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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
7df3744b-aabc-4930-9257-5234074e0261
|
knowledge-injected-prompt-based-fine-tuning
|
2210.03304
| null |
https://arxiv.org/abs/2210.03304v2
|
https://arxiv.org/pdf/2210.03304v2.pdf
|
Knowledge Injected Prompt Based Fine-tuning for Multi-label Few-shot ICD Coding
|
Automatic International Classification of Diseases (ICD) coding aims to assign multiple ICD codes to a medical note with average length of 3,000+ tokens. This task is challenging due to a high-dimensional space of multi-label assignment (tens of thousands of ICD codes) and the long-tail challenge: only a few codes (common diseases) are frequently assigned while most codes (rare diseases) are infrequently assigned. This study addresses the long-tail challenge by adapting a prompt-based fine-tuning technique with label semantics, which has been shown to be effective under few-shot setting. To further enhance the performance in medical domain, we propose a knowledge-enhanced longformer by injecting three domain-specific knowledge: hierarchy, synonym, and abbreviation with additional pretraining using contrastive learning. Experiments on MIMIC-III-full, a benchmark dataset of code assignment, show that our proposed method outperforms previous state-of-the-art method in 14.5% in marco F1 (from 10.3 to 11.8, P<0.001). To further test our model on few-shot setting, we created a new rare diseases coding dataset, MIMIC-III-rare50, on which our model improves marco F1 from 17.1 to 30.4 and micro F1 from 17.2 to 32.6 compared to previous method.
|
['Hong Yu', 'Avijit Mitra', 'Bhanu Pratap Singh Rawat', 'Shufan Wang', 'Zhichao Yang']
|
2022-10-07
| null | null | null | null |
['medical-code-prediction']
|
['medical']
|
[ 2.62842327e-01 -4.08542827e-02 -4.97094572e-01 -2.53282130e-01
-1.06132829e+00 -3.99280429e-01 1.61267325e-01 4.83213335e-01
-3.97338957e-01 8.17784727e-01 4.98922646e-01 1.45731959e-02
-3.00339997e-01 -4.80107039e-01 -2.15495139e-01 -3.01606804e-01
-1.92630693e-01 7.23709941e-01 1.94499284e-01 -1.91457253e-02
6.01366647e-02 -1.57487705e-01 -1.40940404e+00 6.04948163e-01
1.09578156e+00 8.25971544e-01 6.06308989e-02 4.73532021e-01
-1.85891807e-01 1.00422418e+00 -4.05390769e-01 -6.40412688e-01
-1.41853979e-02 -3.98717850e-01 -9.80677247e-01 -1.65793940e-01
1.91669896e-01 -2.10481137e-01 -9.54603031e-02 1.06352901e+00
6.39500439e-01 1.41583830e-01 1.09226680e+00 -9.48729813e-01
-9.67348337e-01 7.60233104e-01 -6.32575750e-01 2.19854161e-01
6.12325668e-01 -2.56253392e-01 1.20949781e+00 -7.91818798e-01
8.23459506e-01 1.08436513e+00 9.32990372e-01 8.22086513e-01
-9.71840501e-01 -9.61681843e-01 -1.62726894e-01 2.26624101e-01
-1.67000663e+00 -2.04336718e-01 1.57858685e-01 -7.31540024e-01
1.18010259e+00 5.36396615e-02 1.25761852e-01 1.07447255e+00
1.55718967e-01 4.17047024e-01 7.76726961e-01 -4.67143714e-01
3.06025118e-01 -9.93478298e-03 2.32837260e-01 8.18019569e-01
1.79906353e-01 -1.13693461e-01 -1.39993310e-01 -4.77243334e-01
3.70019466e-01 5.41462123e-01 -3.01843546e-02 7.76552632e-02
-1.44583666e+00 9.09467936e-01 2.29155108e-01 2.36134082e-01
-4.56488132e-01 -6.33232528e-03 7.85782456e-01 2.13753596e-01
3.17102134e-01 6.06050193e-01 -6.51548564e-01 -4.57042098e-01
-8.49859178e-01 3.55013877e-01 7.13619471e-01 1.22588694e+00
4.06815290e-01 -3.99648935e-01 -5.91262579e-01 1.40519464e+00
-1.34479731e-01 4.74477470e-01 9.19812858e-01 -8.04474354e-01
5.79157591e-01 5.90497255e-01 3.75599526e-02 -7.26638317e-01
-6.63267493e-01 -5.80928445e-01 -1.23631752e+00 -5.98041117e-01
5.82869388e-02 -3.56919825e-01 -1.11574328e+00 1.75111449e+00
6.49776831e-02 5.22635460e-01 7.84743652e-02 5.66568077e-01
1.02254391e+00 4.43160564e-01 3.37314457e-01 -3.00295413e-01
1.77262568e+00 -8.98092568e-01 -8.26332510e-01 1.54097557e-01
1.07040203e+00 -7.04631865e-01 1.05144942e+00 2.10880756e-01
-7.18559861e-01 -4.88734782e-01 -4.43459779e-01 1.19697802e-01
-1.97473347e-01 3.50381806e-02 5.45261562e-01 3.59079659e-01
-5.12123704e-01 3.49311203e-01 -4.02095258e-01 -5.73929071e-01
7.67704189e-01 8.36281702e-02 -2.31056347e-01 -4.72750098e-01
-1.65369785e+00 6.79067731e-01 6.31829321e-01 -1.02190852e+00
-7.59839177e-01 -1.27403665e+00 -8.41470599e-01 2.62725800e-01
3.32195282e-01 -7.77272165e-01 1.09918189e+00 -3.92400712e-01
-9.14086521e-01 1.13355982e+00 8.63240063e-02 -4.59477812e-01
2.29981229e-01 -2.42505267e-01 -8.87943983e-01 2.83236682e-01
5.48147738e-01 7.65896261e-01 2.74397403e-01 -6.70073450e-01
-8.44764948e-01 4.53390107e-02 1.42378241e-01 1.42339408e-01
-6.09608293e-01 6.80576265e-02 -3.28512460e-01 -1.27952623e+00
-3.45416576e-01 -1.05894160e+00 -7.12575018e-02 -1.97901055e-01
-3.00427407e-01 -3.48080963e-01 -9.99605656e-02 -4.16552365e-01
1.85924923e+00 -2.22340870e+00 -1.54143453e-01 -6.37371689e-02
2.10226804e-01 1.43919915e-01 -1.17418263e-02 5.11059582e-01
-2.26835921e-01 6.33043051e-02 -5.36946893e-01 -1.76399186e-01
-7.98004791e-02 3.39249432e-01 -2.40929246e-01 1.87648118e-01
1.04657775e-02 6.75461292e-01 -1.15939510e+00 -8.36443186e-01
-1.22884296e-01 1.07610516e-01 -1.00100291e+00 9.38908383e-02
2.01561809e-01 -1.01275882e-02 -1.37972638e-01 1.03922415e+00
5.41022301e-01 -6.99557602e-01 -3.29451747e-02 7.01761767e-02
3.13646078e-01 1.94372665e-02 -1.10213697e+00 1.99215972e+00
-4.87853527e-01 -8.96429643e-02 -7.61849046e-01 -9.84768271e-01
6.00754678e-01 5.76350152e-01 7.92742610e-01 -2.98128158e-01
1.87622961e-02 1.39046282e-01 5.48056997e-02 -8.65372598e-01
1.11826353e-01 -5.48474789e-01 -4.98460263e-01 2.49946833e-01
2.88062721e-01 2.22886652e-01 2.92109698e-01 3.96234661e-01
1.58054256e+00 -5.55599868e-01 8.05372894e-01 -2.34680578e-01
2.23125368e-01 -9.74432305e-02 8.12769115e-01 8.64455998e-01
-3.28185081e-01 8.79076958e-01 4.74227309e-01 -5.25760174e-01
-7.29720473e-01 -1.05245197e+00 -5.70399284e-01 1.22760677e+00
-1.42890528e-01 -4.49738890e-01 -5.51257014e-01 -9.87507999e-01
2.97471762e-01 6.28929377e-01 -8.91823530e-01 -2.68505633e-01
-1.03457429e-01 -8.93069685e-01 9.32164311e-01 5.27484238e-01
1.75355777e-01 -1.16166782e+00 -4.94358271e-01 3.45382959e-01
-4.96896982e-01 -1.24491024e+00 -8.20043921e-01 1.54942527e-01
-5.33775449e-01 -1.11707044e+00 -1.11931527e+00 -8.79147589e-01
6.10004425e-01 -9.95611176e-02 1.34891057e+00 4.52135131e-02
-4.74202335e-01 3.33883278e-02 -8.88477683e-01 -4.09815520e-01
-1.71420306e-01 4.47732121e-01 2.37629116e-01 -8.47878680e-02
5.98039389e-01 -3.37731928e-01 -5.14441371e-01 8.89934599e-02
-9.61909413e-01 1.92243345e-02 6.45939946e-01 1.06379068e+00
4.30317909e-01 -1.40306875e-01 9.53783810e-01 -1.37627864e+00
4.97312605e-01 -8.99142742e-01 -1.33496448e-02 2.71847606e-01
-6.99080586e-01 -1.10920165e-02 7.04609692e-01 -5.34840882e-01
-7.15885937e-01 3.86483371e-02 -2.75103480e-01 -2.84562856e-01
-3.06097507e-01 7.58770227e-01 3.23281437e-01 3.70767623e-01
9.63901699e-01 2.08342206e-02 -5.08819282e-01 -5.15508711e-01
1.94793314e-01 1.07533324e+00 4.95225132e-01 -3.46539527e-01
3.74810934e-01 3.12590957e-01 -2.60186106e-01 -1.47820517e-01
-1.39477098e+00 -9.63562429e-01 -4.04404223e-01 3.37665558e-01
8.80052805e-01 -1.24149895e+00 -4.93315607e-01 1.41526848e-01
-9.21791911e-01 -8.95990208e-02 -2.34700814e-01 6.70290470e-01
-4.19465363e-01 3.05740297e-01 -7.43759453e-01 -1.80489048e-01
-5.05386353e-01 -8.29569221e-01 1.00847363e+00 -2.36027315e-01
-8.32456648e-01 -8.77343416e-01 3.76544207e-01 2.57299393e-01
1.24663927e-01 3.73232573e-01 1.20283508e+00 -7.73950756e-01
4.22029883e-01 -1.03817489e-02 -4.01919156e-01 1.79446340e-02
5.32448769e-01 -5.06271005e-01 -6.81828201e-01 -1.22190900e-01
-3.62826943e-01 -5.96009374e-01 1.03083837e+00 3.72180611e-01
1.45190895e+00 -2.20792562e-01 -4.77435738e-01 5.95550060e-01
1.54440367e+00 2.53618509e-01 3.05385679e-01 1.40660450e-01
7.94011474e-01 2.69718766e-01 7.54357338e-01 1.09438622e+00
5.80241501e-01 8.20726812e-01 1.52731493e-01 2.07090333e-01
-1.47328168e-01 1.12866573e-01 -8.28096420e-02 1.00281274e+00
9.67347026e-02 1.06530767e-02 -1.40363455e+00 9.51045692e-01
-1.86151838e+00 -1.05888665e+00 1.67242706e-01 1.97627020e+00
1.52644253e+00 2.26046398e-01 1.75243109e-01 7.93720633e-02
8.26066196e-01 -4.03873712e-01 -5.05701602e-01 -2.95346141e-01
2.24356860e-01 2.40009397e-01 4.01765138e-01 2.35350907e-01
-1.31446373e+00 7.13198185e-01 6.01926804e+00 1.14994931e+00
-6.70766652e-01 3.82111311e-01 4.57490414e-01 -2.62880594e-01
-9.15683359e-02 -4.96884167e-01 -7.92851329e-01 1.05985320e+00
1.13177848e+00 -2.34791979e-01 1.95916325e-01 8.12379777e-01
-3.08805376e-01 2.04034045e-01 -9.93933320e-01 1.44908214e+00
2.10039139e-01 -1.28051460e+00 2.65204102e-01 -1.62562773e-01
1.21959257e+00 9.48326513e-02 -5.25139309e-02 8.30021679e-01
4.06287909e-01 -8.26652765e-01 4.31564450e-01 4.98319805e-01
1.47554636e+00 -8.63646686e-01 1.06205511e+00 1.94927588e-01
-1.24973714e+00 -3.68608385e-01 -4.98406291e-01 -8.29173699e-02
1.61190972e-01 8.95346642e-01 -6.87049925e-01 4.69235122e-01
6.76196396e-01 1.06765616e+00 -3.86282027e-01 1.08720160e+00
1.91597119e-01 6.52126431e-01 6.12778924e-02 2.44045302e-01
2.93571144e-01 7.18529582e-01 -1.69374757e-02 1.68334746e+00
5.40389836e-01 4.12101001e-01 4.00843441e-01 3.80784392e-01
-4.84086126e-01 1.55788407e-01 -3.78054470e-01 1.69521108e-01
6.84270978e-01 1.06776309e+00 -6.35042489e-01 -7.61384964e-01
-4.97155130e-01 9.78173614e-01 3.03683639e-01 9.63207036e-02
-1.27842760e+00 -9.05332327e-01 5.23077309e-01 -8.39480609e-02
3.81179810e-01 6.24667406e-01 -1.35208711e-01 -1.27983117e+00
-3.65432113e-01 -8.98972332e-01 9.91739511e-01 -5.35611629e-01
-1.58265626e+00 5.37471712e-01 -1.09230116e-01 -1.74403906e+00
-2.08749771e-01 -1.84957191e-01 5.49841896e-02 3.76139790e-01
-1.46294010e+00 -9.25972581e-01 -2.80496985e-01 6.58430696e-01
6.41764700e-01 -4.46131200e-01 1.32522810e+00 9.16245997e-01
-2.14915782e-01 1.10963845e+00 3.50936532e-01 2.85161942e-01
1.19433928e+00 -1.33997846e+00 5.64628467e-02 2.42744222e-01
-8.29314440e-02 4.52873945e-01 3.42072934e-01 -6.47945404e-01
-6.64798081e-01 -1.53861284e+00 1.26121652e+00 -5.54469526e-01
5.29697061e-01 -6.01537228e-02 -8.43815923e-01 3.20507735e-01
-1.84569821e-01 3.43128502e-01 1.28162169e+00 1.34554967e-01
-5.34048736e-01 -9.09969732e-02 -1.47473884e+00 2.63536692e-01
1.24204290e+00 -4.37262535e-01 -8.63643050e-01 4.62638825e-01
9.58199263e-01 -3.97448242e-01 -1.22414076e+00 4.29689169e-01
6.38900459e-01 -4.05281991e-01 9.74628270e-01 -7.05599070e-01
8.53537023e-01 2.48479750e-02 -3.29813570e-01 -1.30464423e+00
-7.88128376e-01 -3.22876632e-01 -2.34203845e-01 1.16183996e+00
3.58030826e-01 -2.94666827e-01 3.04696500e-01 2.78604418e-01
-1.78390712e-01 -8.52774382e-01 -9.82172728e-01 -9.89238799e-01
-1.59996899e-03 -2.58470535e-01 4.94140416e-01 1.57972717e+00
4.52553421e-01 2.62978286e-01 -6.96924925e-01 -1.43205956e-01
3.79063278e-01 9.18932781e-02 6.87151477e-02 -1.44435120e+00
-2.65977859e-01 -3.17606002e-01 -3.41278195e-01 -3.77278566e-01
6.73720464e-02 -1.28591311e+00 -5.52784093e-03 -1.54650319e+00
7.57537425e-01 -4.42942619e-01 -9.08491135e-01 8.26135159e-01
-5.10607064e-01 2.87542224e-01 5.72211184e-02 3.38091373e-01
-1.10698664e+00 1.41429245e-01 9.10425246e-01 -1.55228227e-01
-3.95943373e-02 -1.62394226e-01 -8.92107844e-01 6.02123857e-01
6.71208978e-01 -1.12788975e+00 -3.83467406e-01 -2.69968092e-01
2.95657843e-01 8.60592648e-02 -1.93456382e-01 -1.20152378e+00
1.07950993e-01 -1.01400949e-01 -3.35097536e-02 -3.17369461e-01
-1.52099267e-01 -8.10365796e-01 6.81603188e-03 9.85896885e-01
-8.90247941e-01 4.88034040e-02 1.70368239e-01 6.63599551e-01
-1.10538252e-01 -4.57418263e-01 7.42101967e-01 -2.25589752e-01
-7.42895067e-01 4.11475897e-01 -3.21340472e-01 8.37164044e-01
1.01108885e+00 1.75684586e-01 -2.19650920e-02 3.66076035e-03
-8.54455411e-01 1.49987563e-01 5.84611520e-02 5.08195639e-01
5.75884461e-01 -1.74116445e+00 -9.10384774e-01 -8.54403228e-02
9.00165081e-01 -3.33395958e-01 5.40463507e-01 7.63332009e-01
-5.00792384e-01 5.43784916e-01 -2.41700962e-01 -4.07241762e-01
-1.18735337e+00 7.17741430e-01 -1.76602677e-01 -6.91210926e-01
-6.42366469e-01 9.29418802e-01 1.40747696e-01 -5.01545608e-01
4.25040036e-01 -6.95509493e-01 -3.59440237e-01 3.21524918e-01
7.93290436e-01 4.50105011e-01 -9.18476954e-02 -4.12369370e-01
-7.24696338e-01 7.51796007e-01 -3.58107209e-01 2.48291120e-01
1.28834224e+00 7.22958939e-03 8.62421766e-02 2.87848592e-01
1.35191834e+00 -2.45389208e-01 -6.49973392e-01 -4.45603192e-01
1.26325667e-01 -2.60736048e-01 -2.06939250e-01 -1.18893635e+00
-8.11838090e-01 6.38646483e-01 7.23199844e-01 -1.07950039e-01
1.10855043e+00 4.90920879e-02 1.14503396e+00 3.66893828e-01
2.30390713e-01 -1.05316377e+00 1.77410170e-01 6.47029698e-01
6.87195480e-01 -1.42841005e+00 -1.60344005e-01 -1.40458569e-01
-9.81986582e-01 7.05754757e-01 4.28096414e-01 7.27808252e-02
8.46476138e-01 1.15689203e-01 -1.19303316e-01 -2.08245769e-01
-8.60823691e-01 -2.17572704e-01 3.94927859e-01 4.95014787e-01
4.85470712e-01 3.24714750e-01 -2.91772604e-01 9.41370189e-01
4.76414859e-02 4.97056246e-01 4.16342705e-01 7.19044507e-01
-3.49308968e-01 -8.15221667e-01 -9.62441638e-02 9.52621400e-01
-1.05643427e+00 -5.52749157e-01 2.99736142e-01 2.74948120e-01
6.45237803e-01 7.35923231e-01 7.33515695e-02 -3.53936940e-01
2.88661957e-01 1.04613841e-01 1.44000594e-02 -1.20254576e+00
-6.78796768e-01 -7.72363245e-02 -1.02960274e-01 -2.52355218e-01
-4.38894093e-01 -5.33463895e-01 -1.49653816e+00 -1.51396900e-01
-9.21680480e-02 7.16461912e-02 -1.95826218e-01 7.80078530e-01
6.42968416e-01 8.74135494e-01 3.28058481e-01 1.35145308e-02
-5.74386775e-01 -9.62895215e-01 -7.38298476e-01 9.12691951e-01
4.17963296e-01 -8.81788433e-01 -1.93105996e-01 3.57630193e-01]
|
[8.014490127563477, 6.8608832359313965]
|
a0115296-1800-4476-9676-39bcdd81292c
|
prototype-based-counterfactual-explanation
|
2105.00703
| null |
https://arxiv.org/abs/2105.00703v3
|
https://arxiv.org/pdf/2105.00703v3.pdf
|
Causality-based Counterfactual Explanation for Classification Models
|
Counterfactual explanation is one branch of interpretable machine learning that produces a perturbation sample to change the model's original decision. The generated samples can act as a recommendation for end-users to achieve their desired outputs. Most of the current counterfactual explanation approaches are the gradient-based method, which can only optimize the differentiable loss functions with continuous variables. Accordingly, the gradient-free methods are proposed to handle the categorical variables, which however have several major limitations: 1) causal relationships among features are typically ignored when generating the counterfactuals, possibly resulting in impractical guidelines for decision-makers; 2) the counterfactual explanation algorithm requires a great deal of effort into parameter tuning for dertermining the optimal weight for each loss functions which must be conducted repeatedly for different datasets and settings. In this work, to address the above limitations, we propose a prototype-based counterfactual explanation framework (ProCE). ProCE is capable of preserving the causal relationship underlying the features of the counterfactual data. In addition, we design a novel gradient-free optimization based on the multi-objective genetic algorithm that generates the counterfactual explanations for the mixed-type of continuous and categorical features. Numerical experiments demonstrate that our method compares favorably with state-of-the-art methods and therefore is applicable to existing prediction models. All the source codes and data are available at \url{https://github.com/tridungduong16/multiobj-scm-cf}.
|
['Guandong Xu', 'Qian Li', 'Tri Dung Duong']
|
2021-05-03
| null | null | null | null |
['counterfactual-explanation']
|
['miscellaneous']
|
[ 1.86155677e-01 4.65931475e-01 -6.68896079e-01 -4.67317611e-01
-5.34662008e-01 -9.37925428e-02 6.00201964e-01 -1.37541756e-01
3.06956992e-02 1.34113896e+00 3.39226663e-01 -7.91125655e-01
-5.14926076e-01 -7.72151828e-01 -8.18723857e-01 -6.72096908e-01
-1.50422603e-01 2.34843865e-01 -5.57663858e-01 -4.53508943e-02
5.88466167e-01 6.19100481e-02 -1.75560057e+00 1.81028917e-01
1.51312470e+00 7.61008501e-01 2.24373229e-02 1.99651614e-01
-5.45809604e-02 3.72594774e-01 -3.18605483e-01 -7.36379325e-01
3.91359836e-01 -6.48016214e-01 -5.90139389e-01 -1.95026278e-01
9.98158976e-02 -2.80667633e-01 -1.27113074e-01 1.06734121e+00
4.36172873e-01 2.58087099e-01 7.76226521e-01 -1.83923697e+00
-9.93408501e-01 8.70824039e-01 -6.05135262e-01 -3.00702732e-02
1.93962350e-01 2.94261843e-01 1.16770899e+00 -8.02466929e-01
9.44934040e-02 1.45083499e+00 4.81691241e-01 7.85628915e-01
-1.17734814e+00 -9.41048861e-01 5.09662449e-01 4.62734759e-01
-7.88785398e-01 -1.41820401e-01 1.06504476e+00 -2.68307716e-01
5.16058505e-01 7.46335626e-01 6.66939497e-01 1.10280621e+00
2.95853049e-01 9.45497572e-01 1.11581171e+00 -4.10702109e-01
4.56423521e-01 1.85973153e-01 -1.02596983e-01 6.48958504e-01
7.08887100e-01 5.69989383e-01 -3.10991198e-01 -4.33971554e-01
4.74880725e-01 3.24307621e-01 -6.33021593e-01 -4.49160725e-01
-1.09566510e+00 1.30273628e+00 4.64513272e-01 -4.59669344e-02
-5.42244732e-01 1.98765129e-01 2.56099433e-01 2.76623547e-01
6.06175780e-01 7.10363448e-01 -7.37672627e-01 1.21234745e-01
-4.82980222e-01 6.35404289e-01 6.16213441e-01 5.88017821e-01
5.26928961e-01 1.63735256e-01 -3.15134734e-01 6.97944641e-01
5.29397547e-01 3.70028704e-01 8.31047237e-01 -9.32050645e-01
6.92318797e-01 6.60401583e-01 4.40560132e-01 -9.61227596e-01
-1.59464672e-01 -4.97017711e-01 -8.57769251e-01 3.48346174e-01
2.89397955e-01 -4.30856913e-01 -7.17506349e-01 1.78370333e+00
4.26573873e-01 3.06166828e-01 -2.04087421e-02 1.09309208e+00
3.63217086e-01 4.43576753e-01 1.62398964e-01 -5.86299598e-01
8.57959688e-01 -8.79860103e-01 -7.44271278e-01 -2.11516097e-01
4.29969400e-01 -4.66712266e-01 1.40460587e+00 1.61705002e-01
-8.02305341e-01 -3.31535786e-01 -1.05557394e+00 4.93340820e-01
-3.07152063e-01 -1.49196029e-01 1.14847231e+00 7.10398018e-01
-5.27067125e-01 8.35344195e-01 -3.79246294e-01 2.38489270e-01
5.41077793e-01 4.50587273e-01 1.04928426e-01 9.18654650e-02
-1.48507309e+00 5.92915118e-01 5.55739343e-01 1.34117827e-01
-6.25808418e-01 -1.01920176e+00 -8.49494040e-01 2.70767510e-01
5.09841859e-01 -1.07471895e+00 1.25448453e+00 -1.21031809e+00
-1.45699453e+00 2.60294259e-01 -1.57521814e-01 -5.66400945e-01
8.75386298e-01 -1.27162218e-01 -3.72996807e-01 -4.65743870e-01
2.21181527e-01 5.28662622e-01 6.97453082e-01 -1.38808084e+00
-7.83820271e-01 -3.84391755e-01 4.51112650e-02 3.80664974e-01
-4.92512658e-02 -4.38533127e-01 1.48538113e-01 -9.26594794e-01
-9.99937952e-02 -7.67375112e-01 -4.74013180e-01 -1.55589566e-01
-7.46531427e-01 -6.13555536e-02 7.79010713e-01 -3.42163831e-01
1.34266663e+00 -1.73414195e+00 -2.70690024e-01 2.09476240e-02
-3.63552719e-02 1.53772652e-01 4.07290161e-02 1.53916329e-01
-4.26699251e-01 6.00873709e-01 -5.06581068e-01 -1.38573602e-01
2.18094617e-01 -1.35560066e-01 -6.15598023e-01 3.24739695e-01
-2.20937822e-02 8.07499170e-01 -1.01259470e+00 -2.40951687e-01
3.26817602e-01 -2.44846307e-02 -6.78114951e-01 3.46456200e-01
-3.14061552e-01 2.45675668e-01 -7.03624368e-01 4.39666241e-01
6.97889268e-01 -9.01170075e-02 2.02530146e-01 1.95175558e-01
-8.88436660e-02 3.44734550e-01 -1.30749393e+00 1.03514004e+00
-3.93328696e-01 1.93552896e-01 -4.73395556e-01 -1.25541127e+00
7.68548548e-01 3.82462233e-01 2.86491483e-01 -3.03240269e-01
1.53730094e-01 4.13561076e-01 1.33122146e-01 -4.56334144e-01
1.74611062e-01 -5.36902606e-01 -5.20729087e-02 5.67033172e-01
-5.41050792e-01 5.76163903e-02 -4.00346331e-02 -2.81107754e-01
5.40840983e-01 4.92157927e-03 7.49717593e-01 -2.48616546e-01
4.95070606e-01 -6.01659454e-02 9.79850650e-01 9.45815027e-01
-1.39191344e-01 4.58216220e-01 4.91659284e-01 -5.83309412e-01
-7.84359396e-01 -9.88859475e-01 -1.31297350e-01 5.10942698e-01
1.17902584e-01 3.54905814e-01 -4.90117520e-01 -1.13455224e+00
4.83083904e-01 1.56853378e+00 -7.56745338e-01 -4.55191374e-01
-1.60184383e-01 -1.17350769e+00 4.58138250e-02 3.03898782e-01
7.30088592e-01 -1.09211695e+00 -4.85702962e-01 1.44291505e-01
-2.54554361e-01 -1.00374937e-01 -5.01765490e-01 -2.62468934e-01
-1.07895887e+00 -1.24335241e+00 -4.05460924e-01 -8.59252065e-02
7.21294343e-01 2.55354255e-01 7.93967783e-01 1.05734661e-01
5.21085598e-02 -1.63900424e-02 -1.05422989e-01 -8.72357547e-01
-1.98787943e-01 -3.35969269e-01 1.81913093e-01 2.18922898e-01
3.88008475e-01 -5.56638420e-01 -8.75911951e-01 1.70545101e-01
-7.04724371e-01 3.53877217e-01 5.80869377e-01 1.31457245e+00
4.70271528e-01 1.13137729e-01 1.00778878e+00 -1.15776145e+00
8.75132680e-01 -9.31510210e-01 -5.93774855e-01 2.74905533e-01
-1.24390721e+00 2.25451931e-01 1.03916538e+00 -5.06439388e-01
-1.49253905e+00 -2.74042070e-01 2.96518207e-01 -3.68920952e-01
-9.40083712e-02 4.73681003e-01 -4.62253571e-01 4.89786416e-01
4.35176551e-01 1.36247873e-01 -7.89807588e-02 -4.86074656e-01
4.64886904e-01 6.86294973e-01 1.91580325e-01 -4.43591118e-01
6.96900189e-01 4.01818603e-01 -6.04310334e-02 -1.70709506e-01
-8.91657352e-01 4.00350727e-02 -1.37155429e-01 -3.50691341e-02
3.68774623e-01 -4.49989706e-01 -8.11127424e-01 1.81398585e-01
-9.29445326e-01 -1.26114666e-01 -3.27494025e-01 7.52492547e-01
-7.81189680e-01 1.32415205e-01 3.11695021e-02 -1.07637250e+00
-3.86163235e-01 -9.09562051e-01 5.43089569e-01 3.89584452e-01
-3.16363931e-01 -1.09697342e+00 -4.76372875e-02 4.40511733e-01
8.57872516e-02 4.98974830e-01 1.12689698e+00 -6.69005215e-01
-5.29797614e-01 -2.55220048e-02 -1.94551162e-02 7.86682367e-02
2.67787784e-01 -1.11279503e-01 -7.89617181e-01 -2.28844821e-01
9.05554518e-02 -7.38019943e-02 7.44053304e-01 7.78657317e-01
1.58557284e+00 -1.12662745e+00 -4.41920638e-01 5.71585119e-01
1.39224732e+00 5.31756461e-01 4.56666917e-01 5.18511593e-01
3.50030869e-01 6.39300585e-01 1.08468997e+00 7.13845491e-01
3.25086981e-01 4.76379633e-01 6.46065354e-01 -1.98544174e-01
4.27630424e-01 -6.56678736e-01 1.53829098e-01 -2.67103687e-02
-1.26878083e-01 -2.77613610e-01 -5.23878634e-01 6.77079380e-01
-2.26297951e+00 -1.29852450e+00 -2.14745656e-01 2.43325257e+00
7.89019406e-01 1.96740720e-02 -1.52974710e-01 1.11964189e-01
8.04066062e-01 1.59921095e-01 -1.04268610e+00 -7.66911089e-01
1.99237436e-01 -3.02004665e-01 4.55366164e-01 4.50902462e-01
-9.68444526e-01 4.45424378e-01 5.69875908e+00 6.69928789e-01
-1.06925714e+00 -7.23391101e-02 1.00336921e+00 -3.18118930e-01
-9.21951294e-01 2.62763143e-01 -5.17189741e-01 8.97729933e-01
7.76853859e-01 -8.97091866e-01 4.54745591e-01 9.34585452e-01
9.25819695e-01 9.75363851e-02 -1.05462492e+00 5.95443130e-01
-3.69413555e-01 -1.37175298e+00 3.60096782e-01 4.78980429e-02
8.92497957e-01 -4.51494157e-01 2.40344107e-01 3.51199269e-01
5.34256279e-01 -1.00918460e+00 7.79017210e-01 5.21969974e-01
6.76466703e-01 -8.17810178e-01 7.83735573e-01 4.74925250e-01
-5.81573844e-01 -4.73944336e-01 -4.20863748e-01 -3.58384460e-01
-2.69606207e-02 8.06180120e-01 -9.80136216e-01 6.72849357e-01
4.90389585e-01 4.01439965e-01 -7.94594213e-02 1.01573658e+00
-5.87398767e-01 8.19933236e-01 8.44771788e-02 -2.57648289e-01
8.46621394e-02 -2.11753964e-01 7.19227970e-01 7.00738132e-01
5.65230906e-01 1.10463932e-01 -1.37510926e-01 1.22006786e+00
-1.02430031e-01 1.20976649e-01 -7.13724732e-01 2.33341470e-01
7.73930311e-01 7.38872707e-01 -2.13906392e-01 -2.69321740e-01
-2.50272095e-01 6.62837923e-01 2.03718737e-01 4.83959407e-01
-1.06452620e+00 -3.16302896e-01 8.56750011e-01 -4.22690362e-02
8.75273943e-02 5.67918003e-01 -8.13296616e-01 -1.29644859e+00
8.82342309e-02 -1.04941583e+00 7.43451357e-01 -5.96077383e-01
-1.38930655e+00 6.50230199e-02 2.00585261e-01 -1.28377342e+00
-5.04456282e-01 -2.83871591e-01 -1.10664749e+00 8.72646213e-01
-1.45695460e+00 -7.71299720e-01 4.32698093e-02 2.52414286e-01
6.04743659e-01 -1.40479714e-01 6.76554322e-01 -1.45019010e-01
-7.32488036e-01 6.21171117e-01 3.92642945e-01 -4.05972689e-01
3.75648499e-01 -1.42098093e+00 2.11234301e-01 8.19872618e-01
-1.38546512e-01 7.86122501e-01 1.08011043e+00 -6.91026807e-01
-9.45367396e-01 -1.17714882e+00 9.80679095e-01 -1.93231867e-03
3.31051677e-01 -4.72128112e-03 -7.46827245e-01 6.66450381e-01
-5.71449567e-03 -3.95568073e-01 7.98783123e-01 2.73800850e-01
2.06914153e-02 -2.01585740e-01 -1.42268610e+00 1.07782316e+00
1.02270305e+00 1.75111994e-01 -9.04221356e-01 1.98160768e-01
7.17033207e-01 -6.92471117e-02 -4.03990895e-01 3.97728711e-01
5.69737792e-01 -1.01936042e+00 8.40464592e-01 -9.53819275e-01
7.49443352e-01 -2.59286076e-01 6.05652183e-02 -1.72816324e+00
-3.19945157e-01 -7.43935108e-01 -1.54404268e-01 1.23755491e+00
6.93781197e-01 -1.22549534e+00 6.99639320e-01 1.02955830e+00
9.36930776e-02 -1.18485129e+00 -9.29462194e-01 -7.51132607e-01
5.85375726e-03 -3.42305213e-01 1.32387102e+00 1.16662431e+00
1.61679134e-01 1.54264241e-01 -4.96996254e-01 1.02561422e-01
8.93056989e-01 6.23197794e-01 7.58243978e-01 -9.08079684e-01
-4.43875700e-01 -5.27381480e-01 9.73186940e-02 -5.77892423e-01
2.59187877e-01 -7.44227350e-01 -2.13003084e-01 -1.27585137e+00
3.23894054e-01 -4.61624354e-01 -3.99385393e-01 6.49411261e-01
-7.70301700e-01 -4.44199890e-01 1.24720089e-01 1.96969911e-01
1.56480111e-02 1.07601535e+00 1.22431135e+00 -3.66962962e-02
-2.83681363e-01 4.58613455e-01 -1.28687263e+00 8.54566514e-01
1.07401180e+00 -6.66524470e-01 -7.30179250e-01 1.80900991e-02
-2.65489876e-01 1.73057482e-01 5.27117431e-01 -3.31221133e-01
-3.75051558e-01 -9.10895526e-01 3.05760235e-01 -3.22281301e-01
-1.09879807e-01 -6.54844165e-01 4.17687595e-01 7.71677494e-01
-5.51348031e-01 -6.94633573e-02 -1.00029176e-02 7.98433959e-01
-1.54260173e-01 -3.61588240e-01 6.14983141e-01 -1.46609783e-01
-4.42350894e-01 3.08677375e-01 -1.33838028e-01 5.89370616e-02
1.08988214e+00 -1.21547684e-01 -3.26270401e-01 -5.50050020e-01
-1.51824012e-01 5.24221659e-01 2.80538768e-01 4.89956856e-01
6.27337575e-01 -1.51945758e+00 -8.20508480e-01 -1.12435676e-01
-1.04390070e-01 -2.43316203e-01 3.16919893e-01 5.17486572e-01
-1.82214677e-02 7.14107752e-01 -7.71250054e-02 1.22815184e-01
-9.85970616e-01 6.69557631e-01 4.60260451e-01 -4.96898741e-01
-2.02057749e-01 4.50317889e-01 3.68889093e-01 -6.45310938e-01
-1.63884029e-01 -1.31935686e-01 -2.00405523e-01 -2.23860160e-01
3.81435156e-01 6.49637461e-01 -3.90353650e-01 2.51065902e-02
-1.81298301e-01 -1.36047408e-01 1.23264175e-02 1.88563317e-02
1.44607961e+00 -1.27372295e-01 1.18993446e-01 3.23791742e-01
1.13980722e+00 -1.68056563e-01 -1.39831424e+00 1.25670671e-01
1.16764242e-02 -1.00513530e+00 -3.99294160e-02 -9.78356123e-01
-8.63631606e-01 5.33972442e-01 4.11774874e-01 2.09179029e-01
1.22271574e+00 -4.22180176e-01 4.55338776e-01 1.21360779e-01
2.79470623e-01 -9.62639570e-01 -3.98510337e-01 -1.01786636e-01
1.27721143e+00 -1.42698252e+00 1.48889020e-01 -3.26018602e-01
-7.39500463e-01 7.13811815e-01 7.55810201e-01 9.14695412e-02
4.54168826e-01 -3.23641777e-01 3.76028717e-02 -1.90631673e-02
-9.96408463e-01 2.29805440e-01 4.35203344e-01 4.06808376e-01
4.64990348e-01 5.11644781e-01 -1.05849957e+00 9.17495131e-01
-4.09251541e-01 5.88135235e-02 4.21966046e-01 4.75510031e-01
-1.13378473e-01 -1.09103370e+00 -4.82777148e-01 8.36860955e-01
-5.93150079e-01 -1.18535511e-01 -3.23324144e-01 9.87515152e-01
1.40532061e-01 1.02367091e+00 -2.15530396e-01 -1.51575074e-01
4.08599705e-01 6.58280849e-02 -1.06152475e-01 -4.34970766e-01
-2.80419499e-01 -4.68263596e-01 1.01443015e-01 -5.86742043e-01
-1.78173795e-01 -9.10244346e-01 -1.19173181e+00 -4.70115721e-01
-4.82729465e-01 4.66418684e-01 4.54774201e-01 9.49897647e-01
5.12659967e-01 3.22018772e-01 1.09942806e+00 -5.64317703e-01
-1.09772348e+00 -8.55736852e-01 -5.21614313e-01 5.75579226e-01
2.28813499e-01 -9.50306594e-01 -8.39864135e-01 -2.28899524e-01]
|
[8.672701835632324, 5.620512962341309]
|
8b63dedd-fcf0-432d-9071-6e2f69322fd1
|
pixel-level-matching-for-video-object
|
1708.05137
| null |
http://arxiv.org/abs/1708.05137v1
|
http://arxiv.org/pdf/1708.05137v1.pdf
|
Pixel-Level Matching for Video Object Segmentation using Convolutional Neural Networks
|
We propose a novel video object segmentation algorithm based on pixel-level
matching using Convolutional Neural Networks (CNN). Our network aims to
distinguish the target area from the background on the basis of the pixel-level
similarity between two object units. The proposed network represents a target
object using features from different depth layers in order to take advantage of
both the spatial details and the category-level semantic information.
Furthermore, we propose a feature compression technique that drastically
reduces the memory requirements while maintaining the capability of feature
representation. Two-stage training (pre-training and fine-tuning) allows our
network to handle any target object regardless of its category (even if the
object's type does not belong to the pre-training data) or of variations in its
appearance through a video sequence. Experiments on large datasets demonstrate
the effectiveness of our model - against related methods - in terms of
accuracy, speed, and stability. Finally, we introduce the transferability of
our network to different domains, such as the infrared data domain.
|
['Seokju Lee', 'In So Kweon', 'Seunghak Shin', 'Jae Shin Yoon', 'Francois Rameau', 'Junsik Kim']
|
2017-08-17
|
pixel-level-matching-for-video-object-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Yoon_Pixel-Level_Matching_for_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Yoon_Pixel-Level_Matching_for_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['feature-compression']
|
['computer-vision']
|
[ 4.58591193e-01 -3.19418579e-01 -2.74793088e-01 -4.11763042e-01
-2.47847512e-01 -3.43925923e-01 8.33279267e-02 2.94977650e-02
-4.60376412e-01 2.83718258e-01 -5.41273952e-01 -7.10626245e-02
-7.27241561e-02 -1.11284292e+00 -7.47636497e-01 -6.43713593e-01
1.49946398e-04 1.79247558e-01 7.34340727e-01 1.01415917e-01
2.54358858e-01 9.30513203e-01 -1.77365899e+00 4.30568814e-01
7.94934213e-01 1.50776350e+00 5.00113070e-01 3.73226881e-01
-2.73858547e-01 5.66285908e-01 -4.74232435e-01 -6.88484404e-04
6.14627421e-01 -2.82950312e-01 -7.57042468e-01 5.12320876e-01
9.42997098e-01 -4.37268108e-01 -4.77073073e-01 1.25044012e+00
-7.00901598e-02 6.78606108e-02 6.09108329e-01 -9.47089195e-01
-2.36460239e-01 -6.46193400e-02 -4.54064995e-01 2.57956654e-01
-1.79232761e-01 1.17611093e-02 7.77499676e-01 -5.57706237e-01
5.78709722e-01 9.85189438e-01 6.02045655e-01 4.35439914e-01
-9.45533454e-01 -5.13480246e-01 2.89206654e-01 3.74121577e-01
-1.22842896e+00 -1.00921668e-01 8.74362290e-01 -5.20209670e-01
5.25041759e-01 6.30919188e-02 7.04074681e-01 4.34552044e-01
2.45814268e-02 6.69517577e-01 7.25301743e-01 -2.99568862e-01
3.70717257e-01 -1.74286459e-02 2.88426280e-01 8.94249320e-01
2.49309048e-01 -9.58761573e-02 -2.53416240e-01 1.22524396e-01
1.14717662e+00 1.77982181e-01 -2.46197745e-01 -6.88259542e-01
-9.05327678e-01 6.27020895e-01 7.90008724e-01 5.47180474e-01
-3.98101687e-01 1.74437717e-01 2.95757324e-01 9.50662121e-02
2.13211060e-01 1.19703516e-01 -5.39173186e-01 3.40650529e-01
-1.16577148e+00 -5.77628650e-02 5.86446702e-01 9.28688705e-01
1.17697513e+00 -7.62868822e-02 -2.43988540e-02 6.55777276e-01
1.13676861e-01 9.70800593e-02 6.17426872e-01 -1.00792623e+00
3.24277192e-01 9.03018475e-01 -1.28053963e-01 -1.03903484e+00
-2.48692736e-01 -3.87822002e-01 -7.13279188e-01 5.12026131e-01
5.75063169e-01 5.52213490e-02 -1.08542573e+00 1.58770978e+00
4.15986240e-01 1.79739133e-01 -1.00001723e-01 9.94168580e-01
9.39535618e-01 5.83315849e-01 -9.03873444e-02 -1.85792912e-02
1.31598961e+00 -9.83822286e-01 -2.12260023e-01 -3.69618267e-01
3.39007616e-01 -3.08011889e-01 5.72561860e-01 1.76381916e-01
-1.09680295e+00 -9.58003402e-01 -1.01633036e+00 9.35680643e-02
-4.77073133e-01 4.10553932e-01 6.15010858e-01 6.17030025e-01
-9.61109817e-01 6.94464862e-01 -7.63399184e-01 -4.60186422e-01
5.85589051e-01 6.12957656e-01 -3.66028100e-01 -8.08324516e-02
-8.18018258e-01 3.34027559e-01 8.07833254e-01 9.52894390e-02
-6.37872696e-01 -5.40835977e-01 -7.65765369e-01 2.70274282e-01
3.21285516e-01 -6.09336853e-01 9.23035145e-01 -1.82275176e+00
-1.21640074e+00 7.22521484e-01 -1.58168733e-01 -2.66405463e-01
5.25564551e-01 1.01494268e-01 -1.05758406e-01 6.54900372e-01
-2.15730127e-02 8.91393900e-01 9.67169762e-01 -1.03446233e+00
-1.14846051e+00 -4.74793136e-01 2.61416584e-01 1.43358380e-01
-3.42757732e-01 -1.55522808e-01 -8.68242860e-01 -5.04401326e-01
5.74397564e-01 -7.62331128e-01 -2.95710623e-01 6.26913846e-01
-1.29969358e-01 -4.95946556e-02 1.21261561e+00 -3.85408372e-01
7.06688285e-01 -2.37961984e+00 -1.50590315e-01 4.28980619e-01
2.14649960e-01 4.70068753e-01 -8.40783417e-02 -1.63893864e-01
-1.34095564e-01 -3.43341604e-02 -3.20983291e-01 1.97786987e-01
-5.28355300e-01 1.93522260e-01 1.55053034e-01 4.97792572e-01
3.41505975e-01 7.80205071e-01 -6.85882390e-01 -6.91491365e-01
4.21393901e-01 3.19692999e-01 -4.00084227e-01 1.26273364e-01
-2.99148738e-01 1.67079702e-01 -5.54272175e-01 6.21297479e-01
7.73171306e-01 -1.78708330e-01 1.22628115e-01 -4.17661607e-01
-1.40741110e-01 5.84557690e-02 -1.28676343e+00 1.50097239e+00
-2.00380176e-01 7.80304492e-01 1.05226159e-01 -1.18766248e+00
8.07636201e-01 -9.98763591e-02 6.78236425e-01 -7.17819810e-01
3.03473055e-01 2.88174987e-01 9.77577344e-02 -4.51456189e-01
4.45700079e-01 3.59749317e-01 2.83075511e-01 1.61534682e-01
5.39922714e-02 3.06845605e-01 4.19613421e-01 -1.13205217e-01
7.26592124e-01 1.02889590e-01 1.38839319e-01 -2.29001373e-01
4.70698148e-01 1.04237095e-01 6.01246119e-01 8.27497482e-01
-2.66235679e-01 7.35644042e-01 3.77727628e-01 -5.85688770e-01
-8.21214437e-01 -7.72582531e-01 -2.68406361e-01 9.73750412e-01
6.03364825e-01 1.23666659e-01 -9.09181774e-01 -7.49683321e-01
1.16895579e-01 1.33478671e-01 -7.72016227e-01 -1.48346247e-02
-5.89802325e-01 -4.50531572e-01 4.63498943e-02 7.90416896e-01
8.77135634e-01 -1.00405729e+00 -1.14058888e+00 1.65186927e-01
2.67223455e-02 -1.16323662e+00 -3.32620144e-01 2.49615818e-01
-1.12255752e+00 -1.14460015e+00 -4.87586290e-01 -1.15468466e+00
8.79806221e-01 5.39612889e-01 7.84997940e-01 1.90424129e-01
-4.97136027e-01 3.71454209e-01 -1.28840417e-01 1.90366022e-02
-1.81663319e-01 -4.33374345e-02 -2.71525621e-01 2.83533335e-01
4.24405456e-01 -3.00319403e-01 -7.35323310e-01 4.21917766e-01
-1.06820953e+00 -2.87531987e-02 6.41652048e-01 4.46295798e-01
6.51501119e-01 3.35719079e-01 6.88440129e-02 -6.43624842e-01
-3.35710533e-02 -1.94807559e-01 -7.44082332e-01 2.65539676e-01
-1.85546473e-01 -1.69174045e-01 2.39236400e-01 -5.26450157e-01
-7.42850423e-01 4.03528422e-01 2.60128736e-01 -5.47065794e-01
-4.19368476e-01 1.74370185e-01 -1.63408428e-01 -3.96690935e-01
2.86981910e-01 3.60884190e-01 1.11978918e-01 -3.92592341e-01
2.27538660e-01 5.62418640e-01 7.43585944e-01 -3.99359822e-01
6.41258776e-01 7.06156969e-01 -2.12772749e-02 -7.15802789e-01
-6.30155563e-01 -6.30992591e-01 -1.10128903e+00 -2.96108365e-01
1.00857615e+00 -8.68116915e-01 -5.08132875e-01 5.37774444e-01
-1.23492169e+00 -1.12461984e-01 -2.75012165e-01 4.43820775e-01
-3.73098254e-01 3.13393265e-01 -5.88305652e-01 -4.83185410e-01
-1.29222438e-01 -1.23291767e+00 8.14280868e-01 4.50251132e-01
3.37840468e-01 -9.25344467e-01 -5.54013610e-01 1.19708598e-01
2.24221379e-01 2.02329472e-01 8.94606113e-01 -6.46514416e-01
-9.34065938e-01 -2.93750614e-01 -5.80762506e-01 4.97466177e-01
2.34015703e-01 1.63852811e-01 -8.63976717e-01 -1.72946379e-01
1.72060784e-02 -1.21585868e-01 1.12042975e+00 6.24925196e-01
1.50177002e+00 -1.51734054e-01 -4.41982985e-01 6.91393077e-01
1.64150202e+00 2.13266075e-01 5.80202401e-01 5.38145661e-01
8.30244660e-01 6.61446571e-01 5.41663349e-01 1.81494534e-01
-3.39660458e-02 7.06718981e-01 6.55786097e-01 -3.26138198e-01
-2.34286159e-01 1.83473602e-01 1.14395306e-01 1.76315531e-01
-5.20957708e-02 6.54410943e-02 -6.14409208e-01 5.76045990e-01
-1.74227273e+00 -1.00120056e+00 -1.22487418e-01 2.29195976e+00
4.47753459e-01 1.65212721e-01 2.92103708e-01 -1.18630016e-02
9.14645970e-01 -1.08525850e-01 -6.47540808e-01 -3.36817443e-01
-4.27514017e-02 1.18312635e-01 7.65549898e-01 7.44179934e-02
-1.37033761e+00 7.45636880e-01 5.84711552e+00 8.41883719e-01
-1.38568997e+00 -2.29602814e-01 8.57259393e-01 1.21919200e-01
1.86135799e-01 -2.66627938e-01 -8.18925977e-01 4.64497507e-01
3.46886516e-01 1.56881884e-01 1.58767357e-01 1.01296496e+00
-4.06605825e-02 -2.37564936e-01 -1.18179762e+00 8.87753844e-01
3.12664568e-01 -1.39740038e+00 9.32991132e-03 -5.47779202e-02
6.04132295e-01 -7.08116498e-03 3.71966213e-02 4.33510505e-02
-2.24057913e-01 -7.47079968e-01 1.00182605e+00 1.74447447e-01
7.37043023e-01 -6.86798036e-01 6.93679452e-01 2.97112167e-01
-1.40863550e+00 -2.94324428e-01 -4.61414874e-01 9.02827308e-02
-3.59663248e-01 3.11110228e-01 -6.33006155e-01 5.08865476e-01
9.06306982e-01 6.57777250e-01 -6.36000514e-01 1.20431972e+00
1.69875413e-01 3.50536197e-01 -3.66578728e-01 2.04608217e-01
4.87686604e-01 -3.60409528e-01 1.61423802e-01 1.18367624e+00
1.80826396e-01 1.69776067e-01 4.68446940e-01 8.92269552e-01
-2.53805332e-02 6.64010271e-02 -3.08668464e-01 2.60652483e-01
2.20200628e-01 1.22503424e+00 -9.92927074e-01 -4.04533297e-01
-5.66242278e-01 1.04938197e+00 2.91771322e-01 3.71136189e-01
-6.69516146e-01 -5.10198891e-01 5.82938671e-01 2.04477668e-01
9.56158042e-01 -2.16757745e-01 -2.85522133e-01 -7.91070104e-01
1.83731794e-01 -5.62698007e-01 2.94384450e-01 -5.83527029e-01
-1.05462182e+00 7.28958726e-01 -1.40848517e-01 -1.39146101e+00
-1.00899734e-01 -8.80515993e-01 -5.01323223e-01 7.24254191e-01
-1.65593851e+00 -1.01262820e+00 -5.48133254e-01 5.82911134e-01
6.29159212e-01 -5.30227795e-02 3.09787601e-01 2.30233520e-01
-5.42994082e-01 4.05478150e-01 1.86919525e-01 5.10032296e-01
1.48893729e-01 -9.66896713e-01 1.14816926e-01 7.06883073e-01
3.33303958e-02 3.34033787e-01 1.98767722e-01 -4.58815247e-01
-9.50306177e-01 -1.38055384e+00 4.88561511e-01 5.09107858e-02
4.62109357e-01 -2.97199786e-01 -1.06392431e+00 3.47320706e-01
-1.15620956e-01 3.59724075e-01 3.38895053e-01 -2.16825649e-01
-4.17626441e-01 -2.90855587e-01 -1.32712364e+00 2.94438273e-01
8.54292035e-01 -3.42735380e-01 -4.77471918e-01 1.46972314e-01
4.36326742e-01 -3.80557537e-01 -5.47928929e-01 5.49213469e-01
7.06290543e-01 -1.10752463e+00 8.87174904e-01 -5.36084533e-01
3.45243245e-01 -3.89523804e-01 -2.27142468e-01 -6.61781967e-01
-5.77655971e-01 -2.97013037e-02 1.65136352e-01 9.20018196e-01
7.19431117e-02 -3.77848774e-01 9.57552671e-01 5.06077051e-01
6.28661066e-02 -7.65928507e-01 -8.80280733e-01 -8.82634521e-01
-2.55862623e-01 -4.57460284e-01 3.80785316e-01 6.20242655e-01
-5.44730544e-01 -1.46793887e-01 4.98573110e-02 4.75677073e-01
5.53018451e-01 5.68439305e-01 5.48924387e-01 -1.37148547e+00
-2.04806983e-01 -5.53085387e-01 -9.25185382e-01 -1.15773880e+00
6.99011162e-02 -6.22011364e-01 1.44231409e-01 -1.50105727e+00
3.23368490e-01 -5.60991406e-01 -4.48101074e-01 6.35188639e-01
-6.32830635e-02 6.28020048e-01 2.43454635e-01 2.84783423e-01
-6.62645936e-01 1.06880106e-01 1.10047138e+00 -2.63563544e-01
-2.61638403e-01 1.88562050e-01 -2.58153468e-01 7.30999053e-01
7.03240871e-01 -3.79335314e-01 -2.62200236e-01 -4.37289089e-01
-5.25396585e-01 -1.01193562e-01 5.29569805e-01 -1.39400101e+00
1.31915271e-01 -2.11677954e-01 6.58554554e-01 -5.29686213e-01
2.33450264e-01 -1.17026532e+00 -3.52444835e-02 6.29737437e-01
-2.92608649e-01 -2.38599554e-01 2.98404962e-01 6.92797720e-01
-1.64409906e-01 -5.62751532e-01 1.02663779e+00 -1.91933855e-01
-1.21039855e+00 5.21537423e-01 -2.91738033e-01 -2.24963352e-01
1.23819911e+00 -7.92222559e-01 -1.12477340e-01 -1.51125090e-02
-5.20934224e-01 -8.45357105e-02 6.42505288e-01 3.45859230e-01
6.39155626e-01 -1.10453188e+00 -4.03588325e-01 5.14767289e-01
1.74274817e-01 -2.39908248e-02 1.74560130e-01 5.29736102e-01
-7.26705372e-01 3.27104479e-01 -4.54907209e-01 -9.51235175e-01
-1.37312639e+00 5.71935892e-01 6.74509048e-01 2.18947664e-01
-6.15518808e-01 7.16808796e-01 5.92159629e-01 1.19003579e-02
4.41458672e-01 -5.30744672e-01 -1.89539194e-01 -1.67422488e-01
5.59455097e-01 1.06425181e-01 5.21455333e-02 -6.80410564e-01
-4.02970910e-01 8.77718270e-01 -1.09315157e-01 3.01996648e-01
1.07989538e+00 2.65538674e-02 -9.74197388e-02 3.62195760e-01
1.33726490e+00 -4.02298152e-01 -1.48066127e+00 -4.42806393e-01
-1.43028378e-01 -6.91684067e-01 3.14720064e-01 -4.69889730e-01
-1.59117258e+00 7.14583158e-01 9.54465151e-01 7.88056999e-02
1.29011607e+00 8.68833624e-03 6.46834433e-01 3.97224426e-01
3.27921003e-01 -1.11834788e+00 1.54209867e-01 2.98870653e-01
2.86570251e-01 -1.25980985e+00 -5.66328540e-02 -5.68288982e-01
-3.68703544e-01 1.48547065e+00 7.41721570e-01 -2.85279572e-01
5.24899602e-01 4.73891050e-02 1.38887703e-01 -5.11049889e-02
-2.49122173e-01 -4.47303176e-01 4.95797455e-01 6.61157906e-01
1.01894863e-01 -1.36213616e-01 -6.35516420e-02 1.33552060e-01
3.34325045e-01 -4.80386913e-02 2.10825175e-01 9.50606525e-01
-8.72102499e-01 -8.12554538e-01 -2.98761278e-01 4.28695768e-01
-3.19751322e-01 1.11982062e-01 -2.06416532e-01 8.00791800e-01
5.11871815e-01 7.25471258e-01 5.40718198e-01 -1.80890411e-01
1.01300180e-01 -2.51595974e-01 5.21820426e-01 -5.43401122e-01
-4.95575190e-01 1.63200945e-01 -3.37678581e-01 -6.11881733e-01
-6.95128262e-01 -4.76573229e-01 -1.30003405e+00 1.49151077e-02
-3.57273579e-01 -3.31986956e-02 8.36939752e-01 9.24085855e-01
2.18475491e-01 5.02828360e-01 7.23196983e-01 -9.01797891e-01
-1.81708351e-01 -6.18487716e-01 -6.86184406e-01 4.23559219e-01
3.56161445e-01 -6.29683793e-01 -9.54280123e-02 2.68543839e-01]
|
[9.378649711608887, -0.024840237572789192]
|
d2736b45-5109-421b-b0e4-7f18f82040b5
|
how-many-answers-should-i-give-an-empirical
|
2306.00435
| null |
https://arxiv.org/abs/2306.00435v1
|
https://arxiv.org/pdf/2306.00435v1.pdf
|
How Many Answers Should I Give? An Empirical Study of Multi-Answer Reading Comprehension
|
The multi-answer phenomenon, where a question may have multiple answers scattered in the document, can be well handled by humans but is challenging enough for machine reading comprehension (MRC) systems. Despite recent progress in multi-answer MRC, there lacks a systematic analysis of how this phenomenon arises and how to better address it. In this work, we design a taxonomy to categorize commonly-seen multi-answer MRC instances, with which we inspect three multi-answer datasets and analyze where the multi-answer challenge comes from. We further analyze how well different paradigms of current multi-answer MRC models deal with different types of multi-answer instances. We find that some paradigms capture well the key information in the questions while others better model the relationship between questions and contexts. We thus explore strategies to make the best of the strengths of different paradigms. Experiments show that generation models can be a promising platform to incorporate different paradigms. Our annotations and code are released for further research.
|
['Dongyan Zhao', 'Yansong Feng', 'Yuxuan Lai', 'Xiao Liu', 'Jiuheng Lin', 'Chen Zhang']
|
2023-06-01
| null | null | null | null |
['reading-comprehension', 'machine-reading-comprehension']
|
['natural-language-processing', 'natural-language-processing']
|
[ 1.46473870e-01 1.89964846e-01 7.36002810e-03 -3.13335389e-01
-9.89987552e-01 -1.02201056e+00 6.56312943e-01 5.69005370e-01
-1.21301889e-01 5.68435013e-01 5.23707628e-01 -6.93213522e-01
-5.79538465e-01 -7.15052187e-01 -4.64751124e-01 -2.83570588e-02
6.93406880e-01 6.09786332e-01 7.86203265e-01 -7.95479357e-01
7.55553663e-01 -2.06248581e-01 -1.84622097e+00 9.08963203e-01
1.16788757e+00 4.07399297e-01 4.30477262e-01 1.04527831e+00
-7.80360281e-01 1.35758543e+00 -9.37253714e-01 -4.87030983e-01
-2.36872956e-01 -7.08271682e-01 -1.50003994e+00 -4.61898476e-01
9.59254444e-01 2.19128672e-02 3.65982242e-02 6.46536887e-01
5.12048006e-01 3.06990426e-02 6.79916561e-01 -1.07368040e+00
-9.73091543e-01 6.13516331e-01 -1.00405682e-02 5.67831933e-01
1.41681290e+00 -1.62994578e-01 1.21463943e+00 -6.50501847e-01
7.84057438e-01 1.31007957e+00 6.29693568e-01 5.93011141e-01
-8.96931469e-01 4.16706726e-02 -2.27469262e-02 8.61682713e-01
-9.22228158e-01 -1.65946350e-01 4.34967577e-01 -3.04699033e-01
1.10423267e+00 8.83893251e-01 1.65441215e-01 8.47288311e-01
1.43552274e-01 8.30389023e-01 1.49462712e+00 -7.47761965e-01
-1.66824564e-01 1.60560191e-01 9.67725933e-01 5.33221781e-01
-6.66179731e-02 -4.71970856e-01 -5.19023597e-01 -1.44311592e-01
3.30139697e-02 -2.95893848e-01 -6.46902561e-01 1.34396702e-01
-1.02612758e+00 7.59392023e-01 1.51295364e-01 6.96918190e-01
3.85571457e-02 -2.00002775e-01 -8.18328117e-04 8.66309702e-01
4.64261416e-03 1.02148867e+00 -7.94738114e-01 -3.76709312e-01
-6.95770979e-01 8.54298234e-01 1.27798557e+00 1.06361115e+00
7.49157786e-01 -7.92276144e-01 -5.15470266e-01 9.46899891e-01
9.22773629e-02 3.00461352e-01 4.99964088e-01 -1.17095268e+00
5.45003593e-01 1.02563810e+00 2.19374955e-01 -1.11677206e+00
-5.74105799e-01 -2.78875321e-01 -1.24596596e-01 -4.54016149e-01
7.37329483e-01 1.00651398e-01 -4.37355369e-01 1.46108091e+00
1.67695388e-01 -3.21927845e-01 5.02378903e-02 5.59832335e-01
1.56237853e+00 4.90085632e-01 -8.66107866e-02 1.75619870e-02
1.76616204e+00 -1.20958006e+00 -9.29120064e-01 -4.29130793e-01
8.50456178e-01 -1.14101028e+00 1.43793368e+00 6.67101070e-02
-1.12849021e+00 -5.55082977e-01 -6.82387829e-01 -5.20563483e-01
-7.58610189e-01 -9.72941667e-02 1.11097805e-01 6.00603461e-01
-9.82077122e-01 1.73184931e-01 7.06051961e-02 -5.77295184e-01
-3.20951641e-01 -1.17263183e-01 -6.32347092e-02 -5.34091771e-01
-1.39708197e+00 1.50059497e+00 1.39230698e-01 -2.61376530e-01
-2.82634050e-01 -7.26924837e-01 -6.60319507e-01 -2.98412461e-02
8.34381819e-01 -9.65418577e-01 1.74168873e+00 -5.45226097e-01
-9.67152715e-01 9.27785158e-01 -5.85500360e-01 -4.90080453e-02
1.84351783e-02 -2.62616962e-01 -4.32046354e-01 2.85767853e-01
1.68989286e-01 3.37976605e-01 6.08308911e-01 -1.21455419e+00
-6.41940296e-01 -2.06130788e-01 7.65711844e-01 2.08202913e-01
1.54408151e-02 4.77506608e-01 -1.72179312e-01 -3.84907365e-01
2.55714152e-02 -6.67215347e-01 2.29815692e-01 -5.19897521e-01
-1.08396128e-01 -7.02130556e-01 6.95748091e-01 -5.97113311e-01
1.87082827e+00 -1.56126225e+00 4.02319878e-01 -4.87989843e-01
5.07647872e-01 2.56679147e-01 -3.83719385e-01 1.00380361e+00
1.31112665e-01 3.53767633e-01 -2.12450475e-01 -7.95391202e-02
2.61364937e-01 4.06071454e-01 -3.96019936e-01 -2.89929539e-01
1.36582911e-01 1.17900074e+00 -8.95889282e-01 -4.33708042e-01
-2.86491513e-01 -1.01265632e-01 -3.95146757e-01 5.30807734e-01
-7.42235661e-01 1.94532931e-01 -3.89334679e-01 6.83939457e-01
5.11854887e-01 -4.13548470e-01 -1.72629997e-01 -2.64449567e-02
-2.06656065e-02 6.53050721e-01 -1.08322036e+00 1.32463658e+00
-3.85551691e-01 6.28827512e-01 1.10851973e-01 -6.17196143e-01
6.51004851e-01 3.24761510e-01 -3.00158083e-01 -7.10863709e-01
-1.84587926e-01 5.16051948e-01 2.91003168e-01 -1.31087780e+00
8.81322503e-01 -1.08349271e-01 -1.78572699e-01 7.14215457e-01
3.56136486e-02 -4.50016111e-01 6.10943973e-01 4.30961996e-01
1.37786341e+00 -1.50007814e-01 4.85418111e-01 -3.06548566e-01
8.64128768e-01 2.28208810e-01 -6.41959384e-02 1.17033112e+00
-1.51959121e-01 6.16759360e-01 5.23681223e-01 -1.81215182e-01
-4.36803013e-01 -6.81113183e-01 4.45123389e-02 1.42964602e+00
6.24459982e-02 -8.21575105e-01 -6.66182935e-01 -7.83963382e-01
-2.08418190e-01 9.01188016e-01 -5.72336078e-01 9.67033654e-02
-8.77852142e-01 -4.35175091e-01 5.28424203e-01 4.27533627e-01
2.21931338e-01 -1.14309311e+00 -8.89917731e-01 2.26447821e-01
-8.88656437e-01 -8.83486331e-01 -1.46196902e-01 1.15963094e-01
-4.90103036e-01 -1.40076876e+00 -5.29398680e-01 -7.67780840e-01
1.87670872e-01 5.62624633e-01 1.92580605e+00 7.68118739e-01
-8.59851111e-03 1.04538226e+00 -1.00274301e+00 -6.04395688e-01
-5.58106720e-01 5.03760636e-01 -7.09628642e-01 -5.66583991e-01
7.40496814e-01 -1.71083406e-01 -4.22493458e-01 4.55433607e-01
-1.08827055e+00 -1.89418793e-01 2.12216333e-01 4.64692652e-01
9.18641239e-02 -3.26897800e-01 1.00543439e+00 -9.98840809e-01
1.34461391e+00 -9.65723395e-01 1.49570942e-01 9.01572168e-01
-3.24237168e-01 1.93113223e-01 4.45970684e-01 -3.29241484e-01
-9.05916035e-01 -8.31419349e-01 -4.48069930e-01 4.50256497e-01
-3.44495624e-01 7.48193622e-01 5.42039871e-02 -1.44040203e-02
7.69178808e-01 -9.15022288e-03 -2.65411675e-01 -6.14355087e-01
6.66720390e-01 7.40745842e-01 1.80010170e-01 -8.87319684e-01
4.91622299e-01 -8.70075002e-02 -3.48126441e-01 -7.88317680e-01
-1.31678343e+00 -8.92046452e-01 -3.44636261e-01 -3.27296317e-01
8.23435843e-01 -3.60057473e-01 -3.17454576e-01 1.95611343e-01
-1.44947267e+00 -1.45065784e-01 -1.95467100e-01 -1.63414136e-01
-5.39818466e-01 4.14081305e-01 -4.81645554e-01 -5.66931367e-01
-1.22976482e-01 -1.04613483e+00 9.23503578e-01 5.28847933e-01
-8.62209916e-01 -1.26588619e+00 2.76462585e-01 9.15506721e-01
8.27347100e-01 -8.29467773e-02 1.61481190e+00 -7.94998944e-01
-5.79851508e-01 2.15321019e-01 -1.07640140e-01 -7.05561042e-02
-3.55425701e-02 -1.55777112e-01 -8.26412261e-01 1.19443893e-01
4.21655715e-01 -5.51931739e-01 7.97574997e-01 -5.90497255e-02
9.87996519e-01 -2.08297685e-01 -4.69553955e-02 -2.93756008e-01
1.27642465e+00 -1.62907451e-01 5.65412521e-01 4.06686008e-01
4.06532794e-01 1.23290193e+00 5.47806919e-01 -1.42824933e-01
1.13654935e+00 5.54410160e-01 3.75226766e-01 4.42989826e-01
-3.94605368e-01 2.84467102e-03 1.09853521e-01 1.30534613e+00
1.52098387e-01 -4.42997187e-01 -1.07329345e+00 6.24657869e-01
-1.79173934e+00 -1.10825431e+00 -9.20940936e-01 1.74110961e+00
1.02062440e+00 -2.59687245e-01 9.44478139e-02 2.27194712e-01
2.68954128e-01 2.22189635e-01 -2.58176103e-02 -7.05354452e-01
-2.90663272e-01 4.15649742e-01 -4.02096063e-01 7.23172188e-01
-5.28644085e-01 6.38589919e-01 7.05380678e+00 6.46918297e-01
-3.85032147e-01 1.42551646e-01 -1.23091139e-01 3.42402518e-01
-8.38475883e-01 2.28124514e-01 -8.98300171e-01 1.56663746e-01
1.01709378e+00 -9.48517770e-02 2.44398370e-01 3.23575109e-01
-3.41860265e-01 -4.33431417e-01 -1.28974056e+00 5.08433521e-01
6.00482047e-01 -1.21230090e+00 8.32915381e-02 -4.95376498e-01
4.92331803e-01 -3.15767705e-01 -4.03339304e-02 7.63979971e-01
1.13053724e-01 -1.12223017e+00 6.14592910e-01 7.77961791e-01
1.16267860e-01 -1.97336882e-01 6.40872002e-01 8.95009995e-01
-1.06189203e+00 -3.51972491e-01 -2.93441623e-01 -3.76900434e-01
8.52248967e-02 1.07599534e-01 -3.04956496e-01 7.39296496e-01
6.87331975e-01 2.92161405e-01 -1.28796089e+00 1.18061101e+00
-5.23768604e-01 4.52798635e-01 2.54342780e-02 -4.99269217e-01
6.91380054e-02 5.37779294e-02 5.03362298e-01 1.12382579e+00
2.94806004e-01 2.24757269e-01 6.35166690e-02 6.89224422e-01
1.54175520e-01 2.65665174e-01 -4.39612091e-01 1.43586382e-01
4.55827743e-01 1.11947155e+00 -3.34401488e-01 -2.67612040e-01
-7.20320582e-01 7.15831399e-01 5.77112198e-01 2.36632556e-01
-5.31876087e-01 -3.65257353e-01 1.01375520e-01 2.65363723e-01
2.95881890e-02 -8.51679072e-02 -1.12114966e-01 -1.32352853e+00
2.85257369e-01 -1.63709569e+00 6.28943861e-01 -1.21205199e+00
-1.54483032e+00 4.27465856e-01 2.34461486e-01 -7.65026271e-01
-3.60329449e-01 -7.25389123e-01 -6.01606131e-01 7.30319798e-01
-1.73208094e+00 -1.12013602e+00 -4.02044326e-01 3.99650037e-01
8.00677955e-01 3.25279117e-01 1.09066236e+00 2.90307760e-01
-2.46040508e-01 3.26202840e-01 -2.86349803e-01 -3.93845677e-01
9.93291259e-01 -1.72422755e+00 -6.54300442e-04 5.20665228e-01
2.79523045e-01 9.83204663e-01 8.48253369e-01 -3.62404495e-01
-1.38063240e+00 -5.24922490e-01 1.48373663e+00 -1.26694262e+00
8.26363087e-01 4.70728846e-03 -1.42102432e+00 5.31054378e-01
9.40337420e-01 -5.35405517e-01 1.09657741e+00 4.19647068e-01
-6.56943321e-01 3.73284131e-01 -8.55380356e-01 6.16745651e-01
7.06454873e-01 -7.80305684e-01 -1.59893715e+00 3.95321757e-01
8.60125184e-01 -4.40552026e-01 -8.20369124e-01 3.45410764e-01
2.19387218e-01 -1.17561460e+00 6.41278505e-01 -7.45071113e-01
7.51938760e-01 -3.35670322e-01 -3.81066322e-01 -1.49567783e+00
-7.24321827e-02 -4.52968687e-01 -5.36831856e-01 1.32891417e+00
6.43070400e-01 -5.25488615e-01 1.58435225e-01 5.95219016e-01
-2.33158171e-01 -9.20269191e-01 -7.79176474e-01 -4.86713856e-01
7.20282018e-01 -3.31273973e-01 6.55891418e-01 7.40829766e-01
2.34362811e-01 7.41978288e-01 1.58996642e-01 -1.45294473e-01
1.08259425e-01 4.47069883e-01 6.67388439e-01 -1.23447824e+00
-3.64095539e-01 -5.85581660e-01 2.37825930e-01 -1.41910529e+00
1.05119035e-01 -8.57101142e-01 -2.14297771e-01 -1.98269784e+00
2.95868009e-01 -1.64315820e-01 1.61883712e-01 7.23067150e-02
-7.76517093e-01 -1.42691314e-01 3.03355455e-01 9.81241390e-02
-1.05207515e+00 1.24405250e-01 1.38584411e+00 1.56949796e-02
2.66282707e-01 -2.53634229e-02 -1.05834973e+00 7.05231428e-01
8.32092226e-01 -4.58154202e-01 -5.16783357e-01 -9.04379845e-01
9.92466748e-01 1.83484375e-01 2.34606296e-01 -8.62460911e-01
5.13009012e-01 -1.10970318e-01 -2.19771594e-01 -6.47824883e-01
6.86175078e-02 -6.61291480e-01 -3.63307297e-01 1.00746430e-01
-5.51183701e-01 5.40547729e-01 1.18679881e-01 4.30536836e-01
-4.67249751e-01 -1.01126575e+00 2.88618714e-01 -4.84312713e-01
-5.35319269e-01 -3.77489537e-01 -7.12827861e-01 8.88383210e-01
7.40196705e-01 -3.25873457e-02 -1.01464403e+00 -5.77521086e-01
-4.46993887e-01 5.82378626e-01 2.66485274e-01 8.29494178e-01
5.83819926e-01 -9.21461284e-01 -8.03860068e-01 -2.93083906e-01
5.33449948e-01 -3.39025557e-01 3.32804441e-01 6.22756898e-01
-3.96589845e-01 6.14692152e-01 1.43333212e-01 -4.17324156e-01
-1.44029593e+00 4.40348327e-01 5.46261489e-01 -5.97800910e-01
-1.05787255e-01 7.50113547e-01 -4.70053732e-01 -1.02385747e+00
5.96168078e-02 -4.47268426e-01 -8.62885773e-01 4.93510604e-01
8.00065398e-01 5.90480387e-01 1.65045649e-01 -4.66661304e-01
-7.18002245e-02 9.23026741e-01 -1.93093836e-01 6.57408535e-02
8.80835176e-01 -4.09291625e-01 -6.83078229e-01 7.48520613e-01
9.59274292e-01 1.26105934e-01 -3.00296277e-01 -2.79299557e-01
5.65651417e-01 -2.23527804e-01 -5.61229646e-01 -1.09475100e+00
-1.80620372e-01 9.18072343e-01 3.84883508e-02 8.08074057e-01
1.00242531e+00 4.04900223e-01 7.86148071e-01 5.90798616e-01
3.13931733e-01 -9.65193689e-01 3.76557827e-01 1.07730365e+00
1.20624852e+00 -1.19949436e+00 -2.11286217e-01 -4.86834377e-01
-3.06372881e-01 1.27481580e+00 8.62802386e-01 2.55144328e-01
4.96817678e-01 8.64224806e-02 3.13524723e-01 -6.50822580e-01
-1.05550063e+00 -4.21160758e-01 4.21473026e-01 4.66682941e-01
7.64375985e-01 -1.19458146e-01 -6.76438689e-01 6.35297835e-01
-4.80988801e-01 -3.09082121e-01 8.81343722e-01 1.29066098e+00
-6.99743986e-01 -1.45727289e+00 -4.98194039e-01 4.78383243e-01
-5.41537285e-01 -9.92907882e-02 -9.72239554e-01 6.76808536e-01
1.64661676e-01 1.74227405e+00 -4.27539229e-01 -1.84235647e-01
6.61238968e-01 3.93576592e-01 7.64518797e-01 -9.77783382e-01
-1.33377481e+00 -7.52420664e-01 3.67507488e-01 -3.34668458e-01
-5.82469642e-01 -4.19695199e-01 -9.06450748e-01 -2.05596179e-01
-4.62015122e-01 2.52488524e-01 2.45909885e-01 1.36359251e+00
3.43940258e-01 5.43164432e-01 -4.19320017e-02 1.24819726e-01
-8.62952948e-01 -1.18513381e+00 5.46653382e-02 3.73024464e-01
4.62753296e-01 -3.76556158e-01 -5.53045034e-01 -9.84958559e-02]
|
[11.379459381103516, 8.125174522399902]
|
9754e569-9dd6-41fe-bf96-68f24f5bdf4f
|
spin-simplifying-polar-invariance-for-neural
|
2111.14507
| null |
https://arxiv.org/abs/2111.14507v3
|
https://arxiv.org/pdf/2111.14507v3.pdf
|
SPIN: Simplifying Polar Invariance for Neural networks Application to vision-based irradiance forecasting
|
Translational invariance induced by pooling operations is an inherent property of convolutional neural networks, which facilitates numerous computer vision tasks such as classification. Yet to leverage rotational invariant tasks, convolutional architectures require specific rotational invariant layers or extensive data augmentation to learn from diverse rotated versions of a given spatial configuration. Unwrapping the image into its polar coordinates provides a more explicit representation to train a convolutional architecture as the rotational invariance becomes translational, hence the visually distinct but otherwise equivalent rotated versions of a given scene can be learnt from a single image. We show with two common vision-based solar irradiance forecasting challenges (i.e. using ground-taken sky images or satellite images), that this preprocessing step significantly improves prediction results by standardising the scene representation, while decreasing training time by a factor of 4 compared to augmenting data with rotations. In addition, this transformation magnifies the area surrounding the centre of the rotation, leading to more accurate short-term irradiance predictions.
|
['Joan Lasenby', 'Philippe Blanc', 'Guillaume Arbod', 'Anthony Hu', 'Quentin Paletta']
|
2021-11-29
| null | null | null | null |
['solar-irradiance-forecasting']
|
['time-series']
|
[ 6.15183830e-01 -2.78619617e-01 7.52034038e-02 -4.87254232e-01
-1.29385129e-01 -9.98130918e-01 8.08486462e-01 -2.32958004e-01
-3.98505241e-01 5.56603968e-01 2.26376772e-01 -3.85339826e-01
-3.61247733e-02 -6.83129728e-01 -8.29614758e-01 -1.02850938e+00
2.66126752e-01 -2.55086184e-01 -1.95011601e-01 -2.00149775e-01
1.79341480e-01 1.03916097e+00 -1.67751074e+00 1.23537593e-01
5.00601232e-01 9.71361876e-01 2.01205269e-01 8.92831504e-01
3.72857809e-01 4.78552610e-01 -4.47986543e-01 4.05602194e-02
6.92379236e-01 -6.66867718e-02 -5.37097394e-01 2.08056882e-01
1.07945979e+00 -4.28788424e-01 -5.40073633e-01 7.55024850e-01
3.57345253e-01 3.77189487e-01 5.61257184e-01 -6.54551625e-01
-8.94373894e-01 -9.79075879e-02 -5.67957520e-01 1.54234141e-01
9.08043459e-02 1.82803810e-01 9.12135005e-01 -6.46716177e-01
4.21473086e-01 7.98594117e-01 8.86158645e-01 1.14106918e-02
-1.35197628e+00 -8.02308843e-02 2.43774891e-01 -1.48276147e-02
-1.25033534e+00 -3.48421812e-01 6.76700532e-01 -4.27226514e-01
1.29574347e+00 4.72082436e-01 6.30864322e-01 7.44694889e-01
2.38269985e-01 9.65927467e-02 1.08479214e+00 -2.81608492e-01
6.74389154e-02 -3.05657715e-01 -1.13777317e-01 4.74818915e-01
3.97119015e-01 1.34762809e-01 -2.99394548e-01 3.96934271e-01
7.96516478e-01 2.99740583e-01 -7.46467710e-01 -3.97462487e-01
-1.23061371e+00 5.93007445e-01 1.21452677e+00 -1.52546838e-01
-5.32725573e-01 1.36132911e-01 4.73975502e-02 2.62433410e-01
4.38047171e-01 6.43488526e-01 -7.19313383e-01 3.72609764e-01
-8.05989504e-01 1.25071228e-01 3.54681402e-01 5.14388084e-01
9.97068405e-01 3.70248377e-01 -1.92373749e-02 5.79334974e-01
1.14256173e-01 9.74632621e-01 4.97877598e-01 -6.51568413e-01
2.27579370e-01 5.53178608e-01 1.54150024e-01 -7.62242675e-01
-7.61890173e-01 -7.75971889e-01 -1.04215050e+00 4.50160712e-01
3.74864578e-01 -2.26410955e-01 -1.39609551e+00 1.53552115e+00
9.55012888e-02 3.59527431e-02 2.65358984e-01 1.01240110e+00
6.17245317e-01 7.54744589e-01 -2.71417975e-01 1.63688004e-01
1.30803823e+00 -9.30665791e-01 1.77145877e-03 -7.05436349e-01
4.24686760e-01 -8.81766915e-01 8.67403150e-01 2.31034324e-01
-5.00916302e-01 -6.72153354e-01 -1.47616589e+00 -4.18336093e-01
-6.98474050e-01 1.92667022e-01 7.58646250e-01 5.26527464e-01
-1.25662923e+00 5.96244514e-01 -5.61764300e-01 -3.44004005e-01
3.72126400e-01 1.10340476e-01 -6.25890911e-01 -3.37466061e-01
-7.32130527e-01 1.00323284e+00 1.29962057e-01 2.06317320e-01
-5.32243490e-01 -9.65530455e-01 -1.10192037e+00 3.44175994e-01
-2.43077099e-01 -7.89856434e-01 1.19912946e+00 -1.36824095e+00
-1.46866357e+00 7.21088052e-01 -3.09429914e-01 -4.92166936e-01
3.18689436e-01 -2.86930233e-01 -1.99636802e-01 -3.05405110e-02
-3.47546399e-01 6.65790915e-01 1.29657125e+00 -9.17171478e-01
-3.02144170e-01 -3.28948349e-01 3.08105499e-01 6.43700540e-01
7.50356680e-03 -2.97476083e-01 -1.26768962e-01 -5.72154641e-01
3.58340293e-01 -1.17604363e+00 -1.96927115e-01 2.80570954e-01
-1.65516496e-01 3.27984214e-01 8.89149368e-01 -6.44816399e-01
2.53349423e-01 -2.14866447e+00 -2.56710220e-02 4.39314581e-02
-1.65359713e-02 2.66251415e-01 -3.22063059e-01 1.36363864e-01
-6.00308716e-01 -5.14381379e-02 -1.40794829e-01 -2.45050415e-01
-3.36922497e-01 2.38644466e-01 -6.49711788e-01 7.59734333e-01
5.22330642e-01 9.50533092e-01 -5.88598907e-01 4.42656755e-01
6.25408590e-01 9.91641760e-01 -3.11385870e-01 -1.29092503e-02
-9.14480910e-02 5.06529450e-01 1.92134798e-01 2.54198343e-01
9.56021845e-01 -2.65371680e-01 9.79867131e-02 -4.24982160e-01
-3.41740847e-01 3.45236093e-01 -8.06436121e-01 1.52088904e+00
-8.24604452e-01 1.25318325e+00 -4.34434652e-01 -9.01721358e-01
1.05628240e+00 7.04354374e-03 2.44859710e-01 -9.29816604e-01
-5.60994335e-02 -3.15795690e-01 -1.57807350e-01 -6.22192807e-02
6.55521870e-01 -3.04735769e-02 4.38661426e-01 2.79257059e-01
-4.58134338e-02 -3.38019401e-01 -2.81372070e-01 -2.90326387e-01
6.25177503e-01 3.27935338e-01 2.58519202e-01 -1.50612220e-01
1.45683467e-01 -7.15325847e-02 3.65233779e-01 5.54749489e-01
1.44376665e-01 1.10716355e+00 3.09604052e-02 -1.11449766e+00
-1.14635277e+00 -9.32327449e-01 -3.38326812e-01 8.51590395e-01
-1.57928374e-02 1.16115071e-01 -3.68463635e-01 -2.76455104e-01
-6.42203912e-02 5.01078248e-01 -5.66886008e-01 -1.88892316e-02
-5.17475963e-01 -9.74597871e-01 1.78105533e-01 6.04190707e-01
7.61789024e-01 -7.02403545e-01 -1.02664745e+00 -1.75350845e-01
8.61609802e-02 -1.19266629e+00 1.63379386e-02 5.55655420e-01
-8.05554867e-01 -1.11504900e+00 -8.41607749e-01 -4.20644701e-01
7.48598099e-01 1.00982940e+00 1.20512474e+00 -1.50243357e-01
-5.47170758e-01 2.12108985e-01 -1.71708405e-01 -5.23090243e-01
1.59046993e-01 -1.13974825e-01 -6.27281144e-03 -9.93267968e-02
2.55553033e-02 -5.23785710e-01 -8.65455508e-01 1.46439865e-01
-9.33370054e-01 3.37854117e-01 5.88443160e-01 6.51952624e-01
5.18089235e-01 -3.98691595e-02 7.16118813e-02 -4.89465058e-01
1.77402887e-02 -1.29176378e-01 -8.69850695e-01 2.44466648e-01
-3.72152448e-01 3.10409874e-01 6.94225848e-01 -1.55290470e-01
-1.27581882e+00 3.03421348e-01 2.10002974e-01 -3.00188243e-01
-3.62120718e-01 4.88499790e-01 3.08621109e-01 -3.83231163e-01
1.09792042e+00 1.61973298e-01 -2.72742867e-01 -3.78610700e-01
7.68410325e-01 3.73740673e-01 7.05954731e-01 -3.96784302e-03
1.13706207e+00 7.50751436e-01 3.86665434e-01 -1.12280202e+00
-8.58797729e-01 -3.62959176e-01 -9.53938842e-01 -2.45213527e-02
9.45256591e-01 -1.32944620e+00 -4.18784261e-01 5.70550978e-01
-9.99001563e-01 -4.03175473e-01 -2.22255811e-01 4.76593643e-01
-1.82039797e-01 2.22312883e-01 -2.93761343e-02 -4.61461812e-01
-3.63606393e-01 -8.68869483e-01 1.02962124e+00 7.60954380e-01
8.67895037e-02 -8.19467843e-01 2.66926378e-01 9.63885710e-02
6.77433670e-01 2.35644668e-01 9.10379708e-01 -1.03614904e-01
-6.81953311e-01 -3.82393897e-01 -6.04461193e-01 4.47048008e-01
3.46572250e-01 3.03455949e-01 -1.51708663e+00 -5.11315882e-01
-2.09035486e-01 -2.25084469e-01 1.17024052e+00 3.91242385e-01
9.81514156e-01 -1.98974088e-01 7.84630924e-02 1.17752862e+00
1.54498672e+00 -7.19147250e-02 6.64333403e-01 4.64202642e-01
7.66976535e-01 2.64103264e-01 5.33386059e-02 1.24701098e-01
1.87425703e-01 5.17157257e-01 6.94016695e-01 -5.53009272e-01
-2.82429934e-01 2.02740524e-02 1.98746681e-01 1.78280517e-01
-4.21875626e-01 -6.77167550e-02 -7.34425187e-01 4.08508182e-01
-1.44448519e+00 -8.51768672e-01 1.46289021e-01 2.48323750e+00
4.60327268e-01 -1.33741006e-01 -5.29892087e-01 6.55007660e-02
3.26931149e-01 6.38259947e-01 -8.39620948e-01 -4.76557136e-01
-5.39360106e-01 4.95336473e-01 8.99827898e-01 3.73320460e-01
-1.29776847e+00 7.33651817e-01 6.74801731e+00 -3.89551260e-02
-1.92055726e+00 -4.91208136e-01 6.17232621e-01 -7.72645250e-02
-1.11088470e-01 4.89298441e-02 -4.65936631e-01 -1.14303678e-01
7.95576751e-01 1.23505905e-01 6.55318558e-01 9.54996943e-01
2.00605661e-01 -1.51334643e-01 -9.91365671e-01 1.02208805e+00
2.84929872e-01 -1.38020527e+00 8.18591565e-02 -8.10014680e-02
1.03518248e+00 7.36995220e-01 3.46098006e-01 9.38625038e-02
1.73612386e-01 -1.36417675e+00 3.90515953e-01 6.38467014e-01
8.62859905e-01 -4.89885747e-01 4.89982516e-01 5.73279187e-02
-1.05147445e+00 4.18003425e-02 -7.94815004e-01 -4.01015610e-01
-2.68350124e-01 3.93360943e-01 -8.98755550e-01 9.23754454e-01
8.79741907e-01 8.73654187e-01 -9.24771130e-01 1.04957652e+00
-3.88283610e-01 2.79998749e-01 -4.71271843e-01 4.94365066e-01
1.51784882e-01 -3.59238535e-01 9.13967788e-02 8.58751774e-01
4.17359769e-01 -1.76470473e-01 -2.56140858e-01 3.85374308e-01
-1.29186586e-01 -3.15760970e-01 -8.74769628e-01 1.89741924e-01
3.93903665e-02 1.51724935e+00 -4.25024778e-01 -5.71018532e-02
-7.01100945e-01 1.10182583e+00 2.41552219e-01 7.10975051e-01
-6.52829111e-01 -4.77244824e-01 1.00730562e+00 -1.05685011e-01
6.52113020e-01 -4.81280416e-01 -3.62286299e-01 -1.38666272e+00
-4.35157791e-02 -5.53816020e-01 -5.70498072e-02 -1.40963316e+00
-8.55020404e-01 5.79365015e-01 -3.15622658e-01 -1.13677061e+00
-2.20136210e-01 -1.04194248e+00 -8.96194875e-01 1.35281229e+00
-2.08027315e+00 -1.33996546e+00 -9.95658219e-01 4.03943837e-01
3.96399707e-01 7.86612555e-02 8.91645968e-01 -1.89992502e-01
-4.25501794e-01 3.56957108e-01 2.94575185e-01 7.31619969e-02
8.02632928e-01 -1.24901044e+00 7.43502378e-01 1.17169881e+00
5.39908469e-01 5.74478686e-01 6.40893877e-01 6.96685538e-02
-1.37085855e+00 -1.31076896e+00 5.74788332e-01 -4.37355161e-01
2.79751599e-01 -1.24771679e-02 -8.81512165e-01 8.02732944e-01
3.56043339e-01 3.14013928e-01 5.96981823e-01 1.84867024e-01
-1.00901949e+00 -3.65452081e-01 -8.58867705e-01 7.68270135e-01
7.52139449e-01 -8.30920875e-01 -2.98774928e-01 4.98254359e-01
6.41389310e-01 -4.09699291e-01 -6.58002377e-01 3.69441807e-01
6.38245225e-01 -1.07095563e+00 1.24064422e+00 -7.43065655e-01
3.97833765e-01 -5.89475811e-01 -3.03389758e-01 -1.51752222e+00
-5.12164056e-01 -3.92054051e-01 2.90660650e-01 5.66207528e-01
3.10870439e-01 -8.53877366e-01 5.48765361e-01 4.89050090e-01
-9.96778905e-02 -4.54943061e-01 -8.01138461e-01 -5.34185529e-01
9.92370248e-02 -1.51395291e-01 5.65687299e-01 9.26687360e-01
-6.56538427e-01 2.73878962e-01 -2.12636292e-01 7.30140746e-01
2.45670944e-01 3.74852121e-01 8.69499028e-01 -1.15430546e+00
-8.22187364e-02 -3.47998738e-01 -5.57762504e-01 -1.14797544e+00
-1.32079899e-01 -6.56720400e-01 -7.19745010e-02 -1.46858478e+00
-1.44348368e-02 -1.56264558e-01 -3.76326025e-01 7.76303589e-01
-1.31571114e-01 6.73730969e-01 2.56926417e-01 1.40415117e-01
-2.08023228e-02 4.07188624e-01 1.04792154e+00 -1.08357742e-01
-8.63406807e-02 -6.46170601e-02 -6.66076183e-01 7.22970605e-01
1.08204138e+00 5.20267785e-02 -4.68098879e-01 -9.44261014e-01
5.76922297e-01 -4.55601662e-01 6.70394242e-01 -1.09335244e+00
-6.81574941e-02 -1.72549143e-01 1.09493613e+00 -4.05293703e-01
4.58940119e-01 -6.46835744e-01 1.66725412e-01 2.79821813e-01
-2.53592338e-02 1.26975387e-01 5.45004666e-01 4.89885181e-01
-1.40224127e-02 -8.18569958e-03 9.56456065e-01 7.14076459e-02
-7.91307569e-01 1.74487934e-01 -2.39134550e-01 -4.44017023e-01
6.93234086e-01 -2.22304061e-01 -9.32758689e-01 -3.55541199e-01
-1.96642458e-01 -3.49207282e-01 7.48921514e-01 4.41893011e-01
2.68144369e-01 -8.98311436e-01 -4.61404294e-01 5.13069808e-01
1.50839597e-01 4.30947185e-01 4.03900236e-01 5.89404106e-01
-1.04686964e+00 6.10323787e-01 -4.73690361e-01 -7.44851887e-01
-1.32448113e+00 3.74285340e-01 6.59465551e-01 1.12121165e-01
-7.24880338e-01 6.52228296e-01 4.51579064e-01 -4.20000255e-01
-2.46825859e-01 -8.01556170e-01 -1.49843082e-01 -1.91895172e-01
4.49850559e-01 -1.24684602e-01 4.90837216e-01 -6.61215544e-01
-3.32788348e-01 8.14220011e-01 3.92053947e-02 1.38887420e-01
1.40138257e+00 2.50984430e-02 1.04381420e-01 2.24222746e-02
1.13896656e+00 -5.92623465e-02 -1.84172964e+00 -3.94341767e-01
-5.11511385e-01 -4.91279572e-01 4.43964124e-01 -9.67220902e-01
-1.02343595e+00 1.04298580e+00 7.87145019e-01 1.66516989e-01
1.24724555e+00 -4.31907147e-01 7.12237731e-02 1.03791356e+00
-9.88508761e-02 -5.01840472e-01 -1.94238558e-01 7.60888517e-01
1.04712892e+00 -1.38052225e+00 4.60089624e-01 2.70143524e-02
-5.74896574e-01 1.33841801e+00 2.68534690e-01 -1.69028759e-01
4.74365562e-01 -1.08116053e-01 4.65688407e-01 1.29705757e-01
-4.06221628e-01 -2.91629881e-01 6.87871814e-01 7.47668505e-01
4.03290987e-01 9.38132964e-03 6.40311062e-01 -1.00205056e-01
-3.84192288e-01 -4.70288664e-01 5.03494024e-01 8.12882960e-01
-5.22698462e-01 -6.21852636e-01 -3.22587103e-01 6.86589852e-02
-6.34791702e-02 -3.28198612e-01 -3.14156532e-01 5.85701048e-01
-1.11303337e-01 6.78799272e-01 5.29520392e-01 -2.40704510e-02
8.31211358e-02 3.80739383e-02 5.29639542e-01 -3.71168405e-01
-1.17940195e-01 -3.19291174e-01 -2.13785261e-01 -5.28191805e-01
-4.13931787e-01 -3.31136972e-01 -6.82962656e-01 -3.53436708e-01
-5.53525910e-02 -5.64411521e-01 1.15866339e+00 8.47512543e-01
4.42752451e-01 5.68286717e-01 7.91040480e-01 -1.44152772e+00
-4.16649371e-01 -8.28574121e-01 -1.55758277e-01 3.58621806e-01
8.26923490e-01 -4.47745204e-01 -4.20688659e-01 2.42686331e-01]
|
[9.943340301513672, -2.6662745475769043]
|
ec037092-5e91-4f02-b9b1-a6f05cb3056b
|
improvement-of-verbnet-like-resources-by
| null | null |
https://aclanthology.org/W16-3809
|
https://aclanthology.org/W16-3809.pdf
|
Improvement of VerbNet-like resources by frame typing
|
Verbenet is a French lexicon developed by {``}translation{''} of its English counterpart {---} VerbNet (Kipper-Schuler, 2005){---}and treatment of the specificities of French syntax (Pradet et al., 2014; Danlos et al., 2016). One difficulty encountered in its development springs from the fact that the list of (potentially numerous) frames has no internal organization. This paper proposes a type system for frames that shows whether two frames are variants of a given alternation. Frame typing facilitates coherence checking of the resource in a {``}virtuous circle{''}. We present the principles underlying a program we developed and used to automatically type frames in VerbeNet. We also show that our system is portable to other languages.
|
['Matthieu Constant', 'Lucie Barque', 'Laurence Danlos']
|
2016-12-01
| null | null | null |
ws-2016-12
|
['stock-market-prediction']
|
['time-series']
|
[ 1.33616984e-01 2.60601968e-01 -1.34511992e-01 -3.01217109e-01
-4.37513173e-01 -1.05341196e+00 7.88361847e-01 1.65308014e-01
-1.73208177e-01 9.68841732e-01 3.38931739e-01 -7.03787088e-01
-2.16542423e-01 -8.64037871e-01 -4.29944396e-01 -8.61354321e-02
1.37193605e-01 2.20221266e-01 6.31431460e-01 -7.37451375e-01
3.20751667e-02 1.57683372e-01 -1.68263113e+00 6.82446241e-01
2.75594532e-01 6.91525459e-01 4.51625615e-01 6.73569858e-01
-3.11339885e-01 1.11592042e+00 -6.42420292e-01 -7.63147354e-01
8.72341245e-02 -6.12063348e-01 -1.29882050e+00 -2.50471771e-01
3.29431057e-01 1.13641068e-01 -1.69725940e-01 9.49143529e-01
7.13493526e-02 -6.17290139e-02 3.67320091e-01 -1.03826892e+00
-4.91322160e-01 1.11028945e+00 4.68586385e-02 3.87942761e-01
9.42106545e-01 8.09137672e-02 1.03717184e+00 -1.00579989e+00
1.16568887e+00 1.32661748e+00 7.90342987e-01 5.87654829e-01
-1.10444653e+00 -2.99022853e-01 -6.59779534e-02 2.34523624e-01
-1.11744952e+00 -5.13328075e-01 3.71328592e-01 -6.44665897e-01
1.43889153e+00 6.01807594e-01 1.07235110e+00 1.01435804e+00
2.39257053e-01 4.35278565e-01 1.26125622e+00 -8.87704968e-01
-1.63982853e-01 -2.11168617e-01 4.20056701e-01 7.31611729e-01
2.68969834e-01 -2.56521683e-02 -5.19945443e-01 -1.49438411e-01
9.35859263e-01 -6.03188097e-01 6.62531629e-02 1.63628384e-01
-1.38419700e+00 5.76837838e-01 -2.51728833e-01 7.85525322e-01
8.95396620e-03 2.18233407e-01 7.54747152e-01 3.12268138e-01
-5.85599989e-02 1.87461659e-01 -4.70303595e-01 -6.52122080e-01
-7.32064426e-01 6.61330760e-01 8.97552311e-01 1.41810346e+00
5.38844049e-01 -1.05230957e-01 -1.30360156e-01 4.45256591e-01
4.57453310e-01 2.01580599e-01 2.11764157e-01 -1.28784418e+00
3.66355538e-01 5.80980539e-01 3.56079936e-01 -6.29996121e-01
-4.20417070e-01 2.37117838e-02 -1.40599027e-01 -5.74618429e-02
9.44531202e-01 -1.24878734e-02 -1.23276368e-01 1.91506970e+00
1.50334969e-01 -2.01516137e-01 2.16642663e-01 5.57047367e-01
1.00719571e+00 4.19515431e-01 2.58974165e-01 -7.60861337e-01
1.78679204e+00 -4.72254872e-01 -1.01898038e+00 1.72439605e-01
7.93646574e-01 -1.23306894e+00 1.40929770e+00 2.54350662e-01
-1.54400527e+00 -2.90275544e-01 -6.70424163e-01 -2.96557993e-01
-2.43688524e-01 1.07331328e-01 7.37336695e-01 8.38787854e-01
-9.23319697e-01 5.07305384e-01 -7.25239396e-01 -6.95537686e-01
-7.23224208e-02 1.92738116e-01 -1.11565910e-01 5.91157615e-01
-1.35850704e+00 1.02025151e+00 6.11197233e-01 -7.57026393e-03
-3.59418809e-01 -3.22497368e-01 -7.76071012e-01 -3.44762415e-01
6.93000734e-01 -4.72873092e-01 1.63358986e+00 -1.21266198e+00
-1.55421567e+00 1.25728285e+00 -2.93979943e-01 -5.06384373e-01
5.37020147e-01 -2.90658772e-01 -7.68861115e-01 -3.93273383e-02
4.46106762e-01 1.29553139e-01 3.46600771e-01 -7.95293093e-01
-8.75706553e-01 1.17603146e-01 7.96525538e-01 -1.66615136e-02
1.62388548e-01 8.95366728e-01 2.07148362e-02 -1.03650451e+00
5.37494570e-02 -8.97668600e-01 4.40780282e-01 -3.24558228e-01
-1.49878472e-01 -6.26362920e-01 6.50527775e-01 -4.83957350e-01
2.06925607e+00 -1.93941820e+00 8.01266804e-02 -1.45365894e-01
-7.34053878e-03 3.08826178e-01 2.13064000e-01 9.46338058e-01
-3.64944249e-01 2.67604828e-01 -2.38241136e-01 5.36647379e-01
2.88898200e-01 6.28968716e-01 -3.01177949e-01 4.95716929e-01
-8.25131759e-02 6.77895248e-01 -1.12200665e+00 -4.74753678e-01
1.69596076e-01 7.59377703e-02 -6.62039161e-01 -2.33159333e-01
-5.79796195e-01 2.11525247e-01 -1.41571775e-01 4.99935210e-01
2.22695351e-01 -6.24378398e-02 7.45384097e-01 -3.57914537e-01
-8.30387294e-01 9.36063170e-01 -1.57574999e+00 1.58423829e+00
-1.61541298e-01 5.98173440e-01 -2.56240934e-01 -4.23982412e-01
4.28902775e-01 7.21991181e-01 2.40739584e-01 -3.90334010e-01
2.69819766e-01 5.09917080e-01 3.39570224e-01 -5.33145189e-01
5.98875403e-01 2.21546050e-02 -4.01640564e-01 4.18676853e-01
-1.17822562e-03 -2.05344111e-01 1.04802680e+00 5.41919246e-02
9.68632579e-01 8.43830764e-01 8.68577003e-01 -8.94064724e-01
9.12153125e-01 2.22041488e-01 7.85159469e-01 4.47935909e-01
-9.83461067e-02 1.76536977e-01 7.80136168e-01 -7.62899280e-01
-1.03858566e+00 -1.08912981e+00 -4.36231196e-01 1.17410862e+00
-1.25619605e-01 -1.33305705e+00 -8.26404572e-01 -5.75649858e-01
-4.97824341e-01 9.33675826e-01 -6.21786475e-01 4.79125500e-01
-1.15080070e+00 -1.97043329e-01 8.93201232e-01 4.92232740e-01
1.97463974e-01 -1.11253381e+00 -1.30195093e+00 4.62851673e-01
-4.55578178e-01 -9.10026431e-01 -1.29275516e-01 -4.95889112e-02
-2.37852395e-01 -1.25974846e+00 1.20494872e-01 -7.00250268e-01
8.66658315e-02 -2.18278199e-01 1.51553857e+00 3.06218117e-01
8.73777792e-02 2.68678576e-01 -5.70392072e-01 -4.57893372e-01
-6.74238443e-01 -2.37614036e-01 6.07622042e-02 -5.01885474e-01
3.08759153e-01 -2.01284885e-01 -4.77234237e-02 2.83467740e-01
-8.59276235e-01 8.19204375e-02 -4.20734078e-01 6.20545268e-01
5.09479105e-01 -2.27498174e-01 1.17110915e-01 -1.08575070e+00
5.64085960e-01 -6.52118251e-02 -6.56255424e-01 2.71649003e-01
-9.48202536e-02 -1.92228422e-01 4.24144179e-01 -1.14026345e-01
-9.45957124e-01 -3.73047978e-01 -3.56215268e-01 2.88631767e-01
-1.53987780e-01 4.58716542e-01 -2.40012452e-01 4.07946706e-01
7.43368983e-01 -3.06942999e-01 -4.95954007e-01 -1.58971027e-01
5.01268208e-01 2.68627673e-01 5.09431362e-01 -1.04208422e+00
4.45431411e-01 3.24124306e-01 9.70892981e-02 -6.69020891e-01
-6.08051240e-01 2.17206664e-02 -8.49164724e-01 -4.81714606e-01
7.73540139e-01 -9.92331862e-01 -6.29652858e-01 6.40924424e-02
-1.38817275e+00 -4.70332086e-01 -6.40654922e-01 3.29875052e-01
-9.14422274e-01 3.82127136e-01 -7.37113655e-01 -6.70440793e-01
2.09562421e-01 -1.03746521e+00 8.57864976e-01 2.49076225e-02
-1.09794319e+00 -9.25941408e-01 6.82821125e-02 -3.24248195e-01
-1.45495459e-01 3.16300601e-01 1.07198083e+00 -3.40561807e-01
-2.06655174e-01 5.64814687e-01 6.02666177e-02 -2.05386832e-01
1.61584213e-01 4.94628698e-01 -5.18214166e-01 1.35472966e-02
5.36103956e-02 2.83130966e-02 -1.41078949e-01 9.66537222e-02
4.51636553e-01 -6.64208591e-01 -6.77142888e-02 2.18303144e-01
1.53954899e+00 4.65370834e-01 7.15152740e-01 7.49408007e-01
2.69105822e-01 5.84166765e-01 7.68922687e-01 2.53537416e-01
5.75187624e-01 8.94775927e-01 -5.92284687e-02 6.68101966e-01
-3.06350827e-01 -1.05467938e-01 5.91229379e-01 8.48626196e-01
-5.31158268e-01 -1.71575293e-01 -9.24280345e-01 5.38671672e-01
-2.08123350e+00 -1.44880307e+00 -9.31049347e-01 1.80356419e+00
1.03576946e+00 2.15745673e-01 5.46382189e-01 3.51138353e-01
7.18642175e-01 8.16215277e-02 5.12864053e-01 -8.42130899e-01
-4.70922023e-01 5.43586314e-01 6.51209280e-02 6.46705449e-01
-9.31565225e-01 1.44585955e+00 6.46646500e+00 6.41545355e-01
-7.80984998e-01 2.85855502e-01 -2.48102739e-01 1.92546338e-01
-4.97533619e-01 4.84700531e-01 -8.17583919e-01 5.93709230e-01
9.13639605e-01 -3.77077162e-01 2.82391518e-01 3.05589169e-01
3.62074733e-01 -2.44551405e-01 -1.22986102e+00 6.50705874e-01
-1.18616866e-02 -1.60469580e+00 -3.32888067e-02 -2.66571015e-01
4.54932421e-01 -3.64558339e-01 -4.41071898e-01 8.34336132e-02
4.37950224e-01 -6.22375190e-01 1.74453115e+00 5.13998270e-01
9.70301211e-01 -4.44454789e-01 3.13466161e-01 4.91926745e-02
-1.51387036e+00 4.50617112e-02 -6.45405874e-02 -5.54597259e-01
3.59471649e-01 2.08979934e-01 -2.43491858e-01 7.30575144e-01
6.65163338e-01 4.87545431e-01 -5.32767594e-01 5.60507059e-01
-6.27508700e-01 7.62715459e-01 -3.88348490e-01 -9.77005288e-02
1.50179207e-01 -4.34217751e-01 8.53374541e-01 1.55640328e+00
3.53348136e-01 2.94125378e-01 1.99390605e-01 5.92843711e-01
4.03521985e-01 2.81833798e-01 -6.29309535e-01 -2.20215078e-02
6.14580989e-01 7.68811524e-01 -1.21843565e+00 -4.55295771e-01
-6.53691947e-01 4.66986626e-01 3.94644449e-04 6.90298975e-02
-9.03696775e-01 -2.28139818e-01 3.92476499e-01 4.40985620e-01
2.36673459e-01 -3.70320797e-01 -3.19699019e-01 -1.08607936e+00
2.42213421e-02 -1.03909659e+00 4.88164365e-01 -8.56213510e-01
-8.92372370e-01 6.70971394e-01 5.08700609e-01 -9.88552511e-01
-4.11269605e-01 -6.66199327e-01 -1.92301050e-01 7.55913258e-01
-5.91426909e-01 -1.38243484e+00 2.36624196e-01 7.34552920e-01
5.57212591e-01 1.55757391e-03 1.05328882e+00 1.81605786e-01
-3.88096333e-01 1.07380055e-01 -5.02661526e-01 1.53531596e-01
4.59835798e-01 -1.31749845e+00 3.86390030e-01 1.24596131e+00
1.32802397e-01 8.85423422e-01 9.87503290e-01 -7.68870592e-01
-1.23101151e+00 -6.52410686e-01 1.71578074e+00 -5.60742080e-01
1.22638571e+00 -3.27178061e-01 -6.38400078e-01 1.19545507e+00
7.91868865e-01 -4.48359609e-01 6.28971100e-01 -1.93217367e-01
-2.47474134e-01 2.21009567e-01 -8.54441941e-01 9.08522606e-01
1.42265868e+00 -7.70841479e-01 -1.11231375e+00 4.45152283e-01
5.56480587e-01 -7.00556338e-01 -1.00266802e+00 3.97542864e-01
5.94728112e-01 -1.41914344e+00 6.11637652e-01 -4.71932560e-01
2.01658458e-01 -6.27219260e-01 -4.04883981e-01 -5.72739720e-01
-1.85099527e-01 -1.27427542e+00 1.33121997e-01 1.39960325e+00
2.46969610e-01 -5.20788670e-01 8.44284669e-02 2.47781321e-01
-4.80403036e-01 -1.38125062e-01 -1.05134249e+00 -5.46037376e-01
2.07944438e-01 -7.83347070e-01 6.14958048e-01 9.51748490e-01
4.39040691e-01 3.36451560e-01 1.29672363e-01 -2.52069175e-01
5.96311316e-02 -8.69838372e-02 5.85028827e-01 -1.19891655e+00
-4.04780775e-01 -3.69027823e-01 -1.45801902e-01 -5.32159984e-01
2.17226490e-01 -1.08635497e+00 -3.69908899e-01 -1.48808086e+00
-2.89364159e-01 -3.43220502e-01 3.03568244e-01 5.49982429e-01
3.20872337e-01 1.75029710e-01 4.92735565e-01 2.25340456e-01
-4.34058428e-01 -2.79053807e-01 8.83056819e-01 4.19078976e-01
-4.11425009e-02 -3.33047420e-01 -4.32496816e-01 1.22028196e+00
5.51518202e-01 -4.83066797e-01 -2.63936996e-01 -4.27743733e-01
1.02394772e+00 1.59283429e-02 4.12577540e-01 -8.53600383e-01
-4.15673740e-02 -3.40526789e-01 -1.85036600e-01 -5.16024649e-01
-1.50396049e-01 -6.16676927e-01 8.02219868e-01 4.15264010e-01
-2.55710147e-02 1.04801810e+00 4.55687940e-01 -2.00040609e-01
8.33263714e-03 -6.86225832e-01 5.77532709e-01 -5.49902916e-01
-7.55248845e-01 -3.57057303e-01 -9.10860538e-01 2.98045784e-01
1.06391895e+00 -2.31209740e-01 -4.74863499e-01 1.31774887e-01
-7.70045757e-01 -2.58828670e-01 7.09951580e-01 1.44843727e-01
8.57342258e-02 -1.39768887e+00 -3.12534273e-01 5.26155047e-02
-2.24344376e-02 -3.02675694e-01 -2.60392874e-01 8.00027251e-01
-1.23277116e+00 2.68061489e-01 -2.97364175e-01 -4.23314601e-01
-1.33053279e+00 3.24365526e-01 1.58838198e-01 -1.85308039e-01
-7.27856755e-01 5.65611839e-01 -3.49153638e-01 -2.43544858e-02
-3.47571224e-01 -5.92415094e-01 -1.80936471e-01 3.63451004e-01
4.24434900e-01 4.81154323e-01 2.04898771e-02 -1.25670397e+00
-6.23558104e-01 5.18217266e-01 4.18453544e-01 -4.90037978e-01
9.67452765e-01 -4.30967599e-01 -7.49774933e-01 8.94005299e-01
6.18121982e-01 4.50616926e-01 -6.76523685e-01 6.61180839e-02
2.68474817e-01 -3.14518958e-01 -7.57162154e-01 -7.84113467e-01
-1.62492558e-01 2.68686026e-01 2.76097536e-01 7.57811069e-01
8.74630868e-01 5.34339473e-02 5.16424000e-01 1.51187941e-01
5.21450758e-01 -1.25974560e+00 -5.20473123e-01 1.00760078e+00
9.04193819e-01 -1.99624717e-01 -1.08965367e-01 -8.02501678e-01
-3.13730806e-01 1.44227934e+00 2.17767596e-01 -7.67841861e-02
5.53989887e-01 6.08517945e-01 3.09554320e-02 -2.80541271e-01
-7.60855258e-01 -5.73261559e-01 -1.14714414e-01 5.58906317e-01
8.84143770e-01 2.19467252e-01 -1.42476058e+00 5.90144515e-01
-8.18057597e-01 2.80693054e-01 8.65675032e-01 1.29990029e+00
-2.09707960e-01 -1.69741559e+00 -5.81447959e-01 -2.13131569e-02
-7.02636182e-01 -2.33134225e-01 -3.17488045e-01 1.33607709e+00
8.10102940e-01 1.08143771e+00 1.28484085e-01 -1.18231714e-01
5.54222703e-01 2.18287393e-01 1.03858459e+00 -9.96074498e-01
-1.26306152e+00 6.14850402e-01 6.78441525e-01 -4.12130654e-01
-1.17366302e+00 -1.16094601e+00 -1.45871544e+00 -4.31946009e-01
-5.86961880e-02 2.61258304e-01 -9.12081152e-02 1.23042488e+00
-3.23357731e-01 5.34755290e-01 -1.81927010e-01 -3.75030816e-01
1.45192504e-01 -5.99664807e-01 -3.37521821e-01 3.92713219e-01
-1.99270636e-01 -8.15876901e-01 3.19549292e-02 5.76168418e-01]
|
[10.132232666015625, 9.340715408325195]
|
263ec71b-7999-46c1-8865-458cf75f5c88
|
persian-semantic-role-labeling-using-transfer
|
2306.10339
| null |
https://arxiv.org/abs/2306.10339v1
|
https://arxiv.org/pdf/2306.10339v1.pdf
|
Persian Semantic Role Labeling Using Transfer Learning and BERT-Based Models
|
Semantic role labeling (SRL) is the process of detecting the predicate-argument structure of each predicate in a sentence. SRL plays a crucial role as a pre-processing step in many NLP applications such as topic and concept extraction, question answering, summarization, machine translation, sentiment analysis, and text mining. Recently, in many languages, unified SRL dragged lots of attention due to its outstanding performance, which is the result of overcoming the error propagation problem. However, regarding the Persian language, all previous works have focused on traditional methods of SRL leading to a drop in accuracy and imposing expensive feature extraction steps in terms of financial resources, time and energy consumption. In this work, we present an end-to-end SRL method that not only eliminates the need for feature extraction but also outperforms existing methods in facing new samples in practical situations. The proposed method does not employ any auxiliary features and shows more than 16 (83.16) percent improvement in accuracy against previous methods in similar circumstances.
|
['Behrouz Minaei Bidgoli', 'Nasim Khozouei', 'Erfan Khedersolh Sadeh', 'Sayyed Ali Hossayni', 'Saeideh Niksirat Aghdam']
|
2023-06-17
| null | null | null | null |
['transfer-learning', 'question-answering', 'sentiment-analysis', 'machine-translation', 'semantic-role-labeling']
|
['miscellaneous', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 5.34580946e-01 6.76526353e-02 -2.16880694e-01 -2.93926448e-01
-8.13768387e-01 -6.56430304e-01 7.70230055e-01 8.67039800e-01
-7.01588035e-01 1.09904826e+00 4.12423372e-01 -2.95234382e-01
-1.22370712e-01 -6.86990857e-01 -1.21048011e-01 -5.50879121e-01
2.92425931e-01 3.10301512e-01 4.38644081e-01 -2.97400922e-01
5.68638206e-01 2.22472996e-01 -1.39013815e+00 7.65626729e-02
1.09188759e+00 7.16499865e-01 1.59424111e-01 2.03952238e-01
-4.58200395e-01 9.18588758e-01 -7.15791404e-01 -3.07477772e-01
-1.09785676e-01 -4.34685618e-01 -9.69681621e-01 4.15448770e-02
-1.15999669e-01 2.09609523e-01 5.37094101e-02 1.06690359e+00
4.22483414e-01 3.34659934e-01 3.77738804e-01 -9.51864898e-01
-2.45416179e-01 5.42589188e-01 -8.20323110e-01 3.54583442e-01
5.58397889e-01 -5.02515495e-01 1.26384580e+00 -7.18216538e-01
4.98916030e-01 1.25544357e+00 1.87169597e-01 2.95534313e-01
-7.55260706e-01 -5.54214895e-01 1.97668478e-01 3.29656839e-01
-9.88976359e-01 -3.60935330e-01 8.48145545e-01 -1.58496335e-01
9.43806767e-01 3.21090072e-01 2.97324091e-01 4.67316478e-01
1.29538432e-01 9.83183622e-01 1.06054211e+00 -6.15140021e-01
2.54305631e-01 1.45254284e-01 4.44302648e-01 4.90657657e-01
4.27424192e-01 -8.57705474e-01 -5.86901128e-01 -2.81885028e-01
2.05642790e-01 -2.15192974e-01 -1.69537559e-01 1.70876443e-01
-1.07241416e+00 8.93209755e-01 -1.80493668e-01 5.37820339e-01
-4.27248627e-01 -3.19652110e-01 6.49951220e-01 9.55141187e-02
6.26345456e-01 6.62785947e-01 -6.80034101e-01 -3.43917876e-01
-8.64165902e-01 1.29491642e-01 6.94981575e-01 5.53435564e-01
3.48271191e-01 -1.73849240e-01 -1.14269815e-01 9.85306084e-01
-4.58446443e-02 2.36528680e-01 7.73891389e-01 -6.41047955e-01
6.44314051e-01 1.11300087e+00 2.97670215e-01 -1.02614403e+00
-5.22395253e-01 -3.98025304e-01 -6.93173289e-01 -5.33691645e-01
1.75083771e-01 -9.19067264e-02 -6.94233894e-01 1.59216177e+00
5.70538104e-01 -2.42405176e-01 2.85401255e-01 5.88702917e-01
1.02778280e+00 8.13020885e-01 2.76009589e-01 -5.92049181e-01
1.69139957e+00 -7.69467890e-01 -7.12508500e-01 -2.12076962e-01
4.93687332e-01 -1.04307246e+00 9.53206062e-01 3.99104357e-01
-7.88605154e-01 -1.95362449e-01 -9.99443054e-01 -9.24094319e-02
-2.41063252e-01 3.52677882e-01 8.09319079e-01 4.41660404e-01
-6.03328884e-01 3.38286906e-01 -4.39880073e-01 -5.28141081e-01
3.14717382e-01 3.48287106e-01 -5.29194534e-01 -4.28169668e-02
-1.24545753e+00 7.82095730e-01 6.27914071e-01 -7.30164573e-02
-5.06858863e-02 -3.95650655e-01 -9.17788506e-01 2.51638442e-01
8.81521642e-01 -3.10788631e-01 1.17996788e+00 -6.64106131e-01
-1.44091141e+00 5.30733228e-01 -5.71695745e-01 -4.35875654e-01
2.10530922e-01 -5.39002478e-01 -3.91126662e-01 2.50517696e-01
4.24661100e-01 2.24956438e-01 5.70770085e-01 -6.05144441e-01
-9.52311158e-01 -4.07479584e-01 2.16954872e-01 3.85030508e-01
-3.49734306e-01 3.77073526e-01 -3.64870012e-01 -6.56993806e-01
3.74878079e-01 -6.91965699e-01 -1.87138751e-01 -7.38981366e-01
-3.66761446e-01 -8.58157635e-01 7.42804885e-01 -7.33520508e-01
1.20990944e+00 -2.07899022e+00 -2.30487525e-01 -5.64225353e-02
-1.36483088e-02 4.81441706e-01 2.68624842e-01 5.48213601e-01
-4.90372622e-05 2.55276412e-01 -3.92962545e-01 9.03583840e-02
-2.30864793e-01 -2.37461887e-02 -3.06491673e-01 4.14164424e-01
2.38817662e-01 5.24116457e-01 -1.01354158e+00 -7.69748271e-01
1.15707412e-01 1.20958142e-01 -3.29061568e-01 -7.60298669e-02
-1.77719012e-01 3.22949231e-01 -5.83292127e-01 6.14863992e-01
3.94932419e-01 -2.04615623e-01 3.07252496e-01 7.64269531e-02
-2.44320288e-01 8.05831194e-01 -1.16500747e+00 1.38551772e+00
-3.79741490e-01 4.48396921e-01 -1.87367335e-01 -1.31309342e+00
8.67494762e-01 5.07480502e-01 7.05309391e-01 -8.00317705e-01
2.35321715e-01 2.72712231e-01 3.39919589e-02 -3.91898006e-01
7.01564074e-01 -1.03947572e-01 -3.65204930e-01 3.63715708e-01
-2.53814369e-01 1.50852263e-01 7.31755078e-01 2.40591809e-01
1.01190853e+00 -6.74855858e-02 9.47846472e-01 -3.15000415e-01
1.10733581e+00 2.50179440e-01 8.63924503e-01 3.86277199e-01
-2.03397237e-02 2.31654719e-01 5.63813984e-01 -1.52613133e-01
-6.62026882e-01 -4.70109880e-01 1.01188198e-01 8.73011053e-01
1.48946702e-01 -5.15843391e-01 -4.99725938e-01 -6.74944222e-01
-7.41985515e-02 8.33065271e-01 -2.51252562e-01 1.70388997e-01
-7.81186104e-01 -9.77975845e-01 3.91496003e-01 2.98686266e-01
7.90207326e-01 -9.85582948e-01 -6.69696093e-01 4.57050681e-01
-3.23706090e-01 -1.38723409e+00 -1.79892212e-01 3.62188667e-02
-9.66572642e-01 -1.22632277e+00 -3.53222549e-01 -7.16554821e-01
6.46291077e-01 3.90823990e-01 7.33743310e-01 -1.43196601e-02
-1.11909077e-01 -2.30565518e-01 -6.88095391e-01 -6.09536767e-01
-1.53180927e-01 3.14850211e-01 -6.62141154e-03 -5.94266467e-02
4.87701714e-01 -3.73814553e-01 -5.36618173e-01 -4.99369809e-03
-7.44610250e-01 4.72276062e-02 5.52347183e-01 6.81338012e-01
3.32863152e-01 4.32483226e-01 1.01765013e+00 -1.26989245e+00
7.71656275e-01 -3.12421143e-01 -5.11588931e-01 3.08148623e-01
-7.11635828e-01 2.02123404e-01 7.50423312e-01 -3.81562971e-02
-1.32139015e+00 -2.61842236e-02 -2.00276405e-01 6.53671265e-01
-2.21819393e-02 6.52162671e-01 -2.84644186e-01 4.07561451e-01
3.05137157e-01 3.71118575e-01 -8.87115970e-02 -6.61559641e-01
3.41583975e-02 7.36982048e-01 2.49679938e-01 -3.50738585e-01
7.77282596e-01 3.07002604e-01 8.48998949e-02 -1.08765340e+00
-1.18596876e+00 -8.38335454e-01 -4.30397391e-01 2.18608305e-01
4.99769151e-01 -7.69755900e-01 -8.35131347e-01 2.04548210e-01
-9.97536421e-01 5.63694179e-01 -8.06808621e-02 4.69797313e-01
1.39599005e-02 7.19316125e-01 -3.81489903e-01 -8.40457261e-01
-7.78458476e-01 -7.34203756e-01 8.04841042e-01 4.70672101e-01
-4.82209146e-01 -5.27669013e-01 -3.44942927e-01 5.86538315e-01
1.77588344e-01 1.79300398e-01 1.25965500e+00 -1.04169738e+00
-1.08207271e-01 -3.64427656e-01 -2.40402520e-01 3.59148741e-01
3.66317123e-01 -3.81743371e-01 -6.12553954e-01 -9.35580507e-02
1.96894959e-01 -7.43012726e-02 6.98987544e-01 1.03104360e-01
7.77246952e-01 -3.37810069e-01 -6.09616339e-02 -1.37273028e-01
1.38281393e+00 6.13382876e-01 3.24163914e-01 4.90169853e-01
4.92641181e-01 8.18984330e-01 1.10465705e+00 4.19470400e-01
2.53144085e-01 4.37885702e-01 3.94790582e-02 1.13368064e-01
1.47937194e-01 -1.20037690e-01 3.09045464e-01 7.67479897e-01
6.66178092e-02 -2.94383675e-01 -7.12936342e-01 6.63896263e-01
-1.96456909e+00 -9.01262462e-01 -3.24351549e-01 2.06423450e+00
9.74790573e-01 3.58750165e-01 3.19482684e-02 5.91538489e-01
6.96655214e-01 2.43429095e-01 -3.91373336e-01 -3.74741197e-01
-3.25333551e-02 3.10736656e-01 2.94580817e-01 3.40073705e-01
-1.12751567e+00 1.16748726e+00 5.07164097e+00 8.55249941e-01
-9.79459167e-01 1.40597358e-01 4.73952889e-01 1.13449521e-01
6.12241700e-02 5.29784411e-02 -9.18228388e-01 2.92916834e-01
5.72540164e-01 -4.09426004e-01 -7.10331425e-02 8.90645742e-01
3.83668154e-01 -8.08185756e-01 -6.24731243e-01 9.17504191e-01
7.29726776e-02 -1.09866405e+00 5.02849072e-02 -1.96825340e-01
3.84380311e-01 -4.83621269e-01 -4.82019246e-01 2.39192292e-01
-1.47488862e-01 -8.10568035e-01 4.83283043e-01 -2.09220454e-01
4.39644575e-01 -9.69703972e-01 1.08999407e+00 5.10099292e-01
-1.13717711e+00 -1.37176672e-02 -1.66624233e-01 -3.72237682e-01
1.56849355e-01 8.97833347e-01 -1.15272772e+00 6.88516438e-01
4.55091685e-01 4.61350054e-01 -3.19007933e-01 8.10751736e-01
-5.61036468e-01 9.47703838e-01 -2.57789195e-01 -4.01289791e-01
2.92803377e-01 -1.33775771e-01 5.43433428e-01 1.08154464e+00
1.12044148e-01 3.42240304e-01 2.56348193e-01 1.89588189e-01
-7.70493895e-02 7.09927917e-01 -2.65831649e-01 -2.93939263e-01
3.94782275e-01 1.10541606e+00 -1.19679737e+00 -3.75058472e-01
-2.84200221e-01 8.12032104e-01 4.50995192e-02 4.44991514e-03
-5.53418338e-01 -6.74584985e-01 3.94875139e-01 1.37631506e-01
2.35088587e-01 -2.63043910e-01 -3.37309837e-01 -1.01111495e+00
2.39435136e-01 -7.48396575e-01 4.78452623e-01 -2.03704327e-01
-7.53373086e-01 4.32469517e-01 -9.20461491e-03 -9.96407688e-01
-3.03784788e-01 -4.03249055e-01 -3.13804030e-01 7.44685769e-01
-1.57544863e+00 -8.05759311e-01 -2.88189333e-02 2.87985355e-01
1.01821291e+00 -1.59976512e-01 6.36985242e-01 2.81258643e-01
-6.43704474e-01 4.06979978e-01 -4.58149537e-02 5.54706976e-02
6.83435857e-01 -1.10636866e+00 1.36253193e-01 1.11012137e+00
-3.61623126e-04 7.50896275e-01 9.32754219e-01 -5.80464184e-01
-1.16653669e+00 -8.58844519e-01 1.67822921e+00 -6.62438199e-02
4.37912107e-01 -2.27788687e-01 -7.24721491e-01 9.59326103e-02
7.84457549e-02 -5.06873131e-01 7.59404600e-01 2.11394936e-01
-1.37954310e-03 -2.46625721e-01 -1.18078721e+00 4.34668988e-01
6.76930010e-01 -1.67393878e-01 -9.00710046e-01 2.44111255e-01
6.80861115e-01 -2.04465657e-01 -4.91140157e-01 3.76504093e-01
2.79439479e-01 -7.04717159e-01 5.38930237e-01 -3.81704152e-01
4.12055999e-01 -5.95043659e-01 4.44984101e-02 -8.72577548e-01
-1.75915316e-01 -5.46976566e-01 1.18137173e-01 1.47023380e+00
5.26748896e-01 -6.80090547e-01 6.33947968e-01 4.66545671e-01
-1.18792346e-02 -8.35087895e-01 -9.12254989e-01 -5.29151261e-01
-4.98845428e-01 -2.20729470e-01 4.35839236e-01 8.05300415e-01
8.87488574e-02 8.95761728e-01 -2.29208931e-01 -1.74709991e-01
2.95401037e-01 4.65797544e-01 5.37774861e-01 -1.44447136e+00
-4.23418805e-02 -1.43506005e-01 -1.57593474e-01 -8.26224327e-01
1.25758693e-01 -6.44268572e-01 2.81767435e-02 -1.84180224e+00
2.76729912e-01 -2.71816522e-01 -3.73566896e-01 6.13321185e-01
-4.97769594e-01 3.03945150e-02 8.42914507e-02 1.84501618e-01
-7.33017683e-01 4.46578056e-01 9.13528085e-01 1.33772595e-02
-3.35732460e-01 1.30095214e-01 -1.15704215e+00 8.82600665e-01
1.01905286e+00 -7.04905450e-01 -5.66774249e-01 -1.15084179e-01
3.65861833e-01 -1.34408444e-01 -2.55110353e-01 -6.95866168e-01
6.49164245e-02 -2.37421349e-01 1.01851352e-01 -6.05769575e-01
1.40347421e-01 -5.53856552e-01 -2.65718490e-01 4.47535813e-01
-2.78616529e-02 1.49833232e-01 8.73997733e-02 5.04185975e-01
-5.09834468e-01 -5.72976410e-01 5.75967908e-01 -2.10020602e-01
-8.85104895e-01 -2.24888220e-01 -4.05494332e-01 1.09749340e-01
1.03518116e+00 -1.94288969e-01 -3.06136310e-01 -1.82758003e-01
-2.35101685e-01 2.75324076e-01 1.90899149e-01 4.59110588e-01
3.98062885e-01 -7.12761462e-01 -6.63506091e-01 -2.31687367e-01
4.47216555e-02 1.68367237e-01 6.16698265e-02 8.29070926e-01
-5.17727137e-01 8.05113614e-01 1.26077473e-01 -7.42898583e-02
-1.59111273e+00 1.57214880e-01 -5.47632158e-01 -7.21673906e-01
-5.55046380e-01 5.99668026e-01 -7.28380084e-02 6.05034865e-02
5.62296100e-02 -7.79561698e-02 -7.75083363e-01 4.07629192e-01
4.06996906e-01 3.50219935e-01 1.69708505e-01 -5.52300274e-01
-6.42872095e-01 3.11238557e-01 -3.84578139e-01 -8.88180211e-02
1.35662031e+00 9.66283381e-02 -4.08134341e-01 3.89943987e-01
8.41653526e-01 3.38518173e-01 -4.18427557e-01 -2.84796327e-01
6.07018948e-01 -2.37280115e-01 -1.35651790e-02 -7.53970027e-01
-5.67689419e-01 4.79030430e-01 2.00023167e-02 2.50030518e-01
1.17543864e+00 -1.63861625e-02 1.06378722e+00 4.54495549e-01
4.43662852e-01 -1.35264409e+00 -9.47984233e-02 7.47385561e-01
6.30001843e-01 -1.30114305e+00 3.76530468e-01 -7.39338636e-01
-6.67524040e-01 9.12767649e-01 3.85064125e-01 7.95803517e-02
4.73539680e-01 -1.57301635e-01 -2.38570705e-01 -6.80434331e-02
-4.74703848e-01 -1.94848418e-01 1.87360242e-01 3.86947803e-02
8.13640714e-01 2.58889236e-02 -1.25368142e+00 5.98108947e-01
-3.25022221e-01 -1.99551076e-01 4.89442617e-01 1.23438275e+00
-6.57589495e-01 -1.39267445e+00 -1.34783268e-01 5.73595226e-01
-1.00025952e+00 -2.22902462e-01 -3.97055984e-01 7.47636020e-01
1.50961969e-02 1.33060932e+00 -2.71152079e-01 -1.90992057e-02
3.32689375e-01 2.42616072e-01 2.58084685e-01 -9.35918987e-01
-5.77711523e-01 -3.08866296e-02 5.96865118e-01 -1.95714012e-01
-6.95731759e-01 -8.10571790e-01 -1.59467733e+00 -3.17060715e-03
-3.66757661e-01 7.10532725e-01 6.88810945e-01 1.21637642e+00
3.68892550e-01 3.73277754e-01 5.36999404e-01 -1.52355298e-01
-4.03576642e-01 -1.14600539e+00 -4.21736568e-01 4.02290046e-01
6.32377118e-02 -6.45772815e-01 -5.89951128e-02 -7.78052658e-02]
|
[9.808002471923828, 8.76321792602539]
|
7ccdb33f-ed26-4c25-a09d-9db45a8b922e
|
optimizing-cad-models-with-latent-space
|
2303.12739
| null |
https://arxiv.org/abs/2303.12739v1
|
https://arxiv.org/pdf/2303.12739v1.pdf
|
Optimizing CAD Models with Latent Space Manipulation
|
When it comes to the optimization of CAD models in the automation domain, neural networks currently play only a minor role. Optimizing abstract features such as automation capability is challenging, since they can be very difficult to simulate, are too complex for rule-based systems, and also have little to no data available for machine-learning methods. On the other hand, image manipulation methods that can manipulate abstract features in images such as StyleCLIP have seen much success. They rely on the latent space of pretrained generative adversarial networks, and could therefore also make use of the vast amount of unlabeled CAD data. In this paper, we show that such an approach is also suitable for optimizing abstract automation-related features of CAD parts. We achieved this by extending StyleCLIP to work with CAD models in the form of voxel models, which includes using a 3D StyleGAN and a custom classifier. Finally, we demonstrate the ability of our system for the optimiziation of automation-related features by optimizing the grabability of various CAD models. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer review under the responsibility of the scientific committee of the 33rd CIRP Design Conference.
|
['Marco F Huber', 'Steffen Tauber', 'Raoul G. C. Schönhof', 'Jannes Elstner']
|
2023-03-09
| null | null | null | null |
['image-manipulation']
|
['computer-vision']
|
[ 1.77892864e-01 2.66498387e-01 -7.36520737e-02 -1.72685742e-01
-2.28245556e-01 -6.56768680e-01 5.61623394e-01 -2.12581068e-01
-1.68776348e-01 6.76375508e-01 -4.52644616e-01 -3.86616468e-01
-9.71582718e-03 -8.30972552e-01 -7.26564586e-01 -5.28365135e-01
1.35728404e-01 5.57755530e-01 -1.14210257e-02 -2.45292887e-01
2.30104998e-02 1.07355726e+00 -1.47776389e+00 -2.09859446e-01
7.34844744e-01 9.63936508e-01 1.00303754e-01 5.74596286e-01
1.97117068e-02 4.45277393e-01 -7.44632840e-01 -3.11614364e-01
5.41051865e-01 -3.70030880e-01 -3.89083505e-01 1.72902316e-01
2.39594430e-01 -1.92349076e-01 -2.33299091e-01 9.30118203e-01
1.68570191e-01 -1.21410094e-01 6.95805609e-01 -1.33319783e+00
-4.80224729e-01 2.24579453e-01 -7.30252936e-02 -4.04051304e-01
2.44972944e-01 3.26790631e-01 5.70344806e-01 -5.65725446e-01
6.39944911e-01 9.78359938e-01 1.58285931e-01 4.84380871e-01
-1.35026610e+00 -7.29296386e-01 1.19493201e-01 -1.68502837e-01
-1.31766856e+00 -1.65399790e-01 9.38170671e-01 -6.15430832e-01
5.31077743e-01 2.63409168e-01 7.97276199e-01 1.06957448e+00
3.92156005e-01 4.23843235e-01 1.13438332e+00 -3.96996617e-01
4.75932628e-01 3.58904839e-01 -5.07573783e-01 9.11655366e-01
1.95461407e-01 2.16761351e-01 1.66740686e-01 -2.86891329e-04
1.19982588e+00 1.43974628e-02 -2.14257926e-01 -8.47659469e-01
-9.77987289e-01 1.02394950e+00 4.98896539e-01 3.53799313e-01
-1.12177722e-01 3.26032937e-01 -1.97303742e-02 5.73531747e-01
2.54419833e-01 9.05521274e-01 -4.54240412e-01 7.79019073e-02
-8.65403235e-01 3.41981709e-01 1.09918261e+00 8.19337606e-01
7.91627884e-01 2.24872380e-01 4.04323816e-01 5.56098700e-01
2.60917813e-01 3.03765029e-01 3.15047979e-01 -1.05385423e+00
3.37252736e-01 5.66924155e-01 1.17518581e-01 -1.01521409e+00
-2.33965784e-01 -3.26661944e-01 -8.91000211e-01 7.42022753e-01
2.49369264e-01 -3.54025275e-01 -1.06186974e+00 1.43480599e+00
1.09079674e-01 -2.29401484e-01 -9.09296572e-02 6.51308596e-01
3.28954399e-01 5.50900936e-01 -2.63052702e-01 1.21337719e-01
1.00109637e+00 -5.89905262e-01 -4.57913846e-01 -7.19761774e-02
2.72889286e-01 -8.55331600e-01 1.06529224e+00 4.02384520e-01
-9.59895551e-01 -5.64319670e-01 -1.41433704e+00 2.95231700e-01
-5.61471283e-01 -5.02432920e-02 7.05850840e-01 6.68333650e-01
-8.65700245e-01 6.92318141e-01 -1.00808930e+00 -1.12592585e-01
5.05556107e-01 7.49166071e-01 -2.88084805e-01 5.58399633e-02
-9.46665287e-01 1.10783207e+00 2.47831404e-01 1.19618826e-01
-8.91666472e-01 -5.90080738e-01 -9.83951807e-01 -1.00978911e-01
4.81954664e-01 -6.31453514e-01 9.54264164e-01 -1.01631093e+00
-1.81118262e+00 5.96615434e-01 5.29491842e-01 -2.28828222e-01
7.76997864e-01 -1.06108025e-01 -1.97478756e-01 -9.59256850e-03
-1.06870085e-01 7.76601970e-01 1.18923569e+00 -1.06842279e+00
-1.16479196e-01 -1.25768363e-01 4.03337955e-01 -2.50323325e-01
-1.91430613e-01 -3.16610605e-01 -9.80931371e-02 -5.81715286e-01
-7.64686093e-02 -1.31565905e+00 -3.24636102e-01 1.74820125e-01
-4.83418345e-01 2.89141208e-01 1.12866724e+00 -3.44202250e-01
7.37028062e-01 -1.92611778e+00 3.55581313e-01 3.44540328e-01
-7.58554786e-02 2.07890600e-01 3.41737866e-02 2.75465757e-01
-1.79527104e-01 5.21887600e-01 -2.69168735e-01 6.76617771e-02
-8.53993371e-02 1.19496882e-01 3.67119350e-02 5.75841844e-01
4.37555552e-01 7.30956256e-01 -6.57845318e-01 -4.42229718e-01
6.09126270e-01 2.28597388e-01 -4.89403099e-01 1.74500898e-01
-4.04291362e-01 5.65087557e-01 -7.87987411e-01 5.10632634e-01
2.64820039e-01 -2.03101640e-03 1.79100648e-01 -6.41279966e-02
-1.42498255e-01 -3.20947528e-01 -1.30868053e+00 1.46006966e+00
-6.56755447e-01 5.90274155e-01 1.51140258e-01 -6.75940454e-01
8.94632816e-01 4.03801858e-01 3.61332059e-01 -1.23382859e-01
2.72939265e-01 3.49018663e-01 2.25772470e-01 -1.74832165e-01
2.50184476e-01 -2.83429623e-01 -2.91508168e-01 2.82167673e-01
7.40710273e-02 -1.26461112e+00 3.98943499e-02 -1.45381004e-01
1.02586007e+00 3.84921283e-01 4.06972349e-01 -1.81898717e-02
3.52245301e-01 3.65247056e-02 2.83499181e-01 5.03762543e-01
2.06598446e-01 6.48300409e-01 6.08729541e-01 -3.20291877e-01
-1.25977504e+00 -8.42945874e-01 -3.30608152e-02 9.90614668e-02
1.15494803e-02 -4.08119857e-02 -7.50176311e-01 -7.61588156e-01
1.04022846e-01 9.32207644e-01 -5.30373394e-01 -8.46964121e-02
-4.38037246e-01 -1.44602507e-01 5.32088399e-01 2.98305601e-01
2.75311738e-01 -7.00239301e-01 -7.72844911e-01 1.01836964e-01
4.00878787e-01 -9.86612022e-01 -1.29317641e-01 4.15482014e-01
-8.28722477e-01 -1.02627075e+00 -7.89285362e-01 -6.46438897e-01
8.14666867e-01 -2.89137870e-01 7.71500647e-01 2.19455939e-02
-3.24558586e-01 5.23190796e-01 -2.07019210e-01 -6.32064402e-01
-6.56975329e-01 1.57434970e-01 -1.03156693e-01 -2.63445348e-01
-1.52857602e-01 -6.95711672e-01 -4.05106604e-01 4.49262738e-01
-9.71987486e-01 6.25075325e-02 7.88192093e-01 9.18712795e-01
5.49594820e-01 3.11773419e-01 2.57753819e-01 -6.53615832e-01
5.84197879e-01 -4.73706722e-02 -1.19132841e+00 -6.68231919e-02
-6.29766703e-01 2.20321134e-01 7.35910356e-01 -5.14135301e-01
-7.90295720e-01 3.09394240e-01 5.28455041e-02 -8.43434930e-01
-3.89132589e-01 2.05247074e-01 -6.26766801e-01 -2.55492926e-01
5.06901920e-01 -3.02957922e-01 2.99155176e-01 -3.01159620e-01
4.62704599e-01 4.34186041e-01 2.60417819e-01 -5.53020537e-01
1.20797372e+00 2.59321094e-01 4.12684888e-01 -6.80252910e-01
-5.13083339e-01 3.25790882e-01 -5.44175684e-01 -1.53463140e-01
8.95955324e-01 -5.50847113e-01 -4.45598722e-01 4.49722618e-01
-8.37688386e-01 -4.19629782e-01 -7.03843236e-01 3.91169012e-01
-8.08725417e-01 1.97074816e-01 -3.05757850e-01 -4.45607364e-01
1.14770845e-01 -1.52181304e+00 7.22592771e-01 5.29555492e-02
-4.65229183e-01 -7.75502563e-01 -4.15205479e-01 1.28708377e-01
4.35595334e-01 6.43027365e-01 1.15410376e+00 -4.97323215e-01
-7.99443781e-01 -6.31595552e-01 2.88956434e-01 7.19667017e-01
1.56864151e-01 2.80350864e-01 -7.72265136e-01 -3.66722308e-02
2.18993425e-01 -2.22545341e-01 1.13452680e-01 3.93448442e-01
1.23306108e+00 -1.31501392e-01 -2.78711468e-01 4.82903570e-01
1.31044066e+00 4.11451697e-01 6.18240297e-01 3.83932918e-01
7.23413289e-01 5.96896827e-01 6.54710472e-01 7.24198595e-02
-2.67979175e-01 1.01349652e+00 6.17381573e-01 -8.90813917e-02
1.76706448e-01 -2.45925263e-01 1.88942865e-01 3.19541693e-01
-1.34629071e-01 -7.07933381e-02 -8.24367464e-01 3.64993930e-01
-1.37500548e+00 -6.00269079e-01 5.88164032e-02 2.03654790e+00
7.23869085e-01 2.66414315e-01 -2.50617564e-01 2.79592216e-01
6.58113122e-01 6.75097704e-02 -4.04147625e-01 -6.67430878e-01
2.75892049e-01 6.76478922e-01 5.13567746e-01 2.40844071e-01
-1.09996569e+00 7.96810210e-01 4.79737473e+00 6.07393324e-01
-1.34886086e+00 -6.86640814e-02 5.17091215e-01 -5.76599054e-02
-2.40397587e-01 1.37344360e-01 -2.13772371e-01 3.99617940e-01
7.70765483e-01 5.99800274e-02 4.98148799e-01 1.13552666e+00
1.55281186e-01 -5.79513535e-02 -1.21896458e+00 5.33816695e-01
-7.38582388e-02 -1.06274271e+00 -1.10534564e-01 3.14551592e-01
5.67888916e-01 -4.32412714e-01 3.20894241e-01 3.20592336e-02
3.11038345e-01 -1.25280857e+00 7.99076974e-01 3.46495032e-01
9.16255116e-01 -7.77106881e-01 7.00853229e-01 3.60582620e-01
-6.57215834e-01 5.48488311e-02 -1.97688520e-01 6.40460774e-02
9.60043967e-02 5.56417763e-01 -7.12514818e-01 4.18613762e-01
2.83177167e-01 3.10956031e-01 -3.32728088e-01 8.94542515e-01
-4.88599211e-01 3.95413071e-01 -3.71801704e-01 -9.80110765e-02
1.71877220e-01 -2.36829415e-01 4.61966336e-01 5.38886547e-01
4.66985464e-01 -1.86549067e-01 8.26826170e-02 1.02711904e+00
-1.44911513e-01 -1.31282628e-01 -7.94547200e-01 -2.60123312e-01
1.61185581e-02 1.08325279e+00 -8.65874469e-01 1.49562489e-02
-2.89701402e-01 8.82686734e-01 -1.99037731e-01 9.56450030e-02
-9.22855198e-01 -5.19195914e-01 3.73813987e-01 3.25719804e-01
5.12531757e-01 -1.96139589e-01 -2.69439638e-01 -1.13503122e+00
-4.38442640e-02 -8.24890971e-01 -1.63814843e-01 -8.19604039e-01
-8.26032996e-01 4.73746181e-01 3.10589671e-01 -1.36546052e+00
-5.27182341e-01 -7.58133352e-01 -2.41058931e-01 9.27616000e-01
-1.02761745e+00 -9.97753322e-01 -2.09899217e-01 4.10595119e-01
4.72140878e-01 -4.28000599e-01 9.78496253e-01 1.26850437e-02
-1.64816722e-01 2.81260699e-01 1.09344311e-01 5.72286174e-02
2.90293217e-01 -1.04445589e+00 2.69876629e-01 5.66390216e-01
6.32677749e-02 5.33227682e-01 8.37655485e-01 -4.05398130e-01
-1.38972473e+00 -9.78394449e-01 2.78659493e-01 -2.43433252e-01
6.60633922e-01 -3.86458397e-01 -5.09970129e-01 7.28431344e-01
1.25859363e-03 -2.78469101e-02 5.53745389e-01 2.29290817e-02
-5.39962947e-03 6.16993681e-02 -1.33075917e+00 7.18212128e-01
6.72657430e-01 -2.97889411e-01 -5.88263392e-01 1.70100600e-01
3.73492390e-01 -5.57527959e-01 -1.10021126e+00 4.11882371e-01
6.08557642e-01 -6.30660057e-01 8.51363480e-01 -3.27197790e-01
6.27558112e-01 -3.33642215e-01 -2.17402861e-01 -1.46059906e+00
-1.02473468e-01 -4.31205750e-01 5.37437685e-02 1.13524532e+00
4.14632320e-01 -8.47082436e-01 8.38126481e-01 5.60225606e-01
-2.74069875e-01 -9.02698934e-01 -7.35990703e-01 -9.10454333e-01
7.47840106e-02 -4.76515293e-01 8.28952193e-01 6.56933069e-01
-4.11371529e-01 1.11886188e-02 -1.47955254e-01 1.56410649e-01
2.26103008e-01 1.82921469e-01 8.70054066e-01 -1.22295308e+00
-4.58361804e-01 -5.13770044e-01 -8.92111242e-01 -4.42181855e-01
3.11542660e-01 -6.52287543e-01 -1.18864803e-02 -1.06863117e+00
-4.61484432e-01 -7.81001270e-01 -3.56842540e-02 3.53048533e-01
2.11857200e-01 -3.85533832e-02 4.78357494e-01 3.03581208e-01
-4.19147201e-02 3.63891035e-01 1.30131924e+00 -5.09199917e-01
-2.22030059e-01 2.31989846e-01 -5.85475504e-01 6.81581795e-01
9.70054090e-01 -5.74051559e-01 -4.33430225e-01 -1.65536389e-01
-1.66356295e-01 5.63411489e-02 3.69952321e-01 -1.25933814e+00
-3.17102522e-01 -2.90278494e-01 4.71813470e-01 -5.29316664e-02
6.60752714e-01 -1.25542426e+00 4.69130784e-01 5.61601460e-01
-8.01949427e-02 4.26499695e-02 2.10814625e-01 2.61437267e-01
-3.65822166e-01 -3.95728111e-01 8.79127204e-01 -3.73010248e-01
-4.41228896e-01 1.08082078e-01 -4.33205217e-01 -2.13517189e-01
1.19198501e+00 -3.23606491e-01 4.69053388e-02 -4.67260271e-01
-5.66307902e-01 2.72339620e-02 8.94992828e-01 2.46675864e-01
4.83955055e-01 -1.03019130e+00 -3.90036225e-01 2.68839240e-01
4.03329805e-02 1.02767639e-01 -9.87863094e-02 3.93466681e-01
-7.02854276e-01 3.38027328e-01 -2.64658630e-01 -5.08857489e-01
-9.56902623e-01 6.99899435e-01 4.67362493e-01 -2.24244371e-01
-5.50776541e-01 4.09950733e-01 2.38199428e-01 -3.48501116e-01
-7.96180442e-02 -3.22408080e-01 2.93506086e-01 -1.32580072e-01
-1.28178999e-01 1.45922303e-01 -1.70420762e-02 -4.55655128e-01
-1.33727998e-01 4.55462337e-01 7.67261395e-03 -1.99473277e-01
1.24311376e+00 4.11877930e-01 -4.80441190e-02 3.71953309e-01
1.05450916e+00 1.80529416e-01 -1.00932658e+00 6.39803588e-01
-3.40968072e-01 -4.31145370e-01 1.22810334e-01 -6.70041621e-01
-1.32555985e+00 8.68933201e-01 7.02745914e-01 6.44294173e-02
9.39032316e-01 -1.42719477e-01 4.86104012e-01 2.21026763e-01
6.09805763e-01 -8.76975834e-01 1.12414127e-02 2.72511363e-01
8.47372234e-01 -8.26098144e-01 1.13773145e-01 -4.93095726e-01
-5.49210429e-01 1.25036621e+00 3.17721277e-01 -2.88305968e-01
7.60767639e-01 4.83645737e-01 3.16121802e-02 -2.60712001e-02
-3.76855850e-01 4.27975915e-02 -7.30296131e-03 5.21353185e-01
2.01773956e-01 2.07467973e-01 -4.08676304e-02 2.39462018e-01
-2.81363428e-01 2.21385121e-01 5.63225865e-01 1.18037128e+00
5.77218495e-02 -1.38456440e+00 -4.80150908e-01 5.64582109e-01
-4.72902119e-01 2.06127509e-01 -2.35461861e-01 1.23979378e+00
2.35640809e-01 8.97991776e-01 -2.53691196e-01 -4.20695812e-01
3.71996045e-01 -1.84671767e-02 7.08497584e-01 -7.54079103e-01
-5.43847680e-01 -1.71435431e-01 1.85086709e-02 -2.65835822e-01
-1.03111453e-01 -6.96968913e-01 -8.28855217e-01 -1.12706184e-01
-3.76658320e-01 1.38226584e-01 9.24190283e-01 6.68352842e-01
2.89561480e-01 6.03739738e-01 5.25159061e-01 -1.02461672e+00
-5.60846269e-01 -6.80818617e-01 -6.64759099e-01 1.33486077e-01
5.12280203e-02 -8.17212701e-01 -3.71775806e-01 -3.55545171e-02]
|
[11.55858039855957, -0.531907856464386]
|
886f5509-1eca-4a09-8506-0ce5cd39e3e0
|
elo-system-for-skat-and-other-games-of-chance
|
2104.05422
| null |
https://arxiv.org/abs/2104.05422v1
|
https://arxiv.org/pdf/2104.05422v1.pdf
|
ELO System for Skat and Other Games of Chance
|
Assessing the skill level of players to predict the outcome and to rank the players in a longer series of games is of critical importance for tournament play. Besides weaknesses, like an observed continuous inflation, through a steadily increasing playing body, the ELO ranking system, named after its creator Arpad Elo, has proven to be a reliable method for calculating the relative skill levels of players in zero-sum games. The evaluation of player strength in trick-taking card games like Skat or Bridge, however, is not obvious. Firstly, these are incomplete information partially observable games with more than one player, where opponent strength should influence the scoring as it does in existing ELO systems. Secondly, they are game of both skill and chance, so that besides the playing strength the outcome of a game also depends on the deal. Last but not least, there are internationally established scoring systems, in which the players are used to be evaluated, and to which ELO should align. Based on a tournament scoring system, we propose a new ELO system for Skat to overcome these weaknesses.
|
['Stefan Edelkamp']
|
2021-04-07
| null | null | null | null |
['card-games']
|
['playing-games']
|
[-4.36845839e-01 1.11434951e-01 -8.46315697e-02 8.31263047e-03
-5.58600247e-01 -6.72999859e-01 3.59640986e-01 2.35427544e-01
-8.16610873e-01 8.07779074e-01 2.51155198e-01 -3.44534159e-01
-1.07426441e+00 -8.74108076e-01 -1.87646374e-01 -5.74810684e-01
-2.13988610e-02 7.67241299e-01 6.94999576e-01 -7.98375428e-01
3.63646865e-01 2.21728176e-01 -1.67044091e+00 1.24061108e-01
7.47038186e-01 8.93529534e-01 9.09291506e-02 8.06288838e-01
3.85666460e-01 1.39504349e+00 -8.06119680e-01 -9.75981295e-01
4.69346493e-01 -5.96172869e-01 -9.39036012e-01 -3.11366588e-01
-1.55040517e-01 -2.13484123e-01 -1.38104051e-01 7.74457633e-01
4.32809383e-01 7.51608834e-02 5.83921492e-01 -9.37634408e-01
7.35399276e-02 9.45535421e-01 -1.44516647e-01 3.04240495e-01
5.62539697e-01 1.00049198e-01 1.40134418e+00 -3.55533868e-01
3.00581247e-01 6.77182674e-01 6.67455375e-01 4.46914323e-02
-9.22418773e-01 -5.51449120e-01 -8.70370269e-02 5.00494599e-01
-1.14216959e+00 1.06278017e-01 6.06312335e-01 -6.87400818e-01
2.41919741e-01 4.44856137e-01 1.15005612e+00 6.57877982e-01
1.31093651e-01 2.48329058e-01 1.46224558e+00 -3.54768753e-01
1.99925303e-01 4.68398705e-02 -6.08710870e-02 1.70649290e-01
4.42467213e-01 3.10801536e-01 -8.66377175e-01 4.03329618e-02
6.64472640e-01 -5.36070049e-01 8.90815407e-02 -1.92225128e-01
-8.88353944e-01 8.63633752e-01 8.71966854e-02 2.62841165e-01
-4.22364146e-01 3.79262515e-03 3.59257519e-01 4.44685161e-01
1.99836284e-01 8.25595081e-01 -3.72627556e-01 -9.01298881e-01
-1.06647646e+00 6.67756379e-01 5.99407375e-01 -9.78998914e-02
4.23517138e-01 -2.13963732e-01 -4.80149947e-02 5.73621690e-01
6.42573237e-02 1.53323144e-01 4.12786663e-01 -1.05764163e+00
4.83573347e-01 8.80430043e-01 2.00114444e-01 -1.10199022e+00
-6.18112445e-01 -7.07136095e-01 -4.28543508e-01 1.01738429e+00
1.04998159e+00 -8.21512416e-02 -1.46799281e-01 1.68347645e+00
-1.73398823e-01 -1.90580472e-01 1.49855642e-02 9.58610952e-01
5.30754805e-01 1.26633141e-02 -1.80800989e-01 6.11123517e-02
1.22610486e+00 -3.44069719e-01 -4.59082514e-01 -2.63139427e-01
5.37814736e-01 -6.27802968e-01 8.90814483e-01 8.92866194e-01
-1.41949379e+00 -5.79380631e-01 -8.31723869e-01 4.14754301e-01
-1.50856256e-01 -7.33583719e-02 7.33100295e-01 8.66833210e-01
-8.38125169e-01 8.81737709e-01 -4.77210015e-01 1.78925693e-01
-5.89243397e-02 5.88501096e-01 -1.95539296e-01 6.43233299e-01
-1.64460278e+00 1.19215930e+00 7.08867550e-01 1.85838342e-01
-6.10649467e-01 -4.95670050e-01 -5.28389513e-01 3.57675493e-01
7.94453263e-01 -7.76927918e-02 9.44443583e-01 -6.31912589e-01
-1.54274714e+00 9.36299145e-01 5.74569106e-01 -4.63949025e-01
1.05112445e+00 -5.55684455e-02 -4.05332804e-01 -1.27658561e-01
2.97987789e-01 -1.44569308e-01 1.17364809e-01 -7.18967378e-01
-1.10232508e+00 -3.11872900e-01 7.54455686e-01 6.83390141e-01
-8.27323049e-02 3.12315017e-01 -1.48054406e-01 -1.56321749e-01
1.62206516e-01 -7.01895416e-01 -3.33370030e-01 -7.97233522e-01
-5.41145243e-02 -4.17636245e-01 1.36665869e-04 -5.47050774e-01
1.68484867e+00 -2.00720739e+00 1.42314598e-01 4.47672874e-01
4.12045300e-01 -5.05657494e-02 4.56610858e-01 4.28387463e-01
-9.12008882e-02 -8.37642178e-02 4.23282027e-01 2.55681008e-01
2.25680113e-01 -1.21276997e-01 9.25457925e-02 3.14227492e-01
-1.90001920e-01 7.31666744e-01 -8.19919467e-01 -3.98575217e-01
2.44433105e-01 -2.38493800e-01 -5.01634657e-01 -7.98546970e-02
4.72638726e-01 4.12788659e-01 -3.71736765e-01 2.04300031e-01
1.91089913e-01 4.83650640e-02 1.62176654e-01 5.75212538e-01
-5.05064428e-01 4.47831303e-01 -1.68901980e+00 1.06260455e+00
-1.18564991e-02 5.21459401e-01 -1.12128295e-01 -1.05556095e+00
1.02389359e+00 4.14131999e-01 5.79400182e-01 -7.75246501e-01
1.28811300e-01 5.09792268e-01 4.83599097e-01 -3.32933992e-01
9.06559348e-01 -2.29807779e-01 -6.46609783e-01 2.50510067e-01
-2.02828925e-02 -1.86892211e-01 7.66623318e-01 2.83525556e-01
1.03242981e+00 2.44947821e-01 4.01166260e-01 -2.99930751e-01
6.48683727e-01 1.95854396e-01 6.66596591e-01 9.61383224e-01
-1.53875694e-01 4.78609085e-01 1.14272022e+00 -3.56216162e-01
-9.49138045e-01 -9.51367021e-01 -9.26717073e-02 1.17094588e+00
2.64781505e-01 -4.85262126e-01 -5.47142029e-01 -1.53026730e-01
-2.32991785e-01 1.41307160e-01 -6.34450316e-01 -2.01843113e-01
-3.48490208e-01 -6.19853199e-01 5.11887372e-01 2.97506511e-01
2.96740234e-01 -1.12156534e+00 -7.48257577e-01 4.46849257e-01
-3.75799775e-01 -7.07795143e-01 1.05283502e-02 2.97256887e-01
-4.47839320e-01 -1.35554576e+00 -6.10057116e-01 -3.90083373e-01
6.10086285e-02 -1.76404655e-01 1.35556090e+00 2.09748283e-01
2.49707818e-01 1.94325238e-01 -4.82866257e-01 -6.59761190e-01
-4.17285979e-01 2.44719654e-01 3.10329735e-01 -6.09699711e-02
2.27377117e-01 -5.19623220e-01 -5.22878468e-01 6.19604707e-01
-5.60633361e-01 -5.19350618e-02 6.00683570e-01 6.90225363e-01
9.35304239e-02 4.93762791e-01 3.86448681e-01 -8.02068651e-01
7.87173808e-01 -2.25593463e-01 -6.51654303e-01 6.99880049e-02
-6.12237990e-01 -2.95986563e-01 5.06984890e-01 -2.22563878e-01
-8.60543847e-01 -2.55697489e-01 -2.85152972e-01 4.36283886e-01
-7.50479773e-02 9.20243144e-01 6.50003850e-02 1.16692528e-01
9.40835297e-01 -1.06651215e-02 -1.16472945e-01 -4.93761867e-01
-1.89346239e-01 5.85953057e-01 7.88627803e-01 -5.52631736e-01
9.74151611e-01 2.11894006e-01 1.84729025e-01 -2.72695571e-01
-7.40902126e-01 -5.13152122e-01 -7.73513734e-01 -9.32977080e-01
7.61573851e-01 -8.28509092e-01 -1.37345159e+00 6.20074213e-01
-6.69450879e-01 -2.83417374e-01 -5.55353343e-01 9.96296406e-01
-6.18630290e-01 2.16965586e-01 -2.34603539e-01 -9.57442403e-01
3.17927480e-01 -1.19098067e+00 3.66630852e-01 4.13612753e-01
-3.83292168e-01 -9.59305227e-01 2.81459242e-01 7.74213612e-01
1.94170594e-01 2.19241008e-01 3.38589609e-01 -4.96829808e-01
-2.07078487e-01 -5.75310588e-01 2.93635666e-01 1.75802127e-01
-2.31902838e-01 -2.81486809e-01 -4.60728884e-01 -1.37337372e-01
5.28247915e-02 -3.77499253e-01 4.31274146e-01 4.45553482e-01
5.76026976e-01 2.93388516e-01 2.45229691e-01 1.61452159e-01
1.24041247e+00 1.61576390e-01 7.25809574e-01 9.51667964e-01
1.89210176e-02 8.62785339e-01 8.11503112e-01 5.10209501e-01
2.03290164e-01 1.01382601e+00 6.19007051e-01 -8.37388188e-02
1.49075642e-01 -2.67077595e-01 5.46665609e-01 8.03364217e-01
-9.98898149e-01 1.64412588e-01 -6.19879544e-01 4.81876135e-01
-1.81512272e+00 -1.38252699e+00 -7.05233932e-01 2.53269720e+00
4.43574995e-01 9.06944811e-01 7.02133298e-01 7.63350725e-01
4.41619724e-01 4.75543812e-02 -6.13830388e-02 -4.86957699e-01
-2.80508220e-01 3.51765037e-01 6.79752409e-01 5.21066606e-01
-5.46156108e-01 5.49045444e-01 6.66203165e+00 1.19668806e+00
-7.05742776e-01 8.91688466e-02 3.73923987e-01 -2.84256041e-01
3.45753580e-02 3.41560543e-01 -6.73283398e-01 6.88960314e-01
7.02067435e-01 -6.19904935e-01 2.84874469e-01 6.15332007e-01
5.20976007e-01 -6.24611497e-01 -4.35152352e-01 6.21700823e-01
-3.02412480e-01 -9.55340743e-01 -5.91308594e-01 5.20511925e-01
4.81169760e-01 -4.07917440e-01 3.23020443e-02 4.86094385e-01
7.04773366e-01 -1.10162413e+00 1.32255590e+00 4.66273785e-01
6.09142601e-01 -8.66632760e-01 1.23564947e+00 4.17660147e-01
-1.19701099e+00 -3.05047691e-01 -3.59473556e-01 -1.02474618e+00
3.11390996e-01 4.44126815e-01 -1.54414609e-01 8.37161958e-01
8.30274403e-01 -5.77814272e-03 -4.99763340e-01 1.20727289e+00
-6.54907942e-01 8.69615436e-01 -1.94479406e-01 -5.23876883e-02
4.18757409e-01 -4.30910408e-01 5.34786284e-01 2.53694087e-01
3.04119945e-01 1.16759144e-01 2.89520696e-02 4.09939319e-01
5.45979321e-01 2.13901728e-01 -1.91148058e-01 3.30347568e-01
1.46992281e-01 1.08370602e+00 -7.06812620e-01 -1.03961423e-01
-4.62182820e-01 4.61497635e-01 7.98661336e-02 -2.03234524e-01
-8.07503164e-01 -2.11320847e-01 6.08501136e-01 5.16208827e-01
-1.25771686e-01 -7.69662708e-02 -3.98216873e-01 -7.16814339e-01
5.76908290e-02 -8.28116119e-01 5.22719026e-01 -5.92579305e-01
-8.66779506e-01 4.32860911e-01 1.38804734e-01 -1.58435464e+00
-2.50658333e-01 -6.83891773e-01 -6.77851260e-01 9.63917375e-01
-1.02157366e+00 -7.10578918e-01 -1.65554658e-01 3.77326071e-01
9.41143036e-02 -3.40899140e-01 3.63308400e-01 2.41983548e-01
-3.03200215e-01 2.82950431e-01 1.90408632e-01 2.31289744e-01
4.24833208e-01 -1.59924674e+00 -9.82304141e-02 6.14076257e-01
2.53771663e-01 8.54051039e-02 9.81835842e-01 -5.25550604e-01
-4.93551344e-01 2.16317060e-03 8.82829845e-01 -6.61119640e-01
9.80133653e-01 -2.13505074e-01 -4.01729941e-01 3.13784719e-01
-2.51638055e-01 -6.46378934e-01 7.20630586e-01 7.26118445e-01
4.13429946e-01 -8.68123397e-02 -5.05815208e-01 5.12264252e-01
9.40736294e-01 -3.65110576e-01 -7.79663801e-01 -2.46720091e-02
-1.69623286e-01 -7.53809154e-01 -9.57008660e-01 1.71431184e-01
9.27549601e-01 -1.60549939e+00 8.88332188e-01 -6.22219443e-01
6.31051362e-01 -2.37876713e-01 4.56112117e-01 -1.31556058e+00
-7.71687329e-01 -4.01267618e-01 6.88819051e-01 1.06193531e+00
5.73696434e-01 -4.86043066e-01 1.32470345e+00 6.18784070e-01
7.51876310e-02 -6.97689950e-01 -1.14938939e+00 -1.01951993e+00
2.05528617e-01 -7.49330163e-01 6.23124421e-01 5.61311781e-01
5.00876486e-01 2.63757855e-01 -9.00800228e-01 -9.62913111e-02
5.25344372e-01 -2.30307594e-01 9.05777276e-01 -1.86837268e+00
-8.10575962e-01 -7.50677466e-01 -8.73651683e-01 -6.57791913e-01
-4.68042791e-01 -7.56220162e-01 -3.76730561e-01 -1.29426932e+00
3.56431663e-01 -2.96851844e-01 -5.55140913e-01 1.04256853e-01
-4.01288062e-01 5.55151641e-01 4.42841589e-01 2.40128204e-01
-6.63967431e-01 -2.41346192e-02 1.15660572e+00 2.13811353e-01
-2.98737556e-01 4.84050095e-01 -8.05085897e-01 8.91455650e-01
7.13855386e-01 -6.12206459e-01 -2.40326636e-02 -4.11547236e-02
1.20546103e+00 2.95225918e-01 2.84436703e-01 -1.29447162e+00
1.83969155e-01 -3.98556620e-01 -2.22131121e-03 -3.41810614e-01
2.90616721e-01 -6.37077749e-01 5.92106044e-01 7.43398547e-01
-1.84792116e-01 1.40836125e-03 -2.52154171e-01 -1.71001945e-02
-5.57079315e-01 -8.16243112e-01 5.03774285e-01 -2.98874557e-01
-7.30400383e-01 1.50388815e-02 -6.87293589e-01 6.86259707e-03
1.02650964e+00 -6.87501550e-01 7.31890090e-03 -7.81253040e-01
-9.15102959e-01 3.50260794e-01 2.99333334e-01 5.54881692e-02
-2.22506262e-02 -1.29249001e+00 -1.14640880e+00 -3.85761976e-01
7.83044845e-02 -3.80337387e-01 6.37907982e-01 1.06316566e+00
-7.25008726e-01 3.73323232e-01 -4.89466012e-01 -4.59423773e-02
-1.47390842e+00 1.21999286e-01 3.00034434e-01 -1.11116266e+00
-2.77645499e-01 6.95061266e-01 5.95646240e-02 -2.04040721e-01
-1.40381828e-01 2.87450165e-01 -8.56910706e-01 3.95870715e-01
4.11897272e-01 6.87272251e-01 1.01444669e-01 -6.98979318e-01
-1.25748813e-01 3.47108930e-01 2.88505912e-01 -3.20056289e-01
1.42262495e+00 -7.72578269e-02 5.43631660e-03 5.73371708e-01
1.57856509e-01 6.41402245e-01 -9.91150200e-01 3.78921330e-02
2.97734737e-02 -5.13180494e-01 -1.10702775e-01 -7.67674029e-01
-8.21806431e-01 5.16225517e-01 2.47709304e-01 8.15127134e-01
9.13097858e-01 -1.21991195e-01 1.80055231e-01 -1.81030661e-01
6.16508067e-01 -1.49637878e+00 9.79138464e-02 5.30867577e-01
5.03725827e-01 -9.78915572e-01 8.98992866e-02 -9.87517610e-02
-7.52419055e-01 9.75509346e-01 4.84423757e-01 -3.07669863e-03
2.71995932e-01 1.73626140e-01 1.40623942e-01 -3.90087396e-01
-4.47598755e-01 -5.80916941e-01 2.36047864e-01 3.94058853e-01
2.49446213e-01 2.26095691e-01 -1.30343056e+00 9.92397487e-01
-9.48729932e-01 -1.72128789e-02 8.29694927e-01 3.15214127e-01
-4.81414706e-01 -1.39924359e+00 -8.32803071e-01 6.18763983e-01
-7.98769355e-01 1.91638529e-01 -5.12422025e-01 8.74829471e-01
5.58664560e-01 1.01434994e+00 -1.36583924e-01 -5.43065488e-01
6.11319244e-01 -2.50723034e-01 1.69571370e-01 -4.28587556e-01
-9.43836272e-01 -4.15043116e-01 2.88739860e-01 -3.73776525e-01
-3.09692234e-01 -7.77003229e-01 -6.40488625e-01 -5.64479530e-01
-5.65151155e-01 7.86549509e-01 2.91796744e-01 1.09873295e+00
-5.07414401e-01 5.42571425e-01 7.02162385e-01 -7.01202750e-01
-5.45454741e-01 -1.01935554e+00 -1.42083097e+00 5.14412820e-01
-4.45561349e-01 -1.00957143e+00 -5.83022535e-01 -6.91650391e-01]
|
[6.459214687347412, 0.4212234318256378]
|
ce05f80a-690e-4820-965d-603b8fa68eb7
|
cospgd-a-unified-white-box-adversarial-attack
|
2302.02213
| null |
https://arxiv.org/abs/2302.02213v2
|
https://arxiv.org/pdf/2302.02213v2.pdf
|
CosPGD: a unified white-box adversarial attack for pixel-wise prediction tasks
|
While neural networks allow highly accurate predictions in many tasks, their lack of robustness towards even slight input perturbations hampers their deployment in many real-world applications. Recent research towards evaluating the robustness of neural networks such as the seminal projected gradient descent(PGD) attack and subsequent works have drawn significant attention, as they provide an effective insight into the quality of representations learned by the network. However, these methods predominantly focus on image classification tasks, while only a few approaches specifically address the analysis of pixel-wise prediction tasks such as semantic segmentation, optical flow, disparity estimation, and others, respectively. Thus, there is a lack of a unified adversarial robustness benchmarking tool(algorithm) that is applicable to all such pixel-wise prediction tasks. In this work, we close this gap and propose CosPGD, a novel white-box adversarial attack that allows optimizing dedicated attacks for any pixel-wise prediction task in a unified setting. It leverages the cosine similarity between the distributions over the predictions and ground truth (or target) to extend directly from classification tasks to regression settings. We outperform the SotA on semantic segmentation attacks in our experiments on PASCAL VOC2012 and CityScapes. Further, we set a new benchmark for adversarial attacks on optical flow, and image restoration displaying the ability to extend to any pixel-wise prediction task.
|
['Margret Keuper', 'Steffen Jung', 'Shashank Agnihotri']
|
2023-02-04
| null | null | null | null |
['disparity-estimation']
|
['computer-vision']
|
[ 5.02848268e-01 2.06448473e-02 1.12690963e-03 -1.84048221e-01
-8.09306264e-01 -8.89097095e-01 6.35634601e-01 -3.77530813e-01
-5.50249398e-01 6.93795264e-01 -5.82557358e-02 -5.63522279e-01
2.05087513e-02 -7.28059709e-01 -1.00419343e+00 -8.66082132e-01
-8.91703889e-02 -1.54983804e-01 3.95864904e-01 -3.61316174e-01
2.36221194e-01 7.03665733e-01 -1.11062777e+00 -7.73009704e-03
7.19591975e-01 1.02228200e+00 -4.81006294e-01 8.48973095e-01
5.30149281e-01 1.02719140e+00 -8.25531721e-01 -9.53810811e-01
7.82296240e-01 -1.98156551e-01 -7.29081631e-01 -3.67813408e-01
1.01310205e+00 -3.65637153e-01 -9.00148451e-01 1.36770892e+00
8.40946972e-01 1.57252908e-01 4.53615904e-01 -1.64274716e+00
-5.98484516e-01 3.64560276e-01 -2.72551686e-01 3.29376012e-01
-1.94504056e-02 7.05277681e-01 8.84474635e-01 -2.87684053e-01
4.44834590e-01 1.27933037e+00 8.32712531e-01 1.01803577e+00
-1.03205669e+00 -7.57663488e-01 1.36770085e-01 3.28887880e-01
-7.93556690e-01 -3.17445189e-01 7.53465712e-01 -4.17599201e-01
7.67435431e-01 3.04591298e-01 1.13379858e-01 1.72374797e+00
1.64343968e-01 7.97188282e-01 1.14581275e+00 3.71262357e-02
1.31889924e-01 -8.56713727e-02 8.21411796e-03 4.65837717e-01
1.66523755e-01 4.86107975e-01 -3.25634837e-01 4.29698043e-02
5.46295047e-01 -3.86209726e-01 -6.44043624e-01 -4.25490499e-01
-8.98151159e-01 9.10423338e-01 5.00496387e-01 -2.36557633e-01
4.15934548e-02 4.04524714e-01 7.30235755e-01 3.25875849e-01
4.79769051e-01 5.71799338e-01 -5.49609661e-01 -3.41462418e-02
-7.08837211e-01 4.36015069e-01 7.85654187e-01 5.52334964e-01
5.97204924e-01 4.30612713e-01 -2.71096945e-01 4.06570196e-01
3.63553688e-02 4.76787060e-01 3.21844518e-01 -1.27203596e+00
7.56775260e-01 1.03755251e-01 -4.92715389e-02 -1.29619527e+00
-2.38707036e-01 -3.16586733e-01 -9.36669946e-01 6.79881632e-01
7.57551134e-01 -5.12512386e-01 -8.14577520e-01 2.14130330e+00
2.18394578e-01 5.96695960e-01 3.42928350e-01 1.13830340e+00
7.10473657e-01 3.33647430e-01 -3.06106023e-02 2.08305940e-01
8.60986829e-01 -1.14258289e+00 -2.54751444e-01 -3.42595130e-01
4.82101113e-01 -6.46291554e-01 9.33699191e-01 4.02121603e-01
-1.01034582e+00 -6.40422046e-01 -1.23461366e+00 -7.98090696e-02
-4.21263367e-01 -4.19789165e-01 5.72466016e-01 1.06378186e+00
-9.42629635e-01 1.01203132e+00 -8.55195642e-01 -1.99684221e-02
7.58054614e-01 3.09983641e-01 -4.59129602e-01 2.66269594e-02
-1.41561615e+00 9.77876425e-01 1.36510983e-01 1.62229627e-01
-1.04912472e+00 -1.00908113e+00 -1.00810444e+00 -2.83111036e-01
1.87483594e-01 -6.83764338e-01 8.79745364e-01 -1.01043046e+00
-1.69543362e+00 8.44465375e-01 3.33579957e-01 -1.12231874e+00
1.09052575e+00 -3.83953482e-01 -2.80913562e-01 1.59719631e-01
-1.17157377e-01 8.58534276e-01 1.04038346e+00 -1.05706429e+00
-3.58350366e-01 -1.84009671e-01 3.79151672e-01 2.84966473e-02
-3.14389259e-01 -2.80475095e-02 -2.44914711e-01 -8.97136211e-01
-4.20598954e-01 -1.02323318e+00 -3.66565108e-01 2.05515131e-01
-7.79446423e-01 1.77012920e-01 1.01825309e+00 -5.47763169e-01
8.88365686e-01 -2.33287668e+00 1.57537255e-02 -5.22568710e-02
-3.61991152e-02 7.33279705e-01 -3.71238440e-01 9.68211293e-02
-4.05467868e-01 2.19724417e-01 -6.09715581e-01 -4.69832093e-01
1.67586744e-01 3.03364128e-01 -1.02747118e+00 7.70546913e-01
2.94736177e-01 9.92109776e-01 -7.17823148e-01 -1.91840410e-01
3.64282638e-01 6.40055895e-01 -6.35288596e-01 1.67697504e-01
-8.51892829e-02 5.88766158e-01 -1.56724036e-01 2.57096827e-01
8.44763339e-01 1.99461788e-01 -5.02741039e-01 -1.65671468e-01
2.63525814e-01 1.62049189e-01 -1.00172448e+00 1.54117537e+00
-3.31336945e-01 1.00475490e+00 -1.03394091e-01 -1.23969793e+00
6.87394738e-01 2.34277979e-01 4.85593289e-01 -6.55810475e-01
-2.04862617e-02 -2.57347757e-03 -3.51693965e-02 -3.05482656e-01
3.15986216e-01 1.06272176e-01 -2.00043619e-02 2.32835695e-01
-2.57512219e-02 -2.95138478e-01 4.46819142e-02 1.76527441e-01
1.22665775e+00 2.79765457e-01 -1.72494724e-01 6.25465810e-02
6.05018079e-01 -6.74730092e-02 6.93340480e-01 9.63323593e-01
-7.46540248e-01 1.10685277e+00 7.15004325e-01 -5.61342239e-01
-9.33358967e-01 -1.16975296e+00 -2.30514273e-01 7.32613683e-01
2.91192442e-01 -7.58747384e-02 -1.05228055e+00 -1.19055986e+00
7.07466155e-02 6.23980463e-01 -7.85194218e-01 -3.92725199e-01
-6.77729964e-01 -9.31129217e-01 1.26580691e+00 3.79566044e-01
8.91527534e-01 -1.08568347e+00 -5.04337907e-01 -1.25858277e-01
-7.44222403e-02 -1.63096595e+00 -4.09333736e-01 -1.29173636e-01
-6.20928645e-01 -1.42875504e+00 -5.97720087e-01 -5.13764322e-01
3.06953013e-01 -5.79603985e-02 1.08812714e+00 -1.76532343e-01
-2.88554341e-01 5.25270462e-01 -1.02085546e-01 -4.05442536e-01
-5.21018386e-01 -2.49413133e-01 -6.97238231e-03 1.36061490e-01
4.69912402e-03 -4.93801564e-01 -8.81481886e-01 6.23700559e-01
-1.01758051e+00 -3.30401152e-01 1.87286064e-01 7.51313627e-01
3.74238849e-01 2.74883788e-02 3.44936460e-01 -9.07838941e-01
4.64674413e-01 -3.62180769e-01 -6.94894373e-01 -2.80666295e-02
-3.12109202e-01 -1.17883071e-01 9.83365536e-01 -3.49890202e-01
-7.31944084e-01 -1.30791083e-01 -4.75697637e-01 -7.56354272e-01
-3.31844568e-01 -3.31022553e-02 -1.40971974e-01 -4.71052468e-01
9.82136548e-01 1.36404827e-01 -1.55131100e-02 -8.23825002e-02
4.30627286e-01 1.79306522e-01 9.78227556e-01 -4.73190695e-01
1.24934769e+00 6.13514423e-01 4.19177920e-01 -6.21782124e-01
-8.26937079e-01 -5.32686077e-02 -2.90550709e-01 9.95323882e-02
9.28075254e-01 -8.48827004e-01 -8.35461140e-01 8.56613994e-01
-1.08185720e+00 -5.61481714e-01 -3.83294284e-01 3.09546679e-01
-7.57528663e-01 7.73277819e-01 -6.53066456e-01 -4.03637141e-01
-3.90347868e-01 -1.52067363e+00 7.49483585e-01 1.22361831e-01
7.20929801e-02 -1.24495685e+00 9.13646370e-02 5.08183956e-01
3.54524314e-01 6.74964368e-01 5.73701680e-01 -7.26308048e-01
-5.39165199e-01 -2.13913813e-01 -2.89682299e-01 1.02585375e+00
-1.95334405e-01 1.50112718e-01 -1.25086308e+00 -4.94501024e-01
1.40012115e-01 -5.26874363e-01 1.13706768e+00 3.90588582e-01
1.40538704e+00 -3.75767827e-01 1.61649078e-01 1.19427776e+00
1.31012297e+00 -2.55711526e-01 9.73940492e-01 5.41538596e-01
8.24167073e-01 5.53433120e-01 3.83626968e-01 3.61816064e-02
-9.61898081e-03 6.54637933e-01 1.07355702e+00 -1.03334665e-01
-3.13950330e-01 -1.33795008e-01 4.50745910e-01 5.02647385e-02
7.22512603e-02 -3.87796193e-01 -6.76060736e-01 2.58420646e-01
-1.70361173e+00 -1.02869391e+00 -9.90363397e-03 2.19905448e+00
6.92979872e-01 3.47450197e-01 -1.41446099e-01 2.36785546e-01
6.92915142e-01 5.45999408e-01 -8.68741512e-01 -5.42519808e-01
-3.64328802e-01 3.49672586e-01 8.17076623e-01 3.92092377e-01
-1.64592826e+00 1.17824340e+00 6.22368670e+00 9.22896147e-01
-1.24777544e+00 -6.61171079e-02 9.38299656e-01 9.25961584e-02
3.49258631e-02 -1.58219919e-01 -5.77593327e-01 5.18649042e-01
7.01653183e-01 2.48315543e-01 4.43460196e-01 7.70537317e-01
1.15650333e-01 2.51301020e-01 -1.02881014e+00 9.07369673e-01
5.56148216e-02 -1.23388398e+00 8.27917084e-02 -9.09819007e-02
9.81714010e-01 1.64910257e-01 6.88773215e-01 1.78589255e-01
5.29190540e-01 -1.26060891e+00 3.37451816e-01 2.55232066e-01
6.05550647e-01 -7.70351827e-01 5.90276003e-01 1.25184447e-01
-5.94216406e-01 -8.77640247e-02 -3.82081479e-01 9.19657424e-02
-3.93067412e-02 3.90058845e-01 -5.32991707e-01 4.30033892e-01
7.61175036e-01 8.63725066e-01 -5.30254245e-01 9.46045697e-01
-6.07086718e-01 8.23863328e-01 -4.33596969e-02 4.74669725e-01
4.38687652e-01 -5.35913417e-03 7.87963212e-01 1.10464907e+00
-6.42574281e-02 -3.74956340e-01 -1.48991525e-01 7.46803224e-01
-2.36176208e-01 -1.49175406e-01 -6.25297010e-01 3.62918407e-01
2.00341031e-01 1.06715465e+00 -3.77109349e-01 1.11964718e-01
-4.51841503e-01 1.08596301e+00 2.19565049e-01 6.05400980e-01
-1.19034505e+00 -4.40794766e-01 1.43402052e+00 -4.15239841e-01
2.46467710e-01 -1.57554999e-01 -5.21616101e-01 -1.09798861e+00
6.26491234e-02 -1.16240942e+00 3.34514320e-01 -5.82433581e-01
-1.47130966e+00 4.57977980e-01 -3.27133775e-01 -1.28118646e+00
-2.59292722e-01 -8.94516170e-01 -8.65468144e-01 8.55280817e-01
-1.76816714e+00 -8.35975349e-01 -1.84868559e-01 9.84364986e-01
3.26440275e-01 -3.29794407e-01 6.88198388e-01 2.22736433e-01
-1.00947952e+00 1.03548431e+00 -2.10377220e-02 6.40741348e-01
9.18967843e-01 -1.28306806e+00 8.55858445e-01 1.45159543e+00
2.61019140e-01 2.57330209e-01 7.92012572e-01 -2.05104768e-01
-1.07004297e+00 -1.37446928e+00 1.87675357e-01 -6.61342740e-01
8.30223918e-01 -1.41508147e-01 -7.19151437e-01 6.33336067e-01
1.45797208e-01 5.43384194e-01 2.64051437e-01 -5.57012439e-01
-6.27886713e-01 -7.24790618e-02 -1.25431252e+00 9.36044693e-01
1.07168043e+00 -6.24630749e-01 -1.69894889e-01 3.83436233e-01
8.82958174e-01 -7.22742081e-01 -6.86467230e-01 6.32187724e-01
2.29840830e-01 -1.38754797e+00 1.50048757e+00 -8.71912837e-01
7.71993399e-01 -1.49709880e-01 -2.03931496e-01 -1.32383871e+00
1.45928860e-01 -9.76035237e-01 -7.83691630e-02 1.01298916e+00
2.66663164e-01 -9.86998975e-01 1.05060089e+00 6.35214269e-01
-2.57380426e-01 -7.36471593e-01 -1.08971179e+00 -7.35034347e-01
4.99702722e-01 -8.06799531e-01 3.59603435e-01 9.29247618e-01
-6.67582154e-01 -3.36158544e-01 -5.26944220e-01 5.88448167e-01
7.99852014e-01 -1.50904462e-01 1.02744448e+00 -5.71287096e-01
-4.77363884e-01 -6.57637358e-01 -9.94454801e-01 -9.47837830e-01
5.15805066e-01 -7.94903457e-01 -1.14118002e-01 -9.72884178e-01
-3.35405916e-01 -3.11539561e-01 -3.69299799e-01 4.75396931e-01
-3.94742519e-01 6.93609118e-01 3.01046968e-01 8.30554068e-02
-2.66052574e-01 3.26595068e-01 1.32046998e+00 -4.08353418e-01
1.38462827e-01 3.52992505e-01 -6.39466882e-01 7.41490781e-01
8.79960477e-01 -4.45583224e-01 -4.62959677e-01 -5.73709249e-01
-4.06193398e-02 -2.98963428e-01 8.20680797e-01 -1.32795513e+00
1.64103180e-01 2.72258315e-02 2.25522533e-01 9.87493768e-02
3.11269552e-01 -6.33569062e-01 -3.23392421e-01 5.39353848e-01
-4.74056453e-01 -1.80614963e-01 3.00578952e-01 5.54200113e-01
-2.78413713e-01 -8.69437829e-02 1.06931651e+00 7.99320564e-02
-8.22620571e-01 6.73513591e-01 1.31624550e-01 6.72560811e-01
1.10005414e+00 -3.13772321e-01 -6.30372465e-01 -4.08650756e-01
-6.35632932e-01 4.94231246e-02 3.94171566e-01 5.19431353e-01
5.86016119e-01 -9.52721834e-01 -8.29796135e-01 3.03564042e-01
-2.14085340e-01 6.43255785e-02 2.25395009e-01 5.45726299e-01
-8.24432373e-01 2.33653530e-01 -3.63660425e-01 -5.47664523e-01
-1.06619072e+00 7.02511966e-01 7.22355485e-01 -1.67330980e-01
-5.96855760e-01 1.04405999e+00 3.82980138e-01 -4.81870592e-01
5.19721627e-01 8.22588056e-02 -1.67142943e-01 -3.72845441e-01
3.44263017e-01 4.63060081e-01 3.50269005e-02 -7.71945655e-01
-1.49729431e-01 6.65648043e-01 1.99330717e-01 4.75929640e-02
1.09336364e+00 9.43221375e-02 1.63127482e-01 -1.35216013e-01
1.52824867e+00 -7.71079063e-02 -1.94871020e+00 -3.45986485e-02
-4.75596368e-01 -5.78088999e-01 -7.97341317e-02 -7.05639064e-01
-1.61992466e+00 1.11871970e+00 6.34418905e-01 1.37295067e-01
1.15590966e+00 -4.98277962e-01 9.87313926e-01 3.49268347e-01
7.48564601e-02 -7.51045823e-01 1.17528535e-01 5.98793685e-01
7.13835001e-01 -1.23720121e+00 -2.24041790e-01 -4.55482125e-01
-6.39735401e-01 9.85689282e-01 6.10041857e-01 -4.72459346e-01
5.41164696e-01 2.00485319e-01 3.66882831e-01 4.28708464e-01
-3.98503482e-01 1.93593244e-03 2.60725856e-01 9.46622193e-01
-5.93692809e-03 -2.83229798e-01 1.56561688e-01 1.95924506e-01
-4.53323990e-01 -3.87613654e-01 5.37776709e-01 5.45083821e-01
7.81100914e-02 -9.90456462e-01 -3.06408256e-01 2.53173765e-02
-9.12865222e-01 -1.33169428e-01 -2.20318258e-01 7.96178818e-01
8.05170089e-02 1.03058827e+00 -4.68243249e-02 -4.27789658e-01
3.78196448e-01 -2.14702383e-01 3.82316709e-01 -1.24840908e-01
-7.04403281e-01 -7.30112076e-01 -9.23943818e-02 -1.10516179e+00
-2.15244144e-01 -5.02811134e-01 -7.81475782e-01 -4.82771754e-01
5.74547499e-02 -1.68168545e-01 6.48581445e-01 8.74344409e-01
1.53044358e-01 4.28329080e-01 8.80258143e-01 -8.89799893e-01
-7.39136875e-01 -4.61223125e-01 -2.05407634e-01 8.45461965e-01
4.87428784e-01 -3.61028731e-01 -7.39922762e-01 -2.34736264e-01]
|
[5.505273342132568, 7.942113876342773]
|
602c963a-a36f-4e15-b7b2-b3b36b517cf0
|
abolitionist-networks-modeling-language
|
2103.07538
| null |
https://arxiv.org/abs/2103.07538v1
|
https://arxiv.org/pdf/2103.07538v1.pdf
|
Abolitionist Networks: Modeling Language Change in Nineteenth-Century Activist Newspapers
|
The abolitionist movement of the nineteenth-century United States remains among the most significant social and political movements in US history. Abolitionist newspapers played a crucial role in spreading information and shaping public opinion around a range of issues relating to the abolition of slavery. These newspapers also serve as a primary source of information about the movement for scholars today, resulting in powerful new accounts of the movement and its leaders. This paper supplements recent qualitative work on the role of women in abolition's vanguard, as well as the role of the Black press, with a quantitative text modeling approach. Using diachronic word embeddings, we identify which newspapers tended to lead lexical semantic innovations -- the introduction of new usages of specific words -- and which newspapers tended to follow. We then aggregate the evidence across hundreds of changes into a weighted network with the newspapers as nodes; directed edge weights represent the frequency with which each newspaper led the other in the adoption of a lexical semantic change. Analysis of this network reveals pathways of lexical semantic influence, distinguishing leaders from followers, as well as others who stood apart from the semantic changes that swept through this period. More specifically, we find that two newspapers edited by women -- THE PROVINCIAL FREEMAN and THE LILY -- led a large number of semantic changes in our corpus, lending additional credence to the argument that a multiracial coalition of women led the abolitionist movement in terms of both thought and action. It also contributes additional complexity to the scholarship that has sought to tease apart the relation of the abolitionist movement to the women's suffrage movement, and the vexed racial politics that characterized their relation.
|
['Jacob Eisenstein', 'Lauren Klein', 'Sandeep Soni']
|
2021-03-12
| null | null | null | null |
['diachronic-word-embeddings']
|
['natural-language-processing']
|
[-2.16699857e-02 3.53364795e-01 -6.08439624e-01 -1.11923717e-01
-1.34417608e-01 -9.30877864e-01 1.43619835e+00 5.88291228e-01
-9.14415777e-01 5.62043190e-01 1.61089337e+00 -1.07906008e+00
-3.11451167e-01 -9.93709445e-01 -3.76105189e-01 -5.44765592e-01
2.59765267e-01 5.55767477e-01 -1.69285730e-01 -8.68266821e-01
6.84810162e-01 1.88224241e-01 -6.66743457e-01 -3.21030885e-01
6.47787511e-01 -2.93352455e-02 -2.87309170e-01 9.42583680e-02
-4.06679690e-01 9.48306561e-01 -6.18640900e-01 -4.08066362e-01
1.43406451e-01 -6.77663743e-01 -6.80410922e-01 -3.01746637e-01
3.40406269e-01 2.15064399e-02 -7.79203832e-01 9.70265388e-01
3.60439628e-01 1.87724363e-02 5.36486089e-01 -4.14012492e-01
-1.00503922e+00 1.37143064e+00 -6.66071832e-01 8.00406396e-01
1.05542570e-01 -2.02337712e-01 1.31980550e+00 -5.93107581e-01
1.54743707e+00 1.66678953e+00 5.16453207e-01 -4.26791944e-02
-1.35692811e+00 -9.37019408e-01 1.76938817e-01 -5.34133352e-02
-1.03764999e+00 -6.85273230e-01 7.14149714e-01 -8.46706688e-01
4.90906805e-01 3.41250181e-01 1.20634186e+00 1.04637480e+00
4.10456270e-01 -1.33227631e-01 1.18223906e+00 -4.23304439e-01
-2.44667068e-01 -9.49426889e-02 -4.43896884e-03 4.99530286e-01
8.50948751e-01 -2.53821313e-02 -3.24548423e-01 -7.75173426e-01
4.36001003e-01 2.53403131e-02 -1.48058429e-01 2.85672009e-01
-1.21222103e+00 1.38670528e+00 4.78032678e-01 1.15443563e+00
-1.46444306e-01 8.26440528e-02 4.23700184e-01 4.45713878e-01
8.30477476e-01 5.31831086e-01 -1.35161668e-01 -2.65265018e-01
-6.45388305e-01 1.22241341e-01 8.49347055e-01 -2.04368740e-01
3.60703856e-01 -2.71720529e-01 2.69820422e-01 5.50069749e-01
5.25144517e-01 6.51036918e-01 2.34049596e-02 -6.97563350e-01
4.69390333e-01 6.21140122e-01 -1.66456476e-01 -1.71922290e+00
-2.24987656e-01 -3.35272580e-01 -4.79597807e-01 1.12346485e-01
6.20880187e-01 -1.67831510e-01 -6.12566650e-01 1.86219752e+00
3.09112936e-01 -6.60236061e-01 -3.70661080e-01 7.74261475e-01
3.60234052e-01 8.44214916e-01 2.28517160e-01 -2.62759119e-01
1.39757550e+00 -4.92913537e-02 -6.61772907e-01 -1.34155229e-01
3.45569402e-01 -1.06533015e+00 6.85696423e-01 -3.61061543e-01
-8.64777923e-01 -8.77179671e-03 -8.15063715e-01 -1.82831511e-01
-3.20003837e-01 -8.69340241e-01 3.98582667e-01 3.47741723e-01
-8.25469792e-01 7.19559789e-01 -4.90775853e-01 -1.04126513e+00
7.55243242e-01 -3.21526021e-01 -1.04965888e-01 1.22134514e-01
-1.16887045e+00 1.08291745e+00 -2.50039585e-02 -2.44129989e-02
-4.32995148e-02 -4.55465049e-01 -7.38129973e-01 -2.44818479e-01
5.31919479e-01 -3.68690908e-01 7.03128934e-01 -1.22058368e+00
-7.25415587e-01 1.13774776e+00 4.68436405e-02 -9.24914554e-02
3.93662721e-01 1.06330082e-01 -7.95221806e-01 -5.87876774e-02
6.52293921e-01 5.40154353e-02 4.64940280e-01 -1.03087175e+00
-6.10830903e-01 -5.74190021e-01 3.17704767e-01 -1.84008908e-02
-4.22166109e-01 5.27412593e-01 2.63997555e-01 -9.15692270e-01
3.40892613e-01 -8.02565873e-01 -1.28439277e-01 -4.87657309e-01
-2.90659487e-01 -3.66454780e-01 6.60429716e-01 -9.30073977e-01
1.54601109e+00 -2.38842392e+00 4.47046049e-02 4.26364571e-01
8.64402413e-01 -4.56680328e-01 3.94563943e-01 1.13375568e+00
1.39502764e-01 8.93734217e-01 -1.14827687e-02 4.11822647e-01
2.75364786e-01 3.64517182e-01 -4.80526626e-01 1.03958309e+00
-4.74219292e-01 7.73285866e-01 -1.11369812e+00 -2.33844697e-01
-2.04892486e-01 3.82407635e-01 -4.19893712e-01 -1.00652504e+00
2.29863629e-01 2.30946168e-01 -5.15597701e-01 4.84896958e-01
2.86059707e-01 -9.90144387e-02 7.90132821e-01 -1.10743538e-01
-8.61168623e-01 5.33527315e-01 -3.29177976e-01 1.20685649e+00
-1.63815427e-03 1.24085915e+00 5.70806742e-01 -5.40466368e-01
6.26445413e-01 -3.69231477e-02 2.72290587e-01 -8.40058506e-01
6.60772085e-01 2.90201485e-01 7.41673052e-01 -3.00731361e-01
7.73746610e-01 -7.38877773e-01 -5.55128992e-01 8.54989886e-01
-4.76560146e-01 -2.63446718e-02 1.11522205e-01 5.90448141e-01
1.05185437e+00 -1.28003389e-01 5.39002538e-01 -9.34969246e-01
-5.94709292e-02 2.53253579e-01 7.86788106e-01 4.26093161e-01
-1.75513580e-01 -1.00628950e-01 4.98843223e-01 -8.43028069e-01
-1.23890352e+00 -9.47053730e-01 -3.02436769e-01 1.13639128e+00
3.00005972e-01 -3.58685523e-01 -9.17869061e-02 -3.88930231e-01
3.21191043e-01 9.09899354e-01 -8.39370489e-01 -5.75681925e-02
-7.74839282e-01 -7.54036725e-01 2.87410915e-01 -2.28414625e-01
3.96806210e-01 -7.21569419e-01 -9.61802065e-01 1.79623187e-01
-3.23992670e-01 -5.79251289e-01 -3.99697363e-01 1.24661261e-02
-6.33509219e-01 -1.37625706e+00 -2.74620831e-01 -6.18433714e-01
7.01539278e-01 -1.91629604e-01 8.35302472e-01 3.64474691e-02
-4.38195944e-01 1.74648181e-01 -2.50551909e-01 -6.21664345e-01
-9.99859154e-01 8.95998105e-02 6.29015788e-02 -2.70634115e-01
3.12464714e-01 -6.93804920e-01 -2.72945464e-01 -2.65054882e-01
-9.54190254e-01 -9.34002101e-02 7.44580552e-02 2.55317479e-01
-1.69284925e-01 8.42665508e-02 5.02026498e-01 -1.29782462e+00
7.63985634e-01 -9.33449447e-01 -1.56158626e-01 -1.09830253e-01
-5.82726061e-01 -1.85105637e-01 3.39053541e-01 -1.69389173e-01
-9.44962919e-01 -1.21038079e+00 1.64254561e-01 7.02120960e-01
4.11407024e-01 7.39840209e-01 4.68382955e-01 3.55065584e-01
9.34552670e-01 -3.57391655e-01 1.76643804e-01 -4.76707816e-01
4.83627439e-01 6.15018249e-01 2.79991031e-01 -4.77788568e-01
1.18018711e+00 8.54114175e-01 -4.08874512e-01 -1.15325999e+00
-6.79353714e-01 -1.93855837e-01 -3.02065402e-01 -2.62763470e-01
1.01140881e+00 -8.95908833e-01 -4.63513613e-01 5.30402660e-02
-1.13136590e+00 -2.97196716e-01 -5.92968106e-01 3.74901116e-01
2.28088573e-01 7.35034794e-02 -6.69468582e-01 -5.90509534e-01
1.45115973e-02 -5.15842736e-01 -5.05789444e-02 1.04349047e-01
-8.65432322e-01 -1.28514719e+00 4.59034860e-01 3.39069963e-01
7.18249619e-01 8.74618471e-01 1.51572990e+00 -6.89238608e-01
1.01052031e-01 3.28744143e-01 -7.09485039e-02 -3.37400645e-01
7.63271391e-01 -6.80763721e-02 -2.77712315e-01 -2.32691631e-01
3.08185726e-01 1.18971072e-01 6.10399902e-01 1.80812493e-01
-1.84773311e-01 -8.47855330e-01 -4.90876824e-01 -1.13593131e-04
1.61740911e+00 4.67978328e-01 3.52587759e-01 8.18223953e-01
3.58480513e-01 3.78779382e-01 -1.61776394e-01 4.77571338e-01
3.59096944e-01 2.27989957e-01 -1.62060231e-01 -2.29247704e-01
-1.11565076e-01 -3.00112665e-01 2.46913090e-01 1.11110222e+00
-3.11669499e-01 4.66697775e-02 -1.27005744e+00 7.78356671e-01
-1.55770779e+00 -1.26055014e+00 -3.66600364e-01 1.79243886e+00
8.53034496e-01 2.71042407e-01 -2.20044255e-02 -1.62885368e-01
6.26553595e-01 9.84027147e-01 -1.14663787e-01 -5.10132432e-01
-8.99948925e-02 5.01504004e-01 7.38905430e-01 1.05686557e+00
-6.45379066e-01 1.06789100e+00 6.47415686e+00 3.92114401e-01
-9.17005658e-01 4.08633173e-01 4.61649805e-01 -2.00942814e-01
-1.01427162e+00 3.31950814e-01 -3.60056758e-01 3.90230268e-01
5.05531609e-01 -7.99519777e-01 2.63536870e-01 7.48284161e-02
4.26309317e-01 -3.47825706e-01 -2.88550138e-01 3.30863267e-01
2.56594330e-01 -1.66706550e+00 -1.34352267e-01 4.59424943e-01
1.00536096e+00 2.89080173e-01 -9.25064161e-02 -2.53895342e-01
6.81159079e-01 -7.70854235e-01 1.23964930e+00 1.96780577e-01
7.56834567e-01 -4.80822623e-01 3.45502347e-01 1.23222014e-02
-7.45153368e-01 -2.69826204e-01 -1.88968390e-01 -7.22958565e-01
5.31084836e-01 7.24676132e-01 -3.30896616e-01 2.00893939e-01
4.04532939e-01 3.88338715e-01 -3.00537199e-01 8.88080895e-02
-4.24957812e-01 9.37003136e-01 -8.47034752e-02 8.30431506e-02
6.55866563e-01 -6.13333642e-01 1.11099291e+00 7.71596909e-01
-4.11115959e-03 3.38503778e-01 -3.43704782e-02 7.70695746e-01
-1.40143946e-01 1.74377412e-01 -8.67241204e-01 -6.75053775e-01
6.07396543e-01 9.97146368e-01 -1.23728728e+00 -6.57034218e-02
-3.06623548e-01 5.10314643e-01 1.37380540e-01 4.99687403e-01
-7.57817924e-01 -4.32543397e-01 8.50581765e-01 1.05167758e+00
-1.71789706e-01 -5.66825390e-01 -1.16357626e-02 -7.37078667e-01
-5.25758207e-01 -8.11583579e-01 8.43928158e-02 9.39660296e-02
-1.01079655e+00 3.48581314e-01 -1.13567606e-01 1.80407211e-01
5.31766891e-01 1.96574956e-01 -5.06381750e-01 6.54227734e-01
-9.14733350e-01 -8.95935953e-01 3.15985113e-01 4.20610830e-02
3.54075909e-01 -8.28656331e-02 4.68995929e-01 7.61188418e-02
-1.21334150e-01 2.99690850e-02 3.38344127e-01 1.52169392e-01
7.31077433e-01 -5.43305516e-01 5.38220406e-01 7.49643087e-01
2.40893275e-01 7.90761232e-01 7.98890889e-01 -1.16432285e+00
-1.02320838e+00 -5.21129191e-01 1.58914733e+00 -5.80123663e-01
1.55343425e+00 -3.23642552e-01 -4.26129848e-01 8.40773463e-01
6.14329576e-01 -9.75094259e-01 6.25238955e-01 4.75570381e-01
-3.18976671e-01 1.84918746e-01 -8.92796218e-01 9.88032758e-01
1.40120196e+00 -5.79141855e-01 -1.00623095e+00 3.29602897e-01
8.45555604e-01 4.51815397e-01 -5.15315294e-01 -2.03597888e-01
6.77487552e-01 -4.80221719e-01 6.53880358e-01 -4.98215735e-01
7.51169622e-01 1.38870627e-01 -2.55247593e-01 -1.34466338e+00
-1.11616182e+00 -8.60189795e-01 9.09845233e-01 1.28966963e+00
4.33904082e-01 -1.17428303e+00 3.12538296e-01 3.41207474e-01
-2.21664347e-02 -3.47527802e-01 -1.15396202e+00 -1.82174712e-01
3.66785914e-01 8.17552656e-02 1.76119953e-01 1.73021889e+00
2.43865415e-01 5.66689193e-01 1.58987418e-01 -3.34558338e-01
4.61268038e-01 -1.32422790e-01 6.08728647e-01 -1.61075997e+00
1.54331595e-01 -8.37870836e-01 -3.55499953e-01 -2.57393956e-01
7.15388134e-02 -1.40077341e+00 -3.70889693e-01 -1.83233654e+00
3.41139615e-01 -3.80529702e-01 2.54728287e-01 2.39460871e-01
3.47604930e-01 -2.21871175e-02 4.62572038e-01 5.67029536e-01
1.60453945e-01 2.76343286e-01 1.22543609e+00 -1.60226226e-01
-4.96572644e-01 -8.79433870e-01 -1.62597132e+00 9.86045003e-01
6.87720597e-01 -7.11567998e-01 -7.27306008e-02 -5.23278773e-01
1.11802173e+00 -4.73167270e-01 1.94106534e-01 -6.51284754e-01
1.00345619e-01 -4.31885034e-01 1.63219020e-01 1.41959384e-01
-2.87169218e-01 -8.65822673e-01 5.74935079e-01 9.10358846e-01
-1.89474598e-01 1.33963004e-01 1.91590533e-01 3.09703588e-01
-4.80491631e-02 1.68846637e-01 6.82146490e-01 -3.64926606e-02
-2.91013658e-01 -2.35710159e-01 -8.23464751e-01 5.64663231e-01
7.38175035e-01 -2.03462467e-01 -7.79090226e-01 -2.76609987e-01
-5.53950608e-01 -1.81225494e-01 9.15829837e-01 3.88225734e-01
6.52938336e-02 -1.37273443e+00 -1.16187298e+00 -1.22850783e-01
-4.46258366e-01 -7.15177417e-01 -1.13554299e-01 1.12054896e+00
-6.76602721e-01 -1.56902510e-03 1.99813917e-02 3.97344470e-01
-1.06177759e+00 1.13746874e-01 -1.29526839e-01 -5.83523735e-02
-1.08917356e+00 6.98321044e-01 3.77387971e-01 -3.00783515e-01
-4.82454151e-01 -1.83983356e-01 -3.31197418e-02 8.54573011e-01
2.26602018e-01 7.29779959e-01 -8.94150019e-01 -1.05322576e+00
-4.91110206e-01 5.03634155e-01 1.49303123e-01 -4.16668028e-01
1.52821052e+00 -3.43844205e-01 -7.70736039e-01 1.03598332e+00
1.15836966e+00 6.88679159e-01 -2.52973944e-01 -1.03664823e-01
-1.11484043e-01 -7.46151626e-01 -1.48775145e-01 -9.70283031e-01
-5.81911623e-01 1.38660207e-01 -6.88351765e-02 5.07006168e-01
7.26418123e-02 4.55054283e-01 7.37872183e-01 -2.62821794e-01
1.62482381e-01 -1.23731697e+00 -3.09043884e-01 3.52305621e-01
1.10158575e+00 -6.75818563e-01 3.45045060e-01 -1.02237724e-01
-3.22603643e-01 7.11716354e-01 -1.32436424e-01 -1.96464106e-01
8.61869574e-01 6.24096729e-02 2.28813365e-02 -7.46406555e-01
-1.77293390e-01 2.25384533e-01 -1.64453223e-01 -3.84429954e-02
5.72597444e-01 4.33464408e-01 -1.36469233e+00 3.97387892e-01
-5.09126306e-01 -4.13873106e-01 6.51276767e-01 7.84300387e-01
-8.32260728e-01 -8.83292913e-01 -6.25181496e-01 5.51456630e-01
-6.58361137e-01 -1.83916196e-01 -9.77209568e-01 1.29669833e+00
6.70703709e-01 8.45855117e-01 4.33545798e-01 -2.40041211e-01
-5.35524078e-02 7.55437687e-02 3.49113107e-01 -5.34660101e-01
-9.11152601e-01 4.27424043e-01 5.64729273e-01 -1.63924947e-01
-3.20384085e-01 -9.07262087e-01 -1.23082721e+00 -1.16089666e+00
-8.86205956e-02 3.71446311e-01 5.46136975e-01 9.32547867e-01
4.42590192e-02 3.33344549e-01 3.77873689e-01 -1.50517941e-01
-2.43859544e-01 -9.67512965e-01 -9.07011628e-01 3.41098666e-01
3.77877662e-03 -1.99627727e-01 -6.59884453e-01 -2.39337519e-01]
|
[9.002531051635742, 10.006030082702637]
|
f0883ad1-f457-4040-9097-7bdab1d76073
|
scene-flow-to-action-map-a-new-representation
|
1702.08652
| null |
http://arxiv.org/abs/1702.08652v3
|
http://arxiv.org/pdf/1702.08652v3.pdf
|
Scene Flow to Action Map: A New Representation for RGB-D based Action Recognition with Convolutional Neural Networks
|
Scene flow describes the motion of 3D objects in real world and potentially
could be the basis of a good feature for 3D action recognition. However, its
use for action recognition, especially in the context of convolutional neural
networks (ConvNets), has not been previously studied. In this paper, we propose
the extraction and use of scene flow for action recognition from RGB-D data.
Previous works have considered the depth and RGB modalities as separate
channels and extract features for later fusion. We take a different approach
and consider the modalities as one entity, thus allowing feature extraction for
action recognition at the beginning. Two key questions about the use of scene
flow for action recognition are addressed: how to organize the scene flow
vectors and how to represent the long term dynamics of videos based on scene
flow. In order to calculate the scene flow correctly on the available datasets,
we propose an effective self-calibration method to align the RGB and depth data
spatially without knowledge of the camera parameters. Based on the scene flow
vectors, we propose a new representation, namely, Scene Flow to Action Map
(SFAM), that describes several long term spatio-temporal dynamics for action
recognition. We adopt a channel transform kernel to transform the scene flow
vectors to an optimal color space analogous to RGB. This transformation takes
better advantage of the trained ConvNets models over ImageNet. Experimental
results indicate that this new representation can surpass the performance of
state-of-the-art methods on two large public datasets.
|
['Pichao Wang', 'Zhimin Gao', 'Philip Ogunbona', 'Chang Tang', 'Yuyao Zhang', 'Wanqing Li']
|
2017-02-28
|
scene-flow-to-action-map-a-new-representation-1
|
http://openaccess.thecvf.com/content_cvpr_2017/html/Wang_Scene_Flow_to_CVPR_2017_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2017/papers/Wang_Scene_Flow_to_CVPR_2017_paper.pdf
|
cvpr-2017-7
|
['3d-human-action-recognition']
|
['computer-vision']
|
[ 3.60010237e-01 -6.23180807e-01 -1.44734517e-01 -2.88957328e-01
-5.69583662e-02 -4.40700233e-01 7.36342788e-01 -3.09379250e-01
-5.83448648e-01 4.55789298e-01 1.77571088e-01 5.51639907e-02
-1.75712302e-01 -7.88903356e-01 -4.22850817e-01 -8.86253953e-01
1.05646029e-01 -8.59588757e-02 4.62786764e-01 -5.55479415e-02
3.07897061e-01 9.85589802e-01 -1.70196533e+00 3.96672994e-01
3.30715895e-01 1.14846075e+00 -1.60263553e-01 8.39357972e-01
-4.85766739e-01 1.31277359e+00 -5.42460382e-01 -6.38242345e-03
4.05064940e-01 -8.47715616e-01 -9.02443230e-01 3.98284912e-01
4.57601458e-01 -4.54222381e-01 -6.61506772e-01 7.96647787e-01
1.83962405e-01 2.71096706e-01 5.67579746e-01 -1.44441724e+00
-2.03291446e-01 -8.68433639e-02 -2.80092031e-01 4.56658453e-01
5.86551130e-01 3.08157265e-01 5.19519985e-01 -4.77742493e-01
7.29220927e-01 1.14633679e+00 2.99135834e-01 4.38388973e-01
-8.49627137e-01 -4.00311261e-01 7.47340694e-02 8.85362387e-01
-1.17299592e+00 -3.07834059e-01 1.09052169e+00 -5.33813655e-01
1.04864669e+00 1.95936158e-01 1.26574552e+00 1.06049287e+00
1.95797339e-01 8.12008023e-01 1.08695388e+00 -4.90879148e-01
2.29877725e-01 -2.09508300e-01 7.75593594e-02 5.69340289e-01
-6.91571236e-02 -7.08571523e-02 -7.91605771e-01 4.11740810e-01
9.70618367e-01 1.04341537e-01 -2.87910461e-01 -6.11906469e-01
-1.31775391e+00 5.74476242e-01 4.22221661e-01 6.63496017e-01
-4.25828040e-01 3.34377348e-01 3.26139539e-01 1.54296666e-01
3.94743055e-01 5.07106856e-02 -3.21904540e-01 -6.05423808e-01
-6.01501524e-01 -2.22771242e-02 6.03541911e-01 4.32230711e-01
7.69925058e-01 1.50027931e-01 -5.10749184e-02 3.77712339e-01
2.13364631e-01 7.04524577e-01 5.33671439e-01 -1.08416355e+00
5.22329926e-01 1.04246950e+00 -8.34244266e-02 -1.27874076e+00
-5.37806034e-01 2.02254713e-01 -8.87908340e-01 2.66986012e-01
8.21438193e-01 1.93245545e-01 -7.41768479e-01 1.24759293e+00
4.07884091e-01 6.56455934e-01 2.52088636e-01 9.42448676e-01
5.97053111e-01 6.24369025e-01 1.83030069e-02 -9.84036177e-02
9.78877187e-01 -6.52849495e-01 -7.75207102e-01 -3.16734761e-02
9.65281129e-01 -4.20073271e-01 7.32224345e-01 4.02096212e-01
-7.17011213e-01 -7.71104991e-01 -1.00010014e+00 8.35649669e-03
-6.75167501e-01 3.14648718e-01 7.36874878e-01 6.78085089e-01
-8.73430431e-01 6.29377067e-01 -1.18425858e+00 -5.45417190e-01
3.04569602e-01 2.33517602e-01 -7.81412601e-01 -3.16518210e-02
-1.04629779e+00 9.43090796e-01 5.29336751e-01 3.90556455e-01
-5.82849681e-01 -2.83444375e-01 -8.44284654e-01 -2.40816221e-01
1.69322059e-01 -4.10873055e-01 7.43361712e-01 -1.17700994e+00
-1.73900044e+00 6.07169926e-01 -3.49023230e-02 -3.33369911e-01
3.71408433e-01 -1.28851727e-01 -3.51701230e-01 5.69749415e-01
-2.71358550e-01 4.07873213e-01 7.15707719e-01 -7.86909223e-01
-7.42573321e-01 -4.48839098e-01 3.73594373e-01 2.88434148e-01
-6.21838927e-01 -1.13174275e-01 -5.43007076e-01 -2.31981739e-01
2.47239679e-01 -8.94876719e-01 -5.55587448e-02 2.19707593e-01
-3.51500437e-02 -3.75350937e-02 1.04002559e+00 -5.47465801e-01
1.01549041e+00 -2.08680940e+00 2.31308028e-01 1.47893235e-01
-1.00179151e-01 4.33156133e-01 -1.31065696e-01 1.03464358e-01
-1.69647783e-01 -2.84459293e-01 -2.50400275e-01 -1.11800972e-02
-4.02810812e-01 5.26814818e-01 -1.22625679e-02 8.60492289e-01
2.56888300e-01 7.89613962e-01 -9.36339915e-01 -4.76891816e-01
8.88944030e-01 7.99033940e-01 -4.18426245e-01 1.03463978e-01
1.98929772e-01 7.29681730e-01 -6.26742721e-01 3.41433585e-01
5.88768125e-01 1.99939478e-02 2.20009144e-02 -5.03136694e-01
-1.08464614e-01 -4.97598276e-02 -1.45046115e+00 1.92676127e+00
-3.29953492e-01 8.65903914e-01 -4.75365520e-01 -1.20377207e+00
1.02567148e+00 1.11438237e-01 1.24187553e+00 -9.37712371e-01
3.63803715e-01 -1.37125561e-02 -1.73834383e-01 -6.87068760e-01
3.57912481e-01 1.68363050e-01 1.01905093e-01 3.59423548e-01
1.59733951e-01 -7.09247515e-02 2.72746563e-01 -8.63089338e-02
1.17473316e+00 6.07199132e-01 2.43199885e-01 1.92852929e-01
1.03984952e+00 9.33285579e-02 3.93278390e-01 3.52867216e-01
-3.75778437e-01 5.20004511e-01 5.16508162e-01 -6.81146920e-01
-7.23004341e-01 -6.37943685e-01 2.17299849e-01 5.02691448e-01
3.43866974e-01 -3.35559666e-01 -7.25346446e-01 -7.76633799e-01
-1.34723946e-01 3.41764539e-01 -6.10961854e-01 -2.41483271e-01
-8.84953082e-01 -5.45393765e-01 5.09267569e-01 5.98330438e-01
1.04007554e+00 -1.02558362e+00 -1.13109517e+00 1.75269559e-01
-2.56816655e-01 -1.52227890e+00 -1.06987670e-01 -5.67873418e-02
-9.12421346e-01 -1.38959360e+00 -5.73598683e-01 -1.10784039e-01
4.35899258e-01 4.49260205e-01 5.34626722e-01 -1.57725632e-01
-3.22053850e-01 9.06005979e-01 -7.06607044e-01 5.41783422e-02
-3.00101042e-01 -2.86728680e-01 -1.47142783e-02 5.53505898e-01
4.92076457e-01 -5.01489401e-01 -6.65345907e-01 3.78317147e-01
-1.17058539e+00 3.64965089e-02 2.76109397e-01 3.29972774e-01
4.12343651e-01 -7.09163919e-02 -2.55187273e-01 -3.73106688e-01
3.43602686e-03 -9.53103323e-03 -4.13973749e-01 2.22793221e-01
-7.69141689e-02 6.80479854e-02 4.27307516e-01 -3.04763943e-01
-1.08638883e+00 6.01396203e-01 -3.34656499e-02 -8.02784264e-01
-4.36731637e-01 2.12531984e-02 -1.93200141e-01 -3.59118581e-01
4.94575292e-01 3.04636478e-01 1.10257410e-01 -4.55452681e-01
5.64492524e-01 5.32138944e-01 3.65324110e-01 -1.60050258e-01
6.11390293e-01 9.03870881e-01 4.28458601e-01 -1.02669072e+00
-6.02247179e-01 -7.81183422e-01 -1.35110545e+00 -7.17884183e-01
1.38511693e+00 -6.42777026e-01 -7.86099017e-01 9.52398956e-01
-1.37363458e+00 -2.86349386e-01 -4.60184753e-01 8.95468712e-01
-8.26795101e-01 5.32353878e-01 -2.74054497e-01 -6.78602040e-01
9.72102284e-02 -1.12356031e+00 9.88768101e-01 1.57526225e-01
2.30274305e-01 -1.07379067e+00 2.36794725e-01 2.94422179e-01
1.58522248e-01 5.18350899e-01 4.08499092e-01 -2.44103119e-01
-7.23155260e-01 -1.96740538e-01 -2.00068787e-01 6.80301845e-01
3.11108887e-01 2.03613818e-01 -9.31729674e-01 1.72675624e-01
1.53508231e-01 6.46060631e-02 8.84072185e-01 3.57114077e-01
1.10559547e+00 1.18519060e-01 -1.22133851e-01 8.39657962e-01
1.42182124e+00 4.18346584e-01 9.44250762e-01 4.96935368e-01
1.13510060e+00 5.29052436e-01 5.58494568e-01 6.53537214e-01
1.30707413e-01 9.22810197e-01 4.79363918e-01 2.02576499e-02
-3.98250461e-01 -4.88916179e-03 6.48797691e-01 7.73143768e-01
-6.30967140e-01 -6.74925074e-02 -9.28307414e-01 2.18578488e-01
-1.92911339e+00 -1.11881840e+00 -2.76019752e-01 2.00022864e+00
2.72128135e-01 -5.99024631e-02 4.33604196e-02 4.58368391e-01
3.46432835e-01 3.76835614e-01 -3.40794384e-01 -2.42645517e-01
-2.11747855e-01 2.12303832e-01 6.98694944e-01 2.06555858e-01
-1.20653415e+00 8.78153205e-01 5.66602993e+00 5.66126227e-01
-1.33873594e+00 -2.69856323e-02 1.82725921e-01 5.26949093e-02
2.79253572e-01 1.12753406e-01 -7.57135153e-01 1.72598764e-01
9.07236874e-01 1.46537915e-01 1.98793307e-01 6.60026848e-01
4.17110711e-01 -3.47006708e-01 -1.02040017e+00 1.30959094e+00
3.62771332e-01 -1.17877424e+00 9.21388417e-02 4.42741513e-02
5.14926314e-01 -2.32303828e-01 -3.35052967e-01 -7.93334469e-02
-1.71135858e-01 -5.85755706e-01 6.03174984e-01 9.88462269e-01
4.90428776e-01 -4.84739959e-01 5.46603203e-01 1.45128474e-01
-1.35285759e+00 -1.52305081e-01 -2.30345726e-01 -1.64430976e-01
3.02547038e-01 1.99272752e-01 -5.38971543e-01 7.40254045e-01
6.15051866e-01 1.45249009e+00 -8.67806792e-01 1.05163288e+00
-5.54695912e-02 3.50223184e-01 -2.95387983e-01 2.65955627e-02
3.98289979e-01 -3.54159802e-01 3.93159956e-01 1.03052437e+00
4.52548832e-01 6.66005611e-02 -6.14606515e-02 5.87098420e-01
4.21427876e-01 9.27500054e-02 -7.66635776e-01 -9.46973413e-02
-3.01520050e-01 1.10684335e+00 -9.40558136e-01 -2.47215793e-01
-5.04385173e-01 1.15326893e+00 -2.16512054e-01 4.47091430e-01
-8.67088318e-01 -1.32599205e-01 7.19276071e-01 -6.47496060e-02
2.85993427e-01 -6.68439984e-01 6.58261329e-02 -1.26869941e+00
-1.17603568e-02 -5.15464902e-01 3.47892612e-01 -8.99569929e-01
-7.65134096e-01 4.24945056e-01 3.47143471e-01 -1.75818598e+00
-3.89096856e-01 -1.09020746e+00 -3.39799464e-01 4.07527804e-01
-1.53525734e+00 -1.06936872e+00 -7.03475416e-01 1.12542152e+00
4.14394587e-01 -1.49177015e-01 5.62849402e-01 2.24498779e-01
-4.29495364e-01 6.22160407e-03 6.97922930e-02 3.51859778e-01
3.22700590e-01 -9.13056672e-01 3.37954983e-02 9.65485215e-01
5.96344709e-01 -1.38636589e-01 2.85449147e-01 -2.80909926e-01
-1.71522379e+00 -9.65397418e-01 5.38097978e-01 -5.95217109e-01
5.69919705e-01 -1.02512777e-01 -7.07751513e-01 3.89758885e-01
-1.09701581e-01 3.29653054e-01 3.98973256e-01 -5.08927524e-01
-1.34541364e-02 -5.34763515e-01 -7.96544015e-01 1.58251256e-01
1.13866007e+00 -6.00621581e-01 -2.58693427e-01 2.21325278e-01
2.13604361e-01 -3.49554420e-01 -8.67756188e-01 3.79241198e-01
6.43953562e-01 -1.24204707e+00 1.07499468e+00 -5.15695691e-01
2.93349743e-01 -4.75340754e-01 -2.94547379e-01 -1.09961200e+00
4.97954153e-02 -1.26948029e-01 -1.70817241e-01 8.71357799e-01
-2.94666260e-01 -4.30905879e-01 8.59949052e-01 4.89402562e-01
1.14815414e-01 -3.61070395e-01 -1.16055608e+00 -7.43780673e-01
-2.70825297e-01 -8.06427479e-01 4.95468199e-01 8.22056592e-01
-2.37302735e-01 -8.84259716e-02 -2.96328992e-01 -8.00756216e-02
4.15635288e-01 1.54239554e-02 1.02907479e+00 -1.00452077e+00
-5.35724945e-02 -4.68768090e-01 -1.28209114e+00 -1.06294167e+00
2.85531223e-01 -6.42209649e-01 -1.53597787e-01 -1.53671622e+00
-9.83756185e-02 -1.51031137e-01 -3.37719053e-01 4.62904841e-01
2.47339353e-01 4.47269708e-01 5.12145519e-01 1.35971844e-01
-5.07740855e-01 7.25007832e-01 1.47915137e+00 -2.33963162e-01
-3.23880434e-01 -1.24770463e-01 1.44267008e-01 7.87297606e-01
6.47801876e-01 -1.46197453e-01 -4.39397037e-01 -3.62761527e-01
-1.68565467e-01 7.13765919e-02 5.97850621e-01 -1.49862504e+00
3.16397488e-01 -4.50128764e-01 4.98974353e-01 -5.18913209e-01
6.30548060e-01 -1.19198096e+00 2.75159121e-01 3.21641266e-01
-1.64870396e-01 6.76643383e-03 1.29646793e-01 5.31256378e-01
-4.33373541e-01 -9.86270234e-02 4.90199685e-01 -1.80811033e-01
-1.27888393e+00 3.38651568e-01 -3.75829548e-01 -3.40116501e-01
1.26132298e+00 -7.30642140e-01 -1.77571610e-01 -2.49139115e-01
-6.93845093e-01 -3.47289503e-01 2.09551126e-01 6.07304156e-01
7.52638936e-01 -1.53376973e+00 -3.40217173e-01 4.05442208e-01
5.39521873e-02 -3.39426041e-01 4.33663130e-01 9.67438161e-01
-8.08548510e-01 5.61738610e-01 -7.04080522e-01 -9.04270053e-01
-1.30508566e+00 3.30196023e-01 5.65330088e-01 -8.09458867e-02
-6.92131758e-01 3.30143064e-01 -2.33600438e-02 7.08000436e-02
1.01856507e-01 -5.61164379e-01 -6.83942616e-01 2.00209588e-01
4.99855131e-01 4.77229655e-01 4.63298447e-02 -1.24701703e+00
-5.56878328e-01 1.12028837e+00 5.63423276e-01 -3.49675268e-02
1.25328946e+00 -1.26072228e-01 -3.77147608e-02 5.53662896e-01
1.50427973e+00 -5.78190327e-01 -1.35587656e+00 -2.99192183e-02
-1.46642730e-01 -9.20044363e-01 2.39745632e-01 -1.61791846e-01
-1.53008115e+00 1.10805857e+00 8.96450996e-01 2.08492056e-01
1.44562197e+00 -2.31203437e-01 5.19089818e-01 3.80020529e-01
3.21053594e-01 -1.03678632e+00 2.87114114e-01 5.94827712e-01
6.62145138e-01 -9.97349083e-01 7.08448738e-02 -3.63121867e-01
-6.18032694e-01 1.59230971e+00 6.25013530e-01 -4.68098484e-02
7.16952980e-01 -1.02496959e-01 8.18160325e-02 -1.38259590e-01
-3.39899391e-01 -5.74573636e-01 4.01414573e-01 6.67313039e-01
8.22852924e-02 -2.12525651e-01 -6.30333200e-02 -8.46177414e-02
2.76586920e-01 2.31918901e-01 5.41018069e-01 9.60908413e-01
-1.83190018e-01 -1.15643072e+00 -2.48537585e-01 8.29042196e-02
7.38282129e-03 4.72918123e-01 -3.85750979e-01 8.29214931e-01
3.64092201e-01 7.36560822e-01 2.40197152e-01 -6.32087529e-01
6.36975288e-01 1.29025325e-01 8.80460799e-01 -1.32349536e-01
-2.20138520e-01 -2.65772939e-01 -1.30020082e-01 -1.13294446e+00
-1.38974667e+00 -7.12455928e-01 -1.10804331e+00 -1.41730949e-01
-4.22777701e-03 -3.75982016e-01 9.42860603e-01 1.14161003e+00
1.58352390e-01 4.36227322e-01 7.71196127e-01 -1.00912333e+00
-4.34906036e-03 -7.35147834e-01 -7.14296997e-01 7.85700500e-01
2.61119545e-01 -9.65658784e-01 -4.45689738e-01 4.94291812e-01]
|
[7.9055399894714355, 0.3626278042793274]
|
abb206c3-24c9-42ec-9526-bff55ee19fc8
|
a-single-stream-network-for-robust-and-real
|
2007.06811
| null |
https://arxiv.org/abs/2007.06811v2
|
https://arxiv.org/pdf/2007.06811v2.pdf
|
A Single Stream Network for Robust and Real-time RGB-D Salient Object Detection
|
Existing RGB-D salient object detection (SOD) approaches concentrate on the cross-modal fusion between the RGB stream and the depth stream. They do not deeply explore the effect of the depth map itself. In this work, we design a single stream network to directly use the depth map to guide early fusion and middle fusion between RGB and depth, which saves the feature encoder of the depth stream and achieves a lightweight and real-time model. We tactfully utilize depth information from two perspectives: (1) Overcoming the incompatibility problem caused by the great difference between modalities, we build a single stream encoder to achieve the early fusion, which can take full advantage of ImageNet pre-trained backbone model to extract rich and discriminative features. (2) We design a novel depth-enhanced dual attention module (DEDA) to efficiently provide the fore-/back-ground branches with the spatially filtered features, which enables the decoder to optimally perform the middle fusion. Besides, we put forward a pyramidally attended feature extraction module (PAFE) to accurately localize the objects of different scales. Extensive experiments demonstrate that the proposed model performs favorably against most state-of-the-art methods under different evaluation metrics. Furthermore, this model is 55.5\% lighter than the current lightest model and runs at a real-time speed of 32 FPS when processing a $384 \times 384$ image.
|
['Huchuan Lu', 'Youwei Pang', 'Lihe Zhang', 'Lei Zhang', 'Xiaoqi Zhao']
|
2020-07-14
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4160_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123670647.pdf
|
eccv-2020-8
|
['rgb-d-salient-object-detection', 'thermal-image-segmentation']
|
['computer-vision', 'computer-vision']
|
[ 2.55678147e-02 -1.03431195e-01 5.69240078e-02 -3.26847166e-01
-6.27796352e-01 -1.38769209e-01 4.26128000e-01 -7.60157183e-02
-6.36654735e-01 1.34694353e-01 1.78890362e-01 -1.00025699e-01
2.61705011e-01 -9.24693942e-01 -6.58854425e-01 -6.30436778e-01
9.57468003e-02 -2.74709851e-01 7.61829674e-01 -1.69584930e-01
1.83017671e-01 4.01570112e-01 -1.72191906e+00 3.79994392e-01
8.19867730e-01 1.69720936e+00 6.77817345e-01 5.11381328e-01
-1.60257027e-01 8.79763424e-01 -1.78890601e-01 -2.35065430e-01
4.39318657e-01 -2.25793898e-01 -4.08335149e-01 1.81689277e-01
3.30397666e-01 -1.01088798e+00 -8.05131197e-01 1.10767508e+00
7.63010502e-01 -1.79885358e-01 1.78168431e-01 -1.08439481e+00
-3.79341841e-01 2.72055954e-01 -9.29916084e-01 4.58112121e-01
2.87014395e-01 4.99505222e-01 7.09231734e-01 -1.07352686e+00
3.41781825e-01 1.22829711e+00 2.85523474e-01 3.21936965e-01
-5.74359357e-01 -7.70749509e-01 5.80660939e-01 3.70834172e-01
-1.30750692e+00 -5.07961512e-01 9.13604319e-01 -3.26173790e-02
9.40401971e-01 -1.11199416e-01 7.68084943e-01 7.08825409e-01
3.55495960e-02 1.17691040e+00 9.97217536e-01 -1.47026777e-01
1.51189744e-01 -1.82841167e-01 -8.26483816e-02 9.99671996e-01
-9.44211613e-03 7.22898617e-02 -8.71077359e-01 3.81092966e-01
1.03517222e+00 3.96029502e-01 -4.28386301e-01 -1.82742521e-01
-1.28717911e+00 5.00558138e-01 9.15550828e-01 1.62237480e-01
-4.95986521e-01 3.45655203e-01 1.84791997e-01 1.32404240e-02
2.87572086e-01 -1.07899331e-01 -3.35897714e-01 -3.69005129e-02
-7.25635111e-01 -9.71853286e-02 1.60706177e-01 1.04962206e+00
9.85665917e-01 -1.44704014e-01 -2.01543868e-01 4.79720592e-01
6.57980859e-01 6.11431539e-01 3.46888036e-01 -9.45977688e-01
8.34699512e-01 8.51996005e-01 6.44875616e-02 -9.37369168e-01
-4.09797817e-01 -3.97232741e-01 -7.24886835e-01 2.94439614e-01
3.00808728e-01 1.73875224e-02 -9.42494452e-01 1.47383189e+00
5.28719366e-01 1.67718843e-01 -5.38636232e-03 1.24988210e+00
8.66678298e-01 6.19388700e-01 3.78044061e-02 -1.68973822e-02
1.46761572e+00 -1.08584881e+00 -4.20542270e-01 -5.66613913e-01
4.19253647e-01 -5.84993780e-01 1.05772424e+00 2.67879069e-01
-1.32112670e+00 -7.50954688e-01 -1.28377879e+00 -6.13266647e-01
-2.13902071e-01 2.76396155e-01 7.97964036e-01 2.72909164e-01
-9.12820458e-01 4.79963303e-01 -1.21982884e+00 -6.92143664e-02
6.46564841e-01 4.43606108e-01 -2.69859701e-01 -4.32002246e-01
-9.92772698e-01 6.09605908e-01 2.55127758e-01 3.73554051e-01
-9.18510675e-01 -5.75017214e-01 -9.56542730e-01 -4.09054607e-02
2.85847366e-01 -7.82909811e-01 1.12750351e+00 -8.18705142e-01
-1.51816630e+00 5.72962165e-01 -3.67999315e-01 -1.03292033e-01
3.53022039e-01 -3.71202379e-01 -1.16937317e-01 5.94265282e-01
7.42001012e-02 8.40121746e-01 6.27015769e-01 -1.07127106e+00
-1.18066752e+00 -5.80627024e-01 3.36531252e-01 4.66310829e-01
-5.88444233e-01 -1.64876357e-01 -9.35979664e-01 -4.42703575e-01
6.12346292e-01 -4.53734398e-01 -1.41808122e-01 4.32064652e-01
-2.42044434e-01 -7.37980679e-02 7.96841860e-01 -5.08973420e-01
1.11002862e+00 -2.29653859e+00 6.64025918e-02 -2.10046709e-01
3.68807822e-01 2.15613917e-01 5.10514081e-02 8.19740593e-02
2.82697231e-01 -2.39557326e-01 -3.76885906e-02 -7.90777087e-01
-1.43725976e-01 1.18675470e-01 -2.65456557e-01 6.29121184e-01
3.32786918e-01 9.81479645e-01 -9.86923754e-01 -5.42429030e-01
4.20818895e-01 5.17638564e-01 -6.15839601e-01 3.48712891e-01
-2.51298770e-02 2.20617399e-01 -7.64174163e-01 9.40465331e-01
8.72772098e-01 -3.32806677e-01 -2.36690819e-01 -4.49450076e-01
-3.32882166e-01 4.88998413e-01 -1.19061053e+00 2.27158117e+00
-3.97317618e-01 3.35655242e-01 1.07186086e-01 -5.74694633e-01
7.38022685e-01 -1.22732446e-01 3.28249484e-01 -1.10421586e+00
4.41315562e-01 3.29260230e-01 -1.63630128e-01 -3.86415184e-01
2.28549659e-01 8.44872594e-02 8.20581615e-03 2.70495802e-01
1.17998295e-01 8.93071592e-02 -7.56286383e-02 1.46148309e-01
1.21050406e+00 2.82253563e-01 2.01621745e-02 1.28238037e-01
5.77024996e-01 -4.31360960e-01 5.73826730e-01 4.24970061e-01
-3.98073167e-01 7.70868897e-01 3.25811416e-01 -2.99314022e-01
-6.78764105e-01 -1.22959948e+00 1.60431102e-01 1.00235462e+00
8.50014746e-01 -3.26407254e-01 -3.48046422e-01 -6.96449399e-01
-1.95848659e-01 2.23399982e-01 -5.70180833e-01 -3.05729955e-01
-6.05314255e-01 -4.43810105e-01 1.83892921e-01 8.78259003e-01
1.11957395e+00 -7.64431894e-01 -1.10928261e+00 1.81523889e-01
-2.88483113e-01 -1.34417570e+00 -3.62698317e-01 3.30412537e-01
-9.16563690e-01 -8.45608532e-01 -6.77900791e-01 -6.39116108e-01
5.57835579e-01 7.99393237e-01 7.24262357e-01 -2.66230740e-02
-8.88202786e-02 1.10486805e-01 -5.33687472e-01 -3.34298372e-01
3.07332963e-01 1.36822030e-01 -2.75573611e-01 6.47810325e-02
4.13070738e-01 -6.28347099e-01 -1.23426318e+00 2.61572599e-01
-9.40342486e-01 4.48138684e-01 8.94943476e-01 6.14516556e-01
6.55698717e-01 -1.21864185e-01 3.29092920e-01 -1.30601197e-01
-2.13195726e-01 -3.07039887e-01 -4.19550240e-01 2.77028717e-02
-3.35801333e-01 3.54440063e-02 3.84415984e-01 -9.86523703e-02
-1.03180695e+00 2.79102147e-01 -2.88039744e-01 -7.13541329e-01
2.29725763e-02 -1.67148095e-02 -4.65927988e-01 -6.25078678e-02
2.03206524e-01 4.06772524e-01 -1.78931979e-03 -5.95993459e-01
3.12773556e-01 7.39845872e-01 5.36182940e-01 -2.22569615e-01
7.01396167e-01 9.60473418e-01 -1.59116492e-01 -4.28711951e-01
-8.81914675e-01 -4.70817834e-01 -4.90400940e-01 -2.43568361e-01
8.02511930e-01 -1.35107672e+00 -9.18715358e-01 7.27523506e-01
-1.17120993e+00 -2.94062525e-01 -1.65637344e-01 5.57235956e-01
-4.01783615e-01 3.38932693e-01 -7.65585005e-01 -6.38217211e-01
-3.92653614e-01 -1.18461609e+00 1.48157334e+00 4.74343508e-01
5.05959749e-01 -2.95334578e-01 -5.33766866e-01 1.29305035e-01
3.70750040e-01 5.67648709e-02 5.04487872e-01 5.00530861e-02
-1.00758564e+00 9.58688557e-03 -8.15344870e-01 2.38700494e-01
5.96620049e-03 -2.98203498e-01 -1.29636776e+00 -1.72416821e-01
1.23609208e-01 -2.21601889e-01 1.19184995e+00 1.30742073e-01
1.21049488e+00 1.41048193e-01 -3.19273949e-01 9.84331667e-01
1.42357349e+00 4.79617156e-02 6.73069894e-01 4.17688102e-01
9.12584126e-01 4.63622183e-01 8.04751933e-01 6.41623437e-01
8.65442157e-01 4.72909629e-01 9.24009085e-01 -3.18355918e-01
-3.76619041e-01 -4.16966885e-01 4.79060024e-01 6.76067591e-01
-1.57227479e-02 -1.52949914e-01 -6.64372444e-01 4.62669104e-01
-1.81395221e+00 -7.00532138e-01 1.50995478e-01 1.88770401e+00
7.83692598e-01 3.60279053e-01 -8.88266340e-02 2.41133198e-01
4.62766320e-01 2.60596484e-01 -6.21025026e-01 1.36540875e-01
-1.45371914e-01 1.71408325e-01 6.35311186e-01 2.68068045e-01
-1.12934387e+00 7.27224231e-01 4.95320606e+00 8.64986658e-01
-1.30334926e+00 6.05164357e-02 5.57255268e-01 -5.53082168e-01
-3.05416346e-01 -9.30480659e-02 -9.84726906e-01 6.05642200e-01
5.63009381e-01 2.01736435e-01 9.50924009e-02 8.36708784e-01
1.35408223e-01 -3.80999684e-01 -1.08402491e+00 1.18789589e+00
-2.45847143e-02 -1.14275157e+00 -1.90076411e-01 1.00527100e-01
3.58280659e-01 2.88636357e-01 3.08185760e-02 1.53759047e-01
-9.05864686e-02 -6.88457489e-01 1.13885343e+00 4.66695517e-01
7.49618530e-01 -8.52279902e-01 6.84363425e-01 3.27537715e-01
-1.48928618e+00 -3.88194591e-01 -4.93457913e-01 -2.11911708e-01
2.35320732e-01 8.34510624e-01 -2.93198764e-01 6.29617870e-01
9.95073318e-01 9.70493555e-01 -6.70656562e-01 9.59337115e-01
-3.32762450e-01 6.92896247e-02 -6.21021748e-01 1.65208861e-01
3.39246452e-01 2.84459621e-01 2.05541730e-01 9.09884870e-01
4.24921364e-01 3.08675587e-01 6.79587126e-02 6.58980370e-01
4.62647602e-02 -2.75649279e-01 -2.42971033e-01 2.74843007e-01
4.31623220e-01 1.16959715e+00 -7.43072331e-01 -3.00907433e-01
-6.74884796e-01 1.28644502e+00 2.67743021e-01 1.95600331e-01
-9.70948339e-01 -5.25376081e-01 7.01933801e-01 1.39849231e-01
7.48278737e-01 -2.05505595e-01 -3.76893580e-01 -1.28112602e+00
3.62747490e-01 -4.55157846e-01 2.55618483e-01 -9.33809817e-01
-1.04449940e+00 7.14504242e-01 -3.08978528e-01 -1.44057286e+00
1.25011012e-01 -5.97957492e-01 -2.88650095e-01 9.38317060e-01
-2.09939551e+00 -1.19553173e+00 -7.76311278e-01 7.91964352e-01
4.14153218e-01 2.69135475e-01 3.22492450e-01 5.91879010e-01
-6.58769667e-01 5.51797926e-01 -4.13294703e-01 9.35406983e-02
5.31130791e-01 -8.99799466e-01 3.42646182e-01 1.10144448e+00
-2.21215278e-01 4.12093222e-01 2.21085623e-01 -3.01963598e-01
-1.73700190e+00 -1.05615854e+00 4.93359804e-01 -5.96137755e-02
3.11479032e-01 -4.30673063e-01 -6.71079814e-01 2.80252695e-01
-6.89935088e-02 5.64847648e-01 2.28618488e-01 -5.08382618e-01
-3.68904799e-01 -4.61150497e-01 -8.38383913e-01 3.88739198e-01
1.27419090e+00 -5.82007349e-01 -3.85644883e-01 -1.59552380e-01
1.02353740e+00 -5.89763761e-01 -6.42022431e-01 5.71582854e-01
4.96001720e-01 -1.46139228e+00 1.06803417e+00 2.67028213e-01
7.51714826e-01 -6.65532470e-01 -3.53263646e-01 -6.50267243e-01
-1.24283984e-01 -2.60113686e-01 -4.06691879e-01 1.16100514e+00
-9.15610343e-02 -4.63324159e-01 6.89502478e-01 4.33088362e-01
-4.26143020e-01 -1.25474989e+00 -9.73197460e-01 -2.73459733e-01
-5.02614617e-01 -7.16514230e-01 6.04209721e-01 4.15036827e-01
-1.40476331e-01 1.74810857e-01 -5.30416518e-02 3.38998228e-01
5.60043097e-01 3.47588003e-01 6.19296968e-01 -7.53737390e-01
-4.04926956e-01 -4.15449619e-01 -5.53098321e-01 -1.82379127e+00
-3.35342765e-01 -4.83729601e-01 1.81873634e-01 -1.58117402e+00
-6.93101436e-02 -5.44152796e-01 -4.84172046e-01 5.83402276e-01
-2.72614151e-01 4.08023268e-01 3.24558705e-01 1.46074474e-01
-7.90512323e-01 8.55110109e-01 1.38958478e+00 8.86983648e-02
-1.02629758e-01 -3.29172909e-01 -9.58377004e-01 8.46642375e-01
3.45637769e-01 -2.92275578e-01 -4.70628917e-01 -8.49842429e-01
1.00464620e-01 3.87829170e-02 5.67304671e-01 -1.12531209e+00
4.69412953e-01 3.75993326e-02 7.53238797e-01 -8.46290350e-01
5.12648702e-01 -7.74271846e-01 -5.34954846e-01 3.81113231e-01
1.51215523e-01 -5.27198017e-02 1.76347777e-01 6.35215044e-01
-2.80174285e-01 2.63381690e-01 7.89036572e-01 -4.92823794e-02
-1.13504672e+00 5.70457518e-01 1.86709404e-01 -2.30192870e-01
1.16651845e+00 -3.96979660e-01 -3.94061208e-01 -2.26110406e-02
-2.47855231e-01 3.79314542e-01 6.31854832e-01 3.36552769e-01
9.22303438e-01 -1.27188277e+00 -3.70841593e-01 5.08853197e-01
5.91770336e-02 4.33789015e-01 5.72922647e-01 1.03971517e+00
-5.42717457e-01 2.02563837e-01 -2.43610829e-01 -7.36150801e-01
-6.78927422e-01 4.92702663e-01 3.13595861e-01 1.20589063e-02
-8.22747648e-01 1.22817338e+00 5.70151925e-01 3.37468475e-01
3.40865731e-01 -4.49548036e-01 1.31246410e-02 7.98240080e-02
8.46484959e-01 2.18207046e-01 1.40409827e-01 -5.02672434e-01
-5.84863186e-01 6.84886277e-01 -2.74746981e-03 -8.51757228e-02
1.35460353e+00 -5.54029465e-01 3.92474085e-02 2.73356646e-01
1.38698936e+00 -2.14102700e-01 -1.88033903e+00 -4.57019895e-01
-6.13191485e-01 -7.66627073e-01 3.95304859e-01 -5.48920035e-01
-1.39767587e+00 1.23846924e+00 8.03118944e-01 -1.09468460e-01
1.65685499e+00 1.32207351e-03 1.07033646e+00 4.58245864e-03
4.68317628e-01 -7.73900092e-01 2.42473915e-01 2.62895107e-01
5.71249247e-01 -1.30183160e+00 4.88988161e-02 -4.89587158e-01
-5.41832983e-01 1.04151881e+00 8.57264698e-01 -1.07504174e-01
5.78063965e-01 3.48393321e-01 -7.96983391e-02 -1.05011731e-01
-6.93245828e-01 -6.41712725e-01 1.07965544e-01 4.95372117e-01
1.47829384e-01 -3.58097106e-01 8.71038958e-02 7.90790081e-01
1.23123102e-01 1.20854050e-01 1.92576081e-01 1.07303834e+00
-5.85050106e-01 -7.15522826e-01 -8.09309632e-02 2.02754289e-01
-3.30453277e-01 -1.32659107e-01 9.64747444e-02 5.75458407e-01
5.08103311e-01 8.89250875e-01 1.07511587e-01 -6.53761029e-01
4.30305421e-01 -3.08481634e-01 4.68897671e-01 -4.52128917e-01
-3.95168006e-01 2.18687877e-01 -3.01401407e-01 -1.10672951e+00
-4.82534170e-01 -4.98494446e-01 -1.41836488e+00 -2.91599423e-01
-2.72889346e-01 -4.35368240e-01 6.19447589e-01 9.53480542e-01
6.07110739e-01 7.00841546e-01 7.77347088e-01 -1.27240479e+00
-2.71235049e-01 -7.65845358e-01 -5.40074825e-01 6.16545454e-02
5.40546000e-01 -8.22706878e-01 -3.02573532e-01 -2.61958450e-01]
|
[9.601926803588867, -0.8941361904144287]
|
d163a666-a7c9-46f9-b51c-732bfa8a4fe8
|
matching-with-transformers-in-melt
|
2109.07401
| null |
https://arxiv.org/abs/2109.07401v1
|
https://arxiv.org/pdf/2109.07401v1.pdf
|
Matching with Transformers in MELT
|
One of the strongest signals for automated matching of ontologies and knowledge graphs are the textual descriptions of the concepts. The methods that are typically applied (such as character- or token-based comparisons) are relatively simple, and therefore do not capture the actual meaning of the texts. With the rise of transformer-based language models, text comparison based on meaning (rather than lexical features) is possible. In this paper, we model the ontology matching task as classification problem and present approaches based on transformer models. We further provide an easy to use implementation in the MELT framework which is suited for ontology and knowledge graph matching. We show that a transformer-based filter helps to choose the correct correspondences given a high-recall alignment and already achieves a good result with simple alignment post-processing methods.
|
['Heiko Paulheim', 'Jan Portisch', 'Sven Hertling']
|
2021-09-15
| null | null | null | null |
['ontology-matching']
|
['knowledge-base']
|
[ 2.08485961e-01 2.77202781e-02 -3.10006589e-01 -4.17242736e-01
-3.38193625e-01 -5.56158364e-01 1.02862930e+00 9.52533901e-01
-5.98988891e-01 1.95533499e-01 3.22058350e-01 -3.91386241e-01
-6.80295527e-01 -1.27122438e+00 -1.31520107e-01 -5.48124239e-02
2.35829204e-01 9.65538740e-01 5.83618402e-01 -6.20364249e-01
2.44664401e-01 3.18644941e-01 -2.03638697e+00 7.16499448e-01
6.98473215e-01 8.30321074e-01 1.52467981e-01 3.15202117e-01
-1.08060980e+00 8.26160789e-01 -2.80015111e-01 -8.20162237e-01
-3.34662832e-02 -4.45266783e-01 -1.23207772e+00 -3.20250481e-01
4.93121207e-01 3.64171058e-01 -2.04032972e-01 1.17427599e+00
1.01063870e-01 -8.73150826e-02 6.63592994e-01 -1.17548621e+00
-2.09387466e-01 9.27096009e-01 2.09357385e-02 -2.70739309e-02
1.06077039e+00 -6.72595561e-01 1.27158368e+00 -8.00791800e-01
9.03749824e-01 1.42398226e+00 7.67839611e-01 1.27982289e-01
-1.15634871e+00 -2.68247545e-01 -2.16634840e-01 6.40651882e-01
-1.37622416e+00 -5.32057345e-01 4.21580851e-01 -5.00640273e-01
1.14520788e+00 5.64317584e-01 5.62486529e-01 6.25418186e-01
4.73478101e-02 3.22394252e-01 1.04851317e+00 -1.08941364e+00
3.74770239e-02 3.03703845e-01 3.85602564e-01 6.48105383e-01
3.05248678e-01 -2.87761748e-01 -4.54816967e-01 -3.18568975e-01
3.55432123e-01 -1.72035903e-01 -9.90120322e-02 -5.09288073e-01
-1.13130164e+00 7.07186341e-01 5.13457842e-02 1.08668494e+00
-1.95321634e-01 -2.48101249e-01 6.37075365e-01 8.14588308e-01
2.61774182e-01 4.68345195e-01 -3.06775719e-01 2.64889058e-02
-1.00225055e+00 4.59572077e-01 1.06475186e+00 1.11923957e+00
8.98026407e-01 -6.23189867e-01 -2.03315094e-02 7.65641451e-01
4.38720256e-01 2.77070552e-01 6.53056383e-01 -5.12456536e-01
2.44447425e-01 9.49604034e-01 -1.58516467e-02 -1.30434871e+00
-3.96427095e-01 -2.20896170e-01 -2.21271619e-01 1.23736691e-02
7.14033961e-01 6.08977377e-01 -5.49352050e-01 1.42216814e+00
2.08907261e-01 -1.90469772e-01 2.04616249e-01 3.35844696e-01
7.91106880e-01 6.18110448e-02 1.80035129e-01 -2.42709100e-01
1.78138340e+00 -3.02648932e-01 -8.69621515e-01 -2.26666451e-01
1.15255845e+00 -1.00008416e+00 9.48112428e-01 2.08516475e-02
-7.97676682e-01 -2.80927479e-01 -8.83288324e-01 -3.33399512e-02
-9.56618190e-01 -3.48654181e-01 4.33262229e-01 7.42179036e-01
-8.43253195e-01 9.22229528e-01 -3.94070297e-01 -1.11060739e+00
-1.62763819e-01 2.70253599e-01 -6.02138042e-01 -9.34009701e-02
-1.29098034e+00 1.39392591e+00 9.37174976e-01 -3.44442368e-01
2.10260674e-01 -3.86503756e-01 -8.44538093e-01 4.22076099e-02
4.57318544e-01 -7.62001336e-01 1.13314044e+00 -1.10620761e+00
-1.16902900e+00 1.28156435e+00 1.13785481e-02 -5.31665027e-01
4.43892837e-01 2.36593232e-01 -6.49058819e-01 2.10999325e-02
7.81886876e-02 1.46685332e-01 3.73862743e-01 -9.07936454e-01
-8.31718862e-01 -4.94033605e-01 1.90530404e-01 -2.45406870e-02
-5.14691889e-01 5.34168661e-01 -2.83556908e-01 -3.83552074e-01
1.71611950e-01 -6.01971388e-01 1.09506324e-01 -2.84656078e-01
4.46850732e-02 -3.85511249e-01 5.18497825e-01 -6.83259130e-01
1.58959436e+00 -1.70125103e+00 -2.95205452e-02 6.05569899e-01
2.36923054e-01 7.15548471e-02 -4.11574505e-02 9.90917444e-01
-1.89512953e-01 9.53246281e-02 -1.42195374e-01 3.17624584e-02
3.69846135e-01 4.60826308e-01 -2.02546895e-01 2.35004768e-01
-1.81481332e-01 7.33078480e-01 -9.81787503e-01 -7.68746793e-01
2.45488092e-01 1.32338285e-01 -2.35905573e-01 1.33848644e-03
-2.39454940e-01 -2.67372072e-01 -4.93421018e-01 2.69703090e-01
1.86671451e-01 7.77701214e-02 8.58671546e-01 -4.40272421e-01
-1.66243538e-01 7.29076564e-01 -1.30061412e+00 1.67578530e+00
-5.89912355e-01 5.19637406e-01 -3.54265094e-01 -1.13155711e+00
1.08288944e+00 4.30627435e-01 3.49669248e-01 -8.45544875e-01
1.27117395e-01 5.00677288e-01 -4.39526001e-03 -5.16469121e-01
6.92152739e-01 -3.10123175e-01 -7.94722661e-02 2.54139364e-01
4.31055613e-02 -1.17460907e-01 5.94121337e-01 1.22373521e-01
1.06564927e+00 2.50637174e-01 8.93443108e-01 -6.58850014e-01
7.53706336e-01 3.04746091e-01 3.32077742e-01 5.63915849e-01
6.19732738e-01 1.94194630e-01 4.17714030e-01 -5.32124221e-01
-1.08656549e+00 -4.80541587e-01 -3.23464602e-01 9.25994456e-01
-6.75617158e-02 -1.17559099e+00 -6.15769625e-01 -6.03028238e-01
1.02538019e-01 7.73531914e-01 -3.73772651e-01 1.15438730e-01
-4.41029936e-01 -5.07666349e-01 5.80667078e-01 2.73574919e-01
1.95985381e-03 -7.67972529e-01 -5.24313033e-01 4.97675329e-01
-1.86649740e-01 -1.26559317e+00 1.57007322e-01 4.25923876e-02
-8.50284159e-01 -1.17840028e+00 -2.06489023e-02 -6.23462677e-01
4.06678885e-01 -1.29837319e-01 1.38013232e+00 3.47864777e-01
-6.50024740e-04 3.22368234e-01 -6.15148067e-01 -3.23196888e-01
-8.03149045e-01 8.68042856e-02 -1.19337298e-01 -8.20142552e-02
7.82706797e-01 -6.04794264e-01 1.65548161e-01 2.92244256e-01
-1.00800478e+00 -1.08943552e-01 2.35900611e-01 7.26331651e-01
2.65937477e-01 1.05597757e-01 2.06841957e-02 -1.16372597e+00
7.26241469e-01 -1.88461527e-01 -4.79360014e-01 6.74337506e-01
-9.28795695e-01 4.21207845e-01 4.88649786e-01 -1.95324302e-01
-7.02969670e-01 -3.08694709e-02 -2.49978200e-01 2.86913291e-02
-1.75794169e-01 1.00882757e+00 -2.34020159e-01 -1.25887260e-01
7.00393319e-01 1.25699220e-02 -8.59238654e-02 -7.69371510e-01
3.97740364e-01 7.79643774e-01 4.14791077e-01 -8.37476015e-01
7.89384127e-01 2.65595019e-01 3.47036533e-02 -7.56081164e-01
-5.43704093e-01 -6.18879914e-01 -8.35230529e-01 -1.28981650e-01
5.56617796e-01 -6.48020864e-01 -5.10047376e-01 -2.47341814e-04
-1.19406152e+00 1.13513656e-01 -4.41169977e-01 3.78650904e-01
-5.21009803e-01 6.97447956e-01 -1.80044860e-01 -7.60037184e-01
-2.00687662e-01 -6.78253829e-01 8.02024126e-01 -1.12264089e-01
-6.56582117e-01 -1.11847472e+00 1.64370701e-01 1.33777544e-01
3.63542765e-01 2.35919766e-02 1.46199787e+00 -1.17466509e+00
-8.74481350e-02 -4.43028241e-01 -1.05605498e-01 -1.59109220e-01
6.61120117e-02 9.66304839e-02 -7.71004438e-01 -8.00446123e-02
-2.64750570e-01 2.34912053e-01 3.44800651e-01 -3.57431591e-01
5.57559431e-01 -1.72743171e-01 -5.42869925e-01 5.19004799e-02
1.65709031e+00 -1.13402642e-02 9.10962939e-01 6.89237833e-01
4.94512469e-01 1.00322139e+00 5.36418915e-01 1.88882262e-01
4.39380825e-01 1.30343664e+00 7.14251772e-02 1.72197521e-01
-2.46865880e-02 -3.19851756e-01 6.53041154e-02 9.42020118e-01
-8.30418896e-03 -9.55585539e-02 -1.25528967e+00 5.75957835e-01
-2.18676448e+00 -1.13107407e+00 -5.60214162e-01 2.45823669e+00
7.99389482e-01 2.89141327e-01 1.72079191e-01 3.80008638e-01
5.67407966e-01 -1.97642267e-01 4.61152583e-01 -4.24026728e-01
-1.82470620e-01 4.26933616e-01 4.95953053e-01 6.20326936e-01
-7.06348062e-01 7.82722473e-01 6.42413378e+00 8.73376727e-01
-6.26108646e-01 -4.99229729e-02 -3.02740544e-01 5.47597885e-01
-5.46006024e-01 4.21446681e-01 -6.42821670e-01 3.20333481e-01
1.02598882e+00 -4.52510029e-01 1.97913498e-01 5.75489879e-01
-1.48101762e-01 3.14113759e-02 -1.39831817e+00 9.89117861e-01
-5.34642152e-02 -1.31385744e+00 3.40893149e-01 -8.12896416e-02
-1.90671082e-04 -1.96000993e-01 -7.72596836e-01 3.48547399e-02
3.03226650e-01 -8.34348261e-01 9.03513908e-01 7.46121407e-01
5.26494145e-01 -4.84928012e-01 1.03763819e+00 8.02645013e-02
-1.42366374e+00 4.82618809e-03 -3.57330352e-01 -1.16232984e-01
-7.94274136e-02 5.11831403e-01 -4.93919849e-01 1.18988216e+00
5.15829802e-01 6.85145974e-01 -6.18457735e-01 1.07191777e+00
-1.33781657e-01 1.12177148e-01 -4.11951005e-01 -6.93888366e-02
-1.11547321e-01 -2.12477759e-01 5.42944551e-01 1.54928744e+00
3.59904557e-01 -2.47054204e-01 3.18546712e-01 5.62064469e-01
2.81320661e-01 7.79115677e-01 -6.43423796e-01 -1.18420616e-01
5.93610704e-01 1.15270662e+00 -6.57747686e-01 -5.86529255e-01
-6.34804964e-01 5.55838525e-01 2.69389093e-01 -2.02578217e-01
-2.27882162e-01 -6.49644434e-01 6.52713180e-01 4.19666171e-01
3.48082148e-02 -6.80385455e-02 7.59012550e-02 -1.12672675e+00
1.34197235e-01 -1.17779624e+00 7.00593412e-01 -7.42207050e-01
-1.30412841e+00 6.03457451e-01 4.13642645e-01 -1.16487038e+00
-5.50088286e-01 -8.30361068e-01 -3.37413073e-01 8.28500330e-01
-1.25624001e+00 -1.12468779e+00 -1.83391422e-01 5.34605384e-01
1.01339836e-02 1.20358188e-02 1.23817790e+00 5.19898593e-01
-6.76144753e-03 2.27835640e-01 -2.40612458e-02 2.81858802e-01
6.18027389e-01 -1.28583574e+00 3.18784684e-01 6.96504414e-01
5.55471659e-01 8.58170033e-01 1.17757535e+00 -5.56277931e-01
-1.29617071e+00 -5.10834038e-01 1.87238348e+00 -4.45260108e-01
1.08817923e+00 -2.23990276e-01 -1.11453176e+00 5.33730328e-01
2.21731633e-01 -3.88851821e-01 6.88353360e-01 2.37857774e-01
-6.53134584e-01 -1.49861351e-01 -9.16346073e-01 3.91651124e-01
1.17466438e+00 -8.75310421e-01 -1.21985722e+00 2.33039036e-01
2.01144159e-01 -1.44754440e-01 -1.20504379e+00 2.95698166e-01
7.76516438e-01 -8.04545999e-01 6.91480875e-01 -7.88825750e-01
-8.99326578e-02 -4.03153449e-01 -3.00323814e-01 -9.49069738e-01
-2.24832565e-01 -4.15816277e-01 5.42068303e-01 1.35822439e+00
4.86759931e-01 -7.36588061e-01 5.11301219e-01 4.28082168e-01
3.08004677e-01 2.32995804e-02 -8.65212202e-01 -9.40245628e-01
-2.60358036e-01 -5.37701309e-01 9.28068995e-01 1.30362093e+00
8.04356277e-01 3.79252285e-01 5.82719669e-02 -8.87047723e-02
2.92745560e-01 2.50744730e-01 6.56424284e-01 -1.89276123e+00
-1.79354012e-01 -6.26810908e-01 -1.01817858e+00 -4.66579258e-01
2.41246432e-01 -1.23472357e+00 -2.79337794e-01 -1.66871095e+00
4.59182523e-02 -3.71834636e-01 -1.40009776e-01 6.64445639e-01
2.44647413e-01 -3.47358473e-02 6.22667335e-02 2.31162295e-01
-3.29965711e-01 1.04597114e-01 3.56233507e-01 -1.26447245e-01
2.21244693e-01 -3.85587603e-01 -2.96143025e-01 5.70979118e-01
6.11764967e-01 -8.84801805e-01 -2.35550746e-01 -2.18532592e-01
7.64902890e-01 -1.04288176e-01 2.10906401e-01 -8.49706590e-01
4.96463031e-01 -6.45249337e-02 -1.78585127e-01 -9.38281938e-02
-1.34577408e-01 -1.12539327e+00 6.54390156e-01 5.03240526e-01
-3.79111558e-01 5.00066876e-01 9.75275505e-03 3.26633692e-01
-3.98494124e-01 -7.33438492e-01 4.49523538e-01 -2.33202532e-01
-7.57326543e-01 -1.42329559e-01 -4.66734141e-01 -8.91181901e-02
4.26428348e-01 -3.66460949e-01 -1.51449233e-01 -1.85601354e-01
-5.90079069e-01 -5.74828945e-02 6.77503645e-01 5.51555097e-01
2.11011127e-01 -1.30095673e+00 -6.88136399e-01 -1.56957731e-01
6.20180309e-01 -7.02714086e-01 -4.93201375e-01 7.64386714e-01
-3.80204320e-01 4.24388081e-01 -3.77054483e-01 -2.37208202e-01
-1.76162195e+00 5.62303782e-01 4.82340574e-01 -4.43098962e-01
-6.54521942e-01 4.40553539e-02 -3.16939443e-01 -4.81813848e-01
-1.28960505e-01 -2.20859692e-01 -5.78904569e-01 4.82902825e-01
4.99799103e-01 2.22934917e-01 6.39695704e-01 -7.86147714e-01
-6.90323293e-01 7.82448530e-01 1.19257547e-01 -7.40254596e-02
1.27375245e+00 -1.35730812e-02 -5.90340495e-01 3.33575517e-01
7.79480875e-01 1.25043049e-01 7.04567507e-02 -6.15351558e-01
9.47467983e-01 -5.34609616e-01 -2.44112704e-02 -4.72551346e-01
-5.08603454e-01 5.80602467e-01 4.05590385e-01 6.62899435e-01
8.02194238e-01 -6.54446334e-02 3.35293025e-01 4.80653793e-01
4.25598234e-01 -1.05996835e+00 -5.48702180e-01 5.18923640e-01
5.42332232e-01 -7.21724987e-01 1.11586288e-01 -7.71954656e-01
3.37711461e-02 1.65889120e+00 1.15624806e-02 2.94959426e-01
3.37955296e-01 4.13025975e-01 8.88263881e-02 -5.03772736e-01
-7.05920994e-01 -7.83598125e-01 4.91823226e-01 6.58111691e-01
7.26695538e-01 -1.45280808e-01 -1.02491844e+00 2.72645980e-01
-3.03597212e-01 2.26352196e-02 3.54048222e-01 9.38009977e-01
-5.80254018e-01 -1.86388922e+00 -2.66121835e-01 4.21021193e-01
-4.70481962e-01 -2.32023641e-01 -6.33637130e-01 9.83108044e-01
-3.75415012e-02 9.41783547e-01 8.81504640e-02 -5.01767218e-01
5.93076468e-01 6.53907835e-01 7.69290626e-01 -6.44656479e-01
-9.39940810e-01 -2.71833301e-01 8.23739290e-01 -4.97691751e-01
-6.90821826e-01 -7.04096854e-01 -1.07194555e+00 -4.50024456e-01
-4.81936216e-01 4.29294854e-01 6.04658842e-01 1.37506270e+00
1.16968945e-01 1.56550422e-01 1.06000327e-01 -1.67463988e-01
-4.23860669e-01 -7.49419391e-01 -4.91067886e-01 6.55614078e-01
-4.60259795e-01 -5.43778539e-01 -2.56762467e-02 1.03519291e-01]
|
[9.312644958496094, 8.151619911193848]
|
b59772dd-5b61-4f32-8a99-a63d8c289a03
|
190406726
|
1904.06726
| null |
http://arxiv.org/abs/1904.06726v1
|
http://arxiv.org/pdf/1904.06726v1.pdf
|
VORNet: Spatio-temporally Consistent Video Inpainting for Object Removal
|
Video object removal is a challenging task in video processing that often
requires massive human efforts. Given the mask of the foreground object in each
frame, the goal is to complete (inpaint) the object region and generate a video
without the target object. While recently deep learning based methods have
achieved great success on the image inpainting task, they often lead to
inconsistent results between frames when applied to videos. In this work, we
propose a novel learning-based Video Object Removal Network (VORNet) to solve
the video object removal task in a spatio-temporally consistent manner, by
combining the optical flow warping and image-based inpainting model.
Experiments are done on our Synthesized Video Object Removal (SVOR) dataset
based on the YouTube-VOS video segmentation dataset, and both the objective and
subjective evaluation demonstrate that our VORNet generates more spatially and
temporally consistent videos compared with existing methods.
|
['Ya-Liang Chang', 'Winston Hsu', 'Zhe Yu Liu']
|
2019-04-14
| null | null | null | null |
['one-shot-visual-object-segmentation', 'video-inpainting']
|
['computer-vision', 'computer-vision']
|
[ 3.53644401e-01 -4.92710918e-01 -1.47903949e-01 8.07502121e-02
-4.75590348e-01 -3.77554297e-01 1.48540556e-01 -5.92047215e-01
-3.22795540e-01 7.93751121e-01 1.03064135e-01 1.11077003e-01
1.84609547e-01 -3.63538265e-01 -9.51246858e-01 -6.40100121e-01
1.11428142e-01 -8.82676467e-02 4.50361490e-01 2.03553692e-01
1.84257418e-01 3.79423469e-01 -1.28338492e+00 3.52247179e-01
8.65528286e-01 9.39623237e-01 4.86930847e-01 6.93717897e-01
3.65512967e-02 1.46963382e+00 -6.98012531e-01 -3.93047705e-02
6.68817878e-01 -7.04292417e-01 -7.36878753e-01 6.37564421e-01
8.93996298e-01 -8.23873818e-01 -9.65051711e-01 1.13139272e+00
2.12247282e-01 4.92828548e-01 2.68084049e-01 -1.36322773e+00
-6.92488253e-01 3.25151503e-01 -8.54400814e-01 5.26846468e-01
3.62673908e-01 2.96781600e-01 4.14409369e-01 -9.21683133e-01
9.08685565e-01 1.18487453e+00 4.48163599e-01 7.57311225e-01
-1.21026611e+00 -7.51030684e-01 2.73835599e-01 4.90027338e-01
-1.43192542e+00 -5.16691208e-01 9.24264848e-01 -3.92938823e-01
6.06608152e-01 2.67291725e-01 9.10726905e-01 9.11623955e-01
3.55753750e-01 1.12149823e+00 5.56338012e-01 -1.47879303e-01
2.33075589e-01 -4.33704287e-01 -3.77946168e-01 4.94959384e-01
1.36832312e-01 9.32206865e-04 -6.18037403e-01 2.21909404e-01
1.13994682e+00 4.28431690e-01 -6.73026860e-01 -2.10437417e-01
-1.07697320e+00 5.50247848e-01 2.14842960e-01 1.81134537e-01
-4.90818381e-01 4.69925553e-01 3.80485773e-01 3.68351042e-01
7.17809319e-01 1.41939759e-01 -2.58314252e-01 5.96000031e-02
-1.59490979e+00 5.54703653e-01 5.23929238e-01 1.14956176e+00
6.10079229e-01 5.36602318e-01 -3.61101836e-01 6.99107587e-01
1.26895592e-01 2.86729038e-01 3.81162256e-01 -1.51247072e+00
3.40168387e-01 1.77761182e-01 3.20794374e-01 -1.25985789e+00
2.63423532e-01 8.35847706e-02 -8.67878377e-01 3.96999538e-01
2.17395678e-01 -1.85291648e-01 -1.12935114e+00 1.56747651e+00
3.57881933e-01 9.37841594e-01 -9.47777703e-02 1.23247385e+00
9.82807577e-01 1.01753199e+00 1.21734053e-01 -6.87493920e-01
7.57286370e-01 -1.44233251e+00 -1.19334507e+00 -2.52572447e-01
1.61846176e-01 -7.64047861e-01 5.33551157e-01 4.62351888e-01
-1.41113377e+00 -6.34282291e-01 -9.26621079e-01 -3.16176087e-01
2.15905324e-01 -1.97131082e-01 2.42674068e-01 4.17414345e-02
-1.04860866e+00 5.50149441e-01 -8.07236373e-01 -6.01035058e-02
6.72921658e-01 2.08455682e-01 -5.26455641e-01 -2.43746623e-01
-8.73324394e-01 5.05792439e-01 5.86308420e-01 2.02444315e-01
-1.42050540e+00 -8.63969564e-01 -8.97952557e-01 1.03042394e-01
7.95194805e-01 -7.08001614e-01 1.19188297e+00 -1.59059298e+00
-1.34687364e+00 4.84894216e-01 -1.81553081e-01 -5.29481113e-01
8.27040613e-01 -5.79134166e-01 -2.22979501e-01 4.99405235e-01
2.22572371e-01 1.04713368e+00 1.49120557e+00 -1.31950164e+00
-7.66673625e-01 1.24795713e-01 -1.25264719e-01 3.01160097e-01
-9.83004719e-02 1.75934166e-01 -8.79624069e-01 -1.27927673e+00
-5.01127802e-02 -6.85814977e-01 -6.53188825e-02 3.51990789e-01
-2.97261119e-01 5.15114516e-02 1.60636878e+00 -1.19008529e+00
1.27447963e+00 -2.15990186e+00 5.00268817e-01 -4.27631646e-01
4.73212600e-01 6.03366017e-01 -4.32261080e-01 8.24068934e-02
-1.71310335e-01 -3.91019545e-02 -4.00478542e-01 -5.09394050e-01
-6.63038492e-01 2.35901758e-01 -5.24389207e-01 5.36136508e-01
1.87287673e-01 8.95016670e-01 -1.00683594e+00 -7.52241552e-01
5.33086717e-01 4.98903692e-01 -7.65208602e-01 3.73090804e-01
-4.61443782e-01 5.84915400e-01 -2.05787465e-01 7.37185478e-01
8.07484090e-01 1.47383109e-01 2.28297096e-02 -3.49312782e-01
-1.10961311e-01 -3.91350508e-01 -1.06408894e+00 1.86335933e+00
-1.87350288e-02 1.07508290e+00 4.25724059e-01 -7.63410866e-01
2.75659651e-01 2.75428027e-01 9.50540662e-01 -6.03498876e-01
1.52478263e-01 -8.68474916e-02 -1.25043958e-01 -9.66678202e-01
6.73341036e-01 1.62382387e-02 5.64460754e-01 2.26071045e-01
1.17241606e-01 -1.46827400e-01 5.09455025e-01 2.16573760e-01
9.96311128e-01 4.45792854e-01 -4.18464430e-02 -5.39196394e-02
4.39744830e-01 -1.31287187e-01 1.03851581e+00 6.25458181e-01
-4.19190288e-01 1.09529138e+00 2.22706109e-01 -6.53105974e-01
-1.09282529e+00 -8.54009926e-01 2.62041450e-01 6.51961386e-01
5.56013405e-01 -3.26937109e-01 -1.01662338e+00 -7.30448067e-01
-1.45027772e-01 4.75453109e-01 -3.51141661e-01 -1.51063964e-01
-1.00092041e+00 -2.93962419e-01 2.28781149e-01 4.62913603e-01
7.35854328e-01 -1.31185818e+00 -5.62691450e-01 3.34035039e-01
-5.01862824e-01 -1.41434085e+00 -1.06196308e+00 -5.39972126e-01
-8.75382900e-01 -1.22128677e+00 -9.34075356e-01 -1.11470485e+00
6.57448351e-01 8.05797160e-01 8.23429286e-01 3.56187552e-01
-4.06537086e-01 2.03777835e-01 -3.17229301e-01 -1.83835682e-02
-4.16248769e-01 -4.08358514e-01 1.93126872e-02 2.18777597e-01
-1.29416995e-02 -3.50251198e-01 -6.56980395e-01 2.02704340e-01
-1.58294249e+00 3.31770211e-01 2.62287050e-01 7.36171842e-01
6.42028570e-01 3.44124705e-01 3.60473216e-01 -5.79980791e-01
3.40818882e-01 -3.29465151e-01 -5.89141071e-01 1.28496319e-01
-8.41163546e-02 -5.10008335e-01 5.78271627e-01 -6.14299238e-01
-1.09076285e+00 9.18578282e-02 1.70874193e-01 -1.23358059e+00
7.33067691e-02 1.77802995e-01 -1.87159300e-01 -3.00921321e-01
1.96731135e-01 4.85501945e-01 6.08581416e-02 -4.01297152e-01
1.57113642e-01 3.43661368e-01 6.54621243e-01 -1.70488730e-01
7.98945665e-01 6.49959445e-01 -2.81244069e-01 -9.13577557e-01
-6.33489907e-01 -2.22555861e-01 -6.16872251e-01 -4.83039826e-01
1.13516331e+00 -1.10496092e+00 -3.69688153e-01 7.09724605e-01
-1.38531685e+00 -4.85830873e-01 -2.86739886e-01 4.76459771e-01
-6.28781915e-01 7.40291893e-01 -8.38428795e-01 -4.13128048e-01
-3.49387467e-01 -1.26174641e+00 9.90371644e-01 3.45550776e-01
-9.97585207e-02 -8.37912977e-01 -1.59259811e-01 3.82615387e-01
2.07296118e-01 3.98361921e-01 5.75992584e-01 8.11366662e-02
-1.13800037e+00 1.12635858e-01 -2.30447367e-01 6.94303393e-01
3.88462365e-01 2.17412502e-01 -4.82672095e-01 -4.39683080e-01
2.79644698e-01 -3.28354612e-02 1.15283787e+00 7.12276578e-01
1.38595092e+00 -4.52761889e-01 -2.62309015e-01 8.61001253e-01
1.44284320e+00 5.88268340e-01 8.58024478e-01 2.06410661e-01
1.13230550e+00 2.80223489e-01 7.02687740e-01 2.41854936e-01
-1.44374102e-01 6.96393728e-01 4.17488009e-01 -1.92093223e-01
-5.86814404e-01 -1.07182860e-01 4.78179336e-01 6.94730639e-01
-8.99609402e-02 -5.58749199e-01 -4.21171516e-01 5.31438768e-01
-2.20712090e+00 -1.30547357e+00 4.71663512e-02 1.92931259e+00
8.24870765e-01 -2.43044913e-01 -1.28969654e-01 -8.93977210e-02
9.37172651e-01 3.81792247e-01 -6.20598316e-01 6.94980696e-02
-1.59562677e-01 -1.55166954e-01 3.08218032e-01 6.38077796e-01
-1.26453888e+00 1.24178553e+00 6.38722038e+00 8.58501494e-01
-1.24019563e+00 2.15975776e-01 6.77028358e-01 -5.16014040e-01
4.31500338e-02 -1.40262246e-01 -3.54453802e-01 7.16029525e-01
2.58369982e-01 -1.69121981e-01 6.78858817e-01 5.32733500e-01
6.85599685e-01 -3.50954235e-01 -1.07778060e+00 1.38339531e+00
4.53903168e-01 -1.61619639e+00 3.11583489e-01 -3.51325035e-01
1.11793613e+00 -3.26204062e-01 -8.67815390e-02 -8.77252128e-03
-2.82353163e-01 -9.20919955e-01 1.11449730e+00 5.10971308e-01
6.49857044e-01 -6.81298077e-01 4.36710835e-01 1.03689529e-01
-1.12500203e+00 -1.39227882e-01 -2.26288870e-01 -3.13240141e-02
5.24263918e-01 4.18067932e-01 -2.83581167e-01 3.11049432e-01
9.51468647e-01 1.14732015e+00 -2.73130357e-01 1.33388698e+00
-1.21202931e-01 4.49472576e-01 1.22520782e-01 6.76014066e-01
2.01992676e-01 -5.07657230e-01 9.39121604e-01 9.54572320e-01
2.13913947e-01 2.47329384e-01 2.99867988e-01 1.13422859e+00
-4.27299947e-01 -1.38178930e-01 -6.28879309e-01 -1.30811453e-01
1.70983329e-01 1.18093777e+00 -8.56518090e-01 -5.16080081e-01
-5.06922126e-01 1.32676041e+00 1.00683812e-02 7.96579659e-01
-1.16729701e+00 -2.28242412e-01 8.43064427e-01 1.29678473e-01
4.74901289e-01 -3.43994260e-01 1.21743731e-01 -1.34183121e+00
1.76862076e-01 -8.99799764e-01 -1.66421216e-02 -1.03005314e+00
-1.00608897e+00 3.65758538e-01 4.94175591e-03 -1.38146567e+00
-1.67809382e-01 -1.88207403e-01 -6.00618124e-01 5.41725695e-01
-1.43265116e+00 -8.26590121e-01 -5.00099540e-01 6.88431799e-01
1.23747945e+00 -1.82324186e-01 5.20769022e-02 7.63164341e-01
-7.74186254e-01 1.01107895e-01 6.00359775e-02 2.81181842e-01
7.35048950e-01 -6.75222456e-01 1.76853031e-01 1.33341169e+00
-2.28268988e-02 2.85648942e-01 8.21765661e-01 -1.07921851e+00
-1.53694522e+00 -1.63429523e+00 3.46017510e-01 -1.27194062e-01
1.83894217e-01 -1.05946712e-01 -9.87004101e-01 8.75730276e-01
6.68817282e-01 2.88243622e-01 1.19606353e-01 -1.03485513e+00
1.46258220e-01 -5.63873239e-02 -1.16427827e+00 7.98004329e-01
1.14464808e+00 -2.61593789e-01 -4.63638425e-01 3.96525681e-01
9.10998583e-01 -6.10592306e-01 -3.89482588e-01 4.85900789e-01
2.41838858e-01 -8.00306320e-01 9.89373982e-01 -5.88874161e-01
7.50786722e-01 -7.77471602e-01 1.22737825e-01 -1.08422112e+00
-2.94971466e-01 -1.01339650e+00 -4.86467540e-01 1.11203539e+00
-4.49439496e-01 4.25018296e-02 8.73210788e-01 6.18219733e-01
-2.98059970e-01 -4.69818532e-01 -7.68083096e-01 -6.77981436e-01
-3.91654432e-01 -2.84783691e-01 9.76059213e-02 9.94375825e-01
-6.48603141e-01 -5.48328720e-02 -8.81117761e-01 -1.13954945e-02
6.75692797e-01 1.41469268e-02 7.45709002e-01 -6.33059621e-01
-1.56883806e-01 -2.76327044e-01 -4.18294489e-01 -1.09235573e+00
4.33352530e-01 -5.25600493e-01 3.18715423e-01 -1.52466750e+00
2.72046506e-01 1.48530871e-01 2.09772419e-02 3.83317471e-01
-2.91475862e-01 5.27955174e-01 4.64191258e-01 3.72031540e-01
-8.16799819e-01 6.22436762e-01 1.56106162e+00 -4.47346747e-01
-4.22534078e-01 -3.93070042e-01 -3.10142428e-01 8.03070843e-01
3.84889781e-01 -5.30619621e-01 -4.79236364e-01 -7.95249581e-01
-2.26526350e-01 2.49821275e-01 4.77811217e-01 -9.78343785e-01
2.34997854e-01 -3.80680263e-01 6.54169858e-01 -6.00849271e-01
2.53829956e-01 -9.11377311e-01 3.73798400e-01 5.35246372e-01
-1.26499787e-01 5.59702031e-02 4.02690619e-01 7.38275826e-01
-5.25373340e-01 -1.06147550e-01 9.17871952e-01 -2.12154537e-01
-1.18543959e+00 6.59008503e-01 -5.48401475e-01 1.16963133e-01
1.34041989e+00 -3.66937369e-01 -9.36839730e-03 -4.30823147e-01
-6.58122599e-01 1.61058515e-01 5.38623750e-01 6.56092286e-01
1.13562393e+00 -1.30256343e+00 -7.41373360e-01 3.03772151e-01
-4.68350261e-01 1.94911763e-01 4.71494406e-01 8.03949594e-01
-1.12982380e+00 8.46611261e-02 -4.29082096e-01 -5.93808353e-01
-1.37642217e+00 7.48768210e-01 3.41810346e-01 2.93545902e-01
-9.52870369e-01 8.66519034e-01 6.67587101e-01 4.20085967e-01
3.84756714e-01 -3.11210334e-01 -2.71525178e-02 -1.36258021e-01
8.68715882e-01 5.42798460e-01 -1.74157009e-01 -6.70807898e-01
-2.70571969e-02 4.08436000e-01 -3.31829876e-01 8.90423059e-02
1.29466283e+00 -3.35128039e-01 -3.00610781e-01 -2.36256011e-02
1.15586126e+00 -1.74786150e-01 -1.83590615e+00 -7.18929395e-02
-3.27698618e-01 -9.99474883e-01 8.86335745e-02 -2.53529102e-01
-1.68518746e+00 4.79152977e-01 5.02549231e-01 5.68269528e-02
1.27754796e+00 -4.25505936e-01 1.35495222e+00 3.25576924e-02
1.47257179e-01 -1.17646313e+00 3.86234313e-01 3.52408767e-01
1.02849281e+00 -1.14340031e+00 1.71908274e-01 -6.42132223e-01
-4.63994086e-01 1.13651228e+00 9.04291034e-01 -3.72129291e-01
3.64364028e-01 3.22383285e-01 5.17210476e-02 2.33239412e-01
-5.85420549e-01 2.04566494e-01 8.63401592e-02 5.45419633e-01
1.65939003e-01 -4.15243059e-01 -2.89321512e-01 1.75824106e-01
4.91863340e-01 3.48521680e-01 7.25381792e-01 1.05933046e+00
-3.30096662e-01 -8.55756223e-01 -4.72037405e-01 2.49602586e-01
-6.00930393e-01 -1.14192158e-01 -2.12097675e-01 6.83106661e-01
2.76701480e-01 8.13463151e-01 1.30277067e-01 -1.97296783e-01
-4.50506248e-03 -2.79791355e-01 6.60009503e-01 -5.25761783e-01
-3.05928200e-01 4.36994612e-01 -3.48291963e-01 -9.48964298e-01
-7.25277603e-01 -5.73149979e-01 -1.34199166e+00 -2.71102220e-01
-3.61594334e-02 -2.41695493e-01 1.86818063e-01 9.13831353e-01
3.49276513e-01 7.77429104e-01 3.81319255e-01 -1.35694528e+00
4.92982715e-02 -6.55116856e-01 -7.00066745e-01 7.20868528e-01
6.10367477e-01 -5.48849463e-01 -2.53505766e-01 7.62778461e-01]
|
[10.776630401611328, -1.2917225360870361]
|
f1d98b22-f073-46b9-af86-970c5c002c19
|
from-clickbait-to-fake-news-detection-an
| null | null |
https://aclanthology.org/W17-4215
|
https://aclanthology.org/W17-4215.pdf
|
From Clickbait to Fake News Detection: An Approach based on Detecting the Stance of Headlines to Articles
|
We present a system for the detection of the stance of headlines with regard to their corresponding article bodies. The approach can be applied in fake news, especially clickbait detection scenarios. The component is part of a larger platform for the curation of digital content; we consider veracity and relevancy an increasingly important part of curating online information. We want to contribute to the debate on how to deal with fake news and related online phenomena with technological means, by providing means to separate related from unrelated headlines and further classifying the related headlines. On a publicly available data set annotated for the stance of headlines with regard to their corresponding article bodies, we achieve a (weighted) accuracy score of 89.59.
|
['Peter Bourgonje', 'Georg Rehm', 'Julian Moreno Schneider']
|
2017-09-01
| null | null | null |
ws-2017-9
|
['rumour-detection', 'clickbait-detection']
|
['natural-language-processing', 'natural-language-processing']
|
[-2.41707996e-01 4.22372013e-01 -4.08314914e-01 1.23295531e-01
-9.22796369e-01 -9.74363208e-01 9.01133060e-01 8.29136968e-01
-3.94368410e-01 7.11690009e-01 4.89570260e-01 -3.44700813e-01
1.37651607e-01 -7.36379862e-01 -8.24270904e-01 -2.56312937e-01
5.05627096e-01 5.52873731e-01 7.93224037e-01 -6.38937652e-01
9.86660421e-01 3.73534858e-01 -1.32710421e+00 5.83548188e-01
7.51627445e-01 8.16014290e-01 -4.41385567e-01 4.76709872e-01
-9.44346562e-02 1.10072565e+00 -9.33511078e-01 -8.05752099e-01
-5.29888906e-02 -4.21647817e-01 -8.83445978e-01 -1.08051799e-01
7.61705041e-01 -1.85464621e-01 -1.84068874e-01 1.33388281e+00
-1.69757791e-02 -6.15772486e-01 5.68297088e-01 -1.08772707e+00
-3.57785612e-01 6.05353057e-01 -3.38279992e-01 8.10079873e-01
6.45737827e-01 -2.87387937e-01 1.14914477e+00 -6.41377628e-01
1.30924571e+00 9.57175732e-01 6.15583420e-01 4.38270718e-02
-8.09985816e-01 -1.53694972e-01 -4.13824439e-01 2.95249820e-01
-8.89340997e-01 -3.04904878e-01 4.64193344e-01 -8.64573658e-01
2.08809584e-01 6.40997469e-01 6.37848258e-01 1.37588358e+00
1.91095009e-01 7.56423712e-01 1.35733414e+00 -4.70914662e-01
9.22858417e-02 6.20429099e-01 6.67338073e-01 5.77138782e-01
8.69992852e-01 -4.09229040e-01 -5.90061426e-01 -5.46957254e-01
1.82163805e-01 -4.57212925e-01 -1.98439226e-01 1.81716070e-01
-1.11930692e+00 1.17250621e+00 3.99328053e-01 8.35619807e-01
-4.36356604e-01 1.58568248e-02 8.17598164e-01 4.91835833e-01
9.75985944e-01 8.83956015e-01 -2.09399685e-01 -2.23648638e-01
-1.06825972e+00 6.19884551e-01 1.31225514e+00 8.10379028e-01
2.62516141e-01 -6.10498071e-01 -3.17807347e-01 3.81866902e-01
7.88081288e-02 4.05549109e-01 4.04090613e-01 -6.33625507e-01
7.38093257e-01 6.83209360e-01 6.49775684e-01 -1.55612683e+00
-1.80345133e-01 -6.72468483e-01 -9.94079262e-02 -4.18236256e-01
6.70922220e-01 -2.93816812e-02 -2.76601374e-01 6.83029354e-01
4.33202744e-01 -3.85174692e-01 -4.70676035e-01 9.01135743e-01
7.33281910e-01 4.56806093e-01 -2.81789124e-01 -2.19819859e-01
1.83800721e+00 -6.63600445e-01 -1.16533303e+00 2.02135723e-02
9.68775690e-01 -1.29950178e+00 7.95885682e-01 5.48943937e-01
-7.10830033e-01 2.01931313e-01 -1.10982656e+00 -1.41736612e-01
-8.97044122e-01 -5.91536202e-02 1.91243172e-01 5.53169250e-01
-3.32324684e-01 6.47659659e-01 -2.53001124e-01 -3.90381873e-01
4.61483628e-01 -5.43908060e-01 -1.66244179e-01 2.98362285e-01
-1.36116648e+00 1.42737353e+00 2.24324375e-01 -3.13800067e-01
-3.78516883e-01 -3.74516994e-01 -2.12852523e-01 -3.73488098e-01
5.58348298e-01 -2.19686087e-02 1.13293898e+00 -7.49372602e-01
-6.70030832e-01 1.27078772e+00 3.34052861e-01 -7.76649356e-01
1.18009186e+00 -3.57106149e-01 -9.05040681e-01 3.29923451e-01
5.89125991e-01 -4.52259451e-01 1.14280975e+00 -8.94594133e-01
-7.30089724e-01 -2.41031215e-01 -3.07401508e-01 -4.57541555e-01
-3.38017702e-01 4.18205857e-01 2.17383504e-02 -7.37999320e-01
2.03536615e-01 -7.69738078e-01 5.20688474e-01 -3.38954806e-01
-7.42129564e-01 -3.79280984e-01 9.79240358e-01 -1.39383626e+00
1.64781487e+00 -1.72368932e+00 -2.39293918e-01 1.88953504e-01
7.14325964e-01 3.68569613e-01 4.60703254e-01 8.86619627e-01
4.07700658e-01 4.38444048e-01 3.57130170e-01 2.85015196e-01
1.74584150e-01 -2.89315760e-01 -6.02169573e-01 9.94175851e-01
-1.78591609e-01 8.90434206e-01 -1.05002451e+00 -3.84966254e-01
-3.57529104e-01 -1.20861746e-01 -2.04269901e-01 -2.11447999e-01
-2.84433752e-01 4.53858711e-02 -7.37354577e-01 7.04212189e-01
4.38563108e-01 -4.01440889e-01 -1.19108953e-01 -2.54261166e-01
-4.34719265e-01 8.28259945e-01 -6.23612881e-01 4.42947328e-01
3.77218053e-02 1.23723245e+00 5.08304639e-03 -5.38620412e-01
7.63258696e-01 2.35455334e-01 1.10529676e-01 -5.28192937e-01
4.07666206e-01 7.93887675e-01 -1.98027894e-01 -6.85724258e-01
8.85675430e-01 2.69137889e-01 -2.86021739e-01 5.70821345e-01
-1.80904031e-01 3.75357270e-02 3.13423812e-01 5.63675106e-01
1.14081860e+00 -3.16622496e-01 4.27955419e-01 -2.55660623e-01
4.35872227e-01 4.41420883e-01 -2.00007170e-01 6.78441882e-01
-4.28045124e-01 2.04217777e-01 1.01946580e+00 -6.58951104e-01
-1.49220192e+00 -2.88291663e-01 -2.53102124e-01 8.49674582e-01
-3.54480483e-02 -3.87930393e-01 -8.66509557e-01 -1.07145035e+00
2.78637886e-01 8.50582540e-01 -7.90253997e-01 9.42433476e-02
-4.65296805e-01 -5.33824444e-01 6.47917986e-01 -3.19322407e-01
4.30903494e-01 -5.18870413e-01 -5.38452923e-01 3.64194334e-01
-7.19142437e-01 -1.31670165e+00 -1.21409424e-01 -2.00412080e-01
-3.91604334e-01 -1.45836818e+00 -5.38258314e-01 -3.44072819e-01
3.64778817e-01 2.17941210e-01 7.91557908e-01 4.55640435e-01
2.01340437e-01 -6.90507889e-02 -6.27181292e-01 -2.95459747e-01
-1.08533752e+00 1.60131976e-01 -1.83236122e-01 -7.61699006e-02
3.32907230e-01 6.95179328e-02 -3.01117450e-01 5.73935986e-01
-9.39545989e-01 -3.98704797e-01 7.90998563e-02 5.92312574e-01
2.09463527e-03 -3.01480889e-01 4.74265903e-01 -1.42084873e+00
1.07953680e+00 -8.17980289e-01 -6.97279334e-01 4.60760817e-02
-7.54780889e-01 -3.13437551e-01 4.08302039e-01 -2.06992298e-01
-6.23083353e-01 -6.77027225e-01 -1.94287241e-01 2.83901334e-01
1.58496350e-01 5.94696939e-01 5.50194204e-01 -3.45006406e-01
1.07485747e+00 -1.66690201e-01 -4.08959277e-02 -7.02088714e-01
4.45824742e-01 1.10809469e+00 2.95971662e-01 6.48048893e-03
1.00506663e+00 6.10796213e-01 -1.15359560e-01 -7.67090142e-01
-1.49276805e+00 -1.05969405e+00 -3.83568823e-01 -4.60300118e-01
3.93702328e-01 -6.62600398e-01 -4.80078667e-01 1.99230984e-01
-1.34881246e+00 4.97337788e-01 6.61504269e-02 1.62551060e-01
-2.35983476e-01 7.29154646e-01 -9.16729212e-01 -4.27229285e-01
-2.45859951e-01 -5.28356135e-01 7.62549281e-01 -3.66349846e-01
-6.44776285e-01 -8.80912423e-01 2.46624485e-01 1.03960240e+00
4.64457870e-01 5.97336948e-01 4.70474869e-01 -1.46019423e+00
-1.02388740e-01 -1.05828345e+00 -2.93641537e-01 2.95080036e-01
-2.70945668e-01 1.76109567e-01 -6.58898175e-01 -2.50045732e-02
1.88020304e-01 -1.32184178e-01 6.34155333e-01 -2.23441124e-01
4.21745896e-01 -1.22598267e+00 -3.72520506e-01 -3.63171756e-01
1.18390489e+00 -5.78989923e-01 6.20547712e-01 1.11183834e+00
4.29194957e-01 8.97387207e-01 9.27039444e-01 6.21991158e-01
-3.58672440e-02 7.26293445e-01 5.35724103e-01 4.01907325e-01
-1.52515620e-01 -6.02410257e-01 4.98421416e-02 8.59244525e-01
1.89212095e-02 -3.84793371e-01 -7.79909968e-01 5.58418453e-01
-1.79986250e+00 -1.22231007e+00 -1.20971859e+00 1.90638924e+00
5.43015361e-01 5.54405391e-01 4.93648976e-01 2.90647984e-01
1.06743586e+00 -1.49886236e-02 1.54146791e-01 -2.81625599e-01
-1.86931714e-01 -4.61384624e-01 1.18206024e+00 4.44943666e-01
-1.03149188e+00 8.68646204e-01 6.21699429e+00 9.64326382e-01
-8.84755194e-01 5.24786294e-01 2.55105436e-01 3.80977035e-01
-6.71111494e-02 -7.78232981e-03 -1.00560617e+00 9.91553783e-01
9.58306253e-01 -7.18858317e-02 1.30951963e-03 9.52375472e-01
5.56233883e-01 -1.68807477e-01 -5.12080073e-01 3.99038792e-01
4.16286439e-01 -1.85883510e+00 -3.10250193e-01 3.00370902e-01
6.52931571e-01 1.33823603e-01 -2.82302350e-01 -9.83416438e-02
7.40461703e-03 -2.69618809e-01 1.23055267e+00 4.08678263e-01
2.60878503e-01 -1.74435511e-01 1.18049788e+00 4.40695524e-01
-6.22917786e-02 2.06671521e-01 -7.44149759e-02 -5.20663820e-02
2.37459227e-01 9.71142054e-01 -1.08905745e+00 -1.36398710e-02
4.48618025e-01 7.73090899e-01 -8.85775268e-01 1.22391295e+00
-5.50862730e-01 7.79069126e-01 -1.38775274e-01 -6.61424041e-01
2.79778302e-01 3.79378386e-02 9.78410125e-01 1.18428648e+00
1.38537958e-01 -4.62102532e-01 -3.67199630e-01 7.18308747e-01
-2.38337725e-01 2.20598921e-01 -3.69647890e-01 -6.07441723e-01
4.93558943e-01 1.29499340e+00 -1.05249310e+00 -6.96866870e-01
1.68285915e-03 8.23976219e-01 2.04597801e-01 -1.96546048e-01
-9.71093476e-01 -3.46079886e-01 -3.27078477e-02 8.95757556e-01
3.17680925e-01 -7.21817371e-03 -1.20751068e-01 -1.21113968e+00
6.52105436e-02 -1.03905547e+00 3.53323519e-01 -6.99423611e-01
-1.41056275e+00 5.11607826e-01 -2.62394160e-01 -1.46484017e+00
1.55507624e-01 -4.92692590e-01 -2.38156766e-01 3.30314636e-01
-1.36862481e+00 -1.06789136e+00 -2.50774413e-01 2.03805238e-01
1.37718022e-01 1.84396476e-01 2.50538707e-01 3.43249828e-01
-3.44221503e-01 3.16580269e-03 3.11849177e-01 2.20751613e-01
9.14416313e-01 -9.11957383e-01 3.95080954e-01 6.71746790e-01
-3.70927155e-02 4.43830907e-01 1.46738982e+00 -1.01257360e+00
-9.27958727e-01 -8.47524524e-01 1.48750377e+00 -9.16098595e-01
1.72972345e+00 -2.38267988e-01 -7.87396789e-01 4.14822787e-01
1.46823958e-01 -3.84476840e-01 4.68285978e-01 -1.05228499e-01
-6.23655438e-01 4.63639498e-01 -1.35301733e+00 1.94169715e-01
6.25567973e-01 -4.92088586e-01 -1.09550750e+00 1.36684859e+00
4.52597648e-01 -2.93180943e-01 -7.96699584e-01 -3.75824958e-01
3.79891187e-01 -7.95470059e-01 4.33865488e-01 -7.40803123e-01
8.84789765e-01 -2.41274580e-01 3.00642222e-01 -1.02958846e+00
-2.81539828e-01 -6.26631439e-01 -2.86835164e-01 1.02260673e+00
3.59459728e-01 -1.00241470e+00 5.85349917e-01 9.23336595e-02
-1.37135042e-02 -1.11791883e-02 -1.10891449e+00 -7.91519463e-01
-1.98726058e-01 2.46906504e-02 -5.01506031e-03 1.12504470e+00
3.53069425e-01 3.72305036e-01 -5.03181458e-01 -1.14289418e-01
5.37004411e-01 9.00928378e-02 7.02013969e-01 -1.34603214e+00
1.83307170e-03 -5.55287659e-01 -5.05852997e-01 -7.56138563e-01
-3.19267571e-01 -5.89834213e-01 -3.42670828e-01 -1.20422530e+00
-9.08428207e-02 -1.77206188e-01 3.64690244e-01 -1.04180723e-01
3.01135987e-01 6.41291499e-01 -1.99709497e-02 9.95162070e-01
-8.35771441e-01 -1.59455940e-01 1.14915359e+00 -4.45115566e-02
3.88378412e-01 8.66025090e-02 -5.92932403e-01 8.95829499e-01
6.24659359e-01 -9.27987397e-01 6.74075365e-01 3.53715062e-01
1.03500235e+00 -2.02665463e-01 4.22795653e-01 -6.88296497e-01
9.01325941e-02 -4.47026938e-02 -1.60329014e-01 -6.20049417e-01
-1.41019896e-01 -6.57450914e-01 2.10434031e-02 8.88401330e-01
-4.93846297e-01 -1.02391437e-01 -4.35951442e-01 8.72572958e-01
-2.42768869e-01 -7.34217763e-01 6.83225513e-01 -2.74630666e-01
-2.13783294e-01 -3.38685095e-01 -8.26319039e-01 2.52780795e-01
9.96364117e-01 4.91971783e-02 -1.25166106e+00 -5.71063876e-01
-6.60688996e-01 -3.21643919e-01 5.75651050e-01 3.57454896e-01
2.47748196e-02 -9.12565112e-01 -8.11291754e-01 -3.27800274e-01
4.86634851e-01 -1.19611287e+00 -3.52298766e-01 1.30962694e+00
-1.18730581e+00 6.30882740e-01 -6.43242449e-02 6.68738186e-02
-1.23250675e+00 7.10069537e-01 -3.66297998e-02 -2.32303113e-01
-3.52267087e-01 4.26569700e-01 -8.16168845e-01 1.45806074e-01
-1.76472306e-01 -8.72065574e-02 -5.03141522e-01 7.19266653e-01
7.49322712e-01 9.58712399e-01 5.27389050e-01 -1.02355587e+00
-2.00634018e-01 -1.20879821e-01 -1.07555747e-01 1.52055353e-01
1.27840483e+00 -2.31684610e-01 -5.66086411e-01 4.15106595e-01
1.35347819e+00 7.39851356e-01 -2.86472231e-01 -3.31834368e-02
6.14982247e-01 -7.69660711e-01 1.29547760e-01 -8.73985767e-01
-4.96934116e-01 1.51107401e-01 -2.03129426e-02 1.36078429e+00
5.74322082e-02 2.79264778e-01 8.75299633e-01 7.77992532e-02
3.14098030e-01 -1.40021944e+00 1.20300815e-01 3.28126371e-01
1.09010136e+00 -1.23675048e+00 4.20608461e-01 -7.64352977e-01
-2.62546360e-01 1.28985107e+00 -1.39782995e-01 -4.16565508e-01
4.24732685e-01 -2.34738186e-01 -1.81975916e-01 -8.09882939e-01
-4.46576476e-01 1.31026149e-01 5.37560284e-01 1.01461634e-01
4.17286754e-01 6.76709935e-02 -1.53063321e+00 2.09871545e-01
-2.83457935e-01 -2.77233452e-01 1.10030556e+00 8.09029460e-01
-8.94229174e-01 -6.69157088e-01 -5.91713965e-01 8.09126556e-01
-1.04766905e+00 -1.58669010e-01 -8.75704288e-01 8.01966846e-01
-1.59990899e-02 9.29721475e-01 -3.87741297e-01 -1.22328870e-01
2.46148229e-01 -1.15862735e-01 3.02554257e-02 -3.38005215e-01
-9.47262168e-01 1.44273769e-02 1.02502835e+00 -4.24079627e-01
-3.86096478e-01 -6.53672397e-01 -4.76396501e-01 -6.06305778e-01
-8.32610488e-01 5.12829959e-01 9.84296322e-01 9.78823662e-01
3.15024585e-01 -3.73159442e-03 4.47374791e-01 -1.93755299e-01
-7.93907940e-01 -1.04255068e+00 -5.23097217e-01 7.55878985e-01
3.74565423e-01 -5.65694630e-01 -9.98758733e-01 -8.21485072e-02]
|
[8.142444610595703, 10.164408683776855]
|
6cdea61a-e06c-4e1a-bdaf-3125b2a7797c
|
fine-grained-opinion-summarization-with
|
2110.08845
| null |
https://arxiv.org/abs/2110.08845v1
|
https://arxiv.org/pdf/2110.08845v1.pdf
|
Fine-Grained Opinion Summarization with Minimal Supervision
|
Opinion summarization aims to profile a target by extracting opinions from multiple documents. Most existing work approaches the task in a semi-supervised manner due to the difficulty of obtaining high-quality annotation from thousands of documents. Among them, some use aspect and sentiment analysis as a proxy for identifying opinions. In this work, we propose a new framework, FineSum, which advances this frontier in three aspects: (1) minimal supervision, where only aspect names and a few aspect/sentiment keywords are available; (2) fine-grained opinion analysis, where sentiment analysis drills down to the sub-aspect level; and (3) phrase-based summarization, where opinion is summarized in the form of phrases. FineSum automatically identifies opinion phrases from the raw corpus, classifies them into different aspects and sentiments, and constructs multiple fine-grained opinion clusters under each aspect/sentiment. Each cluster consists of semantically coherent phrases, expressing uniform opinions towards certain sub-aspect or characteristics (e.g., positive feelings for ``burgers'' in the ``food'' aspect). An opinion-oriented spherical word embedding space is trained to provide weak supervision for the phrase classifier, and phrase clustering is performed using the aspect-aware contextualized embedding generated from the phrase classifier. Both automatic evaluation on the benchmark and quantitative human evaluation validate the effectiveness of our approach.
|
['Jiawei Han', 'Sharon Wang', 'Yu Meng', 'Jiaxin Huang', 'Suyu Ge']
|
2021-10-17
| null | null | null | null |
['fine-grained-opinion-analysis']
|
['natural-language-processing']
|
[ 2.35365912e-01 2.46166319e-01 -6.25649154e-01 -4.31356877e-01
-1.08257639e+00 -7.94075489e-01 6.02081060e-01 9.66600001e-01
-3.57395083e-01 5.97494185e-01 1.01835477e+00 1.09614819e-01
1.10287927e-01 -8.40543330e-01 -2.66189605e-01 -8.44708502e-01
2.27011710e-01 5.00044167e-01 7.86245540e-02 -3.51563960e-01
5.49614251e-01 -6.72166198e-02 -1.41744578e+00 4.35589522e-01
1.04541194e+00 1.15297687e+00 -2.08566204e-01 4.82626975e-01
-7.60613620e-01 4.82582450e-01 -9.91757870e-01 -7.10363507e-01
-1.32220760e-01 -3.97639424e-01 -7.14176655e-01 3.62243444e-01
3.31836962e-03 2.28364214e-01 7.16078937e-01 1.23634410e+00
4.37935889e-01 7.73511082e-02 7.19767451e-01 -8.49272251e-01
-8.23520482e-01 8.21765423e-01 -6.97084546e-01 -1.04577258e-01
5.34577072e-01 -3.68376642e-01 1.76349473e+00 -1.16198480e+00
3.77433836e-01 1.14833629e+00 5.20981371e-01 2.26681903e-01
-1.02537048e+00 -1.38713792e-01 5.52805722e-01 -1.54899344e-01
-8.94842148e-01 -7.32977092e-02 9.00351822e-01 -2.09557161e-01
1.14729559e+00 3.73595148e-01 9.42110777e-01 7.97817290e-01
3.86501253e-01 1.09545529e+00 1.01174259e+00 -3.01461577e-01
7.05027759e-01 5.80890894e-01 8.27999294e-01 2.36808121e-01
6.03138804e-01 -7.76839256e-01 -7.34638631e-01 -4.49896425e-01
-2.05577269e-01 -8.68163630e-02 -3.12510759e-01 -1.83166310e-01
-1.08790159e+00 9.99716938e-01 8.42603594e-02 3.80610317e-01
-6.95008516e-01 -3.85466546e-01 6.86912537e-01 1.49054632e-01
1.19096231e+00 8.30792129e-01 -8.14091623e-01 -2.68658493e-02
-8.21092010e-01 2.08142072e-01 1.05604577e+00 8.58351409e-01
9.40791607e-01 -7.12597445e-02 -2.67257512e-01 9.46652651e-01
3.30683291e-01 6.11837208e-01 8.54297221e-01 -5.22503793e-01
4.90964383e-01 1.17332613e+00 1.29149795e-01 -1.14433420e+00
-2.87020981e-01 -3.89734745e-01 -8.12582314e-01 -2.20825851e-01
-5.12482584e-01 -4.17330354e-01 -7.43231654e-01 1.24473441e+00
3.74076277e-01 -4.90175813e-01 4.98501241e-01 6.48748636e-01
1.19317031e+00 8.47158492e-01 -4.82721580e-03 -5.72246790e-01
1.76635635e+00 -1.32569563e+00 -8.42556298e-01 -3.62712830e-01
3.83567959e-01 -6.77757144e-01 1.09757054e+00 2.81084836e-01
-8.92685831e-01 -3.05647492e-01 -1.02865350e+00 1.54582649e-01
-7.39177048e-01 -2.26508901e-02 5.53942323e-01 5.89205265e-01
-1.04101562e+00 1.26372710e-01 -4.05700594e-01 -2.58098304e-01
2.93691754e-01 3.43887776e-01 -4.52343136e-01 1.34995610e-01
-1.17731702e+00 6.00862563e-01 3.03057462e-01 -1.94530457e-01
6.16407692e-02 -5.20696819e-01 -1.22738969e+00 2.47285575e-01
3.64637077e-01 -8.97998095e-01 9.24637198e-01 -1.11572945e+00
-1.36976242e+00 7.42879570e-01 -6.67463601e-01 -1.84924290e-01
-3.59353393e-01 -3.54155332e-01 -4.82775897e-01 4.74458486e-01
5.82520247e-01 4.42455888e-01 9.10582721e-01 -1.54179645e+00
-9.67132688e-01 -5.16340554e-01 2.28205398e-01 5.60787499e-01
-9.19113100e-01 1.21839486e-01 -4.94239718e-01 -7.37107456e-01
1.73702668e-02 -6.57568634e-01 -5.48810899e-01 -8.61494839e-01
-5.61919630e-01 -6.81938231e-01 5.84949195e-01 -2.97877431e-01
1.55504274e+00 -1.89370918e+00 2.51135200e-01 4.94536944e-02
1.55897960e-01 -1.01609090e-02 -7.07273558e-02 4.92145211e-01
-1.61410924e-02 3.14691126e-01 -3.92845184e-01 -4.86102790e-01
2.18896568e-01 1.71512917e-01 -4.99395132e-01 -8.85778591e-02
4.07169014e-01 9.54683721e-01 -1.22654760e+00 -6.50445342e-01
-2.23242566e-01 2.09882230e-01 -5.24039865e-01 9.18629542e-02
-3.10453594e-01 4.70450558e-02 -8.59531879e-01 8.48525524e-01
3.78337860e-01 -1.66103482e-01 -8.19105059e-02 -4.96205419e-01
-1.78184658e-01 3.39864761e-01 -8.83613825e-01 1.25063264e+00
-4.93117362e-01 1.17302150e-01 -6.26132116e-02 -9.33851004e-01
8.27605546e-01 5.05979478e-01 6.43086910e-01 -7.11977258e-02
2.09071547e-01 1.37793213e-01 -5.66624224e-01 -5.21368027e-01
1.18263578e+00 -2.94529259e-01 -7.53314853e-01 6.88703954e-01
1.15110442e-01 -6.25957310e-01 7.42840707e-01 1.51630744e-01
1.01865244e+00 -2.03767449e-01 8.26968014e-01 -3.74614775e-01
7.39634335e-01 4.33626384e-01 5.72724640e-01 3.93875331e-01
7.00475473e-04 7.19866455e-01 5.93974590e-01 -2.58952022e-01
-5.49780965e-01 -8.35100830e-01 4.79321145e-02 1.18632829e+00
2.39775956e-01 -9.04112935e-01 -4.92752135e-01 -1.04543221e+00
-1.74491227e-01 7.77016819e-01 -7.41921127e-01 1.37495557e-02
-1.59470081e-01 -1.07126331e+00 -9.57945436e-02 5.98892629e-01
2.94217795e-01 -1.10569632e+00 -2.23494992e-01 2.05486834e-01
-3.24944675e-01 -8.32507789e-01 -5.55868864e-01 4.05976832e-01
-6.31466568e-01 -7.82969475e-01 -7.90297687e-01 -9.31304395e-01
1.00111854e+00 2.17041269e-01 1.28029287e+00 -4.91679758e-01
5.20277917e-01 4.78089750e-01 -9.72955525e-01 -6.69621766e-01
-3.33684832e-02 2.23125055e-01 1.87711075e-01 2.48723671e-01
7.97712624e-01 -4.07116503e-01 -5.78989446e-01 -1.67272016e-01
-1.00670826e+00 -5.10923862e-01 8.17541003e-01 7.36490309e-01
8.42878222e-01 4.29141462e-01 5.65334618e-01 -1.12783170e+00
1.27614188e+00 -4.23483878e-01 1.36677861e-01 1.72609881e-01
-6.07581377e-01 -1.57356095e-02 9.29428935e-01 -2.95119256e-01
-8.99607241e-01 -2.89483011e-01 -2.90070623e-01 2.13304341e-01
-1.22353010e-01 9.12888944e-01 -2.16393173e-01 5.46651661e-01
3.65696520e-01 5.06105661e-01 -5.02134562e-01 -8.71509165e-02
5.76612711e-01 9.31025445e-01 1.88555762e-01 -4.56281602e-01
7.44134963e-01 5.57424366e-01 -4.89609957e-01 -1.00882661e+00
-1.52980947e+00 -9.99117613e-01 -4.82082695e-01 -2.49471087e-02
9.58271325e-01 -9.09265816e-01 -1.75702676e-01 4.99500781e-02
-1.09725988e+00 4.86332595e-01 -7.46257424e-01 4.01730925e-01
-2.62942195e-01 5.43043792e-01 -3.66817832e-01 -7.95086920e-01
-9.97821987e-01 -9.39460695e-01 1.44380224e+00 4.12963033e-01
-9.27469254e-01 -1.08968675e+00 3.73175830e-01 3.75369608e-01
2.33012080e-01 5.22840135e-02 9.17216241e-01 -9.82024074e-01
2.05371961e-01 -5.98277748e-01 1.02411807e-01 7.63435185e-01
5.15214622e-01 -1.40061185e-01 -7.99965203e-01 -7.83998370e-02
2.28231281e-01 -2.99969852e-01 1.02304077e+00 3.10110450e-01
5.93990982e-01 -6.49099350e-01 -1.53754666e-01 -5.37866987e-02
1.25450146e+00 -4.57869172e-02 4.31330688e-02 3.85750711e-01
6.69823050e-01 8.16239655e-01 7.74249136e-01 2.12097675e-01
6.08227670e-01 1.69964716e-01 8.43874812e-02 -1.95724387e-02
4.06488121e-01 4.21461724e-02 6.95104122e-01 1.50847793e+00
1.02157079e-01 -2.91937500e-01 -3.90453637e-01 9.50804353e-01
-1.77108574e+00 -9.08013701e-01 2.80007180e-02 1.65490401e+00
9.51597393e-01 3.76152188e-01 2.93649044e-02 2.17235208e-01
4.49666739e-01 7.10467160e-01 -3.35948765e-01 -8.39596331e-01
-2.99914062e-01 2.57014096e-01 1.61396623e-01 3.99403930e-01
-1.21339810e+00 9.44633484e-01 5.26030207e+00 7.53164411e-01
-8.35535467e-01 -9.33627710e-02 5.45605361e-01 2.15168521e-01
-8.92307222e-01 5.01670986e-02 -8.80140960e-01 3.17213714e-01
5.47052681e-01 -5.23516059e-01 -5.56152403e-01 1.16155088e+00
1.51740491e-01 -2.66104072e-01 -6.54608130e-01 4.78666067e-01
7.66671002e-01 -1.13548493e+00 5.82028270e-01 -2.77051419e-01
1.06690717e+00 -2.56574124e-01 3.15933935e-02 2.81597972e-01
2.91842878e-01 -4.96823013e-01 5.23772657e-01 1.33427829e-01
3.84894907e-01 -9.08759773e-01 1.28753388e+00 1.28731668e-01
-1.43433237e+00 1.81200966e-01 -3.88933629e-01 1.35439992e-01
3.78921360e-01 1.02897251e+00 -3.90085369e-01 9.39123869e-01
7.86721587e-01 1.01706338e+00 -4.34736997e-01 5.49486518e-01
-6.22545600e-01 6.52810216e-01 -8.06537643e-02 -4.81088728e-01
4.61125314e-01 -5.81987798e-01 7.48877108e-01 1.42981172e+00
1.58655524e-01 2.49426346e-03 2.98348784e-01 2.56770134e-01
-4.20655496e-02 5.93633592e-01 -4.97310966e-01 -2.58243233e-01
9.86092091e-02 1.61537719e+00 -9.38948452e-01 -6.32467210e-01
-4.75398898e-01 1.01766706e+00 -1.28446862e-01 3.09304297e-01
-1.05191059e-01 -7.36587584e-01 6.12530351e-01 -4.12117034e-01
5.36025584e-01 1.81740850e-01 -5.08161843e-01 -1.40007973e+00
3.41309935e-01 -8.17665279e-01 3.11428398e-01 -5.41857123e-01
-1.32169759e+00 1.02112639e+00 -1.17343016e-01 -1.23111475e+00
-2.99261510e-01 -5.41352630e-01 -9.02088344e-01 5.88390648e-01
-1.58676636e+00 -1.07006550e+00 1.39478207e-01 9.63204354e-02
9.33970451e-01 -9.55536962e-02 1.21414876e+00 -2.74775475e-01
-5.74639142e-01 3.24340671e-01 -8.67935643e-02 -5.34626432e-02
6.58961117e-01 -1.71328378e+00 1.38058022e-01 6.68556213e-01
-1.43710207e-02 8.86984646e-01 9.99801934e-01 -4.73238587e-01
-1.09452605e+00 -1.13693380e+00 1.62820768e+00 -6.39118195e-01
9.02466297e-01 -1.84370443e-01 -7.20645070e-01 3.10918570e-01
5.42488098e-01 -6.15328908e-01 1.43801761e+00 4.32758778e-01
-2.30098605e-01 -1.55724332e-01 -8.47369432e-01 7.72216439e-01
2.78840840e-01 -3.20492297e-01 -1.31726480e+00 3.42508018e-01
1.11348891e+00 2.07294822e-01 -8.42521012e-01 3.73372167e-01
4.27226633e-01 -8.60309005e-01 6.30690932e-01 -4.72831905e-01
7.72506833e-01 -4.61203367e-01 -2.15720579e-01 -1.82635832e+00
-6.19998500e-02 -3.13304007e-01 -2.10258484e-01 1.48982632e+00
8.55024576e-01 -5.18143952e-01 7.95627296e-01 9.97096896e-02
-2.37324923e-01 -1.14523244e+00 -4.60903704e-01 -2.79289484e-01
-4.58937287e-02 -4.41712141e-01 6.17474198e-01 6.24808073e-01
4.52627927e-01 1.01156259e+00 -1.06276255e-02 2.22248569e-01
3.64836007e-01 7.01376736e-01 5.64391673e-01 -1.01092660e+00
-1.01217613e-01 -5.19440234e-01 -3.15034896e-01 -1.05245769e+00
3.40064853e-01 -6.22594416e-01 1.36786759e-01 -2.01553011e+00
3.67322475e-01 2.71094680e-01 -3.52056533e-01 2.54797846e-01
-6.11907899e-01 2.65335351e-01 -2.03508779e-01 1.06908560e-01
-1.14393914e+00 8.22639406e-01 9.17138100e-01 -4.88449872e-01
-2.63240188e-01 2.77801722e-01 -1.55727565e+00 1.13282096e+00
8.29106450e-01 -4.38252985e-01 -2.62534052e-01 -1.65790752e-01
7.68480062e-01 -5.39475858e-01 -5.00696421e-01 -5.80176473e-01
1.89100817e-01 -4.20221649e-02 5.61202727e-02 -9.52584982e-01
3.26685250e-01 -6.48087680e-01 -8.10098648e-01 6.84298947e-02
-2.78223693e-01 1.96618408e-01 -7.53962472e-02 5.76992750e-01
-7.98818946e-01 -2.46373683e-01 2.87839502e-01 -1.34538054e-01
-6.67212009e-01 1.44335374e-01 -4.82622385e-01 5.91499992e-02
6.61462247e-01 -1.82523176e-01 -1.95585430e-01 -4.63478565e-01
-4.91287291e-01 6.24763191e-01 4.67025816e-01 3.70215446e-01
6.72790349e-01 -1.31365669e+00 -6.90382957e-01 5.15730819e-04
7.25493371e-01 1.80228442e-01 -7.64141278e-03 7.32937813e-01
-6.87156850e-03 5.09369195e-01 4.17278767e-01 -3.48406017e-01
-1.23762989e+00 3.90506566e-01 -3.31218421e-01 -6.74202442e-01
-2.47510225e-01 8.62592340e-01 1.65479124e-01 -6.24507964e-01
-3.12500000e-02 -3.76228094e-01 -9.97673988e-01 8.08711827e-01
7.03719079e-01 -7.00172316e-03 -4.58714105e-02 -9.89683330e-01
-2.66716152e-01 1.03921068e+00 -2.01370388e-01 -1.56096295e-01
1.51584244e+00 -3.32589269e-01 -7.36115754e-01 8.08341861e-01
1.35759628e+00 6.13480449e-01 -6.72025859e-01 -2.11467624e-01
1.98831722e-01 1.20447002e-01 -1.36279255e-01 -5.01982868e-01
-6.66597843e-01 3.39267522e-01 -2.33503178e-01 6.93034470e-01
1.25101769e+00 2.90410995e-01 8.70144308e-01 5.35987854e-01
9.81279742e-03 -1.33983111e+00 1.72688499e-01 7.85496593e-01
7.74522126e-01 -1.22294021e+00 1.59918666e-01 -2.51305342e-01
-1.03950942e+00 9.05554533e-01 2.01908022e-01 -1.80318639e-01
7.70247579e-01 4.33972143e-02 3.48237276e-01 -4.99266207e-01
-7.87130892e-01 -3.92015398e-01 4.97200698e-01 3.54896396e-01
4.54587132e-01 7.83866495e-02 -7.31184840e-01 1.09292412e+00
-6.68610811e-01 -6.62066460e-01 4.03158396e-01 8.97380233e-01
-8.28231573e-01 -1.04580557e+00 -2.63115138e-01 7.82147288e-01
-7.40240037e-01 -3.00256550e-01 -8.68430257e-01 2.37766251e-01
1.25425771e-01 1.48159778e+00 -1.43311605e-01 -2.86213249e-01
5.98220885e-01 8.85563940e-02 -2.45532408e-01 -1.26837087e+00
-7.43297875e-01 2.18278319e-01 2.49665514e-01 -3.85194689e-01
-8.19230080e-01 -5.24237931e-01 -1.02737498e+00 1.65071189e-01
-4.13000584e-01 8.70016575e-01 5.14294028e-01 1.09943020e+00
2.28218824e-01 4.73902225e-01 9.39270437e-01 -8.05677056e-01
-4.24465299e-01 -9.63886857e-01 -6.31709099e-01 4.75704283e-01
3.83763731e-01 -6.42101616e-02 -6.48341835e-01 -3.59601192e-02]
|
[11.421252250671387, 6.69329833984375]
|
b459c04b-0cb0-4419-b007-0644ccc93dd6
|
flowtext-synthesizing-realistic-scene-text
|
2305.03327
| null |
https://arxiv.org/abs/2305.03327v1
|
https://arxiv.org/pdf/2305.03327v1.pdf
|
FlowText: Synthesizing Realistic Scene Text Video with Optical Flow Estimation
|
Current video text spotting methods can achieve preferable performance, powered with sufficient labeled training data. However, labeling data manually is time-consuming and labor-intensive. To overcome this, using low-cost synthetic data is a promising alternative. This paper introduces a novel video text synthesis technique called FlowText, which utilizes optical flow estimation to synthesize a large amount of text video data at a low cost for training robust video text spotters. Unlike existing methods that focus on image-level synthesis, FlowText concentrates on synthesizing temporal information of text instances across consecutive frames using optical flow. This temporal information is crucial for accurately tracking and spotting text in video sequences, including text movement, distortion, appearance, disappearance, shelter, and blur. Experiments show that combining general detectors like TransDETR with the proposed FlowText produces remarkable results on various datasets, such as ICDAR2015video and ICDAR2013video. Code is available at https://github.com/callsys/FlowText.
|
['Weiqiang Wang', 'Jiahong Li', 'Zhuang Li', 'Weijia Wu', 'Yuzhong Zhao']
|
2023-05-05
| null | null | null | null |
['text-spotting']
|
['computer-vision']
|
[ 2.33239934e-01 -5.89514911e-01 -1.96483150e-01 -9.68517661e-02
-3.91160399e-01 -5.63875198e-01 6.75287664e-01 -1.83168352e-01
-1.91703826e-01 6.80708289e-01 1.97410181e-01 -1.78315461e-01
4.36309755e-01 -3.95086437e-01 -6.45538449e-01 -5.90148866e-01
5.50227582e-01 5.91114573e-02 3.81825864e-01 1.12357967e-01
5.16450942e-01 1.58736944e-01 -1.50542486e+00 3.11900109e-01
9.38674688e-01 5.83520889e-01 4.67070282e-01 8.85168850e-01
-3.34054321e-01 1.00834012e+00 -5.44920027e-01 -4.49765295e-01
2.42937684e-01 -7.57456779e-01 -4.28089321e-01 5.31130791e-01
7.22309113e-01 -8.12691271e-01 -6.98089004e-01 1.08155119e+00
3.81028503e-01 1.58738211e-01 4.93068516e-01 -1.33677256e+00
-5.31981289e-01 5.28882980e-01 -7.10903585e-01 1.85459435e-01
6.93679273e-01 4.74555850e-01 5.65458596e-01 -1.08653080e+00
8.71197820e-01 1.08810127e+00 4.96405274e-01 7.29940951e-01
-9.46717203e-01 -7.55732417e-01 9.65509862e-02 2.24809527e-01
-1.28921938e+00 -8.15313935e-01 8.01237404e-01 -5.83307326e-01
5.71834981e-01 2.02070221e-01 5.96645713e-01 1.38292038e+00
1.75861329e-01 1.23527348e+00 7.54701853e-01 -3.49870622e-01
-2.31115613e-02 -5.13000488e-02 -2.61291087e-01 7.78983653e-01
2.39584759e-01 -4.76382151e-02 -7.64662027e-01 2.79144317e-01
8.84499729e-01 2.27249295e-01 -7.88672984e-01 -6.66415244e-02
-1.51588893e+00 2.71895409e-01 -2.17185065e-01 1.85150594e-01
-5.52592240e-02 2.23090574e-01 5.24246752e-01 2.69187480e-01
5.01147807e-01 1.35952956e-03 -1.58195212e-01 -4.34150517e-01
-1.57941663e+00 8.27346221e-02 4.94918853e-01 1.16069210e+00
6.08806074e-01 3.89378339e-01 -2.91288137e-01 6.74987078e-01
3.32410127e-01 8.16686392e-01 7.25087166e-01 -8.55181396e-01
9.14337039e-01 4.99791354e-01 2.00824395e-01 -9.74460244e-01
-1.32710576e-01 3.92393172e-01 -7.18713284e-01 1.58503190e-01
5.17748713e-01 -3.98502380e-01 -1.00086462e+00 1.08715510e+00
3.72034013e-01 3.02131295e-01 -2.30530903e-01 1.06315684e+00
8.39011967e-01 9.04953122e-01 -2.03633472e-01 -4.24326211e-01
9.43069875e-01 -1.12130463e+00 -1.28739011e+00 2.04160493e-02
6.40583634e-01 -1.31737864e+00 1.00682914e+00 5.13152719e-01
-1.15842867e+00 -4.54432458e-01 -9.33565915e-01 -2.56259739e-01
-7.09234849e-02 5.03364027e-01 1.63741603e-01 6.25336289e-01
-9.17126298e-01 4.04515147e-01 -8.98768008e-01 -4.49401051e-01
3.15908283e-01 1.01156443e-01 -2.37069607e-01 -8.94056931e-02
-8.56549680e-01 3.50554854e-01 2.60506272e-01 1.31364703e-01
-8.48016739e-01 -5.24009287e-01 -7.50793874e-01 -1.50656432e-01
5.95442712e-01 -5.12458146e-01 1.17447102e+00 -1.09784651e+00
-1.79067206e+00 5.19991279e-01 -3.50666702e-01 -2.45973557e-01
1.19745350e+00 -3.33565027e-01 -4.03687775e-01 4.22724515e-01
-8.36368576e-02 7.03718424e-01 1.48441541e+00 -8.82157505e-01
-6.42997622e-01 1.19527662e-02 -3.66010666e-01 3.20252508e-01
-3.51883680e-01 9.22933593e-02 -6.65220678e-01 -1.02064693e+00
-1.87363431e-01 -7.04651415e-01 2.39055499e-01 4.30419415e-01
-5.09141862e-01 -7.24594221e-02 1.51842654e+00 -8.12491477e-01
1.35160375e+00 -1.91229975e+00 -2.26436183e-02 -4.86315936e-01
1.62760347e-01 5.83627880e-01 -2.01439988e-02 5.06290436e-01
1.74629673e-01 1.03217609e-01 -1.34798616e-01 -3.94120783e-01
-3.22225899e-01 -2.44651943e-01 -3.72876704e-01 5.86178303e-01
1.17984414e-02 8.88494492e-01 -7.91882515e-01 -8.74331534e-01
8.93214405e-01 5.07656097e-01 -4.11370665e-01 1.14240505e-01
-5.23612559e-01 4.19435740e-01 -5.15571475e-01 7.83645272e-01
7.60931015e-01 -2.21671641e-01 -7.09094480e-02 -1.53715417e-01
-1.70691103e-01 -1.14771746e-01 -1.13491929e+00 1.77670598e+00
-2.34721154e-01 1.31098807e+00 4.40559201e-02 -6.56957448e-01
6.89912438e-01 5.15405178e-01 7.03815818e-01 -5.43739319e-01
5.34145713e-01 -3.74993943e-02 -5.56840301e-01 -1.01101410e+00
8.53236318e-01 3.38701218e-01 3.54046434e-01 6.21355593e-01
-9.70119238e-02 -2.39081964e-01 6.03242695e-01 3.37570846e-01
9.77961361e-01 5.79083145e-01 -8.05019811e-02 6.23349622e-02
6.81427002e-01 -6.59601539e-02 3.83143187e-01 4.19647723e-01
-4.11453694e-01 9.01482046e-01 1.93686366e-01 -3.65757227e-01
-1.29249597e+00 -5.99640250e-01 1.04266845e-01 6.40753508e-01
4.10928845e-01 -6.11585200e-01 -8.93679798e-01 -7.06423283e-01
-2.98666865e-01 4.09789443e-01 -3.25623363e-01 2.19629377e-01
-7.33015597e-01 -3.98827076e-01 4.95167911e-01 1.82797894e-01
7.48152435e-01 -9.89242494e-01 -6.27603590e-01 5.08490615e-02
-7.19447911e-01 -1.47951221e+00 -1.08569646e+00 -5.41471958e-01
-9.42705095e-01 -1.10530949e+00 -1.20901310e+00 -7.56590307e-01
7.49895990e-01 6.36416316e-01 6.47797942e-01 3.02589595e-01
-4.53971863e-01 2.14182869e-01 -5.76405287e-01 9.46237668e-02
-5.72551012e-01 -1.83643594e-01 -1.71885401e-01 2.73934692e-01
8.19904581e-02 -6.85468270e-03 -7.40351439e-01 4.89513159e-01
-1.21058166e+00 5.59546232e-01 2.99894631e-01 7.06574619e-01
2.46378064e-01 -6.16326556e-02 1.68849155e-01 -6.19020045e-01
5.07076263e-01 -1.04757465e-01 -7.19636500e-01 3.00487101e-01
-4.22990173e-01 -1.96041644e-01 9.31349158e-01 -6.18643403e-01
-1.42248833e+00 2.84096807e-01 1.96975529e-01 -7.92206347e-01
-2.23025292e-01 -3.78327221e-02 1.45789996e-01 1.16071396e-01
6.21380806e-01 6.66276038e-01 -1.08110815e-01 -1.56648517e-01
2.98329383e-01 8.98940682e-01 3.47344130e-01 -2.00395316e-01
7.78007567e-01 7.58887231e-01 -3.33769858e-01 -1.18022633e+00
-3.43716860e-01 -4.95024294e-01 -5.78932881e-01 -4.76400256e-01
8.05110455e-01 -8.10639381e-01 -5.70323944e-01 8.57070982e-01
-1.09615076e+00 -4.93324548e-01 3.64929698e-02 5.51427007e-01
-5.99939883e-01 8.38525891e-01 -6.19604945e-01 -6.86896741e-01
-3.74104887e-01 -1.16060758e+00 1.15954697e+00 2.50347406e-01
1.01731904e-01 -9.70446765e-01 -1.30661383e-01 4.72244442e-01
1.47842735e-01 2.90327352e-02 1.71895057e-01 -7.64611037e-03
-9.27582204e-01 -1.62958369e-01 -3.14491004e-01 1.86088234e-01
4.25058335e-01 6.18487835e-01 -7.58282363e-01 -2.78483152e-01
-2.18415022e-01 -1.90240130e-01 8.15950394e-01 5.05830109e-01
1.05589283e+00 -3.56775850e-01 -3.30356419e-01 6.80390418e-01
1.26667583e+00 3.78991544e-01 6.06216669e-01 1.19401239e-01
9.76902962e-01 4.53144044e-01 8.87621224e-01 5.64248145e-01
2.09207013e-01 6.50448143e-01 1.34287670e-01 1.16478717e-02
-6.15202665e-01 -2.61867315e-01 6.57415211e-01 7.28144825e-01
1.73656959e-02 -8.49086881e-01 -7.57710755e-01 4.18880731e-01
-2.00581431e+00 -1.18119371e+00 -3.65938604e-01 2.01162815e+00
7.14361489e-01 -5.92870340e-02 -3.82855884e-03 3.70493770e-01
1.13857448e+00 3.44188124e-01 -5.07285416e-01 2.74677929e-02
-1.49861813e-01 -2.94188410e-01 5.45696437e-01 4.67272848e-01
-1.07679927e+00 1.25264215e+00 5.48524904e+00 8.36582303e-01
-1.33256304e+00 -8.22845325e-02 4.13395375e-01 -2.00663835e-01
-7.68922344e-02 -1.55057147e-01 -7.37045050e-01 8.31449807e-01
6.61334634e-01 -2.93494284e-01 3.57626408e-01 3.06528151e-01
9.19343352e-01 -2.67523110e-01 -8.41553032e-01 1.39255714e+00
3.65648121e-01 -1.48013926e+00 2.48319790e-01 -3.31639469e-01
9.16160762e-01 -2.06457511e-01 -1.22261392e-02 -2.27302700e-01
1.65508874e-02 -4.97767270e-01 7.79795408e-01 4.32580858e-01
1.06308305e+00 -2.43607163e-01 3.21855098e-01 8.29349831e-02
-1.24130356e+00 1.86617240e-01 -1.51165709e-01 2.52564818e-01
3.20214719e-01 4.20965821e-01 -8.70120287e-01 4.69383389e-01
5.29269755e-01 1.41378582e+00 -5.21171153e-01 1.23760509e+00
-3.04557621e-01 5.94095528e-01 -1.66382313e-01 -1.52319714e-01
7.78493797e-03 -3.25917482e-01 6.15265787e-01 1.25052249e+00
4.10474509e-01 -1.26023337e-01 5.63969836e-02 6.10861242e-01
-2.47233823e-01 1.78165153e-01 -5.37047327e-01 -3.43929082e-01
2.57389516e-01 1.25043046e+00 -1.04249811e+00 -5.89909673e-01
-4.56120819e-01 1.28457689e+00 -4.19413388e-01 5.84716856e-01
-1.00509942e+00 -6.08490586e-01 1.99527249e-01 2.56218553e-01
2.65353531e-01 -2.78446853e-01 -9.62079540e-02 -1.76223981e+00
1.57362193e-01 -7.55316675e-01 1.28751397e-01 -1.18428016e+00
-5.60982287e-01 3.21349353e-01 -3.00443649e-01 -1.72793472e+00
-1.46988004e-01 -4.04941350e-01 -5.09379089e-01 3.83030921e-01
-1.39566088e+00 -1.01575541e+00 -7.08586276e-01 8.89391899e-01
1.37251508e+00 -2.46781688e-02 1.06440917e-01 4.94816154e-01
-7.94966102e-01 4.11908150e-01 3.18088442e-01 2.32657209e-01
1.14542198e+00 -8.69844079e-01 5.18177509e-01 1.14743555e+00
1.67648986e-01 6.92313462e-02 8.27533066e-01 -8.39926064e-01
-1.40793085e+00 -1.06001806e+00 5.92324436e-01 -3.05917680e-01
6.54003859e-01 -3.59547555e-01 -8.26436758e-01 6.47900939e-01
3.70174676e-01 3.41020040e-02 4.33257930e-02 -8.65592778e-01
1.10707574e-01 5.53540140e-02 -9.25219417e-01 8.42045426e-01
9.46899116e-01 -3.18918526e-01 -3.00447285e-01 4.43051666e-01
5.35412610e-01 -5.00885248e-01 -4.34498519e-01 4.64620214e-04
6.02725983e-01 -9.72496092e-01 5.54325819e-01 -9.03480221e-03
6.60251141e-01 -5.26143074e-01 2.87618190e-01 -9.62857902e-01
3.90991867e-01 -1.17678261e+00 -1.37766406e-01 1.22997260e+00
3.03406697e-02 -3.24147910e-01 1.06238270e+00 3.24640393e-01
-1.55037018e-02 -1.99713632e-01 -5.42635262e-01 -6.04956388e-01
-3.42615873e-01 -3.19033861e-01 3.29135239e-01 1.13529229e+00
3.20461541e-02 1.44417301e-01 -8.55112553e-01 -2.78902113e-01
6.36811912e-01 -9.40747112e-02 9.34108138e-01 -8.68074477e-01
1.60020649e-01 -3.84092689e-01 -3.32548797e-01 -1.28651023e+00
-1.20854676e-02 -5.57683110e-01 1.05087131e-01 -1.57760739e+00
-9.19818580e-02 -1.24433398e-01 3.22438091e-01 1.39959902e-01
-1.77713186e-01 2.57112592e-01 3.67586434e-01 4.93832648e-01
-6.76939905e-01 6.24490321e-01 1.68349206e+00 -1.96854249e-01
-2.34871864e-01 -1.20346881e-01 3.11855972e-02 6.75393820e-01
8.70077848e-01 -3.43055457e-01 -5.12834847e-01 -6.18246019e-01
5.41270375e-02 4.86190528e-01 8.99570882e-02 -9.54365134e-01
5.05022705e-01 -3.34540397e-01 5.26431620e-01 -8.40005338e-01
2.48947769e-01 -7.84597337e-01 6.42671287e-02 4.13328230e-01
-3.45445126e-01 6.47180378e-02 8.16560984e-02 5.55717349e-01
-2.57434815e-01 -2.87736386e-01 7.18801379e-01 -4.48302962e-02
-7.07083583e-01 5.10299265e-01 -6.03479862e-01 7.85764828e-02
1.17213666e+00 -4.31131184e-01 -6.27924204e-01 -5.10174453e-01
-3.80262136e-01 2.41215765e-01 6.79572582e-01 7.50315726e-01
8.87246013e-01 -9.64968681e-01 -6.20773315e-01 2.38970220e-01
-8.81889835e-02 -1.13604143e-01 3.97336572e-01 7.80513108e-01
-9.73941922e-01 5.28611541e-01 -2.17720479e-01 -7.80943215e-01
-1.41911185e+00 5.42601407e-01 2.51054972e-01 1.74266368e-01
-8.54734719e-01 5.16959429e-01 1.74101204e-01 1.06389917e-01
1.91358030e-01 -3.15368444e-01 -2.45203860e-02 8.35063979e-02
6.98645234e-01 5.30783713e-01 -6.11062534e-02 -5.43973446e-01
-9.35356319e-02 8.77232194e-01 -1.23525657e-01 -9.58105624e-02
7.19363749e-01 -6.95367575e-01 3.90707046e-01 2.59380937e-01
9.66862619e-01 -8.30804333e-02 -1.73970020e+00 -1.54396042e-01
-6.92643002e-02 -8.83994162e-01 6.43932447e-02 -4.48590845e-01
-1.29599428e+00 9.59450185e-01 3.56188715e-01 2.19625514e-02
1.07940686e+00 -5.42918801e-01 9.96172488e-01 2.82928705e-01
1.24724917e-01 -1.21776485e+00 3.74142349e-01 1.79939672e-01
6.30358219e-01 -1.42742383e+00 1.23107672e-01 -3.08834016e-01
-6.27940953e-01 1.41101003e+00 5.89156687e-01 -3.77658345e-02
3.12617689e-01 3.13263863e-01 2.58711696e-01 2.60364801e-01
-8.77036989e-01 2.02196389e-02 2.88684778e-02 2.84732431e-01
4.52131748e-01 -3.20077568e-01 -2.96987563e-01 -4.28367138e-01
1.52555378e-02 2.68563926e-01 1.15296161e+00 9.44430411e-01
-3.26092273e-01 -9.85669255e-01 -4.79895025e-01 3.82363796e-01
-6.83502614e-01 -1.12752363e-01 -2.20343292e-01 7.01277852e-01
-3.04198354e-01 9.55137193e-01 -5.01485309e-03 -1.46524653e-01
-2.72613787e-03 -7.09360987e-02 4.09386367e-01 -2.81651050e-01
-2.44192913e-01 4.34426129e-01 -7.97009468e-02 -4.46745932e-01
-7.89333642e-01 -6.93828344e-01 -1.21381724e+00 -5.23796320e-01
-4.82328147e-01 -1.12487294e-01 7.50705004e-01 6.69264376e-01
6.17009327e-02 4.29265380e-01 5.97025454e-01 -8.44760597e-01
4.76166904e-02 -8.91564846e-01 -4.30205554e-01 5.14828324e-01
6.68505073e-01 -3.82822812e-01 -4.84409064e-01 9.29316282e-01]
|
[10.826600074768066, -0.7795295119285583]
|
cc79b5ca-ff8d-4544-80ba-8f988ea20fe8
|
re2g-retrieve-rerank-generate-2
|
2207.06300
| null |
https://arxiv.org/abs/2207.06300v1
|
https://arxiv.org/pdf/2207.06300v1.pdf
|
Re2G: Retrieve, Rerank, Generate
|
As demonstrated by GPT-3 and T5, transformers grow in capability as parameter spaces become larger and larger. However, for tasks that require a large amount of knowledge, non-parametric memory allows models to grow dramatically with a sub-linear increase in computational cost and GPU memory requirements. Recent models such as RAG and REALM have introduced retrieval into conditional generation. These models incorporate neural initial retrieval from a corpus of passages. We build on this line of research, proposing Re2G, which combines both neural initial retrieval and reranking into a BART-based sequence-to-sequence generation. Our reranking approach also permits merging retrieval results from sources with incomparable scores, enabling an ensemble of BM25 and neural initial retrieval. To train our system end-to-end, we introduce a novel variation of knowledge distillation to train the initial retrieval, reranker, and generation using only ground truth on the target sequence output. We find large gains in four diverse tasks: zero-shot slot filling, question answering, fact-checking, and dialog, with relative gains of 9% to 34% over the previous state-of-the-art on the KILT leaderboard. We make our code available as open source at https://github.com/IBM/kgi-slot-filling/tree/re2g.
|
['Alfio Gliozzo', 'Pengshan Cai', 'Ankita Rajaram Naik', 'Md Faisal Mahbub Chowdhury', 'Gaetano Rossiello', 'Michael Glass']
|
2022-07-13
|
re2g-retrieve-rerank-generate-1
|
https://aclanthology.org/2022.naacl-main.194
|
https://aclanthology.org/2022.naacl-main.194.pdf
|
naacl-2022-7
|
['zero-shot-slot-filling', 'slot-filling', 'open-domain-dialog']
|
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 1.45288467e-01 2.56293684e-01 -2.82433182e-01 -8.43694136e-02
-1.71755219e+00 -7.26171970e-01 6.12552822e-01 2.89742053e-01
-4.62388694e-01 8.98091972e-01 4.59904432e-01 -5.50101221e-01
7.59710148e-02 -9.01112914e-01 -7.92741776e-01 4.55304086e-02
1.43201917e-01 1.22181237e+00 4.95412350e-01 -5.48282981e-01
1.34633437e-01 -2.83637464e-01 -1.24973619e+00 6.37003899e-01
1.05065119e+00 8.92768741e-01 1.40336320e-01 9.73507762e-01
-3.45755011e-01 8.95274043e-01 -6.50497079e-01 -5.07748067e-01
2.15352833e-01 -4.11726147e-01 -1.05593419e+00 -3.33995879e-01
4.29522902e-01 -4.71639484e-01 -3.95336926e-01 4.21324641e-01
7.82842636e-01 5.42956948e-01 3.76316518e-01 -9.44260716e-01
-9.76869762e-01 1.06456923e+00 -2.95665383e-01 5.06656058e-02
5.27651489e-01 2.40905270e-01 1.31166887e+00 -9.89850521e-01
6.99813724e-01 1.36350918e+00 6.44119978e-01 5.92169940e-01
-1.26309156e+00 -5.94256997e-01 -2.25911532e-02 1.69303015e-01
-1.19265079e+00 -4.71100301e-01 6.77344948e-02 -2.21927017e-01
1.49688065e+00 3.31007987e-01 3.73729527e-01 8.27536047e-01
-5.72015703e-01 1.15794444e+00 6.73025310e-01 -5.97502649e-01
1.22385897e-01 -3.13589722e-02 3.20702165e-01 6.45459175e-01
-5.52358702e-02 -3.28869978e-03 -7.20863998e-01 -5.31630397e-01
6.06003761e-01 -3.11723202e-01 -1.37140542e-01 2.56787449e-01
-1.12561107e+00 1.03485584e+00 3.40107679e-01 -3.10080424e-02
-3.07343990e-01 1.85992658e-01 3.54680538e-01 2.57704586e-01
6.39305890e-01 8.45240355e-01 -6.15440071e-01 -4.20321584e-01
-1.23677325e+00 6.16130710e-01 1.15865815e+00 9.67595279e-01
7.56676495e-01 -2.92078733e-01 -9.68254983e-01 1.10318363e+00
1.38430977e-02 4.46907699e-01 6.46090567e-01 -1.09605765e+00
7.58600950e-01 3.89514714e-01 3.40223014e-01 -3.85166794e-01
-2.30251461e-01 -3.57297808e-01 -3.93263489e-01 -5.10450721e-01
3.90969008e-01 -3.08738798e-01 -1.41223264e+00 1.73463643e+00
3.22678357e-01 2.92270482e-01 1.55245647e-01 7.32866943e-01
8.09927523e-01 9.55998719e-01 2.11708635e-01 1.80883244e-01
1.41776824e+00 -1.46817183e+00 -4.01617944e-01 -4.18906659e-01
9.02331293e-01 -9.72664535e-01 1.25104332e+00 9.73579213e-02
-1.20530236e+00 -3.17821145e-01 -6.86573327e-01 -5.51705122e-01
-3.26431632e-01 7.58420955e-03 7.24841356e-01 4.34921980e-01
-1.24503934e+00 5.22128165e-01 -7.64547288e-01 -2.61572927e-01
2.06532523e-01 2.40681637e-02 5.25206700e-02 -2.00824976e-01
-1.70535314e+00 9.03462887e-01 6.66454434e-01 -1.63141221e-01
-7.55521953e-01 -1.01356339e+00 -6.72638834e-01 6.05589114e-02
5.30353129e-01 -1.08483970e+00 1.93841970e+00 -2.81818032e-01
-1.54603457e+00 5.06789446e-01 -3.19125831e-01 -8.66259158e-01
4.60292429e-01 -6.27065420e-01 5.85437426e-03 -5.43626472e-02
2.31932461e-01 9.21702027e-01 3.15593719e-01 -8.01567733e-01
-5.81430495e-01 6.55414462e-02 8.43378901e-02 4.85494465e-01
-2.68711690e-02 -7.54673779e-02 -8.38902354e-01 -4.92442369e-01
-8.06627795e-02 -1.01817083e+00 -3.47063184e-01 -5.59535503e-01
-5.75725436e-01 -4.12586451e-01 2.79739887e-01 -1.08290422e+00
1.45780027e+00 -1.74780524e+00 -7.34965354e-02 -1.20081641e-02
-3.04081850e-02 3.66969854e-01 -4.98729706e-01 7.91849911e-01
4.25059259e-01 3.49944562e-01 -2.51373082e-01 -4.26133424e-01
2.96959311e-01 7.15820119e-02 -5.63415766e-01 -4.14833367e-01
1.76779464e-01 1.36952662e+00 -9.89622712e-01 -4.30752188e-01
-1.66461974e-01 2.40379333e-01 -9.09978211e-01 2.00838491e-01
-8.85164142e-01 1.81655541e-01 -4.41117555e-01 5.07404208e-01
2.22822756e-01 -7.18223810e-01 -2.68052937e-03 2.81912118e-01
3.36748272e-01 8.57382238e-01 -8.07903528e-01 2.04273176e+00
-6.35443211e-01 4.02819246e-01 -3.04615825e-01 -4.27343875e-01
6.25572085e-01 3.51551980e-01 5.75823933e-02 -7.42444336e-01
-1.35080561e-01 4.22547281e-01 -2.37910196e-01 -1.33847341e-01
1.22864306e+00 -1.01818077e-04 -1.93612397e-01 6.01755857e-01
1.73498079e-01 -3.45347941e-01 6.18510306e-01 7.41841435e-01
1.21925187e+00 2.25521009e-02 -1.56741142e-01 2.99964458e-01
4.22672518e-02 2.96159774e-01 1.82262704e-01 1.17410195e+00
4.01805162e-01 6.28931046e-01 3.11863035e-01 5.32532968e-02
-1.15968871e+00 -1.02840519e+00 1.47720337e-01 1.62638283e+00
-8.94732550e-02 -5.50869524e-01 -8.16456974e-01 -3.91355008e-01
3.68237823e-01 1.07152176e+00 -3.63041848e-01 -7.03500956e-02
-7.12009609e-01 -7.76175737e-01 8.79020810e-01 5.59961438e-01
3.89328331e-01 -1.03968871e+00 -2.40958810e-01 5.58796883e-01
-6.55990243e-01 -1.15795541e+00 -5.14111638e-01 1.09507591e-01
-9.37969208e-01 -5.16176403e-01 -8.86554241e-01 -4.55132157e-01
1.52556628e-01 9.57024619e-02 1.74404263e+00 2.28460312e-01
-2.30265662e-01 2.95946926e-01 -5.24003327e-01 -2.49357730e-01
-5.01435220e-01 7.09218800e-01 -3.93125504e-01 -7.05917537e-01
1.54266596e-01 -4.06128615e-01 -7.97942996e-01 2.53477097e-01
-9.24843669e-01 2.52802163e-01 5.50910950e-01 1.02700901e+00
5.79731524e-01 -5.39568424e-01 7.74794459e-01 -1.07621336e+00
9.71627355e-01 -7.18208075e-01 -3.57162058e-01 3.98256660e-01
-6.28192604e-01 3.29933971e-01 1.91616207e-01 -3.78212839e-01
-1.11383867e+00 -5.10005116e-01 -3.71683389e-01 -3.46976250e-01
2.97701895e-01 7.70914853e-01 4.18140560e-01 5.15167773e-01
1.08842719e+00 2.36240074e-01 -2.31045038e-01 -5.43542504e-01
1.08680212e+00 5.88868082e-01 5.61673284e-01 -8.74518037e-01
5.50958157e-01 -9.30050462e-02 -7.12734520e-01 -3.13775748e-01
-1.07041085e+00 -7.03675330e-01 -5.09680435e-02 3.39578241e-01
5.52063823e-01 -1.12051404e+00 -3.67394418e-01 3.14315259e-01
-1.23007870e+00 -9.17059720e-01 -4.90399510e-01 1.13374650e-01
-5.09100199e-01 2.54730403e-01 -1.08512759e+00 -6.77968621e-01
-9.62854683e-01 -7.20604181e-01 1.34466600e+00 2.73779154e-01
-3.58452946e-01 -8.23610961e-01 2.55374014e-01 8.05410326e-01
7.05278099e-01 -2.90956765e-01 9.39319372e-01 -9.69799697e-01
-9.89930749e-01 -2.70724475e-01 -2.98871964e-01 1.32127643e-01
-3.68040681e-01 -4.45654541e-01 -8.18502784e-01 -2.05345646e-01
-6.47813141e-01 -7.50010550e-01 1.21677685e+00 1.37207493e-01
8.57817411e-01 -3.65057290e-01 -3.32782120e-01 3.19343865e-01
1.09307730e+00 7.15427697e-02 5.57663083e-01 3.34895343e-01
4.16201383e-01 3.67784351e-01 7.22816467e-01 3.50223184e-01
5.15577137e-01 7.01517582e-01 -5.83713278e-02 1.43497378e-01
-2.88681597e-01 -6.73828542e-01 1.70404106e-01 9.55161631e-01
2.38898546e-01 -6.92682266e-01 -9.99555349e-01 7.84894705e-01
-2.01542854e+00 -7.39431977e-01 3.99061501e-01 2.13223767e+00
1.43693745e+00 1.61264777e-01 6.96317255e-02 -4.25224930e-01
6.12884343e-01 6.78465217e-02 -7.45600760e-01 -2.48512655e-01
-5.52530820e-03 5.25825083e-01 4.24945444e-01 8.41275871e-01
-8.30567956e-01 1.53006721e+00 5.83455658e+00 1.26280856e+00
-7.89925635e-01 2.71193177e-01 8.91188204e-01 -2.10742280e-01
-5.16226590e-01 2.19234779e-01 -1.14655638e+00 3.65490109e-01
1.31632292e+00 -3.78540367e-01 5.53409815e-01 7.08604455e-01
-2.72825450e-01 -2.66105115e-01 -8.41351628e-01 5.59442759e-01
-9.23923776e-02 -1.55015481e+00 2.52924979e-01 -3.69035751e-01
7.98684359e-01 3.88051987e-01 4.18664478e-02 9.52005446e-01
9.39537048e-01 -9.97911930e-01 5.72818696e-01 4.41016465e-01
7.68414736e-01 -3.76236528e-01 5.18358111e-01 4.74243611e-01
-8.38783264e-01 5.87902553e-02 -3.59158128e-01 1.60542522e-02
6.51689112e-01 6.19845092e-01 -1.46341836e+00 5.58359325e-01
2.79641718e-01 4.93427366e-02 -4.20464307e-01 1.05077016e+00
-2.71490961e-01 8.95715892e-01 -6.14361465e-01 -1.10450536e-01
3.08479577e-01 2.01839074e-01 3.83371413e-01 1.37200844e+00
4.08455312e-01 2.70379394e-01 3.13060999e-01 8.05463910e-01
-3.69780809e-01 1.36395708e-01 -6.18513003e-02 -2.49481201e-01
8.90506864e-01 1.05957496e+00 -5.40218949e-01 -8.19768608e-01
-1.24323793e-01 1.00864840e+00 4.74020839e-01 4.10660774e-01
-7.65109360e-01 -3.93334150e-01 5.17371416e-01 -1.05026830e-03
3.67148519e-01 -1.76442310e-01 -2.24449247e-01 -1.18779528e+00
-6.09537736e-02 -9.27837491e-01 4.97993767e-01 -7.31605589e-01
-1.23912251e+00 7.78134406e-01 1.11798756e-01 -6.63843751e-01
-9.62371707e-01 -1.85396850e-01 -2.39850432e-01 1.21377575e+00
-1.48690140e+00 -1.03517413e+00 -5.51777408e-02 2.46158198e-01
7.26455867e-01 6.43354952e-02 9.73317921e-01 3.73248369e-01
-1.90090001e-01 8.15329790e-01 1.52544975e-01 1.61794648e-01
7.51955211e-01 -1.24064398e+00 1.20074153e+00 6.16694391e-01
3.09963167e-01 7.89551437e-01 5.46603918e-01 -8.38212132e-01
-1.12544262e+00 -9.72475350e-01 1.07872128e+00 -8.39892149e-01
8.01470935e-01 -3.66711020e-01 -9.38263714e-01 7.05766082e-01
1.77005306e-01 -2.82742172e-01 6.67479396e-01 5.29288709e-01
-6.37407601e-01 3.58083278e-01 -6.83940470e-01 6.24917507e-01
1.06331205e+00 -6.79644704e-01 -5.78843474e-01 6.06127501e-01
1.32832551e+00 -9.10162508e-01 -8.22732151e-01 3.38854194e-01
4.54688847e-01 -5.53921521e-01 1.08819127e+00 -5.60966492e-01
4.19438928e-01 -6.64964691e-02 -8.39527249e-02 -1.26786327e+00
-1.90986007e-01 -8.90936911e-01 -3.22643161e-01 1.09340906e+00
1.07253253e+00 -6.52804911e-01 1.01557040e+00 7.56283402e-01
-1.88741580e-01 -8.83130729e-01 -7.58842289e-01 -7.47438312e-01
3.88653040e-01 -4.56689715e-01 6.85144484e-01 5.59469163e-01
2.56292745e-02 7.50612855e-01 -3.47027838e-01 -3.43950093e-01
1.78765357e-01 2.23618820e-01 8.19694340e-01 -9.01250720e-01
-7.19328821e-01 -3.00177515e-01 2.44883463e-01 -1.58622956e+00
-1.63157269e-01 -1.05389667e+00 2.34769195e-01 -1.79838371e+00
2.20275387e-01 -7.43328989e-01 -1.86687559e-01 7.27301478e-01
-5.95534146e-01 1.57488465e-01 4.19187099e-01 2.32165784e-01
-8.85522723e-01 5.55661082e-01 1.14495325e+00 -2.95179822e-02
-4.61540550e-01 -2.05679059e-01 -8.37371171e-01 1.74187779e-01
9.06250358e-01 -3.61183912e-01 -5.01643240e-01 -8.04021955e-01
3.88884872e-01 5.32995522e-01 1.19790681e-01 -8.40305507e-01
3.72889578e-01 2.35365495e-01 4.17985618e-02 -6.10690296e-01
6.44079685e-01 1.71143096e-02 -7.53207058e-02 1.38782769e-01
-7.37156272e-01 6.05290011e-02 5.05921245e-01 5.02627373e-01
-2.42412329e-01 -2.38614857e-01 2.61631668e-01 -3.36852968e-01
-6.48894429e-01 1.88103482e-01 -1.15326427e-01 6.38679743e-01
3.29195946e-01 1.13011315e-01 -8.79020154e-01 -6.80082798e-01
-5.69086492e-01 6.08398557e-01 1.38979837e-01 4.72786963e-01
3.66603643e-01 -1.09238231e+00 -9.75497365e-01 -1.81616396e-01
3.07926945e-02 1.56505451e-01 5.13775110e-01 6.61115766e-01
-4.71005440e-01 8.45695913e-01 3.22192371e-01 -3.93197596e-01
-8.83803189e-01 1.77071333e-01 7.05203414e-02 -9.54348803e-01
-3.77298802e-01 1.24953282e+00 -3.84817347e-02 -8.03238451e-01
1.45540714e-01 -3.64607006e-01 2.08695382e-01 3.22991088e-02
4.74728018e-01 3.89300734e-01 1.02323845e-01 -7.25003183e-02
-2.54803151e-02 5.25534377e-02 -4.81459349e-01 -6.32301748e-01
9.11828995e-01 1.10523209e-01 2.78973226e-02 1.39706910e-01
1.00136948e+00 -1.90912843e-01 -9.65442955e-01 -6.15199864e-01
1.44271076e-01 -1.96537480e-01 -1.39103696e-01 -1.36817765e+00
-5.53467751e-01 6.34588361e-01 1.96164951e-01 9.88608971e-02
9.46768105e-01 7.30080828e-02 1.47854185e+00 6.41330242e-01
4.49107081e-01 -9.00416255e-01 1.16097271e-01 1.03772628e+00
7.59479940e-01 -1.08080411e+00 -2.56531298e-01 -2.11520851e-01
-6.95440829e-01 5.99567413e-01 5.46017110e-01 7.35825002e-02
2.81914294e-01 1.01111345e-01 6.01068437e-02 7.78328702e-02
-1.23682821e+00 -3.93738687e-01 3.50777775e-01 2.18386278e-01
5.28133333e-01 1.86004966e-01 -2.29880050e-01 5.70158124e-01
-5.81611872e-01 2.85440743e-01 9.65336859e-02 9.04345334e-01
-6.32458091e-01 -1.26391387e+00 -8.06610286e-02 6.49624705e-01
-5.13541102e-01 -8.12587857e-01 -1.45569757e-01 5.46152651e-01
-3.28438044e-01 1.04592109e+00 1.66414902e-02 -3.54591548e-01
2.45631188e-01 5.10863185e-01 1.96901649e-01 -8.56010437e-01
-6.83861673e-01 -3.13777365e-02 6.64781630e-01 -5.04591107e-01
1.34635627e-01 -4.71498847e-01 -1.25571311e+00 -2.79597193e-01
-4.27967787e-01 5.76784968e-01 4.39306170e-01 5.42624712e-01
8.04081559e-01 3.99090201e-01 -4.08592299e-02 -5.85412860e-01
-7.71319211e-01 -1.36749196e+00 -7.79040232e-02 2.04591200e-01
3.68595123e-02 -3.93259495e-01 -1.03733294e-01 -1.29627496e-01]
|
[11.469141006469727, 8.141422271728516]
|
ed005107-7d98-4d8c-8ab7-e66d93e96a56
|
dataset-of-natural-language-queries-for-e
|
2302.06355
| null |
https://arxiv.org/abs/2302.06355v1
|
https://arxiv.org/pdf/2302.06355v1.pdf
|
Dataset of Natural Language Queries for E-Commerce
|
Shopping online is more and more frequent in our everyday life. For e-commerce search systems, understanding natural language coming through voice assistants, chatbots or from conversational search is an essential ability to understand what the user really wants. However, evaluation datasets with natural and detailed information needs of product-seekers which could be used for research do not exist. Due to privacy issues and competitive consequences, only few datasets with real user search queries from logs are openly available. In this paper, we present a dataset of 3,540 natural language queries in two domains that describe what users want when searching for a laptop or a jacket of their choice. The dataset contains annotations of vague terms and key facts of 1,754 laptop queries. This dataset opens up a range of research opportunities in the fields of natural language processing and (interactive) information retrieval for product search.
|
['Norbert Fuhr', 'Ahmet Aker', 'Alfred Sliwa', 'Daniel Hienert', 'Dagmar Kern', 'Andrea Papenmeier']
|
2023-02-13
| null | null | null | null |
['conversational-search']
|
['natural-language-processing']
|
[-1.50837600e-01 -1.95260234e-02 -6.94810331e-01 -7.43198335e-01
-5.78351855e-01 -9.18636143e-01 7.34159172e-01 5.62616944e-01
-5.89941204e-01 4.99367982e-01 1.67746991e-01 -4.88464981e-01
-2.68456608e-01 -7.33621418e-01 -1.46086663e-01 -7.39270896e-02
2.38706723e-01 8.81778300e-01 2.30907634e-01 -5.36170602e-01
3.94570380e-01 2.75971860e-01 -1.46101046e+00 4.89042222e-01
7.79133916e-01 1.28590941e+00 4.31392431e-01 3.23558092e-01
-5.80526590e-01 5.72715044e-01 -2.59074658e-01 -8.81462932e-01
1.76854461e-01 -8.57740827e-03 -1.40580344e+00 -1.09980749e-02
8.71417746e-02 -3.16396028e-01 -1.69371188e-01 1.06408572e+00
3.52375686e-01 2.84356028e-01 5.57513759e-02 -1.15179527e+00
-8.00931633e-01 7.27865696e-01 2.25456096e-02 1.92870364e-01
1.10759795e+00 3.17982547e-02 1.49870872e+00 -5.44274509e-01
9.31405962e-01 1.34592605e+00 1.95943400e-01 1.47190258e-01
-1.05332494e+00 -3.52094173e-01 2.21541710e-02 1.67579293e-01
-1.44629622e+00 -2.77965456e-01 6.94341242e-01 -2.17310429e-01
8.73688281e-01 6.26426756e-01 2.31665492e-01 1.27508187e+00
-2.38830552e-01 1.03362966e+00 7.23739386e-01 -5.03094018e-01
1.40349671e-01 1.08890891e+00 6.70153379e-01 4.25058246e-01
8.12969208e-02 -1.73303246e-01 -6.35204136e-01 -8.12891603e-01
4.69074190e-01 3.34189236e-01 2.79133916e-02 -2.15764433e-01
-8.79558384e-01 1.01519310e+00 5.09659871e-02 6.81467831e-01
-5.30239701e-01 -6.02156758e-01 4.41649109e-01 6.15501702e-01
1.86684757e-01 7.20901668e-01 -8.80044281e-01 -3.86250585e-01
-3.98264289e-01 5.86466789e-01 1.73419976e+00 1.54074323e+00
7.21755385e-01 -1.03741062e+00 6.64816946e-02 1.04232907e+00
3.93791705e-01 2.96550125e-01 5.90469182e-01 -9.10617232e-01
3.93906802e-01 9.50351834e-01 4.89614338e-01 -1.30274558e+00
-2.07822666e-01 -7.31914863e-02 -3.19851190e-01 -7.88342714e-01
5.92793405e-01 1.51518255e-01 -1.69150874e-01 1.08088934e+00
6.55046999e-02 -8.03977728e-01 -6.96494132e-02 1.05682743e+00
1.12872660e+00 3.12272429e-01 9.12722051e-02 -1.67403206e-01
2.16531682e+00 -6.64971471e-01 -1.02071893e+00 -3.96565616e-01
5.60245335e-01 -1.12261260e+00 1.45639741e+00 4.34467345e-01
-8.75247955e-01 -2.37330705e-01 -3.45178872e-01 -4.37109232e-01
-1.01333392e+00 -1.36995330e-01 1.12282825e+00 6.62168562e-01
-4.08539236e-01 2.04310969e-01 -3.02292287e-01 -9.99294698e-01
6.01768717e-02 4.92174812e-02 2.72291861e-02 -2.79214144e-01
-1.66825318e+00 7.08810329e-01 1.58661187e-01 -7.48170242e-02
-1.76511303e-01 -5.28768122e-01 -8.19664657e-01 2.13081941e-01
1.02360165e+00 -4.67258990e-01 1.71788692e+00 -3.53053361e-01
-1.11680949e+00 1.14308298e+00 -3.49340975e-01 -4.20099616e-01
3.47043961e-01 -2.32594833e-01 -6.38388515e-01 -1.02540493e-01
4.77238774e-01 2.13744119e-01 3.78872573e-01 -9.71689999e-01
-7.94127584e-01 -6.76384926e-01 3.09977472e-01 1.01046205e-01
3.07573425e-03 3.96840602e-01 -8.61509621e-01 -2.17395350e-01
1.51835918e-01 -7.24835932e-01 -1.62216067e-01 -2.74372637e-01
-4.22830582e-01 -7.39092767e-01 7.05784380e-01 -6.24943078e-01
1.40609264e+00 -2.08127952e+00 -6.21923089e-01 1.52358934e-01
6.93578795e-02 1.32971421e-01 2.60750324e-01 9.08029556e-01
4.28169936e-01 5.07395685e-01 4.55475122e-01 1.19045623e-01
5.72512090e-01 2.21163005e-01 -4.59887892e-01 -2.36986145e-01
-3.00555110e-01 1.07332623e+00 -9.38647389e-01 -6.70308828e-01
5.42594865e-02 1.74287651e-02 -4.55846339e-01 9.28731710e-02
-3.79107982e-01 1.17565058e-01 -1.13456666e+00 1.01007712e+00
3.65559846e-01 -2.67786562e-01 6.67167827e-02 -1.07280329e-01
-6.38527498e-02 6.58095121e-01 -1.05351567e+00 1.61194873e+00
-6.39195085e-01 3.30501080e-01 3.96351486e-01 -4.96802956e-01
6.53496802e-01 1.86220393e-01 3.75260890e-01 -1.09285641e+00
1.81233034e-01 2.66686559e-01 -6.00632966e-01 -9.14067268e-01
5.92240810e-01 2.31273413e-01 -4.76642638e-01 4.92821902e-01
-1.26138344e-01 -1.42206103e-01 4.15442795e-01 3.66816521e-02
1.07971263e+00 -4.61459696e-01 3.61613393e-01 -4.20555234e-01
6.29161954e-01 3.24801177e-01 2.09164202e-01 1.21551418e+00
-3.32043231e-01 -2.55004633e-02 1.52804419e-01 -8.76907587e-01
-5.21407962e-01 -6.86363101e-01 -2.38545597e-01 1.62403727e+00
3.09288234e-01 -5.82392156e-01 -3.89338225e-01 -2.91979074e-01
2.44740210e-02 9.02738929e-01 2.16129664e-02 9.70528275e-02
-1.73926726e-01 -5.37791848e-02 1.40419900e-01 -1.69667542e-01
7.75448322e-01 -1.25535691e+00 -5.13868093e-01 1.99125960e-01
-7.06587791e-01 -1.56596720e+00 -7.63266802e-01 -1.80352870e-02
-7.49859512e-01 -1.02967227e+00 -4.02089894e-01 -9.36523378e-01
1.58383146e-01 2.98923045e-01 1.59594381e+00 1.60192430e-01
-5.85354924e-01 5.24020135e-01 -4.07578707e-01 -3.31305176e-01
-1.32953882e-01 4.29903716e-01 -1.15204722e-01 -1.65945634e-01
1.44032228e+00 -1.85108975e-01 -5.88175237e-01 8.56698632e-01
-8.66911829e-01 -5.90908229e-01 3.11891288e-01 6.22877896e-01
6.89957812e-02 4.67783928e-01 1.41983479e-01 -1.10068333e+00
1.31967282e+00 -6.95384443e-01 -6.48958325e-01 3.26744378e-01
-9.13618207e-01 9.32281539e-02 2.23920956e-01 -3.13088685e-01
-1.19682050e+00 3.39409225e-02 -9.18487981e-02 2.71873504e-01
-7.27020860e-01 6.78909659e-01 -6.08853921e-02 3.07297856e-01
7.85541236e-01 5.93111105e-02 -3.57428521e-01 -9.58190084e-01
4.23205197e-01 1.12114334e+00 3.33786637e-01 -6.21451497e-01
1.83196083e-01 1.46862343e-01 -7.93853223e-01 -1.22497511e+00
-7.28341043e-01 -1.37349641e+00 -2.86765426e-01 1.98646605e-01
5.03869414e-01 -4.25011903e-01 -1.58158171e+00 8.33425205e-03
-1.10704029e+00 2.73260862e-01 -4.81178723e-02 1.44386396e-01
-3.37490052e-01 3.90887201e-01 -6.76616371e-01 -1.21767056e+00
-3.51184517e-01 -9.49757695e-01 1.15958440e+00 2.67537117e-01
-7.72837222e-01 -8.22702408e-01 -3.80640030e-01 1.04261303e+00
6.00404978e-01 -4.96574610e-01 1.02551496e+00 -1.18142784e+00
-6.04793072e-01 -8.13858867e-01 -1.31175891e-01 -2.43335530e-01
2.91433573e-01 -5.15654504e-01 -7.21041381e-01 1.69123352e-01
4.28148180e-01 -4.42281365e-01 3.41237485e-01 1.46791697e-01
9.73732531e-01 -7.46959686e-01 -4.48103577e-01 -2.95538336e-01
1.29087627e+00 5.52619874e-01 4.66182530e-01 2.96274722e-01
-9.14212242e-02 1.10724247e+00 7.71337926e-01 6.32675767e-01
5.04255712e-01 7.70442843e-01 7.75029510e-02 3.98090303e-01
7.88517773e-01 -5.65209568e-01 -1.61174238e-01 -2.53231693e-02
2.55454451e-01 -2.77806878e-01 -8.97157311e-01 5.64987183e-01
-1.66125965e+00 -8.81918192e-01 2.27438897e-01 1.99480045e+00
9.55626488e-01 2.12105393e-01 6.10113516e-02 -7.42289349e-02
5.87449133e-01 -1.09281614e-01 -5.80775321e-01 -3.69428575e-01
2.60408491e-01 -4.60893102e-02 4.07200485e-01 5.36177039e-01
-9.05150414e-01 9.55692053e-01 6.06461811e+00 7.45125771e-01
-3.51255536e-01 -1.42861709e-01 7.32279658e-01 3.90496194e-01
-3.21132779e-01 1.37843862e-02 -1.01995945e+00 3.65485430e-01
7.27258265e-01 -1.99016437e-01 8.16449702e-01 1.15595639e+00
1.89411759e-01 -4.85105544e-01 -1.46473885e+00 1.48840249e+00
-2.95694023e-01 -1.10778260e+00 -1.65113762e-01 8.26582983e-02
4.58571091e-02 -3.15816581e-01 -7.29903132e-02 5.67714095e-01
3.32835436e-01 -7.64045060e-01 1.97524786e-01 4.52197880e-01
5.45426048e-02 -2.52142340e-01 8.59606504e-01 1.08151197e+00
-7.97110856e-01 -1.40118822e-01 1.80253610e-02 -6.52164891e-02
2.02182487e-01 3.36140603e-01 -9.19082403e-01 1.41417027e-01
1.04822898e+00 2.18210056e-01 -4.01323467e-01 6.83745146e-01
4.47632819e-01 2.00542852e-01 -5.97263217e-01 -6.38617039e-01
2.39491358e-01 -4.19191241e-01 3.46939474e-01 1.04991257e+00
-9.36103091e-02 3.02220672e-01 1.58022761e-01 1.28921366e+00
-5.76485693e-02 4.11897361e-01 -9.31449354e-01 -6.55173898e-01
4.70311761e-01 1.18017292e+00 -7.09766924e-01 -1.57992885e-01
-5.87762654e-01 1.13057685e+00 -3.28656256e-01 4.31009591e-01
-2.06047997e-01 -4.91287827e-01 7.25714087e-01 4.88104701e-01
-1.58972695e-01 -6.20589666e-02 2.03315131e-02 -1.05636346e+00
3.56727332e-01 -1.13185787e+00 6.43515587e-01 -6.38553023e-01
-1.63004732e+00 4.76396322e-01 6.49365708e-02 -6.17735028e-01
-5.98599970e-01 -5.15962303e-01 -8.04906264e-02 8.88625503e-01
-1.12398887e+00 -5.33633351e-01 -1.92138672e-01 4.96267766e-01
8.63446534e-01 -2.23792996e-02 1.01543033e+00 7.01685786e-01
-2.20649480e-03 1.84665814e-01 -4.30406809e-01 1.92068860e-01
6.03101075e-01 -7.86665320e-01 3.48798901e-01 -1.94784701e-01
2.51853466e-01 1.28510559e+00 8.04380715e-01 -4.96468812e-01
-1.70312262e+00 -1.80260256e-01 1.76082909e+00 -7.61374772e-01
6.96622312e-01 -5.80117106e-01 -7.48724163e-01 5.08695066e-01
3.51809025e-01 -4.85833079e-01 8.27141166e-01 4.11160082e-01
-1.66220516e-02 -7.56641999e-02 -1.34044707e+00 6.16855443e-01
8.80737126e-01 -8.51900160e-01 -6.75396800e-01 7.49692082e-01
7.00902998e-01 -2.32309714e-01 -8.19361508e-01 2.30570938e-02
5.55114925e-01 -6.38448000e-01 1.12619781e+00 -7.96673298e-01
-1.76759735e-01 3.22596163e-01 -1.84265867e-01 -6.83354199e-01
1.99330419e-01 -9.67489600e-01 3.34151030e-01 1.33063924e+00
6.90371692e-01 -7.34158158e-01 7.64762819e-01 1.82830453e+00
6.14028513e-01 -4.05783534e-01 -5.36803603e-01 -4.45346415e-01
-4.90119725e-01 -5.65519333e-01 7.45939434e-01 9.38299000e-01
4.09454733e-01 5.84642231e-01 4.11038956e-04 -2.02938810e-01
2.85211533e-01 3.39562505e-01 4.21942592e-01 -1.44574976e+00
-5.78573020e-03 -4.78084713e-01 -1.36104617e-02 -1.60225523e+00
-8.17038119e-02 -6.65267169e-01 -1.37221158e-01 -1.21914840e+00
7.42832795e-02 -4.09361035e-01 3.18350971e-01 2.13485315e-01
4.95557547e-01 -4.50511217e-01 -1.06762402e-01 2.68801570e-01
-8.41065109e-01 1.47072643e-01 1.11093330e+00 -1.39979318e-01
-5.40725052e-01 6.47002220e-01 -9.95034754e-01 7.12792754e-01
6.82875395e-01 -1.99486718e-01 -4.30662006e-01 -1.89103425e-01
7.02104390e-01 3.10493141e-01 1.88004762e-01 -1.13552347e-01
5.70536971e-01 -2.38794044e-01 -2.26124421e-01 -5.30326962e-01
2.63052404e-01 -1.47773719e+00 3.52020003e-02 -1.47474650e-02
-8.57085645e-01 -9.20152962e-02 -2.29925871e-01 5.47521174e-01
-4.87367868e-01 -6.66936755e-01 4.06461619e-02 -7.37753093e-01
-7.09579110e-01 2.41523728e-01 -6.55081868e-01 2.77783871e-01
5.85289657e-01 -7.12065026e-02 -5.82284294e-02 -8.14975142e-01
-8.28793228e-01 5.92444718e-01 -1.60659645e-02 8.46099794e-01
3.57426643e-01 -1.18394303e+00 -9.62282717e-02 2.72767931e-01
4.09364074e-01 -2.61285692e-01 -3.34852278e-01 2.75376737e-01
-2.00572833e-01 1.23023260e+00 2.38388106e-01 -3.54740471e-01
-9.36086476e-01 6.64977074e-01 -5.50531186e-02 -1.64844841e-01
-1.45349607e-01 6.30011618e-01 -3.13301593e-01 -6.69180572e-01
6.86862767e-01 -5.76402187e-01 -2.66267329e-01 2.44010121e-01
5.69139361e-01 2.55411744e-01 8.42774138e-02 -4.60599661e-01
-3.09076488e-01 3.79521959e-02 -2.80057937e-01 -1.17006518e-01
9.29125071e-01 -7.11782157e-01 -3.01510632e-01 4.63843733e-01
1.36621797e+00 -2.81536698e-01 -1.24687791e-01 -8.46751809e-01
8.18379939e-01 -7.68980920e-01 -8.32918063e-02 -7.58987665e-01
-7.86647856e-01 3.03845465e-01 3.98019105e-01 1.05882895e+00
9.41992044e-01 6.32237077e-01 1.01896715e+00 1.24479735e+00
8.81353974e-01 -1.46361423e+00 -2.35332131e-01 5.28124750e-01
8.99692655e-01 -1.81797016e+00 -4.86148268e-01 -6.54361188e-01
-7.81668186e-01 9.70421314e-01 2.54419327e-01 5.83276093e-01
1.04104304e+00 -1.42594650e-01 1.46299690e-01 -6.89878166e-01
-6.27847672e-01 -4.03944403e-01 1.75011665e-01 2.17041761e-01
6.56974435e-01 -2.52101481e-01 -5.60717821e-01 1.10664439e+00
-2.99803674e-01 1.42052129e-01 -2.44088754e-01 1.13634300e+00
-3.07896614e-01 -1.23872268e+00 -6.27756640e-02 5.35685420e-01
-6.93776906e-01 -2.73597747e-01 -7.64815867e-01 4.63946521e-01
-4.28853035e-01 1.55739379e+00 -2.95709670e-01 -4.65244092e-02
6.21965110e-01 5.52883208e-01 -1.37732118e-01 -6.49183750e-01
-5.84020197e-01 -2.82459825e-01 3.93048376e-01 -7.81094074e-01
-2.50453383e-01 -5.56612849e-01 -8.96400988e-01 -3.34719539e-01
-4.51321065e-01 5.26290715e-01 7.71266937e-01 7.49575078e-01
4.27892685e-01 -5.17357230e-01 2.92821676e-01 -1.88029289e-01
-5.44922113e-01 -9.19578850e-01 -7.08568215e-01 6.81917131e-01
1.49122611e-01 -4.71871138e-01 -3.45125288e-01 7.52857476e-02]
|
[12.181660652160645, 7.810238361358643]
|
2efaa7f5-330d-4633-b3c9-a20054431ee3
|
machine-learning-applications-in-diagnosis
|
2203.02794
| null |
https://arxiv.org/abs/2203.02794v3
|
https://arxiv.org/pdf/2203.02794v3.pdf
|
Machine Learning Applications in Lung Cancer Diagnosis, Treatment and Prognosis
|
The recent development of imaging and sequencing technologies enables systematic advances in the clinical study of lung cancer. Meanwhile, the human mind is limited in effectively handling and fully utilizing the accumulation of such enormous amounts of data. Machine learning-based approaches play a critical role in integrating and analyzing these large and complex datasets, which have extensively characterized lung cancer through the use of different perspectives from these accrued data. In this article, we provide an overview of machine learning-based approaches that strengthen the varying aspects of lung cancer diagnosis and therapy, including early detection, auxiliary diagnosis, prognosis prediction and immunotherapy practice. Moreover, we highlight the challenges and opportunities for future applications of machine learning in lung cancer.
|
['Yuan Luo', 'Guoqian Jiang', 'Ping Yang', 'Xin Wu', 'Yawei Li']
|
2022-03-05
| null | null | null | null |
['lung-cancer-diagnosis']
|
['medical']
|
[ 3.79813582e-01 -4.83225018e-01 -1.00978518e+00 1.49525225e-01
-1.05034649e+00 -4.96348143e-01 3.39576989e-01 5.46359122e-01
-4.08845842e-01 7.85657704e-01 2.40813434e-01 -7.21116662e-01
-2.66857356e-01 -6.07319176e-01 8.56929272e-02 -1.19203377e+00
1.97469950e-01 9.11986530e-01 2.53066242e-01 1.07076682e-01
-3.59617442e-01 8.67994785e-01 -9.97197509e-01 5.05752683e-01
7.10264325e-01 5.64705431e-01 6.73231602e-01 9.71240163e-01
-2.67828822e-01 8.46943438e-01 1.73646942e-01 -6.06278814e-02
-2.77246088e-01 -4.91136879e-01 -7.29391754e-01 -9.51208696e-02
-1.09824175e-02 -3.78035426e-01 -4.52177942e-01 3.67595851e-01
6.04853809e-01 -5.71385443e-01 4.97620255e-01 -7.34177947e-01
9.80623737e-02 4.50504236e-02 -1.84705168e-01 5.33607066e-01
-7.05058947e-02 3.94272745e-01 7.76591420e-01 -9.37959909e-01
7.00217903e-01 3.52543861e-01 5.80830276e-01 9.05539095e-01
-8.61023724e-01 -5.75069189e-02 -4.48770463e-01 2.71164209e-01
-9.48772132e-01 -5.30768335e-02 5.83445989e-02 -6.41091645e-01
7.55955935e-01 5.74925661e-01 9.68356609e-01 9.71793056e-01
5.27321219e-01 7.01598585e-01 9.59396005e-01 -3.68976802e-01
-2.09561393e-01 2.70794183e-01 1.14780910e-01 7.94802785e-01
5.06103933e-01 4.36709434e-01 -3.49453151e-01 -4.98054892e-01
4.65889484e-01 7.53772557e-01 -3.31507444e-01 -1.18244588e-01
-1.49887192e+00 8.13578665e-01 2.32383251e-01 6.04564786e-01
-3.72544348e-01 3.13681029e-02 6.70102656e-01 -3.43418151e-01
6.66295439e-02 -1.58784151e-01 -4.12181675e-01 1.74458802e-01
-8.53298008e-01 -3.41669947e-01 4.65218186e-01 4.77987915e-01
1.63067430e-02 -2.61090964e-01 -3.94266337e-01 4.89321858e-01
1.71221986e-01 3.78224522e-01 8.62411618e-01 -5.53996623e-01
-8.32702890e-02 6.00816965e-01 -2.70265847e-01 -7.43330270e-02
-7.79019952e-01 -5.00294209e-01 -1.01573455e+00 -2.35961884e-01
4.44855362e-01 3.69340897e-01 -5.59929252e-01 1.18111253e+00
6.02790952e-01 3.20391983e-01 -8.53325520e-03 5.33917844e-01
7.31973410e-01 -2.23972425e-01 5.67452431e-01 -4.06063944e-01
1.69536221e+00 -1.01490057e+00 -5.60607255e-01 1.70983285e-01
1.26996696e+00 -4.37267095e-01 6.89870358e-01 1.87445000e-01
-6.05093181e-01 -6.61172420e-02 -4.76601422e-01 -1.24574482e-01
-3.72249037e-01 3.58647883e-01 1.05323005e+00 7.27074444e-01
-8.10836554e-01 2.11473703e-01 -1.01492310e+00 -8.64153445e-01
8.03259134e-01 5.50758302e-01 -5.63281476e-01 -5.17942309e-01
-8.99538040e-01 9.17971849e-01 5.41830838e-01 -6.62257597e-02
-1.01166797e+00 -1.09557903e+00 -1.30603433e-01 -1.89065337e-01
5.20094931e-01 -1.38679075e+00 1.17469132e+00 -1.40645891e-01
-1.05045533e+00 1.14267480e+00 -3.83142412e-01 -3.94265771e-01
3.72168839e-01 2.29109734e-01 -1.67519346e-01 4.41985816e-01
-1.81974068e-01 2.68745482e-01 1.23427592e-01 -6.06117308e-01
-1.15262270e+00 -7.50213206e-01 -8.20641696e-01 3.03855333e-02
-3.85280371e-01 -3.91691849e-02 -2.57200956e-01 -3.76724392e-01
-1.72671735e-01 -9.32515204e-01 -6.41038418e-01 5.06517649e-01
1.57424271e-01 -1.59210786e-01 9.37841475e-01 -5.18210948e-01
1.10442567e+00 -1.87340641e+00 4.51778322e-02 -5.47581203e-02
7.89235353e-01 4.84574825e-01 3.14018250e-01 3.98158669e-01
9.95648503e-02 4.96210307e-01 1.05657175e-01 2.02331707e-01
-6.14146888e-01 2.95692742e-01 1.46398485e-01 4.04440522e-01
-1.09425455e-01 1.54992807e+00 -1.01293766e+00 -8.36826205e-01
4.19849932e-01 2.93382227e-01 -1.38504118e-01 3.33806902e-01
-2.01851577e-01 9.30687547e-01 -8.00296843e-01 1.04980755e+00
1.21728271e-01 -6.39774501e-01 7.73900986e-01 1.79578841e-01
-2.62063760e-02 -4.80188318e-02 4.07873094e-03 1.35515428e+00
-3.02016377e-01 1.00445829e-01 -7.07432115e-03 -6.44826651e-01
2.36178041e-01 7.67417490e-01 9.40477610e-01 -3.36274862e-01
1.33761823e-01 3.71078938e-01 2.02511281e-01 -7.47311115e-01
-3.93228292e-01 -6.55543089e-01 5.21240115e-01 4.22118664e-01
-1.31342471e-01 -1.34930685e-01 8.84265378e-02 9.36995074e-02
1.37971330e+00 -5.88719964e-01 9.05704498e-01 1.10196486e-01
8.03461015e-01 3.06309521e-01 3.32507670e-01 5.07273257e-01
-5.94469845e-01 2.23548144e-01 3.59091550e-01 -3.17793846e-01
-9.06483471e-01 -9.81156170e-01 -4.58010584e-01 7.68777788e-01
-5.35313308e-01 -9.76420045e-02 -1.55743957e-01 -1.05267668e+00
2.25855693e-01 2.47692272e-01 -7.42428601e-01 -6.18113093e-02
-4.52657342e-01 -1.21871543e+00 6.45393550e-01 7.56010711e-01
-6.80410266e-02 -5.33388317e-01 -3.86248916e-01 3.80610645e-01
-1.39348686e-01 -1.09467483e+00 2.22262945e-02 3.54707271e-01
-1.34126604e+00 -1.54543984e+00 -6.94907725e-01 -6.47074580e-01
4.67482030e-01 6.43615127e-01 9.71550941e-01 6.40444219e-01
-1.07063186e+00 3.72315198e-01 -1.12309881e-01 -6.62009597e-01
-6.57034755e-01 -2.61513679e-03 -1.70610234e-01 -4.46573108e-01
3.58807236e-01 -1.73099324e-01 -4.08932805e-01 3.99350822e-02
-7.95518816e-01 1.16789684e-01 1.20503199e+00 1.29751182e+00
1.00624144e+00 2.17053846e-01 6.36348426e-01 -1.19435358e+00
1.60364717e-01 -1.00997996e+00 -2.20508352e-01 5.04195273e-01
-5.17372787e-01 -3.38576645e-01 4.36852843e-01 -1.08648889e-01
-1.03166389e+00 7.76743097e-03 -2.43576244e-01 -8.85228906e-03
-2.32103392e-01 6.27070427e-01 8.98220614e-02 -2.82359213e-01
4.27859247e-01 2.54258871e-01 3.53365928e-01 -1.15672790e-01
-1.38512403e-01 4.64817405e-01 3.23652804e-01 -1.16203047e-01
7.04325557e-01 8.35124433e-01 6.74052060e-01 -8.18506479e-01
-1.13897145e+00 -9.35097992e-01 -8.57695997e-01 -2.53596455e-01
8.47096205e-01 -6.14181042e-01 -4.69146609e-01 5.08395672e-01
-4.51436579e-01 -2.31980056e-01 -5.24963737e-01 8.79722834e-01
-3.43675852e-01 3.24224561e-01 -6.45674050e-01 -4.10656780e-01
-5.40831208e-01 -9.37262535e-01 1.06730890e+00 1.35054022e-01
-1.09561533e-01 -1.35227799e+00 4.79344547e-01 7.25825727e-01
5.09023011e-01 6.66701719e-02 1.44408977e+00 -9.78730023e-01
-8.02044809e-01 -5.17862260e-01 -1.16215684e-01 -1.75647065e-01
4.14579391e-01 1.83606267e-01 -7.59234786e-01 -2.91344766e-02
1.85961336e-01 -5.22842944e-01 9.29150224e-01 3.42565000e-01
1.29987884e+00 2.59186924e-01 -1.03549790e+00 5.29318750e-01
1.43111968e+00 8.75271410e-02 3.71983886e-01 3.54688317e-02
5.89950681e-01 5.19263148e-01 5.49308956e-01 2.83567846e-01
7.31684789e-02 2.19454005e-01 4.66312617e-01 -2.38667175e-01
-3.04470271e-01 -2.19582096e-01 -4.35565948e-01 4.97911334e-01
-1.40043676e-01 -4.91621256e-01 -1.10096824e+00 2.19609171e-01
-1.36850405e+00 -9.67545688e-01 -4.34090078e-01 2.07082200e+00
8.08764994e-01 -2.25914996e-02 -8.01362544e-02 3.27632017e-02
5.17968833e-01 -2.57528484e-01 -5.42477846e-01 1.96953714e-01
-1.11602079e-02 1.86218590e-01 2.50181049e-01 1.36658959e-02
-8.03618252e-01 6.68693900e-01 8.03980350e+00 1.07132292e+00
-1.15849936e+00 2.42393270e-01 7.14824378e-01 5.75313009e-02
-2.33207002e-01 -1.59694836e-01 -7.69874990e-01 1.62327677e-01
9.79496002e-01 -2.63460428e-01 -1.18702818e-02 5.51837385e-01
4.87444907e-01 -4.96343136e-01 -1.18308413e+00 6.14144623e-01
-2.72395730e-01 -1.69381046e+00 -1.33054689e-01 5.25436461e-01
5.36370695e-01 4.24354374e-01 1.78321362e-01 -1.12083457e-01
8.47768635e-02 -1.15041506e+00 -3.52641225e-01 7.96689093e-01
9.74290192e-01 -1.38408676e-01 1.16439867e+00 6.03310704e-01
-9.65640545e-01 -6.96275234e-02 -5.31587601e-02 3.75093460e-01
-1.12500772e-01 8.90726745e-01 -1.54819703e+00 8.12607169e-01
2.36743569e-01 5.30291319e-01 -7.09481299e-01 9.73959148e-01
1.60596520e-02 7.04505980e-01 -1.92359701e-01 -1.95762850e-02
-2.11780965e-01 5.57289459e-02 1.91791859e-02 1.06700075e+00
3.47822368e-01 3.99641782e-01 2.13368908e-01 2.80880719e-01
1.98481902e-01 1.71078920e-01 -3.91969949e-01 -6.58736050e-01
3.37557942e-01 1.63617778e+00 -6.76860809e-01 -4.14761275e-01
-7.99773335e-01 4.32889253e-01 2.34289914e-01 -2.12974712e-01
-7.40449309e-01 7.55402267e-01 2.69523501e-01 3.59657407e-01
-6.75740987e-02 -6.53260201e-03 -1.81580141e-01 -8.92124474e-01
-6.28604949e-01 -8.41978312e-01 1.00131917e+00 -2.59913534e-01
-1.12955737e+00 1.76476568e-01 -4.08904582e-01 -9.20037627e-01
-7.56156147e-02 -7.11008608e-01 -7.65007973e-01 5.11865854e-01
-1.60877740e+00 -1.36629915e+00 -3.62501711e-01 2.31396258e-01
4.82203454e-01 -2.62796938e-01 1.15589535e+00 2.21126806e-02
-7.98967659e-01 2.61570930e-01 5.77630341e-01 -3.49980652e-01
6.75251842e-01 -1.06662536e+00 -5.09440839e-01 -9.37700048e-02
-3.29986095e-01 2.86114365e-02 1.80980172e-02 -5.95739484e-01
-1.68676615e+00 -1.20524144e+00 8.30766678e-01 -6.56364977e-01
7.81119883e-01 1.60700992e-01 -7.54329383e-01 6.67822838e-01
-3.22239071e-01 3.53845477e-01 1.52112138e+00 -2.99795359e-01
1.02133535e-01 5.96964471e-02 -1.28673935e+00 4.49776083e-01
7.38276780e-01 -4.85504150e-01 -9.40211564e-02 7.20274746e-01
2.86404252e-01 -5.62026873e-02 -1.14795101e+00 7.04620123e-01
6.34953082e-01 -8.10050786e-01 1.06473589e+00 -7.93094397e-01
4.62207273e-02 -7.14052022e-02 1.93393305e-01 -6.05059862e-01
-6.11404419e-01 1.24922216e-01 -2.68803202e-02 6.22363389e-01
3.44332069e-01 -5.57685435e-01 1.33486807e+00 3.20314616e-01
4.12238715e-03 -1.44556510e+00 -9.24574375e-01 -3.16282511e-01
2.23640591e-01 -3.72118577e-02 4.11736369e-02 5.57489812e-01
1.21909022e-01 -2.63553977e-01 1.87103853e-01 -2.33126119e-01
5.40113568e-01 -1.03618175e-01 4.09877628e-01 -1.25488162e+00
-2.47890919e-01 -4.01392967e-01 -3.19688052e-01 -2.13415936e-01
-4.94848564e-03 -1.11195767e+00 -5.42673707e-01 -1.49898756e+00
1.15785336e+00 -1.80210173e-01 -4.23327714e-01 1.61374062e-01
-4.37033057e-01 7.67448023e-02 -1.99189633e-01 5.91975451e-01
-3.94947052e-01 2.27672666e-01 1.35808945e+00 -8.70068744e-02
3.17933023e-01 2.49513581e-01 -6.20800972e-01 8.13717782e-01
9.14927602e-01 -5.07250667e-01 -1.94683552e-01 5.82107250e-03
-2.53651977e-01 3.60688537e-01 4.75724876e-01 -7.52562642e-01
4.01401103e-01 -7.51208127e-01 6.43626392e-01 -1.02465987e+00
1.19740508e-01 -1.03402627e+00 4.39525396e-01 1.39780486e+00
-1.15157887e-01 -3.25671613e-01 1.95099767e-02 6.74743295e-01
-1.33063287e-01 -2.64639199e-01 8.53185594e-01 -4.78947103e-01
-6.89967871e-01 6.31545663e-01 -8.22337270e-01 -2.42108449e-01
1.54162443e+00 -2.22519398e-01 -4.70913559e-01 1.94791853e-01
-9.47775066e-01 2.78987706e-01 3.10355902e-01 -2.63522625e-01
3.65227610e-01 -1.15555930e+00 -9.81203318e-01 1.22489417e-02
1.85547426e-01 -1.17711045e-01 7.41036773e-01 1.46921027e+00
-5.34344733e-01 1.06177521e+00 7.93253556e-02 -6.68464959e-01
-1.55776978e+00 9.67325807e-01 5.88267624e-01 -1.01420426e+00
-3.53819609e-01 5.70198596e-01 5.53906381e-01 -2.27332249e-01
-2.87918374e-02 2.73200780e-01 -2.87925005e-01 -6.07989095e-02
3.10176820e-01 5.19941986e-01 1.80715874e-01 -2.12972566e-01
-4.43784386e-01 1.81182086e-01 -2.97494859e-01 2.30751604e-01
1.08260214e+00 -6.20744303e-02 -4.09762710e-01 3.18230748e-01
8.76337349e-01 -6.91312104e-02 -5.21397293e-01 -1.61669627e-01
2.10954234e-01 -2.65028119e-01 -1.74738690e-02 -9.62973058e-01
-8.74020338e-01 8.64081919e-01 4.94941980e-01 -3.26864809e-01
1.19969368e+00 3.76315206e-01 7.84628630e-01 1.98536023e-01
2.00472534e-01 -6.89960182e-01 2.19465658e-01 1.57231927e-01
4.42647725e-01 -1.30861115e+00 1.69641450e-01 -9.14975822e-01
-4.19956923e-01 1.27050555e+00 6.54591978e-01 5.28350294e-01
6.60950243e-01 5.97158849e-01 9.40606818e-02 -9.78967324e-02
-1.54093957e+00 -4.31969464e-01 -1.01021819e-01 6.47635043e-01
6.01676226e-01 2.07936898e-01 -5.67747474e-01 7.13878512e-01
2.89722711e-01 3.90928060e-01 1.83990061e-01 1.08824277e+00
-8.49106073e-01 -1.52048862e+00 -4.95016783e-01 7.48541951e-01
-7.28916049e-01 -1.66843891e-01 -5.97648919e-01 1.00479567e+00
6.55625388e-02 4.23905104e-01 -3.60627383e-01 -1.11676604e-01
-1.93327904e-01 2.83615410e-01 5.33805788e-01 -6.84019208e-01
-3.51414621e-01 3.44495505e-01 -1.45072773e-01 -1.16335079e-01
-3.72141600e-01 -8.11939001e-01 -1.38266599e+00 -9.01853740e-02
-5.05317926e-01 2.12226901e-03 5.39395690e-01 1.01129150e+00
6.18132651e-02 7.08604872e-01 7.92417645e-01 -9.41522643e-02
-6.26888037e-01 -5.05097806e-01 -5.87913811e-01 -3.06624144e-01
-1.76396035e-02 -1.98097005e-01 -3.87726158e-01 -8.61089583e-03]
|
[15.256296157836914, -2.640519618988037]
|
630d8be5-47b1-4521-8bfd-3fd8b7081b2c
|
miriam-exploiting-elastic-kernels-for-real
|
2307.04339
| null |
https://arxiv.org/abs/2307.04339v1
|
https://arxiv.org/pdf/2307.04339v1.pdf
|
Miriam: Exploiting Elastic Kernels for Real-time Multi-DNN Inference on Edge GPU
|
Many applications such as autonomous driving and augmented reality, require the concurrent running of multiple deep neural networks (DNN) that poses different levels of real-time performance requirements. However, coordinating multiple DNN tasks with varying levels of criticality on edge GPUs remains an area of limited study. Unlike server-level GPUs, edge GPUs are resource-limited and lack hardware-level resource management mechanisms for avoiding resource contention. Therefore, we propose Miriam, a contention-aware task coordination framework for multi-DNN inference on edge GPU. Miriam consolidates two main components, an elastic-kernel generator, and a runtime dynamic kernel coordinator, to support mixed critical DNN inference. To evaluate Miriam, we build a new DNN inference benchmark based on CUDA with diverse representative DNN workloads. Experiments on two edge GPU platforms show that Miriam can increase system throughput by 92% while only incurring less than 10\% latency overhead for critical tasks, compared to state of art baselines.
|
['Guoliang Xing', 'Nan Guan', 'Neiwen Ling', 'Zhihe Zhao']
|
2023-07-10
| null | null | null | null |
['autonomous-driving', 'management']
|
['computer-vision', 'miscellaneous']
|
[-4.28878725e-01 -4.13755864e-01 -3.23214084e-01 -5.00290036e-01
-3.85417551e-01 -4.30711776e-01 4.34295148e-01 -2.03821093e-01
-7.77530074e-01 5.92629254e-01 -1.24295302e-01 -8.27687979e-01
3.40069830e-01 -7.49399722e-01 -7.66918361e-01 -5.35939574e-01
2.40810007e-01 5.98084390e-01 7.89423883e-01 2.87828714e-01
-2.35638127e-01 3.84185523e-01 -1.84064186e+00 3.18765163e-01
6.76686704e-01 9.60111499e-01 4.16807264e-01 9.97213781e-01
-2.38985583e-01 1.16945195e+00 -7.70624459e-01 5.97905107e-02
1.95894510e-01 2.39107162e-01 -5.43542981e-01 -4.87412006e-01
1.09587932e+00 -9.91849542e-01 -3.25453281e-01 7.81319559e-01
7.54005849e-01 -3.19930092e-02 6.28867149e-02 -1.62439418e+00
1.95700541e-01 4.45757776e-01 -6.43079340e-01 7.78695583e-01
-5.78242362e-01 1.04239598e-01 6.98294044e-01 -4.81171191e-01
5.12994528e-01 1.07102716e+00 6.11133218e-01 6.17263258e-01
-1.15090024e+00 -1.06546307e+00 3.08618456e-01 3.04564331e-02
-9.13734555e-01 -8.20712149e-01 -1.18605783e-02 -1.42327949e-01
1.64077890e+00 1.59298927e-01 4.98682708e-01 1.05135858e+00
3.66596192e-01 8.26366723e-01 8.02558422e-01 -3.13061513e-02
4.09116238e-01 -6.20462358e-01 8.27364624e-01 5.04900515e-01
5.56722343e-01 -1.70792013e-01 -7.72830367e-01 -3.38105053e-01
9.41088498e-01 -1.32471353e-01 1.57746449e-01 3.94863367e-01
-1.24728763e+00 5.85043430e-01 9.07098129e-02 -2.39363074e-01
-5.36281109e-01 9.91679370e-01 1.24540222e+00 -4.17604297e-02
5.42368710e-01 -3.02830517e-01 -6.33067012e-01 -5.87638617e-01
-9.83880579e-01 3.95077854e-01 9.85621154e-01 1.33474386e+00
5.36936581e-01 2.30235949e-01 -4.48600233e-01 4.81167525e-01
3.48946273e-01 5.89761078e-01 9.74119753e-02 -1.23189831e+00
4.28883523e-01 3.40946883e-01 -4.37087297e-01 -2.20592812e-01
-6.87930048e-01 -2.94394940e-01 -8.60952914e-01 5.10531545e-01
2.07878113e-01 -6.52402759e-01 -6.85540617e-01 1.75161624e+00
5.94947278e-01 6.93938911e-01 -8.11795294e-02 9.97452140e-01
9.10645962e-01 8.51602852e-01 2.73214102e-01 3.01226139e-01
1.67411339e+00 -1.52015448e+00 -4.84652638e-01 -4.24816877e-01
6.93232775e-01 -8.33067656e-01 7.81934381e-01 4.12240922e-01
-1.18170810e+00 -6.19172871e-01 -1.01288044e+00 -6.40142679e-01
1.18121646e-01 2.27309510e-01 1.25282598e+00 6.26618743e-01
-1.59599423e+00 5.32319844e-02 -1.69986999e+00 -1.62180141e-01
3.69840562e-01 4.73890692e-01 4.71374057e-02 1.04246631e-01
-4.58289385e-01 3.20590585e-01 3.10920447e-01 1.11676507e-01
-1.01806986e+00 -1.35561359e+00 -3.13694715e-01 3.92311532e-03
2.01308906e-01 -1.00030267e+00 1.61002684e+00 -4.68250781e-01
-1.51899433e+00 5.98412454e-01 -2.35693797e-01 -3.78915161e-01
4.76510525e-01 -3.96211326e-01 -3.99219871e-01 -1.88906327e-01
1.11017346e-01 8.83938313e-01 5.55410087e-01 -6.40488744e-01
-1.02395105e+00 -3.06040019e-01 1.42124787e-01 5.16726449e-02
-1.54875383e-01 2.63450086e-01 -9.57587004e-01 -2.68124074e-01
-2.77762592e-01 -1.08307922e+00 -1.68372750e-01 2.65059769e-01
-2.85530746e-01 -2.65486360e-01 1.16245723e+00 7.32096508e-02
8.35071325e-01 -2.21109247e+00 -3.41154605e-01 -1.07832916e-01
8.06150436e-01 3.92956287e-01 4.61026877e-02 -2.52021074e-01
3.63769472e-01 -4.67956543e-01 5.10381401e-01 -7.02256918e-01
8.22630003e-02 7.09870934e-01 -3.88795137e-01 1.94278985e-01
-3.19771677e-01 6.01099789e-01 -7.41081953e-01 -3.70081335e-01
1.50358185e-01 5.84600031e-01 -8.81684661e-01 4.17587049e-02
-3.03479254e-01 -4.72678691e-02 -3.21442813e-01 5.89670837e-01
9.68760490e-01 -4.09028620e-01 3.50504905e-01 -3.29540193e-01
-3.64868820e-01 4.74250346e-01 -9.95081842e-01 1.66430056e+00
-6.91903710e-01 1.13402557e+00 5.10204315e-01 -3.14282715e-01
4.67082113e-01 1.44229144e-01 3.33815105e-02 -9.96840477e-01
2.27735996e-01 2.19366491e-01 7.53921494e-02 1.18192546e-01
9.69144821e-01 5.52964568e-01 4.41738367e-02 6.72330439e-01
-1.34016797e-01 5.90123177e-01 2.39576951e-01 2.12381423e-01
1.56927502e+00 -1.88090920e-01 -4.79142636e-01 -7.16654897e-01
-7.15325996e-02 2.41103142e-01 7.58633256e-01 9.69369531e-01
-4.51469958e-01 -6.55293390e-02 6.83530033e-01 -8.54525745e-01
-1.04316783e+00 -1.02778983e+00 -3.22435409e-01 1.75626349e+00
-1.04207471e-02 -5.32821894e-01 -1.04281282e+00 -1.73902869e-01
-7.57332752e-03 6.76622033e-01 -8.55505243e-02 2.05138102e-01
-7.00540006e-01 -1.10082161e+00 9.67749298e-01 9.70876575e-01
8.35373163e-01 -8.82058203e-01 -1.40065372e+00 4.72851962e-01
1.97112232e-01 -1.68391466e+00 -5.03146052e-01 4.27409202e-01
-9.92150187e-01 -7.80871570e-01 -3.17309313e-02 -5.70067286e-01
3.68450850e-01 6.93770945e-01 1.75980997e+00 -6.19511940e-02
-4.98290569e-01 -1.27961323e-01 1.13210402e-01 -3.27842832e-01
-1.52942643e-01 4.31857288e-01 1.01222023e-01 -7.05701470e-01
5.01828492e-01 -7.80681252e-01 -7.19722927e-01 2.77203232e-01
-7.59689867e-01 6.15892828e-01 3.87553930e-01 6.58279955e-01
7.70873427e-01 -2.39704296e-01 1.72599316e-01 -1.02124298e+00
5.80182314e-01 -3.14280272e-01 -1.25916374e+00 -1.27649009e-01
-3.58899713e-01 -1.00696690e-01 8.01538110e-01 -4.35424000e-01
-1.03434992e+00 -1.55801982e-01 -2.61962056e-01 -5.37689924e-01
-1.00079767e-01 -1.90300960e-02 -1.75736502e-01 6.91666007e-02
5.64005077e-01 5.62065132e-02 -1.41253427e-01 -1.14954397e-01
2.36114934e-01 2.70754158e-01 7.33549476e-01 -1.10481513e+00
-1.68936595e-01 6.06025994e-01 -2.45813876e-02 -7.34517097e-01
-3.99444878e-01 -3.03941458e-01 -4.22487855e-02 -2.27184385e-01
8.93958151e-01 -1.51419568e+00 -1.41948962e+00 9.06871676e-01
-1.58920562e+00 -1.12536216e+00 2.78197646e-01 3.61684889e-01
-1.62206262e-01 -2.23054141e-01 -1.20481384e+00 -3.18374932e-01
-8.38351607e-01 -1.69637287e+00 1.27942073e+00 3.53018522e-01
-8.11552331e-02 -7.50572324e-01 6.38976768e-02 3.96691054e-01
7.27479398e-01 -1.46158516e-01 6.11800194e-01 -2.74992436e-01
-1.13377726e+00 3.42761695e-01 -9.98382151e-01 7.41384700e-02
-3.90132815e-01 1.66832641e-01 -1.09327328e+00 -3.67521942e-01
-3.01765174e-01 -3.89630765e-01 7.57446885e-01 5.45374334e-01
1.31323123e+00 1.23856962e-01 -6.37408257e-01 1.21348941e+00
1.59982169e+00 1.84021816e-01 4.90776271e-01 2.07282916e-01
1.16848934e+00 -3.44138891e-01 1.20889150e-01 5.64514935e-01
5.74536800e-01 6.34574652e-01 6.05177820e-01 -1.05040275e-01
-1.57022178e-01 4.84956354e-01 5.03507972e-01 1.04474723e+00
5.99708175e-03 -5.12261450e-01 -1.00308287e+00 3.24080437e-01
-2.10411000e+00 -4.36353683e-01 -5.89304924e-01 1.82869530e+00
5.66532135e-01 4.02694970e-01 4.16084826e-02 -4.24367368e-01
3.26020241e-01 3.03415805e-01 -1.06219792e+00 -6.56387866e-01
2.19439790e-01 2.89057106e-01 1.02292740e+00 2.09024966e-01
-8.55419993e-01 1.16947877e+00 6.38936090e+00 1.01538408e+00
-1.19117105e+00 6.02345407e-01 5.56566298e-01 -5.59322953e-01
6.27319142e-02 -1.33368716e-01 -1.58767760e+00 6.37626052e-01
1.30699861e+00 1.72248617e-01 2.32063353e-01 1.34546924e+00
1.91540882e-01 -3.72335017e-01 -1.09460366e+00 1.04176927e+00
-3.21135223e-01 -1.58866715e+00 -1.21428311e-01 3.21920514e-01
6.86334193e-01 1.06121159e+00 -3.76528859e-01 3.18001419e-01
1.01928008e+00 -5.70658684e-01 7.38338590e-01 1.93230733e-01
7.02025771e-01 -1.05901861e+00 6.81301594e-01 1.88800141e-01
-1.20061684e+00 6.02132976e-01 -5.48364282e-01 -4.03078288e-01
2.61417478e-01 8.67470801e-01 -3.76174778e-01 -8.94968137e-02
8.66503656e-01 2.33340219e-01 -1.51430979e-01 5.47802091e-01
8.62138197e-02 7.32211888e-01 -6.31847322e-01 -1.91664081e-02
3.31463695e-01 -4.44749855e-02 5.99468276e-02 1.37066150e+00
4.57527563e-02 5.65712377e-02 2.36969978e-01 6.81761980e-01
-2.56671041e-01 -4.37833935e-01 -8.66941437e-02 4.60917860e-01
7.25221157e-01 1.60889399e+00 -1.03663516e+00 -6.30518436e-01
-8.69811237e-01 1.26910806e+00 5.53803682e-01 1.23232096e-01
-1.41797733e+00 -1.10853642e-01 1.84093010e+00 -2.21853718e-01
7.24897608e-02 -6.02151871e-01 -5.27661622e-01 -1.09060597e+00
-8.16804469e-02 -5.97753406e-01 2.87734680e-02 -7.15838611e-01
-9.03850079e-01 9.02176321e-01 -4.28338915e-01 -4.38756198e-01
1.59238856e-02 -8.48924279e-01 -6.32404506e-01 8.24190140e-01
-1.41785860e+00 -9.24031675e-01 -7.75525689e-01 5.71222305e-01
5.87401986e-01 -1.55684844e-01 7.17407763e-01 9.59467709e-01
-1.22236598e+00 7.37269640e-01 1.04642011e-01 2.21719909e-02
6.31859899e-01 -1.10049200e+00 1.29417396e+00 8.94704282e-01
-2.73357302e-01 3.72791499e-01 4.90686506e-01 -5.60373604e-01
-1.81380832e+00 -1.35680568e+00 3.97835463e-01 -2.73761123e-01
7.14368463e-01 -7.15485930e-01 -7.61589110e-01 9.48587537e-01
2.68527269e-01 6.66122019e-01 6.51129305e-01 3.22285444e-01
-3.94728810e-01 -3.91976058e-01 -5.75562060e-01 8.92465234e-01
1.27154887e+00 -3.92099649e-01 2.52205074e-01 4.89319146e-01
8.30348969e-01 -9.90718424e-01 -5.67173779e-01 -2.45149106e-01
4.95065272e-01 -1.15521884e+00 8.69500279e-01 -3.32082003e-01
2.70656228e-01 -2.96955228e-01 -1.41567796e-01 -8.70682776e-01
-3.84574562e-01 -2.91534930e-01 -3.43209237e-01 1.01240647e+00
3.00744898e-03 -6.05796099e-01 1.19881499e+00 6.76837027e-01
-6.32541537e-01 -4.47614998e-01 -8.65017295e-01 -8.25379014e-01
-4.74600285e-01 -1.03005803e+00 4.27632391e-01 6.13027453e-01
-6.71021461e-01 5.91607690e-01 -8.87136683e-02 3.67759407e-01
8.09395373e-01 -1.43680453e-01 1.17491031e+00 -9.25322592e-01
-3.40449274e-01 -5.07492661e-01 -2.95432597e-01 -1.34223247e+00
4.54376370e-01 -7.99903810e-01 5.51725179e-02 -1.28862238e+00
2.48745695e-01 -5.26449740e-01 9.14364308e-02 7.29379773e-01
2.59997379e-02 1.78072289e-01 1.43202767e-01 1.60956383e-01
-9.87076759e-01 1.07584000e-01 8.55286956e-01 3.28806669e-01
-2.17812248e-02 -3.49802226e-01 -3.23574603e-01 8.29397380e-01
4.68278885e-01 -3.52201611e-01 -6.77405596e-01 -1.38732851e+00
2.47886240e-01 -3.01608220e-02 5.05267739e-01 -1.43312275e+00
8.03431630e-01 9.95936319e-02 6.10876828e-02 -7.89044559e-01
3.70972484e-01 -6.08078718e-01 2.74309337e-01 4.60075200e-01
3.09455991e-01 7.10562885e-01 8.30244780e-01 3.12997699e-01
6.87921494e-02 4.11044121e-01 5.97410738e-01 2.32326478e-01
-1.25938046e+00 7.41519392e-01 -6.93793356e-01 3.12038720e-01
8.43259513e-01 2.37571336e-02 -8.31834912e-01 3.82170826e-01
1.30188968e-02 3.64308178e-01 2.95474470e-01 1.17513038e-01
1.58951998e-01 -1.05370128e+00 -5.82332313e-01 1.36745721e-01
-1.09741293e-01 4.82953221e-01 6.77861571e-01 6.16825163e-01
-1.45369267e+00 4.20086592e-01 -3.88943762e-01 -7.94057429e-01
-1.34140122e+00 1.65309593e-01 2.07687303e-01 -2.48432785e-01
-9.13031518e-01 1.05525637e+00 4.16682154e-01 -2.34006867e-01
1.28800467e-01 -7.62233198e-01 5.62416553e-01 -1.19271986e-01
7.97474265e-01 5.53526938e-01 4.76951212e-01 -1.42971426e-01
-5.30149519e-01 -3.64493504e-02 -4.53326285e-01 -3.62665877e-02
1.08661258e+00 1.63788974e-01 -4.25992012e-01 -4.02647331e-02
9.56665337e-01 -5.02279401e-01 -1.49539983e+00 -1.06509216e-01
-2.63131320e-01 -1.25434607e-01 8.82409632e-01 -4.30660933e-01
-1.51355839e+00 6.27642035e-01 4.94409263e-01 -3.85222524e-01
1.10501766e+00 -3.06602865e-01 1.37876546e+00 4.28633481e-01
5.63624859e-01 -1.07658172e+00 -2.32922927e-01 9.35282528e-01
-3.33510414e-02 -1.14474058e+00 -2.37523824e-01 -4.34245884e-01
-7.35680610e-02 1.06766343e+00 1.36856616e+00 -1.59137785e-01
6.77150249e-01 1.35503030e+00 2.69981027e-01 -2.22377107e-01
-1.37498724e+00 1.04684860e-01 -4.33850884e-01 2.19797432e-01
3.74911964e-01 2.99345136e-01 -1.07845008e-01 4.06827658e-01
-1.25137433e-01 3.05062294e-01 3.66315991e-01 1.03958988e+00
-1.71450257e-01 -9.93652225e-01 -8.74327421e-02 5.29850841e-01
-3.69576782e-01 -4.57599074e-01 5.13057172e-01 7.68385291e-01
7.76879787e-02 4.30270761e-01 1.11097145e+00 -1.56157464e-01
-3.53010036e-02 -4.15092498e-01 2.89974213e-01 -5.03360629e-01
-8.11051071e-01 -4.44354229e-02 2.97283620e-01 -8.27885687e-01
1.69224218e-01 -4.02941972e-01 -1.29874384e+00 -1.29274297e+00
1.34276524e-01 -4.88432765e-01 1.03161621e+00 7.31775880e-01
9.36646223e-01 1.15635145e+00 -2.69078970e-01 -8.74044299e-01
3.62167470e-02 -4.46516812e-01 -4.99383360e-01 -2.96309173e-01
4.24932092e-02 -3.63121361e-01 1.29922882e-01 -2.72024900e-01]
|
[8.451339721679688, 3.11051607131958]
|
dfa766f6-0268-421f-923d-d4a32cc0917d
|
prequant-a-task-agnostic-quantization
|
2306.00014
| null |
https://arxiv.org/abs/2306.00014v1
|
https://arxiv.org/pdf/2306.00014v1.pdf
|
PreQuant: A Task-agnostic Quantization Approach for Pre-trained Language Models
|
While transformer-based pre-trained language models (PLMs) have dominated a number of NLP applications, these models are heavy to deploy and expensive to use. Therefore, effectively compressing large-scale PLMs becomes an increasingly important problem. Quantization, which represents high-precision tensors with low-bit fix-point format, is a viable solution. However, most existing quantization methods are task-specific, requiring customized training and quantization with a large number of trainable parameters on each individual task. Inspired by the observation that the over-parameterization nature of PLMs makes it possible to freeze most of the parameters during the fine-tuning stage, in this work, we propose a novel ``quantize before fine-tuning'' framework, PreQuant, that differs from both quantization-aware training and post-training quantization. PreQuant is compatible with various quantization strategies, with outlier-aware parameter-efficient fine-tuning incorporated to correct the induced quantization error. We demonstrate the effectiveness of PreQuant on the GLUE benchmark using BERT, RoBERTa, and T5. We also provide an empirical investigation into the workflow of PreQuant, which sheds light on its efficacy.
|
['Rui Yan', 'Dongyan Zhao', 'Yunsen Xian', 'Wei Wu', 'Jingang Wang', 'Yang Yang', 'Qifan Wang', 'Jiahao Liu', 'Zhuocheng Gong']
|
2023-05-30
| null | null | null | null |
['quantization']
|
['methodology']
|
[-5.20771518e-02 -3.50707054e-01 -2.82270461e-01 -4.03531969e-01
-1.15018749e+00 -6.39411747e-01 6.04295611e-01 3.15887332e-01
-5.85187078e-01 4.38372135e-01 1.05360016e-01 -6.27827108e-01
-2.34196410e-01 -5.85676491e-01 -9.32025373e-01 -5.94660103e-01
-5.25145829e-02 7.46630907e-01 2.97713429e-01 -1.79495871e-01
3.65702629e-01 2.84860760e-01 -1.38000035e+00 4.54976141e-01
1.13907003e+00 1.12985456e+00 3.26579958e-01 5.79258621e-01
-2.83291727e-01 5.00905097e-01 -6.36998713e-01 -7.99581826e-01
3.10596228e-01 1.87463582e-01 -8.28105688e-01 4.46273908e-02
6.08828843e-01 -2.17062667e-01 -1.85037181e-02 1.07701480e+00
3.66494775e-01 5.54609597e-02 3.83732021e-01 -1.17485154e+00
-6.15912914e-01 9.64520216e-01 -1.08828984e-01 2.73933530e-01
-1.97095945e-01 3.08800489e-01 1.10681880e+00 -1.11319864e+00
3.42525631e-01 1.23334146e+00 7.05128968e-01 1.79623201e-01
-1.29496932e+00 -5.92471361e-01 -9.10304710e-02 4.20794517e-01
-1.57552278e+00 -5.76422155e-01 4.46446776e-01 -4.97979969e-01
1.25627351e+00 8.21110606e-02 4.42209899e-01 1.01732028e+00
2.92598635e-01 6.75522387e-01 9.43240464e-01 -3.92183632e-01
4.73688394e-01 1.20966144e-01 1.29293902e-02 6.64401472e-01
1.39612496e-01 -3.51117626e-02 -7.18848169e-01 -2.86095947e-01
6.03092372e-01 -1.11805938e-01 -1.65903881e-01 -4.70570356e-01
-1.34441364e+00 8.01171958e-01 2.91214049e-01 3.30703288e-01
-4.04406905e-01 3.31608415e-01 7.22847879e-01 3.47534955e-01
6.13479018e-01 6.41600192e-01 -6.81849957e-01 -5.96633375e-01
-1.40261686e+00 3.75272542e-01 5.78266919e-01 1.01339865e+00
8.94911051e-01 6.90204203e-02 -4.41243619e-01 8.79372299e-01
1.52096435e-01 3.17497641e-01 6.99155509e-01 -9.16042686e-01
7.60634780e-01 5.86060345e-01 -9.52781141e-02 -8.38658333e-01
-2.46034250e-01 -6.14092529e-01 -1.07742083e+00 -3.55276406e-01
3.18151146e-01 1.40357614e-01 -9.72691596e-01 1.37908065e+00
2.63780922e-01 2.18462810e-01 -1.84611216e-01 8.64671111e-01
1.81457102e-01 6.74840212e-01 9.75500618e-04 -7.65283182e-02
1.37121820e+00 -1.04964089e+00 -7.14062989e-01 -5.90460338e-02
9.54690397e-01 -8.50283563e-01 1.77305305e+00 7.59919107e-01
-1.03045511e+00 -3.47811073e-01 -1.05961108e+00 -4.16438520e-01
-3.47431332e-01 1.67068213e-01 7.20245898e-01 7.25548327e-01
-1.10799873e+00 8.30743730e-01 -1.10624743e+00 -4.41931300e-02
3.49154681e-01 3.72506708e-01 -2.00956091e-01 -1.18508592e-01
-1.27675259e+00 8.65857303e-01 7.00927198e-01 2.09544525e-01
-9.07324076e-01 -1.06112897e+00 -4.87634867e-01 2.69050241e-01
4.92175728e-01 -5.82131505e-01 1.32758570e+00 -3.09282154e-01
-1.79386747e+00 4.69798893e-01 -8.97162780e-02 -7.11780727e-01
4.46629226e-01 -3.32894653e-01 -1.49332806e-01 8.32876249e-04
-2.75824398e-01 7.27594852e-01 1.18112028e+00 -6.59178972e-01
-4.20165747e-01 -8.79772082e-02 6.44180775e-02 -8.24505538e-02
-7.55346358e-01 -1.70119165e-03 -6.48884594e-01 -8.47544551e-01
6.78598583e-02 -7.56139219e-01 -2.28167370e-01 -3.63003135e-01
-3.18222612e-01 -4.25016850e-01 4.84916091e-01 -5.64459145e-01
1.57424879e+00 -2.11861849e+00 2.41006806e-01 5.53216003e-02
1.99588761e-01 4.30734009e-01 -2.59535402e-01 4.66787457e-01
2.24859387e-01 4.98289347e-01 -2.05988452e-01 -8.45736742e-01
5.49458981e-01 5.34547389e-01 -5.69743037e-01 2.85215437e-01
2.81151503e-01 7.29170322e-01 -8.97318840e-01 -5.74259639e-01
2.60924935e-01 5.69342077e-01 -9.87051427e-01 9.93323997e-02
-6.28102303e-01 1.93150908e-01 -2.76729554e-01 7.36841500e-01
6.77943051e-01 -4.11673963e-01 1.15280151e-01 -4.03249890e-01
-1.00322701e-01 8.15330267e-01 -1.12437260e+00 1.81227505e+00
-5.41845739e-01 3.09418380e-01 -8.92722458e-02 -9.36412215e-01
6.18072748e-01 2.91000426e-01 2.69050717e-01 -6.73742235e-01
7.55100176e-02 4.19634014e-01 -2.41860047e-01 -2.73674458e-01
9.58014727e-01 7.53159747e-02 -5.57029396e-02 3.04961860e-01
2.70547062e-01 -4.48086649e-01 6.65093184e-01 2.15158507e-01
1.15959334e+00 1.66341066e-02 9.62258056e-02 -2.33187661e-01
3.07082176e-01 -8.70570838e-02 5.88791251e-01 4.69379097e-01
-1.09855652e-01 5.79171300e-01 4.56874251e-01 -3.61904621e-01
-1.26145470e+00 -9.15803671e-01 -2.91774839e-01 1.24814379e+00
-4.69361752e-01 -1.11706901e+00 -8.63622725e-01 -5.48366368e-01
-8.71578902e-02 7.50016630e-01 -3.81393462e-01 -9.41938683e-02
-6.03697181e-01 -8.50260794e-01 6.64880455e-01 2.39836767e-01
4.24295992e-01 -6.52439475e-01 -3.72765243e-01 3.99150670e-01
-4.19988871e-01 -1.24538064e+00 -6.23625100e-01 3.79957825e-01
-1.08215368e+00 -7.31022537e-01 -3.05161089e-01 -2.10975468e-01
3.32139820e-01 1.12395566e-02 1.43852007e+00 2.42180414e-02
2.24742338e-01 -1.56194672e-01 -6.35124564e-01 -1.83278650e-01
-4.46992487e-01 6.29532695e-01 1.84326127e-01 -5.29643632e-02
2.67986178e-01 -7.18696892e-01 -3.00151765e-01 2.34506249e-01
-1.21744442e+00 5.78310341e-02 7.11275458e-01 8.76412392e-01
8.44269335e-01 1.77500024e-01 1.02840088e-01 -7.37032712e-01
7.29775608e-01 -1.63001284e-01 -9.18003738e-01 4.00493473e-01
-8.49280119e-01 4.53712314e-01 9.28798318e-01 -5.20243227e-01
-4.37668502e-01 -2.63473958e-01 -1.32735103e-01 -9.64015663e-01
3.42149168e-01 7.01221764e-01 -1.38029531e-01 -5.32304570e-02
6.27549231e-01 1.65179998e-01 -5.46475530e-01 -8.06364059e-01
5.04788876e-01 5.19516826e-01 4.42355186e-01 -8.85892332e-01
8.29815030e-01 3.74947004e-02 -2.49068514e-01 -6.28380418e-01
-8.84024858e-01 -3.35288107e-01 -7.08108187e-01 2.56071240e-01
4.40707564e-01 -7.81891167e-01 -5.33895671e-01 2.60104150e-01
-1.17821288e+00 -6.18442416e-01 -3.21122110e-01 4.64871973e-01
-4.10664082e-01 3.69725108e-01 -6.07024908e-01 -3.64899844e-01
-5.34463942e-01 -1.52928960e+00 1.23801124e+00 -4.83169943e-01
-8.68305191e-02 -9.88212287e-01 -4.77290712e-02 5.04052162e-01
5.99875689e-01 -3.10555667e-01 1.01279020e+00 -4.55124676e-01
-7.86379278e-01 9.82290208e-02 -1.77780062e-01 6.27480924e-01
-2.00222299e-01 1.02472536e-01 -8.18914354e-01 -5.57181954e-01
-3.42831984e-02 -4.14486736e-01 5.56729615e-01 3.03919576e-02
1.46648324e+00 -5.07984877e-01 1.61232688e-02 9.98203874e-01
1.30897391e+00 -4.50803995e-01 5.91638267e-01 4.68045980e-01
7.87083089e-01 5.13481721e-02 6.31447732e-01 6.11698389e-01
4.64539409e-01 7.84089625e-01 3.88629764e-01 4.11272675e-01
2.49406472e-02 -3.04435790e-01 5.50377488e-01 1.44699168e+00
3.75769734e-02 -1.34361669e-01 -1.03948629e+00 4.55263466e-01
-1.54606724e+00 -4.72288311e-01 5.38257509e-02 2.26410103e+00
1.26139152e+00 2.48529583e-01 -1.66163042e-01 4.02473360e-01
3.37825686e-01 2.99424697e-02 -2.71451861e-01 -4.86410648e-01
9.98389050e-02 5.17650247e-01 7.85596013e-01 3.39606792e-01
-9.96273637e-01 1.18744588e+00 6.26184702e+00 1.25887620e+00
-1.37933755e+00 4.69878972e-01 4.17016298e-01 -2.39517495e-01
-3.80434304e-01 7.78243095e-02 -1.10568035e+00 6.41858339e-01
1.51399279e+00 -1.23911195e-01 7.78891027e-01 9.03168440e-01
1.99867949e-01 1.99300781e-01 -1.23095405e+00 9.98260140e-01
-2.56619006e-01 -1.46251595e+00 3.75138164e-01 -4.62162122e-02
6.11058652e-01 3.13139290e-01 2.26453632e-01 6.39867842e-01
3.16789836e-01 -8.69802415e-01 9.61237371e-01 1.30352736e-01
9.70782220e-01 -7.13641644e-01 6.04387760e-01 4.32256609e-01
-1.16678166e+00 1.03734136e-01 -7.89257288e-01 1.25117496e-01
6.69752285e-02 1.06642163e+00 -1.12121606e+00 5.43407619e-01
8.70979369e-01 3.88105482e-01 -8.44446242e-01 9.44566488e-01
-2.74056911e-01 9.58537579e-01 -4.85517949e-01 2.40446746e-01
4.23384875e-01 -1.75081477e-01 2.96542227e-01 1.19387996e+00
4.12137240e-01 -3.42454016e-01 2.09321186e-01 7.89600492e-01
-1.91919923e-01 2.22388804e-01 5.48050590e-02 -3.67047668e-01
7.52697468e-01 1.22002435e+00 -4.52132851e-01 -5.03964067e-01
-1.31471053e-01 7.91451991e-01 6.54265463e-01 8.35834742e-02
-9.92816985e-01 -1.05815373e-01 5.73399305e-01 9.34941545e-02
5.72735786e-01 -4.94841367e-01 -3.69135678e-01 -1.36330843e+00
1.52207062e-01 -1.11125457e+00 2.12353930e-01 -5.17468512e-01
-1.06704414e+00 5.29244900e-01 6.18307032e-02 -1.19974113e+00
-3.32080156e-01 -4.89902556e-01 6.22795038e-02 7.84604609e-01
-1.69470215e+00 -1.05607760e+00 4.66034077e-02 4.61337715e-01
4.66603190e-01 1.31027833e-01 7.90886760e-01 6.47118688e-01
-6.45813107e-01 1.02211738e+00 2.95305640e-01 -2.12398589e-01
7.62246370e-01 -1.29184926e+00 5.47851384e-01 8.82832229e-01
1.13286823e-01 8.67704988e-01 7.17623532e-01 -3.62743706e-01
-1.79478967e+00 -1.36573005e+00 1.27928185e+00 -3.75553608e-01
9.63297367e-01 -6.38565540e-01 -1.06084597e+00 7.12290645e-01
-7.93909132e-02 1.14879176e-01 4.55760419e-01 2.45580971e-01
-5.30714333e-01 -2.99107879e-01 -9.00257230e-01 5.76926708e-01
8.08655798e-01 -7.36175776e-01 -4.70757216e-01 6.26986086e-01
1.02847087e+00 -6.64103687e-01 -1.33624268e+00 2.72182584e-01
1.35554880e-01 -8.32498252e-01 9.49331582e-01 -2.33349130e-01
2.15943888e-01 -1.84702948e-01 -2.38198176e-01 -1.41109121e+00
-3.30674171e-01 -8.64446044e-01 -3.71131837e-01 1.25316823e+00
2.66018778e-01 -4.40136760e-01 7.39717960e-01 4.91025180e-01
-4.64649349e-01 -6.66694283e-01 -1.09221101e+00 -8.88302207e-01
3.16401929e-01 -8.46790016e-01 8.70272458e-01 9.54527855e-01
-1.34493738e-01 2.31032223e-01 -3.63656610e-01 1.39680907e-01
4.39988464e-01 -2.31549218e-01 8.58180821e-01 -9.77038383e-01
-4.19792354e-01 -4.53329116e-01 -1.16273835e-01 -1.06829822e+00
5.51593192e-02 -1.02588916e+00 2.05583014e-02 -1.03930628e+00
-1.53857425e-01 -7.21139848e-01 -3.19801211e-01 7.05085278e-01
-1.05192542e-01 1.32346943e-01 2.02012181e-01 4.09136921e-01
-6.78911388e-01 7.23128617e-01 1.22763896e+00 -2.85876811e-01
-8.54153186e-02 -2.53422379e-01 -3.27411175e-01 4.27193046e-01
6.74075007e-01 -5.73757052e-01 -4.90237504e-01 -8.17825437e-01
5.38481176e-01 -3.07584167e-01 1.38360083e-01 -1.02775669e+00
3.00702572e-01 -8.76258835e-02 -1.57018527e-01 -7.82443523e-01
3.05896372e-01 -7.17426717e-01 -1.66709628e-02 2.14537323e-01
-2.30357856e-01 3.67322266e-01 2.16627419e-01 1.08455054e-01
-2.84834206e-01 -1.98569253e-01 7.89343834e-01 2.91430876e-02
-6.68077886e-01 5.12050927e-01 -1.24460891e-01 2.23112255e-01
4.63796914e-01 1.36222720e-01 -2.77494937e-01 1.80474743e-01
-3.33533019e-01 8.31751972e-02 5.18618524e-01 2.52892703e-01
4.39966589e-01 -1.30881262e+00 -6.78594291e-01 2.69605309e-01
6.97610527e-02 3.02835017e-01 1.63421795e-01 9.61824358e-01
-5.01289546e-01 7.80153811e-01 1.72419637e-01 -7.06205726e-01
-8.43706191e-01 7.79021919e-01 1.79014325e-01 -8.18221688e-01
-4.97053593e-01 9.54720080e-01 -1.93784893e-01 -6.17166877e-01
3.57521057e-01 -1.07485664e+00 1.59517035e-01 -1.62750438e-01
5.21069169e-01 3.72053653e-01 7.35127211e-01 -3.48494589e-01
-1.63950816e-01 2.08857566e-01 -2.21073627e-01 -6.70004338e-02
1.18389022e+00 3.33166942e-02 -3.82722229e-01 4.48077172e-01
1.16183400e+00 -1.71334922e-01 -1.13370907e+00 -4.86547202e-01
2.58645862e-01 -4.32050228e-01 4.20638084e-01 -6.00890100e-01
-9.70043480e-01 1.16149640e+00 2.74123192e-01 1.29159480e-01
1.09864640e+00 -5.21045804e-01 1.01104784e+00 6.27183795e-01
5.99968672e-01 -1.27205420e+00 3.22346129e-02 8.28358233e-01
8.70715976e-01 -9.24801230e-01 3.61985676e-02 -3.88284504e-01
-1.92755267e-01 9.28395748e-01 3.07709008e-01 1.60380587e-01
5.30651450e-01 3.20633888e-01 -1.17051937e-01 5.91409691e-02
-1.18092978e+00 2.34419942e-01 3.64950627e-01 2.17258185e-01
3.81897718e-01 6.43651038e-02 -1.26115292e-01 4.91638839e-01
-6.34233117e-01 2.69969046e-01 2.47631401e-01 8.00368130e-01
-3.15775692e-01 -1.34931588e+00 -5.71617663e-01 3.79442692e-01
-4.12326396e-01 -5.71891069e-01 6.75365776e-02 5.29078662e-01
2.17871651e-01 7.66102493e-01 8.92383009e-02 -6.07417464e-01
2.65301138e-01 2.02594727e-01 4.88263488e-01 -6.21633530e-01
-8.28780890e-01 -1.74754905e-03 -9.88902897e-02 -7.19373941e-01
-1.37226090e-01 -4.95176107e-01 -9.88642454e-01 -7.31810868e-01
-2.53368616e-01 3.49934489e-01 8.51954401e-01 1.08177149e+00
6.38473272e-01 5.32899857e-01 4.18902516e-01 -8.85883927e-01
-1.20894420e+00 -1.28413355e+00 -4.41869706e-01 3.21895570e-01
6.92409500e-02 -7.56669223e-01 -3.31871212e-01 4.59244028e-02]
|
[8.70685863494873, 3.5061452388763428]
|
c14989c8-d98d-45cf-b644-520beac80307
|
ariann-low-interaction-privacy-preserving
|
2006.04593
| null |
https://arxiv.org/abs/2006.04593v4
|
https://arxiv.org/pdf/2006.04593v4.pdf
|
ARIANN: Low-Interaction Privacy-Preserving Deep Learning via Function Secret Sharing
|
We propose AriaNN, a low-interaction privacy-preserving framework for private neural network training and inference on sensitive data. Our semi-honest 2-party computation protocol (with a trusted dealer) leverages function secret sharing, a recent lightweight cryptographic protocol that allows us to achieve an efficient online phase. We design optimized primitives for the building blocks of neural networks such as ReLU, MaxPool and BatchNorm. For instance, we perform private comparison for ReLU operations with a single message of the size of the input during the online phase, and with preprocessing keys close to 4X smaller than previous work. Last, we propose an extension to support n-party private federated learning. We implement our framework as an extensible system on top of PyTorch that leverages CPU and GPU hardware acceleration for cryptographic and machine learning operations. We evaluate our end-to-end system for private inference between distant servers on standard neural networks such as AlexNet, VGG16 or ResNet18, and for private training on smaller networks like LeNet. We show that computation rather than communication is the main bottleneck and that using GPUs together with reduced key size is a promising solution to overcome this barrier.
|
['David Pointcheval', 'Pierre Tholoniat', 'Théo Ryffel', 'Francis Bach']
|
2020-06-08
| null | null | null | null |
['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning']
|
['methodology', 'natural-language-processing']
|
[-1.69552937e-01 2.66515613e-01 -4.49797697e-02 -1.04040658e+00
-7.55030751e-01 -1.08620775e+00 3.39309037e-01 3.21438685e-02
-1.17729580e+00 4.45513397e-01 -1.54943600e-01 -6.96294606e-01
2.40692407e-01 -9.57423091e-01 -1.22201538e+00 -9.86579657e-01
-4.22267348e-01 2.28662968e-01 2.40462095e-01 4.26277891e-03
-3.17631006e-01 5.04048049e-01 -1.50957608e+00 4.62411702e-01
-1.95839405e-01 1.21956706e+00 -5.95449746e-01 9.78063762e-01
2.91308820e-01 6.54822826e-01 -3.27498823e-01 -1.02251852e+00
9.22910273e-01 4.71116364e-01 -1.03868151e+00 -8.99293482e-01
8.73397291e-01 -1.05851614e+00 -5.47657609e-01 9.83789623e-01
6.60577357e-01 -7.52369016e-02 -1.99509755e-01 -1.61831129e+00
-1.79204375e-01 1.53117442e+00 1.66534260e-01 -2.59097338e-01
-5.09369612e-01 4.28466916e-01 9.48136330e-01 -3.57347392e-02
7.23301888e-01 1.04665935e+00 1.01302195e+00 9.37127411e-01
-1.15911317e+00 -9.97484744e-01 -1.79189146e-01 5.19936346e-02
-1.19256294e+00 -7.41738021e-01 9.60555747e-02 1.87686786e-01
1.20854771e+00 7.00791836e-01 2.84385622e-01 1.18055606e+00
7.52608851e-02 7.50517070e-01 1.22619200e+00 -1.93982106e-02
3.33410621e-01 2.15308636e-01 6.92794859e-01 5.43718576e-01
3.50520849e-01 6.24985322e-02 -4.55253005e-01 -9.00501370e-01
-8.52740835e-03 1.65483430e-01 -3.69561404e-01 -2.43972950e-02
-9.35959697e-01 8.06743205e-01 4.70801800e-01 -1.33389533e-01
1.38754129e-01 9.93209004e-01 1.09261572e+00 8.52613032e-01
3.36034268e-01 1.65828932e-02 -1.09318161e+00 7.37296715e-02
-6.73776746e-01 5.71802437e-01 1.67020810e+00 6.90711021e-01
8.84762228e-01 -6.04986489e-01 -1.63380634e-02 7.34972535e-03
2.76978076e-01 4.40657884e-01 2.88164496e-01 -1.30850077e+00
2.88100868e-01 -2.52966546e-02 -3.25782627e-01 -3.63864243e-01
-1.61094949e-01 1.28038824e-01 -7.41130412e-01 4.48136568e-01
5.73960781e-01 -5.74075878e-01 -5.15442312e-01 1.93600726e+00
6.23621464e-01 2.88796216e-01 6.58811629e-01 9.06906426e-01
7.60239661e-01 3.73272359e-01 -1.12938732e-02 3.75134200e-01
1.76791012e+00 -1.09986508e+00 -2.51445085e-01 2.94087399e-02
1.25913012e+00 -2.24993035e-01 6.42550409e-01 1.90511972e-01
-1.03192317e+00 2.32833609e-01 -1.07314897e+00 -8.76077533e-01
-6.10134363e-01 -2.40021795e-01 1.10669470e+00 1.04794014e+00
-1.63368201e+00 1.14593327e+00 -1.27549660e+00 1.84434965e-01
7.87361503e-01 8.48417580e-01 -9.71641421e-01 2.83772886e-01
-1.13739109e+00 2.46056527e-01 6.38503075e-01 -4.73638810e-02
-7.64597952e-01 -9.20278668e-01 -6.52441442e-01 2.49241263e-01
-2.70035714e-02 -6.80845141e-01 1.41872537e+00 -5.27036905e-01
-1.59064531e+00 1.27263689e+00 4.38959599e-01 -1.23436189e+00
5.91034472e-01 7.60973841e-02 2.77272642e-01 3.80523577e-02
-7.50351429e-01 4.68642771e-01 5.83767295e-01 -7.55955219e-01
-5.73588252e-01 -8.63995194e-01 4.20754015e-01 -2.28020176e-01
-5.74155331e-01 1.08976260e-01 -1.17163351e-02 1.24336723e-02
-6.43632263e-02 -1.22236609e+00 -2.52334744e-01 5.55661559e-01
-3.41094553e-01 -2.42031157e-01 1.13052583e+00 -4.33645368e-01
1.86599612e-01 -2.40289283e+00 -5.80793321e-01 5.11218190e-01
6.22787058e-01 3.75198156e-01 1.21508703e-01 1.34680748e-01
3.86620790e-01 -5.55346124e-02 -1.73527986e-01 -1.13792634e+00
6.03728473e-01 5.78113079e-01 -7.38614976e-01 9.72875416e-01
-5.83632469e-01 8.70348275e-01 -5.85750937e-01 -2.01159790e-01
-4.90976036e-01 6.93796337e-01 -6.66269422e-01 6.74750581e-02
-4.02590096e-01 -1.99597925e-01 -3.20994347e-01 2.23470330e-01
1.26104403e+00 -1.94547132e-01 3.70019168e-01 -3.99775319e-02
5.34130260e-02 6.30753875e-01 -1.03124714e+00 1.74032259e+00
-3.65822613e-01 5.00888109e-01 8.99573267e-01 -1.28419563e-01
3.74515891e-01 5.58837414e-01 -4.95958067e-02 1.33799925e-01
5.21738291e-01 2.54502624e-01 -5.03617942e-01 -7.22545609e-02
4.44467247e-01 4.32164639e-01 -4.91157882e-02 1.17776775e+00
1.41452119e-01 5.65793030e-02 -6.16238058e-01 1.46339074e-01
1.40297902e+00 -1.69228643e-01 -4.21686828e-01 -1.95544481e-01
3.04329395e-01 -3.34447861e-01 4.08085197e-01 9.06264305e-01
-2.42361635e-01 2.81816218e-02 6.69339299e-01 -8.37550282e-01
-7.77070701e-01 -5.41521013e-01 -8.56674165e-02 1.64810693e+00
-2.32603282e-01 -6.26190841e-01 -1.28114045e+00 -8.55512083e-01
2.75380999e-01 4.22545612e-01 -4.06398833e-01 -6.43073246e-02
-4.80291128e-01 -6.15204096e-01 1.48538005e+00 4.39513683e-01
8.13807011e-01 -7.53204882e-01 -9.38343465e-01 -2.92360574e-01
4.16310787e-01 -1.14859116e+00 -3.31888020e-01 6.53832078e-01
-7.77065098e-01 -7.82928824e-01 -5.07633202e-02 -5.55724800e-01
6.42367840e-01 1.59958363e-01 7.09860325e-01 2.35582680e-01
2.89680772e-02 8.09630230e-02 2.62136579e-01 -3.64905953e-01
-4.08441454e-01 3.37451220e-01 -2.12705079e-02 -8.91280323e-02
3.47615272e-01 -8.32313478e-01 -7.84269750e-01 -8.97976384e-02
-1.01086807e+00 -1.69152185e-01 1.36780187e-01 7.09169388e-01
4.92878228e-01 -1.91244408e-01 -5.98107755e-01 -1.33072710e+00
3.59686553e-01 -2.25053266e-01 -1.14603198e+00 1.26318991e-01
-7.75347829e-01 3.39057058e-01 1.01638615e+00 -3.50480795e-01
-6.62248731e-01 3.28242064e-01 -4.42405432e-01 -5.15987933e-01
-1.44854739e-01 -1.34266540e-01 -6.78378195e-02 -7.53251553e-01
7.15425193e-01 5.22273593e-03 4.09332901e-01 -4.21429127e-01
5.36732376e-01 7.33809233e-01 7.42533028e-01 -6.83656693e-01
7.09065497e-01 9.55153763e-01 2.87644178e-01 3.62256207e-02
-4.14241672e-01 -3.79347391e-02 -1.00224510e-01 6.03938222e-01
6.64544523e-01 -1.11396897e+00 -2.06429935e+00 1.11696398e+00
-1.16749096e+00 -7.80761421e-01 -4.70916629e-01 4.18910414e-01
-3.15600961e-01 2.37322390e-01 -1.30390894e+00 -6.30551457e-01
-1.35562849e+00 -1.20846605e+00 9.89851236e-01 9.67658907e-02
3.79662275e-01 -8.25777650e-01 6.00935705e-03 4.75992888e-01
7.29338586e-01 2.66476661e-01 3.32164854e-01 -1.29603684e+00
-9.14085567e-01 -3.52289528e-01 -1.79304108e-01 5.77575326e-01
-8.39663148e-01 -2.24596679e-01 -1.90643716e+00 -6.24071598e-01
3.10368270e-01 -8.26472402e-01 1.08032858e+00 -3.49333227e-01
1.57622433e+00 -1.06592250e+00 -9.08520520e-02 1.51843524e+00
1.40034962e+00 -7.77109981e-01 4.57585603e-01 1.71933457e-01
8.11763883e-01 4.23036814e-01 -1.50698945e-01 5.45141697e-01
7.36455202e-01 2.59070128e-01 5.26923895e-01 6.64028823e-02
5.21661460e-01 3.35167311e-02 4.64272767e-01 3.81273687e-01
4.10540625e-02 2.32262909e-01 -5.30803025e-01 1.45371988e-01
-1.86982656e+00 -6.96285844e-01 -1.19436651e-01 2.17319107e+00
1.26744449e+00 -5.36286794e-02 -2.01416701e-01 -8.24109465e-02
2.63372418e-02 2.45638311e-01 -4.60725963e-01 -7.38871157e-01
-1.05254300e-01 6.10602498e-01 1.78461254e+00 4.70042169e-01
-1.11998463e+00 9.33474958e-01 5.29921913e+00 7.03504622e-01
-1.21784306e+00 7.02903867e-01 7.66956866e-01 -4.13916558e-01
-3.49573910e-01 2.36848116e-01 -1.03590286e+00 3.46519113e-01
1.51129699e+00 3.29144374e-02 8.85619462e-01 1.37770641e+00
-7.02788472e-01 3.92366022e-01 -1.30329812e+00 8.94236863e-01
-4.91119891e-01 -1.40347302e+00 -6.97548807e-01 2.53013045e-01
2.41438031e-01 8.99518549e-01 1.33788601e-01 8.61815959e-02
8.43173206e-01 -8.81639123e-01 7.02520013e-01 2.42692493e-02
7.09236860e-01 -9.16847527e-01 9.08784628e-01 2.78387904e-01
-5.86256385e-01 7.14193285e-02 -6.83078349e-01 1.81691393e-01
-3.52916926e-01 2.67510682e-01 -8.55001986e-01 -5.55910654e-02
1.05439317e+00 -6.12182394e-02 -3.39538902e-01 2.42482126e-01
-2.90685445e-01 6.79019213e-01 -1.02048147e+00 -4.12047766e-02
3.13600421e-01 8.79276544e-02 1.87801093e-01 1.06974614e+00
-1.05341665e-01 1.26607448e-01 -2.04490706e-01 7.03849018e-01
-7.97619164e-01 -1.19067408e-01 -5.38969040e-01 2.67181396e-01
4.45498109e-01 1.66215312e+00 -2.62499928e-01 -3.93200547e-01
-4.41972762e-02 1.03059137e+00 5.13594508e-01 -1.27654508e-01
-6.16893232e-01 -4.39807773e-01 1.27998161e+00 -3.58048677e-01
3.41123790e-01 -5.78709841e-02 -2.00387642e-01 -1.19306803e+00
2.93194562e-01 -8.28177869e-01 5.75046003e-01 4.29109409e-02
-1.18519700e+00 5.05754113e-01 -1.84605509e-01 -1.63937896e-01
-1.41474485e-01 -7.33656645e-01 -4.99155998e-01 8.32705677e-01
-1.66940701e+00 -1.46898460e+00 5.64628653e-02 1.07203555e+00
-7.30065107e-01 -2.28676759e-02 1.34422171e+00 2.23289102e-01
-2.42625058e-01 1.54778862e+00 4.06846642e-01 3.66213381e-01
6.31409347e-01 -1.18463731e+00 1.15211439e+00 6.62415564e-01
-5.14491908e-02 1.01979005e+00 5.62145710e-01 -9.35501009e-02
-2.25312161e+00 -1.15326726e+00 9.41234350e-01 -2.90465623e-01
6.12710476e-01 -1.12736666e+00 -8.53456855e-01 1.13451219e+00
8.18305314e-02 7.90096462e-01 8.38013411e-01 3.70650999e-02
-1.09035814e+00 -6.11365676e-01 -1.80670393e+00 4.94473308e-01
8.36638033e-01 -9.59945858e-01 2.58957744e-01 8.95768642e-01
1.23530364e+00 -1.07033515e+00 -1.09891415e+00 -2.15464309e-01
8.17698896e-01 -8.27222645e-01 7.44876206e-01 -5.00275016e-01
-2.50383377e-01 -9.11231637e-02 -2.16317758e-01 -6.53991878e-01
2.66288817e-01 -1.39491177e+00 -3.38637024e-01 9.14740086e-01
3.38351578e-01 -1.47237873e+00 1.23127770e+00 1.40384078e+00
3.54104429e-01 -4.11876649e-01 -1.28167677e+00 -3.18212122e-01
-7.12082395e-03 -4.78188396e-01 1.22269797e+00 8.54768932e-01
-3.08637291e-01 -3.23482722e-01 -1.58999145e-01 4.70425069e-01
1.02530932e+00 7.60060968e-03 1.10115504e+00 -9.21954155e-01
-8.00445795e-01 1.11185566e-01 -5.83144844e-01 -3.77957851e-01
7.76190400e-01 -1.14316261e+00 -1.17707953e-01 -4.82922047e-01
-3.62661704e-02 -5.08938789e-01 -2.32468069e-01 1.46089303e+00
5.35526216e-01 3.78604412e-01 2.01736927e-01 7.00072125e-02
-3.39574546e-01 9.51611996e-02 4.75191206e-01 -7.86666423e-02
1.46118999e-01 1.19465865e-01 -6.02045655e-01 4.87209290e-01
7.17391193e-01 -7.70590961e-01 -1.26862869e-01 -6.87909842e-01
3.61107141e-01 -3.84625226e-01 8.32140148e-01 -8.15681875e-01
8.10414732e-01 4.49981123e-01 -1.08875692e-01 -7.05491230e-02
3.65252912e-01 -1.10364079e+00 4.00474310e-01 4.89639610e-01
-5.24730146e-01 -7.17898160e-02 4.47883792e-02 2.93162525e-01
4.37719166e-01 -1.53313681e-01 6.22230947e-01 -5.02872691e-02
-2.62959242e-01 7.23173797e-01 2.64943033e-01 -2.19712749e-01
9.60936666e-01 4.37018335e-01 -1.01359236e+00 -3.32670771e-02
-3.94493520e-01 3.07628125e-01 8.25351298e-01 -1.52180851e-01
2.75686771e-01 -7.87626684e-01 -5.62980175e-01 2.77328819e-01
-5.79726547e-02 1.13953799e-01 1.65842071e-01 6.46314204e-01
-9.50142205e-01 1.37863114e-01 -8.58513415e-02 -2.05155268e-01
-1.87710142e+00 2.63399303e-01 3.71250868e-01 -2.23616615e-01
-9.29041266e-01 1.27727580e+00 -3.68685961e-01 -8.57192457e-01
6.12702250e-01 -6.10254467e-01 5.78429699e-01 -3.83892298e-01
9.11493480e-01 1.56540632e-01 4.68990237e-01 -2.27642640e-01
-3.25691760e-01 -2.47195587e-01 -4.30932611e-01 -1.90023154e-01
1.60548687e+00 2.59634376e-01 -8.82466435e-01 -9.80568901e-02
1.79981697e+00 -2.00952247e-01 -1.27670956e+00 -3.77586156e-01
-4.77019399e-01 -2.42797613e-01 4.48177099e-01 -3.89907181e-01
-1.60503840e+00 5.02363026e-01 8.33985090e-01 -2.95891792e-01
9.58139598e-01 -1.83766618e-01 1.53456128e+00 1.34392762e+00
7.98105359e-01 -9.33832765e-01 -1.25552070e+00 4.60864991e-01
-3.51413190e-02 -1.00218725e+00 4.36969697e-02 -2.02773642e-02
-1.75860569e-01 1.08578181e+00 2.53399938e-01 -1.69929907e-01
8.83591413e-01 9.45335507e-01 2.96087950e-01 -5.63529544e-02
-1.22659492e+00 4.89153892e-01 -4.40391183e-01 1.00686684e-01
-2.75412560e-01 2.50769794e-01 6.11696392e-02 7.16807544e-01
-4.18649167e-01 -3.01438197e-02 1.93027779e-01 1.30907679e+00
1.76475585e-01 -1.30600786e+00 -2.38983393e-01 1.53321743e-01
-1.20618236e+00 -1.60391808e-01 -1.05872698e-01 3.03079039e-01
1.26220673e-01 3.41281980e-01 -6.74742088e-02 -4.84142751e-01
-2.80361593e-01 7.98455104e-02 2.69522309e-01 -1.32602766e-01
-1.58774614e+00 -7.08721161e-01 1.03141449e-01 -1.18540728e+00
5.01334853e-02 -3.84325266e-01 -1.46781266e+00 -1.02574146e+00
-3.01925570e-01 -7.99707770e-02 1.41531265e+00 5.82742095e-01
7.05438852e-01 -4.62484360e-01 7.50095487e-01 -6.79706991e-01
-1.31835985e+00 -2.86127687e-01 -6.51778877e-01 1.47449628e-01
4.81063217e-01 4.62222964e-01 -6.79144144e-01 -1.90323785e-01]
|
[5.87473726272583, 6.831081390380859]
|
f121ea04-5973-4365-afe1-a9288431270e
|
equivalent-transformation-and-dual-stream
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Chao_Equivalent_Transformation_and_Dual_Stream_Network_Construction_for_Mobile_Image_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Chao_Equivalent_Transformation_and_Dual_Stream_Network_Construction_for_Mobile_Image_CVPR_2023_paper.pdf
|
Equivalent Transformation and Dual Stream Network Construction for Mobile Image Super-Resolution
|
In recent years, there has been an increasing demand for real-time super-resolution networks on mobile devices. To address this issue, many lightweight super-resolution models have been proposed. However, these models still contain time-consuming components that increase inference latency, limiting their real-world applications on mobile devices. In this paper, we propose a novel model for singleimage super-resolution based on Equivalent Transformation and Dual Stream network construction (ETDS). ET method is proposed to transform time-consuming operators into time-friendly ones such as convolution and ReLU on mobile devices. Then, a dual stream network is designed to alleviate redundant parameters yielded from ET and enhance the feature extraction ability. Taking full advantage of the advance of ET and the dual stream network structure, we develop the efficient SR model ETDS for mobile devices. The experimental results demonstrate that our ETDS achieves superior inference speed and reconstruction quality compared to prior lightweight SR methods on mobile devices. The code is available at https://github.com/ECNUSR/ETDS.
|
['Lydia Dehbi', 'Zhenbing Zeng', 'Zhengfeng Yang', 'Jiali Gong', 'Hongfan Gao', 'Zhou Zhou', 'Jiahao Chao']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['image-super-resolution']
|
['computer-vision']
|
[ 3.59208792e-01 -1.94939122e-01 -2.04379827e-01 -2.93426007e-01
-4.65598345e-01 -8.78180042e-02 2.45956600e-01 -5.87446392e-01
-3.45706403e-01 5.94075918e-01 2.78932989e-01 -2.20924273e-01
-1.03602745e-02 -1.02250898e+00 -4.79266375e-01 -3.36379677e-01
1.92239061e-01 -1.79176614e-01 6.90828562e-01 -7.27134198e-02
-9.13110524e-02 2.70724028e-01 -1.63569117e+00 4.54997689e-01
1.06482577e+00 1.16519046e+00 6.55980468e-01 4.46858168e-01
-2.64232159e-01 6.20755374e-01 -2.08538339e-01 -3.39180946e-01
1.19021192e-01 -2.20766947e-01 -4.71699655e-01 -3.15540910e-01
2.82373995e-01 -7.92613208e-01 -6.75580621e-01 1.18560159e+00
7.96295762e-01 -2.78957356e-02 -1.18054762e-01 -8.89404893e-01
-4.21773523e-01 8.94948661e-01 -7.13470042e-01 6.01459086e-01
3.23120296e-01 -1.55824587e-01 6.53708220e-01 -1.18201983e+00
5.59503257e-01 1.33359659e+00 8.59609187e-01 3.51138234e-01
-9.07007515e-01 -1.19421625e+00 1.13568306e-02 4.27352577e-01
-1.62751245e+00 -7.74530828e-01 5.05424201e-01 2.46846676e-01
6.81780398e-01 2.16969460e-01 4.62501287e-01 1.09134173e+00
-2.44495571e-01 8.09059918e-01 9.38882411e-01 4.78248969e-02
-1.39400750e-01 2.26233136e-02 -1.16838492e-01 6.73151612e-01
3.98978323e-01 -1.92420974e-01 -8.15401256e-01 -3.09213605e-02
1.44194925e+00 2.40351886e-01 -3.69333893e-01 1.98592320e-01
-1.01964736e+00 2.95930564e-01 4.38633353e-01 3.96804392e-01
-4.33355540e-01 4.61103655e-02 4.38654512e-01 8.12159404e-02
6.55023098e-01 -3.74835998e-01 -1.99081451e-01 -3.38866830e-01
-1.01471412e+00 -2.23167147e-02 4.25802737e-01 1.17472470e+00
5.69346726e-01 -7.94817135e-03 -6.13129921e-02 9.17863369e-01
3.93956676e-02 4.97211069e-01 3.70275944e-01 -1.06187427e+00
7.00286448e-01 4.63279307e-01 -7.22803250e-02 -1.29903066e+00
-3.02900642e-01 -6.55826330e-01 -1.34093165e+00 -4.27662373e-01
-1.79226082e-02 6.72550201e-02 -3.24148089e-01 1.45474875e+00
4.45915192e-01 9.64019477e-01 -1.22585960e-01 9.49438095e-01
1.07306147e+00 8.01963925e-01 -2.38590315e-01 -4.21461225e-01
1.47189057e+00 -7.34700918e-01 -9.28596139e-01 1.20001778e-01
2.64264762e-01 -7.52845049e-01 1.15639341e+00 4.97649014e-01
-1.49275517e+00 -7.68252194e-01 -1.13327026e+00 -4.91235673e-01
1.67495161e-01 2.23661616e-01 7.10516214e-01 3.95073235e-01
-8.56488049e-01 7.29659021e-01 -1.05703998e+00 -1.10232115e-01
8.16083848e-01 4.93437231e-01 1.16989903e-01 -8.63597542e-02
-1.26134527e+00 2.63790578e-01 1.31251186e-01 3.05212706e-01
-3.83211732e-01 -7.06328034e-01 -6.44157469e-01 1.48175493e-01
6.39079750e-01 -7.07287252e-01 1.35788488e+00 -5.88042736e-01
-1.55921626e+00 3.19744647e-01 -5.81341565e-01 -3.41107816e-01
3.93817782e-01 -3.14925462e-01 -6.64856791e-01 2.41915688e-01
1.52474772e-02 -3.27747650e-02 8.42421174e-01 -9.25692916e-01
-8.66745293e-01 -4.10003006e-01 8.13405141e-02 2.85515845e-01
-7.54994452e-01 2.35553309e-01 -7.61391282e-01 -6.20653510e-01
1.87255293e-01 -5.31273723e-01 -9.93165076e-02 1.99522436e-01
-1.30913079e-01 -6.56173006e-02 9.13658977e-01 -7.36110985e-01
1.77366447e+00 -2.36882710e+00 3.10533531e-02 1.28063664e-01
5.43543935e-01 5.86759627e-01 1.30184203e-01 -6.98641539e-02
2.82571852e-01 9.07732770e-02 -9.56685990e-02 -5.41597009e-01
-4.46028948e-01 1.96851283e-01 -3.61875147e-01 6.77784085e-02
-9.28907841e-02 6.13019586e-01 -8.15171123e-01 -8.51399958e-01
-1.00741405e-02 1.01825345e+00 -4.46674317e-01 -6.59254640e-02
1.53357744e-01 4.58768904e-01 -7.69833148e-01 6.16069913e-01
1.12578356e+00 -6.52176738e-01 2.05182835e-01 -5.15387535e-01
-2.95424193e-01 6.11261010e-01 -1.60203159e+00 1.91817176e+00
-6.68355048e-01 1.12882376e-01 3.36239189e-01 -5.71873844e-01
7.28137374e-01 2.47537121e-01 2.64512360e-01 -9.42484260e-01
1.33196354e-01 4.52773869e-01 -4.75313246e-01 -3.78886431e-01
7.41410613e-01 2.03247234e-01 3.75473678e-01 4.14952576e-01
-3.82222116e-01 4.62785631e-01 -5.77227287e-02 2.65078753e-01
9.45457101e-01 3.38484377e-01 1.78859636e-01 -3.11457198e-02
7.37859786e-01 -5.87628603e-01 9.84616160e-01 4.48892683e-01
1.60851777e-01 5.76283455e-01 2.24544797e-02 -3.89361650e-01
-8.31042111e-01 -9.54820693e-01 -1.03897236e-01 8.53363276e-01
5.24923444e-01 -8.43247175e-01 -7.71154523e-01 -1.79359719e-01
-4.56108868e-01 1.36649668e-01 1.92791857e-02 6.81741685e-02
-9.40028250e-01 -7.40348518e-01 5.32868922e-01 6.46276712e-01
1.22990143e+00 -7.07558155e-01 -5.80127954e-01 1.55559748e-01
-5.54591596e-01 -1.39912486e+00 -6.75280571e-01 -6.05100572e-01
-1.18105137e+00 -6.09535575e-01 -5.89625597e-01 -7.18718827e-01
4.12884891e-01 8.56012106e-01 8.79864931e-01 2.94156939e-01
-7.77249038e-02 -1.50583848e-01 -2.68432200e-01 -7.89760128e-02
1.54265285e-01 3.81837368e-01 1.63919091e-01 1.38057902e-01
1.23392120e-01 -1.13573062e+00 -9.61288452e-01 3.92058164e-01
-9.26671207e-01 4.90504622e-01 6.04712486e-01 5.33434808e-01
9.49387193e-01 3.21691543e-01 3.98829818e-01 -8.62585127e-01
5.66053331e-01 -5.11195660e-01 -5.92804849e-01 -6.39027730e-03
-6.79419279e-01 -7.68804364e-03 7.60115564e-01 -6.19634628e-01
-1.52855527e+00 -1.70340523e-01 -1.82048172e-01 -4.23896015e-01
3.43639433e-01 4.02716249e-01 -3.01486254e-01 1.26508594e-01
3.64843845e-01 3.54732543e-01 -6.96674585e-02 -7.98600554e-01
1.36880800e-01 9.19218540e-01 5.95092356e-01 -4.22519296e-01
8.03273857e-01 7.70803630e-01 -7.03916997e-02 -7.97229350e-01
-8.68617892e-01 -1.61033884e-01 -2.21665084e-01 1.06518969e-01
5.50626397e-01 -1.22046471e+00 -6.60224557e-01 4.45053995e-01
-1.22340536e+00 -1.46414831e-01 -7.14151412e-02 4.69356775e-01
-1.83736145e-01 4.49194640e-01 -8.15888047e-01 -6.09244525e-01
-7.65202343e-01 -1.01813745e+00 9.80686843e-01 4.90490913e-01
1.67780504e-01 -4.72481251e-01 -3.75049710e-01 2.09568352e-01
6.90638483e-01 -2.56808162e-01 2.65003771e-01 1.60919800e-01
-1.00748289e+00 3.21229875e-01 -8.33487391e-01 9.65235457e-02
1.76941454e-01 -3.57374758e-01 -9.26125765e-01 -1.49523959e-01
1.66001379e-01 1.77333936e-01 6.97663784e-01 3.08120012e-01
1.61979282e+00 -3.24156374e-01 -4.40982282e-01 1.19908297e+00
1.37908113e+00 -2.37045446e-04 7.48101771e-01 2.74746090e-01
7.81724334e-01 1.06988572e-01 6.87689304e-01 6.41396582e-01
6.76363289e-01 8.80018651e-01 2.61840463e-01 -7.76137561e-02
-2.56919354e-01 -3.20595562e-01 3.44469130e-01 1.23985159e+00
-6.30833447e-01 3.75239030e-02 -3.93599570e-01 1.78124085e-01
-1.98020947e+00 -1.03907466e+00 -2.76546001e-01 2.17770696e+00
9.35514152e-01 2.09417582e-01 1.51485831e-01 9.02367011e-02
7.86674082e-01 2.80741066e-01 -6.80889666e-01 8.18467960e-02
-1.01075724e-01 4.24376488e-01 3.38078022e-01 3.28074813e-01
-7.77789652e-01 7.38935888e-01 5.02940702e+00 1.14877856e+00
-1.11372292e+00 5.14167786e-01 3.83293748e-01 -1.85314715e-01
-4.36269790e-01 -2.49699935e-01 -1.14129770e+00 6.27490222e-01
8.31845880e-01 -2.53088385e-01 6.03148341e-01 5.59877455e-01
4.70567375e-01 1.73536181e-01 -5.26571751e-01 1.40085196e+00
-1.55593753e-01 -1.42785370e+00 -8.76150429e-02 1.32731795e-01
3.37364584e-01 -9.09182280e-02 5.62920384e-02 3.23335379e-02
-1.51125669e-01 -6.77120149e-01 3.50815535e-01 4.27081943e-01
1.26374733e+00 -8.10610056e-01 7.02765644e-01 3.13350320e-01
-1.86297977e+00 8.40086415e-02 -4.65933770e-01 -2.12028995e-01
3.52265388e-01 8.29483509e-01 -4.25638497e-01 7.30777204e-01
9.86945868e-01 1.04402184e+00 -2.70627469e-01 6.77255869e-01
-3.96835916e-02 5.03622472e-01 -4.95460153e-01 2.84224331e-01
-3.63432556e-01 -1.27528131e-01 4.07597125e-01 1.06800199e+00
6.54144347e-01 3.51126194e-01 -5.95959909e-02 8.27933669e-01
-1.95054665e-01 -1.86583191e-01 -3.02346051e-01 2.38298804e-01
9.75378036e-01 1.36762154e+00 -7.42107928e-01 -3.74694586e-01
-6.66587710e-01 1.15403128e+00 6.17196609e-04 1.56346574e-01
-1.10391915e+00 -3.51230234e-01 6.60765648e-01 4.92849916e-01
3.75796020e-01 -3.19237113e-01 -4.05064136e-01 -1.45007241e+00
4.40096557e-01 -7.91621149e-01 2.34144285e-01 -6.39052868e-01
-8.71583581e-01 7.40285516e-01 -1.80300429e-01 -1.38092232e+00
1.74052402e-01 -1.04405984e-01 -3.18134457e-01 8.33306551e-01
-1.70794189e+00 -9.80729878e-01 -6.90561295e-01 7.93777883e-01
6.34618342e-01 1.42476648e-01 5.32565534e-01 8.81832421e-01
-7.70775080e-01 6.82190359e-01 -4.56772931e-02 -1.21927708e-01
4.47011560e-01 -6.54092193e-01 5.29492259e-01 1.04282343e+00
-2.63450772e-01 9.27287281e-01 3.33296597e-01 -6.17653668e-01
-1.48464882e+00 -9.76488352e-01 7.21717596e-01 2.43354738e-02
4.94441450e-01 -4.17070270e-01 -1.11937356e+00 5.00282407e-01
-2.03597814e-01 2.70445347e-01 3.46594125e-01 -5.27288131e-02
-2.62516588e-01 -4.69640702e-01 -1.15621483e+00 5.34218609e-01
1.81757450e+00 -6.31392956e-01 -7.49554038e-02 -4.83605787e-02
9.92778182e-01 -6.43633187e-01 -1.04096746e+00 4.65807050e-01
7.24576414e-01 -1.02100468e+00 1.28557754e+00 2.34060094e-01
4.74421948e-01 -5.63403189e-01 4.39362749e-02 -5.54482877e-01
-2.95461774e-01 -8.14847231e-01 -6.18700385e-01 1.22885072e+00
-4.05544750e-02 -7.07904398e-01 7.09504724e-01 2.97668815e-01
-5.86805260e-03 -8.66330504e-01 -8.40966344e-01 -6.60226583e-01
-7.90539980e-01 -3.41679394e-01 8.23814631e-01 7.75187790e-01
-4.21244889e-01 4.50387836e-01 -4.11941290e-01 3.56099516e-01
7.63736606e-01 1.36701047e-01 6.84835553e-01 -1.28289258e+00
-3.82190019e-01 -8.45067948e-02 7.27088526e-02 -1.56864941e+00
-3.04559946e-01 -7.20167100e-01 -4.21601206e-01 -1.48619354e+00
3.66852522e-01 -7.03210115e-01 -1.33801654e-01 2.18050450e-01
-1.12132646e-01 3.20562840e-01 2.16921773e-02 5.38073003e-01
-7.56377876e-01 4.41606253e-01 1.43139255e+00 4.75902587e-01
-3.20106626e-01 1.62084009e-02 -8.23662281e-01 8.70076835e-01
9.78780866e-01 -2.69498765e-01 -6.70404673e-01 -7.44771421e-01
5.73512077e-01 1.14362225e-01 3.87539595e-01 -9.27200198e-01
3.90183538e-01 6.09206297e-02 1.94910750e-01 -5.98447144e-01
5.09483933e-01 -6.98169589e-01 3.56670558e-01 4.43436354e-01
8.28513503e-02 1.27563015e-01 7.67766982e-02 3.74152362e-01
-7.11595938e-02 9.24746618e-02 5.15137076e-01 -1.13489822e-01
-6.32257044e-01 6.44557834e-01 1.82858482e-01 -1.96265876e-01
4.26993996e-01 -3.99028659e-01 -3.39858681e-01 -3.62125725e-01
-4.76663619e-01 1.05658062e-01 3.39864701e-01 2.43582919e-01
9.10155118e-01 -1.35638809e+00 -6.20845973e-01 2.22277164e-01
-4.60581839e-01 6.78043664e-01 5.84073067e-01 1.00284278e+00
-3.67876619e-01 7.86322057e-02 5.98645620e-02 -4.02389526e-01
-1.47969949e+00 3.09560806e-01 -2.00376939e-02 -3.84864390e-01
-1.02398074e+00 7.83516884e-01 1.12349994e-01 -7.41273910e-02
1.16408102e-01 -3.54349643e-01 -7.61454105e-02 -1.24220207e-01
1.09346318e+00 8.02638710e-01 -1.73951343e-01 -3.87883633e-01
-1.09514281e-01 6.38614833e-01 -1.79728925e-01 1.21464081e-01
1.42211425e+00 -5.53441405e-01 -1.74572095e-01 1.81861848e-01
8.42266142e-01 1.08065456e-01 -1.14641607e+00 -6.72885060e-01
-3.43033701e-01 -6.32348597e-01 2.98561633e-01 -2.91525394e-01
-1.25641572e+00 7.23312676e-01 6.46596491e-01 3.42479907e-02
1.48294008e+00 -2.34755933e-01 1.58347189e+00 7.10612237e-02
7.40542412e-01 -9.32511687e-01 -2.29380921e-01 3.00033092e-01
5.93774199e-01 -9.17875826e-01 1.45725876e-01 -1.03976572e+00
-9.69708785e-02 8.65784705e-01 5.42978704e-01 -2.92755757e-02
5.61041832e-01 5.91803730e-01 -4.99861360e-01 4.15819213e-02
-3.82057041e-01 2.97912098e-02 -1.02239572e-01 4.18797821e-01
3.80398780e-01 -7.18081892e-02 -5.44543803e-01 7.94256628e-01
-1.14258021e-01 6.84902310e-01 3.68493438e-01 8.07200909e-01
-2.15277553e-01 -1.12477612e+00 -1.59421474e-01 4.54657555e-01
-5.73106825e-01 -4.08074290e-01 3.39006543e-01 3.80176544e-01
1.89256340e-01 7.50140488e-01 1.25375316e-01 -5.52928865e-01
2.80735791e-01 -5.41079164e-01 5.02115905e-01 -5.26209831e-01
-2.46778190e-01 1.44954503e-01 1.13110371e-01 -9.61889982e-01
-6.29115522e-01 -5.63672006e-01 -1.51177537e+00 -6.36869431e-01
-3.36784244e-01 -1.78924069e-01 4.46079761e-01 5.83943009e-01
9.08243597e-01 6.63126647e-01 5.24580240e-01 -8.53332818e-01
-3.05925250e-01 -7.89924145e-01 -4.01713580e-01 -9.23071150e-03
1.67541429e-01 -5.50268233e-01 -1.34486640e-02 1.21662915e-02]
|
[11.036898612976074, -1.8269202709197998]
|
1eff5ebb-076b-435d-b5fa-68e52d48a33e
|
chemberta-2-towards-chemical-foundation
|
2209.01712
| null |
https://arxiv.org/abs/2209.01712v1
|
https://arxiv.org/pdf/2209.01712v1.pdf
|
ChemBERTa-2: Towards Chemical Foundation Models
|
Large pretrained models such as GPT-3 have had tremendous impact on modern natural language processing by leveraging self-supervised learning to learn salient representations that can be used to readily finetune on a wide variety of downstream tasks. We investigate the possibility of transferring such advances to molecular machine learning by building a chemical foundation model, ChemBERTa-2, using the language of SMILES. While labeled data for molecular prediction tasks is typically scarce, libraries of SMILES strings are readily available. In this work, we build upon ChemBERTa by optimizing the pretraining process. We compare multi-task and self-supervised pretraining by varying hyperparameters and pretraining dataset size, up to 77M compounds from PubChem. To our knowledge, the 77M set constitutes one of the largest datasets used for molecular pretraining to date. We find that with these pretraining improvements, we are competitive with existing state-of-the-art architectures on the MoleculeNet benchmark suite. We analyze the degree to which improvements in pretraining translate to improvement on downstream tasks.
|
['Bharath Ramsundar', 'Gabriel Grand', 'Seyone Chithrananda', 'Elana Simon', 'Walid Ahmad']
|
2022-09-05
| null | null | null | null |
['molecular-property-prediction']
|
['miscellaneous']
|
[ 5.98668635e-01 8.60658512e-02 -6.62275255e-01 -5.12824297e-01
-7.00107813e-01 -8.57627034e-01 7.35309362e-01 7.88225651e-01
-5.31318247e-01 1.19555414e+00 3.35332513e-01 -6.96430087e-01
8.76840204e-03 -7.12265730e-01 -1.17976654e+00 -5.58557093e-01
-2.27748662e-01 5.65067232e-01 1.35223055e-02 -3.64616126e-01
3.47239465e-01 5.42368770e-01 -1.06623983e+00 8.73643398e-01
7.67078400e-01 6.20770633e-01 1.47239223e-01 3.68482262e-01
-7.39504993e-02 4.59298015e-01 -2.50266522e-01 -2.77536303e-01
1.13853969e-01 -2.73676634e-01 -1.06930172e+00 -5.92764139e-01
5.89012921e-01 1.82369068e-01 -2.39055499e-01 7.05731869e-01
6.42525733e-01 3.65528464e-01 7.64673114e-01 -4.48377341e-01
-5.39012492e-01 9.96594548e-01 -3.01735491e-01 4.12251711e-01
1.85422733e-01 4.97234762e-01 1.15689039e+00 -7.24090099e-01
1.01615548e+00 1.32504416e+00 5.81933498e-01 6.92322493e-01
-1.61289811e+00 -8.59808445e-01 8.89199749e-02 1.30000472e-01
-9.35590088e-01 -6.00245595e-01 4.25761670e-01 -5.22038937e-01
1.72180319e+00 -1.56590238e-01 1.78180873e-01 1.28010762e+00
5.41017413e-01 3.64853650e-01 1.06066072e+00 -3.40288162e-01
2.37232119e-01 -9.97930989e-02 1.99453488e-01 8.53442430e-01
8.16938430e-02 2.97387183e-01 -7.39428639e-01 -2.26346195e-01
3.97920012e-01 -8.76546800e-02 -4.30187136e-02 -2.94509321e-01
-1.21765149e+00 1.09431219e+00 8.10284317e-01 1.89655721e-01
-8.88485387e-02 2.43533045e-01 5.95933855e-01 4.61389869e-01
4.87344533e-01 1.22791874e+00 -9.77779090e-01 8.78706649e-02
-7.66074836e-01 1.50189385e-01 7.67222345e-01 6.20192349e-01
6.95293546e-01 -4.28735502e-02 -1.62196234e-02 6.61827803e-01
1.74068376e-01 1.04069859e-01 5.59040785e-01 -4.22277123e-01
6.58300161e-01 3.22170198e-01 -3.23809236e-01 -3.39375317e-01
-6.45340979e-01 -4.75074381e-01 -5.75713813e-01 2.37921402e-01
4.46605414e-01 -2.44968832e-01 -1.37579787e+00 1.80149472e+00
1.13955893e-01 -6.97129145e-02 2.89465547e-01 4.94666845e-01
1.06020796e+00 1.03652501e+00 8.11369419e-01 -4.36164550e-02
1.02508473e+00 -9.44028378e-01 -1.02854185e-01 -1.30907908e-01
1.12289560e+00 -7.48786926e-01 8.37424099e-01 5.80533206e-01
-7.89985538e-01 -6.50931954e-01 -1.33282495e+00 -1.94364622e-01
-8.70582461e-01 -3.32958788e-01 1.35857642e+00 5.87567866e-01
-6.89525127e-01 1.46722496e+00 -7.27266014e-01 -3.61227781e-01
7.60777056e-01 7.94319570e-01 -5.87877154e-01 -1.21155389e-01
-1.30194151e+00 1.22413194e+00 8.49382162e-01 -4.91564631e-01
-1.32990408e+00 -1.14186573e+00 -6.19263113e-01 1.08254971e-02
2.45734036e-01 -8.15215051e-01 1.15841234e+00 -7.32767463e-01
-1.57191622e+00 8.82886410e-01 -6.40461082e-03 -7.44414508e-01
4.56083454e-02 2.10702550e-02 -3.35652322e-01 -1.06616177e-01
-1.14545755e-01 1.14244401e+00 5.41274130e-01 -8.27259481e-01
-2.29795277e-01 -1.94014236e-01 1.09341972e-01 1.23149820e-01
-9.75847244e-02 1.58466976e-02 2.21984580e-01 -4.16755527e-01
-4.00872916e-01 -8.94042671e-01 -7.00304151e-01 -4.26933169e-01
-4.55146581e-01 -2.93568730e-01 4.37349737e-01 -1.19917601e-01
7.05663741e-01 -1.60006332e+00 4.05825436e-01 2.12330639e-01
3.27726603e-01 4.22070235e-01 -6.12411559e-01 6.20883882e-01
-7.70155191e-01 3.72484386e-01 -7.99706802e-02 -1.47039056e-01
-2.64082372e-01 4.66511771e-02 -3.38298261e-01 3.01047713e-01
3.25591415e-01 1.06321669e+00 -9.76013243e-01 -1.14225514e-01
7.73985013e-02 3.15891325e-01 -8.87180865e-01 1.71962231e-02
-1.00757122e+00 6.15480185e-01 -5.35901189e-01 6.25590563e-01
1.79860950e-01 -4.04299676e-01 2.35033795e-01 -1.51659787e-01
-3.14882904e-01 8.52800429e-01 -3.09956908e-01 2.11934853e+00
-4.29862052e-01 3.11550945e-01 -4.09636110e-01 -1.02899802e+00
8.19297373e-01 1.44502148e-01 4.75084931e-01 -4.63134378e-01
-5.52398190e-02 1.40222296e-01 6.79151297e-01 -2.54699022e-01
3.13943535e-01 -5.48682511e-01 1.26970366e-01 2.48408601e-01
4.90712762e-01 -3.12015682e-01 4.66806829e-01 1.94059595e-01
1.07927120e+00 3.96029145e-01 3.92417699e-01 -5.06918967e-01
3.03184509e-01 2.78764963e-01 1.97875902e-01 6.08195841e-01
2.18745023e-01 1.28069043e-01 3.64922196e-01 -6.82762861e-01
-1.12690639e+00 -6.86105967e-01 -4.31044072e-01 1.63033187e+00
-4.52411622e-01 -7.72907376e-01 -4.30352062e-01 -6.07029378e-01
3.13657522e-01 6.36285722e-01 -5.75263858e-01 -3.32074314e-01
-3.20975691e-01 -1.27079988e+00 5.03021300e-01 3.53963703e-01
-3.06819808e-02 -1.27170086e+00 1.11343227e-01 5.37222445e-01
6.68848813e-01 -7.40062833e-01 -2.56724089e-01 1.05112243e+00
-1.06483316e+00 -9.85191226e-01 -3.25662494e-01 -5.69394827e-01
3.19489807e-01 -9.99873504e-02 1.44429862e+00 -2.39143550e-01
-3.83119941e-01 -3.04783374e-01 1.88444112e-03 -5.00696421e-01
-6.09618008e-01 7.21375227e-01 1.37238562e-01 -6.57686710e-01
1.88028991e-01 -8.68539929e-01 -5.12363315e-01 -1.49863034e-01
-7.45649457e-01 1.54387206e-01 7.45111823e-01 8.25310290e-01
6.15098894e-01 -3.43656003e-01 7.98204660e-01 -1.49339437e+00
3.96170586e-01 -6.92090511e-01 -5.08561373e-01 -6.14829920e-02
-6.52926266e-01 6.91837013e-01 9.05414104e-01 -3.46316397e-01
-8.41581285e-01 3.25844318e-01 -4.09666777e-01 -1.99011788e-02
-3.26616228e-01 9.23561633e-01 -7.31829703e-02 -5.11948466e-01
9.89678919e-01 -2.30482563e-01 -3.38637412e-01 -5.60022712e-01
8.48516941e-01 1.22540623e-01 2.03299776e-01 -1.06769109e+00
5.44457495e-01 7.16323033e-02 4.72370028e-01 -6.59519732e-01
-1.12749994e+00 -4.25176352e-01 -8.00477743e-01 5.76237261e-01
8.49206030e-01 -9.80237484e-01 -9.12172377e-01 -1.11434972e-02
-1.02528703e+00 -6.97820663e-01 -1.14298135e-01 2.98337042e-01
-5.38930237e-01 2.34446391e-01 -8.03067505e-01 -1.04521893e-01
-4.23465014e-01 -1.22434998e+00 8.03303301e-01 -1.18546829e-01
-3.52017999e-01 -1.14271140e+00 3.92854452e-01 3.54329556e-01
2.70251811e-01 4.03392732e-01 1.46313310e+00 -9.58353877e-01
-5.54167807e-01 2.74823546e-01 -1.09802663e-01 -8.48501176e-02
1.96061447e-01 -2.37627879e-01 -9.36407983e-01 -3.27850103e-01
-6.31457925e-01 -8.68643463e-01 1.61182046e+00 1.72051117e-01
1.38964105e+00 -4.82115671e-02 -6.50191009e-01 9.82636511e-01
1.21450913e+00 2.72959381e-01 5.84396899e-01 4.17383164e-01
9.12479401e-01 3.57599407e-01 2.28511244e-01 9.40840542e-02
-8.39788392e-02 4.50412184e-01 3.83817613e-01 -2.80833930e-01
-6.73536658e-02 -5.67318022e-01 3.27472836e-01 4.51997727e-01
-1.37821838e-01 -3.62844825e-01 -9.90870357e-01 1.28365951e-02
-1.54008651e+00 -8.26031029e-01 2.37403452e-01 1.98805130e+00
1.47607565e+00 3.22938472e-01 -3.11117657e-02 -2.99911439e-01
1.81496635e-01 2.81727642e-01 -8.83025348e-01 -5.54266989e-01
-3.93929817e-02 9.27072942e-01 8.67848277e-01 5.25054634e-01
-1.40594149e+00 1.40099633e+00 6.65797615e+00 9.21897829e-01
-1.36150193e+00 -5.18060364e-02 9.00325775e-01 -1.27858981e-01
-2.77395070e-01 1.09802194e-01 -1.19749880e+00 2.57123023e-01
1.58058989e+00 -4.88342382e-02 2.88482308e-01 8.80802095e-01
2.37824336e-01 3.51080447e-01 -1.70906353e+00 7.60509014e-01
-2.11764410e-01 -2.04226470e+00 4.08112794e-01 7.53047466e-02
9.53129649e-01 5.19131362e-01 1.89188898e-01 5.32887459e-01
6.24564588e-01 -1.73647738e+00 1.74691945e-01 2.14782983e-01
8.95748675e-01 -8.29002917e-01 2.82184184e-01 5.51236682e-02
-7.92643666e-01 2.30893672e-01 -5.36609113e-01 -1.16383411e-01
-1.29022926e-01 5.02215624e-01 -1.17612100e+00 4.38003778e-01
1.24387406e-01 1.08897483e+00 -5.06214380e-01 8.86959553e-01
-2.08970174e-01 6.66721463e-01 -2.94456273e-01 -8.34621787e-02
5.64707935e-01 -3.87055911e-02 1.97063267e-01 1.41654158e+00
-3.15335691e-02 8.33168924e-02 5.72915137e-01 7.36027062e-01
-7.06834912e-01 2.70209193e-01 -6.39725566e-01 -5.62372446e-01
1.44889221e-01 1.09121764e+00 -4.72359300e-01 -4.99672174e-01
-2.69775718e-01 5.95602274e-01 5.46915412e-01 2.26271719e-01
-5.65296113e-01 -1.93578497e-01 8.24631035e-01 1.74513143e-02
1.79586098e-01 -4.15412277e-01 -2.63275981e-01 -1.00934005e+00
-7.96154559e-01 -1.10039425e+00 4.42733198e-01 -4.42053676e-01
-1.40180254e+00 4.11845177e-01 -1.99265122e-01 -4.28303897e-01
2.25216039e-02 -1.19016623e+00 -5.93661547e-01 8.61658037e-01
-1.62994552e+00 -1.13136649e+00 3.88032198e-01 2.32977480e-01
4.97420251e-01 -3.95066500e-01 1.19367599e+00 1.11258097e-01
-6.67110860e-01 5.30867338e-01 3.72178346e-01 -5.00677288e-01
1.03843582e+00 -1.22260308e+00 7.29635477e-01 2.27902949e-01
1.08169757e-01 1.06972826e+00 6.11298442e-01 -7.50561476e-01
-1.48523474e+00 -1.29012525e+00 4.63544279e-01 -6.76990032e-01
8.16128910e-01 -5.06856382e-01 -8.81337047e-01 6.46269202e-01
1.95258200e-01 -1.14767917e-01 1.04252315e+00 6.66660309e-01
-7.30072916e-01 2.16035396e-01 -8.49593580e-01 3.53888541e-01
1.30367196e+00 -5.06314874e-01 -5.03240049e-01 8.66152704e-01
1.00410318e+00 -4.78392184e-01 -1.25854719e+00 4.18629825e-01
2.83918113e-01 -3.99848163e-01 1.32029939e+00 -1.54860294e+00
8.74121130e-01 4.37883325e-02 2.01673470e-02 -1.54268479e+00
-5.02507091e-01 -7.03467071e-01 3.61493230e-02 8.72532308e-01
1.18389225e+00 -6.14599168e-01 9.93990660e-01 2.52871633e-01
-5.63533902e-01 -8.74071598e-01 -6.75169826e-01 -6.01534247e-01
7.82676935e-01 2.18132138e-03 4.42911148e-01 1.03162968e+00
2.91159451e-01 9.38689888e-01 -2.32208654e-01 -3.19123656e-01
3.78823876e-01 2.96637416e-01 5.39083183e-01 -1.21931207e+00
-6.96645260e-01 -4.37373757e-01 -9.45360586e-02 -1.11968112e+00
4.03689206e-01 -1.61458588e+00 -2.00564206e-01 -1.13229513e+00
5.81502795e-01 -4.55072343e-01 -6.56495512e-01 9.07633722e-01
-9.74171460e-02 -3.86567265e-02 2.69846665e-03 8.20549950e-02
-5.57224512e-01 4.78071809e-01 1.22309029e+00 -4.36037421e-01
-2.71666110e-01 -3.47172290e-01 -8.55938971e-01 4.84638274e-01
9.20295000e-01 -4.87664044e-01 -3.47834855e-01 -2.81688154e-01
2.09128037e-01 -2.59683281e-01 -1.51166469e-01 -8.33688676e-01
3.42359133e-02 -3.24854642e-01 7.18390882e-01 -4.03593838e-01
4.37340498e-01 -1.26770064e-01 -1.22401342e-01 5.05031824e-01
-8.33555460e-01 -2.56441861e-01 6.92916870e-01 6.37518167e-01
1.10453308e-01 -1.24877848e-01 9.68406260e-01 -5.32159746e-01
-6.43696666e-01 6.88110411e-01 -3.33730251e-01 -1.31878287e-01
6.02559686e-01 2.86565095e-01 -6.49685025e-01 1.00709900e-01
-7.30625093e-01 1.19609818e-01 4.15008008e-01 2.15199172e-01
3.30503106e-01 -9.02950048e-01 -4.44680810e-01 -1.19315058e-01
7.79607892e-02 -2.38304213e-01 -2.29944706e-01 2.82065630e-01
-3.80456746e-01 1.00583327e+00 -3.20901811e-01 -2.64461666e-01
-1.20994461e+00 8.72752607e-01 2.36516669e-01 -5.43961108e-01
-1.51811495e-01 9.55497682e-01 4.18245107e-01 -4.75648493e-01
3.66320871e-02 -5.14541090e-01 -1.23634525e-01 -2.77478974e-02
2.82539904e-01 -2.39893690e-01 2.22594097e-01 -2.10915267e-01
-3.13960999e-01 2.98922777e-01 -6.17727280e-01 4.34532732e-01
1.70106673e+00 6.97661221e-01 2.52422858e-02 1.16752177e-01
1.40295780e+00 -2.20961988e-01 -1.07366157e+00 -1.23257034e-01
1.70289323e-01 8.13644230e-02 7.25499094e-02 -1.04374111e+00
-6.03918612e-01 1.11071777e+00 4.23697799e-01 -6.38652325e-01
4.26395178e-01 -2.75199097e-02 7.13988245e-01 1.18290985e+00
3.74016732e-01 -7.50122905e-01 2.69454062e-01 7.51617372e-01
6.33699894e-01 -1.40766430e+00 3.48268867e-01 -3.10327590e-01
-3.00669849e-01 1.15893209e+00 4.48510289e-01 -4.49370891e-02
3.54629785e-01 1.65564254e-01 -3.32898349e-01 -3.64779353e-01
-1.03351390e+00 -2.00068966e-01 2.88454115e-01 5.53225756e-01
1.13126647e+00 1.13016516e-01 -1.02634236e-01 3.40013564e-01
-2.68461972e-01 -1.35773316e-01 1.96395054e-01 9.12082791e-01
-8.02306712e-01 -1.62660873e+00 -1.83928888e-02 4.96107280e-01
-6.76494062e-01 -8.13182354e-01 -6.56047583e-01 5.40161967e-01
1.83000743e-01 6.16048694e-01 -4.22325552e-01 -2.33955741e-01
1.44891351e-01 3.07432681e-01 8.29628527e-01 -1.24501419e+00
-6.77451074e-01 -2.88845003e-01 4.59756523e-01 -4.27323282e-01
-2.88398117e-01 -4.37133580e-01 -1.34276414e+00 -2.93427467e-01
-1.19454034e-01 3.36971074e-01 5.67102909e-01 8.21360111e-01
4.78935272e-01 6.20400608e-01 2.75847882e-01 -1.18556845e+00
-6.26364112e-01 -9.38160479e-01 5.72416596e-02 2.05335006e-01
1.22081153e-01 -6.59954906e-01 7.32062310e-02 1.37024760e-01]
|
[5.004796504974365, 5.849133014678955]
|
0eb67ea0-a1bd-47e2-b172-669e20ff77f2
|
modelling-disease-impact-lifespan-reduction
|
2305.06808
| null |
https://arxiv.org/abs/2305.06808v1
|
https://arxiv.org/pdf/2305.06808v1.pdf
|
Modelling disease impact: lifespan reduction is greatest for young adults in an exogenous damage model of disease
|
We model the effects of disease and other exogenous damage during human aging. While the exogenous damage is repaired at the end of acute disease, propagated secondary damage remains. We consider both short-term mortality effects due to (acute) exogenous damage and long-term mortality effects due to propagated damage within the context of a generic network model (GNM) of individual aging. Across a wide range of disease durations and severities we find that while excess short-term mortality is highest for the oldest individuals, the long-term years of life lost are highest for the youngest individuals. These appear to be universal effects of human disease. We support this conclusion with a phenomenological model coupling damage and mortality. Our results are qualitatively consistent with existing observational studies, though these are mostly limited to short time-horizons. Short-time horizon studies may have significant limitations for understanding the lifetime impacts of disease on both individuals and populations.
|
['Andrew D. Rutenberg', 'Glen Pridham', 'Rebecca Tobin']
|
2023-05-11
| null | null | null | null |
['human-aging']
|
['miscellaneous']
|
[-1.51011959e-01 1.55449480e-01 -3.51435155e-01 4.99877274e-01
1.97304979e-01 -2.68905967e-01 7.10237086e-01 6.39767230e-01
-5.88761210e-01 1.06311691e+00 6.41811967e-01 -2.75097519e-01
-4.93969947e-01 -9.09848392e-01 -2.02552572e-01 -5.51953495e-01
-7.24779069e-01 2.21721217e-01 8.43256861e-02 -3.44796330e-01
-2.54605681e-01 2.56524473e-01 -1.20622861e+00 -7.45612621e-01
8.89374614e-01 1.94411501e-01 -1.81862712e-01 5.11407137e-01
6.70129776e-01 4.10818934e-01 -7.02472031e-01 -4.54102486e-01
-7.35486438e-03 -3.93671334e-01 -3.70363772e-01 -1.90563619e-01
-3.31165493e-02 -6.02087379e-01 -8.85839224e-01 3.62739474e-01
7.48411953e-01 -1.22552581e-01 8.12683582e-01 -1.39835775e+00
-7.06494391e-01 2.10491672e-01 -3.62940341e-01 4.87842143e-01
2.81538665e-01 3.89182806e-01 7.08631992e-01 -4.09376383e-01
6.52342856e-01 1.53928387e+00 1.21267581e+00 5.64025521e-01
-1.48856354e+00 -2.15719029e-01 8.50435570e-02 -2.30211347e-01
-1.11420846e+00 -5.09128988e-01 1.93163231e-01 -6.61416531e-01
8.23765814e-01 1.55784711e-01 9.82769430e-01 1.14080167e+00
1.16934371e+00 -3.05378199e-01 8.28255832e-01 -1.91320539e-01
2.37701535e-01 -5.12005925e-01 4.72027063e-01 5.17041802e-01
1.23696232e+00 6.32094204e-01 -3.65242749e-01 -5.76790690e-01
8.04443359e-01 2.70256877e-01 -3.57027143e-01 2.74213493e-01
-9.97657299e-01 3.61380458e-01 2.74453759e-01 1.29405782e-01
-6.67928040e-01 6.24711633e-01 2.08463252e-01 4.62067783e-01
6.70015514e-01 -5.31362332e-02 -6.19670153e-01 1.34289816e-01
-7.95251429e-01 4.04272169e-01 7.64775097e-01 1.49849132e-01
5.62948287e-01 -1.59431338e-01 -2.41931796e-01 5.40113032e-01
3.91974360e-01 7.84604013e-01 -4.50478531e-02 -1.11562121e+00
6.50972351e-02 2.86545426e-01 1.52797714e-01 -7.72822261e-01
-8.19511414e-01 -9.85100806e-01 -1.03319895e+00 3.16536546e-01
6.65506542e-01 -7.26935565e-01 -6.74599826e-01 2.20079494e+00
1.62579954e-01 7.65448660e-02 -2.37168625e-01 3.54687929e-01
2.38103852e-01 4.11007851e-01 5.78270137e-01 -6.47894919e-01
1.34377921e+00 -4.02017832e-01 -5.87231994e-01 -4.64627802e-01
6.34500861e-01 -1.26811147e-01 3.92869025e-01 -1.31870285e-01
-1.20113862e+00 -5.37144616e-02 -6.49859786e-01 1.94461733e-01
-1.84350878e-01 -5.78325689e-01 3.34941298e-01 7.18267381e-01
-1.30975246e+00 7.98864961e-01 -8.94786239e-01 -1.20410311e+00
1.42299220e-01 -3.63719948e-02 -2.38892585e-02 -1.54694274e-01
-1.44978213e+00 1.15974557e+00 -5.19802272e-01 8.66030902e-02
-1.27319539e+00 -1.21170771e+00 -2.27780387e-01 -2.89883167e-02
3.07925809e-02 -1.58085930e+00 7.50559986e-01 -3.91698420e-01
-3.95622760e-01 5.03455460e-01 -5.99144027e-02 -3.57288748e-01
3.74360532e-01 -1.73951879e-01 -3.97405386e-01 2.60041282e-02
3.38512391e-01 1.92170233e-01 1.75711378e-01 -1.15744555e+00
2.59460621e-02 -7.52862096e-01 3.20210643e-02 3.57239217e-01
-3.68488491e-01 1.62366480e-01 3.41168582e-01 -7.65150964e-01
-4.64641124e-01 -8.44678938e-01 -5.70653558e-01 2.76829392e-01
-2.45858088e-01 1.33662224e-01 3.07496041e-01 -9.32801425e-01
1.51805675e+00 -1.69741023e+00 2.15928674e-01 -3.44590604e-01
6.31835699e-01 -2.17259735e-01 -1.66250486e-02 1.14287961e+00
5.36266454e-02 3.83076549e-01 -2.05098093e-01 -3.83118242e-01
-2.39143282e-01 3.12583119e-01 7.74418041e-02 6.78075969e-01
-7.48757794e-02 7.47334361e-01 -8.59711111e-01 -1.43285736e-01
-3.59967470e-01 3.61401588e-01 -1.06697030e-01 -3.38144809e-01
2.05661491e-01 2.26847231e-01 -2.88004041e-01 6.36013091e-01
3.27265739e-01 -1.37909010e-01 1.81492016e-01 4.89796102e-01
-1.58353597e-01 -8.11801758e-03 -3.05025607e-01 7.84964681e-01
-4.56280820e-02 5.27368523e-02 2.56960154e-01 -6.80558920e-01
1.39057368e-01 5.62243521e-01 3.67883563e-01 -3.44038665e-01
-6.31246790e-02 1.59030348e-01 5.09905219e-01 -2.59183109e-01
1.50657818e-01 -6.43385530e-01 -1.20030425e-01 8.15126479e-01
-1.06783025e-01 5.33830464e-01 2.34374896e-01 5.71535289e-01
1.69559646e+00 -5.16063094e-01 3.14036727e-01 -5.74772894e-01
-2.27714792e-01 -1.35542005e-01 7.97098517e-01 5.87816238e-01
-6.67415798e-01 -5.51373400e-02 8.41852129e-01 4.64615189e-02
-1.08287728e+00 -1.48053241e+00 -2.90786892e-01 7.22484350e-01
4.09431159e-02 -3.25773031e-01 -4.77028221e-01 -1.89137578e-01
5.56244671e-01 3.48940313e-01 -7.23030746e-01 -6.60261810e-01
-2.43134558e-01 -1.44786716e+00 8.44161332e-01 2.58583158e-01
1.47867635e-01 -8.02563190e-01 -4.24679130e-01 2.78057635e-01
1.42568573e-01 -4.23051417e-01 -3.76180261e-01 -2.70996541e-01
-1.33387029e+00 -1.12334096e+00 -1.16860151e+00 -2.98825055e-01
7.37218797e-01 6.64621145e-02 1.06522238e+00 9.17677283e-01
-4.93478268e-01 8.18098128e-01 1.79281622e-01 -2.00426221e-01
-3.52576494e-01 -1.11606494e-01 6.96703553e-01 -6.40082538e-01
-3.84439707e-01 -7.54250228e-01 -1.09717822e+00 2.66339239e-02
-5.28664052e-01 -7.32487619e-01 5.19008696e-01 3.14306021e-01
-3.04006398e-01 2.56184757e-01 1.32759190e+00 -2.79763937e-01
7.94806123e-01 -1.04200947e+00 1.76317006e-01 2.01444507e-01
-1.06977236e+00 -5.06580830e-01 2.69560099e-01 -4.66658086e-01
-1.04931927e+00 -7.67944276e-01 4.26276207e-01 6.31506741e-01
5.76273203e-02 7.97591150e-01 1.00434721e-01 3.46211821e-01
5.61645687e-01 -2.56208837e-01 3.89843702e-01 -5.19724309e-01
1.42671332e-01 2.42929876e-01 3.17968905e-01 -6.15058303e-01
6.94627702e-01 4.18729216e-01 4.78749692e-01 -9.46222246e-01
-3.26330066e-01 2.87829161e-01 -2.93724805e-01 -7.18796134e-01
6.01309955e-01 -1.24159694e+00 -4.93898422e-01 8.20707798e-01
-6.29289031e-01 -8.48112941e-01 -2.40775108e-01 4.55653965e-01
-1.19682506e-01 3.12727392e-01 -1.07751095e+00 -1.00625658e+00
-2.61784106e-01 -3.18013906e-01 4.42876101e-01 1.30243570e-01
-5.15334189e-01 -1.73104262e+00 5.71805775e-01 -2.77937502e-02
4.25880194e-01 5.83703101e-01 1.14266360e+00 1.17857352e-01
-3.34637463e-01 -8.93331617e-02 1.84299462e-02 -1.56719163e-01
2.33054787e-01 1.72046855e-01 -6.50980175e-02 -5.50170898e-01
-2.19989106e-01 -5.44243008e-02 9.50906813e-01 8.92785966e-01
2.33332328e-02 -4.01750803e-01 -9.44763958e-01 -4.62341085e-02
1.42230642e+00 -1.34466022e-01 6.23588741e-01 1.32163092e-01
1.76928028e-01 8.42690527e-01 3.37630898e-01 3.48002434e-01
6.68148458e-01 4.35131073e-01 2.40921915e-01 -1.89274043e-01
-4.61769611e-01 -6.80337623e-02 6.32386088e-01 5.60941041e-01
-2.31152490e-01 -5.21718025e-01 -1.03092980e+00 9.90671217e-01
-1.62733066e+00 -1.14263487e+00 -5.71139693e-01 2.61159444e+00
7.10155308e-01 4.26600426e-01 6.14494860e-01 -5.62094860e-02
7.39356637e-01 -2.27535497e-02 -5.42126834e-01 -1.37857825e-01
-2.18002051e-01 -1.05175249e-01 7.95618296e-01 6.65166378e-01
-3.00824255e-01 8.96659642e-02 8.30151272e+00 8.50299373e-02
-4.32940245e-01 2.53736168e-01 7.05710530e-01 -4.76869017e-01
-2.70886213e-01 3.79996091e-01 -7.01617181e-01 5.46598256e-01
1.57052112e+00 -5.57243586e-01 2.74451792e-01 -2.26054981e-01
7.74016201e-01 -4.23604578e-01 -7.57116973e-01 -1.63269252e-01
-3.91330302e-01 -4.67535198e-01 -4.54076350e-01 4.95046943e-01
7.21216261e-01 4.22302186e-02 -1.61379799e-01 6.55958205e-02
5.35588622e-01 -6.95216000e-01 6.38398945e-01 1.10857952e+00
1.03265548e+00 -5.47251523e-01 4.32082057e-01 3.49940062e-01
-8.60369623e-01 -2.96841234e-01 3.80317145e-03 -8.33539784e-01
6.90284967e-01 1.32821262e+00 6.71176147e-03 2.11561292e-01
3.72493356e-01 6.83436990e-01 -7.36979604e-01 1.19923878e+00
-7.91811869e-02 7.87731826e-01 -3.03717166e-01 5.21242321e-01
-4.74839836e-01 6.65835589e-02 6.12000644e-01 4.74106491e-01
4.13344085e-01 2.42623035e-02 -3.62391502e-01 6.46310866e-01
9.41020474e-02 -7.43601441e-01 -5.65172672e-01 -2.48048261e-01
7.98345089e-01 1.04587126e+00 -5.51687598e-01 -3.68633807e-01
-3.13431919e-01 6.91087246e-01 -2.06904635e-02 6.86157405e-01
-3.77750158e-01 -3.15200448e-01 9.21777129e-01 6.59468055e-01
-5.36433995e-01 -2.69647360e-01 -3.28985453e-01 -9.09565747e-01
-2.43784532e-01 -2.39489719e-01 4.14325476e-01 -5.56276917e-01
-1.36436343e+00 -2.62977839e-01 1.44253522e-01 -4.92917448e-01
1.11400522e-01 -1.20094426e-01 -9.59495068e-01 8.13568592e-01
-9.05243933e-01 -7.00525105e-01 2.01185063e-01 -1.83889200e-03
-1.55337900e-01 4.35102493e-01 6.49623454e-01 4.34720069e-01
-1.07458591e+00 1.81608155e-01 2.91195005e-01 -4.95609581e-01
7.59370148e-01 -1.01990497e+00 4.99208421e-01 8.33747506e-01
-1.05950201e+00 8.97916615e-01 8.76563787e-01 -1.51866412e+00
-8.80521297e-01 -1.15204883e+00 1.28216887e+00 -4.87246960e-01
6.86798692e-01 -2.36587524e-01 -5.26741445e-01 7.95811176e-01
1.94676653e-01 -4.16755646e-01 5.75306535e-01 3.36390316e-01
1.07843868e-01 -7.17529655e-02 -1.20424271e+00 8.42567623e-01
1.40045214e+00 -3.37005079e-01 -2.44607136e-01 1.89749673e-01
7.00227737e-01 6.89935267e-01 -1.27771199e+00 3.68907511e-01
8.27663183e-01 -6.84852839e-01 1.21439588e+00 -3.72032404e-01
4.53467339e-01 5.40623702e-02 3.70243728e-01 -1.39612913e+00
-7.16995418e-01 -3.05980682e-01 -1.93959936e-01 1.28894913e+00
3.68865669e-01 -1.06671059e+00 1.25713766e-01 5.50904095e-01
9.24628302e-02 -7.30372250e-01 -8.49520743e-01 -1.13698792e+00
5.17096460e-01 4.20540273e-01 2.79812604e-01 8.25316131e-01
1.43791467e-01 2.24773347e-01 -1.65680349e-01 2.23442167e-01
9.83840108e-01 -7.77669251e-01 2.29319856e-01 -1.73491275e+00
5.31289689e-02 -3.08842093e-01 -1.85389027e-01 -1.08362444e-01
-1.71732649e-01 -3.65406692e-01 -1.97969720e-01 -1.80907774e+00
7.46025503e-01 -5.08149087e-01 -3.12629700e-01 2.31751025e-01
-3.52828205e-01 3.86106707e-02 -1.05301186e-01 3.12859118e-01
-8.01524520e-02 4.40701604e-01 1.00444067e+00 3.09802681e-01
-9.29620564e-02 -1.42061174e-01 -8.98761570e-01 3.76863718e-01
8.80007207e-01 -5.21996140e-01 -4.96089846e-01 -6.13910854e-02
4.19748127e-01 6.13435268e-01 1.01349926e+00 -1.00321317e+00
-1.74535498e-01 -3.79889309e-01 3.04098219e-01 -2.03008026e-01
2.57039696e-01 -3.33642364e-01 6.26828730e-01 1.32917416e+00
-6.03820570e-02 6.39899522e-02 -8.43436643e-02 9.04922426e-01
8.04021001e-01 4.60277796e-02 5.60660899e-01 -8.76918659e-02
3.66296768e-01 3.67266953e-01 -1.07856381e+00 2.79424768e-02
1.02290046e+00 4.15661447e-02 -1.03745115e+00 -5.59925139e-01
-1.10611081e+00 3.56265843e-01 9.86988842e-01 2.20051799e-02
5.19769154e-02 -1.18872750e+00 -5.94038188e-01 -5.56793928e-01
-3.06857109e-01 -8.47792268e-01 3.66578072e-01 1.18102682e+00
-2.33071819e-01 4.17083114e-01 -1.40701681e-01 2.92045921e-01
-1.04587746e+00 7.33257294e-01 5.73760271e-01 -2.49572128e-01
-4.91474658e-01 5.42368114e-01 2.08500773e-01 1.05019465e-01
-4.07235175e-02 1.61086097e-01 3.61666054e-01 2.07986042e-01
3.80098343e-01 1.23814332e+00 -6.08260512e-01 -5.91833830e-01
-4.42761451e-01 4.24120575e-01 2.43128449e-01 -1.63531423e-01
1.12004983e+00 -8.95317733e-01 -6.05466187e-01 7.29880154e-01
5.33197463e-01 -2.60450318e-02 -1.10311306e+00 1.89336404e-01
-2.40536466e-01 1.81185573e-01 -8.30458179e-02 -9.42709982e-01
-8.16289306e-01 4.85450923e-01 5.29044330e-01 5.27096570e-01
1.06114864e+00 8.63197595e-02 8.49367738e-01 -3.03687334e-01
2.81807512e-01 -6.70893729e-01 -1.71541259e-01 -7.04937875e-02
7.76970565e-01 -3.07681590e-01 4.33470190e-01 -2.94250816e-01
1.44053385e-01 3.62703353e-01 4.44138914e-01 -2.53208727e-01
9.48798001e-01 -7.12218881e-02 -3.92465591e-01 -1.77508771e-01
-1.32469046e+00 -2.04986036e-01 -2.02486083e-01 8.53064001e-01
4.25196558e-01 2.48161629e-01 -1.32008755e+00 4.75415796e-01
1.17035531e-01 -1.93174742e-02 9.44921553e-01 6.97152317e-01
-5.21203518e-01 -1.20125163e+00 -4.86871809e-01 9.06655192e-01
-6.96851909e-01 -3.10711339e-02 -5.54438412e-01 7.97654331e-01
3.67472112e-01 1.10906374e+00 2.69659907e-01 1.47727579e-01
-5.88869490e-02 1.83190092e-01 5.32319188e-01 -5.34955025e-01
-3.69244277e-01 -2.50613600e-01 4.58165914e-01 -1.73663184e-01
-3.70403826e-01 -1.19870937e+00 -1.04691136e+00 -8.75258803e-01
-2.42714986e-01 -4.46577579e-01 1.31091680e-02 8.60695243e-01
4.84128833e-01 5.42939126e-01 4.02750105e-01 -4.16670352e-01
-3.29324722e-01 -9.81130540e-01 -1.20134580e+00 -4.17450927e-02
6.27847731e-01 -7.68091142e-01 -8.33937526e-01 -2.63666641e-02]
|
[6.106392860412598, 4.45125150680542]
|
55379fc1-376e-4ab7-b683-588de41954e4
|
unlocking-the-potential-of-chatgpt-a
|
2304.02017
| null |
https://arxiv.org/abs/2304.02017v5
|
https://arxiv.org/pdf/2304.02017v5.pdf
|
Unlocking the Potential of ChatGPT: A Comprehensive Exploration of its Applications, Advantages, Limitations, and Future Directions in Natural Language Processing
|
Large language models have revolutionized the field of artificial intelligence and have been used in various applications. Among these models, ChatGPT (Chat Generative Pre-trained Transformer) has been developed by OpenAI, it stands out as a powerful tool that has been widely adopted. ChatGPT has been successfully applied in numerous areas, including chatbots, content generation, language translation, personalized recommendations, and even medical diagnosis and treatment. Its success in these applications can be attributed to its ability to generate human-like responses, understand natural language, and adapt to different contexts. Its versatility and accuracy make it a powerful tool for natural language processing (NLP). However, there are also limitations to ChatGPT, such as its tendency to produce biased responses and its potential to perpetuate harmful language patterns. This article provides a comprehensive overview of ChatGPT, its applications, advantages, and limitations. Additionally, the paper emphasizes the importance of ethical considerations when using this robust tool in real-world scenarios. Finally, This paper contributes to ongoing discussions surrounding artificial intelligence and its impact on vision and NLP domains by providing insights into prompt engineering techniques.
|
['Walid Hariri']
|
2023-03-27
| null | null | null | null |
['medical-diagnosis']
|
['medical']
|
[ 3.20333570e-01 2.75578052e-01 -8.26484412e-02 -2.33440712e-01
-5.49464166e-01 -5.99765301e-01 7.28635907e-01 -3.95924672e-02
-2.65151083e-01 9.20112491e-01 3.50773275e-01 -2.37578433e-02
7.55137429e-02 -7.44961202e-01 -1.52528495e-01 -6.43073618e-01
2.95872182e-01 6.93108499e-01 -6.82700351e-02 -4.41894293e-01
2.87391454e-01 1.03991263e-01 -1.38636780e+00 5.88933766e-01
8.84496152e-01 2.89826989e-01 4.85598236e-01 5.69171786e-01
-6.16013050e-01 1.13723171e+00 -1.05145478e+00 -6.26523912e-01
-3.05424750e-01 -8.36526632e-01 -9.99268115e-01 -1.11383326e-01
-3.53658229e-01 8.91136825e-02 1.76249057e-01 6.98153257e-01
7.06057310e-01 1.91681728e-01 4.24986005e-01 -1.16345811e+00
-1.01574659e+00 6.80268824e-01 -1.87024146e-01 -7.08779171e-02
6.69056654e-01 2.78756857e-01 7.79497087e-01 -6.03840947e-01
9.28074419e-01 1.47907996e+00 4.55596387e-01 9.49223459e-01
-9.37877655e-01 -4.03435290e-01 -7.77649805e-02 2.23386079e-01
-9.60271060e-01 -6.37922734e-02 6.33062422e-01 -3.28308493e-01
1.13087332e+00 4.42617148e-01 5.86500168e-01 1.57651830e+00
4.71689105e-01 1.02137887e+00 1.29014313e+00 -7.72827625e-01
2.23917052e-01 6.40812814e-01 -2.33244419e-01 2.56585509e-01
-3.83048683e-01 -1.21978834e-01 -5.53608716e-01 -2.27512524e-01
5.23909569e-01 -2.88638562e-01 -1.43200569e-02 3.47615898e-01
-1.18408203e+00 1.26848495e+00 3.57119203e-01 6.33418977e-01
-5.28338492e-01 -7.42765144e-02 4.09912109e-01 3.34236801e-01
7.00242519e-01 7.15379059e-01 -1.73336953e-01 -5.21357715e-01
-3.55547220e-01 4.68893230e-01 8.90150011e-01 7.41083741e-01
2.60788560e-01 1.21734388e-01 -3.75467628e-01 1.33830607e+00
2.16442958e-01 3.98602575e-01 7.41956830e-01 -9.79612172e-01
1.06857300e-01 5.47589242e-01 -1.75750226e-01 -1.01465058e+00
-2.94853091e-01 -5.48103824e-02 -7.58936405e-01 -1.04860798e-01
7.13060051e-02 -3.09301913e-01 -4.79828149e-01 1.49968767e+00
1.80901587e-01 -1.33513257e-01 2.27301225e-01 6.35993242e-01
1.13287401e+00 9.30486500e-01 2.42482156e-01 -1.89230025e-01
1.23721004e+00 -8.57712626e-01 -9.18554366e-01 -5.52408755e-01
4.63101715e-01 -1.18021131e+00 1.09538364e+00 3.02363813e-01
-8.31367910e-01 -2.24752709e-01 -2.90986627e-01 5.46753965e-02
-4.39646214e-01 -1.11865789e-01 7.80764639e-01 7.09655225e-01
-1.07903492e+00 2.01853320e-01 -2.90370703e-01 -9.25464809e-01
2.18062103e-01 9.33668166e-02 -4.48725633e-02 -7.32777938e-02
-1.35322583e+00 1.12146592e+00 1.82043850e-01 -1.08002447e-01
-1.03651538e-01 -3.65635872e-01 -5.63475311e-01 -2.94761270e-01
3.61326754e-01 -8.68122101e-01 1.47341061e+00 -1.07833207e+00
-2.01395369e+00 8.97346497e-01 -1.18556753e-01 -6.35642648e-01
3.95608366e-01 -1.55063212e-01 -2.25416631e-01 -3.07213347e-02
1.37007758e-01 8.86605024e-01 4.66666520e-01 -6.48718119e-01
-4.09768850e-01 2.00712718e-02 2.61253472e-02 2.70309329e-01
-1.22089215e-01 5.90200603e-01 -1.41118944e-01 -7.10011005e-01
-4.41137433e-01 -9.80081201e-01 -3.76465887e-01 -3.02270234e-01
-2.20787838e-01 -5.83182693e-01 7.06505060e-01 -4.77100104e-01
9.91293073e-01 -1.77208793e+00 -1.30660608e-01 -1.92165837e-01
5.63677028e-03 4.75696564e-01 -2.32936233e-01 1.08501637e+00
4.93750453e-01 2.69701779e-01 5.36569878e-02 8.15219209e-02
-1.81102186e-01 3.39021951e-01 -3.54391217e-01 -2.29522228e-01
4.37065184e-01 1.24056888e+00 -1.02519906e+00 -3.51042509e-01
4.66777533e-01 6.29716456e-01 -4.44781274e-01 1.96175352e-01
-5.96983314e-01 6.29806638e-01 -5.74569821e-01 2.41974875e-01
1.08845055e-01 -2.02794939e-01 2.33440280e-01 5.42721093e-01
-1.97196469e-01 3.88079286e-01 -5.59824049e-01 1.16646862e+00
-6.32307649e-01 9.26848352e-01 -3.06892335e-01 -7.71966517e-01
1.26688087e+00 5.33253133e-01 2.88468570e-01 -7.01894641e-01
1.94834843e-01 7.45458603e-02 1.04983591e-01 -9.12731588e-01
5.76054096e-01 -2.82873422e-01 -7.66475201e-02 8.20393801e-01
9.34307370e-03 -5.23192585e-01 2.57458448e-01 2.86475718e-01
7.04926848e-01 1.20801382e-01 4.80295867e-01 1.31547928e-01
4.70632464e-01 1.68347865e-01 2.26137489e-01 7.02419162e-01
1.78390611e-02 6.05798006e-01 3.83998424e-01 -4.48292762e-01
-7.93807328e-01 -4.11358535e-01 3.04920882e-01 1.11139762e+00
-1.97444484e-01 -3.10126811e-01 -7.87868321e-01 -3.31931859e-01
-5.67136645e-01 1.05085325e+00 -5.00707686e-01 -1.55309454e-01
-3.45294744e-01 -7.85424173e-01 6.44841909e-01 3.92149627e-01
3.88811380e-01 -1.77084887e+00 -6.51999831e-01 5.32431066e-01
-8.43334615e-01 -1.24018395e+00 -1.17296763e-01 -3.01755786e-01
-6.77356362e-01 -7.89936483e-01 -8.51008296e-01 -7.84241378e-01
4.28644627e-01 2.96955377e-01 1.10630882e+00 -4.20753621e-02
-2.79649377e-01 5.08856475e-01 -8.27573240e-01 -8.47945511e-01
-1.15017223e+00 1.48551419e-01 -2.42523000e-01 -4.92938459e-02
5.19721448e-01 -2.67389983e-01 -2.60848831e-02 2.55532980e-01
-8.54506373e-01 4.05911118e-01 5.58127820e-01 1.02127600e+00
8.47610831e-02 -2.77930528e-01 9.50233459e-01 -1.31423461e+00
1.40818250e+00 -4.90423650e-01 -9.03697312e-02 3.14923555e-01
-4.74656731e-01 -2.00345859e-01 4.46770430e-01 -5.74119985e-01
-1.24513638e+00 -3.91697198e-01 -3.47366869e-01 1.02514893e-01
-3.76806617e-01 6.69828534e-01 1.86644837e-01 8.34812298e-02
8.92953038e-01 1.76717386e-01 2.98543215e-01 -3.82050872e-01
4.50776517e-01 9.22355413e-01 8.91859084e-02 -3.84964526e-01
3.75186682e-01 6.04564212e-02 -4.62962478e-01 -1.25185347e+00
-5.25502086e-01 -1.92300647e-01 -2.06048131e-01 -3.96380395e-01
7.64541745e-01 -4.83487099e-01 -5.91547847e-01 4.39339966e-01
-1.41520607e+00 -3.33094865e-01 -2.01756001e-01 2.44436130e-01
-3.65781426e-01 1.77816197e-01 -6.83076143e-01 -1.02178097e+00
-7.61419296e-01 -9.11095202e-01 5.95420599e-01 5.96058667e-01
-8.73873830e-01 -1.26906657e+00 1.15797929e-01 7.10562110e-01
7.18036771e-01 2.75919825e-01 1.05108547e+00 -7.58961856e-01
-2.76853800e-01 -3.30466330e-01 1.13083050e-01 3.68435025e-01
1.17560498e-01 2.28446752e-01 -9.36442494e-01 2.15935767e-01
-1.35324925e-01 -5.82135081e-01 1.39553487e-01 2.72983849e-01
4.62738097e-01 -7.10588932e-01 -2.44447246e-01 -7.71603063e-02
9.76421237e-01 4.71538365e-01 5.89375615e-01 3.55279326e-01
2.42011726e-01 1.05216980e+00 6.39969945e-01 2.79405504e-01
2.85901159e-01 6.36614621e-01 7.10477978e-02 1.24754449e-02
-2.26603299e-02 -3.43432635e-01 4.59529519e-01 9.14610922e-01
-1.38204947e-01 -6.02421582e-01 -7.52890050e-01 3.11988294e-01
-1.90185297e+00 -1.13216996e+00 -2.54453272e-01 1.72048032e+00
7.26995230e-01 -2.06530273e-01 7.80438781e-02 -2.45811686e-01
6.04726672e-01 -1.30663496e-02 -2.25288510e-01 -1.02617359e+00
-3.58327329e-01 3.05108249e-01 -2.63974696e-01 4.22265470e-01
-5.87049961e-01 1.19036448e+00 7.20242310e+00 6.92025900e-01
-1.33622897e+00 2.10043877e-01 6.21338964e-01 1.93599179e-01
-4.02673841e-01 -1.28118873e-01 -4.14244741e-01 4.04679388e-01
9.08734560e-01 -5.10914862e-01 4.66883749e-01 8.16208124e-01
5.45878112e-01 -2.50956953e-01 -6.73348606e-01 7.75961101e-01
3.50090712e-01 -1.57695115e+00 1.63736373e-01 -1.46991074e-01
6.82003796e-01 9.38542113e-02 3.46234143e-02 3.60685855e-01
3.71426344e-01 -9.96765792e-01 4.61040497e-01 5.26189432e-02
5.01657665e-01 -6.39072299e-01 1.05317330e+00 6.55183852e-01
-4.92089450e-01 1.14854679e-01 -2.67424315e-01 -5.03855884e-01
3.78384292e-01 3.94227982e-01 -1.60591495e+00 4.49362516e-01
5.03393948e-01 4.25460488e-01 -2.53328383e-01 8.24684620e-01
-5.14054716e-01 7.13347673e-01 -2.12401420e-01 -7.89979219e-01
2.84240812e-01 -3.50527138e-01 5.59177935e-01 1.21693027e+00
2.01034248e-01 1.12586468e-01 2.46479124e-01 9.23188806e-01
2.81432122e-01 4.73508298e-01 -7.79407740e-01 -4.73545194e-01
4.58022416e-01 1.03409421e+00 -7.44310319e-01 -2.60821134e-01
-3.59234393e-01 8.94158542e-01 6.59670122e-03 2.98689604e-01
-6.34420335e-01 -3.22968572e-01 4.54899758e-01 -1.46843037e-02
-2.74890125e-01 2.29604855e-01 -4.17042553e-01 -7.87787974e-01
-3.16094279e-01 -1.26848114e+00 2.92986602e-01 -1.07061899e+00
-1.20607758e+00 1.05388176e+00 1.81397330e-02 -1.10355413e+00
-8.42838287e-01 -3.83824706e-01 -6.31045520e-01 8.39694083e-01
-8.45378339e-01 -1.31102574e+00 -3.03087868e-02 3.50190997e-01
1.08752477e+00 -2.15189010e-01 1.21085882e+00 2.23360653e-03
-3.59667957e-01 3.92467737e-01 -4.98979837e-02 -3.66423801e-02
7.44510651e-01 -7.13698328e-01 4.64966804e-01 6.73651993e-01
2.84619004e-01 6.95418715e-01 9.67972815e-01 -4.95569050e-01
-1.03919220e+00 -1.02163136e+00 1.42198682e+00 -4.81153518e-01
4.21042830e-01 -1.65295973e-01 -8.22006106e-01 6.29134238e-01
5.62713385e-01 -7.98351943e-01 8.66928756e-01 -2.16281787e-02
1.07959218e-01 1.52535707e-01 -1.13687098e+00 8.86490583e-01
4.64704633e-01 -2.71344632e-01 -5.00813186e-01 6.73611224e-01
5.15163541e-01 -4.77647543e-01 -3.98789763e-01 -6.56431764e-02
4.78038788e-01 -9.90579844e-01 5.99799633e-01 -4.24657375e-01
5.21481991e-01 2.05685183e-01 4.26105738e-01 -1.37833714e+00
-4.33814287e-01 -1.16870797e+00 3.32057774e-01 1.48910427e+00
3.95167172e-01 -9.94456589e-01 5.11639893e-01 7.61883438e-01
6.11597337e-02 -7.23891020e-01 -6.40320063e-01 -4.25410211e-01
-8.03178758e-04 -5.15265048e-01 3.66636485e-01 9.19942796e-01
2.62722999e-01 8.42581272e-01 -7.10227728e-01 -5.07024884e-01
3.48606966e-02 -2.65617631e-02 9.58280504e-01 -1.10979855e+00
-2.58518755e-01 -4.75546539e-01 -1.49079055e-01 -7.28725553e-01
-1.14292406e-01 -7.29762971e-01 1.23093180e-01 -1.77914214e+00
9.65016112e-02 -2.91330397e-01 3.13619882e-01 4.19279039e-01
-1.31848156e-01 3.17738891e-01 3.94091666e-01 1.54832155e-01
-1.11229680e-01 2.01399073e-01 1.24610150e+00 -1.13556042e-01
-4.17641342e-01 2.86372632e-01 -9.48514223e-01 6.59909070e-01
1.12369549e+00 -5.80938697e-01 -5.37179172e-01 -3.05727065e-01
2.96266168e-01 -2.67792016e-01 8.46956223e-02 -4.43608642e-01
-5.96205294e-02 -5.35306275e-01 -7.22354501e-02 -1.17954463e-01
3.24502409e-01 -3.60674471e-01 3.85820061e-01 4.42706764e-01
-5.63107729e-01 1.42426461e-01 1.17935024e-01 2.73325711e-01
-2.93023735e-01 -3.56438130e-01 6.49996579e-01 -5.16395748e-01
-7.64260590e-01 -1.78553015e-01 -1.23135006e+00 6.19503893e-02
1.20723927e+00 -3.11937213e-01 -2.36532703e-01 -8.61277163e-01
-2.33670428e-01 1.25096038e-01 2.25343734e-01 1.03801203e+00
6.57331228e-01 -9.32013929e-01 -7.88851023e-01 -1.07440520e-02
1.64887816e-01 -3.01366746e-01 4.02727634e-01 8.80574703e-01
-6.46155655e-01 4.93129700e-01 -1.18713081e-01 -5.69631934e-01
-1.56875873e+00 3.76375943e-01 -9.16940272e-02 -4.12500650e-01
-8.29994082e-01 8.92407298e-01 9.70911682e-02 -4.15131152e-01
1.09184913e-01 -6.09329902e-02 -5.42123199e-01 -1.82335332e-01
7.42623150e-01 1.78544834e-01 -1.58643290e-01 -5.56059957e-01
-2.61358261e-01 2.81919688e-01 -1.38220400e-01 -3.70451152e-01
1.18930519e+00 -1.45774618e-01 -2.54695982e-01 6.07837558e-01
5.33143163e-01 -1.73085779e-01 -4.40604001e-01 -8.53576809e-02
-5.04712090e-02 -2.45266527e-01 -4.39973772e-01 -1.13265336e+00
-6.01740599e-01 9.84360874e-01 1.64283097e-01 3.86925012e-01
8.87305498e-01 2.96431370e-02 9.57562327e-01 3.30076993e-01
4.67869580e-01 -1.12464035e+00 2.37051770e-01 7.46218204e-01
1.06252289e+00 -9.70535398e-01 -3.68574739e-01 -3.18680704e-01
-1.33033621e+00 1.13743281e+00 4.54775512e-01 2.51956046e-01
4.59940918e-02 2.27678776e-01 8.04776311e-01 1.28201514e-01
-1.08318710e+00 1.57124847e-01 -2.46754885e-02 1.13469112e+00
8.03632796e-01 1.54523194e-01 -7.08563328e-01 3.02066803e-01
-4.10641044e-01 2.95245916e-01 7.07079768e-01 8.86933446e-01
-3.30229074e-01 -1.49027014e+00 -4.63059366e-01 2.60987818e-01
-4.81677115e-01 4.94991504e-02 -8.85293365e-01 5.79711080e-01
-7.40967039e-03 1.29238760e+00 -3.51630300e-01 -4.35972542e-01
5.51321432e-02 2.18547821e-01 2.34200507e-01 -8.49170804e-01
-1.04235423e+00 -1.08915837e-02 3.30615222e-01 -1.85447305e-01
-5.50589919e-01 -4.27485287e-01 -1.01480198e+00 -3.73776495e-01
-3.56383950e-01 4.66037631e-01 5.51821113e-01 1.05343270e+00
5.93683302e-01 4.06483769e-01 4.17118847e-01 -5.63021958e-01
-2.69413888e-01 -1.24501646e+00 -7.63272569e-02 2.78190494e-01
-2.74216741e-01 -1.94663972e-01 1.19076945e-01 2.30169982e-01]
|
[12.204853057861328, 8.49516487121582]
|
f674bbf9-30f9-4abb-b577-7a7ccae3b9bc
|
athena-2-0-contextualized-dialogue-management-1
|
2111.02519
| null |
https://arxiv.org/abs/2111.02519v1
|
https://arxiv.org/pdf/2111.02519v1.pdf
|
Athena 2.0: Contextualized Dialogue Management for an Alexa Prize SocialBot
|
Athena 2.0 is an Alexa Prize SocialBot that has been a finalist in the last two Alexa Prize Grand Challenges. One reason for Athena's success is its novel dialogue management strategy, which allows it to dynamically construct dialogues and responses from component modules, leading to novel conversations with every interaction. Here we describe Athena's system design and performance in the Alexa Prize during the 20/21 competition. A live demo of Athena as well as video recordings will provoke discussion on the state of the art in conversational AI.
|
['Marilyn Walker', 'Adwait Ratnaparkhi', 'Rohan Pandey', 'Jeshwanth Bheemanpally', 'Phillip Lee', 'Eduardo Zamora', 'Cecilia Li', 'Angela Ramirez', 'Rishi Rajasekaran', 'Omkar Patil', 'Wen Cui', 'Vrindavan Harrison', 'Lena Reed', 'Kevin K. Bowden', 'Juraj Juraska']
|
2021-11-03
|
athena-2-0-contextualized-dialogue-management
|
https://aclanthology.org/2021.emnlp-demo.15
|
https://aclanthology.org/2021.emnlp-demo.15.pdf
|
emnlp-acl-2021-11
|
['dialogue-management']
|
['natural-language-processing']
|
[-3.56656194e-01 7.87067413e-01 3.85843217e-01 -4.67821300e-01
-3.46405566e-01 -7.70616591e-01 1.09207714e+00 -4.78694402e-02
-1.26704067e-01 1.14007521e+00 7.36561716e-01 -6.72825798e-02
-1.89891160e-02 -5.38829029e-01 6.34652898e-02 3.24624814e-02
-1.94759727e-01 9.21005368e-01 2.93795526e-01 -1.34475815e+00
4.92882103e-01 7.24238008e-02 -1.47771025e+00 6.18074000e-01
4.83872861e-01 2.88544267e-01 8.21785331e-02 1.40855920e+00
-3.63020450e-01 1.08589482e+00 -9.54648316e-01 -6.39185071e-01
1.56835258e-01 -9.03555155e-01 -1.71119690e+00 -3.42499048e-01
-1.34121999e-02 -2.06387937e-01 -4.01223242e-01 4.13995177e-01
6.59304440e-01 4.62139785e-01 5.40470667e-02 -1.52236474e+00
-1.23534612e-01 1.16514897e+00 3.70608002e-01 1.99713692e-01
9.54073071e-01 3.21819335e-01 1.39745212e+00 -4.09286380e-01
9.93444204e-01 1.40968657e+00 5.66922843e-01 1.19200158e+00
-8.03063631e-01 -3.30080807e-01 -2.75685817e-01 3.00007403e-01
-7.06217170e-01 -7.29518831e-01 7.63294578e-01 -2.84811378e-01
1.07084477e+00 6.09025121e-01 1.21680737e+00 1.10366344e+00
-7.02075437e-02 1.14402092e+00 8.47924173e-01 -5.21902800e-01
1.61371246e-01 2.89911240e-01 1.16804682e-01 4.93888080e-01
-4.78792906e-01 -5.24922669e-01 -7.07897127e-01 -6.23589635e-01
5.90678155e-01 -7.07858562e-01 1.23542212e-02 -1.14244960e-01
-1.20554042e+00 9.87883627e-01 3.65559161e-02 6.15585506e-01
-2.18555942e-01 8.97453055e-02 7.76055038e-01 7.54245400e-01
2.39995733e-01 1.08801305e+00 -1.74720317e-01 -1.12302554e+00
1.28614187e-01 7.64063537e-01 1.62822306e+00 6.91562593e-01
3.92724544e-01 -2.62576282e-01 -1.27209023e-01 1.16674697e+00
3.33486915e-01 -2.46012360e-01 4.02788788e-01 -1.62204885e+00
5.62295020e-02 7.79031157e-01 4.43890721e-01 -8.14296126e-01
-5.77957869e-01 3.21384311e-01 -5.10622151e-02 -6.23628078e-03
7.37944663e-01 -7.46938586e-01 1.21502787e-01 1.57666707e+00
2.87683517e-01 -1.54845387e-01 2.27760568e-01 7.24559188e-01
1.37655175e+00 5.54633260e-01 -2.26724461e-01 -4.92238887e-02
1.21807933e+00 -9.97392714e-01 -7.26242959e-01 -2.05097795e-01
8.16590428e-01 -1.10937774e+00 9.67983961e-01 2.73586631e-01
-1.60412514e+00 -1.89957365e-01 -8.75406444e-01 1.49426103e-01
-2.30247304e-01 -5.47362208e-01 8.61714602e-01 7.76839018e-01
-1.11685991e+00 5.33738315e-01 -2.95928478e-01 -1.05542648e+00
-3.44316989e-01 3.79199088e-01 -4.61611897e-02 6.04799926e-01
-1.70358694e+00 1.24299395e+00 6.25931844e-02 -2.50434667e-01
-3.17766488e-01 -3.26321661e-01 -3.57763916e-01 -2.70005405e-01
3.82915407e-01 -4.63784933e-01 2.03560710e+00 -6.06164515e-01
-2.47515559e+00 1.22099447e+00 5.06811857e-01 -4.49691027e-01
6.19334936e-01 -1.01338953e-01 -3.69379640e-01 -1.88388862e-02
-5.40602803e-02 5.89951217e-01 -1.43081725e-01 -6.96807802e-01
-7.85353303e-01 1.56013921e-01 8.16828966e-01 5.59478760e-01
-6.10653125e-02 3.42793614e-01 -1.41692758e-01 1.89461131e-02
-5.05319595e-01 -9.69011784e-01 -2.75144398e-01 -7.26278186e-01
-9.13172066e-02 -1.00104702e+00 5.46535015e-01 -7.28977621e-02
8.31676662e-01 -1.77211642e+00 3.19318771e-01 -2.45717004e-01
5.18634319e-01 4.16917801e-01 -2.63270773e-02 1.19518542e+00
3.60407144e-01 -4.59994785e-02 3.91767532e-01 -1.31354749e-01
3.83284450e-01 1.19333237e-01 2.72483323e-02 1.91433374e-02
-3.51132184e-01 8.47762108e-01 -1.26915491e+00 -3.05404574e-01
3.92171293e-01 -1.83908474e-02 -6.40906274e-01 4.26788867e-01
-5.27993858e-01 5.97135603e-01 -5.16788602e-01 -1.47682458e-01
2.90794447e-02 -3.38729948e-01 5.08297861e-01 3.75994235e-01
-5.07278979e-01 7.54199922e-01 -6.50851488e-01 1.84780586e+00
-2.83534408e-01 9.15807366e-01 5.28152704e-01 -5.80001175e-01
1.04932213e+00 6.54885650e-01 5.74128509e-01 -4.06694651e-01
5.50786972e-01 2.73915797e-01 3.45248580e-01 -5.96003175e-01
1.01975822e+00 1.29086897e-01 -4.79967296e-01 1.08079791e+00
4.51992042e-02 -7.43583918e-01 3.59541655e-01 7.02233553e-01
1.19713151e+00 -1.30814031e-01 2.87013203e-01 -4.81391251e-01
6.21098459e-01 3.39231312e-01 1.94119230e-01 1.03207982e+00
-8.45113873e-01 2.70556927e-01 5.44306993e-01 -8.93732488e-01
-1.10218680e+00 -7.32595503e-01 3.37326229e-01 1.35171807e+00
1.27194151e-01 -1.08589768e+00 -8.03078890e-01 -5.93734622e-01
-4.23989326e-01 5.87289929e-01 -5.07926881e-01 3.02259121e-02
-9.23734307e-01 -9.26936939e-02 7.99596906e-01 -2.71384627e-01
7.85404861e-01 -1.65539515e+00 -6.67432487e-01 4.89325017e-01
-8.03856671e-01 -9.71452534e-01 -2.17754453e-01 -2.70765930e-01
-1.00332066e-01 -1.12388134e+00 -3.71600509e-01 -6.69080913e-01
-3.39042753e-01 2.18458310e-01 1.23984492e+00 1.80191964e-01
-3.47568631e-01 7.28505671e-01 -6.05712354e-01 -3.87230396e-01
-1.03384662e+00 3.78543526e-01 -7.46184215e-02 -2.30737150e-01
3.35357279e-01 -5.84300816e-01 -4.03350890e-01 4.47591215e-01
-2.89089411e-01 2.82875389e-01 -1.98500320e-01 6.85897708e-01
-8.99336278e-01 -8.29503953e-01 8.28868270e-01 -1.16255403e+00
1.33334374e+00 -4.04893011e-01 -7.96215460e-02 2.50340346e-03
-6.82866052e-02 -5.13283491e-01 3.27056646e-01 5.68983853e-02
-1.09058666e+00 -2.16838270e-01 -9.68953595e-03 5.57042718e-01
1.68646425e-02 1.48255512e-01 3.82875681e-01 2.26181801e-02
1.04472399e+00 -2.06341550e-01 2.70620435e-01 -8.85148272e-02
4.89126652e-01 1.00920665e+00 3.22741270e-01 -7.61110246e-01
1.94562525e-01 1.13918401e-01 -4.53916460e-01 -1.21677089e+00
-6.78606689e-01 -3.73959720e-01 -2.68444926e-01 -8.26476634e-01
7.47255802e-01 -4.98220325e-01 -1.61621249e+00 4.81130868e-01
-1.49609828e+00 -5.90420902e-01 -2.79535860e-01 2.09332064e-01
-7.12198853e-01 1.31634191e-01 -8.06321502e-01 -9.29337800e-01
-3.47597629e-01 -7.94583619e-01 2.29342073e-01 6.93438590e-01
-1.17107999e+00 -1.08478522e+00 5.12924135e-01 8.80483747e-01
8.53405178e-01 -1.44843180e-02 3.43140364e-01 -9.86359596e-01
-1.91139236e-01 -2.89561838e-01 2.73379236e-02 -2.41862312e-02
9.12242848e-03 1.06300257e-01 -8.40800762e-01 2.13474371e-02
-2.35031858e-01 -7.35819757e-01 -5.17055951e-03 -1.77046418e-01
5.83924279e-02 -1.71984091e-01 -1.50671035e-01 -5.36499977e-01
3.45303863e-01 3.68080407e-01 5.67174971e-01 3.67180377e-01
1.53582886e-01 9.35683668e-01 3.56616676e-01 7.32307434e-01
6.73992574e-01 9.54677403e-01 4.63435762e-02 6.12308323e-01
-3.39725614e-02 5.18656746e-02 3.48255426e-01 1.15343142e+00
-1.60027251e-01 -1.16362564e-01 -9.43058550e-01 5.50905645e-01
-2.49246597e+00 -1.07021308e+00 -2.86636293e-01 1.62628353e+00
7.82166958e-01 1.15044869e-01 6.45245492e-01 -3.06019574e-01
8.89794290e-01 3.76744598e-01 -2.29417402e-02 -1.03005517e+00
-2.90330928e-02 1.70541823e-01 -4.00296539e-01 9.04172182e-01
-7.55720317e-01 1.21613300e+00 7.51282310e+00 3.98730904e-01
-7.08287537e-01 2.43869081e-01 6.89317733e-02 -1.74310654e-01
-4.62079830e-02 2.16046825e-01 -5.80889702e-01 1.00059502e-01
9.67952549e-01 -1.03786254e+00 7.01810420e-01 6.05531275e-01
1.23574333e-02 -8.62507895e-02 -1.00763011e+00 6.35295749e-01
1.75218612e-01 -1.84393418e+00 -4.37260598e-01 -1.61341280e-01
4.88606602e-01 3.61636370e-01 -6.09640181e-01 5.94735801e-01
1.15491450e+00 -9.27400112e-01 1.90595120e-01 1.20490104e-01
-1.12663418e-01 -5.86923242e-01 4.76988763e-01 3.06509316e-01
-6.14603937e-01 -5.80651984e-02 -1.79518182e-02 -7.74151742e-01
3.36243540e-01 -2.39683792e-01 -1.25871599e+00 1.60504922e-01
5.64029336e-01 3.98302078e-01 1.09850042e-01 8.21211040e-01
-7.76881203e-02 3.85603249e-01 -3.89133930e-01 -9.14489627e-01
3.84515733e-01 -1.44363299e-01 1.11685324e+00 1.09758902e+00
-5.07267356e-01 6.41759038e-01 4.69617546e-01 4.79948878e-01
-1.68416709e-01 2.70320565e-01 -6.32457078e-01 -7.07027689e-02
6.88851058e-01 1.45886707e+00 -6.09795153e-01 -3.17506850e-01
-3.81204933e-02 6.64084315e-01 2.46295795e-01 -2.61204600e-01
-4.52761501e-01 -5.13928056e-01 7.44503736e-01 -1.07820824e-01
-3.68410200e-01 -2.87569910e-01 -1.14130117e-01 -8.46184969e-01
-4.70495313e-01 -1.08875906e+00 3.53268743e-01 -6.87262237e-01
-1.13078499e+00 9.45617199e-01 -1.33613840e-01 -6.66175067e-01
-6.26584649e-01 5.29786898e-03 -1.04222012e+00 4.22832429e-01
-4.13537264e-01 -9.42235708e-01 -2.68338829e-01 3.16393942e-01
8.92858624e-01 -4.54248995e-01 1.28097832e+00 6.22948632e-02
-3.76094759e-01 3.74430537e-01 -2.90892780e-01 -1.03393756e-02
9.55353379e-01 -1.16672504e+00 8.21146190e-01 -1.20349452e-01
-1.52606994e-01 3.79601687e-01 1.20954561e+00 -2.74353474e-01
-1.44350255e+00 -1.09926816e-02 1.25285435e+00 -7.67203331e-01
1.02158618e+00 -4.07148451e-01 -3.75018269e-01 6.92651093e-01
7.36022532e-01 -7.32939363e-01 6.52007401e-01 3.60581517e-01
4.53004763e-02 3.21253926e-01 -1.06332135e+00 1.07921124e+00
9.96379972e-01 -4.23600167e-01 -7.72698224e-01 6.95340872e-01
7.15494514e-01 -6.90996289e-01 -6.16783261e-01 7.18052313e-02
8.11401486e-01 -1.13085091e+00 5.82376719e-01 -8.72687340e-01
5.27741909e-01 2.24587053e-01 -8.98858979e-02 -1.44433594e+00
-1.46037430e-01 -1.84521735e+00 4.21965092e-01 1.24747181e+00
3.01891446e-01 -5.37702322e-01 1.02757227e+00 1.09990287e+00
-1.53689831e-01 -1.48572624e-01 -8.17740977e-01 -4.13745910e-01
3.15661393e-02 -1.61564037e-01 2.83453763e-01 1.18504846e+00
1.14891994e+00 1.01622653e+00 -7.18623221e-01 -7.61806011e-01
1.95788398e-01 -1.47845715e-01 1.51961148e+00 -1.52097273e+00
-4.75447148e-01 -7.51143456e-01 -2.73571223e-01 -8.93455982e-01
-4.91252951e-02 -6.79230332e-01 1.23062637e-02 -1.33064961e+00
-2.07795948e-01 -1.68826401e-01 3.70712399e-01 2.17073202e-01
1.88212648e-01 3.55983615e-01 5.76939642e-01 2.49550328e-01
-9.53395903e-01 5.00022471e-01 1.29375589e+00 1.76044524e-01
-6.97275639e-01 1.49338946e-01 -8.71017992e-01 7.10887432e-01
1.05520201e+00 2.35816874e-02 -1.57035694e-01 3.52579821e-03
3.84367883e-01 3.11125159e-01 -6.75830096e-02 -7.81029105e-01
5.13734400e-01 -2.97079861e-01 -4.12319660e-01 -2.06690636e-02
9.15250003e-01 -9.72528607e-02 9.66258273e-02 3.03261191e-01
-7.56855071e-01 -2.41639182e-01 2.66997099e-01 6.23974800e-02
3.43182683e-02 -1.93343684e-01 6.60773575e-01 -3.91240299e-01
-5.17448425e-01 -4.11017567e-01 -1.06657588e+00 3.61478984e-01
1.21322322e+00 -1.30456686e-01 -7.15819418e-01 -1.08412051e+00
-9.13838983e-01 6.35576725e-01 1.53163582e-01 7.99614072e-01
8.89892057e-02 -9.58585262e-01 -1.09627128e+00 -2.51678169e-01
4.21775766e-02 -5.10908902e-01 2.50811487e-01 4.15433586e-01
-5.81729352e-01 3.33443880e-01 -5.89243412e-01 -3.65605146e-01
-1.67268384e+00 -4.62603211e-01 3.43227357e-01 -3.27720553e-01
-8.65860045e-01 9.32859063e-01 -2.59291440e-01 -9.03332889e-01
8.41796547e-02 7.73549616e-01 -4.82038498e-01 -5.02929352e-02
7.19730794e-01 5.36586881e-01 -3.35729897e-01 -1.85258955e-01
-2.73987383e-01 -6.16144016e-02 -3.27234298e-01 -7.57435918e-01
1.35958827e+00 -4.15450275e-01 -3.19845647e-01 9.15435255e-01
4.57358450e-01 -6.66300431e-02 -4.95353520e-01 -2.34797359e-01
6.08211495e-02 -3.38054180e-01 -5.29734492e-01 -1.03507710e+00
-1.71432883e-01 4.68357295e-01 -1.39180705e-01 1.03416777e+00
1.33671671e-01 3.66265909e-03 1.12115061e+00 8.34723771e-01
3.64817023e-01 -1.37335432e+00 3.42338085e-01 1.29058492e+00
1.24029148e+00 -8.53002608e-01 -3.08786631e-01 -3.91527951e-01
-9.49948549e-01 1.21830559e+00 8.54149103e-01 -2.12999985e-01
4.26106334e-01 -1.24331946e-02 4.14707005e-01 -5.83974957e-01
-1.35598516e+00 1.27116248e-01 -5.21447003e-01 4.21492368e-01
9.35896218e-01 2.54433841e-01 -7.66356945e-01 4.42545593e-01
-6.23003662e-01 -9.70491022e-02 1.16486371e+00 1.14987421e+00
-7.35421360e-01 -1.58909786e+00 -2.74784770e-02 8.37933645e-02
-2.48251602e-01 4.12501454e-01 -1.49466825e+00 6.24376237e-01
-5.67914367e-01 1.37257016e+00 -1.10269554e-01 -5.01275599e-01
4.38822955e-01 3.89422268e-01 5.43387771e-01 -8.64977062e-01
-1.46470058e+00 -4.67514783e-01 1.14778554e+00 -4.10740972e-01
-1.95058405e-01 -6.45746291e-01 -1.34748197e+00 -1.21800911e+00
-1.54131442e-01 1.12963438e+00 6.53113663e-01 9.37892914e-01
5.13408303e-01 2.59467214e-01 8.55645180e-01 -8.84638608e-01
-3.67801368e-01 -1.29723024e+00 -2.58980811e-01 3.53473008e-01
-2.18859375e-01 -2.11763486e-01 -3.14634293e-01 -4.63985711e-01]
|
[12.748807907104492, 7.945132255554199]
|
b3b8d60c-682a-4d9a-908a-e32542783668
|
audio-declipping-performance-enhancement-via
|
2104.03074
| null |
https://arxiv.org/abs/2104.03074v1
|
https://arxiv.org/pdf/2104.03074v1.pdf
|
Audio declipping performance enhancement via crossfading
|
Some audio declipping methods produce waveforms that do not fully respect the physical process of clipping, which is why we refer to them as inconsistent. This letter reports what effect on perception it has if the solution by inconsistent methods is forced consistent by postprocessing. We first propose a simple sample replacement method, then we identify its main weaknesses and propose an improved variant. The experiments show that the vast majority of inconsistent declipping methods significantly benefit from the proposed approach in terms of objective perceptual metrics. In particular, we show that the SS PEW method based on social sparsity combined with the proposed method performs comparable to top methods from the consistent class, but at a computational cost of one order of magnitude lower.
|
['Ondřej Mokrý', 'Pavel Rajmic', 'Pavel Záviška']
|
2021-04-07
| null | null | null | null |
['audio-declipping']
|
['audio']
|
[ 3.98530900e-01 -3.61519158e-02 6.85196891e-02 -4.43475172e-02
-9.43179607e-01 -6.04950070e-01 3.50176871e-01 -1.12330779e-01
-5.72203100e-02 8.29500318e-01 4.78551328e-01 1.88138857e-01
-6.43810749e-01 -3.78457844e-01 -6.89243615e-01 -7.04027772e-01
-1.73769072e-01 -3.49241555e-01 3.86626452e-01 -3.35259348e-01
5.39174557e-01 1.45140976e-01 -2.01609874e+00 3.75106752e-01
8.65179718e-01 1.20102191e+00 -1.74689487e-01 5.26205480e-01
1.25688285e-01 4.68700111e-01 -7.87212789e-01 -4.94339854e-01
3.55225354e-01 -6.61181211e-01 -3.16957951e-01 1.33346811e-01
5.87680936e-01 4.46357280e-02 1.73003644e-01 1.33405459e+00
7.07860768e-01 -8.23295023e-03 3.95835400e-01 -1.24353421e+00
-4.96651351e-01 9.55278337e-01 -5.35588801e-01 -1.19425757e-02
6.74179196e-01 -3.77467334e-01 1.12292886e+00 -1.12299037e+00
4.80848372e-01 9.10694301e-01 1.09920108e+00 -6.44660443e-02
-1.41728330e+00 -8.20492744e-01 2.85470426e-01 3.96472424e-01
-1.74518120e+00 -7.85736680e-01 1.18021715e+00 -7.25703686e-02
3.85161310e-01 7.18067288e-01 4.04944688e-01 1.02443027e+00
6.50446340e-02 1.96584687e-01 1.39127862e+00 -5.66138506e-01
4.12090898e-01 8.37979373e-03 4.09300297e-01 2.25216836e-01
5.49882710e-01 -3.69316749e-02 -1.00075209e+00 -4.91362542e-01
1.50611382e-02 -5.57301104e-01 -4.99842495e-01 -9.87668261e-02
-9.94957864e-01 5.78633964e-01 -3.38789165e-01 3.97889167e-01
-2.05731943e-01 -1.10043883e-01 2.75662899e-01 6.36559546e-01
5.95132291e-01 6.09286487e-01 -7.44329393e-02 -1.23090565e-01
-1.20680535e+00 4.49659407e-01 7.32423306e-01 8.54755819e-01
3.80644083e-01 3.41804206e-01 5.99005669e-02 9.99108255e-01
-1.51895045e-03 5.47023237e-01 3.23531032e-01 -1.16454995e+00
2.40874752e-01 -1.14055008e-01 3.95857155e-01 -1.41455257e+00
-4.33776915e-01 -7.34420240e-01 -6.85992599e-01 1.72493577e-01
1.45088255e-01 -6.07976271e-03 -2.23353714e-01 1.82273662e+00
-6.97820038e-02 5.91821790e-01 -9.73198116e-02 8.40553582e-01
7.25064456e-01 6.81554079e-01 -3.75744641e-01 -7.57250965e-01
1.13328445e+00 -6.31618738e-01 -1.34789944e+00 3.07878166e-01
-7.73700774e-02 -1.36326170e+00 8.75207961e-01 1.27083302e+00
-1.50512516e+00 -8.20493400e-01 -1.67186463e+00 5.07555425e-01
2.10133605e-02 7.58462027e-02 3.75030726e-01 1.11189067e+00
-1.18840408e+00 7.90157616e-01 -1.19338892e-01 -1.97134584e-01
-6.43598586e-02 1.61224827e-01 -1.28224358e-01 3.58519495e-01
-1.08141327e+00 6.03206456e-01 -8.82909149e-02 4.38103378e-02
-2.55755663e-01 -6.94452941e-01 -5.24296463e-01 1.99710742e-01
4.21337426e-01 -3.67611080e-01 1.06514585e+00 -8.30860734e-01
-1.78744221e+00 3.03837270e-01 -3.53159130e-01 -5.69859326e-01
4.96256411e-01 -5.49578726e-01 -9.67684686e-01 3.27817440e-01
-2.63064295e-01 7.60529935e-02 1.59646416e+00 -1.41362906e+00
-4.66701329e-01 1.85033172e-01 -3.05525333e-01 -2.09442362e-01
-4.68800277e-01 8.41942057e-02 -7.95286223e-02 -1.07742810e+00
5.65909505e-01 -9.43150103e-01 1.34412393e-01 -2.29061365e-01
-5.87658763e-01 2.14941174e-01 8.66795957e-01 -5.80144465e-01
1.73030376e+00 -2.33795595e+00 -1.77279636e-02 4.99836981e-01
2.22241521e-01 1.98667660e-01 -1.15114793e-01 9.14993405e-01
-3.06194723e-01 1.47161826e-01 -1.99241668e-01 -5.25823593e-01
4.77304533e-02 -8.97020698e-02 -8.24656904e-01 6.63657904e-01
6.23029210e-02 -2.46156231e-02 -6.01594329e-01 -3.87482822e-01
7.24209622e-02 3.80052775e-01 -7.49212861e-01 9.70057771e-03
2.82492965e-01 2.16959998e-01 2.08691612e-01 5.53518653e-01
1.17569101e+00 1.88926324e-01 2.81480342e-01 -6.55645967e-01
-5.74276328e-01 2.51896679e-01 -1.66442585e+00 1.42674196e+00
-9.74983796e-02 6.78953588e-01 2.23250389e-01 -1.02082205e+00
1.07461226e+00 4.25553113e-01 3.76918644e-01 -4.71757054e-01
-1.08070008e-01 3.76828760e-01 -6.84394315e-03 -5.28478980e-01
7.39663780e-01 -1.48975521e-01 4.05552387e-02 2.04719916e-01
3.19387540e-02 -1.05248123e-01 3.08505856e-02 2.17966884e-01
9.09454584e-01 -1.99367091e-01 2.40366057e-01 -5.52279830e-01
5.70214212e-01 -6.04916096e-01 6.56225741e-01 1.13326597e+00
-4.14685905e-01 9.35128629e-01 5.62303662e-01 7.59300664e-02
-9.67805564e-01 -1.20062947e+00 -2.17800379e-01 7.89215207e-01
3.44101399e-01 -7.23298490e-01 -4.79097068e-01 -1.39309630e-01
-9.00850073e-02 5.56875467e-01 -3.39219809e-01 -1.29001170e-01
-3.89277458e-01 -5.32488406e-01 6.87314332e-01 4.28694226e-02
3.20214450e-01 -3.91290307e-01 -3.26117605e-01 1.44549161e-01
-4.19125229e-01 -1.18989229e+00 -3.12432885e-01 -5.92374615e-02
-6.42717183e-01 -7.14354038e-01 -6.15256011e-01 -5.17837226e-01
2.11937621e-01 5.49680710e-01 6.81061745e-01 6.18151724e-02
2.60682076e-01 2.62765557e-01 -8.30438495e-01 -3.88429940e-01
-2.74630398e-01 -1.92855462e-01 4.31829423e-01 7.02722549e-01
-1.52700171e-01 -1.11471260e+00 -4.64577198e-01 2.41008222e-01
-7.85557747e-01 -2.12426022e-01 3.59257162e-01 6.22738600e-01
1.98794544e-01 3.26259881e-01 1.07239377e+00 -7.51697659e-01
8.37819636e-01 -3.12209308e-01 -3.12529206e-01 -1.90352961e-01
-6.93661332e-01 -2.00975925e-01 7.41059601e-01 -3.86505574e-01
-1.09140754e+00 -1.50597885e-01 -2.60835737e-01 -1.65840507e-01
1.59448400e-01 3.76743138e-01 -4.39701267e-02 -2.65356511e-01
5.42886913e-01 -3.57782021e-02 -1.59372553e-01 -6.21715367e-01
1.58668041e-01 4.84129548e-01 6.64769053e-01 -4.21299070e-01
1.03450334e+00 6.44647837e-01 -1.15903094e-02 -1.08439493e+00
-6.68659985e-01 -2.85234451e-01 -8.90494138e-02 -5.02711117e-01
3.26236486e-01 -6.18861437e-01 -5.26237130e-01 5.26957214e-01
-1.10338008e+00 3.70113134e-01 -2.21713141e-01 6.52666628e-01
-4.34504688e-01 6.64574623e-01 -4.51355636e-01 -9.69934106e-01
-4.10400890e-03 -6.93487883e-01 6.55319333e-01 -2.31211513e-01
-4.73875821e-01 -3.93214673e-01 1.26557559e-01 -3.17407101e-02
6.56648159e-01 1.26541212e-01 8.67375314e-01 -4.93607640e-01
-1.82342678e-01 2.49486435e-02 -2.71614082e-02 4.65786994e-01
5.18182740e-02 2.41849914e-01 -1.26928806e+00 -2.80435950e-01
3.46915662e-01 1.69382766e-01 6.91311002e-01 5.72612107e-01
1.02289176e+00 -2.96636134e-01 -1.96241979e-02 4.69970584e-01
1.47868872e+00 5.21627031e-02 7.80001640e-01 8.29878170e-03
-2.33023204e-02 7.16235220e-01 7.18740284e-01 7.19317198e-01
-3.02476108e-01 8.45988631e-01 3.52169305e-01 1.48907453e-02
-2.35414162e-01 -2.31655985e-01 5.25136232e-01 1.47904336e+00
-7.02335984e-02 -7.43172392e-02 -1.88094690e-01 4.82122004e-01
-1.79128385e+00 -1.29918373e+00 -2.96535194e-01 2.33878660e+00
7.25952923e-01 3.21112275e-01 1.54192477e-01 1.01228654e+00
9.36675370e-01 3.27495307e-01 2.39228249e-01 -6.58994615e-01
-4.20460701e-01 5.03671229e-01 2.76537448e-01 7.14773297e-01
-8.39144707e-01 2.50856668e-01 7.38786221e+00 9.18490350e-01
-1.00012338e+00 1.08976230e-01 -7.83859100e-03 -4.97187451e-02
-3.16730440e-01 7.96496049e-02 -5.03699064e-01 6.49680614e-01
7.93798149e-01 -2.45945811e-01 1.45396188e-01 3.69655788e-01
5.79572082e-01 -9.97659266e-02 -8.38058352e-01 1.11243629e+00
3.28314155e-01 -1.21746433e+00 -1.76777586e-01 -2.74575740e-01
6.26501679e-01 -5.53696573e-01 4.09717560e-01 -2.88854353e-02
-5.46776116e-01 -6.41992807e-01 1.10301697e+00 5.71226835e-01
3.14321369e-01 -7.78653443e-01 6.16997838e-01 8.78872443e-03
-1.12107968e+00 -1.44578710e-01 -2.08200783e-01 -3.81597131e-01
3.38856190e-01 1.04303348e+00 -3.22166055e-01 7.07580984e-01
7.38505542e-01 4.49158370e-01 -5.19528031e-01 1.37288201e+00
9.22669657e-03 1.11289799e+00 -1.75771087e-01 1.68894708e-01
-1.46374390e-01 -1.99158520e-01 1.06444860e+00 1.24075770e+00
9.02821720e-01 -1.20887227e-01 -1.35520726e-01 6.50582552e-01
3.22530657e-01 1.68236047e-01 -7.26277590e-01 3.57486308e-01
8.69650781e-01 8.55039775e-01 -5.71819723e-01 -2.12491274e-01
-4.43525761e-01 5.50304532e-01 -2.27855980e-01 2.61800468e-01
-9.31657851e-01 -7.70796418e-01 3.76980126e-01 4.26371098e-02
5.18292665e-01 -1.76181570e-01 -4.33188856e-01 -9.16798472e-01
2.68210262e-01 -9.09855187e-01 1.91401273e-01 -8.63772690e-01
-1.30656874e+00 4.92044985e-01 1.38882473e-01 -1.93448138e+00
1.34399356e-02 -1.67649716e-01 -5.08968711e-01 4.93435204e-01
-1.54699063e+00 -5.95645845e-01 -1.44623429e-01 4.47901636e-01
3.87559652e-01 -1.08644762e-03 7.49459326e-01 5.87396622e-01
-2.03296617e-01 7.21552610e-01 8.69530514e-02 -6.12537920e-01
9.86319602e-01 -7.75332510e-01 -1.00228712e-01 1.02475226e+00
2.38672569e-01 7.17238486e-01 1.56107235e+00 -2.94237077e-01
-1.11334014e+00 -8.20868790e-01 1.14492619e+00 2.69315038e-02
6.30005300e-01 -3.51306111e-01 -9.53367949e-01 1.77372947e-01
6.15665793e-01 -2.85633355e-01 7.79790223e-01 2.36455694e-01
-5.35265505e-01 -4.18134391e-01 -1.12858844e+00 6.32774651e-01
1.05067825e+00 -4.42711502e-01 -9.39043760e-01 4.34314786e-03
8.17721844e-01 7.12816278e-03 -6.26130164e-01 4.43716645e-01
7.16999769e-01 -1.60316670e+00 8.93126667e-01 -8.88584927e-02
9.60065499e-02 -5.32524526e-01 -5.30059516e-01 -1.14701617e+00
-3.74118179e-01 -1.17596209e+00 6.11551553e-02 1.50801253e+00
4.55241203e-01 -9.15065229e-01 3.67551565e-01 -4.51005623e-02
-3.01928848e-01 -3.65748346e-01 -1.11991918e+00 -1.34769726e+00
-3.54062051e-01 -8.16147566e-01 5.64377546e-01 9.07571018e-01
4.21326399e-01 8.65175389e-03 -8.92925382e-01 3.71618003e-01
8.30468893e-01 1.51014149e-01 5.70570469e-01 -1.20512056e+00
-4.63091314e-01 -2.98725426e-01 -4.03538167e-01 -7.11843431e-01
6.62357407e-03 -4.01637882e-01 -1.91292763e-01 -6.52606308e-01
-8.03003609e-02 -2.85936058e-01 -5.26519299e-01 -1.10188983e-01
1.45199463e-01 6.67440832e-01 3.06172818e-01 1.48211375e-01
-5.04556298e-01 4.17526096e-01 6.20918632e-01 -7.36378357e-02
-2.28852049e-01 1.72082216e-01 -9.22765076e-01 9.03272390e-01
6.90693080e-01 -6.03566051e-01 -4.30757731e-01 -5.48868030e-02
5.67397833e-01 -5.29952496e-02 4.15976554e-01 -1.54066575e+00
6.34473637e-02 3.58695984e-02 -2.11034015e-01 -5.19694388e-01
6.61563218e-01 -8.21625233e-01 3.64673853e-01 3.33958417e-01
-5.81400096e-01 6.84957132e-02 9.78658870e-02 5.57539165e-01
-5.48157930e-01 -4.91703451e-01 8.33271623e-01 4.07377422e-01
-2.01373443e-01 -4.37965244e-01 -5.91595590e-01 -1.05379999e-01
6.28449976e-01 -2.28565887e-01 -3.28223377e-01 -8.54540586e-01
-7.71410823e-01 -5.30846000e-01 2.34276190e-01 3.51074368e-01
5.52905440e-01 -1.39223754e+00 -7.01787651e-01 1.67878196e-01
-4.57193144e-02 -1.11644959e+00 4.30379510e-01 1.32814145e+00
-6.66984767e-02 2.80732036e-01 -4.63898294e-02 -5.35677075e-01
-1.53527546e+00 3.82340461e-01 -1.20843641e-01 1.95319250e-01
-6.67108297e-01 7.70220995e-01 -1.74151495e-01 2.87044108e-01
3.68548334e-01 -5.30228555e-01 -7.00347200e-02 1.51302919e-01
5.65287411e-01 4.98295337e-01 4.51913297e-01 -6.28456056e-01
-2.76532531e-01 6.91290081e-01 2.52008319e-01 -4.39180881e-01
1.15082264e+00 -3.02816629e-01 -3.85627076e-02 4.33841854e-01
1.06838691e+00 1.00984979e+00 -6.46268487e-01 -1.09903872e-01
-1.82617262e-01 -8.75155032e-01 -1.21462479e-01 -4.69810873e-01
-7.43505597e-01 4.75979298e-01 6.93562269e-01 9.13666189e-01
1.39047241e+00 -1.44918919e-01 5.99662125e-01 2.11959600e-01
4.90042061e-01 -1.26056933e+00 1.77881867e-02 1.15034357e-01
1.11604834e+00 -6.09892607e-01 3.03126454e-01 -8.70020449e-01
-3.35544765e-01 8.92672002e-01 -1.70669723e-02 -3.50734413e-01
8.08972538e-01 4.77802783e-01 -2.55790949e-02 1.48532256e-01
-8.64487112e-01 -4.06551659e-02 1.27449363e-01 8.54932249e-01
2.79500961e-01 -1.05756670e-01 -1.01800060e+00 1.08174860e+00
-3.69544178e-01 -1.67134926e-01 8.23534608e-01 7.35036671e-01
-6.71904564e-01 -1.27256274e+00 -7.78816402e-01 2.43450999e-01
-6.09036088e-01 2.03172732e-02 -3.04516017e-01 6.83429360e-01
2.62877315e-01 1.44817245e+00 -3.00656408e-01 -6.95616722e-01
5.05791605e-01 1.24448873e-01 4.22780812e-01 -1.65232658e-01
-5.43092310e-01 2.70871162e-01 4.70268399e-01 -5.53593278e-01
-7.85148084e-01 -8.73467624e-01 -7.04705775e-01 -2.53156900e-01
-5.59298933e-01 2.48798877e-01 3.69922549e-01 6.81193054e-01
4.43317264e-01 3.78799587e-01 9.46348608e-01 -8.54226708e-01
-6.86319649e-01 -7.71425724e-01 -7.73643851e-01 5.44933140e-01
4.60783303e-01 -8.08767974e-01 -8.58818531e-01 1.03382491e-01]
|
[15.481310844421387, 5.5913405418396]
|
777728c8-37c6-42ab-bc12-84718098709c
|
aircraft-environmental-impact-segmentation
|
2306.13830
| null |
https://arxiv.org/abs/2306.13830v1
|
https://arxiv.org/pdf/2306.13830v1.pdf
|
Aircraft Environmental Impact Segmentation via Metric Learning
|
Metric learning is the process of learning a tailored distance metric for a particular task. This advanced subfield of machine learning is useful to any machine learning or data mining task that relies on the computation of distances or similarities over objects. In recently years, machine learning techniques have been extensively used in aviation and aerospace engineering to make predictions, extract patterns, discover knowledge, etc. Nevertheless, metric learning, an element that can advance the performance of complex machine learning tasks, has so far been hardly utilized in relevant literature. In this study, we apply classic metric learning formulations with novel components on aviation environmental impact modeling. Through a weakly-supervised metric learning task, we achieve significant improvement in the newly emerged problem of aircraft characterization and segmentation for environmental impacts. The result will enable the more efficient and accurate modeling of aircraft environmental impacts, a focal topic in sustainable aviation. This work is also a demonstration that shows the potential and value of metric learning in a wide variety of similar studies in the transportation domain.
|
['Dimitri N. Mavris', 'Zhenyu Gao']
|
2023-06-24
| null | null | null | null |
['metric-learning', 'metric-learning']
|
['computer-vision', 'methodology']
|
[ 5.50746977e-01 -3.65644336e-01 -1.51676759e-01 -6.36313438e-01
-6.35262668e-01 -4.91710633e-01 4.80247676e-01 4.85818535e-01
-3.43664855e-01 7.73646295e-01 -1.17495015e-01 -4.95666355e-01
-1.17641711e+00 -8.55476618e-01 -3.08332115e-01 -6.43007636e-01
-3.22980642e-01 3.89323384e-01 8.57058764e-02 -3.34664077e-01
5.08231878e-01 9.83120561e-01 -1.66831303e+00 3.56649786e-01
5.38871765e-01 1.11028528e+00 -3.52425724e-02 4.56732303e-01
7.79716000e-02 6.15125477e-01 -3.51823419e-01 -2.24699855e-01
3.84454817e-01 -4.48220342e-01 -6.68353021e-01 -3.16616803e-01
6.25738651e-02 3.14786285e-01 -1.00359380e-01 7.78849185e-01
3.55765343e-01 2.05881894e-01 9.71257746e-01 -1.41302192e+00
9.25262570e-02 4.05660510e-01 -3.33126485e-01 1.73482239e-01
1.95527658e-01 -1.22117959e-01 1.08270276e+00 -6.96296453e-01
1.70645669e-01 8.46557140e-01 5.90799689e-01 1.26203761e-01
-9.89421964e-01 -5.39899945e-01 -4.24457677e-02 6.70842767e-01
-1.23057401e+00 8.12143013e-02 9.24568415e-01 -6.86122417e-01
7.06107438e-01 6.00536585e-01 4.42614228e-01 4.70889330e-01
3.53752702e-01 4.96209025e-01 1.33910179e+00 -4.26650167e-01
2.43694812e-01 1.87159002e-01 -4.59348224e-02 6.67238951e-01
1.39389962e-01 6.03194773e-01 -2.62679815e-01 9.29775536e-02
9.71977320e-03 2.68045098e-01 -5.73121198e-03 -4.57882941e-01
-1.14100337e+00 8.54704440e-01 3.81274134e-01 5.92631638e-01
-1.22231603e-01 -1.99135970e-02 3.91060889e-01 7.88745463e-01
5.44947088e-01 9.63962197e-01 -7.40216553e-01 -1.62202924e-01
-1.13701451e+00 5.09575367e-01 9.84822512e-01 7.74713099e-01
7.93828964e-01 -1.73479840e-01 -1.13138407e-01 4.75647569e-01
1.25392273e-01 3.71575356e-01 5.94163463e-02 -6.69441044e-01
2.35540628e-01 9.44041193e-01 1.25038162e-01 -1.02639246e+00
-6.00686550e-01 -5.25573492e-01 -3.85823458e-01 6.23682380e-01
3.72880548e-01 -3.23628277e-01 -4.57417935e-01 1.28745687e+00
5.60451031e-01 1.75426334e-01 -3.24324280e-01 7.33871281e-01
2.25625351e-01 4.75389093e-01 -1.30257905e-01 -1.04219481e-01
8.91469240e-01 -5.38980842e-01 -4.01679397e-01 2.17914388e-01
1.04733121e+00 -9.76948380e-01 8.45354915e-01 6.44805312e-01
-5.27886391e-01 -4.73139465e-01 -1.33113623e+00 4.27175552e-01
-1.08145380e+00 -2.72007644e-01 7.64465928e-01 8.90520751e-01
-3.96461874e-01 1.14857173e+00 -3.31741035e-01 -3.21513236e-01
3.89684856e-01 5.03363729e-01 -1.11235440e-01 -3.74075621e-01
-1.25840747e+00 1.33227837e+00 4.21677858e-01 -7.82539323e-02
-8.87492418e-01 -1.04099798e+00 -8.23986411e-01 -1.20661147e-01
5.43827951e-01 -4.72547770e-01 9.65752602e-01 -4.99163926e-01
-1.01956749e+00 5.75471818e-01 4.10252512e-01 -4.75085765e-01
3.28668505e-01 -1.49034023e-01 -7.70827770e-01 -5.79977512e-01
-2.22330183e-01 5.95494211e-02 6.11472726e-01 -8.50721300e-01
-1.00999212e+00 -4.30883229e-01 2.43867442e-01 -8.66251811e-02
-2.60458887e-01 8.32179412e-02 4.37993675e-01 -4.67505962e-01
-2.53839314e-01 -9.84579623e-01 -4.27868366e-01 -1.41078115e-01
-1.86291009e-01 -3.64558220e-01 7.75070190e-01 -3.10168535e-01
1.45303285e+00 -1.98417330e+00 3.22554559e-01 5.53417563e-01
-1.64478249e-03 2.63846308e-01 1.67148672e-02 5.52644193e-01
-1.77480564e-01 -7.83557072e-02 -6.82249904e-01 2.30851799e-01
2.99997807e-01 2.87786778e-02 -8.85004736e-03 5.10938525e-01
2.61624932e-01 6.02946877e-01 -9.25980389e-01 -3.22946370e-01
4.86263365e-01 2.64875367e-02 -2.79213339e-01 2.15577424e-01
-1.49062335e-01 3.59306246e-01 -4.56329733e-01 7.36252606e-01
3.19075316e-01 4.30814624e-01 -3.48006189e-02 -3.07098746e-01
-4.83504921e-01 -1.36494428e-01 -1.09464145e+00 1.80004084e+00
-6.79641128e-01 6.80491686e-01 -2.91061133e-01 -1.48177588e+00
1.00164855e+00 8.66296515e-02 1.17258871e+00 -8.00460994e-01
2.73691446e-01 2.69492179e-01 3.19663614e-01 -7.28567302e-01
2.97548175e-01 -3.92422587e-01 -3.09116900e-01 5.26908994e-01
-2.24467590e-01 -3.19265544e-01 3.78396958e-01 -3.74363750e-01
1.09858572e+00 9.02748704e-02 2.29974866e-01 -4.04615939e-01
8.95532727e-01 3.52451652e-01 4.74052191e-01 8.38972852e-02
-1.54594094e-01 1.37419268e-01 2.18348965e-01 -4.83361185e-01
-9.94855523e-01 -8.73342395e-01 -3.41340661e-01 1.04435813e+00
-2.85440534e-02 -2.85908610e-01 -6.38710558e-01 -8.57413650e-01
3.22992504e-01 7.54405379e-01 -4.60029721e-01 -5.95465243e-01
-6.68712437e-01 -7.73640275e-01 2.86962748e-01 4.28588122e-01
4.34988514e-02 -7.75479376e-01 -5.19997597e-01 2.26866096e-01
2.86346346e-01 -8.72785866e-01 -1.46334007e-01 2.20644161e-01
-8.93968463e-01 -1.53444135e+00 -4.15035725e-01 -5.74494183e-01
3.77987206e-01 1.44986376e-01 1.04063118e+00 -1.88413069e-01
-7.02503800e-01 2.96996742e-01 -3.17533702e-01 -8.70528400e-01
-2.84368724e-01 2.49484673e-01 1.82331368e-01 7.81765357e-02
6.05445981e-01 -4.35333252e-01 -5.49525976e-01 7.91557252e-01
-8.13960850e-01 -5.09559035e-01 6.26993120e-01 6.04264200e-01
5.90531290e-01 6.06776953e-01 8.24470043e-01 -9.31725204e-01
6.20780408e-01 -7.36936152e-01 -5.44428468e-01 3.08258086e-01
-1.08230484e+00 1.50380284e-01 5.87890148e-01 -6.45714253e-03
-7.77083278e-01 -1.31547600e-01 -1.90443266e-02 5.47377244e-02
-3.11614126e-01 7.79415607e-01 -3.30487251e-01 -2.86021769e-01
6.77912772e-01 -3.11449319e-01 -1.16458252e-01 -5.30825794e-01
3.18412125e-01 7.19114840e-01 -1.55848702e-02 -5.54514587e-01
9.04007852e-01 2.72027463e-01 7.66758442e-01 -5.56959212e-01
-9.22943771e-01 -6.74334764e-01 -1.02571440e+00 -4.90021288e-01
8.23857427e-01 -2.20184594e-01 -4.43317831e-01 -1.73315987e-01
-6.88649714e-01 -1.24737069e-01 -6.46858811e-01 8.07908595e-01
-7.20447361e-01 -6.08500503e-02 1.68678656e-01 -6.79012060e-01
8.63133073e-02 -9.05858874e-01 6.53688967e-01 4.99481261e-02
-3.68692219e-01 -1.26730728e+00 4.26970303e-01 3.77414465e-01
4.75435197e-01 8.31228256e-01 1.32499373e+00 -7.69788027e-01
-3.87195855e-01 -4.51349765e-01 -4.77652550e-02 5.99264443e-01
4.66640651e-01 -1.22489929e-01 -7.96647787e-01 -2.77955532e-01
3.04087028e-02 -9.02526006e-02 7.38771796e-01 1.27305642e-01
1.16301644e+00 1.10141777e-01 -4.11025345e-01 3.21104079e-01
1.44196367e+00 5.00374317e-01 2.37488508e-01 2.46947721e-01
5.24622023e-01 1.39249241e+00 1.46429384e+00 5.55690110e-01
-6.96323961e-02 5.92846394e-01 6.45406008e-01 -2.93711185e-01
2.53283214e-02 1.91615373e-01 -1.18068140e-02 5.59445620e-01
-1.01834260e-01 1.82868809e-01 -1.02057207e+00 4.92648780e-01
-1.87954247e+00 -9.13154602e-01 -6.72039613e-02 2.34397125e+00
4.55841750e-01 1.97803974e-01 1.96313575e-01 6.65418804e-01
5.33902287e-01 7.57984295e-02 -5.61080277e-01 -6.60741389e-01
3.08498353e-01 3.18455011e-01 4.86459792e-01 2.28715062e-01
-1.17347121e+00 5.45401931e-01 5.99994421e+00 6.73044205e-01
-7.62891710e-01 -4.57780398e-02 2.72015184e-01 -3.39056134e-01
-2.05320120e-01 -2.70647079e-01 -4.69277471e-01 3.45586658e-01
1.31825829e+00 -3.54093879e-01 4.86835808e-01 8.12742710e-01
4.17328030e-01 8.01941305e-02 -1.61909080e+00 8.54528964e-01
4.94312160e-02 -1.05414915e+00 -2.19527289e-01 1.10416345e-01
8.81111681e-01 -1.33890361e-01 2.21176788e-01 3.31708848e-01
-1.02289591e-03 -1.26434553e+00 2.25210205e-01 8.56367290e-01
6.71912611e-01 -1.38216484e+00 8.65337312e-01 2.83947378e-01
-1.10399902e+00 -4.51036811e-01 -4.21272039e-01 -1.27556249e-01
3.13775569e-01 8.42403889e-01 -9.51002300e-01 9.46210980e-01
4.46804851e-01 8.68055284e-01 -3.10800344e-01 1.51959944e+00
1.88095510e-01 5.21529555e-01 2.01479923e-02 -7.48538971e-02
4.67496753e-01 -4.04369593e-01 5.57804823e-01 1.25720787e+00
5.16946912e-01 -2.71647811e-01 9.01502594e-02 5.50547838e-01
1.56872913e-01 2.33222961e-01 -1.08559644e+00 -4.74113524e-02
6.48131892e-02 1.15441859e+00 -4.50463027e-01 2.24757880e-01
-4.92296040e-01 3.27574283e-01 -2.54869044e-01 1.59833988e-03
-1.01571047e+00 -8.71406913e-01 1.03792429e+00 3.06852490e-01
1.19611025e-01 -4.89395440e-01 -4.40380573e-01 -2.68232882e-01
-1.64475113e-01 -7.19447374e-01 4.82634515e-01 -1.27513200e-01
-1.28857100e+00 3.18868399e-01 1.58636540e-01 -1.62485158e+00
-2.96236783e-01 -1.04944658e+00 -5.96058667e-01 6.40845597e-01
-1.72920227e+00 -9.92651105e-01 -1.90034211e-01 6.14443958e-01
4.48951453e-01 -4.53975260e-01 7.39470065e-01 6.07238412e-01
-3.75870973e-01 1.35946691e-01 3.73670936e-01 -3.67056131e-01
8.95328045e-01 -1.40597081e+00 -8.20511729e-02 5.15972912e-01
2.26171926e-01 2.79813614e-02 6.79371059e-01 -3.92528296e-01
-1.25668824e+00 -1.34489143e+00 8.90287876e-01 -6.67659819e-01
8.31214905e-01 -1.55701399e-01 -3.72114360e-01 1.75606683e-01
-5.45112304e-02 -3.16450119e-01 1.38742900e+00 2.40303203e-01
-7.95484781e-02 -7.54399240e-01 -1.30777609e+00 4.32541937e-01
1.03435898e+00 -5.15025556e-01 -4.45802599e-01 5.46505988e-01
5.91138959e-01 8.03239122e-02 -1.27781272e+00 6.02914155e-01
6.51114464e-01 -7.34058261e-01 9.39491332e-01 -1.18632364e+00
2.76412696e-01 -5.37562251e-01 -4.42118168e-01 -1.37797356e+00
-3.63396972e-01 -3.62618536e-01 -1.24061346e-01 1.17637873e+00
6.55682385e-01 -1.92560673e-01 7.81539559e-01 6.70590460e-01
-3.86501968e-01 -1.02147031e+00 -9.37197685e-01 -1.05607224e+00
1.95314348e-01 -7.76046574e-01 9.83191133e-01 9.72143054e-01
-8.29936713e-02 -6.29224256e-02 -3.74358445e-01 -2.26314878e-03
7.64629245e-01 2.01927513e-01 4.28978264e-01 -1.64791834e+00
-4.24895659e-02 -5.28582215e-01 -7.11226225e-01 -1.65238291e-01
1.73122391e-01 -1.19497943e+00 -1.08880825e-01 -1.36767793e+00
-3.55314314e-01 -6.32614911e-01 -1.01884627e+00 -1.67357009e-02
2.28660628e-01 1.12803921e-01 4.07227548e-03 -3.09163660e-01
-3.47546905e-01 3.71309936e-01 1.25512505e+00 -5.02902806e-01
-4.52764407e-02 5.92164516e-01 -7.12201536e-01 7.04843879e-01
9.53541040e-01 -8.31477880e-01 -7.22374618e-01 -1.35385290e-01
3.52110595e-01 -3.63940150e-01 2.80514117e-02 -1.21202934e+00
2.56166369e-01 -5.33271730e-01 1.52856469e-01 -4.53753024e-01
2.30848063e-02 -1.44629455e+00 9.47902203e-02 5.77277005e-01
-1.62440121e-01 2.79767960e-01 -1.27893230e-02 5.70463598e-01
-3.94884735e-01 -4.88455027e-01 5.74884832e-01 5.85044064e-02
-9.32421088e-01 5.39529979e-01 -2.28237420e-01 -1.79185465e-01
1.65064108e+00 -2.92400956e-01 2.46441260e-01 2.64716208e-01
-6.90633059e-01 2.61580855e-01 1.02371790e-01 4.91569281e-01
5.99223852e-01 -1.37368679e+00 -7.84519672e-01 -1.45389214e-01
2.75891155e-01 -2.84145743e-01 5.48434742e-02 8.71580064e-01
-3.12950820e-01 6.34034693e-01 -3.80513459e-01 -4.03696746e-01
-1.21621943e+00 7.00345039e-01 1.15664527e-01 -4.27185714e-01
-1.33399200e-02 4.35770333e-01 -3.04016858e-01 -4.19915140e-01
1.80000857e-01 -3.17966431e-01 -2.99969941e-01 3.16952169e-01
3.95148933e-01 1.11785448e+00 3.92532259e-01 -2.86343306e-01
-4.42644626e-01 7.28431702e-01 7.31680095e-02 3.35118696e-02
1.41023636e+00 8.97073448e-02 5.72188059e-04 6.10484838e-01
1.16548729e+00 -3.18612993e-01 -9.91196573e-01 -1.12761103e-01
7.40206718e-01 -4.95818079e-01 3.10736671e-02 -9.87105846e-01
-8.03029358e-01 9.97425973e-01 7.73073733e-01 1.91790029e-01
1.29502404e+00 -2.39602461e-01 7.41033971e-01 6.05069101e-01
5.47254443e-01 -1.32373881e+00 2.18707174e-02 4.03930217e-01
8.41675162e-01 -1.31871140e+00 9.91259366e-02 -1.15778103e-01
-3.11767399e-01 1.02020550e+00 3.31560582e-01 8.18598121e-02
1.12145829e+00 4.05950338e-01 -3.49232048e-01 -2.29615659e-01
-4.10870701e-01 -4.31981146e-01 6.30123198e-01 7.87509143e-01
5.86220980e-01 1.68933153e-01 -4.33914930e-01 3.93640578e-01
-7.65009522e-02 -1.81540504e-01 4.74506877e-02 1.06166494e+00
-6.42989933e-01 -1.52858365e+00 -8.87047052e-02 5.79241216e-01
-2.86965072e-01 4.66016494e-02 -5.24331510e-01 7.69561589e-01
5.39738894e-01 1.04731452e+00 -2.02716634e-01 -6.85170949e-01
7.22410500e-01 8.28496739e-03 5.15401959e-01 -6.85474575e-01
-6.71354234e-01 -6.42226577e-01 2.46536985e-01 -7.68918931e-01
-5.03030777e-01 -7.36792445e-01 -9.64450717e-01 -2.99190640e-01
-2.08410993e-01 3.56349945e-01 1.12228119e+00 1.01883626e+00
1.08332664e-01 8.73302817e-01 1.17795908e+00 -4.56584275e-01
-5.41403055e-01 -7.19621539e-01 -8.56613636e-01 1.97093889e-01
1.36175111e-01 -8.19308817e-01 -2.25181147e-01 -1.56196475e-01]
|
[8.335380554199219, 4.2160491943359375]
|
6a013663-b11e-4f5a-9654-24f9fc5ee0a8
|
neural-scene-decoration-from-a-single
|
2108.01806
| null |
https://arxiv.org/abs/2108.01806v2
|
https://arxiv.org/pdf/2108.01806v2.pdf
|
Neural Scene Decoration from a Single Photograph
|
Furnishing and rendering indoor scenes has been a long-standing task for interior design, where artists create a conceptual design for the space, build a 3D model of the space, decorate, and then perform rendering. Although the task is important, it is tedious and requires tremendous effort. In this paper, we introduce a new problem of domain-specific indoor scene image synthesis, namely neural scene decoration. Given a photograph of an empty indoor space and a list of decorations with layout determined by user, we aim to synthesize a new image of the same space with desired furnishing and decorations. Neural scene decoration can be applied to create conceptual interior designs in a simple yet effective manner. Our attempt to this research problem is a novel scene generation architecture that transforms an empty scene and an object layout into a realistic furnished scene photograph. We demonstrate the performance of our proposed method by comparing it with conditional image synthesis baselines built upon prevailing image translation approaches both qualitatively and quantitatively. We conduct extensive experiments to further validate the plausibility and aesthetics of our generated scenes. Our implementation is available at \url{https://github.com/hkust-vgd/neural_scene_decoration}.
|
['Duc Thanh Nguyen', 'Phuoc-Hieu Le', 'Sai-Kit Yeung', 'Binh-Son Hua', 'Yingshu Chen', 'Hong-Wing Pang']
|
2021-08-04
| null | null | null | null |
['scene-generation']
|
['computer-vision']
|
[ 7.45607495e-01 5.06346673e-03 5.33416092e-01 -4.71733123e-01
-3.32625896e-01 -8.65641415e-01 6.74909890e-01 -3.28451246e-01
1.83065131e-01 6.30184054e-01 3.67773473e-01 -4.35341895e-01
1.09182067e-01 -1.07680809e+00 -9.74604666e-01 -4.05348629e-01
5.07214367e-01 4.27489989e-02 -3.29026371e-01 -2.31921047e-01
1.09211132e-01 6.40797913e-01 -1.67260718e+00 3.38005930e-01
7.99160182e-01 8.63087058e-01 7.21523702e-01 8.46394956e-01
-1.65094480e-01 5.60131669e-01 -5.61711311e-01 -4.24520910e-01
4.61324453e-01 -6.99872851e-01 -8.62254739e-01 4.53593642e-01
5.92666090e-01 -2.94866383e-01 -5.58009855e-02 9.56910014e-01
4.22819942e-01 3.36597741e-01 5.80573976e-01 -1.10680699e+00
-1.21149766e+00 2.98952252e-01 -2.73020893e-01 -5.21084905e-01
5.14860988e-01 3.20399046e-01 9.42748487e-01 -7.65801787e-01
7.07262158e-01 1.28577459e+00 4.62203503e-01 7.36442149e-01
-1.60419965e+00 -4.89671737e-01 2.40318209e-01 -3.61517668e-01
-1.35964346e+00 -4.31323677e-01 1.11143124e+00 -3.62893820e-01
3.81035209e-01 7.48647034e-01 9.45708156e-01 1.13481569e+00
-1.35839343e-01 7.31192887e-01 1.38232136e+00 -4.39195216e-01
5.37329733e-01 1.47829250e-01 -4.74253684e-01 6.44811034e-01
-8.04948155e-04 -3.65845747e-02 -1.67399555e-01 1.17172264e-01
1.22300971e+00 1.38999388e-01 -4.04712498e-01 -5.92103004e-01
-1.18774247e+00 5.18237114e-01 9.48147893e-01 2.48135149e-01
-4.33121145e-01 5.42634726e-01 -1.17050610e-01 2.06309520e-02
3.29342723e-01 9.22996759e-01 -2.47838080e-01 1.04361758e-01
-7.41389573e-01 5.43312490e-01 4.87593770e-01 1.13889229e+00
5.70662081e-01 6.41172677e-02 -2.01209903e-01 8.35682392e-01
1.37450948e-01 4.71346855e-01 4.22365731e-03 -1.32334840e+00
1.57853439e-01 4.11687523e-01 2.65085697e-01 -9.57149327e-01
-4.56422605e-02 -3.16399187e-01 -9.06289458e-01 4.27087992e-01
1.09058850e-01 -4.21514781e-03 -8.65806282e-01 1.76335359e+00
2.94255704e-01 5.52863441e-02 4.69016656e-03 1.04466462e+00
8.70053232e-01 7.91898489e-01 1.36619121e-01 3.07303816e-01
1.28445673e+00 -1.08071578e+00 -5.54057240e-01 -2.85475284e-01
1.26935691e-01 -9.46941733e-01 1.67028153e+00 9.15229023e-02
-1.19124770e+00 -8.08980286e-01 -9.41953063e-01 -3.87508303e-01
-3.24107379e-01 2.21652508e-01 7.69510686e-01 6.18568599e-01
-1.20734811e+00 6.03603125e-01 -2.93187797e-01 -5.90090334e-01
4.96250570e-01 -3.41477580e-02 -2.02534169e-01 3.48312110e-02
-6.83008194e-01 5.15069962e-01 1.32799610e-01 1.36256352e-01
-7.43283391e-01 -7.38505960e-01 -8.96703184e-01 -9.94687006e-02
1.13417238e-01 -1.28779912e+00 1.41818511e+00 -1.08015668e+00
-1.50962019e+00 8.81527603e-01 -7.32204989e-02 4.39894535e-02
4.78775114e-01 -2.03179345e-01 -1.05263799e-01 -3.52728903e-01
2.37813637e-01 9.12444115e-01 6.10983431e-01 -2.03268671e+00
-2.83743143e-01 -1.13864318e-01 5.10378003e-01 5.07526398e-01
1.98562384e-01 -4.62072015e-01 -3.98756981e-01 -8.80519629e-01
2.74047226e-01 -8.42782259e-01 -3.92560631e-01 3.84208560e-01
-6.64714456e-01 5.28060019e-01 5.30385196e-01 -7.17959464e-01
9.35220063e-01 -2.32099175e+00 1.12302512e-01 1.70328006e-01
-1.13590240e-01 -2.31701180e-01 -2.07080528e-01 5.60832262e-01
-2.16163233e-01 3.19393396e-01 -4.66405421e-01 -6.67703629e-01
1.44167259e-01 4.57813293e-02 -4.82878000e-01 1.39435977e-01
2.31790338e-02 1.15219414e+00 -9.70177054e-01 -2.28002295e-01
6.04264379e-01 6.60308123e-01 -8.30496311e-01 4.77543235e-01
-4.36686605e-01 8.57579470e-01 -3.63629222e-01 5.14361739e-01
6.36604846e-01 -2.49087617e-01 6.53624907e-02 -3.85651439e-01
-2.16733992e-01 1.76532790e-01 -1.18999290e+00 2.21644115e+00
-8.16544712e-01 6.60381615e-01 -7.06179887e-02 -5.45353115e-01
1.02468324e+00 -2.94320006e-02 1.89043637e-02 -7.58298159e-01
2.54776865e-01 4.56228144e-02 -5.61161399e-01 -4.11124855e-01
8.38122845e-01 -3.23661774e-01 -2.34199315e-01 3.06211889e-01
-4.15951014e-01 -9.46668744e-01 -2.45148376e-01 -5.22120260e-02
8.73270690e-01 7.82731891e-01 2.29394689e-01 -1.70114741e-01
2.41466239e-01 4.32335353e-03 -4.37923335e-02 5.79976737e-01
2.98621327e-01 1.15203834e+00 1.74452752e-01 -4.37295854e-01
-1.42136896e+00 -1.44422996e+00 1.08347408e-01 7.46616304e-01
2.42447302e-01 -2.36504182e-01 -1.01322424e+00 -2.53818631e-01
-2.44225487e-01 1.12882578e+00 -7.15811789e-01 2.94296001e-03
-4.61813688e-01 -1.36847079e-01 -1.73301455e-02 2.72399545e-01
7.39033341e-01 -1.23209119e+00 -8.33603323e-01 -8.92060995e-02
-2.98586756e-01 -8.83117735e-01 -5.74962735e-01 -1.24425538e-01
-6.96327269e-01 -7.41115153e-01 -8.64086866e-01 -9.80319142e-01
1.12518501e+00 4.64694589e-01 1.12302792e+00 1.41503513e-01
-4.06697333e-01 3.93605828e-01 -1.75260931e-01 -2.17008144e-01
-4.94265139e-01 -2.12659866e-01 -1.77566260e-01 1.01677142e-01
-6.06744349e-01 -9.34312284e-01 -9.53561246e-01 1.49388298e-01
-1.09060609e+00 1.12152326e+00 3.75153065e-01 4.98655081e-01
6.55882716e-01 1.41804576e-01 6.12287037e-02 -9.90813971e-01
7.37912834e-01 -1.46976948e-01 -4.70504403e-01 1.68803647e-01
1.13895827e-03 5.26626371e-02 7.78343320e-01 -4.73572910e-02
-1.26048911e+00 2.82645106e-01 -2.15946138e-01 -3.47514212e-01
-4.40696150e-01 7.60628879e-02 -4.11242127e-01 1.85984999e-01
7.07771719e-01 2.30295151e-01 -3.68159145e-01 -6.39159560e-01
9.65736032e-01 4.87215757e-01 8.61528158e-01 -8.19475949e-01
1.03134656e+00 5.93187690e-01 -7.57594183e-02 -7.31258452e-01
-6.77574992e-01 9.98600796e-02 -7.73671627e-01 -3.66594672e-01
9.08704221e-01 -7.48600304e-01 -5.22791624e-01 2.50654578e-01
-1.35151386e+00 -7.15948820e-01 -7.20102489e-01 -1.99944466e-01
-7.69639075e-01 7.55624548e-02 -6.09098487e-02 -7.77640283e-01
-1.27569035e-01 -1.03515482e+00 1.34122074e+00 1.63132146e-01
-4.54929382e-01 -6.96301222e-01 -1.28223523e-02 4.08111751e-01
4.46184188e-01 9.01243687e-01 9.79263723e-01 4.87478852e-01
-8.78042996e-01 1.43647969e-01 -4.90355045e-01 1.96586177e-01
6.34979427e-01 2.18738541e-02 -1.16736794e+00 6.55667111e-02
-2.69338369e-01 -2.22851299e-02 3.35106790e-01 3.23331803e-01
1.47466040e+00 -4.05828476e-01 -2.70141512e-02 7.83673704e-01
1.65009153e+00 3.90338331e-01 6.85627460e-01 3.11983436e-01
8.42329562e-01 7.65927315e-01 3.24280083e-01 4.15344298e-01
3.66279125e-01 6.76617026e-01 3.07933927e-01 -5.33830047e-01
-5.93119502e-01 -8.13811302e-01 -1.77688271e-01 5.05897284e-01
-1.00753054e-01 -3.09840739e-01 -5.97702026e-01 4.51587826e-01
-1.66536462e+00 -7.85989761e-01 -2.42582448e-02 2.05738950e+00
7.28708923e-01 -2.05441058e-01 -2.33166561e-01 6.17070608e-02
5.50618947e-01 2.18203157e-01 -4.31032628e-01 -5.96267283e-01
-1.44113764e-01 3.03660065e-01 2.24416777e-01 4.97726560e-01
-9.65041220e-01 1.01933289e+00 5.77674532e+00 4.77174968e-01
-9.63390708e-01 -8.52057338e-02 8.27832818e-01 -3.22178006e-02
-8.00588191e-01 -4.52896282e-02 -5.52601293e-02 3.12377483e-01
4.63797450e-01 4.78964113e-02 8.86295438e-01 8.23177576e-01
4.33172047e-01 -5.56560792e-02 -1.15295553e+00 1.21088827e+00
-3.42207290e-02 -1.68499517e+00 2.65373498e-01 -8.35487843e-02
1.04727721e+00 -6.34218693e-01 2.82199919e-01 8.32008105e-03
4.36151445e-01 -9.68919814e-01 1.33744121e+00 6.35716558e-01
1.03387260e+00 -5.15136600e-01 3.27277035e-02 -3.99193838e-02
-1.41943288e+00 9.63299051e-02 -2.61821486e-02 -2.72169501e-01
3.27932507e-01 2.25238666e-01 -5.50921559e-01 4.01962996e-01
6.80039346e-01 4.19434249e-01 -6.70089006e-01 1.01504481e+00
-4.57509577e-01 6.62903413e-02 -4.04275097e-02 -7.92460814e-02
1.42726287e-01 -3.17440122e-01 1.74655154e-01 8.92613411e-01
5.09726822e-01 1.77795187e-01 -4.80043404e-02 1.56944597e+00
-1.93067029e-01 1.77000731e-01 -1.06870365e+00 1.44893453e-01
2.55536944e-01 1.09054923e+00 -8.57680559e-01 -2.31514305e-01
4.03110832e-02 1.64827156e+00 2.80770481e-01 5.85570097e-01
-1.09123611e+00 -2.82931268e-01 6.42748475e-01 2.48774856e-01
7.54044801e-02 -3.88163477e-01 -7.10305631e-01 -9.24070835e-01
5.37649803e-02 -5.32948554e-01 -3.78365517e-01 -1.62427986e+00
-9.99962151e-01 7.10507631e-01 -8.81263462e-04 -1.22059274e+00
1.64883330e-01 -3.42119545e-01 -6.09522879e-01 8.21938455e-01
-1.10061240e+00 -1.47087061e+00 -6.56449854e-01 3.56409103e-01
8.25442433e-01 1.90314308e-01 9.10221636e-01 2.72757679e-01
-3.26157331e-01 1.27379373e-01 -1.68170109e-02 -1.08418420e-01
2.57594556e-01 -1.33457220e+00 1.01706469e+00 7.67540574e-01
1.63310841e-01 6.19677663e-01 8.57708097e-01 -6.30466640e-01
-1.43481278e+00 -1.37760603e+00 6.24692619e-01 -4.79738683e-01
1.94985166e-01 -7.57878900e-01 -3.55574906e-01 5.45124650e-01
3.16510320e-01 -3.50107431e-01 5.32682896e-01 -2.79278517e-01
-1.64911374e-01 8.75637010e-02 -1.15674484e+00 1.31072783e+00
1.59542084e+00 -4.67837155e-01 -4.09397572e-01 3.09210598e-01
1.02445197e+00 -3.76732051e-01 -5.20350456e-01 1.55975178e-01
6.03529513e-01 -1.13327396e+00 1.18375468e+00 -1.66377217e-01
7.46228993e-01 -6.25708818e-01 -5.18344343e-01 -1.34105980e+00
-5.32932580e-01 -6.46741748e-01 4.19995964e-01 1.21277022e+00
3.18190008e-01 -3.99323761e-01 7.26832747e-01 9.35380936e-01
-3.43554318e-01 -5.70337772e-01 -4.85747695e-01 -6.02832317e-01
-2.83599645e-01 -5.93340635e-01 1.10554731e+00 7.49967158e-01
-5.86916268e-01 3.55265409e-01 -3.42286795e-01 1.64184049e-01
6.08650267e-01 5.88813365e-01 1.15475166e+00 -7.79337466e-01
-3.13951701e-01 -5.06210983e-01 3.53324711e-02 -1.09157264e+00
-1.51392907e-01 -8.39188039e-01 1.57835126e-01 -2.13745022e+00
1.67574193e-02 -7.46147990e-01 1.89245060e-01 3.40707123e-01
1.16071619e-01 4.73038882e-01 4.22860891e-01 -2.90209670e-02
-2.26904660e-01 7.37886846e-01 1.68237829e+00 -6.55830503e-02
-2.86533684e-01 -2.78200179e-01 -1.03688216e+00 5.33731103e-01
9.77358758e-01 8.27459246e-02 -6.14137828e-01 -6.60961509e-01
1.43796444e-01 -1.22191124e-01 7.17680991e-01 -1.12523270e+00
-2.80553192e-01 -3.34996819e-01 5.35939991e-01 -3.49885941e-01
6.89179897e-01 -9.50618505e-01 9.54935014e-01 4.41525012e-01
-3.03627521e-01 4.16302904e-02 2.63392359e-01 2.24545911e-01
2.84254432e-01 9.42549258e-02 7.04197764e-01 -3.28003734e-01
-7.56895542e-01 1.12314872e-01 4.60607596e-02 -3.62430543e-01
8.81054044e-01 -4.97520089e-01 -1.49049640e-01 -3.81672412e-01
-5.65011621e-01 -3.38664830e-01 1.00700533e+00 6.48802996e-01
8.45157325e-01 -1.67287683e+00 -3.75280857e-01 3.60770226e-01
1.10740095e-01 1.86752841e-01 4.94890481e-01 1.02225043e-01
-9.54390049e-01 1.62020102e-01 -3.25216264e-01 -3.51738960e-01
-1.09707415e+00 7.09678769e-01 2.24246204e-01 3.11188042e-01
-7.60785401e-01 8.25277090e-01 8.54431629e-01 -7.29028940e-01
-1.51304854e-02 -7.08440602e-01 1.85620800e-01 -5.58652878e-01
2.03503847e-01 -4.27469835e-02 -3.48137528e-01 -5.87122083e-01
9.78897959e-02 7.68350542e-01 6.33879364e-01 -2.56635576e-01
1.21228409e+00 -3.19409311e-01 3.25721852e-03 4.38834310e-01
1.14649308e+00 -1.42875746e-01 -1.32282364e+00 1.00672901e-01
-5.68240643e-01 -9.03201640e-01 -1.18484631e-01 -8.99778187e-01
-9.19411719e-01 6.07527852e-01 6.20303333e-01 8.80341902e-02
1.21992576e+00 -1.78875402e-02 7.43356287e-01 5.48091792e-02
6.10092640e-01 -7.03424633e-01 1.57204553e-01 -1.58836339e-02
1.54468966e+00 -9.74864900e-01 -9.92950797e-02 -5.20922840e-01
-5.92586935e-01 7.17214704e-01 4.79914576e-01 -1.81702211e-01
4.44967330e-01 1.74694899e-02 -2.32145097e-02 -3.54332268e-01
-2.49494568e-01 -1.80550486e-01 3.97860587e-01 5.93035460e-01
3.97981465e-01 4.26117539e-01 9.65407565e-02 4.33040798e-01
-8.82355511e-01 -4.13067751e-02 3.05392206e-01 1.00109267e+00
-1.64890915e-01 -9.99333382e-01 -4.54696089e-01 7.16152489e-02
1.37640938e-01 -1.15506180e-01 -4.25462157e-01 4.69129801e-01
3.74845266e-01 6.75545156e-01 2.90290769e-02 -4.12529707e-01
7.31659412e-01 -2.59132266e-01 5.74907839e-01 -6.53561532e-01
-4.15566176e-01 -4.52990718e-02 1.47662520e-01 -6.07085884e-01
-2.72688776e-01 -5.75322151e-01 -1.05282032e+00 -3.81525338e-01
2.30311632e-01 -2.63486117e-01 9.05958712e-01 3.63693655e-01
4.58389133e-01 8.30577552e-01 7.09456742e-01 -1.21781731e+00
3.24826837e-01 -4.21345562e-01 -4.78741378e-01 6.10850275e-01
2.88470179e-01 -4.75528598e-01 1.18140494e-02 4.87283558e-01]
|
[9.362787246704102, -2.9688239097595215]
|
91dfaf20-d2e2-49e4-a0db-35545b635c3e
|
solov2-dynamic-faster-and-stronger
|
2003.10152
| null |
https://arxiv.org/abs/2003.10152v3
|
https://arxiv.org/pdf/2003.10152v3.pdf
|
SOLOv2: Dynamic and Fast Instance Segmentation
|
In this work, we aim at building a simple, direct, and fast instance segmentation framework with strong performance. We follow the principle of the SOLO method of Wang et al. "SOLO: segmenting objects by locations". Importantly, we take one step further by dynamically learning the mask head of the object segmenter such that the mask head is conditioned on the location. Specifically, the mask branch is decoupled into a mask kernel branch and mask feature branch, which are responsible for learning the convolution kernel and the convolved features respectively. Moreover, we propose Matrix NMS (non maximum suppression) to significantly reduce the inference time overhead due to NMS of masks. Our Matrix NMS performs NMS with parallel matrix operations in one shot, and yields better results. We demonstrate a simple direct instance segmentation system, outperforming a few state-of-the-art methods in both speed and accuracy. A light-weight version of SOLOv2 executes at 31.3 FPS and yields 37.1% AP. Moreover, our state-of-the-art results in object detection (from our mask byproduct) and panoptic segmentation show the potential to serve as a new strong baseline for many instance-level recognition tasks besides instance segmentation. Code is available at: https://git.io/AdelaiDet
|
['Chunhua Shen', 'Lei LI', 'Rufeng Zhang', 'Xinlong Wang', 'Tao Kong']
|
2020-03-23
| null |
http://proceedings.neurips.cc/paper/2020/hash/cd3afef9b8b89558cd56638c3631868a-Abstract.html
|
http://proceedings.neurips.cc/paper/2020/file/cd3afef9b8b89558cd56638c3631868a-Paper.pdf
|
neurips-2020-12
|
['real-time-instance-segmentation']
|
['computer-vision']
|
[ 2.72093892e-01 -2.04445794e-01 -4.17687893e-02 -3.19116086e-01
-8.46409440e-01 -8.24808896e-01 3.87163937e-01 -4.76670116e-02
-5.90364039e-01 1.97734416e-01 -5.88871002e-01 -4.22677040e-01
3.16780537e-01 -9.06986117e-01 -1.00846195e+00 -8.41261625e-01
6.21157475e-02 6.33079708e-01 6.98989272e-01 1.33541077e-01
3.34897488e-01 7.36387432e-01 -1.57241058e+00 3.96950185e-01
8.40236127e-01 1.24023056e+00 4.69449699e-01 8.89186144e-01
-9.71754566e-02 3.62535208e-01 -5.72475314e-01 -2.83709109e-01
6.60457253e-01 1.49196610e-02 -7.02604711e-01 3.09019089e-01
6.47620440e-01 -2.79289514e-01 -2.39200145e-01 9.91700709e-01
4.65666473e-01 4.40809205e-02 4.00696039e-01 -9.57987130e-01
-2.11366266e-01 5.46212971e-01 -9.70986843e-01 2.03224018e-01
-2.73644298e-01 6.70822382e-01 9.59244549e-01 -1.02616441e+00
3.29698473e-01 8.97661328e-01 5.61330020e-01 3.47694844e-01
-1.49359250e+00 -7.40091622e-01 3.07834864e-01 -4.32734303e-02
-1.51807773e+00 -4.14326727e-01 1.02010757e-01 -4.97826576e-01
8.96854877e-01 5.68478167e-01 6.51407659e-01 5.00531733e-01
-1.95310459e-01 1.06493580e+00 1.05658245e+00 -2.64935523e-01
1.19030096e-01 1.58493161e-01 2.46020645e-01 8.94185424e-01
1.46492377e-01 -1.92421019e-01 -3.49575102e-01 4.02996764e-02
9.14714217e-01 -4.73802015e-02 -1.94293499e-01 -1.31050646e-01
-1.30322313e+00 6.55246317e-01 4.47877735e-01 -1.43698491e-02
-2.73021430e-01 2.74454236e-01 1.63933650e-01 -8.46102387e-02
4.85279888e-01 4.50035363e-01 -5.42619824e-01 -9.70041379e-02
-1.40029907e+00 2.59478509e-01 9.06782687e-01 9.15810704e-01
1.02467108e+00 -1.83086365e-01 -4.97386664e-01 7.87390471e-01
9.00904089e-02 6.79449677e-01 5.11796959e-02 -1.20744646e+00
4.14470971e-01 6.48845077e-01 1.04441695e-01 -6.27136886e-01
-4.39267457e-01 -7.17314124e-01 -5.39470196e-01 2.80377507e-01
4.64980930e-01 -2.86016434e-01 -1.15368700e+00 1.28737855e+00
7.30892479e-01 6.05832517e-01 -3.49367678e-01 9.59005833e-01
6.99900746e-01 8.79851758e-01 -2.39641994e-01 1.52582362e-01
1.58246934e+00 -1.29355752e+00 -1.94671854e-01 -3.75992715e-01
5.29996037e-01 -9.30665016e-01 1.10823750e+00 5.34912705e-01
-1.27478957e+00 -6.83718085e-01 -8.09541941e-01 -4.78623360e-02
-3.88910770e-01 4.46099639e-01 8.84720206e-01 5.21610737e-01
-9.62669194e-01 5.48581183e-01 -1.15527868e+00 -4.82212193e-02
7.12783515e-01 5.81204236e-01 1.34297356e-01 2.51565516e-01
-5.30317605e-01 4.59417731e-01 3.91090900e-01 2.20028341e-01
-8.24542820e-01 -9.44042027e-01 -5.69097579e-01 1.45513043e-01
8.17767024e-01 -5.44831336e-01 1.34923768e+00 -7.02683508e-01
-1.41957438e+00 9.78042483e-01 -3.08403760e-01 -6.43720925e-01
6.05574608e-01 -4.38791513e-01 -3.20013463e-02 2.82502562e-01
7.31894746e-02 1.07616079e+00 9.88560200e-01 -1.21057081e+00
-9.31257188e-01 -1.98441267e-01 1.06517635e-01 4.01306190e-02
-1.53497532e-01 3.17820199e-02 -1.12714207e+00 -4.86711919e-01
1.33437380e-01 -1.02941370e+00 -4.13672686e-01 2.17936933e-01
-6.18070543e-01 4.55063842e-02 6.50516272e-01 -5.44007361e-01
1.36890137e+00 -2.37731838e+00 -1.88397557e-01 1.11528508e-01
3.50706190e-01 6.67384267e-01 -4.40406539e-02 1.20238356e-01
2.52522081e-01 7.09912367e-03 -5.98909855e-01 -4.70735878e-01
5.49782142e-02 1.71565339e-01 -4.20855641e-01 4.58163798e-01
3.83980423e-01 9.56827819e-01 -6.37222230e-01 -6.16561055e-01
3.32139879e-01 3.54604244e-01 -6.12406015e-01 1.86645269e-01
-3.97982329e-01 3.99156719e-01 -1.95612893e-01 7.78437495e-01
1.08909976e+00 -4.81422544e-01 -8.66399109e-02 -3.43770981e-01
-5.98397255e-01 3.41119051e-01 -1.48249876e+00 1.55252707e+00
-1.40625909e-01 4.44979876e-01 4.01602030e-01 -8.04260790e-01
4.95245576e-01 -1.36614680e-01 4.29743975e-01 -3.34823191e-01
1.16453683e-02 1.80685014e-01 -2.26814449e-02 -3.18638921e-01
4.91794169e-01 5.08791983e-01 2.13914290e-01 2.50262111e-01
-6.47242516e-02 -2.45020658e-01 5.88325679e-01 3.21863085e-01
8.29356313e-01 2.35985249e-01 7.89811462e-02 -2.98357666e-01
3.80829126e-01 9.14672241e-02 4.64670688e-01 1.03884280e+00
3.33495587e-02 7.42155135e-01 4.95911449e-01 -2.27448955e-01
-6.28562272e-01 -1.20233941e+00 -4.61073905e-01 1.06453085e+00
1.58026233e-01 -4.17159915e-01 -1.11240602e+00 -4.96416658e-01
1.91944689e-01 4.34186816e-01 -3.89242917e-01 4.19212103e-01
-7.10422933e-01 -9.28840280e-01 6.10762537e-01 6.39685273e-01
7.08654702e-01 -9.29825664e-01 -7.79462039e-01 1.94175944e-01
1.05934605e-01 -1.27695465e+00 -6.31482720e-01 1.99287847e-01
-7.57923841e-01 -9.75757420e-01 -4.53080744e-01 -6.35141075e-01
7.17859685e-01 5.24268031e-01 9.78554308e-01 1.99756891e-01
-7.12992072e-01 3.36339511e-02 3.68075911e-04 -3.79747093e-01
9.71773192e-02 3.16190600e-01 -2.83017248e-01 3.11640710e-01
1.07080422e-01 -4.58163351e-01 -8.83197844e-01 4.60096300e-01
-8.98745298e-01 3.91428113e-01 7.20842838e-01 4.96622622e-01
9.32640612e-01 -2.99406379e-01 -1.00209072e-01 -1.08977687e+00
-1.03458911e-01 -2.48846948e-01 -1.28519380e+00 1.52618080e-01
-4.55063134e-01 -2.02801436e-01 3.85297209e-01 -3.81003380e-01
-8.29894364e-01 3.15807194e-01 -1.22902654e-01 -3.48698199e-01
-1.64669350e-01 3.53109278e-02 5.58896102e-02 -2.13762805e-01
4.73426014e-01 3.60550433e-01 -6.20477870e-02 -6.36801124e-01
4.59265113e-01 5.68475723e-01 6.15093768e-01 -6.23421073e-01
8.67667615e-01 7.59471178e-01 -2.38843523e-02 -7.62651920e-01
-9.94858563e-01 -7.83038735e-01 -6.94445491e-01 -3.91824245e-02
8.07906687e-01 -9.87998784e-01 -9.87756908e-01 6.97747350e-01
-1.14687991e+00 -7.52873957e-01 -3.51770669e-01 1.83011129e-01
-3.00005496e-01 1.21221386e-01 -8.62947702e-01 -6.30809784e-01
-3.92105043e-01 -1.28342938e+00 1.28084862e+00 3.75079989e-01
2.72953838e-01 -4.62730676e-01 -2.62731969e-01 3.74527782e-01
4.35840428e-01 -4.05462012e-02 3.08680892e-01 -4.61430341e-01
-1.08524036e+00 3.94279063e-02 -7.08876073e-01 4.37389582e-01
-1.88028768e-01 2.75166273e-01 -1.07156432e+00 -2.64363557e-01
-9.99469757e-02 2.99710948e-02 1.18239748e+00 5.17161250e-01
1.55546033e+00 -2.37976477e-01 -4.69090432e-01 1.16758418e+00
1.48599017e+00 -2.23977473e-02 4.56373215e-01 1.12193853e-01
8.76912713e-01 3.49040538e-01 6.48105621e-01 5.81034362e-01
2.00982779e-01 7.15193272e-01 3.35409671e-01 -5.12261748e-01
-2.20727324e-01 1.73665762e-01 1.54163897e-01 5.00782609e-01
-5.96708581e-02 -1.63515106e-01 -9.79366302e-01 3.98180544e-01
-1.96718347e+00 -7.26735294e-01 -4.67207074e-01 2.23494649e+00
8.24600399e-01 2.86622524e-01 1.39042214e-01 -1.95460498e-01
6.76320255e-01 1.93559110e-01 -6.17539704e-01 -4.46488187e-02
1.04734153e-02 4.93685335e-01 9.55512702e-01 6.63815975e-01
-1.39363062e+00 1.26142907e+00 5.76651287e+00 1.11543000e+00
-1.14175856e+00 9.90989581e-02 6.60046935e-01 -3.75208437e-01
2.05899760e-01 6.99543506e-02 -1.35047638e+00 5.48844159e-01
7.09521472e-01 4.06464875e-01 6.60955906e-01 6.77539766e-01
1.81085184e-01 -3.68923396e-01 -7.59054482e-01 8.66179645e-01
-2.02203095e-01 -1.44197416e+00 -5.95558584e-02 1.24586545e-01
6.38439357e-01 5.07797360e-01 1.05342172e-01 1.70056403e-01
1.51545122e-01 -7.21276224e-01 7.92216718e-01 2.93964744e-01
7.37208366e-01 -4.91175681e-01 2.76892185e-01 3.35806489e-01
-1.45082200e+00 9.30694491e-02 -4.09905940e-01 -2.86196843e-02
3.09780389e-01 9.00175273e-01 -8.34966004e-01 2.53451288e-01
6.72113121e-01 4.83627647e-01 -6.79485500e-01 1.22621775e+00
-1.84554115e-01 1.02368414e+00 -6.58633530e-01 2.65651971e-01
4.68415082e-01 -3.28888923e-01 5.04045248e-01 1.66147351e+00
2.17665210e-02 1.84028804e-01 5.65279245e-01 1.05828762e+00
-1.17385127e-01 -3.91521864e-02 -7.95007497e-02 1.63865119e-01
4.75450426e-01 1.62743497e+00 -1.21654451e+00 -7.47509778e-01
-3.84591043e-01 1.04241490e+00 2.35916659e-01 3.13244164e-01
-1.21405232e+00 -3.80102187e-01 7.85195410e-01 1.80451453e-01
7.89513350e-01 -2.39531845e-01 -5.42402446e-01 -1.13541234e+00
8.54494944e-02 -7.41486967e-01 1.87185422e-01 -4.16491002e-01
-9.35711563e-01 3.34140986e-01 -5.14083020e-02 -1.04511476e+00
3.19698572e-01 -8.21272433e-01 -4.98044908e-01 8.75686109e-01
-1.66405094e+00 -1.02772915e+00 -3.46367091e-01 4.34187382e-01
6.03659272e-01 3.34903628e-01 4.91253823e-01 4.36447531e-01
-8.68001938e-01 4.82712477e-01 3.69671583e-02 2.66222507e-01
5.19255042e-01 -1.28567731e+00 8.71821344e-01 9.53064084e-01
3.63928050e-01 6.01415038e-01 4.24312294e-01 -4.81766194e-01
-1.37907100e+00 -1.11887586e+00 4.75860596e-01 -3.69026452e-01
6.58818185e-01 -6.71280861e-01 -8.58192742e-01 6.43092394e-01
-7.88212493e-02 3.89778405e-01 3.31939638e-01 -1.37904957e-01
-2.16223821e-01 -3.05096775e-01 -9.30148423e-01 5.13905644e-01
1.07193851e+00 -2.06878871e-01 -7.33636469e-02 6.30679548e-01
8.43844593e-01 -8.77693594e-01 -5.03481388e-01 4.10767555e-01
5.22814035e-01 -1.01325119e+00 1.03184748e+00 -1.85654595e-01
1.50060013e-01 -6.51490867e-01 -2.02015620e-02 -6.54461563e-01
-3.61043394e-01 -8.83971512e-01 -3.13100517e-01 1.28727305e+00
6.05273306e-01 -8.18547785e-01 8.12908173e-01 4.55636859e-01
-3.12289417e-01 -1.08107722e+00 -7.96574354e-01 -7.82935560e-01
-3.63308936e-01 -5.68201900e-01 5.26373863e-01 5.39781868e-01
-5.95558822e-01 6.92997202e-02 2.08371412e-02 5.70941746e-01
5.67721963e-01 5.64336002e-01 9.46695149e-01 -9.81262684e-01
-7.62767732e-01 -6.08348787e-01 -3.39212157e-02 -1.61452460e+00
-2.54805893e-01 -9.10142004e-01 7.54733533e-02 -1.36052454e+00
2.50465661e-01 -5.27355611e-01 -2.72527337e-01 6.61263227e-01
-3.09621572e-01 4.86366272e-01 5.02382219e-01 2.34059796e-01
-7.27013350e-01 -2.63044983e-02 1.17245162e+00 -1.36014983e-01
-2.47293606e-01 2.14636490e-01 -4.96818811e-01 7.80976713e-01
7.86956787e-01 -5.68266094e-01 2.67848354e-02 -6.67687535e-01
-1.73645705e-01 -2.57032096e-01 5.01878500e-01 -1.12841535e+00
3.28941405e-01 -1.77551940e-01 1.64411038e-01 -8.40638340e-01
4.82196599e-01 -5.97590387e-01 -4.19124477e-02 5.82764864e-01
1.98757928e-02 -5.16128242e-02 3.18430215e-01 1.06683739e-01
5.53499162e-02 -2.33582139e-01 8.64776909e-01 -2.20484927e-01
-6.24591291e-01 4.82517511e-01 -1.50161088e-01 1.19771056e-01
9.73793864e-01 -2.35855654e-01 -3.15582633e-01 2.62738019e-01
-5.16629696e-01 4.52458590e-01 4.51304704e-01 -8.47777650e-02
2.43756562e-01 -7.24037647e-01 -7.28313029e-01 3.48547965e-01
-2.87043184e-01 5.08856177e-01 9.93919969e-02 1.20774603e+00
-7.80647993e-01 3.82582098e-01 3.58303607e-01 -9.98260975e-01
-1.09053707e+00 4.57589865e-01 1.66127712e-01 -2.63399541e-01
-7.44035542e-01 1.16797602e+00 5.26703954e-01 -3.12015742e-01
3.08327883e-01 -5.13519764e-01 3.29590946e-01 1.43351614e-01
6.19780719e-01 6.11389935e-01 5.91692813e-02 -2.97880322e-01
-3.69838923e-01 6.56914294e-01 -1.19765781e-01 -1.89259529e-01
1.27082908e+00 1.73976511e-01 -2.35465437e-01 3.05464000e-01
8.51084173e-01 1.18041217e-01 -1.64285469e+00 -2.00171828e-01
-1.67447656e-01 -5.29329717e-01 -3.41327041e-02 -8.77102077e-01
-1.25866032e+00 9.03812349e-01 4.95557785e-01 1.24271579e-01
1.09947968e+00 6.29503950e-02 1.01994324e+00 3.48248005e-01
2.00502291e-01 -9.63176250e-01 -3.60480398e-01 4.98311937e-01
4.16828305e-01 -1.18870831e+00 7.41780400e-02 -7.73491979e-01
-3.64218771e-01 8.59008908e-01 6.82643652e-01 -2.28502691e-01
5.55790782e-01 7.19649255e-01 9.30591971e-02 -2.81591922e-01
-4.97293532e-01 -4.48822439e-01 3.80198896e-01 1.23578995e-01
1.95552707e-01 3.12545538e-01 -2.00220436e-01 3.14025104e-01
-5.00121666e-03 -1.74644500e-01 1.31316319e-01 7.89473057e-01
-6.77625060e-01 -1.00140452e+00 -4.85869795e-01 6.03532493e-01
-4.28132653e-01 -4.48451549e-01 -4.25580628e-02 7.29399502e-01
3.23106617e-01 6.77890778e-01 1.90579116e-01 -1.04389310e-01
3.56974393e-01 -8.23034346e-02 2.84165740e-01 -8.83741975e-01
-9.37705934e-01 2.18637511e-01 -1.61398232e-01 -8.91990542e-01
-2.10861653e-01 -6.30709410e-01 -1.49244809e+00 -2.13928431e-01
-5.34397840e-01 -8.62849131e-02 6.66805983e-01 7.52342641e-01
5.70503056e-01 4.03158396e-01 4.23153907e-01 -1.23339891e+00
-4.51472014e-01 -7.71988034e-01 -3.90141398e-01 9.58179459e-02
2.71642506e-01 -3.32627714e-01 -3.05067569e-01 6.78497553e-02]
|
[9.351895332336426, 0.017526468262076378]
|
e419f2bc-642b-4566-8e90-84ef7b5537fe
|
optimization-based-improvement-of-face-image
|
2305.14856
| null |
https://arxiv.org/abs/2305.14856v1
|
https://arxiv.org/pdf/2305.14856v1.pdf
|
Optimization-Based Improvement of Face Image Quality Assessment Techniques
|
Contemporary face recognition (FR) models achieve near-ideal recognition performance in constrained settings, yet do not fully translate the performance to unconstrained (realworld) scenarios. To help improve the performance and stability of FR systems in such unconstrained settings, face image quality assessment (FIQA) techniques try to infer sample-quality information from the input face images that can aid with the recognition process. While existing FIQA techniques are able to efficiently capture the differences between high and low quality images, they typically cannot fully distinguish between images of similar quality, leading to lower performance in many scenarios. To address this issue, we present in this paper a supervised quality-label optimization approach, aimed at improving the performance of existing FIQA techniques. The developed optimization procedure infuses additional information (computed with a selected FR model) into the initial quality scores generated with a given FIQA technique to produce better estimates of the "actual" image quality. We evaluate the proposed approach in comprehensive experiments with six state-of-the-art FIQA approaches (CR-FIQA, FaceQAN, SER-FIQ, PCNet, MagFace, SDD-FIQA) on five commonly used benchmarks (LFW, CFPFP, CPLFW, CALFW, XQLFW) using three targeted FR models (ArcFace, ElasticFace, CurricularFace) with highly encouraging results.
|
['Vitomir Štruc', 'Naser Damer', 'Žiga Babnik']
|
2023-05-24
| null | null | null | null |
['face-image-quality', 'face-recognition', 'image-quality-assessment', 'face-image-quality-assessment']
|
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
|
[ 1.42445311e-01 -3.24560434e-01 2.07838416e-01 -8.13086450e-01
-8.77369940e-01 -4.28377628e-01 5.90133071e-01 -3.55702758e-01
-4.17803414e-02 6.05331719e-01 2.48480197e-02 7.49647245e-02
-5.41986465e-01 -6.75391853e-01 -4.11473036e-01 -5.16964734e-01
-1.35581315e-01 6.86411142e-01 -3.11923802e-01 -3.35210800e-01
2.89891571e-01 9.81955767e-01 -2.07556462e+00 6.06212318e-01
9.71250236e-01 1.33953643e+00 -2.53289551e-01 7.24768579e-01
-2.11119398e-01 6.46927178e-01 -6.84803247e-01 -8.52250278e-01
5.39022744e-01 -4.22150910e-01 -8.49510193e-01 1.70694619e-01
1.16900325e+00 -2.13545397e-01 -7.48979300e-02 1.06378484e+00
5.15334606e-01 -5.37993610e-02 7.47524559e-01 -1.40537107e+00
-6.78073943e-01 2.37353705e-02 -3.53803217e-01 3.92047912e-01
8.92478526e-01 4.98196602e-01 8.92815888e-01 -1.18102741e+00
5.94782829e-01 1.86770272e+00 4.82358426e-01 7.15048254e-01
-1.21144998e+00 -8.17909718e-01 -2.30518788e-01 3.24997932e-01
-1.34228587e+00 -9.26469803e-01 5.04368722e-01 -2.86988229e-01
8.22989881e-01 4.20106322e-01 2.82488763e-01 8.33387375e-01
-1.92458881e-03 5.87286234e-01 1.56664002e+00 -3.29352200e-01
1.16226442e-01 4.83321026e-02 1.95905685e-01 9.06101286e-01
-3.22182119e-01 3.82884383e-01 -7.68299401e-01 -2.28257388e-01
4.86000031e-01 -3.55379075e-01 -3.26121300e-01 -1.18721768e-01
-6.53194189e-01 6.53321743e-01 5.57442546e-01 5.58108687e-01
-4.95174706e-01 -1.62348285e-01 6.26223981e-02 8.09585869e-01
3.80483299e-01 2.69929290e-01 -2.05715448e-01 -1.03510334e-03
-1.13951099e+00 2.55783617e-01 7.03469932e-01 5.33857167e-01
1.00367510e+00 2.99663335e-01 -6.22421086e-01 9.50500071e-01
4.79536593e-01 8.23630989e-01 3.10548872e-01 -1.04379678e+00
3.04255009e-01 5.10613322e-01 1.73289999e-01 -1.05363429e+00
-1.30905241e-01 -3.17334116e-01 -5.89891553e-01 7.04811037e-01
4.50837821e-01 2.92427510e-01 -1.28482997e+00 1.44343102e+00
1.77719727e-01 2.01961756e-01 6.39180094e-02 9.37391579e-01
9.54471290e-01 5.55502415e-01 3.14129502e-01 -3.13893050e-01
1.28737640e+00 -7.54221737e-01 -6.72721446e-01 9.05463323e-02
-5.55597059e-02 -9.99153674e-01 1.18780708e+00 7.54470110e-01
-1.05819762e+00 -1.01663280e+00 -9.31087554e-01 4.55139339e-01
-1.83236256e-01 2.98573673e-01 4.17743385e-01 1.31349111e+00
-1.46983695e+00 5.89024842e-01 -2.71103442e-01 -1.97055802e-01
8.79860282e-01 7.42515981e-01 -8.31323266e-01 -5.73884249e-01
-8.91931117e-01 8.58108997e-01 -5.09325275e-03 1.66784301e-01
-1.39405847e+00 -9.00743723e-01 -4.24786717e-01 6.44309148e-02
3.59137714e-01 -4.41596299e-01 1.03140366e+00 -1.65119755e+00
-1.61364388e+00 9.50827837e-01 -1.64761588e-01 -1.89310029e-01
5.28837442e-01 -1.71457902e-01 -9.11205113e-01 3.72498006e-01
-3.77694070e-01 8.42994034e-01 1.27604282e+00 -1.52440953e+00
-3.55300754e-01 -6.67984426e-01 6.54748157e-02 6.03905879e-02
-9.55301821e-02 3.70453209e-01 -2.13619024e-01 -3.24963927e-01
-3.97048056e-01 -3.77549380e-01 3.52495819e-01 3.13499868e-01
1.68004129e-02 -3.15569043e-01 8.53262722e-01 -6.60299778e-01
8.48591626e-01 -2.14143038e+00 5.64837940e-02 5.06032765e-01
2.79552788e-02 8.20688963e-01 -6.48465276e-01 1.26920760e-01
-1.05870560e-01 3.26512307e-02 -2.54330367e-01 -3.97371382e-01
-1.10988289e-01 3.62757206e-01 2.07502946e-01 4.58152801e-01
7.37024367e-01 6.86587989e-01 -6.14341259e-01 -4.79726523e-01
2.79274881e-01 7.01585770e-01 -6.44875348e-01 5.77975929e-01
-1.55719832e-01 3.16447496e-01 -1.57495037e-01 8.97752643e-01
1.03273249e+00 1.93498880e-01 -8.49040076e-02 -5.18691540e-01
3.85079235e-01 -4.72995073e-01 -1.60348225e+00 1.25058544e+00
-5.51783800e-01 3.04503143e-01 1.40853077e-01 -7.38578498e-01
1.07186973e+00 3.19008201e-01 4.91599619e-01 -9.62959588e-01
1.18245922e-01 1.66821778e-01 6.99596852e-02 -4.73831564e-01
1.10929534e-01 -3.20352048e-01 6.35331511e-01 1.68448791e-01
7.73384690e-01 1.15463458e-01 2.86545068e-01 -1.75500140e-01
9.97895598e-01 -1.43776387e-01 8.48687962e-02 -3.82506579e-01
1.19698012e+00 -5.05545616e-01 3.07752818e-01 5.19924343e-01
-5.75845540e-01 7.06676304e-01 3.01860601e-01 -3.60235959e-01
-8.86857450e-01 -1.25526464e+00 -1.48960724e-01 9.15695548e-01
-2.78045267e-01 -1.37669355e-01 -9.55664039e-01 -9.41302896e-01
5.74094504e-02 3.23618740e-01 -7.17721224e-01 -5.34112342e-02
-4.65007633e-01 -6.98132694e-01 6.38670087e-01 5.78602143e-02
6.65519357e-01 -1.36087286e+00 -3.02002162e-01 9.17030126e-03
1.14562780e-01 -9.67117906e-01 -2.52313614e-01 -4.82422858e-01
-4.84353185e-01 -1.33095419e+00 -6.07438743e-01 -4.37536359e-01
5.63491464e-01 -6.43612146e-02 1.48275244e+00 5.63492537e-01
-5.27213573e-01 5.40781260e-01 -4.84837502e-01 2.05476917e-02
-5.32923818e-01 -4.48759347e-01 8.24733675e-02 5.24550021e-01
3.16350669e-01 -3.30249444e-02 -4.55719620e-01 6.01495981e-01
-1.02957356e+00 -6.53693378e-01 5.77091277e-01 1.07290041e+00
4.03455794e-01 1.12689920e-01 5.46036363e-01 -8.86506915e-01
7.22805142e-01 -2.17584074e-01 -4.37323660e-01 6.19127512e-01
-7.48818934e-01 8.88421685e-02 5.54659486e-01 -2.37876117e-01
-1.32948792e+00 -1.29346281e-01 -6.65707588e-01 -6.38304591e-01
-3.37088257e-01 1.38094604e-01 -4.40152168e-01 -6.43495977e-01
9.11202252e-01 3.64317074e-02 3.14344853e-01 -3.95248204e-01
2.29664236e-01 6.22220218e-01 5.88724494e-01 -6.82285786e-01
7.60555506e-01 2.71104574e-01 -9.14916024e-02 -7.23530769e-01
-4.94529545e-01 -3.38057697e-01 -4.65578258e-01 -5.62945068e-01
4.51536000e-01 -6.84828103e-01 -8.25636208e-01 5.48310280e-01
-8.27672958e-01 1.07132941e-02 -2.13697463e-01 1.05119430e-01
-4.34844524e-01 3.77746046e-01 -3.89691114e-01 -1.12724185e+00
-6.19377434e-01 -1.43425965e+00 1.17818093e+00 3.30315351e-01
2.98703343e-01 -5.66474140e-01 -6.40003830e-02 6.52918518e-01
6.95059776e-01 4.78554219e-02 6.99719131e-01 -2.82038182e-01
-3.47343653e-01 8.09567720e-02 -4.34896260e-01 7.96565235e-01
8.97204652e-02 3.80782992e-01 -1.22444177e+00 -4.15089518e-01
-2.67480433e-01 -5.45078516e-01 6.11996651e-01 8.56214389e-02
9.60942149e-01 -2.57607132e-01 2.19362766e-01 3.95072311e-01
1.69483256e+00 7.87364990e-02 1.14192879e+00 -2.83467233e-01
1.95725083e-01 5.55659473e-01 8.55586588e-01 2.52742946e-01
-2.27336928e-01 8.18731070e-01 5.64312339e-01 5.49060255e-02
-6.10536456e-01 1.32680178e-01 6.22975469e-01 2.16210976e-01
-1.71883449e-01 -3.22261810e-01 -7.50947177e-01 7.71977454e-02
-1.21337306e+00 -1.05897152e+00 9.72374305e-02 1.94535065e+00
5.24891853e-01 4.10670973e-02 2.95622647e-01 5.49920082e-01
5.34242392e-01 -1.11526601e-01 -3.14493448e-01 -5.79149365e-01
-2.00374871e-01 8.14300954e-01 -1.23024866e-01 5.73108852e-01
-8.44131351e-01 8.69750559e-01 6.77736235e+00 9.99344945e-01
-1.03214157e+00 1.74512297e-01 9.62373495e-01 2.74102122e-01
-1.04358524e-01 -4.93339509e-01 -6.98314488e-01 2.88088650e-01
1.15743685e+00 1.62592754e-01 7.98599422e-01 7.37257659e-01
-4.02559042e-02 4.03581746e-02 -1.12738764e+00 1.24260712e+00
3.04588348e-01 -1.04677057e+00 3.20707738e-01 -1.31359994e-01
5.80374181e-01 -2.24441513e-01 3.91072363e-01 2.45606571e-01
7.13652894e-02 -1.64917314e+00 4.91427541e-01 7.84600139e-01
1.12579882e+00 -8.99485111e-01 9.47019279e-01 -2.04357550e-01
-1.01531076e+00 -2.84846127e-01 -5.76782584e-01 4.04203147e-01
-3.47032517e-01 6.54773176e-01 -7.60608375e-01 9.02982116e-01
7.73801267e-01 2.28608951e-01 -9.44455445e-01 9.57661331e-01
1.45041421e-01 4.59327459e-01 -1.75192595e-01 2.91245520e-01
6.29455224e-02 -1.30574163e-02 2.66354799e-01 1.02513027e+00
3.21177900e-01 9.72718745e-02 -1.28004909e-01 7.82773793e-01
-2.27999493e-01 3.39666992e-01 -1.70341447e-01 2.41952506e-03
2.17499927e-01 1.38898325e+00 -3.58089954e-01 -6.97519183e-02
-2.77903795e-01 7.68139303e-01 1.91394567e-01 1.46161551e-02
-5.21811306e-01 1.92630142e-01 9.23102200e-01 1.72836632e-01
1.25199765e-01 8.99838731e-02 2.98223168e-01 -8.54713976e-01
-1.10596552e-01 -1.58244801e+00 6.30961180e-01 -8.04898322e-01
-1.56605458e+00 1.15887535e+00 -2.13095739e-01 -9.53209639e-01
-2.31225327e-01 -9.36108947e-01 -3.60460073e-01 1.05357075e+00
-1.71892965e+00 -1.12528205e+00 -6.27591133e-01 1.01330388e+00
5.46989739e-01 -6.20560765e-01 7.91615903e-01 6.80454671e-01
-2.90323019e-01 1.02116978e+00 -1.47395521e-01 -8.41112211e-02
7.30756521e-01 -1.14650488e+00 9.98924952e-03 6.94089234e-01
3.53899568e-01 2.22910479e-01 7.13667333e-01 -3.27920139e-01
-1.60131049e+00 -8.38289917e-01 2.45751277e-01 -4.70390290e-01
2.57365555e-02 8.31369460e-02 -9.88857865e-01 -8.25178772e-02
9.80075970e-02 4.03122663e-01 6.09584212e-01 -3.06545142e-02
-6.07802629e-01 -5.94688475e-01 -1.97988391e+00 1.68433494e-03
1.03274834e+00 -5.39966702e-01 -4.03591782e-01 1.12178184e-01
1.81875713e-02 6.05565757e-02 -1.13208818e+00 5.01806021e-01
5.90918899e-01 -1.36615872e+00 1.12209129e+00 -5.78057647e-01
6.52516112e-02 -2.80458897e-01 -2.40945756e-01 -1.41469538e+00
-2.23827869e-01 -2.64910668e-01 -2.43724920e-02 1.32524812e+00
9.67396796e-02 -3.51474792e-01 8.79561186e-01 3.88614804e-01
1.48809940e-01 -5.82160711e-01 -1.23849046e+00 -7.38457322e-01
-1.67293400e-01 -3.14373016e-01 1.00067186e+00 5.38663983e-01
-6.75642312e-01 -1.12239487e-01 -3.02534521e-01 2.18124226e-01
7.12626994e-01 -2.49288902e-01 7.05754638e-01 -1.42000568e+00
-2.82562077e-01 -4.32779402e-01 -8.07066321e-01 -2.17042997e-01
1.79215714e-01 -6.73053205e-01 -1.14636105e-02 -9.96076584e-01
1.19348466e-01 -4.82767493e-01 -3.34894896e-01 4.16204363e-01
-1.43160775e-01 6.82372570e-01 4.00409818e-01 6.91363439e-02
-4.36680019e-01 4.19903189e-01 1.35461569e+00 -1.59097791e-01
3.46807450e-01 -2.01545745e-01 -3.43050987e-01 9.53845233e-02
4.16204572e-01 -1.41893998e-01 -3.49143445e-01 -1.82924986e-01
-1.59561932e-01 1.39861286e-01 3.04274499e-01 -1.47013724e+00
-1.57796681e-01 -4.42301407e-02 6.87141120e-01 -5.68663329e-02
3.67863715e-01 -9.77229059e-01 5.01322091e-01 4.54410404e-01
-7.02145994e-02 1.67170972e-01 9.38255638e-02 1.63752839e-01
-5.15796423e-01 -3.61612797e-01 1.26146674e+00 -5.66179911e-03
-7.72619724e-01 5.47396421e-01 1.07052594e-01 -2.40473673e-01
8.17613840e-01 -2.85311133e-01 -3.84916902e-01 -3.81374359e-01
-7.30947435e-01 3.11078709e-02 3.06600809e-01 4.88764912e-01
9.23488140e-01 -1.29068220e+00 -1.05576789e+00 6.07941985e-01
1.91318929e-01 -7.25764334e-01 4.78302777e-01 4.39839125e-01
-5.85434258e-01 2.41570562e-01 -9.13668692e-01 -5.75371325e-01
-1.69007957e+00 5.85712433e-01 7.20420003e-01 -3.38023692e-01
2.42956411e-02 8.65127742e-01 -1.44961715e-01 -3.45093817e-01
1.40195027e-01 1.99730530e-01 -4.59708929e-01 -1.30555658e-02
8.21448922e-01 3.36999893e-01 6.12037420e-01 -1.07572126e+00
-3.45067352e-01 5.92970312e-01 1.63830481e-02 9.01775956e-02
1.21739066e+00 2.19047219e-01 1.39340702e-02 -2.57553399e-01
1.25583708e+00 -1.89143643e-01 -1.16063154e+00 -5.46215698e-02
4.48964117e-03 -1.01335895e+00 1.02420256e-01 -1.23022616e+00
-1.54162204e+00 8.21344376e-01 1.51060581e+00 1.41532281e-02
1.67782736e+00 -2.54221886e-01 2.71575034e-01 7.86083043e-02
5.09326398e-01 -8.27877462e-01 5.31959534e-01 -3.45664434e-02
1.12024295e+00 -1.29818761e+00 -1.26483575e-01 -5.00070572e-01
-5.34310460e-01 1.30553794e+00 6.84015691e-01 -6.40951991e-02
6.20354295e-01 3.95524390e-02 7.64185935e-02 -3.51429254e-01
-6.19366944e-01 -2.57385254e-01 7.50542819e-01 7.59241223e-01
2.88788736e-01 -6.03560917e-02 -1.68681756e-01 1.58856064e-01
7.05765337e-02 1.75078705e-01 1.52682886e-01 5.55104494e-01
-3.81047904e-01 -1.40477133e+00 -6.79329336e-01 7.61599243e-01
-6.04226172e-01 3.29473376e-01 -3.88904393e-01 6.05614603e-01
3.81564647e-01 1.17434633e+00 -1.36498019e-01 -4.55010206e-01
5.97355485e-01 1.87993184e-01 8.42288911e-01 -3.93223554e-01
-1.06506634e+00 -1.82391673e-01 1.10766619e-01 -8.08452785e-01
-6.79040492e-01 -6.54798985e-01 -7.37623990e-01 -5.23082852e-01
-3.31325710e-01 1.70167014e-01 6.59271240e-01 8.24153602e-01
1.45578131e-01 2.51699984e-01 8.46876502e-01 -5.69898248e-01
-5.38902938e-01 -9.09711480e-01 -4.60385203e-01 7.96281695e-01
1.94019794e-01 -9.52024162e-01 -2.41005108e-01 -1.27107367e-01]
|
[13.057344436645508, 0.752301812171936]
|
7c1d67ff-8b0d-4fd0-abdb-49407ca7225c
|
differentiable-digital-signal-processing
|
2202.00200
| null |
https://arxiv.org/abs/2202.00200v1
|
https://arxiv.org/pdf/2202.00200v1.pdf
|
Differentiable Digital Signal Processing Mixture Model for Synthesis Parameter Extraction from Mixture of Harmonic Sounds
|
A differentiable digital signal processing (DDSP) autoencoder is a musical sound synthesizer that combines a deep neural network (DNN) and spectral modeling synthesis. It allows us to flexibly edit sounds by changing the fundamental frequency, timbre feature, and loudness (synthesis parameters) extracted from an input sound. However, it is designed for a monophonic harmonic sound and cannot handle mixtures of harmonic sounds. In this paper, we propose a model (DDSP mixture model) that represents a mixture as the sum of the outputs of multiple pretrained DDSP autoencoders. By fitting the output of the proposed model to the observed mixture, we can directly estimate the synthesis parameters of each source. Through synthesis parameter extraction experiments, we show that the proposed method has high and stable performance compared with a straightforward method that applies the DDSP autoencoder to the signals separated by an audio source separation method.
|
['Kazunobu Kondo', 'Yu Takahashi', 'Hiroshi Saruwatari', 'Daichi Kitamura', 'Tomohiko Nakamura', 'Masaya Kawamura']
|
2022-02-01
| null | null | null | null |
['audio-source-separation']
|
['audio']
|
[ 3.78098041e-02 -2.50490248e-01 4.91898984e-01 9.28374827e-02
-4.13565308e-01 -7.10495412e-01 4.50438678e-01 -4.21540141e-01
-8.46173987e-03 2.87537545e-01 3.44286561e-01 3.37568782e-02
1.38389561e-02 -7.92123139e-01 -7.87967503e-01 -8.80718350e-01
3.67564112e-01 9.31682661e-02 -6.57525435e-02 -2.47482792e-01
-3.12269926e-01 4.44243491e-01 -1.76699436e+00 2.62539327e-01
7.08668649e-01 9.76255655e-01 2.54096746e-01 1.14162171e+00
-1.66793212e-01 6.26946449e-01 -1.14808238e+00 -2.00826511e-01
3.63634378e-01 -9.35516357e-01 -1.64839521e-01 -1.61334306e-01
2.96655118e-01 -3.68605286e-01 -4.55152839e-01 1.18603659e+00
6.39872074e-01 3.72693956e-01 8.35059345e-01 -9.57037866e-01
-7.88523734e-01 1.15162373e+00 1.09949812e-01 -3.02468017e-02
1.45441085e-01 -1.05853461e-01 9.47812378e-01 -7.08013773e-01
4.92043793e-02 1.22608578e+00 6.62091732e-01 3.78298908e-01
-1.30899131e+00 -8.53887677e-01 -4.37924206e-01 -8.34713578e-02
-1.16817176e+00 -3.69634271e-01 1.29387760e+00 -4.83498067e-01
8.18354785e-01 3.70813370e-01 6.59665227e-01 1.11534452e+00
2.82594889e-01 5.77581465e-01 6.35013759e-01 -7.60507643e-01
2.38101274e-01 -7.94649683e-03 -2.98546344e-01 2.83054590e-01
-5.11451840e-01 3.44207585e-01 -4.72289056e-01 -2.61545807e-01
1.01941001e+00 -2.16658726e-01 -4.21411037e-01 1.41074330e-01
-1.12945735e+00 5.67263305e-01 2.35028654e-01 5.21478593e-01
-5.57862997e-01 3.27671975e-01 2.25991786e-01 6.08968437e-01
8.29192400e-02 5.80593705e-01 -3.05112153e-01 -2.14816518e-02
-1.04615450e+00 2.30091006e-01 9.55028951e-01 4.64608103e-01
2.19435081e-01 9.77546811e-01 4.37687933e-02 1.04980433e+00
1.23280950e-01 6.80234075e-01 1.00508273e+00 -1.30074477e+00
-1.18045591e-01 -1.41921476e-01 4.72199731e-02 -9.24127221e-01
-1.95360735e-01 -4.12854582e-01 -8.47521305e-01 3.80904198e-01
2.66450047e-01 -4.96733457e-01 -6.33175135e-01 1.68934345e+00
3.47726084e-02 4.75599051e-01 3.63139540e-01 7.32650161e-01
1.08064854e+00 1.10215640e+00 -2.78803557e-01 -2.32990041e-01
9.52068448e-01 -9.44852769e-01 -1.15292716e+00 2.10325420e-01
-3.88397634e-01 -1.01055288e+00 9.89761770e-01 6.69924796e-01
-1.20428753e+00 -1.19610131e+00 -1.38016260e+00 4.99857217e-02
-2.58311003e-01 3.22176248e-01 1.79928139e-01 6.87345326e-01
-7.70226896e-01 9.66554523e-01 -6.39690757e-01 3.10471028e-01
-5.38416505e-01 2.98818290e-01 -5.26656397e-03 6.84001267e-01
-1.50095987e+00 5.00082493e-01 4.06111032e-01 -9.19054523e-02
-7.87374973e-01 -8.90390098e-01 -6.67496145e-01 4.13593590e-01
-1.99229538e-01 -7.45714009e-01 1.54317498e+00 -1.39883840e+00
-2.56086731e+00 5.48422746e-02 2.85572261e-01 -4.55365330e-01
7.16625676e-02 -3.29097778e-01 -9.76281226e-01 6.84384853e-02
-5.32951176e-01 5.10521829e-01 1.48423064e+00 -9.55880463e-01
-4.93826210e-01 2.05402151e-01 -4.67932492e-01 7.87619594e-03
-2.58935183e-01 -1.35719329e-01 2.17200205e-01 -1.08137119e+00
1.16319919e-03 -5.95837414e-01 2.49549881e-01 -4.79932010e-01
-3.98219377e-01 -2.74761356e-02 6.17186248e-01 -8.87059867e-01
1.04732263e+00 -2.74264884e+00 5.12654543e-01 -1.42161585e-02
-1.43219247e-01 3.26372087e-01 -2.62694389e-01 4.09787208e-01
-1.33183107e-01 -3.36551338e-01 -8.01254585e-02 -3.75668168e-01
2.94510096e-01 2.43835803e-02 -7.57847369e-01 -5.97399436e-02
5.16704917e-02 3.92393947e-01 -8.02678466e-01 3.18905488e-02
2.51609534e-01 9.59786713e-01 -5.87566972e-01 5.28959692e-01
-3.29192668e-01 1.50756285e-01 4.10615385e-01 1.51313066e-01
5.29275417e-01 5.01776099e-01 1.06025323e-01 -5.27056277e-01
-1.27390087e-01 2.46787980e-01 -1.45458555e+00 1.34944975e+00
-7.03717470e-01 7.13414729e-01 1.53992355e-01 -4.98431325e-01
1.26476884e+00 7.61535823e-01 3.62904638e-01 -1.85537077e-02
3.29352081e-01 4.93872762e-01 2.97550797e-01 -3.80615801e-01
4.29410398e-01 -5.83798945e-01 1.33786395e-01 3.45282316e-01
8.26021791e-01 -5.28910756e-01 -1.70602515e-01 -5.54510832e-01
6.16891086e-01 8.29400588e-03 1.04297899e-01 9.25969630e-02
3.44158798e-01 -7.06861258e-01 3.98919523e-01 5.27533710e-01
1.94396809e-01 7.47616887e-01 5.41017801e-02 -2.75244832e-01
-9.66407478e-01 -1.47247827e+00 1.77428246e-01 1.17336404e+00
-2.50966996e-01 -2.02040493e-01 -8.66650581e-01 2.36329913e-01
3.77665535e-02 9.50461090e-01 -2.44225413e-01 -4.50777441e-01
-4.38534051e-01 -4.76728857e-01 1.00874126e+00 4.21854049e-01
3.63824785e-01 -1.28269565e+00 -3.00294489e-01 5.53045511e-01
-2.38112912e-01 -7.95738995e-01 -7.12940991e-01 3.03129405e-01
-4.47058678e-01 -4.52362597e-01 -7.67098069e-01 -1.03348196e+00
-3.03071320e-01 -1.79529145e-01 7.43617237e-01 -6.54156923e-01
1.85751021e-02 2.99320787e-01 -7.48205259e-02 -5.94653845e-01
-9.78054881e-01 -1.92421660e-01 5.48146427e-01 5.24212003e-01
4.24757786e-02 -1.16733468e+00 -2.64750063e-01 -1.93547547e-01
-1.12880468e+00 -4.25889827e-02 3.45457524e-01 4.57753211e-01
5.62796652e-01 5.96363723e-01 8.51683915e-01 -2.88397372e-01
1.27184701e+00 -3.38194281e-01 -3.61726224e-01 2.33076066e-02
-3.66991162e-02 3.27913910e-02 1.09109926e+00 -1.03682029e+00
-1.11824226e+00 2.90925521e-02 -4.00145590e-01 -8.53547633e-01
-3.49277407e-01 3.83901566e-01 -2.34735683e-01 2.02189267e-01
7.39642143e-01 3.90502959e-01 1.24991961e-01 -8.03031802e-01
5.32671750e-01 9.64829326e-01 1.24991024e+00 -3.48576933e-01
5.65778136e-01 4.45551798e-02 -3.99024725e-01 -1.02131522e+00
-4.04122442e-01 5.06895185e-02 -4.90338534e-01 -2.45091081e-01
6.69263959e-01 -6.88066542e-01 -7.34143555e-01 7.56985962e-01
-1.25915909e+00 -3.31041723e-01 -4.78551418e-01 8.08698177e-01
-7.26070106e-01 1.11664437e-01 -9.70134199e-01 -7.96707094e-01
-4.10491526e-01 -9.08886373e-01 6.04185224e-01 3.94353837e-01
-2.10514590e-01 -7.79121280e-01 4.20333445e-01 -3.00564647e-01
5.31735957e-01 1.92267507e-01 9.68787551e-01 -4.14612055e-01
8.56877491e-02 -2.36752033e-01 6.65714562e-01 1.04884481e+00
5.60059965e-01 6.01125836e-01 -1.12582970e+00 1.02938995e-01
5.26671112e-01 6.34858608e-02 7.62485385e-01 5.39546192e-01
1.01678872e+00 -4.89630282e-01 4.78365391e-01 7.38377571e-01
1.12086129e+00 6.01930141e-01 5.47939241e-01 -1.13503367e-01
5.20832956e-01 2.34542593e-01 -2.58157909e-01 3.89477134e-01
-2.74308957e-02 4.35059637e-01 4.47608577e-03 2.23807003e-02
-4.09736454e-01 -4.00479525e-01 7.40749896e-01 1.35793102e+00
-2.41486311e-01 -2.12935001e-01 -2.85663873e-01 2.15535119e-01
-1.40292907e+00 -1.19482708e+00 1.91698059e-01 2.01691532e+00
9.55287635e-01 6.90422952e-02 3.55285734e-01 7.29282796e-01
7.47796655e-01 2.46209744e-02 -4.63029593e-01 -8.11930895e-01
-1.33818671e-01 7.04784334e-01 -1.76337406e-01 7.31719613e-01
-9.90973532e-01 6.57099485e-01 6.69700384e+00 7.48365760e-01
-1.33984077e+00 -1.96511060e-01 -2.34593779e-01 -2.05712482e-01
-2.81530499e-01 -5.99886537e-01 -2.24270672e-01 5.50525486e-01
1.09255505e+00 -2.38801405e-01 1.10140157e+00 5.67127466e-01
-8.86829495e-02 4.44233090e-01 -1.08261788e+00 1.13890648e+00
5.68100438e-02 -1.06611931e+00 2.72106141e-01 -5.03833354e-01
5.69134176e-01 -3.44742388e-01 6.53383970e-01 3.27191889e-01
1.75115272e-01 -9.80008185e-01 9.48940814e-01 7.63208091e-01
5.17169297e-01 -9.63010192e-01 4.19575721e-01 5.07025957e-01
-9.12144661e-01 -2.06518516e-01 -2.65879750e-01 -2.06628889e-01
-1.98983029e-02 4.70630705e-01 -6.90486789e-01 1.73661634e-01
5.49930692e-01 2.03997344e-01 1.19025253e-01 9.92058098e-01
-1.40879005e-01 9.03464496e-01 -2.82120645e-01 8.20379630e-02
-7.21061379e-02 -3.03644925e-01 7.35578656e-01 1.30923378e+00
6.65271759e-01 8.44341293e-02 -2.82012850e-01 1.19042099e+00
-1.67398006e-01 -6.99269725e-03 -1.69013038e-01 -3.66189569e-01
5.93225896e-01 1.02044272e+00 -2.28806168e-01 -2.56060392e-01
-9.96485259e-03 1.04148471e+00 -2.36546800e-01 5.91494858e-01
-5.93171179e-01 -7.70249844e-01 7.00733602e-01 -3.97466213e-01
5.08335769e-01 -1.46929950e-01 3.35133411e-02 -9.08129454e-01
-4.17482585e-01 -1.07621646e+00 3.92655395e-02 -1.00602961e+00
-1.34229541e+00 9.76784348e-01 -2.95744449e-01 -1.30818367e+00
-7.95356929e-01 -5.41560829e-01 -7.31876493e-01 1.04013610e+00
-9.23299909e-01 -6.89310253e-01 1.56248761e-02 4.74758327e-01
5.02419949e-01 -3.90919954e-01 1.24058390e+00 1.60822868e-01
-2.37408489e-01 4.80105519e-01 4.10940140e-01 1.01387508e-01
5.19394994e-01 -1.37142324e+00 2.81873733e-01 4.44801301e-01
4.61801410e-01 3.98834884e-01 1.04240441e+00 -1.06146269e-01
-9.27112460e-01 -8.61994207e-01 7.15276122e-01 5.27457483e-02
5.44719279e-01 -2.05765292e-01 -9.98831570e-01 4.85208750e-01
4.93129253e-01 -3.39183390e-01 1.14293194e+00 -3.64974111e-01
-3.90223324e-01 -2.89618015e-01 -9.77671862e-01 4.71590519e-01
2.83530980e-01 -7.73425937e-01 -1.05050015e+00 -1.51643932e-01
1.01312244e+00 -5.03319502e-01 -9.63573694e-01 2.47197792e-01
7.12721527e-01 -1.06444287e+00 1.09451711e+00 -4.21393573e-01
5.97071946e-01 -4.85069066e-01 -3.46827477e-01 -1.90819252e+00
-6.84788823e-01 -6.84788704e-01 -5.05091667e-01 1.17022097e+00
1.73415154e-01 -3.40431660e-01 -8.93719941e-02 2.45293081e-02
-9.96957794e-02 9.52406414e-03 -6.35758579e-01 -9.03294086e-01
1.49101168e-01 -4.19146329e-01 1.12739766e+00 8.36217284e-01
1.42025873e-02 3.13435167e-01 -4.77334321e-01 4.86188322e-01
4.94710594e-01 3.05691481e-01 5.54234505e-01 -1.25431061e+00
-9.86131012e-01 -6.84702754e-01 -8.54307339e-02 -7.76636541e-01
1.88026756e-01 -6.51772320e-01 2.00933799e-01 -9.67716992e-01
-6.20361269e-01 2.24525005e-01 -5.34801841e-01 3.95515561e-02
-8.81209895e-02 1.13414384e-01 4.24399167e-01 -6.09278493e-02
3.23087752e-01 7.70089328e-01 1.01390135e+00 -2.65938789e-01
-7.37226605e-01 4.49048996e-01 -3.01652044e-01 8.52631032e-01
9.10050511e-01 -3.16030592e-01 -3.52466494e-01 -3.96084666e-01
-2.05261856e-02 4.31258261e-01 2.32253119e-01 -1.44953525e+00
2.28408605e-01 6.04092628e-02 4.54267114e-01 -3.27980340e-01
4.86303180e-01 -6.73218846e-01 7.53618419e-01 3.27941149e-01
-4.93248701e-01 -3.76046598e-01 2.71082193e-01 3.79916579e-01
-5.38312674e-01 -3.45200479e-01 7.06683517e-01 -1.13167297e-02
-2.22869247e-01 -2.75691390e-01 -6.49659872e-01 -5.52398086e-01
3.26308310e-01 1.50568247e-01 3.30661424e-02 -6.78911090e-01
-9.91924524e-01 -6.29677057e-01 -1.10768624e-01 4.43835855e-01
5.58742702e-01 -1.77541542e+00 -6.21606529e-01 5.88421106e-01
-5.82792103e-01 -2.48794034e-01 2.55081981e-01 1.64769575e-01
-4.93931532e-01 3.70039605e-02 -4.76110786e-01 -4.57975455e-02
-8.57030511e-01 7.25572646e-01 9.48514223e-01 4.49491352e-01
-5.03318310e-01 8.78224969e-01 1.21077895e-02 -5.97377956e-01
4.63405639e-01 -5.88669479e-01 -2.86674649e-01 3.09676304e-02
7.55400062e-01 5.84882855e-01 -7.23695531e-02 -4.45682049e-01
1.01654544e-01 3.42825085e-01 5.78202128e-01 -5.45834661e-01
1.40894318e+00 2.13596776e-01 -1.45291924e-01 9.98277307e-01
1.17903769e+00 3.36864829e-01 -1.05493176e+00 -1.05209090e-01
-5.32025516e-01 4.30079885e-02 2.53139347e-01 -7.96127260e-01
-1.07029903e+00 8.25500131e-01 5.11915267e-01 6.29042864e-01
1.24269438e+00 -2.41477683e-01 9.94370341e-01 3.20432782e-01
-2.24343702e-01 -1.19389582e+00 2.05387324e-01 4.16902423e-01
1.17058432e+00 -5.05202770e-01 -7.66368628e-01 1.25654610e-02
-4.92033958e-01 1.56660128e+00 2.88080066e-01 -5.22588015e-01
1.00377059e+00 5.65509558e-01 3.02986979e-01 3.17367643e-01
-5.10713816e-01 1.84566006e-01 4.97965932e-01 5.72132349e-01
2.39835784e-01 3.63646090e-01 2.46810272e-01 1.33014417e+00
-9.77769196e-01 -5.86227365e-02 6.24701202e-01 1.43124461e-01
-5.13122320e-01 -8.61532688e-01 -8.41819584e-01 7.95072019e-02
-4.07787412e-01 -4.75289747e-02 -3.93323302e-01 1.49320185e-01
4.65143412e-01 8.88939917e-01 4.51607853e-01 -8.42362106e-01
4.48811620e-01 5.08267879e-01 6.24948561e-01 -2.69567341e-01
-5.62861323e-01 4.96601492e-01 -5.30549645e-01 3.57416086e-02
-3.30841601e-01 -3.47356230e-01 -1.23727608e+00 -8.15779716e-02
-1.26690596e-01 1.66184142e-01 7.34011292e-01 6.37876868e-01
1.90694910e-03 1.03921711e+00 8.49036992e-01 -1.02476656e+00
-7.17003167e-01 -1.13568830e+00 -1.13180578e+00 1.51326016e-01
8.43229532e-01 -2.08805442e-01 -7.49859035e-01 5.11415422e-01]
|
[15.53277587890625, 5.937432765960693]
|
44e2e515-cd42-4da8-af04-54175dfa8419
|
astbert-enabling-language-model-for-code
|
2201.07984
| null |
https://arxiv.org/abs/2201.07984v4
|
https://arxiv.org/pdf/2201.07984v4.pdf
|
AstBERT: Enabling Language Model for Financial Code Understanding with Abstract Syntax Trees
|
Using the pre-trained language models to understand source codes has attracted increasing attention from financial institutions owing to the great potential to uncover financial risks. However, there are several challenges in applying these language models to solve programming language-related problems directly. For instance, the shift of domain knowledge between natural language (NL) and programming language (PL) requires understanding the semantic and syntactic information from the data from different perspectives. To this end, we propose the AstBERT model, a pre-trained PL model aiming to better understand the financial codes using the abstract syntax tree (AST). Specifically, we collect a sheer number of source codes (both Java and Python) from the Alipay code repository and incorporate both syntactic and semantic code knowledge into our model through the help of code parsers, in which AST information of the source codes can be interpreted and integrated. We evaluate the performance of the proposed model on three tasks, including code question answering, code clone detection and code refinement. Experiment results show that our AstBERT achieves promising performance on three different downstream tasks.
|
['Zhen Huang', 'Yuze Liu', 'Yujie Lu', 'Tiehua Zhang', 'Xin Chen', 'Rong Liang']
|
2022-01-20
| null | null | null | null |
['code-search', 'code-search']
|
['computer-code', 'computer-vision']
|
[-1.84215024e-01 1.77850798e-01 -2.51972973e-01 -5.16800344e-01
-8.12651992e-01 -7.94454098e-01 1.25584111e-01 4.60972428e-01
1.20860830e-01 -5.01751155e-02 2.07266331e-01 -7.84478724e-01
2.33993918e-01 -6.85905337e-01 -7.19178617e-01 1.35452405e-01
5.83657287e-02 -2.57097065e-01 3.59213024e-01 -2.06185073e-01
8.12138498e-01 -1.22197665e-01 -1.20292866e+00 7.30592012e-01
1.50784886e+00 6.99190974e-01 4.34769392e-01 6.25556350e-01
-9.62210059e-01 1.42728448e+00 -2.71355599e-01 -1.04882812e+00
-3.34867393e-03 -3.00324172e-01 -1.01645851e+00 -3.42360228e-01
-7.55142123e-02 -2.74676252e-02 2.74401903e-01 1.51504648e+00
-2.06086472e-01 -6.00291371e-01 2.72728950e-01 -1.17372561e+00
-1.11668193e+00 1.14354825e+00 -6.70221865e-01 5.89224584e-02
4.65580672e-01 -3.76906730e-02 1.27520585e+00 -9.64796543e-01
4.54516023e-01 1.24149275e+00 7.59711802e-01 4.25007850e-01
-8.18617463e-01 -5.07405221e-01 1.76028937e-01 2.06059277e-01
-1.09806514e+00 -1.89630389e-02 7.49785960e-01 -1.10728729e+00
1.16334486e+00 -1.47849480e-02 2.05256432e-01 5.19683540e-01
2.52998054e-01 8.00001860e-01 6.54387057e-01 -7.39274442e-01
-4.79762480e-02 6.87896132e-01 7.21035302e-01 1.05152726e+00
1.81710437e-01 -2.81310737e-01 -3.55074108e-01 -3.27671170e-01
1.38506457e-01 -6.62683025e-02 -2.29821786e-01 -2.83402264e-01
-6.61147237e-01 1.05864251e+00 2.23661408e-01 4.61065829e-01
1.79012567e-01 1.20752834e-01 5.83002925e-01 3.16794276e-01
3.02591175e-01 5.51245987e-01 -8.74298692e-01 -3.13055694e-01
-4.92387772e-01 9.76576209e-02 1.09472585e+00 1.37089205e+00
8.53660643e-01 -2.43971065e-01 2.41746426e-01 8.37202251e-01
8.83069694e-01 1.87488452e-01 5.00282347e-01 -6.19344950e-01
1.00726044e+00 1.24305427e+00 -2.04084918e-01 -1.16140234e+00
-3.79874595e-02 -3.34336042e-01 -1.89784020e-01 -2.50614770e-02
3.40291828e-01 4.69077267e-02 -2.54168332e-01 1.42031276e+00
-2.39528101e-02 -2.87489116e-01 2.62731284e-01 3.71786267e-01
8.75961781e-01 5.97090960e-01 1.48798943e-01 2.49115631e-01
1.46184158e+00 -1.19859326e+00 -3.01556855e-01 -4.53077823e-01
1.10570765e+00 -8.12119901e-01 1.13794231e+00 1.48942143e-01
-7.55635202e-01 -6.49186015e-01 -8.62593114e-01 -4.10831183e-01
-3.77462178e-01 3.17821443e-01 5.04420519e-01 8.76182020e-01
-6.85999632e-01 1.51617840e-01 -8.11055839e-01 -2.90585637e-01
3.03499669e-01 -2.25569323e-01 5.08236513e-02 -1.35624036e-01
-8.38902414e-01 4.48277116e-01 4.75508749e-01 -9.10091698e-02
-5.03682733e-01 -9.70508337e-01 -1.18393254e+00 4.24565732e-01
3.78589660e-01 -1.84993118e-01 1.44175494e+00 -1.18144190e+00
-1.27083015e+00 8.52391183e-01 -2.12723196e-01 -2.54885018e-01
1.60175204e-01 -4.21438456e-01 -1.72119439e-01 -1.28122985e-01
2.00176448e-01 -7.27873594e-02 4.36063319e-01 -1.08327878e+00
-6.99548721e-01 -2.53700942e-01 3.48374188e-01 -3.12418252e-01
-4.40118194e-01 8.30694020e-01 -3.81918103e-01 -6.75416350e-01
-1.13825157e-01 -6.63403928e-01 -6.13228083e-02 -1.29885212e-01
-2.50252306e-01 -3.38132769e-01 3.84221971e-01 -1.22902393e+00
1.47204280e+00 -2.33011818e+00 2.08920255e-01 1.17767304e-01
3.22980344e-01 5.15430868e-02 -9.91430879e-02 3.05202574e-01
-1.04515754e-01 4.89146978e-01 -5.37504077e-01 -4.53456454e-02
1.19883381e-01 2.84751449e-02 -5.26542962e-01 -1.34514719e-01
3.83900821e-01 8.87825966e-01 -7.79821277e-01 -4.61393178e-01
-3.92353565e-01 -5.11938371e-02 -1.15560079e+00 6.08303905e-01
-5.98937750e-01 2.97643304e-01 -8.42437387e-01 7.29553938e-01
7.58889496e-01 -2.33208671e-01 1.97753400e-01 4.74407583e-01
-3.68628502e-01 5.81166029e-01 -8.47612381e-01 1.82907975e+00
-6.38620138e-01 3.93530846e-01 -6.38554245e-02 -1.02072763e+00
1.08819616e+00 1.06766693e-01 -4.10588235e-02 -4.97225255e-01
-1.96034923e-01 4.19068664e-01 3.95276621e-02 -1.13929021e+00
2.80324250e-01 3.11627910e-02 -4.67922360e-01 5.35047710e-01
3.13971341e-02 6.96049258e-03 2.35434145e-01 3.49916041e-01
1.08761346e+00 2.92399853e-01 3.88427794e-01 -3.97795141e-01
1.11588609e+00 3.38043511e-01 6.88647211e-01 5.90891004e-01
1.82667732e-01 2.45179296e-01 1.12932372e+00 -4.11924541e-01
-7.34206378e-01 -7.17099905e-01 -3.47167887e-02 1.38238096e+00
-1.22130990e-01 -8.82306576e-01 -8.91722083e-01 -1.10587823e+00
-9.77479368e-02 7.44395852e-01 -3.55459988e-01 -7.62424469e-02
-7.82170773e-01 -5.48347235e-01 7.11576939e-01 4.98788685e-01
4.09783065e-01 -8.31924081e-01 -4.46562856e-01 1.35327086e-01
-2.25435585e-01 -8.63808692e-01 -3.96470606e-01 -1.66007504e-02
-6.35499954e-01 -1.38733637e+00 -2.27776349e-01 -1.01379442e+00
7.12303638e-01 -2.03173354e-01 1.27066708e+00 6.86475754e-01
-1.54560208e-01 1.95356593e-01 -7.13666320e-01 -4.51030344e-01
-1.11837387e+00 3.31923932e-01 -8.35118711e-01 -1.99555591e-01
5.99523485e-01 -2.13506594e-01 1.46898672e-01 -8.34908709e-02
-9.30520952e-01 1.91708639e-01 5.64852893e-01 3.54926109e-01
-1.40824065e-01 -1.58146992e-01 3.55532080e-01 -1.34591758e+00
5.72314680e-01 -7.50352800e-01 -9.51255858e-01 6.78145885e-01
-5.06551027e-01 3.99735332e-01 8.41478944e-01 3.87958549e-02
-1.45196986e+00 -2.24566191e-01 -4.24018115e-01 3.25292826e-01
5.57209644e-03 1.21270013e+00 -3.26330453e-01 6.53705001e-02
6.05309844e-01 2.56142199e-01 -3.17227721e-01 -8.33499551e-01
2.14511573e-01 7.62136638e-01 4.24658865e-01 -1.03044844e+00
8.66408288e-01 -3.50431018e-02 -4.87033635e-01 -1.95303068e-01
-8.14750552e-01 -3.66643846e-01 -7.39878595e-01 2.39351645e-01
8.82062674e-01 -7.45850444e-01 -3.68178070e-01 6.10997379e-01
-1.59198236e+00 1.85336042e-02 2.33862951e-01 1.59613729e-01
-2.39198968e-01 6.32394850e-01 -7.44641304e-01 -5.79691648e-01
-1.72478944e-01 -1.42401981e+00 8.17821026e-01 1.71515763e-01
-9.55645926e-03 -1.02397108e+00 2.05738604e-01 3.83600384e-01
3.81628752e-01 -1.44610807e-01 1.92442405e+00 -7.11329043e-01
-8.83540273e-01 2.19657389e-03 -5.12507975e-01 5.04211068e-01
-3.33001912e-02 1.44303799e-01 -6.07930601e-01 1.42248631e-01
2.98772395e-01 -2.14399830e-01 4.81415093e-01 -3.34980845e-01
1.25044525e+00 -4.57447976e-01 -2.19059438e-02 6.32501006e-01
1.61314154e+00 1.86320052e-01 5.20227551e-01 5.03388405e-01
7.86437690e-01 7.46391177e-01 2.30699778e-01 4.15470093e-01
8.35412800e-01 2.97637790e-01 2.50243664e-01 4.01420921e-01
1.08154222e-01 -4.50943828e-01 6.17356122e-01 1.38177896e+00
2.31294021e-01 3.31634372e-01 -1.55983627e+00 4.94712085e-01
-1.77151477e+00 -4.69680846e-01 -5.68407178e-01 1.75802994e+00
1.09888768e+00 -1.88389868e-02 -2.73118794e-01 -3.42532009e-01
5.72501063e-01 -1.84061199e-01 -2.95250535e-01 -5.86838484e-01
3.49951506e-01 -7.81545714e-02 8.40661153e-02 3.44817042e-01
-7.77167499e-01 7.10489750e-01 5.31303406e+00 5.21560252e-01
-8.49740386e-01 2.49597058e-01 3.19877863e-01 7.38226175e-01
-5.88634253e-01 5.91925621e-01 -8.89101028e-01 5.26232302e-01
1.02733529e+00 -6.20956719e-01 4.95082647e-01 1.40353632e+00
-1.92201734e-01 -1.30694080e-02 -1.31412733e+00 5.53732514e-01
1.25352755e-01 -1.26237953e+00 -6.81445152e-02 -3.56995344e-01
6.19136930e-01 -2.56154705e-02 -3.27415526e-01 8.16548765e-01
4.32057351e-01 -7.98462570e-01 1.11417496e+00 5.79534411e-01
3.15247864e-01 -5.80329835e-01 8.64238620e-01 7.02195168e-01
-1.37281859e+00 -4.34390217e-01 -3.97105187e-01 -1.23110652e-01
-2.87522584e-01 5.45510650e-01 -5.92999041e-01 7.62181938e-01
7.14278519e-01 1.00146973e+00 -1.08127248e+00 9.26342189e-01
-4.58619744e-01 5.98320127e-01 2.42897779e-01 -8.72504637e-02
-6.00733869e-02 -1.43190578e-01 7.54871815e-02 1.42249107e+00
4.45729941e-01 -5.75426593e-02 3.22414458e-01 1.54084229e+00
-1.97697490e-01 3.36519271e-01 -3.21550101e-01 -3.21128517e-01
6.76406473e-02 1.03578413e+00 -4.07321006e-01 -2.95631081e-01
-1.21161759e+00 3.79941761e-01 4.80809540e-01 2.39001662e-01
-8.33196342e-01 -8.08284700e-01 4.53235120e-01 -8.93019140e-02
1.66828305e-01 -2.39269957e-01 -4.91237253e-01 -1.60801041e+00
3.56432915e-01 -1.14428639e+00 5.69189966e-01 -8.53594005e-01
-1.07062590e+00 5.29890537e-01 -7.87812918e-02 -8.17722499e-01
-1.30853862e-01 -6.53245926e-01 -5.91990411e-01 9.83823240e-01
-1.89285409e+00 -9.59653616e-01 -2.14027315e-01 1.75791711e-01
6.46497607e-01 -3.13883901e-01 5.45038164e-01 4.28779274e-01
-5.10246158e-01 4.90890563e-01 -7.62994885e-02 7.58275628e-01
2.78053045e-01 -1.22607434e+00 5.53425610e-01 1.16161048e+00
-3.53001878e-02 1.10850668e+00 3.28457177e-01 -6.64330006e-01
-1.63454676e+00 -1.20720291e+00 1.28693342e+00 -7.47532070e-01
9.63960290e-01 -5.42837381e-01 -1.31048775e+00 7.73319662e-01
5.32283634e-02 -2.03975558e-01 7.04247713e-01 1.43239619e-02
-8.92971933e-01 2.60161042e-01 -7.81515658e-01 1.11546136e-01
6.83277130e-01 -8.12039077e-01 -9.77739215e-01 8.30539316e-02
9.38915968e-01 -3.47326010e-01 -7.54594803e-01 8.53566304e-02
2.30368495e-01 -1.02316582e+00 5.97687721e-01 -6.19132042e-01
1.09103656e+00 -2.99828708e-01 -2.12858781e-01 -8.83068085e-01
1.24714680e-01 -2.35766843e-01 1.84551433e-01 1.57204676e+00
5.56927621e-01 -4.54897881e-01 4.22856569e-01 8.63976538e-01
-2.96757489e-01 -4.85827237e-01 -3.91320050e-01 -5.84153354e-01
3.78331423e-01 -7.60665238e-01 7.35794723e-01 9.70375597e-01
5.12237728e-01 1.68656558e-01 1.21887766e-01 3.89709651e-01
1.79251894e-01 6.01830840e-01 7.33037591e-01 -1.27676463e+00
-7.19873130e-01 -3.93535554e-01 -1.26345500e-01 -1.11235535e+00
4.97256488e-01 -1.46266019e+00 1.05787985e-01 -1.23307860e+00
5.92677772e-01 -5.44938266e-01 1.21203519e-01 5.51581025e-01
-3.25065970e-01 -6.51143610e-01 2.64738858e-01 3.07290375e-01
-4.39812064e-01 3.57186377e-01 4.82493520e-01 -8.28228891e-02
6.18287027e-02 -5.84323071e-02 -9.67603028e-01 1.10981119e+00
4.58218127e-01 -9.73286390e-01 -8.41632038e-02 -9.26786304e-01
7.56715417e-01 2.97384709e-01 1.76388383e-01 -6.37556493e-01
1.83504567e-01 -2.60288745e-01 -4.15124029e-01 -8.92762914e-02
-6.88904643e-01 -8.00091982e-01 -2.59035259e-01 5.92406631e-01
-6.45228863e-01 2.27433234e-01 1.80413872e-01 4.70275283e-01
-3.84797633e-01 -1.11808968e+00 6.44484103e-01 -4.87788469e-01
-7.90755570e-01 -1.04643285e-01 -4.10687000e-01 4.95306611e-01
7.53672719e-01 3.39537323e-01 -4.68949139e-01 2.12536365e-01
-2.20364749e-01 1.75077558e-01 4.42065597e-01 7.85122156e-01
3.61073673e-01 -9.34355915e-01 -4.77203488e-01 5.06015241e-01
5.33319831e-01 -1.44360363e-01 -1.59524903e-01 6.49202347e-01
-7.81093717e-01 5.12757480e-01 6.15729019e-02 -3.89044791e-01
-1.08652639e+00 7.22754300e-01 2.35203847e-01 -4.74900752e-01
-3.55259180e-01 8.74749124e-01 3.82015496e-01 -9.06101584e-01
1.30717784e-01 -8.53053927e-01 -3.58307809e-01 -3.71868521e-01
6.52342677e-01 7.79675171e-02 4.69745807e-02 -2.64380157e-01
-2.78440416e-01 7.26626277e-01 -1.44824952e-01 4.18528646e-01
1.46908927e+00 -2.32231691e-02 -8.65032136e-01 2.44934529e-01
1.28620076e+00 4.19714749e-01 -8.73335838e-01 -4.08930391e-01
1.02519226e+00 -6.24761462e-01 -4.17563915e-01 -6.33712292e-01
-9.64854896e-01 1.12323654e+00 9.28520337e-02 2.59433299e-01
7.64304519e-01 3.84111434e-01 6.81308091e-01 4.95969772e-01
5.39333880e-01 -6.71985686e-01 1.03671320e-01 8.07151616e-01
7.47472942e-01 -1.10212874e+00 -4.97307897e-01 -7.51246691e-01
-4.60178077e-01 1.51174510e+00 7.56344497e-01 1.04579367e-01
6.89712167e-01 4.77242410e-01 5.16830273e-02 -2.14615703e-01
-6.82395816e-01 1.17467277e-01 1.97810322e-01 3.32634330e-01
7.31854677e-01 -3.97345454e-01 -2.92188436e-01 1.28802466e+00
-1.71146601e-01 1.95598945e-01 8.69088411e-01 1.10246003e+00
-5.94104886e-01 -1.48771584e+00 -3.17167461e-01 2.19682559e-01
-7.34637499e-01 -4.86623257e-01 -2.56188244e-01 4.11249071e-01
2.43853211e-01 8.30235302e-01 -3.48978609e-01 -1.50532275e-02
3.16456974e-01 3.04963052e-01 -3.85081507e-02 -1.26067281e+00
-8.29005480e-01 -3.94124627e-01 -1.17845103e-01 -3.42546225e-01
-2.14975417e-01 -5.99255741e-01 -1.51652110e+00 3.68932039e-02
-1.44106165e-01 4.51256543e-01 6.68699443e-01 1.02224576e+00
3.68823856e-01 4.23302054e-01 4.20591354e-01 6.54576421e-02
-5.91159284e-01 -5.73980033e-01 -2.07946092e-01 3.67833704e-01
2.00735077e-01 -1.58407137e-01 -2.86707669e-01 5.97268343e-01]
|
[7.595324993133545, 7.936840534210205]
|
3a07ea76-3a75-43b2-8030-327b380da709
|
explainable-ai-for-time-series-via-virtual
|
2303.06365
| null |
https://arxiv.org/abs/2303.06365v1
|
https://arxiv.org/pdf/2303.06365v1.pdf
|
Explainable AI for Time Series via Virtual Inspection Layers
|
The field of eXplainable Artificial Intelligence (XAI) has greatly advanced in recent years, but progress has mainly been made in computer vision and natural language processing. For time series, where the input is often not interpretable, only limited research on XAI is available. In this work, we put forward a virtual inspection layer, that transforms the time series to an interpretable representation and allows to propagate relevance attributions to this representation via local XAI methods like layer-wise relevance propagation (LRP). In this way, we extend the applicability of a family of XAI methods to domains (e.g. speech) where the input is only interpretable after a transformation. Here, we focus on the Fourier transformation which is prominently applied in the interpretation of time series and LRP and refer to our method as DFT-LRP. We demonstrate the usefulness of DFT-LRP in various time series classification settings like audio and electronic health records. We showcase how DFT-LRP reveals differences in the classification strategies of models trained in different domains (e.g., time vs. frequency domain) or helps to discover how models act on spurious correlations in the data.
|
['Wojciech Samek', 'Grégoire Montavon', 'Sebastian Lapuschkin', 'Johanna Vielhaben']
|
2023-03-11
| null | null | null | null |
['time-series-classification']
|
['time-series']
|
[ 7.09030688e-01 5.89975476e-01 1.13343112e-01 -4.14766490e-01
-3.41444850e-01 -5.77878296e-01 7.19089925e-01 3.50613922e-01
8.94650444e-02 5.68278491e-01 2.78634101e-01 -6.26478314e-01
-5.64853370e-01 -5.60694456e-01 -6.86976016e-01 -4.48134989e-01
-3.32207501e-01 1.90331563e-01 -3.04200351e-01 -2.24018067e-01
1.57631431e-02 2.50372738e-01 -1.46795809e+00 3.86991769e-01
6.43010974e-01 1.09162569e+00 -1.65885970e-01 5.34421563e-01
3.46411504e-02 7.95723617e-01 -8.66486371e-01 -1.05749182e-01
-5.30510098e-02 -3.93333048e-01 -1.19710112e+00 -2.38116831e-01
-1.47787973e-01 1.47588536e-01 1.05748273e-01 7.54141212e-01
-5.29059619e-02 1.48068920e-01 6.27414465e-01 -1.51913512e+00
-7.41369069e-01 9.23867285e-01 -2.28132337e-01 3.16977173e-01
4.85795468e-01 -1.60051882e-01 1.14611888e+00 -5.78929007e-01
5.65112710e-01 1.45307767e+00 7.67596841e-01 3.80754352e-01
-1.43588233e+00 -2.29879484e-01 4.48637962e-01 5.13839841e-01
-8.50621104e-01 -5.88044012e-03 1.05468297e+00 -3.70303601e-01
9.29742932e-01 5.74690223e-01 4.29909706e-01 1.45497513e+00
2.83612698e-01 7.63874948e-01 1.12322462e+00 -5.74558914e-01
4.78744179e-01 -6.85864314e-02 4.90080178e-01 1.05557874e-01
-1.01972312e-01 7.29691703e-03 -6.35166764e-01 -1.84108093e-01
5.61166406e-01 -3.08841299e-02 -3.14359337e-01 2.61303008e-01
-1.47617173e+00 7.62823403e-01 3.90798509e-01 3.96135330e-01
-5.81959665e-01 1.74028665e-01 4.68853354e-01 7.41218269e-01
8.83197248e-01 9.22346234e-01 -6.65086210e-01 -3.40961777e-02
-7.04646468e-01 3.50571796e-02 6.74200058e-01 5.33927381e-01
3.29007834e-01 -6.92556202e-02 -1.64790705e-01 5.41567504e-01
3.94631386e-01 9.89269763e-02 5.15076935e-01 -8.47725570e-01
3.31044495e-01 3.67754340e-01 -2.90411822e-02 -1.03620148e+00
-5.37285089e-01 -4.27557766e-01 -7.68810928e-01 2.49441296e-01
3.79857630e-01 -1.16925813e-01 -7.51205027e-01 1.85636270e+00
1.92449704e-01 3.33444029e-01 1.06893547e-01 9.50296640e-01
5.13065219e-01 6.67252481e-01 -7.94785190e-03 -2.61560529e-01
1.33523333e+00 -6.40934110e-01 -8.21625888e-01 -1.54845625e-01
2.24457264e-01 -4.34229851e-01 1.05955899e+00 6.59441888e-01
-1.03854084e+00 -6.84492052e-01 -1.04894054e+00 -2.14606449e-02
-4.46504653e-01 -1.29164472e-01 5.56156397e-01 1.65261269e-01
-1.07720828e+00 9.74503636e-01 -1.01697648e+00 -4.25395042e-01
2.13478789e-01 3.80556375e-01 -2.84625262e-01 3.37591976e-01
-1.45432615e+00 9.14570749e-01 2.10632354e-01 3.26182067e-01
-4.92533475e-01 -8.15285742e-01 -6.94065750e-01 9.88601968e-02
2.41800919e-01 -7.86169291e-01 1.27141321e+00 -1.28989136e+00
-1.46619833e+00 4.28992599e-01 -9.19336975e-02 -8.00792933e-01
4.36450392e-01 -2.54940391e-01 -4.48006451e-01 1.60149768e-01
-1.35505637e-02 4.46528345e-01 1.16177189e+00 -9.83834326e-01
-3.76233697e-01 -3.23641837e-01 3.76717031e-01 -2.40829483e-01
-2.92642325e-01 2.40913965e-02 1.50870666e-01 -8.33065271e-01
2.06663579e-01 -9.18323934e-01 -6.09944984e-02 6.26227483e-02
-5.45640528e-01 -4.86003757e-01 9.28699076e-01 -7.71545172e-01
1.16882670e+00 -2.15625834e+00 4.52520341e-01 4.76461798e-01
2.64353067e-01 -5.11328056e-02 -9.26018208e-02 6.00790262e-01
-5.91993332e-01 3.79709244e-01 -4.11695212e-01 -4.64935124e-01
1.01929270e-01 5.09994447e-01 -8.43903720e-01 2.01045543e-01
6.55159295e-01 7.38196313e-01 -1.09187365e+00 -2.02426668e-02
1.09270729e-01 7.14679599e-01 -3.32138151e-01 8.67157951e-02
-4.46102440e-01 9.59550500e-01 -3.69211197e-01 3.13852787e-01
1.75945178e-01 -4.68427598e-01 -4.65290397e-02 -2.53015578e-01
-1.79956764e-01 5.48013330e-01 -7.23015189e-01 1.63173616e+00
-6.41953766e-01 1.03314030e+00 -3.28541785e-01 -1.25910521e+00
8.05827856e-01 5.88499069e-01 3.72538239e-01 -3.84741396e-01
-6.47621974e-02 4.87818718e-02 -6.06458969e-02 -4.73375142e-01
2.77059764e-01 -1.17708281e-01 1.15289651e-01 6.00417852e-01
-1.38767481e-01 7.68651292e-02 -2.34842151e-01 -1.71221539e-01
1.21247160e+00 2.93164551e-01 5.47845185e-01 -8.56736600e-02
5.08574545e-01 5.09295426e-02 7.57086724e-02 7.17750788e-01
4.19915840e-02 8.25704455e-01 6.16724133e-01 -4.67337906e-01
-8.58585596e-01 -9.61413801e-01 -2.60976404e-01 1.02607930e+00
-1.04053847e-01 -4.14574325e-01 -5.29056787e-01 -5.57492971e-01
-2.07246721e-01 9.33474183e-01 -9.31052268e-01 -3.88463020e-01
-5.06923676e-01 -5.61289966e-01 4.26496565e-01 5.54632962e-01
1.32111190e-02 -1.28251672e+00 -7.72896349e-01 3.89782935e-01
-4.50327039e-01 -8.06093276e-01 1.80072896e-02 4.51314807e-01
-1.13182092e+00 -6.83238745e-01 -3.81156743e-01 -3.07498544e-01
5.19385159e-01 -1.38700902e-01 1.22659314e+00 -6.64054602e-02
-1.94210589e-01 6.61086023e-01 -6.55807436e-01 -7.53753662e-01
-5.13995707e-01 -9.33953933e-03 -6.87422138e-03 1.57912001e-01
1.85509145e-01 -7.83702195e-01 -4.58015561e-01 1.76422507e-01
-1.06093872e+00 5.23951761e-02 2.82399505e-01 8.79840851e-01
5.55600703e-01 5.86600676e-02 7.26610243e-01 -7.86640823e-01
9.60247219e-01 -8.32917035e-01 -1.81060344e-01 2.97258914e-01
-7.34359145e-01 4.29259688e-01 8.56047571e-01 -5.25408208e-01
-9.35230494e-01 -4.35000807e-01 -5.64058200e-02 -5.74315131e-01
-2.50516117e-01 1.08691990e+00 3.04476678e-01 3.45442146e-01
8.00776064e-01 -1.48990020e-01 5.54451160e-03 -5.79937100e-01
3.16402704e-01 5.36717951e-01 6.73018038e-01 -4.38924909e-01
6.62527204e-01 5.33622324e-01 3.44693512e-02 -6.98472023e-01
-9.31611896e-01 -1.12380438e-01 -5.49873888e-01 -1.66269615e-01
7.39210069e-01 -4.35055107e-01 -6.70258582e-01 1.27006769e-02
-1.49662352e+00 -1.20190732e-01 -6.77223325e-01 5.67342937e-01
-6.97391212e-01 1.88498273e-01 -3.89700115e-01 -8.72970402e-01
-4.31204528e-01 -8.17286611e-01 1.28963578e+00 -9.98048633e-02
-8.79316688e-01 -1.35191846e+00 -4.34615612e-02 -8.95565934e-03
3.65296036e-01 5.37810922e-01 1.26769316e+00 -7.93779612e-01
-2.69513130e-01 -5.20926490e-02 7.22753331e-02 2.57119477e-01
1.98037446e-01 3.85811715e-03 -1.42719138e+00 1.03306673e-01
4.08513755e-01 9.00961924e-03 7.69566357e-01 5.34104228e-01
1.37786806e+00 -4.67155457e-01 -1.70293748e-01 3.36240560e-01
9.33712959e-01 2.44095087e-01 5.97607255e-01 5.06549060e-01
4.59453195e-01 1.02057326e+00 8.14984024e-01 5.57233155e-01
2.31488675e-01 7.46903479e-01 5.58329642e-01 -1.61128551e-01
6.32364824e-02 -7.28446767e-02 3.67305487e-01 7.19529808e-01
-4.46767360e-01 -2.86946803e-01 -7.91589081e-01 4.83505636e-01
-2.07517123e+00 -9.61598754e-01 -9.79148224e-02 2.10909820e+00
8.79251301e-01 6.35525361e-02 -1.50857925e-01 6.87579572e-01
4.71498191e-01 -1.24047371e-02 -5.82016528e-01 -6.15093589e-01
1.10759258e-01 3.31364363e-01 -1.46525845e-01 4.22924727e-01
-1.13041675e+00 1.82699680e-01 5.98313761e+00 4.46664304e-01
-1.26184058e+00 -3.05737369e-02 7.14043558e-01 1.94588706e-01
-5.50598443e-01 -7.73137882e-02 -1.14678149e-03 3.14594716e-01
1.07807481e+00 -1.97683647e-01 5.25420368e-01 7.00875998e-01
5.21727622e-01 3.72125447e-01 -1.72608578e+00 8.16420734e-01
-3.13043773e-01 -1.01200843e+00 -1.74830228e-01 -9.24793258e-02
1.46447480e-01 -2.72782475e-01 2.95378178e-01 6.42187968e-02
-2.19558388e-01 -1.15800905e+00 9.51637805e-01 6.68413401e-01
5.82916439e-01 -3.55599135e-01 7.25022018e-01 1.33417204e-01
-1.03256869e+00 -9.02272388e-02 -1.29731223e-01 -3.41991097e-01
2.95381024e-02 6.67616427e-01 -1.19852769e+00 7.89415121e-01
7.38229573e-01 9.61283624e-01 -3.25212628e-01 6.92700803e-01
-3.86308134e-01 9.83986318e-01 -3.85505617e-01 2.67630130e-01
-1.28637953e-03 -1.65970072e-01 8.14993262e-01 9.75395024e-01
3.62918377e-01 -1.00449972e-01 -1.81150943e-01 1.14654517e+00
4.09122884e-01 -3.62738013e-01 -6.03413165e-01 -6.89093545e-02
2.59744793e-01 1.00209892e+00 -6.22045577e-01 -2.35589892e-01
-2.57742494e-01 9.54717755e-01 9.29574296e-02 4.88648593e-01
-9.35822129e-01 -2.24697188e-01 5.90027809e-01 -4.74316515e-02
1.69345081e-01 1.80573959e-03 -4.47398573e-01 -9.44008648e-01
1.93316326e-01 -7.96315372e-01 3.81379217e-01 -1.09618950e+00
-1.56541181e+00 8.70575190e-01 4.45726067e-01 -1.49730575e+00
-7.67638922e-01 -5.23934484e-01 -7.04578459e-01 9.93197620e-01
-1.57441962e+00 -9.69823956e-01 -1.72335893e-01 4.08446431e-01
6.38414919e-01 1.84039295e-01 9.88864541e-01 -5.45002371e-02
-3.16795021e-01 3.40163916e-01 -9.47704464e-02 -3.32961291e-01
4.56985861e-01 -1.40938926e+00 5.77515900e-01 6.87462687e-01
2.75069356e-01 8.31985891e-01 1.09132516e+00 -3.44503522e-01
-1.23326159e+00 -1.13262212e+00 1.02387857e+00 -4.86809164e-01
8.73270631e-01 -7.43070766e-02 -1.17457688e+00 7.45485187e-01
2.51879156e-01 -1.39826596e-01 6.02262080e-01 2.14194521e-01
-3.61744791e-01 -9.35665816e-02 -9.95785534e-01 6.67399347e-01
9.14046586e-01 -6.59345865e-01 -8.35488379e-01 3.71868193e-01
1.00490391e+00 -1.23106711e-01 -9.08339202e-01 2.87825048e-01
4.77148920e-01 -6.60638392e-01 1.13798487e+00 -6.78001463e-01
4.92758840e-01 -3.81811440e-01 7.85034373e-02 -1.50564444e+00
-2.27585971e-01 -7.61101425e-01 -2.82954067e-01 1.10595179e+00
5.43654859e-01 -8.25016260e-01 1.65129274e-01 7.00279236e-01
-2.10069790e-01 -6.09488487e-01 -8.27949405e-01 -7.08058298e-01
-3.07583302e-01 -8.67520154e-01 6.75548196e-01 8.81881297e-01
2.31977209e-01 2.48268411e-01 -3.23164165e-01 2.54402518e-01
2.87743747e-01 2.62326747e-01 2.19997823e-01 -1.55907476e+00
-6.21710598e-01 -1.98733106e-01 -4.95247215e-01 -6.80722833e-01
1.10046156e-01 -7.36666858e-01 -1.00040007e-02 -1.26747739e+00
-3.37158710e-01 -2.70315945e-01 -4.52237427e-01 8.39870453e-01
-1.96355298e-01 9.36798602e-02 1.75602525e-01 4.49551642e-01
-1.53202757e-01 3.96232069e-01 9.58943605e-01 -2.69897938e-01
-2.71974444e-01 3.12635422e-01 -6.47465348e-01 7.04114854e-01
8.21689665e-01 -3.93719584e-01 -7.56278455e-01 -4.58988190e-01
3.76043499e-01 1.10353731e-01 6.41210794e-01 -7.79740632e-01
3.28991264e-02 9.75583971e-04 9.28696841e-02 -2.29535758e-01
3.39611918e-01 -9.45806384e-01 3.29716176e-01 1.82868034e-01
-7.51785636e-01 1.61512598e-01 1.94653764e-01 7.19720602e-01
-5.32334268e-01 -2.05107227e-01 1.30263254e-01 1.92882903e-02
-5.66793323e-01 -1.00626916e-01 -3.13658565e-01 -2.04780921e-01
6.03522182e-01 -2.63045967e-01 -2.48234719e-01 -5.20213187e-01
-9.56667244e-01 -5.25235161e-02 -9.20426175e-02 5.58960497e-01
6.20405436e-01 -1.08667839e+00 -6.07878923e-01 8.41031969e-02
1.81732908e-01 -9.71668735e-02 -8.32046345e-02 1.14218497e+00
-5.70614338e-02 5.12110531e-01 8.26201215e-03 -7.12974310e-01
-1.17020202e+00 5.56042373e-01 6.84624389e-02 -3.76913130e-01
-8.02876830e-01 5.65574765e-01 3.28355014e-01 -1.92369223e-01
2.47102663e-01 -9.63084519e-01 -6.06135905e-01 5.14240041e-02
5.30022442e-01 2.67038465e-01 1.49613112e-01 -3.02793622e-01
-4.50883180e-01 4.50207204e-01 9.40940082e-02 -3.36890250e-01
1.50490725e+00 -2.33444795e-01 -1.58387825e-01 9.42343295e-01
1.00385582e+00 -3.59560341e-01 -1.07709277e+00 -4.19342741e-02
2.18483612e-01 -4.19871099e-02 -1.83608338e-01 -8.42759252e-01
-6.05649054e-01 9.91258264e-01 4.19648498e-01 8.65245879e-01
1.59613311e+00 1.22957259e-01 3.48181278e-01 3.04170579e-01
3.87652665e-01 -6.71584725e-01 -8.22368115e-02 4.49826151e-01
1.24667430e+00 -1.01349378e+00 -1.86691925e-01 -3.06694031e-01
-5.02605915e-01 1.45865285e+00 -9.95008796e-02 1.37110874e-01
5.30225396e-01 3.38455997e-02 1.27341850e-02 -2.55851775e-01
-9.39247549e-01 2.59024017e-02 5.75089097e-01 7.62788832e-01
5.40830970e-01 6.66804165e-02 -1.23710603e-01 8.57700229e-01
-3.82540405e-01 1.19497553e-01 4.97619480e-01 7.25215554e-01
6.30430505e-02 -1.02442884e+00 -5.46040237e-01 2.77007997e-01
-4.94058609e-01 -1.19174607e-01 -4.88908738e-01 6.11778140e-01
2.33627204e-02 1.24923754e+00 3.04207467e-02 -4.17567641e-01
2.15757132e-01 2.00221688e-01 1.49566710e-01 -5.48531771e-01
-8.58451188e-01 -7.66285434e-02 9.36989784e-02 -5.34895182e-01
-7.09662616e-01 -6.46959543e-01 -1.39956105e+00 1.71629235e-01
-1.61445931e-01 9.89459679e-02 7.70585358e-01 1.17479396e+00
3.49042386e-01 9.02087629e-01 5.85426629e-01 -8.18567157e-01
-4.73329961e-01 -8.02041709e-01 -2.78858185e-01 4.97805059e-01
8.00719321e-01 -4.62432921e-01 -5.91053784e-01 4.33307141e-01]
|
[7.291958808898926, 3.2109427452087402]
|
933b2264-db38-4ce9-9d29-1b31db79da7a
|
empirical-analysis-of-indirect-internal
|
2002.12274
| null |
https://arxiv.org/abs/2002.12274v1
|
https://arxiv.org/pdf/2002.12274v1.pdf
|
Empirical Analysis of Indirect Internal Conversions in Cryptocurrency Exchanges
|
Algorithmic trading is well studied in traditional financial markets. However, it has received less attention in centralized cryptocurrency exchanges. The Commodity Futures Trading Commission (CFTC) attributed the $2010$ flash crash, one of the most turbulent periods in the history of financial markets that saw the Dow Jones Industrial Average lose $9\%$ of its value within minutes, to automated order "spoofing" algorithms. In this paper, we build a set of methodologies to characterize and empirically measure different algorithmic trading strategies in Binance, a large centralized cryptocurrency exchange, using a complete data set of historical trades. We find that a sub-strategy of triangular arbitrage is widespread, where bots convert between two coins through an intermediary coin, and obtain a favorable exchange rate compared to the direct one. We measure the profitability of this strategy, characterize its risks, and outline two strategies that algorithmic trading bots use to mitigate their losses. We find that this strategy yields an exchange ratio that is $0.144\%$, or $14.4$ basis points (bps) better than the direct exchange ratio. $2.71\%$ of all trades on Binance are attributable to this strategy.
|
['Damon McCoy', 'Tobias Lauinger', 'Paz Grimberg']
|
2020-02-27
| null | null | null | null |
['algorithmic-trading']
|
['time-series']
|
[-5.89958608e-01 3.25098448e-02 -1.33100785e-02 2.51798660e-01
-5.67321181e-01 -1.38706791e+00 8.14674675e-01 -1.54712111e-01
-4.40405518e-01 8.73312354e-01 -1.52995721e-01 -8.94165397e-01
2.94505246e-02 -9.71800566e-01 -4.74055976e-01 -4.98204321e-01
-4.21960145e-01 7.55459130e-01 2.23740533e-01 -3.86193961e-01
8.94144773e-01 5.13526559e-01 -4.33213979e-01 6.10706210e-02
3.13707083e-01 1.62019587e+00 -6.88989758e-01 2.61335433e-01
-6.82475418e-02 9.96277571e-01 -8.07072401e-01 -9.71811771e-01
1.21961081e+00 -5.04600942e-01 -5.73065042e-01 -4.43826407e-01
-2.68572092e-01 -8.96199644e-01 -8.27198848e-02 1.34374118e+00
-1.49015322e-01 -5.14121354e-01 4.65997964e-01 -1.27415669e+00
-3.44229311e-01 1.27835941e+00 -7.78904080e-01 4.07633185e-01
-3.13488215e-01 3.53769958e-01 1.65104759e+00 -5.42351425e-01
7.57740617e-01 6.87133372e-01 4.78354603e-01 1.03387162e-02
-1.40788496e+00 -1.21233451e+00 -4.14124817e-01 -5.18061638e-01
-8.64815652e-01 -7.04025105e-02 6.04266584e-01 -3.92727107e-01
8.75206292e-01 4.30341125e-01 1.03661156e+00 3.80279005e-01
6.43951416e-01 3.33041251e-01 1.47294819e+00 4.54340912e-02
3.37532461e-01 1.09339587e-01 1.92507625e-01 -1.26856893e-01
9.94078994e-01 6.15567088e-01 -3.83666694e-01 -8.29335213e-01
9.18589234e-01 -2.18206272e-01 -1.09008491e-01 -1.44790271e-02
-1.06907296e+00 1.21413755e+00 5.35225332e-01 4.04418945e-01
-5.43848515e-01 4.88527477e-01 2.62651414e-01 8.00033569e-01
3.93359423e-01 8.13078463e-01 -6.29889429e-01 -5.03228366e-01
-9.38977599e-01 7.41648555e-01 1.41150630e+00 5.76729596e-01
3.68948787e-01 1.07548170e-01 5.05877614e-01 -1.08723059e-01
3.55713457e-01 4.94587034e-01 3.00542057e-01 -1.25389791e+00
6.29941881e-01 1.38699085e-01 5.64731777e-01 -1.01500893e+00
6.82312474e-02 -3.45333815e-01 -4.88258779e-01 6.19073749e-01
1.03442430e+00 -3.65804583e-01 -4.07139249e-02 1.00945699e+00
-3.39944452e-01 -3.78056377e-01 -1.64096922e-01 7.32625008e-01
-6.86731219e-01 3.05352360e-01 -4.95178461e-01 -3.64058465e-01
1.40139019e+00 -7.01610148e-01 -3.97013724e-01 3.54492366e-01
3.15242797e-01 -1.07386231e+00 2.61844724e-01 6.92211449e-01
-1.19433355e+00 4.11376745e-01 -9.33209002e-01 7.61989653e-01
-1.87962741e-01 -9.16461408e-01 6.44922674e-01 8.77701044e-01
-7.96923339e-01 9.93857324e-01 -8.04648936e-01 4.94783282e-01
3.82093698e-01 2.84856319e-01 4.62767959e-01 1.06617248e+00
-1.16545212e+00 8.05085421e-01 9.80094820e-02 -2.81891525e-02
-8.36159527e-01 -8.31226945e-01 7.14234635e-02 1.78671390e-01
5.34741938e-01 -1.30988419e-01 1.54293823e+00 -8.32330108e-01
-1.37588954e+00 5.21109760e-01 7.84401476e-01 -1.37933815e+00
1.14115262e+00 -2.70792544e-01 -2.64801979e-01 1.06074445e-01
5.99056631e-02 -3.81138474e-02 5.52566528e-01 -8.43011439e-01
-6.88719630e-01 -3.31479996e-01 -7.47280940e-02 -3.29115629e-01
4.55480367e-02 4.54034001e-01 7.47841656e-01 -1.27025414e+00
3.05772126e-01 -1.13356006e+00 -9.97273847e-02 -2.90726364e-01
-4.55071926e-01 5.08144647e-02 5.90679109e-01 -7.25027323e-01
1.00526702e+00 -1.65491891e+00 -6.74995065e-01 6.50541544e-01
3.88472438e-01 -6.48412034e-02 6.86675310e-01 6.47700787e-01
-5.60385128e-03 8.53737593e-01 -2.46371642e-01 3.36945802e-01
4.78963047e-01 -3.86677146e-01 -1.05035317e+00 4.06704783e-01
-1.54251665e-01 1.08147776e+00 -5.91897249e-01 2.73825139e-01
-1.79665819e-01 -2.25794658e-01 -4.07466561e-01 -9.03216079e-02
-2.56377935e-01 2.20733881e-01 -4.87477094e-01 1.02063763e+00
9.43463564e-01 -3.07329327e-01 4.36407447e-01 1.10904053e-01
-6.13119602e-01 5.27048588e-01 -1.02044868e+00 7.83567131e-01
4.71236497e-01 5.52478492e-01 2.03412011e-01 -2.96377689e-01
8.96085441e-01 9.87718031e-02 4.67341006e-01 -7.05342114e-01
3.36436450e-01 8.84375632e-01 2.83378720e-01 3.90690833e-01
6.34851575e-01 -5.63465476e-01 -5.37640825e-02 1.48574972e+00
-7.83892632e-01 -4.42691207e-01 -1.02730785e-02 7.87396729e-02
1.26895380e+00 -2.16064721e-01 2.43929833e-01 -6.57591760e-01
-9.67622846e-02 3.53773654e-01 5.81594348e-01 7.08335042e-01
-5.32279193e-01 9.04511362e-02 9.93959606e-01 -8.04795504e-01
-1.11759007e+00 -1.07133579e+00 -2.75854487e-02 2.99187988e-01
1.37493923e-01 -3.80057059e-02 -7.51313925e-01 -5.75314820e-01
5.35631001e-01 7.25965083e-01 -1.25156313e-01 3.81196618e-01
-8.10297370e-01 -1.00531459e+00 7.74910748e-01 1.32260382e-01
9.88175273e-01 -1.10088599e+00 -1.12256122e+00 5.37003398e-01
1.20872639e-01 -5.35047352e-01 -5.06371081e-01 8.72796029e-02
-8.86444688e-01 -1.25886989e+00 -5.99901021e-01 1.63520783e-01
1.76087663e-01 -7.55698001e-03 1.13723743e+00 4.48619314e-02
-7.62511119e-02 -4.45188910e-01 -1.81631893e-01 -3.87188047e-01
-3.67712110e-01 -1.79484174e-01 1.90286562e-02 4.45022434e-02
4.45271492e-01 -3.81216735e-01 -9.14177239e-01 3.90504509e-01
-6.80574894e-01 -4.78682041e-01 3.49751949e-01 4.42195386e-01
5.74024804e-02 2.00349092e-01 3.93784851e-01 -8.09720695e-01
6.35019243e-01 -4.42276537e-01 -1.47154546e+00 -1.53759317e-02
-1.17711139e+00 -3.46595198e-02 7.55102485e-02 -9.28031430e-02
-8.30403388e-01 -7.07564235e-01 5.50232768e-01 -2.21930727e-01
5.59340715e-01 3.57462615e-01 4.43822265e-01 -3.63231629e-01
-2.99580861e-03 -3.71156894e-02 8.73129591e-02 -5.83574951e-01
6.00808635e-02 2.81381309e-01 2.02379063e-01 -4.43946481e-01
1.08993089e+00 6.87603712e-01 -3.89234811e-01 -2.38063395e-01
2.40060617e-03 2.84086615e-01 3.18065286e-02 4.98798676e-02
4.99297857e-01 -4.11627978e-01 -1.29305530e+00 9.86810803e-01
-1.03927910e+00 -3.57623577e-01 -3.52714658e-01 6.75519884e-01
-4.40060228e-01 5.89814745e-02 -1.32985771e+00 -9.96792614e-01
-4.29023921e-01 -8.00646424e-01 3.51191238e-02 2.82430314e-02
-6.00690544e-01 -5.94367802e-01 5.94151795e-01 3.89308900e-01
9.84280229e-01 4.44550157e-01 6.47468746e-01 -1.10689652e+00
-1.24729609e+00 1.11552000e-01 -3.76650751e-01 3.55600148e-01
3.84375423e-01 5.62486574e-02 -4.56638247e-01 -3.68818671e-01
9.22337174e-01 9.35202166e-02 6.22946858e-01 1.56552091e-01
1.72240719e-01 -7.50905037e-01 -1.33354664e-02 3.45831752e-01
1.59169412e+00 8.74310195e-01 4.84950662e-01 9.18773532e-01
-1.15795285e-01 4.41630542e-01 4.74526793e-01 7.18624234e-01
-6.40370697e-02 3.46032709e-01 4.26899135e-01 5.39017618e-01
4.37956333e-01 -1.84588775e-01 4.80388552e-01 6.67760253e-01
-2.97799230e-01 2.54264563e-01 -8.89812768e-01 3.07358712e-01
-1.24268878e+00 -9.53603864e-01 1.04170799e-01 1.99558592e+00
7.92613149e-01 7.58626521e-01 4.36800182e-01 -3.37357782e-02
8.37872326e-01 1.06400505e-01 -6.49850488e-01 -2.90253162e-01
-1.21061489e-01 3.82618099e-01 1.36705041e+00 2.56818980e-01
-6.83923125e-01 7.32770264e-01 6.53085232e+00 6.00664854e-01
-1.03339863e+00 -5.35152741e-02 1.08639622e+00 -2.12706134e-01
-4.60087955e-01 5.06474078e-01 -5.52563250e-01 9.63744521e-01
1.12098134e+00 -7.43264735e-01 7.81467140e-01 5.60133398e-01
1.69732600e-01 -1.13686614e-01 -6.97732747e-01 5.22261679e-01
-5.57844400e-01 -1.79154778e+00 -7.46725649e-02 8.57279062e-01
6.05280519e-01 6.54790597e-03 2.87622601e-01 -1.96863189e-01
8.98223996e-01 -8.34143698e-01 1.31114876e+00 4.98623729e-01
1.47467583e-01 -9.76305783e-01 8.05131555e-01 -1.11730229e-02
-1.05508053e+00 -6.79288656e-02 6.25837818e-02 -2.76911885e-01
4.36675847e-01 5.80709338e-01 -4.47780102e-01 5.00673473e-01
9.62181270e-01 2.14846790e-01 -1.39501497e-01 6.68065310e-01
2.19503790e-02 6.20508671e-01 -5.80051124e-01 -3.79324555e-02
6.58169091e-01 -1.03421390e+00 7.56226540e-01 3.16569328e-01
4.77720052e-01 4.01018620e-01 -5.46482921e-01 1.47575617e+00
-4.51355040e-01 -3.32091868e-01 -4.64233100e-01 -5.61147094e-01
7.24760473e-01 8.86988699e-01 -1.28961194e+00 -2.87704557e-01
-1.91121176e-01 6.14647746e-01 -4.03916925e-01 1.30734909e-02
-8.08349609e-01 -5.37203014e-01 7.45102584e-01 2.21784636e-01
5.76637089e-01 -3.24394286e-01 -8.32782805e-01 -1.21261942e+00
1.30170956e-01 -1.05772531e+00 1.89415246e-01 -2.82579780e-01
-1.27567446e+00 4.84194517e-01 -4.50758100e-01 -1.20238149e+00
-3.01134229e-01 -3.04054290e-01 -6.45479977e-01 7.14228332e-01
-1.36940002e+00 -4.13981438e-01 7.52121210e-01 3.26626599e-01
3.12801972e-02 -4.54771996e-01 1.65752977e-01 -1.29507512e-01
-1.10697508e-01 3.21666628e-01 4.26826000e-01 5.62814236e-01
3.15247506e-01 -1.24541426e+00 9.07396555e-01 6.50532186e-01
3.18461716e-01 9.09812391e-01 6.99527323e-01 -1.10442090e+00
-1.29660463e+00 -4.69077528e-01 9.62224483e-01 -5.52190244e-01
1.64678049e+00 -8.39742497e-02 -5.91746688e-01 9.69599426e-01
6.16257548e-01 -5.48594117e-01 2.66217172e-01 -6.51414156e-01
-5.55540502e-01 -1.54836804e-01 -1.43016601e+00 4.17245597e-01
3.05655479e-01 -4.40519542e-01 -9.74641085e-01 -1.75180078e-01
7.72402465e-01 1.58687800e-01 -1.00753248e+00 -1.82960480e-02
8.89315903e-01 -1.54685891e+00 6.28946483e-01 -2.64904827e-01
8.74570757e-02 -6.53472990e-02 -1.53127223e-01 -9.33366358e-01
2.81924367e-01 -1.52318263e+00 8.42150152e-02 1.07316518e+00
4.92554069e-01 -1.46010447e+00 9.34135377e-01 1.02372575e+00
6.56138182e-01 -1.47019461e-01 -1.20268512e+00 -9.82302368e-01
6.44864380e-01 -8.99830163e-02 1.00687027e+00 1.05702472e+00
1.34690357e-02 -3.99001390e-01 -2.32864901e-01 -3.90624076e-01
1.20751143e+00 6.29888475e-01 4.30375904e-01 -8.79267514e-01
-3.66835445e-01 -9.46782112e-01 7.03804120e-02 -6.40101194e-01
-5.62282979e-01 -6.11384273e-01 -6.47205114e-01 -3.30606014e-01
-2.67380793e-02 -5.77333212e-01 -5.13668716e-01 1.65875763e-01
7.42773414e-01 2.97589421e-01 8.24458241e-01 9.24298286e-01
1.16793580e-01 1.85513124e-01 7.78531611e-01 -1.96637496e-01
-5.11715421e-04 1.10172868e-01 -7.88924694e-01 6.39771879e-01
9.34831977e-01 -6.81119859e-01 3.53700370e-01 -1.41761795e-01
5.04405081e-01 3.40604931e-01 5.35528243e-01 -5.14801204e-01
2.69437999e-01 -3.25523943e-01 -9.38438475e-02 -7.45188475e-01
-2.70008504e-01 -9.23278689e-01 8.42562079e-01 1.39615750e+00
-1.37258589e-01 4.68404979e-01 -3.11497778e-01 3.94827932e-01
-2.64685780e-01 -2.55915791e-01 6.17806613e-01 -6.05250120e-01
1.97675020e-01 6.01126999e-02 -4.97052372e-01 2.03033477e-01
1.26796484e+00 4.71215807e-02 -5.31121850e-01 -2.35426337e-01
-4.99990553e-01 -1.13800373e-02 6.86034501e-01 1.11891694e-01
8.40707198e-02 -1.25142443e+00 -7.38117516e-01 1.16561070e-01
-7.50690341e-01 -6.73068047e-01 -4.61713016e-01 8.62166882e-01
-1.47820115e+00 5.48516214e-01 -3.81781667e-01 9.11123753e-02
-5.28795242e-01 2.46696889e-01 5.48723638e-01 -5.71512818e-01
-6.02914453e-01 4.44527268e-01 -6.53852522e-02 1.03236228e-01
-3.00416917e-01 -5.78075349e-01 4.56131786e-01 5.66467285e-01
7.43273973e-01 7.90082216e-01 -1.55979767e-01 -1.67834595e-01
-2.69699126e-01 2.88874745e-01 -1.50840566e-01 -8.11432123e-01
1.34689486e+00 8.85105878e-02 -8.35558653e-01 3.53946894e-01
8.44115973e-01 2.44149134e-01 -1.20718110e+00 3.78121715e-03
8.36995482e-01 -8.05754662e-01 -4.99776185e-01 -9.01079297e-01
-1.50022483e+00 3.92482102e-01 -1.64176729e-02 1.15673292e+00
4.30501163e-01 -3.16223383e-01 1.25534785e+00 1.94654271e-01
9.25319195e-01 -9.37639296e-01 -3.30643177e-01 3.54810357e-01
7.36707926e-01 -4.37982619e-01 -1.09235190e-01 1.57563463e-01
-4.28233355e-01 1.00171077e+00 -1.07651815e-01 -8.24017227e-01
1.02962518e+00 5.34386158e-01 1.12121992e-01 -3.16960156e-01
-6.16432726e-01 6.58792496e-01 -7.57369041e-01 -2.82927930e-01
-1.69518888e-01 4.99512106e-01 -6.36139512e-01 9.05338228e-01
-5.69367349e-01 -9.55080148e-03 8.32868636e-01 1.19954216e+00
-4.79720742e-01 -1.13027561e+00 -5.30754030e-01 5.76366901e-01
-9.67486382e-01 -4.04460162e-01 -7.97363222e-01 1.02182472e+00
-3.96803230e-01 7.64598370e-01 4.86670405e-01 -1.38696060e-01
1.06203996e-01 -6.22011945e-02 -1.87994078e-01 1.07690386e-01
-1.47367501e+00 4.35299367e-01 -2.41933204e-02 -7.53297985e-01
-2.07105458e-01 -8.52743387e-01 -1.16755641e+00 -1.27458060e+00
-3.88691396e-01 5.86817086e-01 3.81183922e-01 4.51732665e-01
5.06905764e-02 -3.32871407e-01 9.62601364e-01 -5.63074410e-01
-1.25655687e+00 -6.37907267e-01 -1.40634573e+00 -4.24312241e-02
2.85892367e-01 -3.99470448e-01 -1.21817124e+00 -2.42970169e-01]
|
[4.698049545288086, 4.112616062164307]
|
a2c1a3a5-4aa9-415f-a235-d2d28b8ca1b5
|
sibylvariant-transformations-for-robust-text
|
2205.05137
| null |
https://arxiv.org/abs/2205.05137v1
|
https://arxiv.org/pdf/2205.05137v1.pdf
|
Sibylvariant Transformations for Robust Text Classification
|
The vast majority of text transformation techniques in NLP are inherently limited in their ability to expand input space coverage due to an implicit constraint to preserve the original class label. In this work, we propose the notion of sibylvariance (SIB) to describe the broader set of transforms that relax the label-preserving constraint, knowably vary the expected class, and lead to significantly more diverse input distributions. We offer a unified framework to organize all data transformations, including two types of SIB: (1) Transmutations convert one discrete kind into another, (2) Mixture Mutations blend two or more classes together. To explore the role of sibylvariance within NLP, we implemented 41 text transformations, including several novel techniques like Concept2Sentence and SentMix. Sibylvariance also enables a unique form of adaptive training that generates new input mixtures for the most confused class pairs, challenging the learner to differentiate with greater nuance. Our experiments on six benchmark datasets strongly support the efficacy of sibylvariance for generalization performance, defect detection, and adversarial robustness.
|
['Miryung Kim', 'Nanyun Peng', 'Muhammad Ali Gulzar', 'Fabrice Harel-Canada']
|
2022-05-10
| null |
https://aclanthology.org/2022.findings-acl.140
|
https://aclanthology.org/2022.findings-acl.140.pdf
|
findings-acl-2022-5
|
['defect-detection', 'classification']
|
['computer-vision', 'methodology']
|
[ 7.30792940e-01 1.81043133e-01 -1.27931386e-01 -4.16903198e-01
-6.12636745e-01 -1.08573210e+00 7.48367310e-01 1.22488715e-01
-2.43922830e-01 9.20670986e-01 -6.87889084e-02 -3.94185871e-01
-1.43613979e-01 -8.38367522e-01 -7.19711423e-01 -8.70962083e-01
4.59523112e-01 6.39169097e-01 1.98576242e-01 -2.28252858e-01
1.18074361e-02 5.40319920e-01 -1.71197641e+00 4.28469151e-01
1.32155597e+00 6.77174091e-01 -2.34143376e-01 3.56461853e-01
-2.87747741e-01 4.63448972e-01 -9.28357661e-01 -7.99978197e-01
4.34580296e-01 -5.79168379e-01 -5.18226683e-01 -2.58189082e-01
4.17531669e-01 8.86753127e-02 1.27672762e-01 1.13749111e+00
7.10720301e-01 1.18222229e-01 1.16708946e+00 -1.62766385e+00
-7.57490396e-01 9.81646359e-01 -3.67014855e-01 -3.21897864e-02
3.50622952e-01 1.17682502e-01 7.81073391e-01 -6.87901437e-01
5.10119557e-01 1.34076464e+00 6.58913136e-01 7.67788172e-01
-1.57608032e+00 -9.45797503e-01 1.41253501e-01 -1.85786486e-01
-1.23017085e+00 -2.41310686e-01 7.91448951e-01 -4.39025193e-01
6.57523155e-01 8.11553001e-01 2.15821922e-01 1.40581977e+00
2.03736350e-01 8.73512626e-01 1.17028344e+00 -6.64868236e-01
2.95120895e-01 4.92394984e-01 4.02911678e-02 3.27220231e-01
2.85705149e-01 2.05612093e-01 -4.18848783e-01 -2.59399444e-01
2.94170350e-01 -6.71583191e-02 -3.80878657e-01 -5.26246428e-01
-1.05617917e+00 8.28366160e-01 4.54870872e-02 1.89093262e-01
2.31911600e-01 -2.23098040e-01 3.10692012e-01 6.43884659e-01
4.05097246e-01 9.77261841e-01 -5.22655904e-01 1.70932442e-01
-5.70707679e-01 3.70483428e-01 7.77099967e-01 1.05319989e+00
6.56398296e-01 4.64754328e-02 -5.61236441e-01 9.28680837e-01
9.09969807e-02 6.23343945e-01 8.26029241e-01 -6.97714329e-01
5.82302272e-01 5.09839296e-01 -2.38120779e-01 -6.77175283e-01
-1.91288531e-01 -5.31046212e-01 -7.34925985e-01 2.87594706e-01
2.68010855e-01 -2.41822869e-01 -1.07911599e+00 1.96218288e+00
3.99375349e-01 -1.44328251e-01 2.74757862e-01 2.72632241e-01
6.90719008e-01 4.36452270e-01 4.76748087e-02 -1.80761851e-02
8.58529449e-01 -6.93261743e-01 -6.58462167e-01 1.96746849e-02
6.36349380e-01 -6.87113404e-01 1.32351148e+00 3.99975330e-01
-9.27960277e-01 -4.53696340e-01 -1.11330664e+00 1.68822721e-01
-7.86530554e-01 -2.17021018e-01 6.04578376e-01 1.23269415e+00
-7.62419283e-01 7.14507341e-01 -4.09693331e-01 -4.08108830e-02
3.67063940e-01 3.93120021e-01 -3.54112566e-01 9.96681750e-02
-1.33395576e+00 8.08093011e-01 6.92110658e-01 -4.57869500e-01
-5.03635883e-01 -1.15613830e+00 -8.51023316e-01 3.31894830e-02
1.56883135e-01 -6.93559468e-01 9.94355738e-01 -1.02399254e+00
-1.68885815e+00 7.45870590e-01 3.23753208e-01 -2.58141339e-01
9.01893258e-01 -3.67304124e-02 -4.19897050e-01 -3.05856287e-01
-3.42088714e-02 8.06894898e-01 9.42841887e-01 -1.41826200e+00
-5.25139511e-01 -1.86093703e-01 -3.57439905e-01 3.07610333e-01
-7.09134161e-01 -1.69931263e-01 -9.91758183e-02 -1.13790596e+00
-1.01358011e-01 -7.32690632e-01 1.28932327e-01 -1.44403309e-01
-6.84621572e-01 -2.05385149e-01 8.13929200e-01 -1.64187297e-01
1.04849374e+00 -2.29698968e+00 1.98671967e-01 1.55217931e-01
1.85192212e-01 4.72349554e-01 -2.58461565e-01 4.16267157e-01
-3.41295034e-01 3.75725478e-01 -5.53821146e-01 -3.87203634e-01
2.46644422e-01 2.33778849e-01 -7.35555828e-01 1.11689121e-01
3.08273464e-01 8.91787648e-01 -8.16513896e-01 -3.34166318e-01
2.13373434e-02 3.03503603e-01 -7.07261264e-01 4.44897749e-02
-3.28784466e-01 3.91243041e-01 -2.92342573e-01 5.69261491e-01
7.13039935e-01 1.68504819e-01 -1.25463277e-01 6.13064468e-02
3.46236259e-01 4.72840220e-02 -1.11361635e+00 1.36478007e+00
-1.01874493e-01 3.86434466e-01 -4.96620953e-01 -7.09238112e-01
1.04374158e+00 1.18962392e-01 2.60908246e-01 -3.83854657e-01
2.10800558e-01 1.50356039e-01 -2.53983233e-02 -2.50056475e-01
4.43766236e-01 -4.27785635e-01 -1.63062081e-01 4.97569472e-01
1.40710652e-01 -3.43500435e-01 2.23794565e-01 -6.83203042e-02
9.76468384e-01 1.08288340e-01 2.02291757e-01 -1.36011258e-01
3.79023761e-01 -2.09517524e-01 5.80520868e-01 1.00071442e+00
-9.89163741e-02 7.45524168e-01 4.58154023e-01 -7.68399835e-02
-8.33798289e-01 -1.60894620e+00 -3.45714986e-01 1.13357079e+00
5.03034890e-02 4.26564249e-04 -6.36601090e-01 -1.07500839e+00
3.35273325e-01 1.21162605e+00 -6.68610632e-01 -6.37858510e-01
-2.23962232e-01 -8.81321728e-01 1.09601676e+00 5.60213745e-01
3.16581577e-01 -1.02899599e+00 -1.83927521e-01 -1.48330703e-01
-1.46002457e-01 -7.24118829e-01 -5.67712545e-01 5.51472187e-01
-6.85485423e-01 -8.02952290e-01 -6.35471225e-01 -6.97247922e-01
7.19545722e-01 -2.90479898e-01 9.65215921e-01 -2.48097196e-01
-1.71293065e-01 1.45419821e-01 -4.69165832e-01 -6.75699592e-01
-1.00007296e+00 5.79097755e-02 1.43717691e-01 -6.60879090e-02
3.02635670e-01 -7.23945379e-01 7.71078095e-02 3.81703734e-01
-1.26474881e+00 -1.16901539e-01 5.41789591e-01 8.45243216e-01
5.15769184e-01 4.30557460e-01 7.24479198e-01 -1.40116680e+00
1.01939952e+00 -3.64185035e-01 -1.53964207e-01 6.31283700e-01
-8.56163442e-01 3.65134329e-01 8.06594551e-01 -1.01595533e+00
-1.08447969e+00 -9.91226286e-02 -1.18881240e-01 -5.20945132e-01
-2.65573382e-01 1.86573818e-01 -6.38522089e-01 5.08810626e-03
9.69256103e-01 3.91816318e-01 -4.11537103e-02 -3.09984058e-01
6.52783334e-01 5.05728543e-01 6.58357143e-01 -8.86521399e-01
1.06794977e+00 1.28073633e-01 -2.55910337e-01 -3.50401044e-01
-4.43148702e-01 -5.36361709e-02 -6.20842576e-01 1.42797768e-01
6.22875333e-01 -5.00318289e-01 -3.78938407e-01 6.17987454e-01
-9.12940562e-01 -2.82186002e-01 -8.02653372e-01 2.07824498e-01
-3.76803696e-01 3.18822920e-01 -3.04136395e-01 -5.43323576e-01
-2.91163713e-01 -1.06567383e+00 8.12443018e-01 7.18416646e-02
-3.59116733e-01 -9.76364970e-01 4.09935676e-02 1.80296034e-01
2.95770496e-01 4.66516972e-01 1.40312243e+00 -1.29968309e+00
5.57755530e-02 -3.61881852e-01 2.02316627e-01 5.75644076e-01
4.83260036e-01 1.43087516e-02 -1.11754715e+00 -3.22666705e-01
6.60273507e-02 -2.85287529e-01 7.48322189e-01 -1.25278369e-01
1.08999586e+00 -3.06542933e-01 -4.71061230e-01 7.11801410e-01
9.79083061e-01 5.14222920e-01 6.20546877e-01 2.04247981e-01
6.02377713e-01 5.47825873e-01 3.65166783e-01 1.29169226e-01
-1.81307033e-01 4.38787073e-01 1.46187946e-01 -1.11446500e-01
-2.08093941e-01 -4.78194773e-01 3.85575324e-01 5.46425521e-01
3.90464008e-01 -7.85668612e-01 -6.70694113e-01 2.49861434e-01
-1.41371596e+00 -8.58756781e-01 2.71576673e-01 2.17042232e+00
1.31405354e+00 2.70376980e-01 -1.39589444e-01 4.47979391e-01
9.13901329e-01 -1.54797539e-01 -6.28162980e-01 -3.20317030e-01
-2.10575372e-01 3.48694354e-01 4.10828561e-01 3.55890125e-01
-1.05218017e+00 9.21025574e-01 6.78425598e+00 1.16053963e+00
-9.49697316e-01 -8.83603171e-02 6.03616536e-01 -8.94877166e-02
-7.70057023e-01 -3.87599707e-01 -8.49472165e-01 6.39516413e-01
4.02893633e-01 -2.65125424e-01 3.06166381e-01 6.60495639e-01
-5.35508335e-01 4.41479325e-01 -1.38201535e+00 6.66325092e-01
2.26361856e-01 -9.90905702e-01 6.17759466e-01 -2.06378505e-01
9.27843690e-01 -6.04548216e-01 4.57051843e-01 4.43725377e-01
6.80717885e-01 -1.07615089e+00 6.87797308e-01 3.49562168e-01
1.02629507e+00 -8.59510958e-01 4.12996709e-01 4.84167576e-01
-6.15396678e-01 -1.83580667e-01 -2.04405740e-01 3.55122507e-01
-1.37273550e-01 5.16871214e-01 -1.00422835e+00 6.09128952e-01
4.94275749e-01 3.53137016e-01 -6.85133576e-01 8.03764820e-01
-4.04997826e-01 3.88484120e-01 -2.88390398e-01 -6.84567317e-02
-2.97827929e-01 3.25299054e-02 7.36402094e-01 1.09061515e+00
2.85335839e-01 -2.71480352e-01 -1.17190126e-02 1.04669392e+00
-1.44163027e-01 -2.44541485e-02 -5.55900514e-01 -1.00473009e-01
8.81156564e-01 8.88234317e-01 -7.25498497e-01 -1.09905243e-01
1.28476590e-01 1.07125080e+00 9.80604813e-02 3.71295035e-01
-8.95629048e-01 -6.38540804e-01 6.88824236e-01 -4.89486568e-02
1.01916194e-01 2.94034302e-01 -5.59149563e-01 -9.92153823e-01
8.12898427e-02 -1.24767792e+00 4.89445776e-01 -4.55619633e-01
-1.59489703e+00 6.51194751e-01 2.13387385e-01 -1.21679294e+00
-9.59692970e-02 -5.62545657e-01 -6.08197510e-01 8.78074706e-01
-1.16261673e+00 -1.04063368e+00 -1.62777305e-01 5.84347308e-01
4.18532729e-01 -5.17801344e-01 9.40771878e-01 2.39919052e-01
-4.78089035e-01 1.27622187e+00 2.10672721e-01 -6.07699044e-02
1.07790911e+00 -1.40232980e+00 3.25859040e-01 8.25999439e-01
9.10929888e-02 7.26366818e-01 7.71026492e-01 -6.91171408e-01
-9.52864170e-01 -1.26427758e+00 6.41514003e-01 -8.06487203e-01
4.04144764e-01 -5.94174385e-01 -8.42517436e-01 6.91022277e-01
-2.28551105e-01 -3.30221027e-01 9.13534522e-01 -5.52566014e-02
-6.62382483e-01 -1.76879391e-01 -1.53905880e+00 7.54824758e-01
9.16698277e-01 -3.91758442e-01 -7.83567607e-01 3.26364785e-01
1.13822699e+00 -4.56744075e-01 -7.09513426e-01 6.73841834e-01
3.10129434e-01 -7.56567895e-01 1.06051135e+00 -7.02485681e-01
4.35628742e-01 -1.48787245e-01 -1.89691484e-01 -1.46095908e+00
-2.67230421e-01 -6.29634678e-01 1.97539311e-02 1.69179118e+00
6.58061743e-01 -1.27202570e+00 5.85519671e-01 5.96156061e-01
-1.74885511e-01 -6.23637676e-01 -8.08162212e-01 -1.11500549e+00
5.94954789e-01 -2.94919133e-01 1.14385283e+00 1.28647137e+00
-9.95323062e-02 1.12707175e-01 -9.37554389e-02 5.65879159e-02
2.91642517e-01 -1.72776341e-01 6.03237629e-01 -1.26655841e+00
-5.03688693e-01 -6.64203107e-01 -4.45929050e-01 -7.95762420e-01
2.31288701e-01 -1.25625622e+00 8.41467306e-02 -9.37868893e-01
1.89049002e-02 -7.56577909e-01 -3.71747941e-01 7.19146729e-01
-4.32981223e-01 6.66493997e-02 8.40658322e-02 1.87564149e-01
-1.12621836e-01 6.03478968e-01 1.08513141e+00 -3.03983063e-01
-4.12597448e-01 1.77916810e-01 -1.07050061e+00 4.41616952e-01
8.19110394e-01 -7.07365990e-01 -9.31456923e-01 -4.00266349e-01
5.92271052e-02 -4.02955174e-01 3.64561826e-02 -9.73286867e-01
6.16952144e-02 -1.99518219e-01 5.81507027e-01 -2.17147931e-01
1.55442178e-01 -7.70934999e-01 2.49793187e-01 3.71854275e-01
-7.27366805e-01 -1.28514051e-01 3.27746868e-01 5.26314437e-01
1.32413376e-02 -3.62455100e-01 8.13479722e-01 2.53426760e-01
-2.82269567e-01 6.39246032e-02 -1.35489181e-01 3.22604954e-01
1.26773334e+00 -3.12902212e-01 -5.63202083e-01 -1.21866455e-02
-6.38742983e-01 3.39344516e-02 4.86775964e-01 7.52390325e-01
3.38823080e-01 -1.37172103e+00 -6.81376994e-01 5.78599453e-01
7.15451166e-02 -4.13006134e-02 -1.24267109e-01 2.48737410e-01
-2.55410731e-01 7.09867999e-02 -1.93949625e-01 -5.49427211e-01
-1.21548605e+00 7.73522139e-01 4.63732511e-01 -1.58742994e-01
-5.45207500e-01 1.18731344e+00 3.98511738e-01 -9.80649233e-01
2.99580514e-01 -1.71239361e-01 -4.71152104e-02 1.58520062e-02
4.26205099e-01 2.44841039e-01 1.00324407e-01 -1.15459718e-01
-1.90492749e-01 1.50140241e-01 -2.66753137e-01 -9.72520560e-02
1.04251504e+00 2.74940342e-01 -2.66401656e-02 4.40465778e-01
9.99124587e-01 8.46751407e-02 -1.14262319e+00 -1.18434571e-01
-6.60054982e-02 -4.83915240e-01 -4.52622116e-01 -1.15925610e+00
-9.46928620e-01 6.23735309e-01 6.55185044e-01 3.40246260e-01
1.27641237e+00 -1.20106168e-01 4.65775579e-01 2.94349730e-01
1.65327445e-01 -8.76519322e-01 5.62466495e-02 3.71360719e-01
8.67692471e-01 -8.18912268e-01 -1.47763968e-01 -6.48646355e-01
-7.24614084e-01 8.77456963e-01 7.51789987e-01 3.84370536e-01
3.97606462e-01 5.27844667e-01 1.13802508e-01 2.38063902e-01
-4.94987339e-01 2.57149071e-01 3.53336573e-01 1.04120374e+00
2.89543092e-01 2.35727280e-02 -1.74223065e-01 8.39226782e-01
-5.49946606e-01 -3.84181857e-01 1.75369903e-01 8.58205855e-01
-2.48824239e-01 -1.41270006e+00 -4.17969733e-01 5.07244051e-01
-1.85780004e-01 -1.87681317e-01 -7.00016141e-01 8.48234236e-01
4.84473616e-01 9.25822675e-01 1.54286074e-02 -4.91601974e-01
4.09806579e-01 4.60619032e-01 5.24291694e-01 -6.57413781e-01
-6.77314818e-01 -3.96942437e-01 -1.84136897e-01 6.00778908e-02
5.89508303e-02 -7.37176478e-01 -1.15477526e+00 -1.53486595e-01
-4.10407364e-01 1.13445066e-01 4.12056267e-01 9.65385318e-01
1.94397584e-01 7.56915987e-01 5.75487733e-01 -3.30552727e-01
-9.20374334e-01 -8.22740078e-01 -5.78734577e-01 6.00368321e-01
4.33198214e-02 -6.62562013e-01 -6.53349876e-01 1.15619339e-01]
|
[10.223586082458496, 3.2241275310516357]
|
e0636bc0-9ba1-4589-a929-09e6ceb232cd
|
amortized-synthesis-of-constrained
|
2106.09019
| null |
https://arxiv.org/abs/2106.09019v2
|
https://arxiv.org/pdf/2106.09019v2.pdf
|
Amortized Synthesis of Constrained Configurations Using a Differentiable Surrogate
|
In design, fabrication, and control problems, we are often faced with the task of synthesis, in which we must generate an object or configuration that satisfies a set of constraints while maximizing one or more objective functions. The synthesis problem is typically characterized by a physical process in which many different realizations may achieve the goal. This many-to-one map presents challenges to the supervised learning of feed-forward synthesis, as the set of viable designs may have a complex structure. In addition, the non-differentiable nature of many physical simulations prevents efficient direct optimization. We address both of these problems with a two-stage neural network architecture that we may consider to be an autoencoder. We first learn the decoder: a differentiable surrogate that approximates the many-to-one physical realization process. We then learn the encoder, which maps from goal to design, while using the fixed decoder to evaluate the quality of the realization. We evaluate the approach on two case studies: extruder path planning in additive manufacturing and constrained soft robot inverse kinematics. We compare our approach to direct optimization of the design using the learned surrogate, and to supervised learning of the synthesis problem. We find that our approach produces higher quality solutions than supervised learning, while being competitive in quality with direct optimization, at a greatly reduced computational cost.
|
['Szymon Rusinkiewicz', 'Ryan P. Adams', 'Tianju Xue', 'Xingyuan Sun']
|
2021-06-16
| null |
http://proceedings.neurips.cc/paper/2021/hash/9d38e6eab92b2aeb0a83b570188d5a1a-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/9d38e6eab92b2aeb0a83b570188d5a1a-Paper.pdf
|
neurips-2021-12
|
['physical-simulations']
|
['miscellaneous']
|
[ 2.85846055e-01 1.94434240e-01 4.48050769e-03 -3.01153541e-01
-6.32650077e-01 -4.47677225e-01 5.33997476e-01 -1.37397274e-01
8.16023722e-02 7.50594735e-01 -6.53241798e-02 -1.61911249e-01
-5.84977448e-01 -8.78126025e-01 -1.13142192e+00 -7.17629552e-01
2.07582712e-01 9.12051857e-01 -5.36097705e-01 -2.89633304e-01
2.84880996e-01 6.48390293e-01 -1.45540571e+00 1.30990103e-01
4.10319328e-01 1.28920376e+00 4.61955130e-01 7.64816642e-01
9.88072008e-02 5.57444751e-01 -4.65393841e-01 -9.04532671e-02
2.70784378e-01 -4.06535268e-01 -6.88065290e-01 3.10577631e-01
6.42330274e-02 -2.34879062e-01 1.28921038e-02 9.52060759e-01
3.49946022e-01 1.65500671e-01 9.11037982e-01 -9.21754897e-01
-8.14108133e-01 5.61140180e-01 1.57323450e-01 -5.86845160e-01
3.06576610e-01 3.89554530e-01 8.81115258e-01 -9.66435909e-01
7.59715021e-01 1.22054255e+00 4.29380864e-01 2.98610866e-01
-1.41292500e+00 -9.04034898e-02 -1.19278491e-01 -2.01230243e-01
-1.17158127e+00 -5.65233409e-01 7.53729105e-01 -7.99370885e-01
1.03058302e+00 -2.21890211e-01 4.89221305e-01 8.59715044e-01
8.75677109e-01 4.24852163e-01 7.88734198e-01 -3.79831702e-01
7.21283734e-01 -5.74550033e-02 -4.02887136e-01 7.65736878e-01
-2.49671564e-02 6.82622015e-01 -1.12938777e-01 9.44396704e-02
9.79841590e-01 -1.33888423e-01 -1.30530730e-01 -5.37959874e-01
-1.05417335e+00 7.76528835e-01 4.65081215e-01 5.68951219e-02
-6.20295465e-01 5.01238823e-01 -5.48367202e-02 7.52722621e-01
-1.11233346e-01 1.39935350e+00 -6.05796695e-01 1.17517971e-01
-7.29363859e-01 5.58196187e-01 1.28903079e+00 1.13967395e+00
7.76892781e-01 4.26378965e-01 4.11495477e-01 6.11772537e-01
4.89827156e-01 2.63611138e-01 1.84107438e-01 -1.26421785e+00
2.06972420e-01 1.17913246e-01 3.09035838e-01 -7.50338316e-01
-2.91617453e-01 -4.49722528e-01 -6.08408511e-01 8.27757180e-01
2.10526153e-01 -4.38193560e-01 -1.00034297e+00 1.49473310e+00
1.15061283e-01 -2.94030994e-01 1.62408590e-01 9.39312279e-01
3.58290583e-01 1.01549554e+00 -2.87678719e-01 -2.12510973e-01
6.61212981e-01 -8.96718979e-01 -6.52635217e-01 -2.40174726e-01
4.01749611e-01 -1.03137565e+00 7.46258378e-01 6.40990853e-01
-1.46772647e+00 -5.85543334e-01 -1.50706661e+00 2.05480337e-01
-3.20699334e-01 2.72792876e-01 4.24692243e-01 1.36194617e-01
-9.16312516e-01 1.46031940e+00 -7.52774417e-01 1.02438003e-01
-1.46818506e-02 8.61945748e-01 -9.37291235e-02 1.01204552e-01
-7.92904377e-01 1.36989450e+00 4.81721103e-01 2.51438886e-01
-9.31189775e-01 -7.93910801e-01 -8.85730922e-01 1.74228117e-01
5.57994664e-01 -8.91402900e-01 1.64230454e+00 -9.32909667e-01
-2.19566369e+00 1.70655534e-01 2.78746605e-01 -2.53236108e-02
2.28381351e-01 -2.37245718e-03 -2.40522370e-01 -3.12327921e-01
-8.15504342e-02 6.09474659e-01 1.03774619e+00 -1.42040861e+00
-2.32424304e-01 2.86170165e-04 7.17728809e-02 1.17258221e-01
4.29702878e-01 -3.78219843e-01 -6.76053017e-02 -5.04585326e-01
1.89276874e-01 -1.03122067e+00 -5.98577321e-01 2.12428078e-01
-4.63232309e-01 1.99969620e-01 5.78114867e-01 -6.79308474e-01
6.76261663e-01 -1.78480887e+00 8.15210462e-01 3.83749336e-01
2.65397877e-02 -2.33428165e-01 -1.81123346e-01 6.33876085e-01
-1.30175091e-02 -1.82247743e-01 -2.44676292e-01 -2.76501179e-01
1.78109884e-01 3.66792172e-01 -6.27143309e-02 4.53719735e-01
6.22038782e-01 9.62927997e-01 -7.62226105e-01 3.44387465e-03
3.05951118e-01 1.46349445e-01 -6.35009110e-01 4.48005736e-01
-9.02185798e-01 3.55667949e-01 -5.75693130e-01 5.57890534e-01
1.44468471e-02 -2.60275692e-01 2.52594471e-01 -2.57253587e-01
-2.11659178e-01 1.62588388e-01 -1.31746697e+00 1.69379056e+00
-9.06557202e-01 3.78389388e-01 4.41426277e-01 -1.21347988e+00
1.18869829e+00 2.33376026e-01 4.72953856e-01 -4.89700228e-01
5.93255877e-01 4.98788089e-01 2.70240545e-01 -4.77802455e-01
3.96596938e-01 -5.38356245e-01 -2.45309368e-01 3.75950873e-01
3.11948746e-01 -9.64433610e-01 -6.71595931e-02 -4.55448389e-01
1.06785095e+00 2.94331551e-01 1.49711862e-01 -3.88914913e-01
1.63447455e-01 1.47678271e-01 4.27639037e-01 4.19005513e-01
5.53953886e-01 6.45041645e-01 5.65750837e-01 -3.81449997e-01
-1.55287027e+00 -1.10848951e+00 2.18908310e-01 3.53209794e-01
4.79365923e-02 5.16975559e-02 -5.24728954e-01 -7.15544894e-02
1.70994967e-01 8.33220124e-01 -4.03711826e-01 -2.68605441e-01
-8.45993042e-01 -7.58816674e-02 -1.27333164e-01 4.15478170e-01
-1.27468720e-01 -1.10532808e+00 -4.75588411e-01 7.19463706e-01
6.56462431e-01 -8.56300056e-01 -2.92641401e-01 5.29385686e-01
-9.06846523e-01 -8.26769292e-01 -4.05826032e-01 -1.06842887e+00
7.11713016e-01 -5.65447569e-01 1.04701233e+00 -1.62433192e-01
-4.15081382e-01 1.23748928e-01 1.19768776e-01 -2.11541653e-01
-1.11945665e+00 -3.65506671e-02 1.86464056e-01 -3.58366162e-01
-4.59197611e-01 -6.91940248e-01 -2.65249372e-01 1.95031613e-01
-6.17638230e-01 1.74506426e-01 8.83493960e-01 1.00253129e+00
7.26110578e-01 4.50915813e-01 7.23713338e-01 -3.69195938e-01
7.60131240e-01 -3.87233138e-01 -8.72665048e-01 1.82993606e-01
-5.92874825e-01 5.78182638e-01 1.11376572e+00 -4.38634604e-01
-8.32253456e-01 5.42113841e-01 -2.45463073e-01 -6.85095787e-01
-7.45794103e-02 6.11671805e-01 -4.60864365e-01 -1.95300028e-01
3.40331286e-01 -1.63584515e-01 5.10770440e-01 -4.44322556e-01
6.21061862e-01 2.91017026e-01 4.92084920e-01 -9.31232035e-01
6.53908908e-01 -2.47033224e-01 3.44808131e-01 -7.86351025e-01
-2.70079464e-01 3.28177869e-01 -4.66563940e-01 -3.05727631e-01
7.24647641e-01 -4.40835685e-01 -7.29733050e-01 1.77357614e-01
-1.23973250e+00 -6.18194640e-01 -6.34819627e-01 4.37654287e-01
-1.27726352e+00 -3.54945034e-01 -5.76185048e-01 -4.49544638e-01
2.45358348e-02 -1.68626523e+00 1.06043696e+00 -2.39295065e-02
-5.53743541e-01 -9.34247136e-01 -6.16140962e-02 -2.63897926e-01
4.17863309e-01 6.11295819e-01 1.28850567e+00 -1.73116997e-01
-8.25266421e-01 -2.41580904e-01 2.85920620e-01 4.11147445e-01
3.38943005e-01 2.44609937e-02 -4.33383167e-01 -1.97849244e-01
3.49167168e-01 -3.01316559e-01 2.81882137e-01 6.13032758e-01
1.00210083e+00 -5.44114232e-01 -1.27161667e-01 3.93266499e-01
1.60156167e+00 6.02023065e-01 4.13592249e-01 -2.67205536e-02
4.57583457e-01 7.96362996e-01 4.73313779e-01 1.81918740e-01
-1.84621215e-01 7.77083993e-01 3.75634134e-01 1.83748007e-01
-1.38736427e-01 -2.62536138e-01 2.76316285e-01 8.99766922e-01
1.90781876e-01 -7.46768937e-02 -8.34311962e-01 3.10456306e-01
-1.84170032e+00 -5.65587461e-01 4.06658590e-01 1.89509630e+00
6.62781477e-01 3.12739164e-01 -1.31818280e-01 1.03025608e-01
4.83443111e-01 -2.52586722e-01 -6.80950224e-01 -9.06548500e-01
3.36479515e-01 4.47614163e-01 3.69489819e-01 6.77821934e-01
-7.58531511e-01 5.35394728e-01 6.46693611e+00 5.81667483e-01
-1.37714088e+00 -3.80613595e-01 5.63386559e-01 -1.00205727e-01
-2.31176704e-01 -8.19479600e-02 -5.00953317e-01 4.66761678e-01
1.18210232e+00 -7.30252713e-02 9.48690116e-01 1.01479506e+00
2.58998334e-01 7.75918067e-02 -1.79192805e+00 6.97233737e-01
-2.58586466e-01 -1.91134036e+00 -2.34021619e-01 1.06186189e-01
1.00971222e+00 -4.43461597e-01 1.44738138e-01 1.44002274e-01
5.25354862e-01 -1.31885147e+00 1.13900316e+00 6.48401678e-01
6.27440810e-01 -8.02948534e-01 3.39270264e-01 4.28810239e-01
-9.48995709e-01 -3.26320708e-01 -2.53193197e-04 -1.47674322e-01
5.62818348e-01 4.33411330e-01 -8.41976225e-01 3.84832114e-01
-1.59727875e-02 5.06138325e-01 2.64663875e-01 8.44596028e-01
-6.25745878e-02 4.17779386e-02 -3.57896179e-01 -5.23632407e-01
3.32747459e-01 -4.16449338e-01 4.39615935e-01 7.25511432e-01
5.94046116e-01 1.54910848e-01 3.82860601e-01 1.43563533e+00
-1.44591883e-01 -3.94657612e-01 -8.56437683e-01 -3.24025959e-01
3.23576093e-01 6.86492682e-01 -6.39482379e-01 -3.69836651e-02
-1.01516984e-01 5.38833499e-01 5.41892350e-02 2.66498983e-01
-6.65320992e-01 -3.38142157e-01 5.59827566e-01 1.20103896e-01
4.18533683e-01 -6.16052687e-01 -6.36914253e-01 -5.34675360e-01
4.23133224e-02 -1.02959037e+00 -4.97917295e-01 -8.43061209e-01
-1.11543357e+00 4.21924770e-01 -5.59930317e-02 -1.06057632e+00
-6.79490268e-01 -8.86520982e-01 -5.48334002e-01 7.86166608e-01
-9.52972889e-01 -7.80894220e-01 2.68933952e-01 -1.85546637e-01
8.61879945e-01 -3.63502800e-01 7.37329543e-01 2.23283231e-01
-3.43297333e-01 6.12425730e-02 2.42647901e-01 -3.65560323e-01
1.41735747e-01 -1.21614003e+00 3.82281750e-01 3.18260103e-01
-2.41914302e-01 5.12671471e-01 9.33642924e-01 -7.95883834e-01
-2.18397927e+00 -1.02762306e+00 4.75575745e-01 -1.71219885e-01
8.08373809e-01 -1.19712286e-01 -5.62337041e-01 5.37551701e-01
1.11154970e-02 -4.09956992e-01 3.54080349e-02 -1.51067063e-01
4.54605192e-01 2.35207126e-01 -1.11712933e+00 6.97459519e-01
7.34503567e-01 -2.25942343e-01 -6.40408516e-01 3.06800067e-01
7.73660541e-01 -5.78420401e-01 -1.22383976e+00 3.67675483e-01
6.33533359e-01 -2.83021063e-01 9.68005180e-01 -5.89212894e-01
1.19843483e+00 -2.47435197e-01 -3.09779465e-01 -1.70803916e+00
-4.52152371e-01 -8.12306702e-01 -3.34106207e-01 8.67565215e-01
7.31288314e-01 -3.82507592e-01 8.48861933e-01 7.59060442e-01
-6.14626586e-01 -1.35388732e+00 -6.21351421e-01 -1.03855920e+00
2.11387545e-01 -3.30913030e-02 8.62205803e-01 5.36953330e-01
-3.46900702e-01 4.87669766e-01 -2.14947954e-01 2.12867737e-01
5.33596933e-01 3.17711830e-01 4.94802028e-01 -9.41706240e-01
-7.93863714e-01 -6.03525817e-01 -1.36443138e-01 -1.13356400e+00
2.11326182e-01 -7.26963222e-01 7.30950594e-01 -1.51626039e+00
-6.00071669e-01 -6.10594034e-01 2.58604914e-01 -6.71103643e-03
5.61002791e-01 -6.32310271e-01 8.75063837e-02 -6.55737743e-02
1.49942413e-01 7.93313503e-01 1.50136387e+00 -3.59440416e-01
-4.08284694e-01 -1.53904166e-02 -4.65859711e-01 6.26355290e-01
7.74166584e-01 -3.88978004e-01 -3.63066375e-01 -6.37017250e-01
2.79224634e-01 7.33998835e-01 1.45233184e-01 -1.15074718e+00
3.00523967e-01 -4.06788468e-01 4.28685904e-01 -3.70340914e-01
6.21694207e-01 -9.12975490e-01 5.64252794e-01 5.45703471e-01
-4.47967887e-01 1.73994154e-01 4.00998443e-02 3.03949088e-01
-7.17420727e-02 -4.58683670e-01 9.18574154e-01 -3.05404246e-01
-4.37921077e-01 1.76889777e-01 -4.29239631e-01 -5.19491196e-01
7.65629888e-01 -1.44609421e-01 1.38204962e-01 -2.70626962e-01
-9.96160746e-01 2.32987657e-01 5.85807085e-01 4.75358456e-01
6.85844719e-01 -1.38992715e+00 -5.17087281e-01 3.21610719e-01
-4.34618205e-01 2.30494961e-01 -1.85345694e-01 1.78626001e-01
-6.43263876e-01 3.29501301e-01 -3.70351851e-01 -4.56144214e-01
-4.79974717e-01 5.44882298e-01 4.68684793e-01 -6.98996484e-02
-3.97093713e-01 6.54575884e-01 -2.10914671e-01 -5.31521976e-01
-2.47296374e-02 -5.26770055e-01 4.29691344e-01 -2.30030179e-01
-4.11451720e-02 3.63233745e-01 2.23250687e-01 -3.01809639e-01
9.09854919e-02 6.51958287e-01 2.18172073e-01 -3.46738935e-01
1.62397766e+00 4.68437880e-01 -5.88174835e-02 3.12819004e-01
1.40884852e+00 -6.47652388e-01 -1.39240396e+00 1.30558878e-01
4.44819443e-02 -1.17937572e-01 2.09375232e-01 -7.07760274e-01
-1.08427596e+00 4.99808311e-01 2.49873027e-01 7.36993924e-02
7.26986647e-01 -1.27461091e-01 8.85765374e-01 7.74947882e-01
4.33263034e-01 -1.31154096e+00 3.46248269e-01 5.57636917e-01
1.37467396e+00 -9.35415745e-01 7.09524751e-02 -2.48517111e-01
-2.95267850e-01 1.45599508e+00 4.53471601e-01 -6.74322367e-01
7.52870560e-01 8.75446379e-01 -2.55236059e-01 -3.79123211e-01
-7.87457228e-01 2.47260019e-01 3.22929770e-01 1.84221327e-01
7.92821720e-02 7.09567890e-02 1.47892639e-01 3.42296660e-01
-5.49471021e-01 1.51470423e-01 3.80037814e-01 1.24373865e+00
-5.21214545e-01 -1.37437892e+00 -3.04496765e-01 5.54894328e-01
5.41998446e-02 4.52627897e-01 -2.82540321e-01 7.54526317e-01
8.86819512e-02 7.24836826e-01 -6.53797295e-03 -4.58619565e-01
5.13959467e-01 -6.97917268e-02 8.65319669e-01 -8.25660586e-01
-4.98243600e-01 5.65974414e-02 4.52650696e-01 -5.90041757e-01
2.45516092e-01 -6.53978109e-01 -1.31306267e+00 -2.77084619e-01
-4.15856898e-01 -2.37446465e-02 1.08419597e+00 9.73665059e-01
2.42328495e-01 8.62673402e-01 8.56194079e-01 -1.35010731e+00
-9.65569258e-01 -2.91378319e-01 -4.27553743e-01 -6.23650998e-02
5.74765146e-01 -8.85637581e-01 6.17738590e-02 1.59397498e-01]
|
[5.8661112785339355, 3.303075075149536]
|
7ef65090-3f8a-4c01-80d3-e5756744979f
|
a-hierarchical-interactive-network-for-joint
|
2208.11283
| null |
https://arxiv.org/abs/2208.11283v1
|
https://arxiv.org/pdf/2208.11283v1.pdf
|
A Hierarchical Interactive Network for Joint Span-based Aspect-Sentiment Analysis
|
Recently, some span-based methods have achieved encouraging performances for joint aspect-sentiment analysis, which first extract aspects (aspect extraction) by detecting aspect boundaries and then classify the span-level sentiments (sentiment classification). However, most existing approaches either sequentially extract task-specific features, leading to insufficient feature interactions, or they encode aspect features and sentiment features in a parallel manner, implying that feature representation in each task is largely independent of each other except for input sharing. Both of them ignore the internal correlations between the aspect extraction and sentiment classification. To solve this problem, we novelly propose a hierarchical interactive network (HI-ASA) to model two-way interactions between two tasks appropriately, where the hierarchical interactions involve two steps: shallow-level interaction and deep-level interaction. First, we utilize cross-stitch mechanism to combine the different task-specific features selectively as the input to ensure proper two-way interactions. Second, the mutual information technique is applied to mutually constrain learning between two tasks in the output layer, thus the aspect input and the sentiment input are capable of encoding features of the other task via backpropagation. Extensive experiments on three real-world datasets demonstrate HI-ASA's superiority over baselines.
|
['Zhongshi He', 'Fuzhen Zhuang', 'Zhao Zhang', 'Jinglong Du', 'Wei Chen']
|
2022-08-24
| null |
https://aclanthology.org/2022.coling-1.611
|
https://aclanthology.org/2022.coling-1.611.pdf
|
coling-2022-10
|
['aspect-extraction']
|
['natural-language-processing']
|
[ 2.48062938e-01 -6.09118380e-02 -1.93273231e-01 -6.68452680e-01
-5.28678179e-01 -4.97889280e-01 6.43858075e-01 -5.32904342e-02
-4.21540827e-01 5.19796550e-01 2.68367022e-01 -1.59303144e-01
-2.99525913e-02 -6.87896729e-01 -4.80445832e-01 -7.92327642e-01
2.14406654e-01 1.94125459e-01 3.45945776e-01 -2.82497764e-01
2.95742780e-01 -2.38076180e-01 -1.51382983e+00 7.06072032e-01
9.36592638e-01 1.27891684e+00 1.63297039e-02 4.50120419e-01
-5.03874004e-01 6.34202957e-01 -5.57970285e-01 -4.40432757e-01
1.13961205e-01 -4.62213069e-01 -6.54152393e-01 2.23018631e-01
-1.64937705e-01 9.56672356e-02 1.16131324e-02 1.04815769e+00
3.04663867e-01 1.16013080e-01 6.20602190e-01 -1.28763211e+00
-6.52906358e-01 7.27155864e-01 -9.10092652e-01 -9.93886292e-02
9.92368907e-02 1.61958054e-01 1.23656452e+00 -9.14759040e-01
7.75156245e-02 1.08740282e+00 6.24154985e-01 1.67495295e-01
-9.44449663e-01 -7.38994300e-01 8.09987128e-01 -8.16181377e-02
-9.57532108e-01 -2.65611708e-01 1.10404170e+00 -3.15818161e-01
1.06426489e+00 1.73144922e-01 9.61147904e-01 7.82104850e-01
3.99146348e-01 1.22638977e+00 1.16692722e+00 -3.49648530e-03
-5.37133180e-02 5.85901916e-01 6.44209027e-01 4.69430506e-01
2.41045743e-01 -1.06818818e-01 -7.62208700e-01 -9.74040255e-02
4.62915689e-01 1.74968671e-02 -1.51916444e-01 -6.77830428e-02
-1.28327346e+00 6.55890226e-01 4.14232254e-01 4.24108446e-01
-3.57506901e-01 -2.66712487e-01 6.32874310e-01 5.77025115e-01
6.30762160e-01 3.90009433e-01 -9.32614207e-01 1.51769131e-01
-7.31294453e-01 2.61741914e-02 7.97429264e-01 9.71219778e-01
1.01217091e+00 -9.05899405e-02 -5.57291389e-01 7.37445593e-01
5.12632847e-01 1.34045228e-01 7.57239163e-01 -9.87871364e-02
7.49960721e-01 1.24466288e+00 -2.40998253e-01 -1.00051498e+00
-5.28959692e-01 -6.03318572e-01 -8.90963852e-01 -1.11875042e-01
2.93828603e-02 -3.74007612e-01 -9.32424664e-01 1.77559257e+00
3.72943372e-01 -1.80988103e-01 2.18595594e-01 8.18121552e-01
9.84351456e-01 5.17211318e-01 2.14649752e-01 -2.31565952e-01
1.74372506e+00 -1.31439555e+00 -7.40591228e-01 -5.79262674e-01
4.53115642e-01 -8.60164046e-01 1.19617593e+00 2.43697360e-01
-9.65212047e-01 -6.50181174e-01 -1.22394562e+00 -5.48299365e-02
-7.05340028e-01 2.22830951e-01 9.22714770e-01 3.35187793e-01
-7.42627680e-01 1.82695791e-01 -5.64565003e-01 1.66306004e-01
3.34526807e-01 6.74305081e-01 -3.03112209e-01 3.32979143e-01
-1.49608564e+00 4.06135112e-01 4.47336435e-01 4.18783695e-01
-3.99410367e-01 -4.30445015e-01 -9.22747254e-01 2.04679266e-01
3.44197243e-01 -8.56239676e-01 1.02697861e+00 -1.36069405e+00
-1.34708071e+00 5.65481007e-01 -2.55998492e-01 2.23969251e-01
2.30293065e-01 -4.11215872e-01 -4.87013310e-01 -3.32013041e-01
2.10535303e-01 5.45266449e-01 8.93410206e-01 -1.25619555e+00
-7.81663597e-01 -5.75959265e-01 2.58874208e-01 7.49961793e-01
-8.04952979e-01 -1.17888287e-01 -8.68546426e-01 -5.79696476e-01
2.92568594e-01 -6.49182498e-01 -4.08685267e-01 -4.48848844e-01
-6.70972824e-01 -3.22630525e-01 8.92121851e-01 -1.55747205e-01
1.21195042e+00 -2.11098409e+00 1.74369887e-01 8.90192240e-02
1.48798466e-01 2.15677410e-01 -4.01915431e-01 2.51366675e-01
-1.67775169e-01 -1.94726121e-02 -3.46456438e-01 -6.02988422e-01
-1.46527752e-01 -9.33208223e-03 -2.95289546e-01 1.08329192e-01
3.27979147e-01 1.05958140e+00 -7.19330370e-01 -5.45311332e-01
-1.79699704e-01 3.90741974e-01 -5.37592232e-01 4.01604980e-01
-2.13236630e-01 3.38542432e-01 -9.05790865e-01 6.35968983e-01
7.16576219e-01 -2.47463420e-01 2.17265934e-02 -5.00647664e-01
-4.96427454e-02 4.49863762e-01 -1.20162797e+00 1.51555979e+00
-5.90165734e-01 1.76735952e-01 -1.92357413e-02 -8.90109301e-01
8.82204592e-01 3.70346934e-01 3.44927520e-01 -4.74237323e-01
3.06865990e-01 1.24160573e-02 -1.85143985e-02 -4.98129308e-01
4.52726036e-01 -2.78007865e-01 -3.50802451e-01 8.58145833e-01
1.61059812e-01 -1.32010445e-01 1.49419084e-01 -8.66698008e-03
7.14279175e-01 1.91830486e-01 2.77887255e-01 -2.61203706e-01
9.79633212e-01 -6.94582686e-02 7.39520609e-01 5.08182347e-01
-1.71654150e-01 6.22545302e-01 8.73334944e-01 -4.75953072e-01
-3.45003754e-01 -6.76535547e-01 1.16357371e-01 1.21511161e+00
4.19621766e-01 -4.29687500e-01 -5.37380815e-01 -1.24682736e+00
-1.85464859e-01 4.17312980e-01 -8.71522963e-01 -3.85063797e-01
-2.86789238e-01 -1.03900313e+00 2.18517825e-01 5.66549480e-01
7.54219770e-01 -1.33780575e+00 -3.14207375e-01 1.27442047e-01
-4.74643484e-02 -1.07356513e+00 -6.09365284e-01 6.16014183e-01
-8.29993963e-01 -1.01444185e+00 -3.25644314e-01 -8.43956709e-01
9.13228571e-01 4.13575947e-01 1.11914158e+00 1.93717346e-01
2.33680680e-01 -8.71626809e-02 -2.49998346e-01 -2.94310123e-01
3.32104236e-01 4.76508021e-01 -9.40258279e-02 3.30299497e-01
7.55854309e-01 -7.43056953e-01 -7.28193641e-01 3.44948977e-01
-1.06958747e+00 1.84970334e-01 9.51448262e-01 1.00482714e+00
5.60513139e-01 1.89929992e-01 6.03287458e-01 -1.32057023e+00
9.78450894e-01 -4.84631956e-01 -2.95588583e-01 2.23218367e-01
-6.48725569e-01 -3.73577178e-02 8.72971833e-01 -3.32313538e-01
-1.27547133e+00 6.65733516e-02 -1.39760777e-01 -1.37639463e-01
-4.77072261e-02 8.33829641e-01 -5.62814772e-01 3.71238083e-01
1.03303403e-01 6.66763425e-01 -2.03568339e-01 -7.04460591e-02
2.11565852e-01 8.08113277e-01 5.59620596e-02 -3.57494920e-01
7.42391765e-01 3.61436725e-01 -4.69688773e-01 -3.61542493e-01
-1.29310608e+00 -4.55712527e-01 -7.49420285e-01 -5.72363883e-02
8.35563362e-01 -1.07542467e+00 -6.18035436e-01 7.43439734e-01
-1.07641780e+00 1.03286542e-01 -2.29410127e-01 4.83853787e-01
-2.84814775e-01 -2.10891617e-03 -5.14521122e-01 -7.10392118e-01
-5.61091363e-01 -1.46634829e+00 1.39545655e+00 5.67970753e-01
-1.72926605e-01 -1.04316819e+00 8.29442311e-03 3.30509633e-01
3.35940897e-01 -2.92391926e-01 9.47336793e-01 -9.00717199e-01
-2.81328440e-01 -3.32623720e-01 -4.06863421e-01 3.73206913e-01
4.45423007e-01 -2.99522728e-01 -1.19631827e+00 -3.15457471e-02
3.94857973e-01 -3.56229097e-01 1.07565963e+00 1.14435069e-01
9.91772413e-01 -3.40661645e-01 -2.65342951e-01 5.46355844e-01
1.25629807e+00 6.51735812e-02 3.30948591e-01 4.38042253e-01
8.09493005e-01 7.68310249e-01 6.44549608e-01 1.33039936e-01
6.44066632e-01 2.32049331e-01 1.65602162e-01 -2.52849191e-01
7.87132159e-02 -1.16715647e-01 5.73125720e-01 1.32290530e+00
1.59749612e-01 7.37426756e-03 -4.36160624e-01 4.61636633e-01
-1.99994111e+00 -5.23731232e-01 -9.14795622e-02 1.94680643e+00
1.04720902e+00 5.61072111e-01 -1.02298200e-01 9.73941684e-02
3.98831636e-01 5.92912257e-01 -7.61815131e-01 -4.42050874e-01
-2.33584587e-02 -3.88656914e-01 -4.49199183e-03 2.37998456e-01
-1.26603329e+00 9.64722812e-01 5.84112024e+00 6.95353985e-01
-1.06917894e+00 3.39147747e-02 8.52002084e-01 -1.35503858e-01
-7.28074551e-01 9.04769450e-02 -8.20732951e-01 2.85797745e-01
2.50712395e-01 4.18586358e-02 7.75813460e-02 7.72473216e-01
-1.56366318e-01 -1.45273224e-01 -9.25189078e-01 5.00478923e-01
4.72037643e-02 -7.55443454e-01 4.08348441e-01 -2.83350587e-01
8.18879664e-01 -1.77310884e-01 8.04602429e-02 6.67998910e-01
1.39379859e-01 -6.81375027e-01 4.44378763e-01 4.38651413e-01
4.63741928e-01 -8.74256611e-01 9.68483448e-01 3.17373782e-01
-1.61100471e+00 1.21089891e-01 -3.36363047e-01 -1.11600257e-01
7.88929537e-02 8.73355031e-01 -1.45772934e-01 7.24952042e-01
6.26276672e-01 8.27510118e-01 -4.56479728e-01 5.41222751e-01
-4.16742027e-01 4.13664162e-01 -1.01340257e-01 -6.33372813e-02
4.71615374e-01 -3.51960629e-01 5.82908571e-01 1.14209962e+00
-1.34017289e-01 -8.30324367e-02 2.51898289e-01 6.45628631e-01
1.25840649e-01 2.27147192e-01 -5.22398174e-01 -1.15287781e-01
-1.02982984e-03 1.56630182e+00 -8.32933545e-01 -3.26611847e-01
-7.54039407e-01 8.51776958e-01 5.13099074e-01 5.29956162e-01
-7.00171173e-01 -7.28291273e-01 8.00305188e-01 -3.81636620e-01
3.77152830e-01 -4.07640301e-02 -8.72610569e-01 -1.31366575e+00
3.47138613e-01 -8.77635121e-01 3.63411218e-01 -3.81195366e-01
-1.29695284e+00 1.04120517e+00 -3.63041222e-01 -1.41577148e+00
5.86311072e-02 -4.11209553e-01 -1.01551342e+00 9.81344223e-01
-1.76946723e+00 -1.37983012e+00 -1.07063510e-01 6.67312086e-01
7.62489080e-01 -2.41763279e-01 6.62417650e-01 8.59507769e-02
-7.14165270e-01 7.30918407e-01 -3.94892544e-01 6.77548721e-02
5.79621673e-01 -1.22220933e+00 3.17746490e-01 6.86602235e-01
-5.21797650e-02 8.55907440e-01 3.18518311e-01 -6.95148647e-01
-1.38248885e+00 -1.08178461e+00 1.02911365e+00 -2.69560039e-01
6.76209807e-01 -5.61101139e-01 -8.31166506e-01 6.09493673e-01
4.01807964e-01 -2.18256488e-01 1.01770055e+00 4.65309680e-01
-5.44131219e-01 -3.72133434e-01 -6.12625599e-01 7.22370863e-01
8.49718153e-01 -5.18856525e-01 -9.24341023e-01 1.45364851e-01
1.10751724e+00 -7.66541138e-02 -5.95176935e-01 7.09287465e-01
6.73575640e-01 -1.20071387e+00 5.99221110e-01 -6.37319505e-01
7.68013716e-01 -4.66549009e-01 -9.27007347e-02 -1.39274120e+00
-2.44586289e-01 -3.11293811e-01 1.60767860e-03 1.59926701e+00
7.78034449e-01 -8.24849427e-01 5.70102155e-01 4.79042977e-01
-1.54270045e-02 -1.31305027e+00 -4.73545820e-01 -3.58401299e-01
-2.19498560e-01 -4.25849110e-01 8.99275541e-01 8.63637984e-01
1.68801501e-01 9.87747610e-01 -2.46581078e-01 5.24496809e-02
1.58485904e-01 7.33680487e-01 5.87737501e-01 -9.74846601e-01
-4.54315484e-01 -7.51709044e-01 6.92621991e-02 -1.06687653e+00
1.51401132e-01 -7.55964279e-01 2.24563584e-01 -1.52143598e+00
4.84565079e-01 -5.78988373e-01 -7.26575255e-01 6.56638980e-01
-9.22579706e-01 1.16017111e-01 -4.57917042e-02 2.93914616e-01
-6.61095381e-01 9.64379489e-01 1.55925500e+00 -1.78721383e-01
-4.94548649e-01 2.84344494e-01 -1.39204669e+00 8.98309469e-01
6.63377464e-01 -4.66777384e-01 -7.27994323e-01 -6.04308367e-01
4.25703794e-01 -2.37726986e-01 -3.47734272e-01 -8.54223430e-01
4.69476938e-01 -4.61624451e-02 6.05007887e-01 -6.60467088e-01
2.09735528e-01 -8.36295724e-01 -4.83659774e-01 1.62457928e-01
-3.89073968e-01 -1.06994636e-01 2.35018268e-01 5.54803908e-01
-6.71922624e-01 1.34863588e-03 3.02748382e-01 -1.10537313e-01
-3.96209121e-01 5.57904303e-01 -2.66044796e-01 -1.22444734e-01
8.53097856e-01 -1.26004159e-01 -2.89558798e-01 -1.87487051e-01
-4.10185456e-01 6.80358887e-01 4.51014675e-02 6.77430689e-01
4.36962366e-01 -1.25114977e+00 -5.25326431e-01 6.16654575e-01
1.76327959e-01 2.52471626e-01 2.95299411e-01 1.03024352e+00
2.73469299e-01 2.92962730e-01 -4.42361385e-02 -4.90122795e-01
-9.88512099e-01 4.53477144e-01 1.12790093e-01 -9.07799840e-01
-1.81154430e-01 1.07181299e+00 8.24722946e-01 -8.25305939e-01
1.10069729e-01 1.11374043e-01 -8.01156580e-01 4.75932866e-01
5.21480381e-01 -3.54609460e-01 6.16605394e-02 -4.91159260e-01
-3.81125867e-01 8.01091611e-01 -5.65041244e-01 -1.58059299e-01
1.29621220e+00 -2.58367777e-01 -3.83184612e-01 7.45355844e-01
1.27309442e+00 -1.54575020e-01 -1.13481653e+00 -3.99432957e-01
-1.50424138e-01 -1.16716653e-01 1.95047278e-02 -7.31383860e-01
-1.42504120e+00 8.54243994e-01 -6.63289707e-03 4.32643205e-01
1.53639984e+00 -1.63453221e-01 8.56873691e-01 3.10398042e-01
-8.12571943e-02 -1.02598786e+00 9.11580175e-02 7.74541140e-01
7.03211248e-01 -1.20788324e+00 5.37572950e-02 -5.75415254e-01
-9.95110095e-01 9.12480950e-01 1.16549599e+00 -1.96714818e-01
9.96902168e-01 4.40461189e-01 2.61840016e-01 -3.40841144e-01
-1.19402313e+00 -3.16837132e-01 5.81153810e-01 2.73012400e-01
6.71176910e-01 -1.15742519e-01 -5.54587185e-01 1.44409847e+00
-2.18463004e-01 -2.10815027e-01 -7.80098587e-02 9.35009420e-01
-2.13320300e-01 -1.04986775e+00 -6.38562217e-02 7.70217061e-01
-4.34356809e-01 -4.27783847e-01 -4.74821329e-01 4.98218238e-01
7.64025748e-02 9.15871382e-01 1.34204142e-02 -7.62334526e-01
4.47928607e-01 1.10296212e-01 -1.43850446e-01 -6.58848882e-01
-1.26961267e+00 3.16746861e-01 3.29336561e-02 -5.18991530e-01
-3.69177818e-01 -4.84722733e-01 -1.11066234e+00 3.52284640e-01
-4.84048098e-01 4.75171387e-01 5.06030858e-01 1.31696653e+00
4.35276747e-01 9.01565313e-01 9.72558081e-01 -6.80231273e-01
-3.32865983e-01 -9.61967111e-01 -6.37869060e-01 2.26506293e-01
2.63603926e-01 -5.61263621e-01 -4.11103040e-01 -1.22202411e-01]
|
[11.46137809753418, 6.601786136627197]
|
df17dc00-da3a-4aae-b9a0-2d87ab647007
|
embrace-evaluation-and-modifications-for
|
2305.08433
| null |
https://arxiv.org/abs/2305.08433v1
|
https://arxiv.org/pdf/2305.08433v1.pdf
|
EMBRACE: Evaluation and Modifications for Boosting RACE
|
When training and evaluating machine reading comprehension models, it is very important to work with high-quality datasets that are also representative of real-world reading comprehension tasks. This requirement includes, for instance, having questions that are based on texts of different genres and require generating inferences or reflecting on the reading material. In this article we turn our attention to RACE, a dataset of English texts and corresponding multiple-choice questions (MCQs). Each MCQ consists of a question and four alternatives (of which one is the correct answer). RACE was constructed by Chinese teachers of English for human reading comprehension and is widely used as training material for machine reading comprehension models. By construction, RACE should satisfy the aforementioned quality requirements and the purpose of this article is to check whether they are indeed satisfied. We provide a detailed analysis of the test set of RACE for high-school students (1045 texts and 3498 corresponding MCQs) including (1) an evaluation of the difficulty of each MCQ and (2) annotations for the relevant pieces of the texts (called "bases") that are used to justify the plausibility of each alternative. A considerable number of MCQs appear not to fulfill basic requirements for this type of reading comprehension tasks, so we additionally identify the high-quality subset of the evaluated RACE corpus. We also demonstrate that the distribution of the positions of the bases for the alternatives is biased towards certain parts of texts, which is not necessarily desirable when evaluating MCQ answering and generation models.
|
['Johan Boye', 'Dmytro Kalpakchi', 'Mariia Zyrianova']
|
2023-05-15
| null | null | null | null |
['reading-comprehension', 'machine-reading-comprehension']
|
['natural-language-processing', 'natural-language-processing']
|
[ 4.57213074e-01 3.38529378e-01 1.27125248e-01 -4.83799726e-01
-1.01539016e+00 -7.61159778e-01 5.56092501e-01 8.14690590e-01
-5.07427454e-01 7.10995376e-01 4.03210223e-01 -1.04368532e+00
-4.36888546e-01 -9.93676066e-01 -8.07695389e-01 -1.53171331e-01
4.98788923e-01 5.57478786e-01 2.50037432e-01 -5.70586324e-01
6.31276190e-01 -1.64345607e-01 -1.94437206e+00 3.67487907e-01
1.65142918e+00 7.86176503e-01 3.55589509e-01 9.02008772e-01
-3.22577864e-01 8.13848019e-01 -1.01588607e+00 -8.86478245e-01
-3.51295978e-01 -8.65060806e-01 -1.45702302e+00 -3.05132687e-01
6.81787312e-01 -1.59238979e-01 1.84519038e-01 8.88829708e-01
3.01523745e-01 3.61398846e-01 9.34813619e-01 -9.50572491e-01
-6.19682550e-01 9.25904870e-01 -1.44930361e-02 4.65371788e-01
9.43271041e-01 -7.37513900e-02 1.30888844e+00 -6.29351616e-01
3.80195141e-01 9.59419012e-01 2.49772333e-02 5.96019983e-01
-7.59988964e-01 -3.07818383e-01 4.57502231e-02 5.03901184e-01
-9.89304841e-01 -4.36317623e-01 5.16726553e-01 -3.89226109e-01
6.96786761e-01 5.67677975e-01 4.90076810e-01 1.27015138e+00
1.49432674e-01 6.59513474e-01 1.26293731e+00 -9.94068146e-01
3.68810952e-01 8.19176212e-02 7.48026848e-01 2.95491070e-01
2.58442938e-01 -1.58287555e-01 -5.28746128e-01 9.39573944e-02
1.57196030e-01 -8.67609262e-01 -5.60982406e-01 3.01666539e-02
-1.21856117e+00 8.71173620e-01 7.22464025e-02 3.28113288e-01
3.62858512e-02 -3.94968867e-01 1.85913041e-01 5.26248038e-01
5.54126427e-02 6.89261913e-01 -6.00508928e-01 -3.77928078e-01
-5.42765975e-01 4.20446187e-01 9.72536623e-01 1.03493047e+00
4.31411803e-01 -4.42640185e-01 -2.74522096e-01 9.28169608e-01
1.34824499e-01 6.02288544e-01 5.55395007e-01 -8.03173780e-01
8.61775219e-01 6.83921754e-01 1.80163816e-01 -8.00395310e-01
-3.88273746e-01 -4.38503385e-01 -5.65547228e-01 -2.33132452e-01
1.11550128e+00 2.92314589e-02 -3.67401481e-01 2.00493884e+00
2.82477289e-01 -2.09569663e-01 4.00800705e-01 6.09200895e-01
1.19600725e+00 6.22140586e-01 1.49477765e-01 -2.41040602e-01
1.72791100e+00 -6.86836302e-01 -6.28072441e-01 -5.54441400e-02
9.04560089e-01 -7.97503293e-01 1.71259058e+00 3.43250424e-01
-1.29713166e+00 -6.98009491e-01 -1.09787321e+00 -3.49794805e-01
-1.84979916e-01 2.54699010e-02 1.02993682e-01 8.20889950e-01
-5.59424102e-01 2.17356101e-01 -1.99974313e-01 -2.52805687e-02
-5.87634221e-02 -1.53753579e-01 -1.07864581e-01 -2.11484641e-01
-1.54090405e+00 1.06683958e+00 3.21185261e-01 -1.29556492e-01
-6.85913146e-01 -5.34176111e-01 -1.00467837e+00 3.32115293e-01
4.15608913e-01 -5.16268432e-01 1.58043981e+00 -1.06417775e+00
-1.47181153e+00 1.14391100e+00 -2.77227342e-01 1.14140688e-02
5.69129944e-01 -2.43956581e-01 -4.78593498e-01 3.12808573e-01
1.82482749e-01 2.85398483e-01 7.15906560e-01 -7.68425107e-01
-7.00549603e-01 -2.13665068e-01 4.50117201e-01 2.80027360e-01
-3.08905821e-02 1.31918699e-01 3.88590597e-05 -5.42591453e-01
3.05252578e-02 -5.75196207e-01 2.64493227e-01 -6.06057525e-01
-4.57408398e-01 -8.33055377e-01 -1.56097770e-01 -7.26115584e-01
1.37719059e+00 -1.97125065e+00 -9.84206945e-02 1.59626931e-01
1.01301596e-01 9.52847488e-03 -4.00780141e-01 4.27477807e-01
-1.48503125e-01 4.04113412e-01 -6.99387342e-02 4.25605429e-03
3.69053185e-01 -2.38171946e-02 -3.56558144e-01 2.42609471e-01
1.60572976e-01 8.29699337e-01 -9.05151904e-01 -5.71745276e-01
-6.02402166e-02 -2.49214754e-01 -3.22839290e-01 4.14241642e-01
-5.47931731e-01 3.87143880e-01 -4.51731831e-01 2.06769034e-01
4.54913378e-01 -1.01508386e-01 -2.13855624e-01 2.67206728e-01
5.95688410e-02 1.08437669e+00 -9.95164454e-01 1.28632116e+00
-4.45567578e-01 5.18256187e-01 -3.81947398e-01 -7.95832515e-01
8.77709091e-01 1.67259172e-01 -3.10956478e-01 -1.02492976e+00
3.05871785e-01 3.05247724e-01 4.58984822e-01 -8.46801937e-01
8.07485819e-01 -5.72324842e-02 -3.18542212e-01 8.30509007e-01
-8.95069167e-02 -3.43041837e-01 8.53447318e-01 3.18816721e-01
8.88737142e-01 -1.39059573e-01 2.23510787e-01 -5.56862056e-01
8.81528735e-01 -2.38490589e-02 3.03056777e-01 9.23531890e-01
-3.18039395e-02 6.01941109e-01 8.52326035e-01 1.72106996e-01
-9.60870504e-01 -1.10559762e+00 -2.80305892e-01 1.29251158e+00
1.03367589e-01 -5.51379740e-01 -9.55611229e-01 -6.71508551e-01
-6.51016414e-01 1.39939260e+00 -6.31075025e-01 -6.24467134e-02
-5.93820691e-01 -2.05238551e-01 5.05757749e-01 4.00339782e-01
3.89703989e-01 -9.93902981e-01 -1.09456861e+00 -3.41773443e-02
-8.33662748e-01 -8.94523144e-01 -1.45921364e-01 3.63887027e-02
-4.78811383e-01 -1.65140676e+00 -3.90193254e-01 -8.72345090e-01
5.76197147e-01 1.48890063e-01 1.67893684e+00 6.22894228e-01
3.53066593e-01 3.63640934e-01 -9.72067177e-01 -6.30316675e-01
-9.89910662e-01 3.58676523e-01 -4.67917621e-01 -4.64003563e-01
4.78250533e-01 -1.17076620e-01 -2.44602591e-01 3.52937907e-01
-1.00947917e+00 2.92133600e-01 2.60234952e-01 7.79690385e-01
1.01878837e-01 5.51255681e-02 7.97613382e-01 -8.46314967e-01
9.47558641e-01 -4.36009109e-01 -5.35961449e-01 7.14857161e-01
-1.64313719e-01 -5.22714779e-02 7.49525547e-01 -3.14680487e-01
-1.19153583e+00 -6.97644889e-01 -5.76416135e-01 6.07738912e-01
-5.63268185e-01 8.58573735e-01 -3.27579200e-01 6.18438542e-01
9.16235626e-01 1.43112481e-01 -1.88821658e-01 -3.79214913e-01
1.73076466e-01 7.00564146e-01 6.17653906e-01 -1.15023530e+00
6.45054400e-01 -4.04260099e-01 -1.66547745e-01 -1.00807130e+00
-1.29297721e+00 -2.52956897e-01 -5.27581692e-01 -2.47527897e-01
6.74195111e-01 -5.46004176e-01 -7.59558082e-01 3.44925195e-01
-9.39366043e-01 -5.21789432e-01 -1.67114705e-01 4.91662115e-01
-6.63551867e-01 3.35736543e-01 -6.11270308e-01 -7.22094178e-01
1.80279464e-02 -1.14679360e+00 6.49883807e-01 4.03949201e-01
-7.08458185e-01 -9.36912179e-01 -1.70317084e-01 8.17230225e-01
2.12851897e-01 -1.03937976e-01 1.92954683e+00 -9.77964461e-01
-7.88362473e-02 1.20084956e-01 1.78510174e-01 2.42303714e-01
-2.83352315e-01 2.71199226e-01 -8.36994052e-01 -5.04870191e-02
2.47929826e-01 -7.73407757e-01 5.13783216e-01 1.96580887e-01
1.08929920e+00 -1.11735485e-01 2.92919070e-01 -1.84238732e-01
1.14050460e+00 -9.40161049e-02 7.12337077e-01 3.25956434e-01
1.84977159e-01 1.19407403e+00 6.68034256e-01 1.21676438e-01
7.33359337e-01 3.61264646e-01 3.02944660e-01 3.36888969e-01
2.23314762e-01 -5.03016591e-01 3.69434059e-01 1.15873837e+00
2.72741526e-01 -5.12754500e-01 -1.07768381e+00 7.23177612e-01
-1.31843579e+00 -8.91412556e-01 -8.57805848e-01 2.35316610e+00
1.12553573e+00 2.78730005e-01 -3.60799953e-02 5.24276793e-01
5.63569129e-01 1.24891074e-02 -2.00393975e-01 -6.72237158e-01
-2.34653920e-01 6.44357443e-01 -3.94144058e-01 5.80916166e-01
-6.58459723e-01 5.33740819e-01 6.04256821e+00 7.78632164e-01
-5.57408214e-01 -2.31822759e-01 7.14690447e-01 5.26934862e-01
-7.01671004e-01 1.90776754e-02 -6.19810462e-01 2.09411725e-01
1.10661423e+00 -2.32495919e-01 1.14400640e-01 4.35350180e-01
6.40508682e-02 -7.57376313e-01 -1.33881855e+00 4.35838580e-01
1.93083599e-01 -8.12558830e-01 7.45728333e-03 -3.70356202e-01
6.88055217e-01 -4.67831373e-01 -1.25432223e-01 5.11938334e-01
1.55247867e-01 -1.31981933e+00 1.02093613e+00 2.41079032e-01
5.52012324e-01 -7.26684153e-01 8.48175645e-01 8.82347107e-01
-5.62760592e-01 -1.05284490e-02 -4.64207441e-01 -4.27300245e-01
-1.01638906e-01 3.70650232e-01 -2.97426403e-01 5.95861137e-01
5.16803801e-01 7.80254975e-02 -1.09438324e+00 8.24651301e-01
-8.36236477e-01 1.04663002e+00 -1.42957479e-01 -6.06998265e-01
-7.39560975e-03 -1.13307029e-01 2.79575109e-01 8.18460822e-01
3.04906875e-01 2.86125094e-01 -2.35095009e-01 8.52852345e-01
-1.57576635e-01 6.01712942e-01 -1.05667874e-01 7.93472901e-02
6.11255467e-01 8.29128027e-01 -5.12554049e-01 -4.74830300e-01
-5.81801176e-01 4.01005536e-01 5.21973848e-01 2.20160782e-01
-6.67521894e-01 -4.51204568e-01 1.56294748e-01 2.08302327e-02
-5.42458966e-02 -3.40388492e-02 -3.57962102e-01 -1.18358254e+00
1.56089321e-01 -1.38199306e+00 5.21609068e-01 -9.14053500e-01
-1.43748403e+00 4.23787475e-01 1.98508263e-01 -8.83604765e-01
-2.70474970e-01 -6.13149226e-01 -6.52229786e-01 1.13312578e+00
-1.45889640e+00 -5.47803938e-01 -5.12827337e-01 5.53410172e-01
6.52427793e-01 2.95898914e-01 8.23236942e-01 -9.72841680e-02
-5.35736799e-01 7.50524402e-01 -2.30932191e-01 1.50976360e-01
6.70758247e-01 -1.43671906e+00 2.20747665e-01 9.41578805e-01
1.54830173e-01 7.77805150e-01 9.28981364e-01 -4.46923554e-01
-9.50675964e-01 -5.91464877e-01 1.51613724e+00 -5.23628891e-01
5.09257913e-01 -1.54453844e-01 -1.26227844e+00 3.78261119e-01
4.53167468e-01 -9.04993892e-01 9.27873492e-01 3.14671606e-01
-2.60582447e-01 3.21347415e-01 -7.05866933e-01 7.30257571e-01
7.34022617e-01 -3.66688520e-01 -1.51905751e+00 2.45382413e-01
4.64195281e-01 -6.05110466e-01 -6.84852600e-01 3.50589305e-01
1.34393811e-01 -1.23396015e+00 3.48128438e-01 -8.02673221e-01
1.05564439e+00 -1.40722156e-01 -6.05862811e-02 -1.38772750e+00
1.14389084e-01 -1.66090816e-01 2.75695741e-01 1.22766495e+00
6.36593103e-01 -2.85272926e-01 5.22180378e-01 8.42883229e-01
-2.75298595e-01 -6.79864109e-01 -9.02483761e-01 -4.33305353e-01
7.77363658e-01 -6.21364057e-01 6.82502389e-01 8.86610150e-01
3.93417448e-01 7.36558318e-01 2.63387144e-01 -1.27725631e-01
2.10366949e-01 2.88755506e-01 8.20041895e-01 -1.11258233e+00
-2.10784093e-01 -7.38334477e-01 2.70334721e-01 -1.47598290e+00
4.00171727e-01 -6.87478006e-01 1.99098244e-01 -1.42394328e+00
1.50243342e-01 -4.32893455e-01 2.12728247e-01 2.67129503e-02
-6.97688639e-01 -2.40973964e-01 5.57657797e-03 -1.04238741e-01
-5.17283320e-01 5.91690302e-01 1.39687836e+00 -1.16425296e-02
1.44520193e-01 3.01318407e-01 -8.45945418e-01 6.71305537e-01
9.26119685e-01 -6.22687116e-02 -6.51119471e-01 -4.50215071e-01
7.43782699e-01 3.57057899e-01 1.70923918e-01 -5.96618474e-01
1.55852795e-01 -3.91661018e-01 1.96792096e-01 -5.73332489e-01
-2.48044387e-01 -4.57056493e-01 -3.43896300e-01 3.30792248e-01
-8.77357662e-01 3.56907099e-01 -1.32548258e-01 9.44328606e-02
-3.29844266e-01 -1.05451345e+00 6.20591402e-01 -6.65613413e-02
-4.30818766e-01 -4.37089115e-01 -6.78265035e-01 7.68566906e-01
8.08533907e-01 -1.69387758e-01 -6.52980447e-01 -6.57871068e-01
-3.62464607e-01 4.03340638e-01 4.22427922e-01 5.11732817e-01
4.59721804e-01 -8.43269348e-01 -1.15872025e+00 6.71124160e-02
4.48667198e-01 2.01957166e-01 2.68139362e-01 5.97149968e-01
-6.46371305e-01 4.02410239e-01 -1.07430622e-01 -3.03096741e-01
-1.14519596e+00 4.17642504e-01 1.59479946e-01 -3.23573232e-01
-3.07910800e-01 6.88988268e-01 1.02924590e-03 -3.25559884e-01
2.64003992e-01 -5.67457616e-01 -7.21073508e-01 6.23022169e-02
7.60627508e-01 4.35627073e-01 2.60991842e-01 -6.69854701e-01
1.02291539e-01 3.43248129e-01 1.49151400e-01 -1.10087097e-01
7.43014038e-01 -3.31476390e-01 -1.63948566e-01 6.58049345e-01
7.62321293e-01 4.19198513e-01 -6.26450777e-01 -1.17812209e-01
2.14424685e-01 -2.89307803e-01 -5.75903654e-01 -1.06745863e+00
-3.39436740e-01 9.85275328e-01 -1.30643472e-01 3.29585046e-01
1.08044720e+00 8.95088390e-02 6.28913879e-01 3.34395796e-01
2.51006424e-01 -1.22155726e+00 1.01459973e-01 9.14836287e-01
8.16862464e-01 -1.16617191e+00 -2.31351629e-01 -6.07792675e-01
-4.39766049e-01 1.17263865e+00 9.23619449e-01 3.98555100e-01
1.05811886e-01 -2.92357117e-01 2.45325059e-01 -5.98873496e-02
-9.07189548e-01 -1.29590094e-01 3.14982831e-01 4.11872089e-01
7.47743726e-01 1.21046454e-01 -9.42753077e-01 7.84935057e-01
-1.00280762e+00 -6.43643618e-01 8.91854107e-01 7.70370305e-01
-5.70794523e-01 -1.03086472e+00 -5.51563680e-01 6.32563114e-01
-4.98831004e-01 -1.96536645e-01 -4.11048740e-01 6.72759652e-01
-1.61427423e-01 1.66854095e+00 -3.46717946e-02 -1.49792850e-01
4.11283463e-01 1.29424006e-01 6.01164579e-01 -6.42885268e-01
-8.01331758e-01 -4.68832493e-01 4.20435965e-01 2.81536225e-02
-3.13767016e-01 -5.23002923e-01 -1.09217262e+00 -5.41537762e-01
-6.49247766e-01 6.65864229e-01 3.69633764e-01 1.35642540e+00
-3.02979112e-01 3.71856272e-01 3.35368365e-01 1.51547298e-01
-9.02139962e-01 -1.24054849e+00 -3.96073788e-01 8.11464727e-01
9.38178301e-02 -3.69448781e-01 -4.13219690e-01 1.29230013e-02]
|
[11.461197853088379, 8.184136390686035]
|
6cb1a8df-3eb7-412c-be95-6389c098d6aa
|
clenshaw-graph-neural-networks
|
2210.16508
| null |
https://arxiv.org/abs/2210.16508v2
|
https://arxiv.org/pdf/2210.16508v2.pdf
|
Clenshaw Graph Neural Networks
|
Graph Convolutional Networks (GCNs), which use a message-passing paradigm with stacked convolution layers, are foundational methods for learning graph representations. Recent GCN models use various residual connection techniques to alleviate the model degradation problem such as over-smoothing and gradient vanishing. Existing residual connection techniques, however, fail to make extensive use of underlying graph structure as in the graph spectral domain, which is critical for obtaining satisfactory results on heterophilic graphs. In this paper, we introduce ClenshawGCN, a GNN model that employs the Clenshaw Summation Algorithm to enhance the expressiveness of the GCN model. ClenshawGCN equips the standard GCN model with two straightforward residual modules: the adaptive initial residual connection and the negative second-order residual connection. We show that by adding these two residual modules, ClenshawGCN implicitly simulates a polynomial filter under the Chebyshev basis, giving it at least as much expressive power as polynomial spectral GNNs. In addition, we conduct comprehensive experiments to demonstrate the superiority of our model over spatial and spectral GNN models.
|
['Zhewei Wei', 'Yuhe Guo']
|
2022-10-29
| null | null | null | null |
['node-classification-on-non-homophilic']
|
['graphs']
|
[-2.11001956e-03 3.73843014e-01 1.19265549e-01 1.73795462e-01
2.05404595e-01 -4.08196658e-01 6.71970606e-01 9.86624658e-02
-4.28030528e-02 4.09318686e-01 -1.10932611e-01 -6.67075455e-01
-1.04500294e-01 -1.18538916e+00 -7.88644493e-01 -6.90328717e-01
-5.56445777e-01 -1.13223597e-01 2.70326406e-01 -5.92078447e-01
4.75108549e-02 6.34254575e-01 -1.06848216e+00 1.52985707e-01
1.00229573e+00 7.02663362e-01 -3.37742940e-02 1.05998003e+00
-2.37158030e-01 8.06586981e-01 -4.02783066e-01 -4.03759509e-01
3.28840047e-01 -3.47359985e-01 -5.97639501e-01 -1.51086390e-01
5.65190256e-01 -1.13702251e-03 -1.03378308e+00 1.17401874e+00
2.35118404e-01 2.00062260e-01 4.72015649e-01 -1.33813417e+00
-1.01495826e+00 8.70374441e-01 -5.17967522e-01 1.73950374e-01
3.93044055e-02 7.87439931e-04 1.16950572e+00 -7.34024823e-01
2.97740251e-01 1.28934133e+00 1.29271519e+00 2.66415805e-01
-1.48091173e+00 -5.56110084e-01 3.09738129e-01 -8.81044865e-02
-1.45751894e+00 2.46742457e-01 9.58298922e-01 -1.93239242e-01
1.06139219e+00 3.34640563e-01 1.06632960e+00 7.73630798e-01
1.88327059e-01 5.48643649e-01 5.52785277e-01 -4.09495622e-01
-6.46846071e-02 -3.57014865e-01 3.79245937e-01 1.10412955e+00
5.19835889e-01 1.66536923e-02 -1.02949023e-01 -3.00856501e-01
1.36721182e+00 3.26752394e-01 -5.46675146e-01 -5.62674165e-01
-9.08280194e-01 9.77481127e-01 1.21825016e+00 2.81759441e-01
-1.14666194e-01 7.65332818e-01 2.03543171e-01 3.00757766e-01
4.74377632e-01 2.96543449e-01 1.70500055e-01 5.95544696e-01
-6.43322349e-01 4.64144908e-02 9.13986385e-01 1.08105409e+00
1.06268501e+00 3.63717347e-01 2.85306294e-02 6.69043779e-01
1.68361813e-01 3.33755761e-01 2.10510626e-01 -5.70410669e-01
1.89022571e-01 9.77339685e-01 -6.02117419e-01 -1.47572911e+00
-6.94947958e-01 -8.83362353e-01 -1.38740861e+00 -1.03528813e-01
6.79365471e-02 9.19891242e-03 -9.75616157e-01 1.67544127e+00
-6.16941750e-02 5.14878094e-01 -4.18870859e-02 6.83674574e-01
9.72828150e-01 6.40722752e-01 -2.50778794e-02 3.92347306e-01
8.35413158e-01 -1.10415065e+00 -3.55774879e-01 -8.57587717e-03
9.14074004e-01 -4.08041507e-01 8.74470294e-01 -9.29845423e-02
-9.08946574e-01 -3.89211774e-01 -1.29933321e+00 -1.66067436e-01
-6.04422152e-01 -7.04624364e-03 1.29637015e+00 6.41805708e-01
-1.70556664e+00 9.29816723e-01 -6.95899487e-01 -4.95272875e-01
2.30492175e-01 3.90477210e-01 -3.17001700e-01 3.92550640e-02
-1.37709010e+00 4.85425860e-01 3.20945650e-01 3.13850164e-01
-6.19172275e-01 -7.24279463e-01 -1.09081495e+00 3.95349503e-01
3.88788767e-02 -8.05343330e-01 8.46190631e-01 -9.65504825e-01
-1.12256670e+00 5.37217200e-01 2.86540926e-01 -8.38983059e-01
3.21498543e-01 2.45183960e-01 -5.08089483e-01 2.78739333e-01
-4.67753142e-01 4.02220309e-01 6.65073037e-01 -9.86977816e-01
2.80733630e-02 -1.33885980e-01 4.51763391e-01 1.94275185e-01
-4.35004950e-01 -5.92778683e-01 -4.38117117e-01 -6.96514845e-01
3.04791868e-01 -7.57186294e-01 -6.05459452e-01 -1.44613668e-01
-4.48172629e-01 -1.51430309e-01 7.88866937e-01 -3.00670922e-01
1.48277938e+00 -2.17046833e+00 -6.83071166e-02 6.32319689e-01
1.02529812e+00 4.48257059e-01 -4.59570020e-01 7.89183080e-01
-3.52040410e-01 9.26883891e-02 -2.72617638e-01 8.44102632e-03
-1.12530313e-01 3.24016035e-01 -2.27500007e-01 5.64750969e-01
2.99384356e-01 1.11218762e+00 -1.01783025e+00 -6.74577430e-02
1.89661831e-01 7.67352462e-01 -7.48526037e-01 -1.44453362e-01
-2.86694527e-01 -1.04004376e-01 -3.05401832e-01 1.21617086e-01
8.92438531e-01 -7.13543773e-01 2.74183840e-01 -2.27962911e-01
7.56213441e-02 1.99877188e-01 -1.07282305e+00 1.45955276e+00
-2.51587063e-01 6.11272633e-01 2.18809202e-01 -1.16579187e+00
8.13964486e-01 -2.28917859e-02 2.91520745e-01 -4.67612267e-01
-1.79097027e-01 1.60831779e-01 4.58246889e-03 1.43786445e-01
6.79313064e-01 1.75556868e-01 2.80623376e-01 2.19599038e-01
1.89584970e-01 1.25301674e-01 2.69700766e-01 9.78115678e-01
1.29706860e+00 -1.80749640e-01 3.78539041e-02 -6.62877977e-01
4.31665123e-01 -2.83405125e-01 1.33235186e-01 8.34547639e-01
1.47618979e-01 5.14876962e-01 8.98676276e-01 -4.72771972e-01
-9.83460486e-01 -9.75861549e-01 1.84888393e-01 9.42777216e-01
1.92536056e-01 -8.74415457e-01 -7.68644333e-01 -5.22422194e-01
9.84252617e-02 3.27966303e-01 -6.48718953e-01 -4.14093077e-01
-4.53709483e-01 -8.77330065e-01 7.86535919e-01 5.70758402e-01
6.96360052e-01 -6.15873873e-01 5.76590449e-02 1.89079389e-01
2.50308782e-01 -6.69406235e-01 -6.16956055e-01 2.05904290e-01
-8.83156776e-01 -1.17105210e+00 -7.77082205e-01 -9.17466164e-01
8.84899974e-01 6.87537670e-01 1.24871826e+00 7.22195923e-01
-2.55190760e-01 4.14923906e-01 -9.31350887e-02 1.10457301e-01
-2.91752577e-01 1.46562457e-01 -3.05698872e-01 -9.37966928e-02
1.66553289e-01 -9.61043060e-01 -7.16499031e-01 -4.74081263e-02
-1.08354747e+00 3.11252773e-01 5.00523508e-01 8.23027432e-01
2.81253755e-01 8.22956115e-02 2.43056446e-01 -1.22831845e+00
7.39935577e-01 -3.40455711e-01 -7.10860014e-01 1.58123598e-01
-6.22554660e-01 2.77105063e-01 8.87985766e-01 -3.43745232e-01
-3.16734165e-01 -1.99756160e-01 -8.67119506e-02 -4.93676841e-01
5.27984440e-01 8.01658392e-01 1.45709559e-01 -8.93756807e-01
7.48185039e-01 1.28448635e-01 9.79855657e-02 -2.54657596e-01
6.52871847e-01 -6.20306656e-03 5.13304710e-01 -4.31365550e-01
8.07341158e-01 4.58315611e-01 3.41197610e-01 -1.09041476e+00
-4.97543603e-01 -4.37404126e-01 -3.12779188e-01 -9.73089039e-02
4.00937766e-01 -6.99585795e-01 -9.04533863e-01 4.67520863e-01
-1.04969287e+00 -5.32155752e-01 -1.73115745e-01 3.02197069e-01
-2.78905272e-01 7.12919235e-01 -1.03071904e+00 -7.03103483e-01
-3.53979945e-01 -6.32227123e-01 7.43004918e-01 7.94673860e-02
2.85643578e-01 -1.41869938e+00 -3.10697425e-02 -5.28697908e-01
7.36994267e-01 2.32223764e-01 1.23592234e+00 -5.61996937e-01
-5.94630837e-01 -1.60290778e-01 -6.11206055e-01 1.90394402e-01
-3.64570647e-01 -3.68444202e-03 -7.25076795e-01 -4.95144725e-01
-2.77017832e-01 2.03905359e-01 1.34902465e+00 3.20758581e-01
1.22107840e+00 -3.10564667e-01 -2.84297019e-01 1.10807228e+00
1.68568814e+00 -3.12274605e-01 8.28714788e-01 -1.30813822e-01
1.12155175e+00 -3.80122941e-03 -7.25942731e-01 3.58628407e-02
4.09443289e-01 4.23714779e-02 6.74072802e-01 -6.13181353e-01
-2.99075246e-01 -4.44194108e-01 2.00792149e-01 1.16451848e+00
-2.63900161e-01 -2.80213326e-01 -9.18836057e-01 2.76682049e-01
-2.00478101e+00 -8.59903991e-01 -5.94890177e-01 2.05990362e+00
3.15012544e-01 9.08744857e-02 -6.01350330e-02 6.53751567e-03
9.82576132e-01 4.49559480e-01 -4.12244886e-01 -4.21080172e-01
-4.31897193e-01 3.28840435e-01 8.58935595e-01 5.52711308e-01
-1.07033277e+00 8.81582260e-01 6.50392294e+00 7.71359563e-01
-9.64609742e-01 -2.95700908e-01 2.64809877e-01 3.93704265e-01
-7.15347469e-01 1.96812615e-01 -4.64471728e-01 1.49473265e-01
9.20800090e-01 -1.61124185e-01 8.12054932e-01 6.74937189e-01
-3.74647379e-01 3.94585103e-01 -9.30961490e-01 8.31082642e-01
3.22723165e-02 -1.57331336e+00 3.78955603e-01 1.20998643e-01
7.67248213e-01 9.44131389e-02 5.62310293e-02 3.96538258e-01
7.41717100e-01 -1.25176752e+00 2.26536900e-01 4.47435260e-01
7.22466946e-01 -8.10059726e-01 4.52834934e-01 1.71019398e-02
-1.57824206e+00 1.33122683e-01 -6.77258611e-01 -2.18964830e-01
-2.99186766e-01 6.49514914e-01 -5.56719601e-01 8.69689107e-01
3.01213503e-01 8.12756836e-01 -6.53952360e-01 1.11753142e+00
-1.18410155e-01 5.43419361e-01 -3.82833987e-01 -1.41842186e-01
5.89853525e-01 -5.61604679e-01 4.46082383e-01 1.26606631e+00
3.11101019e-01 7.92576149e-02 2.87190616e-01 9.97117758e-01
-3.41293156e-01 2.53034644e-02 -7.78885663e-01 -4.63323504e-01
7.52197430e-02 1.23506713e+00 -7.85529792e-01 -1.57415703e-01
-7.06291854e-01 9.04252350e-01 6.27984762e-01 9.12957489e-01
-5.93908310e-01 -7.28799701e-01 4.63485688e-01 2.20369190e-01
3.78046244e-01 -4.22030360e-01 -9.67637822e-02 -1.18392801e+00
-3.70658010e-01 -5.77308953e-01 5.00072479e-01 -7.38048434e-01
-1.39382565e+00 5.95633686e-01 -4.83182222e-01 -7.65216529e-01
1.82701558e-01 -8.85415196e-01 -7.49493122e-01 1.03158236e+00
-1.42633891e+00 -1.18835735e+00 -3.18474203e-01 8.01563323e-01
-3.95144522e-01 1.69582233e-01 8.18428159e-01 2.39751369e-01
-4.53572005e-01 4.91632521e-01 3.44255567e-02 3.78725916e-01
2.35559404e-01 -1.38819516e+00 6.76357508e-01 8.84951293e-01
5.16841672e-02 1.14432704e+00 4.09678608e-01 -7.06271648e-01
-1.61800969e+00 -1.30151701e+00 5.77275038e-01 1.11368984e-01
1.01293147e+00 -5.42213202e-01 -1.14876890e+00 9.54362154e-01
1.66278005e-01 2.66297787e-01 3.83527458e-01 3.53350312e-01
-6.35065913e-01 -6.41857311e-02 -7.31418669e-01 9.58159924e-01
1.35154176e+00 -6.85709238e-01 -1.07522540e-01 4.55428421e-01
9.51043010e-01 -3.81665617e-01 -6.51179910e-01 3.94109100e-01
2.25631088e-01 -9.63083267e-01 9.13635850e-01 -6.99204087e-01
1.70456082e-01 -2.76277006e-01 1.35939457e-02 -1.40703762e+00
-6.63990140e-01 -8.81663740e-01 -2.74531394e-01 6.10174060e-01
3.42150360e-01 -1.01972747e+00 7.70934045e-01 2.13532433e-01
-3.79536688e-01 -7.01950908e-01 -4.15380865e-01 -7.12305486e-01
1.38383567e-01 -3.32057059e-01 7.17009246e-01 9.23786819e-01
1.15015939e-01 4.01045382e-01 -1.99012831e-01 2.70223320e-01
5.53830683e-01 8.29442218e-02 6.52048528e-01 -1.36051428e+00
-2.39849523e-01 -8.22500467e-01 -8.63242447e-01 -1.19381368e+00
3.58050540e-02 -1.38484693e+00 -4.11982685e-01 -1.76777542e+00
2.12845027e-01 -2.49057889e-01 -4.28850055e-01 2.45456606e-01
-7.20563829e-02 2.57304311e-01 -2.01691985e-02 6.12433441e-03
-5.22582591e-01 3.71800452e-01 1.29012084e+00 -3.73925865e-02
-4.91789207e-02 -3.61708142e-02 -7.80513167e-01 6.92976117e-01
6.93966091e-01 1.61277518e-01 -6.41581774e-01 -4.25377339e-01
6.92244887e-01 -1.38775930e-01 7.06870854e-01 -1.01885450e+00
4.54751283e-01 3.28168362e-01 3.53394657e-01 -4.09150064e-01
7.44199604e-02 -5.27713060e-01 3.14442456e-01 5.06977975e-01
-2.35816255e-01 1.91329047e-01 2.52983093e-01 8.92511129e-01
-3.34664322e-02 1.45339072e-01 7.40225136e-01 -1.27055675e-01
-5.77563763e-01 7.07331598e-01 -1.58060059e-01 -1.30102128e-01
4.79141980e-01 -1.73805550e-01 -6.71337903e-01 -6.08518720e-01
-7.04841733e-01 3.50767016e-01 5.24275541e-01 -1.19181395e-01
6.36310577e-01 -1.50590146e+00 -4.70768034e-01 6.04684889e-01
-1.23106618e-03 -1.25729248e-01 3.72197419e-01 9.83442903e-01
-9.33434129e-01 3.48977506e-01 8.06551427e-02 -4.29094195e-01
-7.21361816e-01 7.42645264e-01 5.93554556e-01 -3.95238131e-01
-1.02580094e+00 9.10837531e-01 6.49950922e-01 -5.49948871e-01
2.86462277e-01 -4.36441272e-01 7.39261834e-03 -4.15738344e-01
2.88866282e-01 2.91583896e-01 8.70003402e-02 -2.89131194e-01
-1.23009041e-01 2.78823495e-01 2.27577854e-02 2.91892350e-01
1.10792649e+00 1.11985840e-01 -5.02713323e-01 1.61157250e-01
1.18493354e+00 2.42237113e-02 -1.15874863e+00 -2.53577083e-01
-2.17596188e-01 1.05085305e-03 -1.12360045e-02 -2.75416404e-01
-1.14476824e+00 1.12706792e+00 5.54972291e-02 9.04444218e-01
1.02590179e+00 -2.26453960e-01 6.69018507e-01 3.70409012e-01
1.51230037e-01 -7.35000014e-01 -1.03204183e-01 9.46465135e-01
8.33209634e-01 -6.41653776e-01 -1.08971469e-01 -6.30591750e-01
-5.78226112e-02 1.37571299e+00 5.03171980e-01 -6.38475776e-01
8.65588963e-01 4.88689318e-02 -3.39726239e-01 -4.21342164e-01
-6.16323531e-01 -3.79607856e-01 3.02225143e-01 5.33157766e-01
4.69419122e-01 2.41290167e-01 -3.68456423e-01 3.40746582e-01
-2.43566424e-01 -4.63703305e-01 5.32467365e-01 4.42507356e-01
-4.51524973e-01 -6.83588624e-01 3.23006026e-02 5.55805802e-01
-1.46391764e-01 -5.83547711e-01 -5.12695789e-01 9.14107502e-01
-3.68799627e-01 6.32743061e-01 1.33823171e-01 -5.09607136e-01
1.31699651e-01 -2.04035208e-01 4.95701224e-01 -5.37405610e-01
-6.37258291e-01 7.08306283e-02 -3.47565301e-02 -6.11815155e-01
-1.46286488e-01 1.38428286e-01 -1.40509462e+00 -9.11773980e-01
-2.81335860e-01 2.43904948e-01 3.99813026e-01 5.22926390e-01
4.62473392e-01 7.68470585e-01 3.66924018e-01 -7.34509289e-01
-4.33686614e-01 -7.92213380e-01 -9.85194921e-01 3.31178755e-01
4.02633935e-01 -2.69322783e-01 -5.54138303e-01 -5.38983285e-01]
|
[6.894405364990234, 6.1327290534973145]
|
3cb908cb-322c-42b2-8bd0-e4f364bf229f
|
classifier-calibration-how-to-assess-and
|
2112.10327
| null |
https://arxiv.org/abs/2112.10327v2
|
https://arxiv.org/pdf/2112.10327v2.pdf
|
Classifier Calibration: A survey on how to assess and improve predicted class probabilities
|
This paper provides both an introduction to and a detailed overview of the principles and practice of classifier calibration. A well-calibrated classifier correctly quantifies the level of uncertainty or confidence associated with its instance-wise predictions. This is essential for critical applications, optimal decision making, cost-sensitive classification, and for some types of context change. Calibration research has a rich history which predates the birth of machine learning as an academic field by decades. However, a recent increase in the interest on calibration has led to new methods and the extension from binary to the multiclass setting. The space of options and issues to consider is large, and navigating it requires the right set of concepts and tools. We provide both introductory material and up-to-date technical details of the main concepts and methods, including proper scoring rules and other evaluation metrics, visualisation approaches, a comprehensive account of post-hoc calibration methods for binary and multiclass classification, and several advanced topics.
|
['Peter Flach', 'Meelis Kull', 'Raul Santos-Rodriguez', 'Miquel Perello-Nieto', 'Hao Song', 'Telmo Silva Filho']
|
2021-12-20
| null | null | null | null |
['classifier-calibration', 'classifier-calibration']
|
['computer-vision', 'miscellaneous']
|
[ 4.73556042e-01 -1.72047362e-01 -6.19619071e-01 -1.02911127e+00
-8.81563127e-01 -7.14430332e-01 4.90627199e-01 6.64443314e-01
-1.81618467e-01 9.26458776e-01 -2.18130037e-01 -6.33382618e-01
-6.00739777e-01 -5.23212492e-01 -8.58886242e-02 -7.39563763e-01
-1.83788851e-01 6.64960921e-01 6.70031309e-02 -4.36501503e-02
4.45730656e-01 6.94826424e-01 -1.76137865e+00 2.62601972e-01
6.71341598e-01 1.35608542e+00 -4.39267546e-01 6.41225100e-01
-6.02854788e-02 2.32641473e-01 -7.99733639e-01 -8.59538853e-01
-1.04455709e-01 -3.63729686e-01 -6.07112706e-01 -2.81255364e-01
5.05316198e-01 3.63521576e-01 2.82328427e-01 8.47357273e-01
4.70965773e-01 -1.37512952e-01 9.91843998e-01 -1.55325890e+00
-3.87539744e-01 4.88629550e-01 -3.46160442e-01 2.99998730e-01
1.46750256e-01 -1.06107689e-01 8.57302189e-01 -6.22034013e-01
9.75053534e-02 1.04005110e+00 1.02464783e+00 3.40092152e-01
-1.25801420e+00 -7.64767349e-01 1.99526682e-01 9.06745493e-02
-1.25788450e+00 -7.13132918e-02 2.81040639e-01 -7.34153271e-01
6.65237427e-01 6.88887179e-01 7.10491180e-01 8.96309197e-01
2.68066019e-01 3.82649690e-01 1.31930673e+00 -8.40702295e-01
3.52935940e-01 8.19815874e-01 4.58698124e-01 2.43324518e-01
5.51354587e-01 3.13066095e-01 -3.27339381e-01 -2.58567542e-01
2.38069847e-01 -1.87583134e-01 -2.59736869e-02 -7.34465420e-01
-9.70773458e-01 9.85884607e-01 3.53722245e-01 3.29314262e-01
2.02760175e-01 -3.69610101e-01 3.98880094e-01 2.77778953e-01
3.23267162e-01 6.57137930e-01 -5.41147351e-01 -1.01546668e-01
-1.16233289e+00 1.33814499e-01 8.72855186e-01 9.24874127e-01
4.56665128e-01 -3.02597672e-01 -1.76535007e-02 1.08586252e+00
-2.03412130e-01 3.40315849e-01 3.95861000e-01 -8.44610631e-01
1.28911301e-01 4.83519435e-01 1.30118122e-02 -7.83815384e-01
-5.39455712e-01 -7.71413803e-01 -8.23460340e-01 3.11922580e-01
3.35081905e-01 1.03086732e-01 -7.48991072e-01 1.51371288e+00
2.50481009e-01 -4.41843271e-01 -2.39887819e-01 6.36112690e-01
6.21133089e-01 1.81245189e-02 3.27908009e-01 -3.49735647e-01
1.21821725e+00 -5.31570911e-01 -5.41045606e-01 -4.08070713e-01
4.98664945e-01 -8.09978306e-01 8.11577797e-01 4.21113640e-01
-6.54952943e-01 -5.36264598e-01 -1.45999193e+00 1.30681813e-01
-8.61509085e-01 -5.60958721e-02 9.86182630e-01 1.40242612e+00
-5.30846655e-01 7.28627980e-01 -5.79732418e-01 -6.54802382e-01
4.47107762e-01 4.30953175e-01 -4.05858070e-01 -1.81138635e-01
-1.15173352e+00 1.37053239e+00 3.96943957e-01 -1.22441567e-01
-4.75591570e-02 -7.24718571e-01 -6.33580863e-01 -2.02229321e-01
1.47454843e-01 -4.53262955e-01 1.31535470e+00 -8.78362238e-01
-1.02471554e+00 1.00673437e+00 7.73085132e-02 -1.70372918e-01
7.50675201e-01 4.15091924e-02 -5.98680317e-01 -2.24418044e-01
3.73789109e-02 5.91242731e-01 4.37877208e-01 -1.00450265e+00
-1.13289356e+00 -3.63991708e-01 -1.22072756e-01 5.41306734e-02
-2.01240465e-01 2.19182789e-01 -1.04132377e-01 -6.39821947e-01
1.44293517e-01 -8.54610443e-01 -7.06070885e-02 -2.02108058e-03
-8.39210376e-02 2.59331167e-02 5.49893677e-01 -4.07381535e-01
1.53081715e+00 -2.16553783e+00 -2.39752650e-01 4.33992773e-01
-1.35197982e-01 1.16760693e-01 3.52738112e-01 3.67180973e-01
-6.49347246e-01 1.94880724e-01 -7.90860474e-01 -9.94821191e-02
-3.08522322e-05 1.13280110e-01 -3.27402443e-01 5.65516770e-01
1.95863977e-01 6.15077317e-01 -9.35251236e-01 -4.72757906e-01
4.89119798e-01 4.67978626e-01 1.80582609e-02 1.58084948e-02
2.97165632e-01 2.27543816e-01 1.60034359e-01 9.78936374e-01
6.17921889e-01 -2.90674329e-01 1.35413662e-01 -7.12190643e-02
-3.22871357e-02 1.71894178e-01 -1.29102051e+00 9.84267056e-01
-2.91062891e-01 6.06749594e-01 -1.54234201e-01 -1.11058939e+00
1.08991230e+00 5.89173399e-02 4.94992346e-01 -2.73189336e-01
1.17502868e-01 5.67990839e-01 -1.29843965e-01 7.79763907e-02
2.74379313e-01 -3.01096767e-01 -2.20493913e-01 3.39012146e-01
-1.51934428e-02 -4.73216981e-01 1.76715612e-01 -3.71747911e-02
6.05939686e-01 -9.04280171e-02 8.91045094e-01 -2.93391615e-01
5.60611665e-01 1.76377103e-01 3.46612453e-01 7.89124966e-01
-1.87691018e-01 6.32338762e-01 5.02090871e-01 -3.94618154e-01
-8.15388501e-01 -8.19863260e-01 -9.75600541e-01 9.40132618e-01
-2.02026546e-01 -2.63010621e-01 -6.65432930e-01 -5.37395656e-01
5.43015778e-01 8.29121590e-01 -9.05667305e-01 -1.51968583e-01
-2.91275650e-01 -1.24175632e+00 2.43550256e-01 8.03964138e-01
2.23785117e-01 -5.71237087e-01 -6.27953291e-01 1.40643278e-02
-3.12107448e-02 -8.42271745e-01 5.25056273e-02 7.22947299e-01
-1.15292931e+00 -1.37291229e+00 -5.03033102e-01 -5.88446677e-01
5.04462898e-01 -3.02600232e-03 1.27470660e+00 1.71844721e-01
-4.64611560e-01 4.33880925e-01 -2.39544109e-01 -6.75608456e-01
-4.05185312e-01 2.05158636e-01 -5.93180619e-02 -3.49986523e-01
7.43884861e-01 -2.38943473e-01 -3.24895710e-01 6.54651701e-01
-5.50948679e-01 -2.77909249e-01 5.88702857e-01 1.01406634e+00
6.01188958e-01 -2.94594523e-02 4.59742337e-01 -1.14167118e+00
7.98101425e-01 -3.43699366e-01 -7.57681310e-01 7.57103622e-01
-1.40419197e+00 -1.51855499e-01 2.77706385e-02 -1.81023598e-01
-6.66129649e-01 1.55068845e-01 9.93253291e-02 6.05032705e-02
-1.53503671e-01 4.66869026e-01 -8.37328434e-02 -1.34414583e-01
1.05489993e+00 -4.38196570e-01 -5.34816496e-02 -4.46656317e-01
3.19735616e-01 1.08752787e+00 6.40738785e-01 -6.10950649e-01
4.67534930e-01 1.55307606e-01 4.16905791e-01 -4.48418289e-01
-9.72281635e-01 -5.13405502e-01 -1.24820232e+00 -1.74372271e-01
3.63304108e-01 -4.56844538e-01 -5.17497659e-01 1.60476476e-01
-6.70500994e-01 -8.64701867e-02 -3.07550877e-01 3.06329936e-01
-4.66136962e-01 2.16892764e-01 5.39294677e-03 -8.70042503e-01
-1.83453724e-01 -1.30455673e+00 7.16275454e-01 2.18488395e-01
-5.68918824e-01 -1.19778121e+00 -1.13185188e-02 2.80485690e-01
3.46639842e-01 5.41196287e-01 1.00344455e+00 -6.70324981e-01
-1.02148555e-01 -9.65243161e-01 -1.90267205e-01 4.07956839e-01
1.54280007e-01 1.81915894e-01 -1.37540972e+00 -3.17822963e-01
-2.84197837e-01 -1.93105340e-01 7.21697807e-01 4.19308573e-01
1.43770516e+00 1.77476719e-01 -6.97874725e-01 5.21495759e-01
1.12022221e+00 1.67214632e-01 2.22452343e-01 4.85664815e-01
6.30185977e-02 9.09856975e-01 8.80103588e-01 2.90121157e-02
1.16645701e-01 7.33689547e-01 2.14055955e-01 -8.25792551e-02
6.62112758e-02 7.53853843e-02 -3.04744810e-01 3.77666861e-01
-4.96669412e-02 5.83477318e-02 -9.91512597e-01 1.80442795e-01
-1.65853822e+00 -8.89532626e-01 -1.60391077e-01 2.80790019e+00
6.35918438e-01 2.96665519e-01 2.62328595e-01 7.34954119e-01
9.88081932e-01 -3.42216641e-01 -5.82441509e-01 -5.71260154e-01
-2.13795707e-01 1.55713990e-01 5.87548554e-01 5.65453470e-01
-1.28739309e+00 4.73769277e-01 8.38466454e+00 7.25579023e-01
-1.04065526e+00 -1.76056549e-01 1.06965959e+00 8.69038180e-02
9.57707763e-02 3.65917869e-02 -9.25244153e-01 3.47198635e-01
1.17189300e+00 -1.79330528e-01 9.58399251e-02 1.10466421e+00
-3.75161320e-01 -5.10111988e-01 -1.34249306e+00 1.10342205e+00
3.16331573e-02 -1.21616983e+00 -3.88145804e-01 -1.10400774e-01
4.64385420e-01 -3.62665296e-01 1.34376600e-01 3.34090143e-01
2.15730146e-01 -1.06237090e+00 6.49768770e-01 2.67539501e-01
1.35299921e+00 -7.13401794e-01 9.70612526e-01 -7.94757828e-02
-9.82163370e-01 -3.85098696e-01 -2.21184537e-01 4.10917066e-02
-1.97199941e-01 9.30329025e-01 -4.57406431e-01 5.07264316e-01
9.41076577e-01 6.47229254e-01 -9.51903343e-01 1.35268438e+00
-1.90273568e-01 5.19717276e-01 -2.83950806e-01 -2.16735210e-02
-2.39359245e-01 -1.16568757e-02 2.62176488e-02 1.52936900e+00
2.90877800e-02 8.03175420e-02 -1.93155304e-01 3.20372909e-01
4.44816142e-01 7.54357427e-02 -1.54404923e-01 9.19262990e-02
7.48387814e-01 1.33018529e+00 -9.30915654e-01 -3.93033773e-01
-4.30618316e-01 3.67821902e-01 2.67296076e-01 -1.13076689e-02
-6.63433254e-01 -5.17281771e-01 5.30049145e-01 9.91990045e-02
-5.74901551e-02 1.67288467e-01 -8.80982697e-01 -7.28853881e-01
-1.36889935e-01 -9.02544975e-01 1.11628020e+00 -5.31675875e-01
-1.49517345e+00 4.54614788e-01 5.14200509e-01 -1.26933682e+00
-3.10631126e-01 -9.90426540e-01 -2.79830992e-01 8.96031916e-01
-1.25155616e+00 -7.71982014e-01 -7.90024102e-01 3.93736511e-02
4.45945002e-02 -1.60476282e-01 1.18974769e+00 2.25924924e-01
-4.61190373e-01 8.49678814e-01 3.04050654e-01 1.95108261e-02
1.18831122e+00 -1.35409343e+00 1.24692731e-01 4.12036091e-01
-2.18934417e-01 5.91652036e-01 7.96986520e-01 -4.09454256e-01
-5.39908111e-01 -7.32084513e-01 7.87406504e-01 -9.77642715e-01
4.84285444e-01 -2.23307028e-01 -7.86272705e-01 5.98920345e-01
-2.64744312e-01 -1.06698550e-01 1.21237814e+00 4.65782017e-01
-5.40548027e-01 -5.54564059e-01 -1.36636531e+00 9.61688086e-02
7.01061487e-01 -2.78454691e-01 -3.37308079e-01 4.63942975e-01
1.53356865e-01 -5.93225777e-01 -1.15148938e+00 6.04948938e-01
9.90886688e-01 -1.13880491e+00 8.64432514e-01 -5.18490732e-01
-1.05015136e-01 -4.92316745e-02 -3.27790976e-01 -1.11970937e+00
-4.95757580e-01 -3.57379198e-01 1.15240932e-01 9.10627425e-01
6.42927110e-01 -8.27250302e-01 5.83347976e-01 1.03760910e+00
-1.20324157e-01 -8.19430172e-01 -7.52466738e-01 -7.17254221e-01
3.94742817e-01 -5.91504812e-01 7.65838206e-01 9.45471704e-01
1.88254103e-01 6.27982616e-02 1.65300980e-01 -1.03491701e-01
7.19170272e-01 1.66836143e-01 6.11579418e-01 -1.67086887e+00
5.93997054e-02 -7.35475481e-01 -7.87538230e-01 -4.62158561e-01
-3.16429228e-01 -7.61635065e-01 -2.44247034e-01 -1.21507490e+00
1.94869623e-01 -7.30003834e-01 -4.74397629e-01 4.22794968e-01
-1.79416344e-01 3.56922626e-01 -1.01748168e-01 3.27710360e-01
-1.15341552e-01 -3.00202221e-01 5.01727462e-01 -2.41692364e-01
-2.65696049e-02 4.10339981e-01 -9.66192663e-01 3.55770081e-01
8.52206171e-01 -5.20595610e-01 -3.22125942e-01 1.83567300e-01
2.06891730e-01 -1.84505224e-01 2.33365342e-01 -1.16564226e+00
1.00457191e-01 -2.31763661e-01 8.71332824e-01 -5.35443962e-01
3.90502453e-01 -8.85843098e-01 3.83165717e-01 4.14696097e-01
-5.25968611e-01 3.37709874e-01 2.93787479e-01 5.29055297e-01
-3.31513822e-01 -3.72733682e-01 1.31531203e+00 3.84001993e-02
-5.93541563e-01 3.12194247e-02 -3.08334026e-02 -4.11572754e-02
1.27675402e+00 -4.41420317e-01 -3.24047506e-01 -3.49447459e-01
-9.03411388e-01 2.29494706e-01 4.85813975e-01 4.93416935e-01
1.24695122e-01 -1.14064693e+00 -4.11811709e-01 1.84103921e-01
5.60231030e-01 -4.68570381e-01 -2.23744318e-01 8.50416362e-01
-3.37757677e-01 7.44366944e-01 -2.43494734e-01 -8.03091347e-01
-1.47873652e+00 5.27622283e-01 6.09582186e-01 2.33842582e-02
-1.58537433e-01 7.10627139e-01 -2.65040040e-01 -3.10921699e-01
4.45877165e-01 -1.38072610e-01 -3.07674319e-01 2.62445539e-01
4.55869526e-01 6.24035180e-01 6.70289218e-01 -4.95945692e-01
-6.24248922e-01 6.14369690e-01 7.22070858e-02 -5.92418872e-02
1.03401911e+00 4.95930761e-02 -5.09668924e-02 7.17133343e-01
8.46783221e-01 -4.37662691e-01 -9.09541786e-01 1.17969453e-01
2.16942191e-01 -6.37843490e-01 4.26676385e-02 -1.36546373e+00
-7.03905284e-01 1.08655274e+00 8.69397819e-01 2.09589407e-01
1.03773665e+00 -2.83822045e-02 -1.91354021e-01 2.53276736e-01
4.00146484e-01 -1.19579327e+00 -1.62501097e-01 1.23546332e-01
8.25523794e-01 -1.39273453e+00 5.53897977e-01 -5.36865175e-01
-5.70043921e-01 1.30888343e+00 4.38648582e-01 4.62658256e-01
1.12695169e+00 3.08361799e-01 3.12399805e-01 2.06752896e-01
-4.05836344e-01 1.92893431e-01 4.68089849e-01 1.02441871e+00
7.04968631e-01 2.34962091e-01 -2.49577940e-01 5.50609767e-01
-5.30520439e-01 -2.73013055e-01 1.47976011e-01 9.41497982e-01
-4.99005437e-01 -1.38048422e+00 -6.02624714e-01 8.03905308e-01
-1.71648428e-01 2.13882208e-01 -4.64532763e-01 1.07028401e+00
2.16257811e-01 1.10167754e+00 1.70752242e-01 -4.12414163e-01
4.32526350e-01 2.85709947e-01 4.60350126e-01 -5.33369243e-01
-5.19568861e-01 -3.05208772e-01 5.73620051e-02 -2.54405618e-01
-4.54076439e-01 -1.10402143e+00 -6.37772977e-01 -4.85398531e-01
-6.57094896e-01 2.94219702e-01 1.03539002e+00 6.09595776e-01
6.75853118e-02 2.42321059e-01 4.06842560e-01 -6.08979106e-01
-6.70693398e-01 -9.95754898e-01 -5.25135696e-01 4.55319397e-02
1.94332317e-01 -1.02441072e+00 -6.97448552e-01 -1.67773962e-01]
|
[8.451921463012695, 4.255292892456055]
|
215df617-ab35-4708-b3fc-6d5bc6785b59
|
fexgan-meta-facial-expression-generation-with
|
2203.05975
| null |
https://arxiv.org/abs/2203.05975v1
|
https://arxiv.org/pdf/2203.05975v1.pdf
|
FExGAN-Meta: Facial Expression Generation with Meta Humans
|
The subtleness of human facial expressions and a large degree of variation in the level of intensity to which a human expresses them is what makes it challenging to robustly classify and generate images of facial expressions. Lack of good quality data can hinder the performance of a deep learning model. In this article, we have proposed a Facial Expression Generation method for Meta-Humans (FExGAN-Meta) that works robustly with the images of Meta-Humans. We have prepared a large dataset of facial expressions exhibited by ten Meta-Humans when placed in a studio environment and then we have evaluated FExGAN-Meta on the collected images. The results show that FExGAN-Meta robustly generates and classifies the images of Meta-Humans for the simple as well as the complex facial expressions.
|
['J. Rafid Siddiqui']
|
2022-02-17
| null | null | null | null |
['facial-expression-generation']
|
['computer-vision']
|
[-2.60157622e-02 4.52186950e-02 3.32426637e-01 -8.40289950e-01
-1.01825267e-01 -2.63142854e-01 6.55442894e-01 -9.65509713e-01
-1.66791752e-01 6.44906580e-01 2.13564411e-02 4.05986339e-01
4.02213395e-01 -5.68183243e-01 -3.04326087e-01 -7.94016302e-01
-1.12756290e-01 1.67984248e-03 -6.04109049e-01 -6.19116008e-01
-2.92583555e-01 6.71787262e-01 -1.79096317e+00 5.70262074e-01
7.79413953e-02 1.13432503e+00 -5.65797150e-01 5.89382946e-01
8.89210552e-02 1.10535622e+00 -1.11786592e+00 -6.43984854e-01
4.51175272e-01 -7.99363792e-01 -5.29889524e-01 5.54053843e-01
5.84979415e-01 -4.49648321e-01 -2.09537759e-01 1.01728356e+00
5.89062691e-01 3.24007459e-02 7.33697534e-01 -1.80338073e+00
-6.72239840e-01 4.92985882e-02 -8.04930329e-01 -2.13826299e-01
3.82907361e-01 2.66165793e-01 1.99880421e-01 -5.83056211e-01
7.90860116e-01 1.66547275e+00 5.21319747e-01 1.10965216e+00
-8.29957306e-01 -1.24567080e+00 -3.71863276e-01 -1.96416184e-01
-1.39044118e+00 -6.95824742e-01 9.50379550e-01 -3.77634913e-01
4.54738975e-01 2.16217473e-01 8.06478560e-01 1.46756148e+00
3.16412933e-02 5.11120379e-01 1.74438345e+00 -4.12363946e-01
7.05388782e-04 2.45064020e-01 -3.63777965e-01 8.62003684e-01
-3.62614572e-01 2.09490433e-01 -5.88056266e-01 -1.15661927e-01
7.40785480e-01 -2.29157299e-01 7.00875372e-02 4.37649280e-01
-7.24328160e-01 7.72110879e-01 3.87970954e-01 3.93414319e-01
-5.70680082e-01 2.62979984e-01 5.02381563e-01 5.15198171e-01
5.94673097e-01 4.05402035e-01 -1.77043959e-01 -1.66250497e-01
-8.59223008e-01 3.99433583e-01 4.04114991e-01 9.59098458e-01
8.05076480e-01 5.21531165e-01 -1.36202469e-01 1.09504819e+00
1.12313576e-01 6.93668365e-01 5.01144588e-01 -1.16041458e+00
-1.58499673e-01 6.49867475e-01 -3.05462349e-02 -1.57007694e+00
-2.80578345e-01 1.71960667e-01 -8.60897660e-01 7.90497959e-01
2.04741195e-01 -7.34588623e-01 -9.20402765e-01 1.99758339e+00
3.67952526e-01 -3.65360111e-01 8.71107653e-02 1.09137106e+00
1.15070021e+00 8.14515948e-01 2.18950883e-01 -3.24795283e-02
1.19184017e+00 -8.06434035e-01 -8.50148082e-01 8.17779079e-02
3.79362851e-01 -6.84186161e-01 8.37336779e-01 3.57731909e-01
-1.04701054e+00 -8.19968820e-01 -9.44600761e-01 2.18982980e-01
-2.01757923e-01 2.80710906e-01 8.90910625e-01 7.20433116e-01
-1.20668042e+00 3.54504108e-01 -6.74273893e-02 -3.79264593e-01
5.30636430e-01 3.90365213e-01 -1.07782841e+00 3.79292846e-01
-1.09497941e+00 7.36888170e-01 8.96911472e-02 3.48991811e-01
-8.88475597e-01 -3.71777177e-01 -7.69141555e-01 -4.26148564e-01
-1.29202589e-01 -2.50257611e-01 1.28838205e+00 -2.12153769e+00
-1.80813241e+00 1.55302060e+00 -8.63128714e-03 6.07321411e-02
6.46168470e-01 8.03473219e-02 -7.15686858e-01 3.80179554e-01
-9.78607610e-02 1.11764634e+00 1.11286366e+00 -1.24177718e+00
-1.00563437e-01 -5.52462220e-01 -1.63918182e-01 -3.09161603e-01
2.41476372e-02 5.39866924e-01 7.81630427e-02 -5.23243487e-01
-2.59376466e-01 -1.06266606e+00 1.08268522e-01 2.19100520e-01
-3.06435972e-01 -2.33753085e-01 1.09406722e+00 -3.94561589e-01
6.07643008e-01 -2.28619075e+00 -2.53233463e-01 2.79988885e-01
2.34872803e-01 5.54913461e-01 -2.70848811e-01 3.42094332e-01
-2.90757686e-01 1.45766228e-01 2.20808864e-01 -2.46921793e-01
-3.55427712e-02 3.93743038e-01 2.51245648e-02 5.04536331e-01
2.28476718e-01 9.11061168e-01 -6.92771375e-01 -6.57544076e-01
-8.95617064e-03 5.55933475e-01 -2.69697756e-01 6.48262560e-01
9.69045609e-03 5.06369472e-01 -3.92094254e-01 9.21807051e-01
8.01686287e-01 1.52062118e-01 -1.14378311e-01 -2.71621048e-01
2.33248144e-01 -6.15906060e-01 -4.74271357e-01 1.04616165e+00
-3.46085101e-01 9.00055528e-01 1.84872240e-01 -4.95069563e-01
1.35586059e+00 4.49158996e-01 3.66100132e-01 -8.59816849e-01
6.44189775e-01 1.37256654e-02 -2.92266086e-02 -8.33463311e-01
2.62295067e-01 -7.19813168e-01 -9.31115001e-02 5.20520031e-01
3.73551190e-01 -7.82835484e-02 -5.89037351e-02 -1.04688041e-01
5.79467595e-01 -3.84204276e-03 1.93905085e-01 -1.14334524e-01
4.84805942e-01 -3.08968157e-01 2.94926673e-01 3.42900157e-01
-5.41588664e-01 5.28598666e-01 5.24798572e-01 -7.78840125e-01
-1.09350216e+00 -6.75497174e-01 3.32469761e-01 1.32075810e+00
-3.75082582e-01 -2.49350369e-01 -1.07060766e+00 -6.27829254e-01
-1.80845723e-01 1.77056655e-01 -1.01656687e+00 -1.84623465e-01
-1.43785417e-01 -5.88725388e-01 1.10407805e+00 2.20787525e-01
9.74674940e-01 -1.51502347e+00 -6.52823031e-01 -9.42038223e-02
-1.25329450e-01 -1.20030832e+00 -1.65349826e-01 -2.54910678e-01
-1.82365447e-01 -9.33656216e-01 -7.20502317e-01 -6.29167557e-01
8.37790728e-01 -1.25409335e-01 9.98109400e-01 2.36504361e-01
-5.47840238e-01 6.61538169e-02 -4.71557945e-01 -7.19513595e-01
-6.25597596e-01 -6.68887794e-01 7.13838544e-03 5.66237032e-01
5.79348743e-01 -4.14899141e-01 -2.97880858e-01 3.39793622e-01
-8.38077426e-01 -8.63592997e-02 4.18414772e-01 5.30306339e-01
2.86889821e-02 -3.01025435e-02 4.76998746e-01 -7.42382169e-01
6.59355938e-01 -2.80209929e-01 -2.30335832e-01 -1.09085344e-01
1.31072044e-01 -2.60604918e-01 1.83137685e-01 -5.92027545e-01
-1.15816009e+00 -5.53695969e-02 -2.84059525e-01 -4.28020895e-01
-4.79427934e-01 -5.23470826e-02 -9.42740515e-02 -3.56234580e-01
6.12611055e-01 -1.59099013e-01 5.47084451e-01 2.34139692e-02
1.39990643e-01 9.15773213e-01 6.11120105e-01 -6.19596481e-01
6.47190809e-01 5.77672422e-01 1.44483745e-01 -1.27696598e+00
-7.29113102e-01 1.00420542e-01 -5.10050416e-01 -8.70773494e-01
1.07976747e+00 -9.82521594e-01 -8.40759099e-01 1.13443291e+00
-1.13684094e+00 -4.20743346e-01 1.39178604e-01 1.66197366e-03
-4.18374002e-01 -6.45428523e-02 -6.77014172e-01 -8.92122805e-01
-4.47405130e-01 -9.30791259e-01 1.15963078e+00 2.34609738e-01
-8.04988682e-01 -8.52774024e-01 -8.16658363e-02 4.30558205e-01
5.37202418e-01 1.18200433e+00 5.31746924e-01 8.70721787e-03
1.01685680e-01 -1.83395043e-01 -1.28950030e-01 6.02247536e-01
3.93234342e-01 6.52715445e-01 -1.33914113e+00 4.43977155e-02
-7.19763041e-02 -8.97134066e-01 1.80381626e-01 7.47833401e-02
9.95592296e-01 -5.72522044e-01 2.72516429e-01 8.10862124e-01
1.03068364e+00 2.58758068e-01 8.80262792e-01 1.53046727e-01
4.42211270e-01 9.66689765e-01 4.86703247e-01 4.71690029e-01
-1.14474893e-01 6.87194049e-01 2.49810427e-01 -5.05377710e-01
4.45797183e-02 -2.75793165e-01 5.79042971e-01 5.03348470e-01
-4.83132064e-01 -2.69028964e-03 -3.74683201e-01 1.96456313e-01
-1.32171571e+00 -1.23575437e+00 -4.35414761e-02 1.30914557e+00
8.69967818e-01 -5.47104061e-01 4.29554999e-01 1.24453269e-01
6.73417807e-01 3.53425413e-01 -1.05326377e-01 -1.06499064e+00
-3.55865300e-01 4.19596046e-01 -8.51368606e-02 9.73261222e-02
-9.50425029e-01 9.35755491e-01 7.62274647e+00 4.98911649e-01
-1.76782358e+00 -1.50466319e-02 1.00998926e+00 -1.84463784e-01
2.62646407e-01 -6.32205665e-01 -4.64989126e-01 4.61704880e-01
9.16651666e-01 -1.30640998e-01 1.05062321e-01 1.23584521e+00
1.89300284e-01 1.13652542e-01 -8.94018352e-01 1.33506560e+00
3.42144251e-01 -7.97153175e-01 1.45381212e-01 4.59609106e-02
9.99733686e-01 -4.32493359e-01 3.03012639e-01 2.36728862e-01
2.34110370e-01 -1.54283535e+00 7.84680307e-01 3.84344012e-01
9.19359088e-01 -7.92765617e-01 7.02551484e-01 6.11891374e-02
-5.25990903e-01 2.63223741e-02 -2.94383705e-01 -3.47726345e-01
-2.53750514e-02 2.38377303e-01 -6.98428154e-01 -2.14921180e-02
7.56595552e-01 3.29049498e-01 -4.79663789e-01 4.74022189e-03
-3.57489198e-01 5.07366240e-01 -9.87527743e-02 -2.32945561e-01
4.23107356e-01 -1.61410332e-01 7.35382661e-02 1.40468991e+00
4.08881813e-01 2.46004239e-01 -1.85284585e-01 9.48595047e-01
-1.54487148e-01 1.27156496e-01 -7.88187146e-01 -2.13316277e-01
-2.70828586e-02 1.66789782e+00 -9.84085500e-02 -4.01660979e-01
-6.56886399e-02 9.99460280e-01 6.38946593e-02 3.03854674e-01
-8.35654140e-01 1.47808192e-03 8.28321576e-01 1.81955189e-01
-2.16116011e-01 3.78836058e-02 2.39272699e-01 -8.29613090e-01
-1.47033945e-01 -1.35069108e+00 1.22334592e-01 -1.42796135e+00
-1.26937032e+00 1.15363264e+00 -5.98801933e-02 -8.59426856e-01
-5.48460066e-01 -7.72559941e-01 -6.75371647e-01 7.51731992e-01
-9.74705577e-01 -1.61257803e+00 -9.91033494e-01 8.88777256e-01
4.97350283e-02 -5.86526990e-01 8.19632947e-01 2.07875654e-01
-2.93863893e-01 7.74983406e-01 -8.17168713e-01 4.87272024e-01
6.31273925e-01 -7.11944997e-01 -1.08892083e-01 4.33455557e-01
-5.03887087e-02 5.17414451e-01 6.78852260e-01 -1.13885418e-01
-1.18272972e+00 -1.14063215e+00 5.56933820e-01 -1.24146797e-01
2.76715577e-01 -3.95877868e-01 -5.50541937e-01 6.94759548e-01
3.69180679e-01 2.36437634e-01 8.84070516e-01 -3.35940033e-01
-4.81588423e-01 -2.73720592e-01 -1.57718480e+00 4.73428041e-01
7.67294109e-01 -6.42872691e-01 -3.21785092e-01 1.46853566e-01
4.52896878e-02 -9.58982203e-03 -7.69089818e-01 3.86588335e-01
9.26446378e-01 -1.37396693e+00 5.18280149e-01 -8.24251056e-01
7.75745630e-01 -1.41868472e-01 -2.59283245e-01 -1.37611592e+00
-1.21805027e-01 -8.78390849e-01 4.67531770e-01 1.34537792e+00
-1.97903942e-02 -5.61455488e-01 7.58692563e-01 6.01376951e-01
4.18986589e-01 -4.87585872e-01 -5.93550503e-01 -5.97460747e-01
5.24709821e-02 -2.93817632e-02 9.42495584e-01 1.08408380e+00
-1.85720190e-01 4.94064316e-02 -8.28734398e-01 -3.20456475e-01
4.67585921e-01 -9.05077904e-02 1.42668414e+00 -8.98962080e-01
-2.13449467e-02 -2.32686102e-01 -9.84266698e-01 -2.28182331e-01
5.72428346e-01 -5.98417222e-01 -2.68294722e-01 -6.44073844e-01
1.60292625e-01 -1.54823944e-01 8.57848376e-02 6.74760401e-01
8.23219642e-02 8.01824510e-01 2.01363847e-01 -1.18130617e-01
-1.83091268e-01 5.42664349e-01 1.47355890e+00 -3.77037958e-03
3.16291958e-01 -3.29504013e-01 -7.51967907e-01 8.83561790e-01
8.37319911e-01 -1.77628621e-01 -3.02803606e-01 -7.32162371e-02
5.83729967e-02 -1.14851095e-01 5.26575983e-01 -9.70476985e-01
-4.39282715e-01 -2.54179835e-01 9.15495515e-01 -2.50125200e-01
5.71520507e-01 -5.65370142e-01 5.11889577e-01 1.54730991e-01
-2.61754900e-01 2.28384599e-01 8.80989805e-02 -3.22176993e-01
-5.70108593e-01 1.39555037e-01 1.27121294e+00 -5.86349607e-01
-9.91385162e-01 2.85898477e-01 -4.41118121e-01 -1.11936063e-01
1.19580114e+00 -2.21130654e-01 -1.72768831e-01 -8.25215220e-01
-5.75076222e-01 -3.04717630e-01 5.43636084e-01 6.25438213e-01
6.19915128e-01 -1.70150208e+00 -8.15525770e-01 4.14321840e-01
2.83783257e-01 -4.84651268e-01 2.57934451e-01 4.40659076e-01
-6.59186780e-01 -6.68507889e-02 -1.07315266e+00 -4.90960389e-01
-1.86487675e+00 2.83491582e-01 6.93798363e-01 3.41636300e-01
-2.11518347e-01 9.33626711e-01 3.20711195e-01 -4.61093694e-01
-1.97648823e-01 4.03641090e-02 -3.32464427e-02 1.40439831e-02
8.59590054e-01 1.89708397e-01 -3.17783058e-01 -1.51220310e+00
-1.38974279e-01 5.71031749e-01 2.13787794e-01 -5.87367490e-02
1.26307046e+00 2.18586326e-01 -4.80961859e-01 2.24229097e-01
1.74463117e+00 -8.29527453e-02 -1.02872992e+00 2.69750416e-01
-4.97421920e-01 -6.59671068e-01 -2.65275151e-01 -6.59897625e-01
-1.54907489e+00 9.52541471e-01 6.33667767e-01 -2.65081644e-01
1.33459544e+00 -1.25151426e-01 6.53512418e-01 2.40127996e-01
5.56017280e-01 -1.17468596e+00 6.01411045e-01 1.83557600e-01
1.14019012e+00 -1.13786757e+00 -2.47540191e-01 -7.70515874e-02
-1.01861775e+00 1.15338159e+00 9.26653385e-01 -7.46701732e-02
5.34574628e-01 4.04015064e-01 8.84618163e-01 -5.18740237e-01
-7.83426404e-01 8.51025879e-02 1.28618209e-02 8.48340929e-01
5.38468003e-01 2.40965754e-01 2.47019231e-01 2.14835450e-01
-6.50923014e-01 3.42417777e-01 4.82981563e-01 7.22684562e-01
-2.00063393e-01 -8.32247734e-01 -5.29405415e-01 -1.42866492e-01
-8.18778276e-01 6.09251797e-01 -9.22385573e-01 1.03370368e+00
5.27122915e-01 1.02200508e+00 -1.08859152e-01 -6.67291880e-01
2.57993996e-01 1.17573321e-01 6.61930978e-01 -1.19353585e-01
-6.19910419e-01 -2.21681371e-01 2.44331881e-01 -6.46242261e-01
-8.62045944e-01 -3.89214009e-01 -9.82115924e-01 -6.52762294e-01
2.89701909e-01 1.82892513e-02 6.63728774e-01 6.37868822e-01
1.65671095e-01 9.67754796e-02 8.91345024e-01 -9.82142568e-01
-2.68600076e-01 -1.13446200e+00 -8.46281052e-01 1.21545553e+00
3.32325846e-01 -7.21585751e-01 -5.03787637e-01 2.59489805e-01]
|
[13.49331283569336, 1.7358776330947876]
|
661b69e4-c36b-4896-b203-7fc3b73efc2c
|
heterogeneous-target-speech-separation
|
2204.03594
| null |
https://arxiv.org/abs/2204.03594v1
|
https://arxiv.org/pdf/2204.03594v1.pdf
|
Heterogeneous Target Speech Separation
|
We introduce a new paradigm for single-channel target source separation where the sources of interest can be distinguished using non-mutually exclusive concepts (e.g., loudness, gender, language, spatial location, etc). Our proposed heterogeneous separation framework can seamlessly leverage datasets with large distribution shifts and learn cross-domain representations under a variety of concepts used as conditioning. Our experiments show that training separation models with heterogeneous conditions facilitates the generalization to new concepts with unseen out-of-domain data while also performing substantially higher than single-domain specialist models. Notably, such training leads to more robust learning of new harder source separation discriminative concepts and can yield improvements over permutation invariant training with oracle source selection. We analyze the intrinsic behavior of source separation training with heterogeneous metadata and propose ways to alleviate emerging problems with challenging separation conditions. We release the collection of preparation recipes for all datasets used to further promote research towards this challenging task.
|
['Jonathan Le Roux', 'Paris Smaragdis', 'Aswin Subramanian', 'Gordon Wichern', 'Efthymios Tzinis']
|
2022-04-07
| null | null | null | null |
['speech-separation']
|
['speech']
|
[ 5.60109198e-01 -3.00634116e-01 -3.70503545e-01 -4.24448013e-01
-1.69776797e+00 -1.18087757e+00 7.54045308e-01 1.78531423e-01
-3.48022617e-02 7.51244068e-01 4.81274277e-01 -2.60441191e-02
-5.74690580e-01 -3.40691328e-01 -5.87981880e-01 -9.51680243e-01
-1.75807104e-01 5.25987089e-01 9.97228827e-03 6.49716258e-02
-3.88737172e-02 4.76148814e-01 -1.57475471e+00 2.48333052e-01
7.89232075e-01 9.49294627e-01 2.36199750e-03 6.89488232e-01
1.12147614e-01 2.95612693e-01 -9.51867521e-01 -1.23304501e-01
3.90358895e-01 -4.18443322e-01 -3.96723181e-01 -2.92096525e-01
5.91283798e-01 2.11916625e-01 -1.53938800e-01 9.84632552e-01
1.00810671e+00 1.67326257e-01 9.50773180e-01 -1.39945030e+00
-5.51823556e-01 9.51206982e-01 -6.29179299e-01 4.04429764e-01
2.17555463e-01 -2.68596739e-01 1.07668149e+00 -7.61555552e-01
8.25002491e-02 1.08220470e+00 8.00403297e-01 3.40662062e-01
-1.59353447e+00 -1.26052713e+00 2.04409882e-01 2.29480974e-02
-1.49448597e+00 -9.18027997e-01 7.54614830e-01 -1.68902084e-01
3.38811606e-01 5.20540059e-01 5.64626344e-02 1.73829043e+00
-6.78047955e-01 9.74829912e-01 1.07441735e+00 -2.40507767e-01
3.22096556e-01 3.06052744e-01 1.76885188e-01 -2.46453471e-02
5.56549788e-01 -5.84459715e-02 -9.12671745e-01 -5.26273310e-01
3.22628558e-01 -1.27658173e-01 -4.01408553e-01 -5.53265810e-01
-1.27942824e+00 7.25301802e-01 -2.28392407e-02 3.06808770e-01
1.33921608e-01 -1.89618710e-02 2.23555371e-01 1.14025235e-01
4.18021500e-01 5.46234250e-01 -9.25280511e-01 -2.01050088e-01
-1.11380351e+00 2.08507031e-01 1.02021086e+00 1.20856547e+00
6.26074910e-01 2.13322222e-01 -2.86148816e-01 1.11270308e+00
-3.65063101e-02 1.26467288e+00 5.14739931e-01 -8.59190941e-01
5.23175716e-01 -1.49588495e-01 2.92570144e-02 -6.82051539e-01
-4.86932814e-01 -9.03052688e-01 -6.75236642e-01 -3.31650347e-01
5.80134273e-01 -4.82852638e-01 -9.29064095e-01 2.25445938e+00
3.45418304e-02 5.05037606e-01 3.28272611e-01 4.53503460e-01
7.89216697e-01 4.22741175e-01 1.28239557e-01 -6.69561923e-02
1.17019093e+00 -5.88317394e-01 -4.51587856e-01 -3.89725477e-01
1.76920727e-01 -6.65645361e-01 6.60769522e-01 5.09760022e-01
-8.95095587e-01 -5.20344138e-01 -8.48735809e-01 3.39644611e-01
-4.70344514e-01 1.60377309e-01 8.42360735e-01 1.18316507e+00
-7.76558280e-01 2.51852483e-01 -4.66237038e-01 -2.53896415e-01
5.61610281e-01 3.97683769e-01 -2.60331780e-01 -6.06652088e-02
-1.13795888e+00 2.64925212e-01 1.92344218e-01 -3.73354286e-01
-1.18668234e+00 -1.21910822e+00 -9.51566577e-01 2.19237387e-01
4.79729623e-01 -4.54336494e-01 9.19690132e-01 -9.79572654e-01
-1.35338867e+00 6.77390218e-01 -1.93279669e-01 -2.92161107e-01
5.89672737e-02 -2.82983273e-01 -8.53202879e-01 8.98179635e-02
5.30293763e-01 3.81457239e-01 1.27845573e+00 -1.57149613e+00
-6.93028688e-01 -3.65334541e-01 -3.21514100e-01 1.39646739e-01
-5.69438756e-01 3.41291577e-02 -3.26128393e-01 -9.35441911e-01
5.39157391e-02 -1.08473349e+00 1.60399914e-01 -4.99524027e-01
-6.94913328e-01 1.13026693e-01 3.86908859e-01 -5.75096846e-01
8.64663363e-01 -2.57590985e+00 2.04643473e-01 4.99382228e-01
1.68868095e-01 -5.53591400e-02 -3.99026603e-01 1.36343285e-01
-2.88948685e-01 -1.26880094e-01 -3.19700956e-01 -3.83998930e-01
2.21547067e-01 -2.09972441e-01 -5.87496936e-01 5.96985579e-01
4.49627191e-02 6.44714952e-01 -7.98614442e-01 -1.74932659e-01
-1.54188693e-01 4.49768484e-01 -5.91394246e-01 1.53253809e-01
1.69488758e-01 6.13822460e-01 -2.37624586e-01 9.55811501e-01
9.66417611e-01 -2.20065147e-01 1.80921644e-01 -9.21920761e-02
3.23470384e-01 2.34297872e-01 -1.50332999e+00 1.93147492e+00
-4.99599397e-01 6.99449658e-01 5.25965452e-01 -1.11979306e+00
7.10096240e-01 2.06174999e-01 6.82794392e-01 -4.08497870e-01
4.48197946e-02 2.40028873e-01 5.26605584e-02 -1.31882504e-01
3.17010283e-01 -3.08584213e-01 -5.00124872e-01 4.37655777e-01
5.54114699e-01 -5.46014346e-02 -2.16586575e-01 1.88098922e-01
9.36552167e-01 -2.04676256e-01 1.37878984e-01 -3.88648838e-01
1.75997436e-01 -4.61592317e-01 4.98107791e-01 1.27616405e+00
-2.88058370e-01 7.08839118e-01 1.26424253e-01 4.20537114e-01
-3.77466708e-01 -1.61118555e+00 -4.93071765e-01 1.54440141e+00
1.96570069e-01 -2.67703146e-01 -2.90888935e-01 -6.12089097e-01
3.46615791e-01 5.31058371e-01 -5.98962128e-01 -3.00288767e-01
-2.30335668e-01 -1.09569943e+00 1.11958599e+00 7.21136391e-01
9.72335711e-02 -4.38097268e-01 -3.35341096e-02 -2.19236568e-01
-3.86197418e-01 -1.20909739e+00 -2.56776810e-01 9.19743001e-01
-5.85572362e-01 -8.57879639e-01 -9.43849802e-01 -6.76468551e-01
2.01041073e-01 4.90037352e-01 1.01246989e+00 -5.08422852e-01
-5.60578629e-02 7.46325493e-01 -3.47730041e-01 -5.92231810e-01
2.93633919e-02 2.27066696e-01 3.04700077e-01 1.95769548e-01
3.56974512e-01 -8.67908478e-01 -3.62562180e-01 1.29003316e-01
-8.08406293e-01 -5.85612833e-01 5.86310863e-01 6.67237639e-01
2.22435668e-01 2.90044963e-01 9.23005104e-01 -8.89863193e-01
5.84944725e-01 -9.83170211e-01 -7.66548216e-02 1.72058508e-01
-2.65916318e-01 2.53609240e-01 5.41855216e-01 -7.77861238e-01
-1.23854089e+00 -8.34391788e-02 1.69770956e-01 -2.68226266e-01
-6.41059279e-01 3.16530541e-02 -6.37935400e-01 -5.27512878e-02
7.61604428e-01 2.74365753e-01 -4.93916035e-01 -6.19872868e-01
5.56776404e-01 6.70600891e-01 6.52844071e-01 -1.06832421e+00
1.05530477e+00 7.37043917e-01 -1.77441910e-01 -7.90229619e-01
-8.27449918e-01 -8.83374870e-01 -6.88574493e-01 5.05515993e-01
4.77425963e-01 -1.42178166e+00 -2.11428478e-01 4.32465583e-01
-6.79349303e-01 -2.95888186e-01 -3.47871840e-01 5.97325265e-01
-4.09622103e-01 2.29724571e-01 -1.19713219e-02 -8.26497138e-01
1.80413187e-01 -7.90886760e-01 1.44652033e+00 1.66375205e-01
-1.83004186e-01 -1.14368129e+00 1.77793756e-01 3.12097371e-01
3.32080781e-01 -4.12144847e-02 7.60690272e-01 -1.29683065e+00
-2.43754193e-01 1.46446005e-01 -8.06163028e-02 2.54123867e-01
3.19178700e-01 -4.74649072e-01 -1.61689568e+00 -3.03856015e-01
-3.09382156e-02 -3.13915223e-01 1.04942918e+00 5.50595522e-01
1.22815478e+00 3.59333158e-02 -7.60739267e-01 1.07986712e+00
1.11233473e+00 2.13828776e-02 2.09783867e-01 7.62070268e-02
6.30271196e-01 5.49339652e-01 1.56299755e-01 5.85893869e-01
1.30063742e-01 6.71177268e-01 -8.39623734e-02 -1.71394959e-01
-1.78729340e-01 -6.05872758e-02 4.60182160e-01 2.17373937e-01
3.40106934e-01 -3.15039247e-01 -6.95731044e-01 8.74598622e-01
-1.26261127e+00 -1.18742073e+00 1.16277896e-01 2.26860571e+00
1.14320230e+00 -6.95855841e-02 3.55353951e-01 2.88687974e-01
6.00089967e-01 8.24683979e-02 -4.71745849e-01 2.39056811e-01
-5.28348148e-01 6.40432000e-01 7.84611106e-01 5.14635324e-01
-1.37989163e+00 5.60027659e-01 6.82573271e+00 1.18352473e+00
-1.09840143e+00 2.65655845e-01 2.93546617e-01 -3.77269268e-01
-5.81270814e-01 -2.25285694e-01 -8.78555119e-01 5.55329621e-01
9.54271257e-01 -5.69599085e-02 4.50045586e-01 5.45569301e-01
-3.52567375e-01 -7.40223899e-02 -1.35837018e+00 1.23028219e+00
6.09876871e-01 -7.20532119e-01 -3.37713480e-01 -2.23734766e-01
7.07314134e-01 6.43811896e-02 4.59116727e-01 5.03858209e-01
4.50727969e-01 -1.03114104e+00 8.35360050e-01 1.12731129e-01
7.89528012e-01 -5.63345253e-01 1.83271945e-01 2.61087716e-01
-1.05189359e+00 -3.49701583e-01 -4.31957133e-02 3.14556420e-01
-7.55782500e-02 6.95160568e-01 -6.79318488e-01 7.19992757e-01
7.23514318e-01 6.55443788e-01 -8.37249577e-01 9.26648378e-01
-2.74435338e-03 8.10743213e-01 -5.90746224e-01 4.50095832e-01
-3.68789464e-01 2.23895758e-01 7.50257671e-01 1.45179784e+00
4.74831074e-01 -2.25680515e-01 8.11522231e-02 6.10770226e-01
8.17579404e-03 -1.04451559e-01 -5.45134008e-01 1.01253353e-01
6.96170032e-01 1.11725354e+00 -7.96861291e-01 -1.71598554e-01
-4.08936352e-01 9.25322533e-01 -2.38756631e-02 8.06461573e-01
-9.80091631e-01 -3.85163158e-01 9.69515204e-01 -1.14776991e-01
4.94811594e-01 -1.19332917e-01 -5.03699481e-01 -1.45930278e+00
-3.21887821e-01 -9.79866207e-01 5.84880173e-01 -5.27165353e-01
-1.65555513e+00 2.70129681e-01 4.23993856e-01 -1.15995288e+00
7.53701031e-02 -6.07023001e-01 -3.55153859e-01 8.30581605e-01
-1.67688572e+00 -1.13435400e+00 -9.27515700e-02 1.05609715e+00
2.93947279e-01 -4.93579805e-01 1.01728690e+00 5.62991798e-01
-3.24528098e-01 9.30717289e-01 4.80024695e-01 3.26516718e-01
1.26532638e+00 -1.31120670e+00 -9.58557203e-02 7.33175159e-01
7.18626320e-01 7.05953658e-01 6.19788647e-01 -3.35959494e-01
-1.22437263e+00 -8.91666174e-01 3.64509374e-01 -9.90201414e-01
7.64471054e-01 -8.31813037e-01 -3.34550440e-01 6.72656775e-01
9.31169614e-02 3.10175233e-02 1.49242115e+00 7.73916304e-01
-9.22616243e-01 -4.10607785e-01 -1.03670585e+00 2.54195362e-01
1.04751730e+00 -7.56157756e-01 -4.65360701e-01 1.89969510e-01
5.62124133e-01 -1.51312932e-01 -6.08764410e-01 3.24760526e-01
3.91019076e-01 -7.82295108e-01 1.46615231e+00 -6.07282937e-01
-1.40408948e-01 -1.43838286e-01 -6.00934863e-01 -1.55051291e+00
-4.80011225e-01 -6.84255183e-01 -4.89835441e-02 1.57387567e+00
7.35407054e-01 -6.37102962e-01 4.67530310e-01 4.19292957e-01
2.32001045e-03 3.60772386e-02 -8.69094551e-01 -8.90677333e-01
2.51678109e-01 -6.89612567e-01 6.67646766e-01 9.86925423e-01
-6.98871911e-03 3.62948745e-01 -3.87152374e-01 6.68416381e-01
8.03762257e-01 5.29098213e-01 7.22369373e-01 -1.38724327e+00
-6.57828093e-01 -4.45583254e-01 -2.21654624e-01 -1.15278482e+00
6.02842093e-01 -1.16523623e+00 8.74248669e-02 -9.41781759e-01
3.39396238e-01 -8.76827061e-01 -6.13871336e-01 4.04096276e-01
-2.35773385e-01 4.14870352e-01 1.48357436e-01 8.04109350e-02
-7.82530487e-01 3.63878012e-01 7.38386333e-01 -4.58105326e-01
-1.84486628e-01 3.36426079e-01 -1.56448317e+00 3.40528965e-01
7.76942372e-01 -7.42418230e-01 -6.22612834e-01 -4.78979111e-01
5.42218983e-02 -2.43583828e-01 4.23054487e-01 -1.21843672e+00
-5.92132211e-02 -2.80370861e-01 6.81247115e-01 -1.55476004e-01
4.82133657e-01 -9.64134514e-01 -9.17108729e-02 -2.10648149e-01
-6.24299169e-01 -6.34710968e-01 5.43047905e-01 8.23317111e-01
-1.87962413e-01 9.81015619e-03 5.72545946e-01 2.20078662e-01
-3.99495691e-01 1.43292323e-01 -9.71029326e-02 7.78925598e-01
9.62057948e-01 5.58395684e-02 -2.37980038e-01 -5.50798059e-01
-8.26951683e-01 -1.74377784e-01 7.53898099e-02 5.50042152e-01
1.46842405e-01 -1.35894394e+00 -7.26716876e-01 4.81726974e-01
2.28249222e-01 -1.92375630e-01 8.12050626e-02 8.00028443e-01
3.69805306e-01 3.25614065e-01 4.66567352e-02 -6.26390338e-01
-1.07188511e+00 5.87699473e-01 2.60196537e-01 7.48732984e-02
-1.86169952e-01 1.20267797e+00 5.75851142e-01 -4.87447530e-01
3.25097829e-01 -2.20160872e-01 -1.07653461e-01 4.13841695e-01
4.15206999e-01 4.10914749e-01 8.42140615e-02 -6.63994670e-01
-6.02475703e-01 4.22250539e-01 4.31411713e-01 -3.55877668e-01
1.31863213e+00 -1.99797407e-01 1.43588334e-01 4.69075501e-01
1.21919346e+00 9.91818011e-01 -1.19392812e+00 -3.34468186e-01
-8.80358815e-02 -5.16931355e-01 -2.89971173e-01 -1.10439098e+00
-9.57424700e-01 8.83768737e-01 5.63251376e-01 2.23314807e-01
1.24750495e+00 4.34112757e-01 4.19344366e-01 1.77953213e-01
2.44564280e-01 -9.58911419e-01 3.39560769e-02 4.69770312e-01
4.71708417e-01 -1.27505887e+00 -1.78317681e-01 -3.15652639e-01
-5.90900540e-01 8.90427232e-01 2.83172578e-01 2.11704239e-01
1.03932798e+00 7.65843511e-01 8.31327662e-02 6.26057386e-02
-2.45961323e-01 -4.92151529e-01 4.51236576e-01 1.36707485e+00
3.16772133e-01 3.29195082e-01 4.66680557e-01 1.33213282e+00
-2.73441344e-01 -6.27970159e-01 1.26897410e-01 5.16394615e-01
-1.70890316e-01 -1.21490705e+00 -7.20246315e-01 3.99439931e-01
-4.87255245e-01 -4.11808759e-01 -3.89776051e-01 5.40248275e-01
3.56071830e-01 1.17471302e+00 -3.59859169e-02 -1.49091527e-01
1.11452542e-01 2.74190068e-01 5.99968135e-01 -8.26779425e-01
-3.96723688e-01 2.74521291e-01 -1.54461954e-02 -2.62188584e-01
-5.97797096e-01 -1.07363391e+00 -1.03052235e+00 7.80309141e-02
-1.80738360e-01 1.95171475e-01 3.58282924e-01 8.69773269e-01
4.05504167e-01 4.11153287e-01 4.90688562e-01 -8.93820703e-01
-6.12155080e-01 -8.91283154e-01 -8.90564442e-01 6.39268696e-01
7.40292966e-01 -9.00804996e-01 -6.91031218e-01 1.97409734e-01]
|
[15.342333793640137, 5.494719505310059]
|
1723b7e9-3ee1-49c4-8d4c-d5404e76289a
|
mobile-authentication-of-copy-detection-1
|
2203.02397
| null |
https://arxiv.org/abs/2203.02397v2
|
https://arxiv.org/pdf/2203.02397v2.pdf
|
Mobile authentication of copy detection patterns
|
In the recent years, the copy detection patterns (CDP) attracted a lot of attention as a link between the physical and digital worlds, which is of great interest for the internet of things and brand protection applications. However, the security of CDP in terms of their reproducibility by unauthorized parties or clonability remains largely unexplored. In this respect this paper addresses a problem of anti-counterfeiting of physical objects and aims at investigating the authentication aspects and the resistances to illegal copying of the modern CDP from machine learning perspectives. A special attention is paid to a reliable authentication under the real life verification conditions when the codes are printed on an industrial printer and enrolled via modern mobile phones under regular light conditions. The theoretical and empirical investigation of authentication aspects of CDP is performed with respect to four types of copy fakes from the point of view of (i) multi-class supervised classification as a baseline approach and (ii) one-class classification as a real-life application case. The obtained results show that the modern machine-learning approaches and the technical capacities of modern mobile phones allow to reliably authenticate CDP on end-user mobile phones under the considered classes of fakes.
|
['Slava Voloshynovskiy', 'Slavi Bonev', 'Roman Chaban', 'Taras Holotyak', 'Joakim Tutt', 'Olga Taran']
|
2022-03-04
| null | null | null | null |
['one-class-classification']
|
['miscellaneous']
|
[ 4.18739051e-01 -9.37208682e-02 -4.70469892e-01 2.46289149e-01
-5.28304636e-01 -6.82794213e-01 1.06261539e+00 2.00179681e-01
-1.49497926e-01 6.21922731e-01 -6.11785233e-01 -6.55314445e-01
-3.33158493e-01 -8.86118054e-01 -6.31952405e-01 -7.85608888e-01
-3.84479575e-02 3.88667136e-01 2.03705475e-01 -2.40815103e-01
5.91885209e-01 7.71253765e-01 -1.78190875e+00 4.55211699e-01
7.92670965e-01 1.05954981e+00 -3.78796488e-01 5.37930131e-01
1.57801080e-02 1.09452561e-01 -6.20563984e-01 -9.27888095e-01
3.31786036e-01 -2.84531713e-01 -6.41501069e-01 1.87082529e-01
2.38886371e-01 5.12472279e-02 -3.56009677e-02 1.24096298e+00
1.55284733e-01 -5.89508295e-01 7.47472644e-01 -1.76969039e+00
-5.22390366e-01 3.44605714e-01 -3.69908333e-01 -8.32005367e-02
6.21112406e-01 1.31305605e-01 7.12453306e-01 -5.77155888e-01
3.61434728e-01 1.01917708e+00 7.32597232e-01 9.50160995e-02
-1.12630785e+00 -5.41815102e-01 -4.70871478e-01 4.93421674e-01
-1.32631516e+00 -1.30968302e-01 6.57397389e-01 -7.26245940e-01
3.30269843e-01 3.00863922e-01 4.27183986e-01 1.17692149e+00
6.79774225e-01 3.58103335e-01 1.53233588e+00 -6.67451799e-01
2.31957242e-01 7.85874605e-01 4.34719592e-01 4.17804599e-01
8.31543863e-01 2.75847107e-01 -2.51963943e-01 -3.91322851e-01
4.31885481e-01 -2.41309762e-01 -1.73557043e-01 -2.89399475e-01
-1.14776647e+00 7.75751472e-01 -1.38339520e-01 9.84385192e-01
-9.32273194e-02 -3.47449273e-01 3.61923546e-01 5.28094471e-01
5.01799732e-02 4.20298398e-01 -4.77088630e-01 -4.23610955e-02
-6.64389253e-01 -1.97185561e-01 9.39159036e-01 8.72492194e-01
5.21157503e-01 -7.32575729e-02 3.99444371e-01 3.80119473e-01
1.96118936e-01 7.14523315e-01 6.40706956e-01 -1.12989917e-01
7.30995238e-01 3.93942505e-01 4.32797670e-01 -1.66372359e+00
-1.36853218e-01 -3.31552386e-01 -9.72095966e-01 3.01565111e-01
8.72246265e-01 3.08957607e-01 8.27477034e-03 1.03571784e+00
8.96216631e-02 2.15332732e-01 -7.61142522e-02 2.16801077e-01
-8.79570320e-02 4.59181786e-01 -1.98054835e-01 -2.92715639e-01
1.56530344e+00 -4.65019226e-01 -1.01669371e+00 1.45166725e-01
7.91544676e-01 -1.11334586e+00 1.06269085e+00 9.70429599e-01
-6.16346180e-01 -7.74431705e-01 -1.44534922e+00 6.09560311e-01
-8.77643764e-01 4.34232324e-01 4.13876086e-01 1.67245853e+00
-6.38567090e-01 7.32560992e-01 -3.34234089e-01 -2.11176604e-01
3.14674169e-01 5.01944602e-01 -5.27177930e-01 1.59336403e-01
-1.36969578e+00 9.15834367e-01 2.94803828e-01 1.61907464e-01
-1.76161662e-01 -6.01918735e-02 -2.15046629e-01 -9.14257690e-02
6.32004291e-02 4.90664393e-02 8.13933790e-01 -1.11950243e+00
-1.36852598e+00 1.15104067e+00 3.42661649e-01 -3.82365525e-01
9.81738806e-01 5.47911152e-02 -1.12627733e+00 -5.14865704e-02
-3.18567082e-02 -4.31555480e-01 1.37834096e+00 -1.31084681e+00
-6.25606418e-01 -4.03987795e-01 -3.55988503e-01 -6.38405681e-01
-4.39890653e-01 -2.18990058e-01 7.98358992e-02 -5.07184565e-01
1.56333834e-01 -1.19749296e+00 5.47660232e-01 -3.06614880e-02
-6.06146514e-01 1.62522063e-01 7.29607105e-01 -6.96989715e-01
1.09845686e+00 -2.00819683e+00 -2.19383597e-01 6.24531806e-01
-3.94675553e-01 6.19783401e-01 1.45391002e-01 5.37749231e-01
-3.80235106e-01 2.34838665e-01 -1.47771742e-02 -9.58520733e-03
4.34725694e-02 -1.77208304e-01 -3.93492728e-01 8.89444292e-01
-5.01117529e-03 3.52303743e-01 -7.07090497e-01 -4.00867492e-01
2.40846068e-01 1.77965239e-01 -3.60687226e-02 -2.79363282e-02
8.63868520e-02 3.20975155e-01 -4.93957579e-01 6.74322128e-01
1.11990499e+00 -3.32009271e-02 3.88420433e-01 -1.54369980e-01
-9.29596275e-03 -2.05162883e-01 -1.53934693e+00 8.93662453e-01
-4.91653055e-01 3.28226417e-01 -1.80932537e-01 -9.93677616e-01
9.86115694e-01 2.87169904e-01 2.19043836e-01 -5.91590524e-01
2.54492491e-01 7.29291737e-01 -1.52365685e-01 -4.71343100e-01
4.69225764e-01 -8.18528875e-04 -1.08955689e-01 6.79516792e-01
-2.04193816e-01 3.60829830e-02 -1.16622426e-01 -2.88084865e-01
8.09342742e-01 -9.09135565e-02 5.47894657e-01 -1.78232968e-01
1.08283246e+00 -5.09936392e-01 -1.05063461e-01 6.22966766e-01
-8.80432576e-02 2.35152051e-01 4.87196475e-01 -1.28669649e-01
-8.31037700e-01 -8.24183881e-01 -2.96451271e-01 3.96271080e-01
2.62801677e-01 3.35844159e-01 -7.11963356e-01 -6.14293814e-01
3.00029486e-01 5.16829193e-01 -2.06278309e-01 -2.03843474e-01
-3.96115810e-01 -8.03695440e-01 8.88704538e-01 -1.64251760e-01
6.19478226e-01 -7.36965418e-01 -8.44946131e-02 -3.66986101e-03
-1.43490359e-01 -1.43810177e+00 1.12133779e-01 -1.64664730e-01
-8.27281237e-01 -1.43248105e+00 -5.79218805e-01 -7.64262795e-01
3.57990384e-01 1.20434813e-01 6.22803628e-01 4.49445546e-01
-7.39153847e-02 2.53372997e-01 -3.89800280e-01 -1.67180821e-01
-1.19777155e+00 2.51903147e-01 3.11730534e-01 6.96191788e-01
2.52197891e-01 -3.80508453e-01 -1.07495628e-01 8.97571087e-01
-8.28872383e-01 -4.70146686e-01 6.16103351e-01 5.39788902e-01
-1.04774602e-01 5.80794096e-01 4.96128768e-01 -8.05198252e-01
5.38746715e-01 -5.01034796e-01 -9.45089400e-01 3.82671624e-01
-7.74232090e-01 -4.05631512e-02 5.39125323e-01 -6.38481379e-01
-8.03347111e-01 -1.04368970e-01 -2.72382587e-01 3.20067406e-01
-2.39051208e-01 -1.44242615e-01 -6.89387381e-01 -4.83179718e-01
6.65724516e-01 4.00120139e-01 -8.71608779e-02 -4.81431007e-01
1.64228559e-01 1.13342488e+00 3.93053770e-01 -5.23661792e-01
1.36771715e+00 4.64591444e-01 2.35043660e-01 -1.11024058e+00
-4.65157256e-02 -3.16516012e-01 -7.55756140e-01 -2.87184596e-01
3.53317410e-01 -4.14395958e-01 -1.08589065e+00 1.17197335e+00
-1.34142232e+00 2.73660839e-01 2.06010535e-01 3.92823160e-01
-5.82062781e-01 1.11346030e+00 -6.23331249e-01 -1.08236885e+00
-2.97064055e-03 -1.10017943e+00 8.58061612e-01 -3.73351514e-01
-1.64698258e-01 -8.47762942e-01 -1.65972695e-01 5.43664634e-01
5.09943292e-02 1.91216797e-01 1.27074552e+00 -8.26908648e-01
-7.41762161e-01 -8.91628623e-01 -6.87980503e-02 4.41515177e-01
3.78173709e-01 -1.57187656e-01 -1.09341776e+00 -3.31895709e-01
3.62356484e-01 2.51115710e-01 5.06731384e-02 -2.86997646e-01
7.32272863e-01 -3.69619220e-01 -3.43439996e-01 5.73259369e-02
1.60136044e+00 3.96831423e-01 1.09963417e+00 5.67971706e-01
1.76850751e-01 7.14391887e-01 8.38945627e-01 3.76742095e-01
-5.34230173e-01 1.00849438e+00 6.25763357e-01 2.62003183e-01
4.07896072e-01 -2.05970049e-01 2.19721749e-01 6.86164260e-01
5.99165000e-02 -5.41521549e-01 -7.06337631e-01 1.98142007e-01
-1.33558273e+00 -1.09061396e+00 -5.82253158e-01 2.61498523e+00
3.16309005e-01 5.60227513e-01 1.61959246e-01 1.17977011e+00
1.38029158e+00 -7.81620145e-02 7.92520791e-02 -3.46355051e-01
-2.97149956e-01 -2.66095879e-03 5.17602742e-01 3.53857458e-01
-9.52668369e-01 2.88626671e-01 4.95245504e+00 1.12987554e+00
-1.02595973e+00 3.03877473e-01 5.71481228e-01 9.92401421e-01
2.38010898e-01 -1.17136866e-01 -8.02066505e-01 1.02064764e+00
7.13264942e-01 5.11165857e-01 3.08155835e-01 6.45655572e-01
2.22039834e-01 -1.57414377e-01 -9.05234158e-01 1.15756631e+00
8.04668143e-02 -1.09232295e+00 1.22500919e-01 4.86464500e-01
4.43423599e-01 -6.73181534e-01 3.62034112e-01 -1.52423814e-01
-5.76371312e-01 -6.78440392e-01 7.97635078e-01 5.95141649e-01
7.99836338e-01 -5.72252870e-01 1.09806347e+00 4.55198556e-01
-8.73365700e-01 -2.55648285e-01 -8.41786042e-02 6.29634634e-02
3.45859043e-02 6.75133109e-01 -6.12767160e-01 7.06903219e-01
5.68019599e-02 1.86592087e-01 -5.31467319e-01 6.96221769e-01
-1.70944948e-02 4.82056826e-01 -4.27951999e-02 -2.71042705e-01
-4.90060747e-02 -3.70795310e-01 2.68442184e-01 1.08927846e+00
5.65636933e-01 -4.86434788e-01 -5.59228539e-01 6.17320299e-01
1.40613288e-01 1.92953080e-01 -6.63824022e-01 -4.07485008e-01
2.47982576e-01 1.00346339e+00 -9.95799303e-01 -6.65726978e-03
-1.33326486e-01 9.84103143e-01 -4.27514374e-01 -2.97378562e-02
-8.54242444e-01 -4.33052152e-01 2.76265174e-01 5.68321466e-01
2.53622413e-01 -2.06383914e-01 -3.64309669e-01 -1.09284246e+00
1.84688345e-01 -1.02316833e+00 6.11002892e-02 -4.29623455e-01
-1.32365227e+00 3.98916304e-01 -4.16814953e-01 -1.92081702e+00
1.21870086e-01 -1.03687644e+00 -4.56846267e-01 6.38978124e-01
-1.23047733e+00 -1.14764631e+00 -6.45264238e-02 4.73510802e-01
1.47107080e-01 -5.59396565e-01 8.26600611e-01 3.88311207e-01
-4.78061996e-02 6.64555252e-01 5.95530987e-01 7.96471760e-02
4.05285686e-01 -7.31781721e-01 4.10646290e-01 5.92311502e-01
8.13055709e-02 5.51867306e-01 7.98857629e-01 -7.94360876e-01
-1.37310433e+00 -5.11471987e-01 1.22717369e+00 -4.49252576e-01
8.27182591e-01 -3.43925774e-01 -7.79481232e-01 1.87183619e-01
-7.43185356e-02 -2.13997364e-01 5.33788383e-01 -2.45313764e-01
-3.63253325e-01 -4.64677773e-02 -1.46186733e+00 2.53212422e-01
5.87075710e-01 -8.64415228e-01 -3.80713046e-01 5.42432725e-01
2.90366746e-02 3.22163612e-01 -7.86490440e-01 -1.75140098e-01
8.27376664e-01 -1.17039573e+00 1.10243416e+00 -3.46964866e-01
4.44166288e-02 -2.99431831e-01 -4.47553881e-02 -6.22313559e-01
1.01725273e-01 -8.21294546e-01 9.62047353e-02 1.27897155e+00
1.17481314e-01 -9.41163003e-01 7.40640402e-01 -6.88151941e-02
6.05230272e-01 -3.15887958e-01 -1.25911474e+00 -1.16077197e+00
-1.21163934e-01 -5.19384205e-01 6.53293014e-01 1.19826162e+00
1.56066969e-01 -1.09292321e-01 -5.72073042e-01 4.08955276e-01
5.88108957e-01 -8.76387730e-02 7.89718568e-01 -1.40308583e+00
-5.39218903e-01 -2.30875328e-01 -1.00069427e+00 -6.62506104e-01
3.87952700e-02 -7.50466347e-01 -5.75597048e-01 -4.75975543e-01
-4.19825137e-01 -5.93190968e-01 -5.47189601e-02 -2.60088682e-01
4.18652266e-01 3.16878736e-01 6.01973720e-02 3.93130690e-01
1.12879187e-01 4.36424203e-02 6.77302420e-01 -3.24285030e-01
1.79933578e-01 9.29687917e-01 -1.56860217e-01 6.08618677e-01
8.03465962e-01 -5.44052541e-01 -2.29310393e-01 2.77223289e-01
4.76394743e-01 1.05672315e-01 5.79710901e-01 -1.24309766e+00
-7.83942416e-02 4.47234400e-02 -3.51189785e-02 -4.72492903e-01
2.74046630e-01 -1.27568531e+00 2.28429079e-01 8.93794000e-01
6.80827722e-02 1.51743248e-01 -1.21961378e-01 8.91491830e-01
4.22960222e-02 -5.81220448e-01 8.95394683e-01 -7.39282593e-02
-2.02791229e-01 -3.09506595e-01 -6.68966651e-01 -4.70155984e-01
1.03693032e+00 -7.86146224e-01 -3.26956689e-01 -3.18353385e-01
-6.53025687e-01 -7.93561995e-01 5.98111570e-01 3.87386560e-01
2.56221443e-01 -1.15858829e+00 -1.65068373e-01 3.12160671e-01
1.85279131e-01 -1.07733834e+00 1.05771430e-01 6.96027160e-01
-6.66552544e-01 6.07416511e-01 -4.56358135e-01 -4.74793553e-01
-1.35800338e+00 1.00421095e+00 3.14585656e-01 -2.93807417e-01
5.71286194e-02 -5.64841144e-02 -6.32485986e-01 -9.79558080e-02
1.18563250e-01 -2.46326208e-01 -2.62174606e-01 2.74626911e-01
3.03614348e-01 7.82851994e-01 6.38553441e-01 -1.02626395e+00
-1.38606384e-01 8.27999294e-01 2.78885067e-01 4.24696580e-02
7.08131850e-01 -1.23903491e-01 -1.08421594e-01 3.78950655e-01
1.33442187e+00 2.55437642e-01 -4.11653966e-01 -8.95569250e-02
5.71767867e-01 -7.57305086e-01 -6.01942837e-01 -6.14236832e-01
-6.03797138e-01 9.63810861e-01 9.79130447e-01 7.51909196e-01
5.77806115e-01 -4.66111720e-01 7.07513213e-01 5.62287152e-01
8.72817934e-01 -1.15925264e+00 2.61334717e-01 1.02996811e-01
5.89340270e-01 -9.23755527e-01 9.05689318e-04 -8.90280724e-01
-3.05676222e-01 1.27938235e+00 -1.96004763e-01 1.29201204e-01
7.88891256e-01 8.01389292e-03 -2.88281232e-01 2.47890040e-01
2.18746647e-01 2.98552155e-01 1.29070312e-01 9.25775766e-01
-8.77311919e-03 4.57095057e-02 -6.05779529e-01 4.47142005e-01
2.48234365e-02 7.22134262e-02 6.62810385e-01 9.82132316e-01
-2.17482984e-01 -1.44206202e+00 -9.10016596e-01 1.98833764e-01
-3.91084939e-01 1.93194792e-01 -2.57480264e-01 9.77525234e-01
5.97049594e-01 1.14162588e+00 -3.93060237e-01 -6.15917385e-01
3.00381392e-01 2.83840775e-01 2.84311533e-01 -4.50407900e-02
-5.60679078e-01 -4.92359817e-01 -2.05518696e-02 -1.19519129e-01
-4.23981279e-01 -6.38977051e-01 -6.15892410e-01 -3.92315716e-01
-8.62251043e-01 2.04958931e-01 9.89731610e-01 1.11957991e+00
-4.99252640e-02 -1.69178560e-01 9.24380362e-01 -6.18198156e-01
-8.89132977e-01 -6.04679406e-01 -1.00547445e+00 5.37960947e-01
2.15098590e-01 -6.77902281e-01 -7.73674369e-01 4.50269878e-02]
|
[12.421416282653809, 1.0402101278305054]
|
8e57472a-d4af-4bcc-b3ac-e71ea6471ec5
|
dce-offline-reinforcement-learning-with
|
2209.13132
| null |
https://arxiv.org/abs/2209.13132v1
|
https://arxiv.org/pdf/2209.13132v1.pdf
|
DCE: Offline Reinforcement Learning With Double Conservative Estimates
|
Offline Reinforcement Learning has attracted much interest in solving the application challenge for traditional reinforcement learning. Offline reinforcement learning uses previously-collected datasets to train agents without any interaction. For addressing the overestimation of OOD (out-of-distribution) actions, conservative estimates give a low value for all inputs. Previous conservative estimation methods are usually difficult to avoid the impact of OOD actions on Q-value estimates. In addition, these algorithms usually need to lose some computational efficiency to achieve the purpose of conservative estimation. In this paper, we propose a simple conservative estimation method, double conservative estimates (DCE), which use two conservative estimation method to constraint policy. Our algorithm introduces V-function to avoid the error of in-distribution action while implicit achieving conservative estimation. In addition, our algorithm uses a controllable penalty term changing the degree of conservatism in training. We theoretically show how this method influences the estimation of OOD actions and in-distribution actions. Our experiment separately shows that two conservative estimation methods impact the estimation of all state-action. DCE demonstrates the state-of-the-art performance on D4RL.
|
['Chun Yuan', 'Kai Xing Huang', 'Chen Zhao']
|
2022-09-27
| null | null | null | null |
['d4rl']
|
['robots']
|
[-3.43941778e-01 3.82705152e-01 -5.94210267e-01 -1.94730729e-01
-7.03517258e-01 -4.99331862e-01 4.16068166e-01 -3.27843241e-02
-7.74131954e-01 1.27595699e+00 -1.97537363e-01 -4.33757246e-01
1.62267108e-02 -5.19468307e-01 -9.39526498e-01 -5.33972859e-01
-3.68073672e-01 2.42945999e-01 3.69451225e-01 -2.70255744e-01
4.15886253e-01 7.50940219e-02 -1.29920030e+00 -6.63513765e-02
1.04939973e+00 1.00036478e+00 4.09406453e-01 5.30736923e-01
1.40808319e-04 1.04504669e+00 -1.18775260e+00 -1.26429321e-02
6.89476907e-01 -5.47908962e-01 -3.35740417e-01 -1.30740076e-01
-4.69154585e-03 -1.10768020e+00 -1.68088809e-01 1.08502817e+00
6.81930542e-01 1.97703958e-01 4.58796293e-01 -1.54086339e+00
-4.51804250e-01 9.07840371e-01 -9.29590166e-01 1.03144154e-01
7.28252605e-02 6.15354717e-01 6.08314157e-01 -2.11956397e-01
5.11848450e-01 1.41456819e+00 3.52401465e-01 7.27433085e-01
-8.71824801e-01 -8.62749696e-01 5.46147704e-01 3.05259023e-02
-1.02218854e+00 -3.40941697e-02 4.89834130e-01 -2.07346186e-01
9.14872944e-01 -6.65070266e-02 6.27889991e-01 8.65326822e-01
2.18481794e-01 9.64028478e-01 1.44333422e+00 -2.33793736e-01
5.65223932e-01 2.21745685e-01 -1.85866699e-01 6.36119187e-01
2.62148470e-01 6.64926529e-01 -3.28059942e-01 -4.41370532e-02
9.96394753e-01 -2.19346777e-01 -1.90026648e-02 -3.43079954e-01
-7.63189137e-01 1.00719786e+00 1.99430585e-01 -3.33279490e-01
-2.70135701e-01 6.58651292e-01 4.70377535e-01 5.68371296e-01
1.00068361e-01 6.30462229e-01 -6.49990499e-01 -6.90691650e-01
-4.64266270e-01 6.50751591e-01 6.38501346e-01 1.17450273e+00
4.93380189e-01 5.33044100e-01 -3.96170497e-01 8.28624666e-01
1.23320818e-01 4.15487230e-01 4.47983444e-01 -1.34565020e+00
7.51738667e-01 3.79469454e-01 8.16640198e-01 -4.03446764e-01
-3.12687695e-01 -4.69612271e-01 -3.28701735e-01 7.85276890e-01
7.34343112e-01 -7.42052317e-01 -8.42145383e-01 1.87633216e+00
3.99300426e-01 -2.89053787e-02 1.66420147e-01 1.15949535e+00
6.35582432e-02 5.64185023e-01 8.98409933e-02 -5.87516069e-01
8.42047989e-01 -1.12618220e+00 -9.97754812e-01 -6.56112209e-02
6.98731005e-01 -4.48129237e-01 1.29239571e+00 6.41041040e-01
-1.20924485e+00 -4.28067774e-01 -1.19141400e+00 5.85130811e-01
-7.34627768e-02 2.12763622e-01 7.32338965e-01 6.03784859e-01
-5.49828768e-01 9.20998633e-01 -8.41535985e-01 4.29053344e-02
3.94466102e-01 5.70372403e-01 2.23489910e-01 5.35345018e-01
-1.38402522e+00 1.17512286e+00 4.35444921e-01 -1.60569459e-01
-1.30113614e+00 -8.27541053e-01 -6.15444541e-01 2.26272903e-02
9.99126852e-01 5.62419333e-02 1.84119654e+00 -1.12484396e+00
-2.20293522e+00 -2.35664435e-02 5.34753501e-01 -8.27134490e-01
9.41763937e-01 -3.96327704e-01 8.24896321e-02 -2.08543658e-01
-1.16411380e-01 7.30630577e-01 9.04374599e-01 -1.14951646e+00
-8.89400303e-01 4.65395562e-02 5.56811452e-01 4.95379835e-01
-2.62679070e-01 -3.49510968e-01 -1.01626404e-01 -5.98262131e-01
-4.73150194e-01 -9.34030831e-01 -4.76791918e-01 -1.40539650e-02
1.40993679e-02 -3.70868176e-01 8.59400570e-01 -2.65079588e-01
1.49136925e+00 -1.92918289e+00 -3.15172434e-01 2.15934888e-02
-1.51325781e-02 1.01509914e-01 -8.21216851e-02 3.91816139e-01
3.55327785e-01 8.48724246e-02 -5.35698282e-03 1.75701365e-01
3.65517825e-01 4.51009512e-01 -4.52280760e-01 4.04946387e-01
1.62692994e-01 3.77624214e-01 -1.07748127e+00 -3.80318016e-01
2.33920425e-01 -1.63315311e-01 -1.03435278e+00 5.39096892e-01
-5.55725992e-01 1.65422857e-01 -6.96484566e-01 5.79121709e-01
7.50025928e-01 3.42274636e-01 4.21390772e-01 9.09810606e-03
-4.83588099e-01 1.10738240e-01 -1.34582627e+00 1.24413931e+00
-4.98098999e-01 1.81361780e-01 7.60425106e-02 -1.06348407e+00
7.07602262e-01 9.39716771e-02 5.32430887e-01 -8.95511270e-01
3.12262893e-01 3.33046049e-01 4.27802265e-01 -3.74931335e-01
3.86173248e-01 2.74717789e-02 -3.55952792e-02 2.96378255e-01
-2.31870309e-01 -3.94057244e-01 4.12769020e-01 -2.08887011e-02
9.26112652e-01 5.95377803e-01 4.20026243e-01 -4.31639403e-01
4.50251512e-02 -7.07319826e-02 8.52643311e-01 8.83182347e-01
-6.84359848e-01 -1.99175209e-01 1.05682898e+00 -2.13874534e-01
-9.43178236e-01 -7.80707419e-01 1.02013476e-01 9.93119478e-01
2.67237782e-01 -7.65856951e-02 -8.48972559e-01 -1.09803784e+00
5.41106403e-01 9.60437715e-01 -6.16744041e-01 -2.57193804e-01
-5.46185851e-01 -5.20826042e-01 5.04110754e-01 6.05250657e-01
5.59662282e-01 -9.18207884e-01 -1.17266667e+00 2.84118563e-01
4.49991554e-01 -7.81155229e-01 -5.69188535e-01 5.14709592e-01
-7.06574142e-01 -9.47902322e-01 -8.16478491e-01 -3.79934520e-01
6.98768258e-01 -1.26794860e-01 7.24675357e-01 -2.05532294e-02
2.26668894e-01 1.46191776e-01 -3.96467865e-01 -8.26292872e-01
-4.00003761e-01 -1.17663845e-01 3.98091078e-01 -5.82741439e-01
2.31453311e-02 -1.60294458e-01 -5.83684862e-01 5.58122873e-01
-4.64532167e-01 -1.62470281e-01 5.44505298e-01 1.03579986e+00
3.06542903e-01 -7.80574307e-02 9.60925102e-01 -7.23251224e-01
9.81979191e-01 -3.53441536e-01 -1.22863603e+00 9.11080539e-02
-8.20239723e-01 5.20843387e-01 1.06782877e+00 -1.09238219e+00
-9.79000270e-01 -9.85698327e-02 1.44389048e-01 -7.44821668e-01
3.62289011e-01 1.85325071e-01 2.64341533e-01 2.30645481e-02
5.03342748e-01 -1.28246963e-01 2.59282857e-01 -1.41005278e-01
1.01894595e-01 6.53327882e-01 -1.01287246e-01 -1.01392269e+00
4.98946875e-01 -4.56420928e-02 -6.52068257e-02 -3.71478558e-01
-7.68411219e-01 1.22363828e-01 1.65509686e-01 -4.33840185e-01
4.04136509e-01 -7.66461015e-01 -1.26271582e+00 2.01489076e-01
-7.56916106e-01 -1.07118464e+00 -4.33608532e-01 6.76044881e-01
-9.03629065e-01 1.51983142e-01 -4.52394515e-01 -1.24317324e+00
9.12296213e-03 -1.42548311e+00 5.07794261e-01 4.00005370e-01
4.54680845e-02 -5.11791348e-01 3.61724161e-02 -6.92250431e-02
3.68324995e-01 1.47208452e-01 4.86229211e-01 -2.27131099e-01
-3.30946445e-01 3.83580297e-01 -6.65729400e-03 3.63018602e-01
6.02751374e-02 -8.52240771e-02 -5.83661616e-01 -5.89040816e-01
-1.41104281e-01 -7.84605622e-01 4.25694525e-01 4.99292910e-01
1.29852641e+00 -6.34602427e-01 9.65497792e-02 2.01268211e-01
1.42518556e+00 8.06174338e-01 4.56981480e-01 6.51048362e-01
1.98256224e-01 3.07689428e-01 1.45391715e+00 1.00941741e+00
3.40410359e-02 7.17713773e-01 6.32445574e-01 2.86420971e-01
6.50457889e-02 -3.42142195e-01 7.34477758e-01 3.29804003e-01
-9.15142968e-02 -3.00463319e-01 -6.39448106e-01 4.36107695e-01
-2.00824594e+00 -7.95417905e-01 3.46585035e-01 2.47790956e+00
1.25671244e+00 5.32150626e-01 3.02839756e-01 -2.60493994e-01
5.49570918e-01 -8.05098265e-02 -1.10183918e+00 -9.25065041e-01
4.53747302e-01 -1.05856940e-01 1.12401056e+00 6.91737413e-01
-7.32185721e-01 1.06705391e+00 6.66966105e+00 1.11206222e+00
-9.54234123e-01 -3.90218459e-02 5.24214506e-01 -2.50255674e-01
-1.19432889e-01 3.56859304e-02 -1.00953424e+00 7.25111067e-01
7.56584048e-01 -4.97045182e-02 6.13077581e-01 1.21957207e+00
6.76461875e-01 -5.89803755e-01 -1.01081276e+00 7.05243289e-01
-4.54895437e-01 -1.03269160e+00 -2.05445185e-01 -1.24604683e-02
8.89090478e-01 -3.52624357e-01 -5.41866012e-02 1.02221632e+00
7.18013942e-01 -6.69700742e-01 9.93343949e-01 2.67336935e-01
7.29087889e-01 -1.11701965e+00 6.78456664e-01 5.14543056e-01
-8.29866409e-01 -4.92160439e-01 -5.85293233e-01 -5.00327706e-01
-2.65028290e-02 2.72376716e-01 -8.08700562e-01 -3.60536687e-02
3.69078726e-01 2.09576592e-01 -5.17702736e-02 6.65331066e-01
-2.73159742e-01 5.96054375e-01 -3.58323991e-01 -5.63293159e-01
5.91022432e-01 -1.89332306e-01 3.51561278e-01 7.16315866e-01
1.01089561e-02 9.34262797e-02 5.61596990e-01 8.03238332e-01
8.15044343e-02 -2.67040849e-01 -3.02363843e-01 -1.26737252e-01
6.24566317e-01 7.25791097e-01 -3.43918175e-01 -2.94921070e-01
-1.95152000e-01 4.45552915e-01 5.89265645e-01 3.13835293e-01
-1.42909503e+00 -4.57528919e-01 6.56056523e-01 -9.17989165e-02
2.69139528e-01 -3.26836228e-01 -4.18719240e-02 -7.30752647e-01
-2.23855168e-01 -1.14774156e+00 1.43467531e-01 -3.77191931e-01
-9.94378448e-01 1.47289867e-02 2.54709303e-01 -1.49359620e+00
-4.43185866e-01 -4.72951829e-01 -4.62693006e-01 4.16743129e-01
-1.44675660e+00 -3.29174846e-01 9.32891965e-02 2.05521703e-01
9.91021633e-01 -2.11128026e-01 2.24051043e-01 2.59912282e-01
-7.38123953e-01 8.48162651e-01 2.94684321e-01 -2.35089183e-01
9.54336464e-01 -1.31197369e+00 -1.77787617e-01 4.91350025e-01
-7.02691495e-01 3.07872921e-01 1.02559090e+00 -6.80995345e-01
-1.33650410e+00 -7.32298315e-01 -1.38827533e-01 -1.97755545e-02
6.63525462e-01 -4.41805385e-02 -3.73266965e-01 4.68763024e-01
1.38646007e-01 -8.22406560e-02 -1.68446582e-02 -2.60945290e-01
7.79413730e-02 -2.78970450e-01 -1.25277793e+00 9.45194602e-01
7.91126907e-01 6.36589304e-02 -2.93821633e-01 1.10852741e-01
8.75178397e-01 -7.33952165e-01 -8.35350692e-01 3.24576855e-01
6.42624855e-01 -7.88487256e-01 5.78227043e-01 -6.48906410e-01
3.52378845e-01 -2.36819491e-01 4.11581285e-02 -1.52718627e+00
1.40162557e-01 -8.79660368e-01 -4.90213305e-01 8.40588808e-01
3.60502809e-01 -5.99094450e-01 8.49277020e-01 5.82079053e-01
-2.26895496e-01 -1.32498145e+00 -6.87341273e-01 -1.35240114e+00
1.36535421e-01 -1.38493523e-01 3.75686109e-01 7.94778585e-01
2.58315593e-01 -1.02435611e-01 -6.50813341e-01 -6.18825518e-02
6.78738356e-01 -1.10813498e-01 9.01654005e-01 -3.46414596e-01
-7.06918657e-01 -2.74788886e-01 2.09984660e-01 -1.36608875e+00
-2.25283690e-02 -1.05830878e-01 2.78656721e-01 -1.15334809e+00
-1.40845388e-01 -5.96265137e-01 -2.68571973e-01 6.03600621e-01
-6.29399568e-02 -3.36425036e-01 3.52826089e-01 -6.08958751e-02
-6.21264577e-01 9.30306613e-01 1.68032491e+00 -6.46987110e-02
-5.75967431e-01 -4.29727323e-02 -3.55138540e-01 7.78850317e-01
1.20547795e+00 -6.22705996e-01 -8.01779032e-01 -2.71709830e-01
9.86766294e-02 2.76626021e-01 -1.70722511e-02 -7.63735592e-01
-1.42024189e-01 -8.56813371e-01 9.48712528e-02 -4.30401862e-01
6.07047491e-02 -8.55901122e-01 -3.63372028e-01 9.89141166e-01
-5.09128928e-01 -8.14408716e-03 4.79124367e-01 5.53986907e-01
6.55518100e-02 -4.63847816e-01 9.90679741e-01 -1.82935610e-01
-6.78839624e-01 3.67980525e-02 -5.39905190e-01 3.04973900e-01
1.26251793e+00 -1.54754877e-01 -3.40912133e-01 -5.33829927e-01
-2.50208795e-01 7.29154170e-01 6.25133440e-02 3.27311486e-01
4.21267927e-01 -1.11287451e+00 -3.73835206e-01 -1.55718923e-01
-3.49733233e-01 -3.26999635e-01 5.35311587e-02 7.36472726e-01
-4.73504096e-01 2.02893540e-01 -3.97392631e-01 -2.03789189e-01
-9.11206722e-01 5.50231814e-01 5.64593613e-01 -6.04005933e-01
-2.09950984e-01 5.49608886e-01 -3.06649171e-02 -3.94387126e-01
5.60148180e-01 -6.30945265e-01 -1.84612989e-03 -4.76853959e-02
1.64167002e-01 5.71126878e-01 -4.14229751e-01 4.22239989e-01
-2.16347620e-01 1.44141167e-01 -2.70374924e-01 -2.69382238e-01
1.15489101e+00 -8.53430387e-03 6.26438498e-01 2.82752514e-01
7.95267642e-01 -7.46413693e-02 -2.03640485e+00 4.18998510e-01
-2.09655777e-01 -5.84261715e-01 1.76708385e-01 -1.09373808e+00
-8.97592485e-01 4.98180598e-01 8.52352500e-01 1.16098255e-01
8.89139116e-01 -6.08244359e-01 6.71205640e-01 5.88789761e-01
6.86676979e-01 -1.86318457e+00 2.47362122e-01 5.19919217e-01
8.55919898e-01 -1.32545614e+00 4.14581209e-01 -5.60175478e-02
-1.07493865e+00 9.27047610e-01 1.36761177e+00 -4.20462221e-01
2.71472186e-01 4.48805302e-01 -3.85836065e-02 2.41053537e-01
-1.02051580e+00 -1.51901007e-01 -3.60420018e-01 4.72816527e-01
9.52835083e-02 2.10092552e-02 -9.19578075e-01 5.06240845e-01
5.47558703e-02 6.02273680e-02 8.33031058e-01 1.16089916e+00
-5.50356030e-01 -1.13098288e+00 -2.24668831e-01 3.58028561e-01
-4.23715919e-01 2.29066342e-01 1.51409600e-02 1.21187890e+00
3.55505869e-02 7.11018860e-01 6.69113100e-02 -2.20703036e-01
3.73932689e-01 -4.14825678e-01 5.80284417e-01 -2.15913951e-01
-6.31919920e-01 8.26095119e-02 2.51343250e-01 -7.98910916e-01
-2.74432898e-01 -3.01307470e-01 -1.62175155e+00 -2.09530920e-01
-5.57007372e-01 2.47434363e-01 6.22508287e-01 7.82055080e-01
1.73474729e-01 6.93266749e-01 7.95805275e-01 -5.74634790e-01
-1.56473422e+00 -8.43655765e-01 -6.76914692e-01 1.46105841e-01
3.04338902e-01 -1.22803915e+00 -5.22378147e-01 -4.45932627e-01]
|
[4.108262538909912, 2.253573417663574]
|
3ea6dd16-2bf7-4e02-aafd-1153680f7912
|
facial-age-estimation-using-convolutional
|
2105.06746
| null |
https://arxiv.org/abs/2105.06746v1
|
https://arxiv.org/pdf/2105.06746v1.pdf
|
Facial Age Estimation using Convolutional Neural Networks
|
This paper is a part of a student project in Machine Learning at the Norwegian University of Science and Technology. In this paper, a deep convolutional neural network with five convolutional layers and three fully-connected layers is presented to estimate the ages of individuals based on images. The model is in its entirety trained from scratch, where a combination of three different datasets is used as training data. These datasets are the APPA dataset, UTK dataset, and the IMDB dataset. The images were preprocessed using a proprietary face-recognition software. Our model is evaluated on both a held-out test set, and on the Adience benchmark. On the test set, our model achieves a categorical accuracy of 52%. On the Adience benchmark, our model proves inferior compared with other leading models, with an exact accuray of 30%, and an one-off accuracy of 46%. Furthermore, a script was created, allowing users to estimate their age directly using their web camera. The script, alongside all other code, is located in our GitHub repository: AgeNet.
|
['Erling Stray Bugge', 'Christian Bakke Vennerød', 'Adrian Kjærran']
|
2021-05-14
| null | null | null | null |
['age-estimation', 'age-estimation']
|
['computer-vision', 'miscellaneous']
|
[-2.13894114e-01 3.58585119e-01 -1.58913620e-02 -7.71901071e-01
-3.81333888e-01 -3.06032449e-01 6.93295419e-01 3.26253921e-02
-7.75793433e-01 4.77947623e-01 -9.80354622e-02 -1.15341358e-01
4.29912034e-04 -8.92301083e-01 -4.64866012e-01 -4.46511775e-01
-8.99595469e-02 6.03864491e-01 -1.72721714e-01 1.39973760e-01
2.20836714e-01 4.52313185e-01 -1.83009088e+00 7.79546425e-02
6.43245757e-01 1.26605535e+00 -5.31997800e-01 8.41235816e-01
2.01141983e-01 5.89395165e-01 -5.79312325e-01 -1.11813271e+00
3.43562454e-01 -2.13981997e-02 -7.91563570e-01 -2.62902558e-01
1.01904798e+00 -7.10657179e-01 -4.17025268e-01 6.49358869e-01
7.60096490e-01 -4.68713820e-01 5.64884067e-01 -1.55013335e+00
-5.59174895e-01 5.82211375e-01 -1.67081788e-01 -1.86926216e-01
3.65164906e-01 3.97901423e-02 5.66343307e-01 -7.58288324e-01
2.80878991e-01 1.28287244e+00 1.17349184e+00 7.62355983e-01
-9.63172019e-01 -7.85162270e-01 -5.04186809e-01 1.79517046e-01
-1.62646687e+00 -6.80074394e-01 2.16141716e-01 -4.86065924e-01
7.83493817e-01 2.06255466e-01 5.64914703e-01 1.28512990e+00
-1.67254955e-01 4.61142838e-01 1.21123242e+00 -4.90831345e-01
2.62291878e-01 3.40741605e-01 2.28131056e-01 7.87073195e-01
3.51984113e-01 2.01043963e-01 -5.24093449e-01 -1.03820533e-01
7.69074023e-01 -5.98063357e-02 1.78562984e-01 -7.27343634e-02
-5.31947076e-01 7.45350897e-01 2.12436751e-01 7.10077062e-02
-2.57123441e-01 8.52415934e-02 1.07699305e-01 2.65469790e-01
6.06203556e-01 1.38153955e-01 -6.16637886e-01 -2.44302124e-01
-9.14716423e-01 3.52675706e-01 1.20575094e+00 5.27678132e-01
5.92892826e-01 -2.17200384e-01 -7.71797299e-02 9.37660098e-01
3.21341574e-01 3.85395408e-01 5.78168154e-01 -1.18168974e+00
3.41567770e-02 9.44403172e-01 -1.20659448e-01 -6.83218658e-01
-3.78908008e-01 -6.62320256e-02 -5.83015025e-01 3.86019468e-01
9.27232385e-01 -3.83331239e-01 -1.12992263e+00 1.68314052e+00
1.54165670e-01 1.92649111e-01 -1.23101324e-01 5.16665041e-01
1.30137873e+00 1.64726019e-01 1.89147126e-02 1.84388176e-01
1.22193742e+00 -5.23170769e-01 -2.85763383e-01 -1.64168790e-01
4.77755040e-01 -4.58211303e-01 5.75609028e-01 6.38132989e-01
-1.26975071e+00 -4.70448673e-01 -9.11598980e-01 -1.07031994e-01
-8.12017679e-01 4.08737600e-01 7.56524920e-01 1.28165758e+00
-1.55668139e+00 8.56090724e-01 -7.09522307e-01 -8.24763238e-01
6.21037722e-01 6.74807131e-01 -6.91436172e-01 -1.19329184e-01
-9.34530497e-01 7.33615398e-01 1.45658091e-01 -5.97712547e-02
-7.76225686e-01 -8.54987383e-01 -9.26761568e-01 -1.55224979e-01
-4.06936631e-02 -4.82605636e-01 1.33755839e+00 -1.03415251e+00
-1.26516640e+00 1.46105981e+00 2.47560263e-01 -5.07497251e-01
9.49810624e-01 -1.91394210e-01 -4.80163664e-01 1.09130690e-06
2.31507956e-03 8.60938311e-01 6.77621067e-01 -7.78483629e-01
-3.92533749e-01 -7.57141650e-01 1.01185150e-01 -3.32242340e-01
-6.84957564e-01 1.53398067e-01 -6.53166294e-01 -3.63191247e-01
-5.43728232e-01 -9.11667347e-01 1.58530697e-01 1.76061332e-01
-1.23353556e-01 -3.92434359e-01 4.50524509e-01 -1.08091915e+00
1.17895925e+00 -1.89460778e+00 -5.95388860e-02 1.09099559e-01
2.36873418e-01 3.95465791e-01 7.82280322e-03 3.43433112e-01
-4.36257809e-01 1.00916214e-01 -2.31983230e-01 -5.99603713e-01
-2.92552523e-02 5.74442968e-02 4.28097337e-01 3.76467198e-01
2.97690570e-01 4.02644008e-01 -5.49170434e-01 -3.78241390e-01
1.40537307e-01 6.53938770e-01 -5.28001666e-01 2.41556615e-01
3.12122077e-01 5.63709699e-02 7.62017369e-02 7.98159957e-01
8.49280596e-01 1.31969064e-01 6.09320737e-02 1.41104728e-01
-1.07403800e-01 -2.31963396e-01 -1.10069013e+00 1.53593898e+00
-4.01189417e-01 5.30922353e-01 8.84435326e-02 -7.46219039e-01
9.53920603e-01 1.92138672e-01 2.67120034e-01 -4.13116008e-01
3.95065576e-01 2.19430268e-01 -1.15451086e-02 -4.40911204e-01
2.92960405e-01 4.24276590e-01 7.59003237e-02 2.92209715e-01
4.13157314e-01 3.07505935e-01 4.49944854e-01 2.09650591e-01
1.20799088e+00 2.55062804e-03 -1.28878906e-01 -2.94352174e-01
5.74453652e-01 -4.77798760e-01 3.91731411e-01 3.65995228e-01
-1.95720077e-01 9.64133203e-01 8.50525022e-01 -9.07888353e-01
-1.09068561e+00 -1.01701581e+00 -4.77909982e-01 8.57902229e-01
-5.89839041e-01 -6.96008444e-01 -1.21318579e+00 -6.03550971e-01
2.17296571e-01 3.36970150e-01 -1.19182515e+00 -9.46791843e-02
-1.80935830e-01 -7.90984273e-01 8.03589642e-01 5.83517015e-01
6.99029028e-01 -1.15684533e+00 -4.14940923e-01 -2.74967104e-01
1.47418424e-01 -1.05963695e+00 -7.02678710e-02 -2.13710025e-01
-5.91948688e-01 -1.46563554e+00 -9.42336082e-01 -6.53343081e-01
7.11179554e-01 -5.72159410e-01 1.38034916e+00 5.53391278e-01
-6.01468265e-01 4.14726555e-01 -1.15130171e-01 -7.88688362e-01
-2.50770509e-01 2.79732406e-01 2.03889996e-01 1.64958179e-01
8.37848783e-01 -4.61811125e-01 -7.33193994e-01 1.95391744e-01
-6.57617748e-01 -1.81593254e-01 5.21732509e-01 4.76541489e-01
-3.49539034e-02 -4.36125696e-01 5.78170478e-01 -7.69822299e-01
2.67208606e-01 -4.56856191e-01 -6.59795284e-01 1.15602717e-01
-8.24163377e-01 -2.83073932e-01 3.09958220e-01 -2.58672476e-01
-7.13963985e-01 1.93915457e-01 -4.38354015e-01 -2.30483770e-01
-5.08283973e-01 1.93869248e-01 -1.28283769e-01 7.78567567e-02
6.55604541e-01 -1.45898893e-01 3.52242023e-01 -7.75071979e-01
-7.79216364e-02 1.18285084e+00 5.93284905e-01 -3.69845361e-01
4.57590282e-01 2.10277736e-01 -1.09323874e-01 -9.08177435e-01
-5.81691921e-01 -1.03067778e-01 -7.26610780e-01 -3.82329851e-01
6.52554750e-01 -1.11857963e+00 -6.70384645e-01 1.32853842e+00
-7.90185273e-01 -5.81066608e-01 8.89585987e-02 3.48301440e-01
-2.16680378e-01 5.16478717e-02 -7.15240538e-01 -7.59263039e-01
-5.14676511e-01 -8.04001272e-01 9.30673480e-01 5.49205720e-01
-2.65975296e-01 -8.66937160e-01 -5.22831380e-02 4.42655057e-01
3.41266572e-01 4.35596883e-01 3.98179322e-01 -7.97999561e-01
2.15101942e-01 -7.16374576e-01 -3.86518300e-01 5.29480040e-01
-1.66858822e-01 3.98807734e-01 -1.31933761e+00 -3.90235066e-01
-6.39899671e-01 -5.74027419e-01 8.87082696e-01 2.61922956e-01
1.42993939e+00 -1.34070307e-01 -2.14441270e-01 5.53318083e-01
1.30071545e+00 -2.52282262e-01 9.71237481e-01 2.52181858e-01
5.19683063e-01 8.53065491e-01 -9.02814642e-02 5.63268423e-01
6.57985806e-01 4.42331135e-01 4.10551459e-01 4.36710678e-02
-1.10735267e-01 -8.65323916e-02 2.14155763e-01 3.32244456e-01
-4.38696772e-01 1.61090091e-01 -1.26716161e+00 6.40310884e-01
-1.63085067e+00 -6.93656087e-01 -3.66214156e-01 2.47646642e+00
7.07838953e-01 2.41948254e-02 6.21084154e-01 3.00243676e-01
7.26285994e-01 -3.79780918e-01 -3.36565793e-01 -6.09052777e-01
2.09437162e-01 5.24701595e-01 4.57254022e-01 2.79137522e-01
-1.05512381e+00 5.73647499e-01 6.95227051e+00 4.26503748e-01
-1.11669230e+00 -7.91658908e-02 1.09166729e+00 -1.55268446e-01
4.94721144e-01 -3.69648784e-01 -8.04411054e-01 7.39568591e-01
1.55313861e+00 1.12485746e-02 5.45682788e-01 8.94289315e-01
-2.64587253e-01 -2.49358088e-01 -1.24618363e+00 1.07875848e+00
4.04352576e-01 -9.46095824e-01 -4.64927197e-01 3.66963625e-01
4.65853006e-01 -5.23417443e-02 2.26805583e-01 4.97717559e-01
2.86532223e-01 -1.40325618e+00 5.61013103e-01 6.76591098e-01
1.18599963e+00 -1.11291528e+00 9.96261060e-01 2.91149672e-02
-7.10032642e-01 -1.76547170e-01 -1.71786427e-01 -4.31669056e-01
-5.60232639e-01 4.98368353e-01 -5.68568051e-01 3.41242433e-01
1.35996866e+00 6.09856367e-01 -1.31223309e+00 1.10451639e+00
-1.44901633e-01 5.07123411e-01 -2.35074475e-01 2.20068544e-01
-2.17796609e-01 2.45862501e-03 -2.84230709e-01 1.07031274e+00
4.82368439e-01 -1.52903106e-02 -4.66375709e-01 5.36863089e-01
-3.65485489e-01 -1.38491824e-01 -4.11372453e-01 -8.77779499e-02
2.89749652e-01 1.77038682e+00 -4.82100755e-01 -2.72161335e-01
-4.23338205e-01 7.59397864e-01 4.16167408e-01 1.65561680e-02
-5.83085477e-01 -6.93056643e-01 4.67512161e-01 3.37528646e-01
1.18020020e-01 1.95362270e-01 -1.94096386e-01 -8.07286322e-01
-2.13196397e-01 -8.16352725e-01 4.19275850e-01 -7.27992475e-01
-1.39392364e+00 5.27093053e-01 1.66434795e-01 -5.76763570e-01
-4.92888480e-01 -1.02856505e+00 -6.89932942e-01 1.01020980e+00
-9.82055366e-01 -1.30735219e+00 -6.83367014e-01 4.77956623e-01
7.04183057e-02 -5.04197001e-01 1.00118577e+00 7.00207233e-01
-1.01188946e+00 9.63323414e-01 -1.32017061e-01 6.48026228e-01
8.07269633e-01 -1.42134178e+00 4.17855561e-01 6.38007700e-01
-3.55334461e-01 5.34291208e-01 5.08720219e-01 -4.79351223e-01
-1.11811328e+00 -1.09908128e+00 1.11967695e+00 -7.62877882e-01
4.54171300e-01 -4.38984692e-01 -7.75805831e-01 7.38970280e-01
3.54803681e-01 -5.03906654e-03 7.87888885e-01 2.84099013e-01
-3.88048142e-01 -1.68398604e-01 -1.54246497e+00 2.37417579e-01
9.10890758e-01 -2.02736780e-01 -3.43307018e-01 1.51612252e-01
1.25639305e-01 -3.86016876e-01 -1.14840794e+00 1.82418406e-01
1.11506546e+00 -1.22793341e+00 9.34314370e-01 -4.37366486e-01
8.67947817e-01 2.62767404e-01 9.99862850e-02 -1.15778363e+00
3.54897492e-02 -2.43224218e-01 -3.27773601e-01 1.66854060e+00
3.03739637e-01 -6.84312344e-01 1.01190007e+00 7.94776380e-01
2.76409864e-01 -7.57597208e-01 -8.27104330e-01 -6.81211174e-01
2.57846743e-01 -3.71538550e-01 7.95158386e-01 7.34108984e-01
-3.97417516e-01 2.24049315e-01 -1.40141830e-01 7.20614521e-03
7.69460976e-01 -5.85211396e-01 9.31727886e-01 -1.74625576e+00
7.67869577e-02 -4.64754343e-01 -5.71516395e-01 1.33865595e-01
1.72938541e-01 -6.34796023e-01 -3.94817561e-01 -1.35983181e+00
2.17971802e-01 -2.20815450e-01 -3.89682911e-02 7.48272538e-01
1.22974344e-01 7.37639248e-01 -4.26289663e-02 -3.72765884e-02
-1.37637809e-01 1.00857608e-01 4.40717191e-01 1.18478471e-02
1.07492998e-01 -4.01106924e-02 -6.60450518e-01 8.05376112e-01
9.50770140e-01 -2.47901708e-01 1.00867324e-01 -3.13257933e-01
1.70472234e-01 -2.36204028e-01 6.01152182e-01 -1.41628349e+00
7.28466082e-03 3.62700522e-01 1.06601477e+00 -3.71476650e-01
5.04161298e-01 -5.63172996e-01 2.43226558e-01 6.51439607e-01
-1.33330345e-01 1.50980324e-01 2.45833829e-01 -6.53505847e-02
1.90155372e-01 -2.64423132e-01 7.09339499e-01 -9.22744647e-02
-5.13927639e-01 4.84108031e-01 -1.44769643e-02 -1.74696922e-01
9.45000648e-01 -1.79917037e-01 -3.90181184e-01 -3.84743214e-01
-6.27354443e-01 2.13397175e-01 6.81512237e-01 5.22493601e-01
3.95007044e-01 -1.35934198e+00 -9.57897186e-01 4.90710378e-01
2.75277346e-01 -3.72908503e-01 2.22873658e-01 6.24526799e-01
-7.74199069e-01 1.94431663e-01 -6.53989911e-01 -4.05386150e-01
-1.41676188e+00 2.63912618e-01 4.45351660e-01 3.64600457e-02
-8.00479054e-02 8.56723905e-01 -3.51781189e-01 -8.65862966e-01
4.15452510e-01 9.01847333e-02 -4.13455904e-01 1.41910523e-01
1.03562558e+00 6.35414124e-01 2.07970321e-01 -6.68451786e-01
-4.42011088e-01 3.20434988e-01 -1.69541277e-02 -8.97117183e-02
1.58588314e+00 2.96069443e-01 -2.30403826e-01 1.81464210e-01
1.19299459e+00 -2.91933239e-01 -1.13319421e+00 1.41028181e-01
-4.81683761e-02 -4.53238338e-01 6.52711987e-02 -1.08854079e+00
-1.31341779e+00 7.67188311e-01 1.12079656e+00 3.19527954e-01
9.81334984e-01 -1.06420971e-01 3.05575758e-01 7.40890577e-02
6.93616271e-02 -1.13188195e+00 -7.33680353e-02 3.77320796e-01
9.23115194e-01 -1.45965612e+00 1.42160922e-01 -9.39846486e-02
-3.23469937e-01 1.11962163e+00 9.25290763e-01 -8.05084407e-02
5.42052090e-01 1.84204385e-01 9.89762247e-02 -1.38831019e-01
-5.44260681e-01 -1.09835714e-01 1.21309683e-01 8.97244930e-01
5.57765365e-01 -6.31187810e-03 -3.36717099e-01 8.84563148e-01
-6.33529305e-01 4.93895859e-01 6.56831145e-01 6.38666451e-01
-2.86008790e-02 -1.26159191e+00 -4.76405233e-01 6.76256835e-01
-8.99518490e-01 1.31123930e-01 -7.53713548e-01 8.87324572e-01
3.89298171e-01 9.29672301e-01 4.27843004e-01 -4.47946370e-01
2.26544783e-01 1.01647414e-01 5.20974100e-01 -2.96922505e-01
-7.76607394e-01 -6.91869438e-01 3.29101264e-01 -4.75329489e-01
-2.45450348e-01 -7.03714132e-01 -7.30582833e-01 -8.33146036e-01
2.43088841e-01 9.96343512e-03 1.01135099e+00 5.68661332e-01
2.64795601e-01 2.77491003e-01 4.76365864e-01 -9.26421523e-01
-1.94885001e-01 -1.26035535e+00 -6.53253257e-01 2.36339375e-01
6.92972541e-03 -4.66548651e-01 -3.26993287e-01 -1.71447396e-01]
|
[13.48942756652832, 1.0402238368988037]
|
490c2d35-76de-4a5a-9d89-530a99900117
|
sketch-less-face-image-retrieval-a-new
|
2302.05576
| null |
https://arxiv.org/abs/2302.05576v1
|
https://arxiv.org/pdf/2302.05576v1.pdf
|
Sketch Less Face Image Retrieval: A New Challenge
|
In some specific scenarios, face sketch was used to identify a person. However, drawing a complete face sketch often needs skills and takes time, which hinder its widespread applicability in the practice. In this study, we proposed a new task named sketch less face image retrieval (SLFIR), in which the retrieval was carried out at each stroke and aim to retrieve the target face photo using a partial sketch with as few strokes as possible (see Fig.1). Firstly, we developed a method to generate the data of sketch with drawing process, and opened such dataset; Secondly, we proposed a two-stage method as the baseline for SLFIR that (1) A triplet network, was first adopt to learn the joint embedding space shared between the complete sketch and its target face photo; (2) Regarding the sketch drawing episode as a sequence, we designed a LSTM module to optimize the representation of the incomplete face sketch. Experiments indicate that the new framework can finish the retrieval using a partial or pool drawing sketch.
|
['Guoyin Wang', 'Shuyin Xia', 'Shiyu Fu', 'Liang Wang', 'Yutang Li', 'Dawei Dai']
|
2023-02-11
| null | null | null | null |
['face-image-retrieval']
|
['computer-vision']
|
[ 1.17051817e-01 -1.10975228e-01 5.84363379e-02 -4.04014468e-01
-4.52424705e-01 -4.19071883e-01 9.33765352e-01 -7.80772507e-01
-1.04669563e-01 3.81568521e-01 1.02527447e-01 2.37854391e-01
-8.19497108e-02 -6.66861117e-01 -3.94292235e-01 -6.26386523e-01
5.92995226e-01 5.24678469e-01 -3.20506752e-01 2.30631322e-01
5.56642830e-01 1.05282938e+00 -1.47722054e+00 4.93452340e-01
2.70590395e-01 9.60132241e-01 4.09723550e-01 4.04008448e-01
-6.06171966e-01 6.03178084e-01 -8.16819131e-01 -7.04353034e-01
3.15952152e-01 -3.92082036e-01 -5.50658047e-01 1.79140285e-01
7.84633934e-01 -1.06199384e+00 -7.31748700e-01 6.94723368e-01
8.00963879e-01 1.50799185e-01 8.35079551e-01 -1.59248948e+00
-1.10504818e+00 4.11235273e-01 -6.31547689e-01 -4.42108303e-01
3.79243672e-01 8.19252357e-02 5.13453901e-01 -1.42982590e+00
8.59447122e-01 1.56361794e+00 3.43360484e-01 1.09399104e+00
-6.96976185e-01 -1.20702326e+00 2.40509540e-01 -1.89353358e-02
-1.71440196e+00 -6.47453308e-01 1.15664124e+00 -1.65226549e-01
4.36347485e-01 2.62616407e-02 7.23142862e-01 1.15930402e+00
-2.82900125e-01 1.24780238e+00 7.37947404e-01 -2.70103902e-01
-1.63288832e-01 1.92477211e-01 -1.01043493e-01 7.21818209e-01
-1.82452440e-01 -4.54717018e-02 -4.38688755e-01 -2.28645101e-01
1.20987165e+00 4.76299971e-01 -6.97728917e-02 -2.05896959e-01
-1.07938790e+00 6.46569967e-01 4.77411896e-01 4.32815254e-01
-3.06873709e-01 3.01477551e-01 3.26318741e-01 4.71876293e-01
2.10239559e-01 1.06556981e-03 1.76366583e-01 3.92072856e-01
-1.27706325e+00 1.94581047e-01 6.73953772e-01 1.17579544e+00
6.37171149e-01 -1.24146171e-01 -5.09085476e-01 1.06804502e+00
4.15886223e-01 7.24325716e-01 2.80451417e-01 -7.81747878e-01
5.24265468e-01 5.16695380e-01 2.09034234e-01 -1.17518067e+00
2.65624672e-01 1.97136313e-01 -1.07457936e+00 -1.04130087e-02
5.72154112e-02 1.27182811e-01 -9.29747403e-01 1.66700470e+00
4.74343105e-04 3.58145773e-01 -4.57387082e-02 1.04935575e+00
1.32956922e+00 7.66802907e-01 2.51822229e-02 -1.69973880e-01
1.24758315e+00 -1.10761142e+00 -9.36650574e-01 6.57292008e-02
-3.85425128e-02 -1.11565220e+00 1.02655184e+00 3.53444278e-01
-1.11574101e+00 -9.57091153e-01 -9.67889428e-01 -3.50115001e-01
-4.10079211e-01 1.06891596e+00 4.66172457e-01 3.93823713e-01
-1.24549699e+00 5.55622399e-01 -1.03654191e-01 -5.17302036e-01
6.41867518e-01 3.44733804e-01 -7.33471572e-01 -3.90654951e-01
-9.95193958e-01 7.66252100e-01 1.90098673e-01 7.28402078e-01
-9.57144856e-01 -3.72089684e-01 -4.40795302e-01 3.06922346e-01
1.56622574e-01 -7.73338616e-01 8.75162005e-01 -1.17756701e+00
-1.80155945e+00 9.74446595e-01 -2.44321704e-01 3.40336442e-01
6.93793774e-01 -2.93081224e-01 -3.32720786e-01 2.98950732e-01
-1.64277583e-01 1.05458367e+00 1.53858554e+00 -1.50689936e+00
-9.67307389e-02 -4.48806316e-01 -1.14130840e-01 2.54748493e-01
-3.04857492e-01 1.43090561e-01 -9.14151967e-01 -6.06859326e-01
-6.34890050e-02 -7.76311994e-01 3.96665722e-01 4.47094858e-01
-1.69141263e-01 -7.07022190e-01 1.03707469e+00 -8.79567564e-01
1.31628692e+00 -2.09523797e+00 1.91616252e-01 3.46437484e-01
5.36450818e-02 6.29077613e-01 -6.98888719e-01 1.01365113e+00
-4.54726033e-02 1.04351021e-01 -1.27938092e-01 -5.69885075e-01
2.88376063e-02 1.88648328e-02 -7.41113842e-01 1.25827059e-01
4.17709082e-01 9.49597538e-01 -7.41134167e-01 -6.38885498e-01
2.16802791e-01 7.77975261e-01 -1.94350973e-01 6.67947710e-01
3.04173250e-02 5.93711808e-03 -6.62258804e-01 5.56490541e-01
1.09194291e+00 1.02971770e-01 1.17845014e-01 -4.74388272e-01
8.13863054e-02 -7.28868902e-01 -1.23107922e+00 2.03319025e+00
-5.23756742e-01 5.47446847e-01 5.32945357e-02 -3.95413727e-01
1.31206858e+00 3.61395001e-01 1.41526356e-01 -5.77492356e-01
7.32634123e-03 1.37216777e-01 -4.88438040e-01 -8.77744615e-01
2.74214029e-01 3.08086872e-02 4.10052478e-01 7.66708016e-01
-1.64971296e-02 -2.47692749e-01 -8.08397308e-02 2.53098041e-01
6.86094165e-01 5.47481894e-01 -1.56662866e-01 1.16358161e-01
8.38728189e-01 -5.24498940e-01 -7.06548095e-02 4.84981924e-01
2.18189165e-01 8.07796240e-01 3.61332983e-01 -7.10225701e-01
-9.68924642e-01 -9.26328599e-01 2.31170878e-01 6.78990781e-01
2.45284215e-01 -2.39214092e-01 -7.35186338e-01 -7.21434057e-01
1.14462495e-01 3.92404586e-01 -4.79525894e-01 -2.22433463e-01
-8.23233962e-01 -2.02994198e-02 6.29998446e-01 3.75586122e-01
7.94201732e-01 -1.68864322e+00 -1.82991475e-01 -8.23669806e-02
-5.03638871e-02 -6.14962459e-01 -1.10606110e+00 -7.38946080e-01
-5.76867819e-01 -1.12651539e+00 -1.49602377e+00 -1.11498737e+00
1.14356351e+00 4.96629685e-01 8.66406262e-01 7.08512723e-01
-4.00728941e-01 4.23652917e-01 -7.49569535e-02 -1.28969461e-01
9.55582634e-02 -5.21620475e-02 -2.10499331e-01 4.83088166e-01
2.71256655e-01 -3.09873581e-01 -8.65089059e-01 1.99545279e-01
-1.10302913e+00 4.58847322e-02 1.02481806e+00 1.04634523e+00
2.03528687e-01 -4.06611681e-01 7.85053909e-01 -6.34966373e-01
1.16374815e+00 -3.78917233e-04 -2.16796950e-01 7.99642026e-01
-2.16027394e-01 3.92143019e-02 6.10621274e-01 -5.81357896e-01
-1.29499745e+00 1.57529101e-01 -1.24688111e-01 -9.73219931e-01
1.08130164e-01 3.20505016e-02 -3.30181122e-01 -8.61773416e-02
-2.21426841e-02 4.96430486e-01 3.39417875e-01 -6.95201218e-01
6.07052684e-01 1.04171741e+00 2.38651320e-01 -7.79054403e-01
9.34868336e-01 2.00696006e-01 -1.80700958e-01 -9.22878325e-01
-2.63946086e-01 -1.60144955e-01 -8.44874859e-01 -3.69790882e-01
5.91841996e-01 -8.16818595e-01 -1.07936525e+00 5.32307088e-01
-1.67594624e+00 -1.65531173e-01 2.45974272e-01 1.25695735e-01
-3.25094134e-01 5.52971482e-01 -4.19579327e-01 -1.05798566e+00
-7.28822470e-01 -8.55962336e-01 1.50856388e+00 3.40377569e-01
7.52686262e-02 -4.84220922e-01 -4.68300842e-02 9.68675781e-03
4.28942651e-01 -1.45548731e-01 9.15888548e-01 -3.11133564e-01
-7.57847011e-01 -2.77018696e-01 -8.59183788e-01 3.12717915e-01
8.78532603e-02 4.39417303e-01 -1.07966268e+00 -5.17186582e-01
-3.16102415e-01 -4.81327534e-01 8.62177968e-01 -1.89394593e-01
1.45081723e+00 -3.64822298e-01 -4.58395123e-01 4.87652063e-01
1.35222268e+00 3.51165146e-01 7.72249103e-01 -4.12929058e-01
6.06096983e-01 5.86478949e-01 6.16141915e-01 1.03129506e-01
1.28161505e-01 6.50161386e-01 -1.64094552e-01 -2.08566934e-01
-5.01782298e-01 -8.72755170e-01 2.09615514e-01 9.71685171e-01
-1.29540160e-01 -2.28991553e-01 -2.31949598e-01 2.71034420e-01
-1.71129417e+00 -1.02967870e+00 5.19317985e-01 2.23406863e+00
4.53439951e-01 -4.56339210e-01 -1.29795820e-01 -2.02874750e-01
8.55339825e-01 2.92074621e-01 -5.11269391e-01 -3.28278452e-01
9.42384899e-02 2.98487663e-01 -2.39044189e-01 3.04938674e-01
-5.71885288e-01 1.26458514e+00 5.95077467e+00 1.08049178e+00
-1.27182436e+00 -2.33069852e-01 3.43751490e-01 -3.08375899e-02
-2.61398703e-01 -8.20173919e-02 -6.56756997e-01 4.81857210e-01
1.03247911e-01 2.11212728e-02 7.99893618e-01 5.72267890e-01
-1.53472468e-01 3.26589018e-01 -1.21603632e+00 1.68625033e+00
5.10687590e-01 -8.65489960e-01 9.02272642e-01 -3.01540762e-01
3.37081313e-01 -8.10934305e-01 2.42320340e-04 4.42905694e-01
-2.93886155e-01 -1.11497748e+00 4.47966009e-01 1.13763416e+00
1.25504494e+00 -6.88136220e-01 4.18911666e-01 2.77428091e-01
-1.28587747e+00 -4.64661345e-02 -7.34627962e-01 3.55370849e-01
-1.26613572e-01 2.96872649e-02 -5.20430982e-01 6.78310633e-01
2.16141745e-01 5.60119450e-01 -6.45342648e-01 1.00160742e+00
-2.39321321e-01 -1.18573189e-01 -1.62294909e-01 -1.15500130e-01
1.88299626e-01 -5.47813594e-01 7.97584355e-02 1.00361741e+00
5.70479274e-01 2.32673943e-01 2.34428123e-02 8.56632769e-01
-4.22433585e-01 2.94580817e-01 -8.89602959e-01 -2.77071476e-01
7.92139113e-01 1.59126163e+00 -5.66360414e-01 -4.55602229e-01
-2.33142361e-01 1.36053252e+00 4.27125305e-01 6.96936965e-01
-6.45937026e-01 -7.36857176e-01 6.27045855e-02 -2.01678932e-01
2.85452995e-02 1.80522129e-01 2.46642873e-01 -1.18674266e+00
2.39655852e-01 -7.19755113e-01 1.45717978e-01 -1.22253287e+00
-1.50227594e+00 7.34494090e-01 -1.43457711e-01 -9.35488224e-01
-9.75703746e-02 -4.35683608e-01 -9.42976415e-01 1.20112669e+00
-1.46164155e+00 -1.70693719e+00 -6.27556384e-01 6.10919118e-01
7.44662285e-01 -4.87060398e-01 7.75904655e-01 6.08350396e-01
-7.07430840e-01 7.55472660e-01 -2.09673718e-01 2.98980176e-01
1.10460150e+00 -6.98092878e-01 3.31716388e-01 2.28567541e-01
3.07138950e-01 1.01910746e+00 4.92400378e-02 -6.96776390e-01
-1.64989316e+00 -8.00590992e-01 9.50876713e-01 -1.60457164e-01
1.15412446e-02 -4.16312844e-01 -7.17094123e-01 6.02365971e-01
3.76585722e-01 -1.70472309e-01 1.50131434e-01 -3.60655099e-01
-2.66964048e-01 -3.46172094e-01 -1.33277559e+00 8.14826787e-01
1.21249437e+00 -8.85517180e-01 -5.90423524e-01 1.33326024e-01
2.95117110e-01 -4.94872257e-02 -6.14687324e-01 2.30188537e-02
1.28060341e+00 -4.92503136e-01 1.14723194e+00 -5.70376813e-01
3.70828032e-01 -3.36203605e-01 1.11633375e-01 -1.03344250e+00
-2.60187179e-01 -4.40080673e-01 -1.10615036e-02 1.57950008e+00
-1.48001283e-01 -4.41423684e-01 6.81269646e-01 6.95099831e-01
3.06453884e-01 -8.46381187e-01 -5.16518712e-01 -4.49150831e-01
-1.02039553e-01 3.73038828e-01 1.05917239e+00 6.81998730e-01
-6.03246629e-01 2.94788480e-01 -6.80880725e-01 -1.12482287e-01
4.42327023e-01 3.16972107e-01 1.03310406e+00 -1.30840373e+00
3.10014606e-01 -4.43935692e-01 -7.74134090e-03 -1.29549992e+00
4.35671270e-01 -9.25711513e-01 -3.38505179e-01 -1.65906119e+00
3.60535890e-01 -4.32056725e-01 -2.22462296e-01 3.45136762e-01
-1.82521746e-01 3.00707757e-01 5.94082654e-01 4.81718242e-01
-2.55451769e-01 7.21353769e-01 1.48611724e+00 -3.72543573e-01
-3.19247767e-02 -5.93694784e-02 -5.29431164e-01 2.79043317e-01
3.84019434e-01 -2.55531639e-01 -6.18313909e-01 -6.21287167e-01
5.29590771e-02 4.32735831e-01 4.63371128e-01 -6.55438125e-01
4.39108223e-01 -1.44523010e-02 9.02716696e-01 -9.84678209e-01
7.40266263e-01 -1.05952942e+00 4.75945979e-01 2.72514910e-01
-4.66009408e-01 2.09364332e-02 -3.27320136e-02 2.92851985e-01
-3.11103106e-01 -5.53744733e-01 5.14564991e-01 -3.16470325e-01
-4.31299835e-01 7.18252420e-01 2.18940392e-01 -4.96231228e-01
8.77386928e-01 -7.87880644e-02 -1.00670733e-01 -2.57766575e-01
-5.84240615e-01 2.50913233e-01 3.10714245e-01 6.77128911e-01
1.08232832e+00 -1.91298950e+00 -6.62991524e-01 3.64611596e-01
-3.22246403e-02 -3.53828669e-01 4.92979974e-01 1.13059476e-01
-5.14161289e-01 3.69278938e-01 -4.00936455e-01 -1.11124955e-01
-1.44563949e+00 8.93799424e-01 1.27710477e-01 1.07745834e-01
-6.18206978e-01 7.23915398e-01 3.26290131e-01 -4.84303117e-01
3.58489245e-01 1.61616102e-01 -3.84272546e-01 4.01187003e-01
7.10835338e-01 2.67301321e-01 -3.16138238e-01 -5.63489795e-01
-1.14847332e-01 1.12361896e+00 -1.55734554e-01 -1.51132762e-01
1.15153551e+00 -1.96221806e-02 -3.81584913e-01 1.58416301e-01
1.49489009e+00 -2.67744631e-01 -1.16424882e+00 -2.04053015e-01
-2.70592302e-01 -7.91767955e-01 -1.93877012e-01 -7.96814740e-01
-1.31880009e+00 1.16216373e+00 5.94188750e-01 -2.45145157e-01
1.06156385e+00 -5.05205393e-01 8.92380834e-01 8.02737296e-01
3.35342795e-01 -9.13451254e-01 2.81220675e-01 2.99882535e-02
1.51863909e+00 -9.95873570e-01 -1.27721718e-02 -1.50414810e-01
-5.48124015e-01 1.62344182e+00 6.15006328e-01 -2.73541480e-01
5.85675299e-01 -2.79243350e-01 5.05688153e-02 -3.21531534e-01
-6.99227214e-01 1.50922462e-01 4.05937582e-01 3.56461704e-01
1.93925291e-01 -3.09383273e-01 -2.50210285e-01 4.75178689e-01
2.15643957e-01 6.04361117e-01 4.87412736e-02 5.40827572e-01
-1.30852625e-01 -1.32518065e+00 -1.91328511e-01 5.34335494e-01
2.32347995e-02 2.93227788e-02 -8.90912354e-01 7.20564008e-01
5.44982255e-02 6.24586403e-01 1.18722953e-02 -1.25296444e-01
3.77786934e-01 3.93878400e-01 7.27958083e-01 -3.04428846e-01
-5.75168431e-01 -2.61445910e-01 -3.63467246e-01 -4.95765150e-01
-2.48316318e-01 -1.38947278e-01 -7.02133954e-01 -1.56809196e-01
-2.65959561e-01 9.46682915e-02 6.37384534e-01 6.51928425e-01
4.28111285e-01 1.96719784e-02 9.04991448e-01 -1.18622720e+00
-5.39466023e-01 -1.10592651e+00 -6.16567671e-01 5.13626456e-01
5.58056645e-02 -6.36689603e-01 -1.24134347e-01 1.57071613e-02]
|
[11.830499649047852, 0.4508243203163147]
|
e13f9f56-32e7-4d92-8604-26966096258e
|
comparison-of-interactive-knowledge-base
|
2010.10472
| null |
https://arxiv.org/abs/2010.10472v1
|
https://arxiv.org/pdf/2010.10472v1.pdf
|
Comparison of Interactive Knowledge Base Spelling Correction Models for Low-Resource Languages
|
Spelling normalization for low resource languages is a challenging task because the patterns are hard to predict and large corpora are usually required to collect enough examples. This work shows a comparison of a neural model and character language models with varying amounts on target language data. Our usage scenario is interactive correction with nearly zero amounts of training examples, improving models as more data is collected, for example within a chat app. Such models are designed to be incrementally improved as feedback is given from users. In this work, we design a knowledge-base and prediction model embedded system for spelling correction in low-resource languages. Experimental results on multiple languages show that the model could become effective with a small amount of data. We perform experiments on both natural and synthetic data, as well as on data from two endangered languages (Ainu and Griko). Last, we built a prototype system that was used for a small case study on Hinglish, which further demonstrated the suitability of our approach in real world scenarios.
|
['Alan W Black', 'Antonios Anastasopoulos', 'Yiyuan Li']
|
2020-10-20
| null | null | null | null |
['spelling-correction']
|
['natural-language-processing']
|
[ 7.39589930e-02 5.62265702e-02 6.88100830e-02 -5.37711918e-01
-6.77880704e-01 -4.96061742e-01 6.73311293e-01 2.81976074e-01
-8.75483990e-01 7.84123302e-01 2.80133814e-01 -2.75624514e-01
4.10885751e-01 -4.28761870e-01 -6.02447033e-01 6.33494521e-04
5.20673464e-04 8.45141232e-01 4.72206116e-01 -6.54194832e-01
3.49846929e-01 4.52858567e-01 -1.46068859e+00 4.79825407e-01
9.69796598e-01 -1.36099145e-01 6.28349006e-01 8.88812065e-01
-5.40117145e-01 6.45962417e-01 -9.59110856e-01 -4.36169416e-01
3.65108252e-01 -1.09361686e-01 -6.33778393e-01 -1.63963631e-01
4.98460740e-01 -3.78956944e-02 -1.15254580e-03 8.12861264e-01
5.98875105e-01 -1.53522403e-03 4.70170736e-01 -9.47471261e-01
-5.70198298e-01 1.04087639e+00 -1.06221884e-01 1.42823741e-01
2.65760690e-01 2.08002537e-01 7.11757898e-01 -8.07686150e-01
6.48398578e-01 1.09247696e+00 8.80283594e-01 8.85138392e-01
-9.41369772e-01 -5.18632591e-01 -7.71829262e-02 1.89447284e-01
-1.16329551e+00 -5.30497968e-01 2.83611596e-01 -3.87434930e-01
1.36726487e+00 3.85407478e-01 4.53229874e-01 6.58069611e-01
-1.55523092e-01 9.30713117e-01 1.02435064e+00 -1.07457614e+00
4.07681614e-03 7.74701715e-01 2.03387320e-01 2.61393964e-01
1.71488822e-01 -2.45043293e-01 -6.91952825e-01 -4.25836295e-02
3.64911884e-01 -2.29365110e-01 -1.56857342e-01 2.34140903e-02
-8.13149631e-01 6.74108148e-01 -9.56202969e-02 6.86290085e-01
-1.05343707e-01 -4.26126197e-02 5.37719965e-01 5.99964201e-01
7.18117535e-01 6.82619154e-01 -8.35387290e-01 -4.63473558e-01
-1.22539401e+00 4.46309775e-01 1.24569035e+00 1.23976743e+00
6.60730243e-01 7.00278357e-02 -1.15879811e-02 1.19591999e+00
7.22627565e-02 3.63800913e-01 7.86417127e-01 -3.05827737e-01
7.10375190e-01 7.61636913e-01 2.41552949e-01 -4.77240533e-01
-3.95025522e-01 -4.49596047e-02 -5.29202759e-01 2.07717165e-01
8.03618252e-01 -1.48563713e-01 -8.68418753e-01 1.60305941e+00
-6.57593906e-02 -2.77369171e-01 7.74134845e-02 6.31472707e-01
6.56424999e-01 7.62448311e-01 7.76016116e-02 -2.98059672e-01
8.81500006e-01 -8.60539138e-01 -7.47690916e-01 -4.17294383e-01
1.05228961e+00 -1.00921750e+00 1.65929961e+00 6.49162829e-01
-1.07850778e+00 -3.10541928e-01 -8.31139505e-01 -3.98091413e-02
-6.23600423e-01 3.31728965e-01 3.82224202e-01 8.32197309e-01
-9.25937653e-01 8.81426036e-01 -6.62664890e-01 -9.25030589e-01
-1.67955995e-01 4.08038259e-01 -2.84144282e-01 1.14339486e-01
-1.16938663e+00 1.07175970e+00 6.85380101e-01 -8.05483460e-02
-4.32320237e-01 -6.67829335e-01 -5.31305730e-01 -6.63226545e-02
2.18479738e-01 5.95322810e-02 1.62945783e+00 -1.23753357e+00
-1.45275402e+00 9.91588593e-01 1.73192024e-01 -5.61045051e-01
8.03365588e-01 -5.39551973e-01 -4.26640540e-01 -5.90791345e-01
-2.01566815e-01 2.98871696e-01 2.80943245e-01 -1.07340848e+00
-7.78842092e-01 -1.25301406e-01 -1.10816739e-01 7.75338262e-02
-7.52959251e-01 7.70443320e-01 -3.46889794e-01 -8.38452578e-01
-5.76880038e-01 -8.41427028e-01 -1.83440566e-01 -4.25313473e-01
-8.57601166e-02 -1.08190954e-01 4.65996951e-01 -1.26187956e+00
1.61678410e+00 -1.78830051e+00 4.14214693e-02 2.54271805e-01
-3.45372438e-01 8.34569275e-01 -2.53522307e-01 7.62400985e-01
2.45817229e-02 3.37763995e-01 -3.34108919e-01 -3.49660635e-01
-1.44186690e-01 3.00393313e-01 -2.09738687e-03 1.01185501e-01
7.71163264e-03 3.95709842e-01 -6.95290089e-01 -3.19176346e-01
-1.08947910e-01 1.33265436e-01 -7.32454658e-01 5.13580322e-01
-4.74447370e-01 -7.17368647e-02 2.11599737e-01 3.86941880e-01
4.05240685e-01 3.05902958e-01 1.53731957e-01 4.33427334e-01
-4.98266220e-01 3.16276371e-01 -1.36766481e+00 1.48911858e+00
-7.49735653e-01 7.18873560e-01 7.10889185e-03 -3.93491924e-01
1.16932094e+00 2.99071223e-01 -1.15930893e-01 -4.25058782e-01
-3.72566422e-03 4.04299080e-01 2.72980064e-01 -7.14044154e-01
9.56849277e-01 1.39798179e-01 -2.93033749e-01 7.38552392e-01
1.03675023e-01 -4.17498201e-01 7.52484620e-01 2.85419345e-01
8.62591267e-01 1.51676849e-01 5.87784469e-01 -2.50873953e-01
4.99699771e-01 2.70085037e-01 5.76791286e-01 9.53051984e-01
-3.74582293e-03 5.07942736e-01 2.08991244e-01 -6.05179012e-01
-1.51900148e+00 -3.48455727e-01 1.73712716e-01 1.51450360e+00
-4.57944393e-01 -7.48117805e-01 -1.13236427e+00 -5.22756577e-01
-3.90439153e-01 1.19366705e+00 -3.82681906e-01 1.20009907e-01
-9.51991975e-01 -7.28242159e-01 8.07977796e-01 3.78204316e-01
5.92934750e-02 -1.50654554e+00 -5.84021151e-01 3.95465761e-01
4.57500853e-02 -6.92678332e-01 -4.97967005e-01 2.49789774e-01
-5.07538676e-01 -6.53694630e-01 -4.44895774e-01 -8.10739875e-01
3.84159982e-01 -1.44965827e-01 1.11882722e+00 4.38554436e-01
-2.78265864e-01 1.39031738e-01 -5.11123717e-01 -9.96069312e-01
-1.16965127e+00 5.09853542e-01 2.70091772e-01 -3.98371756e-01
5.62399268e-01 -3.50638866e-01 2.09915876e-01 5.79609387e-02
-9.64800954e-01 1.76066414e-01 5.07514417e-01 9.26394939e-01
1.08748905e-01 -5.26044071e-01 4.43711817e-01 -1.35166097e+00
9.89144444e-01 -2.88467616e-01 -8.28854561e-01 6.02106869e-01
-5.21317780e-01 7.27366805e-02 1.00094664e+00 -6.06518507e-01
-1.17816925e+00 1.37759820e-01 -2.27393687e-01 1.89492315e-01
-4.47650015e-01 6.06079638e-01 -1.97909653e-01 1.40809134e-01
1.04682708e+00 1.54204160e-01 -3.03525805e-01 -1.02355897e+00
3.46594393e-01 1.26248360e+00 3.68259519e-01 -5.46654761e-01
6.70913637e-01 -3.48902911e-01 -7.13416278e-01 -1.08510113e+00
-3.82731020e-01 -4.73518372e-01 -9.60362554e-01 -2.40987033e-01
1.82638377e-01 -5.93589664e-01 -3.23289901e-01 5.84805369e-01
-1.37029040e+00 -6.82016909e-01 -1.86245188e-01 4.09030020e-01
-3.85866642e-01 2.87346452e-01 -7.13955522e-01 -8.63256574e-01
-6.26441777e-01 -7.34905779e-01 4.62352633e-01 3.68635297e-01
-4.89482492e-01 -9.13824022e-01 3.81286025e-01 6.42765090e-02
6.54700935e-01 -3.20918202e-01 1.00415576e+00 -1.36384106e+00
-3.42807531e-01 -4.03019071e-01 1.21735908e-01 3.92851114e-01
-3.99134047e-02 1.65744990e-01 -8.45630884e-01 -1.77226126e-01
-4.13095474e-01 -5.49812317e-01 3.06936204e-01 -3.56689364e-01
9.32333529e-01 -4.27495539e-01 1.37909546e-01 3.26451093e-01
1.12215257e+00 6.66430667e-02 4.46985960e-01 5.92389703e-01
6.62795663e-01 7.54059315e-01 8.00730705e-01 3.66604269e-01
2.16498211e-01 8.22446048e-01 -1.14067018e-01 2.74632685e-02
-2.36408025e-01 -3.57388079e-01 5.00533223e-01 1.32776916e+00
-9.04216170e-02 -3.64607573e-01 -1.32527912e+00 7.47527778e-01
-1.85020041e+00 -8.22084069e-01 -3.38832140e-01 2.34097099e+00
1.26822817e+00 4.69246842e-02 2.90706366e-01 -6.27255887e-02
5.10614634e-01 -4.04669255e-01 -8.82654861e-02 -9.49856937e-01
-6.13889806e-02 7.91169144e-03 5.93015313e-01 7.27044940e-01
-8.49123061e-01 1.46033704e+00 6.39273119e+00 8.74864757e-01
-1.28551888e+00 1.66872174e-01 3.24369580e-01 -1.18791617e-01
-1.85418591e-01 2.94119883e-02 -1.10741591e+00 4.48779762e-01
1.29099917e+00 -4.61379588e-01 5.77804923e-01 9.17148948e-01
3.95735651e-01 -6.33281916e-02 -9.79447901e-01 6.95564866e-01
3.32515270e-01 -1.18073678e+00 -1.93814218e-01 -3.55197161e-01
3.53193760e-01 3.96012247e-01 -6.44442379e-01 4.22424614e-01
6.22596502e-01 -8.17593157e-01 8.37634265e-01 5.29617906e-01
7.43528306e-01 -7.55912125e-01 8.25852275e-01 9.93747354e-01
-4.37069952e-01 1.52614653e-01 -3.97460639e-01 -2.08697036e-01
1.26297832e-01 2.64525209e-02 -1.48879826e+00 1.68876410e-01
7.11381495e-01 4.07100290e-01 -1.00584960e+00 1.35168874e+00
-4.24687415e-01 1.01691842e+00 -5.50877213e-01 -6.91584408e-01
-1.15632802e-01 -1.88516232e-03 5.10084212e-01 2.05235958e+00
2.14976013e-01 -1.11355551e-01 1.65568545e-01 4.47079480e-01
1.30066380e-01 8.12796950e-01 -5.14552355e-01 1.94546692e-02
2.99254000e-01 1.31893861e+00 -4.25090969e-01 -3.96508545e-01
-4.53492641e-01 8.67018819e-01 6.79801345e-01 6.99525699e-02
-4.84732151e-01 -6.71091259e-01 2.63487548e-01 3.75333816e-01
-3.15298885e-02 -1.10213220e-01 -3.04841816e-01 -1.34926498e+00
-1.43813059e-01 -1.29930019e+00 2.71917969e-01 -6.31819010e-01
-1.16496098e+00 8.11382294e-01 1.28775552e-01 -1.12803435e+00
-5.06230533e-01 -7.01559007e-01 -5.78850865e-01 1.19685709e+00
-1.15279961e+00 -1.21919584e+00 -1.84391979e-02 2.73016065e-01
7.95772910e-01 -6.14912450e-01 9.86471832e-01 5.04163206e-01
-4.75052327e-01 8.72595310e-01 2.40570441e-01 1.81336999e-01
9.38914716e-01 -1.42938709e+00 6.27178252e-01 1.20701671e+00
2.77076006e-01 6.85446024e-01 1.03734219e+00 -9.13441837e-01
-1.02936649e+00 -9.72521603e-01 1.52446151e+00 -4.14482504e-01
7.19902039e-01 -9.09458220e-01 -1.31588984e+00 5.83547652e-01
4.61325914e-01 -4.29985136e-01 8.49001765e-01 1.98354051e-01
-8.14469904e-02 3.91334333e-02 -1.20250678e+00 7.41143763e-01
8.76401544e-01 -1.71637550e-01 -7.20217526e-01 5.45287609e-01
4.73013163e-01 -3.50873381e-01 -6.19995058e-01 -8.21655765e-02
3.52352738e-01 -6.16428137e-01 2.57730812e-01 -8.95272851e-01
2.05022097e-01 -4.00681585e-01 -3.77880931e-02 -1.64599335e+00
-4.80797067e-02 -7.93985426e-01 3.00549388e-01 1.71954620e+00
8.86061192e-01 -1.71569020e-01 5.93330681e-01 8.10558379e-01
9.66486614e-03 -1.53716922e-01 -5.62487185e-01 -7.16452479e-01
2.06857622e-01 -6.21008933e-01 5.43985784e-01 9.18990493e-01
1.80386454e-01 4.07809794e-01 -8.34818363e-01 -1.16838790e-01
1.23236224e-01 -4.11500365e-01 1.19810295e+00 -1.07218277e+00
-5.46949506e-01 -6.31031543e-02 -2.66457766e-01 -5.16300261e-01
4.19844268e-03 -9.75954235e-01 2.61280209e-01 -1.15616393e+00
1.24360994e-01 -5.51980197e-01 2.41516575e-01 7.45809495e-01
-1.82278350e-01 2.74709053e-03 5.63294232e-01 1.46245062e-01
-2.83637971e-01 2.36267686e-01 3.95374566e-01 6.32408932e-02
-5.18529296e-01 2.09727511e-01 -3.09990138e-01 9.83266890e-01
9.28249359e-01 -8.00792992e-01 -6.25157356e-03 -5.84385395e-01
3.70822757e-01 -2.51142621e-01 -3.47511530e-01 -9.02333558e-01
3.63673091e-01 -3.19262296e-01 1.10732421e-01 -4.33234632e-01
2.55690366e-02 -6.83644474e-01 -6.37963042e-02 4.50743496e-01
-5.89323819e-01 3.48734796e-01 3.78438622e-01 -2.16044509e-03
3.95815559e-02 -6.98528945e-01 7.12251246e-01 -2.22532660e-01
-9.23008442e-01 -1.96738541e-02 -5.72046995e-01 9.92439687e-02
7.89475262e-01 -8.07777941e-02 -5.33879884e-02 -3.89737636e-01
-5.70910275e-01 7.42057934e-02 5.80264807e-01 6.98755145e-01
2.23444432e-01 -8.95745099e-01 -1.04347885e+00 2.87542164e-01
4.04198527e-01 -5.14923632e-01 -4.08351332e-01 2.27967352e-01
-1.06451452e+00 1.61451116e-01 -2.68788159e-01 -2.96767414e-01
-1.74221051e+00 4.42850232e-01 2.81830877e-01 -3.27590853e-01
-4.79158938e-01 6.63023829e-01 -4.72544461e-01 -1.13409019e+00
4.02154446e-01 -2.61817455e-01 -4.84151721e-01 -2.42149681e-02
9.46981668e-01 1.93870634e-01 3.33175480e-01 -4.93069887e-01
-1.61687538e-01 1.11715207e-02 -5.47478795e-01 -3.78294766e-01
1.74082839e+00 -1.95263904e-02 -8.79180282e-02 7.82673538e-01
5.52667797e-01 4.25708532e-01 -8.57927740e-01 -4.39511389e-01
5.30299842e-01 -6.04734659e-01 -5.58213770e-01 -1.11046994e+00
-4.91072536e-01 9.57742333e-01 6.48494363e-01 3.28796208e-02
9.19918597e-01 -2.56939858e-01 4.24365550e-01 8.43916774e-01
2.42729202e-01 -1.54097414e+00 -3.25379550e-01 8.90573263e-01
1.10093689e+00 -1.13683534e+00 -8.00081342e-02 -4.76878397e-02
-8.18600655e-01 1.36211574e+00 7.73995936e-01 5.56068122e-02
3.47467542e-01 5.71811616e-01 4.39756721e-01 2.54118145e-01
-8.42657030e-01 1.00669503e-01 -7.67339692e-02 5.68860233e-01
9.82639968e-01 2.34956861e-01 -7.08588779e-01 7.53737569e-01
-5.87140083e-01 1.78860947e-01 1.10392714e+00 9.99624312e-01
-5.89703381e-01 -1.67042887e+00 -4.48899895e-01 4.02105063e-01
-5.99297047e-01 -4.64758039e-01 -7.44297504e-01 9.68799531e-01
5.22043183e-02 6.88239098e-01 -1.32820740e-01 -1.22597635e-01
5.73741376e-01 1.45873979e-01 2.07106382e-01 -1.00632548e+00
-1.30447209e+00 -7.75371790e-02 5.43518245e-01 -1.29529282e-01
-1.08633311e-02 -8.78171742e-01 -9.30712283e-01 -3.49849969e-01
-3.64116132e-01 2.04646647e-01 8.91628802e-01 8.72834444e-01
-4.76953201e-02 5.86897023e-02 1.19238265e-01 -6.16436183e-01
-6.01937652e-01 -1.39817595e+00 -4.04389888e-01 4.91153181e-01
-1.30430922e-01 1.59284234e-01 -5.90793565e-02 3.99000436e-01]
|
[10.879021644592285, 10.295392036437988]
|
9414a548-5554-4457-a4ae-ba377aadddd3
|
multimodal-dialogue-state-tracking-by-qa
|
2007.09903
| null |
https://arxiv.org/abs/2007.09903v1
|
https://arxiv.org/pdf/2007.09903v1.pdf
|
Multimodal Dialogue State Tracking By QA Approach with Data Augmentation
|
Recently, a more challenging state tracking task, Audio-Video Scene-Aware Dialogue (AVSD), is catching an increasing amount of attention among researchers. Different from purely text-based dialogue state tracking, the dialogue in AVSD contains a sequence of question-answer pairs about a video and the final answer to the given question requires additional understanding of the video. This paper interprets the AVSD task from an open-domain Question Answering (QA) point of view and proposes a multimodal open-domain QA system to deal with the problem. The proposed QA system uses common encoder-decoder framework with multimodal fusion and attention. Teacher forcing is applied to train a natural language generator. We also propose a new data augmentation approach specifically under QA assumption. Our experiments show that our model and techniques bring significant improvements over the baseline model on the DSTC7-AVSD dataset and demonstrate the potentials of our data augmentation techniques.
|
['Ian Steenstra', 'Brandyn Sigouin', 'Xiangyang Mou', 'Hui Su']
|
2020-07-20
| null | null | null | null |
['scene-aware-dialogue']
|
['computer-vision']
|
[ 3.62058431e-01 3.91388029e-01 1.96359932e-01 -6.89255178e-01
-1.39779377e+00 -6.68374062e-01 1.01618767e+00 -9.36579555e-02
-2.71140486e-01 6.43015325e-01 8.39749038e-01 -3.29448879e-01
3.19729477e-01 -1.08913198e-01 -4.50519264e-01 -3.62752169e-01
2.65946031e-01 6.18291140e-01 4.70750481e-01 -5.65057814e-01
2.46943563e-01 -3.18405211e-01 -1.14942408e+00 6.85502172e-01
7.60582685e-01 9.94871616e-01 3.85046333e-01 1.37836802e+00
-3.52224022e-01 1.48033047e+00 -8.66586864e-01 -4.02980506e-01
-9.83685255e-02 -8.20346475e-01 -1.63808823e+00 4.51600164e-01
7.61720419e-01 -6.19735897e-01 -6.98453665e-01 8.28309417e-01
6.95815563e-01 2.74114370e-01 4.51060265e-01 -1.43562973e+00
-5.34683645e-01 1.26935840e-01 -3.29840660e-01 3.02644014e-01
1.18935585e+00 3.21632206e-01 8.54766130e-01 -5.52039802e-01
7.57295847e-01 1.53671491e+00 5.31398058e-02 8.45373511e-01
-7.49568820e-01 -1.80728316e-01 1.60649061e-01 6.48404121e-01
-8.25915456e-01 -6.81253314e-01 7.18018174e-01 -1.52151302e-01
8.15268040e-01 2.88783550e-01 3.19580793e-01 1.27303505e+00
-5.68776727e-02 1.23400962e+00 1.02785194e+00 -4.21354413e-01
3.52869965e-02 1.21622356e-02 2.51644880e-01 7.56011069e-01
-8.05820465e-01 -6.98387682e-01 -6.65591836e-01 1.81203876e-02
4.08399522e-01 -5.66307008e-01 -4.74451065e-01 -2.97524601e-01
-1.45238197e+00 7.60576069e-01 -1.36698112e-01 -2.67005023e-02
-6.44720048e-02 -3.79068926e-02 8.14186990e-01 7.24354982e-01
3.21577758e-01 2.04443559e-01 -4.70336586e-01 -6.95762098e-01
-6.14938617e-01 2.26990968e-01 1.03069758e+00 9.86099958e-01
4.19275612e-01 -1.42153483e-02 -5.83497345e-01 8.68947268e-01
3.72365087e-01 4.58923668e-01 3.40074867e-01 -1.41062737e+00
8.72103155e-01 6.46026790e-01 1.60399437e-01 -6.38636589e-01
-8.69134963e-02 3.18926096e-01 -5.56828141e-01 -2.22061098e-01
6.13418579e-01 -3.53851616e-01 -6.89905882e-01 1.68586624e+00
6.28103733e-01 1.52246192e-01 5.57237446e-01 9.35594082e-01
1.21882653e+00 1.30463123e+00 -6.52770177e-02 -3.94602954e-01
1.65543556e+00 -1.31003582e+00 -1.44807923e+00 -1.88127339e-01
4.47631747e-01 -6.24934077e-01 1.13853657e+00 3.93903077e-01
-1.06822491e+00 -4.85625476e-01 -7.70595253e-01 -4.74376112e-01
4.58538160e-02 -9.39065665e-02 1.69360906e-01 3.36782664e-01
-9.98084128e-01 -2.23206982e-01 -6.37137532e-01 -6.03473783e-01
7.52812484e-03 1.79175079e-01 -4.67387527e-01 -2.11999059e-01
-1.31827617e+00 1.00762689e+00 1.58839732e-01 -3.47628966e-02
-1.31517756e+00 -2.33201385e-02 -1.19428015e+00 -1.73830554e-01
8.72608960e-01 -6.46018505e-01 1.92358756e+00 -1.10317206e+00
-1.93558729e+00 7.67612755e-01 -4.66937333e-01 -6.58604324e-01
1.78165317e-01 -4.76363480e-01 -5.23672342e-01 8.61112773e-01
-5.15507571e-02 8.70371699e-01 1.16529167e+00 -8.95049572e-01
-6.16787732e-01 -2.88877934e-01 4.25938785e-01 6.05218172e-01
-7.40165785e-02 3.19378287e-01 -5.11248469e-01 -1.18330009e-01
-2.24270046e-01 -7.03451216e-01 -7.01310635e-02 -2.45419309e-01
-2.93335408e-01 -4.12799239e-01 1.42602801e+00 -8.85250628e-01
1.29606748e+00 -1.87394559e+00 3.71494651e-01 -7.25944638e-01
6.55424595e-02 4.19004828e-01 -2.94553965e-01 7.29098856e-01
-2.06790473e-02 -4.84734952e-01 -2.55460501e-01 -2.39622533e-01
-7.42614642e-03 3.37568611e-01 -3.81675094e-01 3.72093111e-01
4.55888152e-01 8.22414935e-01 -8.18547428e-01 -7.34791577e-01
7.65305459e-02 1.18889935e-01 -4.77908492e-01 8.43780696e-01
-8.08736265e-01 7.94184327e-01 -3.15388381e-01 4.64995503e-01
4.43651676e-01 -3.16377163e-01 -1.09483540e-01 -2.17968315e-01
-1.03108451e-01 4.08050209e-01 -1.03318846e+00 2.43219280e+00
-2.56799906e-01 1.06327474e+00 5.11787653e-01 -1.05867112e+00
7.10142791e-01 8.36865902e-01 1.35429218e-01 -4.71998334e-01
2.07998693e-01 -2.22103760e-01 5.45108430e-02 -1.18593311e+00
6.95250809e-01 7.09117576e-02 -4.36667025e-01 5.62281787e-01
7.69338906e-01 -1.22101098e-01 1.29337043e-01 8.22185755e-01
1.15830469e+00 3.81455421e-01 3.12656432e-01 6.91729262e-02
9.85572100e-01 3.21651518e-01 1.80567294e-01 4.29861128e-01
-5.95326006e-01 6.38008416e-01 8.12163651e-01 -2.05054767e-02
-7.61903644e-01 -6.90590799e-01 3.67879182e-01 1.38360715e+00
1.38086334e-01 -4.23908085e-01 -9.97018516e-01 -9.58723247e-01
-5.90256512e-01 5.92386305e-01 -3.33097339e-01 8.65918547e-02
-4.30099249e-01 1.26779992e-02 7.49633372e-01 1.20898589e-01
9.39647436e-01 -8.32233846e-01 -4.77547884e-01 -4.01745513e-02
-9.04003263e-01 -1.50149238e+00 -6.77475989e-01 -1.47080764e-01
-6.53657675e-01 -1.15580869e+00 -9.09280658e-01 -1.07858479e+00
3.20427388e-01 2.08449692e-01 1.04683173e+00 -3.77907097e-01
1.06168084e-01 1.05551803e+00 -6.27105951e-01 -1.67606607e-01
-8.62912238e-01 -6.00888282e-02 -3.57930452e-01 4.19471294e-01
2.37988204e-01 5.45478910e-02 -5.15395403e-01 3.47831607e-01
-7.03444839e-01 3.11783791e-01 2.05877155e-01 8.36153924e-01
1.88570842e-02 -5.26030123e-01 9.48921800e-01 -5.97090423e-01
9.10153508e-01 -5.79782784e-01 -2.96912134e-01 3.53703380e-01
8.58134106e-02 2.62130022e-01 2.02535495e-01 -4.14843291e-01
-1.69067883e+00 1.73633635e-01 -3.91193867e-01 -2.74750859e-01
-6.08413041e-01 3.90814483e-01 -3.98314357e-01 3.45613092e-01
4.03950900e-01 3.77027363e-01 3.53049517e-01 -3.45738590e-01
4.77613509e-01 1.05481422e+00 1.07643163e+00 -4.08956319e-01
4.95761603e-01 3.08494300e-01 -4.10422742e-01 -9.06257808e-01
-7.90474713e-01 -7.74556756e-01 -4.88511026e-01 -5.59172988e-01
1.30696023e+00 -1.18288112e+00 -9.30864751e-01 4.33125526e-01
-1.50692761e+00 -3.20989609e-01 -8.48840103e-02 1.92930177e-01
-8.13420892e-01 8.45562398e-01 -6.41814947e-01 -1.10370827e+00
-3.23377937e-01 -1.19359910e+00 1.20657253e+00 1.73711002e-01
-8.58087242e-02 -8.48222971e-01 3.08958471e-01 1.10232472e+00
5.89188896e-02 3.54395472e-02 5.72789371e-01 -1.14520764e+00
-6.11724257e-01 -1.11686662e-01 -7.39918724e-02 4.75015849e-01
2.88091488e-02 -3.35437745e-01 -9.98796046e-01 -5.10691777e-02
3.91828924e-01 -8.49625289e-01 6.07700706e-01 1.03119157e-01
5.36359787e-01 -4.05014664e-01 2.02013165e-01 -9.93035287e-02
9.67876554e-01 3.71590853e-01 6.81525826e-01 2.91195326e-02
4.66020256e-01 7.77458489e-01 9.84176636e-01 3.34373266e-01
8.33997071e-01 6.33522689e-01 5.95076084e-01 1.52519643e-01
-2.38617927e-01 -2.14764535e-01 7.93601036e-01 1.11603475e+00
4.19298410e-01 -6.28235519e-01 -8.63325775e-01 7.69244790e-01
-2.01419353e+00 -1.13048875e+00 -4.12115574e-01 1.68838584e+00
8.57287109e-01 -3.69516164e-01 1.69439122e-01 7.95283765e-02
6.87059522e-01 4.60105002e-01 -4.04233545e-01 -6.07189000e-01
-9.46350172e-02 -2.12272316e-01 -2.49826416e-01 5.81184268e-01
-1.11879337e+00 9.17123139e-01 5.84443378e+00 5.81089199e-01
-6.67609572e-01 4.49410021e-01 3.90767366e-01 3.98575485e-01
1.34554729e-01 1.21022552e-01 -6.31597877e-01 8.39226469e-02
1.30737257e+00 1.33852363e-01 -8.13339092e-03 4.92779404e-01
2.72807509e-01 -6.81381702e-01 -1.09578693e+00 8.91114295e-01
6.10753834e-01 -1.26206100e+00 -1.17369793e-01 -3.71446490e-01
4.44026321e-01 -2.55043417e-01 -1.77781761e-01 4.28126037e-01
2.45430648e-01 -7.23205209e-01 4.88309830e-01 4.91262645e-01
6.37895226e-01 -5.01598895e-01 6.48168445e-01 7.08453536e-01
-9.90493953e-01 -9.31115746e-02 7.28463009e-02 -1.56537026e-01
6.70754015e-01 -4.69327867e-01 -1.24333632e+00 5.46137154e-01
4.10800159e-01 4.55133229e-01 -5.96212327e-01 6.77421033e-01
-1.26508102e-01 7.13999987e-01 5.27507775e-02 -8.70755911e-02
3.60721409e-01 5.45960143e-02 7.09106922e-01 1.17925191e+00
-4.66930540e-03 3.15681279e-01 1.67695223e-03 3.96413565e-01
1.06733501e-01 4.03918847e-02 -6.69206202e-01 -1.21488549e-01
1.20671894e-02 1.13040948e+00 -5.94809763e-02 -5.18719554e-01
-7.92046785e-01 1.32397974e+00 -3.96334939e-02 2.02678099e-01
-7.41640151e-01 -2.59086192e-01 3.30955386e-01 -1.96213588e-01
1.85541674e-01 -2.30319887e-01 4.83359426e-01 -1.48522735e+00
-3.65742952e-01 -1.35493612e+00 6.36879563e-01 -1.19622207e+00
-7.33175218e-01 6.61289036e-01 1.05512485e-01 -1.34909534e+00
-6.49124980e-01 -2.99936682e-01 -4.15970266e-01 5.09637356e-01
-1.41414833e+00 -1.11747086e+00 -3.53103876e-01 1.03833079e+00
1.41647053e+00 -3.22847486e-01 8.76016259e-01 2.07797453e-01
-3.75305176e-01 1.08740732e-01 -2.60674089e-01 3.17682326e-01
1.08058453e+00 -1.25628066e+00 1.87689334e-01 8.13454926e-01
1.20634973e-01 1.82735488e-01 9.32058811e-01 -5.55952728e-01
-1.86143911e+00 -8.09234977e-01 7.64191985e-01 -6.68397546e-01
6.61400199e-01 -3.53664011e-01 -9.66212571e-01 7.61947870e-01
1.32381082e+00 -3.87905031e-01 5.94389975e-01 -4.83092844e-01
-1.11447275e-01 1.14910871e-01 -1.02187288e+00 4.12492752e-01
4.95851845e-01 -7.26525366e-01 -1.16817677e+00 3.24698657e-01
9.75212991e-01 -7.10708022e-01 -7.95721591e-01 1.98048979e-01
9.64048654e-02 -8.07049811e-01 6.63365722e-01 -8.07544291e-01
4.56890196e-01 -3.72422338e-01 -3.02075565e-01 -1.03478515e+00
4.83774424e-01 -1.09324253e+00 -4.32034701e-01 1.38370585e+00
1.51572064e-01 3.01407371e-02 5.46579659e-01 4.35641319e-01
-3.59036416e-01 -1.45763025e-01 -1.15049422e+00 -2.17536554e-01
-3.80255908e-01 -3.90657336e-01 9.49596167e-02 7.46288419e-01
4.47678387e-01 1.26307154e+00 -8.50144148e-01 2.19563454e-01
2.28808999e-01 -4.99847606e-02 1.19238997e+00 -8.90862286e-01
-2.44566455e-01 2.85093367e-01 -2.63052404e-01 -1.79091072e+00
1.58767357e-01 -3.68201107e-01 2.57302314e-01 -1.77988482e+00
5.03652245e-02 5.88974595e-01 4.80890542e-01 1.71153948e-01
-1.34453833e-01 -1.20912850e-01 4.05604511e-01 -6.71239272e-02
-1.49342763e+00 8.20693552e-01 1.43005121e+00 -2.80147821e-01
2.78427289e-03 -7.42762312e-02 -3.38969737e-01 6.40802145e-01
4.42239493e-01 -1.03273116e-01 -6.48987293e-01 -5.34807265e-01
7.14437515e-02 1.13923717e+00 3.88925433e-01 -8.63655686e-01
6.19188070e-01 9.74518880e-02 -2.64061391e-01 -9.07420397e-01
7.86557734e-01 -7.55351484e-01 -3.65137786e-01 3.20234239e-01
-6.57261431e-01 1.48264334e-01 1.60884202e-01 8.69955420e-01
-7.28924870e-01 -2.85390586e-01 4.86267656e-01 -1.29028738e-01
-1.05422807e+00 -8.82061664e-03 -8.61809969e-01 3.65081966e-01
1.01250374e+00 -9.68312249e-02 -6.25523210e-01 -1.30208504e+00
-7.95303702e-01 8.25210154e-01 -1.36239767e-01 5.82045078e-01
8.43843699e-01 -1.14508712e+00 -7.33891010e-01 -2.44118005e-01
1.95489198e-01 -2.96451803e-02 4.25659180e-01 8.92234087e-01
-3.47725540e-01 6.08229578e-01 7.41931722e-02 -8.03451955e-01
-1.55562711e+00 3.82214844e-01 2.93210417e-01 -2.27407321e-01
-3.27569574e-01 7.13783741e-01 1.55006662e-01 -4.12485123e-01
5.41085422e-01 -2.16605127e-01 -6.38405442e-01 2.25441843e-01
5.77197909e-01 2.35610425e-01 -1.90154493e-01 -9.72871363e-01
3.75589170e-03 7.34127313e-02 -8.51421654e-02 -5.06266892e-01
1.04578018e+00 -7.52437592e-01 1.03937447e-01 7.82537818e-01
1.13224697e+00 -3.91773671e-01 -1.40878248e+00 -3.65416735e-01
-1.18978880e-01 -1.79444805e-01 -1.99566901e-01 -9.01190579e-01
-5.49890518e-01 1.17748165e+00 6.10050619e-01 3.81742090e-01
1.11657441e+00 2.56345868e-01 9.94372368e-01 6.01657093e-01
4.58825845e-03 -1.20725060e+00 7.00989902e-01 9.19223666e-01
1.10465133e+00 -1.61999583e+00 -3.35699499e-01 -1.83068186e-01
-1.42138553e+00 9.91458893e-01 8.43366325e-01 2.09106430e-01
7.52458796e-02 -1.72368325e-02 3.67382139e-01 -2.53939807e-01
-1.31872654e+00 -2.77521133e-01 9.12681371e-02 6.51020765e-01
2.36555934e-01 -5.87402523e-01 -6.28582537e-02 4.48054880e-01
2.05870792e-01 3.34505900e-03 7.47800946e-01 1.04783654e+00
-6.87206388e-01 -9.72321570e-01 -4.97058600e-01 7.68745914e-02
-5.21876276e-01 4.65976931e-02 -5.43484271e-01 5.92906952e-01
-4.99458551e-01 1.42186201e+00 1.26812100e-01 -2.89215177e-01
2.83584952e-01 5.95014989e-01 3.38317215e-01 -7.15256333e-01
-6.73495829e-01 1.81225091e-01 7.15758801e-01 -4.72042769e-01
-8.65089715e-01 -6.45896256e-01 -1.12270260e+00 1.31873161e-01
-5.13843119e-01 4.09614384e-01 4.70781446e-01 1.14333045e+00
3.82475525e-01 4.53743219e-01 6.03015125e-01 -4.06610280e-01
-5.33873916e-01 -1.22446370e+00 -3.32392529e-02 2.73918658e-01
7.15801001e-01 -2.56727338e-01 -6.67480603e-02 5.08093178e-01]
|
[10.786334037780762, 1.1628167629241943]
|
ad8eb825-3f70-4d4f-b1f1-1b2b266fc2ab
|
a-pilot-study-of-query-free-adversarial
|
2303.16378
| null |
https://arxiv.org/abs/2303.16378v2
|
https://arxiv.org/pdf/2303.16378v2.pdf
|
A Pilot Study of Query-Free Adversarial Attack against Stable Diffusion
|
Despite the record-breaking performance in Text-to-Image (T2I) generation by Stable Diffusion, less research attention is paid to its adversarial robustness. In this work, we study the problem of adversarial attack generation for Stable Diffusion and ask if an adversarial text prompt can be obtained even in the absence of end-to-end model queries. We call the resulting problem 'query-free attack generation'. To resolve this problem, we show that the vulnerability of T2I models is rooted in the lack of robustness of text encoders, e.g., the CLIP text encoder used for attacking Stable Diffusion. Based on such insight, we propose both untargeted and targeted query-free attacks, where the former is built on the most influential dimensions in the text embedding space, which we call steerable key dimensions. By leveraging the proposed attacks, we empirically show that only a five-character perturbation to the text prompt is able to cause the significant content shift of synthesized images using Stable Diffusion. Moreover, we show that the proposed target attack can precisely steer the diffusion model to scrub the targeted image content without causing much change in untargeted image content. Our code is available at https://github.com/OPTML-Group/QF-Attack.
|
['Sijia Liu', 'Yihua Zhang', 'Haomin Zhuang']
|
2023-03-29
| null | null | null | null |
['adversarial-text']
|
['adversarial']
|
[ 5.23470461e-01 -5.90405334e-03 8.56434479e-02 2.54342526e-01
-1.03489232e+00 -1.27453065e+00 6.11594379e-01 -2.08894819e-01
-1.12073362e-01 2.85557687e-01 3.39382529e-01 -5.82669735e-01
6.43952042e-02 -6.37932479e-01 -9.75790918e-01 -7.21341133e-01
9.57230777e-02 -2.55292028e-01 3.51368517e-01 -2.91785598e-01
4.25982028e-01 4.37779069e-01 -7.11112320e-01 3.27078074e-01
5.56813717e-01 8.01607192e-01 -1.99507341e-01 1.11508822e+00
1.47775367e-01 8.70506406e-01 -1.01910770e+00 -8.47269654e-01
6.30460083e-01 -5.49588144e-01 -5.27816176e-01 -9.12996456e-02
4.97879148e-01 -7.53530085e-01 -1.00380433e+00 1.34936655e+00
7.84053087e-01 -4.52523857e-01 5.08648932e-01 -1.38293302e+00
-8.22016060e-01 7.12094247e-01 -5.11899829e-01 2.10390002e-01
4.47145820e-01 4.15890336e-01 8.12593520e-01 -7.23940551e-01
7.02088177e-01 1.23535979e+00 2.54984170e-01 6.05986118e-01
-1.06850159e+00 -8.89879704e-01 -1.74603805e-01 1.87832296e-01
-1.42106664e+00 -7.90838540e-01 9.69924629e-01 -2.99826115e-01
3.80965054e-01 6.94781005e-01 9.03000534e-02 1.49349642e+00
4.18890178e-01 6.05564415e-01 1.01304960e+00 -3.60741645e-01
2.12322976e-02 2.33207017e-01 -3.18507701e-01 5.46950221e-01
1.64760008e-01 4.52219218e-01 -4.99509960e-01 -4.34603125e-01
5.69675386e-01 -2.81830847e-01 -7.36717641e-01 -1.61621824e-01
-1.27348948e+00 9.21053171e-01 3.15959245e-01 4.85537015e-02
-6.34076493e-03 4.00928825e-01 4.86090124e-01 6.96891367e-01
2.85491139e-01 4.64450359e-01 -2.02081621e-01 -1.01983644e-01
-6.72420204e-01 2.10335985e-01 7.35311627e-01 1.02920246e+00
4.24418986e-01 2.97222227e-01 -2.23760068e-01 3.03830802e-01
-1.50458841e-02 1.08910227e+00 3.05370629e-01 -7.31055021e-01
7.96817958e-01 8.45330134e-02 -4.76708310e-03 -1.38179898e+00
9.52747017e-02 -2.22871274e-01 -8.42286944e-01 2.74237752e-01
3.24375808e-01 -5.22057295e-01 -5.46918988e-01 1.82622182e+00
2.73702502e-01 4.03501913e-02 2.86611587e-01 6.65071011e-01
3.70499164e-01 6.78664863e-01 -6.32275641e-01 -7.45752379e-02
1.23930144e+00 -6.58841550e-01 -6.38162255e-01 3.78595442e-02
6.27284944e-01 -1.10023975e+00 8.89429390e-01 1.45087823e-01
-9.94802952e-01 -2.57244408e-01 -1.28612459e+00 2.34198242e-01
-2.38869563e-01 -1.95234805e-01 -1.24476932e-01 1.00566900e+00
-8.23288262e-01 1.50746509e-01 -3.34854931e-01 -1.19695783e-01
1.83405250e-01 2.03960225e-01 -4.91264373e-01 1.53313875e-02
-1.47197509e+00 5.21397471e-01 6.72124922e-02 -5.30367494e-02
-1.07209992e+00 -5.49887061e-01 -4.67958003e-01 -1.58816934e-01
6.37974024e-01 -5.26677310e-01 9.12149191e-01 -8.90878022e-01
-1.60006177e+00 5.71628869e-01 7.04328716e-02 -6.94745958e-01
1.03626323e+00 -1.39069632e-01 -4.30938661e-01 7.21175075e-01
-7.36691132e-02 3.84992361e-01 1.63263535e+00 -1.31489742e+00
-2.00532734e-01 -2.45367929e-01 1.98396653e-01 -1.55301765e-01
-6.18029237e-01 2.58742899e-01 -4.27906483e-01 -1.09311759e+00
-2.02013165e-01 -1.30713391e+00 7.59599879e-02 1.72661901e-01
-1.00141704e+00 5.29215991e-01 1.27135682e+00 -5.69366813e-01
1.36181319e+00 -2.48509574e+00 5.44008724e-02 2.65564144e-01
3.46633762e-01 4.58958238e-01 -2.52673477e-01 9.14477825e-01
-1.81919709e-01 5.67467034e-01 -5.19684702e-02 -6.17976263e-02
-1.03815487e-02 -1.77789032e-01 -1.03995907e+00 6.16325855e-01
-5.99630997e-02 9.83507812e-01 -5.88714898e-01 -2.44704694e-01
1.50279040e-02 3.44179064e-01 -5.51216662e-01 2.13739246e-01
-2.49240194e-02 3.92633557e-01 -4.55711335e-01 3.40994239e-01
8.97893608e-01 9.58992615e-02 -3.72799374e-02 -3.75708431e-01
1.38274521e-01 -1.21893235e-01 -9.77706313e-01 1.25863492e+00
-2.01794699e-01 7.68203437e-01 6.48348778e-02 -5.71331501e-01
6.77571476e-01 3.91436934e-01 9.98289064e-02 -5.84512115e-01
4.66795452e-02 2.44306594e-01 3.58202606e-02 -3.43114167e-01
4.93484199e-01 2.58788258e-01 -3.01758051e-01 5.69273651e-01
-4.02430683e-01 -1.42136157e-01 -2.92155325e-01 6.44107401e-01
1.21706462e+00 -3.32706332e-01 -1.14512876e-01 4.23438363e-02
4.24010605e-01 -3.98541749e-01 2.59965301e-01 1.14954138e+00
-2.33461380e-01 6.49231434e-01 8.06635737e-01 -7.45302513e-02
-1.24330878e+00 -1.05808938e+00 1.80728704e-01 7.11410224e-01
2.98146993e-01 -6.24000013e-01 -9.53374803e-01 -9.03008342e-01
-1.17921434e-01 5.34474432e-01 -6.26365244e-01 -4.80597287e-01
-4.74277496e-01 -3.31301033e-01 1.31140363e+00 8.22420195e-02
6.15534008e-01 -5.66309154e-01 -4.07137841e-01 -3.50724980e-02
-1.84528366e-01 -1.21447980e+00 -9.84921694e-01 -2.07851484e-01
-4.49598432e-01 -1.05302346e+00 -7.98537314e-01 -2.90138543e-01
5.59602559e-01 4.67707306e-01 4.31215703e-01 8.11111405e-02
-1.08060867e-01 5.47457218e-01 -5.19490838e-01 -2.90496439e-01
-1.03727281e+00 1.36314449e-03 -3.09198722e-02 2.72381425e-01
-3.24816883e-01 -5.23670197e-01 -8.00592601e-01 4.92861718e-01
-1.49916673e+00 -2.28747185e-02 3.56290013e-01 6.50824010e-01
6.96629845e-03 2.50100970e-01 4.39039081e-01 -6.86365604e-01
7.54320800e-01 -3.26082826e-01 -5.77845573e-01 1.68352693e-01
-3.67495239e-01 1.23702548e-01 9.18196797e-01 -7.99177527e-01
-5.05948424e-01 -3.01765919e-01 -3.32486093e-01 -6.87112272e-01
1.49387345e-01 1.92731947e-01 -2.82369465e-01 -4.35845017e-01
8.27344358e-01 5.39114237e-01 -5.69419712e-02 -4.51684855e-02
6.86519563e-01 6.88118517e-01 4.73507047e-01 -5.27748883e-01
1.56836236e+00 6.51455760e-01 -7.56563768e-02 -7.75480211e-01
-2.83787787e-01 1.42533243e-01 -9.45804343e-02 -2.23004937e-01
7.03244269e-01 -6.07547581e-01 -6.39257431e-01 8.16554964e-01
-1.13935661e+00 -2.98701108e-01 -2.55677495e-02 5.77055402e-02
-5.65950990e-01 7.60093808e-01 -6.63818598e-01 -4.80275422e-01
-5.39970756e-01 -1.40971565e+00 8.93366218e-01 -2.68213391e-01
9.78244618e-02 -7.72018909e-01 -1.70953467e-01 1.90697163e-01
6.10629976e-01 3.80225599e-01 1.01361394e+00 -7.08208978e-01
-7.26149201e-01 -5.31776547e-01 -1.37232035e-01 4.98876989e-01
1.17158189e-01 4.98974584e-02 -8.46174657e-01 -6.32973731e-01
2.81339288e-01 -3.13339114e-01 5.84245741e-01 -1.28001675e-01
8.64411533e-01 -9.77910936e-01 -2.64034308e-02 7.46796131e-01
1.32973421e+00 1.66816935e-01 8.89738500e-01 3.64621788e-01
7.94796467e-01 2.40726143e-01 3.00228477e-01 4.71543401e-01
-7.27463067e-02 8.70580971e-01 4.35406774e-01 1.95022132e-02
-8.79215524e-02 -5.48067331e-01 6.79983854e-01 6.53460324e-01
5.76177478e-01 -6.41080499e-01 -5.75357199e-01 2.87589192e-01
-1.44406414e+00 -1.02706957e+00 2.25844588e-02 2.31012225e+00
9.28949237e-01 4.86379415e-01 -1.22545294e-01 2.97206134e-01
7.41995335e-01 5.83406985e-01 -5.23681581e-01 -4.69900489e-01
-2.96226293e-01 -1.63379475e-01 8.89337957e-01 7.11034954e-01
-8.91500354e-01 1.02512658e+00 5.59340620e+00 1.20241547e+00
-1.41455364e+00 5.97293600e-02 4.28548694e-01 -4.74219210e-02
-4.52722669e-01 1.87546119e-01 -5.85217535e-01 7.53832161e-01
9.15112555e-01 -5.21395683e-01 4.55454707e-01 5.18564641e-01
2.34115034e-01 3.75703931e-01 -7.65981257e-01 7.19217718e-01
1.97175607e-01 -1.31972528e+00 4.18167919e-01 2.80052215e-01
4.79111016e-01 -3.99881870e-01 5.32347977e-01 -1.67431653e-01
1.65501058e-01 -6.73708081e-01 9.50223923e-01 2.12306395e-01
1.11263299e+00 -7.95105100e-01 2.18768969e-01 2.40898222e-01
-8.72161508e-01 -1.96131364e-01 -2.78049469e-01 4.63612437e-01
5.47016524e-02 3.81492555e-01 -7.69278944e-01 5.38948119e-01
1.80885494e-01 4.25162688e-02 -5.64398050e-01 5.39327204e-01
-3.57708633e-01 8.30576599e-01 -1.97431371e-01 2.46215448e-01
2.21370459e-01 2.17241630e-01 1.02552760e+00 1.11552429e+00
4.59012419e-01 -1.74780235e-01 -1.24805532e-01 6.32626712e-01
-2.96785653e-01 -9.70421582e-02 -1.03395903e+00 -7.94665292e-02
5.65171957e-01 8.48143339e-01 -3.64957094e-01 -1.85883418e-01
-4.30916362e-02 1.67536712e+00 -1.33829460e-01 5.70011437e-01
-1.04066169e+00 -6.78146958e-01 6.31865978e-01 -5.37589192e-02
4.66352046e-01 -3.13180417e-01 -1.79417687e-03 -1.24757493e+00
1.25573650e-01 -1.48645568e+00 5.42500466e-02 -7.51687288e-01
-1.06333256e+00 6.14914119e-01 -1.44810021e-01 -1.15690362e+00
-1.88761681e-01 -2.17845529e-01 -6.01909995e-01 7.49665022e-01
-1.17437005e+00 -1.16999722e+00 9.24111232e-02 1.01232886e+00
3.96087885e-01 -1.19523399e-01 6.53253376e-01 1.57735243e-01
-4.33775812e-01 1.33213568e+00 3.24245453e-01 4.15183336e-01
9.14699733e-01 -9.03069377e-01 8.58182192e-01 1.27826452e+00
1.98172569e-01 6.17731392e-01 9.34789360e-01 -6.56493485e-01
-1.84144795e+00 -1.08946908e+00 4.25135672e-01 -4.35556620e-01
9.62655246e-01 -6.93964005e-01 -7.23246872e-01 5.92804193e-01
3.21275055e-01 -8.73796716e-02 4.29223269e-01 -9.10248876e-01
-8.33040059e-01 -1.08746678e-01 -1.08746588e+00 9.77944553e-01
7.55702674e-01 -9.02770460e-01 -1.01554833e-01 3.58994782e-01
1.16041195e+00 -4.49372351e-01 -6.33592486e-01 2.33888254e-01
4.60751355e-01 -9.11317766e-01 1.11288071e+00 -2.98473299e-01
5.67095459e-01 -2.39723578e-01 -4.13612843e-01 -1.05928671e+00
-1.35902911e-02 -1.31853867e+00 -1.52756155e-01 1.15284145e+00
3.25332999e-01 -8.08945358e-01 5.66229880e-01 1.31321892e-01
2.02885255e-01 -4.54302132e-01 -1.05789590e+00 -8.87776792e-01
3.11915785e-01 -4.17061150e-01 4.34007108e-01 8.15631151e-01
-2.95091867e-01 1.32800013e-01 -1.02679193e+00 5.98305643e-01
5.91472089e-01 -1.92854762e-01 1.14323068e+00 -3.06502163e-01
-4.67534959e-01 -2.34683856e-01 -5.13167322e-01 -1.26002657e+00
-2.33569536e-02 -6.69102073e-01 -2.13672504e-01 -8.02399397e-01
-4.96016182e-02 -1.17427386e-01 9.36099589e-02 1.32078469e-01
-2.81702608e-01 4.55613136e-01 6.78395450e-01 4.04708982e-01
-1.76032826e-01 4.54993933e-01 1.09081650e+00 -2.47683451e-01
9.23604816e-02 -4.18551080e-02 -7.70776272e-01 2.99261540e-01
8.53797257e-01 -7.32743740e-01 -7.04583049e-01 -4.67202276e-01
3.62682641e-01 2.49612182e-01 4.43491995e-01 -9.10429835e-01
2.10857108e-01 -3.37352306e-02 -7.20036626e-02 -2.02151135e-01
2.18757287e-01 -8.40499759e-01 7.28081316e-02 5.67906380e-01
-5.60645580e-01 1.83190390e-01 2.09648132e-01 6.86718881e-01
2.72159055e-02 -2.52037495e-01 7.97303259e-01 1.74400896e-01
-1.13662735e-01 3.23081911e-01 -4.81159210e-01 2.03395292e-01
9.83076692e-01 -7.40108266e-02 -7.51679957e-01 -8.01296353e-01
-3.49665374e-01 -2.37186804e-01 6.79343879e-01 3.82639080e-01
7.92711616e-01 -1.18081975e+00 -9.41648781e-01 2.28776187e-01
9.96210724e-02 -6.96182668e-01 3.25536907e-01 6.84295297e-01
-5.01881301e-01 2.13580936e-01 1.99386626e-01 -3.21823686e-01
-1.40203428e+00 1.01837981e+00 3.56946707e-01 -1.28041580e-01
-7.23076046e-01 6.95756257e-01 4.34000522e-01 1.75155662e-02
2.04358444e-01 8.08245987e-02 4.58536267e-01 -2.47472852e-01
6.79948866e-01 8.52816179e-02 -1.61123529e-01 -5.92665792e-01
-1.94718823e-01 5.06762087e-01 -2.72927642e-01 -4.70946282e-01
5.61492920e-01 -3.39143217e-01 1.20821416e-01 -2.47012246e-02
1.56999350e+00 6.97611868e-01 -1.06479120e+00 -2.14687213e-01
-4.30467457e-01 -8.13375950e-01 -2.00011149e-01 -6.06862605e-01
-1.11602509e+00 8.70302081e-01 7.06890941e-01 5.95393300e-01
1.18108630e+00 -3.78738791e-01 1.09521532e+00 3.21458519e-01
1.60665661e-01 -5.85117698e-01 3.33975226e-01 2.33506337e-01
1.12680101e+00 -7.33668327e-01 -2.51693279e-01 -3.07726860e-01
-5.87967098e-01 9.83627379e-01 1.33138150e-01 2.81270929e-02
4.74441230e-01 3.85611594e-01 2.91897446e-01 4.15643901e-02
-6.87984765e-01 3.90001565e-01 -4.85488288e-02 5.77421188e-01
-1.98157817e-01 -1.84095591e-01 -8.23847111e-03 1.00889660e-01
-3.31029713e-01 -3.51136714e-01 8.87257397e-01 8.57209325e-01
-2.44120032e-01 -1.15809524e+00 -7.27521718e-01 5.14768846e-02
-8.27353477e-01 -4.16526765e-01 -6.34046555e-01 4.07233864e-01
-3.33082378e-01 1.14514613e+00 -4.98986572e-01 -7.55392075e-01
2.93580234e-01 -1.48092568e-01 1.58645466e-01 -4.00474146e-02
-4.65374649e-01 8.69549159e-03 -7.86809325e-02 -5.07043123e-01
1.70320511e-01 -1.56430125e-01 -7.34349489e-01 -8.02490473e-01
-3.34239066e-01 8.14789161e-02 5.82111120e-01 5.03972113e-01
6.32838845e-01 2.25021794e-01 1.32704806e+00 -4.25697118e-01
-9.16175187e-01 -6.14285290e-01 -3.26879054e-01 4.87247229e-01
6.90450668e-01 -3.89452241e-02 -8.93620729e-01 6.08058348e-02]
|
[5.6962480545043945, 7.867449760437012]
|
eef3be21-625d-4685-816d-c53d59551bf0
|
messy-estimation-maximum-entropy-based
|
2306.04120
| null |
https://arxiv.org/abs/2306.04120v1
|
https://arxiv.org/pdf/2306.04120v1.pdf
|
MESSY Estimation: Maximum-Entropy based Stochastic and Symbolic densitY Estimation
|
We introduce MESSY estimation, a Maximum-Entropy based Stochastic and Symbolic densitY estimation method. The proposed approach recovers probability density functions symbolically from samples using moments of a Gradient flow in which the ansatz serves as the driving force. In particular, we construct a gradient-based drift-diffusion process that connects samples of the unknown distribution function to a guess symbolic expression. We then show that when the guess distribution has the maximum entropy form, the parameters of this distribution can be found efficiently by solving a linear system of equations constructed using the moments of the provided samples. Furthermore, we use Symbolic regression to explore the space of smooth functions and find optimal basis functions for the exponent of the maximum entropy functional leading to good conditioning. The cost of the proposed method in each iteration of the random search is linear with the number of samples and quadratic with the number of basis functions. We validate the proposed MESSY estimation method against other benchmark methods for the case of a bi-modal and a discontinuous density, as well as a density at the limit of physical realizability. We find that the addition of a symbolic search for basis functions improves the accuracy of the estimation at a reasonable additional computational cost. Our results suggest that the proposed method outperforms existing density recovery methods in the limit of a small to moderate number of samples by providing a low-bias and tractable symbolic description of the unknown density at a reasonable computational cost.
|
['Nicolas G. Hadjiconstantinou', 'Kamal Youcef-Toumi', 'Mohsen Sadr', 'Tony Tohme']
|
2023-06-07
| null | null | null | null |
['symbolic-regression', 'density-estimation']
|
['knowledge-base', 'methodology']
|
[-2.11060062e-01 6.76349327e-02 -1.46757707e-01 4.14545536e-02
-9.32421088e-01 -2.01227367e-01 6.48219645e-01 3.67094457e-01
-3.90168369e-01 1.23380339e+00 -5.21448731e-01 -1.79866195e-01
-2.42288351e-01 -8.36299539e-01 -8.52387667e-01 -9.87906635e-01
5.27101569e-02 9.51162696e-01 1.51346281e-01 -1.38912261e-01
4.50148135e-01 7.73998499e-01 -1.40130115e+00 -6.17695928e-01
1.08935750e+00 1.36667192e+00 -2.02532604e-01 5.17307997e-01
-1.23622455e-01 4.25151139e-01 -4.42033619e-01 -2.17187330e-01
2.98865177e-02 -5.05922019e-01 -7.12904990e-01 -1.52276129e-01
-1.30682727e-02 -2.19934702e-01 2.24714205e-01 1.34558678e+00
4.22614157e-01 5.41278183e-01 1.13095319e+00 -9.89657402e-01
-7.91598707e-02 2.38044247e-01 -4.14985180e-01 -3.50167155e-02
1.49590090e-01 -1.12417918e-02 5.82572401e-01 -9.52612400e-01
6.54174209e-01 8.19068789e-01 7.76233375e-01 1.38886452e-01
-1.53124213e+00 -3.97481710e-01 -5.62313259e-01 6.19226173e-02
-1.82156265e+00 -3.38182777e-01 8.10701013e-01 -4.81435508e-01
6.08433962e-01 -2.59379670e-02 7.00301230e-01 5.66573799e-01
1.69457033e-01 1.47988722e-01 9.39718604e-01 -5.15906334e-01
8.00418437e-01 5.27736127e-01 1.84971526e-01 8.70609879e-01
4.37451839e-01 -1.90714091e-01 -2.12403819e-01 -6.37197375e-01
6.90190315e-01 -2.32586011e-01 -2.31894299e-01 -3.32616508e-01
-6.40791655e-01 9.43486989e-01 7.18326345e-02 3.52723062e-01
-6.37532175e-01 3.20241988e-01 8.55696276e-02 -1.61376193e-01
6.48536444e-01 3.11205983e-01 4.49205935e-02 -4.31764007e-01
-1.31735539e+00 3.14340264e-01 1.16426551e+00 6.28954768e-01
1.07320976e+00 1.17519513e-01 5.45834079e-02 5.13804615e-01
4.52685915e-02 1.01753843e+00 2.60110915e-01 -8.69180202e-01
1.72858670e-01 3.50533128e-01 6.56172574e-01 -8.22531343e-01
-7.05089048e-02 -1.60126746e-01 -7.81029820e-01 1.46710142e-01
9.08331513e-01 -1.20100729e-01 -4.12076563e-01 1.46250367e+00
6.57792032e-01 2.52928227e-01 -6.28978014e-02 6.75508559e-01
-2.26145685e-02 7.39647746e-01 -3.55623454e-01 -5.15949547e-01
9.77217972e-01 -3.67553651e-01 -5.30199349e-01 4.47488338e-01
4.09285486e-01 -4.85032439e-01 8.13598752e-01 4.15903211e-01
-1.24709272e+00 -1.29331961e-01 -1.02585161e+00 1.95020869e-01
-1.09293289e-03 2.57156312e-01 1.82643026e-01 6.54835701e-01
-7.98424840e-01 1.23091352e+00 -1.02305496e+00 1.36850372e-01
2.39610508e-01 3.33087862e-01 -8.87224898e-02 2.78275520e-01
-8.28112245e-01 7.75256217e-01 1.38317376e-01 3.84949803e-01
-6.82753384e-01 -8.18114102e-01 -5.81898689e-01 3.34290564e-01
4.40965369e-02 -4.45180744e-01 7.82356143e-01 -8.88925850e-01
-1.72745526e+00 2.08218738e-01 -4.25520808e-01 -5.44126689e-01
5.39685845e-01 8.91324729e-02 1.03270588e-02 5.45562804e-01
-7.45097622e-02 2.05362998e-02 1.23859823e+00 -8.24347496e-01
8.51131827e-02 -1.04960643e-01 -4.23016042e-01 -2.31677681e-01
-3.39812264e-02 -5.78869104e-01 2.27490813e-01 -1.25424387e-02
-8.58465061e-02 -8.56716037e-01 -8.80289599e-02 -8.43829513e-02
-3.31984699e-01 -5.05506285e-02 3.98080796e-01 -6.64802790e-01
9.83794272e-01 -2.05819798e+00 2.27274507e-01 6.96052134e-01
3.64404730e-02 -1.32892430e-01 5.83956242e-01 5.14635086e-01
1.58004820e-01 1.25532910e-01 -5.78143835e-01 -4.34440285e-01
-6.73771128e-02 -1.93115667e-01 -2.82404900e-01 9.60205078e-01
2.09390178e-01 3.98278534e-01 -8.36691856e-01 -5.07138252e-01
1.18556298e-01 7.18717873e-01 -5.46352923e-01 1.50820374e-01
-2.52610773e-01 3.29703510e-01 -4.29200411e-01 3.11298013e-01
7.98371017e-01 -4.32569653e-01 7.60649815e-02 -4.30593677e-02
-1.11103855e-01 1.29148260e-01 -1.51838243e+00 1.08964741e+00
-6.47637069e-01 4.04217303e-01 1.12494200e-01 -1.16347349e+00
1.20559227e+00 1.18641339e-01 4.46427345e-01 4.45491299e-02
3.29465032e-01 7.19178200e-01 -2.20161304e-01 -1.06175438e-01
3.32500398e-01 -7.03784227e-01 1.04227833e-01 1.92831859e-01
1.71460643e-01 -2.53054708e-01 2.86657125e-01 7.98176825e-02
8.61580133e-01 -4.39735688e-02 4.82322901e-01 -5.66254437e-01
8.80163670e-01 -2.94722795e-01 -3.93947102e-02 4.76854593e-01
2.65999228e-01 2.55827874e-01 9.67618644e-01 8.33008625e-03
-1.13507795e+00 -7.66879380e-01 -5.08024335e-01 1.05005808e-01
1.97821736e-01 -4.50018346e-02 -8.70722711e-01 -1.91877037e-01
1.76397547e-01 9.18081462e-01 -6.71013534e-01 -3.65678295e-02
-3.87151808e-01 -5.71431100e-01 2.55267233e-01 9.81683098e-03
5.90549588e-01 -8.71487916e-01 -4.48049098e-01 6.10571168e-02
6.33301735e-02 -8.53831410e-01 -1.26771629e-01 1.23198591e-01
-1.08748412e+00 -8.89339209e-01 -9.23600554e-01 -4.24435139e-01
7.32409775e-01 -6.95028424e-01 6.12095356e-01 -1.36457086e-02
7.09915184e-04 3.44388574e-01 1.86926767e-01 3.66458863e-01
-6.42112851e-01 1.03990160e-01 -9.22748297e-02 3.08914781e-01
-1.12727366e-01 -7.32884526e-01 -3.96081775e-01 1.46920279e-01
-6.11400604e-01 -3.38490427e-01 -1.60108842e-02 9.87477005e-01
5.74131370e-01 1.28722057e-01 5.65001249e-01 -3.63948375e-01
6.85685694e-01 -7.49410391e-01 -1.28881514e+00 -3.52083817e-02
-5.26187003e-01 8.07931781e-01 9.05824602e-01 -5.63743234e-01
-8.91461611e-01 7.16324523e-02 -6.43970370e-02 -5.19287288e-01
1.87850401e-01 3.96225780e-01 3.16086262e-01 -4.23366696e-01
6.65631831e-01 2.78800577e-01 1.62045181e-01 -5.98463178e-01
1.22767150e-01 4.13843215e-01 3.92454058e-01 -1.06752932e+00
5.63753724e-01 5.02220452e-01 7.89473414e-01 -1.09102237e+00
-2.76347011e-01 -4.05422598e-01 -2.39450261e-01 -2.28434578e-01
3.38395447e-01 -1.29642174e-01 -1.30240631e+00 4.89383817e-01
-1.06800807e+00 -2.26264909e-01 -7.64413118e-01 4.55377877e-01
-6.87428236e-01 5.23163319e-01 -5.35532773e-01 -1.57856727e+00
-3.42477649e-01 -1.21256876e+00 1.05890715e+00 6.09234534e-03
-9.35743228e-02 -1.13015056e+00 1.71317652e-01 -3.33503276e-01
4.11242306e-01 3.59125644e-01 8.86329174e-01 -4.87811714e-01
-5.61130524e-01 -4.68762219e-01 6.18944839e-02 3.65032345e-01
-2.55226791e-01 8.83853957e-02 -4.94844645e-01 -5.95216192e-02
4.85239297e-01 -1.01433419e-01 5.06644905e-01 5.70093930e-01
6.44936025e-01 -4.54931676e-01 -2.97829449e-01 5.51622570e-01
1.57872057e+00 -4.24240716e-02 4.94627118e-01 -6.49942234e-02
2.96338826e-01 3.74093294e-01 3.61312658e-01 7.95318365e-01
-2.48300716e-01 5.76051772e-01 2.94920541e-02 4.28009421e-01
2.58941859e-01 -2.49768287e-01 2.65554726e-01 5.97755134e-01
-1.49917483e-01 1.65568978e-01 -7.98757315e-01 5.53817332e-01
-1.62822914e+00 -9.64955270e-01 -1.53080346e-02 2.75464487e+00
1.02790940e+00 8.01307410e-02 4.06811208e-01 3.24455410e-01
7.62532473e-01 -4.05054569e-01 -4.79702860e-01 -4.06421989e-01
2.41157860e-01 6.19074225e-01 5.96168220e-01 1.13494015e+00
-4.96515751e-01 3.65669280e-01 5.81988287e+00 1.14082623e+00
-1.30538118e+00 5.17212041e-02 5.08583009e-01 3.11457086e-02
-2.18179256e-01 1.96954697e-01 -8.76894951e-01 8.77520561e-01
1.15976536e+00 -3.55169028e-01 6.16009176e-01 1.02874994e+00
-2.17906367e-02 -5.73367536e-01 -7.13013232e-01 8.46781552e-01
-2.08173767e-01 -1.34879851e+00 -4.52491760e-01 1.31840631e-01
6.11290574e-01 -4.28606898e-01 -2.99644396e-02 -4.28786688e-02
-3.74321252e-01 -8.57295275e-01 6.54935896e-01 9.26123023e-01
7.18455255e-01 -1.04480612e+00 7.93174863e-01 5.13833821e-01
-1.06408882e+00 2.23507941e-01 -1.24822550e-01 -4.45135720e-02
3.92440170e-01 9.96002614e-01 -9.07580376e-01 1.96215063e-01
1.20195292e-01 2.90566862e-01 -1.29594561e-02 9.29185510e-01
4.00062501e-02 8.00382972e-01 -9.77690697e-01 -6.58320963e-01
7.52286091e-02 -8.51812720e-01 7.35969782e-01 8.02748799e-01
6.96586072e-01 -2.38058880e-01 -2.32122287e-01 1.30291486e+00
1.09992221e-01 2.31377363e-01 -3.26426685e-01 -1.88149974e-01
3.88847291e-01 9.45887208e-01 -9.60008442e-01 -3.93587559e-01
2.55945206e-01 7.15944886e-01 3.82589936e-01 6.01748228e-01
-6.90105975e-01 -5.33261955e-01 1.24151476e-01 4.68924820e-01
5.40210128e-01 -3.62447768e-01 -2.49390870e-01 -9.76903439e-01
1.17362976e-01 -3.63451988e-01 -1.14868641e-01 -4.26957101e-01
-9.25583780e-01 4.22142476e-01 4.27365005e-01 -7.78837681e-01
-7.64618754e-01 -3.22690576e-01 -3.16707194e-01 1.13421392e+00
-1.20150733e+00 -3.99041533e-01 1.19144946e-01 3.95154208e-01
-1.99307859e-01 -6.86014742e-02 6.19345188e-01 7.89380670e-02
-2.81311184e-01 3.12170476e-01 5.76645315e-01 -4.36095089e-01
9.18814912e-02 -1.07396078e+00 -2.12866709e-01 5.47069013e-01
-3.77754629e-01 7.94446766e-01 9.86775219e-01 -8.51301491e-01
-1.12404454e+00 -2.56519854e-01 8.15077603e-01 2.05448121e-01
8.77857327e-01 -3.73062670e-01 -1.07528639e+00 1.99888155e-01
-3.04543912e-01 1.78280070e-01 6.01204447e-02 -8.51302519e-02
2.42706954e-01 -1.29092827e-01 -1.59428489e+00 1.90911427e-01
1.89023584e-01 -4.01393980e-01 -4.51726727e-02 3.60585392e-01
2.15765461e-01 -3.30250978e-01 -8.91428411e-01 6.00461960e-02
4.31374401e-01 -8.72228026e-01 7.53346324e-01 1.77936554e-02
2.46593609e-01 -3.51227731e-01 7.91665763e-02 -9.50639904e-01
1.66841745e-01 -1.19800842e+00 -4.36081022e-01 1.06780374e+00
1.64890036e-01 -1.00103295e+00 6.30590737e-01 5.89160800e-01
4.23587322e-01 -1.09448242e+00 -1.29436445e+00 -9.04950082e-01
7.30015635e-02 -2.71221876e-01 4.29137170e-01 4.02912676e-01
1.82085648e-01 9.80581865e-02 -3.14418525e-01 -5.61295217e-03
7.13573039e-01 -1.85203422e-02 4.98751938e-01 -1.10982192e+00
-6.57260895e-01 -3.12392682e-01 -5.39066434e-01 -9.82065260e-01
3.11345667e-01 -7.33225346e-01 5.73795252e-02 -9.61453795e-01
-1.56286135e-01 -6.15664840e-01 3.06984216e-01 -1.34435192e-01
2.18503654e-01 -1.80794090e-01 -2.24241927e-01 1.86331853e-01
-8.15098658e-02 7.57181883e-01 8.89962196e-01 2.96993792e-01
-2.79063255e-01 2.29677007e-01 6.36849478e-02 7.27925539e-01
6.05396926e-01 -7.23089278e-01 -2.52949893e-01 7.01889873e-01
4.60110426e-01 5.79746723e-01 4.75226164e-01 -1.03690517e+00
1.63886756e-01 1.81405433e-02 9.27616358e-02 -4.30308044e-01
5.98146081e-01 -5.42130053e-01 2.54610002e-01 3.02664280e-01
-9.33210924e-02 -3.44990641e-01 1.78695410e-01 5.57460666e-01
-8.29465315e-02 -1.14813936e+00 9.95682061e-01 3.64956796e-01
2.72343159e-01 -1.17348477e-01 -3.73545557e-01 1.02872394e-01
6.86073720e-01 1.08480342e-01 4.08502445e-02 -6.12478495e-01
-6.76186264e-01 -3.45627874e-01 4.96784806e-01 -7.48111427e-01
4.73709553e-01 -1.26255417e+00 -1.80689022e-01 1.45604685e-01
-6.82101429e-01 -4.59744669e-02 1.28660232e-01 1.11344910e+00
-6.94954693e-01 1.01901732e-01 2.11230963e-01 -5.59577405e-01
-5.42414248e-01 3.08059007e-01 6.80231631e-01 -3.78998816e-01
-3.50666583e-01 3.29611033e-01 -4.91103262e-01 4.79638986e-02
9.31548327e-03 -5.88973343e-01 2.05325246e-01 -3.80613320e-02
2.47824267e-01 9.15078521e-01 1.45333150e-04 -5.30917704e-01
-2.93220520e-01 6.63232744e-01 5.12197673e-01 -5.20440757e-01
1.01230955e+00 2.47292548e-01 -1.89763561e-01 6.44204259e-01
1.49781966e+00 2.87488818e-01 -1.08355403e+00 1.72073126e-01
-2.31661364e-01 -2.14003414e-01 -5.57692489e-03 -3.56322527e-01
-7.52788246e-01 7.94370294e-01 4.01743710e-01 4.28289324e-01
8.02316546e-01 -1.15820721e-01 6.48098886e-01 4.13372308e-01
2.40203783e-01 -8.68046701e-01 -2.25226909e-01 2.82035351e-01
6.95915222e-01 -8.40088665e-01 2.03910902e-01 -2.87714154e-01
-1.92486539e-01 1.40861785e+00 2.63117794e-02 -5.80750048e-01
8.92287135e-01 2.31657326e-01 -8.60711038e-01 -1.15586417e-02
-1.41081885e-01 2.64157504e-02 4.28905845e-01 1.78524807e-01
1.49067000e-01 -1.51160374e-01 -4.76740360e-01 4.53835964e-01
-1.37127489e-01 2.06489816e-01 4.09440339e-01 5.06413698e-01
-4.27268505e-01 -9.45401192e-01 -3.87574613e-01 2.81374395e-01
-2.13013068e-01 -7.75672421e-02 1.50617361e-01 8.96277428e-01
-2.83745766e-01 4.84483182e-01 1.33473631e-02 3.62531960e-01
1.16605088e-02 4.28357869e-01 5.95728219e-01 -1.18658081e-01
-2.83928514e-01 5.82320765e-02 4.77909856e-02 -3.88116509e-01
-2.31688738e-01 -7.13447452e-01 -1.41004920e+00 -5.07526755e-01
-4.72984523e-01 5.92380762e-01 8.37644517e-01 1.12415445e+00
1.73773944e-01 -6.24726303e-02 4.43380088e-01 -9.99498188e-01
-8.54730904e-01 -8.79552305e-01 -8.25374901e-01 1.29245058e-01
4.56539690e-01 -8.54031563e-01 -9.52150643e-01 -3.50659817e-01]
|
[6.5835137367248535, 3.874798059463501]
|
99fbacbf-1033-4be8-b2ac-71b36d2da054
|
selfme-self-supervised-motion-learning-for
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Fan_SelfME_Self-Supervised_Motion_Learning_for_Micro-Expression_Recognition_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Fan_SelfME_Self-Supervised_Motion_Learning_for_Micro-Expression_Recognition_CVPR_2023_paper.pdf
|
SelfME: Self-Supervised Motion Learning for Micro-Expression Recognition
|
Facial micro-expressions (MEs) refer to brief spontaneous facial movements that can reveal a person's genuine emotion. They are valuable in lie detection, criminal analysis, and other areas. While deep learning-based ME recognition (MER) methods achieved impressive success, these methods typically require pre-processing using conventional optical flow-based methods to extract facial motions as inputs. To overcome this limitation, we proposed a novel MER framework using self-supervised learning to extract facial motion for ME (SelfME). To the best of our knowledge, this is the first work using an automatically self-learned motion technique for MER. However, the self-supervised motion learning method might suffer from ignoring symmetrical facial actions on the left and right sides of faces when extracting fine features. To address this issue, we developed a symmetric contrastive vision transformer (SCViT) to constrain the learning of similar facial action features for the left and right parts of faces. Experiments were conducted on two benchmark datasets showing that our method achieved state-of-the-art performance, and ablation studies demonstrated the effectiveness of our method.
|
['Hong Yan', 'Ali Raza Shahid', 'Mingjie Jiang', 'Xueli Chen', 'Xinqi Fan']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['micro-expression-recognition']
|
['computer-vision']
|
[-5.42624630e-02 -6.71120435e-02 -1.18217513e-01 -5.71651340e-01
-3.35718840e-01 -2.68048465e-01 6.46840990e-01 -9.96262014e-01
-4.78299022e-01 4.32564914e-01 2.49168947e-01 1.64277792e-01
3.02705139e-01 -2.17970312e-01 -3.37515593e-01 -6.83895290e-01
-2.89748628e-02 -2.41884053e-01 3.46681885e-02 -1.63602740e-01
3.93176347e-01 8.20890665e-01 -1.47965121e+00 3.44759852e-01
3.78029674e-01 8.28646243e-01 -4.21987832e-01 4.06138957e-01
1.63021281e-01 1.07177699e+00 -3.49573582e-01 -4.61069345e-01
2.85160065e-01 -7.63837337e-01 -8.29464555e-01 1.07517727e-01
7.57739544e-01 -7.70177782e-01 -5.57149112e-01 8.26692045e-01
5.15321970e-01 9.47310999e-02 6.55623674e-01 -1.56388605e+00
-4.66733724e-01 -1.80768579e-01 -1.07331991e+00 3.68173689e-01
3.97005707e-01 3.24084193e-01 5.83209574e-01 -1.04060948e+00
1.01779699e+00 1.33235598e+00 5.78518450e-01 1.20402932e+00
-8.41605246e-01 -9.18819487e-01 -1.88957170e-01 3.85768741e-01
-1.18650067e+00 -1.07944858e+00 1.19150341e+00 -4.94244665e-01
7.39380300e-01 2.80342922e-02 6.41267121e-01 1.41246259e+00
2.26541325e-01 9.21190619e-01 1.17450392e+00 -3.05340260e-01
1.10349335e-01 -5.95262647e-02 -1.64091885e-01 1.04859591e+00
-2.26811394e-01 2.63632983e-01 -8.21713209e-01 -1.01898246e-01
8.09198439e-01 -2.84270853e-01 -1.67514279e-01 -3.85941535e-01
-1.00472736e+00 7.73096025e-01 8.66978019e-02 3.56589347e-01
-2.46831805e-01 2.33504221e-01 3.79459620e-01 1.65004551e-01
5.43002546e-01 -3.14523615e-02 1.02399765e-02 -5.29667497e-01
-1.22198427e+00 1.59341730e-02 5.15561163e-01 3.00919414e-01
5.34237087e-01 2.46472225e-01 -1.40064210e-01 5.05655169e-01
2.30252013e-01 1.75076708e-01 4.41600859e-01 -1.22746384e+00
1.82123289e-01 3.44032079e-01 7.81636015e-02 -1.32986701e+00
-5.12543201e-01 1.81295350e-01 -7.76359022e-01 5.20249307e-01
5.02520144e-01 -2.82920152e-01 -6.65534139e-01 1.88426042e+00
4.59020585e-01 3.99996638e-01 -1.52546435e-03 1.16809118e+00
7.47834802e-01 3.84029478e-01 -1.87782589e-02 -3.63744974e-01
8.83211493e-01 -7.77427316e-01 -9.25772190e-01 -4.51165996e-02
6.28167331e-01 -5.42384207e-01 6.90726280e-01 2.93543905e-01
-8.56377900e-01 -2.63647467e-01 -7.38603711e-01 7.73699731e-02
3.93044464e-02 4.17502522e-01 7.71510720e-01 8.16291273e-01
-7.88055599e-01 6.80574358e-01 -1.11007380e+00 -2.67141193e-01
8.75909448e-01 3.33456755e-01 -8.16944599e-01 2.66718775e-01
-8.83120596e-01 7.02903450e-01 -3.98591280e-01 5.54414392e-01
-7.68263221e-01 -4.70320672e-01 -9.91930127e-01 -3.16145033e-01
1.59212068e-01 -3.53371352e-01 9.76562083e-01 -1.45926845e+00
-1.80877662e+00 1.10347652e+00 -6.75515056e-01 -1.51096910e-01
5.95619142e-01 -4.28423584e-01 -4.89423364e-01 5.40854096e-01
3.97427082e-02 9.93564963e-01 1.36764157e+00 -7.95969903e-01
-1.84668917e-02 -5.35994768e-01 -3.03107709e-01 -9.24846157e-02
-4.10707086e-01 4.98738319e-01 -1.44583374e-01 -6.56648159e-01
-8.79755020e-02 -9.84127402e-01 2.85223991e-01 4.30397660e-01
-9.03061405e-02 -1.13495298e-01 1.42424405e+00 -6.84859216e-01
1.00562215e+00 -2.27963901e+00 4.45640832e-03 -2.14436837e-02
2.02285782e-01 5.81119359e-01 -3.07962656e-01 1.28175125e-01
-2.17530683e-01 -1.52394995e-01 -7.24693611e-02 -4.14510757e-01
-2.44551480e-01 -1.24932520e-01 -2.31717408e-01 9.84784305e-01
4.53606516e-01 1.13802660e+00 -6.76175594e-01 -4.39518660e-01
1.65415883e-01 7.35282600e-01 -5.98917484e-01 1.10632271e-01
1.92205086e-01 7.57639706e-01 -1.79509163e-01 7.33279884e-01
6.67776763e-01 2.04446763e-01 -1.57895282e-01 -2.34881043e-01
2.73390383e-01 -1.38948128e-01 -9.64512467e-01 1.61361718e+00
-1.11451052e-01 1.03689849e+00 2.24858329e-01 -9.34296906e-01
7.56848812e-01 2.90240020e-01 8.56322110e-01 -5.80559015e-01
2.05060497e-01 -3.34928627e-03 1.26122190e-02 -8.96407306e-01
1.30063474e-01 -1.34049594e-01 4.20523673e-01 6.34873807e-01
-1.12680055e-01 2.10858002e-01 -1.95685431e-01 -9.86936539e-02
8.18396389e-01 5.34374714e-01 1.00662440e-01 6.34354353e-02
6.73763692e-01 -2.63929367e-01 8.02903771e-01 8.53973180e-02
-7.95630276e-01 4.58096862e-01 5.02373040e-01 -6.28971100e-01
-6.20560825e-01 -6.42028093e-01 1.62793607e-01 7.58148909e-01
3.13844420e-02 -4.43375498e-01 -9.69550550e-01 -8.98277700e-01
-1.99828893e-01 2.51652658e-01 -6.31944358e-01 -1.11578166e-01
-6.72749579e-01 -5.33800840e-01 7.93221831e-01 5.23720503e-01
7.93356121e-01 -1.19090974e+00 -9.08608794e-01 3.87799479e-02
-2.07354084e-01 -1.47333646e+00 -7.84526646e-01 -8.26265097e-01
-5.59216082e-01 -1.19373310e+00 -7.94588149e-01 -5.79103768e-01
6.67740226e-01 2.77079761e-01 3.51818055e-01 -2.16686249e-01
-6.93467617e-01 4.38886076e-01 -7.48079270e-02 -1.20829478e-01
-1.06666249e-03 -4.89750117e-01 4.18594092e-01 5.91490448e-01
5.84899604e-01 -5.78364909e-01 -7.27500498e-01 4.62892622e-01
-6.40746474e-01 -1.47142544e-01 3.06818932e-01 7.90724456e-01
7.36715412e-03 -2.80114830e-01 4.43587482e-01 -4.33409750e-01
5.64253747e-01 -6.37216195e-02 -3.19999844e-01 -8.19925889e-02
-1.09825812e-01 -2.35466268e-02 3.02766621e-01 -4.57642615e-01
-1.08442283e+00 2.45167524e-01 3.25668603e-02 -6.56558573e-01
-3.79548639e-01 -3.97072956e-02 -1.84738822e-02 -3.44121575e-01
3.96329641e-01 1.55423880e-01 5.20040393e-01 -7.80978650e-02
2.94150203e-01 7.26768076e-01 6.65211082e-01 -2.58655846e-01
5.89449406e-01 1.03342402e+00 4.32655752e-01 -1.14945757e+00
-7.19190121e-01 -4.49709922e-01 -8.18294048e-01 -4.32127416e-01
1.04403710e+00 -6.81333363e-01 -9.30138469e-01 7.69978046e-01
-1.14801776e+00 -1.48029670e-01 2.17140540e-01 6.11899972e-01
-7.08450615e-01 7.98473835e-01 -6.55969024e-01 -9.31841075e-01
-4.44414437e-01 -9.88928556e-01 1.12429547e+00 -1.66396275e-02
-5.36584496e-01 -7.68613458e-01 1.54065907e-01 4.48815495e-01
3.56399596e-01 4.96475786e-01 2.22005546e-01 -3.13275568e-02
-2.87186354e-01 -6.42636046e-02 -5.88999316e-03 2.73930162e-01
1.90558508e-01 1.69460163e-01 -1.15081692e+00 -2.80601263e-01
1.52700935e-02 -4.94720817e-01 8.69758487e-01 3.20268482e-01
1.07202303e+00 -3.46308619e-01 -1.97287366e-01 8.43380988e-01
7.49469578e-01 1.69338528e-02 9.02883351e-01 2.43011028e-01
6.52964711e-01 8.98128211e-01 5.97560346e-01 4.40927893e-01
1.27025843e-01 9.62582231e-01 2.60250479e-01 -3.59997340e-02
-1.22663669e-01 -5.47524095e-02 9.92668986e-01 1.65220603e-01
-3.14806014e-01 2.67253369e-01 -6.25345886e-01 2.92102188e-01
-1.76371574e+00 -1.44276416e+00 1.01789072e-01 1.70795500e+00
5.04238427e-01 -3.12232286e-01 3.27203572e-01 4.88051325e-02
5.28223515e-01 2.66290307e-01 -5.30393541e-01 -3.69986922e-01
1.50577249e-02 2.26604179e-01 -8.08814764e-02 2.92802036e-01
-1.28064501e+00 1.24536324e+00 5.84232044e+00 5.85730672e-01
-1.47262836e+00 5.85133247e-02 4.00264680e-01 -3.22266757e-01
1.82805285e-01 -1.05207354e-01 -7.67685354e-01 3.43029916e-01
5.65541863e-01 2.67006367e-01 3.01896691e-01 6.20926201e-01
6.00577176e-01 -4.92078513e-02 -1.01366496e+00 1.53640974e+00
3.69589776e-01 -1.21784246e+00 -2.52785116e-01 2.17760071e-01
5.11149764e-01 -3.68566364e-01 1.62029788e-01 -1.63271174e-01
-3.43843400e-01 -1.20281756e+00 3.94826800e-01 7.15085864e-01
8.78420770e-01 -7.53143489e-01 4.81955171e-01 2.96953231e-01
-9.92994547e-01 7.39488602e-02 -4.97503467e-02 -1.59960702e-01
1.03441358e-01 3.04844737e-01 -4.22714651e-01 1.95181549e-01
5.35730243e-01 1.14173162e+00 -3.63125116e-01 4.36800927e-01
-2.34007955e-01 6.53332651e-01 -2.06319153e-01 2.06795588e-01
2.96139657e-01 -1.52666956e-01 7.12930560e-01 1.08794975e+00
9.65817422e-02 3.13228779e-02 -3.79395217e-01 1.07058978e+00
7.61197060e-02 -1.74219340e-01 -7.82846093e-01 -1.62264854e-01
-4.63166162e-02 1.49122763e+00 -3.64510596e-01 1.71834722e-01
-4.10518825e-01 1.43650508e+00 2.04654366e-01 4.41534966e-01
-7.53754616e-01 -3.45105410e-01 9.05056477e-01 1.79012641e-01
2.25247458e-01 -5.03668845e-01 1.60478711e-01 -1.50739443e+00
2.51633465e-01 -6.28589094e-01 2.35358760e-01 -6.03439867e-01
-1.04397655e+00 3.21047515e-01 -1.36420488e-01 -1.36211216e+00
-5.82951307e-01 -6.06610894e-01 -8.41878414e-01 5.01627624e-01
-1.41929424e+00 -1.18581653e+00 -3.40402603e-01 9.02230620e-01
3.38337034e-01 -5.80218792e-01 6.49301291e-01 -7.87116308e-03
-6.73111498e-01 8.37484598e-01 -2.74742693e-01 5.68202913e-01
9.08915699e-01 -5.22672415e-01 2.51494586e-01 1.01288581e+00
1.65881366e-01 5.27573764e-01 2.23244891e-01 -4.44674760e-01
-1.49555039e+00 -8.65276694e-01 7.00305462e-01 -3.76833051e-01
4.26667750e-01 -2.93759912e-01 -6.07326150e-01 6.35856926e-01
6.86331987e-02 4.54971671e-01 7.01594532e-01 -5.08920312e-01
-2.92218715e-01 -7.15742558e-02 -1.28880537e+00 6.25509799e-01
1.20882428e+00 -6.48719370e-01 -5.38667023e-01 -1.16829894e-01
-2.41701543e-01 -1.44148573e-01 -7.08147407e-01 4.43988532e-01
8.89125049e-01 -1.25092721e+00 7.83741474e-01 -6.15774632e-01
4.68850315e-01 -1.64654374e-01 1.86903492e-01 -1.01966465e+00
6.00603707e-02 -9.77133274e-01 -4.76333886e-01 1.17609727e+00
-3.61201882e-01 -5.22229314e-01 1.01552629e+00 5.77827096e-01
4.30237442e-01 -7.10746706e-01 -1.05831885e+00 -7.08640933e-01
-7.81695619e-02 -3.81393701e-01 2.38233984e-01 1.07300496e+00
3.08642864e-01 2.17969939e-01 -6.69507802e-01 -9.10423473e-02
6.88818216e-01 9.41455439e-02 1.04043698e+00 -9.08964753e-01
1.35473749e-02 -4.83705819e-01 -8.51078153e-01 -8.28475535e-01
8.21206093e-01 -6.78034365e-01 -4.34279740e-01 -9.60550070e-01
2.20494881e-01 1.95523247e-01 2.60395885e-01 6.25344694e-01
7.84668401e-02 4.16800112e-01 1.68568000e-01 3.58410358e-01
-2.39194915e-01 7.14311063e-01 1.34943962e+00 -9.81222168e-02
-2.16514066e-01 -1.87991187e-01 -4.03658986e-01 9.25895274e-01
7.78839409e-01 -2.40132838e-01 -3.62058669e-01 -8.84085968e-02
-1.27740815e-01 -5.57041876e-02 7.78626204e-01 -8.14080954e-01
3.02653611e-01 -1.97210595e-01 5.90519369e-01 -4.42411125e-01
5.95821500e-01 -4.67823952e-01 -1.50959253e-01 3.78388226e-01
-1.61333129e-01 -2.45508999e-01 2.85030216e-01 3.21832508e-01
-3.33402723e-01 2.22132996e-01 9.39719319e-01 -3.97068681e-03
-1.00270188e+00 4.15071100e-01 -4.89868581e-01 -1.08052947e-01
1.28393328e+00 -3.94829899e-01 -1.87078789e-01 -7.62131035e-01
-6.77306592e-01 -1.73285186e-01 3.24520886e-01 4.60614681e-01
1.03299129e+00 -1.29876745e+00 -7.03453958e-01 5.14920354e-01
-6.88758567e-02 -4.84517455e-01 4.79824916e-02 1.12593913e+00
-3.94367576e-01 3.84156525e-01 -5.16558707e-01 -7.10558057e-01
-1.55695915e+00 3.04731607e-01 6.48522973e-01 2.34862939e-01
-8.40254188e-01 5.73316097e-01 1.18469603e-01 -1.74962536e-01
2.37560179e-02 2.52105266e-01 -3.35349143e-01 -6.41719699e-02
7.34473467e-01 5.59544563e-01 -1.40846103e-01 -1.20140159e+00
-5.81274807e-01 8.96993935e-01 -8.91522393e-02 -2.88284957e-01
1.28243315e+00 -1.47537634e-01 -1.56827271e-01 4.97733662e-03
1.47893417e+00 4.31258045e-02 -1.42795897e+00 8.70536864e-02
-1.29809320e-01 -7.73513615e-01 1.26595899e-01 -2.25270823e-01
-1.35253346e+00 1.07252407e+00 5.97371340e-01 -6.92711413e-01
1.14993227e+00 -2.15063408e-01 9.35974181e-01 3.03774863e-01
3.63256633e-01 -1.01353085e+00 3.75593424e-01 2.80227214e-01
7.73949504e-01 -1.28896701e+00 -1.96104810e-01 -3.79238486e-01
-9.20137942e-01 1.47240114e+00 8.11449647e-01 -2.32042074e-02
6.48440063e-01 9.36084762e-02 1.87058806e-01 -2.54457295e-01
-5.02020419e-01 -8.56064036e-02 4.33419049e-01 5.29167533e-01
3.85364741e-01 -3.41852903e-01 -2.89003402e-01 1.76977292e-01
-8.49288404e-02 4.45635706e-01 4.57172453e-01 1.12765729e+00
-4.95674945e-02 -8.12044680e-01 -2.42297471e-01 2.41485611e-02
-6.17923260e-01 2.82720208e-01 -6.36193514e-01 7.77745426e-01
-1.18851118e-01 9.62065279e-01 7.94695690e-02 -2.96130508e-01
-5.09916879e-02 1.10386781e-01 8.82690072e-01 -1.72446787e-01
-1.98895201e-01 -1.47765186e-02 -1.31392926e-01 -1.15632248e+00
-9.38563526e-01 -6.57737315e-01 -1.16030455e+00 -4.15488929e-01
1.29290923e-01 -2.64981300e-01 3.80850494e-01 9.13482368e-01
5.00886261e-01 -2.09850624e-01 7.90996969e-01 -1.06827796e+00
-4.21776801e-01 -8.44997764e-01 -6.21236324e-01 7.73991227e-01
5.71345150e-01 -8.29347908e-01 -4.33963835e-01 1.81736112e-01]
|
[13.574618339538574, 1.7211593389511108]
|
2a7c0b4b-b8ff-4aa0-a1ad-b0aabc0574a8
|
gan-based-image-compression-with-improved-rdo
|
2306.10461
| null |
https://arxiv.org/abs/2306.10461v1
|
https://arxiv.org/pdf/2306.10461v1.pdf
|
GAN-based Image Compression with Improved RDO Process
|
GAN-based image compression schemes have shown remarkable progress lately due to their high perceptual quality at low bit rates. However, there are two main issues, including 1) the reconstructed image perceptual degeneration in color, texture, and structure as well as 2) the inaccurate entropy model. In this paper, we present a novel GAN-based image compression approach with improved rate-distortion optimization (RDO) process. To achieve this, we utilize the DISTS and MS-SSIM metrics to measure perceptual degeneration in color, texture, and structure. Besides, we absorb the discretized gaussian-laplacian-logistic mixture model (GLLMM) for entropy modeling to improve the accuracy in estimating the probability distributions of the latent representation. During the evaluation process, instead of evaluating the perceptual quality of the reconstructed image via IQA metrics, we directly conduct the Mean Opinion Score (MOS) experiment among different codecs, which fully reflects the actual perceptual results of humans. Experimental results demonstrate that the proposed method outperforms the existing GAN-based methods and the state-of-the-art hybrid codec (i.e., VVC).
|
['Huaxiang Zhang', 'Feng Ding', 'Lili Meng', 'Jian Jin', 'Fanxin Xia']
|
2023-06-18
| null | null | null | null |
['image-compression', 'ms-ssim']
|
['computer-vision', 'computer-vision']
|
[ 3.59903425e-01 -3.45106959e-01 7.81273991e-02 2.63989158e-02
-6.79434240e-01 -5.09870052e-02 8.15054998e-02 -1.64064944e-01
-9.15835276e-02 5.36806643e-01 4.41326946e-01 -1.42800435e-02
1.81307688e-01 -6.36735439e-01 -3.34739655e-01 -9.26416516e-01
2.71158040e-01 -2.32797325e-01 5.29645942e-02 1.45618632e-01
3.62416118e-01 -4.71266285e-02 -1.25932980e+00 1.37291268e-01
1.34374523e+00 1.28551018e+00 1.83208644e-01 5.44495881e-01
6.12127420e-04 1.03048921e+00 -5.95632434e-01 -7.99317300e-01
-1.21396845e-02 -1.02607918e+00 -2.95430034e-01 1.66891113e-01
-2.27006048e-01 -6.70887172e-01 -4.52506304e-01 1.38807154e+00
5.00939548e-01 -1.55477300e-01 7.97971785e-01 -9.93516564e-01
-8.36005926e-01 1.67139977e-01 -8.55021954e-01 -5.73905557e-02
1.17813773e-01 1.28781438e-01 4.24468368e-01 -4.84146148e-01
2.10068345e-01 1.24624205e+00 3.51109743e-01 3.00791144e-01
-9.03123975e-01 -7.13793576e-01 -3.63561600e-01 6.49797559e-01
-1.57754087e+00 -4.19889808e-01 9.29817975e-01 -2.27293715e-01
3.58393967e-01 2.31146559e-01 4.71260250e-01 6.80966556e-01
4.53975886e-01 7.01919496e-01 1.30390882e+00 -3.77384067e-01
4.10172164e-01 6.00082986e-02 -5.80371499e-01 5.41096330e-01
9.80814546e-02 4.32964265e-02 -2.74309963e-01 1.15809187e-01
8.78968716e-01 -1.66138530e-01 -4.28310454e-01 -1.40198112e-01
-9.14863050e-01 5.09359539e-01 2.26909265e-01 1.14842847e-01
-2.62005776e-01 2.81628132e-01 2.95998275e-01 3.58344847e-03
3.80218267e-01 -1.90897226e-01 1.27382159e-01 -5.12917280e-01
-1.09813678e+00 -3.01024467e-01 3.76112193e-01 9.78992462e-01
3.34132910e-01 2.97724932e-01 -3.14700246e-01 1.04512620e+00
5.34940600e-01 5.44376731e-01 6.89189255e-01 -1.16054952e+00
5.30230224e-01 2.35698387e-01 -6.46711290e-02 -1.19314098e+00
4.14977551e-01 -3.93459857e-01 -1.39233088e+00 1.95409387e-01
-3.21470350e-01 1.43710390e-01 -7.90495098e-01 1.45147097e+00
-1.77808046e-01 -3.97755615e-02 6.11848235e-02 7.48646855e-01
5.48488796e-01 9.74072516e-01 -2.78739464e-02 -5.58981955e-01
9.71401811e-01 -1.02085471e+00 -1.13524544e+00 1.74518779e-01
1.99881777e-01 -9.64140654e-01 9.86374080e-01 5.51532924e-01
-1.32187200e+00 -8.02265227e-01 -1.32990456e+00 7.80422101e-03
1.77873209e-01 3.30816150e-01 1.08196855e-01 1.06370139e+00
-9.06535625e-01 4.21598613e-01 -7.22682595e-01 5.25684431e-02
4.06864822e-01 -1.73513219e-02 5.49383201e-02 -3.50309283e-01
-7.77497530e-01 6.18601441e-01 3.12336057e-01 -1.08629547e-01
-8.47754180e-01 -2.36425444e-01 -5.97546220e-01 2.98119724e-01
1.67491566e-03 -3.14860106e-01 7.85739720e-01 -8.24851990e-01
-1.71381605e+00 3.41622412e-01 -1.95132136e-01 -8.48576427e-02
5.05563021e-01 5.73695600e-02 -5.68441033e-01 2.77211279e-01
-3.36281687e-01 5.55142343e-01 8.70282412e-01 -1.43584239e+00
-3.07902426e-01 -1.90310851e-01 -2.68741935e-01 2.58724123e-01
-3.10103029e-01 -4.58909348e-02 -6.79711521e-01 -9.83041465e-01
2.51106977e-01 -6.54199541e-01 1.29407763e-01 2.32477158e-01
-3.86485159e-01 3.10013860e-01 7.83303797e-01 -1.16123509e+00
1.57881081e+00 -2.61860681e+00 1.19074404e-01 9.59851593e-02
5.15574403e-02 2.44061753e-01 -3.41259949e-02 1.67275146e-01
1.65531874e-01 3.49796027e-01 -4.89001244e-01 -3.43304873e-01
-1.35203600e-01 4.28168029e-02 -4.66757864e-02 3.41239691e-01
-2.21846089e-01 6.58043683e-01 -5.01100302e-01 -7.92243183e-01
2.49282196e-01 6.33716226e-01 -5.09984434e-01 3.47347468e-01
2.36881316e-01 4.42586780e-01 -1.99480519e-01 6.86383307e-01
1.12276804e+00 5.43782674e-03 -1.82846952e-02 -5.08363903e-01
1.73785254e-01 -1.67531863e-01 -9.41783428e-01 1.64733350e+00
-5.35718858e-01 4.33130085e-01 -7.38980398e-02 -7.33911097e-01
9.20692444e-01 3.73526067e-01 3.26404482e-01 -9.93113279e-01
1.83417469e-01 2.95014977e-01 -2.43154559e-02 -4.89858687e-01
3.55865747e-01 -6.68438477e-03 2.52613068e-01 2.14875445e-01
-4.89654429e-02 -1.14289381e-01 4.31783311e-02 9.30656269e-02
7.46354342e-01 2.90953256e-02 9.59862918e-02 2.12757420e-02
6.01245940e-01 -6.53667629e-01 6.35822773e-01 1.22443147e-01
-3.27245742e-01 9.60513711e-01 4.58383471e-01 2.63925552e-01
-1.39774191e+00 -1.13344288e+00 -7.30690062e-02 2.92232245e-01
6.38855696e-01 -4.65604156e-01 -9.51427102e-01 -2.24210575e-01
-6.39230967e-01 8.79662275e-01 -1.47486910e-01 -4.29631144e-01
-1.88906625e-01 -6.95424199e-01 5.28478801e-01 2.74985790e-01
1.16470575e+00 -8.03906560e-01 -4.07573432e-01 2.07591355e-02
-6.73874497e-01 -9.49086666e-01 -6.93532705e-01 -3.87918681e-01
-8.99165034e-01 -6.31551027e-01 -1.01141882e+00 -6.34743512e-01
5.83978176e-01 1.07822515e-01 7.16233075e-01 7.10020680e-03
-9.93831381e-02 3.44890393e-02 -6.24385774e-01 6.98585436e-02
-6.29729271e-01 -3.88057530e-01 -3.71822000e-01 9.74630415e-02
-6.28224313e-02 -6.36430144e-01 -9.62940812e-01 4.07600790e-01
-1.17335641e+00 3.27449322e-01 8.01409781e-01 5.87689161e-01
8.40371668e-01 6.22333705e-01 1.98657677e-01 -3.41716319e-01
7.15642691e-01 -3.19004446e-01 -3.72338414e-01 4.95388001e-01
-1.01431918e+00 4.39001806e-02 6.30004883e-01 -2.23992303e-01
-1.22086835e+00 -4.33791488e-01 -2.87033737e-01 -3.28886747e-01
5.73430955e-02 4.21765834e-01 -4.72692668e-01 -1.04444169e-01
9.38986242e-03 7.79392600e-01 -2.36718860e-02 -4.94615197e-01
9.45453942e-02 1.01994824e+00 6.46428645e-01 -2.49165699e-01
6.26922667e-01 1.14120170e-01 -4.64618839e-02 -5.31824529e-01
-1.69394031e-01 -5.80672827e-03 -5.63309826e-02 -3.52853954e-01
8.97096515e-01 -1.08403480e+00 -5.36820769e-01 7.93898761e-01
-1.08629787e+00 -2.73395404e-02 3.61069515e-02 6.84963584e-01
-6.90978348e-01 8.53932619e-01 -8.72901380e-01 -8.89370322e-01
-4.36551154e-01 -1.46556866e+00 8.02131236e-01 2.65393019e-01
3.71182561e-01 -6.09741986e-01 -2.06277296e-01 5.09140134e-01
7.40549564e-01 2.24011928e-01 1.21344292e+00 3.64723772e-01
-5.55024981e-01 -6.11226261e-02 -4.53414649e-01 8.31773400e-01
1.64182559e-01 -9.07977670e-02 -7.38514841e-01 -4.21011567e-01
3.79489839e-01 -2.40940630e-01 6.50149703e-01 5.19683659e-01
1.64191914e+00 -4.27646577e-01 1.29723221e-01 9.29067492e-01
1.68209922e+00 6.30611837e-01 1.45238519e+00 -4.07330357e-02
4.38710868e-01 1.44446552e-01 4.57099855e-01 5.16253710e-01
2.05090880e-01 6.67441249e-01 4.05635625e-01 -2.39677671e-02
-5.84692776e-01 -4.59320396e-01 5.15052378e-01 1.57412100e+00
-1.47894993e-01 -6.83134675e-01 -3.60389590e-01 1.05658591e-01
-1.36315835e+00 -7.76404440e-01 7.46728927e-02 2.27998090e+00
7.64703035e-01 1.03957511e-01 -3.31689328e-01 3.85596037e-01
7.77644217e-01 5.60351834e-02 -6.23720348e-01 -4.38456744e-01
-3.35396886e-01 -4.88215163e-02 4.75166410e-01 2.42458239e-01
-6.93594754e-01 2.54194915e-01 6.16483545e+00 1.47298014e+00
-9.83977616e-01 2.40834400e-01 1.15021396e+00 1.26681328e-01
-3.81285161e-01 -1.12866282e-01 -1.29365534e-01 1.07340503e+00
7.50127673e-01 -1.18726321e-01 8.63538802e-01 4.95304078e-01
1.19409412e-01 -2.81945288e-01 -5.30609012e-01 1.43638098e+00
3.31714690e-01 -7.47005284e-01 1.62866995e-01 2.50627249e-01
8.29678178e-01 -5.20699680e-01 3.85339469e-01 1.26138255e-02
-1.49707690e-01 -9.08570826e-01 8.06538761e-01 5.35054147e-01
1.38065696e+00 -8.01900566e-01 8.13538730e-01 2.08092034e-01
-1.08976388e+00 2.45874319e-02 -5.98137438e-01 4.79928374e-01
2.01449677e-01 7.40943670e-01 1.35972798e-01 5.87356091e-01
5.90159416e-01 5.48748493e-01 -6.12227082e-01 1.10820067e+00
-2.46976629e-01 7.09341764e-01 -5.84976971e-02 2.89140731e-01
-1.67920545e-01 -4.10936147e-01 2.35690087e-01 8.25888872e-01
8.63878429e-01 1.06472582e-01 -4.18336898e-01 1.04938185e+00
-1.56226486e-01 2.13177353e-01 -8.88099372e-02 2.11604014e-02
4.00559515e-01 8.89525175e-01 -5.25127351e-01 -1.90767989e-01
-3.36437881e-01 1.42912340e+00 -3.57949048e-01 4.42896098e-01
-9.41488683e-01 -4.72421825e-01 1.47463307e-01 -1.31284958e-02
2.93845594e-01 -2.74601996e-01 -2.73703635e-01 -1.13769376e+00
1.72390416e-01 -8.69886279e-01 -1.65132716e-01 -1.07233655e+00
-8.81347835e-01 5.46056151e-01 -9.03316364e-02 -1.53945839e+00
4.76402603e-02 -8.05026200e-03 -5.47466338e-01 9.25900757e-01
-1.37934828e+00 -8.56080294e-01 -5.35849333e-01 4.02251124e-01
4.86990094e-01 -2.28151068e-01 5.28248370e-01 6.25334382e-01
-5.83157182e-01 8.38366508e-01 6.21492624e-01 -9.86580104e-02
5.83143711e-01 -6.62879765e-01 2.92716343e-02 1.04252005e+00
-2.44520858e-01 5.24166971e-02 5.21676660e-01 -5.37876010e-01
-1.20293736e+00 -9.52530146e-01 2.95429975e-01 3.59260559e-01
-6.87698722e-02 1.82585511e-02 -7.61838019e-01 7.43372925e-03
3.80499661e-01 -3.05064082e-01 6.95052445e-01 -6.60385907e-01
-1.47337079e-01 -2.25862011e-01 -1.41046703e+00 4.38412696e-01
7.27543771e-01 -5.91793776e-01 1.54363334e-01 -2.37284645e-01
7.29536831e-01 -1.39955610e-01 -9.09147143e-01 5.33552229e-01
4.98613328e-01 -1.35912740e+00 6.60000205e-01 4.67283547e-01
9.48684037e-01 -5.11084974e-01 -5.18648863e-01 -1.19310796e+00
-4.39267039e-01 -1.31700546e-01 -2.15394676e-01 1.49652302e+00
4.81921062e-03 -2.17489779e-01 5.04095376e-01 3.69004428e-01
1.02359936e-01 -7.54122496e-01 -8.90243292e-01 -6.76768243e-01
-2.36959085e-01 -8.36640000e-02 6.59519196e-01 3.08247447e-01
-4.50208336e-02 -3.35030407e-02 -7.22721577e-01 -1.64837793e-01
8.28758776e-01 -1.50113925e-01 2.95036525e-01 -5.67349494e-01
-4.81217235e-01 -3.33504051e-01 -5.97651422e-01 -1.27772093e+00
-4.09206659e-01 -4.79866087e-01 1.07791707e-01 -1.41387188e+00
6.12469673e-01 -3.21229964e-01 -4.36992556e-01 -1.28215462e-01
-1.77554861e-01 3.49619955e-01 2.71822929e-01 5.45048594e-01
-5.68035603e-01 1.08618796e+00 1.37453985e+00 -2.85248309e-01
1.32562429e-01 -3.33385080e-01 -4.99189824e-01 3.67577195e-01
7.95985043e-01 -3.55778337e-01 -6.17995024e-01 -6.76995099e-01
3.28868851e-02 3.52968752e-01 1.81020871e-01 -1.45592320e+00
1.08374715e-01 4.01399359e-02 4.54705834e-01 -5.00327706e-01
3.85363609e-01 -8.57418418e-01 3.71866941e-01 6.42422616e-01
-2.33360693e-01 -7.03444034e-02 -1.74185127e-01 6.13565385e-01
-6.37843490e-01 -1.54589027e-01 1.04507327e+00 7.28073567e-02
-3.64594907e-01 2.12085724e-01 -2.82274008e-01 -2.12947845e-01
8.77797902e-01 -3.16145062e-01 -3.07401091e-01 -8.15838754e-01
-1.33685976e-01 -3.09072465e-01 7.86899447e-01 1.19280249e-01
8.77208889e-01 -1.43502808e+00 -7.62928724e-01 3.31015974e-01
4.77345660e-03 -2.92522997e-01 5.08434653e-01 8.42044353e-01
-8.86564314e-01 6.46690801e-02 -2.76682943e-01 -4.03684229e-01
-1.03556752e+00 5.56076288e-01 -1.92280337e-02 -2.06779093e-01
-1.95791274e-01 5.24024844e-01 2.48110861e-01 3.10566604e-01
1.73757091e-01 -4.92907166e-02 3.02056391e-02 -5.60506344e-01
3.76963913e-01 5.67035675e-01 -1.09779634e-01 -6.32724047e-01
-4.08610031e-02 6.12089455e-01 1.79994360e-01 -1.12093955e-01
8.93227398e-01 -6.17602110e-01 -1.83832720e-01 4.80045266e-02
1.37309361e+00 4.76719290e-02 -1.20819545e+00 9.42332149e-02
-7.13971317e-01 -7.67076313e-01 3.20841402e-01 -8.46922338e-01
-1.42110324e+00 1.06970346e+00 1.24827933e+00 -3.10306307e-02
1.64938331e+00 -3.80821019e-01 1.11641014e+00 -4.18286830e-01
4.17027593e-01 -1.03113079e+00 1.73939422e-01 -9.43800956e-02
8.00697505e-01 -9.35289800e-01 2.01486915e-01 -3.78091991e-01
-7.15059936e-01 9.28633034e-01 3.32287312e-01 1.64933592e-01
6.80314720e-01 1.53678060e-01 -2.43617687e-02 4.13738638e-01
-5.08430064e-01 3.20240229e-01 1.48527309e-01 6.01429164e-01
4.71313536e-01 2.86103990e-02 -5.74075937e-01 3.92006993e-01
-8.79439116e-02 1.37543634e-01 5.90720713e-01 5.71123838e-01
-2.68536896e-01 -1.01356030e+00 -3.24385524e-01 4.06451285e-01
-6.83377445e-01 -1.94105953e-01 1.65596843e-01 -4.98105586e-02
3.31747591e-01 1.21841538e+00 -9.59045961e-02 -7.67070711e-01
-1.19646326e-01 -4.33953345e-01 4.67591852e-01 8.33098292e-02
2.36145243e-01 2.86827624e-01 -4.03667718e-01 -4.60555196e-01
-5.64271092e-01 -2.20638901e-01 -7.89339304e-01 -5.44257283e-01
-3.78681511e-01 1.23925917e-01 9.29148197e-01 6.37426019e-01
3.31391692e-01 5.59550822e-01 8.84136438e-01 -3.46156418e-01
-3.88260156e-01 -1.00832522e+00 -8.14646482e-01 4.30449635e-01
5.41272201e-02 -2.61081368e-01 -2.64635026e-01 3.39538097e-01]
|
[11.404329299926758, -1.7599643468856812]
|
e0f013a1-5b0d-4701-a656-6bb12d4a91a4
|
handwritten-digit-recognition-by-elastic
|
1807.09324
| null |
http://arxiv.org/abs/1807.09324v1
|
http://arxiv.org/pdf/1807.09324v1.pdf
|
Handwritten Digit Recognition by Elastic Matching
|
A simple model of MNIST handwritten digit recognition is presented here. The
model is an adaptation of a previous theory of face recognition. It realizes
translation and rotation invariance in a principled way instead of being based
on extensive learning from large masses of sample data. The presented
recognition rates fall short of other publications, but due to its
inspectability and conceptual and numerical simplicity, our system commends
itself as a basis for further development.
|
['C. von der Malsburg', 'Sagnik Majumder', 'Aashish Richhariya', 'Surekha Bhanot']
|
2018-07-24
| null | null | null | null |
['handwritten-digit-recognition']
|
['computer-vision']
|
[ 1.08643606e-01 8.94557498e-03 -4.44250286e-01 -8.51286590e-01
3.17134634e-02 -4.13604558e-01 1.14809120e+00 -4.72082794e-01
-6.60528243e-01 6.14639461e-01 -2.02935144e-01 -4.40276295e-01
-3.75152975e-01 -4.95711058e-01 -2.73506463e-01 -7.16749072e-01
-1.99932650e-01 7.43320167e-01 1.23021662e-01 -1.19822949e-01
6.16209269e-01 1.32144988e+00 -1.58904862e+00 1.61168575e-01
-6.21088408e-02 1.01157534e+00 -1.41396597e-01 5.33606708e-01
-8.09866711e-02 1.18596590e+00 -3.91242653e-01 -4.31536615e-01
3.24914873e-01 -4.88292813e-01 -1.09087193e+00 4.02194947e-01
4.08856571e-01 -3.98405224e-01 -7.17946291e-01 6.69992030e-01
4.36847806e-01 4.35962901e-03 8.18437994e-01 -9.54849243e-01
-1.17967939e+00 2.22937241e-01 -2.66180992e-01 5.60643673e-01
3.20385903e-01 -2.14589968e-01 7.73293614e-01 -1.09348822e+00
4.37301725e-01 1.26371813e+00 6.10044122e-01 9.44487691e-01
-1.10697234e+00 -4.16384339e-01 -2.48371698e-02 2.41710320e-01
-1.41636789e+00 -9.67332602e-01 6.65274322e-01 -2.36212432e-01
1.29697466e+00 4.40940022e-01 4.67106968e-01 9.41749334e-01
2.86929965e-01 8.47544551e-01 1.14839458e+00 -7.99534917e-01
3.96876663e-01 1.35151640e-01 1.30230293e-01 8.18755269e-01
3.27193081e-01 4.42799509e-01 -5.86249530e-01 -6.50694221e-02
1.27076817e+00 -1.53524624e-02 1.23501062e-01 -7.98556209e-01
-1.11625886e+00 7.56890595e-01 2.08147421e-01 6.01872385e-01
-1.45145684e-01 -3.24830450e-02 1.95449039e-01 7.44071364e-01
-2.03953162e-02 1.46144256e-01 -3.17697346e-01 -6.04155622e-02
-9.28819835e-01 -6.10202849e-02 9.45440412e-01 6.99110448e-01
3.31721842e-01 4.56520081e-01 3.69546324e-01 7.33247042e-01
6.85155034e-01 4.37326878e-01 9.62983787e-01 -8.61856461e-01
1.66210979e-02 4.46671635e-01 -1.67269424e-01 -6.69546425e-01
-3.17844301e-01 -4.11576815e-02 -7.24174976e-01 8.98257017e-01
6.04801893e-01 6.91048205e-02 -1.10113227e+00 1.21359789e+00
-1.67868212e-01 -3.79545033e-01 1.05915070e-01 7.42595732e-01
6.58777595e-01 1.71952888e-01 -7.53405988e-02 -1.13054715e-01
1.05858588e+00 -5.34032583e-01 -7.92310417e-01 -1.41773731e-01
1.38434231e-01 -8.01369309e-01 6.34202540e-01 6.00408673e-01
-8.90807688e-01 -5.41173398e-01 -1.16541195e+00 9.59095806e-02
-5.67425430e-01 2.73593694e-01 1.07824636e+00 1.31983662e+00
-1.24172306e+00 7.84961820e-01 -9.83613491e-01 -7.61763692e-01
4.65458453e-01 9.27895606e-01 -6.36933148e-01 -6.08407333e-02
-5.62310636e-01 1.24453807e+00 5.14060616e-01 3.88405859e-01
-4.47423339e-01 1.70104176e-01 -7.36258209e-01 -3.67049575e-01
3.62273753e-02 -2.62631811e-02 1.32010996e+00 -9.96556222e-01
-2.05696154e+00 1.16937792e+00 -2.53126949e-01 -1.13259465e-01
5.85137010e-01 2.39484265e-01 -7.26023257e-01 2.20343098e-01
-7.05312967e-01 5.10737479e-01 1.13876164e+00 -8.44775856e-01
-3.89533043e-01 -5.35495877e-01 -2.92129904e-01 -2.00616077e-01
-4.05259222e-01 2.94665754e-01 -2.02055853e-02 -7.12251008e-01
6.11222208e-01 -6.07001185e-01 -8.56725033e-03 1.29541293e-01
1.12530343e-01 -2.43415371e-01 7.76249886e-01 -4.47340906e-01
7.18194246e-01 -2.19237351e+00 6.59943232e-03 3.77882808e-01
1.12748839e-01 4.61732388e-01 -2.12348893e-01 5.07517457e-01
-5.17768085e-01 -4.95038070e-02 -9.18557122e-03 -2.00357020e-01
1.54881537e-01 2.81100035e-01 -4.10433173e-01 7.25589633e-01
4.11466569e-01 8.15142930e-01 -6.43469572e-01 -2.11333469e-01
4.08937067e-01 4.76148874e-01 -4.25006524e-02 -2.41573215e-01
3.96392941e-01 -3.33020501e-02 -3.66139501e-01 1.08677924e+00
5.19089520e-01 -1.45725772e-01 3.02280724e-01 2.68868580e-02
-1.08093344e-01 2.10332990e-01 -1.34826338e+00 1.19313228e+00
1.67375252e-01 6.94814801e-01 -1.49716556e-01 -1.31322861e+00
1.33759248e+00 5.23306012e-01 2.96332479e-01 -5.79202175e-01
1.94079667e-01 4.00533736e-01 2.15525955e-01 -3.71846616e-01
1.57470703e-01 -3.88548553e-01 3.64503175e-01 7.59129107e-01
4.28715706e-01 -2.34610010e-02 1.42729759e-01 -1.66135177e-01
6.67372108e-01 2.14854404e-01 5.58991671e-01 -4.28489178e-01
7.71695733e-01 -4.17212069e-01 -9.07721184e-03 6.80597544e-01
-2.74915248e-01 3.61781567e-01 2.04917118e-01 -9.79726255e-01
-9.91972387e-01 -1.03465366e+00 -4.22169775e-01 9.17908728e-01
-3.34973931e-01 1.18107637e-02 -5.29861689e-01 -5.59051156e-01
9.07203779e-02 3.21510375e-01 -7.60280252e-01 1.59131899e-01
-4.34580296e-01 -9.59435582e-01 6.78356111e-01 6.77126348e-01
4.17889684e-01 -1.53111982e+00 -5.21375597e-01 -2.88781598e-02
5.30815125e-01 -7.22738564e-01 1.61131561e-01 3.63272309e-01
-1.20270610e+00 -1.16260767e+00 -9.19443130e-01 -1.08738863e+00
8.03816736e-01 1.88825086e-01 6.20755315e-01 9.55700949e-02
-4.77367848e-01 6.31093502e-01 -1.92222968e-01 -3.60330701e-01
-4.74121690e-01 -8.48416835e-02 5.38287044e-01 6.61317483e-02
9.84886229e-01 -5.30930638e-01 -8.64502937e-02 3.46965998e-01
-7.43859231e-01 -7.24962354e-01 8.52949560e-01 8.52161288e-01
1.01503432e-01 -6.16855994e-02 7.14870989e-01 -5.88314116e-01
6.42214656e-01 6.96660206e-02 -6.40704334e-01 3.64021748e-01
-6.98674858e-01 -3.69408093e-02 1.58667073e-01 -4.05397803e-01
-9.94103849e-01 4.75520462e-01 -1.36599019e-01 4.46511991e-02
-6.40986502e-01 2.15227231e-01 -3.23466435e-02 -7.45849252e-01
8.21318746e-01 6.26557827e-01 4.55327868e-01 -5.96412539e-01
-6.46105483e-02 7.77649760e-01 5.42671561e-01 -2.93799192e-01
7.58845210e-01 3.85710239e-01 6.95499182e-02 -1.17424130e+00
4.72779572e-02 -1.01709468e-02 -1.10147071e+00 -2.21734568e-02
3.16174865e-01 -5.07144690e-01 -1.02997291e+00 6.06403112e-01
-8.84661853e-01 -1.30966708e-01 -3.14704865e-01 8.39584827e-01
-7.05032170e-01 7.38008678e-01 -7.68957138e-01 -1.03116751e+00
-4.44076099e-02 -7.41796315e-01 5.49571276e-01 -7.98604861e-02
-2.26338267e-01 -1.04603374e+00 5.51206851e-03 3.67511921e-02
3.07215244e-01 -1.58967763e-01 5.82512498e-01 -8.14526737e-01
-3.86548817e-01 -8.46037209e-01 -6.99744225e-02 5.69373608e-01
5.50194681e-01 1.28447369e-01 -1.21703434e+00 -5.23997128e-01
2.65086740e-01 -6.60758853e-01 8.27780604e-01 2.98192482e-02
7.81824470e-01 -1.80677325e-01 -9.76857170e-02 1.00882642e-01
1.29744565e+00 4.59731638e-01 7.51512527e-01 4.11919892e-01
1.16971694e-01 4.81703699e-01 1.12944633e-01 2.51641542e-01
-1.56857818e-01 3.39688033e-01 2.32270434e-01 -7.87929818e-03
-1.05131336e-01 1.85133278e-01 2.85502076e-01 5.05279660e-01
-7.32940018e-01 9.19965059e-02 -8.99182856e-01 1.92912832e-01
-1.46757424e+00 -1.30111086e+00 2.18995377e-01 2.16888165e+00
3.36535990e-01 2.12532669e-01 4.30710167e-01 6.47416353e-01
4.81216550e-01 -3.69506665e-02 -3.29158336e-01 -3.27436656e-01
-2.86855936e-01 3.11493456e-01 2.10128903e-01 4.37743962e-01
-1.09024084e+00 7.39923477e-01 9.18516350e+00 4.84837264e-01
-1.18205702e+00 -4.25527006e-01 4.84192848e-01 3.14483136e-01
3.45282942e-01 -4.67258900e-01 -9.22651529e-01 -1.38357431e-01
8.10822785e-01 3.52181345e-02 6.14863634e-01 7.22721159e-01
-4.09640849e-01 5.74502461e-02 -1.56970751e+00 9.11806107e-01
5.15639782e-01 -1.13660109e+00 4.07494098e-01 2.01378062e-01
1.07156314e-01 -9.90333110e-02 2.86876202e-01 1.14174634e-01
1.78109869e-01 -1.28492105e+00 6.03951633e-01 3.41274530e-01
6.15848660e-01 -6.02232695e-01 7.67312348e-01 3.82767439e-01
-7.98718154e-01 -3.03952605e-01 -7.61398137e-01 -6.49209559e-01
-3.87207776e-01 -1.45287300e-03 -7.73362041e-01 2.88331807e-01
2.20539004e-01 4.01752353e-01 -7.10615993e-01 9.61349010e-01
-2.83513796e-02 2.89923698e-01 -2.06658289e-01 -2.78123826e-01
1.03434145e-01 1.07380949e-01 2.22294312e-02 1.28966582e+00
2.82395035e-01 1.42083436e-01 -7.00582266e-02 4.09555554e-01
2.17145056e-01 2.10530370e-01 -7.62161076e-01 -2.66082376e-01
2.40803644e-01 1.05236280e+00 -1.13541698e+00 -3.91965747e-01
-5.95338464e-01 9.08371329e-01 2.91882515e-01 1.03433318e-01
-1.56719118e-01 -3.96477342e-01 1.77571669e-01 -1.58896923e-01
6.94561362e-01 -3.38758290e-01 -3.53848964e-01 -1.21870410e+00
-9.16111097e-02 -8.94547284e-01 2.74215013e-01 -1.87356219e-01
-1.24340236e+00 6.59430563e-01 -1.28618926e-01 -9.92290437e-01
-6.88829422e-01 -1.39770496e+00 -2.91214824e-01 8.16274464e-01
-1.21514928e+00 -1.06289601e+00 1.19984366e-01 7.35620022e-01
4.36679214e-01 -8.70459259e-01 1.47931612e+00 1.03277266e-01
-3.77620995e-01 7.41179585e-01 1.33060470e-01 5.26674271e-01
5.15484989e-01 -9.67487037e-01 5.17618716e-01 5.37813604e-01
5.44392586e-01 8.33016038e-01 6.50422037e-01 -2.53725380e-01
-1.40368044e+00 -5.20818412e-01 1.25633693e+00 -6.07584834e-01
5.17689824e-01 -3.11547041e-01 -6.86904252e-01 9.49786365e-01
6.67766407e-02 5.96296974e-02 8.02091241e-01 1.76741838e-01
-5.76144040e-01 -2.59798020e-01 -1.32517147e+00 3.30639124e-01
8.07495356e-01 -5.82269013e-01 -1.06688666e+00 2.96966851e-01
-2.67711937e-01 -4.39840145e-02 -7.32680202e-01 2.26985663e-01
1.05255914e+00 -9.02847648e-01 1.05120301e+00 -6.96589053e-01
2.67213434e-02 1.02149196e-01 -3.35731298e-01 -5.64466953e-01
-8.19445431e-01 -4.69880104e-01 -2.23409206e-01 9.05890822e-01
2.72289366e-01 -8.53560030e-01 8.26735377e-01 5.84830403e-01
3.55472982e-01 -2.24356383e-01 -1.20480549e+00 -1.32110548e+00
9.02463868e-02 -1.93030760e-01 3.99267018e-01 1.03784180e+00
4.01508719e-01 3.23995888e-01 -3.31118524e-01 -1.22152217e-01
7.91738153e-01 9.64932367e-02 3.84849221e-01 -1.61621833e+00
-2.84379989e-01 -6.54989064e-01 -8.72173071e-01 -8.58868778e-01
8.92965198e-02 -6.65296018e-01 -1.71222895e-01 -9.24604297e-01
5.48260026e-02 -3.44287120e-02 -6.88063502e-01 8.14743817e-01
6.26914084e-01 9.05683100e-01 1.45573676e-01 2.64449358e-01
-2.90187716e-01 1.22299239e-01 7.88171291e-01 -2.26852223e-01
-3.32687050e-02 2.87699938e-01 -4.55091119e-01 8.16654861e-01
1.00178504e+00 -2.12070465e-01 -3.62874389e-01 -5.81536964e-02
-2.77189940e-01 -2.15459928e-01 2.18074039e-01 -1.06090593e+00
8.68936032e-02 -1.89830899e-01 1.09230077e+00 -2.19060838e-01
5.13384044e-01 -8.24961483e-01 -1.82222780e-02 9.46978450e-01
-2.79759169e-01 -2.65980214e-02 1.64835438e-01 2.64225245e-01
-1.18083097e-01 -4.79641080e-01 9.03306961e-01 -3.50181609e-01
-1.09980452e+00 1.30855158e-01 -8.19810450e-01 -7.01739192e-01
9.10834074e-01 -6.21132553e-01 -5.88451549e-02 -2.09009215e-01
-9.90496635e-01 -7.39796340e-01 3.01890731e-01 5.30119538e-01
7.85066545e-01 -1.34380329e+00 -5.35180390e-01 7.94257879e-01
6.01778179e-03 -9.53859091e-01 -1.84574395e-01 4.84680593e-01
-5.54632604e-01 9.41051126e-01 -8.24145555e-01 -3.20499361e-01
-1.35473347e+00 8.93281698e-01 2.45101988e-01 2.19122946e-01
-7.27886438e-01 5.81746519e-01 -2.30942696e-01 -4.18365389e-01
4.52586234e-01 -9.41588655e-02 -3.65634322e-01 -1.47355974e-01
1.01208210e+00 2.47292444e-01 2.12910622e-01 -5.71807623e-01
-5.87739885e-01 4.63858813e-01 -1.84513360e-01 -2.69016862e-01
1.41778207e+00 1.18837185e-01 -2.89283156e-01 6.59387171e-01
6.65707111e-01 -2.86161005e-01 -9.24542129e-01 -2.23176956e-01
2.93563128e-01 -4.74308848e-01 -2.76186764e-01 -7.92417228e-01
-8.67756605e-01 7.37009585e-01 7.21938491e-01 9.75077674e-02
1.07051933e+00 -1.05925173e-01 -9.33030546e-02 1.40401804e+00
4.37596738e-01 -1.09221268e+00 9.13397223e-02 6.12592280e-01
9.65592265e-01 -1.16950178e+00 3.44250649e-01 -2.14083120e-01
-1.46476135e-01 1.74433911e+00 3.26487869e-01 -2.68873632e-01
7.81226695e-01 4.73581374e-01 3.31738949e-01 -9.02638286e-02
-5.60693562e-01 1.45570084e-01 3.13401312e-01 1.09664333e+00
6.97123230e-01 -2.53566653e-01 -3.86460662e-01 9.49351117e-02
-9.42437723e-02 3.31894249e-01 4.06153798e-01 1.22694707e+00
-6.95665061e-01 -1.26956081e+00 -4.82464135e-01 1.79490268e-01
-4.14790988e-01 2.62785584e-01 -6.72983944e-01 1.11854088e+00
-1.22210659e-01 6.32663369e-01 -8.62370245e-03 -3.40840489e-01
1.39683649e-01 4.64137316e-01 1.11266589e+00 -3.60486448e-01
-1.02934390e-01 -9.84302685e-02 -2.21649215e-01 -3.39821637e-01
-6.14088297e-01 -8.94951284e-01 -9.90826964e-01 -4.72731590e-01
3.16369138e-03 -4.00661789e-02 7.26549029e-01 1.02838135e+00
-1.80130720e-01 3.87196727e-02 5.61102271e-01 -1.02385247e+00
-1.05674469e+00 -9.92771924e-01 -1.09712124e+00 -5.41742332e-02
5.11981070e-01 -5.51179886e-01 -2.37180591e-01 2.47193262e-01]
|
[9.999953269958496, 2.0656611919403076]
|
6d94a53a-7213-41a6-a4fe-6d8241e08c36
|
a-modified-ctgan-plus-features-based-method
|
2302.02269
| null |
https://arxiv.org/abs/2302.02269v2
|
https://arxiv.org/pdf/2302.02269v2.pdf
|
A Modified CTGAN-Plus-Features Based Method for Optimal Asset Allocation
|
We propose a new approach to portfolio optimization that utilizes a unique combination of synthetic data generation and a CVaR-constraint. We formulate the portfolio optimization problem as an asset allocation problem in which each asset class is accessed through a passive (index) fund. The asset-class weights are determined by solving an optimization problem which includes a CVaR-constraint. The optimization is carried out by means of a Modified CTGAN algorithm which incorporates features (contextual information) and is used to generate synthetic return scenarios, which, in turn, are fed into the optimization engine. For contextual information we rely on several points along the U.S. Treasury yield curve. The merits of this approach are demonstrated with an example based on ten asset classes (covering stocks, bonds, and commodities) over a fourteen-and-half year period (January 2008-June 2022). We also show that the synthetic generation process is able to capture well the key characteristics of the original data, and the optimization scheme results in portfolios that exhibit satisfactory out-of-sample performance. We also show that this approach outperforms the conventional equal-weights (1/N) asset allocation strategy and other optimization formulations based on historical data only.
|
['Arturo Cifuentes', 'Domingo Ramírez', 'Omar Larré', 'Fernando Suárez', 'José-Manuel Peña']
|
2023-02-05
| null | null | null | null |
['synthetic-data-generation', 'synthetic-data-generation', 'portfolio-optimization']
|
['medical', 'miscellaneous', 'time-series']
|
[ 2.60282569e-02 1.36635959e-01 -5.74770160e-02 -2.72454053e-01
-7.79368341e-01 -8.79386067e-01 1.03283966e+00 -5.68104647e-02
-3.51336062e-01 1.00148439e+00 1.70195848e-01 -4.35392380e-01
-6.28724217e-01 -1.42923546e+00 -3.92834961e-01 -6.42353892e-01
-1.40294641e-01 7.67808616e-01 -2.90558428e-01 -3.83763164e-01
3.14996839e-01 5.99126339e-01 -1.23107827e+00 7.98273981e-02
5.89932740e-01 1.25233078e+00 -1.16859235e-01 4.96860385e-01
-1.36652008e-01 4.75322545e-01 -6.82537675e-01 -8.90878856e-01
1.05303347e+00 -2.71252304e-01 -3.02455455e-01 1.33583531e-01
-1.68864444e-01 -4.97099489e-01 4.41044718e-02 8.76311362e-01
3.80376518e-01 2.08944649e-01 7.18527317e-01 -9.10074472e-01
-8.42620283e-02 7.75448620e-01 -3.07296127e-01 8.43087807e-02
1.43731207e-01 -3.50862625e-03 1.48445058e+00 -7.53779352e-01
4.13342744e-01 1.05950773e+00 4.91303027e-01 -3.20153832e-02
-1.32857370e+00 -2.22566769e-01 -5.80024011e-02 -3.76987964e-01
-9.25400198e-01 -8.73705372e-02 7.92311311e-01 -6.21712267e-01
7.87235320e-01 5.62091231e-01 1.00653970e+00 6.33958638e-01
2.34939218e-01 5.39987028e-01 1.05688548e+00 -6.22661471e-01
3.66936952e-01 2.45909169e-01 -1.35779023e-01 5.25495820e-02
6.25054717e-01 6.96113765e-01 -2.30144411e-01 -4.69108999e-01
6.26512945e-01 -6.97241798e-02 -4.45617735e-02 -3.82986128e-01
-1.14195192e+00 1.22540796e+00 -1.86934173e-02 1.80337697e-01
-7.73819208e-01 9.47473124e-02 9.75942090e-02 6.10534132e-01
6.43626332e-01 5.22168815e-01 -3.91014367e-01 2.27877144e-02
-1.40541589e+00 7.30633974e-01 1.13733768e+00 7.14610338e-01
4.63370711e-01 5.45576334e-01 -3.01541418e-01 5.23378968e-01
4.52365041e-01 7.27234185e-01 3.23462605e-01 -1.02062893e+00
9.29597199e-01 3.47312927e-01 4.52131152e-01 -9.46933448e-01
-2.42530406e-01 -8.08525443e-01 -4.43341851e-01 5.56727529e-01
5.89105308e-01 -3.04107696e-01 -4.02132839e-01 1.56760454e+00
9.47965160e-02 -2.98665792e-01 2.03519732e-01 3.89111489e-01
-6.24500625e-02 6.45456076e-01 -3.74315441e-01 -5.04496932e-01
1.23304403e+00 -6.63972437e-01 -4.24746662e-01 4.03814852e-01
1.26794338e-01 -5.36712646e-01 4.92869109e-01 6.07248545e-01
-1.41488028e+00 -3.35503548e-01 -9.85979259e-01 1.00111282e+00
-3.69621634e-01 -7.91310892e-02 5.14982700e-01 1.26964986e+00
-8.21729481e-01 8.89033377e-01 -3.91853929e-01 4.10168737e-01
2.64220327e-01 3.51998717e-01 1.03403531e-01 4.58385497e-01
-1.28257918e+00 7.37592936e-01 7.27050304e-01 -6.93407003e-03
-7.38996804e-01 -8.56145322e-01 -5.44572353e-01 3.60906512e-01
3.47557724e-01 -7.03916371e-01 1.07392323e+00 -1.14166665e+00
-1.51802671e+00 3.61512214e-01 6.89738452e-01 -8.23322654e-01
1.08336067e+00 5.11139780e-02 -4.98077065e-01 1.64608702e-01
-1.21782362e-01 -4.40917611e-02 7.41046667e-01 -1.00912166e+00
-6.58635497e-01 -9.89492163e-02 1.07229285e-01 -5.86749353e-02
-2.16072962e-01 2.22858146e-01 7.10952058e-02 -1.56040967e+00
-1.69907093e-01 -8.00749958e-01 -3.37466508e-01 -5.86253464e-01
-4.54410464e-01 5.51382661e-01 9.98414978e-02 -9.13388729e-01
1.39109659e+00 -1.48403692e+00 1.22926831e-02 1.01854801e+00
-1.98895767e-01 -6.40386194e-02 1.08165145e-01 6.68276370e-01
-5.15892744e-01 1.57554895e-01 -5.78454316e-01 -2.08996460e-01
4.65803236e-01 -1.21924035e-01 -6.23710513e-01 2.35437781e-01
1.42451629e-01 9.90314722e-01 -6.13551378e-01 1.36068717e-01
1.89171985e-01 -1.31846309e-01 -4.88890469e-01 2.38349572e-01
-5.20602047e-01 5.63432835e-02 -4.58639055e-01 7.47112691e-01
5.93844175e-01 1.50385220e-02 3.11719805e-01 4.74300608e-02
-2.21127480e-01 6.18258342e-02 -1.29976368e+00 1.16541016e+00
-2.30993181e-01 1.56679094e-01 -4.19457257e-01 -9.90821004e-01
1.08533466e+00 3.19329321e-01 7.75511205e-01 -6.16437495e-01
-6.29112422e-02 5.81702590e-01 -6.08975329e-02 2.92509962e-02
5.73214054e-01 -4.55743790e-01 -2.35985324e-01 1.12243581e+00
-7.54916593e-02 -1.25987247e-01 4.75088060e-01 -2.10207969e-01
8.69058192e-01 3.08362871e-01 1.92619592e-01 -5.51033616e-01
4.53402042e-01 -5.32478169e-02 4.08872575e-01 7.64432311e-01
6.06699705e-01 5.99763215e-01 8.18375349e-01 -3.75255078e-01
-1.46586323e+00 -1.03384912e+00 -1.21895880e-01 4.32225347e-01
-8.77948165e-01 1.86008036e-01 -7.80026317e-01 -6.05527282e-01
4.32827979e-01 1.10949159e+00 -7.86000192e-01 2.77041823e-01
-4.47199225e-01 -1.55041897e+00 2.48118490e-01 1.20318584e-01
4.80390251e-01 -1.05713356e+00 -7.95568109e-01 7.20013499e-01
3.19885939e-01 -4.24640417e-01 -2.47408986e-01 -6.43214807e-02
-8.67849946e-01 -1.01790440e+00 -1.24121439e+00 2.36470327e-01
4.69350487e-01 -3.89382958e-01 1.45823145e+00 -2.71120191e-01
4.73682396e-03 4.59190398e-01 -3.30808997e-01 -6.23480678e-01
-4.21982795e-01 -2.91586597e-03 -1.91138148e-01 5.16729355e-01
-3.28243375e-02 -4.69447583e-01 -4.74710554e-01 1.03588782e-01
-9.69547451e-01 -3.13038290e-01 4.30901349e-01 1.02786219e+00
3.98248553e-01 1.70387566e-01 9.04892504e-01 -1.04149210e+00
6.00029707e-01 -7.61161923e-01 -1.34840524e+00 6.19489729e-01
-9.73903954e-01 1.54495880e-01 1.53793618e-01 -1.74175873e-01
-1.34476781e+00 -1.59593493e-01 2.46419117e-01 -1.89236730e-01
6.53908551e-01 8.20674837e-01 -2.89091706e-01 -4.16133553e-02
6.62187785e-02 1.17705062e-01 1.51756048e-01 -7.46510029e-01
1.93072259e-01 2.96931863e-01 3.95166934e-01 -8.24838161e-01
9.98587012e-01 2.26834998e-01 1.88621163e-01 -2.63985723e-01
-5.30617297e-01 1.73747078e-01 -4.66883391e-01 -3.38348150e-01
2.58753121e-01 -7.34383404e-01 -3.87334526e-01 5.98612607e-01
-7.36539602e-01 -2.47524738e-01 -9.38934684e-01 7.51873016e-01
-8.03134859e-01 1.63041919e-01 -2.50070333e-01 -1.14150059e+00
-4.65929478e-01 -8.82102668e-01 4.46856230e-01 -3.66464071e-02
-8.97867084e-02 -1.31069481e+00 4.86532360e-01 4.71019559e-02
6.05766892e-01 8.16821098e-01 1.02966440e+00 -1.03694689e+00
-7.83401132e-01 -3.04529488e-01 1.87483206e-02 4.73623455e-01
6.20964356e-02 1.44277707e-01 -6.90512300e-01 -4.88884181e-01
3.20019692e-01 1.37184694e-01 8.21124434e-01 4.31002110e-01
7.21757531e-01 -6.10121012e-01 3.46894324e-01 5.99920809e-01
1.77251625e+00 4.11646008e-01 6.28570914e-01 7.88083553e-01
9.74414870e-02 9.76354122e-01 8.39039683e-01 1.01781869e+00
6.84826151e-02 8.90406966e-01 5.17704189e-01 1.26075998e-01
4.13960844e-01 1.13640130e-02 3.43548179e-01 5.58073044e-01
-2.89107949e-01 -3.82178426e-01 -7.07934737e-01 5.24703801e-01
-1.66694617e+00 -1.24485552e+00 9.58590209e-02 2.61793780e+00
6.45509839e-01 3.23308259e-01 5.07098675e-01 1.63190141e-01
3.91876698e-01 3.06778640e-01 -4.02354419e-01 -3.62687469e-01
-4.69555020e-01 5.74541509e-01 8.16273153e-01 4.26953971e-01
-8.79777551e-01 1.39871195e-01 7.07789564e+00 7.00945973e-01
-5.40145397e-01 -2.09749177e-01 9.61159289e-01 -3.04599851e-01
-1.06250346e+00 7.36042205e-03 -7.45725691e-01 8.26909065e-01
1.13856542e+00 -6.84744835e-01 3.48036677e-01 5.74337065e-01
1.60642251e-01 2.36310754e-02 -8.33401144e-01 3.19188595e-01
-1.40072078e-01 -1.52038395e+00 2.00363204e-01 5.31042278e-01
8.60237300e-01 -3.77378672e-01 3.23198318e-01 4.10533249e-02
3.72954249e-01 -7.95765698e-01 1.17105579e+00 1.22700310e+00
7.51828551e-01 -1.21373260e+00 8.64565134e-01 4.18179370e-02
-1.02279890e+00 -3.62855077e-01 -2.71244437e-01 2.33246133e-01
3.97804171e-01 8.27323139e-01 -3.74441862e-01 1.03724265e+00
3.36088389e-01 3.56843561e-01 -3.12286049e-01 1.06427932e+00
1.26011327e-01 4.24664348e-01 -4.26758260e-01 1.64995819e-01
3.54853332e-01 -9.01364923e-01 6.81761384e-01 9.22328055e-01
9.95654821e-01 7.56039843e-02 -3.82868707e-01 1.09106374e+00
7.52478242e-02 3.77329788e-03 -5.72411716e-01 -3.91080566e-02
3.23953152e-01 8.10808122e-01 -3.65385652e-01 -1.90572798e-01
-5.75558960e-01 2.30136678e-01 -3.83064240e-01 3.52961183e-01
-5.02667606e-01 -4.04487967e-01 4.41922426e-01 2.68357638e-02
6.20395660e-01 6.73673823e-02 -2.65806407e-01 -1.05525362e+00
9.66438577e-02 -1.11219871e+00 5.46648562e-01 -4.86367196e-01
-1.07848144e+00 3.68369520e-01 4.74664330e-01 -1.50995648e+00
-9.82306063e-01 -7.21819580e-01 -7.81182766e-01 1.33955908e+00
-1.50684237e+00 -7.70480156e-01 2.76474178e-01 2.91822463e-01
1.69018328e-01 -1.12480891e+00 6.93322182e-01 1.37990505e-01
-4.82857347e-01 5.53619742e-01 7.43802547e-01 -5.23205213e-02
-1.27900541e-01 -1.55889547e+00 7.11171389e-01 8.07382107e-01
1.09074466e-01 4.16409522e-01 5.45818806e-01 -8.43420148e-01
-9.43972528e-01 -1.07516646e+00 7.42623150e-01 -1.89465716e-01
9.12357152e-01 6.19972646e-02 -5.52792430e-01 6.15224957e-01
2.17177629e-01 -4.66077715e-01 8.17459762e-01 -5.82267582e-01
9.96573642e-02 -2.73471743e-01 -1.35333872e+00 1.95258215e-01
3.04700434e-01 -8.61079693e-02 -6.47559524e-01 1.48658752e-01
5.28074026e-01 -5.23424074e-02 -1.33712554e+00 2.62711138e-01
7.81856596e-01 -1.10001576e+00 1.17325699e+00 -6.86753571e-01
1.75226748e-01 6.66810125e-02 -4.41898644e-01 -1.18783128e+00
-1.30065709e-01 -9.67297614e-01 -2.52815843e-01 1.36342227e+00
6.36238217e-01 -1.08375621e+00 9.01324093e-01 6.73588574e-01
3.77129912e-01 -6.37686312e-01 -1.06072021e+00 -8.91788065e-01
2.10479200e-01 -2.41227776e-01 1.40860415e+00 5.90874195e-01
-6.28470659e-01 -5.67634106e-01 -4.72824872e-01 -2.28257895e-01
1.15761185e+00 4.94728446e-01 4.95641321e-01 -1.23871827e+00
-7.43908167e-01 -7.45678782e-01 4.85948883e-02 -3.40205312e-01
-1.40385628e-01 -6.15274429e-01 -6.77387178e-01 -7.17781186e-01
4.02684277e-03 -7.73314416e-01 -6.56403303e-01 8.23043063e-02
8.56461301e-02 -5.68301901e-02 4.65701312e-01 1.79140076e-01
2.80423790e-01 5.30906260e-01 6.18974090e-01 -1.32009298e-01
5.05480357e-02 6.29647195e-01 -5.67594826e-01 4.28769201e-01
9.39791858e-01 -4.84982401e-01 -2.46835470e-01 1.21078439e-01
4.93955165e-01 4.99076992e-01 3.19360673e-01 -5.81268430e-01
-3.18664163e-01 -3.89140129e-01 2.72110671e-01 -9.39971507e-01
2.06975147e-01 -7.58158863e-01 1.09803486e+00 7.60973394e-01
-2.74817020e-01 5.16849995e-01 -4.51883860e-02 4.61620808e-01
-3.39268297e-01 -9.24663305e-01 4.81466025e-01 -3.39624137e-01
1.98399033e-02 2.47805923e-01 -1.17476977e-01 3.70645919e-03
1.06064141e+00 -5.00456840e-02 -6.96334243e-02 -3.90820622e-01
-6.73352122e-01 2.50130028e-01 5.57602882e-01 -2.23360900e-02
2.81242520e-01 -1.68297660e+00 -1.26160848e+00 2.44856596e-01
-2.30544001e-01 -4.35242862e-01 -1.25409737e-01 3.49652022e-01
-7.82311022e-01 5.94125509e-01 -3.04661542e-01 5.23715876e-02
-6.30345583e-01 3.24427783e-01 6.42890692e-01 -9.80771780e-01
-1.95647597e-01 5.69045603e-01 -9.23401341e-02 -2.01838240e-01
-6.20469078e-02 -5.90790920e-02 -3.32785994e-01 7.88057566e-01
4.83323514e-01 5.87686658e-01 1.19832091e-01 -4.99114454e-01
1.64452508e-01 5.32691717e-01 1.99563459e-01 -7.33923495e-01
1.70932877e+00 1.40657559e-01 -7.08618537e-02 3.38034630e-01
7.81146705e-01 1.33092865e-01 -1.35935223e+00 -2.70460874e-01
4.37132955e-01 -7.09747672e-01 -2.76186347e-01 -7.75288999e-01
-1.55866671e+00 5.92612743e-01 3.08426678e-01 6.44084394e-01
1.13354158e+00 -7.84364760e-01 4.01278734e-01 1.52462885e-01
6.15509272e-01 -1.22165811e+00 -1.42683282e-01 -2.19804570e-02
1.20125806e+00 -6.02198601e-01 4.18468416e-01 5.39880171e-02
-4.49890018e-01 1.29027665e+00 -3.72887194e-01 -1.87893838e-01
7.28094459e-01 8.86569396e-02 -3.72496992e-01 2.20175437e-03
-7.28676856e-01 1.64671600e-01 3.39310974e-01 3.19017529e-01
-1.19731255e-01 2.26475418e-01 -4.31727529e-01 7.63324022e-01
-4.48583722e-01 -4.31965351e-01 6.21556163e-01 8.51813138e-01
-2.91335255e-01 -1.60933328e+00 -5.37231445e-01 8.06424379e-01
-9.34252679e-01 -8.93316418e-02 -3.39638814e-02 1.06558311e+00
-2.72748649e-01 4.36402559e-01 1.88249752e-01 -8.93918574e-02
5.67050695e-01 2.55132373e-02 2.48084947e-01 -4.20820922e-01
-1.02187467e+00 2.16722652e-01 3.65195960e-01 -3.56381804e-01
-3.78821552e-01 -1.34415817e+00 -3.79405707e-01 -4.18536723e-01
-3.78797561e-01 2.82379478e-01 5.58258712e-01 4.70017850e-01
-2.01180667e-01 6.02193654e-01 1.30069864e+00 -9.76529360e-01
-1.42445266e+00 -7.18811035e-01 -1.10740376e+00 5.26408441e-02
4.97252122e-02 -6.25948012e-01 -4.09208059e-01 -1.59932166e-01]
|
[4.90438175201416, 4.011074066162109]
|
bcfd3eab-3f7a-45ed-8b27-d926fcf40f1b
|
query-driven-knowledge-base-completion-using
|
2212.01923
| null |
https://arxiv.org/abs/2212.01923v3
|
https://arxiv.org/pdf/2212.01923v3.pdf
|
Query-Driven Knowledge Base Completion using Multimodal Path Fusion over Multimodal Knowledge Graph
|
Over the past few years, large knowledge bases have been constructed to store massive amounts of knowledge. However, these knowledge bases are highly incomplete, for example, over 70% of people in Freebase have no known place of birth. To solve this problem, we propose a query-driven knowledge base completion system with multimodal fusion of unstructured and structured information. To effectively fuse unstructured information from the Web and structured information in knowledge bases to achieve good performance, our system builds multimodal knowledge graphs based on question answering and rule inference. We propose a multimodal path fusion algorithm to rank candidate answers based on different paths in the multimodal knowledge graphs, achieving much better performance than question answering, rule inference and a baseline fusion algorithm. To improve system efficiency, query-driven techniques are utilized to reduce the runtime of our system, providing fast responses to user queries. Extensive experiments have been conducted to demonstrate the effectiveness and efficiency of our system.
|
['Daisy Zhe Wang', 'Yang Peng']
|
2022-12-04
| null | null | null | null |
['knowledge-base-completion', 'knowledge-base-completion']
|
['graphs', 'knowledge-base']
|
[-2.24665716e-01 1.14121653e-01 -3.80736232e-01 -4.94210511e-01
-1.08057594e+00 -7.50927806e-01 3.05631965e-01 5.18034279e-01
-3.00336450e-01 8.80705953e-01 4.88633990e-01 -1.37447879e-01
-4.01088476e-01 -1.22307527e+00 -6.23897672e-01 -1.00929596e-01
1.69992104e-01 7.06788778e-01 6.06257975e-01 -5.57022154e-01
-4.77236882e-02 1.40701666e-01 -1.59626889e+00 6.82859719e-01
1.14303792e+00 8.50063980e-01 1.44667958e-03 5.58268070e-01
-6.19886816e-01 9.33247566e-01 -3.31592113e-01 -1.01317513e+00
-2.59200305e-01 -4.82813977e-02 -1.11809182e+00 -2.78809249e-01
3.48910958e-01 -4.53068137e-01 -6.49076104e-01 1.06992793e+00
3.95288467e-01 1.49808437e-01 3.35233271e-01 -1.08062804e+00
-5.20254254e-01 7.78076112e-01 -2.16719210e-01 -1.67477131e-01
1.15893626e+00 -4.82691020e-01 1.15948582e+00 -1.00986314e+00
6.16640031e-01 1.48559499e+00 2.66893059e-01 6.80826306e-02
-8.93823147e-01 -4.14046496e-01 7.28000104e-02 4.74884689e-01
-1.73958218e+00 -4.09569293e-01 6.50089025e-01 -5.27774058e-02
6.37840509e-01 5.64748943e-01 2.90624410e-01 5.04718661e-01
-2.61230350e-01 8.46475065e-01 4.36621457e-01 -4.57225978e-01
7.38640956e-04 2.01128140e-01 4.54155415e-01 1.11012709e+00
3.77568901e-01 -5.64342558e-01 -7.69100845e-01 -6.86427891e-01
5.00888050e-01 2.08241746e-01 -1.94535598e-01 -3.74719143e-01
-1.33825660e+00 6.97171867e-01 4.14123416e-01 -4.33954559e-02
-4.54556406e-01 -2.07891971e-01 1.18043616e-01 1.68013990e-01
-1.84049040e-01 -1.92190968e-02 -2.30366737e-01 1.17199339e-01
-4.28905725e-01 4.62800413e-01 1.25039387e+00 1.13265371e+00
1.26648915e+00 -6.32324696e-01 -2.64027476e-01 1.10431612e+00
5.44684350e-01 7.80558944e-01 4.98419814e-02 -1.24799657e+00
9.52651083e-01 1.29565728e+00 4.67684627e-01 -1.46823168e+00
-3.32504004e-01 2.19240785e-01 -5.69097459e-01 -7.67000020e-01
3.96370143e-01 -1.04707085e-01 -7.57927716e-01 1.46184230e+00
6.86265171e-01 -2.94384390e-01 2.99963742e-01 8.30994546e-01
1.18793893e+00 6.44225061e-01 1.17519774e-01 -5.58217876e-02
1.59213543e+00 -6.98870897e-01 -9.09118891e-01 1.36345118e-01
5.19128203e-01 -6.87663317e-01 7.43296862e-01 2.22957090e-01
-1.04599679e+00 -3.35083425e-01 -7.51052797e-01 -1.92977995e-01
-4.74531829e-01 3.71738896e-02 6.65799677e-01 6.56298637e-01
-6.79838240e-01 -9.93561074e-02 -7.78959572e-01 -4.01068300e-01
9.88651216e-02 4.40287620e-01 -4.74143535e-01 -8.50687742e-01
-1.35081518e+00 5.80898106e-01 7.70287871e-01 6.35437369e-02
-4.91062552e-01 -3.61404330e-01 -8.93159568e-01 1.54012233e-01
8.12187016e-01 -9.99455869e-01 1.12456715e+00 -2.87233561e-01
-1.05774021e+00 2.58591354e-01 -5.14591634e-01 3.99645567e-02
-7.97562152e-02 -2.36205935e-01 -6.42896473e-01 5.39969862e-01
1.20112754e-01 4.12292361e-01 1.72567502e-01 -1.49225092e+00
-9.44613993e-01 -6.24849916e-01 4.42130655e-01 4.02772009e-01
-4.73076403e-01 -5.69501556e-02 -1.38665128e+00 -1.46987095e-01
3.46275002e-01 -7.95487881e-01 -3.32929492e-02 -5.02291381e-01
-2.99248606e-01 -3.77195299e-01 6.22578800e-01 -8.90409231e-01
1.59611332e+00 -1.80863988e+00 3.61646302e-02 6.66299045e-01
2.65260845e-01 1.28164172e-01 -1.43398449e-01 7.92139947e-01
6.01489186e-01 -7.49306679e-02 -1.57111898e-01 1.30180299e-01
1.52794555e-01 6.79926634e-01 -4.40964580e-01 -1.53524831e-01
-2.25494072e-01 9.78039920e-01 -9.40810502e-01 -9.00956452e-01
-1.72848642e-01 3.66305858e-01 -4.74092036e-01 1.00951657e-01
-4.62364733e-01 -1.00537151e-01 -8.22176099e-01 1.12430453e+00
5.27577400e-01 -3.97954553e-01 5.13536692e-01 -4.64094728e-01
2.55440384e-01 -1.51946336e-01 -1.22210693e+00 1.86668456e+00
-6.33891076e-02 -1.35886148e-01 3.14288288e-01 -7.08806753e-01
6.45479560e-01 4.16513920e-01 5.21036923e-01 -7.83974886e-01
-1.65405318e-01 6.73477054e-02 -3.49644393e-01 -7.65485406e-01
9.38541174e-01 2.22893387e-01 -2.11660296e-01 1.85666800e-01
1.03789195e-02 2.02447921e-01 5.79158902e-01 8.57141197e-01
1.26199579e+00 -1.80814356e-01 -4.89903279e-02 4.34087187e-01
6.81266904e-01 4.18180913e-01 5.40312469e-01 7.17756867e-01
2.35212192e-01 1.81003883e-01 2.86117166e-01 -1.96575105e-01
-3.34327728e-01 -1.28063178e+00 3.25866163e-01 1.24742568e+00
5.42242348e-01 -8.68616998e-01 -4.21401232e-01 -6.58001602e-01
2.82368749e-01 4.02587056e-01 -8.19645897e-02 -2.03657933e-02
-4.29993540e-01 -5.22011995e-01 7.10640311e-01 3.80547374e-01
5.76604366e-01 -6.97361529e-01 7.70954266e-02 2.79262573e-01
-9.34840441e-01 -1.25621235e+00 -1.66771814e-01 -6.81452632e-01
-6.85216010e-01 -1.44655716e+00 -4.63848263e-01 -7.25265682e-01
7.92198956e-01 4.99701381e-01 1.09942877e+00 2.92297482e-01
-4.57685255e-03 9.93896484e-01 -6.03165150e-01 -1.77042514e-01
-6.81831613e-02 -2.77018100e-02 -1.12751901e-01 4.41575795e-02
2.69084692e-01 -3.02515566e-01 -6.02487624e-01 5.81765234e-01
-1.39028776e+00 -2.46751174e-01 6.30308151e-01 5.00798106e-01
4.80360091e-01 1.82542413e-01 6.33601189e-01 -8.12666655e-01
7.72520781e-01 -6.29548013e-01 -5.00035405e-01 9.36530292e-01
-3.14887971e-01 1.58732146e-01 1.89100906e-01 -4.19191942e-02
-1.44502151e+00 1.18345499e-01 -2.14947611e-02 -1.42998204e-01
4.06399034e-02 1.18080032e+00 -3.47724199e-01 -1.98507592e-01
4.29983258e-01 3.74136530e-02 -2.09289506e-01 -5.09790242e-01
7.62580991e-01 6.35712445e-01 7.99574614e-01 -9.26251829e-01
6.34004056e-01 5.03180981e-01 -2.39257351e-01 -6.04399800e-01
-8.74822557e-01 -8.23120952e-01 -3.38137299e-01 -1.87519640e-01
6.15189791e-01 -1.09335423e+00 -1.23646450e+00 -9.09726471e-02
-1.15433145e+00 4.09257233e-01 3.14444393e-01 5.34239709e-01
-1.32623047e-01 8.11264277e-01 -5.29424548e-01 -7.51628101e-01
-3.20125967e-01 -6.48500681e-01 8.04073989e-01 3.95832509e-01
7.83645362e-02 -7.80416667e-01 2.59962827e-01 9.98154879e-01
1.25566170e-01 8.94550327e-03 1.13867986e+00 -6.23092771e-01
-9.71881449e-01 -3.31022710e-01 -5.12225688e-01 -2.22825348e-01
6.96913600e-02 -2.01687589e-01 -3.93748969e-01 -6.58104867e-02
-6.89870119e-01 -4.39476132e-01 8.31763625e-01 -2.61351079e-01
7.93515801e-01 -3.26847017e-01 -5.57407379e-01 6.35165907e-03
1.36642754e+00 6.05152063e-02 4.47928846e-01 -3.61702703e-02
7.63094902e-01 6.91969156e-01 7.43043840e-01 5.29422939e-01
1.25757253e+00 4.36009973e-01 2.70861059e-01 2.40344524e-01
1.14744715e-01 -4.03727561e-01 2.15378344e-01 1.03031290e+00
-1.18941240e-01 -2.91448206e-01 -1.17361557e+00 6.93002403e-01
-2.27488637e+00 -9.83425140e-01 -4.21957597e-02 2.15980649e+00
8.26344490e-01 -3.16919476e-01 -4.21716861e-04 -1.41638026e-01
6.61114037e-01 -2.61752307e-01 -3.07889372e-01 2.76319146e-01
-5.28952070e-02 -1.81359991e-01 2.52551675e-01 5.94257236e-01
-7.83562839e-01 8.47647250e-01 6.42710066e+00 7.33954906e-01
-4.55367595e-01 -9.92238000e-02 -1.57327667e-01 8.35360214e-02
-7.80826986e-01 1.71512142e-01 -8.00295055e-01 2.19867125e-01
6.46536350e-01 -3.41964424e-01 4.92790222e-01 5.39159000e-01
-2.76644260e-01 -4.53529477e-01 -8.29437017e-01 1.16005349e+00
2.19552711e-01 -1.38475990e+00 4.15777028e-01 -1.26344562e-01
6.44213974e-01 -1.12454586e-01 -5.01008511e-01 6.41757786e-01
7.60339618e-01 -5.84706008e-01 1.90326333e-01 1.03631938e+00
3.63903850e-01 -1.04563546e+00 6.58513248e-01 5.03343880e-01
-1.40644014e+00 -1.89018086e-01 -2.94309139e-01 3.06688994e-01
3.82819325e-01 6.34107471e-01 -7.05253243e-01 1.32754123e+00
6.25547051e-01 1.06553987e-01 -6.53295577e-01 1.15912604e+00
-4.12191562e-02 4.52101022e-01 -5.32460630e-01 1.21863112e-02
-2.14445770e-01 -2.35147178e-01 3.03220451e-01 9.78615999e-01
2.87622452e-01 6.31204963e-01 5.98083794e-01 3.62271845e-01
-4.63047683e-01 2.70639151e-01 -5.03514111e-01 -1.90968663e-01
6.62449419e-01 1.30799556e+00 -4.13594097e-01 -4.99411076e-01
-7.71773040e-01 5.74524045e-01 4.40233827e-01 7.13634968e-01
-7.36748099e-01 -7.95580268e-01 2.13369325e-01 -1.53093517e-01
1.69783190e-01 -3.36052179e-01 2.04122007e-01 -1.38042641e+00
2.80263394e-01 -7.37396419e-01 1.06293678e+00 -8.15546930e-01
-1.18870354e+00 2.20439225e-01 1.30538374e-01 -5.90443492e-01
-1.67373389e-01 -2.31718421e-01 -5.81580289e-02 5.23870170e-01
-1.28709376e+00 -1.15376759e+00 -4.16610420e-01 1.04786396e+00
-1.35737389e-01 -4.46156040e-02 8.60213161e-01 6.30687714e-01
-4.75466132e-01 3.56461406e-01 -1.25921983e-02 2.71454722e-01
9.11892414e-01 -9.86360908e-01 -3.29938382e-01 4.88530725e-01
2.24727824e-01 1.05927682e+00 3.07408810e-01 -9.63928699e-01
-2.15489936e+00 -8.48701000e-01 9.28358972e-01 -4.98875529e-01
4.87593889e-01 1.45739466e-02 -1.14031184e+00 5.54718018e-01
1.15483291e-01 -3.59022826e-01 1.02529538e+00 4.78401005e-01
-5.58127046e-01 -3.92844856e-01 -9.33314919e-01 5.14236212e-01
8.00549507e-01 -6.95137084e-01 -9.90891278e-01 2.77503133e-01
7.63354719e-01 -3.09944242e-01 -1.18429971e+00 6.64235532e-01
4.90471512e-01 -5.91249287e-01 1.03182554e+00 -5.21154284e-01
2.92275976e-02 -6.06761754e-01 -4.24298823e-01 -7.38834739e-01
-1.09418757e-01 -3.57770741e-01 -6.24551594e-01 1.29177296e+00
6.18272543e-01 -3.23623925e-01 9.26406920e-01 1.22440875e+00
2.38230348e-01 -4.36788261e-01 -7.31397510e-01 -2.60110468e-01
-6.57735229e-01 -3.03146541e-01 6.41991496e-01 8.53600979e-01
4.71210301e-01 5.68355203e-01 -2.06445575e-01 5.90696931e-01
5.14120877e-01 5.00099480e-01 8.40852261e-01 -1.21536243e+00
1.26836479e-01 6.23036847e-02 -6.09033778e-02 -1.01246631e+00
-1.54736983e-02 -8.69358063e-01 -1.82770282e-01 -2.10881090e+00
4.66357887e-01 -2.42696777e-01 -1.92408487e-01 7.12513566e-01
-3.30965400e-01 -2.80539673e-02 5.45605421e-02 2.28075355e-01
-1.40396118e+00 6.03762388e-01 1.06614947e+00 -1.29365072e-01
-1.49747461e-01 -3.38057816e-01 -7.96119392e-01 6.17477238e-01
4.26507324e-01 -2.16114521e-01 -5.55160761e-01 -6.28090024e-01
7.89587498e-01 7.01232135e-01 2.41062105e-01 -7.48305082e-01
9.05832112e-01 -1.61179215e-01 2.20952272e-01 -9.76873517e-01
5.79261959e-01 -8.91964912e-01 1.17328689e-01 4.45859320e-02
-3.24764028e-02 -1.65890362e-02 4.78522293e-02 9.01335895e-01
-8.10152233e-01 1.27564639e-01 1.57167315e-02 -3.90394218e-02
-8.55658770e-01 3.98702741e-01 -1.27647042e-01 -1.43171558e-02
7.48919547e-01 1.93303302e-01 -5.08963704e-01 -5.91397941e-01
-8.14741790e-01 9.35885847e-01 2.40777731e-01 3.93709809e-01
9.86868799e-01 -1.62492192e+00 -4.40961570e-01 -2.90719897e-01
4.37614620e-01 2.03087628e-02 4.20910984e-01 6.28190398e-01
-5.51632345e-01 5.85861564e-01 4.89126369e-02 -1.56659037e-01
-1.29923499e+00 6.29057169e-01 -2.13854298e-01 -1.58341199e-01
-5.26581556e-02 4.78122026e-01 -1.89817414e-01 -7.48320222e-01
5.43035448e-01 3.22205052e-02 -3.06391060e-01 1.15055665e-01
6.79480910e-01 4.96351600e-01 8.29973221e-02 -4.18071717e-01
-3.47678334e-01 3.26075494e-01 -1.27068087e-01 -3.12748075e-01
9.43316579e-01 -4.25985783e-01 -4.59032863e-01 2.31102780e-02
7.50839233e-01 3.60030472e-01 -3.73388559e-01 -7.02653885e-01
1.13220990e-01 -3.60955715e-01 -2.55937994e-01 -9.36730742e-01
-7.01853752e-01 3.38649720e-01 9.27511416e-03 1.10737763e-01
1.09353948e+00 1.27670094e-01 9.82273698e-01 1.15301192e+00
5.15546381e-01 -1.07372582e+00 -1.24280244e-01 6.37654066e-01
7.23797739e-01 -1.34008503e+00 3.44670676e-02 -6.85974300e-01
-6.30371869e-01 9.41282034e-01 5.12187362e-01 5.39029121e-01
5.12014091e-01 -3.42244595e-01 8.45394060e-02 -4.12226558e-01
-8.55630338e-01 -5.35084307e-01 5.25358558e-01 2.83049464e-01
6.42172471e-02 -1.02344118e-01 -2.94452608e-01 8.34298551e-01
1.61794275e-01 1.02984808e-01 -4.42213304e-02 1.21290147e+00
-8.14549506e-01 -1.29227102e+00 -7.51506686e-01 3.00247014e-01
-4.23153728e-01 7.51551166e-02 -5.95420182e-01 5.65540552e-01
-1.65491864e-01 1.46542680e+00 -4.74560171e-01 -4.01904255e-01
5.60736299e-01 3.03194463e-01 4.42880988e-01 -3.98322016e-01
-2.39601180e-01 -1.57547414e-01 4.71557200e-01 -5.81942379e-01
-4.66240138e-01 -2.01079026e-01 -1.63602972e+00 -3.60320717e-01
-2.06066489e-01 5.92705190e-01 4.52873975e-01 9.54368234e-01
5.24586558e-01 5.26228212e-02 2.27451012e-01 -1.73194930e-01
-1.61940128e-01 -4.92931426e-01 -3.58924568e-01 4.82818842e-01
-6.09214604e-02 -5.05659401e-01 4.13432896e-01 1.94564357e-01]
|
[10.449646949768066, 7.879647254943848]
|
e3e859d2-3039-4fa1-86fc-2cb509adac45
|
knowledge-based-paranoia-search-in-trick
|
2104.05423
| null |
https://arxiv.org/abs/2104.05423v1
|
https://arxiv.org/pdf/2104.05423v1.pdf
|
Knowledge-Based Paranoia Search in Trick-Taking
|
This paper proposes \emph{knowledge-based paraonoia search} (KBPS) to find forced wins during trick-taking in the card game Skat; for some one of the most interesting card games for three players. It combines efficient partial information game-tree search with knowledge representation and reasoning. This worst-case analysis, initiated after a small number of tricks, leads to a prioritized choice of cards. We provide variants of KBPS for the declarer and the opponents, and an approximation to find a forced win against most worlds in the belief space. Replaying thousands of expert games, our evaluation indicates that the AIs with the new algorithms perform better than humans in their play, achieving an average score of over 1,000 points in the agreed standard for evaluating Skat tournaments, the extended Seeger system.
|
['Stefan Edelkamp']
|
2021-04-07
| null | null | null | null |
['card-games']
|
['playing-games']
|
[-3.78318906e-01 3.27531368e-01 -3.44953500e-02 1.90841570e-01
-9.63800073e-01 -1.08551252e+00 2.80132443e-01 4.22761776e-02
-8.07552159e-01 1.00225604e+00 -8.21574852e-02 -6.60596609e-01
-9.46745574e-01 -9.95960057e-01 -2.58109897e-01 -2.41878703e-01
-1.19031124e-01 1.45666456e+00 1.13299465e+00 -1.03652132e+00
6.69636250e-01 9.92763266e-02 -1.32369256e+00 6.46938801e-01
4.93993610e-01 1.03545463e+00 -1.33908734e-01 8.91051173e-01
2.15975251e-02 1.56812274e+00 -7.67824233e-01 -1.23291457e+00
7.38759875e-01 -4.35505748e-01 -1.52286923e+00 -8.38643312e-01
-2.12138981e-01 -3.79666597e-01 -1.70852304e-01 1.09374452e+00
3.93680274e-01 3.76945615e-01 4.01240110e-01 -1.08444738e+00
5.02523124e-01 1.39272380e+00 -1.31195530e-01 5.08671522e-01
8.67250681e-01 2.26947561e-01 1.22701681e+00 -8.60226825e-02
6.88105464e-01 9.64894950e-01 7.12972403e-01 3.17834944e-01
-7.36550331e-01 -5.42550981e-01 -3.34299535e-01 7.74209559e-01
-1.52461147e+00 -1.55166566e-01 4.24428254e-01 -1.51254982e-01
1.21188688e+00 6.88433528e-01 1.17899418e+00 3.91863823e-01
2.61507004e-01 6.76526785e-01 1.28309965e+00 -5.78311145e-01
5.92714369e-01 2.32409146e-02 1.15352288e-01 3.60052943e-01
6.52232826e-01 5.36503971e-01 -9.28638637e-01 -7.36281693e-01
8.19146037e-01 -6.79242611e-01 2.82156795e-01 -3.34562033e-01
-6.44950807e-01 8.71818304e-01 -9.24093649e-02 9.55602601e-02
-6.11506224e-01 1.20503791e-01 3.96850854e-01 5.52250743e-01
-1.72210842e-01 1.11230934e+00 -4.86096114e-01 -1.03931189e+00
-9.31315660e-01 1.06570518e+00 1.35036290e+00 4.33510929e-01
3.31147790e-01 -2.40381926e-01 1.53158143e-01 1.63534239e-01
-1.43398598e-01 8.66185650e-02 7.23171309e-02 -1.33076024e+00
4.30573940e-01 7.89781153e-01 5.10630846e-01 -8.98901582e-01
-4.61654663e-01 -3.49379033e-01 -5.56372367e-02 8.93993974e-01
8.62913907e-01 -3.90318334e-02 -2.82226861e-01 1.25169420e+00
1.05028696e-01 -2.33246565e-01 2.40306944e-01 8.44323575e-01
5.07650077e-01 1.97055817e-01 -3.47128779e-01 5.66872172e-02
1.59845579e+00 -4.62601960e-01 -3.54103059e-01 -4.04386550e-01
3.82684350e-01 -2.10538566e-01 6.28553331e-01 1.37509477e+00
-1.69779301e+00 -7.07853138e-02 -1.06394637e+00 3.39908987e-01
-1.34448960e-01 -7.32319415e-01 1.05403769e+00 9.86287355e-01
-8.03156078e-01 5.41009128e-01 -5.70159078e-01 5.74735664e-02
5.42497747e-02 5.18135488e-01 -1.31368577e-01 9.71251074e-03
-1.53592753e+00 1.28917503e+00 9.76264477e-01 -1.47310480e-01
-7.85751402e-01 -1.93503827e-01 -4.15959686e-01 1.45024940e-01
1.41038215e+00 -3.52805525e-01 1.44362867e+00 -4.34516519e-01
-1.42530942e+00 9.04918671e-01 5.61588109e-01 -6.97318435e-01
7.65138328e-01 -2.61578441e-01 -8.93259048e-02 3.47606651e-02
1.99830130e-01 1.63610950e-02 2.21783537e-02 -8.70967507e-01
-9.66116369e-01 -2.93504685e-01 9.57114339e-01 6.66435480e-01
5.08553147e-01 3.44326168e-01 -2.99860179e-01 -3.92043255e-02
1.75630689e-01 -5.78500211e-01 -5.34581006e-01 -8.69086385e-01
-1.96771249e-01 -3.20741773e-01 -4.50684577e-01 -6.28496110e-01
1.62815559e+00 -1.50606894e+00 1.68969378e-01 8.18630099e-01
1.82148427e-01 -8.20595324e-02 3.51160288e-01 6.60050750e-01
1.12033576e-01 -4.58312631e-02 5.62508047e-01 7.06791997e-01
3.31389010e-01 1.78344786e-01 -2.75297523e-01 -5.37176728e-02
-7.25289643e-01 9.92993832e-01 -9.07196224e-01 -5.56101382e-01
-1.34692043e-01 -7.58978605e-01 -6.84976339e-01 -1.39930263e-01
-2.78802533e-02 -4.18004990e-01 -5.81124425e-01 4.65221107e-01
2.67633915e-01 8.61494243e-02 5.20181179e-01 4.26041096e-01
-4.66206949e-03 5.34045994e-01 -1.75940156e+00 1.60056734e+00
3.37431848e-01 4.55811471e-02 5.74215613e-02 -5.30766368e-01
5.48044801e-01 1.56096235e-01 4.84248362e-02 -7.66027451e-01
2.67276227e-01 2.97841966e-01 3.45024765e-01 -1.10314250e-01
7.71614492e-01 -3.45980048e-01 -8.34556282e-01 6.16338611e-01
-1.27997007e-02 -4.61553037e-01 4.11546171e-01 5.11942983e-01
1.60403633e+00 1.19725391e-01 7.59908855e-01 -3.04072738e-01
2.67504513e-01 8.52045953e-01 6.44228101e-01 1.65534699e+00
-2.03485906e-01 -9.49881785e-03 9.37221229e-01 -9.43193376e-01
-4.62277949e-01 -1.01254642e+00 5.60892940e-01 1.38258290e+00
5.01021385e-01 -8.53451192e-01 -5.72641551e-01 -4.57366019e-01
-2.41464496e-01 1.04629540e+00 -4.86030281e-01 -1.38282463e-01
-3.60460997e-01 -3.74907553e-01 1.05911553e+00 4.47259814e-01
6.36924565e-01 -1.03059709e+00 -1.17805040e+00 4.38530624e-01
-5.30859053e-01 -4.75906342e-01 6.99242949e-02 6.39745951e-01
-4.64521080e-01 -1.61650920e+00 4.09757085e-02 -1.95013419e-01
-1.87671304e-01 -3.00358921e-01 1.28113186e+00 1.74065530e-01
-3.22555564e-02 1.47844523e-01 -5.79955697e-01 -5.17748594e-01
-2.79476285e-01 -1.68106571e-01 2.77353101e-03 -9.44428742e-01
6.15141332e-01 -3.58201981e-01 -3.34500462e-01 5.86202085e-01
-2.90015131e-01 -1.94468461e-02 3.10551167e-01 8.63025069e-01
1.39054745e-01 8.86321127e-01 -4.55102772e-02 -7.61185467e-01
1.13515949e+00 -7.56521374e-02 -8.31583858e-01 4.13665682e-01
-5.95821083e-01 2.99544055e-02 3.52664530e-01 -1.55429438e-01
-9.35617566e-01 -2.39791244e-01 7.31258318e-02 2.29095504e-01
1.69659823e-01 7.15394914e-01 -1.03509210e-01 -3.03658098e-01
1.41246629e+00 3.05646539e-01 -3.43248546e-01 -1.61636010e-01
3.06135386e-01 2.21518129e-01 7.04625905e-01 -1.20733535e+00
6.05172157e-01 1.22901099e-02 -2.10572332e-01 5.79212494e-02
-4.57785100e-01 -4.85102952e-01 -2.63920307e-01 -5.90688586e-01
2.95833468e-01 -5.73865891e-01 -1.57353044e+00 4.20386702e-01
-9.11091566e-01 -3.75553876e-01 -6.82047963e-01 2.84153134e-01
-8.88162196e-01 2.94755161e-01 -6.38659656e-01 -1.28877497e+00
-1.05727002e-01 -9.43130553e-01 3.91645104e-01 2.10058719e-01
-5.68470299e-01 -3.50039572e-01 3.65065366e-01 8.34182680e-01
1.34765863e-01 -4.58015762e-02 7.48406351e-01 -1.36484003e+00
-3.85648906e-01 -6.36052489e-01 2.45980114e-01 -2.07899392e-01
-4.57271636e-01 -8.46450269e-01 -3.04527760e-01 -8.68459344e-02
-1.02869220e-01 -6.42307460e-01 4.47337359e-01 3.44794601e-01
3.72932523e-01 -1.96079195e-01 -2.28864878e-01 1.00770183e-01
1.21524549e+00 8.04109871e-01 9.93813336e-01 1.02487063e+00
-3.17237467e-01 5.75124741e-01 9.25853729e-01 6.69355571e-01
2.51533151e-01 7.81318903e-01 3.86911631e-01 6.94234192e-01
4.09380019e-01 -1.62874907e-01 1.22024380e-01 4.76870947e-02
-1.02228463e+00 -8.86853710e-02 -9.74395156e-01 4.93332922e-01
-1.99634337e+00 -1.44206285e+00 1.96772039e-01 2.29590225e+00
8.89495492e-01 1.07816315e+00 6.61932111e-01 4.88795757e-01
3.84957552e-01 -1.08969741e-01 -4.14416611e-01 -7.35868692e-01
-1.28204763e-01 9.67653513e-01 7.48858273e-01 6.81368113e-01
-5.29657185e-01 1.44767356e+00 7.64839411e+00 1.64563847e+00
4.07610312e-02 -5.73313087e-02 2.30063543e-01 -4.16613251e-01
-1.69669762e-01 5.36174953e-01 -4.59540933e-01 1.15357362e-01
7.14497685e-01 -6.18965566e-01 9.13762748e-01 1.06485820e+00
-4.91549075e-01 -7.61962652e-01 -5.47061026e-01 7.94017613e-01
-1.48385212e-01 -1.64711010e+00 -1.89118281e-01 6.37025759e-02
1.83976039e-01 -4.72959191e-01 -3.33197474e-01 7.21363246e-01
1.43248081e+00 -1.13700557e+00 1.22749102e+00 3.18709880e-01
5.44734240e-01 -1.05162132e+00 1.05725646e+00 6.15068316e-01
-9.88372266e-01 -3.70376647e-01 -2.42238566e-01 -8.92491221e-01
-4.34786221e-03 -1.33962274e-01 -9.37755942e-01 7.86667824e-01
7.37489879e-01 -5.48582554e-01 -2.18748748e-01 1.20150721e+00
-3.60328257e-01 6.03540897e-01 -7.89350390e-01 -4.14624751e-01
3.14778477e-01 -6.85719252e-02 6.65665150e-01 5.43457508e-01
-1.22582108e-01 8.63629162e-01 6.28866106e-02 6.74435616e-01
7.48909831e-01 -1.43522441e-01 -9.70028639e-02 3.28036606e-01
7.02192128e-01 7.57273018e-01 -9.98819888e-01 -2.61472255e-01
3.91177356e-01 8.07050228e-01 7.44421110e-02 -1.31747484e-01
-5.69510639e-01 -6.39830649e-01 3.65952045e-01 1.46235481e-01
1.86564177e-01 1.69363976e-01 -3.50000679e-01 -6.44469321e-01
-1.13433927e-01 -1.35059714e+00 1.06300139e+00 -9.26954508e-01
-8.95530462e-01 6.29128635e-01 5.33983409e-01 -7.29398608e-01
-7.94067323e-01 -7.35093474e-01 -7.59806871e-01 7.49365270e-01
-4.51546997e-01 -7.36652911e-01 1.24387786e-01 7.08898723e-01
2.27860168e-01 -4.15118694e-01 7.89855242e-01 -4.89073336e-01
1.19862564e-01 4.88201737e-01 -4.13072824e-01 1.09582856e-01
-5.79052418e-02 -1.45959485e+00 3.80455941e-01 7.14331746e-01
1.06299371e-01 4.90075082e-01 9.76300240e-01 -8.80779266e-01
-1.20479405e+00 2.82746613e-01 3.86162013e-01 -6.66847467e-01
5.80793202e-01 -1.54898129e-02 -2.37686917e-01 4.98799026e-01
-1.26898766e-01 -8.84158492e-01 5.90739965e-01 5.56335866e-01
-3.19561481e-01 5.82883805e-02 -1.15633929e+00 6.05918109e-01
1.11258972e+00 -3.94787520e-01 -1.40759051e+00 -6.56157080e-03
-2.79694051e-02 -8.44677329e-01 -5.27375877e-01 7.94456061e-03
9.33184147e-01 -1.40527880e+00 1.03932667e+00 -9.90529478e-01
-4.46860120e-02 -3.00174773e-01 -1.55155271e-01 -1.05531681e+00
-5.33486664e-01 -9.48819876e-01 2.80551136e-01 3.57011795e-01
2.50570595e-01 -4.78442043e-01 1.39822376e+00 8.77293169e-01
2.58350015e-01 -5.03527403e-01 -1.41826427e+00 -8.41809571e-01
4.83285896e-02 -9.56977963e-01 6.63986921e-01 6.77849591e-01
9.01573837e-01 1.67030290e-01 -5.83558738e-01 -1.34970620e-01
8.19024146e-01 7.30215460e-02 7.95950532e-01 -1.34867942e+00
-9.28409338e-01 -7.15651393e-01 -6.85857475e-01 -9.56887424e-01
-4.65478361e-01 -5.93147695e-01 -7.31442645e-02 -1.46071243e+00
2.93229520e-01 -3.06716561e-01 -3.46084028e-01 7.30881333e-01
1.23191230e-01 -5.82246892e-02 1.76442102e-01 -3.56742553e-02
-1.20737851e+00 -4.37808573e-01 8.39041293e-01 -6.26882911e-03
-2.78799862e-01 2.52984911e-01 -1.20713425e+00 9.38996077e-01
5.31947911e-01 -4.43318367e-01 -2.89146990e-01 2.12041393e-01
1.33062553e+00 7.02161372e-01 2.19287544e-01 -1.16113126e+00
9.28125024e-01 -7.14716256e-01 5.03557511e-02 -5.94880283e-01
4.35340255e-01 -4.86389965e-01 7.01169908e-01 9.17778313e-01
-2.71350086e-01 -3.44489962e-02 3.09380502e-01 3.15667033e-01
-9.16505307e-02 -6.02949619e-01 8.30016881e-02 -5.48719525e-01
-1.09160161e+00 -4.03971523e-01 -8.45975935e-01 2.66268849e-01
9.20636654e-01 -1.04174209e+00 -2.62747973e-01 -7.63630033e-01
-8.01125348e-01 3.09037507e-01 1.70435682e-01 -3.67472947e-01
4.47797507e-01 -8.65423441e-01 -7.37328827e-01 -2.53401369e-01
2.79220343e-02 -1.71541095e-01 3.80508780e-01 4.49596554e-01
-1.00849390e+00 3.05434257e-01 -6.67333961e-01 3.93954277e-01
-1.23309660e+00 2.16110632e-01 5.78558922e-01 -1.06295657e+00
-3.97337615e-01 1.31760406e+00 -2.96784014e-01 -1.55899003e-01
-5.42355292e-02 4.29052152e-02 -2.45638356e-01 -2.52039671e-01
6.15217388e-01 6.76797092e-01 2.20755279e-01 1.84775237e-02
-6.68312013e-01 2.33649127e-02 -1.51566684e-01 -6.62242353e-01
1.05163169e+00 2.82641530e-01 1.34754032e-02 8.83066133e-02
-3.90913576e-01 2.56962389e-01 -5.95037282e-01 -2.92459782e-02
2.05107033e-01 -5.76806724e-01 -3.53895277e-02 -1.47960722e+00
-4.87433732e-01 6.01399727e-02 -1.62690878e-02 6.23653591e-01
9.44638610e-01 7.13789416e-03 4.81127113e-01 9.93074000e-01
1.47133553e+00 -1.20037961e+00 -2.29689211e-01 6.37971103e-01
4.99512285e-01 -5.18874049e-01 3.19083333e-01 -6.71379194e-02
-1.03843200e+00 9.92454827e-01 6.83235049e-01 -1.98807433e-01
2.45752390e-02 4.22352135e-01 -5.35382852e-02 -6.94073319e-01
-9.65218723e-01 -2.24259764e-01 -1.02762923e-01 6.48784459e-01
-5.86810768e-01 3.08619171e-01 -5.65784216e-01 1.63483274e+00
-1.00196898e+00 8.80702063e-02 6.02895439e-01 1.15121281e+00
-8.54740858e-01 -1.01701093e+00 -7.79385030e-01 5.46667099e-01
-5.69209874e-01 -2.40658313e-01 -9.08026993e-01 9.54511344e-01
1.60348997e-01 1.14062858e+00 -3.55404317e-01 -6.81479692e-01
5.87181866e-01 -3.65080461e-02 8.44826519e-01 -3.48616391e-01
-1.10456085e+00 -2.66338110e-01 6.90019310e-01 -9.68025446e-01
2.89236531e-02 -5.97733796e-01 -1.04980695e+00 -7.98448622e-01
-6.51973724e-01 9.52703655e-01 -3.06152329e-02 9.14278448e-01
-2.20851585e-01 3.34776863e-02 -1.14474513e-01 -3.73734951e-01
-7.83076584e-01 -4.56053793e-01 -1.21248496e+00 1.81014109e-02
-7.08545506e-01 -9.39135611e-01 -2.19875619e-01 -6.10174179e-01]
|
[3.419816255569458, 1.5107909440994263]
|
975c377d-06ed-46c7-a631-051affaa9555
|
alignment-free-cross-lingual-semantic-role
| null | null |
https://aclanthology.org/2020.emnlp-main.319
|
https://aclanthology.org/2020.emnlp-main.319.pdf
|
Alignment-free Cross-lingual Semantic Role Labeling
|
Cross-lingual semantic role labeling (SRL) aims at leveraging resources in a source language to minimize the effort required to construct annotations or models for a new target language. Recent approaches rely on word alignments, machine translation engines, or preprocessing tools such as parsers or taggers. We propose a cross-lingual SRL model which only requires annotations in a source language and access to raw text in the form of a parallel corpus. The backbone of our model is an LSTM-based semantic role labeler jointly trained with a semantic role compressor and multilingual word embeddings. The compressor collects useful information from the output of the semantic role labeler, filtering noisy and conflicting evidence. It lives in a multilingual embedding space and provides direct supervision for predicting semantic roles in the target language. Results on the Universal Proposition Bank and manually annotated datasets show that our method is highly effective, even against systems utilizing supervised features.
|
['Mirella Lapata', 'Rui Cai']
| null | null | null | null |
emnlp-2020-11
|
['multilingual-word-embeddings']
|
['methodology']
|
[ 4.67597634e-01 5.68695664e-01 -9.42699671e-01 -6.58433795e-01
-1.15288579e+00 -8.97281468e-01 5.82940042e-01 6.21731639e-01
-9.11803067e-01 8.08920264e-01 6.53184950e-01 -3.05391431e-01
2.30171025e-01 -5.05873978e-01 -7.02135146e-01 -4.23003823e-01
2.72615969e-01 8.01145434e-01 4.40279879e-02 -3.20199817e-01
-4.02651094e-02 -5.44812307e-02 -1.02151990e+00 3.27728629e-01
7.07792819e-01 6.54721856e-01 3.84901941e-01 3.16424221e-01
-3.55746567e-01 1.06304181e+00 -3.29557419e-01 -5.71079671e-01
2.41958961e-01 -3.03191781e-01 -1.00374389e+00 -1.39994726e-01
8.12600181e-02 -5.37466705e-02 -8.65228921e-02 9.56489384e-01
2.39428818e-01 6.96926787e-02 1.97706386e-01 -1.06576061e+00
-7.81297326e-01 1.06169713e+00 4.16714214e-02 5.16740903e-02
2.53195286e-01 -1.68873370e-01 1.79006338e+00 -9.87617016e-01
1.03750265e+00 1.36010146e+00 5.14280260e-01 8.19241464e-01
-1.20230520e+00 -3.79036158e-01 2.28722841e-01 2.03201786e-01
-8.05871844e-01 -6.02563322e-01 6.79617763e-01 -1.70837387e-01
1.26916409e+00 -2.11790040e-01 4.24515992e-01 9.94354486e-01
-2.05724671e-01 8.45885754e-01 8.13086152e-01 -8.41898501e-01
1.77585155e-01 2.40530238e-01 3.73840243e-01 9.81175363e-01
6.94192573e-02 -3.19909692e-01 -9.46449280e-01 -3.25603068e-01
3.80561292e-01 -5.17376363e-01 -3.37242447e-02 -4.61208940e-01
-1.29431069e+00 9.54500556e-01 -9.73885208e-02 4.86938685e-01
-8.49411190e-02 2.13387594e-01 7.76783407e-01 5.52194297e-01
7.79956639e-01 8.50922942e-01 -1.08006859e+00 -1.32745251e-01
-5.25253475e-01 -1.96651340e-01 6.97081864e-01 7.11791694e-01
6.32387459e-01 -1.45400435e-01 2.57532328e-01 1.10261977e+00
4.00108755e-01 2.29470685e-01 8.18746686e-01 -1.16347110e+00
5.60647905e-01 7.38043189e-01 1.72774240e-01 -3.18468630e-01
-1.91981092e-01 -1.35411546e-01 1.20757975e-01 -4.31725472e-01
3.52681249e-01 1.09105244e-01 -4.95234758e-01 2.06711912e+00
4.34812307e-01 -9.86401960e-02 3.22947264e-01 6.20622814e-01
1.84538171e-01 5.24514437e-01 5.91341972e-01 1.24190664e-02
1.40130436e+00 -1.34032083e+00 -6.37098908e-01 -5.40371895e-01
1.30108488e+00 -5.71108878e-01 1.45323992e+00 -5.04402965e-02
-9.69696522e-01 -3.09337497e-01 -1.12946761e+00 -5.91930926e-01
-5.00305176e-01 1.70927152e-01 4.34111655e-01 3.95798057e-01
-1.20508862e+00 4.96873051e-01 -7.15411901e-01 -4.02273268e-01
2.10890383e-01 2.36545056e-01 -7.39300191e-01 -6.15078270e-01
-1.62015450e+00 1.19532359e+00 4.76447642e-01 -2.57150441e-01
-9.61319208e-01 -6.74943864e-01 -1.45744312e+00 -1.02188580e-01
4.33739036e-01 -4.37495530e-01 1.45013463e+00 -1.09637582e+00
-1.18156314e+00 1.08405721e+00 -3.23282331e-01 -5.53219199e-01
-3.87486182e-02 -3.07380021e-01 -1.56915516e-01 2.74044007e-01
8.08311462e-01 4.96877879e-01 6.36252880e-01 -9.55707610e-01
-1.02894855e+00 -4.66310591e-01 1.19969741e-01 5.48437655e-01
-4.17775869e-01 4.91953045e-01 -2.51016200e-01 -6.67172074e-01
-6.08491153e-02 -8.27565134e-01 -3.07866901e-01 -1.53145581e-01
-2.28405017e-02 -8.27969432e-01 5.05915463e-01 -1.11125326e+00
8.77290428e-01 -2.04752612e+00 1.55422091e-01 -4.52472270e-03
-5.91548309e-02 -1.20177604e-01 -4.53799784e-01 1.57642260e-01
-2.20681906e-01 2.80230701e-01 -3.62268090e-01 -5.83590329e-01
1.36958167e-01 8.10746074e-01 -2.40963310e-01 6.04423344e-01
4.22500968e-01 7.63142467e-01 -1.40456986e+00 -5.18180609e-01
-5.60188890e-02 6.46103695e-02 -4.90788430e-01 1.74907610e-01
-4.82033193e-01 3.40076238e-01 -2.34043807e-01 7.45411336e-01
-8.50724876e-02 -1.09225087e-01 1.00195849e+00 4.43013273e-02
-6.68332353e-02 1.27543497e+00 -7.13341415e-01 2.11760831e+00
-8.84343088e-01 2.55875289e-01 1.72580704e-01 -1.31821299e+00
5.87688506e-01 6.02501929e-01 5.94493568e-01 -7.38578141e-01
-6.74959123e-02 5.95951498e-01 -3.47072929e-02 -2.13199168e-01
3.41528922e-01 -3.37934583e-01 -5.37409484e-01 9.65491652e-01
5.50655365e-01 -1.86635870e-02 3.54964167e-01 1.61714926e-01
1.25398433e+00 3.06046247e-01 5.81175208e-01 -3.73137176e-01
7.50284076e-01 3.73352170e-01 8.72470617e-01 2.75461704e-01
-8.80261511e-02 -1.32422969e-01 4.73679245e-01 -4.49459970e-01
-1.29755759e+00 -7.33079493e-01 1.68892927e-02 1.77375698e+00
-1.62856523e-02 -5.38434446e-01 -5.18433392e-01 -1.26811969e+00
-3.58991027e-02 9.27061379e-01 -3.99746299e-01 -7.89344758e-02
-1.16747522e+00 -5.13827562e-01 7.05216408e-01 5.45468509e-01
1.34863146e-02 -9.61140752e-01 -6.27530143e-02 5.83502889e-01
-9.68914106e-02 -1.50608921e+00 -3.26159954e-01 5.44997692e-01
-8.27831686e-01 -1.35206902e+00 -3.20878886e-02 -1.02050591e+00
7.99291968e-01 -1.52402535e-01 1.29572499e+00 -2.10406646e-01
3.03458422e-01 2.04735458e-01 -4.27262485e-01 -7.31672347e-02
-6.74290001e-01 2.39012346e-01 3.35522741e-01 -8.24248865e-02
5.46543300e-01 -4.03295875e-01 -4.85850684e-02 1.73493892e-01
-5.64378858e-01 -1.11351639e-01 4.19599831e-01 1.07254326e+00
7.75337696e-01 -1.83179751e-01 6.00565374e-01 -1.28679967e+00
5.44190705e-01 -7.05968261e-01 -5.40112257e-01 3.65690112e-01
-7.74320900e-01 4.85105217e-01 8.76395881e-01 -2.52055619e-02
-1.28292847e+00 1.51617557e-01 -1.11395761e-01 2.05201089e-01
1.53747052e-01 5.28010309e-01 -6.51748598e-01 4.00595307e-01
5.92627883e-01 1.02023296e-01 -1.56069458e-01 -8.54703724e-01
7.59752393e-01 4.60461557e-01 4.62684959e-01 -8.80046248e-01
6.47128224e-01 3.57119858e-01 -4.60921466e-01 -5.17549217e-01
-1.38202095e+00 -5.72699904e-01 -8.90463173e-01 3.11508507e-01
8.69589150e-01 -1.17840993e+00 1.70610815e-01 1.07051760e-01
-1.29131556e+00 -5.08002877e-01 -5.12200415e-01 3.48659098e-01
-3.62530977e-01 1.45123661e-01 -7.24417984e-01 -2.33993873e-01
-2.21432015e-01 -7.73868144e-01 9.85912502e-01 -2.61018902e-01
-3.86730522e-01 -1.39053404e+00 1.68353289e-01 8.07902813e-01
3.70005295e-02 -4.20977741e-01 1.47205484e+00 -1.27917445e+00
-3.62637371e-01 -1.37070030e-01 -9.68903080e-02 7.47362196e-01
1.57303050e-01 -7.33071327e-01 -7.80725062e-01 -6.30289465e-02
6.28122911e-02 -6.66846812e-01 5.69994271e-01 -1.19653739e-01
7.12728798e-01 -5.46792030e-01 -1.02713548e-01 2.00099960e-01
1.20482135e+00 8.03196952e-02 -7.94088654e-03 3.32904994e-01
9.86150205e-01 9.59030569e-01 8.77456367e-01 -1.87872663e-01
7.67091393e-01 4.99911517e-01 4.10901988e-03 1.70164153e-01
-1.29472420e-01 -6.36884511e-01 6.92835450e-01 1.08464003e+00
2.73870617e-01 1.50208071e-01 -1.06113255e+00 8.21018934e-01
-1.89870632e+00 -5.11682272e-01 2.87123650e-01 1.89399052e+00
1.39482665e+00 -7.58057162e-02 -2.14415312e-01 1.09920382e-01
5.04343748e-01 2.22765684e-01 -6.17840409e-01 -3.54032069e-01
-8.50928575e-02 4.91150111e-01 7.89329529e-01 7.53966093e-01
-1.05276048e+00 1.55520797e+00 5.85191250e+00 7.09379017e-01
-7.03867257e-01 1.04675090e+00 3.96018982e-01 4.09797654e-02
-5.69417417e-01 2.25055531e-01 -7.97035396e-01 2.91128546e-01
1.08048546e+00 -1.31900385e-01 2.53852904e-01 1.08933246e+00
1.18209392e-01 -6.10277392e-02 -1.27887785e+00 8.00659180e-01
2.84553587e-01 -1.28239727e+00 -4.46461253e-02 -1.94951802e-01
3.74538451e-01 4.34139788e-01 -2.82377094e-01 4.05522883e-01
1.02257121e+00 -8.62519145e-01 9.29219007e-01 -2.80318648e-01
1.04407156e+00 -4.87652391e-01 7.35957623e-01 3.54676723e-01
-1.10652912e+00 -1.32397771e-01 -3.06701630e-01 1.15946252e-02
2.10761949e-01 3.15966576e-01 -1.11941576e+00 2.80786872e-01
2.14366958e-01 1.14022923e+00 -4.41870034e-01 4.37582619e-02
-9.75696623e-01 7.16258287e-01 1.99223422e-02 2.68636227e-01
3.64209026e-01 4.04202417e-02 3.57849538e-01 1.07645833e+00
7.08257128e-03 -2.42355853e-01 5.90859354e-01 2.52461135e-01
-5.10472059e-01 3.78554642e-01 -6.22078419e-01 -5.16686559e-01
5.58340430e-01 1.00728869e+00 -3.91961932e-01 -5.23820758e-01
-7.23598719e-01 9.87559438e-01 4.51121956e-01 1.41857630e-02
-4.62857604e-01 1.10240430e-01 8.65281522e-01 1.74227536e-01
-3.46770436e-01 -3.25470716e-01 -3.40832502e-01 -1.33235514e+00
-6.57536909e-02 -9.83610928e-01 6.53918266e-01 -5.47086537e-01
-1.19724870e+00 3.29016656e-01 -2.27232650e-01 -6.95864260e-01
-4.28415000e-01 -8.72536361e-01 9.95499641e-02 8.57831419e-01
-1.90495920e+00 -1.48901582e+00 5.73233545e-01 6.48751616e-01
1.05055070e+00 -5.19455135e-01 1.04734528e+00 4.69124407e-01
-3.77408117e-01 3.35653275e-01 -2.80766994e-01 3.29896390e-01
8.69681597e-01 -1.42661560e+00 4.72526699e-01 8.76873195e-01
2.89038867e-01 6.38360739e-01 4.40496147e-01 -6.07060313e-01
-1.13139176e+00 -1.22952008e+00 1.71673822e+00 -6.96728230e-01
1.21883500e+00 -5.20291448e-01 -7.55097032e-01 8.62685502e-01
1.16329186e-01 9.76379886e-02 1.13829923e+00 1.83788583e-01
-6.73351943e-01 4.82258610e-02 -9.19563115e-01 2.24284694e-01
1.24879754e+00 -1.12058759e+00 -9.65831935e-01 4.65703189e-01
1.09674823e+00 -2.44273022e-02 -6.09946549e-01 -8.91585946e-02
3.53972018e-01 -1.12657011e-01 8.03616524e-01 -9.37731862e-01
4.75822210e-01 -2.31591284e-01 -3.50184649e-01 -1.33184719e+00
-3.24747637e-02 -3.57717037e-01 1.38617933e-01 8.52520227e-01
8.43215227e-01 -4.62345272e-01 4.85246003e-01 7.94675052e-01
-3.35385144e-01 -4.50555652e-01 -1.01988411e+00 -6.79249883e-01
-1.91808090e-01 -6.76419616e-01 3.67753208e-01 1.35593188e+00
-1.45484162e-02 9.30957556e-01 -6.06756359e-02 1.67651638e-01
7.05320001e-01 -1.43572688e-01 2.90468544e-01 -1.18881869e+00
-2.90166080e-01 2.03822881e-01 1.93282142e-01 -6.16295457e-01
1.12004006e+00 -1.48825026e+00 6.99331686e-02 -1.61683786e+00
3.18856686e-02 -1.05719388e+00 -5.21175683e-01 1.21900845e+00
5.97099550e-02 5.28034419e-02 -6.90826029e-02 3.87586445e-01
-7.38262177e-01 4.08187389e-01 7.23495960e-01 -2.26839110e-01
-2.51667723e-02 -4.94319439e-01 -9.49256718e-01 1.13745165e+00
9.16679978e-01 -1.04698837e+00 -4.60897863e-01 -8.52283061e-01
6.05718911e-01 -6.38355315e-02 1.12290420e-01 -3.23323816e-01
5.16316928e-02 -6.27192259e-02 -1.75151154e-01 1.53211772e-01
2.14984685e-01 -7.82336771e-01 -4.83322024e-01 2.24589691e-01
-6.29219830e-01 1.04668766e-01 -2.36742988e-01 5.85086346e-01
-4.84328419e-01 -5.24245918e-01 6.60539567e-01 -4.27928269e-01
-1.11040580e+00 7.70865288e-03 -2.57099599e-01 2.45184273e-01
6.27557158e-01 2.20223442e-01 -2.49350503e-01 1.15392897e-02
-7.82067835e-01 2.63399541e-01 5.78789353e-01 6.48370862e-01
2.37834215e-01 -1.30706751e+00 -5.06520867e-01 1.64329648e-01
3.30904871e-01 -1.36655018e-01 -3.69165838e-01 4.57612067e-01
-1.72444060e-01 5.83754778e-01 -6.07258193e-02 6.12374730e-02
-1.28077996e+00 1.75215170e-01 8.34227130e-02 -7.78675139e-01
-8.13887790e-02 1.04722142e+00 -8.38423297e-02 -9.30581450e-01
-9.00905505e-02 -1.65116891e-01 -3.13188404e-01 3.69698137e-01
2.90101111e-01 -1.26067758e-01 1.57747045e-01 -8.71649742e-01
-4.52330858e-01 -3.46620083e-02 6.88042715e-02 -5.27326286e-01
1.60129809e+00 -6.88253939e-02 -6.27102554e-01 5.32152951e-01
1.10835981e+00 3.23897421e-01 -8.75628829e-01 -6.56792521e-01
8.48398924e-01 -3.05403054e-01 2.37101614e-02 -7.01083720e-01
-5.90753675e-01 5.87725639e-01 8.81209597e-03 -3.22658211e-01
6.49736106e-01 4.23553795e-01 1.05922616e+00 4.74095762e-01
5.66333592e-01 -1.70026791e+00 2.22492427e-01 8.27730536e-01
6.27871156e-01 -1.14354897e+00 -4.18690085e-01 -3.01996529e-01
-6.53144240e-01 6.68791890e-01 4.56080973e-01 1.19708642e-01
5.49922049e-01 2.05708042e-01 1.75146371e-01 -7.49255717e-02
-9.16249871e-01 -2.32230842e-01 -1.24104761e-01 5.05102515e-01
5.27603865e-01 9.78748649e-02 -5.20897090e-01 8.88808548e-01
-1.03042848e-01 -3.21571141e-01 3.06384563e-01 9.68952775e-01
-3.08775365e-01 -2.02444768e+00 4.02202308e-02 1.48844033e-01
-7.57475436e-01 -6.02659523e-01 -2.89063990e-01 4.88734215e-01
4.63300109e-01 7.31519520e-01 1.16322011e-01 -1.12329051e-02
1.00676134e-01 8.51170778e-01 2.97022551e-01 -1.43797719e+00
-4.11288321e-01 -7.72198886e-02 8.22929978e-01 -7.51672447e-01
-6.86166584e-01 -6.35612786e-01 -1.36716342e+00 3.50181907e-01
1.93185255e-01 3.24006230e-01 7.33537495e-01 1.13588130e+00
2.38369197e-01 2.02953056e-01 2.79635578e-01 -2.35815436e-01
-4.68329251e-01 -8.08225036e-01 -4.59870607e-01 3.07096392e-01
6.88254088e-02 -5.83543360e-01 -1.65267587e-01 4.02831286e-01]
|
[10.421709060668945, 9.502047538757324]
|
da08bcaa-01b6-4edf-8a85-2b0ffa722d57
|
neural-sentence-ordering-based-on-constraint
|
2101.11178
| null |
https://arxiv.org/abs/2101.11178v2
|
https://arxiv.org/pdf/2101.11178v2.pdf
|
Neural Sentence Ordering Based on Constraint Graphs
|
Sentence ordering aims at arranging a list of sentences in the correct order. Based on the observation that sentence order at different distances may rely on different types of information, we devise a new approach based on multi-granular orders between sentences. These orders form multiple constraint graphs, which are then encoded by Graph Isomorphism Networks and fused into sentence representations. Finally, sentence order is determined using the order-enhanced sentence representations. Our experiments on five benchmark datasets show that our method outperforms all the existing baselines significantly, achieving a new state-of-the-art performance. The results demonstrate the advantage of considering multiple types of order information and using graph neural networks to integrate sentence content and order information for the task. Our code is available at https://github.com/DaoD/ConstraintGraph4NSO.
|
['Zhicheng Dou', 'Shengchao Liu', 'Jian-Yun Nie', 'Kun Zhou', 'Yutao Zhu']
|
2021-01-27
| null | null | null | null |
['sentence-ordering']
|
['natural-language-processing']
|
[ 1.20442532e-01 -1.62875298e-02 -2.92768776e-01 -7.00286388e-01
-3.44755948e-01 -6.07680321e-01 3.78738433e-01 7.44703710e-01
-2.73133248e-01 4.21148688e-01 7.25887716e-01 -4.65512395e-01
-2.64796197e-01 -8.28696072e-01 -5.80689669e-01 -5.19712865e-02
-2.14886785e-01 5.32577991e-01 2.23413438e-01 -4.37803388e-01
5.66349804e-01 3.38960111e-01 -1.20550013e+00 6.90288365e-01
9.84642088e-01 6.17995203e-01 3.13236326e-01 7.27233708e-01
-2.74427205e-01 8.05429578e-01 -4.78037030e-01 -4.32164937e-01
6.78276345e-02 -4.95071143e-01 -1.03775215e+00 2.36200780e-01
4.80280817e-01 -2.70313501e-01 -6.95170045e-01 1.12785792e+00
1.70679539e-01 3.22221577e-01 3.77314270e-01 -7.24082887e-01
-1.09683120e+00 1.03754306e+00 -1.30712107e-01 4.51242149e-01
9.41910982e-01 -3.82617354e-01 1.77137637e+00 -8.62336040e-01
6.17651701e-01 1.20087743e+00 2.25824088e-01 3.19619656e-01
-1.14230275e+00 -2.35661313e-01 5.30367434e-01 5.66995919e-01
-1.33735907e+00 -3.04879010e-01 1.03063166e+00 -2.08151907e-01
1.36834478e+00 4.14060950e-01 4.99699473e-01 6.98772669e-01
2.59225368e-01 7.85202503e-01 7.81414568e-01 -5.20037413e-01
-1.05894387e-01 -2.79762506e-01 8.03343475e-01 9.01165903e-01
3.06589842e-01 -3.93437445e-01 -6.50862992e-01 8.44982937e-02
1.83075950e-01 6.35959134e-02 -4.99323666e-01 -3.61072570e-02
-1.03281283e+00 8.13889563e-01 7.59104908e-01 5.36884665e-01
-5.79335503e-02 -1.87805191e-01 4.20072824e-01 4.89518166e-01
5.09027660e-01 4.99630421e-01 -3.36825401e-01 2.47716159e-01
-6.06512189e-01 -4.89013530e-02 9.16507900e-01 9.85842049e-01
7.82410741e-01 -5.23568094e-01 -3.24924320e-01 8.58703971e-01
4.48364705e-01 -7.48915896e-02 2.36248359e-01 -5.22484481e-01
1.06780159e+00 9.89708185e-01 -3.56585711e-01 -1.48858583e+00
-6.36683881e-01 -4.35164034e-01 -8.01188588e-01 -5.88909328e-01
7.72087872e-02 1.62397817e-01 -7.25170910e-01 1.62489450e+00
2.20146626e-02 2.76453525e-01 -2.56424528e-02 9.16792810e-01
1.16361403e+00 7.83455491e-01 -2.05185533e-01 -1.46045119e-01
1.36032820e+00 -9.72750425e-01 -7.14203000e-01 -2.70656347e-01
8.09353590e-01 -6.42888129e-01 1.11208630e+00 1.05890259e-01
-1.11442935e+00 -5.67288995e-01 -1.18645084e+00 -4.00957167e-01
-3.60468596e-01 -4.85596284e-02 6.30685568e-01 2.84223169e-01
-1.34250772e+00 9.60034192e-01 -5.74387968e-01 -3.07541132e-01
1.72462150e-01 3.92588049e-01 -3.18744451e-01 -1.90486819e-01
-1.58859122e+00 7.81901658e-01 6.27590179e-01 4.32453871e-01
-1.15955137e-02 -5.60240805e-01 -1.23358703e+00 3.26227963e-01
3.88273776e-01 -9.02244508e-01 8.93031478e-01 -4.16922897e-01
-1.32224536e+00 8.53792369e-01 -4.77461219e-01 -3.64792198e-01
-7.53847286e-02 -1.33183345e-01 -4.53782678e-01 3.14783156e-01
1.56831190e-01 2.83446401e-01 1.85193002e-01 -9.08326447e-01
-3.10317814e-01 -2.87034601e-01 5.53312957e-01 6.23877227e-01
-4.14621860e-01 1.42326206e-01 -6.98631585e-01 -3.07118297e-01
4.51146096e-01 -7.86994874e-01 -1.72191843e-01 -6.34772182e-01
-8.64692092e-01 -6.52001679e-01 2.41111293e-01 -6.71928108e-01
1.65391755e+00 -1.85736012e+00 6.63727939e-01 9.47687328e-02
4.82803792e-01 -2.82665808e-02 -3.13579172e-01 9.33435977e-01
-9.67113897e-02 5.79813540e-01 -2.95610517e-01 -4.91659850e-01
2.03518465e-01 1.94919154e-01 -3.89908068e-03 2.57176965e-01
3.23096246e-01 9.52171981e-01 -9.99880373e-01 -4.84878391e-01
1.03584483e-01 2.97332436e-01 -4.76946980e-01 1.90945193e-01
-1.05889052e-01 3.60028237e-01 -2.57875830e-01 2.81545371e-01
5.07945716e-01 -6.68873072e-01 7.59791613e-01 -1.11905448e-01
2.90836662e-01 9.06183302e-01 -9.50459838e-01 1.71494019e+00
-2.45506555e-01 4.59630549e-01 -4.47874963e-01 -9.56291974e-01
9.32417154e-01 -6.76016062e-02 2.01236218e-01 -7.69626796e-01
4.21598963e-02 -1.46331014e-02 2.37036660e-01 -5.79836488e-01
6.50135219e-01 2.73140132e-01 -3.12201917e-01 3.83350074e-01
-5.22856675e-02 -8.36348999e-03 7.84774899e-01 8.07965696e-01
1.18297637e+00 -3.41763765e-01 3.31774831e-01 -2.45293558e-01
7.52085268e-01 -4.43354070e-01 5.51720023e-01 5.77912867e-01
1.41339973e-01 5.22607565e-01 8.88268650e-01 -4.44204718e-01
-7.43333936e-01 -9.46821034e-01 3.78694311e-02 1.00447011e+00
2.56121129e-01 -9.56767738e-01 -3.26358974e-01 -8.27915311e-01
-1.62618924e-02 7.07438707e-01 -5.52962184e-01 1.83500405e-02
-7.08470881e-01 -3.78673792e-01 1.36639491e-01 4.48184580e-01
1.81959361e-01 -9.74995792e-01 7.04422593e-02 8.45718756e-02
-4.47703511e-01 -1.35067403e+00 -8.93294811e-01 -9.15291458e-02
-8.24009240e-01 -1.16318834e+00 -1.93973169e-01 -9.98858452e-01
9.95329261e-01 4.66458976e-01 1.40689802e+00 6.89368308e-01
9.29947495e-02 2.18931004e-01 -6.90942109e-01 1.10919885e-01
-2.06548661e-01 3.27799678e-01 -1.86548685e-03 4.21953127e-02
3.52926850e-01 -5.50918043e-01 -5.36036670e-01 -1.72834173e-01
-8.08114707e-01 -3.47964279e-02 2.99262136e-01 7.08445787e-01
4.75177526e-01 3.17980424e-02 3.67099822e-01 -1.22567523e+00
1.06986022e+00 -4.39749897e-01 -6.16453409e-01 4.71907109e-01
-4.71952200e-01 2.42639989e-01 7.80124307e-01 1.01832062e-01
-6.18507087e-01 -1.94838524e-01 -6.81975633e-02 -6.10494688e-02
-5.38568310e-02 1.09864163e+00 -2.53589094e-01 2.96160132e-01
2.25102633e-01 1.18977897e-01 -2.67371744e-01 -2.17487842e-01
2.69138336e-01 3.79584134e-01 1.07786708e-01 -3.93554330e-01
5.00775576e-01 1.96288243e-01 1.54009745e-01 -6.72030449e-01
-1.13470209e+00 -6.36058688e-01 -1.07427955e+00 -1.14833511e-01
7.51121998e-01 -5.74494660e-01 -6.00907564e-01 1.03813328e-01
-1.57095432e+00 -1.03787884e-01 3.08716208e-01 3.94471765e-01
-7.62673989e-02 9.68636036e-01 -9.09418166e-01 -4.34228122e-01
-3.32236171e-01 -9.51250792e-01 7.69943714e-01 1.50938004e-01
-1.48948297e-01 -1.24730480e+00 1.56773701e-02 4.16959882e-01
-4.54956442e-02 -1.16554216e-01 8.68639290e-01 -9.50662196e-01
-6.86066389e-01 -2.11280555e-01 -2.57104427e-01 1.01300836e-01
2.03212857e-01 -2.16484964e-02 -2.63667285e-01 -2.82031804e-01
-1.14136368e-01 -1.07250303e-01 1.19696259e+00 2.90967584e-01
1.03633308e+00 -3.82475108e-01 -1.68665290e-01 5.60697258e-01
1.42768097e+00 -1.22944795e-01 5.57549834e-01 1.60558388e-01
1.00219214e+00 6.31768405e-01 3.19101542e-01 3.12581182e-01
8.33050370e-01 5.35495520e-01 3.56962979e-01 2.63298661e-01
-1.12687685e-01 -1.98592275e-01 9.66949090e-02 1.57971847e+00
2.40859985e-01 -5.78951597e-01 -9.81353104e-01 4.46808308e-01
-1.99691057e+00 -9.82553899e-01 -4.61519629e-01 1.93075347e+00
7.01837361e-01 3.89859825e-01 -1.26670659e-01 7.56953433e-02
9.29052770e-01 4.15186167e-01 -1.78153440e-01 -6.79648101e-01
-1.12583809e-01 1.18088117e-02 1.82421193e-01 1.02774775e+00
-1.16042137e+00 9.34716523e-01 5.86148405e+00 2.76066154e-01
-7.00729370e-01 -3.14946443e-01 6.20117843e-01 7.93222114e-02
-5.98697662e-01 1.36456251e-01 -8.20353150e-01 4.53831196e-01
8.66831243e-01 -3.97389442e-01 5.84218204e-01 2.74701655e-01
2.37043694e-01 1.02574408e-01 -1.15769219e+00 7.40958095e-01
4.79482412e-01 -1.55022788e+00 9.50793102e-02 -2.25608096e-01
4.80537653e-01 1.25777945e-01 -2.73086220e-01 1.73494309e-01
2.40638003e-01 -8.59718740e-01 3.78469318e-01 5.46297610e-01
4.08448040e-01 -6.37390912e-01 8.33720684e-01 2.05600664e-01
-1.55262601e+00 5.58833517e-02 -5.51095724e-01 -5.51857650e-01
3.29484493e-01 7.75382102e-01 -8.23820949e-01 1.18550670e+00
4.19088840e-01 1.24866998e+00 -8.03271294e-01 8.93090367e-01
-7.55004048e-01 3.84432495e-01 -8.15688074e-02 -5.37919700e-01
9.38605070e-02 -3.58245075e-01 3.79644752e-01 1.26696265e+00
6.23758920e-02 2.64287561e-01 4.11162496e-01 6.33039474e-01
-4.26285714e-01 2.85448581e-01 -7.26594687e-01 -5.71565814e-02
6.40377164e-01 1.27846396e+00 -8.58640134e-01 -2.66728044e-01
-6.27699077e-01 1.00767028e+00 9.64095235e-01 2.48611152e-01
-7.64098644e-01 -4.99181509e-01 2.50263453e-01 -3.16247314e-01
4.06449676e-01 -5.96983790e-01 -3.16368848e-01 -1.43744147e+00
4.19909716e-01 -5.07180154e-01 7.34071314e-01 -5.08714020e-01
-1.61405683e+00 8.60377669e-01 -8.24377686e-02 -1.18569851e+00
7.48789823e-03 -6.81451142e-01 -7.43782878e-01 6.67933464e-01
-1.57385588e+00 -7.91738391e-01 -5.75548373e-02 2.88086116e-01
5.51295459e-01 2.00959414e-01 7.67795980e-01 2.12019295e-01
-6.51746213e-01 4.82628703e-01 -1.93923995e-01 4.46558774e-01
2.82751381e-01 -1.39161205e+00 5.96317112e-01 9.95059252e-01
5.86061478e-01 7.59081244e-01 3.92476350e-01 -7.14494050e-01
-1.37739229e+00 -8.54980826e-01 1.65752041e+00 -4.07533467e-01
9.36712563e-01 -5.92863739e-01 -1.09801948e+00 6.03982210e-01
5.95291674e-01 -1.62313089e-01 8.54947746e-01 6.45946801e-01
-4.19713587e-01 -1.92756820e-02 -4.66942072e-01 7.51044273e-01
1.38976443e+00 -6.97262049e-01 -9.01701987e-01 6.38859272e-01
1.15620327e+00 -5.59328020e-01 -8.68245363e-01 2.66531795e-01
6.06247317e-03 -9.24700499e-01 7.47835875e-01 -7.53744543e-01
6.20210588e-01 -1.77539825e-01 -2.08824277e-01 -1.32977939e+00
-7.92295694e-01 -5.06023526e-01 -1.28215626e-01 1.11707735e+00
8.33092034e-01 -6.86219513e-01 6.30268097e-01 5.02842247e-01
-3.51827472e-01 -1.00284052e+00 -8.95472109e-01 -8.67523849e-01
-9.81404856e-02 -1.59854591e-01 7.55091608e-01 1.00865746e+00
5.08701980e-01 9.34864879e-01 -1.29062086e-01 3.80881667e-01
3.27447534e-01 4.34182167e-01 2.44636267e-01 -1.16971302e+00
-2.56791711e-01 -5.82967877e-01 -4.71309185e-01 -1.18199956e+00
6.44222796e-01 -1.51451814e+00 -5.39462827e-02 -2.26839137e+00
2.93906569e-01 8.83447528e-02 -5.51442325e-01 4.28626388e-01
-4.99505103e-01 -1.36024252e-01 4.20161366e-01 9.43807587e-02
-1.06452656e+00 5.72510660e-01 1.25290620e+00 -3.03847671e-01
-8.74500498e-02 -1.88134387e-01 -8.29437613e-01 4.11944538e-01
1.21803188e+00 -4.49751586e-01 -4.95916128e-01 -9.31199133e-01
4.93630081e-01 3.10375877e-02 1.93754956e-02 -8.28461349e-01
6.85197115e-01 -2.87874527e-02 1.44573867e-01 -6.86478078e-01
2.10828409e-01 -6.07146144e-01 -3.65366757e-01 3.83375883e-01
-6.22852862e-01 4.13705885e-01 9.13806036e-02 6.20789766e-01
-3.44175607e-01 -3.44084173e-01 1.97387874e-01 -7.84794614e-02
-6.54703021e-01 3.36918473e-01 -4.36532199e-02 6.10469505e-02
6.39087200e-01 2.08644588e-02 -6.26064301e-01 -4.33951706e-01
-7.58878946e-01 4.60151523e-01 1.79920703e-01 5.42445481e-01
9.38989580e-01 -1.29694414e+00 -8.93174231e-01 6.45984262e-02
2.73087900e-02 -1.34103313e-01 2.00863689e-01 6.36088908e-01
-3.92779678e-01 5.04975557e-01 1.83648184e-01 -5.88386834e-01
-1.41933620e+00 6.47195041e-01 1.05906710e-01 -6.58247232e-01
-5.36620915e-01 8.93338501e-01 -1.76535755e-01 -6.21236563e-01
3.42128589e-03 -3.68328750e-01 -5.21305144e-01 -5.21142073e-02
3.14360529e-01 -4.79466096e-02 1.25405326e-01 -6.16040826e-01
-6.06927037e-01 6.04413807e-01 -3.91210407e-01 2.02587590e-01
1.36127126e+00 -2.07101241e-01 -7.53703475e-01 4.24063265e-01
1.32383108e+00 6.80249408e-02 -6.73692822e-01 -3.21661711e-01
4.58480299e-01 -5.70433915e-01 -3.16009879e-01 -4.20281559e-01
-8.25908542e-01 6.30968988e-01 -3.34003150e-01 7.21002221e-01
1.15496278e+00 1.70007855e-01 5.57764709e-01 5.38688481e-01
1.41879246e-01 -8.84789646e-01 2.90337235e-01 1.02350497e+00
1.14814937e+00 -1.25817800e+00 2.91578919e-01 -9.54726160e-01
-6.61820948e-01 1.25572813e+00 5.60590386e-01 -3.41143548e-01
6.76550388e-01 -6.86105713e-02 -1.39772043e-01 -2.12372005e-01
-9.04100776e-01 -3.75113428e-01 7.54779458e-01 2.08435476e-01
7.92875290e-01 8.33939388e-02 -6.88814878e-01 5.39793670e-01
-1.52839199e-01 -4.21986878e-01 5.68829179e-01 7.61533558e-01
-3.99634898e-01 -1.53576207e+00 6.84188306e-02 6.00639641e-01
-1.53316975e-01 -5.92925191e-01 -7.76786745e-01 4.37359303e-01
-4.51641977e-01 1.40105164e+00 5.91478013e-02 -5.79559922e-01
4.30039763e-01 -1.60059005e-01 4.08900082e-01 -1.02802002e+00
-5.59286654e-01 -1.97894290e-01 4.51389164e-01 -5.88119507e-01
-2.29505122e-01 -5.44638455e-01 -1.55902791e+00 -3.70678782e-01
-1.86808586e-01 1.65188611e-01 2.21090347e-01 8.87574911e-01
7.54983306e-01 8.50229621e-01 8.33362877e-01 -5.93939364e-01
-3.05016756e-01 -8.31450582e-01 -3.89833540e-01 7.26444960e-01
1.46073699e-01 -2.85341918e-01 -4.23373848e-01 -3.07687759e-01]
|
[11.119565963745117, 8.78867244720459]
|
2e53ef37-dba4-4f5e-aef4-f91b466ac54b
|
stau-a-spatiotemporal-aware-unit-for-video
|
2204.09456
| null |
https://arxiv.org/abs/2204.09456v1
|
https://arxiv.org/pdf/2204.09456v1.pdf
|
STAU: A SpatioTemporal-Aware Unit for Video Prediction and Beyond
|
Video prediction aims to predict future frames by modeling the complex spatiotemporal dynamics in videos. However, most of the existing methods only model the temporal information and the spatial information for videos in an independent manner but haven't fully explored the correlations between both terms. In this paper, we propose a SpatioTemporal-Aware Unit (STAU) for video prediction and beyond by exploring the significant spatiotemporal correlations in videos. On the one hand, the motion-aware attention weights are learned from the spatial states to help aggregate the temporal states in the temporal domain. On the other hand, the appearance-aware attention weights are learned from the temporal states to help aggregate the spatial states in the spatial domain. In this way, the temporal information and the spatial information can be greatly aware of each other in both domains, during which, the spatiotemporal receptive field can also be greatly broadened for more reliable spatiotemporal modeling. Experiments are not only conducted on traditional video prediction tasks but also other tasks beyond video prediction, including the early action recognition and object detection tasks. Experimental results show that our STAU can outperform other methods on all tasks in terms of performance and computation efficiency.
|
['Wen Gao', 'Siwei Ma', 'Shanshe Wang', 'Xinfeng Zhang', 'Zheng Chang']
|
2022-04-20
| null | null | null | null |
['video-prediction']
|
['computer-vision']
|
[-7.23673552e-02 -4.52071548e-01 -4.76425231e-01 -2.75607347e-01
8.61946680e-03 -6.85967803e-02 3.42527896e-01 -3.20792407e-01
-1.34369388e-01 4.56727594e-01 5.40008187e-01 6.84068128e-02
3.39433588e-02 -4.64702159e-01 -6.47729456e-01 -9.99382794e-01
-2.93240875e-01 -3.14799964e-01 1.03265965e+00 7.68871792e-03
2.26477340e-01 1.45061463e-01 -1.40494120e+00 3.86209518e-01
7.59053767e-01 1.34636569e+00 6.42820239e-01 3.00759822e-01
-4.29250561e-02 1.34288633e+00 5.36449347e-03 3.54374945e-01
5.71751930e-02 -4.10451114e-01 -3.37956488e-01 1.80775762e-01
1.78013563e-01 -6.50986671e-01 -9.27064478e-01 9.96170640e-01
1.69366300e-01 4.99943256e-01 1.21709131e-01 -1.09999323e+00
-6.74394727e-01 2.51898617e-01 -5.93459129e-01 6.50473475e-01
2.43462279e-01 4.21698928e-01 7.54319906e-01 -7.44382620e-01
5.01506269e-01 1.30956495e+00 2.16027573e-01 4.49207574e-01
-6.12823367e-01 -7.89198041e-01 9.77536380e-01 9.15150940e-01
-1.18752706e+00 -2.88635522e-01 9.33438599e-01 -5.68520069e-01
9.31767523e-01 2.96575613e-02 1.01090944e+00 7.34144509e-01
2.12102979e-01 1.23039961e+00 6.81424618e-01 8.85155573e-02
-1.09355533e-02 -3.22651029e-01 -9.94009525e-02 6.69903398e-01
-4.22584593e-01 1.58043340e-01 -5.79368472e-01 3.01319033e-01
1.05523837e+00 5.16116738e-01 -4.58689213e-01 -2.69486487e-01
-1.54816449e+00 3.46995920e-01 4.54495400e-01 3.92414719e-01
-6.13438129e-01 2.58437514e-01 4.65685785e-01 -1.48279816e-01
4.53235924e-01 2.14906316e-02 -4.09742534e-01 -2.23504186e-01
-8.02761018e-01 -1.08329073e-01 -4.95246612e-03 1.08940852e+00
9.17662978e-01 1.71401843e-01 -6.17239118e-01 5.83641291e-01
3.93227369e-01 1.57672077e-01 6.27242923e-01 -1.07688904e+00
7.62052059e-01 7.30175972e-01 4.32317823e-01 -1.23340487e+00
-1.63978726e-01 -7.56201595e-02 -8.06982100e-01 -1.96387872e-01
1.96887404e-01 -7.06297234e-02 -8.94256055e-01 1.79687476e+00
2.94341326e-01 1.08302343e+00 -2.08431780e-01 1.17368710e+00
7.39462674e-01 1.16431785e+00 3.75523150e-01 -6.91904008e-01
1.29170096e+00 -1.32076061e+00 -9.01252270e-01 -2.72759110e-01
4.49274123e-01 -5.78028202e-01 9.07385707e-01 1.26646742e-01
-1.10899043e+00 -1.09285200e+00 -8.87275338e-01 -3.66487005e-03
9.72686037e-02 2.97712982e-01 5.52220881e-01 -1.57528445e-01
-8.94539773e-01 4.23034072e-01 -1.15239322e+00 -2.27130517e-01
4.26873356e-01 3.54782969e-01 -7.61199743e-02 -1.59168124e-01
-1.38133752e+00 5.13965964e-01 6.09978914e-01 2.65836537e-01
-6.99964643e-01 -4.33120400e-01 -6.83476985e-01 2.51725435e-01
5.86220503e-01 -2.28585899e-01 9.13580060e-01 -1.14209712e+00
-1.18932211e+00 4.76866812e-02 -6.96985424e-01 -1.83246225e-01
2.63776392e-01 -1.17841475e-01 -4.51531380e-01 2.77351618e-01
-6.18239865e-02 8.56780112e-01 5.31627059e-01 -8.36860895e-01
-1.10366786e+00 -1.78660333e-01 5.90445735e-02 4.71379131e-01
-6.69297159e-01 -1.06310032e-01 -9.66157198e-01 -8.60430419e-01
9.82743800e-02 -6.50586188e-01 -2.75412351e-01 1.72622874e-01
4.82081994e-02 -4.57375497e-01 1.25357354e+00 -8.49574447e-01
1.62062097e+00 -2.30632448e+00 4.00885671e-01 -1.27883285e-01
1.07227758e-01 4.54557329e-01 -2.23973364e-01 1.36564925e-01
-1.03896372e-01 -1.39454007e-01 2.48149768e-01 6.75017461e-02
-4.77235973e-01 3.07790309e-01 -4.59445924e-01 1.35390922e-01
9.52358544e-02 8.99727106e-01 -1.17733014e+00 -7.67302811e-01
4.81553674e-01 2.22090945e-01 -4.59059536e-01 3.71280193e-01
-3.95695239e-01 7.55131125e-01 -1.01563370e+00 3.60770106e-01
4.92646873e-01 -2.69445121e-01 8.28794315e-02 -4.24525470e-01
-3.11944872e-01 1.40275255e-01 -9.84837830e-01 1.57697833e+00
-1.95207760e-01 6.02967203e-01 -3.00141126e-01 -1.08866739e+00
7.94726074e-01 4.61561829e-01 7.67831624e-01 -9.79923010e-01
-1.93818510e-01 -9.82895046e-02 1.19081929e-01 -8.86739552e-01
3.44339669e-01 2.97580902e-02 3.69125366e-01 2.63878614e-01
-1.20964237e-01 7.25835085e-01 1.66496515e-01 8.01576898e-02
8.19560289e-01 6.06759429e-01 1.97041288e-01 -5.58759496e-02
7.00802505e-01 -1.71508983e-01 1.22519374e+00 2.77022213e-01
-6.69966340e-01 3.70179236e-01 4.04569298e-01 -7.50849783e-01
-8.02820385e-01 -7.78516352e-01 3.11907172e-01 1.08981037e+00
7.72123635e-01 -4.58149701e-01 -4.13061410e-01 -6.47274911e-01
-3.84065360e-01 4.58612561e-01 -7.62554646e-01 -2.84477025e-01
-8.87276769e-01 -3.99604052e-01 -1.54925793e-01 9.75936174e-01
7.40989447e-01 -1.32940674e+00 -4.83388007e-01 4.38148171e-01
-3.92176300e-01 -1.21713269e+00 -9.66562092e-01 -5.33631325e-01
-9.79787052e-01 -9.42031085e-01 -8.34610105e-01 -8.84890914e-01
4.05300200e-01 6.50154650e-01 5.58200955e-01 3.20983350e-01
4.51887995e-02 2.49432877e-01 -5.89775205e-01 -9.36824903e-02
2.70071507e-01 -4.25684571e-01 3.50023843e-02 4.17271525e-01
4.70880330e-01 -6.55318677e-01 -8.96922708e-01 6.74149513e-01
-7.17738092e-01 4.23631459e-01 4.55703467e-01 7.40697682e-01
6.94545150e-01 1.89598516e-01 4.27336484e-01 -2.73980558e-01
1.00791054e-02 -6.41971171e-01 -3.93100858e-01 5.53101480e-01
9.66933556e-03 -1.21124953e-01 6.43925965e-01 -8.24021995e-01
-1.33520377e+00 1.43345073e-01 6.90314472e-02 -7.97949553e-01
-1.67891502e-01 4.17603284e-01 -3.63016307e-01 2.43452191e-01
-1.53352425e-01 8.53572190e-01 -2.86242157e-01 -2.82838315e-01
-1.07300855e-01 2.95884758e-01 2.13731915e-01 -3.33001912e-01
3.49002033e-01 3.97396207e-01 -1.05591759e-01 -6.50486052e-01
-6.93642318e-01 -4.43345994e-01 -7.12005436e-01 -5.55690050e-01
1.25683355e+00 -1.06542587e+00 -5.92789829e-01 5.79527617e-01
-1.05914462e+00 -4.04013515e-01 1.11810148e-01 8.57071817e-01
-5.65711498e-01 8.01592112e-01 -7.84991443e-01 -8.10795367e-01
1.49660021e-01 -1.37237418e+00 9.60834444e-01 4.76467937e-01
2.24430449e-02 -9.88937378e-01 -1.51176319e-01 1.68848857e-01
8.44635740e-02 -1.63388968e-01 8.23077679e-01 -1.67244777e-01
-1.08675313e+00 4.27998006e-02 -5.14326394e-01 3.19184065e-02
2.60352641e-01 1.40487805e-01 -6.20990634e-01 -7.22785741e-02
-3.55434455e-02 7.74958283e-02 9.45011318e-01 6.58783913e-01
1.51547611e+00 -3.19519371e-01 -5.08623719e-01 4.50756907e-01
9.85561848e-01 8.57172906e-01 8.59375656e-01 1.45896971e-01
8.68483722e-01 6.71317816e-01 1.12219965e+00 4.23020422e-01
4.06418115e-01 1.00182188e+00 4.09285158e-01 6.37300834e-02
-2.17184052e-03 -4.31104720e-01 4.67050046e-01 7.69937873e-01
-5.40741384e-01 -9.18624774e-02 -6.37980819e-01 4.31211919e-01
-2.43763828e+00 -1.40469921e+00 -3.70398872e-02 1.95261371e+00
4.98137087e-01 -5.84014207e-02 5.19130602e-02 -3.89615387e-01
8.52954209e-01 4.38381404e-01 -7.92112589e-01 1.83121502e-01
6.59571886e-02 -5.86634576e-01 -1.13236383e-02 1.97182536e-01
-1.14987957e+00 9.32362080e-01 5.45416880e+00 1.00743234e+00
-1.00496876e+00 5.51718175e-02 8.69655073e-01 -3.47102523e-01
-1.39910176e-01 1.19719759e-01 -6.38843060e-01 9.70808864e-01
3.44309419e-01 8.32717270e-02 3.12879205e-01 7.90722489e-01
5.32812953e-01 -9.96092111e-02 -9.25268054e-01 1.03001821e+00
-1.62408724e-01 -1.26987207e+00 2.27772266e-01 -1.87708866e-02
7.40273654e-01 -2.16243044e-01 -4.11733352e-02 3.38042706e-01
-2.82214489e-02 -7.42728293e-01 7.47393429e-01 9.93263245e-01
4.31104302e-01 -6.11735702e-01 6.29661024e-01 6.12167358e-01
-1.83518016e+00 -2.73404539e-01 -4.99786437e-01 -1.48791954e-01
4.27859247e-01 1.64590731e-01 1.09555595e-01 4.11730528e-01
8.03072751e-01 1.57240391e+00 -3.88525277e-01 1.06628168e+00
-9.85201895e-02 6.19143546e-01 -4.75952066e-02 5.41116893e-02
5.04946113e-01 -4.53472346e-01 3.17625582e-01 9.12129998e-01
6.76132679e-01 6.92000985e-01 5.82420170e-01 5.77785432e-01
3.90375406e-01 5.08043654e-02 -3.43859494e-01 6.17372878e-02
3.32942307e-01 9.23105180e-01 -5.14394879e-01 -5.61225712e-01
-7.75406659e-01 8.71236026e-01 1.98307037e-01 6.92318320e-01
-1.08166492e+00 -2.98331911e-03 7.15661883e-01 7.60009512e-02
5.61263800e-01 -3.52338165e-01 1.97440788e-01 -1.49072158e+00
2.13922963e-01 -2.99238563e-01 3.80953729e-01 -1.02324307e+00
-1.02047265e+00 4.00900543e-01 1.07018240e-01 -1.73788679e+00
-3.96805406e-02 -4.21336055e-01 -1.02015829e+00 8.94090772e-01
-1.36948752e+00 -1.02547657e+00 -3.38198543e-01 6.99531615e-01
8.40627551e-01 -1.20483339e-01 2.47648507e-01 2.40877017e-01
-8.63799155e-01 2.24444002e-01 -5.45568466e-02 1.50985524e-01
5.48794448e-01 -7.51725733e-01 5.38434368e-03 9.99246001e-01
-9.17146727e-02 5.48346221e-01 3.45301986e-01 -7.98714221e-01
-1.01445293e+00 -1.08023798e+00 6.32429421e-01 -5.45261279e-02
6.53228045e-01 5.86468726e-02 -1.21741092e+00 7.01827884e-01
1.46440789e-01 3.23395163e-01 5.03857434e-01 -1.42654955e-01
-1.58659354e-01 -2.50880271e-01 -4.27556425e-01 7.52406120e-01
1.23480308e+00 -3.78274411e-01 -3.75680476e-01 2.41980508e-01
6.20409608e-01 -1.61328286e-01 -7.19328344e-01 4.56893981e-01
6.12491548e-01 -1.10748792e+00 9.83409226e-01 -6.62277043e-01
6.37754202e-01 -6.44458592e-01 -1.18790038e-01 -8.46972167e-01
-7.78035343e-01 -1.37223050e-01 -4.77514237e-01 1.17053449e+00
-1.39660060e-01 -3.49742144e-01 7.55458772e-01 6.45720005e-01
-2.86533684e-01 -1.10803699e+00 -1.00912929e+00 -5.81484735e-01
-3.55608016e-01 -3.77293110e-01 4.90353972e-01 7.08773971e-01
9.76367146e-02 1.59954503e-01 -8.98710012e-01 8.71468186e-02
1.16982944e-01 4.45565015e-01 3.67361277e-01 -6.98187590e-01
-3.53360951e-01 -4.90411460e-01 -4.72900391e-01 -1.78430569e+00
2.67460030e-02 -2.00458258e-01 1.32970080e-01 -1.51626241e+00
5.64983189e-01 -2.87226349e-01 -6.95034087e-01 3.97607177e-01
-6.62910700e-01 -9.12420377e-02 3.03013623e-01 5.72009563e-01
-1.07645893e+00 8.90422046e-01 1.59393108e+00 -3.68743353e-02
-1.25557825e-01 -7.42564648e-02 -1.75291345e-01 8.96873951e-01
5.37533939e-01 6.96558505e-02 -7.71784306e-01 -4.80342925e-01
-2.01383993e-01 4.38499182e-01 3.63655031e-01 -1.20318365e+00
2.80903727e-01 -7.63577402e-01 6.31385386e-01 -7.53450871e-01
4.80567485e-01 -8.41395974e-01 -1.53431743e-01 4.50245172e-01
-2.49490678e-01 -7.53216073e-02 1.17122829e-01 1.04513288e+00
-4.61883008e-01 2.61710823e-01 5.41145325e-01 -1.39095366e-01
-1.51883042e+00 9.28835928e-01 -5.37825584e-01 -3.17348689e-01
1.42926133e+00 -4.95098889e-01 -6.57943711e-02 -5.41747093e-01
-8.73241603e-01 5.65373480e-01 3.16201925e-01 6.24208570e-01
8.25369477e-01 -1.56433403e+00 -2.56647944e-01 2.61095375e-01
4.38561291e-02 -1.92209646e-01 1.05678809e+00 1.08765650e+00
-2.22708464e-01 4.10609961e-01 -5.30255318e-01 -6.01838350e-01
-1.30181408e+00 1.01365554e+00 2.89786994e-01 -2.38798440e-01
-4.27368194e-01 6.82980776e-01 1.00345790e+00 2.79874861e-01
2.00198278e-01 -3.80398661e-01 -6.92185760e-01 -2.87414305e-02
8.98375928e-01 2.18783557e-01 -8.26858044e-01 -8.70622337e-01
-2.86147267e-01 7.44975805e-01 3.77752222e-02 3.59621614e-01
1.21984947e+00 -5.00180364e-01 -5.11465818e-02 4.10485983e-01
1.03151298e+00 -3.49309623e-01 -1.89224684e+00 -4.47950602e-01
-3.26239169e-01 -7.42994487e-01 -7.71685988e-02 -3.13415587e-01
-1.47228348e+00 1.31545007e+00 4.70184535e-01 1.10853106e-01
1.49830997e+00 -1.72459781e-01 1.00240624e+00 5.36272302e-02
4.59280133e-01 -9.69807029e-01 3.58000278e-01 5.17050266e-01
6.07692599e-01 -1.10322952e+00 -6.92880750e-02 -5.15185535e-01
-9.91032958e-01 1.15381920e+00 1.17941070e+00 -7.33761042e-02
7.51407027e-01 -1.72135070e-01 -3.27882558e-01 2.15588018e-01
-9.87516046e-01 -1.55513197e-01 6.23308480e-01 5.38788974e-01
4.04060572e-01 -2.29016498e-01 -8.84691700e-02 7.29778707e-01
7.02670038e-01 7.40902051e-02 -1.66096035e-02 6.59039497e-01
-5.79903305e-01 -9.68414545e-01 -8.82225931e-02 4.65157181e-01
-9.65809897e-02 -4.61577438e-02 3.49196084e-02 4.53326166e-01
5.52646220e-01 7.21538186e-01 2.66212136e-01 -7.05099702e-01
1.27281308e-01 -2.11520150e-01 2.31022581e-01 -4.05155152e-01
1.55788632e-02 3.81290764e-01 -1.39665291e-01 -8.30863416e-01
-6.74589038e-01 -7.54075408e-01 -1.20505083e+00 -8.00286606e-02
-7.87261799e-02 2.84771509e-02 -2.46928230e-01 1.13594508e+00
4.14712727e-01 5.59484959e-01 3.89237642e-01 -9.78906691e-01
4.55421321e-02 -9.62696791e-01 -5.91377020e-01 3.77619267e-01
1.26551732e-01 -9.71697569e-01 2.95606673e-01 1.74534112e-01]
|
[8.641632080078125, 0.4760168194770813]
|
1283242b-4afe-412f-aa45-19ce8508d97b
|
mt-quality-estimation-the-cmu-system-for
| null | null |
https://aclanthology.info/papers/W13-2246/w13-2246
|
https://www.aclweb.org/anthology/W13-2246
|
MT Quality Estimation: The CMU System for WMT’13
| null |
['Stephan Vogel', 'Silja Hildebrand']
|
2013-08-01
| null | null | null |
ws-2013-8
|
['cross-lingual-information-retrieval']
|
['natural-language-processing']
|
[-2.44508207e-01 3.89024585e-01 -2.65282035e-01 -2.15905145e-01
-8.60921741e-02 -7.76765764e-01 4.48510379e-01 -7.23253429e-01
-5.48377395e-01 1.31954515e+00 3.66348401e-02 -9.49533224e-01
-2.40340635e-01 -1.05564880e+00 -8.44053447e-01 -8.75781775e-01
-7.42435038e-01 6.86515033e-01 1.44298598e-01 -6.52004302e-01
8.47113907e-01 6.29996777e-01 -1.62287033e+00 5.77558100e-01
6.81926727e-01 5.49597681e-01 5.66466339e-02 1.04480565e+00
6.13576770e-02 1.59847176e+00 -6.05450153e-01 -4.56729174e-01
1.43710867e-01 -1.88022718e-01 -5.35770595e-01 -3.36825520e-01
7.91536197e-02 -5.88986814e-01 -3.10633808e-01 6.07356608e-01
1.14270830e+00 -5.53557873e-02 1.07025254e+00 -1.51067197e+00
-7.97060013e-01 5.44234335e-01 2.30399013e-01 1.20632313e-01
6.10446513e-01 -1.79472014e-01 3.51212233e-01 -1.36621380e+00
6.32680655e-01 8.64870071e-01 9.47046995e-01 5.99273622e-01
-1.20647264e+00 -5.60552299e-01 -7.17137158e-01 -2.57835120e-01
-1.59567785e+00 -6.30042374e-01 6.23617955e-02 -3.34735900e-01
1.69142139e+00 6.72587395e-01 1.35647905e+00 1.47341585e+00
1.36649036e+00 4.72517729e-01 1.13904178e+00 -3.16340059e-01
3.65351558e-01 6.43754780e-01 -2.29148138e-02 5.51889002e-01
1.19686353e+00 6.76312149e-01 -5.60917914e-01 -7.92779744e-01
1.11365557e+00 -3.12949389e-01 2.60771513e-01 -7.06989706e-01
-1.06566191e+00 7.88466871e-01 -1.33808568e-01 2.20882341e-01
-3.46203089e-01 -9.92989019e-02 2.24548295e-01 4.12961632e-01
-2.94082850e-01 4.77862865e-01 -8.89795244e-01 -1.87686309e-01
-1.01472771e+00 1.91863477e-01 1.30615628e+00 1.45811689e+00
4.72722203e-03 3.70100170e-01 1.66493103e-01 2.89523482e-01
5.29429734e-01 1.05927479e+00 4.36951250e-01 -1.33365822e+00
-9.55119133e-02 4.20878716e-02 6.12505674e-01 -1.13473701e+00
-7.12930143e-01 -2.41300568e-01 -9.54724610e-01 5.18771529e-01
-1.59270123e-01 -4.59180593e-01 -7.59686172e-01 4.79171664e-01
-1.81984559e-01 -3.19489211e-01 5.45619786e-01 2.69257158e-01
7.28577793e-01 1.45816103e-01 9.25893057e-03 -4.73371416e-01
8.38296890e-01 -1.26899445e+00 -1.11426139e+00 2.33304992e-01
6.82289064e-01 -1.11704338e+00 4.06794518e-01 4.29345578e-01
-1.87200487e+00 -9.69799384e-02 -1.08317137e+00 1.19022481e-01
-7.97166348e-01 -2.57194549e-01 6.66624963e-01 1.49922979e+00
-1.63972652e+00 6.47176981e-01 -6.05021298e-01 5.40335439e-02
-4.73218322e-01 9.07614470e-01 -3.47142309e-01 2.90655226e-01
-1.28751957e+00 9.52449858e-01 -1.04219764e-01 3.03568155e-01
-3.30324143e-01 -1.84501112e-01 -1.03385210e+00 -6.24127567e-01
-2.59417385e-01 -1.07312346e+00 1.36302292e+00 1.09815992e-01
-1.30379498e+00 1.00962746e+00 -4.14315462e-01 -1.14109404e-01
6.61359608e-01 -1.52096570e-01 -8.18174660e-01 1.58321753e-01
2.87593510e-02 3.47163439e-01 5.30214727e-01 -9.69232559e-01
-6.31655395e-01 -1.00153096e-01 -2.54188716e-01 2.48816490e-01
1.43904146e-03 2.05205917e-01 6.38460100e-01 -9.53604504e-02
1.91062525e-01 -8.61961663e-01 -3.21419090e-01 -7.61404037e-01
-1.77353188e-01 -3.84736449e-01 5.59762537e-01 -2.92163283e-01
1.33848476e+00 -1.84365213e+00 -4.54683900e-01 2.94456817e-03
2.53527164e-01 3.89016122e-02 2.33101144e-01 1.23831403e+00
-2.43849114e-01 1.06667292e+00 3.98064762e-01 9.52541176e-03
3.13382268e-01 5.07273376e-01 2.86674555e-02 2.51206756e-01
9.43330750e-02 1.15280068e+00 -1.09214199e+00 -5.20940363e-01
5.57267785e-01 4.60968643e-01 -3.84588838e-01 6.24593735e-01
9.30012882e-01 2.99172718e-02 -8.94021094e-02 1.33241999e+00
1.05377018e+00 -2.90812284e-01 -2.46628653e-02 3.44544262e-01
-6.44541264e-01 6.00222833e-02 -7.41804183e-01 9.49429095e-01
1.31349964e-02 1.99757457e-01 1.74998537e-01 -6.80728197e-01
3.94654661e-01 1.10545933e+00 4.45659161e-01 -9.24297392e-01
-1.69569388e-01 9.00310338e-01 1.03920102e-01 -9.62983370e-01
6.40755475e-01 -1.72503099e-01 7.69825354e-02 1.40852645e-01
-3.61913115e-01 -3.84120375e-01 5.52458428e-02 4.19574350e-01
1.00224769e+00 -1.33707494e-01 5.48685491e-01 -8.72328460e-01
6.88946426e-01 -2.01589778e-01 3.00597936e-01 1.23978651e+00
-2.90862411e-01 9.21478033e-01 3.19022417e-01 -5.37574828e-01
-5.72780192e-01 -1.20098352e+00 -4.80556756e-01 9.86326516e-01
-1.58388391e-01 -3.87671381e-01 -7.10155964e-01 -2.63155133e-01
-1.85071185e-01 4.93483543e-01 -5.93591571e-01 5.23159266e-01
-5.14705896e-01 -6.85349941e-01 7.65277922e-01 3.17972392e-01
6.49624616e-02 -1.49228525e+00 -5.70813596e-01 3.34754556e-01
-1.07132711e-01 -5.39811492e-01 -1.57590687e-01 6.12205923e-01
-1.19366693e+00 -1.01526415e+00 -8.75530541e-01 -1.03603637e+00
7.18889177e-01 2.44572356e-01 1.18463767e+00 2.84042627e-01
-2.87709564e-01 6.49904788e-01 -1.31030381e-02 -2.67939389e-01
-3.16364132e-02 -5.78750819e-02 5.01899779e-01 -1.07851994e+00
6.46393359e-01 -3.92214775e-01 -1.00622165e+00 5.35368681e-01
-3.85821253e-01 -2.04611585e-01 7.85597622e-01 1.04281425e+00
1.42742008e-01 -1.33182362e-01 2.41971418e-01 -1.16735983e+00
1.11498392e+00 -4.85333532e-01 1.30341025e-02 2.62960047e-01
-8.53362501e-01 -4.53584343e-01 2.27881029e-01 -1.33037493e-02
-6.33760452e-01 -2.51423597e-01 -4.14545417e-01 3.22794706e-01
-4.06866640e-01 3.34393084e-02 2.24873364e-01 -5.26000679e-01
5.21005154e-01 1.50258854e-01 -4.07556817e-02 1.21735156e-01
6.30975887e-02 3.75316888e-01 3.67786407e-01 -7.01224983e-01
5.19809723e-01 1.93325520e-01 -3.00995037e-02 -1.28791058e+00
2.66786069e-01 -1.64763972e-01 -6.60750508e-01 -2.69126564e-01
7.75121748e-01 -9.50198710e-01 -6.35705650e-01 5.21708906e-01
-9.54387486e-01 -1.36340424e-01 -5.25335252e-01 6.30965471e-01
-7.02474535e-01 2.94757709e-02 -4.17852998e-01 -1.17976546e+00
-4.78847146e-01 -1.38869667e+00 9.59160328e-01 3.85633826e-01
-2.05811903e-01 -1.19790471e+00 3.38975668e-01 8.99848714e-02
4.29175526e-01 7.39965662e-02 5.03531873e-01 -2.80004859e-01
-2.98304528e-01 -2.85470933e-01 1.59573574e-02 -1.65876746e-01
-3.09365243e-02 4.80770379e-01 -5.48391879e-01 -4.37962860e-01
2.44738594e-01 -4.86500829e-01 -7.51581416e-02 7.11018145e-01
4.09077644e-01 -6.02920949e-01 -7.12544680e-01 5.56552589e-01
1.27709007e+00 4.72497940e-01 5.88340998e-01 6.77552521e-01
9.28869769e-02 5.17871320e-01 4.48484391e-01 2.08113909e-01
1.07506551e-01 1.08483166e-01 1.14217520e-01 -2.17137560e-01
9.45900455e-02 -1.75644696e-01 4.44304377e-01 1.09980536e+00
-8.23837519e-01 -2.15223029e-01 -4.88283783e-01 6.68056428e-01
-1.69156313e+00 -1.31570101e+00 -6.70152605e-01 9.76623476e-01
3.78074080e-01 3.38884830e-01 1.58041969e-01 1.21232189e-01
5.47049284e-01 -4.76693302e-01 -1.27305150e-01 -9.16352808e-01
-3.82233024e-01 4.21136379e-01 7.14838982e-01 6.51479006e-01
-6.10602558e-01 5.08361042e-01 1.25505590e+01 5.49717486e-01
-5.10860123e-02 1.27314463e-01 3.14815581e-01 1.99914098e-01
-2.24020526e-01 1.62632689e-01 -1.04165065e+00 9.26906094e-02
1.64754760e+00 -4.36735392e-01 3.93212438e-01 6.44444764e-01
3.94769877e-01 -1.96429595e-01 -9.78620768e-01 5.21999180e-01
1.12709433e-01 -1.50397491e+00 -3.76861334e-01 9.64888573e-01
5.80093563e-01 -1.81895152e-01 6.81941509e-01 4.33209062e-01
7.39425182e-01 -1.14519978e+00 7.88427055e-01 3.18626672e-01
8.87266219e-01 -8.18826377e-01 1.07315242e+00 2.76328743e-01
-1.04355454e+00 1.15487054e-02 -5.90443075e-01 -9.22581375e-01
1.62244976e-01 2.02188566e-01 -5.61583102e-01 4.06869203e-01
6.19966745e-01 2.15250537e-01 -8.53338003e-01 1.45581341e+00
-4.29231405e-01 4.22687940e-02 -3.00976187e-01 -4.39922899e-01
4.83290881e-01 5.28384298e-02 3.26606274e-01 1.02295005e+00
2.89785862e-01 5.45581102e-01 -2.62208551e-01 2.92286038e-01
6.63503349e-01 5.84383309e-03 -1.38496268e+00 -8.68683681e-02
3.54620427e-01 9.69217539e-01 -3.99494410e-01 -3.56748998e-02
-6.11456513e-01 6.62115753e-01 -1.66942239e-01 5.75146377e-01
-2.92761147e-01 -8.89406264e-01 4.33969021e-01 -2.30117049e-02
-2.88469017e-01 -1.35338321e-01 -3.23372036e-01 -1.02459288e+00
-6.06416404e-01 -3.90122980e-01 7.32179061e-02 -8.63867998e-01
-1.79313409e+00 6.85844719e-01 -3.50327820e-01 -1.03178740e+00
-9.29034173e-01 -9.72984850e-01 -2.90242791e-01 6.67755604e-01
-6.59234166e-01 -1.30476952e+00 1.77773401e-01 4.65253919e-01
1.82401046e-01 -5.88519633e-01 1.09945750e+00 -1.42670095e-01
-3.18711132e-01 1.08882773e+00 3.51253241e-01 -3.05974871e-01
8.32965672e-01 -1.16271269e+00 4.23785180e-01 1.09189451e-02
-6.59259140e-01 1.28460062e+00 6.81860268e-01 -6.81785107e-01
-1.16639841e+00 -1.52893141e-01 1.38110602e+00 -9.15031374e-01
5.27874053e-01 -1.02810413e-01 1.46300867e-01 6.57190859e-01
9.07351732e-01 -5.09461224e-01 1.08935249e+00 -1.13616258e-01
4.93123859e-01 8.23448122e-01 -1.38805819e+00 5.99272728e-01
1.21222687e+00 -4.81854081e-01 -8.27020168e-01 3.28937650e-01
9.94227231e-01 -5.60786903e-01 -1.34066856e+00 6.99696779e-01
1.14132881e+00 -1.04883564e+00 1.54237401e+00 -1.16912270e+00
4.08816934e-01 1.70770600e-01 -2.18455836e-01 -6.52935803e-01
-5.24122715e-01 -8.07741940e-01 -3.97186369e-01 3.40213925e-01
8.64144087e-01 -1.16800272e+00 5.53630650e-01 1.39166510e+00
-3.58625412e-01 -4.51715857e-01 -8.63314152e-01 -1.06149745e+00
-4.66870656e-03 1.81816339e-01 3.85523647e-01 8.18638563e-01
7.09820271e-01 2.02331603e-01 -2.05201268e-01 -4.37674336e-02
6.59316301e-01 -4.87512946e-01 5.82342803e-01 -8.74011815e-01
2.49149442e-01 -3.85976404e-01 -2.94489831e-01 -2.90531307e-01
-3.26795012e-01 -4.87107635e-01 -8.59293997e-01 -1.53084815e+00
8.49664882e-02 1.97367743e-01 -2.33256146e-01 3.44385266e-01
6.77823246e-01 2.45789096e-01 -2.27382466e-01 1.17054820e-01
-1.86927155e-01 -9.06788632e-02 9.18564737e-01 1.10557920e-03
-2.17724726e-01 3.92412305e-01 -1.01716614e+00 9.60534811e-01
7.85404518e-02 -6.09580100e-01 -5.51487744e-01 1.64939865e-01
5.71885824e-01 3.57775450e-01 4.49405432e-01 -9.70509052e-01
9.87445056e-01 -4.32846308e-01 6.12257063e-01 -1.43385470e+00
-2.65485227e-01 -9.00021493e-01 3.61826897e-01 6.29400074e-01
4.08042461e-01 3.53796333e-01 2.26811334e-01 -9.13986787e-02
-2.01378226e-01 -6.44658387e-01 4.43001747e-01 -7.45609999e-01
-4.55086857e-01 -3.90635371e-01 -1.02313221e+00 -9.67486948e-03
7.88939476e-01 -4.61360693e-01 -6.47767961e-01 -1.95199717e-02
-1.44608676e+00 7.30553791e-02 8.33678424e-01 1.34732248e-02
6.98022604e-01 -1.44444084e+00 -1.11748077e-01 6.18802965e-01
-5.42432606e-01 -4.98194307e-01 2.20674574e-01 1.06156552e+00
-1.31528914e+00 1.26915324e+00 -5.21896303e-01 -2.05424115e-01
-8.96585643e-01 7.19095051e-01 3.15837234e-01 1.75098896e-01
-3.93049836e-01 7.00151443e-01 1.12239346e-02 -6.58594191e-01
1.24011397e-01 2.97559589e-01 -4.87763733e-01 -2.52128989e-01
9.22277749e-01 1.17918742e+00 1.32526740e-01 -3.91715407e-01
-5.58191121e-01 4.39446121e-01 2.08604142e-01 -6.12734079e-01
1.38662541e+00 -2.20989168e-01 -7.44191706e-01 5.00318348e-01
7.50003338e-01 -7.33495504e-02 -3.28966863e-02 1.27011871e+00
4.80256975e-03 -5.47239244e-01 -3.11336160e-01 -7.82959402e-01
-3.42039734e-01 5.93214929e-01 6.17733955e-01 4.59080935e-01
9.69997942e-01 -6.47576571e-01 7.70057857e-01 7.64313757e-01
7.83133745e-01 -1.22997761e+00 -6.36951804e-01 3.69483382e-01
9.39836800e-01 -9.30032432e-01 1.30400375e-01 -3.14221263e-01
-5.55534840e-01 1.08689797e+00 5.38173854e-01 -1.62142932e-01
1.17091513e+00 6.96656525e-01 -1.75611794e-01 -4.53250527e-01
-1.04549897e+00 9.83817875e-02 2.10727617e-01 1.32608259e+00
8.26061428e-01 1.41275764e-01 -1.40426433e+00 8.46834600e-01
-6.34404242e-01 4.76996809e-01 9.47268784e-01 1.55018985e+00
-5.12773991e-01 -1.41692734e+00 -4.41162825e-01 2.31142953e-01
-2.82854587e-01 -3.49930972e-01 -8.62966061e-01 1.09929097e+00
1.97083559e-02 1.37330377e+00 -4.48758155e-01 -4.63731140e-01
8.27405751e-01 3.19600284e-01 4.41484600e-01 -1.25142381e-01
-8.94688129e-01 6.48750722e-01 3.84649158e-01 -1.05672014e+00
-7.94483483e-01 -1.27049911e+00 -1.08188879e+00 -1.16740942e+00
-3.07217389e-01 4.65777248e-01 2.31674016e-01 3.19533348e-01
-2.09303293e-02 1.56313553e-03 6.67447627e-01 -1.00111675e+00
-1.97397545e-01 -6.07395887e-01 -1.20775104e+00 -5.56946039e-01
3.94425273e-01 -5.85869551e-01 -1.15223050e+00 -6.24734610e-02]
|
[-1.5391778945922852, 15.869178771972656]
|
f46e51c1-e368-4840-b8d9-77a94ba98d5e
|
dynamic-survival-transformers-for-causal
|
2210.15417
| null |
https://arxiv.org/abs/2210.15417v1
|
https://arxiv.org/pdf/2210.15417v1.pdf
|
Dynamic Survival Transformers for Causal Inference with Electronic Health Records
|
In medicine, researchers often seek to infer the effects of a given treatment on patients' outcomes. However, the standard methods for causal survival analysis make simplistic assumptions about the data-generating process and cannot capture complex interactions among patient covariates. We introduce the Dynamic Survival Transformer (DynST), a deep survival model that trains on electronic health records (EHRs). Unlike previous transformers used in survival analysis, DynST can make use of time-varying information to predict evolving survival probabilities. We derive a semi-synthetic EHR dataset from MIMIC-III to show that DynST can accurately estimate the causal effect of a treatment intervention on restricted mean survival time (RMST). We demonstrate that DynST achieves better predictive and causal estimation than two alternative models.
|
['Jeffrey Regier', 'Zhenke Wu', 'Yixin Wang', 'Prayag Chatha']
|
2022-10-25
| null | null | null | null |
['survival-analysis']
|
['miscellaneous']
|
[ 5.13238087e-02 7.36877024e-02 -4.54456389e-01 -5.12207508e-01
-8.53109837e-01 -2.44402781e-01 5.69522023e-01 5.17034650e-01
-1.47548720e-01 1.18594134e+00 8.84390593e-01 -8.15922558e-01
-3.73156607e-01 -9.01192904e-01 -5.11345387e-01 -6.94101930e-01
-6.78350210e-01 7.07100570e-01 -2.80186951e-01 2.91984200e-01
-3.23193073e-01 1.29334405e-01 -6.40998840e-01 4.95592989e-02
7.05799639e-01 2.94002384e-01 -4.53956008e-01 8.05677593e-01
3.32337976e-01 1.14057517e+00 -3.42351377e-01 -3.74057472e-01
-3.71798217e-01 -4.95638281e-01 -6.69226587e-01 -6.98639572e-01
-2.87321180e-01 -5.13214469e-01 -9.81158495e-01 2.83411413e-01
5.61658084e-01 -4.75209385e-01 1.07578039e+00 -1.50710535e+00
-5.46130002e-01 1.02681851e+00 -1.46995187e-01 2.48581991e-01
3.65545303e-01 2.65653610e-01 7.54852474e-01 -1.85115039e-01
5.17747104e-01 1.36429703e+00 1.35925865e+00 5.17302513e-01
-1.68202448e+00 -5.07696033e-01 -2.99838424e-01 -9.36904103e-02
-1.23074162e+00 -3.44672769e-01 1.19023524e-01 -6.61487103e-01
5.19871354e-01 2.93993056e-01 7.86608934e-01 1.44782770e+00
1.00394058e+00 6.24837995e-01 8.67364705e-01 1.01090025e-03
2.56682336e-01 -5.26330888e-01 3.92248958e-01 3.80381107e-01
1.63057521e-01 7.86583126e-01 -3.63377631e-01 -1.01462793e+00
8.33098233e-01 4.27108467e-01 -2.30499491e-01 1.62123591e-02
-1.31801188e+00 8.39813709e-01 1.57779694e-01 -3.57715562e-02
-7.10443020e-01 9.73091602e-01 4.64744419e-01 2.20893413e-01
4.68148142e-01 -1.99589282e-01 -8.50061119e-01 -3.34132701e-01
-7.61776567e-01 3.66744608e-01 6.87860966e-01 6.49593472e-01
-7.68179744e-02 -1.46358788e-01 -8.21636617e-01 1.45221859e-01
1.96924075e-01 5.80772579e-01 1.16884999e-01 -7.77398467e-01
-2.80019403e-01 3.41896772e-01 2.75248736e-01 -3.38666201e-01
-7.43894160e-01 -4.37705934e-01 -1.06735682e+00 -4.94812399e-01
6.30915225e-01 -3.28659117e-01 -1.04837930e+00 1.97404277e+00
2.19039559e-01 6.53785169e-01 5.96174486e-02 1.28870055e-01
7.18114257e-01 3.32505733e-01 8.32816362e-01 -5.56650281e-01
1.13849199e+00 3.29342633e-02 -9.00646389e-01 3.13687921e-01
1.19036973e+00 -3.35013755e-02 6.98351562e-01 -3.04918531e-02
-9.97745097e-01 2.73923814e-01 -6.81824088e-02 2.72976160e-01
9.74423513e-02 -3.18240166e-01 9.56288397e-01 5.48943579e-01
-1.15814650e+00 7.96008229e-01 -1.19375372e+00 -3.59683245e-01
7.09589899e-01 3.12669367e-01 -1.97563425e-01 -7.28667825e-02
-1.46839488e+00 5.12487233e-01 7.75714144e-02 -2.98648924e-01
-1.47339427e+00 -1.60838819e+00 -6.06209099e-01 3.05374712e-01
-2.19355766e-02 -1.64208281e+00 1.38402116e+00 -4.78172213e-01
-8.28609347e-01 5.92171729e-01 -3.22249174e-01 -7.44356334e-01
5.99221885e-01 -3.36263105e-02 -3.53409708e-01 -2.87983090e-01
-7.47750849e-02 3.65329869e-02 3.16236943e-01 -7.86531329e-01
-4.19109195e-01 -5.27463019e-01 -6.11735404e-01 -2.09732011e-01
-9.40672830e-02 6.97422326e-02 1.10419475e-01 -7.79548943e-01
-4.74872440e-01 -8.39339375e-01 -6.82314515e-01 -2.27388337e-01
-7.01896489e-01 -1.33949727e-01 3.28816533e-01 -8.04421365e-01
1.67384315e+00 -2.01491857e+00 2.36493200e-02 -2.56193727e-01
5.26869833e-01 -4.13796812e-01 1.36643022e-01 7.25337267e-01
-2.98696578e-01 2.57221282e-01 -3.64938289e-01 -2.93379426e-01
-3.73540550e-01 2.93073326e-01 -2.60918528e-01 7.25474179e-01
-3.31105798e-01 1.32239950e+00 -9.85935211e-01 -5.55082440e-01
3.03304978e-02 7.11907923e-01 -5.95045149e-01 1.82457164e-01
-1.29430145e-01 6.22708559e-01 -4.63420063e-01 5.34340620e-01
1.57181084e-01 -7.48244643e-01 2.61559874e-01 4.03458029e-01
2.15678647e-01 2.34870479e-01 -3.62859160e-01 1.18973827e+00
-1.63454100e-01 3.04166198e-01 -5.53815544e-01 -7.02140629e-01
3.64410549e-01 8.12129140e-01 1.06909895e+00 -1.07617766e-01
3.02094370e-01 -1.58608958e-01 -1.02348670e-01 -2.73069948e-01
-2.03988537e-01 -6.97860122e-01 -4.07824993e-01 3.82855207e-01
-1.86840519e-01 4.09827024e-01 -4.97410923e-01 5.00173926e-01
1.92394733e+00 -3.40793967e-01 5.48240483e-01 -2.53472865e-01
-1.90818325e-01 1.93827778e-01 8.30802739e-01 8.94635558e-01
-2.66119093e-01 2.49661043e-01 7.58622229e-01 -6.99452937e-01
-9.37778950e-01 -1.49181676e+00 -6.11558497e-01 4.29429770e-01
-3.59041959e-01 -3.58608335e-01 -3.54194432e-01 -6.85881555e-01
5.17712772e-01 1.08432829e+00 -1.08569241e+00 -4.87873733e-01
-1.16989635e-01 -1.55843174e+00 7.97881603e-01 5.68940639e-01
-2.27861121e-01 -8.49853337e-01 -3.66719484e-01 4.78227913e-01
-3.13161403e-01 -3.98374617e-01 -4.46401209e-01 4.16199937e-02
-1.22720993e+00 -1.20086193e+00 -5.35346806e-01 -1.36822030e-01
3.07873040e-01 -4.28639859e-01 1.41812682e+00 2.38004383e-02
-3.63740712e-01 5.43604791e-01 6.15619980e-02 -3.49231839e-01
-7.41268635e-01 -1.94844842e-01 7.31405616e-02 -3.71413916e-01
4.66696382e-01 -6.52206719e-01 -1.05340695e+00 -9.15099084e-02
-6.22112334e-01 6.42704666e-02 2.41852656e-01 1.05720603e+00
4.60495293e-01 1.35094464e-01 9.51624215e-01 -1.20718920e+00
3.15903664e-01 -1.00392067e+00 -3.05368990e-01 2.11960644e-01
-9.85489786e-01 4.40620743e-02 4.31771368e-01 -5.02179921e-01
-9.60172057e-01 7.48422295e-02 3.51511240e-02 -3.27451766e-01
-1.14032567e-01 7.42818952e-01 1.74612775e-01 6.56534791e-01
4.20206428e-01 2.01844573e-01 -1.16589293e-01 -3.86709183e-01
1.73404977e-01 2.93295622e-01 5.68163753e-01 -2.93684632e-01
3.35467488e-01 5.85176587e-01 3.12881231e-01 -3.63669991e-02
-6.98590636e-01 -7.29158819e-02 -2.57871419e-01 -7.42665604e-02
7.85946786e-01 -1.08088207e+00 -1.43075931e+00 6.18121386e-01
-7.61551082e-01 -8.82161796e-01 -4.66385603e-01 5.36260307e-01
-7.25005627e-01 -2.40950331e-01 -9.49854195e-01 -8.96784246e-01
-3.67681444e-01 -5.84341228e-01 1.06878424e+00 -6.85384199e-02
-4.45431560e-01 -1.73233831e+00 4.53515291e-01 -2.11900234e-01
2.15441048e-01 8.08977425e-01 1.30230331e+00 -4.60298747e-01
-2.48524591e-01 -3.97381574e-01 1.26253637e-02 -5.79337418e-01
3.03436488e-01 1.55389383e-01 -6.12436473e-01 -4.13080513e-01
-3.27328533e-01 1.70686200e-01 7.90795982e-01 1.40094149e+00
1.40176034e+00 -4.99415845e-01 -1.06045413e+00 8.68747890e-01
1.30483723e+00 3.28110695e-01 7.90221214e-01 -1.57855675e-02
5.52638113e-01 4.00551736e-01 3.29710245e-01 7.51365244e-01
8.34329128e-01 3.88832480e-01 2.70522475e-01 -2.77898073e-01
2.11873487e-01 -7.20740795e-01 2.53212303e-01 2.68219858e-01
1.59221917e-01 -5.39830506e-01 -1.26198030e+00 8.59291673e-01
-1.88478887e+00 -9.23637807e-01 -9.36910450e-01 2.40340590e+00
1.10340095e+00 -7.32340962e-02 3.75470400e-01 -1.81713387e-01
4.62381661e-01 -5.52244961e-01 -8.48404706e-01 1.73617508e-02
-1.71896055e-01 5.26655419e-03 1.00247276e+00 4.62690741e-01
-8.72249603e-01 5.28754771e-01 8.60527802e+00 3.84585530e-01
-5.82217455e-01 2.97198504e-01 1.24719274e+00 -1.87424853e-01
-6.29981816e-01 3.12853456e-01 -4.72618192e-01 6.01410985e-01
1.66939199e+00 -7.82269835e-01 -5.35704829e-02 2.60930300e-01
5.35057187e-01 1.20969847e-01 -1.39469886e+00 5.61622143e-01
-8.26996982e-01 -1.39561999e+00 -2.35299841e-01 4.07369047e-01
6.43736780e-01 -1.95753962e-01 6.45923913e-02 3.21083397e-01
1.63869762e+00 -1.35086155e+00 2.94654638e-01 1.08441293e+00
1.27313948e+00 -7.48550355e-01 7.19753325e-01 2.83887923e-01
-6.43811285e-01 -2.91743785e-01 1.49010703e-01 -2.46708654e-03
7.77265668e-01 1.04587567e+00 -1.09100497e+00 4.54587728e-01
6.44714594e-01 9.14615750e-01 -2.81846672e-01 8.68380725e-01
3.34934779e-02 1.48095500e+00 -2.71708928e-02 5.38708568e-01
-3.09559077e-01 3.12521130e-01 3.73077482e-01 8.80618572e-01
6.27983093e-01 4.15798575e-01 -2.95852095e-01 5.89955986e-01
2.14824434e-02 -2.50619382e-01 -5.70468128e-01 -1.90921262e-01
5.50898790e-01 6.42700911e-01 -2.81353235e-01 -4.61802930e-01
-2.40879133e-01 6.09776199e-01 -1.85925085e-02 3.72380555e-01
-9.49876189e-01 4.04226720e-01 7.49284625e-01 6.59141064e-01
-2.90050000e-01 1.34776011e-01 -6.47567272e-01 -1.02338672e+00
-9.95143592e-01 -5.20483196e-01 1.15685248e+00 -7.34993994e-01
-1.50314474e+00 9.98055488e-02 -5.65672480e-02 -7.27965593e-01
-3.56214374e-01 -4.01388519e-02 -5.33502579e-01 1.01691580e+00
-9.58538055e-01 -1.06057310e+00 1.40587064e-02 7.37517178e-01
1.72601715e-02 3.12250137e-01 9.63270068e-01 8.19307789e-02
-8.16026986e-01 5.22217453e-01 2.86240697e-01 -1.46593034e-01
8.28032196e-01 -1.44524717e+00 6.11918867e-01 2.46099591e-01
-5.60183942e-01 4.48765308e-01 9.22586739e-01 -1.31631327e+00
-1.25695658e+00 -1.28337598e+00 1.07123280e+00 -9.59783077e-01
7.62253284e-01 -1.76234599e-02 -9.28689063e-01 1.15017223e+00
-1.38094172e-01 -1.12298504e-01 9.59294677e-01 5.07590234e-01
-5.88702708e-02 1.29378244e-01 -1.24743187e+00 5.72267950e-01
1.11351347e+00 -1.87538400e-01 -2.88491815e-01 2.67270714e-01
1.11223841e+00 6.24248609e-02 -1.36947989e+00 6.52527392e-01
6.00764334e-01 -5.76437593e-01 1.17962396e+00 -1.25040281e+00
6.71207130e-01 1.64321274e-01 1.41972646e-01 -1.38502085e+00
-6.97294712e-01 -7.80410409e-01 -3.58197421e-01 8.78216267e-01
2.37742066e-01 -7.19420612e-01 5.77340603e-01 8.66272688e-01
2.20376939e-01 -5.75858951e-01 -1.09818971e+00 -5.29697537e-01
4.60740805e-01 -2.81161785e-01 1.00111079e+00 1.21425700e+00
7.78511465e-02 1.95860863e-01 -6.27181828e-01 2.05152228e-01
9.92878377e-01 -1.84025660e-01 5.13227522e-01 -1.60571229e+00
-3.35745543e-01 -8.95260647e-02 -2.07514614e-01 -2.84030765e-01
-4.56039011e-02 -5.15297294e-01 -2.50026494e-01 -1.69325960e+00
9.58743632e-01 -7.90823102e-01 -6.05947196e-01 6.81466579e-01
-6.31516695e-01 -1.92306668e-01 -5.78132808e-01 1.90622211e-01
-5.47296107e-02 6.68997109e-01 1.05254591e+00 1.99189678e-01
-1.98979750e-01 1.84545293e-01 -6.83362484e-01 4.13924962e-01
6.69572949e-01 -8.76520932e-01 -3.37386250e-01 1.17932446e-01
3.63396823e-01 1.25332999e+00 7.68051624e-01 -2.54060209e-01
-1.29652116e-02 -5.10864794e-01 3.02789032e-01 -6.71872079e-01
-2.52157062e-01 -4.86015528e-01 1.18416965e+00 1.04208148e+00
-4.78831291e-01 1.94766149e-01 2.53990233e-01 1.00255501e+00
1.05639234e-01 5.39908290e-01 4.71515149e-01 1.21986739e-01
9.48889181e-02 7.60726392e-01 -6.65766180e-01 -8.14139694e-02
1.00749993e+00 3.58715087e-01 -4.48022544e-01 -6.56895399e-01
-1.10490656e+00 4.50509965e-01 5.04790843e-01 2.06516273e-02
4.28729802e-01 -1.47155988e+00 -9.87799823e-01 -2.68245727e-01
3.22171822e-02 -2.19707385e-01 7.80492604e-01 1.09351957e+00
-1.84915468e-01 3.77945900e-01 3.20968211e-01 -5.36194205e-01
-1.17891419e+00 1.00671208e+00 5.47880352e-01 -6.60450578e-01
-8.60678911e-01 4.82693225e-01 7.59356260e-01 -1.75773144e-01
-1.00862101e-01 5.29684015e-02 7.59985521e-02 -2.18823284e-01
5.11092961e-01 3.53707284e-01 -4.65681672e-01 -2.10220411e-01
-3.00656408e-01 -3.41417283e-01 1.57614276e-01 -1.98976308e-01
1.51720226e+00 -2.17892751e-01 -9.18493569e-02 7.39659488e-01
8.30636442e-01 -5.43351352e-01 -1.29631329e+00 -1.91786960e-01
6.96522668e-02 -2.88117170e-01 2.82397032e-01 -1.11521482e+00
-8.06940079e-01 5.71736276e-01 5.08541167e-01 8.60701129e-02
1.26937771e+00 1.44508883e-01 7.49003768e-01 -5.34847081e-01
3.32843423e-01 -1.74780473e-01 -3.95423591e-01 -2.34383829e-02
5.72184026e-01 -9.08677936e-01 -2.80623138e-01 -2.04755187e-01
-2.82930821e-01 5.39823830e-01 2.38087233e-02 1.99442044e-01
1.02924645e+00 6.52262211e-01 -1.61510289e-01 -4.27100122e-01
-1.51573062e+00 3.21318150e-01 -6.59169927e-02 5.22741020e-01
6.59918070e-01 6.53575659e-01 -3.44844401e-01 8.25635135e-01
-1.29670188e-01 6.52735233e-01 6.06181741e-01 4.73545462e-01
2.68041253e-01 -1.08249724e+00 -3.65023017e-01 9.64416802e-01
-8.35188985e-01 -3.56191933e-01 -1.31727353e-01 5.16902924e-01
-3.17363203e-01 6.84209645e-01 2.11613223e-01 -2.36675158e-01
1.48710772e-01 1.72437996e-01 1.99333847e-01 -2.95990646e-01
-4.37472790e-01 -5.80905564e-02 2.81187519e-02 -4.65745330e-01
-6.97376058e-02 -1.19329035e+00 -1.04219902e+00 -1.03120720e+00
2.29629409e-03 2.44436506e-02 1.25036553e-01 6.99062765e-01
4.28493500e-01 1.07130241e+00 5.75021565e-01 1.09576099e-01
-4.83286798e-01 -7.22169697e-01 -6.15342081e-01 3.42857987e-01
5.64912975e-01 -5.28022528e-01 -4.22397614e-01 3.44628274e-01]
|
[7.918551445007324, 5.616941928863525]
|
3a71083f-720f-4180-b1d5-c34f7e59b3d7
|
optimal-planning-of-hybrid-energy-storage
|
2212.05662
| null |
https://arxiv.org/abs/2212.05662v1
|
https://arxiv.org/pdf/2212.05662v1.pdf
|
Optimal Planning of Hybrid Energy Storage Systems using Curtailed Renewable Energy through Deep Reinforcement Learning
|
Energy management systems (EMS) are becoming increasingly important in order to utilize the continuously growing curtailed renewable energy. Promising energy storage systems (ESS), such as batteries and green hydrogen should be employed to maximize the efficiency of energy stakeholders. However, optimal decision-making, i.e., planning the leveraging between different strategies, is confronted with the complexity and uncertainties of large-scale problems. Here, we propose a sophisticated deep reinforcement learning (DRL) methodology with a policy-based algorithm to realize the real-time optimal ESS planning under the curtailed renewable energy uncertainty. A quantitative performance comparison proved that the DRL agent outperforms the scenario-based stochastic optimization (SO) algorithm, even with a wide action and observation space. Owing to the uncertainty rejection capability of the DRL, we could confirm a robust performance, under a large uncertainty of the curtailed renewable energy, with a maximizing net profit and stable system. Action-mapping was performed for visually assessing the action taken by the DRL agent according to the state. The corresponding results confirmed that the DRL agent learns the way like what a human expert would do, suggesting reliable application of the proposed methodology.
|
['Jonggeol Na', 'J. Jay Liu', 'Won Bo Lee', 'Haider Niaz', 'Sumin Hwangbo', 'Doeun Kang', 'Dongju Kang']
|
2022-12-12
| null | null | null | null |
['energy-management']
|
['time-series']
|
[-4.91018444e-01 -8.27623606e-02 8.93022344e-02 1.32407218e-01
-5.17628014e-01 -4.88348305e-01 7.17221677e-01 1.29046589e-01
-3.99902165e-01 1.39956605e+00 -7.31041655e-02 -3.23611826e-01
-6.95142627e-01 -9.63501096e-01 -5.49100995e-01 -1.29343760e+00
-1.28716201e-01 4.34522361e-01 -2.83168256e-01 -9.07152295e-02
2.30793938e-01 6.51109874e-01 -1.45480347e+00 -5.19335508e-01
1.15787685e+00 1.38514686e+00 7.45133579e-01 2.71482408e-01
1.47689402e-01 2.56558657e-01 -7.26981938e-01 3.35780233e-01
2.12241545e-01 -2.16148540e-01 -3.04675661e-02 -8.69397223e-02
-8.34935546e-01 -4.22145605e-01 -3.26162994e-01 1.16242814e+00
8.40052903e-01 4.30921614e-01 5.68291664e-01 -1.46217716e+00
-4.11237329e-01 7.38205791e-01 -2.63982505e-01 1.52268767e-01
5.76070882e-03 5.88578641e-01 5.34244895e-01 -7.16455579e-01
1.53849736e-01 8.95686388e-01 -2.67252680e-02 2.45936349e-01
-7.67524540e-01 -5.61292171e-01 1.03717990e-01 7.44535744e-01
-1.36431241e+00 -8.55228826e-02 8.48794997e-01 -1.47330374e-01
8.91989052e-01 1.33070037e-01 1.13009501e+00 7.28681445e-01
5.71209669e-01 5.40280163e-01 1.43936825e+00 -3.29542667e-01
9.85358179e-01 1.20339140e-01 -6.16501153e-01 1.08637467e-01
5.89200139e-01 5.60339332e-01 -5.87743223e-02 3.35838258e-01
3.11198264e-01 -1.76632896e-01 -6.29154593e-02 -4.38363105e-01
-1.04714644e+00 5.27065396e-01 5.79945624e-01 5.09816229e-01
-7.71388233e-01 2.67951395e-02 2.92553365e-01 1.69138744e-01
1.25247627e-01 5.14989793e-01 -2.61773199e-01 -2.58297324e-02
-8.60995591e-01 -1.65940717e-01 6.73766077e-01 8.50794494e-01
2.15866849e-01 9.73991573e-01 -3.51299316e-01 9.92590338e-02
4.23951060e-01 9.07746434e-01 5.92490852e-01 -8.86553526e-01
2.17297748e-01 5.66238165e-01 7.74989963e-01 -5.43043554e-01
-8.19014847e-01 -6.29578292e-01 -1.15478384e+00 6.27889037e-01
1.39834909e-02 -5.70211947e-01 -4.30213422e-01 1.53389323e+00
2.04985112e-01 -2.12591261e-01 3.79453123e-01 8.40921283e-01
5.92598692e-02 9.38597560e-01 2.02907816e-01 -7.37506628e-01
1.19445491e+00 -4.61624920e-01 -1.22388589e+00 -1.16626330e-01
2.43018910e-01 -3.92426252e-02 6.00167513e-01 3.03651690e-01
-1.21634066e+00 -4.06658769e-01 -1.34448886e+00 8.47589195e-01
-7.81546652e-01 3.22207272e-01 2.37759113e-01 4.03957814e-01
-1.03251934e+00 6.55519187e-01 -6.83623791e-01 -3.21716845e-01
3.40906173e-01 2.78276980e-01 1.24053523e-01 4.26686019e-01
-1.52720129e+00 1.54771626e+00 1.06378508e+00 6.44092441e-01
-1.16324747e+00 -3.76586258e-01 -5.93469381e-01 4.54808086e-01
5.03614247e-01 -3.58843863e-01 1.01567602e+00 -4.80434537e-01
-1.74878466e+00 9.50284360e-04 4.10286844e-01 -8.27463686e-01
7.60867000e-01 2.52852552e-02 -5.50091267e-01 1.82373941e-01
-2.97365665e-01 1.76702648e-01 5.94211757e-01 -1.38422370e+00
-8.32242846e-01 -2.72037745e-01 -1.65749654e-01 3.19139302e-01
-3.96114737e-01 -5.38861334e-01 5.69894552e-01 -2.56426185e-01
-4.49428439e-01 -7.84182549e-01 -2.57402927e-01 -2.83095539e-01
-3.19752872e-01 -5.82849264e-01 8.58630121e-01 -7.09831476e-01
1.35385954e+00 -1.80994654e+00 2.17157125e-01 1.71591982e-01
-3.67296427e-01 2.92051375e-01 2.65764743e-01 8.77858281e-01
1.06185570e-01 3.56164612e-02 -1.46719888e-01 -7.13086948e-02
5.10557473e-01 2.04086393e-01 -2.08462670e-01 2.98593342e-01
-9.07197520e-02 9.62534308e-01 -1.10072589e+00 -1.07087560e-01
6.47144794e-01 5.55893928e-02 5.15922368e-01 4.16007668e-01
-5.85694492e-01 3.84513170e-01 -7.28816032e-01 6.09767258e-01
4.41568613e-01 -1.65036172e-01 2.45277613e-01 -2.56895721e-01
-4.94714499e-01 -5.68174481e-01 -1.26320970e+00 1.39088404e+00
-7.85647750e-01 4.61587965e-01 3.46357912e-01 -1.17445517e+00
9.90659058e-01 2.78884619e-01 6.60555780e-01 -1.20799577e+00
1.97026491e-01 4.91090536e-01 -6.25673607e-02 -5.70187628e-01
2.78032869e-01 -1.21697389e-01 6.18399084e-02 3.99135202e-01
-2.25425199e-01 -3.07004899e-01 2.52071649e-01 -9.51196402e-02
6.74274206e-01 2.00972050e-01 6.89095438e-01 -5.14450669e-01
6.92349672e-01 -1.98017150e-01 5.04689634e-01 4.84857649e-01
-1.86641335e-01 -3.24223101e-01 1.65345743e-01 -1.36637866e-01
-9.06971335e-01 -8.66101444e-01 1.76352132e-02 3.40900272e-01
3.56447011e-01 4.40864533e-01 -3.68055046e-01 -4.56261665e-01
2.00170860e-01 1.64780462e+00 -2.75224119e-01 -4.10848081e-01
-1.46231040e-01 -8.74893308e-01 -2.06195831e-01 5.27946770e-01
8.07168722e-01 -1.04268742e+00 -1.19918454e+00 4.85641927e-01
3.87829691e-01 -9.80572343e-01 -4.21864167e-02 2.76916027e-01
-4.72402364e-01 -9.40807462e-01 -8.11901629e-01 -2.27070585e-01
5.49161613e-01 -1.11712500e-01 7.97185302e-01 -1.20047651e-01
7.72592006e-03 5.07679105e-01 -7.16319457e-02 -4.01195794e-01
-5.35224915e-01 -2.25673616e-01 6.08149767e-01 -8.65033865e-02
-9.28725749e-02 -5.03593206e-01 -1.00924480e+00 3.44015837e-01
-7.74899662e-01 3.76281217e-02 8.31725836e-01 7.73888350e-01
6.05466187e-01 8.58291566e-01 1.32854855e+00 1.15545966e-01
8.59444857e-01 -6.62854016e-01 -1.12167633e+00 6.35399818e-01
-1.13472700e+00 6.99313954e-02 9.58957016e-01 -1.50567859e-01
-1.20519996e+00 -2.57348537e-01 2.58191705e-01 -4.07615274e-01
3.59071009e-02 3.08542162e-01 -5.39114475e-01 1.17687076e-01
-1.90489337e-01 6.54842198e-01 -1.47821829e-01 -1.25524685e-01
2.88672835e-01 6.27698720e-01 2.76368350e-01 -4.55415040e-01
8.66894484e-01 1.43196955e-01 5.52964687e-01 -3.76973212e-01
-3.51541609e-01 2.39913166e-01 -3.29736024e-01 -7.36605167e-01
7.49601483e-01 -7.28960037e-01 -1.27155948e+00 5.60169756e-01
-7.91366041e-01 -3.18140000e-01 -6.74333870e-01 5.25271177e-01
-7.42200196e-01 8.13506171e-02 1.06071159e-01 -1.40303409e+00
-4.32101846e-01 -9.76946294e-01 3.70106012e-01 6.70418561e-01
5.10392189e-01 -1.05057979e+00 -1.34554371e-01 -1.95616364e-01
7.74656594e-01 4.32976663e-01 9.68135178e-01 -7.54878044e-01
-7.57037520e-01 -8.16246793e-02 -1.06144501e-02 3.94046843e-01
1.97961673e-01 -1.40497357e-01 -5.32557011e-01 -7.14287996e-01
5.61921299e-02 -1.81055412e-01 8.18123892e-02 3.69675338e-01
1.17068505e+00 -3.97581398e-01 -1.82270229e-01 6.02987930e-02
1.78273118e+00 8.44166577e-01 3.98556292e-01 5.87239504e-01
6.29252614e-03 3.79583955e-01 9.03074563e-01 8.99024189e-01
4.87037212e-01 5.05178392e-01 9.85564530e-01 2.11322337e-01
3.20334494e-01 -2.32853189e-01 3.53629172e-01 4.78318125e-01
5.97541593e-02 -6.03302836e-01 -4.78322804e-01 3.68781179e-01
-2.06749463e+00 -9.34924841e-01 5.42384803e-01 2.33683968e+00
4.19861972e-01 3.10185939e-01 -9.10351202e-02 5.39431944e-02
6.64919615e-01 2.13905886e-01 -1.23333025e+00 -3.39985192e-01
-4.33838278e-01 -4.08570886e-01 6.69798017e-01 8.25124979e-02
-3.96216124e-01 2.42086679e-01 5.26056528e+00 1.08921754e+00
-9.36255932e-01 -1.63680255e-01 6.67837858e-01 -9.18112844e-02
-4.65405226e-01 -2.97988057e-01 -5.95139563e-01 1.12725651e+00
1.30362940e+00 -9.35267389e-01 7.72659957e-01 4.60176498e-01
1.24571931e+00 -6.82897270e-01 -8.70371699e-01 6.83179259e-01
-5.39326906e-01 -1.42324591e+00 -2.02035323e-01 1.04561254e-01
9.26032841e-01 -1.68270156e-01 -3.03934813e-01 3.40297163e-01
3.73723388e-01 -7.41640568e-01 8.10421467e-01 1.24055481e+00
2.88706809e-01 -9.73040640e-01 9.43956077e-01 6.30266011e-01
-1.24665082e+00 -8.03173304e-01 -1.29702240e-01 2.31967241e-01
4.77005154e-01 6.74549699e-01 -4.40080434e-01 8.95574033e-01
6.20301008e-01 3.71953487e-01 -2.71852702e-01 8.42566311e-01
-4.19387519e-01 3.93485963e-01 -4.74531531e-01 -5.43432653e-01
1.76353738e-01 -6.78642452e-01 5.24682164e-01 6.44342363e-01
8.00409079e-01 5.03484830e-02 2.03976199e-01 9.53571796e-01
6.82935491e-02 -2.83859789e-01 -5.25245905e-01 -3.77731860e-01
9.86146390e-01 1.39132190e+00 -7.54070759e-01 -2.16848403e-01
-1.51767761e-01 4.33280557e-01 -1.60116911e-01 4.69516218e-01
-7.64017820e-01 -1.48364216e-01 2.04603449e-01 -2.69272953e-01
2.44904578e-01 -4.06527728e-01 -2.53245622e-01 -6.08849406e-01
5.83636574e-02 -2.70654440e-01 3.16759706e-01 -1.12619734e+00
-1.27873445e+00 2.36616090e-01 1.32502258e-01 -1.38027442e+00
-4.21691298e-01 -4.76256937e-01 -9.88700569e-01 7.11969733e-01
-1.93027103e+00 -7.84699202e-01 -1.78923666e-01 3.48204404e-01
6.29172325e-01 -4.49089617e-01 5.38486183e-01 -4.52058949e-02
-1.00944698e+00 -3.26799974e-02 8.39192569e-01 -6.43553555e-01
2.47575324e-02 -1.58629322e+00 -5.04814684e-01 8.22814941e-01
-6.31952643e-01 -1.66197270e-01 9.18762863e-01 -5.80415606e-01
-1.89232039e+00 -6.55925572e-01 9.62952524e-02 4.70046043e-01
1.17635465e+00 9.47582498e-02 -7.52174497e-01 1.06413208e-01
8.58975947e-01 -1.27844676e-01 1.89406455e-01 -8.94830883e-01
6.83942854e-01 -4.76866335e-01 -1.47649193e+00 4.57625896e-01
6.08033657e-01 -5.23717701e-02 -3.26838553e-01 3.07851017e-01
4.94601727e-01 -6.57686815e-02 -1.06021106e+00 5.05817115e-01
2.67861307e-01 -6.26958668e-01 6.21775389e-01 -4.45172280e-01
-8.91150311e-02 -5.51223576e-01 -2.21717894e-01 -2.00475597e+00
-5.65003529e-02 -6.61815822e-01 -9.18024898e-01 1.36900985e+00
6.69885948e-02 -1.00564611e+00 3.26486737e-01 6.55994177e-01
-1.72369272e-01 -9.84030724e-01 -1.52367139e+00 -1.01544356e+00
1.34110544e-02 1.09738573e-01 8.15272629e-01 3.09871972e-01
-6.70881793e-02 -4.19771791e-01 -2.21072063e-01 5.41901648e-01
8.53305876e-01 2.04486817e-01 8.83656740e-02 -6.98303759e-01
1.11993320e-01 -4.95374292e-01 -2.28901431e-02 -3.35178137e-01
1.65232316e-01 -4.41177458e-01 1.01478256e-01 -2.02256060e+00
-3.46804529e-01 -2.41914794e-01 -1.01760447e+00 2.21524701e-01
6.98621497e-02 -4.18878168e-01 4.71555442e-01 2.21414603e-02
-5.34954369e-01 1.35713804e+00 1.15447223e+00 -2.91875482e-01
-1.49403501e-03 4.67098385e-01 -3.82062137e-01 5.50081313e-01
9.99744713e-01 -8.38701874e-02 -5.65362155e-01 -6.96122125e-02
5.29563248e-01 4.96924579e-01 2.80850142e-01 -1.11241889e+00
5.21582484e-01 -4.88637596e-01 4.38448936e-01 -8.86548877e-01
1.02826826e-01 -1.36058342e+00 5.18884361e-01 8.67098570e-01
-1.69725064e-02 2.31305808e-01 2.46604711e-01 7.19147921e-01
3.64069524e-03 -2.23322958e-01 7.66676068e-01 -2.07602922e-02
-1.12545097e+00 8.28124583e-02 -5.18780708e-01 -4.01034862e-01
1.74318123e+00 -5.27643748e-02 -3.65330338e-01 -4.09775555e-01
-6.42772377e-01 1.15532863e+00 1.14901729e-01 3.30834687e-01
4.89942342e-01 -1.26660573e+00 -5.74167848e-01 -1.77158073e-01
-3.77282709e-01 -3.82492214e-01 5.34193993e-01 6.50520265e-01
3.83873396e-02 4.46275353e-01 -2.77090639e-01 -2.94911116e-01
-3.39565367e-01 6.37467384e-01 7.45669067e-01 -4.45438266e-01
-4.26122963e-01 -1.07621826e-01 -5.12430727e-01 8.20569694e-03
1.74654424e-01 2.16790847e-02 -2.48910189e-01 5.64946294e-01
2.87416339e-01 8.79348576e-01 1.22786000e-01 -9.60744098e-02
-2.95162916e-01 3.16639096e-01 6.35776401e-01 2.28495561e-02
1.43477356e+00 -5.63131750e-01 2.76433021e-01 4.40152854e-01
4.18685198e-01 -3.74321163e-01 -1.72009718e+00 1.08552389e-01
1.46037459e-01 -1.24631837e-01 3.30318332e-01 -1.31990898e+00
-1.13918459e+00 5.39526105e-01 8.16635907e-01 7.91505694e-01
1.35455322e+00 -4.52222228e-01 3.78305346e-01 5.52409589e-01
8.97379279e-01 -1.63052440e+00 -2.61887312e-01 4.06097732e-02
1.03120613e+00 -1.08572161e+00 1.96749300e-01 6.02906704e-01
-5.47273397e-01 1.15733683e+00 5.83520353e-01 5.21203913e-02
6.29840136e-01 1.96611807e-01 -4.81170207e-01 1.98182270e-01
-6.51321054e-01 -2.52746493e-01 -1.36189729e-01 3.65437686e-01
-5.59812427e-01 3.24669540e-01 -4.25327331e-01 5.46082079e-01
2.91730553e-01 9.78304967e-02 4.99547243e-01 8.89443099e-01
-5.83750963e-01 -5.28966010e-01 -3.26453328e-01 2.88441956e-01
2.60982484e-01 5.09011567e-01 4.10520375e-01 1.00460160e+00
1.67524870e-02 9.80305374e-01 -8.28935206e-03 1.46848798e-01
5.16074955e-01 -7.46312439e-02 1.15614049e-01 -5.22192987e-03
-1.52489901e-01 -2.57347971e-01 -2.27152631e-01 -4.82095659e-01
-3.16929400e-01 -5.40767848e-01 -1.57088113e+00 -5.87753728e-02
-2.72708416e-01 3.94343436e-01 1.13770688e+00 1.37048805e+00
4.38792050e-01 8.58897209e-01 1.36241972e+00 -9.89379644e-01
-1.18795109e+00 -8.98951948e-01 -1.03488505e+00 -1.51905090e-01
8.78184512e-02 -7.18279719e-01 -7.50930607e-01 -8.24565291e-01]
|
[5.595440864562988, 2.4994571208953857]
|
a86053ec-4863-4898-ba51-c04ac0cc6e15
|
knowledge-transfer-for-surgical-activity
|
1711.05848
| null |
http://arxiv.org/abs/1711.05848v1
|
http://arxiv.org/pdf/1711.05848v1.pdf
|
Knowledge transfer for surgical activity prediction
|
Lack of training data hinders automatic recognition and prediction of
surgical activities necessary for situation-aware operating rooms. We propose
using knowledge transfer to compensate for data deficit and improve prediction.
We used two approaches to extract and transfer surgical process knowledge.
First, we encoded semantic information about surgical terms using word
embedding which boosted learning process. Secondly, we passed knowledge between
different clinical datasets of neurosurgical procedures using transfer
learning. Transfer learning was shown to be more effective than a simple
combination of data, especially for less similar procedures. The combination of
two methods provided 22% improvement of activity prediction. We also made
several pertinent observations about surgical practices.
|
['Xavier Morandi', 'Pierre Jannin', 'Olga Dergachyova']
|
2017-11-15
| null | null | null | null |
['activity-prediction', 'activity-prediction']
|
['computer-vision', 'time-series']
|
[ 3.31800103e-01 6.95804238e-01 -5.61814070e-01 -4.68799949e-01
-7.12284803e-01 -2.56940275e-01 4.62586850e-01 5.26990891e-01
-9.76189315e-01 9.76888597e-01 1.05731142e+00 -7.52723515e-01
-6.05057240e-01 -8.45040321e-01 -6.37075305e-01 -4.81121719e-01
-8.92655253e-02 4.18001235e-01 3.66573930e-02 -2.69673198e-01
5.36966503e-01 4.70749080e-01 -1.12525499e+00 8.15894306e-01
6.04938626e-01 6.43785954e-01 4.46045458e-01 3.93298239e-01
-4.15544003e-01 9.66118693e-01 -2.72336066e-01 -1.17483035e-01
2.18457401e-01 -3.12678218e-01 -1.06915081e+00 -3.64090919e-01
-1.94510184e-02 -3.75849903e-02 -4.17762280e-01 6.24690175e-01
5.78086615e-01 3.16440612e-02 8.20177555e-01 -5.09380639e-01
-6.64831281e-01 4.68958735e-01 1.02843061e-01 5.21095037e-01
3.12454730e-01 -7.29343221e-02 5.34667790e-01 -7.88420081e-01
7.90941834e-01 5.77331305e-01 7.53057241e-01 8.38033080e-01
-8.83290410e-01 -6.30621374e-01 -4.44894165e-01 2.06252575e-01
-7.98871815e-01 -1.01414554e-01 4.35718030e-01 -5.68039775e-01
1.01125443e+00 8.50164890e-02 9.91457939e-01 1.02388823e+00
7.72806764e-01 4.13287520e-01 1.25646114e+00 -6.84520900e-01
-1.25360548e-01 5.39307714e-01 -1.12161092e-01 9.80593562e-01
5.03523827e-01 3.87499839e-01 -6.23644948e-01 -1.33927509e-01
7.42114663e-01 3.23293149e-01 -1.98979944e-01 -3.23018730e-01
-1.48078179e+00 8.27152789e-01 7.04850912e-01 9.03489828e-01
-5.21571875e-01 -1.98302090e-01 6.98868394e-01 4.22932506e-01
2.75640517e-01 1.03508878e+00 -6.73166275e-01 -1.53166294e-01
-4.83630389e-01 -5.04953384e-01 9.07972693e-01 5.02462864e-01
4.84844685e-01 -4.26926762e-01 -2.13737845e-01 5.70816040e-01
-1.89494520e-01 9.41254199e-03 1.06103897e+00 -5.21792173e-01
3.26905668e-01 6.41526699e-01 -2.93786228e-01 -7.68179655e-01
-5.87556958e-01 -1.62883952e-01 -5.18074930e-01 1.21303789e-01
1.35666981e-01 -2.56164461e-01 -1.29957461e+00 1.15470481e+00
-2.43726835e-01 1.44742072e-01 3.95424455e-01 5.62836289e-01
6.85490251e-01 1.72373876e-01 5.96553266e-01 -1.18140668e-01
1.35011411e+00 -1.00118136e+00 -8.94456804e-01 -2.27521449e-01
1.41964793e+00 -7.58050442e-01 4.65742588e-01 2.15345666e-01
-8.94584537e-01 -1.80753946e-01 -8.62146378e-01 1.11574814e-01
-6.65609479e-01 -5.60215861e-03 1.08015227e+00 4.15301800e-01
-1.02396417e+00 8.42749596e-01 -7.99202263e-01 -5.91446579e-01
6.54225111e-01 6.39887989e-01 -8.83615255e-01 -3.47068235e-02
-1.03143167e+00 1.56213129e+00 7.67762780e-01 -3.29043329e-01
-5.21978259e-01 -9.33480322e-01 -1.03419554e+00 9.06790644e-02
3.11235845e-01 -9.55604196e-01 1.07183588e+00 -9.41155791e-01
-1.57115412e+00 1.16031432e+00 1.15362003e-01 -5.81869960e-01
9.02214795e-02 -3.26575369e-01 -5.59506953e-01 8.51023421e-02
-2.65706599e-01 4.57659662e-01 4.56827819e-01 -9.74571764e-01
-5.25317311e-01 -2.75436163e-01 -2.12260216e-01 4.98305798e-01
-7.03612268e-01 -2.47274205e-01 1.64448142e-01 -5.59749186e-01
-9.09876078e-02 -9.52784717e-01 -6.03890061e-01 4.65036333e-02
3.62035692e-01 -3.03500295e-02 3.66284907e-01 -9.15176451e-01
9.96418953e-01 -1.93761373e+00 1.61774486e-01 2.30438456e-01
1.71197176e-01 5.77552319e-01 8.09289962e-02 6.66294992e-01
-4.30932879e-01 1.37432024e-01 1.14956670e-01 2.22397149e-01
-6.02614284e-01 5.08438885e-01 1.07185915e-01 1.30407602e-01
2.08906233e-01 1.03320444e+00 -1.22864544e+00 -7.33756304e-01
5.48802316e-01 4.17506963e-01 -7.19940364e-01 1.47556022e-01
4.98066962e-01 6.75747573e-01 -3.54893804e-01 4.59536284e-01
-4.05233689e-02 -6.37067407e-02 2.53038287e-01 -2.89114177e-01
2.26004511e-01 4.90572512e-01 -2.87972927e-01 2.20421839e+00
-9.22998428e-01 5.07681072e-01 -4.13589120e-01 -1.03357792e+00
8.69641542e-01 5.50064981e-01 9.60363746e-01 -5.50541759e-01
3.30157369e-01 1.99440733e-01 2.83474326e-01 -7.41983533e-01
2.31040895e-01 -7.72742331e-01 1.22645408e-01 1.21110126e-01
4.91812795e-01 -2.98146665e-01 -3.31945539e-01 -6.32667318e-02
1.20488214e+00 -6.60277754e-02 8.85479093e-01 -3.28026086e-01
4.36000109e-01 1.21852860e-01 2.66176790e-01 4.50125694e-01
-4.87717330e-01 1.08071476e-01 1.77592129e-01 -7.17674494e-01
-6.15164280e-01 -1.10668993e+00 -8.39066058e-02 8.62929225e-01
-1.38440579e-01 -4.00374234e-01 -3.08046222e-01 -9.21353340e-01
-7.65941590e-02 8.17185402e-01 -1.15436471e+00 -8.45788777e-01
-5.02911210e-01 -4.37276185e-01 1.81189150e-01 9.28040922e-01
-9.60173234e-02 -1.24312723e+00 -5.95895171e-01 1.33041412e-01
8.02369341e-02 -8.28177094e-01 -2.43369952e-01 6.28735363e-01
-1.29383743e+00 -1.21160376e+00 -5.96942782e-01 -1.11772633e+00
9.23304141e-01 -1.21492110e-02 7.92216182e-01 1.48505777e-01
-5.87811768e-01 4.90236163e-01 -3.81733209e-01 -8.37891400e-01
-6.34485960e-01 1.39634788e-01 3.68965343e-02 -4.19939041e-01
6.33676589e-01 -2.47646451e-01 -8.69961381e-01 3.52558382e-02
-7.33688474e-01 5.10347225e-02 1.40515757e+00 1.18039393e+00
1.66909635e-01 -5.60443997e-01 3.38644028e-01 -1.00803781e+00
8.11742783e-01 -6.22115731e-01 2.67371893e-01 1.91196024e-01
-1.05128551e+00 2.97295272e-01 -4.31837328e-03 -3.11854422e-01
-1.14496803e+00 1.00374475e-01 1.34789675e-01 -2.09831804e-01
-1.94459051e-01 6.34676337e-01 6.48005247e-01 -2.78188288e-01
8.44153345e-01 -1.21555462e-01 3.99612993e-01 -1.84525758e-01
6.14043735e-02 5.25627375e-01 3.00230354e-01 -3.11751872e-01
4.71287400e-01 3.34298611e-01 3.54197584e-02 -4.85750079e-01
-7.56350219e-01 -7.45415807e-01 -7.19356298e-01 3.10711041e-02
1.22929263e+00 -6.22585595e-01 -3.91497105e-01 -6.97296858e-02
-7.70797193e-01 -2.94106424e-01 -5.30463696e-01 1.20217824e+00
-7.33532369e-01 4.32921983e-02 -6.67042255e-01 -2.11634099e-01
-5.49491405e-01 -8.23216200e-01 6.58994317e-01 -3.56073678e-02
-3.54785621e-01 -1.43461227e+00 4.39892590e-01 2.67562896e-01
5.82010508e-01 8.43750015e-02 9.02001023e-01 -9.61511910e-01
-3.98282474e-03 -5.09432554e-01 -1.02751456e-01 4.21926409e-01
9.45473075e-01 -5.06547451e-01 -5.69745302e-01 -2.20006555e-01
-6.48342073e-02 -4.87260558e-02 7.39015579e-01 2.07182169e-01
1.31191611e+00 -8.84657726e-02 -8.02635193e-01 4.74427491e-01
1.39387286e+00 3.16249579e-01 8.32416892e-01 4.91343796e-01
4.63780016e-01 6.22138619e-01 5.92045307e-01 5.77535592e-02
1.84304014e-01 3.27488571e-01 1.41849071e-01 -2.00739354e-01
-2.87185192e-01 -3.00360471e-01 -2.25193650e-02 1.24741817e+00
-7.40657985e-01 3.39169025e-01 -1.37279594e+00 6.74835443e-01
-1.45175278e+00 -8.27686906e-01 4.90104198e-01 2.08520079e+00
1.02903235e+00 1.90466300e-01 -4.03225183e-01 -1.98254645e-01
2.77499795e-01 -3.83705586e-01 -5.15216142e-02 -5.99185705e-01
5.14488339e-01 7.81339228e-01 1.05235577e+00 3.66052836e-01
-8.26118410e-01 9.29039180e-01 7.47019339e+00 4.99111861e-01
-9.79113579e-01 2.76117206e-01 -1.04505289e-02 -6.69583678e-02
-4.59149927e-02 -1.27770752e-01 -2.61477441e-01 1.77959308e-01
1.31475997e+00 -3.88446808e-01 -6.43919036e-02 6.79441810e-01
-2.06699297e-01 -1.04202934e-01 -1.14254868e+00 8.35954845e-01
3.02753627e-01 -1.57252860e+00 1.45018220e-01 -6.31988943e-02
6.16181135e-01 -1.10275820e-02 -1.50717527e-01 4.34259444e-01
3.07547390e-01 -1.16675282e+00 -3.07028174e-01 8.00665498e-01
5.39535284e-01 -4.43401933e-01 1.26351142e+00 1.71443149e-02
-7.14777946e-01 -2.02832803e-01 3.18937562e-03 -9.90643501e-02
-2.93050241e-03 -8.39681700e-02 -1.69479942e+00 7.08998799e-01
7.10360825e-01 7.30979800e-01 -3.00821096e-01 1.09008169e+00
-1.11955270e-01 4.52542752e-01 1.36897266e-01 -6.46522082e-03
2.46605650e-01 8.25205371e-02 1.48478359e-01 1.20281172e+00
4.10227805e-01 1.37608677e-01 -7.46432990e-02 4.24746349e-02
1.38668999e-01 3.89595270e-01 -1.12899518e+00 -2.13544369e-01
-1.36332154e-01 9.60917652e-01 -6.67668223e-01 -2.39095986e-01
-4.93273288e-01 1.06981206e+00 3.14193517e-01 -8.34557861e-02
-3.68990541e-01 -3.08860660e-01 3.61564159e-01 2.57247657e-01
9.29707885e-02 -1.07310601e-01 -4.17089999e-01 -9.20068204e-01
-3.98664832e-01 -3.74767065e-01 4.86348957e-01 -5.85633993e-01
-1.07966828e+00 6.10766351e-01 -2.64097992e-02 -1.37208080e+00
-1.52898550e-01 -1.00944638e+00 -4.32049990e-01 7.06226408e-01
-1.54680872e+00 -1.14453077e+00 -3.36310327e-01 6.26824975e-01
4.84622091e-01 -2.70749122e-01 1.47532153e+00 2.19771266e-01
1.70774892e-01 2.53621966e-01 -2.18157023e-01 2.82915503e-01
1.32885480e+00 -1.30919051e+00 -3.96714419e-01 -3.80511545e-02
-1.22049950e-01 6.27826631e-01 4.99778450e-01 -6.55719399e-01
-1.06377316e+00 -8.46463680e-01 9.15229321e-01 -6.97002709e-01
8.75428319e-01 2.36252725e-01 -8.92772853e-01 9.12807405e-01
4.34426993e-01 -2.12899879e-01 1.53502512e+00 1.38119310e-01
-7.77392164e-02 8.72743577e-02 -1.07386994e+00 5.67566991e-01
1.02393651e+00 -6.20553911e-01 -1.45894134e+00 5.40803909e-01
6.00940108e-01 -3.14181149e-01 -1.46804476e+00 7.35037923e-01
4.18116987e-01 -3.30333322e-01 1.00608134e+00 -1.21706820e+00
6.18710160e-01 4.22240734e-01 8.90632719e-02 -1.68784618e+00
-2.77444422e-01 -1.55691832e-01 2.02129394e-01 2.85844803e-01
5.87944508e-01 -8.14854503e-01 7.67475724e-01 6.96700037e-01
-4.50838596e-01 -8.04840326e-01 -6.20483041e-01 -6.58382475e-01
4.44798440e-01 1.35272652e-01 -1.41353179e-02 1.31679273e+00
6.84984446e-01 7.45593980e-02 -8.96485150e-02 -1.94470078e-01
-6.79620132e-02 -3.32303286e-01 1.54300436e-01 -1.16509306e+00
-1.08083963e-01 -3.60789150e-01 -9.00421977e-01 6.03154209e-03
1.73532426e-01 -1.39998114e+00 -1.16359517e-01 -1.95474136e+00
2.03853682e-01 -3.10752928e-01 -1.09052181e+00 7.58914351e-01
-1.53390363e-01 -1.78441312e-02 -2.91597843e-03 1.18177399e-01
-6.57566339e-02 2.78865874e-01 1.48628426e+00 4.49387431e-02
-2.34467939e-01 -6.85897022e-02 -7.46600688e-01 7.24453688e-01
1.12719440e+00 -6.53887570e-01 -4.66110051e-01 -2.98852921e-01
8.38641971e-02 3.79447863e-02 6.07767701e-02 -1.07759595e+00
4.19855624e-01 -1.00543365e-01 3.38379562e-01 -1.06947273e-01
1.79726303e-01 -1.09049237e+00 -1.20790094e-01 1.20769000e+00
-4.70407665e-01 -1.59017533e-01 6.00161970e-01 5.90355277e-01
-4.42454815e-01 -2.81084269e-01 3.90608430e-01 -2.55497396e-01
-9.88326967e-01 9.74377543e-02 -5.47358513e-01 -3.37326676e-01
1.48311985e+00 -2.73264080e-01 4.48270589e-02 -7.01869503e-02
-1.24229324e+00 -1.61262214e-01 1.93522915e-01 7.17258394e-01
7.19566941e-01 -1.17837512e+00 -5.21486998e-01 1.04250059e-01
3.03913325e-01 -3.86161685e-01 1.83013543e-01 1.12798429e+00
-8.83392632e-01 7.93490052e-01 -7.40989029e-01 -8.80996436e-02
-1.26198411e+00 7.89700389e-01 1.97895378e-01 -5.34766614e-01
-6.72691822e-01 6.20067656e-01 -1.41148806e-01 -5.96977711e-01
1.76423207e-01 -4.17558432e-01 -3.40833515e-01 -2.90876087e-02
3.74160022e-01 -1.32556394e-01 2.81766295e-01 -2.62003653e-02
-4.11599278e-01 4.62773621e-01 -4.75668490e-01 1.41847283e-01
1.51192021e+00 4.27130461e-01 8.15630406e-02 4.89263296e-01
1.23360717e+00 -2.65228391e-01 -6.38839424e-01 -3.94324183e-01
2.96781838e-01 -4.40532655e-01 3.03854942e-01 -1.27877426e+00
-8.14299464e-01 8.16207469e-01 7.25631297e-01 -4.14248466e-01
1.08512783e+00 7.81253800e-02 5.44907510e-01 5.95817745e-01
5.51903248e-01 -1.01591945e+00 8.26737210e-02 2.21439153e-01
7.91840792e-01 -1.63298094e+00 4.84749973e-02 -5.09889245e-01
-8.20965528e-01 1.31634545e+00 3.95004988e-01 -1.00304246e-01
9.93464351e-01 2.05184206e-01 1.73006713e-01 -3.83509934e-01
-5.78565598e-01 1.34060625e-02 5.22906661e-01 5.61011732e-01
5.43232203e-01 1.98259562e-01 -7.52152324e-01 4.10992533e-01
2.59079151e-02 6.84266329e-01 3.53189230e-01 1.47325373e+00
-2.89779246e-01 -1.18702352e+00 1.67360857e-01 8.57982755e-01
-5.31574607e-01 -3.80959213e-01 -2.34304413e-01 9.15082574e-01
-1.41727356e-02 4.49535936e-01 -7.98676312e-02 -4.09076929e-01
3.66809964e-01 3.91245306e-01 7.46715188e-01 -1.00918126e+00
-7.79906750e-01 -4.37092602e-01 1.87174931e-01 -7.77549803e-01
-6.03476346e-01 -4.12785947e-01 -1.27686405e+00 1.79384828e-01
-2.24065870e-01 2.96649367e-01 1.00498104e+00 7.87107229e-01
3.47673446e-01 1.09077156e+00 2.01403454e-01 -5.02032280e-01
-5.22540450e-01 -1.01728392e+00 -3.51393402e-01 4.62963343e-01
1.32950783e-01 -6.78282440e-01 -3.29432636e-01 2.32602388e-01]
|
[14.124971389770508, -3.3795742988586426]
|
90cbdf69-8abb-47ea-ab4a-dd5734390bd8
|
towards-stability-of-autoregressive-neural
|
2306.10619
| null |
https://arxiv.org/abs/2306.10619v1
|
https://arxiv.org/pdf/2306.10619v1.pdf
|
Towards Stability of Autoregressive Neural Operators
|
Neural operators have proven to be a promising approach for modeling spatiotemporal systems in the physical sciences. However, training these models for large systems can be quite challenging as they incur significant computational and memory expense -- these systems are often forced to rely on autoregressive time-stepping of the neural network to predict future temporal states. While this is effective in managing costs, it can lead to uncontrolled error growth over time and eventual instability. We analyze the sources of this autoregressive error growth using prototypical neural operator models for physical systems and explore ways to mitigate it. We introduce architectural and application-specific improvements that allow for careful control of instability-inducing operations within these models without inflating the compute/memory expense. We present results on several scientific systems that include Navier-Stokes fluid flow, rotating shallow water, and a high-resolution global weather forecasting system. We demonstrate that applying our design principles to prototypical neural networks leads to significantly lower errors in long-range forecasts with 800\% longer forecasts without qualitative signs of divergence compared to the original models for these systems. We open-source our \href{https://anonymous.4open.science/r/stabilizing_neural_operators-5774/}{code} for reproducibility.
|
['Jed Brown', 'Shashank Subramanian', 'Peter Harrington', 'Michael McCabe']
|
2023-06-18
| null | null | null | null |
['weather-forecasting']
|
['miscellaneous']
|
[-1.79482013e-01 -2.44079635e-01 5.14048159e-01 -1.59634605e-01
-2.74956226e-01 -5.33510804e-01 4.07499671e-01 -4.97660339e-02
-2.58152992e-01 1.01833451e+00 -2.30585039e-02 -8.95037115e-01
-2.42715627e-01 -7.36622870e-01 -6.12562180e-01 -8.19749296e-01
-7.07412422e-01 1.89738676e-01 8.34483206e-02 -5.46271026e-01
2.06043050e-01 8.94322634e-01 -1.56834280e+00 -2.28275016e-01
1.01407373e+00 1.06825495e+00 -2.54768759e-01 9.26837564e-01
1.11800820e-01 9.37617421e-01 -4.07196105e-01 4.37185705e-01
5.91069996e-01 -3.78110111e-01 -5.21215498e-01 -5.24005711e-01
3.75160605e-01 -1.50783062e-01 -2.35578418e-01 8.94172311e-01
6.87680483e-01 7.63151884e-01 5.86752236e-01 -9.45475817e-01
-3.98970455e-01 2.50892222e-01 -1.57532886e-01 5.90898573e-01
-2.95198321e-01 4.97491062e-01 3.97118866e-01 -8.94026220e-01
3.66867959e-01 1.09641874e+00 1.22685742e+00 3.12675059e-01
-1.39422202e+00 -6.91729963e-01 -8.84778202e-02 -3.56668264e-01
-1.50361419e+00 -8.76359880e-01 4.39399242e-01 -7.92260766e-01
1.33033657e+00 4.56444412e-01 5.94534934e-01 4.35534775e-01
6.55448616e-01 -2.16658175e-01 7.43798971e-01 -1.20251931e-01
3.80817920e-01 -2.13526130e-01 3.46291095e-01 6.17137253e-01
1.23327248e-01 6.58865571e-01 -4.94202256e-01 -4.27591145e-01
9.91525114e-01 -3.31481218e-01 -5.31834960e-01 7.83174261e-02
-8.63417685e-01 7.93508947e-01 3.74743074e-01 2.33974680e-01
-6.08417392e-01 2.98697412e-01 3.21557373e-01 4.57810223e-01
9.12352979e-01 1.02190840e+00 -6.71821356e-01 -3.58507931e-01
-1.05726528e+00 7.21341729e-01 9.28586960e-01 4.05011207e-01
5.75248897e-01 9.18203890e-01 3.05552572e-01 7.06866562e-01
5.38015598e-03 6.38500273e-01 3.58352810e-01 -1.49572873e+00
-2.16860864e-02 2.14132831e-01 3.93212140e-01 -1.22003508e+00
-7.46789455e-01 -6.39262140e-01 -1.26019788e+00 7.48948395e-01
3.73853624e-01 -8.50212038e-01 -8.29113066e-01 1.62113667e+00
3.04446727e-01 2.06842586e-01 2.48847276e-01 8.42977405e-01
2.86987871e-01 1.20449662e+00 -5.15059233e-02 -4.23237175e-01
7.81507373e-01 -6.49663806e-01 -7.29272485e-01 -6.49418384e-02
8.02731991e-01 -5.11976242e-01 7.53539562e-01 1.11324631e-01
-1.36581624e+00 -2.79179722e-01 -8.11724722e-01 2.76514083e-01
-5.39007545e-01 -2.16954648e-01 5.60205638e-01 1.79164141e-01
-1.37163377e+00 1.27196372e+00 -1.24058497e+00 -2.04927683e-01
-2.36910239e-01 3.14745218e-01 -5.12687564e-02 9.20760691e-01
-1.40143621e+00 1.12282979e+00 1.74285829e-01 5.20135760e-01
-5.46402633e-01 -1.05920076e+00 -7.52691567e-01 3.27653170e-01
-1.05068281e-01 -4.81533647e-01 1.34956563e+00 -7.74328887e-01
-1.57019317e+00 -2.13684812e-02 -3.58751625e-01 -5.92764199e-01
2.37504169e-01 -1.78227589e-01 -3.87775630e-01 -2.56925464e-01
-1.76530987e-01 4.22890455e-01 5.28754354e-01 -9.94510710e-01
-2.35762686e-01 7.74269998e-02 -3.49462748e-01 2.27966338e-01
-3.31758916e-01 9.95772630e-02 2.80452818e-01 -8.41408193e-01
1.90337121e-01 -1.27123904e+00 -6.52716994e-01 2.23451760e-02
1.22817591e-01 2.48326913e-01 6.29367411e-01 -7.32518733e-01
1.35462248e+00 -1.84560490e+00 2.82151089e-03 2.71258086e-01
-9.35991928e-02 3.72249633e-01 1.89701423e-01 4.67124790e-01
-2.71260440e-01 3.02403301e-01 -5.86584270e-01 -1.59307718e-01
-4.25351232e-01 2.61345655e-01 -8.33627999e-01 3.87462527e-01
2.70400286e-01 4.48351681e-01 -5.66932797e-01 1.69715807e-02
1.37201443e-01 5.74889898e-01 -6.31036639e-01 1.60186738e-01
-1.67162418e-01 5.60187638e-01 -1.55968726e-01 3.41328591e-01
5.26882291e-01 -2.51423001e-01 -2.35484764e-01 3.51568907e-01
-7.39751220e-01 1.57710642e-01 -1.28912020e+00 9.37586427e-01
-4.96533960e-01 8.37072372e-01 5.97287416e-01 -9.50948834e-01
7.78746963e-01 4.02755171e-01 3.80553186e-01 -3.18090051e-01
4.32551503e-02 2.97932267e-01 1.44037321e-01 -4.43754792e-01
7.76589811e-01 -3.33414525e-01 3.37331802e-01 3.52122784e-01
-3.63498092e-01 -4.54426289e-01 9.12994072e-02 -9.72697362e-02
7.85363615e-01 7.82484934e-02 -8.30304325e-02 -9.77721095e-01
2.63773292e-01 2.70383954e-01 6.72327757e-01 8.14651549e-01
-7.49313757e-02 3.33838850e-01 5.26538789e-01 -8.58169496e-01
-1.06797099e+00 -6.41042113e-01 -5.05150437e-01 1.11146653e+00
-3.14787447e-01 -2.23276943e-01 -3.19906384e-01 5.03730297e-01
2.51319021e-01 7.74565399e-01 -7.03748167e-01 -1.20062746e-01
-7.12906361e-01 -9.84875917e-01 8.33975554e-01 5.70886791e-01
1.41757742e-01 -9.42481160e-01 -7.80750394e-01 3.50699812e-01
2.82768726e-01 -5.24888337e-01 -4.82718535e-02 3.63811612e-01
-1.20975697e+00 -4.81665522e-01 -7.27651775e-01 -3.38881195e-01
5.31809986e-01 -1.52230501e-01 1.10412204e+00 1.42279059e-01
-6.09191954e-02 -1.33311018e-01 3.01269561e-01 -3.84041935e-01
-5.03055394e-01 6.10013045e-02 6.60891712e-01 -4.03590977e-01
-3.74731094e-01 -7.16506600e-01 -4.45917577e-01 3.30036283e-01
-7.12276816e-01 3.22354697e-02 -2.03001812e-01 8.94926488e-01
1.47068426e-01 -1.44382030e-01 4.49762344e-01 -4.79826093e-01
9.84993994e-01 -6.25072598e-01 -1.28371418e+00 -7.12445751e-02
-9.07988310e-01 2.35392556e-01 8.54768813e-01 -3.90000820e-01
-9.47093070e-01 -4.07954082e-02 -1.01110704e-01 -5.91022432e-01
7.73880631e-02 9.10386980e-01 9.10212398e-01 -3.67827713e-01
1.06892729e+00 1.00947104e-01 3.11389446e-01 -3.66518736e-01
-6.88017532e-02 2.99124330e-01 5.21323323e-01 -6.88641489e-01
5.01976788e-01 2.31950015e-01 3.18674773e-01 -1.25267601e+00
-3.17130864e-01 -5.48050627e-02 -1.27347752e-01 -4.78041172e-01
6.38616800e-01 -7.78229594e-01 -8.41588259e-01 7.72322118e-01
-1.28166091e+00 -7.69171834e-01 -2.80515790e-01 4.33384210e-01
-2.37057924e-01 -1.69421867e-01 -9.14412439e-01 -1.09381390e+00
-3.41108143e-01 -1.14426458e+00 5.74111521e-01 3.26097906e-01
-3.90892655e-01 -1.31007123e+00 4.12306219e-01 -5.87126732e-01
1.28969514e+00 4.40567017e-01 6.18129373e-01 -1.90978914e-01
-7.50713050e-02 4.97087417e-03 1.06019964e-02 2.84326166e-01
-1.91931993e-01 6.03005111e-01 -9.95356560e-01 -1.81719229e-01
1.21591508e-01 -1.65989861e-01 7.32015669e-01 7.25788236e-01
1.03047383e+00 -5.16044557e-01 -1.12072557e-01 8.20569098e-01
1.11650991e+00 3.85800242e-01 1.65025085e-01 -1.71098318e-02
4.84598190e-01 6.65003538e-01 -4.35881354e-02 5.42758942e-01
-9.33660120e-02 2.93371111e-01 6.28000721e-02 -2.18799189e-01
6.31235301e-01 3.76123458e-01 3.39308918e-01 8.69708478e-01
-4.89782155e-01 -5.50505035e-02 -1.57023311e+00 4.62215036e-01
-1.86249781e+00 -1.09407926e+00 -3.89089793e-01 2.09262466e+00
7.67072618e-01 1.03565872e-01 -2.61307955e-01 -9.92403328e-02
3.76646042e-01 2.63636023e-01 -5.97734451e-01 -7.46988118e-01
-1.35726452e-01 2.11976916e-01 8.33871901e-01 1.21062922e+00
-9.82474983e-01 7.18305945e-01 7.32725143e+00 1.70020491e-01
-1.70062983e+00 -8.36258829e-02 6.84081554e-01 -4.34382230e-01
6.49438873e-02 5.30093461e-02 -6.50127828e-01 2.67158061e-01
1.73455381e+00 -4.51365858e-01 5.48690498e-01 7.25868642e-01
8.57907534e-01 -1.76109895e-01 -5.44147909e-01 3.62641513e-01
-4.80615675e-01 -1.75562680e+00 -2.36511528e-01 -1.23033665e-01
7.92060375e-01 3.78160447e-01 -1.54482424e-01 2.41919845e-01
4.52985346e-01 -1.18774796e+00 5.59379876e-01 9.27470505e-01
7.08298683e-01 -6.07081711e-01 5.29367745e-01 3.85962367e-01
-1.19514000e+00 2.02599205e-02 -3.02454382e-01 -8.24155688e-01
4.19901222e-01 7.43904650e-01 -2.15339646e-01 -1.51804055e-03
9.61766541e-01 5.10342121e-01 -2.17209309e-01 7.21596181e-01
4.96360362e-01 8.20000529e-01 -7.84946680e-01 -4.21870360e-03
3.00890774e-01 -4.21979100e-01 6.87069714e-01 9.79263127e-01
6.78007007e-01 4.14365411e-01 -1.13356486e-01 8.07181656e-01
3.96679252e-01 -2.71873981e-01 -9.55689847e-01 5.42371124e-02
3.51770759e-01 9.45632815e-01 -4.81079847e-01 -5.86559713e-01
1.01922058e-01 2.86830515e-01 2.57165223e-01 7.45672584e-01
-7.88298368e-01 -5.47519088e-01 1.13868487e+00 2.34116003e-01
-2.89056078e-03 -6.45957768e-01 -5.29386282e-01 -1.07581818e+00
-1.87776491e-01 -5.86857677e-01 7.15642646e-02 -9.25530314e-01
-9.28337932e-01 7.45413601e-01 2.12349985e-02 -9.73778546e-01
-6.60202086e-01 -5.62415898e-01 -8.04217398e-01 1.35053086e+00
-1.09227514e+00 -2.70769656e-01 -9.51798633e-02 2.23980322e-01
2.04668388e-01 2.50335038e-02 9.85280335e-01 1.67600617e-01
-7.59713709e-01 -2.64914390e-02 5.94011426e-01 -2.38593563e-01
4.76743579e-01 -9.95070934e-01 6.72313809e-01 1.04472125e+00
-7.87298024e-01 9.50165987e-01 1.23791695e+00 -6.92278147e-01
-1.26771522e+00 -1.05895019e+00 9.48732972e-01 -2.58068293e-01
9.37322855e-01 -1.26527026e-01 -1.54333878e+00 6.73576832e-01
4.21244919e-01 1.52958453e-01 2.42701724e-01 2.04278886e-01
1.77704841e-01 -1.07042067e-01 -7.65463173e-01 6.31303132e-01
6.55235827e-01 -3.61212641e-01 -3.17390084e-01 3.14072192e-01
5.38479626e-01 -5.83070993e-01 -1.13588881e+00 6.22440994e-01
5.11994243e-01 -9.27311301e-01 7.21657455e-01 -6.88081801e-01
4.35222805e-01 -4.08975035e-01 2.50750542e-01 -1.54659152e+00
-3.77260506e-01 -9.45114613e-01 -1.61542401e-01 7.73181438e-01
6.33457065e-01 -1.23083115e+00 2.84449160e-01 1.11785734e+00
-3.69067729e-01 -7.69144475e-01 -1.04804599e+00 -8.82788002e-01
7.36969650e-01 -5.43836057e-01 2.54147351e-01 1.20812583e+00
1.50740147e-01 -9.85928476e-02 -3.51572603e-01 6.77905381e-01
3.17384422e-01 -8.56473297e-02 4.37408149e-01 -1.15027857e+00
-6.62538111e-02 -7.40259588e-01 5.24633862e-02 -7.01543570e-01
8.36651549e-02 -3.51679146e-01 3.73199821e-01 -9.73730981e-01
-8.26865375e-01 -6.57105565e-01 -9.61121023e-02 4.98762041e-01
-9.43612680e-03 9.33277886e-03 -8.55237469e-02 3.96696419e-01
2.17704028e-01 5.98571718e-01 6.95528388e-01 4.61500108e-01
-5.16292572e-01 -2.10357860e-01 -3.31652677e-03 7.12652683e-01
9.83823180e-01 -4.06448364e-01 -1.79603368e-01 -5.32712042e-01
2.55795956e-01 4.42547619e-01 3.89581829e-01 -1.48948407e+00
4.69424665e-01 -5.35138845e-01 2.09262922e-01 -2.00286239e-01
2.92030841e-01 -4.45548683e-01 4.61837858e-01 6.87265217e-01
-4.79100257e-01 8.12226415e-01 8.84609222e-01 1.30133465e-01
-2.16463953e-01 1.41456693e-01 9.80056703e-01 -1.17999792e-01
-3.26257199e-01 6.97806403e-02 -1.00734627e+00 9.08120424e-02
7.92010009e-01 2.30530187e-01 -3.66169363e-01 -4.48252439e-01
-6.74190879e-01 5.10950744e-01 5.25600076e-01 1.37893051e-01
7.04781041e-02 -8.80356848e-01 -5.78379452e-01 4.33971286e-01
-4.11103368e-01 3.96409258e-02 3.35472465e-01 7.40548968e-01
-1.14353669e+00 3.34419519e-01 -2.17583239e-01 -5.18226087e-01
-8.44861269e-01 2.17011720e-01 1.15227473e+00 -6.06525578e-02
-4.45256293e-01 8.20866108e-01 -1.08092666e-01 -5.17121315e-01
3.89587916e-02 -7.15561211e-01 1.64898112e-01 3.10155540e-03
3.46902162e-01 5.39050758e-01 1.32702187e-01 -4.94795382e-01
-2.34364897e-01 4.43798125e-01 5.29524922e-01 -4.58056808e-01
1.40768707e+00 1.09350555e-01 -1.70326024e-01 6.91785514e-01
7.46963799e-01 -2.72077918e-01 -1.35633445e+00 1.03823721e-01
-2.65080899e-01 2.96688713e-02 3.28590900e-01 -3.80624831e-01
-1.11228716e+00 9.27381754e-01 4.82536674e-01 6.43772483e-01
9.51402128e-01 -7.36500323e-01 6.34035707e-01 7.14896441e-01
-1.72079071e-01 -1.03939199e+00 -8.40392232e-01 1.25642061e+00
1.09793258e+00 -9.93628502e-01 -1.74659323e-02 1.42885998e-01
-3.55562657e-01 1.20001817e+00 6.18706822e-01 -3.14254045e-01
1.20163167e+00 6.57608211e-01 4.27355170e-01 -1.40092894e-01
-1.10769618e+00 3.37891310e-01 -8.05209503e-02 -1.17435068e-01
6.95478737e-01 -1.34475619e-01 -1.15909025e-01 -1.89292040e-02
-3.23618501e-01 1.01237886e-01 6.34336591e-01 1.09167266e+00
-2.78057635e-01 -4.10826683e-01 -6.52661264e-01 5.22266567e-01
-3.12481344e-01 -3.22254717e-01 1.04809664e-01 4.20477778e-01
-2.70633906e-01 5.87281346e-01 4.07930553e-01 -1.23812363e-01
1.92280665e-01 4.47366744e-01 -4.02751774e-01 -1.07013911e-01
-7.88647711e-01 -1.68769658e-01 2.94369757e-01 -4.93228942e-01
-1.69645846e-01 -7.68037200e-01 -1.30551553e+00 -7.48860717e-01
-8.93579796e-02 4.60762113e-01 5.75754523e-01 6.42613471e-01
8.41301024e-01 6.67271495e-01 2.97244668e-01 -1.53966582e+00
-5.82751572e-01 -1.00335026e+00 -4.49130327e-01 1.20473378e-04
7.69191146e-01 -6.45950615e-01 -7.74754286e-01 5.61397634e-02]
|
[6.5612688064575195, 3.309206485748291]
|
4f0c5b50-4e23-4510-abb3-b7555e67cb9a
|
learning-multi-view-aggregation-in-the-wild
|
2204.07548
| null |
https://arxiv.org/abs/2204.07548v2
|
https://arxiv.org/pdf/2204.07548v2.pdf
|
Learning Multi-View Aggregation In the Wild for Large-Scale 3D Semantic Segmentation
|
Recent works on 3D semantic segmentation propose to exploit the synergy between images and point clouds by processing each modality with a dedicated network and projecting learned 2D features onto 3D points. Merging large-scale point clouds and images raises several challenges, such as constructing a mapping between points and pixels, and aggregating features between multiple views. Current methods require mesh reconstruction or specialized sensors to recover occlusions, and use heuristics to select and aggregate available images. In contrast, we propose an end-to-end trainable multi-view aggregation model leveraging the viewing conditions of 3D points to merge features from images taken at arbitrary positions. Our method can combine standard 2D and 3D networks and outperforms both 3D models operating on colorized point clouds and hybrid 2D/3D networks without requiring colorization, meshing, or true depth maps. We set a new state-of-the-art for large-scale indoor/outdoor semantic segmentation on S3DIS (74.7 mIoU 6-Fold) and on KITTI-360 (58.3 mIoU). Our full pipeline is accessible at https://github.com/drprojects/DeepViewAgg, and only requires raw 3D scans and a set of images and poses.
|
['Loic Landrieu', 'Bruno Vallet', 'Damien Robert']
|
2022-04-15
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Robert_Learning_Multi-View_Aggregation_in_the_Wild_for_Large-Scale_3D_Semantic_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Robert_Learning_Multi-View_Aggregation_in_the_Wild_for_Large-Scale_3D_Semantic_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['colorization']
|
['computer-vision']
|
[ 1.30070925e-01 -2.73633711e-02 -1.38336392e-02 -5.52559137e-01
-1.07653069e+00 -9.91540372e-01 2.73166865e-01 -2.95565405e-04
-2.67683983e-01 9.54853147e-02 -3.21862310e-01 -7.79556111e-02
1.63837031e-01 -9.88105595e-01 -1.21528649e+00 -1.21017173e-01
-4.98908106e-03 9.08231080e-01 5.72733521e-01 -1.15852952e-02
6.63129538e-02 9.89187837e-01 -1.85762084e+00 1.16423823e-01
8.06448877e-01 1.37708580e+00 2.05983341e-01 8.58396173e-01
-3.37656856e-01 -3.56031746e-01 -2.10837260e-01 -1.48601815e-01
9.64204133e-01 2.03527585e-01 -5.66318452e-01 4.37081367e-01
1.37786388e+00 -5.26495636e-01 7.81269446e-02 8.27009380e-01
4.82434034e-01 6.96139410e-02 2.34961122e-01 -1.31501210e+00
-4.36627477e-01 -7.42923468e-02 -7.68892169e-01 -2.51822203e-01
3.73537302e-01 3.39842767e-01 7.65879273e-01 -1.00628340e+00
5.97501695e-01 1.33380532e+00 9.80606556e-01 4.13517863e-01
-1.26166475e+00 -6.92006111e-01 4.66238081e-01 -3.32193851e-01
-1.36226010e+00 -2.08306178e-01 8.98053765e-01 -4.55329359e-01
1.12376046e+00 2.52294779e-01 1.13541365e+00 8.43121469e-01
-3.65212947e-01 6.47600591e-01 1.11694777e+00 -1.95735041e-02
2.15211838e-01 -1.91164136e-01 -6.96590692e-02 8.18914115e-01
6.76598847e-02 4.97919973e-03 -6.33279920e-01 -1.87462583e-01
1.19535959e+00 3.61251950e-01 2.55029276e-02 -8.21565628e-01
-1.19554579e+00 5.63963652e-01 6.95827961e-01 -3.55313450e-01
-2.75734335e-01 3.53287041e-01 -1.48454517e-01 4.26338837e-02
8.77827525e-01 2.16484457e-01 -9.57750618e-01 4.09717262e-02
-8.98590982e-01 2.92423964e-01 3.86987060e-01 1.21547580e+00
1.43454599e+00 -3.49487692e-01 5.78854442e-01 5.93395352e-01
3.34103018e-01 1.04047084e+00 -3.69444966e-01 -1.78298509e+00
5.52028239e-01 8.55859458e-01 8.80455077e-02 -6.67680442e-01
-5.41470706e-01 -2.82734513e-01 -2.86631733e-01 6.46088302e-01
3.47150654e-01 -4.24544737e-02 -1.48347163e+00 1.21089077e+00
7.65220821e-01 3.30666661e-01 -3.18955898e-01 1.04072106e+00
9.53110158e-01 3.99995178e-01 -4.26459730e-01 6.31594360e-01
1.04177248e+00 -8.94817829e-01 1.24906741e-01 -6.58124924e-01
2.94465691e-01 -6.49358094e-01 1.06013882e+00 4.15999830e-01
-1.49034488e+00 -5.66840351e-01 -1.00365114e+00 -6.03858054e-01
-6.14608943e-01 -7.88164288e-02 6.36666179e-01 4.61066514e-01
-1.38831675e+00 5.40962934e-01 -1.03707266e+00 -4.28265005e-01
7.75814295e-01 5.59436560e-01 -4.00446564e-01 -2.56188869e-01
-5.58760107e-01 2.87706554e-01 1.36097118e-01 9.96674970e-02
-6.61670148e-01 -1.07490432e+00 -8.81671667e-01 -2.31910437e-01
5.03676414e-01 -1.18001235e+00 1.11086428e+00 -8.40936303e-01
-1.23742783e+00 1.23002231e+00 -1.30273074e-01 -1.22281350e-01
5.79343379e-01 -6.53913558e-01 2.31096566e-01 3.73280466e-01
2.94059962e-01 1.24579787e+00 4.95769143e-01 -1.65675366e+00
-9.14245248e-01 -9.32267666e-01 2.28874385e-01 5.74930131e-01
2.80506432e-01 -5.32789290e-01 -9.49523985e-01 7.78540894e-02
7.27601647e-01 -1.00103104e+00 -4.00153011e-01 5.88205516e-01
-5.61907351e-01 2.48840023e-02 8.12002182e-01 -3.74137163e-01
1.45044386e-01 -2.00889802e+00 1.69521034e-01 3.79412293e-01
3.34025025e-01 -2.14169323e-01 -1.03067905e-01 -3.27813029e-02
2.88153589e-01 2.02357680e-01 -3.76945943e-01 -9.76498067e-01
-2.03031600e-02 2.62974411e-01 1.77485291e-02 5.47450960e-01
4.61055599e-02 8.82801056e-01 -8.77878010e-01 -4.01433885e-01
8.61982942e-01 6.58076286e-01 -5.91195345e-01 5.98113053e-02
-6.41843915e-01 4.99998748e-01 -4.70160812e-01 1.12351704e+00
1.10312045e+00 -3.71885568e-01 -2.95571178e-01 -2.19591349e-01
-2.70413846e-01 1.02939248e-01 -1.24215662e+00 2.56059647e+00
-5.49936414e-01 3.29566240e-01 3.37264776e-01 -3.73718143e-01
8.02641094e-01 -9.50532556e-02 8.43528867e-01 -4.29705560e-01
8.40526447e-02 3.01857203e-01 -9.07988429e-01 -1.49767175e-01
3.00036877e-01 3.80123615e-01 -1.04923129e-01 1.73341900e-01
1.58346966e-01 -9.78172839e-01 -2.01478273e-01 3.35039087e-02
8.02671552e-01 5.98067582e-01 -3.60138595e-01 2.46205196e-01
5.57804331e-02 3.56385976e-01 4.40193743e-01 6.16294503e-01
1.91622972e-01 1.23508310e+00 2.11895764e-01 -5.26740849e-01
-1.16250861e+00 -1.56467509e+00 -9.61571932e-02 7.96299756e-01
4.35133696e-01 -1.26191378e-01 -5.90205014e-01 -7.18539059e-01
3.55794668e-01 4.97986376e-01 -4.79477584e-01 4.70437020e-01
-5.47291994e-01 -2.44948924e-01 9.47538391e-02 6.79160416e-01
5.18459320e-01 -4.96703863e-01 -6.82371616e-01 -1.08529337e-01
-4.20660898e-02 -1.29950917e+00 -3.58024091e-01 4.02977914e-01
-1.07861030e+00 -1.19245970e+00 -3.96845758e-01 -3.95161957e-01
6.99834168e-01 6.64921045e-01 1.46298575e+00 -1.60047606e-01
-1.41658083e-01 8.50525975e-01 -2.85189122e-01 -4.77379054e-01
1.15156554e-01 3.26948434e-01 -2.67996818e-01 -3.84815067e-01
2.57878184e-01 -9.36237216e-01 -8.70214581e-01 2.69474387e-01
-6.75036490e-01 4.04456168e-01 3.81957918e-01 1.46109253e-01
1.23968863e+00 -4.71837938e-01 -3.47363472e-01 -6.62603915e-01
-2.77010471e-01 -3.55007768e-01 -9.93842781e-01 -8.82845372e-02
-1.96396455e-01 -2.94297546e-01 2.22825050e-01 1.19106874e-01
-7.67922342e-01 7.06971228e-01 -1.86961368e-01 -1.08781791e+00
-6.84718668e-01 -1.84033275e-01 -2.86373764e-01 -2.38291711e-01
5.95280349e-01 -8.92906636e-02 1.79412402e-02 -6.02589011e-01
8.46141338e-01 4.63141024e-01 5.09621263e-01 -5.32389581e-01
9.50420260e-01 1.17909694e+00 9.60978270e-02 -5.59644997e-01
-1.07210958e+00 -6.69448197e-01 -1.26828408e+00 -2.67630786e-01
1.32903039e+00 -1.18981326e+00 -4.66755599e-01 5.11763096e-01
-1.23036218e+00 -5.95968604e-01 -4.00967449e-01 2.09185973e-01
-7.34842777e-01 -3.51321995e-02 -2.91403979e-01 -3.90244722e-01
-2.91869938e-01 -1.08538759e+00 1.86791432e+00 1.85839683e-01
9.44116041e-02 -7.98598528e-01 -1.74545884e-01 7.67052829e-01
-4.50224951e-02 5.82014322e-01 3.44677716e-01 -5.24471030e-02
-1.23991990e+00 -1.22427024e-01 -3.04148108e-01 3.36849362e-01
6.81354329e-02 8.00605416e-02 -1.29151285e+00 5.23613505e-02
-2.80131191e-01 -3.80727768e-01 7.68473327e-01 7.10229933e-01
1.42277479e+00 2.90573359e-01 -3.95516068e-01 1.32096148e+00
1.45138633e+00 -1.76787451e-01 9.11433101e-02 2.86588699e-01
1.27067137e+00 5.67478240e-01 4.49924260e-01 2.25056827e-01
8.26582134e-01 6.03925347e-01 9.75841820e-01 -4.57092881e-01
-2.38245174e-01 -2.91627318e-01 -1.68822423e-01 3.95827115e-01
-1.49294168e-01 -5.75526692e-02 -1.05156219e+00 4.40520018e-01
-1.72302437e+00 -4.25878227e-01 -3.24411124e-01 2.27958202e+00
3.91925246e-01 1.26547799e-01 4.49235179e-02 -4.74200040e-01
3.90896767e-01 1.88288271e-01 -9.88810182e-01 -4.39611115e-02
-1.67477727e-01 3.10084879e-01 1.07348585e+00 7.29685724e-01
-1.32579648e+00 9.83931124e-01 5.36285162e+00 5.46918690e-01
-1.14263606e+00 1.34032533e-01 7.97663689e-01 -6.40545011e-01
-5.07665813e-01 -8.05641264e-02 -7.72493839e-01 1.23590223e-01
5.29574215e-01 5.89945078e-01 4.30113047e-01 1.03422678e+00
1.12102285e-01 -2.47373894e-01 -1.03164804e+00 1.18208361e+00
1.41866878e-02 -1.49695599e+00 -2.30175823e-01 2.29815289e-01
1.07183242e+00 9.15176451e-01 8.69804174e-02 -3.28935295e-01
6.01088226e-01 -8.66098225e-01 9.43008840e-01 4.37618464e-01
8.77201855e-01 -5.43776274e-01 2.89157927e-01 3.33698571e-01
-1.34976184e+00 1.78765461e-01 -1.81093097e-01 8.58695433e-02
3.00243229e-01 6.26026988e-01 -6.62468851e-01 6.57999516e-01
1.12628174e+00 9.79465067e-01 -7.56599665e-01 9.24328864e-01
-9.47054476e-02 -5.42544127e-02 -1.06318736e+00 4.29099470e-01
3.42077255e-01 -3.09874922e-01 4.15986151e-01 6.84390604e-01
5.83531559e-01 -1.93532884e-01 4.20372546e-01 1.07096839e+00
-3.91484164e-02 -3.38154107e-01 -6.06845796e-01 5.27603209e-01
5.75616300e-01 1.36550307e+00 -9.62339878e-01 -3.70068759e-01
-5.35264313e-01 1.04203606e+00 2.35007852e-01 3.98009717e-01
-7.12000787e-01 1.25799060e-01 9.39075410e-01 4.20832217e-01
4.53352362e-01 -7.01466978e-01 -8.00687671e-01 -1.03792477e+00
1.72517955e-01 -2.05038249e-01 1.80617049e-02 -1.06815386e+00
-1.28048444e+00 4.02944952e-01 2.14207470e-01 -1.41514540e+00
1.68893695e-01 -4.67996538e-01 -1.96286842e-01 9.44053769e-01
-1.56093287e+00 -1.46448171e+00 -8.16716492e-01 5.48908830e-01
6.67013645e-01 5.47821641e-01 5.27124703e-01 1.61385432e-01
-1.00826323e-01 -3.52512710e-02 -1.30657181e-01 -9.87483785e-02
4.28291380e-01 -1.54945362e+00 9.31824982e-01 5.58527470e-01
1.14509679e-01 1.04877420e-01 1.57812878e-01 -7.10653663e-01
-1.48856926e+00 -1.22885752e+00 2.83491999e-01 -9.75700140e-01
1.78058326e-01 -7.40339696e-01 -6.31765544e-01 7.49762475e-01
-1.39149144e-01 4.31778014e-01 3.74019265e-01 4.85188179e-02
-2.73397923e-01 -2.32957169e-01 -1.22227430e+00 3.80365700e-01
1.72660291e+00 -3.78599256e-01 -6.54399842e-02 4.93886292e-01
1.08376431e+00 -1.18087220e+00 -9.41019654e-01 3.96104991e-01
4.15878952e-01 -1.29409170e+00 1.51451623e+00 3.23185138e-02
2.62650669e-01 -5.32099009e-01 -3.04504961e-01 -1.12854028e+00
8.42192769e-03 -2.89783418e-01 1.65571377e-01 7.64402390e-01
4.08156335e-01 -5.79362452e-01 1.24981308e+00 7.54115164e-01
-6.03826284e-01 -7.37829864e-01 -1.13157320e+00 -5.95572770e-01
5.63578506e-04 -9.23474193e-01 8.67771327e-01 7.32769489e-01
-1.03261018e+00 -1.08108530e-02 1.99465767e-01 7.23933101e-01
9.08205271e-01 4.32620078e-01 1.20438671e+00 -1.41364419e+00
8.50041360e-02 -4.08421606e-01 -2.90780038e-01 -1.29914415e+00
-6.51005507e-02 -8.20855439e-01 -7.84861594e-02 -1.96325183e+00
-4.17310804e-01 -8.18008721e-01 1.87286720e-01 5.16427338e-01
4.94992808e-02 5.22694290e-01 1.70168921e-01 1.84212431e-01
-6.82516754e-01 3.71750027e-01 1.17333186e+00 7.33759208e-03
-5.25810003e-01 1.76394377e-02 -3.27775627e-01 9.72543180e-01
9.20772374e-01 -2.53024548e-01 -4.07713890e-01 -9.45753694e-01
3.56854171e-01 -5.54307923e-02 8.75417113e-01 -1.28596246e+00
3.03470884e-02 -2.04660535e-01 8.51092696e-01 -1.29074836e+00
1.11356664e+00 -1.10317230e+00 3.62519652e-01 -1.18554510e-01
1.79410100e-01 2.05897037e-02 2.41349041e-01 4.88847822e-01
2.35575199e-01 2.07837999e-01 5.08937657e-01 -7.47546315e-01
-8.23821902e-01 7.78143704e-01 3.73125345e-01 -1.34652257e-01
1.09638774e+00 -7.13280737e-01 -2.28459150e-01 3.41444649e-02
-7.49647200e-01 5.54958284e-01 1.22089028e+00 5.27684331e-01
8.02462220e-01 -1.05892074e+00 -3.49410385e-01 4.81582582e-01
2.55032983e-02 1.18616092e+00 4.83107746e-01 5.70612729e-01
-8.94448519e-01 1.16272673e-01 -4.58769724e-02 -1.24509048e+00
-1.16850185e+00 9.70929712e-02 5.55862367e-01 3.25447619e-01
-7.74866998e-01 1.08426690e+00 1.89243689e-01 -9.68047380e-01
1.24787495e-01 -7.43062317e-01 4.07748789e-01 -7.22527578e-02
-7.31636584e-02 3.62575799e-01 1.77289754e-01 -5.12253702e-01
-3.96405160e-01 1.20652032e+00 3.00277531e-01 -1.61142185e-01
1.45182085e+00 -3.22597951e-01 -8.20779353e-02 5.60356140e-01
1.29655993e+00 -9.57321823e-02 -1.93129981e+00 -9.33263823e-02
-6.45453513e-01 -7.42305577e-01 2.50809968e-01 -6.63729489e-01
-1.31106722e+00 9.16384101e-01 7.12189078e-01 5.49616814e-02
9.10065532e-01 5.39756298e-01 9.94384050e-01 1.33205429e-01
6.29552543e-01 -9.81955528e-01 -6.63583800e-02 4.10506606e-01
6.00140870e-01 -1.53614807e+00 1.02468573e-01 -7.66357780e-01
-2.41268545e-01 1.11402714e+00 8.67218971e-01 -3.10357869e-01
7.38083780e-01 1.81556329e-01 3.06598574e-01 -5.16791284e-01
-2.79241621e-01 -3.31868440e-01 3.44802260e-01 7.15734780e-01
-1.95028141e-01 1.47707269e-01 5.87664783e-01 -7.37147480e-02
-3.63493919e-01 -1.04293719e-01 1.93353564e-01 9.93216157e-01
-3.81250083e-01 -8.91793311e-01 -6.54456317e-01 5.34841657e-01
1.13427855e-01 9.05128121e-02 -3.48176926e-01 5.72243273e-01
6.55895948e-01 6.23809099e-01 6.45064175e-01 -4.38957661e-01
4.71963137e-01 -1.23219445e-01 5.10466218e-01 -8.56229484e-01
-3.22806686e-01 2.27601498e-01 -5.53811938e-02 -1.19763160e+00
-5.96156776e-01 -8.32170427e-01 -1.32961190e+00 -1.80177256e-01
-2.65499409e-02 -4.87269580e-01 1.11948013e+00 5.98224759e-01
8.12851667e-01 2.56228268e-01 5.56898415e-01 -1.65997910e+00
1.86830908e-01 -4.76048142e-01 -3.83980095e-01 2.21491396e-01
4.69121993e-01 -6.26801729e-01 -4.38316703e-01 -1.19049652e-02]
|
[8.36036491394043, -2.960167407989502]
|
a214e67d-7d63-463b-b5c0-7a7c7fc1ae07
|
how-search-engine-marketing-influences-user
|
2301.10086
| null |
https://arxiv.org/abs/2301.10086v1
|
https://arxiv.org/pdf/2301.10086v1.pdf
|
How search engine marketing influences user knowledge gain: Development and empirical testing of an information search behavior model
|
People use search engines to find answers to questions related to their health, finances, or other socially relevant issues. However, most users are unaware that search results are considerably influenced by search engine marketing (SEM). SEM measures are driven by commercial, political, or other motives. Due to these motivations, two questions arise: What information quality is mediated through SEM? And how is collecting documents of different quality affecting user knowledge gain? Both questions are not considered by existing models of information behavior. Hence, the doctoral research project described in this paper aims to develop and empirically test an information search behavior model on the influences of SEM on user knowledge gain and thereby contribute to the search as learning body of research.
|
['Sebastian Schultheiß']
|
2023-01-24
| null | null | null | null |
['marketing']
|
['miscellaneous']
|
[-2.52501637e-01 4.05863136e-01 -9.78669941e-01 1.83644384e-01
-1.75199613e-01 -3.06031734e-01 5.28424680e-01 6.15087807e-01
-7.56465137e-01 2.32018247e-01 5.84386826e-01 -9.66415823e-01
-6.29394829e-01 -6.65255368e-01 -4.56321061e-01 2.36850306e-01
7.61708677e-01 4.01428044e-02 3.34796250e-01 -1.37237117e-01
9.21548247e-01 1.38166519e-02 -1.16743088e+00 -3.43950957e-01
1.19498098e+00 3.94490451e-01 1.43522471e-01 1.52683426e-02
-3.64224523e-01 4.97346252e-01 -2.65936852e-01 -6.73053384e-01
-1.97903216e-01 -4.08553571e-01 -9.52483058e-01 4.28344496e-02
1.92370966e-01 -6.74690485e-01 -1.94389954e-01 1.10533333e+00
2.66952693e-01 2.64085755e-02 5.10046661e-01 -7.09365189e-01
-1.13238740e+00 5.44469893e-01 3.00468896e-02 4.45154756e-01
4.70701128e-01 -4.51614484e-02 9.88197207e-01 -4.66844052e-01
9.41826522e-01 1.18562794e+00 1.05852544e-01 -5.64947538e-02
-1.18624246e+00 -5.95252872e-01 -2.70806551e-02 1.06800295e-01
-8.30803871e-01 -5.37122786e-01 4.24606413e-01 -6.24426305e-01
5.10391414e-01 3.65198761e-01 9.13289547e-01 1.09239030e+00
2.97298282e-01 3.22899520e-01 1.50520909e+00 -4.64180112e-01
1.39127433e-01 1.22770989e+00 6.86257064e-01 2.21421435e-01
1.03942597e+00 9.62785557e-02 -5.66593051e-01 -3.10700327e-01
4.92172629e-01 1.78531669e-02 -2.64553756e-01 4.78827655e-02
-6.59101903e-01 1.21150506e+00 4.85842973e-01 8.00331831e-01
-5.86706042e-01 -4.15066510e-01 -2.96556056e-01 5.57718813e-01
2.80487746e-01 8.43635798e-01 -2.61773229e-01 -3.98966461e-01
-6.96270287e-01 1.49847433e-01 1.29235113e+00 2.69664466e-01
6.68978512e-01 -2.38510072e-01 -2.00767189e-01 6.78323388e-01
1.01312888e+00 5.00483990e-01 6.12384081e-01 -7.19848156e-01
-1.32349551e-01 8.45447481e-01 9.11284983e-02 -1.29588914e+00
-8.27669352e-02 -8.64828706e-01 1.97603866e-01 -2.99254566e-01
4.77582902e-01 -1.58461869e-01 -3.89941216e-01 1.27756000e+00
1.48063868e-01 -8.69497478e-01 -4.40041542e-01 9.41620290e-01
9.28310990e-01 1.04920253e-01 3.57463986e-01 -3.09645653e-01
1.52923107e+00 -3.73482972e-01 -8.26390326e-01 -7.04322577e-01
4.43932891e-01 -9.17619705e-01 1.05395854e+00 -1.77806988e-01
-1.15139508e+00 -4.15120542e-01 -6.84506714e-01 -1.70755520e-01
-7.65613496e-01 -2.88343549e-01 6.50169969e-01 1.13400650e+00
-9.96808767e-01 3.52128446e-01 -5.00148594e-01 -9.11256373e-01
1.72228724e-01 -2.49537349e-01 4.83865827e-01 -8.03921148e-02
-1.26220274e+00 1.19457328e+00 -1.22265533e-01 -3.34745854e-01
4.90099490e-02 -3.98964465e-01 -4.38976556e-01 4.74746525e-02
3.80652368e-01 -9.95532155e-01 1.12440205e+00 -8.86825681e-01
-9.59899485e-01 5.58342099e-01 -1.60892963e-01 1.31761104e-01
1.25326082e-01 -4.45567332e-02 -5.51716447e-01 4.95586134e-02
7.18708873e-01 1.21332049e-01 6.62451923e-01 -7.72216022e-01
-4.39868540e-01 -9.25309300e-01 6.86291307e-02 3.13069552e-01
-6.98029399e-01 2.92848855e-01 -6.34284556e-01 -2.09749356e-01
6.72025084e-02 -5.40850341e-01 9.70480684e-03 -3.59205067e-01
-6.28270349e-03 -3.31444383e-01 2.47200415e-01 -9.83964622e-01
1.58847284e+00 -1.65881264e+00 -4.73536760e-01 4.57152992e-01
2.20883325e-01 2.67521832e-02 2.28093997e-01 7.90293932e-01
7.32391715e-01 1.11310554e+00 9.44108129e-01 7.73999274e-01
1.84235692e-01 -2.02310175e-01 2.27288693e-01 1.51585892e-01
-4.60156500e-01 1.11641085e+00 -6.63956642e-01 -5.02504349e-01
-2.29020834e-01 4.81235892e-01 -3.65588605e-01 -3.74587268e-01
-7.96922445e-02 -4.91966680e-02 -1.04903162e+00 8.70062530e-01
1.58516496e-01 -8.16769361e-01 5.37369996e-02 1.29988924e-01
-3.57337087e-01 8.19932640e-01 -5.49209595e-01 7.73819089e-01
-1.79426402e-01 7.18879282e-01 1.30511031e-01 -4.52400208e-01
3.69463652e-01 7.63245672e-02 1.07284106e-01 -1.39582121e+00
4.11566347e-01 2.90124387e-01 2.53879964e-01 -6.41191244e-01
4.82141852e-01 1.18490625e-02 5.05884945e-01 4.68713343e-01
-5.20925522e-01 3.25783104e-01 2.25421265e-01 2.92151749e-01
1.07297277e+00 -3.37496907e-01 2.82024860e-01 -4.51821655e-01
-9.81558561e-02 3.07178885e-01 3.07295054e-01 1.05847442e+00
-5.30046485e-02 -5.31052828e-01 2.71943361e-01 3.53041530e-01
-5.16834259e-01 -4.84566063e-01 -5.93182504e-01 8.08713138e-01
3.67444903e-01 -2.62187809e-01 -6.62304103e-01 -3.14133078e-01
2.69422948e-01 1.23452377e+00 -1.72464654e-01 -6.88580051e-02
3.66970181e-01 -3.10333252e-01 -3.07105333e-01 -7.68578649e-02
6.49741054e-01 -8.51744950e-01 -6.29785657e-01 2.49832317e-01
-4.67239380e-01 -7.57327080e-01 -4.81643558e-01 -5.73654532e-01
-1.27223659e+00 -1.18122733e+00 -7.73120522e-01 -3.99209350e-01
3.31644326e-01 6.17282331e-01 1.09287322e+00 4.25718337e-01
-8.31029788e-02 9.45266426e-01 -3.82648468e-01 -5.85433066e-01
-2.90716350e-01 2.55609214e-01 -5.81606627e-01 -5.20908713e-01
9.02045667e-01 -1.70147657e-01 -7.67954230e-01 3.16505611e-01
-9.58908737e-01 -5.43208897e-01 9.41799462e-01 1.47554949e-01
-8.52257088e-02 2.59005904e-01 7.45889664e-01 -7.78954685e-01
1.46163404e+00 -9.72114563e-01 -5.45368671e-01 7.77291730e-02
-1.55395341e+00 -1.45582005e-01 -4.59532231e-01 -3.61839980e-01
-1.04079199e+00 -8.98271501e-01 -2.41531562e-02 5.89882314e-01
-6.44816831e-02 1.19380379e+00 3.12697068e-02 -3.71821553e-01
9.30276453e-01 -1.65939137e-01 4.11539853e-01 -5.72423398e-01
-1.44073293e-01 7.57026732e-01 -6.54095590e-01 -1.15732059e-01
6.77279770e-01 1.45949721e-01 -5.32530248e-01 -1.34756529e+00
-8.15961480e-01 -6.67957783e-01 3.58104050e-01 -3.92894626e-01
8.66586685e-01 -7.36426890e-01 -1.07600343e+00 -2.97131091e-01
-3.74436200e-01 -5.43303154e-02 2.13450506e-01 8.30241859e-01
1.66194946e-01 2.39619181e-01 -5.29659808e-01 -1.07549572e+00
-1.78983092e-01 -9.17775095e-01 1.91053659e-01 5.03431201e-01
-5.20479262e-01 -8.79538059e-01 -1.45820275e-01 1.12497663e+00
1.00022030e+00 -3.35088193e-01 1.08049762e+00 -5.80486000e-01
-8.44772637e-01 -3.44562620e-01 -1.28924489e-01 -3.06853652e-01
-2.32638400e-02 -4.90950584e-01 -1.92652434e-01 -1.34549975e-01
3.32992196e-01 -1.63773268e-01 5.75375915e-01 5.59096634e-01
3.43825221e-01 -7.84828722e-01 -5.34190357e-01 -4.01929200e-01
1.48018050e+00 2.79127836e-01 3.28935921e-01 8.67745280e-01
-1.81923181e-01 7.64350235e-01 4.00257289e-01 1.74573362e-01
4.73194927e-01 6.68261886e-01 -1.16975857e-02 4.83972132e-01
7.63680264e-02 -6.25564992e-01 2.40842685e-01 3.39499474e-01
2.59137541e-01 -9.41263065e-02 -7.15117037e-01 7.02581465e-01
-1.25980294e+00 -7.25113869e-01 -3.18976164e-01 2.05411148e+00
4.02428836e-01 2.38621563e-01 2.14361012e-01 -3.10549647e-01
2.22252458e-01 -9.88067314e-02 -6.52886271e-01 -9.99301672e-02
2.86168247e-01 -2.41386089e-02 7.47314513e-01 6.90375090e-01
-2.00027674e-01 5.43757617e-01 6.37968254e+00 2.43152454e-01
-4.74778950e-01 -3.52063701e-02 7.46721208e-01 2.42965966e-01
-1.08195734e+00 4.54972237e-01 -9.90512252e-01 3.78207654e-01
7.56749570e-01 -6.45824492e-01 4.15598154e-01 8.13017070e-01
5.05141854e-01 -3.36554468e-01 -5.54691672e-01 4.92756367e-01
-2.92393919e-02 -9.47257459e-01 -2.59923756e-01 7.92885423e-01
4.53772843e-01 -3.45769256e-01 4.10119593e-01 4.90415275e-01
3.65345687e-01 -6.20898247e-01 4.14029092e-01 3.99378777e-01
-3.92879546e-03 -1.31331876e-01 4.26655710e-01 6.26731575e-01
-1.75865471e-01 3.77518358e-03 -3.50926071e-01 -2.74412572e-01
1.86270699e-01 7.13696241e-01 -7.22934365e-01 -1.07326478e-01
5.44890285e-01 1.99681967e-01 -8.60556066e-01 1.05324614e+00
-1.13337003e-02 1.06011879e+00 -3.19729708e-02 -6.83055460e-01
2.43713453e-01 -2.65425503e-01 6.50989592e-01 4.52201515e-01
3.19967151e-01 -5.87779433e-02 -5.46223521e-01 1.30269611e+00
7.65996054e-02 6.51516914e-01 -5.59489608e-01 -9.44181859e-01
5.05765259e-01 1.01373708e+00 -8.99925590e-01 6.97811553e-03
-8.23610783e-01 5.70202231e-01 -1.33770019e-01 7.23064482e-01
-3.42934988e-02 2.49640439e-02 5.34253716e-01 1.08312011e+00
1.88645720e-02 7.52923861e-02 -4.31354135e-01 -9.34027672e-01
2.99739596e-02 -1.03771973e+00 3.12193185e-01 -4.25844043e-01
-9.32201028e-01 -2.45426446e-01 -8.10543373e-02 -1.52880847e-01
-7.77281672e-02 -2.28434607e-01 -3.33665637e-03 9.40457821e-01
-1.52200866e+00 -3.96371484e-01 -1.72900006e-01 1.84430212e-01
2.28904322e-01 2.22683564e-01 3.01112205e-01 1.47002339e-01
-1.60323754e-01 -5.42575354e-03 1.51033968e-01 -4.26928043e-01
6.73054457e-01 -6.17153764e-01 -8.58808011e-02 4.55034167e-01
-8.94658864e-02 9.53441799e-01 5.44875920e-01 -1.29884279e+00
-1.35146272e+00 -1.30312189e-01 1.56586862e+00 -5.49879074e-01
5.21988034e-01 4.08018500e-01 -5.05502701e-01 4.77326661e-01
4.24856335e-01 -1.38045800e+00 8.93915713e-01 6.34930789e-01
1.31437168e-01 2.53701627e-01 -1.01702595e+00 9.21499014e-01
8.18700552e-01 -4.08608168e-01 -4.37923223e-01 3.20940852e-01
4.49747711e-01 3.34581524e-01 -1.05021238e+00 1.87879428e-02
4.81837630e-01 -8.42064142e-01 1.05576324e+00 -4.08528507e-01
3.94602984e-01 6.72449350e-01 2.86033481e-01 -8.75573039e-01
-6.08864486e-01 -2.42310643e-01 3.53470176e-01 9.11629796e-01
7.14961350e-01 -9.42325532e-01 8.65149081e-01 1.58630872e+00
3.87928188e-01 -1.94504291e-01 -4.65993911e-01 -2.39473417e-01
1.09682985e-01 -8.74871239e-02 1.91445783e-01 9.71152425e-01
2.52602249e-01 5.55391252e-01 1.58286318e-01 -2.02244863e-01
7.37070441e-01 -9.95258912e-02 5.29375434e-01 -1.52849948e+00
-2.75515169e-01 -8.86029303e-01 1.35575831e-01 -1.19618177e+00
-2.27710649e-01 -6.55656815e-01 -7.49610066e-01 -1.98014641e+00
4.47826236e-01 1.16281539e-01 1.76227853e-01 -4.01344092e-04
-3.32217783e-01 -5.35773754e-01 2.12708190e-01 4.16505396e-01
-3.28682810e-01 3.57130557e-01 1.18344057e+00 2.47602463e-01
-3.50245357e-01 3.96444291e-01 -1.75989246e+00 4.57785279e-01
8.74964297e-01 -2.04856426e-01 -5.34384906e-01 -1.28911033e-01
9.45390284e-01 6.80659175e-01 3.54628384e-01 -3.88045341e-01
7.73123428e-02 -5.05471647e-01 3.05980593e-01 -2.49630719e-01
7.74671808e-02 -1.14843118e+00 3.09782445e-01 5.33957243e-01
-5.24339616e-01 1.02930449e-01 1.20754531e-02 3.61082852e-01
-1.20641574e-01 -6.47165060e-01 -7.94857293e-02 -2.01967612e-01
-6.17176779e-02 -1.03847504e-01 -6.85141027e-01 2.05034047e-01
3.56722265e-01 -3.85894924e-01 -2.31394917e-01 -1.01233470e+00
-5.72860599e-01 2.46040374e-02 3.95333558e-01 5.83499908e-01
2.10534155e-01 -1.11202955e+00 -1.95431411e-01 -2.40503550e-01
-1.68636337e-01 -9.24408615e-01 -1.10837713e-01 9.23931181e-01
5.86385988e-02 1.10613632e+00 1.43830642e-01 3.07435989e-01
-9.95529473e-01 2.51516312e-01 -1.63167924e-01 -2.17384957e-02
-3.42517942e-02 4.26680923e-01 -1.98630154e-01 1.42679317e-02
1.79064840e-01 1.62287936e-01 -5.24755895e-01 2.30313256e-01
3.42367709e-01 7.96880782e-01 -2.69923359e-01 -3.09821635e-01
-3.24267298e-02 -5.19745313e-02 -1.53110489e-01 -4.61885154e-01
9.00337458e-01 -3.22929800e-01 -1.36399224e-01 1.72711328e-01
1.05880523e+00 1.18479490e-01 -1.60023615e-01 -4.32643741e-01
2.78506726e-01 -9.69418049e-01 2.66936451e-01 -9.83710706e-01
-7.30093002e-01 1.26327321e-01 4.75832343e-01 6.17905438e-01
7.04570413e-01 2.23831981e-01 3.91214222e-01 3.75626236e-01
-1.80014238e-01 -1.44158769e+00 3.39494705e-01 -9.93554592e-02
6.14427567e-01 -1.26020885e+00 -2.46447012e-01 -3.04352760e-01
-4.55982327e-01 6.32781267e-01 4.72121537e-01 8.51733923e-01
1.20380020e+00 -4.42006141e-01 2.09611412e-02 -6.66697145e-01
-3.60233098e-01 -5.14705360e-01 5.40568709e-01 2.54311144e-01
8.35245609e-01 -2.81879246e-01 -1.71836078e+00 4.19389278e-01
-3.63254696e-01 3.51842225e-01 1.94136202e-01 7.44228661e-01
-8.29295814e-01 -1.05094433e+00 -2.63629556e-01 9.68403161e-01
-9.50114727e-01 -6.90228567e-02 -1.00427783e+00 6.94177210e-01
-6.94769621e-01 1.51030612e+00 -3.01024586e-01 -2.31162831e-01
5.07465973e-02 5.82582764e-02 -2.65243351e-01 -2.63636321e-01
-7.27457047e-01 3.54586601e-01 4.16583627e-01 -3.52676064e-01
-3.58573496e-01 -8.03585529e-01 -2.04485595e-01 -3.01613003e-01
-7.35745311e-01 4.71660614e-01 9.40210044e-01 7.22957373e-01
5.97085774e-01 3.54385823e-02 5.55854514e-02 5.14750302e-01
-6.79838121e-01 -1.00829279e+00 -4.21270877e-01 3.45781505e-01
-1.75303668e-01 -2.17753902e-01 -4.37449187e-01 -5.70732176e-01]
|
[10.039582252502441, 6.324141502380371]
|
44d53e6d-942e-47e7-bc8f-e559b1972a75
|
multimodal-chain-of-thought-reasoning-in
|
2302.00923
| null |
https://arxiv.org/abs/2302.00923v4
|
https://arxiv.org/pdf/2302.00923v4.pdf
|
Multimodal Chain-of-Thought Reasoning in Language Models
|
Large language models (LLMs) have shown impressive performance on complex reasoning by leveraging chain-of-thought (CoT) prompting to generate intermediate reasoning chains as the rationale to infer the answer. However, existing CoT studies have focused on the language modality. We propose Multimodal-CoT that incorporates language (text) and vision (images) modalities into a two-stage framework that separates rationale generation and answer inference. In this way, answer inference can leverage better generated rationales that are based on multimodal information. With Multimodal-CoT, our model under 1 billion parameters outperforms the previous state-of-the-art LLM (GPT-3.5) by 16 percentage points (75.17%->91.68% accuracy) on the ScienceQA benchmark and even surpasses human performance. Code is publicly available available at https://github.com/amazon-science/mm-cot.
|
['Alex Smola', 'George Karypis', 'Hai Zhao', 'Mu Li', 'Aston Zhang', 'Zhuosheng Zhang']
|
2023-02-02
| null | null | null | null |
['science-question-answering']
|
['miscellaneous']
|
[-3.95114943e-02 4.88067508e-01 -1.94217891e-01 -4.55118924e-01
-1.43119955e+00 -8.24563324e-01 9.04567719e-01 1.72134787e-01
-1.84215039e-01 5.92825890e-01 7.22719073e-01 -7.89588869e-01
3.03729802e-01 -5.00301600e-01 -8.46816123e-01 -1.52973175e-01
5.58218420e-01 5.97456872e-01 -2.14361050e-03 -1.73033014e-01
4.61210936e-01 -1.35371596e-01 -1.20364511e+00 1.06770933e+00
8.07703912e-01 7.21414626e-01 -1.37773767e-01 1.22407067e+00
-5.81475198e-01 1.67864048e+00 -3.47053409e-01 -1.03152752e+00
-1.88588858e-01 -2.86501616e-01 -1.05101764e+00 -4.45879221e-01
5.85839331e-01 -5.03107190e-01 -1.93053633e-01 6.87573791e-01
4.13378447e-01 -1.60505041e-01 5.86833000e-01 -1.30518520e+00
-1.01008034e+00 8.92778456e-01 -6.65793478e-01 -1.41974480e-03
9.78946328e-01 6.13813341e-01 1.30105042e+00 -1.30159211e+00
5.35256803e-01 1.67229962e+00 2.97956228e-01 7.31454134e-01
-1.18114746e+00 -4.25849080e-01 1.55810565e-01 3.20787817e-01
-1.08032978e+00 -6.50736511e-01 4.37637001e-01 -4.14608747e-01
1.35644221e+00 4.74016756e-01 3.33855271e-01 1.20887280e+00
3.03427637e-01 1.22258806e+00 1.16567075e+00 -3.96299750e-01
1.58860043e-01 -8.48761871e-02 1.90589055e-01 1.12340868e+00
1.66928336e-01 -4.52722371e-01 -9.88912284e-01 -2.48466194e-01
3.56122047e-01 -2.94129997e-01 -5.65354265e-02 2.08477825e-01
-1.42598927e+00 8.84747803e-01 4.98250663e-01 -2.44338781e-01
-6.47544920e-01 6.46702051e-01 7.96996206e-02 5.56189083e-02
1.10143796e-01 6.32456541e-01 -2.45979667e-01 -3.34058076e-01
-7.16929913e-01 4.67015773e-01 1.00144494e+00 7.83127725e-01
5.20815551e-01 -2.27114543e-01 -7.05591679e-01 5.22667766e-01
8.70685399e-01 8.28683436e-01 -7.51752928e-02 -1.41349030e+00
8.56380403e-01 9.33265984e-01 3.27085733e-01 -8.58593345e-01
-1.88789442e-01 -1.42041326e-01 -3.92954141e-01 -1.18694484e-01
3.77364755e-01 -2.82429725e-01 -8.58906090e-01 1.59032869e+00
2.84463465e-01 -1.53402984e-01 3.52786511e-01 9.23326135e-01
1.41861916e+00 7.27105498e-01 3.67432117e-01 3.34980577e-01
1.58846533e+00 -1.30992019e+00 -5.77688396e-01 -4.87142801e-01
5.13573229e-01 -9.34490144e-01 1.56489098e+00 5.15475333e-01
-1.38736880e+00 -2.93024540e-01 -7.48153508e-01 -5.11668026e-01
-7.97538236e-02 1.53609350e-01 8.05669129e-01 3.12866390e-01
-1.34031248e+00 -2.42175221e-01 -4.22090471e-01 -2.97349632e-01
4.58182752e-01 9.25507769e-03 -3.71975042e-02 -5.63398480e-01
-1.07459855e+00 8.22243154e-01 1.08588196e-01 1.54372275e-01
-1.07962489e+00 -7.84540653e-01 -7.58446276e-01 -4.86279353e-02
6.40565574e-01 -1.49472177e+00 1.81911767e+00 -4.68632281e-01
-1.53127956e+00 8.28843534e-01 -5.39095640e-01 -3.76255572e-01
8.29711497e-01 -7.15672433e-01 -2.54657030e-01 4.21322525e-01
1.95309788e-01 1.22840393e+00 7.35653520e-01 -1.34317315e+00
-2.90604651e-01 -3.42208482e-02 6.20392561e-01 1.75932914e-01
1.89400285e-01 3.61336349e-03 -7.14921176e-01 -4.19938341e-02
-1.67021155e-01 -7.94563711e-01 -1.92581788e-02 -1.30847916e-01
-6.49223506e-01 -4.53004420e-01 2.30555907e-01 -7.19772637e-01
1.22133589e+00 -1.56503296e+00 2.28394106e-01 -2.16978401e-01
5.44331729e-01 7.80258253e-02 -3.23217779e-01 8.24416220e-01
2.99308687e-01 3.22641224e-01 -3.51158530e-02 -6.36051178e-01
4.34804112e-01 1.12668820e-01 -7.40485728e-01 -2.18696296e-01
4.16872770e-01 1.46546590e+00 -8.85930359e-01 -7.56261587e-01
2.37582289e-02 3.44084859e-01 -7.26571679e-01 3.21614206e-01
-8.99396837e-01 3.62713277e-01 -5.26647627e-01 8.48333478e-01
3.45078111e-01 -8.68647516e-01 8.09080154e-02 -1.80266425e-01
6.59157988e-03 2.99555957e-01 -5.76106906e-01 1.97515357e+00
-4.90264565e-01 5.65627337e-01 -9.69890580e-02 5.90895377e-02
4.73693728e-01 2.74732351e-01 -2.64932245e-01 -6.95865214e-01
7.95192495e-02 2.31935054e-01 -3.55718322e-02 -9.10339653e-01
4.78773803e-01 1.20018655e-02 -1.96919113e-01 4.58813787e-01
-1.96731403e-01 -3.95536304e-01 3.21356148e-01 8.41398060e-01
1.27216709e+00 2.50741631e-01 -4.96171601e-02 2.61191845e-01
6.79804087e-01 3.60375136e-01 8.80518705e-02 1.02491105e+00
1.30104020e-01 4.86004740e-01 6.70993924e-01 -1.66636094e-01
-6.20498180e-01 -1.22272539e+00 5.92822671e-01 9.79355991e-01
-5.25045507e-02 -8.48821104e-01 -6.60621822e-01 -5.84202826e-01
1.60466969e-01 1.43646646e+00 -4.74841177e-01 9.15300474e-02
-2.95613617e-01 -3.36258590e-01 9.67732072e-01 6.46208167e-01
6.34103298e-01 -9.05182064e-01 -7.19011188e-01 -1.38533786e-01
-6.40586317e-01 -1.22526443e+00 -1.64922550e-01 -5.50832570e-01
-5.82270801e-01 -9.13451314e-01 -5.33399522e-01 8.66407603e-02
4.68149811e-01 5.28562851e-02 1.60933626e+00 3.39248508e-01
8.50173384e-02 9.33332026e-01 -1.29802316e-01 -3.75333637e-01
-4.70196724e-01 -1.88258424e-01 -4.01614308e-01 -1.53831914e-01
2.81916618e-01 -7.54620805e-02 -7.66456246e-01 -9.01642218e-02
-8.05491924e-01 8.11325610e-01 8.78261030e-01 4.28552687e-01
3.08791697e-01 -7.68661678e-01 5.06113350e-01 -6.79692030e-01
8.74550700e-01 -6.48550987e-01 -2.92379677e-01 7.46092439e-01
-6.32024527e-01 3.53783697e-01 1.19761512e-01 -1.16701275e-01
-1.61491919e+00 -3.87367904e-01 -1.05532348e-01 -1.82907119e-01
-8.10887963e-02 7.41836786e-01 1.82411686e-01 4.60191101e-01
4.86587197e-01 -1.19375177e-01 -2.40504026e-01 -2.36362919e-01
9.23324585e-01 4.22198474e-01 6.04039609e-01 -1.14323831e+00
5.17462373e-01 4.36170131e-01 -1.59123480e-01 -1.89387530e-01
-1.09435356e+00 -2.56987184e-01 -9.34957042e-02 -4.53866869e-01
1.08731246e+00 -1.08848357e+00 -1.14109361e+00 1.72567517e-01
-1.54771888e+00 -5.31685650e-01 2.75372177e-01 2.25292102e-01
-4.58252430e-01 2.16074020e-01 -9.13742006e-01 -1.05495656e+00
-7.23767817e-01 -1.06714177e+00 1.11758280e+00 5.10322988e-01
-6.48175657e-01 -8.45943570e-01 -2.37279311e-02 1.29278922e+00
4.53264356e-01 5.90336099e-02 1.14739978e+00 -3.67153257e-01
-9.67221081e-01 5.65501302e-02 -5.25923967e-01 -2.70300001e-01
-4.40525025e-01 2.10451439e-01 -1.16141474e+00 3.23126286e-01
-3.55399162e-01 -7.41142631e-01 9.83919203e-01 4.23990935e-02
8.34006250e-01 -3.70876908e-01 -1.07464686e-01 -1.43861040e-01
1.27220953e+00 -1.98559314e-01 5.88495255e-01 7.71771595e-02
7.14002430e-01 5.66615283e-01 5.01476586e-01 3.22880894e-01
1.07671750e+00 1.08974881e-01 5.69812655e-01 2.50961315e-02
-4.20561612e-01 -5.21236062e-01 6.04291141e-01 6.00548804e-01
1.83519311e-02 -5.07455647e-01 -1.47470176e+00 4.51775283e-01
-2.31694674e+00 -7.32693493e-01 -6.47393525e-01 1.64772820e+00
1.06619024e+00 1.21207707e-01 -2.63500154e-01 -4.98449832e-01
2.50489146e-01 1.44391870e-02 -6.84663236e-01 -5.47889233e-01
-7.81172663e-02 -1.80041313e-01 -1.26728788e-01 7.88675845e-01
-4.50913191e-01 9.69807565e-01 5.70657635e+00 6.16191149e-01
-6.58658326e-01 1.87827215e-01 6.45952106e-01 -2.50527769e-01
-1.05400944e+00 2.57021695e-01 -7.01895714e-01 -6.29650280e-02
9.77069199e-01 2.37589091e-01 3.79233778e-01 3.22947979e-01
7.05804974e-02 -6.81652427e-01 -1.26689708e+00 1.02205241e+00
2.30369568e-01 -1.68869197e+00 4.20810044e-01 -2.68397540e-01
7.65728891e-01 4.48910259e-02 1.75599322e-01 5.19166112e-01
6.36649311e-01 -1.21504378e+00 1.14479399e+00 1.04870737e+00
5.01802385e-01 -3.87298584e-01 6.54358804e-01 3.96859318e-01
-9.05125558e-01 -8.76194388e-02 2.01989576e-01 -1.77765623e-01
3.25481564e-01 6.03127718e-01 -1.18175781e+00 6.63755655e-01
5.88281155e-01 2.60932386e-01 -1.01810932e+00 5.27312696e-01
-7.74407268e-01 7.49117494e-01 -8.41837972e-02 -1.53887466e-01
3.66387904e-01 2.70460874e-01 3.18857759e-01 1.18275666e+00
-5.16589247e-02 2.30211616e-01 -5.92447408e-02 1.30876911e+00
-3.35539877e-01 -2.33758032e-01 -2.86985099e-01 -5.53728521e-01
3.42673600e-01 1.35393560e+00 -4.23209012e-01 -6.22192800e-01
-5.46847939e-01 8.69944274e-01 3.94261807e-01 5.31331241e-01
-1.11669791e+00 5.46517186e-02 2.81307787e-01 -3.07872057e-01
-1.68155804e-01 -2.07547098e-01 -5.69441378e-01 -1.29876995e+00
-5.05840667e-02 -1.04910183e+00 6.79520309e-01 -1.68846166e+00
-1.33822167e+00 4.88440305e-01 1.11086607e-01 -3.89428765e-01
-5.12880743e-01 -6.50063753e-01 -4.51080799e-01 9.10475254e-01
-1.53277028e+00 -1.58814216e+00 -3.38869840e-01 4.31943625e-01
6.38656795e-01 1.49268240e-01 7.86858201e-01 -1.17296197e-01
-4.79530305e-01 1.89005375e-01 -8.14594865e-01 -2.94505116e-02
8.24038684e-01 -1.24918544e+00 3.74089241e-01 8.44518125e-01
8.96373540e-02 1.00249648e+00 7.57515728e-01 -8.64918351e-01
-1.88218486e+00 -5.96573830e-01 1.10741806e+00 -1.28094268e+00
8.33039463e-01 -2.55052179e-01 -6.17838562e-01 7.58634686e-01
8.95587027e-01 -5.60921669e-01 7.77993679e-01 1.04276240e-01
-8.81615222e-01 1.59018636e-01 -9.48096216e-01 1.16016757e+00
7.74383545e-01 -6.73545897e-01 -8.71789098e-01 1.11425593e-01
8.82814467e-01 -4.07851696e-01 -6.69095576e-01 3.12865108e-01
6.88551307e-01 -9.32830751e-01 9.47227776e-01 -7.35367358e-01
1.17058384e+00 -5.23752630e-01 -4.81918871e-01 -6.97860003e-01
6.20876588e-02 -7.27783680e-01 -5.39297998e-01 1.23675227e+00
7.62229979e-01 -5.37073553e-01 2.72139698e-01 1.15312934e+00
1.02290034e-01 -9.52087700e-01 -4.38130885e-01 -2.62845252e-02
-1.18455596e-01 -7.02567637e-01 6.34985209e-01 5.22425115e-01
-5.93114495e-02 7.29544342e-01 -2.00374736e-04 1.64797530e-01
5.82842231e-01 3.42987657e-01 8.73600304e-01 -7.97574878e-01
-3.45491022e-01 -5.80118001e-01 5.86823821e-01 -1.05072129e+00
1.94117948e-01 -7.61874795e-01 -6.71111466e-03 -2.23499870e+00
4.31212008e-01 3.40029858e-02 -3.06515563e-02 8.23285639e-01
-4.17545348e-01 9.52483714e-02 5.07418752e-01 1.33178785e-01
-1.04955184e+00 4.28519368e-01 1.16908777e+00 -1.66752592e-01
2.54209578e-01 -5.29219627e-01 -1.02669621e+00 8.05734634e-01
8.79788697e-01 -2.53527045e-01 -6.01214886e-01 -9.12863970e-01
1.06893194e+00 4.20221090e-01 7.48235166e-01 -5.37792861e-01
5.03093779e-01 -2.82592833e-01 3.39704633e-01 -7.58746266e-01
6.21188939e-01 -2.08608106e-01 4.35821302e-02 3.82054150e-01
-8.82671535e-01 4.14184421e-01 4.05715138e-01 5.36134899e-01
3.26486304e-02 -3.96426618e-02 -1.21381275e-01 -2.75397539e-01
-8.07318449e-01 -2.81519026e-01 -3.40990901e-01 1.98398575e-01
4.50051367e-01 3.59290242e-01 -1.05511355e+00 -7.18602717e-01
-1.65863663e-01 7.08346188e-01 1.49910018e-01 5.02713978e-01
9.04732943e-01 -1.11753857e+00 -9.97692466e-01 -4.92477506e-01
3.84265244e-01 -9.47902203e-02 3.92306775e-01 1.22602260e+00
-4.33705449e-01 6.76769853e-01 2.86285818e-01 -6.71586275e-01
-1.10278916e+00 2.30643898e-01 1.15874052e-01 -2.49788955e-01
-2.16596410e-01 1.15893161e+00 7.64653981e-02 -4.55790043e-01
-8.11657868e-03 -3.92302424e-01 1.29084410e-02 -1.42468825e-01
6.21937931e-01 2.77796626e-01 -2.08474934e-01 -1.11755252e-01
-5.79982519e-01 3.90252501e-01 5.29198833e-02 -6.98454618e-01
8.52772772e-01 -1.26178503e-01 -3.75889033e-01 5.71229100e-01
6.19670391e-01 2.88858950e-01 -1.12297928e+00 -1.32560551e-01
1.26469642e-01 -2.91730702e-01 -1.76383451e-01 -1.58765781e+00
-5.03438771e-01 1.05823195e+00 -1.36800185e-01 -1.66734412e-01
8.82042766e-01 3.60220134e-01 6.74950063e-01 5.67804515e-01
2.00861618e-01 -6.06379330e-01 5.92362285e-01 5.81987739e-01
1.34571576e+00 -1.39593375e+00 -1.49079397e-01 -1.32552713e-01
-1.15495086e+00 1.18069375e+00 7.78275073e-01 3.31460357e-01
9.48533323e-03 2.47132778e-01 3.23493510e-01 -5.73478401e-01
-1.53440189e+00 -5.86737432e-02 5.29589057e-01 -3.80444643e-03
6.81103766e-01 1.98497191e-01 2.91057676e-02 7.68112719e-01
2.63927504e-02 2.18361810e-01 4.42934781e-01 8.98348510e-01
-1.61432818e-01 -7.74401784e-01 -4.75659162e-01 1.14437953e-01
-3.16328973e-01 -5.29422998e-01 -8.06856036e-01 4.84261721e-01
-2.14850456e-01 1.62378645e+00 -3.18587661e-01 -2.56814033e-01
6.52664751e-02 3.67264271e-01 6.13100410e-01 -5.26459932e-01
-7.45524883e-01 -1.75225616e-01 7.18064427e-01 -8.12092185e-01
-4.47540373e-01 -3.77409667e-01 -1.77887011e+00 -5.45570791e-01
2.95213640e-01 1.83942448e-02 5.15093863e-01 9.71410036e-01
6.74509346e-01 4.88932282e-01 -2.90663064e-01 -4.08435464e-01
-3.72049242e-01 -7.22927570e-01 4.21788931e-01 4.56514210e-02
3.02010119e-01 -2.47824654e-01 -9.42195728e-02 1.22116797e-01]
|
[10.847780227661133, 1.8965524435043335]
|
1e4392ba-c457-427a-ba5c-681416478861
|
mixed-norm-regularization-for-brain-decoding
|
1403.3628
| null |
http://arxiv.org/abs/1403.3628v1
|
http://arxiv.org/pdf/1403.3628v1.pdf
|
Mixed-norm Regularization for Brain Decoding
|
This work investigates the use of mixed-norm regularization for sensor
selection in Event-Related Potential (ERP) based Brain-Computer Interfaces
(BCI). The classification problem is cast as a discriminative optimization
framework where sensor selection is induced through the use of mixed-norms.
This framework is extended to the multi-task learning situation where several
similar classification tasks related to different subjects are learned
simultaneously. In this case, multi-task learning helps in leveraging data
scarcity issue yielding to more robust classifiers. For this purpose, we have
introduced a regularizer that induces both sensor selection and classifier
similarities. The different regularization approaches are compared on three ERP
datasets showing the interest of mixed-norm regularization in terms of sensor
selection. The multi-task approaches are evaluated when a small number of
learning examples are available yielding to significant performance
improvements especially for subjects performing poorly.
|
['Rémi Flamary', 'Marco Congedo', 'Ronald Phlypo', 'Nisrine Jrad', 'Alain Rakotomamonjy']
|
2014-03-14
| null | null | null | null |
['brain-decoding', 'brain-decoding']
|
['medical', 'miscellaneous']
|
[ 6.00804031e-01 -6.14603721e-02 6.35234118e-02 -5.56689382e-01
-1.15386534e+00 -1.30135134e-01 4.33241874e-01 2.46316880e-01
-7.84950197e-01 1.06181777e+00 1.60015464e-01 4.05503660e-01
-6.99862361e-01 -9.94031355e-02 -4.36899990e-01 -8.50206316e-01
-1.87328428e-01 -3.51346955e-02 -2.25369707e-01 -1.24532111e-01
3.90823901e-01 4.85359758e-01 -1.63903129e+00 2.82303870e-01
9.58323181e-01 9.28194165e-01 4.15836543e-01 1.34143338e-01
3.59338224e-01 2.15069186e-02 -5.43742418e-01 1.99893229e-02
2.17540756e-01 -1.45116299e-01 -3.74711156e-01 -1.38836235e-01
3.93489867e-01 4.85011011e-01 3.58554870e-01 1.00833416e+00
1.02701068e+00 3.94931227e-01 7.64406681e-01 -1.17396724e+00
-6.24322072e-02 1.80483669e-01 -6.13202929e-01 3.73185217e-01
5.32132506e-01 -3.83200228e-01 8.97839308e-01 -1.00239623e+00
2.21391290e-01 9.55660999e-01 5.95000267e-01 3.50481033e-01
-1.63272500e+00 -6.46854341e-01 8.59024003e-02 4.03367609e-01
-1.35935724e+00 -4.72954541e-01 1.08940721e+00 -5.17921150e-01
8.76417458e-01 7.94309303e-02 3.88438582e-01 1.35864806e+00
4.18217182e-01 6.96094990e-01 1.44025195e+00 -3.80615771e-01
4.50879902e-01 4.54566658e-01 6.39331400e-01 -5.87664992e-02
4.64750379e-01 6.01820974e-03 -8.32845688e-01 -2.07671806e-01
2.31767133e-01 -3.00931752e-01 -2.93877453e-01 -2.26710543e-01
-8.61524999e-01 7.71227717e-01 -6.19150028e-02 7.92206228e-01
-9.88111377e-01 -3.03344697e-01 7.18742251e-01 3.99532557e-01
6.27980053e-01 5.26630998e-01 -4.71409321e-01 -3.11595742e-02
-1.06072390e+00 2.99567878e-01 5.41057467e-01 6.21657550e-01
7.56204009e-01 3.51283193e-01 -3.56147766e-01 1.10546160e+00
1.88211635e-01 2.11273238e-01 7.17676103e-01 -4.23135936e-01
5.30037642e-01 4.20146793e-01 -5.13842963e-02 -9.33367372e-01
-7.14964688e-01 -3.56939971e-01 -6.33727729e-01 3.89213949e-01
2.12099493e-01 -5.38799822e-01 -5.32361448e-01 1.87628174e+00
8.42126738e-03 3.35902870e-01 2.59832740e-01 7.37796426e-01
4.70369667e-01 2.65952229e-01 3.25812280e-01 -5.99341869e-01
1.34924912e+00 -2.56400168e-01 -9.83673811e-01 -3.78407449e-01
4.69923139e-01 -4.13227826e-01 9.78381872e-01 7.58449078e-01
-8.69354069e-01 -5.65950751e-01 -1.20421922e+00 4.73514587e-01
-5.95512271e-01 2.07481161e-01 6.23487353e-01 7.26052046e-01
-7.72392213e-01 4.03587043e-01 -3.84579659e-01 -1.82282850e-01
4.73998219e-01 5.75528085e-01 -4.81625378e-01 1.44465849e-01
-1.09884548e+00 1.17311883e+00 5.51793337e-01 6.48415014e-02
-4.34616804e-01 -5.69080889e-01 -6.08100832e-01 -2.00184152e-01
2.52739787e-01 -1.84499562e-01 4.75913018e-01 -1.23959053e+00
-1.30606472e+00 8.81025553e-01 -2.11373135e-01 -2.87803620e-01
2.40335256e-01 -2.66746163e-01 -5.03818631e-01 -9.16083679e-02
1.09326802e-01 1.76951900e-01 1.19306386e+00 -1.14005387e+00
-3.71517062e-01 -8.18313062e-01 -4.31104481e-01 2.28799745e-01
-4.52218473e-01 2.96540320e-01 2.68247783e-01 -8.03794444e-01
-8.41181204e-02 -6.62966609e-01 -9.87088084e-02 -6.44683897e-01
-9.39690247e-02 -5.37857831e-01 7.62160480e-01 -7.28370249e-01
1.00485575e+00 -2.06350112e+00 5.32866836e-01 5.06902575e-01
-1.29172370e-01 9.69412178e-02 -2.71670282e-01 -2.29835156e-02
-5.00470638e-01 -2.06648707e-01 -3.16932380e-01 -2.57700175e-01
-1.90341666e-01 7.84595013e-02 3.20215195e-01 5.76357245e-01
1.57713845e-01 4.35476035e-01 -4.15450066e-01 -2.81546712e-01
6.96212500e-02 4.76624638e-01 -1.96895882e-01 2.90451407e-01
2.74441272e-01 6.69950366e-01 -4.38128859e-01 5.22789359e-01
6.35827482e-01 4.11695331e-01 9.23294388e-03 -4.49584424e-01
-5.99682629e-02 -3.42146903e-02 -1.64648449e+00 1.89049721e+00
-5.47117710e-01 5.57212830e-01 3.46514434e-01 -1.66334975e+00
1.00069308e+00 6.73262596e-01 8.58996093e-01 -6.71405852e-01
3.28196287e-01 3.87132972e-01 2.03148514e-01 -6.97215497e-01
1.23833731e-01 -1.40779927e-01 -2.57517323e-02 3.39240342e-01
2.75403768e-01 2.70746723e-02 3.29868227e-01 -3.34006548e-01
7.72207499e-01 1.24908917e-01 5.79266250e-01 -7.26957202e-01
6.14598572e-01 -4.65465873e-01 6.46670878e-01 6.38600588e-01
-7.58690462e-02 1.84693694e-01 2.39036709e-01 -6.18138127e-02
-5.15772045e-01 -8.49788189e-01 -5.67854583e-01 1.21341705e+00
-2.40736350e-01 -8.00357088e-02 -4.09584492e-01 -4.02248353e-01
2.06056647e-02 7.58201003e-01 -5.27097642e-01 -2.27558598e-01
-4.50767130e-01 -1.14740491e+00 2.72493362e-01 3.97376269e-01
2.36336112e-01 -1.28232789e+00 -9.74592090e-01 4.25104946e-01
3.73804756e-02 -9.61371541e-01 2.37342902e-02 8.35313737e-01
-1.06857860e+00 -1.00489140e+00 -7.57876754e-01 -7.43219793e-01
3.59764278e-01 -7.66867027e-02 6.41004860e-01 -6.00939751e-01
-3.85318160e-01 8.20147395e-01 -3.48980069e-01 -7.92812943e-01
7.65034854e-02 -1.40688673e-01 3.34012538e-01 3.51497233e-01
5.88966370e-01 -9.20789838e-01 -1.86063915e-01 8.78578238e-03
-8.90514910e-01 -4.96570647e-01 5.80922961e-01 9.59118307e-01
4.97273445e-01 -1.79818496e-01 1.26430452e+00 -6.16932273e-01
1.34230697e+00 -6.00734711e-01 -3.34694803e-01 5.42795122e-01
-7.16364026e-01 1.01793647e-01 4.01637644e-01 -8.74052167e-01
-1.16774654e+00 6.31229803e-02 1.36063829e-01 -8.41613207e-03
-3.23238790e-01 6.12665176e-01 -2.68062115e-01 -5.45523226e-01
7.65001714e-01 5.72133549e-02 -1.64501116e-01 -5.17536521e-01
2.00339537e-02 7.89463341e-01 8.24711695e-02 -6.74721599e-01
1.43251166e-01 -1.32684514e-01 9.06392336e-02 -1.06461084e+00
-5.44575930e-01 -4.90622282e-01 -7.41140783e-01 -3.47164661e-01
7.56750822e-01 -8.63662302e-01 -5.73123336e-01 4.10654038e-01
-1.16249919e+00 9.64808986e-02 -1.53563023e-01 9.18612897e-01
-6.78726375e-01 1.21006660e-01 2.95627546e-02 -1.00951707e+00
-4.32516545e-01 -1.26320589e+00 8.21966171e-01 2.28682175e-01
-2.94821084e-01 -9.61783767e-01 3.56856324e-02 8.75893384e-02
3.46713185e-01 2.92355180e-01 8.36886942e-01 -1.21842480e+00
2.45455846e-01 -3.17944765e-01 -7.03622326e-02 5.08057952e-01
2.10018530e-01 -7.18366683e-01 -1.16701567e+00 -3.86952370e-01
6.05474055e-01 -5.00675559e-01 6.24794245e-01 5.75037122e-01
1.06834388e+00 9.88856331e-02 -1.41455293e-01 2.78340876e-01
1.53170669e+00 3.00577521e-01 5.15291035e-01 2.18983904e-01
4.24198896e-01 8.85039032e-01 5.70502639e-01 6.92628384e-01
-2.08462179e-02 8.10293972e-01 5.81398793e-02 1.87880650e-01
2.75705844e-01 5.24030566e-01 2.63246149e-01 5.86774707e-01
-2.35434011e-01 1.66208625e-01 -7.27616131e-01 4.91166592e-01
-1.81264901e+00 -7.57834435e-01 -1.18936814e-01 2.37175989e+00
6.91819906e-01 -2.90237032e-02 7.93721825e-02 3.44642073e-01
7.19387949e-01 8.56828392e-02 -6.77557945e-01 -5.26219249e-01
-3.05980712e-01 6.78455770e-01 3.49881172e-01 2.26152569e-01
-1.00797796e+00 3.60852808e-01 5.94496107e+00 8.64227116e-01
-1.08844543e+00 4.58920628e-01 1.91673920e-01 -1.08998060e-01
8.22756365e-02 -3.57985824e-01 -7.06662893e-01 5.18803120e-01
9.59166765e-01 -2.51974553e-01 3.62542093e-01 5.56700230e-01
6.33938193e-01 -5.79552293e-01 -1.02354980e+00 1.62026036e+00
3.82160664e-01 -7.61770785e-01 -4.68178719e-01 -1.77747428e-01
4.67138320e-01 -8.40133056e-02 -1.97304174e-01 1.70024797e-01
-5.24916172e-01 -7.91616797e-01 4.69106585e-01 7.06429362e-01
4.94656116e-01 -4.92023796e-01 6.33610308e-01 1.50102302e-01
-1.11220264e+00 -4.32481319e-01 -1.26626953e-01 1.22851029e-01
1.96774304e-01 5.88399351e-01 -3.27479988e-01 7.37173617e-01
5.70121229e-01 9.41643238e-01 -4.84614909e-01 1.18992019e+00
-1.23694748e-01 4.11647826e-01 -3.32676351e-01 -9.85769108e-02
-1.07486889e-01 -4.95054066e-01 8.79887640e-01 1.28858435e+00
3.24455321e-01 6.34923205e-02 1.13031365e-01 6.97187543e-01
2.80822545e-01 5.28149188e-01 -8.26197684e-01 3.05723518e-01
2.78788745e-01 1.29513574e+00 -3.64673376e-01 3.79126482e-02
-5.03761590e-01 1.05784142e+00 2.75871396e-01 5.03657997e-01
-4.82808203e-01 -4.33366656e-01 2.19879374e-01 -2.20519081e-01
2.67192665e-02 -1.85352311e-01 -7.21553326e-01 -9.92039561e-01
3.58595103e-01 -8.64126205e-01 4.14987832e-01 -4.20842856e-01
-1.35382271e+00 5.18086374e-01 4.48789269e-01 -9.40969348e-01
-2.13072509e-01 -7.30382442e-01 -7.72357225e-01 9.84276652e-01
-1.52797496e+00 -8.97846937e-01 -1.17145084e-01 8.78216684e-01
5.72489083e-01 -6.17103636e-01 9.22824442e-01 6.09666944e-01
-6.63159847e-01 5.79311371e-01 8.64389911e-02 -5.98053932e-01
7.96836972e-01 -1.01996398e+00 -8.32590401e-01 6.59129262e-01
5.87017508e-03 3.73123050e-01 6.89934075e-01 -4.85409915e-01
-1.05581689e+00 -6.66911364e-01 7.53436267e-01 1.94609523e-01
4.14509833e-01 -2.71254569e-01 -1.04856372e+00 2.57220387e-01
2.50161141e-01 -1.26332238e-01 8.61046851e-01 3.04713428e-01
3.33813652e-02 -5.08255720e-01 -1.34929883e+00 3.16840202e-01
6.27862871e-01 -5.82754850e-01 -7.68895626e-01 3.67072672e-01
-2.03276232e-01 3.11990619e-01 -1.09984374e+00 4.37123001e-01
4.03952360e-01 -5.40915608e-01 7.70849049e-01 -6.95529699e-01
-3.25737804e-01 2.26506010e-01 -2.25994930e-01 -1.67968535e+00
-4.93151583e-02 -5.46630621e-01 1.68729231e-01 1.13690412e+00
4.22431976e-01 -7.54839540e-01 4.13915396e-01 7.87508488e-01
-3.35815959e-02 -7.60009944e-01 -1.29033208e+00 -7.54959881e-01
-4.33197320e-02 -5.43481708e-01 -1.55084774e-01 7.39419699e-01
2.05680579e-01 3.63966793e-01 -5.55741489e-01 -2.37410769e-01
7.08881736e-01 -3.30283821e-01 1.03633054e-01 -1.61390340e+00
-2.77283601e-02 -2.86612272e-01 -5.06602526e-01 -3.99411649e-01
6.13927364e-01 -1.02684331e+00 1.87307343e-01 -1.08326101e+00
9.26183984e-02 -2.70662844e-01 -6.87853813e-01 5.09065628e-01
-2.78445989e-01 -8.34902599e-02 -1.98577787e-03 -3.53014539e-03
-3.20408165e-01 5.15204370e-01 5.59305847e-01 -2.52424870e-02
-6.15777373e-01 3.55944484e-01 -6.17565036e-01 6.18064106e-01
1.16220140e+00 -6.51168466e-01 -3.87192219e-01 -4.84948978e-02
-5.87441474e-02 4.71533574e-02 2.21386060e-01 -1.28090024e+00
1.70053199e-01 1.67102411e-01 3.25355530e-01 -2.26838589e-01
5.76922357e-01 -9.82427955e-01 -7.88834989e-02 8.53716955e-02
-4.72884029e-01 5.88441826e-02 3.67525727e-01 5.60794652e-01
-2.62069285e-01 -5.85128367e-01 9.12868679e-01 -1.87939964e-02
-8.65746677e-01 -6.86009526e-02 -5.63592494e-01 1.12682834e-01
1.25541699e+00 -6.31955743e-01 5.21953940e-01 -3.07071675e-02
-1.05724025e+00 -5.07239848e-02 -5.41919291e-01 3.76446992e-01
8.75372887e-01 -1.20579934e+00 -8.02802563e-01 1.38265237e-01
1.52295917e-01 -6.08382106e-01 4.34909128e-02 1.19234931e+00
6.23346090e-01 2.88782209e-01 -6.50098145e-01 -4.32620138e-01
-1.56397676e+00 5.84743060e-02 4.06646013e-01 -1.39660925e-01
-3.00885379e-01 6.68483377e-01 -3.45328718e-01 -5.77678680e-02
5.07502317e-01 8.87246951e-02 -7.22351134e-01 8.83332372e-01
3.02257210e-01 6.40176952e-01 3.01419526e-01 -7.27600753e-01
-5.33864617e-01 5.15156031e-01 -6.53391005e-03 -2.30974346e-01
1.75862813e+00 3.85249257e-02 -1.49203137e-01 6.37369156e-01
1.28828955e+00 -2.43027940e-01 -9.70486760e-01 -1.87782884e-01
6.95500016e-01 -2.27035388e-01 2.84762412e-01 -7.39393532e-01
-8.18918347e-01 7.49793172e-01 1.16918945e+00 -1.99977726e-01
1.30148530e+00 -2.83720106e-01 1.56634137e-01 4.77697402e-01
4.52848464e-01 -1.60915804e+00 1.63019329e-01 2.57196248e-01
1.11029768e+00 -1.35465980e+00 1.00221001e-01 3.12253740e-03
-7.99965501e-01 1.28943384e+00 4.68132496e-01 -2.54252553e-01
8.20064723e-01 1.42114490e-01 -3.45551550e-01 -2.91382879e-01
-2.21101865e-01 -3.14317644e-01 6.11550093e-01 7.24744916e-01
6.11387432e-01 4.08790819e-02 -1.08123159e+00 1.05443680e+00
4.88247603e-01 8.37721974e-02 3.03146273e-01 8.96476984e-01
-2.46246383e-01 -1.21797037e+00 -3.57798755e-01 8.23854327e-01
-6.08208299e-01 8.23872804e-04 1.76377594e-02 5.22437572e-01
1.77302063e-01 1.23596692e+00 -4.28352177e-01 -1.79661408e-01
6.60606802e-01 4.25812393e-01 5.46134293e-01 -6.40950680e-01
-9.68230844e-01 1.56943589e-01 1.69081345e-01 -5.59413612e-01
-8.41767788e-01 -8.69389594e-01 -9.00099576e-01 6.82730556e-01
-3.80537570e-01 8.88610780e-02 9.82780218e-01 1.08864975e+00
2.74181664e-01 5.69421828e-01 6.44291759e-01 -1.07558453e+00
-7.19069839e-01 -1.09368575e+00 -9.77239311e-01 5.77001393e-01
8.56790915e-02 -1.00192273e+00 -3.02131534e-01 -1.64690271e-01]
|
[13.045869827270508, 3.4375088214874268]
|
66770c1e-3387-444d-88bd-fefd1586fbdb
|
key-information-extraction-in-purchase
|
2210.03453
| null |
https://arxiv.org/abs/2210.03453v1
|
https://arxiv.org/pdf/2210.03453v1.pdf
|
Key Information Extraction in Purchase Documents using Deep Learning and Rule-based Corrections
|
Deep Learning (DL) is dominating the fields of Natural Language Processing (NLP) and Computer Vision (CV) in the recent times. However, DL commonly relies on the availability of large data annotations, so other alternative or complementary pattern-based techniques can help to improve results. In this paper, we build upon Key Information Extraction (KIE) in purchase documents using both DL and rule-based corrections. Our system initially trusts on Optical Character Recognition (OCR) and text understanding based on entity tagging to identify purchase facts of interest (e.g., product codes, descriptions, quantities, or prices). These facts are then linked to a same product group, which is recognized by means of line detection and some grouping heuristics. Once these DL approaches are processed, we contribute several mechanisms consisting of rule-based corrections for improving the baseline DL predictions. We prove the enhancements provided by these rule-based corrections over the baseline DL results in the presented experiments for purchase documents from public and NielsenIQ datasets.
|
['Javier Lorenzo', 'Héctor Corrales', 'Elena Martínez', 'Javier Yebes', 'Roberto Arroyo']
|
2022-10-07
| null |
https://aclanthology.org/2022.pandl-1.2
|
https://aclanthology.org/2022.pandl-1.2.pdf
|
pandl-coling-2022-10
|
['line-detection', 'key-information-extraction']
|
['computer-vision', 'natural-language-processing']
|
[-6.00700732e-03 5.58077022e-02 -3.05023611e-01 -5.09796143e-01
-5.69007695e-01 -7.43493080e-01 8.33316565e-01 1.04867399e+00
-6.08135581e-01 6.44784808e-01 1.80065736e-01 -2.98300624e-01
-5.58690988e-02 -8.95313859e-01 -8.75053167e-01 -2.51516163e-01
3.17300111e-02 4.13362026e-01 1.88672051e-01 -1.92075055e-02
5.29232502e-01 7.59638965e-01 -1.54042864e+00 6.39871418e-01
8.70168924e-01 1.14275396e+00 2.53138959e-01 6.01489365e-01
-7.90171921e-01 8.51280749e-01 -8.57736096e-02 -5.73002696e-01
4.27542418e-01 9.10452083e-02 -6.55065656e-01 -1.08300708e-02
4.44760114e-01 -2.88636893e-01 6.64722994e-02 1.13148916e+00
1.80958018e-01 7.53638335e-03 7.86401987e-01 -1.10130191e+00
-8.22187245e-01 7.07925677e-01 -5.69842696e-01 2.51495535e-03
5.50687551e-01 -3.33592772e-01 1.26326036e+00 -1.07812107e+00
7.66519129e-01 1.06933546e+00 7.33812273e-01 6.90289140e-02
-9.98676956e-01 -2.26008847e-01 2.24928990e-01 4.00582582e-01
-1.31862748e+00 -1.99163571e-01 5.87879062e-01 -5.57262778e-01
1.16490352e+00 -2.24694200e-02 3.41215253e-01 6.26292825e-01
1.60966992e-01 1.15966916e+00 8.82376432e-01 -8.86860251e-01
3.39621484e-01 7.93784916e-01 4.14850116e-01 6.41756415e-01
5.54399014e-01 -2.28372276e-01 -3.48674387e-01 1.41198143e-01
5.37379563e-01 7.37482831e-02 -4.96577881e-02 -5.60096443e-01
-1.03742373e+00 9.36261773e-01 3.25655282e-01 4.78438586e-01
-7.81818628e-01 -3.80167574e-01 5.50952673e-01 -1.83018837e-02
4.45806950e-01 6.58412158e-01 -8.09604883e-01 1.49387449e-01
-1.10051000e+00 2.15506643e-01 1.07629156e+00 1.10704327e+00
8.72081578e-01 -4.59184468e-01 9.24856309e-03 7.39579678e-01
4.61638570e-01 2.74057090e-01 4.58736032e-01 -2.46657282e-01
7.06122160e-01 9.75563943e-01 2.53484964e-01 -1.39453602e+00
-5.82653761e-01 -1.90558121e-01 -6.00177944e-01 5.78591414e-02
2.57329077e-01 -5.19559234e-02 -9.46524799e-01 1.18291461e+00
6.79835454e-02 -1.89108104e-01 2.95950562e-01 4.99320835e-01
7.35562623e-01 8.33919585e-01 9.46708396e-02 -2.40793183e-01
1.37383723e+00 -8.54943097e-01 -8.29459548e-01 5.44231236e-02
9.29865003e-01 -8.78644526e-01 6.26031935e-01 6.66911125e-01
-5.83560467e-01 -7.05235958e-01 -9.74026084e-01 -2.72200108e-01
-1.08723617e+00 5.16732216e-01 6.68730497e-01 4.75038916e-01
-7.89120495e-01 6.44676387e-01 -4.64475363e-01 -7.74302602e-01
1.61823437e-01 3.33045959e-01 -4.32130635e-01 -3.06814141e-03
-1.03691268e+00 9.47332740e-01 7.65527666e-01 1.87773719e-01
-2.30446607e-01 -4.14374769e-01 -1.07673752e+00 4.89307614e-03
5.37845969e-01 -3.29542041e-01 1.10363460e+00 -9.38198745e-01
-1.08506894e+00 7.98127592e-01 9.10755023e-02 -9.29607153e-01
4.47508276e-01 -6.64717555e-01 -5.25074601e-01 1.09840199e-01
3.56990062e-02 7.45630383e-01 5.75696409e-01 -1.30308807e+00
-1.21036947e+00 -3.10567826e-01 8.02260917e-03 5.85371554e-02
-2.88119227e-01 9.62099805e-02 -5.12118936e-01 -4.67892855e-01
3.68670970e-02 -7.71524489e-01 -4.62973528e-02 -1.03460886e-01
-6.26469731e-01 -4.98997808e-01 4.17562515e-01 -1.03069854e+00
1.28554773e+00 -2.10212564e+00 -2.02645138e-01 3.14367473e-01
1.21139824e-01 4.01927114e-01 1.26928743e-02 4.78666604e-01
2.25816611e-02 1.87069282e-01 4.54491898e-02 -3.35042536e-01
2.86905289e-01 1.83153167e-01 -2.33423531e-01 1.08363092e-01
1.33408964e-01 7.79255509e-01 -6.09166622e-01 -5.82962096e-01
3.19879204e-01 6.68513179e-02 -2.84352541e-01 2.69251913e-02
-4.89806503e-01 -2.29769319e-01 -4.51725096e-01 6.24562204e-01
6.96951628e-01 1.17112786e-01 1.23765275e-01 -6.35863781e-01
-4.63099658e-01 2.12127760e-01 -1.46470046e+00 1.48248196e+00
-4.77627426e-01 6.38029873e-01 -3.68954390e-01 -1.04551220e+00
1.03439760e+00 4.45745476e-02 4.81564105e-01 -6.09054923e-01
1.99072003e-01 2.29723811e-01 -2.51850575e-01 -5.88491678e-01
8.40747595e-01 4.82534975e-01 -5.40743619e-02 2.08213687e-01
-6.18626550e-03 2.84054518e-01 4.59419638e-01 2.06120133e-01
7.32378960e-01 3.43459189e-01 6.21864617e-01 -1.76434293e-01
7.78842568e-01 2.52243519e-01 3.78891289e-01 7.23466814e-01
9.46880970e-03 2.63007998e-01 2.79223770e-01 -5.93719959e-01
-1.43399441e+00 -6.52221978e-01 -6.73447028e-02 8.59404147e-01
-6.01267666e-02 -4.97591347e-01 -6.46400452e-01 -8.81107926e-01
2.73946494e-01 8.69312525e-01 -5.62666297e-01 1.02257505e-01
-3.79243463e-01 -4.42296118e-01 3.92975897e-01 7.05491245e-01
6.03319824e-01 -1.06321418e+00 -3.71239752e-01 5.22006154e-01
2.51807630e-01 -1.26469243e+00 -1.22147277e-01 5.33649385e-01
-6.73305213e-01 -9.69523132e-01 -5.97258270e-01 -8.73187065e-01
6.21129334e-01 -4.79123406e-02 9.54759240e-01 -7.36724511e-02
-1.85879767e-01 2.61589527e-01 -7.05243528e-01 -6.12828135e-01
-5.20418227e-01 2.27525324e-01 2.06327349e-01 1.02356791e-01
8.88432980e-01 1.04327574e-01 -2.40784630e-01 -1.41764805e-01
-8.49501371e-01 7.63914362e-02 9.14761424e-01 4.23903286e-01
5.73847175e-01 2.35600024e-01 5.64538166e-02 -9.55842853e-01
6.78525448e-01 -7.33201951e-02 -7.05110192e-01 6.21849537e-01
-9.27702725e-01 4.53683943e-01 6.31947994e-01 -2.89536148e-01
-1.09525263e+00 5.02648771e-01 -1.51442230e-01 -3.40496711e-02
-6.69778883e-01 7.04761028e-01 -2.86946595e-01 1.43104196e-01
3.82134527e-01 2.04639584e-01 -5.55839419e-01 -8.42732728e-01
6.51521266e-01 8.44471514e-01 4.40675855e-01 -2.77527869e-01
5.95408916e-01 1.20689636e-02 -2.78853059e-01 -7.00950623e-01
-8.55619788e-01 -6.97693527e-01 -1.19682372e+00 1.32528603e-01
9.53691363e-01 -6.76432312e-01 -5.06767631e-01 4.06702012e-01
-1.37769508e+00 3.40316683e-01 -2.26107687e-01 6.91678345e-01
-1.44993901e-01 4.67532575e-01 -6.76036775e-01 -8.23457778e-01
-3.11761588e-01 -8.37588787e-01 9.20919955e-01 3.25876623e-01
-1.06447853e-01 -7.76493430e-01 6.60309568e-02 1.61332786e-01
5.76229207e-03 7.48611242e-02 1.12749445e+00 -1.29267848e+00
-3.78853977e-01 -5.09338915e-01 -4.48923677e-01 7.09986687e-01
1.92255512e-01 1.79006845e-01 -6.96504772e-01 1.47119388e-01
-3.20760995e-01 1.71507925e-01 8.51208985e-01 3.62867624e-01
6.77748859e-01 -3.35116148e-01 -5.89564681e-01 2.15419695e-01
1.66308784e+00 5.42367160e-01 6.57093227e-01 7.90365458e-01
8.40828836e-01 6.53178632e-01 7.83397615e-01 5.09184420e-01
5.06320775e-01 6.14117146e-01 2.43426174e-01 -1.13058031e-01
1.15748949e-01 -3.43001038e-01 2.99229234e-01 5.70235014e-01
9.94661171e-03 -3.69845301e-01 -9.02642965e-01 5.73256016e-01
-1.98567748e+00 -7.22015738e-01 -3.29943478e-01 2.08513689e+00
8.61468196e-01 3.73366237e-01 3.76228616e-02 1.60030812e-01
8.35273743e-01 -2.92757511e-01 -4.11639273e-01 -6.50784969e-01
-1.02295607e-01 -1.45995602e-01 1.01448894e+00 2.80471236e-01
-1.36024892e+00 1.13412356e+00 5.79803419e+00 6.63659096e-01
-8.96996021e-01 -2.21884266e-01 4.93637472e-01 4.12155628e-01
-2.09258288e-01 -1.34859458e-01 -1.44110179e+00 2.63852179e-01
7.12418497e-01 4.18814383e-02 1.38449758e-01 1.08246839e+00
8.63921549e-03 -3.61691803e-01 -1.32688785e+00 9.89234328e-01
2.57583827e-01 -1.36062407e+00 2.48109594e-01 1.47305146e-01
5.99905849e-01 -3.00261736e-01 -2.80815095e-01 3.68204951e-01
3.79103869e-01 -5.33920467e-01 8.23300660e-01 6.74326122e-01
1.37253121e-01 -8.76616657e-01 1.19319427e+00 3.15478891e-01
-1.16315746e+00 -1.04887374e-01 -5.14324844e-01 1.20946236e-01
2.40411498e-02 7.71659851e-01 -1.04391193e+00 7.78390765e-01
5.59126914e-01 6.56901956e-01 -6.03827298e-01 1.02403903e+00
-3.58894646e-01 1.93896160e-01 -3.97764295e-01 -3.53024811e-01
3.25590402e-01 -1.50052637e-01 9.84925255e-02 1.45107520e+00
2.94919461e-01 -9.48738009e-02 2.09901571e-01 7.44735897e-01
-8.40345770e-02 6.18863344e-01 -5.34827709e-01 -3.68327111e-01
2.47967482e-01 1.32381368e+00 -9.61488307e-01 -5.26590407e-01
-6.28483653e-01 1.05984902e+00 6.93652555e-02 4.99834865e-02
-6.22873664e-01 -4.77559417e-01 3.62803280e-01 -1.90143008e-02
6.35280371e-01 -3.60876948e-01 -3.01604509e-01 -1.02597010e+00
6.82141706e-02 -6.84094310e-01 2.59172797e-01 -7.00554430e-01
-1.22302580e+00 5.76027870e-01 -1.43925235e-01 -1.12345028e+00
-2.88182199e-01 -8.81992280e-01 1.11251019e-01 4.98915166e-01
-1.69041681e+00 -1.15551102e+00 -1.23516442e-02 4.10291761e-01
5.97108781e-01 -2.48900205e-01 6.88044429e-01 4.23116684e-01
-4.66526866e-01 3.89931262e-01 3.29592794e-01 4.01241452e-01
7.46129751e-01 -1.53150642e+00 3.15146714e-01 9.80617881e-01
5.64236045e-01 5.72786987e-01 7.72097945e-01 -8.30908775e-01
-1.11616874e+00 -8.76044452e-01 1.35164058e+00 -3.21251959e-01
6.79596841e-01 -4.69269812e-01 -7.36621678e-01 6.10615611e-01
2.29075417e-01 -3.30916077e-01 6.10383868e-01 7.50666037e-02
-4.00155902e-01 -1.72672078e-01 -1.10636055e+00 3.71828318e-01
5.83815098e-01 -3.42060238e-01 -7.87779391e-01 3.47993851e-01
7.14507759e-01 -1.44673377e-01 -7.39324450e-01 1.75160795e-01
4.17451680e-01 -6.08438849e-01 7.54910052e-01 -7.16453612e-01
3.22351903e-01 -4.00383979e-01 -2.42151603e-01 -1.12424898e+00
-2.36587524e-01 -1.02065571e-01 1.01788104e-01 1.62286663e+00
6.97338343e-01 -2.69956350e-01 5.85359037e-01 7.55137563e-01
-1.86935719e-02 -2.51061112e-01 -3.93325210e-01 -8.07329237e-01
-2.14402482e-01 -6.07675195e-01 5.56161344e-01 8.71896207e-01
-4.81221639e-02 2.14205906e-01 -3.12860161e-01 3.26907247e-01
2.17153206e-01 1.70641974e-01 6.72251821e-01 -1.58023071e+00
-8.64316747e-02 -2.63471454e-01 -5.91108143e-01 -9.86293554e-01
-7.17261583e-02 -7.39974678e-01 2.36907244e-01 -1.92676389e+00
5.97556448e-03 -2.98068762e-01 -4.11887676e-01 6.82100117e-01
2.37906843e-01 -1.10278092e-01 3.41262788e-01 1.19994819e-01
-6.86711013e-01 2.84626391e-02 5.89058578e-01 -1.74902007e-01
-3.27770323e-01 -1.82645276e-01 -5.80855191e-01 8.58970881e-01
7.69349694e-01 -3.90951872e-01 3.43287773e-02 -1.29054800e-01
6.24802768e-01 -4.85877901e-01 1.10830009e-01 -8.98247004e-01
4.39635456e-01 1.59003213e-01 5.53028882e-01 -9.05088067e-01
-2.07167685e-01 -1.20612645e+00 -2.90766746e-01 3.76737416e-01
-5.48859417e-01 2.02737093e-01 2.50951380e-01 4.81183618e-01
-1.76340446e-01 -4.52771872e-01 4.00362492e-01 -2.59683698e-01
-1.27732396e+00 -8.93287063e-02 -4.65189070e-01 -5.68015397e-01
9.84323859e-01 -3.66488218e-01 -1.20145984e-01 -1.87134951e-01
-7.16848493e-01 -3.70547585e-02 1.36749730e-01 5.47636390e-01
4.58523810e-01 -1.15348482e+00 -5.04668951e-01 1.03389516e-01
2.78623402e-01 -2.38628685e-01 -1.81168497e-01 7.73837984e-01
-7.25684285e-01 9.26604271e-01 -1.70859292e-01 -3.21933419e-01
-1.29142094e+00 9.80255604e-01 -4.26503532e-02 -4.42779839e-01
-4.58802253e-01 6.95508003e-01 -2.19119415e-01 -2.10840344e-01
5.18927395e-01 -8.66583705e-01 -7.46367574e-01 3.79006028e-01
5.01195788e-01 1.11163065e-01 2.90353060e-01 -5.09838879e-01
-4.08337086e-01 6.09528661e-01 -5.88822961e-01 1.17809713e-01
1.43715441e+00 -2.95204818e-01 -8.23735911e-03 2.80316055e-01
9.88293529e-01 1.28416777e-01 -8.85427654e-01 -4.06079203e-01
6.65574014e-01 -1.64118856e-01 1.79304332e-01 -9.73358631e-01
-8.10431182e-01 6.66462779e-01 7.38707006e-01 5.17193437e-01
1.01990902e+00 -1.51444762e-03 8.48381996e-01 5.92419088e-01
2.87255794e-01 -1.42556286e+00 -4.39278841e-01 4.27064002e-01
6.54202342e-01 -1.29239440e+00 7.16334954e-02 -4.94922936e-01
-6.47069097e-01 1.42472780e+00 2.27116629e-01 -4.62144576e-02
6.15500331e-01 4.02477562e-01 -2.61406094e-04 -1.31367892e-01
-4.20496225e-01 -6.09384000e-01 5.00682294e-01 5.28881550e-01
4.90050077e-01 -5.17634302e-02 -5.42612851e-01 6.67177618e-01
-4.47901785e-02 2.28083476e-01 2.88267255e-01 8.71846974e-01
-6.71422780e-01 -1.22574580e+00 -2.71097034e-01 3.77561599e-01
-3.70908588e-01 -4.97698098e-01 -5.51488578e-01 1.02063704e+00
5.88972092e-01 8.27677846e-01 1.41850546e-01 -3.20621967e-01
4.09582168e-01 1.84372529e-01 3.82738590e-01 -6.47175372e-01
-5.31843066e-01 9.56958905e-02 2.00666130e-01 -3.17209661e-01
-4.91157502e-01 -8.70557606e-01 -1.33018625e+00 -4.93318550e-02
-4.54366863e-01 2.15387437e-02 9.47534740e-01 1.24925005e+00
3.46725583e-01 1.78727373e-01 3.45744103e-01 -3.81612957e-01
-2.53183901e-01 -8.39178205e-01 -6.89267755e-01 4.33093607e-01
1.08956970e-01 -4.41833913e-01 -1.88781306e-01 3.93828601e-01]
|
[11.642945289611816, 2.931990623474121]
|
48416ca3-fd25-4627-90a5-46ddc304aa37
|
doubly-stochastic-matrix-models-for
|
2304.02458
| null |
https://arxiv.org/abs/2304.02458v1
|
https://arxiv.org/pdf/2304.02458v1.pdf
|
Doubly Stochastic Matrix Models for Estimation of Distribution Algorithms
|
Problems with solutions represented by permutations are very prominent in combinatorial optimization. Thus, in recent decades, a number of evolutionary algorithms have been proposed to solve them, and among them, those based on probability models have received much attention. In that sense, most efforts have focused on introducing algorithms that are suited for solving ordering/ranking nature problems. However, when it comes to proposing probability-based evolutionary algorithms for assignment problems, the works have not gone beyond proposing simple and in most cases univariate models. In this paper, we explore the use of Doubly Stochastic Matrices (DSM) for optimizing matching and assignment nature permutation problems. To that end, we explore some learning and sampling methods to efficiently incorporate DSMs within the picture of evolutionary algorithms. Specifically, we adopt the framework of estimation of distribution algorithms and compare DSMs to some existing proposals for permutation problems. Conducted preliminary experiments on instances of the quadratic assignment problem validate this line of research and show that DSMs may obtain very competitive results, while computational cost issues still need to be further investigated.
|
['Josu Ceberio', 'Valentino Santucci']
|
2023-04-05
| null | null | null | null |
['combinatorial-optimization']
|
['methodology']
|
[ 4.02883202e-01 -2.59602726e-01 -1.98677987e-01 -5.73223174e-01
-5.42293370e-01 -3.67761791e-01 3.16684157e-01 -7.71960691e-02
-2.03603461e-01 9.74745095e-01 -8.86536837e-02 -1.67044967e-01
-1.02410555e+00 -8.40200961e-01 -4.65613812e-01 -8.30020487e-01
-2.20766038e-01 9.32739675e-01 5.22124246e-02 -2.20128745e-01
8.54365051e-01 4.92570817e-01 -2.03659010e+00 5.36014251e-02
1.16044009e+00 6.00951970e-01 3.61527294e-01 3.94834071e-01
-3.01299930e-01 1.95653498e-01 -4.94320631e-01 -6.96536243e-01
3.09485197e-01 -4.96740401e-01 -6.30561411e-01 1.74121886e-01
4.38896269e-02 3.73212814e-01 -6.06202222e-02 1.10247850e+00
6.50136590e-01 2.76279569e-01 8.94345224e-01 -1.67443120e+00
-5.77784002e-01 8.97959709e-01 -7.54291117e-01 1.10206135e-01
4.47907835e-01 -3.48904371e-01 1.18514478e+00 -7.02490568e-01
5.05914211e-01 1.18310225e+00 5.79388559e-01 2.53497809e-01
-1.36608565e+00 -3.94094676e-01 2.29815051e-01 8.54354203e-01
-1.50861359e+00 -1.07038997e-01 9.34209287e-01 -1.60065800e-01
7.21135080e-01 8.40593278e-01 5.54376483e-01 8.16266835e-01
9.08375829e-02 1.11309397e+00 1.34589577e+00 -6.43522680e-01
3.62307131e-01 4.13975775e-01 1.92160517e-01 2.98874766e-01
6.00851297e-01 3.67416590e-02 -3.96759868e-01 -3.18690121e-01
-9.76667088e-03 -5.79839170e-01 -1.41896531e-01 -9.73376274e-01
-7.74137735e-01 8.77589703e-01 -1.67419985e-01 2.83241004e-01
-3.03426415e-01 -8.22014734e-02 2.22065374e-01 2.62854487e-01
4.29036617e-01 5.72088122e-01 -3.47661108e-01 -2.07040772e-01
-1.04503787e+00 5.38825333e-01 1.03945160e+00 9.77916777e-01
6.05697095e-01 -1.32886425e-01 -1.63501173e-01 8.84663880e-01
2.80815184e-01 2.47077793e-01 1.11709557e-01 -7.91798174e-01
4.31862235e-01 5.21327972e-01 1.33999243e-01 -1.44795561e+00
-3.68791491e-01 -5.96257150e-01 -6.01673901e-01 -1.16092794e-01
2.95514107e-01 6.42563477e-02 -3.08366656e-01 1.57751584e+00
4.38432008e-01 1.02359489e-01 -2.14739054e-01 6.42623305e-01
2.06715003e-01 6.61798298e-01 -4.32233900e-01 -7.32009828e-01
9.90463912e-01 -9.26916599e-01 -7.50902236e-01 1.30989641e-01
3.85535747e-01 -8.99782360e-01 7.60412037e-01 6.71206295e-01
-1.12899923e+00 -4.26285446e-01 -7.84350336e-01 6.52839005e-01
-3.51245046e-01 2.07597211e-01 8.82214069e-01 1.32797754e+00
-1.10758054e+00 6.70952201e-01 -6.61703050e-01 -4.39884573e-01
-3.55362035e-02 7.06477940e-01 2.39587843e-01 2.66504865e-02
-1.05008423e+00 1.00406146e+00 4.46890056e-01 3.91886204e-01
-2.41522014e-01 -4.96751696e-01 -4.32193041e-01 1.13436960e-01
4.41407442e-01 -7.79705226e-01 8.37302506e-01 -7.89755702e-01
-1.66068459e+00 6.00465000e-01 -2.10778624e-01 -3.35755736e-01
6.11019492e-01 1.54525831e-01 -3.59469950e-01 -3.63296956e-01
-6.95365593e-02 2.77886420e-01 6.31723404e-01 -1.40250564e+00
-6.92739308e-01 -4.06224430e-01 2.21059099e-02 1.56482100e-01
-6.59357190e-01 2.95982331e-01 -3.19796175e-01 -5.17924905e-01
4.71023284e-02 -1.23474324e+00 -4.14033562e-01 -5.60009420e-01
-4.74840790e-01 -3.87167990e-01 2.42603019e-01 -3.20373267e-01
1.67778790e+00 -1.90649807e+00 7.53494561e-01 6.75023496e-01
-5.01297534e-01 1.21799774e-01 2.24273442e-03 7.93789387e-01
-1.37303814e-01 3.38112898e-02 -4.04642522e-01 -2.59509593e-01
4.98498738e-01 3.22304457e-01 -1.59767941e-01 4.95610803e-01
-8.51188898e-02 5.69924176e-01 -7.12486684e-01 -5.11324704e-01
1.13074057e-01 5.36615960e-02 -8.77233148e-01 -1.04064256e-01
-2.78193206e-01 6.16925843e-02 -6.04910016e-01 6.98700070e-01
8.45503211e-01 1.12076968e-01 4.95320708e-01 -7.13622048e-02
-3.87265056e-01 -2.40361035e-01 -1.79823315e+00 1.44123518e+00
-1.88256741e-01 2.72435695e-01 4.69019786e-02 -1.48517919e+00
8.75613093e-01 -8.38166773e-02 6.94034457e-01 -3.70026261e-01
1.97862834e-01 4.40119356e-01 3.55202228e-01 -5.31058073e-01
9.16787267e-01 3.92676108e-02 -2.56897118e-02 4.05543685e-01
-3.51659954e-01 -1.32753208e-01 7.26678073e-01 -2.81596124e-01
7.57016242e-01 4.25406963e-01 2.30398431e-01 -2.87828267e-01
8.72838855e-01 3.02050173e-01 5.21440685e-01 7.75934637e-01
1.50399908e-01 3.78243536e-01 3.45901638e-01 -1.15592964e-01
-9.19672251e-01 -9.19000149e-01 -4.03266340e-01 1.03735280e+00
1.58717066e-01 -3.15500051e-01 -6.12926424e-01 -9.16971043e-02
1.83953181e-01 8.72950375e-01 -4.08434957e-01 -1.41752616e-01
-5.74144602e-01 -1.49187255e+00 1.48609862e-01 2.51136571e-01
1.61768407e-01 -8.08960974e-01 -2.93259859e-01 5.89734972e-01
5.28022014e-02 -5.55970907e-01 -1.03097022e-01 1.30166724e-01
-9.22758102e-01 -1.00447941e+00 -7.08819568e-01 -7.25456595e-01
5.37126005e-01 2.90215105e-01 8.94992709e-01 -3.57943118e-01
-5.17726302e-01 5.62891543e-01 -5.35154581e-01 -4.49889690e-01
-1.67208478e-01 3.60982627e-01 7.44005814e-02 2.13206559e-01
3.46715391e-01 -7.96180248e-01 -2.57259846e-01 6.23735368e-01
-6.63220704e-01 -3.94394338e-01 8.02232683e-01 7.45420575e-01
3.68493497e-01 4.27323043e-01 4.71841782e-01 -9.32646096e-01
7.97318876e-01 -5.81789672e-01 -7.91373372e-01 5.41559577e-01
-7.07766056e-01 2.09657192e-01 5.42307854e-01 -1.39545023e-01
-1.09079003e+00 2.44100466e-02 -2.77230907e-02 -9.40056592e-02
8.81791040e-02 5.89174449e-01 -2.20672458e-01 -2.66581923e-01
2.71736711e-01 5.26107848e-01 -4.87158783e-02 -3.74321133e-01
2.24933907e-01 7.27039695e-01 7.35210180e-02 -9.81667578e-01
7.29243279e-01 2.31468290e-01 3.06711465e-01 -7.25546300e-01
-4.54563171e-01 -4.51501220e-01 -2.18865946e-01 -2.36893505e-01
4.54986721e-01 -2.18635693e-01 -6.68183684e-01 2.89783508e-01
-7.09238827e-01 3.20414752e-01 -4.73277457e-02 5.21930516e-01
-1.00228500e+00 5.29601634e-01 -2.68864240e-02 -1.17082548e+00
1.63756683e-01 -1.33072484e+00 7.48440504e-01 2.43829444e-01
-2.66686320e-01 -8.29949319e-01 1.88607141e-01 4.61755514e-01
4.86334831e-01 2.53711324e-02 1.04747903e+00 -5.62101007e-01
-7.31725454e-01 -4.62841466e-02 9.10328552e-02 5.39308600e-02
-2.33582836e-02 3.29060197e-01 -4.98503387e-01 -3.09363842e-01
3.90819311e-02 2.39925571e-02 4.62095320e-01 6.07556462e-01
1.46407413e+00 1.58157438e-01 -6.44482076e-01 5.25781393e-01
1.42426300e+00 5.40627182e-01 5.93557119e-01 6.86806560e-01
8.09023231e-02 8.41961265e-01 1.07993686e+00 7.08818674e-01
4.09849167e-01 9.25986171e-01 4.33328092e-01 3.88808757e-01
2.23155886e-01 7.48304650e-02 9.13010314e-02 8.85388076e-01
-2.38127112e-01 -5.13456464e-01 -7.38740504e-01 4.26321119e-01
-2.04947948e+00 -1.17721260e+00 -5.04535735e-01 2.21766257e+00
5.80858529e-01 -2.87323035e-02 2.05960423e-01 4.42419887e-01
1.16067398e+00 -4.15012985e-02 -2.12529764e-01 -6.82403624e-01
-2.03730002e-01 1.48763552e-01 3.02494884e-01 3.72597992e-01
-9.74730074e-01 5.20014822e-01 6.32842112e+00 1.11740816e+00
-5.75052142e-01 -2.02354401e-01 4.51419771e-01 5.71045652e-02
-6.16777956e-01 7.22536370e-02 -1.01331425e+00 7.64145732e-01
7.35263824e-01 -3.98852974e-01 5.10291517e-01 7.32551813e-01
2.38465980e-01 -1.56012893e-01 -9.54985559e-01 9.62844491e-01
1.67391554e-01 -8.84113431e-01 -8.14235210e-02 1.91567868e-01
1.09625459e+00 -6.33508265e-01 3.75532210e-01 3.31008613e-01
5.14565744e-02 -8.89937043e-01 5.30748546e-01 5.67438602e-01
-1.39071168e-02 -1.03078687e+00 8.46355796e-01 3.86308163e-01
-8.78360450e-01 -3.02100807e-01 -6.92063093e-01 -1.00688478e-02
3.05494964e-01 7.04025924e-01 -6.42910600e-01 1.21374953e+00
5.05139410e-01 3.22854817e-01 -3.34923357e-01 1.85191178e+00
1.03784828e-02 4.57125664e-01 -4.23665076e-01 -6.93613112e-01
2.56911546e-01 -6.33833528e-01 7.15126932e-01 8.83540332e-01
5.44555962e-01 -1.59954667e-01 1.78333566e-01 7.10778475e-01
3.25209439e-01 4.84883845e-01 -3.03084135e-01 1.92709006e-02
4.89774168e-01 1.05333638e+00 -8.45215499e-01 1.38913423e-01
-3.22192430e-01 7.06609309e-01 1.68395892e-01 3.45336832e-02
-1.17260838e+00 -3.30794185e-01 5.03815114e-01 -1.95054606e-01
3.48961353e-01 -2.57435143e-01 -3.03560406e-01 -8.91494036e-01
9.20784548e-02 -8.65005434e-01 3.71953964e-01 -4.20220762e-01
-1.27319908e+00 2.46326029e-01 4.88554060e-01 -1.14729345e+00
-1.37307525e-01 -7.07233131e-01 -4.89887774e-01 6.36591792e-01
-1.34455693e+00 -5.49713492e-01 4.92026098e-02 1.64978653e-01
5.72963059e-01 -2.80104071e-01 6.07050359e-01 4.86816972e-01
-6.14541292e-01 4.31073546e-01 6.87415540e-01 -7.24843621e-01
5.73857605e-01 -1.27015388e+00 -9.77398083e-02 6.37925088e-01
4.00860250e-01 6.22633219e-01 1.16639531e+00 -3.79532307e-01
-1.80916941e+00 -6.20532572e-01 1.00262594e+00 -2.45127231e-01
6.16507590e-01 -3.63926515e-02 -5.80184579e-01 2.45569393e-01
1.71162382e-01 -6.42619431e-01 9.34318066e-01 4.38154072e-01
3.74481708e-01 -2.69555241e-01 -1.05855000e+00 6.54241860e-01
1.30236959e+00 1.09602831e-01 -6.92510426e-01 4.25009847e-01
-1.09547973e-01 -2.20454305e-01 -5.96409023e-01 4.99563038e-01
4.52527612e-01 -1.06432843e+00 1.12023592e+00 -4.49817598e-01
2.02404186e-01 -4.51590985e-01 -2.22240940e-01 -1.47296834e+00
-4.94762957e-01 -5.12145460e-01 1.48142353e-01 1.38060844e+00
4.01526481e-01 -6.99613333e-01 1.18218732e+00 4.00508493e-01
-2.97187209e-01 -8.32380831e-01 -8.70650649e-01 -1.01294494e+00
-8.36448148e-02 -3.84272814e-01 5.19187689e-01 7.50739336e-01
-2.22684890e-01 -1.55218348e-01 -6.20288730e-01 1.59601092e-01
7.61911273e-01 7.52020061e-01 6.96223021e-01 -1.06191325e+00
-6.32605970e-01 -6.93253577e-01 -4.87803876e-01 -7.18523860e-01
3.37924123e-01 -1.07942533e+00 -1.18664630e-01 -1.47691751e+00
3.47062409e-01 -5.70846379e-01 -1.72091976e-01 -1.23488218e-01
-2.21948013e-01 1.09347701e-01 -3.03517748e-02 -1.49734408e-01
-5.42439342e-01 7.03176498e-01 8.86237383e-01 -1.62985176e-01
-6.78467825e-02 4.48011965e-01 -6.39943719e-01 5.14304578e-01
1.08537889e+00 -5.38779259e-01 -5.64734101e-01 -3.27278972e-01
4.40003812e-01 -1.72187462e-02 -1.88160583e-01 -9.04137313e-01
5.16534567e-01 -4.00376767e-01 6.68670759e-02 -7.33777344e-01
2.75886774e-01 -8.76429915e-01 7.30812907e-01 5.50435245e-01
-1.15850985e-01 2.91003376e-01 -3.81478779e-02 7.01809287e-01
-4.29853290e-01 -7.71110773e-01 4.37281519e-01 -1.85671896e-02
-8.94334793e-01 5.75896725e-02 -4.67983454e-01 -4.59378175e-02
1.45446265e+00 -5.54553032e-01 2.04906523e-01 -9.47408080e-02
-9.16093230e-01 3.85987073e-01 3.18920106e-01 2.15724230e-01
3.92667770e-01 -1.09520817e+00 -7.06787169e-01 -2.04437047e-01
-2.69254968e-02 -5.99827468e-01 3.12268943e-01 9.39302504e-01
-6.07024431e-01 6.76557600e-01 -2.15818048e-01 -4.77067739e-01
-1.42162144e+00 7.14084923e-01 -4.10028175e-02 -2.49718487e-01
3.76065858e-02 1.00839305e+00 -3.67766291e-01 -5.92781484e-01
3.24308962e-01 2.35466078e-01 -2.92255640e-01 4.41440970e-01
3.35817412e-02 8.84185791e-01 1.57858953e-02 -2.09996849e-01
-5.87367654e-01 5.98083794e-01 4.30513769e-02 1.42261565e-01
1.60106218e+00 -1.78165093e-01 -4.97557700e-01 2.35158548e-01
9.22460377e-01 1.22962087e-01 -5.25440753e-01 1.08791038e-01
4.73078102e-01 -8.78878534e-01 -3.53683829e-01 -6.24425352e-01
-8.31075311e-01 4.60566580e-01 3.61924648e-01 3.76241565e-01
1.32452571e+00 -2.85054237e-01 1.95683852e-01 2.94946343e-01
8.68647337e-01 -1.23565722e+00 -2.81747580e-01 3.85662675e-01
6.31593704e-01 -8.26029003e-01 1.74198568e-01 -9.07929659e-01
-3.35305780e-01 1.31168234e+00 4.40343738e-01 -2.09547013e-01
4.19583201e-01 2.25427926e-01 -5.96661627e-01 1.16094582e-01
-7.18180418e-01 -2.84977973e-01 2.11075082e-01 5.12004018e-01
3.47583860e-01 7.15995058e-02 -1.32634175e+00 3.59287083e-01
-3.26751024e-01 -1.57961249e-01 5.76096475e-01 9.48244989e-01
-3.79521638e-01 -1.87160563e+00 -5.58544517e-01 4.45279479e-01
-2.34233797e-01 1.68124344e-02 -3.61087739e-01 7.95555770e-01
1.97932273e-01 8.12733054e-01 -2.62226939e-01 -2.38162979e-01
4.36489522e-01 -3.82527485e-02 9.24737453e-01 -4.08783108e-01
-6.02164924e-01 -2.92742044e-01 4.08348769e-01 -2.63915330e-01
-6.87549710e-01 -1.22996068e+00 -4.34305102e-01 -2.94921875e-01
-5.19555151e-01 6.79492176e-01 7.23875344e-01 5.48420727e-01
1.48191050e-01 5.03843844e-01 8.39035928e-01 -7.03229964e-01
-1.02335179e+00 -4.80839312e-01 -7.87475884e-01 -1.29726902e-01
-6.17270529e-01 -1.00960672e+00 -3.80298674e-01 -6.74423158e-01]
|
[6.0711588859558105, 3.9594013690948486]
|
adb50ecb-4668-4d68-9a9d-8ba47c023bb0
|
single-shot-implicit-morphable-faces-with
|
2305.03043
| null |
https://arxiv.org/abs/2305.03043v1
|
https://arxiv.org/pdf/2305.03043v1.pdf
|
Single-Shot Implicit Morphable Faces with Consistent Texture Parameterization
|
There is a growing demand for the accessible creation of high-quality 3D avatars that are animatable and customizable. Although 3D morphable models provide intuitive control for editing and animation, and robustness for single-view face reconstruction, they cannot easily capture geometric and appearance details. Methods based on neural implicit representations, such as signed distance functions (SDF) or neural radiance fields, approach photo-realism, but are difficult to animate and do not generalize well to unseen data. To tackle this problem, we propose a novel method for constructing implicit 3D morphable face models that are both generalizable and intuitive for editing. Trained from a collection of high-quality 3D scans, our face model is parameterized by geometry, expression, and texture latent codes with a learned SDF and explicit UV texture parameterization. Once trained, we can reconstruct an avatar from a single in-the-wild image by leveraging the learned prior to project the image into the latent space of our model. Our implicit morphable face models can be used to render an avatar from novel views, animate facial expressions by modifying expression codes, and edit textures by directly painting on the learned UV-texture maps. We demonstrate quantitatively and qualitatively that our method improves upon photo-realism, geometry, and expression accuracy compared to state-of-the-art methods.
|
['Sameh Khamis', 'Gordon Wetzstein', 'Leonidas Guibas', 'Umar Iqbal', 'Eric R. Chan', 'Jan Kautz', 'Koki Nagano', 'Connor Z. Lin']
|
2023-05-04
| null | null | null | null |
['face-model', 'face-reconstruction']
|
['computer-vision', 'computer-vision']
|
[ 3.40588272e-01 4.05427307e-01 1.42758980e-01 -7.05823839e-01
-3.17148060e-01 -6.55664802e-01 6.13150537e-01 -6.31894231e-01
3.04079950e-01 3.86829376e-01 1.15354575e-01 2.74480641e-01
2.46562168e-01 -1.03434658e+00 -8.66271853e-01 -4.27459151e-01
1.67757392e-01 6.95240796e-01 -4.08311367e-01 -2.42940530e-01
-4.49757986e-02 1.16051197e+00 -1.61269414e+00 3.21420193e-01
4.29910988e-01 9.78149295e-01 -3.18367809e-01 6.31112576e-01
-1.72625065e-01 5.36003292e-01 -2.19049066e-01 -5.74610472e-01
3.32378864e-01 -4.76273865e-01 -4.21989501e-01 3.26825321e-01
8.77834260e-01 -6.76897526e-01 -3.02876741e-01 6.29666865e-01
3.27800572e-01 4.89542261e-02 8.19800675e-01 -1.25566232e+00
-1.23698890e+00 -2.54233062e-01 -6.25976264e-01 -6.65616810e-01
4.36393201e-01 2.73263931e-01 5.90971828e-01 -1.04346192e+00
1.20338762e+00 1.67871368e+00 7.76527524e-01 1.27884912e+00
-1.78937340e+00 -7.60432005e-01 1.29367545e-01 -5.33784091e-01
-1.10754144e+00 -7.11859643e-01 9.47584331e-01 -4.91627753e-01
5.61578393e-01 4.73856211e-01 1.13766134e+00 1.23604250e+00
9.58574116e-02 2.60561138e-01 1.12679541e+00 -2.21325800e-01
1.67301521e-01 -9.56114084e-02 -7.70376444e-01 1.28774703e+00
-4.35877234e-01 2.74655104e-01 -6.97176516e-01 -3.23334038e-01
1.59676099e+00 3.44649656e-03 -2.11819798e-01 -7.97140360e-01
-8.72231662e-01 8.50678384e-01 3.55640680e-01 -4.50558960e-01
-1.98741734e-01 7.10809231e-01 -3.81595865e-02 1.71139583e-01
8.46157372e-01 4.44415122e-01 -3.07365596e-01 -1.19406849e-01
-7.60483682e-01 4.38584536e-01 6.48588717e-01 9.15608644e-01
1.05271053e+00 4.20106888e-01 1.26851201e-01 7.36375451e-01
3.52611631e-01 9.46494460e-01 -2.65924573e-01 -1.66389644e+00
-2.61786520e-01 4.89541024e-01 -1.19947359e-01 -1.09847856e+00
2.15991773e-02 2.53481030e-01 -7.88634419e-01 1.03205216e+00
-6.76693674e-03 1.39454186e-01 -1.16408563e+00 1.98445666e+00
5.53987801e-01 1.06187545e-01 -4.54677403e-01 8.41862559e-01
7.44922459e-01 7.28713572e-01 -1.03209745e-02 2.65752207e-02
9.37877774e-01 -5.76376379e-01 -6.02360904e-01 9.94658768e-02
1.38898253e-01 -6.79380059e-01 1.18286002e+00 1.73354521e-01
-1.55404902e+00 -2.73980826e-01 -7.14764595e-01 -5.23531079e-01
-3.23271789e-02 -2.39061534e-01 7.37755597e-01 5.22782028e-01
-1.37893867e+00 6.53883040e-01 -8.72687817e-01 -1.76185191e-01
5.21389425e-01 4.14346248e-01 -7.37505317e-01 2.45995924e-01
-6.33474112e-01 6.89394772e-01 -6.48056328e-01 4.65498231e-02
-1.12356555e+00 -9.58850980e-01 -1.00460339e+00 -2.03976452e-01
-2.88055748e-01 -1.12104487e+00 9.89020467e-01 -1.54067445e+00
-2.21047401e+00 1.38437831e+00 -1.13434903e-01 2.92928517e-01
6.48519814e-01 1.15390174e-01 -3.71402688e-02 2.45329648e-01
-2.07542300e-01 1.10434389e+00 1.16580498e+00 -1.61551452e+00
2.92584211e-01 -4.02772486e-01 2.02462912e-01 5.77307679e-02
-1.20394923e-01 -9.97960791e-02 -3.32483411e-01 -8.10245693e-01
1.17756128e-01 -1.01342905e+00 -4.37147543e-02 1.38953054e+00
-5.18079959e-02 4.57533836e-01 1.22847414e+00 -6.42124772e-01
3.63842219e-01 -1.97816002e+00 6.18289709e-01 3.72209013e-01
3.13051522e-01 -2.33845208e-02 -3.80379736e-01 8.39540288e-02
-2.17524450e-02 3.07931542e-01 -3.43204498e-01 -7.11524546e-01
-1.46599859e-02 4.45414662e-01 -3.21691900e-01 4.59832340e-01
3.18270385e-01 1.03538907e+00 -7.98591554e-01 -3.38572025e-01
2.63708770e-01 1.19788504e+00 -8.77877057e-01 4.19228107e-01
-5.50381720e-01 1.04481995e+00 -3.95216644e-01 6.28331721e-01
7.85852313e-01 -1.03803143e-01 1.52546078e-01 -2.03563616e-01
-7.97415804e-03 -5.07317334e-02 -7.68073678e-01 1.97492743e+00
-8.30761909e-01 7.12154746e-01 5.04697502e-01 -4.28984463e-01
1.08590734e+00 2.53326267e-01 5.85327983e-01 -6.45560384e-01
5.81911318e-02 5.04823774e-02 -7.43267596e-01 -3.04884851e-01
2.22365737e-01 -6.12484038e-01 1.26625732e-01 7.74758279e-01
-1.86443217e-02 -8.71177733e-01 -6.90588117e-01 2.02752091e-02
7.77711093e-01 6.21551096e-01 -3.94884527e-01 -6.61253333e-02
1.34349912e-01 -5.30263424e-01 2.20337585e-01 -4.54977639e-02
4.84822661e-01 8.51451695e-01 4.37449932e-01 -8.81227791e-01
-1.46955383e+00 -1.60706460e+00 -1.41880378e-01 7.81066179e-01
-1.27302825e-01 -1.90254837e-01 -8.85261357e-01 -3.24317455e-01
2.74589509e-01 3.63991261e-01 -1.11433125e+00 -2.04404116e-01
-7.05016971e-01 6.29214048e-02 6.08617544e-01 2.65911877e-01
2.91908860e-01 -9.05828416e-01 -4.63875800e-01 -6.54699430e-02
9.94905829e-02 -9.08875406e-01 -5.93956530e-01 -5.11929750e-01
-9.10605729e-01 -7.66341686e-01 -6.98090136e-01 -6.46867454e-01
1.09752917e+00 -1.06173232e-01 1.13406003e+00 2.72441238e-01
-4.95766610e-01 6.93144083e-01 1.17825262e-01 -7.79345334e-02
-7.19620466e-01 -5.08024752e-01 4.29580547e-02 3.33981723e-01
-5.65315187e-01 -1.12397194e+00 -6.86300874e-01 2.87075430e-01
-9.18624282e-01 5.67330420e-01 -8.67925119e-04 7.09668517e-01
7.99975872e-01 -9.83698308e-01 1.61754504e-01 -1.01004755e+00
2.44009405e-01 -4.37681610e-03 -5.48156202e-01 9.59096774e-02
-3.21229458e-01 2.85793394e-01 4.16905284e-01 -6.46510184e-01
-1.20734978e+00 1.99933276e-01 -2.18068704e-01 -9.05847728e-01
9.86673236e-02 -9.90281925e-02 -2.16001004e-01 -7.82404304e-01
6.81772709e-01 -1.89637750e-01 4.47774112e-01 -3.53249013e-01
9.30966258e-01 2.61727184e-01 3.91535759e-01 -8.79638195e-01
1.06183910e+00 1.04684246e+00 3.85109961e-01 -7.86553383e-01
-4.15244311e-01 5.29044390e-01 -7.40951002e-01 -4.09691691e-01
9.28498149e-01 -8.13053668e-01 -8.50591838e-01 4.43730831e-01
-1.28285122e+00 -7.89636850e-01 -6.18620932e-01 -9.83914733e-02
-1.05549169e+00 8.48117098e-02 -7.04757333e-01 -5.74344635e-01
-1.75703496e-01 -1.09027052e+00 1.56122887e+00 -8.92577991e-02
-5.16336977e-01 -9.85593259e-01 1.97536349e-01 1.99576035e-01
5.44124007e-01 1.02378047e+00 1.20077610e+00 9.40166473e-01
-8.56279731e-01 1.37647033e-01 -4.72983234e-02 1.57218635e-01
3.49110961e-01 4.56308275e-01 -1.12446201e+00 -1.64469387e-02
-3.70568782e-01 -5.39211452e-01 3.39457244e-01 2.61480272e-01
1.46024251e+00 -6.24593735e-01 -1.46632656e-01 1.36361825e+00
1.07785428e+00 -2.30501741e-01 6.35868669e-01 -2.72414923e-01
9.19895053e-01 7.32615292e-01 -9.43810716e-02 5.14645696e-01
2.81057060e-01 9.55578625e-01 5.80478430e-01 -3.49640220e-01
-3.69356871e-01 -6.65482640e-01 3.31753731e-01 5.36950767e-01
-5.76093316e-01 1.86490551e-01 -3.90999258e-01 4.14944775e-02
-1.49443567e+00 -8.58292818e-01 2.34917104e-01 2.11194253e+00
1.03308487e+00 -4.28309530e-01 -1.63830504e-01 -3.76619726e-01
3.36848974e-01 1.96045473e-01 -6.87015593e-01 -8.92092526e-01
-1.78459555e-01 8.03306222e-01 5.54245822e-02 7.98723817e-01
-5.03864646e-01 1.05485141e+00 6.54360485e+00 2.68063992e-01
-1.50539887e+00 1.86568230e-01 7.37150311e-01 -4.21197236e-01
-1.18338645e+00 1.44311609e-02 -9.71098989e-02 -2.36162171e-02
6.05028272e-01 -1.62789947e-03 8.37865233e-01 7.13186562e-01
2.25348487e-01 4.53868359e-01 -1.32952738e+00 1.17824221e+00
2.02719063e-01 -1.77951717e+00 6.69595778e-01 2.11104706e-01
1.11957729e+00 -6.07063532e-01 5.14531195e-01 -1.65973559e-01
3.66093904e-01 -1.47930551e+00 1.02508950e+00 8.47011685e-01
1.58595848e+00 -6.10704064e-01 -3.35236698e-01 -1.06622487e-01
-9.39971805e-01 4.71752822e-01 -1.71915457e-01 3.76102552e-02
3.38920891e-01 1.86690301e-01 -2.01820076e-01 -1.15243480e-01
5.99880815e-01 6.58698261e-01 -3.33867520e-02 2.21899837e-01
-2.17129767e-01 5.86242452e-02 -2.95693755e-01 2.80291408e-01
-1.22299746e-01 -4.28274095e-01 4.08390433e-01 7.95912862e-01
4.13663715e-01 3.59672010e-01 -2.66206622e-01 1.51930320e+00
-5.13245523e-01 -4.05757762e-02 -9.10230100e-01 -7.31771365e-02
2.72971075e-02 1.23782718e+00 -3.06202769e-01 -4.89546545e-02
-7.69574121e-02 1.32815397e+00 5.96607029e-01 6.06185496e-01
-8.36077631e-01 1.55427530e-01 1.14535129e+00 5.23064792e-01
6.62211515e-03 -6.32464945e-01 -2.92653263e-01 -1.18055618e+00
-1.05242848e-01 -6.71472251e-01 -4.23582047e-01 -1.37312555e+00
-1.12240946e+00 6.88044190e-01 -1.43939063e-01 -9.10428584e-01
-2.26475671e-01 -6.55801237e-01 -4.03306067e-01 8.93291652e-01
-1.18657780e+00 -1.66843283e+00 -5.67129791e-01 6.89496517e-01
2.15193763e-01 9.25814137e-02 1.26085961e+00 1.84665062e-02
1.74293086e-01 5.71302950e-01 -2.58202076e-01 1.98361464e-02
5.97541809e-01 -7.82345295e-01 6.55747473e-01 1.17295101e-01
1.96611822e-01 4.39198494e-01 4.41378355e-01 -4.72847283e-01
-1.80368006e+00 -9.38938200e-01 2.09329262e-01 -7.11401701e-01
1.38502613e-01 -7.51272023e-01 -6.84200466e-01 1.00858462e+00
-3.94306742e-02 4.14991021e-01 5.64881921e-01 -1.22890152e-01
-8.38207483e-01 -6.94843903e-02 -1.43370235e+00 7.98834503e-01
1.52827084e+00 -9.25685048e-01 -4.69093248e-02 2.56002426e-01
4.59823519e-01 -7.26791024e-01 -9.14111018e-01 2.01207206e-01
1.01885629e+00 -9.31593478e-01 1.07975078e+00 -7.09760964e-01
5.72951674e-01 -1.07070561e-02 -4.00415063e-02 -1.30182111e+00
-2.54937023e-01 -9.48826909e-01 -9.12389532e-02 9.08609092e-01
2.54422039e-01 -4.04307097e-01 9.37979400e-01 9.97007251e-01
5.52265085e-02 -7.88444281e-01 -8.89147818e-01 -4.72713739e-01
2.12960541e-01 -1.80876851e-01 8.88059914e-01 1.16715109e+00
-4.09384340e-01 -1.11293025e-01 -5.11434317e-01 -4.78621796e-02
6.70101941e-01 2.48775244e-01 1.02901149e+00 -1.28713799e+00
-2.62983680e-01 -3.02884787e-01 -5.15867412e-01 -1.11875105e+00
6.23581111e-01 -9.32976305e-01 -1.81164861e-01 -1.26397347e+00
-2.51039844e-02 -7.95421124e-01 5.50142407e-01 6.37718618e-01
5.29999018e-01 6.77804947e-01 1.71666682e-01 1.14961870e-01
1.54373690e-01 9.91921186e-01 1.75080240e+00 -1.14190258e-01
-7.72209093e-02 -4.57682163e-01 -3.65872294e-01 8.76762033e-01
3.87836039e-01 -2.90984154e-01 -4.65296239e-01 -8.55739772e-01
4.80306447e-01 3.01698267e-01 7.19690621e-01 -5.31747043e-01
-2.59830743e-01 -4.81842816e-01 6.70514762e-01 1.79941788e-01
1.05632460e+00 -7.26333022e-01 8.07677865e-01 -2.60639843e-03
-4.01406378e-01 1.58805490e-01 2.53157556e-01 4.07607436e-01
2.02590853e-01 3.62028271e-01 9.56663489e-01 -3.04937959e-01
-1.85119867e-01 1.00439370e+00 -8.39282945e-02 5.76127358e-02
8.32259476e-01 -4.38372374e-01 -5.96583560e-02 -7.54257083e-01
-9.34315383e-01 -3.38870466e-01 1.28668392e+00 5.13571680e-01
9.82004404e-01 -1.71821332e+00 -6.40185833e-01 6.74775183e-01
1.20757325e-02 1.70853034e-01 3.09434295e-01 1.18567474e-01
-1.08354080e+00 -3.95101786e-01 -6.35889113e-01 -6.45913064e-01
-1.23140061e+00 1.59606174e-01 7.72237122e-01 3.08393419e-01
-7.46452510e-01 8.66582274e-01 6.45018578e-01 -1.00644195e+00
-1.31447375e-01 -1.71938181e-01 5.47592580e-01 -4.62314785e-01
2.44747147e-01 -1.13150567e-01 -3.42647135e-01 -9.06464219e-01
-9.38048661e-02 1.30666232e+00 5.63077927e-01 -5.01634717e-01
1.40687990e+00 8.97517130e-02 -3.68198633e-01 2.64889151e-01
1.28931451e+00 2.83132523e-01 -1.76845789e+00 2.34110445e-01
-9.49964285e-01 -8.99817824e-01 -1.96810767e-01 -6.37942672e-01
-1.46282721e+00 1.17497706e+00 4.30584729e-01 -3.25577289e-01
8.73683333e-01 -5.51789720e-03 7.44163930e-01 8.79609957e-03
5.13137937e-01 -7.12524652e-01 5.07427454e-01 2.80249625e-01
1.36030257e+00 -7.84701049e-01 -4.89404462e-02 -6.15656853e-01
-3.27445805e-01 1.37307835e+00 5.58995605e-01 -3.03641945e-01
9.08718944e-01 5.22613645e-01 1.84669405e-01 -3.76580656e-01
-5.54284394e-01 5.98989844e-01 2.32830361e-01 8.86299908e-01
3.32861006e-01 1.76690251e-01 5.08249938e-01 -1.42134085e-01
-3.69376332e-01 8.28515887e-02 3.79323184e-01 6.38108492e-01
1.67613700e-02 -1.16354012e+00 -7.78015032e-02 -1.55777587e-02
7.32882991e-02 2.13705465e-01 -5.49753487e-01 6.02870286e-01
1.13613889e-01 2.45694473e-01 3.61560374e-01 -1.80142596e-01
3.15470964e-01 -2.44911723e-02 1.14700472e+00 -6.08002722e-01
-2.40621343e-01 -2.88564265e-01 -2.31756195e-02 -9.01853025e-01
-4.33057874e-01 -4.80193913e-01 -1.22215557e+00 -7.93334544e-01
2.21313953e-01 -3.06715488e-01 7.57384241e-01 5.07099390e-01
6.92532599e-01 1.00682333e-01 7.52265930e-01 -1.40155113e+00
-6.46385998e-02 -3.00848037e-01 -5.88483453e-01 6.60301030e-01
4.09010828e-01 -7.98205793e-01 -3.25260520e-01 3.81439298e-01]
|
[12.694307327270508, -0.3731805086135864]
|
d083ca98-0940-47b4-b172-87d98ac56c23
|
antbo-towards-real-world-automated-antibody
|
2201.12570
| null |
https://arxiv.org/abs/2201.12570v4
|
https://arxiv.org/pdf/2201.12570v4.pdf
|
AntBO: Towards Real-World Automated Antibody Design with Combinatorial Bayesian Optimisation
|
Antibodies are canonically Y-shaped multimeric proteins capable of highly specific molecular recognition. The CDRH3 region located at the tip of variable chains of an antibody dominates antigen-binding specificity. Therefore, it is a priority to design optimal antigen-specific CDRH3 regions to develop therapeutic antibodies. However, the combinatorial nature of CDRH3 sequence space makes it impossible to search for an optimal binding sequence exhaustively and efficiently using computational approaches. Here, we present \texttt{AntBO}: a combinatorial Bayesian optimisation framework enabling efficient \textit{in silico} design of the CDRH3 region. Ideally, antibodies are expected to have high target specificity and developability. We introduce a CDRH3 trust region that restricts the search to sequences with favourable developability scores to achieve this goal. For benchmarking, \texttt{AntBO} uses the \texttt{Absolut!} software suite as a black-box oracle to score the target specificity and affinity of designed antibodies \textit{in silico} in an unconstrained fashion~\citep{robert2021one}. The experiments performed for $159$ discretised antigens used in \texttt{Absolut!} demonstrate the benefit of \texttt{AntBO} in designing CDRH3 regions with diverse biophysical properties. In under $200$ calls to black-box oracle, \texttt{AntBO} can suggest antibody sequences that outperform the best binding sequence drawn from 6.9 million experimentally obtained CDRH3s and a commonly used genetic algorithm baseline. Additionally, \texttt{AntBO} finds very-high affinity CDRH3 sequences in only 38 protein designs whilst requiring no domain knowledge. We conclude \texttt{AntBO} brings automated antibody design methods closer to what is practically viable for in vitro experimentation.
|
['Amos Storkey', 'Rahmad Akbar', 'Puneet Rawat', 'Eva Smorodina', 'Philippe A. Robert', 'Antoine Grosnit', 'Haitham Bou-Ammar', 'Jun Wang', 'Dany Bou-Ammar', 'Rasul Tutunov', 'Victor Greiff', 'Kamil Dreczkowski', 'Derrick-Goh-Xin Deik', 'Alexander I. Cowen-Rivers', 'Asif Khan']
|
2022-01-29
| null | null | null | null |
['bayesian-optimisation']
|
['methodology']
|
[ 4.54680890e-01 -4.58501140e-03 5.32972720e-03 -4.11611021e-01
-8.14720690e-01 -8.82072031e-01 1.39653683e-01 1.95011988e-01
-4.74654734e-01 1.34267592e+00 -3.05144221e-01 -9.58496988e-01
-3.82172436e-01 -4.71161693e-01 -9.83791590e-01 -9.29674864e-01
1.71365574e-01 9.67834413e-01 1.98503584e-01 -3.46222550e-01
3.58235210e-01 8.71840000e-01 -1.29653931e+00 1.19680762e-01
7.92010307e-01 6.00058973e-01 4.10324395e-01 6.67486429e-01
-1.07565895e-01 1.74278319e-01 -5.23837626e-01 -3.89519721e-01
1.86123237e-01 -7.98596025e-01 -5.52715421e-01 -5.57640314e-01
1.23893552e-01 2.58728981e-01 3.80597770e-01 6.70298338e-01
9.45752382e-01 -1.30743161e-01 9.36004698e-01 -5.79436421e-01
-3.08494091e-01 -1.22458272e-01 -2.67909467e-01 2.44216546e-01
6.34635210e-01 5.34181297e-01 1.22019029e+00 -8.00654709e-01
9.40165758e-01 9.73882914e-01 4.35025036e-01 6.34894371e-01
-1.71045840e+00 -8.87070179e-01 9.09346491e-02 1.40441552e-01
-1.38119256e+00 -6.04583584e-02 2.69564480e-01 -5.58989644e-01
1.51331329e+00 8.19958746e-01 7.43901312e-01 8.90989363e-01
7.37365305e-01 4.45871919e-01 1.38195014e+00 -2.72073925e-01
4.85866785e-01 -1.61662370e-01 8.33360255e-02 5.80799520e-01
3.16506326e-02 4.34805185e-01 -6.40710533e-01 -6.92414582e-01
5.67175329e-01 -6.63445506e-04 -1.99901126e-02 -5.92402339e-01
-7.06666231e-01 1.08760285e+00 2.30194628e-01 -2.12731466e-01
-4.95227069e-01 -1.01368397e-01 -9.32543278e-02 3.26291382e-01
-3.21825325e-01 7.58817494e-01 -7.67270625e-01 -4.75611351e-02
-4.36624140e-01 5.89396894e-01 7.33177006e-01 9.69765544e-01
7.93438077e-01 -2.77028918e-01 1.37895241e-01 7.02088714e-01
2.38700181e-01 5.50656378e-01 -2.31135651e-01 -4.35259908e-01
-3.04901659e-01 5.56266248e-01 4.42342579e-01 -2.24388897e-01
-8.37955296e-01 -4.76008207e-01 -3.12225372e-01 3.45655173e-01
4.76260722e-01 -2.48698950e-01 -9.56819594e-01 1.55526459e+00
5.21233857e-01 -6.76601171e-01 -7.98595995e-02 9.52242732e-01
5.36709845e-01 5.62647164e-01 2.50170827e-01 -5.70521712e-01
1.55705655e+00 -3.80069256e-01 -1.15529798e-01 4.82790172e-02
3.27548444e-01 -1.12257969e+00 6.24028206e-01 6.39815211e-01
-9.72324133e-01 -1.10292099e-01 -1.25895619e+00 4.84925449e-01
-2.98452199e-01 -6.81990623e-01 8.10953856e-01 9.38596666e-01
-7.68527925e-01 1.44669965e-01 -5.41208506e-01 -5.26920669e-02
-8.25767070e-02 1.12321615e+00 -2.81734526e-01 -1.45466775e-01
-1.23363113e+00 1.16710913e+00 4.71126467e-01 -7.59175792e-02
-9.65005636e-01 -7.39963531e-01 -3.43687892e-01 -4.76820618e-01
2.82125890e-01 -8.89427185e-01 8.38029325e-01 -6.61155760e-01
-1.54563165e+00 6.51081324e-01 -1.01518951e-01 -3.77122104e-01
1.85983658e-01 1.89238280e-01 -3.00726920e-01 -1.02374643e-01
-2.78100580e-01 8.39532256e-01 4.58859026e-01 -7.31648862e-01
-6.64188206e-01 -3.17210615e-01 -2.66980261e-01 5.33388630e-02
6.04791105e-01 3.74056190e-01 -6.47466257e-03 -5.51672995e-01
-2.04876721e-01 -1.22803378e+00 -5.78738093e-01 -8.24789166e-01
-1.50775984e-01 -3.97467673e-01 9.41148102e-02 -2.48363197e-01
8.70339453e-01 -1.40043032e+00 3.49583685e-01 7.82542706e-01
4.65629138e-02 3.06561559e-01 -9.28119197e-03 8.85048926e-01
-2.57030100e-01 -2.82709867e-01 1.23738805e-02 9.85229492e-01
7.59507865e-02 2.21555699e-02 -7.80725926e-02 5.84546566e-01
2.39211336e-01 9.67526793e-01 -5.50474942e-01 -8.14744532e-02
1.14125066e-01 5.57625353e-01 -9.27158952e-01 2.75981754e-01
-7.90507317e-01 7.98709869e-01 -6.61957622e-01 8.57133687e-01
5.71511745e-01 -1.87523037e-01 6.58636630e-01 1.75083011e-01
-1.30547255e-01 9.49099809e-02 -8.63127410e-01 1.12126327e+00
1.70665994e-01 -7.60369077e-02 -3.71081769e-01 -5.63329399e-01
1.32522333e+00 1.55896768e-01 5.38077831e-01 -8.99548769e-01
2.08814904e-01 4.25651044e-01 5.13491511e-01 1.56281933e-01
-2.86160558e-01 -5.86194634e-01 -1.19137257e-01 3.28512460e-01
-9.66891870e-02 -1.35336474e-01 -3.12477220e-02 -2.47321680e-01
1.39593339e+00 3.17576975e-01 2.89578974e-01 -5.98820090e-01
7.52761781e-01 3.69695336e-01 6.19499743e-01 7.75011122e-01
6.99256584e-02 3.36184353e-01 4.30215627e-01 -8.30410242e-01
-1.42867041e+00 -1.02620959e+00 -3.94044638e-01 1.39624262e+00
-2.18374357e-01 -3.72404233e-03 -6.42429054e-01 -3.99146855e-01
7.14072213e-02 4.43522274e-01 -3.97638291e-01 1.15597278e-01
-6.20620847e-01 -1.05636632e+00 6.29714847e-01 -8.25111344e-02
-2.49379903e-01 -1.01118708e+00 -6.64071441e-01 7.12861538e-01
4.58526015e-01 -2.36742020e-01 -3.25848460e-01 8.85294974e-01
-2.29592904e-01 -9.95666265e-01 -9.11294937e-01 -7.48795748e-01
5.20017624e-01 -3.61990780e-01 9.78688598e-01 -1.20167650e-01
-7.04506040e-01 -2.01945245e-01 -2.86512017e-01 -7.98613191e-01
-2.90067285e-01 -3.89525592e-02 -8.00701454e-02 -5.03139198e-01
1.03143740e+00 -5.97850561e-01 -1.18683481e+00 8.58969569e-01
-6.69480801e-01 -7.66059682e-02 1.06594348e+00 1.24043691e+00
9.60339963e-01 -3.29329759e-01 7.81917036e-01 -9.34063077e-01
4.51370031e-01 -9.05109048e-02 -9.39900815e-01 1.95454448e-01
-6.93549693e-01 3.60775858e-01 6.27161026e-01 -3.30425054e-01
-7.43912280e-01 4.28393155e-01 -7.66986489e-01 2.64278919e-01
-8.93303230e-02 1.92223325e-01 -3.15964460e-01 -3.60947460e-01
8.27732682e-01 4.37320888e-01 2.35108256e-01 -3.27834874e-01
1.41010076e-01 4.85314965e-01 7.12239444e-02 -9.14297521e-01
3.79876167e-01 9.10215676e-02 2.11859882e-01 -8.13109934e-01
-2.14711696e-01 -4.19930607e-01 3.71065028e-02 1.62264720e-01
6.81525528e-01 -3.92800987e-01 -1.53817213e+00 -7.93868452e-02
-7.03244984e-01 -1.81664199e-01 2.37379283e-01 4.48045224e-01
-7.82418311e-01 2.27389544e-01 -2.77836770e-01 -6.76405132e-01
-4.07891303e-01 -1.78318715e+00 9.75499690e-01 -4.70957421e-02
-7.75913477e-01 -5.78653455e-01 3.19583058e-01 6.60390437e-01
1.42568082e-01 3.78150553e-01 1.28087068e+00 -1.01258934e+00
-7.26264119e-01 -2.18629777e-01 8.94427821e-02 -1.61734954e-01
-2.79447734e-01 -2.79694408e-01 -4.87268507e-01 -6.22823715e-01
-3.21671277e-01 -3.04905415e-01 4.20664579e-01 4.37063813e-01
3.83591741e-01 -1.98868796e-01 -3.84460688e-01 4.51930732e-01
1.43374944e+00 1.11499381e+00 6.98984563e-01 5.56002259e-01
1.80735976e-01 7.03852296e-01 9.17265952e-01 3.85530591e-01
-1.79296315e-01 9.14724469e-01 2.84779519e-01 -3.15351337e-02
5.84645629e-01 1.54521853e-01 1.66066080e-01 -3.25011276e-02
-6.04628772e-02 -1.59285218e-01 -1.00343895e+00 9.35979560e-02
-1.47505403e+00 -7.63436079e-01 -1.32778138e-01 2.15561366e+00
1.24413145e+00 3.14779222e-01 5.10682046e-01 -3.85742784e-01
4.19425696e-01 -5.57105482e-01 -9.11608219e-01 -8.34556282e-01
-3.29371363e-01 8.03993225e-01 7.90500820e-01 7.81528234e-01
-6.22183263e-01 6.66235149e-01 6.06965113e+00 1.04574358e+00
-8.69057298e-01 -3.33228827e-01 4.43100929e-01 -3.11426818e-01
-6.18841171e-01 3.85604650e-01 -1.27022719e+00 3.27589273e-01
1.17786372e+00 1.26986891e-01 3.01132590e-01 3.41180593e-01
1.63998097e-01 -9.39493626e-02 -1.02191782e+00 5.82503676e-01
-3.50729346e-01 -1.78454208e+00 2.88903974e-02 4.79529351e-01
6.80191815e-01 -1.80379391e-01 1.21793501e-01 1.01904638e-01
7.58267760e-01 -1.52462327e+00 3.70732486e-01 4.12209868e-01
6.47577584e-01 -1.27624393e+00 5.63834190e-01 2.03871265e-01
-6.85401857e-01 2.58395493e-01 -3.96636128e-01 2.46563300e-01
-1.57644581e-02 4.02122796e-01 -1.24889183e+00 3.53545547e-01
5.91680765e-01 -4.03162986e-01 -1.80209130e-01 6.49554372e-01
-5.36088925e-03 1.46920472e-01 -2.59234279e-01 -7.77367175e-01
3.79542559e-01 -3.37709427e-01 5.15489578e-01 1.26679528e+00
-3.02515715e-01 5.42498946e-01 1.57848358e-01 4.59212065e-01
3.23059201e-01 4.17423755e-01 -2.45092288e-01 2.90322751e-02
2.36454397e-01 6.46760762e-01 -5.90454519e-01 1.58438414e-01
-2.54968971e-01 5.89155436e-01 2.01541916e-01 4.61386025e-01
-8.20547342e-01 -4.67986763e-01 8.97883773e-01 1.73159793e-01
6.78038538e-01 9.59344506e-02 -1.92910209e-02 -1.22703403e-01
-3.46294224e-01 -1.50462639e+00 6.33231282e-01 -4.44158822e-01
-1.19925988e+00 6.17747962e-01 -1.99683994e-01 -5.79501390e-01
-1.77115172e-01 -9.20345724e-01 2.28220806e-01 1.31825590e+00
-1.00483406e+00 -9.06416118e-01 5.43760478e-01 2.92244315e-01
3.70478988e-01 -2.48359159e-01 8.44768643e-01 -1.28371105e-01
-2.74662912e-01 6.85032964e-01 4.34649259e-01 -7.53338635e-01
6.40595853e-01 -1.29968643e+00 2.37761706e-01 1.57964021e-01
-6.07732296e-01 1.21190679e+00 1.33146167e+00 -9.79871690e-01
-1.75795352e+00 -6.14960790e-01 5.24628162e-01 -6.87011659e-01
6.31365061e-01 -4.62311924e-01 -6.85245097e-01 4.68333066e-01
1.33924410e-01 -8.29547465e-01 1.27834630e+00 -3.76414768e-02
-1.71791062e-01 1.47675425e-01 -1.11878633e+00 7.56099582e-01
8.84385705e-01 -1.56669199e-01 -5.17381132e-01 4.69893038e-01
6.11682832e-01 -3.20077509e-01 -1.34633386e+00 5.76510787e-01
9.26589012e-01 -9.89372492e-01 1.17247820e+00 -8.31419587e-01
-3.19030344e-01 -5.92550516e-01 -4.95188646e-02 -7.95859873e-01
-4.49742526e-01 -1.19617653e+00 3.62422109e-01 5.21100402e-01
8.97526801e-01 -8.99619222e-01 1.07682443e+00 5.66050291e-01
4.56618611e-03 -1.02633047e+00 -1.23901320e+00 -7.44825125e-01
3.12413275e-01 6.42585158e-02 5.51655233e-01 3.20404351e-01
7.96735361e-02 5.26981950e-01 -2.34957710e-01 -7.92382285e-02
5.82225442e-01 2.60678768e-01 9.63203847e-01 -9.13236797e-01
-5.23568928e-01 -5.66155970e-01 -4.08881366e-01 -9.98763442e-01
-2.58127123e-01 -6.87089562e-01 -7.69487396e-02 -1.08336568e+00
3.44946831e-01 -5.43828130e-01 -2.99809188e-01 2.02274501e-01
-9.89870280e-02 4.67621861e-03 -4.01709437e-01 1.26329092e-02
-3.49927932e-01 1.64035439e-01 1.00341833e+00 -1.18906736e-01
-1.90546975e-01 -3.12632956e-02 -7.39436984e-01 1.45198420e-01
5.51384091e-01 -3.92994553e-01 -2.44101629e-01 4.85985816e-01
6.49382234e-01 1.00831635e-01 -4.11597416e-02 -2.79668003e-01
8.24983940e-02 -7.50165582e-01 6.64628983e-01 -8.40682447e-01
2.50364631e-01 -8.35295737e-01 1.06633019e+00 8.20929408e-01
1.35951743e-01 2.21742034e-01 2.36291662e-01 4.67985362e-01
3.82639378e-01 -1.70810036e-02 8.73622239e-01 -1.17825560e-01
-3.35679233e-01 -3.29574607e-02 -7.98625231e-01 -1.91067830e-01
1.24929237e+00 -6.51046872e-01 -1.76614076e-01 1.07521854e-01
-8.21206331e-01 1.88427851e-01 7.60456145e-01 1.29126564e-01
6.67884946e-01 -5.86536229e-01 -5.78139246e-01 3.27065945e-01
4.23081726e-01 -3.11548918e-01 3.05932164e-01 7.57118225e-01
-1.03890550e+00 1.11179614e+00 -3.03398907e-01 -8.14939737e-01
-1.35009480e+00 8.91317189e-01 5.30045271e-01 -1.14048079e-01
-1.06354676e-01 1.04241097e+00 4.28249180e-01 -4.62375849e-01
-7.48817176e-02 3.83944154e-01 1.56609997e-01 -3.17426413e-01
3.94701630e-01 7.95269385e-02 2.91746974e-01 -3.50442171e-01
-7.46463060e-01 4.03082371e-01 -4.35477763e-01 1.33124543e-02
1.26599360e+00 2.60253727e-01 -6.16994388e-02 -3.48113775e-01
9.73345101e-01 -4.88002487e-02 -1.24077022e+00 9.72665921e-02
2.93052197e-01 -2.89940864e-01 -4.73952204e-01 -1.28001511e+00
-8.58329758e-02 1.67164937e-01 7.87542939e-01 -3.97867471e-01
8.21836412e-01 4.97222766e-02 5.08307993e-01 4.40530270e-01
7.40589917e-01 -8.79930854e-01 -1.54536948e-01 2.44435579e-01
7.80183494e-01 -6.48477256e-01 2.00974822e-01 -8.41155574e-02
-6.07643068e-01 6.55850172e-01 5.27903914e-01 1.14753924e-01
9.02061164e-02 1.89759105e-01 6.03597462e-02 -6.32783473e-01
-1.04466975e+00 -1.96643814e-01 1.19409882e-01 6.28178596e-01
5.08152604e-01 -5.74287865e-03 -8.66396606e-01 1.94634020e-01
-2.68998265e-01 -4.36172366e-01 -8.27515647e-02 1.35961878e+00
-6.91185951e-01 -1.76941419e+00 -6.84796214e-01 1.10851608e-01
-5.96409559e-01 -1.44105732e-01 -9.54596639e-01 6.82593405e-01
-2.30699535e-02 9.16698992e-01 -3.91738087e-01 -1.02478359e-02
6.02666259e-01 1.35516480e-01 7.39456534e-01 -2.05644533e-01
-9.26172733e-01 7.36008584e-01 1.48291230e-01 -1.17756754e-01
9.08820406e-02 -5.30258060e-01 -1.47444057e+00 -3.91900122e-01
-4.53099459e-01 8.04905713e-01 4.57332253e-01 7.15828896e-01
4.70448762e-01 3.82131845e-01 4.99570012e-01 -2.79238313e-01
-2.65442967e-01 -5.11699498e-01 -3.79553020e-01 -1.70299888e-01
-5.28619699e-02 -6.57677352e-01 1.78904772e-01 -3.18406433e-01]
|
[4.752732276916504, 5.5995564460754395]
|
6033a932-2ae8-4d39-8154-a5b3f6caaa62
|
adaptively-lighting-up-facial-expression
|
2203.14045
| null |
https://arxiv.org/abs/2203.14045v1
|
https://arxiv.org/pdf/2203.14045v1.pdf
|
Adaptively Lighting up Facial Expression Crucial Regions via Local Non-Local Joint Network
|
Facial expression recognition (FER) is still one challenging research due to the small inter-class discrepancy in the facial expression data. In view of the significance of facial crucial regions for FER, many existing researches utilize the prior information from some annotated crucial points to improve the performance of FER. However, it is complicated and time-consuming to manually annotate facial crucial points, especially for vast wild expression images. Based on this, a local non-local joint network is proposed to adaptively light up the facial crucial regions in feature learning of FER in this paper. In the proposed method, two parts are constructed based on facial local and non-local information respectively, where an ensemble of multiple local networks are proposed to extract local features corresponding to multiple facial local regions and a non-local attention network is addressed to explore the significance of each local region. Especially, the attention weights obtained by the non-local network is fed into the local part to achieve the interactive feedback between the facial global and local information. Interestingly, the non-local weights corresponding to local regions are gradually updated and higher weights are given to more crucial regions. Moreover, U-Net is employed to extract the integrated features of deep semantic information and low hierarchical detail information of expression images. Finally, experimental results illustrate that the proposed method achieves more competitive performance compared with several state-of-the art methods on five benchmark datasets. Noticeably, the analyses of the non-local weights corresponding to local regions demonstrate that the proposed method can automatically enhance some crucial regions in the process of feature learning without any facial landmark information.
|
['Lin Xiong', 'Licheng Jiao', 'Dandan Yan', 'Shuiping Gou', 'GuangHui Shi', 'Shasha Mao']
|
2022-03-26
| null | null | null | null |
['facial-expression-recognition']
|
['computer-vision']
|
[ 9.58265141e-02 -1.25590429e-01 -1.58346847e-01 -6.45334959e-01
-4.60991472e-01 1.56986475e-01 1.72034770e-01 -2.31812984e-01
-4.82679784e-01 5.52386880e-01 2.75840431e-01 5.76871753e-01
-1.99652717e-01 -6.32075191e-01 -4.89306003e-01 -1.06713080e+00
1.10018775e-01 -1.72883853e-01 6.20622300e-02 -5.42970359e-01
5.47189824e-03 6.99368536e-01 -1.65885317e+00 3.41492921e-01
6.02210343e-01 1.63626826e+00 6.58904389e-02 -1.85689673e-01
-2.09020540e-01 6.00725770e-01 -3.13859910e-01 -2.62982756e-01
-3.93810980e-02 -4.00395542e-01 -4.86355633e-01 3.41028035e-01
2.44222879e-01 -3.68901223e-01 1.65188983e-02 1.33705425e+00
5.63791931e-01 1.56132936e-01 4.68620092e-01 -1.33327651e+00
-4.50070322e-01 8.49556550e-02 -1.03724313e+00 2.56598890e-01
6.30594417e-02 -7.36812130e-03 8.94663513e-01 -1.20711327e+00
5.00656188e-01 1.33940184e+00 4.57983196e-01 2.18364358e-01
-6.61299586e-01 -1.14190853e+00 5.47693253e-01 4.76124167e-01
-1.73000586e+00 -6.10825777e-01 1.28672993e+00 -2.37444639e-01
3.90704155e-01 -1.60951480e-01 6.90564156e-01 7.78392553e-01
7.99901336e-02 7.81326234e-01 1.03515136e+00 -1.16803825e-01
-1.50477946e-01 1.46442920e-01 -5.52125722e-02 1.13803625e+00
-4.74877834e-01 -1.64026871e-01 -5.42498648e-01 1.16167404e-02
8.87757123e-01 1.56623229e-01 -3.06448072e-01 -1.55103937e-01
-4.76513386e-01 7.16277421e-01 8.51225376e-01 6.09903991e-01
-5.95503867e-01 -7.43047753e-03 5.67032814e-01 -2.37270277e-02
5.98555088e-01 -3.08645934e-01 -5.73041856e-01 1.30162816e-02
-6.78579748e-01 -2.24320382e-01 4.34855372e-02 6.28835559e-01
1.36436820e+00 6.53244406e-02 -2.75130391e-01 1.13491905e+00
4.55541849e-01 4.25221259e-03 3.33785355e-01 -7.00567007e-01
4.20429736e-01 9.50344920e-01 -1.13603160e-01 -1.45506954e+00
-3.87191176e-01 -3.85630667e-01 -9.90033627e-01 7.59843960e-02
5.65247424e-02 -2.12243363e-01 -6.83008671e-01 1.97382963e+00
5.38942635e-01 2.07317948e-01 -1.23894282e-01 1.12085295e+00
9.55134809e-01 5.70236742e-01 4.61444288e-01 -5.11065364e-01
1.52332294e+00 -9.92465258e-01 -9.05912757e-01 2.70731095e-02
3.38182956e-01 -5.63953102e-01 9.13217604e-01 8.22948888e-02
-7.42862284e-01 -8.44090641e-01 -7.18517125e-01 -2.22929902e-02
-3.01605910e-01 4.87804025e-01 7.09005296e-01 8.61332491e-02
-7.80173838e-01 1.41475067e-01 -3.94050330e-01 -1.79081291e-01
9.99898851e-01 4.82872367e-01 -6.81461573e-01 -6.34153038e-02
-1.44320321e+00 5.38493991e-01 2.38195628e-01 8.49101722e-01
-6.61845982e-01 -4.14812058e-01 -1.01898646e+00 2.11750001e-01
3.95793468e-01 -1.60895750e-01 8.23690355e-01 -1.75045824e+00
-1.58336532e+00 7.36765862e-01 -3.48650336e-01 3.68913561e-01
1.93859652e-01 -5.41945547e-02 -4.52598363e-01 4.69522834e-01
1.61200687e-02 7.49629259e-01 7.77588964e-01 -1.06281424e+00
-6.60266995e-01 -6.33998632e-01 -1.95579872e-01 3.43147039e-01
-7.10751295e-01 3.28505456e-01 -7.25840628e-01 -4.85341102e-01
9.83713660e-03 -4.20275688e-01 -5.19087538e-02 3.22816491e-01
2.57061031e-02 -5.99757016e-01 1.00422776e+00 -6.64919794e-01
1.11454940e+00 -2.31934261e+00 1.22877017e-01 3.76105458e-01
5.01611829e-02 1.88064337e-01 -2.95557618e-01 -1.26568705e-01
-1.70746475e-01 -7.08671287e-02 -7.49675557e-02 -9.17939246e-02
-1.82870284e-01 9.23127979e-02 1.66762307e-01 5.30932784e-01
6.02006972e-01 8.76298189e-01 -7.49765754e-01 -8.56399477e-01
2.26377890e-01 7.91518927e-01 -3.11211675e-01 1.32555380e-01
1.34821385e-01 5.85792422e-01 -8.80724490e-01 9.72213089e-01
8.42265904e-01 4.92919572e-02 -9.16227698e-02 -5.60754955e-01
9.51324999e-02 -4.80252206e-01 -1.02645791e+00 1.61928225e+00
-3.86152953e-01 2.23460019e-01 6.07850790e-01 -1.21523321e+00
1.18248153e+00 3.19812655e-01 5.24689615e-01 -9.35919166e-01
6.49777651e-01 9.52105746e-02 -1.86220363e-01 -9.01163757e-01
5.02940714e-02 -3.47384542e-01 1.69393420e-01 -2.10250039e-02
2.79533237e-01 5.01863599e-01 -4.67157215e-02 -2.66577721e-01
4.65424746e-01 2.25779697e-01 2.80751944e-01 -1.96178406e-01
1.10257161e+00 -7.32741475e-01 1.06327450e+00 -9.64099690e-02
-5.23127198e-01 2.02813774e-01 5.12928545e-01 -5.51210463e-01
-4.66105253e-01 -5.68596363e-01 -2.40373105e-01 1.41977775e+00
4.60631400e-01 -3.69359791e-01 -7.22027481e-01 -9.04296219e-01
-1.54899970e-01 7.51660913e-02 -9.73316491e-01 -3.07510406e-01
-3.71278852e-01 -6.66366279e-01 3.07997704e-01 6.85898244e-01
1.00221717e+00 -1.40399492e+00 -1.90731660e-01 6.13218658e-02
-6.18249141e-02 -1.04429483e+00 -4.78253841e-01 -1.38091221e-01
-5.97322524e-01 -8.19507778e-01 -8.34168077e-01 -8.37807417e-01
1.06666565e+00 4.88579720e-02 5.58548987e-01 3.57650906e-01
-2.22077191e-01 4.91435565e-02 -2.58867770e-01 -2.61757612e-01
3.19941193e-01 -1.96035236e-01 -2.72017121e-01 9.11556542e-01
6.11725330e-01 -5.89074016e-01 -6.67857528e-01 5.11434674e-01
-8.89663041e-01 -9.66381058e-02 7.97995985e-01 9.03163433e-01
7.27103412e-01 1.22280508e-01 6.17162824e-01 -4.65261310e-01
3.68766606e-01 -5.51409721e-01 -2.65988648e-01 1.72524974e-01
-2.69392617e-02 -2.03927755e-01 5.12593389e-01 -4.21631485e-01
-1.32687914e+00 1.02133058e-01 -2.37667114e-01 -5.11767924e-01
-2.69558728e-01 5.12182772e-01 -7.79434443e-01 -2.67243505e-01
7.38927424e-02 1.28555194e-01 1.97972283e-01 -2.66521335e-01
6.11485727e-02 6.39029801e-01 2.28131592e-01 -7.94121802e-01
3.79954398e-01 3.45238328e-01 6.40027076e-02 -6.16363645e-01
-9.93187726e-01 -3.11251938e-01 -6.74455643e-01 -4.71503317e-01
7.87397563e-01 -1.00024605e+00 -7.30387449e-01 5.53174913e-01
-9.14741337e-01 3.70855965e-02 3.08461990e-02 1.80632606e-01
-3.15458596e-01 3.07186246e-01 -6.03708088e-01 -7.81569958e-01
-3.65806311e-01 -1.38011181e+00 1.36114109e+00 7.90891588e-01
9.67453793e-02 -8.29479218e-01 -4.27668869e-01 2.46857390e-01
2.76806265e-01 3.10781121e-01 6.85415268e-01 -1.39171377e-01
-1.95736319e-01 -7.25941360e-02 -5.92014194e-01 4.96761739e-01
3.95395100e-01 1.66679993e-01 -1.06172168e+00 2.12785490e-02
-1.40591770e-01 -6.04299009e-01 7.76974440e-01 2.11434454e-01
1.52772570e+00 -3.43967319e-01 -2.32565984e-01 6.67107582e-01
1.23913443e+00 1.83632717e-01 6.66027248e-01 1.34726778e-01
6.27017617e-01 8.82188797e-01 9.42222178e-01 5.85777760e-01
3.37002307e-01 5.78962624e-01 5.08498549e-01 -3.97970200e-01
3.33357781e-01 -1.41208440e-01 3.58836442e-01 4.53557879e-01
-4.49525535e-01 3.56012166e-01 -3.53728354e-01 4.09607559e-01
-1.90099013e+00 -8.63028407e-01 3.63004327e-01 1.79543459e+00
9.43042159e-01 -1.84194341e-01 -7.89285898e-02 -1.30521983e-01
7.87137151e-01 2.38116667e-01 -4.76501942e-01 -2.37083197e-01
-2.66715944e-01 3.35120946e-01 -1.68399781e-01 2.15374172e-01
-1.20362329e+00 1.03843689e+00 4.46395254e+00 1.17798841e+00
-1.35099161e+00 1.45751268e-01 9.85179007e-01 1.65470690e-02
-3.67647666e-03 -3.20720494e-01 -7.24355459e-01 5.08588076e-01
4.31372643e-01 1.97669894e-01 1.35264784e-01 1.07507718e+00
2.29411200e-01 5.01092896e-03 -5.83338797e-01 1.17530632e+00
1.40687436e-01 -8.11607480e-01 2.58562446e-01 -4.48984392e-02
5.48158944e-01 -4.59333152e-01 5.79687692e-02 3.67336929e-01
-2.42073029e-01 -1.04641056e+00 4.30042386e-01 8.77223611e-01
8.24078441e-01 -1.30475342e+00 1.07690609e+00 1.75219014e-01
-1.53332806e+00 -1.87950730e-01 -6.22032523e-01 1.38943885e-02
-2.65815035e-02 3.54304105e-01 -1.17194690e-01 5.65957665e-01
9.44905996e-01 8.83265972e-01 -4.98535752e-01 5.21451771e-01
-5.41264415e-01 2.35629171e-01 -3.03410709e-01 1.42236963e-01
3.99531484e-01 -4.07059103e-01 6.59126267e-02 1.04839683e+00
3.76641423e-01 3.71004015e-01 1.08679056e-01 7.42837012e-01
-2.37106249e-01 5.73511720e-01 -3.75806063e-01 2.24406779e-01
3.63348126e-02 1.89711297e+00 -4.15357709e-01 -2.66240835e-01
-4.55366194e-01 9.61931407e-01 5.27368069e-01 4.82507616e-01
-9.10677850e-01 -4.90559846e-01 7.00101912e-01 -9.40489843e-02
2.52925754e-01 1.25894427e-01 3.03772956e-01 -9.24663424e-01
2.70496663e-02 -6.22316182e-01 5.25947094e-01 -9.24755275e-01
-1.31144416e+00 7.78819144e-01 -2.28619188e-01 -1.01352477e+00
1.80212691e-01 -5.55759072e-01 -7.52066255e-01 9.50924516e-01
-1.82648134e+00 -1.30288243e+00 -7.59521306e-01 9.60582733e-01
2.78882176e-01 -6.62934408e-02 7.10963845e-01 3.78807575e-01
-8.18399072e-01 8.61439705e-01 -4.28465158e-01 3.06135178e-01
6.52178407e-01 -6.53232515e-01 -7.64955103e-01 5.24065375e-01
-1.37469843e-01 6.84097946e-01 7.97361434e-02 -3.79087061e-01
-8.20402384e-01 -9.89655435e-01 6.67001307e-01 1.74434438e-01
5.60117662e-01 -1.68115094e-01 -9.65955555e-01 4.49660242e-01
7.50134215e-02 6.97120249e-01 6.19447708e-01 4.18807603e-02
-2.02861041e-01 -6.55120730e-01 -1.10598838e+00 4.24897313e-01
8.61983836e-01 -5.52172363e-01 -2.43007109e-01 1.03668518e-01
2.77610064e-01 -1.39497548e-01 -9.68581259e-01 7.74767995e-01
6.03495598e-01 -8.78626287e-01 5.42517126e-01 -5.57774305e-01
5.76857448e-01 -4.66789305e-01 -5.38094118e-02 -9.56273377e-01
-2.57120907e-01 -3.63219589e-01 3.10278863e-01 1.55085754e+00
-7.52223879e-02 -4.51547980e-01 7.54085541e-01 3.56415004e-01
-1.10674955e-01 -1.08816075e+00 -1.05264413e+00 -1.47533119e-01
-2.69183606e-01 -9.68299881e-02 6.07312977e-01 9.14764881e-01
-1.65715143e-01 5.22606969e-01 -2.99611866e-01 4.31490503e-02
2.24841982e-01 2.25432470e-01 6.39093280e-01 -1.18024874e+00
2.41180465e-01 -4.97566640e-01 -5.69028795e-01 -8.76544654e-01
5.94602525e-01 -8.62558126e-01 -6.71250969e-02 -1.15149856e+00
4.33693647e-01 -2.30096340e-01 -9.30328965e-01 7.89524019e-01
-4.81975049e-01 5.25640547e-01 -4.48143482e-02 -6.94575980e-02
-8.06898534e-01 1.20661116e+00 1.62153268e+00 -1.47274226e-01
-2.76287012e-02 -2.73342788e-01 -8.34855020e-01 1.04852390e+00
5.68383038e-01 -1.40226558e-01 -3.30873400e-01 -1.14645556e-01
-8.40994120e-02 6.23845356e-03 3.97150964e-01 -7.47559309e-01
1.74880594e-01 -1.84764564e-01 8.54154646e-01 -3.06408435e-01
4.13188279e-01 -9.79517579e-01 -3.92705530e-01 -4.48750407e-02
-1.17395602e-01 -2.94191539e-01 2.24656656e-01 4.10674661e-01
-6.69746459e-01 2.09077005e-03 9.53339875e-01 -9.05352756e-02
-1.08496153e+00 8.61819386e-01 -2.72611715e-02 -2.46188790e-01
1.16045034e+00 -1.68651357e-01 9.76793319e-02 -2.81818330e-01
-8.90269279e-01 2.57956982e-01 5.33932038e-02 4.10538346e-01
6.19096637e-01 -1.59021592e+00 -4.73754019e-01 4.03561324e-01
3.63594085e-01 6.71041606e-04 7.36844361e-01 1.11179340e+00
-1.05448343e-01 -1.69353969e-02 -6.51755512e-01 -6.00069940e-01
-1.49064481e+00 4.11507756e-01 6.06492102e-01 -1.09716639e-01
-1.63040236e-01 1.13351262e+00 6.09825194e-01 -2.27591306e-01
-5.95811056e-04 7.50026554e-02 -6.91271365e-01 3.67008656e-01
6.14693999e-01 -6.87490404e-02 -2.63431013e-01 -1.34034896e+00
-4.81535107e-01 1.13660157e+00 -1.05416894e-01 3.23831558e-01
1.42053592e+00 -1.54871210e-01 -5.64677894e-01 7.24229813e-02
1.55941331e+00 -1.21226478e-02 -1.47574103e+00 -5.79130650e-01
-3.11123759e-01 -4.45966840e-01 1.32456869e-01 -4.56313640e-01
-1.76784575e+00 1.20588791e+00 6.49443924e-01 -7.32639849e-01
1.75040185e+00 -4.68297973e-02 5.17347574e-01 -2.85314713e-02
3.60391945e-01 -1.17513442e+00 2.80447692e-01 2.80030280e-01
1.00966001e+00 -1.14422393e+00 -9.35342088e-02 -4.15979058e-01
-6.44740880e-01 1.25821924e+00 1.05916286e+00 1.12553332e-02
8.11831594e-01 5.25416918e-02 8.42653066e-02 -3.87218088e-01
-3.19120675e-01 -1.99741855e-01 3.23694348e-01 2.17716426e-01
3.33581537e-01 -3.13512772e-01 -2.41970778e-01 1.12036216e+00
1.65091932e-01 1.97862238e-01 -2.70966977e-01 7.36241400e-01
-4.09524679e-01 -8.12127769e-01 -1.47369042e-01 1.61955506e-01
-6.25124037e-01 8.78563449e-02 -2.77377963e-01 9.71584380e-01
7.28712857e-01 7.56053627e-01 8.56120139e-02 -4.23254877e-01
1.64681956e-01 5.53356297e-02 4.04803872e-01 -1.97647020e-01
-5.12804508e-01 4.74414438e-01 -2.14996144e-01 -8.10438395e-01
-7.61787653e-01 -4.15014893e-01 -1.50130808e+00 5.40078618e-02
-2.74840891e-01 2.70609915e-01 2.83705026e-01 9.71313775e-01
3.40636373e-01 4.97945189e-01 7.76964724e-01 -8.87273490e-01
1.24995653e-02 -1.04011977e+00 -8.01663220e-01 5.80308676e-01
1.47190854e-01 -1.00527811e+00 -2.74357587e-01 -6.10869043e-02]
|
[13.626585006713867, 1.5921097993850708]
|
23852dc1-7e98-4a91-8b67-8d8fcf0df3a4
|
bayesian-nonparametric-estimation-of-coverage
|
2209.02135
| null |
https://arxiv.org/abs/2209.02135v1
|
https://arxiv.org/pdf/2209.02135v1.pdf
|
Bayesian nonparametric estimation of coverage probabilities and distinct counts from sketched data
|
The estimation of coverage probabilities, and in particular of the missing mass, is a classical statistical problem with applications in numerous scientific fields. In this paper, we study this problem in relation to randomized data compression, or sketching. This is a novel but practically relevant perspective, and it refers to situations in which coverage probabilities must be estimated based on a compressed and imperfect summary, or sketch, of the true data, because neither the full data nor the empirical frequencies of distinct symbols can be observed directly. Our contribution is a Bayesian nonparametric methodology to estimate coverage probabilities from data sketched through random hashing, which also solves the challenging problems of recovering the numbers of distinct counts in the true data and of distinct counts with a specified empirical frequency of interest. The proposed Bayesian estimators are shown to be easily applicable to large-scale analyses in combination with a Dirichlet process prior, although they involve some open computational challenges under the more general Pitman-Yor process prior. The empirical effectiveness of our methodology is demonstrated through numerical experiments and applications to real data sets of Covid DNA sequences, classic English literature, and IP addresses.
|
['Matteo Sesia', 'Stefano Favaro']
|
2022-09-05
| null | null | null | null |
['data-compression']
|
['time-series']
|
[ 8.32887948e-01 -1.61857277e-01 -3.56455564e-01 -1.03485622e-01
-8.82162392e-01 -3.37969095e-01 5.49103200e-01 3.01130325e-01
-5.31378746e-01 1.27674079e+00 -8.25584531e-02 -2.15415269e-01
-3.51105243e-01 -8.59610736e-01 -7.69556940e-01 -1.18600321e+00
-1.05316043e-01 1.20334995e+00 1.10744119e-01 2.95187384e-01
7.03603864e-01 3.91195267e-01 -1.56572378e+00 -4.49847877e-01
6.29010081e-01 5.51111162e-01 2.29087681e-01 8.58524501e-01
-1.75838128e-01 2.04841763e-01 -6.94195032e-01 -6.34316206e-01
-9.20169428e-02 2.67507657e-02 -3.02081525e-01 -4.10472676e-02
1.81296587e-01 -5.62153995e-01 -1.73619792e-01 1.13565278e+00
3.21005046e-01 -2.96813369e-01 1.19554257e+00 -1.23310220e+00
-2.25246176e-01 5.80662012e-01 -1.04218853e+00 1.55861601e-01
3.88410777e-01 -2.96077698e-01 8.11557233e-01 -9.52772737e-01
4.44059283e-01 1.08530653e+00 7.61119425e-01 -1.26472250e-01
-1.46353698e+00 -5.65988839e-01 -6.66397989e-01 4.29935269e-02
-1.83164024e+00 -4.42671806e-01 4.08777446e-01 -5.01374483e-01
4.68207210e-01 7.74670765e-02 2.20403492e-01 1.03961623e+00
2.39159316e-01 4.65425283e-01 7.83539653e-01 -4.99046236e-01
5.62138379e-01 8.66549835e-02 3.60511422e-01 3.82129133e-01
1.27290344e+00 -3.33550535e-02 -5.88717818e-01 -1.07359481e+00
7.05552399e-01 4.19279963e-01 -1.73368752e-01 -3.92845422e-01
-1.02807403e+00 1.06575549e+00 -8.46076906e-01 1.45830378e-01
-3.80668998e-01 1.55689448e-01 2.57299036e-01 -1.90630004e-01
4.22192752e-01 -3.50232363e-01 -1.10183328e-01 -3.29108268e-01
-1.24256742e+00 4.19243068e-01 1.35753155e+00 1.36208212e+00
6.15838110e-01 -1.29418463e-01 -5.22607267e-02 7.16010094e-01
2.24355906e-01 1.15605497e+00 -7.13829100e-02 -6.23744905e-01
5.25540411e-01 -8.17579255e-02 6.43308818e-01 -1.05913770e+00
8.72154012e-02 6.54039308e-02 -1.06340814e+00 -4.66242552e-01
7.89980888e-01 7.34656900e-02 -5.88819325e-01 1.74076557e+00
3.12480688e-01 4.51325715e-01 -1.35374758e-02 2.88054109e-01
2.68894732e-01 8.92869234e-01 -1.10578574e-01 -5.43797553e-01
1.61156452e+00 3.85827348e-02 -9.49600756e-01 6.90494254e-02
4.78318240e-03 -8.07042718e-01 5.55483341e-01 6.27492428e-01
-1.05678940e+00 -2.47160286e-01 -9.39242125e-01 9.32171792e-02
-2.22823583e-02 1.86068147e-01 4.53765124e-01 9.45656717e-01
-6.62236810e-01 4.90196019e-01 -7.16159046e-01 -3.41990143e-01
4.20843422e-01 1.32185206e-01 -9.01186466e-02 -4.33110416e-01
-9.30408537e-01 4.89374310e-01 2.78708875e-01 -1.71822552e-02
-7.33065605e-01 -6.13466978e-01 -7.12761045e-01 3.25226575e-01
5.41439474e-01 -3.17203701e-01 9.41252053e-01 1.23179682e-01
-1.00191784e+00 6.73534095e-01 -5.72567642e-01 -5.97953677e-01
3.55888784e-01 2.30380381e-03 8.08780715e-02 3.76528203e-01
-1.44551575e-01 -1.14938118e-01 9.66857493e-01 -9.07429934e-01
-3.06425184e-01 -4.69113797e-01 -7.72542179e-01 -3.99769455e-01
-6.87607825e-02 -3.92622918e-01 -3.29867780e-01 -7.78652012e-01
2.86681559e-02 -7.88415372e-01 -1.62673235e-01 -9.47036296e-02
-4.31354553e-01 -1.08095668e-01 3.63645136e-01 -8.02866697e-01
1.10108232e+00 -2.04312325e+00 9.45250988e-02 5.71427226e-01
1.07579686e-01 3.03596631e-02 9.24691707e-02 9.08287585e-01
5.19162655e-01 4.47013881e-03 -8.96211207e-01 -2.67667234e-01
4.59575169e-02 3.77409905e-01 -6.64791465e-01 1.05921471e+00
1.50661960e-01 2.46216610e-01 -8.11146319e-01 -6.48364782e-01
-1.13197761e-02 6.12667859e-01 -3.76188904e-01 2.72219688e-01
-9.32337791e-02 4.81674485e-02 -2.91472703e-01 6.12147152e-01
1.27386498e+00 -4.55627084e-01 5.00415325e-01 1.82180494e-01
7.53622800e-02 -9.72385332e-02 -1.58761132e+00 1.27677548e+00
-1.55338764e-01 3.18532854e-01 7.62375146e-02 -1.12038302e+00
1.13719571e+00 2.78896987e-01 3.34879071e-01 5.45566119e-02
8.34346265e-02 2.33985662e-01 -3.56518388e-01 -3.00581038e-01
9.64179516e-01 -3.33045244e-01 -2.69119501e-01 6.83525562e-01
-8.78803339e-03 -2.00919867e-01 3.52603644e-01 2.56470114e-01
1.09724808e+00 -1.94229096e-01 8.52034807e-01 -2.75275230e-01
3.29161435e-01 -2.07397282e-01 4.77795482e-01 1.23144162e+00
9.35299695e-02 5.28562129e-01 7.83767402e-01 -6.74910247e-02
-1.57887173e+00 -1.22271669e+00 -7.27957904e-01 4.11653101e-01
2.22708195e-01 -1.57499403e-01 -6.16275132e-01 -4.83560562e-03
2.30049714e-01 5.83656490e-01 -4.94301051e-01 2.12486893e-01
-3.29040766e-01 -1.10401106e+00 6.27389848e-01 1.11460164e-01
-4.83281501e-02 -8.19060326e-01 -6.77285254e-01 3.90554458e-01
-4.06876326e-01 -1.19892788e+00 -3.35995317e-01 1.34764925e-01
-9.29463744e-01 -1.22151959e+00 -9.50455666e-01 -3.62391710e-01
5.16856313e-01 2.61897445e-01 9.00715292e-01 1.02705888e-01
-6.07568145e-01 2.95073628e-01 -2.20131844e-01 -3.54501158e-01
-5.52476943e-01 -1.58225223e-01 4.49262559e-02 -1.23242557e-01
7.23203421e-01 -6.75419927e-01 -2.80428082e-01 3.81239414e-01
-1.28181136e+00 -3.97742957e-01 5.24285316e-01 1.15285552e+00
6.72942042e-01 2.24536043e-02 5.69977283e-01 -1.11439991e+00
3.73045534e-01 -8.57325554e-01 -8.33231211e-01 1.91473499e-01
-3.05232108e-01 2.63023913e-01 4.35561597e-01 -4.24035966e-01
-9.73679125e-01 -2.66239166e-01 1.51589550e-02 -1.99943438e-01
-6.48349971e-02 4.05200571e-01 -1.00070395e-01 3.11181635e-01
2.26633430e-01 6.46624088e-01 2.10102215e-01 -5.75665534e-01
8.00982639e-02 9.22257900e-01 7.02289939e-01 -8.38187695e-01
5.76848745e-01 8.83476436e-01 4.56230611e-01 -1.25822651e+00
-3.51667970e-01 -8.25631559e-01 -4.74605262e-01 3.93031687e-01
3.16818029e-01 -7.28212357e-01 -9.01308239e-01 4.59589988e-01
-1.25876939e+00 1.54093385e-01 -1.97438076e-01 5.00088751e-01
-8.11733067e-01 1.05215728e+00 -5.26434243e-01 -1.35404491e+00
-2.36677870e-01 -8.67200196e-01 1.26830125e+00 -2.62647979e-02
-4.20358591e-02 -8.61382008e-01 4.40840304e-01 -2.16076784e-02
4.05747145e-02 1.11123271e-01 1.04523134e+00 -7.48601556e-01
-4.94621009e-01 -4.72414404e-01 -2.52835214e-01 1.01650521e-01
-2.25159116e-02 1.51647136e-01 -6.41916215e-01 -4.01277959e-01
1.62550300e-01 -7.82607421e-02 7.35952973e-01 6.80539489e-01
1.13776684e+00 -3.78475040e-01 -5.29574573e-01 2.62759358e-01
1.67002141e+00 -1.26287669e-01 9.37934875e-01 -4.57457781e-01
5.50737567e-02 5.89372456e-01 7.80281663e-01 1.32443154e+00
-1.64850593e-01 5.45776665e-01 1.97260287e-02 4.82727051e-01
3.73274684e-01 -4.68912005e-01 -1.98183805e-02 8.56203794e-01
1.75515324e-01 -5.91329694e-01 -6.41870201e-01 7.71609604e-01
-1.51214898e+00 -1.45957112e+00 -2.53795832e-01 2.80679798e+00
1.06773901e+00 -8.67945179e-02 1.54413864e-01 4.25502300e-01
1.32233095e+00 -5.38342185e-02 -6.02262497e-01 -9.78491083e-02
-8.07691365e-02 4.34816062e-01 9.00294423e-01 3.85461062e-01
-7.38821089e-01 1.63765743e-01 6.70616531e+00 1.39671707e+00
-2.24193826e-01 4.97988574e-02 5.61397672e-01 2.69707441e-01
-3.08794588e-01 1.22094281e-01 -1.15193164e+00 8.24226975e-01
1.18635762e+00 -1.78412065e-01 2.24868119e-01 5.32395661e-01
-5.75399213e-02 -7.34671950e-01 -1.01811159e+00 1.07472348e+00
3.92117389e-02 -1.13214433e+00 -7.73486942e-02 4.85294819e-01
5.58953226e-01 -5.55608571e-01 1.27676427e-01 -6.56274557e-02
2.78491080e-01 -8.24607909e-01 4.22981799e-01 8.62834394e-01
1.07344258e+00 -8.88680398e-01 9.95151460e-01 6.53323472e-01
-8.05517316e-01 2.75341302e-01 -9.43423510e-01 3.68581153e-02
3.48022044e-01 1.14776015e+00 -1.10666442e+00 3.34883064e-01
3.22990179e-01 4.83453035e-01 1.96727335e-01 1.34853411e+00
1.97323114e-01 8.79710495e-01 -6.28519356e-01 -4.59158421e-01
-4.05821860e-01 -3.51193666e-01 5.31436682e-01 1.27555919e+00
8.34496856e-01 7.72651285e-02 -3.55039656e-01 8.54552329e-01
-4.64811958e-02 -8.07283297e-02 -7.50872374e-01 -1.74722984e-01
1.00084722e+00 8.52746665e-01 -7.96883643e-01 -4.11045760e-01
-1.63427681e-01 4.24289256e-01 1.63119193e-02 3.12758297e-01
-7.13623345e-01 -5.20731211e-01 3.03583682e-01 2.70464242e-01
8.52582097e-01 -2.12537557e-01 -2.21708551e-01 -9.69171464e-01
-1.27740726e-01 -6.79178715e-01 2.96318680e-01 -1.97762296e-01
-1.60484052e+00 -6.67871758e-02 4.89134938e-01 -1.15979600e+00
-2.27473632e-01 -4.07242656e-01 -2.34925523e-01 9.74640727e-01
-1.29393125e+00 -2.98272938e-01 4.57628584e-03 8.44846964e-02
2.21274912e-01 -5.66852354e-02 7.00738609e-01 6.69577867e-02
-1.64580837e-01 4.58205670e-01 9.43949819e-01 -2.66523212e-01
4.70532894e-01 -9.44840491e-01 3.52818608e-01 5.24064183e-01
-1.81870889e-02 7.08232701e-01 1.19129074e+00 -9.48024571e-01
-1.60732305e+00 -6.91430151e-01 7.68802047e-01 -2.30803072e-01
4.83271897e-01 -5.28221846e-01 -1.01679599e+00 2.64609188e-01
-3.10040861e-01 -2.78489083e-01 8.24598968e-01 -1.85848579e-01
-2.61152089e-01 2.88881123e-01 -1.64953363e+00 2.77004223e-02
7.32965529e-01 -2.08072141e-01 -5.76695502e-01 2.28503779e-01
3.93019676e-01 8.64663646e-02 -8.19158196e-01 9.20082480e-02
7.42563725e-01 -7.88385689e-01 1.16301572e+00 -2.19616577e-01
3.02949607e-01 -2.04515144e-01 -5.07802904e-01 -5.69103658e-01
2.04052299e-01 -6.01225495e-01 -2.43692383e-01 1.21204436e+00
-1.01962656e-01 -5.55580616e-01 8.02449048e-01 1.47112995e-01
6.10765040e-01 -2.73716867e-01 -1.38733125e+00 -7.59084284e-01
-1.62298471e-01 -1.99562043e-01 6.30627334e-01 5.45126438e-01
3.49415950e-02 -2.66417991e-02 -9.73595142e-01 1.72287330e-01
1.31213367e+00 1.70495324e-02 9.89506483e-01 -1.55625236e+00
-5.59476674e-01 9.26210359e-02 -3.88063103e-01 -1.18879104e+00
-5.47439232e-02 -4.69322145e-01 4.34890777e-01 -9.48266268e-01
8.04005921e-01 -4.69520718e-01 3.49829197e-01 -2.92507261e-01
-7.41224289e-02 3.57829258e-02 -3.26975048e-01 4.12554324e-01
-3.28449428e-01 5.84101140e-01 7.69096315e-01 3.35184634e-02
2.99768746e-01 2.04374120e-01 -3.15319508e-01 5.71585357e-01
4.47587252e-01 -9.54775751e-01 -2.35980555e-01 2.07128629e-01
3.30567956e-01 6.65357828e-01 3.60907257e-01 -7.12437749e-01
2.26912692e-01 -1.87450051e-01 5.04362099e-02 -1.25594938e+00
4.87845778e-01 -7.57048070e-01 4.85427797e-01 5.77914894e-01
-1.58701584e-01 -1.50477111e-01 -1.22527238e-02 1.33259237e+00
5.30541316e-02 -9.35965180e-01 6.10444605e-01 1.04141682e-01
1.76200215e-02 1.49086922e-01 -4.72270161e-01 1.14116050e-01
8.35155070e-01 -2.34135017e-01 -4.57448751e-01 -4.75846261e-01
-2.80136555e-01 -2.68373549e-01 5.23823738e-01 -4.10136014e-01
7.45056987e-01 -1.05245793e+00 -9.41264391e-01 1.87462479e-01
1.23690903e-01 -9.75558013e-02 4.86902803e-01 8.22186232e-01
-4.98518735e-01 6.26161933e-01 1.01205990e-01 -8.08190465e-01
-1.09554505e+00 6.98454380e-01 -5.14568627e-01 -3.26179028e-01
-4.23791856e-01 3.47080708e-01 -3.07204947e-02 4.08604406e-02
3.64276260e-01 -2.54820406e-01 -1.06595010e-02 -2.12969765e-01
8.67316663e-01 7.08577514e-01 -1.55873463e-01 -3.89448255e-01
-1.27374902e-01 5.17153561e-01 -3.07871457e-02 -3.39833409e-01
1.33992696e+00 -2.06415981e-01 -1.76851943e-01 6.63300395e-01
1.10017824e+00 2.24903181e-01 -1.11490130e+00 -7.05698907e-01
1.64523553e-02 -8.63260567e-01 -5.55142105e-01 -1.07456923e-01
-4.71048027e-01 9.88310158e-01 1.93956614e-01 3.26842129e-01
6.40825033e-01 6.32229224e-02 6.08850181e-01 3.42889965e-01
6.97581112e-01 -6.76816940e-01 -2.25719973e-01 1.26598939e-01
4.81888235e-01 -8.24009299e-01 4.00357455e-01 -4.75873798e-01
-9.15958136e-02 1.01586676e+00 -2.06239432e-01 -1.07958309e-01
7.25864530e-01 5.78292966e-01 -9.63351250e-01 -1.64188929e-02
-7.17207193e-01 1.74516693e-01 -2.28374332e-01 7.04571187e-01
1.62018403e-01 7.98392594e-02 -7.02258945e-01 5.68577588e-01
-1.54471304e-02 1.66102156e-01 1.00504076e+00 1.02594900e+00
-6.78821683e-01 -9.91969883e-01 -8.31820726e-01 7.74069488e-01
-6.32083178e-01 -2.68189311e-01 2.56966293e-01 6.16448045e-01
-2.78996795e-01 8.75334322e-01 2.79288173e-01 1.78796530e-01
-2.20764637e-01 4.48680157e-03 4.67928708e-01 -4.12038654e-01
3.81915569e-01 6.17099255e-02 -1.42980859e-01 2.52494991e-01
-3.81538659e-01 -1.10078287e+00 -8.46396685e-01 -8.73544633e-01
-4.29570377e-01 4.34535384e-01 9.13784146e-01 7.10877419e-01
-4.08021593e-03 -2.12712258e-01 4.41154450e-01 -6.47286057e-01
-1.15392435e+00 -1.10395718e+00 -1.20769441e+00 6.61262199e-02
3.58864129e-01 -8.09062362e-01 -6.13396287e-01 1.40152335e-01]
|
[7.091291904449463, 4.2426228523254395]
|
76b64a46-2f2a-4442-9d44-4b1c819dea13
|
zero-shot-action-recognition-with-transformer
|
2203.05156
| null |
https://arxiv.org/abs/2203.05156v2
|
https://arxiv.org/pdf/2203.05156v2.pdf
|
End-to-End Semantic Video Transformer for Zero-Shot Action Recognition
|
While video action recognition has been an active area of research for several years, zero-shot action recognition has only recently started gaining traction. In this work, we propose a novel end-to-end trained transformer model which is capable of capturing long range spatiotemporal dependencies efficiently, contrary to existing approaches which use 3D-CNNs. Moreover, to address a common ambiguity in the existing works about classes that can be considered as previously unseen, we propose a new experimentation setup that satisfies the zero-shot learning premise for action recognition by avoiding overlap between the training and testing classes. The proposed approach significantly outperforms the state of the arts in zero-shot action recognition in terms of the the top-1 accuracy on UCF-101, HMDB-51 and ActivityNet datasets. The code and proposed experimentation setup are available in GitHub: https://github.com/Secure-and-Intelligent-Systems-Lab/SemanticVideoTransformer
|
['Yasin Yilmaz', 'Keval Doshi']
|
2022-03-10
| null | null | null | null |
['zero-shot-action-recognition']
|
['computer-vision']
|
[ 2.93484241e-01 -5.92815802e-02 -4.52325881e-01 -3.26752603e-01
-5.32393515e-01 -7.25697950e-02 6.97855949e-01 -3.79112035e-01
-4.47456956e-01 5.96479297e-01 4.64697152e-01 -2.73598228e-02
-5.88652259e-03 -5.04527152e-01 -5.79393685e-01 -7.34215319e-01
1.26338035e-01 3.12513262e-01 5.81444740e-01 7.49472231e-02
9.67078581e-02 1.57274351e-01 -1.72312975e+00 4.28342521e-01
4.20332640e-01 1.17492127e+00 -6.91985562e-02 5.92838407e-01
2.68003821e-01 1.29520762e+00 -3.22299719e-01 -2.00569391e-01
2.74880290e-01 -6.15016282e-01 -9.31960583e-01 2.20302224e-01
4.24368322e-01 -6.73358083e-01 -6.69215739e-01 8.78603995e-01
5.82609177e-01 2.78366089e-01 3.05313468e-01 -1.40182340e+00
-6.48399651e-01 2.26900548e-01 -2.99880773e-01 5.39039552e-01
4.51157004e-01 2.00863570e-01 8.35844159e-01 -6.60965025e-01
6.68862879e-01 9.58266258e-01 3.26101214e-01 7.65004039e-01
-6.15642011e-01 -5.45241416e-01 1.97424069e-01 8.56800854e-01
-1.23903823e+00 -5.70404649e-01 6.72808766e-01 -3.52259099e-01
1.24171937e+00 -5.40063195e-02 7.41416037e-01 1.69104004e+00
3.21986862e-02 1.25228620e+00 8.29528809e-01 -3.55025172e-01
3.53473336e-01 -1.95210561e-01 1.26419574e-01 5.23722053e-01
-5.15739024e-02 -7.97279365e-03 -6.51286840e-01 1.83579862e-01
7.70175338e-01 4.03317243e-01 -2.27292910e-01 -6.20340705e-01
-1.16227031e+00 5.38919747e-01 3.12184151e-02 6.34149015e-01
-3.94091725e-01 2.14972779e-01 6.59018755e-01 1.60490230e-01
4.42723811e-01 -1.52045026e-01 -4.43567246e-01 -8.48595917e-01
-8.66470277e-01 -1.01552278e-01 5.42511344e-01 8.44118118e-01
2.23734766e-01 1.99842259e-01 -3.35614443e-01 7.15641320e-01
1.55968621e-01 1.22667193e-01 7.55734921e-01 -7.74472952e-01
3.76920968e-01 6.51548028e-01 7.05151185e-02 -6.45692647e-01
-1.07738614e-01 -2.73873985e-01 -6.62298918e-01 1.36858359e-01
3.79842043e-01 4.87204008e-02 -1.12526202e+00 1.56704426e+00
3.49460065e-01 9.83939409e-01 3.02273035e-01 1.03691196e+00
7.73000836e-01 4.04852867e-01 1.43806353e-01 -7.48504922e-02
1.28312612e+00 -1.20446467e+00 -7.83209443e-01 -1.06677711e-01
7.58956730e-01 -3.83030474e-01 1.00438499e+00 3.38470846e-01
-8.60285878e-01 -6.07377887e-01 -1.05055964e+00 -6.25985712e-02
-4.82395798e-01 2.10568961e-03 5.82210004e-01 4.81624067e-01
-8.55346501e-01 5.18553019e-01 -1.20203733e+00 -7.70573974e-01
8.23744953e-01 1.11386523e-01 -5.18090069e-01 -1.38836995e-01
-1.19364393e+00 9.58622098e-01 4.53237921e-01 -7.70232677e-02
-1.40421319e+00 -3.92971307e-01 -7.31559336e-01 2.79859174e-04
7.59941280e-01 -3.87870073e-01 1.24156737e+00 -1.12595737e+00
-1.49895251e+00 7.66468227e-01 -5.02512616e-04 -7.24606752e-01
5.61547339e-01 -4.79847401e-01 -5.58825135e-01 3.38778704e-01
-8.20234045e-02 5.24608552e-01 6.96815073e-01 -7.04314411e-01
-6.80333078e-01 -5.06509662e-01 3.78746271e-01 1.80084109e-01
-4.00628328e-01 -1.86621281e-03 -3.94562393e-01 -6.06915176e-01
-1.49235517e-01 -9.05802488e-01 -1.00232521e-03 -4.85958569e-02
-1.88059911e-01 -2.83752650e-01 1.15940273e+00 -4.18492079e-01
9.39806581e-01 -2.21598530e+00 2.36347243e-02 -4.85402763e-01
-2.73133814e-01 6.86662972e-01 -3.70806605e-02 3.59727532e-01
-2.33675435e-01 -2.18755350e-01 -1.30231559e-01 -3.13034981e-01
7.44652599e-02 3.49130541e-01 -1.63086936e-01 5.76154888e-01
3.51649299e-02 8.65030825e-01 -8.32492650e-01 -4.73790824e-01
7.23613441e-01 6.77456081e-01 -2.85530925e-01 1.47992089e-01
3.89124751e-02 5.43279231e-01 -4.67194706e-01 8.87144685e-01
3.76594394e-01 -3.05062413e-01 9.01130810e-02 2.56207827e-02
3.10864747e-02 -1.48335695e-02 -1.28973401e+00 2.01778769e+00
-1.23573735e-01 5.55635810e-01 -4.70870346e-01 -1.25701773e+00
4.58445042e-01 7.65284717e-01 8.66057694e-01 -8.35111499e-01
4.76569414e-01 1.42243253e-02 -6.70533553e-02 -5.84227741e-01
1.62917405e-01 -9.72095579e-02 4.43890803e-02 1.66390717e-01
5.83640814e-01 6.58404350e-01 2.09996805e-01 -1.37183117e-02
1.42474842e+00 6.30583107e-01 3.72109711e-01 8.92533958e-02
5.45580268e-01 -1.90551996e-01 7.68782139e-01 5.18900514e-01
-7.57168114e-01 6.17959023e-01 3.88394386e-01 -5.23900390e-01
-7.06753314e-01 -8.81593227e-01 -3.72940674e-02 1.07032120e+00
2.59273440e-01 -3.65686357e-01 -7.78076828e-01 -8.00633669e-01
-3.18231791e-01 7.62344837e-01 -7.65036523e-01 -1.52086705e-01
-3.44249815e-01 -2.55504131e-01 7.53922403e-01 7.37543404e-01
8.68306100e-01 -1.21993399e+00 -1.08652496e+00 2.70957109e-02
-2.93362200e-01 -1.45114732e+00 -1.58277094e-01 4.26602028e-02
-7.66332388e-01 -1.27511370e+00 -7.78888524e-01 -5.39213121e-01
3.66352826e-01 2.69250423e-01 7.65011489e-01 -1.54818222e-01
-4.60826427e-01 6.90287650e-01 -8.39130223e-01 -3.06141973e-01
9.12032351e-02 -1.20500728e-01 -8.04359764e-02 4.41642612e-01
7.89659023e-01 -6.63214326e-01 -8.60635042e-01 3.85908157e-01
-8.92283857e-01 2.36442555e-02 5.21499157e-01 6.23294473e-01
5.92143834e-01 -2.21553877e-01 4.48607862e-01 -5.39013267e-01
4.14625555e-02 -5.90004802e-01 -2.51497537e-01 2.55864203e-01
-2.92359471e-01 -1.91444397e-01 4.51661229e-01 -4.89241719e-01
-1.11794460e+00 2.40358368e-01 -1.34212747e-01 -8.47215950e-01
-6.83598757e-01 4.83198054e-02 -2.60330409e-01 7.31793791e-02
3.09954703e-01 4.13326800e-01 -2.45100826e-01 -5.05251050e-01
1.69769421e-01 7.38112926e-01 2.66677350e-01 -1.66094020e-01
3.91767323e-01 8.49804819e-01 -1.46220967e-01 -8.24647784e-01
-9.63908851e-01 -7.42962658e-01 -8.18886280e-01 -5.07075667e-01
1.28005135e+00 -9.99233902e-01 -5.07649422e-01 7.40281165e-01
-8.83833766e-01 -3.39363754e-01 -4.84415889e-01 6.72986865e-01
-8.67356122e-01 3.75637561e-01 -4.25073802e-01 -7.61316061e-01
-2.36911476e-01 -1.12886310e+00 1.11055052e+00 1.98747963e-01
-2.11772010e-01 -8.32323849e-01 1.96340740e-01 5.55075824e-01
2.18840078e-01 4.59879011e-01 3.33623320e-01 -8.29655111e-01
-5.48705399e-01 -3.86633635e-01 1.08116835e-01 5.46338856e-01
-4.09039818e-02 -2.17217922e-01 -1.04491305e+00 -2.58854598e-01
5.35349995e-02 -5.73751926e-01 8.77736032e-01 3.19058180e-01
1.03084064e+00 -4.53039538e-03 -3.01045120e-01 5.04321754e-01
1.39423525e+00 4.08404320e-01 1.12248540e+00 2.53970474e-01
5.37339151e-01 1.38599411e-01 7.93362975e-01 7.16565013e-01
2.55631149e-01 8.18718016e-01 6.33634746e-01 3.74151394e-02
-1.92588016e-01 -2.31704265e-01 4.45299536e-01 4.07980621e-01
-4.86110479e-01 -4.66749519e-01 -7.44821906e-01 7.21906543e-01
-2.20550013e+00 -1.54409218e+00 3.34805325e-02 2.08897281e+00
2.84971803e-01 2.04705760e-01 1.75864205e-01 1.48279727e-01
4.82315660e-01 6.89377010e-01 -5.17843187e-01 -3.55387144e-02
1.16947792e-01 2.90826589e-01 4.46001709e-01 1.89130470e-01
-1.35058200e+00 1.07619750e+00 5.09198475e+00 7.69273400e-01
-1.13432634e+00 5.55420399e-01 3.97028565e-01 -4.17448699e-01
4.21849519e-01 6.04736842e-02 -7.99818456e-01 5.06406128e-01
1.01458156e+00 1.04791209e-01 -1.46524208e-02 9.97994781e-01
2.49720901e-01 -1.96365118e-01 -1.05503070e+00 1.03708541e+00
4.62226629e-01 -1.18025064e+00 -3.96583192e-02 2.65354812e-02
5.77709794e-01 2.30239883e-01 -1.15093470e-01 4.83592778e-01
1.17700519e-02 -8.56276989e-01 6.12990379e-01 6.26479089e-01
6.36021435e-01 -5.08108795e-01 7.27449238e-01 3.84042293e-01
-1.18096709e+00 -2.54705518e-01 -2.49141499e-01 -2.96155483e-01
2.63671845e-01 6.89392164e-02 -3.29329312e-01 5.19934535e-01
8.37539911e-01 1.13525033e+00 -4.44572985e-01 1.02822769e+00
-2.31367841e-01 6.85954869e-01 -3.58094135e-03 1.49164036e-01
3.91075224e-01 -2.26895511e-02 4.14023459e-01 9.48241413e-01
4.23567474e-01 3.39300483e-01 1.97344184e-01 3.68131727e-01
8.84110555e-02 -4.27795798e-02 -7.23318696e-01 -2.48480290e-01
-1.01340733e-01 9.67830300e-01 -7.65941203e-01 -4.71038073e-01
-7.73954093e-01 1.28009188e+00 1.36856690e-01 2.77411878e-01
-1.31847203e+00 -2.18245268e-01 7.10779727e-01 1.32507131e-01
6.82103872e-01 -1.68626219e-01 2.70501703e-01 -1.53460872e+00
1.92791253e-01 -7.62333453e-01 5.29136419e-01 -6.58099115e-01
-7.58147299e-01 4.44265753e-01 9.04278904e-02 -1.72791862e+00
-1.20241106e-01 -6.10565007e-01 -5.49322367e-01 1.52136728e-01
-1.37143838e+00 -1.25341630e+00 -4.34508890e-01 8.46368372e-01
1.01173556e+00 -2.24075586e-01 9.47742283e-01 5.88923633e-01
-6.65625572e-01 4.44034815e-01 6.42708391e-02 3.86307240e-01
5.46130955e-01 -7.50356674e-01 1.55405343e-01 1.03128898e+00
4.33661163e-01 1.80213928e-01 5.84013164e-01 -4.96452957e-01
-1.28288412e+00 -9.31841671e-01 6.88357174e-01 -3.77268404e-01
6.45139217e-01 -2.63034970e-01 -8.02723289e-01 8.96890938e-01
4.45951641e-01 4.32041436e-01 7.63530552e-01 -1.40498325e-01
-3.34540814e-01 -7.22379088e-02 -9.73757148e-01 3.89850169e-01
1.37904155e+00 -4.42369193e-01 -5.98999083e-01 4.71381605e-01
3.00798684e-01 -2.49808475e-01 -8.47615957e-01 4.82628107e-01
6.81675375e-01 -1.06084800e+00 7.74890065e-01 -7.98606217e-01
4.18094039e-01 -3.00428927e-01 -3.66913825e-01 -9.41176653e-01
-1.78331673e-01 -3.11311483e-01 -4.28424239e-01 9.41639304e-01
1.48786185e-02 -4.30779845e-01 9.98986065e-01 2.87461609e-01
-1.91319078e-01 -8.32539976e-01 -1.20663702e+00 -1.02299929e+00
-3.12795132e-01 -5.46789050e-01 2.55189985e-01 9.16913688e-01
8.71602967e-02 2.88502067e-01 -7.66897202e-01 -5.28218485e-02
5.32062113e-01 -1.01389453e-01 6.71343327e-01 -9.85652506e-01
-4.50673461e-01 -2.11004123e-01 -1.10124147e+00 -9.81291950e-01
1.17977083e-01 -5.81236482e-01 -1.11461885e-01 -1.72305655e+00
2.65216529e-01 3.18206191e-01 -7.77563512e-01 7.02863693e-01
3.32754344e-01 4.36596543e-01 2.13638738e-01 6.97435960e-02
-1.21212804e+00 8.18197489e-01 1.06731868e+00 -3.57225537e-02
1.95264280e-01 -2.19621994e-02 -2.70047426e-01 7.97414243e-01
8.51766586e-01 -4.33246583e-01 -7.33659089e-01 -4.15211201e-01
-3.94798726e-01 4.62854914e-02 5.96463025e-01 -1.59036851e+00
1.29255965e-01 -1.25455797e-01 3.15191597e-01 -5.71427107e-01
6.93591654e-01 -9.82129931e-01 1.59134150e-01 3.70645851e-01
-3.19835514e-01 -3.84444952e-01 4.29826342e-02 7.94052780e-01
-3.58219981e-01 -8.86906013e-02 8.06265771e-01 -2.21869707e-01
-1.13817215e+00 5.04920006e-01 -3.22216749e-01 4.06444520e-02
1.68169689e+00 -5.13018727e-01 -4.19166952e-01 -2.55939156e-01
-8.84171546e-01 1.04009733e-03 3.72806102e-01 7.54080236e-01
6.94895923e-01 -1.29254401e+00 -4.99728322e-01 7.67907873e-02
3.04652900e-01 -5.24497032e-01 6.89512491e-01 1.15318799e+00
-3.55928391e-01 5.02850235e-01 -4.99301136e-01 -3.77669424e-01
-1.47154844e+00 4.95650649e-01 3.86766821e-01 -2.16406435e-01
-8.77909482e-01 7.45372474e-01 -2.48925481e-02 -8.18947330e-02
5.79943836e-01 -7.90421665e-02 -2.27822706e-01 -4.20645289e-02
6.26530945e-01 3.70975941e-01 -1.56177565e-01 -9.10626769e-01
-6.76746428e-01 2.84408122e-01 -2.10598223e-02 -1.21399621e-02
1.35233438e+00 1.06555015e-01 4.33770686e-01 7.23757267e-01
1.23591352e+00 -7.37547338e-01 -1.38585770e+00 -1.74700439e-01
2.02788040e-03 -7.29051232e-01 2.37080157e-02 -6.62090659e-01
-1.12662518e+00 1.06911123e+00 1.05236185e+00 -2.78282650e-02
1.05888629e+00 1.00195222e-02 8.35488856e-01 3.09216797e-01
5.05689979e-01 -1.28826034e+00 3.09237242e-01 5.61098158e-01
5.48941731e-01 -1.40728426e+00 -1.94763079e-01 -4.33880575e-02
-8.71874213e-01 8.53538096e-01 7.91170359e-01 -2.20201075e-01
6.75843954e-01 6.25177622e-02 -3.59584764e-02 -2.21440941e-01
-8.08224380e-01 -5.78819335e-01 -5.64892665e-02 5.81040502e-01
4.14895833e-01 -6.49227425e-02 -1.80373758e-01 4.05128717e-01
5.09695530e-01 5.94115794e-01 3.11691433e-01 1.27492785e+00
-3.71532053e-01 -9.44220960e-01 1.69891790e-01 2.97575146e-01
-6.28641248e-01 7.54685849e-02 -2.72214770e-01 8.06275368e-01
2.24656329e-01 8.72369826e-01 5.97145446e-02 -4.43817228e-01
3.87753040e-01 4.00630504e-01 5.05133688e-01 -5.12153625e-01
-2.14225695e-01 -1.78566784e-01 1.76216722e-01 -9.53915179e-01
-8.31865191e-01 -7.18416333e-01 -1.15849984e+00 -1.42560126e-02
-4.19146158e-02 -1.39946252e-01 2.70542026e-01 1.13655508e+00
5.02322018e-01 5.84219396e-01 2.07504660e-01 -6.50870681e-01
-3.13397527e-01 -1.10580599e+00 -5.55211008e-01 6.10491037e-01
2.16067676e-02 -1.03916776e+00 -2.34241247e-01 2.20910728e-01]
|
[8.345847129821777, 0.6749891042709351]
|
460dfd02-4267-4f79-8df2-65c9758568ae
|
development-of-personalized-sleep-induction
|
2212.05669
| null |
https://arxiv.org/abs/2212.05669v1
|
https://arxiv.org/pdf/2212.05669v1.pdf
|
Development of Personalized Sleep Induction System based on Mental States
|
Sleep is an essential behavior to prevent the decrement of cognitive, motor, and emotional performance and various diseases. However, it is not easy to fall asleep when people want to sleep. There are various sleep-disturbing factors such as the COVID-19 situation, noise from outside, and light during the night. We aim to develop a personalized sleep induction system based on mental states using electroencephalogram and auditory stimulation. Our system analyzes users' mental states using an electroencephalogram and results of the Pittsburgh sleep quality index and Brunel mood scale. According to mental states, the system plays sleep induction sound among five auditory stimulation: white noise, repetitive beep sounds, rainy sound, binaural beat, and sham sound. Finally, the sleep-inducing system classified the sleep stage of participants with 94.7 percent and stopped auditory stimulation if participants showed non-rapid eye movement sleep. Our system makes 18 participants fall asleep among 20 participants.
|
['Heon-Gyu Kwak', 'Gi-Hwan Shin', 'Young-Seok Kweon']
|
2022-12-12
| null | null | null | null |
['sleep-quality-prediction']
|
['medical']
|
[-3.62438411e-01 -4.47885275e-01 1.58578083e-01 -2.18827099e-01
1.85311690e-01 -2.69710124e-01 -3.07155758e-01 -1.46943629e-01
-6.40672982e-01 1.08405936e+00 3.91112417e-01 -3.46916407e-01
1.29403263e-01 -5.33099353e-01 1.46812961e-01 -5.41892171e-01
2.11530283e-01 -6.58529326e-02 3.18374127e-01 -3.59217554e-01
3.63528758e-01 1.54805958e-01 -1.75979972e+00 -7.88516998e-02
1.10293877e+00 9.39885318e-01 4.51054305e-01 4.49379295e-01
-5.14747910e-02 3.07993382e-01 -9.53930378e-01 5.71254551e-01
-9.81624573e-02 -7.61230707e-01 -2.75096953e-01 -1.48082078e-01
-6.31878018e-01 -5.26363291e-02 5.66135012e-02 1.15797448e+00
8.53370309e-01 4.02389735e-01 1.93799958e-02 -1.52145898e+00
-5.10657549e-01 -1.87046230e-02 -1.15266830e-01 9.07677412e-01
7.31009483e-01 3.00046444e-01 -2.09159385e-02 -6.02033973e-01
-4.34785664e-01 8.29553545e-01 4.21521068e-01 8.13714623e-01
-9.77428257e-01 -1.29485476e+00 -4.59119320e-01 7.70919025e-01
-1.43979442e+00 -7.87721455e-01 4.09218103e-01 -1.21056564e-01
1.01107681e+00 5.06329834e-01 1.21024323e+00 6.65664136e-01
1.01529694e+00 2.94560157e-02 1.33712435e+00 -1.51426405e-01
6.82240486e-01 3.54893148e-01 3.73607159e-01 1.80603817e-01
3.45336705e-01 -1.09935842e-01 -7.99963355e-01 1.33453816e-01
-2.35568658e-02 3.08637559e-01 -4.45470721e-01 5.23843050e-01
-6.55827820e-01 1.40847743e-01 1.21267438e-02 5.79622149e-01
-2.03464955e-01 -2.42397174e-01 8.07104930e-02 1.63062572e-01
1.79199025e-01 1.36944294e-01 -2.14851588e-01 -4.96729910e-01
-9.56966043e-01 -4.92909908e-01 5.85933864e-01 4.33232874e-01
7.21176565e-01 9.28521901e-02 -1.70138478e-01 7.60664940e-01
4.63868052e-01 1.02489626e+00 1.01441431e+00 -6.77643776e-01
-3.87511492e-01 4.73160505e-01 3.77203226e-01 -7.42798626e-01
-9.39705729e-01 -6.25193194e-02 -6.69291794e-01 1.56867251e-01
-2.17906386e-01 -1.38531968e-01 -3.62060040e-01 1.42295706e+00
-1.79884925e-01 2.63183177e-01 -1.79126173e-01 7.15997159e-01
1.06548405e+00 2.69620866e-01 -6.84419274e-03 -1.05898142e+00
1.60248435e+00 -5.94211936e-01 -1.47316146e+00 -5.37821591e-01
-7.71382079e-02 -6.26380503e-01 1.56587446e+00 5.85062504e-01
-9.36125875e-01 -6.17234230e-01 -1.09560192e+00 3.21907759e-01
-2.99300820e-01 -2.22075209e-01 9.98549834e-02 1.24358642e+00
-1.31702292e+00 1.91794515e-01 -1.06376028e+00 -6.05675876e-01
4.03581075e-02 5.53466260e-01 -6.98528886e-02 4.13006186e-01
-1.15238595e+00 9.46827412e-01 -1.66246742e-01 -3.10078058e-02
-4.27839458e-01 -3.58331144e-01 -6.03434086e-01 2.39407703e-01
-1.93493187e-01 -7.40052998e-01 1.03599465e+00 -4.76376891e-01
-1.66569161e+00 6.02570832e-01 -8.64111245e-01 -1.18700802e-01
-6.79608941e-01 -1.44346401e-01 -1.10515666e+00 9.94833559e-02
3.74582678e-01 2.70064431e-03 3.68065387e-01 -5.19758701e-01
-3.40299875e-01 -6.25770330e-01 -3.73311192e-01 2.01977909e-01
-3.23145390e-01 4.49327320e-01 3.08817208e-01 1.77159309e-01
5.61359152e-02 -8.76709163e-01 -1.65615678e-02 -6.28644884e-01
-3.60526711e-01 -4.95301455e-01 4.97732490e-01 -2.68880218e-01
1.59920430e+00 -2.45845985e+00 -7.90983379e-01 5.06290831e-02
6.18134737e-02 8.02950710e-02 3.59964550e-01 8.61328095e-03
-7.08778098e-04 2.26366892e-01 4.24056619e-01 -3.21298778e-01
1.96941748e-01 2.75370598e-01 1.23621039e-01 4.17028815e-01
-5.23647785e-01 4.32972163e-01 -8.76877666e-01 -3.67885321e-01
1.86136290e-01 1.75080866e-01 -3.21037650e-01 2.23231748e-01
6.95274949e-01 3.87923717e-01 -5.44487797e-02 5.71907401e-01
5.60223758e-01 8.20516571e-02 -6.11326993e-01 2.16673553e-01
-7.17631638e-01 8.63404274e-01 -7.26120532e-01 1.36821771e+00
-2.92641163e-01 5.28446436e-01 7.72887245e-02 -2.18834817e-01
9.28084791e-01 4.53071088e-01 1.05929293e-01 -1.15232038e+00
3.55116546e-01 -1.60148628e-02 6.21313341e-02 -8.20758700e-01
2.24310100e-01 -7.22554922e-01 1.32437155e-01 6.98532522e-01
-4.35394883e-01 -1.25586942e-01 2.70320356e-01 1.35315046e-01
1.25006795e+00 -5.82208037e-01 5.41924596e-01 -6.22406006e-01
3.25167835e-01 -7.97338188e-01 7.99665332e-01 3.83888870e-01
-6.98968649e-01 1.11291982e-01 7.88333640e-02 6.61417320e-02
2.58838926e-02 -1.26324332e+00 -4.06570211e-02 1.08489704e+00
3.73646051e-01 -5.78393340e-01 -6.57953739e-01 1.01753391e-01
-6.17744863e-01 1.17883813e+00 -2.99161911e-01 -7.47887790e-01
3.75460774e-01 -7.58686543e-01 2.17432067e-01 3.15308012e-02
6.92687213e-01 -1.28920698e+00 -6.72260284e-01 3.41523327e-02
-5.43302119e-01 -5.37787557e-01 -6.25245392e-01 2.60433882e-01
-4.70250458e-01 -7.46428907e-01 -9.89160035e-03 -5.49381375e-01
2.13686481e-01 5.61934531e-01 7.66963005e-01 1.51973486e-01
-2.51032472e-01 4.04791117e-01 -2.40551800e-01 -7.52589285e-01
3.29272486e-02 -4.64175999e-01 9.26208556e-01 -1.88970521e-01
1.07672668e+00 -1.13039327e+00 -1.14726150e+00 5.06509662e-01
-2.60077804e-01 -3.19968134e-01 1.10845692e-01 -1.61552593e-01
3.66553038e-01 3.59926343e-01 7.18884766e-01 2.64652401e-01
9.49233830e-01 -5.31949461e-01 1.82269290e-01 -2.59574175e-01
-9.31438267e-01 -5.39825022e-01 3.90140623e-01 -3.58777195e-02
-7.49486029e-01 -4.21015739e-01 -2.27093115e-01 2.75990367e-01
-6.45653605e-01 -2.31309131e-01 -4.45027500e-01 3.01080734e-01
8.47525120e-01 6.05523348e-01 6.56215325e-02 -2.85451502e-01
-5.85576057e-01 1.16546261e+00 5.66745281e-01 3.57134104e-01
3.94552231e-01 3.29995453e-01 -3.72652888e-01 -1.17396355e+00
-8.67156982e-01 -7.06846058e-01 -6.65408522e-02 -3.30144495e-01
1.20164227e+00 -7.19012618e-01 -1.25438356e+00 1.82786837e-01
-6.54595077e-01 -2.29563117e-01 6.57438710e-02 8.18043411e-01
-1.82433531e-01 4.22928512e-01 -3.08007270e-01 -1.10146987e+00
-7.15734184e-01 -9.30006027e-01 2.75375247e-01 9.14732158e-01
-6.06071770e-01 -3.93010408e-01 3.51871341e-01 1.44214094e-01
6.13784373e-01 -4.14721847e-01 3.75825703e-01 -2.18420550e-01
1.14312237e-02 2.08518907e-01 4.56140250e-01 3.94717187e-01
8.44512403e-01 -4.03841734e-01 -9.45560098e-01 5.93841076e-02
7.57580638e-01 3.68931368e-02 1.05915278e-01 5.40485263e-01
8.96908939e-01 -8.29670653e-02 -2.52211452e-01 5.32606959e-01
9.05136168e-01 1.04733384e+00 1.14101410e+00 4.01205361e-01
-1.40278697e-01 -2.49569239e-05 2.63703257e-01 4.79679376e-01
4.33809787e-01 1.58420935e-01 2.64689326e-01 8.05947781e-02
1.25874713e-01 2.49984100e-01 7.15549648e-01 8.79973292e-01
-2.66824216e-02 -7.23885139e-03 -6.35732651e-01 2.91370541e-01
-9.13638413e-01 -1.16648304e+00 -4.63067710e-01 2.18767262e+00
8.43258977e-01 2.76416510e-01 2.77966708e-01 2.74652988e-01
5.04405379e-01 -4.93935883e-01 -4.39855576e-01 -5.30811965e-01
4.61668074e-02 5.79296172e-01 5.94144054e-02 2.47106642e-01
-3.07802230e-01 4.54730213e-01 6.82299852e+00 8.51006955e-02
-1.06724405e+00 2.67041206e-01 -3.14297080e-02 -6.78622842e-01
-1.09202750e-01 -8.34061652e-02 -8.26913774e-01 1.28833449e+00
1.49438560e+00 -5.11207223e-01 7.36491144e-01 4.89713997e-01
1.18444324e+00 -9.29975092e-01 -6.69726074e-01 1.35740125e+00
3.54343355e-01 -5.62184393e-01 -7.72316515e-01 -4.31048386e-02
-1.33753806e-01 -3.83889414e-02 -1.21569000e-01 4.20814902e-01
-2.42623612e-01 -9.64092731e-01 2.63054103e-01 9.41856444e-01
6.74773991e-01 -6.81528270e-01 4.56182480e-01 4.82211113e-01
-9.70682442e-01 3.18776071e-02 -1.89488888e-01 -7.46817410e-01
1.87028110e-01 6.26766086e-01 -5.71856797e-01 -2.20078453e-01
1.10768259e+00 3.38291526e-01 -6.48431718e-01 1.26203418e+00
-4.01398718e-01 1.02009058e+00 -4.98711050e-01 -4.41398084e-01
-5.21197796e-01 -3.04576725e-01 4.10007954e-01 5.95332921e-01
5.16191542e-01 5.21078944e-01 -1.94350570e-01 7.96037436e-01
5.99481642e-01 -6.94326609e-02 -4.69472826e-01 3.90942782e-01
5.53134859e-01 1.03755403e+00 -1.02839947e+00 -5.13844155e-02
-2.97929585e-01 8.45488667e-01 -6.19582951e-01 3.23380053e-01
-5.93552828e-01 -6.00587428e-01 5.90445578e-01 1.95674255e-01
-5.26698589e-01 -6.61056265e-02 -7.02981353e-01 -8.56135964e-01
-2.89549112e-01 -5.17016530e-01 -2.28914190e-02 -1.36614072e+00
-6.95437729e-01 5.12016058e-01 -2.57852018e-01 -1.11866498e+00
2.68534780e-01 4.22644280e-02 -1.22968209e+00 8.53144228e-01
-8.65200877e-01 9.50525329e-02 -6.21693909e-01 7.60322511e-01
4.69270587e-01 -6.62655383e-02 1.04738522e+00 4.20297921e-01
-8.29245150e-01 4.06149685e-01 -2.85522848e-01 -5.14667690e-01
1.08473980e+00 -1.02754140e+00 -5.47623754e-01 6.93311691e-01
-5.78740895e-01 9.93173480e-01 1.00404274e+00 -5.28738737e-01
-7.61431396e-01 -4.69805241e-01 1.23623645e+00 -3.27036291e-01
4.32330251e-01 -2.41970435e-01 -4.82843250e-01 2.61095434e-01
5.37926257e-01 -5.83275378e-01 1.57276273e+00 8.08769390e-02
6.10531926e-01 -4.41759676e-01 -1.04872715e+00 6.74242198e-01
6.50301874e-01 -4.78076577e-01 -1.05234873e+00 1.85901314e-01
5.93634486e-01 2.92373151e-01 -3.29352289e-01 -1.63356200e-01
3.79827559e-01 -1.35097301e+00 4.52941418e-01 -4.10497747e-02
-6.80057257e-02 -5.23125410e-01 1.05392121e-01 -1.45665765e+00
-5.27107358e-01 -9.99530196e-01 4.25561905e-01 1.02246761e+00
-1.73525931e-03 -9.19795096e-01 4.03692424e-01 1.00392294e+00
-6.85803056e-01 -1.13311216e-01 -8.13225985e-01 -7.31910765e-01
-5.48368633e-01 -3.24100137e-01 6.11118078e-01 4.48360354e-01
9.38196719e-01 8.90362859e-01 -1.09205939e-01 4.90630753e-02
1.87934503e-01 -1.04536943e-01 5.39673567e-01 -1.04147136e+00
5.90866394e-02 -1.42825529e-01 -3.79191577e-01 -7.02915430e-01
5.93830682e-02 -7.29513943e-01 8.54489431e-02 -1.92702138e+00
3.03860575e-01 2.16274083e-01 -6.77526534e-01 6.57059371e-01
-1.55700788e-01 3.06245148e-01 -2.27568209e-01 -3.19775254e-01
-7.14404404e-01 7.22682774e-01 9.40832257e-01 2.09215611e-01
-7.39011168e-01 3.46488595e-01 -1.02556920e+00 8.60146284e-01
1.16233695e+00 -5.28510511e-01 -6.65004373e-01 1.46858692e-01
4.61462945e-01 5.64490333e-02 1.00234576e-01 -1.32581472e+00
4.50189918e-01 -2.83090293e-01 2.59104013e-01 -8.19077969e-01
3.62132072e-01 -6.48184538e-01 2.08457261e-01 7.54687726e-01
3.96602362e-01 6.94977567e-02 3.69762570e-01 2.08192289e-01
3.95407408e-01 -1.99847236e-01 8.85517597e-01 3.29942517e-02
-3.61809820e-01 -2.02029377e-01 -1.39798212e+00 -6.90856576e-02
8.14759791e-01 -7.26845741e-01 -3.85386139e-01 -5.85997224e-01
-1.11267734e+00 3.72166276e-01 4.13704067e-01 1.24366269e-01
5.99641442e-01 -1.21756005e+00 -7.13291541e-02 6.20980442e-01
-1.58655718e-01 -5.83555400e-01 6.68292761e-01 1.42105067e+00
-1.43093035e-01 3.66863817e-01 -6.33687973e-01 -3.59669507e-01
-1.48889649e+00 3.78292233e-01 3.38521332e-01 4.28112000e-01
-4.01170701e-01 6.69068396e-01 1.18201390e-01 5.50154090e-01
2.09147573e-01 -5.17119884e-01 -4.40438986e-01 -1.92114174e-01
1.12948573e+00 6.30046844e-01 2.13502973e-01 -1.51963234e-01
-7.75013864e-01 9.65103954e-02 4.40428287e-01 -4.14267406e-02
8.51511657e-01 -7.15132833e-01 -5.66742778e-01 8.53168190e-01
6.47560477e-01 4.65273499e-01 -4.59783107e-01 4.88789201e-01
-4.38773751e-01 -2.96859622e-01 8.74358322e-03 -8.25533211e-01
-3.69281173e-01 6.20372653e-01 1.20667326e+00 7.32502699e-01
1.56518269e+00 -1.58117130e-01 1.11488473e+00 3.97723436e-01
6.08404875e-01 -1.21756399e+00 1.20780036e-01 3.03073883e-01
5.20604074e-01 -6.25778139e-01 -1.20104805e-01 1.33939981e-01
-3.65340769e-01 5.77931941e-01 5.88833511e-01 -1.32996123e-02
1.16453421e+00 3.85780573e-01 6.07270151e-02 -1.09496377e-01
-7.53090858e-01 -3.96175981e-01 -7.39283673e-03 8.47366750e-01
2.83927947e-01 1.06263645e-01 -7.67485857e-01 1.32850623e+00
-8.86979222e-01 3.43416542e-01 7.79194295e-01 5.13830483e-01
-1.00565827e+00 -6.38823628e-01 -5.85331559e-01 7.69574344e-01
-5.95250428e-01 -1.43315032e-01 -1.96065307e-01 1.88911125e-01
5.35213113e-01 1.81633866e+00 2.98339456e-01 -6.02659106e-01
4.83691573e-01 4.11448270e-01 9.01634395e-02 -1.02914119e+00
-3.39177251e-01 1.73390105e-01 -2.23410666e-01 -6.91180885e-01
-5.89693069e-01 -5.66448808e-01 -1.69309986e+00 -2.93942541e-01
-3.81892659e-02 5.59127867e-01 5.49238563e-01 1.05569923e+00
3.41224253e-01 5.67360997e-01 4.69917417e-01 -5.21208882e-01
3.36274981e-01 -1.28728437e+00 -1.19501495e+00 -4.31095995e-03
3.90290022e-01 -8.49461317e-01 -7.66665280e-01 2.35137288e-02]
|
[13.501275062561035, 3.4647045135498047]
|
4c5d4877-df08-443c-8c60-33e7afc55b96
|
semi-automatic-definite-description
|
1712.08933
| null |
http://arxiv.org/abs/1712.08933v1
|
http://arxiv.org/pdf/1712.08933v1.pdf
|
Semi-automatic definite description annotation: a first report
|
Studies in Referring Expression Generation (REG) often make use of corpora of
definite descriptions produced by human subjects in controlled experiments.
Experiments of this kind, which are essential for the study of reference
phenomena and many others, may however include a considerable amount of noise.
Human subjects may easily lack attention, or may simply misunderstand the task
at hand and, as a result, the elicited data may include large proportions of
ambiguous or ill-formed descriptions. In addition to that, REG corpora are
usually collected for the study of semantics-related phenomena, and it is often
the case that the elicited descriptions (and their input contexts) need to be
annotated with their corresponding semantic properties. This, as in many other
fields, may require considerable time and skilled annotators. As a means to
tackle both kinds of difficulties - poor data quality and high annotation costs
- this work discusses a semi-automatic method for the annotation of definite
descriptions produced by human subjects in REG data collection experiments. The
method makes use of simple rules to establish associations between words and
meanings, and is intended to facilitate the design of experiments that produce
REG corpora.
|
['Ivandre Paraboni', 'Alex Gwo Jen Lan', 'Danillo da Silva Rocha']
|
2017-12-24
| null | null | null | null |
['referring-expression-generation']
|
['computer-vision']
|
[ 3.16717386e-01 1.24377176e-01 7.53631815e-03 -5.45943022e-01
-8.07756484e-01 -7.72217989e-01 7.34757602e-01 5.99854708e-01
-5.57555914e-01 1.07132900e+00 3.13601732e-01 -2.42378980e-01
1.47629231e-02 -6.50591791e-01 -3.14151853e-01 -4.40674305e-01
3.58028233e-01 7.47017026e-01 3.23898315e-01 -3.37828517e-01
4.25621837e-01 3.67540032e-01 -1.49289298e+00 1.80347323e-01
7.17966318e-01 3.83588523e-01 5.21126926e-01 3.87227237e-02
-6.49970770e-01 8.97311211e-01 -9.17504489e-01 -6.73105597e-01
-3.07960480e-01 -6.05316043e-01 -1.15189624e+00 5.07995248e-01
-2.46645376e-01 2.33780012e-01 3.40498298e-01 1.21466041e+00
3.88506979e-01 4.26987350e-01 5.67687392e-01 -1.13402963e+00
-4.49738503e-01 7.11022675e-01 -1.68191686e-01 1.91245392e-01
7.42020488e-01 -2.47677237e-01 1.00624526e+00 -7.81960189e-01
9.17773306e-01 1.16893816e+00 3.63435805e-01 7.01830983e-01
-1.44731104e+00 -3.16499829e-01 -1.52786121e-01 1.17327990e-02
-1.45506394e+00 -6.29967093e-01 8.33387017e-01 -4.97196287e-01
7.51251876e-01 1.94518998e-01 2.10014597e-01 1.16126204e+00
-2.84999281e-01 4.05989915e-01 9.81332898e-01 -8.68521929e-01
2.48455063e-01 7.27157712e-01 9.46634933e-02 1.57921329e-01
4.42207247e-01 -2.96512395e-01 -3.86159122e-01 -1.39833421e-01
5.53362846e-01 -6.05636120e-01 -5.29262960e-01 -2.49538198e-01
-1.34665203e+00 7.73330569e-01 9.63901579e-02 1.14613974e+00
-4.75581944e-01 -2.50478506e-01 6.93450391e-01 1.03797555e-01
2.84984171e-01 7.98726022e-01 -4.08873916e-01 -3.38465989e-01
-5.73519349e-01 4.85568792e-01 9.81440306e-01 1.21461332e+00
4.79088366e-01 -2.47212455e-01 1.85151815e-01 1.00948238e+00
1.78450257e-01 7.00182701e-03 6.73455119e-01 -6.14124060e-01
4.61001933e-01 7.69057810e-01 5.60908616e-01 -1.07992172e+00
-3.79303135e-02 1.65201113e-01 -5.08124053e-01 -3.34484167e-02
7.16002882e-01 1.60178304e-01 -3.83072376e-01 1.82539392e+00
2.92037934e-01 -7.54534423e-01 2.53436655e-01 9.48910832e-01
9.37426746e-01 3.95010769e-01 5.61919391e-01 -7.26278365e-01
1.59397995e+00 -2.62838900e-01 -1.21558464e+00 -9.96915698e-02
8.43888402e-01 -9.26530242e-01 1.52028942e+00 4.87875426e-03
-1.15472925e+00 -2.65355617e-01 -5.84100068e-01 -1.34301305e-01
-5.49992263e-01 -1.23844035e-01 4.64456469e-01 2.55257249e-01
-6.59830451e-01 4.58543986e-01 -2.70273179e-01 -6.01011276e-01
-8.75972807e-02 2.40592986e-01 -4.21575189e-01 1.43265188e-01
-1.24102461e+00 1.11367273e+00 6.05179310e-01 2.08248600e-01
1.95698496e-02 -6.92656338e-02 -9.57210302e-01 -8.90142173e-02
7.17229247e-01 -2.21602857e-01 1.47546959e+00 -1.25394773e+00
-9.24516201e-01 1.33612633e+00 -3.15083861e-01 -2.67026965e-02
3.22796643e-01 9.61480513e-02 -6.25877738e-01 -1.35443792e-01
5.97958982e-01 1.81037679e-01 3.09400231e-01 -1.21478641e+00
-3.53546053e-01 -4.00081336e-01 1.10520765e-01 1.78851470e-01
2.55184680e-01 6.24988735e-01 -8.26700181e-02 -7.39830673e-01
2.21368074e-01 -6.23180211e-01 -1.86820343e-01 -3.79565269e-01
-2.72136748e-01 -5.88024497e-01 5.76146781e-01 -2.38585308e-01
1.17882216e+00 -2.10602856e+00 3.59065719e-02 7.23375659e-03
1.26268834e-01 2.96306580e-01 2.90948957e-01 5.76654196e-01
-1.55725300e-01 4.46559370e-01 -2.43690625e-01 -1.83156710e-02
1.99254289e-01 3.80723357e-01 -1.37346789e-01 -7.86042120e-03
2.72448629e-01 5.91371238e-01 -9.20319736e-01 -8.76681209e-01
8.41523558e-02 1.92470551e-01 5.53192869e-02 3.13479185e-01
-3.72779459e-01 3.41277540e-01 -7.87902653e-01 3.06660891e-01
8.57690722e-03 -1.42972678e-01 2.09216818e-01 -7.11397156e-02
-2.85341561e-01 7.89315462e-01 -1.09649014e+00 1.39553618e+00
-4.00574207e-01 5.55930138e-01 -1.11017235e-01 -1.07925594e+00
8.50408792e-01 8.03780913e-01 7.50983059e-02 -3.93042326e-01
4.05811906e-01 6.11984551e-01 4.10757102e-02 -8.14001679e-01
7.15211153e-01 -6.91940725e-01 -3.74591351e-01 5.00530183e-01
-1.00704975e-01 -4.47534651e-01 6.35518789e-01 1.83199439e-02
7.27590442e-01 2.22207949e-01 8.55895698e-01 -2.64035821e-01
7.29892671e-01 4.38067228e-01 8.01643193e-01 1.92301244e-01
-9.93336961e-02 1.83993071e-01 8.25916231e-01 -4.38984603e-01
-1.00027061e+00 -5.98069370e-01 -2.34685510e-01 7.61630177e-01
9.62158516e-02 -3.40233982e-01 -7.21459806e-01 -5.39637983e-01
-6.09053195e-01 1.14003825e+00 -4.26169187e-01 1.79291651e-01
-3.90754879e-01 -5.65909564e-01 2.87153780e-01 3.45941752e-01
2.58185118e-01 -1.71621013e+00 -8.13037038e-01 6.89045429e-01
-4.71351415e-01 -1.33323586e+00 -1.68051004e-01 1.31623194e-01
-5.61603367e-01 -1.20878792e+00 -3.44332367e-01 -8.67435277e-01
8.45593512e-01 -1.95548926e-02 1.23724771e+00 2.79503018e-01
1.35687947e-01 -1.08893439e-02 -7.42485940e-01 -5.84954977e-01
-7.97015727e-01 -1.12879284e-01 -2.10800380e-01 -2.65776992e-01
9.16035116e-01 -4.08826381e-01 3.32502902e-01 3.66503775e-01
-1.11094379e+00 4.13942104e-03 1.47115171e-01 7.68318713e-01
5.25393188e-01 7.28004277e-02 6.97412848e-01 -1.09326196e+00
1.03358305e+00 -2.64468431e-01 -6.19870305e-01 2.55679071e-01
-2.74517030e-01 1.67541578e-01 5.73567450e-01 -4.78376240e-01
-1.02045345e+00 -6.95287660e-02 -1.29389599e-01 7.62540549e-02
-7.19345272e-01 6.13323152e-01 -6.24322891e-01 2.98638433e-01
8.69036317e-01 -1.91242639e-02 -2.18326792e-01 -4.13400322e-01
-3.86464875e-04 8.15875113e-01 4.15835947e-01 -8.05734575e-01
5.76886237e-01 -2.39396706e-01 -1.00174591e-01 -7.96992481e-01
-9.87041295e-01 -4.11475182e-01 -7.61399329e-01 -1.42318577e-01
7.18497097e-01 -4.72305357e-01 -1.69804931e-01 1.72431543e-02
-1.40219927e+00 -2.95475274e-01 -4.49529767e-01 4.38097984e-01
-6.23200953e-01 1.71143040e-01 -2.48339385e-01 -7.60769844e-01
-3.31995264e-02 -1.16357005e+00 7.81944096e-01 9.79260504e-02
-9.66738939e-01 -1.09525442e+00 -3.56851786e-01 1.84209600e-01
2.48573378e-01 3.84783268e-01 1.21231103e+00 -9.80145216e-01
-9.92716253e-02 -3.31840038e-01 -2.01952010e-01 1.19013794e-01
3.55081797e-01 -4.93385158e-02 -6.56749547e-01 3.65867049e-01
2.06348166e-01 -4.96261120e-01 -1.47143096e-01 -6.72559738e-02
6.27781332e-01 -4.29396421e-01 -6.68562949e-02 -4.40575689e-01
1.34025609e+00 5.15492380e-01 4.76114422e-01 3.41921657e-01
3.69426817e-01 1.17104983e+00 8.38537216e-01 1.95721135e-01
1.93963528e-01 9.72394288e-01 -1.10042766e-02 2.36315921e-01
3.09863687e-01 4.10816558e-02 -1.08897090e-01 7.20187366e-01
-1.11981198e-01 -3.40832323e-01 -9.13049698e-01 8.45602751e-01
-1.67768419e+00 -9.95415032e-01 -5.64406216e-01 2.19206500e+00
1.14499354e+00 4.63943705e-02 4.50069793e-02 5.91450989e-01
8.66911709e-01 -5.58644906e-02 9.99094844e-02 -4.90187138e-01
-1.00694597e-01 -7.84509778e-02 -2.18206927e-01 3.47061098e-01
-6.72592223e-01 9.77120161e-01 5.49667120e+00 4.34879839e-01
-8.23027492e-01 -7.68129826e-02 1.31149948e-01 5.10210931e-01
-3.27813327e-01 9.15743038e-02 -6.24197483e-01 5.21562696e-01
7.80977845e-01 -4.44830030e-01 1.13155246e-01 6.90241098e-01
4.48853225e-01 -4.96946961e-01 -1.26728547e+00 7.98871636e-01
-8.64902362e-02 -8.62551689e-01 -8.66539851e-02 -9.61726829e-02
2.17919424e-01 -6.31460547e-01 -6.06202543e-01 -6.09538592e-02
3.25078703e-02 -6.65866554e-01 9.11531866e-01 1.56085581e-01
6.63224697e-01 -5.49398303e-01 1.19564998e+00 3.97131115e-01
-8.66763890e-01 3.49257529e-01 -3.31722468e-01 -7.52652138e-02
3.90714586e-01 3.75477433e-01 -7.06502795e-01 2.53840953e-01
4.06423807e-01 1.76600277e-01 -2.22567514e-01 8.02906632e-01
-5.97154140e-01 4.77535069e-01 -1.51623964e-01 -6.46862388e-01
1.49880156e-01 -9.97689664e-02 5.89743137e-01 1.05882347e+00
1.32756621e-01 5.14851928e-01 -8.86993425e-04 1.15423346e+00
-1.61921121e-02 6.09584570e-01 -8.73282075e-01 -3.11910599e-01
5.14936268e-01 9.77350056e-01 -8.85462642e-01 -3.50059479e-01
-3.48557085e-01 6.92313373e-01 1.38939455e-01 1.62135929e-01
-4.96356666e-01 -4.61904645e-01 2.58117884e-01 4.86415058e-01
-1.68828830e-01 -1.31319195e-01 -2.28196532e-02 -1.00246334e+00
2.76121289e-01 -9.28410649e-01 1.50017649e-01 -1.01981568e+00
-1.21055984e+00 8.09516609e-01 2.87905693e-01 -1.11383271e+00
-6.95956767e-01 -3.71145576e-01 -2.79312521e-01 9.04675305e-01
-8.70860100e-01 -7.57252157e-01 -9.43368897e-02 4.83793139e-01
6.42140508e-01 8.10234621e-03 1.03113496e+00 2.81361997e-01
-2.58360267e-01 -8.21177587e-02 -4.72792894e-01 2.26541713e-01
5.50773740e-01 -1.14820886e+00 -6.94009066e-02 6.67540133e-01
2.63159096e-01 6.84588909e-01 1.10852730e+00 -5.39470851e-01
-7.60549724e-01 -6.80669188e-01 1.90030706e+00 -2.49809295e-01
9.06099916e-01 -1.16443321e-01 -1.20004845e+00 5.15850484e-01
-2.29484737e-02 -9.98791978e-02 7.88743734e-01 1.44318983e-01
-3.40812393e-02 2.58586437e-01 -1.14320481e+00 5.98576367e-01
8.49317074e-01 -6.03395998e-01 -1.26600718e+00 4.81335104e-01
4.77147490e-01 -3.37989897e-01 -6.84985757e-01 -8.51897821e-02
4.24200855e-02 -5.88934898e-01 2.97187597e-01 -7.15717971e-01
2.47310445e-01 -5.27626872e-01 -7.48370364e-02 -1.11711907e+00
4.89520989e-02 -4.86505866e-01 8.60606670e-01 1.73379397e+00
5.68145812e-01 -2.68643588e-01 1.90006167e-01 1.07956684e+00
9.99291688e-02 -1.77879080e-01 -7.37242579e-01 -6.78515494e-01
-3.18601847e-01 -6.42088830e-01 6.88942075e-01 1.12582982e+00
4.70444620e-01 8.85852993e-01 1.64147556e-01 -2.90822953e-01
3.17967504e-01 5.05065732e-02 5.51617146e-01 -1.44254315e+00
3.26137096e-01 -3.55312258e-01 -4.42834377e-01 -3.30179870e-01
3.66736680e-01 -4.52735722e-01 4.86352772e-01 -1.35350204e+00
-2.63106853e-01 -5.95247030e-01 2.94501811e-01 4.75126356e-01
-6.74669668e-02 -8.04545134e-02 -2.46100896e-03 2.10823134e-01
-4.64705378e-01 3.48462582e-01 1.01036251e+00 3.59115660e-01
-5.02113402e-01 6.79742917e-02 -7.34809279e-01 1.00241959e+00
7.38382280e-01 -7.35875487e-01 -4.63403851e-01 -1.30211502e-01
4.25387412e-01 9.44922715e-02 4.38889027e-01 -5.91306627e-01
-6.10402934e-02 -4.29200023e-01 -2.22004533e-01 -1.93953380e-01
-4.97492328e-02 -1.14866912e+00 4.50813353e-01 -7.76991695e-02
-4.22535747e-01 2.78681159e-01 2.64461096e-02 7.09694698e-02
-5.17905116e-01 -8.52456331e-01 6.41523361e-01 -4.82533276e-01
-6.66336715e-01 -3.88343364e-01 -6.59266353e-01 2.69486427e-01
9.76168036e-01 -2.66115874e-01 9.26831365e-02 -2.52369702e-01
-8.10540795e-01 -7.60107934e-02 7.06851125e-01 2.64434546e-01
3.37341011e-01 -1.28273225e+00 -3.83715272e-01 -1.13069616e-01
2.90425748e-01 2.92035729e-01 -2.75617331e-01 8.31367195e-01
-3.95641208e-01 2.26852283e-01 -1.29873484e-01 -3.56194228e-02
-1.19578123e+00 6.92491770e-01 1.08326137e-01 -1.62875980e-01
-5.55128932e-01 4.14034396e-01 2.79278420e-02 -1.19301639e-01
9.61574838e-02 -2.18006000e-01 -4.96684641e-01 4.41190451e-01
5.40155530e-01 -8.31949711e-02 2.24477146e-02 -1.22074771e+00
-1.78419575e-01 2.22914100e-01 1.74682274e-01 -3.77321571e-01
1.26818907e+00 -2.65073955e-01 -3.85231584e-01 8.91826332e-01
8.84305537e-01 1.32537425e-01 -4.20241058e-01 -4.44414794e-01
5.56420028e-01 -3.85511845e-01 -2.96323478e-01 -5.73182583e-01
-5.88804305e-01 6.23699665e-01 -3.35775286e-01 6.83196068e-01
1.13479888e+00 2.97735155e-01 3.59153360e-01 4.13129538e-01
6.73417687e-01 -1.15249813e+00 -3.99632931e-01 2.04615220e-01
1.13337231e+00 -1.21299911e+00 -2.49039784e-01 -7.05826759e-01
-7.72637963e-01 1.20504177e+00 3.80237520e-01 1.97295800e-01
2.07098216e-01 3.79829109e-01 2.01241270e-01 -4.00759041e-01
-6.86617017e-01 -3.54432970e-01 1.77681342e-01 5.86687982e-01
8.45785975e-01 -8.01200122e-02 -1.13093507e+00 6.69019520e-01
-2.99879313e-01 8.82307589e-02 7.22748220e-01 1.02299953e+00
-2.16836333e-01 -1.34443307e+00 -5.67761242e-01 2.52078176e-01
-8.26590717e-01 4.83398736e-02 -7.55726337e-01 1.13515246e+00
1.99169770e-01 1.10042465e+00 -2.11880319e-02 7.71169886e-02
7.52815545e-01 1.95528090e-01 1.71632931e-01 -7.66672909e-01
-6.64058983e-01 1.48518994e-01 6.41612053e-01 -3.13388377e-01
-9.85021591e-01 -9.17829454e-01 -1.29870307e+00 -2.03193694e-01
-5.74756444e-01 7.01299250e-01 5.35151780e-01 1.30943167e+00
-3.00707877e-01 2.32620895e-01 7.59989843e-02 -4.77237761e-01
-3.17452997e-01 -1.06479907e+00 -7.98566461e-01 7.93465436e-01
-7.08118975e-02 -7.25295901e-01 -2.96491653e-01 5.14877617e-01]
|
[10.324409484863281, 9.16994571685791]
|
9fede14e-0865-45f2-be42-19a414961048
|
bent-broken-bicycles-leveraging-synthetic
|
2304.07883
| null |
https://arxiv.org/abs/2304.07883v1
|
https://arxiv.org/pdf/2304.07883v1.pdf
|
Bent & Broken Bicycles: Leveraging synthetic data for damaged object re-identification
|
Instance-level object re-identification is a fundamental computer vision task, with applications from image retrieval to intelligent monitoring and fraud detection. In this work, we propose the novel task of damaged object re-identification, which aims at distinguishing changes in visual appearance due to deformations or missing parts from subtle intra-class variations. To explore this task, we leverage the power of computer-generated imagery to create, in a semi-automatic fashion, high-quality synthetic images of the same bike before and after a damage occurs. The resulting dataset, Bent & Broken Bicycles (BBBicycles), contains 39,200 images and 2,800 unique bike instances spanning 20 different bike models. As a baseline for this task, we propose TransReI3D, a multi-task, transformer-based deep network unifying damage detection (framed as a multi-label classification task) with object re-identification. The BBBicycles dataset is available at https://huggingface.co/datasets/GrainsPolito/BBBicycles
|
['Fabrizio Lamberti', 'Lia Morra', 'Lorenzo Lanari', 'Alessandro Sebastian Russo', 'Filippo Gabriele Pratticò', 'Luca Piano']
|
2023-04-16
| null | null | null | null |
['fraud-detection']
|
['miscellaneous']
|
[ 2.82109916e-01 -4.22143042e-01 -6.64767623e-02 -2.19938681e-01
-8.31837833e-01 -6.07045650e-01 6.41045153e-01 -3.37875150e-02
1.02998633e-02 4.14456874e-01 2.41180286e-01 2.93050051e-01
1.33627607e-02 -6.28386497e-01 -9.80941534e-01 -6.50434911e-01
1.45599440e-01 4.85556126e-01 1.00726493e-01 -8.44634417e-03
3.51299673e-01 7.35546768e-01 -1.90606427e+00 5.01341283e-01
5.24588525e-01 1.10497582e+00 -7.61351213e-02 6.17416084e-01
5.88851273e-01 4.53937709e-01 -3.75016481e-01 -6.82129204e-01
3.23698938e-01 -1.63903654e-01 -1.07023859e+00 4.90183651e-01
1.10563076e+00 -4.37650561e-01 -4.04916018e-01 1.04330182e+00
5.30582845e-01 2.64895968e-02 9.40912366e-01 -1.53806055e+00
-8.17125618e-01 7.10500479e-02 -8.48499477e-01 5.24817884e-01
3.92468199e-02 3.94563466e-01 7.84962773e-01 -9.28847015e-01
4.65442687e-01 1.31179297e+00 8.87112677e-01 2.70086348e-01
-1.47466516e+00 -6.80907249e-01 -2.54309118e-01 7.69620597e-01
-1.43087137e+00 -5.49750745e-01 8.70812833e-01 -9.30415690e-01
7.87458539e-01 1.80346832e-01 3.65267605e-01 1.27315378e+00
5.51124215e-02 7.49553084e-01 1.19752467e+00 -3.34006518e-01
-2.98073173e-01 -1.06544062e-01 3.16320091e-01 6.35939479e-01
3.46663952e-01 4.01046485e-01 -5.11628747e-01 9.32967663e-02
3.73766452e-01 1.63942605e-01 -2.99082100e-01 -5.19438624e-01
-1.31156611e+00 5.34178674e-01 5.71261048e-01 1.60484299e-01
-3.99836361e-01 2.62705654e-01 6.77911460e-01 3.16180497e-01
5.45699537e-01 1.76240191e-01 -1.12309791e-01 2.04815388e-01
-7.70480216e-01 3.67662758e-01 7.01985285e-02 5.03857791e-01
8.24576318e-01 -9.15030614e-02 -2.29626626e-01 1.15414476e+00
-1.74398702e-02 4.04569477e-01 4.79842782e-01 -7.99777925e-01
4.31635886e-01 4.89305526e-01 2.18433410e-01 -1.22183299e+00
-5.36733508e-01 -2.26715073e-01 -1.05585706e+00 3.71497035e-01
3.94681603e-01 4.12623197e-01 -1.03358138e+00 1.43978083e+00
5.26305914e-01 2.13232189e-02 -3.60365599e-01 9.65672374e-01
7.28789389e-01 1.70684710e-01 -6.09807931e-02 3.49917620e-01
1.58982956e+00 -1.12338281e+00 -1.42490551e-01 -1.81122750e-01
3.83160233e-01 -4.36401784e-01 9.69196081e-01 6.50409579e-01
-8.12531650e-01 -7.17343152e-01 -9.18634534e-01 -1.11602031e-01
-5.25261879e-01 3.37046981e-01 -9.15036276e-02 4.30116475e-01
-8.66031170e-01 4.86106426e-01 -4.27080780e-01 -5.47203362e-01
6.31748557e-01 -1.00769408e-01 -7.75284469e-01 -1.74191356e-01
-8.76758933e-01 1.18315089e+00 5.27688980e-01 1.90942705e-01
-1.50957704e+00 -7.32506931e-01 -7.93122590e-01 -1.35389492e-01
3.30394864e-01 -5.17729700e-01 1.01237023e+00 -8.36021543e-01
-8.21127892e-01 1.61697304e+00 1.86102211e-01 -4.70045835e-01
5.90176940e-01 -1.66732550e-01 -5.72281420e-01 1.63163468e-01
2.09186435e-01 5.16353667e-01 1.31889927e+00 -1.50583124e+00
-5.22580862e-01 -6.37817800e-01 -2.62832284e-01 -2.08940253e-01
-9.64565575e-02 3.25316995e-01 -5.33780223e-03 -1.02447021e+00
-2.36509189e-01 -9.64970350e-01 5.18408716e-01 1.59027964e-01
-7.56372511e-01 -2.46429682e-01 6.90283239e-01 -1.27345479e+00
9.20008242e-01 -2.14930010e+00 3.40458423e-01 3.66434194e-02
3.30448031e-01 2.85503626e-01 -4.09228474e-01 2.88083345e-01
-6.19183838e-01 -1.03054345e-01 -4.78283048e-01 -4.51654673e-01
2.01970078e-02 -1.70706496e-01 -2.61135370e-01 6.81610465e-01
2.94862509e-01 1.15147400e+00 -5.89053690e-01 -1.66494802e-01
1.27835900e-01 9.15934965e-02 -5.38269505e-02 1.38645574e-01
8.55982900e-02 4.52284247e-01 2.69927353e-01 1.04462612e+00
7.49861836e-01 2.72898972e-02 -2.78232187e-01 -7.65022099e-01
-4.22085356e-03 -9.37271118e-02 -9.30184543e-01 1.53424561e+00
-3.03840011e-01 5.56684375e-01 3.16921063e-02 -1.44094849e+00
8.05342078e-01 9.69778821e-02 5.61830223e-01 -8.46452117e-01
1.01819448e-02 1.29706815e-01 -4.29617763e-01 -7.05281138e-01
5.19480109e-01 -1.71791419e-01 -1.82941034e-01 8.26577425e-01
-1.09962476e-02 -9.14406702e-02 1.82437629e-01 -7.97773004e-02
9.59456921e-01 -4.18655574e-02 4.01660874e-02 -1.37701482e-01
3.90054435e-01 5.98870628e-02 2.95406461e-01 4.51477021e-01
-4.63021964e-01 8.81414771e-01 1.01579137e-01 -7.02756703e-01
-1.35525787e+00 -9.55629647e-01 -1.36947513e-01 8.52185130e-01
1.42895535e-01 2.16283873e-01 -5.53286254e-01 -8.11984956e-01
5.58212399e-01 4.62762088e-01 -8.79727781e-01 -5.14718711e-01
-4.86411721e-01 -9.46430922e-01 7.61395156e-01 4.31263268e-01
7.10590839e-01 -1.22289407e+00 -4.33832616e-01 7.88101852e-02
-5.67481697e-01 -7.23892331e-01 -6.83956325e-01 -1.69843420e-01
-3.08168024e-01 -1.46521270e+00 -9.02006805e-01 -6.67041361e-01
3.42808425e-01 5.21926403e-01 1.29856086e+00 1.16238199e-01
-9.94123340e-01 6.24715149e-01 -3.04053247e-01 -1.49138600e-01
-4.25473571e-01 -5.64953014e-02 1.84724644e-01 4.57336277e-01
1.39862508e-01 -5.77126324e-01 -7.30039835e-01 5.61507344e-01
-9.86016512e-01 -1.28656417e-01 3.37420404e-01 9.64354217e-01
5.52983046e-01 1.25622451e-01 7.59071112e-01 -2.05544621e-01
2.72728711e-01 -4.85072941e-01 -2.73363739e-01 5.58742881e-01
-6.22364044e-01 -3.03824544e-01 -5.52820675e-02 -3.31991047e-01
-9.85876441e-01 -5.58324680e-02 2.50955671e-01 -5.19384563e-01
-5.11942208e-01 3.02399218e-01 -2.13561103e-01 -1.94319144e-01
7.97157466e-01 3.13809007e-01 9.45416465e-02 -8.66970897e-01
4.27197963e-01 8.66261601e-01 1.12075806e+00 -4.11227643e-01
7.89614737e-01 5.27129769e-01 -2.48316139e-01 -5.38029730e-01
-8.80085051e-01 -7.11601734e-01 -9.64234233e-01 -3.87868941e-01
7.91919470e-01 -9.89088833e-01 -5.60647607e-01 1.51515245e+00
-1.08648813e+00 -6.43823206e-01 -2.03086585e-01 -2.04549834e-01
-6.01612508e-01 6.71784997e-01 -3.32063168e-01 -5.22821426e-01
-4.82283443e-01 -7.60118723e-01 1.37371802e+00 -1.44498050e-01
5.62258959e-02 -5.23255885e-01 2.70723522e-01 8.79189193e-01
4.82283309e-02 5.90622425e-01 9.84985590e-01 -2.24315763e-01
-2.75487691e-01 -1.93363801e-01 -5.95647275e-01 6.36016726e-01
4.13406432e-01 -2.30589256e-01 -9.27901864e-01 -5.32875419e-01
-3.56326550e-01 -7.73041010e-01 1.22387588e+00 6.20238250e-03
1.31298232e+00 -3.04372966e-01 -2.95643091e-01 5.17802417e-01
1.28146267e+00 -1.57427773e-01 9.20661747e-01 6.68107688e-01
1.04479325e+00 8.32618177e-01 6.00735068e-01 3.44471842e-01
6.14094198e-01 1.13483679e+00 7.69827485e-01 -6.39670566e-02
-7.47740448e-01 1.13743946e-01 3.63178939e-01 1.73998237e-01
9.47775915e-02 -2.66160280e-01 -1.12310565e+00 1.18802786e+00
-1.57158852e+00 -1.22658813e+00 -3.10949534e-01 2.06550956e+00
8.33481312e-01 -3.36622417e-01 5.26421309e-01 3.88483495e-01
1.09336853e+00 7.54961446e-02 -9.03406680e-01 -4.17367145e-02
-2.54960716e-01 -2.76198536e-01 4.07340735e-01 1.31153807e-01
-1.47556531e+00 6.02253377e-01 5.02203989e+00 7.84034252e-01
-9.75629687e-01 3.96777958e-01 6.67763472e-01 -1.04699053e-01
1.55339137e-01 -5.40871263e-01 -6.09852016e-01 7.80506074e-01
6.07334137e-01 6.65204301e-02 5.59396863e-01 4.06748116e-01
1.57793965e-02 -1.92892969e-01 -9.81096268e-01 1.08001506e+00
5.86181998e-01 -9.83208477e-01 -3.16034406e-02 -1.16049223e-01
7.82746255e-01 2.08929017e-01 1.24941289e-01 1.10517330e-02
1.89318582e-01 -9.36269104e-01 1.08942568e+00 8.63981664e-01
1.15809643e+00 -6.72032177e-01 4.20439094e-01 1.83284953e-01
-1.16450143e+00 -2.72351295e-01 1.10594397e-02 3.56952429e-01
1.71570763e-01 5.39877594e-01 -6.51932180e-01 7.40166485e-01
1.34301364e+00 9.89307702e-01 -1.00915480e+00 1.30055416e+00
5.76057611e-03 3.93385857e-01 1.35029647e-02 9.41579878e-01
-2.75196880e-01 -1.36419870e-02 6.57521307e-01 1.11545694e+00
2.47552499e-01 -2.75573939e-01 8.17125067e-02 8.11737597e-01
-1.50702000e-01 -5.08104324e-01 -5.02682149e-01 2.20913962e-01
1.01225279e-01 1.25088060e+00 -5.07864118e-01 -3.74219686e-01
-9.76800472e-02 1.48479664e+00 3.50729793e-01 1.02857217e-01
-7.98496723e-01 -3.48130226e-01 8.88849139e-01 1.46552831e-01
4.12451625e-01 1.39205724e-01 -1.00956105e-01 -1.25728250e+00
1.73361823e-01 -9.26917791e-01 5.86342335e-01 -1.07119262e+00
-1.87966740e+00 2.75653839e-01 9.21639986e-03 -1.16529834e+00
3.90379131e-02 -5.78683555e-01 -5.48527360e-01 7.13576138e-01
-1.59530008e+00 -1.81232738e+00 -8.12489688e-01 5.64751267e-01
5.28895617e-01 -1.78205203e-02 4.41185445e-01 6.80232346e-01
-8.40580225e-01 7.67326415e-01 1.95042193e-01 1.72187880e-01
1.04137552e+00 -1.04428351e+00 7.00812340e-01 9.31466877e-01
6.42916858e-02 -2.30059370e-01 3.89597625e-01 -7.30184615e-01
-7.66457736e-01 -1.61066115e+00 7.62429893e-01 -8.41312647e-01
6.33361280e-01 -1.71894923e-01 -1.17892301e+00 5.84444523e-01
-2.17858404e-01 2.18059659e-01 2.24506676e-01 -4.00571823e-01
-6.17079914e-01 -1.30606756e-01 -1.26746440e+00 4.76712659e-02
1.33891034e+00 -8.43518674e-01 -4.91559774e-01 4.83013809e-01
3.33001435e-01 -2.08855584e-01 -9.61639404e-01 5.18547356e-01
5.99828303e-01 -1.01283705e+00 1.45325792e+00 -6.40366495e-01
4.80995148e-01 -4.05786663e-01 -3.12805474e-02 -1.38775039e+00
-5.02453268e-01 -1.14913642e-01 -1.10106811e-01 1.39707661e+00
-1.82648316e-01 -5.19424558e-01 4.03476536e-01 4.44938779e-01
-3.68705124e-01 -2.71325469e-01 -1.08831632e+00 -1.07098484e+00
9.14776772e-02 -3.59282464e-01 8.23860526e-01 9.73768353e-01
-4.42072004e-01 -2.17336297e-01 -6.60854280e-01 2.12332383e-01
7.28304505e-01 3.40705425e-01 8.31249237e-01 -1.45101106e+00
-1.36611909e-02 -4.94321585e-01 -5.99941730e-01 -4.60262686e-01
2.89469779e-01 -1.13220119e+00 2.01056570e-01 -1.24434161e+00
5.25140882e-01 -5.72108746e-01 -3.28247726e-01 8.82025361e-01
-7.33066574e-02 1.03067005e+00 7.27883652e-02 5.11009693e-01
-3.70617241e-01 7.05398142e-01 9.11834300e-01 -7.38643646e-01
2.96539992e-01 -1.43666774e-01 -4.76231903e-01 4.98073846e-01
7.69173026e-01 -4.50678498e-01 2.86272049e-01 -5.32161534e-01
2.55175773e-02 -2.70963103e-01 1.49580848e+00 -9.46113646e-01
-2.24448189e-01 7.02808127e-02 1.03210449e-01 -7.26194322e-01
1.88015386e-01 -7.15696514e-01 4.07251656e-01 3.73518646e-01
-2.15304166e-01 -3.35255973e-02 1.57920450e-01 7.23115802e-01
-1.36150628e-01 -3.67526889e-01 8.50701571e-01 2.16604527e-02
-9.71212387e-01 4.00437236e-01 -2.36642808e-01 -1.59160674e-01
1.14109790e+00 -2.99561173e-01 -5.64416170e-01 1.05336353e-01
-7.72383273e-01 -9.83903650e-03 6.77820742e-01 7.00654447e-01
4.73236471e-01 -1.63321602e+00 -9.54427898e-01 8.28947574e-02
7.01334119e-01 -1.08546622e-01 7.36202717e-01 8.69359076e-01
-2.01962978e-01 1.72087625e-01 -4.91726637e-01 -5.44424176e-01
-1.41072834e+00 6.85954452e-01 6.16703868e-01 -1.26528487e-01
-7.08628893e-01 8.00873637e-01 1.77395388e-01 -4.02051508e-01
-9.35355481e-03 -1.13108940e-01 -2.82038450e-01 3.94516945e-01
4.93268937e-01 7.94447541e-01 5.53392231e-01 -1.08630276e+00
-2.57351667e-01 5.51220417e-01 3.89124714e-02 2.79013813e-01
1.29893625e+00 -5.08204937e-01 -2.36178502e-01 1.45134330e-01
1.07291365e+00 -6.29122496e-01 -1.39758456e+00 -3.80299568e-01
2.18507260e-01 -5.91378212e-01 6.46837130e-02 -1.17699218e+00
-1.29388213e+00 7.43924081e-01 1.08235395e+00 8.57592449e-02
1.07162941e+00 2.26826936e-01 1.02372766e+00 1.68251425e-01
4.94281590e-01 -9.17725146e-01 5.48087299e-01 2.03751728e-01
1.48743713e+00 -1.58006728e+00 -1.53483003e-01 3.62387300e-02
-7.03209579e-01 7.95774162e-01 6.21160984e-01 -1.19228885e-02
3.80784243e-01 -3.76976699e-01 -1.45052224e-01 -5.47115266e-01
-2.34634504e-01 -2.87111819e-01 5.18497169e-01 8.77247214e-01
-3.72009814e-01 1.31740123e-01 3.01696301e-01 4.89048988e-01
5.74439578e-03 -9.11223367e-02 4.16200131e-01 5.24534166e-01
-2.41298035e-01 -8.48436177e-01 -6.43273354e-01 5.49742877e-01
-1.09377906e-01 1.04003973e-01 -4.21909720e-01 3.73362720e-01
4.80050176e-01 9.60932493e-01 1.47335351e-01 -4.07305717e-01
5.25134087e-01 1.09009996e-01 4.63024735e-01 -5.11101067e-01
-4.82091814e-01 -6.37460113e-01 2.01039508e-01 -4.76255238e-01
-6.08429253e-01 -1.06181252e+00 -5.03313243e-01 -2.05758989e-01
-3.04273874e-01 -5.45992970e-01 5.31324387e-01 8.47087860e-01
4.94289249e-01 2.88716435e-01 7.67885387e-01 -1.38198674e+00
-4.65743601e-01 -1.07829666e+00 -6.86790347e-01 8.15286577e-01
5.58355510e-01 -1.12914383e+00 -2.25433737e-01 5.49885273e-01]
|
[14.654884338378906, 0.9732877016067505]
|
acf23301-687c-4297-a356-711c2dd0ac87
|
people-talking-and-ai-listening-how
|
2305.10201
| null |
https://arxiv.org/abs/2305.10201v4
|
https://arxiv.org/pdf/2305.10201v4.pdf
|
Echoes of Biases: How Stigmatizing Language Affects AI Performance
|
Electronic health records (EHRs) serve as an essential data source for the envisioned artificial intelligence (AI)-driven transformation in healthcare. However, clinician biases reflected in EHR notes can lead to AI models inheriting and amplifying these biases, perpetuating health disparities. This study investigates the impact of stigmatizing language (SL) in EHR notes on mortality prediction using a Transformer-based deep learning model and explainable AI (XAI) techniques. Our findings demonstrate that SL written by clinicians adversely affects AI performance, particularly so for black patients, highlighting SL as a source of racial disparity in AI model development. To explore an operationally efficient way to mitigate SL's impact, we investigate patterns in the generation of SL through a clinicians' collaborative network, identifying central clinicians as having a stronger impact on racial disparity in the AI model. We find that removing SL written by central clinicians is a more efficient bias reduction strategy than eliminating all SL in the entire corpus of data. This study provides actionable insights for responsible AI development and contributes to understanding clinician behavior and EHR note writing in healthcare.
|
['Ritu Agarwal', 'Guodong Gordon Gao', 'Weiguang Wang', 'Yizhi Liu']
|
2023-05-17
| null | null | null | null |
['mortality-prediction']
|
['medical']
|
[ 2.84867138e-01 7.93017447e-01 -1.75973386e-01 -3.38522851e-01
-7.45366752e-01 -3.21528733e-01 3.75260979e-01 5.40110707e-01
-3.53871882e-01 3.86016876e-01 1.39356399e+00 -1.01478457e+00
-9.07719582e-02 -5.08462608e-01 -6.24541104e-01 -3.46426666e-01
3.76081288e-01 5.38692534e-01 -9.93180633e-01 8.49485844e-02
2.62392998e-01 4.90258545e-01 -3.69002521e-01 6.81585193e-01
1.08354104e+00 2.32826754e-01 -5.79927683e-01 3.33958477e-01
1.95560814e-03 1.81664121e+00 -8.04588675e-01 -7.56185293e-01
3.59118551e-01 -5.95380366e-01 -5.47865152e-01 -4.45893645e-01
6.63430870e-01 -4.78974611e-01 -4.79600191e-01 7.74791420e-01
7.60699570e-01 -7.35038817e-01 9.20752108e-01 -8.96859109e-01
-1.17406762e+00 1.41447937e+00 -9.76380855e-02 1.98205695e-01
3.78070362e-02 6.65132701e-01 6.91830754e-01 -5.20471156e-01
7.55864263e-01 1.18012762e+00 1.17954063e+00 6.03485107e-01
-1.30221879e+00 -1.03319573e+00 -1.47199675e-01 -1.85817629e-01
-1.15235114e+00 -7.28842676e-01 7.30474532e-01 -9.94934797e-01
7.14706182e-01 4.55490321e-01 1.07983124e+00 1.28429413e+00
4.90631223e-01 5.96699059e-01 8.94322217e-01 -3.89215827e-01
7.03937113e-02 5.22416770e-01 3.98815900e-01 6.76336348e-01
8.94268990e-01 8.94270912e-02 -4.54835713e-01 -9.70002711e-01
6.42641246e-01 3.57707322e-01 -2.55944312e-01 3.12973469e-01
-1.26020932e+00 8.41527522e-01 5.54729402e-01 1.59823045e-01
-6.68743134e-01 9.03060474e-03 1.69414461e-01 2.39596307e-01
2.72586495e-01 1.07611907e+00 -2.43783191e-01 -1.55906066e-01
-7.48673379e-01 1.05018131e-01 7.85134315e-01 6.09221458e-01
-3.56201790e-02 1.87785566e-01 -7.30405092e-01 5.42470336e-01
3.17952894e-02 7.13756084e-01 2.62275636e-01 -1.03893793e+00
3.41417104e-01 9.92018163e-01 6.45551011e-02 -1.03233409e+00
-4.63626146e-01 -6.52981579e-01 -9.70812440e-01 -3.10741872e-01
6.20811939e-01 -4.76147979e-01 -8.18414271e-01 1.53072631e+00
-1.74863636e-01 -3.03633809e-01 2.08742861e-02 6.83886647e-01
6.03837252e-01 1.58272013e-02 6.24702215e-01 -4.62969160e-03
1.34261203e+00 -4.27397162e-01 -8.25872004e-01 -3.18547279e-01
1.21146107e+00 -5.59218526e-01 9.34125304e-01 1.73404634e-01
-1.08860159e+00 1.28839582e-01 -4.73557353e-01 -2.22540960e-01
1.29612371e-01 -7.72099793e-02 3.31416637e-01 6.63172781e-01
-9.02678549e-01 2.39457041e-01 -6.40416980e-01 -2.21527696e-01
1.32446599e+00 1.28020585e-01 -2.86474195e-03 -1.27510518e-01
-9.22277927e-01 1.07630181e+00 -5.97697914e-01 -7.92684928e-02
-6.25690699e-01 -1.45403862e+00 -6.53139114e-01 2.59397149e-01
1.68403178e-01 -1.24920654e+00 1.12225282e+00 -1.27872694e+00
-5.47406435e-01 8.61667693e-01 -2.27984443e-01 -6.77530527e-01
6.76766753e-01 -2.97355056e-01 -3.02983224e-01 -2.36609623e-01
1.70813482e-02 3.07883352e-01 6.87320352e-01 -1.23411214e+00
-1.18195243e-01 -7.41956115e-01 -7.47309506e-01 -1.68695614e-01
-6.41847908e-01 -2.85220370e-02 8.00655067e-01 -9.17859495e-01
-5.16021013e-01 -9.08832133e-01 -4.52819616e-01 -1.23663489e-02
-5.93891442e-01 -1.18731745e-01 3.08267009e-02 -1.05766904e+00
1.56790984e+00 -2.11387992e+00 -3.35752130e-01 3.74814391e-01
1.15414178e+00 4.27084655e-01 1.24269314e-01 4.25958574e-01
8.75410140e-02 6.05172157e-01 -1.46981806e-01 -3.28356177e-02
-3.37351471e-01 -9.09322351e-02 -2.48345152e-01 3.13035935e-01
4.49134588e-01 1.18396342e+00 -7.32042789e-01 -4.43228692e-01
-2.16004923e-01 7.22050786e-01 -9.60539639e-01 2.11284906e-01
2.25153506e-01 4.77359802e-01 -3.70583236e-01 7.34174907e-01
3.84763628e-01 -6.03639722e-01 4.04232353e-01 1.83739513e-03
-1.32455617e-01 5.84009767e-01 -1.53275087e-01 8.38466704e-01
-7.94888511e-02 6.67876124e-01 -1.00987561e-01 -5.13076186e-01
9.15827930e-01 2.07710281e-01 5.12669861e-01 -6.22635007e-01
2.25626037e-01 1.50095969e-01 7.22661972e-01 -6.80126190e-01
1.62240744e-01 -4.03309107e-01 1.13448054e-01 5.68037450e-01
-8.49372923e-01 1.41775787e-01 -7.35378027e-01 4.95534688e-01
1.53822291e+00 -9.80600715e-01 3.04450750e-01 -5.39847732e-01
-1.38759911e-01 6.01988196e-01 7.81147599e-01 1.02949929e+00
-5.79925895e-01 3.35577130e-01 5.80661952e-01 -8.96028221e-01
-1.34531391e+00 -7.38509893e-01 -3.15946966e-01 3.83765161e-01
-7.16091514e-01 -2.05967903e-01 -6.89983189e-01 -5.05890369e-01
5.55483699e-01 1.25361872e+00 -7.85361707e-01 -7.24885166e-01
-5.79383552e-01 -5.31245112e-01 1.15617776e+00 5.97789586e-01
4.67342064e-02 -9.80299175e-01 -1.00630462e+00 8.85809809e-02
-8.42063054e-02 -7.01896787e-01 -6.94145083e-01 -3.08875203e-01
-8.04259300e-01 -1.15348339e+00 -8.38864803e-01 -3.13985616e-01
9.01607275e-01 -1.82691589e-01 1.38010490e+00 2.82688320e-01
-6.72349334e-01 4.62512255e-01 8.45909715e-02 -1.15716124e+00
-1.05558920e+00 -4.02955525e-02 -1.18234448e-01 -4.55274999e-01
9.91446137e-01 1.82185888e-01 -9.57436144e-01 -3.51961762e-01
-7.47460365e-01 1.15027927e-01 6.22604251e-01 8.60373318e-01
-1.70888662e-01 -7.35449135e-01 5.92548132e-01 -1.71189284e+00
9.73982990e-01 -6.37996018e-01 5.16987853e-02 2.04072103e-01
-1.17559266e+00 7.10424632e-02 6.02057099e-01 -3.89425188e-01
-9.65302169e-01 -2.72038043e-01 1.34415373e-01 -1.80129722e-01
-8.58643204e-02 4.35087621e-01 5.63480437e-01 2.92048424e-01
9.88477468e-01 -3.08227986e-01 3.42571855e-01 -4.02741879e-01
-1.24340050e-01 9.93571877e-01 -1.26432315e-01 -9.36484784e-02
4.01939929e-01 6.24580145e-01 -4.11976218e-01 -7.02219486e-01
-8.44521880e-01 -8.17354769e-02 -1.15433916e-01 1.04431003e-01
1.06474435e+00 -1.13049877e+00 -7.30111063e-01 1.22575805e-01
-8.44651461e-01 -4.30242330e-01 -4.34788465e-01 6.31039679e-01
2.31555015e-01 -2.65986413e-01 -8.23708594e-01 -9.77468908e-01
-6.51653528e-01 -9.85316634e-01 7.70908535e-01 3.42668407e-02
-1.17534232e+00 -7.09318519e-01 3.34637016e-01 7.83072352e-01
6.04538441e-01 3.53032351e-02 1.64401031e+00 -9.89983380e-01
-3.65300775e-01 8.94563869e-02 -4.29335922e-01 2.42564231e-02
1.89695686e-01 -4.51268107e-02 -8.55018914e-01 -6.71304241e-02
3.80485728e-02 -1.27856871e-02 6.63232207e-01 5.79091966e-01
9.97016430e-01 -9.14397657e-01 -5.09701550e-01 6.67345285e-01
1.10737085e+00 5.89157045e-01 4.69536453e-01 1.28247097e-01
1.09113646e+00 5.14659524e-01 -1.44009709e-01 6.41030073e-01
7.13584840e-01 -1.17409870e-01 -3.25428307e-01 -5.56213915e-01
-3.70896906e-02 -4.44500357e-01 1.44928172e-01 8.85951936e-01
1.08520232e-01 7.22958371e-02 -1.70341635e+00 7.52277672e-01
-1.40404642e+00 -7.17140973e-01 -2.91636646e-01 1.93275845e+00
9.10283923e-01 -2.62338579e-01 2.87502054e-02 -3.15227449e-01
2.75848210e-01 -2.36495912e-01 -8.23372602e-01 -5.84390879e-01
-2.56172299e-01 7.28356792e-03 7.72180676e-01 3.37705433e-01
-3.29857051e-01 4.48197603e-01 7.00045395e+00 -3.84902894e-01
-1.02917373e+00 8.56848899e-03 1.27558601e+00 -3.84122074e-01
-1.11276400e+00 -3.73251438e-01 -3.52515131e-01 4.38362151e-01
1.06745398e+00 -4.18613583e-01 2.07101062e-01 6.51111186e-01
5.15698314e-01 4.17426586e-01 -1.28804326e+00 7.43622959e-01
2.43341923e-01 -1.66341400e+00 4.79322433e-01 4.11213845e-01
8.35631073e-01 2.25124066e-03 4.46657091e-01 -3.04023456e-02
7.85066485e-01 -1.41786313e+00 5.19558311e-01 7.69679785e-01
9.06968415e-01 -4.33515579e-01 8.33131611e-01 -5.68335468e-04
-8.96175355e-02 -5.61848044e-01 -1.40262201e-01 -2.40535423e-01
-1.13187291e-01 6.64712310e-01 -1.32863581e+00 -2.66381562e-01
5.78386128e-01 5.70259154e-01 -6.03079438e-01 4.62870091e-01
5.43176830e-01 1.15846622e+00 1.71970263e-01 2.12344795e-01
7.50441551e-02 9.96692851e-02 5.33263326e-01 1.21661627e+00
2.20330387e-01 5.45654655e-01 -3.59201372e-01 1.15049505e+00
-4.34121251e-01 1.02605678e-01 -1.17961144e+00 -8.07220459e-01
6.18856132e-01 6.07869744e-01 -1.91255853e-01 -6.44404769e-01
-4.33327973e-01 4.49503541e-01 3.15213591e-01 3.84380311e-01
-3.95172179e-01 1.82806507e-01 9.34510827e-01 7.23098814e-01
-2.67980397e-01 1.90930814e-01 -1.14991498e+00 -1.03691900e+00
-4.56653059e-01 -1.57677996e+00 6.04665816e-01 -5.41873038e-01
-1.46517372e+00 3.08151305e-01 -7.04601049e-01 -5.63792109e-01
-2.26090141e-02 -2.30297834e-01 -8.01458284e-02 8.72852802e-01
-1.27978849e+00 -9.50368047e-01 -2.07623869e-01 3.97256166e-01
1.78501323e-01 -4.34183359e-01 8.05625439e-01 3.47249568e-01
-7.32579887e-01 1.09916222e+00 -6.73658699e-02 4.80700344e-01
1.03773105e+00 -8.55208635e-01 3.47704619e-01 3.70386660e-01
-1.25821561e-01 1.10551834e+00 3.80109042e-01 -1.17601812e+00
-1.40333533e+00 -1.05423832e+00 1.31588733e+00 -1.17633998e+00
4.05911386e-01 1.31590879e-02 -9.87592280e-01 1.16536820e+00
1.80285349e-02 -8.05686772e-01 1.26366127e+00 1.48559287e-01
-5.66213965e-01 1.62680075e-01 -1.17498922e+00 9.14903045e-01
1.20056486e+00 -7.37725735e-01 -6.46782875e-01 9.72542688e-02
8.46446276e-01 2.38431051e-01 -1.03683686e+00 2.47544065e-01
6.31365538e-01 -7.50001132e-01 7.65299380e-01 -1.20970297e+00
1.20774126e+00 4.42414016e-01 4.46502209e-01 -1.28288913e+00
-9.06074703e-01 -6.07734621e-01 1.89435571e-01 7.49379635e-01
6.39097631e-01 -8.66484821e-01 6.32815540e-01 1.52787721e+00
-5.64336516e-02 -5.81100941e-01 -5.51670492e-01 7.81619698e-02
4.90509301e-01 6.90328181e-02 9.17811453e-01 1.67503452e+00
2.62462925e-02 5.48991933e-02 -1.34377226e-01 -5.62252514e-02
6.89366639e-01 -2.54333019e-01 5.11992037e-01 -1.23130929e+00
-2.52843238e-02 -6.00350440e-01 -2.60779168e-02 -1.90715685e-01
-2.39520937e-01 -8.83863747e-01 -3.99557024e-01 -1.54534686e+00
6.95715070e-01 -8.81782353e-01 -2.95744151e-01 5.70190549e-01
-5.44417024e-01 -2.47059599e-01 4.63990867e-01 5.76505661e-01
9.20773894e-02 -1.24050654e-03 1.14966416e+00 -3.16947401e-01
-2.26273835e-01 -4.18040901e-01 -1.62446618e+00 6.02873921e-01
6.31147981e-01 -7.79948175e-01 -1.44905807e-03 -9.13650393e-01
4.34127837e-01 8.18057507e-02 3.77636701e-01 -3.27232927e-01
1.93495303e-01 -1.71295881e-01 4.99683917e-01 -2.97547155e-03
-2.61810422e-01 -9.21263814e-01 2.77656943e-01 1.15606308e+00
-1.07333624e+00 3.74898881e-01 1.01878546e-01 1.51518241e-01
3.04753244e-01 3.64215106e-01 4.73625541e-01 -3.17697734e-01
4.65763003e-01 -8.05557445e-02 -7.90881336e-01 5.02046466e-01
6.37267113e-01 1.46247119e-01 -5.29820919e-01 -3.39154452e-01
-3.70610893e-01 3.67563516e-02 6.63287759e-01 1.46204531e-01
4.68177915e-01 -9.35248375e-01 -1.14708245e+00 4.90975678e-01
8.87088701e-02 -1.56165898e-01 2.37611204e-01 9.74529088e-01
-5.90647876e-01 5.66923141e-01 -1.87502559e-02 -7.31921792e-02
-1.21336925e+00 4.00656074e-01 4.48444307e-01 -1.10780515e-01
-8.53674889e-01 9.45432127e-01 2.57802278e-01 -1.07063048e-01
2.94019341e-01 -5.26677728e-01 8.99282470e-02 -1.49676889e-01
5.16245425e-01 6.25809908e-01 -2.86863863e-01 -3.44922125e-01
-3.46136510e-01 -1.74321439e-02 -3.97782892e-01 2.42390513e-01
1.33946252e+00 1.70947328e-01 -1.54137418e-01 4.85163718e-01
7.80058861e-01 2.67845631e-01 -7.22663939e-01 -8.03722069e-03
4.50216569e-02 -5.34850419e-01 -3.89471874e-02 -1.16206157e+00
-8.17613006e-01 7.85990834e-01 3.87805998e-01 -1.08657286e-01
6.17370307e-01 -1.85119957e-02 9.92351770e-01 1.38773158e-01
-1.15911037e-01 -8.43355000e-01 -1.97285622e-01 -1.80789724e-01
6.85547590e-01 -1.33783460e+00 -5.89975864e-02 -4.85115275e-02
-1.16649389e+00 4.15873915e-01 5.53100288e-01 1.59435257e-01
6.46926641e-01 4.84286040e-01 7.76457906e-01 -4.93179649e-01
-8.79881322e-01 6.85020566e-01 1.54026679e-03 4.22961622e-01
8.94523323e-01 5.33738971e-01 -3.71978700e-01 6.79832697e-01
-2.74598897e-01 3.88782889e-01 6.29106224e-01 6.26979172e-01
2.28003487e-01 -7.31635809e-01 -4.50890243e-01 1.25959623e+00
-7.08835900e-01 -7.02922940e-01 -1.05406153e+00 3.48959178e-01
1.55788347e-01 6.04261756e-01 4.39517319e-01 -2.79734492e-01
2.54458398e-01 4.17521924e-01 -7.19463220e-03 -5.18343210e-01
-1.39637733e+00 -3.09039056e-01 3.74489158e-01 -3.35875392e-01
2.49451295e-01 -7.54134357e-01 -9.81561959e-01 -5.61937332e-01
1.45975426e-01 -1.02921315e-01 1.77002758e-01 6.60567582e-01
1.10450637e+00 7.40929127e-01 1.67442292e-01 4.55597609e-01
-8.03086817e-01 -7.93310046e-01 -1.18871212e-01 8.54806304e-01
5.51819086e-01 7.63609409e-02 -3.01285475e-01 1.35408361e-02]
|
[7.969786643981934, 6.172063827514648]
|
b47327e3-6bee-4897-bfff-32d993c30e37
|
unsupervised-3d-human-mesh-recovery-from
|
2107.07539
| null |
https://arxiv.org/abs/2107.07539v2
|
https://arxiv.org/pdf/2107.07539v2.pdf
|
Self-supervised 3D Human Mesh Recovery from Noisy Point Clouds
|
This paper presents a novel self-supervised approach to reconstruct human shape and pose from noisy point cloud data. Relying on large amount of dataset with ground-truth annotations, recent learning-based approaches predict correspondences for every vertice on the point cloud; Chamfer distance is usually used to minimize the distance between a deformed template model and the input point cloud. However, Chamfer distance is quite sensitive to noise and outliers, thus could be unreliable to assign correspondences. To address these issues, we model the probability distribution of the input point cloud as generated from a parametric human model under a Gaussian Mixture Model. Instead of explicitly aligning correspondences, we treat the process of correspondence search as an implicit probabilistic association by updating the posterior probability of the template model given the input. A novel self-supervised loss is further derived which penalizes the discrepancy between the deformed template and the input point cloud conditioned on the posterior probability. Our approach is very flexible, which works with both complete point cloud and incomplete ones including even a single depth image as input. Compared to previous self-supervised methods, our method shows the capability to deal with substantial noise and outliers. Extensive experiments conducted on various public synthetic datasets as well as a very noisy real dataset (i.e. CMU Panoptic) demonstrate the superior performance of our approach over the state-of-the-art methods.
|
['Li Cheng', 'Minglun Gong', 'Qiang Sun', 'Sen Wang', 'Xinxin Zuo']
|
2021-07-15
| null | null | null | null |
['human-mesh-recovery']
|
['computer-vision']
|
[-2.26233900e-02 1.17708474e-01 2.20079333e-01 -2.44380385e-01
-9.66446579e-01 -2.17522696e-01 5.66397965e-01 1.99983373e-01
-3.25251877e-01 4.79989231e-01 -3.27537626e-01 4.69811410e-01
-1.00932360e-01 -8.79782081e-01 -1.00773144e+00 -5.63426733e-01
2.32263252e-01 1.42656994e+00 4.81117964e-01 -4.53350646e-03
2.04143345e-01 5.70525646e-01 -1.73839617e+00 -1.30500540e-01
1.03674471e+00 8.47744286e-01 2.46240586e-01 8.91293064e-02
-1.09123312e-01 -2.08018675e-01 -5.23677170e-01 -5.37231982e-01
4.49789226e-01 6.46290481e-02 -4.39357311e-01 4.05025929e-01
5.40042281e-01 -5.32189338e-03 -4.56634797e-02 1.18520188e+00
4.95008141e-01 -1.76020414e-02 9.42246258e-01 -1.32039869e+00
-1.37099370e-01 -9.67937708e-02 -7.26208210e-01 -5.30533850e-01
5.77733994e-01 1.71961468e-02 7.57632375e-01 -1.26232302e+00
7.60853946e-01 1.05953300e+00 9.13463116e-01 2.05301598e-01
-1.30124116e+00 -6.66830182e-01 -2.09289923e-01 -2.02926458e-03
-1.73284805e+00 -1.02707088e-01 1.08657229e+00 -7.50401616e-01
3.29414129e-01 1.58736527e-01 8.38248193e-01 9.07995880e-01
9.03201252e-02 6.12325013e-01 9.21743512e-01 -3.60645711e-01
3.60302299e-01 1.77109037e-02 -3.18197429e-01 5.51546633e-01
2.55150080e-01 2.18500957e-01 -4.90403891e-01 -6.77729011e-01
9.04459298e-01 1.49768949e-01 -5.37429899e-02 -1.18003917e+00
-1.40965986e+00 5.22684872e-01 2.84400880e-01 -8.57386291e-02
-4.37026650e-01 -1.00567624e-01 -9.64006037e-03 -2.26029590e-01
5.39616406e-01 -7.01833144e-02 -3.27945262e-01 1.06369136e-02
-1.09638047e+00 4.73077267e-01 7.06293643e-01 1.12093067e+00
9.36042905e-01 -3.24009955e-01 3.31474155e-01 6.75287426e-01
7.11211085e-01 8.17737043e-01 7.83594474e-02 -8.26869965e-01
5.66048801e-01 7.14143097e-01 3.98090035e-01 -1.24887514e+00
-2.39398479e-01 -1.57037571e-01 -9.79327798e-01 5.28474033e-01
5.33627033e-01 1.56573430e-01 -8.58165979e-01 1.19483888e+00
6.37747049e-01 3.60691458e-01 -2.18545347e-01 8.85492444e-01
6.33006632e-01 4.17131901e-01 -3.20814908e-01 -1.35663599e-01
8.64736319e-01 -4.08486485e-01 -5.40451527e-01 -1.12670138e-01
1.32501826e-01 -9.71290231e-01 7.77921438e-01 5.79295278e-01
-1.04706693e+00 -6.33139074e-01 -9.52942014e-01 2.36813158e-01
7.13779479e-02 2.31203720e-01 1.60683692e-01 2.15922952e-01
-6.03985846e-01 7.88566947e-01 -1.06331253e+00 -2.76047349e-01
4.03664082e-01 3.59321684e-01 -6.78974271e-01 -1.03520509e-02
-5.89937210e-01 8.15960467e-01 3.41164649e-01 2.01161027e-01
-3.94619316e-01 -5.03377020e-01 -7.81436145e-01 -4.26778287e-01
3.51328135e-01 -8.60674679e-01 7.97997892e-01 -6.84460819e-01
-1.38736463e+00 1.01969790e+00 -1.48624361e-01 -1.04814962e-01
1.06078362e+00 -3.76455873e-01 -2.49217283e-02 -1.17275804e-01
2.60913014e-01 4.93230194e-01 1.06361628e+00 -1.87830520e+00
-4.17870522e-01 -6.26814008e-01 -6.36214733e-01 2.60751694e-01
3.43261898e-01 -2.90324956e-01 -6.29590869e-01 -6.62693322e-01
9.94032681e-01 -1.07833135e+00 -2.21964478e-01 3.68031949e-01
-5.93495429e-01 -2.92454511e-01 9.05790985e-01 -5.44209838e-01
6.14518344e-01 -2.13891411e+00 1.30763620e-01 6.73918784e-01
2.24022102e-02 -8.82815123e-02 3.58429313e-01 3.53338689e-01
9.85361356e-03 -2.14728743e-01 -7.29700804e-01 -6.81572258e-01
5.81711018e-03 3.65211427e-01 -1.40959978e-01 9.95607018e-01
-4.96851131e-02 5.20969927e-01 -8.51251602e-01 -6.48731470e-01
3.95668536e-01 5.68063140e-01 -2.64335513e-01 2.73582131e-01
-2.25984961e-01 6.87300682e-01 -3.35350394e-01 7.11074710e-01
1.00722015e+00 3.74932699e-02 -3.40390742e-01 -2.09099963e-01
1.79658771e-01 -4.00157601e-01 -1.79746807e+00 1.97453058e+00
3.86313759e-02 -1.11386172e-01 -4.65133302e-02 -7.34537601e-01
1.33280087e+00 4.37273204e-01 7.79279709e-01 3.13355923e-02
1.10488221e-01 3.31089795e-01 -2.86845624e-01 -2.83046395e-01
2.27214932e-01 -3.31517011e-01 2.67590955e-02 2.42571071e-01
1.90074947e-02 -5.65059900e-01 -4.50175345e-01 -1.25275537e-01
6.16884232e-01 6.71675205e-01 2.03790367e-01 -1.80909839e-02
4.58903879e-01 1.31896988e-01 7.39411771e-01 3.56631041e-01
3.02823018e-02 1.35937119e+00 9.76940691e-02 -3.22159886e-01
-1.22493255e+00 -1.34788382e+00 -3.65374118e-01 1.60407826e-01
4.95407790e-01 -2.61352658e-01 -6.76792800e-01 -5.92183292e-01
2.75534362e-01 4.83864844e-01 -3.60998422e-01 1.03189737e-01
-4.87462670e-01 -5.51285148e-01 1.47147208e-01 4.29412156e-01
2.77189016e-01 -8.13469768e-01 -3.80946010e-01 8.57101977e-02
-2.55467147e-01 -1.08065367e+00 -3.25249940e-01 -2.73034215e-01
-1.09997714e+00 -1.15921104e+00 -5.80546618e-01 -6.98724091e-01
1.12526464e+00 -1.92492008e-01 9.90048587e-01 6.24342263e-02
-1.16467908e-01 4.75150943e-01 -1.11895598e-01 -3.80355388e-01
-3.38319540e-01 -3.75032991e-01 3.34473670e-01 2.95722187e-01
2.32526898e-01 -6.32525623e-01 -3.95924687e-01 6.40702307e-01
-6.71483636e-01 -6.07945696e-02 3.76523793e-01 7.34273195e-01
1.22562790e+00 1.85330555e-01 3.96927558e-02 -6.68014407e-01
1.22606695e-01 -4.83532518e-01 -7.20427334e-01 6.40661567e-02
-3.24077785e-01 -8.81343186e-02 2.19346687e-01 -4.21589643e-01
-9.60703254e-01 7.39138782e-01 3.17085832e-02 -8.32239270e-01
-3.95675749e-01 2.16680706e-01 -4.91488129e-01 -4.39714603e-02
6.23644531e-01 1.65224910e-01 4.78687733e-01 -6.77995026e-01
1.57735214e-01 5.22938251e-01 9.20208275e-01 -7.93643475e-01
1.39746273e+00 8.57714951e-01 2.16739088e-01 -6.58649445e-01
-4.72346544e-01 -7.11527109e-01 -1.42482030e+00 -2.64145613e-01
7.08521128e-01 -8.71556103e-01 -6.11806750e-01 5.23608267e-01
-1.38311660e+00 2.62206465e-01 -2.82343388e-01 5.34391284e-01
-8.79708707e-01 7.29239702e-01 -7.84432143e-02 -8.44957650e-01
-1.26970202e-01 -1.13222229e+00 1.39847839e+00 -2.22778752e-01
-2.68500268e-01 -8.13569486e-01 2.22971663e-01 4.25149471e-01
-2.67916560e-01 8.20898235e-01 4.15095091e-01 -5.79326332e-01
-6.68016732e-01 -8.00034404e-01 1.66200757e-01 3.59823912e-01
8.97527486e-02 1.47498250e-01 -7.29117453e-01 -2.69455791e-01
2.66497135e-01 -7.09542483e-02 2.43096039e-01 1.41489118e-01
1.11992395e+00 -1.00672562e-02 -5.86350739e-01 5.24929523e-01
1.44813669e+00 -3.34792644e-01 6.43285036e-01 2.84270793e-01
9.37421143e-01 6.51947856e-01 1.02989650e+00 5.68828106e-01
4.15318310e-01 7.53841579e-01 7.18060672e-01 1.03639632e-01
2.35557303e-01 -5.78325748e-01 -3.20014656e-02 7.12016523e-01
-3.07413816e-01 1.46112666e-01 -1.28133619e+00 4.25286859e-01
-1.99849904e+00 -7.10253775e-01 -4.33825165e-01 2.74046373e+00
5.95186889e-01 2.17084125e-01 3.11262328e-02 3.44505042e-01
7.63220310e-01 -1.73664361e-01 -4.25664008e-01 4.02162343e-01
-1.59944475e-01 5.97973987e-02 3.32773775e-01 4.44751799e-01
-1.13611531e+00 6.78678989e-01 5.71718264e+00 7.99563885e-01
-7.17673779e-01 3.30579281e-02 3.11837364e-02 1.78518042e-01
-8.49453434e-02 1.20066836e-01 -6.18023694e-01 6.79461956e-01
3.13499719e-01 -5.33516631e-02 -4.15182561e-02 9.16664124e-01
1.01486385e-01 -2.88411975e-01 -1.28952253e+00 1.34045768e+00
3.13990355e-01 -9.39841449e-01 -5.65640032e-02 1.99961260e-01
6.83110654e-01 -3.33525240e-02 -2.03186870e-01 -2.08561838e-01
5.60934059e-02 -8.84188056e-01 9.00053978e-01 9.99426544e-01
7.27665842e-01 -6.94909036e-01 8.61784458e-01 9.47670996e-01
-1.01229763e+00 5.18346786e-01 -4.84080136e-01 9.81411338e-02
2.44369000e-01 7.56630957e-01 -7.90257633e-01 8.38226676e-01
9.10444140e-01 6.88466966e-01 -4.71587569e-01 1.26743424e+00
-1.93661511e-01 1.80327564e-01 -7.73005545e-01 5.10529935e-01
-4.19061363e-01 -6.33564234e-01 8.44463348e-01 5.13876438e-01
3.88403714e-01 1.33229822e-01 5.45693040e-01 9.26265597e-01
3.08308274e-01 2.03820065e-01 -6.01063490e-01 7.47627497e-01
5.54664314e-01 1.02377427e+00 -7.36555159e-01 -9.98631567e-02
-7.48592019e-02 1.02908146e+00 1.70558065e-01 1.08430229e-01
-5.99550188e-01 6.32371083e-02 2.77556956e-01 6.26354098e-01
1.27126753e-01 -3.87422115e-01 -6.72053814e-01 -1.00650966e+00
5.89642465e-01 -5.75037062e-01 1.15436174e-01 -8.19190264e-01
-1.57389498e+00 5.63474298e-01 2.53094196e-01 -1.98517764e+00
-4.55860607e-02 -2.88341224e-01 -6.40636802e-01 8.89734507e-01
-9.66887653e-01 -1.22978795e+00 -4.49259520e-01 4.20105666e-01
2.36335814e-01 -1.74238086e-01 7.39335239e-01 6.43503070e-02
-9.18204188e-02 2.49068484e-01 1.39607683e-01 3.98046114e-02
7.86794722e-01 -1.25032139e+00 2.98546582e-01 6.19521856e-01
1.22972041e-01 4.43258256e-01 8.33441079e-01 -1.13043642e+00
-1.26046693e+00 -9.91409063e-01 5.70255101e-01 -9.33976531e-01
3.56333286e-01 -3.28905940e-01 -1.17313910e+00 6.49187744e-01
-4.08367276e-01 4.08306539e-01 3.31795156e-01 -2.45217204e-01
5.58718145e-02 4.52625118e-02 -1.47700477e+00 2.45232105e-01
1.00307465e+00 -1.46638900e-01 -8.34163904e-01 5.01391172e-01
2.10892007e-01 -9.36922252e-01 -9.85732138e-01 7.44174302e-01
2.05931261e-01 -1.00082994e+00 1.18598998e+00 1.89562123e-02
1.18847065e-01 -7.74825275e-01 -1.38991281e-01 -1.27849543e+00
-3.67960222e-02 -5.39489269e-01 -1.90560788e-01 1.14332032e+00
-3.71613428e-02 -4.21420336e-01 1.21452868e+00 5.48111379e-01
6.81160763e-02 -8.28741372e-01 -1.34137452e+00 -8.82748127e-01
-5.03578177e-03 -5.58330119e-01 6.23629272e-01 7.95309961e-01
-3.87696445e-01 -4.98423278e-02 -2.49827772e-01 6.45641744e-01
1.27472794e+00 -9.24018398e-02 1.21503162e+00 -1.80134869e+00
1.59332156e-02 -2.51925420e-02 -8.14056933e-01 -8.08186829e-01
3.27516824e-01 -7.95410454e-01 3.99163753e-01 -1.50592983e+00
-1.19393654e-01 -8.40231597e-01 4.34859842e-01 1.87809750e-01
-1.09015830e-01 1.84468895e-01 -5.15503809e-02 6.57615781e-01
-2.02863529e-01 7.86941290e-01 1.03906846e+00 3.50833684e-02
-8.26720521e-02 4.64739233e-01 1.01691529e-01 1.26121998e+00
4.77194220e-01 -6.20663166e-01 -5.76625541e-02 -3.55751753e-01
-4.63433675e-02 5.83723709e-02 6.15364850e-01 -1.24155426e+00
5.61178565e-01 -1.43165156e-01 5.46930552e-01 -1.17473412e+00
5.69666922e-01 -1.32585979e+00 6.28242135e-01 1.77934870e-01
3.50533754e-01 1.11608505e-01 -1.80064902e-01 9.92306411e-01
-2.23947346e-01 -2.11115003e-01 7.47310877e-01 -1.46793321e-01
-2.14070633e-01 6.89072311e-01 3.66686791e-01 -9.41310599e-02
1.21170473e+00 -6.47236228e-01 4.47124720e-01 -1.87770814e-01
-9.75458801e-01 2.18063861e-01 9.91157293e-01 3.74126047e-01
9.55183148e-01 -1.44579291e+00 -8.27544153e-01 4.15145636e-01
3.65780562e-01 9.27894592e-01 3.96342576e-02 6.22019589e-01
-5.54529369e-01 -1.75624460e-01 1.43198073e-01 -1.32694483e+00
-1.14735329e+00 3.05838525e-01 3.95034850e-01 2.78465956e-01
-9.67040896e-01 6.21265054e-01 -2.21929789e-01 -9.24134970e-01
3.41291368e-01 -2.17825189e-01 1.42155200e-01 -2.28216738e-01
-3.34936604e-02 5.04108369e-01 2.26560950e-01 -1.18903124e+00
-3.44752222e-01 1.05955100e+00 3.58482540e-01 -1.43651947e-01
1.24248993e+00 4.76807691e-02 -2.48143926e-01 6.66542888e-01
1.00102091e+00 7.05316141e-02 -1.38556147e+00 -5.15144825e-01
8.22618231e-02 -7.85586834e-01 -2.52574176e-01 -3.97861868e-01
-8.72432590e-01 5.51757336e-01 5.57137489e-01 -3.34619492e-01
5.30762315e-01 1.40695602e-01 6.90941215e-01 5.90337589e-02
8.06632876e-01 -1.05477870e+00 -2.49098483e-02 1.58233374e-01
1.17924309e+00 -1.46114039e+00 3.50918442e-01 -8.47796023e-01
-5.82304180e-01 9.92976665e-01 6.41798019e-01 -5.09362936e-01
9.64210272e-01 1.21853195e-01 6.47460148e-02 -3.01078409e-01
-8.44844803e-02 1.78903952e-01 5.52581668e-01 7.42096603e-01
-1.65445954e-01 6.67355582e-02 -5.16184233e-02 4.62620884e-01
-6.15594089e-01 1.18007418e-02 2.57917970e-01 1.02370703e+00
-3.60237837e-01 -1.12249875e+00 -1.09301114e+00 4.35247242e-01
-2.20587105e-03 4.52685267e-01 -1.88993663e-01 6.87175989e-01
3.36457551e-01 4.85289305e-01 2.98981875e-01 -2.82659560e-01
6.83135211e-01 -2.69703977e-02 3.91266078e-01 -8.27976465e-01
-2.02793315e-01 2.33863950e-01 -5.36770523e-01 -4.89144593e-01
-4.70045716e-01 -1.00102520e+00 -1.37594378e+00 7.20399618e-03
-3.99518490e-01 8.54731426e-02 6.55421793e-01 9.72762048e-01
1.78000927e-01 -1.79735780e-01 5.37420273e-01 -1.29186845e+00
-6.55014992e-01 -8.33655834e-01 -6.08691454e-01 7.41542339e-01
1.10088311e-01 -1.09476745e+00 -5.96463382e-01 1.48819208e-01]
|
[8.063830375671387, -3.0441715717315674]
|
c200d20e-386b-4309-9eb8-ade6cda5686b
|
denoising-relation-extraction-from-document
|
2011.03888
| null |
https://arxiv.org/abs/2011.03888v1
|
https://arxiv.org/pdf/2011.03888v1.pdf
|
Denoising Relation Extraction from Document-level Distant Supervision
|
Distant supervision (DS) has been widely used to generate auto-labeled data for sentence-level relation extraction (RE), which improves RE performance. However, the existing success of DS cannot be directly transferred to the more challenging document-level relation extraction (DocRE), since the inherent noise in DS may be even multiplied in document level and significantly harm the performance of RE. To address this challenge, we propose a novel pre-trained model for DocRE, which denoises the document-level DS data via multiple pre-training tasks. Experimental results on the large-scale DocRE benchmark show that our model can capture useful information from noisy DS data and achieve promising results.
|
['Leyu Lin', 'Fen Lin', 'Maosong Sun', 'Zhiyuan Liu', 'Xu Han', 'Ruobing Xie', 'Yuan YAO', 'Chaojun Xiao']
|
2020-11-08
| null |
https://aclanthology.org/2020.emnlp-main.300
|
https://aclanthology.org/2020.emnlp-main.300.pdf
|
emnlp-2020-11
|
['document-level-relation-extraction']
|
['natural-language-processing']
|
[ 3.44583511e-01 5.33580482e-01 -1.92980856e-01 -3.03089797e-01
-1.15474939e+00 -3.53759676e-01 5.70599377e-01 -5.08302525e-02
-3.65091920e-01 1.04096293e+00 6.74641132e-01 -2.75820255e-01
1.09533727e-01 -7.06191659e-01 -5.32320619e-01 -3.20798725e-01
2.34463409e-01 5.22774637e-01 1.22469723e-01 -5.50620615e-01
-5.56974888e-01 3.14673841e-01 -8.60552669e-01 5.22099257e-01
1.02615047e+00 6.34966552e-01 -1.32634476e-01 5.39553285e-01
6.03783987e-02 1.04595876e+00 -1.14831769e+00 -9.20159996e-01
-7.83274695e-02 -5.59452534e-01 -1.18844056e+00 -2.03255668e-01
-1.05592301e-02 -2.12208897e-01 -4.78007436e-01 1.13648736e+00
4.05851722e-01 -1.18048145e-02 6.75809085e-01 -9.14567232e-01
-8.64243686e-01 1.30696487e+00 -3.42806488e-01 3.97611439e-01
2.81855434e-01 -5.99786043e-01 1.15442729e+00 -9.54996109e-01
9.30571377e-01 1.27812552e+00 5.54577470e-01 6.49572134e-01
-1.12390053e+00 -6.19033694e-01 7.56612569e-02 2.78051049e-01
-1.55804932e+00 -4.48739707e-01 7.76117325e-01 1.82655407e-03
1.32740664e+00 2.19275936e-01 1.95711806e-01 1.53148437e+00
2.08492398e-01 8.74342918e-01 8.48957241e-01 -4.61233377e-01
-1.57414109e-01 -1.79958791e-01 1.85447246e-01 1.95847020e-01
3.14640313e-01 -1.27971753e-01 -5.17429829e-01 1.34937674e-01
3.77168894e-01 -3.86459589e-01 -3.93611103e-01 3.74739766e-01
-9.43424404e-01 6.26336277e-01 3.93814087e-01 4.91626143e-01
-2.75908470e-01 -3.47785681e-01 5.61947048e-01 3.49411905e-01
8.57361853e-01 6.13318980e-01 -7.30415702e-01 -2.12882474e-01
-7.12932944e-01 1.16476834e-01 8.22289109e-01 1.23583543e+00
2.28198946e-01 -2.27953047e-02 -4.91729736e-01 9.46410239e-01
-1.47622541e-01 2.89040476e-01 4.12118882e-01 -5.72617531e-01
1.05476761e+00 5.47004163e-01 -7.04732537e-02 -8.21283758e-01
-2.00674042e-01 -7.77922392e-01 -1.44731796e+00 -5.78642249e-01
-7.06825256e-02 -3.34373415e-01 -9.64315653e-01 1.37459219e+00
9.51877385e-02 -3.47927026e-02 6.28432453e-01 7.40225971e-01
1.33810103e+00 6.33251846e-01 4.01692279e-03 -3.21326762e-01
1.17510235e+00 -1.24061656e+00 -1.38348019e+00 -5.99779665e-01
8.23798060e-01 -5.61844647e-01 7.77180314e-01 5.13287306e-01
-9.00501966e-01 -5.10798037e-01 -1.29386783e+00 -4.57252145e-01
-2.61487395e-01 3.99291128e-01 4.64957684e-01 1.94709003e-01
-5.74060142e-01 5.26325822e-01 -8.73157859e-01 6.76012859e-02
5.45499146e-01 -1.45889334e-02 -4.73965853e-01 -2.42552251e-01
-1.83341801e+00 1.16809499e+00 6.02617800e-01 4.95884448e-01
-7.06394553e-01 -5.49213648e-01 -1.03291547e+00 2.68590450e-01
8.92466247e-01 -4.64816481e-01 1.24611497e+00 -1.96499914e-01
-1.54042125e+00 4.41044480e-01 -3.60526472e-01 -5.24602354e-01
2.46631846e-01 -6.83652878e-01 -8.96531045e-01 -8.94489512e-02
1.28956884e-01 1.28627017e-01 4.45452541e-01 -1.24781489e+00
-2.29026243e-01 -1.34249404e-01 -1.24575406e-01 5.95749021e-02
-4.10584003e-01 3.57598066e-01 -3.18532646e-01 -9.60322261e-01
-2.02500261e-02 -5.32141447e-01 -3.09741557e-01 -1.12401736e+00
-9.13988233e-01 -6.40982866e-01 6.87940061e-01 -8.81011426e-01
1.59261668e+00 -1.93237460e+00 1.73683241e-01 -2.10701466e-01
3.67078245e-01 7.28529513e-01 -2.84485817e-01 5.45082152e-01
-3.26468855e-01 3.01446259e-01 -2.14160979e-01 -4.88757908e-01
-3.04232150e-01 2.71325856e-01 -4.72829342e-01 -2.57584155e-01
7.41867185e-01 1.40598083e+00 -1.16186666e+00 -4.57052082e-01
-3.71936291e-01 4.67233121e-01 -7.60001168e-02 4.27317977e-01
-8.83533359e-02 1.85243189e-01 -3.53122056e-01 4.11643535e-01
6.19056880e-01 -3.16913366e-01 4.73624736e-01 -3.15425485e-01
6.07912838e-01 9.27277446e-01 -8.91661108e-01 1.42345273e+00
-4.80872780e-01 4.52590883e-01 -3.63674372e-01 -9.20821071e-01
1.12059617e+00 4.14544523e-01 1.12646148e-01 -5.32841086e-01
1.08692236e-01 1.02956317e-01 -4.65983525e-02 -3.00299555e-01
7.59198427e-01 -1.75637096e-01 -2.96286494e-01 7.46100992e-02
1.53915212e-01 -3.55359882e-01 2.27785692e-01 5.77456176e-01
1.47538197e+00 2.14295927e-02 4.96891081e-01 1.35175660e-01
4.70327288e-01 -2.35550888e-02 1.06702507e+00 4.01267081e-01
1.39825031e-01 7.80613303e-01 6.61296785e-01 -2.37858808e-03
-6.78701758e-01 -9.36123610e-01 -4.13064063e-02 5.19098341e-01
-1.80258825e-02 -1.14848495e+00 -7.81006277e-01 -1.50505185e+00
-4.41904664e-01 7.67233372e-01 -3.15979153e-01 -4.66794610e-01
-6.74055696e-01 -1.01950336e+00 7.55506217e-01 7.02340961e-01
5.73900402e-01 -9.76757228e-01 4.54990238e-01 4.77860332e-01
-7.49101996e-01 -1.82674336e+00 -3.10435802e-01 4.62485015e-01
-5.87382436e-01 -7.67561257e-01 -4.01946187e-01 -6.40739024e-01
5.11807203e-01 1.74600363e-01 1.54573107e+00 -1.27402395e-01
2.91773856e-01 -4.92269427e-01 -8.86683166e-01 -3.10989112e-01
-8.06334317e-01 5.29512286e-01 4.58069257e-02 -4.36248958e-01
8.17114294e-01 -4.73398745e-01 -1.34157792e-01 2.04249069e-01
-9.10838604e-01 -7.95930922e-02 7.05195129e-01 1.09632289e+00
5.00979841e-01 4.84110564e-01 1.03339839e+00 -1.38067472e+00
9.73939776e-01 -3.44053507e-01 -1.02213360e-01 5.07155299e-01
-8.39982092e-01 1.53750807e-01 8.71643722e-01 -2.67439306e-01
-1.20668077e+00 -2.49730229e-01 -3.03198338e-01 -4.55006510e-02
8.01881626e-02 9.16752994e-01 -6.79721117e-01 6.11807585e-01
6.76495731e-01 -1.09366581e-01 -4.99968588e-01 -6.88721418e-01
4.66207862e-01 1.05951369e+00 9.43395197e-01 -3.13324034e-01
9.81882274e-01 -9.17596146e-02 -4.02082540e-02 -2.58796751e-01
-1.78242111e+00 -1.94909349e-01 -6.64821148e-01 5.38943529e-01
7.40285456e-01 -1.20630193e+00 1.06839500e-01 3.25921863e-01
-1.50050128e+00 -1.18855298e-01 -4.79456306e-01 2.27447554e-01
1.25252575e-01 3.98066580e-01 -9.31044340e-01 -4.57313597e-01
-6.57980978e-01 -8.12646270e-01 1.09690166e+00 2.44439933e-02
-3.60061198e-01 -8.77895713e-01 1.72583193e-01 6.38365507e-01
-4.55680229e-02 5.66556938e-02 7.12987959e-01 -6.70659661e-01
-2.34177247e-01 -4.94950294e-01 -3.64049852e-01 8.49403560e-01
4.36160594e-01 -8.81689340e-02 -9.57785130e-01 3.27629671e-02
-1.39492720e-01 -5.99081397e-01 9.81980264e-01 -1.74326599e-01
1.12122881e+00 -2.91768104e-01 -3.87479037e-01 3.07699770e-01
9.11335945e-01 -1.88280165e-01 7.53508747e-01 7.55129531e-02
1.04518855e+00 3.79683405e-01 9.12720442e-01 -1.47067448e-02
8.30202401e-01 8.23894799e-01 -3.08459047e-02 -3.48267287e-01
-4.89068627e-01 -5.52485943e-01 5.04398286e-01 1.37029696e+00
-1.56324059e-01 -6.44578993e-01 -8.23243141e-01 5.83675742e-01
-1.76891077e+00 -6.34599864e-01 -3.82025898e-01 1.45592952e+00
1.49059415e+00 3.59964669e-01 -2.46042579e-01 4.74957615e-01
5.73511183e-01 1.08698532e-01 -5.58201969e-02 -3.86961550e-01
-3.97604853e-01 5.11253715e-01 2.13306561e-01 4.42793310e-01
-1.15616512e+00 1.31373584e+00 6.62038708e+00 1.18976939e+00
-6.02809191e-01 2.92411715e-01 5.14490783e-01 1.33524373e-01
-3.60276848e-01 -5.54276593e-02 -1.19156241e+00 2.28115976e-01
1.22005522e+00 -1.46971241e-01 8.18054080e-02 7.06558228e-01
4.99709845e-02 1.54293790e-01 -1.22260463e+00 7.84342885e-01
1.43958628e-01 -1.17975390e+00 1.26689732e-01 -8.72856081e-02
1.01146126e+00 -1.23032570e-01 -4.48979288e-01 6.83924913e-01
7.94392765e-01 -1.25905263e+00 2.21246585e-01 2.50534624e-01
8.15753102e-01 -8.81071329e-01 1.38694727e+00 4.92043525e-01
-1.10921109e+00 2.81938732e-01 -5.04730940e-01 -6.09511100e-02
3.40087056e-01 1.36264670e+00 -1.06248653e+00 1.24935913e+00
6.06740654e-01 9.91073489e-01 -6.80361450e-01 2.63544112e-01
-8.64912212e-01 7.91125119e-01 -1.65368170e-01 1.74253196e-01
-1.38631538e-01 -2.90247984e-02 4.83080208e-01 1.35188448e+00
6.51740059e-02 1.25462249e-01 -1.48525625e-01 6.46548629e-01
-4.57851619e-01 -7.22251162e-02 -4.36601460e-01 -2.48500213e-01
5.63083112e-01 1.37674177e+00 -2.43849874e-01 -4.29465950e-01
-2.58618683e-01 1.15455258e+00 7.25840151e-01 3.23623717e-01
-5.49806356e-01 -5.11208534e-01 3.48557502e-01 -2.42703035e-01
3.84922862e-01 -1.93702146e-01 -3.26552898e-01 -1.57162046e+00
3.21919024e-01 -1.10348403e+00 4.31616783e-01 -5.59635341e-01
-1.48606229e+00 1.17719150e+00 -1.72497764e-01 -1.18378508e+00
-4.41905588e-01 -1.51841283e-01 -2.21228287e-01 8.11741173e-01
-1.55428886e+00 -1.24343431e+00 -9.37650353e-02 6.30072951e-02
4.36280757e-01 -1.73820972e-01 9.57706809e-01 4.91763115e-01
-1.00868893e+00 1.06479561e+00 -1.58770708e-03 5.46207726e-01
7.80936062e-01 -1.44500649e+00 8.42241585e-01 1.18016124e+00
5.05728364e-01 5.19009173e-01 4.30688292e-01 -9.74694312e-01
-8.89072001e-01 -1.41417408e+00 1.57378793e+00 -9.42581415e-01
6.92319095e-01 -6.83191359e-01 -1.00241852e+00 7.14958847e-01
5.48555925e-02 2.40294695e-01 5.53222895e-01 5.05766034e-01
-4.55381960e-01 -2.78758168e-01 -8.90449643e-01 5.08401453e-01
1.34350765e+00 -7.24922597e-01 -9.18181479e-01 1.85465693e-01
9.61821735e-01 -6.30274355e-01 -1.37454081e+00 9.51042950e-01
-8.55467394e-02 -3.13649923e-01 8.98790598e-01 -8.26936603e-01
9.17359471e-01 -9.99358967e-02 1.99103177e-01 -1.79330778e+00
-3.01052988e-01 -7.36061037e-01 -6.23400569e-01 1.94639742e+00
7.50166416e-01 -2.67056823e-01 4.82490212e-01 3.55269164e-01
-1.29959464e-01 -8.16959321e-01 -9.03973579e-01 -9.81900394e-01
1.93176806e-01 -3.81754011e-01 7.90808737e-01 8.47383201e-01
1.31226987e-01 1.15860701e+00 -5.35849333e-01 2.08216682e-01
2.63385385e-01 -1.62280444e-02 5.89774191e-01 -1.02261245e+00
-3.02977353e-01 1.68711096e-01 -1.68325514e-01 -1.31430089e+00
3.70926857e-01 -1.05686951e+00 3.44335526e-01 -1.65486562e+00
3.25094402e-01 -3.40483934e-01 -2.21724465e-01 5.49040437e-01
-8.99023116e-01 3.07356060e-01 -1.22792728e-01 2.76158936e-02
-6.37834966e-01 1.02083576e+00 1.59770954e+00 -1.61547393e-01
-2.49792840e-02 -7.62871727e-02 -1.05223739e+00 4.88038659e-01
5.45301437e-01 -7.68133938e-01 -4.44991231e-01 -3.56429070e-01
4.21333551e-01 -1.46636680e-01 -3.70334163e-02 -6.24557257e-01
-4.78971228e-02 -2.48650983e-02 2.13056087e-01 -6.64633512e-01
1.02967329e-01 -5.14442325e-01 -2.46661142e-01 -1.49813995e-01
-2.27504015e-01 -2.22055867e-01 -1.00173235e-01 5.66422701e-01
-6.83504641e-01 -2.05392063e-01 2.88032711e-01 2.30438456e-01
-1.23017147e-01 1.76977351e-01 -1.03398770e-01 4.68030632e-01
4.69113946e-01 3.38693053e-01 -5.92034042e-01 -3.23695332e-01
-6.49414062e-01 2.30747372e-01 -1.48203611e-01 5.73756039e-01
4.63130057e-01 -1.44631326e+00 -1.11770892e+00 -3.96382995e-03
8.65003392e-02 7.52042949e-01 -7.17349425e-02 4.20090765e-01
1.07180752e-01 4.38598633e-01 5.34911931e-01 -7.60621205e-02
-1.11018205e+00 4.85609859e-01 8.12735856e-02 -1.11665559e+00
-8.74575496e-01 1.05552149e+00 -1.91157341e-01 -4.17607218e-01
-1.06709376e-02 -4.70372975e-01 -4.24990058e-01 -8.81393924e-02
5.39731920e-01 1.85794055e-01 5.91037631e-01 -4.08353657e-01
-3.67971063e-01 1.22568935e-01 -5.67547679e-01 6.30597547e-02
1.49608779e+00 -1.39166161e-01 -2.60818213e-01 2.01432019e-01
9.25984621e-01 3.14375579e-01 -9.12207663e-01 -4.94482726e-01
4.46932107e-01 -5.95671274e-02 2.59378284e-01 -1.02342427e+00
-1.10601377e+00 7.47113645e-01 -2.63942212e-01 2.55891711e-01
1.03726995e+00 1.73115745e-01 1.36592615e+00 5.33880591e-01
2.27216735e-01 -1.11645055e+00 1.47492802e-02 8.64872336e-01
1.04460299e+00 -1.23025894e+00 1.49745300e-01 -1.08819759e+00
-8.13617766e-01 8.03959489e-01 7.30045259e-01 7.67720118e-02
6.15043819e-01 8.77860487e-01 2.40409851e-01 -6.84828982e-02
-8.03332925e-01 -1.98174149e-01 5.18742204e-01 6.52006924e-01
5.24713278e-01 1.94163751e-02 -4.57854331e-01 1.48542917e+00
-5.43761194e-01 3.33878621e-02 6.32432044e-01 6.71306372e-01
2.67686695e-01 -1.63408160e+00 1.45958662e-01 4.15632486e-01
-6.87736452e-01 -4.98103648e-01 -7.71692753e-01 3.11838508e-01
7.13725435e-03 1.24826980e+00 -4.11527932e-01 -6.86637938e-01
4.83701169e-01 2.88771968e-02 3.26826066e-01 -1.02263606e+00
-6.16245568e-01 -6.46069050e-02 9.60924029e-01 -3.17387998e-01
-3.86688709e-01 -2.98701853e-01 -1.13773525e+00 -5.38223572e-02
-6.08297169e-01 5.09390831e-01 9.34401453e-02 1.29928052e+00
4.15060461e-01 9.46795940e-01 7.06176400e-01 -1.08607195e-01
-6.05195224e-01 -1.53229666e+00 -3.33614737e-01 3.53670359e-01
2.54594386e-01 -5.43523371e-01 -4.33364660e-01 2.31995191e-02]
|
[9.34582233428955, 8.707720756530762]
|
c430d9a7-2428-457e-a0ae-62ba510ac853
|
the-age-of-synthetic-realities-challenges-and
|
2306.11503
| null |
https://arxiv.org/abs/2306.11503v1
|
https://arxiv.org/pdf/2306.11503v1.pdf
|
The Age of Synthetic Realities: Challenges and Opportunities
|
Synthetic realities are digital creations or augmentations that are contextually generated through the use of Artificial Intelligence (AI) methods, leveraging extensive amounts of data to construct new narratives or realities, regardless of the intent to deceive. In this paper, we delve into the concept of synthetic realities and their implications for Digital Forensics and society at large within the rapidly advancing field of AI. We highlight the crucial need for the development of forensic techniques capable of identifying harmful synthetic creations and distinguishing them from reality. This is especially important in scenarios involving the creation and dissemination of fake news, disinformation, and misinformation. Our focus extends to various forms of media, such as images, videos, audio, and text, as we examine how synthetic realities are crafted and explore approaches to detecting these malicious creations. Additionally, we shed light on the key research challenges that lie ahead in this area. This study is of paramount importance due to the rapid progress of AI generative techniques and their impact on the fundamental principles of Forensic Science.
|
['Anderson Rocha', 'Sébastien Marcel', 'Fernanda Andaló', 'Shiqi Wang', 'Haoliang Li', 'Daniel Moreira', 'Renjie Wan', 'Rafael Padilha', 'Jing Yang', 'João Phillipe Cardenuto']
|
2023-06-09
| null | null | null | null |
['misinformation']
|
['miscellaneous']
|
[ 7.15959489e-01 2.80109257e-01 -7.95922894e-03 3.25832188e-01
-5.50242007e-01 -8.76723826e-01 1.31266367e+00 2.06779584e-01
-7.64896423e-02 7.22508848e-01 6.65710747e-01 -5.39079070e-01
1.68810964e-01 -8.62835586e-01 -5.90920627e-01 -4.20879394e-01
-1.19374044e-01 9.38356668e-02 -1.81225434e-01 -2.60095507e-01
9.08041537e-01 5.83551228e-01 -1.30242312e+00 3.75444800e-01
6.52316749e-01 4.71990436e-01 -4.45507646e-01 6.61404669e-01
-4.28999402e-02 1.16474390e+00 -1.23278344e+00 -1.13076687e+00
1.33665353e-01 -6.64721668e-01 -6.39600217e-01 9.90979746e-02
2.65977383e-01 -6.46445870e-01 -3.93943816e-01 1.22528028e+00
3.78031284e-01 -3.15045893e-01 4.28039044e-01 -1.43818021e+00
-9.15990233e-01 6.68105066e-01 -5.70167720e-01 7.26441741e-01
6.08044505e-01 5.30467570e-01 4.23156381e-01 -5.93609095e-01
1.06873775e+00 1.41481626e+00 6.50314569e-01 4.23760891e-01
-8.45935404e-01 -9.68693316e-01 -4.32276279e-01 3.73085499e-01
-1.03874004e+00 -7.34689593e-01 1.13084698e+00 -7.36215234e-01
3.69289368e-01 2.01548874e-01 6.80449724e-01 1.67349613e+00
1.43691912e-01 6.53530180e-01 1.13997042e+00 -6.15736723e-01
1.86523929e-01 2.93791562e-01 -8.87385383e-02 3.88459355e-01
7.69201458e-01 2.16633752e-01 -8.79709184e-01 -5.95048010e-01
5.31998873e-01 -1.98205560e-01 -1.64905906e-01 3.67255390e-01
-1.17691863e+00 1.13287127e+00 -1.50145993e-01 6.12091362e-01
-6.77509129e-01 2.00206384e-01 6.55492365e-01 -1.27263039e-01
4.69211817e-01 9.92470503e-01 6.26719594e-01 -5.89103341e-01
-9.53349233e-01 2.65302092e-01 6.43212736e-01 3.37860227e-01
2.10220665e-01 2.88931996e-01 2.69450992e-01 1.67356297e-01
2.17592135e-01 3.78298670e-01 3.04506153e-01 -9.63820100e-01
3.85421902e-01 4.79058236e-01 1.49967834e-01 -1.64724588e+00
4.28290188e-01 -1.90949276e-01 -2.07789078e-01 2.45790780e-02
1.51690736e-01 -2.84772478e-02 -4.85864013e-01 1.16207349e+00
4.69991118e-01 4.18676406e-01 -1.13863118e-01 7.34347880e-01
5.59492648e-01 4.57669705e-01 1.50567248e-01 -1.05890557e-01
1.40140033e+00 -2.20231757e-01 -1.05466127e+00 -2.29178816e-01
3.31513792e-01 -7.85461009e-01 7.04572856e-01 5.30325890e-01
-1.07099342e+00 1.63017884e-01 -1.15999138e+00 1.04523510e-01
-5.43748379e-01 -4.72883821e-01 5.27892053e-01 1.31337619e+00
-4.38743383e-01 4.87306029e-01 -5.14649868e-01 -1.98445603e-01
8.88888836e-01 -3.86832893e-01 -1.28664732e-01 -1.36393443e-01
-1.21665943e+00 9.02591050e-01 3.47483695e-01 6.02734648e-02
-8.38748336e-01 -6.88998401e-01 -5.83632231e-01 -6.36851668e-01
5.16655445e-01 -6.45212531e-02 9.88960564e-01 -9.43299174e-01
-8.23944330e-01 1.10254633e+00 1.03530832e-01 -6.66876912e-01
7.31239080e-01 -2.19423726e-01 -7.97435284e-01 5.08682668e-01
4.10485893e-01 -4.92924750e-02 1.08341670e+00 -1.62476075e+00
-9.37006921e-02 -4.50419277e-01 1.89818628e-03 -4.34687346e-01
-5.37658572e-01 6.94227397e-01 4.89146888e-01 -9.47435200e-01
-3.11174840e-01 -7.03951180e-01 1.09834544e-01 -2.57174134e-01
-7.35895872e-01 3.35322022e-01 1.18090189e+00 -1.09226620e+00
1.29166174e+00 -2.04907608e+00 -4.71736521e-01 8.13061297e-02
6.33317828e-01 5.67697585e-01 3.04169208e-01 1.12184072e+00
1.79052681e-01 7.87365794e-01 -2.09799156e-01 3.57026346e-02
-4.72811460e-02 -2.18449179e-02 -8.32644582e-01 7.35924006e-01
3.58543098e-01 9.35427308e-01 -1.19506240e+00 -3.12754065e-01
-2.48891488e-02 4.14745927e-01 -5.99889755e-02 -1.58180520e-01
-2.18021646e-01 5.73022366e-01 -6.00656748e-01 8.57915878e-01
6.55481756e-01 3.93051505e-02 2.08311781e-01 7.03937337e-02
-2.89003462e-01 3.88650417e-01 -6.68507814e-01 8.08985829e-01
8.74297097e-02 1.21115458e+00 -2.52803657e-02 -7.16978312e-01
8.37078333e-01 3.54625553e-01 2.36628518e-01 -5.94810724e-01
2.68204629e-01 1.50764287e-01 -1.08459666e-01 -8.09362888e-01
9.16388094e-01 -3.79140824e-01 3.65278125e-02 1.28713119e+00
-4.51931477e-01 -1.26336053e-01 1.14306010e-01 4.73910511e-01
1.35459030e+00 -3.79230641e-02 3.37399840e-01 3.81129354e-01
1.21437438e-01 2.75272697e-01 1.60348728e-01 7.67426968e-01
-3.70285630e-01 3.44942480e-01 6.05846941e-01 -4.25993383e-01
-1.35345912e+00 -8.67877483e-01 8.07258592e-04 5.05275548e-01
1.10878237e-01 -3.83439094e-01 -6.72930002e-01 -6.43119454e-01
5.26140770e-03 1.17544889e+00 -7.11542308e-01 -4.16227102e-01
-6.88934624e-01 -8.57784510e-01 1.32088351e+00 9.50807557e-02
3.47200513e-01 -1.29589534e+00 -1.04250455e+00 2.59631783e-01
-5.10803461e-01 -1.29733789e+00 2.01804444e-01 -8.10836494e-01
-5.50445557e-01 -1.41001987e+00 -3.23222429e-01 4.33699554e-03
4.68477517e-01 3.21768016e-01 8.22438776e-01 2.73789972e-01
-4.46811646e-01 5.33130169e-01 -6.75003886e-01 -7.19838858e-01
-1.30837071e+00 -6.50660276e-01 -1.98168814e-01 2.96347082e-01
3.56905997e-01 -6.69654369e-01 -1.80600956e-01 -5.73954806e-02
-1.61235535e+00 -1.37422234e-01 2.76843667e-01 4.22627598e-01
-1.64429143e-01 2.38214254e-01 5.65713704e-01 -1.10660720e+00
1.25188220e+00 -1.26586068e+00 1.08932287e-01 2.15197697e-01
-4.22997862e-01 -4.13348109e-01 3.32981229e-01 -7.87707746e-01
-1.09551668e+00 -9.45082307e-01 3.59996915e-01 -3.83539498e-01
-1.68839693e-01 5.68901360e-01 1.65970117e-01 -5.00226878e-02
1.01377821e+00 1.93153188e-01 9.51654837e-02 -3.73334944e-01
4.90115583e-01 8.36080790e-01 7.51676261e-01 -4.57495570e-01
1.09323919e+00 1.08768523e+00 -2.23779097e-01 -1.08061779e+00
-4.75433707e-01 -2.41843201e-02 -4.29109782e-01 -8.60049784e-01
3.53920221e-01 -3.37600052e-01 -2.78600574e-01 5.55701077e-01
-1.38927138e+00 1.42967880e-01 -3.49910766e-01 6.65057972e-02
-2.10796550e-01 8.71819556e-01 -6.32226408e-01 -1.10759044e+00
-2.86683738e-01 -8.31427574e-01 6.81398451e-01 -8.09014812e-02
-5.90915978e-01 -8.32706571e-01 1.35811552e-01 9.78137851e-01
2.53120333e-01 1.06879330e+00 6.81235492e-01 -8.37135911e-01
-5.10380983e-01 -5.91931224e-01 -1.36031181e-01 1.15655087e-01
8.39980617e-02 2.42291957e-01 -8.95785630e-01 2.05073074e-01
2.29884401e-01 -3.21548611e-01 2.54196048e-01 -3.17626119e-01
2.89056093e-01 -1.00447524e+00 -8.01463276e-02 -3.79862785e-02
1.11977506e+00 5.85776746e-01 9.81237590e-01 7.67579913e-01
4.73355651e-01 9.11340892e-01 3.67912710e-01 8.88902068e-01
1.80459052e-01 3.14112544e-01 4.75855023e-01 3.24834883e-01
-2.57883728e-01 -5.92918098e-01 4.01343584e-01 4.80088502e-01
-1.47135645e-01 -4.20056224e-01 -1.17052162e+00 6.39852941e-01
-1.44858932e+00 -1.60454035e+00 -4.60339427e-01 1.87657332e+00
5.33178389e-01 1.03283621e-01 1.68969497e-01 4.14244622e-01
1.02925277e+00 3.64377111e-01 -3.26318830e-01 -5.33098936e-01
-2.61323065e-01 2.24470973e-01 1.55621752e-01 -3.40016373e-02
-7.20703721e-01 8.81277919e-01 6.16527414e+00 7.69497454e-01
-1.05449057e+00 1.86523691e-01 4.57485557e-01 9.32264850e-02
-6.14410520e-01 1.25729501e-01 -1.93190858e-01 8.89499426e-01
1.05984402e+00 -2.69866556e-01 5.17710149e-01 3.75798523e-01
4.38120097e-01 -3.95293772e-01 -4.63184416e-01 6.04072213e-01
6.33350134e-01 -1.53365660e+00 -7.14818835e-02 4.38510656e-01
6.71709001e-01 -5.17298460e-01 1.21234715e-01 -4.07996476e-01
3.78379166e-01 -9.92216289e-01 1.18620932e+00 3.57648820e-01
5.81425846e-01 -5.48707366e-01 5.24544418e-01 1.53638780e-01
-2.60159105e-01 2.28725187e-02 7.34114051e-02 -2.08543241e-01
6.20586812e-01 7.18368828e-01 -1.13053024e+00 1.63114756e-01
3.30524743e-01 5.18952906e-01 -5.04523575e-01 7.86701620e-01
-5.29057562e-01 1.05778039e+00 -1.12844095e-01 -5.98998182e-03
1.47701010e-01 -1.36276007e-01 9.78475690e-01 1.25434685e+00
3.21701646e-01 1.31810322e-01 -5.16789317e-01 1.14539886e+00
-4.09959890e-02 -3.49964082e-01 -9.84606922e-01 -1.16475010e+00
7.59386420e-01 9.04252887e-01 -9.46136177e-01 -2.22030789e-01
4.46598902e-02 7.44968176e-01 -1.64118171e-01 1.45522192e-01
-7.78720796e-01 -3.15814614e-02 3.32525671e-01 4.89285618e-01
-4.77228425e-02 -3.28145683e-01 -5.22571325e-01 -1.09472060e+00
4.37285714e-02 -1.25322855e+00 6.94769844e-02 -7.16306865e-01
-1.33383608e+00 3.93595338e-01 4.27211858e-02 -9.17522013e-01
-4.42706317e-01 -5.24199661e-03 -8.11318040e-01 1.38021633e-01
-1.01097238e+00 -1.42312026e+00 -1.94264293e-01 1.56032383e-01
1.52579054e-01 -9.93094370e-02 4.22143519e-01 1.53924420e-01
-3.66761118e-01 1.61100551e-01 -2.07879767e-02 2.28945628e-01
4.13352907e-01 -4.32883710e-01 7.37975240e-01 1.16526163e+00
2.44403809e-01 6.42131388e-01 9.88216698e-01 -1.32330489e+00
-1.54626226e+00 -5.22277355e-01 8.21652651e-01 -6.12557113e-01
1.21515858e+00 -3.75690132e-01 -7.68468559e-01 6.46982908e-01
2.11323462e-02 -5.56841612e-01 9.76866961e-01 -3.56268883e-01
-7.19184756e-01 8.26737881e-01 -1.60807300e+00 6.27994657e-01
1.06740820e+00 -7.51779020e-01 -7.47269690e-01 3.93214703e-01
3.76968175e-01 1.20785050e-01 -5.99828184e-01 -9.56534967e-02
7.63463020e-01 -1.16123152e+00 9.87100601e-01 -7.89234340e-01
8.93067956e-01 7.04190731e-02 1.83348760e-01 -7.39352405e-01
1.65105551e-01 -1.00171447e+00 -4.42937195e-01 1.51779377e+00
-1.38342857e-01 -7.67583430e-01 6.79861069e-01 6.54594898e-01
1.18976697e-01 -3.31706464e-01 -7.62634277e-01 -7.02979565e-01
-6.97666481e-02 -6.75267577e-01 5.57438970e-01 1.62280536e+00
5.48914708e-02 -1.32865250e-01 -4.91687536e-01 -1.09123699e-01
9.07375157e-01 -2.13164136e-01 9.15725410e-01 -9.70667899e-01
2.70721093e-02 -2.92457283e-01 -7.37932563e-01 1.35194119e-02
-1.21605329e-01 -3.90951127e-01 -4.74625558e-01 -9.89100933e-01
3.08853358e-01 -2.10615814e-01 4.73583877e-01 -2.06107330e-02
6.92425519e-02 6.94432497e-01 5.79830408e-01 6.43418491e-01
-1.79187208e-01 2.31510878e-01 1.12606061e+00 -4.46410922e-05
3.45654637e-02 -3.58897060e-01 -1.10420072e+00 8.15411568e-01
9.59827900e-01 -7.21225798e-01 -1.30293146e-01 -3.80314738e-02
4.75526571e-01 -1.22289240e-01 7.68322945e-01 -7.70906091e-01
6.58218563e-03 -3.53520602e-01 -2.07069851e-02 -2.71215945e-01
5.05135059e-01 -3.16202492e-01 4.98941451e-01 4.06108677e-01
-1.82433084e-01 2.23718673e-01 -4.80925031e-02 6.54731989e-01
-6.01327792e-02 -5.07880867e-01 6.27401829e-01 -3.78776163e-01
-4.47139502e-01 -2.67115116e-01 -7.84053206e-01 4.15101081e-01
1.27525210e+00 -6.29933298e-01 -8.47809672e-01 -7.83636928e-01
-3.86746168e-01 -5.84508479e-01 7.21016705e-01 3.49563181e-01
6.46925330e-01 -1.04735911e+00 -7.59489417e-01 -1.38141379e-01
-1.45528689e-01 -7.35690117e-01 1.21545129e-01 4.50904667e-01
-7.16470718e-01 2.18894631e-02 -2.97091872e-01 3.19307834e-01
-1.09072900e+00 5.58045387e-01 -2.72435665e-01 -5.00655212e-02
-6.44830465e-01 4.65596497e-01 -1.85611650e-01 2.70531774e-01
-3.70370567e-01 6.67666137e-01 -1.51958764e-02 1.65294319e-01
9.54079449e-01 1.06448138e+00 -3.74098778e-01 -1.15609205e+00
-2.30367824e-01 -2.02041999e-01 2.66791862e-02 -4.59215760e-01
1.41222012e+00 -6.18017130e-02 -3.16771150e-01 1.48052245e-01
9.71159160e-01 5.77225745e-01 -8.09532642e-01 6.14104420e-02
3.36836994e-01 -9.24094141e-01 -1.78038523e-01 -8.99435103e-01
-6.89554453e-01 7.75574327e-01 -1.02078483e-01 6.23147488e-01
6.93805516e-01 2.15143323e-01 1.22110379e+00 -3.21454346e-01
3.40027928e-01 -8.85596633e-01 5.16928792e-01 -7.58595169e-02
1.03821802e+00 -7.10908532e-01 3.39719117e-01 -3.64965469e-01
-7.12887704e-01 1.15436327e+00 8.58732983e-02 -5.93231879e-02
2.62974620e-01 3.41352165e-01 -9.35184285e-02 -4.97917742e-01
-2.12833077e-01 1.99974626e-01 -1.14635117e-01 9.15231645e-01
9.86413136e-02 3.03022955e-02 -5.99348545e-01 3.59983653e-01
-4.43553180e-01 2.77531031e-03 1.23340476e+00 1.28926957e+00
-1.66012242e-01 -9.73924160e-01 -1.17022312e+00 2.66606301e-01
-9.95966554e-01 -2.83736968e-03 -1.19035256e+00 7.06107020e-01
3.90284538e-01 1.20270276e+00 -1.11489244e-01 -4.84259069e-01
-3.46510410e-01 -5.79085127e-02 2.59308726e-01 -2.36860216e-01
-7.35463738e-01 -3.57319981e-01 5.49827158e-01 -1.92585483e-01
-3.90882790e-01 -1.05561769e+00 -7.93636858e-01 -8.05056751e-01
-3.67722511e-01 -4.67180088e-02 1.04771841e+00 8.90057623e-01
4.60928380e-01 -1.08627819e-01 2.82263607e-01 -6.92916751e-01
-1.38243765e-01 -6.78263307e-01 -3.40632886e-01 5.71197510e-01
1.58801273e-01 -6.32048905e-01 -4.03961837e-01 2.39607096e-01]
|
[12.427656173706055, 1.1120774745941162]
|
95dc75b9-cb1f-4506-abff-09e09b442d39
|
transeditor-transformer-based-dual-space-gan
|
2203.17266
| null |
https://arxiv.org/abs/2203.17266v1
|
https://arxiv.org/pdf/2203.17266v1.pdf
|
TransEditor: Transformer-Based Dual-Space GAN for Highly Controllable Facial Editing
|
Recent advances like StyleGAN have promoted the growth of controllable facial editing. To address its core challenge of attribute decoupling in a single latent space, attempts have been made to adopt dual-space GAN for better disentanglement of style and content representations. Nonetheless, these methods are still incompetent to obtain plausible editing results with high controllability, especially for complicated attributes. In this study, we highlight the importance of interaction in a dual-space GAN for more controllable editing. We propose TransEditor, a novel Transformer-based framework to enhance such interaction. Besides, we develop a new dual-space editing and inversion strategy to provide additional editing flexibility. Extensive experiments demonstrate the superiority of the proposed framework in image quality and editing capability, suggesting the effectiveness of TransEditor for highly controllable facial editing.
|
['Wayne Wu', 'Bo Dai', 'Chen Change Loy', 'Chengyao Zheng', 'Qianyi Wu', 'Liming Jiang', 'Yueqin Yin', 'Yanbo Xu']
|
2022-03-31
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Xu_TransEditor_Transformer-Based_Dual-Space_GAN_for_Highly_Controllable_Facial_Editing_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Xu_TransEditor_Transformer-Based_Dual-Space_GAN_for_Highly_Controllable_Facial_Editing_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['facial-editing']
|
['computer-vision']
|
[ 4.27979559e-01 3.54999155e-02 -1.70844495e-02 -3.06131393e-01
-5.06261349e-01 -6.04509532e-01 7.55242646e-01 -6.76780581e-01
2.22460002e-01 7.61141181e-01 4.15873289e-01 1.33714631e-01
-2.21685320e-01 -7.63016522e-01 -4.91612107e-01 -7.79770195e-01
6.99460089e-01 -1.15799353e-01 -5.01121223e-01 -3.68268251e-01
-2.34767944e-01 4.00467098e-01 -1.21309876e+00 -1.51246175e-01
1.31135201e+00 7.55280137e-01 1.30232543e-01 1.12000972e-01
2.98571885e-02 4.37188506e-01 -3.09255928e-01 -7.13312626e-01
2.33968422e-01 -7.07032263e-01 -1.46421313e-01 3.04046065e-01
3.20023865e-01 -4.07100171e-01 -4.49898034e-01 1.12975430e+00
4.66138482e-01 -6.62157312e-02 4.62226778e-01 -1.35719824e+00
-9.24563110e-01 4.42984790e-01 -7.78093219e-01 -4.63897467e-01
3.25690448e-01 1.40116274e-01 1.02570868e+00 -8.32477927e-01
4.36362654e-01 1.37804520e+00 2.68148303e-01 6.54115021e-01
-1.34272838e+00 -1.09393489e+00 2.27291018e-01 -2.70610806e-02
-1.23688614e+00 -5.67398429e-01 1.14974332e+00 -6.40814975e-02
1.95256993e-01 4.44339246e-01 7.95769989e-01 1.39841259e+00
-7.44175464e-02 7.47253537e-01 1.29056954e+00 -2.60490090e-01
-2.59799391e-01 1.78217500e-01 -6.48084879e-01 7.25576878e-01
4.45576496e-02 5.51583022e-02 -5.94385862e-01 4.38398160e-02
1.19220614e+00 1.96510956e-01 -3.25108916e-01 -3.76655400e-01
-1.44172049e+00 7.10528791e-01 4.02166963e-01 9.44528580e-02
-3.87252748e-01 1.18560001e-01 1.80721089e-01 1.56226069e-01
5.33819675e-01 4.71061885e-01 7.78933764e-02 -2.06520364e-01
-8.06971967e-01 8.74544978e-02 1.83397427e-01 1.20789897e+00
4.59958404e-01 5.01032531e-01 -5.12069166e-01 9.24633265e-01
1.32051662e-01 6.14004016e-01 1.99908152e-01 -9.77481961e-01
4.56435770e-01 6.05407834e-01 -5.70756290e-03 -1.22658658e+00
1.12720281e-01 -6.16775990e-01 -1.25131297e+00 1.41964585e-01
-6.45029992e-02 -5.47791123e-02 -8.23013783e-01 2.05794883e+00
2.15876117e-01 2.56921947e-01 -1.41327232e-01 7.74452746e-01
4.41731542e-01 6.64771199e-01 4.13612165e-02 -1.52092248e-01
1.28951609e+00 -8.95450175e-01 -1.04795802e+00 2.26759404e-01
5.91205619e-02 -7.76728511e-01 1.24326503e+00 1.99404448e-01
-1.11975837e+00 -4.86425221e-01 -1.02519524e+00 -1.09806649e-01
1.87300116e-01 4.50003833e-01 8.19271326e-01 5.85212111e-01
-8.21892262e-01 2.95484722e-01 -7.35686898e-01 -1.94334164e-02
6.11164689e-01 3.61559004e-01 -5.68519235e-01 3.72201242e-02
-1.12014115e+00 5.71115851e-01 2.72539314e-02 3.43142301e-01
-7.27716982e-01 -7.30767488e-01 -7.83643723e-01 2.14433923e-01
4.11679119e-01 -8.68344903e-01 7.86355019e-01 -9.67227340e-01
-2.05664110e+00 5.34967422e-01 -1.38171360e-01 1.16209127e-01
8.28424335e-01 -2.46391207e-01 -4.66173440e-01 -3.77073102e-02
-2.15362310e-02 6.99525356e-01 1.25322211e+00 -1.25072575e+00
-2.69660115e-01 -3.27397615e-01 3.29033524e-01 3.24229181e-01
-8.67567837e-01 -1.71338335e-01 -5.22018075e-01 -1.16828203e+00
3.44999768e-02 -1.12361932e+00 -6.60191551e-02 2.94642895e-01
-6.11394465e-01 2.17880771e-01 8.47285867e-01 -4.26797360e-01
9.54606414e-01 -2.32543993e+00 4.95667696e-01 -3.27992253e-02
5.68897188e-01 3.30964655e-01 -2.21619308e-01 2.48494580e-01
6.49890804e-04 8.34637061e-02 -2.08902329e-01 -5.99007487e-01
4.67125848e-02 1.74525753e-01 -4.22399640e-01 3.55377465e-01
4.15894836e-01 9.01372671e-01 -7.32890487e-01 -4.23395127e-01
2.94577658e-01 9.50369835e-01 -7.22270846e-01 3.12930107e-01
-3.31206843e-02 9.33983684e-01 -7.83191502e-01 6.87456191e-01
8.21502209e-01 3.70967970e-03 1.17629461e-01 -5.35342097e-01
-5.00404611e-02 -1.94945514e-01 -9.03044522e-01 1.66952670e+00
-7.46046066e-01 5.08613765e-01 2.21341774e-01 -6.41136169e-01
1.06257629e+00 3.20770562e-01 4.13476229e-01 -6.80206239e-01
2.52378643e-01 1.80077255e-02 -2.11578891e-01 -1.76783651e-01
6.55771613e-01 -4.11202997e-01 -1.30282976e-02 1.72998965e-01
-5.77081135e-03 -3.96427102e-02 -2.44033575e-01 7.68870115e-02
3.63574654e-01 4.51521307e-01 8.94079208e-02 -1.79516524e-01
5.87565362e-01 -7.51509190e-01 7.92708814e-01 2.10057452e-01
4.02335525e-02 7.95195043e-01 4.38706428e-01 2.38394272e-02
-9.91362274e-01 -8.80349040e-01 7.71859959e-02 8.15813780e-01
2.55039215e-01 -5.19612968e-01 -7.44745791e-01 -4.33752567e-01
-2.04681635e-01 5.74268460e-01 -5.20210564e-01 -2.90900618e-01
-5.86942852e-01 -5.11189759e-01 7.40825236e-01 5.58732510e-01
7.96775341e-01 -5.61873972e-01 -2.66138583e-01 3.72010283e-02
-3.35534751e-01 -1.18110919e+00 -7.75659859e-01 -4.21677679e-01
-6.81711674e-01 -5.14175951e-01 -8.93707752e-01 -5.13476312e-01
7.85762370e-01 2.18855500e-01 5.35246968e-01 -7.98215941e-02
8.19848999e-02 2.49936774e-01 -2.14067906e-01 -1.72312438e-01
-2.66992480e-01 1.25795633e-01 1.41566023e-01 4.60083663e-01
-3.77029270e-01 -1.04562020e+00 -7.17482507e-01 3.14749718e-01
-9.50623035e-01 4.83628362e-01 5.87634802e-01 9.32223916e-01
3.89950693e-01 -9.20529217e-02 5.28540790e-01 -9.37106848e-01
6.57671690e-01 -2.50968665e-01 -5.47355533e-01 3.16752344e-01
-6.41138315e-01 1.64528757e-01 8.53203654e-01 -4.92924243e-01
-1.45923126e+00 -1.42562732e-01 -2.28881866e-01 -6.63245976e-01
2.51724571e-01 2.56606996e-01 -6.92622125e-01 -2.18331754e-01
9.43520740e-02 4.09581512e-01 2.77522206e-01 -4.61901337e-01
7.52563179e-01 3.74561846e-01 5.47707856e-01 -8.19152296e-01
9.92422104e-01 5.60445905e-01 1.94242120e-01 -5.50872445e-01
-4.01154608e-01 3.54487330e-01 -3.71529520e-01 -7.93964937e-02
6.14929974e-01 -1.17892194e+00 -7.98385084e-01 6.93602860e-01
-8.90725195e-01 5.70126213e-02 -2.57859796e-01 2.79624701e-01
-4.42534000e-01 3.66903603e-01 -4.46074396e-01 -5.30574679e-01
-3.29149306e-01 -1.46361518e+00 1.05544579e+00 3.38664979e-01
-1.22864377e-02 -8.44780982e-01 -3.19848895e-01 3.24912876e-01
7.20544517e-01 4.90189254e-01 7.98327029e-01 1.08357228e-01
-6.75837040e-01 1.67328194e-02 -3.76901865e-01 2.48654753e-01
6.30683243e-01 1.29586488e-01 -8.65375459e-01 -3.25861424e-01
1.36564039e-02 -2.17067301e-01 6.73393309e-01 3.68917435e-02
1.27613890e+00 -2.74978250e-01 -2.34741285e-01 1.10775638e+00
1.15414131e+00 5.41639552e-02 6.38808608e-01 1.27919093e-01
1.12190032e+00 3.08040380e-01 5.80020130e-01 6.83132052e-01
3.74935806e-01 8.94514084e-01 2.84284294e-01 -3.92575443e-01
-2.36191273e-01 -6.72043502e-01 3.19928229e-01 9.10420001e-01
-3.29361141e-01 -2.58290946e-01 -2.84506023e-01 3.24538350e-01
-1.51148236e+00 -7.73129761e-01 3.61943662e-01 1.85869813e+00
9.11929548e-01 -1.18998736e-01 -5.39740212e-02 -9.78201032e-02
7.02954113e-01 4.47521061e-01 -5.92329681e-01 -4.83358540e-02
-2.53604174e-01 1.55706614e-01 1.85134232e-01 3.63407671e-01
-6.92437410e-01 1.01344562e+00 5.93304205e+00 1.03289688e+00
-1.48376215e+00 1.11043274e-01 5.41012168e-01 -4.52696681e-02
-1.03230155e+00 -7.41325505e-03 -6.53240919e-01 5.87465763e-01
1.41023099e-01 -2.77788550e-01 4.37164873e-01 6.97542071e-01
2.59154886e-01 4.36538279e-01 -7.24601209e-01 1.04441762e+00
-4.59818915e-03 -1.32354999e+00 5.29610872e-01 1.35636702e-01
7.53014266e-01 -8.18017066e-01 6.11743808e-01 1.39817551e-01
-1.27003580e-01 -9.11966205e-01 7.93751836e-01 5.56096911e-01
1.43582797e+00 -8.76298130e-01 1.45850122e-01 -7.71535859e-02
-1.11069274e+00 6.71191663e-02 -1.56656533e-01 1.00372434e-01
1.85740501e-01 4.90684330e-01 -3.78208786e-01 7.65671313e-01
2.00028449e-01 9.03972208e-01 -4.28152293e-01 3.21137190e-01
-3.32279682e-01 3.58023107e-01 -8.17985982e-02 3.03581715e-01
1.42158074e-02 -7.05855012e-01 7.33232081e-01 7.52693951e-01
5.63940048e-01 2.85747260e-01 -7.04500079e-02 1.13657892e+00
-2.00872347e-01 -1.32724524e-01 -5.94182491e-01 -2.64090300e-01
6.34068966e-01 1.37298226e+00 -3.62383217e-01 -1.29420742e-01
-2.18540311e-01 1.25137806e+00 1.77350953e-01 3.69981319e-01
-1.19933832e+00 -2.46124715e-01 9.89874959e-01 -2.12261230e-02
1.73474938e-01 -4.31197733e-01 -2.51636446e-01 -1.53733933e+00
1.27081692e-01 -1.05890286e+00 -1.64011985e-01 -6.39549792e-01
-9.58413780e-01 7.66452014e-01 -5.25271520e-02 -1.33954871e+00
7.30414391e-02 -1.18728779e-01 -5.63484907e-01 8.38639557e-01
-1.57188940e+00 -1.82778251e+00 -4.84328091e-01 7.98192084e-01
3.95984024e-01 -2.62353957e-01 6.47327065e-01 3.77387017e-01
-6.83765054e-01 1.08017099e+00 1.67218447e-01 -3.17128748e-01
7.54000902e-01 -9.62977648e-01 1.76629618e-01 8.39194119e-01
-5.82186729e-02 8.12048912e-01 5.67646265e-01 -4.28636670e-01
-1.72589600e+00 -1.11727810e+00 1.95601642e-01 -1.96776643e-01
2.80803204e-01 -5.00623047e-01 -6.39039814e-01 6.73747420e-01
2.78295934e-01 -1.60516530e-01 5.96829236e-01 -1.16506785e-01
-3.64517450e-01 -4.71952379e-01 -1.08368886e+00 8.07985663e-01
1.20784426e+00 -7.23385274e-01 -1.93299547e-01 -6.47439286e-02
8.61268401e-01 -4.14208204e-01 -8.78587306e-01 6.17473245e-01
8.56432855e-01 -9.09523666e-01 8.63501668e-01 -8.06276351e-02
6.42351866e-01 -3.49513680e-01 -1.44165069e-01 -1.45240891e+00
-4.06365633e-01 -9.89676595e-01 1.37812883e-01 1.82732797e+00
-1.54996198e-02 -8.21442485e-01 6.72990859e-01 4.73630726e-01
1.39514543e-02 -7.59292066e-01 -6.08473420e-01 -6.27197802e-01
-8.36442783e-02 3.24155055e-02 1.01616216e+00 1.07352102e+00
-2.99820453e-01 2.53377289e-01 -1.02682602e+00 -3.35059129e-02
5.71906090e-01 2.21016347e-01 8.26011181e-01 -8.36188018e-01
-2.63375968e-01 -3.09585929e-01 -3.19507480e-01 -1.08918405e+00
3.22479755e-01 -5.79568207e-01 -2.50782073e-01 -1.11521435e+00
2.82436997e-01 -6.13933206e-01 -2.53233820e-01 5.08812010e-01
-3.50391537e-01 4.72260416e-01 4.12294120e-01 2.12900028e-01
-1.11855648e-01 1.31851292e+00 1.70786333e+00 -9.89097655e-02
1.99278165e-02 -2.87080616e-01 -1.05381978e+00 3.84683549e-01
6.35479569e-01 -1.72945708e-01 -7.71005452e-01 -6.75499916e-01
1.49535552e-01 1.64348572e-01 2.62939900e-01 -7.66880274e-01
-6.83245063e-02 -2.83806413e-01 2.79900014e-01 -9.15231705e-02
6.38084531e-01 -8.58521044e-01 6.15624726e-01 5.78726083e-02
-2.91339904e-01 4.60791998e-02 9.04862285e-02 6.02927506e-01
-4.42654967e-01 4.93484885e-01 6.89128220e-01 1.25708103e-01
-3.90545070e-01 5.84651470e-01 5.71342446e-02 -2.84426391e-01
9.95869756e-01 -2.22810134e-01 3.04485112e-03 -5.87634265e-01
-5.36146760e-01 5.82714491e-02 6.78054154e-01 5.54421604e-01
6.54539585e-01 -1.65472901e+00 -6.73955441e-01 6.47124946e-01
2.71901060e-02 -2.83894390e-01 4.69251186e-01 8.47258985e-01
-2.41825581e-01 2.14584157e-01 -6.29206359e-01 -4.35989857e-01
-1.22955930e+00 2.41345793e-01 5.11033796e-02 4.20938618e-03
-6.63733661e-01 7.12010443e-01 6.57439053e-01 -2.93877602e-01
-2.63761342e-01 -5.34720123e-02 -8.88642669e-02 -3.42370607e-02
3.50590408e-01 2.08195537e-01 -2.25745454e-01 -6.79041326e-01
-1.59566075e-01 6.80454850e-01 -7.56139159e-02 -5.29102609e-03
1.36190963e+00 -3.61042231e-01 -2.52524257e-01 1.25784516e-01
1.00960255e+00 4.10447001e-01 -1.56036830e+00 -9.79046971e-02
-7.43572772e-01 -9.32503045e-01 5.35517670e-02 -4.80523944e-01
-1.36891317e+00 7.18670309e-01 4.29271758e-01 -1.35254353e-01
1.43761587e+00 -4.00964707e-01 9.23162222e-01 -1.78045228e-01
3.16327423e-01 -5.33563375e-01 1.33842140e-01 1.04434989e-01
9.68086064e-01 -9.26284611e-01 -2.78676581e-02 -7.34275520e-01
-8.26907039e-01 9.71245646e-01 6.77382767e-01 1.06286980e-01
3.71315897e-01 1.50822550e-01 -1.14940703e-01 -5.98246157e-02
-4.08211976e-01 1.98230505e-01 2.69748598e-01 4.64805514e-01
4.94169623e-01 2.32885331e-01 -2.14119941e-01 3.31588417e-01
-2.41161212e-01 8.99616852e-02 4.87531394e-01 5.78904033e-01
2.88121492e-01 -1.43996894e+00 -1.28242463e-01 2.07329601e-01
-3.42367262e-01 -1.42776305e-02 -4.29882295e-02 6.16183519e-01
9.01032761e-02 6.28671706e-01 -3.07932973e-01 -5.54381847e-01
2.75005072e-01 -3.20235968e-01 7.39509463e-01 -3.45898628e-01
-2.20243424e-01 2.29724646e-01 -1.10276751e-01 -5.56698859e-01
-4.70088452e-01 -4.14422691e-01 -6.80853486e-01 -4.32520926e-01
-4.91657108e-01 7.52741992e-02 3.97401363e-01 8.44667494e-01
7.22238243e-01 5.29754162e-01 1.00461566e+00 -6.64559722e-01
-5.28181553e-01 -6.54523134e-01 -6.51562929e-01 3.17072093e-01
3.23974490e-01 -8.45885754e-01 1.22211063e-02 4.53585610e-02]
|
[12.489534378051758, -0.26609405875205994]
|
ff585670-b302-4bca-b404-7e2d3baa68b9
|
pixelrl-fully-convolutional-network-with
|
1912.07190
| null |
https://arxiv.org/abs/1912.07190v1
|
https://arxiv.org/pdf/1912.07190v1.pdf
|
PixelRL: Fully Convolutional Network with Reinforcement Learning for Image Processing
|
This paper tackles a new problem setting: reinforcement learning with pixel-wise rewards (pixelRL) for image processing. After the introduction of the deep Q-network, deep RL has been achieving great success. However, the applications of deep reinforcement learning (RL) for image processing are still limited. Therefore, we extend deep RL to pixelRL for various image processing applications. In pixelRL, each pixel has an agent, and the agent changes the pixel value by taking an action. We also propose an effective learning method for pixelRL that significantly improves the performance by considering not only the future states of the own pixel but also those of the neighbor pixels. The proposed method can be applied to some image processing tasks that require pixel-wise manipulations, where deep RL has never been applied. Besides, it is possible to visualize what kind of operation is employed for each pixel at each iteration, which would help us understand why and how such an operation is chosen. We also believe that our technology can enhance the explainability and interpretability of the deep neural networks. In addition, because the operations executed at each pixels are visualized, we can change or modify the operations if necessary. We apply the proposed method to a variety of image processing tasks: image denoising, image restoration, local color enhancement, and saliency-driven image editing. Our experimental results demonstrate that the proposed method achieves comparable or better performance, compared with the state-of-the-art methods based on supervised learning. The source code is available on https://github.com/rfuruta/pixelRL.
|
['Ryosuke Furuta', 'Naoto Inoue', 'Toshihiko Yamasaki']
|
2019-12-16
| null | null | null | null |
['local-color-enhancement']
|
['computer-vision']
|
[ 3.51683050e-01 -8.80819485e-02 -1.34458005e-01 -1.57613188e-01
-1.60540164e-01 -1.24577224e-01 2.11645499e-01 -2.80340277e-02
-6.41911805e-01 7.03777850e-01 -3.06670278e-01 -3.63668203e-01
2.91359518e-02 -9.38580632e-01 -9.00300682e-01 -1.02642024e+00
1.72238350e-01 -1.82523906e-01 2.16953799e-01 -2.88595259e-01
4.49638158e-01 5.19180834e-01 -1.50699568e+00 1.49001330e-01
9.20398474e-01 9.15693700e-01 7.24954903e-01 4.13627207e-01
-1.97791040e-01 9.79935110e-01 -6.82325184e-01 2.47736517e-02
1.73065543e-01 -6.47767961e-01 -5.98975897e-01 3.79145086e-01
2.50700042e-02 -4.37551558e-01 -1.70483693e-01 1.35459995e+00
4.86236900e-01 2.72380352e-01 1.04871839e-01 -1.31272674e+00
-8.61217320e-01 6.47908211e-01 -8.81546855e-01 1.73472971e-01
-2.00429279e-02 3.80117506e-01 8.42918396e-01 -6.54957116e-01
5.10975420e-01 1.27661324e+00 1.30666062e-01 5.25147259e-01
-8.77887964e-01 -6.03116214e-01 4.35431033e-01 5.81983984e-01
-9.83018756e-01 -3.56145874e-02 1.14565003e+00 5.70699833e-02
5.75487018e-01 1.85641602e-01 8.92649472e-01 6.41680658e-01
4.08941180e-01 1.25441253e+00 1.38713515e+00 -5.21298885e-01
4.86806273e-01 -3.14796455e-02 -2.47252911e-01 8.24775338e-01
-9.26351771e-02 2.08264768e-01 -2.32107252e-01 3.89409930e-01
1.08688426e+00 4.28427100e-01 -2.48418972e-01 -3.32784534e-01
-1.30976355e+00 7.39826918e-01 8.82594109e-01 2.21721545e-01
-6.35494232e-01 5.27967691e-01 2.34541222e-01 2.25714028e-01
7.35546201e-02 5.45613647e-01 -4.39435631e-01 1.15094870e-01
-7.92429090e-01 2.65843375e-03 1.39475763e-01 5.81337750e-01
1.06270123e+00 3.17895979e-01 -2.13419318e-01 8.25025797e-01
5.07409498e-02 3.49212706e-01 6.09171629e-01 -1.32641482e+00
4.90575545e-02 5.02877057e-01 1.48129702e-01 -1.02427793e+00
-3.86982322e-01 -2.92139083e-01 -1.00348842e+00 9.03668940e-01
1.80286795e-01 -2.59017646e-01 -1.07362366e+00 1.59161150e+00
2.77102292e-01 3.07370901e-01 1.94834486e-01 9.90937889e-01
6.23245955e-01 8.94653141e-01 9.92389768e-02 -3.54245871e-01
1.23765516e+00 -1.39504409e+00 -9.31147575e-01 -3.03068995e-01
2.50309497e-01 -5.47055781e-01 1.38293338e+00 5.18187582e-01
-1.11268616e+00 -7.28325367e-01 -9.63675737e-01 5.80928884e-02
-3.08921814e-01 3.84793252e-01 7.63613403e-01 3.03651631e-01
-9.89835739e-01 6.76420093e-01 -9.15565014e-01 -1.04720350e-02
5.44784784e-01 2.82537907e-01 7.97515064e-02 6.21553138e-02
-1.23099506e+00 7.36871243e-01 3.13056946e-01 4.48561341e-01
-8.57021391e-01 -2.85788208e-01 -7.16639698e-01 1.42746210e-01
6.77312195e-01 -4.11211014e-01 1.31688476e+00 -1.52397454e+00
-1.62452054e+00 3.91855985e-01 -1.73369721e-01 -5.50818920e-01
4.60206121e-01 -9.91431326e-02 -2.11954385e-01 1.91134408e-01
-3.35827060e-02 9.29200828e-01 1.06674612e+00 -1.24502981e+00
-9.02038693e-01 -1.92559943e-01 5.24677098e-01 5.00584245e-01
-3.61850023e-01 -1.85140893e-01 -6.75628543e-01 -7.85417914e-01
-2.00169850e-02 -8.31794500e-01 -5.24271786e-01 2.99183488e-01
-2.16489017e-01 -2.33936042e-01 9.15060103e-01 -4.80844438e-01
9.64515924e-01 -2.19687557e+00 -1.09536522e-04 -5.58725856e-02
7.62157962e-02 4.32702094e-01 -2.59029627e-01 1.98780954e-01
3.63818090e-03 1.44222677e-01 -4.25481230e-01 -6.21643588e-02
-2.52182662e-01 3.12098235e-01 -1.02892384e-01 1.87531844e-01
2.64631182e-01 1.02834308e+00 -9.47670877e-01 -5.17446816e-01
4.96900588e-01 3.47699702e-01 -3.58608752e-01 4.45790440e-02
-3.79263282e-01 4.49083388e-01 -4.85689551e-01 4.59299922e-01
7.39947140e-01 -2.41120055e-01 7.60475099e-02 -2.32829168e-01
-2.07520440e-01 -1.65522769e-01 -1.30341172e+00 1.52284646e+00
-6.06908321e-01 6.26514018e-01 -7.43887424e-02 -1.12132025e+00
1.04384255e+00 -6.51378110e-02 3.95695746e-01 -1.06070220e+00
2.02914886e-02 -5.34499586e-02 5.80918901e-02 -5.22191703e-01
4.65702444e-01 2.43204057e-01 2.35454589e-01 5.17623901e-01
-3.16676378e-01 -8.73612836e-02 4.61524218e-01 3.31353135e-02
8.27440619e-01 4.14151579e-01 2.66918123e-01 3.68010700e-02
6.60767376e-01 6.19837381e-02 8.38586807e-01 9.08759534e-01
-3.31961483e-01 2.65296638e-01 5.53907514e-01 -5.57009161e-01
-7.68277824e-01 -7.69780993e-01 2.02359796e-01 1.03416836e+00
6.02137268e-01 1.85453549e-01 -8.24666679e-01 -6.12856686e-01
-1.90967709e-01 8.13728809e-01 -5.66897094e-01 -2.93593317e-01
-6.32388532e-01 -7.62889624e-01 1.07593313e-01 4.57848340e-01
1.04511571e+00 -1.86881244e+00 -1.07719934e+00 2.12853178e-01
-3.54598835e-02 -8.04713249e-01 -3.71275842e-01 1.41457036e-01
-1.05104053e+00 -8.95375192e-01 -7.83442557e-01 -1.07860291e+00
9.75193501e-01 4.85639364e-01 8.51686299e-01 3.24692935e-01
-1.69855863e-01 5.93139045e-02 -4.18338597e-01 -2.96236068e-01
-3.34685504e-01 -1.53753772e-01 -2.87000448e-01 1.51324393e-02
2.41537974e-03 -2.25745484e-01 -8.92391741e-01 3.40312600e-01
-1.14054120e+00 3.50475669e-01 8.87309790e-01 9.19733047e-01
8.62407029e-01 5.22703767e-01 6.27792478e-01 -1.12835085e+00
6.61025286e-01 8.19986612e-02 -7.79152513e-01 4.23440486e-01
-6.56949997e-01 1.48310333e-01 7.52347767e-01 -3.73850167e-01
-1.31863952e+00 1.93550438e-01 -1.37656312e-02 -3.56096506e-01
-1.46460697e-01 4.18468982e-01 -2.89849797e-03 -1.20770279e-02
3.60184252e-01 3.08525056e-01 2.09814131e-01 -1.28905505e-01
4.53351647e-01 4.27504718e-01 3.94063443e-01 -1.19822383e-01
4.74451244e-01 4.37747061e-01 -1.01372480e-01 -4.53086734e-01
-4.84046489e-01 -6.42462727e-03 -2.22354636e-01 -3.18550199e-01
8.32697213e-01 -7.43483603e-01 -9.70275879e-01 7.06578672e-01
-9.98074234e-01 -6.25564277e-01 -3.85981500e-01 1.92917570e-01
-5.87162912e-01 2.25485206e-01 -6.00810289e-01 -8.25416505e-01
-3.51341069e-01 -1.52282178e+00 7.12899923e-01 6.95104301e-01
3.51754278e-01 -9.91983354e-01 -4.64512885e-01 1.73742682e-01
2.45573416e-01 8.31723735e-02 8.33012283e-01 9.18916911e-02
-7.04652965e-01 1.73147753e-01 -1.74421087e-01 4.65525478e-01
3.04679364e-01 9.85414237e-02 -6.40383303e-01 -2.66511291e-01
-1.14007622e-01 -1.99706376e-01 1.01751685e+00 6.26425803e-01
1.68058777e+00 -2.37379327e-01 -8.02775547e-02 4.10175085e-01
1.47209525e+00 5.82382560e-01 9.35889721e-01 6.80080116e-01
5.97184241e-01 4.29281116e-01 9.45971668e-01 3.03184271e-01
2.23318353e-01 5.60328305e-01 8.68143618e-01 -8.05800080e-01
-2.53025532e-01 -9.30110812e-02 4.90725815e-01 4.46345687e-01
1.01434430e-02 -1.97963655e-01 -5.68267941e-01 4.11732167e-01
-2.13009357e+00 -8.73210251e-01 1.22171290e-01 1.87333310e+00
7.16178596e-01 8.66754800e-02 -2.28037238e-01 1.96885362e-01
9.00849462e-01 2.91415453e-01 -1.11453342e+00 -6.02383673e-01
-1.26575693e-01 2.07710192e-01 4.78607655e-01 3.38363349e-01
-1.02173126e+00 1.21571362e+00 5.25348568e+00 8.46735656e-01
-1.41225255e+00 -1.81294978e-01 9.36458766e-01 1.46392241e-01
-1.13354780e-01 -1.17690980e-01 -3.95617932e-01 3.91271383e-01
2.06213489e-01 7.43077919e-02 8.07438314e-01 7.94473290e-01
6.15869045e-01 -5.00147939e-01 -7.26281941e-01 8.70106876e-01
-4.16446924e-02 -1.32034588e+00 4.89890762e-02 -3.40832084e-01
8.83923948e-01 -3.32165867e-01 3.74805033e-01 1.42115161e-01
2.12291226e-01 -7.36224771e-01 5.58152318e-01 4.56944376e-01
3.45112473e-01 -8.95045340e-01 7.09642053e-01 3.10442775e-01
-8.38674426e-01 -3.59563231e-01 -5.22740662e-01 -6.09525852e-02
-1.50188342e-01 5.95224679e-01 -8.04935098e-01 2.80770659e-01
8.79037321e-01 9.09054577e-01 -6.16356373e-01 9.62019384e-01
-8.97356451e-01 4.30832714e-01 1.13644987e-01 -3.54197621e-01
2.84088433e-01 -3.35116118e-01 2.14670852e-01 7.84465551e-01
2.93635011e-01 -3.46055217e-02 7.52111673e-02 9.27536905e-01
-1.30789518e-01 6.30708039e-02 -3.48380208e-01 2.25580484e-01
3.13562125e-01 1.36860263e+00 -9.29697931e-01 -3.66281092e-01
-3.04781735e-01 1.23083425e+00 2.03585297e-01 6.68007672e-01
-8.93958092e-01 -5.61786413e-01 4.99041796e-01 -2.36749560e-01
4.78923500e-01 -6.41873106e-02 -3.76695871e-01 -8.37559402e-01
-3.59996147e-02 -9.39754188e-01 1.73646435e-01 -1.17307901e+00
-7.68281698e-01 5.64352214e-01 -3.05977166e-01 -1.17720616e+00
-1.24464817e-01 -4.19898212e-01 -7.15092480e-01 5.32595575e-01
-1.94393611e+00 -6.58204436e-01 -2.34272420e-01 6.24374509e-01
7.41221011e-01 -1.04647599e-01 5.37879884e-01 -7.50225550e-03
-7.91000724e-01 2.97702581e-01 1.67459801e-01 8.70786235e-02
4.12113100e-01 -1.27522969e+00 2.03208774e-01 9.99634564e-01
1.89342782e-01 2.48733953e-01 5.52695513e-01 -3.01887274e-01
-1.32449460e+00 -1.20786834e+00 3.31863493e-01 3.57217818e-01
3.00475121e-01 -2.36211233e-02 -7.40765750e-01 3.43223631e-01
4.68298972e-01 8.37193877e-02 7.77941644e-02 -3.00445169e-01
3.22673947e-01 -5.86547196e-01 -1.14815128e+00 1.01014483e+00
6.87647164e-01 -6.30953535e-02 -3.04821432e-01 3.61775815e-01
7.04745710e-01 -3.22161824e-01 -4.13679749e-01 2.49915913e-01
2.65868992e-01 -1.00558877e+00 9.22713637e-01 -1.86593652e-01
6.29980028e-01 -7.30173230e-01 3.35671127e-01 -1.55056918e+00
-3.89085263e-01 -3.61308187e-01 1.25303596e-01 9.45315540e-01
3.55782330e-01 -6.62147164e-01 7.58450270e-01 5.11425853e-01
-7.04850396e-03 -9.48077261e-01 -6.16695583e-01 -4.49871957e-01
-1.75548375e-01 -1.90275833e-01 6.62411869e-01 6.34983540e-01
-3.86360079e-01 -1.16836511e-01 -5.00966728e-01 2.87358493e-01
4.76361454e-01 4.88973498e-01 6.24793768e-01 -6.35916591e-01
-2.51277298e-01 -4.68521506e-01 -2.79791147e-01 -1.11171353e+00
1.19783789e-01 -5.80969095e-01 2.15224236e-01 -1.79806447e+00
6.75029904e-02 -5.02232850e-01 -6.98491693e-01 9.57796991e-01
-4.72860634e-01 3.29703301e-01 5.76654911e-01 6.71719983e-02
-6.50620222e-01 6.67167246e-01 1.86952710e+00 -4.11257803e-01
-4.47346836e-01 6.30510747e-02 -6.50381386e-01 6.96025014e-01
1.31630230e+00 -3.53748322e-01 -3.80931854e-01 -5.64152062e-01
5.31362444e-02 5.23540117e-02 4.27491397e-01 -1.00398445e+00
3.38323981e-01 -4.38345313e-01 4.76350456e-01 -3.07022631e-01
1.79413781e-01 -8.93274784e-01 -8.50479007e-02 8.19735467e-01
-5.17101705e-01 1.10507920e-01 1.92380726e-01 4.14370269e-01
-3.20352733e-01 -4.10904795e-01 9.14271832e-01 -3.95578116e-01
-1.22788179e+00 1.10697053e-01 -4.46930945e-01 -3.87157232e-01
1.19446361e+00 -1.67815045e-01 -2.32826546e-01 -5.80135465e-01
-6.31567419e-01 4.05653417e-01 3.34927201e-01 4.23753887e-01
9.65186834e-01 -1.17882967e+00 -3.88966411e-01 2.35277295e-01
-2.40339920e-01 -1.68671124e-02 2.85877287e-01 6.49594307e-01
-5.91761887e-01 -1.51046291e-01 -6.11208439e-01 -4.78802055e-01
-1.09357536e+00 6.37527764e-01 3.24764490e-01 -1.89131260e-01
-4.76211697e-01 5.25325477e-01 3.04458141e-01 -2.05657840e-01
1.95747256e-01 -4.15897846e-01 -4.65369999e-01 -2.40558177e-01
4.68319207e-01 3.31484526e-01 -1.35005414e-01 -1.57186180e-01
-1.09298155e-01 6.28370464e-01 -2.17051014e-01 5.21762529e-03
1.36960232e+00 -2.62500256e-01 -2.64641851e-01 3.93767446e-01
8.26488733e-01 -3.29129487e-01 -1.71250486e+00 -1.26638263e-01
-3.72271299e-01 -3.62091333e-01 4.19641286e-01 -9.20498729e-01
-1.77834237e+00 9.94310141e-01 8.69646966e-01 8.77013877e-02
1.67212522e+00 -3.74387175e-01 7.64442444e-01 4.77664769e-01
3.39188188e-01 -1.33713448e+00 2.21440405e-01 3.45838577e-01
9.14453208e-01 -1.34380841e+00 1.33141473e-01 -1.93307638e-01
-9.90017593e-01 1.17420959e+00 7.31030047e-01 -2.48771071e-01
4.08200055e-01 5.74888699e-02 3.31795365e-01 1.30724564e-01
-5.81276536e-01 -3.37263435e-01 -1.62379086e-01 4.84199464e-01
1.16275944e-01 1.03046313e-01 -4.26483959e-01 6.32685721e-02
1.81703761e-01 1.60566613e-01 6.81966007e-01 8.93219113e-01
-5.29619098e-01 -1.22214735e+00 -3.31986219e-01 2.34050497e-01
-2.26049304e-01 -1.23884819e-01 2.74484269e-02 6.93680346e-01
2.31840491e-01 9.86451805e-01 -5.30091720e-03 -7.88225979e-02
1.97113752e-01 -4.15181905e-01 2.95689017e-01 -3.39880764e-01
-4.16292101e-01 4.43713926e-02 -4.69050974e-01 -5.77561557e-01
-7.24786162e-01 -5.14830232e-01 -1.82010496e+00 -3.45276482e-02
-1.27227157e-01 -2.59261597e-02 7.91091502e-01 6.49120748e-01
3.26007426e-01 9.79401410e-01 9.26805377e-01 -7.47728705e-01
-2.56175935e-01 -5.54961622e-01 -5.44459641e-01 3.14338863e-01
3.33665162e-01 -6.16931617e-01 -1.11294560e-01 1.96261004e-01]
|
[11.224810600280762, -1.3185428380966187]
|
7e5e5242-72d1-4b72-a9ab-e5696e311342
|
hprnet-hierarchical-point-regression-for
|
2106.04269
| null |
https://arxiv.org/abs/2106.04269v2
|
https://arxiv.org/pdf/2106.04269v2.pdf
|
HPRNet: Hierarchical Point Regression for Whole-Body Human Pose Estimation
|
In this paper, we present a new bottom-up one-stage method for whole-body pose estimation, which we call "hierarchical point regression," or HPRNet for short. In standard body pose estimation, the locations of $\sim 17$ major joints on the human body are estimated. Differently, in whole-body pose estimation, the locations of fine-grained keypoints (68 on face, 21 on each hand and 3 on each foot) are estimated as well, which creates a scale variance problem that needs to be addressed. To handle the scale variance among different body parts, we build a hierarchical point representation of body parts and jointly regress them. The relative locations of fine-grained keypoints in each part (e.g. face) are regressed in reference to the center of that part, whose location itself is estimated relative to the person center. In addition, unlike the existing two-stage methods, our method predicts whole-body pose in a constant time independent of the number of people in an image. On the COCO WholeBody dataset, HPRNet significantly outperforms all previous bottom-up methods on the keypoint detection of all whole-body parts (i.e. body, foot, face and hand); it also achieves state-of-the-art results on face (75.4 AP) and hand (50.4 AP) keypoint detection. Code and models are available at \url{https://github.com/nerminsamet/HPRNet}.
|
['Emre Akbas', 'Nermin Samet']
|
2021-06-08
| null | null | null | null |
['2d-human-pose-estimation']
|
['computer-vision']
|
[-5.12550473e-01 4.91147861e-02 -1.49141803e-01 3.11009632e-03
-8.13554406e-01 -3.79047215e-01 2.50506699e-01 -2.22738355e-01
-3.66792023e-01 2.64056355e-01 1.79966137e-01 6.15525961e-01
2.76485860e-01 -4.51042086e-01 -7.42269754e-01 -5.74520826e-01
1.75571479e-02 8.72470140e-01 3.93810749e-01 -3.94452780e-01
-5.50996028e-02 4.11760539e-01 -1.36946702e+00 -1.91590950e-01
2.44715631e-01 9.79604959e-01 -3.45060319e-01 5.66286623e-01
5.02321661e-01 8.80609557e-04 -5.74974358e-01 -9.19634163e-01
2.77070552e-01 -1.25078470e-01 -4.99976635e-01 -1.20953610e-02
1.11404395e+00 -4.87386018e-01 -2.36642063e-01 7.36591518e-01
8.78553391e-01 2.94108391e-02 7.37489879e-01 -1.28685224e+00
2.02395782e-01 1.03747509e-01 -1.50125861e+00 -8.23722184e-02
5.80384254e-01 -1.01375654e-01 8.21831048e-01 -1.32010424e+00
6.19717836e-01 1.55406690e+00 1.09654355e+00 6.59248412e-01
-9.40371573e-01 -9.58487391e-01 2.30874360e-01 -2.25617569e-02
-1.83806682e+00 -4.85388547e-01 6.95859909e-01 -5.78409016e-01
4.66232866e-01 8.76876563e-02 9.22655046e-01 7.57435262e-01
5.02213895e-01 7.57209480e-01 6.75859988e-01 -4.03280199e-01
-3.42924058e-01 -4.59074199e-01 2.09970362e-02 1.15924609e+00
4.67808932e-01 -1.67540222e-01 -9.26486433e-01 -2.84635782e-01
1.09972453e+00 -9.20710191e-02 1.86611652e-01 -5.67008495e-01
-1.08710027e+00 5.19039094e-01 5.06410360e-01 -2.27867663e-01
-4.13572192e-01 5.48768461e-01 1.79952651e-01 -4.28657472e-01
4.21517104e-01 -4.28869605e-01 -5.92990041e-01 2.80882251e-02
-1.00605035e+00 6.00211442e-01 6.48606181e-01 1.04370701e+00
5.81362784e-01 -2.98412263e-01 -1.82382360e-01 8.84557962e-01
6.73219979e-01 8.32578421e-01 1.41235188e-01 -7.86103427e-01
5.38833320e-01 5.88240445e-01 4.70582396e-02 -9.12485659e-01
-6.68933988e-01 -3.27464640e-01 -6.95529282e-01 9.66932774e-02
8.80479038e-01 -3.13592404e-01 -1.11937761e+00 1.63799262e+00
1.16070497e+00 -1.77975655e-01 -7.59007871e-01 1.08436430e+00
9.27744448e-01 2.04539016e-01 1.44880265e-01 1.94776118e-01
2.19736505e+00 -9.64226425e-01 -3.80779088e-01 -5.11035681e-01
1.29434004e-01 -8.49365830e-01 6.89267695e-01 4.37933087e-01
-1.22798026e+00 -6.75514162e-01 -6.69815838e-01 -2.50423104e-01
-4.91692461e-02 5.91467679e-01 3.97152871e-01 5.54930866e-01
-7.08127260e-01 2.81350136e-01 -9.88292456e-01 -3.77190024e-01
1.27721176e-01 6.50055707e-01 -6.30307615e-01 1.57153651e-01
-1.07414091e+00 9.06658411e-01 -7.87158459e-02 3.52426440e-01
-5.77264726e-01 -4.26018208e-01 -9.75054204e-01 -3.58741611e-01
7.18231499e-01 -1.10528719e+00 1.30371034e+00 -4.69238967e-01
-1.50023663e+00 1.09758687e+00 -3.71669084e-01 7.56267160e-02
7.69361436e-01 -8.60802948e-01 -8.81377384e-02 2.54254818e-01
2.89791942e-01 8.98438752e-01 1.34580588e+00 -9.44672406e-01
-6.94643438e-01 -8.95585656e-01 -4.29619908e-01 4.42611933e-01
1.40425265e-01 2.47433871e-01 -1.36110067e+00 -8.48854542e-01
3.60230446e-01 -1.33107734e+00 5.89523800e-02 3.87945235e-01
-7.24327743e-01 -4.74915504e-01 4.36034381e-01 -1.09366870e+00
1.19903064e+00 -1.97401869e+00 4.31698918e-01 3.88954192e-01
2.72044212e-01 -9.22405645e-02 3.15187395e-01 2.13491753e-01
-7.29347160e-03 -3.27479482e-01 1.65370271e-01 -5.44847608e-01
6.29675165e-02 -1.09996431e-01 3.82895112e-01 9.48670387e-01
-2.61810869e-01 8.45402420e-01 -4.01532084e-01 -9.51741874e-01
1.49702519e-01 6.84331298e-01 -3.75503719e-01 -2.13039592e-01
2.74462402e-01 3.52768272e-01 -3.81286263e-01 1.13176382e+00
6.75411105e-01 -1.50964066e-01 -2.75891349e-02 -5.39067924e-01
1.48698822e-01 -2.53095955e-01 -1.60359943e+00 1.62811708e+00
5.86571880e-02 -8.46595615e-02 2.78818667e-01 -6.96600899e-02
5.23476541e-01 3.44517678e-01 6.09332860e-01 -2.18158722e-01
2.59844899e-01 -5.22257350e-02 -1.07947461e-01 1.62390038e-01
2.16497645e-01 -1.42295882e-01 -2.83163667e-01 -2.74926070e-02
1.66180491e-01 1.36655301e-01 1.35350734e-01 -2.50610728e-02
4.91120100e-01 4.33456272e-01 5.46708286e-01 -1.43232152e-01
5.20387590e-01 -2.21011490e-01 5.77830732e-01 3.20195198e-01
-2.40996301e-01 8.39722633e-01 2.87886053e-01 -2.65441746e-01
-6.44519687e-01 -1.17334259e+00 -7.28484057e-03 1.46444857e+00
2.17050776e-01 -9.25572634e-01 -1.06045389e+00 -6.27501428e-01
4.99941051e-01 -2.72985518e-01 -7.56964982e-01 4.99787889e-02
-8.51047277e-01 -3.75553548e-01 4.93776649e-01 7.67335892e-01
2.78054446e-01 -7.27582514e-01 -6.77138388e-01 -1.45973405e-02
-3.65369797e-01 -9.48750556e-01 -9.27750409e-01 -3.47567409e-01
-7.33499527e-01 -1.26336062e+00 -1.17939472e+00 -3.86880994e-01
7.14237332e-01 -4.80333641e-02 1.04535067e+00 -9.17247310e-03
-6.52510345e-01 5.66022933e-01 -3.37659717e-02 -5.31213880e-01
4.67481822e-01 -1.15909912e-01 4.54166740e-01 -6.23496696e-02
3.52996658e-03 2.37434823e-02 -1.00032556e+00 7.37667561e-01
-2.28753258e-02 -1.73821777e-01 5.41354239e-01 5.99132776e-01
8.37006569e-01 -8.52945074e-02 -2.39477664e-01 -4.75621045e-01
-9.25848484e-02 -8.32839906e-02 -4.47580367e-01 1.63770437e-01
-8.25316906e-02 -3.30643386e-01 -2.74506584e-02 -4.35754597e-01
-8.16781223e-01 4.79744434e-01 -2.79889703e-01 -4.13581073e-01
-2.00863462e-02 7.12151006e-02 -2.84328789e-01 -2.43074581e-01
3.36607218e-01 -3.57073918e-02 1.97216541e-01 -7.25690067e-01
3.53374928e-01 1.06991336e-01 7.08308220e-01 -7.02638865e-01
9.69339907e-01 5.94608188e-01 2.54090846e-01 -8.93850029e-01
-7.07373917e-01 -7.05378354e-01 -1.17579186e+00 -4.79267269e-01
8.13185573e-01 -1.15441191e+00 -9.99939263e-01 7.45477438e-01
-9.06765103e-01 -1.17204398e-01 6.61582453e-03 4.48632538e-01
-4.78603184e-01 4.23855782e-01 -7.49406636e-01 -7.09652305e-01
-7.35398293e-01 -8.35306466e-01 1.85069489e+00 2.03484863e-01
-7.21617520e-01 -5.05589187e-01 6.99099153e-02 4.87304389e-01
-4.60161537e-01 2.90888786e-01 2.58558095e-01 -1.45289809e-01
-3.62605006e-02 -5.69013178e-01 1.02404945e-01 -1.25216067e-01
-8.18949342e-02 1.59811288e-01 -5.58058381e-01 -5.77340543e-01
-3.68938118e-01 -1.43041328e-01 4.83305931e-01 7.53016591e-01
6.52294517e-01 -1.62041321e-01 -6.48215950e-01 6.16853893e-01
8.90882850e-01 -3.92797977e-01 3.72059584e-01 1.14636116e-01
9.48355377e-01 7.98734665e-01 9.70395267e-01 6.68831050e-01
5.93578398e-01 1.09010851e+00 2.73606509e-01 -1.80097818e-01
-3.37967783e-01 -3.90947521e-01 4.83769864e-01 2.45705709e-01
-7.67231405e-01 2.20298424e-01 -9.43387687e-01 2.32537061e-01
-1.64299953e+00 -6.24918699e-01 -2.51929611e-01 2.27024937e+00
6.62012696e-01 -6.52020350e-02 9.04182076e-01 -1.15559146e-01
9.12658870e-01 -1.39623418e-01 -5.05723596e-01 4.03011918e-01
3.16981375e-01 2.18747467e-01 5.63896060e-01 2.57243544e-01
-1.38094866e+00 1.07690644e+00 6.11608982e+00 7.85064995e-01
-8.38450253e-01 1.47434771e-01 2.15723380e-01 -3.80483299e-01
7.08182693e-01 -3.49421203e-01 -1.53794897e+00 4.51418906e-01
3.87002587e-01 4.24989372e-01 -7.70580098e-02 9.66245234e-01
-1.89389020e-01 -2.51360625e-01 -9.10285950e-01 1.28076637e+00
1.15971893e-01 -5.13654947e-01 -1.76429793e-01 2.16402560e-01
1.98984593e-01 -2.12396905e-01 -4.31093387e-02 1.72952443e-01
-9.20892432e-02 -7.11344898e-01 1.10075605e+00 4.02459890e-01
9.80671942e-01 -8.21172953e-01 4.99658853e-01 3.00102293e-01
-1.67083514e+00 2.01836243e-01 -2.17675492e-01 1.11669555e-01
1.86551735e-01 3.58648032e-01 -3.65900874e-01 4.68164772e-01
1.12077904e+00 4.19338912e-01 -6.84495926e-01 9.37663376e-01
-6.29504502e-01 2.91730434e-01 -7.36396134e-01 4.27662075e-01
-3.89245570e-01 1.61816061e-01 5.39467216e-01 1.02304053e+00
1.62741765e-01 1.39316112e-01 2.70687610e-01 2.18332753e-01
1.80988032e-02 2.61559576e-01 6.66870922e-02 5.96856236e-01
3.00655454e-01 1.57934117e+00 -7.95602620e-01 -2.66622514e-01
-3.36980373e-01 1.15714645e+00 1.73872337e-01 4.32974733e-02
-1.03498816e+00 -1.33124799e-01 6.79180980e-01 7.03278661e-01
3.66123259e-01 -1.88837528e-01 1.21091478e-01 -1.22123098e+00
3.03362668e-01 -8.71755779e-01 8.40133965e-01 -5.77871382e-01
-1.13195181e+00 1.13850497e-01 2.88771510e-01 -1.12813759e+00
-3.39841515e-01 -6.47991121e-01 -3.51168245e-01 8.34529340e-01
-6.71832025e-01 -1.67466450e+00 -3.35804701e-01 9.75483835e-01
3.74693334e-01 3.42690527e-01 6.19242370e-01 2.29014501e-01
-5.78404844e-01 9.80989397e-01 -5.72272182e-01 6.34080887e-01
1.08755469e+00 -1.08330643e+00 6.06670380e-01 6.00438535e-01
9.03631300e-02 8.88126016e-01 5.54698288e-01 -1.13893044e+00
-1.59999931e+00 -7.43313193e-01 6.40087783e-01 -8.67993057e-01
3.44237804e-01 -5.72315991e-01 -3.62359107e-01 8.81971121e-01
-3.86993974e-01 2.60049880e-01 5.76697052e-01 4.62302148e-01
-4.22702938e-01 -1.36695400e-01 -1.03140783e+00 6.37423217e-01
1.00670969e+00 -1.30398542e-01 -5.89616835e-01 1.74373746e-01
7.24171847e-03 -8.91105771e-01 -9.94209290e-01 6.23984516e-01
1.31651103e+00 -7.15220213e-01 1.54433000e+00 -1.92056999e-01
1.76307205e-02 -4.08688426e-01 2.52284616e-01 -8.47512841e-01
-4.85864908e-01 -4.34547246e-01 -7.17510462e-01 9.67682481e-01
-2.49102470e-02 -3.51548463e-01 1.18985343e+00 5.47696233e-01
2.88615704e-01 -8.69463563e-01 -1.05869806e+00 -5.89250982e-01
-1.22932121e-01 2.12244596e-03 1.55074149e-01 4.17948663e-01
-1.26974970e-01 4.37544525e-01 -6.75103664e-01 2.39015028e-01
9.58451569e-01 1.64483814e-03 1.14573550e+00 -1.30219364e+00
-2.63531506e-01 -2.69710779e-01 -5.82590044e-01 -1.10103345e+00
-2.46775538e-01 -2.28987396e-01 5.58925187e-03 -1.29335415e+00
3.65192860e-01 7.76983723e-02 2.75344122e-02 7.26109087e-01
-3.99953991e-01 8.09367001e-01 2.85741240e-01 3.25375378e-01
-3.93779039e-01 1.07583255e-01 1.22158599e+00 1.25312969e-01
8.76652524e-02 2.32861623e-01 -4.22054738e-01 1.20796537e+00
5.03659964e-01 -3.68663162e-01 1.73442096e-01 -4.48160507e-02
2.15242207e-02 2.45512977e-01 7.42590725e-01 -1.13149285e+00
3.03094596e-01 1.57201260e-01 1.22186160e+00 -1.07613647e+00
7.54197299e-01 -7.05628395e-01 3.14472020e-01 7.38453150e-01
4.08686727e-01 8.51174444e-02 1.16023973e-01 4.00186062e-01
9.34128389e-02 1.78361833e-01 9.16648388e-01 -2.82948077e-01
-5.70516646e-01 6.34118974e-01 -1.99778918e-02 -1.95644572e-02
1.11588240e+00 -3.96346211e-01 7.04341009e-02 -2.66415298e-01
-1.09336019e+00 2.87062138e-01 4.23245817e-01 4.07049745e-01
5.21739066e-01 -1.21337259e+00 -7.47101605e-01 9.86230373e-02
8.42882842e-02 1.54408351e-01 4.60468739e-01 1.23001015e+00
-4.51363474e-01 2.00142726e-01 -1.41412541e-01 -7.30887234e-01
-1.85799587e+00 2.51490146e-01 3.80795002e-01 -4.46880609e-02
-7.06225157e-01 1.12558973e+00 3.15376401e-01 -2.65247911e-01
2.81242430e-01 -5.24070002e-02 -3.01008336e-02 3.67836773e-01
4.75272685e-01 7.03604102e-01 -9.34811160e-02 -1.27668178e+00
-8.22791100e-01 1.51382971e+00 -3.92886549e-02 -4.94908430e-02
8.61238480e-01 -5.35243787e-02 -1.35921672e-01 8.39919969e-02
8.79679918e-01 4.17556047e-01 -1.28734660e+00 -8.30109790e-02
-3.45012516e-01 -4.34230655e-01 -4.42593932e-01 -7.30028391e-01
-1.09051239e+00 7.99766839e-01 6.86379254e-01 -6.19488180e-01
8.60952139e-01 3.40661526e-01 7.12020338e-01 -1.17993221e-01
6.66353345e-01 -1.23442650e+00 1.27614215e-01 3.84812862e-01
9.74329412e-01 -1.05468071e+00 5.52279294e-01 -9.49987113e-01
-5.50301671e-01 1.00686884e+00 8.94394279e-01 -2.21576139e-01
6.61777139e-01 1.61838755e-01 -1.97742246e-02 -4.60679948e-01
-7.71095753e-02 -2.55343705e-01 9.48415756e-01 2.81495333e-01
6.05502725e-01 1.58842370e-01 -2.54350305e-01 6.25947833e-01
-3.85883421e-01 -2.21982419e-01 -3.79464567e-01 1.04919767e+00
-3.24211895e-01 -1.03624856e+00 -1.05899203e+00 2.52965569e-01
-7.05128074e-01 1.69628516e-01 -5.74792504e-01 1.20822549e+00
2.92862743e-01 4.87958401e-01 -8.77571851e-02 -2.82469004e-01
6.31184757e-01 1.75637692e-01 8.69426966e-01 -6.71159267e-01
-5.26428878e-01 6.87888443e-01 -9.80727468e-03 -8.75720322e-01
-5.50517328e-02 -8.58890116e-01 -1.40249455e+00 -3.69986415e-01
-3.65291983e-01 -1.30657062e-01 4.54772651e-01 6.19358599e-01
8.45536590e-02 5.02536520e-02 -1.68367609e-01 -1.30491507e+00
-4.86053467e-01 -9.98401701e-01 -7.92919099e-01 3.10003847e-01
4.04655635e-02 -1.17540765e+00 -2.32326910e-02 7.96810314e-02]
|
[7.057460784912109, -0.8918129205703735]
|
3bdd8e1d-e2a9-4ad3-b8d5-ebf725f73693
|
a-novel-brain-decoding-method-a-correlation
|
1712.01668
| null |
http://arxiv.org/abs/1712.01668v1
|
http://arxiv.org/pdf/1712.01668v1.pdf
|
A Novel Brain Decoding Method: a Correlation Network Framework for Revealing Brain Connections
|
Brain decoding is a hot spot in cognitive science, which focuses on
reconstructing perceptual images from brain activities. Analyzing the
correlations of collected data from human brain activities and representing
activity patterns are two problems in brain decoding based on functional
magnetic resonance imaging (fMRI) signals. However, existing correlation
analysis methods mainly focus on the strength information of voxel, which
reveals functional connectivity in the cerebral cortex. They tend to neglect
the structural information that implies the intracortical or intrinsic
connections; that is, structural connectivity. Hence, the effective
connectivity inferred by these methods is relatively unilateral. Therefore, we
proposed a correlation network (CorrNet) framework that could be flexibly
combined with diverse pattern representation models. In the CorrNet framework,
the topological correlation was introduced to reveal structural information.
Rich correlations were obtained, which contributed to specifying the underlying
effective connectivity. We also combined the CorrNet framework with a linear
support vector machine (SVM) and a dynamic evolving spike neuron network (SNN)
for pattern representation separately, thus providing a novel method for
decoding cognitive activity patterns. Experimental results verified the
reliability and robustness of our CorrNet framework and demonstrated that the
new method achieved significant improvement in brain decoding over comparable
methods.
|
['Badong Chen', 'Hao Wu', 'Yongqiang Ma', 'Siyu Yu', 'Nanning Zheng']
|
2017-12-01
| null | null | null | null |
['brain-decoding', 'brain-decoding']
|
['medical', 'miscellaneous']
|
[ 3.80934894e-01 -3.24380189e-01 1.15904003e-01 -2.74011403e-01
4.70403016e-01 -4.49943513e-01 6.68011665e-01 -9.53332782e-02
-1.76180601e-01 7.41751432e-01 1.40168846e-01 -8.65569711e-02
-5.47473848e-01 -7.78573036e-01 -4.52855855e-01 -9.55773473e-01
-3.81156564e-01 -5.49574532e-02 2.13806242e-01 -1.35313064e-01
5.26841044e-01 2.16729835e-01 -1.50686431e+00 2.56183058e-01
9.74931240e-01 1.10979033e+00 5.00766158e-01 1.50423825e-01
-2.83569008e-01 5.58148742e-01 -4.76707757e-01 1.10101394e-01
-8.39472190e-02 -7.69873083e-01 -3.23851556e-01 -1.52563214e-01
-2.77203619e-01 3.40863585e-01 -4.70466316e-01 1.38989651e+00
3.10876817e-01 -7.28224069e-02 6.09333038e-01 -1.26344013e+00
-5.85832894e-01 7.65912712e-01 -4.18945700e-01 7.88267016e-01
3.75857323e-01 1.11738116e-01 8.07739794e-01 -6.91228271e-01
4.43309605e-01 1.01583493e+00 3.58433902e-01 1.73205853e-01
-1.21488202e+00 -8.59446526e-01 1.85570821e-01 5.64842522e-01
-1.22910964e+00 -1.27524525e-01 9.97307837e-01 -8.37556899e-01
7.93168664e-01 3.00087065e-01 1.45216966e+00 1.13926625e+00
6.33381486e-01 4.56934482e-01 1.61231077e+00 -1.61701012e-02
2.49153942e-01 -2.95083284e-01 3.62746239e-01 5.71418464e-01
3.10442656e-01 9.53740552e-02 -7.13428915e-01 7.29385689e-02
1.02815497e+00 3.11229937e-02 -5.29896617e-01 -2.38825083e-01
-1.78004050e+00 4.80211169e-01 7.13224888e-01 7.13619947e-01
-3.73759240e-01 -1.44238830e-01 4.03963357e-01 2.17245966e-01
2.73573101e-01 3.22360933e-01 -1.47499546e-01 -3.22161503e-02
-9.33045506e-01 -1.98841542e-01 6.16704822e-01 3.71899754e-01
5.43554485e-01 2.81568915e-01 -2.24082530e-01 7.71399617e-01
2.51548856e-01 4.59771842e-01 8.21372449e-01 -6.69603944e-01
1.21304609e-01 7.99078405e-01 -4.75063443e-01 -1.42880261e+00
-7.23432899e-01 -6.72006667e-01 -1.25321925e+00 -6.05376847e-02
1.55381680e-01 1.03853203e-01 -5.47522485e-01 1.81448615e+00
-9.80200693e-02 4.49886531e-01 -2.13243902e-01 1.12281883e+00
6.66074514e-01 3.84667873e-01 -1.56833977e-01 -5.30379415e-01
1.25147855e+00 -6.46718085e-01 -9.28225577e-01 5.26077067e-03
3.03438395e-01 -2.01023027e-01 7.43929803e-01 3.72391135e-01
-9.33982790e-01 -5.20966828e-01 -1.23120403e+00 5.16580701e-01
-2.59888649e-01 -2.65061319e-01 9.12715912e-01 3.47759873e-01
-1.05595589e+00 5.88376343e-01 -8.85048747e-01 -2.03063935e-01
4.82155055e-01 3.23530138e-01 -4.98668790e-01 1.18404530e-01
-1.15560317e+00 9.02382314e-01 5.44777215e-01 4.34498459e-01
-5.53898394e-01 -3.87884587e-01 -6.00928545e-01 2.28657871e-01
5.67701794e-02 -5.62663198e-01 3.52181405e-01 -1.06723857e+00
-1.34428072e+00 3.16262096e-01 -2.82891273e-01 -1.89608812e-01
1.86229855e-01 3.32327485e-01 -5.96147835e-01 1.63421184e-01
-6.65934309e-02 5.68307281e-01 4.65734333e-01 -1.04341614e+00
1.10884093e-01 -4.34777349e-01 -3.81940454e-01 1.75309211e-01
-3.18964243e-01 -2.97446847e-01 -3.71397063e-02 -5.97416461e-01
8.58201146e-01 -6.21306181e-01 7.45273083e-02 -1.50929391e-01
-2.23284319e-01 -3.64937596e-02 5.42861760e-01 -8.26967955e-01
1.25008488e+00 -2.17840052e+00 6.23111367e-01 6.40801609e-01
5.38569987e-01 -6.62324354e-02 -1.60308525e-01 1.36245340e-01
-3.31333786e-01 -8.86922032e-02 -6.00178838e-01 4.34368372e-01
-3.16517085e-01 1.66558906e-01 -3.37902345e-02 6.41936541e-01
1.33846134e-01 1.08171070e+00 -9.76390779e-01 -3.08408856e-01
1.12960048e-01 5.04481912e-01 -4.64236557e-01 3.92023660e-02
1.45260543e-01 1.00211084e+00 -1.87477872e-01 6.97404683e-01
5.76362669e-01 -1.80552110e-01 3.42014194e-01 -5.27881324e-01
-1.45447776e-01 5.23811877e-02 -8.65020096e-01 1.87255657e+00
-6.83558509e-02 7.41994023e-01 -1.83310270e-01 -1.60365355e+00
1.26775205e+00 2.11833239e-01 7.29096591e-01 -1.12793660e+00
3.67625594e-01 1.65376186e-01 8.88585150e-01 -6.70445323e-01
-2.52458811e-01 1.56154037e-01 2.95150846e-01 4.13049936e-01
2.17256799e-01 4.28705603e-01 9.84971449e-02 -3.86407264e-02
1.22653401e+00 1.58571620e-02 2.60881573e-01 -5.74489713e-01
6.20683908e-01 -3.43521059e-01 5.32290041e-01 3.23098093e-01
-2.73669094e-01 3.62372071e-01 7.70878851e-01 -2.60157436e-01
-8.17264915e-01 -1.17922759e+00 -3.43519211e-01 3.66445333e-01
3.72458369e-01 -2.38930434e-01 -6.78208351e-01 -2.02121466e-01
-3.40411276e-01 2.14358658e-01 -6.89792275e-01 -2.68057674e-01
-5.11589408e-01 -9.93623257e-01 4.38288331e-01 4.09723759e-01
8.07222486e-01 -1.17258346e+00 -5.40940106e-01 2.82385916e-01
-3.26003999e-01 -9.26120877e-01 -2.27561519e-01 1.83600783e-01
-1.11654472e+00 -1.22563374e+00 -4.93424088e-01 -6.86722398e-01
9.24134552e-01 2.73253918e-01 5.65140605e-01 2.14254647e-01
-3.91859591e-01 8.65016580e-02 -2.14743704e-01 2.04497740e-01
1.24404840e-01 -3.13613266e-01 2.10758731e-01 2.23522201e-01
2.75153816e-01 -1.13361025e+00 -7.04344213e-01 3.49879622e-01
-8.37529123e-01 4.04229373e-01 7.56619453e-01 1.04633927e+00
4.02597725e-01 6.43558726e-02 6.48529232e-01 -4.36339945e-01
7.75292277e-01 -7.13339090e-01 -3.57441276e-01 2.65353799e-01
-6.23918653e-01 2.18651459e-01 4.47818667e-01 -5.99570036e-01
-8.68750215e-01 -2.92250067e-01 9.26962271e-02 -2.00483933e-01
4.67257798e-02 8.92008901e-01 -3.57640535e-01 -1.29327372e-01
2.80902773e-01 8.44076931e-01 2.22691774e-01 -2.20660895e-01
-1.84720773e-02 3.45995665e-01 6.17164016e-01 -5.29702723e-01
2.00865433e-01 2.76306480e-01 2.31574818e-01 -6.19537234e-01
-2.31617749e-01 -5.06483689e-02 -9.33404624e-01 -5.75824201e-01
7.62906730e-01 -6.57176673e-01 -8.74166489e-01 4.40986037e-01
-1.11680520e+00 4.27132696e-02 1.97040319e-01 8.05884063e-01
-3.42714667e-01 4.97553170e-01 -5.41170180e-01 -6.90137863e-01
6.11956697e-03 -1.08150208e+00 4.26853895e-01 6.96932077e-02
-1.47751063e-01 -1.15307903e+00 8.71326029e-02 1.18568070e-01
5.83287835e-01 4.45383757e-01 1.05330527e+00 -3.66131425e-01
-4.63700086e-01 6.87687546e-02 -4.58964676e-01 7.83856586e-02
3.36913504e-02 -2.40141258e-01 -6.63594782e-01 2.60671284e-02
2.01495633e-01 2.60167569e-01 8.62972140e-01 3.44566941e-01
1.35360372e+00 -2.97859371e-01 -3.68613452e-01 5.79273283e-01
1.27335012e+00 4.13103074e-01 7.57407367e-01 7.85993487e-02
6.20522738e-01 8.26000631e-01 -6.32869452e-02 2.72391319e-01
2.17104509e-01 5.54832518e-01 3.48507494e-01 1.53617114e-01
-1.39556751e-01 -1.77861810e-01 4.39487666e-01 1.45947576e+00
-5.26606858e-01 2.14210555e-01 -8.57977331e-01 2.45924979e-01
-1.98089314e+00 -1.13397086e+00 -3.19845408e-01 1.96462572e+00
6.50374413e-01 1.33963764e-01 -2.12647632e-01 2.86886752e-01
8.52510393e-01 4.19280566e-02 -6.31240070e-01 -1.09184615e-01
-5.25949836e-01 1.47840336e-01 1.52406678e-01 1.58822730e-01
-3.99208635e-01 4.82998818e-01 6.63069534e+00 5.30624568e-01
-1.28357959e+00 4.13116664e-01 1.45400003e-01 -7.76763856e-02
-4.41636652e-01 -7.25620165e-02 -1.48180753e-01 8.66147757e-01
8.09312761e-01 -2.01463372e-01 8.18583727e-01 3.37696731e-01
2.59863496e-01 -1.80011362e-01 -9.11643922e-01 1.23572040e+00
1.05161063e-01 -1.23831367e+00 4.99215163e-03 2.10194096e-01
5.17595232e-01 -1.34898871e-01 -8.41892511e-02 -6.36930298e-03
-1.80637121e-01 -1.13765740e+00 5.33445060e-01 1.05381131e+00
5.03521383e-01 -2.49530166e-01 6.60925448e-01 4.16729689e-01
-1.24805570e+00 -1.51793852e-01 -3.39800864e-01 -4.25012946e-01
1.37605906e-01 8.17163825e-01 -1.46387294e-01 4.63347584e-01
5.95306337e-01 1.28063679e+00 -5.59429467e-01 1.21906364e+00
-1.64239764e-01 6.28502011e-01 -2.15145089e-02 -9.78112370e-02
-1.02722153e-01 -5.06764650e-01 5.18579304e-01 1.12074208e+00
5.34451663e-01 9.80532244e-02 -1.35461241e-01 1.22772956e+00
2.94538409e-01 2.61279717e-02 -6.43515944e-01 -5.12615032e-02
4.39832360e-01 1.34066069e+00 -1.27933836e+00 -1.60984442e-01
-3.40253085e-01 8.73086870e-01 3.25690448e-01 3.68055344e-01
-7.71358073e-01 -1.05269313e-01 3.29484820e-01 -3.65103339e-03
-3.22660580e-02 -6.11548245e-01 -6.88382208e-01 -1.54787946e+00
1.69705540e-01 -3.79889816e-01 -9.74561796e-02 -8.71450245e-01
-1.23177886e+00 7.73323596e-01 1.46286279e-01 -1.20846820e+00
2.38765195e-01 -6.52810335e-01 -6.86249435e-01 8.50575328e-01
-1.19125378e+00 -6.05891347e-01 -5.14010668e-01 7.96996534e-01
3.35110635e-01 -1.93153977e-01 7.03078866e-01 3.60508502e-01
-7.80270517e-01 1.36373013e-01 -8.90148878e-02 -3.77892442e-02
1.73545182e-01 -9.57720637e-01 -2.88216561e-01 8.04787755e-01
2.59680897e-01 7.61497796e-01 2.88586855e-01 -6.91339850e-01
-1.34065259e+00 -5.94019592e-01 5.69085181e-01 -8.29620883e-02
8.79819751e-01 -5.36906481e-01 -1.14504147e+00 2.60882348e-01
2.44901136e-01 1.07257880e-01 8.06006789e-01 -1.64985046e-01
-2.63487488e-01 -4.60597947e-02 -8.96312535e-01 5.14476478e-01
1.41516066e+00 -4.42562163e-01 -7.06298470e-01 -1.50096265e-03
3.58752489e-01 1.31923527e-01 -8.31250131e-01 3.10463697e-01
7.18542874e-01 -9.84451413e-01 7.93078661e-01 -1.36113539e-01
3.54004562e-01 -4.89430845e-01 -2.09724337e-01 -1.44977546e+00
-6.38117194e-01 1.79978237e-01 -3.48132133e-01 8.51013303e-01
1.69614628e-01 -9.10113335e-01 1.94480389e-01 2.27106512e-01
-1.36206284e-01 -6.60613000e-01 -9.73093331e-01 -7.62325764e-01
-2.95383036e-01 -3.57553214e-01 5.05031824e-01 1.11934114e+00
4.33434337e-01 1.58793584e-01 -2.39988163e-01 -4.49074209e-02
5.79377770e-01 -2.85657376e-01 -6.36002421e-02 -1.42959619e+00
-2.01391608e-01 -9.04240012e-01 -8.42581213e-01 -6.79387629e-01
4.28784817e-01 -1.39971483e+00 -2.05572054e-01 -1.55801582e+00
3.21369827e-01 -1.43237606e-01 -8.18378389e-01 3.92351210e-01
4.58285138e-02 2.69079357e-01 -3.41988876e-02 5.50495982e-01
-1.21342897e-01 6.22689962e-01 1.39769399e+00 -2.04776242e-01
-1.18377842e-01 -4.93072420e-01 -5.77323496e-01 4.53667641e-01
7.79877067e-01 -2.21847370e-01 -5.39659679e-01 -2.46843055e-01
2.36503229e-01 2.18453199e-01 6.09737217e-01 -1.24657464e+00
5.16532719e-01 1.92513373e-02 6.43914044e-01 -1.51632458e-01
1.65645137e-01 -7.07743049e-01 4.06693906e-01 7.11978793e-01
-2.08782390e-01 9.44114402e-02 -1.75110660e-02 6.25469446e-01
-4.66056168e-01 2.18584925e-01 4.37249422e-01 -5.73536381e-02
-7.12848306e-01 1.52721941e-01 -4.43026483e-01 -3.87697726e-01
9.42380369e-01 -6.32548571e-01 -3.63867939e-01 1.74033433e-01
-8.60105693e-01 -9.19455662e-02 -5.51358275e-02 3.98865074e-01
8.80289912e-01 -1.53471529e+00 -5.04080594e-01 5.90672076e-01
-1.63176954e-01 -6.15916371e-01 3.10403436e-01 1.58217049e+00
-2.65535355e-01 4.68185723e-01 -9.67670619e-01 -8.65727663e-01
-8.71775389e-01 4.42954123e-01 2.60609925e-01 1.55121580e-01
-7.54849970e-01 4.11915541e-01 1.87308311e-01 -4.01785783e-02
-7.41545409e-02 -2.54366904e-01 -7.40704775e-01 3.03446501e-01
3.82795125e-01 2.13708401e-01 -5.28001823e-02 -7.73181856e-01
-5.52186608e-01 6.16828382e-01 3.66009831e-01 -2.28231177e-01
1.30719531e+00 -1.72889128e-01 -6.54093862e-01 7.70192564e-01
1.02693880e+00 -4.73989576e-01 -1.12524414e+00 -2.26971403e-01
-3.70427445e-02 -4.44552183e-01 1.59084722e-01 -7.58767128e-01
-1.32847536e+00 9.88308430e-01 6.11711681e-01 8.18605796e-02
1.24066627e+00 -1.81244895e-01 4.35741037e-01 1.78575471e-01
7.12275982e-01 -7.27690101e-01 2.74110585e-01 2.73006678e-01
8.97099972e-01 -9.49007213e-01 -1.73033968e-01 -3.16115469e-01
-5.16033709e-01 1.37513876e+00 7.20063925e-01 -2.17731267e-01
9.81377065e-01 6.28682002e-02 -5.00949383e-01 -3.94215822e-01
-6.06453478e-01 4.99048419e-02 4.58709151e-01 4.60093230e-01
5.03961921e-01 2.53588021e-01 -7.92281449e-01 9.17186022e-01
-2.53633887e-01 -1.01832159e-01 2.90270299e-01 7.29389071e-01
-5.69491744e-01 -8.04110348e-01 -3.04260433e-01 7.04599023e-01
4.16241921e-02 -1.66813001e-01 -2.67190859e-02 3.75606447e-01
3.14800948e-01 7.18995035e-01 2.64164686e-01 -7.11304724e-01
-3.85254249e-02 1.36417195e-01 7.92683363e-01 -3.79142731e-01
-3.15428406e-01 4.06056680e-02 -4.21283454e-01 -6.50681734e-01
-7.64055431e-01 -7.19964564e-01 -1.20944679e+00 -6.85525164e-02
-1.98127374e-01 1.17988445e-01 8.41819108e-01 1.16424966e+00
2.22572580e-01 9.42810178e-01 5.11403680e-01 -8.27498794e-01
2.85687506e-01 -1.08570659e+00 -8.11104357e-01 1.16722293e-01
-4.22575958e-02 -1.07587385e+00 -3.41014534e-01 3.75316702e-02]
|
[12.620474815368652, 3.4133052825927734]
|
319f5bd7-2d85-418d-82af-d2d7fe265292
|
neural-implicit-dense-semantic-slam
|
2304.14560
| null |
https://arxiv.org/abs/2304.14560v2
|
https://arxiv.org/pdf/2304.14560v2.pdf
|
Neural Implicit Dense Semantic SLAM
|
Visual Simultaneous Localization and Mapping (vSLAM) is a widely used technique in robotics and computer vision that enables a robot to create a map of an unfamiliar environment using a camera sensor while simultaneously tracking its position over time. In this paper, we propose a novel RGBD vSLAM algorithm that can learn a memory-efficient, dense 3D geometry, and semantic segmentation of an indoor scene in an online manner. Our pipeline combines classical 3D vision-based tracking and loop closing with neural fields-based mapping. The mapping network learns the SDF of the scene as well as RGB, depth, and semantic maps of any novel view using only a set of keyframes. Additionally, we extend our pipeline to large scenes by using multiple local mapping networks. Extensive experiments on well-known benchmark datasets confirm that our approach provides robust tracking, mapping, and semantic labeling even with noisy, sparse, or no input depth. Overall, our proposed algorithm can greatly enhance scene perception and assist with a range of robot control problems.
|
['Jean-Philippe Thiran', 'Luc van Gool', 'Suryansh Kumar', 'Yasaman Haghighi']
|
2023-04-27
| null | null | null | null |
['simultaneous-localization-and-mapping', 'semantic-slam']
|
['computer-vision', 'computer-vision']
|
[ 1.91662282e-01 -1.91404462e-01 7.19075371e-03 -4.33629632e-01
-3.68444264e-01 -9.90227878e-01 3.67158234e-01 2.76149005e-01
-5.77244222e-01 2.38618195e-01 -4.94736999e-01 -2.45011806e-01
1.47461772e-01 -6.97146118e-01 -1.16227782e+00 -3.01279038e-01
2.72985488e-01 7.38190591e-01 7.01923847e-01 -6.91895559e-02
3.69933218e-01 8.95647109e-01 -1.50122881e+00 -3.53195369e-01
4.74120855e-01 1.16821837e+00 1.04462421e+00 5.07634163e-01
-6.47679865e-02 6.88583076e-01 -1.49009749e-01 3.20686251e-01
6.11960173e-01 1.40162379e-01 -7.89388835e-01 4.10408854e-01
7.35136747e-01 -4.50762600e-01 -5.51918328e-01 1.10297322e+00
1.38164148e-01 3.62148464e-01 -8.83057863e-02 -1.15537190e+00
-2.94770420e-01 -6.27724156e-02 -7.08410561e-01 -1.34202734e-01
8.35399210e-01 2.62900889e-01 3.28203678e-01 -8.14557850e-01
8.94058228e-01 1.28850198e+00 8.59291077e-01 2.00486422e-01
-1.07431090e+00 -4.53393161e-01 4.94359463e-01 -3.22562568e-02
-1.41930544e+00 -1.02103807e-01 8.49186897e-01 -3.75809014e-01
1.09157145e+00 -2.35157460e-01 9.25925672e-01 6.49221420e-01
1.28167883e-01 5.19392014e-01 9.85070586e-01 -2.86579013e-01
4.68025953e-01 -1.55508950e-01 -2.06385091e-01 1.12717557e+00
1.67872887e-02 8.71927142e-02 -6.88735008e-01 1.63196266e-01
1.22125196e+00 4.42722589e-01 -1.54310420e-01 -1.20999265e+00
-1.61171484e+00 4.34330106e-01 9.22026277e-01 -2.36397088e-01
-2.22448334e-01 5.29266596e-01 -1.66471992e-02 1.09554000e-01
7.31229708e-02 2.90039986e-01 -5.20931900e-01 1.19917728e-01
-5.74247241e-01 1.00646965e-01 5.65484226e-01 1.41006410e+00
1.39556944e+00 -2.39323393e-01 4.67931151e-01 1.90380290e-01
2.44557500e-01 9.22716677e-01 5.74339405e-02 -1.66155589e+00
2.46932134e-01 6.94088459e-01 2.86384702e-01 -1.19183278e+00
-7.49573886e-01 6.13625981e-02 -4.47536379e-01 5.63960612e-01
2.56301910e-01 2.04534486e-01 -1.22142613e+00 1.42221105e+00
7.00900674e-01 3.87682766e-01 -8.99291635e-02 9.80126262e-01
6.15777969e-01 3.78821939e-01 -4.56952155e-01 1.53973430e-01
8.51036429e-01 -1.11898613e+00 -3.97107631e-01 -9.69578505e-01
2.59273559e-01 -5.44392645e-01 7.37372816e-01 3.15447509e-01
-7.03098238e-01 -6.98312759e-01 -8.94599974e-01 -5.19352734e-01
-4.12932217e-01 -2.74765547e-02 9.39426005e-01 1.28044769e-01
-1.35930097e+00 3.49432170e-01 -1.23078740e+00 -7.32416272e-01
3.98811340e-01 3.83196473e-01 -8.51151407e-01 -4.91610199e-01
-4.11893785e-01 9.22554612e-01 7.27430165e-01 7.02322721e-02
-1.07932472e+00 -4.59545016e-01 -1.25897872e+00 -6.00372970e-01
5.18720567e-01 -8.35686803e-01 1.21760404e+00 -3.35097283e-01
-1.46279943e+00 8.66979778e-01 -3.54142696e-01 -3.01570803e-01
3.87815177e-01 -4.70162451e-01 4.66838360e-01 2.40076900e-01
3.10887307e-01 1.21841228e+00 4.97335643e-01 -1.42061746e+00
-8.91284466e-01 -8.18615913e-01 1.67734846e-01 5.53408206e-01
2.86579996e-01 -6.08614624e-01 -8.53779078e-01 1.73198745e-01
1.03275084e+00 -1.05579364e+00 -6.03254557e-01 5.88642895e-01
-3.54804903e-01 -5.60743846e-02 8.82688105e-01 -3.61419678e-01
1.78355664e-01 -1.98604369e+00 2.98433453e-01 1.47007510e-01
1.14962175e-01 -3.44541579e-01 1.82191357e-01 1.20213553e-01
5.54248989e-01 -4.53718424e-01 -1.38404727e-01 -6.86453819e-01
-1.57263309e-01 5.11932194e-01 -2.89817363e-01 8.12323153e-01
-2.86657304e-01 9.24834549e-01 -1.27910793e+00 -3.05658728e-01
8.68952990e-01 5.75252831e-01 -4.80027080e-01 1.69376254e-01
-5.04360735e-01 8.59944880e-01 -3.53461981e-01 8.23080540e-01
8.00108433e-01 -3.08759481e-01 -4.22705412e-02 -1.07161157e-01
-4.31072950e-01 1.00170836e-01 -1.21057117e+00 3.07381511e+00
-4.23248887e-01 8.13514650e-01 6.65440336e-02 -6.05160475e-01
1.00479305e+00 -3.76617014e-01 4.71531779e-01 -6.81837320e-01
2.91550905e-01 1.77078739e-01 -8.87826800e-01 -1.49627402e-01
6.90591812e-01 4.02499974e-01 -1.38248116e-01 2.48911962e-01
7.48892128e-02 -6.24191284e-01 -1.06443428e-01 1.17270902e-01
1.18625140e+00 6.70662165e-01 2.72875339e-01 1.95005655e-01
1.31728619e-01 6.71789765e-01 4.70642656e-01 8.63736391e-01
-1.42966047e-01 4.71269071e-01 -2.85565674e-01 -6.64556980e-01
-9.68729138e-01 -1.17372155e+00 9.44477096e-02 7.57857680e-01
1.08450091e+00 -5.82840405e-02 -2.65588015e-01 -4.22196716e-01
2.48334199e-01 2.26864874e-01 -4.14941162e-01 1.25340983e-01
-5.58721364e-01 1.14639886e-01 -2.11018920e-02 6.26509607e-01
6.56233907e-01 -8.64898503e-01 -1.36928296e+00 3.94952595e-02
-1.66268826e-01 -1.61965704e+00 -2.53670037e-01 6.43619776e-01
-8.63582373e-01 -1.16868544e+00 -2.18457937e-01 -1.16449261e+00
9.04721379e-01 9.89659846e-01 7.65134931e-01 -3.22413146e-01
-2.71123976e-01 6.59357965e-01 -2.31568798e-01 -1.17491640e-01
-5.36914580e-02 -7.42194131e-02 1.56571940e-01 -5.59076130e-01
2.36373693e-01 -4.99827445e-01 -4.52225626e-01 1.06410414e-01
-4.25791889e-01 2.36881092e-01 4.69553947e-01 2.91103244e-01
1.22036004e+00 -7.12453872e-02 -3.51284206e-01 -5.22837877e-01
-1.58916131e-01 -1.42868981e-01 -1.19840813e+00 1.65994465e-02
-4.70089138e-01 -1.01385906e-01 2.69359380e-01 -2.01541036e-01
-7.67347634e-01 1.10841429e+00 1.68149963e-01 -8.55298936e-01
-4.76528347e-01 8.39661881e-02 -3.88226053e-03 -6.03610337e-01
5.60057878e-01 2.78817028e-01 -2.17641387e-02 -5.26413202e-01
7.51335979e-01 3.17962110e-01 1.08076036e+00 -2.12012723e-01
9.48510528e-01 1.06391716e+00 1.27248392e-01 -2.98423469e-01
-9.52275276e-01 -9.73234415e-01 -1.38581324e+00 -1.56920746e-01
8.38478923e-01 -1.33671844e+00 -9.44543242e-01 5.09817660e-01
-1.39389133e+00 -5.86626887e-01 -1.25150949e-01 4.28652763e-01
-8.41084778e-01 1.80445582e-01 -2.66491532e-01 -5.41201949e-01
-1.11141512e-02 -1.12364745e+00 1.48058176e+00 3.70104939e-01
5.82712777e-02 -8.89099419e-01 1.45202979e-01 1.29280046e-01
-7.58876652e-02 6.11584961e-01 1.68131977e-01 7.12276846e-02
-1.19326437e+00 -2.43073516e-02 -2.92863965e-01 -1.49634838e-01
2.90167451e-01 -5.35677493e-01 -9.03577387e-01 -4.18071568e-01
-2.72566140e-01 -3.42433512e-01 8.14776480e-01 2.53273040e-01
9.48672473e-01 1.99481592e-01 -7.44966686e-01 1.18158996e+00
1.69383764e+00 3.27276349e-01 2.51631439e-02 7.30029404e-01
1.05266023e+00 3.12130630e-01 8.75099838e-01 3.95669997e-01
7.61325359e-01 5.37038088e-01 1.01521862e+00 -1.35535091e-01
8.91999528e-02 -5.78532219e-01 1.29549682e-01 4.09743994e-01
4.59237754e-01 1.08776338e-01 -1.07453215e+00 5.00313044e-01
-1.94301105e+00 -4.17261302e-01 6.46422058e-02 2.07251358e+00
5.44168830e-01 7.20583722e-02 -3.41618180e-01 -1.51623994e-01
5.60185194e-01 -2.77106371e-03 -1.09474564e+00 1.07660428e-01
-1.88709050e-02 -7.71507993e-02 1.13752389e+00 7.29465842e-01
-1.29715753e+00 1.37247729e+00 5.80046463e+00 -2.98520885e-02
-1.06974292e+00 1.19615234e-02 4.15044371e-03 6.58327527e-03
-2.99401898e-02 8.38684067e-02 -8.61324549e-01 -8.58392566e-03
2.57217318e-01 2.06144542e-01 6.62964642e-01 1.17912292e+00
-7.71418065e-02 -5.79682708e-01 -1.19164979e+00 1.43071306e+00
1.59500882e-01 -1.47641408e+00 -3.41337144e-01 -2.14468781e-02
8.71223688e-01 6.33630812e-01 -5.13433032e-02 -2.19713211e-01
7.23164856e-01 -6.90893471e-01 1.07560217e+00 3.70020568e-01
7.13093221e-01 -5.87421179e-01 5.58586776e-01 6.80479169e-01
-1.34234524e+00 -2.28080675e-01 -4.56561446e-01 -2.44100139e-01
1.66765451e-01 2.61381745e-01 -1.05840003e+00 3.54331732e-01
1.09020269e+00 1.14038014e+00 -6.33110344e-01 1.16261446e+00
-3.77519280e-01 -4.92098719e-01 -7.54234791e-01 1.70877934e-01
2.27494553e-01 -2.36845110e-02 3.48474115e-01 6.23050511e-01
3.35049897e-01 3.69420536e-02 7.90921211e-01 9.45496976e-01
1.34219229e-01 -4.61270422e-01 -8.92147779e-01 3.68133843e-01
9.34514642e-01 1.23458636e+00 -1.22323430e+00 2.56525949e-02
-1.73377424e-01 1.36296129e+00 4.55588311e-01 2.29367584e-01
-4.72251445e-01 -2.56925255e-01 4.93685812e-01 -1.41195819e-01
3.52323174e-01 -9.68430698e-01 -2.88098514e-01 -9.03623164e-01
-1.73664745e-02 -1.68574840e-01 -2.00665165e-02 -1.06670594e+00
-5.63712180e-01 4.94922310e-01 -2.54296571e-01 -1.17866564e+00
-1.06162474e-01 -5.99606335e-01 8.04331079e-02 6.12462044e-01
-1.94666135e+00 -1.23236644e+00 -1.12830937e+00 8.33842874e-01
6.36082411e-01 2.81084836e-01 5.86624265e-01 -6.74750358e-02
7.82953389e-03 -2.46078014e-01 -8.00475851e-02 1.26902491e-01
5.08209527e-01 -1.27145052e+00 6.17979884e-01 8.30210447e-01
2.46476963e-01 4.31673974e-01 4.27949727e-01 -7.05687642e-01
-1.81332040e+00 -1.31155789e+00 3.98480147e-01 -8.06909740e-01
3.12320143e-01 -6.10079229e-01 -4.90212321e-01 1.02675200e+00
-1.56249702e-01 3.38475108e-01 -9.13119242e-02 -3.86138946e-01
-2.16856062e-01 -1.63231447e-01 -1.09496605e+00 2.28715405e-01
1.47769570e+00 -6.01680696e-01 -3.94304574e-01 4.96136010e-01
1.18537331e+00 -1.33903968e+00 -5.41373372e-01 2.58020937e-01
4.06574845e-01 -9.34048116e-01 1.24077940e+00 2.12233633e-01
-1.81204349e-01 -8.34938407e-01 -4.62715983e-01 -1.15185964e+00
-1.79595381e-01 -5.55088162e-01 -1.73141900e-02 6.80760324e-01
-1.74281463e-01 -4.06999618e-01 9.90761042e-01 4.05939192e-01
-3.49534065e-01 -2.75550157e-01 -9.49455202e-01 -6.44174993e-01
-4.87075686e-01 -5.59086740e-01 5.10760248e-01 7.15605378e-01
-3.89272928e-01 2.62867194e-02 -8.42714757e-02 7.53757000e-01
1.02269423e+00 4.37614441e-01 1.11771977e+00 -1.23618388e+00
1.68818370e-01 -3.98348831e-02 -7.50847220e-01 -1.61688864e+00
3.36953580e-01 -7.97215104e-01 6.61581516e-01 -1.96547782e+00
-1.00515626e-01 -7.24520862e-01 2.01698113e-02 7.40977287e-01
2.90104955e-01 2.53115237e-01 9.56822783e-02 2.94412136e-01
-1.10531604e+00 3.30430388e-01 1.17603540e+00 -9.13313106e-02
-4.18257087e-01 -2.72354037e-01 -1.66430444e-01 8.50585043e-01
5.80885649e-01 -4.05996025e-01 -4.44105178e-01 -9.69711781e-01
2.75339514e-01 -2.01271567e-02 7.00187385e-01 -1.26799452e+00
8.90965521e-01 -2.60054499e-01 7.03290105e-01 -1.01876557e+00
6.82991564e-01 -1.18479013e+00 1.59749761e-01 5.10412157e-01
6.36462867e-02 2.74410725e-01 2.08652705e-01 9.43969071e-01
5.69162257e-02 1.93609610e-01 5.43918967e-01 -6.48052692e-01
-1.72151315e+00 4.70733196e-01 9.61139202e-02 -2.73431480e-01
1.25157714e+00 -4.51311767e-01 -1.83460444e-01 -1.04732908e-01
-5.06720841e-01 4.97201562e-01 1.12590086e+00 5.94374955e-01
1.04507005e+00 -1.19070077e+00 6.26903847e-02 5.38141489e-01
2.82906860e-01 1.08182085e+00 -5.86061133e-03 6.45581245e-01
-1.10786390e+00 4.87034321e-01 -2.94031948e-01 -1.34157145e+00
-9.48541164e-01 5.65771699e-01 2.62501597e-01 5.20134032e-01
-9.47919667e-01 1.00960696e+00 5.62658906e-02 -9.39066529e-01
5.52474737e-01 -4.99683648e-01 2.06863165e-01 -4.67831343e-01
3.23645413e-01 3.99819732e-01 -7.25390241e-02 -8.83600533e-01
-6.38257027e-01 9.38993871e-01 3.93652499e-01 4.36160415e-02
1.17507946e+00 -7.67101467e-01 -2.95265198e-01 7.31063187e-01
1.10880554e+00 -3.34395677e-01 -1.84676492e+00 -4.92878526e-01
-6.66807592e-02 -7.32436359e-01 3.24690968e-01 -5.01576364e-01
-8.57086718e-01 7.37584174e-01 7.89790988e-01 -3.49297047e-01
8.98424208e-01 2.84886092e-01 8.21048379e-01 8.03146183e-01
1.23629248e+00 -9.62500870e-01 2.19583005e-01 8.18140566e-01
4.31564093e-01 -1.49441791e+00 3.76813821e-02 -3.25719863e-01
-1.96084216e-01 1.14610267e+00 8.59263361e-01 -1.05875641e-01
4.17074025e-01 3.21867794e-01 2.61702269e-01 -1.23866320e-01
-2.17488036e-01 -3.87975335e-01 -9.08921883e-02 9.82728541e-01
-4.97866422e-01 -2.34188065e-01 9.23280895e-01 -2.75792211e-01
-2.61238962e-01 -1.42646164e-01 4.30075735e-01 1.30575967e+00
-8.47448349e-01 -6.72363281e-01 -4.14180338e-01 -2.03579426e-01
2.67870277e-01 1.53265372e-01 -3.52042168e-01 6.87758207e-01
3.40019971e-01 8.98685932e-01 3.86798590e-01 -5.51623106e-01
2.76726872e-01 -2.79785037e-01 7.84698009e-01 -8.22577953e-01
3.58913210e-03 -6.94828480e-02 -6.23350143e-01 -1.26308024e+00
-6.48131251e-01 -6.31894946e-01 -1.82352865e+00 -1.26671135e-01
-2.27190241e-01 -3.93914402e-01 1.16486549e+00 1.01261961e+00
4.53430623e-01 3.16172540e-01 5.71935654e-01 -1.30170977e+00
9.10369381e-02 -4.13785160e-01 -4.26303893e-01 9.83260348e-02
7.97490895e-01 -7.76644409e-01 -5.44184521e-02 1.80805683e-01]
|
[7.643080234527588, -2.383608102798462]
|
5e0d5358-8efa-4344-83fa-a8adcd39a830
|
distilbert-a-distilled-version-of-bert
|
1910.01108
| null |
https://arxiv.org/abs/1910.01108v4
|
https://arxiv.org/pdf/1910.01108v4.pdf
|
DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
|
As Transfer Learning from large-scale pre-trained models becomes more prevalent in Natural Language Processing (NLP), operating these large models in on-the-edge and/or under constrained computational training or inference budgets remains challenging. In this work, we propose a method to pre-train a smaller general-purpose language representation model, called DistilBERT, which can then be fine-tuned with good performances on a wide range of tasks like its larger counterparts. While most prior work investigated the use of distillation for building task-specific models, we leverage knowledge distillation during the pre-training phase and show that it is possible to reduce the size of a BERT model by 40%, while retaining 97% of its language understanding capabilities and being 60% faster. To leverage the inductive biases learned by larger models during pre-training, we introduce a triple loss combining language modeling, distillation and cosine-distance losses. Our smaller, faster and lighter model is cheaper to pre-train and we demonstrate its capabilities for on-device computations in a proof-of-concept experiment and a comparative on-device study.
|
['Julien Chaumond', 'Lysandre Debut', 'Victor Sanh', 'Thomas Wolf']
|
2019-10-02
| null | null | null |
neurips-2019-12
|
['linguistic-acceptability']
|
['natural-language-processing']
|
[ 1.40172884e-01 2.85400659e-01 -4.45098639e-01 -5.56550801e-01
-1.15701246e+00 -6.03245735e-01 4.17661607e-01 3.02613318e-01
-7.79283166e-01 6.95409358e-01 2.02738836e-01 -7.33019650e-01
1.60132229e-01 -8.10228646e-01 -9.67556655e-01 -9.27296951e-02
6.59394339e-02 7.09481478e-01 8.02074652e-03 -1.09421626e-01
-8.23495537e-02 5.28666139e-01 -9.92894053e-01 3.99641484e-01
6.86626077e-01 9.57246363e-01 2.84768641e-01 7.73827851e-01
-1.25856220e-03 8.02635252e-01 -3.78205597e-01 -4.12108451e-01
2.08974212e-01 6.09712973e-02 -8.36650193e-01 -6.59380972e-01
3.34544808e-01 -3.71314436e-01 -2.95889765e-01 5.58198214e-01
9.88400638e-01 2.65116423e-01 4.61519569e-01 -7.75574207e-01
-5.61287165e-01 7.83837259e-01 -3.62727791e-01 1.02494538e-01
1.24412015e-01 1.14738531e-01 9.79637206e-01 -8.76638532e-01
3.19850713e-01 1.26440990e+00 9.57967460e-01 6.34187579e-01
-1.40641499e+00 -1.05838764e+00 1.27804771e-01 -3.32806528e-01
-1.52847815e+00 -6.00580513e-01 5.27167916e-01 -1.95505574e-01
1.45391011e+00 -2.42934763e-01 -2.64107808e-03 1.01574039e+00
8.16848427e-02 8.81482244e-01 9.07590330e-01 -6.64440513e-01
8.30641016e-02 5.46968222e-01 1.40924677e-01 7.91521430e-01
2.45080039e-01 1.19522680e-02 -4.15602684e-01 -1.83348373e-01
6.77041888e-01 -1.18106447e-01 1.88746434e-02 -3.64558935e-01
-1.03546488e+00 1.01473510e+00 6.95254982e-01 4.33638275e-01
-1.78431049e-01 3.80548745e-01 6.70860529e-01 2.77425826e-01
7.13305593e-01 6.64604664e-01 -8.06099415e-01 -2.56264508e-01
-1.18883026e+00 7.65157565e-02 9.12299931e-01 1.02086437e+00
6.89333916e-01 -1.65857807e-01 -2.27074221e-01 9.33667421e-01
3.83881852e-02 6.07289553e-01 5.01643002e-01 -4.74434078e-01
8.14550161e-01 3.55397463e-01 -1.61729738e-01 -4.88081366e-01
-6.30078018e-01 -5.87909877e-01 -8.40125144e-01 -1.84264615e-01
1.79766893e-01 -4.39574510e-01 -9.62040544e-01 1.84112024e+00
-1.42869223e-02 1.98244750e-01 -3.49968113e-02 5.31331480e-01
4.86044347e-01 7.76392221e-01 3.55404913e-01 2.28013456e-01
1.23825252e+00 -9.23880935e-01 -2.98471868e-01 -3.62423271e-01
1.21662331e+00 -6.52372122e-01 1.20544779e+00 4.03511018e-01
-1.20456314e+00 -7.13653743e-01 -1.16644645e+00 -6.18125796e-01
-4.47439075e-01 3.11929226e-01 9.50823247e-01 7.80769467e-01
-1.17026365e+00 6.51894629e-01 -9.12653744e-01 -3.58615071e-01
5.36110580e-01 6.58035636e-01 -3.09614744e-02 -4.05405432e-01
-1.23051989e+00 1.07570529e+00 3.63267750e-01 -2.75977194e-01
-9.01668906e-01 -1.20710897e+00 -8.43722880e-01 1.42387301e-01
1.82640448e-01 -9.66029406e-01 1.22854578e+00 -6.37188911e-01
-1.77430630e+00 9.17641997e-01 -5.78131825e-02 -7.00829208e-01
3.03679228e-01 -4.77467179e-01 -4.83891249e-01 -1.17830358e-01
-2.15699986e-01 8.98227394e-01 6.98358774e-01 -8.42134595e-01
-4.32389706e-01 -1.22005470e-01 3.13670933e-01 1.47344880e-02
-6.24416828e-01 -7.14916587e-02 -5.31384230e-01 -6.99630380e-01
-4.51928496e-01 -9.97987628e-01 -2.81057179e-01 9.86845344e-02
-2.58822292e-01 -1.13655567e-01 4.19847786e-01 -6.57688618e-01
1.28114295e+00 -2.07718754e+00 -1.71067193e-01 3.09866071e-01
1.13877676e-01 5.88526309e-01 -3.78109872e-01 1.91085920e-01
1.35388196e-01 1.36158511e-01 -2.32908592e-01 -4.74207461e-01
-3.59835885e-02 1.87346548e-01 -4.84816164e-01 2.26876378e-01
2.57779598e-01 1.08600628e+00 -8.20049822e-01 -3.12088192e-01
1.66783556e-01 7.98098147e-01 -1.05645347e+00 2.20211715e-01
-2.39012718e-01 3.17784727e-01 -4.27303523e-01 2.75056809e-01
5.89974165e-01 -4.41700697e-01 2.31991723e-01 -2.90041685e-01
2.66343087e-01 7.58887827e-01 -7.44746029e-01 2.28482533e+00
-1.46470451e+00 7.16040611e-01 6.08823113e-02 -1.22367358e+00
7.27065265e-01 1.33292630e-01 2.60179192e-01 -9.82851148e-01
2.03249767e-01 2.52576500e-01 -3.13607827e-02 -1.77767351e-01
4.75927323e-01 -2.97495753e-01 -2.18281642e-01 6.31716907e-01
3.17055851e-01 -2.83734053e-01 -2.25800365e-01 2.13887110e-01
1.05841029e+00 3.54337245e-02 1.85332268e-01 -4.02629673e-01
2.91782022e-01 -3.49007547e-01 4.03783955e-02 7.74831295e-01
1.59234270e-01 4.74607646e-01 -1.51153550e-01 -2.27877155e-01
-1.04172432e+00 -1.24065471e+00 -2.41587549e-01 1.53232491e+00
-2.60651231e-01 -3.26971442e-01 -5.95035315e-01 -6.61137819e-01
7.29309171e-02 1.00872159e+00 -2.24141896e-01 -3.76579762e-01
-7.04076529e-01 -7.60599315e-01 9.62772667e-01 8.28498244e-01
5.66566527e-01 -7.67450213e-01 -2.67957717e-01 3.20553899e-01
1.24471113e-01 -1.23378277e+00 -4.48377639e-01 2.86956310e-01
-9.54736829e-01 -4.72727776e-01 -7.68005788e-01 -8.19690704e-01
7.25072742e-01 -2.52212882e-02 1.46601844e+00 -1.17647193e-01
-3.12777400e-01 3.17683756e-01 -1.08698502e-01 -5.47690868e-01
-2.98545271e-01 5.83528221e-01 6.37476891e-02 -4.50486571e-01
4.56677914e-01 -6.88522995e-01 -4.49034989e-01 3.09582520e-02
-6.76477075e-01 1.72278155e-02 8.94299686e-01 8.20426702e-01
4.18153077e-01 -1.24878787e-01 5.26010394e-01 -9.65866446e-01
7.27286637e-01 -5.10673046e-01 -4.89893407e-01 3.04446101e-01
-6.79972947e-01 5.67353070e-01 6.60590291e-01 -6.74145043e-01
-1.00539148e+00 -8.03422704e-02 -3.53721559e-01 -3.00067276e-01
3.09812665e-01 5.37544310e-01 1.82983745e-02 -1.26315221e-01
8.00352633e-01 -2.85016559e-02 -2.78837174e-01 -6.22891784e-01
8.56345773e-01 6.59141362e-01 2.26965860e-01 -1.03246236e+00
7.25975990e-01 4.21066165e-01 -3.06998193e-01 -8.51004720e-01
-1.00015688e+00 -4.66525733e-01 -5.25098264e-01 6.13987803e-01
7.12968946e-01 -1.40411246e+00 -6.24160826e-01 -4.15445715e-02
-1.20026648e+00 -7.55451858e-01 -4.00059789e-01 7.17340112e-01
-2.86582679e-01 -1.20058067e-01 -4.83015776e-01 -6.21606827e-01
-7.91389585e-01 -7.52814651e-01 1.24290979e+00 -1.84485078e-01
-3.50614518e-01 -1.33520293e+00 4.92220744e-03 2.03162238e-01
8.98508549e-01 -3.62706214e-01 1.07001770e+00 -7.45657146e-01
-3.48611027e-01 -2.67836720e-01 -4.91549283e-01 5.51010191e-01
-1.67549402e-01 -7.51515925e-01 -1.20901024e+00 -5.55445671e-01
-1.43893763e-01 -6.24641776e-01 9.55217361e-01 2.83253103e-01
1.43133926e+00 -1.37213334e-01 -5.34957767e-01 7.36322641e-01
1.35569239e+00 -2.64364421e-01 5.78002572e-01 -8.98446888e-02
6.49910748e-01 2.31027722e-01 1.29567772e-01 2.32270479e-01
4.23853964e-01 7.02072501e-01 -4.13058370e-01 -3.33624899e-01
-2.34971225e-01 -5.18709481e-01 4.90115911e-01 9.92431879e-01
1.10746726e-01 -3.05997163e-01 -9.39850926e-01 4.96175796e-01
-1.44687593e+00 -6.15034997e-01 7.25999832e-01 2.29576778e+00
1.11044168e+00 5.41569054e-01 -1.51257709e-01 -2.49187320e-01
4.07459497e-01 -1.43711135e-01 -6.24111414e-01 -7.23732710e-01
2.29572877e-01 7.77859330e-01 7.27822661e-01 6.44347906e-01
-8.95996749e-01 1.01226544e+00 6.54020023e+00 1.14142787e+00
-1.37624705e+00 4.19496953e-01 9.20595407e-01 -4.04257566e-01
-3.46457094e-01 -1.72090217e-01 -1.04325688e+00 1.29964754e-01
1.47416532e+00 -1.13051616e-01 5.25687575e-01 1.00403595e+00
-4.51464429e-02 1.14041686e-01 -1.57694852e+00 1.10230231e+00
1.33916140e-01 -1.25252986e+00 1.43944472e-01 -1.46113634e-01
7.04728961e-01 5.10546505e-01 2.04827376e-02 1.02408218e+00
4.72828031e-01 -1.14618504e+00 4.15197760e-01 1.99139550e-01
1.19745672e+00 -6.51301742e-01 5.29643536e-01 4.75207269e-01
-1.04912984e+00 -6.59731850e-02 -4.01971489e-01 -2.03842431e-01
1.82046697e-01 9.37447786e-01 -1.12913084e+00 2.79942453e-01
3.14768761e-01 3.60798776e-01 -3.22719783e-01 6.11434579e-01
-2.37838641e-01 8.71540248e-01 -6.11189663e-01 -1.82014495e-01
7.22315088e-02 3.41579109e-01 2.46307440e-02 1.53771448e+00
2.93197274e-01 -7.03997491e-03 1.55315146e-01 8.98773849e-01
-7.10855305e-01 -2.59161331e-02 -5.39020598e-01 -1.08246431e-01
5.45608401e-01 1.02352798e+00 -2.83696979e-01 -6.61070287e-01
-3.97158772e-01 1.00464582e+00 5.72336972e-01 2.64009655e-01
-1.02629638e+00 -6.00267172e-01 4.99242723e-01 2.98786610e-01
2.87543416e-01 -3.88385832e-01 -4.39417005e-01 -1.11262894e+00
1.33696884e-01 -5.89085817e-01 2.56553531e-01 -6.05111480e-01
-1.37200654e+00 6.32786393e-01 -6.82849251e-03 -8.25943589e-01
-3.39187562e-01 -7.41849780e-01 -3.82495910e-01 1.12881744e+00
-1.78949583e+00 -1.36013412e+00 5.11902235e-02 4.53361332e-01
3.54705274e-01 -3.58627774e-02 1.04796052e+00 6.99445963e-01
-1.37226209e-01 1.08398283e+00 1.92128330e-01 2.29874820e-01
7.21804261e-01 -8.39003742e-01 5.02161026e-01 3.69403154e-01
1.46675944e-01 1.00565815e+00 3.31348956e-01 -2.44373709e-01
-1.48174405e+00 -1.27623904e+00 9.98940587e-01 -6.06785655e-01
6.47376597e-01 -9.26894367e-01 -6.06302977e-01 7.50875533e-01
-6.31279126e-02 1.97215706e-01 6.61537409e-01 6.00305974e-01
-6.45751774e-01 -2.29359210e-01 -1.18910396e+00 3.93523902e-01
1.13697088e+00 -1.02723646e+00 -4.61303204e-01 5.07063866e-01
1.06600869e+00 -3.24752331e-01 -9.08351958e-01 4.00412351e-01
6.12979293e-01 -2.50338435e-01 1.25066555e+00 -5.82747519e-01
2.95381576e-01 2.42794022e-01 -1.27544433e-01 -1.22650301e+00
-1.79526284e-01 -5.31064689e-01 -7.11947754e-02 1.25021291e+00
8.21820676e-01 -6.15826190e-01 7.25437641e-01 5.81338882e-01
3.57582308e-02 -7.85616994e-01 -7.86606014e-01 -8.81521940e-01
5.99763632e-01 -7.76151776e-01 3.12905252e-01 8.16364944e-01
1.04077898e-01 7.78668404e-01 -2.70094305e-01 -8.86460096e-02
1.80950418e-01 -1.44295767e-01 6.72091126e-01 -9.94422615e-01
-5.78277767e-01 -2.39328608e-01 -9.74559560e-02 -1.61047471e+00
2.42362708e-01 -1.28919923e+00 6.95186332e-02 -1.31598794e+00
8.27979520e-02 -9.78855133e-01 -4.27851915e-01 5.82937181e-01
6.94571659e-02 8.13767314e-02 1.30571261e-01 -9.74562392e-02
-6.37134194e-01 4.34408396e-01 9.13372219e-01 -4.25613225e-01
-3.05783391e-01 -1.27180696e-01 -8.54365826e-01 5.29304266e-01
7.09795237e-01 -4.64602351e-01 -5.86137056e-01 -8.29980910e-01
2.80619830e-01 -3.03478330e-01 7.44695291e-02 -1.20372427e+00
2.16305062e-01 5.55483341e-01 2.11332768e-01 -1.65313631e-01
4.37318981e-01 -8.10898960e-01 -5.19381642e-01 3.43886584e-01
-6.36910677e-01 -1.57214865e-01 8.11626196e-01 3.26672852e-01
5.71218990e-02 -2.51029115e-02 7.37606585e-01 4.73688953e-02
-5.65683901e-01 2.17779666e-01 -3.52886841e-02 4.29943711e-01
6.55015647e-01 2.13867933e-01 1.22963404e-02 -2.69016445e-01
-3.63282979e-01 3.91094089e-02 -5.17453365e-02 4.50252146e-01
3.40958446e-01 -1.14894950e+00 -6.36556506e-01 3.06504726e-01
5.75408638e-02 9.57271233e-02 1.66561171e-01 6.29595876e-01
-4.55215335e-01 8.09778988e-01 1.83303207e-01 -3.74417901e-01
-7.24071801e-01 6.99478865e-01 1.78129181e-01 -8.97632778e-01
-4.79807884e-01 1.10068917e+00 2.98636377e-01 -7.98978925e-01
3.03549141e-01 -7.65573800e-01 3.75911653e-01 -3.61808151e-01
5.58143437e-01 1.47326544e-01 3.18896651e-01 -1.75216384e-02
-3.96178722e-01 4.58003461e-01 -2.41928130e-01 -1.22799769e-01
1.32988095e+00 1.11737631e-01 2.71835625e-01 3.51923794e-01
1.72604978e+00 7.78122544e-02 -9.65303361e-01 -6.63498342e-01
-1.59930363e-01 2.48015430e-02 2.87087053e-01 -1.03088295e+00
-8.49568784e-01 1.41330767e+00 6.91526830e-01 -4.16491270e-01
1.07884312e+00 -5.62164374e-02 1.09119225e+00 8.91518593e-01
6.07656837e-01 -1.21859729e+00 9.04025137e-02 7.14064598e-01
7.04825044e-01 -1.28155220e+00 2.44840402e-02 -2.72539496e-01
-4.09429729e-01 8.13302517e-01 2.93346167e-01 -1.51855975e-01
1.03205168e+00 5.81503212e-01 -2.22614512e-01 8.96099396e-03
-6.54939473e-01 1.02008589e-01 2.83585638e-01 6.29024208e-01
7.42525220e-01 1.78723574e-01 1.07628018e-01 6.61408246e-01
-2.63154507e-01 3.79870623e-01 -1.91871360e-01 8.88177276e-01
-1.92080334e-01 -1.09273911e+00 2.45349091e-02 5.21419287e-01
-4.10064191e-01 -7.06646383e-01 1.52912959e-01 6.47123337e-01
1.25944525e-01 7.98169553e-01 1.50075182e-01 -3.97777438e-01
2.77276039e-01 6.81057423e-02 6.52644575e-01 -8.72550189e-01
-6.26868188e-01 -4.57328975e-01 3.55322331e-01 -5.12295902e-01
-1.27172306e-01 -7.77157545e-02 -1.45051062e+00 -4.52359229e-01
-3.81697118e-01 -7.32137710e-02 9.01761591e-01 9.89992321e-01
7.72819161e-01 6.78055048e-01 2.63176292e-01 -8.89716446e-01
-7.39424527e-01 -9.80702937e-01 -2.50822574e-01 2.06889987e-01
1.96701869e-01 -5.06150544e-01 -7.48344287e-02 -1.01713344e-01]
|
[10.62611198425293, 8.453210830688477]
|
55e8eb62-eb3a-4182-bb83-f798e2b84409
|
coherence-based-frequency-subset-selection
|
2205.08985
| null |
https://arxiv.org/abs/2205.08985v1
|
https://arxiv.org/pdf/2205.08985v1.pdf
|
Coherence-Based Frequency Subset Selection For Binaural RTF-Vector-Based Direction of Arrival Estimation for Multiple Speakers
|
Recently, a method has been proposed to estimate the direction of arrival (DOA) of a single speaker by minimizing the frequency-averaged Hermitian angle between an estimated relative transfer function (RTF) vector and a database of prototype anechoic RTF vectors. In this paper, we extend this method to multi-speaker localization by introducing the frequency-averaged Hermitian angle spectrum and selecting peaks of this spatial spectrum. To construct the Hermitian angle spectrum, we consider only a subset of frequencies, where it is likely that one speaker is dominant. We compare the effectiveness of the generalized magnitude squared coherence and two coherent-to-diffuse ratio (CDR) estimators as frequency selection criteria. Simulation results for estimating the DOAs of two speakers in a reverberant environment with diffuse-like babble noise using binaural hearing devices show that using the binaural effective-coherence-based CDR estimate as a frequency selection criterion yields the best performance.
|
['Simon Doclo', 'Daniel Fejgin']
|
2022-05-18
| null | null | null | null |
['direction-of-arrival-estimation']
|
['audio']
|
[-1.09876625e-01 -5.30420482e-01 8.80380273e-01 -2.11524755e-01
-1.24542415e+00 -6.16768301e-01 2.36454591e-01 4.12681401e-02
-2.02831253e-01 5.03951371e-01 5.50887942e-01 -1.50731206e-01
-3.45422417e-01 -4.32955056e-01 -3.66165072e-01 -9.62909222e-01
-4.86624092e-01 -3.61443102e-01 1.35239661e-01 1.92719344e-02
2.99173892e-02 3.19523484e-01 -1.65168524e+00 -1.39006361e-01
6.82253659e-01 9.84773517e-01 3.65878463e-01 9.51994538e-01
4.91317511e-01 2.92490304e-01 -1.27068055e+00 5.74603193e-02
3.26070219e-01 -5.34144640e-01 -7.05818757e-02 -3.40549439e-01
4.99035776e-01 -5.39135896e-02 2.92219128e-02 9.99883175e-01
1.04794073e+00 2.75294036e-01 6.78295970e-01 -7.15795100e-01
2.22763047e-01 4.11690205e-01 -1.66022956e-01 4.31804627e-01
7.72257507e-01 -2.47313246e-01 7.07794845e-01 -1.10089922e+00
2.22702995e-01 8.67575824e-01 8.39038730e-01 -1.30313516e-01
-9.23714101e-01 -6.25149906e-01 -5.50322950e-01 7.33448714e-02
-1.86326706e+00 -5.81374824e-01 1.03650939e+00 -4.04754788e-01
7.66319454e-01 4.42668706e-01 6.18311226e-01 3.43847960e-01
2.50270665e-01 -7.57244676e-02 1.11693656e+00 -8.92499626e-01
3.64813745e-01 1.93807110e-01 -2.27515232e-02 4.22535241e-01
1.21953562e-01 3.44711781e-01 -7.14662075e-01 -5.92461944e-01
4.31529790e-01 -8.38863492e-01 -1.07154489e+00 -1.85636252e-01
-1.17080522e+00 5.47612965e-01 1.90932989e-01 6.80825830e-01
-4.33323950e-01 4.78173569e-02 -3.46810102e-01 2.80223995e-01
3.78211290e-01 6.99541271e-01 -4.23579589e-02 -1.23702683e-01
-8.57082605e-01 1.17663607e-01 1.16188824e+00 5.71384847e-01
5.24248183e-01 4.76282746e-01 2.23657921e-01 1.21258080e+00
6.73906446e-01 9.46763396e-01 4.28199917e-02 -6.76563323e-01
1.73745379e-01 -7.56893277e-01 4.14650291e-01 -1.01775014e+00
-3.16441745e-01 -1.11190259e+00 -1.61059067e-01 -3.02752247e-03
5.59962869e-01 -6.17048085e-01 -4.03313607e-01 1.83251560e+00
6.41423762e-01 2.19142273e-01 1.19444408e-01 1.14169919e+00
3.05930138e-01 6.67673945e-01 -5.21030545e-01 -5.89124978e-01
1.12960458e+00 -3.20380569e-01 -8.38095367e-01 1.75997820e-02
2.44020045e-01 -1.47397029e+00 5.46796441e-01 6.22061431e-01
-8.63367498e-01 -5.76037526e-01 -1.14288759e+00 8.02249670e-01
5.83660090e-03 -1.64786682e-01 -9.84324738e-02 1.19876814e+00
-1.02165782e+00 6.49875998e-02 -3.92330587e-01 2.88735088e-02
-8.30481052e-01 -1.16310865e-01 -1.01349622e-01 1.49928406e-01
-1.22035980e+00 8.05751801e-01 -4.61560279e-01 2.02972695e-01
-6.09200120e-01 -7.86475122e-01 -7.26152062e-01 2.29398813e-02
-3.52846265e-01 -3.36319625e-01 1.34797835e+00 -3.95619303e-01
-1.42348778e+00 1.05168847e-02 -2.50106156e-01 -2.33940363e-01
1.04373686e-01 -2.58036464e-01 -1.05245662e+00 2.99557090e-01
4.85083312e-02 -1.81244224e-01 8.90117288e-01 -1.33005226e+00
-5.07414222e-01 -3.23181823e-02 -4.61962193e-01 3.78870577e-01
-1.18864588e-02 5.58133908e-02 4.17916507e-01 -4.84002650e-01
6.97665691e-01 -5.80713749e-01 3.88039798e-02 -3.69962454e-01
-3.38386953e-01 4.35434766e-02 2.74275035e-01 -7.55818427e-01
1.34391618e+00 -2.33582544e+00 -5.10471702e-01 5.78371584e-01
-1.56300828e-01 -1.57038331e-01 9.99279879e-03 5.54415107e-01
-1.88040167e-01 -5.36828041e-01 4.99069877e-02 1.59262806e-01
-2.46983618e-01 -6.76994741e-01 -2.13733628e-01 9.04002428e-01
-3.49105954e-01 -4.34052080e-01 -9.03307557e-01 -1.38034210e-01
1.03850290e-01 7.46317089e-01 -5.66354454e-01 4.61556077e-01
6.60306156e-01 2.73956269e-01 -1.13764614e-01 4.06173557e-01
9.66217518e-01 5.26530087e-01 -3.81110013e-02 -3.78401667e-01
-4.83513653e-01 4.11408037e-01 -1.67550194e+00 9.50271547e-01
-8.64510655e-01 8.57106209e-01 5.72959721e-01 -4.26031560e-01
1.24285316e+00 6.35717750e-01 2.87594140e-01 -2.97882408e-01
-1.87710762e-01 6.60195529e-01 4.36182052e-01 -3.84998173e-01
2.30215266e-01 -4.09509122e-01 1.60355508e-01 3.24444711e-01
3.39972526e-01 -3.93455833e-01 -2.85464197e-01 -2.64489412e-01
9.11651611e-01 -3.42176497e-01 5.69740355e-01 -5.71175516e-01
5.84717631e-01 -7.78022647e-01 3.72643352e-01 6.06370807e-01
-2.82842606e-01 7.47625589e-01 -1.57368347e-01 1.13231599e-01
-6.16948068e-01 -1.37893367e+00 -5.04873633e-01 6.57379448e-01
5.76994456e-02 -1.74089581e-01 -7.78629839e-01 1.85114384e-01
-5.42239249e-02 1.03351176e+00 1.17565833e-01 1.12503432e-01
-5.58454990e-01 -3.86249691e-01 5.12884259e-01 -1.49327591e-01
2.46812105e-01 -1.58818543e-01 -4.66258228e-01 4.04468358e-01
-5.64234078e-01 -8.60355914e-01 -7.40695775e-01 2.24188805e-01
-2.45561793e-01 -6.71252072e-01 -8.22389901e-01 -6.83352649e-01
3.15895736e-01 4.94573832e-01 7.67020226e-01 -7.50325203e-01
-2.12450355e-01 9.36467230e-01 -1.86706185e-01 -5.42572320e-01
-2.74364382e-01 -8.55182707e-01 4.03454572e-01 2.23425865e-01
-1.45510063e-01 -6.80623293e-01 -8.77592027e-01 7.23639905e-01
-2.35188112e-01 -6.70750141e-01 2.23381054e-02 5.96437156e-01
9.77092013e-02 1.40355662e-01 8.62867296e-01 2.64160931e-01
8.86884332e-01 -4.36763376e-01 -6.05224788e-01 3.26379053e-02
-2.41194367e-01 -2.99825191e-01 5.18834174e-01 -5.10649681e-01
-1.25538194e+00 -1.66467726e-01 -1.97687015e-01 -1.32848486e-01
-2.03217939e-01 4.14546251e-01 3.30409221e-02 -3.40014756e-01
1.04439247e+00 3.89558107e-01 -4.50526655e-01 -4.18249190e-01
5.18133976e-02 9.77476478e-01 4.66012895e-01 -3.05812746e-01
7.22627044e-01 6.17206320e-02 9.48968306e-02 -1.59300768e+00
-3.63484204e-01 -1.01033461e+00 8.81802756e-03 -5.33538878e-01
5.43157399e-01 -9.72133100e-01 -5.01957059e-01 4.30807620e-01
-1.21582401e+00 1.12552963e-01 6.21455163e-02 1.71253312e+00
-4.19759959e-01 4.61532354e-01 -1.45137385e-01 -1.44539320e+00
-1.13903083e-01 -8.73990834e-01 7.89929450e-01 9.67715904e-02
-2.71877021e-01 -8.16723049e-01 5.61607897e-01 5.30937091e-02
5.85609615e-01 1.24476627e-01 4.40657854e-01 -3.94436955e-01
-3.21013778e-01 -3.15786481e-01 5.40963292e-01 4.08108115e-01
2.19815969e-01 -2.70825773e-01 -1.23350763e+00 -2.32627869e-01
7.11709023e-01 2.45525807e-01 1.66918442e-01 9.67976093e-01
4.20105696e-01 -1.47315517e-01 -1.18361942e-01 6.44976497e-01
1.51415026e+00 5.83580136e-01 2.79574126e-01 -1.67351186e-01
2.24739593e-02 5.86321890e-01 6.78102434e-01 5.75444579e-01
-3.26385139e-03 6.64345503e-01 4.69198525e-02 1.60316110e-01
-1.66763067e-01 -3.72222476e-02 2.00582922e-01 1.18904865e+00
-4.28787358e-02 -5.36367655e-01 -7.67767191e-01 6.50986433e-01
-8.53387654e-01 -8.53795469e-01 -1.03178419e-01 2.78028917e+00
5.95179081e-01 -2.38887325e-01 -3.48085128e-02 3.14213663e-01
8.66021395e-01 1.80694629e-02 1.07235186e-01 -2.83798575e-01
-5.08612767e-02 2.72314519e-01 5.31686544e-01 1.11274505e+00
-7.98762798e-01 -9.90220308e-02 6.63434696e+00 6.54567719e-01
-1.30326676e+00 7.04339966e-02 4.33494244e-03 1.73397481e-01
-3.33538860e-01 -2.11676821e-01 -7.28581190e-01 2.24083290e-01
1.21914554e+00 -3.19454014e-01 2.97791809e-01 7.13969469e-01
3.26938272e-01 -3.55900913e-01 -7.19071746e-01 9.89519477e-01
2.03445271e-01 -4.93413061e-01 -6.39056921e-01 7.53715411e-02
4.20375615e-01 -1.93898961e-01 1.19979233e-01 -2.22053632e-01
-2.75126189e-01 -6.05545461e-01 8.40262294e-01 5.29615104e-01
6.02360964e-01 -6.07118309e-01 6.12464011e-01 1.83255255e-01
-1.49079823e+00 -3.51505019e-02 -2.16563836e-01 1.99195012e-01
1.96632624e-01 1.11458516e+00 -1.62606144e+00 4.33514535e-01
5.13775349e-01 -2.43565783e-01 1.70045078e-01 1.90204465e+00
-4.18172451e-03 9.86283481e-01 -8.35688055e-01 -2.97820240e-01
-5.53335398e-02 -9.33985114e-02 1.18274593e+00 1.38430250e+00
1.02090168e+00 7.52366036e-02 -2.86074966e-01 4.66868490e-01
3.55600417e-01 3.77035588e-01 -6.18671119e-01 3.97844076e-01
7.96580374e-01 9.27272558e-01 -3.82331967e-01 2.10215062e-01
-3.01787078e-01 3.68202627e-01 -6.67239010e-01 7.86073565e-01
-6.53767765e-01 -9.46262419e-01 6.01302922e-01 2.41892576e-01
4.09321696e-01 -4.20656383e-01 4.10259739e-02 -7.06266880e-01
-9.15898010e-03 -6.31792426e-01 1.51843317e-02 -9.79356825e-01
-8.49688113e-01 7.03930020e-01 8.41874182e-02 -1.76594114e+00
-4.92877066e-01 -2.14712843e-01 -5.54656923e-01 1.40957820e+00
-1.21096611e+00 -4.19651657e-01 6.04909807e-02 4.81456816e-01
1.83832139e-01 -4.21901755e-02 9.46024239e-01 4.16959196e-01
3.28248560e-01 5.14877617e-01 4.78018224e-01 -2.80108303e-01
8.65455210e-01 -1.28403723e+00 -2.81491607e-01 7.40677774e-01
-9.18164402e-02 9.11392927e-01 1.38770318e+00 -2.70653099e-01
-1.08501780e+00 -6.25208974e-01 9.34564829e-01 1.11708030e-01
4.97151375e-01 -5.00791907e-01 -5.87980092e-01 8.07655305e-02
1.89723805e-01 -2.79612765e-02 9.50150907e-01 -5.36921248e-03
-3.50467980e-01 -5.93177319e-01 -1.18188429e+00 2.97368735e-01
4.67832953e-01 -5.49128234e-01 -5.55781305e-01 3.94445926e-01
4.82490897e-01 -2.20015019e-01 -9.84458983e-01 2.65142441e-01
7.36884058e-01 -9.99633491e-01 1.03774464e+00 5.09839416e-01
-4.59850907e-01 -6.52906120e-01 -7.09696114e-01 -1.82499611e+00
-2.25571886e-01 -1.09369862e+00 3.83323699e-01 1.05788171e+00
3.95494074e-01 -1.12817323e+00 -5.35817519e-02 -2.92501628e-01
-1.36777207e-01 -3.67894262e-01 -1.19347954e+00 -9.01299834e-01
-3.38610590e-01 -3.34956408e-01 3.04571658e-01 6.34785712e-01
2.09484547e-01 3.11104506e-01 -2.59543657e-01 8.19966316e-01
1.05519390e+00 8.69823769e-02 5.25646567e-01 -1.00936103e+00
-6.00895464e-01 6.85491934e-02 -4.94672269e-01 -1.15779126e+00
-5.22693276e-01 -2.54205197e-01 6.73583448e-01 -1.19650781e+00
-7.35437036e-01 -2.37170890e-01 -4.31568533e-01 -5.12537241e-01
5.18227741e-02 -1.62878990e-01 -8.49766210e-02 -1.72984213e-01
3.01367253e-01 2.47277066e-01 9.69535887e-01 2.43973628e-01
-3.03037524e-01 7.09216416e-01 -2.24390224e-01 7.95164883e-01
3.58869880e-01 -5.92101872e-01 -4.68521684e-01 -1.92922857e-02
-1.25775009e-01 6.77881420e-01 5.61249554e-02 -1.44370246e+00
3.76090646e-01 2.03363553e-01 2.19235778e-01 -6.86182737e-01
6.87743485e-01 -6.91046834e-01 3.37975383e-01 4.56390798e-01
-2.12846622e-01 -3.38720739e-01 1.61097571e-02 6.44694388e-01
-4.18313205e-01 -1.59139246e-01 1.09270597e+00 1.59763023e-01
-5.26389927e-02 -4.18457866e-01 -7.56943166e-01 -2.69032240e-01
5.76081872e-01 -1.73426881e-01 -2.00793922e-01 -1.19797742e+00
-4.80475098e-01 -6.07518375e-01 -2.47129411e-01 -2.03497082e-01
6.38183534e-01 -9.74666357e-01 -8.08416963e-01 2.79222429e-01
-1.49108708e-01 -7.37546623e-01 3.00653577e-01 7.66515911e-01
-5.33444881e-01 6.81915939e-01 1.20970316e-01 -8.51421714e-01
-1.46121073e+00 -3.60488556e-02 9.23694134e-01 2.55338967e-01
1.36487195e-02 1.12043822e+00 3.56053114e-01 -2.71050781e-01
1.78839177e-01 -3.50777596e-01 -2.08506271e-01 -1.96186334e-01
5.60717642e-01 4.95564729e-01 3.07326496e-01 -7.63135493e-01
-4.96994615e-01 9.06036079e-01 7.61127770e-01 -8.90612423e-01
8.59121978e-01 -4.04216141e-01 6.84619620e-02 6.42090380e-01
1.54034162e+00 1.06162369e+00 -6.90616965e-01 -1.13582551e-01
-3.86551559e-01 -7.27812350e-01 2.07566351e-01 -8.12075913e-01
-3.80855411e-01 8.96028578e-01 1.07989180e+00 5.79084814e-01
1.27880645e+00 -2.37650439e-01 3.29984725e-01 2.34617323e-01
5.34063458e-01 -7.45715618e-01 2.00358592e-02 2.10892245e-01
8.87010932e-01 -4.31779742e-01 -2.25318149e-01 -6.05444610e-01
-9.25897136e-02 1.26401496e+00 1.71001956e-01 -4.35237288e-02
1.23003876e+00 2.72767335e-01 5.05819738e-01 2.76042044e-01
-2.84904599e-01 -4.12911773e-02 3.28497052e-01 8.04064393e-01
6.92983985e-01 2.86292374e-01 -4.87056702e-01 1.04251668e-01
-8.45978916e-01 -5.37770510e-01 6.42418981e-01 7.25577116e-01
-9.30580020e-01 -5.46504378e-01 -1.17909205e+00 5.77985607e-02
-5.44878602e-01 -1.86077535e-01 1.69113666e-01 2.84846187e-01
-2.78393123e-02 1.56404400e+00 1.37033537e-01 -3.06726873e-01
7.11163402e-01 1.98120251e-01 4.39367622e-01 -2.52163291e-01
-4.60919410e-01 9.13134694e-01 5.00830114e-01 -8.72170106e-02
-2.07700938e-01 -6.18751705e-01 -7.81262934e-01 9.68067050e-02
-8.67909253e-01 6.57968283e-01 1.20108998e+00 4.99546200e-01
6.05681958e-03 3.06434870e-01 1.28207231e+00 -4.16937053e-01
-6.21604383e-01 -1.06542397e+00 -1.12284708e+00 -1.84880435e-01
7.97068655e-01 -4.30014551e-01 -1.08748031e+00 -1.83570474e-01]
|
[15.152446746826172, 5.762768268585205]
|
b89f754a-e3fc-4f24-99c6-1eacf8d23674
|
lay-text-summarisation-using-natural-language
|
2303.14222
| null |
https://arxiv.org/abs/2303.14222v1
|
https://arxiv.org/pdf/2303.14222v1.pdf
|
Lay Text Summarisation Using Natural Language Processing: A Narrative Literature Review
|
Summarisation of research results in plain language is crucial for promoting public understanding of research findings. The use of Natural Language Processing to generate lay summaries has the potential to relieve researchers' workload and bridge the gap between science and society. The aim of this narrative literature review is to describe and compare the different text summarisation approaches used to generate lay summaries. We searched the databases Web of Science, Google Scholar, IEEE Xplore, Association for Computing Machinery Digital Library and arXiv for articles published until 6 May 2022. We included original studies on automatic text summarisation methods to generate lay summaries. We screened 82 articles and included eight relevant papers published between 2020 and 2021, all using the same dataset. The results show that transformer-based methods such as Bidirectional Encoder Representations from Transformers (BERT) and Pre-training with Extracted Gap-sentences for Abstractive Summarization (PEGASUS) dominate the landscape of lay text summarisation, with all but one study using these methods. A combination of extractive and abstractive summarisation methods in a hybrid approach was found to be most effective. Furthermore, pre-processing approaches to input text (e.g. applying extractive summarisation) or determining which sections of a text to include, appear critical. Evaluation metrics such as Recall-Oriented Understudy for Gisting Evaluation (ROUGE) were used, which do not consider readability. To conclude, automatic lay text summarisation is under-explored. Future research should consider long document lay text summarisation, including clinical trial reports, and the development of evaluation metrics that consider readability of the lay summary.
|
['Zoe Tieges', 'David McMinn', 'Gordon Morison', 'Mark David Jenkins', 'Oliver Vinzelberg']
|
2023-03-24
| null | null | null | null |
['abstractive-text-summarization']
|
['natural-language-processing']
|
[ 7.82398701e-01 6.03441417e-01 -5.72031558e-01 1.17917443e-02
-1.27930856e+00 -6.91183746e-01 8.12158227e-01 8.90003920e-01
-5.27016878e-01 1.12789118e+00 1.33965492e+00 -6.02214456e-01
-4.10741150e-01 -4.85178590e-01 -4.46769536e-01 -1.42804697e-01
3.47293109e-01 2.23114580e-01 -1.64625496e-01 -7.33835697e-02
1.15275633e+00 2.81880528e-01 -1.18602443e+00 6.56604171e-01
1.32152689e+00 2.20726371e-01 1.29807785e-01 1.03438866e+00
-4.39211994e-01 9.21173930e-01 -1.23977292e+00 -4.99254853e-01
-3.39339495e-01 -8.50259185e-01 -1.00862491e+00 -2.21017569e-01
6.04442298e-01 -2.58843303e-01 -5.39560281e-02 5.87287784e-01
1.03896832e+00 -1.77912995e-01 7.07939088e-01 -6.31270111e-01
-6.90325201e-01 1.01258922e+00 -4.83093828e-01 6.39751136e-01
8.17035973e-01 1.01253100e-01 8.98261547e-01 -5.60139537e-01
8.04056942e-01 1.10834467e+00 5.80478132e-01 4.50502425e-01
-1.03317583e+00 -5.74986279e-01 -3.62832248e-01 -1.85167253e-01
-6.46011472e-01 -8.58468592e-01 4.43546504e-01 -5.11847019e-01
1.46669209e+00 7.45205164e-01 8.44979346e-01 1.04494250e+00
8.53164077e-01 3.87640178e-01 9.89888370e-01 -6.80109322e-01
8.59638155e-02 6.27561510e-02 3.84579241e-01 9.55951735e-02
1.08007133e+00 -3.35109293e-01 -5.55045784e-01 -2.86532581e-01
9.37571749e-02 -3.83439392e-01 -1.56151846e-01 5.08964777e-01
-1.35681581e+00 8.74399602e-01 -6.11091629e-02 3.61895293e-01
-7.05071807e-01 -2.16099750e-02 1.13433647e+00 2.71322072e-01
8.01733315e-01 1.03036904e+00 7.41099045e-02 -3.38600695e-01
-1.77866626e+00 4.40105021e-01 9.42522883e-01 7.12989688e-01
-9.60581098e-03 -2.21623518e-02 -7.55362928e-01 6.92776084e-01
1.14779308e-01 4.06538844e-01 7.39670157e-01 -8.83749783e-01
8.18286717e-01 7.30431795e-01 -2.31769711e-01 -1.08166277e+00
-4.12063926e-01 -2.48570129e-01 -7.22905755e-01 -5.11426747e-01
-3.32596093e-01 -4.49900001e-01 -7.82903194e-01 9.52266276e-01
-2.47253552e-01 -6.21505857e-01 4.40266877e-01 3.09757113e-01
1.69524455e+00 8.76837611e-01 1.86301708e-01 -6.69176042e-01
1.47172284e+00 -6.47142887e-01 -1.12355173e+00 -2.32694000e-01
8.34286094e-01 -1.03973913e+00 5.41570842e-01 2.33285710e-01
-1.62775373e+00 -1.02542035e-01 -1.27466631e+00 -2.78020680e-01
-3.28184396e-01 2.27722034e-01 2.39604473e-01 7.98494101e-01
-1.18736649e+00 5.37137568e-01 -6.83420599e-01 -6.98006392e-01
6.38645232e-01 1.95978090e-01 -2.16939628e-01 1.46697626e-01
-1.12662315e+00 1.24658644e+00 6.27758682e-01 -1.22284524e-01
-2.58565873e-01 -9.28429782e-01 -8.45280588e-01 -1.09316967e-01
4.37256433e-02 -1.19363153e+00 1.17909598e+00 -5.17037868e-01
-1.13109469e+00 8.74180794e-01 -2.83110291e-01 -7.68127322e-01
3.60730588e-01 -1.41520321e-01 -1.82924464e-01 5.68598449e-01
4.33409899e-01 5.19181371e-01 3.09867382e-01 -7.22622454e-01
-5.01773596e-01 -2.82993734e-01 -2.88893491e-01 3.63083839e-01
-1.59882233e-01 6.41670108e-01 2.73160309e-01 -7.26170480e-01
-4.14642006e-01 -3.90218526e-01 -1.86806560e-01 -7.53967583e-01
-7.21045911e-01 -4.22127962e-01 4.74470675e-01 -1.18217432e+00
1.69194579e+00 -1.41739321e+00 -1.76743358e-01 -2.68135905e-01
4.07578170e-01 4.88675863e-01 9.13369507e-02 1.21080148e+00
-2.69955304e-02 7.88042545e-01 -2.91839421e-01 4.54843566e-02
-1.60766244e-01 -2.28861570e-01 -3.44449818e-01 4.18604076e-01
2.68993020e-01 1.12123978e+00 -9.30003583e-01 -7.60547340e-01
1.52243882e-01 2.62407422e-01 -2.23323643e-01 -1.61135450e-01
1.48753315e-01 -3.69713418e-02 -4.89183545e-01 1.22341886e-01
3.53894711e-01 2.07252633e-02 -1.50490955e-01 -9.35513303e-02
-6.05915546e-01 1.00781488e+00 -6.62076890e-01 1.44200611e+00
-3.61965477e-01 1.06167746e+00 -3.31699520e-01 -9.61069822e-01
8.93840909e-01 6.16030812e-01 3.45089525e-01 -5.41588247e-01
1.23911440e-01 4.34206486e-01 1.62690803e-02 -6.29308164e-01
8.27412963e-01 -1.65091664e-01 -2.10144874e-02 5.86046219e-01
1.11000706e-02 -6.92311585e-01 6.90648079e-01 6.14351153e-01
1.01544678e+00 -6.67235926e-02 8.01922917e-01 -3.60891849e-01
3.62493992e-01 4.15451288e-01 -1.38505045e-02 8.84168208e-01
2.78459191e-01 6.58209443e-01 7.95997977e-01 -1.57469213e-02
-1.21192873e+00 -3.96846265e-01 -3.22227299e-01 3.11574787e-01
-5.62122822e-01 -8.10339153e-01 -1.02737188e+00 -2.93974280e-01
-4.92147744e-01 1.37503958e+00 -5.22181511e-01 -4.47391868e-01
-4.60732460e-01 -8.04299235e-01 7.89981425e-01 1.06076531e-01
3.01596493e-01 -1.28456485e+00 -1.22426236e+00 3.68202388e-01
-2.87268013e-01 -7.64868736e-01 -3.65866631e-01 -1.08158566e-01
-1.12958443e+00 -8.88170362e-01 -1.20813036e+00 -4.73961174e-01
4.28636193e-01 2.10396163e-02 9.09769773e-01 -1.40188277e-01
-3.42030287e-01 4.81055617e-01 -5.19008338e-01 -9.71314490e-01
-8.93146217e-01 3.13025057e-01 -3.71836454e-01 -9.66610193e-01
3.18436235e-01 -1.66846365e-01 -6.96309984e-01 -6.06112242e-01
-1.18102860e+00 2.59379953e-01 1.03747904e+00 6.02101743e-01
1.35991856e-01 -2.93440402e-01 1.13006520e+00 -9.90801513e-01
1.62806785e+00 -4.21076745e-01 2.58682787e-01 1.83739424e-01
-8.56348217e-01 -5.92466146e-02 2.61911631e-01 -1.28728032e-01
-8.96883547e-01 -7.55549610e-01 -1.21931821e-01 5.14014125e-01
3.09846438e-02 1.03045332e+00 3.37215245e-01 4.49762523e-01
1.00623226e+00 2.23735616e-01 2.98242182e-01 -7.86210150e-02
1.75845802e-01 9.02441978e-01 1.89818263e-01 -1.47097513e-01
1.80982143e-01 -9.88972336e-02 -2.01664388e-01 -1.23470080e+00
-6.59056485e-01 -5.00577986e-01 -1.85085103e-01 -1.37750313e-01
7.41852939e-01 -7.72837400e-01 -9.90666300e-02 -7.98523799e-02
-1.32799876e+00 -5.71401045e-02 -5.46749353e-01 5.72789907e-01
-4.58197504e-01 5.50234735e-01 -3.56815726e-01 -5.14397383e-01
-1.35162413e+00 -8.82906795e-01 1.09428275e+00 3.62282425e-01
-1.01433456e+00 -8.39839339e-01 3.36331517e-01 5.50404966e-01
3.06351215e-01 5.40287793e-01 8.95869493e-01 -1.06616449e+00
3.75678957e-01 -4.26740915e-01 -1.59309968e-01 2.34086111e-01
4.18455094e-01 2.20325410e-01 -4.27431911e-01 -2.47202292e-02
2.96791876e-03 -1.32462487e-01 8.60081434e-01 9.67271984e-01
7.33312964e-01 -1.14701188e+00 -3.96929920e-01 -1.69138178e-01
1.21289837e+00 2.81795472e-01 8.90040398e-01 4.60244060e-01
5.43165863e-01 8.75313818e-01 3.20273101e-01 2.94218689e-01
2.73794472e-01 1.62667885e-01 -3.92355889e-01 1.15091726e-01
-3.43473285e-01 -1.14949763e-01 4.15846616e-01 1.07154357e+00
4.36486257e-03 -3.77454191e-01 -8.38986695e-01 7.05505192e-01
-1.40045094e+00 -1.17882252e+00 -4.48983520e-01 2.05302167e+00
1.05270731e+00 3.86030406e-01 2.70425826e-01 1.59327328e-01
4.73693788e-01 1.77624226e-01 -1.75727606e-01 -1.19511139e+00
-7.52106681e-02 4.20523077e-01 6.48604870e-01 2.80015200e-01
-5.33758163e-01 4.89688069e-01 5.98990059e+00 6.50568426e-01
-9.58236635e-01 -1.11614943e-01 5.25224984e-01 -1.12781674e-01
-5.63289106e-01 8.54499489e-02 -5.24367571e-01 4.55880553e-01
1.68090045e+00 -1.09556615e+00 -3.23231220e-01 6.24384731e-02
8.57968628e-01 -4.93817329e-01 -7.81387687e-01 4.13777232e-01
4.10981715e-01 -1.74849689e+00 3.15184176e-01 -2.22604405e-02
7.47520447e-01 -1.98901325e-01 -2.33258620e-01 1.43731073e-01
3.47961001e-02 -1.05483460e+00 5.94748735e-01 6.39405668e-01
8.81301641e-01 -5.69722414e-01 1.03669465e+00 1.20583288e-01
-4.29315418e-01 3.51881295e-01 -2.78813720e-01 4.48982753e-02
1.91581637e-01 7.61852264e-01 -1.18890619e+00 1.15518379e+00
2.84804791e-01 8.39371920e-01 -8.38702798e-01 1.22474182e+00
5.02145570e-03 1.00768983e+00 6.26681075e-02 -5.56114674e-01
2.22279012e-01 8.99433047e-02 8.88643682e-01 1.72177601e+00
2.99902141e-01 1.14694498e-01 -4.85809505e-01 6.21481776e-01
-2.46066786e-02 5.19864500e-01 -8.42208326e-01 -6.07075870e-01
2.57165760e-01 8.97267759e-01 -8.77988279e-01 -4.64997470e-01
-2.45864421e-01 6.10084593e-01 -4.13552821e-01 9.43177417e-02
-2.29063258e-01 -7.66845286e-01 -3.45360190e-01 2.96325833e-01
-1.74166888e-01 2.55672574e-01 -7.16554523e-01 -5.79397976e-01
9.41232685e-03 -1.05834281e+00 4.06059295e-01 -8.51282299e-01
-7.29147017e-01 4.57092851e-01 5.83066940e-01 -9.46054876e-01
-2.86376774e-01 -5.45753725e-02 -7.75043786e-01 1.03510928e+00
-8.31251383e-01 -7.87100077e-01 2.89410144e-01 -3.88125151e-01
9.52151656e-01 -3.27839926e-02 6.32020056e-01 -8.23009312e-02
-5.14087796e-01 2.55722255e-01 1.61870122e-01 -2.41671845e-01
6.65379941e-01 -1.24466574e+00 3.80791485e-01 5.23307562e-01
-4.67676967e-01 8.68934393e-01 1.03888476e+00 -1.21038604e+00
-1.12863851e+00 -9.76113081e-01 1.49254167e+00 -4.09739703e-01
4.30226028e-01 1.55681297e-01 -6.50429785e-01 3.99435580e-01
9.46656942e-01 -1.17901611e+00 8.26115370e-01 -2.43436038e-01
4.20657247e-01 1.88469261e-01 -9.47475612e-01 8.48639190e-01
4.73581553e-01 -1.06649637e-01 -1.32700539e+00 4.64613140e-01
8.49240899e-01 -3.38423282e-01 -1.01742768e+00 2.98054218e-01
5.21846116e-01 -4.51153606e-01 7.78407335e-01 -4.68780518e-01
1.17580605e+00 2.07224444e-01 5.58443964e-01 -1.47410417e+00
-5.97128868e-02 -7.36439228e-01 1.88473031e-01 1.41260493e+00
7.01750040e-01 -5.80673099e-01 4.79799420e-01 4.22019511e-01
-5.20234764e-01 -8.25019181e-01 -7.40713358e-01 -2.11148232e-01
4.13386136e-01 6.28163368e-02 1.70793548e-01 5.02023041e-01
5.03346562e-01 8.16160977e-01 2.72667687e-02 -6.41588211e-01
2.69050330e-01 -3.88088495e-01 4.77891833e-01 -1.10783160e+00
5.50296068e-01 -9.88399923e-01 -2.98035502e-01 -2.03098074e-01
-3.70987616e-02 -1.09760249e+00 -2.58659184e-01 -2.86323714e+00
5.50851464e-01 4.14912075e-01 4.01628941e-01 1.25740632e-01
-3.15007448e-01 -2.33323157e-01 -7.94499964e-02 2.08441749e-01
-3.14926207e-01 4.04169083e-01 1.40485787e+00 -2.85371244e-01
-3.48794609e-01 -2.88877860e-02 -1.50973928e+00 1.62475839e-01
9.48202252e-01 -6.12518191e-01 -5.17689943e-01 4.54898626e-02
5.66384912e-01 2.16077924e-01 1.18081450e-01 -6.19862318e-01
3.54878485e-01 -7.39379227e-02 2.24791452e-01 -7.68528283e-01
-4.41564053e-01 -1.35081470e-01 1.27925366e-01 5.91073096e-01
-1.03582501e+00 3.70501935e-01 7.02265978e-01 1.85222983e-01
-6.25446364e-02 -7.75663853e-01 2.45024294e-01 -2.76222199e-01
9.85122323e-02 -4.03907955e-01 -8.85578454e-01 4.20073330e-01
7.64449477e-01 -4.85599458e-01 -6.08516574e-01 -5.35936773e-01
-1.68825194e-01 3.32432270e-01 1.27363697e-01 2.29105651e-01
6.74110353e-01 -7.78919041e-01 -1.43041205e+00 -6.43100202e-01
-8.04064721e-02 -2.84167305e-02 3.51676881e-01 1.09230614e+00
-9.45640206e-01 9.81401861e-01 -1.52749777e-01 -4.55685556e-02
-1.28363299e+00 2.84758806e-01 -1.18501961e-01 -5.27810633e-01
-1.06464434e+00 2.47765645e-01 -2.68290430e-01 -1.67821988e-01
6.21303283e-02 -4.08011377e-01 -6.91730678e-01 5.36079824e-01
8.51297021e-01 8.22403848e-01 5.03587484e-01 -4.78811502e-01
-2.09705055e-01 1.75526768e-01 -2.88096607e-01 -3.56112272e-01
1.37449884e+00 -1.61634475e-01 -4.71541196e-01 5.62920153e-01
1.22230148e+00 1.72600612e-01 -1.60506144e-01 3.38591248e-01
2.76940793e-01 2.57904410e-01 1.60410002e-01 -1.03629386e+00
-3.13425004e-01 5.81389844e-01 8.88152793e-03 4.59176272e-01
1.06317294e+00 -1.90896362e-01 4.79492664e-01 1.77212179e-01
-4.21331316e-01 -1.22065461e+00 -2.01034069e-01 8.60654935e-02
1.30823600e+00 -7.75250971e-01 8.73736084e-01 3.69367041e-02
-7.94363737e-01 1.24687541e+00 -1.08624786e-01 5.97908944e-02
9.71591696e-02 -6.86888322e-02 -3.15000534e-01 -5.96205294e-01
-7.45418131e-01 3.77044410e-01 6.44523382e-01 2.88859993e-01
9.01522040e-01 -5.24721518e-02 -1.46265936e+00 5.26710868e-01
-7.50082374e-01 2.49135882e-01 1.07851815e+00 9.63247240e-01
-2.74670869e-01 -8.23038042e-01 -4.27611262e-01 1.38727403e+00
-1.09650648e+00 -2.86844701e-01 -8.65063012e-01 7.14010060e-01
-5.84016204e-01 1.31655324e+00 -6.07501864e-02 1.83389589e-01
4.36293572e-01 -4.33892710e-03 3.77148837e-01 -9.24665451e-01
-1.11366546e+00 -4.83425334e-02 7.70053506e-01 1.08115584e-01
-7.30327070e-01 -9.64769185e-01 -9.61884022e-01 -3.71106178e-01
-3.28156739e-01 5.92875719e-01 7.42728949e-01 9.49732840e-01
4.60089594e-01 9.84861195e-01 2.13778406e-01 -5.75716197e-01
-3.78682137e-01 -1.29684913e+00 -7.87016898e-02 -2.88977206e-01
4.44898397e-01 -4.18053977e-02 -2.50660628e-01 1.42555371e-01]
|
[12.348066329956055, 9.589519500732422]
|
3cc17aa6-29de-48bc-8a56-38ce744b3f38
|
least-to-most-prompting-enables-complex
|
2205.10625
| null |
https://arxiv.org/abs/2205.10625v3
|
https://arxiv.org/pdf/2205.10625v3.pdf
|
Least-to-Most Prompting Enables Complex Reasoning in Large Language Models
|
Chain-of-thought prompting has demonstrated remarkable performance on various natural language reasoning tasks. However, it tends to perform poorly on tasks which requires solving problems harder than the exemplars shown in the prompts. To overcome this challenge of easy-to-hard generalization, we propose a novel prompting strategy, least-to-most prompting. The key idea in this strategy is to break down a complex problem into a series of simpler subproblems and then solve them in sequence. Solving each subproblem is facilitated by the answers to previously solved subproblems. Our experimental results on tasks related to symbolic manipulation, compositional generalization, and math reasoning reveal that least-to-most prompting is capable of generalizing to more difficult problems than those seen in the prompts. A notable finding is that when the GPT-3 code-davinci-002 model is used with least-to-most prompting, it can solve the compositional generalization benchmark SCAN in any split (including length split) with an accuracy of at least 99% using just 14 exemplars, compared to only 16% accuracy with chain-of-thought prompting. This is particularly noteworthy because neural-symbolic models in the literature that specialize in solving SCAN are trained on the entire training set containing over 15,000 examples. We have included prompts for all the tasks in the Appendix.
|
['Olivier Bousquet', 'Claire Cui', 'Ed Chi', 'Quoc Le', 'Dale Schuurmans', 'Xuezhi Wang', 'Nathan Scales', 'Jason Wei', 'Le Hou', 'Nathanael Schärli', 'Denny Zhou']
|
2022-05-21
| null | null | null | null |
['arithmetic-reasoning']
|
['reasoning']
|
[ 3.60670239e-01 2.85689205e-01 2.80681159e-02 -4.19102460e-01
-7.29773462e-01 -8.56858671e-01 4.63590890e-01 3.67834061e-01
-1.59340113e-01 7.16988921e-01 -2.47412622e-01 -8.72749627e-01
-5.60447395e-01 -7.47276247e-01 -7.11553395e-01 -2.90552318e-01
-1.59995705e-01 7.87345111e-01 1.45176291e-01 -6.12226605e-01
6.33405209e-01 4.88831997e-01 -1.47674644e+00 6.17416739e-01
1.18088937e+00 6.95335090e-01 2.50458270e-01 6.37159288e-01
-4.58647788e-01 8.27305675e-01 -7.63736248e-01 -3.17818314e-01
3.87426943e-01 -3.15190226e-01 -1.31250274e+00 -3.32933187e-01
7.81999171e-01 -2.49037594e-01 -3.07435375e-02 9.32711720e-01
8.91646650e-03 5.19483328e-01 4.85847563e-01 -1.36953485e+00
-5.05756497e-01 8.35531116e-01 -2.09497020e-01 4.25357968e-01
1.06575549e+00 3.33314657e-01 9.60230768e-01 -6.51999772e-01
5.22644043e-01 1.30752349e+00 6.28098488e-01 6.22016013e-01
-1.57115221e+00 -6.97182477e-01 4.28426504e-01 2.84763992e-01
-1.36492753e+00 9.27697122e-02 2.41701618e-01 -3.89977545e-01
1.53069603e+00 4.29640830e-01 4.66434509e-01 7.51175940e-01
2.82413632e-01 5.83915174e-01 1.19518387e+00 -2.74747819e-01
2.06490204e-01 -3.02259445e-01 6.21899545e-01 7.61004746e-01
1.63366303e-01 7.63616338e-02 -4.68819916e-01 -3.80609632e-01
7.10293233e-01 2.43181754e-02 -2.31713116e-01 2.66367108e-01
-1.16776645e+00 6.51786029e-01 4.22735006e-01 2.49500588e-01
-3.26268405e-01 1.89802721e-01 3.58552784e-01 7.57645249e-01
-2.87734885e-02 1.21534765e+00 -7.08455205e-01 -8.94649625e-02
-1.11381304e+00 9.69488144e-01 1.13434696e+00 1.16791701e+00
6.99603975e-01 6.64296001e-02 -1.96583450e-01 4.29570407e-01
-3.32678139e-01 3.86425704e-01 4.39837962e-01 -1.29348767e+00
6.20506465e-01 7.34750330e-01 3.46356183e-02 -8.52349579e-01
-6.93243682e-01 -6.01625204e-01 -4.65496629e-01 2.08365589e-01
8.17662716e-01 -5.83213866e-02 -7.11929500e-01 1.92487502e+00
1.70308188e-01 8.82046744e-02 1.43805057e-01 7.58682728e-01
8.08307528e-01 8.82957041e-01 1.95593432e-01 -1.88660055e-01
1.25922370e+00 -1.12970412e+00 -1.96654692e-01 -3.39682072e-01
9.42468762e-01 -5.04406869e-01 1.46090817e+00 7.64218986e-01
-1.29327261e+00 -6.71355128e-01 -8.72107983e-01 -1.86783746e-01
-3.55556488e-01 -3.60326231e-01 9.73049998e-01 1.63048968e-01
-1.09626555e+00 8.10787022e-01 -4.85051394e-01 -2.74407506e-01
-9.78429019e-02 4.40324396e-01 -3.30884218e-01 -3.67517501e-01
-1.11384821e+00 1.17537999e+00 5.04622459e-01 -2.50919610e-01
-6.62013531e-01 -1.12091887e+00 -8.06972384e-01 4.50636566e-01
8.32300186e-01 -7.22695649e-01 1.79641068e+00 -6.93350315e-01
-1.18760598e+00 5.99436164e-01 -2.78619200e-01 -4.25857604e-01
2.80450881e-01 -3.64454657e-01 -1.53749362e-01 2.94898897e-01
2.59067178e-01 7.57013798e-01 6.00762188e-01 -9.23698246e-01
-4.66028214e-01 9.84481256e-03 6.83109403e-01 2.30611283e-02
1.56015247e-01 -3.82800475e-02 1.54252863e-03 -6.80076301e-01
4.35687304e-01 -9.55743909e-01 -2.93536186e-01 -3.51349562e-01
-1.94951475e-01 -6.48923218e-01 4.57866192e-01 -5.49490213e-01
1.23381138e+00 -1.98444855e+00 3.87787580e-01 9.99700651e-02
2.68917650e-01 2.23264083e-01 -3.36128205e-01 5.72669089e-01
-6.36408448e-01 1.01480577e-02 -2.16435075e-01 1.89667717e-01
2.36196086e-01 3.83102030e-01 -7.86765516e-01 -1.43068328e-01
2.04382196e-01 1.01940978e+00 -1.15352094e+00 -3.46609890e-01
-1.46079600e-01 -4.23143804e-01 -9.16922152e-01 2.55149424e-01
-8.29307377e-01 2.10750312e-01 -1.23817883e-01 5.90467095e-01
4.02312279e-01 -4.46496874e-01 1.75390411e-02 3.16097707e-01
2.38231674e-01 6.35481954e-01 -1.01989973e+00 1.87608302e+00
-3.91857564e-01 3.21004629e-01 -2.08412766e-01 -1.04095542e+00
7.66283810e-01 2.85700172e-01 -6.09073415e-02 -6.47834301e-01
-2.89895296e-01 4.56806213e-01 4.11382347e-01 -8.03802371e-01
5.01286566e-01 -3.11804861e-01 -3.27500165e-01 7.20293820e-01
-4.93509248e-02 -8.89088511e-01 8.64243031e-01 6.04487300e-01
1.37762368e+00 1.05286323e-01 4.58755493e-01 -3.40002477e-01
6.20690703e-01 3.65281731e-01 4.13708895e-01 1.05083334e+00
2.34382361e-01 2.56648213e-01 7.35105693e-01 -6.57437623e-01
-7.94478774e-01 -9.36428130e-01 1.92010745e-01 1.33132362e+00
-6.86810911e-02 -8.84711504e-01 -7.48101830e-01 -6.11276329e-01
2.06974670e-01 1.48049784e+00 -3.83807033e-01 -3.20501000e-01
-8.42328787e-01 -9.40678045e-02 6.15071297e-01 7.29372680e-01
4.14643854e-01 -1.06659400e+00 -8.78092110e-01 1.77002966e-01
-1.24791734e-01 -9.11312819e-01 -1.64710909e-01 5.04101038e-01
-1.17845094e+00 -1.10627079e+00 -4.89715576e-01 -6.16377115e-01
7.84416437e-01 1.14978112e-01 1.27748811e+00 6.20034695e-01
-1.97081700e-01 4.89829451e-01 -3.58589709e-01 -1.05463862e-01
-3.51535499e-01 1.97040930e-01 -2.30782241e-01 -9.13851619e-01
2.91152596e-01 -6.32796526e-01 5.95094003e-02 1.38850912e-01
-8.27367663e-01 1.46657094e-01 4.02366698e-01 8.86498392e-01
2.00154200e-01 3.23863737e-02 5.27996421e-01 -8.36240113e-01
1.00665605e+00 -5.66836774e-01 -4.70679879e-01 3.61037076e-01
-5.46393096e-01 3.12755376e-01 1.07980883e+00 -6.35459244e-01
-7.87300766e-01 -2.02481791e-01 1.74744036e-02 -1.07645273e-01
-4.24288601e-01 8.60518336e-01 4.07163709e-01 -5.35045378e-02
1.04548740e+00 3.59178722e-01 -2.83898085e-01 -2.34972879e-01
2.96157837e-01 -8.97395983e-02 7.51563013e-01 -1.61662877e+00
8.11768413e-01 -3.24406207e-01 4.10667449e-01 -5.79562485e-01
-1.09040117e+00 -1.18759103e-01 -1.70155853e-01 4.90301624e-02
3.33231568e-01 -4.06098723e-01 -9.52395856e-01 2.62970507e-01
-1.35534024e+00 -7.82432854e-01 -4.07241195e-01 1.45596370e-01
-7.33832955e-01 3.66483241e-01 -7.32883573e-01 -6.20218039e-01
-1.56857401e-01 -1.31301463e+00 8.12693059e-01 3.02022099e-01
-1.05831945e+00 -7.90691972e-01 -3.58681589e-01 9.35510024e-02
5.25109470e-01 -2.59299260e-02 1.67578793e+00 -1.16392696e+00
-5.65810382e-01 -1.58813726e-02 -1.47721142e-01 1.19878583e-01
-2.29935706e-01 -3.13834995e-01 -3.28024209e-01 -1.26668200e-01
2.30289519e-01 -6.46871746e-01 5.62104523e-01 -1.22036301e-01
1.32898545e+00 -2.98742801e-01 -1.30682021e-01 5.72735488e-01
9.91711140e-01 2.60709137e-01 3.32938462e-01 4.09340590e-01
1.76937386e-01 4.45585847e-01 7.94007599e-01 3.62027586e-02
3.19599748e-01 5.41105330e-01 2.09360585e-01 3.60296249e-01
1.34189427e-01 -8.91196504e-02 1.67679727e-01 2.43469134e-01
-9.07771960e-02 1.53160647e-01 -1.33090031e+00 1.11309767e-01
-1.68700087e+00 -9.23185229e-01 -2.73379415e-01 1.82248664e+00
1.14384544e+00 3.90349418e-01 2.46324427e-02 4.38118875e-01
2.36123785e-01 -1.85661763e-01 -3.95449132e-01 -8.66047323e-01
2.37494841e-01 7.06350029e-01 -2.09790096e-01 6.94014013e-01
-5.11101961e-01 1.19699764e+00 6.89409399e+00 8.18110824e-01
-1.15386271e+00 -2.95718580e-01 1.51522577e-01 -9.19215083e-02
-1.72581092e-01 1.75732926e-01 -7.21251607e-01 1.56163186e-01
8.60430539e-01 -3.51783156e-01 9.38519180e-01 9.06771719e-01
-1.74079463e-01 -5.79371154e-01 -1.87151158e+00 7.23276734e-01
2.69124061e-02 -1.19323277e+00 2.44909734e-01 -7.15481520e-01
6.47356033e-01 -6.35357022e-01 7.10413456e-02 7.88666189e-01
1.60629690e-01 -1.27168787e+00 8.13726783e-01 2.98779756e-01
5.71258783e-01 -4.56805021e-01 3.47292781e-01 8.57753575e-01
-9.09429073e-01 -3.76226485e-01 -1.90531164e-01 -8.45612764e-01
-1.23916119e-02 2.60646105e-01 -8.81600857e-01 5.19197464e-01
3.91732723e-01 3.12726706e-01 -6.26385629e-01 1.01727128e+00
-8.27087522e-01 4.34600770e-01 -4.44542170e-01 -1.09640375e-01
3.31589520e-01 1.29258409e-01 5.15118122e-01 1.25763023e+00
3.89884859e-01 7.85823226e-01 3.44372034e-01 1.18546927e+00
4.52103525e-01 -2.07474962e-01 -2.97289193e-01 -1.78477257e-01
5.10162771e-01 8.30113649e-01 -8.45884264e-01 -7.36117184e-01
-1.08111158e-01 7.23206758e-01 5.03767908e-01 6.58449292e-01
-7.19383597e-01 -5.61366856e-01 1.67855725e-01 2.30010659e-01
-6.39020652e-02 -6.41669750e-01 -5.36861777e-01 -9.64494169e-01
1.90600334e-03 -1.46186996e+00 4.63756204e-01 -1.22309208e+00
-1.06585205e+00 4.55582947e-01 8.44624639e-01 -4.43636239e-01
-5.65252006e-01 -9.63307500e-01 -7.91790247e-01 1.16532028e+00
-1.09682965e+00 -5.45196831e-01 -4.04608428e-01 6.27601981e-01
7.75154769e-01 2.60298867e-02 1.07780600e+00 -7.96294212e-02
-2.35715240e-01 4.74730343e-01 -6.31940663e-01 -2.98150748e-01
5.32911479e-01 -1.37075579e+00 4.42529619e-01 6.20565414e-01
-1.46755114e-01 1.32925105e+00 1.02589250e+00 -6.91057563e-01
-1.23841143e+00 -4.52157259e-01 1.13793457e+00 -2.13353142e-01
7.74311602e-01 -2.70540081e-02 -1.16699803e+00 9.81128037e-01
2.42453697e-03 -3.33765805e-01 3.87905568e-01 2.07664952e-01
-6.51031971e-01 1.29423812e-01 -1.02374649e+00 7.65076339e-01
1.18098438e+00 -4.20175612e-01 -1.51144946e+00 6.31906033e-01
8.65636885e-01 -9.89395320e-01 -5.18464983e-01 2.36114532e-01
1.18783362e-01 -9.34071898e-01 9.85716581e-01 -1.08280134e+00
9.70234215e-01 -1.36712715e-01 -1.45257831e-01 -1.37079871e+00
-4.23758328e-01 -7.37137437e-01 -9.45232362e-02 8.19423974e-01
4.12567526e-01 -9.69128788e-01 5.66635191e-01 9.32577431e-01
-3.92074585e-01 -1.04577839e+00 -5.42357564e-01 -1.03573966e+00
4.49969381e-01 -4.48798269e-01 7.19655573e-01 8.78792763e-01
5.59626937e-01 2.78520793e-01 3.25330228e-01 -6.92226412e-03
3.26092213e-01 5.51039577e-01 8.34838688e-01 -1.32180643e+00
-6.67904198e-01 -4.97116029e-01 1.33936688e-01 -1.46151173e+00
5.81486166e-01 -1.38244116e+00 -1.36674881e-01 -1.40474463e+00
1.09661713e-01 -4.54795063e-01 1.37953997e-01 9.54917729e-01
-3.26745212e-01 -4.37921286e-01 3.14527184e-01 1.77807696e-02
-4.91422027e-01 -2.32912949e-03 1.38613379e+00 -8.47979933e-02
-6.86597973e-02 -1.20170727e-01 -7.56246209e-01 7.36262083e-01
6.37018979e-01 -5.68745852e-01 -4.69485670e-01 -4.15640235e-01
6.70895934e-01 4.07317132e-01 6.52544558e-01 -1.05905521e+00
5.10897994e-01 -5.98541141e-01 2.34108344e-02 -4.56345528e-01
2.79050052e-01 -7.08539367e-01 1.19048260e-01 7.21826017e-01
-7.15524852e-01 5.35951495e-01 7.07124591e-01 -1.16096884e-02
-5.74429631e-02 -7.60997772e-01 4.29688662e-01 -4.20573562e-01
-7.62434781e-01 -3.29522133e-01 -5.24185479e-01 4.26283240e-01
1.02514219e+00 -1.01814687e-01 -6.47636116e-01 -2.75625467e-01
-8.57437611e-01 4.10209477e-01 3.89793813e-01 5.04929163e-02
4.74315822e-01 -1.00482130e+00 -5.06335080e-01 1.24516539e-01
-1.22654825e-01 3.71931165e-01 1.55760851e-02 8.99733424e-01
-6.08943343e-01 7.90114164e-01 -2.19714493e-01 -5.55737078e-01
-1.19693410e+00 9.59176302e-01 1.50504112e-01 -3.38556826e-01
-6.47475302e-01 1.08163691e+00 6.24790415e-02 -4.86028671e-01
3.04723114e-01 -8.96082640e-01 3.70866686e-01 -3.58606905e-01
4.26119685e-01 4.42641705e-01 2.81433668e-02 1.35533720e-01
-3.00182313e-01 5.11505246e-01 -2.37808675e-01 -1.38807688e-02
1.28354657e+00 5.26295006e-01 -3.73265415e-01 3.41302276e-01
5.94933331e-01 -2.91595459e-01 -6.40639186e-01 -1.94188893e-01
2.22087070e-01 -4.40543920e-01 -5.73928475e-01 -1.10669518e+00
-4.48373377e-01 8.99765491e-01 -5.10273159e-01 4.59126532e-01
9.60000396e-01 -5.79353906e-02 6.47066534e-01 1.13335454e+00
5.39946079e-01 -6.85882449e-01 3.29081327e-01 1.07198286e+00
1.24026215e+00 -7.01941729e-01 1.85691342e-02 -5.59894443e-01
-2.75784254e-01 1.41380024e+00 1.00747871e+00 -2.77413696e-01
-8.51057619e-02 4.06058788e-01 -3.20872903e-01 -3.10570568e-01
-9.84386444e-01 4.44271863e-01 8.38227868e-02 4.15938020e-01
4.49024171e-01 -1.53480783e-01 -3.37114781e-01 9.11726773e-01
-6.72992051e-01 8.62577260e-02 4.25403118e-01 1.11623359e+00
-5.82055569e-01 -1.13579547e+00 -7.39762187e-01 3.80156487e-01
-3.26815508e-02 -3.67104679e-01 -3.91486883e-01 9.08206105e-01
2.75634252e-03 8.11195314e-01 -2.41348073e-01 -2.19083965e-01
2.94508338e-01 5.84347010e-01 9.43930924e-01 -1.02850127e+00
-1.01144135e+00 -5.50413668e-01 1.13850676e-01 -8.08877707e-01
1.51168570e-01 -6.18644416e-01 -1.71249878e+00 -6.62097573e-01
9.73273888e-02 3.02983195e-01 2.46270031e-01 1.27925754e+00
1.80049296e-02 6.70627832e-01 -1.97225258e-01 -1.06836665e+00
-1.28544247e+00 -7.03902483e-01 -2.51489639e-01 4.46890771e-01
1.46102995e-01 -6.76957667e-01 -4.84244764e-01 -1.48736864e-01]
|
[9.406899452209473, 7.279869556427002]
|
74253061-17f5-42d9-bf48-8bccceeb6591
|
substructure-aware-graph-neural-networks
| null | null |
https://ojs.aaai.org/index.php/AAAI/article/view/26318
|
https://ojs.aaai.org/index.php/AAAI/article/view/26318/26090
|
Substructure Aware Graph Neural Networks
|
Despite the great achievements of Graph Neural Networks (GNNs) in graph learning, conventional GNNs struggle to break through the upper limit of the expressiveness of first-order Weisfeiler-Leman graph isomorphism test algorithm (1-WL) due to the consistency of the propagation paradigm of GNNs with the 1-WL.Based on the fact that it is easier to distinguish the original graph through subgraphs, we propose a novel framework neural network framework called Substructure Aware Graph Neural Networks (SAGNN) to address these issues. We first propose a Cut subgraph which can be obtained from the original graph by continuously and selectively removing edges. Then we extend the random walk encoding paradigm to the return probability of the rooted node on the subgraph to capture the structural information and use it as a node feature to improve the expressiveness of GNNs. We theoretically prove that our framework is more powerful than 1-WL, and is superior in structure perception. Our extensive experiments demonstrate the effectiveness of our framework, achieving state-of-the-art performance on a variety of well-proven graph tasks, and GNNs equipped with our framework perform flawlessly even in 3-WL failed graphs. Specifically, our framework achieves a maximum performance improvement of 83% compared to the base models and 32% compared to the previous state-of-the-art methods.
|
['H', '& Qu', 'M.', 'Zhang', 'L.', 'Zhou', 'Chen', 'W.', 'Liu', 'D.', 'Zeng']
|
2023-06-26
| null | null | null |
proceedings-of-the-aaai-conference-on-6
|
['graph-learning', 'graph-regression']
|
['graphs', 'graphs']
|
[ 2.70103216e-01 4.52917784e-01 -3.04215610e-01 -2.25797556e-02
-4.53157313e-02 -4.76689726e-01 3.12729120e-01 -9.75189544e-03
-8.61492008e-02 4.72901106e-01 -1.63029820e-01 -7.28404760e-01
-5.18720686e-01 -1.38622284e+00 -9.11737561e-01 -6.04173362e-01
-5.00993252e-01 3.92369837e-01 3.88772875e-01 -3.44410777e-01
-1.21457949e-01 6.28100157e-01 -1.21075499e+00 -3.35545763e-02
8.50397527e-01 7.57567883e-01 1.04853086e-01 6.78137898e-01
-1.24188989e-01 7.70797133e-01 -2.25412294e-01 -4.03498173e-01
3.55581760e-01 -4.44959193e-01 -9.31228817e-01 -9.86648276e-02
4.63041216e-01 -2.20849097e-01 -8.93792093e-01 1.26312566e+00
2.85758048e-01 -6.56989217e-02 4.52152342e-01 -1.50121844e+00
-9.36754942e-01 1.10626686e+00 -5.61936438e-01 -2.22670622e-02
2.93343723e-01 -9.63205285e-03 1.53844273e+00 -4.19532478e-01
7.08816230e-01 1.11923838e+00 9.34127510e-01 4.25613582e-01
-1.14817321e+00 -5.66188276e-01 4.84675467e-01 3.30953121e-01
-1.25985801e+00 4.35082577e-02 9.81968939e-01 -8.34948793e-02
8.84498119e-01 2.16596246e-01 7.86437094e-01 1.11883819e+00
1.67780355e-01 6.16972208e-01 8.13256502e-01 -3.52584004e-01
-7.88509399e-02 -5.59259892e-01 3.70056480e-01 1.36534381e+00
6.21072114e-01 1.90358818e-01 -1.78449735e-01 2.49568731e-01
9.16851759e-01 -5.31281978e-02 -4.60124820e-01 -6.95133507e-01
-8.59935403e-01 7.18655109e-01 1.16584074e+00 4.11030114e-01
-2.25389991e-02 3.93951267e-01 2.28107482e-01 4.30052698e-01
1.41116112e-01 4.35750127e-01 -5.79673238e-02 4.02722090e-01
-6.42539024e-01 -1.20206662e-01 8.87498438e-01 9.63931799e-01
7.85321712e-01 2.60278285e-01 -1.62247270e-01 3.34173322e-01
2.18744963e-01 1.65113002e-01 7.56520703e-02 -3.76264513e-01
3.55406046e-01 1.06975341e+00 -6.40596032e-01 -1.41695130e+00
-7.57906318e-01 -1.00765729e+00 -1.28206289e+00 -6.52890578e-02
3.42693835e-01 1.59044713e-01 -1.13435662e+00 1.99714482e+00
8.78688544e-02 3.67058456e-01 -1.61014438e-01 5.35804272e-01
9.96908665e-01 3.25620979e-01 -2.71306783e-01 1.08623922e-01
8.83185327e-01 -1.13158631e+00 -2.93092042e-01 -2.75063664e-01
8.62810433e-01 1.90444514e-01 9.93571758e-01 1.39562115e-01
-7.99934387e-01 -5.53630888e-01 -1.30935943e+00 1.55508503e-01
-4.28972840e-01 -4.54803407e-01 1.00613344e+00 7.59041250e-01
-1.35146821e+00 9.52037454e-01 -7.46749461e-01 -3.64423305e-01
5.02901971e-01 4.20594782e-01 -6.96886599e-01 -1.81423038e-01
-1.04229963e+00 4.35380489e-01 7.62469292e-01 2.93779314e-01
-8.54928613e-01 -4.62907672e-01 -1.11758447e+00 4.16997433e-01
7.65668988e-01 -7.93393195e-01 6.74023330e-01 -8.35873425e-01
-1.13192868e+00 6.95231140e-01 1.61049619e-01 -6.45259202e-01
3.24806690e-01 1.78721502e-01 -4.78567928e-01 3.19183201e-01
-1.46167085e-01 2.84314781e-01 4.65255678e-01 -1.19490004e+00
-2.90644675e-01 -2.96191216e-01 4.90970701e-01 -1.38816714e-01
-4.23190325e-01 -4.99417216e-01 -5.18100083e-01 -4.38900054e-01
2.76071131e-01 -8.32226336e-01 -1.19506471e-01 -1.45316273e-01
-9.01134253e-01 -2.16006815e-01 6.51595771e-01 -3.25538337e-01
1.33787656e+00 -1.97359145e+00 2.19531938e-01 5.82140744e-01
1.17647254e+00 3.94829482e-01 -6.16021156e-01 6.50325537e-01
-2.72618562e-01 3.55555683e-01 -2.20589489e-01 3.54511663e-02
1.35857582e-01 3.34525257e-01 1.63721684e-02 5.12639523e-01
2.25557983e-02 1.25504160e+00 -1.02617216e+00 -2.60246754e-01
-1.02316414e-03 1.33575127e-01 -4.92207617e-01 1.46328971e-01
-7.71801397e-02 1.32640064e-01 -3.43259782e-01 5.19658566e-01
7.69033015e-01 -7.24854410e-01 5.82522869e-01 -1.48899227e-01
5.24784088e-01 1.56706512e-01 -1.07689869e+00 1.51351178e+00
-6.25042170e-02 3.99726689e-01 -1.77329313e-02 -1.28482592e+00
8.91341448e-01 -1.47472113e-01 2.21447557e-01 -6.51091456e-01
-1.13516524e-02 3.08032539e-02 4.44052994e-01 -1.13588773e-01
1.62389517e-01 2.31549755e-01 1.81001514e-01 2.94262856e-01
2.22317070e-01 3.53228211e-01 5.06362259e-01 4.40519512e-01
1.60632432e+00 5.15350178e-02 3.33582163e-01 -3.69663507e-01
4.53590304e-01 -5.53298473e-01 4.00634497e-01 1.12717569e+00
-1.75967664e-01 3.88800293e-01 9.67637300e-01 -5.92576265e-01
-6.80194676e-01 -1.30798900e+00 3.09874266e-01 1.06295466e+00
3.22396189e-01 -7.38444626e-01 -7.33983874e-01 -1.00318193e+00
-7.42416177e-03 4.56063122e-01 -8.07199180e-01 -5.73376417e-01
-5.95186889e-01 -6.29458904e-01 7.52913952e-01 7.20662057e-01
6.70476854e-01 -9.01295483e-01 -3.15206908e-02 1.57058731e-01
1.32790670e-01 -1.27155566e+00 -1.97680458e-01 1.48202881e-01
-7.35628664e-01 -1.45432353e+00 -2.42742971e-01 -1.01401913e+00
7.98588634e-01 3.16078186e-01 1.24881756e+00 6.93513751e-01
3.63869971e-04 2.27771163e-01 -3.89747381e-01 7.55908415e-02
-4.84815896e-01 4.54521924e-01 -1.99580550e-01 -5.88355847e-02
-1.76800787e-01 -1.10958660e+00 -3.82274717e-01 1.60324514e-01
-9.45080996e-01 2.39647478e-01 6.64347172e-01 8.56764257e-01
4.16050225e-01 3.76872361e-01 4.64650005e-01 -1.32195103e+00
6.57214463e-01 -3.59243959e-01 -5.57290673e-01 3.85621428e-01
-9.08341169e-01 3.11161011e-01 1.03991151e+00 -1.51892632e-01
-3.46319228e-01 -3.14694136e-01 -6.81561083e-02 -1.89790860e-01
1.25289410e-01 8.10323238e-01 -2.57119596e-01 -4.33248430e-01
4.03382987e-01 2.41378054e-01 -1.21610440e-01 -2.68937290e-01
3.99718642e-01 -3.66182141e-02 6.72918022e-01 -5.79281271e-01
1.14100838e+00 3.42739284e-01 6.87724471e-01 -5.42967439e-01
-7.89964199e-01 -1.31959990e-01 -6.01627469e-01 -1.69398174e-01
4.90665466e-01 -4.41100180e-01 -1.01471043e+00 5.63404024e-01
-8.87967467e-01 -2.89186746e-01 1.14868015e-01 8.45273584e-02
-3.02908003e-01 8.24825048e-01 -8.64142239e-01 -5.93529105e-01
-3.78512710e-01 -9.62124884e-01 6.44951582e-01 4.27732430e-03
3.26336116e-01 -1.26061606e+00 -8.70796070e-02 -1.60959944e-01
3.78540754e-01 6.32174253e-01 1.55961990e+00 -9.36990142e-01
-8.09390903e-01 -1.35896951e-01 -5.15100718e-01 1.45387545e-01
3.20445523e-02 -9.16036293e-02 -5.84994912e-01 -6.83570981e-01
-4.28859383e-01 -1.51522040e-01 1.27501261e+00 1.96052834e-01
1.27115512e+00 -3.04660976e-01 -4.84951824e-01 9.69621122e-01
1.75107384e+00 -8.44796598e-02 7.94211447e-01 1.90334201e-01
1.15261126e+00 1.96788356e-01 -3.36404443e-01 -1.39707297e-01
4.85898197e-01 3.06050360e-01 1.00406575e+00 -1.80497721e-01
-2.74197131e-01 -6.76633596e-01 2.22782895e-01 9.91755247e-01
-2.88261831e-01 -7.25579262e-01 -8.38597476e-01 3.16908777e-01
-1.83470237e+00 -7.28308856e-01 -2.76948839e-01 2.00470042e+00
2.29440287e-01 4.93140578e-01 7.93094281e-04 3.00021946e-01
8.15279663e-01 5.97431421e-01 -4.97709960e-01 -2.54507720e-01
-1.65636197e-01 3.84403110e-01 6.08133912e-01 4.19795424e-01
-9.58356380e-01 1.00095665e+00 6.45686245e+00 7.92549312e-01
-8.24695826e-01 -2.94350147e-01 3.59170973e-01 5.77350318e-01
-4.35426861e-01 1.26346901e-01 -4.22249377e-01 1.75566643e-01
8.37612569e-01 -2.29915395e-01 7.05883980e-01 7.39943087e-01
-5.04586577e-01 4.49207902e-01 -1.39758468e+00 6.84818149e-01
1.05611324e-01 -1.28117728e+00 3.73793632e-01 1.61510348e-01
5.16493261e-01 5.00946715e-02 -1.75564587e-01 5.12634099e-01
4.40304101e-01 -1.29359353e+00 2.98838913e-01 3.43999445e-01
7.62272120e-01 -7.23511934e-01 7.56094337e-01 3.85517120e-01
-1.56510973e+00 -5.22283837e-02 -4.18588609e-01 -2.16800421e-01
-2.08576813e-01 5.56425333e-01 -7.75164425e-01 1.31417120e+00
3.21367651e-01 7.95151830e-01 -7.75910497e-01 1.03585756e+00
-4.61387336e-01 5.81452549e-01 -1.97850540e-01 -4.48516048e-02
4.28096384e-01 -2.05963805e-01 6.37736857e-01 9.33667302e-01
1.79231808e-01 -3.17711085e-01 3.44321162e-01 9.95025873e-01
-6.23672426e-01 -1.18301712e-01 -8.63735080e-01 -3.64630878e-01
3.89868528e-01 1.12262321e+00 -9.92922843e-01 4.18739393e-02
-2.36954793e-01 8.53661716e-01 9.26933646e-01 4.42091674e-01
-8.35731328e-01 -7.23779142e-01 2.00431153e-01 -6.47709565e-03
5.43389916e-01 -2.82622486e-01 2.06155434e-01 -1.00390160e+00
2.60774344e-01 -8.53126764e-01 4.58357930e-01 -6.14963830e-01
-1.40314388e+00 1.02204275e+00 -1.70524716e-01 -7.48167694e-01
2.60464326e-02 -1.01124513e+00 -8.03970456e-01 4.11110371e-01
-1.50412107e+00 -1.46619797e+00 -4.61162865e-01 5.58842599e-01
-2.13200763e-01 1.22888284e-02 8.68296742e-01 1.18109807e-01
-4.74729031e-01 8.30369830e-01 -1.24372989e-01 5.81105649e-01
1.10121354e-01 -1.51578212e+00 7.74707854e-01 1.20612335e+00
5.23063242e-01 5.56261718e-01 3.02814096e-01 -6.59312606e-01
-1.67972779e+00 -1.03394306e+00 5.15481532e-01 2.89939009e-02
8.38535368e-01 -6.02374375e-01 -8.89108777e-01 9.50917542e-01
-3.31913531e-02 1.04759939e-01 2.11985976e-01 4.14278150e-01
-8.84386003e-01 -2.63335913e-01 -8.14127028e-01 8.02592039e-01
1.80905879e+00 -5.22826374e-01 -3.88246268e-01 2.90831447e-01
1.13292134e+00 -3.99125606e-01 -7.24187374e-01 7.49642193e-01
4.54468399e-01 -1.13924897e+00 8.76007140e-01 -8.39050412e-01
3.35925370e-01 -1.36469573e-01 -4.55967896e-02 -1.28824484e+00
-7.47288108e-01 -7.86697209e-01 -2.34471917e-01 8.97516906e-01
3.44392449e-01 -9.98810589e-01 9.13087249e-01 -1.95640307e-02
-2.78749436e-01 -9.67236400e-01 -8.93807650e-01 -1.08714116e+00
-5.74647747e-02 -4.49889868e-01 7.65976012e-01 8.81426156e-01
-2.46545538e-01 5.74615538e-01 -2.31168807e-01 3.88494968e-01
6.65363014e-01 3.27240556e-01 7.24599421e-01 -1.54236388e+00
-5.61240196e-01 -8.03668201e-01 -8.24860752e-01 -1.18110120e+00
4.30576771e-01 -1.48232555e+00 -3.42015356e-01 -2.04917431e+00
3.48695606e-01 -1.32386506e-01 -6.51742458e-01 6.57732069e-01
-1.21952899e-01 9.71407518e-02 2.35174611e-01 -1.78794131e-01
-7.62018442e-01 4.53818947e-01 1.46916926e+00 -3.37610334e-01
8.37399811e-02 -8.70176852e-02 -8.93827856e-01 5.89213669e-01
5.11561155e-01 -3.66690725e-01 -6.05586648e-01 -3.71451855e-01
4.92386043e-01 -1.00365490e-01 5.50411463e-01 -1.07629704e+00
2.34942615e-01 1.38173595e-01 1.01350777e-01 -3.70039612e-01
-1.19680017e-01 -7.10949481e-01 1.30241215e-01 7.12701261e-01
-2.46083215e-01 1.46992981e-01 9.63703613e-04 9.16974306e-01
3.13097984e-02 -2.81640049e-02 4.28220749e-01 -2.69018747e-02
-7.11345434e-01 6.73919439e-01 1.69021517e-01 6.88259155e-02
7.02074409e-01 -3.78324121e-01 -7.45998979e-01 -4.31382179e-01
-5.79422295e-01 2.84833014e-01 3.45050633e-01 1.90654203e-01
6.29795253e-01 -1.27323747e+00 -6.60837173e-01 3.90287668e-01
1.80998236e-01 -6.53570890e-02 2.09302723e-01 8.11104238e-01
-6.50778949e-01 1.94009230e-01 -2.44159549e-01 -4.60951358e-01
-9.96483982e-01 7.94831812e-01 4.76386666e-01 -7.73656845e-01
-9.64442909e-01 7.38613009e-01 4.76951182e-01 -5.02697527e-01
2.09307238e-01 -4.35732007e-01 3.48745435e-02 -6.49440229e-01
1.18079998e-01 3.22564542e-01 9.73290354e-02 -3.57758194e-01
-3.13771963e-01 4.34333265e-01 -1.26066357e-01 5.58051705e-01
1.41429794e+00 2.36348197e-01 -3.75157744e-01 1.48780540e-01
1.13695908e+00 9.40010697e-02 -8.81175399e-01 -2.36683011e-01
2.26981919e-02 -1.44828737e-01 -1.70678377e-01 -7.23329902e-01
-1.15216935e+00 6.40242159e-01 6.28524050e-02 6.21014953e-01
1.21207154e+00 5.56670595e-03 8.69392335e-01 7.46411622e-01
3.93517077e-01 -5.31172216e-01 -8.74480605e-02 5.45506299e-01
6.23752236e-01 -8.71108830e-01 -9.50468034e-02 -7.64170647e-01
6.76613394e-03 1.20237398e+00 5.66714466e-01 -3.94277990e-01
5.96733093e-01 1.47497132e-01 -5.90466976e-01 -4.33367819e-01
-5.78032672e-01 -3.72781932e-01 5.16055167e-01 7.99224317e-01
4.80234716e-03 1.79309770e-01 -7.77620450e-02 5.81582785e-01
-1.95338160e-01 -3.28172833e-01 4.97451514e-01 4.53129470e-01
-3.50174189e-01 -1.03596377e+00 4.75295693e-01 4.31989610e-01
-2.01290697e-01 -2.55728066e-01 -6.36247218e-01 1.30391145e+00
-1.99247286e-01 7.92146802e-01 -1.70528084e-01 -8.43600690e-01
1.88495144e-01 -2.33327866e-01 6.78055644e-01 -5.42398334e-01
-3.78560901e-01 -3.65251631e-01 2.40913272e-01 -7.54557014e-01
-3.43338788e-01 2.39559069e-01 -1.22666383e+00 -6.03559911e-01
-3.66036028e-01 6.40741959e-02 7.71023929e-02 9.26197767e-01
4.43014562e-01 7.86170125e-01 5.98255873e-01 -4.32069540e-01
-3.93938482e-01 -6.95836961e-01 -9.27039742e-01 4.42539901e-01
3.08927774e-01 -5.64242303e-01 -5.38477302e-01 -6.09386265e-01]
|
[6.982449531555176, 6.222543716430664]
|
48f50e1a-719f-4d6b-a9f3-96bab99a44a2
|
confidence-intervals-for-error-rates-in
|
2306.01198
| null |
https://arxiv.org/abs/2306.01198v1
|
https://arxiv.org/pdf/2306.01198v1.pdf
|
Confidence Intervals for Error Rates in Matching Tasks: Critical Review and Recommendations
|
Matching algorithms are commonly used to predict matches between items in a collection. For example, in 1:1 face verification, a matching algorithm predicts whether two face images depict the same person. Accurately assessing the uncertainty of the error rates of such algorithms can be challenging when data are dependent and error rates are low, two aspects that have been often overlooked in the literature. In this work, we review methods for constructing confidence intervals for error rates in matching tasks such as 1:1 face verification. We derive and examine the statistical properties of these methods and demonstrate how coverage and interval width vary with sample size, error rates, and degree of data dependence using both synthetic and real-world datasets. Based on our findings, we provide recommendations for best practices for constructing confidence intervals for error rates in matching tasks.
|
['Pietro Perona', 'Pratik Patil', 'Riccardo Fogliato']
|
2023-06-01
| null | null | null | null |
['face-verification']
|
['computer-vision']
|
[ 3.99373889e-01 -3.89093235e-02 -4.67080891e-01 -9.12712455e-01
-7.72629142e-01 -6.83085263e-01 4.46281254e-01 5.15804589e-01
-4.15544659e-01 6.65403485e-01 8.83204788e-02 -3.56138945e-01
-3.12651098e-01 -6.36693358e-01 -8.26199532e-01 7.95310289e-02
-2.71715254e-01 3.14589143e-01 -3.00380915e-01 5.47958612e-01
3.42483819e-01 4.78465080e-01 -1.83361781e+00 1.54337108e-01
6.72020614e-01 1.28991342e+00 -5.88717222e-01 4.45394129e-01
1.69154197e-01 4.89889346e-02 -7.57988751e-01 -1.10947907e+00
4.00928199e-01 -2.81347692e-01 -3.57693404e-01 1.44022867e-01
8.85006428e-01 -5.19272804e-01 8.29260573e-02 1.08971763e+00
3.34140301e-01 -7.78513327e-02 1.15821612e+00 -1.90834308e+00
-6.51044488e-01 5.27540565e-01 -7.97037601e-01 1.72832534e-01
8.54388714e-01 -6.87739924e-02 8.60749900e-01 -8.70782673e-01
2.48124644e-01 1.32202232e+00 1.04036391e+00 3.87418211e-01
-1.29325449e+00 -1.29314780e+00 -1.29066318e-01 -1.47854641e-01
-1.65273404e+00 -8.10542285e-01 1.90976217e-01 -5.84930182e-01
3.26071829e-01 1.62256330e-01 3.76505822e-01 9.96576905e-01
2.25005478e-01 3.48179638e-01 1.15479052e+00 -4.97964680e-01
6.97765499e-02 3.74488145e-01 1.68154672e-01 5.48806369e-01
9.79021072e-01 4.38075632e-01 -8.38454843e-01 -6.44487083e-01
7.48065591e-01 -8.39331299e-02 1.39382213e-01 -1.06446169e-01
-7.21053600e-01 9.61855710e-01 -7.32052699e-02 -1.99145868e-01
-2.09523048e-02 -1.05357453e-01 2.08492994e-01 2.25616530e-01
4.86848503e-01 2.27892756e-01 -5.45889214e-02 4.65196073e-02
-1.10857916e+00 5.44568062e-01 8.53788972e-01 1.22509694e+00
3.76226038e-01 -4.58975792e-01 -4.07795429e-01 9.09361839e-01
2.87624478e-01 6.06918037e-01 1.07742988e-01 -9.17203307e-01
2.95677185e-01 2.13627338e-01 4.23309743e-01 -1.17633533e+00
-2.24453598e-01 6.51704222e-02 -4.30931389e-01 1.07906200e-01
8.29640388e-01 -3.01159352e-01 -7.55746663e-01 1.84000361e+00
2.72840768e-01 1.37636364e-01 -3.19265932e-01 6.54281497e-01
6.44097805e-01 1.27162248e-01 4.23173487e-01 -4.11972433e-01
1.56680226e+00 -1.80474669e-01 -6.28176272e-01 -2.39576668e-01
3.36372435e-01 -1.02012706e+00 8.62385869e-01 5.92880733e-02
-1.13470817e+00 -4.55336154e-01 -8.29045773e-01 1.92701653e-01
-2.14349516e-02 -1.14160255e-02 5.67715764e-01 1.24163604e+00
-6.95547760e-01 6.75512612e-01 -3.05292845e-01 -3.44912440e-01
5.60924828e-01 3.40613067e-01 -4.07312453e-01 -1.78604007e-01
-1.00442159e+00 7.81238914e-01 -9.93007049e-02 -1.19868957e-01
-3.45411539e-01 -9.39774871e-01 -9.91077423e-01 9.60346013e-02
7.92734772e-02 -4.33306932e-01 1.28992903e+00 -8.68471146e-01
-6.59912944e-01 1.11315894e+00 -3.51795435e-01 -3.54976594e-01
5.07007003e-01 2.18668431e-02 -6.22029483e-01 -3.51219058e-01
2.89734840e-01 6.41770720e-01 9.26490963e-01 -9.89738286e-01
-7.13053584e-01 -5.51683247e-01 -2.35529378e-01 -1.87546648e-02
-4.81826216e-02 4.97051060e-01 -1.49845034e-01 -6.89162433e-01
-9.43138972e-02 -9.67843711e-01 2.21755598e-02 5.16129851e-01
-7.03242570e-02 -3.24254125e-01 3.42637211e-01 -4.36349958e-01
1.20476079e+00 -2.22219515e+00 -7.06947327e-01 5.36155343e-01
1.56587902e-02 -2.13641360e-01 2.93070711e-02 5.42043038e-02
-1.63882859e-02 2.15701148e-01 -1.31486654e-02 -2.73340851e-01
7.09055513e-02 -1.74824402e-01 -1.93881407e-01 6.83892608e-01
1.94237173e-01 5.84106207e-01 -4.49439049e-01 -6.67543113e-01
-1.30388690e-02 3.97162557e-01 -3.40920985e-01 2.19122842e-01
2.30345339e-01 6.56819418e-02 1.79679226e-02 6.65992200e-01
8.19766879e-01 -3.19715261e-01 7.45411962e-02 -1.50839597e-01
2.23144293e-01 2.61031296e-02 -1.33841348e+00 9.74366546e-01
-1.38329417e-01 6.22633576e-01 -1.09141566e-01 -5.14486849e-01
1.00117683e+00 9.41840634e-02 3.09207320e-01 -5.75398028e-01
1.79160446e-01 1.07885644e-01 -3.59533727e-02 -1.10770509e-01
3.38346660e-01 -5.16740859e-01 -1.29680842e-01 6.29293084e-01
-3.44773442e-01 -1.11757740e-01 1.84196100e-01 -1.85194016e-01
6.34195805e-01 -4.42151636e-01 3.95554781e-01 -2.64223695e-01
1.26393586e-01 -2.55496055e-01 4.84361261e-01 8.49718094e-01
-3.38428825e-01 6.41933203e-01 5.39110541e-01 -3.08793485e-01
-1.22453380e+00 -1.26758265e+00 -8.48958671e-01 7.31228650e-01
1.43688515e-01 -2.37012297e-01 -8.02763104e-01 -5.99991918e-01
7.13432729e-01 5.80647290e-01 -9.34058189e-01 -1.07749060e-01
3.15204300e-02 -5.60194135e-01 6.98841453e-01 6.75394475e-01
-1.64267775e-02 -7.21627355e-01 -4.97659117e-01 -2.57069051e-01
8.85892659e-03 -1.23858738e+00 -7.93871522e-01 -6.12220764e-01
-6.42814100e-01 -1.53547347e+00 -6.43190682e-01 -4.79138285e-01
9.42674160e-01 1.30255252e-01 1.35776258e+00 3.20497781e-01
-2.52254665e-01 6.74125314e-01 -5.10216244e-02 -7.72324920e-01
-2.51716673e-01 -3.91366869e-01 3.60347986e-01 5.61331697e-02
7.23618388e-01 -1.20244242e-01 -5.26479900e-01 6.98049426e-01
-7.88759351e-01 -3.56580675e-01 3.44015568e-01 5.36673188e-01
5.27027130e-01 -8.72519836e-02 6.43283963e-01 -9.87170577e-01
9.77325916e-01 -5.94178855e-01 -7.95767486e-01 4.66833293e-01
-1.04828322e+00 -8.57364386e-02 -8.37210268e-02 -8.37624550e-01
-8.23871434e-01 -1.00219451e-01 3.28802437e-01 -3.80370319e-01
-1.31884903e-01 2.83387750e-01 3.62648159e-01 -1.75532043e-01
7.87927926e-01 -2.77528375e-01 2.58445740e-01 -8.90017115e-03
-1.76677510e-01 7.87019014e-01 4.24020857e-01 -8.45873654e-01
5.39340615e-01 3.62592369e-01 2.45783664e-02 -4.91408527e-01
-9.21549439e-01 -2.48021916e-01 -2.51282066e-01 -3.41613621e-01
4.01687056e-01 -9.02703166e-01 -1.05171132e+00 6.13695905e-02
-6.84493542e-01 -1.12418771e-01 -1.88171580e-01 6.47617579e-01
-5.23430467e-01 2.74489403e-01 -3.67655545e-01 -1.07121146e+00
-2.23966032e-01 -1.04804730e+00 1.13559222e+00 4.24972057e-01
-7.41752803e-01 -7.56559014e-01 -3.69606875e-02 3.36215824e-01
2.77859271e-01 1.47082880e-01 6.38198555e-01 -5.35765529e-01
-7.30272308e-02 -4.14952666e-01 -5.58902442e-01 -1.63332909e-01
2.13640898e-01 2.23516554e-01 -7.63865888e-01 -3.63161683e-01
-3.12081605e-01 -3.54674071e-01 3.12364727e-01 6.60403907e-01
1.46639693e+00 -5.27235568e-01 -4.12324905e-01 2.95644283e-01
1.10509372e+00 6.56780303e-02 4.04450357e-01 -1.46695092e-01
-5.32467756e-03 9.62713063e-01 8.72960091e-01 7.25945830e-01
2.74595171e-01 5.68068862e-01 -1.96908593e-01 1.71279833e-01
7.86501989e-02 -5.21457493e-01 -3.08564641e-02 -2.22803310e-01
2.40000397e-01 -2.67327344e-03 -7.08232164e-01 5.21485984e-01
-1.45427895e+00 -1.03217661e+00 2.44096905e-01 2.83726478e+00
8.45965385e-01 7.60487467e-02 4.86620605e-01 7.23344088e-02
1.06782424e+00 -2.98310608e-01 -6.34522080e-01 -4.36813593e-01
1.46498233e-01 -2.69624796e-02 5.69816291e-01 4.07398760e-01
-7.27446973e-01 3.37392032e-01 8.31912136e+00 5.41766346e-01
-6.56332791e-01 -2.18873575e-01 1.08693767e+00 -2.13625878e-01
-3.28191042e-01 -1.02064386e-01 -1.06737304e+00 7.43256688e-01
1.09885907e+00 -5.53559721e-01 1.15812287e-01 6.52231395e-01
-3.37484367e-02 -3.71958911e-01 -1.56980705e+00 1.13654447e+00
2.27014348e-01 -1.05414701e+00 -1.73989862e-01 1.71067983e-01
6.70783103e-01 -5.98403275e-01 3.76211673e-01 3.75594422e-02
2.68509328e-01 -1.45034862e+00 6.76624298e-01 4.26249653e-01
1.33120596e+00 -7.45171487e-01 6.12920940e-01 3.96329090e-02
-1.03278923e+00 -7.68088698e-02 -3.11527669e-01 -1.23443641e-01
-2.88025178e-02 8.42975259e-01 -8.64568233e-01 -1.05161108e-01
7.91589141e-01 2.09489018e-01 -3.99452358e-01 1.12918651e+00
1.63896769e-01 5.07803082e-01 -5.25763631e-01 4.70754467e-02
-4.57402647e-01 -4.54794988e-02 -1.97762586e-02 1.05769670e+00
5.22921383e-01 2.33437985e-01 -2.20496893e-01 9.72696006e-01
-3.09693515e-01 8.57814401e-02 -8.62269223e-01 6.65593371e-02
1.03319168e+00 8.43089104e-01 -5.92875659e-01 -1.03965685e-01
-5.62485337e-01 4.63331759e-01 2.40169570e-01 1.10212848e-01
-8.84396374e-01 -6.33376613e-02 8.80681157e-01 3.70744407e-01
-6.01183400e-02 8.78723785e-02 -5.68676651e-01 -7.23282695e-01
1.90054923e-01 -8.08943450e-01 7.40931392e-01 -6.61646247e-01
-1.71300542e+00 2.20241219e-01 5.11308849e-01 -1.07184088e+00
-4.22685325e-01 -5.21932006e-01 -2.98614681e-01 8.62881064e-01
-9.24374700e-01 -7.53832579e-01 -1.36258721e-01 4.28346455e-01
1.71127915e-01 4.24683057e-02 7.23905206e-01 1.84263915e-01
-4.73807722e-01 1.37275958e+00 -1.97378829e-01 2.98339903e-01
9.91554618e-01 -6.04504168e-01 4.21094418e-01 6.11035466e-01
2.08189204e-01 8.24069798e-01 8.59481812e-01 -8.30196023e-01
-9.65677440e-01 -6.61217034e-01 7.33004093e-01 -5.35791636e-01
3.49230856e-01 -1.67256758e-01 -6.94867015e-01 7.35922158e-01
-3.87508243e-01 1.12931691e-01 1.18908143e+00 5.30420899e-01
-7.69884408e-01 -1.46678239e-01 -1.87735200e+00 4.62255090e-01
1.05186403e+00 -5.85385799e-01 -2.43541211e-01 3.04226339e-01
2.70570129e-01 -5.80224514e-01 -1.27919269e+00 4.68626797e-01
1.35471487e+00 -9.98754978e-01 9.58637416e-01 -6.61793888e-01
1.85243323e-01 4.65903431e-02 -1.73022807e-01 -9.71810341e-01
-9.59783047e-02 -3.08471382e-01 1.15719020e-01 1.13499069e+00
6.05124891e-01 -6.11006618e-01 8.29652786e-01 1.69662035e+00
7.41473854e-01 -5.19933343e-01 -9.26994503e-01 -7.71557808e-01
-3.10603194e-02 -6.50263250e-01 9.62748349e-01 8.89413834e-01
5.42626269e-02 -5.28431125e-02 -2.77634084e-01 3.16008806e-01
1.04924202e+00 1.17930673e-01 7.33172119e-01 -1.35402513e+00
-8.38124752e-03 -3.62482786e-01 -5.70500493e-01 -4.52364117e-01
3.55634928e-01 -3.07201445e-01 -1.93796158e-01 -8.43364716e-01
6.17696464e-01 -8.25479865e-01 1.04214512e-01 3.28842968e-01
-2.01824233e-01 3.44385386e-01 1.71630979e-01 3.61371301e-02
-1.54702336e-01 1.21240757e-01 6.69900477e-01 4.54125889e-02
1.72816247e-01 3.51369619e-01 -9.50936496e-01 6.46357179e-01
6.57293022e-01 -6.69943750e-01 -3.74684900e-01 -1.88950613e-01
2.10710928e-01 4.60527599e-01 1.95432574e-01 -8.66537869e-01
2.77986139e-01 -4.57646936e-01 8.54744732e-01 -4.14413840e-01
2.52169758e-01 -1.01262164e+00 4.46406841e-01 3.03083986e-01
-6.73415840e-01 4.35002983e-01 4.53337252e-01 6.72885120e-01
4.65865508e-02 -3.12478483e-01 7.79241681e-01 2.29989052e-01
-2.02272147e-01 5.16065657e-01 5.09913862e-02 1.38042107e-01
1.12745535e+00 -3.72221768e-01 -2.46186584e-01 -6.34590924e-01
-4.82206345e-01 2.52055466e-01 5.18443704e-01 4.87623215e-01
7.21850812e-01 -1.56333232e+00 -8.40334713e-01 4.63783085e-01
4.39494550e-01 -4.80301142e-01 7.28267431e-03 5.34122288e-01
2.02685073e-02 1.02765918e-01 -2.70530641e-01 -5.54489255e-01
-1.68195689e+00 5.32216728e-01 2.15003341e-01 2.21772477e-01
4.93542366e-02 7.20967174e-01 1.19570449e-01 -1.77837968e-01
2.80750811e-01 -1.01632141e-01 1.63183242e-01 -6.94869608e-02
8.69901836e-01 3.19745541e-01 -1.51428342e-01 -6.43842399e-01
-5.16962886e-01 5.58218241e-01 -1.65699571e-01 -3.42299640e-01
6.60766304e-01 -1.22617483e-01 9.35737714e-02 2.95468569e-01
9.87358868e-01 9.22077000e-02 -1.07314217e+00 -1.08893201e-01
-1.38442427e-01 -1.00449526e+00 -4.30001438e-01 -6.44793510e-01
-8.63878608e-01 4.68768418e-01 6.96836293e-01 2.28180721e-01
1.01968169e+00 -3.31407264e-02 3.16849947e-01 -1.34156913e-01
5.41467190e-01 -7.79824793e-01 -3.80574763e-01 -1.07804075e-01
7.84526765e-01 -1.63704014e+00 4.31205451e-01 -5.78785777e-01
-5.42841256e-01 7.60687411e-01 6.02306545e-01 3.21177840e-02
9.91900086e-01 4.11886156e-01 -2.07808450e-01 5.00189774e-02
-6.69264376e-01 1.53080776e-01 6.28144860e-01 7.04239964e-01
6.46436989e-01 3.28720331e-01 -5.54136634e-01 5.89655995e-01
-4.26363230e-01 7.90866464e-02 3.80189776e-01 6.50358021e-01
-7.60781616e-02 -9.88299012e-01 -7.31646121e-01 9.40956950e-01
-6.13254786e-01 1.59861490e-01 -3.06408316e-01 7.02872276e-01
-2.03522533e-01 1.16061711e+00 5.88300705e-01 -4.75294739e-01
3.68902504e-01 -6.03570230e-02 6.98291779e-01 -4.45161015e-01
-2.53142178e-01 -6.07125700e-01 2.95684874e-01 -4.92387950e-01
-4.29307640e-01 -9.55069184e-01 -8.71969521e-01 -8.18698287e-01
-4.25356865e-01 1.01950325e-01 6.14101052e-01 8.46826077e-01
5.19939899e-01 -3.86173278e-01 5.47936440e-01 -3.96814764e-01
-6.42530441e-01 -7.48554409e-01 -5.82636356e-01 7.58022785e-01
1.26964346e-01 -9.79486644e-01 -3.73413891e-01 -1.56391174e-01]
|
[13.061836242675781, 1.2022353410720825]
|
fca69d28-cdca-4e43-bec1-6b4a97675c19
|
class-incremental-learning-with-repetition
|
2301.11396
| null |
https://arxiv.org/abs/2301.11396v2
|
https://arxiv.org/pdf/2301.11396v2.pdf
|
Class-Incremental Learning with Repetition
|
Real-world data streams naturally include the repetition of previous concepts. From a Continual Learning (CL) perspective, repetition is a property of the environment and, unlike replay, cannot be controlled by the agent. Nowadays, the Class-Incremental (CI) scenario represents the leading test-bed for assessing and comparing CL strategies. This scenario type is very easy to use, but it never allows revisiting previously seen classes, thus completely neglecting the role of repetition. We focus on the family of Class-Incremental with Repetition (CIR) scenario, where repetition is embedded in the definition of the stream. We propose two stochastic stream generators that produce a wide range of CIR streams starting from a single dataset and a few interpretable control parameters. We conduct the first comprehensive evaluation of repetition in CL by studying the behavior of existing CL strategies under different CIR streams. We then present a novel replay strategy that exploits repetition and counteracts the natural imbalance present in the stream. On both CIFAR100 and TinyImageNet, our strategy outperforms other replay approaches, which are not designed for environments with repetition.
|
['Damian Borth', 'Vincenzo Lomonaco', 'Davide Bacciu', 'Lorenzo Pellegrini', 'Julio Hurtado', 'Antonio Carta', 'Andrea Cossu', 'Hamed Hemati']
|
2023-01-26
| null | null | null | null |
['class-incremental-learning']
|
['computer-vision']
|
[ 1.66118696e-01 -3.33375394e-01 -1.03817925e-01 -7.13382214e-02
-3.11812311e-01 -6.26145899e-01 1.03066576e+00 6.21132135e-01
-6.04723871e-01 8.22484910e-01 4.72827628e-02 -9.78244692e-02
-2.31747240e-01 -8.46056521e-01 -9.06766176e-01 -7.48972714e-01
-7.02557445e-01 4.84563202e-01 4.87680495e-01 -3.87805402e-01
1.05192192e-01 2.79310018e-01 -1.99192691e+00 3.64465415e-01
5.05686402e-01 7.89684474e-01 1.92968637e-01 8.32777083e-01
-1.42221034e-01 8.98222089e-01 -1.26664436e+00 -9.76338685e-02
3.63855779e-01 -8.39722574e-01 -2.71450400e-01 2.92782068e-01
7.28627518e-02 -1.83758095e-01 7.65774101e-02 6.45121992e-01
6.53907895e-01 9.96428579e-02 3.05397958e-01 -1.76912665e+00
8.26612562e-02 1.24421990e+00 -4.33899194e-01 5.78948021e-01
5.73777378e-01 1.96374744e-01 7.62451053e-01 -4.10865515e-01
5.12334585e-01 1.02825296e+00 5.93042791e-01 2.97882587e-01
-1.20646393e+00 -7.25161612e-01 4.95067745e-01 3.13118279e-01
-1.01219869e+00 -2.80929953e-01 7.75937438e-01 -1.76042765e-01
6.50044084e-01 2.34319165e-01 7.94883192e-01 1.28997743e+00
7.77254552e-02 9.10257757e-01 9.12745476e-01 -5.06706417e-01
8.47089469e-01 2.05105871e-01 1.31711021e-01 -8.05179253e-02
3.36910874e-01 2.08056182e-01 -8.98959935e-01 -2.96641350e-01
3.52152765e-01 1.60375144e-02 -4.89852965e-01 -2.87168056e-01
-1.40796256e+00 6.52392209e-01 -6.02503717e-02 1.33346274e-01
-4.77216810e-01 1.54066771e-01 6.18707299e-01 7.41879284e-01
2.17022955e-01 2.28221193e-01 -2.60854691e-01 -4.42562014e-01
-8.18748236e-01 6.58617854e-01 1.00551569e+00 1.15814137e+00
6.67793930e-01 2.29360759e-01 -1.58393100e-01 5.25442004e-01
-2.10802004e-01 4.00683463e-01 9.90417778e-01 -6.56152844e-01
4.00672048e-01 2.21635640e-01 1.65811330e-01 -5.32897592e-01
-4.77744132e-01 -6.85322285e-01 -7.15571523e-01 9.54165533e-02
6.42676234e-01 -3.30270380e-01 -3.02374870e-01 1.77741373e+00
3.30153823e-01 8.64388585e-01 2.14750096e-01 5.28989315e-01
4.43695813e-01 5.40313184e-01 -3.43274698e-02 -6.94768965e-01
8.91012311e-01 -6.20590270e-01 -5.99359155e-01 5.85361533e-02
4.45273519e-01 -4.25047457e-01 1.08328021e+00 8.39552939e-01
-9.37349439e-01 -4.50435817e-01 -1.22796249e+00 1.02732909e+00
-1.71352282e-01 -4.53682244e-01 4.23510700e-01 8.61691415e-01
-7.92350471e-01 5.49948573e-01 -6.42481089e-01 -2.13909954e-01
1.83787107e-01 -1.72658846e-01 1.00239813e-01 1.08247645e-01
-1.18954647e+00 3.48907053e-01 3.83054703e-01 -2.55571008e-01
-1.10318303e+00 -9.29010093e-01 -5.70692718e-01 2.68292669e-02
6.17767453e-01 -3.77715021e-01 1.52056611e+00 -1.19757640e+00
-1.52468407e+00 1.14853881e-01 -5.92811778e-02 -1.07807696e+00
1.06448293e+00 -2.57706821e-01 -2.53998518e-01 1.11200981e-01
-2.44417280e-01 2.83743024e-01 1.09964764e+00 -1.44460440e+00
-7.44932234e-01 7.34340996e-02 1.49184763e-01 3.04108351e-01
-9.74393785e-02 -2.24027365e-01 1.17581934e-01 -6.70194447e-01
-2.98483312e-01 -7.14806318e-01 -2.17509896e-01 -3.77808213e-01
-2.20885560e-01 -2.11785585e-01 7.23833740e-01 4.48428184e-01
1.25731051e+00 -2.06246901e+00 -3.04171175e-01 9.42392042e-04
-1.95425406e-01 1.84576228e-01 -1.69006258e-01 7.17786849e-01
-1.58279434e-01 2.31737699e-02 -2.17216179e-01 -5.51661730e-01
-1.53061539e-01 2.37356260e-01 -8.52594554e-01 3.79745901e-01
-6.77629709e-02 5.38384974e-01 -1.22043657e+00 1.35808047e-02
4.48695086e-02 5.83311021e-02 -6.42857313e-01 2.51565188e-01
-5.85144579e-01 1.31966874e-01 -1.98954903e-03 2.19398603e-01
6.65458381e-01 -2.18069833e-03 5.54528385e-02 3.61508638e-01
-1.58552766e-01 -8.16297084e-02 -1.53112328e+00 1.34742868e+00
-3.88342470e-01 4.64722723e-01 -4.01484698e-01 -9.19234753e-01
9.62134600e-01 2.61583120e-01 5.91095924e-01 -7.51246154e-01
-7.29138628e-02 -8.96649994e-03 9.56732258e-02 -5.01296520e-01
4.87175137e-01 -1.82268113e-01 -2.55296975e-02 1.07481027e+00
-1.21398918e-01 1.72953308e-02 5.92658043e-01 2.74608105e-01
1.34119844e+00 -8.07916746e-02 3.87408644e-01 -1.03719942e-01
4.73764420e-01 -2.84596860e-01 5.84431231e-01 1.43274772e+00
-2.05949873e-01 7.60735571e-01 7.35650241e-01 -5.97149491e-01
-8.84340882e-01 -1.09903622e+00 4.73185852e-02 9.83273745e-01
4.46754217e-01 -3.84595096e-01 -4.69223857e-01 -5.64766467e-01
-1.11302286e-02 9.24365222e-01 -5.67379653e-01 -1.50938183e-01
-5.75680256e-01 -1.25526977e+00 6.28354013e-01 1.04828373e-01
6.25099421e-01 -1.35339463e+00 -1.31156468e+00 4.74376082e-01
-1.66876957e-01 -1.17820418e+00 -1.16057195e-01 2.09673643e-01
-6.84151649e-01 -1.09456563e+00 -4.34868157e-01 -1.44016057e-01
1.54556453e-01 3.67591798e-01 1.25264442e+00 1.07239164e-01
-6.69626668e-02 6.30088091e-01 -7.90511608e-01 -1.03941917e+00
-7.04403281e-01 6.47849664e-02 1.07773133e-01 3.19936723e-01
9.26740691e-02 -6.97734773e-01 -4.01814461e-01 3.78971249e-01
-1.27281940e+00 -2.05945253e-01 2.76249796e-01 6.86068356e-01
3.60681951e-01 3.76738578e-01 1.27133000e+00 -9.94882047e-01
8.74998093e-01 -9.19942617e-01 -3.77785951e-01 2.49797210e-01
-4.39525843e-01 -1.15556493e-01 9.75828409e-01 -9.14762795e-01
-9.17429090e-01 -1.52442351e-01 1.46307960e-01 -4.06963527e-01
-3.89656126e-01 1.86320096e-01 -7.00677037e-02 5.93542099e-01
9.98852670e-01 5.43024242e-01 5.11370562e-02 -1.52916312e-01
1.86819360e-01 4.02012020e-01 4.51788515e-01 -6.69686973e-01
5.65965891e-01 6.96307957e-01 -2.50550449e-01 -1.17492354e+00
-5.99503160e-01 -3.11437696e-01 -3.29185992e-01 -4.64263201e-01
7.92432576e-02 -7.86532104e-01 -6.39791489e-01 8.94610465e-01
-1.10984790e+00 -6.63295925e-01 -8.46465886e-01 4.08583790e-01
-8.09646428e-01 3.12629253e-01 -2.08606705e-01 -8.59923780e-01
-1.06216267e-01 -9.42828834e-01 6.88089907e-01 1.61885813e-01
-6.84516206e-02 -8.50693941e-01 3.14599097e-01 -4.79334891e-01
5.29219151e-01 2.52151966e-01 6.11616731e-01 -9.49040473e-01
-5.25838017e-01 -4.19946201e-02 2.49094248e-01 1.53254837e-01
-9.31558548e-04 -3.45347263e-02 -9.93501425e-01 -3.48422527e-01
1.12925395e-01 -2.93104708e-01 6.93858206e-01 1.93321079e-01
1.02519369e+00 -3.49425524e-01 1.17400423e-01 3.05189997e-01
1.43544197e+00 4.43466604e-01 6.34807885e-01 5.94210267e-01
2.68132705e-02 4.07097518e-01 3.94567579e-01 1.14832413e+00
3.28052372e-01 4.37637478e-01 5.97841024e-01 3.17917913e-01
5.30799516e-02 -1.95230991e-01 5.25555134e-01 3.01004112e-01
1.32388204e-01 -7.55370378e-01 -7.04503655e-01 4.12529379e-01
-1.80692780e+00 -1.31074905e+00 -7.99923483e-03 2.57296824e+00
7.38360763e-01 4.84079421e-01 4.27005678e-01 6.32493198e-01
6.75644696e-01 1.83538161e-02 -5.41515946e-01 -1.96456298e-01
-3.24878544e-01 -3.10094538e-03 1.71180695e-01 3.03184092e-01
-9.53305483e-01 5.37822366e-01 6.73319864e+00 6.18702829e-01
-1.15191007e+00 -1.04227357e-01 3.05785656e-01 -3.54361653e-01
-3.47102851e-01 -1.43090025e-01 -7.03358412e-01 7.36505449e-01
1.14132071e+00 -7.36083031e-01 2.80333191e-01 6.40395105e-01
3.73348594e-01 -4.86927420e-01 -1.39456069e+00 9.73756313e-01
2.90355146e-01 -1.24333560e+00 2.81734437e-01 -3.49167138e-01
5.46880960e-01 6.03574291e-02 -5.56861721e-02 3.49889219e-01
2.09416971e-01 -5.97974479e-01 1.04813004e+00 4.69451427e-01
3.43838543e-01 -8.60035598e-01 6.76698446e-01 7.10918427e-01
-8.73896956e-01 -3.47677916e-01 -1.79612339e-01 -1.19118981e-01
2.50968516e-01 7.95501232e-01 -7.97596097e-01 5.47858596e-01
6.65166974e-01 6.06305540e-01 -6.61774457e-01 1.38789332e+00
1.43668735e-02 8.93516779e-01 -5.11374116e-01 -8.32982510e-02
-1.63721666e-02 1.48755893e-01 1.01021039e+00 1.30960572e+00
3.28862667e-01 -2.31198341e-01 1.74511969e-01 5.44321120e-01
8.58202651e-02 3.86715829e-02 -6.06498539e-01 4.79649961e-01
8.02653015e-01 7.69906163e-01 -8.96269262e-01 -5.37731588e-01
-2.85216630e-01 5.14323294e-01 -9.12707075e-02 4.06820983e-01
-8.39952588e-01 -2.73989320e-01 4.97721821e-01 2.43361011e-01
2.02456966e-01 -1.17336959e-01 -8.41797441e-02 -1.19644201e+00
7.95839950e-02 -1.05746651e+00 4.30957317e-01 -4.08418328e-01
-1.30378306e+00 7.99165070e-01 4.83339608e-01 -1.58202827e+00
-6.53641880e-01 1.38835117e-01 -8.99785936e-01 1.99439019e-01
-1.62195969e+00 -5.11438370e-01 -7.01326132e-01 7.14860618e-01
8.80039573e-01 -3.39795083e-01 5.10000706e-01 1.94495082e-01
-5.04044414e-01 6.05191171e-01 -3.13035063e-02 -3.01646650e-01
5.86300731e-01 -1.11615896e+00 2.52763838e-01 8.34886491e-01
3.07121873e-01 3.02265078e-01 1.10433114e+00 -3.71253312e-01
-1.11420810e+00 -1.12740958e+00 3.90011132e-01 -1.94649816e-01
6.88221991e-01 -4.72198278e-01 -1.09889448e+00 4.40653592e-01
2.11612195e-01 -9.15713981e-03 4.52062964e-01 -2.89328694e-01
-3.15179408e-01 -3.39091182e-01 -8.97202849e-01 4.09006864e-01
9.30517733e-01 9.99651849e-02 -3.54438365e-01 1.26585647e-01
7.66541362e-01 -1.74689963e-01 -2.90120810e-01 2.48172626e-01
4.10943478e-01 -1.48487651e+00 7.13357627e-01 -3.80559176e-01
2.98141181e-01 -3.23495090e-01 7.61580130e-04 -1.50435698e+00
2.92737275e-01 -8.97404194e-01 -1.20283134e-01 1.20184517e+00
2.44843572e-01 -7.40281343e-01 6.20324612e-01 2.38963254e-02
9.08713713e-02 -5.06753981e-01 -9.11907971e-01 -1.19517982e+00
-2.43217070e-02 -9.10199046e-01 9.01484907e-01 5.66268027e-01
-3.94654982e-02 4.59353290e-02 -3.44788015e-01 -8.13659057e-02
8.59813094e-01 1.61768869e-01 1.08283305e+00 -1.03857279e+00
-6.14847720e-01 -4.34013456e-01 -4.35618609e-01 -8.77930045e-01
-1.03115104e-01 -6.10490024e-01 8.01362917e-02 -7.92999864e-01
-5.64521290e-02 -6.92918003e-01 -2.10141599e-01 6.91328049e-02
-4.38633673e-02 -5.58587685e-02 4.76592958e-01 3.87718409e-01
-8.07182372e-01 7.73958802e-01 7.50818253e-01 1.43046230e-01
-6.11769915e-01 4.95972931e-01 -4.77052361e-01 5.36305845e-01
9.34020042e-01 -4.97503251e-01 -8.97040606e-01 -4.20915633e-02
3.57227683e-01 9.77277569e-03 3.48047614e-01 -1.29038525e+00
3.52536976e-01 -1.11088663e-01 -5.45800552e-02 -5.36848366e-01
-1.68938264e-01 -5.76394379e-01 2.23285586e-01 3.17962170e-01
-6.40413105e-01 3.01857591e-01 1.76729709e-01 8.29241455e-01
-2.00244322e-01 -3.69624376e-01 6.73210919e-01 -1.92695796e-01
-4.77502316e-01 1.94918483e-01 -6.09505653e-01 4.83986825e-01
1.28408027e+00 -2.73463838e-02 -4.50555176e-01 -6.10029280e-01
-5.51694214e-01 4.01673496e-01 1.95146725e-01 5.56583226e-01
6.66103125e-01 -1.02087867e+00 -1.00119102e+00 3.36999714e-01
8.79580826e-02 1.33660063e-01 2.23543286e-01 5.25239527e-01
-3.40526998e-01 -1.50565162e-01 -1.49393320e-01 -8.47388983e-01
-9.60583985e-01 7.58431673e-01 3.95309418e-01 -1.85025543e-01
-9.46340442e-01 3.58331263e-01 -4.28347886e-02 5.50426990e-02
5.04968882e-01 -1.57541290e-01 -4.09540027e-01 5.06671250e-01
9.86644089e-01 3.86852622e-01 2.66472667e-01 -1.11792684e-01
-2.66469810e-02 -1.03009725e-02 1.53432069e-02 -3.78675222e-01
1.30402541e+00 -2.73767978e-01 2.85087198e-01 9.84903872e-01
7.33494401e-01 -2.14301065e-01 -1.29832780e+00 -3.57638508e-01
1.25766873e-01 -3.95577282e-01 -5.14285386e-01 -4.77246940e-01
-8.65473211e-01 5.73441029e-01 4.35901135e-01 6.12711906e-01
1.22096634e+00 -1.94057018e-01 4.74038035e-01 3.96719724e-01
6.32141888e-01 -9.00739968e-01 3.54174584e-01 3.94434333e-01
9.63221192e-01 -8.46621692e-01 1.00290030e-02 -3.25330347e-02
-8.18548977e-01 1.08301592e+00 4.71442789e-01 -2.23797575e-01
8.52060556e-01 6.72406793e-01 1.18365720e-01 2.86837161e-01
-1.36051202e+00 -2.56673187e-01 -5.75952768e-01 6.54250264e-01
-1.07456939e-02 -1.67888969e-01 -2.49632344e-01 4.93763059e-01
-4.15236264e-01 8.91188681e-02 1.29625964e+00 9.92412865e-01
-3.75801533e-01 -1.00291133e+00 -4.14226443e-01 3.11111391e-01
-1.83104530e-01 1.70592979e-01 1.13833584e-02 7.75988400e-01
1.74053654e-01 1.01235521e+00 3.73743504e-01 -2.19372094e-01
5.03958941e-01 6.29367307e-02 2.13222474e-01 -3.32040250e-01
-6.67485833e-01 4.13653590e-02 -1.49857804e-01 -4.52158868e-01
-6.06096089e-01 -1.04691112e+00 -1.10788131e+00 -1.29286662e-01
-5.50103672e-02 2.11998001e-01 4.36126024e-01 9.28196430e-01
9.59530622e-02 5.23914337e-01 9.90010202e-01 -7.99038768e-01
-8.22683454e-01 -7.56803334e-01 -6.41500771e-01 6.16603315e-01
5.80330968e-01 -6.33903146e-01 -6.57441676e-01 1.91957176e-01]
|
[7.6926727294921875, 3.060680866241455]
|
53520bb8-7688-491b-b04c-c78b102208b7
|
re-attention-transformer-for-weakly
|
2208.01838
| null |
https://arxiv.org/abs/2208.01838v2
|
https://arxiv.org/pdf/2208.01838v2.pdf
|
Re-Attention Transformer for Weakly Supervised Object Localization
|
Weakly supervised object localization is a challenging task which aims to localize objects with coarse annotations such as image categories. Existing deep network approaches are mainly based on class activation map, which focuses on highlighting discriminative local region while ignoring the full object. In addition, the emerging transformer-based techniques constantly put a lot of emphasis on the backdrop that impedes the ability to identify complete objects. To address these issues, we present a re-attention mechanism termed token refinement transformer (TRT) that captures the object-level semantics to guide the localization well. Specifically, TRT introduces a novel module named token priority scoring module (TPSM) to suppress the effects of background noise while focusing on the target object. Then, we incorporate the class activation map as the semantically aware input to restrain the attention map to the target object. Extensive experiments on two benchmarks showcase the superiority of our proposed method against existing methods with image category annotations. Source code is available in \url{https://github.com/su-hui-zz/ReAttentionTransformer}.
|
['Lechao Cheng', 'Mingli Song', 'Zhiwei Chen', 'Yue Ye', 'Hui Su']
|
2022-08-03
| null | null | null | null |
['weakly-supervised-object-localization']
|
['computer-vision']
|
[ 9.54810679e-02 1.40208021e-01 -2.23519549e-01 -4.49148417e-01
-6.87102437e-01 -2.73365378e-01 5.79137564e-01 2.06617385e-01
-3.74099910e-01 3.94173265e-01 3.27140540e-01 6.22303039e-02
8.41563642e-02 -5.09919286e-01 -7.09727407e-01 -7.42183208e-01
2.23226994e-01 -2.36189887e-02 6.22848630e-01 9.20293108e-02
2.58839697e-01 2.29260489e-01 -1.32654321e+00 4.05617476e-01
7.81865895e-01 1.11334300e+00 6.36876285e-01 -3.88927981e-02
-3.08991373e-01 9.89913523e-01 -4.73577410e-01 -1.84352353e-01
2.28051752e-01 -1.67049646e-01 -8.07184458e-01 -6.35861903e-02
4.17832404e-01 -2.68124491e-01 -3.77823859e-01 1.43915784e+00
3.67291540e-01 1.60078853e-01 3.30053359e-01 -1.28571391e+00
-8.11696410e-01 7.83068717e-01 -8.43121409e-01 4.96408522e-01
-2.23914366e-02 1.45136014e-01 1.01706684e+00 -1.26955128e+00
3.27928156e-01 1.25681484e+00 3.55917990e-01 4.30775166e-01
-1.01394987e+00 -7.34281063e-01 7.25869477e-01 4.12519306e-01
-1.72472298e+00 -3.97746772e-01 9.70806658e-01 -3.21719855e-01
4.62359756e-01 1.01357080e-01 2.66459972e-01 8.42424691e-01
-8.56638476e-02 1.11315274e+00 1.03876483e+00 -2.75627762e-01
5.02599441e-02 1.68728173e-01 2.61875749e-01 6.74054861e-01
1.57177389e-01 -1.90173030e-01 -4.18139189e-01 4.64759134e-02
6.54889107e-01 3.88065606e-01 -2.97551811e-01 -4.02802825e-01
-1.43907154e+00 5.77364743e-01 1.08000064e+00 3.85767907e-01
-4.34360862e-01 2.29207769e-01 4.74862784e-01 -2.84430444e-01
5.46593249e-01 1.66600689e-01 -4.13787633e-01 2.89263874e-01
-8.58709157e-01 6.50590509e-02 1.47202402e-01 1.03664804e+00
9.22862232e-01 -2.74888426e-02 -7.30218649e-01 8.55881095e-01
2.91663349e-01 7.79293403e-02 4.89935279e-01 -6.42688155e-01
3.16003799e-01 8.92146885e-01 5.14059775e-02 -8.56090903e-01
-2.10198924e-01 -8.21876526e-01 -7.44696200e-01 7.06224516e-02
1.19470477e-01 3.13216597e-01 -1.22067583e+00 1.74043715e+00
4.82809335e-01 3.39148641e-01 -3.10629398e-01 1.25216830e+00
1.02625132e+00 4.65371013e-01 4.89389360e-01 2.38918588e-01
1.43665421e+00 -1.36424696e+00 -5.44788241e-01 -2.93659717e-01
4.99954432e-01 -6.92415953e-01 1.34935343e+00 8.36203694e-02
-8.40522051e-01 -6.53343022e-01 -8.13213944e-01 -2.05345050e-01
-4.36899900e-01 3.48453104e-01 4.77161348e-01 9.79045555e-02
-9.80838954e-01 2.91840702e-01 -8.79566610e-01 -3.91185522e-01
9.69713390e-01 3.00530791e-01 -3.02174091e-01 -2.32146561e-01
-9.67037678e-01 7.18988895e-01 6.32211149e-01 3.36738735e-01
-1.21154356e+00 -6.60903931e-01 -7.34581113e-01 3.04480165e-01
6.18851244e-01 -3.80829424e-01 1.16412807e+00 -1.20107746e+00
-1.05783701e+00 7.36222148e-01 -2.94551849e-01 -2.89321929e-01
3.69989783e-01 -3.58221263e-01 -9.61229876e-02 9.34619531e-02
5.09034634e-01 9.05199587e-01 7.45839775e-01 -1.40950179e+00
-8.88090849e-01 -2.97565252e-01 1.70141011e-01 3.04405689e-01
-4.96756852e-01 1.20773852e-01 -8.09363306e-01 -8.40405285e-01
1.90257892e-01 -5.55083394e-01 -2.40563750e-01 -5.00335731e-02
-7.27262199e-01 -5.18653810e-01 1.08208537e+00 -4.35143411e-01
1.12108231e+00 -2.22038460e+00 1.47653744e-02 -7.89365396e-02
5.07341862e-01 1.93466812e-01 -3.16661566e-01 7.44145364e-02
-2.10022628e-01 1.19905122e-01 -1.00657515e-01 -3.98416132e-01
-3.05522773e-02 -4.32463773e-02 -2.88697600e-01 4.43972498e-01
3.62186670e-01 1.09966087e+00 -9.54828501e-01 -6.21341646e-01
2.89871305e-01 4.36585665e-01 -4.05149370e-01 1.73221052e-01
-1.78808242e-01 4.33996409e-01 -6.26128852e-01 9.70739067e-01
6.80668235e-01 -4.93749231e-01 -3.15017283e-01 -5.51400185e-01
-2.26433709e-01 2.62602866e-01 -8.95494759e-01 1.76832688e+00
-1.22037366e-01 2.66779751e-01 1.15368441e-01 -1.05407691e+00
7.46866703e-01 5.47886034e-03 5.09791791e-01 -7.65199423e-01
3.08968604e-01 4.15370949e-02 -1.17769418e-02 -2.78935671e-01
3.23078960e-01 1.48699552e-01 -1.12937633e-02 8.78096297e-02
6.24439828e-02 5.86798072e-01 1.10312481e-03 3.92542601e-01
9.70619142e-01 3.52980584e-01 2.04899356e-01 -4.89152342e-01
6.14895642e-01 4.70135398e-02 8.30014229e-01 6.95516944e-01
-4.65689927e-01 6.51744723e-01 2.92821050e-01 -3.64779234e-01
-8.57888520e-01 -1.04034710e+00 8.71277321e-03 1.39400446e+00
5.95114529e-01 -3.94557148e-01 -7.47974992e-01 -9.51992154e-01
-5.04379869e-02 4.54763591e-01 -8.87410939e-01 -2.72554696e-01
-4.43766773e-01 -5.88563800e-01 1.93363786e-01 6.37829423e-01
7.63025880e-01 -1.13269651e+00 -4.33350593e-01 -2.16225088e-02
-3.21888149e-01 -9.35770988e-01 -6.59804940e-01 2.24385485e-01
-5.93095243e-01 -8.34329903e-01 -9.23904002e-01 -9.66679931e-01
9.78245676e-01 5.41081488e-01 8.66547406e-01 1.53407335e-01
-2.12783560e-01 1.92797795e-01 -4.60307419e-01 -4.06205326e-01
2.46816114e-01 3.20922583e-01 -6.83378503e-02 2.41762385e-01
5.44639587e-01 -3.81810635e-01 -1.00473928e+00 4.57916319e-01
-7.94286430e-01 2.71779239e-01 8.25002372e-01 7.58373082e-01
8.64067972e-01 1.86325423e-03 5.28379679e-01 -7.72097528e-01
1.70136333e-01 -6.33326292e-01 -4.95876402e-01 2.60889620e-01
-2.96543270e-01 -4.09276001e-02 4.19754118e-01 -5.33571124e-01
-1.07683647e+00 2.17044130e-01 3.11457850e-02 -6.79896116e-01
-2.02679515e-01 3.33719522e-01 -4.25841302e-01 -3.81876603e-02
2.81345814e-01 3.70606452e-01 -4.84812021e-01 -7.32186198e-01
3.02292556e-01 4.29960907e-01 5.69848418e-01 -4.84665126e-01
7.91471660e-01 6.83725953e-01 -3.96333039e-01 -3.67764592e-01
-1.37645710e+00 -5.98198652e-01 -6.32544518e-01 -2.25158870e-01
8.08535099e-01 -1.00532794e+00 -4.90422070e-01 2.14003056e-01
-1.08137774e+00 -3.36016417e-01 -3.51772904e-01 3.31953883e-01
-1.74743995e-01 2.14822531e-01 -3.54252011e-01 -5.86099029e-01
-3.11727822e-01 -1.09533238e+00 1.26029968e+00 4.48620319e-01
1.82784483e-01 -5.98725140e-01 -3.85928333e-01 2.58904666e-01
5.03586113e-01 1.17470220e-01 7.54139960e-01 -7.39503324e-01
-9.09516633e-01 -1.62476793e-01 -7.50085056e-01 2.93220818e-01
2.57552236e-01 -4.48857754e-01 -1.09406066e+00 -2.55379379e-01
-1.83437839e-01 -1.73515663e-01 1.17247689e+00 2.74158746e-01
1.54071045e+00 -3.66661131e-01 -6.22791469e-01 5.47691882e-01
1.44466650e+00 -8.48327652e-02 3.68184268e-01 3.56272519e-01
1.16602778e+00 4.91711348e-01 7.61560500e-01 2.52986282e-01
4.14059579e-01 6.81699872e-01 7.32496738e-01 -4.80283529e-01
-2.94245213e-01 -4.13522035e-01 8.16113129e-02 3.96084309e-01
6.56604841e-02 -2.23297238e-01 -8.83773923e-01 7.61371791e-01
-1.98100841e+00 -7.00342119e-01 9.51571912e-02 2.07694316e+00
6.37762189e-01 2.80013144e-01 -8.59019347e-03 -1.46359265e-01
1.05861485e+00 1.48068577e-01 -6.04159713e-01 2.67381698e-01
7.47047588e-02 -2.04080403e-01 4.89727169e-01 2.94313610e-01
-1.33247960e+00 1.14939868e+00 4.77082586e+00 1.10849655e+00
-1.23172498e+00 5.02687931e-01 6.45924807e-01 -4.14337516e-02
-1.03632189e-01 -1.04862535e-02 -9.15365458e-01 6.44546926e-01
3.21503729e-01 -7.23576471e-02 4.99677882e-02 1.04085660e+00
3.62282157e-01 -1.36104301e-01 -8.82166624e-01 8.09571862e-01
6.71148896e-02 -1.02076566e+00 1.76058993e-01 -1.34214759e-01
5.75998425e-01 1.71858743e-01 1.27254874e-01 4.65847224e-01
5.94955757e-02 -8.12144876e-01 1.03715003e+00 4.87770170e-01
6.72576487e-01 -6.28972113e-01 7.63245165e-01 2.08825365e-01
-1.51809835e+00 -2.04450086e-01 -3.93630624e-01 1.45292625e-01
-1.20944455e-01 6.07732654e-01 -8.28073978e-01 3.13976645e-01
1.03560543e+00 7.60003805e-01 -7.94873476e-01 1.29495645e+00
-2.86737233e-01 6.40079439e-01 -1.58404976e-01 1.62100747e-01
3.87574226e-01 2.33801275e-01 5.60616434e-01 1.15921748e+00
5.81052490e-02 5.51327802e-02 4.39282298e-01 1.11034083e+00
-1.39209792e-01 1.02917790e-01 -2.63674051e-01 2.15666100e-01
4.28784192e-01 1.54495692e+00 -1.16585219e+00 -4.02471155e-01
-4.56159681e-01 1.06504726e+00 5.04171431e-01 4.35563684e-01
-9.06024754e-01 -4.64479715e-01 4.24521089e-01 3.56762499e-01
4.91114855e-01 -1.02061220e-01 -2.19270006e-01 -1.00818491e+00
1.42592877e-01 -4.77821499e-01 3.90476733e-01 -8.69564831e-01
-1.08246076e+00 6.80760860e-01 7.49636767e-03 -1.13772357e+00
5.20003140e-01 -3.78504574e-01 -7.09501207e-01 9.18505430e-01
-1.69864225e+00 -1.41526210e+00 -6.13049150e-01 5.29047370e-01
8.11735809e-01 1.41226172e-01 3.42339277e-01 5.23801744e-01
-7.03940988e-01 5.65943003e-01 -2.01529250e-01 2.45349377e-01
6.72596872e-01 -1.15230381e+00 1.81160048e-02 9.81472969e-01
-1.37976468e-01 8.10754478e-01 4.54516053e-01 -6.69965208e-01
-8.59657884e-01 -1.49633169e+00 6.61824167e-01 -4.63005155e-01
4.68602538e-01 -5.85846066e-01 -1.09089744e+00 6.58924282e-01
1.03856474e-01 4.25302476e-01 1.96196422e-01 -7.26920590e-02
-3.82528096e-01 -2.68332094e-01 -9.47532654e-01 4.68264639e-01
9.97020602e-01 -3.10500503e-01 -5.53241372e-01 3.83150905e-01
9.49736714e-01 -3.02088231e-01 -4.24571157e-01 6.35090053e-01
2.53471971e-01 -7.73649335e-01 9.56207573e-01 -3.31712574e-01
3.06328654e-01 -7.81226575e-01 9.29360650e-03 -8.34533274e-01
-5.36533892e-01 -2.19798684e-01 -4.05945331e-02 1.48931801e+00
1.64727181e-01 -4.14068907e-01 7.58623838e-01 3.23395818e-01
-3.37558538e-01 -9.69520032e-01 -7.44218588e-01 -6.21320844e-01
-2.40165636e-01 -1.56484812e-01 5.92589617e-01 8.60005260e-01
-2.65402406e-01 2.93310255e-01 -2.31632337e-01 3.59339625e-01
7.04570293e-01 1.11352749e-01 4.29190695e-01 -9.38553989e-01
1.41156420e-01 -5.07608354e-01 -4.96737123e-01 -1.03183699e+00
2.94529442e-02 -1.05095518e+00 3.41717631e-01 -1.67169178e+00
5.35404682e-01 -5.60799956e-01 -9.24380779e-01 9.43728149e-01
-3.62634033e-01 5.50269246e-01 1.47333801e-01 3.82755369e-01
-1.11860585e+00 7.46704876e-01 1.13870251e+00 -2.88886279e-01
4.67377566e-02 -4.18814458e-02 -9.77358997e-01 8.14589918e-01
8.41463685e-01 -5.97556353e-01 -3.21175218e-01 -5.96013367e-01
-2.47613043e-01 -4.41389531e-01 8.03628147e-01 -9.52209592e-01
3.73818696e-01 -9.19842348e-02 4.04126108e-01 -8.24942768e-01
1.74973071e-01 -7.56822467e-01 -3.28103542e-01 3.42761844e-01
-4.80750352e-01 -1.19361402e-02 2.01487258e-01 5.52706301e-01
-3.13256741e-01 1.74820758e-02 9.13393974e-01 -2.12560728e-01
-1.05676472e+00 5.49052477e-01 -7.49673173e-02 -9.58209932e-02
9.80380416e-01 -1.24054857e-01 -2.56420523e-01 7.95500651e-02
-5.65973461e-01 5.04271865e-01 3.22390884e-01 6.60262764e-01
6.70216382e-01 -1.44248867e+00 -5.71026444e-01 2.25290637e-02
3.76446754e-01 1.88342810e-01 4.65311229e-01 9.91590381e-01
-1.25388369e-01 6.05956733e-01 -3.07307113e-02 -6.60858452e-01
-1.19908333e+00 8.11913252e-01 4.18297619e-01 1.05723910e-01
-7.03296661e-01 1.11987102e+00 1.00768888e+00 -1.48381040e-01
4.62608188e-01 -3.37190449e-01 -2.88407475e-01 -2.07820922e-01
5.03146529e-01 1.23747200e-01 -2.26872247e-02 -7.17585564e-01
-6.92168236e-01 3.48940134e-01 -4.35425282e-01 2.40037903e-01
1.19811857e+00 -3.35669816e-01 -1.32113263e-01 2.20819980e-01
1.10100913e+00 -1.97713777e-01 -1.38496077e+00 -6.32703245e-01
8.46191645e-02 -5.07787406e-01 2.56702125e-01 -8.04505110e-01
-1.26562440e+00 8.02374840e-01 8.10280204e-01 -1.13600053e-01
1.19444692e+00 2.48177230e-01 5.31399608e-01 1.17029190e-01
1.95780337e-01 -8.32907915e-01 2.59450674e-01 3.50273818e-01
8.85710657e-01 -1.35723615e+00 -1.59913346e-01 -5.06479144e-01
-5.87732375e-01 5.57540417e-01 1.11667204e+00 -2.44941220e-01
6.90310955e-01 1.16865784e-01 -3.28787751e-02 -2.18371004e-01
-4.62318629e-01 -4.67294395e-01 3.35873842e-01 4.18472707e-01
3.29796731e-01 -1.55345872e-01 -1.84102044e-01 8.21739733e-01
2.98062205e-01 -1.06728546e-01 1.02512993e-01 9.20701385e-01
-6.37126744e-01 -8.17047417e-01 -3.60844344e-01 4.04595345e-01
-4.99248803e-01 -3.33587170e-01 -2.51145661e-01 6.27059937e-01
4.86192018e-01 6.00064218e-01 -2.50952803e-02 -2.18399391e-01
2.60948151e-01 -1.82801142e-01 8.02162886e-02 -8.93081665e-01
-3.97628427e-01 2.67818511e-01 -4.97883111e-01 -6.56816125e-01
-4.28489268e-01 -4.99155849e-01 -1.29728341e+00 1.77554324e-01
-4.55995411e-01 2.41799220e-01 3.40807825e-01 7.55252063e-01
4.54857349e-01 7.60866702e-01 3.90029460e-01 -1.02926099e+00
-1.03319220e-01 -9.98367012e-01 -4.54619974e-01 1.82231039e-01
4.87924606e-01 -9.01396334e-01 -1.90433711e-01 4.28029336e-02]
|
[9.594327926635742, 0.8424345254898071]
|
f1d2c376-18f0-4390-8c11-eb684b7ebcfa
|
3d-hand-pose-detection-in-egocentric-rgb-d
|
1412.0065
| null |
http://arxiv.org/abs/1412.0065v1
|
http://arxiv.org/pdf/1412.0065v1.pdf
|
3D Hand Pose Detection in Egocentric RGB-D Images
|
We focus on the task of everyday hand pose estimation from egocentric
viewpoints. For this task, we show that depth sensors are particularly
informative for extracting near-field interactions of the camera wearer with
his/her environment. Despite the recent advances in full-body pose estimation
using Kinect-like sensors, reliable monocular hand pose estimation in RGB-D
images is still an unsolved problem. The problem is considerably exacerbated
when analyzing hands performing daily activities from a first-person viewpoint,
due to severe occlusions arising from object manipulations and a limited
field-of-view. Our system addresses these difficulties by exploiting strong
priors over viewpoint and pose in a discriminative tracking-by-detection
framework. Our priors are operationalized through a photorealistic synthetic
model of egocentric scenes, which is used to generate training data for
learning depth-based pose classifiers. We evaluate our approach on an annotated
dataset of real egocentric object manipulation scenes and compare to both
commercial and academic approaches. Our method provides state-of-the-art
performance for both hand detection and pose estimation in egocentric RGB-D
images.
|
['Maryam Khademi', 'Deva Ramanan', 'Jose Maria Martinez Montiel', 'James S. Supancic III', 'Gregory Rogez']
|
2014-11-29
| null | null | null | null |
['hand-detection']
|
['computer-vision']
|
[ 8.76614824e-02 -3.24328035e-01 -8.78574103e-02 -8.98186788e-02
-5.50804079e-01 -5.31248808e-01 3.58888984e-01 -5.69967031e-01
-4.66215938e-01 5.52334189e-01 3.39819074e-01 3.67691755e-01
-7.77210519e-02 -1.49255648e-01 -5.65280139e-01 -4.39684063e-01
1.42549545e-01 7.85546601e-01 1.64323598e-01 -1.21704705e-01
2.84346253e-01 9.54936266e-01 -1.69030678e+00 4.67356257e-02
9.70297903e-02 6.60055995e-01 4.29939121e-01 1.01263320e+00
5.96166253e-01 5.58401704e-01 -6.41577482e-01 -2.28852868e-01
4.31809783e-01 -7.61268511e-02 -5.23399770e-01 5.35011590e-01
8.10416162e-01 -9.37288880e-01 -7.62138069e-01 7.11404085e-01
1.09098172e+00 -5.49350306e-02 5.14681518e-01 -1.18614519e+00
2.07621634e-01 -9.25720856e-02 -6.61321163e-01 8.28572884e-02
1.33655417e+00 4.10448849e-01 6.15723372e-01 -1.00689423e+00
1.06645465e+00 1.24487197e+00 4.78287280e-01 7.16132760e-01
-1.09845376e+00 -3.44689459e-01 2.24962026e-01 8.43090340e-02
-1.51746535e+00 -4.31637049e-01 9.40928161e-01 -6.71576977e-01
1.10875750e+00 3.22764486e-01 9.48107541e-01 1.61161327e+00
1.71186149e-01 1.05155396e+00 9.16032553e-01 -6.74909830e-01
9.67692584e-02 -1.92202292e-02 -2.22570539e-01 4.67084527e-01
1.91577509e-01 -5.32732829e-02 -8.86809766e-01 -5.79780303e-02
1.39058268e+00 3.42456728e-01 -5.44045508e-01 -1.17612696e+00
-1.24194145e+00 6.34022579e-02 2.96760529e-01 6.10588901e-02
-6.32277966e-01 9.72220674e-02 4.78566475e-02 -1.00885302e-01
4.64941144e-01 1.17633097e-01 -4.86116469e-01 -5.10632157e-01
-6.33496404e-01 6.30674183e-01 7.08349943e-01 1.56698966e+00
6.29228652e-02 -3.74842882e-01 -1.68500140e-01 3.03347170e-01
3.79372567e-01 4.99430120e-01 -1.02493219e-01 -8.59870434e-01
8.59947920e-01 5.90282381e-01 4.33452219e-01 -5.61337233e-01
-5.01640320e-01 -2.57480472e-01 -7.54145905e-02 3.68192315e-01
9.02605474e-01 1.09366909e-01 -7.20316291e-01 1.37422276e+00
6.20924473e-01 -6.02203369e-01 -6.36694252e-01 1.45501089e+00
3.74007881e-01 -2.68738121e-01 3.21570747e-02 -5.71215339e-02
1.55018461e+00 -5.29514611e-01 -6.93627834e-01 -3.97619724e-01
1.67985693e-01 -7.91660190e-01 1.14355147e+00 8.36884737e-01
-1.07517004e+00 -3.70586306e-01 -6.41999304e-01 -1.02340966e-01
-1.13319211e-01 4.59151924e-01 6.41418338e-01 8.58056724e-01
-4.87245619e-01 4.67898875e-01 -1.11509860e+00 -6.28326774e-01
1.61492452e-01 5.47121465e-01 -6.85611904e-01 -3.91559154e-02
-6.09013021e-01 1.03495932e+00 5.03130332e-02 3.01010191e-01
-8.56843710e-01 -5.01981795e-01 -8.43554914e-01 -4.40003335e-01
6.15507245e-01 -8.23934197e-01 1.20146203e+00 -1.82551876e-01
-1.53624392e+00 1.31956959e+00 -2.03589246e-01 3.14260393e-01
1.20370448e+00 -8.77742708e-01 2.15742141e-01 2.83729732e-01
-1.76476613e-01 1.36164755e-01 1.07160449e+00 -1.26727581e+00
-1.93371385e-01 -1.34725034e+00 3.15985054e-01 6.04364455e-01
-8.70054811e-02 1.50122553e-01 -8.00461292e-01 -6.38136148e-01
4.86205429e-01 -1.08847368e+00 1.31973475e-02 4.58365858e-01
-4.86449927e-01 -1.65534049e-01 7.88122892e-01 -9.13925290e-01
6.68136477e-01 -1.60141730e+00 6.61553442e-01 1.21964686e-01
1.83745682e-01 -3.37182954e-02 4.63677406e-01 2.64261305e-01
-5.04475465e-05 -8.24792802e-01 3.40548396e-01 -6.63105965e-01
-6.32713661e-02 -3.19880657e-02 2.22647600e-02 1.01145756e+00
-2.53597766e-01 5.38972914e-01 -9.71712351e-01 -4.68124539e-01
7.09638178e-01 8.30563366e-01 -4.59828824e-01 7.26243436e-01
-1.93782836e-01 1.06005323e+00 -3.76235902e-01 1.02515888e+00
5.43838620e-01 2.62539666e-02 2.17437282e-01 -2.64170408e-01
1.28547817e-01 1.70125082e-01 -1.47105253e+00 2.52464604e+00
-4.14604366e-01 3.14869732e-01 3.30673069e-01 -2.56317496e-01
2.87818581e-01 6.21445417e-01 4.91248339e-01 -1.11073643e-01
4.98983890e-01 1.28091546e-02 -3.84116858e-01 -7.94697106e-01
3.48222464e-01 -4.18418720e-02 2.94537544e-01 3.98239762e-01
6.85246289e-02 -3.16726893e-01 -2.37017870e-01 -2.32321888e-01
8.33795488e-01 9.76774693e-01 3.35954070e-01 1.13440938e-01
3.54383200e-01 -1.96911916e-01 -4.25005071e-02 4.12570238e-01
-2.50372499e-01 9.15658355e-01 1.66504696e-01 -3.85835052e-01
-1.10675979e+00 -1.23102379e+00 -5.11684455e-02 9.71697032e-01
3.92188691e-03 -3.10096890e-01 -9.28697467e-01 -4.94883984e-01
3.10932606e-01 2.07625300e-01 -5.27507305e-01 2.61211365e-01
-7.44128346e-01 -1.99521214e-01 1.01066865e-01 9.50887978e-01
6.72838911e-02 -8.53247583e-01 -1.13480389e+00 -9.03070271e-02
-3.91101763e-02 -1.28759325e+00 -3.98856848e-01 -1.13673724e-01
-8.85455132e-01 -1.39242196e+00 -1.37990248e+00 -3.90616417e-01
7.46206701e-01 2.52084911e-01 1.06657648e+00 -7.20744252e-01
-8.66641760e-01 1.10408747e+00 -1.59714431e-01 -3.88275743e-01
1.91020831e-01 -7.25666955e-02 4.90951031e-01 -2.85387069e-01
5.45771956e-01 -5.68942845e-01 -1.02628160e+00 4.51857001e-01
-1.19640166e-02 -2.81489462e-01 2.99692690e-01 4.35556561e-01
2.03023165e-01 -6.71771169e-01 -4.10212129e-01 -3.44791114e-01
3.89459908e-01 1.33296981e-01 -4.68080848e-01 9.66075510e-02
1.15073405e-01 -3.25034201e-01 -1.15312085e-01 -5.46426952e-01
-1.19181180e+00 6.57842696e-01 3.61464061e-02 -7.54285514e-01
-3.47049952e-01 -3.48436385e-01 -4.98110354e-01 -1.00810610e-01
8.22839737e-01 -8.91484469e-02 9.86911356e-02 -5.81784248e-01
3.85585487e-01 9.49490845e-01 6.75599158e-01 -7.88662314e-01
5.88272214e-01 8.71186018e-01 3.45636643e-02 -9.73468125e-01
-4.66817975e-01 -7.31205523e-01 -1.50551486e+00 -5.64733207e-01
8.71979535e-01 -1.08161414e+00 -1.41512322e+00 6.45067394e-01
-1.47261775e+00 7.52447732e-03 -1.37058750e-01 7.72545457e-01
-1.04066217e+00 6.02911890e-01 -3.98945153e-01 -1.33337831e+00
-1.36179104e-01 -1.25888252e+00 1.88491011e+00 -1.62347883e-01
-6.87989652e-01 -6.18214369e-01 -1.26239538e-01 6.01522863e-01
-2.37885624e-01 4.68264818e-01 1.09480560e-01 1.88255653e-01
-6.57827497e-01 -6.99813068e-01 1.80813462e-01 1.55614084e-03
1.86795652e-01 -5.68479061e-01 -1.45779252e+00 -4.61641520e-01
8.82156268e-02 -3.14254344e-01 3.15905660e-02 4.43973124e-01
1.06875145e+00 2.12714463e-01 -3.30197364e-01 5.26724517e-01
9.38696921e-01 -2.02711642e-01 4.00432825e-01 2.85055369e-01
1.02299225e+00 8.79018068e-01 8.11230779e-01 8.54685187e-01
1.05388626e-01 1.21888018e+00 5.05333900e-01 3.21847409e-01
-9.89064202e-02 -2.24855885e-01 -2.08651777e-02 8.88181105e-02
-9.76216614e-01 1.44222707e-01 -9.99086082e-01 3.24413687e-01
-1.55592740e+00 -5.72166502e-01 -5.72865456e-02 2.33169627e+00
6.56828463e-01 -2.60747876e-03 5.33696711e-01 6.54829144e-01
3.81243855e-01 -1.86149970e-01 -6.69624388e-01 3.67123246e-01
3.61918747e-01 1.66872352e-01 4.40992862e-01 3.25785547e-01
-1.12069893e+00 7.82611549e-01 5.98855114e+00 1.39649585e-01
-7.70736158e-01 1.08778834e-01 -3.90225440e-01 -4.65725809e-01
4.54590082e-01 -3.99294287e-01 -7.25192249e-01 -1.61032714e-02
2.20159907e-03 6.02363467e-01 1.89903527e-01 1.19120491e+00
7.21005127e-02 -3.65534157e-01 -1.61536634e+00 1.58342600e+00
4.51680660e-01 -3.50388914e-01 -3.80970180e-01 2.04318419e-01
3.81073505e-01 -2.74946570e-01 -1.12265088e-01 -2.73781419e-01
-3.83400023e-01 -8.11468661e-01 7.72733688e-01 4.85767424e-01
8.92543018e-01 -4.26698416e-01 3.46842200e-01 6.26430452e-01
-1.10757780e+00 4.60613817e-02 -1.41414374e-01 -4.79521394e-01
1.88458964e-01 2.45117843e-01 -8.65987122e-01 2.46185556e-01
7.18656242e-01 5.91593921e-01 -2.62662262e-01 7.38803267e-01
-4.21002358e-01 -3.46734166e-01 -3.88192356e-01 1.02358647e-01
-4.76432949e-01 1.96628831e-02 8.05200994e-01 9.01332200e-01
-1.00790352e-01 1.36963278e-01 1.31279137e-02 6.95305049e-01
2.46498033e-01 -2.37811640e-01 -7.67745256e-01 3.07554841e-01
4.15172353e-02 8.95915747e-01 -6.57149971e-01 1.34074427e-02
-1.33921355e-01 1.43248808e+00 1.17543124e-01 3.83179426e-01
-3.89148206e-01 -1.98842183e-01 8.36797237e-01 5.48435628e-01
1.16733707e-01 -6.70335293e-01 -1.96175531e-01 -1.44581342e+00
8.33754122e-01 -8.56241047e-01 2.91797183e-02 -1.00384092e+00
-8.65318716e-01 3.26563716e-01 3.06451023e-01 -1.63081396e+00
-5.69068968e-01 -1.21173191e+00 4.63965647e-02 1.16829085e+00
-1.02820969e+00 -1.37594259e+00 -9.71535027e-01 7.98659742e-01
7.37681091e-01 -3.58691812e-02 9.80648577e-01 2.97645535e-02
5.66810481e-02 3.93092752e-01 -4.43995833e-01 1.06476746e-01
7.71761119e-01 -1.36114061e+00 2.91574359e-01 4.90217686e-01
-2.31738407e-02 9.06534553e-01 8.57079089e-01 -5.89370668e-01
-2.19803143e+00 -2.04687431e-01 4.85550076e-01 -1.62333584e+00
1.96252793e-01 -8.37966800e-01 -1.15897797e-01 8.81705225e-01
-4.82011616e-01 2.00869128e-01 3.20145249e-01 1.97249576e-01
-3.89568239e-01 1.44188330e-01 -1.22519422e+00 4.98037875e-01
1.73446381e+00 -8.64641666e-01 -8.32216442e-01 7.65398264e-01
-1.25367045e-01 -1.14089823e+00 -9.67656434e-01 1.25018090e-01
1.35503650e+00 -9.40625548e-01 1.34778333e+00 -5.72325647e-01
1.93881571e-01 -1.65960908e-01 -2.42705658e-01 -8.32579136e-01
7.58199245e-02 -6.28473461e-01 -4.99087423e-01 6.36436820e-01
-6.07695103e-01 -1.03800885e-01 1.42439735e+00 6.21807039e-01
5.28241456e-01 -4.50917900e-01 -1.00800753e+00 -8.34032059e-01
-4.76895362e-01 -6.02294922e-01 1.27401546e-01 3.04855257e-01
2.64913499e-01 3.70526202e-02 -3.50529194e-01 1.55218661e-01
1.19755816e+00 -5.63709214e-02 1.42353392e+00 -1.26094902e+00
-2.90616482e-01 -5.89160584e-02 -9.16519046e-01 -1.26375163e+00
6.71529770e-02 -1.79502085e-01 8.69935155e-02 -1.19726038e+00
2.78781980e-01 3.73968557e-02 3.98455024e-01 -5.73505722e-02
-7.93703943e-02 3.87089163e-01 1.81563705e-01 3.31719786e-01
-1.78093046e-01 3.07760328e-01 1.40583789e+00 2.25683540e-01
-1.48217216e-01 3.23964924e-01 9.40721184e-02 9.31776464e-01
1.56784743e-01 -2.22197458e-01 -3.74848455e-01 -3.69382769e-01
1.59734428e-01 2.91294754e-01 6.89001322e-01 -1.02068675e+00
2.11356595e-01 4.65478115e-02 6.05721414e-01 -9.45411086e-01
9.73148286e-01 -1.10398018e+00 1.69996515e-01 6.14227831e-01
-1.17002176e-02 -1.19921722e-01 -4.92459685e-02 4.77166057e-01
3.07549059e-01 1.16713420e-01 3.93133968e-01 -6.01662636e-01
-4.69375253e-01 2.72284597e-01 7.32199550e-02 -4.76899967e-02
1.07934535e+00 -7.13814259e-01 1.46090299e-01 -5.80876350e-01
-1.11518705e+00 -2.06115618e-01 6.89243078e-01 5.61662197e-01
7.37658501e-01 -1.01541519e+00 -4.90477145e-01 5.34758627e-01
3.86251390e-01 2.73762017e-01 5.97775020e-02 9.36090052e-01
-6.29160285e-01 5.96830308e-01 -3.32601190e-01 -1.14489925e+00
-1.67415857e+00 3.84706110e-01 2.63346642e-01 2.38781348e-01
-7.77707219e-01 9.33060646e-01 1.38292015e-01 -4.87934291e-01
7.98828602e-01 -4.31205362e-01 2.08350092e-01 -1.35565117e-01
5.84074855e-01 7.18767047e-01 2.79080778e-01 -7.04740345e-01
-4.76962984e-01 1.07178080e+00 2.16212600e-01 -2.17709377e-01
1.18608415e+00 7.07296841e-03 1.27539948e-01 3.68491143e-01
8.49586129e-01 6.37031943e-02 -1.63556635e+00 -2.43352428e-01
-3.82875592e-01 -1.08559752e+00 -2.92249322e-01 -5.93644857e-01
-6.34368539e-01 1.17831492e+00 7.02921987e-01 -5.46640694e-01
7.59023607e-01 3.08617324e-01 2.27022082e-01 5.63637555e-01
1.05263865e+00 -1.13723564e+00 2.98325032e-01 2.53831923e-01
1.38834083e+00 -1.23751903e+00 5.45461357e-01 -8.12808394e-01
-3.13330173e-01 1.00925589e+00 4.90079492e-01 -1.79212213e-01
6.00618124e-01 2.56497771e-01 -7.52320811e-02 -3.30185860e-01
1.89688131e-01 -1.12445809e-01 4.98825788e-01 9.44904387e-01
3.91199052e-01 1.57349765e-01 1.48868039e-01 1.48173839e-01
-2.51232296e-01 1.35706395e-01 -2.55934130e-02 1.61508119e+00
1.04303300e-01 -9.90051806e-01 -8.59489381e-01 8.89648423e-02
-3.13949525e-01 6.16709709e-01 -4.84526664e-01 1.01499915e+00
-3.41951801e-03 6.28167808e-01 -7.50345364e-02 -1.59857765e-01
9.41782355e-01 1.33025378e-01 1.67124927e+00 -8.30859601e-01
-4.53050584e-01 3.33325595e-01 -8.09053034e-02 -9.69920337e-01
-5.63154578e-01 -9.17442918e-01 -4.93491858e-01 -4.69096452e-02
-3.66453260e-01 -6.29366755e-01 9.62102294e-01 9.68903542e-01
-6.69867545e-02 3.41483444e-01 1.70120597e-01 -1.92973173e+00
-9.71612751e-01 -1.10318828e+00 -9.17344391e-01 5.76679826e-01
3.56764942e-01 -1.22215736e+00 -1.17851935e-01 -1.24352224e-01]
|
[6.548900604248047, -0.8780843019485474]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.