.36900878e-01 -2.38994464e-01
8.80978525e-01 -4.59591925e-01 -3.08978051e-01 9.82868850e-01
-9.77991283e-01 -3.07772249e-01 -7.27028191e-01 2.48796389e-01
-3.65406424e-01 9.51528966e-01 -7.48128295e-01 -8.73615146e-01
-2.75953442e-01 -1.19052505e+00 1.14024115e+00 1.14130296e-01
4.09357905e-01 -8.13317716e-01 -3.27723831e-01 5.21988928e-01
3.06755602e-01 3.22542608e-01 1.27634847e+00 -3.03025395e-01
-7.12906778e-01 -4.14768696e-01 -3.32625657e-01 1.15637600e+00
5.50828993e-01 -3.58869463e-01 -6.49680734e-01 -3.42395872e-01
5.67126453e-01 -2.04398468e-01 1.08734441e+00 4.08833653e-01
1.29054177e+00 -8.17096606e-02 -2.06275046e-01 5.06805480e-01
1.27078247e+00 4.27296042e-01 8.17592978e-01 4.70434606e-01
9.43727374e-01 7.05088139e-01 9.64150071e-01 3.40643674e-01
-3.83271389e-02 3.46328408e-01 8.12824011e-01 -4.75919843e-01
-2.07350269e-01 -6.42144158e-02 3.20764691e-01 1.07806647e+00
7.56226704e-02 -7.27355108e-02 -4.28507000e-01 7.33223200e-01
-1.54070973e+00 -5.99269271e-01 -3.64000835e-02 2.19862008e+00
5.24815559e-01 3.66081119e-01 -4.19756323e-01 6.12475693e-01
8.40918720e-01 7.70895481e-02 -8.89891803e-01 -2.30970114e-01
-2.65059531e-01 1.86841428e-01 2.06752755e-02 1.14789270e-02
-1.08148205e+00 6.12240016e-01 4.97758818e+00 1.04051864e+00
-9.84233677e-01 2.67274156e-02 7.23939419e-01 4.23115849e-01
-1.99341312e-01 -1.61994353e-01 -2.62141794e-01 5.27442873e-01
4.53813635e-02 6.51804328e-01 1.34228453e-01 8.43828917e-01
3.35758954e-01 -3.65391523e-01 -6.26193166e-01 8.77077103e-01
2.88920283e-01 -7.12253869e-01 1.10919878e-01 -2.05167010e-02
6.32562220e-01 -3.93083274e-01 1.59408092e-01 9.90649834e-02
-2.81593144e-01 -7.35193253e-01 5.46628058e-01 6.50721490e-01
9.15993392e-01 -6.90064788e-01 9.07272398e-01 8.64852741e-02
-1.06294465e+00 -4.24468756e-01 -6.37978077e-01 2.02408716e-01
7.52772763e-02 1.08235371e+00 -3.54500443e-01 8.25307786e-01
8.13745022e-01 1.02944851e+00 -5.15787363e-01 1.14682186e+00
-2.23855823e-01 6.59036577e-01 -2.55787633e-02 7.25588381e-01
6.29848987e-02 -2.74084717e-01 8.32360208e-01 6.88205004e-01
3.48893613e-01 -2.44602591e-01 2.87582874e-01 7.88915217e-01
2.27601513e-01 -9.20573846e-02 -5.50042331e-01 1.49498537e-01
-7.39815682e-02 1.27871370e+00 -8.73922527e-01 -1.19114043e-02
-5.05409181e-01 1.39254165e+00 7.33876452e-02 3.85042667e-01
-3.22173148e-01 -5.25828004e-01 5.10242105e-01 1.94462702e-01
4.46704060e-01 -1.12629145e-01 -2.08767965e-01 -1.05553198e+00
1.81916878e-01 -9.20031667e-01 3.25075567e-01 -5.30827522e-01
-1.60603583e+00 4.23966438e-01 -5.24009109e-01 -1.40334773e+00
4.53790873e-01 -6.49911106e-01 -7.37985969e-01 8.32621098e-01
-1.79461086e+00 -9.72216070e-01 -3.54713827e-01 3.61358583e-01
1.02875543e+00 -3.51845354e-01 4.32260841e-01 4.02021497e-01
-1.14241374e+00 3.23244095e-01 3.22923549e-02 3.48698385e-02
3.39803785e-01 -1.17242825e+00 3.76608893e-02 1.19637108e+00
-3.92930925e-01 2.26207823e-01 4.34508264e-01 -9.04951155e-01
-1.25124216e+00 -1.35095084e+00 1.73564300e-01 1.47888198e-01
1.73682526e-01 -9.85239372e-02 -1.37822950e+00 2.62390703e-01
-1.29682422e-01 3.96807373e-01 1.48918241e-01 -5.94000638e-01
7.91961327e-02 -2.42092326e-01 -1.26641560e+00 2.06773832e-01
7.54462898e-01 -3.56346637e-01 -4.74899083e-01 1.50636986e-01
5.61256170e-01 -2.69754052e-01 -6.63546920e-01 7.11849928e-01
1.10307239e-01 -1.02941442e+00 7.91034222e-01 -1.03137232e-01
4.43211019e-01 -5.36196947e-01 -1.46144778e-01 -1.09365177e+00
-3.31899703e-01 -1.18902974e-01 -2.28685327e-02 1.26992202e+00
-8.24284106e-02 -7.73422599e-01 8.47984016e-01 3.90309095e-03
-9.08855140e-01 -1.11905682e+00 -8.82710159e-01 -4.69821393e-01
-4.71358985e-01 -3.31097662e-01 1.84310913e-01 7.33255327e-01
-6.00303590e-01 -9.02697146e-02 -6.80734888e-02 3.57206851e-01
6.25018239e-01 2.09510669e-01 2.63509333e-01 -1.31014884e+00
-3.37319106e-01 -2.38539651e-01 -7.33195484e-01 -1.22486782e+00
-1.52821675e-01 -8.59520853e-01 4.94334698e-01 -1.57922399e+00
-2.91881002e-02 -6.40594840e-01 -6.29881620e-01 3.64191920e-01
-2.72298276e-01 4.24097180e-01 -4.50647265e-01 2.84434080e-01
-5.63548326e-01 8.43213737e-01 1.56413376e+00 -3.55688900e-01
-2.15226904e-01 1.39489576e-01 -5.16416788e-01 6.71444774e-01
7.23637462e-01 -2.74952143e-01 -2.22870708e-01 -3.24344665e-01
-7.07321316e-02 -1.90448433e-01 5.67781210e-01 -1.01020896e+00
-7.68166548e-03 6.02802485e-02 4.98490334e-01 -6.95512533e-01
1.56134143e-01 -7.41252244e-01 -5.42472124e-01 4.44637865e-01
1.63517952e-01 -1.59031004e-01 1.60209332e-02 1.03500104e+00
-6.58397615e-01 -2.30786100e-01 1.02221763e+00 -2.12942049e-01
-8.44378054e-01 3.73516738e-01 -3.92302781e-01 -2.89918065e-01
9.27018046e-01 -7.38366663e-01 8.58052075e-02 2.33839154e-01
-5.86671948e-01 9.77276489e-02 5.27262330e-01 3.39552879e-01
1.24380577e+00 -1.00543571e+00 -5.77509642e-01 9.31147754e-01
1.48923188e-01 5.88259876e-01 8.35735559e-01 1.00641048e+00
-6.52905583e-01 -4.57687676e-02 -2.18379021e-01 -7.73480952e-01
-7.95406580e-01 4.74205732e-01 3.31922770e-01 -2.83947945e-01
-9.00234044e-01 8.09896529e-01 4.17709976e-01 -1.81086123e-01
1.89468358e-02 -3.15265864e-01 -2.21621931e-01 -3.08826298e-01
2.00624809e-01 4.39005166e-01 3.56754810e-01 -6.33520544e-01
-1.13888849e-02 5.50409794e-01 -1.82234243e-01 4.89072472e-01
1.37557733e+00 -4.54558760e-01 -4.17039543e-01 3.71280283e-01
1.00403285e+00 6.16141642e-03 -1.55035043e+00 -4.39965315e-02
-2.30550729e-02 -6.15126371e-01 2.99582273e-01 -5.93409479e-01
-1.49349141e+00 9.95262861e-01 8.61108959e-01 4.12071109e-01
1.59343612e+00 -1.63089573e-01 1.11274731e+00 -1.04852818e-01
2.25420713e-01 -1.06541741e+00 5.95361054e-01 1.86291516e-01
8.41593385e-01 -1.37856793e+00 -1.71839565e-01 -6.72686398e-01
-4.61308718e-01 9.67007041e-01 8.19898903e-01 -2.56110132e-01
3.90149742e-01 -1.32067511e-02 3.39275994e-03 -4.65993792e-01
-1.56654805e-01 -8.41488838e-02 3.04280668e-01 7.03026295e-01
-1.97472945e-02 -4.42560136e-01 7.14277336e-03 6.21049106e-01
6.14488721e-01 -3.96992266e-01 1.22274190e-01 1.05222571e+00
-5.93289316e-01 -9.55100834e-01 -2.86848038e-01 8.13178480e-01
-5.15150070e-01 1.20209284e-01 2.04346702e-01 4.17983651e-01
4.90177691e-01 1.03639829e+00 1.96686655e-01 -3.84674460e-01
5.28855443e-01 -2.28174612e-01 2.84216791e-01 -1.02428615e+00
-2.46419117e-01 5.81034459e-02 -3.72900963e-01 -5.92718422e-01
-1.93800643e-01 -5.68321347e-01 -1.24348533e+00 4.34300601e-01
-5.54334760e-01 -8.06724355e-02 3.24201196e-01 8.70802701e-01
3.21379483e-01 8.52751255e-01 9.61323619e-01 -9.54882443e-01
-3.23286623e-01 -9.72804844e-01 -1.13248396e+00 7.09532559e-01
5.32284617e-01 -9.57576334e-01 -5.16351700e-01 5.98728750e-03]
|
[7.496016025543213, 1.8929733037948608]
|
8e0daa68-fefc-4256-a1d9-e845e609cfd0
|
gencomparesum-a-hybrid-unsupervised
| null | null |
https://aclanthology.org/2022.bionlp-1.22
|
https://aclanthology.org/2022.bionlp-1.22.pdf
|
GenCompareSum: a hybrid unsupervised summarization method using salience
|
Text summarization (TS) is an important NLP task. Pre-trained Language Models (PLMs) have been used to improve the performance of TS. However, PLMs are limited by their need of labelled training data and by their attention mechanism, which often makes them unsuitable for use on long documents. To this end, we propose a hybrid, unsupervised, abstractive-extractive approach, in which we walk through a document, generating salient textual fragments representing its key points. We then select the most important sentences of the document by choosing the most similar sentences to the generated texts, calculated using BERTScore. We evaluate the efficacy of generating and using salient textual fragments to guide extractive summarization on documents from the biomedical and general scientific domains. We compare the performance between long and short documents using different generative text models, which are finetuned to generate relevant queries or document titles. We show that our hybrid approach out-performs existing unsupervised methods, as well as state-of-the-art supervised methods, despite not needing a vast amount of labelled training data.
|
['Sophia Ananiadou', 'Qianqian Xie', 'Jennifer Bishop']
| null | null | null | null |
bionlp-acl-2022-5
|
['extractive-summarization']
|
['natural-language-processing']
|
[ 6.32914126e-01 4.65126783e-01 -1.76545829e-01 -2.29931742e-01
-1.21788216e+00 -5.80073655e-01 8.88154328e-01 6.81992471e-01
-4.83436435e-01 9.76684809e-01 8.83870959e-01 3.01134512e-02
-2.81826202e-02 -5.36379635e-01 -4.99325603e-01 -5.77349365e-01
2.90404528e-01 8.42500687e-01 3.06485593e-01 -2.35745996e-01
8.00228596e-01 3.54238659e-01 -1.46269631e+00 5.89344919e-01
1.39694035e+00 2.46417537e-01 4.15918827e-01 8.33614707e-01
-6.77222073e-01 6.16534650e-01 -9.90633070e-01 -2.32237801e-01
-4.29568410e-01 -9.76303935e-01 -1.02813256e+00 1.07961237e-01
2.64772058e-01 5.16232960e-02 1.69995397e-01 7.93973863e-01
8.87666285e-01 1.07226223e-01 9.23183560e-01 -5.64532340e-01
-2.56117404e-01 7.69557416e-01 -4.99738634e-01 3.33271414e-01
6.24419749e-01 -3.32058549e-01 1.12522459e+00 -8.99900854e-01
9.81947362e-01 1.23581123e+00 2.62630463e-01 6.64662480e-01
-1.31603551e+00 -8.83110985e-02 -1.60604380e-02 -1.01409175e-01
-9.17937100e-01 -6.88885212e-01 9.03039277e-01 -2.53330976e-01
1.04656446e+00 4.31241900e-01 4.64995682e-01 1.08887172e+00
4.37162817e-01 1.12931299e+00 6.71714783e-01 -9.43872392e-01
4.18217182e-01 2.08696991e-01 1.11094326e-01 3.87729615e-01
2.73306340e-01 -6.95964158e-01 -6.39721453e-01 -3.82464826e-01
1.45386830e-01 -2.08464861e-01 -3.42352271e-01 -2.40994226e-02
-1.13740146e+00 1.08802533e+00 -4.12250385e-02 6.89196944e-01
-6.49081469e-01 -3.33731234e-01 5.98311722e-01 -1.95408612e-02
7.63640225e-01 9.09261882e-01 -2.09285527e-01 -2.92802993e-02
-1.57242000e+00 3.08482170e-01 9.54263628e-01 7.78277636e-01
4.00434792e-01 -2.42947459e-01 -7.40427434e-01 9.38205123e-01
-7.68577168e-03 2.40337655e-01 8.99669409e-01 -5.83328903e-01
6.57122254e-01 8.08183491e-01 -1.09938331e-01 -8.10903847e-01
-3.22010368e-01 -3.76690924e-01 -8.12502503e-01 -3.71304423e-01
-1.79329231e-01 -2.06591457e-01 -9.11032021e-01 1.27108932e+00
7.84467533e-03 -4.63708341e-01 3.69948208e-01 3.69506180e-01
9.97545600e-01 1.02297235e+00 7.91546851e-02 -7.21240819e-01
1.28576553e+00 -9.93988872e-01 -8.66457820e-01 -2.63628751e-01
6.92884743e-01 -9.80019808e-01 9.95475113e-01 4.25585747e-01
-1.26871359e+00 -4.78390485e-01 -8.43115151e-01 -1.87907502e-01
-3.25914472e-01 4.05331105e-01 5.69142327e-02 3.23593974e-01
-9.13965225e-01 6.81603670e-01 -6.39993906e-01 -6.32560313e-01
4.18026894e-01 1.86432004e-01 -8.28674212e-02 7.42415860e-02
-1.00646138e+00 7.82383561e-01 7.03409433e-01 -2.79056996e-01
-5.31564713e-01 -4.10446525e-01 -6.94141328e-01 3.10397685e-01
4.20918256e-01 -9.00894225e-01 1.23374450e+00 -6.54708147e-01
-1.46352422e+00 7.09499478e-01 -4.06233698e-01 -6.51560605e-01
3.45332563e-01 -4.59216893e-01 -3.68091874e-02 7.18535841e-01
2.87886202e-01 7.37262428e-01 8.06605399e-01 -1.16425836e+00
-5.09610713e-01 -2.14694947e-01 -2.47478202e-01 3.58729988e-01
-4.16823894e-01 1.82406113e-01 -4.09237295e-01 -7.91324675e-01
-1.80634111e-01 -7.31417596e-01 -4.80705470e-01 -8.56814086e-01
-8.41964841e-01 -4.79201287e-01 7.29968071e-01 -7.18002260e-01
1.59577143e+00 -1.82783210e+00 3.23564231e-01 4.38146852e-02
8.48702341e-02 4.15822357e-01 -7.25350231e-02 9.98142362e-01
2.06015110e-01 3.41980249e-01 -3.93344134e-01 -4.25621510e-01
-1.25138447e-01 4.96226400e-02 -5.30991137e-01 -1.54789478e-01
2.26213858e-01 8.62177134e-01 -1.10626841e+00 -1.00234890e+00
-4.49832994e-03 1.93439797e-01 -3.71344179e-01 1.74625427e-01
-5.39931118e-01 4.44156885e-01 -7.99205899e-01 1.44548625e-01
1.33428499e-01 -2.10350230e-01 6.04608767e-02 -2.52553783e-02
-1.72067910e-01 5.31553090e-01 -6.44371331e-01 1.67983520e+00
-3.29034299e-01 5.92730761e-01 -5.09851873e-01 -1.13711381e+00
9.26065564e-01 4.06303048e-01 3.95784050e-01 -4.79760617e-01
-7.44140670e-02 3.21974695e-01 -2.96009690e-01 -7.54845321e-01
7.54673481e-01 -3.09123769e-02 -2.47100338e-01 6.37900829e-01
1.39106274e-01 -4.52685446e-01 8.85376334e-01 6.81324065e-01
9.62200105e-01 5.74955605e-02 6.73290372e-01 -1.84757963e-01
7.14799225e-01 1.97579473e-01 9.24175978e-02 9.57857490e-01
5.09142697e-01 9.32361782e-01 7.98440278e-01 -1.56160807e-02
-1.02806580e+00 -4.96580273e-01 2.45911241e-01 1.11630619e+00
-2.41263449e-01 -8.51009607e-01 -1.08863878e+00 -9.27084804e-01
-3.82039398e-01 1.15989602e+00 -5.38125992e-01 -2.45819673e-01
-6.57676041e-01 -7.87133217e-01 5.04153907e-01 2.56878287e-01
2.02282727e-01 -1.42347980e+00 -7.49679148e-01 4.17266816e-01
-5.94688892e-01 -8.33961248e-01 -4.75136071e-01 1.35551348e-01
-1.01781809e+00 -7.10116208e-01 -1.10764492e+00 -7.19160318e-01
9.37424541e-01 7.41063356e-02 1.12326670e+00 -1.66502640e-01
-1.18030362e-01 2.69032776e-01 -6.85802698e-01 -6.69074655e-01
-8.19919288e-01 7.10634172e-01 -3.11477244e-01 -1.37110949e-01
1.32955432e-01 -3.68531674e-01 -4.27124470e-01 -1.97244450e-01
-1.26300299e+00 2.36336038e-01 1.07407808e+00 7.84625292e-01
5.74449778e-01 -1.12567648e-01 9.33140099e-01 -1.30521727e+00
1.24277520e+00 -2.80609578e-01 -9.20917168e-02 4.41526502e-01
-6.19089723e-01 5.62307298e-01 7.88931727e-01 -3.77560794e-01
-1.09741795e+00 -1.08532205e-01 -2.31420785e-01 1.17417887e-01
-2.27341533e-01 8.31954479e-01 3.39791290e-02 4.58253771e-01
8.82349789e-01 5.99621773e-01 -2.28751317e-01 -6.28827691e-01
1.86340749e-01 7.99869597e-01 2.61181027e-01 -3.25621575e-01
4.19277728e-01 1.65701836e-01 -1.21018246e-01 -1.17097247e+00
-1.09219587e+00 -6.55317545e-01 -6.88600004e-01 -8.67728889e-02
7.11338878e-01 -3.60909283e-01 5.01166359e-02 -7.69215673e-02
-1.34217644e+00 9.74270850e-02 -4.72292602e-01 3.65582943e-01
-5.62464297e-01 6.84286833e-01 -4.61496204e-01 -6.63557231e-01
-1.06370556e+00 -7.79596031e-01 1.49334371e+00 2.89930493e-01
-6.68759108e-01 -6.76650941e-01 3.03079426e-01 4.40817505e-01
1.66772977e-01 2.68456429e-01 9.92731750e-01 -1.17572808e+00
-4.46068309e-02 -4.16366577e-01 1.54516339e-01 1.31889969e-01
1.33134529e-01 -9.13092960e-03 -6.99410856e-01 -1.90302670e-01
-7.51630217e-02 -3.72257411e-01 1.22791266e+00 5.62439382e-01
1.00703025e+00 -5.38265407e-01 -5.96799254e-01 -7.89136514e-02
9.98365045e-01 1.99443743e-01 6.30960703e-01 3.10419768e-01
4.83454227e-01 8.45227420e-01 6.30381942e-01 3.61070454e-01
-3.11800819e-02 3.97140920e-01 -6.78585693e-02 -5.49881868e-02
4.07071523e-02 -2.58581787e-01 3.65589887e-01 9.90307748e-01
-9.21225920e-02 -5.20781457e-01 -6.76130533e-01 5.31569242e-01
-1.95546055e+00 -1.10807037e+00 -1.00708701e-01 1.97608542e+00
1.10852182e+00 3.63942057e-01 1.86968178e-01 1.88760430e-01
5.45226157e-01 2.90207267e-01 -1.96370080e-01 -5.22700787e-01
-2.86754989e-03 2.02795222e-01 7.23038390e-02 2.99637407e-01
-8.07021618e-01 9.75416124e-01 5.90419817e+00 1.06869853e+00
-1.03297973e+00 -1.73577160e-01 5.34810722e-01 -2.10507423e-01
-3.79064828e-01 -3.62403318e-02 -8.80163908e-01 4.00216907e-01
1.17040503e+00 -4.61790174e-01 -1.98969215e-01 6.77239776e-01
6.38042808e-01 -3.41701359e-01 -9.58897591e-01 5.17101347e-01
5.53083062e-01 -1.37831247e+00 5.87110996e-01 -6.12848774e-02
8.03426206e-01 -3.17660213e-01 -4.47323233e-01 2.10004419e-01
2.49330816e-03 -7.81751156e-01 5.27186036e-01 6.43130302e-01
4.24135387e-01 -7.62109458e-01 1.03151560e+00 6.58234060e-01
-6.09241724e-01 2.30974108e-01 -4.86329108e-01 4.76985306e-01
2.70129621e-01 8.40143561e-01 -1.04814279e+00 8.03265512e-01
3.37634504e-01 3.97543311e-01 -7.49970853e-01 9.67907548e-01
-4.20635194e-01 8.22111011e-01 -9.58853066e-02 -4.05429602e-01
4.15214390e-01 -5.77445440e-02 7.56320000e-01 1.66622901e+00
3.57441694e-01 1.98468119e-02 1.82199642e-01 7.01236844e-01
-1.67248100e-01 6.50379717e-01 -5.76147676e-01 -4.49345440e-01
1.08880386e-01 1.24457431e+00 -9.99718249e-01 -7.62019813e-01
-1.56504009e-02 1.03754294e+00 2.55129486e-02 2.86771655e-01
-2.63681889e-01 -7.91629076e-01 -5.50147295e-01 3.88292447e-02
3.48879009e-01 1.36677131e-01 -6.90117702e-02 -1.10239136e+00
7.12916371e-04 -8.97310257e-01 4.94446069e-01 -7.77252197e-01
-8.65105331e-01 8.25801313e-01 1.45525992e-01 -1.20169091e+00
-6.85948968e-01 7.28687868e-02 -8.01126480e-01 5.53903639e-01
-1.20451450e+00 -9.95021164e-01 5.10623418e-02 9.90877002e-02
1.16650367e+00 -1.54392228e-01 8.26214969e-01 -1.19159363e-01
-4.93416786e-01 1.40894964e-01 2.16841519e-01 -1.51905134e-01
8.07594359e-01 -1.44134998e+00 3.35538596e-01 7.29443610e-01
3.04695606e-01 7.02560902e-01 1.03871572e+00 -7.55676210e-01
-9.83860672e-01 -1.00265980e+00 1.52553296e+00 -1.46013007e-01
2.71643311e-01 -2.59406000e-01 -8.92165005e-01 2.32453972e-01
6.21489823e-01 -7.42162168e-01 7.62110531e-01 -5.96498698e-02
2.83284962e-01 1.13643579e-01 -8.96429658e-01 7.46506035e-01
5.96508503e-01 -7.56697357e-02 -1.18485188e+00 5.22426665e-01
6.49855018e-01 -1.51944891e-01 -4.20087904e-01 1.25322476e-01
2.24882871e-01 -7.66129494e-01 7.31739283e-01 -5.61796427e-01
8.86783123e-01 -7.24755451e-02 4.62890178e-01 -1.57054996e+00
-2.76559919e-01 -6.80013597e-01 -1.05585538e-01 1.48394537e+00
6.69071317e-01 -3.59365344e-01 7.03879833e-01 2.78164223e-02
-3.06375861e-01 -7.41627574e-01 -4.71607655e-01 -4.27076161e-01
-1.64160281e-01 1.90706730e-01 1.92023501e-01 5.53866982e-01
3.85430574e-01 1.03068411e+00 -2.19278902e-01 -3.54080558e-01
3.52609187e-01 3.83330286e-01 7.03061461e-01 -1.27881873e+00
3.14711547e-03 -6.71086013e-01 3.69328223e-02 -9.06736255e-01
8.54214802e-02 -8.08438778e-01 2.96769172e-01 -2.10743856e+00
4.82119679e-01 3.02051991e-01 6.15095235e-02 2.96012044e-01
-4.02230173e-01 -8.47176984e-02 -5.95620484e-04 3.75904053e-01
-8.70667875e-01 6.80136740e-01 1.12487745e+00 -2.69237638e-01
-6.36809528e-01 1.05061643e-01 -9.23570275e-01 6.95364058e-01
8.96886587e-01 -5.90508342e-01 -5.14069259e-01 5.58634289e-02
1.16254315e-01 6.98756129e-02 -1.98998168e-01 -9.13575768e-01
3.84285152e-01 -8.21997821e-02 4.20758367e-01 -1.02678609e+00
-7.43860975e-02 -1.40330091e-01 -2.13286445e-01 4.59116966e-01
-8.48104358e-01 -4.26564030e-02 8.54565501e-02 4.13341671e-01
-3.39087874e-01 -8.63799334e-01 4.65819329e-01 -3.64158422e-01
-1.83615759e-01 -1.66686997e-01 -6.12946391e-01 1.60100132e-01
6.00481212e-01 -4.66859974e-02 -2.21352324e-01 -4.74397838e-01
-4.82443929e-01 1.23962142e-01 2.42670476e-01 1.65937051e-01
6.45446241e-01 -8.50492537e-01 -9.67697978e-01 -8.91463459e-02
1.86616391e-01 1.84054121e-01 6.57396019e-02 8.61651778e-01
-5.64494371e-01 7.84369648e-01 9.66439173e-02 -3.76795083e-01
-1.36449099e+00 5.92508137e-01 -3.59822601e-01 -8.06650817e-01
-7.90371954e-01 3.23251247e-01 8.07200149e-02 -4.81362678e-02
4.84928936e-02 -1.96135700e-01 -8.35265994e-01 4.92503434e-01
6.50155008e-01 3.07388246e-01 2.12068051e-01 -3.51414293e-01
-2.40022759e-03 4.33424860e-01 -4.87418354e-01 -2.87780434e-01
1.51013601e+00 -6.75119534e-02 -1.76386461e-01 3.37425977e-01
9.81138051e-01 3.25677782e-01 -5.20646870e-01 -1.31405249e-01
4.53794837e-01 9.04545859e-02 -7.64560252e-02 -6.28454864e-01
-4.53708053e-01 6.90942049e-01 -1.38212547e-01 5.94306111e-01
1.11634433e+00 2.43766472e-01 8.75432551e-01 5.13037980e-01
-7.66477063e-02 -1.32900584e+00 2.01201588e-01 4.27981079e-01
1.12696409e+00 -1.00723577e+00 3.47723365e-01 -3.17451060e-01
-7.16675639e-01 1.20068145e+00 1.73333153e-01 -4.25945781e-02
-5.13972007e-02 -1.15858175e-01 -1.92447260e-01 -3.68249834e-01
-7.92126656e-01 -2.18959749e-01 5.24848282e-01 1.65425047e-01
6.32787049e-01 -4.09566551e-01 -9.09017861e-01 3.49652290e-01
-2.84971356e-01 2.89834179e-02 5.66204786e-01 1.18010771e+00
-7.07623005e-01 -1.07222140e+00 -4.12537456e-01 7.54015386e-01
-7.66629755e-01 -1.99094296e-01 -9.21460211e-01 5.20811915e-01
-3.35988313e-01 9.67303693e-01 -3.45202714e-01 1.43987924e-01
3.24630260e-01 3.39317888e-01 3.02649051e-01 -9.73472655e-01
-5.59524655e-01 5.17552912e-01 3.11488837e-01 9.95434541e-03
-5.47873437e-01 -7.48246312e-01 -1.16835833e+00 1.84371740e-01
-4.32089120e-01 7.29484141e-01 4.56863284e-01 1.11949861e+00
5.17951310e-01 6.58468544e-01 5.49629867e-01 -1.01816928e+00
-5.34428895e-01 -1.37821913e+00 -1.81099355e-01 3.94558609e-01
1.83874875e-01 -1.97707504e-01 -2.90344387e-01 4.86211509e-01]
|
[12.480443954467773, 9.492705345153809]
|
44a45a50-164f-4157-82df-8a357ab7875e
|
towards-a-unified-view-on-visual-parameter
|
2210.00788
| null |
https://arxiv.org/abs/2210.00788v2
|
https://arxiv.org/pdf/2210.00788v2.pdf
|
Towards a Unified View on Visual Parameter-Efficient Transfer Learning
|
Parameter efficient transfer learning (PETL) aims at making good use of the representation knowledge in the pre-trained large models by fine-tuning a small number of parameters. Recently, taking inspiration from the natural language processing (NLP) domain, popular PETL techniques such as prompt-tuning and Adapter have also been successfully applied to the vision domain. However, prefix-tuning remains under-explored for vision tasks. In this work, we intend to adapt large vision models (LVMs) to downstream tasks with a good parameter-accuracy trade-off. Towards this goal, we propose a framework with a unified view of PETL called visual-PETL (V-PETL) to investigate the effects of different PETL techniques, data scales of downstream domains, positions of trainable parameters, and other aspects affecting the trade-off. Specifically, we analyze the positional importance of trainable parameters and differences between NLP and vision tasks in terms of data structures and pre-training mechanisms while implementing various PETL techniques, especially for the under-explored prefix-tuning technique. Based on a comprehensive understanding of the differences between NLP and vision data, we propose a new variation of the prefix-tuning module called parallel attention (PATT) for vision downstream tasks. An extensive empirical analysis on vision tasks via different frozen LVMs has been carried and the findings show that the proposed PATT can effectively contribute to other PETL techniques. An effective scheme Swin-BAPAT derived from the proposed V-PETL framework achieves significantly better performance than the state-of-the-art AdaptFormer-Swin with slightly more parameters and outperforms full-tuning with far fewer parameters. Code and data are available at: https://github.com/bruceyo/V-PETL.
|
['Chang Wen Chen', 'Qi Tian', 'Lingbo Liu', 'Jianlong Chang', 'Bruce X. B. Yu']
|
2022-10-03
| null | null | null | null |
['video-recognition']
|
['computer-vision']
|
[-5.15556335e-02 -2.03253537e-01 -1.48038238e-01 -3.51655960e-01
-7.06103623e-01 -5.63552499e-01 7.32548594e-01 -1.55170739e-01
-7.83187628e-01 4.95962918e-01 1.62911564e-01 -3.29591393e-01
-2.30570763e-01 -4.51256424e-01 -9.03620064e-01 -7.86997736e-01
3.63697708e-01 4.49849039e-01 4.76610631e-01 -2.05038756e-01
3.83796364e-01 4.57179248e-01 -1.43173897e+00 2.77625263e-01
7.56330431e-01 6.78674281e-01 6.29181981e-01 4.63109463e-01
-3.07705253e-01 3.31462622e-01 -2.38019630e-01 -3.37470531e-01
4.33827430e-01 -1.88166142e-01 -7.42497921e-01 -1.58643290e-01
5.17542779e-01 -2.32797027e-01 -4.88506034e-02 8.17438960e-01
6.99431360e-01 7.69239515e-02 7.25968480e-01 -1.26335394e+00
-8.27629745e-01 5.12501240e-01 -6.22539222e-01 3.05315763e-01
-2.12715268e-01 5.96877396e-01 8.24496269e-01 -9.78855431e-01
4.23247039e-01 1.38341784e+00 8.32273364e-01 6.46619320e-01
-1.16049314e+00 -5.70380092e-01 2.90772647e-01 4.97509271e-01
-1.26050663e+00 -3.79598975e-01 4.80846286e-01 -5.92683971e-01
1.22154498e+00 -2.67379820e-01 3.91118348e-01 1.25159252e+00
2.33185619e-01 7.03141212e-01 1.20459235e+00 -7.02898622e-01
1.04896031e-01 4.42762583e-01 2.83872545e-01 6.71273112e-01
1.24867454e-01 2.13540897e-01 -5.34362733e-01 1.04133405e-01
8.34835649e-01 -2.96586126e-01 -3.39965433e-01 -4.62202460e-01
-1.09027827e+00 9.32906151e-01 5.06445050e-01 4.18739289e-01
-1.59083053e-01 3.06009293e-01 5.40237248e-01 3.65622431e-01
2.98470080e-01 4.51222122e-01 -8.95654440e-01 7.14475289e-02
-7.88361371e-01 -2.04782113e-02 4.87085253e-01 9.10229266e-01
9.46656108e-01 -7.30469525e-02 -6.17904007e-01 9.37957823e-01
4.83554661e-01 3.15904975e-01 7.73281634e-01 -8.01703036e-01
5.34642518e-01 5.14849007e-01 -7.19405562e-02 -3.94920826e-01
-4.49379861e-01 -4.34317917e-01 -5.99724591e-01 2.43609250e-01
4.76860732e-01 -1.33286878e-01 -1.09389329e+00 1.91190994e+00
3.27084064e-01 1.08775653e-01 6.41280934e-02 8.41337025e-01
8.20181012e-01 7.39493787e-01 5.15148342e-01 3.96001823e-02
1.49921095e+00 -1.37841654e+00 -3.31888914e-01 -3.90289217e-01
6.32589698e-01 -7.85323501e-01 1.60379601e+00 8.93955529e-02
-7.50976980e-01 -9.75752175e-01 -7.58323193e-01 -2.91355222e-01
-4.70246702e-01 3.55541289e-01 4.17241126e-01 5.02637088e-01
-1.36398530e+00 5.47521472e-01 -6.87879086e-01 -8.26467454e-01
4.89573389e-01 3.73223394e-01 -1.54834867e-01 -9.38721001e-02
-1.02545071e+00 1.02267027e+00 5.10862768e-01 8.95199403e-02
-9.97560263e-01 -9.30312872e-01 -4.99784976e-01 7.77211934e-02
4.95007664e-01 -1.08370984e+00 1.28522193e+00 -1.06175673e+00
-1.70127761e+00 8.42131913e-01 1.02289841e-01 -4.43879217e-01
4.83073711e-01 -4.54826325e-01 1.82265937e-01 -2.28358107e-03
-6.00429960e-02 1.10544813e+00 1.15139639e+00 -1.13814771e+00
-3.69400769e-01 -2.88541436e-01 1.46930099e-01 3.00178885e-01
-4.35945958e-01 -1.16734587e-01 -7.07032621e-01 -4.23246264e-01
-6.42034173e-01 -9.19493258e-01 -1.25016004e-01 1.26061797e-01
-2.53537688e-02 -4.54811931e-01 7.35810101e-01 -2.49216020e-01
9.65520203e-01 -2.07046080e+00 1.57916769e-01 -2.89747149e-01
7.31429309e-02 8.79387796e-01 -6.60128474e-01 4.87743050e-01
2.15265438e-01 -8.31021294e-02 -1.00498281e-01 -4.59618717e-01
9.14945267e-03 4.31593955e-01 -1.45182267e-01 1.93973914e-01
1.66536406e-01 1.18710017e+00 -5.26160121e-01 -5.18317044e-01
3.99011672e-01 5.36270082e-01 -6.02023959e-01 2.88118988e-01
-4.46739823e-01 5.49698532e-01 -6.06982768e-01 4.18689191e-01
4.17563081e-01 -2.83207268e-01 -2.41553411e-01 -5.07594764e-01
-3.63011777e-01 -1.21374100e-01 -7.20455110e-01 1.76158369e+00
-5.45592248e-01 5.47202110e-01 -4.36091945e-02 -9.37451422e-01
8.05082023e-01 8.46743286e-02 6.51390776e-02 -8.15790355e-01
2.82421589e-01 4.68584560e-02 -5.53991981e-02 -7.89436102e-01
2.09675819e-01 -1.26275510e-01 2.67864913e-01 1.28837198e-01
4.79497343e-01 5.58496192e-02 1.45168051e-01 -5.45270033e-02
9.24651623e-01 5.45840621e-01 4.42013472e-01 -2.43819982e-01
7.40508199e-01 9.59726796e-02 4.26521778e-01 8.85912716e-01
-4.76313382e-01 3.79022092e-01 2.12448537e-01 -2.11782366e-01
-1.08181131e+00 -7.30228484e-01 -4.32548746e-02 1.46332872e+00
-5.17938510e-02 -2.43469581e-01 -9.28135157e-01 -6.01824760e-01
3.90186831e-02 6.71409547e-01 -6.70606911e-01 -1.54719844e-01
-4.43486243e-01 -6.96266949e-01 6.01382613e-01 6.36408031e-01
7.47672379e-01 -1.38343835e+00 -7.43338823e-01 -8.31120554e-03
-5.16266190e-03 -1.23568940e+00 -4.88361686e-01 2.17682764e-01
-9.50705945e-01 -7.96228170e-01 -8.62961173e-01 -7.24531114e-01
3.76697004e-01 2.80322999e-01 1.10363686e+00 -2.12290391e-01
-1.31840870e-01 6.47409260e-01 -5.53878844e-01 -5.49303114e-01
-3.69609207e-01 3.46285522e-01 -1.03813723e-01 3.04272268e-02
4.60865587e-01 -4.16065872e-01 -6.97478592e-01 3.75109732e-01
-7.78479278e-01 2.10097805e-01 1.02946043e+00 7.50267327e-01
6.92755580e-01 -5.06577313e-01 5.03000021e-01 -8.57440770e-01
6.02052033e-01 -3.06022674e-01 -7.04981148e-01 5.84346056e-01
-7.78100014e-01 4.74600911e-01 6.31093979e-01 -6.43233895e-01
-1.16073406e+00 -6.40461966e-02 -1.77760571e-01 -8.22116137e-01
-3.23754340e-01 2.88452804e-01 -2.25049883e-01 -2.40714148e-01
6.59543872e-01 2.03870028e-01 -1.32448614e-01 -6.04289412e-01
7.22601593e-01 4.76308614e-01 3.31116050e-01 -6.30479276e-01
7.44033635e-01 3.29020023e-01 -2.00808927e-01 -7.23963976e-01
-1.02264524e+00 -5.00076532e-01 -7.19121754e-01 2.29498800e-02
8.98884654e-01 -7.42794156e-01 -4.20528352e-01 6.74175501e-01
-1.01740932e+00 -7.71268785e-01 -3.39147657e-01 4.80157048e-01
-6.56500280e-01 3.54435116e-01 -4.52948689e-01 -4.44408894e-01
-6.42652810e-01 -1.28605270e+00 9.52762008e-01 3.52791488e-01
4.15742546e-02 -1.05369771e+00 3.03598344e-01 5.24356842e-01
6.92596495e-01 -3.20812196e-01 1.08073032e+00 -6.88576996e-01
-4.90895927e-01 4.50291365e-01 -5.64495742e-01 4.66331273e-01
-1.27790257e-01 -1.41834348e-01 -1.20383739e+00 -4.79221076e-01
1.33230593e-02 -5.22625506e-01 1.10672891e+00 6.93814874e-01
9.80412126e-01 -1.96855113e-01 -2.70075470e-01 9.80744600e-01
1.74715626e+00 3.69003718e-03 4.72145051e-01 7.00634181e-01
7.64369249e-01 6.70666158e-01 6.77890420e-01 3.07438791e-01
3.43160212e-01 5.92114210e-01 4.32108939e-01 -1.13985632e-02
-3.28879237e-01 -2.21791908e-01 5.24799764e-01 6.31384671e-01
-1.20252199e-01 -2.35920623e-01 -8.31767559e-01 4.90048140e-01
-1.91486371e+00 -5.30191362e-01 1.21766590e-01 2.05520248e+00
7.87505627e-01 5.21052256e-02 5.60162216e-02 -3.01253587e-01
5.15228868e-01 1.05086036e-01 -7.76433051e-01 -5.77441335e-01
1.05108112e-01 1.15997501e-01 6.58191800e-01 4.48850930e-01
-9.41304088e-01 1.29822385e+00 5.47634602e+00 1.07630241e+00
-1.16765273e+00 4.44776922e-01 2.88868636e-01 2.63352543e-02
-1.78607538e-01 5.94035052e-02 -1.27863085e+00 2.42974043e-01
1.02744949e+00 1.50345303e-02 3.57116699e-01 8.26113224e-01
4.08998013e-01 2.12523364e-03 -1.18977523e+00 9.45959210e-01
-5.44002687e-05 -1.20342720e+00 3.68654281e-01 -1.32508859e-01
4.65355039e-01 6.57968223e-01 9.03792381e-02 6.83787823e-01
7.97930360e-02 -7.25502133e-01 7.54795134e-01 3.75784606e-01
7.49219120e-01 -3.51496786e-01 5.87855697e-01 3.45816582e-01
-1.02670097e+00 -3.31181109e-01 -6.62812889e-01 1.42964229e-01
-1.16100699e-01 3.93947780e-01 -7.82989860e-01 4.26168710e-01
9.21652257e-01 6.50826991e-01 -7.25327492e-01 1.05306852e+00
-3.30020338e-01 7.56695986e-01 -1.76671237e-01 4.14457545e-02
5.70112169e-01 -2.11698472e-01 3.40031266e-01 1.42357659e+00
1.93526968e-01 -1.15768299e-01 -7.83905834e-02 1.01814544e+00
1.13328382e-01 1.46237075e-01 -4.28261220e-01 6.23025298e-02
3.95118684e-01 1.32997918e+00 -4.45597112e-01 -1.86775297e-01
-6.32512748e-01 7.70185888e-01 6.45611823e-01 5.69908082e-01
-9.39808369e-01 1.28695488e-01 5.54350853e-01 8.08793493e-03
6.54674947e-01 -2.30222374e-01 -1.03567779e-01 -9.59947288e-01
-1.86383665e-01 -7.22604811e-01 3.30226481e-01 -9.19839740e-01
-1.46870124e+00 7.03890622e-01 2.30173439e-01 -9.86145020e-01
2.17549168e-02 -8.23180079e-01 -5.88922679e-01 7.66662419e-01
-1.98458266e+00 -1.58764780e+00 -4.51378196e-01 9.34394300e-01
9.21863019e-01 -2.52314746e-01 5.63403726e-01 -7.32413260e-03
-6.89151645e-01 7.24135578e-01 2.05698594e-01 -2.46292204e-01
1.03451288e+00 -1.01588011e+00 3.08034778e-01 7.30818748e-01
1.58096373e-01 4.69872355e-01 5.67280233e-01 -2.98283905e-01
-1.19655967e+00 -1.20235300e+00 6.07786894e-01 -4.29093510e-01
7.02994287e-01 -1.57247603e-01 -1.03070974e+00 5.99646568e-01
4.33960289e-01 -1.06610164e-01 3.98008764e-01 -3.58502492e-02
-5.16496956e-01 -2.48620138e-01 -9.97573495e-01 4.49504465e-01
9.39099133e-01 -3.19817305e-01 -8.15902770e-01 1.53238490e-01
8.09927404e-01 -4.27339226e-02 -5.02914727e-01 3.47255260e-01
2.99038470e-01 -1.03440988e+00 1.03216946e+00 -3.23419780e-01
2.70742357e-01 -1.71923071e-01 -1.16792373e-01 -1.33182669e+00
-6.38140202e-01 -4.88953650e-01 7.48485997e-02 1.44492638e+00
2.59855300e-01 -8.07793736e-01 4.27854747e-01 3.44909132e-01
-2.34194845e-01 -8.66283417e-01 -7.74987757e-01 -7.74042368e-01
3.33394259e-01 -3.73367637e-01 1.33752376e-01 5.76195478e-01
-7.11423278e-01 6.40830934e-01 -2.50166833e-01 6.43846467e-02
5.93177021e-01 3.14783938e-02 7.89548397e-01 -1.01335943e+00
-6.21367991e-01 -4.12215590e-01 7.36653060e-02 -9.61422741e-01
5.90427294e-02 -7.70590007e-01 1.66433081e-02 -1.55611491e+00
3.00908834e-01 -4.25497442e-01 -4.73262966e-01 9.27002966e-01
-2.03254178e-01 8.39208215e-02 4.98637319e-01 5.21889806e-01
-5.13806045e-01 7.33941615e-01 1.24576199e+00 2.28679972e-04
-2.83716202e-01 -1.19699888e-01 -5.74215949e-01 8.01154733e-01
8.68434012e-01 -4.08709407e-01 -7.17674017e-01 -7.95131207e-01
-4.36906666e-02 -4.03137207e-01 3.71517539e-01 -9.29418147e-01
2.00986490e-01 -1.62268624e-01 8.50486308e-02 -3.43712807e-01
2.18893319e-01 -6.85059071e-01 -2.89171785e-01 3.87892574e-01
-2.57768929e-01 -2.72409525e-02 5.67836463e-01 5.34048617e-01
-2.57942895e-03 -3.68067950e-01 9.99675155e-01 -2.04080462e-01
-1.18906319e+00 2.36219168e-01 -2.41182163e-01 1.11859165e-01
9.65606451e-01 -3.10429603e-01 -5.50837934e-01 1.43978074e-01
-5.16486883e-01 3.25793833e-01 3.33638757e-01 4.50573146e-01
4.40226436e-01 -8.84577274e-01 -6.50067925e-01 2.04542130e-01
3.16907525e-01 -1.69641688e-01 3.28570098e-01 1.02905726e+00
-3.12850624e-01 6.70905590e-01 -5.11348128e-01 -6.26539469e-01
-1.25245106e+00 7.45039284e-01 2.99047738e-01 -4.50808614e-01
-5.65429449e-01 1.02309310e+00 7.09060669e-01 -5.14048219e-01
2.71243870e-01 -5.38215935e-01 -2.91271836e-01 4.38336432e-02
2.61549503e-01 2.26461425e-01 -9.43779293e-03 -3.25822771e-01
-3.04506361e-01 1.04758215e+00 -1.41711846e-01 7.06373900e-02
1.37195706e+00 -4.12176192e-01 1.40230104e-01 2.80359894e-01
1.03088009e+00 -4.07840401e-01 -1.69691050e+00 -4.36024606e-01
-1.68269709e-01 -9.03975368e-02 1.24503307e-01 -9.08934176e-01
-1.00458109e+00 1.06015873e+00 7.95363724e-01 -3.42543930e-01
1.27898765e+00 1.00361399e-01 5.65135479e-01 3.42291862e-01
4.18555170e-01 -9.63710189e-01 1.95723161e-01 6.28064454e-01
9.35538471e-01 -1.30902529e+00 -3.12342912e-01 -1.20119803e-01
-7.40205288e-01 9.29451108e-01 8.64228070e-01 -1.12214424e-01
4.89859819e-01 2.01170400e-01 2.04493359e-01 -7.99734741e-02
-8.31548691e-01 -4.20714498e-01 2.72548795e-01 7.65443265e-01
2.30669156e-01 -2.77812243e-01 -2.84837514e-01 3.70325744e-01
1.30468354e-01 2.83745915e-01 1.29753187e-01 5.58559358e-01
-4.50015128e-01 -1.26691377e+00 -3.70836377e-01 1.09306619e-01
-1.90154836e-01 -2.19259530e-01 -2.65754938e-01 9.37839210e-01
3.43027949e-01 6.88071012e-01 -2.40587786e-01 -1.69165522e-01
2.96328515e-01 9.56584439e-02 6.62692249e-01 -5.85468411e-01
-8.49526405e-01 1.32327661e-01 -2.10555911e-01 -6.67225480e-01
-7.13104427e-01 -2.71592885e-01 -9.99025822e-01 -5.03973253e-02
-2.48901486e-01 -1.26276538e-01 5.97394764e-01 1.07883012e+00
5.51472604e-01 5.54167032e-01 1.31543070e-01 -1.12607801e+00
-7.82426357e-01 -1.03394163e+00 -2.82901287e-01 1.91507787e-01
1.81921065e-01 -7.28733659e-01 -2.60928988e-01 1.73910353e-02]
|
[10.176244735717773, 1.9558099508285522]
|
07bac89c-7d88-458e-a1cc-2e0f2b5ff6cc
|
opp-miner-order-preserving-sequential-pattern
|
2202.03140
| null |
https://arxiv.org/abs/2202.03140v2
|
https://arxiv.org/pdf/2202.03140v2.pdf
|
OPP-Miner: Order-preserving sequential pattern mining
|
A time series is a collection of measurements in chronological order. Discovering patterns from time series is useful in many domains, such as stock analysis, disease detection, and weather forecast. To discover patterns, existing methods often convert time series data into another form, such as nominal/symbolic format, to reduce dimensionality, which inevitably deviates the data values. Moreover, existing methods mainly neglect the order relationships between time series values. To tackle these issues, inspired by order-preserving matching, this paper proposes an Order-Preserving sequential Pattern (OPP) mining method, which represents patterns based on the order relationships of the time series data. An inherent advantage of such representation is that the trend of a time series can be represented by the relative order of the values underneath the time series data. To obtain frequent trends in time series, we propose the OPP-Miner algorithm to mine patterns with the same trend (sub-sequences with the same relative order). OPP-Miner employs the filtration and verification strategies to calculate the support and uses pattern fusion strategy to generate candidate patterns. To compress the result set, we also study finding the maximal OPPs. Experiments validate that OPP-Miner is not only efficient and scalable but can also discover similar sub-sequences in time series. In addition, case studies show that our algorithms have high utility in analyzing the COVID-19 epidemic by identifying critical trends and improve the clustering performance.
|
['Xindong Wu', 'Xingquan Zhu', 'Lei Guo', 'Yan Li', 'Qian Hu', 'Youxi Wu']
|
2022-01-09
| null | null | null | null |
['sequential-pattern-mining']
|
['natural-language-processing']
|
[ 9.17155147e-02 -7.82656193e-01 -2.22416341e-01 -1.52948156e-01
3.18618417e-01 -5.60835898e-01 3.38700980e-01 4.78801519e-01
-9.40041840e-02 5.89582324e-01 2.64456511e-01 -3.86785179e-01
-8.34689438e-01 -1.04007053e+00 -1.32723823e-01 -5.83384395e-01
-6.75073504e-01 3.47442210e-01 3.11834514e-01 -2.08598375e-01
3.83867979e-01 4.71923620e-01 -1.88529158e+00 4.42394853e-01
7.70526767e-01 1.22353005e+00 -6.36784211e-02 -1.72906853e-02
-5.22464216e-01 5.62534034e-01 -7.35411108e-01 9.85502154e-02
6.29144013e-01 -4.87013191e-01 -8.49566683e-02 -7.28002638e-02
-7.54515767e-01 1.33328419e-02 -1.74892649e-01 1.10930395e+00
2.79231742e-02 -1.22721858e-01 3.63539279e-01 -1.76646805e+00
1.35908015e-02 7.81172514e-01 -9.18421090e-01 5.52002549e-01
3.26876700e-01 -3.22192371e-01 7.09474981e-01 -3.62694204e-01
4.97747093e-01 1.07017589e+00 8.29341352e-01 -1.69399872e-01
-8.55751336e-01 -9.96547043e-01 2.02203579e-02 5.85691154e-01
-1.52919197e+00 1.47805288e-01 1.02515650e+00 -2.85573006e-01
6.86171532e-01 6.56109333e-01 9.30219889e-01 2.79418945e-01
3.89939517e-01 4.35182154e-01 1.19055462e+00 -2.18151972e-01
2.50510246e-01 -2.75458187e-01 3.70832264e-01 -3.85491177e-02
4.21711117e-01 1.14535756e-01 -2.63165385e-01 -6.07452810e-01
2.68398970e-01 7.83321679e-01 -2.41297558e-01 1.55287609e-01
-1.18374670e+00 6.13430381e-01 -7.82956555e-02 8.09151947e-01
-5.95940053e-01 -5.59069455e-01 6.38925791e-01 9.54735041e-01
3.31582904e-01 5.53418361e-02 -6.10255361e-01 -2.17317030e-01
-1.09188485e+00 4.55111027e-01 7.50318766e-01 6.56449020e-01
5.43742836e-01 -9.85861197e-03 1.45280674e-01 2.24704713e-01
1.36624128e-01 4.79348511e-01 9.36892092e-01 -4.28182244e-01
3.29606205e-01 1.14066482e+00 -6.03872687e-02 -1.61246097e+00
-5.09860814e-01 -3.36308628e-01 -1.08102119e+00 -2.04690650e-01
1.12539716e-01 2.59417333e-02 -5.87385714e-01 1.40345609e+00
3.15152228e-01 4.73745286e-01 1.67834922e-03 4.31191385e-01
3.08637619e-01 9.25161600e-01 -3.13033223e-01 -1.20104408e+00
1.24025917e+00 5.45774922e-02 -8.14847231e-01 6.26852214e-01
4.29429173e-01 -7.48775959e-01 3.35336059e-01 3.99566084e-01
-5.60033798e-01 -3.94279957e-01 -8.31852973e-01 9.72795188e-01
-2.84847587e-01 -2.72075713e-01 3.36538315e-01 3.39421928e-01
-5.19717515e-01 7.91096926e-01 -6.59530878e-01 -2.99259961e-01
-1.37773305e-01 1.91257641e-01 -1.10404648e-01 3.05502713e-01
-1.24746788e+00 3.72288913e-01 8.73712540e-01 -1.76025391e-01
-3.98572870e-02 -9.73807216e-01 -4.07590985e-01 8.65226388e-02
3.08836818e-01 -8.47153962e-02 7.17004359e-01 -8.46696377e-01
-7.22637713e-01 2.81992495e-01 -3.26170653e-01 -7.55318463e-01
3.17701727e-01 4.94731903e-01 -1.40116966e+00 -3.79774533e-02
9.31544676e-02 -4.22659338e-01 6.84328139e-01 -6.15868926e-01
-9.87596393e-01 -4.36616063e-01 -6.26391709e-01 -1.84682950e-01
-5.30086279e-01 3.02592963e-01 1.15945429e-01 -8.46132576e-01
4.60658193e-01 -6.38427079e-01 -3.25788140e-01 -6.22152627e-01
-1.46970958e-01 -4.95770782e-01 1.33467078e+00 -6.15886986e-01
2.07454681e+00 -2.34834409e+00 -2.77659178e-01 1.04664934e+00
9.30354968e-02 -2.21502893e-02 2.99011379e-01 8.90641928e-01
-2.91042209e-01 1.20998055e-01 -5.89288831e-01 4.09700423e-01
-2.93641359e-01 5.52676737e-01 -9.48400915e-01 4.46384579e-01
-1.98836729e-01 3.20717931e-01 -6.40307486e-01 -4.07890260e-01
4.94206436e-02 -1.63755760e-01 -4.00678106e-02 -1.71933242e-03
-1.30431250e-01 2.91876018e-01 -3.93440723e-01 5.17842293e-01
7.95346916e-01 -2.76764259e-02 3.06592196e-01 -2.69789606e-01
-7.49244213e-01 -5.02717346e-02 -1.49720526e+00 7.63760567e-01
3.16999793e-01 1.05082065e-01 -3.25854123e-01 -1.23148263e+00
1.43336916e+00 4.58729953e-01 1.26801813e+00 -8.18124056e-01
1.39768541e-01 4.22318429e-01 2.85332799e-01 -4.97563779e-01
3.40358526e-01 -7.92553425e-02 -1.53408229e-01 7.36015141e-01
-7.53299713e-01 3.82058024e-01 6.57424688e-01 -1.84539869e-01
1.17053771e+00 -4.99440670e-01 5.70013344e-01 -4.01498854e-01
6.21194661e-01 4.06166911e-01 1.14808142e+00 9.68722254e-02
1.45769700e-01 8.48432407e-02 5.31239748e-01 -9.15942729e-01
-1.18367219e+00 -6.92288280e-01 -1.62686244e-01 3.28083664e-01
-9.41483863e-03 -6.30667150e-01 1.48053840e-01 -2.69322008e-01
2.64578581e-01 3.77572447e-01 -4.93597031e-01 1.28876284e-01
-8.39652300e-01 -1.05430174e+00 2.93705583e-01 2.23358467e-01
4.13120747e-01 -1.03858185e+00 -7.91246176e-01 5.55581391e-01
-2.24252176e-02 -5.56341410e-01 -2.92125255e-01 -9.36352164e-02
-9.59837914e-01 -1.21239758e+00 -2.60154903e-01 -5.10909200e-01
5.91811717e-01 3.44225675e-01 6.73567653e-01 1.57531947e-01
-1.10414294e-04 -2.20980093e-01 -6.73506439e-01 -6.88315630e-01
-3.06612134e-01 -3.35985094e-01 2.36934289e-01 2.71817654e-01
7.79148221e-01 -1.11675048e+00 -3.81340712e-01 4.78021622e-01
-1.02771807e+00 -2.55941868e-01 2.14586481e-01 3.68994355e-01
7.39677668e-01 1.16656613e+00 6.15112424e-01 -3.82360905e-01
1.01837671e+00 -9.23734903e-01 -6.31376803e-01 2.95962721e-01
-9.55335200e-01 -1.98426247e-01 9.24641490e-01 -4.75766599e-01
-6.22709513e-01 -1.57750234e-01 3.44671607e-01 -6.15243495e-01
1.98456362e-01 1.12323678e+00 1.69367492e-01 4.79412407e-01
1.67644903e-01 8.32953811e-01 2.38738239e-01 -5.90850413e-01
-3.08583409e-01 6.75908923e-01 3.88345569e-01 -2.62968689e-01
8.85453880e-01 6.42203867e-01 2.71410525e-01 -5.90535164e-01
-2.25961078e-02 -7.48512983e-01 -2.74374515e-01 -1.23173408e-01
2.60537773e-01 -5.72479725e-01 -6.82614207e-01 2.93313354e-01
-8.36136758e-01 5.97206354e-01 -3.24415684e-01 5.98756909e-01
-3.27504911e-02 5.51597893e-01 -3.61069411e-01 -9.76509809e-01
-4.26651657e-01 -3.54929477e-01 2.61946380e-01 2.12006554e-01
-5.49531817e-01 -6.79068685e-01 2.84599692e-01 -6.46788001e-01
3.35195541e-01 6.62468314e-01 1.13904595e+00 -1.13940430e+00
-1.18610248e-01 -2.38664076e-01 9.36671793e-02 -2.40074188e-01
4.06958580e-01 2.38129437e-01 -7.42593110e-02 -3.21464121e-01
3.16308230e-01 5.79242826e-01 4.29279685e-01 1.92064762e-01
1.10586572e+00 -6.29668593e-01 -6.01289272e-01 3.23446333e-01
1.19947052e+00 1.04436255e+00 4.85188812e-01 4.43779469e-01
2.08856672e-01 8.58860254e-01 7.53049731e-01 8.81594121e-01
1.58577263e-01 4.64978635e-01 4.07787003e-02 4.24345374e-01
4.92303222e-01 -2.62737572e-01 2.04487458e-01 1.22764683e+00
-8.93687159e-02 8.50183070e-02 -8.43208313e-01 6.78275228e-01
-2.07392573e+00 -1.52229464e+00 -3.84855419e-01 2.06614709e+00
7.60135531e-01 2.36217916e-01 5.20375729e-01 1.10474420e+00
8.33941698e-01 1.33924801e-02 -4.44396347e-01 -3.35195929e-01
-4.00381774e-01 -5.34440018e-02 3.32487345e-01 -1.31460592e-01
-7.13575065e-01 3.87227200e-02 5.46508551e+00 8.13993037e-01
-1.23757041e+00 -2.38286436e-01 3.43968511e-01 1.59135416e-01
-5.59464276e-01 2.42889710e-02 -5.09858787e-01 1.01099563e+00
8.41586351e-01 -9.99014556e-01 1.45591736e-01 6.32116735e-01
5.92423081e-01 3.35635781e-01 -5.65485656e-01 9.85422969e-01
-3.24749112e-01 -1.17810905e+00 5.99832796e-02 1.65149331e-01
6.27084851e-01 -2.36315385e-01 -2.17789069e-01 -2.37495564e-02
1.66241564e-02 -5.96363604e-01 5.50594032e-01 5.49390495e-01
3.28363508e-01 -8.55614185e-01 7.65638471e-01 4.68041778e-01
-1.82823479e+00 -3.83911401e-01 -2.05091342e-01 -4.57331389e-01
3.63850474e-01 9.93744135e-01 -6.41474009e-01 1.08604193e+00
9.31405365e-01 1.09088027e+00 5.26952483e-02 1.39017224e+00
3.23044240e-01 6.96406424e-01 -8.37717354e-01 -5.94165325e-02
-1.54921906e-02 -5.61254978e-01 6.97454751e-01 7.96993911e-01
9.09337759e-01 3.17507416e-01 3.67304742e-01 3.94050270e-01
3.33842486e-01 2.86997527e-01 -5.04135668e-01 -1.69138491e-01
1.01054823e+00 6.78527832e-01 -9.29129481e-01 -3.56170654e-01
-2.52134681e-01 7.59413242e-02 -6.71527326e-01 1.01283900e-01
-5.28579295e-01 -4.58140731e-01 5.41862190e-01 3.72185707e-01
3.03940058e-01 -3.52491409e-01 -2.83562630e-01 -7.97283888e-01
3.43559802e-01 -9.73352432e-01 8.99970710e-01 -1.65067717e-01
-1.44989002e+00 6.80918157e-01 3.56106639e-01 -1.98797417e+00
-4.53021049e-01 -1.09616488e-01 -9.60863531e-01 6.28786206e-01
-1.16777956e+00 -5.67648470e-01 -5.18370382e-02 9.16382790e-01
3.22522432e-01 -2.27895662e-01 5.54194391e-01 4.08338547e-01
-4.16966110e-01 8.72714445e-02 3.02025318e-01 2.77809538e-02
8.53295028e-02 -5.95636249e-01 2.25721866e-01 9.95985568e-01
9.18958895e-03 6.91347361e-01 9.16640997e-01 -1.11318743e+00
-1.10875523e+00 -9.62299347e-01 1.02244270e+00 2.82789171e-01
9.21036303e-01 2.09660009e-01 -1.12084913e+00 2.84731597e-01
-4.98685837e-02 -3.02969933e-01 7.67657816e-01 -1.60031959e-01
-2.91093826e-01 -5.39877176e-01 -1.24240792e+00 4.91043895e-01
9.10281897e-01 -6.15300685e-02 -1.03859675e+00 -1.43192604e-01
6.58751965e-01 2.34436095e-01 -1.02009571e+00 7.25348234e-01
5.63039243e-01 -8.52437973e-01 7.06272423e-01 -3.13291967e-01
-7.47154653e-02 -8.65675271e-01 -1.14539891e-01 -1.06533718e+00
-3.90935719e-01 -7.55121112e-01 -2.31861278e-01 1.23484182e+00
8.18964392e-02 -7.99659133e-01 5.54885745e-01 9.93326008e-02
2.17257798e-01 -4.95713979e-01 -1.19823408e+00 -1.08841670e+00
-6.16617858e-01 -3.24502587e-01 1.54259634e+00 1.25651920e+00
3.50359321e-01 -2.80971795e-01 -5.14150500e-01 -7.70416530e-03
6.88249767e-01 8.64091218e-01 4.31261331e-01 -1.73265910e+00
-3.22723724e-02 -6.17764652e-01 -5.21850109e-01 -2.80549258e-01
-2.41042301e-01 -6.20885193e-01 -5.09521186e-01 -1.04438639e+00
-2.72862554e-01 -5.91042817e-01 -5.90122998e-01 3.04962903e-01
3.22233617e-01 -1.44351527e-01 7.51279294e-02 8.71869624e-01
-1.57764047e-01 3.47108334e-01 7.75454581e-01 -6.23752661e-02
-5.30975997e-01 2.64985651e-01 -2.17183262e-01 5.11720002e-01
1.01953459e+00 -6.16427660e-01 -5.14255226e-01 2.36850098e-01
3.51805687e-01 1.72170937e-01 -1.63698375e-01 -9.30539846e-01
5.35100460e-01 -5.72819591e-01 2.15341985e-01 -1.23334479e+00
-3.29376757e-01 -1.35983598e+00 1.12868941e+00 1.08152854e+00
1.75651237e-01 8.24425399e-01 1.69800028e-01 4.97609764e-01
-4.99949723e-01 -1.08091310e-02 3.57183032e-02 8.86072442e-02
-9.13164437e-01 3.13721597e-01 -3.89371037e-01 -4.18363839e-01
1.19414115e+00 -5.06804705e-01 -2.30983570e-01 -1.09415583e-01
-3.97457451e-01 4.67403680e-01 2.48003483e-01 4.40693259e-01
6.58627689e-01 -1.44842482e+00 -7.68149674e-01 3.77934247e-01
1.02425188e-01 -3.64974141e-01 1.80776492e-01 1.00264478e+00
-3.22507054e-01 2.81998485e-01 -5.39392173e-01 -5.18218935e-01
-1.50118935e+00 8.65916491e-01 -1.17428504e-01 -4.14188921e-01
-7.03031182e-01 -1.18575864e-01 -4.72220689e-01 6.82327971e-02
-9.17776749e-02 -4.56571311e-01 -6.32361412e-01 4.50777888e-01
1.00141931e+00 5.98830938e-01 2.86408439e-02 -3.91362697e-01
-5.11685073e-01 7.81019509e-01 4.95625921e-02 3.64288837e-02
1.53981793e+00 -7.06461519e-02 -7.61368036e-01 5.67826748e-01
9.42614675e-01 1.43419579e-01 -6.93280041e-01 -3.27504933e-01
6.53121769e-01 -5.49560905e-01 -5.61117113e-01 -2.21122190e-01
-8.53903353e-01 5.69613762e-02 3.48259985e-01 9.50446367e-01
1.54340076e+00 -2.46281922e-01 9.49562490e-01 1.13547906e-01
5.14575303e-01 -7.55873799e-01 -5.04377782e-01 2.72119880e-01
6.65316820e-01 -5.46668708e-01 4.58245212e-03 -3.51378918e-01
-2.55563885e-01 1.21543050e+00 1.92657784e-02 -1.36286005e-01
1.10163438e+00 4.97410774e-01 -3.07469159e-01 -7.52763376e-02
-7.39360571e-01 -1.56292662e-01 7.41011575e-02 2.93669730e-01
3.23219746e-02 2.14981943e-01 -1.04427385e+00 7.32472658e-01
-5.39674342e-01 1.83330372e-01 2.76930600e-01 1.05927324e+00
-5.63476622e-01 -1.26584148e+00 -6.70695543e-01 7.28407145e-01
-3.93001199e-01 2.05980435e-01 -2.06307888e-01 5.88961840e-01
3.29220384e-01 1.02136958e+00 4.52154368e-01 -8.47769618e-01
5.19592702e-01 4.72708195e-02 -3.95687409e-02 -6.72222441e-03
-5.57578504e-01 1.12499371e-01 -2.46035680e-01 -3.50441188e-01
-6.04108691e-01 -8.29228580e-01 -1.40980256e+00 -6.84012651e-01
1.27304032e-01 7.61184454e-01 2.69957870e-01 8.40070605e-01
4.34489667e-01 2.59540886e-01 1.12622511e+00 1.45724744e-01
-3.70805532e-01 -7.91723967e-01 -8.34517062e-01 4.95235384e-01
8.08528811e-02 -4.16511744e-01 -3.61254841e-01 -4.77582626e-02]
|
[7.313724994659424, 3.375871419906616]
|
62c66527-a0b9-4b9b-8945-d88b64f51ae2
|
from-clozing-to-comprehending-retrofitting
|
2212.04755
| null |
https://arxiv.org/abs/2212.04755v2
|
https://arxiv.org/pdf/2212.04755v2.pdf
|
From Clozing to Comprehending: Retrofitting Pre-trained Masked Language Model to Pre-trained Machine Reader
|
We present Pre-trained Machine Reader (PMR), a novel method for retrofitting pre-trained masked language models (MLMs) to pre-trained machine reading comprehension (MRC) models without acquiring labeled data. PMR can resolve the discrepancy between model pre-training and downstream fine-tuning of existing MLMs. To build the proposed PMR, we constructed a large volume of general-purpose and high-quality MRC-style training data by using Wikipedia hyperlinks and designed a Wiki Anchor Extraction task to guide the MRC-style pre-training. Apart from its simplicity, PMR effectively solves extraction tasks, such as Extractive Question Answering and Named Entity Recognition. PMR shows tremendous improvements over existing approaches, especially in low-resource scenarios. When applied to the sequence classification task in the MRC formulation, PMR enables the extraction of high-quality rationales to explain the classification process, thereby providing greater prediction explainability. PMR also has the potential to serve as a unified model for tackling various extraction and classification tasks in the MRC formulation.
|
['Lidong Bing', 'Luo Si', 'Wai Lam', 'Meng Zhou', 'Wenxuan Zhang', 'Xin Li', 'Weiwen Xu']
|
2022-12-09
| null | null | null | null |
['machine-reading-comprehension']
|
['natural-language-processing']
|
[ 6.15876973e-01 1.05534744e+00 -2.12287590e-01 -4.28490371e-01
-1.23058712e+00 -4.52354550e-01 4.59441423e-01 4.49839681e-01
-4.22723204e-01 8.25782835e-01 6.16231740e-01 -9.25817430e-01
-1.17062837e-01 -5.94833255e-01 -9.55016315e-01 2.60563940e-01
2.69210339e-01 4.68810737e-01 8.71173069e-02 -3.23859841e-01
3.60699207e-01 -1.95434570e-01 -1.31665552e+00 8.08785796e-01
1.54486620e+00 5.18074632e-01 6.48041368e-01 9.30624843e-01
-4.53246862e-01 1.19442809e+00 -5.43263793e-01 -7.68746495e-01
-3.82541388e-01 -2.78358042e-01 -1.38288176e+00 -3.20907235e-01
4.07996148e-01 -8.08920935e-02 -1.20277599e-01 4.30445462e-01
3.15051526e-01 4.25079651e-03 6.36023998e-01 -8.93545747e-01
-1.26562369e+00 1.16958833e+00 -2.73723632e-01 -2.77952235e-02
5.85929811e-01 -2.24386945e-01 1.37153947e+00 -1.17743647e+00
5.94762027e-01 9.71223891e-01 5.82676113e-01 1.03047848e+00
-1.11378801e+00 -4.16975290e-01 1.87774643e-01 4.18398350e-01
-1.06652856e+00 -4.54659849e-01 2.75040835e-01 -3.52938771e-01
1.35549378e+00 3.57674956e-01 3.73930670e-02 1.24119651e+00
2.03138795e-02 1.14206839e+00 1.00980330e+00 -1.00027883e+00
-3.64945680e-02 2.67701983e-01 5.83768368e-01 7.75205791e-01
2.10950628e-01 -2.82982469e-01 -8.79083753e-01 -9.27063078e-03
2.81915009e-01 -4.92026538e-01 -6.82251573e-01 -4.39272486e-02
-1.31822860e+00 6.56191230e-01 2.72159755e-01 -8.47826600e-02
-3.29246849e-01 -2.67653614e-01 5.85189881e-03 1.61675423e-01
3.90152842e-01 1.27897131e+00 -1.19953287e+00 -1.07523696e-02
-8.88515174e-01 -4.40078825e-02 1.04931259e+00 1.37316799e+00
5.35691738e-01 -4.71770316e-01 -4.87974942e-01 7.53520012e-01
4.98075187e-01 4.61883426e-01 5.09244740e-01 -6.71382308e-01
1.00352716e+00 7.41703689e-01 1.64420977e-01 -7.61774540e-01
-4.41389561e-01 -7.22699106e-01 -5.99508286e-01 -5.09201825e-01
3.72344166e-01 5.08017186e-03 -9.03309822e-01 1.84005344e+00
4.43461724e-02 -2.22472325e-01 6.08442605e-01 4.54386622e-01
1.20413029e+00 7.89852679e-01 4.06725138e-01 -1.26790777e-02
1.46086991e+00 -1.29842520e+00 -6.53172433e-01 -6.26860321e-01
9.49931741e-01 -5.91212690e-01 1.46602857e+00 2.96752691e-01
-9.19703305e-01 -5.92064857e-01 -9.86923099e-01 -5.61590314e-01
-3.82779568e-01 3.63732308e-01 5.01154065e-01 2.53209203e-01
-9.48114038e-01 2.10229263e-01 -5.00397563e-01 -3.75765026e-01
2.68071592e-01 2.78817397e-02 -4.23399746e-01 -4.70630854e-01
-1.36665154e+00 1.24710262e+00 5.22618711e-01 1.25715479e-01
-6.23904824e-01 -1.14463031e+00 -9.24086809e-01 1.05820090e-01
5.32687604e-01 -8.73014510e-01 1.61462975e+00 -4.81790066e-01
-1.28851390e+00 8.34947288e-01 -6.10421419e-01 -3.65316689e-01
9.79890078e-02 -6.71004474e-01 -5.02274036e-01 -1.55598506e-01
2.80583680e-01 7.23028839e-01 5.68816066e-01 -1.12027645e+00
-4.67603534e-01 -2.96396613e-02 5.69343492e-02 2.41380647e-01
-1.50587514e-01 -1.07988536e-01 -2.93148667e-01 -4.66342658e-01
7.35651329e-02 -4.52448934e-01 -1.27133429e-01 -4.66677010e-01
-7.31949747e-01 -3.48680794e-01 5.74201196e-02 -1.24689174e+00
1.60284817e+00 -1.59070873e+00 3.93469810e-01 -1.65579706e-01
6.62001133e-01 4.35264409e-01 -5.54405630e-01 5.37901819e-01
-4.15942632e-03 4.07991409e-01 -3.02939415e-01 -2.52720177e-01
7.47247264e-02 1.23844109e-01 -5.42785048e-01 -4.92271543e-01
6.87460244e-01 1.22759986e+00 -1.06853151e+00 -4.10926461e-01
-2.80865252e-01 4.11987312e-05 -5.49452186e-01 8.17648888e-01
-7.53998637e-01 4.47789103e-01 -4.67094034e-01 4.19593036e-01
3.36934716e-01 -6.96794391e-01 1.37696281e-01 1.99305154e-02
1.54211357e-01 9.28463936e-01 -6.37906075e-01 1.72693646e+00
-6.99777305e-01 7.48649716e-01 -2.58223951e-01 -5.66041291e-01
9.83055949e-01 2.39209771e-01 -3.17684948e-01 -8.29786777e-01
-4.05645251e-01 2.29680389e-01 -1.05820522e-01 -9.08961177e-01
7.79966116e-01 1.49148375e-01 -9.55472663e-02 5.50319731e-01
1.63356975e-01 2.11547494e-01 9.52180251e-02 6.74411535e-01
1.44642508e+00 3.07158917e-01 6.81951165e-01 -4.74867634e-02
7.09253073e-01 3.48624498e-01 2.10855350e-01 9.89099920e-01
2.12082759e-01 3.74493152e-01 2.84772422e-02 6.94079921e-02
-9.03201282e-01 -9.62668061e-01 3.62893194e-02 1.38501096e+00
-2.08776578e-01 -8.48659933e-01 -7.57868171e-01 -9.22279656e-01
-9.13922712e-02 1.34209967e+00 -5.32217681e-01 -2.19475076e-01
-5.00079155e-01 -4.39899176e-01 6.95239663e-01 6.94280863e-01
5.25182486e-01 -1.04641497e+00 -1.99847773e-01 3.60704303e-01
-7.18550146e-01 -1.13959277e+00 -2.18233556e-01 3.60007823e-01
-6.48650110e-01 -1.08739293e+00 -3.16537678e-01 -7.41302371e-01
7.74472177e-01 1.93413317e-01 1.75918806e+00 4.07417178e-01
2.98928861e-02 3.42659324e-01 -6.40411675e-01 -5.52133560e-01
-8.78916204e-01 7.14840353e-01 -3.14068377e-01 -5.75197816e-01
9.24865603e-01 -1.29897907e-01 -1.55424267e-01 7.47044235e-02
-7.25489974e-01 8.43813777e-01 9.30390120e-01 8.86749148e-01
3.45486820e-01 -5.52865863e-01 1.01810980e+00 -1.14138341e+00
8.23958158e-01 -5.99351943e-01 -1.62432685e-01 1.10437989e+00
-8.76262605e-01 5.47101438e-01 5.25290608e-01 -1.73406586e-01
-1.29564381e+00 -2.70073503e-01 -2.21343130e-01 5.71430564e-01
-2.08690092e-01 9.54197466e-01 -4.75664020e-01 4.54951018e-01
8.86859477e-01 1.00571528e-01 -1.56903073e-01 -9.45589244e-01
9.86766458e-01 9.62278724e-01 7.73964942e-01 -5.42495310e-01
8.55461895e-01 -5.39058506e-01 -5.83901227e-01 -5.09078860e-01
-1.57088339e+00 -4.69973236e-01 -9.64327514e-01 1.63289458e-01
9.78304267e-01 -1.07967138e+00 -1.60209745e-01 2.65928693e-02
-1.40727913e+00 -3.97266060e-01 -1.60159305e-01 2.01909259e-01
-3.35905641e-01 4.18111831e-02 -5.36098719e-01 -5.97246945e-01
-4.56921875e-01 -5.75875521e-01 8.86776090e-01 2.94729233e-01
-7.49067724e-01 -1.16162610e+00 1.23441033e-02 1.11787808e+00
3.71971369e-01 -3.11918229e-01 1.70428753e+00 -1.07797694e+00
-5.93714118e-01 8.60607773e-02 -3.31023067e-01 2.00428367e-01
-1.27209276e-01 -3.15527260e-01 -9.77489412e-01 1.24699555e-01
-3.67752582e-01 -5.86864889e-01 6.69980288e-01 -9.62755755e-02
1.09638071e+00 -5.24592042e-01 -2.41824806e-01 1.83797792e-01
1.03092396e+00 -2.13454172e-01 6.52780771e-01 5.88203907e-01
8.30888212e-01 8.92752409e-01 6.69903696e-01 -1.19356126e-01
9.66525555e-01 4.06832963e-01 5.44141158e-02 -6.86191022e-02
-3.79909992e-01 -9.59246457e-01 4.09474611e-01 1.34144616e+00
3.60250771e-01 -3.68428618e-01 -1.29296815e+00 4.06163633e-01
-1.78709865e+00 -5.92706084e-01 -4.18904811e-01 1.80823553e+00
1.42052639e+00 -1.40805528e-01 -5.51283538e-01 -2.53911644e-01
3.99527550e-01 -2.87160099e-01 -5.36956847e-01 -3.27704430e-01
-1.16655201e-01 3.60655755e-01 2.15035588e-01 6.65197194e-01
-7.74627388e-01 1.16396999e+00 6.56615543e+00 6.88879669e-01
-4.48615223e-01 2.70929843e-01 2.40686104e-01 4.35466260e-01
-6.24136090e-01 1.19784243e-01 -1.10406864e+00 -2.43137982e-02
1.30892205e+00 -6.69176430e-02 2.85114676e-01 7.91878402e-01
1.09048732e-01 -2.91158501e-02 -1.40880978e+00 5.14180601e-01
2.15001196e-01 -1.53977215e+00 3.90087813e-01 -3.86139631e-01
4.56036806e-01 -8.18678364e-02 -2.05378309e-01 9.26403761e-01
3.41183960e-01 -1.38611317e+00 5.87864816e-01 5.71242690e-01
7.05474436e-01 -2.94948936e-01 7.48992622e-01 7.75728166e-01
-6.85061455e-01 -2.86208112e-02 -3.38491291e-01 -1.49988413e-01
1.06216200e-01 3.84102970e-01 -1.21565855e+00 6.51429653e-01
2.72520065e-01 5.41897953e-01 -1.15397584e+00 6.65000618e-01
-9.55562532e-01 8.91603410e-01 1.66800603e-01 -2.70949274e-01
-2.40798667e-02 4.22181815e-01 3.22176486e-01 1.34564257e+00
-6.32952601e-02 2.21410170e-01 -8.69247243e-02 9.97093022e-01
-3.13807994e-01 2.63487041e-01 -1.72351345e-01 -3.87393177e-01
7.36421406e-01 1.03480852e+00 -1.64379589e-02 -3.59852493e-01
-4.84659523e-01 9.13764179e-01 9.26576138e-01 3.35398257e-01
-6.11297011e-01 -4.78850991e-01 3.33617389e-01 1.38693377e-02
4.31954861e-02 -2.68589526e-01 -4.21199322e-01 -1.52376950e+00
6.26935810e-02 -1.39093041e+00 2.84155637e-01 -1.16885221e+00
-1.29631650e+00 6.99741125e-01 -1.00523680e-01 -7.49572694e-01
-3.92597675e-01 -8.30906153e-01 -2.61878073e-01 1.04842389e+00
-1.80541170e+00 -1.26970041e+00 -3.66194397e-01 1.27083495e-01
7.72845745e-01 -2.01463282e-01 1.05517066e+00 -1.17554683e-02
-7.46145368e-01 7.47114897e-01 -8.71299114e-03 1.45707041e-01
8.79932702e-01 -1.45891821e+00 5.28533161e-01 1.13011825e+00
3.42206389e-01 1.13819432e+00 6.72651291e-01 -7.70089984e-01
-1.20265305e+00 -1.25605869e+00 1.55995810e+00 -1.16120803e+00
6.96379423e-01 -4.97617036e-01 -1.32427227e+00 8.37395549e-01
2.43093893e-01 -7.73766935e-01 1.16996002e+00 5.26122034e-01
-6.24264598e-01 4.34814245e-01 -6.88639641e-01 6.94790959e-01
9.59104061e-01 -8.59026432e-01 -1.39503491e+00 4.27475035e-01
1.11274111e+00 -4.35951769e-01 -9.39820349e-01 2.06240490e-01
2.48100579e-01 -3.01431626e-01 6.07728601e-01 -1.26301157e+00
8.91963243e-01 -2.98274934e-01 -7.82492906e-02 -1.35023475e+00
-3.02922934e-01 -6.72240376e-01 -4.83194172e-01 1.47278082e+00
1.40677023e+00 -2.81726122e-01 4.19980794e-01 1.12985480e+00
-2.40558505e-01 -8.11637402e-01 -4.97490376e-01 -4.64452475e-01
1.10409118e-01 -5.91898978e-01 5.66369414e-01 7.72179425e-01
3.33107889e-01 9.83519614e-01 -1.54110476e-01 3.98520738e-01
4.67284024e-01 -2.17351943e-01 8.04867685e-01 -1.15946507e+00
-4.49539721e-01 3.91940773e-02 2.75816202e-01 -1.59517169e+00
3.31793219e-01 -1.33435726e+00 4.09698844e-01 -1.90079844e+00
4.37686443e-01 -5.07163763e-01 -1.24389790e-01 8.02182794e-01
-7.27851033e-01 -3.47381204e-01 -3.61397639e-02 2.24909842e-01
-7.85991371e-01 4.12102848e-01 1.07911646e+00 6.28527179e-02
-8.10452327e-02 -2.01764554e-01 -1.15602291e+00 5.46869338e-01
6.00214899e-01 -5.64431190e-01 -5.15648365e-01 -8.79637182e-01
5.30207336e-01 -1.48146087e-02 1.98202819e-01 -6.34246707e-01
3.40817302e-01 -9.77993011e-03 3.27974886e-01 -4.93535101e-01
-2.61307001e-01 -3.92686039e-01 -2.71464974e-01 1.46643724e-02
-1.25786078e+00 3.65234613e-02 1.51653200e-01 5.30587018e-01
-6.03033453e-02 -4.95769829e-01 3.47842723e-01 -7.01938495e-02
-8.52402925e-01 -2.66000837e-01 -4.53000933e-01 5.21970749e-01
4.65275526e-01 7.16422722e-02 -8.80181909e-01 -4.26033914e-01
-6.98196411e-01 5.84708929e-01 7.53976330e-02 8.40354621e-01
6.88914299e-01 -8.56969059e-01 -7.98538625e-01 3.43527012e-02
5.44489563e-01 9.36226621e-02 -1.01325259e-01 6.46217108e-01
-2.28472546e-01 7.66636074e-01 3.07868142e-03 -3.16962391e-01
-1.13102901e+00 4.34134096e-01 -3.72697599e-02 -6.13085866e-01
-5.96887708e-01 9.98149633e-01 -1.32213861e-01 -8.75568628e-01
2.73477256e-01 -3.16841871e-01 -4.22233343e-01 -1.94814384e-01
8.89827847e-01 9.69127193e-02 1.18070178e-01 -1.31171837e-01
-1.99262872e-01 3.88366580e-02 -3.74521047e-01 -7.37710297e-02
1.33237934e+00 -5.38221776e-01 -1.58677250e-01 3.56386960e-01
8.82049322e-01 1.53992876e-01 -9.03419793e-01 -4.61015135e-01
8.01880896e-01 1.46673635e-01 -3.75287123e-02 -1.57942450e+00
-1.77540272e-01 1.02961075e+00 -1.45135194e-01 -2.71161884e-01
8.87670100e-01 9.38490406e-02 8.26845407e-01 8.10653389e-01
2.52459437e-01 -9.14231956e-01 7.46427700e-02 8.43758464e-01
1.00344121e+00 -1.27121389e+00 -2.58952230e-01 -5.72669804e-01
-7.77957737e-01 1.15564585e+00 1.03123605e+00 6.73395634e-01
1.21538572e-01 9.52793192e-03 1.51619896e-01 -4.87714559e-02
-1.04031157e+00 -1.42286703e-01 8.55797768e-01 7.89279759e-01
8.92103732e-01 -1.16355099e-01 -1.70731798e-01 1.20781589e+00
-4.16499197e-01 -6.01356402e-02 5.69797754e-01 8.46138537e-01
-6.22130275e-01 -1.13999784e+00 1.80454314e-01 4.41619068e-01
-1.19506203e-01 -7.38175631e-01 -5.67305624e-01 7.32275367e-01
-1.45805359e-01 1.28421867e+00 -3.23953360e-01 -5.16016781e-01
3.57065827e-01 5.72258115e-01 1.98984504e-01 -1.12701070e+00
-6.89306915e-01 -6.72997177e-01 6.34312391e-01 -4.41091985e-01
-1.97439179e-01 -1.32318974e-01 -1.26006985e+00 1.02477796e-01
-4.78529990e-01 5.15797257e-01 4.27290976e-01 1.28784025e+00
6.89508915e-01 5.32823503e-01 1.19980909e-01 -1.04530260e-01
-7.33404994e-01 -1.11677623e+00 1.07321769e-01 2.45733395e-01
2.37927541e-01 -1.91898748e-01 -1.76185727e-01 2.42007360e-01]
|
[11.024075508117676, 8.110283851623535]
|
4dd08158-edbc-4815-a535-76bb0bbd4428
|
graph-neural-networks-go-forward-forward
|
2302.05282
| null |
https://arxiv.org/abs/2302.05282v1
|
https://arxiv.org/pdf/2302.05282v1.pdf
|
Graph Neural Networks Go Forward-Forward
|
We present the Graph Forward-Forward (GFF) algorithm, an extension of the Forward-Forward procedure to graphs, able to handle features distributed over a graph's nodes. This allows training graph neural networks with forward passes only, without backpropagation. Our method is agnostic to the message-passing scheme, and provides a more biologically plausible learning scheme than backpropagation, while also carrying computational advantages. With GFF, graph neural networks are trained greedily layer by layer, using both positive and negative samples. We run experiments on 11 standard graph property prediction tasks, showing how GFF provides an effective alternative to backpropagation for training graph neural networks. This shows in particular that this procedure is remarkably efficient in spite of combining the per-layer training with the locality of the processing in a GNN.
|
['François Fleuret', 'Bálint Máté', 'Mathieu Alain', 'Daniele Paliotta']
|
2023-02-10
| null | null | null | null |
['graph-property-prediction']
|
['graphs']
|
[ 4.41195875e-01 4.31628823e-01 -1.83009543e-02 -3.85252208e-01
3.38013798e-01 -3.28314155e-01 8.70561182e-01 5.10648668e-01
-5.75192869e-01 7.65933871e-01 -2.37068504e-01 -7.16276765e-01
-1.95537239e-01 -1.35613000e+00 -1.16082871e+00 -6.65780246e-01
-8.98005962e-01 4.82220501e-01 5.19691169e-01 -3.04242104e-01
2.20631763e-01 8.79760861e-01 -1.31614387e+00 3.01606953e-01
4.93161045e-02 6.66199267e-01 2.11016592e-02 1.05597377e+00
-2.53286302e-01 1.23195374e+00 -2.64865726e-01 -6.87200308e-01
1.79922879e-02 -2.84465373e-01 -1.13218665e+00 -2.75073916e-01
2.53497273e-01 1.16361052e-01 -5.15823960e-01 8.57786059e-01
1.09759666e-01 6.13058619e-02 6.42235994e-01 -1.27212787e+00
-8.38305414e-01 7.41818190e-01 -1.45351410e-01 2.25005284e-01
3.48375797e-01 6.33581579e-02 1.04041183e+00 -8.90151680e-01
9.19056773e-01 1.12340569e+00 9.75192487e-01 6.88890696e-01
-1.48049653e+00 -2.51662940e-01 2.90338993e-01 1.37643233e-01
-1.02277374e+00 -2.61029840e-01 6.45005941e-01 -3.27262193e-01
1.60653985e+00 1.32823065e-01 1.01463151e+00 6.63422942e-01
5.23910999e-01 7.16050029e-01 7.70744026e-01 -4.97803420e-01
2.39080727e-01 -1.92526996e-01 3.85268897e-01 1.20015967e+00
4.47392255e-01 2.93557644e-01 -4.58238006e-01 -1.89093664e-01
8.43987823e-01 -2.46760622e-02 -2.55956143e-01 -4.30586219e-01
-9.59139585e-01 6.65205300e-01 1.07888353e+00 2.48874292e-01
-3.20307881e-01 7.93694198e-01 3.73092175e-01 7.50956178e-01
4.56724405e-01 3.37124109e-01 -7.06014752e-01 5.09961963e-01
-5.45620263e-01 -8.62080157e-02 1.11671448e+00 7.61330605e-01
1.12980282e+00 1.18165106e-01 1.75293639e-01 3.55646878e-01
7.14243531e-01 4.98640090e-02 3.63326788e-01 -4.31126952e-01
2.03496903e-01 6.36286736e-01 -5.27845681e-01 -1.14658952e+00
-7.11795449e-01 -5.36082089e-01 -9.36527371e-01 5.22366941e-01
4.62913245e-01 -8.33455995e-02 -1.06839693e+00 1.75090432e+00
1.02077417e-01 2.16358647e-01 6.47170916e-02 3.74008685e-01
7.84556866e-01 7.03982651e-01 2.97492296e-01 -1.16731599e-02
9.73260283e-01 -8.31599355e-01 -2.31143713e-01 -4.69980299e-01
1.11858928e+00 -5.14862984e-02 6.23432338e-01 3.50625455e-01
-1.05143964e+00 -3.00871044e-01 -1.03742814e+00 -8.61700326e-02
-1.01563573e+00 -5.12198865e-01 1.34295130e+00 5.98109245e-01
-1.73925674e+00 1.30795527e+00 -6.69876695e-01 -3.88462782e-01
5.43259978e-01 7.66706526e-01 -8.34629178e-01 1.43631725e-02
-1.26902080e+00 9.41623032e-01 7.74235606e-01 1.72288284e-01
-7.40324736e-01 -4.17902797e-01 -1.10540116e+00 4.21720326e-01
-1.07416958e-01 -9.60056126e-01 9.49721813e-01 -1.12596929e+00
-1.42418647e+00 8.22515428e-01 5.29181957e-02 -8.01875889e-01
2.02971742e-01 4.60013032e-01 -2.04901621e-01 3.58100943e-02
-5.72434068e-01 8.00319970e-01 8.18906546e-01 -9.18567777e-01
-1.90251142e-01 -2.64568597e-01 -8.67207441e-03 -7.58425072e-02
-2.25100189e-01 -2.20851287e-01 -1.59484044e-01 -2.07765743e-01
1.69985041e-01 -7.00511038e-01 -4.18801188e-01 3.30455214e-01
-2.11320043e-01 -1.59315646e-01 4.32477474e-01 -2.46368006e-01
8.96665633e-01 -1.97140944e+00 9.49403271e-02 6.09396935e-01
6.41312599e-01 1.67950794e-01 -2.80016184e-01 9.11104083e-01
-5.82866848e-01 1.81304589e-01 -3.22302550e-01 -1.61507338e-01
-1.93325803e-01 4.77742165e-01 -5.90923578e-02 6.25165045e-01
3.90367597e-01 1.24672294e+00 -1.02055681e+00 -2.74169832e-01
4.16184328e-02 3.89400601e-01 -4.10120428e-01 -9.39671472e-02
-4.64616627e-01 -1.36363491e-01 -6.76160678e-02 2.23313794e-01
4.02915120e-01 -6.87574267e-01 3.55228215e-01 3.01856071e-01
2.56428987e-01 3.09097141e-01 -8.14849138e-01 1.44694781e+00
-1.74497321e-01 5.52580476e-01 -4.57433388e-02 -1.06289268e+00
9.08781469e-01 1.95245340e-01 -1.47287585e-02 -5.12173176e-01
2.04408601e-01 1.37813628e-01 6.20407127e-02 -1.20383054e-01
1.74528137e-02 -3.97227228e-01 3.16771358e-01 2.82024652e-01
6.78603709e-01 2.78783113e-01 4.74863380e-01 4.22759026e-01
1.42322564e+00 1.53417066e-01 3.32548082e-01 -2.27959126e-01
6.50279939e-01 -1.86255336e-01 1.20425910e-01 7.57502913e-01
1.15076583e-02 2.20032558e-01 8.07035208e-01 -6.73565149e-01
-5.69722533e-01 -9.34833825e-01 3.49024743e-01 1.33998907e+00
-3.24893624e-01 -5.84727228e-01 -5.14587402e-01 -8.25067580e-01
6.25074655e-02 3.97625387e-01 -8.96905839e-01 -4.10681129e-01
-4.40459639e-01 -7.52410233e-01 6.62371397e-01 4.18905854e-01
2.81306952e-01 -1.33247864e+00 -5.58010265e-02 2.24052981e-01
5.87318778e-01 -5.57913780e-01 2.00362623e-01 6.84786797e-01
-1.36449063e+00 -1.15461409e+00 -5.52679181e-01 -1.19653428e+00
9.45021272e-01 -6.10121824e-02 1.25517464e+00 6.94660842e-01
-1.20001854e-02 5.22857159e-02 2.08911169e-02 -2.37920016e-01
-5.42918622e-01 2.03372717e-01 -4.41517770e-01 -1.25943735e-01
1.57291219e-01 -8.79851341e-01 -2.43410647e-01 -3.44572544e-01
-7.25583494e-01 2.37597391e-01 6.39896631e-01 9.37051058e-01
3.38614762e-01 8.66541266e-02 5.33389211e-01 -1.63772321e+00
6.14671946e-01 -5.60632288e-01 -6.82822466e-01 2.07387745e-01
-9.20404673e-01 3.02280158e-01 1.04436767e+00 4.34467569e-03
-5.88375390e-01 2.13280335e-01 -4.88088846e-01 -9.55271050e-02
-1.78828448e-01 7.39655554e-01 2.31227413e-01 -6.63096249e-01
8.67535472e-01 3.22203666e-01 1.86325207e-01 -3.29901338e-01
5.63974917e-01 -2.20354483e-01 3.17313373e-01 -6.12388253e-02
6.97679460e-01 2.82769561e-01 5.59132874e-01 -7.83813357e-01
-1.50891498e-01 -1.78272948e-01 -5.73165178e-01 -2.94712871e-01
6.74493015e-01 -4.91421670e-01 -9.45213318e-01 7.16183424e-01
-1.14433873e+00 -6.32249355e-01 -3.37731421e-01 3.19054067e-01
-5.97487330e-01 2.05993310e-01 -1.25176799e+00 -5.48631966e-01
-2.49421462e-01 -3.83980453e-01 3.25229645e-01 -2.86377668e-02
1.19210310e-01 -1.65868056e+00 1.05906464e-01 -5.48315823e-01
4.87123638e-01 1.30492032e-01 1.28626740e+00 -1.04880583e+00
-5.24313092e-01 -3.22838843e-01 -3.04442257e-01 9.25267264e-02
-1.77271605e-01 1.04286708e-02 -1.14343524e+00 -2.90724277e-01
-3.78271788e-01 -2.22060695e-01 1.27706242e+00 1.59633204e-01
9.41075981e-01 -5.39942145e-01 -5.22236884e-01 7.84786761e-01
1.80351222e+00 -4.20271426e-01 7.65065610e-01 1.67900458e-01
7.41693795e-01 4.64284092e-01 -2.76416719e-01 -1.41213298e-01
3.23086768e-01 -3.37995082e-01 6.37374401e-01 -3.68186921e-01
-3.53550851e-01 -5.02068341e-01 2.44695485e-01 6.64020836e-01
-1.34751052e-01 -5.19711137e-01 -9.86836135e-01 3.63985091e-01
-1.76993823e+00 -9.25125480e-01 -4.66818422e-01 1.99514627e+00
5.88394582e-01 5.21606624e-01 3.22156958e-02 4.02418733e-01
6.62960887e-01 1.70910239e-01 -1.49519831e-01 -7.58883238e-01
9.00010616e-02 5.85984468e-01 8.75385284e-01 6.52314901e-01
-7.76131868e-01 9.94310260e-01 7.64977598e+00 5.85939348e-01
-1.18447900e+00 -2.69509971e-01 4.02278244e-01 4.57754791e-01
-3.42101157e-01 4.27897014e-02 -4.11375374e-01 6.02985993e-02
1.45002913e+00 1.11279175e-01 7.00432539e-01 7.21100032e-01
-4.77734268e-01 7.49897882e-02 -1.36757290e+00 5.38220465e-01
-1.97153345e-01 -1.73874390e+00 2.07585305e-01 -1.85472414e-01
1.44220039e-01 4.96597290e-01 -3.34313333e-01 5.10450065e-01
5.10391951e-01 -1.19933462e+00 3.80946636e-01 4.22246099e-01
4.50549901e-01 -4.87154245e-01 5.00786364e-01 5.47744453e-01
-1.14408314e+00 -8.15450680e-03 -4.50888038e-01 -5.40321648e-01
-1.76935613e-01 5.59073150e-01 -1.11772430e+00 4.10851091e-01
3.53090137e-01 8.73424530e-01 -7.67911136e-01 1.05183768e+00
-5.01221299e-01 5.61343431e-01 -3.02699447e-01 -4.67922568e-01
1.87531605e-01 2.77286731e-02 1.23680986e-01 1.63523304e+00
-1.73210397e-01 -2.74367720e-01 -7.95368999e-02 7.96196282e-01
-2.79967427e-01 8.42811689e-02 -1.03543746e+00 -3.21661055e-01
1.13553442e-01 1.25061786e+00 -1.17395592e+00 -4.08125162e-01
-4.24148917e-01 9.37355220e-01 8.40053439e-01 4.20051217e-01
-3.40273798e-01 -8.91050398e-01 6.16145059e-02 1.00528494e-01
5.06683052e-01 -3.46758872e-01 -1.52201504e-01 -8.16989422e-01
-3.77235562e-01 -3.48268598e-01 5.59103429e-01 -7.50059366e-01
-1.27434909e+00 5.69455206e-01 -3.03260952e-01 -5.40316701e-01
-2.75180757e-01 -1.17856407e+00 -6.65943146e-01 9.38818932e-01
-1.92278492e+00 -1.18398559e+00 4.96830605e-02 7.64359772e-01
-1.63457319e-01 2.41236106e-01 1.19738960e+00 1.69677421e-01
-2.25128546e-01 2.61567652e-01 -3.00076634e-01 3.76296997e-01
-6.50437735e-03 -1.49946856e+00 1.06517804e+00 7.12998748e-01
4.02662963e-01 8.42171073e-01 3.42643410e-01 -6.33371234e-01
-1.56213105e+00 -1.03453565e+00 1.28445077e+00 -7.87336975e-02
8.45193624e-01 -6.11088753e-01 -1.19321609e+00 1.01064932e+00
1.36003733e-01 4.45407867e-01 5.57342172e-01 4.93697137e-01
-4.62026864e-01 7.60779306e-02 -1.06181359e+00 3.75107437e-01
1.21567237e+00 -5.19259453e-01 -5.13238370e-01 5.02425611e-01
5.89386940e-01 -1.77424937e-01 -8.10187936e-01 6.36247322e-02
2.80194998e-01 -9.28652883e-01 9.03029382e-01 -8.31201315e-01
1.46754667e-01 -2.31326267e-01 2.74087429e-01 -1.22152042e+00
-7.95849800e-01 -6.28908694e-01 -3.11025679e-01 7.68161416e-01
9.23933983e-01 -1.18713355e+00 1.03815603e+00 2.24442825e-01
4.20258343e-02 -8.29231441e-01 -5.31841159e-01 -6.83698416e-01
7.28595406e-02 -4.85036612e-01 4.56577569e-01 7.86516309e-01
1.76120043e-01 7.08523393e-01 -7.94426799e-02 6.37235045e-02
3.55966926e-01 -1.54415026e-01 4.12642092e-01 -1.46985996e+00
-7.31330395e-01 -4.42422807e-01 -7.72928178e-01 -9.20179963e-01
1.38290077e-01 -1.56608045e+00 2.63330247e-02 -1.79761267e+00
-1.52312577e-01 -3.56980801e-01 -4.78381276e-01 9.08688962e-01
1.86729565e-01 3.03484142e-01 -2.60608923e-02 -4.12803441e-02
-3.63792151e-01 9.26542431e-02 1.12234008e+00 1.92091949e-02
2.19545782e-01 -2.13733828e-03 -4.05685335e-01 8.79061818e-01
9.14584637e-01 -7.65759885e-01 -5.40203929e-01 -5.07842720e-01
7.92905927e-01 -1.32987082e-01 5.09392500e-01 -9.04328883e-01
5.03469586e-01 1.54197156e-01 7.39545524e-01 -2.04524383e-01
2.97607362e-01 -7.54488409e-01 3.77829559e-02 9.40818250e-01
-4.05129790e-01 1.05333947e-01 1.85551718e-01 8.17000329e-01
5.74591458e-02 -2.88725257e-01 6.55212343e-01 -5.65041423e-01
-9.52701867e-01 4.92177069e-01 -6.26934171e-01 -2.95446455e-01
6.53687358e-01 -4.35268402e-01 -4.49623108e-01 -8.32485035e-02
-1.05942738e+00 6.29438236e-02 4.67705131e-01 -1.39339373e-01
7.69523740e-01 -8.69807720e-01 -5.11128902e-01 7.83526361e-01
-4.11539413e-02 -3.30873489e-01 -1.88023552e-01 7.16562867e-01
-9.41539466e-01 2.95566291e-01 -2.77825773e-01 -3.26010048e-01
-1.08860600e+00 1.13594222e+00 4.59285915e-01 -2.24761054e-01
-7.15912104e-01 1.34328616e+00 -1.01887668e-02 -6.27940059e-01
2.35184133e-01 -2.93293417e-01 -2.96051472e-01 -3.18797171e-01
3.89210612e-01 8.58204514e-02 3.23675752e-01 -1.81831077e-01
-4.47439939e-01 1.86809778e-01 4.46056090e-02 6.90021738e-02
1.34799981e+00 2.87948847e-01 -5.59855878e-01 4.41414803e-01
1.22329533e+00 -3.39912176e-01 -8.09736848e-01 -4.52789757e-03
4.04007524e-01 -5.68550378e-02 2.28226539e-02 -7.98571348e-01
-1.08392036e+00 1.00250864e+00 6.31970689e-02 5.31455457e-01
9.77977574e-01 -1.55791461e-01 5.31787574e-01 7.79948771e-01
4.46239561e-01 -5.55852473e-01 -3.05569351e-01 6.63028836e-01
5.73731184e-01 -7.48459756e-01 8.29012319e-02 -5.67101181e-01
7.46018291e-02 1.51832259e+00 3.11502188e-01 -7.36921608e-01
9.20429528e-01 3.37031305e-01 -3.37603927e-01 -5.35293460e-01
-1.21919370e+00 -1.17545590e-01 1.91583842e-01 6.85929537e-01
5.02735436e-01 -1.02288179e-01 -1.60123691e-01 -5.94906397e-02
1.38499171e-01 2.31521204e-01 4.31046844e-01 1.05680227e+00
-6.51958823e-01 -1.14391398e+00 2.34667704e-01 6.29297018e-01
-3.18179190e-01 -3.39026988e-01 -6.70447826e-01 1.00798941e+00
-3.36189829e-02 4.90411848e-01 -9.92877558e-02 -5.33183932e-01
6.58656880e-02 3.17547232e-01 8.26195776e-01 -7.71374941e-01
-8.36277485e-01 -4.82048213e-01 3.11850309e-01 -6.77582324e-01
-1.96532339e-01 -2.06047386e-01 -1.61429107e+00 -5.28565288e-01
-5.30724406e-01 1.65893376e-01 7.46596277e-01 7.73538172e-01
2.96918035e-01 6.12864256e-01 3.68840903e-01 -8.22354734e-01
-2.61555314e-01 -5.68543434e-01 -7.31088758e-01 1.33875370e-01
4.84380245e-01 -1.63983479e-01 -3.83990943e-01 -1.16114818e-01]
|
[6.927856922149658, 6.178539752960205]
|
3666a733-1a37-474d-8527-4daeca71d8ee
|
3d-object-recognition-with-ensemble-learning
|
1904.08159
| null |
https://arxiv.org/abs/1904.08159v2
|
https://arxiv.org/pdf/1904.08159v2.pdf
|
3D Object Recognition with Ensemble Learning --- A Study of Point Cloud-Based Deep Learning Models
|
In this study, we present an analysis of model-based ensemble learning for 3D point-cloud object classification and detection. An ensemble of multiple model instances is known to outperform a single model instance, but there is little study of the topic of ensemble learning for 3D point clouds. First, an ensemble of multiple model instances trained on the same part of the $\textit{ModelNet40}$ dataset was tested for seven deep learning, point cloud-based classification algorithms: $\textit{PointNet}$, $\textit{PointNet++}$, $\textit{SO-Net}$, $\textit{KCNet}$, $\textit{DeepSets}$, $\textit{DGCNN}$, and $\textit{PointCNN}$. Second, the ensemble of different architectures was tested. Results of our experiments show that the tested ensemble learning methods improve over state-of-the-art on the $\textit{ModelNet40}$ dataset, from $92.65\%$ to $93.64\%$ for the ensemble of single architecture instances, $94.03\%$ for two different architectures, and $94.15\%$ for five different architectures. We show that the ensemble of two models with different architectures can be as effective as the ensemble of 10 models with the same architecture. Third, a study on classic bagging i.e. with different subsets used for training multiple model instances) was tested and sources of ensemble accuracy growth were investigated for best-performing architecture, i.e. $\textit{SO-Net}$. We also investigate the ensemble learning of $\textit{Frustum PointNet}$ approach in the task of 3D object detection, increasing the average precision of 3D box detection on the $\textit{KITTI}$ dataset from $63.1\%$ to $66.5\%$ using only three model instances. We measure the inference time of all 3D classification architectures on a $\textit{Nvidia Jetson TX2}$, a common embedded computer for mobile robots, to allude to the use of these models in real-life applications.
|
['Tarek El-Gaaly', 'Łukasz Chechliński', 'Daniel Koguciuk']
|
2019-04-17
| null | null | null | null |
['3d-object-recognition', '3d-classification']
|
['computer-vision', 'computer-vision']
|
[-2.79816717e-01 -2.80804873e-01 2.82905877e-01 -3.11048627e-01
-5.72475433e-01 -5.05500615e-01 3.36732119e-01 -5.09218052e-02
-4.58658248e-01 3.77498478e-01 -1.13040340e+00 -6.91002846e-01
-3.41305614e-01 -9.30378556e-01 -1.13514698e+00 -7.22812653e-01
-5.18257320e-01 8.80599916e-01 1.89314067e-01 -3.14539135e-01
2.55555332e-01 8.02357495e-01 -2.05129623e+00 2.84549445e-01
6.52209461e-01 1.69938302e+00 1.83413446e-01 5.40535450e-01
-1.29416406e-01 3.36339414e-01 -9.07829523e-01 -2.08333448e-01
5.53393781e-01 3.10518056e-01 -2.37736031e-01 -4.86473322e-01
9.55476582e-01 -3.01191956e-01 -6.91243680e-03 9.14371908e-01
5.98218322e-01 -8.53194222e-02 8.95021677e-01 -1.62449014e+00
-1.85597569e-01 1.58511847e-01 -6.31811976e-01 3.47637713e-01
-2.56565154e-01 4.09106225e-01 6.53573155e-01 -1.18932807e+00
1.52232170e-01 1.16357601e+00 9.23079669e-01 4.36330378e-01
-1.08120978e+00 -1.36241651e+00 1.60144508e-01 -1.34878188e-01
-1.52802777e+00 -5.82318790e-02 5.27624726e-01 -8.04668903e-01
1.62001562e+00 2.28190705e-01 6.60060406e-01 7.26800799e-01
3.93933415e-01 6.16827130e-01 1.03361166e+00 -3.91154081e-01
2.08828956e-01 2.24749092e-02 5.07787585e-01 8.31537306e-01
5.52626431e-01 5.65105975e-01 -2.13973895e-01 -1.14802867e-01
7.85165250e-01 -2.70815333e-03 2.41917729e-01 -3.63734782e-01
-7.54305601e-01 8.43440473e-01 7.04911232e-01 1.64746493e-01
-1.41378701e-01 4.62323606e-01 3.62747312e-01 2.11620897e-01
8.21856856e-01 5.01025736e-01 -8.22609305e-01 1.60881728e-01
-5.65330744e-01 5.83703220e-01 6.54747128e-01 1.42301714e+00
1.02541637e+00 3.98356587e-01 4.62655008e-01 9.33799446e-01
3.84717971e-01 9.68052864e-01 1.02292895e-01 -8.34769130e-01
5.36429822e-01 9.27437127e-01 1.32232666e-01 -8.05300832e-01
-5.64855218e-01 -3.48498285e-01 -9.15051758e-01 1.17725480e+00
2.31523633e-01 -2.14954495e-01 -1.40070629e+00 1.22896183e+00
7.35853687e-02 8.72520506e-02 -2.68764377e-01 5.47289073e-01
1.11163008e+00 5.78909397e-01 1.05696492e-01 5.28450072e-01
1.19555175e+00 -6.47327304e-01 2.50404447e-01 -4.81326550e-01
9.51081455e-01 -4.05915529e-01 6.50553584e-01 4.77731794e-01
-9.78159070e-01 -1.04338551e+00 -1.36367249e+00 3.94656956e-01
-7.72548378e-01 2.33759895e-01 5.84465861e-01 7.55584240e-01
-1.06501698e+00 6.37368441e-01 -9.44485128e-01 -1.81181982e-01
7.19190419e-01 9.50863719e-01 -4.98921610e-02 -1.75381556e-01
-6.35094881e-01 1.14330292e+00 3.18527967e-01 9.70050879e-03
-9.97585773e-01 -7.63563752e-01 -5.45366764e-01 -2.94408619e-01
-8.05962682e-02 -7.82809973e-01 1.08798563e+00 -6.20586991e-01
-1.07437968e+00 1.11232030e+00 1.67029455e-01 -5.28844535e-01
1.43213391e-01 -3.12858105e-01 -2.52459824e-01 -2.05169812e-01
-1.37934545e-02 9.64288831e-01 8.30955982e-01 -1.41479170e+00
-1.06613553e+00 -8.99376750e-01 1.58855096e-01 -1.09761104e-01
2.30243549e-01 -2.10205749e-01 -9.31871086e-02 -2.86312729e-01
4.00741816e-01 -1.32794476e+00 -1.08896658e-01 -8.33341330e-02
-1.35258362e-01 -5.01342237e-01 9.77516174e-01 -6.78672493e-02
6.50295079e-01 -1.95986056e+00 -6.24533072e-02 2.69471616e-01
3.18753004e-01 3.35800827e-01 1.01566374e-01 -8.67849663e-02
-1.93884313e-01 3.64408642e-01 -2.63264596e-01 -5.29692411e-01
-8.89514163e-02 1.84302866e-01 -9.85439271e-02 3.84821266e-01
1.79110363e-01 4.44086701e-01 -3.58995348e-01 -2.28810385e-01
6.66376472e-01 3.15596849e-01 -5.03445208e-01 -1.40385076e-01
-4.43202823e-01 1.68346912e-01 -5.16869247e-01 8.50219667e-01
9.90951180e-01 -2.30554268e-01 -5.30014098e-01 -1.23630911e-01
-2.20068693e-01 6.13566078e-02 -1.02043700e+00 1.57500231e+00
-3.78401846e-01 4.76074964e-01 4.64653298e-02 -1.03809738e+00
1.42487979e+00 -4.80632931e-02 7.49341786e-01 -5.52147746e-01
3.45564842e-01 5.28181434e-01 3.39404434e-01 -7.12286383e-02
4.37862933e-01 -8.15405101e-02 -9.79214758e-02 2.99802721e-01
3.13778311e-01 -7.43021667e-01 -1.53756738e-01 -2.31244937e-01
1.14166451e+00 2.55738139e-01 -2.17502460e-01 -4.95657980e-01
1.06126145e-01 3.64684850e-01 1.26957204e-02 1.08664095e+00
-2.10004121e-01 4.55434322e-01 -2.41456125e-02 -8.37998927e-01
-1.10345209e+00 -1.03739333e+00 -3.88164312e-01 1.20021975e+00
2.24554852e-01 -1.93374932e-01 -6.76432788e-01 -6.63749278e-01
4.80130851e-01 8.99606824e-01 -3.79326522e-01 -1.52547330e-01
-7.15907753e-01 -9.69431162e-01 7.16202796e-01 7.55529881e-01
6.37550831e-01 -1.01856184e+00 -9.68853593e-01 2.93556657e-02
3.66157115e-01 -8.30418229e-01 5.76211333e-01 6.56646430e-01
-1.27784133e+00 -1.08269942e+00 -2.61580467e-01 -5.54717839e-01
3.89570624e-01 1.21549256e-01 1.59261048e+00 4.15097624e-01
-2.36248598e-01 5.29504657e-01 -3.76328707e-01 -1.46943438e+00
-1.35861322e-01 1.72388628e-01 2.37628460e-01 -7.02967167e-01
8.27171385e-01 -5.08724928e-01 -4.36300635e-01 5.72371185e-01
-6.31851137e-01 -1.88463867e-01 4.54508215e-01 6.72456861e-01
7.14492917e-01 5.86584210e-03 2.64487684e-01 -4.01465565e-01
-2.06899140e-02 -4.16610569e-01 -8.23185444e-01 -2.72428393e-01
-5.47739506e-01 -3.08525383e-01 1.72294095e-01 -3.43019933e-01
-3.21935356e-01 -2.49292627e-02 -1.74918085e-01 -1.20960844e+00
-3.31512004e-01 1.18364409e-01 1.55307263e-01 -3.45108390e-01
9.26641345e-01 3.15055549e-02 1.06789641e-01 -4.53775257e-01
-5.29849343e-02 5.30721724e-01 1.12116605e-01 -8.17528725e-01
5.29627025e-01 4.89469349e-01 1.22208960e-01 -8.69848192e-01
-3.44178975e-01 -1.73925295e-01 -7.37960279e-01 -2.10767746e-01
8.17845702e-01 -1.00589335e+00 -8.59350741e-01 8.11601341e-01
-1.28882468e+00 -4.41721648e-01 2.79186526e-03 4.38106775e-01
-6.07840419e-01 -3.94600093e-01 -1.09408297e-01 -1.07813859e+00
-3.88255239e-01 -1.44146788e+00 1.42569149e+00 -2.93071065e-02
-6.19179867e-02 -6.93450153e-01 -3.26025665e-01 8.05581659e-02
2.85105437e-01 4.15864021e-01 9.73886967e-01 -8.29908609e-01
-7.31186509e-01 -5.26955724e-01 -3.11461478e-01 6.86518908e-01
-1.86944947e-01 7.31394365e-02 -1.11889923e+00 -3.86860222e-01
1.58086419e-01 -1.74526319e-01 8.50632012e-01 4.78534490e-01
1.39287567e+00 1.37524247e-01 -7.96531677e-01 3.62197965e-01
1.42568517e+00 7.14838386e-01 6.40147865e-01 4.31056678e-01
6.93466544e-01 1.52249932e-01 3.93554032e-01 1.66205898e-01
2.54565537e-01 5.80825984e-01 1.08041298e+00 1.79346830e-01
2.71384269e-01 3.22920918e-01 1.01939347e-02 2.66386449e-01
-5.34903288e-01 -2.28481337e-01 -1.50057936e+00 3.48566741e-01
-1.39127910e+00 -6.55443132e-01 -1.32946447e-01 1.99716961e+00
-2.82553993e-02 5.27714312e-01 1.41035914e-01 2.12664768e-01
6.45847917e-01 -2.23589122e-01 -7.44393110e-01 -2.08837941e-01
7.43819922e-02 8.03317547e-01 4.79295999e-01 3.01930159e-02
-1.43882918e+00 7.03736544e-01 4.71700382e+00 7.63631821e-01
-1.24540472e+00 1.22798802e-02 6.85915947e-01 -3.62901658e-01
4.49647784e-01 -3.69694769e-01 -1.25129640e+00 7.21769094e-01
9.23971415e-01 3.90568972e-01 -2.46473160e-02 1.44406438e+00
-4.57762271e-01 -2.50396162e-01 -1.52235186e+00 1.28423512e+00
1.54731870e-01 -1.37380946e+00 -1.36354119e-01 2.37954125e-01
5.23671508e-01 7.16714621e-01 2.85293818e-01 8.10643435e-01
4.52965647e-01 -1.28219330e+00 9.92038071e-01 2.80754209e-01
8.25101197e-01 -7.15048134e-01 8.41503620e-01 6.38839900e-01
-1.10731363e+00 -2.51722068e-01 -3.58153462e-01 -1.44782543e-01
-2.60860682e-01 3.96638513e-01 -8.94902647e-01 5.96405864e-01
1.66628349e+00 4.81036812e-01 -5.21690845e-01 7.74324179e-01
3.61091584e-01 2.50428349e-01 -9.88516569e-01 -3.37775052e-01
3.26022536e-01 1.74696669e-01 6.71943545e-01 1.00360727e+00
6.70910716e-01 4.37549591e-01 1.18349865e-01 8.03024113e-01
6.28587380e-02 -4.77674931e-01 -9.76449847e-01 4.72492158e-01
3.67282152e-01 7.83253193e-01 -6.44151092e-01 -3.47035676e-01
-9.27913189e-02 3.02755654e-01 2.76085436e-01 8.52324069e-02
-9.03325796e-01 -3.14192384e-01 9.37019765e-01 2.83128947e-01
4.84731436e-01 -4.72371429e-01 -7.76212990e-01 -5.87874711e-01
-2.04213113e-02 -6.54893577e-01 3.62098545e-01 -1.12653768e+00
-1.48590684e+00 8.61641109e-01 4.06844467e-01 -1.54394996e+00
-5.82660511e-02 -1.38595068e+00 -3.74740750e-01 9.81757104e-01
-8.88558924e-01 -1.04610097e+00 -5.65200210e-01 3.09910893e-01
4.33940500e-01 -5.46689928e-01 8.08190107e-01 2.69038498e-01
-1.86965033e-01 4.53807652e-01 7.77481645e-02 1.51068226e-01
1.94097072e-01 -9.90985870e-01 5.15425026e-01 1.11616969e-01
-7.55638489e-03 4.67763036e-01 4.15114343e-01 -5.31475425e-01
-1.33264565e+00 -1.37345624e+00 1.38758540e-01 -1.07451880e+00
1.09945841e-01 -3.72286648e-01 -7.91868091e-01 9.65198219e-01
-2.21881926e-01 1.95625685e-02 3.71579587e-01 1.22351542e-01
-2.55772799e-01 -2.93624967e-01 -1.51714790e+00 2.10538849e-01
1.20357895e+00 -1.33577764e-01 -4.63116616e-01 9.84799862e-02
6.49683058e-01 -8.66895199e-01 -9.47401702e-01 1.02021837e+00
5.88495374e-01 -1.28951490e+00 1.24945557e+00 -4.39690500e-01
4.02847707e-01 -1.61475465e-01 -6.39695287e-01 -1.24240947e+00
-1.99226081e-01 3.24093848e-01 -1.15539446e-01 5.71988940e-01
5.51748693e-01 -8.24071825e-01 9.78773177e-01 4.80986625e-01
-8.38696659e-01 -9.20158207e-01 -1.32450032e+00 -8.58581126e-01
5.80550671e-01 -9.11784649e-01 8.07764113e-01 6.22523308e-01
-7.50418723e-01 9.59517881e-02 3.29679728e-01 3.03731680e-01
7.07057476e-01 -1.69838235e-01 1.13858593e+00 -1.57811105e+00
9.63329598e-02 -6.56409144e-01 -7.58166969e-01 -8.22694838e-01
-5.12342807e-03 -9.33735073e-01 -1.64145857e-01 -1.30547619e+00
-3.67779046e-01 -1.35465026e+00 -2.74136275e-01 3.57698113e-01
3.07393134e-01 2.09429905e-01 4.04621691e-01 2.94475377e-01
-3.71688485e-01 1.90609217e-01 8.49677444e-01 -3.94869208e-01
-6.82840273e-02 7.28585571e-02 -2.04954937e-01 1.03040183e+00
6.31440938e-01 -4.60295975e-01 1.48485200e-02 -9.68967438e-01
9.01783034e-02 -2.06341818e-01 9.31392193e-01 -1.60361087e+00
1.73182189e-01 3.01283866e-01 7.55952060e-01 -1.29284513e+00
8.36665928e-01 -9.55203176e-01 1.81990206e-01 3.30400139e-01
3.91563594e-01 2.13461593e-01 9.96464133e-01 2.57198304e-01
2.14408502e-01 -2.69861430e-01 9.62762356e-01 -4.62799013e-01
-7.23530769e-01 3.97013903e-01 7.88137689e-02 -3.91006798e-01
1.17261899e+00 -6.26175404e-01 -2.84171432e-01 2.25621253e-01
-7.44727314e-01 1.48019284e-01 3.83844763e-01 4.61484402e-01
5.89397252e-01 -1.18341160e+00 -5.96467972e-01 4.21086848e-01
1.56563371e-01 8.77940655e-01 2.70676732e-01 3.76478374e-01
-5.74293852e-01 4.46642131e-01 -2.47321218e-01 -1.54270637e+00
-1.01342225e+00 3.13726932e-01 7.76654601e-01 -2.31962069e-03
-1.21445209e-01 1.24036968e+00 3.69852893e-02 -8.27172101e-01
1.66194960e-01 -6.04771554e-01 1.55338272e-01 -2.26582423e-01
7.09090605e-02 7.27011204e-01 6.44134223e-01 -4.04598266e-01
-5.33099473e-01 7.80660987e-01 -7.18953495e-04 3.03880632e-01
1.31409001e+00 4.25367177e-01 -1.41499132e-01 5.67131162e-01
1.18971276e+00 -8.75515163e-01 -8.91104281e-01 2.72358179e-01
-1.21472046e-01 -1.37183115e-01 -2.73076594e-01 -9.96735275e-01
-9.38534975e-01 1.04881930e+00 1.25498962e+00 2.36085162e-01
7.86992908e-01 8.62883404e-02 1.36666626e-01 5.91624439e-01
1.11823070e+00 -7.77899504e-01 5.01168147e-02 9.10018384e-01
9.47114706e-01 -1.50792360e+00 1.05550207e-01 4.94237430e-02
-2.69446638e-03 8.93281996e-01 1.17761135e+00 -6.41642034e-01
1.04359341e+00 3.39084715e-01 -1.86331034e-01 -6.29234612e-01
-5.13727367e-01 5.33866584e-02 1.47308603e-01 6.89953506e-01
7.30422139e-02 1.17457151e-01 5.24010539e-01 6.09449565e-01
-4.61901426e-01 -6.52177557e-02 -5.05505465e-02 1.12196016e+00
-5.07587254e-01 -4.98428941e-01 -5.81447899e-01 1.00441217e+00
-1.05740473e-01 -2.04769000e-02 -2.65469909e-01 1.38653469e+00
7.40511060e-01 9.32835519e-01 7.55081594e-01 -6.73927307e-01
6.27144575e-01 3.52972567e-01 7.20055401e-01 -7.03751504e-01
-7.78052330e-01 -3.30536157e-01 -9.89039242e-02 -3.75015408e-01
-3.24674010e-01 -5.17056823e-01 -1.17205119e+00 -4.53401566e-01
-3.85380328e-01 -8.46771672e-02 1.09796524e+00 7.20871150e-01
4.77370828e-01 4.74880874e-01 2.64418095e-01 -1.65675390e+00
-4.31979448e-01 -1.19703722e+00 -4.14658993e-01 3.26332971e-02
2.08909154e-01 -1.17972767e+00 -5.99011123e-01 -2.01263621e-01]
|
[7.925760746002197, -3.4585959911346436]
|
13fa5cb3-e157-4cd9-8db5-affcf2b09288
|
multi-organ-cancer-classification-and
|
1606.00897
| null |
http://arxiv.org/abs/1606.00897v2
|
http://arxiv.org/pdf/1606.00897v2.pdf
|
Multi-Organ Cancer Classification and Survival Analysis
|
Accurate and robust cell nuclei classification is the cornerstone for a wider
range of tasks in digital and Computational Pathology. However, most machine
learning systems require extensive labeling from expert pathologists for each
individual problem at hand, with no or limited abilities for knowledge transfer
between datasets and organ sites. In this paper we implement and evaluate a
variety of deep neural network models and model ensembles for nuclei
classification in renal cell cancer (RCC) and prostate cancer (PCa). We propose
a convolutional neural network system based on residual learning which
significantly improves over the state-of-the-art in cell nuclei classification.
Finally, we show that the combination of tissue types during training increases
not only classification accuracy but also overall survival analysis.
|
['Thomas Fuchs', 'Peter Schüffler', 'Peter Wild', 'Stefan Bauer', 'Joachim M. Buhmann', 'Nicolas Carion']
|
2016-06-02
| null | null | null | null |
['nuclei-classification']
|
['medical']
|
[ 1.21544331e-01 8.45541209e-02 -2.88080901e-01 -9.84986946e-02
-8.82044792e-01 -6.25710666e-01 4.09080565e-01 5.63147485e-01
-5.87670565e-01 1.05450380e+00 -1.76204279e-01 -4.26544636e-01
6.25960827e-02 -8.52346778e-01 -4.14871782e-01 -1.12053776e+00
1.86504964e-02 7.53479958e-01 1.71940774e-02 3.10728569e-02
-9.16304588e-02 1.07486498e+00 -9.65688348e-01 2.21362621e-01
7.49425411e-01 8.12879264e-01 8.29271823e-02 9.70789194e-01
-1.00046173e-01 1.11849034e+00 -9.48993936e-02 -2.84545422e-01
1.15740113e-01 -1.92994729e-01 -7.20795393e-01 -2.41478041e-01
5.80818355e-01 -7.78405368e-02 -5.93424737e-01 1.04944301e+00
7.65182734e-01 -4.65210319e-01 8.99661541e-01 -8.44338775e-01
-3.26364040e-01 5.24028838e-01 -2.73595452e-01 3.11804473e-01
-4.57545221e-01 3.34367067e-01 6.22351170e-01 -5.34336805e-01
1.06246579e+00 5.48741579e-01 9.58262682e-01 8.49532843e-01
-1.32136381e+00 -6.00105107e-01 -5.00910282e-01 5.32389618e-02
-1.30113065e+00 -3.14278334e-01 2.44136781e-01 -5.67490458e-01
7.02167511e-01 1.65498585e-01 8.91096890e-01 8.21479023e-01
4.41998512e-01 7.66954541e-01 9.77413416e-01 -1.01968460e-01
1.84115022e-01 5.97428344e-02 3.76272082e-01 6.87054455e-01
5.99071741e-01 -2.19217315e-01 7.98594058e-02 4.64939624e-02
9.99664962e-01 3.47437590e-01 -3.60872090e-01 -2.63119727e-01
-1.27139854e+00 6.36044621e-01 5.05487561e-01 6.22012198e-01
-2.24769861e-01 4.32531744e-01 6.83773220e-01 -3.85113470e-02
2.03779548e-01 2.52238452e-01 -3.36617351e-01 2.14819163e-01
-9.66263890e-01 1.23245291e-01 8.40752482e-01 4.28411454e-01
3.47606361e-01 -1.36654526e-01 -4.37608242e-01 6.74232721e-01
2.33639941e-01 2.98537850e-01 5.37904203e-01 -8.20153475e-01
-2.92131394e-01 8.86026323e-01 -1.74166277e-01 -5.89133799e-01
-9.23824787e-01 -9.51077580e-01 -1.45573878e+00 2.81019866e-01
8.92007351e-01 2.02621296e-01 -1.28777039e+00 1.50159395e+00
3.30384642e-01 1.52864963e-01 1.63189247e-01 6.76486611e-01
1.03211975e+00 1.47382960e-01 2.33836517e-01 -1.30291209e-01
1.39873827e+00 -8.25858414e-01 -6.16655529e-01 -1.57242686e-01
1.20955873e+00 -3.97747368e-01 2.64318228e-01 1.42651573e-01
-8.43074083e-01 -3.10187135e-02 -8.31480801e-01 -4.77239460e-01
-5.95514655e-01 4.33907658e-01 8.39861035e-01 4.84102666e-01
-1.15564454e+00 6.08612359e-01 -1.06677413e+00 -6.85539067e-01
1.03632379e+00 7.68124402e-01 -7.72110820e-01 -2.57335633e-01
-6.20142877e-01 9.86137569e-01 2.26496354e-01 1.40086919e-01
-8.14193070e-01 -1.10457373e+00 -6.12219989e-01 2.38065079e-01
-1.86158806e-01 -9.20804441e-01 1.17519355e+00 -3.32153618e-01
-1.02298200e+00 1.37540936e+00 1.06470190e-01 -5.35174966e-01
6.04716599e-01 4.52143759e-01 6.76624104e-02 8.05875435e-02
-1.58813536e-01 9.26103771e-01 -9.69697759e-02 -1.09233856e+00
-5.36812186e-01 -5.06605208e-01 -3.96306366e-01 -7.43448511e-02
-8.97095874e-02 -4.31504518e-01 -4.88249570e-01 -1.65506274e-01
-1.97629221e-02 -1.13149476e+00 -6.44146442e-01 5.26723504e-01
-4.46577817e-01 -1.02838971e-01 7.69438267e-01 -6.86346114e-01
5.76270163e-01 -2.10457134e+00 2.05452606e-01 2.60871530e-01
6.15073025e-01 3.60667825e-01 1.87417082e-02 -1.32313937e-01
-2.41916757e-02 3.52645427e-01 -6.38829917e-03 -4.46986347e-01
-8.62032101e-02 3.41227293e-01 3.65614772e-01 8.91400993e-01
-7.68224616e-03 1.05907822e+00 -7.80849278e-01 -7.37263918e-01
2.96126395e-01 6.95108891e-01 -3.91795874e-01 -1.13746986e-01
1.46814495e-01 4.23651010e-01 -1.01655960e-01 9.59257364e-01
6.61435544e-01 -8.23334813e-01 4.90507454e-01 -2.39133537e-01
3.84794205e-01 -1.55795500e-01 -6.80519104e-01 1.35205567e+00
-8.67177993e-02 7.14845181e-01 4.05509204e-01 -7.03445852e-01
5.69053888e-01 3.55999321e-01 5.98548532e-01 -3.60696882e-01
3.57351393e-01 5.40889144e-01 3.36359024e-01 -8.02414119e-02
1.61634937e-01 -3.42378825e-01 1.23646729e-01 8.75573903e-02
3.46878678e-01 1.34819448e-01 5.60889184e-01 8.39807019e-02
1.55462563e+00 -5.97508132e-01 5.71156621e-01 -3.46408337e-01
5.88356197e-01 8.87209252e-02 8.24001312e-01 6.18143082e-01
-6.76830590e-01 6.33756995e-01 7.86074698e-01 -5.24018228e-01
-1.09189224e+00 -8.39670658e-01 -5.52988648e-01 5.68587601e-01
-1.27278522e-01 1.03818700e-01 -5.11098921e-01 -8.44258249e-01
2.99578518e-01 -8.12219456e-02 -1.04670942e+00 9.07329544e-02
-5.35781562e-01 -1.05917764e+00 9.20840979e-01 7.97307611e-01
2.05602139e-01 -1.04596782e+00 -1.26661703e-01 4.92333174e-02
1.32822365e-01 -9.81118023e-01 3.37469727e-02 6.54537320e-01
-8.75136852e-01 -1.52072275e+00 -7.97804832e-01 -9.38622177e-01
9.73478556e-01 -8.00738931e-02 1.06405079e+00 4.61258143e-01
-8.00418377e-01 7.96408057e-02 -1.08019891e-03 -3.94406825e-01
-6.76070452e-01 5.62496901e-01 -1.53030202e-01 -3.74340773e-01
3.73011291e-01 -2.57685810e-01 -6.71810091e-01 -9.78616923e-02
-8.83316159e-01 4.89990078e-02 1.03608632e+00 1.06262374e+00
8.42225313e-01 -1.21817730e-01 4.77357417e-01 -1.20437729e+00
2.90295660e-01 -5.05290508e-01 -4.25993413e-01 2.29828879e-01
-3.01861137e-01 -1.30279496e-01 6.16818070e-01 -2.13543531e-02
-6.80819452e-01 3.51079524e-01 -3.18248838e-01 -2.30567425e-01
-2.93067247e-01 6.30926073e-01 1.02449402e-01 -4.94273692e-01
7.28604674e-01 8.36900547e-02 1.95090353e-01 -3.40854265e-02
-1.61286160e-01 2.70641088e-01 7.63988018e-01 -6.75653890e-02
4.79155958e-01 6.60055459e-01 5.77419102e-01 -6.06383622e-01
-6.65885031e-01 -6.43301845e-01 -7.54439116e-01 -2.81673402e-01
9.99894738e-01 -8.06499064e-01 -7.98604071e-01 7.69803762e-01
-7.86515832e-01 -4.56901819e-01 -3.74506176e-01 4.42906976e-01
-3.28732997e-01 1.47453994e-01 -1.28800750e+00 -3.24558139e-01
-5.46136677e-01 -1.00145054e+00 9.59966004e-01 5.94847977e-01
4.39441502e-02 -1.30604398e+00 3.75759095e-01 3.06480557e-01
5.13889790e-01 3.96598220e-01 1.13395238e+00 -8.20665359e-01
-8.60649526e-01 -5.67335784e-01 -3.49867910e-01 7.85526633e-02
1.30826369e-01 3.10920656e-01 -1.06082511e+00 -3.64265949e-01
-6.28254890e-01 -4.48203713e-01 1.23619998e+00 7.70764351e-01
1.17386246e+00 -2.24472564e-02 -1.02299833e+00 7.96954453e-01
1.70229936e+00 -4.47627753e-02 9.64191735e-01 2.74060786e-01
5.39977610e-01 3.26884270e-01 1.06349580e-01 -2.75697112e-02
1.17311120e-01 -5.92490882e-02 4.39554840e-01 -7.13860512e-01
-2.29040638e-01 3.07094812e-01 -4.38437790e-01 4.56260324e-01
-1.07535116e-01 -3.36287975e-01 -1.31636953e+00 8.27364087e-01
-1.77251184e+00 -9.27980185e-01 -1.41304597e-01 1.67161047e+00
8.99462700e-01 -1.06884293e-01 -3.47826451e-01 2.38103926e-01
6.85697019e-01 -3.97242457e-01 -5.80631852e-01 2.41189733e-01
-8.36762190e-02 2.67563283e-01 7.19955742e-01 8.36403668e-02
-1.15941191e+00 6.17411315e-01 6.91393042e+00 7.64222741e-01
-1.21356964e+00 -1.08272947e-01 1.17631495e+00 9.77156386e-02
1.82387516e-01 -4.51344460e-01 -8.26846361e-01 3.81391868e-02
7.17869163e-01 -2.83098477e-03 4.47459519e-02 6.83920622e-01
-1.43693775e-01 -2.74044927e-02 -1.33815968e+00 7.33295739e-01
-2.01287895e-01 -1.88143516e+00 -3.05007875e-01 5.76780796e-01
6.87814593e-01 4.02154237e-01 -5.17057404e-02 4.32824343e-01
4.09318954e-01 -1.38807988e+00 -1.75211295e-01 8.01447034e-01
7.73578286e-01 -6.08971357e-01 1.53242731e+00 2.27883965e-01
-8.95310581e-01 3.82972136e-02 -2.85628498e-01 4.16616976e-01
-2.75278181e-01 6.81301892e-01 -1.08273697e+00 2.40193710e-01
4.51496989e-01 6.24170542e-01 -7.78376877e-01 1.35589123e+00
3.09252977e-01 5.39286196e-01 -4.09163296e-01 -1.48932591e-01
2.02032514e-02 1.77467167e-01 6.51307106e-02 1.46373641e+00
4.36314017e-01 3.04329507e-02 8.14870745e-02 7.45339036e-01
-5.26926100e-01 -4.23645005e-02 -3.57221961e-01 -3.21964473e-01
4.21440363e-01 1.99747527e+00 -1.19776654e+00 -2.44731978e-01
-1.88457385e-01 3.05813730e-01 5.19915342e-01 1.52802750e-01
-5.60683131e-01 -4.56667356e-02 4.69560355e-01 2.52923936e-01
7.93396011e-02 1.15282468e-01 -5.85160434e-01 -8.64632428e-01
-5.16634166e-01 -4.52386260e-01 4.72330660e-01 -2.05429584e-01
-1.67199576e+00 1.57496065e-01 -7.54271984e-01 -1.08959019e+00
1.05019112e-03 -1.01207781e+00 -5.25017917e-01 5.49963474e-01
-1.65218246e+00 -1.34949577e+00 -5.39666116e-01 1.76807940e-01
-1.61365241e-01 -5.71121871e-02 1.05577683e+00 3.13404202e-01
-6.93834484e-01 4.45856869e-01 3.05079907e-01 5.75695217e-01
7.53338993e-01 -1.64237154e+00 -1.01371840e-01 4.55031455e-01
-5.99013627e-01 6.27427220e-01 4.51785326e-01 -4.72825319e-01
-1.23730636e+00 -1.55030370e+00 8.78242850e-01 -1.62458912e-01
4.10525769e-01 -2.60740221e-01 -8.14353764e-01 7.14654028e-01
1.72236428e-01 6.82850361e-01 1.25834072e+00 1.13207638e-01
2.89331307e-03 -1.59379855e-01 -1.43692064e+00 5.68305016e-01
4.73647714e-01 -4.62162316e-01 9.32859406e-02 4.64501530e-01
1.42320693e-01 -6.04134917e-01 -1.23828590e+00 5.69853187e-01
5.20188451e-01 -7.76783288e-01 8.26289833e-01 -3.61193508e-01
3.58048677e-01 -5.29943109e-01 2.81320643e-02 -9.57450807e-01
-6.04864717e-01 1.08734816e-01 -7.56443143e-02 9.23721433e-01
4.46128845e-01 -3.96074414e-01 1.39437044e+00 4.76910144e-01
-2.89684474e-01 -9.25685167e-01 -9.93012965e-01 -2.56945997e-01
4.56923872e-01 1.18838057e-01 7.96606094e-02 1.04451239e+00
1.27339199e-01 2.62980074e-01 1.58040226e-01 2.56083161e-02
6.77428722e-01 -2.44848058e-01 7.19889879e-01 -1.46696067e+00
7.15202466e-03 -6.19922161e-01 -9.96978641e-01 -2.16490954e-01
1.38356835e-01 -1.31787956e+00 -1.44754916e-01 -1.90676987e+00
7.85343170e-01 -4.08854693e-01 -5.61635375e-01 7.32290030e-01
-1.19037524e-01 7.11283743e-01 -4.84908605e-03 1.90130532e-01
-7.32868135e-01 1.07738815e-01 1.28541255e+00 -5.22026122e-01
1.27521276e-01 -2.51822531e-01 -7.04839051e-01 4.70868558e-01
7.17657387e-01 -4.19214964e-01 7.99264312e-02 2.36719489e-01
2.06131265e-01 8.33031163e-02 5.64205527e-01 -1.32510328e+00
5.56057453e-01 -6.43195119e-03 1.10881650e+00 -8.68050098e-01
2.17037469e-01 -6.37771368e-01 2.45742038e-01 8.26776922e-01
-2.94617802e-01 -5.32859564e-01 3.35541785e-01 5.59764981e-01
-1.96916535e-02 -7.57930353e-02 1.14802337e+00 -4.35600519e-01
-1.24977350e-01 4.63616937e-01 -6.71398640e-01 -3.87598246e-01
1.11031067e+00 -2.77229905e-01 -9.11426723e-01 1.20284468e-01
-9.67402697e-01 4.89905000e-01 6.90530181e-01 -2.84154385e-01
2.62915283e-01 -1.30571830e+00 -7.52196431e-01 -4.22389172e-02
1.66552186e-01 2.32102171e-01 7.03701556e-01 1.11698055e+00
-1.02926660e+00 7.12242365e-01 -4.48930740e-01 -9.84184742e-01
-1.30892396e+00 2.13160738e-01 9.12062645e-01 -1.13896549e+00
-2.35012293e-01 8.35780680e-01 1.07822210e-01 -9.38758254e-01
1.63580239e-01 -4.76099581e-01 -4.21436101e-01 -2.23651335e-01
2.83511728e-01 2.14534372e-01 2.60560662e-01 -3.07872117e-01
-2.66880274e-01 7.96363205e-02 -6.09868050e-01 6.05894923e-01
1.29664993e+00 3.97281557e-01 -3.88855606e-01 2.93968052e-01
1.04558301e+00 -3.76253158e-01 -1.02096748e+00 9.40411091e-02
-9.06064957e-02 4.21483517e-02 2.53245711e-01 -1.08626163e+00
-1.20410490e+00 5.80808342e-01 9.07982469e-01 -7.25912005e-02
9.37286615e-01 1.05899349e-02 6.06014192e-01 6.86663806e-01
1.35387778e-01 -9.50708747e-01 -3.55001241e-01 3.59137356e-01
1.93501025e-01 -1.19134903e+00 7.31945485e-02 -4.24613059e-01
-1.85809955e-01 1.23034024e+00 5.80004930e-01 -1.36608794e-01
6.50742352e-01 7.79767334e-01 3.65867138e-01 -3.92799675e-01
-1.09363866e+00 -1.43745646e-01 -8.19492247e-03 6.02321565e-01
7.50698984e-01 1.32475868e-01 -2.07903057e-01 6.85302258e-01
1.24238655e-02 3.59970987e-01 7.35381246e-01 9.88298059e-01
-4.67426419e-01 -9.58633125e-01 -7.56415948e-02 9.38998818e-01
-7.97014832e-01 7.51548931e-02 -6.59761250e-01 1.06291950e+00
1.38681456e-01 1.20021626e-01 6.53561726e-02 -6.33505732e-02
-2.25904748e-01 -1.29468977e-01 5.79958200e-01 -5.25410950e-01
-7.97894180e-01 1.47744566e-01 -1.09697528e-01 -9.41324160e-02
-3.88723135e-01 -6.73859060e-01 -1.55969167e+00 -4.31762308e-01
-4.17587429e-01 -2.61007756e-01 5.41129410e-01 8.93646240e-01
1.22820705e-01 8.50622356e-01 1.05178080e-01 -5.94548643e-01
-2.02916920e-01 -1.04447401e+00 -1.05986392e+00 7.26468563e-02
4.10450011e-01 -3.25120389e-01 -2.88126171e-01 1.03332162e-01]
|
[15.061395645141602, -3.0759661197662354]
|
eb459a0c-c58c-4d72-b93c-74d438b7794a
|
cross-domain-3d-hand-pose-estimation-with
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Lin_Cross-Domain_3D_Hand_Pose_Estimation_With_Dual_Modalities_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Lin_Cross-Domain_3D_Hand_Pose_Estimation_With_Dual_Modalities_CVPR_2023_paper.pdf
|
Cross-Domain 3D Hand Pose Estimation With Dual Modalities
|
Recent advances in hand pose estimation have shed light on utilizing synthetic data to train neural networks, which however inevitably hinders generalization to real-world data due to domain gaps. To solve this problem, we present a framework for cross-domain semi-supervised hand pose estimation and target the challenging scenario of learning models from labelled multi-modal synthetic data and unlabelled real-world data. To that end, we propose a dual-modality network that exploits synthetic RGB and synthetic depth images. For pre-training, our network uses multi-modal contrastive learning and attention-fused supervision to learn effective representations of the RGB images. We then integrate a novel self-distillation technique during fine-tuning to reduce pseudo-label noise. Experiments show that the proposed method significantly improves 3D hand pose estimation and 2D keypoint detection on benchmarks.
|
['Angela Yao', 'Linlin Yang', 'Qiuxia Lin']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['3d-hand-pose-estimation', 'keypoint-detection', 'hand-pose-estimation', '3d-hand-pose-estimation', 'pseudo-label']
|
['computer-vision', 'computer-vision', 'computer-vision', 'graphs', 'miscellaneous']
|
[ 3.10607284e-01 -1.09780088e-01 -3.66882682e-01 -2.40601555e-01
-1.14655542e+00 -6.92398250e-01 3.35690230e-01 -4.00068581e-01
-5.57999492e-01 8.71802926e-01 3.11147302e-01 2.51782052e-02
2.42459383e-02 -4.78676826e-01 -9.05063927e-01 -6.18964672e-01
3.94003838e-01 8.40568841e-01 1.84345454e-01 -1.28711909e-01
6.40311465e-02 6.76919699e-01 -1.55715406e+00 4.13959563e-01
6.25933230e-01 9.12754357e-01 5.37424922e-01 6.35068178e-01
1.45649254e-01 6.59859717e-01 -5.35326183e-01 -2.18727693e-01
4.23855364e-01 -1.40978947e-01 -1.03922331e+00 2.87297428e-01
7.32892036e-01 -7.58872330e-01 -5.00443339e-01 7.99609900e-01
1.17032230e+00 -6.58060238e-02 7.84758151e-01 -1.32304919e+00
-3.74144763e-01 2.53195316e-01 -6.79405570e-01 -9.68662277e-02
4.88439679e-01 5.07464230e-01 7.97304213e-01 -8.80978048e-01
7.94041693e-01 1.26621509e+00 5.72472334e-01 8.17299366e-01
-1.27825570e+00 -5.48992634e-01 2.74249762e-01 9.59758982e-02
-1.27633309e+00 -1.57971263e-01 1.03920984e+00 -4.43208992e-01
8.56609225e-01 8.00257996e-02 6.94649279e-01 1.85287154e+00
-4.24722701e-01 1.31194603e+00 1.35238349e+00 -7.71902263e-01
-1.37954459e-01 2.27842443e-02 -3.84655148e-01 6.84687555e-01
8.68634954e-02 1.47904426e-01 -6.78960025e-01 1.55171782e-01
1.11926591e+00 -3.56641896e-02 -4.17997211e-01 -7.51770496e-01
-1.29828095e+00 5.05304873e-01 6.62940323e-01 6.44734427e-02
-5.11391342e-01 5.27943708e-02 3.85869116e-01 1.11053355e-01
3.96475315e-01 2.67096341e-01 -8.15353394e-01 2.82687973e-02
-8.18083465e-01 4.54462796e-01 4.78976250e-01 9.14923728e-01
4.93789256e-01 -3.61227989e-01 -2.52105325e-01 9.55202699e-01
4.85491127e-01 7.53438592e-01 3.38234544e-01 -7.78011560e-01
9.70176578e-01 6.11213565e-01 8.09488744e-02 -3.80439252e-01
-5.28988183e-01 -3.82394940e-01 -5.21942139e-01 4.41351682e-01
9.34287727e-01 1.35942250e-01 -1.39158630e+00 1.68605864e+00
3.76049310e-01 -3.27927142e-01 -2.83702254e-01 1.21994019e+00
7.24242330e-01 3.56648955e-03 2.21203074e-01 2.22304806e-01
1.09341741e+00 -1.15532088e+00 -4.44388270e-01 -4.43699777e-01
3.35131854e-01 -6.26649380e-01 1.45463574e+00 3.34723294e-01
-8.43088686e-01 -6.27409637e-01 -7.90763497e-01 -1.93883017e-01
-4.17755216e-01 4.29770082e-01 3.83183748e-01 5.27154744e-01
-5.87798774e-01 3.44841480e-01 -1.00736475e+00 -2.57286251e-01
8.21437240e-01 4.19486284e-01 -6.38220727e-01 -1.86347008e-01
-1.03846538e+00 8.47995162e-01 6.21623755e-01 1.42179638e-01
-9.15068686e-01 -6.07688069e-01 -7.73102164e-01 -5.01628995e-01
5.39339483e-01 -8.26538801e-01 1.11814940e+00 -7.59480715e-01
-1.58615899e+00 1.23661256e+00 1.85170025e-01 -5.41482121e-03
8.71084690e-01 -4.76105630e-01 1.37535125e-01 1.95184410e-01
2.35324100e-01 8.35299671e-01 1.15524364e+00 -1.57966924e+00
-5.11506021e-01 -1.06265593e+00 7.43593797e-02 3.30810308e-01
-3.40150654e-01 -3.46332878e-01 -6.15148008e-01 -9.46874261e-01
2.23657951e-01 -1.10272944e+00 -5.48913516e-02 1.74153298e-01
-7.81553209e-01 -1.72616795e-01 7.79872954e-01 -7.36916602e-01
4.85066861e-01 -1.53320670e+00 6.79391384e-01 3.11402589e-01
1.39356881e-01 2.53427029e-01 -2.21419305e-01 -1.65382382e-02
-6.49704337e-02 -4.78508502e-01 -2.97749639e-01 -7.05858409e-01
-8.08683261e-02 2.06889853e-01 -2.00589210e-01 3.78097534e-01
2.51336664e-01 1.06933868e+00 -8.97513449e-01 -7.38175213e-01
5.09984434e-01 8.56831789e-01 -4.17127550e-01 4.65846151e-01
-5.78231156e-01 1.04408324e+00 -3.41569394e-01 8.62342536e-01
6.05019808e-01 -3.21393192e-01 6.36578584e-03 -4.44846332e-01
3.06749582e-01 1.45518824e-01 -1.20158625e+00 2.49353647e+00
-6.72728479e-01 3.21975887e-01 -1.41725972e-01 -7.27230668e-01
4.20398086e-01 2.27157146e-01 3.77875417e-01 -6.07685208e-01
3.41502458e-01 3.44443291e-01 -4.79524314e-01 -6.14725649e-01
1.29835770e-01 -1.03147127e-01 2.68338740e-01 5.86297512e-01
2.60029107e-01 -3.05975705e-01 -2.18139961e-01 -1.37676641e-01
5.76164365e-01 8.32795620e-01 -4.14518006e-02 3.24029297e-01
4.07720327e-01 -1.92478653e-02 -2.93174461e-02 4.98029321e-01
-2.14006081e-01 8.77529263e-01 1.52124196e-01 -3.80065113e-01
-1.07933676e+00 -1.17842698e+00 1.01379745e-01 1.33229339e+00
-4.29346412e-02 -4.95947227e-02 -8.06908727e-01 -1.21491075e+00
2.42906734e-01 7.50151947e-02 -8.92640412e-01 6.88548461e-02
-7.21674383e-01 -3.99100721e-01 4.82152790e-01 1.08508229e+00
4.98969853e-01 -1.16302407e+00 -8.38456035e-01 -2.30825227e-02
-3.84216666e-01 -1.23597431e+00 -3.32217485e-01 4.20143664e-01
-6.90972805e-01 -1.13458502e+00 -1.45594501e+00 -9.22706068e-01
6.20683491e-01 -4.89202095e-03 9.73890126e-01 -4.36348051e-01
-5.55431068e-01 5.84930480e-01 -2.54703075e-01 -3.32194656e-01
-1.86328769e-01 6.76410019e-01 1.32056892e-01 -2.21176460e-01
1.69964686e-01 -5.51746309e-01 -7.63177395e-01 2.85082519e-01
-5.02593696e-01 -9.76520330e-02 8.30852568e-01 1.03403139e+00
6.03629708e-01 -6.44829988e-01 3.14109981e-01 -5.07464230e-01
4.55653727e-01 2.38824710e-01 -4.00270343e-01 4.74731028e-01
-3.41597289e-01 3.84260565e-01 1.91487297e-01 -7.10399508e-01
-1.27990830e+00 6.40324295e-01 -2.24992543e-01 -7.13157296e-01
-2.76691079e-01 8.15192387e-02 -3.43661129e-01 -3.05118114e-01
9.18623447e-01 1.11292779e-01 1.40276611e-01 -6.06162190e-01
5.49180210e-01 9.41610992e-01 7.63967335e-01 -8.29956949e-01
7.31903732e-01 7.57368445e-01 -4.90014255e-02 -3.99520695e-01
-9.97364521e-01 -2.93546081e-01 -1.14973474e+00 -1.75394014e-01
7.78605223e-01 -1.03741181e+00 -8.80751550e-01 8.05323184e-01
-1.32242107e+00 -5.90196431e-01 -2.76808828e-01 5.57826340e-01
-9.47905779e-01 2.46277660e-01 -3.53170961e-01 -8.10415268e-01
-3.26900482e-01 -1.26902604e+00 1.68656433e+00 -3.15453932e-02
-9.35302600e-02 -6.56429768e-01 2.66738266e-01 6.36071622e-01
8.57582167e-02 3.39106709e-01 4.39461380e-01 -2.23958358e-01
-4.99257296e-01 -2.64212281e-01 -4.54503685e-01 3.53567779e-01
2.77232975e-01 -6.49993718e-01 -1.28255272e+00 -4.27399486e-01
-4.06966001e-01 -1.00750756e+00 8.33142638e-01 3.82457584e-01
1.27904582e+00 9.07903686e-02 -4.07180339e-01 6.10147178e-01
9.51005459e-01 -3.97121340e-01 2.94389993e-01 5.46982288e-01
1.16139901e+00 8.59053910e-01 6.26077890e-01 4.70754564e-01
4.72908199e-01 9.69985604e-01 5.53954363e-01 -6.72766045e-02
-4.35614288e-01 -4.38762665e-01 -1.64757431e-01 4.06853780e-02
-5.67473829e-01 9.18281823e-02 -1.11903512e+00 4.62033391e-01
-1.57906997e+00 -6.02751672e-01 3.57474834e-01 1.97238660e+00
1.06593788e+00 9.26498845e-02 6.64019465e-01 5.13831437e-01
5.81543982e-01 -1.46058444e-02 -7.69744515e-01 6.31684482e-01
-6.50802702e-02 4.68372166e-01 5.54698288e-01 3.84594828e-01
-1.35157144e+00 1.10157776e+00 5.31026411e+00 6.49644017e-01
-1.24914587e+00 2.07086071e-01 1.16335183e-01 -8.04260597e-02
1.04778223e-01 -6.39423668e-01 -6.09430850e-01 1.90790430e-01
1.58051088e-01 6.54908240e-01 3.57845813e-01 9.73330855e-01
-1.82889178e-01 7.34204128e-02 -1.25694788e+00 1.39471674e+00
2.11738661e-01 -8.08808982e-01 1.99516509e-02 -2.41907574e-02
6.86029673e-01 -1.85577024e-03 3.60085428e-01 -2.06279475e-02
3.72456014e-02 -9.99373496e-01 7.48146772e-01 3.21907192e-01
1.13503206e+00 -5.64508796e-01 4.24395889e-01 3.47753644e-01
-1.17310059e+00 -1.56819507e-01 1.52649805e-01 1.53093368e-01
1.67051956e-01 -5.40319597e-04 -1.08167350e+00 3.74610096e-01
7.28016376e-01 6.40914321e-01 -5.69883764e-01 6.84860885e-01
-5.13500512e-01 -8.22096094e-02 -4.41900730e-01 1.82975098e-01
-2.58148145e-02 4.63417470e-01 4.40601289e-01 9.27143216e-01
-1.97424188e-01 -2.49517262e-01 -5.25659323e-02 7.48764336e-01
1.01491816e-01 -1.34514973e-01 -4.60332543e-01 1.15706891e-01
2.85495073e-01 7.47389674e-01 -7.08849132e-01 -8.74663070e-02
-1.03463069e-01 1.50254512e+00 5.21540582e-01 3.36222887e-01
-5.69011927e-01 -2.71438688e-01 3.67742419e-01 1.58258379e-01
2.04508185e-01 -2.98869818e-01 -4.61268157e-01 -1.21962798e+00
3.41065258e-01 -9.12456214e-01 3.59090418e-01 -6.28426850e-01
-1.32872236e+00 6.44951582e-01 -1.77241676e-02 -1.24637556e+00
-4.28613901e-01 -1.05410028e+00 8.45897868e-02 8.55646312e-01
-1.72922957e+00 -1.80063295e+00 -7.55395532e-01 1.04590046e+00
5.27628481e-01 -2.83839256e-01 8.83244574e-01 2.91054249e-01
-1.91180915e-01 8.90916288e-01 -3.39654326e-01 2.85303533e-01
1.02903688e+00 -1.35948479e+00 4.24842656e-01 2.77532607e-01
2.15092719e-01 4.91541237e-01 5.16611457e-01 -5.80402076e-01
-1.21115637e+00 -9.06538546e-01 2.32014403e-01 -8.38581741e-01
1.39702484e-01 -4.27098006e-01 -5.00216603e-01 7.07775474e-01
-4.04711008e-01 4.29096282e-01 3.34893435e-01 1.64835006e-01
-7.77642488e-01 6.42122254e-02 -1.27657855e+00 3.68662477e-01
1.52763748e+00 -9.33620691e-01 -6.36798203e-01 5.92378855e-01
3.35770905e-01 -8.22067142e-01 -8.00542474e-01 5.00222087e-01
1.12769580e+00 -7.17287719e-01 1.38271618e+00 -8.15243363e-01
3.63994986e-01 -5.21743223e-02 -3.21463823e-01 -1.25483525e+00
2.57476479e-01 -1.48405954e-01 -3.08888704e-01 7.68338919e-01
1.02217302e-01 -2.73391455e-01 1.43963611e+00 2.60380119e-01
2.45877832e-01 -6.59590602e-01 -9.41740334e-01 -7.87859142e-01
1.52466357e-01 -4.62928861e-01 4.61855084e-01 6.21547580e-01
-1.64927855e-01 3.76391739e-01 -4.44447517e-01 1.14460126e-01
1.10971153e+00 2.24690288e-01 1.14163494e+00 -1.32504547e+00
-3.57011557e-01 -2.54860222e-01 -4.59465921e-01 -1.34884667e+00
4.11454558e-01 -6.32889569e-01 8.68292525e-02 -1.44784975e+00
1.20437883e-01 -4.44838434e-01 4.64966670e-02 5.70573747e-01
-2.59116054e-01 7.55475461e-01 1.31649271e-01 2.68145233e-01
-3.81294787e-01 5.96018434e-01 1.64144838e+00 -2.28697926e-01
-3.48555028e-01 1.67642698e-01 -1.98287651e-01 5.82630813e-01
5.36578059e-01 -2.58502692e-01 -3.25468570e-01 -4.20188457e-01
-1.32498927e-02 2.40772534e-02 7.95708001e-01 -8.75368118e-01
4.26815823e-02 5.34225479e-02 5.57845891e-01 -8.52490962e-01
6.35828733e-01 -1.00097930e+00 -4.16943192e-01 4.97054815e-01
-3.74309301e-01 -4.18478340e-01 7.04504848e-02 5.88357151e-01
1.24178849e-01 2.54809171e-01 7.42910922e-01 -2.34578967e-01
-5.99978089e-01 4.24599320e-01 3.05786669e-01 1.73996955e-01
9.86488044e-01 -3.74962687e-01 -1.33248018e-02 -2.93343544e-01
-1.03998184e+00 1.75230905e-01 2.93138832e-01 5.54259419e-01
5.86097002e-01 -1.43148243e+00 -5.52908063e-01 4.26547647e-01
4.46478635e-01 4.39576983e-01 2.58326769e-01 6.65312350e-01
-3.14522058e-01 5.39647520e-01 -6.22380018e-01 -8.98566663e-01
-1.29353225e+00 5.25438845e-01 3.99655879e-01 -1.40606731e-01
-5.06606579e-01 1.23665857e+00 -8.16106796e-02 -8.69949460e-01
8.63860250e-01 -2.59957910e-01 1.33825779e-01 9.97520238e-02
3.41264516e-01 3.58435214e-01 1.64614975e-01 -7.62079239e-01
-5.37573695e-01 8.88623476e-01 -1.12247527e-01 -3.39954227e-01
1.37739170e+00 1.08350150e-01 3.01654518e-01 4.55687165e-01
1.37599671e+00 -3.14534485e-01 -1.64930558e+00 -4.57246304e-01
-1.97383314e-01 -6.21252537e-01 -7.85347074e-02 -1.17873359e+00
-1.03460097e+00 1.08407533e+00 9.83609200e-01 -4.66523468e-01
1.04750311e+00 3.44933927e-01 7.22108543e-01 6.02171481e-01
4.16999489e-01 -1.16859889e+00 7.19545245e-01 2.80825049e-01
1.07264721e+00 -1.68753159e+00 -1.30243659e-01 -4.83579695e-01
-6.13689363e-01 1.04008520e+00 8.62037122e-01 5.45423403e-02
4.88561660e-01 2.16132760e-01 2.12742120e-01 -1.12542257e-01
-2.94720149e-03 -5.29133320e-01 4.04677361e-01 9.83524919e-01
2.07040623e-01 -3.79429199e-02 2.53787994e-01 4.19504762e-01
-4.77738082e-02 3.25295210e-01 -3.33789229e-01 1.22120249e+00
2.07898133e-02 -1.24778223e+00 -7.29680419e-01 2.52712250e-01
-2.13346213e-01 2.42488340e-01 -4.40584064e-01 8.95684123e-01
2.14614481e-01 3.43530446e-01 -2.69500494e-01 -4.88796473e-01
6.47963643e-01 1.44278407e-01 1.42084265e+00 -5.72825015e-01
-3.75341475e-01 3.92504856e-02 -3.33093613e-01 -4.48737323e-01
-6.52582407e-01 -4.55213219e-01 -8.49595010e-01 1.76293656e-01
-4.78879958e-01 -3.85035336e-01 7.05450714e-01 1.15231907e+00
8.66319463e-02 5.71415067e-01 3.30241024e-01 -1.57929826e+00
-8.32940519e-01 -1.19793940e+00 -3.47188383e-01 3.44436407e-01
6.25747383e-01 -1.30797637e+00 -1.75425097e-01 -7.63269290e-02]
|
[6.7218828201293945, -0.8396722674369812]
|
9d0c893f-0f9f-4852-ac76-ff30d0a8bae5
|
domain-agnostic-learning-with-disentangled
|
1904.12347
| null |
http://arxiv.org/abs/1904.12347v1
|
http://arxiv.org/pdf/1904.12347v1.pdf
|
Domain Agnostic Learning with Disentangled Representations
|
Unsupervised model transfer has the potential to greatly improve the
generalizability of deep models to novel domains. Yet the current literature
assumes that the separation of target data into distinct domains is known as a
priori. In this paper, we propose the task of Domain-Agnostic Learning (DAL):
How to transfer knowledge from a labeled source domain to unlabeled data from
arbitrary target domains? To tackle this problem, we devise a novel Deep
Adversarial Disentangled Autoencoder (DADA) capable of disentangling
domain-specific features from class identity. We demonstrate experimentally
that when the target domain labels are unknown, DADA leads to state-of-the-art
performance on several image classification datasets.
|
['Zijun Huang', 'Kate Saenko', 'Xingchao Peng', 'Ximeng Sun']
|
2019-04-28
| null | null | null | null |
['multi-target-domain-adaptation']
|
['computer-vision']
|
[ 3.91514450e-01 1.95348859e-01 -1.37598649e-01 -5.52944064e-01
-8.39932501e-01 -9.09433722e-01 7.88599432e-01 -3.48749787e-01
-3.19152415e-01 1.00244641e+00 -6.94943294e-02 -1.71923652e-01
-7.61053562e-02 -7.05228150e-01 -6.90265119e-01 -7.02028692e-01
6.80851936e-02 8.92772675e-01 -1.11692391e-01 -1.60687685e-01
-2.32057929e-01 6.29132211e-01 -9.59501922e-01 2.14734808e-01
6.78333580e-01 7.33303607e-01 -4.18484569e-01 4.38482970e-01
3.26675475e-01 7.48769581e-01 -6.01538241e-01 -4.39636022e-01
5.21901786e-01 -4.25425917e-01 -9.35388863e-01 1.02768056e-01
4.02588099e-01 -5.56774676e-01 -5.14714181e-01 1.13034010e+00
2.78933346e-01 8.48138239e-04 1.23825431e+00 -1.58260322e+00
-1.13837183e+00 3.45705599e-01 -2.68384844e-01 1.32622719e-01
-2.96777248e-01 1.33144677e-01 9.72151101e-01 -7.61200070e-01
5.81682563e-01 9.37525690e-01 5.46771049e-01 9.22596216e-01
-1.78293717e+00 -1.26883721e+00 8.16572160e-02 5.90642504e-02
-1.32711565e+00 -3.82509083e-01 1.05294228e+00 -8.14812183e-01
5.24292946e-01 -3.08982700e-01 -5.19518368e-02 1.78671634e+00
-2.26128101e-01 6.16502166e-01 1.33893383e+00 -3.59458357e-01
4.20524508e-01 3.49972278e-01 7.33268186e-02 4.21957374e-01
5.22380352e-01 5.57485580e-01 -3.85041922e-01 -1.62476540e-01
7.45505691e-01 -2.35997066e-01 9.88869928e-03 -9.74543869e-01
-9.75449145e-01 1.17945659e+00 2.18840614e-01 6.16104156e-02
-2.06823796e-01 -1.56459078e-01 3.31388086e-01 6.10341847e-01
5.78864276e-01 8.19506049e-01 -7.43575096e-01 3.80215943e-01
-4.65958267e-01 6.67635724e-02 9.56003487e-01 1.01574600e+00
9.19919729e-01 3.55678469e-01 3.34900290e-01 5.42638183e-01
5.11939563e-02 5.84690809e-01 5.64853668e-01 -8.14181268e-01
2.77262151e-01 5.26298165e-01 1.34162784e-01 -4.75966871e-01
-5.95144406e-02 -3.73398036e-01 -7.62807190e-01 5.02168775e-01
6.64691508e-01 -5.62036395e-01 -1.13503265e+00 2.11993289e+00
1.16717175e-01 3.34250480e-01 6.91341877e-01 7.95796990e-01
6.85644686e-01 3.49087328e-01 2.83110350e-01 3.23275536e-01
9.51633990e-01 -6.31981015e-01 -3.50198328e-01 -7.00957298e-01
1.96712881e-01 -3.33859354e-01 7.66141891e-01 3.27534467e-01
-5.09631813e-01 -5.18734753e-01 -1.31808555e+00 -5.77641316e-02
-5.75753629e-01 -1.68294106e-02 6.29277170e-01 7.42898583e-01
-5.50439358e-01 3.46288174e-01 -8.16200733e-01 -3.44931483e-01
7.94991195e-01 5.86373389e-01 -9.33451414e-01 -1.93879530e-01
-1.28743029e+00 1.00370824e+00 7.34699190e-01 -5.41720927e-01
-1.47122848e+00 -7.43328929e-01 -9.71223712e-01 1.67585567e-01
2.86962122e-01 -7.63493121e-01 1.33811402e+00 -1.49268854e+00
-1.46147287e+00 1.36539114e+00 1.67348772e-01 -6.47430182e-01
3.06844682e-01 -1.91676527e-01 -6.26115143e-01 1.23591870e-01
1.03825338e-01 4.26408827e-01 1.20788574e+00 -1.37241924e+00
-4.09294009e-01 -5.29807687e-01 2.23212689e-01 -5.38577549e-02
-5.74200511e-01 -2.80208886e-01 2.83848524e-01 -7.90564358e-01
-3.13135535e-01 -9.83973145e-01 -6.48847595e-02 1.37118593e-01
-3.34912807e-01 3.95058058e-02 8.59714746e-01 -4.38140750e-01
3.52631062e-01 -2.35227299e+00 3.34607393e-01 -4.81517576e-02
3.93184364e-01 5.02084672e-01 -3.16552430e-01 3.16530049e-01
-4.35346097e-01 -1.43817484e-01 -3.51293981e-01 -1.73434734e-01
1.05364822e-01 5.39346099e-01 -7.12127388e-01 5.92997432e-01
7.02292502e-01 7.60109425e-01 -9.23325419e-01 -1.62807539e-01
-5.09294048e-02 3.73938620e-01 -4.35559571e-01 5.52716374e-01
-1.88987255e-01 7.44467020e-01 -5.14098942e-01 3.26221228e-01
7.61451960e-01 -5.15289187e-01 3.95425439e-01 5.52695952e-02
6.68352664e-01 6.84079677e-02 -8.36462438e-01 1.60475922e+00
-3.24059010e-01 8.28209817e-01 -1.95625313e-02 -1.31350756e+00
9.16012526e-01 3.18040937e-01 2.23259285e-01 -3.36479872e-01
2.06850901e-01 3.87273654e-02 1.89011887e-01 -1.14079632e-01
-2.42765266e-02 -7.95742095e-01 -3.17276686e-01 3.95152867e-01
6.75134480e-01 2.23754835e-03 -4.12879407e-01 5.12322411e-02
1.10026443e+00 3.25844623e-02 4.54176784e-01 -2.01825753e-01
3.51685643e-01 2.56757736e-01 7.04986274e-01 6.24410570e-01
-4.60539818e-01 7.32622266e-01 4.99943912e-01 -3.33622068e-01
-1.15660834e+00 -1.88079619e+00 6.06939197e-02 1.09128237e+00
1.18286973e-02 3.27841580e-01 -5.55817306e-01 -1.13935757e+00
3.56058240e-01 8.51390302e-01 -9.04351711e-01 -6.34065807e-01
-3.21466237e-01 -5.53619266e-01 6.91295862e-01 7.10696578e-01
3.78339887e-01 -7.05287099e-01 -1.30292580e-01 -1.33263152e-02
1.64542586e-01 -1.38274097e+00 -9.86096710e-02 5.34077108e-01
-7.78603613e-01 -1.30075264e+00 -8.34905446e-01 -9.47891116e-01
8.32379162e-01 -5.32165691e-02 1.26013947e+00 -8.59022200e-01
4.96816114e-02 4.48017091e-01 -2.29814142e-01 -4.94464695e-01
-6.32402539e-01 5.60405105e-02 2.58893341e-01 2.86185831e-01
7.54808903e-01 -8.79085541e-01 -4.06547636e-01 1.90646812e-01
-9.37209964e-01 -2.76357502e-01 6.21810853e-01 1.10849071e+00
4.36909050e-01 4.95080315e-02 8.50611329e-01 -1.32504940e+00
5.98999798e-01 -8.18007231e-01 -5.39145052e-01 1.72139063e-01
-5.44283330e-01 2.66498089e-01 1.12514639e+00 -8.85087192e-01
-1.11017704e+00 3.51576984e-01 2.40332603e-01 -7.52881885e-01
-6.42471552e-01 2.60469496e-01 -5.83977461e-01 1.13325389e-02
9.48678851e-01 2.06361532e-01 6.38927892e-02 -5.95735669e-01
6.11218452e-01 5.35746574e-01 8.00901532e-01 -7.38507569e-01
1.17853713e+00 6.67335868e-01 -3.36994231e-02 -4.10003155e-01
-1.03025591e+00 -2.99821734e-01 -8.99372041e-01 4.74946469e-01
8.88692677e-01 -1.33192015e+00 -1.61338240e-01 4.21755433e-01
-9.90985334e-01 -3.77083361e-01 -5.19849300e-01 4.49353397e-01
-7.02199101e-01 3.94484587e-02 -2.55777359e-01 -1.69665918e-01
-2.37979796e-02 -9.14109051e-01 5.51861584e-01 7.91745633e-02
-3.69144440e-01 -1.21553588e+00 2.46807739e-01 2.69865036e-01
1.33921281e-01 3.53584290e-01 1.07554960e+00 -1.70029414e+00
-3.58996153e-01 -3.03680629e-01 -1.95883825e-01 8.24883759e-01
3.23105574e-01 -7.00083852e-01 -1.32944620e+00 -4.55242336e-01
1.20318569e-01 -6.84025824e-01 7.23898053e-01 -9.94822383e-02
8.84075701e-01 -3.79604816e-01 -2.89400071e-01 6.90882266e-01
1.34542394e+00 1.31403163e-01 2.29163304e-01 3.22855860e-01
6.05016470e-01 5.30507326e-01 3.51331651e-01 2.90896922e-01
1.12326078e-01 3.76206100e-01 3.39676023e-01 -1.15581661e-01
-1.20280005e-01 -4.71329540e-01 2.24536225e-01 2.29446381e-01
4.77999270e-01 -2.77394295e-01 -9.42867696e-01 9.15981710e-01
-1.48732769e+00 -6.85317755e-01 4.39398855e-01 1.90141582e+00
9.06318426e-01 9.22747105e-02 -7.21323341e-02 -2.44421735e-01
6.63192511e-01 5.64475954e-02 -1.06199801e+00 -1.89363956e-01
-5.91631271e-02 5.72773635e-01 5.47448516e-01 3.05023491e-01
-1.43559313e+00 8.90465438e-01 6.38353300e+00 5.12657702e-01
-1.13854575e+00 2.50156701e-01 3.26610148e-01 1.08669691e-01
-6.33996576e-02 4.97073494e-02 -5.18957496e-01 2.57918566e-01
9.20465112e-01 -5.45381904e-01 2.39688873e-01 1.23207963e+00
-5.52927375e-01 5.56904733e-01 -1.72664320e+00 7.95862079e-01
1.23839334e-01 -9.75196660e-01 2.11814448e-01 1.14559896e-01
8.95171344e-01 1.06144108e-01 4.66308922e-01 5.98200202e-01
1.07682431e+00 -1.08351767e+00 2.75798172e-01 1.57051444e-01
1.00968456e+00 -8.35257649e-01 4.48391050e-01 4.13533211e-01
-3.51687729e-01 1.53636765e-02 -1.11149281e-01 -7.70294666e-02
-3.12449247e-01 3.10969353e-01 -1.18321919e+00 4.07981664e-01
2.65727669e-01 7.42430627e-01 -3.72405320e-01 5.48663735e-01
-4.34443057e-01 7.20716178e-01 -5.57912737e-02 5.84819555e-01
1.07468799e-01 4.18764278e-02 5.07480681e-01 9.11742747e-01
-3.99924442e-02 1.20157525e-01 -4.14173771e-03 1.10349345e+00
-4.75067884e-01 -4.82893258e-01 -9.88328516e-01 -3.73432249e-01
3.27029109e-01 6.85237348e-01 -3.15691739e-01 -3.02635252e-01
-4.97768432e-01 1.28104174e+00 4.99059021e-01 6.91073418e-01
-5.92187941e-01 -3.82293642e-01 1.17147017e+00 -1.07844882e-01
5.44357955e-01 -1.04702465e-01 -2.50267893e-01 -1.53385139e+00
-2.69708842e-01 -1.03070426e+00 4.55774218e-01 -5.40029585e-01
-2.09559202e+00 4.14534181e-01 -1.99630447e-02 -1.38107061e+00
-4.02484328e-01 -9.46553469e-01 -3.56392026e-01 1.09158599e+00
-1.58000278e+00 -1.42020214e+00 8.39883909e-02 9.05420482e-01
2.65281349e-01 -7.32429147e-01 1.31464481e+00 8.35644230e-02
-2.77325004e-01 8.15855682e-01 6.64485157e-01 7.83360720e-01
1.08679450e+00 -1.24753237e+00 5.25505543e-01 8.00386369e-01
3.09480369e-01 5.48027813e-01 7.90667415e-01 -3.22990000e-01
-1.13221383e+00 -1.23124170e+00 5.04479587e-01 -7.61004567e-01
8.69697511e-01 -5.36056519e-01 -1.00490487e+00 1.22003341e+00
5.70143573e-02 5.32282054e-01 1.12265420e+00 1.63829550e-01
-1.12345850e+00 -2.01464966e-01 -1.39815593e+00 3.13125491e-01
5.67757428e-01 -1.03621900e+00 -1.02188241e+00 1.82834059e-01
6.98686779e-01 -1.39173582e-01 -8.95689189e-01 3.62075239e-01
4.40838754e-01 -6.90606833e-01 1.13011050e+00 -1.39114785e+00
5.30608773e-01 1.46131115e-02 -2.85245091e-01 -1.57592857e+00
-2.32914403e-01 -1.96737424e-01 -2.01335073e-01 1.18478227e+00
2.70192474e-01 -8.74462664e-01 9.57024872e-01 7.63943970e-01
2.82036275e-01 -7.15954751e-02 -9.48142290e-01 -1.10484362e+00
7.29741037e-01 -6.56615868e-02 4.97941226e-01 1.48864090e+00
-2.88517147e-01 6.54008746e-01 -3.48076671e-01 6.09640300e-01
7.94382513e-01 2.82905847e-01 8.01369011e-01 -1.55349791e+00
-4.44856495e-01 -1.03189521e-01 -5.94273090e-01 -8.57410908e-01
8.94025981e-01 -9.37407613e-01 -2.33556807e-01 -9.87047076e-01
2.49508753e-01 -2.83839524e-01 -6.65751457e-01 5.78937888e-01
-1.44518735e-02 1.98502332e-01 8.64511207e-02 5.03640957e-02
-4.36243206e-01 6.41407609e-01 1.03697276e+00 -3.98209155e-01
2.25554779e-01 6.04739077e-02 -1.07475877e+00 7.63546169e-01
7.61747539e-01 -9.15511847e-01 -6.79741085e-01 -4.50420201e-01
-4.01919961e-01 -7.74396360e-02 5.99714458e-01 -9.28787112e-01
3.34681332e-04 -2.96009094e-01 5.34386635e-01 1.28851021e-02
5.26476979e-01 -1.20669258e+00 -3.73726450e-02 2.96460152e-01
-4.39123482e-01 -4.68048990e-01 4.35158551e-01 8.27271760e-01
-3.66839230e-01 -1.57610163e-01 9.15123582e-01 1.50461365e-02
-1.00616622e+00 2.86579072e-01 -1.34185597e-01 5.07623374e-01
1.12794590e+00 4.76246327e-02 -2.99988717e-01 -3.03368926e-01
-1.00355053e+00 -4.75611212e-03 5.59985757e-01 5.51845372e-01
4.97051537e-01 -1.29433370e+00 -7.68966019e-01 4.53223497e-01
4.44043338e-01 -1.04941644e-01 -3.64545360e-02 3.82501557e-02
-7.88279176e-02 2.89079398e-01 -5.71020544e-01 -3.24592501e-01
-1.04267967e+00 9.19789135e-01 4.22245979e-01 -2.74504542e-01
-3.15070421e-01 7.90199220e-01 8.47297490e-01 -7.59604275e-01
-3.52615602e-02 1.99079007e-01 6.30983114e-02 -1.62983909e-01
1.63020611e-01 3.26659270e-02 -2.11790115e-01 -6.93185449e-01
-2.40717798e-01 1.37918025e-01 -3.51721972e-01 -7.06687868e-02
1.33228683e+00 1.03406601e-01 4.11057055e-01 3.51490289e-01
1.49503350e+00 -1.83705628e-01 -1.56450438e+00 -4.73541886e-01
-1.20342270e-01 -3.54803950e-01 -3.02070826e-01 -1.12909961e+00
-8.34537268e-01 1.09387743e+00 7.33008087e-01 -1.64929345e-01
1.09220266e+00 4.07091789e-02 5.25825441e-01 3.87736976e-01
2.36523315e-01 -6.05585337e-01 1.85827777e-01 5.59162676e-01
6.40988767e-01 -1.66938150e+00 -1.71571642e-01 -7.67159760e-02
-8.64373744e-01 8.31878185e-01 7.27688551e-01 -5.74596703e-01
8.03575873e-01 9.71393362e-02 1.37486503e-01 1.32754259e-02
-4.99914110e-01 -2.19402060e-01 4.19477731e-01 1.26506186e+00
-7.11819008e-02 1.57546684e-01 3.48587304e-01 1.02890122e+00
-1.25031341e-02 6.88297600e-02 3.90364170e-01 7.35650957e-01
3.96947982e-03 -1.31240714e+00 -1.26792178e-01 1.15612820e-01
-6.11176193e-01 -3.56061151e-03 -6.46052897e-01 1.02959883e+00
4.44594920e-02 7.16703415e-01 -8.27409327e-02 -2.76669651e-01
2.90448457e-01 2.72393823e-01 4.83000964e-01 -8.68017316e-01
-2.05574602e-01 -5.03603578e-01 -5.19623756e-02 -1.09844036e-01
-3.47523600e-01 -3.62807453e-01 -9.36414182e-01 3.92078571e-02
3.83135937e-02 1.45670608e-01 3.12921435e-01 9.71881330e-01
5.13309240e-01 2.83471197e-01 4.56298292e-01 -4.59920347e-01
-1.07003474e+00 -7.13055491e-01 -6.93202734e-01 8.28634083e-01
7.97023714e-01 -8.38209331e-01 -4.55895811e-01 3.66128564e-01]
|
[10.261934280395508, 3.0280961990356445]
|
32b62954-edfd-48a7-bc24-1d115358365c
|
estimating-treatment-effects-using
|
2211.04370
| null |
https://arxiv.org/abs/2211.04370v3
|
https://arxiv.org/pdf/2211.04370v3.pdf
|
NESTER: An Adaptive Neurosymbolic Method for Treatment Effect Estimation
|
Treatment effect estimation from observational data is a central problem in causal inference. Methods based on potential outcomes framework solve this problem by exploiting inductive biases and heuristics from causal inference. Each of these methods addresses a specific aspect of treatment effect estimation, such as controlling propensity score, enforcing randomization, etc., by designing neural network architectures and regularizers. In this paper, we propose an adaptive method called Neurosymbolic Treatment Effect Estimator (NESTER), a generalized method for treatment effect estimation. NESTER brings together the ideas used in existing methods based on multi-head neural networks for treatment effect estimation into one framework. To perform program synthesis, we design a Domain Specific Language (DSL) for treatment effect estimation based on inductive biases used in literature. We also theoretically study NESTER's capability for treatment effect estimation. Our comprehensive empirical results show that NESTER performs better than state-of-the-art methods on benchmark datasets without compromising run time requirements.
|
['Vineeth N Balasubramanian', 'Abbavaram Gowtham Reddy']
|
2022-11-08
| null | null | null | null |
['program-synthesis']
|
['computer-code']
|
[ 4.41746384e-01 2.70624936e-01 -1.32446587e+00 -5.33489227e-01
-6.18767381e-01 -1.85241207e-01 5.28361559e-01 2.67533213e-01
-2.02322543e-01 1.27464426e+00 1.03491998e+00 -8.18151593e-01
-6.39172912e-01 -1.15668309e+00 -9.68816936e-01 -5.73819041e-01
-2.08478183e-01 3.33896816e-01 -3.60093594e-01 1.86601907e-01
5.97188532e-01 3.54489744e-01 -1.35015094e+00 3.24107468e-01
1.10916615e+00 3.35374594e-01 -3.44709277e-01 2.99322933e-01
1.51035354e-01 1.13867950e+00 -9.93921757e-02 -2.33917847e-01
-6.38082400e-02 -2.38986865e-01 -8.51567805e-01 -8.18711698e-01
5.88158429e-01 -6.08837306e-01 -2.38449231e-01 7.53821015e-01
7.60073900e-01 -2.34607905e-02 1.14335144e+00 -1.56704569e+00
-9.80789900e-01 1.51009107e+00 -5.48942208e-01 -7.76323080e-02
2.33492643e-01 1.33781627e-01 8.88996840e-01 -5.40583074e-01
5.45238018e-01 1.51979828e+00 9.54683006e-01 6.57713354e-01
-1.42156315e+00 -1.07662141e+00 4.10201997e-02 1.25950560e-01
-7.84711838e-01 -4.56983536e-01 6.04753137e-01 -6.73164010e-01
8.05780113e-01 4.34270382e-01 4.06136185e-01 1.53701413e+00
5.43352664e-01 9.16093171e-01 1.21916962e+00 -5.40293932e-01
4.57592905e-01 -2.60660887e-01 5.12015939e-01 4.82089907e-01
4.21663880e-01 9.76975381e-01 -3.75724524e-01 -5.57101130e-01
9.91299808e-01 -2.78512808e-03 -1.91833973e-01 -4.73919958e-01
-1.23319757e+00 1.39754105e+00 7.02281117e-01 -8.28654096e-02
-6.55373454e-01 6.09428704e-01 5.98076999e-01 -1.13105476e-01
5.43038666e-01 4.49162185e-01 -6.87064469e-01 3.34290534e-01
-7.42106199e-01 6.52324915e-01 7.39162803e-01 8.61997306e-01
1.39624253e-01 8.55288003e-03 -1.03702009e+00 6.87033951e-01
2.01832235e-01 4.33543712e-01 3.07847649e-01 -9.56989348e-01
2.71774530e-01 5.94630420e-01 8.00595060e-02 -7.86616802e-01
-6.66423082e-01 -8.43251944e-02 -1.25063717e+00 3.94937675e-03
4.56097037e-01 -5.35860896e-01 -1.03873456e+00 2.04358411e+00
4.61763620e-01 1.80615678e-01 -1.49001390e-01 6.53641045e-01
1.12523270e+00 5.85742831e-01 8.12116802e-01 -4.20554847e-01
1.24626184e+00 -7.62531817e-01 -9.40539420e-01 -1.03218056e-01
6.88265622e-01 -2.55748719e-01 7.73667514e-01 1.88531801e-01
-1.15652955e+00 -8.79845470e-02 -6.25479519e-01 -2.32825186e-02
-4.71572638e-01 -5.15777096e-02 1.19198096e+00 8.87296200e-01
-7.41487205e-01 6.50746942e-01 -4.30413425e-01 -1.51587874e-01
5.77924967e-01 6.78466082e-01 -2.25687891e-01 2.12709814e-01
-1.60065961e+00 9.40619588e-01 5.08809090e-01 -2.89725631e-01
-1.25789309e+00 -1.57641506e+00 -9.57905352e-01 3.56142759e-01
3.76154363e-01 -1.25081468e+00 1.49561036e+00 -8.53572369e-01
-1.46773171e+00 5.87657332e-01 -4.11170274e-02 -5.01347125e-01
2.68356562e-01 -1.13951907e-01 -4.47665825e-02 -7.17996716e-01
1.03918187e-01 6.91836476e-01 4.32704240e-01 -1.06214404e+00
-7.18493819e-01 -4.90529358e-01 1.07207475e-02 -8.01020563e-02
-2.87353367e-01 2.68504173e-01 2.65877396e-01 -7.50110686e-01
-5.58906138e-01 -6.10392094e-01 -5.95074177e-01 -4.61227179e-01
-6.26515210e-01 -5.80310225e-01 1.05561160e-01 -4.37119961e-01
1.65298212e+00 -1.60681653e+00 1.10688075e-01 3.29151079e-02
4.21264261e-01 -4.80695143e-02 -5.92041872e-02 3.74264628e-01
-6.86278284e-01 3.13378006e-01 -3.85178328e-01 1.63836941e-01
2.43436709e-01 -7.91470855e-02 -2.69967645e-01 5.49798906e-01
-1.22283511e-01 8.64264548e-01 -9.33214247e-01 -6.44219458e-01
3.18020791e-01 1.59357607e-01 -1.13826501e+00 4.91435081e-01
-1.82585437e-02 1.52428418e-01 -4.60452735e-01 5.88352084e-01
7.31304824e-01 -9.17971320e-03 2.86595166e-01 -3.52451243e-02
-4.48121250e-01 4.75495636e-01 -1.14144671e+00 1.32253790e+00
-7.14367390e-01 1.45925790e-01 -1.75407767e-01 -1.22196507e+00
5.11719167e-01 6.28534794e-01 3.37101281e-01 -4.84060645e-01
2.90134162e-01 -1.95192248e-02 1.74541809e-02 -8.05618584e-01
2.79679179e-01 -4.53553945e-01 -2.39919081e-01 3.62132818e-01
7.67524615e-02 3.45945537e-01 4.14177254e-02 -1.77278906e-01
1.03259289e+00 -1.40985157e-02 1.01070845e+00 -6.76670551e-01
2.44715929e-01 -4.87089604e-02 7.61399209e-01 1.36086607e+00
-1.69270530e-01 2.52827704e-01 7.02184677e-01 -8.14246714e-01
-9.42001402e-01 -9.96295333e-01 -5.63024402e-01 1.28832245e+00
-5.89591086e-01 8.92372727e-02 -7.50088573e-01 -8.78605604e-01
3.22034240e-01 1.12705636e+00 -1.26075077e+00 -1.23724371e-01
-5.85268617e-01 -1.26564872e+00 6.51862860e-01 8.83379698e-01
2.74520606e-01 -1.48736656e+00 -4.57060784e-01 2.50411421e-01
-4.18331623e-02 -1.31474376e-01 -1.23588011e-01 3.02574426e-01
-1.07646167e+00 -9.70402241e-01 -6.62270427e-01 -4.01402444e-01
2.76696056e-01 -3.03669840e-01 1.52976537e+00 -1.66288093e-01
6.65408298e-02 -1.72265917e-01 1.30326912e-01 -8.25647712e-01
-4.28645402e-01 2.48120993e-01 5.18008769e-02 -5.86162746e-01
5.99085391e-01 -7.47128248e-01 -6.76610827e-01 -8.37893486e-02
-8.78088534e-01 4.76224124e-02 5.39904058e-01 1.18899012e+00
-2.90293787e-02 -4.00786400e-01 8.95479500e-01 -1.45918047e+00
8.58164251e-01 -8.80881906e-01 -8.60300064e-01 1.28292188e-01
-9.37167048e-01 2.74769932e-01 4.55051929e-01 -4.54623103e-01
-1.39123535e+00 -7.09486753e-02 -7.83375055e-02 1.78752795e-01
-4.43643838e-01 7.42722929e-01 -1.92074627e-01 4.25270915e-01
1.02995777e+00 -4.91172075e-01 -3.31226647e-01 -2.32663333e-01
6.06377065e-01 8.03037703e-01 2.49161810e-01 -7.91741312e-01
1.21864239e-02 2.26487115e-01 3.72495987e-02 1.39778271e-01
-8.63975704e-01 -5.98220639e-02 -1.70489714e-01 3.15510407e-02
8.36168110e-01 -6.07376873e-01 -1.38617802e+00 1.82443693e-01
-1.20167184e+00 -6.13225222e-01 -1.44495990e-03 6.90192163e-01
-5.65824449e-01 -4.10744727e-01 -6.45318449e-01 -7.21023083e-01
-5.15730381e-01 -1.27773583e+00 9.59523916e-01 1.73816547e-01
-4.96418059e-01 -1.23972428e+00 4.81266469e-01 6.06100149e-02
4.03205991e-01 2.99308360e-01 1.30165625e+00 -4.14253354e-01
-1.99001655e-01 -1.14559345e-01 -3.82185787e-01 -3.49490702e-01
6.33889288e-02 9.18273926e-02 -8.62222314e-01 1.61686420e-01
-2.13793367e-01 -9.92409587e-02 8.33540976e-01 1.66250467e+00
1.69565594e+00 -6.74486279e-01 -7.04077363e-01 5.77661872e-01
1.53745663e+00 2.58881807e-01 7.64419854e-01 3.93032253e-01
8.05609047e-01 8.41025651e-01 2.55047888e-01 4.90793973e-01
3.47905606e-01 4.25795048e-01 4.90824401e-01 -4.10578996e-01
2.62680531e-01 -3.75206262e-01 8.29234943e-02 5.71512096e-02
-1.46175086e-01 -3.15023422e-01 -1.09251189e+00 8.30141008e-01
-2.16657019e+00 -9.34781551e-01 -7.01679587e-01 2.30823851e+00
1.24348545e+00 -1.35495827e-01 1.78087324e-01 -2.25721881e-01
8.90351236e-01 -9.85084996e-02 -2.79635429e-01 -8.15005600e-01
4.02097225e-01 3.78352582e-01 1.07651448e+00 4.57356513e-01
-1.25702298e+00 7.12844551e-01 7.36384296e+00 9.46101189e-01
-7.55342782e-01 2.33274847e-01 7.14141905e-01 1.44017518e-01
-3.39747518e-01 2.05318913e-01 -6.74336791e-01 3.47676873e-01
1.10932708e+00 -2.66056597e-01 1.02701515e-01 7.84612179e-01
5.98034322e-01 -4.18528616e-02 -1.47047949e+00 2.91878492e-01
-4.61629927e-01 -1.61445367e+00 -4.76468774e-03 2.22627865e-03
9.99817848e-01 -2.44033217e-01 6.09884821e-02 5.08305073e-01
1.22153616e+00 -1.43163300e+00 1.84728146e-01 4.12614346e-01
6.42363071e-01 -7.65570104e-01 9.06812608e-01 1.19621634e-01
-3.99372905e-01 -4.10101354e-01 -2.62603819e-01 -4.92212415e-01
7.56306499e-02 9.52681780e-01 -7.04887033e-01 5.61008573e-01
5.93910456e-01 7.48346984e-01 -1.08549573e-01 1.15139687e+00
-4.79045242e-01 1.11640453e+00 7.51692383e-03 -9.61599126e-02
1.92537218e-01 2.19690904e-01 2.21312404e-01 1.32231271e+00
2.62783974e-01 1.27944753e-01 -2.15574741e-01 1.12217093e+00
-2.75265127e-01 2.02741995e-01 -1.05841696e+00 4.10754591e-01
5.70887268e-01 8.81269693e-01 -2.22404495e-01 -5.70349097e-01
-1.48705214e-01 3.84760983e-02 2.64719516e-01 2.96556741e-01
-9.96175289e-01 9.46595371e-02 2.90132821e-01 -1.43625915e-01
-2.95792341e-01 6.76764786e-01 -8.80808771e-01 -7.81293154e-01
-6.45563364e-01 -1.13878191e+00 9.07677054e-01 -7.10547268e-01
-1.46220899e+00 -3.07996571e-01 5.78423738e-01 -9.01732624e-01
-1.83906719e-01 -5.48015714e-01 -5.85425317e-01 8.17608297e-01
-1.11074460e+00 -1.15188253e+00 8.88535306e-02 3.42185140e-01
4.62286174e-01 1.67595267e-01 9.27492499e-01 3.93428653e-01
-8.67342949e-01 5.21773160e-01 -1.45990439e-02 -8.10497552e-02
1.01463282e+00 -1.54542649e+00 8.22410807e-02 4.15171146e-01
-6.74740016e-01 8.89496326e-01 7.19791412e-01 -8.80318582e-01
-1.21086764e+00 -1.05694044e+00 1.18054259e+00 -4.96692866e-01
6.73354447e-01 -6.36506677e-02 -6.52024329e-01 9.56157863e-01
5.25215447e-01 -4.91266519e-01 6.76128626e-01 1.05095255e+00
-2.16594413e-01 -6.05828427e-02 -1.07024372e+00 9.29910004e-01
1.02741992e+00 9.35069695e-02 -7.54803061e-01 4.07499641e-01
7.15210736e-01 -4.39957678e-01 -8.78294945e-01 8.65114212e-01
6.52599335e-01 -8.31980348e-01 1.17249298e+00 -1.25701630e+00
1.50813603e+00 1.61081836e-01 2.21720666e-01 -1.51333761e+00
-7.09063053e-01 -1.97382838e-01 7.03602508e-02 1.20277679e+00
5.72658062e-01 -5.05904198e-01 6.23508334e-01 8.94820929e-01
4.74665686e-02 -4.14877266e-01 -7.03176558e-01 -2.45386809e-01
6.61006391e-01 -3.62964123e-01 9.59940672e-01 1.60568464e+00
8.49181041e-02 4.82392609e-01 -7.59491742e-01 1.25156567e-01
8.30483079e-01 3.22481021e-02 6.51105404e-01 -1.21227431e+00
-2.28952598e-02 -6.47674501e-01 2.57986546e-01 -3.29567045e-01
5.43691397e-01 -8.98581386e-01 1.59435049e-01 -1.46717358e+00
7.89012372e-01 -4.03828979e-01 -3.67121965e-01 7.24078178e-01
-6.43380225e-01 -3.56496215e-01 -4.26668465e-01 -4.59727347e-01
1.28708199e-01 5.14659226e-01 9.87398624e-01 -2.00403422e-01
-2.24452436e-01 3.88945825e-03 -7.51853764e-01 8.64692986e-01
8.94203842e-01 -1.10679781e+00 -2.69720197e-01 -2.55721360e-01
2.57495135e-01 6.37517810e-01 5.59740365e-01 -3.63220841e-01
1.38577014e-01 -7.80966938e-01 2.41664812e-01 -4.05677050e-01
-6.32149220e-01 -6.74655557e-01 1.70172274e-01 6.04754090e-01
-1.09652984e+00 1.07669175e-01 1.80605993e-01 2.47973263e-01
8.85643810e-02 -5.16003489e-01 4.14037347e-01 -4.79729660e-03
-3.33411962e-01 1.05536088e-01 -6.73612595e-01 -2.07201585e-01
5.78813195e-01 4.04303432e-01 -5.18610120e-01 -1.50065765e-01
-3.83614808e-01 4.41146702e-01 -2.06060573e-01 3.39575738e-01
1.49545059e-01 -1.49623728e+00 -1.00007033e+00 -2.04629317e-01
1.03630997e-01 -4.51754272e-01 4.32159752e-01 8.86041522e-01
-1.23367250e-01 5.76453745e-01 -2.25185752e-01 -2.25793079e-01
-1.18276548e+00 1.05450130e+00 2.95861751e-01 -6.51490450e-01
-2.88215011e-01 5.38471699e-01 7.37627625e-01 -1.02006900e+00
3.08856457e-01 -3.20765436e-01 -5.29318094e-01 -9.61017907e-02
4.25619096e-01 4.77574080e-01 -1.80538297e-01 3.26997519e-01
-1.75529078e-01 -6.05412610e-02 1.41263604e-01 2.60579903e-02
1.56245697e+00 4.06917274e-01 -4.80697542e-01 2.73156017e-01
9.18621302e-01 -3.17141354e-01 -6.65288806e-01 3.35395075e-02
-5.74263632e-02 -3.09152663e-01 5.73915601e-01 -1.09384894e+00
-8.75250161e-01 5.48777401e-01 7.62741983e-01 1.38303563e-01
1.24969482e+00 -5.28800309e-01 5.68244010e-02 1.15081780e-01
-8.22110027e-02 -9.97295380e-01 -6.96269512e-01 2.23031312e-01
9.65242207e-01 -1.43590939e+00 1.15261674e-01 -3.82854730e-01
-1.53998792e-01 7.91230917e-01 5.56513250e-01 -1.79241359e-01
6.08154476e-01 4.47146684e-01 -3.00616145e-01 -2.80908883e-01
-7.71673322e-01 1.33107975e-01 3.40847820e-01 5.43344319e-01
1.10894191e+00 5.27667284e-01 -1.08965492e+00 5.13212085e-01
3.88866030e-02 7.22066462e-01 4.81052309e-01 6.59224153e-01
1.40974507e-01 -1.18929017e+00 -6.34882212e-01 6.98750973e-01
-6.47046328e-01 -6.57450795e-01 -2.96068609e-01 1.09711945e+00
3.78073826e-02 7.74940610e-01 -3.26003917e-02 9.28053558e-02
3.28261018e-01 -4.69082072e-02 3.32583010e-01 -2.63418704e-01
-1.00458741e+00 -2.23000973e-01 4.29648846e-01 -4.33906198e-01
-8.58775258e-01 -4.78947759e-01 -6.44227803e-01 -6.61191702e-01
-2.07955271e-01 -1.68894231e-02 3.99616539e-01 8.71906757e-01
5.83013333e-02 8.93073380e-01 4.04386461e-01 -6.02176905e-01
-4.97725517e-01 -1.12267983e+00 -1.72615901e-01 1.60197020e-01
2.18194097e-01 -8.23248208e-01 -1.15012698e-01 -1.19551271e-01]
|
[8.037425994873047, 5.434096336364746]
|
8f99a276-d0f8-4da4-8a8b-4772d24eb8ad
|
parallel-attention-network-with-sequence
|
2105.08481
| null |
https://arxiv.org/abs/2105.08481v1
|
https://arxiv.org/pdf/2105.08481v1.pdf
|
Parallel Attention Network with Sequence Matching for Video Grounding
|
Given a video, video grounding aims to retrieve a temporal moment that semantically corresponds to a language query. In this work, we propose a Parallel Attention Network with Sequence matching (SeqPAN) to address the challenges in this task: multi-modal representation learning, and target moment boundary prediction. We design a self-guided parallel attention module to effectively capture self-modal contexts and cross-modal attentive information between video and text. Inspired by sequence labeling tasks in natural language processing, we split the ground truth moment into begin, inside, and end regions. We then propose a sequence matching strategy to guide start/end boundary predictions using region labels. Experimental results on three datasets show that SeqPAN is superior to state-of-the-art methods. Furthermore, the effectiveness of the self-guided parallel attention module and the sequence matching module is verified.
|
['Rick Siow Mong Goh', 'Joey Tianyi Zhou', 'Liangli Zhen', 'Wei Jing', 'Aixin Sun', 'Hao Zhang']
|
2021-05-18
| null |
https://aclanthology.org/2021.findings-acl.69
|
https://aclanthology.org/2021.findings-acl.69.pdf
|
findings-acl-2021-8
|
['video-grounding']
|
['computer-vision']
|
[ 2.37912416e-01 -2.57232696e-01 -6.12079322e-01 -2.73086667e-01
-8.87167811e-01 -3.64519864e-01 7.13629901e-01 -1.57728568e-01
-4.33491945e-01 1.63798794e-01 7.84644842e-01 3.70775089e-02
1.92919046e-01 -4.83292520e-01 -7.44425833e-01 -2.06518307e-01
-5.70102669e-02 2.95877486e-01 3.73979539e-01 -1.18393280e-01
3.12452763e-01 1.23795308e-01 -1.58448029e+00 8.82210553e-01
6.67716622e-01 1.29331851e+00 6.39528513e-01 5.90373218e-01
-4.11944896e-01 1.30973852e+00 -8.50922316e-02 -1.90865800e-01
-3.28140855e-02 -7.68528938e-01 -1.21835113e+00 2.72041202e-01
4.43097591e-01 -6.17928803e-01 -9.00308847e-01 9.94819820e-01
3.31692398e-01 6.92046225e-01 3.88471991e-01 -1.37162554e+00
-7.36657917e-01 6.64980412e-01 -5.69414079e-01 6.41890585e-01
9.59332466e-01 2.42728740e-01 1.04771411e+00 -8.86801183e-01
8.78253520e-01 1.20731199e+00 3.89291734e-01 6.54233396e-01
-5.88310659e-01 -5.76400340e-01 6.17631435e-01 7.18811452e-01
-1.50654018e+00 -4.86124247e-01 9.22996163e-01 -6.73284650e-01
8.52300048e-01 -8.06933641e-02 6.20191574e-01 1.24564373e+00
7.88281783e-02 1.31916440e+00 3.99631679e-01 -1.27518728e-01
-1.01994880e-01 -4.89354014e-01 -1.11083344e-01 8.78329992e-01
-7.81034410e-01 1.66395903e-02 -7.71326303e-01 8.53355452e-02
7.60105371e-01 2.44483188e-01 -5.58933675e-01 -1.81671590e-01
-1.74164248e+00 6.04070127e-01 2.65320182e-01 5.55462122e-01
-5.51942647e-01 2.26197466e-01 7.94387281e-01 2.74355197e-03
5.29235661e-01 7.22412765e-02 -1.48347855e-01 -1.24049164e-01
-1.25048494e+00 1.44322395e-01 4.11479026e-01 1.24180257e+00
6.87528610e-01 -1.60520539e-01 -8.49566996e-01 6.32318020e-01
3.60154986e-01 3.31300110e-01 8.54705870e-01 -1.15117681e+00
6.86805964e-01 5.62482417e-01 8.38275179e-02 -1.19003594e+00
-3.21767658e-01 1.20533690e-01 -7.69765437e-01 -6.04970992e-01
1.32555321e-01 1.58605561e-01 -7.74142444e-01 1.78016007e+00
1.56374842e-01 9.37691629e-01 -3.35528590e-02 1.35318148e+00
1.10431075e+00 1.01561558e+00 3.68637800e-01 -3.36110055e-01
1.37970865e+00 -1.28424716e+00 -8.54186833e-01 -3.76068741e-01
5.45731544e-01 -4.85367030e-01 1.12841117e+00 -1.72982693e-01
-1.13479364e+00 -8.58231127e-01 -7.40802526e-01 -2.12425321e-01
-1.93428360e-02 -2.38001123e-02 2.51890570e-01 -3.70228022e-01
-9.35285866e-01 4.19480711e-01 -5.97109973e-01 -5.16706407e-01
2.32012704e-01 3.35569913e-03 -3.37792605e-01 -8.94816145e-02
-1.57127309e+00 3.46107334e-01 7.50262797e-01 4.75863460e-03
-1.05807269e+00 -7.61978686e-01 -1.24016440e+00 -4.53886986e-02
5.22698045e-01 -8.23784769e-01 1.42417955e+00 -1.25001347e+00
-1.34329402e+00 1.20646858e+00 -4.93632197e-01 -6.57496154e-01
3.86644930e-01 -2.59881735e-01 -5.87487459e-01 7.44621217e-01
4.28989291e-01 1.05169094e+00 7.64022648e-01 -9.58577991e-01
-8.38704646e-01 -1.87128693e-01 1.26223877e-01 4.03851032e-01
-3.90084833e-02 2.62068093e-01 -1.02643001e+00 -8.79159093e-01
3.81231830e-02 -5.50399840e-01 -1.75073341e-01 -2.00331956e-01
-2.97941118e-01 -3.47123384e-01 7.42389262e-01 -9.95186031e-01
1.47309065e+00 -2.05559850e+00 2.07976624e-01 -2.49659956e-01
1.26968235e-01 -7.09723309e-02 -3.91540170e-01 3.30825269e-01
-1.60560548e-01 3.40100452e-02 -1.55287674e-02 -1.91663548e-01
-5.88364489e-02 6.21304624e-02 -6.76967323e-01 5.05115092e-01
8.65632147e-02 1.17535794e+00 -1.22690809e+00 -8.77040803e-01
1.61019325e-01 2.33880550e-01 -4.95434165e-01 6.31231248e-01
-6.14040136e-01 5.81484616e-01 -4.18983459e-01 6.29647076e-01
1.20843500e-01 -4.56567764e-01 -2.80732810e-02 -4.74446774e-01
-8.95230472e-02 1.29945114e-01 -7.03690827e-01 2.45343661e+00
-2.76443392e-01 6.88640356e-01 -2.37846047e-01 -9.88338053e-01
8.41351330e-01 5.65132916e-01 8.75469983e-01 -1.10808516e+00
2.12248743e-01 -3.06739062e-01 -5.05979836e-01 -9.99093890e-01
7.85132468e-01 1.37983076e-02 -1.72906891e-01 4.91479784e-01
2.89621949e-01 4.18120086e-01 2.79921114e-01 3.64473373e-01
8.25654328e-01 5.12029648e-01 2.39920154e-01 -8.30720887e-02
8.36903512e-01 -1.10800564e-01 8.35957885e-01 5.92714965e-01
-6.36027694e-01 7.84203291e-01 4.28637445e-01 -5.44070661e-01
-1.04987848e+00 -7.66806066e-01 6.53775573e-01 1.59619677e+00
7.19359219e-01 -4.21161175e-01 -6.71629965e-01 -7.36809373e-01
-4.13057119e-01 4.57595736e-01 -7.15060771e-01 -2.05744028e-01
-7.59332716e-01 1.16970055e-01 3.32729727e-01 7.56656229e-01
7.15507090e-01 -1.38772428e+00 -5.24757028e-01 2.18549460e-01
-1.02635920e+00 -1.51497841e+00 -1.21283019e+00 -6.28856838e-01
-5.41937351e-01 -1.08618486e+00 -9.06661928e-01 -1.08847332e+00
2.43621469e-01 3.90033513e-01 1.33255637e+00 -2.77329851e-02
5.57128675e-02 6.89287663e-01 -5.52463770e-01 2.76966274e-01
-1.53955400e-01 5.73885976e-04 -1.44633770e-01 3.34582567e-01
3.99231851e-01 -3.82714391e-01 -8.71170044e-01 3.76819074e-01
-1.01799381e+00 4.94546413e-01 2.63726085e-01 6.66250288e-01
8.39330018e-01 -4.93057340e-01 5.20068169e-01 -3.98625225e-01
3.55176955e-01 -8.23902190e-01 -3.58482182e-01 4.16100025e-01
-1.27046043e-02 -5.05088419e-02 5.51411033e-01 -4.55108553e-01
-1.01250327e+00 9.12690088e-02 -6.64331913e-02 -1.13151085e+00
-2.42046311e-01 6.74756348e-01 -2.57938415e-01 4.50041890e-01
2.41528869e-01 6.15752935e-01 -9.50039923e-02 -1.96889117e-01
4.31891680e-01 4.78171915e-01 1.03169131e+00 -5.64500749e-01
2.27471814e-01 6.12700999e-01 -5.32939613e-01 -5.37099600e-01
-1.28834701e+00 -9.18584526e-01 -6.42172992e-01 -6.37850225e-01
1.24090385e+00 -1.08947957e+00 -8.44051957e-01 1.96393475e-01
-1.36359978e+00 -5.52057326e-01 -1.19468190e-01 3.20291907e-01
-1.23665190e+00 7.52229214e-01 -5.85154712e-01 -4.68490779e-01
-4.83207434e-01 -9.77318764e-01 1.30107796e+00 1.99407175e-01
-1.52579591e-01 -9.97222960e-01 1.48058355e-01 5.13601780e-01
-1.93954021e-01 2.00016245e-01 4.21386451e-01 -6.98130250e-01
-6.74336195e-01 1.10464051e-01 -4.00722712e-01 -2.90783077e-01
-2.00924233e-01 -2.45174348e-01 -5.81952095e-01 -1.60750374e-01
-1.95234388e-01 -3.81429195e-01 9.64094937e-01 2.94111907e-01
1.34342504e+00 -5.10627806e-01 -3.64064187e-01 8.34291160e-01
1.26525700e+00 3.32084864e-01 6.71002448e-01 3.74103159e-01
7.52082646e-01 5.66897631e-01 9.65863168e-01 6.57358825e-01
6.76557064e-01 6.79598987e-01 4.05510753e-01 2.24411160e-01
-7.93143287e-02 -6.26143456e-01 4.68338102e-01 7.74069488e-01
2.56624132e-01 -3.49307746e-01 -1.09591663e+00 9.13958788e-01
-2.40214992e+00 -1.62212288e+00 4.94399399e-01 1.73534441e+00
6.54151976e-01 -1.58332959e-01 2.90070385e-01 -2.96946853e-01
1.03206718e+00 5.43335199e-01 -5.44904292e-01 7.48424828e-02
-1.62263978e-02 -6.08361900e-01 1.63645610e-01 3.66174787e-01
-1.43712294e+00 1.10338390e+00 5.66133070e+00 8.48493516e-01
-1.01041341e+00 2.51809359e-01 5.77207804e-01 6.42239079e-02
-3.17640275e-01 -2.21599519e-01 -6.35963857e-01 6.82496607e-01
8.93855274e-01 -3.99390459e-01 4.43382710e-01 5.74470878e-01
4.37418282e-01 1.53422728e-01 -1.18551409e+00 1.27481115e+00
5.87640345e-01 -1.61932981e+00 2.80498713e-01 -5.04168689e-01
8.06498230e-01 1.43816888e-01 -2.16962814e-01 3.80121440e-01
2.27941293e-02 -7.44716287e-01 1.05853498e+00 9.39154506e-01
8.13739002e-01 -7.52884865e-01 5.26007652e-01 3.52825373e-01
-1.77219546e+00 -2.25078583e-01 -3.78662273e-02 1.72269955e-01
5.81181645e-01 3.57686915e-03 -4.51544255e-01 7.29618132e-01
6.56594634e-01 1.18872249e+00 -2.43803337e-01 1.10771477e+00
-3.46670635e-02 3.28028738e-01 1.18469968e-01 2.45314315e-01
6.26359046e-01 -9.97067392e-02 5.88143647e-01 1.48128402e+00
1.77016720e-01 2.03264341e-01 6.26832843e-01 8.52393866e-01
-1.01676352e-01 1.42475799e-01 -7.05923438e-01 -9.21746865e-02
3.91819358e-01 9.27269340e-01 -8.26063812e-01 -5.73310316e-01
-6.20760143e-01 1.27062345e+00 3.89369667e-01 6.91972613e-01
-1.10150957e+00 -2.26120576e-01 6.17074788e-01 8.48256126e-02
4.08263355e-01 -7.91774169e-02 2.04468325e-01 -1.40694964e+00
-1.08285435e-01 -7.20739901e-01 7.91234136e-01 -1.27032077e+00
-1.18781245e+00 6.13875091e-01 -1.48831913e-02 -1.50895190e+00
-4.44959521e-01 -1.83227167e-01 -5.82090616e-01 5.39416611e-01
-1.37939847e+00 -1.27240431e+00 -4.62397546e-01 1.00113797e+00
1.06049955e+00 -9.72012058e-02 2.65828609e-01 3.66706878e-01
-4.09080684e-01 5.22891045e-01 -3.09406072e-01 6.67220950e-01
5.40386498e-01 -6.94261134e-01 2.83721715e-01 9.96550083e-01
1.64463088e-01 3.97941232e-01 3.88841659e-01 -8.49455655e-01
-1.23863614e+00 -1.42150581e+00 8.05775583e-01 7.27520511e-02
8.57722521e-01 -4.50572446e-02 -9.96639967e-01 8.69047463e-01
3.69886100e-01 4.56130989e-02 5.44288218e-01 -3.90606731e-01
-4.84313786e-01 2.67904755e-02 -5.62788904e-01 7.53897190e-01
1.45958281e+00 -9.88670170e-01 -7.59277761e-01 3.59909892e-01
1.03788102e+00 -5.44994295e-01 -7.64274657e-01 4.05876696e-01
4.25721854e-01 -7.70851672e-01 9.98904645e-01 -8.56608272e-01
7.69509256e-01 -2.70097166e-01 -2.13502169e-01 -7.90836573e-01
-3.91345263e-01 -8.48415613e-01 -4.30633008e-01 1.29578292e+00
-8.42770115e-02 1.32801250e-01 8.15252542e-01 3.40727568e-01
-4.65249181e-01 -7.07026660e-01 -8.69711399e-01 -6.65728450e-01
-3.33913028e-01 -6.18865371e-01 4.83343422e-01 1.01943898e+00
2.79506683e-01 5.17423034e-01 -6.30077481e-01 1.08318150e-01
3.08279783e-01 6.78380549e-01 5.08415163e-01 -7.99801111e-01
-9.29898117e-03 -6.31522238e-01 -4.49220836e-01 -1.61560380e+00
9.07838941e-01 -9.12804425e-01 3.53208154e-01 -1.59516537e+00
4.51155722e-01 3.00775915e-01 -3.71070594e-01 2.94071019e-01
-1.92864835e-01 1.09072417e-01 3.92762691e-01 3.08984220e-01
-1.43530083e+00 7.40433693e-01 1.26697457e+00 -4.16901529e-01
-4.56604175e-02 -3.94247085e-01 -2.81194955e-01 7.10865617e-01
5.63375235e-01 -1.35388613e-01 -5.42437911e-01 -4.64372665e-01
1.48537487e-01 6.41675591e-01 4.11249280e-01 -1.01180005e+00
5.18696606e-01 -4.56434727e-01 9.90803391e-02 -1.10576975e+00
2.00095907e-01 -6.98221385e-01 -1.09308623e-01 1.86541572e-01
-7.41987467e-01 2.27087200e-01 3.57152298e-02 7.64711618e-01
-4.43836212e-01 -5.57055846e-02 5.25847793e-01 -2.05960378e-01
-1.61033309e+00 9.14153099e-01 -3.99180919e-01 4.99004990e-01
1.15955400e+00 -1.87236786e-01 -1.03681184e-01 -6.32412434e-01
-8.20956051e-01 6.71276808e-01 4.13965791e-01 7.78551519e-01
9.82188404e-01 -1.69695055e+00 -5.92920423e-01 -5.66288605e-02
3.37920755e-01 -1.70341432e-01 6.37831092e-01 8.23165536e-01
-4.24997300e-01 5.66437483e-01 -1.32322684e-01 -7.13124931e-01
-1.01415336e+00 1.13473403e+00 4.11433190e-01 -1.55217022e-01
-7.38335371e-01 7.34210432e-01 5.00456452e-01 -4.66347449e-02
4.09574121e-01 -2.78299034e-01 -4.70423460e-01 1.35821745e-01
8.88874292e-01 1.79301754e-01 -4.82249200e-01 -1.14617789e+00
-2.35238016e-01 6.44994676e-01 6.29322901e-02 -1.03340060e-01
8.01049888e-01 -5.87927699e-01 -1.64933335e-02 6.59677565e-01
1.35595655e+00 -6.12438738e-01 -1.56187904e+00 -5.89640617e-01
3.86962928e-02 -4.18988705e-01 -1.08637474e-01 -3.50048453e-01
-1.12769997e+00 7.79368758e-01 3.20687860e-01 -7.22793341e-02
1.23401058e+00 3.05240899e-01 1.18719172e+00 1.60061345e-01
1.68793306e-01 -1.03805971e+00 4.44945484e-01 8.17640424e-01
9.34895158e-01 -1.27432203e+00 -4.34655488e-01 -4.32137698e-02
-8.06799352e-01 8.05209041e-01 1.00020826e+00 2.72300746e-02
4.66069341e-01 -1.83598250e-01 -4.10654582e-03 -1.21657841e-01
-9.38738823e-01 -5.28073609e-01 5.62796474e-01 4.79658961e-01
3.77626717e-01 -4.33553576e-01 -1.16979286e-01 6.99915290e-01
3.21261168e-01 3.57868105e-01 1.24972962e-01 8.42789412e-01
-5.09980857e-01 -4.69284177e-01 -2.75687009e-01 1.84354797e-01
-4.39681977e-01 -9.24826637e-02 -1.79492459e-01 4.13762748e-01
-6.47940934e-02 7.45065570e-01 6.00493371e-01 -6.20008707e-01
1.20741889e-01 2.14753106e-01 2.57083505e-01 -4.21007127e-01
-4.43564296e-01 1.56132549e-01 -2.31511712e-01 -1.06155384e+00
-6.38637602e-01 -5.35112321e-01 -1.54930794e+00 -1.10447682e-01
1.77836463e-01 1.48531541e-01 -3.09676398e-02 1.10481155e+00
7.11848140e-01 5.13351679e-01 4.73729968e-01 -1.02901793e+00
6.32086694e-02 -7.79091775e-01 -2.00986192e-01 7.85702646e-01
4.01075155e-01 -6.25482678e-01 1.27957582e-01 5.59730113e-01]
|
[10.117977142333984, 0.8568747639656067]
|
5537cdae-9abc-4595-8bde-bcd7a1274144
|
saliency-based-segmentation-of-dermoscopic
|
2011.13179
| null |
https://arxiv.org/abs/2011.13179v3
|
https://arxiv.org/pdf/2011.13179v3.pdf
|
Saliency-based segmentation of dermoscopic images using color information
|
Skin lesion segmentation is one of the crucial steps for an efficient non-invasive computer-aided early diagnosis of melanoma. This paper investigates how color information, besides saliency, can be used to determine the pigmented lesion region automatically. Unlike most existing segmentation methods using only the saliency in order to discriminate against the skin lesion from the surrounding regions, we propose a novel method employing a binarization process coupled with new perceptual criteria, inspired by the human visual perception, related to the properties of saliency and color of the input image data distribution. As a means of refining the accuracy of the proposed method, the segmentation step is preceded by a pre-processing aimed at reducing the computation burden, removing artifacts, and improving contrast. We have assessed the method on two public databases, including 1497 dermoscopic images. We have also compared its performance with classical and recent saliency-based methods designed explicitly for dermoscopic images. The qualitative and quantitative evaluation indicates that the proposed method is promising since it produces an accurate skin lesion segmentation and performs satisfactorily compared to other existing saliency-based segmentation methods.
|
['Giuliana Ramella']
|
2020-11-26
| null | null | null | null |
['skin-lesion-segmentation']
|
['medical']
|
[ 9.32743549e-01 5.31378239e-02 -1.45490184e-01 -1.81662038e-01
-5.05921125e-01 -2.06913501e-01 4.71622378e-01 6.97590292e-01
-6.84826672e-01 5.53044558e-01 -6.01025075e-02 -2.74580598e-01
-2.54677206e-01 -4.27398562e-01 -1.06502399e-01 -8.95194471e-01
3.86619359e-01 5.30727841e-02 7.19251692e-01 -4.22787815e-02
7.88248420e-01 5.30759215e-01 -1.80305362e+00 3.45404632e-02
1.43458378e+00 8.66636336e-01 4.03434992e-01 6.77768648e-01
-1.94429830e-01 3.58291119e-01 -5.35791695e-01 -3.07881445e-01
-3.94887365e-02 -7.70586431e-01 -8.66278708e-01 6.26120269e-01
2.55636513e-01 1.11685336e-01 5.93762577e-01 1.41456747e+00
4.30415213e-01 -4.46214154e-02 7.15736032e-01 -7.63978660e-01
-3.13000917e-01 -1.07357122e-01 -8.20566893e-01 2.80983686e-01
3.37176859e-01 4.31335010e-02 5.74659407e-01 -4.77547348e-01
7.62555122e-01 7.40217805e-01 4.20932949e-01 4.39359874e-01
-1.00677478e+00 4.13622931e-02 -6.19520582e-02 5.08415461e-01
-1.23228788e+00 -2.61877745e-01 9.19273496e-01 -4.25766766e-01
2.56809831e-01 8.19577754e-01 8.92651260e-01 4.15832877e-01
1.36081427e-01 6.67638481e-01 1.59275758e+00 -8.66751015e-01
4.86899078e-01 4.98469949e-01 4.15843315e-02 6.49941802e-01
3.41743410e-01 -9.94985402e-02 -1.77479923e-01 -1.17490225e-01
6.11080527e-01 -2.22723871e-01 -3.08723897e-01 -5.13231575e-01
-8.87151361e-01 6.10295892e-01 6.10124230e-01 5.69400072e-01
-6.37365937e-01 -2.53896534e-01 2.16972560e-01 -4.51377749e-01
5.77793598e-01 4.41430002e-01 2.90026844e-01 1.73782215e-01
-1.33063972e+00 -2.29292914e-01 4.16597366e-01 7.61653334e-02
3.71782720e-01 -2.67395079e-01 -3.24584365e-01 8.06080282e-01
1.93129733e-01 3.17160696e-01 5.82284153e-01 -5.07914662e-01
-2.55322993e-01 9.15554881e-01 5.18408567e-02 -9.52021301e-01
-3.34242731e-01 -2.25048497e-01 -5.29238462e-01 5.08234322e-01
4.75968659e-01 7.27721676e-02 -1.39117765e+00 1.01625395e+00
6.67737603e-01 -9.66953337e-02 -7.21181855e-02 1.21210599e+00
7.56540179e-01 1.60241067e-01 4.74645525e-01 -2.80098915e-01
1.39504588e+00 -7.74591565e-01 -7.89498210e-01 2.99969725e-02
1.84394911e-01 -1.01146114e+00 9.14079249e-01 3.81009221e-01
-1.09806049e+00 -2.79854029e-01 -1.05503225e+00 1.12886965e-01
-4.15079832e-01 4.68956947e-01 4.68814969e-01 8.69568408e-01
-1.03911328e+00 2.27856830e-01 -6.00521028e-01 -8.16964805e-01
3.72176975e-01 2.54817337e-01 -1.87587216e-01 9.99634713e-02
-8.73406172e-01 1.19569123e+00 3.84516209e-01 2.74858475e-01
-1.41050637e-01 -3.74213517e-01 -6.90093279e-01 -1.48735359e-01
3.31113964e-01 -4.18882459e-01 8.98198962e-01 -1.47660506e+00
-1.44377100e+00 1.02208066e+00 -5.85837424e-01 -4.03041869e-01
5.65624595e-01 2.41701439e-01 -2.66575575e-01 8.40146244e-01
5.08510566e-04 6.13584399e-01 9.70353842e-01 -1.37321949e+00
-8.15318286e-01 -3.41395855e-01 -9.91057828e-02 5.14759779e-01
-1.82515651e-01 5.57257682e-02 -5.31892121e-01 -3.96518975e-01
3.15216333e-02 -6.65498912e-01 -3.77926528e-01 9.94340256e-02
-6.44376576e-01 -2.41990909e-02 4.89194691e-01 -8.36093485e-01
1.03220534e+00 -2.11898947e+00 -9.76119339e-02 6.53385282e-01
2.04580814e-01 5.76685429e-01 7.20215887e-02 7.61703625e-02
7.30476379e-02 4.59989309e-02 -5.65742254e-01 1.03600342e-02
-3.27038825e-01 -2.52184629e-01 4.02164012e-01 6.67123795e-01
4.83134329e-01 6.22291923e-01 -8.78812015e-01 -9.76897538e-01
6.48431659e-01 5.32981515e-01 4.27027158e-02 -2.01988474e-01
-1.03344366e-01 2.72302598e-01 -2.93208987e-01 7.84010410e-01
7.19370186e-01 -8.38711299e-03 1.14458874e-01 -1.68087229e-01
-1.71564698e-01 -2.15907499e-01 -9.46497440e-01 1.24456477e+00
-1.37153938e-01 5.85170627e-01 1.84087101e-02 -7.19996572e-01
7.79839635e-01 1.66743442e-01 5.48815489e-01 -8.14579487e-01
2.84879476e-01 3.28267545e-01 8.33169147e-02 -6.77060187e-01
5.90234458e-01 -1.94665059e-01 4.64428306e-01 8.19100216e-02
-2.97149926e-01 -2.06490800e-01 5.58577538e-01 2.66230870e-02
5.57791531e-01 1.47594018e-02 6.54827714e-01 -4.00058448e-01
7.82891154e-01 4.64293122e-01 2.78846890e-01 2.57275671e-01
-4.64017600e-01 5.48531532e-01 4.09433722e-01 5.67975864e-02
-6.61391497e-01 -9.50329483e-01 -3.19825262e-01 6.09887242e-01
6.68197691e-01 2.45743394e-01 -1.12212682e+00 -5.58250129e-01
-1.88025862e-01 5.40617168e-01 -9.14934933e-01 1.23317009e-02
-3.37167494e-02 -8.68598342e-01 -2.27202885e-02 1.09988816e-01
6.59671545e-01 -9.19023812e-01 -9.82069850e-01 -3.23775969e-02
-7.51863271e-02 -6.61160529e-01 -3.45893055e-01 -9.44750309e-02
-7.28567660e-01 -1.42071080e+00 -1.19572222e+00 -8.63338470e-01
1.21367884e+00 3.31571490e-01 6.35857761e-01 4.01656479e-01
-8.94909143e-01 9.27939117e-02 -3.77205729e-01 -4.72095430e-01
-3.39300334e-01 1.18878335e-02 -6.38118386e-01 2.49801353e-01
4.55149293e-01 8.53928626e-02 -7.58179247e-01 9.95566174e-02
-1.04194295e+00 1.88700140e-01 1.09160757e+00 5.95348418e-01
6.44388676e-01 1.66941687e-01 2.61155188e-01 -1.21238458e+00
6.63404346e-01 -2.19680160e-01 -4.44164366e-01 5.08595288e-01
-5.87690473e-01 -1.89889014e-01 2.00762525e-01 -2.41702631e-01
-1.19963503e+00 3.56467187e-01 -8.17929655e-02 1.29702374e-01
-4.46354955e-01 2.71406561e-01 1.57329187e-01 -6.26823604e-01
7.08538413e-01 3.64081562e-01 3.12539488e-01 -2.69449055e-01
1.83050901e-01 6.18048251e-01 4.94602323e-01 2.45000482e-01
4.56351399e-01 5.78636348e-01 2.13034883e-01 -9.55370367e-01
-4.01368171e-01 -8.70100319e-01 -7.12731540e-01 -5.49857199e-01
8.44365358e-01 -2.74708599e-01 -5.03613055e-01 4.62828875e-01
-8.29968989e-01 5.39706349e-02 -1.98283777e-01 2.80128002e-01
-2.48139381e-01 6.86489701e-01 -1.78963557e-01 -1.03775299e+00
-4.01844770e-01 -1.13921857e+00 9.56952631e-01 8.13702404e-01
-2.11508334e-01 -1.20415342e+00 7.62316957e-02 3.73444587e-01
5.02145946e-01 5.26339531e-01 8.57292771e-01 -3.69626820e-01
-1.62088901e-01 -2.85314292e-01 -4.78307843e-01 1.33278698e-01
4.37644124e-01 3.72739971e-01 -8.29028845e-01 5.00571392e-02
-9.59021449e-02 2.12798700e-01 9.45590317e-01 6.68268204e-01
8.62223983e-01 -8.82696882e-02 -4.24955130e-01 1.93909690e-01
1.81618512e+00 3.24174345e-01 7.15133190e-01 3.27118844e-01
3.84167075e-01 8.67464185e-01 9.53410089e-01 1.29854962e-01
2.50147939e-01 3.00409496e-01 6.47274733e-01 -9.42904770e-01
-3.70350718e-01 1.57380700e-01 -1.27763972e-01 1.83840394e-01
-2.15761364e-01 1.30550675e-02 -7.58581042e-01 9.22848582e-01
-1.45652747e+00 -6.53908849e-01 -3.39325696e-01 2.22752094e+00
8.20868731e-01 1.89818125e-02 4.59288567e-01 4.13357079e-01
1.06905556e+00 -6.10984676e-02 -4.85212684e-01 -6.46432042e-01
-8.78163502e-02 4.58105296e-01 6.50396109e-01 5.95878780e-01
-1.16913211e+00 7.60532737e-01 6.52617598e+00 9.03022408e-01
-1.53791463e+00 -1.36602908e-01 6.57712877e-01 9.58205387e-02
-2.61850089e-01 -1.71922639e-01 -2.83668816e-01 5.99474669e-01
4.78891641e-01 -5.19140474e-02 -3.68953729e-03 4.99622047e-01
2.23312005e-01 -1.14133620e+00 -4.77071404e-01 5.17441094e-01
3.47762227e-01 -1.16114104e+00 7.36539960e-02 -1.04079075e-01
7.08725810e-01 -6.20364428e-01 3.20116639e-01 -6.78365946e-01
-2.81307459e-01 -9.70232725e-01 4.31456298e-01 9.09879744e-01
6.43967152e-01 -7.15296507e-01 1.06900620e+00 -6.97009936e-02
-8.16969693e-01 2.67909110e-01 9.41404328e-02 2.09857762e-01
-8.52178335e-02 6.89576209e-01 -1.17778397e+00 4.69690800e-01
3.52762282e-01 4.07451302e-01 -1.08987451e+00 1.79849029e+00
-2.09188119e-01 3.15885305e-01 -8.46814290e-02 -5.77106655e-01
2.63902009e-01 -7.17497766e-02 5.55686176e-01 1.09953165e+00
-3.34232487e-02 -3.10690075e-01 -2.28046581e-01 8.26214612e-01
6.51513875e-01 6.13762856e-01 -1.53121874e-01 -4.98182774e-02
2.23757908e-01 1.30589950e+00 -1.34217036e+00 -1.94947660e-01
-4.53144731e-03 1.05792499e+00 -3.56633127e-01 2.84949332e-01
-5.28167844e-01 -7.60179639e-01 2.52664592e-02 2.77906209e-01
1.06388964e-01 2.45981738e-01 -5.47273815e-01 -4.36085671e-01
-6.31220778e-03 -4.90276366e-01 2.21461743e-01 -7.57939756e-01
-8.12489092e-01 6.13070011e-01 -1.84740588e-01 -1.10250604e+00
-1.01371408e-01 -5.13401031e-01 -8.15975070e-01 1.07761562e+00
-1.74153626e+00 -1.19865537e+00 -4.56310213e-01 4.32180792e-01
3.04601282e-01 3.03750098e-01 6.70915425e-01 -9.20049325e-02
-4.81078148e-01 2.59650141e-01 2.60492209e-02 -2.44037256e-01
6.05428874e-01 -1.60511088e+00 -2.54240990e-01 1.11701441e+00
-2.01581329e-01 2.26604819e-01 8.47696900e-01 -6.60632133e-01
-6.40471399e-01 -6.94266021e-01 7.96613514e-01 9.15678516e-02
4.10771310e-01 1.51781395e-01 -7.10253716e-01 -2.46385306e-01
4.31123942e-01 -3.58430326e-01 7.97352076e-01 -2.61924416e-01
3.78219426e-01 -5.46987243e-02 -1.55606604e+00 6.78891242e-01
2.82470077e-01 -2.01913789e-01 -4.99808013e-01 9.87699106e-02
8.25885236e-02 -3.18943143e-01 -6.32022023e-01 4.52210665e-01
3.53347033e-01 -1.11817408e+00 7.93934822e-01 1.23253223e-02
3.84797364e-01 -3.34528476e-01 4.57208723e-01 -1.09780324e+00
-2.43826032e-01 -2.96861142e-01 4.75554168e-01 1.00029409e+00
3.58238012e-01 -5.46610892e-01 9.46745336e-01 4.52377826e-01
2.06074134e-01 -8.52614462e-01 -8.49361777e-01 -2.07317263e-01
-4.00380880e-01 2.44885251e-01 1.95461065e-01 5.27460992e-01
8.77749547e-02 -2.54184961e-01 4.73773256e-02 2.56329924e-02
7.41599083e-01 1.71402827e-01 2.60379314e-01 -1.13627088e+00
1.55682579e-01 -7.67566860e-01 -6.94354713e-01 -1.41590863e-01
-4.02065843e-01 -4.99808133e-01 2.38674536e-01 -1.72296286e+00
2.71926671e-01 -2.59339005e-01 -4.41511124e-01 3.46844971e-01
-5.75322449e-01 6.75321400e-01 4.64236960e-02 -2.45843660e-02
-4.55282748e-01 -2.04158593e-02 1.50885773e+00 -4.49560396e-02
-3.04517299e-01 1.74330965e-01 -8.46262574e-01 7.50205696e-01
8.11358750e-01 -1.10949472e-01 -2.85680622e-01 1.99913129e-01
-3.51248533e-01 -2.73955494e-01 6.13412321e-01 -1.01515293e+00
3.75195891e-01 -2.45731533e-01 3.84472907e-01 -4.98837054e-01
1.81108788e-01 -7.39182770e-01 -1.91473469e-01 8.10106456e-01
-3.03923339e-01 -5.49447894e-01 1.94208622e-01 4.39510196e-01
-3.61903518e-01 -5.58834434e-01 1.11600411e+00 -5.85046550e-03
-1.04663694e+00 -3.45061988e-01 -6.16599917e-01 -3.14353108e-01
1.40523875e+00 -6.59037828e-01 -2.58832633e-01 -7.22763166e-02
-8.33956480e-01 -1.43090487e-01 6.77748799e-01 6.50912803e-03
6.46733582e-01 -8.66888821e-01 -4.34392631e-01 1.17347375e-01
1.77613154e-01 -3.36264253e-01 4.36348081e-01 1.36603689e+00
-9.20434773e-01 3.34017277e-01 -5.85356951e-01 -6.43480122e-01
-1.77639055e+00 4.33086544e-01 1.98149532e-01 -1.04527026e-01
-1.49520174e-01 6.60593033e-01 -2.99235970e-01 2.82207340e-01
1.37994781e-01 -4.69626427e-01 -6.43471301e-01 2.54900400e-02
2.44507506e-01 4.21853691e-01 4.09465842e-02 -7.00459898e-01
-4.72602308e-01 7.27549553e-01 2.47951923e-03 -7.32287467e-02
7.52380371e-01 -1.68234274e-01 -3.54040325e-01 3.01443517e-01
7.25329280e-01 9.07724947e-02 -8.76032948e-01 -5.93704730e-02
1.68508604e-01 -6.70640230e-01 2.24030718e-01 -1.17474389e+00
-9.07253325e-01 7.85442233e-01 9.10188437e-01 4.80231494e-01
1.66060603e+00 -2.63489485e-01 4.73877877e-01 -3.82602423e-01
1.29361764e-01 -1.34657872e+00 -2.17820287e-01 -2.73644298e-01
5.60212672e-01 -1.39113772e+00 2.08753407e-01 -9.07676876e-01
-9.34104621e-01 1.00186217e+00 3.61145020e-01 -1.83842361e-01
4.45079774e-01 -1.19678542e-01 2.58603156e-01 -2.96625886e-02
-9.45295021e-02 -7.84211457e-01 9.12810743e-01 7.39793956e-01
2.55859375e-01 4.11773287e-02 -9.86037850e-01 1.26065284e-01
2.38527045e-01 8.58017206e-02 3.92297655e-01 8.85377884e-01
-6.86486721e-01 -1.00466442e+00 -4.23493832e-01 5.21321177e-01
-5.26109576e-01 -5.66532882e-03 -8.17665756e-01 8.64762783e-01
3.59801441e-01 9.70963240e-01 -3.33600603e-02 -5.05808741e-02
-8.35432671e-03 -2.71060318e-01 4.80037928e-01 -6.25787914e-01
-6.09351873e-01 2.15980604e-01 2.15996765e-02 -4.42849994e-01
-8.83173168e-01 -5.39113402e-01 -1.14256871e+00 1.70871958e-01
-4.62305069e-01 2.31589496e-01 1.06617165e+00 9.25388396e-01
7.95101151e-02 4.97246832e-01 6.23776972e-01 -6.68396354e-01
1.18377479e-02 -6.96507454e-01 -6.26202583e-01 4.90409970e-01
3.50697964e-01 -5.59635222e-01 -2.65599042e-01 1.80116251e-01]
|
[15.612696647644043, -2.999173641204834]
|
2ffa66ff-f3dc-4d82-857c-a2417ec0c764
|
prix-lm-pretraining-for-multilingual-1
| null | null |
https://openreview.net/forum?id=y1DoH6Y75rK
|
https://openreview.net/pdf?id=y1DoH6Y75rK
|
Prix-LM: Pretraining for Multilingual Knowledge Base Construction
|
Knowledge bases (KBs) contain plenty of structured world and commonsense knowledge. As such, they often complement distributional text-based information and facilitate various downstream tasks. Since their manual construction is resource- and time-intensive, recent efforts have tried leveraging large pretrained language models (PLMs) to generate additional monolingual knowledge facts for KBs. However, such methods have not been attempted for building and enriching multilingual KBs. Besides wider application, such multilingual KBs can provide richer combined knowledge than monolingual (e.g., English) KBs. Knowledge expressed in different languages may be complementary and unequally distributed: this implies that the knowledge available in high-resource languages can be transferred to low-resource ones. To achieve this, it is crucial to represent multilingual knowledge in a shared/unified space. To this end, we propose a unified representation model, Prix-LM, for multilingual KB construction and completion. We leverage two types of knowledge, monolingual triples and cross-lingual links, extracted from existing multilingual KBs, and tune a multilingual language encoder XLM-R via a causal language modeling objective. Prix-LM integrates useful multilingual and KB-based factual knowledge into a single model. Experiments on standard entity-related tasks, such as link prediction in multiple languages, cross-lingual entity linking and bilingual lexicon induction, demonstrate its effectiveness, with gains reported over strong task-specialised baselines.
|
['Anonymous']
|
2021-11-16
| null | null | null |
acl-arr-november-2021-11
|
['cross-lingual-entity-linking']
|
['natural-language-processing']
|
[-6.20127320e-01 2.08124325e-01 -9.40706074e-01 -2.10796788e-01
-1.05943680e+00 -7.17959523e-01 6.07758045e-01 3.92465174e-01
-6.20166421e-01 1.51780760e+00 6.56196654e-01 -4.57236916e-01
1.39848337e-01 -9.05861437e-01 -1.14106572e+00 -2.02583708e-02
2.19449192e-01 6.51587844e-01 1.06323116e-01 -6.46346092e-01
-4.28300768e-01 2.96008829e-02 -8.13531339e-01 4.05694515e-01
1.23461914e+00 4.67305750e-01 2.99449772e-01 -4.01316509e-02
-5.60366094e-01 1.18043327e+00 -3.92106563e-01 -1.22316790e+00
-1.65227607e-01 -2.21790433e-01 -1.10949087e+00 -6.67589068e-01
1.14543460e-01 8.71729553e-02 -3.09464544e-01 8.94638300e-01
4.22812343e-01 -1.75514802e-01 6.22647643e-01 -9.98374224e-01
-1.13194394e+00 1.62678516e+00 -4.66761887e-01 1.85853288e-01
4.33415681e-01 -1.91088051e-01 1.33745611e+00 -9.16379094e-01
1.00633907e+00 1.33102989e+00 7.20282018e-01 2.00569764e-01
-1.20122826e+00 -9.24063027e-01 2.07256183e-01 5.63812315e-01
-1.60366488e+00 -3.75660449e-01 6.03412867e-01 -3.51296276e-01
1.33160770e+00 -1.60513893e-01 5.19882083e-01 1.37457502e+00
-1.27594054e-01 9.12082374e-01 1.03291464e+00 -6.70360744e-01
-5.56344390e-01 4.62593615e-01 1.27596781e-01 7.22959101e-01
6.11543357e-01 -2.21664310e-01 -7.26630509e-01 1.40080690e-01
5.10882020e-01 -5.86879611e-01 -3.65715116e-01 4.08577360e-02
-1.52557480e+00 7.40847170e-01 4.19034272e-01 5.50192058e-01
-4.08889830e-01 2.21121665e-02 5.77285945e-01 2.62434453e-01
4.70553219e-01 5.10643661e-01 -9.79638040e-01 1.66106164e-01
-5.52516878e-01 -1.14848921e-02 8.82354856e-01 1.29393756e+00
1.00073266e+00 -2.40327135e-01 1.25638610e-02 1.15647304e+00
2.57985950e-01 7.50589073e-01 6.28175914e-01 -3.07660222e-01
9.63331461e-01 6.18221045e-01 -6.79706633e-02 -8.60563099e-01
-3.39887589e-01 -2.93917298e-01 -7.80741692e-01 -7.79697359e-01
1.08901046e-01 -2.64221549e-01 -6.05740666e-01 2.03238177e+00
3.28940153e-01 1.18043326e-01 6.92518473e-01 5.39154649e-01
1.08089805e+00 6.20333731e-01 4.70834017e-01 -1.31390259e-01
1.50404358e+00 -8.04425657e-01 -8.85803640e-01 -2.70494103e-01
9.78910625e-01 -7.30476022e-01 1.11166537e+00 -7.12209046e-02
-7.61817276e-01 -3.17682087e-01 -9.02229965e-01 -4.45258826e-01
-9.27747190e-01 2.37524047e-01 7.96057761e-01 2.91740388e-01
-6.36887729e-01 6.79411143e-02 -4.62014794e-01 -2.22849622e-01
1.54985711e-01 -1.18345968e-01 -6.25590086e-01 -4.08219844e-01
-2.25629115e+00 1.39103699e+00 1.22136319e+00 1.35630861e-01
-5.93527615e-01 -1.03576314e+00 -1.19920945e+00 -2.92323917e-01
7.01507211e-01 -8.95319462e-01 8.75317037e-01 -5.05740821e-01
-1.13361990e+00 8.76653314e-01 4.97781225e-02 -3.58933002e-01
3.29125486e-02 -4.20312762e-01 -8.90051901e-01 -3.53718877e-01
5.52806914e-01 5.47829628e-01 1.63982317e-01 -1.29340804e+00
-7.59060919e-01 -8.80121663e-02 1.96924627e-01 3.65882128e-01
-4.67564136e-01 1.74659759e-01 -7.04168916e-01 -7.75763690e-01
-5.47841012e-01 -6.90369546e-01 4.29413840e-02 -9.44461763e-01
-7.89931774e-01 -2.41739452e-01 2.60460794e-01 -1.06680655e+00
1.46577644e+00 -1.51976430e+00 2.28296787e-01 1.70159176e-01
-1.19563036e-01 3.75201851e-01 -1.45760357e-01 4.81646538e-01
-6.16905279e-02 2.66403556e-01 4.75362204e-02 5.51117538e-03
1.60503775e-01 7.12656021e-01 -5.05653560e-01 -1.65471643e-01
4.14523751e-01 1.54836929e+00 -1.28642654e+00 -7.83048093e-01
-1.13801360e-01 4.71731216e-01 -4.07305092e-01 -8.27002823e-02
-2.68843442e-01 4.75594401e-01 -3.47192287e-01 6.40050948e-01
8.49655364e-04 -1.51930124e-01 6.24309897e-01 -6.60713494e-01
9.95973274e-02 8.05452108e-01 -9.81133103e-01 1.82011712e+00
-1.00882471e+00 2.49768272e-01 -5.47482491e-01 -7.48196006e-01
5.65374076e-01 4.94986057e-01 1.07644714e-01 -7.59450138e-01
-1.27971441e-01 5.24989843e-01 -1.53184578e-01 -5.49233913e-01
6.86323941e-01 -5.69312453e-01 -4.28809285e-01 1.99824676e-01
5.58975697e-01 -2.96115084e-03 4.94371325e-01 5.05228162e-01
7.79956400e-01 2.81036139e-01 7.76521385e-01 -4.81520221e-02
6.23313189e-01 2.16290519e-01 6.46930456e-01 2.22151175e-01
4.14157540e-01 -3.57245654e-01 2.62123495e-01 -2.60061286e-02
-8.22031558e-01 -9.46354926e-01 -2.84842193e-01 1.10427892e+00
6.49911389e-02 -7.52046287e-01 -7.69121423e-02 -9.76229608e-01
2.08561033e-01 9.06323850e-01 -4.55071926e-01 -8.33637044e-02
-7.15218246e-01 -9.63306367e-01 1.10566258e+00 6.24255717e-01
2.63935059e-01 -1.02890432e+00 2.85503477e-01 5.59283495e-01
-7.03243971e-01 -1.63875210e+00 -1.97981700e-01 2.19114229e-01
-2.53522456e-01 -1.17521441e+00 -3.58012885e-01 -6.06457591e-01
3.35937470e-01 -1.31007031e-01 1.72686422e+00 -4.73333180e-01
4.29655053e-03 1.98718429e-01 -4.98726279e-01 -3.69655013e-01
-5.48522651e-01 2.69736111e-01 3.42285097e-01 -3.68170917e-01
5.90510190e-01 -4.50227231e-01 6.98011518e-02 1.08406786e-02
-8.90927851e-01 1.33561447e-01 7.18390584e-01 9.90939617e-01
7.68387973e-01 1.30183965e-01 1.05397987e+00 -1.08139670e+00
6.97996497e-01 -7.69635379e-01 -3.70040029e-01 6.72432065e-01
-3.75197977e-01 4.16522264e-01 7.07053483e-01 -4.00321990e-01
-1.41053081e+00 -4.62007821e-01 -5.28686084e-02 -4.35162298e-02
2.09242508e-01 1.40738308e+00 -6.04519486e-01 2.94863760e-01
6.92588389e-01 3.31319943e-02 -7.12541044e-01 -4.74770963e-01
1.27292502e+00 4.75941420e-01 6.31884277e-01 -1.21263969e+00
6.93705857e-01 -9.97260734e-02 -4.26389396e-01 -4.88764703e-01
-1.33294570e+00 -4.69252229e-01 -8.77333820e-01 1.52882695e-01
7.65728951e-01 -1.43039751e+00 -1.74736619e-01 1.29954606e-01
-1.32219577e+00 -3.68830860e-01 -2.10606307e-01 5.62896371e-01
-6.58765510e-02 6.75873309e-02 -7.10150361e-01 -1.80097386e-01
-1.78606421e-01 -7.35570133e-01 8.96751404e-01 -9.84727964e-02
-1.97397962e-01 -1.53431153e+00 4.21340853e-01 4.55376446e-01
-1.28907198e-02 -8.79950821e-03 1.28045952e+00 -6.74063563e-01
-3.41110229e-01 2.84578279e-02 -3.99909079e-01 3.98629278e-01
2.74008632e-01 -2.73876935e-01 -5.54301620e-01 4.39405143e-02
-8.22279096e-01 -9.82094646e-01 7.62334943e-01 -1.26663268e-01
4.76711363e-01 -4.74808842e-01 -5.17861724e-01 3.49877030e-01
1.46593618e+00 -2.29695484e-01 3.06910962e-01 4.09557939e-01
1.17477798e+00 5.57977080e-01 4.90713030e-01 -1.56358108e-02
1.34457457e+00 6.99962854e-01 -1.91610992e-01 -1.41225487e-01
-4.15183127e-01 -6.71727836e-01 6.29786730e-01 1.58865190e+00
-1.81798831e-01 -4.33867844e-03 -1.30835128e+00 9.11063612e-01
-1.80350423e+00 -8.43536317e-01 -1.99774012e-01 1.89368582e+00
1.94619179e+00 -4.32662889e-02 -2.38299578e-01 -4.34330612e-01
4.70025390e-01 -2.04292580e-01 -3.10555726e-01 1.61989689e-01
-7.34316170e-01 1.82570249e-01 5.26448846e-01 6.07019186e-01
-9.54840958e-01 1.71947110e+00 4.92325115e+00 1.21275389e+00
-9.33458984e-01 4.75057781e-01 1.45366386e-01 1.96138084e-01
-6.84908509e-01 1.77940741e-01 -1.33364701e+00 3.26678485e-01
8.74415040e-01 -4.84180719e-01 1.30939648e-01 5.59183121e-01
-3.34928542e-01 1.10694997e-01 -1.00168788e+00 9.03026819e-01
2.67621465e-02 -1.39148831e+00 3.54120642e-01 -2.04441622e-01
9.26625490e-01 3.86907935e-01 -3.39274287e-01 1.04081368e+00
1.10551262e+00 -9.98536468e-01 6.65523708e-01 4.25171345e-01
1.06871283e+00 -7.76466668e-01 8.42751324e-01 2.54114658e-01
-1.39725518e+00 3.92706335e-01 -4.32402819e-01 4.09183174e-01
5.95613778e-01 9.63756621e-01 -8.25895488e-01 1.30202758e+00
4.86430258e-01 9.36203480e-01 -6.05670869e-01 4.19915944e-01
-9.31766331e-01 5.77217400e-01 -2.17408940e-01 2.49499351e-01
2.16078758e-01 1.50165424e-01 3.06558579e-01 1.70181549e+00
1.26523310e-02 5.97788431e-02 3.26697379e-01 7.43315220e-01
-6.18540347e-01 6.42737925e-01 -7.00778127e-01 -2.89104134e-01
6.27314925e-01 1.25243235e+00 -8.01483542e-02 -6.22326732e-01
-7.34529376e-01 8.07751417e-01 1.03391230e+00 3.88687193e-01
-8.35409939e-01 -2.32098922e-01 4.78120923e-01 -3.95989865e-01
-8.75935480e-02 -1.72370449e-01 1.56982228e-01 -1.76148868e+00
3.63908410e-02 -8.51313710e-01 6.06398046e-01 -6.10480428e-01
-1.83611715e+00 6.25107765e-01 1.60214648e-01 -7.44254351e-01
-4.56940711e-01 -6.00048006e-01 2.31861964e-01 1.13763332e+00
-2.17921638e+00 -1.92339742e+00 2.66838163e-01 7.72588253e-01
1.19678520e-01 -3.08811933e-01 8.44382405e-01 7.08318293e-01
-6.39106393e-01 7.23989248e-01 -2.40965635e-01 5.19515038e-01
1.17022049e+00 -1.34064794e+00 7.00670257e-02 8.30336690e-01
5.12711823e-01 9.93659079e-01 2.27519527e-01 -1.12513793e+00
-1.20817566e+00 -1.47255313e+00 1.38330972e+00 -7.87916601e-01
1.38790619e+00 -1.74294189e-01 -1.17018294e+00 1.11278391e+00
2.72151977e-01 -4.33552116e-02 1.17534029e+00 9.49102998e-01
-8.99685204e-01 2.13857368e-02 -3.72501135e-01 7.35836506e-01
9.76512313e-01 -9.25154865e-01 -9.45952892e-01 2.87331700e-01
8.36899579e-01 -3.36589009e-01 -1.40209734e+00 5.85850954e-01
3.17826718e-01 -2.24167421e-01 1.08162093e+00 -1.06835735e+00
6.04736567e-01 -3.55542183e-01 -2.86302477e-01 -1.82202375e+00
-1.08200334e-01 -2.17814103e-01 -2.68456608e-01 1.71333301e+00
9.03855324e-01 -5.20572603e-01 1.73618887e-02 2.13570654e-01
-1.39013454e-01 -5.65851748e-01 -7.45924771e-01 -9.89794314e-01
3.97910655e-01 -6.93577349e-01 5.75366557e-01 1.73078859e+00
5.04742801e-01 1.11528635e+00 -4.29548413e-01 3.55267406e-01
3.42981309e-01 2.55104080e-02 6.87788188e-01 -1.06949687e+00
-1.80144385e-01 -4.12118852e-01 -1.98904276e-01 -7.18607008e-01
1.00076425e+00 -1.56018150e+00 -1.72867224e-01 -1.56193411e+00
4.12000507e-01 -1.01774013e+00 -3.80242556e-01 1.18076277e+00
-7.48131573e-01 1.37681499e-01 -1.64754540e-01 1.23092070e-01
-6.43398762e-01 7.94592083e-01 1.06292105e+00 -9.66608450e-02
-1.17504410e-01 -8.89683306e-01 -8.52174461e-01 7.01100290e-01
4.90082890e-01 -3.98755610e-01 -4.00542319e-01 -6.77772403e-01
9.83230650e-01 -5.01403548e-02 1.80702254e-01 -3.55806112e-01
3.09006006e-01 -2.11609662e-01 1.56764537e-01 -2.66031355e-01
-2.66300775e-02 -5.89791596e-01 1.48193479e-01 -1.44435436e-01
-2.41698816e-01 -5.96082769e-02 4.58735108e-01 4.73890960e-01
-6.09225452e-01 -9.13591757e-02 2.50997812e-01 -1.53413624e-01
-1.00232172e+00 1.92973107e-01 2.33063832e-01 6.51437104e-01
7.49036133e-01 4.28351164e-01 -6.17450356e-01 -1.41358534e-02
-5.28433800e-01 4.48774934e-01 1.66970029e-01 6.39983714e-01
2.02084377e-01 -1.72013068e+00 -9.98788774e-01 -2.53068238e-01
6.39369369e-01 -6.31198585e-02 -1.40289739e-02 9.06556666e-01
1.73949432e-02 8.00316393e-01 -3.96460947e-03 -4.68393713e-02
-6.89725339e-01 6.87417626e-01 1.02761574e-01 -7.61652350e-01
-3.51306349e-01 7.55973101e-01 8.60930458e-02 -8.62649560e-01
-2.27282286e-01 -1.64333001e-01 -3.75382513e-01 3.01166385e-01
3.68566424e-01 -2.89618932e-02 1.41723260e-01 -8.82927716e-01
-3.73241156e-01 2.64095962e-01 -1.06351383e-01 -1.83343515e-01
1.12955368e+00 -2.84999818e-01 -5.70815206e-01 6.37762308e-01
8.34519267e-01 6.88669443e-01 -4.23126519e-01 -7.86773980e-01
4.27163392e-01 4.70698997e-02 -4.32898365e-02 -1.19823205e+00
-8.12921584e-01 5.75905561e-01 -5.03460526e-01 -3.28301847e-01
8.05768132e-01 4.01985019e-01 8.49634528e-01 5.11823535e-01
7.58315861e-01 -1.03617430e+00 -2.47909293e-01 9.89900410e-01
8.76255631e-01 -1.38651931e+00 -1.50681660e-01 -5.44203341e-01
-8.51935446e-01 6.61447346e-01 6.90462291e-01 5.34337699e-01
5.89689255e-01 3.73555869e-01 1.69907555e-01 -2.39562467e-01
-7.50168085e-01 -6.29834950e-01 6.98126912e-01 5.49091399e-01
7.89093852e-01 3.93132687e-01 -3.72280866e-01 1.17947483e+00
-4.13253933e-01 -9.96982828e-02 6.69917241e-02 4.90980595e-01
8.99431035e-02 -1.51528311e+00 -6.05863072e-02 2.40654513e-01
-5.16376555e-01 -8.64517868e-01 -1.50530845e-01 9.24435675e-01
5.31363189e-01 6.25380933e-01 -5.74716032e-01 -2.81807929e-01
2.69106567e-01 3.70331585e-01 5.41954756e-01 -7.90339530e-01
-3.35898280e-01 -3.11356187e-01 6.75909996e-01 -3.12078834e-01
-6.74956620e-01 -3.62049758e-01 -1.17910933e+00 -8.21892396e-02
-3.19321036e-01 1.55303597e-01 3.95944893e-01 1.18966305e+00
2.18892470e-01 6.59570694e-01 -9.87274572e-02 -2.15162575e-01
7.11192116e-02 -8.92708480e-01 -3.20357323e-01 3.15547228e-01
-2.32747674e-01 -7.80507803e-01 2.67189682e-01 2.85140067e-01]
|
[9.49140739440918, 8.800393104553223]
|
84468633-0e55-40d9-bd73-19dc2bdf10b2
|
gnmr-a-provable-one-line-algorithm-for-low
|
2106.12933
| null |
https://arxiv.org/abs/2106.12933v3
|
https://arxiv.org/pdf/2106.12933v3.pdf
|
GNMR: A provable one-line algorithm for low rank matrix recovery
|
Low rank matrix recovery problems, including matrix completion and matrix sensing, appear in a broad range of applications. In this work we present GNMR -- an extremely simple iterative algorithm for low rank matrix recovery, based on a Gauss-Newton linearization. On the theoretical front, we derive recovery guarantees for GNMR in both the matrix sensing and matrix completion settings. Some of these results improve upon the best currently known for other methods. A key property of GNMR is that it implicitly keeps the factor matrices approximately balanced throughout its iterations. On the empirical front, we show that for matrix completion with uniform sampling, GNMR performs better than several popular methods, especially when given very few observations close to the information limit.
|
['Boaz Nadler', 'Pini Zilber']
|
2021-06-24
| null | null | null | null |
['low-rank-matrix-completion']
|
['methodology']
|
[ 6.17582321e-01 -3.83423641e-02 -2.87140399e-01 1.61066279e-01
-1.04258525e+00 -5.98251045e-01 2.69814700e-01 -1.75094102e-02
-2.22833514e-01 7.27837861e-01 5.47704637e-01 -4.51757908e-01
-7.50689924e-01 -2.47745782e-01 -6.86478436e-01 -7.01016188e-01
-5.73950648e-01 3.23263377e-01 -3.37947100e-01 -4.83120263e-01
1.39252812e-01 3.47598165e-01 -6.42454982e-01 8.09522066e-03
6.84404612e-01 7.55574524e-01 1.34966508e-01 7.23243117e-01
6.11188829e-01 9.01828170e-01 1.56778127e-01 -2.10763171e-01
8.64461958e-01 -3.12724948e-01 -5.14512897e-01 2.21033052e-01
3.12352479e-01 -4.33316201e-01 -9.61207688e-01 1.31826723e+00
4.35045779e-01 3.41068923e-01 4.43897754e-01 -9.15987909e-01
-4.20418352e-01 1.05695927e+00 -1.09785318e+00 8.27391520e-02
8.49567890e-01 -4.73566651e-01 1.09673190e+00 -1.31557631e+00
5.24079621e-01 1.40330899e+00 1.15703678e+00 -9.38041229e-03
-1.48361254e+00 -6.21408939e-01 -9.99189690e-02 -5.19273728e-02
-1.79517519e+00 -9.11682010e-01 2.89102882e-01 -6.04190052e-01
3.50976616e-01 2.82947779e-01 1.98769137e-01 6.63017929e-01
-9.44157317e-02 8.15159380e-01 1.16874778e+00 -5.31507075e-01
8.18390341e-04 -1.69318959e-01 2.75431454e-01 4.71365482e-01
7.35135496e-01 4.37685251e-01 -6.90484345e-01 -7.48965204e-01
9.15317118e-01 -8.74811143e-04 -6.50131822e-01 -5.53538620e-01
-1.53409159e+00 9.02199507e-01 1.54566661e-01 2.35299647e-01
-6.16339386e-01 3.98033738e-01 1.07038796e-01 6.07582927e-01
1.03882581e-01 1.38804898e-01 6.98285326e-02 1.38212889e-01
-1.17373550e+00 2.87434250e-01 1.07395971e+00 1.03226781e+00
9.25900996e-01 2.27076843e-01 -2.05400407e-01 7.13034809e-01
3.70156109e-01 1.15011394e+00 -5.22114672e-02 -9.27624285e-01
6.99057758e-01 -2.87895113e-01 2.59667158e-01 -1.54705834e+00
-3.09602976e-01 -7.53430128e-01 -1.36610711e+00 -3.03539008e-01
2.82033175e-01 -1.67293817e-01 -4.56768215e-01 1.66393661e+00
2.74462283e-01 5.78229725e-01 3.54727417e-01 9.87433612e-01
2.36962184e-01 5.83765030e-01 -7.17761278e-01 -6.28533542e-01
8.73203814e-01 -5.74534833e-01 -1.03906715e+00 -4.98178393e-01
3.68452579e-01 -8.28318000e-01 4.59849387e-01 7.58704245e-01
-1.02168429e+00 -8.71431306e-02 -1.12426078e+00 1.82548314e-01
4.16709006e-01 2.86983311e-01 7.44453371e-01 7.68495023e-01
-1.22166550e+00 6.63459420e-01 -6.65322900e-01 -2.86023647e-01
-6.52571842e-02 5.50838768e-01 -6.94687784e-01 -6.73400283e-01
-8.52770507e-01 4.53585744e-01 2.51513869e-01 4.00101930e-01
-9.96251225e-01 -6.08904779e-01 -8.34865034e-01 -1.72041640e-01
5.28899193e-01 -6.49708569e-01 1.13752413e+00 -4.50306833e-01
-1.18702269e+00 4.07372922e-01 -4.67012584e-01 -6.93088233e-01
4.37640518e-01 -6.30591571e-01 -3.68193984e-01 3.08168679e-01
1.79545373e-01 -2.21461624e-01 1.14796543e+00 -1.19200885e+00
-2.90766537e-01 -3.36222768e-01 -1.06336862e-01 1.39594823e-01
6.43204059e-03 -3.07583719e-01 -5.35605788e-01 -7.19099343e-01
8.29280198e-01 -1.12148607e+00 -8.95433307e-01 -4.07373130e-01
-6.83586061e-01 5.80290675e-01 3.57937843e-01 -6.86970413e-01
1.43795288e+00 -2.29453135e+00 4.13637698e-01 8.83236110e-01
5.00401258e-01 -5.95526807e-02 -1.60602838e-01 1.10179937e+00
-3.40627432e-01 -3.29696506e-01 -3.73469949e-01 -2.44199395e-01
-8.39376450e-02 -8.12004432e-02 -7.02697635e-01 1.17315435e+00
-5.15255570e-01 5.80026984e-01 -9.51933086e-01 4.33131866e-02
1.68799147e-01 2.54436791e-01 -6.69861853e-01 -2.07466692e-01
5.03005445e-01 5.30573070e-01 -4.14072782e-01 3.75642627e-01
1.01190841e+00 -4.24958408e-01 5.14448762e-01 -4.24139947e-01
1.40229374e-01 -1.65660247e-01 -2.07447767e+00 1.79006958e+00
-1.42916277e-01 6.42253101e-01 9.38661695e-01 -1.06344366e+00
4.44647431e-01 4.69217807e-01 7.82302022e-01 -2.20889643e-01
8.52096174e-03 3.92980278e-01 -1.59158766e-01 -6.47119656e-02
7.11027622e-01 -2.09347144e-01 2.62774322e-02 6.29598677e-01
-2.84616917e-01 2.38380432e-01 3.11904162e-01 9.13461447e-01
1.31926942e+00 -5.26298285e-01 8.13147485e-01 -4.99038488e-01
5.14651656e-01 -2.16565624e-01 4.21891600e-01 1.36128235e+00
3.27905983e-01 5.10264456e-01 1.09088672e-02 6.40510470e-02
-8.17162871e-01 -1.10499823e+00 -2.37489548e-02 8.96269560e-01
1.44200251e-01 -6.95119321e-01 -2.90619791e-01 8.97701979e-02
1.45839319e-01 5.37479632e-02 -3.89338642e-01 1.23047858e-01
-3.16022277e-01 -8.79927516e-01 4.01767612e-01 9.26027596e-02
1.69436634e-01 -1.07805438e-01 3.43928248e-01 4.47398663e-01
-2.00992897e-01 -1.49724209e+00 -5.18537343e-01 4.32329178e-02
-1.08161914e+00 -1.10206187e+00 -6.38282716e-01 -3.83064449e-01
7.18366742e-01 1.01220489e+00 7.86095381e-01 8.89261588e-02
8.04185271e-02 7.43700743e-01 -3.98305118e-01 1.22920975e-01
-2.71039099e-01 -1.59917340e-01 6.43459737e-01 4.33499783e-01
-1.00456744e-01 -8.79799187e-01 -3.25607717e-01 2.95328319e-01
-1.00591481e+00 -2.07195982e-01 9.31075275e-01 7.57292330e-01
6.90436959e-01 5.11796661e-02 3.42704743e-01 -1.34952295e+00
8.00445974e-01 -6.61934257e-01 -5.46924710e-01 1.43841850e-02
-4.82470065e-01 1.10499926e-01 4.13765848e-01 -4.41352159e-01
-5.02651691e-01 4.76092368e-01 7.87085742e-02 -4.56755579e-01
6.52415752e-01 8.29895079e-01 -2.55573746e-02 -4.31021899e-01
1.08371496e+00 3.47768396e-01 8.33184868e-02 -5.40482640e-01
7.03378856e-01 4.76632446e-01 8.52368951e-01 -8.07469428e-01
1.58082390e+00 8.92950833e-01 2.85639048e-01 -1.18244624e+00
-1.00322700e+00 -1.05453193e+00 -4.40004736e-01 1.54761747e-01
3.28497440e-02 -1.40944278e+00 -5.65234303e-01 9.24120471e-02
-5.55282056e-01 -3.68651897e-01 -1.90454468e-01 9.27523077e-01
-6.83311284e-01 8.48859549e-01 -6.25563204e-01 -1.11966419e+00
-1.42636999e-01 -7.11928904e-01 8.04671109e-01 -3.25527966e-01
1.42872147e-02 -1.00922143e+00 4.04131800e-01 -1.63577236e-02
3.02284986e-01 1.56952843e-01 1.53394714e-01 -3.00712973e-01
-5.15671194e-01 -2.52101958e-01 -4.13167447e-01 7.49918967e-02
-3.61662209e-02 -7.28947759e-01 -4.82878238e-01 -1.00858915e+00
1.93583310e-01 -1.58085629e-01 9.68720019e-01 5.73939681e-01
5.77024400e-01 -5.62737226e-01 -6.48323774e-01 8.18056047e-01
1.66397250e+00 -3.44021350e-01 7.28376925e-01 -3.73854600e-02
7.73656130e-01 1.71663165e-01 7.88129926e-01 8.69333029e-01
-6.29443228e-02 5.22448361e-01 1.27788380e-01 -1.67598650e-01
1.12664253e-01 -2.32491046e-01 4.91202295e-01 9.91102219e-01
2.55380720e-01 1.10000826e-03 -5.18165112e-01 4.98254031e-01
-2.17948103e+00 -1.08810771e+00 -5.32518327e-01 2.52926898e+00
6.18068516e-01 -4.06009316e-01 1.49796411e-01 2.90983826e-01
7.97906637e-01 2.91226089e-01 -3.75579506e-01 3.01043153e-01
-4.78922069e-01 1.30717024e-01 1.11040354e+00 9.85451221e-01
-9.86244678e-01 6.26382709e-01 7.98942852e+00 8.23036551e-01
-3.99278730e-01 1.90551311e-01 -1.55996785e-01 2.35216762e-03
-3.08957845e-01 4.03665155e-01 -8.08794141e-01 -1.42538905e-01
7.05460310e-01 -2.80027449e-01 9.06424165e-01 6.71795011e-01
3.36079359e-01 -7.38501037e-03 -1.21715307e+00 1.55709815e+00
1.59020185e-01 -1.16470659e+00 -2.31441572e-01 4.68899429e-01
1.09486532e+00 1.55763716e-01 1.15276814e-01 -7.78491516e-03
6.68032765e-01 -1.01794863e+00 3.71987551e-01 4.45311099e-01
1.05144393e+00 -4.87276673e-01 3.73329759e-01 3.33254993e-01
-1.43504620e+00 -2.00834826e-01 -7.45039701e-01 -1.76890656e-01
5.40215254e-01 1.37525904e+00 -5.42748690e-01 9.52745557e-01
2.56084293e-01 1.01788282e+00 -7.06459908e-03 1.19551635e+00
-7.90802538e-02 6.33451283e-01 -7.51038790e-01 6.55583978e-01
3.98771241e-02 -6.28079772e-01 1.01357949e+00 1.33639359e+00
5.98855734e-01 3.58066767e-01 6.33395731e-01 2.71156639e-01
4.70345840e-02 1.63723737e-01 -8.50100636e-01 -7.12220594e-02
5.76014221e-01 1.11936462e+00 -2.40059480e-01 -1.74091458e-01
-2.63082087e-01 8.83399427e-01 1.64747834e-01 6.14489019e-01
-3.17200303e-01 -2.29174227e-01 5.02999067e-01 1.12867005e-01
3.35427254e-01 -6.95826471e-01 -1.15106860e-02 -1.36536753e+00
-1.13404438e-01 -1.32535136e+00 5.07670879e-01 -1.51800856e-01
-1.33673310e+00 6.69989735e-02 -2.69875880e-02 -1.26556146e+00
-3.75836700e-01 -4.42096353e-01 -1.11966990e-01 5.61994910e-01
-1.11768091e+00 -6.43397748e-01 -5.99150620e-02 8.79719615e-01
-8.23368803e-02 -1.35422185e-01 4.14846033e-01 5.02816796e-01
-4.02405590e-01 4.83825892e-01 8.78469646e-01 3.01399105e-03
5.22378385e-01 -1.08107471e+00 1.80319443e-01 1.44556928e+00
5.35452008e-01 1.14581406e+00 1.04014587e+00 -6.78412676e-01
-2.10552931e+00 -8.24444354e-01 3.45893323e-01 -1.36350915e-01
9.30433810e-01 -1.73485368e-01 -6.64247334e-01 1.13732970e+00
-1.34411529e-01 -1.89447016e-01 7.49998927e-01 4.04351413e-01
-3.71553123e-01 -1.08310804e-01 -8.42945218e-01 5.05159199e-01
1.13385177e+00 -7.49839067e-01 -4.35916185e-01 8.44649732e-01
1.38624147e-01 -6.18201017e-01 -9.32626367e-01 2.99955457e-01
5.32156110e-01 -7.51970291e-01 1.32901013e+00 -1.79155529e-01
-1.75881132e-01 -7.09529400e-01 -7.55623817e-01 -9.89921510e-01
-6.35621369e-01 -1.48363101e+00 -4.25007701e-01 6.88239813e-01
2.57717699e-01 -6.15652680e-01 7.00225711e-01 5.54955192e-02
1.34877369e-01 -2.26170838e-01 -8.08633924e-01 -1.03016973e+00
-4.46161509e-01 -6.10985041e-01 1.41037494e-01 8.63896668e-01
4.46145125e-02 5.30174851e-01 -1.26835680e+00 5.87032914e-01
1.21603560e+00 4.88472134e-02 1.07929921e+00 -1.10544991e+00
-9.48465943e-01 1.43624663e-01 -2.08987966e-01 -1.79440033e+00
-7.54405335e-02 -9.90746558e-01 9.99885798e-02 -1.44700146e+00
3.43956590e-01 -5.18346608e-01 -1.55168310e-01 4.38618250e-02
-3.14213745e-02 4.55326647e-01 2.05066472e-01 4.51274306e-01
-6.85598254e-01 2.24046797e-01 8.67446244e-01 6.83330372e-02
-2.54215598e-01 1.09297693e-01 -1.00592542e+00 5.48577964e-01
4.33860630e-01 -5.34807205e-01 -5.54919541e-01 -3.18174779e-01
7.62133896e-01 4.72547531e-01 7.69986631e-03 -1.19292092e+00
4.40230191e-01 6.65868744e-02 -5.52854016e-02 -6.99781179e-01
4.60039526e-01 -7.60853708e-01 6.62039220e-01 6.13234401e-01
-2.63486028e-01 -1.33637086e-01 -1.02538146e-01 1.18556309e+00
-9.52430815e-03 -3.37729901e-01 6.76344454e-01 6.21802323e-02
-4.82474476e-01 5.75537741e-01 -4.92206067e-01 2.20096678e-01
4.56909060e-01 -1.59528121e-01 1.27907600e-02 -1.03407741e+00
-9.53812063e-01 3.96627337e-02 3.38157028e-01 -2.09777027e-01
6.06906593e-01 -1.41448522e+00 -1.05043852e+00 2.42053103e-02
-5.95453894e-03 -4.42450225e-01 3.81888092e-01 1.25854528e+00
-2.46474102e-01 3.92285466e-01 4.10887808e-01 -5.97721875e-01
-8.99146974e-01 6.02897227e-01 6.45045042e-02 -5.06366849e-01
-7.28464127e-01 5.47461152e-01 3.71396780e-01 -3.73187631e-01
-8.92558880e-03 1.21168122e-01 2.85273403e-01 -2.79337317e-01
1.05743396e+00 6.12613499e-01 -1.42765671e-01 -7.06584334e-01
-1.58312231e-01 6.58137918e-01 1.39330905e-02 -4.65390354e-01
1.15098763e+00 -5.40454030e-01 -3.75383079e-01 3.43737215e-01
1.08990192e+00 8.33879352e-01 -9.44913208e-01 -7.80826449e-01
1.01045910e-02 -9.08178985e-01 6.98435307e-02 -1.76632538e-01
-9.57663774e-01 3.14524591e-01 3.64437222e-01 1.21373661e-01
1.02670109e+00 -2.81473547e-01 5.92663109e-01 9.87758815e-01
6.20053709e-01 -7.74982452e-01 -2.02703133e-01 6.25229836e-01
9.08240795e-01 -1.10484767e+00 6.86973333e-01 -6.29109859e-01
-1.53703153e-01 9.34722662e-01 -3.44479948e-01 -4.56970632e-01
6.24311626e-01 2.37769768e-01 -2.94803381e-01 -1.55469403e-01
-3.53868127e-01 -2.17714459e-01 2.90233910e-01 8.31538618e-01
1.95799977e-01 6.89999089e-02 -3.44353914e-01 2.81605959e-01
-3.00348341e-01 -3.06536257e-01 8.08490217e-01 6.32619917e-01
-6.19676888e-01 -1.19270396e+00 -8.56094778e-01 6.24377489e-01
-6.24901533e-01 -3.58775318e-01 -1.16586752e-01 5.72022557e-01
-7.24567533e-01 1.41467738e+00 -6.62843823e-01 -3.35899740e-01
1.19855016e-01 -6.64700508e-01 4.84070987e-01 -5.50087273e-01
-2.55978793e-01 3.39709729e-01 2.20475063e-01 -9.10833001e-01
-4.70254749e-01 -9.46280539e-01 -9.02728856e-01 -7.14651287e-01
-4.35347080e-01 4.42025065e-01 3.60738903e-01 8.76440167e-01
2.19204694e-01 -6.60740212e-02 7.00286090e-01 -7.90547907e-01
-9.49734747e-01 -7.49497235e-01 -1.22588086e+00 1.71305910e-01
5.56107759e-01 -3.61022413e-01 -4.99920070e-01 -2.79096037e-01]
|
[6.965213775634766, 4.644918441772461]
|
af27652d-ed75-47b1-9d89-9d7789776568
|
direct-and-residual-subspace-decomposition-of
|
2207.09733
| null |
https://arxiv.org/abs/2207.09733v2
|
https://arxiv.org/pdf/2207.09733v2.pdf
|
Direct and Residual Subspace Decomposition of Spatial Room Impulse Responses
|
Psychoacoustic experiments have shown that directional properties of the direct sound, salient reflections, and the late reverberation of an acoustic room response can have a distinct influence on the auditory perception of a given room. Spatial room impulse responses (SRIRs) capture those properties and thus are used for direction-dependent room acoustic analysis and virtual acoustic rendering. This work proposes a subspace method that decomposes SRIRs into a direct part, which comprises the direct sound and the salient reflections, and a residual, to facilitate enhanced analysis and rendering methods by providing individual access to these components. The proposed method is based on the generalized singular value decomposition and interprets the residual as noise that is to be separated from the other components of the reverberation. Large generalized singular values are attributed to the direct part, which is then obtained as a low-rank approximation of the SRIR. By advancing from the end of the SRIR toward the beginning while iteratively updating the residual estimate, the method adapts to spatio-temporal variations of the residual. The method is evaluated using a spatio-spectral error measure and simulated SRIRs of different rooms, microphone arrays, and ratios of direct sound to residual energy. The proposed method creates lower errors than existing approaches in all tested scenarios, including a scenario with two simultaneous reflections. A case study with measured SRIRs shows the applicability of the method under real-world acoustic conditions. A reference implementation is provided.
|
['Jens Ahrens', 'Paul Calamia', 'Sebastià V. Amengual Garí', 'Thomas Deppisch']
|
2022-07-20
| null | null | null | null |
['direction-of-arrival-estimation']
|
['audio']
|
[ 3.61135691e-01 -6.11203671e-01 1.09807277e+00 -5.38329147e-02
-8.49085331e-01 -4.68740791e-01 3.00811440e-01 -1.39441201e-02
-3.55065018e-01 2.74375826e-01 8.45500231e-01 -1.23437770e-01
-3.00621629e-01 -4.52890277e-01 -1.97059304e-01 -1.06547034e+00
-2.42132515e-01 -2.81609744e-01 2.56805867e-01 -4.00373966e-01
9.73978564e-02 3.35197419e-01 -1.70269680e+00 3.05188030e-01
5.46875298e-01 9.23511624e-01 4.29307163e-01 9.98616993e-01
-7.81164085e-03 3.52638721e-01 -8.21022034e-01 5.66069782e-01
2.52253950e-01 -3.61115485e-01 -3.62635329e-02 -3.23004574e-02
2.92184830e-01 -1.61571845e-01 -1.59807116e-01 8.97275746e-01
9.92473722e-01 9.12959397e-01 4.44304734e-01 -5.50703466e-01
-2.31434599e-01 1.58367619e-01 -2.62027621e-01 1.74248070e-01
7.14737415e-01 -2.66047537e-01 7.80319631e-01 -1.37858367e+00
-3.34152207e-02 1.29387295e+00 5.51428437e-01 2.52636820e-01
-1.17465031e+00 -4.01274681e-01 5.43872863e-02 -8.29995982e-03
-1.36993134e+00 -7.08738446e-01 1.10804963e+00 -4.41827565e-01
7.42523611e-01 8.91208053e-01 3.42917323e-01 7.07715511e-01
-6.02154993e-03 8.95332471e-02 1.29303038e+00 -7.17921436e-01
4.72407609e-01 1.76984996e-01 3.67568791e-01 2.30562791e-01
-1.97440699e-01 3.51028055e-01 -4.92430717e-01 -4.30153221e-01
4.40380543e-01 -2.07499981e-01 -9.80862379e-01 -1.60851881e-01
-1.06187975e+00 1.85431898e-01 3.54262948e-01 5.16094208e-01
-4.19657618e-01 -1.78417698e-01 5.45477308e-02 -9.98802185e-02
2.70411611e-01 1.28357172e-01 4.39752415e-02 -3.79918031e-02
-6.42357230e-01 1.85253695e-01 1.02102327e+00 4.38625962e-01
3.65211487e-01 4.47003543e-01 -1.40091211e-01 1.49120283e+00
7.29295492e-01 7.96457589e-01 1.70517281e-01 -6.48718417e-01
4.94893938e-01 -8.18052515e-02 3.05903465e-01 -1.09721053e+00
-3.08411062e-01 -6.97887838e-01 -8.92798364e-01 3.37028384e-01
3.72183383e-01 1.01709545e-01 -7.60954320e-01 1.53169990e+00
5.28040111e-01 3.41917872e-01 1.29258752e-01 1.08562458e+00
9.35548246e-01 1.09141505e+00 -3.54578882e-01 -4.40980405e-01
1.23593128e+00 -6.26871824e-01 -8.95211816e-01 -2.07418323e-01
-1.31299570e-01 -1.22105777e+00 1.03078866e+00 6.13602996e-01
-9.56882000e-01 -9.60095227e-01 -1.05492747e+00 3.87976497e-01
1.15240797e-01 -2.10658833e-02 -3.76139730e-01 8.56207609e-01
-1.07029641e+00 6.71432987e-02 -5.10055840e-01 2.52933621e-01
-6.83693826e-01 -7.49201626e-02 -6.21973202e-02 -2.29250222e-01
-9.61215556e-01 4.14622486e-01 -5.56238651e-01 8.43823910e-01
-8.06047201e-01 -7.82480419e-01 -6.98781788e-01 5.01299500e-02
1.10911854e-01 -3.39248359e-01 9.42381442e-01 -4.87994164e-01
-1.44279957e+00 -2.83173881e-02 -5.31496286e-01 2.70130754e-01
2.61758149e-01 -3.07808042e-01 -9.41865861e-01 -1.54658630e-01
-1.72667146e-01 -6.39588475e-01 1.11961174e+00 -1.94282675e+00
-1.67468205e-01 -3.58184725e-01 -4.63232785e-01 3.66381556e-01
-1.55683890e-01 -1.06219754e-01 -2.71682829e-01 -7.47187793e-01
7.37964213e-01 -8.75542641e-01 -3.96661669e-01 -5.92758060e-01
-3.12269419e-01 3.04681987e-01 3.23373407e-01 -9.66041088e-01
1.34884667e+00 -2.81710052e+00 8.57821256e-02 7.04192281e-01
1.41299203e-01 1.00473635e-01 -1.79877087e-01 5.80820262e-01
-2.71887213e-01 -5.25576591e-01 -2.66588539e-01 -4.37932014e-01
-2.14334264e-01 -1.72960192e-01 -4.32545096e-01 5.09751916e-01
-6.10669196e-01 -1.76232055e-01 -7.85399079e-01 6.07389621e-02
2.70869702e-01 9.62762356e-01 -3.39442313e-01 4.88432556e-01
6.31613970e-01 5.33453286e-01 -3.05626422e-01 1.35369942e-01
9.43782568e-01 6.80588365e-01 -1.71199724e-01 -4.25783247e-01
-5.50319254e-01 2.38542706e-01 -1.91496968e+00 9.68233764e-01
-8.92304718e-01 5.42399645e-01 8.90930116e-01 -3.30667049e-01
1.28786623e+00 4.02261406e-01 7.51210302e-02 -6.33452952e-01
-2.30619997e-01 4.56998587e-01 2.36401372e-02 -3.90608191e-01
6.05911911e-01 -2.53881335e-01 3.48785460e-01 2.16551557e-01
-3.78491789e-01 -2.03264132e-01 -3.18779856e-01 8.83809552e-02
9.81394351e-01 -2.26680845e-01 1.26983285e-01 -2.58180678e-01
1.05342603e+00 -8.22277725e-01 2.61948436e-01 5.73520720e-01
-1.73369367e-02 6.93463027e-01 -3.88852715e-01 -4.67777103e-02
-7.16707706e-01 -1.34140027e+00 -1.05797864e-01 1.16744709e+00
4.59165722e-02 -1.96880862e-01 -5.47428131e-01 2.89889246e-01
-2.89177984e-01 9.83449817e-01 -2.88049519e-01 3.30458698e-03
-8.45654786e-01 -3.65711570e-01 1.48449853e-01 2.02153668e-01
2.45411545e-01 -8.85690749e-01 -3.97356242e-01 4.22656506e-01
-5.62596440e-01 -8.86246741e-01 -5.62558293e-01 1.32257327e-01
-5.60469866e-01 -5.28472424e-01 -6.80075467e-01 -6.21496379e-01
5.98464608e-01 8.35846901e-01 6.90656066e-01 -1.15888201e-01
-3.60145748e-01 1.01440108e+00 -2.26900235e-01 -2.60221183e-01
-5.26738346e-01 -1.05990982e+00 2.92536259e-01 5.77277660e-01
-4.22066689e-01 -7.51953781e-01 -6.93009615e-01 6.12549365e-01
-5.99887967e-01 -2.64687210e-01 8.05445537e-02 5.20804942e-01
5.14992595e-01 3.92459959e-01 2.83178121e-01 -1.59546360e-01
9.34841990e-01 -9.41680670e-02 -4.89167958e-01 -1.32779926e-01
-1.46628171e-01 -1.79504186e-01 7.94900775e-01 -3.69384617e-01
-1.66777778e+00 -3.16164792e-01 -2.41092831e-01 -1.16683729e-01
-3.50933135e-01 2.41587162e-01 -3.09313029e-01 -4.16843779e-03
8.39204311e-01 4.83659714e-01 -3.43194783e-01 -7.48947918e-01
-8.02857522e-03 6.82673097e-01 5.81244528e-01 -3.85053545e-01
1.02437985e+00 3.88783813e-01 1.04623668e-01 -1.61556590e+00
-3.14319372e-01 -9.71403837e-01 -2.98688829e-01 -5.20646930e-01
5.39491057e-01 -8.32580328e-01 -4.46007639e-01 4.19580877e-01
-9.55288768e-01 -1.98274150e-01 -2.45343238e-01 1.02143776e+00
-9.86375287e-02 4.47384924e-01 -4.52855200e-01 -1.48877215e+00
-1.17686942e-01 -9.08413649e-01 5.76383471e-01 -9.35067311e-02
-2.08408415e-01 -8.75515401e-01 4.74028587e-01 4.08865899e-01
6.49541259e-01 -7.54041970e-02 8.50027919e-01 -2.87993401e-01
-3.05838227e-01 -2.27880538e-01 3.27499211e-01 5.99750102e-01
3.58340144e-01 -2.21965075e-01 -1.36002243e+00 -4.26735431e-01
6.09554529e-01 3.39919716e-01 6.12490475e-01 6.28260136e-01
5.10922968e-01 -2.17473641e-01 5.52120246e-02 3.87618214e-01
1.49889922e+00 4.34899896e-01 6.41133189e-01 5.87622263e-02
6.90520287e-01 8.36591423e-01 5.71171999e-01 5.55476069e-01
-4.72252071e-02 6.36533082e-01 4.27715957e-01 -1.95685744e-01
-3.56435478e-01 8.44115764e-02 4.78655785e-01 1.38574350e+00
-4.78689939e-01 -1.19693115e-01 -6.97537601e-01 4.26224083e-01
-1.17952871e+00 -9.07548249e-01 -4.21459705e-01 2.68653560e+00
2.78511167e-01 -1.76286027e-01 -3.21627706e-01 6.46525383e-01
4.80161756e-01 4.58192199e-01 4.71880510e-02 -5.34834325e-01
7.11665899e-02 2.48371542e-01 4.34876904e-02 1.07902789e+00
-6.00716949e-01 1.23540044e-01 5.97595119e+00 4.06468183e-01
-8.55146110e-01 -1.67229339e-01 3.32449451e-02 1.20982870e-01
-5.89104235e-01 -3.68804425e-01 -4.55455840e-01 -4.71430644e-02
8.41638148e-01 1.09499954e-01 7.97848165e-01 5.37581205e-01
6.27116740e-01 -1.74048439e-01 -8.27775717e-01 9.03944910e-01
1.31667584e-01 -4.56698060e-01 -2.55483001e-01 2.16082074e-02
3.48267257e-01 -3.73142511e-01 3.76043856e-01 -5.85520305e-02
-1.88822672e-02 -8.65586698e-01 6.87210321e-01 5.44321477e-01
3.89175326e-01 -6.55894578e-01 4.66300130e-01 2.34063298e-01
-1.65272343e+00 -2.11948708e-01 -2.72560000e-01 7.77877355e-03
1.70904890e-01 7.23842680e-01 -9.77689803e-01 5.00391126e-01
7.13624001e-01 4.64233495e-02 -1.29279733e-01 1.06572068e+00
-1.82321042e-01 9.20558989e-01 -2.63235688e-01 8.63962397e-02
-4.15264145e-02 -4.36885625e-01 1.25217581e+00 1.31328344e+00
5.41315317e-01 4.74852026e-01 7.79056773e-02 7.30582118e-01
4.97909069e-01 3.44254643e-01 -3.34557623e-01 5.79087555e-01
4.87909049e-01 1.18239880e+00 -4.04724389e-01 1.20453618e-01
-2.19678566e-01 6.50211394e-01 -4.92126137e-01 1.06023896e+00
-5.03736913e-01 -3.87961268e-01 5.53299487e-01 1.12193994e-01
2.63193995e-01 -5.03548980e-01 -7.09003285e-02 -4.58729327e-01
1.86766982e-01 -8.99284720e-01 -3.80468694e-03 -7.71485329e-01
-7.60254264e-01 9.24087048e-01 -1.28095374e-01 -1.49582112e+00
-1.53872430e-01 -2.88188308e-01 -7.09585607e-01 1.40303779e+00
-1.09879208e+00 -6.24757349e-01 -4.92556810e-01 9.56631124e-01
5.21711290e-01 3.12730521e-02 1.07075346e+00 2.58985549e-01
-2.58260995e-01 3.24017644e-01 3.91380757e-01 -1.68442696e-01
5.63259602e-01 -1.05771446e+00 -5.68167642e-02 9.56777394e-01
-1.16268210e-02 8.34233761e-01 1.24282873e+00 -3.13007087e-01
-1.39121592e+00 -6.42538190e-01 5.48016727e-01 -7.97247738e-02
3.82432371e-01 -5.22469819e-01 -1.09591722e+00 3.75828519e-02
-2.21269205e-01 3.47315259e-02 1.01781392e+00 1.06496163e-01
-4.62750256e-01 -4.85070735e-01 -9.16548908e-01 5.42691410e-01
6.38783753e-01 -6.34629846e-01 -5.96554935e-01 -1.96286276e-01
4.54593480e-01 -1.61529273e-01 -4.76832569e-01 9.28856283e-02
5.96656144e-01 -1.18083775e+00 1.37173641e+00 3.33664924e-01
-5.72458096e-02 -7.10837722e-01 -7.06861079e-01 -1.65081418e+00
-5.93389571e-01 -4.90346044e-01 1.53158233e-01 1.26227164e+00
2.38330930e-01 -7.16866076e-01 2.69953683e-02 4.71740603e-01
-4.14929092e-01 -1.24258988e-01 -1.07517290e+00 -6.31526351e-01
-5.07002592e-01 -6.76660717e-01 1.94587544e-01 4.92811710e-01
-3.02922785e-01 2.79615134e-01 -3.52134883e-01 8.06700349e-01
1.04482591e+00 5.70589043e-02 6.87790513e-01 -1.05024135e+00
-5.89032650e-01 2.91063115e-02 -8.55999365e-02 -1.09606254e+00
-3.11506152e-01 -3.71483564e-01 6.73461676e-01 -1.63872731e+00
-3.30760121e-01 -5.09550393e-01 -5.19007564e-01 -2.65363008e-01
-1.58483252e-01 -5.14069684e-02 1.96230859e-01 -1.88894023e-03
2.92628676e-01 5.26167452e-01 1.18601823e+00 2.92525953e-03
-7.36856818e-01 5.37932932e-01 -2.50187010e-01 8.11164260e-01
3.77321541e-01 -1.74597576e-01 -4.99633133e-01 -1.87351555e-01
-7.31388032e-02 3.31529588e-01 4.52243537e-01 -1.34975958e+00
3.15844983e-01 8.82133394e-02 2.64715254e-01 -6.51306272e-01
7.85633266e-01 -1.10044181e+00 3.79195839e-01 2.58635312e-01
-2.78077304e-01 -3.77173007e-01 3.02640319e-01 8.84657025e-01
-2.74418294e-01 5.57747856e-02 7.60664761e-01 6.81935102e-02
-3.25354487e-01 -3.69058013e-01 -6.54562294e-01 -3.20602298e-01
3.91013175e-01 -3.12483549e-01 1.67947844e-01 -7.10424066e-01
-8.68215680e-01 -4.94810820e-01 -2.26692498e-01 2.23630756e-01
9.92678583e-01 -1.09318125e+00 -8.67139876e-01 4.17582452e-01
-1.91745207e-01 -2.78372347e-01 8.29507291e-01 6.79024994e-01
-1.06041566e-01 1.77354530e-01 2.10104346e-01 -6.16739988e-01
-1.77998078e+00 4.39472735e-01 4.50897634e-01 -5.80623969e-02
-5.48691034e-01 1.01307368e+00 7.63697505e-01 -3.11392277e-01
4.37222779e-01 -4.36651945e-01 -6.19521677e-01 -8.23456645e-02
8.01057100e-01 8.68462324e-01 1.65085137e-01 -1.01924813e+00
-4.43806261e-01 8.86486650e-01 5.53190112e-01 -6.75763607e-01
1.23118544e+00 -5.21486759e-01 -3.20533991e-01 8.98534536e-01
1.29061306e+00 1.02669179e+00 -8.55009079e-01 -3.90242338e-01
-4.20406133e-01 -7.67617404e-01 2.41167992e-01 -7.94658124e-01
-5.59912860e-01 8.89161229e-01 9.14991736e-01 1.97935805e-01
1.50588059e+00 -3.13547671e-01 3.84752899e-01 1.02429949e-01
5.29867113e-01 -7.82441735e-01 3.31950933e-01 4.43342417e-01
1.24037623e+00 -5.14666796e-01 -7.25041404e-02 -6.47584498e-01
-3.32116991e-01 1.08483183e+00 1.23169854e-01 1.35554358e-01
8.87377501e-01 3.44318151e-01 3.91840070e-01 2.62117445e-01
-3.58565301e-01 1.36275902e-01 6.42687261e-01 6.15783036e-01
5.31084657e-01 2.04634711e-01 6.87915906e-02 5.84392667e-01
-3.05579215e-01 -7.31850505e-01 4.10898566e-01 6.78233325e-01
-7.02141166e-01 -6.64592981e-01 -1.36535990e+00 -1.28809765e-01
-3.07445705e-01 -3.19391727e-01 5.02144359e-02 8.84043276e-02
-1.42136827e-01 1.46128356e+00 -2.52739191e-01 -4.31607336e-01
9.52563167e-01 3.31166126e-02 1.68703124e-01 -3.25041860e-01
-5.22534311e-01 8.51302624e-01 1.52449802e-01 -4.82236743e-01
-1.39906198e-01 -5.71658075e-01 -1.14184117e+00 6.00267239e-02
-2.45738298e-01 4.20227706e-01 8.85271013e-01 4.32218760e-01
-1.20312743e-01 1.06809545e+00 1.05856478e+00 -1.01497483e+00
-5.17555416e-01 -8.22555065e-01 -9.20702696e-01 2.51626641e-01
9.47299242e-01 -2.27524444e-01 -9.90124285e-01 -1.22418538e-01]
|
[15.139379501342773, 5.747859001159668]
|
b68e30fb-f085-4650-aeb2-7ed66cea745b
|
network-comparison-study-of-deep-activation
|
2202.03695
| null |
https://arxiv.org/abs/2202.03695v1
|
https://arxiv.org/pdf/2202.03695v1.pdf
|
Network Comparison Study of Deep Activation Feature Discriminability with Novel Objects
|
Feature extraction has always been a critical component of the computer vision field. More recently, state-of-the-art computer visions algorithms have incorporated Deep Neural Networks (DNN) in feature extracting roles, creating Deep Convolutional Activation Features (DeCAF). The transferability of DNN knowledge domains has enabled the wide use of pretrained DNN feature extraction for applications with novel object classes, especially those with limited training data. This study analyzes the general discriminability of novel object visual appearances encoded into the DeCAF space of six of the leading visual recognition DNN architectures. The results of this study characterize the Mahalanobis distances and cosine similarities between DeCAF object manifolds across two visual object tracking benchmark data sets. The backgrounds surrounding each object are also included as an object classes in the manifold analysis, providing a wider range of novel classes. This study found that different network architectures led to different network feature focuses that must to be considered in the network selection process. These results are generated from the VOT2015 and UAV123 benchmark data sets; however, the proposed methods can be applied to efficiently compare estimated network performance characteristics for any labeled visual data set.
|
['Alper Yilmaz', 'Michael Karnes']
|
2022-02-08
| null | null | null | null |
['visual-object-tracking']
|
['computer-vision']
|
[ 8.80279690e-02 -3.13221127e-01 -1.03499912e-01 -4.78179723e-01
1.72718287e-01 -8.22999716e-01 9.41630840e-01 -1.09310150e-01
-5.53752005e-01 5.10415614e-01 -1.48566544e-01 8.73760879e-02
-7.43165672e-01 -6.26982152e-01 -4.22564775e-01 -7.64134049e-01
-2.56362826e-01 2.00108096e-01 2.72140771e-01 -4.16714549e-02
1.53786615e-01 1.17880332e+00 -2.13862777e+00 3.59931231e-01
4.77011502e-01 1.38731194e+00 1.98501945e-01 6.01858854e-01
-1.25802130e-01 5.23505628e-01 -6.92560852e-01 -2.09546402e-01
7.34401405e-01 -1.29997373e-01 -6.75931871e-01 1.10464416e-01
1.00156772e+00 -1.52459309e-01 -5.78090429e-01 1.22362578e+00
3.53046417e-01 3.80532920e-01 9.98061717e-01 -1.52837145e+00
-8.96355629e-01 3.12747866e-01 1.03330694e-01 6.23591900e-01
-1.74016982e-01 3.54578465e-01 9.12190616e-01 -1.03943181e+00
8.55293393e-01 1.28503191e+00 4.54353929e-01 4.52627659e-01
-1.07400608e+00 -5.74259758e-01 3.24661285e-02 7.75492370e-01
-1.30528486e+00 -2.40802869e-01 8.68218303e-01 -7.20313191e-01
1.02060449e+00 1.78112745e-01 8.58069122e-01 1.13274682e+00
1.14707060e-01 6.06069982e-01 8.94447386e-01 -1.84801698e-01
8.54725316e-02 3.32663953e-01 3.67879570e-01 4.43866253e-01
6.04720533e-01 5.77018559e-01 -2.93866873e-01 1.22626752e-01
5.77503562e-01 2.41278097e-01 -3.42512667e-01 -5.96321702e-01
-1.19468856e+00 9.87659574e-01 8.06851923e-01 4.73023504e-01
-3.43511075e-01 -1.92453954e-02 6.31244540e-01 5.10756493e-01
1.69875175e-01 4.18701112e-01 -4.40624684e-01 5.33137731e-02
-5.51738620e-01 8.59402046e-02 6.52848959e-01 9.82255340e-01
7.97180057e-01 4.99400824e-01 -3.23809475e-01 6.32664859e-01
2.71621406e-01 4.35404360e-01 5.02241671e-01 -8.99447441e-01
1.32311791e-01 1.02064180e+00 -3.71914685e-01 -1.23138154e+00
-4.92876887e-01 -5.63324869e-01 -8.45642626e-01 6.48416400e-01
5.13663292e-01 -8.74850899e-02 -1.16046190e+00 1.51282156e+00
3.47755671e-01 -1.90504640e-01 2.49229640e-01 9.96725738e-01
1.09039140e+00 4.56967026e-01 -1.24887362e-01 1.94359839e-01
1.18453181e+00 -6.94793999e-01 -4.72872436e-01 1.56804219e-01
5.23870707e-01 -7.66321599e-01 5.20242810e-01 3.78615856e-01
-4.58053172e-01 -9.51850414e-01 -1.11207306e+00 1.69266611e-01
-9.24169540e-01 2.70776093e-01 6.90324962e-01 4.56940234e-01
-1.08278465e+00 7.81692266e-01 -5.37198246e-01 -7.14062691e-01
7.02263832e-01 6.00411713e-01 -5.16606569e-01 -1.19928733e-01
-9.60066199e-01 1.22758842e+00 9.05561507e-01 2.77500123e-01
-1.37481999e+00 -5.63594878e-01 -7.42656827e-01 -3.17247510e-02
1.68601856e-01 -4.51795876e-01 7.10945070e-01 -1.60018337e+00
-1.08916748e+00 7.27146268e-01 5.51628828e-01 -4.73545969e-01
2.60042310e-01 4.36009429e-02 -6.74109578e-01 1.73041508e-01
-2.62957126e-01 8.92399967e-01 1.00143051e+00 -1.13223839e+00
-7.72622943e-01 -2.67036349e-01 1.07213564e-01 2.61164438e-02
-5.45449555e-01 -1.21603325e-01 2.78295606e-01 -6.01122618e-01
-1.83647200e-01 -9.32473481e-01 1.37484983e-01 4.89226371e-01
-2.22251549e-01 -5.16249001e-01 1.40080249e+00 -3.22990537e-01
7.66979098e-01 -2.33007026e+00 3.25116903e-01 2.75333494e-01
3.73013318e-01 7.03114808e-01 -4.47350472e-01 2.64460027e-01
-2.16376245e-01 -1.18625015e-01 5.14923483e-02 2.02592865e-01
-4.12327610e-02 1.45073205e-01 7.67242769e-03 6.44809544e-01
4.51468050e-01 7.96088994e-01 -7.39984095e-01 -2.04693109e-01
4.83587146e-01 6.23953819e-01 -7.38150626e-02 8.70619416e-02
-1.16878413e-01 7.98804685e-02 -3.50783527e-01 7.72117674e-01
5.82127213e-01 1.13679655e-01 -2.44124100e-01 -6.51891530e-01
-3.10967024e-02 -4.15215433e-01 -1.02528667e+00 1.54439974e+00
6.94397017e-02 1.34386075e+00 -1.77353784e-01 -1.09907854e+00
1.16979122e+00 7.82070309e-03 4.13676411e-01 -4.86407548e-01
4.63480234e-01 8.42466652e-02 6.66173816e-01 -6.26250505e-01
3.65296662e-01 1.67574599e-01 5.48110962e-01 -1.19857565e-01
7.93360770e-01 2.96515882e-01 3.98225367e-01 -2.44912639e-01
9.34493959e-01 -4.31319177e-02 2.18498975e-01 -6.00818753e-01
6.03067636e-01 3.08096796e-01 4.69804198e-01 5.63891053e-01
-5.52578866e-01 4.23650324e-01 1.07955918e-01 -9.72920656e-01
-8.82289946e-01 -1.28542781e+00 -4.80894595e-01 8.07523787e-01
-1.19582629e-02 -2.26534773e-02 -4.31226224e-01 -9.06709969e-01
1.21998675e-01 4.94873464e-01 -8.54423463e-01 -4.90263045e-01
-2.60623097e-01 -5.58619201e-01 7.24121332e-01 5.12389064e-01
5.64226985e-01 -1.35038602e+00 -9.00904179e-01 -3.73063907e-02
3.80306721e-01 -1.11655891e+00 -3.46769728e-02 4.36239898e-01
-9.48715687e-01 -1.56092000e+00 -6.68346167e-01 -1.01782274e+00
6.32101119e-01 2.63752013e-01 8.65585506e-01 -8.92915204e-02
-8.78020048e-01 8.53268027e-01 -5.35487831e-01 -5.26078582e-01
-3.55946600e-01 -5.34469821e-02 2.81823635e-01 1.64454013e-01
7.91165471e-01 -3.70788515e-01 -5.81552744e-01 2.51545519e-01
-9.51176941e-01 -5.27890742e-01 6.56274259e-01 8.12771857e-01
4.05182272e-01 5.55945933e-02 3.84678930e-01 -1.99026585e-01
4.26261216e-01 -4.60121930e-01 -7.49596596e-01 2.18345225e-01
-3.68728757e-01 1.16977088e-01 6.56256557e-01 -8.26302290e-01
-7.17463970e-01 4.75471355e-02 2.95656055e-01 -7.08135605e-01
-5.85646391e-01 2.97500581e-01 -1.51125252e-01 -4.65837449e-01
9.76098657e-01 1.60116121e-01 1.83926269e-01 -2.65870214e-01
3.93582106e-01 4.00473803e-01 2.81329542e-01 -1.43995211e-01
8.14136982e-01 4.55259353e-01 1.90726444e-01 -1.23909318e+00
-5.30257523e-01 -3.10408920e-01 -9.96777475e-01 -4.71428335e-01
1.14645100e+00 -6.50145948e-01 -5.58637559e-01 5.15102148e-01
-1.18895936e+00 7.14467913e-02 -6.53010249e-01 6.70888901e-01
-3.28334242e-01 1.82265162e-01 -8.06660503e-02 -5.91098189e-01
-3.04907858e-01 -1.20340908e+00 5.21723032e-01 4.58694667e-01
-9.96274576e-02 -1.11164427e+00 3.43093798e-02 -8.68343189e-02
4.85324621e-01 5.06561100e-01 9.16618526e-01 -1.10002398e+00
-5.29927909e-01 -7.38081187e-02 -3.98407817e-01 8.08704197e-01
4.36093658e-01 2.70093858e-01 -1.24906933e+00 -4.82773632e-01
-1.62330925e-01 -2.09649384e-01 1.08327723e+00 3.48090589e-01
8.72610867e-01 -1.89634860e-01 -2.61954218e-01 8.28237534e-01
1.44556797e+00 3.64175171e-01 5.16197503e-01 5.58666766e-01
8.94776106e-01 6.93552911e-01 3.85890931e-01 2.12650925e-01
-5.09621203e-02 4.44073141e-01 9.95645702e-01 4.01421301e-02
-3.34007472e-01 3.26475471e-01 5.55064917e-01 5.14304042e-01
-2.20216826e-01 -1.68771893e-01 -7.92218447e-01 6.43762946e-01
-1.69610465e+00 -9.11651492e-01 -1.83821116e-02 1.90034509e+00
1.80462182e-01 -5.75531460e-02 5.30748703e-02 6.15849644e-02
7.07080364e-01 -5.56291267e-02 -7.42349207e-01 -3.07843179e-01
-4.25188333e-01 6.93227500e-02 3.71044934e-01 -2.09312290e-01
-1.39426327e+00 7.45769322e-01 5.48505020e+00 7.73851573e-01
-1.32478666e+00 -1.87616393e-01 1.14024892e-01 -7.12712780e-02
2.69417346e-01 -2.81745613e-01 -7.54994988e-01 1.56405002e-01
9.69290912e-01 -5.30062318e-02 2.02563420e-01 1.17297029e+00
-2.38906264e-01 1.72473222e-01 -1.38220656e+00 1.24395466e+00
2.74385691e-01 -1.41114366e+00 6.04718924e-01 8.39493349e-02
7.41879761e-01 3.73789757e-01 2.42274135e-01 1.97708428e-01
1.05080307e-01 -1.14906633e+00 4.87507433e-01 5.62200189e-01
5.85221291e-01 -7.67726958e-01 8.57045233e-01 -9.07238796e-02
-1.14004290e+00 -4.51700449e-01 -8.28095257e-01 8.49547908e-02
-2.74088740e-01 2.91565418e-01 -8.64811540e-01 6.17906034e-01
1.02875769e+00 1.11148846e+00 -1.00237739e+00 1.34692597e+00
1.33899778e-01 3.35981041e-01 -2.32791722e-01 -2.31441766e-01
4.40775871e-01 -1.73243925e-01 7.75071859e-01 1.14166486e+00
3.50434661e-01 -3.89094502e-01 3.24052870e-02 1.06393075e+00
-1.34541560e-02 -1.63830426e-02 -1.02322137e+00 -2.95350730e-01
1.68391481e-01 1.59609878e+00 -8.26455116e-01 -1.49731264e-01
-4.49447483e-01 7.66141891e-01 2.62000829e-01 4.90738302e-01
-6.37541473e-01 -5.31069756e-01 1.18930626e+00 -1.08977199e-01
6.06744230e-01 -2.38757297e-01 1.15803726e-01 -8.02474201e-01
-1.72068834e-01 -7.81399071e-01 3.75407368e-01 -4.85694826e-01
-1.66662097e+00 9.25460994e-01 1.95005566e-01 -1.57115710e+00
6.95712715e-02 -1.35796022e+00 -5.66904128e-01 5.85762739e-01
-1.33554101e+00 -1.09480166e+00 -5.66663027e-01 7.88122594e-01
5.78435302e-01 -8.59477103e-01 7.34282553e-01 2.98869938e-01
-2.94978738e-01 3.80952090e-01 2.89543599e-01 4.40009981e-01
4.69294369e-01 -9.66933429e-01 1.98914334e-02 8.26176286e-01
4.07763302e-01 4.63982046e-01 3.17412525e-01 -3.12551707e-01
-1.38107681e+00 -1.44296134e+00 2.94866651e-01 -5.26315868e-01
4.35514539e-01 -2.33038098e-01 -8.38048160e-01 4.59906191e-01
4.43030864e-01 3.11894149e-01 5.58118761e-01 -3.99065197e-01
-2.53951967e-01 -2.59164214e-01 -1.14353251e+00 4.13871884e-01
1.04141271e+00 -4.21719909e-01 -6.53884828e-01 -2.81233452e-02
5.01497328e-01 7.17119426e-02 -7.86958337e-01 6.91356540e-01
4.40495670e-01 -9.02005851e-01 1.00232184e+00 -9.37604487e-01
1.03645518e-01 -5.60468495e-01 -3.00430685e-01 -1.42183089e+00
-5.18792033e-01 -5.78466011e-03 -2.80038774e-01 1.10476279e+00
2.50122726e-01 -4.99237418e-01 4.81972694e-01 1.42936721e-01
-2.31503636e-01 -5.64412832e-01 -1.00918031e+00 -1.02012944e+00
-2.62268092e-02 -2.80725777e-01 2.69924551e-01 9.68873858e-01
-7.02269435e-01 2.86812127e-01 3.95205580e-02 3.02249449e-03
6.78427756e-01 -1.12503834e-01 5.73301017e-01 -1.79204059e+00
2.47218505e-01 -7.14148521e-01 -1.32710266e+00 -4.84007478e-01
2.53086746e-01 -1.39107454e+00 -3.36151361e-01 -1.36321104e+00
4.25196178e-02 -1.50164872e-01 -5.05370259e-01 4.12803978e-01
3.42281222e-01 1.92287728e-01 4.41893101e-01 -2.73244977e-02
-3.83122802e-01 8.49626005e-01 1.23546469e+00 -5.96512496e-01
-4.08229381e-02 -1.42496750e-01 -2.77509958e-01 8.40942323e-01
8.27089608e-01 -4.50466901e-01 -6.12073064e-01 -3.99658412e-01
-1.42552137e-01 -8.15224528e-01 7.19005466e-01 -1.36105633e+00
8.71144682e-02 -3.83045785e-02 9.29135203e-01 -6.08078718e-01
2.66611218e-01 -1.38681161e+00 2.38352522e-01 5.49164593e-01
-4.95417491e-02 1.91442534e-01 5.46117067e-01 5.83193183e-01
-3.55130643e-01 -2.46985480e-01 9.12554860e-01 -7.25430846e-02
-1.37641919e+00 5.37665367e-01 -4.23809737e-01 5.38755879e-02
1.38302851e+00 -6.91481769e-01 -3.59424025e-01 7.51511231e-02
-6.95312977e-01 -3.67361046e-02 1.65167078e-01 9.04507518e-01
9.24901307e-01 -1.55145836e+00 -6.17706180e-01 2.08499104e-01
2.79551655e-01 -7.75005445e-02 1.96127936e-01 6.42608583e-01
-5.97251832e-01 3.83951217e-01 -8.84291649e-01 -1.00413215e+00
-1.28872371e+00 8.12896490e-01 5.79236031e-01 2.16868505e-01
-5.77639699e-01 7.17942476e-01 2.06646636e-01 -2.85232276e-01
3.63728106e-01 -5.06296813e-01 -5.10330975e-01 4.02046084e-01
4.07707989e-01 6.57896101e-01 -2.30935700e-02 -8.85050714e-01
-4.74522322e-01 5.82942426e-01 -7.36495480e-02 4.57016021e-01
1.39333725e+00 9.28247422e-02 -4.12325375e-02 4.08440560e-01
1.40127432e+00 -8.29903901e-01 -1.26264238e+00 -3.27241182e-01
2.97721386e-01 -2.98364311e-01 1.73768345e-02 -7.15413570e-01
-1.40878928e+00 8.48698139e-01 1.36807978e+00 8.36163387e-02
1.04837608e+00 -9.82418805e-02 1.78446025e-01 7.90424645e-01
2.18861341e-01 -1.18948281e+00 1.59822419e-01 6.79727435e-01
1.03163195e+00 -1.40327489e+00 -8.89397413e-02 2.87248543e-03
-5.51865280e-01 1.65189648e+00 8.24091673e-01 -2.34500319e-01
8.06942940e-01 -2.04216748e-01 1.49069920e-01 -5.00540316e-01
-3.26165736e-01 -4.52800393e-01 8.32457006e-01 9.74322200e-01
1.17531240e-01 -9.56521779e-02 3.88984196e-02 4.09771711e-01
-1.61509011e-02 -3.60226780e-01 4.37168211e-01 8.25716674e-01
-3.88177723e-01 -7.95106053e-01 -2.09807247e-01 6.71866000e-01
-1.64841071e-01 5.92608750e-03 -6.14313245e-01 1.15055966e+00
4.30636674e-01 7.08163381e-01 6.06482960e-02 -6.35229707e-01
4.00907397e-01 5.77713922e-02 4.90266204e-01 -5.39637029e-01
-7.73597538e-01 -6.37702644e-01 -6.94007874e-02 -5.01854002e-01
-6.84010029e-01 -5.09410203e-01 -1.05715835e+00 -2.57101227e-02
-4.07212406e-01 -2.33220026e-01 6.51279271e-01 8.41914237e-01
3.74357700e-01 6.68684661e-01 3.33415091e-01 -1.04544985e+00
-4.54012930e-01 -1.07340300e+00 -4.58883673e-01 6.55634046e-01
5.11761248e-01 -1.08064127e+00 -4.75462943e-01 2.19828695e-01]
|
[9.51385498046875, 2.3220107555389404]
|
d28a9ba7-eab0-4e14-83c8-83fe8ca93030
|
fast-fourier-color-constancy
|
1611.07596
| null |
https://arxiv.org/abs/1611.07596v3
|
https://arxiv.org/pdf/1611.07596v3.pdf
|
Fast Fourier Color Constancy
|
We present Fast Fourier Color Constancy (FFCC), a color constancy algorithm which solves illuminant estimation by reducing it to a spatial localization task on a torus. By operating in the frequency domain, FFCC produces lower error rates than the previous state-of-the-art by 13-20% while being 250-3000 times faster. This unconventional approach introduces challenges regarding aliasing, directional statistics, and preconditioning, which we address. By producing a complete posterior distribution over illuminants instead of a single illuminant estimate, FFCC enables better training techniques, an effective temporal smoothing technique, and richer methods for error analysis. Our implementation of FFCC runs at ~700 frames per second on a mobile device, allowing it to be used as an accurate, real-time, temporally-coherent automatic white balance algorithm.
|
['Yun-Ta Tsai', 'Jonathan T. Barron']
|
2016-11-23
|
fast-fourier-color-constancy-1
|
http://openaccess.thecvf.com/content_cvpr_2017/html/Barron_Fast_Fourier_Color_CVPR_2017_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2017/papers/Barron_Fast_Fourier_Color_CVPR_2017_paper.pdf
|
cvpr-2017-7
|
['color-constancy']
|
['computer-vision']
|
[ 1.92235280e-02 -7.81611204e-01 6.79855198e-02 2.30135992e-01
-6.42308295e-01 -6.38796866e-01 3.85940611e-01 -2.60939449e-01
-2.31818587e-01 7.73930252e-01 1.69082075e-01 -4.79477257e-01
3.96312177e-01 -4.84267175e-01 -3.41103703e-01 -7.20719516e-01
-4.42730412e-02 -2.09466025e-01 3.42687547e-01 3.18873450e-02
3.08875144e-01 2.77930111e-01 -1.77300549e+00 -1.37000158e-01
1.15663016e+00 1.05562949e+00 1.23940647e-01 1.13472891e+00
-1.87015012e-02 5.69022715e-01 -6.45317733e-01 -2.10984573e-02
2.24715367e-01 -6.15932763e-01 -3.77299935e-01 -2.57953197e-01
8.21326196e-01 -3.29349279e-01 8.46934877e-03 9.08208013e-01
5.98841786e-01 3.00811708e-01 4.34229940e-01 -1.07777333e+00
-3.74572903e-01 -3.04443687e-01 -9.42906499e-01 1.74000323e-01
4.08880502e-01 2.42153674e-01 6.04231536e-01 -6.02209568e-01
4.07135278e-01 7.79059112e-01 8.29628527e-01 3.65282297e-01
-1.54805899e+00 -5.59186578e-01 -2.42936268e-01 -5.64605668e-02
-1.29529059e+00 -5.91690183e-01 4.63441670e-01 -2.47060046e-01
8.32616925e-01 7.81424820e-01 7.85458446e-01 6.35764956e-01
1.62488222e-01 3.58565211e-01 1.50875115e+00 -5.84136307e-01
3.30005616e-01 -2.25672171e-01 -3.09547573e-01 7.10332274e-01
6.23122789e-02 3.16973418e-01 -8.20175290e-01 -1.12630323e-01
1.13586986e+00 -6.88385546e-01 -6.03646219e-01 -2.71313041e-01
-1.37180150e+00 3.86269629e-01 4.70919967e-01 -5.66323921e-02
-5.10413721e-02 7.07530618e-01 2.63410985e-01 4.58317399e-02
8.53045762e-01 5.24689257e-01 -2.31789082e-01 -7.88497031e-01
-9.72003222e-01 1.84711199e-02 5.64905465e-01 7.61871517e-01
5.99223971e-01 2.73111194e-01 -2.61358023e-01 5.85768461e-01
4.06574346e-02 1.08230495e+00 1.23113237e-01 -1.39839697e+00
1.44807994e-01 -1.65551692e-01 6.43497825e-01 -9.66180265e-01
-4.63681668e-01 -5.18326223e-01 -5.12683868e-01 6.57236636e-01
6.43518329e-01 -4.00673807e-01 -8.09338868e-01 1.70019948e+00
4.72962976e-01 5.46488881e-01 -4.96337086e-01 1.18485856e+00
1.61433578e-01 6.98084772e-01 -1.42069682e-01 -2.62717336e-01
1.21349657e+00 -9.20296192e-01 -7.13642716e-01 -6.96996553e-03
2.45076492e-01 -1.22170186e+00 1.42872286e+00 6.78689241e-01
-1.01413143e+00 -3.63785595e-01 -1.04401255e+00 -2.75817275e-01
-8.70867968e-02 1.65503815e-01 9.80713427e-01 1.10262847e+00
-1.39255285e+00 5.16547263e-01 -8.02511394e-01 -1.72383532e-01
-1.42974794e-01 -5.50819226e-02 4.25237492e-02 1.22660203e-02
-4.64123666e-01 7.20046520e-01 -2.96704382e-01 -3.07248175e-01
-1.27125964e-01 -1.10517764e+00 -7.57068813e-01 -1.91519335e-01
9.13779289e-02 -5.70657849e-01 1.27951705e+00 -1.04809415e+00
-2.10236025e+00 4.29146498e-01 -8.27495933e-01 -1.04327211e-02
4.48392421e-01 -4.55939800e-01 -5.75423896e-01 3.00408274e-01
9.69688669e-02 3.69009793e-01 9.88852084e-01 -1.34467804e+00
-3.91325325e-01 2.21991673e-01 -3.83673728e-01 3.20721000e-01
1.94403529e-02 -1.21058643e-01 -6.92163885e-01 -6.17316425e-01
2.06733927e-01 -8.31232369e-01 2.26085752e-01 4.65746105e-01
-3.19238991e-01 3.80056649e-01 7.46211648e-01 -7.59886920e-01
1.33228862e+00 -2.23696709e+00 -2.61551142e-01 1.32169023e-01
1.38295829e-01 3.99070345e-02 -6.14184812e-02 1.13128550e-01
-1.31715029e-01 -2.20645647e-02 -6.97408617e-02 -4.53752190e-01
-1.48786992e-01 -2.74000078e-01 -1.86428115e-01 7.54097164e-01
-1.40868396e-01 5.07410169e-01 -1.06617820e+00 -9.23356265e-02
6.06680036e-01 6.28362894e-01 -7.00461507e-01 -2.23938320e-02
-3.33259888e-02 6.71401322e-01 1.98070705e-01 7.22556055e-01
1.06899703e+00 -1.32707462e-01 4.87714037e-02 -2.91626990e-01
-7.65359998e-01 2.41033420e-01 -1.21580398e+00 1.79238653e+00
-7.72709966e-01 1.33940148e+00 2.12563157e-01 -1.77166715e-01
5.25214970e-01 -5.42416610e-03 6.00566983e-01 -1.20863891e+00
8.32678005e-02 2.43841261e-01 -5.46829522e-01 -3.11920881e-01
8.24601233e-01 -2.05201328e-01 2.85641491e-01 6.17820442e-01
-4.38902080e-01 -4.19877559e-01 1.53937683e-01 -9.64105409e-03
9.28260267e-01 4.98044908e-01 -2.10098267e-01 -4.82979447e-01
-1.34137601e-01 3.44486325e-03 4.27275926e-01 5.89464486e-01
-2.78734922e-01 9.46646571e-01 1.30627826e-02 -1.87282398e-01
-1.07136488e+00 -1.31461608e+00 -3.85187149e-01 9.79523540e-01
5.30238271e-01 -5.20407438e-01 -6.51944280e-01 1.73063353e-01
-7.19936490e-02 8.28901649e-01 -3.60781282e-01 2.05856755e-01
-3.67845356e-01 -7.13566542e-01 1.24824934e-01 3.43122900e-01
8.63358617e-01 -2.63570338e-01 -9.46955085e-01 2.91671250e-02
-5.12465715e-01 -8.28174412e-01 -8.10615718e-01 1.16106056e-01
-7.13107765e-01 -1.06977594e+00 -8.15197647e-01 -2.32147664e-01
4.52074677e-01 8.04717302e-01 1.41806245e+00 -5.17748203e-03
-5.72459698e-01 3.95682007e-01 -2.20886350e-01 -2.75044978e-01
1.61121234e-01 -4.48369801e-01 7.64782587e-03 -2.88850784e-01
-1.88456457e-02 -3.88702840e-01 -1.01493931e+00 3.24584186e-01
-4.51886326e-01 4.72322315e-01 1.92601178e-02 7.23308444e-01
4.62662369e-01 8.91189091e-03 -1.01941667e-01 -5.03358841e-01
4.81541961e-01 -9.30840299e-02 -9.04779434e-01 4.57187258e-02
-6.56185329e-01 9.92742926e-02 4.27564204e-01 -3.66717190e-01
-1.46856713e+00 -1.62518859e-01 1.71004325e-01 -1.67750135e-01
2.88390785e-01 -5.49236462e-02 5.24807632e-01 -4.49960381e-01
1.07735610e+00 -1.69773158e-02 5.48654459e-02 -2.64569461e-01
5.92651010e-01 2.86794990e-01 9.40888524e-01 -7.02076256e-01
7.71817803e-01 6.11802876e-01 1.28824711e-01 -8.36431026e-01
-4.96855319e-01 -6.74628913e-01 -3.87863785e-01 -3.85878503e-01
6.92181110e-01 -9.79657829e-01 -8.81862283e-01 6.61054671e-01
-8.49294662e-01 -9.00968254e-01 -8.91680568e-02 4.66429383e-01
-4.28886086e-01 2.28070855e-01 -5.06503999e-01 -1.00461483e+00
-1.22224443e-01 -6.55384421e-01 1.24410427e+00 6.12713516e-01
-1.65621057e-01 -9.50743556e-01 2.24527359e-01 -2.90859882e-02
9.08659279e-01 2.75507987e-01 5.98720491e-01 1.05891156e+00
-6.15982533e-01 1.41334742e-01 -6.71863735e-01 -2.21513346e-01
2.02376917e-01 4.40565825e-01 -1.25044322e+00 -2.60140836e-01
-2.16869667e-01 -7.65862912e-02 8.17777634e-01 8.56449425e-01
1.00774348e+00 -2.50504352e-02 -2.90435791e-01 9.73916829e-01
1.77624106e+00 -6.51188120e-02 7.77259827e-01 3.41345429e-01
4.99851048e-01 -1.50032993e-02 6.80534899e-01 6.05692923e-01
2.87967384e-01 8.40043545e-01 2.37334684e-01 -5.43312371e-01
-5.38019717e-01 6.44969493e-02 1.13593735e-01 3.84506136e-01
-3.51485252e-01 -2.00993881e-01 -5.11808097e-01 2.80893385e-01
-1.49655688e+00 -1.04631388e+00 -4.66016918e-01 2.63795328e+00
1.03509843e+00 -2.75744170e-01 2.23085910e-01 7.39312023e-02
6.19942188e-01 1.14350125e-01 -5.01612425e-01 -3.61241281e-01
-5.34669794e-02 4.77689594e-01 8.97108376e-01 7.05690444e-01
-1.05274332e+00 7.80753911e-01 7.80039024e+00 5.63811004e-01
-1.57612538e+00 1.12377666e-01 6.43905103e-01 -3.69912952e-01
-6.00150645e-01 -2.90680621e-02 -1.55242652e-01 5.89541733e-01
8.39694619e-01 7.90210143e-02 9.44351494e-01 4.65400070e-01
5.89109600e-01 -1.08855951e+00 -4.29913521e-01 1.26899815e+00
6.34258315e-02 -1.28359020e+00 -7.74130583e-01 -2.57827550e-01
8.23585868e-01 7.55453035e-02 2.67823040e-01 -1.53031990e-01
2.63087898e-01 -9.45505619e-01 9.61594820e-01 5.40655673e-01
1.41293514e+00 -6.64126575e-01 1.41818777e-01 -2.32618153e-01
-1.39636910e+00 2.73942113e-01 -2.86542773e-01 -1.03739418e-01
2.64694333e-01 1.04507899e+00 -3.96614522e-01 3.56937498e-01
8.55314732e-01 3.64549637e-01 -4.55879301e-01 1.43185532e+00
-1.50935665e-01 6.89639747e-01 -6.54778302e-01 -1.16795719e-01
-2.12374732e-01 -2.40476266e-01 3.99603695e-01 1.18201542e+00
5.87307692e-01 5.72977848e-02 -1.41432464e-01 8.07452321e-01
2.80045092e-01 -1.90068424e-01 -1.64463371e-01 2.42421746e-01
5.70843339e-01 1.23911333e+00 -8.56752932e-01 -5.17947003e-02
-3.31730992e-01 1.36282599e+00 4.16304991e-02 8.46140027e-01
-1.10890698e+00 -6.75241768e-01 9.26406980e-01 2.43844129e-02
8.88828747e-03 -7.04272687e-01 -8.27822983e-01 -1.26029336e+00
-5.86993471e-02 -5.60621142e-01 -8.97810608e-02 -1.32745612e+00
-7.74825573e-01 4.36638236e-01 -2.99167305e-01 -1.38078094e+00
-3.56816836e-02 -8.17177892e-01 -7.81523585e-01 1.19847178e+00
-1.73987508e+00 -6.04912162e-01 -8.30620706e-01 5.69060445e-01
2.19522923e-01 3.81714731e-01 9.55206037e-01 2.25566983e-01
-4.08342898e-01 3.60894293e-01 3.68237406e-01 -4.85750824e-01
1.07663107e+00 -1.48461807e+00 4.04565632e-01 1.20199943e+00
-6.43923208e-02 4.75806743e-01 1.03532779e+00 -5.75359881e-01
-1.56439722e+00 -7.61475444e-01 3.19589853e-01 -3.02483320e-01
5.38234830e-01 -2.69161880e-01 -4.05552715e-01 2.14540958e-03
3.26196671e-01 1.43741429e-01 5.83684146e-01 3.97113353e-01
-5.11776626e-01 -1.28054827e-01 -1.11439514e+00 8.27304482e-01
9.80388522e-01 -7.38632739e-01 3.38658929e-01 1.99260831e-01
5.39821923e-01 -8.86914194e-01 -3.66469145e-01 -1.18227415e-01
7.13928878e-01 -1.36895192e+00 8.47348511e-01 3.19539458e-01
7.47955441e-02 -5.67272186e-01 7.43483454e-02 -1.51323760e+00
-4.76154745e-01 -1.14771271e+00 -4.41726437e-03 9.00477171e-01
2.84825832e-01 -6.66430712e-01 6.45986438e-01 7.51175106e-01
-1.26270548e-01 -1.93413049e-01 -9.43903267e-01 -7.33089149e-01
-3.46099466e-01 -7.04894483e-01 2.90618956e-01 7.59300351e-01
1.04986362e-01 -1.90792128e-01 -3.78200889e-01 1.94837540e-01
6.80023074e-01 3.13936859e-01 6.28328085e-01 -8.95064414e-01
-3.76928389e-01 -4.93863463e-01 1.07000954e-01 -1.25813055e+00
-3.40042204e-01 -2.07943559e-01 3.11974585e-01 -1.34362340e+00
-8.32360163e-02 -8.04152369e-01 3.14517021e-01 3.02806228e-01
-3.33728015e-01 8.27926755e-01 -6.96930885e-02 -6.07829839e-02
-3.32776159e-01 3.41396183e-01 1.22866881e+00 1.17152967e-01
-4.13219213e-01 -2.11078480e-01 -4.60146844e-01 3.94634336e-01
7.74530709e-01 1.29316598e-01 -5.37717462e-01 -7.64717102e-01
2.58191437e-01 -2.51943618e-01 4.67859864e-01 -1.25005376e+00
1.72376379e-01 -3.94763261e-01 6.54263139e-01 -3.90958875e-01
5.61601341e-01 -4.21985716e-01 1.85318783e-01 1.95189297e-01
2.61375278e-01 6.50234744e-02 7.34059095e-01 2.25469843e-01
1.47271886e-01 1.30123034e-01 7.80229032e-01 9.02916342e-02
-8.17568302e-01 -7.57194683e-02 -3.96356225e-01 1.29772216e-01
7.54719317e-01 -3.70973349e-01 -7.64924526e-01 -6.29732609e-01
-1.94189280e-01 -2.56067961e-01 9.46541905e-01 -1.34899184e-01
2.00193480e-01 -1.34985626e+00 -2.94415981e-01 3.13566923e-01
-5.91109246e-02 -2.49922931e-01 3.54595304e-01 8.76131058e-01
-1.04454350e+00 2.42359981e-01 -9.69437789e-03 -7.59172976e-01
-1.05167973e+00 1.32431895e-01 4.63862568e-01 4.07838225e-01
-5.54043531e-01 1.05889738e+00 -1.24239400e-01 1.94937780e-01
-1.69205666e-03 -3.52606624e-01 4.55733925e-01 -3.53406876e-01
6.15174115e-01 6.09618068e-01 1.08935006e-01 -2.58422136e-01
-1.64044112e-01 7.69139946e-01 7.44281352e-01 -6.18188918e-01
9.63721871e-01 -5.73350728e-01 -4.50892933e-02 4.28700119e-01
8.99679065e-01 5.99509895e-01 -1.78532720e+00 2.23503798e-01
-5.03542781e-01 -9.75581229e-01 5.32565773e-01 -1.04292214e+00
-8.78447771e-01 6.74505055e-01 9.15735424e-01 3.30692559e-01
1.43346381e+00 -4.59257066e-01 7.17820048e-01 -2.27344424e-01
4.25921351e-01 -1.24275482e+00 1.76252097e-01 4.85730678e-01
5.66458583e-01 -9.68082249e-01 1.08135834e-01 -6.63838685e-01
-5.57819009e-01 1.21117711e+00 4.37100053e-01 2.05842420e-01
4.81151760e-01 6.32916510e-01 3.42744827e-01 1.68376490e-01
-4.07786965e-01 -3.52462232e-01 4.45399523e-01 9.28861737e-01
7.49444783e-01 1.26833975e-01 6.73183054e-02 -3.60793650e-01
-2.72607476e-01 -8.51638988e-02 4.19988364e-01 6.85979486e-01
-4.58852410e-01 -9.40817893e-01 -5.50831020e-01 2.43516415e-01
-4.25781235e-02 -3.52432579e-01 2.24857107e-01 4.48838562e-01
-9.79967341e-02 1.21120858e+00 3.58822405e-01 -2.38003120e-01
6.78373203e-02 -2.35260099e-01 5.85835397e-01 -2.19099015e-01
-8.94974321e-02 3.36095572e-01 2.81211615e-01 -1.07718551e+00
-4.07599121e-01 -5.69240153e-01 -1.12870860e+00 -8.42691898e-01
-2.92795986e-01 -1.30023882e-01 8.63863111e-01 6.41018450e-01
5.78125834e-01 5.36148548e-01 6.98447108e-01 -1.13782990e+00
2.46402562e-01 -6.51849568e-01 -5.85921049e-01 2.19653055e-01
5.73353648e-01 -7.54996121e-01 -3.92701656e-01 2.12826021e-02]
|
[10.424629211425781, -2.6374261379241943]
|
966338ee-75b6-4839-97f2-0b52c3ea323c
|
deep-neural-networks-for-covid-19-detection
|
2012.07655
| null |
https://arxiv.org/abs/2012.07655v4
|
https://arxiv.org/pdf/2012.07655v4.pdf
|
Deep Neural Networks for COVID-19 Detection and Diagnosis using Images and Acoustic-based Techniques: A Recent Review
|
The new coronavirus disease (COVID-19) has been declared a pandemic since March 2020 by the World Health Organization. It consists of an emerging viral infection with respiratory tropism that could develop atypical pneumonia. Experts emphasize the importance of early detection of those who have the COVID-19 virus. In this way, patients will be isolated from other people and the spread of the virus can be prevented. For this reason, it has become an area of interest to develop early diagnosis and detection methods to ensure a rapid treatment process and prevent the virus from spreading. Since the standard testing system is time-consuming and not available for everyone, alternative early-screening techniques have become an urgent need. In this study, the approaches used in the detection of COVID-19 based on deep learning (DL) algorithms, which have been popular in recent years, have been comprehensively discussed. The advantages and disadvantages of different approaches used in literature are examined in detail. The Computed Tomography of the chest and X-ray images give a rich representation of the patient's lung that is less time-consuming and allows an efficient viral pneumonia detection using the DL algorithms. The first step is the pre-processing of these images to remove noise. Next, deep features are extracted using multiple types of deep models (pre-trained models, generative models, generic neural networks, etc.). Finally, the classification is performed using the obtained features to decide whether the patient is infected by coronavirus or it is another lung disease. In this study, we also give a brief review of the latest applications of cough analysis to early screen the COVID-19, and human mobility estimation to limit its spread.
|
['Ali Narin', 'Walid Hariri']
|
2020-12-10
| null | null | null | null |
['pneumonia-detection']
|
['medical']
|
[ 5.42056337e-02 -5.47638178e-01 -1.62813246e-01 1.66482061e-01
1.03473170e-02 -3.66279483e-01 3.40235353e-01 1.33443370e-01
-6.31267846e-01 7.69473553e-01 -1.34474248e-01 -2.27336258e-01
-9.76705402e-02 -9.68729854e-01 -2.42124483e-01 -1.03696477e+00
-1.45548552e-01 1.07210052e+00 2.13255852e-01 1.01919182e-01
-2.75084406e-01 9.27399695e-01 -1.51752830e+00 2.07363248e-01
1.00251102e+00 4.40047503e-01 8.75913739e-01 9.33692396e-01
2.14870665e-02 2.84512073e-01 -5.46306729e-01 2.20477238e-01
-1.49284452e-01 -4.74617660e-01 -4.86925602e-01 -4.97891933e-01
-1.88066155e-01 -8.86936247e-01 7.92211443e-02 6.94263399e-01
6.38865173e-01 -3.29035491e-01 1.03470147e+00 -9.83493030e-01
-1.32349238e-01 -2.62371868e-01 -2.40194201e-01 5.79855621e-01
2.82163229e-02 6.00898750e-02 4.04397517e-01 -7.82331467e-01
5.69393933e-01 9.10628259e-01 8.05233777e-01 8.81292820e-01
-5.85007012e-01 -5.30781448e-01 -4.58043635e-01 4.30211395e-01
-1.13372052e+00 3.69033843e-01 4.10164624e-01 -9.11283493e-01
9.71579611e-01 2.57392555e-01 1.10036063e+00 1.31250310e+00
4.34962720e-01 6.16402566e-01 6.55042410e-01 -5.21523245e-02
-1.12476526e-02 1.79738954e-01 4.64414842e-02 7.61030793e-01
8.80332649e-01 2.46626541e-01 3.16041082e-01 -4.74887013e-01
5.75570285e-01 9.49904025e-01 -3.81074399e-01 -3.49681437e-01
-8.63234460e-01 1.04670238e+00 3.85314673e-01 7.78681815e-01
-6.50555432e-01 -3.95505935e-01 4.23119247e-01 -3.24864447e-01
1.31260857e-01 3.01128998e-03 -4.49207723e-01 2.09830806e-01
-9.63430941e-01 2.56896526e-01 4.75217372e-01 1.37160718e-01
3.96753103e-01 -1.29220784e-01 -2.54712641e-01 6.87439442e-01
5.32817602e-01 1.14309359e+00 5.14562011e-01 -3.56248349e-01
5.47720753e-02 7.01100111e-01 1.93946481e-01 -7.59125769e-01
-7.54557669e-01 -2.94232130e-01 -1.26963174e+00 6.98443577e-02
5.15759662e-02 -4.71256852e-01 -9.23225224e-01 1.39536583e+00
3.37952405e-01 2.56114215e-01 -2.94165183e-02 6.29087090e-01
8.86068046e-01 9.56081152e-01 1.17065094e-01 -4.51514870e-01
1.77598500e+00 -7.29852617e-01 -7.56857693e-01 -3.54083553e-02
5.36711097e-01 -4.79275405e-01 4.86205608e-01 1.42958701e-01
-4.27211791e-01 -4.41920310e-01 -9.21708703e-01 4.32024568e-01
-5.97629726e-01 1.29753783e-01 2.17075139e-01 7.80185640e-01
-8.68437469e-01 4.10931826e-01 -1.00034404e+00 -9.02962387e-01
4.20662612e-01 4.19831306e-01 6.40503690e-03 3.34991403e-02
-1.32306600e+00 9.41506565e-01 3.03804845e-01 1.54907972e-01
-9.37255681e-01 -3.43654186e-01 -3.68207783e-01 -3.87578979e-02
-5.61436787e-02 -1.17393112e+00 1.00566113e+00 -4.06434685e-01
-9.85962987e-01 9.31710005e-01 -3.51676464e-01 -3.03458929e-01
3.89999479e-01 -4.58079457e-01 -4.26896662e-01 3.78395021e-01
4.27678972e-02 1.98360831e-01 8.40387285e-01 -9.50830340e-01
-8.30079019e-01 -4.98947293e-01 -4.12556708e-01 -1.10467896e-01
-1.43624276e-01 3.28591734e-01 -1.90511048e-01 -4.57742780e-01
-7.33886361e-01 -1.19889331e+00 -1.71285421e-01 -3.03416967e-01
-1.39968440e-01 -5.41373909e-01 1.27528679e+00 -6.82781100e-01
1.13135076e+00 -1.84998417e+00 -2.52736539e-01 5.32355197e-02
5.08485377e-01 1.17891026e+00 1.86136112e-01 5.22637367e-01
1.88973173e-01 1.89578757e-01 -3.22741032e-01 1.77171081e-01
-5.76275587e-01 2.86216408e-01 4.61514816e-02 4.43021894e-01
4.99950945e-02 9.03999090e-01 -8.01795244e-01 -5.81931531e-01
3.18178773e-01 9.83066857e-01 -4.30433512e-01 7.36878574e-01
-1.17372178e-01 5.67915738e-01 -6.83907390e-01 5.33909321e-01
8.13128531e-01 -6.63449347e-01 1.83209151e-01 1.52279427e-02
-5.09277955e-02 -1.19924508e-01 -4.31442618e-01 5.32082438e-01
-2.31262296e-01 3.67167652e-01 9.96500179e-02 -8.18622589e-01
4.52689171e-01 8.22246015e-01 5.83336830e-01 -1.51538119e-01
3.61202657e-01 4.38836008e-01 -1.58556346e-02 -1.14669073e+00
-2.16543823e-01 -1.49754733e-01 5.34030676e-01 3.88576061e-01
-5.52669227e-01 3.15613687e-01 2.02606022e-01 -3.62102896e-01
8.91538739e-01 -3.75719279e-01 4.39269394e-01 4.64644432e-02
8.20485711e-01 4.83044833e-02 5.92486978e-01 6.91014767e-01
-3.06049705e-01 4.76046592e-01 -4.96243276e-02 -5.94403207e-01
-8.53893101e-01 -1.16067624e+00 -9.99542698e-02 6.24308467e-01
-2.41155118e-01 2.05862671e-01 -9.03147042e-01 -6.78335190e-01
-1.28188968e-01 3.86953712e-01 -4.49274331e-01 3.94344963e-02
-1.00945735e+00 -1.06609857e+00 3.84161443e-01 5.15995502e-01
5.30113578e-01 -1.46272206e+00 -1.07228363e+00 1.24996200e-01
-4.24481601e-01 -8.53105009e-01 -1.18952110e-01 1.41530866e-02
-1.08304000e+00 -1.39079714e+00 -1.24095929e+00 -1.17789006e+00
5.89840591e-01 2.28999406e-01 6.29618526e-01 5.03185689e-01
-6.25863433e-01 1.54213876e-01 -2.13308036e-01 -6.10860527e-01
-6.19741440e-01 6.77718520e-02 2.83895433e-01 -3.09531868e-01
6.89896941e-01 -1.72918290e-01 -9.36237454e-01 4.88463044e-02
-7.65143752e-01 -2.23506540e-01 7.33474612e-01 6.08600736e-01
4.70508546e-01 5.78725822e-02 5.71064711e-01 -8.79350841e-01
6.12835169e-01 -6.92650735e-01 -5.27547956e-01 2.13113159e-01
-6.07912004e-01 -3.51836383e-01 7.36205697e-01 -2.40980729e-01
-8.81208479e-01 -1.19492456e-01 -4.06428516e-01 -4.80244249e-01
-2.99814582e-01 1.36867985e-01 1.66698515e-01 4.69414800e-01
3.52849007e-01 3.49721670e-01 1.29194364e-01 -5.91944158e-01
-2.80389398e-01 9.95344043e-01 1.67980149e-01 3.25729817e-01
8.37641239e-01 6.52035892e-01 -9.22653154e-02 -1.23125041e+00
-6.40415490e-01 -7.99538016e-01 -4.81147081e-01 -2.55497396e-01
1.64818239e+00 -6.68989122e-01 -7.12462723e-01 8.04008007e-01
-1.51559639e+00 6.09732233e-03 2.05410719e-01 9.01722431e-01
-2.49925777e-01 4.04241204e-01 -7.20268726e-01 -7.26533234e-01
-1.05825996e+00 -1.13001668e+00 7.31885791e-01 3.67980242e-01
-2.46740431e-01 -1.00315273e+00 7.81276643e-01 4.54418629e-01
5.84544659e-01 1.67644665e-01 1.19384217e+00 -8.13496709e-01
-5.21384001e-01 -2.25157157e-01 -3.17698240e-01 5.92533588e-01
5.37368298e-01 1.62846744e-01 -8.72387350e-01 -4.95775014e-01
4.10427839e-01 2.39506394e-01 8.42494488e-01 7.41312921e-01
7.80851305e-01 -2.88940936e-01 -1.01840949e+00 5.24688900e-01
1.31668389e+00 8.91081512e-01 3.62128973e-01 9.93275344e-02
6.68327272e-01 4.69716042e-01 3.68326217e-01 2.02741727e-01
9.19946060e-02 3.35092694e-01 4.30089504e-01 -2.65275687e-01
1.38897941e-01 1.90016463e-01 1.96879879e-01 9.87220228e-01
-6.54725313e-01 -6.75830305e-01 -9.63443518e-01 4.27231610e-01
-1.49433470e+00 -1.22568536e+00 -3.70434821e-01 1.99852729e+00
3.84626448e-01 -2.46387511e-01 1.99362770e-01 -8.37414563e-02
9.89230871e-01 -3.56661491e-02 -3.27044040e-01 -4.79373813e-01
1.86341435e-01 1.78950921e-01 1.36345029e-01 2.26227388e-01
-1.21165621e+00 2.05787450e-01 6.37530041e+00 3.02234411e-01
-1.33719778e+00 2.23944828e-01 2.97954381e-01 2.25336447e-01
7.89946597e-03 -5.74376404e-01 -8.89848828e-01 5.83956182e-01
7.45808780e-01 1.49045810e-01 1.85517907e-01 7.97903836e-01
3.55683148e-01 7.82462060e-02 -6.98935926e-01 9.72915888e-01
1.17777385e-01 -1.15666842e+00 2.05106344e-02 1.17653683e-01
4.70365554e-01 5.82443118e-01 -2.02360556e-01 1.48667529e-01
-3.38366389e-01 -7.68716276e-01 -2.41759360e-01 6.01066947e-01
7.39685357e-01 -7.28379488e-01 1.29496324e+00 6.14156425e-01
-1.26105666e+00 -6.62520826e-02 -2.85610527e-01 3.30401987e-01
4.16433781e-01 6.76640987e-01 -1.39965141e+00 4.54467796e-02
9.06377137e-01 2.80623108e-01 -1.95251539e-01 1.19127369e+00
-1.91313639e-01 5.16084909e-01 -3.80342185e-01 -4.91455942e-01
4.21101563e-02 -1.91738054e-01 7.33751774e-01 1.41879308e+00
5.63712299e-01 5.63586801e-02 4.72614542e-02 7.17080653e-01
2.84706175e-01 1.22863412e-01 -9.20242012e-01 -2.31531873e-01
5.10902517e-02 1.07279980e+00 -7.50495434e-01 -4.55309659e-01
-4.70230341e-01 9.26911771e-01 -1.21488862e-01 2.50653744e-01
-7.93692172e-01 -3.60434681e-01 5.30324101e-01 5.44562995e-01
5.49598575e-01 1.36024088e-01 4.08710450e-01 -7.78383672e-01
-4.18102533e-01 -6.49939179e-01 5.16086519e-01 -5.45230687e-01
-1.12912393e+00 7.40775406e-01 7.44577721e-02 -1.01496816e+00
-5.05524695e-01 -7.36700714e-01 -8.80009174e-01 7.24713087e-01
-1.33020222e+00 -6.09778643e-01 -3.80260944e-01 3.46268505e-01
4.19811100e-01 -2.32151091e-01 1.04324031e+00 4.04771119e-01
-6.12761259e-01 -3.63477916e-02 4.05733645e-01 1.07608117e-01
2.10013643e-01 -8.62412691e-01 8.24583173e-02 5.93725264e-01
-5.48602223e-01 7.43944645e-01 4.66910124e-01 -1.06298888e+00
-7.47776926e-01 -1.20764875e+00 1.34161878e+00 -2.78725088e-01
1.10290855e-01 -2.47573450e-01 -8.91774416e-01 4.22920376e-01
3.22336704e-01 -3.78586888e-01 7.21196711e-01 -6.26116216e-01
2.96901673e-01 -2.93352865e-02 -1.21527898e+00 4.41456288e-01
6.43328428e-01 -3.86313528e-01 -8.21652234e-01 5.50785363e-01
5.00056088e-01 2.37312451e-01 -2.94894725e-01 8.64938557e-01
6.31405890e-01 -1.11272275e+00 1.07545686e+00 -3.94036591e-01
-1.97942123e-01 -2.15707421e-01 3.72096568e-01 -8.40018213e-01
-5.00388205e-01 -1.72114968e-01 -8.79154056e-02 5.90105057e-01
-3.26896459e-02 -7.91511357e-01 7.77953982e-01 -2.23501295e-01
3.23874176e-01 -9.70654070e-01 -6.76495969e-01 -6.86194658e-01
-2.86520898e-01 2.70546917e-02 3.77831608e-01 6.45951509e-01
-5.83739042e-01 3.61470550e-01 -3.86103362e-01 1.40087485e-01
3.66562039e-01 1.15631893e-01 4.34406936e-01 -1.64859986e+00
-3.22844684e-02 -3.26062858e-01 -4.13329974e-02 -6.64133906e-01
-3.32491040e-01 -6.61893249e-01 -8.89637768e-02 -2.15704513e+00
4.22524780e-01 -1.04244120e-01 -4.17922556e-01 2.44378686e-01
-1.38676941e-01 -3.66938155e-04 -2.17333272e-01 4.39793319e-01
1.30490074e-02 2.62265146e-01 1.31182480e+00 8.56917177e-04
-3.41237128e-01 6.06500387e-01 -6.51548728e-02 1.02360034e+00
1.26595569e+00 -7.51642466e-01 -5.45795739e-01 -9.59931985e-02
2.75709659e-01 -1.50520355e-01 3.77228141e-01 -9.11627948e-01
-1.50414690e-01 -1.69370443e-01 3.27330858e-01 -1.36706460e+00
2.90650994e-01 -9.70849931e-01 1.98597819e-01 1.49409366e+00
5.90458214e-01 1.65184796e-01 -6.44067489e-03 4.44924593e-01
1.21600427e-01 -4.34061199e-01 9.72746015e-01 -1.01752765e-01
-2.38331124e-01 4.22978371e-01 -1.23904276e+00 1.01928376e-01
1.22591412e+00 -1.04388170e-01 -1.52122587e-01 -7.25468919e-02
-4.35294360e-01 -1.06732352e-02 1.88828319e-01 2.41370454e-01
6.86934590e-01 -9.31410193e-01 -7.07855225e-01 4.28584009e-01
-9.66656879e-02 -5.30365109e-02 5.09882271e-01 1.10602176e+00
-1.21191585e+00 7.41771579e-01 -9.52043533e-02 -7.76317120e-01
-1.64243817e+00 1.00220740e+00 4.76149380e-01 -4.88837510e-01
-6.03345275e-01 5.64958453e-01 4.98296767e-01 -3.03901315e-01
1.37202516e-01 -3.66659194e-01 -8.91532183e-01 8.12865421e-02
6.59608662e-01 5.99273860e-01 -1.39997140e-01 -6.50084019e-01
-8.61314237e-01 7.22842395e-01 -1.01203382e-01 6.77402616e-01
1.25959587e+00 1.16655342e-01 -4.37802106e-01 2.19261616e-01
1.26213443e+00 6.68170750e-02 -4.62055981e-01 1.70770109e-01
-3.34650964e-01 7.83958882e-02 -3.36578727e-01 -5.95575511e-01
-8.74818444e-01 1.00323129e+00 1.20474398e+00 3.45324636e-01
1.06064439e+00 1.67859033e-01 1.28017652e+00 4.06434864e-01
1.52344294e-02 -8.02781999e-01 -5.69840036e-02 4.54026043e-01
6.65277004e-01 -1.12937927e+00 -2.10714057e-01 -2.35803336e-01
-1.76095083e-01 9.60818589e-01 2.66560256e-01 -8.78076628e-03
9.37309861e-01 2.15179980e-01 3.39055449e-01 -4.30902451e-01
-3.82352293e-01 -8.21747035e-02 1.52869746e-01 8.87876093e-01
2.28567347e-01 2.60171324e-01 -5.13178289e-01 4.52874660e-01
2.02995598e-01 3.04776162e-01 -4.13180776e-02 8.29232574e-01
-7.78012991e-01 -9.55256462e-01 -5.25272489e-01 7.92889237e-01
-6.01300061e-01 5.19904681e-02 -2.92065203e-01 7.57502794e-01
5.48376560e-01 6.54215336e-01 -4.16410044e-02 -8.57516006e-02
-4.07896265e-02 8.55570957e-02 3.00376117e-01 -6.31191194e-01
-2.19181210e-01 -4.95664254e-02 -1.92576453e-01 -1.12977192e-01
-5.95271051e-01 -4.94487762e-01 -1.59479618e+00 -5.44960797e-02
-1.21968724e-01 2.15522468e-01 4.36099619e-01 9.22003627e-01
1.74080521e-01 3.50796282e-01 5.02842605e-01 -5.27278543e-01
-2.04605684e-01 -6.76465213e-01 -4.85666811e-01 1.62569717e-01
5.67619205e-01 -5.61382353e-01 -5.68333983e-01 -2.00322628e-01]
|
[15.589373588562012, -1.6785900592803955]
|
06facda6-56b2-47c5-ace5-21412f6dbf24
|
spirit-diffusion-self-consistency-driven
|
2304.05060
| null |
https://arxiv.org/abs/2304.05060v1
|
https://arxiv.org/pdf/2304.05060v1.pdf
|
SPIRiT-Diffusion: Self-Consistency Driven Diffusion Model for Accelerated MRI
|
Diffusion models are a leading method for image generation and have been successfully applied in magnetic resonance imaging (MRI) reconstruction. Current diffusion-based reconstruction methods rely on coil sensitivity maps (CSM) to reconstruct multi-coil data. However, it is difficult to accurately estimate CSMs in practice use, resulting in degradation of the reconstruction quality. To address this issue, we propose a self-consistency-driven diffusion model inspired by the iterative self-consistent parallel imaging (SPIRiT), namely SPIRiT-Diffusion. Specifically, the iterative solver of the self-consistent term in SPIRiT is utilized to design a novel stochastic differential equation (SDE) for diffusion process. Then $\textit{k}$-space data can be interpolated directly during the reverse diffusion process, instead of using CSM to separate and combine individual coil images. This method indicates that the optimization model can be used to design SDE in diffusion models, driving the diffusion process strongly conforming with the physics involved in the optimization model, dubbed model-driven diffusion. The proposed SPIRiT-Diffusion method was evaluated on a 3D joint Intracranial and Carotid Vessel Wall imaging dataset. The results demonstrate that it outperforms the CSM-based reconstruction methods, and achieves high reconstruction quality at a high acceleration rate of 10.
|
['Yanjie Zhu', 'Dong Liang', 'Hairong Zheng', 'Sen Jia', 'Jing Cheng', 'Chentao Cao', 'Zhuo-Xu Cui']
|
2023-04-11
| null | null | null | null |
['mri-reconstruction']
|
['computer-vision']
|
[ 5.21583483e-02 -1.97015733e-01 3.51386726e-01 -3.45082551e-01
-4.48106527e-01 -2.46081114e-01 4.36210841e-01 -2.46372759e-01
-3.74445617e-01 5.13319373e-01 3.44257593e-01 -2.25643679e-01
-3.62251937e-01 -4.58924025e-01 -1.96810573e-01 -1.12908936e+00
-1.13770738e-01 3.93550962e-01 4.28912878e-01 -2.87391152e-03
2.07595319e-01 4.44722801e-01 -6.21674061e-01 -1.55706346e-01
1.26324320e+00 7.44216323e-01 7.07487404e-01 1.78772926e-01
-3.38431597e-01 9.62860584e-01 -1.28121763e-01 7.89554715e-02
1.17850013e-01 -8.88745308e-01 -7.24284649e-01 7.02222213e-02
-2.69160599e-01 -3.20027590e-01 -4.67397153e-01 1.28762865e+00
8.44146967e-01 2.25923300e-01 6.79057717e-01 -4.94496465e-01
-5.17692924e-01 5.98464727e-01 -9.20148849e-01 7.12336421e-01
-8.33201408e-02 1.52614161e-01 -6.40060157e-02 -9.96163070e-01
9.14223135e-01 8.55413318e-01 4.49510962e-01 5.50448537e-01
-1.22041702e+00 -5.83611190e-01 -1.08934537e-01 8.20493549e-02
-1.31179667e+00 -2.79056311e-01 1.06870747e+00 -7.04023838e-01
3.87699187e-01 7.10802078e-02 8.42149138e-01 7.31616735e-01
7.12610483e-01 5.78748941e-01 1.73467898e+00 -1.71724856e-01
3.87587190e-01 -9.48292241e-02 2.51496196e-01 5.24378896e-01
1.49659649e-01 1.51688457e-01 -2.14517742e-01 -3.36299688e-01
1.12514544e+00 -1.93677232e-01 -5.80471516e-01 -3.79351377e-01
-1.35297346e+00 7.68594265e-01 4.79440987e-01 7.12705135e-01
-9.12247181e-01 -1.97038233e-01 3.33507061e-01 -6.16375078e-03
5.67425549e-01 1.94524258e-01 2.95998394e-01 1.41250834e-01
-1.16843903e+00 2.42235422e-01 4.90428925e-01 4.22126055e-01
2.59216905e-01 2.27700293e-01 -3.31172466e-01 9.17705536e-01
5.17634273e-01 5.99778652e-01 9.10201311e-01 -8.71049285e-01
1.43218506e-02 9.79714096e-02 -4.14801091e-02 -9.40189838e-01
-6.74908340e-01 -7.45864511e-01 -1.32298720e+00 6.82335347e-02
2.08907276e-01 -3.30190808e-01 -7.23829865e-01 1.42842710e+00
7.71033525e-01 4.00403261e-01 -1.77161738e-01 1.55149353e+00
5.62356949e-01 6.87061012e-01 1.39248922e-01 -6.58986330e-01
1.10171223e+00 -7.98436224e-01 -1.04248166e+00 -2.97136344e-02
5.78304827e-01 -7.35564828e-01 5.05956113e-01 4.14765745e-01
-1.29077184e+00 -2.37671763e-01 -9.22723889e-01 4.68731463e-01
5.69718122e-01 -4.12937999e-01 5.77984452e-01 3.53712916e-01
-1.01047301e+00 5.73990107e-01 -1.12960529e+00 1.38301671e-01
2.28564084e-01 3.84391770e-02 -3.78546417e-02 -2.54098505e-01
-1.12879682e+00 1.05323815e+00 1.50868427e-02 2.63424844e-01
-9.33257699e-01 -8.18490446e-01 -2.96204448e-01 -6.09213114e-01
-4.50805873e-02 -7.33715057e-01 9.16102827e-01 -7.24008441e-01
-1.73467886e+00 6.07740164e-01 -2.36422241e-01 -4.32956159e-01
6.85617745e-01 3.33751142e-01 -5.25812745e-01 4.17156994e-01
1.99437305e-01 3.47853243e-01 1.03079498e+00 -1.23791885e+00
1.25242740e-01 -4.66124207e-01 -4.65379059e-01 2.57972956e-01
-1.92054007e-02 9.54902992e-02 -2.89845437e-01 -7.58916378e-01
7.81328976e-01 -1.09962273e+00 -6.11575603e-01 -9.26371068e-02
-2.90632308e-01 2.97026038e-01 4.00745690e-01 -1.11119413e+00
1.25195456e+00 -2.02581239e+00 3.88103306e-01 3.74279588e-01
6.92908823e-01 1.58877388e-01 9.30411667e-02 -9.53359157e-02
-2.13545144e-01 -4.35173422e-01 -7.74682879e-01 -7.16610700e-02
-6.09254956e-01 1.58448033e-02 8.58196020e-02 8.46738815e-01
-3.55816633e-01 7.51271069e-01 -9.43536997e-01 -5.24774551e-01
2.15067416e-01 7.45893836e-01 -4.39418852e-01 2.09665835e-01
3.16451252e-01 1.53397608e+00 -7.91009068e-01 4.76728678e-02
1.19071269e+00 -2.41020769e-01 3.90547752e-01 -3.47704291e-01
-3.61572921e-01 4.22119945e-02 -1.24216080e+00 2.01506329e+00
-5.08489013e-01 9.32606682e-02 4.81253028e-01 -1.01450336e+00
9.90483522e-01 3.66487861e-01 1.05261028e+00 -9.67775941e-01
1.17820144e-01 6.36644840e-01 4.73500133e-01 -6.37297571e-01
-6.11105794e-03 -4.93129551e-01 5.63824534e-01 5.39062202e-01
-2.61467874e-01 -4.00653452e-01 -7.29020312e-02 2.29225099e-01
8.98052394e-01 -1.62209980e-02 -3.02596211e-01 -9.96666789e-01
6.91296518e-01 -5.84430248e-02 5.41444898e-01 5.67758739e-01
-2.80830383e-01 5.50744772e-01 6.97562248e-02 -4.19152975e-01
-9.43407416e-01 -9.24966931e-01 -5.53841412e-01 -5.53561226e-02
4.32577819e-01 7.79003799e-02 -1.04688644e+00 -3.51997733e-01
-4.65446949e-01 5.19846559e-01 -3.69572699e-01 -7.07484409e-02
-8.25851738e-01 -1.29623187e+00 8.14461708e-02 -9.51608568e-02
7.33415067e-01 -7.89377868e-01 -5.04498422e-01 7.71174192e-01
-4.21596736e-01 -8.85912776e-01 -5.54212749e-01 -7.30048865e-02
-1.22554028e+00 -7.66293526e-01 -1.41242886e+00 -5.07962584e-01
6.31092072e-01 3.29498708e-01 6.27154946e-01 -1.07390061e-01
-1.32794648e-01 1.50628567e-01 -1.54977351e-01 1.74786448e-01
-5.84428787e-01 -2.50546008e-01 1.05022222e-01 4.06388223e-01
-1.10124454e-01 -7.60652304e-01 -1.15797973e+00 3.37282479e-01
-9.50460911e-01 2.35417545e-01 4.89560664e-01 8.30176830e-01
7.80285776e-01 1.84270460e-02 5.23555875e-01 -5.70680797e-01
8.84744167e-01 -6.51661038e-01 -5.62170863e-01 3.45541686e-02
-7.13990748e-01 2.74332106e-01 3.49560022e-01 -5.72258949e-01
-1.19892490e+00 -4.42589261e-02 -4.25715178e-01 -3.97739202e-01
2.19380662e-01 4.29627240e-01 3.35087329e-01 -4.76651311e-01
5.13810098e-01 7.57319510e-01 3.51778865e-01 -4.55193132e-01
1.03114106e-01 4.40834612e-01 1.97476253e-01 -5.27992368e-01
3.11382234e-01 6.94328845e-01 2.31540158e-01 -7.74741232e-01
-3.27500850e-01 -1.92129642e-01 -4.31420475e-01 -5.41213632e-01
9.97017384e-01 -5.56849718e-01 -4.19508040e-01 8.35351825e-01
-1.15683055e+00 -3.02924037e-01 -1.43873125e-01 1.10816038e+00
-3.54473770e-01 6.17282867e-01 -8.53943348e-01 -5.62799931e-01
-5.26358843e-01 -1.94261146e+00 4.55887377e-01 -1.79183148e-02
-1.73691623e-02 -1.15398574e+00 1.29470080e-01 3.19971964e-02
9.80232239e-01 2.21435547e-01 8.09357405e-01 6.84584007e-02
-4.97384191e-01 1.87335014e-01 -1.12078331e-01 2.55501866e-01
-8.58122036e-02 -7.03487992e-01 -5.18770337e-01 -2.32956409e-01
9.35649931e-01 1.63211435e-01 5.80133259e-01 9.93951857e-01
1.04106486e+00 6.19704947e-02 -3.54311317e-01 6.98856831e-01
1.37482882e+00 2.97026396e-01 6.75904274e-01 1.54750854e-01
5.31200469e-01 6.09197497e-01 1.39455408e-01 3.62059832e-01
4.04488146e-01 7.36137688e-01 -3.82237136e-02 -1.88420981e-01
-5.71671844e-01 2.15558782e-02 -4.18920517e-02 1.48518884e+00
-8.04279670e-02 3.34838361e-01 -1.02581656e+00 3.23828310e-01
-1.55109954e+00 -6.10754132e-01 -8.47843528e-01 2.09954405e+00
8.45015705e-01 -1.86921969e-01 -1.64658353e-01 -1.44858867e-01
7.19948292e-01 1.25763509e-02 -7.38090694e-01 2.59194653e-02
-7.51966089e-02 9.30811539e-02 5.07397711e-01 8.26157331e-01
-5.51908076e-01 4.02736932e-01 6.21747351e+00 7.63247013e-01
-1.51889241e+00 8.61675501e-01 5.13197899e-01 1.40282005e-01
-5.78672647e-01 1.09737597e-01 -3.29101056e-01 7.24314094e-01
8.06788802e-01 -7.49225765e-02 5.36966145e-01 3.99945676e-01
7.37554550e-01 -4.04658616e-01 -2.88293660e-01 1.16525185e+00
-1.74098879e-01 -1.21665382e+00 -1.74393594e-01 2.44945392e-01
7.18233228e-01 1.50018141e-01 -1.19057223e-01 -2.54366606e-01
5.52996323e-02 -5.48004866e-01 6.14665210e-01 7.95656204e-01
5.31421185e-01 -3.59436065e-01 2.71569222e-01 4.78834510e-01
-8.43916655e-01 2.97022820e-01 -1.81388319e-01 3.92090261e-01
7.53990054e-01 1.37407374e+00 -3.01005930e-01 5.09840012e-01
5.61186552e-01 6.32940710e-01 -1.60748698e-02 1.11550748e+00
-2.77170911e-03 6.27719045e-01 -1.58420622e-01 3.52377892e-01
2.24748030e-01 -7.47531056e-01 9.12715316e-01 7.81280756e-01
4.47883338e-01 3.99896175e-01 -1.64907798e-03 1.13010073e+00
3.57109874e-01 1.55293822e-01 -2.49579832e-01 4.86352324e-01
1.64485365e-01 1.24089170e+00 -8.29094410e-01 -4.13677931e-01
-2.39795193e-01 1.17170572e+00 2.84628049e-02 4.17429894e-01
-8.76751900e-01 8.29703659e-02 7.78198093e-02 4.14519340e-01
-1.40938789e-01 -6.11845136e-01 -2.72816718e-01 -1.31941664e+00
-6.07540756e-02 -6.31171584e-01 -6.94980249e-02 -6.48946643e-01
-1.29726946e+00 1.01802349e+00 7.76440874e-02 -1.01305556e+00
4.52602208e-02 -5.96920699e-02 -4.53378886e-01 1.31404090e+00
-1.65872669e+00 -5.82739770e-01 -2.49533385e-01 7.23157406e-01
2.55844355e-01 2.38008529e-01 4.80551243e-01 6.35272264e-01
-4.02278185e-01 -2.73713693e-02 3.03504169e-01 -2.65466034e-01
4.75933731e-01 -8.57419193e-01 4.21731509e-02 7.51775384e-01
-5.29214501e-01 6.93553269e-01 5.87539554e-01 -9.36958075e-01
-1.47820222e+00 -7.96237111e-01 4.38528001e-01 1.62346274e-01
6.73173666e-01 6.39531091e-02 -1.16310143e+00 1.25515118e-01
2.42520139e-01 3.50915164e-01 3.11086178e-01 -7.36204922e-01
3.07771862e-01 -4.07100320e-02 -1.37850952e+00 2.22902730e-01
9.77568865e-01 -1.23914599e-01 -3.15878391e-01 3.69590133e-01
4.00826395e-01 -6.55092478e-01 -1.20992780e+00 2.51982927e-01
3.38475406e-01 -8.21033835e-01 8.95620406e-01 1.33000344e-01
2.92421520e-01 -3.01572978e-01 2.47510642e-01 -1.57395840e+00
-6.30598187e-01 -5.87450385e-01 -1.38052315e-01 7.90019095e-01
-1.12491213e-02 -8.64705563e-01 3.25415254e-01 6.78118825e-01
-3.03044379e-01 -6.20897830e-01 -1.19427705e+00 -6.50753736e-01
3.52146059e-01 -2.58156300e-01 4.35819834e-01 1.14608502e+00
-2.10932046e-01 -9.97380465e-02 -2.24235564e-01 1.27294302e-01
1.09996521e+00 -1.04152948e-01 1.05262138e-02 -9.27368224e-01
-4.79569972e-01 -4.06759679e-01 8.54683146e-02 -1.36616158e+00
-5.11352569e-02 -1.11856556e+00 -9.62202996e-02 -1.66500354e+00
5.97250387e-02 -1.09705245e+00 -1.43803775e-01 -2.24870607e-01
-8.01413506e-02 4.51283753e-02 -5.89668676e-02 8.05800498e-01
2.89133321e-02 6.41657233e-01 2.11927462e+00 -7.19337761e-02
-1.72181189e-01 -3.16979527e-01 -3.22751969e-01 5.17523170e-01
5.37305951e-01 -5.51308274e-01 -4.13066626e-01 -6.23470485e-01
-3.44144106e-01 6.65191531e-01 2.44253635e-01 -1.04652143e+00
4.33441460e-01 8.52976739e-03 1.37671113e-01 -2.43309557e-01
3.58152762e-02 -7.79625773e-01 4.42329764e-01 8.23280692e-01
-1.36036724e-01 -1.20385669e-01 -1.99560914e-03 1.89419404e-01
-2.35438630e-01 -3.16796571e-01 1.17038977e+00 -3.11158806e-01
-3.73861730e-01 4.40392941e-01 -5.40105522e-01 4.67746630e-02
8.47959042e-01 1.64546102e-01 3.06086630e-01 -1.00961059e-01
-9.71879542e-01 -1.27629638e-01 3.07429377e-02 1.54158063e-02
6.92700863e-01 -1.35909486e+00 -8.43871653e-01 2.45240957e-01
-3.76998961e-01 -5.53233176e-02 9.61200237e-01 1.71973729e+00
-6.09677732e-01 2.08491743e-01 -6.37393892e-02 -8.41466486e-01
-3.68886948e-01 4.77932125e-01 8.54175210e-01 -4.14121330e-01
-1.10002768e+00 6.79019213e-01 1.69518977e-01 -2.55590171e-01
-3.33829343e-01 -1.36117160e-01 -3.48919630e-03 -3.39173466e-01
6.84569001e-01 3.05922985e-01 2.29400843e-01 -7.67077088e-01
-4.69795913e-01 7.82494605e-01 8.91827606e-03 -4.09034252e-01
1.24553728e+00 -4.73211855e-01 -3.43115032e-01 -7.25691440e-03
1.15325403e+00 -1.03884682e-01 -1.29716170e+00 -3.42348278e-01
-1.67891115e-01 -3.03179204e-01 8.30667615e-01 -7.27971375e-01
-1.42940402e+00 8.86866570e-01 1.01090384e+00 -1.01597831e-01
9.03106689e-01 -3.18511665e-01 1.15183306e+00 -5.11937857e-01
6.68165207e-01 -6.52010679e-01 -2.42983937e-01 1.38580367e-01
1.01387894e+00 -1.01312995e+00 -1.12305775e-01 -3.67721438e-01
-7.45497286e-01 9.72993135e-01 2.47491017e-01 -2.08090648e-01
1.02441645e+00 4.85658705e-01 1.45931169e-02 -3.71529639e-01
3.57204899e-02 3.20956141e-01 1.41057357e-01 3.61118644e-01
5.76964259e-01 2.59537939e-02 -9.49021459e-01 3.98930252e-01
4.28085446e-01 4.50454295e-01 2.84319401e-01 8.11177552e-01
-1.22820862e-01 -1.21023488e+00 -5.22681534e-01 2.40729332e-01
-2.20909432e-01 -1.82084411e-01 3.68923247e-01 5.21110818e-02
-4.33598496e-02 8.21645617e-01 -2.21967161e-01 -2.48443428e-03
1.81148216e-01 -2.11881205e-01 6.98306561e-01 -7.01362640e-02
-4.42179024e-01 3.99514288e-01 -3.81631672e-01 -4.77121532e-01
-6.00487530e-01 -7.41601050e-01 -1.55423534e+00 -2.04470754e-01
-1.32435545e-01 3.96597028e-01 1.02714777e+00 8.77226114e-01
4.96431112e-01 6.03194237e-01 8.08247387e-01 -7.50065923e-01
-4.26410526e-01 -9.53833342e-01 -8.26273203e-01 4.61391807e-01
1.08184807e-01 -8.04136097e-01 -2.41446793e-01 -2.59911299e-01]
|
[13.534701347351074, -2.3960635662078857]
|
53bfa387-6990-4a0b-a9bb-8a04162a9ad0
|
3dfacefill-an-analysis-by-synthesis-approach
|
2110.10395
| null |
https://arxiv.org/abs/2110.10395v1
|
https://arxiv.org/pdf/2110.10395v1.pdf
|
3DFaceFill: An Analysis-By-Synthesis Approach to Face Completion
|
Existing face completion solutions are primarily driven by end-to-end models that directly generate 2D completions of 2D masked faces. By having to implicitly account for geometric and photometric variations in facial shape and appearance, such approaches result in unrealistic completions, especially under large variations in pose, shape, illumination and mask sizes. To alleviate these limitations, we introduce 3DFaceFill, an analysis-by-synthesis approach for face completion that explicitly considers the image formation process. It comprises three components, (1) an encoder that disentangles the face into its constituent 3D mesh, 3D pose, illumination and albedo factors, (2) an autoencoder that inpaints the UV representation of facial albedo, and (3) a renderer that resynthesizes the completed face. By operating on the UV representation, 3DFaceFill affords the power of correspondence and allows us to naturally enforce geometrical priors (e.g. facial symmetry) more effectively. Quantitatively, 3DFaceFill improves the state-of-the-art by up to 4dB higher PSNR and 25% better LPIPS for large masks. And, qualitatively, it leads to demonstrably more photorealistic face completions over a range of masks and occlusions while preserving consistency in global and component-wise shape, pose, illumination and eye-gaze.
|
['Vishnu Boddeti', 'Rahul Dey']
|
2021-10-20
| null | null | null | null |
['facial-inpainting']
|
['computer-vision']
|
[ 1.74600855e-01 2.76215136e-01 4.58471894e-01 -4.36706603e-01
-6.70622766e-01 -6.41451240e-01 7.18102276e-01 -4.82018471e-01
5.21658808e-02 4.75178003e-01 4.01608855e-01 5.97145744e-02
2.94604689e-01 -5.49303830e-01 -7.82936633e-01 -7.36744761e-01
8.80087465e-02 2.02993587e-01 -3.65800798e-01 -1.00084916e-01
-1.27494454e-01 9.34090018e-01 -1.83364391e+00 2.36798063e-01
5.57717085e-01 1.15607488e+00 -3.13351214e-01 6.01276338e-01
2.26532415e-01 3.57693315e-01 -3.87410879e-01 -7.49959111e-01
6.78737819e-01 -3.17134649e-01 -2.80065030e-01 7.89338350e-01
1.19352913e+00 -6.69851005e-01 -3.78509521e-01 8.41144562e-01
4.56343353e-01 -1.10243700e-01 6.04658365e-01 -1.04617429e+00
-6.05530322e-01 -2.94916928e-01 -9.41545606e-01 -4.81930584e-01
5.44711649e-01 4.71564025e-01 6.70769095e-01 -1.35029340e+00
7.06974089e-01 1.63686538e+00 7.36561000e-01 5.30166388e-01
-1.85870469e+00 -5.22756219e-01 -5.82636632e-02 -5.35972297e-01
-1.52877367e+00 -1.03262794e+00 8.32322776e-01 -3.79007131e-01
7.01099813e-01 4.30314600e-01 4.79140401e-01 8.17476213e-01
1.83997989e-01 1.46051526e-01 1.18070126e+00 -4.85281438e-01
-3.10059520e-04 -1.45414963e-01 -5.56597173e-01 7.65079856e-01
-4.93833013e-02 2.65420079e-01 -3.99257511e-01 -3.89398456e-01
1.10131502e+00 -1.98265165e-01 -3.43923271e-01 -4.53092128e-01
-6.72380030e-01 4.88438100e-01 2.32373342e-01 -3.40021938e-01
-4.44194347e-01 1.64642841e-01 -1.62274733e-01 2.32430682e-01
8.28869045e-01 3.30160022e-01 -2.02896491e-01 4.75759417e-01
-9.32891965e-01 5.97735465e-01 8.12655747e-01 8.88570607e-01
1.09256721e+00 3.30034047e-01 -8.18676502e-02 9.18585420e-01
4.15212750e-01 8.09887111e-01 -2.76729524e-01 -1.54385543e+00
2.12677628e-01 2.64343143e-01 2.39183247e-01 -1.08117950e+00
-1.06667668e-01 -1.66583210e-01 -3.97401214e-01 7.13455558e-01
2.64796615e-01 -2.18927309e-01 -1.09163034e+00 1.99784517e+00
6.64882779e-01 -4.45452929e-02 -2.17893556e-01 9.52941895e-01
7.58700013e-01 4.37752485e-01 -1.38087720e-01 -3.59643966e-01
1.62713552e+00 -5.58511376e-01 -7.94123292e-01 -3.69978070e-01
-1.69915482e-01 -1.20025671e+00 8.02449405e-01 3.34994793e-01
-1.73143148e+00 -4.78089690e-01 -9.31525111e-01 -4.03297842e-01
2.38873139e-01 1.56698763e-01 3.05237710e-01 6.81877971e-01
-1.55638814e+00 5.18961370e-01 -6.19458973e-01 -9.99055356e-02
4.01121765e-01 5.41141331e-01 -6.85676396e-01 -2.54621595e-01
-5.10602415e-01 7.39027679e-01 -3.44573528e-01 4.38561253e-02
-8.39141190e-01 -9.08549905e-01 -1.11099267e+00 1.33483633e-01
1.96459979e-01 -7.77448654e-01 1.10690844e+00 -1.19090283e+00
-1.70339274e+00 1.20500708e+00 -4.02798027e-01 1.82764083e-01
5.55597365e-01 -7.49820471e-02 -1.91931173e-01 1.54862255e-01
-1.37548819e-01 9.33166504e-01 1.47972703e+00 -1.43887174e+00
-8.37803110e-02 -5.73499739e-01 -1.46877095e-01 4.36241865e-01
-2.02595398e-01 1.49465889e-01 -8.81086886e-01 -7.28456914e-01
2.04392180e-01 -9.05280411e-01 1.43148527e-02 5.78127503e-01
-1.51899904e-01 3.87550294e-01 9.02771115e-01 -1.11039376e+00
6.34664118e-01 -2.27646351e+00 2.52013773e-01 1.13394201e-01
2.94104695e-01 1.37874618e-01 -3.16143721e-01 3.58720243e-01
-3.68863821e-01 -1.21158093e-01 -4.41211723e-02 -1.16949487e+00
-6.53862134e-02 1.32343307e-01 -1.27473921e-01 8.12788546e-01
4.75455403e-01 7.25800991e-01 -3.82521987e-01 -1.96204826e-01
2.35653758e-01 1.06556594e+00 -9.16642189e-01 3.25599819e-01
-2.36851022e-01 5.80258965e-01 -1.98101625e-02 7.64238715e-01
1.12948024e+00 9.32599828e-02 2.02241868e-01 -2.98075169e-01
-7.10694268e-02 -1.62101723e-02 -1.22425961e+00 1.71679223e+00
-4.18242574e-01 6.49025798e-01 8.77628267e-01 -2.68100381e-01
8.30736816e-01 4.47846949e-01 5.12843728e-01 -5.84362268e-01
2.53229439e-01 2.49695614e-01 -3.91506523e-01 -3.22976947e-01
1.62683189e-01 -3.62323195e-01 5.02378166e-01 2.69726068e-01
1.30337715e-01 -4.28997397e-01 -1.32090449e-01 -1.25551119e-01
8.36097419e-01 2.59873450e-01 -1.46704733e-01 -2.41733760e-01
5.46348989e-01 -6.56843364e-01 4.41484123e-01 -1.55218005e-01
1.73099846e-01 1.21927297e+00 5.20299435e-01 -5.19492328e-01
-1.32493591e+00 -1.30442703e+00 -1.16745964e-01 6.28547907e-01
-2.10174143e-01 -2.10836515e-01 -1.15717030e+00 -8.90396684e-02
1.58756003e-01 4.65533406e-01 -5.87548435e-01 -7.31932884e-03
-6.04192317e-01 -3.72315615e-01 3.88431579e-01 2.13265687e-01
3.14061105e-01 -7.52979338e-01 -3.10581267e-01 -5.26095927e-02
-1.63477644e-01 -1.25038302e+00 -8.48611116e-01 -3.45522821e-01
-8.41830373e-01 -9.46014941e-01 -6.74065828e-01 -6.13573432e-01
1.09042811e+00 1.76532298e-01 1.20541370e+00 1.90525129e-01
-4.35967326e-01 3.89752239e-01 1.48399860e-01 -3.54339868e-01
-4.47801173e-01 -6.34212792e-01 7.14978725e-02 5.97727299e-01
-2.73564279e-01 -1.03789556e+00 -8.91161144e-01 3.17748159e-01
-9.22535002e-01 1.30483866e-01 1.83219030e-01 5.20642996e-01
3.85024369e-01 -2.40479007e-01 -5.46828881e-02 -4.74434316e-01
3.11650306e-01 1.38860017e-01 -8.80864799e-01 -5.81186228e-02
-3.48138869e-01 -1.03588149e-01 3.33311498e-01 -2.24238724e-01
-1.40691960e+00 2.31123075e-01 -1.92152783e-01 -8.96455288e-01
-8.82062763e-02 -2.82719016e-01 -4.50028777e-01 -3.09764862e-01
8.23091745e-01 6.01278059e-03 6.11506879e-01 -6.33482814e-01
4.20989364e-01 4.00356293e-01 6.77316606e-01 -5.91962397e-01
1.13143599e+00 9.73976612e-01 1.50177583e-01 -1.12956524e+00
-4.26389068e-01 8.63633119e-03 -6.65187061e-01 -1.75442338e-01
8.49413455e-01 -1.19325876e+00 -7.80551732e-01 6.60941303e-01
-1.26745105e+00 -2.56982207e-01 -2.20366150e-01 1.72142059e-01
-5.03942013e-01 1.92656398e-01 -6.13341272e-01 -7.53449261e-01
-3.28488171e-01 -1.14329875e+00 1.51659942e+00 3.39241102e-02
-1.58342063e-01 -7.31953263e-01 -1.92749679e-01 4.01748776e-01
3.42932224e-01 6.02789879e-01 7.66008258e-01 3.01657617e-01
-7.48529315e-01 -1.45309970e-01 -1.60374701e-01 6.08931184e-01
2.31068552e-01 1.92597017e-01 -1.52735913e+00 -5.25855064e-01
1.72691464e-01 -1.19968705e-01 4.96201724e-01 4.19700414e-01
7.08352864e-01 -5.94945192e-01 -6.25623092e-02 1.04180002e+00
1.32258964e+00 -9.27465037e-02 9.09595907e-01 -3.89071435e-01
7.81400084e-01 1.03433466e+00 8.99519995e-02 5.42794704e-01
1.28894061e-01 8.29458237e-01 7.08093584e-01 -4.78503048e-01
-6.30995095e-01 -2.09286705e-01 3.76952171e-01 2.16428936e-01
-1.00400761e-01 1.63545292e-02 -4.54468161e-01 2.48706177e-01
-1.20846033e+00 -7.32088923e-01 -1.86824929e-02 2.32243466e+00
8.40070605e-01 -4.07725215e-01 3.11287567e-02 5.03694871e-03
5.04538476e-01 2.39377260e-01 -4.57251012e-01 -3.97812426e-01
-1.00329727e-01 4.19679701e-01 2.34336749e-01 9.04089689e-01
-7.23745883e-01 9.01889861e-01 6.33032513e+00 5.46340525e-01
-1.20554686e+00 3.14123929e-02 9.19162750e-01 -3.61234576e-01
-6.43078804e-01 -3.41030856e-04 -4.46012825e-01 1.16164312e-01
4.04285371e-01 5.03770590e-01 8.81200552e-01 4.47211057e-01
2.48539805e-01 1.10715516e-01 -1.11028922e+00 1.05839467e+00
3.71408314e-01 -1.16913378e+00 -6.78059459e-02 4.93657142e-01
7.80852497e-01 -1.93372831e-01 2.52177954e-01 -3.30359519e-01
-4.38260771e-02 -1.31560385e+00 1.03909647e+00 4.05283988e-01
1.19349241e+00 -7.08333790e-01 6.03581145e-02 1.94354262e-02
-9.17072117e-01 5.70514984e-02 -7.06874803e-02 -8.69763866e-02
-5.01373503e-03 5.44758677e-01 -4.91952121e-01 2.87550747e-01
5.93289256e-01 2.62724012e-01 -2.75845170e-01 5.17178416e-01
-4.17696238e-01 8.82959813e-02 -5.32359600e-01 7.68785059e-01
-1.99922547e-01 -4.66301650e-01 6.13310337e-01 7.47678220e-01
2.62511224e-01 3.56588125e-01 2.20467104e-03 1.09187841e+00
-2.30687216e-01 -1.42692000e-01 -4.80998218e-01 2.38160297e-01
3.32958370e-01 1.37982571e+00 -5.33042729e-01 4.58063520e-02
-3.16782683e-01 1.10479641e+00 1.41170755e-01 7.30589926e-01
-7.02342331e-01 3.22588789e-03 1.32891953e+00 7.00656056e-01
3.81118834e-01 -3.26615989e-01 -3.03823888e-01 -9.48189676e-01
3.53919744e-01 -1.06153309e+00 -2.06659481e-01 -9.22697544e-01
-9.14470077e-01 6.71965003e-01 -1.63536176e-01 -7.50653684e-01
-1.81219220e-01 -6.01193666e-01 -6.01666749e-01 1.24148047e+00
-1.39424527e+00 -1.39287329e+00 -3.39479357e-01 6.95190072e-01
2.85430908e-01 1.71427354e-01 7.41529465e-01 3.81580859e-01
-3.29857886e-01 6.21972144e-01 -2.18962356e-01 -8.68927985e-02
7.78782547e-01 -7.99329996e-01 6.63326442e-01 8.09078753e-01
5.74978031e-02 6.10994160e-01 8.84880900e-01 -4.43361461e-01
-1.79812551e+00 -8.64731371e-01 7.44791090e-01 -6.13575280e-01
4.05344218e-02 -6.78368688e-01 -6.51190341e-01 6.51661575e-01
2.33493626e-01 3.43189269e-01 2.62161255e-01 -1.53923601e-01
-6.63226306e-01 -3.68034780e-01 -1.44508779e+00 8.93902779e-01
1.02403545e+00 -5.68433642e-01 -2.34007053e-02 2.77466774e-01
5.18370926e-01 -7.19929218e-01 -7.44671762e-01 2.38398939e-01
8.06659460e-01 -1.27025235e+00 1.04901814e+00 -1.92745984e-01
3.51171225e-01 -3.49640608e-01 -2.47451559e-01 -1.06940269e+00
-1.99800476e-01 -1.18689466e+00 6.39908388e-02 1.27202559e+00
1.54171735e-01 -6.23124540e-01 7.35422134e-01 8.99020374e-01
-1.38063163e-01 -6.76604807e-01 -8.26879025e-01 -4.25744593e-01
-1.23976484e-01 -2.80846894e-01 6.59806848e-01 7.21828759e-01
-5.28648913e-01 -1.03463000e-02 -3.89638990e-01 2.89716452e-01
7.51985312e-01 -1.11989595e-01 9.56651390e-01 -1.14022434e+00
-1.92144796e-01 -1.81971937e-01 -6.84816018e-02 -9.29892898e-01
6.39919490e-02 -5.03516197e-01 9.63319317e-02 -9.78353858e-01
-1.48330137e-01 -1.83715120e-01 5.29612362e-01 5.74374974e-01
-4.47967760e-02 6.79736197e-01 2.11093456e-01 1.29801065e-01
3.07182789e-01 5.28150201e-01 1.40679324e+00 2.50101238e-01
-2.60876358e-01 -3.23798954e-01 -8.85203958e-01 8.32219481e-01
3.09738249e-01 -2.32261062e-01 -4.76011097e-01 -8.81248236e-01
-7.16697192e-03 1.85349047e-01 5.07368147e-01 -8.77115548e-01
-1.35502011e-01 -9.60355923e-02 7.77904451e-01 -2.12823879e-02
9.96855795e-01 -7.81944335e-01 5.70929527e-01 1.70848206e-01
-4.22138348e-02 1.06674060e-01 3.78671914e-01 3.23155612e-01
1.63838975e-02 2.35489979e-01 9.59062159e-01 -2.21258905e-02
1.17820658e-01 5.93477666e-01 -5.15088700e-02 -7.43929297e-02
7.32592702e-01 -4.26589251e-01 3.13641168e-02 -5.84049046e-01
-5.36281586e-01 -1.21754415e-01 9.63571429e-01 3.55807662e-01
5.05498886e-01 -1.41213667e+00 -1.02905774e+00 9.22056675e-01
-2.58086085e-01 9.22834203e-02 2.52729625e-01 7.14190602e-01
-8.11160862e-01 -6.20445125e-02 4.81657498e-03 -5.85721791e-01
-1.51746655e+00 2.53084749e-01 4.84780043e-01 2.24794403e-01
-4.51537549e-01 1.10300410e+00 5.90973198e-01 -3.63373011e-01
1.99425638e-01 -4.25589643e-03 3.83393526e-01 4.03413028e-02
6.35453105e-01 1.31086916e-01 2.31164992e-01 -1.01181388e+00
-2.20920742e-01 7.90823579e-01 1.66435897e-01 -3.90272319e-01
1.35525489e+00 -1.12390660e-01 -2.74023026e-01 -3.93105716e-01
1.38185835e+00 4.43855792e-01 -1.99494791e+00 -1.51531417e-02
-6.22946203e-01 -8.71539712e-01 6.45861104e-02 -6.13309383e-01
-1.45338714e+00 7.89515436e-01 3.18464011e-01 -2.79672951e-01
1.44176245e+00 -2.41880625e-01 7.09583879e-01 -2.49564305e-01
1.41329661e-01 -6.96134806e-01 1.17071770e-01 2.47655064e-01
1.31893647e+00 -7.93486178e-01 2.38301352e-01 -6.95832312e-01
-2.42431417e-01 9.99315262e-01 2.89541900e-01 -1.00527532e-01
6.92219079e-01 4.71752316e-01 2.63154298e-01 -2.54398465e-01
-5.29309034e-01 1.55810937e-01 6.01780832e-01 6.43286586e-01
3.33520204e-01 -2.38617212e-01 2.12883830e-01 -6.06522663e-03
-2.39557490e-01 -3.20871711e-01 1.32906288e-01 6.04821563e-01
-1.13522150e-01 -1.03646708e+00 -8.29144537e-01 -3.42546143e-02
-4.27817017e-01 -1.23890214e-01 -3.20955336e-01 6.96113706e-01
2.82721132e-01 9.08452690e-01 2.06118435e-01 -1.26783252e-01
4.09028679e-01 1.23599870e-02 7.92267382e-01 -6.43699050e-01
-4.67778802e-01 6.42879009e-01 9.06749070e-02 -8.06086957e-01
-1.01546109e-01 -6.25585794e-01 -9.31911647e-01 -4.94243562e-01
-2.02446580e-01 -4.08760905e-01 7.67827392e-01 5.74410677e-01
6.37647688e-01 2.14415357e-01 7.88523197e-01 -1.43768179e+00
-4.25829858e-01 -7.42858827e-01 -6.84693694e-01 4.83944416e-01
5.87895811e-01 -5.46962321e-01 -4.57466304e-01 2.82905340e-01]
|
[12.82443618774414, -0.22173534333705902]
|
b2e89043-b462-4f73-b3e9-1a5832250183
|
randomly-projected-additive-gaussian
|
1912.12834
| null |
https://arxiv.org/abs/1912.12834v1
|
https://arxiv.org/pdf/1912.12834v1.pdf
|
Randomly Projected Additive Gaussian Processes for Regression
|
Gaussian processes (GPs) provide flexible distributions over functions, with inductive biases controlled by a kernel. However, in many applications Gaussian processes can struggle with even moderate input dimensionality. Learning a low dimensional projection can help alleviate this curse of dimensionality, but introduces many trainable hyperparameters, which can be cumbersome, especially in the small data regime. We use additive sums of kernels for GP regression, where each kernel operates on a different random projection of its inputs. Surprisingly, we find that as the number of random projections increases, the predictive performance of this approach quickly converges to the performance of a kernel operating on the original full dimensional inputs, over a wide range of data sets, even if we are projecting into a single dimension. As a consequence, many problems can remarkably be reduced to one dimensional input spaces, without learning a transformation. We prove this convergence and its rate, and additionally propose a deterministic approach that converges more quickly than purely random projections. Moreover, we demonstrate our approach can achieve faster inference and improved predictive accuracy for high-dimensional inputs compared to kernels in the original input space.
|
['Ian A. Delbridge', 'Andrew Gordon Wilson', 'David S. Bindel']
|
2019-12-30
| null |
https://proceedings.icml.cc/static/paper_files/icml/2020/4272-Paper.pdf
|
https://proceedings.icml.cc/static/paper_files/icml/2020/4272-Paper.pdf
|
icml-2020-1
|
['small-data']
|
['computer-vision']
|
[ 1.24404199e-01 9.30516049e-02 6.92299157e-02 -4.41601090e-02
-9.78957415e-01 -9.01472032e-01 7.50259399e-01 -2.18616068e-01
-2.81117737e-01 8.72820437e-01 1.01640271e-02 -4.95227814e-01
-3.68065476e-01 -1.08164835e+00 -8.89349878e-01 -1.23279774e+00
5.67846559e-02 9.10340965e-01 3.51883508e-02 2.20744565e-01
-2.68125031e-02 4.63152707e-01 -1.25784624e+00 -1.99881524e-01
8.10586572e-01 7.77481019e-01 -5.55191822e-02 9.67096031e-01
-5.99937178e-02 3.90082449e-01 -1.72281265e-01 -4.93368149e-01
3.33141059e-01 -2.98708124e-04 -4.44568634e-01 -4.04312849e-01
3.05657268e-01 1.25572935e-01 -4.23377991e-01 1.09890592e+00
5.16399860e-01 3.46108377e-01 1.03995514e+00 -1.24374735e+00
-1.08597219e+00 4.98011887e-01 -1.96370929e-01 -2.77845383e-01
-6.57443888e-03 3.26125056e-01 8.61939728e-01 -1.04667950e+00
1.90417185e-01 1.27273059e+00 1.12166965e+00 5.70658803e-01
-2.04330587e+00 -3.61662120e-01 2.05251984e-02 -6.92749739e-01
-1.18941319e+00 -8.96269903e-02 2.90333927e-01 -6.49906754e-01
6.85170054e-01 1.84438139e-01 3.88157606e-01 1.30754399e+00
1.48510098e-01 5.73392868e-01 1.07016850e+00 -1.43252239e-01
6.84610307e-01 2.21908659e-01 2.61939853e-01 3.52611303e-01
3.30620825e-01 5.71351126e-03 -2.08115175e-01 -7.61960685e-01
7.85905182e-01 4.08617198e-01 -4.21013743e-01 -6.51545346e-01
-1.25969779e+00 9.83361661e-01 1.71022207e-01 -1.08865008e-01
-4.46718395e-01 3.63667428e-01 6.64534094e-03 3.29664677e-01
5.81298709e-01 6.55243754e-01 -7.03870296e-01 -4.36833918e-01
-6.99859500e-01 4.46785569e-01 1.44260502e+00 9.65055406e-01
7.25486040e-01 -1.91792563e-01 -2.82271862e-01 6.88992500e-01
-9.37375352e-02 1.10497844e+00 2.13116959e-01 -7.99704552e-01
5.05076885e-01 4.08093184e-01 3.67194206e-01 -6.48684263e-01
-3.72705758e-01 -1.79269090e-01 -1.08479393e+00 1.25412837e-01
7.71084547e-01 -5.92288077e-01 -1.02582896e+00 1.78222227e+00
1.66634232e-01 6.26258105e-02 2.48831451e-01 5.20033240e-01
-7.88628459e-02 8.93696547e-01 2.18293592e-01 8.21668655e-02
9.98986423e-01 -6.44070029e-01 -2.85181195e-01 -1.49647459e-01
5.31617582e-01 -4.54778939e-01 1.49724746e+00 3.70405257e-01
-9.67610836e-01 -2.48877957e-01 -5.98166645e-01 8.58434364e-02
-4.73905832e-01 -2.35408768e-02 1.01497412e+00 8.77454102e-01
-1.19087648e+00 8.32220137e-01 -1.00406623e+00 -2.51867473e-01
5.62903523e-01 6.57315731e-01 -2.79805064e-01 -1.20887034e-01
-9.60972011e-01 7.92640805e-01 2.44476140e-01 -1.26859009e-01
-6.19288385e-01 -1.27366912e+00 -4.36254442e-01 1.85610101e-01
1.44651845e-01 -8.44287276e-01 1.05053377e+00 -3.16454977e-01
-1.60877395e+00 3.37429971e-01 -2.89380044e-01 -4.86051321e-01
5.81367612e-01 -4.85981703e-01 -4.41283397e-02 -2.07260013e-01
-1.77839711e-01 4.05231237e-01 1.14489472e+00 -9.01701927e-01
-3.55345219e-01 -4.84800488e-01 -3.04670110e-02 8.80402103e-02
-4.73154694e-01 -2.10087970e-01 -3.50051373e-01 -2.49219447e-01
1.45984769e-01 -1.35899007e+00 -5.77058375e-01 -2.55792239e-03
-3.02773088e-01 -1.10898003e-01 4.60816562e-01 -2.83930808e-01
1.06644881e+00 -2.14106011e+00 4.24290270e-01 3.00190181e-01
3.52999657e-01 1.15883686e-01 1.61196947e-01 3.64611149e-01
2.95448676e-02 2.34461382e-01 -4.92184252e-01 -4.06262696e-01
2.36356348e-01 4.38257813e-01 -8.02530944e-01 3.57068777e-01
4.83244330e-01 1.09819818e+00 -1.00410175e+00 4.59463932e-02
1.19255912e-02 4.93801028e-01 -5.27492404e-01 2.51286142e-02
-1.06105827e-01 1.77766308e-01 -4.37228590e-01 4.16354299e-01
6.57984674e-01 -5.51761985e-01 -2.74142593e-01 2.11524203e-01
3.24378282e-01 -1.57534368e-02 -1.23985708e+00 1.15922821e+00
-6.63596809e-01 4.23811883e-01 -2.19856367e-01 -7.63814032e-01
8.99305999e-01 2.24742964e-01 2.94887453e-01 1.85441971e-01
-3.02967280e-01 2.57392287e-01 -1.52224302e-01 -1.88317951e-02
2.65190721e-01 -4.55886424e-01 -1.60480052e-01 6.19214833e-01
2.26801991e-01 -3.64930272e-01 -1.00262292e-01 -1.01287372e-01
1.23047042e+00 6.11709729e-02 -7.04636946e-02 -3.89844686e-01
2.72186041e-01 -1.32556453e-01 2.48509407e-01 1.10830581e+00
1.66806012e-01 6.29823148e-01 1.06782377e+00 -5.74633837e-01
-1.29337120e+00 -1.55491734e+00 -4.25804764e-01 1.22012079e+00
-4.04135466e-01 -2.46764477e-02 -6.67095363e-01 -6.46071136e-01
6.16499722e-01 7.74648190e-01 -8.77845466e-01 -2.64888465e-01
-3.23200792e-01 -1.44010961e+00 6.86769783e-01 6.59833848e-01
-7.60413483e-02 -7.20251739e-01 9.36358571e-02 1.09204918e-01
5.36704421e-01 -8.09403718e-01 -2.33836219e-01 6.07732832e-01
-1.09201837e+00 -8.21234882e-01 -9.12571728e-01 -4.11912203e-01
7.57581115e-01 -1.48232847e-01 9.62916970e-01 -8.16842437e-01
1.99048012e-01 3.77572119e-01 3.42220217e-01 -4.26332653e-01
-3.36414307e-01 3.29955906e-01 3.55272323e-01 -2.22202614e-01
6.72068834e-01 -8.85542274e-01 -3.32245499e-01 2.40734100e-01
-6.95844471e-01 -2.42385507e-01 5.29865384e-01 9.68907595e-01
5.21995604e-01 -1.55260775e-03 5.12597501e-01 -1.04827774e+00
1.07877862e+00 -6.75590932e-01 -6.81742191e-01 1.45137921e-01
-5.49353182e-01 5.48026085e-01 8.75894368e-01 -9.46978748e-01
-7.77223885e-01 2.51634214e-02 2.63958186e-01 -6.25174940e-01
-8.85736849e-03 1.08944066e-01 2.46978682e-02 3.24759930e-02
1.05184722e+00 3.44454706e-01 8.70522335e-02 -3.91183555e-01
7.48373151e-01 4.86181766e-01 4.41934526e-01 -1.04434812e+00
9.66504335e-01 4.46523577e-01 4.20005798e-01 -6.14428520e-01
-8.40910316e-01 -2.11515546e-01 -7.46804416e-01 4.19665724e-01
5.73652208e-01 -8.44128966e-01 -7.26631701e-01 2.64267743e-01
-8.29971194e-01 -5.28347611e-01 -6.63175702e-01 5.76559126e-01
-8.45176101e-01 -1.45479754e-01 -6.11718237e-01 -8.66477072e-01
-3.08890462e-01 -9.02633190e-01 1.02664471e+00 4.74613532e-02
-2.75030613e-01 -1.35086036e+00 3.35725486e-01 -4.80506003e-01
4.97146338e-01 5.00185415e-03 1.15867507e+00 -8.94504845e-01
-3.34518611e-01 -4.58410621e-01 -1.95509821e-01 5.16909063e-01
1.52696714e-01 -8.22441354e-02 -9.33410943e-01 -1.03302628e-01
1.67349353e-01 -7.75161386e-02 7.98648059e-01 3.89379710e-01
1.13045192e+00 -3.07440281e-01 -3.63570333e-01 8.55454683e-01
1.29907310e+00 -2.20215499e-01 4.74331528e-01 -1.13576567e-02
8.61133993e-01 3.21076751e-01 7.97779635e-02 1.42675087e-01
1.13284197e-02 8.56358930e-02 -1.74412057e-01 1.08000472e-01
4.16048646e-01 -4.97367591e-01 3.43631268e-01 5.13445497e-01
-3.15419376e-01 1.14647903e-01 -1.17799437e+00 3.51779252e-01
-1.94578743e+00 -7.89963543e-01 -9.14090350e-02 2.72619843e+00
1.02187538e+00 1.98803470e-01 1.26801834e-01 -2.43661329e-01
5.41809499e-01 -2.13493690e-01 -8.78727138e-01 -2.72996455e-01
-1.81451142e-01 5.42467237e-01 9.35685337e-01 4.57119167e-01
-1.28405011e+00 7.84207821e-01 7.69322968e+00 6.85208499e-01
-9.70879912e-01 3.96203659e-02 5.85547090e-01 -2.47649476e-01
-3.73865902e-01 -2.27254145e-02 -1.10737169e+00 7.06148207e-01
1.27516818e+00 -4.52759624e-01 5.47230124e-01 1.41605103e+00
-1.22521698e-01 1.05462678e-01 -1.41150761e+00 9.39232647e-01
-4.01874095e-01 -1.07264197e+00 1.60915982e-02 3.64482075e-01
1.01804769e+00 1.40039489e-01 3.64214778e-01 6.34258091e-01
9.22127366e-01 -1.34585917e+00 -1.50054498e-02 7.19311059e-01
5.90806246e-01 -9.02343273e-01 5.80455959e-01 5.64490438e-01
-5.59526086e-01 -1.97604686e-01 -9.79037762e-01 -9.07592922e-02
-1.07611790e-02 7.70004332e-01 -8.03065956e-01 -1.51802478e-02
3.56650054e-01 3.55743796e-01 -3.36335540e-01 8.64928424e-01
-2.34763116e-01 8.05035055e-01 -7.89408028e-01 -2.61591405e-01
1.40334025e-01 -5.38248658e-01 4.76463377e-01 1.16544533e+00
6.27145827e-01 -5.38993739e-02 -7.62639940e-02 1.01623976e+00
-6.57991692e-02 -1.02349035e-01 -9.18215692e-01 -2.40714058e-01
4.72585738e-01 1.04594660e+00 -3.45295638e-01 -3.38129014e-01
-3.93840224e-01 8.21673155e-01 5.49572945e-01 7.65775025e-01
-6.57304227e-01 -3.05801600e-01 1.11922514e+00 2.32173447e-02
4.43910629e-01 -2.82325387e-01 -4.93618816e-01 -1.03234982e+00
9.18725133e-02 -5.22572875e-01 8.65811631e-02 -4.57314104e-01
-1.72364056e+00 1.77716210e-01 -5.47223613e-02 -1.00970745e+00
-3.34608465e-01 -8.70254219e-01 -5.63816667e-01 1.36853635e+00
-9.28330839e-01 -9.09076810e-01 2.36797675e-01 3.72872382e-01
-5.73571026e-02 2.12733960e-03 1.18359613e+00 -2.06135109e-01
-3.27339858e-01 4.48483109e-01 7.61787355e-01 -1.48148552e-01
6.50685549e-01 -1.74300480e+00 7.47041643e-01 4.56925720e-01
4.32133116e-02 9.75322664e-01 5.95481992e-01 -5.64936161e-01
-1.86054718e+00 -1.11168778e+00 5.57626665e-01 -1.10119808e+00
1.23178768e+00 -5.34575224e-01 -1.18428147e+00 8.72367859e-01
-6.19444788e-01 2.28803515e-01 8.50990176e-01 6.73902631e-01
-2.62418121e-01 1.55413777e-01 -1.01514351e+00 8.51757586e-01
8.75476360e-01 -7.24921584e-01 -5.50601840e-01 5.96823275e-01
8.81919503e-01 -3.46892029e-01 -1.22418272e+00 6.62111565e-02
4.75798100e-01 -3.59452963e-01 1.06675804e+00 -1.12142634e+00
3.66252840e-01 -3.02247196e-01 -1.75424501e-01 -1.44712389e+00
-3.44326943e-01 -9.55189943e-01 -6.40743613e-01 9.79354858e-01
7.04294741e-01 -1.00897694e+00 8.32319081e-01 1.26993787e+00
3.94968003e-01 -9.43383634e-01 -6.77561998e-01 -8.55690002e-01
5.97038269e-01 -6.56143308e-01 6.87586725e-01 5.70350349e-01
-5.53265326e-02 3.28611165e-01 -2.41428971e-01 3.57343376e-01
7.25995004e-01 6.29320145e-02 9.94934797e-01 -1.45525205e+00
-5.11093199e-01 -3.94415289e-01 -5.18850863e-01 -1.09385848e+00
-4.70650941e-02 -8.42207849e-01 -2.80457083e-02 -1.15036297e+00
1.04516774e-01 -8.56159806e-01 -2.34063670e-01 2.82655209e-01
-6.05739415e-01 -6.10714667e-02 -3.15587074e-02 3.71604919e-01
-2.17240795e-01 5.34429908e-01 1.09738088e+00 2.10382566e-01
-4.93862867e-01 5.67009568e-01 -7.74210870e-01 8.78478289e-01
8.11912656e-01 -4.75247115e-01 -4.73154575e-01 -1.02008335e-01
4.59440500e-01 -2.39345118e-01 3.12483430e-01 -1.02674043e+00
3.27217191e-01 -2.35330403e-01 7.58918464e-01 -3.71424437e-01
4.53569591e-01 -5.25847018e-01 1.74236163e-01 6.01154603e-02
-3.84644479e-01 -1.74177587e-01 1.26795948e-01 9.37682509e-01
2.00676307e-01 -1.67502552e-01 5.28703809e-01 6.05600439e-02
3.66912484e-02 5.35439432e-01 -2.01587498e-01 2.34114215e-01
9.72072303e-01 -5.78663275e-02 -1.86875775e-01 -2.96746135e-01
-9.38967049e-01 1.54756948e-01 7.72194624e-01 1.38584018e-01
2.90375412e-01 -1.38684809e+00 -4.09146756e-01 2.50709772e-01
-1.58228517e-01 2.63486147e-01 4.96514663e-02 8.82518709e-01
-3.88302989e-02 3.76806080e-01 1.35110945e-01 -7.15103149e-01
-6.44132972e-01 7.96558440e-01 1.05201237e-01 -3.20684433e-01
-7.27498055e-01 7.49954760e-01 3.87737542e-01 -7.67965257e-01
9.07983780e-02 -4.83171523e-01 2.55958855e-01 -3.19219567e-02
6.35844171e-01 3.82288903e-01 -2.68652648e-01 2.30308529e-02
1.63279131e-01 5.76498628e-01 6.77851886e-02 -1.62668869e-01
1.32997108e+00 2.54344106e-01 -1.18128121e-01 9.22706187e-01
1.15582037e+00 -1.12399757e-01 -1.73598731e+00 -3.32629323e-01
-1.64018303e-01 -4.36706066e-01 -3.63592297e-01 -2.98078924e-01
-3.93351585e-01 1.10891879e+00 2.79848009e-01 3.93079668e-01
6.25135124e-01 7.49786869e-02 3.49254251e-01 9.37599480e-01
2.56508112e-01 -8.94572437e-01 -4.37357813e-01 6.92636967e-01
5.39327621e-01 -1.00866771e+00 -1.48688853e-01 -2.66291678e-01
-7.45879829e-01 1.09850097e+00 1.34290069e-01 -3.85415405e-01
8.17877769e-01 3.81396472e-01 -4.55947727e-01 7.41343806e-03
-9.29781258e-01 3.86033091e-03 1.25698939e-01 1.01681352e+00
6.23458698e-02 3.63851994e-01 3.70610595e-01 5.18764853e-01
-4.59801912e-01 1.98464803e-02 2.03019604e-01 6.14491105e-01
-4.77561533e-01 -1.02839339e+00 -3.76179665e-01 8.90900910e-01
-3.45768422e-01 -3.74165148e-01 4.02553938e-02 7.08309650e-01
-4.39805746e-01 2.29097292e-01 2.17314899e-01 -7.10566267e-02
2.25401998e-01 6.65961504e-01 3.34143102e-01 -5.38987577e-01
-9.60305482e-02 -4.07415241e-01 -2.53572553e-01 -3.77846420e-01
2.37365589e-01 -7.98254669e-01 -8.33274126e-01 -4.98152763e-01
5.12110628e-02 8.71051662e-03 6.34662807e-01 6.94036007e-01
4.21153247e-01 9.46188644e-02 4.75198150e-01 -8.05640399e-01
-1.24989247e+00 -9.28629637e-01 -6.21906221e-01 1.98656499e-01
1.85629189e-01 -5.57647407e-01 -6.41544461e-01 1.61304623e-02]
|
[7.26861572265625, 3.8154168128967285]
|
6e30ca52-d6ee-4bb5-b19d-bfc4195f1908
|
unsupervised-contrastive-photo-to-caricature
|
2011.04965
| null |
https://arxiv.org/abs/2011.04965v1
|
https://arxiv.org/pdf/2011.04965v1.pdf
|
Unsupervised Contrastive Photo-to-Caricature Translation based on Auto-distortion
|
Photo-to-caricature translation aims to synthesize the caricature as a rendered image exaggerating the features through sketching, pencil strokes, or other artistic drawings. Style rendering and geometry deformation are the most important aspects in photo-to-caricature translation task. To take both into consideration, we propose an unsupervised contrastive photo-to-caricature translation architecture. Considering the intuitive artifacts in the existing methods, we propose a contrastive style loss for style rendering to enforce the similarity between the style of rendered photo and the caricature, and simultaneously enhance its discrepancy to the photos. To obtain an exaggerating deformation in an unpaired/unsupervised fashion, we propose a Distortion Prediction Module (DPM) to predict a set of displacements vectors for each input image while fixing some controlling points, followed by the thin plate spline interpolation for warping. The model is trained on unpaired photo and caricature while can offer bidirectional synthesizing via inputting either a photo or a caricature. Extensive experiments demonstrate that the proposed model is effective to generate hand-drawn like caricatures compared with existing competitors.
|
['Ran He', 'Aihua Zheng', 'Mandi Luo', 'Xin Ma', 'Yuhe Ding']
|
2020-11-10
| null | null | null | null |
['photo-to-caricature-translation', 'caricature']
|
['computer-vision', 'computer-vision']
|
[ 6.42157376e-01 1.28814518e-01 1.38903618e-01 -3.51861626e-01
-5.00744343e-01 -6.90337658e-01 8.01001906e-01 -6.05049074e-01
1.93375826e-01 6.21849895e-01 1.27586693e-01 -3.24346386e-02
2.34830841e-01 -7.49330401e-01 -1.02274823e+00 -5.26148617e-01
7.51731336e-01 4.18591321e-01 -1.64040431e-01 -1.86271101e-01
4.85459656e-01 8.98276687e-01 -1.29808092e+00 3.38288695e-01
1.03936863e+00 8.50643635e-01 1.70505464e-01 7.69379199e-01
-2.59776026e-01 5.61104178e-01 -6.43023670e-01 -7.56363511e-01
4.98564869e-01 -6.56090617e-01 -3.48865390e-01 5.32117486e-01
9.20886159e-01 -5.65570056e-01 -2.78044313e-01 1.01070130e+00
2.70097017e-01 -1.07302539e-01 9.63782132e-01 -1.36037898e+00
-1.26303482e+00 2.32339114e-01 -8.17045450e-01 -5.53918779e-01
4.12265241e-01 3.37777764e-01 5.84417999e-01 -1.12841320e+00
9.39314544e-01 1.70615685e+00 4.95815635e-01 5.63319266e-01
-1.42405701e+00 -5.10653079e-01 -5.20322286e-02 -1.82813227e-01
-1.47454476e+00 -2.57757425e-01 1.44456065e+00 -3.38027477e-01
1.53665602e-01 6.89670265e-01 8.40384841e-01 1.07375062e+00
6.91845194e-02 8.29981506e-01 1.18268275e+00 -3.75995159e-01
1.39647141e-01 1.81020796e-01 -6.82105899e-01 4.79220182e-01
-2.21442923e-01 1.70713380e-01 -5.56333549e-02 -1.50958046e-01
1.56602812e+00 1.87768891e-01 -3.19565892e-01 -3.37601215e-01
-1.29625189e+00 5.27500331e-01 5.18469036e-01 -3.92960384e-02
-2.71142244e-01 2.83261061e-01 9.93653312e-02 1.59390882e-01
4.63102549e-01 5.68270087e-01 9.02706310e-02 8.51304233e-02
-1.02013111e+00 5.05450845e-01 3.07826161e-01 1.33545852e+00
7.18916774e-01 2.88355708e-01 -3.13498497e-01 9.08374012e-01
2.14594588e-01 6.08205378e-01 2.32195318e-01 -9.88609970e-01
7.66362488e-01 6.81183636e-01 3.14577967e-01 -1.23431504e+00
3.02838504e-01 2.96788346e-02 -1.06989264e+00 7.30108202e-01
1.07306771e-01 8.08900595e-02 -7.33409464e-01 1.24211919e+00
3.15576434e-01 1.71086326e-01 -1.68003872e-01 1.10513449e+00
4.80845124e-01 9.48109090e-01 -1.08895175e-01 -1.15785629e-01
1.13008869e+00 -1.14111662e+00 -9.40236211e-01 -9.10491273e-02
1.64468169e-01 -1.39975202e+00 1.55265760e+00 3.49154770e-01
-1.53491640e+00 -7.64262795e-01 -1.14944816e+00 -6.94991946e-01
5.56560531e-02 6.64616525e-01 3.25499922e-02 2.69508302e-01
-8.11274469e-01 7.74107337e-01 -4.44769979e-01 1.41944647e-01
3.99596453e-01 -1.08934984e-01 -1.83384836e-01 1.65522143e-01
-7.76943386e-01 8.03988278e-01 7.11740404e-02 1.83640435e-01
-5.08911788e-01 -8.38597894e-01 -7.20108628e-01 -2.92626787e-02
-2.53501553e-02 -8.78028512e-01 8.13804686e-01 -1.38859570e+00
-2.03175497e+00 8.54787052e-01 8.62583518e-02 8.89275596e-02
1.25004649e+00 -8.19974616e-02 -3.32550973e-01 -8.46944153e-02
-2.88207203e-01 1.00609052e+00 1.38172317e+00 -1.69841826e+00
-3.66478473e-01 -1.67996123e-01 -2.07144573e-01 4.90941852e-01
-2.26798609e-01 -2.37286538e-01 -5.90544343e-01 -1.33834565e+00
1.64537638e-01 -8.26396644e-01 -1.00324256e-02 6.42398655e-01
-7.70115137e-01 1.51349396e-01 1.30309427e+00 -8.76490176e-01
1.01625836e+00 -2.09959507e+00 5.25926292e-01 1.49191320e-01
-7.71529377e-02 1.41529977e-01 -3.35056901e-01 3.68577987e-01
-1.79535747e-01 7.80127272e-02 -3.95190150e-01 -6.49875402e-01
-1.14098780e-01 1.34990618e-01 -5.47482491e-01 1.87737048e-01
6.13348007e-01 1.07008016e+00 -8.94739211e-01 -5.02348721e-01
2.83886671e-01 7.83436537e-01 -6.65338874e-01 4.61182684e-01
-2.79411614e-01 7.60797501e-01 -4.47697043e-01 4.43690211e-01
1.18699086e+00 2.99683422e-01 -2.52386928e-02 -4.74672407e-01
-2.25377128e-01 -2.12022632e-01 -1.20983672e+00 1.82140791e+00
-4.68297809e-01 6.04929507e-01 -1.35005772e-01 -4.19324845e-01
1.34854496e+00 1.32735908e-01 -9.02070031e-02 -3.90834838e-01
1.76071420e-01 3.31863672e-01 -4.02220428e-01 -4.80712771e-01
5.91290891e-01 -2.18087822e-01 1.94656208e-01 2.15879187e-01
-5.16001821e-01 -1.00570691e+00 -2.78822631e-01 -9.00260508e-02
3.08293283e-01 6.52629972e-01 -4.52565700e-02 -1.75926656e-01
7.78029859e-01 -1.11069502e-02 2.94612378e-01 9.10317600e-02
3.72684538e-01 1.28930104e+00 4.65868056e-01 -7.12472558e-01
-1.83112562e+00 -9.37220037e-01 9.43194479e-02 3.63786638e-01
2.53302097e-01 9.46559384e-02 -1.16701686e+00 -3.82296681e-01
-9.60586518e-02 8.18566680e-01 -5.94570816e-01 -8.58620256e-02
-9.34241414e-01 9.38657597e-02 4.62692201e-01 5.81352949e-01
5.76766491e-01 -1.07828319e+00 -7.73958638e-02 3.28468792e-02
-2.01458000e-02 -8.50912094e-01 -1.22952831e+00 -8.49391103e-01
-1.01247215e+00 -6.24889672e-01 -1.30864620e+00 -1.03222370e+00
1.11990821e+00 8.67819712e-02 8.56395841e-01 2.10146829e-01
-2.21160099e-01 -2.25637957e-01 8.11018571e-02 -1.42288879e-01
-8.14491987e-01 -3.43511164e-01 -9.39681530e-02 4.04210240e-01
-4.92421448e-01 -9.65537906e-01 -9.68750119e-01 4.59693789e-01
-1.26531422e+00 7.46435940e-01 5.67994118e-01 7.88003385e-01
8.06583583e-01 -3.54568809e-01 1.36326179e-01 -8.92144918e-01
6.68117344e-01 1.02186367e-01 -5.67082882e-01 2.13940546e-01
-3.59456986e-01 1.10922612e-01 1.04709411e+00 -8.32789898e-01
-1.27084672e+00 1.93780065e-01 -7.48433173e-02 -1.05569434e+00
-3.85698378e-02 -1.88822508e-01 -4.40210521e-01 -1.44658625e-01
5.79078078e-01 4.52799201e-01 1.91509277e-01 -6.11256182e-01
8.53430569e-01 5.55440366e-01 8.43523085e-01 -6.51183963e-01
1.34096026e+00 5.53361952e-01 -2.97561921e-02 -5.78794658e-01
-1.48259103e-01 3.75034630e-01 -7.68374503e-01 -1.56826377e-01
7.51865268e-01 -6.42568290e-01 -5.75018287e-01 4.72641855e-01
-1.56487691e+00 -1.24184713e-01 -4.08101231e-01 3.64504270e-02
-8.31128001e-01 5.08057833e-01 -4.77681905e-01 -6.59881055e-01
-3.64554286e-01 -1.24900854e+00 1.40123904e+00 1.81521878e-01
-1.61914960e-01 -6.95281982e-01 -6.72718883e-02 2.95929343e-01
2.62536943e-01 5.91938555e-01 1.06681013e+00 2.88485795e-01
-7.47390628e-01 -2.95120716e-01 -3.40506703e-01 5.49842298e-01
2.96851784e-01 5.26859522e-01 -7.89074779e-01 -9.74626765e-02
-2.86994219e-01 -2.00685859e-02 2.90638000e-01 1.07714662e-03
1.38350618e+00 -7.47688651e-01 -3.70343514e-02 8.29368412e-01
1.32549846e+00 3.10509503e-01 1.19572806e+00 -8.80057737e-02
1.02296412e+00 3.80082548e-01 5.17059922e-01 2.97230721e-01
4.64200154e-02 8.31064641e-01 2.99650759e-01 -2.72260040e-01
-4.77931619e-01 -1.00826681e+00 1.12831257e-01 7.94091403e-01
-3.83544058e-01 -1.49017768e-02 -2.07140550e-01 2.54246980e-01
-1.39222908e+00 -8.41286778e-01 -2.67849535e-01 2.25920081e+00
8.72772992e-01 -1.34535551e-01 -8.64003152e-02 1.26363620e-01
9.13609982e-01 1.19594209e-01 -5.99448800e-01 -6.27641201e-01
-1.97044045e-01 4.46004653e-03 1.63016215e-01 5.79700172e-01
-4.97113556e-01 1.00677991e+00 5.55633974e+00 1.27002847e+00
-1.34771729e+00 -2.29207098e-01 9.55103219e-01 1.01529211e-01
-9.30858314e-01 6.75287331e-03 -1.51419967e-01 5.72994471e-01
-4.81108613e-02 -3.68024893e-02 7.13458657e-01 8.24797273e-01
5.50030231e-01 4.48827863e-01 -1.15234625e+00 1.13234842e+00
9.29564908e-02 -1.56859660e+00 7.66843379e-01 -3.41110110e-01
1.09507751e+00 -8.99202883e-01 4.00660872e-01 -1.01755641e-01
-1.72085404e-01 -9.12853658e-01 1.18604016e+00 7.84081876e-01
1.29762256e+00 -6.43886387e-01 1.05283558e-01 2.94413179e-01
-9.68251288e-01 3.54069859e-01 -5.04201829e-01 3.53937373e-02
2.14469314e-01 1.91425487e-01 -4.18098241e-01 5.02647817e-01
1.50701240e-01 4.13284302e-01 -2.98570395e-01 6.78467333e-01
-3.60583127e-01 -6.07985305e-03 -1.92515664e-02 8.46839249e-02
1.63690358e-01 -7.71810651e-01 6.28967762e-01 8.35102618e-01
7.02666163e-01 2.47365639e-01 -1.75513193e-01 1.47865093e+00
-2.77108908e-01 4.24263895e-01 -6.00796402e-01 -7.82775879e-02
5.92134953e-01 1.33207345e+00 -2.54206091e-01 -5.37064075e-01
1.74551122e-02 1.62756133e+00 1.12288535e-01 4.58950669e-01
-9.09054339e-01 -5.21768630e-01 3.73167366e-01 2.95534015e-01
-7.28353336e-02 -3.14371809e-02 -6.96542203e-01 -1.04493737e+00
1.94319412e-01 -8.14434409e-01 -3.45293671e-01 -1.46323478e+00
-1.11274588e+00 8.02949846e-01 -3.01856279e-01 -1.72343647e+00
5.66355977e-03 -2.57526606e-01 -1.11142278e+00 1.16298425e+00
-1.27904260e+00 -1.63036156e+00 -5.32177448e-01 4.52403247e-01
8.04499209e-01 -4.38229591e-02 5.27665555e-01 2.41987079e-01
-2.64837950e-01 6.91976607e-01 5.91910891e-02 -2.81177368e-02
8.17588389e-01 -8.89329612e-01 7.82007396e-01 6.29488051e-01
-5.40018529e-02 3.80324125e-01 7.65171707e-01 -5.98399282e-01
-1.49808276e+00 -1.18073070e+00 6.72892690e-01 -4.31568772e-01
1.63744524e-01 -2.20828027e-01 -8.05813611e-01 6.60328925e-01
3.87330890e-01 -4.78761308e-02 -5.06570823e-02 -8.25565696e-01
-3.33030969e-01 5.55686429e-02 -1.38672352e+00 1.07586193e+00
1.20729029e+00 -3.21497351e-01 -4.91999686e-01 1.94945097e-01
7.06825435e-01 -7.38509953e-01 -7.47730434e-01 1.28665611e-01
6.38466299e-01 -7.86899149e-01 1.09717333e+00 -4.75654393e-01
1.21402895e+00 -5.18239796e-01 2.61865050e-01 -1.47021639e+00
-4.54154760e-01 -1.12483191e+00 2.82657206e-01 1.37180340e+00
2.77963847e-01 -2.78452665e-01 7.93743491e-01 6.86618328e-01
-2.53402591e-01 -6.51971042e-01 -4.75443512e-01 -7.65861988e-01
3.43170434e-01 3.97715569e-02 1.03262186e+00 1.07524061e+00
-3.81320059e-01 7.84266442e-02 -8.09793234e-01 6.19318932e-02
5.81358433e-01 2.21300989e-01 1.06604111e+00 -7.30414212e-01
-2.01635957e-01 -5.85839510e-01 -1.79656312e-01 -1.32779431e+00
-1.34243757e-01 -7.56860077e-01 -2.24336013e-01 -1.10391986e+00
-6.61825091e-02 -5.74556649e-01 4.82441723e-01 4.33055088e-02
-2.59051979e-01 4.88763392e-01 4.10201252e-01 4.82015342e-01
3.78144771e-01 8.21206212e-01 2.20847082e+00 -3.52841109e-01
-2.92828768e-01 -1.63661659e-01 -6.22592688e-01 7.83206999e-01
3.80217850e-01 -9.32026878e-02 -4.28225905e-01 -7.64026880e-01
4.07465994e-02 3.38353306e-01 3.56848121e-01 -7.47581840e-01
-1.43830508e-01 -4.44325119e-01 5.31433105e-01 -5.52886069e-01
4.81109500e-01 -7.22255290e-01 6.15274429e-01 2.66167045e-01
-6.41005039e-01 2.75769711e-01 -4.66869175e-02 3.87545586e-01
-1.69847444e-01 -8.15423727e-02 9.22740519e-01 -2.93530989e-02
-2.05854818e-01 4.71951216e-01 2.55236596e-01 -2.10871518e-01
8.47674012e-01 -2.71319181e-01 -9.20013338e-02 -3.53462219e-01
-5.71596026e-01 -1.81199342e-01 8.43734026e-01 5.24918318e-01
8.72171581e-01 -2.08929348e+00 -9.18674827e-01 5.53003907e-01
5.97126074e-02 5.91300540e-02 5.11818647e-01 3.56642693e-01
-1.11366975e+00 1.52828814e-02 -4.63739544e-01 -3.35417449e-01
-1.14096880e+00 7.22779632e-01 1.78911731e-01 8.00433978e-02
-7.26859093e-01 6.28320694e-01 4.53143597e-01 -4.01591122e-01
4.57565580e-03 -5.50057709e-01 4.63770144e-02 -4.66994852e-01
4.60748494e-01 3.94315004e-01 -2.95083672e-01 -4.63292390e-01
1.59510702e-01 9.41988945e-01 1.52982533e-01 -2.72070497e-01
1.01994538e+00 -7.58151188e-02 -5.76215014e-02 2.55895611e-02
1.31373823e+00 2.23344415e-01 -1.73842061e+00 -3.59790474e-02
-6.22040749e-01 -8.85649860e-01 -1.93378687e-01 -6.46983087e-01
-1.10141551e+00 1.01479447e+00 4.38181877e-01 -1.05953656e-01
9.87759948e-01 -5.01145244e-01 1.05072057e+00 -8.45062062e-02
1.02909982e-01 -9.84541118e-01 2.35217452e-01 -1.21713936e-01
1.77150857e+00 -8.30954671e-01 -2.07374960e-01 -7.46174812e-01
-7.93754101e-01 1.28396404e+00 4.54517722e-01 -6.16252959e-01
1.04346097e-01 1.43150762e-01 1.62207395e-01 2.08210081e-01
-3.40753585e-01 6.49322987e-01 5.75895965e-01 5.04003048e-01
1.65140361e-01 1.37295589e-01 -5.32324374e-01 3.25347036e-01
-3.07158172e-01 1.25372797e-01 4.37051088e-01 4.35923845e-01
-8.85812789e-02 -1.24284744e+00 -6.52322948e-01 -8.66384059e-03
1.42594516e-01 -2.04257384e-01 -5.31089842e-01 5.99295795e-01
1.17884144e-01 3.93834591e-01 1.76115051e-01 -3.37594360e-01
5.93898714e-01 -8.35251510e-02 5.80471992e-01 -1.57601669e-01
-4.56523180e-01 1.80740476e-01 -2.23396599e-01 -3.40466589e-01
3.74365188e-02 -2.53584415e-01 -9.85619366e-01 -4.53393102e-01
-5.08493222e-02 -2.67034441e-01 5.39170325e-01 5.55439651e-01
4.41049665e-01 3.64905834e-01 1.10814953e+00 -1.26741540e+00
-5.57643354e-01 -8.38524580e-01 -5.30014813e-01 1.01034367e+00
-3.57779711e-02 -3.48817855e-01 -3.68395239e-01 4.60184515e-01]
|
[12.116382598876953, -0.3787212371826172]
|
e7fa2663-e008-4ab9-b170-f6f0a39cbded
|
redi-efficient-learning-free-diffusion
|
2302.02285
| null |
https://arxiv.org/abs/2302.02285v1
|
https://arxiv.org/pdf/2302.02285v1.pdf
|
ReDi: Efficient Learning-Free Diffusion Inference via Trajectory Retrieval
|
Diffusion models show promising generation capability for a variety of data. Despite their high generation quality, the inference for diffusion models is still time-consuming due to the numerous sampling iterations required. To accelerate the inference, we propose ReDi, a simple yet learning-free Retrieval-based Diffusion sampling framework. From a precomputed knowledge base, ReDi retrieves a trajectory similar to the partially generated trajectory at an early stage of generation, skips a large portion of intermediate steps, and continues sampling from a later step in the retrieved trajectory. We theoretically prove that the generation performance of ReDi is guaranteed. Our experiments demonstrate that ReDi improves the model inference efficiency by 2x speedup. Furthermore, ReDi is able to generalize well in zero-shot cross-domain image generation such as image stylization.
|
['Lei LI', 'William Yang Wang', 'Xianjun Yang', 'Kexun Zhang']
|
2023-02-05
| null | null | null | null |
['image-stylization']
|
['computer-vision']
|
[ 1.31266654e-01 -5.48882931e-02 -3.93330485e-01 2.19552130e-01
-1.16179240e+00 -7.73077905e-01 8.36760044e-01 3.34334113e-02
-1.55878127e-01 9.23659980e-01 1.95969984e-01 -2.33323947e-01
-1.65250853e-01 -1.22758281e+00 -7.88649321e-01 -5.62347829e-01
1.05085358e-01 9.47649717e-01 3.74257028e-01 -5.83182648e-03
2.42885441e-01 4.64418828e-01 -1.18247926e+00 8.40148032e-02
9.92075145e-01 7.27601230e-01 4.32725251e-01 9.44259644e-01
-2.62090296e-01 8.09865594e-01 -5.62155247e-01 -5.65609217e-01
9.66151133e-02 -7.25292444e-01 -7.89012074e-01 6.61170706e-02
1.84186772e-01 -7.22382843e-01 -4.43137676e-01 1.07842362e+00
6.12912536e-01 3.51591706e-01 7.81382561e-01 -1.10470128e+00
-8.96841228e-01 7.32066333e-01 -4.82022822e-01 3.19634937e-02
4.33916718e-01 8.35014135e-02 7.84511387e-01 -1.02025902e+00
1.16346967e+00 1.08887649e+00 2.95437902e-01 6.86756849e-01
-1.04361808e+00 -7.04326868e-01 3.52188163e-02 -1.36984542e-01
-1.39911020e+00 -3.21659088e-01 4.47283208e-01 -2.73269415e-01
6.77638233e-01 8.54554027e-02 8.71457994e-01 1.21244788e+00
-1.01615347e-01 1.06931841e+00 9.10437405e-01 -1.10120557e-01
4.58637595e-01 -1.52478814e-01 -1.60135046e-01 1.02031863e+00
2.22135797e-01 1.48975238e-01 -6.96490824e-01 -5.08302033e-01
1.04390645e+00 -6.14305846e-02 -2.29569286e-01 -1.60017405e-02
-1.19211745e+00 8.99462104e-01 2.86825329e-01 -1.53041154e-01
-4.57153261e-01 5.66694438e-01 1.04469225e-01 2.31425345e-01
6.60688579e-01 3.03662628e-01 -3.48504633e-02 -5.65758169e-01
-1.34190857e+00 6.84997857e-01 9.01297092e-01 1.30072117e+00
8.79422903e-01 -8.03443119e-02 -6.18987978e-01 6.34187639e-01
-9.30386558e-02 9.46113467e-01 2.48170108e-01 -1.09095597e+00
3.92823905e-01 2.41816580e-01 3.03645194e-01 -7.49055564e-01
3.32622170e-01 -2.03581825e-01 -1.00119889e+00 -2.61332870e-01
3.48299295e-01 -3.11553627e-01 -1.00508904e+00 1.59387445e+00
5.20717144e-01 3.93970549e-01 -1.41283842e-02 7.28376210e-01
3.37361038e-01 9.78526831e-01 -1.13280751e-01 -4.11393374e-01
1.01389873e+00 -1.15807617e+00 -4.19110537e-01 2.23432049e-01
3.48555058e-01 -7.61641860e-01 1.02717602e+00 3.33920240e-01
-1.49843383e+00 -3.31400394e-01 -6.21190965e-01 -8.43838006e-02
2.18593851e-02 -9.45092004e-04 7.70473659e-01 3.98044258e-01
-9.64960575e-01 6.91683173e-01 -8.24538708e-01 -1.46240622e-01
6.31784439e-01 -1.12943016e-01 -6.55382592e-03 -5.44932783e-01
-1.05411363e+00 3.72377992e-01 2.37065360e-01 -2.47457132e-01
-1.34456050e+00 -9.59624469e-01 -4.27248389e-01 -1.13838902e-02
4.78023946e-01 -1.26123810e+00 1.38340223e+00 -5.84505618e-01
-1.68268681e+00 1.65187076e-01 -6.36092365e-01 -5.36578894e-01
9.68136132e-01 -3.16833347e-01 -4.59776558e-02 2.93340594e-01
3.26851875e-01 7.57215142e-01 1.12177014e+00 -1.15519798e+00
-6.12627447e-01 -5.72019704e-02 3.02449256e-01 2.97944784e-01
-2.09319025e-01 -3.44209552e-01 -1.04672146e+00 -7.09185004e-01
-1.87186450e-01 -1.17083144e+00 -4.72589046e-01 6.04041219e-02
-5.33321381e-01 -2.01541677e-01 4.66521144e-01 -3.69757146e-01
1.34901953e+00 -1.62978053e+00 2.02914655e-01 2.86140084e-01
1.80720299e-01 1.61957473e-01 -1.48585960e-01 7.01149046e-01
7.29354560e-01 3.71383369e-01 -2.93495238e-01 -4.11383867e-01
-6.27065375e-02 6.07548505e-02 -6.27610326e-01 1.51002361e-02
-1.51878387e-01 1.16054618e+00 -1.31293893e+00 -6.37259007e-01
-7.13690519e-02 2.55708575e-01 -6.65449858e-01 1.47422746e-01
-6.95215106e-01 2.85335094e-01 -6.35432959e-01 5.43588817e-01
6.96720183e-01 -5.98120630e-01 -1.56154960e-01 4.41439152e-02
2.64503092e-01 -1.23383710e-02 -1.00019610e+00 2.13001966e+00
-6.55927896e-01 5.51025093e-01 -4.45347041e-01 -2.60948330e-01
5.66598296e-01 1.58051655e-01 2.56941617e-01 -4.67843324e-01
-3.72297972e-01 1.35317862e-01 -5.17563820e-01 -3.21455270e-01
1.02532017e+00 -3.11719347e-02 3.30982544e-02 1.00158525e+00
-1.77744433e-01 -3.56491536e-01 3.86966109e-01 7.06714332e-01
1.06857014e+00 1.91702604e-01 -2.79577281e-02 2.59596944e-01
8.71032029e-02 1.93948522e-01 2.93181747e-01 1.10390210e+00
2.89025068e-01 7.14483321e-01 3.61766458e-01 -1.55865885e-02
-1.12540257e+00 -1.43920028e+00 3.37444484e-01 7.23319590e-01
4.00832206e-01 -7.22729385e-01 -9.22366560e-01 -6.64553821e-01
-9.59652737e-02 7.95870841e-01 -4.40001428e-01 -5.69782443e-02
-4.19781625e-01 -5.90709209e-01 7.27975905e-01 4.72588837e-01
6.42332554e-01 -9.18561816e-01 -2.81299144e-01 2.40326166e-01
-2.82126129e-01 -7.95087993e-01 -7.88259327e-01 -7.29301214e-01
-1.00717843e+00 -9.85744238e-01 -1.13324308e+00 -4.22081053e-01
8.09314966e-01 3.84307951e-01 1.03318465e+00 8.13667029e-02
-1.31368086e-01 2.94984132e-01 -1.73030436e-01 2.51950566e-02
-5.85384488e-01 3.25111598e-01 -3.42568070e-01 -3.21751058e-01
-5.68822250e-02 -2.67626345e-01 -7.42862284e-01 3.77126184e-04
-9.91643429e-01 2.07600534e-01 4.65463281e-01 9.16240990e-01
8.50873113e-01 2.52834380e-01 5.80663025e-01 -1.03137577e+00
1.21304226e+00 -6.52537644e-01 -6.66192114e-01 3.73507500e-01
-6.84181929e-01 1.49543211e-01 5.97708046e-01 -6.32193327e-01
-1.32869136e+00 -2.30433762e-01 1.71234505e-03 -6.18562758e-01
1.45392343e-01 6.52320921e-01 3.29386801e-01 4.27547276e-01
7.86464155e-01 5.56799412e-01 -2.79699918e-04 -2.75362283e-01
8.30697238e-01 3.44628692e-01 2.43797228e-01 -8.29349875e-01
8.73692811e-01 6.28967106e-01 -8.35981220e-02 -7.40857244e-01
-7.16427207e-01 -1.61197931e-01 -1.53407708e-01 -1.74824670e-01
5.15380204e-01 -9.44993734e-01 -5.57947636e-01 5.10892093e-01
-1.22892368e+00 -6.04282618e-01 -4.48814243e-01 3.14049780e-01
-5.75966597e-01 4.39406276e-01 -7.64367640e-01 -9.16013122e-01
-6.63678288e-01 -9.78042424e-01 1.16430974e+00 1.77057758e-01
-1.66734546e-01 -1.09275270e+00 1.82144001e-01 1.25108168e-01
3.73542398e-01 1.15835434e-03 8.95002246e-01 -1.41945660e-01
-1.28247225e+00 -1.72556639e-01 -2.59010077e-01 -1.87919158e-02
2.86720973e-02 1.24370657e-01 -4.67338115e-01 -2.36898333e-01
-5.27955830e-01 -2.76002735e-01 1.07822609e+00 2.89037168e-01
1.09606743e+00 -3.91165555e-01 -6.24805868e-01 4.84017640e-01
1.39194906e+00 -1.70499925e-02 6.15936577e-01 -1.85505599e-01
5.86890340e-01 1.40370488e-01 8.27849746e-01 5.12660801e-01
3.35085958e-01 3.37855190e-01 -4.23945449e-02 1.20398469e-01
-3.45237941e-01 -7.48740792e-01 3.71448487e-01 8.00147355e-01
-1.75524592e-01 -5.38409352e-01 -6.55282497e-01 8.71723711e-01
-1.98923314e+00 -1.24664330e+00 1.33910522e-01 2.29767108e+00
1.06541860e+00 -3.07337828e-02 1.19940810e-01 -3.34552944e-01
5.83244085e-01 1.90124661e-01 -8.38641346e-01 -6.54036626e-02
3.49350572e-02 3.29839140e-01 4.87665564e-01 7.08392799e-01
-5.33342600e-01 1.51333988e+00 7.07419872e+00 1.50581539e+00
-7.93706059e-01 1.97990850e-01 4.87987936e-01 -4.33886766e-01
-8.50921035e-01 1.73735708e-01 -8.08367610e-01 5.97937107e-01
8.13047528e-01 -6.60484135e-01 7.54203439e-01 8.46426427e-01
1.23578578e-01 -2.75967002e-01 -8.40667546e-01 8.14390242e-01
3.57409418e-02 -2.01106906e+00 4.99177098e-01 -3.52453664e-02
1.23388958e+00 9.06632096e-03 9.06559527e-02 1.30828038e-01
7.21640348e-01 -8.91663671e-01 5.38017511e-01 6.99985325e-01
9.10252154e-01 -9.47750628e-01 9.03978646e-02 6.33809328e-01
-9.81697440e-01 2.12995037e-01 -6.80404484e-01 2.60816693e-01
5.68809152e-01 9.58370447e-01 -1.00652039e+00 4.86508906e-01
1.95899576e-01 4.35771763e-01 -2.06612512e-01 8.85450840e-01
-3.66978437e-01 6.33040309e-01 -3.29918593e-01 -1.06749758e-01
1.76013499e-01 -3.77138317e-01 6.19349658e-01 1.09488344e+00
8.38272750e-01 7.25153312e-02 5.00932001e-02 1.18048656e+00
-2.98807323e-01 -1.90845340e-01 -7.73703277e-01 -2.60419995e-01
8.42866540e-01 1.04701614e+00 -6.19725585e-01 -8.58026564e-01
1.16618201e-01 1.54076564e+00 4.20513600e-01 5.92569172e-01
-1.06527662e+00 -2.79328167e-01 5.34825981e-01 6.11407831e-02
4.80354220e-01 -4.01586890e-01 1.49868250e-01 -1.22088492e+00
-1.39252022e-01 -5.35266340e-01 2.58552581e-02 -9.06201124e-01
-1.24429786e+00 6.27538085e-01 1.40265405e-01 -9.24271822e-01
-6.54809833e-01 1.37930736e-01 -4.03501332e-01 9.78106558e-01
-1.54200459e+00 -1.12956405e+00 -3.23891431e-01 6.88669324e-01
7.09522784e-01 -4.13035415e-02 7.78405964e-01 8.07505194e-03
-2.21558616e-01 4.60143656e-01 2.59830534e-01 -2.04817906e-01
5.74578941e-01 -1.25560391e+00 6.91048086e-01 6.82843804e-01
2.01818958e-01 8.47254276e-01 4.01076406e-01 -1.00362837e+00
-1.34942937e+00 -1.28493023e+00 9.02672052e-01 -1.09388433e-01
5.12540400e-01 -1.16081117e-02 -5.27227581e-01 4.39810008e-01
1.31398723e-01 -3.62064153e-01 4.64553833e-01 -1.29037365e-01
-2.73478597e-01 1.22727886e-01 -9.48752642e-01 1.08988416e+00
1.35673654e+00 -6.69325590e-01 -1.43314198e-01 5.80503583e-01
6.53536797e-01 -6.19581163e-01 -9.69192803e-01 -1.29185528e-01
5.80586791e-01 -8.24853182e-01 1.10712624e+00 -4.36237782e-01
7.67808437e-01 -1.80021971e-01 1.62262306e-01 -1.29985249e+00
-1.74055263e-01 -1.14585376e+00 -6.79626465e-01 1.17154479e+00
5.30132234e-01 -3.63613188e-01 9.00950432e-01 4.69336838e-01
2.82746941e-01 -8.00710678e-01 -5.38829327e-01 -9.17218626e-01
6.90026581e-02 -4.01391566e-01 8.05227995e-01 4.37695920e-01
-3.25863391e-01 4.13771123e-01 -5.99090993e-01 2.41808146e-02
8.76981318e-01 6.23223364e-01 1.08631742e+00 -9.15075123e-01
-4.85349774e-01 -2.53189534e-01 3.67814869e-01 -1.77518308e+00
-7.37223634e-03 -9.74839509e-01 -9.59494263e-02 -1.85191488e+00
2.11694553e-01 -7.74343848e-01 1.46046937e-01 -8.86266828e-02
-5.11510491e-01 8.10828581e-02 2.02441081e-01 5.17683804e-01
-5.79863608e-01 5.76904714e-01 1.78449297e+00 -1.19682893e-01
-1.33899733e-01 4.36968841e-02 -5.56095302e-01 4.95156169e-01
6.52419746e-01 -5.66333473e-01 -1.02480841e+00 -4.80323553e-01
4.24904019e-01 5.26246786e-01 1.86785370e-01 -7.10351586e-01
4.62198645e-01 -3.00865173e-01 2.12760434e-01 -7.46298373e-01
5.62580347e-01 -3.38383585e-01 4.63904321e-01 4.05831099e-01
-4.51007515e-01 -5.51884472e-02 -1.09586142e-01 1.04782557e+00
-3.22271069e-03 -3.70700121e-01 3.62156838e-01 -3.41728449e-01
-3.85565966e-01 8.96906197e-01 -2.60098785e-01 2.74925530e-01
9.53279734e-01 -1.94413513e-01 -2.49500394e-01 -6.64584816e-01
-6.03005588e-01 -2.25547403e-02 6.67411566e-01 1.59695238e-01
6.49287760e-01 -1.42165768e+00 -4.89795506e-01 -6.63055032e-02
-1.03200011e-01 2.35196397e-01 2.96755791e-01 5.02474368e-01
-4.70772564e-01 1.38904303e-01 2.94768661e-01 -4.62887615e-01
-9.11766291e-01 6.32363558e-01 -9.94422734e-02 -5.70911646e-01
-6.86749995e-01 8.94995630e-01 7.23363161e-02 5.40297255e-02
-4.47542667e-02 -1.80062577e-01 3.94549340e-01 -6.12175204e-02
6.24337733e-01 7.14289308e-01 -4.87232149e-01 -1.32385790e-01
1.19777679e-01 5.20968258e-01 -1.66622385e-01 -7.11320400e-01
8.91712725e-01 -1.35699376e-01 9.96769127e-03 2.33049616e-01
1.08675718e+00 2.14826062e-01 -1.40755153e+00 -1.75431579e-01
-3.41826797e-01 -7.53114164e-01 -6.53396249e-02 -7.26771355e-01
-1.02816713e+00 6.80503428e-01 -3.99131104e-02 1.00530311e-01
8.97893429e-01 -1.17362767e-01 1.40428710e+00 4.96766508e-01
5.31112254e-01 -1.14513481e+00 2.88775533e-01 3.58316153e-01
7.90054560e-01 -7.54729569e-01 7.29293078e-02 -3.96264642e-01
-7.79173076e-01 8.19015205e-01 2.13183850e-01 -1.46957695e-01
6.05528235e-01 1.54478550e-01 -5.04819989e-01 -6.39446378e-02
-1.02280056e+00 -4.50775661e-02 2.13670097e-02 6.80646360e-01
7.17766359e-02 -2.43101940e-02 -2.79675186e-01 1.73159108e-01
-1.32044777e-01 4.60648209e-01 4.50387895e-01 6.60107970e-01
-4.18299973e-01 -1.20183671e+00 5.51653579e-02 3.60867381e-01
-2.20884964e-01 -3.62199515e-01 -1.22652017e-01 5.26166737e-01
-2.20654577e-01 8.06053281e-01 4.48607616e-02 2.63329949e-02
-1.40354736e-02 -1.98532529e-02 7.52168119e-01 -3.92774045e-01
-3.68319184e-01 7.78345391e-02 1.74275130e-01 -6.08646989e-01
-2.68947184e-01 -5.46579957e-01 -1.27803922e+00 -7.89294839e-01
-4.43173885e-01 2.09862247e-01 5.48389673e-01 5.95172644e-01
8.63991022e-01 2.77951151e-01 4.80451256e-01 -4.74463403e-01
-4.92003471e-01 -7.65670300e-01 -4.50873792e-01 3.80308896e-01
1.00566342e-01 -5.15154183e-01 -2.49282476e-02 2.39374340e-01]
|
[11.143110275268555, -0.37346193194389343]
|
03baa82a-9a22-43e6-ab0b-5f95e2207980
|
coreface-sample-guided-contrastive
|
2304.11668
| null |
https://arxiv.org/abs/2304.11668v1
|
https://arxiv.org/pdf/2304.11668v1.pdf
|
CoReFace: Sample-Guided Contrastive Regularization for Deep Face Recognition
|
The discriminability of feature representation is the key to open-set face recognition. Previous methods rely on the learnable weights of the classification layer that represent the identities. However, the evaluation process learns no identity representation and drops the classifier from training. This inconsistency could confuse the feature encoder in understanding the evaluation goal and hinder the effect of identity-based methods. To alleviate the above problem, we propose a novel approach namely Contrastive Regularization for Face recognition (CoReFace) to apply image-level regularization in feature representation learning. Specifically, we employ sample-guided contrastive learning to regularize the training with the image-image relationship directly, which is consistent with the evaluation process. To integrate contrastive learning into face recognition, we augment embeddings instead of images to avoid the image quality degradation. Then, we propose a novel contrastive loss for the representation distribution by incorporating an adaptive margin and a supervised contrastive mask to generate steady loss values and avoid the collision with the classification supervision signal. Finally, we discover and solve the semantically repetitive signal problem in contrastive learning by exploring new pair coupling protocols. Extensive experiments demonstrate the efficacy and efficiency of our CoReFace which is highly competitive with the state-of-the-art approaches.
|
['Feng Wang', 'Youzhe Song']
|
2023-04-23
| null | null | null | null |
['face-recognition']
|
['computer-vision']
|
[ 3.37591201e-01 -6.42188266e-02 -9.35492143e-02 -7.44444251e-01
-5.10790706e-01 -4.05997276e-01 4.89621997e-01 -6.55922949e-01
-2.30573416e-01 4.50007766e-01 -5.40185571e-02 1.90266445e-01
-2.67920345e-01 -5.73180676e-01 -7.79142916e-01 -9.97184873e-01
2.09790036e-01 5.21425717e-02 -3.65334064e-01 -9.54024643e-02
1.18425153e-01 4.59877849e-01 -1.77010381e+00 4.21151966e-01
7.64395058e-01 1.43625271e+00 3.54070291e-02 1.26541004e-01
-9.15366113e-02 7.05507100e-01 -3.55004609e-01 -4.81299132e-01
5.18422484e-01 -7.90077567e-01 -6.31847799e-01 3.19944590e-01
6.89216793e-01 -2.87866324e-01 -3.54914010e-01 1.35062337e+00
6.52876258e-01 -1.98165439e-02 6.36362970e-01 -1.60137820e+00
-1.17451537e+00 3.61484110e-01 -6.92727506e-01 6.29965737e-02
2.07067087e-01 8.63666758e-02 1.06817317e+00 -1.31340075e+00
3.65459949e-01 1.34823298e+00 6.35506570e-01 9.03320312e-01
-1.24144125e+00 -1.06834507e+00 3.73700798e-01 3.87891710e-01
-1.65257978e+00 -8.41925919e-01 1.06774211e+00 -2.10852623e-01
2.57812202e-01 2.88324714e-01 4.15120900e-01 1.11811757e+00
-4.24211502e-01 6.26354635e-01 1.09213936e+00 -4.73963201e-01
-1.07853584e-01 2.09557548e-01 1.42208906e-02 1.01500070e+00
1.99731756e-02 3.83602887e-01 -6.15548790e-01 -9.90251750e-02
7.72363842e-01 2.10151300e-01 -4.41282213e-01 -3.16999912e-01
-9.16819692e-01 6.99692965e-01 4.03114110e-01 9.41313133e-02
-9.26250145e-02 2.42533498e-02 3.06668699e-01 7.19794273e-01
4.22709882e-01 1.78130955e-01 -1.96893841e-01 4.01440442e-01
-7.81894028e-01 -1.66472897e-01 4.89379764e-01 8.19678009e-01
9.23101187e-01 1.35601029e-01 -3.64729911e-01 1.01335967e+00
5.11537135e-01 5.05755067e-01 4.85636741e-01 -9.68215585e-01
1.94258869e-01 5.85529387e-01 -4.29845065e-01 -8.31731319e-01
-2.79351901e-02 -5.64724445e-01 -8.98770332e-01 3.48288566e-01
3.66358519e-01 1.58629343e-01 -7.04565465e-01 2.05428457e+00
2.76050836e-01 5.93012094e-01 4.13073637e-02 1.09449756e+00
7.84490347e-01 3.06853712e-01 -2.31036514e-01 -3.95174801e-01
1.31283605e+00 -9.43600595e-01 -6.65786088e-01 1.24175683e-01
4.22585875e-01 -6.24500155e-01 1.15685093e+00 2.57383019e-01
-8.92572105e-01 -7.09201694e-01 -1.36092508e+00 -2.71356869e-02
9.07156244e-03 2.51387417e-01 3.84802192e-01 7.10927248e-01
-7.28234708e-01 6.15939140e-01 -5.23930073e-01 9.27414671e-02
7.78297603e-01 6.13774478e-01 -5.72047889e-01 -1.23965226e-01
-1.09532034e+00 5.13905168e-01 4.85397056e-02 4.19864029e-01
-8.68067682e-01 -7.63736665e-01 -9.28146183e-01 -4.01222287e-03
2.20348626e-01 -4.76056755e-01 7.16309190e-01 -1.62728775e+00
-1.69729161e+00 1.09071779e+00 -1.23451471e-01 1.03767827e-01
3.84709686e-01 -7.22900406e-03 -4.66345400e-01 1.70737013e-01
1.17362142e-02 6.25692904e-01 1.29112661e+00 -1.42875373e+00
-3.07809323e-01 -5.13301969e-01 -8.92099366e-02 1.40290141e-01
-6.87925935e-01 1.01982549e-01 -3.17732960e-01 -5.67450762e-01
3.21516961e-01 -8.23775709e-01 2.05376223e-01 5.09846747e-01
-1.47127092e-01 -1.42970458e-01 7.92989612e-01 -4.28994954e-01
9.47691500e-01 -2.55800223e+00 1.08861357e-01 4.20898229e-01
2.18326464e-01 7.75200650e-02 -6.15883052e-01 -1.76323563e-01
-3.82456779e-01 -5.50070703e-02 -2.80175716e-01 -4.30837095e-01
-6.96898997e-02 2.61544824e-01 -4.57715362e-01 7.60151207e-01
5.85512877e-01 7.40515232e-01 -7.76300371e-01 -4.89219338e-01
-1.72330827e-01 7.05340385e-01 -8.31517339e-01 5.26231408e-01
1.22879177e-01 6.60527229e-01 -3.98445368e-01 6.09083533e-01
1.03814256e+00 -2.16361880e-01 1.84512839e-01 -6.71626151e-01
1.01121388e-01 1.78888723e-01 -1.29093790e+00 1.65932012e+00
-3.02466571e-01 2.03321174e-01 2.85002589e-01 -1.43573070e+00
1.06347632e+00 7.66272321e-02 4.24888581e-01 -7.44015217e-01
8.19709301e-02 3.36815119e-01 1.46737501e-01 -4.77035463e-01
-1.47344455e-01 -2.46452108e-01 3.73730332e-01 4.81407076e-01
3.18095595e-01 2.50136375e-01 -4.21305746e-01 -1.98391289e-01
7.14979410e-01 3.62495393e-01 -1.03577882e-01 -4.64929521e-01
9.69612479e-01 -7.14370847e-01 8.91875565e-01 3.86361629e-01
-3.53300154e-01 9.07645643e-01 3.40999633e-01 -3.17278713e-01
-7.66142786e-01 -1.02102005e+00 -4.90276128e-01 1.17666805e+00
3.69830608e-01 -2.08317488e-01 -8.50026727e-01 -9.93594825e-01
4.98103211e-03 5.74216805e-02 -8.02178681e-01 -5.62943459e-01
-6.39827192e-01 -7.11494386e-01 6.48749590e-01 4.67078388e-01
7.79059768e-01 -8.29536617e-01 1.02083519e-01 -2.15758443e-01
-5.06638214e-02 -8.34923506e-01 -8.79737198e-01 5.16919307e-02
-4.08289552e-01 -1.07348192e+00 -5.20041585e-01 -1.07885826e+00
1.09985626e+00 2.49859616e-01 7.23400414e-01 3.83285731e-01
-2.29233965e-01 4.80516076e-01 -1.28374889e-01 3.72542553e-02
-2.45470554e-01 -2.24233240e-01 2.64969856e-01 6.23026013e-01
3.20045620e-01 -7.72417307e-01 -8.01406085e-01 4.22529906e-01
-7.67749131e-01 -3.26571196e-01 4.45725411e-01 1.41672003e+00
5.60859740e-01 -3.21315736e-01 8.06343675e-01 -6.42915070e-01
3.85453463e-01 -3.55758965e-01 -4.38581884e-01 4.80390072e-01
-7.44824946e-01 2.96890974e-01 8.00357699e-01 -7.69141078e-01
-1.08188057e+00 5.73381856e-02 -4.09805365e-02 -9.21754897e-01
2.53699213e-01 2.62818523e-02 -5.85377038e-01 -6.12587869e-01
4.91402060e-01 4.42021072e-01 5.22806406e-01 -1.86938882e-01
3.74193072e-01 6.74595475e-01 4.58781332e-01 -8.51177514e-01
9.48700964e-01 5.67489803e-01 -2.29472835e-02 -6.35240555e-01
-9.96883929e-01 1.43382717e-02 -3.69544327e-01 -8.86370987e-02
4.65252966e-01 -9.80199456e-01 -1.03893232e+00 4.40568268e-01
-1.12688196e+00 1.64862722e-01 -4.86140460e-01 5.16717076e-01
-5.79293847e-01 4.29591864e-01 -6.09965384e-01 -5.96032619e-01
-2.64222562e-01 -1.13929033e+00 9.52656090e-01 1.18916258e-01
2.65621156e-01 -7.41285920e-01 -1.37830764e-01 2.57614881e-01
2.76765317e-01 -2.05106258e-01 6.18667066e-01 -6.11968338e-01
-5.32158375e-01 2.99355090e-01 -4.86042202e-01 7.89144933e-01
2.71269202e-01 -2.41870642e-01 -1.42512584e+00 -3.94074589e-01
4.54646796e-01 -6.42890751e-01 9.42041755e-01 -9.95642468e-02
1.46911502e+00 -4.15647537e-01 -1.63962841e-01 1.00277650e+00
1.14384258e+00 -4.81286906e-02 7.94158518e-01 1.52556924e-02
8.37104321e-01 9.02123749e-01 3.03811997e-01 3.99605989e-01
1.32431254e-01 5.96022487e-01 2.13370770e-01 -2.51435470e-02
-2.80477554e-01 -3.75661761e-01 5.75758576e-01 8.53652000e-01
-2.49202307e-02 2.41896287e-01 -2.77558416e-01 1.00237750e-01
-1.70651853e+00 -1.14742410e+00 4.90437806e-01 2.23191619e+00
1.11878121e+00 -2.03982532e-01 -1.61989689e-01 1.24248467e-01
9.26570833e-01 8.37710872e-02 -6.23337924e-01 -7.25306347e-02
-2.58048058e-01 1.36568785e-01 9.70320925e-02 4.72400576e-01
-9.47721541e-01 7.02319562e-01 6.05129051e+00 1.04806757e+00
-1.38698804e+00 1.56025052e-01 8.01188827e-01 -5.67554273e-02
-5.53078532e-01 -5.45121208e-02 -9.17623162e-01 5.62355578e-01
4.78099972e-01 2.80633532e-02 6.42516375e-01 6.82336390e-01
-1.30741969e-01 6.34603441e-01 -1.54450595e+00 1.31948900e+00
5.00311017e-01 -1.03717268e+00 1.94396392e-01 4.52288240e-02
5.14017761e-01 -3.96483630e-01 3.84231687e-01 2.97384679e-01
-2.14559168e-01 -1.13758206e+00 7.60748863e-01 5.64790428e-01
1.01063359e+00 -5.94803989e-01 3.19927692e-01 4.24511321e-02
-1.20221841e+00 -2.23015085e-01 -4.03277397e-01 9.09500644e-02
-3.31217527e-01 4.27524418e-01 -1.79152921e-01 3.80414456e-01
5.28454661e-01 8.35225642e-01 -4.10328060e-01 5.21926582e-01
-1.27703354e-01 4.23910707e-01 -2.61751860e-01 3.56131464e-01
-6.47116303e-02 -4.44944769e-01 4.41217005e-01 8.44398260e-01
2.67467231e-01 7.93646500e-02 2.71272302e-01 1.33798873e+00
-4.40324008e-01 6.06215149e-02 -6.89189434e-01 1.09121874e-01
5.99264503e-01 1.13044548e+00 -1.20717868e-01 -7.97153264e-02
-6.47279441e-01 1.20296764e+00 6.60629272e-01 3.38939726e-01
-8.03645968e-01 -2.70149976e-01 8.01163495e-01 1.48702031e-02
3.57672155e-01 1.41630515e-01 -6.02546930e-02 -1.24835157e+00
4.59115714e-01 -1.08316672e+00 2.82664299e-01 -1.78030506e-01
-1.64591873e+00 6.06838226e-01 -2.71471292e-01 -1.43470573e+00
9.30465162e-02 -5.72679162e-01 -6.53099298e-01 7.64232576e-01
-1.87606227e+00 -1.28548074e+00 -1.69848427e-01 8.58739734e-01
1.35855734e-01 -3.51707935e-01 7.65360534e-01 6.73899770e-01
-6.89224958e-01 1.35951936e+00 -3.58876884e-02 2.81511128e-01
9.25282776e-01 -6.80796206e-01 -3.11392218e-01 6.04112327e-01
1.31835714e-01 7.01821983e-01 1.18470930e-01 -1.30096436e-01
-1.51973939e+00 -1.02788675e+00 4.45380419e-01 -2.18482003e-01
4.89421397e-01 -5.45389235e-01 -1.15023398e+00 5.03368318e-01
1.92829128e-02 6.95440471e-01 9.33398068e-01 -3.54849943e-03
-1.00085509e+00 -6.65344477e-01 -1.33753133e+00 4.02264297e-01
1.56136024e+00 -9.02223825e-01 -4.45306480e-01 7.87120908e-02
7.40446031e-01 1.60408355e-02 -7.71991432e-01 5.39807498e-01
6.47644401e-01 -8.50759089e-01 9.23928916e-01 -5.90307653e-01
2.78037190e-01 -3.78216714e-01 -2.85048038e-01 -1.09805608e+00
-5.39028347e-01 -6.38218760e-01 -6.34302869e-02 1.59973872e+00
2.48343647e-01 -8.33562315e-01 6.60878062e-01 3.91780525e-01
-6.25513047e-02 -6.27823412e-01 -1.17380941e+00 -9.56687748e-01
1.58957019e-01 4.37629083e-03 6.36907756e-01 1.23166203e+00
-1.32523747e-02 3.18362147e-01 -4.27309841e-01 1.74269184e-01
8.76158178e-01 2.93095648e-01 3.46099794e-01 -1.20991123e+00
-3.76114011e-01 -4.58523154e-01 -5.57849765e-01 -1.14910650e+00
7.63502121e-01 -1.20364499e+00 -6.17885217e-02 -5.34713805e-01
4.79747355e-01 -6.02762461e-01 -6.03518069e-01 5.82904398e-01
-1.75805286e-01 4.16048616e-01 9.97813884e-03 4.70304400e-01
-5.54584742e-01 1.04484260e+00 1.51100087e+00 -4.23957735e-01
6.64158314e-02 -3.93039942e-01 -9.40277874e-01 4.34953690e-01
4.23282653e-01 -3.46610129e-01 -5.42155743e-01 -3.29671770e-01
-1.15088783e-02 -4.85351443e-01 4.28267479e-01 -7.21111834e-01
2.92584687e-01 -3.25608067e-02 4.83015776e-01 1.37117401e-01
2.58287340e-01 -8.69474351e-01 -7.97448531e-02 3.72469693e-01
-5.24083018e-01 -3.57162654e-01 -5.20924926e-02 4.13419396e-01
-3.73786271e-01 -2.62564778e-01 9.75334764e-01 1.50664732e-01
-4.43743676e-01 6.27738655e-01 2.84450144e-01 1.86216444e-01
8.75778973e-01 -2.35488862e-01 -3.44334215e-01 2.18186174e-02
-5.72589338e-01 3.78759831e-01 2.86160916e-01 4.17376012e-01
9.12551343e-01 -1.88107526e+00 -9.08466041e-01 8.96574318e-01
2.90189117e-01 -2.50451893e-01 1.05326280e-01 8.25722218e-01
4.59465086e-02 -2.83134431e-01 -2.97278017e-01 -5.95428169e-01
-1.07893300e+00 5.27526975e-01 6.93780601e-01 1.96545526e-01
-5.25687099e-01 9.26300645e-01 6.11785948e-01 -6.04224086e-01
2.85896897e-01 2.73336828e-01 -1.44994065e-01 2.51769591e-02
7.51450717e-01 5.72289675e-02 -8.89932066e-02 -6.37325406e-01
-5.13892531e-01 7.87858844e-01 -3.69519770e-01 1.01391181e-01
1.13404059e+00 -1.38419732e-01 -3.15290064e-01 2.18062609e-01
1.71244907e+00 -1.66097030e-01 -1.48284161e+00 -3.66613537e-01
-3.50159526e-01 -5.18078983e-01 -8.70222505e-03 -3.61144185e-01
-1.34101653e+00 8.02720070e-01 9.88480687e-01 -1.18882068e-01
1.19253409e+00 -8.21549147e-02 5.52721918e-01 4.21428323e-01
1.06425516e-01 -1.06558883e+00 3.33815396e-01 2.84329802e-01
1.10968375e+00 -1.50397766e+00 -1.35555252e-01 -6.37045205e-01
-3.75049978e-01 1.13874269e+00 8.60033512e-01 -1.92888483e-01
9.71690238e-01 2.18876243e-01 4.43728678e-02 -1.25857964e-01
-6.72960043e-01 -1.03823490e-01 3.25842291e-01 5.77891171e-01
4.05002147e-01 -2.71727979e-01 -2.59733260e-01 9.10866916e-01
1.76037401e-01 -1.26218185e-01 -3.62983905e-02 6.49558306e-01
-1.85093731e-01 -1.26724100e+00 -2.64470994e-01 2.05672979e-01
-2.78842747e-01 -1.04478166e-01 -2.91182429e-01 3.10233325e-01
4.13061023e-01 8.02743137e-01 2.09149599e-01 -4.94913548e-01
2.64497101e-01 1.30282253e-01 9.36028004e-01 -4.85219538e-01
-3.98600787e-01 -7.47756511e-02 -2.24870503e-01 -5.63946784e-01
-4.89466071e-01 -4.99034077e-01 -1.33761585e+00 -1.30793512e-01
-5.67031622e-01 1.84273198e-01 2.91884214e-01 7.97951698e-01
5.77940285e-01 2.50851691e-01 1.42899299e+00 -6.52762771e-01
-1.06237864e+00 -7.01489329e-01 -6.62594080e-01 7.91335344e-01
4.14050907e-01 -8.99753451e-01 -7.75710285e-01 8.74090493e-02]
|
[13.219524383544922, 0.5464556217193604]
|
a690903b-c917-4015-8589-433664273daf
|
detecting-vanishing-points-using-global-image
|
1608.05684
| null |
http://arxiv.org/abs/1608.05684v1
|
http://arxiv.org/pdf/1608.05684v1.pdf
|
Detecting Vanishing Points using Global Image Context in a Non-Manhattan World
|
We propose a novel method for detecting horizontal vanishing points and the
zenith vanishing point in man-made environments. The dominant trend in existing
methods is to first find candidate vanishing points, then remove outliers by
enforcing mutual orthogonality. Our method reverses this process: we propose a
set of horizon line candidates and score each based on the vanishing points it
contains. A key element of our approach is the use of global image context,
extracted with a deep convolutional network, to constrain the set of candidates
under consideration. Our method does not make a Manhattan-world assumption and
can operate effectively on scenes with only a single horizontal vanishing
point. We evaluate our approach on three benchmark datasets and achieve
state-of-the-art performance on each. In addition, our approach is
significantly faster than the previous best method.
|
['Menghua Zhai', 'Scott Workman', 'Nathan Jacobs']
|
2016-08-19
|
detecting-vanishing-points-using-global-image-1
|
http://openaccess.thecvf.com/content_cvpr_2016/html/Zhai_Detecting_Vanishing_Points_CVPR_2016_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2016/papers/Zhai_Detecting_Vanishing_Points_CVPR_2016_paper.pdf
|
cvpr-2016-6
|
['horizon-line-estimation']
|
['computer-vision']
|
[-1.66349262e-01 -2.59236008e-01 1.16121352e-01 -3.95167232e-01
-5.04720688e-01 -6.11021638e-01 8.34763765e-01 1.45741418e-01
-4.56257015e-01 1.45570427e-01 9.98629704e-02 -1.54769018e-01
1.51874840e-01 -8.53671908e-01 -6.84754372e-01 -4.74464655e-01
-1.28815398e-01 3.34247261e-01 7.75479198e-01 -6.15529954e-01
6.97683454e-01 5.49847901e-01 -1.22752154e+00 -6.53852448e-02
6.05133355e-01 9.17283475e-01 -3.15006584e-01 7.26972938e-01
4.96879429e-01 6.68121934e-01 -3.39364976e-01 -3.78138304e-01
7.73861766e-01 -2.85493433e-01 -6.81819081e-01 2.86066681e-01
8.05877984e-01 -4.54530925e-01 -1.69589818e-01 9.34775710e-01
3.50098580e-01 1.69548869e-01 5.12397051e-01 -7.73744524e-01
-1.17724538e-01 -7.75738358e-02 -7.90908337e-01 2.01979488e-01
3.01264733e-01 1.28133446e-01 1.22360992e+00 -9.74246860e-01
7.84987390e-01 7.66485572e-01 9.36110616e-01 -1.20530307e-01
-1.01666486e+00 -1.15141317e-01 1.38092339e-01 1.12661190e-01
-1.52384329e+00 -5.60765564e-01 7.29312241e-01 -3.74448746e-01
1.07582116e+00 3.41376394e-01 6.44772887e-01 4.54201519e-01
3.71911258e-01 6.16915166e-01 7.29054928e-01 -5.10344744e-01
1.79929659e-01 -3.59672815e-01 -1.23051904e-01 6.86006606e-01
2.23153681e-01 -1.03983343e-01 -3.18918735e-01 -1.45728961e-01
8.36253762e-01 -8.40561911e-02 -2.36441404e-01 -8.05017948e-01
-1.34036183e+00 8.39922011e-01 6.03370786e-01 6.13405183e-02
-2.05200478e-01 9.49043557e-02 2.62616187e-01 2.74703026e-01
4.43075508e-01 7.63549745e-01 -2.23156556e-01 1.81256846e-01
-1.18419039e+00 4.87259179e-01 7.62380481e-01 7.89762080e-01
7.69207418e-01 -2.01840073e-01 2.57432044e-01 4.53255385e-01
1.27277970e-01 1.79566577e-01 9.98137742e-02 -9.47892606e-01
2.65445977e-01 3.22064817e-01 3.75066608e-01 -1.20883369e+00
-6.29227400e-01 -5.61983585e-01 -4.16269600e-01 3.42255741e-01
5.13485014e-01 1.03489831e-02 -1.02096021e+00 1.12879658e+00
5.25797248e-01 2.69197106e-01 -8.54911208e-02 1.10880923e+00
3.00093323e-01 4.29222167e-01 -7.53179967e-01 6.50504082e-02
9.30335760e-01 -1.21446240e+00 -2.55318254e-01 -2.93770850e-01
6.36476338e-01 -1.13008881e+00 7.81006336e-01 5.23832321e-01
-7.17625678e-01 -1.12265185e-01 -1.30603182e+00 -1.97476268e-01
-2.17378929e-01 -6.08342551e-02 6.22840285e-01 4.77951974e-01
-9.61839139e-01 7.92287290e-01 -8.51241589e-01 -4.40064639e-01
-2.16867924e-01 2.44383559e-01 -1.44539937e-01 2.09660426e-01
-7.97547102e-01 8.44840765e-01 3.09196591e-01 1.54306099e-01
-5.56798458e-01 -1.69307753e-01 -8.44886839e-01 -1.45857602e-01
5.66476643e-01 -7.86197007e-01 1.38217020e+00 -1.02987564e+00
-1.46960151e+00 8.39556336e-01 -2.19641626e-01 -6.34666979e-01
9.52163517e-01 -6.06011868e-01 -4.14485484e-01 2.04923481e-01
3.75645757e-02 4.23457295e-01 7.75669396e-01 -1.28987765e+00
-7.69189358e-01 -6.01269826e-02 8.26286674e-02 3.45801443e-01
2.14223247e-02 1.05556406e-01 -9.19771492e-01 -4.62268800e-01
7.99031377e-01 -1.19277310e+00 -4.83439177e-01 -1.05313987e-01
-6.01340711e-01 1.41993929e-02 5.67451358e-01 -4.80888009e-01
1.06546581e+00 -2.02129602e+00 -1.47126958e-01 5.95687032e-01
3.04425716e-01 -6.38919100e-02 2.20993429e-01 5.61215103e-01
2.12504596e-01 -1.49358287e-01 -2.39236772e-01 -3.59883130e-01
-4.74903099e-02 -6.42950162e-02 -4.97090667e-01 9.98927057e-01
1.07415274e-01 4.22714412e-01 -9.53186393e-01 -1.25540167e-01
5.32110870e-01 2.59694815e-01 -7.08652079e-01 -1.01854570e-01
-3.50067168e-02 3.91867161e-01 -2.15148672e-01 5.14741659e-01
8.01769197e-01 -5.36982529e-03 8.94017518e-02 3.85775650e-03
-5.54577708e-01 5.18652141e-01 -1.41560149e+00 1.47531021e+00
-1.06597669e-01 6.93210185e-01 -2.79586434e-01 -5.10189176e-01
1.10856366e+00 8.45341459e-02 5.23549676e-01 -6.23583257e-01
8.25516209e-02 4.10628855e-01 1.76885203e-02 -2.60754973e-01
8.53890717e-01 1.29809722e-01 3.84076359e-03 -1.89918235e-01
-1.78923860e-01 -3.29856247e-01 1.27754211e-01 6.98008463e-02
1.10918713e+00 2.87166119e-01 5.32495320e-01 -3.98334205e-01
2.80022115e-01 -2.32981816e-02 8.63987982e-01 8.34121406e-01
-4.47481275e-01 1.13817930e+00 6.75949693e-01 -1.14017749e+00
-1.33552289e+00 -1.01158607e+00 -1.71188310e-01 7.54635513e-01
4.30082530e-01 -6.77501082e-01 -5.62783241e-01 -5.45451880e-01
-1.47542775e-01 5.60032189e-01 -5.41595578e-01 1.09014221e-01
-6.82591259e-01 -6.00230098e-01 2.29407668e-01 2.22962022e-01
6.88694000e-01 -6.23996079e-01 -1.01224625e+00 1.09310172e-01
-2.36121953e-01 -1.14817441e+00 -3.64818305e-01 9.79685485e-02
-8.72070670e-01 -1.06199515e+00 -3.62746179e-01 -5.22013187e-01
6.61567390e-01 5.40565729e-01 1.34922063e+00 3.10569759e-02
3.49595174e-02 1.62656663e-03 -3.44454497e-01 -3.36592555e-01
1.77825868e-01 5.56067098e-03 6.13101274e-02 7.08859712e-02
2.42150143e-01 -4.26428884e-01 -9.01188433e-01 4.87563938e-01
-6.13495290e-01 3.86693887e-02 3.97661418e-01 5.91588140e-01
5.94333351e-01 3.98652330e-02 -1.21475056e-01 -7.01127648e-01
3.54167223e-02 -3.71844113e-01 -9.89598513e-01 -5.39509095e-02
-3.08432192e-01 -1.29983768e-01 6.28218472e-01 2.26098448e-01
-7.08616614e-01 4.18888867e-01 -5.32286102e-03 -3.36774170e-01
-3.10936481e-01 4.76643801e-01 4.48436774e-02 -2.80940503e-01
7.22252131e-01 -1.31723762e-01 -7.06409276e-01 -3.86580825e-01
4.30804431e-01 2.04278663e-01 9.27296281e-01 -2.06266597e-01
1.05900347e+00 1.02083445e+00 1.52965769e-01 -1.19456553e+00
-7.50295937e-01 -9.94630754e-01 -1.13152075e+00 -1.50770187e-01
6.96083605e-01 -7.94914484e-01 -2.77448595e-01 4.44222003e-01
-1.10005271e+00 -3.21933515e-02 1.08953327e-01 5.21600664e-01
-4.06556338e-01 5.28649330e-01 -3.44412237e-01 -8.52381229e-01
-2.22205237e-01 -9.47557271e-01 1.15970111e+00 1.05240606e-02
-1.57267392e-01 -7.29049683e-01 4.30994302e-01 5.46828546e-02
1.26770854e-01 6.71149731e-01 1.57562688e-01 -4.37101603e-01
-7.84062624e-01 -5.46147645e-01 -1.35445492e-02 -9.15700570e-02
-5.97906299e-02 3.77087206e-01 -9.50976849e-01 -4.41680908e-01
1.25568852e-01 9.52868909e-03 1.29892051e+00 3.94134104e-01
7.03562856e-01 -1.90978676e-01 -5.70405088e-02 1.18114352e+00
1.45506167e+00 -4.99019846e-02 7.25592911e-01 1.00165164e+00
8.17487478e-01 5.45840323e-01 5.67123532e-01 4.64091212e-01
4.47613716e-01 8.21518958e-01 8.51137459e-01 -3.44453633e-01
4.96147662e-01 5.73974685e-04 3.29502374e-01 5.71040034e-01
-3.47120285e-01 -1.44803718e-01 -1.07592738e+00 7.30390787e-01
-2.13711238e+00 -8.90151501e-01 -3.67322624e-01 2.67137480e+00
2.25312747e-02 4.79534537e-01 1.36592239e-01 9.78894532e-02
3.06632578e-01 3.87790054e-01 -2.39387140e-01 -4.00754482e-01
-1.29640147e-01 -1.47412822e-01 9.03709352e-01 5.81884086e-01
-1.80367100e+00 1.12888622e+00 6.51351261e+00 9.20074359e-02
-1.42372930e+00 -3.20626497e-01 3.06055367e-01 -2.24205643e-01
8.69135857e-02 3.04159671e-01 -6.69705391e-01 -1.22974440e-01
4.99564379e-01 1.72513932e-01 1.81374744e-01 8.63049984e-01
2.35572726e-01 -2.67556399e-01 -9.44089949e-01 7.10952699e-01
1.61630601e-01 -1.25881577e+00 -4.27675247e-01 9.10276622e-02
9.75829899e-01 7.13470221e-01 -1.09626308e-01 -1.97055012e-01
2.87080675e-01 -6.93221211e-01 9.99460459e-01 3.93508762e-01
2.06594616e-01 -8.39088500e-01 7.90773332e-01 7.93178529e-02
-1.27067280e+00 2.77998805e-01 -4.42832083e-01 -3.53812099e-01
4.18145210e-02 6.59026146e-01 -9.46199000e-01 7.86199212e-01
8.99825573e-01 7.03525782e-01 -6.69723153e-01 1.63904488e+00
-5.00010848e-01 5.23053885e-01 -7.51271486e-01 2.06023932e-01
6.08108640e-01 -2.89525568e-01 7.39006460e-01 1.22432792e+00
3.36975336e-01 -2.63847500e-01 5.80050409e-01 3.83557767e-01
1.17391601e-01 3.18444997e-01 -8.74114692e-01 6.64466321e-01
2.37309426e-01 1.17564726e+00 -1.18767846e+00 -2.18550175e-01
-6.62026167e-01 1.17995751e+00 3.32560360e-01 2.76278645e-01
-7.45443702e-01 -5.54299355e-01 3.26017767e-01 1.60365224e-01
5.14963210e-01 -6.89543962e-01 -3.42663586e-01 -1.51263571e+00
3.98877054e-01 -6.58638597e-01 4.28402156e-01 -5.49895585e-01
-8.47003996e-01 4.85091060e-01 -2.56731778e-01 -1.47998846e+00
-1.67597026e-01 -5.25923014e-01 -9.50678408e-01 6.35734260e-01
-1.48144889e+00 -1.15098917e+00 -4.23937261e-01 4.14439201e-01
4.98761177e-01 3.39837104e-01 5.50640941e-01 3.90121490e-02
-4.19231057e-01 1.75232783e-01 5.52754879e-01 1.57071650e-01
9.93362725e-01 -1.51074946e+00 1.04454052e+00 1.38739550e+00
3.47495466e-01 7.78499186e-01 1.14444041e+00 -5.83871067e-01
-1.19408238e+00 -9.97602701e-01 9.35158491e-01 -6.00464344e-01
6.73309207e-01 -4.51600820e-01 -8.18943441e-01 8.99252772e-01
1.78222403e-01 1.06995299e-01 2.06234604e-01 3.48380864e-01
-5.40149868e-01 -3.68612469e-03 -7.56827772e-01 8.80802751e-01
7.39941955e-01 -2.52347767e-01 -5.70824921e-01 4.98052984e-01
5.35558224e-01 -8.52436900e-01 -3.01636219e-01 5.71616888e-01
5.70445001e-01 -1.48034072e+00 8.70167851e-01 -2.44149566e-01
3.13252360e-01 -7.13576615e-01 -1.08528391e-01 -1.24325752e+00
-4.65066254e-01 -8.51153851e-01 1.28499597e-01 6.13960683e-01
3.82985055e-01 -5.79517186e-01 7.17835248e-01 2.03064203e-01
-4.24946547e-01 -6.28701150e-01 -8.38152528e-01 -8.78301978e-01
-1.26424685e-01 -4.59972084e-01 3.38921845e-01 9.59514201e-01
-1.62776306e-01 1.94269583e-01 -7.34925032e-01 6.29160464e-01
6.16301954e-01 2.37499952e-01 1.26266074e+00 -1.16129160e+00
-3.28393221e-01 -4.86542672e-01 -5.84765553e-01 -1.04853594e+00
-3.66914958e-01 -5.62145829e-01 1.91656440e-01 -1.31551135e+00
-2.52449751e-01 -5.19550666e-02 -4.51001406e-01 2.47570619e-01
-6.79282174e-02 4.47019815e-01 1.56472638e-01 4.98130053e-01
-8.80363047e-01 3.73360276e-01 7.23706543e-01 1.99880123e-01
-4.81752843e-01 -5.53946868e-02 -2.34527290e-01 1.41008210e+00
8.77990901e-01 -2.09011242e-01 -7.57083446e-02 -6.07816458e-01
4.11339134e-01 -3.94936204e-01 4.72955137e-01 -1.42814326e+00
2.10535452e-01 -2.33249024e-01 5.87247133e-01 -7.93094695e-01
1.51453808e-01 -5.96292973e-01 -2.07859382e-01 3.24045688e-01
1.39434487e-01 2.48603538e-01 2.05444079e-02 3.45460445e-01
-1.09916620e-01 -1.28694251e-01 9.53178287e-01 4.37111259e-02
-9.61572886e-01 7.20691010e-02 -1.24878973e-01 -3.12264472e-01
8.64054739e-01 -9.81245786e-02 -2.91913182e-01 -5.01672089e-01
-3.37119192e-01 2.52448767e-01 9.73954976e-01 4.29856330e-01
5.61604261e-01 -1.03574908e+00 -5.29284179e-01 2.03275576e-01
3.90800208e-01 5.59346043e-02 -2.93889374e-01 9.64578986e-01
-1.16832697e+00 2.52672225e-01 9.67274308e-02 -7.04093277e-01
-1.06597841e+00 4.66895908e-01 6.15846217e-01 -1.79073155e-01
-1.04207158e+00 6.52632117e-01 2.94251323e-01 -4.05225456e-01
1.51887489e-02 -4.30351675e-01 -4.28332649e-02 -2.53871471e-01
3.47248942e-01 2.84113795e-01 3.65573794e-01 -8.77312660e-01
-3.60308290e-01 7.11521804e-01 -1.61343560e-01 -2.88610131e-01
1.22930062e+00 -7.64253289e-02 -8.10782462e-02 5.29524028e-01
1.01554561e+00 4.62787330e-01 -1.26357925e+00 -1.62911072e-01
2.37430781e-01 -6.98455691e-01 2.23828167e-01 -3.40951353e-01
-7.77218044e-01 6.45037770e-01 4.64298815e-01 1.27954334e-01
9.47010040e-01 -3.91915619e-01 6.78617895e-01 7.82561719e-01
2.19411448e-01 -1.19910514e+00 -1.55366078e-01 1.02247941e+00
8.04368079e-01 -1.36965907e+00 1.79582715e-01 -4.70081121e-01
-4.12097394e-01 1.31621754e+00 4.93383050e-01 -6.10971093e-01
4.16873455e-01 -1.46880046e-01 3.67639124e-01 -3.70523661e-01
-3.26601267e-01 -2.53529131e-01 4.03302014e-01 3.13688032e-02
5.54597676e-01 -2.77292319e-02 -2.86660433e-01 -1.79263696e-01
-4.76685971e-01 -3.02914530e-01 6.92938447e-01 1.17962062e+00
-7.91128933e-01 -7.81350732e-01 -6.97539508e-01 1.91268831e-01
-4.61704731e-01 -7.40054548e-02 -5.34103394e-01 8.10605109e-01
1.33875594e-01 8.66633356e-01 1.17129482e-01 -4.57923651e-01
5.07790208e-01 -2.04858035e-01 1.76359504e-01 -2.87219286e-01
-3.87983292e-01 6.36915386e-01 1.21817678e-01 -8.36592138e-01
-1.08961694e-01 -7.99140871e-01 -1.25363362e+00 -3.83054137e-01
-3.70350927e-01 -8.99806991e-02 4.88945335e-01 8.80983293e-01
1.56347603e-01 3.15728366e-01 8.52298915e-01 -1.07474971e+00
-2.20768571e-01 -4.90524948e-01 -3.69357228e-01 5.29203296e-01
6.80341065e-01 -5.13029099e-01 -3.69951487e-01 -1.71654716e-01]
|
[8.025334358215332, -2.038414716720581]
|
178be14a-c474-4c88-a0a8-2a6a1e3c5a19
|
dart-distribution-aware-retinal-transform-for
|
1710.10800
| null |
http://arxiv.org/abs/1710.10800v3
|
http://arxiv.org/pdf/1710.10800v3.pdf
|
DART: Distribution Aware Retinal Transform for Event-based Cameras
|
We introduce a generic visual descriptor, termed as distribution aware
retinal transform (DART), that encodes the structural context using log-polar
grids for event cameras. The DART descriptor is applied to four different
problems, namely object classification, tracking, detection and feature
matching: (1) The DART features are directly employed as local descriptors in a
bag-of-features classification framework and testing is carried out on four
standard event-based object datasets (N-MNIST, MNIST-DVS, CIFAR10-DVS,
NCaltech-101). (2) Extending the classification system, tracking is
demonstrated using two key novelties: (i) For overcoming the low-sample problem
for the one-shot learning of a binary classifier, statistical bootstrapping is
leveraged with online learning; (ii) To achieve tracker robustness, the scale
and rotation equivariance property of the DART descriptors is exploited for the
one-shot learning. (3) To solve the long-term object tracking problem, an
object detector is designed using the principle of cluster majority voting. The
detection scheme is then combined with the tracker to result in a high
intersection-over-union score with augmented ground truth annotations on the
publicly available event camera dataset. (4) Finally, the event context encoded
by DART greatly simplifies the feature correspondence problem, especially for
spatio-temporal slices far apart in time, which has not been explicitly tackled
in the event-based vision domain.
|
['Shihao Zhang', 'Bharath Ramesh', 'Garrick Orchard', 'Cheng Xiang', 'Ngoc Anh Le Thi', 'Hong Yang']
|
2017-10-30
| null | null | null | null |
['event-based-vision']
|
['computer-vision']
|
[ 1.04529308e-02 -4.44627434e-01 -1.13376021e-01 -1.58199981e-01
-8.98536921e-01 -7.62370586e-01 9.34496343e-01 3.73071283e-01
-6.35983288e-01 4.75843459e-01 -1.41123846e-01 2.31640771e-01
-4.92585599e-01 -3.13380361e-01 -6.48444593e-01 -8.78572881e-01
-4.20291543e-01 2.71001965e-01 5.65129519e-01 1.75989315e-01
9.71957669e-02 9.48365152e-01 -1.79236042e+00 -2.34072488e-02
2.67432690e-01 1.55707133e+00 8.15142244e-02 6.23766243e-01
4.29426461e-01 6.75568938e-01 -2.84427404e-01 -3.15865189e-01
6.19556010e-01 -1.97301954e-01 -2.12864935e-01 1.57758921e-01
6.44965768e-01 -1.08079724e-01 -3.56823832e-01 1.03888476e+00
5.11726320e-01 2.27479994e-01 8.68557811e-01 -1.45420074e+00
-1.36205286e-01 -2.10054934e-01 -5.91888428e-01 3.86740416e-01
2.48814717e-01 3.32381487e-01 9.46717322e-01 -1.06612682e+00
6.92712665e-01 1.10617542e+00 7.40311503e-01 -3.42638046e-03
-1.19164526e+00 -7.21148670e-01 -9.44247562e-03 3.69104534e-01
-1.70394933e+00 -1.73743606e-01 5.39227247e-01 -7.73764968e-01
6.53267562e-01 8.02541971e-02 1.03346837e+00 9.95811045e-01
3.53491873e-01 4.64278042e-01 9.54635978e-01 -2.34819934e-01
3.68211389e-01 -1.77931655e-02 3.15037146e-02 4.75445032e-01
4.16345000e-01 7.32714236e-01 -6.64289713e-01 -2.40499109e-01
6.60104513e-01 2.13629797e-01 -6.65024063e-03 -9.48468685e-01
-1.33045423e+00 8.77022922e-01 5.64645410e-01 1.01517722e-01
-4.59760070e-01 5.31264991e-02 5.26703715e-01 2.09037244e-01
2.93860286e-01 1.90117374e-01 -3.64482254e-01 3.20113264e-02
-8.97069216e-01 3.58174652e-01 4.89765525e-01 9.51144934e-01
6.47741079e-01 1.85644418e-01 -4.78151858e-01 3.02927524e-01
4.04073447e-01 7.41776049e-01 2.54299343e-01 -5.49585342e-01
6.95332140e-02 4.81908441e-01 2.14328215e-01 -9.41243947e-01
-5.46165109e-01 -5.15318930e-01 -6.36441767e-01 5.28384805e-01
5.89366496e-01 4.20673825e-02 -6.13447845e-01 1.64589095e+00
7.34084308e-01 5.44960856e-01 -3.28943990e-02 9.21407461e-01
6.25084996e-01 1.72588184e-01 1.74088866e-01 -3.87979418e-01
1.75588882e+00 -2.96980292e-01 -3.01554024e-01 -1.09694928e-01
3.63164335e-01 -6.15038633e-01 1.88058466e-01 2.67214358e-01
-4.86299843e-01 -7.26385713e-01 -1.04043293e+00 3.48699659e-01
-5.01163840e-01 1.98081404e-01 5.44060469e-01 4.24066454e-01
-6.39555573e-01 3.92871231e-01 -8.63315463e-01 -6.80111587e-01
5.56788027e-01 3.38133633e-01 -5.40570080e-01 8.80572274e-02
-9.32339787e-01 8.55724990e-01 6.57807946e-01 -1.82610318e-01
-1.07280290e+00 -7.38192320e-01 -9.01183546e-01 -1.92737222e-01
3.46128672e-01 -3.64969671e-01 8.33505571e-01 -6.53108537e-01
-1.13182855e+00 8.98663700e-01 5.32552265e-02 -6.78515911e-01
5.83885908e-01 1.59903973e-01 -3.78554642e-01 1.90020725e-01
2.16213167e-01 6.58991158e-01 1.11671162e+00 -8.14242482e-01
-9.22158241e-01 -4.54407394e-01 -2.71537602e-01 1.04442634e-01
-1.02233976e-01 1.54525712e-01 -3.20915967e-01 -8.87414038e-01
4.49406281e-02 -9.74713385e-01 -2.87509132e-02 4.85307276e-01
-1.40415773e-01 -2.23665729e-01 9.73499537e-01 -3.80619317e-01
6.52795196e-01 -2.44619727e+00 -6.75557479e-02 1.99865788e-01
3.78804952e-02 1.95379600e-01 -2.57636961e-02 1.66980863e-01
-2.43946195e-01 -7.53748238e-01 1.35693401e-01 -6.07389696e-02
-1.23835459e-01 -2.75673922e-02 -3.13502043e-01 1.02955222e+00
3.40567738e-01 9.00851071e-01 -8.83345366e-01 -6.30044103e-01
5.43797314e-01 4.28338349e-01 -2.54179925e-01 -5.42222746e-02
6.37680516e-02 6.65992737e-01 -4.31891203e-01 7.19422042e-01
5.88076472e-01 -2.41580769e-01 -2.01428354e-01 -4.90007937e-01
-3.64036202e-01 -3.01833242e-01 -1.46100211e+00 1.55346227e+00
1.39008462e-01 6.22736990e-01 -2.44647682e-01 -9.06266928e-01
8.47526014e-01 3.08597833e-01 8.74106288e-01 -7.29144096e-01
1.54685184e-01 4.53893244e-02 -1.53841376e-01 -2.94764757e-01
3.16122204e-01 4.14788872e-02 -1.42412364e-01 1.60944667e-02
4.83097702e-01 2.62516379e-01 1.53810859e-01 4.44694646e-02
1.20621669e+00 1.47099435e-01 7.01847732e-01 -3.05888385e-01
3.96171451e-01 1.50962546e-01 6.86995149e-01 8.59099925e-01
-3.35552752e-01 5.25624871e-01 1.32451952e-01 -3.78116250e-01
-8.44326973e-01 -1.20194793e+00 -6.27767384e-01 8.53588581e-01
2.00466216e-01 -2.64675498e-01 -2.52168179e-01 -5.39278865e-01
3.46946120e-01 3.04287583e-01 -6.81080163e-01 -1.98356658e-01
-2.90142953e-01 -6.15881920e-01 4.31798816e-01 6.63096607e-01
3.03359985e-01 -6.75498009e-01 -1.26547611e+00 3.15275371e-01
2.11655080e-01 -1.38107491e+00 -4.58778858e-01 4.81852561e-01
-7.16895938e-01 -1.32113695e+00 -6.60020590e-01 -3.57112497e-01
2.84876704e-01 3.57935935e-01 6.87288761e-01 -4.84477282e-01
-8.51938725e-01 9.99635100e-01 -3.77036989e-01 -5.37749946e-01
2.43013829e-01 -5.90365589e-01 4.09073859e-01 3.35349560e-01
4.15029824e-01 -3.93943191e-01 -6.52983606e-01 4.88563448e-01
-5.07472873e-01 -3.65532935e-01 5.84451437e-01 8.19915533e-01
8.74822497e-01 -7.40841553e-02 4.13829535e-01 -5.48806265e-02
-2.08658367e-01 -3.38195890e-01 -1.20547688e+00 2.48003900e-01
-3.36520195e-01 -1.53015211e-01 2.30479851e-01 -9.61476326e-01
-6.87940657e-01 4.26117629e-01 4.74704146e-01 -8.85292888e-01
1.59482453e-02 1.09598137e-01 -6.34310618e-02 -5.37456155e-01
7.31464684e-01 2.27361009e-01 -5.16701266e-02 -2.40322307e-01
1.89770326e-01 2.77813822e-01 7.32247114e-01 -3.86706203e-01
9.82728362e-01 8.44572008e-01 3.63659918e-01 -8.27172339e-01
-7.53857076e-01 -7.92845547e-01 -8.14744592e-01 -3.65702242e-01
1.13086760e+00 -1.28335428e+00 -8.20395589e-01 4.25025672e-01
-1.00084507e+00 1.15998477e-01 -7.83808529e-01 1.10017443e+00
-7.39144683e-01 2.07124978e-01 -6.12039417e-02 -1.02154279e+00
-1.20410740e-01 -9.63692367e-01 1.25276256e+00 4.25531358e-01
6.37924671e-02 -7.43429661e-01 9.12177116e-02 -2.73140758e-01
1.76725000e-01 6.15792394e-01 4.02450860e-01 -9.38729286e-01
-7.65339494e-01 -4.94867116e-01 -3.87005210e-01 1.14280060e-01
-1.88534170e-01 -2.00350940e-01 -1.01186740e+00 -6.53527617e-01
1.28132822e-02 -2.60332435e-01 5.47571540e-01 5.31689823e-01
7.43984163e-01 2.08530262e-01 -4.98104692e-01 7.46618867e-01
1.42009890e+00 2.96718687e-01 1.92368060e-01 2.80597717e-01
4.94468451e-01 2.64470994e-01 9.12383914e-01 8.25045586e-01
3.12643766e-01 1.02771604e+00 4.53255951e-01 1.14699051e-01
-2.87523955e-01 -3.88059467e-02 5.03348768e-01 -4.07177545e-02
1.17477238e-01 1.49066165e-01 -7.57988155e-01 5.31243384e-01
-2.07653952e+00 -1.02869022e+00 -1.20126255e-01 2.56036878e+00
2.81864703e-01 2.11567599e-02 1.90857306e-01 -1.33874014e-01
8.04659307e-01 1.04449559e-02 -5.79229951e-01 6.05078876e-01
-2.17359409e-01 1.52523056e-01 7.67984152e-01 -1.96480900e-01
-1.41928911e+00 5.81786811e-01 5.29659128e+00 7.85415113e-01
-1.16159511e+00 3.28453749e-01 -4.39116284e-02 -1.24223858e-01
8.14997137e-01 2.37376139e-01 -1.24126875e+00 5.38388371e-01
8.37109029e-01 -2.72009760e-01 1.10050313e-01 7.90665746e-01
9.00188461e-03 -2.20156252e-01 -1.35274303e+00 1.30857098e+00
1.84165701e-01 -1.14378035e+00 -3.89630377e-01 1.82192385e-01
3.49682212e-01 2.96344459e-01 -2.57168841e-02 3.42599124e-01
3.12988274e-02 -5.39770246e-01 1.04825842e+00 7.31850207e-01
8.50362122e-01 -6.15244508e-01 4.30331528e-01 2.14161336e-01
-1.66938639e+00 -2.99969465e-01 -3.13979149e-01 4.69220519e-01
2.59128213e-01 4.46263134e-01 -6.33916557e-01 9.19935048e-01
9.61212695e-01 1.02963948e+00 -7.68495381e-01 1.62774563e+00
1.54249698e-01 2.93089330e-01 -6.37592018e-01 4.00577188e-01
4.31559533e-02 -5.09847254e-02 1.00599766e+00 1.08888185e+00
4.13962334e-01 -1.16427749e-01 5.22189617e-01 6.67723060e-01
2.48018146e-01 -1.87698811e-01 -5.69584846e-01 2.51133651e-01
4.06680882e-01 1.44974291e+00 -8.82621050e-01 -1.28359154e-01
-4.21511978e-01 5.77054024e-01 -5.15412875e-02 2.48052895e-01
-9.39517617e-01 -2.37706900e-01 5.32903552e-01 9.21620876e-02
1.07635188e+00 -1.40256360e-01 2.52442509e-01 -1.11060381e+00
1.03370717e-03 -6.26748204e-01 5.79060853e-01 -6.79906249e-01
-1.51607013e+00 2.98974395e-01 4.30015683e-01 -1.81814992e+00
-2.79559642e-01 -6.69588029e-01 -5.30875504e-01 5.96156716e-01
-1.43527472e+00 -1.43304098e+00 -4.26186204e-01 9.52490747e-01
2.67459124e-01 -4.38805431e-01 4.96535271e-01 5.31566024e-01
-2.85582781e-01 4.68541950e-01 -4.55532745e-02 2.89981693e-01
8.01140785e-01 -1.00361419e+00 -8.96180794e-02 8.05785954e-01
1.94169387e-01 2.56498754e-01 6.60854995e-01 -6.91159248e-01
-1.60010076e+00 -1.51127553e+00 3.38583320e-01 -5.27540505e-01
8.80577385e-01 -4.54496771e-01 -7.34474182e-01 5.59276700e-01
-3.57527792e-01 9.81626809e-01 4.54453737e-01 -2.52480417e-01
-4.11921024e-01 -2.74622738e-01 -1.14250612e+00 -4.22055051e-02
7.74577260e-01 -5.55821776e-01 -4.91554081e-01 5.20115733e-01
2.77423710e-01 -4.45676506e-01 -9.60792243e-01 4.07295078e-01
6.34961307e-01 -5.45970500e-01 1.19084394e+00 -4.63125676e-01
-5.61856449e-01 -7.60329306e-01 -4.37356204e-01 -8.22606146e-01
-3.17764282e-01 -3.37652087e-01 -2.82136649e-01 1.26484215e+00
-3.35431635e-01 -5.66592813e-01 4.32809383e-01 7.01184431e-03
9.30240974e-02 -2.59022325e-01 -1.62061834e+00 -1.33713269e+00
-4.21525747e-01 -4.41462755e-01 1.60290748e-01 6.82156265e-01
-6.32139504e-01 1.81475505e-01 -2.24656194e-01 3.96676064e-01
1.22678638e+00 7.53458356e-03 6.85783982e-01 -1.71267712e+00
-4.18563098e-01 -1.27810851e-01 -1.12665772e+00 -7.30004430e-01
3.65977325e-02 -9.60431635e-01 5.39350659e-02 -7.91832745e-01
2.39549994e-01 -3.85529429e-01 -3.98497611e-01 3.68622333e-01
2.14002341e-01 3.08223963e-01 3.10600609e-01 3.20139587e-01
-9.41545546e-01 5.98895133e-01 6.47314548e-01 2.66744085e-02
1.88072860e-01 1.76684689e-02 -1.45548061e-01 6.61436737e-01
3.37611698e-02 -8.22248399e-01 -6.62048087e-02 2.51737535e-01
-1.09577425e-01 3.42504203e-01 1.20910156e+00 -1.32844758e+00
5.54121733e-01 2.20840778e-02 6.94359124e-01 -9.26419377e-01
4.51116294e-01 -1.04405224e+00 4.40640152e-01 4.92489576e-01
4.43656668e-02 3.92771550e-02 1.96654990e-01 1.10203910e+00
-1.05375260e-01 -1.75308734e-01 1.14150918e+00 2.73403466e-01
-8.82508636e-01 4.86275554e-01 -1.24147780e-01 4.02719155e-02
1.63792992e+00 -2.88333148e-01 -6.58575073e-02 1.70118809e-02
-5.10489225e-01 1.39838278e-01 3.26155961e-01 3.99649829e-01
3.20251971e-01 -1.61758721e+00 -6.32378340e-01 2.21491069e-01
6.02072835e-01 -1.50936037e-01 1.75795406e-01 1.31181681e+00
2.21191067e-02 3.88064116e-01 -2.27655008e-01 -1.23884964e+00
-1.20033348e+00 6.50717616e-01 4.46854204e-01 -2.49935195e-01
-1.00106907e+00 5.51614165e-01 2.98866391e-01 1.17528565e-01
2.86187291e-01 -9.03112963e-02 -1.54907718e-01 5.31393528e-01
5.98756850e-01 2.68833786e-01 3.07778239e-01 -8.48098218e-01
-6.29501164e-01 7.19876707e-01 -1.03213027e-01 3.87312211e-02
1.32232642e+00 2.40210518e-02 2.70067543e-01 3.77792984e-01
9.83005106e-01 -4.08932388e-01 -1.66654575e+00 -5.12769341e-01
2.53198922e-01 -4.74185795e-01 1.23272583e-01 -4.78401214e-01
-9.68565166e-01 5.96922100e-01 1.41407800e+00 3.59383635e-02
8.30953598e-01 7.64850676e-02 3.30751479e-01 3.50241542e-01
5.49835801e-01 -9.29784894e-01 1.46604046e-01 5.61738610e-01
6.79519951e-01 -1.35073388e+00 2.04730615e-01 1.70771211e-01
-5.93201399e-01 9.87022400e-01 3.19324613e-01 -2.09559724e-01
7.72596300e-01 2.13241652e-01 -4.74682361e-01 -3.66752386e-01
-5.45920789e-01 -5.76904535e-01 5.64712644e-01 6.44662738e-01
-1.83978543e-01 -2.55444318e-01 7.52271861e-02 4.79517102e-01
1.78606659e-01 -7.02095926e-02 -3.27326842e-02 1.02780473e+00
-3.60542625e-01 -6.78913355e-01 -5.46549439e-01 3.34542364e-01
-2.26497695e-01 2.91262239e-01 4.70293649e-02 9.21016812e-01
3.06061834e-01 6.26982689e-01 2.43079811e-01 -1.38868779e-01
4.27245378e-01 -1.01947710e-02 5.19090772e-01 -3.32902163e-01
-6.09138310e-01 1.85536891e-01 -3.77243698e-01 -7.62540162e-01
-4.91306275e-01 -1.09996605e+00 -1.02683079e+00 1.83712780e-01
-4.71887797e-01 -1.10462345e-01 7.07383037e-01 9.79814112e-01
3.35365176e-01 5.24363160e-01 5.27143717e-01 -1.20199573e+00
-7.01350927e-01 -6.54502034e-01 -7.54278243e-01 4.42249000e-01
3.97652507e-01 -1.32368755e+00 -4.09454703e-01 3.78860999e-03]
|
[6.437562465667725, -2.1005494594573975]
|
053c62ab-84cc-491b-b2c4-e48c73bb90c1
|
separate-and-diffuse-using-a-pretrained
|
2301.10752
| null |
https://arxiv.org/abs/2301.10752v2
|
https://arxiv.org/pdf/2301.10752v2.pdf
|
Separate And Diffuse: Using a Pretrained Diffusion Model for Improving Source Separation
|
The problem of speech separation, also known as the cocktail party problem, refers to the task of isolating a single speech signal from a mixture of speech signals. Previous work on source separation derived an upper bound for the source separation task in the domain of human speech. This bound is derived for deterministic models. Recent advancements in generative models challenge this bound. We show how the upper bound can be generalized to the case of random generative models. Applying a diffusion model Vocoder that was pretrained to model single-speaker voices on the output of a deterministic separation model leads to state-of-the-art separation results. It is shown that this requires one to combine the output of the separation model with that of the diffusion model. In our method, a linear combination is performed, in the frequency domain, using weights that are inferred by a learned model. We show state-of-the-art results on 2, 3, 5, 10, and 20 speakers on multiple benchmarks. In particular, for two speakers, our method is able to surpass what was previously considered the upper performance bound.
|
['Lior Wolf', 'Eliya Nachmani', 'Shahar Lutati']
|
2023-01-25
| null | null | null | null |
['audio-source-separation', 'speech-separation', 'multi-speaker-source-separation']
|
['audio', 'speech', 'speech']
|
[ 4.38185543e-01 3.02803725e-01 2.88376331e-01 -6.28805608e-02
-1.24706674e+00 -8.12681317e-01 7.10315883e-01 -2.01953188e-01
-5.64018339e-02 2.93105364e-01 4.16079283e-01 -3.35935563e-01
-5.95939830e-02 -1.47244528e-01 -6.23103440e-01 -1.08054852e+00
-8.07525888e-02 5.13771117e-01 2.17972264e-01 -1.48997515e-01
-2.30649963e-01 4.27653670e-01 -1.31716096e+00 3.08399588e-01
6.42756701e-01 6.55022442e-01 -6.32697493e-02 1.24503112e+00
-4.17440981e-02 4.53933746e-01 -9.72643614e-01 -1.96448356e-01
3.08471829e-01 -1.05499423e+00 -5.32990217e-01 1.07461840e-01
3.61384213e-01 7.11471289e-02 -2.71726489e-01 1.07503510e+00
7.51147091e-01 -2.46955734e-02 7.87712038e-01 -1.09720600e+00
-4.63814616e-01 1.24689639e+00 -3.69426817e-01 3.47788006e-01
1.70343548e-01 -2.18799278e-01 7.64095485e-01 -5.18491983e-01
3.10947031e-01 1.21643746e+00 4.68229949e-01 6.80308998e-01
-1.57095313e+00 -6.71045482e-01 2.17603907e-01 -2.98807442e-01
-1.17750430e+00 -9.03778911e-01 8.32630515e-01 -5.48609674e-01
9.67106998e-01 3.74564022e-01 2.70120919e-01 1.27710021e+00
-9.89052281e-02 7.98294604e-01 1.10821152e+00 -6.39780104e-01
3.18478763e-01 3.79412621e-01 3.94923717e-01 2.91287869e-01
1.09578520e-01 2.98479170e-01 -6.41688704e-01 -3.35947394e-01
3.03682446e-01 -5.50400496e-01 -4.87195373e-01 -1.34621367e-01
-8.57972980e-01 7.73985207e-01 2.01030504e-02 6.09994829e-01
-1.90330237e-01 -3.27367224e-02 3.05265654e-02 2.96624839e-01
5.80966234e-01 2.40526661e-01 -4.24453579e-02 -9.88022760e-02
-1.45624304e+00 5.79831377e-02 1.42695379e+00 8.79684329e-01
7.14741945e-02 4.29539412e-01 -1.71397462e-01 6.78671479e-01
3.27333182e-01 7.72983491e-01 5.64577103e-01 -7.15847075e-01
2.05206454e-01 -2.63593942e-01 -4.57052812e-02 -5.32265544e-01
-1.98589563e-01 -1.01052248e+00 -6.42666698e-01 2.88720936e-01
7.34847724e-01 -4.63676304e-01 -1.15998363e+00 1.78687131e+00
1.67855129e-01 5.72943687e-01 4.63658094e-01 7.79411495e-01
4.77248192e-01 7.28188574e-01 -5.26175261e-01 -3.44919413e-01
1.04413223e+00 -1.13401282e+00 -7.08687425e-01 -5.77142477e-01
-7.96952397e-02 -9.63773310e-01 3.39701653e-01 7.56276309e-01
-1.04172015e+00 -4.71606761e-01 -1.24290907e+00 4.19237643e-01
-1.98287547e-01 7.08486438e-02 1.35638446e-01 1.07395792e+00
-9.38223720e-01 5.45602143e-01 -9.13472772e-01 -1.66860476e-01
1.95083618e-02 2.94156879e-01 -1.28630146e-01 2.28087366e-01
-1.05872655e+00 7.16624618e-01 1.69797279e-02 2.98441909e-02
-1.42480469e+00 -4.74283934e-01 -6.34741962e-01 2.25879401e-01
1.54522464e-01 -5.27760565e-01 1.30668032e+00 -1.00506747e+00
-1.75389910e+00 5.98898530e-01 -3.44659537e-01 -8.40777695e-01
4.92089689e-01 -4.30802763e-01 -7.39975274e-01 4.39365059e-02
-2.01645941e-01 4.90088239e-02 1.39377630e+00 -1.61021399e+00
-5.00855923e-01 -8.80842954e-02 -2.79257536e-01 -9.00902078e-02
2.52883248e-02 1.26228034e-01 -3.60952854e-01 -6.34382665e-01
1.29491821e-01 -1.26274788e+00 5.02462909e-02 -8.84345174e-01
-6.99107945e-01 1.94012389e-01 5.45231462e-01 -9.81506884e-01
1.13379514e+00 -2.53958440e+00 5.70540607e-01 2.96539545e-01
8.71609002e-02 2.76233852e-01 -4.09873500e-02 4.61209834e-01
-2.70085186e-01 -6.34034202e-02 -3.41048151e-01 -9.01363671e-01
2.78382599e-01 -1.52329043e-01 -6.92945302e-01 5.28478324e-01
-2.07546473e-01 2.98798710e-01 -7.34006464e-01 -6.47323877e-02
-1.63565278e-01 7.47570217e-01 -3.61909926e-01 2.54108012e-01
7.73725510e-02 2.90016532e-01 2.06949726e-01 -2.97352169e-02
6.23758554e-01 2.73515433e-02 1.83337927e-01 2.10335448e-01
1.50416881e-01 4.95665550e-01 -1.26330316e+00 1.69520724e+00
-4.03533131e-01 9.04413283e-01 5.79121590e-01 -5.85437238e-01
7.49834239e-01 6.71525836e-01 1.50566742e-01 1.09911524e-01
2.34983698e-01 4.53923821e-01 4.71081495e-01 -1.28682200e-02
-1.99056268e-02 -4.56825733e-01 1.60103070e-03 5.47178924e-01
4.67337191e-01 -2.80228823e-01 1.26666084e-01 2.05881000e-01
9.30857956e-01 -2.40350351e-01 -9.33603346e-02 -2.00673521e-01
3.33656937e-01 -3.85141701e-01 1.47834554e-01 8.48785520e-01
4.22420278e-02 6.77407384e-01 5.15259326e-01 4.40827250e-01
-5.24437308e-01 -1.39912164e+00 1.83852524e-01 1.00588489e+00
-1.91964015e-01 -3.02874327e-01 -1.24863505e+00 -3.67404491e-01
-3.13217416e-02 9.76529896e-01 -6.59892619e-01 -3.54288042e-01
-4.69479740e-01 -8.51051629e-01 8.35675716e-01 2.97222435e-01
-1.27186596e-01 -6.14400625e-01 -3.47394526e-01 1.74599111e-01
1.28812537e-01 -1.17738712e+00 -6.00246191e-01 6.52428687e-01
-5.30842662e-01 -5.87289751e-01 -9.42172229e-01 -6.20368600e-01
2.56073982e-01 1.39195636e-01 9.06128228e-01 -4.41014737e-01
1.86407626e-01 2.62142360e-01 -1.13974161e-01 -5.83605647e-01
-9.91297126e-01 1.89079061e-01 3.04087937e-01 3.34745795e-01
2.24963009e-01 -7.81486928e-01 -1.33495197e-01 1.08583935e-01
-9.54396725e-01 -1.62488088e-01 5.04379213e-01 6.04418516e-01
1.41229704e-01 2.24528387e-01 5.27721941e-01 -6.73777699e-01
8.09684336e-01 -4.19067442e-01 -4.60989505e-01 1.81025550e-01
-3.82921159e-01 4.37394887e-01 5.79820514e-01 -7.59442985e-01
-1.03514802e+00 1.39873832e-01 -6.95962235e-02 -4.93425786e-01
-2.52874672e-01 1.15781836e-01 -2.53704339e-01 1.75624177e-01
6.52492106e-01 2.55479246e-01 -1.09411225e-01 -6.54597878e-01
5.59031546e-01 8.56303096e-01 7.16152012e-01 -2.85340220e-01
7.76496530e-01 4.86145705e-01 -2.97090948e-01 -8.96009505e-01
-7.89667487e-01 -4.56674188e-01 -4.75817442e-01 2.34551162e-01
6.34522557e-01 -7.85291910e-01 -3.97355646e-01 4.49977279e-01
-1.18190849e+00 -3.87641519e-01 -3.17108810e-01 7.39696562e-01
-4.22393352e-01 9.44422409e-02 -5.71410358e-01 -1.25784206e+00
-1.64062932e-01 -1.05769587e+00 9.60492253e-01 2.41747051e-01
-2.70502657e-01 -9.25014317e-01 4.81536210e-01 3.02210063e-01
4.25659716e-01 -3.15899178e-02 5.88882744e-01 -1.26730728e+00
-1.55472964e-01 -1.74440265e-01 4.18966621e-01 7.21735895e-01
2.03417927e-01 -7.52662262e-03 -1.50896442e+00 -2.26541743e-01
5.27224898e-01 3.39720279e-01 1.25565135e+00 4.25428569e-01
2.07058057e-01 -1.54663011e-01 -5.32440901e-01 5.13312936e-01
1.01165318e+00 4.05907333e-01 4.09137964e-01 -2.27714092e-01
5.04044175e-01 5.18369913e-01 -3.37885052e-01 6.74899481e-03
-1.00928828e-01 6.56367123e-01 -5.81633896e-02 -5.51609695e-02
-5.41978419e-01 -1.44461274e-01 6.93924963e-01 9.76879835e-01
1.77431330e-01 -4.37833637e-01 -8.41327727e-01 6.80198729e-01
-1.60374331e+00 -1.05515099e+00 4.50021029e-02 2.21667171e+00
8.34731042e-01 5.60421109e-01 3.84221107e-01 4.38097298e-01
5.83340764e-01 1.31497145e-01 -3.49683404e-01 -4.14476007e-01
-1.18558772e-01 3.18834364e-01 4.80549991e-01 1.13839591e+00
-1.15901327e+00 7.09161699e-01 7.25041866e+00 9.05166626e-01
-1.16841805e+00 3.32179844e-01 1.47343099e-01 -4.93461370e-01
-2.57126331e-01 -8.57270285e-02 -9.71417010e-01 4.33292001e-01
1.42750132e+00 -3.02227736e-01 7.55128324e-01 4.86044407e-01
-9.41818058e-02 3.44141833e-02 -1.39474750e+00 7.89436400e-01
4.90247101e-01 -6.70659959e-01 -3.43205899e-01 1.87854394e-01
6.65070772e-01 3.83180641e-02 2.87797660e-01 1.74398512e-01
3.91314596e-01 -1.16073179e+00 8.90865862e-01 4.49814647e-01
2.20884919e-01 -6.18912578e-01 5.41491151e-01 6.09003246e-01
-8.59032273e-01 1.26619667e-01 1.17136359e-01 2.80260980e-01
3.36405307e-01 8.34939480e-01 -9.48897898e-01 6.54528081e-01
2.12636247e-01 9.00000259e-02 -2.06297219e-01 8.61479104e-01
-3.69769394e-01 9.74989772e-01 -5.75052440e-01 2.35935479e-01
2.06016570e-01 1.22658305e-01 1.03176153e+00 1.67627227e+00
2.78762102e-01 -3.39819491e-01 -7.57469758e-02 8.62092137e-01
1.19306304e-01 -3.16566825e-01 -4.85083282e-01 -1.83870986e-01
1.87731802e-01 1.06373692e+00 -8.05866122e-01 -3.05739999e-01
-4.89778742e-02 1.07452297e+00 5.31908982e-02 5.67002594e-01
-9.24415946e-01 -5.84041834e-01 5.82266033e-01 -4.21855599e-02
6.41041398e-01 -3.14323157e-01 -1.03811897e-01 -1.03151643e+00
-1.67688042e-01 -1.04872417e+00 -5.68471253e-02 -3.58877301e-01
-1.17053437e+00 1.12725282e+00 5.19036055e-02 -8.49514723e-01
-6.22499466e-01 -4.91865009e-01 -5.34486771e-01 1.26928627e+00
-1.06396246e+00 -6.23194456e-01 1.67633250e-01 2.84123182e-01
4.79874402e-01 -2.50775754e-01 8.99277568e-01 2.17246160e-01
-5.28401732e-01 4.58764493e-01 2.79625952e-01 1.62284046e-01
6.01543963e-01 -1.36918104e+00 5.61217189e-01 1.17458665e+00
7.70171225e-01 5.48991561e-01 1.21469092e+00 -2.85738409e-01
-9.47811246e-01 -5.84551454e-01 7.63330519e-01 -6.34847164e-01
5.60643792e-01 -7.75649488e-01 -9.02020454e-01 6.37482762e-01
6.51479781e-01 -2.17304125e-01 9.98976827e-01 5.38502373e-02
-4.17360246e-01 -5.63965179e-03 -7.94299960e-01 2.35107496e-01
7.36299098e-01 -6.13850236e-01 -8.38739097e-01 9.82991979e-02
6.98215663e-01 -5.60186327e-01 -4.76342291e-01 -2.33884498e-01
6.39250934e-01 -9.91475582e-01 9.25042391e-01 -5.83286226e-01
-3.16844927e-03 -2.89741933e-01 -1.64797872e-01 -1.73691428e+00
-1.62206843e-01 -1.17178702e+00 -4.60674286e-01 1.25547230e+00
7.67132044e-01 -5.42003036e-01 2.39810720e-01 4.18296963e-01
2.80646998e-02 -3.50553691e-01 -9.77397382e-01 -1.24604559e+00
2.52959788e-01 -3.76762122e-01 3.94647211e-01 5.02565086e-01
2.17636041e-02 6.54616892e-01 -3.98556471e-01 5.18419266e-01
7.69762754e-01 1.56200439e-01 5.86749732e-01 -1.06382596e+00
-9.59366679e-01 -7.70062029e-01 -3.10525484e-02 -1.21238410e+00
2.91048229e-01 -9.00249779e-01 3.27808082e-01 -1.46125400e+00
-2.60696411e-01 6.41389862e-02 -3.42941791e-01 -4.91779123e-04
-1.13956772e-01 -1.68150991e-01 5.64991951e-01 6.76000267e-02
-8.87399018e-02 2.22164601e-01 5.71490407e-01 -2.63534576e-01
-3.40526640e-01 2.88317978e-01 -9.00318980e-01 8.06960583e-01
7.37762153e-01 -8.89578819e-01 -3.63249987e-01 -3.88417095e-01
-1.44489661e-01 1.10120259e-01 2.63358921e-01 -1.24254692e+00
4.88546908e-01 3.35450530e-01 6.68490753e-02 -9.61398631e-02
6.73987150e-01 -7.01905787e-01 3.66760075e-01 3.98991436e-01
-4.29050446e-01 -6.17660880e-01 3.54834557e-01 6.08915985e-01
-2.61272728e-01 -4.03638631e-01 6.62917972e-01 2.47323766e-01
7.25819319e-02 -3.70110989e-01 -3.69263500e-01 9.71764848e-02
7.36218572e-01 2.13345259e-01 -1.69565424e-01 -6.70662880e-01
-1.16859007e+00 -1.55357808e-01 -1.00734882e-01 4.10428077e-01
2.65997142e-01 -1.10189259e+00 -9.92615461e-01 3.35989505e-01
-4.81411725e-01 -4.11418498e-01 4.53789085e-02 1.03505874e+00
4.12033200e-02 4.39180106e-01 3.16896141e-01 -4.29782331e-01
-1.37997639e+00 7.90080905e-01 5.83257198e-01 -3.83878142e-01
-2.96726167e-01 9.77589786e-01 3.11379462e-01 9.54994485e-02
3.88529181e-01 -4.69612539e-01 7.71384761e-02 9.34949368e-02
6.65918171e-01 4.01565343e-01 7.69016370e-02 -6.48380518e-01
-6.03137910e-01 4.68853801e-01 9.42256823e-02 -9.36442733e-01
1.08690321e+00 6.60611093e-02 1.86972678e-01 7.84557998e-01
1.24908102e+00 7.27565229e-01 -1.10767984e+00 -1.89959958e-01
-1.44297168e-01 -2.78872382e-02 6.93742633e-02 -8.75425458e-01
-8.71365726e-01 1.05917442e+00 5.16851008e-01 8.30655813e-01
1.08560002e+00 9.32197794e-02 6.50563538e-01 2.37673782e-02
1.98011436e-02 -7.96581030e-01 -4.70161051e-01 4.68570203e-01
9.43862915e-01 -7.85840034e-01 -4.07622933e-01 -3.57172579e-01
-5.64393163e-01 9.45991457e-01 -2.29829550e-01 -3.36587846e-01
8.92393649e-01 9.16689098e-01 1.01215467e-01 1.42478034e-01
-7.02433765e-01 -2.75093228e-01 6.20389640e-01 5.26101172e-01
5.16161442e-01 2.46042028e-01 1.83068737e-01 1.03350234e+00
-4.44222242e-01 -2.31713027e-01 2.80182272e-01 6.76958501e-01
-5.47266781e-01 -1.03393924e+00 -6.91289246e-01 -7.42259547e-02
-5.86320400e-01 -3.46812844e-01 -7.06430137e-01 3.62600178e-01
-2.71773729e-02 1.48563850e+00 -1.78865924e-01 -2.97882438e-01
3.27709168e-01 5.02343476e-01 5.54595351e-01 -5.73830783e-01
-7.32494056e-01 6.32000268e-01 -6.42664963e-03 -2.33394597e-02
-4.24146086e-01 -6.35481477e-01 -1.08337319e+00 -6.38118945e-03
-3.31580937e-01 5.95443428e-01 7.15700328e-01 9.26662803e-01
3.65641415e-01 9.25255239e-01 3.59942734e-01 -8.33612263e-01
-7.35613823e-01 -1.23096883e+00 -7.40119755e-01 8.24043974e-02
8.03968847e-01 -2.75183886e-01 -9.48007882e-01 2.73103863e-01]
|
[15.224556922912598, 5.765683650970459]
|
9db1baf0-db33-447a-b453-c0cb8e9311ad
|
a-graph-neural-network-approach-to
|
2303.13773
| null |
https://arxiv.org/abs/2303.13773v1
|
https://arxiv.org/pdf/2303.13773v1.pdf
|
A Graph Neural Network Approach to Nanosatellite Task Scheduling: Insights into Learning Mixed-Integer Models
|
This study investigates how to schedule nanosatellite tasks more efficiently using Graph Neural Networks (GNN). In the Offline Nanosatellite Task Scheduling (ONTS) problem, the goal is to find the optimal schedule for tasks to be carried out in orbit while taking into account Quality-of-Service (QoS) considerations such as priority, minimum and maximum activation events, execution time-frames, periods, and execution windows, as well as constraints on the satellite's power resources and the complexity of energy harvesting and management. The ONTS problem has been approached using conventional mathematical formulations and precise methods, but their applicability to challenging cases of the problem is limited. This study examines the use of GNNs in this context, which has been effectively applied to many optimization problems, including traveling salesman problems, scheduling problems, and facility placement problems. Here, we fully represent MILP instances of the ONTS problem in bipartite graphs. We apply a feature aggregation and message-passing methodology allied to a ReLU activation function to learn using a classic deep learning model, obtaining an optimal set of parameters. Furthermore, we apply Explainable AI (XAI), another emerging field of research, to determine which features -- nodes, constraints -- had the most significant impact on learning performance, shedding light on the inner workings and decision process of such models. We also explored an early fixing approach by obtaining an accuracy above 80\% both in predicting the feasibility of a solution and the probability of a decision variable value being in the optimal solution. Our results point to GNNs as a potentially effective method for scheduling nanosatellite tasks and shed light on the advantages of explainable machine learning models for challenging combinatorial optimization problems.
|
['Leandro dos Santos Coelho', 'Eduardo Augusto Bezerra', 'Eduardo Camponogara', 'Cezar Antônio Rigo', 'Laio Oriel Seman', 'Bruno Machado Pacheco']
|
2023-03-24
| null | null | null | null |
['combinatorial-optimization']
|
['methodology']
|
[ 3.78546476e-01 3.37714702e-01 -3.19579989e-01 -1.20276049e-01
-1.73650101e-01 -2.83438146e-01 1.30299523e-01 2.79827237e-01
-3.62962008e-01 9.56459999e-01 -3.59818667e-01 -4.43775445e-01
-1.00360382e+00 -7.86677122e-01 -9.54478443e-01 -9.57221031e-01
-6.06846154e-01 8.26075435e-01 -6.18809462e-01 -2.93138117e-01
1.05936863e-01 6.54780269e-01 -1.55746746e+00 -1.55085921e-01
8.66493762e-01 1.09705532e+00 5.33785284e-01 3.66603464e-01
6.70484975e-02 3.13964963e-01 -7.26544142e-01 3.50655913e-01
3.02627683e-01 1.97147369e-01 -6.99450374e-01 8.86336640e-02
-3.86319488e-01 3.00597191e-01 -2.86699265e-01 5.40565848e-01
2.57394195e-01 3.56385797e-01 5.96404791e-01 -1.72934031e+00
-2.42476389e-01 5.49244404e-01 -2.13153362e-01 4.40854698e-01
-1.70932822e-02 7.49102086e-02 8.40766966e-01 3.35646421e-02
2.88321257e-01 6.53688550e-01 2.75388598e-01 1.81332290e-01
-1.11911559e+00 -4.01570112e-01 6.92164302e-02 3.41378301e-01
-1.10018027e+00 -1.18254013e-01 3.65955442e-01 -2.37888813e-01
1.43768728e+00 5.16420722e-01 7.04238892e-01 5.53435087e-01
8.67980063e-01 5.73855281e-01 9.59169209e-01 -6.24727726e-01
5.20565748e-01 -1.85897037e-01 2.80053094e-02 5.86049616e-01
2.51474053e-01 2.63432115e-01 -5.50755024e-01 7.28920773e-02
3.00401181e-01 -2.23987445e-01 -1.99002862e-01 -1.87338606e-01
-1.00358021e+00 7.81559885e-01 5.92530727e-01 4.92846906e-01
-5.84794343e-01 5.61332345e-01 2.33585671e-01 4.06711042e-01
5.44900239e-01 9.83695388e-01 -5.47617733e-01 2.88733721e-01
-9.18022215e-01 2.40665704e-01 7.97845602e-01 7.73513794e-01
9.34029222e-01 3.42288911e-01 -4.40881044e-01 5.61186373e-02
-2.38100253e-03 5.57459712e-01 -5.73907718e-02 -7.97191501e-01
4.22717959e-01 4.85428154e-01 1.98817283e-01 -6.89551532e-01
-1.04821634e+00 -7.98913181e-01 -5.73629141e-01 1.17071107e-01
2.32014656e-02 -4.15090412e-01 -9.97283041e-01 1.53095257e+00
1.36713013e-01 1.06674783e-01 -8.91120285e-02 1.08571458e+00
2.69521683e-01 9.81634200e-01 1.52554885e-01 -3.35663736e-01
1.23393965e+00 -7.74997771e-01 -6.27452433e-01 -4.94775176e-01
6.39057398e-01 -2.91489661e-01 5.13074040e-01 2.31237188e-01
-9.75055218e-01 -2.01149508e-01 -1.06309605e+00 3.18644643e-01
-6.17332220e-01 1.28535986e-01 1.27510345e+00 5.29193759e-01
-1.21174085e+00 7.19799399e-01 -8.53396714e-01 -2.89103925e-01
4.16828878e-02 1.14497840e+00 -3.48613784e-02 -5.19600734e-02
-1.19312489e+00 1.16634715e+00 5.15771389e-01 5.43244362e-01
-9.82673109e-01 -5.74094832e-01 -7.44994044e-01 7.14513838e-01
8.06287587e-01 -9.34349060e-01 1.00919950e+00 -9.54733133e-01
-9.84130859e-01 2.28473529e-01 -2.50917822e-02 -7.10752130e-01
-2.22776428e-01 4.49713945e-01 -2.64625251e-01 -1.82466775e-01
-5.91134690e-02 3.97630811e-01 7.61174440e-01 -9.12894785e-01
-5.01940489e-01 -2.13359922e-01 4.93596345e-01 3.86794925e-01
-1.81926861e-01 -2.72253335e-01 -1.37062063e-02 -9.55493897e-02
-1.99114591e-01 -1.20622742e+00 -3.60636383e-01 -6.01625919e-01
-3.57219815e-01 -4.05978233e-01 4.63520259e-01 -4.54651922e-01
1.01727521e+00 -1.50105929e+00 5.95470846e-01 5.61696529e-01
1.05478287e-01 -1.15995146e-01 -1.47974327e-01 6.31866992e-01
8.30574054e-03 2.71605514e-02 -1.99708238e-01 5.65642584e-03
3.43384087e-01 5.71672618e-01 1.34718329e-01 4.00050640e-01
1.75991565e-01 7.51428366e-01 -7.29991972e-01 -4.07108888e-02
2.20782310e-01 1.10258587e-01 -6.89186528e-02 8.45522657e-02
-6.51308239e-01 2.57236063e-01 -7.57098675e-01 7.72748053e-01
1.77880451e-01 -2.39749715e-01 3.55615228e-01 -6.60964400e-02
-2.72162020e-01 4.81686089e-03 -7.70002961e-01 1.30624795e+00
-8.57418180e-01 8.58852863e-01 2.28108421e-01 -1.39765882e+00
6.88928962e-01 9.06855837e-02 8.87761593e-01 -1.01266074e+00
2.91796237e-01 3.87477241e-02 -6.11498617e-02 -6.39794707e-01
6.17575705e-01 -1.34504929e-01 -1.59495547e-01 4.59798574e-01
-1.24357313e-01 8.53979066e-02 4.51993406e-01 -9.51387268e-03
1.18934202e+00 5.32429703e-02 -1.41203523e-01 -6.50322735e-01
1.01653583e-01 3.94858479e-01 3.00685316e-01 5.81525624e-01
2.79444307e-01 -1.18763009e-02 7.73626029e-01 -6.05237782e-01
-7.80995846e-01 -3.97082180e-01 2.10348386e-02 1.18834591e+00
1.20131612e-01 2.13673204e-01 -7.29769409e-01 -3.44200850e-01
2.12620094e-01 9.69428420e-01 -7.30132222e-01 -1.18336961e-01
-3.51678312e-01 -1.13763094e+00 -4.99837585e-02 2.29629159e-01
-1.58653617e-01 -1.15553498e+00 -8.87274921e-01 2.17620432e-01
-2.43203014e-01 -9.47399199e-01 -1.80632785e-01 9.41299021e-01
-6.40372097e-01 -1.04971015e+00 -3.78869772e-01 -5.20456433e-01
9.50994313e-01 2.95614213e-01 1.19298327e+00 2.84900874e-01
-3.91424060e-01 5.00451684e-01 -2.20903456e-01 -4.97808069e-01
1.23920646e-02 4.99410212e-01 2.38435730e-01 -3.29390198e-01
1.32009372e-01 -1.08765066e-01 -3.53543788e-01 3.58108699e-01
-1.06110990e+00 -3.42764310e-03 8.06485653e-01 5.03226697e-01
5.29049993e-01 4.34473813e-01 6.52883589e-01 -5.75157404e-01
6.29864395e-01 -7.52228856e-01 -8.01033795e-01 5.16409576e-01
-9.13603842e-01 4.14326578e-01 5.85366726e-01 6.71094283e-02
-6.15984857e-01 -3.18141729e-02 4.76474732e-01 -4.23446923e-01
2.19900638e-01 8.23417664e-01 5.20362817e-02 -4.17971075e-01
3.50078791e-01 1.12538524e-01 -2.90485173e-02 8.99300650e-02
-1.53989971e-01 3.60684127e-01 1.13112405e-01 -7.11978674e-01
6.11300349e-01 1.20906711e-01 4.60984886e-01 -6.69507980e-01
-6.58806980e-01 -3.57593715e-01 -2.09744677e-01 -4.18552518e-01
6.74366474e-01 -5.99464715e-01 -1.15055382e+00 -1.23089470e-01
-8.34003866e-01 -5.04348218e-01 -9.28244069e-02 3.99373472e-01
-6.88539982e-01 2.49839462e-02 -4.70802858e-02 -9.41224873e-01
-2.95349985e-01 -1.14948523e+00 8.32180083e-01 4.27292645e-01
1.62190661e-01 -1.06347835e+00 -3.28831285e-01 5.06789207e-01
5.83312929e-01 4.88133103e-01 1.15577519e+00 -4.18191582e-01
-8.21925521e-01 -1.37795294e-02 -4.04947586e-02 -1.40561879e-01
-1.87484264e-01 -3.24613690e-01 -6.06722176e-01 -6.60402894e-01
-1.83196545e-01 -1.24176122e-01 6.24944746e-01 8.62521052e-01
1.12290239e+00 -3.03748965e-01 -5.23636281e-01 4.94540036e-01
1.52285242e+00 7.39545897e-02 3.95807773e-01 6.50619924e-01
2.31610268e-01 7.77315855e-01 7.95967519e-01 5.75841725e-01
3.66053671e-01 6.30132675e-01 1.06715894e+00 -1.26573294e-01
4.15547252e-01 4.17074084e-01 2.56987214e-01 3.82242143e-01
-3.21366549e-01 -7.67823339e-01 -9.30417657e-01 4.80552316e-01
-1.80658889e+00 -6.13419950e-01 1.23218028e-02 2.15486860e+00
1.02570742e-01 1.34089127e-01 -1.33433565e-01 -2.73837093e-02
6.90451264e-01 1.51222795e-01 -6.77550733e-01 -7.95846164e-01
1.51109904e-01 7.91032538e-02 1.04926455e+00 2.30177030e-01
-8.20884109e-01 4.76579875e-01 6.02777243e+00 6.96896017e-01
-1.09162891e+00 -2.49338880e-01 7.18815565e-01 -3.45283240e-01
-4.60153967e-01 -3.58814038e-02 -5.92307091e-01 4.40308183e-01
1.24006796e+00 -2.85021871e-01 1.22149575e+00 5.19018054e-01
6.50597870e-01 -4.69196141e-01 -1.09010732e+00 4.18969959e-01
-4.14025262e-02 -1.39589310e+00 -4.17976022e-01 1.89152703e-01
7.77004898e-01 8.93254876e-02 1.46648148e-02 5.25800288e-01
-2.97260098e-02 -1.22055709e+00 5.27632236e-01 6.08394384e-01
5.72766125e-01 -9.20984745e-01 9.14931834e-01 2.86365539e-01
-9.90029633e-01 -3.68554384e-01 -4.27062452e-01 -2.21685126e-01
6.73822267e-03 6.55694008e-01 -1.15243387e+00 9.91351008e-01
6.02638721e-01 2.27993447e-02 2.27790396e-03 9.98349130e-01
1.28697708e-01 3.37098956e-01 -5.98107636e-01 -4.14322913e-01
6.68680191e-01 -1.45589620e-01 3.86778653e-01 8.38902593e-01
4.57728624e-01 1.08088769e-01 1.86058268e-01 8.87604833e-01
1.44927412e-01 -3.19697380e-01 -4.33976740e-01 -2.80938506e-01
2.80169159e-01 1.49242115e+00 -9.37014341e-01 2.48502031e-01
-1.03696734e-01 2.61575848e-01 1.58896059e-01 5.55498600e-01
-9.67695415e-01 -2.55917609e-01 3.95168751e-01 1.36647061e-01
1.46093681e-01 -3.97793829e-01 -2.55642146e-01 -5.70224285e-01
-3.31132039e-02 -5.29641807e-01 2.96876043e-01 -9.16507423e-01
-7.62825429e-01 4.55675513e-01 2.11173534e-01 -9.35404301e-01
-2.21584231e-01 -6.76933110e-01 -6.66336298e-01 8.18057537e-01
-1.92611051e+00 -9.22925293e-01 -2.18146518e-01 2.61152923e-01
4.70915645e-01 -1.32506654e-01 4.91449952e-01 9.53889862e-02
-6.63040102e-01 -3.74869518e-02 3.84968728e-01 -7.31030822e-01
-9.99011938e-03 -1.15532780e+00 3.90784666e-02 4.81183589e-01
-1.84865296e-01 7.28619248e-02 1.02798140e+00 -5.58223426e-01
-2.14102936e+00 -8.54273558e-01 7.66784489e-01 3.05917840e-02
6.22590005e-01 -1.26823798e-01 -3.46487671e-01 6.30365372e-01
2.42743134e-01 -2.98959821e-01 4.13407952e-01 3.60762089e-01
7.95648336e-01 -1.44714907e-01 -9.97802794e-01 1.28854573e-01
6.16957545e-01 3.88152152e-02 8.23184289e-03 9.06483710e-01
5.35694182e-01 -5.82455695e-01 -7.66447246e-01 4.03289974e-01
1.78758249e-01 -5.37863672e-01 6.08764410e-01 -6.58426166e-01
1.21109910e-01 -1.18014887e-01 1.18349098e-01 -1.62676060e+00
-3.83465171e-01 -5.65644205e-01 -1.06821336e-01 5.70822418e-01
5.73233545e-01 -4.19315368e-01 7.47996509e-01 8.67827117e-01
-5.22125065e-01 -1.12654364e+00 -1.25355947e+00 -6.77193284e-01
-4.71245676e-01 -2.90844262e-01 6.46216750e-01 7.92735994e-01
-1.86210692e-01 -6.03910983e-02 -3.73119831e-01 5.65841079e-01
5.53868473e-01 3.28221381e-01 2.62424171e-01 -1.07912612e+00
-1.60309523e-01 -3.61593552e-02 -1.66318059e-01 -4.11647826e-01
5.14303982e-01 -1.03496635e+00 1.47154927e-01 -1.82985032e+00
-1.28840327e-01 -7.80097246e-01 -4.87527132e-01 6.65614367e-01
2.09277272e-01 -4.21464056e-01 3.32144529e-01 -6.90106526e-02
-6.80409372e-01 4.65236098e-01 1.02242374e+00 -4.33269471e-01
-4.25266214e-02 2.86798179e-01 -4.35659349e-01 5.99418767e-02
8.67293358e-01 -7.00635731e-01 -4.15948182e-01 -8.25716019e-01
6.41408563e-01 4.95559722e-01 2.68237054e-01 -1.03872263e+00
4.47175384e-01 -4.81670111e-01 1.65775627e-01 -2.86614180e-01
2.43668064e-01 -1.16865313e+00 4.00700450e-01 5.03058255e-01
-1.06446259e-01 2.40942642e-01 4.14939880e-01 7.51484990e-01
2.02228904e-01 -5.44016719e-01 1.31819472e-01 -6.81749508e-02
-8.54318202e-01 1.98582008e-01 -6.22389674e-01 -4.43873823e-01
1.38360310e+00 -1.50900066e-01 -3.03757101e-01 -1.56076193e-01
-1.01084542e+00 9.72990394e-01 1.41629800e-01 3.13549012e-01
1.29949942e-01 -7.64971495e-01 -6.01397991e-01 1.14558339e-02
-2.66709626e-01 -1.01824880e-01 4.09220248e-01 7.91157842e-01
-5.20614386e-01 8.33472133e-01 -4.34826285e-01 -4.91438955e-01
-7.02668786e-01 4.82788563e-01 4.79044080e-01 -5.69751263e-01
-5.57082072e-02 3.57983261e-01 -1.98338032e-01 -3.38061571e-01
2.64919609e-01 -4.23574418e-01 -2.52232831e-02 2.81326054e-03
-2.38142207e-01 5.60866058e-01 4.93011773e-01 -6.02910295e-03
-3.65822792e-01 1.71959132e-01 3.80513936e-01 4.03268576e-01
1.46832061e+00 -6.97939144e-03 -1.94322526e-01 -2.15778798e-02
5.91506541e-01 -5.34793913e-01 -1.01262033e+00 1.50790602e-01
4.60732542e-02 9.90551803e-03 5.46563029e-01 -9.12892282e-01
-1.24271309e+00 4.52294528e-01 5.14021873e-01 7.09971189e-01
1.17388928e+00 -2.37140268e-01 5.35398901e-01 6.93448663e-01
5.31816423e-01 -1.42950857e+00 -3.80564362e-01 4.56951350e-01
6.16560578e-01 -1.10931432e+00 2.11560547e-01 -1.02202743e-01
-4.10741419e-01 1.26424563e+00 5.43942392e-01 1.94068119e-01
2.03716189e-01 3.68588239e-01 -4.22421604e-01 -4.34168667e-01
-8.97484243e-01 -2.99524456e-01 1.26246259e-01 3.14193934e-01
-3.25276591e-02 4.03467625e-01 -3.11491400e-01 3.97728682e-01
1.48467749e-01 1.06803216e-01 5.79390407e-01 9.56638396e-01
-5.94360530e-01 -7.09250987e-01 -5.21762729e-01 7.70502746e-01
-1.62094980e-01 1.04053989e-01 9.41762403e-02 8.95892322e-01
1.59013435e-01 9.30713534e-01 1.94344401e-01 -1.82813585e-01
2.37156615e-01 -9.61673781e-02 3.25727701e-01 -4.01810139e-01
-7.64597833e-01 -3.80864054e-01 3.52395415e-01 -3.96949679e-01
-4.11107808e-01 -3.54738623e-01 -1.40645146e+00 -3.57855171e-01
-4.99263972e-01 4.70980406e-01 1.28957236e+00 1.31118584e+00
4.42005873e-01 1.09049773e+00 8.15484047e-01 -1.27080500e+00
-4.37795937e-01 -5.52127600e-01 -6.81555212e-01 -4.22419488e-01
1.69885576e-01 -7.72597909e-01 -3.39991361e-01 -4.64184552e-01]
|
[5.218470573425293, 2.889234781265259]
|
27836d8e-800a-45f0-ab90-1611d36eb434
|
diverse-projection-ensembles-for
|
2306.07124
| null |
https://arxiv.org/abs/2306.07124v1
|
https://arxiv.org/pdf/2306.07124v1.pdf
|
Diverse Projection Ensembles for Distributional Reinforcement Learning
|
In contrast to classical reinforcement learning, distributional reinforcement learning algorithms aim to learn the distribution of returns rather than their expected value. Since the nature of the return distribution is generally unknown a priori or arbitrarily complex, a common approach finds approximations within a set of representable, parametric distributions. Typically, this involves a projection of the unconstrained distribution onto the set of simplified distributions. We argue that this projection step entails a strong inductive bias when coupled with neural networks and gradient descent, thereby profoundly impacting the generalization behavior of learned models. In order to facilitate reliable uncertainty estimation through diversity, this work studies the combination of several different projections and representations in a distributional ensemble. We establish theoretical properties of such projection ensembles and derive an algorithm that uses ensemble disagreement, measured by the average $1$-Wasserstein distance, as a bonus for deep exploration. We evaluate our algorithm on the behavior suite benchmark and find that diverse projection ensembles lead to significant performance improvements over existing methods on a wide variety of tasks with the most pronounced gains in directed exploration problems.
|
['Matthijs T. J. Spaan', 'Wendelin Böhmer', 'Moritz A. Zanger']
|
2023-06-12
| null | null | null | null |
['distributional-reinforcement-learning']
|
['methodology']
|
[ 2.52119862e-02 9.28971693e-02 -1.67820349e-01 -5.99131584e-01
-9.96926248e-01 -8.20990562e-01 7.22928703e-01 1.20125808e-01
-6.57780766e-01 1.14679360e+00 2.74379373e-01 -4.71773654e-01
-3.68186116e-01 -8.87471855e-01 -8.30520511e-01 -8.98128867e-01
-1.15759134e-01 8.71840417e-01 -1.98384255e-01 -2.07666472e-01
4.25192177e-01 4.73483562e-01 -1.39181840e+00 3.10418103e-02
9.21035171e-01 1.07673025e+00 -3.03276032e-01 3.20713401e-01
-1.25220522e-01 4.38588083e-01 -7.25316703e-01 -3.46910477e-01
4.09196854e-01 -4.66856629e-01 -5.68030357e-01 -3.03917944e-01
1.71168242e-02 -3.07643294e-01 1.35068327e-01 1.28238988e+00
4.49167758e-01 4.01811659e-01 1.25338745e+00 -1.06219304e+00
-3.32233071e-01 1.07439113e+00 -4.83461827e-01 7.60527700e-02
7.40154609e-02 1.79684237e-01 1.19124341e+00 -7.04552233e-01
4.33976859e-01 1.23327267e+00 6.33412898e-01 4.08507347e-01
-1.88870037e+00 -6.40190005e-01 1.66289151e-01 -2.46858418e-01
-1.20608950e+00 -2.07049921e-01 6.95879579e-01 -3.78103584e-01
6.72491789e-01 1.47473358e-03 3.93899262e-01 1.27337575e+00
3.55583668e-01 6.88457489e-01 1.34545457e+00 -3.49209130e-01
9.27807152e-01 5.06254613e-01 -8.67826939e-02 3.02011669e-01
4.01409179e-01 4.43245143e-01 -3.75879735e-01 -5.18927872e-01
3.29939008e-01 -1.14843898e-01 -2.83577770e-01 -9.54936147e-01
-8.05799484e-01 1.16859126e+00 3.46010953e-01 -1.23676077e-01
-3.45789015e-01 4.08344626e-01 3.07669163e-01 4.16420430e-01
3.99182022e-01 8.11125040e-01 -3.36198747e-01 -3.10298920e-01
-7.61328161e-01 5.78218222e-01 1.11891150e+00 4.66605484e-01
8.29937279e-01 8.76084119e-02 -2.06857145e-01 6.96154654e-01
2.23881617e-01 5.45077682e-01 3.29282701e-01 -8.82451892e-01
4.64243054e-01 2.85621285e-01 3.56654823e-01 -6.19707525e-01
-1.87000126e-01 -5.95270813e-01 -5.03922820e-01 7.52363265e-01
7.11114228e-01 -5.49671650e-01 -8.19196820e-01 2.08621192e+00
1.61491901e-01 -3.56764913e-01 1.79623775e-02 6.98572159e-01
-5.61909303e-02 4.94467109e-01 1.61539078e-01 -6.03561588e-02
5.55477083e-01 -3.91889542e-01 -2.11901337e-01 -1.02094300e-01
4.60551202e-01 -1.90413132e-01 1.24784219e+00 5.55330038e-01
-8.40502143e-01 -6.26391247e-02 -1.11867177e+00 4.90827084e-01
-2.73589462e-01 -2.77756035e-01 4.90204781e-01 8.47496092e-01
-8.90942812e-01 1.08395326e+00 -7.16184735e-01 8.39264914e-02
5.38386703e-01 3.69151920e-01 -5.55823296e-02 2.35672772e-01
-9.33190942e-01 1.06348336e+00 7.05893457e-01 -1.79606661e-01
-1.04204428e+00 -6.39653325e-01 -5.88319361e-01 2.31592506e-01
3.24333668e-01 -4.38582480e-01 1.16839492e+00 -9.54565585e-01
-1.58525634e+00 2.35170275e-01 3.36890578e-01 -8.04603696e-01
7.96173215e-01 -2.53128678e-01 2.99340393e-02 -3.56115639e-01
-3.33027959e-01 5.82784235e-01 9.13268924e-01 -1.22312212e+00
-4.91661578e-01 -3.18261296e-01 2.34389342e-02 3.47096264e-01
-2.82298237e-01 -5.53344309e-01 2.81641543e-01 -4.22899127e-01
-2.46745616e-01 -1.08547103e+00 -3.75841349e-01 -2.67245889e-01
-4.19186294e-01 -3.21128339e-01 2.95588046e-01 -8.32705349e-02
1.17017770e+00 -2.03303742e+00 3.57187271e-01 7.85165191e-01
4.10891175e-02 -2.60251850e-01 -6.19446784e-02 3.91864419e-01
-1.08088329e-01 1.00197509e-01 -6.49610221e-01 -1.15119569e-01
5.59659600e-01 3.10546130e-01 -7.88825393e-01 5.91638684e-01
-2.01300159e-02 6.02870643e-01 -9.23000932e-01 7.63134751e-03
-3.78971361e-02 1.55867249e-01 -7.18221664e-01 1.43051729e-01
-6.51851654e-01 2.23940045e-01 -4.41816121e-01 2.47794047e-01
4.25222993e-01 -7.64186680e-03 1.51303619e-01 3.82845223e-01
2.16475233e-01 3.40458244e-01 -1.20257199e+00 1.60869384e+00
-3.76604140e-01 3.09015214e-01 -2.73127049e-01 -1.12922263e+00
1.13718045e+00 -2.09994823e-01 2.40662336e-01 -3.22680622e-01
1.65098399e-01 3.26690137e-01 1.92861557e-01 2.16110691e-01
4.98086333e-01 -4.44021434e-01 -1.69192791e-01 8.72701526e-01
2.82111485e-02 -3.63235027e-01 7.20093325e-02 -1.17565855e-01
1.01483750e+00 4.57739800e-01 1.92478478e-01 -7.48801172e-01
4.40322794e-02 -1.41148120e-01 5.32845914e-01 9.04313028e-01
-2.51827873e-02 4.28858966e-01 9.75261748e-01 -3.59505117e-01
-1.13234854e+00 -1.64201176e+00 -4.15742397e-01 1.13561523e+00
-9.60208252e-02 -1.00134544e-01 -5.95410466e-01 -8.99158895e-01
5.02598703e-01 1.40713477e+00 -8.49943459e-01 -4.74154800e-01
-1.64517432e-01 -9.45159614e-01 4.33308810e-01 6.19516134e-01
-3.95352095e-02 -9.82237220e-01 -7.58789122e-01 1.66807488e-01
3.90148968e-01 -2.94415593e-01 -1.70216352e-01 8.03954482e-01
-8.81742179e-01 -7.83124328e-01 -6.28611505e-01 -1.44982040e-01
5.37598372e-01 -6.37756288e-01 1.33862770e+00 -6.06987059e-01
2.89918762e-02 3.55287373e-01 -4.22254615e-02 -7.30068505e-01
-3.15095782e-01 8.40659067e-02 1.91929668e-01 -3.85845721e-01
5.33180594e-01 -8.63522649e-01 -4.78341043e-01 2.71671899e-02
-8.73301148e-01 -5.89711547e-01 6.45251274e-01 1.14430249e+00
5.50202429e-01 -9.23663527e-02 7.18582749e-01 -9.60498571e-01
1.17662191e+00 -7.05634773e-01 -8.35977674e-01 1.72918901e-01
-9.46771920e-01 8.50055993e-01 6.41514242e-01 -4.20945227e-01
-1.03896093e+00 -8.76689851e-02 6.00585602e-02 -2.93017149e-01
-8.47236365e-02 5.62361836e-01 3.09134983e-02 2.63541102e-01
1.01034880e+00 2.78238565e-01 1.13185279e-01 -3.04320544e-01
6.02778673e-01 2.97676235e-01 3.43742579e-01 -1.22109652e+00
5.97826540e-01 1.72517955e-01 4.15231613e-03 -5.23820877e-01
-7.27203965e-01 1.09725863e-01 -3.19227964e-01 4.72018719e-02
3.43923986e-01 -5.09131789e-01 -5.23154616e-01 -1.57851517e-01
-6.29094541e-01 -4.60331172e-01 -8.63047183e-01 6.38628721e-01
-8.85942340e-01 4.27758135e-02 -8.62135515e-02 -1.04625964e+00
-2.81190574e-01 -1.18732536e+00 7.13241518e-01 1.29911914e-01
-5.21794021e-01 -1.02068090e+00 5.82358778e-01 -4.55477387e-01
4.99285698e-01 4.43117887e-01 1.25010276e+00 -1.05075526e+00
-1.77930117e-01 1.44920155e-01 1.20472915e-01 5.32574654e-01
-1.50651962e-01 -1.56532601e-01 -8.08259308e-01 -4.29613471e-01
-1.74419582e-01 -6.34967804e-01 1.18188119e+00 4.88613188e-01
1.31941414e+00 -5.22574745e-02 -1.45114079e-01 6.70521736e-01
1.39921653e+00 3.06159228e-01 5.21121621e-01 4.41893905e-01
5.62423319e-02 4.50369090e-01 2.85937041e-01 7.67689049e-01
-9.09978002e-02 8.28152150e-02 4.60245013e-01 5.99298835e-01
6.01002336e-01 -5.11790812e-01 4.28539932e-01 1.82247192e-01
1.28634751e-01 -1.20941073e-01 -8.66770983e-01 4.88091737e-01
-1.67291665e+00 -9.64537621e-01 8.42476130e-01 2.54788327e+00
9.87629771e-01 4.37913686e-01 2.85478324e-01 -5.83017319e-02
4.36331332e-01 1.36178240e-01 -1.06721294e+00 -7.10362792e-01
5.59468791e-02 4.57664698e-01 5.32025635e-01 4.05508071e-01
-8.06911349e-01 3.79064918e-01 7.00312757e+00 8.03955972e-01
-1.05322838e+00 -3.81314605e-01 9.27434504e-01 -4.25631821e-01
-8.47996056e-01 -7.71394521e-02 -7.19792128e-01 5.41944921e-01
1.01092672e+00 -4.67425197e-01 5.64265370e-01 1.13490808e+00
-4.58097339e-01 -2.01658741e-01 -1.59818709e+00 7.55692899e-01
-3.76401484e-01 -1.18987405e+00 4.31425646e-02 3.03160727e-01
8.97859991e-01 9.70745236e-02 4.87816185e-01 6.09259665e-01
9.58738208e-01 -1.51711977e+00 7.15614200e-01 6.50004625e-01
6.30557358e-01 -1.35675359e+00 5.87212324e-01 4.73919421e-01
-4.92843568e-01 -2.03234166e-01 -5.58005691e-01 8.63327086e-02
-3.11960012e-01 5.80256343e-01 -7.79642105e-01 5.33746257e-02
5.51946104e-01 2.78617859e-01 -1.39472231e-01 8.85803223e-01
-2.08822876e-01 7.63315320e-01 -7.58425176e-01 -4.85074639e-01
4.03530121e-01 -5.37095070e-01 6.50792480e-01 9.86229837e-01
4.65694070e-01 -2.06183761e-01 -3.23828543e-03 1.34525764e+00
-2.43128151e-01 -4.55150455e-02 -8.19303215e-01 -2.12912396e-01
5.63233435e-01 9.49459672e-01 -3.92584205e-01 -8.63124579e-02
1.95834227e-02 3.64128053e-01 6.06384516e-01 4.15619254e-01
-7.78384805e-01 -5.09517729e-01 7.40752161e-01 -1.90845415e-01
4.92084712e-01 -8.27013552e-02 -4.32732075e-01 -9.03889358e-01
4.34392914e-02 -9.51911867e-01 4.53987569e-01 -1.25118360e-01
-1.46870732e+00 6.12473488e-01 1.72435731e-01 -9.55247760e-01
-7.42677629e-01 -6.84586883e-01 -6.55574203e-01 1.05651391e+00
-1.15438342e+00 -2.82090694e-01 3.20658386e-01 3.27204049e-01
9.67634991e-02 -3.79412711e-01 8.33239913e-01 -3.51543486e-01
-3.73737484e-01 6.12200201e-01 7.94906080e-01 -1.42764792e-01
5.32461166e-01 -1.79801559e+00 4.81268838e-02 4.25170243e-01
3.22187871e-01 5.99435687e-01 1.10158205e+00 -3.08073848e-01
-1.30167282e+00 -9.39489782e-01 -2.78852247e-02 -5.89452982e-01
7.69730031e-01 -2.40090266e-01 -7.35151052e-01 7.08934188e-01
2.33084068e-01 -9.64955762e-02 7.02254951e-01 4.43875462e-01
-5.01736879e-01 -1.47925630e-01 -1.19558227e+00 5.20511568e-01
6.27504945e-01 -4.01888996e-01 -7.92872310e-01 -6.82416633e-02
1.78121954e-01 -2.27657944e-01 -8.02521765e-01 3.29606652e-01
6.31523430e-01 -1.15737927e+00 7.41786838e-01 -8.66198182e-01
5.00183463e-01 -9.84033644e-02 -5.12924194e-01 -1.99561357e+00
-4.72634882e-02 -5.64651489e-01 -1.67423397e-01 8.24513197e-01
6.40373111e-01 -7.52943993e-01 9.51260626e-01 7.42211699e-01
2.50142306e-01 -1.02982080e+00 -8.64026427e-01 -7.85924852e-01
8.45093489e-01 -3.55740607e-01 7.49006271e-01 4.77310181e-01
1.81706935e-01 2.21326381e-01 -1.40527189e-01 -2.15863690e-01
9.69881713e-01 4.10744309e-01 6.50834382e-01 -1.21437037e+00
-5.50726771e-01 -7.46967316e-01 -1.31320924e-01 -9.07902181e-01
4.04614449e-01 -8.93931627e-01 1.92328125e-01 -9.58591878e-01
8.48738775e-02 -5.50980747e-01 -6.13749862e-01 1.48365170e-01
9.15749446e-02 -3.45765710e-01 -1.50989071e-01 -1.87446386e-01
-5.15107036e-01 9.97442245e-01 8.14330637e-01 -9.65094268e-02
-2.45779067e-01 3.05789672e-02 -8.48698676e-01 8.09322476e-01
8.77808452e-01 -4.70172942e-01 -7.12699771e-01 -2.39132702e-01
3.85824353e-01 -4.29860614e-02 8.23356956e-02 -8.94216537e-01
-1.36156678e-02 -3.36533189e-01 7.59713233e-01 -4.77792680e-01
2.35827416e-01 -5.93237221e-01 -2.80495472e-02 4.19670999e-01
-8.67018342e-01 4.41541597e-02 1.79417029e-01 8.57226253e-01
-1.41158804e-01 -2.58904397e-01 8.12891126e-01 7.94150587e-03
-3.82329732e-01 1.13311179e-01 -1.55311510e-01 4.13740158e-01
9.53147769e-01 -4.05086484e-03 -1.45919472e-01 -4.98840481e-01
-5.84204197e-01 2.13110685e-01 4.93866563e-01 1.03494577e-01
4.58411872e-01 -1.23159873e+00 -7.45310366e-01 5.12249023e-02
3.67141142e-02 1.80087518e-02 -3.28039289e-01 4.24940556e-01
-2.12714449e-01 1.54871434e-01 -3.58898431e-01 -5.30121922e-01
-5.97893655e-01 3.01062465e-01 4.91936356e-01 -2.89948374e-01
-4.39016044e-01 8.63893569e-01 1.91712156e-01 -5.79956353e-01
4.38312739e-01 -3.37560296e-01 1.78510666e-01 -1.73164587e-02
3.73613149e-01 3.43055457e-01 -1.26685411e-01 6.59412816e-02
-1.81575999e-01 3.41406427e-02 -1.41341209e-01 -3.70875359e-01
1.50640881e+00 1.91632047e-01 1.13893747e-01 6.99307978e-01
1.13649857e+00 -8.65525082e-02 -1.59653592e+00 -1.96868256e-01
2.75872678e-01 -4.21951473e-01 -1.63317531e-01 -9.53438163e-01
-8.11311066e-01 8.80219996e-01 5.25181413e-01 1.57378286e-01
8.32120597e-01 -2.33012646e-01 2.16490477e-01 8.25575769e-01
4.84303415e-01 -1.31143928e+00 -1.54191867e-01 4.53946710e-01
9.61191416e-01 -1.14179373e+00 1.20614655e-01 5.40770888e-01
-7.69935489e-01 1.16540098e+00 5.38620591e-01 -4.28233415e-01
6.46629453e-01 4.55268860e-01 -2.06453145e-01 9.85819623e-02
-9.44599450e-01 2.35213503e-01 2.67452300e-01 5.54236531e-01
3.02916199e-01 2.42852390e-01 -5.21508045e-02 4.68589306e-01
-4.78057861e-01 -3.89248997e-01 3.80114228e-01 6.78844571e-01
-6.36167645e-01 -1.05305684e+00 -1.83888212e-01 7.75444150e-01
-3.91116798e-01 -3.13661876e-03 -1.74597666e-01 7.40761697e-01
-3.01726639e-01 4.42564756e-01 1.46209687e-01 -2.16680020e-01
1.52855009e-01 2.43009791e-01 5.22261262e-01 -4.13844407e-01
-2.59518504e-01 -2.81974375e-01 4.82687019e-02 -5.19057870e-01
1.10095330e-01 -7.99342394e-01 -1.00119996e+00 -2.77939349e-01
-8.64772648e-02 6.04143202e-01 6.65262222e-01 8.91959488e-01
1.86074525e-01 2.77667642e-01 5.76750517e-01 -6.94139481e-01
-1.76443756e+00 -9.10199821e-01 -9.47317719e-01 4.15381283e-01
2.94520676e-01 -9.06285822e-01 -5.84046245e-01 -5.38197458e-01]
|
[4.123377323150635, 2.5773816108703613]
|
555cb637-c4c7-4920-aefe-e8eb66e32620
|
automated-metrics-for-medical-multi-document
|
2305.13693
| null |
https://arxiv.org/abs/2305.13693v1
|
https://arxiv.org/pdf/2305.13693v1.pdf
|
Automated Metrics for Medical Multi-Document Summarization Disagree with Human Evaluations
|
Evaluating multi-document summarization (MDS) quality is difficult. This is especially true in the case of MDS for biomedical literature reviews, where models must synthesize contradicting evidence reported across different documents. Prior work has shown that rather than performing the task, models may exploit shortcuts that are difficult to detect using standard n-gram similarity metrics such as ROUGE. Better automated evaluation metrics are needed, but few resources exist to assess metrics when they are proposed. Therefore, we introduce a dataset of human-assessed summary quality facets and pairwise preferences to encourage and support the development of better automated evaluation methods for literature review MDS. We take advantage of community submissions to the Multi-document Summarization for Literature Review (MSLR) shared task to compile a diverse and representative sample of generated summaries. We analyze how automated summarization evaluation metrics correlate with lexical features of generated summaries, to other automated metrics including several we propose in this work, and to aspects of human-assessed summary quality. We find that not only do automated metrics fail to capture aspects of quality as assessed by humans, in many cases the system rankings produced by these metrics are anti-correlated with rankings according to human annotators.
|
['Byron C. Wallace', 'Erin Bransom', 'Bailey E. Kuehl', 'Thinh Hung Truong', 'Jay DeYoung', 'Yulia Otmakhova', 'Lucy Lu Wang']
|
2023-05-23
| null | null | null | null |
['multi-document-summarization', 'document-summarization']
|
['natural-language-processing', 'natural-language-processing']
|
[ 3.56334120e-01 3.16095650e-01 -4.70200151e-01 -2.49011904e-01
-1.50555933e+00 -1.04605591e+00 6.24420524e-01 1.17986095e+00
-4.76089180e-01 9.66871202e-01 1.12458837e+00 -3.27101588e-01
-4.09720004e-01 -3.14730376e-01 -9.61635783e-02 -1.07894629e-01
3.42795402e-01 5.25967598e-01 -6.72473907e-02 1.57635678e-02
1.02995300e+00 2.18994230e-01 -1.25660825e+00 6.49427772e-01
1.31822729e+00 2.50422716e-01 5.85053349e-03 9.33803618e-01
-8.93610790e-02 6.67933226e-01 -1.01606321e+00 -4.39001590e-01
-3.80025238e-01 -6.82379603e-01 -9.36314464e-01 -1.86225280e-01
5.61826527e-01 -5.94922081e-02 3.29942137e-01 8.26679707e-01
7.20470071e-01 -1.67409644e-01 1.13455486e+00 -9.63477254e-01
-5.64994335e-01 8.38756502e-01 -4.85982835e-01 5.88687360e-01
8.19199562e-01 -7.02445861e-03 1.44717598e+00 -6.28150225e-01
1.09598994e+00 1.29405534e+00 5.80075264e-01 1.88694820e-01
-1.21772289e+00 -4.04251546e-01 -1.56957194e-01 -1.66969933e-02
-7.80382335e-01 -6.51925325e-01 3.48236799e-01 -6.02770150e-01
1.03993750e+00 4.76768613e-01 3.84141177e-01 8.63231122e-01
5.02883732e-01 4.88715082e-01 9.72458601e-01 -4.24252123e-01
2.73553610e-01 8.70786756e-02 4.81936693e-01 3.17930281e-01
1.04084539e+00 -7.02982247e-01 -5.61185181e-01 -7.84486532e-01
-1.91757884e-02 -6.29115045e-01 -3.02585065e-01 1.20601781e-01
-1.61138546e+00 8.15349996e-01 -2.03615725e-01 5.33468068e-01
-5.34073293e-01 -2.71404237e-01 7.29538381e-01 1.30204991e-01
6.41448200e-01 1.19869649e+00 -4.43055272e-01 -4.93294299e-01
-1.53166962e+00 5.18953681e-01 1.01998699e+00 5.97490489e-01
1.94224417e-01 -3.65392148e-01 -4.21642065e-01 9.16408479e-01
4.16961163e-02 2.93511361e-01 6.30304039e-01 -1.18096697e+00
5.40797651e-01 7.38416493e-01 1.67925015e-01 -1.46921754e+00
-7.28839934e-01 -2.75814712e-01 -6.33556187e-01 -3.01623315e-01
1.40859112e-01 -5.52352518e-02 -2.29828238e-01 1.41523480e+00
-7.04009458e-02 -7.42036819e-01 1.54909894e-01 5.46709776e-01
1.27923214e+00 3.86193156e-01 6.89280927e-02 -7.26229608e-01
1.34240770e+00 -4.80339766e-01 -9.85111237e-01 -3.27929109e-02
9.03042018e-01 -1.13059556e+00 9.56394136e-01 6.47474170e-01
-1.43280494e+00 -1.99924797e-01 -1.31449652e+00 -8.02857652e-02
-2.54363865e-01 1.51898757e-01 2.48290926e-01 5.89691043e-01
-1.10275030e+00 7.94337630e-01 -5.72737753e-01 -5.93654811e-01
4.29800838e-01 8.06887299e-02 -3.75961751e-01 1.84148759e-01
-9.16026890e-01 1.43733633e+00 1.74343795e-01 -2.88763970e-01
-2.62678802e-01 -6.99665129e-01 -7.16058135e-01 -5.09836860e-02
1.22944981e-01 -9.07681763e-01 1.22396779e+00 -5.10499001e-01
-7.44655192e-01 8.34084928e-01 -3.23575020e-01 -7.41826743e-02
2.94526577e-01 1.37768432e-01 -3.71896386e-01 4.61984217e-01
7.23202825e-01 4.18994963e-01 -7.81656131e-02 -9.98862684e-01
-5.82092643e-01 -3.26024979e-01 -1.77270100e-01 2.43331999e-01
-3.62999171e-01 3.84403944e-01 4.12864797e-02 -6.96254849e-01
-4.03711125e-02 -6.11863017e-01 -2.04874650e-01 -5.05855978e-01
-5.82341492e-01 -4.08705890e-01 1.73155844e-01 -7.47008562e-01
1.80159307e+00 -1.36556852e+00 3.54004391e-02 1.98613405e-01
3.86729360e-01 7.70684406e-02 -2.71825105e-01 8.31565142e-01
9.25892293e-02 7.93319106e-01 -3.32088381e-01 -4.95736860e-02
-2.29140855e-02 -2.05648378e-01 -1.28195465e-01 4.55266684e-01
3.26485485e-01 7.40853548e-01 -1.24503529e+00 -9.13625658e-01
-3.11553568e-01 -1.09587004e-02 -4.96777892e-01 -1.98692739e-01
-1.04049370e-01 -1.16316741e-03 -4.27400053e-01 4.85685170e-01
2.76200235e-01 -3.90257418e-01 2.54904121e-01 -2.73781657e-01
-3.43077779e-01 9.48549688e-01 -9.23222780e-01 1.50479472e+00
-6.86776266e-02 7.01400459e-01 -2.93743819e-01 -7.13985801e-01
8.08646977e-01 3.53777319e-01 4.92706358e-01 -3.55955660e-01
-1.02518804e-01 5.26838303e-01 2.53433377e-01 -4.52480614e-01
9.64507520e-01 -1.11395714e-03 -1.99350178e-01 8.90493572e-01
-4.08128719e-04 -5.19797981e-01 8.63142848e-01 7.12529421e-01
1.53761578e+00 -2.90445119e-01 9.60462987e-01 -6.16370380e-01
4.32658345e-01 4.89874870e-01 4.58677024e-01 8.14775050e-01
5.29611818e-02 9.32072699e-01 9.51497555e-01 1.15495160e-01
-1.11007774e+00 -5.83524823e-01 -3.04391563e-01 5.72369576e-01
-4.44415361e-01 -1.13888979e+00 -6.97489977e-01 -5.78647375e-01
-1.03106797e-01 8.57906401e-01 -5.25054693e-01 -3.43531184e-02
-1.74079046e-01 -9.17482913e-01 7.66392231e-01 2.41462976e-01
-2.86420226e-01 -1.01427424e+00 -8.42280090e-01 5.24334371e-01
-3.76056045e-01 -8.62878025e-01 -5.63529789e-01 -3.56441736e-02
-8.11131179e-01 -1.42877436e+00 -7.87960351e-01 -3.34770769e-01
3.36593121e-01 1.22893788e-01 1.44657981e+00 2.15403318e-01
-1.33547992e-01 6.11398399e-01 -4.38399792e-01 -6.15569413e-01
-8.90846550e-01 3.15659225e-01 7.71248862e-02 -9.05540764e-01
4.45050806e-01 -2.94377536e-01 -5.80237865e-01 3.54905166e-02
-1.08871830e+00 -1.82805747e-01 6.75350010e-01 7.39188612e-01
3.59586358e-01 -3.54877800e-01 1.30852783e+00 -9.68355417e-01
1.56070077e+00 -4.85013962e-01 1.32793799e-01 4.52644914e-01
-9.40568447e-01 -2.49140034e-03 4.17059183e-01 -1.13409981e-01
-6.23283327e-01 -6.66576743e-01 4.34968546e-02 5.12828231e-01
6.49540946e-02 1.13522911e+00 1.48000836e-01 3.23601723e-01
1.04241884e+00 -2.48613834e-01 4.36964035e-02 -1.73894584e-01
2.30530292e-01 8.91999424e-01 2.82752663e-01 -5.01833558e-01
3.43091309e-01 8.93514678e-02 -7.73723423e-02 -8.93619835e-01
-8.61585796e-01 -5.50646365e-01 -3.96907657e-01 -2.37840265e-02
7.45056570e-01 -7.55038738e-01 -4.47537094e-01 -3.00274432e-01
-1.41815853e+00 1.29857659e-01 -1.26265109e-01 4.88749802e-01
-4.77278739e-01 7.62420475e-01 -3.56293499e-01 -6.48190141e-01
-7.04708338e-01 -1.17703164e+00 1.01416123e+00 8.74331892e-02
-1.36901140e+00 -9.51409519e-01 5.05161285e-01 3.91045064e-01
3.00861448e-01 4.43308532e-01 1.18567860e+00 -1.01061428e+00
1.47003159e-01 -2.27716193e-01 -3.31870019e-02 1.49259329e-01
4.52950060e-01 4.19296533e-01 -5.13236344e-01 1.71648171e-02
-1.51872098e-01 -3.13040704e-01 7.71919966e-01 7.29911268e-01
7.52322137e-01 -5.60560226e-01 -3.90811414e-01 -3.35515529e-01
1.01919043e+00 1.24023758e-01 3.23655456e-01 4.11164701e-01
3.78816992e-01 9.06581759e-01 7.20623195e-01 3.38866860e-01
6.20063841e-01 3.97751182e-01 -3.68199348e-01 1.89845577e-01
7.76341036e-02 3.84018086e-02 2.62067974e-01 1.28934443e+00
2.08255917e-01 -4.48192954e-01 -1.18326247e+00 6.87651634e-01
-1.71447647e+00 -1.00952864e+00 -3.21945459e-01 1.90581179e+00
1.06506693e+00 4.11318094e-01 3.74885708e-01 8.38461071e-02
5.45877755e-01 1.76964700e-01 -2.66079545e-01 -8.04170251e-01
-4.11396980e-01 1.16440132e-01 2.40637347e-01 3.20265085e-01
-6.25756860e-01 3.90939921e-01 7.03439713e+00 5.50050735e-01
-5.41795015e-01 -1.87933087e-01 6.48893118e-01 -1.31145164e-01
-8.69256318e-01 7.28461444e-02 -4.24192935e-01 3.29766065e-01
1.12315893e+00 -8.69191289e-01 -2.15893701e-01 4.74072486e-01
5.96941054e-01 -4.06042278e-01 -1.28674734e+00 5.74219823e-01
3.53159547e-01 -1.57229936e+00 1.74114943e-01 4.66051921e-02
8.90256226e-01 -9.10036266e-02 -2.52487302e-01 -2.02576280e-01
3.97757500e-01 -9.25781965e-01 3.85505408e-01 5.68720043e-01
5.30298829e-01 -6.50576770e-01 8.26532364e-01 1.27064332e-01
-6.56154335e-01 2.88485289e-01 -3.84550095e-01 1.88961461e-01
1.58643067e-01 8.55525196e-01 -1.02343464e+00 6.80225432e-01
2.25967839e-01 8.80111873e-01 -9.21755612e-01 1.12594116e+00
3.13592143e-02 5.59150636e-01 7.00171664e-02 -5.09408474e-01
1.15639001e-01 5.60131781e-02 8.18000555e-01 1.62205100e+00
4.11179960e-01 -5.67708723e-02 -1.48275778e-01 6.43456042e-01
-3.96357253e-02 4.90850985e-01 -7.61222422e-01 -5.59508622e-01
5.43560863e-01 1.25633097e+00 -9.06303108e-01 -5.51082850e-01
-1.74138695e-01 3.73898178e-01 9.49485525e-02 1.56503264e-02
-1.93284616e-01 -4.71703559e-01 2.37885952e-01 -9.65580493e-02
-1.87796667e-01 -3.16483341e-02 -7.59550452e-01 -9.54073370e-01
-1.26354294e-02 -1.36911511e+00 5.89986563e-01 -8.54559720e-01
-1.36688232e+00 5.75924814e-01 9.14827436e-02 -1.22016048e+00
-3.38830531e-01 -2.14163080e-01 -6.61670804e-01 7.34419227e-01
-1.04750359e+00 -5.30752599e-01 1.35990322e-01 -2.77060986e-01
5.16301990e-01 -8.59058723e-02 7.95819700e-01 -6.59016818e-02
-3.84643257e-01 3.61377537e-01 2.96787247e-02 -3.70043397e-01
1.17152286e+00 -1.46006560e+00 2.55375057e-01 5.39874732e-01
3.31726596e-02 9.60035622e-01 1.11231720e+00 -1.01008856e+00
-9.65447903e-01 -5.55502176e-01 1.25778449e+00 -7.53604352e-01
6.96213722e-01 4.02053386e-01 -9.11319733e-01 1.19435713e-01
4.77603078e-01 -8.97187948e-01 1.19802380e+00 3.55168432e-01
-1.56343132e-01 2.93416679e-01 -9.94660437e-01 6.13348246e-01
7.52127647e-01 -3.57997507e-01 -1.04107463e+00 5.30523121e-01
3.58683169e-01 -9.84674692e-02 -1.13550162e+00 2.86337912e-01
6.25113010e-01 -4.96526986e-01 6.42347515e-01 -7.36473620e-01
1.02511835e+00 -3.45312268e-01 2.90726181e-02 -1.60924673e+00
-3.02679896e-01 -3.82843941e-01 2.20512226e-01 1.35322046e+00
8.37272346e-01 -2.67417967e-01 3.93395066e-01 6.22921884e-01
-4.03979927e-01 -7.34664917e-01 -6.93299770e-01 -3.91496241e-01
3.41128290e-01 -1.56841502e-01 4.00492638e-01 9.85812783e-01
7.32574999e-01 6.55562937e-01 2.11916775e-01 -3.11082929e-01
4.42873240e-01 -1.29636914e-01 6.61734939e-01 -1.39400220e+00
1.77417353e-01 -9.57976937e-01 -2.90070146e-01 -1.49804413e-01
4.94409464e-02 -9.89287913e-01 1.77458860e-03 -2.36104155e+00
5.74202538e-01 -1.91954616e-02 -2.84518272e-01 1.92138672e-01
-4.44685251e-01 -1.51500162e-02 -1.15596555e-01 2.52239406e-01
-9.16532576e-01 2.03124389e-01 1.07953477e+00 -1.25250950e-01
-2.11322367e-01 -4.01994705e-01 -1.46454644e+00 6.47529781e-01
6.32024407e-01 -6.75162077e-01 -2.20465094e-01 -1.07448876e-01
7.35930502e-01 9.63716358e-02 -2.25519389e-02 -8.82147908e-01
3.42155069e-01 -2.51320034e-01 2.78677344e-01 -8.16573024e-01
-3.23794186e-01 -2.47213654e-02 4.51183431e-02 2.83856094e-01
-8.22796881e-01 6.26602232e-01 1.94591776e-01 2.84547299e-01
-1.62550166e-01 -5.19116640e-01 2.10435808e-01 -2.71502793e-01
8.35640058e-02 -3.75709832e-01 -6.84046090e-01 3.56910348e-01
3.76625031e-01 -1.22880064e-01 -8.34873617e-01 -5.16648412e-01
-2.38165915e-01 3.54815304e-01 6.04381621e-01 3.40782583e-01
5.24374008e-01 -1.03262866e+00 -1.11829436e+00 -8.65859926e-01
3.99429590e-01 -3.69974941e-01 1.11331314e-01 1.08374989e+00
-4.58882093e-01 7.40019321e-01 -1.18693456e-01 -3.10105443e-01
-1.34817457e+00 2.42010951e-01 -3.33776206e-01 -6.16046369e-01
-3.88608664e-01 2.43629485e-01 -1.45195857e-01 -1.98679775e-01
1.71433464e-02 -4.74132329e-01 -5.63578486e-01 7.22629905e-01
7.64998555e-01 6.50545418e-01 3.25830609e-01 -5.91105819e-01
-6.01422071e-01 2.02298164e-01 -2.58701295e-01 -3.83492380e-01
1.51563656e+00 -5.00905737e-02 -3.11568886e-01 6.25842154e-01
1.04925561e+00 3.09284300e-01 -9.82574821e-02 1.95675060e-01
5.43462515e-01 -2.00125407e-02 -1.08596692e-02 -1.06943345e+00
-2.33047619e-01 4.16094691e-01 -3.58290263e-02 4.34713960e-01
7.15041697e-01 -1.10155828e-01 3.80471468e-01 6.04061723e-01
-1.36271507e-01 -1.39036584e+00 1.39018580e-01 2.13125095e-01
1.17141616e+00 -1.05658770e+00 7.05094039e-01 -7.68970549e-02
-8.58427167e-01 1.20951116e+00 1.58330172e-01 1.59885168e-01
2.11600646e-01 3.40891853e-02 2.13807654e-02 -5.48847973e-01
-1.01028419e+00 1.61836103e-01 6.52313232e-01 2.88606256e-01
9.69378471e-01 -6.98194727e-02 -1.30938494e+00 7.00632751e-01
-4.22129273e-01 -1.97900794e-02 1.15888298e+00 8.74505818e-01
-4.54625189e-01 -1.13247359e+00 -3.22002172e-01 1.18762612e+00
-8.35260749e-01 -1.21290646e-01 -8.49460006e-01 5.17316461e-01
-4.42647308e-01 1.37113154e+00 -2.38872349e-01 -1.20212577e-01
3.99734110e-01 -9.77500454e-02 2.88657248e-01 -8.98702383e-01
-9.37961221e-01 9.19370130e-02 7.77982652e-01 -1.02566659e-01
-7.82675326e-01 -9.80925381e-01 -9.09692109e-01 -1.30695850e-01
-3.25750321e-01 6.03772461e-01 6.22378230e-01 9.16275978e-01
4.38338071e-01 5.85060298e-01 2.22226053e-01 -5.09527028e-01
-4.64294940e-01 -1.09810209e+00 -2.82162726e-01 2.72966266e-01
2.77670026e-01 -4.51395750e-01 -4.01801705e-01 -6.35170043e-02]
|
[12.30379581451416, 9.564273834228516]
|
f14de9d4-cf87-447b-83ff-d6fe31e267b4
|
overview-generalizations-of-multi-agent-path
|
1702.05515
| null |
http://arxiv.org/abs/1702.05515v1
|
http://arxiv.org/pdf/1702.05515v1.pdf
|
Overview: Generalizations of Multi-Agent Path Finding to Real-World Scenarios
|
Multi-agent path finding (MAPF) is well-studied in artificial intelligence,
robotics, theoretical computer science and operations research. We discuss
issues that arise when generalizing MAPF methods to real-world scenarios and
four research directions that address them. We emphasize the importance of
addressing these issues as opposed to developing faster methods for the
standard formulation of the MAPF problem.
|
['Tansel Uras', 'Sven Koenig', 'Nora Ayanian', 'Wolfgang Hoenig', 'Liron Cohen', 'Craig Tovey', 'T. K. Satish Kumar', 'Hong Xu', 'Guni Sharon', 'Hang Ma']
|
2017-02-17
| null | null | null | null |
['multi-agent-path-finding']
|
['playing-games']
|
[ 7.47645125e-02 6.50239214e-02 -3.33360434e-01 -1.30291581e-01
-2.48010635e-01 -6.10003948e-01 7.49219298e-01 4.60387170e-01
-7.59740055e-01 1.03991318e+00 -5.54966442e-02 -5.84415019e-01
-7.50865400e-01 -1.03162611e+00 -4.55013275e-01 -4.18252379e-01
-9.24475610e-01 8.91012907e-01 5.31018496e-01 -4.56423253e-01
7.62961566e-01 7.14712560e-01 -1.15078771e+00 -4.73869413e-01
6.26718104e-01 3.20859671e-01 3.76413465e-01 6.88810766e-01
-1.17310196e-01 2.26124465e-01 -6.32248759e-01 -7.69298226e-02
4.09497678e-01 -2.09430948e-01 -1.11565435e+00 6.67375773e-02
-2.72884876e-01 -1.70215398e-01 -4.74071875e-02 1.02846563e+00
1.45321384e-01 4.64363009e-01 7.73491740e-01 -2.04390836e+00
-1.52080357e-01 3.15684199e-01 -8.02295208e-01 4.47301805e-01
7.26841450e-01 -1.78811654e-01 6.52554929e-01 -4.42917258e-01
9.62397099e-01 1.44745159e+00 5.62234104e-01 2.46317863e-01
-6.34960592e-01 -4.37237062e-02 3.22100788e-01 4.67119932e-01
-1.17190015e+00 -2.30902120e-01 1.93441719e-01 -7.79291242e-02
1.18627262e+00 3.08009565e-01 5.70995927e-01 3.03028911e-01
7.60255337e-01 7.18365312e-01 1.08690989e+00 -5.49451292e-01
4.54209417e-01 -2.05580458e-01 2.42830247e-01 4.64287192e-01
6.37738049e-01 4.40145701e-01 7.07083009e-03 -3.51036936e-01
9.87308681e-01 -3.83358389e-01 -1.29168212e-01 -5.99177539e-01
-1.53827703e+00 1.27673483e+00 2.97383428e-01 1.34846210e-01
-7.61961877e-01 2.94750929e-01 2.52390265e-01 4.64493513e-01
-1.37826679e-02 5.86168110e-01 -4.54175889e-01 -3.58415246e-01
1.05657168e-02 8.57589602e-01 1.15482473e+00 1.19474816e+00
8.58476520e-01 -3.56009096e-01 8.30572724e-01 3.81307423e-01
4.96751994e-01 1.08534716e-01 -1.39871791e-01 -1.56104863e+00
2.53863335e-01 3.11547875e-01 5.39773047e-01 -1.17996919e+00
-7.96922863e-01 1.17413282e-01 -2.49798447e-01 6.20456934e-01
3.72745216e-01 -1.79802522e-01 -3.17968875e-01 1.29007435e+00
6.16099179e-01 1.45740703e-01 3.25832009e-01 7.72257328e-01
2.75631905e-01 8.95597160e-01 -3.50709349e-01 -5.39706588e-01
1.09398425e+00 -1.37716484e+00 -6.68489456e-01 -5.10110497e-01
5.59685647e-01 -7.60222673e-01 4.78124112e-01 3.59866083e-01
-1.08964622e+00 3.07607681e-01 -1.07099950e+00 2.17861935e-01
-6.46710455e-01 -8.37879300e-01 9.69875216e-01 3.82701933e-01
-1.40846026e+00 3.99409413e-01 -7.96954393e-01 -1.05860937e+00
-1.29130244e-01 5.75713038e-01 -3.23436499e-01 -2.36855730e-01
-8.64947379e-01 1.48596168e+00 5.61653554e-01 -1.75026461e-01
-4.98832196e-01 -8.50658864e-02 -8.29707682e-01 -2.79605657e-01
8.03580463e-01 -6.98411763e-01 1.49296224e+00 -2.79523194e-01
-1.56279910e+00 3.86947930e-01 -2.38645434e-01 -5.41100085e-01
3.40189755e-01 1.96954980e-01 -4.36536789e-01 -6.36955053e-02
4.71725672e-01 3.73380989e-01 1.91655114e-01 -1.11728859e+00
-1.03875208e+00 -1.19461223e-01 4.20184851e-01 4.92167115e-01
2.17763573e-01 1.99002355e-01 -4.42908034e-02 -2.50975847e-01
1.74733281e-01 -1.17093885e+00 -1.10290766e+00 -6.10559992e-02
1.18349539e-02 -4.34926927e-01 8.35725784e-01 5.94011247e-02
7.76788890e-01 -1.56838095e+00 2.23728001e-01 4.05279219e-01
-1.64156288e-01 -1.84945062e-01 -3.21416557e-01 1.09150386e+00
4.97186154e-01 1.13815535e-03 -4.47870344e-02 2.26785406e-01
5.92235401e-02 6.55182600e-01 6.87947050e-02 7.46551394e-01
-1.69702634e-01 7.55476475e-01 -1.30358386e+00 -3.09545040e-01
3.99317980e-01 -1.57125577e-01 -3.97025257e-01 -4.80385214e-01
-2.05195710e-01 5.11841699e-02 -7.99075246e-01 6.05852425e-01
5.38359761e-01 -4.54784520e-02 3.11356515e-01 6.64965272e-01
-6.25474274e-01 2.59113818e-01 -1.47671139e+00 1.63900030e+00
-2.65055120e-01 5.82439423e-01 3.29031229e-01 -9.89325881e-01
5.51414371e-01 5.23857363e-02 7.98982024e-01 -5.39185405e-01
1.03405453e-01 3.71889889e-01 1.53079182e-01 -3.21125269e-01
6.81446433e-01 -1.28750265e-01 -2.29322031e-01 1.04005587e+00
-4.34947908e-01 -1.92040473e-01 5.28893054e-01 -2.06082370e-02
1.32191718e+00 -1.90294266e-01 9.23105836e-01 -7.69597173e-01
5.35245299e-01 7.51471996e-01 2.82729924e-01 9.62540686e-01
-8.84658813e-01 -2.02390507e-01 1.95202045e-02 -8.07598531e-01
-6.46879971e-01 -8.49590123e-01 7.35821500e-02 9.28238630e-01
9.15474951e-01 -4.79389906e-01 -4.63307470e-01 -6.20790005e-01
1.28186271e-01 5.21574199e-01 -4.98211354e-01 2.76340753e-01
-1.00272465e+00 -9.12288964e-01 -1.26395728e-02 1.72561601e-01
2.26348877e-01 -1.06114447e+00 -1.05969322e+00 6.21441483e-01
-1.90009296e-01 -1.13803208e+00 -1.56483650e-01 -1.67273562e-02
-6.81163549e-01 -1.36240673e+00 -4.13585752e-01 -9.73744214e-01
6.37417138e-01 1.01229835e+00 8.51349592e-01 4.22113910e-02
-2.08537295e-01 6.57665133e-01 -5.37691116e-01 -6.72960699e-01
-5.25407672e-01 8.39560181e-02 1.28038585e-01 -6.57267809e-01
3.50338995e-01 -3.37271303e-01 -3.99588525e-01 7.41166770e-01
-4.62411284e-01 -2.20592380e-01 5.22050798e-01 1.84992313e-01
3.41529906e-01 5.77704549e-01 7.48720348e-01 -4.36152816e-01
1.04901111e+00 -7.32200801e-01 -6.39075935e-01 7.37833083e-02
-7.91001141e-01 -2.38465816e-01 2.44835079e-01 -2.21124291e-01
-5.07288814e-01 -1.75517157e-01 1.24676526e-01 3.79493594e-01
-1.27398670e-01 6.32261276e-01 1.87695131e-01 -8.33528042e-01
4.46513325e-01 -3.72432806e-02 3.77956629e-01 4.59442027e-02
3.64659876e-01 4.23036635e-01 3.11977834e-01 -6.44697666e-01
6.31530583e-01 7.14332819e-01 2.97782809e-01 -8.37610126e-01
1.74077526e-02 -6.82426274e-01 -3.86355758e-01 -2.66003698e-01
4.87437695e-01 -2.79996067e-01 -8.53551149e-01 1.12469725e-01
-1.36588311e+00 -1.65250748e-01 -5.76005057e-02 4.85159636e-01
-1.10336006e+00 2.34898314e-01 -3.28923583e-01 -8.24560046e-01
9.21823680e-02 -1.31291163e+00 7.50100970e-01 2.95678407e-01
-1.87035754e-01 -1.33023357e+00 5.02676010e-01 -8.53536129e-02
5.40729344e-01 3.87148768e-01 6.48342967e-01 -3.74914706e-01
-5.94814301e-01 -2.45082621e-02 -7.17277601e-02 -6.86797619e-01
2.33994082e-01 -1.54872611e-01 -1.18043445e-01 -3.31739873e-01
3.32647236e-04 2.19676822e-01 3.37975115e-01 7.84576118e-01
2.26181626e-01 -3.38360727e-01 -8.96448255e-01 3.47956903e-02
1.69225955e+00 8.46476912e-01 3.48876089e-01 1.28060377e+00
-1.34063438e-01 9.53522503e-01 1.11910331e+00 2.84695446e-01
9.71874416e-01 8.30450952e-01 6.05691910e-01 2.82913059e-01
3.64285499e-01 2.94120848e-01 1.90919548e-01 4.16646391e-01
-1.83904842e-01 -4.52445775e-01 -1.32901275e+00 6.39693916e-01
-2.51336312e+00 -8.19290876e-01 -4.51259702e-01 1.68962157e+00
-1.46346465e-01 -2.46622711e-02 5.61262310e-01 2.18742847e-01
9.17899489e-01 -1.06235005e-01 -3.12077463e-01 -7.84507573e-01
-4.35075238e-02 -2.71722227e-01 8.35718036e-01 9.26584721e-01
-1.13431013e+00 9.47038174e-01 8.18771648e+00 1.73974961e-01
-6.72852337e-01 2.03554943e-01 -1.32957578e-01 3.93547148e-01
-7.09149092e-02 2.64283746e-01 -3.35990757e-01 -5.62139452e-02
9.28351164e-01 -8.52822483e-01 8.54209721e-01 8.35767269e-01
3.56279373e-01 -4.79120046e-01 -7.23206878e-01 4.32715863e-01
-2.11693510e-01 -1.31148469e+00 -2.92341590e-01 5.58973730e-01
5.53432286e-01 2.19577670e-01 -3.85280520e-01 8.26740265e-03
6.25687003e-01 -8.56821239e-01 5.47912478e-01 -1.73285887e-01
-1.60911679e-01 -8.78472984e-01 6.02374434e-01 4.73310232e-01
-1.23400414e+00 -3.31499189e-01 -4.72535849e-01 -8.20407093e-01
7.92450726e-01 2.00468287e-01 -1.29803443e+00 7.80783474e-01
4.96637255e-01 6.65352345e-01 4.38388139e-02 1.50795972e+00
7.03316331e-02 -2.28536710e-01 -5.54992497e-01 -5.34198940e-01
6.90711677e-01 -5.35896420e-01 8.52301776e-01 8.61449003e-01
2.03994468e-01 -1.15629919e-02 4.76751119e-01 2.66857326e-01
9.02807951e-01 -1.17673250e-02 -8.76470089e-01 3.94525081e-02
5.23296475e-01 9.01310980e-01 -1.39510727e+00 1.48736000e-01
-4.97965813e-01 6.03630066e-01 5.12558967e-02 2.99111038e-01
-5.95076501e-01 -4.15253490e-01 8.94549608e-01 1.20385356e-01
-1.91587787e-02 -7.78439283e-01 -4.40268308e-01 -5.88244319e-01
-1.76838323e-01 -4.51392531e-01 4.83133018e-01 -3.64437759e-01
-9.82533097e-01 5.13707995e-01 5.37183762e-01 -1.00885379e+00
-5.43824375e-01 -6.22340441e-01 -8.19175422e-01 2.43041337e-01
-1.82937586e+00 -7.83224523e-01 -1.87577829e-02 4.42030400e-01
6.26948714e-01 -1.35249421e-01 9.78507400e-01 1.34973094e-01
-4.50659364e-01 -1.11231349e-01 1.57895774e-01 -5.64068854e-01
3.12246650e-01 -1.11150014e+00 9.32428777e-01 1.01017642e+00
-2.60669500e-01 5.33455908e-01 1.00958574e+00 -5.00882804e-01
-1.91109431e+00 -8.68697941e-01 7.15802133e-01 -2.48690799e-01
1.02411175e+00 8.11857656e-02 -2.66631365e-01 1.03436565e+00
3.24170142e-01 -2.74645150e-01 3.11713606e-01 -1.79918054e-02
4.84504253e-01 4.58106190e-01 -1.35560560e+00 6.75495982e-01
1.08006513e+00 3.69518846e-01 -3.60046536e-01 3.77462626e-01
5.92003882e-01 -2.84787565e-01 -6.55681312e-01 2.16603726e-01
4.86022234e-01 -7.43877709e-01 1.10329592e+00 -3.67089957e-01
-1.94519833e-01 -4.21034366e-01 -1.56462789e-01 -1.56038833e+00
-4.60593164e-01 -8.85195494e-01 1.86898902e-01 4.87518102e-01
2.39062712e-01 -1.42680800e+00 9.03840542e-01 1.79963008e-01
-2.25239396e-01 -7.04647183e-01 -1.21799195e+00 -9.93162334e-01
3.99466723e-01 -2.59406477e-01 7.70617783e-01 9.57809925e-01
5.17976165e-01 1.16337389e-01 -1.07973076e-01 3.83811921e-01
6.22864962e-01 1.81246907e-01 8.95171344e-01 -1.28533852e+00
6.91974312e-02 -6.48927033e-01 -6.16138518e-01 -9.25737083e-01
2.33140394e-01 -3.20824593e-01 2.74475992e-01 -2.14679289e+00
-3.15407693e-01 -7.94988036e-01 9.19898525e-02 2.40383551e-01
2.94254184e-01 -1.74998194e-01 3.16422969e-01 1.49647713e-01
-6.84584916e-01 7.24632144e-02 1.34659052e+00 2.58995201e-02
-1.47708490e-01 1.88711911e-01 -6.51465654e-01 7.58962750e-01
1.13950241e+00 -5.14910698e-01 -6.12033129e-01 -5.09011447e-01
2.99198240e-01 3.43838513e-01 8.49300474e-02 -6.52765632e-01
6.29483104e-01 -1.10215902e+00 -6.32910311e-01 -3.47722441e-01
2.62836069e-01 -1.15134275e+00 1.58714294e-01 8.82091105e-01
3.03606987e-01 7.33076394e-01 1.10186776e-02 5.68322003e-01
-6.28372235e-03 -6.59143925e-01 5.65406084e-01 -5.43068171e-01
-1.25445688e+00 1.15786791e-01 -9.02894497e-01 -1.50552675e-01
1.89136505e+00 -3.99421334e-01 -6.97187722e-01 -5.23365319e-01
-4.20307755e-01 6.87791646e-01 7.62859046e-01 4.54805166e-01
6.80726171e-01 -1.06108546e+00 -5.37674010e-01 -6.74986243e-02
-2.25164723e-02 -7.45276138e-02 -2.82026798e-01 8.11513424e-01
-1.13063633e+00 5.46725750e-01 -4.63556349e-01 -1.71994522e-01
-8.44289482e-01 7.40384936e-01 1.59517244e-01 -1.66080520e-01
-6.61857724e-01 4.09092814e-01 -1.03190608e-01 -5.76819539e-01
1.77124783e-01 8.70513394e-02 -4.12204303e-02 -3.91091257e-01
5.32986403e-01 1.06821597e+00 -1.42330468e-01 -4.63347882e-01
-8.68977726e-01 5.80178201e-01 5.53677976e-02 -4.50401545e-01
1.48395455e+00 -5.25433362e-01 -3.87078285e-01 2.93124765e-01
7.00294495e-01 -3.93523574e-01 -3.93222511e-01 1.45137906e-01
4.95157033e-01 -3.72027040e-01 -1.48346707e-01 -5.77355921e-01
-5.08798957e-01 3.01906794e-01 1.31741256e-01 6.65805638e-01
9.16624546e-01 -4.92888801e-02 6.13358855e-01 5.53708315e-01
1.27937222e+00 -9.36351180e-01 -2.80709028e-01 7.58177102e-01
8.76064181e-01 -1.05715501e+00 1.95352942e-01 -7.61200547e-01
-4.31426495e-01 1.25938618e+00 4.66406763e-01 -6.15522265e-01
9.06083882e-01 5.60430050e-01 1.90231681e-01 -4.09412175e-01
-6.58303916e-01 -2.93981552e-01 -4.68833029e-01 1.00443268e+00
-2.63594538e-01 5.16913943e-02 -9.20297325e-01 -1.07235409e-01
-3.01410377e-01 -3.93346213e-02 1.16711080e+00 1.71607971e+00
-7.65661061e-01 -1.38617265e+00 -6.86650753e-01 2.83855885e-01
-3.93231958e-02 4.24583733e-01 -5.30951321e-01 1.16629231e+00
-1.70644999e-01 1.25928247e+00 2.61300243e-02 -1.70019373e-01
2.04027280e-01 -5.94202936e-01 5.51390350e-01 -3.88985068e-01
-2.07476124e-01 -4.41508330e-02 4.97458309e-01 -6.08719647e-01
-8.07060122e-01 -9.49561059e-01 -1.55537212e+00 -6.05823815e-01
-2.92558044e-01 5.02380490e-01 9.39736724e-01 1.07280982e+00
2.32041836e-01 1.44938022e-01 5.11779666e-01 -1.05899334e+00
-1.24647655e-01 -3.37033927e-01 -4.26531792e-01 -3.90236497e-01
2.96239108e-01 -1.08347189e+00 -1.71181157e-01 -6.22934818e-01]
|
[4.977100372314453, 1.7156450748443604]
|
1302acb7-52f4-465b-97b1-1a88f4154915
|
bayesian-neural-networks-essentials
|
2106.13594
| null |
https://arxiv.org/abs/2106.13594v1
|
https://arxiv.org/pdf/2106.13594v1.pdf
|
Bayesian Neural Networks: Essentials
|
Bayesian neural networks utilize probabilistic layers that capture uncertainty over weights and activations, and are trained using Bayesian inference. Since these probabilistic layers are designed to be drop-in replacement of their deterministic counter parts, Bayesian neural networks provide a direct and natural way to extend conventional deep neural networks to support probabilistic deep learning. However, it is nontrivial to understand, design and train Bayesian neural networks due to their complexities. We discuss the essentials of Bayesian neural networks including duality (deep neural networks, probabilistic models), approximate Bayesian inference, Bayesian priors, Bayesian posteriors, and deep variational learning. We use TensorFlow Probability APIs and code examples for illustration. The main problem with Bayesian neural networks is that the architecture of deep neural networks makes it quite redundant, and costly, to account for uncertainty for a large number of successive layers. Hybrid Bayesian neural networks, which use few probabilistic layers judicially positioned in the networks, provide a practical solution.
|
['Daniel T. Chang']
|
2021-06-22
| null | null | null | null |
['probabilistic-deep-learning']
|
['computer-vision']
|
[-3.79388332e-01 3.93767238e-01 -4.40147035e-02 -8.48351181e-01
-3.72923315e-01 -4.15436208e-01 7.05011189e-01 -6.34050667e-01
-3.49486977e-01 6.89790308e-01 1.59104243e-01 -6.22930646e-01
-4.73488599e-01 -6.83031380e-01 -8.29384327e-01 -6.00973248e-01
-2.07641855e-01 6.13222361e-01 4.31800544e-01 4.20944124e-01
-2.16019433e-02 4.74767387e-01 -1.47001028e+00 1.45879105e-01
2.37126440e-01 1.06722915e+00 -4.15711626e-02 6.66062951e-01
-2.53687590e-01 9.38417017e-01 -4.67876315e-01 -8.01042438e-01
-1.99912414e-01 4.09599066e-01 -4.14895833e-01 -8.55289757e-01
2.47997329e-01 -1.06681263e+00 -6.46775424e-01 1.11948907e+00
1.72848895e-01 2.62803733e-01 1.14861178e+00 -1.36331093e+00
-4.38748717e-01 1.40763807e+00 -4.11856264e-01 9.54195634e-02
-4.23539430e-01 2.63830479e-02 1.05137706e+00 -7.95546889e-01
-6.89089894e-02 1.80541110e+00 1.11428738e+00 7.01066196e-01
-1.26948595e+00 -6.58137023e-01 2.93048859e-01 -5.23644127e-02
-1.46804714e+00 -6.40563428e-01 5.31092405e-01 -5.59106946e-01
9.64976788e-01 -1.77544549e-01 4.47585970e-01 1.70143723e+00
3.28215718e-01 9.57086682e-01 6.34975374e-01 -3.35968584e-02
5.46575665e-01 -2.79699772e-04 6.65551722e-01 5.62414587e-01
4.75445896e-01 5.21737278e-01 -6.10162973e-01 -4.88900632e-01
8.36817563e-01 2.91573316e-01 1.62606746e-01 -1.49123460e-01
-5.64723015e-01 8.38910401e-01 1.18172310e-01 -3.42568547e-01
-1.82799205e-01 1.22482419e+00 1.25983089e-01 -5.79253674e-01
1.15926370e-01 -1.10501289e-01 -7.26073980e-01 -3.45546901e-01
-1.26328087e+00 5.77047944e-01 1.06793964e+00 1.06256020e+00
4.82024729e-01 2.82982677e-01 -1.76661253e-01 7.07643151e-01
1.25376797e+00 5.73563218e-01 -2.38156661e-01 -1.53041375e+00
-1.13244802e-01 -3.94469321e-01 2.70137247e-02 -6.21501565e-01
-2.55577505e-01 -3.68865758e-01 -8.33502352e-01 3.41829926e-01
6.12911701e-01 -4.82957125e-01 -1.02208543e+00 1.88322842e+00
-1.33075103e-01 3.47508162e-01 -1.10630639e-01 5.04113615e-01
7.11623430e-01 7.36074626e-01 2.03399330e-01 4.42482084e-01
1.38652945e+00 -2.20354021e-01 -5.78932583e-01 -2.50858992e-01
-7.01830760e-02 -4.34727699e-01 6.30994141e-01 7.67511845e-01
-1.15212357e+00 -1.20010376e-01 -1.10643613e+00 -1.08015388e-01
-3.22374195e-01 -6.42462522e-02 1.06688714e+00 1.13611746e+00
-1.00380564e+00 7.98441410e-01 -1.53625584e+00 2.34040782e-01
7.07031727e-01 2.92289287e-01 3.18154991e-01 -1.08574130e-01
-1.44224691e+00 9.44691300e-01 7.28775501e-01 4.54581976e-01
-1.11646688e+00 -6.66408002e-01 -8.22161615e-01 5.53152204e-01
2.10248269e-02 -5.29100835e-01 1.54686451e+00 -2.20927656e-01
-1.70390093e+00 5.00693731e-02 -4.35929745e-02 -5.04287779e-01
2.24046439e-01 -2.53210694e-01 7.60633638e-03 -1.19734921e-01
-5.59344530e-01 1.04379845e+00 8.05934429e-01 -9.76513445e-01
-5.45848608e-01 -1.20907828e-01 1.36440605e-01 -3.71504903e-01
-1.55932561e-01 4.65057753e-02 -6.66283131e-01 -3.26222301e-01
2.35805094e-01 -7.69334674e-01 -1.47047967e-01 -2.73458119e-02
-6.11961305e-01 -5.25761366e-01 4.73431408e-01 -3.99365991e-01
8.34153116e-01 -2.04429483e+00 -3.20613533e-01 4.87754583e-01
3.56746316e-01 -2.90643632e-01 1.84459135e-01 -3.60271484e-01
1.53471112e-01 3.66216183e-01 -9.49377045e-02 -5.05639195e-01
7.55171359e-01 6.35866463e-01 -6.08782113e-01 2.57021397e-01
3.45296592e-01 7.40434408e-01 -6.07774734e-01 -4.91969407e-01
8.05684701e-02 8.80096078e-01 -7.51932621e-01 -4.29771543e-02
-5.78803301e-01 -4.44802940e-01 -2.03989998e-01 5.72984815e-01
8.88350189e-01 -1.01383887e-01 4.61299866e-01 -3.66408736e-01
1.89524621e-01 5.76122761e-01 -1.53664637e+00 1.50423717e+00
-1.15889683e-01 7.45816469e-01 -7.94441497e-04 -5.15895665e-01
4.84537095e-01 2.60894299e-01 3.88057418e-02 2.39962667e-01
3.34721833e-01 -2.76670102e-02 -3.28114741e-02 8.88171606e-03
6.20887160e-01 -9.05976221e-02 -1.65146023e-01 6.90930784e-01
5.34038603e-01 -2.35937893e-01 2.41442919e-01 4.05823499e-01
9.33713853e-01 6.26317501e-01 -4.90029037e-01 -3.15686643e-01
-5.37176669e-01 -4.60214198e-01 8.13464284e-01 1.28625882e+00
-1.08172581e-01 7.04657257e-01 1.06414354e+00 -4.18333977e-01
-7.25147367e-01 -1.75128138e+00 -4.69746590e-01 1.07704377e+00
-4.52944934e-01 -3.57497334e-01 -7.30137229e-01 -3.92499745e-01
5.36302887e-02 1.01552641e+00 -4.46001559e-01 -1.55099109e-01
4.08542529e-02 -1.16231763e+00 9.35106814e-01 9.24632013e-01
9.24985111e-02 -4.33249027e-01 -5.47482669e-01 3.31336290e-01
2.28199095e-01 -9.30574834e-01 -4.35265936e-02 6.93154693e-01
-9.96487796e-01 -7.35216618e-01 -5.47381341e-01 7.85448104e-02
1.51681080e-01 -2.79733807e-01 1.23682857e+00 -4.72485751e-01
3.29121351e-02 2.85440594e-01 3.29292178e-01 -7.23378718e-01
-2.41887704e-01 -3.93541493e-02 3.07899296e-01 -5.52828252e-01
5.68603694e-01 -9.35879469e-01 -3.36772621e-01 -1.10580139e-02
-8.27791870e-01 -1.78853557e-01 5.85750222e-01 7.71070242e-01
1.81509733e-01 3.06145161e-01 3.59287299e-02 -7.54274905e-01
4.80028540e-01 -5.47830760e-01 -1.10125279e+00 2.19598457e-01
-6.00048423e-01 5.17268300e-01 -1.34025097e-01 -4.69974697e-01
-1.34413540e+00 -2.10636407e-01 -2.08692819e-01 -5.01550972e-01
-8.11472386e-02 6.97839797e-01 -8.15273672e-02 6.54031813e-01
5.27876318e-01 -4.17375803e-01 -1.63451537e-01 -5.97431779e-01
7.03272820e-01 5.63265800e-01 5.73456526e-01 -1.28840208e+00
2.00474203e-01 3.87227505e-01 -6.85573220e-02 -1.34002909e-01
-9.87417221e-01 4.15836781e-01 -3.83186847e-01 -1.14682168e-01
8.30632687e-01 -1.01485300e+00 -1.26326668e+00 5.12269199e-01
-1.72120023e+00 -3.20394754e-01 -9.79247168e-02 8.14962506e-01
-2.14793995e-01 -1.38238862e-01 -8.90835345e-01 -1.16593742e+00
-1.34987496e-02 -1.39307332e+00 4.96231258e-01 3.86036277e-01
-3.41720164e-01 -1.02214062e+00 -2.45168135e-01 -7.67389759e-02
5.93853533e-01 -2.95587182e-01 9.09137011e-01 -5.80655694e-01
-6.98940337e-01 -1.90153629e-01 -5.84628105e-01 6.13028765e-01
-4.12477106e-01 9.06730056e-01 -1.49162912e+00 3.33847016e-01
-1.61475956e-01 -3.26344103e-01 1.17555833e+00 1.00516474e+00
1.36486745e+00 -2.11949915e-01 -1.96322918e-01 6.10523820e-01
9.68711257e-01 -1.04125850e-01 6.00009739e-01 -3.55804980e-01
6.86739802e-01 4.85248923e-01 -4.38159764e-01 5.42198777e-01
6.23091757e-01 -1.80250090e-02 5.34867585e-01 5.15570045e-01
1.04764804e-01 -7.11646751e-02 4.05549735e-01 5.09741426e-01
-9.52689350e-03 -1.05428390e-01 -1.19140160e+00 2.33820215e-01
-1.85502291e+00 -1.17068982e+00 -9.39062983e-02 1.79333460e+00
1.18319011e+00 7.36892879e-01 -2.42598027e-01 -2.02849269e-01
5.16340077e-01 -9.06021334e-03 -5.54300547e-01 -3.60642225e-01
1.41291484e-01 3.14066082e-01 5.56214750e-01 5.03548026e-01
-1.03995919e+00 6.95027590e-01 7.99918604e+00 1.04702592e+00
-5.72070956e-01 2.85419226e-01 8.17160964e-01 -4.78723675e-01
-5.39138734e-01 1.51974663e-01 -1.27813363e+00 4.62182730e-01
1.46301532e+00 6.52624905e-01 4.30034220e-01 1.00551283e+00
-2.64974069e-02 -3.42420965e-01 -1.38980591e+00 8.73678088e-01
-5.28366446e-01 -1.73808432e+00 -9.12741479e-03 -4.52344632e-03
6.30106091e-01 5.68436623e-01 2.44161859e-01 4.62554514e-01
1.61522079e+00 -1.27202678e+00 1.22185707e+00 1.01059556e+00
4.17827576e-01 -9.18016016e-01 8.11882257e-01 7.48424307e-02
-4.72424716e-01 1.21096797e-01 -5.79087853e-01 -1.81209385e-01
1.69187233e-01 1.23632061e+00 -3.21654648e-01 -2.00704291e-01
1.14464033e+00 4.78705466e-01 6.68636858e-02 6.32168353e-01
-4.39964741e-01 8.47656965e-01 -9.16125238e-01 -1.56009331e-01
1.95997685e-01 -7.67165497e-02 1.05446272e-01 1.17222297e+00
2.38339558e-01 -1.61795691e-01 -3.56331289e-01 1.73318684e+00
-1.13330327e-01 -1.15740299e+00 -2.38157183e-01 -1.54670969e-01
8.84337068e-01 1.03736317e+00 -6.25988126e-01 -3.84190798e-01
-3.41612190e-01 8.39276537e-02 2.55206764e-01 6.64790630e-01
-1.00734806e+00 -4.26624745e-01 9.29431558e-01 -5.43180168e-01
3.32340598e-01 -4.04900342e-01 -5.91070592e-01 -9.45112646e-01
-4.07402009e-01 -4.21941936e-01 1.34596869e-01 -9.59767342e-01
-1.47772264e+00 2.01296449e-01 5.12302101e-01 -2.73525000e-01
-5.64076185e-01 -1.19222355e+00 -3.84199828e-01 1.11922967e+00
-1.22394621e+00 -8.95665348e-01 3.38545322e-01 2.74169028e-01
-1.35946609e-02 -1.10310428e-02 6.94223046e-01 1.01551771e-01
-8.05877507e-01 5.81126809e-01 3.89661878e-01 2.55518556e-01
2.06013158e-01 -1.33781636e+00 3.87132138e-01 7.48010516e-01
5.62383682e-02 1.18890178e+00 7.86127567e-01 -3.10631633e-01
-1.31496310e+00 -6.61486208e-01 3.98003310e-01 -6.98848665e-01
7.64741004e-01 -5.23289621e-01 -6.92414701e-01 7.87767649e-01
-1.42267132e-02 -1.02929108e-01 7.00884759e-01 7.02492893e-01
-9.35332835e-01 -2.44444802e-01 -1.05086458e+00 7.69705474e-01
4.98244673e-01 -6.42161548e-01 -7.25532770e-01 5.33610918e-02
8.75756860e-01 -3.87024671e-01 -9.95305240e-01 1.50784954e-01
1.23295331e+00 -1.05236900e+00 1.06537926e+00 -4.58148271e-01
4.71204609e-01 -2.17192441e-01 -5.18834054e-01 -8.27898443e-01
-7.85560235e-02 -8.00156772e-01 -6.93594813e-01 1.32964122e+00
5.75751007e-01 -5.62895715e-01 8.94474626e-01 1.42094338e+00
-1.36429176e-01 -4.81318742e-01 -9.67854977e-01 -6.58512115e-01
3.34605545e-01 -1.43213773e+00 8.55357111e-01 4.76348191e-01
-3.01007926e-01 -1.87855303e-01 -5.22543751e-02 3.07250917e-01
1.11284304e+00 -5.48558772e-01 1.63420349e-01 -1.47207034e+00
-5.06674886e-01 -8.97126615e-01 4.03085276e-02 -1.16601443e+00
2.51046240e-01 -4.97762322e-01 4.52141792e-01 -1.38269174e+00
3.64635587e-01 -5.40766656e-01 -4.83362108e-01 7.01531053e-01
9.00718719e-02 -2.80694850e-02 -2.29903072e-01 -2.18768716e-02
-2.98451990e-01 5.03728330e-01 2.17428431e-01 -2.28040919e-01
1.56828210e-01 1.91128477e-01 -8.08446765e-01 1.22673368e+00
6.12728477e-01 -8.36825311e-01 -5.06260097e-01 -8.64696443e-01
6.98736608e-01 -2.14203373e-01 5.90308964e-01 -8.73162866e-01
4.94987309e-01 -2.09535658e-01 6.84434891e-01 -1.01135421e+00
6.08974099e-01 -6.16247714e-01 1.59689501e-01 1.41480371e-01
-3.90588671e-01 -2.45123357e-01 3.09777409e-01 6.02114141e-01
1.88721970e-01 -5.73347032e-01 6.60453439e-01 -1.02130182e-01
-2.67935485e-01 3.17477971e-01 -8.48832190e-01 -9.79191139e-02
1.63715899e-01 1.10584937e-01 -2.54538208e-01 -1.20177358e-01
-7.71747768e-01 1.97430700e-01 -7.22017288e-02 -3.47712748e-02
6.33804977e-01 -1.04634392e+00 -3.41510683e-01 -2.73754656e-01
-4.83428270e-01 4.63769495e-01 2.28890046e-01 3.68728548e-01
-3.03135544e-01 2.86769271e-01 8.76672491e-02 -6.84037566e-01
-2.64633864e-01 -5.62393144e-02 4.81109619e-01 3.12699080e-02
-1.62609577e-01 1.24293005e+00 -1.60241932e-01 -5.67212403e-01
8.15576851e-01 -8.54413450e-01 2.37531677e-01 -5.28407060e-02
5.86196244e-01 4.13286448e-01 -1.59839079e-01 4.01987344e-01
-4.08392787e-01 -1.38986155e-01 -1.04406200e-01 -7.09897339e-01
1.42356181e+00 2.05912031e-02 -2.30948254e-01 6.50890708e-01
6.51742458e-01 -6.06939495e-01 -1.90276349e+00 -3.22781622e-01
8.77663717e-02 2.02580109e-01 6.99363291e-01 -8.26669157e-01
-1.09320104e+00 1.44288468e+00 4.47467774e-01 -9.63078812e-02
4.29269284e-01 -9.57245678e-02 4.44532990e-01 8.05437207e-01
2.70380042e-02 -1.08939040e+00 -3.94531012e-01 1.00215161e+00
3.66300434e-01 -9.56923962e-01 2.33564392e-01 3.43311727e-01
-9.46322754e-02 1.34916830e+00 5.03399372e-01 1.18404003e-02
1.36876535e+00 8.63049507e-01 -4.56811219e-01 -1.80462629e-01
-1.06478107e+00 1.95189282e-01 1.22446306e-01 7.40227461e-01
4.45041418e-01 5.99550083e-02 5.98002017e-01 1.26964951e+00
-1.33479327e-01 8.66022632e-02 4.57221746e-01 7.38775253e-01
-2.25377783e-01 -7.42492855e-01 -4.07513231e-01 6.15133047e-01
-5.65078020e-01 -6.39355004e-01 5.18440723e-01 4.65610862e-01
-5.08097149e-02 8.48881900e-01 4.29294944e-01 -2.93176174e-01
-2.56043524e-01 2.14740857e-01 5.28617203e-01 -4.99576479e-01
-2.29391251e-02 -4.03476208e-02 8.91486108e-02 -4.26039338e-01
-6.02669977e-02 -7.16808677e-01 -1.19987690e+00 -6.72683179e-01
-2.25297838e-01 -1.01378798e-01 1.23845994e+00 1.21228373e+00
3.04507196e-01 6.88380480e-01 -1.23446025e-01 -1.14392018e+00
-1.05755186e+00 -1.09750700e+00 -8.83623183e-01 -5.54256499e-01
7.18822256e-02 -8.37479949e-01 -3.17124516e-01 8.53567198e-02]
|
[7.267587661743164, 3.875962018966675]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.