Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files
xet
Community
1
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
f1bdd34b-5e15-475a-b2b2-0a3bde94771b
agreement-among-human-and-automated
null
null
https://openreview.net/forum?id=WL_MC7HgGQ
https://openreview.net/pdf?id=WL_MC7HgGQ
AGREEMENT AMONG HUMAN AND AUTOMATED TRANSCRIPTIONS OF GLOBAL SONGS
Cross-cultural musical analysis requires standardized symbolic representation of sounds such as score notation. However, transcription into notation is usually conducted manually by ear, which is time-consuming and subjective. Our aim is to evaluate the reliability of existing methods for transcribing songs from diverse societies. We had 3 experts independently transcribe a sample of 32 excerpts of traditional monophonic songs from around the world (half a cappella, half with instrumental accompaniment). 16 songs also had pre-existing transcriptions created by 3 different experts. We compared these human transcriptions against one another and against 10 automatic music transcription algorithms. We found that human transcriptions can be sufficiently reliable (~90% agreement, κ ~.7), but current automated methods are not (<60% agreement, κ <.4). No automated method clearly outperformed others, in contrast to our predictions. These results suggest that improving automated methods for cross-cultural music transcription is critical for diversifying MIR.
['Anonymous']
2021-05-24
null
null
null
null
['music-transcription']
['music']
[ 3.91580880e-01 -8.87339637e-02 3.98346782e-01 -1.19305998e-01 -1.38349009e+00 -1.37800848e+00 1.29851848e-01 -6.48287237e-02 -3.18837196e-01 7.57327855e-01 5.35780787e-01 -1.29307210e-01 -3.39159667e-01 -2.16363683e-01 -3.81304741e-01 -3.48504692e-01 1.65265389e-02 6.74598157e-01 3.64939123e-02 -3.27503115e-01 5.28450489e-01 2.36464362e-03 -1.51604772e+00 4.86668766e-01 7.03818202e-01 6.23782754e-01 5.17212339e-02 1.14825821e+00 1.12041701e-02 3.59093428e-01 -1.26542270e+00 -6.79823875e-01 2.43112609e-01 -9.39560115e-01 -7.79555976e-01 -2.58484453e-01 4.61335957e-01 1.90892834e-02 4.69620198e-01 8.35124731e-01 7.18520164e-01 1.59178805e-02 6.45342529e-01 -6.91804528e-01 -6.33680999e-01 1.10954928e+00 -2.67229408e-01 -2.17353299e-01 9.75976706e-01 2.06850395e-02 1.21040392e+00 -4.64408755e-01 7.12748885e-01 9.64908779e-01 1.19555092e+00 2.82220632e-01 -1.31388748e+00 -1.23575962e+00 -6.82860434e-01 -3.96363914e-01 -1.75825024e+00 -5.18683374e-01 6.31544590e-01 -7.62091219e-01 4.82395709e-01 6.74412489e-01 1.30268407e+00 1.03176928e+00 -1.62709355e-01 2.12947756e-01 1.32540905e+00 -4.88720298e-01 2.36320142e-02 1.00730747e-01 -5.97867787e-01 1.39620706e-01 4.08261009e-02 -1.51649237e-01 -9.88121450e-01 -5.34812748e-01 7.82486320e-01 -8.01292956e-01 -3.44076008e-01 3.60391855e-01 -1.47888529e+00 3.50868046e-01 -2.42176831e-01 7.34488189e-01 -2.98375101e-03 9.14020240e-02 5.42887866e-01 6.37299776e-01 5.41601004e-03 1.24499404e+00 -2.86967844e-01 -9.64083135e-01 -1.48407733e+00 3.70898098e-01 8.79272640e-01 7.23230898e-01 1.84038535e-01 2.38522515e-01 2.42917627e-01 1.24362946e+00 5.63008897e-02 6.39583170e-01 6.00710094e-01 -1.34313679e+00 1.61074162e-01 1.16095200e-01 3.60777617e-01 -1.13858855e+00 -1.20096162e-01 -5.43876886e-01 -5.41209400e-01 5.57808429e-02 7.90077865e-01 -8.46782550e-02 -3.91973406e-01 1.55729985e+00 -3.42168719e-01 -1.29545957e-01 -2.77946025e-01 9.56531405e-01 3.99120986e-01 4.53715295e-01 -1.47489801e-01 -3.84435415e-01 1.10315192e+00 -4.05985475e-01 -5.30929983e-01 2.26987302e-01 3.27622294e-01 -1.58969927e+00 1.45222449e+00 1.16291714e+00 -1.28113365e+00 -6.36962593e-01 -1.16773772e+00 1.81507632e-01 3.17036033e-01 2.58748800e-01 2.05095261e-01 1.15386772e+00 -9.24570978e-01 6.71165943e-01 -4.54402566e-01 -4.10670280e-01 -2.02311445e-02 4.43486810e-01 -3.60516340e-01 6.76385820e-01 -1.02120852e+00 5.11126161e-01 2.02746004e-01 4.20859382e-02 -5.37381411e-01 -4.75158095e-01 -1.69232756e-01 -3.84820998e-01 -5.15677929e-02 -2.74137437e-01 1.55353630e+00 -1.41499758e+00 -1.90007317e+00 1.20982265e+00 1.48769319e-01 8.21680948e-03 4.79571551e-01 -4.27605152e-01 -8.02650034e-01 6.85040057e-02 1.73315212e-01 2.51837403e-01 7.43003964e-01 -1.34170520e+00 -5.28249681e-01 6.21025488e-02 -4.30731773e-01 1.45988405e-01 -2.29711264e-01 5.15467644e-01 -2.83502072e-01 -1.11248589e+00 2.73042560e-01 -1.20594883e+00 2.13080496e-01 -3.28457057e-01 -3.32602173e-01 1.90629184e-01 1.35502815e-02 -1.13845408e+00 1.63460898e+00 -2.23084259e+00 6.31985664e-02 6.24472558e-01 -4.36797738e-02 1.64172128e-02 6.63400888e-02 4.96279866e-01 1.23550057e-01 3.07117283e-01 -2.73056537e-01 -8.97587612e-02 2.62055427e-01 -5.24927527e-02 -4.94429380e-01 3.48082513e-01 -4.47515935e-01 4.90416199e-01 -1.08818901e+00 -5.23871422e-01 -3.03071558e-01 4.12586272e-01 -5.34163058e-01 -7.23265558e-02 1.98890686e-01 5.77738464e-01 1.34426624e-01 7.54679561e-01 1.11410134e-01 2.36965045e-01 5.65177500e-01 7.76617676e-02 -2.47681037e-01 5.30258119e-01 -1.41536224e+00 1.98730302e+00 -1.44707993e-01 8.39831293e-01 1.88823476e-01 -2.63289422e-01 1.27519155e+00 5.47775447e-01 9.00542066e-02 -1.30484194e-01 6.23735636e-02 8.44770312e-01 4.85643208e-01 -1.81647658e-01 7.27428257e-01 -7.34644651e-01 -3.55889976e-01 7.39485025e-01 -6.35777861e-02 -8.39531898e-01 1.08484164e-01 -1.42843559e-01 9.38453019e-01 2.09351242e-01 4.25200373e-01 -2.83706248e-01 1.17213055e-01 8.56058970e-02 5.40498316e-01 7.09153652e-01 -1.47964090e-01 1.05925143e+00 3.91651094e-01 -1.07710220e-01 -1.05163074e+00 -1.19560003e+00 1.44927442e-01 1.35217917e+00 -5.36372364e-01 -1.05027318e+00 -7.94618607e-01 2.37817243e-02 -2.82592863e-01 5.69740415e-01 -3.24790627e-01 9.50835496e-02 -3.35332781e-01 -1.60857961e-01 1.38356638e+00 5.11631906e-01 -1.50273427e-01 -9.98012006e-01 -6.85709000e-01 4.08388078e-01 -4.11082894e-01 -5.97283006e-01 -5.68792403e-01 -3.01411431e-02 -5.16074181e-01 -8.02889049e-01 -9.05458748e-01 -7.09340990e-01 -9.64543670e-02 -1.98493749e-01 1.16221750e+00 6.28983006e-02 -6.24773130e-02 2.62890995e-01 -4.19528931e-01 -6.02349818e-01 -7.26442814e-01 2.67102629e-01 4.44899738e-01 -4.07615632e-01 2.39464536e-01 -8.83164287e-01 -3.20157468e-01 5.70917606e-01 -7.02206612e-01 -1.38926789e-01 4.98671860e-01 5.34389913e-01 4.20291245e-01 -1.97234362e-01 5.84322751e-01 -5.91499507e-01 9.07271564e-01 6.08736277e-02 -1.47541881e-01 1.19390100e-01 -9.06283930e-02 -5.17168224e-01 5.80193043e-01 -5.70577741e-01 -7.65227795e-01 -9.91550647e-03 -8.47564172e-03 -7.54918009e-02 -7.13294446e-02 4.39427346e-01 2.48122379e-01 1.84393749e-01 1.01048064e+00 -1.01304539e-01 -3.07989150e-01 -3.91055167e-01 4.24093865e-02 9.42039311e-01 1.02739823e+00 -9.32190239e-01 9.62485731e-01 3.81390681e-03 -4.49247628e-01 -8.95202160e-01 -4.11671609e-01 -2.82741785e-01 -7.56120145e-01 -5.16920567e-01 7.11230397e-01 -7.73089051e-01 -5.84656954e-01 2.60888129e-01 -8.89728189e-01 -3.56820434e-01 -2.49773398e-01 8.33333254e-01 -6.67102814e-01 2.59084612e-01 -6.95721924e-01 -9.19121027e-01 -3.20245445e-01 -7.93517530e-01 1.14022696e+00 -6.90932721e-02 -1.45801890e+00 -4.75910723e-01 6.11852109e-01 5.84723175e-01 4.37961444e-02 3.24417591e-01 5.85525215e-01 -3.89734566e-01 1.54109612e-01 -2.80580908e-01 2.69082457e-01 1.82778135e-01 5.37069619e-01 4.31763262e-01 -1.13833487e+00 -1.20212667e-01 -3.20368648e-01 -6.80413783e-01 1.69819385e-01 -1.27327265e-02 6.41652822e-01 -5.30576520e-02 2.13298947e-01 2.96388447e-01 8.30959976e-01 4.13985312e-01 6.13899410e-01 2.84559011e-01 2.74050504e-01 6.90164208e-01 5.53202748e-01 4.80142564e-01 -1.61196932e-01 6.43927991e-01 -4.34311509e-01 4.32151973e-01 8.33954737e-02 -4.24869657e-01 7.03154087e-01 1.55304861e+00 -7.38068521e-01 2.58277655e-02 -1.14845860e+00 5.38816452e-01 -1.37684155e+00 -1.11765516e+00 -1.49520382e-01 2.21776199e+00 1.09868133e+00 1.80205867e-01 5.89111209e-01 7.39266694e-01 6.93615854e-01 -1.83411866e-01 -3.16771939e-02 -6.40673041e-01 -3.16863447e-01 7.20388412e-01 -3.70024033e-02 2.31556520e-01 -7.14465499e-01 7.89368570e-01 7.88800573e+00 7.82420337e-01 -8.93230140e-01 -2.42778406e-01 3.95724410e-03 -6.13890529e-01 -3.90715718e-01 -1.40157947e-02 -7.48861730e-02 4.06943023e-01 9.73653078e-01 -1.31229371e-01 8.03672731e-01 3.89500409e-01 5.37370853e-02 8.53750482e-02 -8.81815493e-01 1.20215154e+00 2.62091011e-01 -1.00799465e+00 -7.68980710e-03 -5.90159707e-02 9.18484569e-01 -3.60260099e-01 -5.02519570e-02 9.07489061e-02 2.23268270e-01 -1.13619232e+00 1.18656564e+00 5.63239694e-01 1.20746434e+00 -7.38817692e-01 5.25144994e-01 -1.34050086e-01 -1.32487524e+00 2.58905768e-01 3.07301544e-02 -4.53506380e-01 3.71952832e-01 2.45699108e-01 -8.99334967e-01 3.82704049e-01 8.38619351e-01 4.36723709e-01 -6.25622630e-01 9.55372334e-01 -2.18861252e-01 1.18784487e+00 -5.19266009e-01 -1.50094582e-02 -3.32735218e-02 -3.53446007e-01 5.74006438e-01 1.31455123e+00 8.79571259e-01 2.03438904e-02 -2.41416261e-01 4.78652000e-01 2.11358041e-01 4.62612420e-01 -3.85131061e-01 -5.45841813e-01 5.75465798e-01 8.79912853e-01 -1.14479303e+00 -1.09594621e-01 -7.53113851e-02 9.98327553e-01 -1.78489983e-01 1.02938890e-01 -6.87749684e-01 -6.04389787e-01 4.60091799e-01 7.53495172e-02 -7.73192151e-03 -3.55054885e-01 -4.51847613e-01 -7.21557319e-01 -9.33952257e-02 -1.40726745e+00 3.60365629e-01 -1.06212044e+00 -1.23235786e+00 4.78652418e-01 -3.39010388e-01 -1.58580816e+00 -4.66118425e-01 -4.70536739e-01 -2.62602150e-01 7.57127106e-01 -4.75731269e-02 -8.63059938e-01 -5.19185476e-02 1.37904614e-01 1.86486036e-01 -2.03554019e-01 1.13591397e+00 3.94962966e-01 1.75190210e-01 5.77895164e-01 1.40021831e-01 2.03971371e-01 1.41500449e+00 -1.20639503e+00 1.80595055e-01 1.97923899e-01 6.70367777e-01 1.03657722e+00 9.09394801e-01 -6.05303168e-01 -7.43454933e-01 -4.02766407e-01 8.47820282e-01 -5.53646386e-01 9.32619095e-01 -3.20330828e-01 -6.31527781e-01 5.83591104e-01 3.08012217e-01 -9.73354101e-01 1.65019035e+00 3.84286076e-01 -3.95212680e-01 7.38955587e-02 -5.54533005e-01 7.72210658e-01 1.23287797e+00 -1.05869925e+00 -9.44557250e-01 -1.92316934e-01 2.05458611e-01 -2.70777047e-01 -1.12652469e+00 8.80981684e-02 1.61061943e+00 -1.01630211e+00 7.07623541e-01 -3.31424445e-01 3.77238601e-01 -7.55180001e-01 -3.27364087e-01 -1.32177007e+00 -3.13599050e-01 -1.09891605e+00 5.72351754e-01 1.52967691e+00 4.19671059e-01 -2.13468239e-01 3.59363198e-01 3.52123618e-01 -2.90752500e-01 5.32276779e-02 -7.16833413e-01 -8.18572819e-01 -2.56801043e-02 -8.47247541e-01 4.16391164e-01 1.24403286e+00 3.83471847e-01 3.25095683e-01 -4.84545529e-01 -2.97181904e-01 1.98166415e-01 5.32800369e-02 1.00482106e+00 -1.45178580e+00 -7.16370165e-01 -7.90333509e-01 -6.34340942e-01 -2.83855557e-01 -1.10151283e-01 -8.57705772e-01 -8.37541372e-02 -1.08566320e+00 6.72078505e-02 -1.32144451e-01 -1.98261470e-01 3.05575937e-01 1.40326530e-01 1.08960927e+00 3.94117236e-01 4.13019955e-01 -3.08102876e-01 9.58897695e-02 1.00870025e+00 4.29511210e-03 -6.30357504e-01 -1.78179443e-01 -8.29236269e-01 9.89353895e-01 9.39554989e-01 -5.79339981e-01 -2.74634629e-01 -2.70352304e-01 8.47209334e-01 -2.57239550e-01 7.61007741e-02 -1.36597073e+00 1.36563554e-01 -4.52733897e-02 4.23904806e-01 -3.00272226e-01 3.15212280e-01 -3.11605066e-01 8.78677547e-01 2.34931424e-01 -2.97654301e-01 9.34821442e-02 3.58548313e-01 1.62187248e-01 -3.54781657e-01 -3.00519764e-01 3.23800594e-01 -1.54978499e-01 -9.56516564e-02 -7.92530835e-01 -7.94295371e-01 2.82925040e-01 3.50314409e-01 -4.92684692e-01 3.29886675e-02 -7.22223580e-01 -7.56005526e-01 -4.87490296e-01 7.91950345e-01 2.07203984e-01 1.60381153e-01 -1.40694332e+00 -6.85617208e-01 -1.51023909e-01 1.67723835e-01 -5.61869144e-01 -1.04890332e-01 8.58750880e-01 -1.02908266e+00 1.05472192e-01 -3.34125698e-01 -3.53288561e-01 -1.50379324e+00 -7.07676336e-02 -3.01426817e-02 1.88492164e-01 -2.45655984e-01 8.89913559e-01 -2.47945905e-01 -5.11195660e-01 2.95773372e-02 -4.17191982e-01 1.02746166e-01 3.27832371e-01 4.06158179e-01 5.88549376e-01 -1.47620425e-01 -6.84998393e-01 -3.23170930e-01 9.62377369e-01 4.56447333e-01 -1.10485017e+00 9.02464092e-01 1.15526728e-01 -2.25527838e-01 1.39302886e+00 6.53397739e-01 9.08039808e-01 -4.33865070e-01 4.56885189e-01 1.57436728e-02 -5.57029068e-01 -5.04257858e-01 -8.91123831e-01 -1.20130435e-01 5.24029613e-01 1.49176151e-01 2.85762280e-01 1.10597253e+00 -2.46897727e-01 7.24281490e-01 5.39569259e-01 5.66259742e-01 -1.59928071e+00 -5.45479730e-02 3.90135825e-01 1.16483974e+00 -3.75295192e-01 4.50602286e-02 -1.01660892e-01 -7.95877814e-01 1.02648997e+00 1.07795060e-01 6.26224205e-02 2.99453616e-01 1.58018515e-01 4.07202572e-01 4.82229590e-02 -3.77912313e-01 1.38251737e-01 5.61258912e-01 4.91530389e-01 1.13869941e+00 3.11387569e-01 -7.00707734e-01 1.08673465e+00 -1.42685938e+00 3.23630124e-02 3.10664147e-01 7.06300616e-01 -2.32088774e-01 -1.21314847e+00 -9.27351832e-01 3.30515325e-01 -5.23907006e-01 -5.61233386e-02 -1.19561970e+00 7.26930261e-01 4.03303504e-01 1.08718467e+00 -1.02185339e-01 -7.69600272e-01 9.83705670e-02 4.30787385e-01 6.42374933e-01 -5.06239116e-01 -1.13634217e+00 5.70379019e-01 4.39200491e-01 -1.36558622e-01 -5.84703684e-01 -9.96774614e-01 -9.48761165e-01 -1.99562341e-01 -1.56298190e-01 5.67324877e-01 2.52625853e-01 5.07507265e-01 -1.32771626e-01 3.04402292e-01 1.41042903e-01 -7.96448946e-01 -1.10546038e-01 -1.12605584e+00 -9.72096562e-01 3.50802124e-01 -1.29409090e-01 -3.91550988e-01 -5.08211136e-01 5.24535418e-01]
[15.92697525024414, 5.371687889099121]
18e73d3b-2f00-49e1-9bd3-518e04bcd4de
locate-this-not-that-class-conditioned-sound
2203.04197
null
https://arxiv.org/abs/2203.04197v1
https://arxiv.org/pdf/2203.04197v1.pdf
Locate This, Not That: Class-Conditioned Sound Event DOA Estimation
Existing systems for sound event localization and detection (SELD) typically operate by estimating a source location for all classes at every time instant. In this paper, we propose an alternative class-conditioned SELD model for situations where we may not be interested in localizing all classes all of the time. This class-conditioned SELD model takes as input the spatial and spectral features from the sound file, and also a one-hot vector indicating the class we are currently interested in localizing. We inject the conditioning information at several points in our model using feature-wise linear modulation (FiLM) layers. Through experiments on the DCASE 2020 Task 3 dataset, we show that the proposed class-conditioned SELD model performs better in terms of common SELD metrics than the baseline model that locates all classes simultaneously, and also outperforms specialist models that are trained to locate only a single class of interest. We also evaluate performance on the DCASE 2021 Task 3 dataset, which includes directional interference (sound events from classes we are not interested in localizing) and notice especially strong improvement from the class-conditioned model.
['Jonathan Le Roux', 'Zhong-Qiu Wang', 'Gordon Wichern', 'Olga Slizovskaia']
2022-03-08
null
null
null
null
['sound-event-localization-and-detection']
['audio']
[ 1.67828634e-01 -4.76213634e-01 1.62748396e-01 -3.13038439e-01 -1.75333881e+00 -7.02409446e-01 5.56310534e-01 3.85228485e-01 -4.98925567e-01 3.45587403e-01 2.79412895e-01 -8.66451189e-02 -1.54106557e-01 -4.36726362e-01 -8.04579556e-01 -6.72872007e-01 -4.90499675e-01 2.61656433e-01 5.99185884e-01 1.00396387e-01 2.89696306e-01 4.39462870e-01 -1.49734855e+00 7.95896292e-01 2.86622524e-01 1.07707453e+00 5.76602042e-01 1.02567208e+00 4.34819788e-01 7.50992715e-01 -9.80750918e-01 3.02277774e-01 -2.31072888e-01 -6.60872936e-01 -7.24544764e-01 -3.65919799e-01 7.80790567e-01 -1.14970934e-02 -2.42775485e-01 8.72555137e-01 8.68308544e-01 2.39164397e-01 5.86457133e-01 -1.03319788e+00 9.37946886e-02 7.07450151e-01 -3.42063367e-01 8.49225104e-01 4.44299459e-01 -2.90874958e-01 1.23372698e+00 -1.15973043e+00 2.67647535e-01 8.82474601e-01 6.74099565e-01 2.24023446e-01 -1.16726089e+00 -7.45428264e-01 3.41479242e-01 7.04264581e-01 -1.66830420e+00 -8.70088339e-01 8.36044788e-01 -2.76937455e-01 1.25933647e+00 5.76715469e-01 5.61528876e-02 1.11100042e+00 -9.23913792e-02 6.54460728e-01 7.21874058e-01 -6.37532711e-01 4.42262024e-01 -5.39260842e-02 1.28449827e-01 2.08131000e-01 -4.67728615e-01 1.79555669e-01 -1.19793224e+00 -3.20925444e-01 2.79746979e-01 -5.55169284e-01 -5.07853806e-01 1.39756441e-01 -1.36257946e+00 4.09986883e-01 2.67436147e-01 5.23351133e-01 -2.95690089e-01 2.50304133e-01 2.28103921e-01 2.60355085e-01 6.02515876e-01 3.79843205e-01 -4.81018484e-01 -3.55787724e-01 -1.46019673e+00 2.45053887e-01 7.80235410e-01 9.68560219e-01 4.28843439e-01 -7.07797380e-03 -5.10424078e-01 1.09980750e+00 3.56847942e-02 4.29515064e-01 3.22216988e-01 -8.71775448e-01 6.40423656e-01 -3.54640216e-01 1.57553256e-01 -7.37776160e-01 -6.45780385e-01 -8.83600295e-01 -2.77905315e-01 -1.49537072e-01 1.55110836e-01 -2.10810259e-01 -7.38908947e-01 2.06254172e+00 -7.94892665e-03 6.62030280e-01 -1.97438806e-01 8.08814824e-01 6.53427601e-01 1.01466131e+00 -7.90821835e-02 -2.62556940e-01 1.09483910e+00 -7.37756014e-01 -6.34829283e-01 -2.49259531e-01 5.26666105e-01 -8.73749197e-01 6.75363541e-01 6.94415689e-01 -1.02485847e+00 -5.75565159e-01 -9.92699027e-01 1.73063412e-01 -2.78536916e-01 4.13542330e-01 1.70026198e-01 5.58322012e-01 -1.14044189e+00 4.03573483e-01 -8.76774073e-01 -2.66922861e-01 1.01877324e-01 2.19264716e-01 -1.83572426e-01 2.72889227e-01 -1.16923058e+00 5.74984908e-01 -3.02353594e-02 1.22525387e-01 -1.31031358e+00 -7.14913785e-01 -5.97171545e-01 3.98829073e-01 1.42079011e-01 7.76310265e-02 1.69223773e+00 -4.65412706e-01 -1.23667932e+00 3.05726409e-01 -5.91023624e-01 -4.57406789e-01 -9.00539830e-02 -2.50352859e-01 -7.30925381e-01 3.42683196e-01 2.81118214e-01 3.81013632e-01 7.27701962e-01 -1.15490997e+00 -1.15939140e+00 4.11553755e-02 -2.14915127e-02 1.80315569e-01 -3.41878057e-01 4.09039557e-01 -4.64336157e-01 -6.95680618e-01 2.48243168e-01 -6.31208420e-01 1.00046739e-01 -1.71920612e-01 -6.65610194e-01 -2.44538859e-01 7.95908749e-01 -5.74761748e-01 1.46843028e+00 -2.29795432e+00 -1.27017677e-01 1.81145340e-01 -2.68641263e-01 6.86402917e-02 -2.58682430e-01 5.58536053e-01 -3.77915740e-01 -1.24688953e-01 -1.55063927e-01 -6.48988724e-01 1.13469049e-01 -1.61895439e-01 -5.36564291e-01 4.83318567e-01 2.42962077e-01 2.36039639e-01 -8.21439981e-01 -2.80756623e-01 1.34363770e-01 4.59046245e-01 -8.27888250e-01 1.18583977e-01 -3.74942683e-02 2.30325103e-01 -1.57898366e-01 5.78648508e-01 5.85343599e-01 2.47467253e-02 -1.08696129e-02 -3.23335975e-01 -3.09818089e-01 7.73667157e-01 -1.34138179e+00 1.70538747e+00 -7.86146581e-01 9.79914725e-01 9.05452296e-02 -1.03849411e+00 3.63269627e-01 7.65020430e-01 4.26568449e-01 -7.85089612e-01 -3.27147633e-01 2.50598848e-01 7.57532865e-02 -3.20346922e-01 1.99508280e-01 -3.44209000e-02 -2.20204920e-01 4.73478615e-01 3.54436278e-01 1.15004197e-01 1.08593568e-01 2.99567848e-01 1.45469701e+00 -1.68158188e-01 -9.98730436e-02 -2.43788317e-01 3.42303425e-01 -3.86518538e-01 1.78269953e-01 1.23054445e+00 -3.36252488e-02 1.04777837e+00 4.46223855e-01 3.59260589e-01 -3.39551240e-01 -1.35333419e+00 -4.93099362e-01 1.33405340e+00 3.79862115e-02 -5.75960517e-01 -4.85583723e-01 -4.67468858e-01 -3.81274223e-01 1.00196481e+00 -4.48160768e-01 -1.49534671e-02 -6.25226140e-01 -6.00827575e-01 7.36487329e-01 5.88199079e-01 1.83354363e-01 -9.98209476e-01 -4.48104799e-01 3.86790872e-01 -6.20269537e-01 -1.14482224e+00 -6.76986933e-01 8.43116522e-01 -2.01413468e-01 -6.75198078e-01 -5.39509833e-01 -8.86000097e-01 2.83214450e-01 9.82006565e-02 9.53698099e-01 -3.68482500e-01 -1.12950973e-01 4.59153473e-01 -5.37049830e-01 -4.99140233e-01 -3.93189192e-02 7.25301132e-02 -1.10061146e-01 2.89367348e-01 7.12551549e-02 -6.70421422e-01 -5.15156448e-01 2.68507212e-01 -5.48104823e-01 -4.09769833e-01 2.27908790e-01 6.86364233e-01 4.25803155e-01 2.21544594e-01 9.32681262e-01 -6.15926445e-01 3.87226760e-01 -6.93947852e-01 -2.03255400e-01 -6.90888464e-02 -1.90216731e-02 -3.99673432e-01 6.31106555e-01 -4.77154940e-01 -9.56922531e-01 9.85648483e-02 -6.28692091e-01 -1.26427501e-01 -3.15266222e-01 2.95867324e-01 -2.29891196e-01 8.44829530e-02 7.98411906e-01 1.97101772e-01 -7.61838615e-01 -8.23409259e-01 1.13299778e-02 1.08654583e+00 7.20570087e-01 -5.02459168e-01 4.76863414e-01 2.80758440e-01 -2.03525946e-01 -1.01186979e+00 -9.19201910e-01 -7.19326556e-01 -3.96618694e-01 -1.41333327e-01 6.50831223e-01 -9.89882231e-01 -2.51890600e-01 4.83009368e-01 -1.32876492e+00 -2.79765159e-01 -2.33496219e-01 7.64891088e-01 -5.73583782e-01 -7.31680691e-02 -4.74056929e-01 -9.31326091e-01 1.80887267e-01 -9.60615695e-01 1.40605319e+00 -1.33997589e-01 -3.12304258e-01 -6.50578618e-01 5.14703169e-02 -2.67835706e-01 4.16917294e-01 -1.82722345e-01 9.11816478e-01 -8.29875946e-01 -4.27961737e-01 -3.10397714e-01 1.82954028e-01 3.37918311e-01 8.06180909e-02 -4.68271494e-01 -1.51073861e+00 -9.20411050e-02 5.56951687e-02 3.12095694e-03 1.00263286e+00 4.89416122e-01 1.44557559e+00 4.92975339e-02 -4.85872805e-01 4.71394151e-01 1.02129972e+00 5.18663704e-01 4.11153704e-01 1.32380193e-02 3.71184975e-01 3.14269245e-01 5.44093430e-01 3.79491210e-01 2.66575485e-01 1.11521518e+00 4.20229912e-01 -2.87652135e-01 -3.50043923e-01 -2.09270313e-01 2.97643453e-01 6.48888588e-01 3.52013022e-01 -8.36342573e-01 -5.68155766e-01 1.01930642e+00 -1.51792097e+00 -1.19758153e+00 -1.26679778e-01 2.28842473e+00 9.57936168e-01 5.49502410e-02 5.16671427e-02 5.48962951e-01 7.72119343e-01 1.50282711e-01 -8.12567100e-02 -3.16924989e-01 -1.36093274e-01 6.99063361e-01 1.52452160e-02 7.38992512e-01 -1.38320518e+00 6.11413538e-01 6.68120241e+00 1.06345689e+00 -1.22799003e+00 3.48069280e-01 3.57660711e-01 -6.04896843e-01 1.86312214e-01 -7.33083412e-02 -8.35594475e-01 5.71080387e-01 1.25607288e+00 2.08261505e-01 3.80196005e-01 2.64712155e-01 2.94286311e-01 -3.63398194e-01 -1.53880632e+00 8.36769760e-01 5.63862696e-02 -1.06563163e+00 -4.03581113e-01 -2.08706275e-01 4.01491284e-01 4.87031788e-02 1.57310963e-01 2.47720212e-01 -3.11001003e-01 -6.27249658e-01 1.23483419e+00 4.09955084e-01 7.84935772e-01 -7.72999883e-01 3.06761086e-01 5.16942501e-01 -1.26621544e+00 -1.41702279e-01 -1.86186824e-02 5.63130081e-02 2.28591114e-01 5.37467182e-01 -7.25162745e-01 3.42134237e-01 1.00717247e+00 3.92314672e-01 -4.25108254e-01 1.42855406e+00 -2.14285359e-01 1.34777415e+00 -7.14824617e-01 1.32130370e-01 6.92884549e-02 5.72985768e-01 7.56124258e-01 1.59648371e+00 5.94562113e-01 -3.04329488e-02 3.56844105e-02 5.64227164e-01 1.79527968e-01 -2.98326220e-02 -2.68521816e-01 2.88441718e-01 6.56561673e-01 8.71045232e-01 -5.09792268e-01 -1.68696746e-01 -5.04159570e-01 1.03107274e+00 1.64629489e-01 5.76493084e-01 -9.25540149e-01 -9.30967927e-01 4.86691624e-01 -1.06344461e-01 6.57838225e-01 -7.87708014e-02 -1.60593987e-02 -6.53607845e-01 -4.28501554e-02 -5.59790194e-01 3.54400754e-01 -9.85716403e-01 -1.00565588e+00 6.08806312e-01 1.31072819e-01 -1.31982899e+00 -3.50019127e-01 -3.63023520e-01 -7.90888846e-01 9.84756351e-01 -1.37217116e+00 -7.72280633e-01 2.43444815e-01 5.65845490e-01 5.71017861e-01 1.30625710e-01 9.60568011e-01 7.76255548e-01 -3.11271071e-01 8.76290500e-01 1.24232844e-02 -8.83156434e-02 8.83893311e-01 -1.27257097e+00 1.78359851e-01 9.38306332e-01 9.01461959e-01 4.69083875e-01 8.35382700e-01 -2.12677911e-01 -8.93894017e-01 -1.05336344e+00 1.48495317e+00 -4.18262213e-01 5.59498250e-01 -8.00080061e-01 -7.47407675e-01 6.48207903e-01 1.49089796e-02 1.66985929e-01 5.46506584e-01 2.93306112e-01 -1.28044225e-02 -3.03488642e-01 -8.31335604e-01 2.10572943e-01 1.04264426e+00 -8.72736990e-01 -3.86489093e-01 6.66814387e-01 4.72060829e-01 -3.68142784e-01 -4.66241330e-01 2.29497448e-01 3.39680463e-01 -9.21108425e-01 9.96379435e-01 -2.41729423e-01 -1.76241249e-02 -3.91643465e-01 -4.54890341e-01 -1.32302547e+00 -2.97330678e-01 -5.00776410e-01 -1.23662278e-01 1.33958459e+00 7.54418969e-01 -3.20790231e-01 2.27805063e-01 -1.79467395e-01 -4.94060487e-01 -4.19217825e-01 -1.37294376e+00 -7.41528809e-01 -6.92077503e-02 -1.07879245e+00 3.51041257e-01 4.69638139e-01 4.49217521e-02 2.62625456e-01 -3.71560842e-01 8.54097426e-01 3.60682458e-01 1.82554051e-01 1.46370336e-01 -8.04088175e-01 -4.94247168e-01 -2.02964157e-01 -3.11353654e-01 -1.46971822e+00 -5.13759553e-02 -9.61754620e-01 7.13578165e-01 -1.29891372e+00 -2.86555570e-02 -3.91973615e-01 -8.58933568e-01 6.24772072e-01 -8.58744606e-03 4.37497556e-01 -2.12825537e-01 -7.71341622e-02 -6.71212256e-01 3.06754410e-01 6.93417430e-01 -1.88395614e-03 -4.93946150e-02 3.70660573e-01 -7.32966006e-01 5.21867871e-01 5.11034906e-01 -7.54302561e-01 -3.19837600e-01 -4.86353338e-01 -1.66322023e-01 3.22945833e-01 7.70964742e-01 -1.46892869e+00 7.44253933e-01 1.34641245e-01 3.82388979e-01 -8.72266114e-01 6.77865088e-01 -7.05798447e-01 -8.59980360e-02 -4.38848883e-02 -7.01264560e-01 -5.23312509e-01 3.87080461e-01 5.40097058e-01 -3.50813180e-01 -3.34579051e-01 7.60043621e-01 3.66989970e-01 -6.54084802e-01 2.75597852e-02 -6.88925683e-01 2.19922606e-02 6.29791737e-01 2.90541142e-01 -2.67078936e-01 -5.28205872e-01 -1.09071279e+00 -1.49894161e-02 -3.54405642e-01 3.90240639e-01 3.89299214e-01 -1.29962301e+00 -6.62451625e-01 4.33376402e-01 5.97705804e-02 -5.28321505e-01 2.00450048e-01 6.57709539e-01 -9.27752443e-03 5.90256155e-01 3.80115896e-01 -7.80590951e-01 -1.09639108e+00 2.27727830e-01 5.75788081e-01 1.57798588e-01 -4.76074368e-01 1.44338751e+00 3.35854560e-01 -1.39358222e-01 6.20817125e-01 -3.27974051e-01 -2.95543164e-01 7.16563985e-02 5.85418701e-01 3.12849700e-01 5.72082996e-01 -8.02458763e-01 -7.89500356e-01 4.63677049e-01 1.68100998e-01 -6.31048501e-01 1.09851933e+00 -9.30736065e-02 2.75395811e-01 6.77899837e-01 1.36005926e+00 4.28889573e-01 -9.90910351e-01 -3.17262471e-01 -5.54028675e-02 -3.63747597e-01 3.64371270e-01 -1.16631985e+00 -9.76555526e-01 1.19001865e+00 8.69202971e-01 3.72091115e-01 1.39167321e+00 2.93842882e-01 5.22618055e-01 2.78262407e-01 4.58699882e-01 -1.00176060e+00 -8.34731534e-02 5.30486643e-01 9.09278333e-01 -6.82483137e-01 -3.80352974e-01 -2.67432421e-01 -1.88638180e-01 8.87321711e-01 4.02483016e-01 3.22935171e-02 9.87321973e-01 6.23620033e-01 3.59608419e-02 -1.21627770e-01 -9.08310294e-01 -4.25897181e-01 3.65425169e-01 7.12968111e-01 4.30857271e-01 -5.99628948e-02 -3.08940746e-02 8.74217689e-01 -2.89978147e-01 -5.14595687e-01 9.26580578e-02 1.00026453e+00 -7.00967789e-01 -1.06932437e+00 -3.95056576e-01 5.12256205e-01 -5.50813377e-01 -3.46073508e-01 -1.96871176e-01 4.67638910e-01 3.12700063e-01 1.54415798e+00 2.39790872e-01 -4.41790193e-01 5.40963411e-01 1.63685679e-01 3.73100609e-01 -8.84530842e-01 -5.68998814e-01 5.55871010e-01 3.17925811e-01 -6.07391596e-01 -4.23486866e-02 -9.12684143e-01 -1.25990915e+00 4.39866990e-01 -5.65695167e-01 3.86341810e-01 6.53059006e-01 9.00847018e-01 2.15987429e-01 9.64935780e-01 8.94828856e-01 -1.02820015e+00 -3.37428480e-01 -1.03387427e+00 -7.77092338e-01 -5.75054362e-02 7.94547677e-01 -5.52146971e-01 -6.79804742e-01 9.93228927e-02]
[15.198359489440918, 5.198291778564453]
b72b4250-d82f-447e-a9eb-70f8cabb08e1
leveraging-multi-view-image-sets-for
1911.07262
null
https://arxiv.org/abs/1911.07262v1
https://arxiv.org/pdf/1911.07262v1.pdf
Leveraging Multi-view Image Sets for Unsupervised Intrinsic Image Decomposition and Highlight Separation
We present an unsupervised approach for factorizing object appearance into highlight, shading, and albedo layers, trained by multi-view real images. To do so, we construct a multi-view dataset by collecting numerous customer product photos online, which exhibit large illumination variations that make them suitable for training of reflectance separation and can facilitate object-level decomposition. The main contribution of our approach is a proposed image representation based on local color distributions that allows training to be insensitive to the local misalignments of multi-view images. In addition, we present a new guidance cue for unsupervised training that exploits synergy between highlight separation and intrinsic image decomposition. Over a broad range of objects, our technique is shown to yield state-of-the-art results for both of these tasks.
['Stephen Lin', 'Renjiao Yi', 'Ping Tan']
2019-11-17
null
null
null
null
['intrinsic-image-decomposition']
['computer-vision']
[ 4.30127740e-01 -4.58802938e-01 -1.93670020e-02 -3.94759089e-01 -4.19129044e-01 -8.49180341e-01 4.28655714e-01 -3.38890225e-01 1.78384602e-01 1.70139685e-01 -7.93874916e-03 2.31880188e-01 -1.61663201e-02 -5.60838640e-01 -6.97946489e-01 -9.23902631e-01 4.51295614e-01 -5.56911975e-02 1.65629819e-01 -3.09285045e-01 2.47721612e-01 8.30134034e-01 -1.84361923e+00 7.80343533e-01 7.57137179e-01 8.92592490e-01 2.43784532e-01 6.17925763e-01 2.55063742e-01 5.86269498e-01 -2.20429718e-01 -3.97393405e-01 5.89492917e-01 -1.02167530e-02 -3.21700692e-01 7.93700755e-01 1.09490144e+00 -5.96108913e-01 9.27093774e-02 9.24822330e-01 3.52235258e-01 5.61185777e-02 5.52104056e-01 -1.04838371e+00 -7.99051523e-01 7.17051402e-02 -7.91140079e-01 -5.82896769e-02 2.36116469e-01 1.45386502e-01 1.25192082e+00 -1.23444116e+00 4.82167482e-01 9.95531142e-01 3.98923457e-01 1.90651625e-01 -1.56280065e+00 -1.93644524e-01 4.67057556e-01 -3.93951423e-02 -9.55838680e-01 -5.13121486e-01 1.34881735e+00 -5.40199220e-01 8.25175643e-01 1.74422726e-01 6.04772508e-01 8.82155001e-01 1.19955294e-01 7.49579132e-01 1.50598121e+00 -5.76644421e-01 8.65728315e-03 4.21697795e-01 -1.78696271e-02 6.67406857e-01 2.49332115e-01 1.90125644e-01 -6.71403825e-01 7.07969293e-02 7.45455265e-01 2.18177751e-01 -3.08867991e-01 -9.00183082e-01 -1.05289721e+00 4.41640556e-01 3.87287498e-01 -1.36302978e-01 -2.02639699e-01 -1.42636716e-01 -8.79146829e-02 7.01254457e-02 6.73428714e-01 3.28153104e-01 -5.76864839e-01 4.42816049e-01 -6.43023968e-01 1.08360937e-02 4.15109366e-01 8.38925362e-01 9.26544368e-01 1.22317582e-01 1.87644273e-01 1.19169712e+00 3.20660830e-01 9.01279271e-01 -6.16176650e-02 -1.03611541e+00 3.50066513e-01 7.19034553e-01 2.23217964e-01 -9.96782839e-01 -4.75367010e-01 -3.66078883e-01 -5.68611026e-01 5.49353600e-01 3.07546973e-01 2.36653313e-01 -8.70728314e-01 1.40326774e+00 4.14254755e-01 -2.79557616e-01 -2.83396207e-02 9.25795019e-01 5.91121733e-01 1.62187606e-01 -5.97539723e-01 -1.45421952e-01 1.29381907e+00 -1.13038003e+00 -4.39130872e-01 -2.68193752e-01 1.71613872e-01 -1.15411353e+00 1.18761277e+00 7.93814301e-01 -1.29434609e+00 -7.94430077e-01 -1.13281274e+00 -1.50981009e-01 -3.10807586e-01 5.86051404e-01 8.92005205e-01 8.76565635e-01 -8.37561607e-01 3.66459370e-01 -5.45882285e-01 -1.62299439e-01 2.39556521e-01 -4.78565134e-03 -2.70183951e-01 -5.28771698e-01 -3.88898462e-01 5.43407738e-01 -4.62093800e-02 2.49280572e-01 -6.68566227e-01 -7.27690101e-01 -6.88704729e-01 -1.25595763e-01 2.86561489e-01 -7.45987952e-01 7.28492796e-01 -1.14281237e+00 -1.69341707e+00 1.16240919e+00 -2.30521768e-01 1.26028731e-01 2.15313897e-01 -2.88408846e-01 -3.58503520e-01 4.28901613e-01 -2.63037413e-01 3.21646005e-01 1.24425781e+00 -1.97578168e+00 -3.26048434e-01 -6.04838729e-01 1.91749752e-01 4.76494461e-01 -3.70712042e-01 5.93778901e-02 -6.90149069e-01 -6.06383920e-01 4.03973788e-01 -8.60514760e-01 1.60909183e-02 -7.78870359e-02 -3.13187629e-01 4.53989685e-01 7.31313407e-01 -4.05840099e-01 7.19438612e-01 -2.28788638e+00 2.33146787e-01 1.57837391e-01 1.52288094e-01 -9.37854126e-02 -2.15658695e-01 4.24948722e-01 -2.10280538e-01 -2.37853125e-01 1.26256272e-01 -2.92119324e-01 -1.85665667e-01 1.86887324e-01 -4.42156106e-01 5.93030572e-01 2.44014964e-01 6.30662262e-01 -6.21393681e-01 -5.81543110e-02 4.61674660e-01 4.91895944e-01 -5.99665642e-01 3.50771993e-01 -7.22488612e-02 3.81647974e-01 -1.66490853e-01 9.12498236e-01 1.08530426e+00 -3.21559250e-01 1.44522205e-01 -6.95636094e-01 -7.32728988e-02 -2.31421329e-02 -1.19628084e+00 1.50828314e+00 -5.65772116e-01 3.74552906e-01 3.32269758e-01 -6.74988031e-01 7.88881361e-01 -1.39631063e-01 8.00124466e-01 -5.47175825e-01 -2.33429000e-02 -1.53340057e-01 -3.09979796e-01 -2.85686970e-01 6.05532646e-01 7.55899027e-02 5.02558112e-01 4.91352290e-01 -2.28401981e-02 -4.38305765e-01 1.66691229e-01 1.99591532e-01 5.26481748e-01 3.76034558e-01 7.83169866e-02 -6.54817522e-02 5.44383407e-01 -4.40909892e-01 3.92121345e-01 4.51196492e-01 1.05420142e-01 1.05521679e+00 2.47547343e-01 -4.80867445e-01 -9.71479654e-01 -1.28180504e+00 -1.43779367e-01 1.27877235e+00 2.04618856e-01 -1.64041609e-01 -5.83879173e-01 -4.02480304e-01 2.21780106e-01 3.39075536e-01 -5.47417402e-01 1.26566529e-01 -4.00681734e-01 -7.86342680e-01 -1.01417840e-01 5.24173737e-01 3.95612240e-01 -5.34932733e-01 -5.57392359e-01 -1.95462123e-01 -1.10514045e-01 -1.60738587e+00 -4.77577567e-01 5.84361553e-02 -1.04570985e+00 -1.39173150e+00 -5.40313840e-01 -3.43446255e-01 9.70607221e-01 1.30978012e+00 1.37590790e+00 -4.83336821e-02 -4.93796945e-01 1.12111676e+00 -3.73467177e-01 -3.29143226e-01 -1.23506211e-01 -2.38771081e-01 3.01909354e-02 4.79417086e-01 1.86293423e-02 -6.54375494e-01 -7.55439639e-01 6.90619409e-01 -1.09186232e+00 3.44923645e-01 5.85518122e-01 6.87169313e-01 6.70927584e-01 -1.77910522e-01 -3.28149162e-02 -1.07198620e+00 1.70090690e-01 7.88419470e-02 -8.70254397e-01 4.22869384e-01 -7.37543344e-01 -3.32124650e-01 5.40236294e-01 -2.76262850e-01 -1.38073027e+00 2.54745245e-01 3.28845799e-01 -4.80592847e-01 -3.20361733e-01 -1.59142241e-01 -3.06780070e-01 -4.13708597e-01 6.85435832e-01 1.19999230e-01 -5.44732483e-03 -5.09716392e-01 7.61696577e-01 3.93381715e-01 4.06910151e-01 -5.64348280e-01 9.69959676e-01 1.02058625e+00 -1.99550167e-02 -1.08907259e+00 -9.95520830e-01 -8.55209947e-01 -9.34794784e-01 -3.46818238e-01 6.42299652e-01 -1.24669254e+00 -7.40191221e-01 6.06897593e-01 -9.76391613e-01 -2.67795503e-01 -1.11426421e-01 2.68137932e-01 -6.15722716e-01 4.59016800e-01 -5.17837048e-01 -7.34541059e-01 -9.89837497e-02 -1.01060295e+00 1.33549380e+00 1.50971591e-01 5.33842444e-01 -9.36762869e-01 -1.03315197e-01 7.75727570e-01 2.56460160e-01 1.41022637e-01 7.87121713e-01 1.80990428e-01 -8.78851116e-01 1.46554202e-01 -6.12677276e-01 7.37958252e-01 3.54697347e-01 4.46474731e-01 -1.38491559e+00 -4.78311449e-01 8.94729495e-02 -2.85894394e-01 1.12770510e+00 4.73522305e-01 1.05843198e+00 2.37787262e-01 -1.33513585e-01 9.31746244e-01 1.62118113e+00 -2.06691027e-01 6.44925594e-01 2.68485993e-01 1.05564642e+00 7.61187732e-01 5.92030764e-01 5.27605891e-01 3.12718362e-01 9.01559055e-01 6.30287468e-01 -5.84176719e-01 -3.07624459e-01 1.31316140e-01 3.58368397e-01 6.93806708e-01 -3.77853692e-01 -5.74667715e-02 -4.98186767e-01 3.53445083e-01 -1.62521291e+00 -8.89938653e-01 -2.92028457e-01 2.14386225e+00 4.74096626e-01 -1.84273675e-01 1.50046170e-01 -9.71527696e-02 3.49216580e-01 3.73321891e-01 -4.69610900e-01 -9.74268168e-02 -5.86245298e-01 -4.68345024e-02 6.32031977e-01 4.10018831e-01 -1.07406437e+00 8.53356719e-01 7.19949913e+00 2.82103509e-01 -1.10117388e+00 -1.32993728e-01 5.31220198e-01 -8.76584575e-02 -5.70774376e-01 -9.39281583e-02 -8.06707799e-01 3.92610617e-02 1.70372859e-01 5.80145776e-01 7.15290725e-01 7.45680392e-01 2.01776102e-01 -3.05798084e-01 -9.56601441e-01 1.07116032e+00 5.29244781e-01 -1.01804638e+00 -6.24414012e-02 2.21505929e-02 1.10835004e+00 1.73118517e-01 4.73329067e-01 -4.75061178e-01 2.69094765e-01 -6.01212204e-01 5.95425427e-01 2.40679041e-01 6.87881172e-01 -3.86207074e-01 2.12988153e-01 -1.42703131e-01 -1.11832106e+00 -2.27473751e-01 -4.42561328e-01 2.09366828e-01 7.67592266e-02 8.85309219e-01 -6.07197523e-01 8.82687569e-01 7.31282234e-01 1.06852949e+00 -6.93246782e-01 6.22690260e-01 -3.09978902e-01 1.80935547e-01 -2.79688001e-01 5.73044240e-01 -2.10053205e-01 -6.35202169e-01 3.64647985e-01 1.02265978e+00 -1.35452598e-01 -4.66284305e-02 2.61917830e-01 8.63280237e-01 1.63878262e-01 3.88787873e-02 -7.66720355e-01 1.98810399e-01 -2.09374297e-02 1.63113487e+00 -7.35379100e-01 1.29979879e-01 -7.12538064e-01 1.03093958e+00 3.49515736e-01 9.04125214e-01 -6.46221638e-01 1.96450025e-01 5.90668321e-01 1.72554612e-01 6.65064156e-01 -4.36394185e-01 -4.14994806e-01 -1.45713067e+00 2.57047355e-01 -9.53027546e-01 1.26804173e-01 -1.27402151e+00 -1.39121139e+00 2.47857004e-01 -1.69074405e-02 -1.42607009e+00 1.60945058e-01 -1.22931933e+00 -3.48461211e-01 6.73022389e-01 -2.04142451e+00 -1.60710919e+00 -5.65756738e-01 6.33062124e-01 3.57780397e-01 -1.07103236e-01 6.42758548e-01 1.41163141e-01 -3.21260571e-01 2.28594780e-01 3.36498886e-01 -1.91773981e-01 1.00797296e+00 -1.26430035e+00 6.03983365e-02 9.48360145e-01 4.81605768e-01 6.99264526e-01 4.59319085e-01 -1.48271516e-01 -1.76002765e+00 -7.49323964e-01 -8.74395669e-02 -6.56556189e-01 3.55067343e-01 -3.69492143e-01 -5.50134361e-01 5.77151000e-01 1.56803727e-01 2.28761777e-01 8.16939414e-01 3.83413941e-01 -9.06544864e-01 -5.79158962e-01 -7.73480177e-01 3.72152835e-01 8.98010731e-01 -7.64135718e-01 -1.77321225e-01 3.56848747e-01 1.63913980e-01 -5.86501360e-01 -9.12731469e-01 3.43818724e-01 8.59325290e-01 -1.65505195e+00 1.33322001e+00 -3.92096341e-01 5.16886353e-01 -4.59039986e-01 -5.01077116e-01 -1.34205449e+00 -3.39973897e-01 -4.62158501e-01 7.22281337e-02 1.28464162e+00 2.76941985e-01 -6.68406010e-01 6.28340006e-01 4.22599256e-01 -6.15459830e-02 -5.56841373e-01 -2.94599384e-01 -4.61472631e-01 -3.59038681e-01 -2.35360965e-01 1.78013146e-01 7.87569106e-01 -4.59257364e-01 2.68519223e-01 -3.83963436e-01 3.74590486e-01 8.65125120e-01 7.00815916e-01 1.19223547e+00 -1.17647707e+00 -5.13131797e-01 -3.11209410e-01 -1.68224312e-02 -1.10235274e+00 -1.02673799e-01 -5.33106923e-01 -7.36803114e-02 -1.29134083e+00 4.42065716e-01 -2.89733857e-01 -3.23832422e-01 1.97200015e-01 -1.80808678e-01 4.66493070e-01 3.34235221e-01 2.60750741e-01 -5.21408856e-01 3.75399888e-01 1.37205732e+00 -6.06314875e-02 -1.49072334e-01 -8.62944522e-04 -8.25265110e-01 8.03024232e-01 5.26232541e-01 3.44932973e-02 -4.52866584e-01 -6.82946444e-01 4.72117752e-01 -3.75620425e-01 3.57914388e-01 -5.40726244e-01 -1.55780479e-01 -3.36994529e-01 5.77852011e-01 -7.02384293e-01 5.53266108e-01 -1.01741850e+00 -8.06100294e-02 -1.26462385e-01 -8.34519267e-02 -6.75729290e-02 1.33321300e-01 7.06540227e-01 -1.00171626e-01 2.05424130e-01 9.43385184e-01 -2.41248742e-01 -5.62040806e-01 5.20972461e-02 1.47311628e-01 -8.22975039e-02 6.64623260e-01 -3.88952494e-01 -5.10696173e-01 -2.64251202e-01 -2.65601903e-01 -4.35081571e-02 9.28545773e-01 2.97321647e-01 6.16130471e-01 -1.15349948e+00 -4.83760715e-01 4.98304307e-01 3.80059063e-01 -1.32667571e-01 4.44494426e-01 9.09716606e-01 -3.42277676e-01 1.71874359e-01 -3.22833449e-01 -9.26749349e-01 -1.55122972e+00 5.73354602e-01 2.04786733e-01 -1.85133487e-01 -5.14777064e-01 6.27300441e-01 6.45810604e-01 -3.52998883e-01 8.80043674e-03 -4.85510200e-01 -7.50118494e-02 -2.85804067e-02 5.83665192e-01 2.72384971e-01 2.62432069e-01 -6.44721627e-01 -1.04680657e-01 1.09468842e+00 -1.33495718e-01 9.59869474e-02 1.50151265e+00 -4.92928833e-01 -2.86174625e-01 4.58269089e-01 1.02744651e+00 4.39078093e-01 -1.55816054e+00 -2.87298232e-01 -6.55330837e-01 -8.61126125e-01 1.04751319e-01 -6.90968573e-01 -1.17937171e+00 9.69434977e-01 5.56588948e-01 2.59932965e-01 1.45990860e+00 -2.17195272e-01 3.57713133e-01 3.76777738e-01 2.24151343e-01 -1.18047273e+00 5.69091260e-01 3.01830471e-01 7.99028158e-01 -1.52987182e+00 5.19917965e-01 -9.89402533e-01 -7.12504447e-01 1.35265768e+00 5.33049405e-01 8.28564093e-02 4.17099297e-01 1.31590635e-01 3.01704347e-01 -2.63095468e-01 -5.43745995e-01 -5.16102850e-01 6.20188951e-01 7.63572216e-01 3.90417993e-01 -2.13552907e-01 3.24101955e-01 6.53086677e-02 3.16209316e-01 -5.20318210e-01 4.69980448e-01 5.91139734e-01 -3.39610696e-01 -1.03003585e+00 -3.61248583e-01 -2.10767984e-02 -3.86023879e-01 -7.88548738e-02 -3.83654743e-01 6.69932365e-01 -6.56411275e-02 8.73283327e-01 -1.59742460e-01 -2.28356093e-01 5.17922163e-01 -9.32262987e-02 9.72511828e-01 -6.10758245e-01 -3.63595337e-01 5.25860906e-01 -4.59601805e-02 -7.18785644e-01 -8.87153268e-01 -5.20854533e-01 -6.61803484e-01 1.27545297e-01 -5.89382887e-01 -5.11175811e-01 5.89281261e-01 6.46453142e-01 3.02405030e-01 4.74385470e-01 1.13074934e+00 -1.27915764e+00 -2.59284556e-01 -5.52547216e-01 -8.85470688e-01 9.01031315e-01 3.48375618e-01 -8.05214405e-01 -5.00926137e-01 3.87702733e-01]
[9.92387580871582, -2.8187363147735596]
084ce3a1-db39-40ce-89a7-5cfdd50ef071
distributed-layer-partitioned-training-for
1904.06049
null
http://arxiv.org/abs/1904.06049v1
http://arxiv.org/pdf/1904.06049v1.pdf
Distributed Layer-Partitioned Training for Privacy-Preserved Deep Learning
Deep Learning techniques have achieved remarkable results in many domains. Often, training deep learning models requires large datasets, which may require sensitive information to be uploaded to the cloud to accelerate training. To adequately protect sensitive information, we propose distributed layer-partitioned training with step-wise activation functions for privacy-preserving deep learning. Experimental results attest our method to be simple and effective.
['Chun-Nan Chou', 'Chun-Hsien Yu', 'Emily Chang']
2019-04-12
null
null
null
null
['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning']
['methodology', 'natural-language-processing']
[-1.56509221e-01 -1.45478174e-01 -4.08421814e-01 -1.00936842e+00 -6.52558446e-01 -5.75101614e-01 2.71324050e-02 -3.62785570e-02 -7.57010341e-01 9.56492841e-01 -9.46231857e-02 -3.80411029e-01 -4.95191254e-02 -8.50349128e-01 -8.77890825e-01 -9.09447730e-01 -2.21534535e-01 1.51862174e-01 -6.46793982e-03 2.15581104e-01 -3.22371095e-01 5.48414409e-01 -9.98164892e-01 3.18010986e-01 6.66374207e-01 1.18804193e+00 -2.71126121e-01 1.35155782e-01 1.17458194e-01 7.35104501e-01 -7.66783118e-01 -8.29261363e-01 6.67566895e-01 2.02121079e-01 -8.49714220e-01 -4.65010256e-01 3.65645945e-01 -7.88491249e-01 -5.07016897e-01 1.37372184e+00 6.62600815e-01 3.74630801e-02 -1.57247201e-01 -1.33072615e+00 -5.61410069e-01 6.02555037e-01 -5.09350300e-01 1.78503260e-01 -4.43809718e-01 -6.94920868e-02 6.29073620e-01 -4.97341305e-01 2.54249096e-01 7.32650220e-01 6.94045067e-01 1.12555063e+00 -8.79849732e-01 -1.42085624e+00 -2.82979254e-02 2.19716579e-01 -1.31599927e+00 -4.00774211e-01 6.38568997e-01 -3.27607407e-03 6.20826125e-01 5.10291636e-01 3.41193914e-01 1.05473614e+00 4.78014462e-02 9.52852547e-01 9.77274537e-01 1.05035596e-01 3.55110824e-01 6.18873596e-01 -4.38771285e-02 4.98535156e-01 5.63881457e-01 8.53866264e-02 -3.72559637e-01 -3.48930418e-01 4.29339021e-01 5.45831978e-01 -1.01339623e-01 -4.00380671e-01 -3.98540705e-01 1.05940127e+00 5.64030111e-01 3.02784350e-02 -1.52396098e-01 1.16248637e-01 9.60423648e-01 5.72592854e-01 8.35368991e-01 2.43583724e-01 -7.31192052e-01 1.33578897e-01 -8.15370262e-01 2.07629830e-01 5.69948554e-01 9.23836648e-01 7.21680284e-01 -1.11712083e-01 9.96421184e-03 6.48099363e-01 -1.19595490e-01 2.76328996e-02 4.89124238e-01 -8.05411398e-01 6.57603502e-01 4.48868990e-01 -2.30230629e-01 -7.56404817e-01 -1.28909424e-01 -4.76507217e-01 -1.26734829e+00 -9.09918337e-04 -4.25436646e-02 -7.11006939e-01 -5.76487243e-01 1.49715161e+00 4.68080014e-01 2.16609359e-01 3.03421527e-01 1.01944685e+00 4.59547251e-01 5.88425875e-01 3.19425642e-01 9.89303226e-04 1.06871414e+00 -8.23049247e-01 -6.31497264e-01 -7.34655093e-03 8.60978246e-01 5.14670927e-03 1.00127125e+00 1.90140635e-01 -9.33060765e-01 -7.87981600e-02 -9.91288602e-01 -3.33358586e-01 -4.01461959e-01 -1.33326992e-01 9.75042045e-01 1.03169441e+00 -1.03924441e+00 4.78396237e-01 -9.10353422e-01 3.30300242e-01 1.54046416e+00 7.10189939e-01 -6.37316406e-01 -2.52538174e-01 -1.44738877e+00 1.65004432e-01 7.61666656e-01 4.98602428e-02 -8.37408066e-01 -1.03185308e+00 -7.10040927e-01 2.71081716e-01 1.38624981e-02 -4.25956130e-01 1.13746798e+00 -7.56955624e-01 -1.32115555e+00 9.18577433e-01 3.18415016e-01 -8.86655152e-01 7.71905482e-01 -3.30865771e-01 -3.03752810e-01 1.59099624e-01 -3.50280941e-01 2.49828488e-01 7.02118278e-01 -8.72940779e-01 -7.14469433e-01 -7.08373547e-01 -1.38515949e-01 -1.47946224e-01 -1.44639790e+00 3.97122204e-01 -1.29139706e-01 -5.08746147e-01 -3.27463478e-01 -5.49352288e-01 -4.41702604e-01 2.83492714e-01 -3.50934714e-01 -2.33220086e-01 1.40380406e+00 -7.13280141e-01 9.59749877e-01 -2.35142469e+00 -5.84702730e-01 3.77966911e-01 5.49437404e-01 8.61401916e-01 1.14650704e-01 -6.71672001e-02 -6.35063723e-02 1.75420240e-01 -3.13275516e-01 -6.76778436e-01 2.01730490e-01 4.37010437e-01 -6.76286578e-01 6.68867826e-01 -5.02232946e-02 9.15902972e-01 -4.83978778e-01 -5.12799919e-01 6.78915344e-03 6.98186576e-01 -6.13736629e-01 3.69792491e-01 -6.51604757e-02 1.77750602e-01 -7.72249222e-01 3.81529212e-01 1.20093727e+00 4.37579380e-04 -6.09084405e-02 4.14373666e-01 3.71954143e-01 1.37926459e-01 -4.13586348e-01 1.39292943e+00 -3.51285785e-01 5.26338935e-01 3.80843788e-01 -1.12942934e+00 9.43645537e-01 4.25109774e-01 5.21798193e-01 -6.56534255e-01 2.22209752e-01 -1.53573841e-01 -6.81307137e-01 -2.88591355e-01 1.90777034e-01 -1.21246502e-01 -1.50645241e-01 3.86457652e-01 -1.60839185e-01 4.36915398e-01 -7.18836010e-01 1.04248814e-01 9.15097117e-01 -7.12585568e-01 -3.64908725e-01 -2.31301785e-01 3.75177801e-01 -3.75704616e-01 9.01683271e-01 3.70016605e-01 -4.81369019e-01 2.33219922e-01 4.05534118e-01 -9.71129596e-01 -8.17791462e-01 -6.35533392e-01 -2.36105233e-01 1.36888957e+00 -2.29674518e-01 -2.90804416e-01 -1.22821891e+00 -1.06241190e+00 1.48452342e-01 3.95293713e-01 -6.62315607e-01 -5.70952773e-01 -3.83943766e-01 -7.29522407e-01 8.53067160e-01 6.31236494e-01 9.53151584e-01 -7.00629473e-01 -6.77057922e-01 -1.71665445e-01 2.27891698e-01 -8.40002060e-01 -4.93133545e-01 2.82849938e-01 -9.40105498e-01 -6.78116620e-01 -6.78404212e-01 -6.76021695e-01 8.90719354e-01 2.65218675e-01 6.22437477e-01 -1.68727990e-02 -1.77008420e-01 -6.59610480e-02 4.13648635e-02 -7.63501227e-01 1.10292383e-01 2.80370206e-01 1.74236178e-01 1.58081159e-01 7.14456081e-01 -6.42727792e-01 -5.43457806e-01 1.36791408e-01 -1.07810771e+00 -4.32717353e-01 2.30277076e-01 6.78455472e-01 6.44779623e-01 4.61130619e-01 4.31542248e-01 -1.29500449e+00 7.03026414e-01 -6.01949275e-01 -7.63616621e-01 3.13608676e-01 -7.77990341e-01 -1.60373375e-01 9.77629840e-01 -2.97612518e-01 -8.96597207e-01 1.36254638e-01 -1.18448928e-01 -1.05758929e+00 -1.24140643e-01 2.90553659e-01 -4.99753296e-01 -4.09238309e-01 4.64749277e-01 2.21040353e-01 -7.35673755e-02 -6.47737205e-01 -2.21317145e-03 6.38284922e-01 4.92890596e-01 -5.62303543e-01 8.40213120e-01 3.50465000e-01 -2.34338731e-01 -2.10125670e-01 -6.69992208e-01 -6.74643815e-02 -1.77781209e-01 2.96618223e-01 7.30862498e-01 -1.14310324e+00 -8.97695482e-01 5.88566840e-01 -8.09803128e-01 -3.16910833e-01 -3.80310118e-01 3.52048934e-01 1.78239681e-02 -1.66264549e-02 -7.97487140e-01 -7.49485970e-01 -9.01391149e-01 -7.21054196e-01 4.42373604e-01 3.92260522e-01 3.87171835e-01 -1.16371500e+00 -4.11212370e-02 5.55689812e-01 6.37151837e-01 3.02722484e-01 5.25951266e-01 -9.91198838e-01 -5.17266333e-01 -7.00570107e-01 -2.07130402e-01 7.01007426e-01 -8.49649534e-02 -3.75468552e-01 -1.32307279e+00 -6.64433181e-01 2.48323411e-01 -7.84354866e-01 9.19731438e-01 1.03356212e-01 2.40731525e+00 -1.10317969e+00 -3.21055621e-01 1.40952134e+00 1.25549531e+00 -1.90191895e-01 5.40596724e-01 2.81724274e-01 9.61809218e-01 5.55057585e-01 2.21901134e-01 6.58118904e-01 3.17820996e-01 2.07062826e-01 5.46197355e-01 -2.82933384e-01 5.09426057e-01 -4.78564262e-01 1.06578879e-01 3.12889814e-01 2.95172155e-01 1.14384554e-01 -7.17562973e-01 4.34890240e-01 -1.65772569e+00 -8.82136941e-01 3.03069562e-01 2.11827350e+00 1.15078342e+00 -1.04565904e-01 1.47321507e-01 -1.61716104e-01 4.72797692e-01 2.35897809e-01 -9.20721054e-01 -5.67966044e-01 -1.83473863e-02 3.84110868e-01 9.41939235e-01 4.12872992e-02 -1.30547190e+00 9.05607760e-01 6.74134016e+00 8.41818094e-01 -1.39542568e+00 4.71740067e-01 1.17581308e+00 -7.76252866e-01 -5.45899510e-01 -6.29023433e-01 -7.19648063e-01 6.27124488e-01 1.05710685e+00 -6.15107954e-01 1.47516608e-01 1.42279506e+00 -2.24322647e-01 6.37567878e-01 -8.29714775e-01 9.31943834e-01 -5.04092157e-01 -1.34328020e+00 -2.29809895e-01 3.22211415e-01 8.52182567e-01 2.39654765e-01 5.04635572e-01 2.86600530e-01 7.25248754e-01 -1.14583635e+00 2.64264584e-01 7.33871609e-02 9.57367122e-01 -1.47064471e+00 7.18985319e-01 4.82811272e-01 -7.59846091e-01 -2.97798246e-01 -1.02215457e+00 9.14700553e-02 -5.91740310e-01 6.07340336e-01 -7.06063271e-01 2.82086343e-01 1.18926179e+00 5.73991716e-01 -3.96040589e-01 7.52814472e-01 -1.61379073e-02 8.67494583e-01 -3.43585134e-01 -5.94146773e-02 2.22365841e-01 -6.64664656e-02 -2.34876852e-02 9.20478523e-01 8.38961303e-02 1.36899158e-01 8.18513185e-02 7.35569119e-01 -7.90130138e-01 1.73237361e-02 -8.11723709e-01 8.34035128e-02 6.08348429e-01 1.16343570e+00 4.27036472e-02 -2.91703586e-02 -1.90225109e-01 9.74350393e-01 5.76215625e-01 1.95252091e-01 -6.68583333e-01 -4.76915300e-01 1.28940582e+00 -1.10586539e-01 1.92543402e-01 1.08524852e-01 -5.14984488e-01 -1.07439017e+00 3.36838752e-01 -4.87222165e-01 9.24368441e-01 -1.45429835e-01 -1.49053133e+00 5.66745818e-01 -3.97706658e-01 -9.32977080e-01 -4.48779613e-02 -3.45555693e-01 -6.02993309e-01 8.63007605e-01 -1.58222449e+00 -1.12975621e+00 8.93361419e-02 1.39851701e+00 -2.00488418e-01 -4.07024264e-01 9.96523559e-01 6.81869924e-01 -1.02503133e+00 1.79508758e+00 4.52511936e-01 5.79534948e-01 6.37456834e-01 -9.13968146e-01 3.42541367e-01 9.88789320e-01 3.26034389e-02 7.15148509e-01 3.33118409e-01 -2.90285289e-01 -1.45508265e+00 -1.75235403e+00 7.52614498e-01 -3.07550043e-01 3.22471380e-01 -7.18548238e-01 -1.25885952e+00 9.04476583e-01 5.41495485e-03 8.08016419e-01 1.15419197e+00 7.59093314e-02 -9.06829655e-01 -6.81025207e-01 -1.70757341e+00 4.46634032e-02 6.85851157e-01 -8.37102592e-01 8.46112221e-02 5.52089453e-01 1.12316644e+00 -4.94899154e-01 -9.35023904e-01 1.53556228e-01 2.17426643e-01 -7.34667420e-01 9.07189906e-01 -1.13996208e+00 1.91666275e-01 2.87033349e-01 4.24016714e-02 -9.56847072e-01 -1.66591361e-01 -8.25820327e-01 -3.31888735e-01 1.17312014e+00 2.77359754e-01 -1.05760050e+00 1.46300244e+00 1.52720046e+00 7.00561106e-02 -7.80264318e-01 -1.16536236e+00 -6.39639258e-01 4.21967208e-01 -9.73447263e-02 1.13790691e+00 1.56933951e+00 -1.64495841e-01 -3.46355826e-01 -6.14632666e-01 4.13828999e-01 7.48001099e-01 -1.29116312e-01 5.91490507e-01 -1.14011490e+00 8.80539641e-02 -1.46383852e-01 -3.29556346e-01 -2.82026082e-01 6.17920101e-01 -8.10590804e-01 -2.68829256e-01 -9.22664702e-01 1.30107939e-01 -7.97558486e-01 -9.73039985e-01 1.28108716e+00 -1.58665329e-01 2.84677237e-01 -2.00901851e-01 2.66222954e-02 -7.16059685e-01 6.59460962e-01 7.25126386e-01 -1.20032594e-01 7.07071275e-02 3.84515703e-01 -1.04323184e+00 3.19791615e-01 1.18727863e+00 -6.99185789e-01 -6.07908070e-01 -7.62865722e-01 2.15600221e-03 -3.15191388e-01 2.88984358e-01 -9.10859168e-01 3.96205604e-01 -1.75484210e-01 5.94924152e-01 -1.94315478e-01 9.46183577e-02 -1.33930123e+00 8.11423920e-03 5.45229018e-01 -6.63331270e-01 -1.83157906e-01 2.90445566e-01 4.07014668e-01 -2.71619022e-01 2.58141488e-01 9.05401230e-01 2.52455473e-01 -5.48609972e-01 1.26872253e+00 2.40955308e-01 -2.48705801e-02 1.27138472e+00 1.11048080e-01 -3.58368307e-01 -1.61439613e-01 -2.09156513e-01 5.45596957e-01 4.54380393e-01 3.96989256e-01 7.57398248e-01 -1.37374365e+00 -7.10743487e-01 5.75504661e-01 1.06157251e-01 3.83176297e-01 5.53316534e-01 8.92242193e-02 -3.36816281e-01 3.27557504e-01 -3.50032359e-01 -5.02705276e-02 -1.52770388e+00 8.68581414e-01 4.11235273e-01 -4.19274122e-02 -7.18194127e-01 1.55009890e+00 1.30839169e-01 -5.29416144e-01 6.19015276e-01 6.71414752e-03 2.20943302e-01 -2.96564966e-01 1.16819024e+00 6.88045099e-02 1.19349048e-01 1.72328819e-02 -4.42578614e-01 -1.27971441e-01 -5.46144485e-01 2.70809978e-01 1.63917398e+00 1.98821992e-01 -2.14083850e-01 -2.57499635e-01 1.90374792e+00 -2.88475305e-01 -1.64166737e+00 -5.33943832e-01 -3.58251393e-01 -7.24049330e-01 4.09366667e-01 -5.71856201e-01 -1.93710268e+00 1.10973430e+00 7.44483650e-01 -1.15141869e-01 1.44751060e+00 -2.48752221e-01 1.23639429e+00 6.11816883e-01 5.78056335e-01 -1.09585428e+00 -5.39398193e-01 8.64229202e-02 3.90893638e-01 -1.22273791e+00 -1.18593045e-01 -2.11750623e-02 -4.84859079e-01 7.14353561e-01 7.39206851e-01 -7.27371720e-04 9.27478909e-01 5.90467751e-01 1.86457276e-01 8.04855525e-02 -7.52634346e-01 6.49163663e-01 -9.05218348e-02 6.41618788e-01 -1.13954768e-01 1.48763984e-01 1.45523772e-01 1.05428290e+00 -2.74101257e-01 2.37714723e-02 -1.69525146e-01 8.35407972e-01 -1.84680834e-01 -1.21358109e+00 1.06940670e-02 3.71052712e-01 -1.02297568e+00 -5.11764549e-03 -3.97737771e-01 2.43037626e-01 -2.10777316e-02 4.61273134e-01 1.21732689e-01 -7.38993704e-01 1.54749662e-01 -4.93889041e-02 -9.83981863e-02 -1.89311534e-01 -9.30168271e-01 -7.31676400e-01 -2.98072428e-01 -7.19252229e-01 1.90188348e-01 -4.25175369e-01 -1.22917151e+00 -8.38582218e-01 9.40304548e-02 4.09661889e-01 7.01964915e-01 5.41998386e-01 7.97944129e-01 1.47939131e-01 1.13476741e+00 -1.97111055e-01 -8.40608776e-01 -4.34363127e-01 -7.32737243e-01 2.35428110e-01 5.39244115e-01 7.54975975e-02 -1.85580805e-01 -3.97600979e-01]
[5.87939977645874, 6.853515148162842]
f2778e4b-11bc-47a4-afa3-3fee95144cce
real-time-monocular-visual-odometry-for
1806.05842
null
https://arxiv.org/abs/1806.05842v3
https://arxiv.org/pdf/1806.05842v3.pdf
Real-time Monocular Visual Odometry for Turbid and Dynamic Underwater Environments
In the context of robotic underwater operations, the visual degradations induced by the medium properties make difficult the exclusive use of cameras for localization purpose. Hence, most localization methods are based on expensive navigational sensors associated with acoustic positioning. On the other hand, visual odometry and visual SLAM have been exhaustively studied for aerial or terrestrial applications, but state-of-the-art algorithms fail underwater. In this paper we tackle the problem of using a simple low-cost camera for underwater localization and propose a new monocular visual odometry method dedicated to the underwater environment. We evaluate different tracking methods and show that optical flow based tracking is more suited to underwater images than classical approaches based on descriptors. We also propose a keyframe-based visual odometry approach highly relying on nonlinear optimization. The proposed algorithm has been assessed on both simulated and real underwater datasets and outperforms state-of-the-art visual SLAM methods under many of the most challenging conditions. The main application of this work is the localization of Remotely Operated Vehicles (ROVs) used for underwater archaeological missions but the developed system can be used in any other applications as long as visual information is available.
['Pauline Trouvé-Peloux', 'Julien Moras', 'Maxime Ferrera', 'Vincent Creuze']
2018-06-15
null
null
null
null
['monocular-visual-odometry']
['robots']
[-1.25596568e-01 -2.02148240e-02 4.37228233e-01 -2.25687638e-01 -2.72837222e-01 -4.62029308e-01 5.20506203e-01 2.30521917e-01 -1.38300288e+00 7.49053121e-01 -3.82551342e-01 9.88638178e-02 -3.31308961e-01 -6.45920753e-01 -5.92829168e-01 -8.73582900e-01 -2.81849205e-01 6.09980524e-01 6.42473221e-01 -5.34110904e-01 5.37092865e-01 6.34156823e-01 -1.82336736e+00 -6.80273831e-01 6.04595304e-01 6.36192679e-01 8.46324325e-01 7.31417060e-01 -1.33838624e-01 2.13900134e-01 -2.88888425e-01 1.27588985e-02 3.95583510e-01 -3.64433795e-01 -2.22830415e-01 -1.88997954e-01 5.90040505e-01 -4.31813300e-01 -2.06013843e-01 1.14073384e+00 8.58582675e-01 4.63089287e-01 5.88342845e-01 -7.79046178e-01 -2.09953673e-02 1.88903615e-01 -2.40063239e-02 3.99235822e-02 4.07475531e-01 -1.53815210e-01 7.80056953e-01 -1.01140523e+00 7.56994247e-01 7.90471792e-01 8.35000098e-01 4.34056252e-01 -9.10428822e-01 -7.30053261e-02 -2.68247694e-01 4.05642211e-01 -1.36333859e+00 -4.65880901e-01 3.47345024e-01 -5.39501667e-01 5.90533018e-01 -1.85652003e-01 7.94265807e-01 4.89742041e-01 3.52482319e-01 6.88734874e-02 1.10142839e+00 -4.40162599e-01 4.45867062e-01 1.45780779e-02 -1.84961006e-01 8.00176859e-01 7.16807365e-01 -1.98110998e-01 -5.97714663e-01 6.03012517e-02 2.86822259e-01 2.21305847e-01 -7.30047762e-01 -6.86189294e-01 -8.46332073e-01 8.10688794e-01 4.22343969e-01 3.36175174e-01 -3.38649780e-01 2.57320583e-01 2.83182651e-01 3.19370687e-01 1.24973655e-01 4.08976495e-01 -4.94544655e-02 -1.03575855e-01 -9.73260760e-01 9.43059102e-02 1.05106819e+00 7.20145345e-01 1.15365195e+00 3.50608557e-01 7.08856523e-01 5.85993946e-01 8.12830627e-01 1.04904413e+00 6.46022499e-01 -6.63154304e-01 1.40637681e-01 1.15999430e-01 2.37942770e-01 -9.73457575e-01 -7.82319307e-01 -1.42140597e-01 -3.64678621e-01 6.40175521e-01 4.22780812e-01 -1.39282614e-01 -7.43770003e-01 1.17473912e+00 2.19822451e-01 1.09503560e-01 7.23928988e-01 1.06703973e+00 8.44816148e-01 4.94048119e-01 -4.72747356e-01 -2.88316429e-01 1.14693737e+00 -4.96728718e-01 -8.19816709e-01 -3.63131642e-01 5.61047912e-01 -6.55228317e-01 2.74079919e-01 2.61888891e-01 -6.43291175e-01 -1.13962926e-01 -1.33482027e+00 1.65966362e-01 -4.31076169e-01 -6.39165044e-02 1.44560635e-01 6.21645689e-01 -1.25458097e+00 5.46388388e-01 -1.07536721e+00 -8.64113688e-01 -3.76809031e-01 4.75379795e-01 -7.10262597e-01 -2.59037584e-01 -9.39267874e-01 1.25732744e+00 2.89335459e-01 5.37879288e-01 -1.13350260e+00 -8.75554234e-02 -1.16907549e+00 -2.80212790e-01 9.92076099e-02 -3.03091943e-01 9.49635208e-01 -6.31578565e-01 -1.68211877e+00 6.95752203e-01 -2.10298505e-02 -5.41843235e-01 7.47176647e-01 -2.82210976e-01 5.91745712e-02 4.51147586e-01 1.40075222e-01 1.50492460e-01 5.77442110e-01 -1.46026921e+00 -7.59592175e-01 -4.29422051e-01 1.41839776e-02 5.18263400e-01 -2.62861460e-01 -5.78969598e-01 -3.56108457e-01 6.89979568e-02 6.69416845e-01 -1.11553574e+00 -1.06774598e-01 2.91927457e-01 1.20057762e-01 3.33002687e-01 8.21891725e-01 -4.31881249e-01 5.35907149e-01 -1.89925611e+00 3.97421658e-01 -5.73892798e-03 -3.34207296e-01 1.72887012e-01 2.77105331e-01 9.03050005e-01 6.57999635e-01 -3.95089805e-01 -3.22863996e-01 -8.93745542e-01 -9.76118520e-02 5.51964581e-01 2.00724348e-01 1.37580121e+00 -6.76236033e-01 2.30898354e-02 -1.04411793e+00 -6.17963016e-01 4.86331075e-01 5.10780036e-01 -2.47477785e-01 3.89006853e-01 3.00339729e-01 6.38439953e-01 -1.34803906e-01 7.15529025e-01 8.16824794e-01 6.54882550e-01 2.49709174e-01 -2.38612164e-02 -9.33952510e-01 -3.43727142e-01 -1.38539183e+00 1.87539411e+00 -7.44976342e-01 9.15380061e-01 5.99308372e-01 -7.51027822e-01 1.21950531e+00 1.52755320e-01 3.95222515e-01 -4.48195219e-01 8.80712792e-02 8.31763268e-01 -2.75745481e-01 -8.76639247e-01 8.73482049e-01 -4.98443007e-01 1.15415014e-01 -7.59532815e-03 1.37527406e-01 -3.19022655e-01 1.85325041e-01 -1.68306455e-01 8.64300787e-01 4.24430281e-01 5.49102068e-01 -6.63332164e-01 5.61787367e-01 3.36757377e-02 4.19696718e-01 7.17713296e-01 1.86563917e-02 5.97541213e-01 -1.74223468e-01 -5.11528969e-01 -8.93084407e-01 -6.72192514e-01 -3.65236402e-01 6.71045184e-01 8.46359253e-01 5.70884123e-02 -2.30572090e-01 -1.57089591e-01 1.94136724e-01 -8.30293670e-02 -5.46656966e-01 3.84016961e-01 -5.92683852e-01 -6.42646432e-01 5.89643657e-01 -7.25327581e-02 2.34635204e-01 -7.83959448e-01 -1.24473643e+00 4.69644755e-01 1.23207092e-01 -1.26168108e+00 4.45140570e-01 2.51479089e-01 -9.58382070e-01 -1.00129056e+00 -1.16196215e+00 -7.82058001e-01 5.93102574e-01 5.49269676e-01 4.83614832e-01 2.94636041e-01 3.92492022e-03 6.23676300e-01 -9.10719097e-01 -1.13108657e-01 -4.06240582e-01 -5.70689887e-02 5.10914207e-01 2.17062652e-01 -8.41068029e-02 -5.30244768e-01 -6.04490876e-01 4.50211495e-01 -7.92311370e-01 -7.53779173e-01 5.50943077e-01 7.72912741e-01 1.79105073e-01 -5.46144247e-01 1.34142265e-01 -4.14795995e-01 -8.93005356e-02 -4.51195270e-01 -1.08105898e+00 -1.32114366e-01 -4.22848314e-01 1.60374239e-01 5.53657830e-01 1.97658297e-02 -6.13340080e-01 4.24388677e-01 -2.69908965e-01 -2.16644391e-01 1.12354897e-01 5.86444736e-01 1.97313279e-01 -8.00358415e-01 5.51494956e-01 5.32057583e-01 2.60554403e-01 -5.13193488e-01 -2.47999430e-01 7.96800971e-01 2.99503803e-01 8.09878558e-02 1.02356482e+00 1.01812267e+00 5.75901270e-01 -1.63108456e+00 -2.13745609e-01 -9.02166128e-01 -7.12693751e-01 -5.50463557e-01 9.83816206e-01 -1.10009193e+00 -8.48266184e-01 5.50034761e-01 -1.09819698e+00 -1.20041832e-01 1.59131110e-01 9.13232446e-01 -4.38453525e-01 1.03808510e+00 -2.54314333e-01 -1.23212373e+00 -1.20619789e-01 -1.44483471e+00 9.80426431e-01 4.39754754e-01 4.77475286e-01 -1.15090156e+00 6.26093984e-01 -1.66470170e-01 3.35582584e-01 1.71381265e-01 -3.64394635e-01 -3.40418279e-01 -5.98387659e-01 -3.14758301e-01 1.59010977e-01 1.01370662e-01 -7.06064850e-02 -1.34395823e-01 -8.33841681e-01 -6.97780013e-01 -2.10662231e-01 -9.95121226e-02 9.02148366e-01 4.75592352e-02 -2.57486522e-01 2.50794768e-01 -2.91711748e-01 8.50852132e-01 1.99095976e+00 6.51996508e-02 5.33751905e-01 9.05554712e-01 6.65384948e-01 7.83130884e-01 8.83350253e-01 6.60287440e-01 4.92472649e-01 9.63824749e-01 1.16572464e+00 2.30914891e-01 4.13194597e-01 1.48537070e-01 4.86778408e-01 6.27316415e-01 -5.66494048e-01 -3.57586294e-01 -8.16942036e-01 8.29472721e-01 -1.81903911e+00 -7.15431094e-01 -5.21515131e-01 2.46846795e+00 -7.72429630e-03 -2.41742194e-01 -5.91006815e-01 4.62650172e-02 5.58665395e-01 6.26797453e-02 8.78631920e-02 -1.67187691e-01 -1.06469177e-01 -1.20245181e-01 1.32706618e+00 9.35703218e-01 -9.69776392e-01 7.03517675e-01 4.53325081e+00 5.54590933e-02 -1.13682628e+00 2.04839081e-01 -8.86937857e-01 5.23112774e-01 3.62609699e-02 3.73328328e-01 -1.10899889e+00 3.05762440e-01 5.77969491e-01 4.55585688e-01 -7.94790126e-03 8.93970788e-01 2.95055568e-01 -5.84815502e-01 -5.28546035e-01 9.71496463e-01 4.63212609e-01 -9.24462199e-01 -2.74536192e-01 2.71523476e-01 5.56352735e-01 3.42213094e-01 -5.40273309e-01 -2.19476745e-01 -3.19051862e-01 -4.32922095e-01 1.00526035e+00 7.37185419e-01 5.17602205e-01 -3.55183512e-01 1.49594879e+00 2.13203996e-01 -1.33494890e+00 -1.70433819e-01 -6.83311045e-01 -3.35042536e-01 7.77001798e-01 2.25572750e-01 -8.23553145e-01 6.78394318e-01 8.56339991e-01 7.19370306e-01 -2.13117689e-01 1.68468487e+00 -2.76231587e-01 -2.27919817e-02 -7.16119647e-01 -5.10554492e-01 6.11537457e-01 -3.11414152e-01 8.62487853e-01 1.21774900e+00 9.22698736e-01 -1.98563382e-01 1.19216040e-01 7.74359331e-02 2.72191405e-01 4.47821438e-01 -9.42459702e-01 5.05463541e-01 3.05666536e-01 1.16084015e+00 -7.04602838e-01 -4.97864895e-02 -2.44103640e-01 9.06490326e-01 -6.05589673e-02 -6.81855902e-02 -2.50657082e-01 -6.55728281e-01 5.08090019e-01 4.82456312e-02 3.60882491e-01 -8.20663631e-01 4.87218916e-01 -1.11603236e+00 -2.76374519e-01 5.90578056e-05 -1.83994137e-02 -6.64531350e-01 -6.05287671e-01 6.94330812e-01 -5.42015880e-02 -1.90234768e+00 -1.48274854e-01 -7.75245428e-01 -3.44310135e-01 6.46675646e-01 -2.02778244e+00 -9.64058340e-01 -9.42552924e-01 2.28751197e-01 4.52903152e-01 -4.17707413e-02 6.45465732e-01 3.41140419e-01 9.30005610e-02 -3.11470889e-02 6.57003343e-01 -9.58188027e-02 8.61209929e-01 -1.29139209e+00 -2.87242442e-01 1.13472927e+00 8.86544064e-02 3.64259362e-01 1.37038183e+00 -6.85432732e-01 -1.82169116e+00 -3.76173377e-01 7.18763888e-01 5.47780804e-02 7.10222721e-01 -1.61311924e-01 -7.29311705e-01 4.80034292e-01 1.90137699e-01 2.51369298e-01 2.21816286e-01 -5.75148404e-01 4.28808630e-01 -2.19349861e-01 -1.04028642e+00 5.60368635e-02 5.51468194e-01 -1.10574275e-01 -4.36031252e-01 1.76355943e-01 1.48732126e-01 -5.05709946e-01 -8.16150308e-01 4.28111583e-01 7.66177177e-01 -1.25773382e+00 6.88173592e-01 3.17740023e-01 -2.69561648e-01 -7.97429264e-01 -3.41782659e-01 -1.30882621e+00 4.77118880e-01 -5.54563701e-01 6.28208995e-01 7.84335911e-01 6.09672107e-02 -8.73244405e-01 7.00609922e-01 -3.71040910e-01 -5.32725871e-01 1.79157957e-01 -1.51324236e+00 -7.90364861e-01 -4.81840700e-01 -8.51836577e-02 -2.83997923e-01 5.43950081e-01 -1.31519884e-01 -7.39620999e-03 -7.60656774e-01 8.42454314e-01 1.00571966e+00 -6.80079609e-02 1.04316723e+00 -1.45810211e+00 -2.93171823e-01 3.73538211e-02 -1.19248855e+00 -1.02523077e+00 -1.35356691e-02 -3.62062097e-01 8.23137760e-01 -1.71035576e+00 -5.43072164e-01 -3.46858233e-01 3.50838900e-01 4.45510671e-02 4.02575672e-01 8.19881082e-01 2.53168672e-01 3.97492200e-01 -3.61834139e-01 6.29259229e-01 7.65018940e-01 1.91069767e-01 -1.84015650e-02 8.29446092e-02 4.33529228e-01 6.87425911e-01 3.41351062e-01 -6.41960979e-01 8.59687105e-02 -6.34727061e-01 4.19827402e-01 1.92942291e-01 2.86806107e-01 -1.39330089e+00 7.95343399e-01 2.89917648e-01 -2.99935788e-01 -2.07394689e-01 7.06541002e-01 -1.11091053e+00 6.13592714e-02 9.51125920e-01 4.33626026e-01 2.61837453e-01 -2.03576073e-01 8.46508026e-01 -4.57168579e-01 -1.23644817e+00 1.07090437e+00 -4.50835705e-01 -1.45247734e+00 -5.47453910e-02 -5.91952622e-01 -2.51788080e-01 9.74383533e-01 -4.61592942e-01 -2.65955716e-01 -5.72105885e-01 -7.34006822e-01 -9.90666915e-03 8.82043839e-01 -1.04430735e-01 8.47514451e-01 -4.52106237e-01 -5.83156645e-01 -8.34689885e-02 4.58118230e-01 -9.86607149e-02 3.04313064e-01 1.01554346e+00 -1.73617554e+00 -1.14594162e-01 -4.36590493e-01 -8.23257744e-01 -1.19284916e+00 8.12851116e-02 5.60601175e-01 3.75514179e-01 -7.34556794e-01 5.25901318e-01 -3.44595730e-01 -3.15946341e-01 4.73048985e-02 -8.83051157e-02 -5.68546653e-01 3.38456511e-01 3.36761981e-01 7.11837053e-01 3.84485559e-03 -1.13157856e+00 -6.13090217e-01 1.45886803e+00 6.94059193e-01 -3.78831387e-01 1.46226573e+00 -6.20857120e-01 -1.74508780e-01 4.39596742e-01 1.15707827e+00 2.43783802e-01 -1.26367676e+00 1.24413140e-01 -2.53805816e-02 -3.92817914e-01 9.39247906e-02 7.34007508e-02 -7.11516440e-01 1.01935661e+00 9.86195683e-01 3.24121833e-01 5.99861026e-01 -2.51184046e-01 3.52166176e-01 7.75397182e-01 1.04542601e+00 -8.42097938e-01 -2.90442675e-01 6.73363328e-01 4.62188661e-01 -1.34800053e+00 3.62840772e-01 -2.99789310e-02 -5.04793167e-01 1.58903682e+00 1.31084532e-01 -3.36506277e-01 5.37205696e-01 2.72700071e-01 6.05118454e-01 1.06300384e-01 7.25891590e-02 -8.37245226e-01 -4.01561886e-01 4.78117734e-01 -1.70060724e-01 -3.73502165e-01 -7.12624490e-01 -3.70609850e-01 2.52019584e-01 -3.64259928e-01 1.23065782e+00 1.29710257e+00 -9.16764140e-01 -9.59102988e-01 -6.05196834e-01 -3.58969629e-01 -4.77896541e-01 -4.64446507e-02 3.38210791e-01 1.15359306e+00 1.98552594e-01 7.51802683e-01 -1.70320272e-01 -1.61659166e-01 4.65609282e-01 -3.14620376e-01 3.22496653e-01 -4.44290280e-01 -2.67623961e-01 -1.44264638e-01 1.31340902e-02 -2.77881503e-01 -1.05473590e+00 -9.46079850e-01 -1.39270353e+00 2.95488000e-01 -4.58180308e-01 6.45555496e-01 1.51439667e+00 9.44149315e-01 -3.55250835e-01 -3.52761261e-02 4.93031293e-01 -1.61307156e+00 -3.95621568e-01 -1.07034600e+00 -9.67918932e-01 6.34194240e-02 6.29840076e-01 -1.10268009e+00 -9.00206327e-01 -1.27791196e-01]
[7.515535354614258, -1.800980567932129]
a86a7bb0-50bf-437d-95ee-acd92da71d37
pmvos-pixel-level-matching-based-video-object
2009.08855
null
https://arxiv.org/abs/2009.08855v1
https://arxiv.org/pdf/2009.08855v1.pdf
PMVOS: Pixel-Level Matching-Based Video Object Segmentation
Semi-supervised video object segmentation (VOS) aims to segment arbitrary target objects in video when the ground truth segmentation mask of the initial frame is provided. Due to this limitation of using prior knowledge about the target object, feature matching, which compares template features representing the target object with input features, is an essential step. Recently, pixel-level matching (PM), which matches every pixel in template features and input features, has been widely used for feature matching because of its high performance. However, despite its effectiveness, the information used to build the template features is limited to the initial and previous frames. We address this issue by proposing a novel method-PM-based video object segmentation (PMVOS)-that constructs strong template features containing the information of all past frames. Furthermore, we apply self-attention to the similarity maps generated from PM to capture global dependencies. On the DAVIS 2016 validation set, we achieve new state-of-the-art performance among real-time methods (> 30 fps), with a J&F score of 85.6%. Performance on the DAVIS 2017 and YouTube-VOS validation sets is also impressive, with J&F scores of 74.0% and 68.2%, respectively.
['Suhwan Cho', 'Sungjun Jang', 'Sungmin Woo', 'Sangyoun Lee', 'Heansung Lee']
2020-09-18
null
null
null
null
['one-shot-visual-object-segmentation']
['computer-vision']
[ 3.21613789e-01 -2.85136819e-01 -3.02802473e-01 -4.99172419e-01 -8.68743956e-01 -3.04153889e-01 3.56853008e-01 -5.91498427e-02 -5.47484934e-01 3.51261288e-01 -1.67946264e-01 3.76596749e-01 1.90740556e-01 -5.71205735e-01 -8.03552866e-01 -5.69682300e-01 -1.37119424e-02 8.00049528e-02 8.57210040e-01 2.19968893e-02 2.01221138e-01 4.06081796e-01 -1.76934719e+00 3.76247585e-01 9.55610454e-01 1.51764500e+00 3.61707240e-01 5.93702614e-01 -1.30301699e-01 5.44205666e-01 -5.34005165e-01 -2.86923259e-01 5.50241768e-01 -4.31315690e-01 -9.20324504e-01 5.13732135e-01 6.95792913e-01 -3.37132007e-01 -5.54012239e-01 1.00680852e+00 8.37567747e-02 1.80251852e-01 4.67226982e-01 -1.32533479e+00 -2.58626342e-01 1.34850800e-01 -5.81225514e-01 4.69990194e-01 4.75584000e-01 3.25678140e-01 8.87079418e-01 -8.81486475e-01 8.40301991e-01 9.45299983e-01 4.12103593e-01 5.25235295e-01 -1.07382119e+00 -5.65923750e-01 2.29242742e-01 5.18286049e-01 -1.45999634e+00 -3.76700401e-01 6.65581048e-01 -5.26105344e-01 8.29237878e-01 8.35528523e-02 8.14478457e-01 7.21487463e-01 8.46650898e-02 1.11715615e+00 9.13608551e-01 -1.12025343e-01 1.23387463e-01 -9.00885612e-02 1.53490350e-01 5.83286822e-01 -2.24461675e-01 2.49917023e-02 -6.80197239e-01 3.09064120e-01 7.57824183e-01 5.61913429e-03 -3.85640860e-01 -2.04580128e-01 -1.24176919e+00 5.10844111e-01 4.72054571e-01 2.08986253e-01 -5.43549895e-01 -6.37649298e-02 3.49928021e-01 1.47838786e-01 3.77483457e-01 1.96448416e-01 -4.25795555e-01 -3.91388655e-01 -1.27841711e+00 6.69175312e-02 4.50381130e-01 1.16694057e+00 9.14732575e-01 7.31766447e-02 -3.81032318e-01 6.01122320e-01 1.93417281e-01 5.09363711e-01 3.90090168e-01 -1.23458314e+00 4.62692142e-01 6.06510282e-01 1.30352587e-01 -8.77093315e-01 -2.20986381e-01 -2.99625367e-01 -6.15718186e-01 6.25684708e-02 5.40371478e-01 8.69130343e-02 -1.19790661e+00 1.60270119e+00 5.00985742e-01 5.24966300e-01 5.00089303e-02 1.21596587e+00 9.35568571e-01 7.60077775e-01 6.96846694e-02 -2.70815700e-01 1.00987375e+00 -1.03286850e+00 -5.81537604e-01 -2.87391603e-01 2.50014067e-01 -6.24488413e-01 9.19034541e-01 1.25824079e-01 -1.05875468e+00 -1.03931630e+00 -9.68398750e-01 9.96463150e-02 -4.23069671e-02 -9.68051702e-03 2.50777662e-01 4.53259647e-01 -9.22980607e-01 9.08599257e-01 -9.39672112e-01 -3.52062494e-01 8.77474546e-01 4.46242690e-01 -3.74984622e-01 -6.09617122e-02 -9.46957707e-01 4.16507900e-01 3.42564106e-01 2.38466319e-02 -9.67379749e-01 -8.46091628e-01 -7.79900849e-01 9.91547406e-02 6.37704849e-01 -3.83279115e-01 1.01494539e+00 -1.28455341e+00 -1.36707723e+00 7.73527861e-01 -3.09522241e-01 -5.85420966e-01 5.78349590e-01 -2.91404814e-01 -3.37267667e-01 6.15654767e-01 2.94075906e-01 1.07117474e+00 1.16730070e+00 -9.65658128e-01 -9.36330676e-01 -2.35938460e-01 -9.78572071e-02 2.12137401e-01 -1.49643525e-01 1.21402115e-01 -1.05176508e+00 -5.69456816e-01 2.17586800e-01 -9.00633454e-01 -1.29807502e-01 3.35080884e-02 -2.74229199e-01 -2.69561052e-01 1.00688303e+00 -7.57147968e-01 1.08416188e+00 -2.35823250e+00 1.83048509e-02 1.24390692e-01 1.34073004e-01 5.16361117e-01 -2.55453318e-01 -1.12765171e-01 1.34083405e-01 -2.99684219e-02 -3.14071000e-01 -2.66762763e-01 -3.18610758e-01 7.34128505e-02 -6.81530014e-02 4.22799200e-01 6.02134228e-01 1.03259325e+00 -8.19964170e-01 -7.54829466e-01 3.60987604e-01 4.00445342e-01 -5.85744739e-01 3.97409260e-01 -1.88741982e-01 4.48923081e-01 -3.33292484e-01 8.50867808e-01 5.14133394e-01 -2.72297859e-01 -2.20298305e-01 -3.67805332e-01 -2.39416994e-02 -6.55406096e-04 -1.26663458e+00 1.82963037e+00 1.61090165e-01 7.39747167e-01 -2.16112942e-01 -9.85984385e-01 8.59359920e-01 1.88717842e-01 9.66305673e-01 -6.85716927e-01 1.41524971e-01 1.36527911e-01 -4.08456735e-02 -6.16126716e-01 3.04948568e-01 4.08266157e-01 1.97439954e-01 -1.12365996e-02 1.98191077e-01 -1.51101604e-01 5.56055605e-01 2.75413781e-01 1.03608263e+00 4.54319894e-01 4.95535247e-02 -1.90571353e-01 6.20561302e-01 1.75383046e-01 9.70183313e-01 5.31269312e-01 -5.06960213e-01 1.07585609e+00 3.16608101e-01 -3.44579756e-01 -9.36694980e-01 -9.60739315e-01 -7.81586170e-02 6.75411880e-01 5.29860139e-01 -5.21651208e-01 -9.68162656e-01 -9.41731155e-01 -1.04485527e-01 2.62458473e-01 -6.44367874e-01 -9.19976234e-02 -6.83601618e-01 -2.94490755e-01 2.46795118e-02 7.97819972e-01 8.69283557e-01 -1.06127822e+00 -9.11027133e-01 3.61648858e-01 -3.16139311e-01 -1.54827631e+00 -8.41011226e-01 -1.73548669e-01 -8.54717791e-01 -1.37012577e+00 -7.61035323e-01 -6.33082509e-01 5.77330470e-01 4.29625213e-01 9.42586720e-01 9.84269604e-02 -2.98151076e-01 3.63847286e-01 -4.75433171e-01 -3.95419076e-02 -2.00625673e-01 -5.80833666e-02 -8.87534097e-02 4.12719131e-01 2.02077538e-01 -1.53726757e-01 -6.48594558e-01 6.47716224e-01 -8.17673504e-01 2.57676750e-01 5.32310545e-01 6.70954287e-01 9.15785491e-01 -1.33889660e-01 4.36043262e-01 -6.23928726e-01 -2.47057319e-01 -2.65723407e-01 -5.59392154e-01 1.45424709e-01 -3.14187527e-01 -3.62735569e-01 4.86738384e-01 -4.99523461e-01 -7.62590230e-01 2.66515732e-01 -2.68954062e-03 -8.60738695e-01 -2.10639447e-01 2.97599614e-01 -2.28908777e-01 -7.20499605e-02 3.41111124e-01 2.51886785e-01 1.36196941e-01 -3.79210889e-01 -3.41919325e-02 5.83633959e-01 7.53604650e-01 -2.26829037e-01 6.95876539e-01 4.61268008e-01 -5.29084317e-02 -7.93025374e-01 -9.51323271e-01 -7.39947617e-01 -8.24195206e-01 -5.08728921e-01 1.03428197e+00 -8.84865046e-01 -3.95128757e-01 5.69567144e-01 -1.05341876e+00 -3.07560891e-01 -4.66134697e-01 5.55888295e-01 -7.05237687e-01 4.32483792e-01 -3.91540140e-01 -4.81897533e-01 -2.44681373e-01 -1.44611657e+00 1.04137325e+00 5.02413929e-01 -1.56286389e-01 -4.51522231e-01 -4.66950566e-01 4.78660017e-01 2.48054475e-01 2.82268614e-01 3.23282331e-01 -6.02362454e-01 -7.94180214e-01 -1.08464964e-01 -4.51940924e-01 6.44829810e-01 2.72208273e-01 2.87042618e-01 -8.64365041e-01 -2.54281253e-01 -7.33844042e-02 2.03488525e-02 8.33540380e-01 4.87328321e-01 1.10846674e+00 4.44799587e-02 -2.72686124e-01 6.23838425e-01 1.32069170e+00 4.30929005e-01 6.28649890e-01 2.64498979e-01 8.12994123e-01 5.07347763e-01 1.07973099e+00 4.04547274e-01 2.74844289e-01 9.07583416e-01 4.68011916e-01 -9.49989818e-03 -4.67940152e-01 -6.21275650e-03 3.92874628e-01 3.99865180e-01 -6.70462176e-02 -3.48488539e-02 -8.30463111e-01 6.17033362e-01 -1.97176576e+00 -9.30511951e-01 -1.26169235e-01 2.23563671e+00 5.80324054e-01 3.21158797e-01 3.48479718e-01 2.08666593e-01 9.09301221e-01 9.92188379e-02 -7.68837690e-01 1.07568599e-01 -1.97806805e-01 8.12470019e-02 4.23027605e-01 2.10824504e-01 -1.43348014e+00 1.02115214e+00 5.61783791e+00 9.39950287e-01 -1.11595500e+00 -4.29910719e-02 7.01913595e-01 -1.30667984e-01 2.60612398e-01 -8.66155401e-02 -7.80657113e-01 7.90907264e-01 7.25650370e-01 -1.09863523e-02 3.85965109e-01 7.34274507e-01 3.05095375e-01 -4.05257821e-01 -1.17824233e+00 1.20465660e+00 1.11027002e-01 -1.44159603e+00 -3.58893387e-02 -1.77312315e-01 9.78215277e-01 9.24252998e-03 -8.42280090e-02 2.19918564e-01 -3.86882067e-01 -7.43292868e-01 9.44283426e-01 5.16818345e-01 8.19537044e-01 -6.85300589e-01 7.22006977e-01 1.07199922e-01 -1.45550692e+00 2.09691823e-02 -2.03773975e-01 2.11641729e-01 2.86686242e-01 5.10041535e-01 -5.78653753e-01 6.40836358e-01 9.91937101e-01 1.06218779e+00 -6.53308094e-01 1.33379066e+00 -7.18955845e-02 7.77080357e-01 -4.93398726e-01 2.74087250e-01 3.74473363e-01 -1.31636709e-01 5.39388299e-01 1.14899504e+00 1.83649689e-01 7.65940621e-02 4.72554624e-01 6.75536633e-01 1.15325684e-02 1.96667090e-01 -1.36895210e-01 6.85583279e-02 2.08633304e-01 1.11086953e+00 -9.00367975e-01 -6.15307689e-01 -5.45573771e-01 9.80086267e-01 4.81278114e-02 3.64227027e-01 -1.07795036e+00 -3.62940997e-01 8.08410168e-01 2.41943374e-01 6.97227895e-01 -2.45873019e-01 -8.06763172e-02 -1.03430676e+00 1.65642142e-01 -6.74076378e-01 3.84146154e-01 -6.05869710e-01 -9.90081310e-01 5.87530673e-01 -3.81013341e-02 -1.61713481e+00 -1.48952112e-01 -2.95954943e-01 -5.06291509e-01 6.04864895e-01 -1.49854696e+00 -9.79104042e-01 -5.71706772e-01 6.34565055e-01 9.27809060e-01 -1.10779516e-01 1.53519213e-01 4.13753718e-01 -6.78388774e-01 5.67262650e-01 -1.64251238e-01 3.62361014e-01 4.20171857e-01 -9.38769877e-01 4.24156100e-01 1.05416393e+00 3.23896974e-01 5.31492680e-02 5.13427854e-01 -7.52122343e-01 -1.45391417e+00 -1.31350863e+00 5.79648197e-01 -1.32614896e-01 3.73844683e-01 -3.33130732e-02 -1.13654792e+00 4.30359542e-01 -7.41077960e-02 5.97952247e-01 2.85570890e-01 -6.06858253e-01 -1.20281473e-01 -1.63452849e-01 -1.18469632e+00 4.15893644e-01 1.26258719e+00 -3.23547482e-01 -4.26613152e-01 1.41638026e-01 7.60271609e-01 -7.01279700e-01 -9.69462276e-01 6.82784855e-01 5.24134219e-01 -1.02549052e+00 8.72416973e-01 -4.29943353e-01 3.11984271e-01 -5.61438978e-01 -2.43459091e-01 -8.45677793e-01 -2.36514881e-01 -7.26417124e-01 -1.08761877e-01 1.26410878e+00 1.27186283e-01 -2.85159856e-01 1.00654745e+00 6.68092191e-01 -2.30948895e-01 -8.29340875e-01 -1.09770644e+00 -9.95715022e-01 -5.70563972e-01 -7.28893757e-01 5.16143680e-01 5.52523196e-01 -3.50104779e-01 1.89800549e-03 -1.12687998e-01 1.08734645e-01 5.14383674e-01 3.36562932e-01 8.10604513e-01 -1.06392455e+00 -3.88449691e-02 -3.56013954e-01 -7.72329330e-01 -1.18209147e+00 2.80727059e-01 -7.98329115e-01 1.29021972e-01 -1.46928763e+00 1.34420499e-01 -3.70847046e-01 -3.88749599e-01 3.72313738e-01 -3.44359189e-01 5.88661849e-01 5.99140167e-01 3.72219414e-01 -8.99666786e-01 5.24025679e-01 1.34333515e+00 -1.66418701e-01 -3.02965581e-01 7.54211619e-02 -2.01825023e-01 6.62226379e-01 7.50104666e-01 -4.66355085e-01 -1.14669837e-01 -2.40702003e-01 -5.80717325e-01 -3.21816020e-02 2.80595928e-01 -1.40485048e+00 2.09412128e-01 -2.30420798e-01 4.35416043e-01 -7.50177741e-01 4.03808355e-01 -7.70093739e-01 1.62244841e-01 4.95674402e-01 -4.18262556e-02 -5.71564287e-02 1.34914711e-01 5.26105165e-01 -4.17839140e-01 -1.95948511e-01 9.02949154e-01 1.27354249e-01 -1.23111629e+00 7.38598108e-01 -1.66207597e-01 2.88766295e-01 1.21831560e+00 -7.55584002e-01 7.39813372e-02 -1.56553701e-01 -6.52841091e-01 2.71482855e-01 5.33100843e-01 5.21594167e-01 8.28903496e-01 -1.27637398e+00 -6.97453439e-01 3.62594366e-01 5.82100861e-02 2.21971691e-01 3.94737512e-01 9.18475330e-01 -3.11099142e-01 2.80014545e-01 -2.00552180e-01 -1.05843771e+00 -1.36707389e+00 5.19368887e-01 7.49675333e-02 1.73809379e-01 -7.55439997e-01 7.74409294e-01 1.57162875e-01 3.79314780e-01 2.21298233e-01 -3.84504229e-01 -1.83872074e-01 1.17899373e-01 4.75578099e-01 4.41757083e-01 -3.78655791e-02 -1.02859223e+00 -3.90985429e-01 9.30676103e-01 3.39391641e-02 3.31516713e-02 1.14445460e+00 -1.53051287e-01 2.39952058e-01 2.56462246e-01 1.40081549e+00 -3.21088463e-01 -1.87942660e+00 -4.93618816e-01 7.36766532e-02 -9.27214742e-01 -3.12749576e-03 -4.70803916e-01 -1.45123506e+00 6.40076876e-01 6.57109201e-01 -5.34027107e-02 1.15427279e+00 5.13437204e-02 9.85850513e-01 1.12096265e-01 2.97477275e-01 -1.19319749e+00 2.03232244e-01 4.57447797e-01 6.96971655e-01 -1.57019758e+00 -1.98056057e-01 -6.31120861e-01 -7.55028546e-01 1.04658377e+00 8.48007143e-01 -1.63220763e-01 4.69250947e-01 2.81959176e-02 4.28005606e-02 4.07930315e-02 -4.64795560e-01 -3.97484004e-01 6.79603517e-01 5.45380592e-01 3.14345993e-02 -2.87121207e-01 -4.62770835e-02 4.62361008e-01 6.81141987e-02 7.75709674e-02 2.30354369e-01 9.10074234e-01 -3.28615814e-01 -7.04099894e-01 -2.50721037e-01 6.26144469e-01 -4.33138579e-01 2.00621471e-01 4.28165905e-02 7.20073462e-01 6.44198135e-02 9.58026409e-01 2.42502078e-01 -4.66346085e-01 4.11045313e-01 -7.80234188e-02 4.27872628e-01 -5.39975226e-01 -5.98935366e-01 1.83514044e-01 -1.68923810e-01 -1.03663433e+00 -7.58306921e-01 -9.20288444e-01 -1.54020572e+00 -1.77192017e-02 -2.79193491e-01 3.42561565e-02 5.52346945e-01 1.02256358e+00 4.42916214e-01 4.64571655e-01 8.37897897e-01 -1.21449959e+00 -2.44976833e-01 -7.13129044e-01 -3.80758971e-01 7.80108929e-01 2.11094946e-01 -7.01180100e-01 -1.12874091e-01 2.83792019e-01]
[9.13656997680664, -0.09598333388566971]
b9a5b4c6-6a96-45ee-93c0-46db5a3194be
blind-video-deflickering-by-neural-filtering
2303.08120
null
https://arxiv.org/abs/2303.08120v1
https://arxiv.org/pdf/2303.08120v1.pdf
Blind Video Deflickering by Neural Filtering with a Flawed Atlas
Many videos contain flickering artifacts. Common causes of flicker include video processing algorithms, video generation algorithms, and capturing videos under specific situations. Prior work usually requires specific guidance such as the flickering frequency, manual annotations, or extra consistent videos to remove the flicker. In this work, we propose a general flicker removal framework that only receives a single flickering video as input without additional guidance. Since it is blind to a specific flickering type or guidance, we name this "blind deflickering." The core of our approach is utilizing the neural atlas in cooperation with a neural filtering strategy. The neural atlas is a unified representation for all frames in a video that provides temporal consistency guidance but is flawed in many cases. To this end, a neural network is trained to mimic a filter to learn the consistent features (e.g., color, brightness) and avoid introducing the artifacts in the atlas. To validate our method, we construct a dataset that contains diverse real-world flickering videos. Extensive experiments show that our method achieves satisfying deflickering performance and even outperforms baselines that use extra guidance on a public benchmark.
['Qifeng Chen', 'Zhaoxiang Zhang', 'Xuanchi Ren', 'Chenyang Lei']
2023-03-14
null
http://openaccess.thecvf.com//content/CVPR2023/html/Lei_Blind_Video_Deflickering_by_Neural_Filtering_With_a_Flawed_Atlas_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Lei_Blind_Video_Deflickering_by_Neural_Filtering_With_a_Flawed_Atlas_CVPR_2023_paper.pdf
cvpr-2023-1
['video-generation', 'video-temporal-consistency']
['computer-vision', 'computer-vision']
[ 4.00022268e-01 -6.55616522e-01 3.09531223e-02 -2.17126295e-01 -6.91305220e-01 -8.04640830e-01 2.63493598e-01 -6.67252183e-01 -8.27241018e-02 3.10006052e-01 3.52716595e-01 -2.92301048e-02 2.00534210e-01 -2.92167991e-01 -1.06213272e+00 -8.31998944e-01 -3.62542830e-02 -7.17961371e-01 3.06208521e-01 6.25131205e-02 2.90268391e-01 1.20285213e-01 -1.62965882e+00 3.76270890e-01 8.55545640e-01 1.18001211e+00 2.01622561e-01 8.20916772e-01 4.17198569e-01 1.00030458e+00 -7.02077329e-01 -1.76993221e-01 7.52233088e-01 -4.17587131e-01 -3.22223276e-01 1.71467915e-01 1.00080311e+00 -9.36262310e-01 -8.15644324e-01 1.40476406e+00 4.80629474e-01 3.37688714e-01 3.59819740e-01 -1.54670799e+00 -7.85739601e-01 4.41282034e-01 -5.31513691e-01 3.84724587e-01 5.58173478e-01 6.23656332e-01 5.68897843e-01 -8.94099712e-01 3.48028064e-01 9.25396264e-01 8.41997683e-01 4.47008997e-01 -7.12162971e-01 -9.00822997e-01 3.28180581e-01 5.98875940e-01 -1.51784635e+00 -7.06952155e-01 9.71616209e-01 -5.00897884e-01 6.86012268e-01 3.15040320e-01 5.67511439e-01 1.23277044e+00 1.85799450e-01 7.56389678e-01 6.84923232e-01 -1.73962370e-01 2.55511254e-01 -3.65933657e-01 2.32833698e-02 3.39994580e-01 1.71590656e-01 1.41825333e-01 -4.48018879e-01 -3.44643183e-02 6.84439361e-01 2.12022394e-01 -8.84024382e-01 -2.16508046e-01 -1.08823872e+00 2.25081131e-01 2.81274199e-01 -4.82628793e-02 -2.11869478e-01 4.50608194e-01 4.93694991e-01 2.79793620e-01 9.37241018e-02 3.28966051e-01 -4.84379441e-01 -8.50849152e-02 -9.54271615e-01 1.94670096e-01 4.31246340e-01 1.01915574e+00 5.26574671e-01 2.25842252e-01 -6.15100920e-01 6.89913213e-01 1.28478929e-01 5.67936778e-01 2.96461731e-01 -1.42140520e+00 4.46930587e-01 8.80088508e-02 4.61370736e-01 -1.29094398e+00 -6.55566230e-02 -1.77977517e-01 -9.24050987e-01 1.31156713e-01 2.45520383e-01 -2.33510777e-01 -1.02930331e+00 1.74615037e+00 7.45867193e-02 9.74593580e-01 -3.25355440e-01 1.40894139e+00 7.70824969e-01 6.47153556e-01 -1.83967292e-01 -3.55327725e-01 1.21862280e+00 -1.18222404e+00 -1.04597986e+00 -4.13980968e-02 1.03304960e-01 -6.82839990e-01 1.07694113e+00 8.41097593e-01 -8.47142637e-01 -7.47405231e-01 -1.14775372e+00 2.82587875e-02 -2.15046093e-01 2.53290802e-01 2.67209679e-01 4.22061920e-01 -1.05048585e+00 5.90892851e-01 -5.12874067e-01 -1.25408381e-01 1.16724744e-01 1.76717583e-02 -2.41759881e-01 -2.67015606e-01 -1.38003016e+00 4.03328270e-01 2.57641882e-01 3.59769464e-01 -1.18149722e+00 -6.08689368e-01 -8.67829800e-01 7.63485581e-02 8.81713867e-01 -5.25875628e-01 1.32131350e+00 -1.46855521e+00 -1.46754849e+00 3.37891400e-01 -9.97488275e-02 -3.72005522e-01 6.54449224e-01 -8.07769001e-01 -7.34319508e-01 4.28871959e-01 -4.92516207e-03 4.09823000e-01 1.59168386e+00 -1.32287300e+00 -5.36338627e-01 1.95838988e-01 2.17508540e-01 -1.41159957e-02 -3.78232926e-01 1.47065938e-01 -1.15166306e+00 -1.19620883e+00 -2.57492363e-01 -8.63603354e-01 2.92585921e-02 -1.78641811e-01 -4.28164542e-01 1.15092754e-01 1.08656371e+00 -1.05472565e+00 1.78957117e+00 -2.42538285e+00 -2.01181367e-01 1.52222931e-01 2.89286524e-01 3.80877346e-01 -4.63980973e-01 1.15788057e-01 -3.19756985e-01 -3.64795178e-02 -8.90764445e-02 1.61623349e-03 -4.99111824e-02 -2.46947020e-01 -4.45259690e-01 6.54091001e-01 1.51983649e-01 5.41102648e-01 -1.09096432e+00 -1.39628828e-01 3.36416602e-01 4.92794693e-01 -7.91343808e-01 3.33401322e-01 -8.26737061e-02 1.49822384e-01 -1.07429798e-04 7.70755470e-01 9.34556544e-01 -1.61638826e-01 -7.99226239e-02 -8.25143337e-01 -8.04040581e-02 -1.06260285e-01 -1.43662345e+00 1.71422052e+00 6.67347200e-03 7.56150186e-01 1.06996514e-01 -6.73490167e-01 4.02844876e-01 3.31952423e-01 5.11615813e-01 -6.09381735e-01 1.74129516e-01 -5.27762547e-02 -2.63856024e-01 -9.19426143e-01 5.57797074e-01 6.02181315e-01 9.28559825e-02 1.80420354e-01 -1.97935179e-01 3.73806417e-01 3.13750505e-01 3.59264374e-01 1.49429560e+00 4.37623739e-01 1.00530811e-01 1.82528347e-01 4.25625473e-01 -3.33808064e-01 9.52452242e-01 1.03050733e+00 -5.86929083e-01 9.81909931e-01 2.13205308e-01 -3.17176670e-01 -7.59778559e-01 -9.58440900e-01 3.06469798e-01 1.02244902e+00 5.82401335e-01 -6.95593476e-01 -9.55227852e-01 -7.85214424e-01 -1.59381732e-01 3.98583859e-01 -6.39037073e-01 -4.01261359e-01 -5.14782846e-01 -3.14208537e-01 4.41174299e-01 5.95797002e-01 7.42150068e-01 -8.08293343e-01 -5.03016055e-01 4.78085987e-02 -4.05703336e-01 -1.30825186e+00 -1.07271993e+00 -2.65663356e-01 -2.25768521e-01 -1.41037929e+00 -5.90678215e-01 -5.22897363e-01 6.92889690e-01 9.79915142e-01 9.32555616e-01 1.37425378e-01 -2.67791331e-01 3.53377849e-01 -5.10645449e-01 4.20425832e-02 -3.60288210e-02 -4.94478285e-01 1.84578970e-01 1.76017344e-01 3.23167831e-01 -2.81436086e-01 -8.73652518e-01 4.93389636e-01 -1.11295712e+00 -7.19532818e-02 3.65763158e-01 6.16873920e-01 2.45578110e-01 5.62803805e-01 2.01966748e-01 -3.55098337e-01 4.99947816e-01 -5.10113776e-01 -5.46441078e-01 2.28727862e-01 -1.61519811e-01 -4.26008970e-01 9.90403056e-01 -6.29299939e-01 -1.09017527e+00 1.24333128e-01 2.58946627e-01 -1.07810640e+00 -4.30200487e-01 2.09335133e-01 -4.52840686e-01 -3.37060839e-02 5.37762582e-01 1.13912053e-01 -2.64698535e-01 -4.25547570e-01 2.88023919e-01 6.41619146e-01 1.02455235e+00 -7.22783282e-02 9.96002257e-01 3.77761811e-01 -4.28541899e-01 -5.21789491e-01 -7.25341082e-01 -4.53124225e-01 -2.22263396e-01 -6.03980958e-01 6.37435794e-01 -1.24060857e+00 -6.03124857e-01 8.59267592e-01 -1.05265379e+00 -2.26631895e-01 3.37775737e-01 5.24863482e-01 -3.63121837e-01 7.18586922e-01 -6.30449533e-01 -5.66754997e-01 -2.29839936e-01 -1.10263610e+00 8.68871868e-01 3.71064395e-01 -3.61266779e-04 -4.24010187e-01 -4.89464700e-02 2.22075641e-01 3.82125825e-01 1.13989003e-01 1.77312896e-01 -1.91734508e-01 -7.47558355e-01 -2.28444114e-01 -2.71914899e-01 7.89029479e-01 3.44062835e-01 3.61532897e-01 -1.08888495e+00 -4.14897710e-01 -1.58473611e-01 2.72252858e-02 1.06359184e+00 5.98554850e-01 1.69828200e+00 -4.50324237e-01 -1.23876832e-01 9.89129424e-01 1.37011433e+00 4.58349705e-01 8.43736291e-01 2.52049804e-01 8.66149664e-01 1.91180572e-01 6.74749792e-01 5.59953868e-01 6.70254305e-02 7.08023906e-01 4.82241303e-01 -4.26633619e-02 -1.95281550e-01 -1.57314494e-01 9.19486344e-01 6.05856478e-01 2.13238224e-02 -4.43838775e-01 -4.75852638e-01 3.37246656e-01 -1.99060762e+00 -1.36800063e+00 4.56466042e-02 2.03992963e+00 6.08592391e-01 -1.23753674e-01 3.34286690e-03 1.33178264e-01 9.30400670e-01 2.46221989e-01 -7.21171856e-01 1.09571159e-01 -9.38305482e-02 -3.24316919e-01 5.94964087e-01 1.93175629e-01 -1.44169903e+00 7.76052058e-01 6.39443493e+00 6.62517667e-01 -1.11566412e+00 -1.71543479e-01 3.17128271e-01 -3.08406115e-01 -3.92866583e-04 -1.58517808e-01 -3.91870946e-01 9.88328457e-01 3.82176995e-01 8.83243978e-02 9.43591118e-01 6.22234523e-01 7.45449781e-01 -1.03425093e-01 -9.57076788e-01 1.29643273e+00 4.71742779e-01 -1.19220054e+00 -7.06642959e-03 -5.02458274e-01 7.21680105e-01 -2.41734907e-01 -1.27306916e-02 2.58038282e-01 1.31494388e-01 -8.47699940e-01 8.41243446e-01 8.70108604e-01 7.87037134e-01 -4.47953135e-01 6.98568165e-01 -1.18808404e-01 -1.21335351e+00 -1.85777098e-01 2.18820535e-02 3.97138894e-02 1.03632897e-01 4.63903069e-01 -2.10953087e-01 5.10297954e-01 1.13418067e+00 1.08863163e+00 -6.94834590e-01 1.60777676e+00 -1.81158006e-01 7.03614473e-01 -2.80287981e-01 6.36946440e-01 -2.38786601e-02 -9.23218057e-02 4.95711207e-01 1.26091397e+00 5.98032176e-01 1.38776645e-01 1.18973069e-01 4.51026917e-01 -3.75962220e-02 -1.57744795e-01 -6.08146787e-01 -9.41369161e-02 6.50378525e-01 1.21678829e+00 -4.48360443e-01 -3.88706148e-01 -6.82234406e-01 1.20007348e+00 -1.38114244e-01 8.33734810e-01 -1.36251247e+00 -6.32536173e-01 7.93011010e-01 -1.26029506e-01 4.44746405e-01 6.77596852e-02 1.72459304e-01 -1.35155988e+00 2.27228776e-01 -1.13747144e+00 3.53791416e-01 -1.23314524e+00 -1.23644412e+00 2.81083047e-01 -1.07802898e-01 -1.68798912e+00 -2.83927098e-02 -5.32019258e-01 -8.55952621e-01 4.11884516e-01 -1.40507174e+00 -6.36211693e-01 -8.52754712e-01 8.65102887e-01 6.10399902e-01 -5.78675643e-02 1.34256348e-01 6.96961224e-01 -9.94359612e-01 6.79685712e-01 -1.05478726e-01 4.18955624e-01 1.27134943e+00 -1.01125944e+00 1.02716140e-01 1.53419411e+00 -2.11972862e-01 6.65905952e-01 9.43194270e-01 -6.63105726e-01 -1.65207863e+00 -1.29818296e+00 9.56176370e-02 -1.91561058e-01 6.39010668e-01 -7.15464875e-02 -9.27205503e-01 5.80118418e-01 3.34712535e-01 2.23020598e-01 5.27133286e-01 -5.25777280e-01 -4.03996736e-01 -3.66345674e-01 -7.47672915e-01 6.67167544e-01 1.16014767e+00 -4.60963488e-01 -3.77585262e-01 1.16656415e-01 7.98741341e-01 -3.33760917e-01 -4.39230621e-01 4.78059322e-01 5.70505679e-01 -1.16620421e+00 9.36078370e-01 -3.50021273e-01 4.41062242e-01 -1.05312562e+00 -2.04464212e-01 -1.36456716e+00 -5.14499962e-01 -1.01625359e+00 -6.33167386e-01 1.20634890e+00 -2.34045580e-01 -1.51322857e-01 2.73837060e-01 6.23023808e-01 -3.96096855e-01 -2.42706820e-01 -4.89422917e-01 -7.93275476e-01 -6.72747493e-01 -6.56474888e-01 6.63596928e-01 9.26519334e-01 -2.56284088e-01 -1.56203136e-01 -1.04276896e+00 5.42165160e-01 6.10496938e-01 -1.32501766e-01 1.02578330e+00 -6.22320652e-01 -2.66739756e-01 -3.67635310e-01 -1.96895599e-01 -1.06606054e+00 3.55446413e-02 -2.37391204e-01 4.19687599e-01 -1.23964512e+00 2.07892820e-01 2.15257794e-01 -6.47937655e-01 3.87750387e-01 -5.17107368e-01 4.01746273e-01 2.88229138e-01 8.81042480e-02 -1.00835145e+00 3.36396843e-01 9.97310042e-01 -1.20583266e-01 -2.04663604e-01 -4.24468786e-01 -7.31258988e-01 9.83936608e-01 6.79421186e-01 -1.30648151e-01 -4.14848208e-01 -6.45707190e-01 2.48747841e-01 -1.76548883e-01 6.82425678e-01 -1.18036401e+00 2.61835605e-01 -2.48584971e-01 4.89612132e-01 -4.94712144e-01 3.88379507e-02 -1.03514290e+00 4.02267843e-01 1.54773012e-01 -2.47523934e-01 -2.27208566e-02 8.83561969e-02 7.49511957e-01 -2.37754494e-01 2.95526460e-02 6.15305901e-01 7.08168074e-02 -1.29015529e+00 3.05901557e-01 -2.85342932e-01 -7.39000440e-02 1.16584146e+00 -3.37471843e-01 -5.41380525e-01 -5.53515673e-01 -3.12743425e-01 3.23700666e-01 6.79062545e-01 7.03920305e-01 6.94854796e-01 -1.44369018e+00 -3.74327868e-01 4.24281687e-01 1.37395695e-01 -4.09676343e-01 5.82119405e-01 8.57785761e-01 -3.90194833e-01 6.95081055e-02 -5.60016595e-02 -3.10466230e-01 -1.25941908e+00 8.95465910e-01 3.93564701e-01 5.12084842e-01 -6.69472158e-01 6.73582554e-01 2.29487494e-01 4.02057171e-01 6.31960392e-01 -5.52268267e-01 -1.33671761e-01 -9.84613150e-02 9.38275754e-01 4.01965082e-01 -2.07441568e-01 -6.19581163e-01 -3.60524654e-01 5.37965596e-01 1.10964857e-01 3.45914066e-01 8.92839730e-01 -3.93338829e-01 1.35919705e-01 1.19946674e-01 1.17916238e+00 1.78664699e-01 -1.88739610e+00 -1.55751690e-01 -5.42216718e-01 -9.80718791e-01 1.30570367e-01 -8.49026382e-01 -1.42279029e+00 3.79168481e-01 6.19103014e-01 3.14298004e-01 1.70987296e+00 -3.94017041e-01 9.49147642e-01 2.84264356e-01 2.49469817e-01 -1.32824314e+00 1.49634570e-01 3.57953131e-01 9.20204818e-01 -1.19747365e+00 -2.09541216e-01 -3.87161046e-01 -5.50547600e-01 9.89883006e-01 8.80838335e-01 -2.41701201e-01 4.37367588e-01 4.54364985e-01 2.13138774e-01 1.74443677e-01 -7.41200149e-01 -6.43312037e-02 4.22902435e-01 5.23608983e-01 5.25706112e-02 -2.83781826e-01 1.55087588e-02 6.70798838e-01 8.67961124e-02 3.22144926e-01 4.82669026e-01 8.22815955e-01 -3.80457282e-01 -3.40024024e-01 -6.19982719e-01 4.12069231e-01 -4.25931156e-01 -5.78256622e-02 -2.84837723e-01 3.11753988e-01 3.97914648e-01 1.12945473e+00 9.29576829e-02 -7.28006184e-01 2.57868975e-01 -2.28277698e-01 2.69622028e-01 -9.61736068e-02 -4.57647949e-01 3.34837317e-01 -1.91466287e-01 -1.18832028e+00 -5.63556373e-01 -5.09036541e-01 -9.56717670e-01 -2.29349032e-01 -2.56854743e-01 -1.63422137e-01 1.58911139e-01 7.46459305e-01 4.76489395e-01 6.55517995e-01 6.86565161e-01 -9.56746936e-01 -2.67890006e-01 -7.78621852e-01 -6.13271952e-01 8.85632515e-01 7.04282403e-01 -7.27100849e-01 -6.52546227e-01 5.82158506e-01]
[10.725076675415039, -1.5294805765151978]
d66289a5-20be-4b06-afeb-d36e5d9128c8
hla-class-i-binding-prediction-via
1701.00593
null
http://arxiv.org/abs/1701.00593v2
http://arxiv.org/pdf/1701.00593v2.pdf
HLA class I binding prediction via convolutional neural networks
Many biological processes are governed by protein-ligand interactions. One such example is the recognition of self and nonself cells by the immune system. This immune response process is regulated by the major histocompatibility complex (MHC) protein which is encoded by the human leukocyte antigen (HLA) complex. Understanding the binding potential between MHC and peptides can lead to the design of more potent, peptide-based vaccines and immunotherapies for infectious autoimmune diseases. We apply machine learning techniques from the natural language processing (NLP) domain to address the task of MHC-peptide binding prediction. More specifically, we introduce a new distributed representation of amino acids, name HLA-Vec, that can be used for a variety of downstream proteomic machine learning tasks. We then propose a deep convolutional neural network architecture, name HLA-CNN, for the task of HLA class I-peptide binding prediction. Experimental results show combining the new distributed representation with our HLA-CNN architecture achieves state-of-the-art results in the majority of the latest two Immune Epitope Database (IEDB) weekly automated benchmark datasets. We further apply our model to predict binding on the human genome and identify 15 genes with potential for self binding.
['Yeeleng Scott Vang', 'Xiaohui Xie']
2017-01-03
null
null
null
null
['mhc-presentation-prediction']
['medical']
[ 4.72133011e-01 -4.05373991e-01 -4.36159283e-01 -6.30247235e-01 -7.77726531e-01 -6.73325181e-01 1.31694078e-01 4.56629783e-01 -6.36653125e-01 1.14086890e+00 2.24985421e-01 -4.49764997e-01 4.67562266e-02 -9.44528997e-01 -7.62909353e-01 -1.00197828e+00 -1.58947036e-02 1.09431875e+00 1.05970822e-01 -3.40279520e-01 2.06819698e-01 6.29946411e-01 -1.13992822e+00 9.56062913e-01 6.19270086e-01 8.45514894e-01 -2.96857152e-02 3.36199939e-01 -5.66970706e-01 1.14189364e-01 -2.23882481e-01 -1.04721986e-01 5.13650216e-02 -3.41549128e-01 -5.92440069e-01 -7.57365406e-01 -9.33002532e-02 9.77547616e-02 1.97105542e-01 7.07677722e-01 8.42450678e-01 -2.89893001e-01 8.20347846e-01 -6.46031201e-01 -4.36773419e-01 3.60643268e-02 -5.00903189e-01 7.68101811e-02 3.19509476e-01 7.09968805e-02 7.35518456e-01 -8.92742515e-01 1.25264645e+00 1.40430653e+00 6.06770933e-01 7.12618411e-01 -1.50091124e+00 -7.51849174e-01 -5.07232070e-01 3.86954963e-01 -1.31891167e+00 -3.62154134e-02 2.93967277e-01 -6.90421581e-01 1.64483130e+00 -6.23354726e-02 2.93846607e-01 1.10478985e+00 1.07002664e+00 5.03730416e-01 1.03097439e+00 -4.57656533e-01 8.44720826e-02 -4.58600253e-01 3.79303724e-01 6.83165669e-01 -2.47115135e-01 3.51518661e-01 -7.51642585e-01 -8.04008543e-01 2.80790329e-01 -1.81969684e-02 -7.36838952e-02 -8.09728861e-01 -9.99919057e-01 1.21491098e+00 9.84216109e-02 9.73339602e-02 -5.01670539e-01 -2.96097308e-01 7.41043150e-01 3.65725517e-01 3.46675634e-01 1.08221717e-01 -8.67956042e-01 3.00181955e-01 -3.02604228e-01 3.52449507e-01 1.01811838e+00 7.87551627e-02 8.05365920e-01 -5.90554595e-01 -7.30371997e-02 1.22838843e+00 1.37109861e-01 5.27850032e-01 3.12453717e-01 -5.85422367e-02 -2.92903066e-01 3.20775896e-01 -5.46590611e-02 -6.08689308e-01 -9.36847150e-01 -2.74882644e-01 -6.59711421e-01 1.68970510e-01 3.97547692e-01 2.21689213e-02 -6.27573431e-01 1.72294736e+00 6.06026173e-01 -8.75512287e-02 1.72405526e-01 1.14609456e+00 5.14746606e-01 7.28024304e-01 6.26093626e-01 -1.28232643e-01 1.77963555e+00 -6.05692983e-01 -4.55600977e-01 1.41130283e-01 6.34702623e-01 -8.00351262e-01 4.10443217e-01 1.22011557e-01 -5.69940984e-01 -3.59407812e-01 -9.21047151e-01 7.47741312e-02 -6.62767947e-01 -4.81657833e-01 6.67543411e-01 2.90545464e-01 -5.76729596e-01 5.38446784e-01 -6.12347662e-01 -6.47469163e-01 1.76663369e-01 7.04478741e-01 -7.06261039e-01 -1.67877704e-01 -1.75681400e+00 1.14814258e+00 3.87546659e-01 -2.38977343e-01 -5.82672417e-01 -7.96554625e-01 -5.19360423e-01 -7.48836948e-03 -2.33858079e-01 -8.46213043e-01 7.85180628e-01 -8.35318863e-01 -1.24650002e+00 1.43787217e+00 8.20944458e-02 -7.76041090e-01 -3.95692885e-01 -2.05066592e-01 -5.99077880e-01 -1.43219396e-01 -1.08548969e-01 7.99162328e-01 1.81422979e-01 -3.19878221e-01 -5.65373778e-01 -7.66756713e-01 -6.68415368e-01 -1.87066749e-01 4.57329810e-01 4.81632411e-01 1.76721260e-01 -5.40031135e-01 -5.09130180e-01 -9.09265518e-01 -2.32482105e-01 -4.71228287e-02 -9.05810222e-02 -3.40418994e-01 2.45711297e-01 -4.18274105e-01 4.31016654e-01 -1.77747440e+00 2.74337828e-01 2.78742820e-01 -6.14875033e-02 5.64581096e-01 -6.10690832e-01 9.50557411e-01 -3.71286571e-01 -2.46454418e-01 5.06271161e-02 8.44854772e-01 6.89176172e-02 -5.93981631e-02 -5.10010600e-01 5.83947182e-01 5.37312508e-01 1.18153906e+00 -5.07857025e-01 8.32639262e-02 -1.15049928e-01 9.65670943e-01 -4.20535654e-01 3.53689492e-01 -6.53296709e-01 5.86853981e-01 -4.78488505e-01 5.50515771e-01 9.30594027e-01 -1.43511087e-01 8.52049589e-01 -4.01746690e-01 9.87035707e-02 2.36937150e-01 -2.88669348e-01 1.68616211e+00 7.61097297e-02 3.08720231e-01 1.36533543e-01 -9.47673202e-01 1.13382494e+00 -1.27695613e-02 4.27652836e-01 -1.11144435e+00 2.52138712e-02 1.96406856e-01 3.94697875e-01 -2.59915233e-01 -6.18652463e-01 -5.91796517e-01 -7.13495463e-02 6.37644708e-01 1.38859421e-01 8.98979783e-01 -3.25699627e-01 -1.63290396e-01 1.18582213e+00 3.40737134e-01 4.52348500e-01 -6.71795607e-01 6.72749460e-01 3.57903957e-01 8.45725536e-01 4.30460721e-01 -1.81735441e-01 -4.40831296e-03 2.89616048e-01 -1.30888307e+00 -1.11691070e+00 -1.12163758e+00 -4.59943146e-01 1.62331295e+00 -3.81911725e-01 3.78414184e-01 -7.18784750e-01 -4.92908299e-01 3.80519509e-01 1.46728963e-01 -5.60834944e-01 -2.16497391e-01 -6.24498904e-01 -9.97278810e-01 6.69824064e-01 4.07211572e-01 -8.19381624e-02 -1.34725893e+00 -2.01036826e-01 7.36127913e-01 5.73017560e-02 -6.31179094e-01 -3.89224410e-01 7.09275007e-01 -2.95930624e-01 -1.21614456e+00 -5.82507908e-01 -1.01777470e+00 1.28898263e-01 -2.77471155e-01 1.12766385e+00 -3.77770692e-01 -8.09909046e-01 -5.54783456e-02 4.84931329e-03 -6.08929455e-01 -4.39599395e-01 -8.18942208e-03 3.86402845e-01 -1.18197843e-01 1.33532369e+00 -3.27906460e-01 -7.91025460e-01 1.60446659e-01 -1.00770473e+00 6.53451383e-02 8.97274315e-01 1.29119623e+00 9.98815119e-01 -7.28154063e-01 1.16516399e+00 -1.21624601e+00 5.05093694e-01 -6.26925528e-01 -5.45595109e-01 4.87771720e-01 -2.37904146e-01 4.36717719e-01 3.90435517e-01 -3.60361397e-01 -1.07358646e+00 4.69493538e-01 -7.07526565e-01 3.69225889e-01 -5.67099333e-01 5.58771431e-01 -2.86609799e-01 -2.28576168e-01 3.39634299e-01 4.07744139e-01 2.38268733e-01 -3.82676125e-01 3.03162873e-01 5.36640823e-01 5.60688555e-01 -5.88241279e-01 -8.53597820e-02 3.74146402e-01 2.10306585e-01 -6.40252173e-01 -8.86591852e-01 -5.47628284e-01 -5.35476267e-01 3.95974189e-01 1.15229297e+00 -8.11541200e-01 -1.09345675e+00 3.94681603e-01 -1.19309056e+00 -2.57383138e-01 6.17770612e-01 2.64968783e-01 -7.08320796e-01 2.24937245e-01 -1.02418458e+00 -1.59073681e-01 -9.84550357e-01 -7.44287193e-01 1.12522602e+00 -1.95039272e-01 -3.62628430e-01 -8.44289124e-01 1.04666257e+00 3.28854561e-01 4.75966334e-01 8.32760856e-02 1.74797046e+00 -1.37853241e+00 -1.35255918e-01 -1.28137479e-02 -3.72337759e-01 -9.96822715e-02 -2.10633278e-01 -5.60167253e-01 -5.95648885e-01 -3.95232081e-01 -4.02164370e-01 -5.78937650e-01 1.15938139e+00 2.32432440e-01 5.41695833e-01 1.02494145e-02 -6.16340399e-01 6.47901595e-01 1.65437579e+00 4.93319452e-01 7.05094874e-01 2.42171317e-01 3.47808123e-01 1.08954024e+00 9.53311920e-01 1.84437037e-01 -1.23186938e-01 9.09154296e-01 3.50057632e-02 -1.82294577e-01 2.93136269e-01 1.11139894e-01 8.63893777e-02 4.44753170e-01 5.01071513e-01 -3.55215281e-01 -1.26706052e+00 3.70554149e-01 -1.79992270e+00 -6.66666865e-01 -4.31929380e-02 1.95095289e+00 1.10237002e+00 -2.24531993e-01 -2.82050688e-02 -8.25102448e-01 6.41375661e-01 -3.38939011e-01 -6.69597805e-01 -7.55266845e-01 -4.84903514e-01 1.08242965e+00 3.13863724e-01 4.90032077e-01 -1.15257895e+00 7.53354907e-01 6.23250198e+00 8.27589631e-01 -1.10404503e+00 2.22360082e-02 5.42143643e-01 1.95148379e-01 2.37052310e-02 -1.99803770e-01 -9.62627470e-01 2.69282967e-01 1.20804501e+00 1.74784407e-01 1.00715190e-01 4.35900331e-01 -1.65580779e-01 2.41275355e-01 -1.15581965e+00 7.32165873e-01 -1.40770182e-01 -1.73284912e+00 1.23081028e-01 -1.01601787e-01 2.83436239e-01 4.41749275e-01 -9.03383419e-02 3.07497412e-01 2.65902936e-01 -1.33774126e+00 -3.93253565e-01 7.04499424e-01 7.58007109e-01 -1.12651682e+00 1.07396412e+00 1.84690014e-01 -6.93218291e-01 3.22926998e-01 -6.51269257e-01 2.64775246e-01 2.90152933e-02 7.05394685e-01 -8.46983194e-01 1.99834704e-02 3.83164287e-01 5.23996875e-02 1.76995829e-01 4.50285107e-01 2.29055122e-01 2.87613302e-01 1.54023185e-01 -5.94887249e-02 8.84543732e-02 -4.34048139e-02 2.63921797e-01 1.57003570e+00 -1.74274743e-01 5.53154826e-01 4.49266702e-01 6.07950270e-01 4.02169824e-02 6.69704556e-01 -5.15623093e-01 -1.37702659e-01 1.18421562e-01 1.24252689e+00 -2.81555355e-01 -6.46062121e-02 -4.07924563e-01 8.92781675e-01 7.24263966e-01 1.94074005e-01 -3.81304383e-01 -3.52276862e-01 1.34417498e+00 8.56205001e-02 3.90448511e-01 1.37476206e-01 4.26756531e-01 -6.41617715e-01 -7.90242195e-01 -1.13931215e+00 5.87250769e-01 -4.34341937e-01 -1.95825446e+00 4.00880516e-01 -7.07834482e-01 -5.25454402e-01 -3.21444780e-01 -1.09777594e+00 -2.39904359e-01 1.20967007e+00 -1.43573010e+00 -1.29640007e+00 3.88821781e-01 2.39476070e-01 2.67741829e-01 -4.00020063e-01 1.47643995e+00 1.69529468e-01 -1.96334422e-01 2.65070885e-01 3.98848653e-01 2.72040553e-02 1.47979701e+00 -8.95286918e-01 5.36470592e-01 -3.20889384e-01 -3.22761536e-01 7.55105555e-01 5.63918471e-01 -8.20145965e-01 -1.45083570e+00 -1.13794029e+00 1.22464919e+00 -2.92670101e-01 6.74667418e-01 -6.50880039e-01 -1.16749156e+00 3.45987499e-01 2.27439851e-01 -4.91535157e-01 1.61991906e+00 3.77937742e-02 -7.26986825e-01 5.69638684e-02 -1.11346972e+00 1.49296686e-01 5.42713523e-01 -6.05281174e-01 -4.67305988e-01 4.94078070e-01 7.62973905e-01 -5.18718101e-02 -1.13389921e+00 5.20050645e-01 1.06623435e+00 -7.72364676e-01 1.19170582e+00 -1.41038132e+00 2.21028641e-01 -1.71822324e-01 -1.05868466e-01 -1.18816984e+00 -6.14946842e-01 4.49581742e-02 2.96589047e-01 4.82093483e-01 3.56485963e-01 -6.85372353e-01 8.11900914e-01 3.14752236e-02 1.05305858e-01 -8.91940951e-01 -1.06970739e+00 -3.56295168e-01 4.67468530e-01 7.32342526e-02 5.15372097e-01 9.60271120e-01 1.37850389e-01 7.36038148e-01 -2.36822054e-01 -1.86291188e-01 6.82527304e-01 7.02581465e-01 5.51893532e-01 -1.35153139e+00 -5.02481520e-01 -1.82039559e-01 -4.15038675e-01 -5.11935771e-01 4.67253417e-01 -1.03824878e+00 7.37252161e-02 -1.14107907e+00 6.89702451e-01 -4.16077971e-02 -9.11388457e-01 4.41945493e-01 5.46318442e-02 2.09592238e-01 -4.66389537e-01 -1.21305957e-02 -5.28999209e-01 1.12526715e-01 6.62530482e-01 -3.61254781e-01 1.39662743e-01 -4.13956851e-01 -2.31423512e-01 2.29591429e-01 7.84574747e-01 -6.05717123e-01 2.14477926e-01 -7.91624412e-02 3.85928869e-01 9.69589502e-02 1.01728506e-01 -3.64350528e-01 1.22498155e-01 -3.62082779e-01 6.45755887e-01 -8.14018607e-01 7.79218450e-02 -4.09456670e-01 4.82449889e-01 1.04761446e+00 -4.49187338e-01 -6.88541681e-02 4.61200207e-01 7.43965209e-01 -1.92202866e-01 3.76423746e-01 9.25641000e-01 9.16850939e-02 -6.99980438e-01 2.56595820e-01 -8.08349669e-01 -2.73437619e-01 9.98507202e-01 2.08922520e-01 -5.97218633e-01 1.98583975e-01 -9.57569182e-01 5.59104653e-03 3.74271780e-01 4.85540479e-01 5.35912573e-01 -1.18718374e+00 -1.12705982e+00 4.75578129e-01 5.04378021e-01 -1.03058147e+00 4.78291273e-01 5.42915821e-01 -7.57633209e-01 1.16296911e+00 -9.37557995e-01 -6.00433886e-01 -1.39991951e+00 1.03252649e+00 3.00780356e-01 -5.10321140e-01 -8.75359476e-02 8.50238621e-01 9.85414028e-01 -8.52174401e-01 -1.03267908e-01 4.72847402e-01 -2.17214480e-01 -2.33071178e-01 7.63090134e-01 -2.01904953e-01 1.34497181e-01 -5.74119151e-01 -7.33394146e-01 4.83438820e-01 -5.42090535e-01 5.34752071e-01 1.50288856e+00 3.14673424e-01 -8.18182647e-01 -4.22000960e-02 1.38948298e+00 -3.63219649e-01 -4.75452572e-01 -4.65266198e-01 3.44015777e-01 1.42148584e-01 -6.11012995e-01 -1.27511513e+00 -5.02949297e-01 8.93352449e-01 1.19422352e+00 -6.58736408e-01 8.41385484e-01 2.16109622e-02 1.11702478e+00 5.94869256e-01 6.82873607e-01 -9.20021534e-01 -2.58588046e-01 6.76625371e-01 6.57509565e-01 -1.08268785e+00 -4.06264096e-01 -2.65238166e-01 -4.25053746e-01 1.00873542e+00 4.93191630e-01 -1.56343341e-01 5.40689766e-01 4.26352978e-01 3.92428935e-01 -4.66755569e-01 -1.44961810e+00 1.09911501e-01 4.21520114e-01 9.43987012e-01 9.27854419e-01 3.01794469e-01 -8.67633760e-01 9.08048570e-01 5.73120654e-01 -1.62346274e-01 -3.68524909e-01 6.31406844e-01 -8.11477602e-01 -1.88873982e+00 -1.33161038e-01 2.88313597e-01 -8.40856314e-01 -2.65755117e-01 -8.36063325e-01 3.32805008e-01 9.12066847e-02 4.30705547e-01 3.53066236e-01 -2.08121669e-02 1.53821066e-01 6.17871821e-01 4.27673608e-01 -5.54676414e-01 -6.99888229e-01 4.54336762e-01 1.45433277e-01 -6.76791847e-01 -1.72575146e-01 -2.20784724e-01 -1.40264058e+00 -1.48181289e-01 1.81880966e-02 1.39770806e-01 6.44033194e-01 6.41870618e-01 8.59113514e-01 2.71772623e-01 1.64040312e-01 -3.43928546e-01 -4.95397180e-01 -6.47018790e-01 -6.69219971e-01 3.55814517e-01 -9.99626815e-02 -3.77525806e-01 4.97637391e-01 -3.05263698e-01]
[4.733050346374512, 5.612176418304443]
706aed2b-15fe-468c-af43-1870867518f8
autotaskformer-searching-vision-transformers
2304.08756
null
https://arxiv.org/abs/2304.08756v2
https://arxiv.org/pdf/2304.08756v2.pdf
AutoTaskFormer: Searching Vision Transformers for Multi-task Learning
Vision Transformers have shown great performance in single tasks such as classification and segmentation. However, real-world problems are not isolated, which calls for vision transformers that can perform multiple tasks concurrently. Existing multi-task vision transformers are handcrafted and heavily rely on human expertise. In this work, we propose a novel one-shot neural architecture search framework, dubbed AutoTaskFormer (Automated Multi-Task Vision TransFormer), to automate this process. AutoTaskFormer not only identifies the weights to share across multiple tasks automatically, but also provides thousands of well-trained vision transformers with a wide range of parameters (e.g., number of heads and network depth) for deployment under various resource constraints. Experiments on both small-scale (2-task Cityscapes and 3-task NYUv2) and large-scale (16-task Taskonomy) datasets show that AutoTaskFormer outperforms state-of-the-art handcrafted vision transformers in multi-task learning. The entire code and models will be open-sourced.
['Mi Zhang', 'Deng Cai', 'Zebin Ren', 'Quanlu Zhang', 'Kan Ren', 'Yuge Zhang', 'Shen Yan', 'Yang Liu']
2023-04-18
null
null
null
null
['architecture-search']
['methodology']
[ 2.99784467e-02 -3.62192392e-01 2.27902755e-01 -3.27585340e-01 -8.79662573e-01 -6.38917029e-01 4.51264322e-01 -6.01255178e-01 -7.18591213e-01 3.65937710e-01 -1.11976318e-01 -1.14442997e-01 -7.48721436e-02 -2.32081473e-01 -4.55763698e-01 -7.08082497e-01 5.16618252e-01 7.99320161e-01 7.21017957e-01 3.93497273e-02 9.62832123e-02 -3.08197532e-02 -1.50986385e+00 9.80138034e-02 9.18961585e-01 9.05792952e-01 9.83260751e-01 7.83363223e-01 -6.48994073e-02 5.41919708e-01 -6.97283685e-01 -6.54052138e-01 3.40748578e-01 2.26620182e-01 -8.06131840e-01 9.04160589e-02 6.92670643e-01 -2.27032229e-01 -3.53089254e-03 9.28553283e-01 7.85885811e-01 1.57139361e-01 5.29262662e-01 -1.39123750e+00 -7.45694280e-01 3.84218842e-01 -7.04146504e-01 3.73198748e-01 -4.86910701e-01 5.54126441e-01 1.14465177e+00 -1.01342642e+00 -7.26059370e-04 1.23441362e+00 6.00886524e-01 5.61903238e-01 -9.27983284e-01 -1.00585020e+00 3.66858214e-01 3.29036295e-01 -1.07961988e+00 -5.14219582e-01 3.79291147e-01 -5.80726147e-01 1.43110442e+00 -1.20696589e-01 4.84765887e-01 1.24187076e+00 6.07810654e-02 9.32247400e-01 1.03277946e+00 -1.47152334e-01 1.11023434e-01 -1.75094143e-01 1.26016021e-01 1.08619499e+00 2.32748672e-01 -2.36365080e-01 -7.11619854e-01 -4.58670296e-02 8.77701283e-01 8.96900743e-02 1.20739546e-02 -8.18216950e-02 -1.33561218e+00 9.26322460e-01 1.97219521e-01 2.87231773e-01 -3.10833305e-01 3.93002480e-01 5.57728827e-01 9.45408717e-02 3.42815906e-01 5.67475080e-01 -8.47146571e-01 -6.36589993e-03 -7.74201214e-01 1.11744925e-02 5.66421032e-01 1.02908468e+00 9.38566089e-01 3.96706909e-01 -4.51927364e-01 1.13742685e+00 2.75326073e-01 6.79100513e-01 8.77183735e-01 -8.00769866e-01 4.39141870e-01 4.79878992e-01 -5.16726077e-02 -5.01136363e-01 -4.78253990e-01 -3.47670019e-01 -8.78176570e-01 1.46470159e-01 2.80081660e-01 -4.24196929e-01 -1.45790219e+00 1.39477611e+00 2.39547372e-01 4.85244632e-01 7.98357129e-02 1.03174019e+00 1.02252102e+00 3.27353358e-01 8.63364562e-02 2.20729545e-01 1.67314017e+00 -1.81256247e+00 -3.50451171e-01 -9.57184672e-01 2.65596900e-02 -7.48831391e-01 1.29122174e+00 5.23660243e-01 -9.40460622e-01 -7.34067798e-01 -8.23776186e-01 -1.16611138e-01 -3.57995152e-01 3.20963562e-01 8.86967957e-01 4.69072819e-01 -1.17085361e+00 4.37106900e-02 -7.50540078e-01 -2.80514657e-01 5.60571432e-01 2.55939811e-01 -1.11569837e-01 2.82821842e-02 -9.21517551e-01 1.06286192e+00 3.09356213e-01 8.75251964e-02 -1.56338978e+00 -5.61840653e-01 -6.77179515e-01 2.78767049e-01 6.36176586e-01 -7.81149030e-01 1.65844035e+00 -8.45380902e-01 -1.31178224e+00 9.15614605e-01 -1.05061144e-01 -3.16570699e-01 2.91850060e-01 -4.26213324e-01 -2.01780051e-01 -4.05471511e-02 3.61977488e-01 8.07261407e-01 1.39888072e+00 -1.00607836e+00 -9.47579563e-01 -4.48641062e-01 5.26390597e-02 3.14304650e-01 -8.19345713e-01 2.18381122e-01 -9.45570052e-01 -4.71939147e-01 -4.44297314e-01 -8.34909976e-01 -4.52850491e-01 5.13304621e-02 -3.60331059e-01 -4.13491309e-01 8.55443835e-01 -3.25999677e-01 7.82251596e-01 -1.96029782e+00 1.12068228e-01 -3.93421739e-01 5.50281823e-01 5.54486990e-01 -3.43760192e-01 -1.09905675e-01 2.56145447e-01 -5.89976124e-02 1.56641658e-02 -7.35191226e-01 -5.72737046e-02 3.55571806e-01 9.15560424e-02 1.57155871e-01 -1.71288759e-01 9.95284617e-01 -7.76673377e-01 -7.01784313e-01 2.87045479e-01 1.57650426e-01 -4.50276434e-02 1.81140050e-01 -2.05940872e-01 9.62952748e-02 -6.05966091e-01 9.04281318e-01 3.74143660e-01 -7.21034169e-01 -1.99778467e-01 -3.61221492e-01 -7.86807239e-02 -3.35059822e-01 -7.77812481e-01 2.11295342e+00 -6.98218465e-01 5.60573995e-01 2.74965048e-01 -7.90389895e-01 8.34943354e-01 3.39713871e-01 2.09802836e-01 -6.03914082e-01 9.65647399e-02 -2.13542841e-02 -4.98769730e-02 -6.16431594e-01 4.56943005e-01 2.77900577e-01 -2.36083046e-02 4.51509356e-01 5.07663727e-01 -1.29053622e-01 1.70975089e-01 -4.14997414e-02 1.19715357e+00 -1.73712030e-01 2.44217709e-01 -1.17493749e-01 1.05745442e-01 1.37304619e-01 6.59995914e-01 7.98108101e-01 -6.17676795e-01 6.47155344e-01 -1.39270037e-01 -6.20496094e-01 -8.80289018e-01 -9.03864443e-01 3.74736965e-01 1.72533512e+00 1.82960611e-02 -1.28356040e-01 -5.39351702e-01 -6.72555387e-01 9.35928673e-02 3.40639651e-01 -6.34607852e-01 1.53080434e-01 -2.76814371e-01 -8.94649863e-01 7.23795593e-01 7.12327778e-01 8.15906644e-01 -1.22817326e+00 -1.24536395e+00 2.47187614e-01 -1.59109011e-01 -1.44611549e+00 -7.45889723e-01 3.66830140e-01 -5.94765484e-01 -1.26989722e+00 -1.03972638e+00 -9.38536346e-01 2.83023715e-01 7.16903567e-01 1.21261668e+00 -2.12339222e-01 -4.98995721e-01 3.75478417e-01 -1.91575840e-01 -7.57269442e-01 3.98892879e-01 2.46153861e-01 -4.76412103e-02 5.98744713e-02 2.32476562e-01 -4.92458045e-01 -5.82499027e-01 5.10457695e-01 -4.66753721e-01 2.27695212e-01 9.74596679e-01 9.70655859e-01 6.35622978e-01 -1.69608966e-01 5.27698398e-01 -7.49366760e-01 7.68062651e-01 -1.74687937e-01 -7.61553764e-01 8.67739618e-01 -6.85633123e-01 -5.34962080e-02 4.57773775e-01 -6.60958350e-01 -1.10055673e+00 2.79726207e-01 1.60142109e-01 -1.16488504e+00 -1.70568869e-01 1.64693043e-01 1.29720449e-01 -1.61379263e-01 5.69170058e-01 3.79478872e-01 -3.53130519e-01 -4.55038816e-01 4.01128799e-01 5.49459457e-01 7.33414054e-01 -3.77354681e-01 5.64075708e-01 3.22243661e-01 -3.45605344e-01 -6.45903230e-01 -1.13918984e+00 -6.02373421e-01 -5.49568653e-01 -2.89054513e-01 1.22873735e+00 -1.26087463e+00 -7.18838811e-01 9.39305663e-01 -1.18434095e+00 -6.87500656e-01 3.15274775e-01 2.31057018e-01 -3.91038537e-01 5.65080866e-02 -2.42322043e-01 -7.48211384e-01 -1.09582710e+00 -1.31630981e+00 1.23129523e+00 5.50453186e-01 1.83220282e-01 -9.02408302e-01 -1.15733564e-01 6.84929311e-01 6.19305253e-01 -1.70633242e-01 5.53383350e-01 -6.57657683e-01 -5.04234731e-01 3.11163366e-01 -5.25482535e-01 1.82249829e-01 6.88420311e-02 -1.93540081e-01 -1.29223311e+00 -4.56189752e-01 -2.15795726e-01 -8.37766528e-01 1.25212610e+00 6.61614716e-01 1.32553291e+00 9.46526602e-03 -3.75033349e-01 8.00378323e-01 1.40022016e+00 3.52244899e-02 1.68358192e-01 3.46203655e-01 1.01078248e+00 3.41600239e-01 5.10023952e-01 4.92775321e-01 8.22001815e-01 5.37296414e-01 7.55098879e-01 -3.09794098e-01 -4.90148477e-02 2.12181449e-01 1.00185700e-01 5.96964657e-01 -1.20807469e-01 -1.26546532e-01 -1.20827711e+00 8.76136422e-01 -2.18074703e+00 -6.32787168e-01 1.56749174e-01 1.65549695e+00 6.29389226e-01 2.92354077e-01 2.35580996e-01 -4.08324838e-01 6.73555791e-01 2.89576769e-01 -1.20894742e+00 -1.46687865e-01 3.33859287e-02 7.74929672e-02 5.88793218e-01 7.65747726e-02 -1.32554305e+00 1.36092675e+00 6.27856922e+00 9.27826524e-01 -1.08371902e+00 7.03970850e-01 4.62990314e-01 -3.11502635e-01 1.42233148e-01 -3.38663906e-01 -9.60301220e-01 2.70209908e-01 5.82954466e-01 -7.48596191e-02 4.76646125e-01 1.14850831e+00 -1.00822069e-01 -6.78626075e-02 -6.29456282e-01 1.43377340e+00 2.50481397e-01 -1.08115458e+00 -1.15153112e-01 -3.30002636e-01 5.96260309e-01 7.42653310e-01 7.30749667e-02 4.26436871e-01 9.18084919e-01 -8.98135662e-01 7.10890412e-01 1.80645227e-01 9.64175165e-01 -5.08082390e-01 4.73212540e-01 3.02054018e-01 -1.48960745e+00 -4.23329085e-01 -5.15857458e-01 2.50784576e-01 1.91912092e-02 5.14941692e-01 -6.98744953e-01 2.33426780e-01 1.25837314e+00 6.62627697e-01 -7.65447378e-01 1.18916368e+00 -3.03989202e-01 4.17008877e-01 -1.75539419e-01 -1.01540796e-01 4.40067112e-01 1.02298900e-01 9.63824764e-02 1.18666983e+00 2.60323375e-01 -1.39117032e-01 5.20771146e-01 5.95936596e-01 6.83298893e-03 -3.03422570e-01 -3.96521240e-01 -1.34945670e-02 6.09571218e-01 1.88121665e+00 -6.50838017e-01 -3.59208763e-01 -7.75560796e-01 1.18760014e+00 6.15131855e-01 4.74972129e-01 -7.96752691e-01 -2.31911570e-01 9.78490651e-01 -5.90418935e-01 6.44854009e-01 -3.17782074e-01 -4.51013803e-01 -1.19441640e+00 -2.14215219e-01 -7.50066817e-01 4.93942887e-01 -1.06264532e+00 -1.37597990e+00 8.81818831e-01 -1.31234393e-01 -8.04321706e-01 3.84775810e-02 -6.27090394e-01 -9.97737944e-01 8.42949510e-01 -1.71727252e+00 -1.64975464e+00 -6.35807693e-01 1.01189566e+00 1.27236450e+00 -6.45903230e-01 8.64401460e-01 1.43595979e-01 -9.34156001e-01 7.23292351e-01 -2.09006563e-01 1.19064495e-01 8.60587239e-01 -1.28934526e+00 7.74627566e-01 8.63900840e-01 1.29661471e-01 8.89631659e-02 3.82373929e-01 -3.52058351e-01 -1.42019570e+00 -1.28487241e+00 4.39283758e-01 -3.28883410e-01 7.71872461e-01 -4.26295251e-01 -6.34574771e-01 7.32282102e-01 4.72732246e-01 1.16875738e-01 4.81282294e-01 2.91544110e-01 -4.99822289e-01 -2.72576869e-01 -9.64365542e-01 4.21338350e-01 1.10100365e+00 -4.61422831e-01 -4.96429652e-01 5.14350832e-01 8.47663164e-01 -3.79141927e-01 -5.22345066e-01 2.92767491e-02 6.37615323e-01 -7.61183083e-01 9.69168305e-01 -4.92408097e-01 2.59818166e-01 -2.15171836e-02 -8.50800723e-02 -1.51309502e+00 -5.93033493e-01 -4.64788318e-01 -2.67459631e-01 9.54671741e-01 6.63980186e-01 -7.89280593e-01 6.83138430e-01 4.74682570e-01 -5.00792444e-01 -8.82345319e-01 -9.49878275e-01 -7.17123926e-01 -3.42704833e-01 -1.94141313e-01 4.23708946e-01 8.10524404e-01 -6.85700059e-01 7.98166215e-01 -6.40387833e-01 1.52750000e-01 8.05401325e-01 2.28634521e-01 7.12373912e-01 -1.45786476e+00 -5.88767946e-01 -4.50498730e-01 4.80850972e-02 -9.08707500e-01 9.28730667e-02 -5.33389091e-01 1.21223457e-01 -1.69015336e+00 3.48934531e-01 -3.57496887e-01 -3.22372198e-01 1.25888193e+00 -3.78340811e-01 -2.03930661e-02 2.55808532e-01 3.54605466e-01 -9.94334221e-01 5.01762152e-01 1.29645669e+00 -2.97004968e-01 2.45323088e-02 5.05760089e-02 -9.14377749e-01 8.11392605e-01 8.06069493e-01 -3.96089882e-01 -6.20274901e-01 -1.20214891e+00 -1.47917077e-01 -2.00457811e-01 3.54768127e-01 -1.20354807e+00 6.35505438e-01 -3.59874666e-01 3.97890449e-01 -4.92701441e-01 6.67317033e-01 -7.21906126e-01 -2.65182823e-01 2.08069399e-01 -9.84741300e-02 5.56182086e-01 2.44891167e-01 5.03055692e-01 1.62137002e-02 -1.88805893e-01 6.99549139e-01 -5.38510561e-01 -1.39640164e+00 6.32221878e-01 -9.40111876e-02 2.29005381e-01 1.07341647e+00 -9.47091058e-02 -7.45713413e-01 -8.57847109e-02 -5.35279453e-01 8.34462345e-01 -2.40475908e-02 6.05109870e-01 8.15892041e-01 -9.00821209e-01 -6.77754223e-01 -2.45192274e-02 2.47267812e-01 2.78946131e-01 3.21713716e-01 5.74628413e-01 6.08139411e-02 2.74834871e-01 -4.13516074e-01 -5.97631454e-01 -1.50844824e+00 2.95898139e-01 4.77597654e-01 -4.49286699e-01 -5.18888950e-01 1.27013385e+00 3.25800180e-01 -3.71443957e-01 3.90310496e-01 -1.94026932e-01 -1.92319125e-01 2.62497991e-01 5.60185790e-01 2.39432707e-01 -9.27771442e-03 -2.07847074e-01 -3.63747060e-01 5.95594347e-01 -3.57191712e-01 -3.37273162e-03 1.40611899e+00 -1.67646073e-02 2.38606140e-01 3.04352224e-01 6.75377488e-01 -9.13864255e-01 -1.61463261e+00 -4.69127119e-01 -2.17444837e-01 -2.79718548e-01 3.36051434e-01 -1.01686990e+00 -1.46204841e+00 1.02681684e+00 7.16720581e-01 -7.23932311e-02 1.31981719e+00 6.44537620e-03 9.16942537e-01 7.93175459e-01 4.73023981e-01 -1.31326866e+00 5.46539068e-01 7.72953510e-01 8.29006195e-01 -1.53310001e+00 -1.96591064e-01 -7.82670081e-02 -1.11125946e+00 8.33200336e-01 1.34525049e+00 1.12428144e-01 6.11996412e-01 3.51543635e-01 3.07472616e-01 -5.20779610e-01 -1.00782657e+00 -6.10672474e-01 1.09385692e-01 8.03433478e-01 1.15169985e-02 5.38310371e-02 4.38684344e-01 6.72806859e-01 6.69421554e-02 -6.17226660e-02 2.39833415e-01 7.96607673e-01 -6.88783050e-01 -6.32819474e-01 -2.73156852e-01 7.46586204e-01 -3.08851272e-01 -2.73076802e-01 -2.86511302e-01 3.74511242e-01 2.18465790e-01 1.05207205e+00 -7.95154944e-02 -5.22637069e-01 3.51189077e-01 -7.18780160e-02 3.36764455e-01 -7.20530510e-01 -9.12441850e-01 4.96019013e-02 1.04906268e-01 -4.12413508e-01 -4.86538947e-01 -3.03716630e-01 -9.10974562e-01 3.07404157e-03 -3.69161367e-01 -3.73002350e-01 7.09578454e-01 1.03877068e+00 4.04190451e-01 7.84490824e-01 2.57927746e-01 -1.02852571e+00 -5.62024891e-01 -1.31134319e+00 -4.84523833e-01 -1.40384573e-03 2.70254284e-01 -9.17695522e-01 1.00845106e-01 -4.42557735e-03]
[9.741081237792969, 1.5901906490325928]
c6f3e0d9-58ee-4fcd-8dad-503ae991cd5b
trust-aware-resilient-control-and
2305.16818
null
https://arxiv.org/abs/2305.16818v2
https://arxiv.org/pdf/2305.16818v2.pdf
Trust-Aware Resilient Control and Coordination of Connected and Automated Vehicles
We address the security of a network of Connected and Automated Vehicles (CAVs) cooperating to navigate through a conflict area. Adversarial attacks such as Sybil attacks can cause safety violations resulting in collisions and traffic jams. In addition, uncooperative (but not necessarily adversarial) CAVs can also induce similar adversarial effects on the traffic network. We propose a decentralized resilient control and coordination scheme that mitigates the effects of adversarial attacks and uncooperative CAVs by utilizing a trust framework. Our trust-aware scheme can guarantee safe collision free coordination and mitigate traffic jams. Simulation results validate the theoretical guarantee of our proposed scheme, and demonstrate that it can effectively mitigate adversarial effects across different traffic scenarios.
['Wenchao Li', 'Christos G. Cassandras', 'Wei Xiao', 'Ehsan Sabouni', 'H M Sabbir Ahmad']
2023-05-26
null
null
null
null
['navigate']
['reasoning']
[-0.56626046 0.2917245 0.06731325 -0.10059398 -0.18381959 -1.0001694 0.5588755 -0.11051565 -0.37606773 0.95283943 -0.4878059 -0.6376267 0.10025848 -1.086412 -0.71504223 -0.64098656 -0.6472055 0.32367536 0.8336023 -0.8016122 -0.17199557 0.6720342 -0.91107434 -0.64404845 0.9643093 0.8284102 -0.5224172 0.5965378 0.42828998 0.9135233 -0.9386806 -0.7586005 0.6666163 0.3516838 -0.5186075 -0.2778442 -0.0949757 -0.5321629 -0.67439973 1.1331036 0.05193804 -0.05428143 0.4538074 -2.8143303 -0.03512203 0.6008828 -0.75834936 -0.09765644 0.10137592 0.45378524 0.3609793 -0.1412688 0.42737493 1.5116227 0.7217911 0.6978658 -0.8388371 -1.4918151 0.27612826 0.14585622 -1.3997242 -0.5825653 0.62138987 0.02504945 0.37533477 0.2996757 0.40077585 0.96873844 0.73635995 0.47555783 0.6570502 0.3403133 0.4113559 -0.02148376 -0.40912142 0.53980184 1.017484 0.538672 0.01280983 -0.87140924 0.2666844 -0.24237989 -0.0176075 -0.4163802 -1.0339285 0.9092531 0.7007877 -0.31293613 -0.5480487 0.5984092 0.6342177 0.8033495 -0.09135514 -0.07923896 -0.37017655 0.3164757 -0.22023472 0.36472067 0.9832278 1.3657329 0.5987286 0.4135038 0.24182245 -0.27410215 0.77828425 1.1204894 -0.6186047 -1.5595065 0.40814313 0.09465058 0.4628791 -1.5295461 -0.3918473 0.11048639 -0.897093 0.9156946 0.2309495 -0.9559681 -0.35403317 1.9673538 0.7222218 0.43945286 0.51322967 0.84616035 0.14498582 0.3822397 0.3815465 -0.4701416 0.6696105 -0.61444575 -0.929691 -0.13706999 0.42193243 -0.5317254 -0.15759791 -0.08440924 -0.94637483 0.3316142 -1.1091032 0.7665764 -0.16741674 -1.1048796 0.53715307 0.90892285 -0.91063565 0.02937671 -0.8176717 0.03664837 0.5954691 0.6436061 -0.30188072 0.07347324 -1.5689495 1.0128971 0.01945952 0.0470037 -1.4735248 -0.6461594 -0.8748538 -0.5416854 0.9683855 -0.5558884 1.0571469 -0.3996885 -1.016535 0.14967625 0.18666793 -0.9796017 0.84011394 0.21578245 -0.64570004 0.24892087 0.4535776 0.2440288 0.41653475 -1.7557967 -0.61042535 0.01479445 0.4464232 -0.10672963 0.05844888 0.180606 0.18393289 -0.33544493 -0.3095111 -1.3109295 -0.87037265 0.5666873 -0.42081502 0.0459977 1.4487331 0.42184046 0.55973315 -1.865568 -0.5083423 0.6294306 0.6546881 0.44584763 -0.20022802 0.5961209 0.6541273 0.45057267 0.37755704 0.03279196 0.17731561 0.68101513 -0.40952542 1.0106903 -0.11004314 0.63227093 -1.1999608 -0.5183472 0.12537703 0.42368755 -0.4480995 0.15112026 0.10548412 0.60004497 -1.0613595 0.6810428 1.2668983 0.47567263 0.19837399 0.06381286 -0.04421618 -0.32549334 -1.0020964 0.4246852 -0.20181394 0.62310946 0.9570558 -0.6158067 0.6355335 0.5265921 0.60278296 -0.4331968 0.8685119 0.18697827 0.03492289 -0.10798898 0.31852883 -0.04003056 -0.640061 0.6973686 -0.4134547 0.09301594 -0.27060664 0.88846874 1.371386 -0.8316934 -0.02340594 -0.10924219 0.7570761 0.16334671 1.0549273 1.0322758 -1.1189439 -0.63581145 0.40448976 -0.28288773 -0.68580145 -1.1586372 0.23285134 0.6632802 1.0415893 -0.14606741 -0.43877947 -0.91014004 0.4978213 0.58101016 -0.41508505 -0.63099086 -0.47769383 0.01144147 1.2699721 0.07846737 0.8406269 -0.49486944 -0.2169906 0.27853987 -0.15153603 -1.5860691 -0.5496087 -0.48191118 0.12830687 -1.7520484 0.29867864 -0.3195266 0.7315744 1.2325921 0.51212156 0.6459059 0.14514 0.67184883 -0.1940195 -0.81602347 -1.052442 -0.57719517 0.8623989 0.05706337 0.19355032 -0.5438536 -0.6323807 1.0203441 -0.6870781 -0.6425492 -0.14762391 0.17895108 0.07236606 0.5758144 0.85052204 -0.35943666 0.66756344 -0.96914274 -1.010484 -0.00746257 -0.4887994 -0.49819112 0.8631981 -0.33065063 -0.6878711 -0.2571791 0.5242119 -0.5808015 -0.11120522 -0.21412612 -0.22531237 -1.172877 0.39114302 -0.4737023 0.37724486 0.48710704 0.40873826 0.6420976 0.31567827 -0.47439852 1.8328824 0.927407 0.655848 -0.5556036 -0.17534769 0.14102234 -0.01945413 -0.69593513 0.5201124 -1.1315337 -1.5714743 0.6901837 -1.268083 -0.03163547 0.22836176 0.3918473 -0.42528468 0.60933244 -0.53352404 -1.1126845 -0.09774033 -0.9909938 0.12236886 0.38080767 0.17844605 -0.947063 -0.20675345 0.36723018 0.91585034 0.96211994 0.04924516 -0.6066771 -0.9030438 -0.43499365 -0.00894978 -0.1324055 0.05759577 0.2532475 -0.20227113 -0.6075834 -0.19574848 -0.44071537 0.05655846 -0.19365528 0.1963638 -0.98150575 -0.6056834 0.20506474 1.0390828 0.25771454 0.3782502 0.3415093 0.49424145 0.59769905 1.134737 0.45307285 0.9168097 0.42033732 1.3995609 0.0589707 0.42800656 -0.34017584 0.5316517 -0.0106057 0.21245322 -0.41336003 -0.7093642 0.57073975 -1.80491 -0.9751364 -0.5601256 1.8204275 0.49501264 0.33094865 0.18048403 0.13730787 0.91398394 0.07268263 -0.45585257 -0.53393346 -0.02502884 -0.58030534 1.2534525 0.7500697 -0.8881282 1.1942323 6.625389 0.6151201 -0.5215146 0.24683024 -0.03065828 0.11474421 -0.19013998 0.3063847 -0.27369535 0.17885484 0.92634714 -0.9077753 0.4644895 0.8229763 0.61330134 0.00960938 -0.40274847 0.12149535 -0.432273 -1.2101495 -0.14205852 0.19892736 0.78118527 0.3818544 -0.25102124 0.05470434 1.3631974 -0.7580974 0.6840081 -0.00688275 0.2855029 -1.473875 0.71262693 0.3101936 -1.4793419 -0.33461255 -0.24823086 0.06057547 0.6573573 0.06246103 -0.2778338 0.46144897 0.39080063 0.17100087 -0.05827424 0.82081264 -0.35063487 0.28616303 -0.47796735 0.05435361 0.54010767 0.0688021 1.432102 0.52481973 -0.40395612 0.3441809 0.4946058 0.6799023 -0.06432841 -0.58019316 -1.261153 0.5810945 1.3440528 1.1245114 -0.2450977 -0.04045155 -0.1776685 0.44162667 -0.2216701 0.5422905 -1.2253711 -0.70431924 1.5797912 -0.02170294 0.10279873 -0.49528575 -0.09165599 -0.68906754 -0.23848234 -0.70678455 0.10283081 -0.22800621 -1.17885 0.5858178 -0.22949798 -1.5056909 0.17364064 0.37298173 -1.17459 0.14032488 -1.638155 -1.0641314 -0.02080335 1.3297907 -0.30162382 -0.43234706 0.41080198 0.12954281 -0.5880473 0.8737099 -0.23758394 0.18135001 0.42045993 -0.41431513 0.22686057 1.0798086 -0.627807 0.4196101 0.997938 -0.7970334 -1.8221114 -1.6346867 0.4909553 -0.27069446 1.0784649 -0.17489314 -0.38214 0.77928215 0.17805132 0.28828683 0.44585037 -0.783503 -0.5509945 -0.20671558 -1.9880108 0.8720026 0.9243218 -0.28651607 -0.0449959 0.43358767 1.1741495 -0.20454249 -0.52056885 0.32807288 0.2997052 -0.66235137 0.71491045 -0.730391 -0.7994242 -0.77837706 -0.07851094 -1.160936 -0.01268618 -1.4013205 -0.08460588 1.0581448 0.02287273 -1.2064904 0.38543558 0.766526 0.2851613 0.0519101 -1.4417481 -1.114329 0.17010236 -0.6379477 0.81487066 0.8994355 0.14183071 -0.09997126 -0.5308099 1.0761408 1.4538442 -0.58294344 1.1934181 -0.97176725 0.70069206 -0.07302694 -0.7716378 -0.32411665 0.56329894 -0.36294627 0.11274758 -0.9108538 -0.50534314 -0.86707675 -0.0164613 0.66383 0.39137965 0.40961656 0.39333022 0.086356 -1.1472689 0.59766215 1.0558548 -0.27542683 0.0335402 0.27227467 -0.88650274 0.7922306 1.1122296 -0.84853685 -0.55790496 -0.22181419 -0.06619977 0.6208741 0.60517573 -0.8595826 0.6206651 -0.74494165 -0.61144894 -0.36380893 -0.10450902 -1.5165355 0.32044548 0.9688096 0.18153629 -0.19316073 0.4213858 1.2371438 0.06320181 0.7138835 0.8143206 0.45771283 -0.03736999 0.64430827 -1.1881857 0.04187036 1.9271898 0.04902009 -0.81103176 -0.9638365 -0.5650445 1.4664783 0.43954974 0.7141831 0.547531 -1.4277698 -0.7604906 0.07579632 -0.16022244 -0.34708703 0.0651884 0.9131501 -0.31763732 0.10813726 -0.38549712 -0.09723889 -1.4303753 0.6333405 0.558493 0.20315303 -0.04530216 0.2959001 -0.30294192 -0.29447696 0.48326367 0.5065512 0.05818251 -0.5638897 0.518965 0.818374 -0.49606657 -1.3024937 -0.82786196 0.26700005 -0.1480864 0.02267732 0.68327075 -0.7815801 -0.30514315 -0.5426329 0.7983314 0.11370602 -1.1269848 -0.16765416 -0.46466333 -0.7403731 0.12334827 -0.3218403 -1.6853878 -0.07692306 0.10736437 0.56705123 0.65429556 -0.29971474 1.1728928 0.5743414 1.2062128 -0.8436396 -0.29381847 0.6058351 0.47716105 -1.1572356 -0.3777996 -0.8028281 -0.7914302 0.6900814 1.0538831 -0.4934207 0.90495133 0.46858662 0.34784612 -0.36603475 -0.91303253 -0.13374098 -0.6768744 1.1936519 -0.82951146 0.23497112 -0.3559164 0.5244304 0.0284224 -0.30220324 1.1818432 1.1163975 -0.6043756 -1.1672429 -0.5123448 -0.173069 -0.3562476 0.5571562 -0.6072673 0.77437663 0.06741542 1.9504131 -0.21614072 -0.84380996 0.3115256 -0.8861857 -0.3335547 0.25074303 -0.48339874 -0.39927244 0.24317834 -0.9932159 -0.35044232 -0.24041088 -1.6055794 -1.3773035 -0.5205106 0.55693185 0.33566657 0.71322954 0.29109436 0.12182353 1.8642513 -0.3888378 -0.38042784 -0.22279054 -0.49245453 -0.09932379 0.8670407 -0.8540281 -1.0099607 -0.7620034 ]
[5.347891807556152, 7.2378458976745605]
a4edfd43-70b0-4600-bbb5-abaf7025119b
study-of-robust-sparsity-aware-rls-algorithms
2204.08990
null
https://arxiv.org/abs/2204.08990v1
https://arxiv.org/pdf/2204.08990v1.pdf
Study of Robust Sparsity-Aware RLS algorithms with Jointly-Optimized Parameters for Impulsive Noise Environments
This paper proposes a unified sparsity-aware robust recursive least-squares RLS (S-RRLS) algorithm for the identification of sparse systems under impulsive noise. The proposed algorithm generalizes multiple algorithms only by replacing the specified criterion of robustness and sparsity-aware penalty. Furthermore, by jointly optimizing the forgetting factor and the sparsity penalty parameter, we develop the jointly-optimized S-RRLS (JO-S-RRLS) algorithm, which not only exhibits low misadjustment but also can track well sudden changes of a sparse system. Simulations in impulsive noise scenarios demonstrate that the proposed S-RRLS and JO-S-RRLS algorithms outperform existing techniques.
['B. Chen', 'R. C. de Lamare', 'Y. Zakharov', 'L. Lu', 'Y. Yu']
2022-04-09
null
null
null
null
['2d-human-pose-estimation']
['computer-vision']
[ 1.25493497e-01 -4.80903447e-01 8.97863656e-02 8.94065108e-03 -9.77427781e-01 -2.81156301e-01 5.66535778e-02 -3.08011055e-01 9.83643830e-02 7.83093870e-01 -2.22396408e-03 -5.43107092e-02 -6.00328028e-01 -1.70444936e-01 -6.20281816e-01 -8.42008591e-01 -4.16676044e-01 -1.43152222e-01 2.60129541e-01 -3.15935344e-01 1.44301057e-01 5.10474920e-01 -6.91785693e-01 -5.42728722e-01 9.95716572e-01 1.05116355e+00 3.34480017e-01 6.51609123e-01 6.06691360e-01 7.46723294e-01 -1.66473582e-01 4.52758521e-01 9.04286742e-01 -4.55441803e-01 2.26154834e-01 3.68146658e-01 1.22024819e-01 9.29921344e-02 -9.13305700e-01 1.20255637e+00 5.97967923e-01 4.74757463e-01 2.79315114e-01 -9.50941205e-01 -4.72961754e-01 6.24024212e-01 -1.03653240e+00 6.71600819e-01 3.06223601e-01 1.59958974e-01 4.43393111e-01 -1.33452916e+00 5.84181957e-02 1.37741554e+00 1.07381451e+00 2.72837225e-02 -1.42272604e+00 -8.29685569e-01 3.66494864e-01 -1.44232571e-01 -2.02229571e+00 -9.38946187e-01 8.46661985e-01 -3.38168263e-01 4.80914950e-01 3.53947997e-01 9.09948051e-02 7.23720908e-01 -3.54379453e-02 6.43881977e-01 1.04150593e+00 -2.30136216e-01 3.62288028e-01 -3.87960114e-02 4.00574803e-01 6.66894794e-01 9.53089535e-01 -2.39415877e-02 -3.97696823e-01 -4.41467822e-01 1.11186266e+00 8.33402947e-02 -3.45739633e-01 -7.80867785e-02 -9.94141400e-01 5.89150906e-01 -1.60077110e-01 3.59128267e-01 -5.26834071e-01 9.93216559e-02 3.12358528e-01 5.93328834e-01 5.30775964e-01 3.49936277e-01 -2.63266444e-01 1.86162233e-01 -1.10741377e+00 3.77653465e-02 9.21961188e-01 1.24275494e+00 4.86270577e-01 1.27839541e+00 -2.03689322e-01 8.32269192e-01 3.25728357e-01 1.09167349e+00 3.15152049e-01 -1.06818902e+00 6.07537866e-01 8.47473294e-02 5.96131384e-01 -1.38068855e+00 -5.75547457e-01 -1.04998326e+00 -1.33624971e+00 -4.62022215e-01 -1.07637914e-02 -6.68183744e-01 -4.25189167e-01 1.70846128e+00 9.24747363e-02 1.05622792e+00 2.02206671e-01 8.77856970e-01 3.91571045e-01 7.89031923e-01 -4.39938933e-01 -1.15468299e+00 3.94388378e-01 -6.53639197e-01 -1.03654945e+00 -5.08582056e-01 1.63608611e-01 -5.17661929e-01 6.06263399e-01 3.40864509e-01 -1.12226582e+00 -5.01452923e-01 -9.33387935e-01 9.58430588e-01 3.68122548e-01 4.29465622e-01 1.26945391e-01 3.90033662e-01 -9.18982923e-01 2.65220016e-01 -5.50277352e-01 -2.35891953e-01 -7.05280751e-02 3.17422211e-01 4.97371070e-02 2.47059435e-01 -1.28692710e+00 6.63664758e-01 1.22166097e-01 5.31318784e-01 -9.03287709e-01 -7.33198166e-01 -8.34253073e-01 -6.06155731e-02 9.55490112e-01 -2.79851079e-01 9.44937110e-01 -6.45821869e-01 -1.80202425e+00 -1.06936581e-01 -5.08305967e-01 -6.49644494e-01 1.77089840e-01 -3.09040755e-01 -8.87475669e-01 6.28437996e-02 1.89201161e-01 -7.50463009e-01 1.43234885e+00 -1.52497578e+00 -2.39208132e-01 1.14467241e-01 -7.15401471e-01 1.44805193e-01 6.36078119e-02 -2.46159554e-01 -2.72318840e-01 -1.04856730e+00 5.71106970e-01 -9.57890809e-01 -9.82705951e-01 -4.26378012e-01 -4.63220358e-01 2.41117448e-01 1.15081465e+00 -5.51763713e-01 1.69103813e+00 -2.32777858e+00 -5.53252213e-02 6.86092734e-01 -1.58731908e-01 5.87986231e-01 -3.34050626e-01 5.73436022e-01 -1.23677142e-01 -2.97083467e-01 -1.39944643e-01 -1.97623134e-01 -1.55085415e-01 1.76615082e-02 -6.77367091e-01 9.77031469e-01 2.80600041e-02 2.30805337e-01 -8.23702931e-01 6.38119057e-02 2.18601614e-01 9.74851325e-02 -2.13192299e-01 -2.34661456e-02 4.53076094e-01 4.60970044e-01 -7.78262198e-01 6.54842138e-01 6.11079693e-01 -3.17001969e-01 -1.21922076e-01 -3.84813815e-01 -4.28320944e-01 -6.03331268e-01 -1.93560386e+00 8.79794598e-01 -4.77452546e-01 4.40685362e-01 5.96807718e-01 -1.29044771e+00 1.21283042e+00 4.57262635e-01 7.73955762e-01 -4.98091817e-01 1.51780441e-01 3.45384270e-01 -4.16954756e-01 -4.27854478e-01 2.34089941e-01 -2.44843531e-02 -1.06737025e-01 1.54084742e-01 -1.61366925e-01 7.74196014e-02 2.52968013e-01 4.17414039e-01 1.01247656e+00 -4.92491156e-01 8.98707569e-01 -3.99795234e-01 1.11293828e+00 -5.93523562e-01 1.15029919e+00 1.33747530e+00 -2.91897744e-01 2.08454609e-01 -1.35114089e-01 1.32558987e-01 -9.15529788e-01 -9.69184875e-01 -1.46418497e-01 7.19622195e-01 4.72252339e-01 4.38996255e-02 4.27975282e-02 -7.35783204e-03 2.00317919e-01 6.13501608e-01 -2.30396360e-01 -2.63720751e-01 -5.53901196e-01 -6.43384039e-01 4.14747298e-01 1.54846504e-01 2.80463398e-01 -1.04382016e-01 9.13740024e-02 5.51227450e-01 1.32390633e-01 -1.38203907e+00 -8.50006342e-01 2.34874114e-02 -7.28808999e-01 -6.67384744e-01 -6.48530662e-01 -5.91220021e-01 8.89906585e-01 9.32353556e-01 3.23726118e-01 -3.01819533e-01 -1.80143645e-04 9.13401306e-01 -3.16589892e-01 -1.83526680e-01 -1.71141431e-01 -4.94385272e-01 1.03156006e+00 7.40711987e-01 -4.94543642e-01 -7.19292879e-01 -1.70617580e-01 4.26606804e-01 -4.25516903e-01 -4.97430027e-01 6.92830801e-01 7.98006594e-01 9.00539398e-01 2.82414824e-01 1.11744487e+00 -6.89797819e-01 8.30818474e-01 -8.11513782e-01 -1.05614221e+00 2.20658079e-01 -8.37212861e-01 -2.16630965e-01 1.13084340e+00 -8.11300814e-01 -1.00731766e+00 2.43424103e-01 3.62018466e-01 -1.02495372e+00 2.95582294e-01 5.75540006e-01 1.86113164e-01 -8.60153973e-01 4.11167532e-01 7.43951797e-01 -1.60741925e-01 -3.55793387e-01 3.21956694e-01 5.11821985e-01 8.39744687e-01 -2.09012493e-01 1.60662282e+00 3.71992797e-01 2.45459020e-01 -1.68963373e+00 -9.46210086e-01 -1.04432535e+00 -3.15852165e-01 -1.73524007e-01 -1.33537233e-01 -1.32411265e+00 -6.33933127e-01 3.79188657e-01 -6.10260963e-01 -2.31183656e-02 -4.88828540e-01 7.00596869e-01 -6.32607341e-01 7.02747822e-01 -3.85524869e-01 -1.63500023e+00 -2.72658527e-01 -5.92483878e-01 5.80240548e-01 3.17624360e-01 1.15089910e-02 -1.04048240e+00 2.24961489e-01 -2.71758199e-01 5.19938171e-01 2.56060511e-01 -7.83599317e-02 -6.20288551e-01 -3.33176285e-01 -3.84607643e-01 -1.66586488e-01 2.56720215e-01 3.28552015e-02 -3.26133341e-01 -2.54851073e-01 -8.19609582e-01 4.07255560e-01 1.93936788e-02 5.65675676e-01 7.90559292e-01 5.99959314e-01 -9.60846603e-01 -1.84759170e-01 8.38000000e-01 1.83600187e+00 3.08045208e-01 2.47900352e-01 2.55901963e-01 5.75112820e-01 -3.36341918e-01 5.56509614e-01 1.13187432e+00 2.42155530e-02 3.70614350e-01 4.38590385e-02 -6.31610528e-02 3.48823339e-01 9.45460647e-02 7.57933557e-01 1.49991035e+00 9.98018384e-02 8.57166424e-02 -2.12333485e-01 5.17964005e-01 -2.04654026e+00 -1.19409275e+00 -3.50124866e-01 2.15919185e+00 7.06028223e-01 -8.49698558e-02 -2.03638282e-02 1.68356806e-01 1.02126443e+00 4.23033357e-01 -1.08841264e+00 -7.98873976e-02 -4.21726614e-01 -1.37916788e-01 1.00893795e+00 6.06107235e-01 -1.14848876e+00 5.18661737e-01 7.21306562e+00 8.69910061e-01 -8.18749428e-01 -5.86404428e-02 -9.68563408e-02 1.48467153e-01 2.54677594e-01 -8.20316300e-02 -8.83237958e-01 4.29163426e-01 9.59560633e-01 -8.44926178e-01 7.08127081e-01 7.62891531e-01 1.01204884e+00 -2.62958091e-02 -3.03899527e-01 1.21351373e+00 2.24925622e-01 -9.69231308e-01 -1.72127917e-01 -4.77169752e-01 9.97718573e-01 -6.66022971e-02 5.83119318e-02 1.48921371e-01 3.42344016e-01 -3.09092194e-01 6.05285287e-01 1.10518575e+00 6.50961339e-01 -4.02710021e-01 3.95242512e-01 3.06194365e-01 -1.62638378e+00 -5.60196936e-01 -3.38849604e-01 3.74028757e-02 4.70854431e-01 9.04861629e-01 -3.34603339e-01 7.03614056e-01 2.23420858e-01 1.18141198e+00 -3.89794379e-01 1.48293519e+00 9.95680615e-02 1.06955624e+00 -5.70579171e-01 2.19427079e-01 1.48654595e-01 -4.09282625e-01 1.48014939e+00 1.45445383e+00 7.77468860e-01 6.80873990e-01 7.78939188e-01 5.24088740e-01 4.52677459e-01 -1.80167899e-01 -4.40461308e-01 7.75956213e-02 8.61539960e-01 9.78591025e-01 -2.80987918e-01 -2.68246233e-01 -3.68589550e-01 4.75187927e-01 -4.07793939e-01 8.20858300e-01 -6.00007653e-01 -5.97113192e-01 4.46798772e-01 -1.68727800e-01 5.78998208e-01 -6.18269563e-01 -2.57431209e-01 -1.12473834e+00 -4.52019334e-01 -9.22342777e-01 1.94442362e-01 -5.64745069e-01 -1.48355758e+00 2.82912254e-01 -2.66499072e-01 -1.74439776e+00 7.00630844e-02 3.29739630e-01 -5.69000959e-01 5.99553406e-01 -1.40520406e+00 -6.38694108e-01 7.68986270e-02 9.47623312e-01 9.03254688e-01 -6.21236145e-01 3.86846364e-01 2.69402266e-01 -1.23101676e+00 4.09882426e-01 8.58847201e-01 -3.35849762e-01 5.56788743e-01 -7.42327631e-01 -1.08891308e-01 1.38500082e+00 -2.30014831e-01 8.24105561e-01 1.38492787e+00 -7.81378746e-01 -1.70774972e+00 -1.58169806e+00 3.46552014e-01 4.00491983e-01 1.12154579e+00 1.25283048e-01 -9.96301770e-01 6.53090596e-01 -2.47802705e-01 4.72437702e-02 2.63714522e-01 -4.01284337e-01 -1.24595121e-01 -5.00392020e-01 -1.01347148e+00 5.77992857e-01 8.56013000e-01 -4.34411734e-01 -4.97883916e-01 4.34856415e-01 5.99239409e-01 -3.92887354e-01 -8.03223670e-01 4.40096617e-01 6.45612404e-02 -4.17243809e-01 1.19390059e+00 -1.36074886e-01 -6.94854200e-01 -7.65525341e-01 -4.04861331e-01 -1.21485078e+00 -8.98940682e-01 -1.46343505e+00 -6.48017704e-01 1.05539954e+00 2.15530351e-01 -8.01265955e-01 1.13405749e-01 9.15850028e-02 -1.65900454e-01 -2.88378239e-01 -1.11111856e+00 -1.46440351e+00 -8.37521434e-01 -2.55755126e-01 -1.55638382e-01 1.02512383e+00 8.54804274e-03 1.12926222e-01 -1.06531453e+00 1.13580120e+00 1.27163267e+00 -1.21564388e-01 6.71224236e-01 -9.03180361e-01 -3.64462078e-01 -1.16034567e-01 -1.20663308e-01 -1.01273322e+00 9.12778229e-02 -4.21879441e-01 2.69861788e-01 -1.13248897e+00 -2.24559784e-01 -4.14199352e-01 -6.42507076e-01 1.45503953e-01 -2.10442454e-01 1.84978127e-01 1.75881281e-01 4.56721604e-01 -1.14757872e+00 6.47169352e-01 7.44847715e-01 6.07610606e-02 -5.79603553e-01 5.38775384e-01 -7.45895743e-01 8.75410557e-01 6.89290762e-01 -3.69139552e-01 -4.44732159e-01 -7.07323551e-02 -1.48907632e-01 5.71571350e-01 9.03731361e-02 -1.38465071e+00 6.51952207e-01 -5.21909714e-01 3.05349194e-02 -6.42818630e-01 1.93625569e-01 -8.88600826e-01 2.53186673e-01 7.00395942e-01 -2.96574533e-01 -2.86217451e-01 -1.80434324e-02 1.27192080e+00 -9.87200364e-02 -1.64919972e-01 9.92988586e-01 2.20067307e-01 -7.94143856e-01 4.02878523e-01 -8.63085806e-01 1.90824136e-01 9.87297535e-01 -2.66267776e-01 -8.37507844e-02 -7.49629438e-01 -8.86953712e-01 5.45922697e-01 -3.22290540e-01 2.35756487e-02 8.32860231e-01 -1.31330967e+00 -7.31061220e-01 3.21182370e-01 -4.02645737e-01 -5.83144605e-01 3.65732968e-01 1.24252439e+00 2.49000832e-01 4.80975658e-01 4.74115372e-01 -3.67420018e-01 -9.31638181e-01 4.19886708e-01 1.50652960e-01 -1.34162828e-01 -6.67678595e-01 6.31718278e-01 -2.35745654e-01 -7.78042302e-02 2.37003833e-01 1.36116818e-01 -6.68987334e-02 -3.58695984e-01 6.20066524e-01 9.55085754e-01 -4.16422278e-01 -9.15055573e-01 -3.12301666e-01 8.49392235e-01 2.17103258e-01 1.77258655e-01 1.56211507e+00 -9.65727627e-01 -7.12842727e-03 7.56823480e-01 7.71862507e-01 3.03470850e-01 -1.43503523e+00 -7.54817486e-01 -1.58988461e-02 -5.24552822e-01 6.72957823e-02 -2.24342912e-01 -1.01438737e+00 -2.40619436e-01 4.45161968e-01 2.70833969e-01 1.23899007e+00 -5.68043113e-01 7.56408811e-01 7.81824231e-01 5.78394353e-01 -1.07203925e+00 1.67196348e-01 1.11315072e+00 9.89970028e-01 -9.35893595e-01 4.52195555e-01 -4.79682922e-01 -6.65545762e-01 8.78314137e-01 3.98367763e-01 -8.49151492e-01 8.14213276e-01 5.14712632e-01 -2.71710515e-01 3.39712918e-01 -8.60460997e-01 -3.33165675e-01 2.26702899e-01 5.58791637e-01 -1.07726783e-01 -1.99663982e-01 -4.33935612e-01 7.97171235e-01 4.02230531e-01 4.03869525e-03 7.45995343e-01 6.57423615e-01 -8.84099662e-01 -3.07871789e-01 -8.95123184e-01 6.25812054e-01 -1.55489728e-01 1.15691178e-01 8.85119140e-02 3.58386427e-01 -4.07647520e-01 1.16051197e+00 -4.41171259e-01 -4.15507585e-01 7.31777728e-01 -5.46098113e-01 1.94308050e-02 -3.81760269e-01 -1.02680504e-01 6.63357079e-01 -4.85667475e-02 -5.20240963e-01 -3.51177543e-01 -9.54604983e-01 -1.01967955e+00 -2.05565561e-02 -7.81638563e-01 4.46616650e-01 6.68682531e-02 9.55482244e-01 2.61439472e-01 5.65680683e-01 1.52879286e+00 -7.89215207e-01 -1.03743994e+00 -6.05615914e-01 -1.24847305e+00 -1.04061522e-01 7.84455359e-01 -4.53043938e-01 -8.74593794e-01 -5.54066040e-02]
[6.560156345367432, 1.5944578647613525]
fbd28611-1cd1-4214-974a-167ad02fae8b
numerical-approximation-in-cfd-problems-using
2111.02987
null
https://arxiv.org/abs/2111.02987v1
https://arxiv.org/pdf/2111.02987v1.pdf
Numerical Approximation in CFD Problems Using Physics Informed Machine Learning
The thesis focuses on various techniques to find an alternate approximation method that could be universally used for a wide range of CFD problems but with low computational cost and low runtime. Various techniques have been explored within the field of machine learning to gauge the utility in fulfilling the core ambition. Steady advection diffusion problem has been used as the test case to understand the level of complexity up to which a method can provide solution. Ultimately, the focus stays over physics informed machine learning techniques where solving differential equations is possible without any training with computed data. The prevalent methods by I.E. Lagaris et.al. and M. Raissi et.al are explored thoroughly. The prevalent methods cannot solve advection dominant problems. A physics informed method, called as Distributed Physics Informed Neural Network (DPINN), is proposed to solve advection dominant problems. It increases the lexibility and capability of older methods by splitting the domain and introducing other physics-based constraints as mean squared loss terms. Various experiments are done to explore the end to end possibilities with the method. Parametric study is also done to understand the behavior of the method to different tunable parameters. The method is tested over steady advection-diffusion problems and unsteady square pulse problems. Very accurate results are recorded. Extreme learning machine (ELM) is a very fast neural network algorithm at the cost of tunable parameters. The ELM based variant of the proposed model is tested over the advection-diffusion problem. ELM makes the complex optimization simpler and Since the method is non-iterative, the solution is recorded in a single shot. The ELM based variant seems to work better than the simple DPINN method. Simultaneously scope for various development in future are hinted throughout the thesis.
['Balaji Srinivasan', 'Vikas Dwivedi', 'Siddharth Rout']
2021-11-01
null
null
null
null
['physics-informed-machine-learning']
['graphs']
[-1.89467862e-01 -9.22486559e-02 1.35867164e-01 -3.15153711e-02 -1.59138188e-01 -3.28527302e-01 6.83251858e-01 7.30157946e-04 -2.86164671e-01 1.08295739e+00 -2.95941472e-01 -2.95014381e-01 -6.41091883e-01 -7.78158665e-01 -2.04639763e-01 -1.13377810e+00 -2.63210267e-01 7.27408648e-01 -1.24239065e-01 -4.54718590e-01 5.24158180e-01 9.92209017e-01 -1.50282824e+00 -8.26372299e-03 9.29414570e-01 8.87018859e-01 8.53480399e-02 8.02044749e-01 -5.87238848e-01 7.63378024e-01 -3.86620820e-01 2.84915835e-01 5.29222727e-01 -5.63740551e-01 -6.53222561e-01 -4.83291030e-01 9.87659991e-02 1.19218469e-01 2.12337691e-02 7.08587825e-01 1.01298320e+00 8.07854116e-01 1.14855659e+00 -1.21865582e+00 -4.47829634e-01 2.24950761e-01 -7.30842173e-01 5.37608206e-01 1.69074833e-01 2.40669712e-01 1.25877380e-01 -9.16879117e-01 3.92935663e-01 1.05754471e+00 8.69184196e-01 4.18821752e-01 -9.89041746e-01 -4.69062060e-01 -2.92700469e-01 2.69772280e-02 -1.39699686e+00 6.94022551e-02 9.24411297e-01 -4.91623044e-01 1.28166997e+00 4.62311476e-01 5.50650001e-01 5.44055343e-01 7.10094810e-01 3.06075513e-01 1.48479521e+00 -5.49976170e-01 7.12124467e-01 4.15853918e-01 1.84882447e-01 6.78958833e-01 2.35357791e-01 5.14525950e-01 -2.69139618e-01 -2.41295084e-01 4.82219666e-01 -1.91873610e-01 -2.34103158e-01 -9.56647694e-02 -2.78438896e-01 1.11876845e+00 2.78013080e-01 7.98938155e-01 -5.27690887e-01 -1.69959962e-01 4.73149836e-01 3.52604270e-01 5.83754301e-01 8.24735463e-01 -5.23275912e-01 -2.70194232e-01 -1.28550720e+00 6.47915125e-01 1.20952308e+00 2.95636684e-01 5.63917935e-01 7.31517494e-01 8.12573358e-02 7.41416991e-01 2.26170093e-01 3.02062064e-01 5.95725894e-01 -1.07220030e+00 -1.41624629e-01 4.05853152e-01 8.57817158e-02 -9.30618346e-01 -4.23020244e-01 -5.35529196e-01 -8.96800518e-01 1.18492603e+00 4.18486327e-01 -6.95584893e-01 -9.62393343e-01 1.18822479e+00 5.67797959e-01 1.43204093e-01 1.10539772e-01 8.90751183e-01 7.10916460e-01 1.16396832e+00 3.05301659e-02 -3.75397176e-01 7.43059278e-01 -8.80725026e-01 -6.19512081e-01 2.52913415e-01 5.46629190e-01 -7.86103189e-01 6.78908587e-01 6.85023546e-01 -1.30632198e+00 -6.24035716e-01 -9.50923562e-01 3.85232896e-01 -8.19019079e-01 -5.08974075e-01 6.64467156e-01 7.38226473e-01 -9.59659696e-01 1.20553672e+00 -7.76617944e-01 -2.36293554e-01 6.34975731e-02 4.91812170e-01 1.27246901e-01 3.33207607e-01 -1.16759682e+00 1.11530387e+00 4.76244211e-01 9.60632414e-02 -4.85610962e-01 -1.00458920e+00 -5.21111250e-01 -1.50269628e-01 -2.04746142e-01 -6.18052721e-01 7.79820979e-01 -1.15937316e+00 -1.92862928e+00 4.65701073e-01 1.03954598e-01 -3.90814722e-01 7.90837407e-01 4.49344590e-02 -2.54045427e-01 4.45151851e-02 -2.62137413e-01 5.30927658e-01 5.69864810e-01 -1.33162880e+00 -4.10187215e-01 1.24426916e-01 -8.35143700e-02 2.15261966e-01 -8.63302201e-02 -2.22031519e-01 5.27275801e-01 -5.27520478e-01 -6.72175735e-02 -9.93759573e-01 -2.23989218e-01 4.51439694e-02 7.36636994e-03 -3.35526645e-01 1.34364557e+00 -4.15466905e-01 8.75385225e-01 -1.62898016e+00 1.88202471e-01 1.98401511e-01 -9.37460363e-02 4.23485637e-01 3.06366026e-01 8.39471102e-01 -3.27208459e-01 -1.54968060e-03 -7.16621101e-01 -1.28889620e-01 -4.02057171e-02 4.42072630e-01 -2.97992647e-01 6.06959283e-01 -1.29327595e-01 5.52484393e-01 -4.19174314e-01 -4.60827798e-01 3.63515884e-01 8.60105157e-01 -3.24050188e-01 2.55250782e-01 -1.17187284e-01 7.39032686e-01 -3.61073464e-01 4.63063627e-01 9.07534361e-01 2.30531439e-01 -6.76299334e-01 5.86259663e-02 -5.94193876e-01 -5.72142839e-01 -1.46351790e+00 1.49629653e+00 -6.89799368e-01 6.14250958e-01 2.74205148e-01 -1.45573843e+00 1.20257485e+00 5.32431126e-01 6.68785036e-01 -4.90552455e-01 4.51208234e-01 4.23542053e-01 2.47239638e-02 -8.83619368e-01 3.76810312e-01 -8.00311267e-01 5.05773544e-01 4.12248701e-01 -9.10521299e-02 -3.03808361e-01 5.44611700e-02 -2.28053063e-01 6.68329000e-01 5.68422258e-01 1.20065600e-01 -8.38564634e-01 6.70959830e-01 3.73574883e-01 2.69249320e-01 8.30079496e-01 -2.91538864e-01 2.86808819e-01 -2.81590456e-03 -8.33894134e-01 -9.83141541e-01 -8.78495574e-01 -3.79592359e-01 7.67960966e-01 5.06826825e-02 3.18277270e-01 -4.92888272e-01 -8.31438527e-02 -8.20654333e-02 1.01433468e+00 -5.57880998e-01 2.07325574e-02 -7.42470384e-01 -1.07635832e+00 4.91176158e-01 1.55116364e-01 7.22861826e-01 -1.17683136e+00 -8.04163933e-01 2.08616719e-01 3.24130505e-01 -4.26116586e-01 5.16371608e-01 3.54793727e-01 -1.28226435e+00 -5.57356298e-01 -1.00784397e+00 -5.46835661e-01 2.31070995e-01 -4.50902581e-01 8.72919023e-01 6.99194744e-02 -5.39470732e-01 2.82600403e-01 -2.46317312e-01 -5.79526365e-01 -4.54141408e-01 -6.71550333e-02 7.89848194e-02 -4.37164485e-01 1.95600808e-01 -1.01091933e+00 -6.48213148e-01 -1.61724001e-01 -7.46953130e-01 -4.25156176e-01 5.20682573e-01 6.68297589e-01 4.24287230e-01 5.69593608e-01 4.90743577e-01 -6.30612075e-01 8.36101353e-01 -7.67841637e-01 -3.16864699e-01 -2.90302068e-01 -8.85066092e-01 1.12443633e-01 1.00670779e+00 -6.37349963e-01 -1.30274177e+00 -3.53847921e-01 -3.47305328e-01 -4.53223884e-01 -4.77463156e-01 4.25467283e-01 4.71071005e-01 -5.93865037e-01 8.83912504e-01 1.76160187e-01 -1.69691853e-02 -5.28290987e-01 -1.25042751e-01 6.49449766e-01 8.63762796e-02 -8.21535110e-01 6.57180011e-01 4.02196139e-01 6.15340412e-01 -1.28428864e+00 -2.93654144e-01 -1.61396250e-01 -3.86850804e-01 -3.29831481e-01 6.28880501e-01 -1.49087921e-01 -5.51120281e-01 4.72905070e-01 -8.99312139e-01 -3.81467730e-01 -4.59288627e-01 5.69908798e-01 -5.58008611e-01 1.89024031e-01 -6.27066195e-01 -1.38520896e+00 -4.58162487e-01 -9.93576050e-01 3.23103994e-01 5.53008378e-01 -2.27501750e-01 -1.62580216e+00 3.49589229e-01 2.80019324e-02 9.05204713e-01 7.93246984e-01 8.83338213e-01 -5.84945738e-01 -1.55813918e-01 -1.48712248e-01 2.33679250e-01 1.31643012e-01 2.38693748e-02 2.90093511e-01 -9.96232629e-01 -3.36301237e-01 4.45167154e-01 -1.43616229e-01 6.02836370e-01 6.65810704e-01 8.98913145e-01 -1.86334386e-01 -3.40726227e-01 6.36377037e-01 1.99902046e+00 5.20023167e-01 3.82771552e-01 5.39792538e-01 2.86641747e-01 5.86435556e-01 2.48587728e-01 2.73139596e-01 -4.17623460e-01 1.98099196e-01 2.37395853e-01 -2.13931933e-01 -4.56959791e-02 4.16229934e-01 5.62591217e-02 5.92130363e-01 -4.14928585e-01 -4.64075148e-01 -1.07528281e+00 3.29629451e-01 -1.39863122e+00 -1.05254328e+00 -4.52378541e-01 1.73858011e+00 4.17987376e-01 2.68669814e-01 -6.47225454e-02 4.75025386e-01 4.97497171e-01 -5.49939238e-02 -3.92637759e-01 -1.37649310e+00 1.10859223e-01 4.40658480e-01 3.14202547e-01 7.36477196e-01 -8.16846311e-01 2.96691626e-01 6.07233953e+00 9.02217686e-01 -1.72081614e+00 3.52134146e-02 5.27682245e-01 -1.43473536e-01 1.12527683e-01 -2.78750882e-02 -5.72916329e-01 6.29367590e-01 1.15788031e+00 1.04976923e-03 2.53136843e-01 7.92504728e-01 5.21931231e-01 -5.42307317e-01 -6.86659157e-01 7.55112112e-01 -1.63441554e-01 -1.21018684e+00 -1.01863062e-02 -1.41549604e-02 9.48551893e-01 8.42205621e-03 -3.54418554e-03 2.42661893e-01 -1.23326853e-01 -1.20860076e+00 2.83705175e-01 8.96832168e-01 4.11441088e-01 -1.01134360e+00 1.01216161e+00 6.03003740e-01 -9.00674284e-01 -2.18186930e-01 -3.19464475e-01 -6.84511483e-01 3.27428907e-01 5.55452406e-01 -3.32282752e-01 6.31575346e-01 6.35910630e-01 1.48138314e-01 1.12700351e-01 1.12927759e+00 4.18282419e-01 7.05270827e-01 -7.53450811e-01 -2.46678919e-01 7.72826254e-01 -4.68553632e-01 9.73168790e-01 1.32221663e+00 4.62412447e-01 2.04674557e-01 1.38788342e-01 1.15798020e+00 7.08382070e-01 3.10364068e-01 -9.04668748e-01 1.86026260e-01 5.24178110e-02 1.19797695e+00 -9.47900414e-01 -1.75639480e-01 -2.07429025e-02 6.67454422e-01 -5.99630736e-02 5.09335756e-01 -7.44334996e-01 -6.12751305e-01 2.93482095e-01 2.14776650e-01 7.34145865e-02 -1.78563327e-01 -3.76622349e-01 -4.40564394e-01 -4.89113241e-01 -3.68809044e-01 3.72957081e-01 -6.77390397e-01 -1.33016586e+00 7.37621129e-01 4.93709117e-01 -8.16265285e-01 -1.98683187e-01 -7.56431699e-01 -1.19485104e+00 1.36209857e+00 -1.43944812e+00 -7.54159093e-01 -2.90972590e-01 3.56661856e-01 7.51111567e-01 -3.32098126e-01 7.62468100e-01 2.04782218e-01 -5.71623981e-01 1.32099718e-01 3.61158758e-01 -3.65755439e-01 1.86334297e-01 -1.26427245e+00 -4.30021793e-01 4.92836773e-01 -5.25088131e-01 5.25377810e-01 1.34121788e+00 -7.06454158e-01 -1.18722546e+00 -7.28161693e-01 5.87738991e-01 1.61521770e-02 3.86327952e-01 2.14624241e-01 -1.17894650e+00 -4.25545825e-03 6.66120172e-01 -3.37719768e-02 5.20141542e-01 -3.64941418e-01 6.48478031e-01 8.84408876e-02 -1.58445835e+00 1.68306053e-01 3.01945060e-01 5.38273342e-02 -6.94242299e-01 2.65225142e-01 5.66513874e-02 -4.15829301e-01 -8.38368773e-01 3.62238199e-01 4.31233525e-01 -1.12752211e+00 8.57684255e-01 -4.04429674e-01 4.32865769e-01 -1.37639821e-01 7.00853467e-02 -1.29515564e+00 -1.89764008e-01 -7.16825306e-01 -3.21170390e-01 1.12000716e+00 3.03438991e-01 -9.98913586e-01 8.29893947e-01 6.96756303e-01 1.07832272e-02 -1.36766183e+00 -1.12630141e+00 -1.01013911e+00 8.29616368e-01 -1.93071112e-01 1.88314006e-01 1.14271486e+00 -3.37942034e-01 -8.02771896e-02 -1.87871531e-02 -3.74985598e-02 7.04864085e-01 3.15270871e-01 2.77707398e-01 -1.27548373e+00 -3.95277262e-01 -5.15250504e-01 -2.58149117e-01 -3.70276660e-01 3.91416922e-02 -8.49907041e-01 -2.43811771e-01 -1.56212854e+00 -3.50976378e-01 -7.32672811e-01 -1.45956308e-01 -3.22996154e-02 1.48820549e-01 1.42821878e-01 -5.80358543e-02 2.09675789e-01 3.95271868e-01 4.12397206e-01 1.18020964e+00 7.37933442e-04 -6.18062735e-01 -1.01446003e-01 -1.15406632e-01 8.62980425e-01 1.38197768e+00 -6.62066042e-01 -6.29476607e-01 -8.48756544e-03 5.63658290e-02 1.74032986e-01 2.82003015e-01 -1.30033565e+00 2.82263696e-01 -3.52016300e-01 5.41597784e-01 -4.33231205e-01 4.37451422e-01 -8.58700454e-01 1.96482763e-01 7.08364606e-01 -6.09655641e-02 3.03515315e-01 4.28748965e-01 1.57011941e-01 -1.21401109e-01 -7.00318158e-01 1.20983243e+00 -5.55777252e-01 -5.54324806e-01 1.19816788e-01 -5.67391872e-01 1.46859586e-01 1.28840494e+00 -8.41795743e-01 3.33107442e-01 -3.47148597e-01 -8.27359259e-01 -7.62403682e-02 3.14563662e-01 -1.63837865e-01 1.93677723e-01 -8.35348785e-01 -6.15396202e-01 3.51205990e-02 -7.88424253e-01 -1.17144145e-01 2.59499133e-01 9.44499493e-01 -9.69879508e-01 3.17191958e-01 -4.05250072e-01 -4.73056644e-01 -9.30779874e-01 5.51362038e-01 9.94133651e-01 -5.93174957e-02 -8.73562038e-01 9.57721293e-01 -2.60443896e-01 -4.52066749e-01 2.66269082e-03 7.92410150e-02 -2.91045934e-01 -7.51550719e-02 2.12280929e-01 9.92420673e-01 -1.23591349e-01 -5.00525117e-01 -2.07098514e-01 8.53581190e-01 3.85237873e-01 -1.85875878e-01 1.56013179e+00 3.57872322e-02 -2.24457622e-01 6.50257587e-01 1.28090692e+00 -1.10954240e-01 -1.13112366e+00 4.68501419e-01 -2.83769459e-01 -6.24505281e-02 5.38068771e-01 -1.12406826e+00 -1.11648190e+00 1.03244662e+00 9.53635931e-01 3.87475938e-01 1.08241320e+00 -6.98943138e-01 8.27696860e-01 3.74621481e-01 -3.03479023e-02 -1.17383504e+00 -2.70039290e-01 4.53763723e-01 9.14818287e-01 -1.05036974e+00 2.94432104e-01 5.90688065e-02 -3.19484353e-01 1.55975902e+00 6.15429103e-01 -4.90883321e-01 1.00900936e+00 8.61300051e-01 2.12997682e-02 -2.89058894e-01 -3.58740598e-01 3.53730440e-01 -1.71050150e-02 3.91622066e-01 5.06000459e-01 -4.04046625e-01 -6.96064711e-01 4.31502201e-02 -1.09768957e-01 1.69995636e-01 4.62590605e-01 1.20927906e+00 -5.71961164e-01 -9.14315462e-01 -6.04718626e-01 3.01276952e-01 -6.96834266e-01 1.42045632e-01 -2.04425901e-01 1.07381380e+00 5.26047170e-01 8.47937882e-01 -1.89218596e-01 1.43141389e-01 -2.99195666e-02 3.40372443e-01 4.06369716e-01 8.83697420e-02 -7.41635680e-01 -3.20577472e-01 -2.33683735e-01 -1.47043735e-01 -5.62953115e-01 -3.92238766e-01 -1.57472861e+00 -6.10611200e-01 -2.82695413e-01 7.28052258e-01 8.96287143e-01 1.10104227e+00 1.12351812e-01 4.87299383e-01 2.91240931e-01 -1.16746962e+00 -3.40285867e-01 -8.53508115e-01 -6.01433873e-01 -6.57913610e-02 4.12166834e-01 -8.98868263e-01 -6.88986361e-01 -3.80701303e-01]
[6.392177581787109, 3.3687777519226074]
9074b068-8a0e-4f8e-9b41-ed020259cf0a
winners-at-w-nut-2020-shared-task-3
null
null
https://aclanthology.org/2020.wnut-1.79
https://aclanthology.org/2020.wnut-1.79.pdf
Winners at W-NUT 2020 Shared Task-3: Leveraging Event Specific and Chunk Span information for Extracting COVID Entities from Tweets
Twitter has acted as an important source of information during disasters and pandemic, especially during the times of COVID-19. In this paper, we describe our system entry for WNUT 2020 Shared Task-3. The task was aimed at automating the extraction of a variety of COVID-19 related events from Twitter, such as individuals who recently contracted the virus, someone with symptoms who were denied testing and believed remedies against the infection. The system consists of separate multi-task models for slot-filling subtasks and sentence-classification subtasks, while leveraging the useful sentence-level information for the corresponding event. The system uses COVID-Twitter-BERT with attention-weighted pooling of candidate slot-chunk features to capture the useful information chunks. The system ranks 1st at the leaderboard with F1 of 0.6598, without using any ensembles or additional datasets.
['Tejas Vaidhya', 'Ayush Kaushal']
null
null
null
null
emnlp-wnut-2020-11
['sentence-classification']
['natural-language-processing']
[ 1.41663909e-01 1.87379330e-01 -2.55612642e-01 -1.06146201e-01 -1.01904023e+00 -2.14131415e-01 6.95524871e-01 8.13396275e-01 -7.68778980e-01 1.04683471e+00 7.97628343e-01 -2.45200306e-01 6.90945908e-02 -6.65966630e-01 -1.79443404e-01 -3.10828030e-01 -2.75411218e-01 7.64039755e-01 1.30031377e-01 -4.95967358e-01 1.10356463e-02 -2.27743220e-02 -1.15172410e+00 6.49556160e-01 6.48114800e-01 7.31527090e-01 3.66740078e-01 7.62006879e-01 -3.32034975e-01 9.43531573e-01 -9.47005987e-01 -2.49178827e-01 -1.62817389e-01 -1.59078371e-02 -1.00880015e+00 -5.37414312e-01 -3.05904359e-01 -2.78768092e-01 -2.57758558e-01 4.53851312e-01 6.26176119e-01 -9.87734869e-02 4.74939823e-01 -1.31856203e+00 -7.43648112e-02 6.06898010e-01 -2.96927631e-01 7.95162141e-01 5.44058800e-01 4.91883494e-02 9.60303724e-01 -8.07323217e-01 9.42740858e-01 1.03218591e+00 1.10940564e+00 5.51010489e-01 -7.78421879e-01 -7.19872117e-01 -2.17674598e-01 1.96193695e-01 -1.19536567e+00 -3.18057746e-01 2.58976102e-01 -8.98960173e-01 1.71491396e+00 4.52838689e-01 3.92380267e-01 1.31597352e+00 3.31892639e-01 8.13023150e-01 5.73215842e-01 2.25995928e-01 5.99910803e-02 -1.94987617e-02 5.67753494e-01 3.36215973e-01 2.61072338e-01 -3.26025307e-01 -6.88612223e-01 -7.11752176e-01 -4.06196713e-02 4.75776583e-01 -4.26867269e-02 8.96871030e-01 -1.15067065e+00 1.02855313e+00 1.68069527e-01 3.48020762e-01 -7.98935294e-01 -3.72565091e-01 8.47988188e-01 7.61073381e-02 1.23776090e+00 4.06779945e-01 -8.96613419e-01 -2.05988526e-01 -1.03298616e+00 5.46522856e-01 6.95335627e-01 4.67127323e-01 4.59994346e-01 -2.50128418e-01 -7.79532611e-01 4.85269248e-01 -4.08167019e-02 7.88056016e-01 2.96836317e-01 -1.13903314e-01 8.43508422e-01 5.86230159e-01 3.78601074e-01 -9.59982455e-01 -1.26072061e+00 -1.94770962e-01 -9.30416048e-01 -8.22179496e-01 3.07578873e-02 -9.46108937e-01 -9.61510539e-01 1.70063138e+00 4.63834256e-01 5.69047987e-01 -4.87476178e-02 6.08000338e-01 1.25546455e+00 8.43083799e-01 5.88780403e-01 -3.68232310e-01 1.73466051e+00 -7.19284475e-01 -9.80033338e-01 -3.32553416e-01 6.45135283e-01 -5.87591887e-01 3.74066025e-01 -2.30315015e-01 -7.00151682e-01 -1.78388849e-01 -5.77768385e-01 2.73201708e-02 -7.35356808e-01 -3.81340951e-01 1.78694174e-01 3.16168666e-01 -7.83987164e-01 2.82276630e-01 -5.56583583e-01 -7.90144324e-01 3.39019150e-01 -1.48832217e-01 -2.32554838e-01 3.69061440e-01 -1.86163020e+00 9.56249237e-01 3.54879111e-01 -1.21708468e-01 -7.62671232e-01 -1.03630114e+00 -8.06541264e-01 -1.92594100e-02 1.25436589e-01 -7.03898847e-01 1.10552502e+00 -7.17915967e-02 -6.56569004e-01 9.34117317e-01 -5.67467690e-01 -7.29217768e-01 2.18268052e-01 -3.89381081e-01 -6.35687351e-01 1.28214389e-01 6.10261381e-01 4.03270006e-01 7.46341765e-01 -4.29329544e-01 -7.70906568e-01 -4.18065220e-01 -2.21038118e-01 -1.73336398e-02 -2.58097529e-01 7.86986113e-01 -7.85823315e-02 -6.69827759e-01 -6.60517633e-01 -7.10662484e-01 -3.92815411e-01 -7.78152525e-01 -7.23935544e-01 -5.43722987e-01 8.46309662e-01 -1.04330993e+00 1.39222383e+00 -1.93595076e+00 -3.14391673e-01 -2.47889459e-01 3.17249209e-01 3.72334033e-01 -2.77624965e-01 1.03782201e+00 1.31877095e-01 3.41398448e-01 -2.65943315e-02 -5.65751672e-01 -2.23552436e-01 -9.44660157e-02 -7.18860030e-01 1.23497754e-01 5.73472440e-01 9.86567020e-01 -1.10853410e+00 -3.26116711e-01 -2.07682207e-01 4.22040015e-01 -2.48306766e-01 4.03126270e-01 -4.68639165e-01 4.48172152e-01 -6.97351575e-01 2.70575374e-01 6.68458581e-01 -4.90305156e-01 4.91768196e-02 1.52910471e-01 -7.03187287e-02 6.55795515e-01 -4.07235384e-01 1.21694434e+00 -1.43205985e-01 5.29909968e-01 2.41030902e-01 -6.18548095e-01 4.59845394e-01 7.79840589e-01 8.81146550e-01 -5.03812432e-01 6.51963130e-02 -2.25094438e-01 -5.86444318e-01 -8.87877584e-01 8.07735682e-01 -2.90912949e-02 -4.82608259e-01 7.29326487e-01 -1.30336851e-01 2.61494577e-01 1.04556836e-01 3.22707027e-01 1.46400249e+00 -7.11426795e-01 4.60686654e-01 -1.65661007e-01 2.34621301e-01 3.48933488e-01 5.07409036e-01 6.86992109e-01 -3.99466395e-01 4.06845987e-01 4.20393109e-01 -6.99483991e-01 -8.49759698e-01 -6.61395669e-01 -1.56730503e-01 1.34304595e+00 -2.51286477e-01 -7.14016914e-01 -4.87886429e-01 -4.88617748e-01 2.58955956e-02 6.63743913e-01 -7.97194064e-01 3.51628810e-01 -4.35471416e-01 -1.11379898e+00 9.82888937e-01 2.19806060e-01 3.19335759e-01 -1.35011363e+00 -8.06532860e-01 5.11882365e-01 -9.61414516e-01 -1.26732373e+00 -4.05590951e-01 1.37509987e-01 -3.42989445e-01 -1.18996644e+00 -6.10790551e-01 -4.76814687e-01 1.92684978e-01 2.92874128e-01 1.03257775e+00 1.05278946e-01 -4.34712142e-01 7.28889043e-03 -3.26532394e-01 -8.16321492e-01 1.55393735e-01 3.67960989e-01 6.96885064e-02 -4.00508970e-01 6.20312989e-01 1.08720712e-01 -2.71823466e-01 -8.16248134e-02 -7.68759668e-01 7.37378672e-02 -1.55438557e-01 7.60192692e-01 3.48423794e-02 -8.73135403e-02 8.85065794e-01 -1.02176702e+00 9.66094017e-01 -1.29030681e+00 2.00366378e-01 -6.90508485e-02 8.32630247e-02 -4.43638951e-01 3.67543250e-01 1.60483476e-02 -8.58736932e-01 -3.13845575e-01 -4.63035494e-01 2.85214037e-01 -1.27542913e-01 8.83597910e-01 3.71182680e-01 9.00942385e-01 6.30412817e-01 1.63047299e-01 -1.51550636e-01 -4.67216164e-01 5.13580488e-03 1.12373662e+00 1.63210660e-01 -1.03769429e-01 5.07906675e-01 3.69602919e-01 -5.05669177e-01 -1.34582710e+00 -1.43849385e+00 -1.01845074e+00 -6.58310503e-02 3.73024605e-02 1.28298020e+00 -1.19610810e+00 -7.66538501e-01 7.02304304e-01 -1.55174935e+00 -3.50310951e-01 -2.26564966e-02 1.76874414e-01 9.93355215e-02 -5.80794327e-02 -1.05666101e+00 -8.60763729e-01 -9.56116736e-01 -6.82529807e-01 1.21848667e+00 8.47548321e-02 -5.83003342e-01 -9.12038505e-01 5.44022620e-01 5.39441407e-01 8.08144629e-01 4.36277896e-01 7.09561229e-01 -1.41831756e+00 7.59587735e-02 -2.24058956e-01 -3.94154817e-01 -4.46165979e-01 1.04304411e-01 -3.95130664e-01 -1.15338647e+00 -2.64934540e-01 -1.46186322e-01 -6.07274294e-01 1.00749493e+00 4.24815416e-01 8.61970186e-01 -6.31334782e-01 -5.91576636e-01 -5.35658840e-03 9.65066552e-01 -9.31027532e-02 1.77671775e-01 2.61017174e-01 3.68197292e-01 6.61514938e-01 4.69528019e-01 9.89462793e-01 7.77689815e-01 5.92483878e-01 1.81461200e-01 -1.02415696e-01 3.74567330e-01 -1.78854257e-01 3.04310530e-01 6.17018402e-01 1.71054736e-01 -4.42232013e-01 -1.11305642e+00 7.89896131e-01 -1.85053539e+00 -1.28455639e+00 -3.64880621e-01 1.61429644e+00 8.50877106e-01 1.11352466e-01 3.61046523e-01 -2.68087566e-01 6.35468602e-01 6.12947047e-01 -3.50514889e-01 -3.39825362e-01 -6.42472655e-02 1.21578492e-01 2.38605365e-01 5.62111199e-01 -1.60870516e+00 9.37373161e-01 6.88398790e+00 7.76921749e-01 -9.08452928e-01 4.99969572e-01 3.77785414e-01 -6.85009882e-02 -1.28519043e-01 -5.67694962e-01 -9.36186135e-01 6.16190434e-01 1.40177929e+00 -3.20402622e-01 -1.95686892e-02 3.80789429e-01 4.68620569e-01 8.21859017e-02 -4.18089390e-01 4.06949103e-01 -4.31512250e-03 -1.64291751e+00 -2.67809004e-01 -2.15444624e-01 6.12229049e-01 7.30060160e-01 -1.49871901e-01 5.16955674e-01 1.13395095e-01 -8.55268955e-01 4.44552451e-01 6.70829237e-01 6.98184073e-01 -5.54604292e-01 1.18001997e+00 6.96689963e-01 -1.20103168e+00 1.15260676e-01 -1.05749257e-01 -2.72996515e-01 6.91375673e-01 9.23831344e-01 -1.39026666e+00 5.02894521e-01 5.39012849e-01 7.37290263e-01 -2.17103750e-01 9.39227104e-01 -6.34853616e-02 7.53395438e-01 -3.12527210e-01 -2.32489839e-01 2.42902994e-01 5.70930719e-01 8.06261778e-01 1.76044858e+00 1.64454073e-01 3.01178664e-01 3.54144365e-01 3.58293265e-01 -7.76275918e-02 1.60875648e-01 -6.96422815e-01 -1.98346034e-01 5.52099943e-01 1.07450354e+00 -5.53624988e-01 -5.62319696e-01 -1.11342825e-01 7.82867610e-01 -1.31120440e-02 2.17152610e-01 -8.40785325e-01 -3.83202195e-01 9.40265656e-01 1.63698047e-01 2.64476955e-01 -4.04868945e-02 -2.46840373e-01 -1.08065724e+00 -4.19070929e-01 -6.27254605e-01 5.99274337e-01 -5.57201028e-01 -1.41084564e+00 1.07068753e+00 -7.04987869e-02 -6.89351082e-01 -3.97334754e-01 -1.18033491e-01 -9.68552291e-01 8.47115993e-01 -1.32380509e+00 -1.09273815e+00 -1.82641950e-02 5.87379754e-01 7.14951754e-01 -1.09294012e-01 1.05506063e+00 3.43219012e-01 -7.98041642e-01 2.03168049e-01 -5.92416488e-02 2.26974100e-01 6.17701054e-01 -7.70705640e-01 8.50682735e-01 6.05797410e-01 -5.62490582e-01 5.97015798e-01 7.72527754e-01 -1.15061784e+00 -7.37179995e-01 -1.56412792e+00 1.98367035e+00 -6.46621048e-01 9.43559766e-01 -5.78597248e-01 -6.88118577e-01 5.72745800e-01 3.82056415e-01 -6.17907405e-01 1.01968455e+00 2.28032649e-01 -2.04125941e-01 3.57324898e-01 -1.27417731e+00 2.86634356e-01 7.29874611e-01 -8.14649642e-01 -8.87321830e-01 1.03591228e+00 1.36116171e+00 -2.35448822e-01 -6.04565501e-01 1.71385810e-01 1.31761655e-01 -3.46055657e-01 9.90621507e-01 -1.17132580e+00 3.24653149e-01 6.49489313e-02 6.33509606e-02 -1.11371303e+00 -2.27013648e-01 -6.66946411e-01 -2.66334843e-02 1.03669667e+00 6.57502592e-01 -6.09444499e-01 5.00758231e-01 4.71095741e-01 4.11901399e-02 -4.74304467e-01 -1.07153225e+00 -3.94395351e-01 -4.01305974e-01 -7.03117490e-01 6.59017026e-01 1.04115379e+00 4.02135819e-01 6.80294812e-01 -8.45924199e-01 7.00788796e-02 3.71531665e-01 -4.51591164e-02 5.17130136e-01 -1.33717215e+00 1.52876452e-01 -1.22083291e-01 9.90039483e-02 -5.09628415e-01 1.52800038e-01 -8.83229077e-01 1.99481938e-02 -1.59715760e+00 4.94257957e-01 -1.98481813e-01 -1.67650938e-01 9.20958459e-01 -4.81644303e-01 2.60884557e-02 2.71403939e-01 1.73470497e-01 -9.22028124e-01 4.02783632e-01 6.29290044e-01 -3.15204114e-01 -2.72832185e-01 2.06027985e-01 -6.78346395e-01 5.49101591e-01 9.37385738e-01 -7.99602866e-01 -7.95565993e-02 -2.57864624e-01 4.67997432e-01 1.95314363e-01 2.09839121e-01 -5.67298114e-01 1.61538184e-01 -2.20413759e-01 8.56223106e-02 -1.10951138e+00 3.04577172e-01 -4.35182482e-01 2.00405031e-01 8.22756171e-01 -3.43593657e-01 7.79217482e-02 4.39258397e-01 5.51948190e-01 1.32298283e-02 5.86162470e-02 1.91076487e-01 8.19775611e-02 -4.69356954e-01 4.52261776e-01 -9.20135498e-01 5.31710327e-01 8.65879476e-01 5.48172653e-01 -9.10862207e-01 -3.42225134e-01 -7.37936080e-01 6.66435421e-01 -1.51960671e-01 6.05721056e-01 5.75446248e-01 -9.55792427e-01 -1.24662828e+00 1.73070148e-01 1.23394296e-01 -3.26384097e-01 6.13842487e-01 8.56716692e-01 -3.33646655e-01 8.37876737e-01 -1.26578152e-01 -2.90844679e-01 -1.08463562e+00 5.17440796e-01 -2.08942011e-01 -8.45210075e-01 -5.34174919e-01 8.34221423e-01 -2.80522883e-01 -5.64296186e-01 1.24899864e-01 -2.98908830e-01 -5.24250209e-01 6.33957624e-01 1.17315614e+00 4.58797067e-01 1.61192849e-01 -7.41346896e-01 -8.55115950e-01 -4.77127284e-02 -6.85461517e-03 1.00003570e-01 1.59358454e+00 -2.42159385e-02 -3.12316000e-01 2.29943410e-01 1.06051850e+00 -1.13016196e-01 -7.07897246e-01 -3.88463706e-01 3.09814334e-01 -2.19498929e-02 -2.46962011e-01 -7.69102812e-01 -5.31803250e-01 6.50017440e-01 7.97998458e-02 6.30594730e-01 7.65839577e-01 1.84514802e-02 1.39752841e+00 2.19901636e-01 3.44051778e-01 -8.00603747e-01 -3.34103823e-01 1.12094879e+00 9.60194230e-01 -1.28139973e+00 -1.52829722e-01 6.91832155e-02 -7.96395004e-01 6.19360924e-01 8.01879913e-02 3.04215968e-01 1.03927147e+00 2.93549210e-01 1.17930807e-02 -4.11044389e-01 -1.21538615e+00 -3.64861637e-01 2.65130609e-01 6.13901258e-01 3.46139878e-01 4.62422550e-01 -4.04879153e-01 1.00249374e+00 1.67941544e-02 1.43997679e-02 1.99263826e-01 7.80837178e-01 -7.38635242e-01 -6.33995473e-01 -2.23452508e-01 6.70832396e-01 -6.79641306e-01 -4.51756358e-01 -3.47178280e-01 3.97023857e-01 2.44421139e-01 1.22405636e+00 1.81536615e-01 -6.27307475e-01 1.55473992e-01 3.17583919e-01 -4.82726872e-01 -7.32315421e-01 -1.04040384e+00 -2.17085078e-01 5.83700716e-01 -5.53293347e-01 -3.50138128e-01 -7.95067489e-01 -1.34609437e+00 -4.58325922e-01 2.19449192e-01 3.33932579e-01 3.04930359e-01 1.16894329e+00 6.51642323e-01 3.69856089e-01 4.89471734e-01 -7.22866297e-01 -1.17492102e-01 -1.19619179e+00 -3.09496403e-01 2.87613183e-01 7.39230454e-01 -2.74114132e-01 -8.62630382e-02 -2.50392497e-01]
[8.554841995239258, 9.403800010681152]
c6118dc7-7ab8-4cb5-b277-909411a7e0af
mimo-sar-a-hierarchical-high-resolution
2101.09293
null
https://arxiv.org/abs/2101.09293v2
https://arxiv.org/pdf/2101.09293v2.pdf
MIMO-SAR: A Hierarchical High-resolution Imaging Algorithm for mmWave FMCW Radar in Autonomous Driving
Millimeter-wave radars are being increasingly integrated into commercial vehicles to support advanced driver-assistance system features. A key shortcoming for present-day vehicular radar imaging is poor azimuth resolution (for side-looking operation) due to the form factor limits on antenna size and placement. In this paper, we propose a solution via a new multiple-input and multiple-output synthetic aperture radar (MIMO-SAR) imaging technique, that applies coherent SAR principles to vehicular MIMO radar to improve the side-view (angular) resolution. The proposed 2-stage hierarchical MIMO-SAR processing workflow drastically reduces the computation load while preserving image resolution. To enable coherent processing over the synthetic aperture, we integrate a radar odometry algorithm that estimates the trajectory of ego-radar. The MIMO-SAR algorithm is validated by both simulations and real experiment data collected by a vehicle-mounted radar platform.
['Guanbin Xing', 'Sumit Roy', 'Xiangyu Gao']
2021-01-22
null
null
null
null
['radar-odometry']
['robots']
[ 4.58670825e-01 -1.33768544e-01 4.03730303e-01 -6.81045771e-01 -7.30358601e-01 -5.55635631e-01 7.79811382e-01 -7.30113983e-01 -4.37997580e-01 6.64568305e-01 -6.89594224e-02 -6.43271208e-01 -4.73689824e-01 -8.50414872e-01 -6.53963983e-02 -6.41091943e-01 -6.86422512e-02 4.50267851e-01 3.92545387e-02 -3.18128437e-01 1.93119347e-01 1.08269513e+00 -1.51904190e+00 -1.44085392e-01 1.17685270e+00 8.93877625e-01 3.78743798e-01 7.97898829e-01 4.65120763e-01 2.77864814e-01 -2.39897728e-01 6.27595186e-02 7.79490173e-01 -9.19333994e-02 6.92602038e-01 -1.08718216e-01 9.66847897e-01 -4.15973336e-01 -5.79442382e-01 1.03767920e+00 1.91024274e-01 -2.38616034e-01 7.10186720e-01 -1.05983734e+00 -4.52939533e-02 -4.57750231e-01 -5.87389410e-01 4.04306650e-01 -1.77794009e-01 1.94783583e-01 2.70937383e-01 -1.10704124e+00 5.29242516e-01 7.44192600e-01 4.77808148e-01 -1.95888430e-01 -7.05063760e-01 -8.08813155e-01 -4.46715087e-01 5.89521050e-01 -1.47017264e+00 -6.68983161e-01 5.29466987e-01 -5.03467858e-01 5.78976989e-01 3.90522212e-01 5.59441864e-01 4.58957970e-01 9.48541582e-01 -1.88745484e-01 1.44679856e+00 -1.64208978e-01 2.66450405e-01 9.46006104e-02 3.04290354e-01 4.99076635e-01 1.28420293e+00 7.42230773e-01 -1.61320895e-01 2.03317285e-01 4.68235135e-01 9.56552569e-03 -1.38275996e-01 -6.62605226e-01 -6.79035306e-01 1.10937738e+00 1.77120075e-01 1.90668494e-01 -7.86652505e-01 -1.12199575e-01 -2.79100567e-01 4.73171830e-01 2.41021037e-01 5.62070847e-01 -8.84904042e-02 2.31688872e-01 -1.32498217e+00 4.44794446e-01 3.58321965e-01 7.62556672e-01 8.00411701e-01 7.93104410e-01 5.45248128e-02 -1.31689021e-02 6.88750505e-01 1.61767232e+00 -1.93344489e-01 -7.29926765e-01 3.03963155e-01 1.91059440e-01 4.34382074e-02 -9.24259007e-01 -6.28988385e-01 -1.13066566e+00 -7.77196825e-01 7.30928719e-01 3.11519112e-02 -4.21329260e-01 -1.26877499e+00 1.02735126e+00 2.88759768e-01 2.19465107e-01 5.46360970e-01 1.17114758e+00 7.39105225e-01 5.49996018e-01 -4.13477868e-01 -4.92336154e-01 1.60211599e+00 -4.60840702e-01 -8.54717135e-01 -9.41429257e-01 3.17702234e-01 -9.52057779e-01 -1.64494850e-02 4.31587428e-01 -3.40786964e-01 -4.78117675e-01 -1.53690517e+00 6.44570291e-01 -1.02613091e-01 4.67570335e-01 4.62557375e-01 1.10129952e+00 -6.98578775e-01 -5.13235331e-01 -3.80704403e-01 -2.08557501e-01 1.25963062e-01 -4.60668169e-02 -2.55490124e-01 -4.78744417e-01 -1.06948018e+00 1.32948291e+00 -3.34755689e-01 1.92076936e-01 -5.27350545e-01 -1.10358870e+00 -9.36684906e-01 -1.78079456e-01 3.53277586e-02 -5.89158952e-01 7.20937848e-01 -3.71836901e-01 -1.29150343e+00 3.74641210e-01 -1.66420862e-01 -9.69822288e-01 2.27761358e-01 -4.40464467e-01 -1.00860822e+00 3.66801888e-01 1.40625611e-01 2.13446334e-01 1.07054579e+00 -1.23440802e+00 -4.90874499e-01 -8.03480208e-01 -7.54175305e-01 2.06395179e-01 6.56315386e-01 -3.51199090e-01 3.38918418e-01 -1.48408547e-01 5.20545721e-01 -9.78399575e-01 -5.51095843e-01 -5.63656807e-01 2.77915090e-01 8.32957089e-01 1.32525373e+00 -6.05355024e-01 6.88473761e-01 -2.10496664e+00 -4.96578664e-01 4.02407229e-01 9.46038775e-03 4.71528262e-01 3.84986587e-02 3.65142971e-01 1.32778287e-01 -9.88484919e-01 -1.18595988e-01 2.45677471e-01 -5.10809124e-01 -1.18924819e-01 -7.35847890e-01 8.86336088e-01 -2.56914087e-02 7.99665451e-01 -1.33344099e-01 5.89180253e-02 5.20868540e-01 4.91925389e-01 -1.31832063e-01 -3.34803127e-02 2.10613266e-01 4.13868159e-01 -5.03637135e-01 7.60056674e-01 1.64298964e+00 4.89044279e-01 -3.22440602e-02 -4.61667210e-01 -8.71918321e-01 -3.51327688e-01 -1.20254111e+00 9.62955415e-01 -4.16754335e-01 1.15570903e+00 6.05886042e-01 -6.97485626e-01 1.28478789e+00 -2.12834820e-01 9.93513167e-02 -1.36216426e+00 7.05739930e-02 2.13145301e-01 2.76080728e-01 -4.64825600e-01 9.32540953e-01 -3.05043757e-01 -9.77868661e-02 -1.28450289e-01 -5.12624443e-01 -4.45092022e-01 -2.14217350e-01 5.98715469e-02 1.07627976e+00 -1.96663141e-01 6.51738167e-01 -3.17300469e-01 6.64043903e-01 7.36332417e-01 5.08190572e-01 4.48598862e-01 3.32439467e-02 2.30798990e-01 -3.58492136e-01 -4.12008315e-01 -9.64630663e-01 -1.20565152e+00 -3.09999943e-01 4.20957580e-02 2.97080994e-01 2.68710762e-01 -2.06460893e-01 -1.29350394e-01 2.16239974e-01 9.28620756e-01 -1.80592522e-01 1.36257961e-01 -7.47927606e-01 -7.30309963e-01 3.25610816e-01 -2.16415711e-02 4.67138022e-01 -2.62523443e-01 -1.45122969e+00 2.32286796e-01 3.80840182e-01 -1.41633320e+00 2.06552148e-01 -1.93073452e-01 -8.19082737e-01 -8.92394960e-01 -1.10231124e-01 -2.21975788e-01 3.71028394e-01 1.22266340e+00 5.26089847e-01 -7.58484125e-01 -6.60337508e-01 3.68266612e-01 -2.71855652e-01 -5.73337257e-01 -7.12539256e-03 -7.06445992e-01 2.48771951e-01 2.07881868e-01 3.22732031e-01 -5.65184653e-01 -5.31471550e-01 1.31604463e-01 -2.86160499e-01 3.85804959e-02 1.62598479e+00 3.70892167e-01 1.88903287e-01 -1.93337888e-01 6.32477880e-01 -6.00025892e-01 1.07847191e-01 -2.89630383e-01 -1.18265939e+00 -1.84916049e-01 -6.66729808e-01 -3.49854715e-02 3.17372084e-01 1.45439819e-01 -1.28472257e+00 1.15726843e-01 2.37039670e-01 -4.09727961e-01 -1.81756198e-01 6.25473678e-01 -2.26898953e-01 -5.22452712e-01 5.46472609e-01 4.71329600e-01 3.35171103e-01 -8.49987194e-02 4.49873358e-01 6.54604912e-01 6.40852749e-01 5.40102720e-01 1.52822578e+00 8.23980808e-01 6.38714314e-01 -1.68699622e+00 -4.17863756e-01 -5.88692307e-01 -3.76378000e-01 -4.50238645e-01 7.84511805e-01 -1.58828914e+00 -1.81699261e-01 -4.52135280e-02 -9.57403898e-01 1.93214133e-01 1.50143608e-01 1.32319319e+00 -2.88795441e-01 3.02719653e-01 4.21023726e-01 -1.08607399e+00 -5.52445769e-01 -7.28329718e-01 8.04501712e-01 4.29076850e-01 1.39093176e-01 -3.27852190e-01 2.97019869e-01 6.33377969e-01 7.60760307e-01 1.75318152e-01 3.48914564e-01 -1.11314304e-01 -1.32494628e+00 -2.13258013e-01 -5.09030819e-01 -2.49908224e-01 -4.55129713e-01 -5.41194081e-01 -5.91398120e-01 -4.54285026e-01 2.09859967e-01 4.21702981e-01 8.82946253e-01 8.83451164e-01 -1.96573973e-01 2.25728843e-02 -5.37589729e-01 9.80674505e-01 1.85810375e+00 4.35742021e-01 8.20932508e-01 3.69046181e-01 2.04571381e-01 3.84045333e-01 1.54668009e+00 3.05896223e-01 1.50476128e-01 8.34225833e-01 6.04249895e-01 1.95980877e-01 -2.49829754e-01 2.40116030e-01 5.00181675e-01 2.47037485e-01 7.75398239e-02 1.48808673e-01 -4.88932222e-01 4.22114700e-01 -1.50977397e+00 -1.23255157e+00 -6.25840962e-01 2.12195444e+00 -4.17433560e-01 7.33524412e-02 -6.10929966e-01 -3.44508171e-01 1.55318707e-01 2.62234628e-01 -2.68882096e-01 -5.25689542e-01 -1.00871846e-01 4.70425159e-01 1.41183519e+00 8.05376649e-01 -8.05481493e-01 6.37465835e-01 5.24124813e+00 4.95656699e-01 -1.35070348e+00 1.85176790e-01 -4.07883137e-01 1.82765543e-01 -4.26252514e-01 1.08080618e-01 -1.12575424e+00 -5.22965118e-02 1.07950163e+00 -2.13936463e-01 -1.46186605e-01 7.10138917e-01 4.66489255e-01 -5.97698689e-01 -2.89870918e-01 1.01954997e+00 4.26902682e-01 -1.55678236e+00 -5.18608652e-02 5.19587696e-01 3.18652600e-01 2.52495766e-01 1.79344401e-01 2.19638497e-01 -5.82880303e-02 -8.43859494e-01 3.42173189e-01 6.69614136e-01 8.70451212e-01 -8.67450535e-01 6.48373604e-01 3.59025806e-01 -1.22181153e+00 -6.95913881e-02 -1.88249990e-01 -3.91337276e-01 5.87478757e-01 8.58510911e-01 -1.36741650e+00 9.70546186e-01 2.57396370e-01 1.34987161e-01 -4.35453087e-01 9.81101930e-01 -1.01816356e-02 2.94127524e-01 -1.95868075e-01 6.42049536e-02 4.03721660e-01 -8.26830864e-01 1.14200699e+00 1.07780659e+00 1.07958508e+00 4.39010799e-01 -2.73544528e-02 4.91274416e-01 9.23190415e-01 -3.97168398e-01 -1.33952880e+00 2.31492415e-01 5.87193489e-01 1.76024079e+00 -1.79529622e-01 -4.95695435e-02 -4.13127661e-01 1.25097975e-01 -6.39435589e-01 3.63393843e-01 -8.20412278e-01 -5.17144442e-01 1.13944662e+00 5.37656486e-01 6.52643979e-01 -1.03549123e+00 -5.73177159e-01 -5.03683388e-01 -2.05563635e-01 -5.06482720e-01 -1.80500329e-01 -9.54474568e-01 -3.95302683e-01 6.95729256e-01 1.09292954e-01 -1.73001778e+00 -4.66729879e-01 -5.41934729e-01 -4.73839104e-01 1.03182805e+00 -1.77870226e+00 -1.44660378e+00 -6.61022007e-01 2.32688069e-01 4.37993199e-01 -8.12483609e-01 4.74520862e-01 2.61552155e-01 1.32896885e-01 7.62196258e-03 5.65733537e-02 -4.00201470e-01 5.52921712e-01 -3.73851180e-01 2.27143288e-01 1.49814332e+00 -1.17600918e-01 2.90295571e-01 1.40032184e+00 -8.96204412e-01 -2.14024138e+00 -1.34483552e+00 6.45589709e-01 2.78857257e-02 3.71646374e-01 -1.86692163e-01 -3.36425871e-01 6.35380983e-01 3.85701776e-01 -7.04370067e-02 4.00604278e-01 -3.19427133e-01 -2.49445066e-01 -4.09757078e-01 -1.19920778e+00 4.51925755e-01 5.60617864e-01 2.24612392e-02 -8.02597940e-01 -8.13741833e-02 2.94032693e-01 -2.17232496e-01 -3.22302669e-01 1.05975282e+00 7.78478801e-01 -1.04221821e+00 8.68646562e-01 4.00520042e-02 -3.98765653e-01 -8.07077646e-01 -5.43257773e-01 -1.38706684e+00 -5.39321482e-01 -3.04499298e-01 6.08609617e-02 3.16516757e-01 1.62286192e-01 -9.52105343e-01 1.02941453e+00 -5.42381883e-01 -3.79304945e-01 -2.10972294e-01 -1.24846733e+00 -8.11608374e-01 -6.60851359e-01 -3.33810687e-01 -1.19824156e-01 4.50118423e-01 -5.65438628e-01 7.60522723e-01 -4.04515743e-01 1.24660540e+00 1.37685823e+00 6.54413998e-01 1.18185377e+00 -1.53723979e+00 -7.05581605e-02 2.03996107e-01 -8.01582396e-01 -9.27545369e-01 -2.00905278e-01 -4.81938034e-01 -4.45603281e-02 -1.52092111e+00 -3.52572709e-01 -3.93646419e-01 6.17762089e-01 -5.89948535e-01 4.64918286e-01 6.17406249e-01 5.82975075e-02 -7.10328147e-02 1.95679218e-01 4.71132457e-01 8.97198677e-01 9.67818052e-02 -6.85858354e-02 2.21779004e-01 -4.35607821e-01 4.32064414e-01 6.58843935e-01 -2.62302369e-01 -4.22816157e-01 -2.64549524e-01 2.05074683e-01 5.20401418e-01 5.01000047e-01 -1.54134667e+00 6.34192288e-01 -2.16931954e-01 4.65565056e-01 -1.45587277e+00 7.86951125e-01 -9.88188803e-01 6.09009027e-01 8.25692952e-01 8.03363919e-01 4.90982458e-03 1.32591397e-01 5.39473951e-01 -3.23470414e-01 7.55342618e-02 1.15511811e+00 2.95339525e-01 -1.15100932e+00 2.42658779e-01 -7.59704649e-01 -4.87358183e-01 1.54699886e+00 -6.67051673e-01 -5.90468168e-01 -4.43634391e-01 -2.60415703e-01 1.28344655e-01 1.44331113e-01 1.91824824e-01 1.06779075e+00 -1.05172801e+00 -9.94386792e-01 9.72786844e-01 3.01455021e-01 -6.30625725e-01 7.30356514e-01 1.01508522e+00 -6.30174160e-01 1.09268045e+00 -7.38017559e-01 -5.46150804e-01 -1.73492873e+00 1.85478389e-01 1.82465196e-01 -4.32613194e-02 -6.89708233e-01 1.43380389e-01 1.40469611e-01 -2.45859817e-01 -7.50189662e-01 1.93697557e-01 -3.98944885e-01 3.03855129e-02 8.51694345e-01 3.44801366e-01 2.69962907e-01 -8.34661365e-01 -6.28078222e-01 1.07311571e+00 -4.51886617e-02 -5.66015005e-01 1.33809376e+00 -2.92519122e-01 2.97654063e-01 -1.22016668e-01 5.97659230e-01 8.21710706e-01 -9.51232016e-01 -4.86384407e-02 -2.32090503e-01 -5.30866981e-01 5.61509907e-01 -6.14886403e-01 -6.24402225e-01 6.93672836e-01 9.54337358e-01 -2.50215590e-01 9.52184379e-01 -4.05307829e-01 3.51282060e-01 7.06199229e-01 5.24629891e-01 -7.15947330e-01 -6.08797789e-01 7.67742395e-01 6.33543193e-01 -9.24092829e-01 7.11736202e-01 -7.19699264e-01 -6.86978698e-01 1.16343522e+00 3.32025141e-01 -2.98072755e-01 6.49636567e-01 6.88880086e-01 4.83364433e-01 -5.09026110e-01 -6.35582864e-01 -5.07819533e-01 1.80107281e-01 7.97377408e-01 -1.48609832e-01 2.99114168e-01 -5.48144102e-01 1.86734930e-01 -2.29977772e-01 -3.14806670e-01 8.41732323e-01 7.20440328e-01 -1.19054043e+00 -6.56578779e-01 -7.44246125e-01 4.55930442e-01 2.65371799e-01 3.38516012e-02 9.37412158e-02 9.96838570e-01 3.83579470e-02 9.25731719e-01 3.83862078e-01 -3.11475009e-01 4.41988975e-01 -2.88050413e-01 7.41341829e-01 -3.32965553e-01 1.39108628e-01 -1.69411182e-01 3.62759650e-01 -3.44064265e-01 -7.40519464e-02 -9.09874618e-01 -9.60167646e-01 8.64136070e-02 1.24194227e-01 3.81108820e-01 1.21703243e+00 8.07016075e-01 7.40080118e-01 4.60853279e-01 8.39163840e-01 -5.98065913e-01 -5.42842329e-01 -9.93009567e-01 -6.99924052e-01 -8.58424067e-01 4.50482577e-01 -7.45159745e-01 -3.30109954e-01 -4.10180062e-01]
[6.765275478363037, 0.9258217811584473]
81f911c9-f2e3-4903-9a72-3212face78a6
improving-neural-text-summarization-using
null
null
https://openreview.net/forum?id=9nBQn6hjJmg
https://openreview.net/pdf?id=9nBQn6hjJmg
Improving Neural Text Summarization using Knowledge Graphs
In this paper, we propose a method for extractive text summarization using auto-regressive transformers. For better learning procedure we adopt the knowledge graph method to convert our textual data to more informative text and unsupervised training methods for wide use. We feed the informative text to our pre-trained generative model to summarize the text more properly and infer on generating a proper summary. The model is able to summarize input text into adequate information and is capable of performing several Natural Language Processing(NLP) tasks.
['Sumit Kumar', 'Raj Ratn Pranesh', 'Ambesh Shekhar']
2020-10-24
null
null
null
null
['extractive-document-summarization']
['natural-language-processing']
[ 4.17049319e-01 7.13903248e-01 -4.39049155e-01 -4.28127736e-01 -8.39315057e-01 -3.40411246e-01 8.32415581e-01 4.21919316e-01 -1.42894611e-01 1.21109295e+00 1.13314211e+00 -2.14374274e-01 6.30908832e-02 -8.63079965e-01 -5.36789000e-01 -2.45700717e-01 3.13457757e-01 8.11745584e-01 -6.42112643e-02 -3.32507044e-01 5.21815419e-01 4.06319231e-01 -1.16456616e+00 4.00378644e-01 1.27895820e+00 3.25816512e-01 3.57931167e-01 1.13699901e+00 -7.20687687e-01 1.40539384e+00 -1.06832826e+00 -4.42818969e-01 -4.50242102e-01 -8.79911602e-01 -1.37323713e+00 1.24159515e-01 1.73988774e-01 -8.37875605e-02 -1.77947924e-01 9.08358932e-01 1.20802172e-01 4.98228073e-01 1.34477174e+00 -8.01370978e-01 -7.94169486e-01 1.28727376e+00 -1.68747917e-01 2.62953997e-01 6.06234372e-01 -3.44902933e-01 9.99884188e-01 -7.79180646e-01 7.78621912e-01 1.25888097e+00 1.75346091e-01 5.31112671e-01 -9.29430544e-01 -1.14365652e-01 3.67502570e-02 8.29183981e-02 -6.94277585e-01 -5.01083195e-01 1.06267583e+00 -2.66243130e-01 1.24450457e+00 4.98094082e-01 8.00189137e-01 1.15709841e+00 5.92171788e-01 9.50878441e-01 6.65206969e-01 -6.45036876e-01 1.96792185e-01 1.43022224e-01 5.41168392e-01 8.21789384e-01 5.98186016e-01 -7.88000226e-01 -8.51990223e-01 9.62422788e-02 2.78823823e-01 -4.88663092e-02 -2.38783076e-01 2.54135430e-01 -8.49167705e-01 9.44134712e-01 2.14579970e-01 6.19460702e-01 -7.58774102e-01 1.62276804e-01 4.96955276e-01 1.93149984e-01 9.80229676e-01 7.62192547e-01 -2.81880468e-01 -1.43350149e-02 -1.38550997e+00 -3.21568586e-02 1.17051625e+00 9.29521143e-01 7.24562228e-01 4.76678431e-01 -4.64276791e-01 6.85770214e-01 2.14579418e-01 4.79965955e-01 7.30355680e-01 -7.25144029e-01 7.61991322e-01 7.13656247e-01 -1.20909788e-01 -8.42672706e-01 -2.77903199e-01 -3.50430608e-01 -1.14283979e+00 -3.43144476e-01 -5.26105940e-01 -3.27085078e-01 -9.02112901e-01 1.01565337e+00 -3.25743735e-01 -2.17147529e-01 5.44757545e-01 2.32026532e-01 1.29910374e+00 1.38133025e+00 1.02859534e-01 -9.47727263e-01 1.19851935e+00 -1.11317325e+00 -1.31717741e+00 -4.35225159e-01 4.08782810e-01 -4.84252065e-01 9.76294875e-01 5.42876601e-01 -1.09831142e+00 -3.88402551e-01 -1.12113440e+00 -4.89027202e-01 -3.85321587e-01 5.34619331e-01 4.30770576e-01 1.73694611e-01 -1.14426887e+00 7.05964923e-01 -8.40550601e-01 -5.04089415e-01 4.58331823e-01 2.48069987e-01 -3.46100092e-01 1.64945439e-01 -9.98482645e-01 1.17741930e+00 1.01502013e+00 -2.99676150e-01 -5.90645552e-01 -3.59568655e-01 -1.13310385e+00 3.87535781e-01 3.70112985e-01 -1.44402647e+00 1.31430113e+00 -7.77925730e-01 -1.98514724e+00 4.91194487e-01 -3.74993265e-01 -7.29442000e-01 3.10003012e-02 -5.01781702e-01 -1.51035190e-01 4.29815114e-01 -3.45999300e-02 2.72891015e-01 9.57438886e-01 -1.33591878e+00 -2.88931489e-01 -3.35691571e-01 -5.71806371e-01 2.85572052e-01 -5.68213522e-01 -7.11730346e-02 -2.36906126e-01 -8.53815556e-01 -2.60106653e-01 -3.06094289e-01 -1.78857237e-01 -1.19211709e+00 -8.66857767e-01 -6.12383842e-01 7.39061117e-01 -1.15668726e+00 1.63604105e+00 -1.24622250e+00 5.22359371e-01 -5.60680665e-02 3.09686005e-01 3.07647139e-01 1.15477666e-01 9.73323524e-01 1.97149843e-01 3.13578933e-01 -3.93486410e-01 -6.41243279e-01 -2.02465057e-01 3.96709472e-01 -7.18982458e-01 -1.65250435e-01 3.41219574e-01 1.11534429e+00 -9.44954813e-01 -9.25797403e-01 1.69421509e-01 1.70496702e-01 -2.77309030e-01 4.78962779e-01 -5.10497093e-01 2.17774838e-01 -8.36657047e-01 2.62317777e-01 1.90827727e-01 -1.24453254e-01 -2.38911770e-02 -2.10125044e-01 -5.77801764e-02 5.43062806e-01 -4.64884222e-01 1.68681753e+00 -6.05259240e-01 9.71912861e-01 -5.81946194e-01 -1.22621524e+00 1.26636279e+00 1.12596415e-01 5.34467995e-02 -4.56556045e-02 3.54041755e-01 -1.28897175e-01 -6.74660981e-01 -8.80631983e-01 1.18780041e+00 -1.35182068e-01 -1.39689505e-01 6.11274481e-01 5.91369987e-01 -6.83331490e-01 4.90203559e-01 6.91841066e-01 8.73355031e-01 -2.93158870e-02 8.88060987e-01 -2.19228610e-01 6.32630289e-01 2.01317549e-01 5.77354394e-02 7.56136417e-01 6.92037106e-01 2.71800339e-01 5.42245507e-01 -1.21483251e-01 -1.07053792e+00 -7.41499305e-01 4.60993707e-01 1.06055200e+00 -3.91778469e-01 -7.51028299e-01 -7.17253745e-01 -7.61905074e-01 -3.52466464e-01 1.68538380e+00 -3.83688748e-01 -4.17470843e-01 -3.87357682e-01 -4.28645015e-01 3.98119986e-01 4.55560386e-01 4.49258953e-01 -1.35158491e+00 -1.90449581e-01 9.03160796e-02 -5.18957973e-01 -7.29473650e-01 -3.83067340e-01 4.01733220e-01 -1.22405994e+00 -6.21593297e-01 -5.00058413e-01 -1.04255164e+00 6.64926946e-01 -8.25716779e-02 1.16206110e+00 -2.62616128e-01 2.04076171e-01 3.56705159e-01 -4.96884376e-01 -6.64674640e-01 -1.13755131e+00 4.98912811e-01 -1.86710805e-01 -2.04229012e-01 -4.92348394e-04 -6.11811340e-01 3.47802602e-02 -8.18771064e-01 -1.15547788e+00 3.76642972e-01 7.52251506e-01 6.57070696e-01 3.31637770e-01 3.99074741e-02 8.47288191e-01 -1.18120623e+00 1.50406933e+00 -2.94876546e-01 -1.00642852e-01 5.36684871e-01 -8.18654239e-01 9.45591211e-01 1.03610349e+00 -1.40252784e-01 -1.60898364e+00 -1.48686543e-01 -1.41779527e-01 4.50689942e-02 1.02900662e-01 9.54044878e-01 -7.72822201e-02 5.71838319e-01 8.33089769e-01 6.50710285e-01 -3.97429734e-01 -4.01167691e-01 6.39369190e-01 8.44219267e-01 6.97801709e-01 -2.09888399e-01 6.89519048e-01 -4.07081693e-02 -7.75733739e-02 -1.06756115e+00 -1.11546612e+00 -4.09304112e-01 -6.81181133e-01 -2.45626330e-01 8.92209649e-01 -5.10807574e-01 -1.41551763e-01 -6.41106367e-02 -1.48070621e+00 2.10073721e-02 -4.56790775e-01 3.92443120e-01 -6.13296568e-01 6.67200863e-01 -4.05811757e-01 -8.42352629e-01 -1.21983802e+00 -3.20758522e-01 1.01032531e+00 5.20455182e-01 -2.92274356e-01 -1.27605212e+00 5.12619555e-01 2.76440054e-01 9.34213176e-02 -1.18961521e-04 9.13913429e-01 -1.31576061e+00 -1.16208173e-01 -2.25064620e-01 1.61676615e-01 5.99111199e-01 2.64788955e-01 3.31278861e-01 -6.18978024e-01 1.69065788e-01 1.88414663e-01 -3.40970069e-01 1.43792903e+00 4.95536387e-01 1.00302923e+00 -1.12737024e+00 -3.64573598e-01 2.43903622e-01 1.14669406e+00 -7.63897672e-02 7.43893623e-01 7.49127641e-02 5.92824876e-01 5.43005049e-01 3.79696131e-01 2.35504866e-01 3.91293168e-01 -4.66673151e-02 -4.15068902e-02 2.17579752e-01 -7.04419315e-02 -6.79350257e-01 5.29746711e-01 1.57972991e+00 -2.38123357e-01 -8.57782125e-01 -7.13998556e-01 3.14704299e-01 -1.96267247e+00 -1.32806361e+00 -1.19954593e-01 1.57374942e+00 1.18562806e+00 1.47417113e-01 -1.82047129e-01 -1.05835707e-03 6.18553817e-01 2.85576344e-01 -1.52437523e-01 -9.61346030e-01 -6.32436872e-02 3.29116344e-01 9.11966488e-02 7.98488319e-01 -7.25520492e-01 1.31833041e+00 6.74052763e+00 7.31257737e-01 -9.07047927e-01 -3.44267607e-01 4.86095876e-01 3.06170374e-01 -7.16172159e-01 7.67729431e-02 -6.26312971e-01 4.29496355e-02 1.12889707e+00 -9.54729736e-01 1.34654179e-01 6.55279994e-01 3.38869810e-01 -4.16197419e-01 -9.24944103e-01 7.76870072e-01 6.54291332e-01 -1.68044007e+00 9.60091233e-01 -3.98587018e-01 8.02483201e-01 -3.77651632e-01 -4.73710090e-01 5.65085888e-01 4.55470800e-01 -9.89223421e-01 2.42285088e-01 1.10922587e+00 4.38288212e-01 -7.64729142e-01 9.22577202e-01 8.36030245e-01 -6.86114907e-01 2.47376233e-01 -4.98222739e-01 -1.53668210e-01 7.51339942e-02 6.54753566e-01 -1.37455714e+00 1.04053378e+00 -4.65980507e-02 9.23667252e-01 -8.33761930e-01 5.02236485e-01 -6.21824503e-01 8.57081711e-01 4.72257137e-02 -6.41807377e-01 5.43944985e-02 -4.25595164e-01 7.75510132e-01 1.54167771e+00 3.96758795e-01 2.60284245e-01 -8.04763138e-02 7.25871742e-01 -5.03096938e-01 5.28685033e-01 -7.96635270e-01 -5.85220456e-01 1.69980690e-01 1.31294215e+00 -6.82802558e-01 -9.87185776e-01 2.75061518e-01 1.10715902e+00 3.55334312e-01 3.77214581e-01 -3.35949033e-01 -5.52830637e-01 -5.10634840e-01 -3.24053377e-01 7.84466118e-02 -1.61134660e-01 -3.73333246e-01 -1.59992981e+00 -1.82785556e-01 -5.95318735e-01 4.34163630e-01 -1.23478818e+00 -9.53239262e-01 8.43300045e-01 3.21045965e-01 -6.98266625e-01 -6.59366667e-01 -1.64324299e-01 -1.11839080e+00 6.04467213e-01 -1.31520236e+00 -1.07606363e+00 -1.85381979e-01 3.38595718e-01 1.01521909e+00 -4.81113672e-01 9.46371794e-01 -6.63197517e-01 -4.91058618e-01 -5.69211282e-02 -9.07836407e-02 -5.93044274e-02 6.06656611e-01 -1.53546357e+00 1.85962975e-01 1.07512093e+00 5.16928852e-01 6.00101352e-01 1.20922482e+00 -1.00038564e+00 -1.26484323e+00 -1.15830922e+00 1.53530252e+00 -2.61460483e-01 5.91423869e-01 -4.22016659e-04 -9.29276407e-01 8.47894549e-01 9.49649632e-01 -8.81597102e-01 7.07782090e-01 -1.27000902e-02 5.01863733e-02 -1.49217695e-01 -7.82718956e-01 5.65088928e-01 4.85888511e-01 -2.97047287e-01 -1.73341441e+00 4.35312688e-01 9.58661616e-01 -1.04954029e-02 -4.86988604e-01 1.15186833e-01 -3.91989239e-02 -4.97109354e-01 4.13319468e-01 -8.20301294e-01 9.42737103e-01 4.76910919e-02 4.32169497e-01 -1.83562291e+00 -2.23158449e-01 -8.67165506e-01 -4.15392697e-01 1.54753888e+00 5.66612422e-01 -3.82281333e-01 4.43327099e-01 9.88212377e-02 -3.82660806e-01 -4.25802857e-01 -2.80003667e-01 -2.38422543e-01 -1.63650960e-01 -7.33054802e-02 1.69660807e-01 6.59900010e-01 5.41292310e-01 1.26430750e+00 -5.00099659e-01 -3.14703226e-01 2.67626852e-01 2.43621305e-01 6.55398786e-01 -1.23974419e+00 1.15092784e-01 -6.18299961e-01 8.11045617e-02 -9.75899100e-01 4.54176962e-01 -1.17496896e+00 1.60113052e-01 -2.55097127e+00 5.53060055e-01 6.83188021e-01 2.97912210e-01 4.05868381e-01 -3.47014397e-01 -4.27624822e-01 -1.01583160e-01 1.59933582e-01 -7.57007539e-01 7.90745914e-01 1.21004117e+00 -3.73920590e-01 -4.71427828e-01 3.14289629e-01 -1.06352746e+00 5.14943838e-01 9.88111258e-01 -5.87041318e-01 -4.81967628e-01 -5.55882454e-02 4.91857558e-01 4.01949495e-01 -4.95063551e-02 -6.59897983e-01 4.67234194e-01 -2.87823498e-01 4.45302814e-01 -1.08731389e+00 -6.02640696e-02 -3.52804869e-01 -2.00428620e-01 4.46006000e-01 -8.68406832e-01 -1.14870565e-02 4.20395657e-02 4.89085913e-01 -4.73553121e-01 -6.80659473e-01 2.20671207e-01 -1.28939569e-01 -1.57727852e-01 5.10185510e-02 -7.74118483e-01 1.99777797e-01 4.82930660e-01 1.63332552e-01 -3.46756727e-01 -8.80804181e-01 -6.28353179e-01 2.05302522e-01 9.01308805e-02 2.24239916e-01 8.24008286e-01 -1.02475488e+00 -9.95681763e-01 -2.11187273e-01 -1.56361699e-01 -3.81422676e-02 -1.25128970e-01 3.87455910e-01 -8.48162949e-01 7.81486213e-01 -2.27308571e-01 -7.01863691e-02 -1.12322414e+00 4.82075751e-01 -9.17424038e-02 -7.10186541e-01 -7.22522199e-01 2.88706481e-01 -3.28139544e-01 6.71670511e-02 3.16377021e-02 -5.12744129e-01 -9.35991049e-01 2.92681903e-01 5.32012284e-01 4.51133639e-01 2.72770934e-02 -3.25545579e-01 8.56103841e-04 1.71369746e-01 -9.35424343e-02 -9.41228047e-02 1.74348903e+00 -1.30659938e-01 -4.82892960e-01 7.01534450e-01 1.02500582e+00 3.73895258e-01 -6.13559842e-01 -2.34312955e-02 3.01466525e-01 3.68628114e-01 1.48589551e-01 -6.70676053e-01 -4.76210654e-01 7.38541186e-01 -5.26233613e-01 5.41839242e-01 1.02409780e+00 2.73080260e-01 5.61672807e-01 1.04284132e+00 -4.62203652e-01 -1.24525201e+00 3.45286727e-01 7.41332412e-01 1.21718252e+00 -9.12969828e-01 5.96623421e-01 -2.62484461e-01 -1.00001419e+00 1.77743042e+00 2.02784419e-01 -2.18451127e-01 1.91591352e-01 1.85591817e-01 -3.26549441e-01 -4.19843048e-01 -1.00513554e+00 -1.07182913e-01 7.79080987e-01 5.81669688e-01 4.22437549e-01 -3.39241810e-02 -5.42619944e-01 9.14710820e-01 -7.41127491e-01 -5.86084723e-02 9.23727691e-01 5.28002322e-01 -1.05829585e+00 -8.36704493e-01 -1.65574461e-01 6.51025414e-01 -2.40859613e-01 -1.37410656e-01 -9.43448663e-01 5.50137043e-01 -5.89170396e-01 1.05437517e+00 -1.35054737e-01 -5.43604419e-02 1.69504687e-01 4.90247399e-01 5.54636240e-01 -9.84370828e-01 -6.51143670e-01 4.65026312e-02 4.61354166e-01 1.44941896e-01 -5.85951149e-01 -3.68188441e-01 -1.40991318e+00 -1.17189273e-01 -2.80419856e-01 5.89123607e-01 6.29499674e-01 1.18378508e+00 1.39236256e-01 9.52641010e-01 6.36255026e-01 -6.77294493e-01 -4.30132002e-01 -1.45879877e+00 -1.20984748e-01 1.45102412e-01 2.27961525e-01 1.05843112e-01 -3.52118880e-01 6.38441443e-01]
[12.539624214172363, 9.54670524597168]
9a079358-ecb0-4be4-8753-00c6e1b9f977
volumetric-super-resolution-of-multispectral
1705.05745
null
http://arxiv.org/abs/1705.05745v1
http://arxiv.org/pdf/1705.05745v1.pdf
Volumetric Super-Resolution of Multispectral Data
Most multispectral remote sensors (e.g. QuickBird, IKONOS, and Landsat 7 ETM+) provide low-spatial high-spectral resolution multispectral (MS) or high-spatial low-spectral resolution panchromatic (PAN) images, separately. In order to reconstruct a high-spatial/high-spectral resolution multispectral image volume, either the information in MS and PAN images are fused (i.e. pansharpening) or super-resolution reconstruction (SRR) is used with only MS images captured on different dates. Existing methods do not utilize temporal information of MS and high spatial resolution of PAN images together to improve the resolution. In this paper, we propose a multiframe SRR algorithm using pansharpened MS images, taking advantage of both temporal and spatial information available in multispectral imagery, in order to exceed spatial resolution of given PAN images. We first apply pansharpening to a set of multispectral images and their corresponding PAN images captured on different dates. Then, we use the pansharpened multispectral images as input to the proposed wavelet-based multiframe SRR method to yield full volumetric SRR. The proposed SRR method is obtained by deriving the subband relations between multitemporal MS volumes. We demonstrate the results on Landsat 7 ETM+ images comparing our method to conventional techniques.
['Vildan Atalay Aydin', 'Hassan Foroosh']
2017-05-14
null
null
null
null
['pansharpening']
['computer-vision']
[ 9.81114507e-01 -1.03054631e+00 -1.09006464e-01 -9.82350409e-02 -1.07637203e+00 -6.61881208e-01 3.12733501e-01 -4.58772600e-01 -5.24582863e-01 8.19522917e-01 -5.98100238e-02 -1.30834088e-01 -6.36235893e-01 -1.25395155e+00 -1.56745061e-01 -9.33197320e-01 5.90355545e-02 -2.00900406e-01 2.52011478e-01 -4.84624386e-01 -9.64671448e-02 8.19809079e-01 -1.66641104e+00 3.10495377e-01 1.05493426e+00 8.24449480e-01 9.97312903e-01 8.89564276e-01 1.67871431e-01 1.16401669e-02 1.10537582e-03 6.29539251e-01 6.50143266e-01 -5.43953359e-01 -5.64406276e-01 3.93486530e-01 2.48451069e-01 -3.80359054e-01 -9.57398266e-02 1.32648182e+00 8.44525099e-02 2.09514096e-01 2.03008100e-01 -5.65519750e-01 -4.62206870e-01 4.25767839e-01 -1.59008753e+00 4.62696940e-01 3.52705130e-03 -1.65687114e-01 8.29313517e-01 -6.83310509e-01 3.20545346e-01 8.84131610e-01 6.32692397e-01 -2.01051667e-01 -1.36229682e+00 -4.91936058e-01 -5.04229069e-01 1.04397811e-01 -1.41664171e+00 -2.34998599e-01 9.27655220e-01 -1.11517332e-01 3.45002800e-01 6.58611834e-01 7.67659783e-01 1.68164179e-01 2.16978833e-01 1.67966425e-01 1.83073401e+00 -5.93348145e-01 -2.49471769e-01 -5.20473897e-01 3.01180750e-01 6.54780790e-02 2.79194146e-01 6.26338899e-01 -2.60041803e-01 -2.10233912e-01 9.60990250e-01 4.42333013e-01 -6.15978360e-01 3.26618135e-01 -1.13092947e+00 6.65723622e-01 5.31346321e-01 7.12132692e-01 -1.01338410e+00 -3.34170431e-01 -6.86756149e-02 4.86398578e-01 6.27118468e-01 1.04248270e-01 -4.28845465e-01 3.99046510e-01 -1.44676185e+00 -2.22545080e-02 -1.95965737e-01 4.05365288e-01 1.26586783e+00 8.39033425e-02 4.85152155e-01 1.23657644e+00 2.49711499e-01 1.07975602e+00 4.81611311e-01 -9.73175943e-01 1.31119803e-01 3.03730547e-01 1.95766687e-01 -9.42061841e-01 -2.49098092e-01 -2.43778095e-01 -1.06872368e+00 1.17933780e-01 -1.96362182e-01 2.21648574e-01 -9.77632582e-01 1.17126119e+00 1.72648519e-01 1.77283198e-01 4.15006250e-01 1.17185116e+00 4.36678350e-01 1.30863631e+00 -2.05501188e-02 -7.56500483e-01 1.23316360e+00 -4.06166852e-01 -6.54351413e-01 -4.56795543e-01 -1.36357933e-01 -1.05726707e+00 7.65104413e-01 3.05279702e-01 -6.68023527e-01 -6.06365323e-01 -9.68595147e-01 4.37744081e-01 -3.14479679e-01 5.06824136e-01 3.74926895e-01 3.61416668e-01 -8.84705663e-01 6.75109923e-01 -7.29314506e-01 -3.81027013e-01 -5.26694581e-02 -1.51480183e-01 -3.19715351e-01 -2.70530254e-01 -1.11534405e+00 8.04773748e-01 6.27812982e-01 4.45048302e-01 -3.53064120e-01 -5.24419606e-01 -7.10301042e-01 -3.23825508e-01 1.81947634e-01 8.54593813e-02 6.72577620e-01 -1.22376144e+00 -1.09154320e+00 8.50026608e-01 -4.57995683e-02 -2.66024888e-01 -9.62367430e-02 4.64290977e-02 -9.24568176e-01 6.69587731e-01 2.75741786e-01 2.42880336e-03 9.55378175e-01 -1.38334501e+00 -8.49900246e-01 -7.39643157e-01 -3.67719710e-01 1.88296869e-01 7.92300608e-03 3.44699055e-01 1.99305534e-01 -6.82410479e-01 8.12920034e-01 -6.55903757e-01 -1.71047524e-01 -3.79619569e-01 3.91021557e-02 7.92951167e-01 1.05217922e+00 -1.15371001e+00 8.10537159e-01 -2.32723951e+00 -4.26595025e-02 1.13234632e-01 -3.98958713e-01 4.01919812e-01 -4.54691321e-01 2.31085852e-01 -2.80558109e-01 -1.53322950e-01 -8.87124181e-01 4.62419242e-01 -9.28941548e-01 3.76323521e-01 -3.55027288e-01 7.22892523e-01 -1.73642799e-01 3.40219170e-01 -7.25452244e-01 -3.89155686e-01 4.73753959e-01 5.05208790e-01 4.67529774e-01 -1.64183065e-01 2.02890530e-01 4.10114557e-01 -4.06998962e-01 7.14280128e-01 1.45493674e+00 2.29359061e-01 -1.10433418e-02 -4.65990335e-01 -8.02812397e-01 -4.48846072e-01 -1.20222139e+00 1.41058195e+00 -6.63915813e-01 3.48519355e-01 5.33122718e-01 -8.02336097e-01 1.13591444e+00 3.13515872e-01 6.76318347e-01 -7.04266608e-01 -5.09733558e-01 4.88721400e-01 -4.93724376e-01 -4.13956791e-01 9.07820225e-01 -6.48763895e-01 3.96353602e-01 3.45517784e-01 -4.32422012e-01 -5.48348129e-01 4.11587395e-02 -2.97929049e-01 2.45768189e-01 2.73568451e-01 5.18369615e-01 -2.70306379e-01 7.15357184e-01 4.96683985e-01 4.75654393e-01 3.46693695e-01 9.25228819e-02 6.37036026e-01 -4.20656741e-01 -1.96127445e-01 -1.35254800e+00 -9.22163367e-01 -5.06091654e-01 7.96501994e-01 2.11705342e-01 2.95751870e-01 -7.91908950e-02 2.72212774e-01 -1.10450961e-01 5.37023365e-01 -3.96942705e-01 2.35822320e-01 -3.42181087e-01 -1.59608150e+00 2.21610919e-01 3.95681299e-02 1.14588809e+00 -7.21934497e-01 -9.33233619e-01 4.02812660e-01 -2.32737646e-01 -9.01345909e-01 -4.20580246e-02 -3.06139767e-01 -1.26653337e+00 -9.72559452e-01 -9.78288949e-01 -2.41956651e-01 6.26472011e-02 1.26029944e+00 5.49032032e-01 -4.60914999e-01 -3.96817595e-01 3.52283925e-01 -6.56441808e-01 1.88430890e-01 -4.04243737e-01 -5.18946230e-01 -3.52927238e-01 3.81623179e-01 -4.26526070e-02 -8.76073241e-01 -3.72575223e-01 2.40216717e-01 -1.31490874e+00 2.90069580e-01 8.62284541e-01 8.16662967e-01 1.03158748e+00 8.64342690e-01 2.67518610e-01 -4.26685810e-01 1.70506760e-01 -2.23278701e-01 -9.37413037e-01 3.53877187e-01 -4.89398837e-01 -4.83658940e-01 3.44897568e-01 -2.99799323e-01 -1.49331689e+00 1.98544879e-02 9.60858092e-02 -1.59276292e-01 -1.70405775e-01 1.00930154e+00 2.17758968e-01 -3.70547771e-01 8.23336542e-01 9.69498575e-01 -7.21348524e-02 -7.93190956e-01 1.06443703e-01 1.03002226e+00 8.78115654e-01 -1.34904474e-01 9.05013740e-01 9.14410830e-01 7.19071478e-02 -1.55268848e+00 -6.74420357e-01 -8.30154896e-01 -5.69087923e-01 -2.41761729e-01 7.98986733e-01 -1.07268584e+00 1.91181421e-01 7.62132764e-01 -8.11771095e-01 -1.39992520e-01 -8.47396851e-02 7.41083205e-01 -3.97162467e-01 6.64287984e-01 -6.11401975e-01 -1.08199203e+00 -5.17732322e-01 -7.00248480e-01 9.44766343e-01 3.42594594e-01 4.76717800e-01 -5.31135261e-01 2.22939715e-01 3.88768047e-01 6.43626630e-01 4.67935622e-01 3.72612536e-01 3.52764904e-01 -3.84371281e-01 -7.43172541e-02 -7.09918916e-01 4.55289423e-01 5.97286701e-01 1.55200347e-01 -6.98941171e-01 -2.38064036e-01 4.10765707e-01 7.77071565e-02 1.12213075e+00 8.09946120e-01 7.46852458e-01 -1.59754321e-01 -4.84839305e-02 7.40617454e-01 2.20316005e+00 1.83801189e-01 7.19464660e-01 6.66158974e-01 3.71062219e-01 4.53312039e-01 1.09230721e+00 4.30273056e-01 -8.74490067e-02 3.56073588e-01 3.37432057e-01 -2.85218000e-01 4.74962965e-02 2.89717436e-01 1.70583069e-01 2.60380059e-01 -6.66016579e-01 3.94542098e-01 -8.25029016e-01 7.55144179e-01 -1.55784631e+00 -1.30538559e+00 -6.57092750e-01 2.33602285e+00 7.73080647e-01 -5.18600464e-01 -6.36913031e-02 4.85375375e-01 1.11932778e+00 8.37664127e-01 -1.75286531e-01 1.90661669e-01 -7.50025749e-01 3.21139216e-01 1.15754890e+00 6.61123157e-01 -1.13524020e+00 6.47476733e-01 4.92829943e+00 5.68707049e-01 -1.32904387e+00 2.65584677e-01 5.91094419e-02 2.81460732e-01 -3.87123883e-01 1.40111983e-01 -1.36334017e-01 1.39998198e-01 8.74582529e-01 -3.67196232e-01 7.65109837e-01 4.35800970e-01 6.37081325e-01 -7.64161348e-01 4.05535363e-02 1.07067823e+00 -3.63517553e-01 -9.63552117e-01 -8.72105286e-02 -7.15961903e-02 7.87553072e-01 2.62145668e-01 -1.35661855e-01 -4.92841333e-01 2.07715139e-01 -5.25281370e-01 3.39109808e-01 6.86267555e-01 9.91960466e-01 -7.00771511e-01 5.67786455e-01 4.10821795e-01 -1.52370715e+00 -6.64727986e-02 -5.88390470e-01 1.67844996e-01 4.03697602e-02 1.00517201e+00 -4.78110313e-02 1.29313004e+00 8.10143173e-01 9.54023838e-01 -1.73091829e-01 1.00857043e+00 -1.67153984e-01 4.62764472e-01 -4.64583546e-01 7.66587913e-01 2.97577232e-01 -7.39382863e-01 7.16137528e-01 8.50641310e-01 8.25010240e-01 8.66257071e-01 6.71393499e-02 8.29778075e-01 5.21891177e-01 -2.81892531e-02 -4.58248675e-01 -1.67281747e-01 3.68620038e-01 1.20154810e+00 -4.72847104e-01 -2.74776042e-01 -4.36687857e-01 9.78109062e-01 -8.96794975e-01 5.70311368e-01 -3.12002867e-01 -1.07595980e-01 2.64839858e-01 -6.32147267e-02 6.39832318e-02 -4.90419775e-01 -3.39657634e-01 -1.18155241e+00 -3.58123630e-01 -7.04469502e-01 5.67838013e-01 -1.26057673e+00 -8.04854453e-01 5.70398629e-01 1.63145885e-01 -1.43155301e+00 1.42190799e-01 -1.16075732e-01 -3.63896042e-01 1.61474931e+00 -2.15369749e+00 -1.38105011e+00 -5.95304787e-01 7.01564908e-01 5.03968656e-01 2.85247147e-01 7.44494081e-01 3.71052884e-02 -2.20851257e-01 -6.97288275e-01 4.70192164e-01 -3.87736946e-01 3.96735370e-01 -7.48646319e-01 -1.57192245e-01 1.26203918e+00 -3.66382092e-01 2.69643981e-02 8.48587155e-01 -8.74462426e-01 -1.24516773e+00 -1.14826024e+00 6.37664497e-01 6.46673441e-01 6.12946332e-01 7.54710972e-01 -9.89304006e-01 7.40589499e-01 -1.74754798e-01 4.60772067e-02 4.45301980e-01 -4.83541012e-01 -1.64529651e-01 -6.16471827e-01 -1.34380579e+00 6.28368407e-02 3.66051465e-01 -5.93285024e-01 -7.33141303e-01 2.61323750e-01 3.18578601e-01 -2.67664883e-02 -1.06051433e+00 8.31958592e-01 6.30671203e-01 -1.17563224e+00 1.12937570e+00 3.22229236e-01 5.04851997e-01 -5.69312453e-01 -6.29724741e-01 -1.31298304e+00 -5.76755106e-01 9.07231402e-03 8.51189613e-01 7.38326252e-01 1.33060232e-01 -7.47557521e-01 6.81419522e-02 -6.70706257e-02 9.08583775e-02 1.82164133e-01 -7.93759644e-01 -8.32156599e-01 -2.57796854e-01 -2.45953098e-01 6.05586469e-01 1.05050683e+00 -5.80984831e-01 -2.06211358e-01 -4.59148198e-01 9.30304825e-01 1.12234366e+00 9.20678616e-01 2.13227451e-01 -1.40749931e+00 -2.29850382e-01 -1.68170527e-01 1.42853662e-01 -3.04105222e-01 -2.59834617e-01 -3.91404301e-01 -1.13137662e-01 -1.55997241e+00 3.36810201e-01 -3.62513334e-01 -1.78835407e-01 4.68006402e-01 -1.24890722e-01 5.72361410e-01 1.32052884e-01 8.36811304e-01 5.13585448e-01 2.12167740e-01 1.16508174e+00 -1.28035992e-01 -5.08080602e-01 6.38752431e-02 -8.57640803e-02 5.12003541e-01 6.81872368e-01 -3.21620405e-01 -9.39187780e-02 -3.84386718e-01 3.29259899e-04 9.04454231e-01 7.13848293e-01 -9.14767921e-01 -1.65402547e-01 -7.51252770e-01 2.93331295e-01 -9.48210776e-01 3.82080734e-01 -1.04073477e+00 8.60108972e-01 4.06291962e-01 7.17046335e-02 -5.46660185e-01 3.46902907e-01 2.14234814e-01 -4.77662593e-01 -2.51998305e-01 1.41537547e+00 -4.60241795e-01 -1.19122815e+00 3.09200943e-01 -1.61918044e-01 -9.20792699e-01 7.50944078e-01 -2.57704496e-01 -3.37021261e-01 -9.72984508e-02 -6.77849889e-01 -1.13991298e-01 7.33231366e-01 -1.50007114e-01 7.97623038e-01 -1.13121951e+00 -9.68969822e-01 2.76994973e-01 8.28069001e-02 -2.78553277e-01 9.41082656e-01 8.39849114e-01 -6.94158614e-01 1.01207055e-01 -8.03956628e-01 -5.46250403e-01 -1.43063843e+00 3.13510835e-01 5.61775208e-01 -5.65845408e-02 -6.26827002e-01 2.94019848e-01 -4.01653141e-01 -3.10162187e-01 -9.77943182e-01 -5.56790084e-02 -3.57379794e-01 2.44942203e-01 7.46767282e-01 4.77526814e-01 -1.25801265e-01 -1.05813193e+00 -2.73178488e-01 1.13181889e+00 5.46247840e-01 -5.10616362e-01 1.82305205e+00 -4.63213027e-01 -7.33105898e-01 5.66532791e-01 1.04014838e+00 1.39813021e-01 -9.71827805e-01 -5.14824212e-01 -2.28071138e-01 -8.43032300e-01 7.50071585e-01 -6.01377785e-01 -1.16910863e+00 6.25823081e-01 6.40250623e-01 3.41276437e-01 1.99227989e+00 -2.87642330e-01 7.34586120e-01 -3.90348174e-02 4.16230679e-01 -9.87919748e-01 -7.01157928e-01 2.81226132e-02 9.17921245e-01 -1.15782726e+00 4.37110931e-01 -5.69958687e-01 -3.50934833e-01 1.49005437e+00 -2.57078320e-01 -6.11429624e-02 3.90057862e-01 -5.41926250e-02 -1.50387865e-02 -5.48076034e-02 -6.06939457e-02 -4.95741218e-01 5.18201403e-02 3.30436975e-01 1.11647725e-01 4.29348290e-01 -5.56427777e-01 1.69489961e-02 2.53531951e-02 4.10499126e-01 8.38476002e-01 8.03336143e-01 -9.10826027e-01 -7.60226130e-01 -1.22287524e+00 4.86979663e-01 -2.19972998e-01 -7.13553503e-02 2.22321048e-01 5.21593571e-01 -1.04928657e-01 9.96905088e-01 -1.46921590e-01 -1.75678805e-01 9.62716565e-02 -1.39936522e-01 3.49343479e-01 -3.21371585e-01 1.15152471e-01 5.39660811e-01 -8.78930464e-02 -2.75850505e-01 -1.10280287e+00 -9.23697650e-01 -9.26716805e-01 -3.34919184e-01 -2.61029094e-01 1.22206308e-01 8.60965312e-01 7.11090982e-01 -2.59584248e-01 1.34863093e-01 1.02303278e+00 -1.00982249e+00 -3.78878355e-01 -9.79298055e-01 -1.57204378e+00 -4.78501879e-02 8.06274712e-01 -1.89658403e-01 -5.93373716e-01 3.80544849e-02]
[10.114933967590332, -2.0743682384490967]
4f02cf89-2b52-4709-99b2-d99b3b43318e
scalable-logo-recognition-using-proxies
1811.08009
null
http://arxiv.org/abs/1811.08009v1
http://arxiv.org/pdf/1811.08009v1.pdf
Scalable Logo Recognition using Proxies
Logo recognition is the task of identifying and classifying logos. Logo recognition is a challenging problem as there is no clear definition of a logo and there are huge variations of logos, brands and re-training to cover every variation is impractical. In this paper, we formulate logo recognition as a few-shot object detection problem. The two main components in our pipeline are universal logo detector and few-shot logo recognizer. The universal logo detector is a class-agnostic deep object detector network which tries to learn the characteristics of what makes a logo. It predicts bounding boxes on likely logo regions. These logo regions are then classified by logo recognizer using nearest neighbor search, trained by triplet loss using proxies. We also annotated a first of its kind product logo dataset containing 2000 logos from 295K images collected from Amazon called PL2K. Our pipeline achieves 97% recall with 0.6 mAP on PL2K test dataset and state-of-the-art 0.565 mAP on the publicly available FlickrLogos-32 test set without fine-tuning.
['Srikar Appalaraju', 'Istvan Fehervari']
2018-11-19
null
null
null
null
['logo-recognition']
['computer-vision']
[-1.02010253e-03 -4.06659126e-01 -5.85576534e-01 -4.33035553e-01 -9.52126145e-01 -8.40420783e-01 4.09719795e-01 3.39826420e-02 1.06227875e-01 2.72929911e-02 -6.09711558e-02 3.33718449e-01 -8.11159983e-02 -7.83232927e-01 -1.13336611e+00 -4.37439442e-01 -1.28065675e-01 9.96849239e-01 4.57477897e-01 6.11266345e-02 3.42037529e-01 7.32666790e-01 -1.62087905e+00 4.76244241e-01 5.62793851e-01 1.64725935e+00 -1.35251423e-02 8.45901132e-01 7.97913074e-02 9.32854652e-01 -5.72494686e-01 -1.08590536e-01 7.13840127e-01 -1.98636979e-01 -5.16473651e-01 3.47034961e-01 1.30140471e+00 -3.75566959e-01 -4.38226283e-01 9.74701703e-01 3.22249025e-01 4.93825674e-02 1.06589675e+00 -1.14817095e+00 -9.94814336e-01 2.54809648e-01 -5.44862568e-01 4.57973808e-01 7.41533488e-02 2.26317391e-01 1.34155059e+00 -1.19223952e+00 5.09142637e-01 9.75197494e-01 7.81206965e-01 9.39996094e-02 -1.06499612e+00 -7.58476496e-01 -5.89443862e-01 5.63113272e-01 -1.84307623e+00 -4.23391670e-01 4.46575344e-01 -8.05176020e-01 1.48968172e+00 -9.06618759e-02 4.04642910e-01 6.54880822e-01 5.68805277e-01 8.26094568e-01 8.31446350e-01 -3.74462187e-01 1.22556321e-01 5.11294484e-01 3.62352192e-01 7.21973956e-01 2.75457501e-01 2.66850859e-01 -1.19477320e+00 1.25909641e-01 2.44724870e-01 3.95236611e-01 4.23979670e-01 -4.57247406e-01 -4.40114737e-01 7.02152312e-01 7.90427327e-01 -1.26957685e-01 -1.32856846e-01 2.81303316e-01 3.13278854e-01 3.64179313e-01 3.69842172e-01 1.66130513e-01 -2.23232895e-01 9.44540203e-02 -9.89217222e-01 6.23021871e-02 1.01987076e+00 1.16190219e+00 1.04030156e+00 -3.43700230e-01 3.05055052e-01 1.12030375e+00 2.01299071e-01 5.11542439e-01 5.96236825e-01 -3.38351220e-01 3.55273336e-01 7.08765268e-01 -8.45104903e-02 -6.86909676e-01 -1.66333556e-01 -1.73650011e-01 -1.08168781e-01 4.27554071e-01 1.68664843e-01 5.93410015e-01 -1.34812570e+00 5.38976490e-01 1.18913889e-01 -6.91587627e-02 -2.68733352e-01 5.52372754e-01 6.85285211e-01 7.76307881e-01 -4.15010840e-01 5.36393225e-01 1.53455949e+00 -1.49869871e+00 -4.83955085e-01 -9.72013593e-01 6.17029428e-01 -7.17312992e-01 1.20984817e+00 2.82959014e-01 -6.97458461e-02 -3.77392560e-01 -1.36325884e+00 -1.26677752e-01 -8.26193154e-01 2.34349310e-01 7.46750414e-01 4.82562840e-01 -3.53401631e-01 3.83725643e-01 -8.37212682e-01 -6.72062397e-01 1.06501615e+00 2.32906491e-01 -3.72894019e-01 -2.13184208e-01 -6.53560877e-01 5.86673200e-01 8.21199238e-01 -4.32392694e-02 -1.95438039e+00 -5.44226825e-01 -8.54497492e-01 3.39273632e-01 7.59167552e-01 1.25697106e-01 1.11026263e+00 -1.08866799e+00 -9.35289025e-01 1.26550090e+00 -3.37939188e-02 -8.11080694e-01 3.37339878e-01 -6.59574866e-01 -6.42708302e-01 7.84914717e-02 4.17980731e-01 4.58221614e-01 1.18301761e+00 -9.08823490e-01 -9.71808553e-01 -3.50649923e-01 -5.03345013e-01 2.22631227e-02 -7.04094693e-02 8.33339244e-02 -2.46815339e-01 -3.94670486e-01 -2.70495471e-02 -8.92967641e-01 3.02327186e-01 1.29612699e-01 -4.98869807e-01 -3.56637865e-01 7.21209586e-01 -5.11331677e-01 1.24662864e+00 -2.45231700e+00 -9.72250581e-01 -1.49667159e-01 1.33206159e-01 -1.08037181e-02 9.94610935e-02 4.55518305e-01 9.65813547e-02 -1.49384663e-01 1.15255564e-02 -1.30652199e-02 3.19840729e-01 -8.06158874e-03 -5.95195651e-01 7.20171511e-01 7.46065527e-02 1.15173841e+00 -5.18941820e-01 -5.15565336e-01 1.34515792e-01 2.26810738e-01 -2.94522405e-01 1.16321504e-01 -4.70947444e-01 -2.98712552e-01 -1.82692513e-01 1.54630530e+00 5.36578298e-01 -5.07310390e-01 -5.03200769e-01 1.39778450e-01 2.06905454e-01 3.20056736e-01 -9.49033797e-01 1.38651884e+00 -3.83073002e-01 9.70057189e-01 -5.31781971e-01 -5.98297536e-01 1.15836120e+00 -2.45354325e-01 2.42196590e-01 -4.82918918e-01 9.00595039e-02 4.65315551e-01 -4.66920078e-01 -5.14735222e-01 4.97449607e-01 -1.32440895e-01 -1.31011724e-01 2.40416050e-01 2.47698039e-01 1.80748150e-01 1.06216989e-01 -1.75873205e-01 1.33847475e+00 1.14002660e-01 6.52496457e-01 -1.47766709e-01 4.68076393e-02 5.21313310e-01 4.58670706e-01 1.01423383e+00 -4.92269069e-01 9.18614805e-01 3.43664199e-01 -7.61060834e-01 -1.01612246e+00 -1.31797254e+00 -3.63967627e-01 1.56123471e+00 2.54659593e-01 -2.29275301e-01 -3.20809633e-01 -9.48417187e-01 3.89617115e-01 6.32835627e-01 -9.10670042e-01 -1.66032031e-01 -5.87400496e-01 -6.02205217e-01 6.53442562e-01 6.61244214e-01 5.79728544e-01 -1.33832431e+00 -5.08934438e-01 7.83796832e-02 5.42080641e-01 -9.35360193e-01 -1.00329828e+00 7.42848337e-01 -5.94065964e-01 -1.17731297e+00 -3.71822894e-01 -1.32067752e+00 4.74069834e-01 2.61439174e-01 1.15763998e+00 -7.81392217e-01 -8.85518551e-01 1.67281590e-02 -1.47624791e-01 -5.42316437e-01 -3.13802451e-01 1.38417050e-01 -1.30984381e-01 3.93118322e-01 1.18576193e+00 -4.50523496e-01 -5.26920497e-01 5.83370030e-01 -7.46028006e-01 -6.10026360e-01 8.30935776e-01 7.38490760e-01 1.12628043e+00 1.70384303e-01 2.27550089e-01 -9.37404275e-01 8.08183998e-02 -5.86138070e-01 -9.59581792e-01 5.00017881e-01 -9.73996520e-01 -1.93664804e-02 6.22878611e-01 -5.86718023e-01 -8.25589538e-01 8.16630065e-01 3.21875870e-01 -5.75395346e-01 -2.10683346e-01 1.05584137e-01 -3.14920455e-01 7.37652853e-02 1.13953435e+00 3.82335216e-01 -5.01487076e-01 -9.36569333e-01 3.38276148e-01 1.05788076e+00 6.21623635e-01 3.31642270e-01 8.25988233e-01 9.28537071e-01 -3.50197941e-01 -8.49815667e-01 -1.47072732e+00 -1.18187094e+00 -7.95879543e-01 2.94238552e-02 7.94450223e-01 -8.61811996e-01 -6.35341048e-01 6.51356816e-01 -7.52347469e-01 -4.02762502e-01 -5.52156031e-01 1.33983552e-01 -5.52946210e-01 -2.41587609e-01 -3.61637771e-01 -4.74453002e-01 -3.79741848e-01 -5.83004594e-01 1.28080237e+00 4.34697419e-01 2.09127963e-01 -7.01816320e-01 3.28555048e-01 5.91045260e-01 1.73754096e-01 9.23514068e-02 3.51720959e-01 -1.22868598e+00 -1.02807212e+00 -1.02544582e+00 -2.64591455e-01 4.58922416e-01 2.75887903e-02 -4.26189780e-01 -1.20927465e+00 -6.97466582e-02 3.27475145e-02 -7.28265047e-01 1.27689505e+00 1.80982858e-01 9.47056770e-01 -3.26325625e-01 -4.60766256e-01 1.01886964e+00 1.72526634e+00 1.45855278e-01 5.27371645e-01 5.78737676e-01 9.49030876e-01 4.37576175e-01 9.81177032e-01 2.78683484e-01 1.32005677e-01 6.62283540e-01 4.44202006e-01 5.27778268e-01 -1.64243773e-01 -6.90310955e-01 7.33626842e-01 3.26713324e-02 7.50149250e-01 -4.46601033e-01 -1.09905779e+00 7.51604855e-01 -1.60824656e+00 -8.01496923e-01 1.65413842e-01 2.18232298e+00 4.52462077e-01 3.08409631e-01 1.07523508e-01 -1.83001220e-01 7.66249418e-01 4.77782547e-01 -8.21343184e-01 -7.40445182e-02 -1.61957294e-01 -1.57719627e-01 1.44804370e+00 4.66233909e-01 -1.41457701e+00 1.19572043e+00 5.85438061e+00 1.29488587e+00 -9.84749079e-01 4.54438508e-01 4.33430970e-01 -3.78664851e-01 5.18482566e-01 1.19488098e-01 -1.52772319e+00 3.17290932e-01 8.86111856e-01 -1.15394458e-01 6.34491503e-01 1.49937975e+00 -2.79102772e-01 -4.45975959e-01 -1.34309995e+00 1.24128973e+00 7.04010904e-01 -1.20141196e+00 -3.31102431e-01 4.28737998e-01 8.22763920e-01 6.22902989e-01 -1.13660365e-01 3.88744563e-01 1.02959044e-01 -1.62338746e+00 9.40406382e-01 2.22151816e-01 8.68842006e-01 -4.87315804e-01 5.25281012e-01 2.11495414e-01 -1.53094232e+00 -1.55484542e-01 -4.98483926e-01 3.40062678e-01 5.23265861e-02 3.44075978e-01 -1.26747572e+00 -1.70418873e-01 1.09663796e+00 1.04709601e+00 -1.00426531e+00 1.27644610e+00 -3.04547042e-01 8.66533279e-01 -7.19233513e-01 -2.28486136e-01 2.20878974e-01 2.65015304e-01 4.05463010e-01 1.03655195e+00 -8.93664286e-02 -6.08278096e-01 2.74391353e-01 8.39654624e-01 -3.20147038e-01 2.36454010e-02 -6.77638769e-01 -4.48259026e-01 5.24278045e-01 1.08503377e+00 -9.75461185e-01 -3.56832027e-01 -2.89525121e-01 1.37900329e+00 2.38758072e-01 -1.27496183e-01 -8.73225093e-01 -8.65383625e-01 4.34269160e-01 6.41147554e-01 6.70271158e-01 1.15041740e-01 -1.90310016e-01 -8.39063942e-01 3.42751980e-01 -3.48783404e-01 7.23149180e-01 -4.07009542e-01 -1.69862413e+00 2.54676342e-01 -1.54825181e-01 -1.45511818e+00 1.66219324e-01 -1.00371385e+00 -6.39593542e-01 4.09482539e-01 -1.25096142e+00 -1.52335298e+00 -6.90212488e-01 -1.26104802e-01 8.71004522e-01 -2.14480549e-01 2.51279950e-01 5.17606795e-01 -2.86967903e-01 6.44332767e-01 6.40028775e-01 3.56067777e-01 9.29273069e-01 -1.31296647e+00 5.23764312e-01 4.00462151e-01 4.65389282e-01 1.99359775e-01 4.75563288e-01 -7.07162857e-01 -1.22834790e+00 -1.43492198e+00 1.05338967e+00 -9.86839175e-01 1.07341254e+00 -5.88275254e-01 -9.62247252e-01 1.11738288e+00 -5.46956062e-01 7.23344743e-01 5.35876870e-01 -3.40657234e-01 -6.52637780e-01 -7.64544845e-01 -9.08980191e-01 -2.83977598e-01 1.06860399e+00 -7.83365846e-01 -5.36379099e-01 9.42076325e-01 3.66028637e-01 -1.61973044e-01 -6.10805511e-01 -2.46712584e-02 7.72789955e-01 -8.18968534e-01 8.75765622e-01 -4.47196007e-01 3.06336641e-01 -3.90001178e-01 -2.74222553e-01 -6.32056952e-01 2.54180580e-02 -4.52376723e-01 -2.18171701e-01 1.43411171e+00 3.75566244e-01 -7.17819273e-01 1.30546021e+00 -2.00904980e-01 -1.63425893e-01 -5.13140619e-01 -1.00677228e+00 -1.52677441e+00 -2.22760722e-01 -4.91820395e-01 4.10071820e-01 4.82530504e-01 -5.87359846e-01 6.02255762e-01 -5.39633632e-01 2.62230217e-01 6.35843515e-01 2.29128182e-01 8.47105324e-01 -8.27773869e-01 -6.04565144e-01 -2.42022991e-01 -8.54116023e-01 -1.28025401e+00 -3.51347476e-02 -1.19221222e+00 4.67548758e-01 -1.29596162e+00 3.50937754e-01 -3.62311959e-01 -2.44432047e-01 4.63618517e-01 3.52120817e-01 7.01573253e-01 -1.54538810e-01 7.77637005e-01 -9.79817510e-01 5.27173877e-01 4.67810154e-01 -4.87243772e-01 -3.65237832e-01 -2.29437295e-02 -1.66846186e-01 7.02561975e-01 6.40483081e-01 -1.16518807e+00 6.02725595e-02 -1.21821418e-01 2.30276778e-01 -5.65949202e-01 4.97128844e-01 -1.13255107e+00 6.36241436e-02 -5.70132062e-02 2.63708770e-01 -1.19346607e+00 2.37953275e-01 -7.57608891e-01 1.89841554e-01 6.96811616e-01 -1.32351935e-01 -4.92003143e-01 -2.11761951e-01 9.40664351e-01 -2.36029625e-01 -7.06610560e-01 1.06477225e+00 -9.44336131e-02 -1.43244076e+00 3.62018347e-01 -1.47121362e-02 1.20598391e-01 1.27674949e+00 -6.64614618e-01 -4.03182179e-01 1.28324712e-02 -5.74866593e-01 2.48131633e-01 5.97599864e-01 5.50699234e-01 3.05607498e-01 -1.24158156e+00 -4.62532282e-01 1.99115023e-01 1.04114270e+00 1.48714324e-02 -1.17345117e-01 4.67940271e-01 -1.40114176e+00 6.50887728e-01 7.67550152e-03 -7.09371686e-01 -1.27262068e+00 6.75029755e-01 8.63680780e-01 -2.42099062e-01 -6.80121541e-01 9.76104498e-01 5.26108801e-01 -2.95911819e-01 4.61761147e-01 1.20140336e-01 -6.63706139e-02 1.56361908e-01 7.97238469e-01 5.69606721e-01 1.20894633e-01 -6.99477434e-01 -7.51505733e-01 6.44634545e-01 -7.47254565e-02 2.32050464e-01 1.43306255e+00 -7.48636648e-02 1.21593215e-01 7.82351017e-01 1.43723333e+00 -9.52500105e-02 -1.39141738e+00 -3.05747330e-01 5.13048589e-01 -7.84353375e-01 -3.46647292e-01 -8.56466651e-01 -7.66801059e-01 9.61921513e-01 1.11479592e+00 8.98848176e-02 7.78914511e-01 7.56700873e-01 8.11224043e-01 5.76252341e-01 3.09696585e-01 -1.27214193e+00 3.80486131e-01 5.62573791e-01 8.36749375e-01 -1.76218963e+00 8.42716247e-02 -3.78111720e-01 -5.68305254e-01 1.02322984e+00 6.43737555e-01 -5.49105883e-01 7.27498233e-01 -1.63071938e-02 4.11686525e-02 -6.43985927e-01 -5.59552729e-01 -1.95581809e-01 6.07000530e-01 5.26278973e-01 -6.10231943e-02 2.31812283e-01 1.31514594e-01 6.15988255e-01 -3.36016685e-01 -1.50459766e-01 1.64679348e-01 1.01990783e+00 -9.49334562e-01 -4.06376600e-01 -4.52031434e-01 7.26749480e-01 -4.65970874e-01 -1.20247871e-01 -6.97659135e-01 6.48947597e-01 1.44023165e-01 5.42273581e-01 1.93312436e-01 -4.70804542e-01 1.27018422e-01 4.18618508e-02 1.66271597e-01 -9.83022213e-01 -4.72476602e-01 1.58772413e-02 -6.04550056e-02 -5.45362830e-01 4.29616332e-01 -5.12336969e-01 -1.06826806e+00 -1.08451635e-01 -6.55146182e-01 -2.95286298e-01 6.42630517e-01 6.84831858e-01 2.95094520e-01 1.83921576e-01 8.24667990e-01 -7.64160693e-01 -8.13658893e-01 -1.00433993e+00 -1.14053547e+00 4.63929951e-01 2.98693925e-01 -4.34585303e-01 -5.08096039e-01 3.60432714e-01]
[9.31908893585205, 1.306208610534668]
745bc38b-4504-401b-868c-b3ad66151f8b
cross-domain-review-generation-for-aspect
null
null
https://aclanthology.org/2021.findings-acl.421
https://aclanthology.org/2021.findings-acl.421.pdf
Cross-Domain Review Generation for Aspect-Based Sentiment Analysis
null
['Rui Xia', 'Chenggong Gong', 'Jianfei Yu']
null
null
null
null
findings-acl-2021-8
['review-generation']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.399898529052734, 3.6167614459991455]
f0b9aa49-4cdf-4652-9d52-b16ade87fc74
t-recx-tiny-resource-efficient-convolutional
2207.06613
null
https://arxiv.org/abs/2207.06613v2
https://arxiv.org/pdf/2207.06613v2.pdf
T-RECX: Tiny-Resource Efficient Convolutional neural networks with early-eXit
Deploying Machine learning (ML) on milliwatt-scale edge devices (tinyML) is gaining popularity due to recent breakthroughs in ML and Internet of Things (IoT). Most tinyML research focuses on model compression techniques that trade accuracy (and model capacity) for compact models to fit into the KB-sized tiny-edge devices. In this paper, we show how such models can be enhanced by the addition of an early exit intermediate classifier. If the intermediate classifier exhibits sufficient confidence in its prediction, the network exits early thereby, resulting in considerable savings in time. Although early exit classifiers have been proposed in previous work, these previous proposals focus on large networks, making their techniques suboptimal/impractical for tinyML applications. Our technique is optimized specifically for tiny-CNN sized models. In addition, we present a method to alleviate the effect of network overthinking by leveraging the representations learned by the early exit. We evaluate T-RecX on three CNNs from the MLPerf tiny benchmark suite for image classification, keyword spotting and visual wake word detection tasks. Our results show that T-RecX 1) improves the accuracy of baseline network, 2) achieves 31.58% average reduction in FLOPS in exchange for one percent accuracy across all evaluated models. Furthermore, we show that our methods consistently outperform popular prior works on the tiny-CNNs we evaluate.
['Steve Wilton', 'Nikhil P Ghanathe']
2022-07-14
null
null
null
null
['keyword-spotting']
['speech']
[ 1.33100241e-01 5.61865605e-02 -7.17589915e-01 -4.95024502e-01 -5.05333900e-01 -2.42376328e-01 2.15912104e-01 4.64710928e-02 -7.39326596e-01 4.23139125e-01 4.29351777e-02 -7.98643172e-01 1.82834595e-01 -6.50369585e-01 -1.03126287e+00 -1.09817691e-01 -2.75671724e-02 2.20160782e-01 3.09172213e-01 4.43316847e-01 -1.64685488e-01 3.38165820e-01 -1.48032248e+00 6.44598424e-01 2.17048511e-01 1.62553465e+00 3.09852749e-01 1.01310813e+00 -4.61643189e-03 1.27686083e+00 -5.16512752e-01 -4.07600611e-01 3.27822000e-01 1.57420158e-01 -6.54497504e-01 -3.60848278e-01 6.37565315e-01 -5.69399834e-01 -7.42684007e-01 7.64983356e-01 6.94757164e-01 -3.45692128e-01 1.58228606e-01 -1.35183907e+00 -2.39705116e-01 1.12861741e+00 -3.44013214e-01 4.47173387e-01 -3.81310612e-01 3.35801512e-01 8.94300103e-01 -7.45066881e-01 3.15663189e-01 1.23191655e+00 8.98204505e-01 7.04287052e-01 -8.11681509e-01 -1.06623900e+00 -6.72537759e-02 1.94496557e-01 -1.41696811e+00 -1.04743493e+00 9.33466032e-02 2.28931665e-01 1.70605361e+00 3.32712173e-01 6.86809301e-01 1.06279242e+00 3.57761234e-01 9.05427873e-01 6.55439436e-01 -2.57139057e-01 4.65622813e-01 3.08260024e-02 2.63055444e-01 8.94577324e-01 3.84927541e-01 -2.39244133e-01 -7.12949574e-01 -2.89411843e-02 4.43347991e-01 1.31207824e-01 1.19451456e-01 3.06517124e-01 -9.65621173e-01 5.55708885e-01 5.62847435e-01 7.56353140e-03 -3.12133908e-01 1.08263457e+00 7.42792487e-01 1.78936243e-01 5.74158072e-01 5.75570166e-02 -9.04713392e-01 -5.33335328e-01 -1.11744428e+00 -1.82112098e-01 1.01732945e+00 1.26853263e+00 4.09864664e-01 1.40028208e-01 -1.47038683e-01 6.34928465e-01 3.65192324e-01 4.41566616e-01 6.85876012e-01 -9.83689249e-01 7.53726006e-01 5.41710913e-01 -4.87012804e-01 -6.42174006e-01 -4.44020033e-01 -6.02541029e-01 -9.93054509e-01 -2.68378884e-01 -9.69595090e-02 -3.04988086e-01 -1.04146898e+00 1.31987631e+00 8.68056193e-02 5.44298530e-01 5.32453470e-02 3.11199039e-01 9.73510861e-01 7.80117691e-01 2.18666151e-01 1.02009073e-01 1.31799626e+00 -1.40532422e+00 -4.80827272e-01 -6.08113825e-01 1.15848303e+00 -5.53736627e-01 1.02612782e+00 5.24676502e-01 -1.00381279e+00 -5.58746457e-01 -1.34652710e+00 -3.06506902e-01 -3.07324648e-01 1.09482035e-01 9.02265310e-01 8.13332021e-01 -1.38841259e+00 8.43759239e-01 -1.22720134e+00 -2.95546234e-01 9.53466773e-01 8.51091325e-01 -5.18841669e-02 -1.22420393e-01 -6.00125611e-01 3.90665710e-01 4.85857815e-01 -1.80870816e-01 -7.89993227e-01 -8.70387614e-01 -5.99484265e-01 3.25958699e-01 3.48852009e-01 -6.89041436e-01 1.54100454e+00 -6.33402348e-01 -1.32128990e+00 4.27227706e-01 -2.85836339e-01 -1.17195332e+00 3.08876783e-01 -3.53290945e-01 -5.48193157e-01 -4.66683507e-02 -3.72182697e-01 1.13320863e+00 6.76207602e-01 -5.89913487e-01 -7.74068475e-01 -1.08781755e-01 1.82039946e-01 -1.77930474e-01 -1.01247501e+00 3.79944146e-02 -9.19625223e-01 -4.58463192e-01 -1.72576025e-01 -1.00132740e+00 -3.18654239e-01 2.81274348e-01 -4.82076913e-01 -2.63169140e-01 1.31017816e+00 -2.54901201e-01 1.36331117e+00 -2.14166784e+00 -6.44046366e-01 -6.04141094e-02 7.31136024e-01 4.04487759e-01 -6.40699118e-02 -1.22377910e-01 4.00202334e-01 5.59615612e-01 2.99204230e-01 -8.70924294e-01 -1.08278126e-01 4.28226590e-01 -1.21831141e-01 1.77187353e-01 -1.66015178e-01 1.35001409e+00 -4.13469166e-01 -5.42284369e-01 1.48432970e-01 7.56780326e-01 -7.51344860e-01 -2.09373742e-01 -3.01477462e-01 -3.28833878e-01 -1.85424477e-01 8.97328973e-01 3.58334243e-01 -7.92936563e-01 1.19203366e-01 -5.61069608e-01 1.18897274e-01 5.79834223e-01 -7.52453625e-01 1.63656437e+00 -8.06939900e-01 7.88233101e-01 -8.06322396e-02 -5.90747356e-01 5.30307949e-01 3.47040087e-01 2.69731373e-01 -7.74929523e-01 3.29109371e-01 2.22704813e-01 -7.30983168e-02 -8.53413343e-02 3.64018083e-01 4.20423150e-01 1.55228287e-01 4.07042742e-01 1.04132339e-01 9.63188410e-02 5.35880625e-02 4.25833613e-01 1.47653294e+00 -4.34634954e-01 -3.42497323e-03 -2.80473921e-02 -3.21247578e-01 -4.10425454e-01 3.00323784e-01 1.04682660e+00 -1.30143344e-01 1.89684510e-01 2.16806322e-01 -6.88673377e-01 -1.03827763e+00 -4.94414955e-01 -6.31406680e-02 1.02519655e+00 -1.51098013e-01 -1.24524629e+00 -8.14774454e-01 -8.03777993e-01 -1.91720501e-01 4.30757105e-01 -1.94135860e-01 -2.44779319e-01 -4.89898145e-01 -9.72637177e-01 9.38865244e-01 9.14892495e-01 8.60654831e-01 -8.56353104e-01 -1.10312104e+00 2.51508623e-01 2.18719453e-01 -1.54602528e+00 -2.57953316e-01 7.50504076e-01 -1.34655893e+00 -6.24042094e-01 -4.06853892e-02 -5.14731824e-01 2.96836376e-01 2.56974608e-01 1.15099144e+00 2.73383886e-01 -2.24030003e-01 7.59882927e-02 -2.49392316e-01 -6.74825847e-01 -2.79315054e-01 4.74749655e-01 2.22547501e-01 -5.39856374e-01 4.30833459e-01 -5.90250671e-01 -6.87615216e-01 3.71419080e-02 -7.89962471e-01 2.97490507e-01 8.86205077e-01 4.59737837e-01 6.79119885e-01 -3.12388549e-03 3.97987187e-01 -8.75447452e-01 2.33961657e-01 -5.30234933e-01 -4.71810907e-01 6.63854554e-02 -1.03501558e+00 1.22441210e-01 6.60997570e-01 -7.37756133e-01 -3.18001360e-01 8.98804292e-02 -3.95953685e-01 -6.73598826e-01 9.16483179e-02 3.48305106e-01 -7.75783658e-02 -2.72173703e-01 4.80463415e-01 -1.72128230e-01 -3.87284130e-01 -4.69627500e-01 2.19632417e-01 9.27681923e-01 2.87923813e-01 -2.29899168e-01 1.39975309e-01 3.50113034e-01 -9.48041528e-02 -6.82484448e-01 -6.25518262e-01 -1.82182282e-01 -2.02592924e-01 1.10056199e-01 4.89897579e-01 -1.25108194e+00 -8.96742642e-01 2.30480745e-01 -1.06380081e+00 -8.43275368e-01 -3.22492152e-01 3.44607115e-01 -9.55031961e-02 -2.11761007e-03 -1.05709267e+00 -5.81743419e-01 -9.27837908e-01 -1.25094569e+00 1.17347670e+00 2.22321168e-01 -1.75081119e-01 -7.28811145e-01 -5.49768627e-01 8.64886865e-02 8.08416069e-01 -3.75017375e-01 7.57055998e-01 -7.06096351e-01 -7.76937008e-01 -3.02905023e-01 -4.42505360e-01 5.00728846e-01 -2.06848472e-01 -1.68478772e-01 -1.29057860e+00 -3.29193741e-01 -9.53998417e-02 -6.23575211e-01 1.13963521e+00 3.96666110e-01 1.80642462e+00 -7.03381240e-01 -6.70706511e-01 9.87443745e-01 1.49281764e+00 1.00139424e-01 5.16378045e-01 8.05486962e-02 8.72419715e-01 -3.48575354e-01 -6.61316589e-02 5.03849328e-01 2.49474660e-01 4.57449943e-01 6.22747898e-01 -1.57893434e-01 -4.26034153e-01 -1.92911163e-01 4.30009961e-01 1.15340757e+00 3.42373222e-01 -6.55622423e-01 -8.08502614e-01 4.28015202e-01 -1.77086592e+00 -3.77632469e-01 1.93272799e-01 1.85734105e+00 7.29681730e-01 7.49874890e-01 -2.32380152e-01 3.14840943e-01 1.84757039e-01 2.87670702e-01 -8.55895221e-01 -4.26226825e-01 -4.06527184e-02 5.56992292e-01 1.08348894e+00 1.73512220e-01 -9.64392722e-01 1.02436101e+00 6.10540915e+00 1.23907757e+00 -1.28177881e+00 6.03231668e-01 1.25457740e+00 -6.31433308e-01 1.50219545e-01 -1.19196646e-01 -1.37441528e+00 4.00104582e-01 1.74616408e+00 8.50936994e-02 3.35577995e-01 1.26924169e+00 -3.42033282e-02 3.72253284e-02 -1.39701581e+00 1.35315263e+00 -1.38273805e-01 -1.71263647e+00 1.83453169e-02 1.23318426e-01 5.88941157e-01 8.25455964e-01 1.97039872e-01 4.06107157e-01 3.40873390e-01 -1.17214465e+00 9.00480747e-01 7.51068369e-02 1.14645123e+00 -8.26191843e-01 7.50782967e-01 4.60506290e-01 -1.32311761e+00 -2.81310558e-01 -6.20110393e-01 -1.70882329e-01 -6.20082840e-02 8.88072968e-01 -1.17110085e+00 -1.47589996e-01 1.01160145e+00 7.42148399e-01 -8.20752680e-01 9.35364664e-01 2.92371571e-01 1.00035012e+00 -7.88229048e-01 -3.28639776e-01 2.52140433e-01 7.15963006e-01 -3.71687068e-03 1.43666315e+00 3.24024141e-01 -2.14578599e-01 -1.58864912e-02 5.40189028e-01 -9.27540123e-01 -6.43484369e-02 -4.99099731e-01 -1.50968861e-02 7.16061950e-01 1.29365098e+00 -9.99580801e-01 -7.31331229e-01 -4.52628404e-01 1.03312469e+00 2.17277318e-01 2.65991837e-02 -9.08849120e-01 -1.05814718e-01 5.36322594e-01 8.68855491e-02 3.34812880e-01 1.81975346e-02 -4.17814106e-01 -9.83278990e-01 1.04512200e-01 -8.09296310e-01 1.27154484e-01 -6.29872918e-01 -7.61451960e-01 8.11263919e-01 -2.57527769e-01 -6.11867070e-01 6.52402490e-02 -5.56737185e-01 -2.16335297e-01 1.42035723e-01 -1.25844085e+00 -1.03662932e+00 -2.07665309e-01 3.42984140e-01 9.28489685e-01 1.39741510e-01 8.33197832e-01 7.86620021e-01 -6.44598424e-01 9.86230135e-01 6.64999261e-02 -2.73108482e-04 4.52112049e-01 -8.23203564e-01 9.64854300e-01 6.60602391e-01 4.27542925e-01 4.65721339e-01 3.47646385e-01 -6.41907275e-01 -1.61068475e+00 -1.56190157e+00 8.49889934e-01 -2.94396043e-01 5.28386831e-01 -7.48380363e-01 -3.70683759e-01 1.04025614e+00 -3.35133082e-04 6.29117191e-01 3.67305309e-01 -1.13901094e-01 -5.22958398e-01 -4.61850345e-01 -1.00850224e+00 5.74465752e-01 1.17955446e+00 -5.39442956e-01 4.86504048e-01 6.07410610e-01 1.29747510e+00 -3.40572804e-01 -8.02252352e-01 5.90004861e-01 6.72461271e-01 -7.86030114e-01 9.01242614e-01 -4.81964648e-01 2.30195448e-01 2.13011369e-01 -3.86949062e-01 -7.06496894e-01 1.21330045e-01 -6.15454078e-01 -8.90472949e-01 9.06091154e-01 5.48176467e-01 -3.28180134e-01 1.08773983e+00 5.85743189e-01 -2.46723831e-01 -1.20987725e+00 -1.08628774e+00 -7.57625222e-01 -3.53252202e-01 -1.12870669e+00 5.81553221e-01 4.47295696e-01 -2.08620593e-01 4.20344591e-01 -5.32318115e-01 -1.66369319e-01 3.52894932e-01 -5.54655135e-01 6.81381762e-01 -1.00567579e+00 -5.03004014e-01 -3.14994276e-01 -4.29249257e-01 -1.58219790e+00 -1.29095092e-01 -8.31081271e-01 -1.35501698e-01 -1.26825058e+00 2.54846901e-01 -7.28496671e-01 -3.16035450e-01 9.44993019e-01 3.24886173e-01 6.90833986e-01 3.67372513e-01 1.31939322e-01 -1.21516335e+00 1.11507103e-01 5.58755994e-01 -2.95111269e-01 -6.03145137e-02 -6.22488856e-02 -6.41907096e-01 7.80908167e-01 8.71961176e-01 -6.91160798e-01 -4.96450484e-01 -7.93694973e-01 4.68868375e-01 -3.29240024e-01 3.32975596e-01 -1.62647331e+00 6.28258705e-01 5.90846717e-01 4.31766331e-01 -4.81457561e-01 4.18521166e-01 -8.38237226e-01 1.22399442e-01 6.54404402e-01 -2.35519409e-01 3.66451204e-01 4.97271836e-01 5.04712045e-01 2.45982900e-01 -4.87611033e-02 6.94133103e-01 -1.95473339e-03 -6.41319394e-01 5.81372559e-01 -2.10112646e-01 -1.60385013e-01 6.86406076e-01 -2.90619209e-02 -5.84079385e-01 -3.23361576e-01 -4.71882850e-01 -6.39004558e-02 2.45138466e-01 3.66741657e-01 7.47204065e-01 -1.03054821e+00 -2.33178496e-01 1.59379363e-01 -3.89533937e-02 8.81261304e-02 -5.32799102e-02 8.37040544e-01 -7.11382329e-01 7.18905091e-01 2.54735351e-01 -5.95631540e-01 -1.36165607e+00 4.66104656e-01 2.81182915e-01 -4.45288152e-01 -7.02453673e-01 1.20407963e+00 -2.54163772e-01 1.03337867e-02 7.83620715e-01 -9.20719028e-01 4.40909326e-01 -5.23517966e-01 7.09005654e-01 3.76596898e-01 5.35793841e-01 -2.31273323e-01 -3.06762844e-01 1.44468635e-01 -2.64264584e-01 1.62016630e-01 1.35749364e+00 5.04433550e-02 2.74090320e-02 3.80866796e-01 1.48200929e+00 -1.29881352e-01 -1.10085177e+00 -2.20667645e-01 1.36974885e-03 7.59427324e-02 4.09240127e-01 -7.79447675e-01 -1.31454372e+00 8.89936090e-01 9.75457549e-01 -9.16151479e-02 1.08261395e+00 2.13519007e-01 1.33492661e+00 8.34487140e-01 4.21030521e-01 -8.70293081e-01 3.69939730e-02 5.48723698e-01 1.55316904e-01 -1.11106122e+00 6.36917651e-02 -2.37258911e-01 -4.20775898e-02 9.50605989e-01 6.51862204e-01 2.81881213e-01 9.58177388e-01 1.04033983e+00 -2.64439940e-01 -1.77194074e-01 -1.29922569e+00 3.03824902e-01 3.56541388e-02 2.35630333e-01 1.90867186e-01 -4.93603982e-02 2.15372413e-01 6.49213195e-01 -2.30712667e-01 3.10473293e-01 1.86951607e-01 1.08648014e+00 -5.29164791e-01 -8.65869045e-01 1.17796816e-01 1.04864609e+00 -8.06993544e-01 -7.55757809e-01 -9.22224596e-02 5.18507242e-01 2.64720321e-01 9.88803983e-01 4.17624295e-01 -8.87641788e-01 -4.26605456e-02 -2.07533147e-02 2.15222344e-01 -5.90040386e-01 -6.38808668e-01 8.30158517e-02 1.04650289e-01 -8.78184438e-01 -1.67914942e-01 6.93501458e-02 -1.23150623e+00 -6.54844463e-01 -5.22342384e-01 -2.63656884e-01 1.13426566e+00 8.14049840e-01 7.99421370e-01 6.14642560e-01 3.24402191e-02 -7.54376650e-01 -3.67125332e-01 -1.07166600e+00 -2.23503858e-01 -3.99611801e-01 2.57077634e-01 -3.24994139e-02 -3.58604431e-01 -2.14122720e-02]
[8.591816902160645, 3.043600559234619]
17cf3457-24a3-4dc1-89ef-d929d016ca2a
towards-safe-autonomous-driving-policies
2307.01316
null
https://arxiv.org/abs/2307.01316v1
https://arxiv.org/pdf/2307.01316v1.pdf
Towards Safe Autonomous Driving Policies using a Neuro-Symbolic Deep Reinforcement Learning Approach
The dynamic nature of driving environments and the presence of diverse road users pose significant challenges for decision-making in autonomous driving. Deep reinforcement learning (DRL) has emerged as a popular approach to tackle this problem. However, the application of existing DRL solutions is mainly confined to simulated environments due to safety concerns, impeding their deployment in real-world. To overcome this limitation, this paper introduces a novel neuro-symbolic model-free DRL approach, called DRL with Symbolic Logics (DRLSL) that combines the strengths of DRL (learning from experience) and symbolic first-order logics knowledge-driven reasoning) to enable safe learning in real-time interactions of autonomous driving within real environments. This innovative approach provides a means to learn autonomous driving policies by actively engaging with the physical environment while ensuring safety. We have implemented the DRLSL framework in autonomous driving using the highD dataset and demonstrated that our method successfully avoids unsafe actions during both the training and testing phases. Furthermore, our results indicate that DRLSL achieves faster convergence during training and exhibits better generalizability to new driving scenarios compared to traditional DRL methods.
['Saber Fallah', 'Mustafa Yıldırım', 'Iman Sharifi']
2023-07-03
null
null
null
null
['decision-making']
['reasoning']
[-8.58821347e-02 2.53771335e-01 -2.75692016e-01 -3.46825033e-01 -3.76675248e-01 -4.94320959e-01 7.53435194e-01 -2.49880850e-01 -4.76327240e-01 1.02076864e+00 -2.83233911e-01 -7.22914815e-01 -5.51566958e-01 -9.81550395e-01 -8.91230464e-01 -4.40276176e-01 -2.51931399e-01 3.03314030e-01 4.74027008e-01 -6.92640245e-01 9.93656814e-02 9.22921598e-01 -2.33655119e+00 -1.72832720e-02 1.16317868e+00 5.36573708e-01 1.25077680e-01 4.40962225e-01 9.11325514e-02 1.14566958e+00 -1.85598955e-01 1.55392364e-01 3.73275816e-01 -6.08394556e-02 -5.77032328e-01 -2.97516137e-01 1.42926142e-01 -4.21622306e-01 -5.75103760e-01 7.65052676e-01 3.35452199e-01 4.81753975e-01 2.19040811e-01 -1.66441357e+00 -1.22810446e-01 4.71959352e-01 1.43158853e-01 5.21897990e-03 2.61058241e-01 7.67102122e-01 4.24205065e-01 -3.88864279e-01 4.48194027e-01 1.42783964e+00 6.19041383e-01 7.95358658e-01 -1.19984806e+00 -7.86057711e-01 2.85037935e-01 5.53033054e-01 -1.35260034e+00 -5.41769743e-01 6.08518243e-01 -3.63009214e-01 1.43440485e+00 -2.43344102e-02 7.91122198e-01 1.34095597e+00 4.97954518e-01 9.03961420e-01 1.19183660e+00 -2.63564855e-01 7.34314084e-01 2.09920540e-01 -1.20953120e-01 4.09557879e-01 2.31038630e-01 1.06173289e+00 -3.69241744e-01 9.11896229e-02 1.92453563e-01 -2.65922189e-01 3.42864007e-01 -5.88379681e-01 -9.13328648e-01 7.43717909e-01 2.97531784e-01 1.21701136e-02 -3.96081656e-01 3.38042021e-01 5.90638518e-01 3.21287125e-01 -1.88530669e-01 4.92967844e-01 -2.62963831e-01 -4.70711797e-01 -6.15320504e-01 7.89596438e-01 8.07560444e-01 9.21632290e-01 7.05818176e-01 4.16349530e-01 5.71755022e-02 4.13592041e-01 3.10286939e-01 7.29238629e-01 2.04187080e-01 -1.38862300e+00 1.34626850e-01 4.92731452e-01 2.65246332e-01 -9.50662792e-01 -5.31758308e-01 -1.18757598e-01 -2.95956284e-01 7.80504465e-01 1.58449277e-01 -2.12205619e-01 -6.31059468e-01 1.60296345e+00 2.97672302e-01 4.37981606e-01 6.95306778e-01 5.81694901e-01 3.84113610e-01 5.83082914e-01 2.43357152e-01 1.58410877e-01 6.62049532e-01 -7.17987478e-01 -8.34636152e-01 -4.97530520e-01 7.64890313e-01 7.33196735e-02 9.75699961e-01 7.35368133e-01 -8.55086207e-01 -6.70927882e-01 -1.12019038e+00 3.27372134e-01 -7.22984135e-01 -3.52943897e-01 7.14882314e-01 5.31137824e-01 -1.16962802e+00 4.50377464e-01 -8.89268100e-01 -3.57714772e-01 3.33420694e-01 6.08413935e-01 -1.66741252e-01 -1.22138828e-01 -1.70801628e+00 1.31074321e+00 7.22713411e-01 2.21365079e-01 -1.25367892e+00 -4.76726174e-01 -1.13930082e+00 -3.84772748e-01 7.77396679e-01 4.24291147e-03 1.46455932e+00 -4.75882947e-01 -1.86639762e+00 2.60050505e-01 2.00715601e-01 -9.08854902e-01 6.48984492e-01 -3.92989516e-01 -6.17960274e-01 -1.65149719e-01 2.73275990e-02 8.02830160e-01 4.44058955e-01 -1.35901678e+00 -8.42369437e-01 1.58742145e-01 3.29304039e-01 4.95325252e-02 1.12968005e-01 -6.17757142e-01 9.63321030e-02 5.80391176e-02 -4.90833253e-01 -1.03996360e+00 -4.26677406e-01 -2.09933639e-01 1.67295799e-01 -4.06342536e-01 1.11982346e+00 -1.51424959e-01 1.09713638e+00 -2.28078341e+00 -1.85992382e-02 3.20789695e-01 -9.46091339e-02 7.66909122e-01 -1.71875745e-01 6.52392566e-01 3.32718551e-01 -2.36373201e-01 -8.56492892e-02 1.00872658e-01 3.32847506e-01 9.71109569e-01 -4.07170445e-01 6.40713051e-02 4.00141031e-01 9.11279976e-01 -1.34292519e+00 -4.98831570e-01 7.82022119e-01 3.98251563e-01 -5.74766695e-01 1.71060905e-01 -3.75156969e-01 5.24069011e-01 -5.45504510e-01 4.38955843e-01 5.21492183e-01 5.91195226e-01 1.48524374e-01 3.68113667e-01 -4.87133205e-01 5.80802746e-02 -1.18857849e+00 1.28202116e+00 -7.67685115e-01 6.54508412e-01 -1.25320122e-01 -8.92965317e-01 1.10950196e+00 2.21987695e-01 2.46964529e-01 -1.35098422e+00 4.07733433e-02 2.48381153e-01 7.01855421e-02 -9.92554665e-01 2.62494802e-01 1.15765659e-02 -2.94171542e-01 7.72208422e-02 -2.04557836e-01 -4.48492616e-01 2.44442478e-01 -2.42008954e-01 9.85456586e-01 6.79448843e-01 1.81080535e-01 -2.41861552e-01 6.95608675e-01 2.20762745e-01 8.06398094e-01 9.17147815e-01 -7.71720350e-01 -4.58032757e-01 3.94070297e-01 -7.25173354e-01 -6.60764873e-01 -9.28640962e-01 -3.24498937e-02 7.72933304e-01 3.10125858e-01 -1.36545241e-01 -7.90505171e-01 -4.91686314e-01 3.03908199e-01 1.52970612e+00 -5.78590989e-01 -6.61321580e-01 -6.49883032e-01 3.92038608e-03 7.59171188e-01 5.88351190e-01 8.08576941e-01 -1.32029390e+00 -1.14841259e+00 4.04322088e-01 1.83916509e-01 -1.28105175e+00 3.71042192e-01 3.50826502e-01 -4.47039723e-01 -9.67386425e-01 3.19914043e-01 -4.87406403e-01 2.14269221e-01 -4.44474742e-02 7.02020168e-01 -1.52766824e-01 -2.63504446e-01 3.80379617e-01 -1.77391976e-01 -5.12944639e-01 -7.55212307e-01 -2.45022196e-02 3.01328629e-01 -1.77744821e-01 7.14698255e-01 -4.05388355e-01 -2.43373156e-01 4.79153037e-01 -6.98441625e-01 -7.20221084e-03 6.57418549e-01 7.16703415e-01 4.82400447e-01 5.49787521e-01 9.06313539e-01 -6.48639977e-01 8.39816034e-01 -5.03282785e-01 -9.92182195e-01 1.35451004e-01 -7.88232148e-01 2.06771970e-01 7.34071136e-01 -3.87131840e-01 -1.03619611e+00 8.52728784e-02 -2.03789800e-01 -2.11389497e-01 -6.58553243e-01 3.70763808e-01 -2.51506686e-01 -1.97264031e-01 6.73753798e-01 2.66036808e-01 3.62423569e-01 9.37471241e-02 2.23653749e-01 7.34068394e-01 4.10150319e-01 -7.55348027e-01 7.79524565e-01 1.30542964e-01 2.26321608e-01 -6.74653530e-01 -5.25511444e-01 9.05679613e-02 -6.35598183e-01 -6.60753548e-01 7.83553302e-01 -9.47486043e-01 -1.16174209e+00 3.66299719e-01 -5.15322864e-01 -1.01363885e+00 -4.95677978e-01 4.70189601e-01 -9.15357351e-01 -8.22424814e-02 -5.88141270e-02 -1.19676673e+00 2.53089815e-01 -1.44429219e+00 6.71895325e-01 2.23804578e-01 -3.92656267e-01 -9.28488433e-01 9.84788314e-03 1.87488675e-01 6.81231976e-01 6.60326600e-01 7.65787363e-01 -3.54827166e-01 -4.78435457e-01 -2.18141034e-01 2.26506144e-01 4.09263879e-01 -8.32696259e-02 2.67509688e-02 -9.12487507e-01 -1.82444036e-01 -3.13007176e-01 -6.48675621e-01 2.38824323e-01 -7.62983318e-03 1.00442362e+00 -1.79158017e-01 -2.85152763e-01 1.39075965e-01 1.26338863e+00 6.30561650e-01 7.43594050e-01 8.38731468e-01 3.88937265e-01 7.91033089e-01 1.15614998e+00 3.26758057e-01 6.80810988e-01 6.17306292e-01 6.97042286e-01 4.08926308e-02 1.44530982e-01 -4.39818054e-01 6.18948519e-01 1.91474244e-01 1.04377165e-01 6.42950386e-02 -1.19102037e+00 5.46915114e-01 -2.29519510e+00 -1.10066628e+00 -8.15574378e-02 2.09852815e+00 6.83767855e-01 6.79569423e-01 1.77610144e-01 3.11965674e-01 1.80461645e-01 -3.16982508e-01 -8.89483392e-01 -1.07991147e+00 1.68999642e-01 1.06697023e-01 5.34140408e-01 7.24460483e-01 -9.29740429e-01 1.21700203e+00 6.64242125e+00 5.75509071e-01 -1.06712186e+00 -2.06649810e-01 4.74192984e-02 4.78098467e-02 -1.21018492e-01 -1.82551846e-01 -8.75656068e-01 3.12489182e-01 1.42711413e+00 2.16198340e-02 7.58508980e-01 1.07048523e+00 7.58454025e-01 -3.12356830e-01 -1.07915711e+00 4.76713955e-01 -4.89443570e-01 -1.04088342e+00 -2.81566322e-01 -5.37279919e-02 4.95897919e-01 3.05658672e-02 -3.85846682e-02 1.00369310e+00 5.88947058e-01 -1.12188208e+00 8.60211551e-01 6.38269126e-01 4.86509234e-01 -1.28641891e+00 7.40943193e-01 6.87217951e-01 -8.83118749e-01 -6.07186139e-01 4.19185916e-03 -4.01513845e-01 8.45018495e-03 1.51036009e-01 -9.84972775e-01 3.91570926e-01 7.18407035e-01 8.72827232e-01 -4.73328024e-01 5.97714365e-01 -3.99251908e-01 6.12201631e-01 -8.35887417e-02 -3.01344097e-01 5.94262004e-01 8.17953721e-02 4.37489659e-01 1.03992808e+00 -1.27403541e-02 -2.09196284e-01 3.49514216e-01 8.92872334e-01 7.46724904e-01 -6.06428742e-01 -1.20685863e+00 1.02330953e-01 4.76697594e-01 7.34447718e-01 -1.64874762e-01 1.84672233e-02 -2.68529385e-01 1.74866751e-01 2.72707194e-01 4.65672374e-01 -9.74725902e-01 -5.67014933e-01 1.06137753e+00 1.79795399e-01 1.97453991e-01 -5.39555073e-01 -1.92887411e-01 -4.21603918e-01 -2.82484591e-01 -1.01609957e+00 -1.03007620e-02 -6.95686102e-01 -6.96556509e-01 5.39263725e-01 5.29821396e-01 -1.25078917e+00 -5.24809897e-01 -6.47285700e-01 -3.64569575e-01 6.01212502e-01 -2.09004688e+00 -9.17033434e-01 -3.95007730e-01 6.11573637e-01 4.75897640e-01 -3.95372063e-01 8.61193061e-01 1.92613468e-01 -6.08355165e-01 4.35463756e-01 -3.79211381e-02 -4.72376376e-01 3.31086934e-01 -1.11541510e+00 3.81892264e-01 8.27603400e-01 -8.16202462e-01 3.53984654e-01 8.95080030e-01 -6.08246624e-01 -1.61081350e+00 -1.23431742e+00 6.32354200e-01 -2.48525813e-01 5.79687655e-01 -4.06612813e-01 -9.90739107e-01 5.47062993e-01 2.08657131e-01 -2.04514995e-01 4.35061216e-01 -2.83267975e-01 2.62892637e-02 -4.35518444e-01 -1.29336739e+00 9.64504063e-01 9.71400082e-01 -3.18938524e-01 -5.45137286e-01 -5.43452539e-02 6.48561239e-01 -3.67310137e-01 -5.45641482e-01 5.95830679e-01 3.75990927e-01 -9.89370167e-01 6.82089806e-01 -3.94649088e-01 7.41047040e-02 -6.80159867e-01 -1.03767544e-01 -1.17146850e+00 -1.81775182e-01 -6.88021600e-01 -3.14569861e-01 8.26498210e-01 2.61577636e-01 -1.00067103e+00 3.45443398e-01 7.79139638e-01 -3.80776882e-01 -8.61500740e-01 -9.90492463e-01 -1.00491750e+00 9.99277383e-02 -8.80614400e-01 7.29709387e-01 5.37398636e-01 -1.57976765e-02 -2.51956761e-01 -3.05404007e-01 2.83601850e-01 5.15212238e-01 -3.01177263e-01 9.94665146e-01 -1.06964278e+00 -1.03052557e-01 -2.93385118e-01 -5.85587382e-01 -4.90740806e-01 7.86300421e-01 -7.91898191e-01 5.03063142e-01 -1.47114766e+00 -5.39955199e-01 -7.38153100e-01 -2.00078860e-01 8.00239444e-01 3.12368631e-01 -2.75296032e-01 -2.05695555e-01 -2.94842601e-01 -5.71621060e-01 5.66872776e-01 1.23312604e+00 -4.97717634e-02 -3.23648125e-01 6.50908574e-02 -4.85363543e-01 5.25421619e-01 1.05763829e+00 -2.29557529e-01 -9.67922747e-01 -1.33569628e-01 2.91984290e-01 -5.18698581e-02 5.03332615e-01 -1.40488505e+00 2.51833230e-01 -7.50645459e-01 -1.58665106e-01 -5.63093722e-01 1.39260307e-01 -1.12704384e+00 1.43645629e-01 8.22200716e-01 -3.95760238e-01 -1.70066401e-01 8.18619788e-01 4.87815261e-01 -1.21600062e-01 7.32396916e-02 7.09835708e-01 1.79787040e-01 -1.38960958e+00 -1.72231093e-01 -9.91398871e-01 -7.05782324e-02 1.73360407e+00 -5.03289282e-01 -1.43791199e-01 -1.59792185e-01 -5.82155645e-01 8.85766447e-01 2.20677778e-01 7.37684429e-01 7.93626487e-01 -1.19958639e+00 -3.67712349e-01 5.87428749e-01 3.28815401e-01 2.16191471e-01 1.45792559e-01 7.03485131e-01 -6.06693506e-01 5.05845368e-01 -5.08252263e-01 -5.85357845e-01 -7.72924900e-01 6.45267129e-01 6.88842893e-01 -1.39440689e-02 -7.92557418e-01 3.07438016e-01 -3.59941095e-01 -8.63836229e-01 5.01615345e-01 -2.46352091e-01 -2.27997094e-01 -3.92979503e-01 4.30200845e-01 5.11357188e-01 1.92564771e-01 -3.96166056e-01 -4.05851692e-01 1.99698940e-01 1.91251375e-02 -9.93381813e-02 1.21208346e+00 -3.51413228e-02 3.98488343e-01 6.16370559e-01 6.39132023e-01 -5.36475778e-01 -1.65274966e+00 3.98422889e-02 2.36297667e-01 -2.62619674e-01 2.35376537e-01 -9.62123692e-01 -5.43747425e-01 7.88627028e-01 7.07674921e-01 5.20639159e-02 8.71894300e-01 -5.06833494e-01 7.21284747e-01 7.81543314e-01 7.70616233e-01 -1.50386298e+00 -7.46513754e-02 7.82690942e-01 7.10379243e-01 -1.24865186e+00 -4.19367075e-01 1.44119384e-02 -8.97980690e-01 1.00049508e+00 1.02402627e+00 -6.22792542e-02 5.97628236e-01 4.69336480e-01 2.98430532e-01 2.51792762e-02 -9.28589404e-01 -3.07730049e-01 -1.16579935e-01 9.85544622e-01 -1.63837761e-01 1.40831888e-01 1.43214106e-03 -3.91305685e-02 -1.21320747e-01 3.19892228e-01 5.26188195e-01 1.43503022e+00 -4.68365818e-01 -1.12051201e+00 -1.18655130e-01 1.44071192e-01 2.36965775e-01 5.07083952e-01 -1.03114523e-01 1.09092546e+00 5.43910742e-01 1.18773544e+00 -1.84130594e-01 -5.82004726e-01 7.09608555e-01 5.53628616e-02 2.44853988e-01 -3.72619927e-01 -3.53667259e-01 -5.30037940e-01 1.06596410e-01 -9.55585361e-01 -2.76787311e-01 -7.67991066e-01 -1.67630029e+00 -3.63127559e-01 1.41531751e-01 2.14144699e-02 6.66816354e-01 1.02557147e+00 5.69891930e-01 8.69380891e-01 6.74015820e-01 -9.23156500e-01 -6.44776344e-01 -1.99313253e-01 -3.63120735e-01 -2.85338219e-02 5.55555046e-01 -1.01105857e+00 -3.37967902e-01 -3.10081005e-01]
[5.244971752166748, 1.2481963634490967]
3d0d5eda-e03b-4d13-9f97-9c0c4d995783
st-detr-spatio-temporal-object-traces
2107.05887
null
https://arxiv.org/abs/2107.05887v2
https://arxiv.org/pdf/2107.05887v2.pdf
ST-DETR: Spatio-Temporal Object Traces Attention Detection Transformer
We propose ST-DETR, a Spatio-Temporal Transformer-based architecture for object detection from a sequence of temporal frames. We treat the temporal frames as sequences in both space and time and employ the full attention mechanisms to take advantage of the features correlations over both dimensions. This treatment enables us to deal with frames sequence as temporal object features traces over every location in the space. We explore two possible approaches; the early spatial features aggregation over the temporal dimension, and the late temporal aggregation of object query spatial features. Moreover, we propose a novel Temporal Positional Embedding technique to encode the time sequence information. To evaluate our approach, we choose the Moving Object Detection (MOD)task, since it is a perfect candidate to showcase the importance of the temporal dimension. Results show a significant 5% mAP improvement on the KITTI MOD dataset over the 1-step spatial baseline.
['Ahmad El-Sallab', 'Eslam Mohamed']
2021-07-13
null
null
null
null
['moving-object-detection']
['computer-vision']
[ 6.00896962e-02 -4.45968598e-01 -8.52421746e-02 -2.87461549e-01 -8.90125692e-01 -6.03641570e-01 1.00594413e+00 -6.97944639e-03 -8.22543263e-01 4.32728469e-01 4.61718678e-01 1.54745197e-02 -2.22341269e-01 -6.11169338e-01 -7.06694484e-01 -7.81707168e-01 -6.32671297e-01 7.10339621e-02 8.01712215e-01 6.67956248e-02 5.64548224e-02 7.45930195e-01 -1.57967293e+00 6.24760449e-01 2.49088466e-01 1.21305466e+00 3.73049766e-01 8.89076591e-01 1.05642611e-02 9.94597614e-01 -3.16540033e-01 -1.35995477e-01 3.10916960e-01 -2.51082778e-01 -7.56665468e-01 -9.77728330e-03 4.57163751e-01 -4.65159655e-01 -6.89431727e-01 6.85534537e-01 3.83014411e-01 3.99634004e-01 3.53038162e-01 -1.17706966e+00 -3.26147467e-01 4.20908391e-01 -6.56796336e-01 8.19304347e-01 3.16310912e-01 3.76220971e-01 1.13344073e+00 -1.06712198e+00 9.26497519e-01 1.35250366e+00 6.15367293e-01 1.71524107e-01 -1.16063070e+00 -1.69672981e-01 2.05878049e-01 7.16581821e-01 -1.32213759e+00 -3.27297330e-01 6.22029960e-01 -6.73939645e-01 1.26904190e+00 1.79899827e-01 6.48627639e-01 1.01417911e+00 1.53652579e-01 1.00000632e+00 8.55804801e-01 -2.67158061e-01 9.70695764e-02 -2.39597961e-01 1.29209951e-01 3.44647199e-01 -4.52197284e-01 4.80751663e-01 -6.71935916e-01 8.03504884e-03 5.98789632e-01 1.22402348e-01 -1.00872502e-01 -4.01851296e-01 -1.58794498e+00 6.15775585e-01 7.43744075e-01 5.18126607e-01 -5.19287884e-01 5.32062054e-01 5.05299389e-01 2.03434736e-01 5.31809688e-01 2.58962929e-01 -4.06032175e-01 -3.94592732e-01 -1.08141077e+00 3.61756325e-01 1.69362187e-01 8.09463680e-01 5.94034135e-01 -1.38290361e-01 -5.77911675e-01 3.86476278e-01 5.97316250e-02 3.28122586e-01 2.90889382e-01 -1.08007920e+00 6.15457535e-01 1.56891823e-01 3.07188481e-01 -9.40398574e-01 -3.04608881e-01 -1.51238874e-01 -3.11790347e-01 1.42124638e-01 5.10914087e-01 1.92106992e-01 -8.22610676e-01 1.75622416e+00 4.01204348e-01 5.75599611e-01 -1.91761628e-01 1.04894280e+00 4.13029552e-01 8.43483567e-01 1.54348373e-01 -1.30884409e-01 1.54032218e+00 -7.73913860e-01 -4.79384154e-01 5.53669110e-02 6.80004179e-01 -4.90742594e-01 8.88803840e-01 -4.06412687e-03 -1.25122702e+00 -7.93410540e-01 -8.75791430e-01 -3.57047647e-01 -6.43337488e-01 1.24015830e-01 4.27672386e-01 8.50387216e-02 -1.00288010e+00 8.66217792e-01 -1.31612897e+00 -3.59005004e-01 2.37787887e-01 2.98893124e-01 -4.43773150e-01 -5.08471802e-02 -1.09766209e+00 6.89104795e-01 4.47863013e-01 3.41145396e-02 -7.90456653e-01 -6.52945697e-01 -7.54171073e-01 1.60071641e-01 2.65046686e-01 -3.72919589e-01 1.06823432e+00 -6.44744873e-01 -7.34768867e-01 7.55376637e-01 -6.25964701e-01 -7.49689400e-01 5.82672179e-01 -3.51414621e-01 -2.70742029e-01 5.40136337e-01 2.02881828e-01 6.69309974e-01 8.19157422e-01 -7.03309596e-01 -9.08843756e-01 -3.49485278e-01 2.23097235e-01 9.94483903e-02 -1.51270688e-01 1.37140661e-01 -5.69198966e-01 -7.84582376e-01 5.98422177e-02 -9.48871017e-01 -3.39274034e-02 1.96129322e-01 -1.52779266e-01 -1.49211690e-01 9.55349267e-01 -6.64324760e-01 1.10221684e+00 -2.51031828e+00 2.60125697e-01 -1.33022830e-01 8.64336044e-02 5.57204708e-02 -1.88754305e-01 3.87101501e-01 -2.52626956e-01 -3.60823870e-02 6.91578910e-03 -5.17731309e-01 -3.29900882e-03 1.54579118e-01 -5.62903821e-01 3.88512522e-01 5.48568606e-01 1.25109255e+00 -1.01910651e+00 -4.72445786e-01 2.71805584e-01 6.27173781e-01 -4.82664317e-01 -6.14191219e-03 -2.83985257e-01 4.02787745e-01 -5.68520129e-01 2.57417887e-01 4.16228473e-01 -2.09083647e-01 -1.69564113e-01 -4.50475067e-01 -4.28248078e-01 4.83939856e-01 -9.36447859e-01 1.68646693e+00 -2.06780449e-01 8.45433116e-01 -3.97052139e-01 -5.10855615e-01 2.67636716e-01 3.60571623e-01 7.83219337e-01 -1.06442094e+00 -1.28630415e-01 -2.99064815e-02 -6.41170964e-02 -6.24795496e-01 7.23748624e-01 3.84607762e-02 8.23807418e-02 4.36719447e-01 1.55029863e-01 5.80629051e-01 2.92954028e-01 3.58572096e-01 1.42536783e+00 3.36333573e-01 2.32202355e-02 -1.80694789e-01 3.17268670e-01 -1.37819629e-02 3.46883565e-01 6.51248991e-01 -2.51254827e-01 6.97225094e-01 5.76506436e-01 -7.60675013e-01 -1.20130706e+00 -1.07871497e+00 1.23728633e-01 1.21769381e+00 4.82352450e-02 -4.63542640e-01 -1.48742080e-01 -7.53479242e-01 4.65529747e-02 4.71023679e-01 -1.02663541e+00 -5.20127686e-03 -1.12303913e+00 -5.35918772e-01 2.45528668e-01 9.53902662e-01 3.69551837e-01 -9.50171292e-01 -1.14312100e+00 3.02290767e-01 -3.53777975e-01 -1.32094729e+00 -7.38028228e-01 3.05293053e-01 -6.94402039e-01 -8.13586891e-01 -6.79849982e-01 -3.48955572e-01 -4.63542603e-02 3.50939125e-01 9.85140383e-01 -2.36744821e-01 -4.89551663e-01 4.14211452e-01 -4.64972794e-01 2.02296767e-02 7.30261877e-02 -7.51371831e-02 -1.14991486e-01 1.94450364e-01 4.75449353e-01 -6.61834955e-01 -7.56162107e-01 2.41296440e-01 -8.09187412e-01 -1.40234411e-01 2.64898598e-01 7.10317254e-01 5.41274250e-01 -2.07256272e-01 9.80280861e-02 -2.49483988e-01 -9.02403742e-02 -4.75255638e-01 -6.54966831e-01 1.22661375e-01 -3.23375985e-02 1.85331061e-01 1.32545099e-01 -4.97643232e-01 -8.23032081e-01 1.12187006e-01 1.71437133e-02 -7.78330386e-01 5.52264750e-02 3.67354751e-01 7.87266120e-02 2.01093480e-01 4.89075661e-01 4.32371020e-01 -2.92536914e-01 -5.47374666e-01 3.91341329e-01 6.65337592e-02 5.18060505e-01 -4.62298334e-01 7.14706659e-01 8.51360381e-01 -8.05475563e-03 -7.62898505e-01 -5.89569986e-01 -5.99446774e-01 -9.03218091e-01 -7.71695673e-02 1.29035139e+00 -8.61152709e-01 -7.30452776e-01 -1.41179478e-02 -1.38098860e+00 -3.64097863e-01 -5.14980137e-01 6.66598022e-01 -6.15823865e-01 2.30712011e-01 -7.35734463e-01 -7.73443997e-01 2.37462237e-01 -1.15881574e+00 1.39864659e+00 -2.64112860e-01 -2.23076373e-01 -8.08811128e-01 7.14243054e-02 -3.59593302e-01 2.47454345e-01 4.82253283e-01 7.46491730e-01 -4.77525204e-01 -1.16399109e+00 -6.81449324e-02 -4.81254041e-01 -1.27754405e-01 -7.24473521e-02 -5.74756451e-02 -1.16131115e+00 -3.34224522e-01 2.95873173e-02 -7.03539997e-02 1.28290868e+00 4.37113792e-01 1.00290644e+00 -1.98218703e-01 -5.26754856e-01 6.08855963e-01 1.39033246e+00 3.13704580e-01 5.75087786e-01 4.50269163e-01 7.18746126e-01 5.83366394e-01 7.44226098e-01 4.15907681e-01 2.80564457e-01 1.17231262e+00 3.95869851e-01 1.46878781e-02 -3.43966573e-01 -1.44119546e-01 3.60100061e-01 4.14277226e-01 -1.09066054e-01 -2.56153524e-01 -9.38333273e-01 9.33615863e-01 -1.88271201e+00 -1.44295156e+00 1.26556251e-02 2.17003369e+00 3.02553654e-01 1.62878737e-01 3.81576717e-01 1.95566162e-01 4.50768232e-01 4.50079024e-01 -3.33044678e-01 3.66920792e-02 -2.53275514e-01 -3.77878100e-02 5.52428782e-01 4.82886225e-01 -1.30850315e+00 7.75438964e-01 6.57358408e+00 7.39738464e-01 -1.22685742e+00 3.15755069e-01 3.55965495e-01 -4.88173902e-01 -2.94572245e-02 -1.70021534e-01 -7.04063237e-01 5.55210412e-01 9.34287310e-01 1.63935330e-02 1.13919832e-01 5.72037339e-01 2.17025697e-01 -8.44144598e-02 -1.46850324e+00 7.57698298e-01 -2.95940012e-01 -1.37573874e+00 -1.15041643e-01 2.13439256e-01 3.58853787e-01 4.16030675e-01 2.00208068e-01 2.23708048e-01 6.88625872e-02 -8.45372498e-01 1.15499389e+00 4.15561378e-01 5.71501613e-01 -4.71713543e-01 3.96913737e-01 2.11110219e-01 -1.61861575e+00 -2.36603886e-01 -1.44244269e-01 -5.60875330e-03 4.75682527e-01 3.47034842e-01 -6.33450210e-01 5.07313430e-01 9.29613769e-01 1.02529740e+00 -7.58253932e-01 1.09675026e+00 2.09609151e-01 5.08823633e-01 -6.14766657e-01 3.55669141e-01 5.97552001e-01 6.51500225e-02 7.42969215e-01 1.33819628e+00 4.54718322e-01 2.44481444e-01 9.33801085e-02 7.09743261e-01 4.16352630e-01 -4.06931072e-01 -7.14756846e-01 6.06282032e-04 1.98904678e-01 8.47445428e-01 -7.54012465e-01 -4.70834702e-01 -5.46684504e-01 1.06384265e+00 2.17794567e-01 5.56782484e-01 -1.10992205e+00 -1.50023028e-01 8.26276958e-01 1.27455756e-01 9.66478527e-01 -5.42195201e-01 1.40488995e-02 -1.05433810e+00 4.59855080e-01 -2.58395225e-01 5.33871710e-01 -8.00334275e-01 -9.20893550e-01 6.86727166e-01 2.26271927e-01 -1.49597168e+00 -4.61793751e-01 -5.56483805e-01 -3.67697805e-01 8.65853667e-01 -1.45128620e+00 -1.08994937e+00 -4.59421091e-02 6.73015058e-01 6.23703301e-01 2.51751602e-01 3.71711820e-01 5.04224360e-01 -1.44800752e-01 3.53289366e-01 -1.63265746e-02 1.03283115e-01 4.43778247e-01 -1.17718089e+00 6.60525501e-01 9.46552157e-01 4.08666909e-01 4.90995556e-01 6.58573091e-01 -4.09490377e-01 -1.32734334e+00 -1.12772048e+00 1.16954625e+00 -8.70354295e-01 7.33882546e-01 -5.25613785e-01 -1.07930446e+00 8.67523551e-01 1.03367612e-01 4.11326528e-01 2.65466720e-01 -1.35886595e-01 -5.07677794e-01 -1.21522276e-02 -7.94216454e-01 5.33293247e-01 1.16900396e+00 -8.40404749e-01 -6.27117395e-01 1.01520002e-01 1.05287886e+00 -2.52551675e-01 -6.31245732e-01 4.80356187e-01 7.12801516e-01 -1.05816162e+00 1.22492874e+00 -4.95845765e-01 5.12958109e-01 -5.65906942e-01 -4.31257576e-01 -8.46786141e-01 -6.24811828e-01 -4.34013337e-01 -1.54993415e-01 9.61506307e-01 1.70743302e-01 -3.05245310e-01 7.25161612e-01 4.19241548e-01 9.00769159e-02 -5.95698059e-01 -1.25973761e+00 -1.00828135e+00 -1.63776249e-01 -7.29563594e-01 5.84263325e-01 7.82960415e-01 -1.48786739e-01 2.13609844e-01 -4.12340015e-01 2.34485671e-01 3.94482940e-01 3.01672816e-01 4.14413095e-01 -8.81833911e-01 -5.63791573e-01 -3.87590468e-01 -7.50825703e-01 -1.23330891e+00 -1.08855225e-01 -6.51268423e-01 6.68371022e-02 -1.13526464e+00 1.05705634e-01 -8.46737549e-02 -5.30984402e-01 3.44792455e-01 -6.79365769e-02 3.51629287e-01 4.14909452e-01 2.63050050e-01 -6.74768746e-01 6.15428746e-01 9.05997455e-01 -1.76969379e-01 6.45560399e-02 -3.31994146e-01 1.35451257e-01 3.53413999e-01 3.80066186e-01 -4.65267658e-01 -3.75821263e-01 -7.72191465e-01 -8.66727307e-02 3.90878379e-01 8.46411169e-01 -1.15582287e+00 1.80910721e-01 -8.99460092e-02 4.30220693e-01 -9.81342256e-01 9.09445226e-01 -9.22330856e-01 9.85197276e-02 3.18116605e-01 -4.41706508e-01 3.88317406e-01 3.32621574e-01 6.52198613e-01 -3.18235517e-01 3.67610455e-01 6.33835256e-01 2.55833790e-02 -1.03043294e+00 3.62230092e-01 -2.71371812e-01 -2.36550257e-01 1.04713082e+00 -3.88411522e-01 -1.84786871e-01 -1.43473715e-01 -1.10370064e+00 2.08670035e-01 3.83345515e-01 5.91484666e-01 4.80521858e-01 -1.57174397e+00 -4.70864207e-01 1.28233895e-01 1.75511509e-01 -5.23591101e-01 3.11651230e-01 1.13896394e+00 -1.92963004e-01 6.07671320e-01 -2.98514396e-01 -9.58422065e-01 -1.17317319e+00 1.07526195e+00 1.56771183e-01 -2.91371137e-01 -8.67131889e-01 9.38197196e-01 5.33304930e-01 3.01415354e-01 1.65239692e-01 -5.14201522e-01 -8.59962553e-02 2.50479698e-01 8.23007762e-01 3.84439498e-01 -5.79145318e-03 -8.80901635e-01 -5.06864071e-01 7.41525769e-01 1.06419839e-01 -6.00722492e-01 1.20312393e+00 -3.16273928e-01 1.46490782e-01 8.15000236e-01 1.63951743e+00 -2.49325261e-01 -1.55926478e+00 -5.15829444e-01 3.91413689e-01 -7.54280865e-01 -1.85273334e-01 -5.05271137e-01 -7.86335826e-01 1.11608136e+00 7.90170193e-01 3.03356349e-01 1.03523445e+00 7.52003193e-02 6.31437957e-01 4.60146442e-02 3.88334334e-01 -6.48088932e-01 1.24405146e-01 5.93174398e-01 8.83070230e-01 -1.20154107e+00 -1.80167317e-01 -2.67250270e-01 -4.66856688e-01 9.23552692e-01 2.52373457e-01 -1.81461483e-01 5.51820397e-01 2.30494797e-01 -2.07546502e-01 -2.51240790e-01 -1.08863974e+00 -6.21201813e-01 4.17061985e-01 3.68795335e-01 2.86551714e-01 -1.83506206e-01 -3.00270505e-02 6.02420606e-02 1.46120459e-01 -6.62932917e-02 4.77857981e-03 9.52463448e-01 -3.26189429e-01 -9.72565055e-01 -2.10083768e-01 1.35748178e-01 -4.19617057e-01 7.27026686e-02 -1.30405158e-01 9.89386916e-01 2.17987329e-01 5.36915123e-01 2.87304223e-01 -4.06501502e-01 3.87413383e-01 2.70017594e-01 4.66173947e-01 -2.81906009e-01 -3.21989119e-01 3.18412781e-01 -1.16902441e-01 -1.14213634e+00 -4.82101649e-01 -9.24603760e-01 -9.57357407e-01 -1.98162459e-02 1.16712362e-01 -3.74966115e-02 4.01633918e-01 8.40394199e-01 3.88105720e-01 5.98294854e-01 3.87440830e-01 -1.29901373e+00 -2.23037332e-01 -7.82612264e-01 -3.72634441e-01 4.55383748e-01 9.48769152e-01 -7.01243281e-01 -1.25928983e-01 2.61700928e-01]
[8.695175170898438, 0.46207454800605774]
023cf483-5f6b-4725-9302-46ef16a7f917
a-probabilistic-translation-method-for
1411.1006
null
http://arxiv.org/abs/1411.1006v2
http://arxiv.org/pdf/1411.1006v2.pdf
A Probabilistic Translation Method for Dictionary-based Cross-lingual Information Retrieval in Agglutinative Languages
Translation ambiguity, out of vocabulary words and missing some translations in bilingual dictionaries make dictionary-based Cross-language Information Retrieval (CLIR) a challenging task. Moreover, in agglutinative languages which do not have reliable stemmers, missing various lexical formations in bilingual dictionaries degrades CLIR performance. This paper aims to introduce a probabilistic translation model to solve the ambiguity problem, and also to provide most likely formations of a dictionary candidate. We propose Minimum Edit Support Candidates (MESC) method that exploits a monolingual corpus and a bilingual dictionary to translate users' native language queries to documents' language. Our experiments show that the proposed method outperforms state-of-the-art dictionary-based English-Persian CLIR.
['Azadeh Shakery', 'Javid Dadashkarimi', 'Heshaam Faili']
2014-11-04
null
null
null
null
['cross-lingual-information-retrieval']
['natural-language-processing']
[-3.63131374e-01 -4.94735271e-01 -6.84336603e-01 -1.96372449e-01 -9.86342549e-01 -1.11965692e+00 7.10396826e-01 5.21649957e-01 -7.82294333e-01 1.05331063e+00 3.53841811e-01 -7.71117568e-01 -4.46577482e-02 -7.31067717e-01 -3.78338307e-01 -1.95525557e-01 5.88210642e-01 1.12135530e+00 2.59285986e-01 -9.14114118e-01 3.46923053e-01 3.42956275e-01 -1.01494730e+00 2.29318604e-01 1.26561701e+00 1.04776442e-01 7.49947548e-01 -4.21757102e-02 -5.68627894e-01 1.19573221e-01 -3.80218297e-01 -7.94335723e-01 2.59759694e-01 -2.64478087e-01 -6.71897531e-01 -5.84592700e-01 1.53361097e-01 1.15693480e-01 -6.29304498e-02 1.39450192e+00 5.71505606e-01 -8.24018642e-02 7.94386506e-01 -6.12079680e-01 -1.27569914e+00 9.31346893e-01 -3.79761755e-01 4.06581730e-01 5.56967139e-01 -7.30930388e-01 1.27190733e+00 -1.40076399e+00 9.36324954e-01 1.30620933e+00 2.15188533e-01 5.37814677e-01 -1.01682770e+00 -6.11933649e-01 5.74903488e-02 1.67513192e-01 -1.99499512e+00 -3.22204888e-01 5.36021173e-01 -2.10767552e-01 1.14672852e+00 3.02968621e-01 2.66215324e-01 9.11018610e-01 4.52864826e-01 5.09383023e-01 1.32588339e+00 -9.60638702e-01 -1.57886967e-01 5.17644644e-01 3.02454025e-01 5.02413511e-01 5.35426021e-01 7.93940425e-02 -5.80277801e-01 -2.39342555e-01 7.17194676e-01 -1.93467885e-01 -1.53203858e-02 9.48344246e-02 -1.41220391e+00 8.55997980e-01 -2.29424059e-01 6.86782837e-01 -3.57085168e-01 -7.89942622e-01 3.08878899e-01 6.57159746e-01 1.94069475e-01 5.39658487e-01 -6.68000460e-01 2.18517989e-01 -5.61771750e-01 1.07562020e-01 8.29040885e-01 1.28189027e+00 7.18243897e-01 -3.44486237e-01 1.81146117e-03 1.22355807e+00 4.82450247e-01 1.17193925e+00 8.30835938e-01 -1.48571089e-01 3.78021032e-01 4.95350212e-01 7.19548911e-02 -9.01342511e-01 -4.54956405e-02 -4.25545186e-01 -4.03408259e-01 -6.39072239e-01 6.94625899e-02 8.58920962e-02 -3.97112012e-01 1.42224658e+00 2.20563546e-01 -6.03186011e-01 4.04987782e-01 9.89570856e-01 7.74074912e-01 8.19828093e-01 -3.44303027e-02 -6.74255192e-01 1.70277369e+00 -8.85101497e-01 -8.19220722e-01 -4.02332574e-01 4.29261059e-01 -1.50550830e+00 1.32310081e+00 3.33210111e-01 -8.55918646e-01 -6.04890525e-01 -9.60266232e-01 -2.20152527e-01 -5.40341496e-01 1.66969895e-01 3.33711296e-01 6.24990821e-01 -1.00189519e+00 -1.26484511e-02 -4.74633455e-01 -4.97523665e-01 -6.50998592e-01 1.89146101e-01 -2.90576607e-01 -3.58795732e-01 -1.66079545e+00 1.27736449e+00 7.16232359e-01 -3.34021389e-01 -6.82654679e-01 -1.94901809e-01 -7.68352807e-01 -3.67149204e-01 1.26987860e-01 -4.19901431e-01 7.94030309e-01 -6.63181543e-01 -9.92272615e-01 1.06175792e+00 -4.18581158e-01 -5.09224199e-02 -3.60312797e-02 -1.31506339e-01 -8.40553463e-01 -4.59178537e-01 5.52594721e-01 5.81782579e-01 3.89526367e-01 -1.08362424e+00 -7.96970725e-01 -4.32059050e-01 -3.04774404e-01 7.13437974e-01 -1.96942493e-01 6.18331373e-01 -8.99178505e-01 -1.12262988e+00 1.17337145e-01 -8.29653800e-01 -1.29739061e-01 -6.36339962e-01 5.46775423e-02 -6.42440677e-01 2.09621564e-01 -1.06406343e+00 1.68542767e+00 -1.82131159e+00 2.64394522e-01 2.74337620e-01 -4.86499459e-01 2.03793332e-01 -2.92004168e-01 8.23297262e-01 3.68803203e-01 5.29080294e-02 1.58434853e-01 -8.02036822e-02 7.50693120e-03 8.50474000e-01 -5.02908051e-01 1.60060272e-01 -2.39939749e-01 8.15273643e-01 -9.87344444e-01 -7.34762788e-01 8.40733647e-02 4.05573279e-01 -2.26321176e-01 -5.86343035e-02 -1.71248823e-01 4.02982205e-01 -3.36816877e-01 9.14158940e-01 4.14561093e-01 4.73682374e-01 5.20576656e-01 6.77357391e-02 -3.97485673e-01 9.19866979e-01 -1.04604220e+00 1.77770507e+00 -7.34192073e-01 6.51901290e-02 -3.08375567e-01 -5.87760866e-01 1.12609947e+00 3.65920007e-01 7.67860711e-02 -1.05805063e+00 1.85754329e-01 8.84240866e-01 3.63776670e-03 -7.68010691e-02 8.05280566e-01 -2.94703037e-01 -2.72989631e-01 4.40619439e-01 1.67152002e-01 2.90000010e-02 4.75989670e-01 -1.11250386e-01 5.67103982e-01 -4.09586653e-02 6.92766011e-01 -8.51690233e-01 8.40899289e-01 3.02713662e-01 6.66710615e-01 6.99609101e-01 1.06041171e-01 1.86895892e-01 -4.66644585e-01 -4.09895748e-01 -9.16460872e-01 -1.18852758e+00 -4.23930436e-01 1.35760963e+00 3.45152318e-01 -6.64493322e-01 -4.25945848e-01 -7.04515100e-01 -2.69554436e-01 9.52470720e-01 9.31746438e-02 2.10024595e-01 -8.49518418e-01 -6.36336267e-01 4.78756130e-01 -5.95388077e-02 1.88141875e-02 -1.19069684e+00 3.02041709e-01 5.31023741e-01 -7.38357425e-01 -9.36848044e-01 -1.03818464e+00 -5.37523478e-02 -6.86621010e-01 -7.99151838e-01 -4.00876701e-01 -1.52581215e+00 5.76269984e-01 4.24829513e-01 1.38582635e+00 5.41002341e-02 1.06171466e-01 6.36843368e-02 -6.98448062e-01 -3.97034645e-01 -7.37567365e-01 2.25394100e-01 5.55791378e-01 -5.33756852e-01 1.22736943e+00 -3.93878430e-01 -1.51100919e-01 4.77352738e-01 -8.81057739e-01 -3.01805109e-01 6.09947026e-01 9.35933053e-01 1.02392459e+00 8.12690053e-03 5.08243740e-01 -7.66187131e-01 1.08783507e+00 -4.49128062e-01 -8.24461639e-01 8.25327039e-01 -1.09795105e+00 4.11714792e-01 5.45048594e-01 -4.94786680e-01 -9.06216025e-01 -2.47130752e-01 -3.74489933e-01 1.33287877e-01 1.25049308e-01 8.52860689e-01 -1.95681125e-01 1.73903465e-01 9.97814596e-01 5.53462565e-01 -4.80898976e-01 -8.53337824e-01 3.87745142e-01 9.61596787e-01 4.16742682e-01 -9.33665335e-01 4.39075142e-01 -1.91260651e-01 -4.95943934e-01 -6.32206976e-01 -5.45989454e-01 -9.63903666e-01 -8.56590688e-01 2.25773394e-01 7.20820785e-01 -1.21791875e+00 -1.17250569e-01 -1.87314570e-01 -1.49933970e+00 3.68699253e-01 5.06242104e-02 7.06552923e-01 3.32032964e-02 4.00842339e-01 -7.64976740e-01 -4.29990530e-01 -7.58376062e-01 -1.11925149e+00 1.00379133e+00 -1.44872621e-01 -2.45729998e-01 -1.05013168e+00 5.63390732e-01 3.27791393e-01 1.82485804e-01 -6.21581554e-01 1.10228801e+00 -9.17875350e-01 -2.84753829e-01 1.07392050e-01 1.39156058e-02 2.27207869e-01 4.68345553e-01 -5.56787431e-01 -2.48794869e-01 -4.10899848e-01 8.70499387e-02 4.23811525e-02 2.65465200e-01 -5.10687009e-02 7.14481696e-02 -5.38944423e-01 -3.26238930e-01 7.33235180e-02 1.54803348e+00 5.10571957e-01 3.29036713e-01 3.68671149e-01 6.18035734e-01 4.82612252e-01 8.47730935e-01 4.87969369e-02 7.07547486e-01 7.85365462e-01 -3.84451240e-01 2.01756015e-01 -1.99617743e-01 -4.69120651e-01 6.50034666e-01 1.97974777e+00 3.28891665e-01 -5.03618121e-02 -9.73732650e-01 8.39389265e-01 -1.61766779e+00 -4.55723643e-01 -5.89843467e-02 2.37571931e+00 1.64423203e+00 -6.04070127e-02 -2.44436070e-01 -3.39751452e-01 6.76704764e-01 -3.12865019e-01 1.76981147e-02 -4.26259011e-01 -5.56249678e-01 6.72229350e-01 6.75423682e-01 1.28971696e+00 -7.64417171e-01 1.65672219e+00 6.12023258e+00 1.21527648e+00 -8.87297690e-01 5.62156379e-01 -1.19444571e-01 5.72435915e-01 -8.30944598e-01 2.93459535e-01 -1.35821331e+00 1.61559522e-01 6.57108366e-01 -4.43295687e-01 8.10311139e-01 7.00741470e-01 -7.98559114e-02 2.38248378e-01 -9.01386619e-01 1.22383416e+00 4.32465464e-01 -8.88066232e-01 6.52111471e-01 -1.57386497e-01 7.06434548e-01 3.59386533e-01 -3.70122731e-01 4.06277329e-01 6.27693355e-01 -7.65634120e-01 5.32912970e-01 1.50912523e-01 9.12623286e-01 -7.46842504e-01 6.29464090e-01 6.38799667e-01 -1.10118413e+00 5.59303403e-01 -8.33591402e-01 1.48115546e-01 4.62722927e-02 3.30868363e-01 -8.82357895e-01 5.60653746e-01 3.15091670e-01 5.34578443e-01 -5.73154926e-01 5.05814791e-01 -5.91657817e-01 4.05335546e-01 -1.81969926e-01 -2.61381179e-01 -2.73497254e-02 -5.92726648e-01 7.49643147e-01 1.39759493e+00 5.04693866e-01 2.33887937e-02 5.46004176e-01 3.65733206e-01 4.41490337e-02 1.02672935e+00 -7.94656396e-01 2.15685256e-02 8.45300078e-01 9.10891116e-01 -5.96898139e-01 -3.00540477e-01 -5.27067661e-01 1.34380293e+00 1.99013531e-01 1.58524647e-01 -2.61653274e-01 -1.48279503e-01 6.21454060e-01 -3.24296057e-02 -1.12339072e-01 -5.58926284e-01 -1.67594984e-01 -1.48279727e+00 1.39888376e-01 -1.39134550e+00 6.20510340e-01 -4.29373950e-01 -1.57175887e+00 9.60442364e-01 -8.14688802e-02 -1.12349653e+00 -2.75090635e-01 -5.34809709e-01 2.53626108e-01 1.21589851e+00 -1.67582679e+00 -1.24833071e+00 6.32661581e-01 9.41898346e-01 8.75160992e-01 -5.60946345e-01 1.21415496e+00 8.44009995e-01 -5.62538505e-02 7.02599585e-01 3.53717744e-01 2.04650342e-01 1.18793952e+00 -1.10269129e+00 2.87348181e-01 1.08059978e+00 8.47718835e-01 1.30697095e+00 7.15728998e-01 -1.07065582e+00 -1.51075113e+00 -9.27325666e-01 2.02602816e+00 -6.35678470e-01 7.36135364e-01 -3.56840134e-01 -6.65906072e-01 6.71108365e-01 6.02667570e-01 -5.07697344e-01 8.55295300e-01 3.40537518e-01 -4.54466492e-01 -1.75414354e-01 -7.63625383e-01 8.23461354e-01 1.02773523e+00 -7.83999324e-01 -1.14294827e+00 4.94667500e-01 7.75200963e-01 -2.92893589e-01 -6.75884187e-01 2.13311359e-01 4.86507237e-01 -2.29096174e-01 9.47960138e-01 -4.24588799e-01 -2.14749023e-01 -6.48001850e-01 -5.45233428e-01 -1.39754510e+00 -3.51969093e-01 -4.42977995e-01 4.23233479e-01 1.12181818e+00 6.41978502e-01 -5.05370200e-01 -8.44703764e-02 7.06496015e-02 -1.13517918e-01 -6.31796196e-02 -1.00758660e+00 -8.73354018e-01 4.18956071e-01 -2.11664796e-01 4.43420917e-01 1.35703599e+00 2.96951920e-01 1.08532894e+00 -4.03246403e-01 2.10843533e-01 4.32386875e-01 6.58391938e-02 2.92711765e-01 -1.30513108e+00 -3.76995683e-01 -3.19227487e-01 8.50497782e-02 -1.24265230e+00 3.13286483e-01 -1.31747735e+00 -9.16882604e-03 -1.35117459e+00 3.09993848e-02 -8.55329990e-01 -3.82252216e-01 5.03197610e-01 -3.19610238e-01 2.36184373e-01 -1.29469469e-01 5.88581443e-01 -4.89529908e-01 3.68230104e-01 9.21704412e-01 -1.98809981e-01 -3.91081214e-01 -3.47322106e-01 -8.85942578e-01 2.65727162e-01 6.93722844e-01 -9.04778183e-01 -4.47888315e-01 -8.32284689e-01 6.31207883e-01 1.79190025e-01 -4.95459795e-01 -4.09459919e-01 3.40992242e-01 -4.58385140e-01 -5.97117171e-02 -5.63725531e-01 -1.37422472e-01 -7.89270401e-01 2.58646190e-01 4.09802675e-01 -6.07148372e-02 9.50941443e-01 1.60544962e-01 4.62543219e-02 -4.29783672e-01 -3.80822420e-01 4.41655397e-01 -3.29642981e-01 -6.81395352e-01 1.74924254e-01 -6.15554810e-01 2.01294765e-01 4.45580870e-01 3.93750966e-01 -3.98678109e-02 1.66609943e-01 -4.06203598e-01 6.37798458e-02 4.73229736e-01 1.04009056e+00 4.89118844e-01 -1.57793653e+00 -1.01324940e+00 4.28679615e-01 5.30667365e-01 -5.25170445e-01 -5.33666432e-01 3.47096235e-01 -6.26910210e-01 7.86100328e-01 -2.29559362e-01 -2.44767949e-01 -1.29658759e+00 7.69540668e-01 -9.14265737e-02 -3.56920332e-01 -4.53134514e-02 6.52839720e-01 -6.46390989e-02 -1.05249500e+00 1.16382957e-01 8.77383724e-02 -3.63390654e-01 -1.75284982e-01 5.23747802e-01 -1.48062095e-01 3.95695120e-01 -1.14431226e+00 -5.13718903e-01 4.54154879e-01 -3.26436609e-01 -6.91228151e-01 6.37552619e-01 -6.19890988e-01 -6.80329263e-01 4.07828152e-01 6.96301699e-01 8.51056218e-01 2.32933640e-01 -8.64277065e-01 4.65396285e-01 -4.45810318e-01 -1.17130227e-01 -1.04236066e+00 -4.30441588e-01 4.77512807e-01 4.87117082e-01 -4.54264998e-01 9.43766952e-01 -6.61972836e-02 9.57609832e-01 7.47543871e-01 1.00670981e+00 -1.24409330e+00 -5.36652505e-01 1.04359782e+00 8.86765361e-01 -1.33211708e+00 -3.65296841e-01 -3.17056835e-01 -5.81213117e-01 9.73865092e-01 3.81973892e-01 2.07190678e-01 9.77305055e-01 5.18334247e-02 5.92690468e-01 -6.16762042e-02 -5.21890104e-01 -2.51244962e-01 5.47504306e-01 4.54173028e-01 7.62351513e-01 3.30507785e-01 -1.49968874e+00 7.09649086e-01 -4.47705209e-01 -4.31141526e-01 3.68989143e-03 8.81005228e-01 -4.18029249e-01 -2.23329806e+00 -4.14492995e-01 2.63353139e-02 -6.38007224e-01 -9.27146852e-01 -5.16901135e-01 3.98709655e-01 3.27652037e-01 1.08916080e+00 -2.27660462e-01 -2.23370746e-01 1.85808152e-01 2.13935450e-01 5.23843169e-01 -8.11762333e-01 -5.50012112e-01 5.07342160e-01 5.49624637e-02 -4.88745496e-02 -2.89985836e-01 -5.71326613e-01 -1.03482771e+00 -3.59197170e-01 -5.03823698e-01 7.45452046e-01 5.99910021e-01 9.96317804e-01 2.28067905e-01 -3.34571630e-01 5.04089355e-01 1.35481030e-01 -3.65590602e-01 -1.02926457e+00 -5.42918444e-01 1.17667653e-01 -1.41917437e-01 -4.19770718e-01 1.89403996e-01 3.04892082e-02]
[11.101367950439453, 9.998368263244629]
7508d7fe-292d-47d6-bb53-529dbdb0471f
a-hierarchical-structured-self-attentive
1805.07799
null
http://arxiv.org/abs/1805.07799v1
http://arxiv.org/pdf/1805.07799v1.pdf
A Hierarchical Structured Self-Attentive Model for Extractive Document Summarization (HSSAS)
The recent advance in neural network architecture and training algorithms have shown the effectiveness of representation learning. The neural network-based models generate better representation than the traditional ones. They have the ability to automatically learn the distributed representation for sentences and documents. To this end, we proposed a novel model that addresses several issues that are not adequately modeled by the previously proposed models, such as the memory problem and incorporating the knowledge of document structure. Our model uses a hierarchical structured self-attention mechanism to create the sentence and document embeddings. This architecture mirrors the hierarchical structure of the document and in turn enables us to obtain better feature representation. The attention mechanism provides extra source of information to guide the summary extraction. The new model treated the summarization task as a classification problem in which the model computes the respective probabilities of sentence-summary membership. The model predictions are broken up by several features such as information content, salience, novelty and positional representation. The proposed model was evaluated on two well-known datasets, the CNN / Daily Mail, and DUC 2002. The experimental results show that our model outperforms the current extractive state-of-the-art by a considerable margin.
['Kamal Al-Sabahi', 'Mohammed Nadher', 'Zhang Zuping']
2018-05-20
null
null
null
null
['extractive-document-summarization']
['natural-language-processing']
[ 1.00105830e-01 2.10802406e-01 -2.06429780e-01 -4.31398183e-01 -5.84262848e-01 -2.00220704e-01 8.68941307e-01 5.66647828e-01 -4.53906208e-01 7.95164704e-01 1.08491898e+00 1.19826376e-01 -6.09722659e-02 -6.93161666e-01 -5.22426069e-01 -5.56636572e-01 -5.96634969e-02 1.93576992e-01 2.13024184e-01 -2.74983048e-01 1.05408382e+00 1.89651340e-01 -1.72082329e+00 6.81878805e-01 9.60852802e-01 8.96111488e-01 5.80931485e-01 8.61696064e-01 -4.34480876e-01 1.17812622e+00 -9.50951517e-01 -8.39535072e-02 -3.04067701e-01 -5.25094211e-01 -8.78096819e-01 2.09148657e-02 4.22768980e-01 -4.86942738e-01 -5.15414178e-01 8.73116314e-01 6.69467270e-01 5.10595441e-01 9.38355684e-01 -6.33926034e-01 -1.37327743e+00 9.52840507e-01 -3.70421886e-01 5.90848565e-01 3.04515570e-01 -4.55849439e-01 1.18522322e+00 -1.02962518e+00 6.17444634e-01 1.26329231e+00 4.95056689e-01 5.02140641e-01 -6.42957926e-01 -1.70192167e-01 2.95465708e-01 5.02064466e-01 -9.54619884e-01 -3.89617443e-01 9.51907396e-01 -3.53504568e-01 1.40949142e+00 3.48714530e-01 4.60656732e-01 9.46440637e-01 4.18741584e-01 1.04413331e+00 5.94640434e-01 -6.60260618e-01 1.05006911e-01 3.91395152e-01 7.35047579e-01 5.54113746e-01 2.91135669e-01 -3.96046430e-01 -5.65654397e-01 -2.38018818e-02 4.64551866e-01 8.32517743e-02 -4.06928331e-01 7.48730600e-02 -1.00128961e+00 1.06175768e+00 5.76773107e-01 6.88515425e-01 -5.25709808e-01 -3.90910804e-02 7.58390009e-01 8.68273303e-02 6.14412010e-01 5.38127959e-01 -4.15300369e-01 7.18647465e-02 -1.10465348e+00 2.07468346e-01 6.82859361e-01 7.67326295e-01 3.49502534e-01 2.37757206e-01 -6.52314961e-01 8.45733881e-01 2.91124821e-01 -1.21709652e-01 1.32214308e+00 -5.21748424e-01 7.48759568e-01 7.71259665e-01 -3.03257331e-02 -1.40044272e+00 -4.26760584e-01 -6.59169137e-01 -1.09668839e+00 -3.77007991e-01 -2.54443437e-01 -4.27131802e-02 -6.58916235e-01 1.48220789e+00 -2.20930547e-01 -3.47566232e-02 3.48533005e-01 5.80742896e-01 1.30660570e+00 1.14143407e+00 -1.82083189e-01 -2.87590116e-01 1.18498182e+00 -1.35172570e+00 -1.20645726e+00 -1.91860706e-01 5.93179047e-01 -6.51923060e-01 7.02916145e-01 1.50468737e-01 -1.27785861e+00 -9.49996889e-01 -1.31186104e+00 -3.11102301e-01 -5.49524903e-01 6.04832768e-01 5.02477825e-01 1.85543522e-01 -1.17382073e+00 8.58797550e-01 -3.85385156e-01 -6.57398701e-01 2.68864065e-01 1.96092412e-01 -1.92886487e-01 3.15845847e-01 -1.22125602e+00 1.08625543e+00 9.78072941e-01 1.59512058e-01 -5.57084501e-01 -9.61457565e-02 -8.54790211e-01 6.51790202e-01 5.42146713e-02 -8.83754194e-01 1.24977410e+00 -1.02132618e+00 -1.43539822e+00 5.48898101e-01 -1.97231799e-01 -7.08504021e-01 -2.09067598e-01 -3.96572798e-01 -2.24003509e-01 2.61378914e-01 -4.78477851e-02 5.36771178e-01 6.70561671e-01 -1.15128982e+00 -5.74113131e-01 -3.51519167e-01 -8.05289000e-02 4.63504553e-01 -8.38948727e-01 4.81913909e-02 -6.76686736e-03 -8.16752136e-01 -1.25612006e-01 -3.10759366e-01 -7.83487707e-02 -6.43478155e-01 -4.21385139e-01 -5.12741923e-01 7.08953083e-01 -8.74332070e-01 1.80571616e+00 -1.93464458e+00 4.27439928e-01 -4.86675113e-01 5.92560209e-02 3.64575207e-01 -1.44926652e-01 9.73639786e-01 -4.47030254e-02 1.25243470e-01 -8.96513388e-02 -4.43012923e-01 -6.13265075e-02 -2.24213488e-02 -4.01847601e-01 4.63786162e-02 1.98934242e-01 7.79314518e-01 -6.74067557e-01 -6.37013495e-01 -1.05177417e-01 3.40429008e-01 -5.41406989e-01 5.12047708e-01 -1.14283077e-01 -1.77128211e-01 -4.34792191e-01 8.45945105e-02 3.56205136e-01 -1.51538223e-01 7.69271106e-02 -2.27971807e-01 -2.21214250e-01 5.14176607e-01 -9.19419825e-01 1.74310315e+00 -6.58012107e-02 8.02711844e-01 -2.10636035e-01 -1.30410564e+00 1.31431866e+00 3.32709074e-01 -5.78171201e-02 -3.86819988e-01 2.86330670e-01 -9.29925069e-02 7.95127377e-02 -7.96190321e-01 1.25056016e+00 1.35351077e-01 -8.60609561e-02 4.79798794e-01 4.65591699e-01 2.24299207e-01 2.56453186e-01 4.92751807e-01 8.67194772e-01 -8.69222283e-02 6.17136717e-01 -3.68308514e-01 7.32168734e-01 -2.04355672e-01 4.23107326e-01 7.36462235e-01 5.95221817e-02 6.21387720e-01 5.16181827e-01 -2.88221121e-01 -8.96352291e-01 -5.43781519e-01 1.43195212e-01 1.19354248e+00 -1.82330489e-01 -5.34444451e-01 -8.53401482e-01 -6.00321472e-01 -2.20145091e-01 1.02562022e+00 -8.72635305e-01 -4.69025701e-01 -5.58689177e-01 -5.45228779e-01 2.14384139e-01 9.52198088e-01 5.38865387e-01 -1.44280076e+00 -4.45353121e-01 2.84185827e-01 -2.79767185e-01 -6.65985465e-01 -5.34932435e-01 2.49703646e-01 -1.12805128e+00 -7.02497900e-01 -6.15806341e-01 -1.21863008e+00 6.39662564e-01 2.36117646e-01 9.25523698e-01 1.52563184e-01 -1.84370086e-01 2.17198893e-01 -5.40573001e-01 -5.40624619e-01 -2.83241510e-01 3.64898622e-01 -1.65824279e-01 4.03753109e-02 3.12894642e-01 -4.82114375e-01 -6.47160590e-01 -4.51728582e-01 -1.02722991e+00 -6.20606132e-02 8.28748345e-01 9.95619059e-01 1.31759301e-01 5.26520312e-02 1.21479213e+00 -8.52688074e-01 1.46937346e+00 -6.07585788e-01 1.40981361e-01 2.34164491e-01 -4.40383703e-01 2.30758518e-01 7.41815984e-01 -1.70107931e-01 -1.45296705e+00 -3.18226486e-01 -8.82886574e-02 1.58685997e-01 -1.57711446e-01 9.10352767e-01 4.36247401e-02 6.19569480e-01 4.94194716e-01 7.18492031e-01 -1.21733874e-01 -7.39789844e-01 3.52289498e-01 1.12331164e+00 4.16807055e-01 -2.31980726e-01 1.98591456e-01 1.68059189e-02 -4.25958484e-01 -9.63734865e-01 -1.06513703e+00 -4.90047604e-01 -8.96189451e-01 4.63824347e-03 8.36977720e-01 -6.15207314e-01 -3.16117138e-01 1.90207645e-01 -1.71423006e+00 3.85065943e-01 -3.52424622e-01 4.52850997e-01 -4.63244528e-01 7.48209655e-01 -6.82302117e-01 -8.76606703e-01 -8.75114560e-01 -7.02340543e-01 8.08432817e-01 3.66626292e-01 -3.44237477e-01 -1.11854696e+00 2.66632617e-01 1.29489169e-01 6.69307947e-01 4.00647484e-02 1.22428918e+00 -1.20655298e+00 -2.59152859e-01 -2.05680609e-01 -1.94715425e-01 6.68650329e-01 -1.80818792e-02 1.03536900e-02 -9.25462425e-01 -1.52646035e-01 3.35790008e-01 -3.54008108e-01 1.44192147e+00 4.46925014e-01 1.24588656e+00 -5.50150275e-01 -1.16491884e-01 2.44593993e-02 1.33813214e+00 2.72270292e-01 7.37055421e-01 4.36986744e-01 2.62295157e-01 7.04487383e-01 5.28275371e-01 5.81692100e-01 4.24220502e-01 2.72520244e-01 4.54956591e-01 2.64277339e-01 -6.59785420e-02 -3.20082217e-01 3.79432917e-01 1.57007408e+00 -6.97104186e-02 -5.46127319e-01 -4.79716063e-01 6.99201703e-01 -2.30880904e+00 -1.34089983e+00 2.62634642e-02 1.79155731e+00 7.47451425e-01 2.65608668e-01 -1.38339937e-01 7.49359429e-02 6.86449409e-01 5.20105302e-01 -1.36675268e-01 -9.48204577e-01 -8.82119909e-02 -8.32381472e-02 -8.36673900e-02 3.01443249e-01 -1.11722410e+00 8.46948326e-01 6.66691208e+00 6.46152914e-01 -8.08770776e-01 -1.39651924e-01 3.68441671e-01 6.02661185e-02 -1.39008224e-01 -3.95485401e-01 -9.67788041e-01 4.35063362e-01 1.16828501e+00 -4.80282515e-01 -1.81149244e-01 9.49579656e-01 1.15393795e-01 -1.20945267e-01 -1.05947256e+00 6.54122710e-01 9.56256807e-01 -1.65556860e+00 5.86774945e-01 -2.73592800e-01 6.09746814e-01 -2.68689275e-01 -1.37417972e-01 6.05468988e-01 -9.63494256e-02 -9.27721024e-01 4.56733912e-01 9.58521724e-01 1.06854424e-01 -7.68482685e-01 1.14846551e+00 7.64258981e-01 -8.28507602e-01 -3.51915717e-01 -8.65435302e-01 -4.44152862e-01 -4.81674774e-03 1.69169039e-01 -8.18246007e-01 7.12181568e-01 3.21375459e-01 1.00007844e+00 -8.02741051e-01 1.06312227e+00 -1.27666175e-01 4.65679735e-01 3.40247512e-01 -6.48096681e-01 3.80706519e-01 -6.52812719e-02 4.48241293e-01 1.51372683e+00 3.41123253e-01 -1.30198270e-01 2.83181015e-02 6.28063440e-01 -1.81628242e-01 4.28952992e-01 -6.52711153e-01 -7.87999183e-02 4.68789250e-01 1.21084785e+00 -5.15673637e-01 -6.12491310e-01 -2.53695369e-01 8.49445403e-01 7.01829195e-01 1.47774041e-01 -5.11081457e-01 -8.14563453e-01 1.01708546e-01 -3.75268795e-02 5.95877349e-01 -1.57875717e-01 -9.96589214e-02 -1.21909142e+00 -4.36304249e-02 -5.83310366e-01 2.67406195e-01 -6.87668204e-01 -1.26786864e+00 9.07824039e-01 -6.99554989e-03 -9.71052110e-01 -4.46643054e-01 -5.58212161e-01 -9.68502223e-01 5.93441367e-01 -1.56350112e+00 -9.43882406e-01 -1.04449876e-01 2.48163715e-01 1.11830759e+00 -5.91520607e-01 1.14356494e+00 8.25352371e-02 -6.85527682e-01 2.69528151e-01 4.49871421e-01 1.65837139e-01 5.91619194e-01 -1.29117644e+00 1.44402817e-01 6.54281139e-01 6.87735155e-02 8.34852755e-01 6.47056162e-01 -4.44761276e-01 -1.00731432e+00 -9.86988187e-01 1.40233207e+00 -2.60647565e-01 4.64386225e-01 -1.73815936e-01 -9.39777672e-01 7.65549302e-01 9.35460091e-01 -4.44720566e-01 9.66567159e-01 3.68085206e-02 2.37005041e-03 -1.67791415e-02 -8.65519047e-01 3.83296758e-01 5.91950536e-01 -1.76255003e-01 -1.51390576e+00 3.20715100e-01 7.96102166e-01 2.33725626e-02 -5.30924082e-01 2.98716798e-02 3.06551635e-01 -9.84591067e-01 7.29030967e-01 -9.36700881e-01 1.00727236e+00 7.79483095e-02 -1.97205842e-01 -1.50643635e+00 -7.52625346e-01 -1.43942609e-01 -5.76474607e-01 1.65642583e+00 3.22582275e-01 -4.01740521e-01 3.81080657e-01 2.16270238e-01 -4.84908551e-01 -9.67737615e-01 -6.78221881e-01 -4.07842577e-01 -1.68110225e-02 2.30761603e-01 3.97630751e-01 7.56577551e-01 2.62800992e-01 1.02579880e+00 -3.46006244e-01 -3.62769693e-01 2.47803584e-01 2.17954203e-01 4.34215814e-01 -1.38800979e+00 -1.17001936e-01 -4.98411238e-01 -4.09441501e-01 -1.26887298e+00 2.82131106e-01 -1.09087741e+00 -1.03367336e-01 -2.12600708e+00 5.85086584e-01 4.04796720e-01 -4.49584365e-01 2.23997142e-02 -2.36649811e-01 -4.22327071e-01 1.96217895e-01 9.20618698e-02 -9.03548062e-01 1.02400756e+00 1.06509781e+00 -1.18727542e-01 -1.02629580e-01 -1.33137882e-01 -1.13487923e+00 8.22987497e-01 1.08347845e+00 -3.89179975e-01 -4.03034687e-01 -6.51196122e-01 5.02654128e-02 -4.76687737e-02 -1.77268754e-03 -8.99282098e-01 4.44839090e-01 2.39837080e-01 6.15982771e-01 -1.18232012e+00 3.63436043e-01 -6.03962600e-01 -6.34690166e-01 3.05539131e-01 -9.84852314e-01 1.27859861e-01 8.48923698e-02 7.08185911e-01 -5.81769526e-01 -8.35725188e-01 4.04704958e-01 -3.20408285e-01 -5.73226511e-01 -2.13898033e-01 -5.39625168e-01 -1.90941975e-01 8.40987563e-01 -1.58434868e-01 -4.29985255e-01 -4.39220607e-01 -6.38483465e-01 1.19375102e-01 -2.48202533e-01 6.56138778e-01 9.63368773e-01 -1.42912447e+00 -9.09094512e-01 3.16916220e-02 -2.10454315e-02 -2.57990807e-01 2.67250955e-01 3.98632616e-01 -2.74944097e-01 8.67345691e-01 -3.12441945e-01 -1.67358860e-01 -1.30057812e+00 6.14217043e-01 -2.47851517e-02 -5.42482853e-01 -4.87521023e-01 6.59799814e-01 2.97028851e-02 -3.07865232e-01 5.33333182e-01 -4.75850195e-01 -1.06414413e+00 3.86282653e-01 8.13696265e-01 4.91689950e-01 -5.89684248e-02 -5.78729272e-01 -1.80051252e-02 4.06775862e-01 -6.03899300e-01 2.93046445e-01 1.66843855e+00 -2.17187226e-01 -3.65973562e-01 5.71087062e-01 1.29638267e+00 -2.76819408e-01 -6.31273150e-01 -3.38662326e-01 2.44602785e-01 -3.25341433e-01 1.50447369e-01 -6.70508921e-01 -6.90342247e-01 1.14642835e+00 9.54134241e-02 5.13406277e-01 8.17866445e-01 -1.46277294e-01 8.24354291e-01 6.09293222e-01 -1.93881944e-01 -1.41941011e+00 4.46624368e-01 7.87183642e-01 1.38339531e+00 -1.07475877e+00 9.16049033e-02 3.84638621e-03 -6.84200704e-01 1.47263503e+00 8.20650578e-01 -4.63245332e-01 4.48731899e-01 1.94169320e-02 -4.43460256e-01 -1.73296928e-01 -1.17491019e+00 1.56372130e-01 4.72317576e-01 3.80294383e-01 8.85613620e-01 -2.09896058e-01 -7.01401591e-01 1.17057717e+00 -2.82846242e-01 -1.19532011e-01 7.42496789e-01 9.50600684e-01 -1.18172073e+00 -6.83944106e-01 -2.27038398e-01 6.62902951e-01 -5.40237069e-01 -8.87891278e-02 -6.32315099e-01 4.99033421e-01 -1.94242373e-01 8.43145669e-01 1.57122165e-01 -2.16898337e-01 4.14058864e-01 3.29895049e-01 2.26370245e-01 -1.02803516e+00 -7.71638691e-01 -2.19944239e-01 2.04591826e-01 -5.75327016e-02 -4.38502252e-01 -4.10908431e-01 -1.08346522e+00 9.40238833e-02 -4.25804377e-01 4.44201231e-01 6.77709818e-01 7.55406380e-01 5.70347369e-01 1.03591990e+00 6.10222697e-01 -1.12124252e+00 -1.02651381e+00 -1.50759053e+00 -5.60158491e-01 2.41005346e-01 2.81721741e-01 -3.00561905e-01 -3.48961145e-01 1.59009576e-01]
[12.287875175476074, 9.24690055847168]
afaf4b2e-2438-4cd0-aea6-5a89b866ce59
deep-learning-for-real-time-gravitational
1711.07966
null
http://arxiv.org/abs/1711.07966v2
http://arxiv.org/pdf/1711.07966v2.pdf
Deep Learning for Real-time Gravitational Wave Detection and Parameter Estimation with LIGO Data
The recent Nobel-prize-winning detections of gravitational waves from merging black holes and the subsequent detection of the collision of two neutron stars in coincidence with electromagnetic observations have inaugurated a new era of multimessenger astrophysics. To enhance the scope of this emergent science, we proposed the use of deep convolutional neural networks for the detection and characterization of gravitational wave signals in real-time. This method, Deep Filtering, was initially demonstrated using simulated LIGO noise. In this article, we present the extension of Deep Filtering using real data from the first observing run of LIGO, for both detection and parameter estimation of gravitational waves from binary black hole mergers with continuous data streams from multiple LIGO detectors. We show for the first time that machine learning can detect and estimate the true parameters of a real GW event observed by LIGO. Our comparisons show that Deep Filtering is far more computationally efficient than matched-filtering, while retaining similar sensitivity and lower errors, allowing real-time processing of weak time-series signals in non-stationary non-Gaussian noise, with minimal resources, and also enables the detection of new classes of gravitational wave sources that may go unnoticed with existing detection algorithms. This approach is uniquely suited to enable coincident detection campaigns of gravitational waves and their multimessenger counterparts in real-time.
['E. A. Huerta', 'Daniel George']
2017-11-21
null
null
null
null
['gravitational-wave-detection']
['miscellaneous']
[-3.71502548e-01 -1.65404961e-01 5.67757010e-01 2.07923651e-02 -5.00450253e-01 -5.31166673e-01 1.12547517e+00 -3.29815656e-01 -5.23880839e-01 3.46068650e-01 -9.62824151e-02 -6.74986124e-01 -2.90450543e-01 -1.04878533e+00 -3.39155823e-01 -8.84345174e-01 -5.96004307e-01 8.41975808e-01 3.00822169e-01 -1.57712817e-01 1.19887151e-01 8.51472437e-01 -1.31143761e+00 -1.73675090e-01 1.52421091e-03 1.11739182e+00 1.78668439e-01 7.85337508e-01 3.37746888e-01 3.63649070e-01 -6.65769041e-01 -2.59784877e-01 6.38095260e-01 -5.47128022e-01 -4.12269652e-01 -5.06186724e-01 2.21560568e-01 -1.70215636e-01 -9.74818170e-01 9.56988275e-01 9.72406447e-01 2.75914729e-01 1.96233898e-01 -8.32493663e-01 -8.56396034e-02 4.96437371e-01 -1.69485182e-01 9.11346316e-01 -2.58421320e-02 8.35621119e-01 8.12611878e-01 -7.36673594e-01 4.16535646e-01 8.44230175e-01 7.92206824e-01 6.28318414e-02 -9.96664345e-01 -7.68451691e-01 -1.07817268e+00 3.08809161e-01 -8.37923348e-01 -5.42269349e-01 5.19032359e-01 -6.49108827e-01 1.38383436e+00 2.47716844e-01 6.72762692e-01 8.90930355e-01 -1.46749914e-01 1.75590977e-01 9.59589303e-01 -5.99030912e-01 2.99910128e-01 -7.00745225e-01 -2.64382452e-01 4.58340913e-01 5.12257159e-01 1.26608396e+00 -7.96509624e-01 -4.02079701e-01 8.39755416e-01 -5.35320997e-01 -3.79525751e-01 2.39052311e-01 -1.43257415e+00 1.05421090e+00 1.43358573e-01 6.48513794e-01 -3.04300666e-01 3.65866452e-01 1.69850498e-01 4.13152397e-01 4.62601751e-01 9.61271048e-01 -6.18665040e-01 -3.64405245e-01 -1.07661080e+00 5.08935392e-01 7.80604720e-01 3.90719116e-01 4.91598934e-01 5.36839664e-01 1.42547533e-01 6.55525848e-02 1.80479825e-01 7.68282115e-01 9.06177223e-01 -5.47492385e-01 3.83009873e-02 -1.57085702e-01 2.18705520e-01 -6.67113900e-01 -1.14832699e+00 -1.03407109e+00 -7.13629365e-01 6.25767708e-01 8.11289132e-01 -5.23179829e-01 -8.70410085e-01 1.36537492e+00 3.00065845e-01 5.27050138e-01 -7.08853006e-02 1.04231346e+00 9.24769998e-01 3.60238254e-01 -3.24605078e-01 -1.15697831e-01 1.55887556e+00 -2.19365552e-01 -2.48434678e-01 -6.55811787e-01 3.33979130e-01 -9.86875355e-01 -4.50905636e-02 3.71574849e-01 -8.67225766e-01 -3.98731232e-01 -9.99827325e-01 4.19099510e-01 -1.76752090e-01 -1.17000222e-01 1.31768239e+00 6.95423603e-01 -6.38537526e-01 8.79473269e-01 -1.11546457e+00 -2.21792176e-01 2.27257177e-01 1.38474688e-01 8.55768546e-02 8.11536133e-01 -1.22825742e+00 6.43002033e-01 5.27330875e-01 -3.69377658e-02 -7.40332067e-01 -9.38504219e-01 -3.63951087e-01 2.59584248e-01 -1.65502176e-01 -6.62337303e-01 1.35377073e+00 -2.35939637e-01 -1.10384643e+00 1.01227880e+00 4.36490238e-01 -1.20461655e+00 5.34650624e-01 1.51187614e-01 -9.34277356e-01 1.68079406e-01 5.27544459e-03 -2.96257704e-01 8.17462206e-01 -2.48316973e-01 -9.98482466e-01 -2.09733039e-01 -3.71359497e-01 -6.10243261e-01 4.88124728e-01 3.39359701e-01 1.35610834e-01 -6.82459891e-01 7.48875916e-01 -5.40172935e-01 -1.40897706e-01 -6.92337036e-01 8.66350234e-02 -2.70570099e-01 5.11113048e-01 -4.63298231e-01 3.86311978e-01 -2.08672047e+00 -4.82044846e-01 -1.20358868e-02 4.44904685e-01 3.24683845e-01 2.84654886e-01 4.51489538e-01 -3.84763151e-01 -2.61459678e-01 1.62194386e-01 3.89630161e-02 1.03838861e-01 -3.30106646e-01 -5.20363808e-01 8.05778086e-01 -1.24056369e-01 1.05553651e+00 -8.60538960e-01 4.63510692e-01 3.44035119e-01 7.04504475e-02 -2.60563940e-01 2.97866970e-01 -7.10116029e-02 8.19129348e-01 -4.08467539e-02 2.19614089e-01 9.77539182e-01 5.29684797e-02 -3.52353930e-01 -1.23949111e-01 -5.89602590e-01 5.65200329e-01 -1.11602724e+00 1.28084660e+00 -2.36860022e-01 1.12807453e+00 4.42366183e-01 -1.28997278e+00 1.16738367e+00 3.38226616e-01 3.62577856e-01 -1.02077389e+00 3.56084704e-01 4.93155152e-01 2.56617278e-01 -9.25397694e-01 3.33839178e-01 -8.74466956e-01 2.79804040e-02 4.35024470e-01 5.51872253e-01 -4.54929829e-01 1.66850597e-01 -1.36861160e-01 1.38169408e+00 -2.23811761e-01 1.56077454e-02 -2.16144368e-01 -1.57104686e-01 -1.44841045e-01 4.71707374e-01 1.48641586e+00 -9.86854807e-02 8.25013340e-01 1.11653194e-01 -9.66343284e-01 -1.07915223e+00 -9.86010015e-01 -4.31205988e-01 6.23253345e-01 -1.51037559e-01 -2.09527388e-01 1.02830902e-01 -3.34382541e-02 2.25176215e-01 6.90610766e-01 -2.91456848e-01 -3.55482578e-01 -5.44676125e-01 -1.72287822e+00 6.76295996e-01 1.27788946e-01 4.38232601e-01 -1.09746301e+00 -9.87927854e-01 4.19744194e-01 5.35640009e-02 -1.02001095e+00 4.72544700e-01 5.90918958e-01 -5.52980602e-01 -1.09670222e+00 -1.93368480e-01 -4.23280180e-01 -1.25742003e-01 3.32903676e-02 1.17319202e+00 -2.04529073e-02 -7.56787181e-01 -4.88325395e-03 -2.91293532e-01 -5.49418986e-01 -4.48792547e-01 -4.41619396e-01 3.21788967e-01 8.03819392e-03 6.48481786e-01 -9.81657088e-01 -5.36305010e-01 1.32409945e-01 -5.42183518e-01 -2.94198602e-01 4.62806642e-01 7.01701999e-01 -3.32241565e-01 3.52699280e-01 6.53076768e-01 -1.01307578e-01 3.03378284e-01 -5.75362027e-01 -1.43048978e+00 -3.82022083e-01 -1.70733705e-01 1.06271066e-01 3.78247738e-01 -3.53799164e-01 -8.30232561e-01 -2.13649139e-01 -4.39230710e-01 -8.00338387e-02 -2.26013273e-01 3.71219575e-01 4.41549182e-01 -3.90403181e-01 1.02588725e+00 1.05516143e-01 -2.74121433e-01 -7.65372336e-01 2.01969057e-01 4.10835683e-01 1.32442009e+00 -3.00928593e-01 1.22250640e+00 7.79437721e-01 3.20003003e-01 -8.59478772e-01 -5.53318679e-01 -7.80123949e-01 -1.92848578e-01 -1.02158874e-01 5.57682157e-01 -9.18564439e-01 -9.66120005e-01 7.62010634e-01 -1.19192016e+00 1.22683190e-01 -4.57622826e-01 1.31446850e+00 -3.27640533e-01 3.23517233e-01 -2.75569171e-01 -9.26896632e-01 -1.40229970e-01 -4.42144871e-01 5.25083482e-01 4.67538625e-01 1.45305365e-01 -7.13124335e-01 3.72758746e-01 1.24604777e-01 7.92677581e-01 2.82401115e-01 2.87964702e-01 -8.67825270e-01 -5.75840831e-01 -7.78690934e-01 -1.66767433e-01 -1.32023871e-01 -2.55817950e-01 -3.61329466e-01 -1.11739028e+00 -3.36439282e-01 8.44039440e-01 1.08791344e-01 1.22163987e+00 6.83547258e-01 3.88174951e-01 2.01841757e-01 -2.44439363e-01 1.05932426e+00 1.34411311e+00 2.00581402e-01 2.43354768e-01 5.70725739e-01 8.77923816e-02 9.22632515e-02 -1.87299065e-02 8.09240758e-01 -4.36073661e-01 4.68936503e-01 5.64848781e-01 -8.39716867e-02 -2.37234786e-01 3.27850133e-01 -1.85714319e-01 4.06110585e-01 -3.98350924e-01 1.14596345e-01 -8.92944217e-01 7.09998131e-01 -1.59794497e+00 -1.45746040e+00 -8.25703919e-01 2.59425807e+00 2.79801916e-02 4.86482441e-01 -5.75585514e-02 -2.60260314e-01 6.93320274e-01 5.38033657e-02 -7.41751567e-02 2.91888695e-02 -4.13796872e-01 8.36744726e-01 9.48063791e-01 4.87926364e-01 -1.15244210e+00 2.40182564e-01 6.44822550e+00 5.59025407e-01 -1.11224043e+00 5.47376871e-01 5.81154823e-02 -4.80209976e-01 -7.98144713e-02 2.09582120e-01 -4.20949191e-01 6.14727020e-01 1.01854813e+00 -3.02570492e-01 4.79053020e-01 4.01805818e-01 3.54492575e-01 -2.52967209e-01 -6.01692319e-01 1.31332350e+00 -4.60698456e-01 -1.70380437e+00 -9.12070274e-01 -5.82443811e-02 5.06498277e-01 9.61888611e-01 -4.87431496e-01 2.61016965e-01 3.58767837e-01 -5.91089964e-01 9.01333809e-01 4.97449756e-01 1.94476068e-01 -6.56209767e-01 8.63333523e-01 4.59791511e-01 -8.29696655e-01 -1.11837938e-01 -6.40141010e-01 -6.88363016e-01 6.25968635e-01 1.34424961e+00 -9.41372514e-01 7.62026131e-01 7.29622006e-01 1.02869049e-01 -3.75347555e-01 1.61107290e+00 -3.07903379e-01 1.12436843e+00 -9.69113827e-01 1.19687011e-02 2.77321547e-01 -4.88547236e-02 1.05025828e+00 1.09747601e+00 6.76789284e-01 5.61703630e-02 -1.85239702e-01 8.96058857e-01 -7.84195866e-03 -6.25884414e-01 -5.14015198e-01 5.18308766e-02 1.38843298e-01 1.36026418e+00 -8.30311894e-01 -3.73295218e-01 -5.28476119e-01 7.01989681e-02 -1.54450849e-01 1.43651709e-01 -6.62328660e-01 -5.22446811e-01 5.51029742e-01 1.71641111e-01 6.10628784e-01 -6.97486281e-01 -4.15732235e-01 -1.32406867e+00 -1.06090300e-01 -1.78103969e-01 3.87839884e-01 -4.98284876e-01 -1.17800617e+00 4.51431632e-01 -3.21521938e-01 -1.10976255e+00 -2.58374929e-01 -5.43881595e-01 -1.26256645e+00 1.03876591e+00 -1.10052264e+00 -5.21597624e-01 -1.24701381e-01 -8.25261772e-02 1.44498823e-02 -5.53113937e-01 4.67572093e-01 3.36468160e-01 -4.27585654e-02 -2.28424743e-01 5.42919993e-01 1.21563420e-01 4.36172545e-01 -1.09773445e+00 1.18348157e+00 1.37677395e+00 4.38101590e-01 4.97558117e-02 1.52803767e+00 -8.00295770e-01 -1.35214400e+00 -5.70717216e-01 8.70481014e-01 -1.74355328e-01 1.13749349e+00 -6.03980541e-01 -7.14916766e-01 4.97206897e-01 1.58390790e-01 3.80978465e-01 9.42439213e-02 4.93829697e-02 -9.97270718e-02 1.57539517e-01 -1.12030423e+00 -1.72261968e-01 8.63594294e-01 -4.43571091e-01 -9.93028343e-01 7.04698920e-01 3.39679658e-01 -2.16766208e-01 -1.48463950e-01 7.09396362e-01 4.47094589e-01 -1.03233039e+00 1.02454197e+00 -7.57939637e-01 -2.24730700e-01 -3.49014640e-01 2.57862419e-01 -1.22844505e+00 -5.99416316e-01 -1.34696448e+00 -3.27878147e-02 8.95129800e-01 9.37016122e-03 -8.24792922e-01 8.23054373e-01 8.59524533e-02 -1.28099561e-01 1.66387230e-01 -1.71512187e+00 -1.19619477e+00 2.46289689e-02 -8.83288860e-01 4.73380417e-01 8.27865958e-01 -4.70115431e-02 -2.47651767e-02 -3.95462364e-01 7.91406572e-01 9.12096977e-01 5.34990489e-01 5.09125710e-01 -1.39020646e+00 -8.79911542e-01 -5.42506278e-01 -8.89421165e-01 -7.70888031e-01 -4.91193473e-01 -1.06772041e+00 2.11713105e-01 -8.75607371e-01 -5.14670730e-01 -1.49491102e-01 9.84933227e-03 1.43760920e-01 1.85341865e-01 6.15712047e-01 -8.66896436e-02 1.22253418e-01 4.52643745e-02 2.42295519e-01 3.92861634e-01 4.80179526e-02 1.77555948e-01 3.56316954e-01 1.43660074e-02 8.52613389e-01 7.52069771e-01 -7.32625723e-01 4.07975912e-01 -4.55219805e-01 7.04137027e-01 2.41901457e-01 7.66943574e-01 -1.55467498e+00 5.04336834e-01 2.30362147e-01 4.51017082e-01 -5.36134541e-01 6.03069589e-02 -1.67013049e-01 3.45750004e-01 6.44634724e-01 4.36999798e-01 -4.27247405e-01 1.83704779e-01 2.40594894e-01 -1.23840705e-01 -8.14078450e-01 9.92194176e-01 -4.11875993e-01 -6.12328708e-01 8.14885050e-02 -7.99927652e-01 1.29675478e-01 6.37307584e-01 5.78214049e-01 -4.11247820e-01 -2.58357704e-01 -9.29772675e-01 -2.06353337e-01 -7.92009011e-02 3.32638085e-01 7.14652613e-02 -1.06049812e+00 -1.11489439e+00 5.19310951e-01 -3.37976694e-01 -3.11400324e-01 1.18363075e-01 6.92695498e-01 -7.39353597e-01 4.08522695e-01 -1.00384288e-01 -3.36009502e-01 -7.75783122e-01 5.34525633e-01 6.92656159e-01 -5.84376510e-03 -8.56526673e-01 1.11339128e+00 3.79314050e-02 -3.65079820e-01 -3.12137097e-01 -1.35085359e-01 3.12941074e-01 1.30523860e-01 8.12149465e-01 2.78356552e-01 6.92881346e-01 -4.52971429e-01 -2.55575866e-01 9.29032937e-02 4.20910507e-01 -8.34221542e-02 1.57992876e+00 6.14848845e-02 -1.55316070e-01 6.76328316e-02 8.39120328e-01 1.69448018e-01 -9.75726485e-01 -2.40118608e-01 1.09462194e-01 -5.35825551e-01 3.74916852e-01 -8.29747975e-01 -1.23651266e+00 7.35331833e-01 7.88959682e-01 9.90508914e-01 6.93566322e-01 7.12553680e-01 6.56998277e-01 3.83534640e-01 6.33863032e-01 -6.56319916e-01 -5.70817173e-01 4.75747705e-01 6.05839074e-01 -9.42909241e-01 -1.13036938e-01 2.18415499e-01 4.64596272e-01 1.60024834e+00 8.09459686e-02 -2.34499276e-01 7.03708708e-01 4.88558918e-01 -9.41220522e-02 -6.99198842e-01 -5.87558389e-01 -4.28409159e-01 -2.67437380e-02 4.70514804e-01 -1.25956893e-01 3.03827882e-01 -4.89670426e-01 5.35299778e-01 -6.12827420e-01 -3.08034420e-01 7.49968708e-01 7.50696123e-01 -8.30794990e-01 -6.95102811e-01 -9.99949813e-01 5.14855862e-01 -6.65384293e-01 -2.08058909e-01 1.30241036e-01 4.35753912e-01 2.84190923e-01 1.05028975e+00 2.60882825e-01 1.99551936e-02 9.76152271e-02 2.54434347e-01 3.93570870e-01 -1.99688330e-01 -6.23362303e-01 1.64603144e-01 2.26116896e-01 -2.44398206e-01 -1.91594467e-01 -7.66486764e-01 -8.80269945e-01 -1.69170097e-01 -5.79753518e-01 4.86058891e-01 9.04789448e-01 1.22176087e+00 1.26369342e-01 4.69598144e-01 6.70797586e-01 -1.31259358e+00 -5.54774642e-01 -1.18150461e+00 -7.66946495e-01 3.71015191e-01 5.58893800e-01 -5.68631172e-01 -1.16200054e+00 -2.51666367e-01]
[7.555856704711914, 3.1289854049682617]
c54462e8-29af-4591-a2ec-762347604889
large-margin-convex-polytope-machine
null
null
http://papers.nips.cc/paper/5511-large-margin-convex-polytope-machine
http://papers.nips.cc/paper/5511-large-margin-convex-polytope-machine.pdf
Large-Margin Convex Polytope Machine
We present the Convex Polytope Machine (CPM), a novel non-linear learning algorithm for large-scale binary classification tasks. The CPM finds a large margin convex polytope separator which encloses one class. We develop a stochastic gradient descent based algorithm that is amenable to massive datasets, and augment it with a heuristic procedure to avoid sub-optimal local minima. Our experimental evaluations of the CPM on large-scale datasets from distinct domains (MNIST handwritten digit recognition, text topic, and web security) demonstrate that the CPM trains models faster, sometimes several orders of magnitude, than state-of-the-art similar approaches and kernel-SVM methods while achieving comparable or better classification performance. Our empirical results suggest that, unlike prior similar approaches, we do not need to control the number of sub-classifiers (sides of the polytope) to avoid overfitting.
['Anthony D. Joseph', 'J. D. Tygar', 'Alex Kantchelian', 'Ling Huang', 'Peter L. Bartlett', 'Michael C. Tschantz']
2014-12-01
null
null
null
neurips-2014-12
['handwritten-digit-recognition']
['computer-vision']
[ 8.21283981e-02 -1.28100961e-01 -6.55189455e-01 -3.77934843e-01 -1.26708817e+00 -6.99874699e-01 3.86986077e-01 3.01714838e-01 -2.86467701e-01 9.14299667e-01 -3.26710075e-01 -7.53215432e-01 -3.36245120e-01 -5.54260433e-01 -8.38124454e-01 -7.01295435e-01 -3.47547859e-01 8.98067713e-01 3.15294027e-01 1.60590738e-01 4.99545723e-01 3.06075126e-01 -1.07762098e+00 7.42726922e-01 9.15355861e-01 1.30282223e+00 -3.14397514e-01 7.36879885e-01 1.97297037e-01 6.80837512e-01 -4.67714578e-01 -4.84409750e-01 4.00030643e-01 1.77051663e-01 -7.49001801e-01 3.91846985e-01 8.34065378e-01 -6.33864403e-02 -3.39532167e-01 8.37197900e-01 1.13636635e-01 -8.81068036e-02 1.15917993e+00 -1.63542438e+00 -5.26507676e-01 2.60216951e-01 -7.44468570e-01 2.48484090e-01 -4.72913234e-04 -1.81561902e-01 1.04215515e+00 -8.91817391e-01 2.37984508e-01 1.26116228e+00 1.29686499e+00 -1.58222113e-02 -1.37882102e+00 -4.46176887e-01 9.40260012e-03 6.42349496e-02 -1.32283378e+00 -3.75454985e-02 2.79070199e-01 -5.64861536e-01 1.05017173e+00 3.56187820e-01 2.20918834e-01 1.01249754e+00 4.33338940e-01 1.09253967e+00 1.30199051e+00 -4.34170038e-01 5.33473432e-01 4.72917944e-01 5.97849905e-01 8.49219143e-01 1.87141612e-01 -1.25648513e-01 -2.58431286e-01 -1.00186169e+00 5.68764567e-01 -1.30103007e-01 1.21684432e-01 -1.00409734e+00 -1.11954224e+00 1.09847784e+00 1.49347842e-01 -1.86229616e-01 -2.89061796e-02 8.44305977e-02 5.49596667e-01 3.13421160e-01 6.82007432e-01 3.82656664e-01 -8.66568983e-01 -1.87697336e-01 -1.00845885e+00 2.14485064e-01 1.35025036e+00 9.01205480e-01 4.73216563e-01 -2.40165278e-01 7.46170357e-02 9.02802944e-01 -2.94186715e-02 3.67371410e-01 3.35254282e-01 -5.93556583e-01 9.45743501e-01 5.38060009e-01 2.42346197e-01 -9.04429018e-01 -4.54113334e-01 -4.31910843e-01 -8.06613624e-01 2.44366914e-01 7.08083153e-01 -3.32654983e-01 -7.40244746e-01 1.02655458e+00 3.44027013e-01 8.62293541e-02 5.93611524e-02 5.27373135e-01 5.91374785e-02 7.15580106e-01 -1.36040570e-03 8.28721896e-02 1.11210585e+00 -1.20220387e+00 -8.98020938e-02 -4.32631969e-01 9.07608807e-01 -6.11616254e-01 1.05069613e+00 7.69306302e-01 -5.75535238e-01 -1.89706296e-01 -1.28714299e+00 2.58918881e-01 -5.63793182e-01 4.41954881e-01 1.04457438e+00 8.45202506e-01 -5.56294858e-01 8.86433005e-01 -9.49616015e-01 -1.08878635e-01 6.02039814e-01 5.53504884e-01 -3.86842519e-01 -8.98054764e-02 -6.88036621e-01 9.87878919e-01 5.25694251e-01 -2.23725647e-01 -6.88483417e-01 -7.31284559e-01 -7.33284354e-01 -5.39356992e-02 1.60326689e-01 -9.08051729e-02 1.03036726e+00 -9.25897956e-01 -1.35436988e+00 9.08137679e-01 6.19886967e-04 -6.46344662e-01 7.24604785e-01 -2.27860093e-01 -1.53042212e-01 7.14528635e-02 -2.87050962e-01 3.04138362e-01 9.85902250e-01 -1.04440951e+00 -8.81138027e-01 -4.55849767e-01 -2.73032427e-01 7.34724998e-02 -7.38455772e-01 1.65210858e-01 -2.46807158e-01 -6.27931535e-01 1.76090710e-02 -1.19937837e+00 -3.98407996e-01 9.63254571e-02 -4.04701442e-01 -4.70591038e-01 8.49591494e-01 -5.09422243e-01 1.09648550e+00 -2.04369044e+00 7.48801157e-02 3.68383825e-01 2.45241925e-01 1.58148751e-01 -2.68210080e-02 2.57724404e-01 1.37706678e-02 -8.83475021e-02 -4.84554857e-01 -3.74942750e-01 3.39077339e-02 1.82092831e-01 -4.07214999e-01 9.78977561e-01 1.73915148e-01 5.11473179e-01 -7.41961360e-01 -3.83524537e-01 -8.64574388e-02 3.15582715e-02 -3.00963938e-01 -3.71438533e-01 -2.18422130e-01 -3.25186819e-01 -4.77404118e-01 9.31792676e-01 9.06055391e-01 -7.24424899e-01 1.12981342e-01 1.61264285e-01 3.03727299e-01 4.01756950e-02 -1.45398533e+00 1.10362899e+00 -3.11510831e-01 8.65617454e-01 1.02409832e-01 -1.58381879e+00 8.41484547e-01 -6.19806163e-02 4.96155560e-01 -1.16369069e-01 3.26382406e-02 4.39307690e-01 -3.72846335e-01 -1.11953855e-01 1.60084024e-01 2.07313702e-01 -7.07889721e-02 3.32646072e-01 -1.37713820e-01 -8.45972896e-02 9.55944788e-03 -1.52919799e-01 1.13328898e+00 -2.65846521e-01 2.79456288e-01 -6.03776395e-01 3.44665051e-01 4.91780996e-01 4.33349282e-01 8.64518344e-01 -2.49018639e-01 5.81390023e-01 5.40361345e-01 -6.97199583e-01 -1.13480401e+00 -1.07959950e+00 -6.81019902e-01 1.33403122e+00 7.66864195e-02 -2.03542396e-01 -6.54487789e-01 -8.87178302e-01 6.65356457e-01 3.42375338e-01 -5.84026456e-01 1.43133119e-01 -6.56904757e-01 -1.46902895e+00 5.57305336e-01 7.64162540e-01 3.41845751e-01 -4.24207389e-01 -2.03361675e-01 1.65310636e-01 1.25666246e-01 -1.28354251e+00 -3.40381503e-01 6.74276710e-01 -1.24521196e+00 -1.15395784e+00 -6.39739811e-01 -1.11610806e+00 7.99643040e-01 6.19491003e-02 1.04444420e+00 -4.83408332e-01 -3.92278612e-01 2.12833062e-01 -1.16368726e-01 -3.15566033e-01 -3.57779443e-01 2.46576816e-01 1.84645712e-01 -3.02214205e-01 6.37902617e-01 -2.25583822e-01 -2.08908930e-01 6.52739584e-01 -4.04503584e-01 -2.89336890e-01 5.79037309e-01 1.12122142e+00 5.66057444e-01 5.38125709e-02 4.29063678e-01 -8.01944315e-01 6.49240792e-01 -7.80690908e-01 -9.43053067e-01 6.46335185e-01 -7.37237334e-01 6.70453683e-02 9.19436753e-01 -9.28577423e-01 -4.38841611e-01 3.00544024e-01 4.72364724e-01 -3.84760588e-01 1.92279313e-02 3.72977704e-01 2.74281502e-01 -6.21515810e-01 1.06670380e+00 3.14384133e-01 -1.61607087e-01 -2.06669167e-01 1.87863648e-01 1.11873436e+00 3.53234112e-01 -7.81045735e-01 7.78817773e-01 5.51375389e-01 9.70506072e-02 -7.68217444e-01 -8.51375222e-01 -5.71122766e-01 -7.23508358e-01 3.50833267e-01 5.17806828e-01 -9.39454675e-01 -7.11559415e-01 4.02642190e-01 -7.58300960e-01 -5.15524745e-01 1.02656037e-01 4.99591172e-01 -7.05351651e-01 4.94078219e-01 -8.04466367e-01 -7.79961586e-01 -3.74181837e-01 -8.03380132e-01 1.05241370e+00 9.12362859e-02 1.05220443e-02 -1.05611682e+00 4.59965952e-02 6.02487564e-01 7.97295943e-02 2.78675348e-01 1.00763321e+00 -9.74270105e-01 -2.53807098e-01 -5.97926855e-01 -3.48167807e-01 5.92315435e-01 -7.39691854e-02 -2.73769312e-02 -7.17153907e-01 -6.84979320e-01 -2.19408751e-01 -9.14278150e-01 7.24822044e-01 3.31982136e-01 1.41311359e+00 -3.71998012e-01 -6.69687271e-01 7.70003140e-01 1.36820793e+00 5.89272231e-02 3.57139856e-01 6.80594862e-01 4.55103397e-01 2.73566902e-01 6.02466583e-01 6.37108445e-01 3.37671041e-01 4.27132934e-01 8.91650394e-02 -1.82492897e-01 4.31539685e-01 -7.20210671e-02 3.99424791e-01 2.37523079e-01 3.86749238e-01 -1.50971174e-01 -1.23638797e+00 4.94424582e-01 -2.05718255e+00 -5.64738154e-01 -2.43474785e-02 2.25684094e+00 1.01538193e+00 4.02354717e-01 2.56969720e-01 2.08195046e-01 6.58309877e-01 -1.07067645e-01 -7.20515728e-01 -6.17282987e-01 -1.66217610e-01 6.02248274e-02 9.83008862e-01 2.41818517e-01 -1.78225279e+00 9.81699526e-01 7.56900167e+00 1.29453862e+00 -9.15180743e-01 -8.11215639e-02 1.01838934e+00 9.32079330e-02 4.29745674e-01 -1.72813728e-01 -1.04918826e+00 3.65643412e-01 9.98576343e-01 -9.95072201e-02 5.46918690e-01 1.44327927e+00 -2.81499326e-01 -5.79423644e-02 -1.51603389e+00 1.08945799e+00 1.03704482e-01 -1.41981399e+00 -3.65061402e-01 7.93735906e-02 9.77994382e-01 3.08154404e-01 3.63390148e-01 4.65914756e-01 7.72822380e-01 -1.19867325e+00 5.32826126e-01 -2.11395875e-01 6.03994012e-01 -7.74126112e-01 4.37083244e-01 5.99965096e-01 -9.29086208e-01 -4.86846656e-01 -7.56472528e-01 6.16268069e-02 -3.98084760e-01 6.88426435e-01 -8.11123788e-01 3.04591265e-02 6.52637303e-01 3.96881372e-01 -5.18869221e-01 1.14353871e+00 2.55594701e-01 6.44186318e-01 -5.74165583e-01 -3.57603520e-01 5.08574784e-01 -1.89581394e-01 3.25085253e-01 1.53367424e+00 -1.96456403e-01 6.38338551e-02 6.36181474e-01 4.37801778e-01 -9.87757146e-02 1.79677874e-01 -3.11196357e-01 2.32647732e-02 3.48842859e-01 1.07177186e+00 -8.90989840e-01 -3.54411334e-01 -5.69644749e-01 1.02537012e+00 6.13769770e-01 2.20848963e-01 -8.46386790e-01 -5.99056423e-01 4.99828815e-01 -1.35023952e-01 6.46300137e-01 -4.46131945e-01 -7.84619451e-01 -1.28805542e+00 3.10907722e-01 -1.17390311e+00 5.85069478e-01 -1.96354449e-01 -1.74461746e+00 4.28393066e-01 -2.03264251e-01 -1.19175589e+00 -7.94582590e-02 -1.16834104e+00 -3.21663022e-01 4.18535888e-01 -1.24152756e+00 -1.03358412e+00 9.99308750e-02 6.64684594e-01 4.11911905e-01 -2.40494579e-01 9.13467944e-01 2.08878871e-02 -5.11442959e-01 9.49525476e-01 1.08793950e+00 2.24639341e-01 6.64061368e-01 -1.38739371e+00 3.71581554e-01 4.08326358e-01 1.33583322e-01 2.12741062e-01 5.84145010e-01 -5.55815339e-01 -1.53558040e+00 -1.00771320e+00 5.11360884e-01 -7.72666216e-01 9.98315394e-01 -6.75504744e-01 -7.07360387e-01 5.76700330e-01 -2.88056672e-01 4.15914297e-01 8.86073112e-01 4.22762185e-01 -6.56907976e-01 -9.05218571e-02 -1.42576528e+00 3.49957049e-01 6.40461862e-01 -4.45756465e-01 -6.38387561e-01 1.12662351e+00 2.21286878e-01 -5.23689806e-01 -8.33993733e-01 3.69439960e-01 6.41552448e-01 -3.66195589e-01 1.13648880e+00 -1.15399516e+00 4.86834466e-01 1.32006750e-01 -3.41570556e-01 -1.23869228e+00 -1.75818369e-01 -5.98729193e-01 -4.72789139e-01 6.30750000e-01 5.93343019e-01 -6.22722924e-01 1.26489806e+00 6.84955001e-01 1.70299619e-01 -1.15881574e+00 -1.09020436e+00 -1.40920472e+00 6.11155987e-01 -2.83478409e-01 -3.51076238e-02 1.02672625e+00 5.34337819e-01 -5.23295589e-02 -3.46239448e-01 3.69860262e-01 9.04818594e-01 3.34396452e-01 7.16150999e-01 -1.29019725e+00 -5.33034801e-01 -2.73302495e-01 -5.35435736e-01 -1.16957903e+00 3.90017629e-01 -8.59794915e-01 -5.09627722e-02 -1.09484160e+00 3.35382313e-01 -7.59155631e-01 -3.37080628e-01 6.51559830e-01 -7.98528641e-02 4.81351502e-02 -1.29955783e-01 1.46774188e-01 -8.28778207e-01 3.02427858e-01 6.39817834e-01 -3.66290987e-01 -3.32133561e-01 3.38043511e-01 -5.13463795e-01 7.81961083e-01 7.41524398e-01 -6.12240493e-01 -1.41615480e-01 -3.42926800e-01 -1.70251533e-01 -1.69575978e-02 1.42027885e-01 -1.10272789e+00 3.75131279e-01 -3.67563844e-01 6.50180340e-01 -4.36786979e-01 1.31953910e-01 -7.26698101e-01 -5.79844952e-01 5.78836262e-01 -4.68525559e-01 -6.84221089e-03 4.11737025e-01 6.98742092e-01 1.78894754e-02 -1.32711008e-01 1.14386272e+00 1.66198865e-01 -2.25464776e-01 2.32355624e-01 -5.40684462e-01 3.05456460e-01 1.31222069e+00 -1.35976627e-01 -5.64449549e-01 -7.36420602e-02 -4.57823843e-01 3.74283522e-01 1.48340017e-01 3.76297772e-01 3.58960837e-01 -1.07669067e+00 -7.73933470e-01 -1.05206400e-01 1.08200736e-01 -2.81539798e-01 -3.38132977e-01 5.29545069e-01 -6.64539516e-01 7.29833663e-01 2.98784301e-03 -6.67850792e-01 -1.34628379e+00 6.91411376e-01 3.30610722e-01 -6.11408353e-01 -5.86376667e-01 9.74547207e-01 -2.30337679e-01 -6.05441153e-01 7.04148829e-01 -4.28466290e-01 2.64350504e-01 -2.04051092e-01 5.49249649e-01 6.96864009e-01 2.67676443e-01 -7.37729296e-02 -4.21710551e-01 4.04990435e-01 -3.25551271e-01 1.37716472e-01 1.48214233e+00 4.22563016e-01 5.55667840e-02 2.20314533e-01 1.50634933e+00 -2.55904555e-01 -1.39435041e+00 -3.28383327e-01 3.06527585e-01 -5.59419453e-01 5.10599129e-02 -7.12678552e-01 -5.52151918e-01 6.16838396e-01 7.10694492e-01 1.92665786e-01 7.28587210e-01 -1.24961957e-01 6.90356493e-01 1.13959026e+00 5.76691270e-01 -1.44101751e+00 -1.81648843e-02 6.89987838e-01 6.04948759e-01 -1.49379539e+00 2.74040997e-01 -5.55370033e-01 -5.99162400e-01 1.50600123e+00 5.82670987e-01 -3.62711549e-01 9.08620238e-01 6.03657246e-01 -2.44561642e-01 1.97199643e-01 -7.86878705e-01 5.27729511e-01 2.47130364e-01 6.00694597e-01 -8.12767744e-02 2.31051907e-01 -6.17455952e-02 5.38192093e-01 -2.03971475e-01 -8.10718015e-02 2.42492825e-01 1.22589207e+00 -4.48075801e-01 -8.73013377e-01 -6.12095952e-01 7.27075160e-01 -4.42620724e-01 -1.72573552e-01 -3.41413438e-01 7.00739741e-01 -1.66284591e-01 8.76089573e-01 7.70835429e-02 -4.32132900e-01 1.48578435e-02 2.00109091e-02 4.41815853e-01 -4.13962424e-01 -4.36676949e-01 -2.74615318e-01 4.90396917e-02 -4.08615381e-01 2.27100387e-01 -7.68503189e-01 -1.02447021e+00 -4.12098885e-01 -4.86549407e-01 3.80170643e-02 7.33865738e-01 8.99006605e-01 2.25489497e-01 -1.74165741e-01 9.06638801e-01 -7.52686203e-01 -1.45815575e+00 -7.19144166e-01 -6.33193433e-01 6.95437863e-02 1.56258091e-01 -6.24248207e-01 -6.44675553e-01 -8.14500451e-02]
[8.334735870361328, 4.040724277496338]
7cfa7237-6a5d-4cd1-885e-64cff8677f7e
film-ensemble-probabilistic-deep-learning-via
2206.00050
null
https://arxiv.org/abs/2206.00050v4
https://arxiv.org/pdf/2206.00050v4.pdf
FiLM-Ensemble: Probabilistic Deep Learning via Feature-wise Linear Modulation
The ability to estimate epistemic uncertainty is often crucial when deploying machine learning in the real world, but modern methods often produce overconfident, uncalibrated uncertainty predictions. A common approach to quantify epistemic uncertainty, usable across a wide class of prediction models, is to train a model ensemble. In a naive implementation, the ensemble approach has high computational cost and high memory demand. This challenges in particular modern deep learning, where even a single deep network is already demanding in terms of compute and memory, and has given rise to a number of attempts to emulate the model ensemble without actually instantiating separate ensemble members. We introduce FiLM-Ensemble, a deep, implicit ensemble method based on the concept of Feature-wise Linear Modulation (FiLM). That technique was originally developed for multi-task learning, with the aim of decoupling different tasks. We show that the idea can be extended to uncertainty quantification: by modulating the network activations of a single deep network with FiLM, one obtains a model ensemble with high diversity, and consequently well-calibrated estimates of epistemic uncertainty, with low computational overhead in comparison. Empirically, FiLM-Ensemble outperforms other implicit ensemble methods, and it and comes very close to the upper bound of an explicit ensemble of networks (sometimes even beating it), at a fraction of the memory cost.
['Konrad Schindler', 'Jan Dirk Wegner', "Stefano D'Aronco", 'Rodrigo Caye Daudt', 'Bernd Bischl', 'Mina Rezaei', 'Hüseyin Anil Gündüz', 'Alexander Becker', 'Mehmet Ozgur Turkoglu']
2022-05-31
null
null
null
null
['probabilistic-deep-learning']
['computer-vision']
[ 7.52334222e-02 4.84852880e-01 4.09533650e-01 -3.43341142e-01 -8.91417742e-01 -7.10982144e-01 9.63296592e-01 1.21025257e-01 -5.74534118e-01 1.15717411e+00 -2.11675861e-03 -1.70185238e-01 -5.17744899e-01 -7.14490175e-01 -9.39328253e-01 -9.21543837e-01 -3.88434343e-02 6.14146411e-01 1.68920457e-02 -2.33102217e-01 2.38403231e-01 1.84278920e-01 -1.50912380e+00 2.21042424e-01 8.62667918e-01 1.41491354e+00 -3.75635862e-01 4.40153867e-01 1.54558748e-01 8.47930670e-01 -7.78506756e-01 -8.41354251e-01 4.82004993e-02 -2.50856400e-01 -7.39043891e-01 -7.59615362e-01 3.36131990e-01 -2.09638420e-02 -1.19650867e-02 1.10438156e+00 6.39975846e-01 3.70444477e-01 7.86530137e-01 -1.10184407e+00 -2.68642247e-01 1.29636407e+00 -1.73561990e-01 1.13803349e-01 -2.19179869e-01 -5.80047481e-02 8.53453219e-01 -6.09409809e-01 1.54495358e-01 1.07520175e+00 1.01054502e+00 4.78788227e-01 -1.56888425e+00 -5.97179890e-01 7.44172279e-03 1.07213311e-01 -1.47549045e+00 -5.15785813e-01 4.90700752e-01 -4.74553585e-01 7.68642306e-01 2.18628973e-01 2.85073757e-01 1.60678363e+00 3.49487811e-01 3.93457711e-01 1.37504864e+00 -3.22166413e-01 7.20107913e-01 2.94928610e-01 1.38652902e-02 2.03782514e-01 3.26823324e-01 4.80168253e-01 -6.20038390e-01 -3.40305090e-01 3.91162246e-01 -4.79959011e-01 -3.80559683e-01 -2.43293092e-01 -1.17496049e+00 7.08787978e-01 3.30656976e-01 4.95056957e-01 -2.67374098e-01 7.45811164e-01 5.15920103e-01 4.15506601e-01 6.83444142e-01 7.52661049e-01 -6.21108055e-01 -2.50167698e-01 -1.05796695e+00 4.01670814e-01 1.04974067e+00 2.36299366e-01 4.94863033e-01 9.48304906e-02 4.21517007e-02 3.45039725e-01 2.08020896e-01 2.34236598e-01 4.19752181e-01 -1.24213946e+00 -5.06341383e-02 1.92883294e-02 2.09345207e-01 -9.57102954e-01 -4.35444444e-01 -8.24783921e-01 -1.05736411e+00 6.35049582e-01 6.24323070e-01 -4.50341851e-01 -5.71303606e-01 2.20453262e+00 8.58666450e-02 6.23131037e-01 1.13209903e-01 6.47086143e-01 1.65613234e-01 2.57341027e-01 5.92433177e-02 -9.30253491e-02 9.87673104e-01 -5.30460596e-01 -4.41103697e-01 -2.46061042e-01 6.68781400e-01 -1.77497998e-01 6.44655585e-01 8.47743213e-01 -7.81572282e-01 -2.38110349e-01 -1.26868832e+00 4.01126057e-01 -5.38216352e-01 -4.16459560e-01 4.90325928e-01 9.91200507e-01 -1.11005521e+00 1.27713382e+00 -6.35818720e-01 3.91652584e-01 4.57689285e-01 4.42733765e-01 -4.35067326e-01 2.63363540e-01 -1.62906408e+00 1.52306139e+00 9.15176153e-01 3.80842268e-01 -9.42852855e-01 -6.87907040e-01 -8.26099634e-01 1.27885446e-01 4.81789500e-01 -7.56885052e-01 1.23391402e+00 -1.31317377e+00 -1.77674055e+00 2.80676186e-01 3.52993488e-01 -6.54990017e-01 8.70380759e-01 -4.07173336e-01 -2.70291805e-01 -5.37116230e-01 -5.37796676e-01 2.85322875e-01 1.09030128e+00 -1.25151670e+00 -8.94693658e-02 -3.58034283e-01 4.70537692e-02 -6.70550093e-02 -5.75623736e-02 -1.99370965e-01 4.67876673e-01 -3.90972763e-01 -1.56124517e-01 -9.06675577e-01 -3.58469129e-01 -2.30042174e-01 -2.18990013e-01 -1.70452818e-01 2.43285149e-01 -3.29848051e-01 1.04735267e+00 -1.91724753e+00 5.54331124e-01 5.24902582e-01 4.10465240e-01 2.43403614e-01 7.68781751e-02 8.32233727e-02 -2.04977468e-01 2.96082526e-01 -6.73824072e-01 -6.36956096e-01 4.17241603e-01 2.26258472e-01 -4.38444227e-01 4.07614380e-01 1.23412378e-01 8.11405897e-01 -9.27998424e-01 -2.59229422e-01 3.75734605e-02 6.26202762e-01 -3.27315092e-01 -2.35526234e-01 -6.26654029e-01 5.79722881e-01 -1.40704781e-01 -1.24830157e-01 4.37741727e-01 -3.57783169e-01 3.91204536e-01 -2.97131967e-02 1.41426012e-01 2.08431557e-01 -1.47239482e+00 1.60318184e+00 -6.19082987e-01 6.57255948e-01 -1.78591102e-01 -1.12480843e+00 6.93952322e-01 5.32380819e-01 -2.42586993e-02 -3.31341214e-02 5.97399831e-01 6.34942234e-01 8.32768679e-02 5.62186465e-02 3.67727220e-01 -2.16321394e-01 -4.39578503e-01 7.46296883e-01 4.15330380e-01 -1.44745976e-01 -2.28773743e-01 -1.48970401e-02 9.56594884e-01 3.23991925e-01 1.62552834e-01 -4.19704556e-01 4.15711939e-01 -5.69558799e-01 2.06487134e-01 9.69630837e-01 -7.65213370e-03 4.09344524e-01 8.07556093e-01 -4.59514111e-01 -8.32727253e-01 -1.05614889e+00 -2.73026943e-01 1.08712447e+00 -1.85934320e-01 -1.40560552e-01 -8.80051196e-01 -5.78872740e-01 -3.84474993e-02 1.10340166e+00 -1.08547401e+00 -3.90750825e-01 -3.38002682e-01 -7.79706478e-01 8.38889241e-01 2.80848354e-01 4.36971188e-01 -7.86055505e-01 -5.40986598e-01 1.25739500e-01 3.63344103e-02 -7.87839115e-01 4.46654670e-02 4.07414645e-01 -6.68980658e-01 -8.41511905e-01 -4.29340243e-01 1.35437608e-01 1.10326968e-01 -7.67488241e-01 1.30701077e+00 -1.39334023e-01 3.98813248e-01 1.67751059e-01 -5.39263524e-02 -5.56775570e-01 -6.21348560e-01 7.41310120e-02 3.85501057e-01 1.71247974e-01 2.29848191e-01 -1.05691302e+00 -3.39700818e-01 -1.29433731e-02 -1.02165866e+00 -1.30612984e-01 4.93745863e-01 1.06486952e+00 2.07227036e-01 -9.89136845e-02 7.85088301e-01 -7.70835161e-01 7.79502869e-01 -6.38090372e-01 -5.97737432e-01 2.79965878e-01 -7.80311584e-01 5.49581945e-01 4.78281498e-01 -4.57618475e-01 -9.82669950e-01 -3.36226016e-01 2.50700135e-02 -3.45560342e-01 8.49506911e-03 7.75990307e-01 -4.65580402e-03 -2.32532978e-01 1.10851860e+00 -2.45463178e-02 1.26053572e-01 -2.38287464e-01 6.05751455e-01 4.96697217e-01 5.68191648e-01 -8.66194427e-01 2.73199767e-01 -9.29182861e-03 1.11247830e-01 -2.80085713e-01 -1.08372772e+00 5.54273546e-01 -5.59288442e-01 -3.88932616e-01 4.65043277e-01 -7.14309335e-01 -8.22889090e-01 6.60911977e-01 -1.24305236e+00 -3.49083424e-01 -3.61703277e-01 6.08841896e-01 -6.25745833e-01 1.29042014e-01 -2.54350901e-01 -8.83679688e-01 -2.71741390e-01 -1.03178549e+00 7.38113821e-01 1.06885200e-02 -3.59116882e-01 -1.16966939e+00 3.63799810e-01 -1.28152817e-01 8.60542417e-01 5.24356484e-01 7.13881195e-01 -1.18496287e+00 -2.24325776e-01 -2.36838087e-01 -3.55637521e-02 6.94564342e-01 -4.93691057e-01 -1.18251100e-01 -1.43672752e+00 5.63246831e-02 1.58914283e-01 -5.10565698e-01 1.03939176e+00 2.78929085e-01 1.29244316e+00 -1.06218152e-01 -3.67760137e-02 2.77388990e-01 1.28124118e+00 -3.87655735e-01 6.45313501e-01 9.05838311e-02 2.48833477e-01 5.21742821e-01 -1.99265465e-01 4.22081053e-01 1.00031331e-01 6.17479622e-01 7.25768805e-01 7.34509349e-01 4.73522455e-01 6.14442527e-02 3.45698744e-01 6.71100616e-01 -3.94244611e-01 -2.53205746e-01 -9.24335241e-01 1.98181510e-01 -1.94405437e+00 -1.13993168e+00 2.72166044e-01 2.44199133e+00 1.00220716e+00 3.59766185e-01 -2.09394336e-01 2.53755093e-01 5.61797202e-01 1.66559651e-01 -5.76095581e-01 -5.18768251e-01 -1.48080572e-01 4.52603042e-01 1.82056919e-01 6.58282459e-01 -1.05153704e+00 4.14750814e-01 6.76134157e+00 9.81474459e-01 -9.91153717e-01 3.92884582e-01 8.04557204e-01 -2.28937358e-01 -4.23393726e-01 -8.95304158e-02 -3.65534991e-01 6.97096288e-01 1.61319363e+00 -2.76715815e-01 4.59295690e-01 8.28862727e-01 -3.47827286e-01 -2.19750896e-01 -1.46796191e+00 8.78609300e-01 -2.07756124e-02 -1.44724941e+00 -3.70240927e-01 -4.43289950e-02 7.51455188e-01 2.11057916e-01 1.38563812e-01 5.28929174e-01 5.68933189e-01 -1.52308047e+00 8.52340341e-01 8.63213658e-01 5.25543213e-01 -8.26128542e-01 1.11493134e+00 6.92256093e-01 -3.72539937e-01 -5.27624600e-02 -1.47845671e-01 -1.98535979e-01 1.74411908e-01 9.74045157e-01 -1.12358660e-01 4.66849834e-01 5.73152483e-01 2.13295341e-01 -2.81539977e-01 7.32244194e-01 -1.03759833e-01 4.17936742e-01 -7.58580446e-01 -1.57283545e-01 1.33339539e-01 -1.02250054e-01 4.84918296e-01 1.01971066e+00 5.58977664e-01 -2.12738290e-01 -4.38642919e-01 1.02115154e+00 -1.81939304e-01 -4.41343546e-01 -5.37257850e-01 2.09939349e-02 5.14822185e-01 9.65458035e-01 -1.41692713e-01 -3.04423302e-01 1.25463083e-01 7.25061834e-01 5.91877937e-01 -2.75165745e-04 -1.00641561e+00 -2.76739091e-01 5.43417156e-01 -4.95703161e-01 1.33616939e-01 -1.08052835e-01 -3.81173342e-01 -9.44841564e-01 -1.12485059e-01 -9.09617364e-01 6.04319870e-02 -7.17453003e-01 -1.43433344e+00 8.88414741e-01 1.71414554e-01 -8.49172890e-01 -6.89871132e-01 -8.03952754e-01 -4.38062489e-01 1.00185430e+00 -1.02581656e+00 -8.09958756e-01 -9.52717587e-02 3.55252594e-01 -1.85343936e-01 -2.13971641e-03 1.24599528e+00 1.20183282e-01 -4.62281734e-01 6.55374050e-01 3.62038732e-01 -2.58047909e-01 6.83632612e-01 -1.40025902e+00 4.71584760e-02 5.83148241e-01 1.49814352e-01 5.15768826e-01 1.09946609e+00 -1.22629166e-01 -8.59012544e-01 -8.80144417e-01 6.00758076e-01 -1.01358569e+00 8.95206869e-01 -2.50381917e-01 -9.54410255e-01 7.82932818e-01 2.81182170e-01 1.55023783e-01 7.80682385e-01 3.20880145e-01 -7.01799452e-01 3.26564722e-02 -1.15580082e+00 4.61426437e-01 7.50231981e-01 -6.00611687e-01 -7.92702973e-01 1.22932233e-01 6.70179307e-01 -5.04367232e-01 -1.12964511e+00 4.40225840e-01 7.74838626e-01 -1.27409673e+00 5.31259835e-01 -6.34427071e-01 4.19920117e-01 -1.73130438e-01 -3.23243231e-01 -1.72735620e+00 7.98298195e-02 -6.91731632e-01 -5.84749162e-01 8.72230530e-01 6.59965873e-01 -9.97866750e-01 4.35528427e-01 8.14284027e-01 -1.35396406e-01 -7.76923180e-01 -1.36729062e+00 -6.86527550e-01 5.76129079e-01 -8.68345797e-01 7.49253392e-01 8.32930028e-01 1.72781833e-02 1.61441624e-01 -4.48005170e-01 1.66281387e-01 7.37891138e-01 -2.45595932e-01 3.45679760e-01 -1.62119901e+00 -6.90759122e-01 -8.01317573e-01 -5.27410090e-01 -4.86446261e-01 4.26172674e-01 -7.79218137e-01 8.30757245e-02 -7.84104943e-01 -1.87173605e-01 -4.93756562e-01 -5.50389647e-01 2.24608958e-01 1.18385136e-01 2.75529861e-01 -1.41573876e-01 -2.93523651e-02 -5.14513016e-01 5.29644012e-01 6.74207032e-01 1.05625140e-02 1.39156610e-01 1.28423488e-02 -7.49257863e-01 8.69793177e-01 8.03867936e-01 -5.70438862e-01 -3.08045924e-01 -4.21069294e-01 9.09867942e-01 -1.49133295e-01 6.73706770e-01 -1.20266294e+00 3.18564445e-01 1.55539930e-01 3.56566459e-01 3.62626985e-02 4.52970386e-01 -7.23153770e-01 7.57479429e-01 2.45780915e-01 -5.61506987e-01 -2.46780276e-01 3.93817276e-01 8.15471768e-01 -1.28011212e-01 -5.40375412e-01 7.13880360e-01 -2.14728311e-01 -4.49848354e-01 6.48701787e-02 -1.71281621e-01 -4.70863655e-02 9.75655973e-01 1.38606131e-01 -5.01232505e-01 -2.14292571e-01 -1.05034816e+00 6.29537776e-02 1.80148900e-01 9.63281542e-02 1.44162551e-01 -1.18240726e+00 -8.29730272e-01 -6.16281740e-02 -1.92424521e-01 4.38487716e-02 4.29327250e-01 1.06910300e+00 -1.66319028e-01 1.11725897e-01 1.93782710e-02 -6.12020612e-01 -4.72513676e-01 2.80045003e-01 9.51533318e-01 -3.66368443e-01 -1.36829138e-01 1.01234579e+00 -2.22959936e-01 -5.78244865e-01 1.41378775e-01 -6.98038116e-02 2.05703340e-02 1.02431603e-01 6.16937220e-01 3.03993970e-01 2.53744364e-01 -2.53605574e-01 -1.56655893e-01 2.96695381e-01 2.24924311e-01 -3.62455845e-01 1.21664703e+00 2.14281380e-01 -4.31058139e-01 8.48166466e-01 7.96487153e-01 -2.14523360e-01 -1.21782112e+00 -2.79167414e-01 1.86421141e-01 -1.68600559e-01 2.90736467e-01 -1.06422377e+00 -7.49966919e-01 9.83091533e-01 6.20409846e-01 2.55416393e-01 7.95790672e-01 -1.52461752e-01 4.49379832e-02 6.60475671e-01 7.91249037e-01 -9.63917971e-01 -1.51285321e-01 6.25998557e-01 9.89907444e-01 -1.19694912e+00 -1.43298740e-02 3.60361099e-01 -4.34822321e-01 1.07210183e+00 3.59838784e-01 -2.17652217e-01 1.04196680e+00 4.69937235e-01 -3.01387072e-01 1.20041203e-02 -9.60871637e-01 3.09118032e-01 1.63770691e-01 4.34619009e-01 2.73733467e-01 8.81212279e-02 -3.35544944e-02 1.01751137e+00 -1.78408369e-01 1.15352057e-01 4.08365816e-01 4.33081478e-01 -4.58515823e-01 -9.16864216e-01 -2.13715613e-01 5.11850893e-01 -5.68100929e-01 -1.86812639e-01 -1.04670804e-02 6.31168306e-01 2.47872204e-01 6.32631600e-01 1.28329590e-01 -5.36700189e-01 -1.02122100e-02 4.58686471e-01 6.27310574e-01 -2.27574900e-01 -6.52873278e-01 -6.70460880e-01 2.77371675e-01 -5.85612774e-01 -4.76645499e-01 -3.84339720e-01 -6.53012633e-01 -5.74711323e-01 -6.05214953e-01 4.04301286e-01 8.04399371e-01 1.26484430e+00 5.22831082e-01 4.52678353e-01 2.47513026e-01 -1.28378034e+00 -1.25693285e+00 -1.38442945e+00 -6.73602939e-01 -1.00863777e-01 3.39695841e-01 -9.67075586e-01 -8.73630643e-01 -3.47076476e-01]
[7.354802131652832, 3.8062994480133057]
b528aee1-6cce-4f25-990e-da095ec05581
classification-of-primitive-manufacturing
2303.09558
null
https://arxiv.org/abs/2303.09558v1
https://arxiv.org/pdf/2303.09558v1.pdf
Classification of Primitive Manufacturing Tasks from Filtered Event Data
Collaborative robots are increasingly present in industry to support human activities. However, to make the human-robot collaborative process more effective, there are several challenges to be addressed. Collaborative robotic systems need to be aware of the human activities to (1) anticipate collaborative/assistive actions, (2) learn by demonstration, and (3) activate safety procedures in shared workspace. This study proposes an action classification system to recognize primitive assembly tasks from human motion events data captured by a Dynamic and Active-pixel Vision Sensor (DAVIS). Several filters are compared and combined to remove event data noise. Task patterns are classified from a continuous stream of event data using advanced deep learning and recurrent networks to classify spatial and temporal features. Experiments were conducted on a novel dataset, the dataset of manufacturing tasks (DMT22), featuring 5 classes of representative manufacturing primitives (PickUp, Place, Screw, Hold, Idle) from 5 participants. Results show that the proposed filters remove about 65\% of all events (noise) per recording, conducting to a classification accuracy up to 99,37\% for subjects that trained the system and 97.08\% for new subjects. Data from a left-handed subject were successfully classified using only right-handed training data. These results are object independent.
['Pedro Neto', 'Laura Duarte']
2023-03-15
null
null
null
null
['action-classification']
['computer-vision']
[ 4.89606827e-01 4.83780392e-02 3.49303126e-01 -4.21114534e-01 -3.25079232e-01 -2.24466443e-01 4.62782174e-01 -5.28727472e-02 -4.36600119e-01 4.84435499e-01 5.50078861e-02 2.83833891e-01 -4.23945814e-01 -3.64466012e-01 -6.01762414e-01 -4.62478340e-01 -3.06484282e-01 3.83606374e-01 2.45681241e-01 -1.80979997e-01 4.24589694e-01 8.84462297e-01 -1.94505370e+00 8.26111197e-01 6.47100508e-01 1.07486606e+00 1.01140356e+00 7.48871326e-01 2.40625873e-01 1.12801301e+00 -1.00196505e+00 4.45593983e-01 3.73820066e-01 -3.08947533e-01 -4.92428213e-01 2.36423731e-01 -2.26537380e-02 -3.98650527e-01 -1.82695314e-01 7.00570524e-01 4.26349282e-01 5.16811132e-01 6.11589968e-01 -1.44862688e+00 -7.37014234e-01 6.16330862e-01 -9.15998518e-02 -1.18877664e-01 5.61333418e-01 4.67956901e-01 4.13335323e-01 -1.00535238e+00 8.04491282e-01 1.17902076e+00 4.70550060e-01 3.49689007e-01 -7.38044083e-01 -7.48999178e-01 2.70911872e-01 5.45227528e-01 -1.07772124e+00 -3.04177821e-01 8.58164251e-01 -5.78956664e-01 1.28300905e+00 6.99795876e-03 8.22945237e-01 1.58448637e+00 7.37246037e-01 8.46968710e-01 9.05622244e-01 -3.00743669e-01 4.36159879e-01 -3.32983524e-01 9.23256949e-02 5.64131737e-01 1.04234636e-01 1.07943557e-01 -7.32212842e-01 1.82081774e-01 1.03648865e+00 6.22324646e-01 -6.14600219e-02 -4.23818558e-01 -1.60622811e+00 3.93409967e-01 1.54832006e-01 3.93977582e-01 -9.82228220e-01 7.17365891e-02 3.49286288e-01 4.77341682e-01 -1.66066781e-01 5.98266959e-01 -6.39162838e-01 -4.64450359e-01 1.16963059e-01 1.84959158e-01 8.99159014e-01 1.43551803e+00 2.60030538e-01 4.62313592e-02 -1.91890210e-01 8.11300695e-01 4.26867343e-02 4.53728676e-01 4.95860189e-01 -1.34254181e+00 3.32864434e-01 5.84763885e-01 4.22914743e-01 -9.63507771e-01 -5.67939699e-01 1.63087323e-01 -6.97658002e-01 5.06216824e-01 1.90038964e-01 -8.35478008e-02 -1.04764056e+00 1.17779875e+00 4.77804057e-02 -4.26206708e-01 -6.58466388e-03 8.98709714e-01 3.15991908e-01 6.63316369e-01 5.76296300e-02 -3.01668733e-01 1.17716777e+00 -8.37383389e-01 -1.33661497e+00 -3.62013936e-01 2.62430191e-01 -8.28132570e-01 9.86584783e-01 1.07609439e+00 -8.35452437e-01 -1.19634342e+00 -1.23636341e+00 1.63069919e-01 -3.28233123e-01 4.81707782e-01 8.68737280e-01 -1.30811170e-01 -3.69201362e-01 8.69646311e-01 -1.16399586e+00 -4.07919854e-01 1.64009601e-01 4.30422813e-01 -4.91603345e-01 4.09709960e-02 -6.72076106e-01 9.56658244e-01 3.92501831e-01 3.02124202e-01 -1.15029418e+00 -1.53878421e-01 -9.18909729e-01 -2.73246318e-01 4.98055637e-01 -2.58821189e-01 1.42694271e+00 -7.03018785e-01 -1.90638196e+00 1.99641064e-01 3.23869795e-01 -4.16924685e-01 2.24229142e-01 -9.76085305e-01 -5.26563525e-01 1.06771007e-01 1.85714185e-01 4.57016796e-01 8.95365179e-01 -1.26888680e+00 -1.00802076e+00 -3.73324066e-01 -4.49342728e-01 5.47847897e-02 -4.72498983e-02 6.83735386e-02 -3.61623679e-04 -6.80183411e-01 4.32615668e-01 -1.19831359e+00 -1.27679467e-01 1.34710157e-02 -2.97225595e-01 -3.61008316e-01 9.16112840e-01 -8.18518579e-01 5.60193539e-01 -2.21832323e+00 1.00240044e-01 -3.82071212e-02 -2.46290356e-01 -9.38134044e-02 -1.00696340e-01 7.03761280e-01 -1.46950558e-01 -8.07376862e-01 2.94425283e-02 -2.04992592e-01 5.02810813e-03 2.11542144e-01 -2.07267195e-01 3.85854542e-01 1.66235983e-01 4.21524137e-01 -9.53617394e-01 -1.07150681e-01 6.52897894e-01 9.20481831e-02 -2.61629045e-01 4.46098924e-01 -5.05402237e-02 6.42689347e-01 -3.84058475e-01 8.45586240e-01 2.98724920e-02 2.54320979e-01 2.77742207e-01 -2.01072767e-01 -2.77693659e-01 -2.01754179e-02 -1.45101833e+00 1.94611013e+00 -5.59086263e-01 4.94061947e-01 2.27843285e-01 -8.26518118e-01 1.25692868e+00 4.60719824e-01 8.48578930e-01 -6.04375005e-01 2.62427300e-01 1.35417446e-01 2.11677790e-01 -1.06774783e+00 4.47391003e-01 5.74325681e-01 -4.30731118e-01 1.07029311e-01 1.91595420e-01 -3.41831237e-01 4.98320572e-02 -3.55517626e-01 1.51830220e+00 3.67150635e-01 2.04364032e-01 6.48940057e-02 1.25837639e-01 2.12217182e-01 6.87637269e-01 9.46625352e-01 -4.89017844e-01 3.37487519e-01 -2.17134640e-01 -6.43235385e-01 -6.56577349e-01 -1.25005829e+00 6.28115058e-01 1.05577457e+00 2.67012268e-01 4.37741205e-02 -1.61622599e-01 -3.78081679e-01 -7.84010068e-02 7.99718082e-01 -1.86737657e-01 -4.24027801e-01 -8.43988597e-01 1.23983668e-02 -2.34828442e-02 8.10924113e-01 5.97685993e-01 -1.64383340e+00 -1.20169473e+00 4.71479267e-01 -4.77151340e-03 -1.16879582e+00 -2.99594074e-01 6.09739304e-01 -7.17893124e-01 -1.26819098e+00 -2.65685946e-01 -1.22388339e+00 6.07191086e-01 2.69005150e-01 4.64220136e-01 -6.14653826e-01 -6.42765582e-01 6.20950758e-01 -4.74332213e-01 -8.09552073e-01 -3.83106947e-01 -5.11103153e-01 5.03934503e-01 -2.21885473e-01 2.70975083e-01 -5.49397588e-01 -3.70217621e-01 4.18676168e-01 -1.72912434e-01 1.71701565e-01 9.90744114e-01 8.26162815e-01 5.21069586e-01 2.88234383e-01 5.95444202e-01 -2.66988426e-01 7.11894512e-01 -8.73861909e-02 -2.85963535e-01 7.94892982e-02 -2.65639722e-01 -3.90699238e-01 5.69556236e-01 -9.91473377e-01 -1.44962335e+00 7.26324141e-01 3.80945146e-01 -6.03329420e-01 -6.58227146e-01 2.18102738e-01 -1.71866328e-01 3.66541892e-01 5.58338463e-01 6.22538431e-03 -4.72211093e-02 -3.33077013e-01 2.54717857e-01 8.43992591e-01 9.27727580e-01 -3.44722927e-01 3.05622190e-01 3.68909627e-01 -2.28437498e-01 -7.81706631e-01 -1.62230238e-01 -5.46219230e-01 -8.09988916e-01 -6.70368910e-01 1.00544679e+00 -8.92129183e-01 -1.19502771e+00 7.72291541e-01 -1.38290393e+00 -4.93162453e-01 -4.39787328e-01 1.10279608e+00 -1.02063727e+00 -3.83396596e-02 -9.03280318e-01 -1.20246589e+00 -1.19093843e-01 -9.78858769e-01 1.05081463e+00 -2.82421261e-02 -7.58171618e-01 1.21799288e-02 -4.61579084e-01 3.64963740e-01 5.48610240e-02 4.62746978e-01 5.92967331e-01 -8.51059854e-01 -5.87129176e-01 -4.08146381e-01 4.02715206e-01 4.96072918e-01 6.52502716e-01 -1.63090140e-01 -5.85944533e-01 -4.60652523e-02 4.91864771e-01 -2.19519600e-01 1.89627439e-01 2.74078339e-01 1.06601346e+00 -2.90963147e-02 -4.62708324e-01 -1.96156681e-01 6.46918833e-01 1.21854353e+00 5.35528660e-01 1.05652802e-01 6.11666799e-01 8.10919821e-01 1.31017053e+00 6.29765928e-01 1.26282815e-02 5.20453990e-01 4.54235643e-01 4.96242553e-01 1.36152357e-02 -3.17771025e-02 8.10615063e-01 9.64452446e-01 -2.96986133e-01 7.37370029e-02 -9.28019464e-01 4.33456838e-01 -2.13774204e+00 -9.57645953e-01 -2.64649868e-01 1.81573844e+00 5.13422370e-01 4.04143602e-01 -2.75303692e-01 4.11269307e-01 8.12032819e-01 -3.94815892e-01 -6.59175098e-01 -3.63477141e-01 3.14825207e-01 2.76611030e-01 2.64057904e-01 -1.41252413e-01 -1.20120943e+00 6.49595022e-01 5.84486723e+00 1.60751104e-01 -8.19737494e-01 -9.63484570e-02 -1.97117120e-01 -2.72108138e-01 8.78865302e-01 -4.12473202e-01 -3.59758526e-01 3.62069905e-01 7.85425365e-01 2.64811367e-01 3.88033569e-01 1.32454765e+00 4.19684500e-01 -3.75058830e-01 -1.61234510e+00 1.27133536e+00 1.21362336e-01 -8.44600916e-01 -3.21090341e-01 -3.11838746e-01 6.36567533e-01 -2.68262625e-01 -2.83159703e-01 6.56516492e-01 5.70797324e-01 -6.63617730e-01 1.01898110e+00 1.01332581e+00 -1.89169105e-02 -6.02658510e-01 4.92631853e-01 6.29730701e-01 -1.06309140e+00 -8.43363464e-01 1.04205698e-01 -5.08415520e-01 4.64728534e-01 3.71518821e-01 -1.05568802e+00 5.76101303e-01 1.08492005e+00 7.48864770e-01 -5.98074682e-02 5.30304074e-01 -2.40050033e-01 -6.51881620e-02 -2.33751789e-01 -2.49802023e-01 -2.59985983e-01 -4.89414968e-02 5.77937245e-01 6.73020661e-01 5.45306504e-01 7.01040328e-02 5.50120711e-01 7.68142343e-01 4.04551297e-01 -5.77148497e-01 -8.46775353e-01 -6.98559955e-02 6.31572485e-01 8.59147787e-01 -8.02251816e-01 -7.67688304e-02 -1.48467571e-01 1.38859892e+00 -1.77274317e-01 3.32828283e-01 -8.26819837e-01 -7.22457111e-01 7.11594105e-01 -4.12160814e-01 2.87601233e-01 -9.73474383e-01 -2.13884726e-01 -3.29745382e-01 2.87184656e-01 -8.36893022e-01 2.78347488e-02 -1.29026401e+00 -1.23279417e+00 1.80994168e-01 1.32028490e-01 -1.55510437e+00 -5.36944449e-01 -8.85408640e-01 -2.80314386e-01 5.46008945e-01 -3.96053106e-01 -8.18484247e-01 -6.07060254e-01 5.37680328e-01 1.20106709e+00 -2.80722201e-01 8.82487118e-01 1.27812609e-01 -3.21471274e-01 -1.13701224e-01 -3.36417973e-01 2.13698088e-03 8.05031717e-01 -8.07000577e-01 1.61616087e-01 5.92589974e-01 -1.14377216e-01 8.66317153e-01 6.37553751e-01 -1.15778720e+00 -1.96629393e+00 -9.88426745e-01 4.65525419e-01 -6.23897851e-01 3.59139323e-01 -3.94775838e-01 -5.04697680e-01 9.37087059e-01 3.17250416e-02 -1.61980480e-01 3.31050038e-01 -2.44106889e-01 4.12609696e-01 -8.66859928e-02 -1.06763315e+00 7.48839140e-01 1.54225564e+00 -4.13773000e-01 -1.14407778e+00 5.17796814e-01 7.23449409e-01 -5.58700025e-01 -9.70657587e-01 7.71599829e-01 6.72940254e-01 -6.51816010e-01 7.44685471e-01 -3.12440902e-01 2.75237024e-01 -3.37289512e-01 -8.30882341e-02 -1.28361285e+00 -4.14224625e-01 -6.44280374e-01 -3.63644153e-01 8.18214476e-01 -8.87468755e-02 -4.69404638e-01 3.85230452e-01 4.69069958e-01 -8.76302183e-01 -5.42393446e-01 -6.49070680e-01 -1.10073185e+00 -9.05165315e-01 -6.25389397e-01 3.05303097e-01 7.41285205e-01 1.58544302e-01 8.62131640e-02 -2.08307073e-01 1.55114710e-01 5.04679121e-02 -7.14311153e-02 9.77418363e-01 -1.34249783e+00 5.81780821e-02 -6.20922185e-02 -4.13410693e-01 -9.03811336e-01 -6.84942305e-02 -3.93842041e-01 7.61514962e-01 -1.88075244e+00 -3.86781901e-01 -1.64114505e-01 -9.13398862e-02 6.37996376e-01 4.71200734e-01 -5.14738798e-01 1.58580795e-01 3.48203242e-01 -4.92807150e-01 6.72795057e-01 1.25639462e+00 -2.13629574e-01 -4.57433969e-01 9.21461284e-02 -2.73389574e-02 8.92961860e-01 7.41361976e-01 -2.30939180e-01 -4.86731499e-01 -3.43361795e-01 -1.80451497e-01 2.35454142e-01 4.08919096e-01 -1.56965911e+00 5.72109580e-01 -3.46434057e-01 7.66251206e-01 -9.45285141e-01 7.56806195e-01 -1.19799852e+00 5.14298618e-01 9.15915668e-01 -2.91664958e-01 2.08341166e-01 1.25414714e-01 7.17410803e-01 -3.22639495e-02 -5.03261164e-02 1.63236782e-01 -1.64370373e-01 -1.36793721e+00 -1.65015846e-01 -9.71037567e-01 -7.97533989e-01 1.48842442e+00 -3.64982426e-01 -1.82961896e-01 -1.89595655e-01 -9.91525829e-01 8.75850320e-02 -8.00622106e-02 1.05673766e+00 8.60083938e-01 -1.30622602e+00 -8.58460814e-02 3.81930411e-01 1.78266063e-01 1.51924431e-01 2.90542066e-01 5.89024961e-01 -3.34504068e-01 2.72070885e-01 -6.81525290e-01 -6.58702791e-01 -1.01468825e+00 4.96432394e-01 -1.54837102e-01 2.34934464e-01 -1.04467654e+00 6.30135715e-01 -3.20310801e-01 -4.60761726e-01 5.96549511e-01 -8.36508811e-01 -1.80805698e-01 -2.96292361e-02 2.76626527e-01 6.38963699e-01 1.01728618e-01 -1.94623604e-01 -3.49664003e-01 8.70257393e-02 1.21067405e-01 -5.38340509e-02 1.44174898e+00 1.44309729e-01 2.54508197e-01 9.84811485e-01 5.07085681e-01 -4.17500228e-01 -1.46645176e+00 2.25900725e-01 2.49324262e-01 -1.78857386e-01 -5.03596485e-01 -1.09754324e+00 -6.45552278e-01 4.13540989e-01 8.88977408e-01 7.52421692e-02 9.43679750e-01 -2.49161661e-01 6.74120188e-01 1.05252075e+00 9.76572931e-01 -1.73899078e+00 5.78196228e-01 8.00794721e-01 1.50724268e+00 -1.10397363e+00 -1.58696294e-01 -5.59719682e-01 -7.55256653e-01 1.08353555e+00 8.63007665e-01 -2.71326929e-01 4.72850889e-01 4.70428109e-01 -1.40340775e-01 -3.24953020e-01 -5.68535864e-01 6.28452078e-02 -9.43614021e-02 8.63118351e-01 6.71419725e-02 5.31996414e-02 4.19942066e-02 1.06555915e+00 -6.67683780e-02 3.34330469e-01 4.06750709e-01 1.80165982e+00 -4.88577545e-01 -3.65735769e-01 -4.63627368e-01 4.96475756e-01 1.24256782e-01 7.20412135e-01 -2.36473888e-01 1.02059221e+00 4.51514959e-01 1.27349234e+00 2.75990695e-01 -6.58257127e-01 9.60166514e-01 1.63904756e-01 6.85161293e-01 -8.94234121e-01 -5.83418190e-01 -1.97279394e-01 2.33816013e-01 -1.00436473e+00 -2.33232841e-01 -9.17564332e-01 -1.68425369e+00 2.65909702e-01 -1.14377094e-02 -3.93449605e-01 9.13203180e-01 7.51706958e-01 4.95261163e-01 1.26990104e+00 3.97987604e-01 -1.40163755e+00 -6.89080060e-01 -1.40580320e+00 -5.51417649e-01 5.27351022e-01 -1.33670375e-01 -1.00336146e+00 -1.72206581e-01 5.11010826e-01]
[4.990148544311523, 0.7335023880004883]
e15e9cf0-59e8-4ee8-914f-426a71d105b3
dual-stream-transformer-for-generic-event
2207.03038
null
https://arxiv.org/abs/2207.03038v3
https://arxiv.org/pdf/2207.03038v3.pdf
Dual-Stream Transformer for Generic Event Boundary Captioning
This paper describes our champion solution for the CVPR2022 Generic Event Boundary Captioning (GEBC) competition. GEBC requires the captioning model to have a comprehension of instantaneous status changes around the given video boundary, which makes it much more challenging than conventional video captioning task. In this paper, a Dual-Stream Transformer with improvements on both video content encoding and captions generation is proposed: (1) We utilize three pre-trained models to extract the video features from different granularities. Moreover, we exploit the types of boundary as hints to help the model generate captions. (2) We particularly design an model, termed as Dual-Stream Transformer, to learn discriminative representations for boundary captioning. (3) Towards generating content-relevant and human-like captions, we improve the description quality by designing a word-level ensemble strategy. The promising results on the GEBC test split demonstrate the efficacy of our proposed model.
['Longyin Wen', 'Libo Zhang', 'YuFei Wang', 'Guang Chen', 'Hanhua Ye', 'Xin Gu']
2022-07-07
null
null
null
null
['boundary-captioning']
['computer-vision']
[ 5.15681028e-01 1.57017097e-01 -2.41384178e-01 -3.28687876e-01 -1.21735620e+00 -5.31820297e-01 7.04462886e-01 -1.69211701e-02 -1.58312216e-01 7.58658648e-01 8.30537736e-01 -9.04442817e-02 4.17920411e-01 -4.34903890e-01 -1.16420949e+00 -3.32084805e-01 -3.59818116e-02 3.98844540e-01 2.94286221e-01 -1.62411615e-01 1.19067378e-01 -5.39594330e-02 -1.43546808e+00 8.89668226e-01 9.09350634e-01 1.10571241e+00 4.05216604e-01 6.94091201e-01 -1.09610453e-01 7.43813336e-01 -5.40823042e-01 -3.20349604e-01 -7.18764216e-02 -7.28128254e-01 -8.18395734e-01 1.98381126e-01 3.34235162e-01 -2.17808679e-01 -3.90803158e-01 6.96919084e-01 4.30598348e-01 -1.16268381e-01 7.12392509e-01 -1.24807942e+00 -6.14134490e-01 7.46771932e-01 -3.51426423e-01 3.78499269e-01 5.90345502e-01 2.17336148e-01 1.23305321e+00 -7.31169820e-01 7.82218993e-01 9.27868068e-01 3.16722602e-01 9.23209548e-01 -6.73030019e-01 -3.56422722e-01 4.61504400e-01 6.28496468e-01 -1.21201611e+00 -4.81586933e-01 7.84751415e-01 -5.91906488e-01 6.22685194e-01 3.80914450e-01 3.75410348e-01 1.66769981e+00 -3.55936855e-01 1.15672505e+00 4.91824955e-01 -3.72813463e-01 9.05103460e-02 -5.02820313e-02 -2.64103562e-02 2.18810216e-01 9.71097201e-02 -3.63585413e-01 -5.23174047e-01 9.83225182e-02 4.66517121e-01 -3.86903971e-01 -7.57004976e-01 -1.14899211e-01 -1.40472972e+00 5.43411076e-01 1.45703495e-01 2.83974618e-01 -6.51580691e-01 2.40857810e-01 6.12467885e-01 -2.57935315e-01 4.58999634e-01 4.63749200e-01 -1.32408768e-01 -4.23940867e-01 -9.54044759e-01 2.59936512e-01 5.55576622e-01 1.07810426e+00 3.78134072e-01 -2.47797474e-01 -1.11772156e+00 8.15348089e-01 3.05244684e-01 3.97643387e-01 4.04931515e-01 -6.87798560e-01 1.21100605e+00 2.18036920e-01 6.23035789e-01 -9.79676723e-01 6.90699369e-02 -3.80655885e-01 -4.83377069e-01 -7.56067216e-01 6.66930154e-02 -2.58742034e-01 -1.03433192e+00 1.72178864e+00 -1.09865069e-02 7.06668675e-01 2.44807884e-01 1.05656469e+00 9.78279650e-01 1.26269495e+00 4.48625147e-01 -1.67085573e-01 1.53363001e+00 -1.19696474e+00 -9.00283754e-01 -3.64603162e-01 5.74268341e-01 -5.21564007e-01 8.10257673e-01 -2.83362232e-02 -1.11765277e+00 -5.16142309e-01 -1.08901048e+00 1.01844274e-01 1.21113040e-01 1.83478788e-01 2.32662514e-01 3.14508289e-01 -9.22427177e-01 2.40369916e-01 -5.14671266e-01 -1.28555655e-01 3.68667334e-01 -1.16716787e-01 -1.43685147e-01 -2.50187010e-01 -1.68957555e+00 5.41136980e-01 5.61129391e-01 1.81309149e-01 -1.20440078e+00 -5.95517933e-01 -1.08528697e+00 3.02384526e-01 3.42859536e-01 -5.69446146e-01 1.39229381e+00 -1.10549641e+00 -1.20915258e+00 6.49321795e-01 -5.51049709e-01 -5.96046746e-01 5.84462285e-01 -3.43977392e-01 -5.16069233e-01 7.39595711e-01 1.56861842e-01 1.06305289e+00 7.97389269e-01 -1.46413183e+00 -7.02516854e-01 3.55344385e-01 1.63126625e-02 5.75542271e-01 -3.79101276e-01 1.04049996e-01 -9.50484395e-01 -6.24039650e-01 -3.42796445e-01 -6.36411011e-01 -5.05082462e-05 -4.16774780e-01 -4.72726017e-01 -2.76356339e-01 7.93812990e-01 -1.18202472e+00 1.40226340e+00 -2.14630914e+00 2.05469280e-01 -1.56558484e-01 -7.49350116e-02 4.91896480e-01 -3.16363186e-01 4.23901200e-01 -1.06234103e-01 3.97049099e-01 -1.61792159e-01 -5.54537237e-01 -5.20619862e-02 1.82680022e-02 -5.40012300e-01 -7.65128061e-02 7.69665241e-01 1.00823236e+00 -1.02848375e+00 -6.28543317e-01 -1.27459064e-01 5.02684057e-01 -6.28585517e-01 5.36622524e-01 -7.16723680e-01 4.54720557e-01 -6.69851780e-01 3.46029133e-01 4.77132589e-01 -5.67349553e-01 9.08857584e-03 -3.22462708e-01 1.78374290e-01 4.01584536e-01 -6.58624709e-01 1.55427039e+00 -3.36526453e-01 6.48909986e-01 -2.26942301e-01 -1.02368975e+00 8.94871056e-01 1.01148748e+00 3.79605383e-01 -5.82160175e-01 -1.17658615e-01 -6.97712079e-02 -5.74179888e-01 -8.40056539e-01 4.64632422e-01 1.66118309e-01 -2.44778693e-01 1.82059109e-01 -9.02170092e-02 3.59795898e-01 3.03840846e-01 2.71902353e-01 1.06101906e+00 3.72905880e-01 -5.71555793e-02 2.97166407e-01 6.56146288e-01 -6.81667682e-03 7.58261859e-01 5.36641121e-01 -3.54868144e-01 1.24032319e+00 7.88037658e-01 -1.47684425e-01 -1.10881615e+00 -7.57260799e-01 2.89071500e-01 6.75682664e-01 3.58679444e-01 -4.76616144e-01 -9.23134923e-01 -8.50214422e-01 -4.71418023e-01 8.24445426e-01 -6.39101207e-01 -1.62105769e-01 -8.05995107e-01 -4.57601547e-01 3.89297605e-01 6.26680434e-01 6.44106567e-01 -1.16310120e+00 -2.99381226e-01 3.12971890e-01 -1.20955241e+00 -1.58519852e+00 -8.60714972e-01 -3.46315742e-01 -4.66978431e-01 -7.33598530e-01 -1.12766171e+00 -8.63627553e-01 4.34440911e-01 1.92315236e-01 9.82924819e-01 -4.16422188e-02 2.47498021e-01 4.68628228e-01 -1.01055586e+00 -1.25744462e-01 -5.14717102e-01 9.77014229e-02 -4.56526816e-01 4.90961939e-01 2.70916313e-01 -2.25806534e-01 -5.63334167e-01 3.77148598e-01 -7.28170812e-01 5.72459698e-01 5.41832566e-01 5.85590184e-01 5.46747088e-01 -4.00570542e-01 8.42553794e-01 -6.03262544e-01 5.14462769e-01 -7.92115450e-01 -2.62644589e-01 4.44987535e-01 -4.63364720e-02 -3.66536938e-02 6.04589641e-01 -3.29497308e-01 -1.24461842e+00 -5.55479787e-02 -1.11961618e-01 -4.74858522e-01 -1.39919937e-01 5.55466473e-01 -3.58650923e-01 6.07463002e-01 3.74383301e-01 4.76263940e-01 -2.92706311e-01 -5.31674922e-01 1.61555797e-01 9.50011611e-01 7.05380678e-01 -7.18182623e-01 6.70937419e-01 1.51799098e-01 -5.47680318e-01 -6.05357230e-01 -9.32933807e-01 -6.67118490e-01 -2.23762020e-01 -5.03091633e-01 1.22373283e+00 -1.33452308e+00 -2.41396502e-01 2.94854730e-01 -1.58103383e+00 -2.22971275e-01 8.21968168e-02 3.06704164e-01 -8.55356872e-01 4.20764446e-01 -4.41985160e-01 -5.87695062e-01 -3.71231854e-01 -1.17935789e+00 1.31426239e+00 4.50024962e-01 -5.00910245e-02 -7.28292108e-01 1.64513374e-04 8.32134247e-01 2.39244327e-01 4.13789243e-01 5.68356574e-01 -8.13786626e-01 -7.96118915e-01 -1.20461695e-01 -4.36413825e-01 3.38154644e-01 -8.44163448e-03 -1.29519016e-01 -8.95513237e-01 -7.47083873e-02 -3.44224274e-01 -3.09016973e-01 6.99943781e-01 1.91360429e-01 1.27259159e+00 -3.49960536e-01 -4.66785192e-01 3.56784910e-01 1.15205741e+00 2.89419532e-01 9.50233698e-01 3.66378784e-01 7.15657771e-01 5.60389161e-01 7.52648950e-01 4.89456028e-01 6.07348800e-01 9.45131660e-01 3.83032888e-01 1.18500300e-01 -3.04196954e-01 -6.53314412e-01 6.59869432e-01 7.39569604e-01 1.97431482e-02 -7.04158545e-01 -8.94652307e-01 9.33425307e-01 -2.00808740e+00 -1.13638437e+00 -6.00927509e-02 1.70264339e+00 8.70017588e-01 8.88353437e-02 -6.52054697e-02 -1.00821212e-01 1.30338168e+00 3.47411186e-01 -2.12626785e-01 -2.68580139e-01 2.01967303e-02 -2.19256967e-01 1.69303238e-01 1.98158190e-01 -1.24197829e+00 9.10061121e-01 5.41333151e+00 8.26951504e-01 -9.98366117e-01 1.85334951e-01 8.87811005e-01 1.99443161e-01 -4.95681405e-01 -8.67520720e-02 -1.03214931e+00 9.30669308e-01 1.09969842e+00 -2.92254418e-01 -1.12931598e-02 5.68743646e-01 5.25606871e-01 1.95340455e-01 -1.07599199e+00 9.07293260e-01 4.40622360e-01 -1.55078447e+00 3.83420050e-01 -2.69830614e-01 5.79656124e-01 -2.23345459e-01 -1.60192966e-01 5.17319322e-01 -1.89229503e-01 -8.90258729e-01 1.00201917e+00 5.23406386e-01 7.97419310e-01 -5.34376085e-01 8.25366795e-01 7.54249319e-02 -1.20630968e+00 4.87135863e-03 -1.23922691e-01 2.91301250e-01 8.28841686e-01 3.73890907e-01 -1.01009250e+00 8.85422766e-01 4.40835953e-01 8.45111310e-01 -5.29199123e-01 1.49002218e+00 -3.55907977e-01 1.02349770e+00 -9.90883112e-02 1.36790862e-02 4.93195176e-01 1.62588015e-01 5.28135896e-01 1.30613196e+00 5.41344166e-01 8.23152587e-02 4.29986529e-02 6.49687529e-01 -2.81064630e-01 8.28939974e-02 -2.80327052e-01 -4.10946518e-01 4.91428345e-01 9.80151415e-01 -4.99010175e-01 -4.23825711e-01 -4.32024747e-01 1.06338918e+00 9.08892527e-02 5.35069048e-01 -1.30842221e+00 -2.24885285e-01 5.29126108e-01 1.29694387e-01 7.79913306e-01 1.13026015e-02 1.10825889e-01 -1.34766495e+00 2.17985228e-01 -8.94259989e-01 3.59492987e-01 -1.04759371e+00 -9.74468768e-01 7.49762058e-01 2.03908563e-01 -1.32788277e+00 -2.48118028e-01 -2.19122291e-01 -7.70832360e-01 6.90679610e-01 -1.73582494e+00 -1.12604117e+00 -4.33405221e-01 3.42790395e-01 6.76617503e-01 1.46133795e-01 3.72507840e-01 6.18810475e-01 -7.39264369e-01 6.35281384e-01 -2.15876684e-01 3.91840696e-01 6.86852098e-01 -9.11662638e-01 5.80423951e-01 1.03991008e+00 9.55959558e-02 5.21236956e-02 9.10876632e-01 -8.05869281e-01 -9.61370528e-01 -1.41718745e+00 1.18002105e+00 -1.88909486e-01 3.93904001e-01 -3.64921957e-01 -9.08914685e-01 6.88932896e-01 2.75618792e-01 -1.41530141e-01 5.11338413e-01 -1.49144948e-01 -2.06110388e-01 2.77459379e-02 -7.61206388e-01 6.76402032e-01 9.67925072e-01 -3.51728410e-01 -8.14557314e-01 4.91037488e-01 1.09129763e+00 -3.21313053e-01 -5.37401021e-01 3.75848621e-01 6.19150177e-02 -5.18548608e-01 7.95798302e-01 -6.64939880e-01 6.68618679e-01 -3.55657041e-01 -6.32043034e-02 -1.05004537e+00 -5.00043370e-02 -8.43540728e-01 -1.76076502e-01 1.61741519e+00 5.27711213e-01 -9.50925890e-03 7.43475616e-01 5.07439017e-01 -4.82290894e-01 -4.92902726e-01 -8.35910738e-01 -7.15426385e-01 -3.26078147e-01 -3.56847256e-01 6.16048455e-01 6.52503014e-01 1.01125054e-01 3.97051960e-01 -7.44895339e-01 1.92874730e-01 2.00599238e-01 -2.87514906e-02 5.53270459e-01 -7.60275364e-01 -2.81239092e-01 -6.53853342e-02 -2.93848425e-01 -1.47711742e+00 1.06034704e-01 -8.07793915e-01 3.82598132e-01 -1.77530909e+00 3.07510763e-01 -2.93576330e-01 -2.02121034e-01 5.06076872e-01 -4.88034844e-01 1.36070877e-01 2.76820928e-01 2.52368391e-01 -1.04672074e+00 6.78403497e-01 1.33978176e+00 -1.34556055e-01 1.47260219e-01 -1.18037455e-01 -7.37237453e-01 4.90168780e-02 6.10073686e-01 -3.37413490e-01 -4.74534601e-01 -5.42255104e-01 2.17076480e-01 4.84405130e-01 3.39725256e-01 -1.17800581e+00 -9.05888975e-02 -1.09224223e-01 5.27059101e-02 -4.70876694e-01 3.99379969e-01 -5.12053788e-01 9.00502130e-02 7.89295137e-02 -4.68127191e-01 -9.32599902e-02 2.55986750e-01 5.97246885e-01 -5.39798141e-01 -4.64725196e-01 4.43306595e-01 3.81673984e-02 -8.36239815e-01 4.07811135e-01 -3.15147310e-01 3.91799778e-01 1.26183331e+00 -7.49644190e-02 -4.14780766e-01 -7.65313148e-01 -7.22541094e-01 5.82510471e-01 3.52854669e-01 6.99422121e-01 4.85130787e-01 -1.49427044e+00 -9.13421929e-01 -2.25872189e-01 3.77722502e-01 2.98216734e-02 4.75712031e-01 5.30618429e-01 -5.43530703e-01 5.91469586e-01 -1.44637987e-01 -5.81691682e-01 -9.72933888e-01 4.60206717e-01 4.93041500e-02 -3.94104123e-01 -6.55847073e-01 9.21423614e-01 1.97039083e-01 3.65244806e-01 1.67747304e-01 6.11936413e-02 -6.60515130e-01 2.81334817e-01 7.86216915e-01 -2.19910696e-01 -1.22922584e-01 -8.01796377e-01 -2.16167018e-01 2.42053837e-01 -3.05659503e-01 -1.03864573e-01 1.28320348e+00 -2.15767920e-01 3.82986963e-01 1.10763900e-01 1.19253850e+00 -3.74462277e-01 -1.68312597e+00 2.78763305e-02 2.51053572e-01 -1.95481032e-01 -1.97244689e-01 -8.83653462e-01 -7.29183614e-01 9.18980598e-01 2.05917507e-01 1.09715648e-02 9.98209894e-01 5.75197972e-02 1.23805904e+00 1.62766073e-02 1.54505685e-01 -9.89120781e-01 1.81252539e-01 4.89626527e-01 1.02600312e+00 -1.03203928e+00 -6.14516914e-01 -3.95052552e-01 -1.08142471e+00 7.88263619e-01 6.59759223e-01 2.11248279e-01 -1.84831433e-02 -2.35524684e-01 3.28090191e-02 1.00271344e-01 -9.18812573e-01 -1.13806024e-01 4.83509690e-01 6.11544013e-01 1.73516646e-01 -6.95499331e-02 -3.98847967e-01 8.15999925e-01 2.42437750e-01 1.46535188e-01 7.15876520e-01 7.29483962e-01 -4.88166958e-01 -1.06509697e+00 -3.05598855e-01 8.69731531e-02 -4.26661193e-01 -1.64835408e-01 -6.58753887e-02 3.95662576e-01 -6.52201772e-02 1.02890122e+00 1.50964692e-01 -2.41680995e-01 1.45261452e-01 3.65249097e-01 8.14507902e-02 -8.25267553e-01 -2.97933429e-01 -1.45387739e-01 5.21171391e-01 -6.05397046e-01 -3.44254851e-01 -7.14729071e-01 -1.01835346e+00 3.22271049e-01 -1.88040640e-02 7.32476354e-01 5.21673024e-01 9.42251086e-01 6.83666229e-01 6.31304741e-01 5.74918091e-01 -8.31141889e-01 -3.05599511e-01 -9.91572917e-01 1.03735190e-03 8.01987410e-01 1.39050305e-01 -6.59478009e-01 -3.51839364e-01 4.95022625e-01]
[10.500754356384277, 0.6569319367408752]
86afeb37-6bc3-45a4-b13e-3a967aca0169
estimating-uncertainty-in-pet-image
2306.04664
null
https://arxiv.org/abs/2306.04664v1
https://arxiv.org/pdf/2306.04664v1.pdf
Estimating Uncertainty in PET Image Reconstruction via Deep Posterior Sampling
Positron emission tomography (PET) is an important functional medical imaging technique often used in the evaluation of certain brain disorders, whose reconstruction problem is ill-posed. The vast majority of reconstruction methods in PET imaging, both iterative and deep learning, return a single estimate without quantifying the associated uncertainty. Due to ill-posedness and noise, a single solution can be misleading or inaccurate. Thus, providing a measure of uncertainty in PET image reconstruction can help medical practitioners in making critical decisions. This paper proposes a deep learning-based method for uncertainty quantification in PET image reconstruction via posterior sampling. The method is based on training a conditional generative adversarial network whose generator approximates sampling from the posterior in Bayesian inversion. The generator is conditioned on reconstruction from a low-dose PET scan obtained by a conventional reconstruction method and a high-quality magnetic resonance image and learned to estimate a corresponding standard-dose PET scan reconstruction. We show that the proposed model generates high-quality posterior samples and yields physically-meaningful uncertainty estimates.
['Damir Seršić', 'Tomislav Matulić', 'Tin Vlašić']
2023-06-07
null
null
null
null
['image-reconstruction']
['computer-vision']
[ 2.93009460e-01 3.47730100e-01 -2.09506974e-02 -5.98701358e-01 -1.47514737e+00 2.20540706e-02 2.80762464e-01 -2.31873706e-01 -5.74860334e-01 1.32805872e+00 1.13159418e-01 -2.08176881e-01 -1.06249340e-01 -7.02391684e-01 -9.57294106e-01 -9.79309916e-01 1.46434829e-01 1.02673614e+00 4.09317873e-02 7.40409851e-01 -1.80926934e-01 4.69193101e-01 -5.02384782e-01 2.42992863e-01 7.19753742e-01 1.08094513e+00 3.26976001e-01 3.78300101e-01 7.43496418e-02 7.65299201e-01 -4.42696184e-01 -1.44128978e-01 -8.02956894e-02 -7.56578803e-01 -8.43899071e-01 -2.60728329e-01 -4.35792834e-01 -8.76610041e-01 -4.42037433e-01 1.28801572e+00 6.40366375e-01 -1.37903631e-01 1.20597911e+00 -9.84742284e-01 -1.11815140e-01 9.49368060e-01 -4.45423961e-01 3.48166049e-01 5.10211289e-02 2.99737453e-02 3.82846385e-01 -9.26517487e-01 2.28875205e-01 7.17785060e-01 8.48923445e-01 6.41896188e-01 -1.55502856e+00 -5.70812285e-01 -5.66006839e-01 -3.09596676e-02 -1.36395192e+00 -3.11462045e-01 4.80382562e-01 -5.85272372e-01 3.23804319e-01 5.07152639e-02 7.76273131e-01 1.42562985e+00 6.49089634e-01 5.29429138e-01 1.48709869e+00 -8.78909901e-02 5.64935446e-01 -7.76288137e-02 -4.36306089e-01 5.99811614e-01 2.22737983e-01 4.80658948e-01 -2.93306801e-02 -4.14744556e-01 1.16041660e+00 -3.99401411e-03 -6.94187343e-01 -3.24140549e-01 -1.20195615e+00 8.82807314e-01 6.27849936e-01 1.21618330e-01 -7.60772526e-01 7.41580129e-01 2.34028280e-01 -2.65536755e-01 2.51648873e-01 2.62851685e-01 -9.60805174e-03 2.51752324e-02 -1.49262476e+00 1.22139871e-01 5.96665144e-01 5.26036203e-01 1.89180866e-01 1.33181781e-01 -3.89972597e-01 4.16834056e-01 7.08065808e-01 6.39654338e-01 6.05751812e-01 -8.84576023e-01 -1.31155923e-01 -5.40790200e-01 3.56407195e-01 -3.08687598e-01 -3.43226939e-01 -3.79772335e-01 -8.45861137e-01 3.20357054e-01 4.15760010e-01 -1.56175166e-01 -1.07500172e+00 1.79103518e+00 2.78609157e-01 4.69413936e-01 -2.20104396e-01 1.06034911e+00 7.45057166e-01 5.54881155e-01 3.89640659e-01 -5.06234884e-01 1.10771358e+00 -4.63417917e-01 -8.16905141e-01 -1.63098112e-01 -7.34258145e-02 -5.11842489e-01 6.64337993e-01 4.03396457e-01 -1.34498918e+00 -8.84669200e-02 -1.19215894e+00 2.80509591e-01 4.02944416e-01 -2.55945981e-01 2.12967813e-01 7.65871048e-01 -7.77481437e-01 9.10381854e-01 -1.21447170e+00 3.31948936e-01 7.87702501e-01 2.71423519e-01 -1.72791213e-01 -1.31154388e-01 -1.32400668e+00 1.25144315e+00 5.90672672e-01 3.94322984e-02 -1.64713633e+00 -8.74453068e-01 -4.76433188e-01 -2.38658607e-01 7.34273717e-02 -9.90469813e-01 1.64737654e+00 -6.06596470e-01 -1.65420711e+00 4.64915007e-01 2.40235105e-01 -7.14427769e-01 1.09806871e+00 1.46568716e-01 -1.69444308e-01 4.96657401e-01 3.65804672e-01 7.17394233e-01 1.35781252e+00 -9.97679293e-01 1.70594260e-01 -3.35696131e-01 -6.08501911e-01 9.91272647e-03 7.45485246e-01 -1.38223633e-01 1.14613153e-01 -4.38871384e-01 6.32914901e-01 -6.49799824e-01 -3.84086013e-01 1.16308749e-01 -3.86555254e-01 5.46976268e-01 1.34862378e-01 -8.62056434e-01 5.64047515e-01 -1.61744356e+00 5.81934750e-02 1.85403392e-01 3.48597497e-01 -3.75216216e-01 3.27414483e-01 -2.13515699e-01 -2.81086206e-01 4.51898091e-02 -8.45004499e-01 6.56016963e-03 -2.45098379e-02 2.28774667e-01 -2.35709891e-01 9.32653964e-01 1.10708304e-01 1.18069589e+00 -1.24459565e+00 -4.89891291e-01 3.13914239e-01 7.21518695e-01 -2.12480322e-01 3.39349240e-01 -2.76513249e-01 1.22564042e+00 -6.13591850e-01 3.64392102e-01 6.97231889e-01 -2.50408620e-01 8.51748586e-02 -5.01739502e-01 2.62598336e-01 5.23433574e-02 -6.36687696e-01 1.79017401e+00 -4.51504260e-01 1.49453297e-01 -1.07747935e-01 -9.16487277e-01 2.71443188e-01 5.93731523e-01 7.33061552e-01 -4.92634952e-01 4.89404410e-01 5.31852365e-01 -6.32958636e-02 -3.64312977e-01 -1.15031354e-01 -1.18044448e+00 -4.49089967e-02 6.88920438e-01 1.47498965e-01 -8.92217457e-01 -5.60801268e-01 1.25200570e-01 1.15430880e+00 3.14028502e-01 2.03891888e-01 -3.44462484e-01 8.39578658e-02 -4.33899909e-01 2.39827752e-01 7.34486163e-01 -4.05869901e-01 9.65723097e-01 4.38036889e-01 -1.81390151e-01 -1.37796998e+00 -1.63912630e+00 -9.10795212e-01 -9.75282863e-02 -2.43832290e-01 3.31817389e-01 -1.01792216e+00 -5.88354111e-01 -1.77198708e-01 1.08016872e+00 -6.10761344e-01 -3.00267667e-01 -2.07440794e-01 -1.22793818e+00 6.69730365e-01 5.61489761e-01 4.71998900e-01 -9.02303457e-01 -6.43472612e-01 5.40491998e-01 -6.67707503e-01 -1.02291822e+00 -2.60411024e-01 4.16154593e-01 -1.13445234e+00 -9.12546515e-01 -1.27073634e+00 1.38729541e-02 6.63194597e-01 -4.83899713e-01 1.15935934e+00 -5.68648756e-01 -2.12774366e-01 3.43099892e-01 1.37941003e-01 -2.93164253e-01 -8.75867605e-01 -5.68272829e-01 1.96433619e-01 -2.39263043e-01 -1.15461141e-01 -7.69483805e-01 -7.56358206e-01 1.94123656e-01 -1.01353335e+00 -5.93517162e-02 6.21874690e-01 1.24127102e+00 1.25429416e+00 1.62250474e-01 6.78580344e-01 -8.86000514e-01 6.00365162e-01 -7.65448451e-01 -6.04767561e-01 2.87259873e-02 -4.28299367e-01 4.71836478e-01 2.57734597e-01 -2.95655072e-01 -1.11711133e+00 -5.35586439e-02 -6.08533084e-01 -4.46763068e-01 5.30695394e-02 5.29322267e-01 1.37411142e-02 -2.56765872e-01 9.71332014e-01 2.12661490e-01 2.69847084e-03 -3.58171687e-02 2.70097643e-01 2.15072304e-01 7.02730715e-01 -6.70042574e-01 1.86517060e-01 5.96632004e-01 2.61301577e-01 -4.37052608e-01 -8.55710089e-01 1.01761274e-01 -3.71333152e-01 -3.29176515e-01 1.02641702e+00 -7.84741879e-01 -3.76296490e-01 2.66495734e-01 -1.01720524e+00 -3.63568723e-01 -5.83860219e-01 9.21001196e-01 -1.14245844e+00 4.84060138e-01 -5.01886606e-01 -6.65253758e-01 -3.49859953e-01 -1.61434937e+00 9.56174910e-01 -1.65528357e-01 -1.51051760e-01 -7.71290123e-01 1.79719999e-01 9.36002806e-02 3.29887748e-01 5.50774455e-01 9.08765435e-01 -1.35901615e-01 -4.46513623e-01 -1.56040058e-01 -3.32915723e-01 3.71809006e-01 -2.41106033e-01 -5.36040723e-01 -1.09197581e+00 -1.28047407e-01 9.20403898e-01 -6.51261687e-01 6.97988510e-01 1.19558179e+00 1.50501025e+00 6.62943348e-02 -2.38671258e-01 7.69012690e-01 1.54145193e+00 1.66632801e-01 8.23172331e-01 -2.57624954e-01 2.45297387e-01 3.20564434e-02 5.99949583e-02 4.56123918e-01 -1.77923635e-01 2.39401534e-01 5.58038116e-01 3.64085436e-01 -2.97714043e-02 -2.78991580e-01 -2.48829965e-02 3.45347911e-01 5.59589379e-02 6.32505715e-02 -8.81063223e-01 2.01354310e-01 -1.52816772e+00 -9.16420281e-01 -2.10487228e-02 2.36489081e+00 1.05275083e+00 1.34229228e-01 -9.82194915e-02 -2.77806967e-02 6.82094634e-01 -3.12569082e-01 -9.44774032e-01 7.26572946e-02 3.08106810e-01 5.82704782e-01 6.47696972e-01 4.69280005e-01 -7.36595392e-01 2.28232473e-01 7.45405197e+00 8.35157394e-01 -9.55120385e-01 7.64843464e-01 8.79480720e-01 2.78685335e-03 -4.83109564e-01 -2.71619648e-01 -1.22948684e-01 7.36968398e-01 1.20854425e+00 -1.08384239e-02 3.78885806e-01 5.38484395e-01 2.00721219e-01 -6.98605120e-01 -1.28679383e+00 1.01237965e+00 -3.04805040e-01 -1.23096681e+00 -2.42132336e-01 -6.27639443e-02 6.98745430e-01 2.89813608e-01 1.70464516e-01 4.87611592e-02 2.88470119e-01 -1.39226854e+00 1.00398910e+00 8.89025629e-01 1.22804582e+00 -8.39160323e-01 6.79036856e-01 5.95860243e-01 -3.37567598e-01 3.13855737e-01 -4.00607616e-01 6.13385320e-01 6.53504491e-01 1.01673460e+00 -1.03241706e+00 3.47220331e-01 4.68919665e-01 5.65710813e-02 2.76866913e-01 1.05224299e+00 -5.37946343e-01 5.35553813e-01 -5.74865043e-01 2.56500095e-01 2.13875696e-01 -2.93433785e-01 5.49855590e-01 8.27701867e-01 5.02171576e-01 3.15607131e-01 -2.11136669e-01 1.59397817e+00 -1.08310185e-01 -3.63382071e-01 -4.23505485e-01 -6.85781315e-02 2.11313128e-01 8.51634681e-01 -8.53944719e-01 -1.39599979e-01 3.77136618e-02 1.06868660e+00 1.39975667e-01 2.61194497e-01 -1.34084773e+00 2.67500162e-01 -5.07884249e-02 3.26236725e-01 -1.34674847e-01 6.83518723e-02 -3.06564450e-01 -1.05466557e+00 -3.39201927e-01 -4.33058858e-01 6.24923818e-02 -1.18217123e+00 -1.23703623e+00 8.50048423e-01 5.02737343e-01 -8.03687453e-01 -5.87657034e-01 -3.43118131e-01 -4.40971076e-01 1.21692991e+00 -1.35370469e+00 -7.30454266e-01 2.33226679e-02 4.48962033e-01 -7.14354217e-02 1.85769141e-01 9.32415366e-01 2.40954652e-01 -1.45542309e-01 1.60256714e-01 2.80443370e-01 -1.92007288e-01 2.90708154e-01 -1.34843993e+00 -3.01808305e-02 6.80591762e-01 -8.00450295e-02 1.62708551e-01 8.60000610e-01 -7.90485740e-01 -8.55850160e-01 -1.01477504e+00 1.92665219e-01 -3.32128823e-01 6.95354640e-01 2.10740387e-01 -7.88246214e-01 7.09978640e-01 -9.71540511e-02 5.06985486e-01 4.18504030e-01 -7.51455665e-01 1.38914198e-01 2.28017583e-01 -1.91110075e+00 2.12664679e-01 3.94839674e-01 -7.07738340e-01 -7.19077766e-01 5.07173896e-01 3.95986259e-01 -7.44198322e-01 -9.65300918e-01 3.75414014e-01 5.09975731e-01 -8.94494772e-01 1.03905106e+00 -3.49916637e-01 3.98397684e-01 -2.58698575e-02 -5.34351580e-02 -1.54636502e+00 -1.35272324e-01 -2.56886989e-01 -6.45694509e-02 2.02548027e-01 2.59150118e-01 -5.50744236e-01 8.63176346e-01 8.44018698e-01 -1.53580621e-01 -7.04558849e-01 -1.33516729e+00 -6.12563550e-01 3.72315675e-01 -7.97515512e-01 4.49623108e-01 5.13538003e-01 -3.32268149e-01 -8.00176784e-02 -2.55416691e-01 1.87016681e-01 1.19472325e+00 -3.49355787e-01 -8.71139020e-02 -8.61684084e-01 -5.37657380e-01 -1.68011665e-01 -1.83093190e-01 -7.28639066e-01 1.18686311e-01 -9.96370137e-01 4.39179063e-01 -1.50327671e+00 3.95819157e-01 -5.46167195e-01 -3.18390608e-01 -8.40849802e-02 2.60251284e-01 3.12039763e-01 -3.53335679e-01 2.29782239e-01 9.23307613e-02 6.47476017e-01 1.48971069e+00 -4.86441702e-02 3.06787997e-01 1.94027230e-01 -2.09825292e-01 9.74638522e-01 6.17112935e-01 -1.07308125e+00 -5.46468019e-01 -1.18107527e-01 3.50460738e-01 8.04666162e-01 7.48773634e-01 -1.01307487e+00 1.31846974e-02 1.13694556e-01 8.88330698e-01 -5.47939241e-01 4.57367897e-01 -9.54481840e-01 8.32678974e-01 7.35208988e-01 -2.25156397e-01 -3.92511278e-01 1.00195773e-01 7.28889823e-01 -3.57968174e-02 -9.22955573e-01 1.40400970e+00 -6.59035921e-01 -9.59159657e-02 5.85853577e-01 -4.47950125e-01 2.13104904e-01 7.16216445e-01 2.39912182e-01 4.79935050e-01 -4.92785752e-01 -1.37647402e+00 -3.79770875e-01 8.19930658e-02 -6.70320749e-01 8.83967876e-01 -1.58572352e+00 -7.65738964e-01 1.15114883e-01 -3.39917153e-01 8.93325508e-02 3.76963794e-01 1.05256796e+00 -8.53482068e-01 2.13772759e-01 -2.70903021e-01 -8.56949925e-01 -1.36786729e-01 2.79201984e-01 9.18516874e-01 -3.84196967e-01 -5.71474254e-01 8.76133263e-01 1.30443826e-01 -1.57888949e-01 -1.07114695e-01 -5.61914027e-01 4.03355688e-01 -3.26828361e-01 4.49448347e-01 2.00777516e-01 1.74149647e-01 -4.81155574e-01 -8.23856369e-02 2.73292828e-02 3.22435230e-01 -7.92158842e-01 1.11197579e+00 1.48787955e-02 -1.23225208e-02 3.20042580e-01 1.03279293e+00 -4.88893211e-01 -1.32773435e+00 -9.22748968e-02 -2.96914101e-01 -3.03665578e-01 7.65387595e-01 -1.08596623e+00 -1.03628576e+00 9.82872307e-01 7.44907677e-01 -1.20714225e-01 7.75825322e-01 -4.54665758e-02 7.46104181e-01 1.82580262e-01 6.78919613e-01 -8.31058860e-01 1.12577878e-01 -8.87776911e-02 1.15430987e+00 -1.30872226e+00 2.76578605e-01 -4.94255461e-02 -5.39855719e-01 9.03357089e-01 1.33547157e-01 -1.51582614e-01 9.35821831e-01 5.23065627e-01 -2.88974553e-01 -1.36232436e-01 -9.68493819e-02 4.86160040e-01 6.68688118e-02 7.16583431e-01 2.31784016e-01 3.70179892e-01 -1.98934853e-01 7.71043539e-01 1.39972016e-01 3.36296618e-01 4.21410471e-01 6.38754904e-01 -3.06073368e-01 -8.61438453e-01 -5.45032680e-01 5.68247557e-01 -6.94401264e-01 -2.86308467e-01 3.87206912e-01 5.27166247e-01 -1.45175666e-01 1.91802219e-01 -9.72779244e-02 4.05283958e-01 -1.13794588e-01 1.47586673e-01 1.15686870e+00 -4.55147177e-01 -7.08751753e-02 2.67137438e-01 -2.56858140e-01 -5.35654306e-01 -3.23364228e-01 -6.64636254e-01 -1.44330537e+00 -1.42417662e-02 -3.15952539e-01 2.13131696e-01 8.93938422e-01 9.03506935e-01 -3.29967409e-01 7.27150798e-01 3.70216966e-01 -8.39670718e-01 -1.08048487e+00 -9.81830835e-01 -9.35760021e-01 -1.20123485e-02 1.55277476e-01 -7.78066337e-01 -3.04585129e-01 -2.92798609e-01]
[13.559350967407227, -2.295027494430542]
552534ef-f985-4502-bbad-52d8665a45d5
disco-efficient-unsupervised-decoding-for
2107.05380
null
https://arxiv.org/abs/2107.05380v2
https://arxiv.org/pdf/2107.05380v2.pdf
DISCO : efficient unsupervised decoding for discrete natural language problems via convex relaxation
In this paper we study test time decoding; an ubiquitous step in almost all sequential text generation task spanning across a wide array of natural language processing (NLP) problems. Our main contribution is to develop a continuous relaxation framework for the combinatorial NP-hard decoding problem and propose Disco - an efficient algorithm based on standard first order gradient based. We provide tight analysis and show that our proposed algorithm linearly converges to within $\epsilon$ neighborhood of the optima. Finally, we perform preliminary experiments on the task of adversarial text generation and show superior performance of Disco over several popular decoding approaches.
['Rudrajit Das', 'Anish Acharya']
2021-07-07
null
null
null
null
['adversarial-text']
['adversarial']
[ 7.11245537e-01 3.06722045e-01 1.95495725e-01 -2.43588611e-01 -1.59871924e+00 -9.71291006e-01 3.16816688e-01 1.07330099e-01 -4.22123760e-01 1.40221751e+00 1.19230285e-01 -6.68190837e-01 -9.55660269e-02 -6.21742547e-01 -9.99825776e-01 -5.59454322e-01 -6.28895685e-02 8.86463583e-01 -4.95795608e-02 -5.04991293e-01 4.76699531e-01 1.53778449e-01 -8.58247042e-01 1.10693231e-01 9.50545371e-01 8.55438948e-01 -4.08109426e-02 1.18141556e+00 1.81711395e-03 3.11603814e-01 -9.52658236e-01 -9.73646402e-01 6.03948712e-01 -7.46545076e-01 -9.24039185e-01 -7.02959821e-02 2.35703021e-01 -1.31311461e-01 -2.81969458e-01 1.41918230e+00 9.84607995e-01 2.87359506e-02 5.59680760e-01 -1.29828870e+00 -4.36970353e-01 1.05315447e+00 -3.62612098e-01 3.95765662e-01 7.48369813e-01 -7.59234652e-02 1.07798254e+00 -5.01294971e-01 5.56085289e-01 1.07038498e+00 5.74215949e-01 6.60579503e-01 -8.88289988e-01 -4.26036179e-01 4.35654484e-02 -5.27275428e-02 -1.23513222e+00 -4.30813253e-01 4.45464820e-01 5.15283979e-02 1.14306200e+00 5.75026393e-01 4.07626033e-01 1.04879558e+00 3.49702805e-01 1.14903080e+00 1.23293364e+00 -7.18510330e-01 1.84713468e-01 -1.07355103e-01 -1.34485245e-01 8.74429762e-01 2.88052291e-01 -7.22860694e-02 -4.59423572e-01 -4.81765658e-01 1.51280746e-01 -7.30090976e-01 -3.95946354e-01 3.94067347e-01 -1.07231867e+00 9.10156131e-01 -2.67222703e-01 4.27510887e-02 3.19936946e-02 5.78429937e-01 2.48016074e-01 6.74818814e-01 6.17887497e-01 4.23830062e-01 -3.92556310e-01 -5.85592508e-01 -1.09405243e+00 5.05786240e-01 1.39840877e+00 1.27111018e+00 7.36764967e-02 2.75768876e-01 -4.99080747e-01 8.24017107e-01 8.95893350e-02 6.07671320e-01 4.34768647e-01 -5.87432206e-01 1.10681510e+00 -2.09030554e-01 1.14012361e-01 -7.04897225e-01 7.24237561e-02 -5.20924866e-01 -7.32301235e-01 -3.24659884e-01 3.06585461e-01 -6.83365464e-01 -8.35348248e-01 1.74278593e+00 1.71528205e-01 3.17107476e-02 -5.91040924e-02 6.63284600e-01 1.78154871e-01 9.37807798e-01 -4.82670099e-01 -5.84295392e-01 8.73990893e-01 -9.97993171e-01 -7.54792392e-01 -3.71182859e-01 5.89325905e-01 -9.82770264e-01 6.38691247e-01 6.75514340e-01 -1.54786944e+00 4.38862778e-02 -1.12487268e+00 8.18650424e-02 -1.44300282e-01 -1.40388176e-01 7.72054970e-01 1.08394551e+00 -1.31092644e+00 6.21380687e-01 -2.87617207e-01 -4.07898836e-02 2.18244597e-01 6.03322566e-01 4.74312641e-02 -2.27823764e-01 -1.08872581e+00 4.26496685e-01 6.55960977e-01 2.68548250e-01 -8.48450959e-01 -3.36435735e-01 -7.47650504e-01 -1.74295425e-01 5.38471282e-01 -7.72798359e-01 1.55267835e+00 -5.79286277e-01 -1.76949215e+00 6.39041364e-01 -4.47105259e-01 -8.57530892e-01 9.18425381e-01 -2.58343726e-01 -1.56380877e-01 -2.08702013e-02 6.15226571e-04 4.51622814e-01 8.30569029e-01 -9.35534954e-01 -6.29599094e-01 -1.09875821e-01 2.41302606e-02 4.05312836e-01 -1.92713052e-01 7.91133940e-03 -4.56038415e-01 -9.72095668e-01 1.95490289e-02 -9.75924850e-01 -5.70011675e-01 -4.57273215e-01 -8.08618307e-01 -2.34486148e-01 2.42340848e-01 -8.31097960e-01 1.45299470e+00 -1.75204074e+00 3.91651571e-01 3.57756108e-01 4.85575013e-02 2.93249451e-02 -1.38819784e-01 7.18624473e-01 1.56445071e-01 2.29107365e-01 -5.53763270e-01 -3.62966329e-01 6.48935437e-01 1.66848630e-01 -6.75585926e-01 5.19704640e-01 -1.76466972e-01 1.20313823e+00 -8.84192348e-01 -5.81600904e-01 -3.67646515e-01 -2.86594987e-01 -5.50422966e-01 -5.95259853e-02 -6.79971337e-01 -2.19471350e-01 -4.36763227e-01 6.02775931e-01 6.90312326e-01 7.21082315e-02 1.57607913e-01 5.28944075e-01 3.11302722e-01 2.73906738e-01 -1.08830965e+00 1.69736433e+00 -3.70754242e-01 6.18620813e-01 1.11013137e-01 -1.18097556e+00 5.49499989e-01 3.62390727e-01 1.87799111e-01 -3.67747039e-01 4.05224532e-01 3.26734662e-01 -1.49133518e-01 -1.85387596e-01 6.95992470e-01 -2.00647831e-01 -3.41314971e-01 6.70554817e-01 -1.16282463e-01 -4.04015392e-01 5.66755593e-01 5.14112949e-01 1.23869240e+00 -2.42205933e-01 1.74563497e-01 -1.86318099e-01 4.14752424e-01 -3.95838022e-02 4.17480528e-01 1.47247458e+00 -1.01731256e-01 6.23203933e-01 8.23486149e-01 9.80018154e-02 -1.08680105e+00 -8.76904488e-01 2.46904306e-02 1.01629698e+00 -1.57936469e-01 -3.49891692e-01 -9.56148446e-01 -8.51469457e-01 -2.26798996e-01 7.38542020e-01 -4.79696661e-01 1.54392555e-01 -6.77404642e-01 -1.24803185e+00 8.79196405e-01 2.82004386e-01 4.05032575e-01 -8.44334304e-01 1.60253599e-01 3.70067388e-01 -2.33810976e-01 -1.32466960e+00 -8.78570437e-01 1.62174046e-01 -7.90149033e-01 -4.82502013e-01 -8.06616187e-01 -1.07437956e+00 6.92398787e-01 -8.78168568e-02 1.05223000e+00 7.31650442e-02 -2.57072538e-01 3.91220152e-01 -4.21749800e-01 -5.03273249e-01 -7.62352407e-01 1.57757103e-01 -2.90668011e-01 -2.09597006e-01 -1.47666587e-02 -4.12404805e-01 -1.66237652e-01 -5.79184629e-02 -9.86082256e-01 -1.58195384e-03 4.72703576e-01 1.00010371e+00 6.77373469e-01 2.36607596e-01 4.87135172e-01 -1.19804037e+00 1.29025412e+00 -5.89515746e-01 -1.00889075e+00 4.01892692e-01 -5.89127302e-01 4.38819319e-01 8.58261287e-01 -2.24914908e-01 -5.58762431e-01 5.82450926e-02 -5.69562972e-01 2.82909244e-01 2.51503050e-01 5.34353375e-01 -2.18264848e-01 -1.40673503e-01 4.18626368e-01 6.90908909e-01 -3.78410697e-01 -4.66926135e-02 1.88853279e-01 6.66006148e-01 3.88182878e-01 -8.69015276e-01 1.07363963e+00 2.65347779e-01 9.67010111e-02 -5.22108555e-01 -1.05812991e+00 -2.04273075e-01 8.03505182e-02 1.98355205e-02 4.83202875e-01 -5.51501989e-01 -6.01486683e-01 3.00781012e-01 -1.11815393e+00 -5.04125774e-01 -2.70112120e-02 1.01177752e-01 -7.81984627e-01 7.54644990e-01 -6.43318713e-01 -8.22342634e-01 -7.87982821e-01 -1.07171834e+00 1.21224332e+00 4.53363732e-02 1.86332852e-01 -1.19403625e+00 3.12467009e-01 5.31648934e-01 1.26824647e-01 2.57220596e-01 7.99174249e-01 -7.37327337e-01 -4.64705169e-01 -5.04394233e-01 1.26562854e-02 3.13305318e-01 -5.36731660e-01 -4.17935967e-01 -5.16075730e-01 -5.74391723e-01 7.55493194e-02 -3.55741054e-01 8.50787103e-01 2.48899341e-01 1.22823513e+00 -5.37131011e-01 -2.91974008e-01 6.95377529e-01 1.63917410e+00 4.47169878e-02 6.12596929e-01 2.43754700e-01 3.75222057e-01 1.33826777e-01 4.50820476e-01 5.97616792e-01 3.88348009e-03 3.25950563e-01 9.49793011e-02 2.31776595e-01 1.64120212e-01 -3.59197736e-01 5.20607471e-01 1.04674327e+00 3.30048919e-01 -1.02588797e+00 -7.03746915e-01 4.96313274e-01 -1.84261477e+00 -8.89710963e-01 -5.65514751e-02 2.14192605e+00 1.12265575e+00 3.80446702e-01 -7.66323954e-02 3.86785477e-01 8.10195088e-01 3.41491252e-02 -2.67413080e-01 -9.55696166e-01 -4.46006328e-01 8.06728423e-01 8.68769884e-01 8.22054863e-01 -1.00809991e+00 1.11974490e+00 7.57421589e+00 1.30668402e+00 -5.14770806e-01 3.29121828e-01 5.31027079e-01 -2.64003158e-01 -5.06478131e-01 -1.71769336e-01 -8.53571653e-01 4.88556355e-01 9.71572220e-01 -6.77403152e-01 8.96480262e-01 5.89313388e-01 -1.97382987e-01 -5.48724793e-02 -1.14025950e+00 1.00690925e+00 4.51684415e-01 -1.11381936e+00 -5.63045070e-02 1.14116758e-01 1.30333757e+00 -1.31687954e-01 1.78190306e-01 1.74410537e-01 7.67752349e-01 -1.27218568e+00 7.49874294e-01 -1.23300411e-01 8.44427943e-01 -9.10922647e-01 7.25610137e-01 4.23354924e-01 -7.94366956e-01 -1.29995599e-01 -5.34912109e-01 5.77130914e-02 4.90966916e-01 8.44445944e-01 -1.04702044e+00 5.43784142e-01 1.19389713e-01 1.22295171e-01 -3.22513342e-01 1.22837341e+00 -4.79423732e-01 8.77670646e-01 -5.11523604e-01 -5.63270748e-01 4.82170582e-01 3.59225310e-02 9.05074179e-01 1.54470193e+00 4.54022735e-01 3.06553662e-01 1.28102541e-01 2.18023524e-01 -6.27094686e-01 2.13174492e-01 -5.23253202e-01 -3.05346429e-01 2.78806925e-01 8.77588749e-01 -8.08010221e-01 -2.84341842e-01 1.17806971e-01 1.42436230e+00 2.86088854e-01 2.64846325e-01 -1.20099533e+00 -7.05122054e-01 1.33911833e-01 -4.25867349e-01 5.36963999e-01 -3.58115286e-01 -4.39824492e-01 -1.20113564e+00 4.63838369e-01 -1.18049073e+00 4.62084264e-01 -3.82471025e-01 -1.16788268e+00 4.69145924e-01 -1.46836013e-01 -8.32441628e-01 -4.33644414e-01 -5.43474495e-01 -4.67835039e-01 8.32274258e-01 -1.18871987e+00 -4.98315334e-01 2.43837699e-01 4.48993474e-01 7.94196546e-01 -2.46808559e-01 5.94960570e-01 1.37304172e-01 -4.72026765e-01 1.24178720e+00 6.80992961e-01 4.59773839e-03 2.89323837e-01 -1.47222185e+00 6.02612674e-01 1.31008685e+00 3.65186244e-01 1.04622945e-01 9.13319767e-01 -6.94132805e-01 -1.79328060e+00 -9.26754236e-01 1.05852556e+00 -3.04217070e-01 7.44111121e-01 -5.47231615e-01 -1.81078196e-01 5.13111353e-01 2.05400646e-01 -5.17078459e-01 5.05460918e-01 -3.48896116e-01 3.10064536e-02 1.24356441e-01 -1.23456931e+00 7.91638374e-01 1.30013847e+00 -2.89429694e-01 -3.79135311e-01 1.07497895e+00 7.43143260e-01 -9.79129910e-01 -3.30493182e-01 1.07792273e-01 2.88160443e-01 -4.13595110e-01 5.94165742e-01 -4.96587902e-01 5.82577348e-01 5.17853498e-02 -2.32920229e-01 -1.20549691e+00 1.38893083e-01 -1.63614106e+00 -2.59525657e-01 9.55842912e-01 8.06737244e-01 -7.56972015e-01 9.88166690e-01 2.74294734e-01 -1.85581669e-01 -1.00801301e+00 -1.26341879e+00 -7.53061950e-01 2.38248438e-01 -8.21780086e-01 2.83207476e-01 4.16456759e-01 3.67395654e-02 2.72262633e-01 -5.84200978e-01 1.48521602e-01 8.49094808e-01 -1.17298566e-01 4.62304473e-01 -4.68552858e-01 -9.87574995e-01 -3.08181345e-01 -2.52169937e-01 -1.40276980e+00 2.91184753e-01 -1.20953810e+00 4.16570216e-01 -1.58870375e+00 1.73644647e-01 -1.50087625e-01 -5.11783734e-02 2.47852087e-01 -2.62003154e-01 4.19370890e-01 5.81227019e-02 -4.63799536e-01 -8.00204515e-01 3.29019397e-01 1.15675402e+00 -3.84263813e-01 1.06010765e-01 2.24083886e-01 -8.18580210e-01 1.47910893e-01 9.53291714e-01 -8.25550616e-01 -4.32102919e-01 -1.02102280e+00 7.03042567e-01 3.50922376e-01 -6.09810427e-02 -8.54236245e-01 2.06610590e-01 4.57414053e-02 2.39523896e-03 -4.93721068e-01 4.75114509e-02 -3.49446803e-01 -2.34870628e-01 6.17306471e-01 -5.20837665e-01 4.69097912e-01 2.13647932e-01 7.56988227e-01 4.82235253e-02 -8.56618762e-01 4.63253975e-01 -1.20061608e-02 -5.58557659e-02 2.89749354e-01 -4.76116240e-01 6.68742716e-01 8.91090512e-01 3.19533572e-02 -3.16036373e-01 -7.02141106e-01 -5.17364085e-01 4.11155045e-01 -3.28668468e-02 3.38235050e-02 6.90600991e-01 -8.96767497e-01 -1.05910289e+00 -2.65052468e-01 -4.79112059e-01 -2.44618937e-01 -9.59376842e-02 5.79208255e-01 -8.85027945e-01 6.46144092e-01 5.15939772e-01 -7.83105344e-02 -1.17590654e+00 2.58359790e-01 1.14794068e-01 -7.45205641e-01 -2.94800609e-01 1.28119791e+00 -4.21988964e-01 -8.63158852e-02 4.04435873e-01 8.43044892e-02 4.86282527e-01 -5.53830683e-01 4.17676866e-01 3.80156547e-01 1.45852715e-01 -1.53815448e-01 -2.95795724e-02 3.65791321e-02 -3.02492589e-01 -8.48211229e-01 1.20235264e+00 1.07667051e-01 -4.21347678e-01 6.70439377e-02 1.21370804e+00 8.82046744e-02 -6.57553554e-01 6.42477795e-02 -1.19228885e-01 -2.74255216e-01 -1.17276639e-01 -1.01279628e+00 -6.11430526e-01 6.33378804e-01 3.97206724e-01 2.11507499e-01 1.17534292e+00 -2.25542754e-01 1.24383795e+00 7.27523446e-01 6.44747555e-01 -1.34201467e+00 -2.23962545e-01 7.36653566e-01 6.91952705e-01 -9.13751543e-01 1.21387303e-01 -4.35152262e-01 -2.19980195e-01 9.69566166e-01 1.86714321e-01 -5.85178807e-02 2.74950504e-01 6.12857342e-01 -3.77578735e-01 3.07820022e-01 -9.62240577e-01 -6.03299588e-03 1.75234914e-01 4.48216379e-01 2.95805961e-01 2.25688629e-02 -1.03975224e+00 5.67027688e-01 -8.85935307e-01 -1.33415714e-01 6.43742681e-01 1.12354481e+00 -3.66118371e-01 -1.45295084e+00 -4.47402030e-01 5.28693616e-01 -9.97781754e-01 -6.55669451e-01 -5.56090236e-01 2.70710915e-01 -2.21189514e-01 1.10695374e+00 -3.53712201e-01 -2.40063086e-01 -1.42589971e-01 1.49256840e-01 9.24190342e-01 -4.44702119e-01 -6.14163935e-01 -7.71155134e-02 2.82384634e-01 -1.31541312e-01 2.45479986e-01 -5.02301872e-01 -1.03352082e+00 -5.29611588e-01 -4.83349591e-01 5.11721253e-01 8.22888374e-01 1.12801814e+00 -5.87418815e-03 2.28481695e-01 8.07686150e-01 -3.64157885e-01 -9.54847395e-01 -7.80627608e-01 -2.57215768e-01 -1.09757492e-02 5.77386320e-02 1.44848168e-01 -4.65974212e-01 -4.41236086e-02]
[11.972678184509277, 9.098523139953613]
63936c75-3194-46a9-a75b-7cd650d2edbb
patch-netvlad-learned-patch-descriptor-and
2202.05738
null
https://arxiv.org/abs/2202.05738v1
https://arxiv.org/pdf/2202.05738v1.pdf
Patch-NetVLAD+: Learned patch descriptor and weighted matching strategy for place recognition
Visual Place Recognition (VPR) in areas with similar scenes such as urban or indoor scenarios is a major challenge. Existing VPR methods using global descriptors have difficulty capturing local specific regions (LSR) in the scene and are therefore prone to localization confusion in such scenarios. As a result, finding the LSR that are critical for location recognition becomes key. To address this challenge, we introduced Patch-NetVLAD+, which was inspired by patch-based VPR researches. Our method proposed a fine-tuning strategy with triplet loss to make NetVLAD suitable for extracting patch-level descriptors. Moreover, unlike existing methods that treat all patches in an image equally, our method extracts patches of LSR, which present less frequently throughout the dataset, and makes them play an important role in VPR by assigning proper weights to them. Experiments on Pittsburgh30k and Tokyo247 datasets show that our approach achieved up to 6.35\% performance improvement than existing patch-based methods.
['Tiantian Feng', 'Chen Ye', 'Fenglin Zhang', 'Jiafeng Cui', 'Junqiao Zhao', 'Yingfeng Cai']
2022-02-11
null
null
null
null
['visual-place-recognition']
['computer-vision']
[-1.25815123e-01 -6.15481317e-01 -2.52616137e-01 -2.35739633e-01 -9.12050009e-01 -4.78010058e-01 6.04680836e-01 4.50045854e-01 -3.95297736e-01 6.01481378e-01 1.73636436e-01 -5.49339280e-02 -1.98577061e-01 -1.03542995e+00 -6.68866873e-01 -6.07164323e-01 -3.95925529e-02 2.85284463e-02 7.09432662e-01 -2.57313460e-01 6.09789670e-01 7.71379769e-01 -1.52048767e+00 2.40749091e-01 8.31766188e-01 1.07308900e+00 6.35181665e-01 1.18318178e-01 -1.43823653e-01 8.00927281e-01 -6.59562826e-01 5.07316776e-02 3.64379376e-01 2.85512134e-02 -5.82696319e-01 -8.95366296e-02 6.53561532e-01 -1.18532181e-01 -6.05491698e-01 9.87667859e-01 5.61672151e-01 6.27972782e-01 6.17494762e-01 -1.21766186e+00 -5.84909618e-01 6.81283176e-02 -9.69688654e-01 4.64222997e-01 4.36279058e-01 -9.31102931e-02 1.07134700e+00 -1.10284185e+00 5.06517053e-01 9.83433247e-01 8.18035305e-01 -2.07041437e-03 -1.10132599e+00 -5.63632965e-01 3.57145399e-01 5.45189202e-01 -2.17174435e+00 -3.88948441e-01 9.82766390e-01 -1.52991176e-01 9.56264317e-01 4.44013506e-01 4.74463999e-01 7.53619671e-01 1.20104596e-01 9.38697994e-01 1.02815688e+00 -2.07117468e-01 2.12030470e-01 -4.67027090e-02 1.11140609e-04 4.06132311e-01 2.66654402e-01 -3.35831493e-01 -6.21224344e-01 -2.85533190e-01 9.11462724e-01 4.17812049e-01 -4.12756890e-01 -5.47752619e-01 -1.41154754e+00 6.15971982e-01 1.01455081e+00 4.45538551e-01 -4.31328773e-01 9.89822596e-02 1.37805805e-01 -1.29793018e-01 1.56075090e-01 3.83935124e-01 -3.84085104e-02 2.62032636e-02 -9.24912691e-01 1.19306870e-01 1.40096813e-01 9.97394085e-01 1.13629961e+00 -4.18258131e-01 -3.39320540e-01 1.42077875e+00 3.85496527e-01 6.37158453e-01 5.05439341e-01 -6.67012453e-01 8.79457772e-01 8.76302719e-01 7.90935159e-02 -1.82025850e+00 -2.57104903e-01 -4.20590043e-01 -7.40683436e-01 -1.72998428e-01 1.02916390e-01 4.88644242e-01 -1.12790835e+00 1.32743740e+00 2.68646300e-01 2.72472471e-01 -4.15719710e-02 9.49943304e-01 1.02857745e+00 9.05086040e-01 -1.80661336e-01 3.11721832e-01 1.24492776e+00 -1.07311881e+00 -3.78911287e-01 -4.71570849e-01 4.27243263e-01 -6.84572399e-01 9.59095180e-01 -1.69919059e-02 -3.76618624e-01 -4.60892707e-01 -1.10084045e+00 -9.69120190e-02 -6.10908806e-01 2.25191608e-01 3.73038113e-01 3.04580092e-01 -1.05545998e+00 1.74942091e-01 -3.20695579e-01 -4.91379470e-01 4.20368761e-01 9.91655141e-03 -4.96340871e-01 -5.75936973e-01 -9.52140033e-01 6.42088950e-01 1.57812610e-01 2.17765465e-01 -8.04566503e-01 -5.46549678e-01 -9.87424612e-01 2.41692495e-02 3.08158100e-01 1.74548905e-02 8.28372121e-01 -6.00669444e-01 -8.24991465e-01 5.75573444e-01 -5.57137847e-01 -2.14057937e-01 3.60266924e-01 1.40698791e-01 -5.09496689e-01 2.06835970e-01 5.13426125e-01 6.52315438e-01 6.18316710e-01 -1.52775264e+00 -8.90396714e-01 -2.53594577e-01 1.94979057e-01 3.97455931e-01 -7.32151568e-02 -2.66321808e-01 -7.91748524e-01 -7.00766504e-01 6.70439661e-01 -7.17113674e-01 -3.72409225e-01 1.92490453e-03 -1.84717998e-01 -2.87427694e-01 7.67862260e-01 -4.38415438e-01 1.25991011e+00 -2.36702895e+00 -5.80345154e-01 6.53414786e-01 4.18128371e-01 3.27943027e-01 -2.78722912e-01 5.91539204e-01 1.20688580e-01 3.49656790e-02 -4.28067967e-02 -1.54728606e-01 -7.10052997e-02 1.76394194e-01 -3.20017934e-01 6.24880552e-01 -1.10902846e-01 7.15152979e-01 -1.06941664e+00 -6.90470457e-01 5.47244668e-01 4.46004212e-01 -3.79871905e-01 -1.47051558e-01 2.99087018e-01 1.56528413e-01 -6.69157505e-01 9.82329726e-01 1.02969956e+00 -1.56643853e-01 -1.28400281e-01 -9.94817764e-02 -3.27128381e-01 2.04261288e-01 -1.30940735e+00 1.57017148e+00 -3.12822640e-01 8.98316205e-01 -2.62986988e-01 -9.11539972e-01 1.33402407e+00 -1.42122597e-01 5.70999682e-01 -1.27351797e+00 -1.83229864e-01 2.05225870e-01 -4.49434519e-01 -7.30974302e-02 1.03649616e+00 4.84281659e-01 -9.76376235e-02 -1.75210893e-01 -3.33023041e-01 3.17307770e-01 -6.33176416e-02 5.25462814e-02 1.35268497e+00 -2.56119728e-01 4.28874731e-01 -1.69616133e-01 5.48386216e-01 2.13064313e-01 9.11397874e-01 1.04770124e+00 -5.94836950e-01 1.15781951e+00 1.12922251e-01 -5.06280541e-01 -7.46877074e-01 -9.26916897e-01 -2.73874581e-01 5.43322444e-01 8.02780509e-01 -6.58553779e-01 -1.33930698e-01 -6.49286032e-01 1.84717588e-02 4.13740396e-01 -4.87578362e-01 1.16701983e-02 -4.86869037e-01 -5.78168094e-01 4.58068401e-01 5.18401384e-01 1.01861620e+00 -8.78181159e-01 -4.25961792e-01 2.41984963e-01 -6.51159346e-01 -1.06427503e+00 -4.80139762e-01 -1.06810510e-01 -4.58389699e-01 -9.80451345e-01 -1.08427596e+00 -9.14150834e-01 8.96570742e-01 1.17277813e+00 9.03549194e-01 -9.30743515e-02 -2.45995373e-01 3.65600973e-01 -6.59985960e-01 6.02881946e-02 4.28881466e-01 1.31314561e-01 -2.60663509e-01 2.21152529e-01 4.66071695e-01 -4.77352679e-01 -9.52459097e-01 7.82131374e-01 -4.69046146e-01 -3.26430738e-01 6.07320726e-01 7.16427445e-01 9.72892582e-01 2.59531558e-01 2.01555744e-01 -3.55471313e-01 3.35590780e-01 -6.48770511e-01 -6.83037817e-01 4.05191034e-01 -3.54315579e-01 -3.86208951e-01 5.49628198e-01 -3.28521281e-01 -6.84946299e-01 1.51122764e-01 -4.99183834e-02 -4.81762469e-01 -2.02300638e-01 3.23219866e-01 -3.36863399e-01 -4.21889186e-01 6.04046762e-01 5.98654926e-01 -5.21327496e-01 -3.64598542e-01 2.32768059e-02 7.47658253e-01 8.05904791e-02 -3.65105718e-01 8.26515913e-01 4.43903893e-01 -5.55932894e-02 -1.13340497e+00 -2.95096159e-01 -1.10824871e+00 -3.68427128e-01 -2.09407151e-01 5.47404230e-01 -1.15701854e+00 -5.02249956e-01 3.05980355e-01 -8.57736826e-01 -3.71764004e-02 -1.49321407e-01 4.87749785e-01 -2.17101350e-01 3.39911491e-01 -7.08755404e-02 -7.10834563e-01 -4.60846163e-03 -1.17206991e+00 1.12524354e+00 3.84382010e-01 7.97800124e-02 -7.82413542e-01 2.58011878e-01 2.51651525e-01 5.34424245e-01 6.64259195e-02 3.35184693e-01 -4.13419008e-01 -9.20335054e-01 -3.00995737e-01 -5.66619813e-01 -1.81424636e-02 1.84913412e-01 -3.29530627e-01 -9.50691283e-01 -2.67410189e-01 -4.77764815e-01 7.65256286e-02 9.11800325e-01 2.55131721e-01 1.22103977e+00 -7.53602087e-02 -7.56733835e-01 6.11998796e-01 1.66634631e+00 4.23001379e-01 1.06976748e+00 6.16073608e-01 9.41895604e-01 1.41215265e-01 1.04040384e+00 4.34610635e-01 7.54813433e-01 1.13038588e+00 3.85763019e-01 -3.78218532e-01 -1.85359746e-01 -5.10246873e-01 2.55421877e-01 3.67847532e-01 -1.46902921e-02 -3.20621341e-01 -1.09528470e+00 1.00431585e+00 -2.04853725e+00 -9.41220939e-01 -1.62409127e-01 2.20868397e+00 3.13797653e-01 -3.23796809e-01 -1.28674373e-01 2.19856873e-02 8.15913260e-01 6.97323143e-01 -2.48240322e-01 1.76128134e-01 -2.45043412e-01 -9.99090597e-02 1.02538919e+00 2.31406063e-01 -1.31541324e+00 9.92709816e-01 5.93943501e+00 1.31805849e+00 -1.21772289e+00 -4.07454595e-02 3.79729092e-01 1.42521635e-01 -1.09529793e-01 8.67967610e-04 -8.90713453e-01 4.53445196e-01 3.00691783e-01 2.37793446e-01 2.93037832e-01 1.05893171e+00 1.00092098e-01 -4.65841204e-01 -4.96245652e-01 1.45888484e+00 2.11492538e-01 -1.09506226e+00 5.96310245e-03 2.07567304e-01 7.89551914e-01 2.50675440e-01 9.59322378e-02 2.13132799e-01 8.94955099e-02 -1.00254273e+00 6.62854970e-01 2.56684661e-01 4.74389941e-01 -8.25388014e-01 7.51028121e-01 2.10720673e-01 -1.79155302e+00 -1.53213590e-01 -9.20339227e-01 2.53711253e-01 -1.46630198e-01 5.41500151e-01 -9.10642385e-01 5.94185948e-01 9.08481896e-01 1.02770877e+00 -9.71860707e-01 1.66986156e+00 -1.99847609e-01 3.22602451e-01 -4.02770072e-01 -8.95192381e-03 3.63491565e-01 2.70645302e-02 5.93467772e-01 1.02577901e+00 4.62840945e-01 -6.57749549e-02 4.30381328e-01 6.84107006e-01 3.51882204e-02 3.66551429e-01 -9.59142208e-01 4.48935866e-01 8.60009611e-01 1.28767025e+00 -9.58094656e-01 -1.10867471e-01 -2.47860909e-01 9.72973347e-01 4.32835996e-01 4.00225997e-01 -9.23303545e-01 -6.73272073e-01 8.14741313e-01 2.41675228e-01 6.26935720e-01 -4.12335962e-01 1.67016499e-02 -1.27070928e+00 2.38802001e-01 -6.33679330e-01 2.07291678e-01 -7.03559577e-01 -1.15560651e+00 4.99474406e-01 -1.47213385e-01 -1.74915814e+00 4.01744992e-01 -3.25711787e-01 -4.39687580e-01 6.49159491e-01 -1.87866163e+00 -1.12427974e+00 -6.09194160e-01 8.35751414e-01 4.43670541e-01 1.13400891e-01 4.80045229e-01 7.04361558e-01 -5.54257214e-01 7.68126547e-01 4.29373711e-01 3.50212097e-01 6.72590077e-01 -9.28834200e-01 3.70641470e-01 1.02385616e+00 3.35525125e-01 7.70606458e-01 7.33601898e-02 -5.91358423e-01 -1.19774711e+00 -1.40384555e+00 9.45161760e-01 -3.29549372e-01 1.68086812e-01 -3.56372684e-01 -6.83420420e-01 3.16514283e-01 -1.63736016e-01 2.99040169e-01 4.28886831e-01 -1.92101358e-03 -4.97953176e-01 -5.45841038e-01 -1.29311192e+00 7.14745343e-01 1.27765536e+00 -6.53227091e-01 -5.02128422e-01 9.36848149e-02 3.20744008e-01 -3.90484393e-01 -5.39886475e-01 2.62733817e-01 2.35657692e-01 -8.97747755e-01 1.28788877e+00 3.56178969e-01 5.93349747e-02 -9.16743517e-01 -5.46081126e-01 -1.34286439e+00 -5.74461818e-01 -4.82837902e-03 2.78684735e-01 1.47853255e+00 2.21815377e-01 -9.05544519e-01 5.58341920e-01 3.20824504e-01 -8.19737166e-02 -3.86677742e-01 -1.01420379e+00 -9.08840656e-01 -4.94156957e-01 -3.44247788e-01 8.18833768e-01 1.01784003e+00 -3.00491571e-01 -6.19174391e-02 -3.11981887e-01 5.00099897e-01 5.49710512e-01 1.75958350e-01 8.03423762e-01 -1.01197863e+00 3.90695959e-01 -3.26965600e-01 -1.06460726e+00 -1.32350683e+00 -1.52803034e-01 -8.46753776e-01 3.45717996e-01 -1.92493534e+00 1.94427371e-01 -9.14077044e-01 -6.41232908e-01 6.70268416e-01 1.72401778e-02 6.00596607e-01 2.14407846e-01 5.76675951e-01 -1.01677060e+00 6.10193193e-01 9.54909086e-01 -3.99362803e-01 -3.04523587e-01 -1.77167848e-01 -5.31074047e-01 4.43467677e-01 7.61471033e-01 -4.58664060e-01 -4.95957822e-01 -3.69128495e-01 -4.81963390e-03 -3.84797901e-01 6.04189515e-01 -1.48746836e+00 4.61719751e-01 -2.67262429e-01 5.95449746e-01 -1.07412887e+00 5.36622524e-01 -1.14182484e+00 1.72861308e-01 1.10082395e-01 -1.11993039e-02 1.12743579e-01 5.06115444e-02 6.50150359e-01 -5.16556799e-01 -7.87882786e-03 5.95663130e-01 -7.37855863e-03 -1.41867733e+00 2.20780864e-01 -4.27740932e-01 -2.12897826e-03 1.32211149e+00 -4.81604576e-01 -4.13040251e-01 -3.17384511e-01 -1.19869173e-01 3.78842175e-01 6.64984882e-01 4.86400902e-01 9.77718532e-01 -1.49405575e+00 -3.27934772e-01 1.85220107e-01 6.76788867e-01 1.42552629e-01 3.81475896e-01 7.44855344e-01 -7.16903508e-01 5.11979699e-01 -1.62662357e-01 -7.33096898e-01 -1.27231967e+00 4.11067724e-01 1.91168830e-01 -2.38242839e-02 -8.17021489e-01 6.36322320e-01 3.10152233e-01 -4.43534702e-01 2.03465044e-01 -3.65635842e-01 -4.87229437e-01 2.17920125e-01 5.49365938e-01 3.85238290e-01 1.23711437e-01 -1.17858386e+00 -9.05255675e-01 9.06732261e-01 -8.55828598e-02 1.12749346e-01 1.17497516e+00 -3.61529082e-01 1.28037542e-01 2.59649366e-01 1.32798731e+00 3.31810266e-01 -1.05723500e+00 -4.26423877e-01 -1.36090234e-01 -9.48848784e-01 1.76403314e-01 -5.36920488e-01 -1.14462602e+00 6.38947904e-01 6.74700499e-01 -8.73586684e-02 9.79594052e-01 3.10286530e-03 7.68316269e-01 4.47588414e-01 9.83503401e-01 -1.03124416e+00 -6.46265373e-02 6.84648871e-01 5.68401933e-01 -1.28258896e+00 1.06611371e-03 -4.12393719e-01 -5.79019189e-01 7.71229386e-01 5.65593958e-01 -2.35451043e-01 4.58338708e-01 -3.57006162e-01 1.45724639e-01 -1.74051538e-01 -1.20805122e-01 -3.27452630e-01 4.16400731e-01 7.84456551e-01 -1.62342370e-01 9.37697813e-02 -1.25316605e-01 2.19711095e-01 1.12985715e-01 -2.76076764e-01 1.52016595e-01 1.22759008e+00 -4.89850760e-01 -8.95128310e-01 -6.30586863e-01 2.83261418e-01 8.78002401e-03 -7.85290599e-02 -1.84233576e-01 6.56732857e-01 1.42791763e-01 9.78332698e-01 3.04859169e-02 -5.74755847e-01 5.38587749e-01 -4.12620842e-01 1.10374168e-01 -3.67720008e-01 -3.58594745e-01 -1.77022785e-01 -1.21538222e-01 -8.68879080e-01 -2.14928076e-01 -6.03584349e-01 -1.14229548e+00 -3.14609289e-01 -2.22336844e-01 7.07516670e-02 5.42840183e-01 6.08958483e-01 7.12588787e-01 3.74217600e-01 9.13843393e-01 -8.80403161e-01 -2.74043847e-02 -5.41244149e-01 -7.50970781e-01 1.46302104e-01 3.71071845e-01 -9.01160955e-01 -2.40250513e-01 -4.17292327e-01]
[7.626209735870361, -1.868905782699585]
04ba02e1-27a3-4069-9c71-a235e31532af
comparative-study-of-subset-selection-methods
2306.17551
null
https://arxiv.org/abs/2306.17551v1
https://arxiv.org/pdf/2306.17551v1.pdf
Comparative study of subset selection methods for rapid prototyping of 3D object detection algorithms
Object detection in 3D is a crucial aspect in the context of autonomous vehicles and drones. However, prototyping detection algorithms is time-consuming and costly in terms of energy and environmental impact. To address these challenges, one can check the effectiveness of different models by training on a subset of the original training set. In this paper, we present a comparison of three algorithms for selecting such a subset - random sampling, random per class sampling, and our proposed MONSPeC (Maximum Object Number Sampling per Class). We provide empirical evidence for the superior effectiveness of random per class sampling and MONSPeC over basic random sampling. By replacing random sampling with one of the more efficient algorithms, the results obtained on the subset are more likely to transfer to the results on the entire dataset. The code is available at: https://github.com/vision-agh/monspec.
['Tomasz Kryjak', 'Konrad Lis']
2023-06-30
null
null
null
null
['3d-object-detection', 'object-detection', 'autonomous-vehicles']
['computer-vision', 'computer-vision', 'computer-vision']
[ 4.49729860e-02 -2.08816484e-01 -5.86147718e-02 -1.01609729e-01 -4.19599831e-01 -7.51562893e-01 7.48504281e-01 -7.61235952e-02 -6.17773294e-01 6.35381460e-01 -4.75846708e-01 -2.00257823e-01 1.33457839e-01 -9.76625919e-01 -7.95509815e-01 -7.19305456e-01 -1.66444197e-01 5.73554575e-01 7.09810495e-01 1.70040399e-01 1.21978484e-01 8.28298569e-01 -2.07407236e+00 -1.01685308e-01 6.35975838e-01 8.45434844e-01 4.41804588e-01 7.86914647e-01 3.53253305e-01 1.72714248e-01 -5.28885722e-01 -1.08500548e-01 6.83741510e-01 -3.40396732e-01 -2.95512140e-01 9.27368477e-02 4.91129607e-01 -5.73955059e-01 -1.61077455e-02 1.09242821e+00 3.80412936e-01 2.06376821e-01 6.47087634e-01 -1.59208190e+00 2.65600652e-01 5.86856343e-02 -4.10824299e-01 1.64158419e-01 2.86031723e-01 4.48342234e-01 7.03866243e-01 -9.44176972e-01 5.97914517e-01 1.10347116e+00 5.08960903e-01 3.78835112e-01 -1.14437902e+00 -7.74401188e-01 1.35468813e-02 1.77427888e-01 -1.53341150e+00 -5.88382721e-01 6.05639815e-01 -4.79889423e-01 5.52280307e-01 3.26080233e-01 7.53589213e-01 8.30886424e-01 -2.64623631e-02 7.61667848e-01 1.12527704e+00 -3.55438352e-01 3.73315692e-01 5.58312356e-01 7.57456869e-02 7.02314496e-01 1.02740562e+00 6.88237607e-01 -2.78582215e-01 -2.52859473e-01 4.09332156e-01 -4.96474952e-02 -2.61626929e-01 -6.85411155e-01 -9.67981517e-01 8.65670323e-01 3.18933725e-01 -9.74981412e-02 -4.12697226e-01 2.19691291e-01 3.28611732e-02 1.21038117e-01 2.81701714e-01 5.34613609e-01 -2.72333592e-01 -3.12827085e-03 -9.65922117e-01 4.28475916e-01 6.04705989e-01 1.09101701e+00 9.87273395e-01 -5.13377152e-02 -1.20805157e-03 4.47075784e-01 2.87696928e-01 8.24735999e-01 2.00981926e-02 -1.02516174e+00 -6.59507886e-02 5.66125453e-01 4.36160028e-01 -6.02871835e-01 -4.53445703e-01 -3.59668672e-01 -2.19181150e-01 6.43056452e-01 4.32043910e-01 -4.58031595e-01 -9.08235610e-01 1.43387949e+00 6.60242438e-01 1.96046717e-02 -7.56160617e-02 9.43265259e-01 5.95073879e-01 6.56054378e-01 -1.08738489e-01 -8.33390430e-02 1.26705062e+00 -7.77472138e-01 -3.09365422e-01 -3.18599820e-01 5.17031729e-01 -6.38930082e-01 7.29741931e-01 3.32325757e-01 -6.18933797e-01 -4.77398753e-01 -1.25926924e+00 3.93227547e-01 -4.58485097e-01 3.95845950e-01 5.95814466e-01 7.08907425e-01 -9.03759062e-01 1.75963730e-01 -7.81166434e-01 -6.87138498e-01 3.21888119e-01 2.34905377e-01 -9.11597461e-02 -1.30963504e-01 -7.63744950e-01 9.28610027e-01 3.31486374e-01 -3.46395731e-01 -9.88026619e-01 -3.33969772e-01 -5.81894457e-01 -3.41773570e-01 4.73798186e-01 -4.07390654e-01 1.36780262e+00 -6.69591546e-01 -1.12400520e+00 6.81022167e-01 -1.49361879e-01 -6.67875528e-01 7.50570774e-01 -2.32118085e-01 3.76234837e-02 1.54603973e-01 2.96892300e-02 8.39933872e-01 8.41972470e-01 -1.47577298e+00 -8.98805737e-01 -2.17644811e-01 1.81445464e-01 -2.98126787e-03 1.63855627e-01 -9.31749791e-02 -3.43934715e-01 -8.19193274e-02 -2.36950755e-01 -1.07559884e+00 -1.68321848e-01 1.88826710e-01 -2.08405986e-01 -1.28303885e-01 9.42292392e-01 -1.31544679e-01 6.84854150e-01 -2.10633206e+00 -2.64412940e-01 6.66062683e-02 8.47631320e-02 3.48782003e-01 -8.40831548e-02 4.31561261e-01 2.73743570e-01 1.90824736e-02 -1.01173520e-02 -8.88261348e-02 -2.30652884e-01 -2.59992294e-02 -7.92592540e-02 7.12445080e-01 1.78509712e-01 4.54090774e-01 -9.54980433e-01 -3.19337994e-01 5.51614285e-01 3.33307117e-01 -4.90150452e-01 -4.41490784e-02 -2.39457309e-01 1.34830132e-01 -4.41738963e-01 8.29404950e-01 8.74522269e-01 3.86092649e-03 -2.26060376e-02 -2.01872930e-01 -2.43030682e-01 1.36497408e-01 -1.24911439e+00 8.17987680e-01 -1.48045197e-01 8.71312320e-01 8.37587472e-03 -5.65919995e-01 7.49032974e-01 2.76082288e-02 3.55779409e-01 -5.16758621e-01 4.04454142e-01 1.49934396e-01 1.04360707e-01 -2.61887491e-01 6.06976449e-01 9.66362432e-02 1.91187963e-01 5.04719734e-01 -1.13094829e-01 -2.79644132e-01 4.59196091e-01 1.03573717e-01 1.02944946e+00 7.57717341e-02 4.46871340e-01 -2.25598291e-01 1.15481019e-01 3.79805744e-01 3.97382289e-01 9.34299231e-01 -3.93629760e-01 4.25744623e-01 1.75500527e-01 -2.30097920e-01 -1.03626347e+00 -8.81701589e-01 -2.91583598e-01 6.44917250e-01 5.71550190e-01 -2.55219609e-01 -6.66097879e-01 -7.82252908e-01 2.53052771e-01 9.17150736e-01 -4.21561003e-01 -2.64522433e-01 -2.17148855e-01 -6.62553012e-01 2.62676775e-01 2.90385664e-01 2.73282111e-01 -8.39374721e-01 -1.42447698e+00 -1.61477581e-01 1.28421977e-01 -1.14191949e+00 -9.96102095e-02 1.61033407e-01 -5.81633270e-01 -1.20117712e+00 -4.02873546e-01 -1.55807778e-01 8.73065650e-01 7.78929532e-01 9.84874964e-01 4.18766469e-01 -3.51925343e-01 5.02382517e-01 -6.35497093e-01 -8.84640813e-01 -4.83734399e-01 -5.95433526e-02 1.59068748e-01 -6.57471418e-02 3.63921523e-01 5.85182235e-02 -5.35602570e-01 5.51095724e-01 -6.05383575e-01 1.07084699e-01 6.12030387e-01 3.96559894e-01 4.10179168e-01 1.43568099e-01 1.66034877e-01 -6.60240352e-01 2.39084110e-01 -4.10462648e-01 -1.24665010e+00 -1.76373914e-01 -6.06057048e-01 -1.05761498e-01 2.39502713e-01 -5.41771889e-01 -4.26438868e-01 3.88934612e-01 1.62763909e-01 -5.34903347e-01 -3.10874760e-01 5.82316220e-02 4.81684506e-02 -2.62151748e-01 6.72027290e-01 9.72621217e-02 8.15900862e-02 -3.25796694e-01 1.41478986e-01 4.75970209e-01 3.74723002e-02 -3.42871815e-01 9.99737203e-01 6.17096663e-01 2.72565018e-02 -1.08795953e+00 -6.62077904e-01 -5.44248104e-01 -2.67362744e-01 -6.61345422e-01 6.04614496e-01 -7.87820816e-01 -3.19543332e-01 3.65874439e-01 -9.40117180e-01 -4.05670077e-01 -3.80563915e-01 6.51549220e-01 -3.44988436e-01 1.10611901e-01 1.83737040e-01 -1.13660693e+00 -2.88895275e-02 -1.32307124e+00 1.12121272e+00 3.93226922e-01 -1.47657208e-02 -6.27670109e-01 -1.50645241e-01 8.81373733e-02 2.29757026e-01 3.72050464e-01 3.15671265e-01 -5.92274964e-01 -9.33747351e-01 -6.98517025e-01 -1.79717720e-01 2.42720768e-01 1.77332375e-03 2.96318531e-01 -1.01514161e+00 -5.58634818e-01 -2.96750784e-01 -1.19957320e-01 9.86890674e-01 4.04411495e-01 6.91349208e-01 -4.08336408e-02 -6.19107485e-01 2.71177977e-01 1.48815513e+00 2.87561059e-01 3.16839367e-01 3.91755044e-01 2.05403969e-01 4.49061990e-01 1.12210822e+00 4.80296552e-01 3.20943981e-01 7.57041872e-01 8.09945405e-01 5.95051199e-02 -2.53530353e-01 -8.92893597e-02 3.86870474e-01 -2.37137824e-01 -8.37400109e-02 -4.20583278e-01 -9.67030883e-01 7.70753086e-01 -1.58297968e+00 -1.07983029e+00 -1.21622331e-01 2.56207538e+00 2.99077749e-01 2.48193875e-01 6.17413580e-01 1.79333687e-01 8.21205199e-01 -5.81945032e-02 -3.99178416e-01 3.24416496e-02 1.33692086e-01 -4.19770271e-01 1.05321980e+00 5.38074613e-01 -1.08414483e+00 6.09644830e-01 6.38057137e+00 5.60475469e-01 -1.11043382e+00 6.70685098e-02 1.86220199e-01 -3.16736877e-01 4.98255678e-02 1.80534959e-01 -1.22544515e+00 6.25970364e-01 9.35500264e-01 -2.88587511e-01 4.36066598e-01 8.65094483e-01 2.36883298e-01 -6.79247200e-01 -1.09887171e+00 7.01930642e-01 -3.26122344e-02 -1.11084270e+00 4.32737954e-02 3.91885459e-01 6.64011836e-01 3.31828415e-01 -3.02711010e-01 1.82372883e-01 3.80325705e-01 -5.26609600e-01 9.84942615e-01 2.34062880e-01 5.19199848e-01 -5.95290959e-01 8.34113777e-01 5.07683516e-01 -9.92647231e-01 -2.58428846e-02 -3.24499816e-01 1.00136124e-01 -8.41365084e-02 6.04205489e-01 -1.17950869e+00 3.47216517e-01 7.83584177e-01 3.08655888e-01 -7.40752995e-01 1.56540751e+00 -1.17368802e-01 7.75514722e-01 -7.39266872e-01 -3.30888331e-01 -2.02152925e-03 -2.27649137e-02 1.02228034e+00 1.10761750e+00 4.15813178e-01 -2.74273485e-01 2.97891200e-01 5.76200008e-01 1.19403943e-01 -4.12065208e-01 -8.71026576e-01 4.60554324e-02 8.74009550e-01 1.27597916e+00 -9.53094125e-01 -4.11334157e-01 -3.96255404e-01 3.81239921e-01 -2.69637983e-02 2.43561044e-01 -1.01274395e+00 -3.32980543e-01 4.60309237e-01 3.80782962e-01 5.40539563e-01 -2.21611917e-01 -9.09176394e-02 -8.02555084e-01 1.32027008e-02 -5.84979177e-01 2.58625567e-01 -6.21506572e-01 -7.27191150e-01 4.65965837e-01 4.73743975e-01 -1.62050796e+00 -1.73944086e-01 -6.13499165e-01 -4.25159216e-01 4.29155499e-01 -1.54592109e+00 -8.64042461e-01 -7.13293076e-01 1.66988134e-01 5.96741676e-01 -2.20127683e-02 2.33290151e-01 2.58117586e-01 -3.64403874e-01 3.43219817e-01 -2.27165166e-02 -1.42087698e-01 3.95973831e-01 -1.01608169e+00 1.78146273e-01 1.04781771e+00 -2.75355913e-02 2.54013270e-01 1.05051196e+00 -4.72444326e-01 -1.45691228e+00 -1.20845449e+00 4.22584057e-01 -4.87829357e-01 4.34452772e-01 -3.86598349e-01 -5.10007203e-01 6.24150872e-01 -9.04312134e-02 5.27661927e-02 9.69319120e-02 -2.78946042e-01 6.36012629e-02 -1.55890971e-01 -1.41692138e+00 5.40832579e-01 8.16523731e-01 6.87470287e-02 -2.20541164e-01 2.78722078e-01 4.18617278e-01 -3.36473048e-01 -5.01087189e-01 5.39897442e-01 6.91847503e-01 -1.09236336e+00 8.35449219e-01 2.10355893e-01 -1.69684663e-01 -7.73469448e-01 -2.99074620e-01 -1.18740964e+00 -4.71618734e-02 -3.63197446e-01 -2.89767818e-03 1.02148306e+00 4.76074427e-01 -9.84680235e-01 7.34308481e-01 1.56106591e-01 -4.73338179e-02 -5.12347758e-01 -8.36205423e-01 -1.18480265e+00 -2.51195461e-01 -4.35080022e-01 5.89873016e-01 4.68539715e-01 -6.57519102e-01 2.88444068e-02 -2.26698756e-01 5.96796274e-01 7.97160625e-01 2.69138187e-01 1.21984339e+00 -1.33263814e+00 -1.12273023e-01 -2.90366828e-01 -5.79949439e-01 -8.15264285e-01 -3.70745152e-01 -4.85663295e-01 2.98815787e-01 -1.57151818e+00 2.35491306e-01 -4.88780051e-01 1.13491178e-01 4.58058268e-01 -7.69494381e-03 3.95023972e-01 3.25081766e-01 2.72906929e-01 -6.98846996e-01 3.87670279e-01 1.04004884e+00 -1.20501578e-01 -1.11555196e-01 3.52917880e-01 -4.32463586e-01 7.70887434e-01 1.04995596e+00 -6.84405446e-01 -3.17020863e-01 -1.54015616e-01 -9.85450745e-02 -1.69025958e-01 7.09542572e-01 -1.43964064e+00 1.47335187e-01 -2.73038119e-01 2.92362690e-01 -9.12084997e-01 5.39034367e-01 -1.03540802e+00 1.99774757e-01 7.30733275e-01 1.83386728e-01 -8.93537179e-02 4.48175192e-01 3.55778456e-01 1.34205773e-01 -3.65801722e-01 1.10509074e+00 -4.90492918e-02 -8.61127377e-01 1.30192652e-01 -5.39870560e-01 -1.74413249e-01 1.40509439e+00 -6.01560712e-01 -4.42318738e-01 -2.38055944e-01 -2.54805624e-01 2.43840054e-01 8.85555744e-01 4.53880429e-01 5.04487634e-01 -1.08838713e+00 -5.75722396e-01 3.17216039e-01 3.92223597e-01 -9.92185250e-02 -2.87597418e-01 8.39163840e-01 -5.31934798e-01 3.95507127e-01 -1.01741768e-01 -8.02300632e-01 -1.44862032e+00 4.08147633e-01 4.53532159e-01 2.66955554e-01 -3.79259378e-01 5.74051142e-01 9.00204852e-02 -2.65795738e-01 7.72035420e-02 -1.78341955e-01 1.02753729e-01 -1.20901413e-01 4.98873055e-01 5.85445523e-01 5.64184785e-03 -5.47849894e-01 -5.22055745e-01 5.75766146e-01 3.97456773e-02 -9.60577093e-03 9.99067783e-01 -9.23536271e-02 2.50936240e-01 2.46854469e-01 7.11563349e-01 1.07443899e-01 -1.49360228e+00 7.48350769e-02 -1.21988155e-01 -5.91680527e-01 3.01700592e-01 -5.49981296e-01 -9.31285501e-01 4.08975571e-01 8.69737446e-01 3.71765614e-01 9.94422019e-01 2.19540939e-01 1.62726328e-01 3.84161323e-01 7.48677671e-01 -7.67220080e-01 -3.72458428e-01 2.52502173e-01 8.16945493e-01 -1.31671607e+00 4.01256740e-01 -3.94840956e-01 -3.45971048e-01 6.94058478e-01 7.23436892e-01 -3.68699491e-01 6.40676916e-01 2.59200335e-01 2.10626218e-02 -2.76516080e-01 -8.99659395e-01 -6.68038309e-01 6.49517328e-02 6.56811416e-01 -7.28563890e-02 1.04853019e-01 -2.66588867e-01 -1.79510847e-01 -2.11990267e-01 5.85408509e-02 7.06125081e-01 1.08695590e+00 -7.08954096e-01 -7.41568625e-01 -5.25908291e-01 5.92807472e-01 1.43417884e-02 3.34836394e-01 -4.75953460e-01 1.08970940e+00 1.87809542e-01 1.12319434e+00 2.43469626e-02 -4.42689151e-01 3.05383593e-01 -2.95670182e-01 4.31272358e-01 -4.81836408e-01 -4.88328822e-02 -4.46964353e-02 2.69383162e-01 -4.85930622e-01 -5.60654223e-01 -1.01342618e+00 -1.06508338e+00 -2.54827738e-01 -6.49033427e-01 9.03827325e-02 8.70723903e-01 6.29999340e-01 3.91281426e-01 2.45198339e-01 6.13531470e-01 -1.27356994e+00 -4.87491488e-01 -7.41232753e-01 -4.25753623e-01 -4.02828818e-03 5.96837103e-01 -1.10098553e+00 -5.52118182e-01 -1.37630299e-01]
[8.102227210998535, -1.2249318361282349]
c9b66b9c-c35c-4c96-af6e-9087ba4aecba
grantrel-grant-information-extraction-via
null
null
https://aclanthology.org/2021.findings-acl.236
https://aclanthology.org/2021.findings-acl.236.pdf
GrantRel: Grant Information Extraction via Joint Entity and Relation Extraction
null
['Shanfeng Zhu', 'Hong Zhou', 'Xiaodi Huang', 'Li Huang', 'Junyi Bian']
null
null
null
null
findings-acl-2021-8
['joint-entity-and-relation-extraction']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.276376724243164, 3.630176067352295]
ebcbb8e3-0cf5-4fdd-94c6-c93c4febf13b
a-local-temporal-difference-code-for
null
null
http://proceedings.neurips.cc/paper/2020/hash/9dd16e049becf4d5087c90a83fea403b-Abstract.html
http://proceedings.neurips.cc/paper/2020/file/9dd16e049becf4d5087c90a83fea403b-Paper.pdf
A Local Temporal Difference Code for Distributional Reinforcement Learning
Recent theoretical and experimental results suggest that the dopamine system implements distributional temporal difference backups, allowing learning of the entire distributions of the long-run values of states rather than just their expected values. However, the distributional codes explored so far rely on a complex imputation step which crucially relies on spatial non-locality: in order to compute reward prediction errors, units must know not only their own state but also the states of the other units. It is far from clear how these steps could be implemented in realistic neural circuits. Here, we introduce the Laplace code: a local temporal difference code for distributional reinforcement learning that is representationally powerful and computationally straightforward. The code decomposes value distributions and prediction errors across three separated dimensions: reward magnitude (related to distributional quantiles), temporal discounting (related to the Laplace transform of future rewards) and time horizon (related to eligibility traces). Besides lending itself to a local learning rule, the decomposition recovers the temporal evolution of the immediate reward distribution, indicating all possible rewards at all future times. This increases representational capacity and allows for temporally-flexible computations that immediately adjust to changing horizons or discount factors.
['Alexandre Pouget', 'Peter Dayan', 'Pablo Tano']
2020-12-01
null
null
null
neurips-2020-12
['distributional-reinforcement-learning']
['methodology']
[-8.64445046e-03 -9.38683525e-02 -1.88406065e-01 -4.65959221e-01 -5.14540255e-01 -8.47672641e-01 6.44184649e-01 4.51093853e-01 -9.69801843e-01 1.08346891e+00 2.85428703e-01 -3.38338882e-01 -1.48093387e-01 -7.66560316e-01 -4.55161422e-01 -9.19517756e-01 -5.02031803e-01 4.48989809e-01 2.06121832e-01 -1.27521038e-01 5.12039483e-01 5.32674491e-01 -1.50505030e+00 2.72105873e-01 4.77876246e-01 1.09448314e+00 3.12943220e-01 5.49765885e-01 7.50088394e-02 7.67230153e-01 -3.84014219e-01 -8.15769508e-02 1.27700478e-01 -6.59657180e-01 -4.25425738e-01 -6.54565811e-01 -2.73790509e-01 -7.14965701e-01 -3.18438530e-01 9.81186092e-01 2.63985544e-01 3.19085062e-01 8.11095059e-01 -9.18975294e-01 -7.50098526e-01 6.64121449e-01 -2.96119809e-01 4.68709975e-01 -2.10382827e-02 4.29195195e-01 9.08703744e-01 -4.65300381e-01 5.41820526e-01 9.41246390e-01 5.08939922e-01 5.77700555e-01 -1.65332508e+00 -1.51347011e-01 3.12970281e-01 3.69990543e-02 -9.33242083e-01 -4.48931545e-01 3.32290918e-01 -4.11741972e-01 1.22564721e+00 -2.22088501e-01 1.06752515e+00 9.29504275e-01 7.64052331e-01 4.97368693e-01 1.26778090e+00 -1.12825081e-01 9.30476308e-01 -3.83048087e-01 -3.74236405e-02 2.70160466e-01 -1.95565790e-01 4.98411745e-01 -5.67715406e-01 -2.54563004e-01 9.53303456e-01 1.26172841e-01 -1.91726252e-01 -3.90473604e-01 -1.19633591e+00 6.53731167e-01 3.35131943e-01 1.34692296e-01 -5.38513422e-01 6.84770465e-01 5.06948888e-01 4.01817739e-01 -4.12668064e-02 3.08992803e-01 -4.79669869e-01 -4.55956221e-01 -1.04785740e+00 3.61031324e-01 4.05977815e-01 3.71249110e-01 6.32753968e-01 2.94830948e-01 -4.66790289e-01 4.46056724e-01 3.99320543e-01 4.32051420e-01 6.71061635e-01 -1.34855950e+00 3.09899777e-01 5.31955473e-02 3.16244215e-01 -4.23853457e-01 -4.40712452e-01 -2.22462788e-01 -4.96566772e-01 6.48155451e-01 9.06106532e-01 -2.74293035e-01 -6.57973111e-01 2.39068913e+00 -2.01569304e-01 -1.40840873e-01 1.05974145e-01 7.21015811e-01 -2.85750091e-01 5.38100541e-01 4.60241288e-01 -5.15041709e-01 7.65681148e-01 -7.65508339e-02 -4.10295218e-01 -3.05353791e-01 7.02823997e-01 -4.27945182e-02 8.62183392e-01 1.75468713e-01 -1.55755222e+00 4.63852324e-02 -9.07771409e-01 -9.23992321e-02 -1.89794824e-01 -1.84528515e-01 8.08651209e-01 2.55136847e-01 -1.36796069e+00 1.22110581e+00 -1.01548207e+00 -2.04509236e-02 5.02465010e-01 4.28170830e-01 -1.28719822e-01 3.59000325e-01 -1.25378501e+00 1.20006323e+00 3.20305079e-01 1.72171801e-01 -1.06086290e+00 -4.18641269e-01 -6.11581385e-01 3.63724470e-01 -1.81694090e-01 -4.91288692e-01 1.35095716e+00 -9.48855162e-01 -1.39755630e+00 5.36028504e-01 -9.99010280e-02 -8.52157474e-01 2.92722642e-01 5.53669274e-01 6.64689168e-02 -4.69895191e-02 -1.91613823e-01 8.07796180e-01 5.35961390e-01 -6.12382770e-01 -2.34457374e-01 -6.45132840e-01 4.48597483e-02 4.66263771e-01 4.81273159e-02 -3.52441669e-01 3.10488999e-01 -5.26446640e-01 1.14226334e-01 -7.95620501e-01 -2.71674722e-01 4.81419683e-01 3.56544733e-01 -9.53693464e-02 -2.44106114e-01 -4.80627447e-01 9.75718856e-01 -2.27833939e+00 1.05793968e-01 1.79496154e-01 -1.11586399e-01 -2.49976054e-01 -1.65851638e-01 4.77146775e-01 -2.27155149e-01 -2.24528179e-01 -5.46663344e-01 3.22234584e-03 3.09920460e-01 4.03250486e-01 -8.21192384e-01 6.66797400e-01 1.09706886e-01 9.63307440e-01 -8.94363523e-01 -8.62863287e-02 -1.57736558e-02 2.21773341e-01 -4.54722226e-01 -3.46035272e-01 -3.55470538e-01 1.83381587e-01 -1.96612597e-01 1.61528438e-01 4.85180885e-01 2.42255464e-01 3.54271352e-01 3.34478408e-01 -3.90135258e-01 6.09528363e-01 -1.01810849e+00 1.75827801e+00 -1.39057592e-01 4.19259667e-01 -1.83134899e-01 -9.39919949e-01 8.20283055e-01 1.74292251e-01 4.98441875e-01 -1.07804191e+00 2.35230282e-01 3.63577098e-01 2.10381836e-01 2.24155560e-01 4.55859393e-01 -4.89486992e-01 -1.64981589e-01 8.78062606e-01 1.27181724e-01 1.53863067e-02 1.86442018e-01 2.54145473e-01 9.84008253e-01 4.76819456e-01 1.69218302e-01 -4.81334239e-01 -8.60354081e-02 -2.42053345e-01 6.13381267e-01 5.17960370e-01 -4.49701369e-01 3.62531632e-01 1.18314719e+00 -2.71380067e-01 -1.16335988e+00 -1.63027275e+00 -2.78704911e-01 1.02790356e+00 -7.76412785e-02 -1.08098201e-01 -5.88519633e-01 -1.57671660e-01 9.13598016e-02 9.89779651e-01 -7.90280342e-01 -6.20510459e-01 -2.60540545e-01 -5.82910657e-01 5.04941523e-01 7.33903289e-01 -5.34189679e-02 -1.36768961e+00 -1.29348254e+00 4.59735483e-01 2.73015320e-01 -1.02394283e-01 -4.61367428e-01 1.04247880e+00 -1.34260881e+00 -6.42837703e-01 -7.78904557e-01 -4.70011055e-01 5.35294592e-01 -3.44892174e-01 8.56848001e-01 -3.12013537e-01 -1.25463873e-01 3.13869715e-01 3.44413668e-01 -7.65007734e-02 9.83253270e-02 -4.78688300e-01 1.13223717e-01 -3.55086148e-01 1.63469404e-01 -9.61883724e-01 -7.23962188e-01 -1.64825395e-01 -6.94893539e-01 -2.97354639e-01 3.18305284e-01 8.40317249e-01 1.01732576e+00 -3.25036973e-01 7.64854074e-01 -4.05597627e-01 9.54309702e-01 -5.95402598e-01 -7.97042489e-01 1.75156996e-01 -6.47836089e-01 7.06870258e-01 6.29240632e-01 -5.34816444e-01 -1.15478098e+00 -9.68582090e-03 -6.37510195e-02 -7.40163326e-02 5.71601726e-02 4.58547652e-01 3.40218186e-01 5.66637337e-01 6.73176825e-01 6.37479424e-01 -9.68922803e-04 -1.00275248e-01 5.43568850e-01 -7.01588988e-02 6.06310189e-01 -9.37147141e-01 3.62306200e-02 2.55782455e-01 8.93266872e-02 -9.20906514e-02 -2.52863973e-01 3.12337607e-01 -5.69234490e-01 -1.89310819e-01 5.95452607e-01 -8.61775339e-01 -1.11146760e+00 4.51767951e-01 -1.03265655e+00 -7.81423748e-01 -1.11267769e+00 5.41844964e-01 -1.25030541e+00 -6.92527518e-02 -7.80909181e-01 -9.96317565e-01 9.09358785e-02 -9.22102630e-01 5.57029426e-01 1.26475796e-01 -2.79800624e-01 -9.25802350e-01 2.57137477e-01 -7.57167280e-01 5.49456418e-01 -7.63487723e-03 1.27431524e+00 -2.05049798e-01 -3.62772435e-01 1.77396402e-01 4.05899668e-03 -1.82102937e-02 -5.26013136e-01 -6.84306771e-02 -7.88881123e-01 -8.59090313e-02 2.45203942e-01 -5.46409190e-01 1.18699634e+00 7.57160187e-01 1.09810197e+00 -3.90971154e-01 -2.11087670e-02 3.19862276e-01 1.34487987e+00 2.91259199e-01 9.89914000e-01 1.25515759e-01 -1.34316385e-01 5.33074081e-01 3.56338382e-01 8.56144249e-01 4.96438324e-01 2.57452130e-01 6.49392545e-01 5.64236462e-01 1.68115348e-01 -3.67696404e-01 7.09890962e-01 3.32184047e-01 -3.06100342e-02 1.72610819e-01 -5.75144768e-01 7.18020201e-01 -1.84212184e+00 -1.63496304e+00 1.86943948e-01 2.72942138e+00 1.22860599e+00 2.39904687e-01 9.06686857e-02 -4.82250936e-02 4.24163610e-01 1.37140736e-01 -1.34067261e+00 -1.15636361e+00 -9.40998197e-02 1.85603380e-01 5.41800082e-01 4.34209675e-01 -3.75965536e-01 5.70111930e-01 7.66153431e+00 6.38275564e-01 -9.81518805e-01 -8.16918239e-02 7.90312529e-01 -6.50349915e-01 -7.53932536e-01 1.98108822e-01 -6.19840264e-01 6.17139935e-01 1.38871968e+00 -4.43663418e-01 1.00509417e+00 7.22969353e-01 7.74924979e-02 -6.57860935e-01 -1.19097137e+00 8.31091166e-01 -6.35151267e-01 -1.33996511e+00 -3.73278618e-01 1.99027374e-01 4.92715031e-01 2.65627980e-01 5.07944286e-01 3.18710446e-01 6.76892936e-01 -1.01837051e+00 1.22754037e+00 8.91406834e-01 7.39065647e-01 -8.78643692e-01 2.73259997e-01 6.05599523e-01 -8.81232142e-01 -4.12937313e-01 -5.98816872e-01 -4.95332092e-01 7.06655085e-02 5.56051910e-01 -1.78325459e-01 -4.28788424e-01 2.98469812e-01 6.45710111e-01 -1.92881122e-01 8.08634281e-01 -1.91556826e-01 2.23558098e-01 -4.57507670e-01 -3.76205817e-02 1.73158929e-01 -2.96705812e-01 -1.05725855e-01 7.51620054e-01 5.72317719e-01 2.41529658e-01 -4.34309065e-01 1.22867310e+00 7.98802227e-02 -3.05313468e-01 -6.14743829e-01 -1.34401485e-01 6.73598468e-01 8.15465510e-01 -6.39903426e-01 6.20648731e-03 4.12491988e-03 7.20543623e-01 6.66235209e-01 4.42708313e-01 -8.69970024e-01 -1.85257718e-01 6.82302833e-01 -7.30455294e-02 3.96396220e-01 -6.45744145e-01 -5.42749286e-01 -8.80776763e-01 -4.52926122e-02 -3.85550559e-01 3.28393310e-01 -7.91893899e-01 -1.02069676e+00 2.62898386e-01 -1.55761600e-01 -9.21221197e-01 -7.08295107e-01 -4.51621383e-01 -6.94977045e-01 1.13350630e+00 -1.42368388e+00 -2.50858158e-01 8.24911356e-01 9.15893435e-01 -2.77802441e-02 2.24911004e-01 1.01719260e+00 -2.49011815e-01 -1.71527877e-01 3.67237985e-01 5.77095330e-01 -2.58181691e-01 4.68620688e-01 -1.28094149e+00 1.00098476e-01 4.31092888e-01 7.39872456e-02 7.91226745e-01 4.89927620e-01 -4.05065030e-01 -1.30719388e+00 -5.33181071e-01 8.68266165e-01 -2.23847955e-01 7.17311859e-01 -1.90273628e-01 -8.74389768e-01 7.09482193e-01 1.15580171e-01 2.21565709e-01 4.66401160e-01 2.55843122e-02 -5.12416124e-01 -1.84463695e-01 -1.31692541e+00 6.74955428e-01 7.35244274e-01 -8.48334134e-01 -5.06810367e-01 -1.33918867e-01 1.12958454e-01 -2.66562700e-01 -7.38990486e-01 -2.75075138e-01 7.65285850e-01 -1.37616599e+00 5.84198773e-01 -4.41132069e-01 1.66827902e-01 -2.90388376e-01 -4.31623638e-01 -1.30211544e+00 -3.40032250e-01 -3.56901169e-01 -1.26139000e-01 9.71760869e-01 4.40734595e-01 -6.89569771e-01 3.73239815e-01 1.15067589e+00 -1.00109661e-02 -9.90340412e-01 -1.38243163e+00 -6.73782945e-01 5.97059488e-01 -4.75192785e-01 5.12240231e-01 5.58640778e-01 4.87243712e-01 -1.47964761e-01 1.36396047e-02 -3.60397547e-01 5.21874011e-01 2.38214806e-01 -4.08506185e-01 -9.87071455e-01 -4.26405698e-01 -7.56631374e-01 -2.30902880e-01 -9.22147036e-01 1.44424304e-01 -1.11101234e+00 2.59998828e-01 -1.41004288e+00 1.91916928e-01 -5.25284290e-01 -6.46856546e-01 8.16073716e-01 4.67314154e-01 -1.79274842e-01 1.96271211e-01 1.14908606e-01 -4.74737734e-01 7.99857736e-01 1.00946212e+00 1.17272891e-01 -8.27636719e-02 -2.98552722e-01 -5.93503475e-01 6.97036982e-01 8.87117088e-01 -6.65458381e-01 -6.57030523e-01 -5.82809985e-01 6.08349741e-01 5.34198284e-01 5.54049432e-01 -9.55391288e-01 2.77548730e-01 -5.41138530e-01 7.01845050e-01 -4.38219815e-01 3.69147927e-01 -4.95264828e-01 3.76098230e-02 5.58238029e-01 -9.22172248e-01 2.89174318e-01 3.03212106e-01 7.77911842e-01 1.76150158e-01 -3.01004559e-01 9.11371469e-01 -2.28534028e-01 -6.36637390e-01 3.40138823e-01 -9.04546618e-01 2.24233493e-01 1.09642017e+00 -2.64646798e-01 -3.69877279e-01 -3.06903034e-01 -1.01522660e+00 8.77822936e-02 6.57973766e-01 -1.62636727e-01 6.60798252e-01 -1.35007203e+00 -2.68519521e-01 1.25452444e-01 -2.88025528e-01 -4.59690183e-01 3.77351373e-01 8.38850677e-01 7.95368627e-02 2.91536063e-01 -7.29733706e-01 -1.24402173e-01 -3.05879861e-01 6.27672255e-01 4.84432369e-01 -1.96937203e-01 -2.54167587e-01 7.46069551e-01 -4.55261096e-02 -4.94353985e-03 7.72711784e-02 -4.49971706e-01 -3.67965251e-02 4.03752863e-01 6.02217197e-01 3.49397138e-02 -1.79862916e-01 -2.44688123e-01 -4.51245934e-01 2.62454540e-01 1.13693997e-01 -6.36643946e-01 1.41791272e+00 -2.27179617e-01 -1.77963316e-01 9.31483209e-01 8.07166696e-01 -5.22529721e-01 -2.08082819e+00 1.00745745e-01 3.98803800e-02 -2.52268344e-01 3.35715339e-02 -1.06096911e+00 -8.88996065e-01 1.35011065e+00 5.26604772e-01 1.69913415e-02 1.12516797e+00 -2.16186419e-01 4.86866951e-01 2.83320874e-01 6.44628882e-01 -1.34456313e+00 -8.99273902e-02 8.94253373e-01 6.70129538e-01 -5.66078246e-01 -1.87570244e-01 8.78749609e-01 -8.98436069e-01 1.22360945e+00 2.39880487e-01 -2.79029518e-01 4.43098277e-01 3.78390849e-01 -3.95558417e-01 3.08746874e-01 -1.23323166e+00 -8.40870067e-02 -2.20020413e-01 8.31932306e-01 4.90151733e-01 5.44311889e-02 -2.70746380e-01 5.00245750e-01 1.07366160e-01 -5.55277243e-02 5.93561232e-01 7.41067469e-01 -7.10566401e-01 -1.07754636e+00 -2.38284245e-02 5.49875975e-01 -1.77924752e-01 -2.82576114e-01 2.77916521e-01 2.56457210e-01 2.90811267e-02 3.86768878e-01 3.39667767e-01 2.14290813e-01 -1.42719746e-01 3.81224245e-01 9.15050149e-01 -4.93041396e-01 -3.50591213e-01 -1.70783177e-01 -2.45974883e-01 -7.78471768e-01 7.34622851e-02 -1.06870329e+00 -2.01558256e+00 -3.82182211e-01 1.78069368e-01 -1.29190087e-01 7.72582173e-01 8.05952489e-01 3.48699391e-01 3.61348301e-01 3.84440631e-01 -8.58148634e-01 -1.15220737e+00 -2.73860276e-01 -1.14075661e+00 7.62633234e-02 5.19937396e-01 -4.77398306e-01 -9.92369875e-02 -1.72187313e-01]
[4.104117393493652, 1.8107656240463257]
01d0823b-d852-4d83-a195-6641ffb7b416
sentence-structure-and-word-relationship
2108.12750
null
https://arxiv.org/abs/2108.12750v1
https://arxiv.org/pdf/2108.12750v1.pdf
Sentence Structure and Word Relationship Modeling for Emphasis Selection
Emphasis Selection is a newly proposed task which focuses on choosing words for emphasis in short sentences. Traditional methods only consider the sequence information of a sentence while ignoring the rich sentence structure and word relationship information. In this paper, we propose a new framework that considers sentence structure via a sentence structure graph and word relationship via a word similarity graph. The sentence structure graph is derived from the parse tree of a sentence. The word similarity graph allows nodes to share information with their neighbors since we argue that in emphasis selection, similar words are more likely to be emphasized together. Graph neural networks are employed to learn the representation of each node of these two graphs. Experimental results demonstrate that our framework can achieve superior performance.
['Wai Lam', 'Haoran Yang']
2021-08-29
null
https://aclanthology.org/2021.ranlp-1.175
https://aclanthology.org/2021.ranlp-1.175.pdf
ranlp-2021-9
['word-similarity']
['natural-language-processing']
[ 2.78546333e-01 8.95092413e-02 -4.78939712e-01 -6.16662204e-01 7.72985667e-02 -1.58963129e-01 2.16500871e-02 7.43495166e-01 -5.29341757e-01 6.24782264e-01 7.09392309e-01 -3.26565951e-01 -6.02129996e-02 -8.83131087e-01 -6.26838440e-03 -4.90032405e-01 -3.35704931e-03 -1.02712333e-01 2.17820778e-01 -5.54560423e-01 7.93259799e-01 2.57816818e-03 -8.70002329e-01 1.40990257e-01 8.87665987e-01 3.90542984e-01 6.91388905e-01 4.71950918e-01 -7.33255565e-01 9.85486031e-01 -9.80552912e-01 -4.67674464e-01 -2.22905487e-01 -7.20461607e-01 -7.97399163e-01 1.22585215e-01 4.73957449e-01 -4.43712622e-03 -3.62309545e-01 1.43847966e+00 4.79341239e-01 5.08317709e-01 4.23892170e-01 -8.20457458e-01 -9.15114820e-01 1.35333312e+00 -6.87542915e-01 7.94136286e-01 3.53104681e-01 -4.10777897e-01 1.65042949e+00 -9.20035660e-01 4.16096836e-01 1.28175950e+00 3.95933539e-01 2.19548732e-01 -7.16842651e-01 -2.17539459e-01 8.30918014e-01 2.81045705e-01 -1.20514834e+00 1.73858747e-01 1.29873919e+00 -1.29522700e-02 1.13711417e+00 2.41922274e-01 7.51709938e-01 6.16718531e-01 5.44659674e-01 7.98462093e-01 4.77139950e-01 -8.03069353e-01 1.08496830e-01 -2.29792923e-01 1.07653153e+00 7.85519600e-01 3.27791125e-01 -6.34071052e-01 -6.89594865e-01 -5.76693192e-02 3.68120998e-01 2.27511004e-01 -4.14240927e-01 4.84990627e-02 -9.77151573e-01 9.74923491e-01 4.44398075e-01 5.98710895e-01 -4.27371919e-01 1.08196378e-01 5.49797297e-01 5.77026486e-01 5.78118920e-01 5.40003419e-01 -2.65797675e-01 2.87184864e-01 -5.26537299e-01 -1.27370611e-01 7.19780445e-01 1.01338291e+00 8.30053151e-01 8.32243413e-02 -5.65452218e-01 8.35747659e-01 5.43885589e-01 9.68895108e-02 6.25830829e-01 -4.47624743e-01 6.33505344e-01 6.41172886e-01 -5.76365471e-01 -1.60828245e+00 -2.78701723e-01 -7.76647866e-01 -7.43944466e-01 -2.87365109e-01 -2.20643893e-01 -4.43038136e-01 -7.38849044e-01 1.68259919e+00 2.58656517e-02 2.18251944e-01 -4.59189489e-02 6.32562041e-01 1.22550011e+00 6.63075626e-01 1.90432578e-01 -5.61344743e-01 1.41007543e+00 -1.18269002e+00 -1.11705959e+00 -3.63955438e-01 8.10718000e-01 -5.58186889e-01 1.01573086e+00 -5.43154255e-02 -1.09114861e+00 -5.44950843e-01 -1.18916881e+00 -1.22694053e-01 -2.03452438e-01 -3.64632636e-01 4.70502377e-01 2.31431365e-01 -1.09616935e+00 5.54039776e-01 -2.66158193e-01 -3.07118267e-01 2.36388035e-02 3.10254276e-01 2.78567970e-02 1.21751964e-01 -1.74301267e+00 1.00234795e+00 5.86464226e-01 6.57307729e-02 -2.69235447e-02 -4.47128296e-01 -1.28148627e+00 5.15615523e-01 3.17940593e-01 -7.11430430e-01 1.05449450e+00 -1.08808637e+00 -1.12388372e+00 5.61682880e-01 -4.84980673e-01 -3.16852689e-01 -4.08374041e-01 -4.42554206e-02 -4.27550405e-01 2.55305320e-01 2.55004759e-03 1.97588891e-01 8.62618089e-01 -9.59064484e-01 -7.41807103e-01 -2.41806954e-01 4.30524975e-01 6.45179570e-01 -7.15202332e-01 2.70677209e-01 -3.01260412e-01 -9.99137878e-01 1.15307592e-01 -3.78300190e-01 -5.06932735e-01 -3.45389843e-01 -5.31900346e-01 -6.18620813e-01 7.44794905e-01 -4.82711136e-01 1.96127927e+00 -2.01498246e+00 3.84130657e-01 2.86910802e-01 6.27252936e-01 1.20150685e-01 -3.97357076e-01 8.53698075e-01 -2.38382578e-01 2.28468105e-01 -3.07947308e-01 -1.77302137e-01 -1.65029272e-01 1.79561377e-01 -1.57489441e-02 2.20142186e-01 1.22703061e-01 7.81585574e-01 -1.28392220e+00 -8.08109820e-01 -1.42982021e-01 3.80708367e-01 -2.71724820e-01 1.63141683e-01 1.01849981e-01 -1.66549310e-01 -5.94499588e-01 2.65944332e-01 3.33819360e-01 -2.82586366e-01 5.47747612e-01 -2.40457803e-01 2.77671158e-01 5.94404221e-01 -8.09355259e-01 1.69383168e+00 -3.55172575e-01 6.58905447e-01 -3.18763703e-01 -1.04748595e+00 1.26639652e+00 1.16195567e-01 7.04971626e-02 -5.17741859e-01 2.82601804e-01 -2.51832247e-01 3.75882953e-01 -7.77369201e-01 8.37637901e-01 -7.94062987e-02 2.09521577e-02 4.67267931e-01 -6.15165895e-03 -1.01033330e-01 6.63728237e-01 7.40111053e-01 9.95917618e-01 -4.53042746e-01 6.35793328e-01 -5.46148002e-01 8.31875086e-01 -5.02595007e-01 6.83118641e-01 4.26373601e-01 -3.83274466e-01 4.72037017e-01 6.26796961e-01 -2.86438525e-01 -6.02218807e-01 -9.28343117e-01 2.73878634e-01 1.25108922e+00 3.86724621e-01 -9.99705970e-01 -5.88855207e-01 -1.03832614e+00 -2.65424788e-01 1.01879466e+00 -7.24022210e-01 -4.42059666e-01 -8.55181336e-01 -1.71083912e-01 -4.75798994e-02 5.92407227e-01 1.16083965e-01 -1.21160758e+00 -2.25943625e-01 -7.18951896e-02 -7.37220645e-02 -7.33777583e-01 -1.21057451e+00 5.60260490e-02 -8.91183674e-01 -8.72305214e-01 -5.79020500e-01 -1.48488939e+00 1.03289425e+00 6.58783674e-01 1.48390090e+00 8.08773041e-01 1.53959826e-01 2.26565331e-01 -7.93140650e-01 -3.75933051e-01 -2.75788978e-02 2.87931502e-01 3.80807109e-02 -4.97348458e-02 7.06856966e-01 -6.31922126e-01 -5.42876720e-01 -2.18340188e-01 -9.21393692e-01 -3.33638161e-01 3.36625576e-01 9.21756208e-01 3.08295429e-01 1.82704166e-01 5.74961305e-01 -1.15955329e+00 1.27466977e+00 -5.55710852e-01 9.87118185e-02 5.31623900e-01 -5.95482171e-01 1.68147668e-01 4.35687751e-01 -2.17421934e-01 -1.01046813e+00 -4.35073912e-01 -2.05499485e-01 -1.02582322e-02 2.54158467e-01 1.31702185e+00 1.37282780e-03 2.39202276e-01 2.51551807e-01 1.72914401e-01 -1.74135312e-01 -2.41151214e-01 1.31520346e-01 4.57681805e-01 5.40602095e-02 -3.17144454e-01 4.70599413e-01 -8.75353292e-02 4.08229008e-02 -1.14612126e+00 -1.07557356e+00 -4.79466796e-01 -8.92700493e-01 -1.93250418e-01 6.36965632e-01 -5.44690788e-01 -2.98253089e-01 -1.34411424e-01 -1.29538774e+00 3.88637304e-01 -2.42194742e-01 6.09559715e-01 1.40804499e-01 1.05959630e+00 -7.44738579e-01 -7.84035683e-01 -5.45506418e-01 -8.60904574e-01 7.93831170e-01 5.29391706e-01 -5.15632093e-01 -1.65914190e+00 1.74555942e-01 -9.00567994e-02 1.98523045e-01 -1.45552024e-01 1.05370724e+00 -8.98308098e-01 2.96300072e-02 -1.54413015e-01 -9.94114727e-02 3.45767826e-01 6.24755383e-01 2.70283848e-01 -2.80919969e-01 -2.71274358e-01 2.80250400e-01 8.22407454e-02 1.08806872e+00 4.40302104e-01 1.02891314e+00 -1.49554059e-01 -3.10129076e-01 3.36950123e-01 1.24400747e+00 1.66554719e-01 5.33779263e-01 1.06052965e-01 9.79534149e-01 9.25507367e-01 5.93261540e-01 2.67483443e-01 3.67142320e-01 1.99296132e-01 2.69852847e-01 -3.82135762e-03 -1.62921220e-01 -1.36088967e-01 2.74130851e-01 1.76620412e+00 1.74372420e-01 -6.46587491e-01 -7.40664899e-01 4.62755322e-01 -1.85004139e+00 -9.64920044e-01 -5.45864940e-01 1.83556914e+00 8.53749216e-01 4.97129232e-01 -3.01811635e-01 1.03320234e-01 1.19084132e+00 8.25739503e-01 -1.13030076e-01 -7.98189998e-01 -2.00214177e-01 2.76728034e-01 2.32624840e-02 9.29832458e-01 -7.45397806e-01 9.51353669e-01 6.70675850e+00 6.60024941e-01 -7.08092749e-01 -2.90526539e-01 5.34350157e-01 1.14301085e-01 -7.97347605e-01 -5.12736849e-02 -7.31987834e-01 5.54191530e-01 7.03958452e-01 -5.77666759e-01 -2.26560205e-01 6.08870208e-01 1.33826926e-01 2.71220971e-02 -7.56798029e-01 7.95257568e-01 6.29769683e-01 -1.22720397e+00 2.91674107e-01 -4.65717524e-01 4.62515265e-01 -3.19793999e-01 -1.09078504e-01 1.76322505e-01 1.63156018e-01 -7.65983820e-01 2.24901766e-01 4.05245155e-01 3.67204368e-01 -9.06035721e-01 8.13765764e-01 1.84221327e-01 -1.52930129e+00 2.43337825e-01 -5.01479566e-01 -5.89689732e-01 3.21759790e-01 7.48454452e-01 -5.60521781e-01 6.81129575e-01 3.06697130e-01 1.08340406e+00 -5.91844738e-01 8.20957065e-01 -8.61790240e-01 5.76751649e-01 3.47736627e-01 -6.98552132e-01 3.84033561e-01 -4.81491953e-01 6.26466453e-01 1.52398872e+00 1.07588090e-01 1.70980781e-01 4.62493449e-01 3.28858465e-01 -1.95140436e-01 4.15772706e-01 -6.87897980e-01 -6.03347979e-02 4.47358280e-01 1.28050089e+00 -8.95619035e-01 -3.40987831e-01 -5.69314778e-01 1.08384621e+00 7.95915544e-01 4.58105415e-01 -5.65224290e-01 -1.07021773e+00 6.60209954e-01 -4.06967044e-01 2.34118983e-01 -5.31953633e-01 -3.65997642e-01 -1.15496480e+00 -1.62228093e-01 -4.42221791e-01 5.85916400e-01 -6.78456306e-01 -1.64409995e+00 8.38943601e-01 -1.99728534e-01 -9.13805425e-01 1.93014905e-01 -5.62980175e-01 -1.24449968e+00 1.12159371e+00 -1.55246282e+00 -5.52174926e-01 4.09563519e-02 2.27319345e-01 1.17997873e+00 -6.67547360e-02 5.12290835e-01 -1.40155688e-01 -7.50211120e-01 5.43394148e-01 -2.57236362e-01 2.90731162e-01 4.38765615e-01 -1.43167377e+00 5.33139527e-01 9.28520620e-01 3.18957746e-01 1.12961507e+00 6.58329129e-01 -1.07885659e+00 -1.13248253e+00 -7.75673032e-01 1.68273342e+00 -1.50906101e-01 6.94803417e-01 -1.31555632e-01 -1.00653338e+00 4.77770329e-01 9.88763869e-01 -2.09630564e-01 1.19818068e+00 3.99891049e-01 -2.11200640e-01 1.60311341e-01 -6.84185028e-01 9.14265990e-01 1.16161954e+00 -6.16253674e-01 -1.27085829e+00 4.28554386e-01 1.27991986e+00 -1.40540555e-01 -5.94435692e-01 2.73982465e-01 1.39816359e-01 -6.49181247e-01 6.74534142e-01 -9.58979785e-01 7.02045023e-01 -2.99042970e-01 -1.71181738e-01 -1.70232046e+00 -9.17939544e-01 -4.63036239e-01 -3.53217065e-01 1.18766344e+00 7.02581227e-01 -4.37256813e-01 7.35587656e-01 2.24328607e-01 -3.13052177e-01 -9.03463423e-01 -3.43592465e-01 -5.84269524e-01 -2.01473162e-01 -6.09296933e-02 5.89767456e-01 1.32454157e+00 5.60026407e-01 1.19404376e+00 -1.36434510e-01 3.12161818e-02 3.64825666e-01 6.36853948e-02 -1.89192221e-02 -1.19178879e+00 -1.52938291e-01 -5.58487356e-01 -1.85775310e-01 -1.15139818e+00 4.25102502e-01 -8.77142489e-01 1.84451967e-01 -1.88945317e+00 2.60432661e-01 1.45964563e-01 -7.54991531e-01 -1.99250802e-02 -1.08834183e+00 -2.08460093e-01 1.84895277e-01 -5.05607277e-02 -6.87056720e-01 8.70474041e-01 1.34726083e+00 -3.42932612e-01 -7.36516193e-02 -1.69242874e-01 -8.24193478e-01 7.71874368e-01 1.01585233e+00 -3.98286521e-01 -8.22377264e-01 -7.64368773e-01 4.52722907e-01 -3.17045063e-01 -6.19168103e-01 -5.18450379e-01 4.10854131e-01 -3.64994705e-01 9.91011783e-02 -6.03698432e-01 3.85185555e-02 -7.66442001e-01 -6.83807492e-01 6.24112844e-01 -8.45902205e-01 5.86865425e-01 -2.98190862e-01 7.28471816e-01 -3.44528586e-01 -8.50927651e-01 3.33835214e-01 -2.77217239e-01 -7.27194607e-01 3.79103571e-01 -4.04520035e-01 -9.93470177e-02 8.15955102e-01 -1.95416391e-01 -1.85337871e-01 -4.87524420e-01 -4.89113510e-01 4.97838587e-01 6.08712733e-02 5.53967118e-01 1.06342578e+00 -1.33594251e+00 -7.02777445e-01 -1.42833471e-01 2.05999464e-02 -1.20280020e-01 1.34739559e-02 4.22305584e-01 -4.55197483e-01 1.03082217e-01 7.30270660e-03 -2.50867277e-01 -1.75464869e+00 7.21802652e-01 -2.36854348e-02 -3.05131525e-01 -5.00614047e-01 1.24480689e+00 3.21863562e-01 -1.25949115e-01 2.96750575e-01 -3.66700560e-01 -9.80398536e-01 1.48531437e-01 8.07742417e-01 1.87616423e-01 -2.00604662e-01 -6.64738894e-01 -2.35897169e-01 5.56183279e-01 -5.58651209e-01 1.11214884e-01 1.16253924e+00 -4.57985699e-01 -7.33936012e-01 7.13567019e-01 1.05274475e+00 1.73347875e-01 -5.24434268e-01 -5.58833063e-01 3.23285669e-01 -6.23240232e-01 2.20169857e-01 -2.16804460e-01 -1.09331334e+00 7.81484127e-01 -2.37515479e-01 6.90236270e-01 1.14375412e+00 -2.13189647e-01 8.09740543e-01 5.86437464e-01 3.34060341e-02 -1.13222265e+00 4.09274518e-01 9.75626290e-01 8.34202826e-01 -9.77351248e-01 2.40693167e-01 -8.62848520e-01 -8.59947860e-01 1.36612582e+00 9.74173427e-01 -2.08696827e-01 7.65291631e-01 2.67448388e-02 6.21231161e-02 -4.34519917e-01 -6.62675679e-01 -3.00188512e-01 4.13038850e-01 5.61043680e-01 9.39204633e-01 -2.95952916e-01 -1.00899482e+00 4.46247637e-01 -2.37459200e-03 -6.68359399e-01 5.44469893e-01 1.20273674e+00 -8.56688380e-01 -1.37707841e+00 -8.89986847e-03 5.62287152e-01 -3.62411588e-01 -6.86417043e-01 -6.17778540e-01 8.06098133e-02 -1.90683782e-01 1.08776760e+00 2.25357115e-01 -4.66986746e-01 1.65671900e-01 -1.60950515e-02 2.76330560e-01 -1.21086311e+00 -8.80526721e-01 -2.38773108e-01 2.72586554e-01 -9.58051998e-03 -4.61857647e-01 -3.53328675e-01 -1.52607405e+00 -1.80950701e-01 -5.14981449e-01 6.15531027e-01 5.01165330e-01 1.02962518e+00 3.34480137e-01 1.08492219e+00 8.38220775e-01 -3.82162631e-01 -3.42133373e-01 -9.83766615e-01 -9.12066221e-01 3.80722135e-01 2.51384765e-01 -3.23672771e-01 -1.22760259e-01 -1.37230799e-01]
[11.159334182739258, 8.769057273864746]
de41dc2a-c21e-444b-905d-176aa2b53266
timexplain-a-framework-for-explaining-the
2007.07606
null
https://arxiv.org/abs/2007.07606v1
https://arxiv.org/pdf/2007.07606v1.pdf
timeXplain -- A Framework for Explaining the Predictions of Time Series Classifiers
Modern time series classifiers display impressive predictive capabilities, yet their decision-making processes mostly remain black boxes to the user. At the same time, model-agnostic explainers, such as the recently proposed SHAP, promise to make the predictions of machine learning models interpretable, provided there are well-designed domain mappings. We bring both worlds together in our timeXplain framework, extending the reach of explainable artificial intelligence to time series classification and value prediction. We present novel domain mappings for the time and the frequency domain as well as series statistics and analyze their explicative power as well as their limits. We employ timeXplain in a large-scale experimental comparison of several state-of-the-art time series classifiers and discover similarities between seemingly distinct classification concepts such as residual neural networks and elastic ensembles.
['Patrick Schäfer', 'Vanja Doskoč', 'Martin Schirneck', 'Felix Mujkanovic', 'Tobias Friedrich']
2020-07-15
null
null
null
null
['value-prediction']
['computer-code']
[ 1.59471691e-01 2.09615275e-01 -3.84112656e-01 -4.97827828e-01 -1.31719381e-01 -8.56423259e-01 8.03435802e-01 -6.01725914e-02 2.78743386e-01 7.27814078e-01 1.09068610e-01 -7.25186527e-01 -7.73221135e-01 -5.71508825e-01 -4.98833567e-01 -5.69812298e-01 -5.26874602e-01 6.13896906e-01 -1.89807877e-01 -5.69465518e-01 1.84082761e-01 3.31488460e-01 -1.53698385e+00 2.34139323e-01 1.01181197e+00 1.33511305e+00 -4.49732453e-01 4.23630714e-01 -1.87863201e-01 9.51536417e-01 -3.71087283e-01 -2.77990222e-01 1.10595010e-01 -3.47434551e-01 -6.13455951e-01 -5.63022733e-01 -1.56126872e-01 1.76979706e-01 -5.30071378e-01 5.37563026e-01 -1.42573491e-01 2.80568540e-01 7.30149209e-01 -1.75264513e+00 -1.16879809e+00 8.65117431e-01 -1.07461251e-01 4.95373547e-01 8.45848769e-02 1.68787222e-02 1.01175606e+00 -4.47425932e-01 5.87691605e-01 7.50160635e-01 1.22024035e+00 6.56170487e-01 -1.57167900e+00 -5.47792196e-01 3.51372510e-01 5.85707486e-01 -7.48718500e-01 -7.53946304e-02 1.08139169e+00 -5.42004764e-01 1.43421602e+00 6.33773208e-01 6.91659212e-01 1.46595871e+00 6.75972342e-01 4.20612097e-01 1.16935015e+00 -3.79112899e-01 4.05295402e-01 -1.78370968e-01 3.61300349e-01 2.42835775e-01 -1.21318161e-01 5.88395536e-01 -4.78095412e-01 -1.76079974e-01 5.89224935e-01 4.33454394e-01 -2.27225333e-01 -2.63118923e-01 -1.36335301e+00 7.89508581e-01 4.17402893e-01 5.70781648e-01 -3.52793813e-01 3.97716612e-01 5.29112458e-01 8.02448213e-01 1.02009642e+00 7.84540832e-01 -1.10760665e+00 -2.54927099e-01 -7.45764792e-01 1.29113272e-01 7.51894653e-01 6.44037366e-01 3.65049154e-01 4.39814240e-01 1.86681643e-01 4.46568042e-01 -2.33725488e-01 2.80717552e-01 8.96575749e-01 -7.61464477e-01 1.34101540e-01 4.03726012e-01 -3.41044515e-02 -8.16665292e-01 -7.90193319e-01 -6.51583910e-01 -9.87994075e-01 1.90290034e-01 3.48432750e-01 1.28747700e-02 -8.69585991e-01 1.82457590e+00 -7.68330842e-02 6.54992640e-01 8.90322309e-03 7.44888723e-01 2.26585299e-01 6.99421823e-01 2.90880561e-01 -5.00081599e-01 1.04278755e+00 -8.13336670e-01 -5.62240303e-01 -3.93847115e-02 4.25853908e-01 -2.90706545e-01 8.67778063e-01 2.14320362e-01 -8.77454638e-01 -6.92910612e-01 -1.02307177e+00 9.29446444e-02 -4.72649455e-01 -4.54505831e-01 1.14190185e+00 2.16472343e-01 -6.61789238e-01 1.40737128e+00 -1.32530582e+00 -2.00587928e-01 3.30935642e-02 4.03129309e-01 -3.12003613e-01 5.96215010e-01 -1.28651679e+00 1.14303887e+00 2.55276173e-01 -1.44538447e-01 -3.19107383e-01 -1.29470015e+00 -4.66939718e-01 2.17068642e-01 -8.85049552e-02 -8.53362858e-01 1.31686378e+00 -1.17870390e+00 -1.20828950e+00 5.85862458e-01 -6.58919662e-02 -1.05007339e+00 3.15566689e-01 3.59460674e-02 -1.02993584e+00 -2.54366666e-01 -1.53924569e-01 8.89623910e-02 8.35115016e-01 -5.58034539e-01 -4.16663349e-01 -1.64778158e-01 -2.50402272e-01 -4.69803423e-01 -1.79889306e-01 -3.82912874e-01 6.82683468e-01 -9.63829339e-01 4.49225813e-01 -9.80807781e-01 -2.80089229e-01 4.23859525e-03 -2.11392373e-01 -2.88162380e-01 8.05590093e-01 -5.14339328e-01 1.32112014e+00 -1.83966875e+00 2.94728190e-01 2.60381550e-01 4.21253979e-01 -3.57606918e-01 -1.28823921e-01 5.25783777e-01 -9.45210218e-01 1.63375184e-01 -1.78688675e-01 -1.04431763e-01 2.83653945e-01 3.04499537e-01 -1.09944892e+00 4.81600791e-01 2.73321807e-01 1.09223831e+00 -5.15648544e-01 1.75427124e-01 3.39839965e-01 2.85717100e-01 -4.18878019e-01 1.95260122e-01 -4.88994688e-01 4.70069528e-01 -5.48161268e-01 4.29870695e-01 3.04393411e-01 -5.03659666e-01 5.40100336e-02 4.15715724e-02 -1.30364150e-01 4.86930251e-01 -4.11609054e-01 1.27698421e+00 -3.40166986e-01 9.08693254e-01 -7.46619463e-01 -1.46258557e+00 1.03581512e+00 4.07119304e-01 9.74309504e-01 -7.79691756e-01 -2.00360969e-01 3.40238869e-01 1.66797563e-01 -4.05837417e-01 2.76943862e-01 -5.79926014e-01 -3.94121669e-02 7.02537715e-01 -7.38160461e-02 4.27994169e-02 -2.53974557e-01 -4.08010095e-01 1.35237229e+00 3.35331738e-01 5.22360265e-01 -4.21877533e-01 2.61458218e-01 2.63222784e-01 4.44854796e-01 4.19920951e-01 -1.36598483e-01 5.35928726e-01 4.38979357e-01 -1.25065112e+00 -1.44480658e+00 -1.07496643e+00 -5.27302742e-01 1.22611618e+00 -2.03672409e-01 -2.70346284e-01 -3.41248900e-01 -4.85271513e-01 5.00214159e-01 9.56053495e-01 -1.10851347e+00 -5.25221407e-01 -6.23207271e-01 -6.16262794e-01 2.78037727e-01 9.15049016e-01 -1.75482303e-01 -1.07389927e+00 -5.16492009e-01 5.12221932e-01 4.30957191e-02 -8.07972074e-01 -1.29602954e-01 8.24395716e-01 -1.23456180e+00 -7.46958435e-01 -2.58905858e-01 -2.97681272e-01 2.10348040e-01 -1.36669382e-01 1.49125004e+00 4.20577340e-02 -6.74416274e-02 2.89728969e-01 -2.07743213e-01 -6.29311025e-01 -4.64761287e-01 1.60753638e-01 4.17175353e-01 -3.84780943e-01 5.65103352e-01 -1.31934667e+00 -5.45457125e-01 4.52992678e-01 -5.07679760e-01 1.70224026e-01 7.13490248e-02 1.00085628e+00 5.29115081e-01 -6.76160008e-02 8.77359152e-01 -6.47656858e-01 6.63947880e-01 -9.38946247e-01 -4.20859873e-01 3.37272644e-01 -1.23838139e+00 4.03951168e-01 1.03334665e+00 -9.63000178e-01 -7.08380520e-01 -4.26062196e-01 1.18067302e-01 -6.78982913e-01 -1.99007709e-02 7.03413308e-01 5.88373959e-01 1.23082481e-01 9.73824739e-01 3.30034703e-01 1.45502567e-01 -5.92757702e-01 3.59428406e-01 1.45169064e-01 7.13990748e-01 -7.26833940e-01 1.05042112e+00 4.63729352e-01 1.30235717e-01 -2.46657997e-01 -7.73618519e-01 -1.26750022e-02 -5.94145417e-01 -1.15115279e-02 5.90079546e-01 -5.69672644e-01 -8.57569873e-01 -1.33163869e-01 -1.13298857e+00 -1.28177166e-01 -8.00105751e-01 4.10699964e-01 -1.02918541e+00 -2.49150336e-01 -5.76399326e-01 -7.91648209e-01 -2.62468368e-01 -4.93698865e-01 7.50372529e-01 7.86759853e-02 -8.16781998e-01 -1.49719703e+00 2.71760255e-01 -1.89146519e-01 9.05058324e-01 6.80225134e-01 1.45815897e+00 -1.20638943e+00 -3.93659204e-01 -2.41693661e-01 1.15223370e-01 -7.59666711e-02 -1.48367956e-01 -1.59351397e-02 -9.74950194e-01 6.59481883e-02 1.62866622e-01 1.53236151e-01 6.95121884e-01 6.33174479e-01 1.33083022e+00 -4.26005542e-01 -4.66930330e-01 7.61967421e-01 1.13314080e+00 5.07794023e-01 5.03389299e-01 4.66383040e-01 3.46781224e-01 6.03913188e-01 2.31337309e-01 4.16963965e-01 8.61153305e-02 5.71166515e-01 2.67122924e-01 3.82264517e-02 1.91890672e-01 -3.92795831e-01 1.58905923e-01 1.06277156e+00 -6.35703385e-01 -4.14610952e-02 -9.61314678e-01 1.47000685e-01 -2.10068750e+00 -1.48715866e+00 -1.36558399e-01 1.76784813e+00 6.60385132e-01 1.66447669e-01 7.42163509e-02 3.96967977e-01 4.39234585e-01 2.46369138e-01 -1.00298250e+00 -4.70993131e-01 -1.56626850e-01 4.23809201e-01 1.45245597e-01 7.65961483e-02 -1.01604092e+00 4.31487709e-01 7.72204447e+00 4.89740849e-01 -1.35648906e+00 7.41869584e-02 7.28581369e-01 -2.25224972e-01 -6.54442966e-01 1.65230393e-01 1.24003492e-01 4.11884129e-01 1.60545588e+00 -8.99626195e-01 8.53419662e-01 1.26676190e+00 1.29964903e-01 7.32534230e-01 -1.55732489e+00 9.46284294e-01 -5.13387263e-01 -1.72450280e+00 -3.43980163e-01 -8.52194652e-02 6.35550439e-01 2.19097108e-01 2.91976571e-01 5.99812508e-01 -4.17203791e-02 -1.37756419e+00 8.93946826e-01 8.60266089e-01 6.91651404e-01 -3.33308756e-01 3.84563595e-01 2.00317517e-01 -1.23738408e+00 -5.19708037e-01 -2.59459943e-01 -6.07903838e-01 2.19148267e-02 3.63977700e-01 -3.16039145e-01 3.94381702e-01 5.29734135e-01 1.21767080e+00 -2.43819311e-01 5.96673489e-01 2.24332288e-01 7.66094685e-01 -1.25038370e-01 9.77233145e-03 -6.63352832e-02 -1.92344755e-01 5.51714897e-01 6.39394343e-01 5.66073537e-01 3.88844311e-01 -2.18411803e-01 1.20858109e+00 3.06372851e-01 -4.71133947e-01 -6.16963565e-01 -3.70981365e-01 4.05525386e-01 8.13455701e-01 -4.76064205e-01 -1.80431336e-01 -3.15154463e-01 5.48853576e-01 2.04569891e-01 5.06940186e-01 -1.16652822e+00 -4.84709404e-02 1.08416665e+00 2.06372082e-01 3.99718154e-03 -2.40256965e-01 -6.96003437e-01 -1.40818322e+00 1.07816361e-01 -8.26425493e-01 4.35695142e-01 -1.09256291e+00 -1.88462925e+00 8.02771509e-01 9.44433585e-02 -1.73360944e+00 -8.13718379e-01 -9.52105701e-01 -7.63968349e-01 7.17754960e-01 -1.24119675e+00 -1.11480749e+00 -4.56469208e-02 4.78509665e-01 4.68540609e-01 -3.49389285e-01 1.09200037e+00 -1.05041966e-01 -2.56811798e-01 2.88149416e-01 4.02303904e-01 -2.78523684e-01 1.61185294e-01 -1.24145269e+00 7.61570334e-01 3.32599968e-01 3.42345685e-01 6.92060053e-01 1.21330285e+00 -3.14099371e-01 -1.51007581e+00 -7.83526957e-01 6.89979911e-01 -7.52234519e-01 1.28019273e+00 -9.28360447e-02 -1.29990745e+00 8.81403625e-01 5.29970601e-02 1.33112848e-01 6.58387959e-01 7.33678341e-01 -7.21220315e-01 -9.68651921e-02 -8.35478842e-01 4.57768649e-01 1.25240624e+00 -9.24239635e-01 -9.43318009e-01 4.72521216e-01 9.41495776e-01 -3.34412873e-01 -1.11234844e+00 3.66389781e-01 8.82649183e-01 -9.93032873e-01 1.11807632e+00 -1.46128547e+00 7.28794396e-01 1.39417738e-01 -1.46504045e-01 -1.43253231e+00 -5.24439394e-01 -9.61387992e-01 -6.23533547e-01 8.25795293e-01 5.27000546e-01 -1.06547320e+00 6.10481501e-01 1.12755358e+00 -2.82163411e-01 -8.87887239e-01 -1.13883185e+00 -1.18766129e+00 3.49310815e-01 -8.27046931e-01 1.11334038e+00 1.18074036e+00 6.38938904e-01 1.32849917e-01 -2.95913547e-01 -1.13518730e-01 4.05660182e-01 5.43155968e-01 2.59778947e-01 -1.85317004e+00 -4.48527038e-01 -6.89783454e-01 -4.98386800e-01 -4.32196140e-01 5.32484055e-01 -1.04523814e+00 -4.72949743e-01 -8.75450432e-01 -3.06826215e-02 -5.72355449e-01 -7.33813524e-01 6.15542293e-01 2.02382699e-01 -2.05143392e-01 1.21729083e-01 6.39191806e-01 -2.21982058e-02 4.92977649e-01 8.65428209e-01 -1.53541178e-01 -2.23991767e-01 2.09153578e-01 -6.14995837e-01 6.30082726e-01 7.61395752e-01 -5.55554211e-01 -6.26599550e-01 -2.10921735e-01 2.03051001e-01 5.80056727e-01 6.00098372e-01 -9.67968464e-01 3.00184339e-02 -6.37832761e-01 4.92244869e-01 -1.64498344e-01 2.60538638e-01 -9.88434970e-01 6.89024508e-01 3.77279490e-01 -6.40474439e-01 5.45293391e-01 4.44315344e-01 7.76829183e-01 -1.38817057e-01 3.98976326e-01 4.22866076e-01 3.00770849e-01 -6.94205165e-01 3.72232467e-01 2.74059121e-02 -2.84053106e-02 9.71736670e-01 -2.88849503e-01 -8.58856916e-01 -5.77703476e-01 -8.95425975e-01 -1.04402013e-01 1.06207214e-01 7.71839440e-01 2.57466406e-01 -1.43405354e+00 -4.96779352e-01 2.69796818e-01 1.34619340e-01 -8.96412313e-01 3.80805284e-01 9.22437072e-01 4.17268537e-02 6.63017988e-01 -4.93723571e-01 -5.60911536e-01 -7.63851762e-01 8.62998664e-01 6.03063047e-01 -2.46658489e-01 -9.24492598e-01 2.82083333e-01 1.23975590e-01 -3.60421687e-01 -1.75148532e-01 -7.37556100e-01 2.12288797e-01 -1.11271560e-01 3.48087132e-01 1.76363111e-01 -1.08746469e-01 -1.23909630e-01 -4.49524939e-01 4.91737187e-01 3.03473085e-01 1.90218404e-01 1.89908576e+00 2.23329768e-01 -3.34888995e-02 7.56655335e-01 1.11085546e+00 -5.84718287e-01 -1.26737797e+00 7.20926821e-02 3.94098312e-01 -5.16884401e-02 -2.57441640e-01 -1.00459862e+00 -6.70956731e-01 8.11181843e-01 4.01769698e-01 1.00845957e+00 1.28841746e+00 3.84428985e-02 7.47608960e-01 1.92406610e-01 3.84057760e-01 -8.68502319e-01 -9.42495689e-02 4.45147187e-01 1.03664923e+00 -1.05683517e+00 -3.10204495e-02 -5.00184186e-02 -4.86733407e-01 1.43523526e+00 3.05821508e-01 -2.82192886e-01 8.11185718e-01 2.97175914e-01 -1.37187570e-01 6.17427798e-03 -1.51625419e+00 1.70780003e-01 7.80622482e-01 7.30208874e-01 5.67806005e-01 2.19189167e-01 -1.65439337e-01 1.23437393e+00 -6.36848629e-01 -2.22059625e-05 2.76513547e-01 4.52709138e-01 -1.52764753e-01 -8.39954317e-01 -1.35306641e-01 5.09025216e-01 -2.89050370e-01 4.76210080e-02 -1.64996572e-02 1.06182528e+00 -2.18498573e-01 7.13217795e-01 3.33749294e-01 -6.88192964e-01 3.16830516e-01 5.48729300e-01 1.47228837e-01 -2.64189821e-02 -6.61977351e-01 -3.13373983e-01 4.92118821e-02 -5.67445517e-01 -2.24090710e-01 -5.81429899e-01 -1.32409286e+00 -8.15535188e-01 -4.91313525e-02 9.82479751e-02 7.03976989e-01 1.13333070e+00 5.40603340e-01 6.62754357e-01 7.26117432e-01 -8.62003803e-01 -9.73451197e-01 -8.27674031e-01 -5.17957568e-01 5.83716035e-01 5.88794947e-01 -6.03707731e-01 -5.33988357e-01 2.15582833e-01]
[7.0882463455200195, 3.175719976425171]
ff47bc4f-a301-4acd-a2ad-c1b2acc916f4
object-detection-and-pose-estimation-from-rgb
2101.07347
null
https://arxiv.org/abs/2101.07347v1
https://arxiv.org/pdf/2101.07347v1.pdf
Object Detection and Pose Estimation from RGB and Depth Data for Real-time, Adaptive Robotic Grasping
In recent times, object detection and pose estimation have gained significant attention in the context of robotic vision applications. Both the identification of objects of interest as well as the estimation of their pose remain important capabilities in order for robots to provide effective assistance for numerous robotic applications ranging from household tasks to industrial manipulation. This problem is particularly challenging because of the heterogeneity of objects having different and potentially complex shapes, and the difficulties arising due to background clutter and partial occlusions between objects. As the main contribution of this work, we propose a system that performs real-time object detection and pose estimation, for the purpose of dynamic robot grasping. The robot has been pre-trained to perform a small set of canonical grasps from a few fixed poses for each object. When presented with an unknown object in an arbitrary pose, the proposed approach allows the robot to detect the object identity and its actual pose, and then adapt a canonical grasp in order to be used with the new pose. For training, the system defines a canonical grasp by capturing the relative pose of an object with respect to the gripper attached to the robot's wrist. During testing, once a new pose is detected, a canonical grasp for the object is identified and then dynamically adapted by adjusting the robot arm's joint angles, so that the gripper can grasp the object in its new pose. We conducted experiments using a humanoid PR2 robot and showed that the proposed framework can detect well-textured objects, and provide accurate pose estimation in the presence of tolerable amounts of out-of-plane rotation. The performance is also illustrated by the robot successfully grasping objects from a wide range of arbitrary poses.
['M. Nicolescu', 'M. T. Chowdhury', 'S. K. Paul']
2021-01-18
null
null
null
null
['real-time-object-detection']
['computer-vision']
[ 2.59269059e-01 -1.03610419e-01 -1.17883310e-02 3.42279486e-02 -1.75584808e-01 -6.43888474e-01 1.22251861e-01 1.30745873e-01 -3.95205766e-01 1.59352690e-01 -7.60228217e-01 3.00131321e-01 -4.12373573e-01 -4.45031613e-01 -7.49582350e-01 -8.62846494e-01 -2.25783631e-01 1.20849979e+00 5.25961220e-01 6.05107322e-02 3.06677371e-01 1.12283325e+00 -1.63378501e+00 -1.48181885e-01 5.10966420e-01 9.59503353e-01 1.12005317e+00 3.03402573e-01 2.44963020e-01 -2.06808671e-02 -6.12924039e-01 1.86161190e-01 4.11539495e-01 3.17114115e-01 -4.04932529e-01 5.46370327e-01 -4.69279662e-02 -5.10577619e-01 8.38747770e-02 9.58867371e-01 1.64179206e-01 1.91223934e-01 5.04856229e-01 -1.08879006e+00 1.12665214e-01 3.29026282e-01 -4.81595337e-01 -3.79421443e-01 4.08856928e-01 1.09845780e-01 5.85603476e-01 -9.01756108e-01 6.38957620e-01 1.32588661e+00 2.17553213e-01 5.47127187e-01 -1.09252632e+00 -4.95449036e-01 1.94908813e-01 1.91000193e-01 -1.20151901e+00 -1.12673752e-01 8.21177244e-01 -4.70742315e-01 3.27904105e-01 -7.81579688e-02 3.21803868e-01 8.03573191e-01 2.15647936e-01 5.40397644e-01 7.71834433e-01 -5.96537709e-01 4.08470452e-01 -2.56271474e-02 1.85964536e-02 4.49886799e-01 5.02697527e-01 -3.20726454e-01 6.96183890e-02 -6.04118034e-02 9.19887960e-01 2.00828850e-01 -2.71223783e-01 -1.02581811e+00 -1.43434536e+00 2.59215713e-01 6.32554233e-01 3.76086354e-01 -8.39701533e-01 1.71805881e-02 1.12902462e-01 -2.11449564e-01 -3.76422763e-01 5.14373064e-01 -4.45429862e-01 5.10180928e-02 1.51296094e-01 2.35068977e-01 8.99942279e-01 1.14116335e+00 5.10583043e-01 -2.57591397e-01 2.04967231e-01 8.11723411e-01 8.54149312e-02 7.74270833e-01 1.62367582e-01 -8.15075874e-01 5.24241328e-01 6.64657176e-01 5.37083626e-01 -1.05711758e+00 -5.03212750e-01 -1.04649290e-01 -4.02216643e-01 4.86434788e-01 6.21152580e-01 2.21879795e-01 -7.82616615e-01 1.42578471e+00 6.70156300e-01 -6.10347807e-01 5.60382567e-02 1.28238273e+00 1.50622651e-01 4.12039906e-01 -1.61295652e-01 -1.93947911e-01 1.53754377e+00 -5.51252842e-01 -3.58039826e-01 -6.60527289e-01 -1.21213503e-01 -7.30612636e-01 7.23900795e-01 7.11178958e-01 -6.08510613e-01 -7.25541472e-01 -9.21515226e-01 5.22084296e-01 -3.15733626e-02 5.99145353e-01 3.50421011e-01 3.21235433e-02 -2.78055459e-01 4.18782473e-01 -1.03811789e+00 -5.96630752e-01 -1.92387193e-01 6.24095917e-01 -5.10380328e-01 -2.89259672e-01 -4.72586781e-01 1.15073574e+00 7.19124436e-01 5.40313601e-01 -7.49219596e-01 1.71086863e-01 -4.61718589e-01 -5.45502640e-02 9.12453055e-01 -1.17307499e-01 1.11574519e+00 -7.75982440e-01 -1.41177464e+00 4.84057367e-01 2.68288404e-01 1.20920837e-01 4.90528882e-01 -3.89934838e-01 1.01870447e-01 4.78016526e-01 1.59585848e-01 3.74137670e-01 1.22288108e+00 -1.55321956e+00 -4.22169775e-01 -9.83290851e-01 1.87325299e-01 3.77449185e-01 -9.56876948e-02 -1.86624676e-01 -5.85961878e-01 -1.92849621e-01 7.08658099e-01 -1.40516520e+00 1.16048371e-02 1.71884656e-01 -1.98102176e-01 -2.16579586e-01 1.22424102e+00 -4.67621595e-01 7.25947469e-02 -2.11095452e+00 5.23168802e-01 3.36733192e-01 -3.36981475e-01 1.92457691e-01 -4.94883023e-02 4.32164282e-01 1.55411869e-01 -6.41157627e-01 3.21931876e-02 1.74013779e-01 -2.56409168e-01 2.95650572e-01 -5.00063971e-02 3.67749870e-01 6.84816316e-02 2.53591508e-01 -8.24673295e-01 -1.24074183e-01 1.83352038e-01 1.74852729e-01 -5.95745444e-02 5.24634421e-01 -3.37086886e-01 5.41432440e-01 -9.18161035e-01 7.32394338e-01 5.71931958e-01 1.95524365e-01 6.16263568e-01 -5.60054541e-01 -3.74731682e-02 -3.34640324e-01 -1.51868331e+00 1.07295907e+00 -4.91477817e-01 9.51910540e-02 5.60813129e-01 -8.92500162e-01 1.35308707e+00 3.08925003e-01 4.96363610e-01 -9.30939317e-02 2.27837875e-01 4.84349698e-01 2.08189264e-01 -7.17308044e-01 2.59450644e-01 2.05901414e-01 7.98340216e-02 2.61050701e-01 -1.61634192e-01 -4.05176491e-01 1.99789122e-01 -3.96074265e-01 8.36855888e-01 1.87069133e-01 2.57950783e-01 1.62887033e-02 5.14072835e-01 -7.97882378e-02 2.45322719e-01 4.00399894e-01 1.80521905e-01 3.63879234e-01 -3.42256250e-03 -3.19629043e-01 -9.62018907e-01 -1.18827999e+00 -1.14322379e-02 7.75149941e-01 5.73519528e-01 5.40580511e-01 -5.32351375e-01 -4.10123885e-01 3.40873808e-01 4.03230786e-01 -2.58209914e-01 -1.03431106e-01 -8.54400992e-01 -2.52164304e-01 -2.04441652e-01 4.89406496e-01 3.09200376e-01 -1.45642984e+00 -1.40851963e+00 4.10697699e-01 -1.63867444e-01 -1.34283340e+00 2.73843366e-03 2.25901335e-01 -9.45956647e-01 -1.31025016e+00 -6.43104076e-01 -1.03791368e+00 1.19449997e+00 4.89419788e-01 2.59531289e-01 -5.76990731e-02 -4.79226828e-01 6.58885300e-01 -6.02757275e-01 -2.81666309e-01 -6.02238536e-01 -1.08402587e-01 3.11094224e-01 4.07392271e-02 3.74023169e-02 -1.08521909e-01 -3.41717780e-01 7.97731400e-01 -7.83510864e-01 -3.05852354e-01 8.36650789e-01 6.67102456e-01 2.77321488e-01 2.33726278e-01 5.76296806e-01 -1.60474628e-01 3.96196693e-01 -9.76580079e-04 -8.44158173e-01 3.07762593e-01 7.51568517e-03 -1.07181907e-01 5.46375513e-01 -9.25679624e-01 -8.72040629e-01 4.57847387e-01 3.89492005e-01 -5.35383344e-01 -2.48702824e-01 4.21534091e-01 -2.92095691e-01 -2.97371354e-02 4.91259962e-01 -1.31677063e-02 5.29971063e-01 -5.40594399e-01 -2.59743482e-02 7.76490331e-01 6.94052637e-01 -6.90923631e-01 8.85697186e-01 1.96941748e-01 1.38748929e-01 -8.66125166e-01 -1.92219019e-01 -4.90247846e-01 -1.06121135e+00 -5.93418837e-01 5.19997358e-01 -4.04016018e-01 -1.19984770e+00 6.98291421e-01 -1.31080127e+00 4.77661490e-02 1.61298484e-01 6.33348167e-01 -4.89282191e-01 3.49988580e-01 -2.69386798e-01 -1.05522048e+00 -3.02961409e-01 -1.48414421e+00 1.05294275e+00 1.41467378e-01 -5.24872057e-02 -1.77545577e-01 -6.75104916e-01 1.24109328e-01 1.21114165e-01 2.71039128e-01 9.35075998e-01 -4.23697621e-01 -7.11925626e-01 -7.41766095e-01 1.35815665e-01 2.25933254e-01 5.50531268e-01 -1.84855029e-01 -4.04820979e-01 -8.24574351e-01 1.68128550e-01 -3.05874676e-01 1.53260678e-01 8.13001916e-02 6.48600280e-01 -7.33366236e-02 -6.20408773e-01 -1.90578699e-01 1.20660424e+00 7.20915079e-01 2.15046197e-01 4.34634924e-01 4.63300377e-01 8.27964902e-01 1.24914515e+00 4.04733747e-01 -2.73889452e-01 9.94789124e-01 8.32686067e-01 5.25059760e-01 3.29929292e-01 3.26323658e-01 2.45330587e-01 2.16438815e-01 -1.64884344e-01 5.14293574e-02 -7.29014516e-01 2.11634502e-01 -1.65949237e+00 -5.39380372e-01 1.97328359e-01 2.43404031e+00 3.87611032e-01 1.64135456e-01 3.49866450e-02 3.23889852e-01 1.06820726e+00 -4.81207758e-01 -8.95563185e-01 -3.04151252e-02 6.39487326e-01 -1.99493930e-01 4.13699865e-01 2.00516820e-01 -7.25110650e-01 7.60361195e-01 4.39957333e+00 1.17153063e-01 -1.29520929e+00 -4.93626416e-01 -2.51736581e-01 4.04745668e-01 5.50560057e-01 -1.62324116e-01 -6.55176282e-01 3.00892681e-01 -1.62485376e-01 3.02801996e-01 4.68798846e-01 1.12396431e+00 -2.17189297e-01 -5.03591835e-01 -1.31024539e+00 6.89022183e-01 5.24842106e-02 -3.14296365e-01 -8.91514197e-02 -3.65047663e-01 3.63026671e-02 -4.55451518e-01 -8.23923945e-02 -5.62062580e-03 -3.41951162e-01 -3.12032998e-01 9.44028080e-01 3.36814582e-01 5.26772559e-01 -4.58241642e-01 7.16218650e-01 7.99246728e-01 -1.05678189e+00 -5.73734701e-01 -4.94976610e-01 -3.33513878e-02 1.51387965e-02 2.39728570e-01 -1.50786340e+00 3.38650703e-01 5.99661052e-01 -5.78215197e-02 -7.00522438e-02 9.54979241e-01 -3.29999447e-01 -1.01698086e-01 -4.88720030e-01 -3.72013301e-01 -1.84844345e-01 -1.54770836e-01 8.56769562e-01 5.74673653e-01 4.35381174e-01 2.32041225e-01 5.41960001e-01 7.21021652e-01 3.65093946e-01 -5.21303192e-02 -2.59891629e-01 8.80049467e-02 7.11854219e-01 1.29264307e+00 -1.02513003e+00 2.86189895e-02 2.19302714e-01 1.00932956e+00 2.47731432e-01 3.36760134e-01 -3.95076990e-01 -3.47056121e-01 1.30457893e-01 -8.77020434e-02 5.35857141e-01 -6.28402233e-01 2.98876047e-01 -7.34031856e-01 6.00631595e-01 -8.39188516e-01 3.18525508e-02 -8.82838726e-01 -8.57238114e-01 5.82776845e-01 2.47092217e-01 -1.32100427e+00 -4.45181668e-01 -9.09623504e-01 -1.11794993e-01 6.63393378e-01 -7.26841331e-01 -9.90732908e-01 -7.45408535e-01 1.55231819e-01 6.69397950e-01 -8.43281597e-02 7.59776115e-01 -2.49398291e-01 -1.47456512e-01 4.46945913e-02 -3.47997062e-02 -1.59158255e-03 6.28133953e-01 -7.85642684e-01 -1.95134640e-01 3.71781409e-01 -3.32201034e-01 4.83592391e-01 9.48051870e-01 -7.72218645e-01 -1.93505597e+00 -6.09078705e-01 1.88620701e-01 -1.51632980e-01 3.71120930e-01 -5.03920376e-01 -9.52145159e-01 5.52599490e-01 -7.34523475e-01 -5.29920794e-02 -3.97580892e-01 -3.69168729e-01 6.39480650e-02 -5.04598439e-01 -1.30929184e+00 4.20964271e-01 5.50162494e-01 1.10882849e-01 -6.74990356e-01 3.32613021e-01 2.70876616e-01 -6.36216521e-01 -6.39830053e-01 7.36494184e-01 1.01119268e+00 -5.37412345e-01 8.90547276e-01 -4.12868224e-02 -1.51467010e-01 -4.37894672e-01 8.33077654e-02 -1.12898326e+00 -3.82865518e-01 -9.36511308e-02 2.27931187e-01 8.32887828e-01 -2.39180356e-01 -5.85399628e-01 7.62249529e-01 4.04291421e-01 7.40939900e-02 -5.59763491e-01 -9.54013228e-01 -9.91908729e-01 -5.92113435e-01 1.01114288e-01 2.56786317e-01 2.98198223e-01 2.84885317e-02 1.43590912e-01 -4.69696783e-02 8.08392286e-01 5.88153780e-01 4.25275743e-01 1.21743536e+00 -1.39123654e+00 -3.30737978e-01 -2.58586509e-03 -6.84075952e-01 -1.01463592e+00 8.04707780e-02 -5.08077621e-01 7.01627493e-01 -1.35966992e+00 2.21373871e-01 -6.80048645e-01 1.53876334e-01 4.06461149e-01 -6.46058246e-02 -1.30793497e-01 4.64013278e-01 5.09544373e-01 -6.64191470e-02 1.01493724e-01 1.23670590e+00 -1.64848939e-01 -3.26470941e-01 6.30057096e-01 2.08645180e-01 6.71833396e-01 6.53678179e-01 -2.34343708e-01 -1.61806449e-01 -3.86117011e-01 -3.34364653e-01 4.10662085e-01 5.05405545e-01 -1.20358062e+00 1.57051861e-01 -1.33495629e-01 4.46906984e-01 -5.91936231e-01 6.14586830e-01 -1.58706212e+00 3.42346609e-01 1.00121593e+00 -9.38205346e-02 1.60217620e-02 1.91371471e-01 4.28364992e-01 1.46257490e-01 -7.28062510e-01 7.20329583e-01 -1.67042956e-01 -6.85037613e-01 -3.04102339e-02 -2.40095168e-01 -8.26044142e-01 1.42421877e+00 -4.27820534e-01 1.34227686e-02 -1.44402266e-01 -1.00351477e+00 1.62928119e-01 5.72875679e-01 6.12810075e-01 6.99690759e-01 -8.53417218e-01 -3.19830596e-01 3.20187241e-01 1.41854793e-01 2.40801200e-01 3.94403301e-02 5.63539624e-01 -5.09742796e-01 1.38886765e-01 -4.43386078e-01 -1.05876243e+00 -1.45880151e+00 8.47782016e-01 1.19138658e-01 3.35702121e-01 -5.27976394e-01 2.57342815e-01 4.05319370e-02 -2.69815743e-01 4.81040806e-01 -4.42477942e-01 -1.51211917e-01 -2.50340223e-01 1.85063675e-01 4.48775977e-01 9.58875790e-02 -4.74655211e-01 -3.39518547e-01 9.11838412e-01 -2.01368090e-02 1.69620693e-01 1.19786048e+00 5.88220470e-02 -1.64774433e-01 4.00230438e-01 7.91161716e-01 -7.48126507e-02 -1.34486973e+00 -2.42619678e-01 3.80410478e-02 -6.07355058e-01 -6.59287810e-01 -8.59429777e-01 -8.08512926e-01 6.14070237e-01 6.83642864e-01 -6.41861632e-02 9.26352859e-01 2.74939537e-01 2.97749192e-01 9.77138042e-01 1.11944950e+00 -9.67711329e-01 5.33635616e-01 3.56943697e-01 1.28747356e+00 -9.62427378e-01 6.31122515e-02 -6.69931889e-01 -4.33340341e-01 1.54141855e+00 8.44296098e-01 -2.05926403e-01 1.83219627e-01 2.06025317e-01 4.80765663e-02 -3.95555459e-02 -4.34419513e-02 2.05398232e-01 2.73954272e-01 5.81551790e-01 -4.05056417e-01 1.95138767e-01 2.33563362e-03 6.91848248e-02 9.49944183e-02 -1.55386671e-01 4.62777525e-01 1.36493981e+00 -9.41481411e-01 -7.99146891e-01 -9.41297352e-01 1.01782858e-01 -1.03009500e-01 7.95632660e-01 -8.94746035e-02 1.02037346e+00 -1.10460669e-01 7.01460004e-01 1.37893304e-01 -1.83573559e-01 7.27021158e-01 -1.08577766e-01 8.54743838e-01 -6.29006863e-01 -9.41435918e-02 2.30348170e-01 -3.40799391e-01 -3.55120897e-01 -2.08777860e-01 -7.76432812e-01 -1.55819523e+00 6.62285626e-01 -5.61512470e-01 2.19190512e-02 1.07858098e+00 8.08594823e-01 -3.00518386e-02 1.65965080e-01 7.98672676e-01 -1.49768877e+00 -1.16511095e+00 -1.04934359e+00 -5.43751121e-01 4.88904983e-01 1.04339831e-01 -1.22644925e+00 -1.97422639e-01 -1.76964298e-01]
[5.933801651000977, -0.8847728967666626]
68fefa1a-14f3-4dee-bf19-52193b4d16c3
robust-vision-using-retro-reflective-markers
2007.12514
null
http://arxiv.org/abs/2007.12514v2
http://arxiv.org/pdf/2007.12514v2.pdf
Robust Vision Using Retro Reflective Markers for Remote Handling in ITER
The International Thermonuclear Experimental Reactor (ITER)'s working environment is characterized by extreme conditions, that deem maintenance and inspection tasks to be carried out through remote handling. 3D Node is a hardware/software module that extracts critical information from the remote environment during fine alignment tasks using an eye-in-hand camera system and updates the models behind the virtual reality-based remote handling platform. In this work we develop a retro-reflective marker-based version of 3D Node that estimates the pose of a planar target, the knuckle of the cassette locking system, using the markers attached to its surface. We demonstrate a pin-tool insertion task using these methods. Results show that our approach works reliably with a single low-resolution camera and outperforms the previously researched stereo depth estimation based approaches. We conclude that retro-reflective marker-based tracking has the potential to be a key enabler for remote handling operations in ITER.
[]
2020-07-27
null
null
null
null
['stereo-depth-estimation']
['computer-vision']
[-1.17463090e-01 -2.78013609e-02 2.24160358e-01 1.27597451e-01 -7.04017460e-01 -5.95639825e-01 6.45995140e-01 -1.13170743e-01 -3.71090591e-01 3.43060881e-01 -1.73000902e-01 -3.87531728e-01 -2.13509411e-01 -2.85392433e-01 -3.97068352e-01 -5.97583354e-01 1.27030447e-01 9.67549920e-01 7.06842780e-01 -3.95339817e-01 8.01043093e-01 1.20547855e+00 -1.11152375e+00 -1.18810229e-01 -2.16302425e-01 8.79105568e-01 6.35803163e-01 8.92177343e-01 4.53633249e-01 7.31669068e-01 -5.91858566e-01 4.39166367e-01 3.27756643e-01 1.93163261e-01 -4.85185593e-01 -3.70032480e-03 -5.96518070e-02 -4.89103675e-01 -4.10754740e-01 4.44654495e-01 5.20039916e-01 -1.59082771e-03 5.55991456e-02 -6.28046751e-01 5.02816677e-01 -2.19480589e-01 -6.32816374e-01 2.37483308e-01 1.08039272e+00 1.10623337e-01 1.64622068e-01 -9.70467746e-01 1.24211025e+00 7.39155293e-01 1.06495321e+00 3.00740331e-01 -9.22143579e-01 -1.39050603e-01 -6.09736681e-01 -1.67797238e-01 -1.33539176e+00 -5.43459535e-01 6.60691202e-01 -6.06982052e-01 1.25466573e+00 4.81628537e-01 5.36253393e-01 6.46394014e-01 9.26659465e-01 -2.93844998e-01 9.76328254e-01 -8.11073542e-01 1.30180016e-01 2.76532590e-01 -5.32678142e-02 5.06905675e-01 1.64676383e-01 5.81189871e-01 -6.25564218e-01 -1.39596388e-01 1.35405040e+00 -2.23563582e-01 -5.84741235e-01 -8.16377103e-01 -1.31769383e+00 2.55012244e-01 2.69582719e-01 3.06501389e-01 -3.06877673e-01 4.50961217e-02 4.22320426e-01 2.65308380e-01 2.82362580e-01 5.70528269e-01 -2.02174574e-01 -8.19078505e-01 -6.40645385e-01 1.12849496e-01 5.54255605e-01 1.13477683e+00 2.36370891e-01 -3.67659591e-02 2.44072616e-01 2.32263505e-01 6.33580804e-01 2.23128200e-01 3.25511210e-02 -8.20095956e-01 3.72310311e-01 4.15572792e-01 4.34908599e-01 -6.01503074e-01 -6.83213115e-01 -1.64686054e-01 1.80856824e-01 9.35543776e-01 -5.54247759e-02 3.67318504e-02 -4.58096355e-01 4.60875869e-01 9.20910120e-01 -1.55060783e-01 -8.65179598e-02 9.37719166e-01 5.86130261e-01 3.07041675e-01 -6.92961514e-01 -1.37079284e-01 1.21859562e+00 -5.14221609e-01 -8.23184669e-01 -1.71516880e-01 8.71361256e-01 -1.30407953e+00 4.00268883e-01 5.36829114e-01 -1.02726769e+00 -3.48207444e-01 -1.65218437e+00 4.54718880e-02 1.59127533e-01 1.53808445e-02 3.09401602e-01 6.57958865e-01 -1.00959229e+00 7.24804640e-01 -1.04899240e+00 -2.88377702e-01 -3.68214428e-01 4.13069874e-01 -7.50100911e-01 4.02520262e-02 -5.36853313e-01 1.51302218e+00 2.62626618e-01 6.69516265e-01 -7.44734228e-01 -4.55130994e-01 -7.25139380e-01 -5.98563135e-01 3.84231657e-01 -5.72160125e-01 1.59505808e+00 2.46338069e-01 -1.98339987e+00 1.05811381e+00 2.54293799e-01 -4.34193984e-02 7.35716283e-01 -5.86286902e-01 -5.31437635e-01 3.77751827e-01 -2.38292869e-02 -4.38299865e-01 5.25162458e-01 -1.39503574e+00 -2.80459225e-01 -3.98590416e-01 -1.53425679e-01 4.77224626e-02 6.86906695e-01 5.18035769e-01 -4.46253479e-01 -6.23763166e-02 6.81841969e-01 -7.70196736e-01 -5.44271767e-02 -2.90525816e-02 -2.66580284e-01 2.30755121e-01 1.49511671e+00 -6.73053443e-01 7.48682201e-01 -1.85515380e+00 3.54659185e-03 2.62318939e-01 1.60675105e-02 9.49768722e-02 9.88893926e-01 7.49607325e-01 -1.33458033e-01 -8.80833864e-01 2.82442957e-01 -2.04157159e-01 -2.45947659e-01 -3.10152084e-01 1.40213951e-01 1.01504695e+00 -8.67762923e-01 3.37189883e-01 -6.88399374e-01 -9.49548334e-02 7.34724641e-01 4.82576966e-01 1.52395189e-01 3.58935297e-02 1.21998422e-01 7.77465165e-01 -3.64464045e-01 8.15436184e-01 9.06374395e-01 4.94981229e-01 2.58418560e-01 -2.15931430e-01 -7.34336674e-01 2.42107719e-01 -1.19077075e+00 2.08724642e+00 -5.40035903e-01 5.23487091e-01 6.85775936e-01 -1.60120934e-01 1.23278236e+00 5.96987188e-01 2.78823256e-01 -8.93306971e-01 2.40208521e-01 3.66591275e-01 -7.32895374e-01 -5.64289808e-01 1.08186948e+00 -4.72189277e-01 -2.85114739e-02 2.35101134e-01 -4.67979759e-01 -5.92564464e-01 -6.41670942e-01 2.32730031e-01 1.29550612e+00 7.25819826e-01 3.20280015e-01 -2.74409503e-01 4.34019864e-01 3.63707125e-01 3.36020768e-01 3.75477850e-01 5.94404824e-02 7.63295174e-01 4.83777560e-02 -3.66106957e-01 -1.27622950e+00 -8.42009366e-01 -3.98351312e-01 4.90170084e-02 3.70397449e-01 -5.98139405e-01 -4.32192624e-01 -3.27512324e-01 4.14939895e-02 6.96697295e-01 -6.74671233e-02 -5.04166372e-02 -7.19051957e-01 8.41711164e-02 -2.25451645e-02 6.49091661e-01 -3.18637758e-01 -5.44546008e-01 -1.04454076e+00 3.94876361e-01 4.22151029e-01 -1.19778490e+00 -8.40986297e-02 1.25875100e-01 -1.18001473e+00 -1.27381754e+00 -3.28332692e-01 -2.45324031e-01 7.73349762e-01 4.51297611e-01 1.02336383e+00 -1.01446360e-02 -4.75132555e-01 7.79576540e-01 -4.68758523e-01 -1.38144165e-01 -6.82130873e-01 -5.25681496e-01 2.59737909e-01 -6.54682636e-01 -1.69810141e-03 -9.37364176e-02 -5.84620237e-01 8.53530228e-01 -3.89583915e-01 1.23763107e-01 3.92481461e-02 3.07318419e-01 4.15954888e-01 -1.03309035e-01 -2.43201837e-01 -6.71829641e-01 9.33108628e-02 -9.22093317e-02 -1.10677540e+00 -1.49101689e-01 -3.09201956e-01 -3.15537870e-01 1.44136205e-01 3.26661050e-01 -1.25972581e+00 -7.25682676e-02 -9.94630978e-02 -3.91661286e-01 6.83003068e-02 2.22084105e-01 -1.28462061e-01 -4.84807670e-01 4.90764499e-01 -3.84991586e-01 1.91665366e-01 -7.08157718e-01 -2.26277784e-01 9.26070988e-01 8.73817384e-01 -3.89865190e-01 1.05799758e+00 8.06726575e-01 1.70415506e-01 -7.04669714e-01 -3.77269894e-01 -8.57864201e-01 -9.98439729e-01 -9.28453326e-01 6.50204897e-01 -9.52555954e-01 -8.69633436e-01 4.19797808e-01 -1.15698457e+00 -2.13817924e-01 -3.06407034e-01 9.24325228e-01 -4.74975258e-01 4.61101085e-01 -4.52407628e-01 -8.46075773e-01 -1.44697353e-01 -1.04208338e+00 1.37485313e+00 1.43665999e-01 -2.58341253e-01 -7.88534760e-01 5.32891512e-01 6.17800176e-01 3.12392205e-01 6.00320399e-01 3.25361304e-02 6.06214963e-02 -8.34304333e-01 -8.85507762e-01 4.53108996e-01 -1.59230575e-01 -1.48513466e-02 9.53373089e-02 -8.55617225e-01 -8.62531781e-01 6.65828764e-01 1.79388538e-01 8.43575895e-02 1.20001815e-01 4.02053922e-01 7.32043743e-01 -8.05157602e-01 5.40393233e-01 1.53577781e+00 4.93704319e-01 9.50290084e-01 1.18690455e+00 5.58378339e-01 2.70502210e-01 1.51811039e+00 5.76805592e-01 -5.72796576e-02 1.50154579e+00 5.98427773e-01 1.33177536e-02 3.11476290e-01 1.63783461e-01 5.53029835e-01 8.28909218e-01 -5.52145660e-01 2.43731007e-01 -1.08215082e+00 -7.58780390e-02 -1.30417144e+00 -7.72177696e-01 -8.60117495e-01 2.80435705e+00 1.64130002e-01 4.18644369e-01 -3.60979676e-01 1.19585566e-01 7.21224427e-01 -2.61289567e-01 8.62281118e-03 -5.59624314e-01 5.20357013e-01 -1.34841055e-02 7.88216293e-01 5.75272560e-01 -4.90144968e-01 4.26693439e-01 6.57937241e+00 2.24205583e-01 -8.71483326e-01 4.59735513e-01 -4.39327568e-01 -1.19925383e-02 1.77810416e-02 5.71552992e-01 -1.08159041e+00 2.42958233e-01 9.31098282e-01 1.68188810e-01 -1.33661851e-01 7.50949383e-01 3.89915556e-01 -9.11601424e-01 -1.07603860e+00 6.49547338e-01 1.08064815e-01 -1.26200223e+00 -9.54628885e-01 2.82463551e-01 2.45672628e-01 9.21892226e-02 -5.52756608e-01 -2.41159603e-01 -8.03856701e-02 -5.36440134e-01 1.10754299e+00 8.07069838e-01 7.88246155e-01 -5.50885201e-01 8.04767370e-01 3.89318913e-01 -1.12498796e+00 3.88106972e-01 -2.60241121e-01 9.80225950e-02 9.60012138e-01 6.04382813e-01 -1.35936570e+00 1.07916033e+00 6.17493272e-01 -8.31043273e-02 -2.99385071e-01 1.13796031e+00 -1.84185114e-02 -9.09261480e-02 -4.88619655e-01 5.72239935e-01 -1.06851220e-01 -2.91703403e-01 8.85880888e-01 7.08772838e-01 5.19484818e-01 -2.87684381e-01 -2.47609645e-01 1.98635682e-01 5.53005457e-01 -5.02464831e-01 -8.28898668e-01 6.19722903e-01 2.25241050e-01 1.31613278e+00 -8.29913914e-01 2.01443896e-01 -1.85418934e-01 7.88739800e-01 -3.55165511e-01 -3.10748994e-01 -9.95402396e-01 -4.22279119e-01 2.02905163e-01 6.39978707e-01 3.59040201e-01 -8.19471180e-01 -1.38877779e-02 -6.80890501e-01 4.16102380e-01 -2.89310545e-01 1.28695086e-01 -1.45106900e+00 -3.05253655e-01 4.46838617e-01 5.65787517e-02 -1.71499407e+00 -1.19227037e-01 -6.90344274e-01 -6.00998223e-01 7.79930592e-01 -9.84990299e-01 -9.43734705e-01 -4.45965976e-01 5.80300316e-02 2.50001967e-01 -1.41539469e-01 7.53888667e-01 5.87224588e-02 -5.09498596e-01 -8.18365961e-02 6.22747004e-01 -5.18261850e-01 7.02583969e-01 -1.02553010e+00 3.65581185e-01 7.36598074e-01 -3.01250607e-01 5.29272079e-01 1.06960690e+00 -1.00928462e+00 -2.29396796e+00 -2.99928516e-01 3.66474390e-01 -8.96546364e-01 6.23061657e-01 -6.34903669e-01 -4.79343325e-01 1.26742685e+00 5.34582175e-02 2.24231824e-01 5.15775122e-02 -8.41273740e-02 4.16719347e-01 -3.07817310e-02 -1.37003541e+00 5.33970371e-02 5.40961683e-01 -9.04590189e-01 -8.09143901e-01 5.39109111e-01 2.12496772e-01 -1.55711401e+00 -1.11115360e+00 1.33709967e-01 5.22595167e-01 -1.07904851e+00 1.16647017e+00 2.14544073e-01 -1.89055398e-01 -8.29599917e-01 2.85689384e-01 -1.00319326e+00 -2.61266958e-02 -1.16261435e+00 4.69434774e-04 1.13584018e+00 -9.56715178e-03 -7.84689069e-01 7.31332362e-01 5.43542027e-01 -7.85588682e-01 -2.33446121e-01 -1.52887464e+00 -7.50149071e-01 -5.77140510e-01 -3.52151573e-01 5.65657429e-02 5.53447902e-01 1.12103373e-01 -1.44239753e-01 -1.04448207e-01 8.19515228e-01 3.99925768e-01 -1.57444179e-01 9.88514841e-01 -1.08558667e+00 -3.56828004e-01 6.49316311e-01 -9.08309937e-01 -6.77642167e-01 -4.91326362e-01 -4.86325532e-01 8.58696178e-02 -1.29137075e+00 -3.01018298e-01 -5.77964723e-01 3.36131185e-01 -2.73491889e-01 7.47197151e-01 -8.06719810e-02 -1.15395889e-01 3.65345985e-01 -4.46835548e-01 1.71456710e-01 9.29552436e-01 5.34751356e-01 7.93370791e-03 -2.15408187e-02 1.17438182e-01 8.18509996e-01 3.32851529e-01 -6.92454100e-01 3.55637163e-01 -2.57055432e-01 5.41351914e-01 6.64982677e-01 6.04912400e-01 -1.27163970e+00 6.98832273e-01 3.55251670e-01 4.12395388e-01 -1.21025145e+00 5.69360614e-01 -1.25744498e+00 1.02426505e+00 5.15048265e-01 6.08975649e-01 6.00095570e-01 1.38653666e-01 4.71988559e-01 -4.75528371e-03 -8.18155408e-01 5.18905818e-01 -3.35287899e-01 -5.41111588e-01 -3.07575613e-01 -4.15356815e-01 -5.03065646e-01 1.68213558e+00 -7.60553062e-01 -5.58263481e-01 3.11406434e-01 -9.59160507e-01 -6.04025014e-02 1.13523245e+00 1.40379459e-01 6.40381217e-01 -1.10894763e+00 -1.84060618e-01 3.04035664e-01 1.09179504e-01 2.09455431e-01 4.22318369e-01 1.20893109e+00 -1.58717346e+00 5.89425564e-01 2.71299426e-02 -1.09759784e+00 -1.46127105e+00 2.75670677e-01 6.32088780e-01 -1.04140982e-01 -1.32539129e+00 5.61938941e-01 -4.98404086e-01 -3.72479141e-01 -3.39030623e-01 -1.52451083e-01 3.06135621e-02 -2.41037861e-01 5.65505743e-01 6.54562593e-01 9.91439342e-01 -7.13093519e-01 -3.89327586e-01 7.29283571e-01 -2.44412929e-01 -1.63613617e-01 1.52287257e+00 -3.33054185e-01 7.45650157e-02 6.44756734e-01 5.98740578e-01 4.59838867e-01 -1.15512586e+00 6.60693124e-02 8.79276544e-02 -8.77608538e-01 4.57026124e-01 -5.53236723e-01 -5.69606304e-01 3.09153050e-01 7.86704779e-01 1.94336288e-02 4.59691405e-01 1.20986432e-01 3.45397860e-01 1.89761728e-01 9.61171150e-01 -1.49381661e+00 -4.18741971e-01 3.98687810e-01 1.01710701e+00 -6.07147932e-01 6.58478975e-01 -5.36149323e-01 -3.31319809e-01 1.52125323e+00 4.95990396e-01 2.07932025e-01 3.14078003e-01 8.81745040e-01 1.71686098e-01 -8.90384257e-01 -4.79800433e-01 6.89514995e-01 -3.53212655e-01 5.93448341e-01 3.46085727e-01 -2.01757997e-01 1.13263391e-01 -3.62378210e-01 1.15626417e-02 -7.04715997e-02 8.48415136e-01 1.94823849e+00 -6.87771261e-01 -1.11496830e+00 -1.27773726e+00 -7.94471949e-02 -9.86116081e-02 4.94070292e-01 2.28036568e-02 1.11869764e+00 -4.28342909e-01 4.74690825e-01 1.51227370e-01 -9.82619971e-02 8.40650499e-01 -5.43478690e-02 7.72584617e-01 -5.55230737e-01 -9.51598823e-01 -7.62071908e-02 4.93641704e-01 -8.82783949e-01 -3.32512945e-01 -9.28552330e-01 -1.37353611e+00 -2.61246502e-01 -7.99426138e-01 7.90565535e-02 1.27095675e+00 8.78543437e-01 1.67406291e-01 5.09280980e-01 8.53012919e-01 -1.41736007e+00 -5.75084507e-01 -8.27889979e-01 -1.02771115e+00 -2.15671867e-01 1.11597955e-01 -1.15287781e+00 -4.55794930e-01 -5.03002405e-01]
[7.818589210510254, -1.8117672204971313]
1eb7b7af-128e-4635-8df7-7f30f38c38d6
time-series-forecasting-with-ensembled
2111.13164
null
https://arxiv.org/abs/2111.13164v6
https://arxiv.org/pdf/2111.13164v6.pdf
Neural network stochastic differential equation models with applications to financial data forecasting
In this article, we employ a collection of stochastic differential equations with drift and diffusion coefficients approximated by neural networks to predict the trend of chaotic time series which has big jump properties. Our contributions are, first, we propose a model called L\'evy induced stochastic differential equation network, which explores compounded stochastic differential equations with $\alpha$-stable L\'evy motion to model complex time series data and solve the problem through neural network approximation. Second, we theoretically prove that the numerical solution through our algorithm converges in probability to the solution of corresponding stochastic differential equation, without curse of dimensionality. Finally, we illustrate our method by applying it to real financial time series data and find the accuracy increases through the use of non-Gaussian L\'evy processes. We also present detailed comparisons in terms of data patterns, various models, different shapes of L\'evy motion and the prediction lengths.
['Tao Liu', 'Jinqiao Duan', 'Yubin Lu', 'Ting Gao', 'Luxuan Yang']
2021-11-25
null
null
null
null
['time-series-prediction']
['time-series']
[-3.76836956e-01 -3.97760361e-01 3.62310171e-01 -9.62223951e-03 7.47949211e-03 -5.09407461e-01 3.02379429e-01 -2.98857868e-01 -4.86139119e-01 1.03403580e+00 -2.40766734e-01 -3.14175665e-01 -4.78410095e-01 -6.89393580e-01 -3.35624069e-01 -9.69281256e-01 -6.32356703e-01 4.96203423e-01 9.00261402e-02 -5.58928847e-01 3.90342504e-01 4.63947862e-01 -1.13306248e+00 -4.58755583e-01 9.75135386e-01 1.24600494e+00 -2.97846228e-01 1.02183175e+00 -2.63066888e-01 9.18892801e-01 -5.53579867e-01 -1.62569225e-01 3.62220794e-01 -6.98368549e-01 -2.68184602e-01 -5.10813653e-01 -5.51950693e-01 -6.87071607e-02 -4.94618356e-01 1.33707857e+00 4.87221211e-01 3.21645707e-01 1.29539478e+00 -1.08182418e+00 -9.13981616e-01 6.50952876e-01 -5.03106177e-01 8.35137725e-01 -2.62810409e-01 3.97638381e-01 2.58179396e-01 -6.58389509e-01 6.31825864e-01 9.82198834e-01 1.37030697e+00 7.21371830e-01 -1.15878761e+00 -5.83156228e-01 -1.80101946e-01 -3.15217346e-01 -1.49187624e+00 2.46234611e-01 9.97881472e-01 -5.53170621e-01 7.92714775e-01 4.65372652e-02 6.43548429e-01 8.04823816e-01 8.25137794e-01 5.14787316e-01 1.20700157e+00 4.02914844e-02 3.54986131e-01 -1.54243493e-02 5.40347338e-01 5.58349311e-01 2.10409805e-01 1.71330258e-01 1.24763593e-01 -3.89450729e-01 8.25534701e-01 3.53523076e-01 -3.70115601e-02 2.87329674e-01 -6.67669475e-01 8.20752382e-01 -1.36850178e-01 6.59621537e-01 -6.61136687e-01 3.54037791e-01 1.21623315e-01 5.51380694e-01 8.37834001e-01 2.04407722e-01 -5.07649839e-01 -3.81898046e-01 -9.23380017e-01 5.14634132e-01 1.09271121e+00 7.77234495e-01 4.30875868e-01 5.23124635e-01 8.01842287e-02 3.75004172e-01 4.35287273e-03 1.07623076e+00 8.36765647e-01 -1.13002002e+00 6.93784803e-02 3.89038086e-01 3.23458225e-01 -1.06346405e+00 -6.22502744e-01 -4.28845465e-01 -1.56819427e+00 2.01953441e-01 4.18973863e-01 -8.02597761e-01 -4.22333091e-01 1.74065232e+00 -8.29856247e-02 4.00451630e-01 4.19406712e-01 5.04805744e-01 8.79302770e-02 1.08839583e+00 -2.98599273e-01 -7.09245026e-01 6.85605705e-01 -3.93531471e-01 -8.64358723e-01 9.10555780e-01 6.81057334e-01 -2.83317924e-01 7.65056670e-01 1.75081864e-01 -1.30581748e+00 -1.64409027e-01 -5.77544570e-01 5.03616095e-01 -4.18151557e-01 -2.13863701e-01 1.83977351e-01 3.92663449e-01 -1.22458661e+00 1.27418172e+00 -8.97895753e-01 2.79163383e-02 8.84414315e-02 3.08754981e-01 4.76937175e-01 8.49408865e-01 -1.36699402e+00 5.71014106e-01 -3.30855250e-02 1.93610504e-01 -6.66575909e-01 -8.35790992e-01 -2.09070429e-01 -8.14441312e-03 -2.49146119e-01 -5.01208663e-01 1.04657364e+00 -9.44751859e-01 -1.59617841e+00 4.20002341e-01 -2.70344287e-01 -9.22208726e-01 9.01240826e-01 1.15104102e-01 -5.60756922e-01 1.09012932e-01 -7.43270665e-02 -9.48545933e-02 7.20828414e-01 -7.80145705e-01 -2.36969814e-01 -1.03643492e-01 -7.76165605e-01 -2.10687578e-01 -1.34581909e-01 -9.96536314e-02 6.13932073e-01 -8.88077796e-01 -1.53455272e-01 -8.95032346e-01 -6.07430875e-01 -2.26763770e-01 -9.36989114e-02 -4.36070442e-01 1.03293490e+00 -5.68917513e-01 1.48573768e+00 -2.05919957e+00 -7.11375624e-02 3.94443274e-01 4.37856525e-01 2.03409910e-01 3.06064934e-01 6.75174296e-01 9.27059650e-02 3.93333882e-01 -8.49910259e-01 -2.98987269e-01 -1.34106055e-01 1.76576495e-01 -8.02481532e-01 5.38560092e-01 7.45351240e-02 7.80923009e-01 -5.87361932e-01 -1.64126903e-01 -3.42477560e-01 5.05097032e-01 -2.06322104e-01 -2.80860871e-01 -6.91947415e-02 4.37628835e-01 -5.62186718e-01 2.37041414e-01 8.04952323e-01 -2.67492145e-01 -6.65889680e-01 6.12841845e-01 -5.38870752e-01 -5.11387825e-01 -9.84917760e-01 7.40704775e-01 -2.87968844e-01 8.55673671e-01 -9.58984345e-02 -1.04005837e+00 1.31794906e+00 1.50179625e-01 5.97406507e-01 -2.28116900e-01 4.17319477e-01 5.70943952e-01 -2.63040662e-02 -4.57630277e-01 4.55186546e-01 -3.84858906e-01 6.22267351e-02 9.32774365e-01 -2.79109299e-01 -1.47324026e-01 1.19349077e-01 2.94330250e-02 9.22417581e-01 -3.34491968e-01 -2.63426095e-01 -9.04578924e-01 6.93118870e-01 3.26166153e-02 3.23844701e-01 9.02005434e-01 -3.00424814e-01 1.95111021e-01 8.87978971e-01 -6.88504875e-01 -1.35026217e+00 -8.95203531e-01 -6.36074692e-02 2.79544562e-01 2.29900524e-01 3.51998359e-01 -8.47875535e-01 1.07010268e-01 1.60378173e-01 7.27078319e-01 -9.28115487e-01 -3.16286683e-01 -8.66298735e-01 -1.21524394e+00 9.40937996e-01 1.88627943e-01 7.45673954e-01 -1.15702605e+00 -5.44960499e-01 3.52776319e-01 3.57141405e-01 -5.65535843e-01 -5.56502640e-01 -9.36274752e-02 -1.16511178e+00 -6.43206120e-01 -1.24969947e+00 -4.77649331e-01 4.44539428e-01 -2.49726012e-01 7.99071908e-01 -1.65464386e-01 -3.17254901e-01 4.14248765e-01 2.22097173e-01 -4.76520330e-01 -6.31987333e-01 -3.82223651e-02 4.65594143e-01 9.73321050e-02 4.24586952e-01 -1.05100977e+00 -1.05247343e+00 2.82267749e-01 -8.94583285e-01 -7.11849928e-01 2.16490123e-02 4.25745606e-01 5.35036266e-01 4.50586587e-01 5.56674898e-01 -2.87443578e-01 1.52894807e+00 -6.59024417e-01 -8.96175265e-01 -1.71749100e-01 -8.10279787e-01 3.44411701e-01 1.00214052e+00 -8.63517165e-01 -8.93490672e-01 -4.82509583e-01 7.04383105e-02 -5.42770445e-01 2.09853843e-01 3.39735329e-01 1.03632593e+00 -6.73566610e-02 5.64581454e-01 9.65115428e-01 3.89934093e-01 -4.00954038e-01 -1.85400434e-02 4.82338578e-01 4.18559700e-01 -3.35336179e-01 5.47021627e-01 7.86036491e-01 3.91141415e-01 -9.66378510e-01 -1.54508263e-01 -2.57103387e-02 -3.70459855e-01 -2.07457677e-01 7.76028514e-01 -3.69725794e-01 -1.41408873e+00 1.06207418e+00 -1.20059347e+00 -4.92474258e-01 -7.16336489e-01 5.15483141e-01 -8.11914265e-01 2.03542277e-01 -1.27260494e+00 -1.59370375e+00 -4.73469079e-01 -6.25270724e-01 3.52689028e-01 4.09722924e-01 -5.98521084e-02 -1.52229750e+00 6.67958558e-01 -8.69483232e-01 8.37315202e-01 4.89258468e-01 7.02280045e-01 -5.97134829e-01 -1.93256542e-01 -3.76309305e-01 1.40045539e-01 4.57091093e-01 -2.13094115e-01 4.63800073e-01 -1.12232886e-01 1.65901750e-01 8.09836447e-01 4.44997400e-01 7.92466760e-01 9.69492018e-01 7.43370652e-01 -5.08107960e-01 -1.15047783e-01 8.10891509e-01 1.49405205e+00 6.53160274e-01 3.33556950e-01 6.39741495e-02 1.41548991e-01 6.02943540e-01 -1.29571423e-01 6.21864319e-01 1.28624469e-01 -2.98580855e-01 -7.96371102e-02 3.21666449e-01 5.44233501e-01 -9.90484804e-02 4.67483610e-01 1.28029156e+00 -4.69959706e-01 -4.22486439e-02 -1.13584244e+00 3.66015494e-01 -1.68996274e+00 -1.34659934e+00 -6.61097348e-01 1.71025729e+00 5.38612902e-01 4.06661630e-02 4.50896531e-01 -9.25192237e-02 8.88681233e-01 -1.03334174e-01 -8.95949423e-01 -6.13650739e-01 -6.86977863e-01 3.14208791e-02 7.85441577e-01 6.91151381e-01 -5.54699421e-01 6.79699957e-01 7.32649565e+00 7.10389972e-01 -1.39236009e+00 7.29817897e-02 9.34605241e-01 -7.58098019e-03 -2.77394772e-01 -3.14427853e-01 -7.40381718e-01 9.63264942e-01 1.21279037e+00 -6.60980999e-01 1.85853407e-01 5.60852528e-01 5.72719216e-01 -1.55321695e-02 -2.26491913e-01 1.01138890e+00 -3.74124587e-01 -1.37029111e+00 -1.39172405e-01 1.20364934e-01 1.09952688e+00 3.14888395e-02 5.44476509e-01 7.32998326e-02 6.64744616e-01 -9.50362265e-01 2.80081958e-01 1.26286197e+00 4.36221540e-01 -9.85458374e-01 7.32877254e-01 6.40863419e-01 -9.85415161e-01 -1.24481879e-01 -4.08188969e-01 -4.51915115e-01 5.07533848e-01 6.97880328e-01 1.34117305e-01 8.92378576e-03 6.45415425e-01 6.98074400e-01 -7.01169074e-02 7.82506227e-01 4.35281724e-01 7.08532870e-01 -6.15764260e-01 -9.43465352e-01 5.37189782e-01 -7.94676423e-01 9.46605742e-01 1.05715477e+00 9.32544410e-01 5.42621791e-01 -5.48051596e-01 1.17563856e+00 4.24390644e-01 9.88193303e-02 -7.89257228e-01 -1.45837292e-01 7.38480911e-02 6.20314419e-01 -9.58167553e-01 -4.07926142e-01 2.12255463e-01 9.30025518e-01 -3.90126377e-01 7.85303116e-01 -6.65165901e-01 -7.82619178e-01 5.64745903e-01 6.61118422e-03 1.92421705e-01 -7.26379037e-01 -5.36486030e-01 -1.18833733e+00 -3.59983072e-02 -1.01549700e-01 2.52660271e-02 -6.19612217e-01 -1.53038108e+00 1.03511369e+00 -1.44155681e-01 -1.19799399e+00 -4.61967289e-01 -6.60528839e-01 -9.97751951e-01 1.07722855e+00 -8.68027389e-01 -3.69459577e-02 4.11475807e-01 7.62663126e-01 2.12135226e-01 -5.26305318e-01 3.07574481e-01 -3.12896632e-02 -5.37664354e-01 1.46428123e-01 1.11473751e+00 -8.35645199e-02 -7.81888068e-02 -1.01924455e+00 5.47013223e-01 4.74599242e-01 -6.43081546e-01 4.65944111e-01 1.32335818e+00 -8.03742290e-01 -1.01368392e+00 -8.11789632e-01 7.16062009e-01 -5.09494066e-01 1.23535943e+00 -1.06589191e-01 -1.13380313e+00 2.98902750e-01 9.59343314e-02 -1.22818917e-01 3.04790258e-01 -7.12922275e-01 3.42922837e-01 1.68581724e-01 -1.32322133e+00 6.37747228e-01 8.02056909e-01 -2.15443701e-01 -3.09028506e-01 1.61082178e-01 8.42038989e-01 8.69874060e-02 -6.93278551e-01 2.01782539e-01 4.00799155e-01 -9.09905851e-01 5.82605183e-01 -6.40555978e-01 4.77850676e-01 8.58197436e-02 1.56253770e-01 -1.08036530e+00 -2.37604156e-02 -1.25176084e+00 -2.37145245e-01 8.06824863e-01 3.83108824e-01 -1.29932892e+00 6.45392835e-01 5.38613617e-01 3.60333830e-01 -8.61536145e-01 -1.18768740e+00 -1.17681897e+00 7.57033885e-01 -3.45870465e-01 4.55535203e-01 8.49390686e-01 -1.64727911e-01 -2.88253605e-01 -5.57756305e-01 -1.25039801e-01 6.84706986e-01 -1.34153187e-01 2.79194742e-01 -1.39513290e+00 -5.89125603e-02 -9.74066556e-01 -2.07250595e-01 -1.07494581e+00 2.42852941e-01 -5.17587423e-01 -7.64307901e-02 -8.19876552e-01 -2.49112904e-01 -3.18775386e-01 -2.52121329e-01 -7.18770921e-01 1.78983569e-01 -4.74012531e-02 -1.04332879e-01 7.93963552e-01 -2.20815718e-01 7.71246374e-01 1.09195256e+00 2.59805053e-01 -7.02612579e-01 4.52011317e-01 -1.15428597e-01 8.93857181e-01 1.01291287e+00 -5.87000012e-01 -3.55165780e-01 5.71675692e-03 4.00519013e-01 4.52240288e-01 2.62264073e-01 -9.97916937e-01 4.30016100e-01 -1.01105437e-01 1.19985387e-01 -7.65686870e-01 -1.38316806e-02 -3.51851672e-01 -2.76819919e-03 9.68959689e-01 -4.50771630e-01 9.02787566e-01 1.88771375e-02 7.09766984e-01 -2.34680489e-01 -3.39913815e-01 7.93419123e-01 -1.46760747e-01 -1.17388345e-01 6.30748689e-01 -9.38555241e-01 2.69396454e-01 1.06426859e+00 -1.18050814e-01 1.29735276e-01 -8.51992846e-01 -1.06418431e+00 1.54867634e-01 1.73716024e-01 -2.76636928e-01 4.87518638e-01 -1.20392358e+00 -5.56123078e-01 1.31866947e-01 -8.03327680e-01 -3.39992762e-01 6.07767880e-01 1.00281096e+00 -8.80137801e-01 1.76451668e-01 1.08915113e-03 -5.84972560e-01 -5.67210555e-01 7.46882617e-01 8.02052736e-01 -3.47995460e-01 -5.67479074e-01 5.85388005e-01 -2.04305768e-01 -3.82060140e-01 -1.71998709e-01 -4.79631335e-01 -6.13752641e-02 3.14228572e-02 4.86007184e-01 8.41807365e-01 -8.78416300e-01 -6.78325593e-01 3.48841492e-03 1.07581532e+00 2.93063432e-01 -5.84622204e-01 1.26801157e+00 -3.83584023e-01 -4.56542552e-01 1.03447497e+00 1.42088616e+00 -2.63598293e-01 -1.44571531e+00 -1.48614168e-01 8.14360529e-02 3.67097110e-01 -6.64091706e-01 -1.37912735e-01 -7.96573460e-01 9.09507513e-01 7.64641821e-01 9.26824987e-01 9.27154064e-01 -2.47884199e-01 1.05872154e+00 4.58545506e-01 5.30089512e-02 -1.18473649e+00 -2.21304938e-01 1.02477169e+00 5.97068131e-01 -7.74280667e-01 -6.14339352e-01 8.83597806e-02 -6.08935833e-01 1.38770747e+00 7.54536763e-02 -1.22690678e+00 1.44347918e+00 4.57292140e-01 1.25628449e-02 -9.51593667e-02 -8.96016777e-01 1.93063721e-01 -3.89324874e-01 3.43386322e-01 5.73677309e-02 -1.06884733e-01 -8.38556409e-01 6.59432948e-01 -3.72298896e-01 1.98035643e-01 8.53205681e-01 4.92586434e-01 -4.50242311e-01 -4.17165577e-01 -2.88917005e-01 3.15622091e-01 -7.59513795e-01 -8.47288668e-02 -1.28561899e-01 8.52202356e-01 -3.71559054e-01 3.05422097e-01 3.89165074e-01 -2.15808660e-01 1.59230739e-01 2.31371030e-01 -1.07536271e-01 3.51454616e-01 -4.03875530e-01 -1.36836350e-01 -8.16257119e-01 4.42883074e-02 -2.13079989e-01 -7.59145319e-01 -1.26681590e+00 -6.85356617e-01 2.62343049e-01 5.79687357e-01 4.95057583e-01 8.81180167e-01 4.55206186e-01 1.94741696e-01 7.66808450e-01 -6.17860913e-01 -9.68017936e-01 -8.71711791e-01 -9.71820951e-01 1.10645860e-01 6.97824895e-01 -2.15270013e-01 -1.06867814e+00 -1.71421662e-01]
[6.719962120056152, 3.453284502029419]
cbb20fab-cd02-4cce-9b04-be7e50bdd301
potential-auto-driving-threat-universal-rain
2211.09959
null
https://arxiv.org/abs/2211.09959v1
https://arxiv.org/pdf/2211.09959v1.pdf
Potential Auto-driving Threat: Universal Rain-removal Attack
The problem of robustness in adverse weather conditions is considered a significant challenge for computer vision algorithms in the applicants of autonomous driving. Image rain removal algorithms are a general solution to this problem. They find a deep connection between raindrops/rain-streaks and images by mining the hidden features and restoring information about the rain-free environment based on the powerful representation capabilities of neural networks. However, previous research has focused on architecture innovations and has yet to consider the vulnerability issues that already exist in neural networks. This research gap hints at a potential security threat geared toward the intelligent perception of autonomous driving in the rain. In this paper, we propose a universal rain-removal attack (URA) on the vulnerability of image rain-removal algorithms by generating a non-additive spatial perturbation that significantly reduces the similarity and image quality of scene restoration. Notably, this perturbation is difficult to recognise by humans and is also the same for different target images. Thus, URA could be considered a critical tool for the vulnerability detection of image rain-removal algorithms. It also could be developed as a real-world artificial intelligence attack method. Experimental results show that URA can reduce the scene repair capability by 39.5% and the image generation quality by 26.4%, targeting the state-of-the-art (SOTA) single-image rain-removal algorithms currently available.
['Yuanjian Zhang', 'Cunjia Liu', 'Jingjing Jiang', 'Zhuoran Hou', 'Jihao Li', 'Jinchegn Hu']
2022-11-18
null
null
null
null
['vulnerability-detection']
['miscellaneous']
[ 4.03428495e-01 -2.17406765e-01 6.02156162e-01 -1.95054337e-01 -1.73037067e-01 -5.12783527e-01 3.48110735e-01 -2.24085942e-01 -3.74748111e-01 6.11999810e-01 -4.92393494e-01 -6.61762297e-01 -8.31137877e-03 -9.40915227e-01 -8.03044736e-01 -1.22983742e+00 -2.06539482e-01 -2.92039990e-01 4.62148815e-01 -7.26433694e-01 4.85122472e-01 8.84014308e-01 -2.13533926e+00 7.84647018e-02 1.14800763e+00 5.79464018e-01 4.76267278e-01 8.60451281e-01 1.58400133e-01 1.04489446e+00 -8.82494152e-01 -1.35979936e-01 3.57763648e-01 -3.70199949e-01 -2.28417441e-01 -2.09148526e-01 9.43157017e-01 -2.99229711e-01 -2.73109823e-01 1.40785277e+00 4.95248884e-01 -1.64838299e-01 2.89481461e-01 -1.05891240e+00 -3.11411232e-01 7.52194002e-02 -5.28673828e-01 9.18812752e-01 -1.16109900e-01 3.85425121e-01 9.58486050e-02 -1.02793980e+00 3.54308486e-01 1.21083224e+00 5.07935941e-01 4.02252764e-01 -5.91716588e-01 -8.24774504e-01 -1.67317957e-01 6.19481981e-01 -1.06803834e+00 -2.76046872e-01 4.56566751e-01 -2.91633219e-01 6.01766586e-01 7.86707103e-01 6.46856666e-01 7.48317659e-01 5.86285353e-01 2.76645005e-01 1.51051724e+00 -3.70946527e-01 3.19755435e-01 1.27456963e-01 5.19799590e-01 6.71648204e-01 9.84063208e-01 7.08921015e-01 -3.74770731e-01 1.88293129e-01 2.91222602e-01 -1.33473039e-01 -4.60507780e-01 1.32348612e-01 -3.73950064e-01 9.97622073e-01 8.78594577e-01 1.36215091e-01 -3.74744207e-01 -1.42823383e-01 -2.88069844e-02 5.81493914e-01 4.33691442e-01 5.78184128e-01 5.91430664e-02 5.37854195e-01 -8.50366831e-01 5.06986499e-01 5.91574252e-01 1.85771838e-01 9.16403949e-01 8.00367117e-01 9.92399007e-02 2.60440916e-01 1.40512973e-01 1.32267094e+00 -7.12801367e-02 -7.75267780e-01 2.52691060e-01 2.53807992e-01 2.13039979e-01 -1.16636920e+00 -3.86613101e-01 -6.00004792e-01 -8.46540213e-01 1.39049065e+00 2.90177882e-01 -8.58331025e-02 -1.28729200e+00 1.13680279e+00 3.89653981e-01 4.27670896e-01 4.55916315e-01 1.16306496e+00 9.01057839e-01 7.10228324e-01 -2.25000352e-01 -3.60379934e-01 1.29300237e+00 -7.11424649e-01 -8.06598902e-01 -6.91132784e-01 9.19453576e-02 -7.53607154e-01 7.05450058e-01 4.85416144e-01 -7.28802204e-01 -6.52505755e-01 -1.51550865e+00 5.23462057e-01 -6.13473237e-01 -4.92896765e-01 5.05126715e-01 1.09876633e+00 -8.77010286e-01 3.29189956e-01 -3.27347368e-01 -3.07355732e-01 2.88697422e-01 1.47062644e-01 -1.23326704e-01 -5.51063895e-01 -1.36088264e+00 1.42748666e+00 1.08498201e-01 6.46118760e-01 -1.18163824e+00 -7.58603334e-01 -6.69373453e-01 -2.82610089e-01 5.13975173e-02 -4.78863746e-01 5.36115587e-01 -1.21168339e+00 -9.70382452e-01 8.05511355e-01 -1.31331310e-01 -1.03209591e+00 2.44635433e-01 -6.45868063e-01 -6.78914070e-01 2.76680499e-01 -1.33579910e-01 1.71678156e-01 1.58032954e+00 -1.93376267e+00 -8.49737108e-01 -4.36974883e-01 5.16472310e-02 2.44899035e-01 2.08473265e-01 1.79926157e-01 2.12329432e-01 -5.73385179e-01 -5.79249300e-02 -1.04407883e+00 -4.55379993e-01 -3.26147467e-01 -6.19995818e-02 4.52880353e-01 1.26139987e+00 -8.09415638e-01 8.68855357e-01 -2.14008856e+00 -2.66379505e-01 3.43860269e-01 2.21434668e-01 1.12723553e+00 -1.72848836e-01 1.17344856e-01 -1.06011912e-01 -6.02235459e-02 -6.47717595e-01 1.31764099e-01 -6.64406240e-01 4.60566282e-01 -9.54307437e-01 8.22815299e-01 2.86947817e-01 5.56422889e-01 -6.19815767e-01 1.64230064e-01 3.77857000e-01 4.92940605e-01 1.15017653e-01 4.61839676e-01 1.92021057e-01 3.04470897e-01 -1.57248765e-01 7.30527878e-01 1.41755927e+00 8.52872670e-01 -2.70032316e-01 -7.77047779e-03 -3.28801036e-01 -3.54319453e-01 -8.78893435e-01 6.24636889e-01 -1.83571160e-01 9.57820058e-01 1.56991094e-01 -7.58716106e-01 1.10121226e+00 -9.53740254e-02 -4.19059187e-01 -1.01102424e+00 1.54947802e-01 1.86017245e-01 2.05332741e-01 -1.05883253e+00 5.90992689e-01 -1.19256422e-01 3.30158234e-01 4.81837206e-02 -6.08483374e-01 -3.63571614e-01 -9.03953686e-02 7.84823000e-02 1.23173475e+00 -3.80373120e-01 -2.10196033e-01 -1.47807121e-01 4.61932629e-01 1.16799630e-01 5.22936702e-01 1.08786893e+00 -3.25810581e-01 6.62696600e-01 -2.20700145e-01 -6.53092146e-01 -7.72813499e-01 -1.10012770e+00 -1.40540406e-01 6.63619697e-01 5.25729656e-01 3.78903359e-01 -9.47251260e-01 -4.46396828e-01 7.89243728e-02 1.02809834e+00 -7.55717635e-01 -3.79394889e-01 -6.73752069e-01 -1.12513065e+00 8.24247360e-01 9.88434032e-02 7.10077465e-01 -1.34484541e+00 -1.10798252e+00 -7.89442807e-02 -5.06400317e-02 -1.24086726e+00 4.97870833e-01 2.37701699e-01 -8.61091912e-01 -1.37284875e+00 -2.70627052e-01 -5.40136039e-01 6.00377619e-01 1.15691900e+00 1.09839356e+00 6.46915913e-01 -8.58349979e-01 5.26337400e-02 -6.22996151e-01 -9.01538551e-01 -5.69071651e-01 -7.55817711e-01 7.86013678e-02 1.11569069e-01 2.67399132e-01 -5.70113957e-01 -6.23923421e-01 1.92908794e-01 -1.19309843e+00 -6.16419613e-01 6.56514287e-01 4.84705925e-01 7.86923394e-02 5.46979070e-01 4.36487198e-01 -8.73828650e-01 4.49681729e-01 -5.98125398e-01 -6.62350535e-01 -4.24597748e-02 -5.40946543e-01 -2.03125939e-01 2.78002650e-01 -6.39775619e-02 -1.25531089e+00 6.16980828e-02 6.78018406e-02 -5.27251005e-01 -4.38533157e-01 4.99304742e-01 -5.25854826e-02 -7.21482396e-01 9.47186232e-01 2.24945858e-01 2.63251588e-02 -1.83851406e-01 4.66495305e-01 4.08328623e-01 7.61122882e-01 3.04299563e-01 1.72402442e+00 1.00989592e+00 2.57326484e-01 -1.33029866e+00 -7.87131548e-01 -4.75943863e-01 -2.87821382e-01 -6.33602560e-01 8.60635400e-01 -9.83398855e-01 -1.91236138e-01 8.91785264e-01 -1.03642821e+00 -1.32558599e-01 3.72176506e-02 1.64618567e-01 1.05097130e-01 6.29329205e-01 -8.16536546e-02 -1.19879198e+00 -6.09229863e-01 -8.14457417e-01 3.02392662e-01 4.50486392e-01 3.61804634e-01 -4.48109657e-01 1.27052262e-01 5.85500062e-01 6.47195399e-01 4.39754099e-01 6.69386923e-01 7.12447912e-02 -7.66754031e-01 -9.26552266e-02 -2.16081440e-01 5.40840864e-01 4.32001427e-02 1.08954690e-01 -1.40556312e+00 -5.20563304e-01 6.10207796e-01 -2.73430045e-03 1.39031005e+00 2.86389738e-01 3.77186269e-01 -1.69952527e-01 6.61383718e-02 7.34979153e-01 1.62468112e+00 2.07717136e-01 1.39414155e+00 7.90034115e-01 6.57519758e-01 7.94325352e-01 1.08709788e+00 1.62422940e-01 -2.83451647e-01 2.38415956e-01 1.42075837e+00 -5.47726929e-01 -2.88751453e-01 4.87081110e-01 6.38291538e-01 2.48956710e-01 -4.18770105e-01 -1.62416905e-01 -6.55702174e-01 6.65607810e-01 -1.54650950e+00 -1.33856487e+00 -7.31107712e-01 2.19975686e+00 2.81326294e-01 3.17474097e-01 -3.90963316e-01 3.48737359e-01 7.05854356e-01 3.11909229e-01 -4.62148815e-01 -6.41351283e-01 -5.57300806e-01 5.41062891e-01 1.00466096e+00 6.03491783e-01 -1.22912085e+00 1.06656432e+00 5.89500427e+00 5.58113098e-01 -1.22618341e+00 -5.29311635e-02 1.67175129e-01 1.10303782e-01 -3.95362191e-02 7.21896589e-02 -7.19386637e-01 3.95878226e-01 9.34094846e-01 1.96052194e-01 3.02218497e-01 6.62106931e-01 4.57573324e-01 -5.11103988e-01 -6.93455562e-02 6.71426356e-01 5.26690543e-01 -1.00793076e+00 1.01930119e-01 -1.46957815e-01 5.17311215e-01 1.97704896e-01 3.68456036e-01 -5.66694736e-02 2.11488113e-01 -1.28537762e+00 5.59134603e-01 5.80771208e-01 3.65301549e-01 -1.03638434e+00 8.84288967e-01 1.90271944e-01 -7.71860898e-01 -3.27860147e-01 -7.77632833e-01 -3.31968278e-01 -4.07667868e-02 9.03635740e-01 -7.38802135e-01 5.89300931e-01 1.23640001e+00 1.87685221e-01 -8.79817486e-01 1.34668791e+00 -6.04727983e-01 9.64633584e-01 -2.30181545e-01 4.79548901e-01 2.62925744e-01 -3.15529466e-01 1.09246230e+00 1.25736928e+00 8.79283771e-02 1.30816340e-01 -1.97874010e-01 3.90414923e-01 4.34287459e-01 -4.88806009e-01 -1.05004704e+00 5.18671215e-01 3.05651128e-01 1.19192076e+00 -5.96033692e-01 -1.97593179e-02 -3.16626690e-02 9.28635418e-01 -3.46848071e-02 5.37516177e-01 -8.81064057e-01 -4.68723238e-01 1.10655808e+00 3.13213058e-02 4.82507288e-01 -1.68237403e-01 -3.62830341e-01 -6.00413322e-01 3.83893214e-02 -1.04860187e+00 -8.43878612e-02 -8.66769791e-01 -1.05069995e+00 9.57124114e-01 -3.09752822e-01 -1.33097553e+00 1.67448878e-01 -5.73365808e-01 -1.05611849e+00 8.92002165e-01 -2.33441734e+00 -1.04687631e+00 -8.01605880e-01 6.77435398e-01 5.11013508e-01 -3.28649223e-01 6.71191514e-01 -4.16236408e-02 -4.55879211e-01 2.35296935e-01 1.89364448e-01 -2.91413993e-01 6.84324384e-01 -9.72779334e-01 3.85312289e-01 1.76226532e+00 -1.09059952e-01 2.81545609e-01 1.37143219e+00 -6.60661519e-01 -1.46481705e+00 -1.42727661e+00 5.89368284e-01 -3.25451165e-01 3.18464428e-01 -2.86194850e-02 -1.30822802e+00 1.63826182e-01 4.65376049e-01 -7.46387839e-02 1.58902630e-01 -3.67134809e-01 -6.45645976e-01 -2.91619509e-01 -1.27268434e+00 5.69916964e-01 4.66275305e-01 -3.43508899e-01 -8.40572238e-01 2.08139896e-01 6.64812863e-01 -1.81143999e-01 -5.63774817e-02 6.73314095e-01 2.50236899e-01 -1.50058246e+00 1.14831185e+00 -3.12531620e-01 2.15112478e-01 -7.04693079e-01 -1.21264584e-01 -1.25806940e+00 -4.16718364e-01 -3.88218135e-01 -3.24626751e-02 8.54503691e-01 8.95558745e-02 -7.54317224e-01 5.73657751e-01 3.48085128e-02 -2.49682054e-01 -1.11693546e-01 -1.10731137e+00 -8.17750156e-01 -1.35663722e-03 -3.04319292e-01 1.18666604e-01 1.00124347e+00 -7.91018963e-01 -2.47277506e-02 -5.15206933e-01 1.24283659e+00 9.91031170e-01 2.96207927e-02 8.93766224e-01 -1.18711746e+00 1.81581795e-01 -1.13657884e-01 -4.97531772e-01 -1.02985978e-01 -1.39514595e-01 -3.06698024e-01 4.54042464e-01 -1.20359135e+00 -1.91801339e-01 -1.88921824e-01 -3.31866562e-01 2.61743546e-01 -5.98625958e-01 5.64269483e-01 4.27556187e-01 3.31287980e-01 4.97864075e-02 3.60455155e-01 9.26809788e-01 -2.42453501e-01 -6.71257405e-03 1.35023609e-01 -3.37983876e-01 8.11768472e-01 1.14428091e+00 -8.71784031e-01 -3.38756979e-01 -3.46802741e-01 1.64245158e-01 -3.49457949e-01 7.58132577e-01 -1.57732892e+00 1.02557451e-01 -2.78247267e-01 -4.83389013e-02 -6.81007028e-01 3.76584440e-01 -8.64564300e-01 2.37370911e-03 8.33132684e-01 5.35677493e-01 -8.22905451e-02 3.85423094e-01 6.88592911e-01 -3.09430838e-01 -3.49397719e-01 1.15444243e+00 -1.35923639e-01 -1.08372700e+00 -2.08970696e-01 -1.09020269e+00 -4.09871042e-01 1.05358183e+00 -3.84336293e-01 -8.06495249e-01 -2.08626345e-01 -3.19631785e-01 -3.42538133e-02 2.77967304e-01 4.69965935e-01 1.09915030e+00 -4.51772690e-01 -1.06675673e+00 3.32979023e-01 -5.29685058e-03 -3.63460541e-01 5.08192658e-01 4.05052602e-01 -7.87881494e-01 -1.07260443e-01 -5.38697302e-01 -3.46710116e-01 -1.95929813e+00 7.38496900e-01 4.06747699e-01 5.19284494e-02 -7.49424338e-01 8.54507565e-01 4.02688265e-01 9.11354125e-02 7.60407373e-02 3.93347405e-02 -6.09754384e-01 -6.10523997e-03 9.86693323e-01 3.75165343e-01 3.79581809e-01 -7.08815753e-01 -2.49193549e-01 4.73189533e-01 -1.38822883e-01 2.79828817e-01 1.19918251e+00 -2.12820217e-01 -2.63149142e-01 1.72031552e-01 4.20253724e-01 2.00615469e-02 -1.05946648e+00 -2.08345260e-02 -2.52074093e-01 -7.13647485e-01 3.49473417e-01 -8.84756446e-01 -1.46234477e+00 9.43328083e-01 1.34567153e+00 3.10826391e-01 1.23988962e+00 -5.99237084e-01 6.34313405e-01 5.92885971e-01 9.18470472e-02 -6.90860808e-01 5.57310358e-02 7.32237637e-01 1.07536626e+00 -1.18939185e+00 1.63307294e-01 -6.35655284e-01 -6.67955458e-01 9.84163821e-01 4.43827808e-01 -5.02844632e-01 6.89289391e-01 3.38651299e-01 7.75836289e-01 -3.93378675e-01 -2.44148418e-01 -5.29596686e-01 -1.71832547e-01 9.89761114e-01 -3.92606944e-01 5.89076467e-02 -1.14497967e-01 1.19489037e-01 -1.28252320e-02 -1.84223607e-01 9.89726186e-01 1.12816548e+00 -1.15401495e+00 -6.21199429e-01 -1.19802880e+00 6.79583922e-02 -1.73727140e-01 -3.85960788e-01 -1.46733209e-01 5.48730850e-01 4.12129223e-01 1.41764247e+00 -2.69393712e-01 -5.11369467e-01 3.68210852e-01 -2.96817601e-01 1.02935120e-01 -2.18352109e-01 -6.88079953e-01 -5.08855462e-01 7.75925145e-02 -4.75647181e-01 -6.01351738e-01 -4.87442344e-01 -8.92985225e-01 -3.50942254e-01 -3.35444361e-01 1.72264501e-02 8.69002879e-01 9.71662462e-01 2.72465706e-01 5.07680893e-01 9.33226883e-01 -9.79114056e-01 -1.65125832e-01 -6.98896170e-01 -8.14681768e-01 1.75148234e-01 7.19107926e-01 -6.55055881e-01 -8.59113514e-01 -5.68790287e-02]
[10.915060043334961, -3.263864517211914]
957b1ef9-cd68-4971-b652-4ba3b734db3a
persistent-dirac-for-molecular-representation
2302.02386
null
https://arxiv.org/abs/2302.02386v1
https://arxiv.org/pdf/2302.02386v1.pdf
Persistent Dirac for molecular representation
Molecular representations are of fundamental importance for the modeling and analysis of molecular systems. Representation models and in general approaches based on topological data analysis (TDA) have demonstrated great success in various steps of drug design and materials discovery. Here we develop a mathematically rigorous computational framework for molecular representation based on the persistent Dirac operator. The properties of the spectrum of the discrete weighted and unweighted Dirac matrices are systemically discussed and used to demonstrate the geometric and topological properties of both non-homology and homology eigenvectors of real molecular structures. This allows us to asses the influence of weighting schemes on the information encoded in the Dirac eigenspectrum. A series of physical persistent attributes, which characterize the spectrum of the Dirac matrices across a filtration, are proposed and used as efficient molecular fingerprints. Finally, our persistent Dirac-based model is used for clustering molecular configurations from nine types of organic-inorganic halide perovskites. We found that our model can cluster the structures very well, demonstrating the representation and featurization power of the current approach.
['Kelin Xia', 'Ginestra Bianconi', 'JunJie Wee']
2023-02-05
null
null
null
null
['topological-data-analysis']
['graphs']
[ 2.83046782e-01 -3.73213410e-01 -1.63497329e-01 -9.29235965e-02 -1.64858624e-01 -4.46240693e-01 6.33700311e-01 3.88725907e-01 -1.31246090e-01 7.68718839e-01 1.21597402e-01 -5.47339559e-01 -6.30238235e-01 -9.17849362e-01 -4.02224332e-01 -1.26434302e+00 -6.92664862e-01 3.31872195e-01 5.82514517e-02 -3.92846376e-01 6.22506261e-01 7.44002044e-01 -1.59746289e+00 3.32972258e-01 1.13972640e+00 7.99504817e-01 9.49415714e-02 3.92585933e-01 -5.74453138e-02 2.83076227e-01 -1.62886128e-01 -1.81144163e-01 -2.18202304e-02 -3.71408373e-01 -6.16752505e-01 -1.31757900e-01 2.83982247e-01 4.66019571e-01 -4.11097974e-01 1.09137726e+00 3.93317908e-01 3.10507357e-01 1.25137627e+00 -8.02547157e-01 -9.06880021e-01 3.76382232e-01 -2.62073368e-01 -8.98384675e-02 5.37223160e-01 -6.19124472e-02 9.73106146e-01 -1.08019233e+00 9.69957054e-01 1.14577925e+00 8.40128362e-01 2.80225396e-01 -1.63216662e+00 -3.93157661e-01 -4.09086049e-01 5.03313541e-01 -1.56536889e+00 -2.64717907e-01 7.16406703e-01 -7.51450241e-01 6.66699350e-01 5.08798063e-01 7.46073186e-01 7.44563103e-01 5.98009229e-01 -7.14871883e-02 1.28057396e+00 -6.91409171e-01 4.97480780e-01 9.45359876e-04 5.97547829e-01 9.49313581e-01 7.73500741e-01 4.76497263e-02 -5.30565858e-01 -6.77018344e-01 4.34349954e-01 4.98476923e-02 -3.24055791e-01 -8.68880689e-01 -1.02953172e+00 8.25579166e-01 2.85169125e-01 4.14131582e-01 -4.81599241e-01 2.18313470e-01 1.22217059e-01 -7.58124366e-02 1.07535221e-01 4.17255789e-01 3.37072015e-01 1.91557392e-01 -5.06356418e-01 1.49572253e-01 6.42595589e-01 5.02550900e-01 6.88525975e-01 -3.88675183e-02 7.42898285e-02 6.03087068e-01 4.52910423e-01 3.68465006e-01 3.93845551e-02 -8.14510167e-01 -5.91501482e-02 5.09741008e-01 -1.04108441e-03 -1.18311930e+00 -4.80125755e-01 -3.34394351e-02 -8.37068439e-01 3.91296118e-01 3.41794014e-01 4.93788719e-01 -7.87648261e-01 1.49454439e+00 3.40130985e-01 -3.35090667e-01 1.76791474e-02 4.98397857e-01 6.77523136e-01 5.77650249e-01 3.07094365e-01 -5.11620939e-01 1.18436801e+00 -2.63562929e-02 -7.03394175e-01 9.44223642e-01 4.52688307e-01 -6.33296013e-01 7.59572268e-01 4.04598475e-01 -1.13002229e+00 -1.88387394e-01 -1.05252421e+00 3.01827341e-01 -5.34669518e-01 -1.20121188e-01 8.89466524e-01 7.61024714e-01 -8.39369893e-01 1.28351629e+00 -8.41388226e-01 -3.25594038e-01 2.58727700e-01 5.92158914e-01 -5.69134295e-01 6.00341037e-02 -9.23385859e-01 8.80407870e-01 4.13021952e-01 2.45139524e-02 -3.86134028e-01 -6.50601208e-01 -4.25349832e-01 -2.44869992e-01 -3.42945963e-01 -4.86768335e-01 3.91502947e-01 -1.78836614e-01 -1.17908502e+00 7.03874111e-01 -1.47519127e-01 -2.34859765e-01 3.32558341e-02 6.03909850e-01 -5.45823574e-01 2.62868196e-01 -8.27907622e-02 2.58873492e-01 4.60538864e-01 -1.36589420e+00 1.65055841e-01 -3.97492856e-01 -3.29673171e-01 -2.49350771e-01 -3.04883182e-01 -3.38302463e-01 3.16787839e-01 -5.79779983e-01 5.14315009e-01 -9.49998081e-01 -3.35529268e-01 -2.05227286e-01 -5.97933114e-01 -1.72811598e-02 4.68980968e-01 -3.41701835e-01 1.30030072e+00 -2.10043049e+00 5.21133304e-01 9.75974500e-01 3.83750767e-01 4.54441123e-02 -1.13439234e-02 1.04115164e+00 -4.40633774e-01 1.69544533e-01 -3.54549825e-01 4.61477429e-01 6.67787790e-02 5.12748808e-02 -1.94914326e-01 7.91532874e-01 -1.98309779e-01 4.22219515e-01 -8.17397237e-01 -2.29044050e-01 9.28527117e-02 7.70522237e-01 -5.97119868e-01 -2.60625213e-01 -8.33705906e-03 3.01557899e-01 -3.98813903e-01 5.65720141e-01 9.00782347e-01 -1.27226472e-01 6.80269957e-01 -6.96473956e-01 -5.26096344e-01 4.41150725e-01 -9.51906681e-01 1.31943047e+00 2.15669155e-01 2.59974480e-01 -7.58044794e-02 -8.19562376e-01 7.76718438e-01 6.46514520e-02 7.54368782e-01 -7.17198312e-01 -3.31763148e-01 3.65025938e-01 4.65788901e-01 -2.81806350e-01 5.34546018e-01 -3.94987077e-01 1.44774690e-01 4.64138657e-01 4.22713831e-02 1.00051731e-01 3.79087359e-01 2.62845784e-01 7.85747409e-01 -1.80641830e-01 2.04251960e-01 -1.00778091e+00 6.02659225e-01 1.25006884e-01 1.81191459e-01 4.03067917e-01 1.93141967e-01 2.59541333e-01 4.80579466e-01 -5.58547258e-01 -1.18243754e+00 -1.33233619e+00 -8.29235315e-01 7.83713281e-01 1.78327739e-01 -7.21632659e-01 -6.06021762e-01 8.65768865e-02 3.12776864e-01 5.67321599e-01 -6.43522263e-01 -4.67193574e-01 -3.14146727e-01 -1.31497931e+00 4.04394716e-01 7.19147548e-02 -1.15100853e-01 -7.27939904e-01 -3.16131443e-01 1.91802591e-01 3.10773671e-01 -5.01986921e-01 -2.01138154e-01 3.94481480e-01 -9.02261972e-01 -1.21400666e+00 -2.51677394e-01 -4.47960824e-01 6.81527913e-01 2.32858613e-01 6.78838193e-01 8.49072784e-02 -7.55444705e-01 5.06317019e-01 -1.20018451e-02 6.50329655e-03 -6.27629101e-01 -3.67842376e-01 7.26589501e-01 5.70624284e-02 3.24307643e-02 -8.99774373e-01 -6.94368303e-01 2.57495642e-01 -8.02599192e-01 -2.63885826e-01 2.61115760e-01 7.16832757e-01 7.87169516e-01 1.15295267e-02 3.30983907e-01 -6.43587828e-01 8.77786279e-01 -3.91597241e-01 -3.93144935e-01 3.69717181e-01 -5.46829462e-01 5.43600500e-01 4.21016604e-01 -2.89180577e-01 -6.25291169e-01 -8.73006210e-02 -9.79464594e-03 1.44519061e-02 2.07749441e-01 4.77866709e-01 -4.99915630e-02 -8.24812591e-01 6.39471889e-01 3.60313296e-01 5.88782430e-02 -5.20743370e-01 4.00871933e-01 4.23883587e-01 3.58400345e-01 -9.88838971e-01 5.09903193e-01 5.50664723e-01 7.90569782e-01 -1.36694920e+00 -6.26374856e-02 -2.14395612e-01 -6.68175757e-01 -5.75122088e-02 8.27557147e-01 -3.36026251e-01 -1.31683719e+00 4.26724032e-02 -9.92723107e-01 3.35180998e-01 -2.66660243e-01 4.48972821e-01 -6.09695792e-01 7.93429196e-01 -5.62620223e-01 -9.23616409e-01 -1.78515419e-01 -1.21425748e+00 5.35072446e-01 -2.37483174e-01 -3.26717317e-01 -1.22078395e+00 4.52852249e-01 2.77724620e-02 3.41541111e-01 5.81480324e-01 1.83093607e+00 -3.30961585e-01 -6.39797091e-01 7.42053911e-02 1.56319886e-01 -1.92059442e-01 1.01230554e-01 4.47482646e-01 -6.08528256e-01 -2.19536275e-01 -3.32305908e-01 3.21759246e-02 9.99781013e-01 3.76873404e-01 1.00332832e+00 -2.28813186e-01 -4.75371122e-01 5.26119649e-01 1.38288271e+00 5.35214067e-01 7.32236862e-01 6.04652129e-02 7.17298925e-01 5.29864609e-01 7.35480860e-02 4.37885702e-01 -9.60691869e-02 7.78846323e-01 3.01084131e-01 1.68639958e-01 9.89992842e-02 -2.39752773e-02 3.87114763e-01 1.12859821e+00 -7.08255053e-01 1.66728169e-01 -9.61567938e-01 8.68700817e-02 -1.58018088e+00 -1.37542105e+00 -6.33031309e-01 2.48584938e+00 7.22429991e-01 -2.36967444e-01 1.53093755e-01 1.83006316e-01 6.38062119e-01 -7.01079741e-02 -2.75199085e-01 -6.51714504e-01 -2.71839023e-01 6.20213032e-01 4.39527899e-01 5.37493467e-01 -8.33389938e-01 3.66109192e-01 8.07449627e+00 7.75757730e-01 -9.13042307e-01 -2.48898298e-01 4.93558608e-02 3.62372637e-01 -6.86054885e-01 3.34442593e-02 -4.22715724e-01 5.53297043e-01 9.89222467e-01 -2.74937391e-01 2.90156901e-01 4.72536266e-01 1.06349669e-01 -8.89953226e-02 -9.74512160e-01 8.64234209e-01 -4.12768751e-01 -1.83623910e+00 6.90281630e-01 5.15059888e-01 6.15771830e-01 -3.01029414e-01 3.69026482e-01 -4.90070313e-01 -3.25190052e-02 -1.18452013e+00 5.21755219e-01 9.25468922e-01 9.09005880e-01 -8.91226113e-01 1.01771012e-01 -2.71834701e-01 -1.25277245e+00 4.89650816e-02 -5.44989228e-01 1.56262934e-01 -7.99088851e-02 5.70976794e-01 -5.01241803e-01 5.64045012e-01 1.02378681e-01 6.46229446e-01 -3.62399071e-01 1.18346429e+00 5.35308480e-01 2.98959732e-01 -1.97163656e-01 -3.48750278e-02 1.67213291e-01 -1.14254022e+00 5.89586198e-01 1.08975291e+00 4.10003960e-01 2.09507972e-01 -6.02686591e-03 1.10537732e+00 8.51701722e-02 1.56854227e-01 -7.99771249e-01 -5.85983872e-01 3.76519442e-01 1.06478035e+00 -7.69599378e-01 7.89567269e-03 2.74681542e-02 4.74195749e-01 -4.07536775e-02 2.89176136e-01 -5.05918562e-01 -4.93931681e-01 1.07152259e+00 3.84881973e-01 2.34624892e-01 -6.39990091e-01 -5.18757850e-02 -9.03075635e-01 -2.71716863e-01 -5.84081948e-01 1.22026153e-01 -3.36460918e-01 -1.24846172e+00 2.80123651e-01 1.90004259e-02 -1.03914452e+00 1.84162900e-01 -1.03121579e+00 -7.92147458e-01 7.81731606e-01 -1.00003123e+00 -6.16595447e-01 6.94943666e-02 6.01307094e-01 -5.19766450e-01 -1.20816782e-01 1.15500581e+00 3.90044510e-01 -5.23868680e-01 2.75266737e-01 8.95768046e-01 -4.44366038e-01 1.85620934e-01 -1.18069959e+00 8.05449709e-02 1.50854394e-01 2.27727331e-02 1.22760785e+00 8.64033341e-01 -6.16390765e-01 -1.93472850e+00 -7.55374610e-01 4.69315588e-01 -3.70280743e-01 7.60160029e-01 -3.26761395e-01 -9.89940345e-01 7.66132101e-02 -1.14581063e-01 -1.77837521e-01 1.07352936e+00 -1.89683214e-02 -3.61590236e-01 6.92788959e-02 -1.10083175e+00 5.20430684e-01 1.15254414e+00 -7.01045811e-01 -3.45915735e-01 7.25642383e-01 2.76266098e-01 2.12833986e-01 -1.20731390e+00 2.67988950e-01 7.54811823e-01 -1.24906635e+00 1.37450325e+00 -9.58764136e-01 -4.72219288e-02 -4.64817971e-01 -5.58975875e-01 -9.79227662e-01 -7.26786256e-01 -6.11418962e-01 5.67408390e-02 6.73090398e-01 2.96399266e-01 -6.82468891e-01 4.80239153e-01 2.78084278e-01 -1.03712752e-01 -6.84091568e-01 -1.06816435e+00 -8.46489131e-01 1.79333761e-01 3.64737995e-02 2.51419693e-01 9.42748308e-01 4.82604027e-01 1.37317464e-01 -1.86199304e-02 -4.85639945e-02 9.53819752e-01 2.47609749e-01 -7.38945901e-02 -1.54156911e+00 8.41458980e-03 -3.83953452e-01 -8.95453751e-01 -1.94864541e-01 6.90975934e-02 -1.36666441e+00 -4.87063885e-01 -1.16887236e+00 3.72124046e-01 -7.65993953e-01 -4.37397391e-01 7.94141144e-02 5.58999836e-01 1.96415752e-01 -4.14795987e-02 5.01822829e-01 -5.65870464e-01 5.18267632e-01 1.10535276e+00 -4.45213132e-02 -1.37621403e-01 -3.43618006e-01 -3.72112662e-01 6.63850784e-01 5.28574586e-01 -3.77211124e-01 -1.75638378e-01 3.63573968e-01 4.00539041e-01 -1.63895279e-01 4.58251476e-01 -1.11485982e+00 6.22224901e-03 -2.37573728e-01 2.67281920e-01 -5.17227232e-01 5.60288727e-01 -7.15862751e-01 7.82104254e-01 8.71878743e-01 -1.98527008e-01 8.69072899e-02 -7.45629743e-02 7.69615650e-01 -1.99278872e-02 -1.71234369e-01 9.65351224e-01 -2.98427157e-02 -4.34431612e-01 2.05445603e-01 -5.08590221e-01 -3.49590033e-01 1.09151149e+00 -4.93706793e-01 -2.89216131e-01 5.51232770e-02 -9.08113778e-01 -4.04732913e-01 7.70829797e-01 -3.00211877e-01 6.67317688e-01 -1.61071050e+00 -1.14928231e-01 2.44801536e-01 9.79888812e-02 -9.61519539e-01 1.47041336e-01 1.04939473e+00 -8.35942864e-01 6.38617933e-01 -6.02289379e-01 -6.51753485e-01 -1.37103558e+00 5.28557777e-01 3.78077596e-01 1.58569649e-01 -4.09273982e-01 2.77784646e-01 -1.29747586e-02 -1.80551365e-01 4.87539880e-02 -2.40213573e-01 -1.20139103e-02 2.48508126e-01 4.10136610e-01 6.06814384e-01 1.18537337e-01 -9.40713704e-01 -6.34175241e-01 1.03531480e+00 6.68808073e-02 1.25022918e-01 1.42705154e+00 2.36412868e-01 -6.85021698e-01 3.65425408e-01 1.15313649e+00 1.11095943e-01 -6.66139305e-01 3.04683149e-01 7.33580142e-02 -2.49320999e-01 -3.30425173e-01 -3.17784816e-01 -4.30444747e-01 1.00332534e+00 6.30866528e-01 3.64577234e-01 5.38234651e-01 -1.09483168e-01 4.05857801e-01 5.84921777e-01 5.54645061e-01 -7.49696434e-01 -4.26241793e-02 3.15732867e-01 6.68379247e-01 -5.32465875e-01 1.74004227e-01 -6.23421609e-01 -3.32170218e-01 1.48121846e+00 -1.81276217e-01 -1.17249906e-01 5.33814907e-01 -6.52907193e-02 -5.63595116e-01 -5.11552274e-01 -4.73803818e-01 -1.03913210e-01 4.77103591e-01 8.17078292e-01 7.69620419e-01 2.72995949e-01 -6.29778147e-01 1.70146316e-01 2.29272842e-02 -4.83935922e-01 6.64051116e-01 9.77195799e-01 -6.75727010e-01 -1.51700854e+00 -5.00620484e-01 1.36162773e-01 -1.38618931e-01 -1.97905049e-01 -6.15870714e-01 5.64514756e-01 1.10053524e-01 4.21764851e-01 -1.97662592e-01 -4.94676083e-01 1.92228585e-01 3.91570538e-01 8.52202773e-01 -2.79810429e-01 -1.74967065e-01 -4.58483696e-01 1.08380847e-01 -4.49873120e-01 -4.99601454e-01 -5.75657308e-01 -1.43708014e+00 -7.30268359e-01 -2.54536092e-01 6.95148766e-01 7.40475535e-01 4.82984900e-01 5.52150607e-01 2.05055267e-01 4.99737322e-01 -9.06593263e-01 -4.72140461e-01 -3.12164605e-01 -1.03718591e+00 5.23505032e-01 2.53959328e-01 -9.41363454e-01 -1.53338507e-01 -1.75400153e-01]
[5.2396087646484375, 5.347587585449219]
7d6f0afa-08fb-46e1-b438-dd9941d017e8
transmrsr-transformer-based-self-distilled
2306.06669
null
https://arxiv.org/abs/2306.06669v1
https://arxiv.org/pdf/2306.06669v1.pdf
TransMRSR: Transformer-based Self-Distilled Generative Prior for Brain MRI Super-Resolution
Magnetic resonance images (MRI) acquired with low through-plane resolution compromise time and cost. The poor resolution in one orientation is insufficient to meet the requirement of high resolution for early diagnosis of brain disease and morphometric study. The common Single image super-resolution (SISR) solutions face two main challenges: (1) local detailed and global anatomical structural information combination; and (2) large-scale restoration when applied for reconstructing thick-slice MRI into high-resolution (HR) iso-tropic data. To address these problems, we propose a novel two-stage network for brain MRI SR named TransMRSR based on the convolutional blocks to extract local information and transformer blocks to capture long-range dependencies. TransMRSR consists of three modules: the shallow local feature extraction, the deep non-local feature capture, and the HR image reconstruction. We perform a generative task to encapsulate diverse priors into a generative network (GAN), which is the decoder sub-module of the deep non-local feature capture part, in the first stage. The pre-trained GAN is used for the second stage of SR task. We further eliminate the potential latent space shift caused by the two-stage training strategy through the self-distilled truncation trick. The extensive experiments show that our method achieves superior performance to other SSIR methods on both public and private datasets. Code is released at https://github.com/goddesshs/TransMRSR.git .
['Bin Sheng', 'TingLi Chen', 'Xiaoer Wei', 'Menghan Hu', 'Tao Tan', 'Xiaohong Liu', 'Shan Huang']
2023-06-11
null
null
null
null
['image-super-resolution', 'image-reconstruction', 'super-resolution']
['computer-vision', 'computer-vision', 'computer-vision']
[ 3.56414109e-01 2.54911603e-03 1.33788332e-01 -4.53018755e-01 -1.35211265e+00 5.77706620e-02 3.37268740e-01 -6.25558913e-01 -3.88010651e-01 7.20000386e-01 4.95768666e-01 -2.15097163e-02 -1.57050818e-01 -6.62971318e-01 -4.62068588e-01 -1.08780992e+00 1.35029897e-01 2.93440193e-01 4.17467594e-01 1.13869449e-02 -4.05517444e-02 5.13580203e-01 -1.03691936e+00 4.60688919e-01 8.07863474e-01 9.06562507e-01 7.66454458e-01 3.50614607e-01 2.25025237e-01 9.20193136e-01 -7.86090493e-02 1.61502287e-02 2.37468481e-01 -5.30782640e-01 -8.97639751e-01 -2.51847208e-02 -2.23353922e-01 -6.97840929e-01 -5.02626359e-01 1.06519508e+00 8.18139195e-01 2.26208270e-02 4.31283504e-01 -4.49154556e-01 -5.26095092e-01 7.26619780e-01 -9.94740844e-01 6.16785109e-01 7.10513145e-02 2.38722682e-01 4.18604344e-01 -9.95139539e-01 6.95589066e-01 9.09691572e-01 3.94660562e-01 7.45476842e-01 -1.15635407e+00 -7.67710924e-01 -2.48796955e-01 1.22009255e-01 -1.38250041e+00 -5.37539363e-01 7.67161906e-01 -4.89929348e-01 4.80989426e-01 1.62296057e-01 3.50993633e-01 1.06781328e+00 4.03049290e-01 4.03276205e-01 1.50320899e+00 1.03316337e-01 -1.00176364e-01 -4.39122885e-01 -3.68983783e-02 3.93057078e-01 -1.24873862e-01 8.33484083e-02 -2.41095960e-01 -7.22353756e-02 1.48278546e+00 1.37980253e-01 -4.89974260e-01 1.13360092e-01 -1.31201637e+00 7.05996692e-01 6.23671114e-01 6.63398981e-01 -6.71604395e-01 -1.40183419e-01 2.72058904e-01 7.23891258e-02 5.72429538e-01 2.06143558e-01 -1.21332981e-01 3.37562472e-01 -1.09074759e+00 -1.71742722e-01 -1.81997512e-02 4.43198562e-01 3.58598143e-01 1.97789028e-01 -2.84893364e-01 1.08074272e+00 2.01547369e-01 2.57459342e-01 8.13372791e-01 -7.89599836e-01 2.51888543e-01 1.41028360e-01 -2.97843963e-01 -6.84222281e-01 -4.70722824e-01 -8.22331429e-01 -1.27378857e+00 -1.41705096e-01 5.27200708e-03 -6.65657073e-02 -1.05824780e+00 1.60260665e+00 4.54353929e-01 1.58907950e-01 -1.51077047e-01 1.45965374e+00 1.16072595e+00 6.39066398e-01 1.43788047e-02 -4.46770906e-01 1.49218822e+00 -9.16003704e-01 -6.59464836e-01 -3.09322357e-01 2.75876641e-01 -6.46072626e-01 7.55701602e-01 6.07234426e-02 -1.39760745e+00 -2.73732513e-01 -9.24843669e-01 -2.86907583e-01 3.29890698e-01 2.18360901e-01 3.89072776e-01 -2.51569715e-03 -1.06424487e+00 3.71610373e-01 -1.14433229e+00 3.18972886e-01 6.31487846e-01 3.24801594e-01 -5.37681282e-01 -3.69979858e-01 -1.17751884e+00 7.99463511e-01 8.75400975e-02 4.06355470e-01 -1.01411796e+00 -9.26804543e-01 -7.60641694e-01 -1.34351104e-01 2.30108172e-01 -7.19326794e-01 8.23435605e-01 -6.76756978e-01 -1.52066362e+00 9.51562405e-01 -1.26534998e-01 -2.06671864e-01 6.12385929e-01 8.06680694e-02 -3.11707020e-01 5.54525614e-01 2.19696015e-01 3.08739901e-01 8.48302245e-01 -1.07635534e+00 6.44002557e-02 -7.07359254e-01 -4.19893801e-01 3.01372141e-01 3.25488091e-01 4.11220789e-01 -3.38640690e-01 -8.77283812e-01 4.40403908e-01 -6.31805420e-01 -4.58097428e-01 -4.75703478e-01 -3.44201833e-01 3.76035303e-01 3.32684606e-01 -1.33286202e+00 9.22783792e-01 -2.03317738e+00 3.40721816e-01 3.00619714e-02 5.72894275e-01 1.21336937e-01 -3.50620747e-02 -1.68904081e-01 -6.25744045e-01 -1.21058077e-01 -4.08151746e-01 -2.14403078e-01 -6.12054467e-01 -1.89568356e-01 -1.70949295e-01 7.00454116e-01 -1.09648108e-02 1.01011026e+00 -7.42982328e-01 -4.81982470e-01 1.36933059e-01 8.74443650e-01 -3.03856581e-01 3.62337470e-01 4.74067003e-01 1.29658091e+00 -6.36154592e-01 3.45734268e-01 8.37765396e-01 -4.30524141e-01 1.79424271e-01 -5.35688937e-01 -1.77658007e-01 4.74663675e-01 -8.64065647e-01 1.94250405e+00 -3.51325840e-01 -5.40688038e-02 3.63714874e-01 -1.01399159e+00 8.29167008e-01 5.08810520e-01 7.71008074e-01 -1.09452343e+00 1.82131976e-01 3.60731721e-01 5.54958321e-02 -4.46882278e-01 -4.19598892e-02 -5.57496548e-01 1.99746668e-01 4.68963414e-01 -2.36097071e-02 1.66523114e-01 -2.68572807e-01 2.08287582e-01 1.11754417e+00 3.35117476e-03 1.49306238e-01 -2.51262575e-01 6.18898809e-01 -5.05339324e-01 6.74822032e-01 3.04839790e-01 1.15631714e-01 1.07099485e+00 3.14227462e-01 -3.81527245e-01 -1.27114570e+00 -9.57013488e-01 -4.03177321e-01 4.84441280e-01 -1.16351172e-01 -9.29128602e-02 -7.55868077e-01 -3.79961193e-01 -6.15397751e-01 2.14752704e-01 -6.09809518e-01 -9.87678096e-02 -9.37858045e-01 -1.16180813e+00 1.28426775e-01 4.61209774e-01 6.23033702e-01 -1.08518410e+00 -5.96608818e-01 1.52745739e-01 -5.27936876e-01 -1.29413009e+00 -5.45044065e-01 2.60610823e-02 -9.83197391e-01 -5.76651394e-01 -1.19060326e+00 -5.02550304e-01 8.07586133e-01 2.61575282e-01 7.87075281e-01 1.65068522e-01 -3.72476220e-01 -2.14647755e-01 -2.39248678e-01 3.71311903e-01 -1.77750200e-01 1.50276467e-01 -2.00132608e-01 5.72077893e-02 -2.08588377e-01 -9.10437763e-01 -1.08927178e+00 3.94545943e-01 -9.08971369e-01 5.75122058e-01 1.00406766e+00 1.05135584e+00 9.73320305e-01 -1.12852193e-01 6.42004490e-01 -8.42771649e-01 2.24791810e-01 -6.26035929e-01 -3.67274046e-01 4.91351075e-02 -2.49391794e-01 1.81916952e-02 5.28679609e-01 -1.83912009e-01 -1.19679630e+00 -8.66082087e-02 -5.74559569e-01 -3.74192536e-01 -3.89094092e-03 4.07793701e-01 -1.93584889e-01 -7.94054493e-02 3.99647653e-01 6.69439495e-01 6.68098181e-02 -6.36844575e-01 -1.08742803e-01 5.88654160e-01 5.98389924e-01 -3.86023879e-01 7.52986848e-01 6.23669147e-01 -8.42107162e-02 -6.49577141e-01 -9.12737787e-01 -2.66911358e-01 -6.30137444e-01 -1.15423426e-01 1.19205499e+00 -1.03421211e+00 -2.80938566e-01 6.04751527e-01 -7.78211892e-01 -4.70283061e-01 -1.75346747e-01 7.04646289e-01 -5.19232690e-01 4.13422137e-01 -1.01891494e+00 -4.02161092e-01 -8.79270852e-01 -1.61940002e+00 1.11433339e+00 2.87421227e-01 3.64065200e-01 -5.45715213e-01 -6.14392348e-02 7.85355389e-01 6.98034048e-01 2.98582762e-01 6.42639697e-01 -3.05604070e-01 -7.84748495e-01 2.82657862e-01 -4.86446381e-01 3.75008851e-01 -9.35222022e-03 -7.56481647e-01 -7.64626443e-01 -4.72710490e-01 5.46317339e-01 -3.34365547e-01 7.77994752e-01 7.66252100e-01 1.41670978e+00 -1.02721997e-01 2.01170403e-03 1.17052960e+00 1.40878749e+00 6.54690713e-02 9.58799243e-01 3.04821044e-01 8.57912838e-01 3.04676056e-01 3.72819513e-01 2.57839739e-01 4.05843943e-01 7.40500867e-01 1.32813349e-01 -3.97847980e-01 -4.39414293e-01 1.56212617e-02 2.58669674e-01 1.30128407e+00 -3.02914202e-01 4.57077175e-01 -9.45484877e-01 4.59469497e-01 -1.53258038e+00 -9.19543803e-01 -4.47623394e-02 2.09417868e+00 1.09172153e+00 -1.33406699e-01 -6.73447326e-02 -1.40962735e-01 6.91528141e-01 2.36234993e-01 -6.50404990e-01 1.67039171e-01 2.87419390e-02 2.83726573e-01 4.21305209e-01 4.50892925e-01 -7.60484517e-01 6.76288009e-01 5.30656528e+00 9.65867698e-01 -1.23782444e+00 7.04749227e-01 8.14653099e-01 -4.80587222e-02 -4.19299066e-01 -1.96401313e-01 -4.71593350e-01 5.29151917e-01 8.11394393e-01 2.52744168e-01 5.61203957e-01 3.10535729e-01 2.91168302e-01 -6.07959516e-02 -4.98094738e-01 1.08212769e+00 -2.70152614e-02 -1.26073313e+00 -8.13855603e-02 8.80667269e-02 4.84072477e-01 4.93286639e-01 1.88345853e-02 -1.26373932e-01 -3.88233587e-02 -1.26516175e+00 4.85318601e-01 7.25643337e-01 1.38128924e+00 -8.15316319e-01 6.40246987e-01 2.29908034e-01 -1.09601867e+00 1.38256475e-01 -3.75058889e-01 5.28027177e-01 3.68715316e-01 8.42919409e-01 -3.05296272e-01 8.48111987e-01 7.20948577e-01 6.35629535e-01 -2.60718316e-01 7.95206547e-01 -2.72911370e-01 3.96579117e-01 -3.00604291e-02 9.64237690e-01 9.80957691e-03 -3.16479385e-01 7.53882170e-01 8.45320046e-01 1.92679167e-01 5.42906582e-01 -1.18412361e-01 1.06549847e+00 1.02100737e-01 -6.21544905e-02 -1.51405141e-01 4.01750743e-01 8.33272338e-02 1.60800076e+00 -8.01142752e-01 -1.05361708e-01 -3.54932755e-01 8.87718797e-01 3.22609425e-01 3.79506499e-01 -7.91320264e-01 -3.50114442e-02 5.92193678e-02 4.60927308e-01 2.09194750e-01 -2.19650909e-01 -3.83131027e-01 -1.59805584e+00 1.90865993e-01 -1.00637293e+00 3.65722060e-01 -9.66620386e-01 -1.12563407e+00 1.01330268e+00 -6.55327588e-02 -8.83642673e-01 -1.80764496e-01 9.74862836e-03 -5.30029118e-01 1.27851999e+00 -1.70549476e+00 -1.26748586e+00 -4.02612925e-01 7.57081270e-01 3.67651701e-01 1.18943803e-01 5.87316692e-01 5.74946165e-01 -7.02357352e-01 3.47078741e-01 -1.02236696e-01 2.39103511e-01 5.42535067e-01 -8.77076030e-01 1.15689307e-01 1.02558172e+00 -5.96132517e-01 6.34983420e-01 3.76672506e-01 -8.44518244e-01 -1.14555860e+00 -9.63742673e-01 5.31284928e-01 -6.68568388e-02 5.55172801e-01 -2.74037302e-01 -1.09945953e+00 7.22032845e-01 -1.59067795e-01 3.66278827e-01 4.07224536e-01 -5.39188862e-01 -1.48019537e-01 -8.05896074e-02 -1.30169821e+00 3.45585436e-01 9.39890385e-01 -5.30624568e-01 -5.66683710e-01 1.63380116e-01 6.57858253e-01 -7.19920635e-01 -1.35429025e+00 5.32131076e-01 4.46477830e-01 -7.78241694e-01 1.16087306e+00 -7.75932521e-02 8.69430602e-01 -2.29752705e-01 2.13103667e-02 -1.11104906e+00 -4.56600130e-01 -5.38517177e-01 1.26869485e-01 8.78548384e-01 1.27770647e-01 -7.44774342e-01 3.79287720e-01 5.42893171e-01 -3.57759029e-01 -1.04910123e+00 -1.05412638e+00 -4.05376166e-01 1.40349984e-01 1.60860643e-02 5.35085440e-01 9.13935244e-01 -3.98654819e-01 3.60980272e-01 -4.19561416e-01 1.29340515e-01 1.01018751e+00 1.69647127e-01 1.52818993e-01 -7.81708658e-01 -3.99485707e-01 -1.93752348e-01 -1.70635030e-01 -8.54684889e-01 -7.75482878e-02 -1.01144314e+00 -1.64458305e-01 -1.59886444e+00 8.10574710e-01 -4.79578435e-01 -2.79288650e-01 2.74745613e-01 -1.24200419e-01 4.51467246e-01 -1.11467853e-01 6.06672645e-01 -2.73006797e-01 6.32807732e-01 1.90105450e+00 3.25619966e-01 1.29774675e-01 -2.60084927e-01 -7.12130189e-01 6.04067683e-01 6.61776483e-01 -4.89342719e-01 -4.22784030e-01 -5.81691861e-01 -4.53387611e-02 5.62026739e-01 5.80148935e-01 -8.34083140e-01 3.45445722e-02 1.02180778e-03 5.66795826e-01 -4.76210386e-01 2.79937088e-01 -5.58573484e-01 3.52490008e-01 3.37093860e-01 -3.27561080e-01 5.15908282e-03 -2.54280090e-01 1.95074789e-02 -1.44867286e-01 9.55096185e-02 1.37711048e+00 -3.46842706e-01 -1.07785866e-01 7.78119981e-01 -4.96219508e-02 2.61250250e-02 7.56619990e-01 9.46687609e-02 -3.32859218e-01 -9.77262557e-02 -9.39359605e-01 7.16514587e-02 3.62613797e-01 1.46025747e-01 8.09435189e-01 -1.20729494e+00 -9.79586720e-01 2.68944055e-01 -3.82685691e-01 4.22239780e-01 1.02511108e+00 1.56654549e+00 -4.95782346e-01 2.34289289e-01 -4.30030286e-01 -4.84168738e-01 -8.67803991e-01 2.57325828e-01 6.81584001e-01 -7.23843753e-01 -1.33948076e+00 6.68320835e-01 6.82051301e-01 -2.80669153e-01 -2.54704446e-01 5.62114641e-02 -3.26611936e-01 -2.98656046e-01 8.72277379e-01 9.97211188e-02 2.21250251e-01 -1.00382853e+00 -4.34443533e-01 5.83112776e-01 -3.97631645e-01 -2.27344289e-01 1.93648303e+00 -3.53147119e-01 -3.82848859e-01 1.32309392e-01 1.20569170e+00 -3.05403233e-01 -1.22956014e+00 -4.15806234e-01 -4.47535068e-01 -3.57684016e-01 6.06386185e-01 -7.99186885e-01 -1.56031442e+00 8.65922928e-01 6.37151599e-01 -4.00006950e-01 1.35493779e+00 1.08775251e-01 1.20783079e+00 -5.27084291e-01 4.32836473e-01 -5.94943047e-01 -1.36522204e-01 2.15928301e-01 1.18479443e+00 -9.35270905e-01 1.00477673e-01 -3.11341524e-01 -7.08738148e-01 9.16274548e-01 5.56622684e-01 -3.06834131e-01 5.28917074e-01 4.64189500e-01 -1.10239662e-01 -4.50929761e-01 -6.39856994e-01 3.40821743e-02 4.21791881e-01 3.70583028e-01 4.58248287e-01 6.09916374e-02 -5.40307701e-01 9.43295062e-01 -3.81144248e-02 1.29662529e-01 3.74187589e-01 6.07838273e-01 -6.18624873e-02 -8.27328503e-01 -2.14447424e-01 4.82008547e-01 -9.29062843e-01 -2.67311007e-01 2.45223850e-01 2.93496639e-01 -8.59515220e-02 4.14816678e-01 -1.70039833e-01 -2.76754916e-01 1.24016451e-02 -2.78637081e-01 5.41888833e-01 -5.69467485e-01 -5.66191435e-01 4.72280383e-01 -1.60920665e-01 -8.58346403e-01 -3.79294634e-01 -6.92450643e-01 -1.35893118e+00 -1.75629091e-02 -4.87727821e-02 -1.93915311e-02 5.95997453e-01 1.02265930e+00 3.25955182e-01 8.71693611e-01 6.85022771e-01 -8.93846989e-01 -4.39893395e-01 -9.57615137e-01 -5.65392613e-01 2.63028622e-01 3.47702324e-01 -5.81934214e-01 -2.39939377e-01 -2.64397472e-01]
[13.629505157470703, -2.4012317657470703]
0228a668-1113-4a0e-8865-49dbaa535772
automatic-sexism-detection-with-multilingual
2106.04908
null
https://arxiv.org/abs/2106.04908v2
https://arxiv.org/pdf/2106.04908v2.pdf
Automatic Sexism Detection with Multilingual Transformer Models
Sexism has become an increasingly major problem on social networks during the last years. The first shared task on sEXism Identification in Social neTworks (EXIST) at IberLEF 2021 is an international competition in the field of Natural Language Processing (NLP) with the aim to automatically identify sexism in social media content by applying machine learning methods. Thereby sexism detection is formulated as a coarse (binary) classification problem and a fine-grained classification task that distinguishes multiple types of sexist content (e.g., dominance, stereotyping, and objectification). This paper presents the contribution of the AIT_FHSTP team at the EXIST2021 benchmark for both tasks. To solve the tasks we applied two multilingual transformer models, one based on multilingual BERT and one based on XLM-R. Our approach uses two different strategies to adapt the transformers to the detection of sexist content: first, unsupervised pre-training with additional data and second, supervised fine-tuning with additional and augmented data. For both tasks our best model is XLM-R with unsupervised pre-training on the EXIST data and additional datasets and fine-tuning on the provided dataset. The best run for the binary classification (task 1) achieves a macro F1-score of 0.7752 and scores 5th rank in the benchmark; for the multiclass classification (task 2) our best submission scores 6th rank with a macro F1-score of 0.5589.
['Matthias Zeppelzauer', 'Alexander Schindler', 'Sven Schlarb', 'Johannes Bogensperger', 'Manuel Hecht', 'Armin Kirchknopf', 'Djordje Slijepčević', 'Daria Liakhovets', 'Jaqueline Boeck', 'Mina Schütz']
2021-06-09
null
null
null
null
['unsupervised-pre-training']
['methodology']
[ 3.17776859e-01 5.38348615e-01 -2.19251633e-01 -5.43529809e-01 -4.50125605e-01 -5.35057724e-01 1.31107056e+00 6.31140947e-01 -6.78451359e-01 7.36637533e-01 1.37485847e-01 -3.19793940e-01 -5.52647054e-01 -8.64006996e-01 -4.71315295e-01 -4.01484638e-01 9.14913416e-03 1.24647689e+00 1.13707013e-01 -4.64734375e-01 4.66667920e-01 3.79161268e-01 -1.77784991e+00 5.95799685e-01 9.85566914e-01 8.59574735e-01 -2.05202296e-01 5.86669922e-01 -1.60301134e-01 4.94394422e-01 -4.09084469e-01 -8.63912225e-01 1.33737788e-01 -2.48529091e-01 -1.23709095e+00 -2.35350549e-01 7.04356194e-01 2.76951969e-01 2.01983571e-01 1.20969665e+00 3.34187686e-01 -1.99014664e-01 8.28167915e-01 -1.35250700e+00 9.84796286e-02 1.13065541e+00 -5.67037642e-01 6.58675954e-02 3.84377748e-01 -2.56944060e-01 1.07585597e+00 -6.90515161e-01 9.70363081e-01 1.75963187e+00 6.50719047e-01 2.89101273e-01 -1.44777322e+00 -8.09219360e-01 -2.48338252e-01 1.72430918e-01 -1.29288554e+00 -3.52421939e-01 3.98325205e-01 -7.74292827e-01 5.48997283e-01 5.33524275e-01 3.69882107e-01 1.27793300e+00 -1.64305389e-01 4.55771714e-01 1.76247621e+00 -4.03650075e-01 -1.67218074e-01 7.75071561e-01 2.38677323e-01 5.63203990e-01 1.88377708e-01 9.49059129e-02 -3.93694371e-01 -4.38702926e-02 2.91950554e-01 -6.22717083e-01 5.96988499e-01 -1.06074037e-02 -1.22044504e+00 1.19023693e+00 1.75615668e-01 8.18896711e-01 -9.65175685e-03 -2.53048629e-01 8.26484799e-01 4.97648925e-01 7.06739426e-01 6.54723644e-01 -2.83865482e-01 -1.36383504e-01 -1.16740537e+00 5.51741898e-01 9.02209520e-01 4.56945747e-01 7.05270708e-01 -6.13886654e-01 -3.02663356e-01 1.31684470e+00 1.64192557e-01 4.20337498e-01 7.12377965e-01 -5.79406559e-01 6.50891662e-01 7.96827972e-01 -4.29640859e-01 -1.08783603e+00 -7.09726453e-01 -5.21772444e-01 -6.29904091e-01 -6.35494245e-03 9.08723712e-01 -5.36016710e-02 -4.65452194e-01 1.78205371e+00 4.24904197e-01 -3.48614037e-01 -2.52938390e-01 5.83802104e-01 9.20498610e-01 5.17790139e-01 -2.90805311e-03 -1.57442659e-01 1.52016842e+00 -6.39874339e-01 -5.09829819e-01 -2.52624620e-02 7.03077257e-01 -9.56866086e-01 6.50636315e-01 7.02745020e-01 -9.00409937e-01 -4.99300003e-01 -9.38426793e-01 -9.79874507e-02 -9.73688602e-01 1.82077840e-01 5.93873024e-01 1.00827336e+00 -9.74948168e-01 6.72575533e-01 -4.29515280e-02 -8.06227148e-01 2.78369766e-02 6.89094007e-01 -5.83467484e-01 2.15922818e-01 -1.37283623e+00 7.45160639e-01 5.86630583e-01 -2.52252162e-01 -3.77622396e-01 -6.08122647e-01 -6.55655861e-01 -2.78249145e-01 4.22377795e-01 -1.97058648e-01 7.16612458e-01 -1.10416567e+00 -1.33493662e+00 1.86397195e+00 3.81452799e-01 -4.59701687e-01 1.07685769e+00 2.37526298e-01 -5.37318587e-01 -1.90144658e-01 5.43148994e-01 7.95754015e-01 6.88844383e-01 -1.08858764e+00 -8.05419028e-01 -7.32175410e-01 3.36011499e-02 -7.32250512e-02 -5.36407292e-01 3.92723620e-01 -1.80981227e-03 -6.76115513e-01 -7.32633397e-02 -1.13850927e+00 3.46915334e-01 -4.40458834e-01 -6.77235901e-01 -5.62402129e-01 5.56873798e-01 -7.52243221e-01 1.13722920e+00 -1.92050540e+00 3.51895839e-01 6.84694827e-01 3.23295116e-01 1.95240319e-01 -9.41850245e-02 4.69459057e-01 -3.89189780e-01 1.09049037e-01 7.34431520e-02 -4.37609881e-01 3.12230706e-01 -5.02517074e-02 -9.45889801e-02 2.85291255e-01 -2.04493701e-01 5.04187226e-01 -1.00074768e+00 -7.62858689e-01 1.08608611e-01 1.56409755e-01 -5.28378785e-01 -1.34760231e-01 -3.70603539e-02 3.66227895e-01 1.90439131e-02 3.81871641e-01 7.07644045e-01 1.71432704e-01 5.97111046e-01 -1.23197474e-01 -3.65146607e-01 1.26520917e-01 -1.20587826e+00 1.13838744e+00 -5.78675807e-01 5.55369258e-01 2.66966879e-01 -9.69161510e-01 1.19781411e+00 -8.65746439e-02 5.32210886e-01 -7.45860577e-01 2.51922727e-01 7.74528801e-01 1.65659443e-01 -4.20275837e-01 6.33517146e-01 -1.44756317e-01 -4.26856309e-01 3.92060071e-01 4.13395822e-01 -1.12078562e-01 9.95456994e-01 1.82872564e-01 7.63639510e-01 -1.70208171e-01 2.65569210e-01 -8.60434651e-01 1.10042822e+00 -3.45613092e-01 2.51783788e-01 5.97980976e-01 -3.34766880e-03 1.17505237e-01 1.12582922e+00 -3.83941203e-01 -9.09560025e-01 -8.26114893e-01 -6.38706148e-01 1.49788046e+00 -2.02799574e-01 -4.39557970e-01 -6.76616907e-01 -9.62337077e-01 2.19469830e-01 4.27990794e-01 -8.14470053e-01 3.36800218e-02 -4.49194372e-01 -9.91680682e-01 7.88612783e-01 -3.58463258e-01 5.45242488e-01 -1.02772629e+00 2.75755405e-01 -1.53572798e-01 -3.42610925e-01 -1.14064157e+00 2.61406861e-02 4.11055721e-02 -4.13676381e-01 -1.23990655e+00 -4.95938540e-01 -6.20257795e-01 2.64161170e-01 -7.26739228e-01 1.06790209e+00 -1.51638374e-01 -3.49467188e-01 -3.23166490e-01 -1.92436561e-01 -1.88824490e-01 -8.55189383e-01 5.82929611e-01 4.87778448e-02 5.51857233e-01 3.06947172e-01 -5.78074217e-01 -1.23175025e-01 4.53826904e-01 -7.89982975e-01 7.20689306e-03 5.62806606e-01 6.61389410e-01 -4.19676937e-02 -9.39703658e-02 5.46684444e-01 -1.46834850e+00 6.31469846e-01 -6.01710260e-01 -6.75752878e-01 -4.82436530e-02 -7.61207283e-01 4.67410944e-02 4.03386474e-01 -1.40308022e-01 -8.61444354e-01 -1.93821236e-01 -4.15887773e-01 3.03280681e-01 -1.41476125e-01 2.46334925e-01 -2.45894134e-01 -1.34325653e-01 8.06803584e-01 -2.22762942e-01 -7.18287230e-02 -6.08581960e-01 1.44119740e-01 1.05617607e+00 2.97444910e-01 -6.13183796e-01 6.48658395e-01 1.42996877e-01 2.39595622e-01 -9.47669089e-01 -8.02776992e-01 -4.13954705e-01 -6.54496908e-01 -3.48806798e-01 8.41211200e-01 -6.91035628e-01 -7.79083431e-01 6.10647082e-01 -7.53167808e-01 -2.35607743e-01 -9.53835249e-02 2.84880161e-01 -4.50385064e-01 2.89391547e-01 -5.45506537e-01 -5.25084257e-01 -2.86821663e-01 -9.30012226e-01 1.04547548e+00 -2.38816142e-01 -6.51142478e-01 -1.08337307e+00 4.14879411e-01 8.82398069e-01 1.76505640e-01 4.29936379e-01 1.11944485e+00 -1.25186563e+00 3.20594519e-01 1.19112935e-02 -5.74980497e-01 1.90377966e-01 -1.04971357e-01 -1.06114913e-02 -1.03942907e+00 -6.19792119e-02 -4.21650678e-01 -4.07813877e-01 9.44592655e-01 -7.47894868e-02 9.32655334e-01 -2.72104084e-01 -2.75063515e-01 1.89087287e-01 1.20274007e+00 -2.90384769e-01 4.68489498e-01 3.58523577e-01 6.35298729e-01 1.21735644e+00 6.82144344e-01 3.29580098e-01 3.96069020e-01 1.07028103e+00 3.56453836e-01 1.08258240e-01 -1.01391286e-01 -1.93560243e-01 2.33717009e-01 3.06123555e-01 -1.46908522e-01 9.26529318e-02 -1.11597514e+00 4.85880077e-01 -1.94079709e+00 -8.79187047e-01 -5.61976314e-01 2.26193619e+00 9.27346826e-01 4.12839293e-01 8.30122411e-01 5.46244085e-01 1.02126300e+00 4.12446894e-02 4.13082987e-02 -8.10112953e-01 -1.27887592e-01 2.00264335e-01 3.34544659e-01 7.90979862e-01 -1.25670719e+00 8.66403103e-01 4.88839865e+00 1.30469179e+00 -1.01125407e+00 1.60803035e-01 6.32459879e-01 -9.87889618e-02 9.49060991e-02 -3.04087371e-01 -1.09134448e+00 8.53420854e-01 1.13987768e+00 6.34420589e-02 5.60886383e-01 3.77445042e-01 6.41151965e-02 -6.71452582e-02 -1.06836200e+00 9.28187907e-01 1.32686451e-01 -1.02214134e+00 -1.67450476e-02 4.81788009e-01 6.64775729e-01 8.28055590e-02 9.56608579e-02 5.90461671e-01 1.07245244e-01 -8.90779674e-01 8.18504870e-01 3.51994187e-01 8.42310369e-01 -8.30428064e-01 9.37370360e-01 3.35951954e-01 -7.61020362e-01 -3.54841739e-01 1.23858400e-01 6.90876245e-02 -1.20962895e-01 9.61334229e-01 -7.12596178e-01 5.42033076e-01 5.33582628e-01 8.52550149e-01 -9.47475314e-01 6.69557095e-01 -1.07670985e-01 1.74412444e-01 -4.14040536e-01 -1.75877884e-01 1.77133590e-01 -1.81789383e-01 8.21377814e-01 1.51205111e+00 5.00624254e-02 -8.06739807e-01 9.34459567e-02 6.21451676e-01 -1.89679310e-01 4.00987476e-01 -3.72703820e-01 -1.54075265e-01 1.13585681e-01 1.71299601e+00 -8.98201108e-01 -3.27445358e-01 1.55298799e-01 4.66260403e-01 2.38379851e-01 -2.61088610e-01 -5.12077332e-01 -4.13884610e-01 1.94981664e-01 6.53871715e-01 -7.02535510e-02 4.36518043e-01 -2.15623230e-01 -9.20818090e-01 -3.64876926e-01 -9.27977920e-01 7.89317191e-01 -2.15073243e-01 -1.42537963e+00 5.74267089e-01 2.67568946e-01 -7.01239884e-01 -6.53358161e-01 -8.92838180e-01 -8.30495730e-02 6.19463027e-01 -1.05948424e+00 -1.56493998e+00 -1.18284874e-01 5.15231371e-01 1.67647094e-01 -3.94800812e-01 6.51220918e-01 7.74335027e-01 -4.40880954e-01 6.55444682e-01 3.39980004e-03 2.67785098e-02 1.02903509e+00 -1.62317014e+00 -1.10782459e-01 2.75911242e-01 -9.62479487e-02 1.97075576e-01 9.11006689e-01 -6.59417272e-01 -6.73008382e-01 -1.11071229e+00 1.68138433e+00 -2.61733145e-01 9.76600051e-01 -9.98986363e-01 -3.01878244e-01 4.10867542e-01 6.38278797e-02 -4.48243886e-01 5.13853490e-01 5.08984327e-01 -5.61506450e-01 -2.78454930e-01 -1.47680545e+00 1.69127136e-01 9.74205077e-01 -5.36399782e-01 -3.50724816e-01 8.59909534e-01 1.61146522e-01 -1.51850134e-01 -1.14740026e+00 2.45569915e-01 5.93818009e-01 -1.17082286e+00 7.45769143e-01 -2.17730835e-01 8.50722015e-01 8.26010630e-02 -8.53197351e-02 -1.13578749e+00 6.61337301e-02 -5.33286035e-01 4.87225324e-01 1.60940003e+00 5.88955939e-01 -8.37993324e-01 5.98626316e-01 5.40239178e-02 3.35718513e-01 -5.08521140e-01 -9.27947521e-01 -7.67322600e-01 2.37368286e-01 -1.23461984e-01 3.80546004e-01 1.20326138e+00 1.19892739e-01 7.72237659e-01 -1.01151861e-01 -4.16865975e-01 6.13927960e-01 1.51289567e-01 8.72325301e-01 -1.74363685e+00 -2.69468606e-01 -8.81932318e-01 -6.61002040e-01 -5.82496226e-02 4.66095835e-01 -1.37673783e+00 -4.08065379e-01 -9.64758337e-01 4.30182338e-01 -4.10980433e-01 6.98853582e-02 3.49837393e-01 4.21163231e-01 7.25248098e-01 1.51975930e-01 -1.41341642e-01 -5.79049528e-01 -3.74620734e-03 7.80046940e-01 -1.98583260e-01 -7.05999509e-02 1.60104886e-01 -5.44931114e-01 5.96604526e-01 6.02780163e-01 -4.84590441e-01 -4.23937105e-02 2.94494629e-01 9.32425201e-01 -2.77504534e-01 4.05719072e-01 -8.81026149e-01 -2.15407923e-01 1.91682145e-01 1.47346675e-01 -6.05241358e-01 3.90619725e-01 -4.53299940e-01 1.25646412e-01 7.59213805e-01 -4.40561771e-01 -1.76142588e-01 1.15142334e-02 2.80767567e-02 -2.26277187e-01 -4.61559743e-01 8.84320855e-01 -7.32711703e-02 -2.09053159e-01 -1.09414486e-02 -2.45773420e-01 3.04432243e-01 9.00609791e-01 5.40073700e-02 -3.97431165e-01 -1.31281978e-02 -1.18564856e+00 1.46112099e-01 2.87267238e-01 6.55599535e-01 -2.23118350e-01 -1.05552113e+00 -1.15434659e+00 2.35617340e-01 2.93695599e-01 -8.59595954e-01 6.14839830e-02 8.67106616e-01 -3.62639159e-01 4.64596897e-01 -6.34100437e-01 -5.81107378e-01 -1.65519583e+00 2.83346146e-01 2.69311070e-01 -1.01419032e+00 2.41468504e-01 6.63162410e-01 -2.88258418e-02 -1.19809711e+00 -1.87867761e-01 8.10042918e-02 -8.37083340e-01 8.33456635e-01 2.77649581e-01 8.16874802e-01 3.28259766e-01 -1.01130450e+00 -3.66025060e-01 5.45621276e-01 -1.46576568e-01 -4.32482541e-01 1.39183068e+00 7.25380853e-02 -7.64798224e-01 3.99660885e-01 1.46366167e+00 2.55235493e-01 -3.20612341e-01 -3.32521982e-02 3.13999355e-01 -3.01626563e-01 -1.12758346e-01 -1.06317902e+00 -8.80364656e-01 6.09724581e-01 4.75728393e-01 7.27473974e-01 5.60829937e-01 2.76903212e-02 2.04197541e-01 3.38581875e-02 4.98210452e-02 -1.36029780e+00 -1.22745618e-01 6.79991245e-01 1.14557016e+00 -1.12875044e+00 9.22291055e-02 -8.41728747e-01 -2.79227823e-01 1.15959108e+00 3.71691018e-01 5.07952347e-02 5.17202556e-01 -2.53756136e-01 -2.73745477e-01 -4.43189859e-01 -3.32782388e-01 -1.37228712e-01 5.10773122e-01 3.02649200e-01 3.96303624e-01 2.08455801e-01 -1.03542101e+00 5.56949615e-01 -7.37204194e-01 -2.57311612e-01 2.36219004e-01 1.40763238e-01 -4.20049718e-03 -1.38149083e+00 -4.07648355e-01 7.51092911e-01 -5.80513060e-01 1.30385131e-01 -9.41113591e-01 9.02603805e-01 5.97672343e-01 9.67377424e-01 9.50578377e-02 -3.95406932e-01 2.36541718e-01 -1.02258608e-01 5.65729201e-01 -5.36414206e-01 -1.30472791e+00 -2.29977846e-01 8.09753299e-01 -5.54007828e-01 -3.16872388e-01 -9.97485876e-01 -7.28325665e-01 -6.38717949e-01 1.20556228e-01 1.95536658e-01 9.76231515e-01 1.04038835e+00 -1.52021170e-01 2.91169345e-01 6.64146781e-01 -7.91006863e-01 -3.96516114e-01 -1.14369011e+00 -8.80094945e-01 5.76923907e-01 3.04887574e-02 -6.74914658e-01 -4.34643924e-01 -3.99220824e-01]
[9.372026443481445, 10.365900993347168]
1456ccaa-cef4-41b9-973e-d70647d2d082
stubborn-lexical-bias-in-data-and-models
2306.02190
null
https://arxiv.org/abs/2306.02190v1
https://arxiv.org/pdf/2306.02190v1.pdf
Stubborn Lexical Bias in Data and Models
In NLP, recent work has seen increased focus on spurious correlations between various features and labels in training data, and how these influence model behavior. However, the presence and effect of such correlations are typically examined feature by feature. We investigate the cumulative impact on a model of many such intersecting features. Using a new statistical method, we examine whether such spurious patterns in data appear in models trained on the data. We select two tasks -- natural language inference and duplicate-question detection -- for which any unigram feature on its own should ideally be uninformative, which gives us a large pool of automatically extracted features with which to experiment. The large size of this pool allows us to investigate the intersection of features spuriously associated with (potentially different) labels. We then apply an optimization approach to *reweight* the training data, reducing thousands of spurious correlations, and examine how doing so affects models trained on the reweighted data. Surprisingly, though this method can successfully reduce lexical biases in the training data, we still find strong evidence of corresponding bias in the trained models, including worsened bias for slightly more complex features (bigrams). We close with discussion about the implications of our results on what it means to "debias" training data, and how issues of data quality can affect model bias.
['Noah A. Smith', 'Jesse Dodge', 'Sofia Serrano']
2023-06-03
null
null
null
null
['natural-language-inference']
['natural-language-processing']
[ 2.97390819e-01 2.03767031e-01 -2.43585110e-01 -7.42257118e-01 -9.75570560e-01 -7.50673473e-01 7.33986855e-01 3.81065875e-01 -7.96296597e-01 7.84029841e-01 5.87033093e-01 -7.18200207e-01 -8.99362490e-02 -6.28047526e-01 -8.93739223e-01 -5.03174543e-01 2.99675196e-01 2.06363365e-01 1.95630863e-01 -1.20883598e-03 5.07749736e-01 1.76200122e-01 -1.54889488e+00 4.69751179e-01 6.68956876e-01 3.10758203e-01 -5.59767783e-02 4.35119867e-01 -1.86720490e-01 8.78876507e-01 -7.53615558e-01 -5.76205730e-01 3.49169783e-02 -4.56921786e-01 -9.22923625e-01 -1.01822175e-01 8.53267848e-01 5.73773831e-02 -5.38412333e-02 9.40189660e-01 3.53553474e-01 -1.01592112e-02 9.33640659e-01 -1.02396345e+00 -5.42951226e-01 7.37087846e-01 -3.29790592e-01 6.21799350e-01 3.54381382e-01 2.65102953e-01 1.38138187e+00 -9.64466631e-01 7.07341254e-01 1.42716944e+00 8.30313206e-01 1.94600463e-01 -1.59607458e+00 -7.55572021e-01 -2.44343579e-02 -1.14898667e-01 -1.17792249e+00 -7.51814425e-01 3.36500615e-01 -5.73085606e-01 9.24014568e-01 3.89796078e-01 1.84379861e-01 1.08599973e+00 1.61976293e-01 3.04151058e-01 1.50629640e+00 -7.23028660e-01 -1.18408106e-01 5.96754253e-01 6.52312577e-01 5.93830407e-01 5.39484680e-01 2.73427904e-01 -5.11243463e-01 -5.60770810e-01 2.06341565e-01 -4.56797242e-01 -1.79537401e-01 1.44431338e-01 -1.12988198e+00 1.02009881e+00 3.09529990e-01 5.42658150e-01 -4.08066844e-04 -2.69012451e-02 1.81977555e-01 4.38363224e-01 5.49538195e-01 9.43028033e-01 -9.30652678e-01 -8.25893357e-02 -8.69109094e-01 6.00296855e-01 9.15593445e-01 5.50293565e-01 8.90429139e-01 -3.87554020e-01 -8.80333707e-02 1.23316598e+00 1.46463960e-01 3.73811454e-01 6.91878080e-01 -8.45552146e-01 4.12952036e-01 5.03847241e-01 2.33052328e-01 -1.17956328e+00 -5.55314064e-01 -8.68536308e-02 -3.54860388e-02 -1.38347015e-01 9.85807240e-01 -2.97149777e-01 -6.36429310e-01 2.02646589e+00 2.42733285e-01 -4.31030512e-01 -2.66963422e-01 6.76220179e-01 5.45052230e-01 4.62836623e-01 3.89337331e-01 -2.12956712e-01 1.38752842e+00 -3.06918055e-01 -5.76874256e-01 -5.92586875e-01 1.31126249e+00 -1.07607687e+00 1.47691178e+00 -7.37763867e-02 -8.05700541e-01 -3.04824084e-01 -7.61782169e-01 -2.20744774e-01 -4.39283520e-01 -2.54589647e-01 5.80002904e-01 6.49370551e-01 -6.09611988e-01 9.49529290e-01 -2.72464752e-01 -4.09815878e-01 -5.24616055e-02 3.09333384e-01 -1.76475719e-01 -1.48623392e-01 -1.33565867e+00 1.41842806e+00 1.50202245e-01 -2.24265084e-01 5.61338589e-02 -6.60284400e-01 -7.44689703e-01 -3.27974372e-02 4.87296700e-01 -1.90221190e-01 1.36472142e+00 -1.04304254e+00 -6.21873677e-01 1.03370953e+00 -6.16562128e-01 1.03837460e-01 1.63033962e-01 5.90992831e-02 -6.19419873e-01 -6.04481876e-01 4.16449726e-01 4.01931554e-01 4.46300536e-01 -1.17909074e+00 -7.32468426e-01 -3.94204229e-01 -2.52685547e-01 1.53783858e-02 -1.63229465e-01 4.58522588e-01 1.57022327e-01 -5.17569959e-01 2.37836048e-01 -1.01219189e+00 1.75831437e-01 -4.99600470e-01 -3.12379986e-01 -5.67704141e-01 1.08179756e-01 -3.96098673e-01 1.41703117e+00 -2.07174897e+00 -4.62711334e-01 3.78805459e-01 6.06842898e-02 -9.98893008e-02 -2.23397270e-01 4.38704878e-01 -1.66594625e-01 6.33737206e-01 -1.63848281e-01 -4.56997938e-02 1.60959866e-02 3.54858935e-01 -2.89078742e-01 4.46907520e-01 4.49184030e-01 7.81814694e-01 -6.75970078e-01 -6.46562636e-01 -2.94110626e-01 2.36851559e-03 -8.86069536e-01 3.37220244e-02 -1.43008411e-01 -6.06197901e-02 -5.21924719e-02 3.91967893e-01 3.43483686e-01 -2.39302590e-01 2.76781350e-01 -1.53391242e-01 -2.37048417e-01 1.26415384e+00 -9.86297131e-01 6.90869927e-01 -3.64196658e-01 8.19114149e-01 -2.32021675e-01 -5.98195553e-01 6.98411882e-01 -6.29476234e-02 -3.87705974e-02 -5.99161685e-01 1.61082135e-03 5.07589519e-01 7.29625404e-01 -7.41129279e-01 6.81390464e-01 -6.06656790e-01 -2.08204016e-02 5.72181404e-01 -4.82978597e-02 -8.43600258e-02 9.41930860e-02 9.03226286e-02 1.16912723e+00 -3.97531748e-01 1.35767773e-01 -5.97392738e-01 4.21477295e-02 2.52656966e-01 7.17265904e-01 1.06235671e+00 1.49659356e-02 4.20002103e-01 8.41128469e-01 -1.62434354e-01 -1.06192839e+00 -7.09321797e-01 -7.40784705e-01 1.31236482e+00 -3.39898050e-01 -5.19958973e-01 -2.95102566e-01 -8.66970718e-01 3.57857406e-01 1.20096850e+00 -7.64646590e-01 -2.74386555e-01 -4.84617889e-01 -1.10829926e+00 5.68737745e-01 2.59728849e-01 -3.03419709e-01 -1.09893453e+00 -2.46296585e-01 8.49683210e-02 -2.52623767e-01 -7.86186874e-01 -3.38316947e-01 5.33872843e-01 -7.76016355e-01 -1.08697152e+00 6.80893809e-02 -6.75777018e-01 5.81915498e-01 2.07666695e-01 1.51877856e+00 5.00935674e-01 -1.06274092e-03 7.71289691e-02 -3.24172407e-01 -4.97326970e-01 -8.98620009e-01 3.69879343e-02 -9.40740928e-02 -3.82769316e-01 1.02464592e+00 -2.61104375e-01 -3.48413289e-02 2.23281637e-01 -9.08851683e-01 -4.49684709e-01 5.68433106e-01 9.74406064e-01 9.52447206e-03 -1.61741614e-01 7.74618804e-01 -1.57047570e+00 8.41777802e-01 -6.80780530e-01 -3.91741574e-01 2.29929030e-01 -7.93730319e-01 4.37916130e-01 4.14725840e-01 -5.81608474e-01 -7.92243421e-01 -5.66802859e-01 4.71187523e-03 2.87858427e-01 -1.75894916e-01 6.81378961e-01 7.20492080e-02 1.57703117e-01 9.94119287e-01 -3.20676297e-01 5.89367747e-02 -5.41397452e-01 8.49030763e-02 8.52049887e-01 4.95213009e-02 -6.81535065e-01 6.81028068e-01 -1.48202002e-01 -3.16810429e-01 -7.68514395e-01 -1.22544146e+00 -3.41264993e-01 -3.33597690e-01 2.27906123e-01 4.91670132e-01 -7.24834383e-01 -3.81570756e-01 -7.71212392e-03 -1.10096610e+00 -2.77257621e-01 -1.13655396e-01 6.90926015e-01 -6.03885502e-02 1.43483251e-01 -5.85527003e-01 -6.40658021e-01 3.61765206e-01 -1.10334694e+00 5.68179727e-01 -1.26391500e-01 -1.00660253e+00 -1.00211167e+00 2.37516493e-01 3.84968698e-01 3.21430027e-01 -2.35430285e-01 1.58482456e+00 -1.26809239e+00 -1.25294790e-01 -2.48509824e-01 -2.17589572e-01 2.81947017e-01 1.97715044e-01 2.24646762e-01 -1.12275064e+00 -8.64186045e-03 1.34501442e-01 -4.74710763e-01 8.11348319e-01 6.70425147e-02 8.60801637e-01 -3.66796553e-01 -2.94777155e-01 5.62559487e-03 1.22024381e+00 -2.21731916e-01 2.35081837e-01 2.67693371e-01 4.83115852e-01 1.01576293e+00 4.65173542e-01 1.62110701e-02 3.27272773e-01 6.31302536e-01 -1.90344274e-01 1.51777908e-01 1.39591724e-01 -3.61698866e-01 3.67697418e-01 8.09610009e-01 5.23898840e-01 -4.53647487e-02 -9.73834634e-01 7.24497318e-01 -1.36552930e+00 -8.81192148e-01 -4.21426415e-01 2.33585429e+00 1.23659658e+00 3.81128967e-01 1.16980731e-01 9.53253941e-04 7.20042765e-01 1.89599916e-01 -1.53367609e-01 -6.96132898e-01 -6.94513991e-02 2.74761111e-01 5.84501386e-01 7.76696265e-01 -6.39427722e-01 7.22131073e-01 7.15080309e+00 4.40057963e-01 -9.43129241e-01 -1.95049748e-01 5.45738518e-01 -1.38566867e-01 -7.56712317e-01 2.41845757e-01 -9.38887119e-01 5.07096469e-01 1.29470754e+00 -1.52993307e-01 2.63981223e-01 5.46999037e-01 2.11815313e-01 -4.95699555e-01 -1.42032301e+00 2.45855168e-01 4.37731855e-02 -9.57576871e-01 -4.12338004e-02 2.86251932e-01 5.30130863e-01 1.42925039e-01 -1.62370548e-01 4.99510974e-01 5.44426024e-01 -1.15393567e+00 6.34328425e-01 1.04680158e-01 3.71511877e-01 -6.88085437e-01 7.21822083e-01 5.56040823e-01 -2.21366525e-01 -1.68738693e-01 -5.27177334e-01 -3.73597533e-01 -2.57769316e-01 8.45537901e-01 -1.01327121e+00 -2.31597602e-01 3.15032333e-01 1.07923865e-01 -9.04456794e-01 6.71886623e-01 -8.42849687e-02 1.09396505e+00 -3.58742326e-01 -3.98447037e-01 1.23272389e-01 1.15943968e-01 2.40725264e-01 1.31825089e+00 4.71233726e-02 4.44195047e-02 -2.18326896e-01 8.85864794e-01 -9.57778543e-02 2.44771749e-01 -8.23187470e-01 -1.60667017e-01 7.39019394e-01 1.15250468e+00 -4.62458938e-01 -4.30779845e-01 -8.74047756e-01 3.30515176e-01 7.82935798e-01 8.03595707e-02 -6.50409877e-01 -1.89969063e-01 6.23539627e-01 4.15958911e-01 -2.61548231e-03 5.61478995e-02 -4.17906106e-01 -1.18402338e+00 1.44559383e-01 -1.22395456e+00 3.18234950e-01 -5.74865639e-01 -1.54327190e+00 1.91982955e-01 2.20967494e-02 -4.43031132e-01 -3.21822703e-01 -5.87945640e-01 -5.42916238e-01 1.07206607e+00 -1.27997434e+00 -3.80466938e-01 3.10456723e-01 2.02091098e-01 2.05683783e-01 2.40594760e-01 7.66274571e-01 1.63767219e-01 -4.80107605e-01 7.94979453e-01 1.70045540e-01 1.61119819e-01 1.26119876e+00 -1.20460355e+00 3.48085999e-01 5.44728220e-01 4.37775016e-01 9.78921652e-01 8.27972889e-01 -7.60502577e-01 -8.11410427e-01 -7.14371741e-01 1.73740721e+00 -9.82536852e-01 8.92975092e-01 -3.06227416e-01 -1.23320758e+00 9.72368836e-01 -5.25460169e-02 -1.82565138e-01 9.47311819e-01 8.38360548e-01 -6.89169347e-01 2.24725500e-01 -1.02198672e+00 7.52766788e-01 9.18864429e-01 -6.29528522e-01 -1.00504386e+00 3.29821557e-01 7.28132069e-01 1.09087370e-01 -7.55278766e-01 3.66452992e-01 5.63632011e-01 -9.96696055e-01 4.73227322e-01 -1.07226408e+00 6.52499020e-01 1.64227426e-01 -2.03428626e-01 -1.44574475e+00 -6.17725492e-01 -1.25835747e-01 7.42359698e-01 1.49901307e+00 9.76035297e-01 -7.84126639e-01 5.20578623e-01 1.14097989e+00 4.05360311e-02 -6.55957878e-01 -6.87456429e-01 -6.32381320e-01 6.22110486e-01 -4.66529489e-01 4.77973104e-01 1.23241591e+00 2.38922700e-01 6.54924214e-01 8.50192532e-02 -2.55976543e-02 3.35699975e-01 1.14064915e-02 7.30400205e-01 -1.08611691e+00 -3.47536832e-01 -1.83057398e-01 2.45905537e-02 -7.56366372e-01 3.23491663e-01 -1.06318724e+00 1.60367459e-01 -8.19829166e-01 3.26271683e-01 -8.26448262e-01 -1.16555221e-01 3.85708064e-01 -6.92056656e-01 -1.00702807e-01 3.56082678e-01 2.24135131e-01 -2.21220717e-01 9.74177942e-03 1.01759446e+00 3.74504268e-01 -1.23201169e-01 -1.08012937e-01 -9.91495669e-01 8.42756450e-01 6.68521762e-01 -9.14916992e-01 -2.90585272e-02 -5.68750024e-01 4.10965085e-01 -2.82975495e-01 4.67368454e-01 -6.35879695e-01 -1.02256916e-01 -2.80201584e-01 4.87091988e-01 -1.08786546e-01 4.26039658e-02 -6.87268078e-01 -1.92367792e-01 2.05862001e-01 -9.31971788e-01 3.72835875e-01 1.46004364e-01 1.85646713e-01 -5.89230806e-02 -7.35036194e-01 5.84635139e-01 -2.48993844e-01 -1.49788901e-01 -4.35324311e-01 -6.41030431e-01 6.53704703e-01 3.65570813e-01 1.11003779e-01 -4.11707103e-01 -1.85567085e-02 -4.59594041e-01 -8.12974796e-02 5.69297254e-01 4.29194003e-01 -5.62569201e-02 -1.18080199e+00 -7.06258953e-01 2.08098099e-01 1.88101456e-01 -3.74275655e-01 -3.57026458e-01 8.04660082e-01 1.40761495e-01 4.88812327e-01 3.42411309e-01 -1.30025223e-01 -1.19590557e+00 3.76174986e-01 2.21321285e-01 -1.62305295e-01 -1.33595541e-01 6.95043564e-01 1.23304226e-01 -7.06635714e-01 -1.05631717e-01 -5.09988427e-01 5.25520667e-02 4.03720975e-01 1.52867734e-01 2.12765709e-01 1.46660313e-01 -6.17594838e-01 -3.62776607e-01 1.77788883e-01 -3.97462726e-01 -3.11893374e-01 1.25860441e+00 -4.02609184e-02 -2.84589767e-01 8.94653141e-01 1.44992244e+00 4.90086257e-01 -6.94231868e-01 -3.84897321e-01 4.45134789e-01 -5.76011896e-01 -6.63017109e-02 -8.57567668e-01 -3.87914211e-01 6.52078509e-01 -1.24180447e-02 4.48568195e-01 2.84932375e-01 5.92495836e-02 3.82293493e-01 3.57457846e-01 1.50850341e-02 -1.03639603e+00 -2.80021429e-01 5.69245160e-01 6.02824032e-01 -1.42356813e+00 9.64540616e-02 -3.65066171e-01 -4.36328292e-01 8.28226864e-01 6.42418742e-01 -1.27502993e-01 5.02807498e-01 1.47787511e-01 1.64289206e-01 -3.25911731e-01 -1.12623298e+00 -7.25443661e-02 1.28924608e-01 2.17774853e-01 9.82342005e-01 -4.85842042e-02 -8.34479928e-01 4.13736850e-01 -6.24959886e-01 -2.31278792e-01 6.84321046e-01 7.99545169e-01 -5.49266577e-01 -1.09671402e+00 -5.62154770e-01 1.11419940e+00 -7.69896746e-01 -5.37068427e-01 -7.66169250e-01 1.07597113e+00 1.63137272e-01 1.07073903e+00 2.54437566e-01 -3.70727181e-01 2.77840853e-01 5.40865242e-01 3.35241079e-01 -9.46557879e-01 -8.68338704e-01 -1.56152129e-01 6.85020983e-01 -3.10010225e-01 -1.77209720e-01 -9.52300906e-01 -8.68388474e-01 -4.61122125e-01 -8.42011154e-01 2.79622644e-01 4.76830393e-01 1.24290943e+00 1.75700277e-01 9.89352539e-02 5.60069442e-01 -1.93992317e-01 -1.12719691e+00 -1.20572019e+00 -5.15550435e-01 8.24039161e-01 2.92158902e-01 -6.69209898e-01 -9.42959189e-01 -3.41788083e-01]
[10.802263259887695, 9.223336219787598]
b04c9152-54e8-4148-8c0f-35458273386f
a-multi-dimensional-deep-structured-state
2306.00331
null
https://arxiv.org/abs/2306.00331v1
https://arxiv.org/pdf/2306.00331v1.pdf
A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models
We propose a multi-dimensional structured state space (S4) approach to speech enhancement. To better capture the spectral dependencies across the frequency axis, we focus on modifying the multi-dimensional S4 layer with whitening transformation to build new small-footprint models that also achieve good performance. We explore several S4-based deep architectures in time (T) and time-frequency (TF) domains. The 2-D S4 layer can be considered a particular convolutional layer with an infinite receptive field although it utilizes fewer parameters than a conventional convolutional layer. Evaluated on the VoiceBank-DEMAND data set, when compared with the conventional U-net model based on convolutional layers, the proposed TF-domain S4-based model is 78.6% smaller in size, yet it still achieves competitive results with a PESQ score of 3.15 with data augmentation. By increasing the model size, we can even reach a PESQ score of 3.18.
['Chin-Hui Lee', 'Sabato Marco Siniscalchi', 'Chao-Han Huck Yang', 'Pin-Jui Ku']
2023-06-01
null
null
null
null
['speech-enhancement']
['speech']
[ 1.25777237e-02 4.27436568e-02 -1.97173804e-02 -2.13842481e-01 -7.14885592e-01 -1.78055823e-01 1.99336156e-01 -3.58235896e-01 -5.06012678e-01 1.88076198e-01 4.88972962e-01 -6.16540551e-01 1.97161973e-01 -4.02217358e-01 -3.75326604e-01 -4.44678992e-01 -2.50063002e-01 -5.27826667e-01 1.85164154e-01 -4.93436426e-01 -2.39602104e-01 5.29085159e-01 -1.34867263e+00 2.38304317e-01 7.12968647e-01 1.22061026e+00 2.96271414e-01 9.69417095e-01 8.83958191e-02 5.30553639e-01 -8.29440117e-01 8.99068117e-02 2.17016190e-01 -1.58348113e-01 -7.09881783e-01 -6.80839568e-02 1.71219066e-01 -3.69954795e-01 -8.29176009e-01 9.82828259e-01 9.35071707e-01 5.66818058e-01 1.44044250e-01 -5.91474771e-01 -7.66236365e-01 5.96300364e-01 -3.48736763e-01 4.68030214e-01 -5.90271614e-02 -1.06713764e-01 9.92933989e-01 -9.06841874e-01 2.86453098e-01 1.43762052e+00 7.50145793e-01 7.98432291e-01 -1.15228450e+00 -5.83460510e-01 3.22439879e-01 4.17280197e-01 -9.16447878e-01 -7.56605387e-01 9.60199177e-01 8.82613435e-02 1.58637142e+00 3.81747633e-01 4.31235433e-01 1.17201114e+00 -7.27142841e-02 8.78607213e-01 9.74453449e-01 -5.89674354e-01 1.00968175e-01 -2.97692478e-01 2.44606644e-01 4.28853393e-01 -5.47133207e-01 3.83694559e-01 -7.64668226e-01 2.95404885e-02 9.23717141e-01 -4.04113919e-01 -2.68548727e-01 3.16421062e-01 -1.08407855e+00 5.55630684e-01 5.00180483e-01 4.75644886e-01 -3.32810014e-01 3.22123438e-01 6.39497519e-01 5.32920063e-01 6.63329661e-01 3.99516433e-01 -8.61718714e-01 -3.76466334e-01 -1.05885673e+00 1.05399184e-01 4.14668351e-01 7.08916366e-01 3.21688652e-01 9.45690274e-01 -3.73070359e-01 1.11256325e+00 7.62836710e-02 4.40949231e-01 7.10842073e-01 -1.09107816e+00 3.86827946e-01 6.16206639e-02 -5.05875498e-02 -3.12917680e-01 -6.87026083e-01 -8.37412298e-01 -1.16903591e+00 6.63151294e-02 1.12039022e-01 -6.06160343e-01 -1.16752541e+00 2.06547284e+00 7.66604468e-02 4.26462203e-01 2.83947170e-01 9.36649442e-01 8.76446724e-01 1.03686440e+00 -1.75175034e-02 -2.77702898e-01 1.33945382e+00 -1.18057287e+00 -1.27496099e+00 -3.12623352e-01 3.91863495e-01 -7.94216037e-01 1.06023431e+00 2.70951688e-01 -1.28933656e+00 -1.02971268e+00 -1.18098950e+00 -2.29919866e-01 -2.83651173e-01 3.86507750e-01 4.50957149e-01 8.52948606e-01 -1.46822047e+00 9.39750016e-01 -9.03169632e-01 -4.70258296e-03 -5.26783504e-02 3.81491631e-01 -2.74489731e-01 5.91214359e-01 -1.88547587e+00 8.42795551e-01 3.19085300e-01 5.21545447e-02 -7.85279453e-01 -7.83759236e-01 -9.30324733e-01 5.49935877e-01 3.18388361e-03 -3.27592731e-01 1.51739979e+00 -3.58609319e-01 -1.99247432e+00 1.36218056e-01 -2.98235923e-01 -6.50930405e-01 -1.43119395e-01 -2.42583200e-01 -9.64384854e-01 2.37952806e-02 -5.06347835e-01 6.08788311e-01 1.08576441e+00 -7.68255591e-01 -5.69755971e-01 -1.23753265e-01 2.09191740e-02 -1.33595644e-02 -8.44557524e-01 3.22532058e-01 -2.06698343e-01 -1.15573955e+00 2.01211095e-01 -9.34028447e-01 -3.65526110e-01 -4.30814594e-01 -2.18556821e-01 -5.96441738e-02 9.31643307e-01 -1.08158576e+00 1.72862852e+00 -2.41636610e+00 1.24015538e-02 -2.04214618e-01 1.17069609e-01 9.82435107e-01 -3.82934421e-01 1.74698278e-01 -4.07911718e-01 1.44907191e-01 -4.32994775e-02 -5.70774436e-01 2.06795469e-01 -1.58681553e-02 -1.86276257e-01 2.01859549e-01 5.79636157e-01 6.68301582e-01 -5.49463868e-01 7.61779323e-02 2.56467879e-01 8.72206271e-01 -9.40138280e-01 1.34132490e-01 1.17839970e-01 3.17223877e-01 -5.79992197e-02 1.91962242e-01 8.24754953e-01 9.94210094e-02 2.06457287e-01 -3.64957452e-01 -3.59851986e-01 8.03765237e-01 -1.23596692e+00 1.90557635e+00 -8.26966763e-01 9.72603440e-01 2.33159497e-01 -7.63784289e-01 1.14672971e+00 8.51784825e-01 5.92492819e-01 -7.55274653e-01 1.45229185e-02 1.45320490e-01 3.53278428e-01 -2.88606472e-02 5.95628440e-01 -2.05015078e-01 -1.69145092e-02 2.48619676e-01 6.45407736e-01 1.40623808e-01 -1.81800887e-01 -2.57374436e-01 1.16557157e+00 -1.10960603e-01 -1.15457028e-01 -3.55421036e-01 5.54309666e-01 -7.52883375e-01 7.73787498e-01 3.95456821e-01 -5.27225614e-01 4.29345638e-01 3.42631489e-01 -3.53911102e-01 -1.20762646e+00 -1.05132008e+00 -2.65657067e-01 1.29133260e+00 -3.78163815e-01 -5.76818168e-01 -9.06593144e-01 -2.17305914e-01 -3.20051283e-01 4.84674245e-01 -3.76030475e-01 -4.63689029e-01 -8.04456770e-01 -5.85965514e-01 9.14514482e-01 7.30479419e-01 7.40112782e-01 -9.89600778e-01 -3.92082065e-01 5.61596513e-01 -2.95908630e-01 -1.14920342e+00 -7.97582924e-01 7.72230566e-01 -8.52594078e-01 -1.95850745e-01 -1.02678084e+00 -8.36423755e-01 -4.78280112e-02 -6.89833285e-03 7.83451080e-01 -4.77479339e-01 2.79838800e-01 -2.26320803e-01 -5.55521905e-01 -4.77319628e-01 -4.57475901e-01 2.63419062e-01 4.20190513e-01 -7.56668532e-03 1.75587311e-01 -6.18560493e-01 -6.83007777e-01 1.26005024e-01 -7.97044337e-01 -1.57950148e-01 3.12676847e-01 1.06104863e+00 2.90031344e-01 1.62148148e-01 1.13973248e+00 -2.55904436e-01 6.97039604e-01 3.76150496e-02 -2.36929074e-01 -6.37696087e-02 -4.74822134e-01 1.74039930e-01 6.27682388e-01 -5.79032362e-01 -1.42894208e+00 -1.16213843e-01 -7.66549468e-01 -7.01277673e-01 5.53621240e-02 3.42903286e-01 -7.02079982e-02 1.22167274e-01 6.72228396e-01 -1.12210847e-02 -3.46182957e-02 -1.02135253e+00 4.62445438e-01 9.97182727e-01 5.08787811e-01 -2.13733330e-01 5.91205895e-01 2.77990028e-02 -3.38300556e-01 -1.01789606e+00 -6.66601002e-01 -5.19181788e-01 -5.50510943e-01 -2.59508025e-02 7.43137836e-01 -1.00678802e+00 -5.71244657e-01 6.65883899e-01 -1.34971058e+00 -4.38411325e-01 -3.80169570e-01 8.00825119e-01 -5.65652907e-01 2.69130677e-01 -1.19193602e+00 -9.00393307e-01 -6.09967887e-01 -1.14054811e+00 9.81566668e-01 1.72951189e-03 7.92841520e-03 -8.50205421e-01 5.24054794e-03 -2.61380952e-02 7.77214944e-01 -2.60542452e-01 8.66671681e-01 -4.75279301e-01 1.40003130e-01 1.26413882e-01 3.16822832e-03 9.95570540e-01 1.77217215e-01 -2.83747554e-01 -1.60772514e+00 -4.28210288e-01 3.38497341e-01 1.62074491e-02 9.77244020e-01 6.32105589e-01 1.34077883e+00 -1.16977632e-01 2.88758993e-01 7.47136891e-01 8.05538833e-01 5.89430869e-01 6.97088778e-01 1.34630069e-01 3.28068197e-01 3.63988519e-01 4.48260516e-01 5.03610313e-01 9.02228504e-02 6.96604848e-01 2.42475376e-01 -5.78597903e-01 -6.58106923e-01 -8.09947178e-02 6.14012301e-01 1.62656856e+00 -1.42661124e-01 -2.07013816e-01 -6.52928054e-01 5.70836008e-01 -1.57123244e+00 -9.34555769e-01 2.18124747e-01 1.78844547e+00 9.86692429e-01 2.26605237e-01 1.04584709e-01 6.72398865e-01 8.06456089e-01 5.54262877e-01 -6.26397252e-01 -9.34862733e-01 -1.22987390e-01 7.62281835e-01 2.73828685e-01 4.32944506e-01 -1.20321798e+00 1.07827377e+00 7.12637377e+00 1.11566913e+00 -1.34567249e+00 2.58818358e-01 5.42481124e-01 -1.76474094e-01 -9.96802188e-03 -4.58793759e-01 -6.78417325e-01 1.23739906e-01 1.60356438e+00 -1.19467899e-01 4.63266313e-01 6.19719446e-01 4.52096611e-01 7.21877337e-01 -6.95671260e-01 9.13135886e-01 -4.31168616e-01 -1.15196896e+00 -4.06882167e-01 8.45193397e-03 6.98213160e-01 1.00868322e-01 4.48443413e-01 6.64708614e-01 -1.11046948e-01 -8.92984867e-01 6.67523503e-01 7.91034028e-02 1.34250605e+00 -1.01339233e+00 4.62241769e-01 1.55810967e-01 -1.46121109e+00 -2.43713662e-01 -3.79775167e-01 -1.52159452e-01 2.10578963e-01 4.10868913e-01 -6.16601229e-01 4.51752424e-01 9.14651573e-01 4.97230411e-01 -1.01682477e-01 5.72552681e-01 1.71987265e-01 9.29183424e-01 -1.80516973e-01 2.36447841e-01 4.48732674e-01 3.07368934e-01 5.99317849e-01 1.43932784e+00 5.34100235e-01 -6.46464750e-02 -3.07058543e-01 4.88472462e-01 -1.64737582e-01 -2.12608710e-01 -1.44651994e-01 -2.32040677e-02 3.87610406e-01 1.06977367e+00 -1.52609870e-02 -2.35389292e-01 -4.84344810e-01 8.76697183e-01 2.36486956e-01 4.85547781e-01 -9.78895187e-01 -7.88164198e-01 1.23473871e+00 -2.10323736e-01 4.73725706e-01 -4.33664352e-01 -2.86136493e-02 -1.07355106e+00 -2.27741018e-01 -9.84624743e-01 -7.73318037e-02 -6.35138452e-01 -8.88754070e-01 1.18075037e+00 -4.45725679e-01 -1.19647717e+00 -3.61623406e-01 -8.13447773e-01 -3.48797053e-01 1.26206684e+00 -1.70329452e+00 -9.17000890e-01 2.49577403e-01 6.58391953e-01 6.56287372e-01 -3.16050053e-01 1.15343428e+00 7.02618122e-01 -6.24314368e-01 8.53864431e-01 1.42161533e-01 1.05425909e-01 6.59734786e-01 -1.34103692e+00 1.22382271e+00 1.03542161e+00 9.27410647e-02 6.06555939e-01 3.57731283e-01 -3.08502883e-01 -1.00800765e+00 -1.23758662e+00 9.18741167e-01 1.64359137e-01 6.71080768e-01 -3.46056610e-01 -1.08529723e+00 4.36928630e-01 3.85972083e-01 9.34156999e-02 5.84626138e-01 3.17670763e-01 -3.49727362e-01 -1.45894632e-01 -9.65574145e-01 7.23854125e-01 1.15051210e+00 -9.61967826e-01 -4.76447463e-01 -1.01045914e-01 1.38626277e+00 -5.13815939e-01 -1.31828511e+00 4.96545672e-01 2.85312116e-01 -8.48922670e-01 1.13667512e+00 -7.04046845e-01 -3.17167044e-02 -6.30330294e-02 -3.21298987e-01 -1.69051027e+00 -8.49488735e-01 -1.11737967e+00 -5.19406199e-01 1.01290607e+00 4.44026083e-01 -4.04825985e-01 5.68362355e-01 1.63518742e-01 -7.04789400e-01 -6.62113130e-01 -1.28037798e+00 -9.83300507e-01 2.24356636e-01 -8.00392032e-01 8.14922452e-01 6.63330853e-01 1.34537295e-01 2.45755613e-01 -6.68145120e-01 2.19312772e-01 1.14319518e-01 -3.96010786e-01 6.16137087e-02 -8.80719423e-01 -2.38906071e-01 -4.31961209e-01 -1.76088005e-01 -1.26576591e+00 2.33942628e-01 -6.62031710e-01 1.92134362e-02 -1.12505603e+00 -4.58394051e-01 -3.40500832e-01 -9.01280999e-01 4.52751994e-01 -2.59753734e-01 7.73311555e-02 3.04922432e-01 -3.12707812e-01 -5.75876748e-03 7.20331550e-01 1.43587887e+00 -6.65193871e-02 -4.24109906e-01 5.78149930e-02 -4.08246130e-01 5.98642707e-01 1.05669022e+00 9.85173658e-02 -3.79363388e-01 -6.01792097e-01 -3.68701965e-01 3.71742159e-01 9.87844244e-02 -1.16764235e+00 1.16936699e-01 2.10665524e-01 1.99182943e-01 -6.08518004e-01 8.62159133e-01 -6.12154365e-01 -2.15312392e-01 4.50505614e-01 -4.88704741e-01 -4.94657233e-02 6.32632911e-01 3.33441705e-01 -6.08021736e-01 5.84429689e-02 1.04093707e+00 1.65309131e-01 -7.53435373e-01 3.57976884e-01 -5.48198223e-01 -2.69130319e-01 3.42310131e-01 9.40356702e-02 -1.85020402e-01 -3.65756601e-01 -1.11668181e+00 -2.67656952e-01 -3.35480988e-01 6.36322081e-01 5.39826930e-01 -1.60639119e+00 -6.52449787e-01 5.30192077e-01 -2.55243719e-01 -5.07219791e-01 7.42107987e-01 4.21389461e-01 -1.12559445e-01 7.69137740e-01 -2.12299407e-01 -3.38751674e-01 -9.70113039e-01 3.65300357e-01 5.34748673e-01 -3.94478232e-01 -6.08362436e-01 1.05584538e+00 2.03820616e-01 -5.59438467e-01 3.56025010e-01 -6.15689397e-01 -3.27291071e-01 -1.16327316e-01 5.49027860e-01 6.34568095e-01 4.83532697e-01 -5.83202481e-01 -3.85853678e-01 2.26507664e-01 1.21825814e-01 -3.90865624e-01 1.45024240e+00 -6.41840175e-02 4.24853325e-01 2.45057359e-01 1.29572129e+00 -9.24281552e-02 -1.36985028e+00 -4.83930111e-01 -1.36764690e-01 -4.99439053e-02 7.51026869e-01 -7.82721937e-01 -1.30722892e+00 1.21972060e+00 9.33401287e-01 3.31475884e-01 1.54681683e+00 -4.08311367e-01 1.16510379e+00 8.73635709e-02 -8.59974697e-02 -1.37308812e+00 1.47107974e-01 1.02705181e+00 9.70317960e-01 -9.08282816e-01 -5.31575680e-01 -1.47580653e-01 -5.15577972e-01 1.08887839e+00 4.52613443e-01 3.16692963e-02 9.78926122e-01 5.69892108e-01 2.20639244e-01 2.65951335e-01 -9.68817770e-01 -4.94559318e-01 5.35459578e-01 7.58900642e-01 7.30142057e-01 8.46145153e-02 5.76721802e-02 7.77080536e-01 -3.24527055e-01 -2.99428344e-01 2.57594019e-01 5.50983012e-01 -4.70710903e-01 -1.33007216e+00 -3.20743710e-01 1.86330944e-01 -6.74658418e-01 -3.06149572e-01 1.84711114e-01 4.25420970e-01 -4.24369052e-02 1.28941083e+00 1.78036898e-01 -8.29037666e-01 6.54796720e-01 1.68948635e-01 1.72174886e-01 -4.39423293e-01 -8.77466381e-01 6.45745695e-01 2.32252955e-01 -5.09665847e-01 -2.77783513e-01 -2.98003793e-01 -1.11576319e+00 -3.76750141e-01 -4.36080962e-01 -4.46324870e-02 7.78597832e-01 5.20691097e-01 4.78775412e-01 1.27886879e+00 8.39560688e-01 -8.46170664e-01 -6.89007759e-01 -1.38504207e+00 -7.54242957e-01 8.90463963e-02 9.29286361e-01 -4.51909751e-01 -2.01428741e-01 1.55582931e-02]
[14.89173698425293, 5.957709312438965]
3c64de12-26b9-495f-abad-120282fc7362
m6doc-a-large-scale-multi-format-multi-type
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Cheng_M6Doc_A_Large-Scale_Multi-Format_Multi-Type_Multi-Layout_Multi-Language_Multi-Annotation_Category_Dataset_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Cheng_M6Doc_A_Large-Scale_Multi-Format_Multi-Type_Multi-Layout_Multi-Language_Multi-Annotation_Category_Dataset_CVPR_2023_paper.pdf
M6Doc: A Large-Scale Multi-Format, Multi-Type, Multi-Layout, Multi-Language, Multi-Annotation Category Dataset for Modern Document Layout Analysis
Document layout analysis is a crucial prerequisite for document understanding, including document retrieval and conversion. Most public datasets currently contain only PDF documents and lack realistic documents. Models trained on these datasets may not generalize well to real-world scenarios. Therefore, this paper introduces a large and diverse document layout analysis dataset called M^6-Doc. The M^6 designation represents six properties: (1) Multi-Format (including scanned, photographed, and PDF documents); (2) Multi-Type (such as scientific articles, textbooks, books, test papers, magazines, newspapers, and notes); (3) Multi-Layout (rectangular, Manhattan, non-Manhattan, and multi-column Manhattan); (4) Multi-Language (Chinese and English); (5) Multi-Annotation Category (74 types of annotation labels with 237,116 annotation instances in 9,080 manually annotated pages); and (6) Modern documents. Additionally, we propose a transformer-based document layout analysis method called TransDLANet, which leverages an adaptive element matching mechanism that enables query embedding to better match ground truth to improve recall, and constructs a segmentation branch for more precise document image instance segmentation. We conduct a comprehensive evaluation of M^6-Doc with various layout analysis methods and demonstrate its effectiveness. TransDLANet achieves state-of-the-art performance on M^6-Doc with 64.5% mAP. The M^6-Doc dataset will be available at https://github.com/HCIILAB/M6Doc.
['Lianwen Jin', 'Kai Ding', 'Jing Li', 'Zecheng Xie', 'Qiyuan Zhu', 'Jiaxin Zhang', 'Sihang Wu', 'Peirong Zhang', 'Hiuyi Cheng']
2023-01-01
null
null
null
cvpr-2023-1
['document-layout-analysis']
['computer-vision']
[-1.04418606e-01 -4.06369567e-01 -3.57886165e-01 -1.47964627e-01 -1.04064047e+00 -1.04317260e+00 6.96394622e-01 3.41510862e-01 -9.19019505e-02 4.64353621e-01 3.81617725e-01 -7.01539218e-01 -4.71862793e-01 -9.09884453e-01 -6.45511329e-01 -3.45527798e-01 8.03525075e-02 6.66131079e-01 2.45796695e-01 7.20634088e-02 5.93428314e-01 6.04083121e-01 -1.03451598e+00 6.52355671e-01 1.02980959e+00 1.03441978e+00 2.82097250e-01 8.76528561e-01 -6.62204862e-01 3.38829935e-01 -8.13419878e-01 -5.99213660e-01 5.87699972e-02 -7.96478838e-02 -6.39582932e-01 3.12234282e-01 5.76429188e-01 -4.12440747e-01 -3.97973150e-01 9.45689023e-01 4.37987179e-01 -2.84587860e-01 9.03050542e-01 -1.07132471e+00 -1.18599987e+00 6.45972610e-01 -1.09866869e+00 -5.49295768e-02 4.24991399e-01 -1.70703188e-01 1.08469737e+00 -1.05716729e+00 7.95601606e-01 1.34942961e+00 4.85349894e-01 -9.49119255e-02 -7.15636194e-01 -4.08724159e-01 1.19669676e-01 8.28745961e-02 -1.58157861e+00 -2.04235334e-02 5.48187673e-01 -5.05433440e-01 6.52276576e-01 4.75459784e-01 4.04508740e-01 1.02305925e+00 3.30716968e-01 1.04595578e+00 1.02037489e+00 -7.32853055e-01 5.19256927e-02 1.22055978e-01 3.16425949e-01 8.19135010e-01 5.20219445e-01 -7.93030977e-01 -1.88723192e-01 2.18790218e-01 6.41545713e-01 3.78303416e-02 -3.02347034e-01 -8.22439343e-02 -1.17037618e+00 4.89809126e-01 1.95504874e-01 4.06356215e-01 8.78369585e-02 -2.30142146e-01 6.56496167e-01 -6.54540509e-02 1.93568692e-01 5.31974077e-01 -2.88896561e-01 -1.51949659e-01 -1.10275865e+00 3.62558186e-01 6.36908412e-01 1.55756974e+00 3.28741014e-01 -2.24740028e-01 -3.62252772e-01 1.16028678e+00 3.34575325e-01 7.61111200e-01 5.33743501e-01 -6.68326378e-01 1.13386345e+00 8.66364002e-01 -7.88926482e-02 -1.00568092e+00 -3.32883596e-01 -3.97155017e-01 -9.40014124e-01 -4.78297085e-01 2.19271138e-01 3.03408742e-01 -9.73160207e-01 8.73816609e-01 -2.10639283e-01 -6.92954540e-01 -8.54368806e-02 5.54294586e-01 1.03683186e+00 7.70967245e-01 -3.55259389e-01 1.37341782e-01 1.70412147e+00 -1.33067715e+00 -8.06798339e-01 -4.99826744e-02 6.46524727e-01 -1.12086689e+00 1.50018346e+00 5.17665207e-01 -9.00357306e-01 -5.41248560e-01 -1.09248650e+00 -3.36111784e-01 -8.34402502e-01 6.39361680e-01 4.84151959e-01 7.79889047e-01 -6.82429552e-01 2.53361613e-02 -3.02825749e-01 -4.90966231e-01 6.00679398e-01 -1.14166234e-02 -3.72678071e-01 -5.96646667e-01 -6.74749017e-01 2.85529554e-01 4.30450946e-01 -1.86903790e-01 -3.38037521e-01 -5.26312113e-01 -5.70092916e-01 1.35520563e-01 4.64973330e-01 -2.98356414e-01 9.67084587e-01 -2.48070180e-01 -9.94781613e-01 7.04438806e-01 -3.74127813e-02 1.03259921e-01 6.48894548e-01 -2.81518340e-01 -8.72220576e-01 3.06061774e-01 2.19024166e-01 5.19678950e-01 3.61749768e-01 -1.34117854e+00 -5.16126096e-01 -4.89963621e-01 -7.55327493e-02 1.44586876e-01 -5.56140125e-01 -1.95295706e-01 -1.39963615e+00 -8.63432467e-01 3.62514615e-01 -6.91958785e-01 3.75883162e-01 -7.58872181e-02 -1.17854011e+00 -1.19092681e-01 1.18774509e+00 -9.79991436e-01 1.66375828e+00 -2.09016943e+00 -3.70845854e-01 4.47244197e-01 5.04732803e-02 1.23017155e-01 -2.07737565e-01 7.94995904e-01 4.04691458e-01 5.60526907e-01 3.18027399e-02 -1.98254853e-01 4.12767768e-01 -4.23923247e-02 -4.63600785e-01 4.17184755e-02 -2.68935472e-01 1.24336219e+00 -4.07077402e-01 -8.53012621e-01 1.45849720e-01 2.53935248e-01 -2.65646905e-01 -2.03281432e-01 -2.55565017e-01 -1.61904112e-01 -6.36829495e-01 1.22316015e+00 8.97462904e-01 -3.97534788e-01 3.11157078e-01 -5.52510917e-01 -1.98861241e-01 -1.67508330e-02 -1.14357436e+00 1.51629829e+00 -3.31619591e-01 9.37951922e-01 -2.18952209e-01 -4.60328549e-01 8.58119071e-01 -1.50681019e-01 3.81997347e-01 -9.62812603e-01 1.18653469e-01 2.45175511e-01 -2.56022096e-01 -5.25473714e-01 1.08941424e+00 9.46618736e-01 -3.12102079e-01 4.25236851e-01 -2.75595099e-01 -1.63058743e-01 6.90647662e-01 5.35110354e-01 1.11360776e+00 -1.23334080e-01 -1.40933931e-01 -1.87165841e-01 3.24070841e-01 -7.58247375e-02 2.58816838e-01 7.98501611e-01 1.07723944e-01 7.06267595e-01 8.12462687e-01 7.84618855e-02 -1.31557572e+00 -1.13445258e+00 -3.52983356e-01 7.82749057e-01 1.39594078e-01 -7.23810196e-01 -8.79525900e-01 -5.90252697e-01 2.29517907e-01 5.95728040e-01 -3.23465586e-01 2.64818072e-01 -4.48264629e-01 -4.87982750e-01 8.80947113e-01 8.50665510e-01 8.62951696e-01 -7.13877022e-01 -2.20712036e-01 -1.11136891e-01 -4.33645844e-02 -1.12798905e+00 -8.69577050e-01 7.05464259e-02 -6.99297011e-01 -1.01959765e+00 -8.72365475e-01 -8.63850892e-01 8.71973634e-01 3.30290228e-01 8.05703580e-01 -7.92371705e-02 -2.45216504e-01 4.28264350e-01 -4.50454563e-01 -3.41576487e-01 7.55736604e-02 2.54305005e-01 -2.99712569e-01 -3.32462937e-01 2.20175654e-01 1.73330382e-01 -5.95022202e-01 3.06503654e-01 -1.14977634e+00 -5.12973517e-02 8.15675378e-01 8.47600937e-01 7.73063898e-01 5.39554238e-01 8.23298544e-02 -1.15463948e+00 9.39846396e-01 -2.38665000e-01 -5.89053333e-01 7.44097531e-01 -9.60238814e-01 -2.33447462e-01 6.75305843e-01 -2.54935682e-01 -9.68842268e-01 -4.30841595e-01 3.67291681e-02 -1.91800132e-01 -1.63848177e-02 8.10414135e-01 -6.31634533e-01 2.85164267e-01 2.67995268e-01 2.88731396e-01 -5.75741410e-01 -8.41494322e-01 4.40786153e-01 1.20866156e+00 7.42534399e-01 -7.12680161e-01 7.13649511e-01 3.06671262e-01 -1.71724603e-01 -8.68040502e-01 -5.47881544e-01 -4.91040438e-01 -8.20207119e-01 -4.41599861e-02 6.00760877e-01 -5.53685427e-01 -5.43394148e-01 5.38138807e-01 -1.05494630e+00 -2.50923336e-01 1.37021542e-01 2.12456837e-01 7.09144771e-02 6.09233260e-01 -7.82740474e-01 -5.41665316e-01 -5.03814578e-01 -1.10279179e+00 1.22783554e+00 3.38413417e-01 -1.36175469e-01 -9.74178910e-01 -2.66348302e-01 5.80996275e-01 1.39816189e-02 -5.90154380e-02 1.45231390e+00 -3.62286240e-01 -7.24778354e-01 -3.66938144e-01 -8.16029489e-01 2.17133150e-01 6.93778172e-02 6.08042359e-01 -5.93634486e-01 -2.36478180e-01 -1.01096988e+00 -1.13843076e-01 9.23431516e-01 4.00288463e-01 1.63233495e+00 -3.83060247e-01 -4.96147722e-01 5.50161898e-01 1.44050133e+00 5.42005241e-01 1.00326872e+00 6.80342495e-01 9.24576402e-01 3.94220054e-01 6.40757263e-01 4.31888670e-01 5.87381124e-01 6.07644796e-01 1.93298250e-01 -5.68199679e-02 -1.89620450e-01 -4.64323759e-01 9.72870216e-02 1.18122673e+00 2.97101885e-01 -9.26945925e-01 -1.11539316e+00 5.84204972e-01 -1.42281914e+00 -8.28810036e-01 -4.03762668e-01 1.80669892e+00 5.71582913e-01 4.44513470e-01 -8.93571451e-02 3.34846050e-01 6.50166094e-01 2.36909658e-01 -4.26110446e-01 -5.61588287e-01 -4.88147497e-01 4.33046073e-02 7.24479616e-01 -1.70444809e-02 -1.09256315e+00 9.39098954e-01 5.74388981e+00 1.21999681e+00 -8.66678655e-01 -9.30142999e-02 5.51666915e-01 3.44846606e-01 -6.39383793e-01 -2.00192899e-01 -9.88356411e-01 7.75696695e-01 3.81512642e-01 1.64878629e-02 1.36430010e-01 7.62144387e-01 -2.31356636e-01 -1.80917114e-01 -7.60344744e-01 1.09432518e+00 9.39531624e-02 -1.61036181e+00 3.81973147e-01 2.38873333e-01 8.89325082e-01 -4.28036779e-01 4.65486050e-01 2.99050689e-01 -8.02665204e-02 -9.79995012e-01 1.02288163e+00 4.20454115e-01 1.27887130e+00 -6.71982706e-01 6.52085602e-01 -3.53306495e-02 -1.29677784e+00 -9.36504230e-02 -2.80340374e-01 7.71536112e-01 3.47583182e-02 7.50806153e-01 -5.34357905e-01 8.35684478e-01 8.65680754e-01 6.50705457e-01 -1.03393066e+00 1.07878470e+00 -1.04167223e-01 5.23660004e-01 -4.12952155e-02 -2.55782008e-01 3.33739549e-01 -3.36454153e-01 8.92238691e-02 1.52286184e+00 5.49582601e-01 -4.16259676e-01 -4.24556155e-03 6.31319582e-01 -4.50742543e-01 2.66912907e-01 -2.36948177e-01 -6.44337833e-01 9.24136341e-01 1.13554072e+00 -1.31652641e+00 -2.41019398e-01 -4.24048066e-01 9.28817451e-01 -2.30938867e-01 5.55024385e-01 -7.71156430e-01 -1.02790248e+00 3.40494327e-02 4.87374999e-02 4.43557173e-01 -3.54207367e-01 -5.79919100e-01 -9.04285908e-01 3.65251541e-01 -9.41540539e-01 3.47194552e-01 -1.05843222e+00 -1.02766192e+00 5.27431130e-01 -1.76023543e-01 -1.23390567e+00 2.24468082e-01 -8.88906002e-01 -3.70290101e-01 6.08423173e-01 -1.14363182e+00 -1.39042819e+00 -5.28153360e-01 2.20361546e-01 6.81777060e-01 -4.88357872e-01 6.90566301e-01 5.88621259e-01 -9.39630508e-01 8.77543926e-01 8.14312339e-01 4.49252695e-01 7.53917754e-01 -1.38693881e+00 4.28168029e-01 5.87057352e-01 2.65664190e-01 7.54149020e-01 1.10029154e-01 -6.10669374e-01 -1.59458780e+00 -1.24189806e+00 6.41080081e-01 -4.61369306e-01 4.95715022e-01 -5.62447846e-01 -8.47094297e-01 6.16612971e-01 1.01556055e-01 -3.55415553e-01 5.56346297e-01 1.44729882e-01 -6.44782364e-01 -3.03703457e-01 -9.23224449e-01 8.52936804e-01 8.13390315e-01 -7.14009464e-01 -1.99149415e-01 4.64830518e-01 5.53472579e-01 -7.13942051e-01 -9.01530445e-01 8.68837535e-02 8.92418385e-01 -6.70421839e-01 8.65244210e-01 -1.34256780e-01 7.20169783e-01 -2.74437129e-01 -3.29959720e-01 -7.49460578e-01 -2.40585566e-01 -1.00471333e-01 -4.41778392e-01 1.52572608e+00 3.93752456e-01 -4.16518986e-01 7.97165692e-01 1.47055432e-01 -5.51777780e-01 -1.11929607e+00 -4.92330819e-01 -1.07310021e+00 9.70764458e-02 -3.22780848e-01 9.60508287e-01 8.34142447e-01 -2.66932666e-01 6.82300478e-02 1.55616431e-02 -8.60638693e-02 2.28462845e-01 2.86372364e-01 8.24325442e-01 -8.94606948e-01 -1.04514763e-01 -6.76798999e-01 -1.79312736e-01 -1.48888338e+00 -8.29651803e-02 -8.97890806e-01 -4.51019943e-01 -2.11652231e+00 3.65553230e-01 -4.42721844e-01 -8.96261930e-02 4.72726643e-01 3.79802614e-01 2.10232928e-01 -1.10198744e-01 4.69506830e-01 -7.44061768e-01 1.88517347e-01 1.44359100e+00 -6.81875587e-01 -2.03031860e-02 -3.99325252e-01 -7.80698538e-01 3.84754509e-01 5.42937279e-01 1.17295012e-02 -3.81916046e-01 -7.12192595e-01 2.16678351e-01 -3.50649625e-01 9.60342139e-02 -8.18275869e-01 2.50836223e-01 -5.48142903e-02 8.19639325e-01 -1.22944343e+00 1.68318495e-01 -5.75864196e-01 -1.33908704e-01 1.95465595e-01 -2.10223749e-01 3.88980538e-01 2.08703414e-01 2.95651466e-01 -2.69095510e-01 -3.36026818e-01 2.56419390e-01 1.77781224e-01 -9.23366904e-01 1.32786185e-01 -2.65153080e-01 1.24589153e-01 8.44776809e-01 -4.66795623e-01 -1.15318537e+00 -1.49898678e-01 9.71334428e-02 2.36633480e-01 5.92406034e-01 5.62809587e-01 6.86188042e-01 -1.26947165e+00 -3.48227352e-01 3.47950310e-02 4.61177796e-01 -8.50312859e-02 2.26059720e-01 7.04120040e-01 -9.99757707e-01 9.32702422e-01 -6.16916306e-02 -4.19497699e-01 -1.21310973e+00 3.53556991e-01 -3.28022271e-01 -3.76324952e-01 -6.08330429e-01 5.64909279e-01 -1.03240110e-01 -3.58113527e-01 3.33141983e-01 -4.00466681e-01 -1.11450627e-01 1.93211854e-01 3.92062068e-01 6.60364747e-01 2.86618263e-01 -5.07187128e-01 -2.17597157e-01 6.60438597e-01 -5.03092587e-01 -1.36584014e-01 9.02810514e-01 -3.38364616e-02 -2.23676831e-01 2.20566124e-01 1.40270591e+00 4.76534963e-01 -8.15938771e-01 2.35319547e-02 2.29589194e-01 -5.45480072e-01 -5.15385382e-02 -1.03650939e+00 -8.85201037e-01 7.82192230e-01 4.15574193e-01 1.34585992e-01 1.15968025e+00 -1.13240287e-01 9.28225994e-01 5.32610953e-01 2.41391763e-01 -1.55982757e+00 4.08275843e-01 6.30597353e-01 9.71293986e-01 -8.19120169e-01 5.11246145e-01 -3.32426637e-01 -5.16011655e-01 1.31292617e+00 4.71865475e-01 4.44652528e-01 4.72122133e-01 2.83892125e-01 7.95829073e-02 -2.33140320e-01 -3.31368893e-01 7.10412785e-02 6.55391276e-01 3.26912910e-01 5.02713084e-01 8.65084678e-02 -2.89565235e-01 8.45390320e-01 -3.67957652e-01 -4.88329947e-01 3.83695602e-01 1.11862195e+00 -1.73446730e-01 -1.11811972e+00 -4.24195349e-01 9.48155940e-01 -2.29748830e-01 -4.62235743e-03 -5.47944844e-01 9.80730534e-01 1.07245602e-01 6.93260431e-01 1.83303207e-01 -3.96410346e-01 3.36502969e-01 -7.59407729e-02 4.77135897e-01 -3.46592933e-01 -1.22541204e-01 2.36821696e-01 4.45422856e-03 -2.08581656e-01 2.16955543e-01 -5.84316313e-01 -1.23983610e+00 -5.45651495e-01 -3.24116081e-01 4.39889403e-03 8.62795830e-01 4.98497486e-01 4.82994139e-01 7.78288424e-01 3.38187963e-01 -2.46631235e-01 -1.48118198e-01 -9.71384346e-01 -8.72599065e-01 2.40537092e-01 -1.77609891e-01 -4.90852416e-01 7.19999671e-02 1.36616647e-01]
[11.685769081115723, 2.5379719734191895]
95d6b910-e26b-462c-aa41-94a23272b47e
learning-to-classify-intents-and-slot-labels
2004.10793
null
https://arxiv.org/abs/2004.10793v1
https://arxiv.org/pdf/2004.10793v1.pdf
Learning to Classify Intents and Slot Labels Given a Handful of Examples
Intent classification (IC) and slot filling (SF) are core components in most goal-oriented dialogue systems. Current IC/SF models perform poorly when the number of training examples per class is small. We propose a new few-shot learning task, few-shot IC/SF, to study and improve the performance of IC and SF models on classes not seen at training time in ultra low resource scenarios. We establish a few-shot IC/SF benchmark by defining few-shot splits for three public IC/SF datasets, ATIS, TOP, and Snips. We show that two popular few-shot learning algorithms, model agnostic meta learning (MAML) and prototypical networks, outperform a fine-tuning baseline on this benchmark. Prototypical networks achieves significant gains in IC performance on the ATIS and TOP datasets, while both prototypical networks and MAML outperform the baseline with respect to SF on all three datasets. In addition, we demonstrate that joint training as well as the use of pre-trained language models, ELMo and BERT in our case, are complementary to these few-shot learning methods and yield further gains.
['Jason Krone', 'Mona Diab', 'Yi Zhang']
2020-04-22
learning-to-classify-intents-and-slot-labels-1
https://aclanthology.org/2020.nlp4convai-1.12
https://aclanthology.org/2020.nlp4convai-1.12.pdf
ws-2020-7
['goal-oriented-dialogue-systems']
['natural-language-processing']
[ 1.22677952e-01 3.19628298e-01 -6.18696332e-01 -3.24673712e-01 -6.92557216e-01 -2.50920206e-01 1.02037597e+00 8.02980810e-02 -6.75250649e-01 8.28273535e-01 6.00592077e-01 -7.51559213e-02 -1.62639856e-01 -6.95378184e-01 -6.95007518e-02 -1.86792105e-01 8.91395733e-02 9.72976863e-01 5.97807527e-01 -8.89394164e-01 2.91784972e-01 -1.77713916e-01 -1.44336176e+00 6.05892479e-01 8.23482156e-01 8.40505719e-01 3.60084437e-02 9.00799692e-01 -3.76390606e-01 1.30037034e+00 -5.29558539e-01 -3.64539653e-01 -1.92719158e-02 -3.58511478e-01 -1.06530643e+00 1.58282705e-02 1.82850450e-01 -5.13689041e-01 -4.69373047e-01 5.56453586e-01 7.24245369e-01 1.08620989e+00 7.12766171e-01 -1.24307919e+00 -2.22576752e-01 7.55250573e-01 -1.98927462e-01 3.54345173e-01 4.12179381e-01 3.80485117e-01 1.21823299e+00 -9.01733577e-01 7.80381978e-01 1.34298837e+00 7.76077867e-01 1.03309405e+00 -1.15327549e+00 -2.33638942e-01 -2.36025006e-02 5.09108961e-01 -7.06383288e-01 -9.17290628e-01 5.87502718e-01 -1.66484237e-01 1.55967033e+00 1.19371884e-01 2.76135027e-01 1.51125169e+00 -1.03448726e-01 1.34691882e+00 7.52465546e-01 -8.30314815e-01 9.35571551e-01 1.42134994e-01 9.62830663e-01 5.97631633e-01 -2.76607692e-01 -9.68932956e-02 -8.89908552e-01 -4.57919002e-01 1.43201604e-01 1.40553653e-01 -5.28738573e-02 -2.81651765e-01 -6.38656199e-01 1.33369827e+00 -1.85254887e-02 4.82782662e-01 -1.32819712e-01 -1.83716089e-01 8.72171521e-01 3.97384346e-01 6.58722162e-01 6.71804726e-01 -4.49846923e-01 -7.04073727e-01 -7.25747824e-01 3.51568043e-01 1.31744564e+00 9.57279742e-01 5.47719777e-01 8.62507820e-02 -9.57289398e-01 1.36077666e+00 -2.40146652e-01 -2.65836746e-01 9.98472393e-01 -1.05514228e+00 4.94275242e-01 4.54855859e-01 7.74627849e-02 -2.68865079e-01 -7.87462354e-01 -9.72282439e-02 -4.78987217e-01 -2.26539776e-01 3.32801253e-01 -2.65027791e-01 -1.12984562e+00 1.65036893e+00 2.55018920e-01 2.41401553e-01 2.94785649e-01 5.54181337e-01 1.12320065e+00 5.81505537e-01 2.71376789e-01 -3.17489773e-01 1.35386968e+00 -1.40349054e+00 -9.68557715e-01 -8.71487081e-01 1.08092070e+00 -1.62226200e-01 1.44633305e+00 7.50490054e-02 -7.97706187e-01 -3.26565027e-01 -1.04537940e+00 -9.11580175e-02 -4.98876750e-01 -5.98781466e-01 8.96552324e-01 6.27411306e-01 -7.28164077e-01 8.91038120e-01 -6.58081293e-01 -7.47459471e-01 3.80389035e-01 2.96900086e-02 -1.03039868e-01 -1.65849939e-01 -1.62818980e+00 1.08145559e+00 6.69335783e-01 -8.84525180e-01 -1.07183826e+00 -9.82448995e-01 -1.20372474e+00 3.73774052e-01 8.82613003e-01 -3.86090189e-01 1.98837137e+00 -2.24745542e-01 -1.68514299e+00 6.85612082e-01 4.22797985e-02 -1.01224267e+00 2.82872200e-01 -1.37976214e-01 -2.77145326e-01 1.30308807e-01 1.99187666e-01 5.03175855e-01 4.89189237e-01 -7.51428485e-01 -7.99963295e-01 -1.25982150e-01 2.69507438e-01 2.54266620e-01 -6.29389048e-01 2.93303821e-02 -2.23281264e-01 -1.95482939e-01 -4.83061969e-01 -5.76629639e-01 -3.14127296e-01 -3.63663912e-01 -3.07566553e-01 -6.27696753e-01 7.81782746e-01 -2.94237524e-01 1.38877976e+00 -1.91556168e+00 -6.26813993e-02 -5.33500016e-01 1.06650800e-01 6.49312675e-01 -1.83373317e-01 6.95090890e-01 2.14924872e-01 -2.33476996e-01 -1.88350782e-01 -5.07190287e-01 1.91488400e-01 5.17503798e-01 -8.16574395e-02 4.81658280e-02 -6.04091957e-02 1.02466500e+00 -1.17655897e+00 -5.74958384e-01 6.58004522e-01 -9.81243625e-02 -7.42356479e-01 3.06025922e-01 -6.56386256e-01 -1.53934464e-01 -4.45802398e-02 5.28162956e-01 4.96200882e-02 -3.38935971e-01 1.68972716e-01 1.77324280e-01 2.49051467e-01 5.10474563e-01 -6.07320130e-01 2.02489257e+00 -6.60427868e-01 3.50084573e-01 -2.82135785e-01 -9.30711567e-01 5.89346945e-01 5.47707856e-01 1.54765204e-01 -8.79920661e-01 1.62497073e-01 -6.64307252e-02 -3.06888688e-02 -2.89745450e-01 7.36081362e-01 -4.86881196e-01 -4.72221345e-01 7.56092668e-01 9.77555513e-01 1.81347370e-01 7.46000886e-01 6.20671332e-01 1.33758438e+00 -2.86723047e-01 7.66137838e-01 -3.76822054e-02 8.42468217e-02 1.45001724e-01 6.84421003e-01 1.32876241e+00 -6.36941254e-01 5.04378021e-01 6.33170187e-01 -4.95193213e-01 -9.70303953e-01 -7.41437316e-01 2.26593558e-02 2.14898491e+00 -4.23303358e-02 -7.85242379e-01 -6.58458710e-01 -9.19299722e-01 -1.08459741e-01 1.50602472e+00 -7.47807860e-01 -6.14755630e-01 -2.55986780e-01 -9.78888035e-01 5.44767678e-01 4.79463905e-01 5.50065815e-01 -1.23259604e+00 -7.33827949e-01 3.98515463e-01 -2.22166970e-01 -1.12918675e+00 -3.36012363e-01 6.39002025e-01 -6.10623837e-01 -9.37084913e-01 -6.35199308e-01 -2.46112645e-01 -2.63596267e-01 3.73697549e-01 1.23338866e+00 -1.10710740e-01 -3.95889729e-01 2.77812064e-01 -7.49444664e-01 -3.18884254e-01 -4.03462917e-01 2.73926735e-01 2.94455409e-01 -2.50307560e-01 6.61677957e-01 -5.98877072e-01 -1.87294886e-01 1.30658373e-01 -5.27132869e-01 1.46744475e-01 9.93195400e-02 1.31162596e+00 -2.85328090e-01 -3.75905186e-01 6.94357395e-01 -1.36126530e+00 9.51891541e-01 -6.68676078e-01 1.82803810e-01 2.69954145e-01 -7.76722014e-01 5.12784421e-02 5.28214335e-01 -4.06804681e-01 -1.32705545e+00 -5.15285552e-01 -1.84404641e-01 -4.61065143e-01 -1.86452284e-01 4.16455567e-01 3.60356458e-02 2.21756384e-01 1.24349487e+00 6.93544596e-02 -6.30276417e-03 -6.50340855e-01 4.26180214e-01 8.80877018e-01 3.82074714e-01 -4.07968819e-01 1.06318362e-01 1.77587390e-01 -5.12015343e-01 -1.08836067e+00 -1.35971367e+00 -9.77593303e-01 -4.86041158e-01 -5.41124754e-02 6.11042619e-01 -7.45770693e-01 -3.57625097e-01 1.80709288e-01 -1.01548874e+00 -7.05830216e-01 -7.61185288e-01 1.44905642e-01 -9.65362787e-01 3.21453631e-01 -9.44845080e-01 -1.08549678e+00 -6.67828321e-01 -6.51023686e-01 8.72870684e-01 3.27281535e-01 -6.15360260e-01 -1.22368002e+00 3.31234485e-01 4.63316083e-01 4.67768937e-01 -3.59212607e-01 1.02237499e+00 -1.64540577e+00 3.57904375e-01 -2.73322314e-01 -5.76503284e-04 -1.69595182e-02 -4.78163362e-02 -6.35307550e-01 -1.18072462e+00 -2.49467388e-01 5.45005426e-02 -1.02446544e+00 1.10844660e+00 2.92944580e-01 4.60014373e-01 -2.60602325e-01 -8.65609720e-02 2.10607395e-01 1.04640651e+00 4.53221142e-01 4.96094614e-01 3.96386564e-01 2.98952252e-01 4.75235224e-01 8.54751289e-01 8.62272978e-01 2.89382726e-01 8.76963556e-01 4.50014956e-02 2.76526064e-01 -6.92463890e-02 -2.90408492e-01 2.02882722e-01 3.94129753e-01 4.73537371e-02 -2.57644743e-01 -1.07591999e+00 5.21299303e-01 -2.34868312e+00 -1.19109535e+00 5.44326901e-01 2.02530813e+00 8.65194440e-01 3.78035188e-01 4.24904823e-01 1.41412497e-01 5.83043754e-01 6.08475447e-01 -6.01750910e-01 -6.45394146e-01 1.88013881e-01 3.70828182e-01 8.47808123e-02 5.09369195e-01 -1.25072503e+00 1.33014882e+00 6.23692560e+00 1.24675441e+00 -3.99874359e-01 6.78023160e-01 6.39444351e-01 -4.91331637e-01 1.92907646e-01 -1.45011514e-01 -1.01861608e+00 2.68002898e-01 1.43970251e+00 -3.97693366e-01 2.42880002e-01 1.35205758e+00 -2.04200387e-01 -7.59400055e-02 -1.11104417e+00 7.60660350e-01 3.03814113e-01 -1.54609072e+00 -7.86185563e-02 -2.91789085e-01 6.21980250e-01 1.16273999e-01 -4.92008746e-01 1.40158772e+00 8.06885898e-01 -7.94009089e-01 1.30219772e-01 2.38998130e-01 5.19759476e-01 -6.91975296e-01 7.38366604e-01 7.53681779e-01 -5.12596786e-01 -3.82740200e-01 -7.18417704e-01 -3.20811361e-01 3.13503027e-01 1.06295153e-01 -8.73444438e-01 3.31864446e-01 4.87480372e-01 6.13184094e-01 -4.37901139e-01 8.83176744e-01 2.47291923e-02 5.46023965e-01 -4.82027791e-02 -2.41876662e-01 6.30577803e-01 3.00489366e-01 5.69400549e-01 1.17734516e+00 -3.49729121e-01 1.35562703e-01 3.87321442e-01 5.34179509e-01 -1.27523616e-01 2.02428550e-02 -5.97480893e-01 -2.26400554e-01 4.72144186e-01 1.27481866e+00 -4.45713431e-01 -7.95026660e-01 -4.86893088e-01 8.28486741e-01 6.14562750e-01 -3.51893380e-02 -4.32135582e-01 -4.58602458e-01 6.18143559e-01 -1.40462413e-01 7.51750767e-02 4.01826173e-01 -9.99384448e-02 -1.39043140e+00 -5.70185900e-01 -8.75592291e-01 9.56625044e-01 -3.44334126e-01 -1.24342120e+00 4.27027255e-01 1.32342413e-01 -1.02119339e+00 -8.51860106e-01 -4.53495115e-01 -1.10440338e+00 4.44723755e-01 -1.12969792e+00 -1.02816010e+00 -1.52093530e-01 4.48049128e-01 1.37548912e+00 -5.43279350e-01 1.18756413e+00 -2.39189100e-02 -7.47689188e-01 6.09876096e-01 1.36619315e-01 7.23612681e-02 7.04824626e-01 -1.26352191e+00 5.76177776e-01 4.55259174e-01 1.14647456e-01 4.06267911e-01 8.60321581e-01 -5.31210065e-01 -1.03545785e+00 -9.32865679e-01 7.30443954e-01 -3.30658883e-01 6.24136627e-01 -6.16672516e-01 -8.96188498e-01 7.59815276e-01 2.02917695e-01 -4.44015190e-02 8.84416878e-01 7.31504500e-01 -4.29675370e-01 5.20824850e-01 -1.25930607e+00 6.59864783e-01 1.08355868e+00 -5.90276182e-01 -1.20179379e+00 8.10024142e-01 1.08325112e+00 -1.45820886e-01 -6.70268595e-01 4.03135598e-01 4.54832703e-01 -1.15471160e+00 1.00343621e+00 -1.42277336e+00 5.83491743e-01 7.69535899e-01 -2.79390991e-01 -1.41762793e+00 -3.38846415e-01 -6.77989364e-01 -6.97714925e-01 1.07827497e+00 2.87607223e-01 -2.04424486e-01 1.01151037e+00 8.50948095e-01 -3.54948223e-01 -6.08573318e-01 -1.09091187e+00 -1.00619531e+00 7.58416504e-02 -4.53894556e-01 2.82390356e-01 8.75190139e-01 7.52399564e-01 8.82533908e-01 -8.56914997e-01 -8.27098489e-01 6.27479672e-01 1.59108657e-02 7.97735393e-01 -1.27551687e+00 -4.32358325e-01 -2.43050173e-01 -2.36009255e-01 -5.33564568e-01 2.92087734e-01 -7.53981590e-01 4.65997271e-02 -1.47447658e+00 3.49673152e-01 -6.79440871e-02 -2.34829143e-01 6.82903111e-01 -3.28376800e-01 2.06630658e-02 2.44557157e-01 3.74965928e-02 -1.25297296e+00 8.97254407e-01 7.30039239e-01 -2.10252106e-01 -4.68870759e-01 6.09757006e-02 -5.97810686e-01 8.12121630e-01 7.58700013e-01 -1.78183421e-01 -4.64148402e-01 3.83521289e-01 -4.41069990e-01 2.86058217e-01 -7.29322955e-02 -1.20152295e+00 4.83074546e-01 -2.40740135e-01 -3.37291919e-02 -2.79148132e-01 7.31506884e-01 -6.98077157e-02 -5.91325104e-01 4.66798633e-01 -6.58224463e-01 -7.72465825e-01 5.47460951e-02 6.22888267e-01 6.06614463e-02 -7.76218832e-01 9.52953994e-01 -5.92938840e-01 -1.28459620e+00 2.29676723e-01 -3.85622740e-01 6.90238774e-01 9.41535652e-01 -9.60302502e-02 -7.45473981e-01 -5.73674679e-01 -8.26125920e-01 4.02498126e-01 9.15754288e-02 6.01736963e-01 5.02278626e-01 -1.13280463e+00 -3.30247194e-01 3.49127725e-02 6.96182251e-01 -3.86989743e-01 5.90918124e-01 8.01846504e-01 -6.50454536e-02 7.85330117e-01 -2.42537498e-01 -2.11384386e-01 -1.06122243e+00 6.53338134e-01 2.74250537e-01 -7.03102589e-01 -7.86191463e-01 8.17419767e-01 5.08326925e-02 -1.00714910e+00 5.16987205e-01 4.23446000e-01 -3.42406362e-01 3.44478548e-01 9.23890412e-01 6.71775341e-01 -7.46279359e-02 -2.10402146e-01 -1.93709470e-02 -3.70434672e-01 -6.71947241e-01 -2.55426556e-01 1.36900437e+00 1.05467178e-01 5.12329698e-01 9.17740703e-01 8.45742822e-01 -9.13561404e-01 -1.04353774e+00 -6.85842216e-01 3.15627187e-01 -3.48742485e-01 9.72305089e-02 -8.55567873e-01 -3.12916249e-01 1.01171470e+00 3.26625317e-01 1.84632897e-01 4.72072631e-01 -3.23447026e-03 8.10481787e-01 6.92739606e-01 6.30319118e-01 -1.64538133e+00 3.79820138e-01 1.03644872e+00 4.55347478e-01 -1.48231220e+00 -1.90899894e-01 -1.05126984e-01 -1.03193021e+00 7.61468649e-01 8.80303800e-01 6.08546995e-02 3.47110659e-01 2.93852929e-02 -2.72662014e-01 -2.44731426e-01 -1.47521889e+00 -4.92500961e-01 -2.10327655e-02 5.08895218e-01 2.63455957e-01 -2.02601716e-01 -1.50148839e-01 9.60581660e-01 4.87976037e-02 1.32144481e-01 4.53269809e-01 1.33681154e+00 -9.04649198e-01 -8.24727714e-01 1.28210977e-01 1.01928413e+00 -2.36348808e-01 -2.93625087e-01 -3.59144062e-01 6.11981630e-01 -5.23889244e-01 1.20836699e+00 8.57206658e-02 -6.15971684e-01 2.66205221e-01 8.08075607e-01 1.60219669e-01 -1.21708059e+00 -7.10484743e-01 -3.13601911e-01 7.19502866e-01 -6.60272598e-01 7.38746747e-02 -2.44546816e-01 -9.01013553e-01 -4.49421763e-01 -4.82752204e-01 3.77355516e-01 2.24718869e-01 1.33202386e+00 2.06655577e-01 5.55317163e-01 3.90987873e-01 -9.37367201e-01 -1.02691627e+00 -1.63113832e+00 -8.30678344e-01 4.93182391e-01 -4.33591753e-02 -1.08279908e+00 -6.68418944e-01 -7.61415005e-01]
[11.977683067321777, 7.682446479797363]
3479e427-0ed3-4fb6-9563-57497664fc8c
just-a-glimpse-rethinking-temporal
2305.18418
null
https://arxiv.org/abs/2305.18418v2
https://arxiv.org/pdf/2305.18418v2.pdf
Just a Glimpse: Rethinking Temporal Information for Video Continual Learning
Class-incremental learning is one of the most important settings for the study of Continual Learning, as it closely resembles real-world application scenarios. With constrained memory sizes, catastrophic forgetting arises as the number of classes/tasks increases. Studying continual learning in the video domain poses even more challenges, as video data contains a large number of frames, which places a higher burden on the replay memory. The current common practice is to sub-sample frames from the video stream and store them in the replay memory. In this paper, we propose SMILE a novel replay mechanism for effective video continual learning based on individual/single frames. Through extensive experimentation, we show that under extreme memory constraints, video diversity plays a more significant role than temporal information. Therefore, our method focuses on learning from a small number of frames that represent a large number of unique videos. On three representative video datasets, Kinetics, UCF101, and ActivityNet, the proposed method achieves state-of-the-art performance, outperforming the previous state-of-the-art by up to 21.49%.
['Bernard Ghanem', 'Chen Zhao', 'Merey Ramazanova', 'Juan Leon Alcazar', 'Lama Alssum']
2023-05-28
null
null
null
null
['class-incremental-learning', 'incremental-learning']
['computer-vision', 'methodology']
[ 7.66217187e-02 -7.16548145e-01 -5.24404168e-01 5.02133705e-02 -4.90173608e-01 -2.42084652e-01 4.73519921e-01 1.65108442e-01 -7.70003498e-01 9.36111629e-01 -1.88419838e-02 -1.77780733e-01 -1.21485889e-02 -3.18122685e-01 -1.15767312e+00 -8.30305398e-01 -5.35190880e-01 6.35913163e-02 7.23972917e-01 1.06187105e-01 2.20186695e-01 1.81141093e-01 -1.94241619e+00 4.71736938e-01 4.76199627e-01 1.07628214e+00 4.68739092e-01 6.69461966e-01 2.34140102e-02 1.17498064e+00 -5.34275234e-01 -1.08347321e-02 1.29825965e-01 -4.45963353e-01 -3.76295030e-01 4.04357642e-01 6.49902344e-01 -6.09940946e-01 -8.32803249e-01 7.91709304e-01 5.38654566e-01 5.14039457e-01 2.59464443e-01 -1.38769615e+00 -3.30189466e-01 4.72702116e-01 -7.39580631e-01 8.57342124e-01 4.13368255e-01 3.29592615e-01 5.26575804e-01 -9.42625761e-01 6.68849826e-01 1.04694068e+00 4.52191830e-01 5.65211236e-01 -8.27524960e-01 -8.14431906e-01 5.33912838e-01 7.48863399e-01 -1.31691515e+00 -5.56455970e-01 6.13416612e-01 -2.93302864e-01 8.66907716e-01 3.20880599e-02 8.04349184e-01 1.29828250e+00 3.42154682e-01 1.07552695e+00 7.77991354e-01 -3.26865494e-01 4.97958958e-01 -1.74658820e-01 1.34641603e-01 6.15854502e-01 3.13846260e-01 -7.52633661e-02 -1.16652668e+00 -9.64640081e-02 6.72295451e-01 6.32334411e-01 -5.86850584e-01 -3.83595049e-01 -1.12840748e+00 5.37397027e-01 -8.41768757e-02 1.40645400e-01 -3.58794153e-01 1.45214781e-01 6.40038967e-01 6.78087950e-01 3.32464129e-01 -1.15975350e-01 -3.26641351e-01 -6.71599805e-01 -1.05800056e+00 1.83802709e-01 5.62640965e-01 9.76829290e-01 6.63352787e-01 5.44179603e-02 -1.61925763e-01 6.94616675e-01 -2.62895942e-01 2.64177322e-01 9.14992988e-01 -9.74860966e-01 4.48205441e-01 2.59914607e-01 1.74660206e-01 -8.73595893e-01 -1.03477955e-01 -3.80114138e-01 -8.47509205e-01 -1.98849663e-02 5.44099748e-01 5.92615753e-02 -7.69589007e-01 1.70997834e+00 2.65347004e-01 8.95652831e-01 -2.23383754e-01 7.73959219e-01 2.36674204e-01 7.35003650e-01 3.14193927e-02 -9.74374175e-01 1.00273633e+00 -9.91078019e-01 -7.20469832e-01 -1.03753746e-01 2.94680119e-01 -5.52648544e-01 1.40563345e+00 6.77752495e-01 -1.04866922e+00 -6.64985955e-01 -1.12042749e+00 3.77684355e-01 7.20161945e-02 -2.33747065e-01 5.26434183e-01 2.77589709e-01 -7.48646259e-01 5.66316307e-01 -1.16492915e+00 -2.82297611e-01 4.57405061e-01 1.61378048e-02 -1.89368442e-01 -4.91571009e-01 -9.38137889e-01 3.17603678e-01 4.94694948e-01 -3.57322723e-01 -1.33082998e+00 -7.73990452e-01 -4.05907393e-01 -3.68260145e-02 9.33134019e-01 -3.73483688e-01 1.26188695e+00 -1.05662227e+00 -1.29574025e+00 2.98294485e-01 -3.25590074e-01 -7.02468514e-01 9.12879884e-01 -7.00148106e-01 -3.59798640e-01 3.79464030e-01 -2.60053843e-01 3.19276392e-01 1.22028196e+00 -9.09834683e-01 -7.82085776e-01 -2.34492660e-01 2.22943693e-01 2.48746037e-01 -9.72922325e-01 -3.77581447e-01 -6.57324374e-01 -6.86278462e-01 -2.74585038e-01 -1.08907187e+00 3.83719325e-01 -6.41069785e-02 1.65819570e-01 -3.10108364e-01 1.13785934e+00 -2.78156996e-01 1.67472517e+00 -2.32686710e+00 1.51231036e-01 -3.79781961e-01 3.48612726e-01 4.07905966e-01 6.18466288e-02 3.16599429e-01 1.09068908e-01 -2.09711611e-01 1.45157548e-02 -3.91833365e-01 -4.88451391e-01 1.64125800e-01 -3.71457577e-01 4.65759844e-01 -3.23130727e-01 5.03601432e-01 -1.03136802e+00 -4.32175398e-01 -7.07404539e-02 2.61809021e-01 -4.79483962e-01 2.45801076e-01 -3.50316226e-01 3.19959760e-01 -1.47969350e-01 5.25976539e-01 4.97654498e-01 -4.78392541e-01 1.71874896e-01 2.11044714e-01 2.81214565e-02 -2.03897491e-01 -1.08824670e+00 1.91217232e+00 -3.13171864e-01 7.44555116e-01 -5.06460011e-01 -7.91285038e-01 4.56664443e-01 2.86441118e-01 5.09167135e-01 -8.48624289e-01 -2.80852079e-01 7.37954676e-02 -1.70090005e-01 -6.90712750e-01 5.07887721e-01 6.87080696e-02 2.13494539e-01 6.27006412e-01 1.18342750e-01 5.96398711e-01 5.88510334e-01 3.09102565e-01 1.41692924e+00 -1.06690332e-01 2.31176645e-01 1.38406590e-01 3.41957867e-01 -3.46309125e-01 8.07654977e-01 1.01113749e+00 -3.60638618e-01 4.18497711e-01 4.61114168e-01 -7.38507032e-01 -9.41668272e-01 -9.96764243e-01 1.82995379e-01 1.38510954e+00 1.76651523e-01 -5.01608789e-01 -5.96579731e-01 -4.72490698e-01 -5.23909852e-02 4.96166587e-01 -4.79409993e-01 -5.57436705e-01 -8.61979961e-01 -7.30061293e-01 2.37192929e-01 4.30293441e-01 6.58406496e-01 -1.12098527e+00 -9.67696071e-01 2.83483058e-01 -1.96157470e-01 -1.11399758e+00 -6.87014818e-01 -2.10886806e-01 -1.08583105e+00 -1.21531117e+00 -8.46949756e-01 -6.31461442e-01 2.55782664e-01 9.10495996e-01 1.15368652e+00 2.35559970e-01 -2.70325422e-01 5.91836214e-01 -4.20046151e-01 -1.02876216e-01 -1.75459459e-01 2.38637298e-01 3.91208678e-01 1.68881223e-01 2.26050258e-01 -6.79531813e-01 -6.89656436e-01 2.49262154e-01 -1.20860374e+00 2.43858714e-02 5.31167328e-01 9.70248759e-01 7.46880829e-01 3.97416413e-01 7.87726283e-01 -9.20280218e-01 4.36938286e-01 -7.76379943e-01 -3.10995638e-01 3.48620117e-01 -5.32582283e-01 -3.73653293e-01 9.12955999e-01 -1.12424469e+00 -9.66337025e-01 -8.97186249e-02 4.24558014e-01 -9.58785474e-01 1.37457982e-01 4.19182837e-01 2.55000517e-02 2.30554372e-01 5.44740081e-01 8.16483021e-01 -8.66532102e-02 -2.83339322e-01 5.65751866e-02 3.19716811e-01 5.63145816e-01 -4.33915198e-01 3.39681625e-01 3.82451564e-01 -1.65326655e-01 -1.12417448e+00 -6.28377557e-01 -4.45763826e-01 -5.02750337e-01 -4.66154724e-01 2.13126615e-01 -1.10658979e+00 -7.57177413e-01 8.85671794e-01 -7.23569036e-01 -6.08351886e-01 -1.93379387e-01 4.87517148e-01 -6.46247566e-01 4.83514994e-01 -7.85824239e-01 -6.91655159e-01 -9.68792364e-02 -1.06605065e+00 4.77968991e-01 2.98139602e-01 1.89445123e-01 -7.11174667e-01 -3.17983404e-02 8.96833837e-02 3.80297661e-01 2.32007951e-02 7.97596931e-01 -5.03666461e-01 -9.23692107e-01 -8.61060619e-02 2.51550525e-01 2.25261331e-01 1.23000808e-01 -3.38719189e-01 -7.15684950e-01 -8.87752831e-01 1.27145842e-01 -5.66093206e-01 1.12807357e+00 1.27962217e-01 1.38410056e+00 -3.45555484e-01 -7.20176026e-02 3.93476218e-01 1.35214269e+00 4.14190799e-01 7.00712562e-01 2.82746792e-01 5.64351022e-01 -1.58283226e-02 8.23945284e-01 8.17012250e-01 2.95537442e-01 3.56884480e-01 3.68294418e-01 5.51273644e-01 -1.97704285e-01 -3.52556467e-01 4.98018116e-01 9.74862516e-01 -9.29804295e-02 -4.50289637e-01 -6.48134410e-01 4.29928660e-01 -2.13432813e+00 -1.24840713e+00 5.26963234e-01 2.71580839e+00 9.24389839e-01 4.31526214e-01 3.53015274e-01 2.61176944e-01 6.77409947e-01 3.21143568e-01 -1.02474129e+00 5.14668286e-01 -1.53828427e-01 -2.61820614e-01 2.94773638e-01 1.03385732e-01 -1.09203601e+00 8.32098126e-01 5.96894979e+00 1.11661780e+00 -1.26078069e+00 2.42324054e-01 8.39862466e-01 -7.83398688e-01 2.63702780e-01 -2.20022842e-01 -8.69322360e-01 1.00203037e+00 1.16558814e+00 -2.84539849e-01 6.19043469e-01 8.03004265e-01 2.01683879e-01 -4.28757787e-01 -1.12149596e+00 1.29494143e+00 3.52445334e-01 -1.38457394e+00 2.74213791e-01 -1.58321977e-01 8.12480748e-01 -5.59166297e-02 2.43980274e-01 4.25555497e-01 -2.48021483e-01 -6.37496591e-01 7.26018727e-01 6.28455162e-01 8.93666327e-01 -7.82061875e-01 3.10794145e-01 6.67926610e-01 -1.07765341e+00 -5.32637179e-01 -7.16342628e-01 -1.19817048e-01 1.27756581e-01 7.48531222e-01 -2.88020879e-01 2.32798979e-02 9.97409284e-01 7.51698434e-01 -6.37289703e-01 1.25729585e+00 2.19273567e-01 9.59261417e-01 -1.98753640e-01 4.78801057e-02 -9.45785865e-02 2.87698746e-01 4.67416286e-01 1.15111351e+00 4.15179342e-01 6.24349480e-03 3.80104154e-01 -5.28992563e-02 -4.27655697e-01 5.01160696e-02 -5.25109947e-01 3.58618097e-03 7.79036999e-01 7.66555846e-01 -7.81029701e-01 -7.13735223e-01 -6.62036240e-01 1.02416861e+00 4.90686506e-01 3.60760808e-01 -9.77277994e-01 -1.24017768e-01 5.63280225e-01 3.65041643e-01 4.04796690e-01 -4.82192278e-01 4.81717110e-01 -1.40414155e+00 2.14119360e-01 -1.01577258e+00 5.66013575e-01 -4.25759614e-01 -1.01776206e+00 4.34125006e-01 2.01064236e-02 -1.35226333e+00 -3.02512676e-01 -1.52457863e-01 -4.62752968e-01 7.53301848e-03 -1.49136722e+00 -6.81724429e-01 -7.35282958e-01 8.72925878e-01 1.16171575e+00 -3.90523076e-01 3.18915755e-01 6.63595855e-01 -5.56954384e-01 7.70837903e-01 2.76398987e-01 -3.18229645e-01 8.82285535e-01 -7.60844767e-01 8.65702108e-02 7.89602041e-01 1.89595297e-01 6.12271786e-01 7.53622353e-01 -6.82709277e-01 -1.86463618e+00 -1.09594846e+00 3.93637031e-01 -9.79084000e-02 5.80190837e-01 -3.06851298e-01 -1.21643865e+00 7.17284560e-01 5.61284414e-03 3.12437594e-01 5.65683186e-01 -3.98289114e-01 -2.78381199e-01 -2.76719481e-01 -7.01815724e-01 5.13167441e-01 1.45995200e+00 -4.59786773e-01 -3.44158828e-01 3.63347173e-01 9.09602225e-01 -4.05457079e-01 -4.76961166e-01 3.72095108e-02 6.52451158e-01 -1.19708848e+00 8.67421865e-01 -5.97518325e-01 3.06951255e-01 -1.66398078e-01 -1.51428446e-01 -1.17733455e+00 -1.80548161e-01 -8.83166254e-01 -9.58538234e-01 1.04793811e+00 -1.18736982e-01 -3.38338763e-01 8.39036882e-01 2.11276516e-01 1.67485923e-01 -7.32411444e-01 -1.03707504e+00 -1.06706250e+00 -2.14257717e-01 -3.40352565e-01 3.77126247e-01 6.68855548e-01 -1.21147014e-01 1.34165436e-01 -9.50704575e-01 -2.44485065e-01 6.03553176e-01 -3.95803750e-02 7.44113207e-01 -8.81777763e-01 -5.83699584e-01 4.01742086e-02 -3.65418911e-01 -1.41799664e+00 1.26514152e-01 -4.02177721e-01 -1.62113458e-01 -6.63941979e-01 5.25191844e-01 -8.66405368e-02 -6.71899676e-01 2.19992340e-01 -2.85969496e-01 1.22799836e-01 3.54811460e-01 5.70463181e-01 -1.35502589e+00 7.20756114e-01 1.08937395e+00 -9.82805565e-02 -3.84708196e-01 1.39374778e-01 -2.78416574e-01 6.12219810e-01 6.54868722e-01 -4.22094136e-01 -8.55019510e-01 -3.50722313e-01 7.17036799e-02 2.80566692e-01 8.19990188e-02 -1.41105068e+00 6.45968676e-01 -2.30153576e-01 3.34014356e-01 -4.86576051e-01 3.13632637e-01 -7.35652626e-01 1.53123274e-01 5.60247064e-01 -4.26861703e-01 4.26225007e-01 1.56562299e-01 1.15650022e+00 -1.28273532e-01 -2.40311660e-02 7.27561772e-01 -3.33182573e-01 -9.32878971e-01 5.99609256e-01 -4.62176859e-01 3.82668465e-01 1.15255308e+00 -1.16308883e-01 -4.53961551e-01 -5.02442539e-01 -6.67812765e-01 1.21833988e-01 5.96595526e-01 6.19340301e-01 7.98231483e-01 -1.30187321e+00 -3.92765909e-01 9.92223024e-02 4.86131310e-02 -1.88647047e-01 7.60715485e-01 7.89974153e-01 -2.45369777e-01 2.30225116e-01 -2.08519265e-01 -7.12673187e-01 -1.32142818e+00 8.82246554e-01 -6.80360496e-02 -1.33596687e-02 -8.55351269e-01 5.55265903e-01 1.56572938e-01 5.85722506e-01 6.12398565e-01 1.58477724e-02 6.92165345e-02 2.35933140e-01 1.08911026e+00 8.38367641e-01 -5.56077324e-02 -2.78845847e-01 -8.93582925e-02 2.03822002e-01 -4.49539006e-01 1.11197464e-01 1.38958144e+00 -3.44266683e-01 3.17055732e-01 9.45687175e-01 1.11978197e+00 -3.69326234e-01 -2.04543495e+00 -6.14503264e-01 -8.06919262e-02 -8.42270076e-01 -1.54024348e-01 -3.70573729e-01 -1.08263493e+00 7.31119812e-01 7.26265192e-01 -8.88142735e-02 1.23597956e+00 -2.23146796e-01 1.09988391e+00 6.68334186e-01 7.74548113e-01 -1.29318666e+00 8.45652878e-01 4.58020359e-01 6.19091928e-01 -1.20178735e+00 7.40280151e-02 8.82084817e-02 -6.12814546e-01 1.05788279e+00 7.17349708e-01 -4.31034118e-02 6.94443285e-01 1.34863958e-01 -2.80502617e-01 2.59394526e-01 -1.20672309e+00 3.78277451e-01 -1.90290660e-01 2.55192012e-01 2.47938827e-01 -1.19687796e-01 -2.07872152e-01 4.50656116e-01 2.60400593e-01 4.48764324e-01 7.82890439e-01 1.24607229e+00 -7.38587797e-01 -6.76987827e-01 -1.52727708e-01 5.20663679e-01 -5.51133692e-01 5.00837713e-02 3.10085386e-01 6.48825288e-01 -1.19816519e-01 7.02875972e-01 1.65952250e-01 -4.23840344e-01 2.99614370e-02 2.59611398e-01 5.55779874e-01 -3.19099575e-01 -2.84799132e-02 1.12148762e-01 -5.46720326e-01 -6.31195366e-01 -4.01761562e-01 -8.92719209e-01 -1.14279938e+00 -7.12641239e-01 -1.01238213e-01 -1.41112328e-01 2.70528048e-01 7.30217755e-01 5.02872109e-01 3.93175513e-01 5.88376462e-01 -8.38057637e-01 -6.37529850e-01 -8.49184453e-01 -7.23218679e-01 5.22363245e-01 5.06625533e-01 -9.38405633e-01 -5.07338524e-01 3.21606576e-01]
[8.605113983154297, 0.7086646556854248]
19136360-54bc-4192-a958-6afdd20e59e0
deep-learned-svt-unrolling-singular-value
2105.06934
null
https://arxiv.org/abs/2105.06934v1
https://arxiv.org/pdf/2105.06934v1.pdf
Deep learned SVT: Unrolling singular value thresholding to obtain better MSE
Affine rank minimization problem is the generalized version of low rank matrix completion problem where linear combinations of the entries of a low rank matrix are observed and the matrix is estimated from these measurements. We propose a trainable deep neural network by unrolling a popular iterative algorithm called the singular value thresholding (SVT) algorithm to perform this generalized matrix completion which we call Learned SVT (LSVT). We show that our proposed LSVT with fixed layers (say T) reconstructs the matrix with lesser mean squared error (MSE) compared with that incurred by SVT with fixed (same T) number of iterations and our method is much more robust to the parameters which need to be carefully chosen in SVT algorithm.
['Sheetal Kalyani', 'Siva Shanmugam']
2021-05-14
null
null
null
null
['low-rank-matrix-completion']
['methodology']
[ 3.40795308e-01 1.32724226e-01 2.75857717e-01 -4.25073504e-01 -9.00186956e-01 -7.20228910e-01 4.81788844e-01 -4.07018304e-01 -6.05123758e-01 6.93624616e-01 4.24500525e-01 -1.87967643e-01 -6.48294091e-01 -2.27565587e-01 -1.24508154e+00 -7.42448092e-01 -2.67091930e-01 6.22634828e-01 -3.94358367e-01 -1.38615757e-01 -1.58550814e-02 3.27767462e-01 -7.77464628e-01 1.71845198e-01 5.76393187e-01 8.10343027e-01 4.58933413e-02 8.93001974e-01 4.72553313e-01 6.99010611e-01 -1.90750793e-01 -1.44410700e-01 1.05379868e+00 -8.09767097e-02 -5.32804191e-01 2.50156701e-01 9.39359546e-01 -4.55601335e-01 -8.33872974e-01 1.24509823e+00 3.12134594e-01 3.92065585e-01 6.71880186e-01 -8.75059843e-01 -5.27667046e-01 9.66369510e-01 -6.49596870e-01 5.27693406e-02 1.54936373e-01 -2.12886304e-01 1.19722795e+00 -1.43164742e+00 5.15531659e-01 1.40252268e+00 9.00896907e-01 8.80618056e-04 -1.42871487e+00 -5.04081488e-01 -4.45676111e-02 7.93935284e-02 -1.36262155e+00 -7.43909717e-01 6.00671768e-01 -4.11260188e-01 5.79881907e-01 2.28778780e-01 2.72357494e-01 1.05871952e+00 5.86486049e-02 7.19559491e-01 1.22761285e+00 -1.97491378e-01 1.81409627e-01 -1.90330401e-01 3.04640800e-01 8.97519827e-01 5.72905660e-01 -3.30627710e-02 -4.00341034e-01 -2.12012276e-01 9.83148694e-01 -8.99787769e-02 -1.66533664e-01 -5.81640840e-01 -1.91056573e+00 6.49629593e-01 8.61813352e-02 7.46280178e-02 -6.61184311e-01 6.83978558e-01 3.89553607e-01 8.65944207e-01 -9.20986235e-02 3.23007852e-01 -5.65004945e-01 2.10038289e-01 -9.05919790e-01 -5.50501309e-02 9.81139600e-01 9.78255808e-01 8.26230168e-01 5.47269523e-01 -2.29599744e-01 6.70152903e-01 3.29791605e-01 7.21408963e-01 2.43119553e-01 -1.17566168e+00 7.29364216e-01 3.36940438e-02 1.59406468e-01 -1.22676909e+00 -3.90285641e-01 -7.89377093e-01 -1.23446965e+00 3.78101901e-03 2.72513181e-01 -5.00335336e-01 -9.92398977e-01 1.81276131e+00 1.17852226e-01 5.78909159e-01 2.76783556e-01 9.89128947e-01 5.33832312e-01 6.51792049e-01 -6.66764677e-01 -1.77774608e-01 8.11736047e-01 -7.06064343e-01 -8.19454074e-01 -3.79036456e-01 3.40466499e-01 -6.32430017e-01 8.81523013e-01 6.53210938e-01 -8.98421824e-01 -1.53033048e-01 -1.43769002e+00 2.65483037e-02 9.20782760e-02 6.26109421e-01 6.86352611e-01 5.88649035e-01 -1.17412984e+00 9.37309504e-01 -9.64905083e-01 -9.56608653e-02 -1.23095186e-03 9.29585457e-01 -9.50361907e-01 6.92668781e-02 -9.88840103e-01 4.46503103e-01 2.99628854e-01 7.25605428e-01 -1.73464453e+00 -3.78842145e-01 -7.29241431e-01 -1.56718180e-01 4.54606593e-01 -6.41790330e-01 9.47170496e-01 -7.62081385e-01 -1.53829157e+00 6.94155335e-01 -1.97236344e-01 -5.22960782e-01 5.76884806e-01 -6.34497941e-01 2.88071129e-02 -9.62091833e-02 7.74304047e-02 6.72203377e-02 1.48674500e+00 -1.12127352e+00 -8.04088637e-02 -3.37033659e-01 -8.82479325e-02 1.58693850e-01 -1.97398439e-01 -2.19262153e-01 -3.90485257e-01 -6.78094566e-01 7.96760321e-01 -1.26133490e+00 -5.35018086e-01 -2.31723040e-01 -5.97706258e-01 3.33084464e-01 6.56506658e-01 -9.12239492e-01 9.37271535e-01 -1.96204925e+00 6.35489762e-01 5.81911087e-01 6.52079046e-01 -1.33527622e-01 -5.59667349e-01 6.82752132e-01 -3.39148968e-01 -3.35876971e-01 -2.56716520e-01 -3.83863509e-01 -3.37974019e-02 4.27568197e-01 -3.52535754e-01 9.49931264e-01 -3.98259044e-01 5.48097312e-01 -7.08320498e-01 -2.73796786e-02 1.78374216e-01 3.66544783e-01 -4.58971024e-01 -1.12212105e-02 1.62010401e-01 4.88011360e-01 -4.28212404e-01 2.36136809e-01 7.08135486e-01 -2.95763046e-01 2.91105837e-01 -8.74357104e-01 1.31390914e-01 -1.37284338e-01 -1.96444893e+00 1.59165215e+00 -3.80618870e-01 7.79166996e-01 5.47817945e-01 -1.22785521e+00 6.03966177e-01 4.94090825e-01 5.11965632e-01 -5.19165583e-03 4.82612461e-01 1.03962436e-01 -4.04668860e-02 -3.14723939e-01 2.11615071e-01 2.89092064e-02 1.62103966e-01 4.59074557e-01 2.19602644e-01 4.68133599e-01 1.23331696e-02 6.94594026e-01 1.31618702e+00 -1.95584029e-01 2.06444278e-01 -4.23725307e-01 5.67316234e-01 -4.30689067e-01 7.93855429e-01 1.13432968e+00 6.88312352e-02 5.07213235e-01 4.36961830e-01 -3.66935849e-01 -1.46298361e+00 -1.01046753e+00 4.45601083e-02 1.09914887e+00 -3.17973554e-01 -3.27654690e-01 -6.63566411e-01 -3.33517224e-01 -2.20712379e-01 1.70411289e-01 -7.62025177e-01 6.88369796e-02 -7.30923772e-01 -7.88514495e-01 4.68309611e-01 3.55437666e-01 7.97343433e-01 -3.73067945e-01 2.70787984e-01 1.21285677e-01 7.51767606e-02 -1.38373756e+00 -7.59237707e-01 3.67334962e-01 -1.11568809e+00 -8.58806372e-01 -5.12527287e-01 -7.70884335e-01 1.05346286e+00 3.76151025e-01 6.50669634e-01 -4.66616005e-01 2.30389997e-01 5.59991777e-01 -8.81314725e-02 5.10212481e-02 -1.35045066e-01 -1.51560143e-01 7.06454575e-01 7.09068179e-01 -2.91016459e-01 -9.82598603e-01 -4.25840497e-01 1.30255267e-01 -1.11662424e+00 8.20046365e-02 1.06651473e+00 1.01611650e+00 7.38011360e-01 -6.68408116e-04 2.44544670e-01 -1.21059597e+00 8.68279219e-01 -1.69750869e-01 -8.39407206e-01 2.19827861e-01 -5.24640739e-01 6.63124084e-01 6.72374785e-01 -6.14396513e-01 -6.99204803e-01 6.19476736e-01 2.83888638e-01 -9.27577972e-01 4.77422923e-01 6.07404053e-01 -8.20352808e-02 -4.44510937e-01 7.84622192e-01 2.67187178e-01 -3.87590453e-02 -6.84288263e-01 6.85526311e-01 2.80872315e-01 7.66115606e-01 -6.57403886e-01 1.36520171e+00 5.87842703e-01 4.52463269e-01 -6.76405668e-01 -1.09640098e+00 -3.78482938e-01 -8.35317731e-01 -5.81141450e-02 5.38552582e-01 -1.16073823e+00 -8.45731437e-01 3.14409614e-01 -7.49556482e-01 -3.01024616e-01 1.28759397e-02 7.98429310e-01 -5.09811640e-01 6.93844736e-01 -7.70554304e-01 -5.29556751e-01 -4.75053370e-01 -9.68274534e-01 9.59544837e-01 -4.08398777e-01 1.68372422e-01 -1.12865853e+00 1.44857302e-01 2.53396124e-01 2.02894613e-01 3.67403179e-01 5.47218740e-01 -5.47970355e-01 -6.96501970e-01 -5.31800807e-01 -2.65787333e-01 8.19083929e-01 -1.72318682e-01 -4.23024625e-01 -8.21888208e-01 -8.80001903e-01 3.59573364e-01 -1.65572643e-01 9.75626409e-01 4.48022217e-01 7.19426274e-01 -9.59525645e-01 1.63025036e-02 9.78233457e-01 1.65796900e+00 -4.91701543e-01 2.54258454e-01 2.79459655e-01 1.21113288e+00 4.29467112e-02 3.66898507e-01 5.51801980e-01 -5.73357008e-02 2.90825397e-01 3.77343774e-01 2.63023768e-02 2.27075517e-01 -1.31905973e-01 7.30991125e-01 1.19652462e+00 -8.72640163e-02 3.64152431e-01 -6.61560297e-01 3.09072256e-01 -2.28559446e+00 -7.06206679e-01 -2.94046342e-01 2.55770326e+00 7.79551208e-01 2.01674923e-02 -1.94227412e-01 2.20280752e-01 6.11453772e-01 1.90727398e-01 -4.92694169e-01 -1.93932533e-01 -3.67623657e-01 -2.82714143e-02 1.28501487e+00 9.47851717e-01 -1.08892632e+00 1.04974520e+00 7.33591700e+00 5.57289898e-01 -8.47938776e-01 1.58973411e-01 1.17750563e-01 -1.93188488e-02 -1.00321323e-01 3.15794706e-01 -4.95132327e-01 -2.57766724e-01 9.24832046e-01 -1.45059511e-01 1.04999101e+00 7.98012435e-01 5.09848177e-01 4.42222208e-01 -1.45115232e+00 1.19348848e+00 1.29492596e-01 -1.03297985e+00 1.59047902e-01 2.16361344e-01 1.08391142e+00 3.56854439e-01 2.54617780e-01 2.56728530e-01 6.92736566e-01 -8.77773166e-01 5.71804523e-01 6.56686723e-01 5.87041736e-01 -4.66226846e-01 5.44278979e-01 1.90193385e-01 -1.06516349e+00 -2.15750009e-01 -7.62424231e-01 7.99384937e-02 1.72308624e-01 1.09183884e+00 -1.01432753e+00 5.40252142e-02 3.95393848e-01 8.61999810e-01 -4.31521088e-01 8.77583086e-01 -8.09034556e-02 7.94687629e-01 -6.66590571e-01 5.68258524e-01 5.01222551e-01 -8.09075892e-01 1.15877283e+00 8.69209170e-01 2.03918114e-01 -2.16789171e-02 3.04081708e-01 4.69039917e-01 -3.63456517e-01 -3.12238466e-02 -6.56109571e-01 -1.28394291e-01 2.56327033e-01 1.41177702e+00 -3.47380966e-01 -4.02220398e-01 -3.21586430e-01 1.15258229e+00 4.50259417e-01 6.34238899e-01 -4.91177440e-01 -1.23771459e-01 4.45084155e-01 -1.36606872e-01 3.06732446e-01 -7.28881419e-01 -2.31883585e-01 -1.36276519e+00 5.54233454e-02 -1.06553841e+00 3.90210360e-01 -6.31003320e-01 -1.19194496e+00 4.33236629e-01 -1.75887257e-01 -1.26699638e+00 1.39063792e-02 -7.66133964e-01 -2.74946481e-01 2.89858818e-01 -7.73731232e-01 -9.01689768e-01 -1.16600834e-01 8.71703982e-01 1.78022996e-01 -3.87245268e-01 3.92356932e-01 4.65031207e-01 -7.99088717e-01 5.59311628e-01 8.46211970e-01 3.12974185e-01 6.64109886e-01 -1.53797030e+00 1.69338465e-01 1.08007669e+00 2.53568232e-01 1.01035690e+00 1.13057089e+00 -6.01052821e-01 -2.12076163e+00 -1.19213629e+00 3.86571586e-01 -5.35476148e-01 9.06182289e-01 -3.85500759e-01 -5.05010247e-01 1.41477203e+00 5.18342704e-02 -3.59815359e-02 3.90808195e-01 1.06765375e-01 -4.98653293e-01 -4.56809163e-01 -8.60296011e-01 6.17359042e-01 8.29327524e-01 -6.39881968e-01 -5.64881325e-01 7.40736008e-01 7.03104496e-01 -4.67336237e-01 -9.95347977e-01 4.41903383e-01 5.28729200e-01 -5.10132909e-01 9.78173018e-01 -7.23857224e-01 -1.69239685e-01 -6.01258516e-01 -5.31139076e-01 -1.09889805e+00 -6.24345779e-01 -1.02118897e+00 -2.15847671e-01 5.19122124e-01 4.28491056e-01 -5.53317606e-01 9.41346169e-01 5.20650089e-01 -1.65388778e-01 -3.91696811e-01 -8.84911299e-01 -8.44606161e-01 -4.49987948e-01 -3.71107787e-01 -6.47988617e-02 1.02729309e+00 -5.00820458e-01 7.04599738e-01 -1.22690821e+00 5.97323954e-01 1.17527342e+00 -3.17481607e-01 8.74455631e-01 -1.11536264e+00 -5.07747889e-01 2.40834817e-01 -2.34528601e-01 -1.14729261e+00 1.29598742e-02 -9.88253593e-01 3.68095227e-02 -1.51639891e+00 2.75169522e-01 4.85528558e-02 -4.51558739e-01 4.99850810e-01 2.23244727e-01 1.79847509e-01 9.84017029e-02 3.10550034e-01 -8.56508255e-01 5.07611930e-01 1.13951898e+00 -1.31969139e-01 -2.33826801e-01 -4.87155952e-02 -5.57368636e-01 7.32743979e-01 3.24198484e-01 -6.85316265e-01 -3.80110115e-01 -6.88308835e-01 7.13454843e-01 4.01646614e-01 -5.07508814e-02 -1.08116853e+00 4.21633840e-01 6.77952543e-02 4.27595466e-01 -5.74508429e-01 3.47645432e-01 -8.16600144e-01 3.25828910e-01 1.96308926e-01 -5.22930264e-01 2.40226448e-01 -1.19479977e-01 9.82336462e-01 5.28277718e-02 -2.09134713e-01 6.85207009e-01 -9.10404623e-02 -3.13128740e-01 6.30251884e-01 -5.09100795e-01 -2.35784456e-01 4.51560825e-01 -1.91188738e-01 9.74854529e-02 -7.68455625e-01 -1.21391726e+00 2.01010168e-01 -2.28180066e-02 1.22310452e-01 7.56402016e-01 -1.52704120e+00 -8.86861503e-01 1.11275003e-03 -1.58370361e-01 -1.25018165e-01 1.18451677e-01 1.08439302e+00 -5.49552202e-01 5.47913671e-01 -5.49249165e-03 -4.14761126e-01 -1.15159476e+00 4.74641263e-01 3.94321173e-01 -2.99300194e-01 -7.74192512e-01 8.09207082e-01 1.14796229e-01 -6.10257804e-01 4.04729962e-01 -7.12736323e-02 -5.69024533e-02 -2.41063222e-01 3.45773578e-01 5.83031952e-01 8.70021209e-02 -6.88688159e-01 1.36233494e-02 6.15188599e-01 -2.87346035e-01 -4.05374646e-01 1.45024431e+00 -2.07430124e-01 -4.99917984e-01 4.45170552e-01 1.42968702e+00 1.98185861e-01 -1.15364587e+00 -7.09195971e-01 1.22993747e-02 -3.66067529e-01 4.14465368e-01 -1.89717531e-01 -1.29734409e+00 3.71509671e-01 7.75837958e-01 -3.06564927e-01 8.69133949e-01 -5.30418277e-01 7.46862292e-01 1.46777964e+00 2.97649741e-01 -1.13743043e+00 1.43030912e-01 7.53425598e-01 1.07423651e+00 -1.13751078e+00 2.58275151e-01 -1.08761221e-01 -5.10697126e-01 8.85726392e-01 1.96511090e-01 -7.43764281e-01 7.78650820e-01 5.64999059e-02 -1.05430104e-01 -2.56696939e-01 -7.52086341e-01 -1.18063428e-02 4.09532160e-01 3.26448381e-01 6.77651465e-02 1.05692614e-02 -2.40193903e-01 2.22809002e-01 -2.55838126e-01 -3.53092551e-01 7.74618268e-01 5.32683492e-01 -4.77266431e-01 -9.97475088e-01 -5.74359298e-01 7.21225500e-01 -5.43007851e-01 -1.98663265e-01 -4.27234590e-01 1.87924609e-01 -4.36861157e-01 6.89274848e-01 -5.74118733e-01 -6.60255551e-01 1.44057617e-01 -3.09882969e-01 5.76454163e-01 -6.04276538e-01 -1.88832924e-01 2.80861020e-01 2.58624405e-01 -8.61609697e-01 -1.85975760e-01 -6.67953789e-01 -1.05555332e+00 -1.98091805e-01 -1.33304179e-01 9.07179713e-02 7.07045376e-01 1.02637851e+00 1.50487259e-01 1.05260938e-01 7.51437724e-01 -8.01036179e-01 -8.96215558e-01 -9.77961004e-01 -8.31103027e-01 3.76785398e-01 6.43067896e-01 -4.14075851e-01 -6.27796352e-01 7.68723106e-03]
[6.990872383117676, 4.590592861175537]
1caec31e-4c6f-4392-b12e-35084f6c9736
generalized-rectifier-wavelet-covariance-1
2203.07902
null
https://arxiv.org/abs/2203.07902v1
https://arxiv.org/pdf/2203.07902v1.pdf
Generalized Rectifier Wavelet Covariance Models For Texture Synthesis
State-of-the-art maximum entropy models for texture synthesis are built from statistics relying on image representations defined by convolutional neural networks (CNN). Such representations capture rich structures in texture images, outperforming wavelet-based representations in this regard. However, conversely to neural networks, wavelets offer meaningful representations, as they are known to detect structures at multiple scales (e.g. edges) in images. In this work, we propose a family of statistics built upon non-linear wavelet based representations, that can be viewed as a particular instance of a one-layer CNN, using a generalized rectifier non-linearity. These statistics significantly improve the visual quality of previous classical wavelet-based models, and allow one to produce syntheses of similar quality to state-of-the-art models, on both gray-scale and color textures.
['Stéphane Mallat', 'Sixin Zhang', 'Antoine Brochard']
2022-03-14
generalized-rectifier-wavelet-covariance
https://openreview.net/forum?id=ziRLU3Y2PN_
https://openreview.net/pdf?id=ziRLU3Y2PN_
iclr-2022-4
['texture-synthesis']
['computer-vision']
[ 3.36427033e-01 1.45299971e-01 -3.50833088e-01 -1.17756158e-01 -6.16877913e-01 -1.78831682e-01 9.12585258e-01 2.59896576e-01 -1.60124794e-01 7.39924908e-01 2.59289891e-01 1.53716087e-01 -2.46315226e-01 -1.38847697e+00 -7.89095700e-01 -9.86516654e-01 -3.70022744e-01 5.96834570e-02 2.19061926e-01 -5.61867416e-01 1.81956261e-01 7.14611113e-01 -1.87016666e+00 6.00812495e-01 3.18675429e-01 1.44094849e+00 -1.91571370e-01 5.85153341e-01 -1.84887156e-01 7.26622999e-01 -2.95090407e-01 -3.23129982e-01 -2.12727159e-01 -4.09488648e-01 -4.55660790e-01 -3.30771729e-02 7.10717022e-01 7.70580620e-02 -4.96423125e-01 1.10575974e+00 1.97713450e-01 8.50207955e-02 8.79294693e-01 -8.00222933e-01 -1.34214020e+00 3.22231412e-01 -2.71887869e-01 2.32969113e-02 3.58212709e-01 -5.43995015e-02 1.21020079e+00 -8.80549490e-01 7.09858119e-01 1.39329731e+00 9.59366679e-01 4.15279180e-01 -1.86013520e+00 -6.96266145e-02 -2.34130636e-01 2.68733084e-01 -1.17020380e+00 -4.17676032e-01 1.01207542e+00 -2.80874401e-01 7.28513241e-01 4.20535386e-01 7.11952269e-01 1.52707541e+00 6.23642206e-01 5.09221971e-01 1.54717398e+00 -5.75039744e-01 3.80723685e-01 -7.17746317e-02 -3.34569663e-01 8.26691568e-01 2.89033502e-01 3.55379552e-01 -8.35938334e-01 1.30348682e-01 1.19190705e+00 -2.23906368e-01 -2.01695636e-01 -4.92413223e-01 -1.19462204e+00 8.36391807e-01 5.81090629e-01 7.03674436e-01 -6.56130075e-01 5.19793630e-01 4.38744396e-01 2.78723449e-01 1.05475843e+00 4.49923992e-01 -4.01106119e-01 -8.63419175e-02 -1.14978886e+00 1.74701199e-01 7.60422587e-01 6.61386549e-01 9.75307286e-01 4.71989334e-01 -4.81697947e-01 7.47232616e-01 -8.49906802e-02 5.49239397e-01 3.69151235e-01 -6.97069228e-01 -1.84728280e-01 4.60857183e-01 -3.51419628e-01 -1.35409522e+00 -2.68584073e-01 -5.15894413e-01 -1.45168841e+00 5.67370176e-01 2.70747691e-01 5.24978757e-01 -1.12210917e+00 1.54647160e+00 -3.03378880e-01 2.09668964e-01 1.46363884e-01 6.13855660e-01 9.54483926e-01 6.05610430e-01 -8.80377963e-02 7.95263797e-03 1.47223878e+00 -4.71168011e-01 -8.96367073e-01 2.66572893e-01 1.39403287e-02 -7.85556495e-01 8.58350754e-01 5.47017276e-01 -1.17233193e+00 -8.82599533e-01 -1.16537476e+00 4.60183211e-02 -7.51385868e-01 1.51665404e-01 7.41889894e-01 6.25011563e-01 -1.40430152e+00 1.03552783e+00 -6.65605605e-01 -2.17694461e-01 6.47994757e-01 -4.18680906e-02 -4.97199476e-01 1.80887319e-02 -1.06647229e+00 1.10506165e+00 2.40324602e-01 1.29183531e-01 -6.33846402e-01 -5.82620919e-01 -1.13880730e+00 4.13655013e-01 2.31627971e-01 -5.22806406e-01 7.60662913e-01 -1.04439175e+00 -1.69437623e+00 9.71830010e-01 -1.56326637e-01 -6.02247000e-01 3.71684134e-01 -2.03292407e-02 -3.31433773e-01 3.24091941e-01 -3.92746329e-01 6.18455112e-01 1.12646270e+00 -1.63382614e+00 -2.04831883e-01 1.47165269e-01 -1.24894314e-01 -6.75283194e-01 -3.06607932e-01 -2.16498151e-01 1.28617406e-01 -1.08738625e+00 1.76928163e-01 -4.68798935e-01 -2.37061948e-01 2.60001719e-01 -3.24076265e-01 -4.57558688e-03 8.28487515e-01 -5.45043766e-01 9.95098412e-01 -1.92665267e+00 2.96394825e-01 3.07260960e-01 2.38647580e-01 4.04758342e-02 -3.76023412e-01 3.71581614e-01 -2.56929487e-01 3.20240557e-01 -3.69267344e-01 -4.08594966e-01 1.73476875e-01 3.04292619e-01 -1.99669272e-01 4.69814777e-01 8.59397590e-01 1.03232467e+00 -6.58170640e-01 -1.91986218e-01 4.42118496e-01 9.50915635e-01 -3.73251975e-01 -1.92863848e-02 -3.70936275e-01 3.25400352e-01 -2.57527918e-01 6.61928415e-01 7.90905833e-01 -8.71293917e-02 1.54650281e-03 -3.01328480e-01 -8.16286951e-02 -1.50646091e-01 -7.56351650e-01 1.57684755e+00 -8.42207849e-01 9.67151880e-01 -3.30090582e-01 -1.37148046e+00 1.31418204e+00 4.80011165e-01 6.00052953e-01 -9.02766109e-01 8.40562358e-02 3.03082764e-01 -2.40663514e-01 -2.95160979e-01 5.13141453e-01 -3.24956149e-01 5.70773743e-02 -2.35172473e-02 5.96220016e-01 -2.96727061e-01 3.40729117e-01 -4.87404913e-01 9.43099797e-01 3.55312079e-01 3.56131464e-01 -4.86802578e-01 5.91305196e-01 -3.73917699e-01 2.36709818e-01 7.13066220e-01 4.04993802e-01 1.02430391e+00 7.22371995e-01 -8.36042881e-01 -1.57892346e+00 -1.14176595e+00 -4.47575510e-01 7.56074607e-01 -2.87123650e-01 -2.79374927e-01 -6.36963964e-01 -1.34059638e-01 -3.84396017e-02 2.83146501e-01 -1.07146299e+00 -2.26010337e-01 -5.83001554e-01 -5.60413897e-01 4.02760804e-01 3.54583144e-01 6.14746451e-01 -1.45304394e+00 -7.83416629e-01 3.16668183e-01 4.20523956e-02 -1.24435186e+00 2.63913780e-01 3.62548769e-01 -9.23627853e-01 -9.32950258e-01 -1.11823630e+00 -4.80257839e-01 4.35003936e-01 -2.79644251e-01 1.54597795e+00 3.75204869e-02 -5.70545912e-01 3.68871987e-01 -6.57903969e-01 -3.10770124e-01 -6.80812299e-01 -2.67886296e-02 -2.03994438e-01 3.40654522e-01 6.61655068e-02 -8.57955277e-01 -5.20515859e-01 -2.10551694e-01 -1.27289343e+00 1.63069889e-01 9.41213727e-01 1.13283908e+00 7.39917219e-01 2.23549500e-01 3.69783968e-01 -6.20246232e-01 6.81529403e-01 -3.35305333e-01 -4.47947025e-01 2.66205043e-01 -3.25778157e-01 2.94914931e-01 9.45504725e-01 -5.75076699e-01 -7.63672173e-01 -1.99531257e-01 -1.31765962e-01 -3.60296607e-01 -5.21219492e-01 6.03402197e-01 3.38419527e-01 -3.84572774e-01 6.92098022e-01 6.14557505e-01 2.66049802e-02 -4.57831621e-01 3.88724953e-01 5.59619367e-02 5.74747920e-01 -8.10364604e-01 7.68159926e-01 6.82527363e-01 5.43439925e-01 -9.63205278e-01 -6.25609517e-01 -7.07383454e-02 -6.86439455e-01 -1.88381791e-01 9.60219860e-01 -5.47882915e-01 -7.24326074e-01 3.87362599e-01 -1.10234344e+00 -1.31037712e-01 -7.47794092e-01 3.05468708e-01 -1.08319259e+00 1.62847452e-02 -5.44572175e-01 -9.77022588e-01 6.07193587e-03 -9.33048248e-01 1.26652861e+00 1.23028092e-01 -1.12697311e-01 -1.36393642e+00 -1.02479734e-01 -5.22586107e-01 9.93217885e-01 8.93531501e-01 1.22948790e+00 3.20743918e-02 -3.90051663e-01 -1.81669399e-01 -2.27566376e-01 6.73335433e-01 1.20400816e-01 1.81407347e-01 -1.09359562e+00 -1.45224065e-01 -1.77353740e-01 -3.38561416e-01 1.45730960e+00 7.60343552e-01 1.28720236e+00 -3.63956213e-01 1.02994323e-01 7.17272341e-01 1.81941211e+00 -2.04363674e-01 9.20286298e-01 3.61434609e-01 1.89887524e-01 4.84238923e-01 -1.21705040e-01 5.14224589e-01 -1.75341293e-01 6.19852364e-01 4.76173908e-01 -7.00794935e-01 -5.92494607e-01 -1.28085732e-01 8.84905905e-02 8.35586250e-01 -6.76546395e-01 4.77537364e-02 -5.41124701e-01 4.03886437e-01 -1.67023122e+00 -9.97118354e-01 -5.70859499e-02 1.97954702e+00 7.75759637e-01 2.98578382e-01 -1.20724037e-01 4.36049700e-01 4.48420674e-01 5.66996872e-01 -1.59934506e-01 -6.08790457e-01 -6.79217935e-01 1.16662169e+00 4.97554362e-01 1.43622220e-01 -1.19660676e+00 9.51524019e-01 7.05307102e+00 1.00346887e+00 -1.20044470e+00 -1.48736328e-01 8.44196558e-01 3.59997690e-01 -3.82467598e-01 -1.61898479e-01 -8.25442672e-02 1.99315488e-01 8.91130328e-01 1.11719459e-01 4.34491664e-01 6.60134733e-01 -5.22365011e-02 -1.61306992e-01 -9.98393118e-01 9.17927027e-01 7.72937387e-02 -1.77996731e+00 4.33063745e-01 8.04583281e-02 7.87866712e-01 -4.75777745e-01 4.82175708e-01 1.14322796e-01 8.83529708e-02 -1.55453837e+00 8.77456784e-01 1.09562922e+00 1.12145042e+00 -7.28862226e-01 9.25618351e-01 -1.58938050e-01 -1.30892217e+00 1.05435267e-01 -7.17559516e-01 2.94326469e-02 -1.31722853e-01 8.39252710e-01 6.67886138e-02 8.19978595e-01 8.49978924e-01 7.78784931e-01 -5.75011194e-01 8.88934791e-01 -1.32844672e-01 5.84202945e-01 1.45478360e-02 -7.15435520e-02 3.96207541e-01 -2.66857624e-01 3.33308131e-01 1.49083674e+00 5.99267066e-01 -1.60533398e-01 -1.24685481e-01 1.20087790e+00 -9.49460343e-02 1.13197910e-02 -8.81200314e-01 -7.24148601e-02 -1.75051689e-01 1.35630953e+00 -9.65927005e-01 -4.07464623e-01 -3.29791695e-01 6.63370073e-01 5.24083972e-02 5.36365569e-01 -4.87779975e-01 -3.17658514e-01 6.86845422e-01 -4.15958203e-02 3.93297017e-01 -3.83753449e-01 -3.84940416e-01 -1.01231849e+00 5.37991822e-02 -8.30783010e-01 -2.06314221e-01 -8.32020581e-01 -1.44935882e+00 7.81620204e-01 -1.82060122e-01 -1.37120807e+00 1.31987289e-01 -1.12025285e+00 -4.32349175e-01 8.06156158e-01 -1.78751707e+00 -1.41218305e+00 -2.50778913e-01 2.87957072e-01 2.66695857e-01 -2.02269450e-01 1.42685187e+00 -1.68482512e-01 1.12738535e-01 2.80801773e-01 3.12657893e-01 1.05223596e-01 2.82398313e-01 -1.44200182e+00 4.88229185e-01 5.28762817e-01 5.51430583e-01 2.94835836e-01 7.11860776e-01 -1.64656758e-01 -1.44357824e+00 -6.50332928e-01 6.19539142e-01 2.31680106e-02 5.76063752e-01 -4.97863233e-01 -9.74919081e-01 1.51861861e-01 3.67473483e-01 3.83705378e-01 4.84407604e-01 2.16306835e-01 -5.71345448e-01 5.04641384e-02 -1.04536533e+00 4.37861413e-01 8.09924245e-01 -6.05530620e-01 -4.35059398e-01 1.02323733e-01 1.86769634e-01 -1.77799255e-01 -1.13022888e+00 5.96664310e-01 8.01206410e-01 -1.33897376e+00 1.23287344e+00 -6.54395223e-01 1.08269501e+00 1.66626379e-01 -1.77190021e-01 -1.55534160e+00 -5.96472502e-01 -5.78452289e-01 -1.73822731e-01 7.80028105e-01 2.53888935e-01 -4.73741621e-01 7.22226143e-01 -3.96288574e-01 1.25545450e-02 -9.42615390e-01 -1.32523668e+00 -9.33264196e-01 2.07060590e-01 -3.54679346e-01 4.71666783e-01 9.08815980e-01 -1.93565339e-01 -7.96367899e-02 -4.74388391e-01 -1.67268217e-01 5.70018411e-01 -1.98227856e-02 3.55500698e-01 -1.59544528e+00 -5.81446337e-03 -1.08294249e+00 -8.53935480e-01 -4.92887288e-01 3.05662066e-01 -8.35402608e-01 5.81919588e-02 -1.58922112e+00 1.25324681e-01 -2.32838213e-01 -5.70896327e-01 4.76547003e-01 2.30451956e-01 7.29672790e-01 8.95390138e-02 -1.19807251e-01 -1.28952026e-01 7.60516584e-01 1.51652753e+00 -3.58907968e-01 3.75730693e-01 -2.46592328e-01 -3.57266277e-01 7.72000670e-01 7.38217115e-01 -4.71674442e-01 6.87442943e-02 1.14898592e-01 4.45520133e-01 4.90188152e-02 6.20610714e-01 -1.33343029e+00 -9.05001611e-02 -9.89245474e-02 9.14054990e-01 -9.68591571e-02 6.29558921e-01 -7.43227243e-01 -7.71783218e-02 3.54748696e-01 -4.08509880e-01 -1.61621869e-01 3.23017091e-01 4.13278848e-01 -7.74876177e-01 5.71010858e-02 9.41657782e-01 -4.05186951e-01 -5.54722309e-01 2.76326329e-01 -6.13245726e-01 -3.70636553e-01 4.96297717e-01 -3.74146670e-01 -3.69700938e-01 -5.08974969e-01 -8.58330429e-01 -7.68414855e-01 1.93475142e-01 3.28635335e-01 8.94841909e-01 -1.62172234e+00 -9.00418341e-01 3.60315681e-01 1.90968469e-01 -3.07857484e-01 2.04748884e-01 7.63279378e-01 -6.39072776e-01 4.39056069e-01 -7.60073125e-01 -6.36652350e-01 -8.71011972e-01 3.81219685e-01 1.31789461e-01 -3.42425823e-01 -7.68476307e-01 7.18200207e-01 -6.37065098e-02 -1.10416703e-01 -8.01604986e-02 -8.81558478e-01 -2.71484286e-01 3.41492593e-02 1.66746810e-01 6.10295869e-02 3.60459997e-03 -6.05896235e-01 -1.03214309e-01 9.24895346e-01 4.62072641e-01 -2.19799299e-02 1.55367589e+00 4.85397607e-01 -3.20684612e-01 6.11329317e-01 1.29103732e+00 -2.00727701e-01 -1.34142864e+00 -3.19432288e-01 1.34200707e-01 -3.14669758e-01 1.74871117e-01 -6.83069110e-01 -1.14541614e+00 1.01204085e+00 5.09289742e-01 6.86738074e-01 1.34877872e+00 -1.59892812e-01 5.31776249e-01 3.86545807e-01 3.28477383e-01 -1.11563349e+00 4.98089254e-01 5.54020286e-01 1.19873774e+00 -1.06502616e+00 2.04027385e-01 -2.93525428e-01 -1.82793811e-01 1.87436402e+00 -4.12239805e-02 -6.90877318e-01 7.91887820e-01 3.52687895e-01 -1.25674933e-01 -2.15982154e-01 -7.25692093e-01 -5.27018189e-01 7.76980639e-01 6.16298556e-01 7.01329827e-01 1.87001556e-01 -1.54917657e-01 9.88797545e-02 -2.64159411e-01 -5.88404462e-02 4.17016298e-01 8.54370475e-01 -4.33034509e-01 -9.41299319e-01 -3.28998774e-01 3.04499984e-01 -6.66740894e-01 -1.99302271e-01 -1.85041487e-01 9.16215360e-01 1.64959490e-01 6.52946353e-01 2.24954054e-01 -3.90661657e-01 2.90559530e-01 8.29906017e-02 8.64685416e-01 -3.11682463e-01 -3.81983250e-01 1.72898509e-02 -2.89800460e-03 -5.45824111e-01 -7.41051853e-01 -4.51066159e-02 -5.45659244e-01 -3.06991249e-01 -1.77160397e-01 -2.20161915e-01 7.86177754e-01 6.86117887e-01 -8.77743810e-02 1.03528666e+00 5.47520518e-01 -1.32222080e+00 -3.52548599e-01 -1.31292546e+00 -8.64497066e-01 5.75242817e-01 5.40308774e-01 -8.69808555e-01 -2.69773722e-01 3.49595964e-01]
[11.301395416259766, -0.6428686380386353]
e412f938-8b1d-4352-be9b-b60efc13040e
what-if-we-do-not-have-multiple-videos-of-the
null
null
http://openaccess.thecvf.com/content_cvpr_2016/html/Sultani_What_If_We_CVPR_2016_paper.html
http://openaccess.thecvf.com/content_cvpr_2016/papers/Sultani_What_If_We_CVPR_2016_paper.pdf
What If We Do Not Have Multiple Videos of the Same Action? -- Video Action Localization Using Web Images
This paper tackles the problem of spatio-temporal action localization in a video without assuming the availability of multiple videos or any prior annotations. Action is localized by employing images downloaded from internet using action name. Given web images, we first mitigate image noise using random walk framework and evade distracting backgrounds within images using image action proposals. Then, given a video, we generate multiple spatio-temporal action proposals. We suppress camera and background generated proposals by exploiting optical flow gradients within proposal. To obtain the most action representative proposal, we propose to reconstruct action proposals in the video by leveraging the action proposal in images. Moreover, we preserve the temporal smoothness of the video by introducing consensus regularization. Consensus regularization enforces consistency among coefficients vectors of multiple frames within proposal. %We reconstruct video action proposals from image action proposals while enforcing consistency across coefficient vectors of multiple frames by consensus regularization. Finally, the video proposal that have the lowest reconstruction cost and is motion salient is considered as final action localization. Our extensive experiments on trimmed as well as untrimmed datasets validate the effectiveness of proposed approach.
['Waqas Sultani', 'Mubarak Shah']
2016-06-01
null
null
null
cvpr-2016-6
['spatio-temporal-action-localization']
['computer-vision']
[ 4.10319299e-01 -1.16217211e-01 -2.77665913e-01 1.57720502e-02 -5.82205951e-01 -5.18527210e-01 3.11474770e-01 -2.87016094e-01 -5.27303040e-01 7.02544332e-01 5.43735385e-01 2.16754094e-01 -1.20477512e-01 -2.33573943e-01 -8.58401299e-01 -8.32814813e-01 6.07113615e-02 -4.44776535e-01 7.69961715e-01 3.50497663e-01 5.86008310e-01 3.79860669e-01 -1.20916057e+00 4.06376660e-01 5.54407001e-01 7.87130117e-01 3.97590667e-01 6.48959100e-01 1.82743996e-01 1.35798216e+00 -3.33043426e-01 4.97706905e-02 6.63005412e-01 -5.98817289e-01 -7.38092422e-01 9.33178008e-01 8.77900362e-01 -5.34294963e-01 -5.11170328e-01 1.17823780e+00 1.17251739e-01 7.04322278e-01 3.34772110e-01 -1.19704676e+00 -4.75764722e-01 1.65382728e-01 -9.84770954e-01 6.31292462e-01 5.47459066e-01 3.80324304e-01 7.69644737e-01 -1.01804960e+00 1.12761962e+00 1.25999093e+00 1.53336212e-01 4.77623791e-01 -1.29696059e+00 -4.43371952e-01 7.74453998e-01 3.86565268e-01 -1.45842505e+00 -7.90768325e-01 1.02310646e+00 -3.07574779e-01 5.22056699e-01 1.10049158e-01 4.46048349e-01 9.01462615e-01 1.45784572e-01 1.05905282e+00 6.96428001e-01 -2.27294922e-01 2.05110833e-01 -2.78597593e-01 -2.56627351e-01 9.87870097e-01 -5.28890565e-02 -4.35843885e-01 -7.62405872e-01 -3.13039154e-01 9.95977938e-01 3.69361639e-01 -6.95026278e-01 -5.02527893e-01 -1.68583250e+00 4.37828273e-01 7.43037760e-02 2.46200457e-01 -8.81004453e-01 3.19568574e-01 2.02838466e-01 -5.07153757e-03 8.46963972e-02 -2.46519461e-01 -2.80388385e-01 1.90578029e-01 -1.10901070e+00 8.98423418e-02 2.17514589e-01 9.64589775e-01 9.08069491e-01 3.73766482e-01 -3.96039993e-01 4.83102053e-01 3.08215618e-01 5.92959285e-01 2.26742804e-01 -1.75072896e+00 5.80666304e-01 4.20772046e-01 5.88873744e-01 -1.43960428e+00 2.52558887e-01 6.00070283e-02 -7.76383698e-01 1.16459161e-01 3.67093325e-01 1.46715743e-02 -6.89440846e-01 1.49778354e+00 5.89262784e-01 7.93826520e-01 -8.00309703e-02 1.17877328e+00 3.11819762e-01 7.28852510e-01 2.62539566e-01 -8.51497889e-01 9.01883781e-01 -1.19106066e+00 -7.66581655e-01 -1.17458411e-01 2.69757181e-01 -9.42632794e-01 5.13258815e-01 4.81018096e-01 -1.19610655e+00 -6.93154275e-01 -6.28228843e-01 1.43643007e-01 2.98070759e-01 3.72435659e-01 3.75472866e-02 -3.05510778e-02 -9.34602618e-01 4.45798069e-01 -9.37302887e-01 -3.44698846e-01 3.30940992e-01 2.71981299e-01 -6.72852874e-01 -2.82669067e-01 -7.08857119e-01 3.97613466e-01 4.72433180e-01 2.93715503e-02 -1.27289116e+00 -3.43761057e-01 -7.78474212e-01 -4.20044720e-01 6.27790630e-01 -6.38787925e-01 6.29983604e-01 -1.60903800e+00 -1.26980889e+00 3.92687649e-01 -4.10577893e-01 -3.96102935e-01 6.75598681e-01 -2.33464926e-01 -2.86544055e-01 6.55528367e-01 2.52832502e-01 6.15555584e-01 1.36816847e+00 -1.24446392e+00 -9.82163787e-01 9.47887730e-03 -2.99923699e-02 3.92085344e-01 -1.39836624e-01 2.39981133e-02 -9.31393981e-01 -6.23777509e-01 4.08221215e-01 -9.32165980e-01 -3.27098131e-01 1.97162822e-01 -2.26617262e-01 6.09623007e-02 1.10183370e+00 -1.05565345e+00 1.33484054e+00 -2.26212406e+00 3.13252777e-01 5.94237261e-02 2.48190667e-02 5.93783669e-02 -3.88821214e-01 9.75708291e-02 1.71113878e-01 -2.32613906e-01 -1.76433008e-04 -1.33870378e-01 -5.02100348e-01 3.36184531e-01 -1.00528158e-01 9.58776593e-01 6.01322763e-02 5.08489966e-01 -1.01653934e+00 -9.03645813e-01 6.53376341e-01 4.23432380e-01 -7.53577948e-01 1.23728365e-01 -1.48280397e-01 8.17155659e-01 -8.42056215e-01 5.35350859e-01 8.16775024e-01 -1.77201152e-01 1.77857518e-01 -5.32961547e-01 -3.14770639e-01 -2.37826928e-01 -1.66943884e+00 1.88697267e+00 -9.85299200e-02 2.81172425e-01 4.41755921e-01 -1.15116990e+00 5.16712368e-01 2.03646928e-01 1.04169476e+00 -3.18577111e-01 -1.97766975e-01 -1.49658650e-01 -2.09413067e-01 -8.21859717e-01 4.23382908e-01 3.80178034e-01 3.56308788e-01 3.55717719e-01 -1.12110361e-01 5.29408395e-01 2.97369897e-01 4.98727530e-01 1.21780086e+00 5.11979103e-01 2.96891242e-01 -1.16990665e-02 1.01095116e+00 -1.69112563e-01 9.37063217e-01 7.24796891e-01 -6.55123889e-01 6.88271999e-01 8.06565881e-02 -5.28589368e-01 -9.12311137e-01 -1.13293993e+00 5.31943679e-01 1.12523997e+00 4.43352342e-01 -1.96916565e-01 -4.73832458e-01 -8.38696122e-01 -2.15150610e-01 1.62148595e-01 -6.30370140e-01 3.00684184e-01 -9.67553794e-01 -2.69337028e-01 4.56069177e-03 1.87196180e-01 7.70589590e-01 -9.80443060e-01 -6.11420453e-01 4.31076258e-01 -6.57399774e-01 -1.25539339e+00 -1.01697087e+00 -4.23747450e-01 -8.99712622e-01 -1.25422394e+00 -7.98040330e-01 -7.81851530e-01 9.80363607e-01 8.48524034e-01 7.26074457e-01 8.69171992e-02 -1.83281809e-01 6.14913106e-01 -2.48148948e-01 6.26917899e-01 -2.02267930e-01 -6.54703975e-01 3.01687628e-01 6.56628549e-01 2.28030924e-02 -3.94046605e-01 -1.05667961e+00 3.95609349e-01 -9.73944485e-01 5.07485010e-02 4.66862172e-01 5.18543184e-01 8.28767955e-01 1.20769918e-01 1.01918295e-01 -3.07812512e-01 4.33900254e-03 -3.32097113e-01 -4.13440734e-01 3.83278072e-01 1.92872733e-01 -1.46493018e-01 4.97652173e-01 -5.07463992e-01 -1.14587271e+00 6.62528217e-01 5.85541308e-01 -1.00379086e+00 -2.75908649e-01 -2.10666165e-01 -1.00880735e-01 -7.95080699e-03 2.95228720e-01 5.42885244e-01 -1.09322429e-01 -2.13536128e-01 5.03641367e-01 8.89532417e-02 4.88471240e-01 -3.70974004e-01 6.80107415e-01 9.21737969e-01 -1.47265598e-01 -1.04455078e+00 -6.26970828e-01 -7.76350856e-01 -8.25197399e-01 -5.16062617e-01 1.28119123e+00 -9.35889244e-01 -6.00139141e-01 8.36016517e-03 -1.26194572e+00 1.60529956e-01 1.34186551e-01 6.86409175e-01 -5.77238560e-01 9.73385274e-01 -3.23570043e-01 -9.46062207e-01 -1.14206128e-01 -1.05629551e+00 9.58967865e-01 5.78713380e-02 -1.19154215e-01 -8.98924708e-01 -2.67473936e-01 2.51520634e-01 -1.50631787e-02 2.39324734e-01 1.93646953e-01 -7.60115013e-02 -1.26008129e+00 1.22242048e-03 -1.12547942e-01 4.75071728e-01 3.70064378e-01 2.64080018e-01 -4.99325305e-01 -2.57852644e-01 -4.19174172e-02 1.90760508e-01 9.19056714e-01 8.15050781e-01 9.29294825e-01 -5.66865146e-01 -4.46161479e-01 3.11001360e-01 1.46406364e+00 1.04176663e-01 4.34347570e-01 2.65269309e-01 8.14415753e-01 5.50420702e-01 8.71041834e-01 5.83690703e-01 -3.11903171e-02 5.29083073e-01 4.67068732e-01 1.60818890e-01 -1.48887143e-01 -3.33167434e-01 7.28138983e-01 6.06616318e-01 -4.31293994e-01 -1.26515225e-01 -3.79558742e-01 6.15286648e-01 -2.26710868e+00 -1.32604313e+00 -1.24591514e-01 2.29453659e+00 2.94699460e-01 -7.03336820e-02 2.08490305e-02 -1.06268182e-01 1.01182246e+00 4.03959036e-01 -5.12946308e-01 3.80678684e-01 -1.64950624e-01 -4.82298404e-01 6.40689075e-01 5.71633577e-01 -1.22628880e+00 1.03779173e+00 5.51909018e+00 7.32479692e-01 -8.69661391e-01 2.43810207e-01 5.15103817e-01 -2.88429111e-01 -3.53382416e-02 1.13291077e-01 -6.17593169e-01 5.17036498e-01 2.02330738e-01 1.22211408e-03 3.80959570e-01 7.14771688e-01 1.00558841e+00 -5.34409761e-01 -8.14429283e-01 9.54003930e-01 3.01401526e-01 -1.37581766e+00 2.02904135e-01 -1.90027803e-01 1.03136075e+00 -2.48620033e-01 -4.99976166e-02 -2.18414500e-01 2.48687714e-01 -2.58337438e-01 7.58351207e-01 6.80872858e-01 1.04401760e-01 -4.41205889e-01 2.13069573e-01 2.55087584e-01 -1.42459202e+00 -2.64040351e-01 -4.92525697e-01 1.16423391e-01 3.88402998e-01 2.49006644e-01 -3.19238603e-01 4.20127004e-01 6.26387775e-01 1.37915850e+00 -4.06098306e-01 9.45279062e-01 -6.91826195e-02 3.60069096e-01 -1.03736319e-01 5.75749636e-01 3.64546955e-01 -5.63789248e-01 6.87641442e-01 9.45662796e-01 2.86331743e-01 4.34913665e-01 8.00587118e-01 4.24805373e-01 2.41058275e-01 3.61245424e-01 -7.06985831e-01 2.81831861e-01 2.76671588e-01 1.03551328e+00 -7.96890199e-01 -5.08666217e-01 -7.46667862e-01 1.57639194e+00 1.76017359e-02 8.21349323e-01 -9.24721241e-01 3.67417634e-01 4.66764331e-01 1.49065629e-01 4.80072290e-01 -3.16471606e-01 4.16885525e-01 -1.35327220e+00 1.20644204e-01 -7.34451830e-01 5.40384710e-01 -7.77140200e-01 -8.77377629e-01 2.27218315e-01 -7.86874518e-02 -1.72187924e+00 1.86120912e-01 -8.59804228e-02 -5.68905711e-01 4.21997428e-01 -1.24282074e+00 -1.02093923e+00 -2.53702879e-01 1.03029454e+00 1.14104545e+00 6.90469369e-02 5.40368930e-02 2.90910065e-01 -5.13008475e-01 -1.25662446e-01 2.53775883e-02 2.76090298e-02 8.56368661e-01 -6.56491280e-01 -1.60467982e-01 1.21822405e+00 1.74752504e-01 6.03286803e-01 6.75909877e-01 -8.68851602e-01 -1.34646034e+00 -1.28511858e+00 6.36085391e-01 -2.60629356e-01 8.08906853e-01 2.12598577e-01 -7.40783751e-01 7.73821712e-01 3.22558671e-01 4.58394557e-01 1.32181048e-01 -7.95962870e-01 9.57594886e-02 7.27758780e-02 -9.04540539e-01 5.81255972e-01 1.06778646e+00 -2.05374971e-01 -4.08477455e-01 4.91359472e-01 4.65942442e-01 -1.13021933e-01 -4.33091283e-01 7.96831027e-02 3.53345275e-01 -9.82012868e-01 1.15630579e+00 -5.55971444e-01 3.00441504e-01 -8.33597958e-01 -3.16225469e-01 -7.20453143e-01 -4.03841347e-01 -7.93482840e-01 -2.35312078e-02 1.12545645e+00 -1.08634129e-01 8.01283494e-02 9.35250044e-01 6.15042686e-01 1.31442010e-01 -1.61679789e-01 -8.29094827e-01 -6.02953613e-01 -4.90954250e-01 -1.13279529e-01 -3.22806507e-01 8.21641624e-01 -3.25838327e-01 -2.16051340e-01 -1.01296759e+00 3.61961305e-01 9.91593719e-01 1.15070585e-02 8.64770770e-01 -5.13181627e-01 -1.82068467e-01 6.81777596e-02 -4.34180796e-01 -1.21941578e+00 3.47411692e-01 -4.55714017e-01 1.93057194e-01 -1.38786876e+00 3.33329350e-01 1.18009523e-01 -5.05421460e-01 2.56069362e-01 -2.01042697e-01 4.28859562e-01 4.04195219e-01 4.33030814e-01 -1.23581254e+00 4.08198714e-01 1.37908936e+00 -1.39743984e-01 -2.74834216e-01 -1.99188888e-01 9.30548385e-02 1.10560131e+00 4.22784746e-01 -4.58454221e-01 -4.50872451e-01 -4.26005214e-01 -4.57853347e-01 4.82880682e-01 4.59545314e-01 -9.74492729e-01 3.15320581e-01 -6.64148867e-01 4.67714220e-01 -5.26632309e-01 4.11721498e-01 -9.15488482e-01 2.31546983e-01 4.82720017e-01 -3.99508327e-01 1.27610177e-01 -1.97934359e-01 1.22046995e+00 -1.84154183e-01 -7.28766769e-02 1.03563142e+00 -4.19439286e-01 -1.18742502e+00 3.96064371e-01 -6.10768020e-01 -6.46270663e-02 1.32961392e+00 -3.99589062e-01 5.69118112e-02 -3.12328100e-01 -1.04628968e+00 9.22400951e-02 5.23876011e-01 4.71748710e-01 8.42228830e-01 -1.24565959e+00 -5.54627776e-01 1.79076284e-01 -1.24933511e-01 -4.67361927e-01 6.25193715e-01 1.18181610e+00 -5.12966156e-01 1.41588941e-01 -2.60516584e-01 -6.86879218e-01 -1.64437723e+00 7.44254470e-01 1.90710545e-01 3.53096664e-01 -8.75306129e-01 7.27052510e-01 4.94803995e-01 3.56937528e-01 2.87694424e-01 -1.95909247e-01 -2.20447153e-01 -2.40003914e-01 7.60264993e-01 5.23511767e-01 -5.18414617e-01 -1.15090358e+00 -5.15419900e-01 7.86984861e-01 -2.10452043e-02 -1.50408238e-01 1.02310932e+00 -7.19935000e-01 1.45055689e-02 -4.55564819e-02 1.24427044e+00 1.34649873e-01 -1.80912268e+00 -3.13436985e-01 -1.73526481e-01 -1.11452734e+00 -1.58159100e-02 -1.45456180e-01 -1.32289517e+00 3.94817114e-01 4.68575388e-01 -3.77091646e-01 1.13148808e+00 -2.98125178e-01 5.21299720e-01 1.64350554e-01 3.17696929e-01 -1.13534129e+00 6.47748530e-01 1.03396967e-01 6.88370824e-01 -1.38613212e+00 2.87637621e-01 -4.31187630e-01 -9.21596766e-01 9.33153331e-01 6.07759535e-01 -4.48723376e-01 5.54457545e-01 -2.25693896e-01 -5.17313965e-02 1.20437115e-01 -7.81136036e-01 -1.29051432e-01 1.87705353e-01 4.91950184e-01 1.52089119e-01 -4.31366503e-01 -4.23170507e-01 -1.91808254e-01 8.55938137e-01 1.25787467e-01 7.17455328e-01 9.06447828e-01 -7.53046751e-01 -6.97404027e-01 -5.32519639e-01 7.83215463e-02 -5.06697834e-01 9.17819440e-02 -1.56836614e-01 5.40837944e-01 2.46806964e-01 1.10920417e+00 -9.63373780e-02 -4.96603027e-02 8.33743364e-02 1.08260438e-02 3.81202430e-01 -3.89963508e-01 -1.20276794e-01 6.48470104e-01 -8.45215246e-02 -9.83572900e-01 -1.02972257e+00 -9.35340524e-01 -1.17650115e+00 -1.55704200e-01 -1.61772117e-01 1.31641984e-01 1.33287638e-01 7.66516864e-01 2.94655442e-01 1.99270099e-01 6.26208305e-01 -8.28449667e-01 -1.85273573e-01 -6.39149308e-01 -6.72226608e-01 8.06536973e-01 3.77912074e-01 -4.95232582e-01 -5.31243622e-01 8.26525807e-01]
[9.269017219543457, -0.25957173109054565]
b67fe4fa-beba-4d9c-8e17-cb9beece79b2
polyglot-distributed-word-representations-for
1307.1662
null
http://arxiv.org/abs/1307.1662v2
http://arxiv.org/pdf/1307.1662v2.pdf
Polyglot: Distributed Word Representations for Multilingual NLP
Distributed word representations (word embeddings) have recently contributed to competitive performance in language modeling and several NLP tasks. In this work, we train word embeddings for more than 100 languages using their corresponding Wikipedias. We quantitatively demonstrate the utility of our word embeddings by using them as the sole features for training a part of speech tagger for a subset of these languages. We find their performance to be competitive with near state-of-art methods in English, Danish and Swedish. Moreover, we investigate the semantic features captured by these embeddings through the proximity of word groupings. We will release these embeddings publicly to help researchers in the development and enhancement of multilingual applications.
['Rami Al-Rfou', 'Steven Skiena', 'Bryan Perozzi']
2013-07-05
polyglot-distributed-word-representations-for-1
https://aclanthology.org/W13-3520
https://aclanthology.org/W13-3520.pdf
ws-2013-8
['multilingual-nlp']
['natural-language-processing']
[-5.75676799e-01 -5.35756499e-02 -7.25799918e-01 -3.99762839e-01 -9.29694533e-01 -7.21320331e-01 8.60815763e-01 4.90137637e-01 -9.01584387e-01 5.01303673e-01 9.36637402e-01 -4.45055425e-01 4.35143150e-02 -7.31428146e-01 -1.52410045e-01 -2.12196678e-01 -7.47406781e-02 5.85958123e-01 -4.48355405e-03 -4.70576584e-01 5.29884771e-02 3.51355880e-01 -1.07215261e+00 -9.86847561e-03 8.43425453e-01 3.77996325e-01 3.14245135e-01 2.34604359e-01 -7.43616283e-01 1.04619823e-01 -4.18279976e-01 -5.53546965e-01 -5.48074841e-02 2.77627587e-01 -6.56839728e-01 -1.91331103e-01 2.39723995e-01 5.54029681e-02 -6.50238812e-01 1.06378174e+00 4.90799874e-01 2.63561249e-01 7.42417753e-01 -9.47981536e-01 -1.41870117e+00 7.24441826e-01 -3.57935727e-01 4.78117883e-01 4.83972818e-01 -3.14530164e-01 1.70145822e+00 -1.26164591e+00 9.17100608e-01 1.27468240e+00 6.62525952e-01 3.59477341e-01 -1.06143117e+00 -3.74021828e-01 1.67618826e-01 1.36564419e-01 -1.69360077e+00 -4.99978244e-01 5.38579226e-01 -5.59184849e-01 1.66070735e+00 -3.27111483e-01 2.90577888e-01 9.80063140e-01 8.67549479e-02 7.00358510e-01 5.90100050e-01 -7.87017584e-01 -1.48210868e-01 2.37369418e-01 4.45109099e-01 7.60400057e-01 4.65838432e-01 -1.52223304e-01 -5.33205569e-01 -3.92328352e-01 4.43764687e-01 -6.25769943e-02 -1.63808718e-01 -2.15233743e-01 -1.34895396e+00 1.33220983e+00 1.98543236e-01 7.03798652e-01 -2.91548669e-01 2.60783851e-01 4.01946068e-01 1.90694705e-01 1.09372377e+00 5.35887122e-01 -7.65927732e-01 -3.22934419e-01 -4.78077859e-01 1.57920718e-01 6.73531950e-01 8.51626337e-01 8.13143373e-01 1.35606620e-02 1.39301464e-01 1.39979064e+00 5.67813158e-01 3.43810707e-01 1.00495601e+00 -4.83969092e-01 4.82959628e-01 4.96963501e-01 1.02950707e-01 -8.30550551e-01 -2.78321296e-01 -6.56304657e-02 2.83741206e-02 -5.79522550e-01 1.35945156e-01 -1.17250703e-01 -7.59213686e-01 1.62039828e+00 1.87767118e-01 2.41700970e-02 3.20430607e-01 5.51327169e-01 7.09905803e-01 9.23667789e-01 3.26394081e-01 2.24776343e-01 1.72772348e+00 -9.65335190e-01 -8.22070003e-01 -5.31116545e-01 9.77569461e-01 -9.16123688e-01 9.74202633e-01 -4.98425700e-02 -6.83109462e-01 -5.17083228e-01 -8.07839334e-01 -4.73868132e-01 -9.02230442e-01 -2.86161844e-02 9.62298810e-01 6.68084145e-01 -9.19937074e-01 2.54581273e-01 -8.35184574e-01 -7.07185209e-01 1.23562425e-01 4.53688204e-02 -6.04252398e-01 -4.24490243e-01 -1.53502369e+00 1.09420919e+00 2.58038729e-01 -6.32473886e-01 -3.64642799e-01 -7.47053325e-01 -1.40631580e+00 -1.20522298e-01 -3.63843739e-01 -1.59537926e-01 9.46412086e-01 -3.05092365e-01 -9.59969223e-01 1.20125735e+00 -3.90126407e-01 -3.09440494e-01 -2.73566455e-01 -3.51716161e-01 -9.01987493e-01 -1.44251838e-01 3.53706241e-01 7.63828039e-01 1.04996160e-01 -7.03755677e-01 -7.09909797e-01 -2.98819214e-01 -3.89593579e-02 1.13134101e-01 -1.00230336e+00 5.85238338e-01 -5.43421507e-01 -9.08359945e-01 -1.37583017e-01 -5.79325676e-01 -3.15486282e-01 -1.02797654e-02 8.43838230e-02 -9.29001629e-01 3.84115368e-01 -8.46878648e-01 1.38181543e+00 -2.46050262e+00 -5.81342131e-02 -1.53419435e-01 4.04267805e-03 2.76228070e-01 -6.31377518e-01 8.79771829e-01 1.22576185e-01 4.43368882e-01 1.71086654e-01 -3.72307837e-01 3.11789542e-01 6.29534721e-01 -2.36182347e-01 6.11486375e-01 4.42793906e-01 9.90569949e-01 -1.17491829e+00 -3.42372149e-01 1.49859190e-01 7.58172095e-01 -4.98058379e-01 -1.00422828e-02 3.86289693e-02 -2.42079288e-01 -4.47432816e-01 5.71096480e-01 3.56571913e-01 3.25224340e-01 5.25846660e-01 7.02122524e-02 -2.64978737e-01 8.80049527e-01 -6.80460095e-01 1.73740411e+00 -8.23844254e-01 7.28885472e-01 -3.17369223e-01 -9.42694962e-01 1.14337289e+00 4.59147096e-01 3.88900220e-01 -5.49622118e-01 -1.07488237e-01 4.46034282e-01 1.50729269e-01 -3.30280304e-01 8.25392783e-01 -1.78630605e-01 -4.19565111e-01 4.96309072e-01 6.32625401e-01 2.95558367e-02 2.78836459e-01 9.40376669e-02 9.03151870e-01 -1.44747257e-01 7.76502073e-01 -5.94533682e-01 2.55180478e-01 5.39595559e-02 5.12730360e-01 2.83379734e-01 -3.39273363e-01 3.94265205e-01 3.28749239e-01 -4.07204300e-01 -1.06460941e+00 -1.24042010e+00 -5.19798577e-01 1.64248955e+00 -2.55575895e-01 -6.51778042e-01 -6.73309416e-02 -6.53753459e-01 4.07015860e-01 8.27403843e-01 -5.26700556e-01 -9.09417123e-02 -5.13194621e-01 -5.72527885e-01 7.34477520e-01 8.24982762e-01 -2.66547441e-01 -8.47164273e-01 2.37352535e-01 4.42613214e-01 4.57793996e-02 -1.35559452e+00 -6.47993028e-01 3.92792374e-01 -5.94437778e-01 -8.53401780e-01 -7.19390929e-01 -1.39943564e+00 3.99632633e-01 2.55717665e-01 1.12515748e+00 -1.87930554e-01 -1.93098530e-01 4.35154170e-01 -5.62501550e-01 -3.22764784e-01 -1.78537920e-01 1.60887003e-01 4.93684351e-01 -4.81261522e-01 1.03118026e+00 -2.35945120e-01 -2.40002424e-02 -1.02928728e-01 -8.57100010e-01 -7.44800925e-01 3.02963704e-03 7.47274935e-01 3.95784289e-01 -3.71257156e-01 4.35411036e-01 -8.53331029e-01 9.92359698e-01 -8.71806979e-01 -2.43349239e-01 1.45394564e-01 -4.70932305e-01 2.49516159e-01 3.85837466e-01 -5.02476037e-01 -7.97699213e-01 -2.27421716e-01 -4.36502457e-01 2.79035233e-02 -4.72783437e-03 7.97476351e-01 5.36107924e-03 1.56769857e-01 3.72686684e-01 3.43928337e-02 -2.40888163e-01 -7.77087629e-01 8.34865808e-01 1.07895553e+00 2.44759396e-01 -7.56350636e-01 5.98389745e-01 1.18107855e-01 -7.56262302e-01 -1.14039314e+00 -6.67786777e-01 -1.01238859e+00 -6.61555290e-01 4.79237556e-01 9.09103274e-01 -1.35281599e+00 3.33987594e-01 -4.21943739e-02 -1.32444382e+00 1.24265157e-01 -2.32138366e-01 7.26190031e-01 -5.91623820e-02 4.74197656e-01 -7.04791188e-01 -5.27064979e-01 6.95787668e-02 -8.28364968e-01 1.06932449e+00 -5.87232076e-02 -4.69929248e-01 -1.75104415e+00 6.04880512e-01 1.07691027e-01 2.99339294e-01 -2.22449109e-01 1.05969894e+00 -1.27997351e+00 2.51396209e-01 -3.26810360e-01 -1.68751031e-01 3.59571666e-01 3.84034634e-01 -5.17097414e-02 -7.89268851e-01 -1.52593032e-01 -7.43151486e-01 -2.24757120e-01 1.03023493e+00 2.12317228e-01 5.97835362e-01 -2.82662865e-02 -4.83526707e-01 5.87327898e-01 1.54860353e+00 -9.47563648e-02 3.41041744e-01 4.37692970e-01 6.06339335e-01 5.94434500e-01 3.45052004e-01 3.66486579e-01 5.91692805e-01 5.14038742e-01 -1.05785996e-01 8.31460804e-02 -1.77308246e-01 -5.35032153e-01 5.27175128e-01 1.45552421e+00 3.04168254e-01 -3.87457877e-01 -1.30345380e+00 1.34493327e+00 -1.58615112e+00 -5.05392969e-01 -7.06658885e-02 1.83792651e+00 9.31508482e-01 -1.42536208e-01 -1.98020816e-01 -3.96946758e-01 6.21113479e-01 6.43675208e-01 2.18045533e-01 -9.24838960e-01 -1.56105921e-01 7.54701674e-01 6.03960574e-01 7.53688514e-01 -1.10849071e+00 1.49581850e+00 7.00486088e+00 6.75598025e-01 -6.60697937e-01 5.39205492e-01 3.29052769e-02 3.99202138e-01 -8.43830347e-01 -3.13442387e-02 -1.01566541e+00 2.19019353e-01 1.11907506e+00 -5.13478577e-01 1.08247913e-01 8.89672101e-01 5.22210561e-02 3.41315657e-01 -9.48839188e-01 6.89241886e-01 2.20009059e-01 -1.12740004e+00 1.56691253e-01 1.68590993e-01 8.69125247e-01 7.22680867e-01 -5.35321645e-02 5.62033355e-01 7.88174510e-01 -1.06295347e+00 2.24193767e-01 3.99226025e-02 9.21831191e-01 -8.86819422e-01 7.86016941e-01 -2.56946380e-03 -1.33677876e+00 6.28554448e-02 -8.91546011e-01 -1.02518611e-01 3.70177597e-01 5.59375703e-01 -7.46844590e-01 3.85373831e-01 3.24743420e-01 9.18866277e-01 -3.70571017e-01 6.61862135e-01 -4.67879981e-01 6.70042217e-01 -2.27657065e-01 -1.38513684e-01 6.45840347e-01 -7.33519569e-02 3.82134914e-01 1.71672034e+00 4.49220181e-01 -3.25149536e-01 3.21772039e-01 5.39598048e-01 -3.73233736e-01 4.52875733e-01 -9.05038536e-01 -8.68470848e-01 7.60315418e-01 1.08211100e+00 -2.29398116e-01 -1.88302264e-01 -9.51383293e-01 9.30124760e-01 7.01825619e-01 2.92375654e-01 -4.10221487e-01 -8.57316256e-01 1.66125584e+00 -7.25228190e-02 4.01198655e-01 -8.25084031e-01 1.88258499e-01 -1.28180611e+00 -1.78075284e-01 -3.78280759e-01 3.26199263e-01 -4.24683154e-01 -1.85428286e+00 5.45242310e-01 -1.61404461e-01 -7.91012228e-01 -3.45065266e-01 -1.20147038e+00 -5.40194213e-01 1.02499413e+00 -1.91143095e+00 -1.14877844e+00 4.04698044e-01 2.77749300e-01 5.53573430e-01 -5.73419094e-01 1.20848906e+00 4.19854909e-01 -3.20603728e-01 5.48021317e-01 5.72462022e-01 6.17881000e-01 8.62653375e-01 -1.25345647e+00 9.21911538e-01 5.65413713e-01 9.47380602e-01 8.94667029e-01 4.30833608e-01 -5.09169877e-01 -1.24527061e+00 -9.79251564e-01 1.95085120e+00 -4.48136061e-01 1.50308764e+00 -5.83350062e-01 -7.35474646e-01 9.18225110e-01 4.02196109e-01 3.68129730e-01 1.17239869e+00 7.77292609e-01 -6.32773757e-01 2.27683708e-01 -8.11337054e-01 3.31210852e-01 1.03782809e+00 -8.42036486e-01 -1.16855562e+00 4.17729437e-01 1.06555176e+00 1.78401977e-01 -9.80720162e-01 -1.74226061e-01 4.26835060e-01 -2.11759686e-01 1.04772878e+00 -1.04797482e+00 2.22705856e-01 1.05302498e-01 -6.30676866e-01 -1.71878350e+00 -5.36172986e-01 -1.12882331e-01 9.61377919e-02 1.48856461e+00 4.58115548e-01 -9.12150383e-01 3.20750833e-01 1.64506361e-02 -2.27253437e-01 -5.55253565e-01 -1.02239811e+00 -1.00601220e+00 6.60900533e-01 -7.24646032e-01 5.38577974e-01 1.16416204e+00 3.62625927e-01 1.47539303e-01 1.31157577e-01 1.84555411e-01 2.54911393e-01 -2.70354360e-01 2.61029035e-01 -1.24896288e+00 -2.97138398e-03 -3.40742171e-01 -8.18985224e-01 -1.11090744e+00 7.74634778e-01 -1.31531489e+00 -1.54386386e-01 -1.85281718e+00 3.93730290e-02 -4.06479120e-01 -6.12652898e-01 5.99570274e-01 -1.68547466e-01 2.89675236e-01 -7.52459541e-02 -4.08428808e-04 -3.93000066e-01 6.59749150e-01 5.17884076e-01 -2.53565669e-01 1.60841540e-01 -6.93832755e-01 -8.92512381e-01 6.58123434e-01 7.90469050e-01 -5.69828928e-01 -2.70464849e-02 -1.05881441e+00 2.45068416e-01 -6.47315681e-01 -1.86876044e-01 -6.18551254e-01 -1.16800524e-01 -1.49385154e-01 7.66044185e-02 -1.57063007e-01 2.43776828e-01 -5.12360096e-01 -6.50071919e-01 2.04372555e-01 -2.66555458e-01 3.25970143e-01 1.48627013e-01 4.68523204e-01 -6.18710995e-01 -4.77030009e-01 4.10154819e-01 5.19867651e-02 -1.23083770e+00 2.11327717e-01 -5.47704220e-01 5.67287862e-01 8.34590018e-01 6.51021823e-02 -1.13617584e-01 -1.42596081e-01 -4.63353634e-01 2.27863073e-01 3.53182435e-01 1.02912128e+00 4.48819280e-01 -1.66863275e+00 -7.34012067e-01 2.73428798e-01 7.02117026e-01 -7.10478067e-01 -2.56544054e-01 2.72859126e-01 -5.34563243e-01 7.46817410e-01 2.44354811e-02 4.61081266e-02 -9.39080119e-01 4.45490211e-01 -1.05775453e-01 -2.91138500e-01 -3.10484886e-01 9.07086432e-01 -4.55704890e-03 -9.50202167e-01 -3.73125635e-02 -3.05486947e-01 -3.02876443e-01 3.36059004e-01 5.20070314e-01 1.40130699e-01 -1.11991040e-01 -7.88686156e-01 -8.38107526e-01 4.05852348e-01 -6.61507472e-02 -2.26218566e-01 1.68583679e+00 -5.61931506e-02 -7.18220472e-02 5.87410748e-01 1.61188185e+00 5.54753006e-01 -6.00712121e-01 -4.59538251e-01 4.46014225e-01 -3.96727651e-01 1.08479172e-01 -3.67141634e-01 -7.85560191e-01 9.41228628e-01 2.33663246e-01 -2.01473129e-03 2.75101215e-01 4.93072391e-01 1.05920660e+00 2.46835798e-01 4.50847745e-01 -1.20905411e+00 -3.67693037e-01 9.77881610e-01 3.93420547e-01 -1.09599829e+00 -7.80145302e-02 -1.77970096e-01 -6.09903514e-01 1.27181625e+00 9.72593054e-02 -3.88688087e-01 9.59926307e-01 3.19581658e-01 2.37361118e-01 -1.00749671e-01 -7.36259758e-01 -5.77759027e-01 4.35729980e-01 1.01986325e+00 1.18611467e+00 3.11036527e-01 -7.86793053e-01 7.31155634e-01 -7.83130750e-02 -4.03651178e-01 2.92064369e-01 7.85275042e-01 -6.80292428e-01 -1.75327301e+00 8.20718147e-03 2.83044666e-01 -5.35227776e-01 -6.44861341e-01 -1.83953792e-01 8.94875467e-01 9.91573781e-02 8.05346131e-01 4.62711632e-01 -6.98178560e-02 1.22496761e-01 4.06789035e-01 2.57674664e-01 -1.14983761e+00 -3.26084375e-01 -1.32812858e-01 4.95630980e-01 -2.91449666e-01 -1.71793282e-01 -6.25321031e-01 -1.19697177e+00 -1.26694888e-01 -7.09773749e-02 2.59304196e-01 8.06509912e-01 7.75726318e-01 3.18093449e-01 2.04224601e-01 4.08375323e-01 -4.26074356e-01 -6.84581697e-01 -1.10631537e+00 -8.52196991e-01 3.48269522e-01 -5.73260263e-02 -7.67519116e-01 -3.06640089e-01 -2.47529238e-01]
[10.77316951751709, 9.65013599395752]
c30f3811-b153-473f-b8de-3011247badfa
can-deep-neural-networks-learn-process-model
2202.11985
null
https://arxiv.org/abs/2202.11985v1
https://arxiv.org/pdf/2202.11985v1.pdf
Can deep neural networks learn process model structure? An assessment framework and analysis
Predictive process monitoring concerns itself with the prediction of ongoing cases in (business) processes. Prediction tasks typically focus on remaining time, outcome, next event or full case suffix prediction. Various methods using machine and deep learning havebeen proposed for these tasks in recent years. Especially recurrent neural networks (RNNs) such as long short-term memory nets (LSTMs) have gained in popularity. However, no research focuses on whether such neural network-based models can truly learn the structure of underlying process models. For instance, can such neural networks effectively learn parallel behaviour or loops? Therefore, in this work, we propose an evaluation scheme complemented with new fitness, precision, and generalisation metrics, specifically tailored towards measuring the capacity of deep learning models to learn process model structure. We apply this framework to several process models with simple control-flow behaviour, on the task of next-event prediction. Our results show that, even for such simplistic models, careful tuning of overfitting countermeasures is required to allow these models to learn process model structure.
['Jochen De Weerdt', 'Seppe vanden Broucke', 'Jari Peeperkorn']
2022-02-24
null
null
null
null
['predictive-process-monitoring']
['time-series']
[ 4.53434408e-01 3.95962119e-01 -9.76897329e-02 -2.90878147e-01 -1.35229826e-01 -2.59292185e-01 1.06727719e+00 4.83445257e-01 -4.54711653e-02 4.17137951e-01 3.11284542e-01 -5.24436295e-01 -6.83007598e-01 -8.73681128e-01 -4.31892961e-01 -4.16160047e-01 -2.89921463e-01 5.93646288e-01 7.88411573e-02 1.38883799e-01 4.33799058e-01 6.78427637e-01 -1.31077242e+00 3.99750501e-01 3.53844762e-01 1.09626806e+00 -1.13720171e-01 7.31252968e-01 -3.56898218e-01 1.81257951e+00 -4.70846653e-01 -3.46966684e-01 1.19941495e-01 -1.99297547e-01 -7.30801702e-01 -2.24812143e-02 -2.67840296e-01 -1.15918860e-01 -4.93371904e-01 3.72611940e-01 2.09961444e-01 5.51983602e-02 6.33142710e-01 -1.03533304e+00 -1.62420347e-01 9.32501495e-01 -2.03213528e-01 5.30257702e-01 1.19172715e-01 5.35374165e-01 9.22929168e-01 -2.79205948e-01 3.43316406e-01 1.29701936e+00 8.41312885e-01 4.58232313e-01 -1.32553530e+00 -4.27500308e-01 3.07834923e-01 1.90919951e-01 -6.15618289e-01 -3.81087244e-01 4.99386907e-01 -4.61243480e-01 1.18572927e+00 1.81191936e-02 6.22657776e-01 1.45453775e+00 6.66137457e-01 9.21401381e-01 1.03485918e+00 -1.72557831e-01 3.70089918e-01 -9.21380669e-02 1.66541591e-01 3.26569021e-01 1.23976238e-01 2.53251135e-01 -3.86076659e-01 -1.73496962e-01 8.96314859e-01 5.66007614e-01 2.75142230e-02 -1.99005723e-01 -1.11114895e+00 5.40430486e-01 1.32295996e-01 5.42155683e-01 -7.57641196e-01 4.76316392e-01 5.89039683e-01 4.90184069e-01 2.59305000e-01 7.71665514e-01 -6.96554840e-01 -5.62637925e-01 -8.54154170e-01 1.81868002e-01 1.37040102e+00 5.66915572e-01 3.42852116e-01 2.00722203e-01 -8.61088574e-01 4.32180464e-01 5.04217073e-02 -7.38296583e-02 6.80042028e-01 -9.26974833e-01 4.45829421e-01 7.73363411e-01 6.84071109e-02 -7.67510772e-01 -5.91323078e-01 -5.51352620e-01 -1.06422818e+00 -1.00143254e-01 3.36190104e-01 -4.23903167e-01 -6.40583515e-01 1.28259897e+00 -3.47192585e-01 4.47938234e-01 3.41479853e-02 1.93881989e-01 -4.78852652e-02 7.89972126e-01 4.93889004e-01 -3.26278478e-01 1.14229465e+00 -9.29807007e-01 -8.77689362e-01 -3.88386697e-01 4.70989168e-01 -3.86402547e-01 5.21877766e-01 4.53700960e-01 -1.19169557e+00 -7.27935314e-01 -5.95159113e-01 5.13135195e-01 -1.93809018e-01 -9.44177359e-02 7.80189812e-01 5.32740474e-01 -8.85560215e-01 1.24850631e+00 -1.13879490e+00 -4.58428472e-01 4.80749369e-01 4.82358307e-01 9.26364213e-02 2.06521302e-01 -1.04045677e+00 9.82029974e-01 3.77173722e-01 2.17472181e-01 -1.18141246e+00 -6.14792109e-01 -3.34019929e-01 7.46196806e-01 5.39380729e-01 -7.61194289e-01 1.56708384e+00 -9.21062231e-01 -1.57524467e+00 2.45537162e-01 -1.14504829e-01 -1.11314821e+00 5.21946788e-01 -2.85562456e-01 -4.78381723e-01 -2.87427396e-01 -6.62370384e-01 1.55871913e-01 9.48125720e-01 -1.00426090e+00 -8.61987710e-01 -3.49809021e-01 -1.22192279e-01 -2.34478578e-01 -2.00237885e-01 2.45022789e-01 7.12929070e-02 -5.27583838e-01 -2.25242116e-02 -7.60710001e-01 -7.13882446e-01 -4.63700205e-01 -3.68977010e-01 -3.92239422e-01 5.39646447e-01 -5.81928492e-01 1.41141844e+00 -1.87448668e+00 -1.74054608e-01 4.01925921e-01 1.40517026e-01 2.67581522e-01 4.61652577e-02 7.60041118e-01 -5.00076115e-02 3.15090299e-01 9.99311805e-02 -3.82022679e-01 2.14318037e-01 3.25255930e-01 -5.27815163e-01 5.73067926e-02 5.74394643e-01 9.74376976e-01 -5.28965354e-01 -1.38843311e-02 3.16168547e-01 3.00810337e-01 -4.45599928e-02 4.58406508e-01 -3.03329170e-01 3.72865647e-01 -5.05140960e-01 5.18046558e-01 1.80512760e-02 -4.40806627e-01 4.39916074e-01 4.49889034e-01 3.34451050e-02 3.47883403e-01 -9.88019407e-01 9.57227468e-01 -6.97560668e-01 4.97145444e-01 -2.08988622e-01 -9.77088809e-01 1.11264682e+00 6.31936550e-01 5.15017569e-01 -6.32588863e-01 3.56516289e-03 -8.60019550e-02 3.72969389e-01 -4.24286395e-01 2.98887014e-01 -2.50939518e-01 1.89390436e-01 6.67497039e-01 -9.10533592e-02 3.39911789e-01 2.21378401e-01 -4.88155901e-01 1.57703936e+00 2.19458267e-01 4.54343945e-01 1.08820580e-01 6.85829103e-01 -4.33786631e-01 5.41444361e-01 1.00593543e+00 -2.51618773e-01 1.90601990e-01 9.92648602e-01 -7.40116358e-01 -1.14085853e+00 -6.55080736e-01 2.70091534e-01 1.17961478e+00 -5.45344770e-01 -4.55101818e-01 -4.73234922e-01 -4.78829920e-01 -3.30411717e-02 8.55098486e-01 -6.13549769e-01 -3.35099518e-01 -7.32205212e-01 -7.25558817e-01 6.31357849e-01 8.34951937e-01 3.48457634e-01 -1.49838722e+00 -7.39277840e-01 9.59903061e-01 6.27218843e-01 -1.00470030e+00 3.62752885e-01 5.99499881e-01 -1.40680742e+00 -9.75175917e-01 -2.25687906e-01 -4.13435698e-01 2.91163996e-02 -4.88503724e-01 1.12146807e+00 -3.21052790e-01 1.19012915e-01 2.22319141e-01 3.84751149e-02 -6.03437304e-01 -6.54740155e-01 3.81201118e-01 -2.35452622e-01 1.27293572e-01 6.12577498e-01 -7.33806849e-01 -3.61579418e-01 1.73887163e-01 -7.42611289e-01 -1.81202769e-01 1.06896603e+00 6.01704061e-01 2.82767534e-01 2.36400396e-01 6.48910642e-01 -8.70104074e-01 1.08766961e+00 -4.10933733e-01 -3.47540230e-01 4.30710763e-01 -8.45578611e-01 3.66839349e-01 7.36704290e-01 -4.66189653e-01 -1.21450830e+00 -1.49552330e-01 5.43982908e-03 -5.53719938e-01 -5.90705752e-01 5.41726470e-01 8.08551628e-03 5.52523315e-01 3.23188484e-01 4.83954966e-01 -2.80850865e-02 -3.48323584e-01 -2.94195767e-02 2.42760152e-01 7.14854896e-02 -4.55686361e-01 4.25684094e-01 3.19369555e-01 1.77072555e-01 -2.85124153e-01 -6.51097953e-01 -3.40730339e-01 -3.97695214e-01 -1.55437797e-01 6.56279147e-01 -6.88563228e-01 -1.10413933e+00 3.03703308e-01 -9.68749523e-01 -5.21080017e-01 -5.43393016e-01 2.87341386e-01 -8.85849714e-01 -2.07210943e-01 -1.07992148e+00 -1.15779853e+00 -5.25761902e-01 -7.86309421e-01 8.27011824e-01 1.14565529e-01 -5.72703362e-01 -1.38431501e+00 3.09592783e-02 -2.02709232e-02 7.07016051e-01 8.91947672e-02 1.13206673e+00 -1.26469278e+00 -6.44921482e-01 -5.26731908e-01 3.93129773e-02 4.07768309e-01 7.81074464e-02 3.57300974e-02 -9.15455222e-01 -1.02137662e-02 4.54507440e-01 1.67905107e-01 6.23931170e-01 4.21717972e-01 1.57276833e+00 -3.06259304e-01 -5.03449559e-01 2.04527184e-01 1.15534437e+00 3.77506614e-01 8.64087224e-01 5.18222213e-01 7.18296707e-01 8.98555815e-01 2.71970004e-01 6.13991380e-01 7.19618127e-02 1.95242643e-01 2.62551337e-01 3.16800922e-01 3.90913993e-01 -2.15398714e-01 5.85993826e-01 6.33515060e-01 -5.27806580e-01 -2.05184400e-01 -1.21667314e+00 1.96254238e-01 -1.93667555e+00 -1.37201393e+00 -2.38865942e-01 2.02793503e+00 2.67764241e-01 6.91298783e-01 -1.54068679e-01 3.91445696e-01 6.17752075e-01 1.58492759e-01 -5.27487636e-01 -7.51478732e-01 1.83968201e-01 2.43070930e-01 4.33039814e-01 -1.13039678e-02 -1.19256485e+00 6.15979910e-01 6.31124115e+00 7.25016415e-01 -9.47237611e-01 3.25444266e-02 1.02316284e+00 -4.43117134e-02 -2.57951245e-02 -5.43243214e-02 -9.42159712e-01 3.19937438e-01 1.70831382e+00 -1.86575260e-02 1.64963365e-01 8.02524686e-01 7.74319291e-01 2.91755140e-01 -1.50237131e+00 7.01601863e-01 -4.16795194e-01 -1.25361216e+00 7.06165805e-02 2.54540980e-01 6.39669299e-01 -2.43434995e-01 -2.44632512e-02 8.76380205e-01 5.45274973e-01 -1.48506331e+00 3.85727108e-01 1.07316399e+00 -2.39425257e-01 -9.62145090e-01 9.18797493e-01 4.02675480e-01 -8.89100254e-01 -8.21642041e-01 -2.31369346e-01 -3.45417261e-01 1.30000561e-01 5.86739779e-01 -9.71115708e-01 4.10292745e-01 4.00863141e-01 9.16831255e-01 -5.45172989e-01 9.83801305e-01 -2.98963577e-01 9.48679745e-01 -1.60050374e-02 -9.00751576e-02 4.90395099e-01 -1.39895886e-01 3.53364795e-01 1.10079098e+00 3.69964957e-01 -5.45290947e-01 1.89124987e-01 8.60362411e-01 1.31311059e-01 -1.32971093e-01 -5.22720933e-01 -2.75857478e-01 9.29469913e-02 1.11288106e+00 -6.99227512e-01 -3.44422966e-01 -2.40762725e-01 4.81585026e-01 1.67422712e-01 3.73606592e-01 -5.71297526e-01 1.78165898e-01 7.13415325e-01 4.96920466e-01 4.14041549e-01 -7.78233409e-02 -5.46392918e-01 -6.46826386e-01 -1.98924512e-01 -9.24102902e-01 4.34916735e-01 -6.17216170e-01 -1.51749921e+00 4.93804574e-01 -3.32385540e-01 -8.44924569e-01 -8.33655894e-01 -5.82955480e-01 -9.92240608e-01 7.48667002e-01 -1.34426689e+00 -1.01070404e+00 8.36484600e-03 3.09427947e-01 6.62893474e-01 -2.56916881e-01 7.35477567e-01 -2.68179983e-01 -7.57227480e-01 -2.54876968e-02 2.29515418e-01 -9.27135721e-02 4.08219099e-01 -1.35613763e+00 4.49312836e-01 4.62864310e-01 -2.07447767e-01 6.68542266e-01 7.63494551e-01 -5.80169082e-01 -1.36932850e+00 -1.33977079e+00 8.51684272e-01 -6.53812110e-01 9.45330203e-01 -1.01853527e-01 -1.14004421e+00 1.00231493e+00 2.01817080e-01 -2.04307094e-01 5.78087211e-01 3.37351561e-01 1.44313335e-01 -2.14580074e-01 -7.86237121e-01 5.28244853e-01 8.89590442e-01 -4.55883741e-01 -6.71137691e-01 1.41477093e-01 3.84656668e-01 1.04478963e-01 -1.18521595e+00 4.69659656e-01 2.77125478e-01 -1.03649235e+00 7.63643444e-01 -8.25330496e-01 6.27922654e-01 1.48305178e-01 2.55514622e-01 -1.04979837e+00 -3.94639343e-01 -7.23991334e-01 -8.79085898e-01 1.15211141e+00 3.11926246e-01 -5.02623439e-01 9.67653394e-01 6.74937546e-01 5.87424226e-02 -1.05575371e+00 -5.20797968e-01 -7.00597525e-01 -1.13237863e-02 -5.11947572e-01 7.51348376e-01 7.63885379e-01 -3.30231279e-01 4.60798502e-01 -3.15853983e-01 1.27371540e-02 2.00319290e-01 4.89572762e-03 4.61890161e-01 -1.59319866e+00 -3.73858899e-01 -7.30521858e-01 -4.04494435e-01 -6.51187599e-01 2.50409514e-01 -5.09573281e-01 -3.56134892e-01 -1.70390666e+00 -2.48231646e-03 -2.55642325e-01 -5.98163784e-01 4.26864684e-01 -7.78151006e-02 -6.95655763e-01 1.81034580e-01 3.28479916e-01 -6.19589448e-01 5.18125832e-01 1.15069115e+00 -2.91772094e-02 -3.26247543e-01 5.67883313e-01 -4.48151797e-01 7.24010468e-01 1.14770257e+00 -3.94219786e-01 -3.73548925e-01 1.33549228e-01 3.49429786e-01 3.58182728e-01 3.99146616e-01 -1.36154723e+00 4.76473600e-01 -1.41675815e-01 6.91631377e-01 -3.09592396e-01 2.28514403e-01 -9.27900493e-01 3.64170283e-01 8.97934079e-01 -7.96531558e-01 2.80117482e-01 5.19599505e-02 9.23800051e-01 -3.41551036e-01 -3.40574145e-01 1.09471753e-01 -3.90574485e-01 -7.74381220e-01 3.99057359e-01 -9.74995077e-01 -4.67521697e-01 1.08835649e+00 -3.26072633e-01 6.32337183e-02 -5.00753999e-01 -1.05563152e+00 1.77207574e-01 -3.28314006e-02 4.71223265e-01 2.90869266e-01 -8.76981497e-01 -4.97887850e-01 -1.51089609e-01 -1.20246470e-01 -8.22521895e-02 9.93127450e-02 8.21313918e-01 -2.65570164e-01 6.25264704e-01 -2.31361270e-01 -3.80071998e-01 -8.18906605e-01 8.73488605e-01 5.90168953e-01 -1.10807300e+00 -4.22065586e-01 1.43460587e-01 -1.07270949e-01 -3.72342944e-01 1.94107041e-01 -6.47899687e-01 -5.83354175e-01 1.62855908e-01 4.92848933e-01 4.59616095e-01 1.04884848e-01 5.86145818e-02 1.75534442e-01 -1.03145987e-01 -1.45661294e-01 2.23494872e-01 1.49579144e+00 2.70753447e-02 -2.25141600e-01 8.28371167e-01 5.51284194e-01 -5.79067647e-01 -1.36462748e+00 -8.80186260e-02 9.83237743e-01 1.50858313e-01 -1.68809411e-03 -6.44842088e-01 -8.63074660e-01 1.15655160e+00 5.95044076e-01 6.16554976e-01 8.88474166e-01 -2.73021400e-01 3.39706004e-01 6.76330566e-01 1.39808103e-01 -1.03051782e+00 7.50025362e-02 9.42904770e-01 5.06520987e-01 -8.48661721e-01 -3.39826435e-01 -3.38568874e-02 -5.14883935e-01 1.40017056e+00 5.94781101e-01 -1.55483320e-01 6.88215137e-01 2.43917137e-01 -3.65966737e-01 -2.65664548e-01 -1.40974748e+00 -1.57219563e-02 -1.33491486e-01 3.64077836e-01 4.82494056e-01 9.92136896e-02 2.82742649e-01 5.89158177e-01 -9.88367200e-02 2.52339423e-01 5.08020043e-01 1.06184769e+00 -3.43849838e-01 -9.60187018e-01 -3.62285912e-01 9.48642790e-01 -7.11353540e-01 4.27955538e-02 -7.89576098e-02 6.90741539e-01 -9.91698727e-02 8.70975256e-01 3.69838387e-01 -2.45853215e-01 4.80042309e-01 4.44107860e-01 1.62063196e-01 -6.02801144e-01 -1.17256916e+00 -1.41702563e-01 2.15280890e-01 -4.95814919e-01 -3.05576831e-01 -7.54857481e-01 -8.77637625e-01 -4.91753787e-01 4.83568311e-02 -1.87408596e-01 4.84179258e-01 1.10455525e+00 2.82227993e-01 1.19042683e+00 2.55936563e-01 -6.52853787e-01 -9.33804214e-01 -1.33118165e+00 -3.99663776e-01 2.52209187e-01 -3.21506746e-02 -1.93507776e-01 7.88228214e-02 -1.16431899e-01]
[8.583480834960938, 5.935131549835205]
fa1b16ae-9d29-4bd5-9605-269b351ccba6
exploration-in-nethack-with-secret-discovery
1711.03087
null
http://arxiv.org/abs/1711.03087v2
http://arxiv.org/pdf/1711.03087v2.pdf
Exploration in NetHack With Secret Discovery
Roguelike games generally feature exploration problems as a critical, yet often repetitive element of gameplay. Automated approaches, however, face challenges in terms of optimality, as well as due to incomplete information, such as from the presence of secret doors. This paper presents an algorithmic approach to exploration of roguelike dungeon environments. Our design aims to minimize exploration time, balancing coverage and discovery of secret areas with resource cost. Our algorithm is based on the concept of occupancy maps popular in robotics, adapted to encourage efficient discovery of secret access points. Through extensive experimentation on NetHack maps we show that this technique is significantly more efficient than simpler greedy approaches and an existing automated player. We further investigate optimized parameterization for the algorithm through a comprehensive data analysis. These results point towards better automation for players as well as heuristics applicable to fully automated gameplay.
['Jonathan C. Campbell', 'Clark Verbrugge']
2017-11-08
null
null
null
null
['nethack']
['playing-games']
[ 1.36630148e-01 2.97061205e-01 1.37714684e-01 1.93427905e-01 -4.95264739e-01 -1.03008032e+00 4.48032796e-01 2.11727962e-01 -8.02289307e-01 1.15157318e+00 -2.28028819e-02 -4.85176831e-01 -5.83746493e-01 -9.53263998e-01 -5.04034340e-01 -4.16971624e-01 -6.64430857e-01 6.89825892e-01 5.19429862e-01 -5.29214323e-01 5.33910751e-01 4.54343140e-01 -1.44298887e+00 -4.44186360e-01 2.62581795e-01 6.30479038e-01 3.65895599e-01 7.58710742e-01 1.07565112e-01 5.39795101e-01 -5.36804795e-01 -5.15144654e-02 6.15213096e-01 -1.80637628e-01 -7.95012176e-01 2.29331423e-02 -6.52787149e-01 -3.31817418e-01 -5.82450405e-02 7.64670849e-01 3.45093310e-01 3.17072302e-01 1.87868729e-01 -1.28343618e+00 2.62479186e-01 5.37848830e-01 -4.80420500e-01 7.21038282e-02 4.86802429e-01 -3.08302958e-02 9.12316442e-01 -2.67668307e-01 8.43231738e-01 5.11505842e-01 5.46083748e-01 4.28183228e-02 -1.21491551e+00 -5.76173007e-01 5.02891503e-02 -2.25589480e-02 -1.56665432e+00 -3.63224715e-01 5.14942765e-01 -2.70072976e-03 8.07318389e-01 6.32017970e-01 1.16411293e+00 5.31197011e-01 1.12502590e-01 5.75891256e-01 1.09995162e+00 -4.75085050e-01 7.03083575e-01 2.29426846e-01 -5.77531338e-01 5.75052440e-01 4.89464730e-01 2.53538340e-01 -4.67812240e-01 -3.61922383e-01 8.95191848e-01 -1.96768075e-01 -2.52286345e-03 -1.16674054e+00 -7.79470086e-01 8.63299072e-01 1.52830154e-01 2.31297225e-01 -3.52757543e-01 4.28301722e-01 8.31432864e-02 3.92831445e-01 1.26574934e-01 1.08130360e+00 -1.51461810e-01 -8.45716178e-01 -8.54697108e-01 7.90852249e-01 1.14779890e+00 1.07058871e+00 8.44369769e-01 -4.94345784e-01 5.51211655e-01 3.37021679e-01 1.55743107e-01 -9.67292339e-02 -2.51245379e-01 -8.97327960e-01 2.80076176e-01 5.35251677e-01 6.11723006e-01 -1.10671711e+00 -6.39581263e-01 -4.07563925e-01 -2.75714993e-01 4.57031429e-01 4.99330282e-01 -2.64213383e-01 -5.10151863e-01 1.42616928e+00 4.35120851e-01 -2.48482719e-01 -7.24699954e-03 7.08795607e-01 2.18323129e-03 1.41248718e-01 -3.22457641e-01 -1.67030200e-01 1.39679384e+00 -3.96506727e-01 -4.54533130e-01 -3.77148151e-01 6.57026172e-01 -6.00762844e-01 6.43470883e-01 5.67252815e-01 -1.08614409e+00 3.82097542e-01 -1.07697201e+00 2.57081807e-01 -5.24131835e-01 -3.59887928e-01 9.45863128e-01 1.10675085e+00 -9.95518148e-01 3.71145070e-01 -7.92536497e-01 -5.72166383e-01 4.37616527e-01 7.00411737e-01 -2.06417099e-01 1.48628324e-01 -8.73815656e-01 8.95956039e-01 7.71621108e-01 -2.06546977e-01 -6.67783797e-01 -2.09487423e-01 -8.01556528e-01 9.39172059e-02 1.18320560e+00 -1.11964360e-01 1.16399789e+00 -1.09986566e-01 -1.31227136e+00 6.15674198e-01 4.00775611e-01 -7.82799840e-01 6.77103996e-01 5.87806702e-02 2.40108669e-01 -1.27551422e-01 2.27288395e-01 4.40193176e-01 1.33373082e-01 -1.10876131e+00 -8.89666975e-01 -1.11327462e-01 6.26728594e-01 6.40948772e-01 -7.90778473e-02 -1.55037925e-01 -2.46330991e-01 -2.52378106e-01 2.28295967e-01 -1.05442619e+00 -1.02411044e+00 -2.05758005e-01 -3.63450766e-01 2.37972990e-01 3.78056079e-01 -2.47741833e-01 1.17964470e+00 -1.91998541e+00 -8.62010196e-02 8.66412580e-01 2.67816305e-01 -2.35864893e-01 2.91722506e-01 9.14318562e-01 4.71902490e-01 1.05668008e-01 -2.81026661e-02 1.43217951e-01 1.99000940e-01 3.50802302e-01 -2.12140247e-01 4.88451391e-01 -3.67561489e-01 7.08984315e-01 -1.11325264e+00 -1.54053748e-01 3.35607588e-01 -2.62337208e-01 -6.99751139e-01 -2.79829621e-01 -2.42378190e-01 1.85748920e-01 -6.15670443e-01 8.06546807e-01 3.85281682e-01 1.35714188e-01 4.59840089e-01 7.76833594e-01 -6.28415406e-01 3.00064892e-01 -1.35232425e+00 1.78263509e+00 -2.17682987e-01 2.83715576e-01 4.94782418e-01 -7.68653452e-01 9.51553762e-01 -6.94166496e-02 6.26922429e-01 -4.45714861e-01 3.85233432e-01 3.26728940e-01 -2.22694531e-01 1.44337509e-02 1.27876389e+00 9.77255125e-03 -5.98261654e-01 7.72344172e-01 -4.63608861e-01 -4.06738281e-01 8.34473372e-02 1.70678034e-01 1.59563851e+00 1.09088786e-01 8.96997154e-01 -5.78533173e-01 -1.36211336e-01 6.62662566e-01 1.87744588e-01 1.17537689e+00 -2.17763290e-01 -1.05399907e-01 5.63622594e-01 -5.64899862e-01 -1.05085516e+00 -7.06973016e-01 1.63669348e-01 9.19137418e-01 7.95569301e-01 -7.88709700e-01 -6.41153872e-01 -2.42904291e-01 -1.64034575e-01 5.69112480e-01 -5.62770784e-01 2.63685822e-01 -3.83636504e-01 -6.29828990e-01 4.42339450e-01 1.26051113e-01 2.67936081e-01 -9.39025521e-01 -1.57400703e+00 6.08494818e-01 -2.51694340e-02 -9.15253222e-01 -3.36815864e-02 6.43191814e-01 -4.40298975e-01 -1.12142575e+00 -2.98970819e-01 -3.50682080e-01 5.23238063e-01 3.94372940e-01 7.54406095e-01 1.54902309e-01 -6.70074701e-01 5.38581431e-01 -5.66829622e-01 -6.74430788e-01 4.89158295e-02 3.62339914e-01 1.37151778e-01 -7.02535152e-01 4.24785733e-01 -6.94271982e-01 -5.06812274e-01 5.40961623e-01 -6.51451409e-01 2.17722468e-02 5.68367541e-01 6.11506760e-01 6.11899912e-01 4.44775432e-01 5.45326509e-02 -6.66621923e-01 7.99645722e-01 -5.84364295e-01 -1.16494000e+00 -4.31108288e-02 -5.28738678e-01 5.01312092e-02 2.13006437e-01 -2.93930527e-02 -4.69374210e-01 3.08728516e-01 1.90548167e-01 -3.66812199e-02 -3.69723828e-04 2.96341956e-01 4.78542410e-02 -5.97141862e-01 7.90668666e-01 1.41656518e-01 -1.87613629e-02 -1.23255052e-01 4.09519613e-01 3.52582753e-01 3.87425274e-01 -7.11427212e-01 9.44564998e-01 7.69442856e-01 1.61145076e-01 -1.01017952e+00 5.77675365e-02 -7.22101986e-01 -2.97044784e-01 -1.85035661e-01 3.23226839e-01 -5.75692177e-01 -1.04071844e+00 -2.85779759e-02 -6.84567869e-01 -3.72259557e-01 -6.41003013e-01 1.93637118e-01 -1.00743079e+00 1.67995602e-01 -5.92864640e-02 -1.33155370e+00 2.67148644e-01 -8.58197808e-01 7.29627967e-01 2.68475860e-01 -4.62448746e-01 -6.79186821e-01 4.11547989e-01 2.57966727e-01 4.53112572e-01 4.96715873e-01 2.28747547e-01 -6.64099574e-01 -1.16987276e+00 -2.34177873e-01 1.70451611e-01 -8.30746710e-01 2.63023209e-02 -7.63470769e-01 -5.81967711e-01 -5.06260216e-01 -1.96998134e-01 -3.83091390e-01 2.98226923e-01 1.30763948e-01 8.03288162e-01 -3.21707606e-01 -6.36083364e-01 4.35988843e-01 1.59449565e+00 2.69720793e-01 7.92761922e-01 1.07797468e+00 -3.15775932e-03 7.83299506e-01 1.16569614e+00 8.74938786e-01 3.34653080e-01 7.20854342e-01 8.51190627e-01 4.89576980e-02 8.82224023e-01 -3.87275666e-01 -2.36633033e-01 -7.20870644e-02 -2.87751526e-01 -1.90305367e-01 -9.76312339e-01 8.19707096e-01 -1.90471601e+00 -8.99033189e-01 4.27091479e-01 2.23054528e+00 4.11009610e-01 3.38385016e-01 5.19856870e-01 2.63098687e-01 4.64216590e-01 6.49436414e-02 -2.77080864e-01 -4.09703940e-01 1.50075585e-01 3.45104098e-01 1.29985869e+00 5.38682997e-01 -9.93860602e-01 1.03219151e+00 6.74997187e+00 8.19337547e-01 -2.41036698e-01 2.03945078e-02 2.18049794e-01 -5.90354860e-01 -2.77099550e-01 4.26434457e-01 -5.81883848e-01 2.26791188e-01 4.23054248e-01 -2.94597477e-01 9.08935547e-01 9.13829744e-01 4.94924327e-03 -7.33681202e-01 -6.09583259e-01 9.21535313e-01 -2.24790722e-01 -1.57563877e+00 -8.12045753e-01 9.35145736e-01 5.02886295e-01 -1.17934033e-01 -1.45868540e-01 -3.82222533e-02 9.69438851e-01 -9.59031641e-01 8.22378278e-01 6.89860806e-02 5.76136410e-01 -1.20960200e+00 5.33135116e-01 5.75983644e-01 -1.36780548e+00 -2.67980397e-01 -3.39476645e-01 -5.89870632e-01 2.78031200e-01 -2.97228787e-02 -1.23624718e+00 4.58892375e-01 8.57044458e-01 -8.22695345e-02 -1.56051433e-03 1.28027880e+00 -8.30696672e-02 -1.12310886e-01 -9.88916755e-01 -6.76168203e-01 6.54978871e-01 -3.28532100e-01 8.14346373e-01 5.85101247e-01 3.05935353e-01 4.31308448e-01 3.79041910e-01 8.05132747e-01 4.20641690e-01 9.71788913e-03 -1.10223234e+00 -1.38719201e-01 7.53372908e-01 1.11626327e+00 -1.32482541e+00 3.77404064e-01 9.61225405e-02 5.95760107e-01 1.07691817e-01 3.31190968e-04 -4.26373690e-01 -4.43009347e-01 6.60285592e-01 4.98817801e-01 3.75066072e-01 -4.22556192e-01 -4.13262546e-01 -5.39543211e-01 1.28468573e-01 -6.27080798e-01 1.66472897e-01 -1.93502173e-01 -2.85816342e-01 2.31787547e-01 1.31839588e-01 -1.18296468e+00 -4.26978439e-01 -2.74310738e-01 -4.15625334e-01 4.01985526e-01 -1.04715490e+00 -9.46897626e-01 -7.17760995e-02 3.42340052e-01 1.85879722e-01 -1.57465532e-01 7.93890715e-01 -1.62095040e-01 1.52912095e-01 2.64813155e-01 2.57039815e-01 -2.63486952e-01 -3.96423265e-02 -1.20007491e+00 5.21874905e-01 8.69821429e-01 1.05390817e-01 4.88859266e-01 9.93711770e-01 -7.16218889e-01 -1.36688054e+00 -4.86993372e-01 3.51788998e-01 -3.65591586e-01 8.01733136e-01 -6.58216596e-01 -1.79017618e-01 5.46338499e-01 -1.36174664e-01 -6.63435757e-01 7.22625494e-01 3.88300270e-01 4.32901800e-01 4.07291353e-01 -1.28039289e+00 8.77762735e-01 1.19651532e+00 -2.93640763e-01 -2.11719468e-01 3.39944027e-02 1.26587436e-01 -8.01741302e-01 -4.53057736e-01 8.15342665e-02 6.55700862e-01 -1.16117501e+00 7.21519887e-01 -9.25263315e-02 -2.53595620e-01 -4.26371902e-01 -2.41861373e-01 -1.11622930e+00 -1.35475188e-01 -1.30740952e+00 2.90915132e-01 7.25492001e-01 1.99982613e-01 -4.45700765e-01 1.52119458e+00 6.43368244e-01 2.28341848e-01 -8.45804811e-01 -1.22048628e+00 -7.30466604e-01 -4.07871544e-01 -5.04057884e-01 6.22088373e-01 6.93294406e-01 4.90105420e-01 -1.73254490e-01 -6.85286880e-01 1.34902447e-01 6.82532609e-01 -7.45062158e-02 1.18028116e+00 -1.16725123e+00 -4.17182714e-01 -2.33931616e-01 -7.21996665e-01 -8.51499856e-01 -5.45832574e-01 -3.58137876e-01 3.31734926e-01 -1.38276279e+00 -1.44608721e-01 -1.04420900e+00 1.73051596e-01 4.84475166e-01 5.22996724e-01 2.71005899e-01 1.54475302e-01 -1.99243538e-02 -8.88074517e-01 2.03858390e-01 8.43298078e-01 2.31376007e-01 -7.13874996e-01 7.13648573e-02 -8.97025764e-01 7.68385947e-01 9.84657764e-01 -5.00117421e-01 -6.37599885e-01 5.37505411e-02 7.08498240e-01 1.65242776e-01 2.31998637e-01 -1.12937903e+00 5.58541656e-01 -3.42522830e-01 -9.34144408e-02 -6.70816422e-01 7.21258700e-01 -1.02922654e+00 5.48482358e-01 6.33437097e-01 6.40227497e-02 -1.30294621e-01 3.25501144e-01 8.72420907e-01 2.47233436e-01 -3.27047020e-01 3.83525163e-01 -5.01798213e-01 -9.20744956e-01 2.27332916e-02 -6.63296044e-01 -1.53095692e-01 1.70397031e+00 -9.35299933e-01 -7.33775795e-02 -6.31501317e-01 -6.79509044e-01 6.66669309e-01 1.03494596e+00 8.96553546e-02 4.24373209e-01 -8.76392901e-01 -2.41027892e-01 2.34735057e-01 1.31362095e-01 1.25171393e-01 4.05536927e-02 6.79865479e-01 -9.19802606e-01 4.10761207e-01 -4.82913256e-01 -3.13310385e-01 -1.33056688e+00 2.68875390e-01 9.77927446e-02 -4.76678401e-01 -6.88413084e-01 7.59860337e-01 -1.48703232e-01 -3.23278427e-01 2.77193397e-01 -3.50845829e-02 4.96059135e-02 -9.24562663e-02 4.27240938e-01 4.02443200e-01 -2.26764947e-01 -1.71460882e-01 -4.28674102e-01 2.69580901e-01 -1.31016180e-01 -5.84101975e-01 1.28875601e+00 -3.05837125e-01 8.68430361e-02 -5.51688299e-02 1.70104623e-01 2.01922163e-01 -1.25002038e+00 3.43212821e-02 2.50376433e-01 -7.65578091e-01 -2.35859111e-01 -4.82154101e-01 -1.97867155e-01 -7.85379335e-02 3.12620372e-01 4.82594252e-01 9.28046584e-01 8.05392803e-04 4.24779207e-01 7.49110579e-01 1.49050677e+00 -1.21751297e+00 -3.76585424e-01 3.25749576e-01 2.58717716e-01 -6.73595369e-01 2.32977033e-01 -3.45812827e-01 -3.85371178e-01 6.68693781e-01 3.98542374e-01 -2.19435051e-01 3.59950900e-01 6.71480656e-01 -2.51299679e-01 -4.98223007e-01 -6.56501949e-01 -3.69964987e-01 -7.13922322e-01 6.28890097e-01 -4.79111701e-01 1.76873118e-01 -3.83367032e-01 4.02035385e-01 -7.35228300e-01 1.43548036e-02 8.41344059e-01 1.61410809e+00 -7.85009682e-01 -1.13775444e+00 -6.68406427e-01 4.76290584e-01 -3.71547371e-01 1.78327337e-01 -5.76303720e-01 1.18633080e+00 1.26793385e-01 9.21866357e-01 4.58716080e-02 -4.57194328e-01 1.31132558e-01 -5.36228180e-01 3.85191798e-01 -5.86706698e-01 -3.95896524e-01 6.48704842e-02 3.99872184e-01 -7.37026930e-01 -9.57000032e-02 -5.94627082e-01 -9.45699632e-01 -4.69290137e-01 -6.04826272e-01 6.46441162e-01 7.73635268e-01 9.25566375e-01 2.18232702e-02 1.64987519e-01 3.52628678e-01 -9.26439404e-01 -2.86112309e-01 -4.06583458e-01 -1.02857554e+00 -4.45390463e-01 1.28620388e-02 -9.16250765e-01 2.36669276e-02 -7.26389527e-01]
[3.777085781097412, 1.5532044172286987]
f22ad567-228d-4e50-9458-ffa2ac66ad5f
toward-negotiable-reinforcement-learning
1701.01302
null
http://arxiv.org/abs/1701.01302v3
http://arxiv.org/pdf/1701.01302v3.pdf
Toward negotiable reinforcement learning: shifting priorities in Pareto optimal sequential decision-making
Existing multi-objective reinforcement learning (MORL) algorithms do not account for objectives that arise from players with differing beliefs. Concretely, consider two players with different beliefs and utility functions who may cooperate to build a machine that takes actions on their behalf. A representation is needed for how much the machine's policy will prioritize each player's interests over time. Assuming the players have reached common knowledge of their situation, this paper derives a recursion that any Pareto optimal policy must satisfy. Two qualitative observations can be made from the recursion: the machine must (1) use each player's own beliefs in evaluating how well an action will serve that player's utility function, and (2) shift the relative priority it assigns to each player's expected utilities over time, by a factor proportional to how well that player's beliefs predict the machine's inputs. Observation (2) represents a substantial divergence from na\"{i}ve linear utility aggregation (as in Harsanyi's utilitarian theorem, and existing MORL algorithms), which is shown here to be inadequate for Pareto optimal sequential decision-making on behalf of players with different beliefs.
['Andrew Critch']
2017-01-05
null
null
null
null
['multi-objective-reinforcement-learning']
['methodology']
[-1.44514605e-01 6.30950034e-01 -5.29168010e-01 -1.03063777e-01 -6.90828979e-01 -6.17411554e-01 4.58199769e-01 3.04400563e-01 -9.82490540e-01 1.11284816e+00 4.18113261e-01 -5.18602848e-01 -5.36299467e-01 -9.55060780e-01 -1.99200250e-02 -8.45481336e-01 7.31570572e-02 9.98983085e-01 1.02378972e-01 -2.02277631e-01 4.70435560e-01 4.38198566e-01 -1.41200137e+00 8.99093971e-02 7.04921246e-01 9.41223264e-01 6.22597225e-02 9.32703197e-01 3.48138958e-02 1.58053923e+00 -7.12045252e-01 -3.50317925e-01 2.26368666e-01 -6.34850919e-01 -1.30498695e+00 1.39673203e-01 -6.85975611e-01 -7.09790587e-01 -9.63975862e-02 1.09949553e+00 1.88907042e-01 3.40644807e-01 1.07683027e+00 -1.46410370e+00 -4.11788851e-01 9.58192885e-01 -3.72837007e-01 -1.92535017e-02 3.86336237e-01 5.16710579e-01 1.25102293e+00 3.01018924e-01 3.07770640e-01 1.47240770e+00 2.38558669e-02 8.01210582e-01 -1.28358853e+00 -1.36179522e-01 3.33205014e-01 9.24356878e-02 -7.86218166e-01 -4.01650190e-01 7.19030619e-01 -2.72004843e-01 7.59387016e-01 4.53264832e-01 8.08205843e-01 3.97908628e-01 3.63205612e-01 1.09768701e+00 1.34132791e+00 -4.62099612e-01 8.23749065e-01 3.01180750e-01 -3.80078584e-01 1.47025526e-01 2.62241840e-01 3.14655036e-01 -2.82509089e-01 -2.72229254e-01 5.96428692e-01 -3.51430267e-01 1.38943970e-01 -5.30650973e-01 -9.06624019e-01 1.01653934e+00 -9.66216028e-02 2.96401083e-01 -8.43942404e-01 4.40790117e-01 2.55158544e-01 7.41188169e-01 9.20498893e-02 8.83145630e-01 -2.22360000e-01 -3.96791995e-01 -4.50936079e-01 4.95739818e-01 9.16293800e-01 5.40250123e-01 5.80461442e-01 1.00231752e-01 -7.12517425e-02 3.63640040e-01 5.61994672e-01 2.71367431e-01 3.20927769e-01 -1.69662905e+00 2.39357352e-01 2.95354307e-01 8.14196050e-01 -5.98732173e-01 -3.02974254e-01 -1.29921451e-01 -2.20114738e-01 9.67019200e-01 5.03030300e-01 -6.50905073e-01 -2.62911111e-01 1.85470760e+00 -6.90639764e-02 -6.18192017e-01 6.16860628e-01 7.05784142e-01 -4.19853404e-02 5.61727166e-01 2.02853441e-01 -6.26589835e-01 1.02125537e+00 -2.21341118e-01 -3.82563770e-01 -6.04247935e-02 6.30324662e-01 -4.05420214e-01 5.99390924e-01 5.03057420e-01 -1.63752079e+00 1.18616834e-01 -7.78517008e-01 4.88216072e-01 2.64194161e-01 -4.94450271e-01 7.08689272e-01 6.85546041e-01 -1.00982702e+00 7.53861010e-01 -6.71875536e-01 -7.69471899e-02 3.32971603e-01 4.11890298e-01 2.58584321e-01 4.72274095e-01 -1.16804063e+00 1.27245986e+00 4.80547994e-01 -7.83541277e-02 -1.32670856e+00 -1.74404513e-02 -3.56406182e-01 4.80126023e-01 8.46896529e-01 -6.55812085e-01 1.90598762e+00 -1.26605642e+00 -1.59422874e+00 5.90984762e-01 1.16168573e-01 -4.26149845e-01 8.99792671e-01 2.76777029e-01 -3.88384126e-02 7.55333900e-02 2.77812570e-01 2.91622609e-01 4.22391534e-01 -1.52922499e+00 -1.50876403e+00 -4.32592869e-01 8.24716806e-01 8.99459660e-01 -4.72029597e-02 2.43215397e-01 3.21332574e-01 6.88437372e-02 -1.95415363e-01 -6.27740324e-01 -5.69959581e-01 -4.15644437e-01 -2.13679135e-01 -5.11227548e-01 2.67797057e-02 4.07626070e-02 1.15655565e+00 -1.53504765e+00 2.18724057e-01 4.21715677e-01 3.81367624e-01 -2.34390482e-01 3.82883586e-02 6.20345771e-01 3.23461384e-01 2.02795386e-01 1.03827633e-01 3.73127982e-02 5.94201446e-01 4.10115451e-01 5.66468798e-02 4.64908451e-01 -1.30596399e-01 5.54673314e-01 -9.29253399e-01 -4.13551658e-01 9.72205326e-02 -5.92047751e-01 -5.94457984e-01 3.48101914e-01 -2.16664076e-01 -1.32024530e-02 -9.47747707e-01 1.98009044e-01 1.28896877e-01 -2.36120597e-02 4.99759257e-01 5.54136455e-01 -3.95678490e-01 2.83066690e-01 -1.38808656e+00 6.10959113e-01 -1.87940598e-01 2.36656144e-01 4.22325611e-01 -1.19643271e+00 3.68543595e-01 5.55494964e-01 7.45660186e-01 -5.03372788e-01 5.14085412e-01 2.89311171e-01 4.18306738e-01 -4.80406344e-01 3.49958688e-01 -9.18865442e-01 -3.51006776e-01 1.14676833e+00 -2.69893557e-01 -1.74299732e-01 1.10375293e-01 2.05688998e-01 9.36457753e-01 -4.70672660e-02 5.66073656e-01 -6.00684106e-01 4.49583590e-01 3.41388613e-01 9.50799584e-01 1.05594444e+00 -5.09705007e-01 -3.57186198e-01 1.21168721e+00 -2.77857512e-01 -1.05559063e+00 -9.44068134e-01 5.80506206e-01 1.41555166e+00 3.07079375e-01 3.92344326e-01 -5.29571533e-01 -4.71548051e-01 1.78777829e-01 1.40956223e+00 -6.83225513e-01 4.08726074e-02 -2.27019250e-01 -4.13701564e-01 4.30181772e-01 4.44655389e-01 2.78296322e-01 -1.15950894e+00 -1.39996076e+00 4.09205556e-01 -6.05971292e-02 -4.39113267e-02 -2.35444337e-01 4.85990494e-01 -4.91307914e-01 -1.06080759e+00 -4.17210281e-01 -1.53392956e-01 3.58825028e-01 5.53888045e-02 8.37899923e-01 -2.49550760e-01 3.77804190e-01 7.34342694e-01 -1.84351414e-01 -8.39279592e-01 -7.22554088e-01 -2.38144949e-01 3.19574207e-01 -3.33694011e-01 4.38476562e-01 -3.66031379e-01 -5.06158173e-01 2.08097383e-01 -6.43817723e-01 1.40721574e-02 3.25790375e-01 5.94310462e-01 8.23630169e-02 7.44445920e-01 6.74026608e-01 -8.96101177e-01 1.21567941e+00 -6.45421803e-01 -6.76935017e-01 4.35417295e-01 -5.85376084e-01 3.32015395e-01 6.35433137e-01 -1.54794678e-01 -1.52518523e+00 -5.32843173e-01 1.83297321e-01 1.66718885e-02 -1.34693801e-01 3.73704612e-01 -4.60109830e-01 5.92766583e-01 4.85372007e-01 1.98779747e-01 3.95117968e-01 -1.68109357e-01 2.36019239e-01 7.16138005e-01 3.87167066e-01 -1.24842548e+00 3.75170767e-01 9.51042026e-02 -5.32217585e-02 -2.52671897e-01 -5.25086820e-01 -2.05793366e-01 -2.22166479e-01 -4.76443559e-01 7.65265882e-01 -6.37688696e-01 -1.69557714e+00 2.58585036e-01 -6.80461287e-01 -4.77125227e-01 -5.85171223e-01 4.71754193e-01 -1.33364773e+00 5.60344346e-02 -2.81551570e-01 -1.73599839e+00 1.32217154e-01 -1.14945996e+00 9.94292200e-02 4.10913140e-01 -3.34121734e-01 -1.08619893e+00 -1.04731560e-01 3.43488634e-01 3.01191092e-01 -2.97394376e-02 1.01104069e+00 -7.88067758e-01 -2.23291501e-01 2.33108670e-01 2.44517788e-01 2.90034503e-01 9.23642516e-02 -1.11798592e-01 -6.14356160e-01 -3.74998122e-01 3.64974976e-01 -6.86226785e-01 2.30393946e-01 4.38169360e-01 6.73858643e-01 -8.09552312e-01 -3.82014699e-02 -3.08636159e-01 1.37633669e+00 9.28283036e-01 3.64860654e-01 5.62910795e-01 -4.76203226e-02 8.44088674e-01 6.79178774e-01 7.30424583e-01 4.73121881e-01 3.22793335e-01 7.42962599e-01 3.31822604e-01 7.87658215e-01 -2.02805981e-01 5.06026387e-01 3.12948115e-02 -5.49676657e-01 -2.78823018e-01 -6.33961320e-01 5.13878703e-01 -2.31905794e+00 -1.61786962e+00 4.31916565e-01 2.28848624e+00 7.72341371e-01 4.04816568e-01 5.70989847e-01 7.26555753e-03 5.87043464e-01 4.11739461e-02 -1.10231173e+00 -9.83883381e-01 1.20149367e-01 -3.09012115e-01 8.29755962e-01 8.22969854e-01 -5.04398406e-01 5.37538230e-01 6.63255405e+00 9.38139319e-01 -3.86317223e-01 2.82652862e-02 8.94894242e-01 -4.99752939e-01 -9.45832610e-01 3.16110879e-01 -4.01237518e-01 3.80150110e-01 8.40692818e-01 -9.38135982e-01 6.84341669e-01 7.90363133e-01 4.99149799e-01 -6.34131968e-01 -1.34759605e+00 4.03773069e-01 -4.76848602e-01 -1.16912282e+00 -3.80200148e-01 5.14349997e-01 5.38590968e-01 -4.38811243e-01 -3.13487232e-01 4.11598325e-01 1.63382876e+00 -1.10789883e+00 1.17221475e+00 5.34909189e-01 5.30060887e-01 -1.48869109e+00 7.43341863e-01 1.06446731e+00 -8.12808275e-01 -6.12101674e-01 -2.47520775e-01 -7.21538305e-01 7.13925660e-02 -1.03239603e-01 -7.00577378e-01 3.47187549e-01 9.97111276e-02 -3.36599857e-01 3.97795349e-01 4.80294168e-01 -2.13252112e-01 4.14864480e-01 -2.36440599e-01 -4.91590470e-01 7.87071645e-01 -5.63867807e-01 4.54935431e-01 4.68730479e-01 1.69420257e-01 6.53618634e-01 3.63022387e-01 7.62645304e-01 1.44466832e-01 -7.69972578e-02 -2.81195611e-01 -1.12155914e-01 7.17108607e-01 7.68804371e-01 -5.53885877e-01 -4.54806596e-01 -2.37697795e-01 4.86969501e-01 2.48871341e-01 2.16549680e-01 -5.19988477e-01 -2.53933638e-01 1.09503686e+00 -4.95273396e-02 -3.34373921e-01 2.27800503e-01 -2.41813883e-01 -7.31418550e-01 -3.68363053e-01 -8.79824102e-01 4.87860143e-01 -6.56898260e-01 -1.02273285e+00 -1.37822581e-02 1.23228543e-01 -7.82952785e-01 -6.58142865e-01 -4.18656617e-01 -8.96126211e-01 1.13088119e+00 -1.12708092e+00 -7.54417360e-01 8.41122389e-01 2.45886818e-01 5.59686840e-01 -3.34151387e-01 6.86036229e-01 -6.23022318e-01 -3.16900253e-01 1.11894138e-01 2.38778934e-01 -1.38032302e-01 -1.52853709e-02 -1.64293253e+00 -3.60573024e-01 5.66784918e-01 -5.31864703e-01 2.72636533e-01 9.93734181e-01 -3.24796200e-01 -1.41472256e+00 -4.01450425e-01 6.76403642e-01 -2.66562879e-01 8.24939549e-01 5.80055416e-01 -4.27147955e-01 6.28921986e-01 4.10336912e-01 -9.10364687e-01 7.43779957e-01 2.20807105e-01 3.18104386e-01 -1.24218334e-02 -1.50743830e+00 8.66348028e-01 7.03962564e-01 -2.43620306e-01 -9.81234848e-01 -6.56849742e-02 9.63854194e-02 8.63986686e-02 -7.60178208e-01 -6.07212484e-02 7.18315363e-01 -1.22136235e+00 4.42532867e-01 -1.20551300e+00 4.10014123e-01 -1.89315349e-01 -1.81714118e-01 -1.67496669e+00 -8.28704894e-01 -7.78615534e-01 4.09544617e-01 7.52946615e-01 1.90833852e-01 -8.82358730e-01 7.95173645e-01 1.27175403e+00 1.52082875e-01 -8.20366442e-01 -1.13945329e+00 -6.01516068e-01 8.33859742e-01 -4.16354924e-01 7.64661968e-01 7.31893718e-01 6.37822270e-01 1.91398323e-01 -3.51590306e-01 -6.15641437e-02 8.44326258e-01 1.31402925e-01 5.30815423e-01 -1.15229213e+00 -5.21623373e-01 -1.07048905e+00 6.25916868e-02 -9.39348578e-01 7.97312409e-02 -4.29617405e-01 1.78153574e-01 -1.71456349e+00 4.02727902e-01 -6.54227555e-01 -5.44790745e-01 6.13764882e-01 6.63088635e-03 -7.17913091e-01 6.37642562e-01 3.26447964e-01 -8.56655836e-01 2.97859907e-01 1.26743412e+00 -3.27637279e-03 -1.40531644e-01 3.11628878e-01 -1.46629679e+00 1.04426885e+00 9.47323024e-01 -2.38508463e-01 -6.89391196e-01 -2.06946477e-01 4.91504073e-01 8.43552411e-01 -9.85770375e-02 -4.82095927e-01 3.78517866e-01 -1.35859942e+00 1.81370050e-01 -1.42814085e-01 -1.08200973e-02 -7.61160910e-01 4.10023451e-01 8.78408194e-01 -8.69101465e-01 -1.13696262e-01 -3.29544932e-01 3.35100114e-01 1.71202481e-01 -6.78241670e-01 9.49973106e-01 -6.28616571e-01 -4.99788791e-01 -1.17315322e-01 -1.22285330e+00 -1.20917540e-02 1.34682381e+00 -2.78451860e-01 -4.13507521e-02 -7.86384344e-01 -7.61047363e-01 1.01448345e+00 5.84820807e-01 -1.65749148e-01 3.45158488e-01 -1.18968618e+00 -7.43479848e-01 -4.28332984e-01 -2.39220545e-01 -2.63094187e-01 1.01239912e-01 4.63537842e-01 -1.84016988e-01 3.70237201e-01 -5.87481260e-01 1.76508397e-01 -1.05445337e+00 4.41365719e-01 6.42545283e-01 -5.01181364e-01 -1.09526306e-01 8.06479752e-01 1.97350949e-01 8.62388760e-02 -5.80534972e-02 1.21483617e-01 -3.00255656e-01 5.22651613e-01 4.50435430e-01 7.14924932e-01 -6.57293797e-01 -4.68530715e-01 -1.90267742e-01 -1.37346759e-02 -3.34608592e-02 -7.34841406e-01 1.28988636e+00 -2.56379783e-01 8.75584260e-02 5.69651186e-01 4.59874302e-01 6.07870659e-03 -1.45748544e+00 -2.31703997e-01 -1.61909074e-01 -7.12303698e-01 -1.03810906e-01 -7.21416056e-01 -5.63390255e-01 5.17302871e-01 -1.37110159e-01 6.94207013e-01 1.14708865e+00 -2.18455419e-01 6.70368373e-02 3.12003940e-01 1.00769782e+00 -1.77358294e+00 -7.66839460e-02 2.71144867e-01 7.05335200e-01 -9.62031364e-01 9.55771133e-02 4.22429234e-01 -1.29883361e+00 9.35149491e-01 4.09354359e-01 3.97145795e-03 1.94155082e-01 1.64611056e-01 -6.47861212e-02 -3.95475663e-02 -1.32177627e+00 -4.58736539e-01 -3.96817654e-01 5.71317732e-01 2.76375264e-02 6.16106868e-01 -4.25357431e-01 3.79263461e-01 -2.40034983e-01 2.66295001e-02 8.80914748e-01 1.02754164e+00 -1.20166290e+00 -1.05743825e+00 -4.90641296e-01 7.09748805e-01 -3.92914146e-01 4.87730175e-01 -2.97188073e-01 6.24389470e-01 1.43732473e-01 1.19859743e+00 1.41636327e-01 1.47275597e-01 2.06031382e-01 -5.36729544e-02 3.63066941e-01 -6.66167796e-01 -2.62324780e-01 3.67075354e-01 3.29022646e-01 -4.68032092e-01 -4.20939386e-01 -9.60566521e-01 -1.40020859e+00 -7.70797729e-01 1.35145724e-01 6.30859256e-01 7.08078966e-02 1.13020265e+00 -3.82772356e-01 4.70619410e-01 8.88492346e-01 -4.05085921e-01 -1.40713859e+00 -6.55598283e-01 -9.63770807e-01 1.66572586e-01 1.97496146e-01 -7.29790509e-01 -5.09994566e-01 -4.58343238e-01]
[4.16765832901001, 2.6879467964172363]
067dbff5-6b14-4b31-a715-fda3d2b7f680
structured-vision-language-pretraining-for
2212.04267
null
https://arxiv.org/abs/2212.04267v2
https://arxiv.org/pdf/2212.04267v2.pdf
Vision and Structured-Language Pretraining for Cross-Modal Food Retrieval
Vision-Language Pretraining (VLP) and Foundation models have been the go-to recipe for achieving SoTA performance on general benchmarks. However, leveraging these powerful techniques for more complex vision-language tasks, such as cooking applications, with more structured input data, is still little investigated. In this work, we propose to leverage these techniques for structured-text based computational cuisine tasks. Our strategy, dubbed VLPCook, first transforms existing image-text pairs to image and structured-text pairs. This allows to pretrain our VLPCook model using VLP objectives adapted to the strutured data of the resulting datasets, then finetuning it on downstream computational cooking tasks. During finetuning, we also enrich the visual encoder, leveraging pretrained foundation models (e.g. CLIP) to provide local and global textual context. VLPCook outperforms current SoTA by a significant margin (+3.3 Recall@1 absolute improvement) on the task of Cross-Modal Food Retrieval on the large Recipe1M dataset. We conduct further experiments on VLP to validate their importance, especially on the Recipe1M+ dataset. Finally, we validate the generalization of the approach to other tasks (i.e, Food Recognition) and domains with structured text such as the Medical domain on the ROCO dataset. The code is available here: https://github.com/mshukor/VLPCook
['Matthieu Cord', 'Nicolas Thome', 'Mustafa Shukor']
2022-12-08
null
null
null
null
['food-recognition']
['computer-vision']
[ 3.8126844e-01 -8.9261524e-02 -2.2779524e-01 -3.9452615e-01 -8.4416741e-01 -7.1249437e-01 6.3662642e-01 3.0353433e-01 -5.9125149e-01 1.5612085e-01 4.1309187e-01 -1.3910040e-01 2.4094589e-01 -6.0710996e-01 -1.2651625e+00 -5.0868177e-01 1.6478612e-01 3.2013464e-01 -1.6550277e-01 -1.6487253e-01 -1.2994841e-01 -2.7017468e-01 -1.3043190e+00 8.4180051e-01 6.7225373e-01 8.0434936e-01 6.9358879e-01 7.9751247e-01 -8.1160292e-02 7.5388098e-01 -4.1455887e-02 -5.6688690e-01 3.1799480e-01 -3.4204116e-01 -5.6112003e-01 9.3875051e-02 8.9603043e-01 -3.6128169e-01 -8.6499296e-02 8.9388895e-01 3.0763611e-01 8.4563427e-02 5.6580222e-01 -1.0496157e+00 -1.2754925e+00 1.0066701e+00 -4.4825095e-01 -1.3407893e-01 4.3377203e-01 4.4155818e-01 1.1647730e+00 -1.1391664e+00 7.2188061e-01 1.3469083e+00 7.4266380e-01 5.4339540e-01 -1.3550243e+00 -5.4701155e-01 1.8382691e-01 2.7318123e-01 -1.1416188e+00 -3.8232523e-01 6.5282661e-01 -4.0460682e-01 1.0884467e+00 6.7497782e-02 5.2310425e-01 1.3498213e+00 -5.5218849e-02 1.4102131e+00 9.5399439e-01 -7.3158711e-01 -1.5857957e-01 1.5040682e-01 3.0019087e-01 9.7314787e-01 1.4905113e-01 2.9532874e-01 -5.3133678e-01 3.4278163e-01 3.5009837e-01 8.0602936e-02 -2.5720549e-01 -5.1126015e-01 -1.5028322e+00 1.1473782e+00 5.5393875e-01 1.5705121e-01 -4.1053137e-01 3.1608626e-01 6.7613047e-01 3.9414629e-01 2.8089285e-01 4.6586573e-01 -5.8101833e-01 2.9514098e-01 -9.7783858e-01 1.7393118e-01 8.0549085e-01 1.1933737e+00 5.9167469e-01 -2.6990595e-01 -5.2078742e-01 9.1603053e-01 4.0055367e-01 8.9509112e-01 2.3479101e-01 -8.1898433e-01 7.1112716e-01 2.4855369e-01 -2.0120846e-01 -8.0777615e-01 -3.0430186e-01 -1.4711457e-01 -6.9850445e-01 -1.0325799e-01 5.3088522e-01 -4.0992260e-02 -1.1900561e+00 1.8031158e+00 2.2470757e-01 -9.0490185e-02 2.0090449e-01 1.0227020e+00 1.1398441e+00 7.3153859e-01 4.8778808e-01 2.7298275e-01 1.6350676e+00 -1.5451149e+00 -5.1422250e-01 -3.4720078e-01 8.3930624e-01 -1.0923448e+00 1.4359612e+00 3.8543305e-01 -8.3748561e-01 -8.0100167e-01 -8.7933630e-01 -4.9072936e-01 -5.3189689e-01 6.2762362e-01 6.6990316e-01 2.4914376e-01 -1.0579352e+00 3.4574655e-01 -8.0688417e-01 -8.8477528e-01 3.5105169e-01 7.4524477e-02 -3.0345380e-01 -6.6414487e-01 -1.1402112e+00 7.9872662e-01 7.7699959e-01 1.1685513e-01 -1.1742350e+00 -9.6895748e-01 -1.1503860e+00 1.0283929e-02 6.8929178e-01 -9.0672225e-01 1.1089607e+00 -1.1305877e+00 -1.1928779e+00 1.2056502e+00 1.5516190e-01 -6.4732438e-01 3.1728402e-01 -4.9733439e-01 -2.4525200e-01 1.6300359e-01 2.0824566e-01 1.3383344e+00 8.4172976e-01 -1.0876001e+00 -5.3706717e-01 -1.6155651e-01 2.0397282e-01 1.6292804e-01 -1.5118039e-01 -2.9550282e-02 -6.7805958e-01 -6.1679775e-01 -5.6834441e-01 -1.0698192e+00 -1.1789351e-01 1.3495517e-01 -4.3717548e-01 -2.9458934e-01 4.0701479e-01 -9.0886325e-01 9.1541886e-01 -2.3860826e+00 1.7886078e-01 -8.2987525e-02 -9.2595413e-02 3.7751809e-01 -7.2778362e-01 4.4728979e-01 4.5943249e-02 -2.2124624e-01 -2.5773802e-01 -3.2126075e-01 2.8347689e-01 1.9139594e-01 -5.1401414e-02 2.0561743e-01 4.3764719e-01 1.3761597e+00 -1.0152465e+00 -5.1976866e-01 4.5278889e-01 4.7593755e-01 -7.2430801e-01 1.9755413e-01 -7.6587260e-01 6.8407640e-02 -3.2760841e-01 6.4071900e-01 4.2638057e-01 -3.4721854e-01 1.2030301e-01 -6.1023581e-01 -9.3461201e-02 -8.1524827e-02 -7.7402562e-01 2.4151244e+00 -5.6069434e-01 4.8692036e-01 1.2801446e-01 -1.0520511e+00 6.0312039e-01 1.4980477e-01 2.9396975e-01 -8.8918400e-01 6.6690616e-02 -3.1678267e-02 -2.6949731e-01 -6.8982023e-01 5.1081693e-01 1.0552406e-02 -2.2502086e-01 1.9369099e-01 4.4112021e-01 4.3267399e-02 6.2282026e-01 3.4539917e-01 8.1784642e-01 6.2236381e-01 2.4538787e-01 -3.8761207e-01 3.8378161e-01 5.6074661e-01 1.9003352e-02 7.5586039e-01 -2.3844089e-01 5.4845142e-01 7.8863129e-02 -2.7650702e-01 -1.1100079e+00 -8.9351583e-01 9.8203391e-02 1.4833032e+00 -8.7318875e-02 -6.2867558e-01 -6.9703889e-01 -9.0403897e-01 1.4302622e-01 9.3566167e-01 -7.5868350e-01 -1.8490881e-01 -5.2391875e-01 -6.3163006e-01 5.9979630e-01 6.3733464e-01 4.0987760e-01 -1.1733326e+00 -6.8143851e-01 2.5714105e-01 -3.0076489e-01 -1.2178071e+00 -8.6017895e-01 3.4577966e-01 -4.5143786e-01 -1.0759928e+00 -7.2613770e-01 -9.2939579e-01 2.3750764e-01 3.6825341e-01 1.6342797e+00 1.8681262e-02 -5.0679743e-01 8.8542193e-01 -7.1463346e-01 -4.1073593e-01 -6.1999983e-01 6.2591955e-02 -5.3488386e-01 -2.5725850e-01 7.5248474e-01 5.5669840e-02 -7.0344478e-01 -6.7569181e-02 -1.1172408e+00 3.9617366e-01 6.5726376e-01 1.1516279e+00 7.0761091e-01 -4.1405872e-01 3.7397760e-01 -8.6077315e-01 3.0304903e-01 -5.8467251e-01 -4.7549751e-01 5.9157306e-01 -4.9186796e-01 2.1284576e-01 5.7924622e-01 -5.7585675e-01 -9.0740263e-01 4.3739963e-01 -2.5387686e-01 -6.5782231e-01 -3.9733166e-01 6.3801074e-01 -1.4764869e-01 4.5122743e-01 5.8171892e-01 2.2652745e-01 -9.2765294e-02 -5.3166014e-01 1.0958269e+00 3.4055403e-01 6.3710201e-01 -7.5845748e-01 4.5861176e-01 2.3740481e-01 -3.1732950e-01 -7.0239037e-01 -9.7584689e-01 -6.2558562e-01 -5.7393622e-01 -1.7679259e-02 1.3439935e+00 -1.2505451e+00 -4.7399390e-01 1.6423607e-01 -9.9006963e-01 -8.2490134e-01 -1.6897802e-01 5.2568853e-01 -4.9008229e-01 2.8744295e-01 -7.3440880e-01 -3.2015920e-01 -6.4020181e-01 -1.1096495e+00 1.2261631e+00 1.1752108e-02 -9.3297817e-02 -1.0712768e+00 4.5818493e-02 5.5487657e-01 8.5256837e-02 2.2335730e-01 9.8549557e-01 -6.1118913e-01 -4.5465520e-01 2.6239488e-01 -3.7258309e-01 3.6699796e-01 -8.6390555e-02 -1.0682087e-01 -8.7519407e-01 -3.9481509e-01 -4.2405957e-01 -7.9246438e-01 1.3796499e+00 4.0942141e-01 7.9881853e-01 -6.5474212e-02 -2.8468075e-01 6.2859261e-01 1.6292942e+00 -1.0637555e-01 3.1748644e-01 3.5913455e-01 9.1677594e-01 5.1176512e-01 9.4106835e-01 1.5350677e-01 6.5417922e-01 7.4277037e-01 5.0656629e-01 -3.9980128e-01 -7.6336586e-01 -5.1056343e-01 7.1173429e-01 6.2959963e-01 3.8155031e-01 -1.9421922e-01 -6.2060624e-01 8.1777632e-01 -1.8394977e+00 -8.7789392e-01 -4.4884045e-02 1.7906450e+00 9.5129573e-01 -2.8819844e-01 1.0479081e-01 -3.9498478e-01 4.2716521e-01 6.7768149e-02 -5.3594011e-01 -4.2954710e-01 7.5963070e-03 2.7099296e-01 5.1160604e-01 3.6131302e-01 -1.4503717e+00 1.2337428e+00 4.6991715e+00 7.6043272e-01 -1.0177881e+00 2.7834246e-01 4.2239639e-01 -1.4691764e-01 -1.4715728e-01 -2.4622214e-01 -8.4534991e-01 3.0825275e-01 8.6597407e-01 4.3393439e-01 7.2883075e-01 8.1723088e-01 8.1791729e-02 -4.2263586e-03 -1.5306867e+00 8.5975468e-01 3.7477508e-01 -1.1650379e+00 1.0623267e-01 -3.0196390e-01 5.9268820e-01 4.0549743e-01 1.2830776e-01 7.6115561e-01 6.2427950e-01 -9.0381908e-01 9.2186069e-01 1.6416883e-01 7.8965461e-01 -3.1378812e-01 4.3303135e-01 1.1187082e-01 -1.2008326e+00 9.4857693e-02 -3.8604364e-01 4.7788003e-01 -4.8135385e-02 5.8257514e-01 -8.8928598e-01 6.6999263e-01 8.8590628e-01 9.9748284e-01 -6.4164418e-01 6.6927212e-01 -2.4010812e-01 6.1122394e-01 -3.7896603e-01 7.6213285e-02 5.8655488e-01 -3.8433537e-02 2.0604944e-01 1.7205448e+00 1.9628406e-01 -1.5211329e-01 4.7406757e-01 1.0271590e+00 -6.1386172e-02 3.4145030e-01 -6.4447021e-01 -2.8402615e-01 -1.5157814e-02 1.4335362e+00 -5.8524859e-01 -3.9729974e-01 -7.9406446e-01 1.1575767e+00 3.8695908e-01 3.0787840e-01 -9.7423482e-01 -1.1202857e-02 4.3225169e-01 -3.5729879e-01 8.1262141e-01 -4.5668695e-02 1.3268958e-01 -1.2912719e+00 -2.1518423e-01 -1.1418850e+00 6.8766600e-01 -8.3668143e-01 -1.4011568e+00 3.3665949e-01 7.6380581e-02 -7.5395781e-01 -1.3277982e-01 -8.4401923e-01 -1.8499826e-01 5.7273281e-01 -1.7311649e+00 -1.8554860e+00 -1.9400887e-01 8.8763088e-01 8.7351161e-01 1.6075821e-01 8.1632364e-01 4.1549298e-01 -3.3827332e-01 6.5671068e-01 5.8622897e-02 3.0659696e-01 9.7405279e-01 -1.2958585e+00 2.5019500e-01 7.8347540e-01 5.8627623e-01 4.7615853e-01 6.6441184e-01 -6.0815018e-01 -1.8061260e+00 -1.4470863e+00 6.6507000e-01 -5.8431774e-01 5.9245437e-01 -5.6996018e-01 -6.7981339e-01 7.5856686e-01 5.9215009e-01 -1.2870580e-01 5.9842902e-01 1.9414811e-01 -7.5119734e-01 -4.7005415e-02 -9.8195946e-01 5.4174995e-01 9.9167907e-01 -4.9335125e-01 -5.9675580e-01 5.1956558e-01 1.0712411e+00 -3.0974668e-01 -9.7590542e-01 2.6550725e-01 6.5983921e-01 -5.7909912e-01 1.2285320e+00 -5.0420618e-01 7.9169446e-01 -1.4205644e-01 -5.2392757e-01 -1.3824465e+00 -3.8038459e-01 -1.0612851e-01 -2.6398739e-03 1.2681239e+00 4.2680550e-01 -7.9536438e-02 4.2238811e-01 2.1776895e-01 -3.2361057e-01 -3.9371297e-01 -3.4936062e-01 -5.6035322e-01 2.1227141e-01 -4.9133065e-01 3.1303105e-01 1.0771204e+00 -1.8688133e-01 5.0143617e-01 -4.2622799e-01 9.7596742e-02 5.7393503e-01 3.7569615e-01 7.7321845e-01 -6.2519121e-01 -6.3412929e-01 -2.2956224e-01 1.5475801e-01 -9.5306164e-01 2.2732463e-01 -1.3950088e+00 3.6505359e-01 -1.4888966e+00 4.7306859e-01 -2.5383425e-01 -4.6094587e-01 8.1711918e-01 -3.6570862e-01 3.2702193e-01 7.7344549e-01 -1.6612019e-01 -8.4108204e-01 2.3119052e-01 1.2350520e+00 -5.8561230e-01 -6.2555782e-02 -5.9879273e-01 -6.5917647e-01 3.5855308e-01 7.3237920e-01 -3.6268356e-01 -5.2008247e-01 -8.5273695e-01 -3.6767583e-02 -1.4169061e-01 6.7706543e-01 -7.9073513e-01 -8.0653429e-02 -9.9335998e-02 3.3349892e-01 -6.0701907e-01 1.8109435e-01 -9.2280042e-01 8.1918754e-02 5.3119379e-01 -6.2153298e-01 -3.4357864e-02 3.8671115e-01 4.8703235e-01 -2.9727036e-02 -2.0744298e-01 4.8821366e-01 -3.5216641e-01 -1.0535686e+00 1.8348859e-01 -3.8735639e-02 1.2975697e-01 7.2845691e-01 2.3836102e-01 -4.0655693e-01 1.3844056e-02 -7.5254107e-01 4.7670788e-01 4.4579616e-01 6.2099069e-01 5.0165963e-01 -1.2438465e+00 -9.6003687e-01 1.8057488e-01 5.6787515e-01 -2.2417165e-01 3.0074701e-01 8.1275272e-01 -3.6595964e-01 7.6226926e-01 -1.0036640e-01 -8.0838305e-01 -1.4316509e+00 1.1265063e+00 6.7508309e-03 -4.9678865e-01 -8.3156037e-01 7.9368562e-01 6.3842303e-01 -3.9843270e-01 2.3279949e-01 -8.0262250e-01 -9.3664132e-02 -2.0523444e-02 5.3282970e-01 -9.0830922e-02 -6.5615959e-02 -4.7392493e-01 -1.8454365e-01 4.9054810e-01 -2.1866126e-01 1.4651318e-01 1.3349550e+00 -1.9788282e-01 1.7896205e-01 8.5591534e-03 1.3077563e+00 -3.4915066e-01 -1.3393766e+00 -3.2753530e-01 1.3051832e-01 -1.2564959e-01 6.7799173e-02 -1.2183830e+00 -1.2012229e+00 1.0103588e+00 8.1796396e-01 -2.4741301e-01 1.2033826e+00 2.5139362e-01 8.9186502e-01 4.4890979e-01 1.7998150e-01 -8.5349226e-01 8.4990047e-02 4.4478503e-01 9.3922991e-01 -1.5126482e+00 -2.5601342e-01 -2.2940457e-01 -8.2439965e-01 9.2333502e-01 4.0259808e-01 -1.6983172e-01 3.8083798e-01 1.6326836e-01 1.8079051e-01 -2.6243562e-01 -8.3182037e-01 -4.8373193e-01 3.9938793e-01 7.0909429e-01 5.0774413e-01 3.7122211e-01 -1.4164796e-01 6.2922168e-01 1.3278697e-01 3.4684339e-01 1.2586226e-01 6.4250338e-01 -1.5613690e-01 -1.0996069e+00 -3.7157205e-01 1.8080640e-01 -3.4327698e-01 -5.5715960e-01 -9.8808713e-02 8.1358045e-01 2.9739031e-01 9.5740169e-01 -2.2853775e-01 -1.4206454e-01 3.9636192e-01 2.1992052e-01 6.5843499e-01 -6.4462191e-01 -9.0444404e-01 2.7291042e-01 2.9272315e-01 -6.1380768e-01 -7.0191389e-01 -7.5787532e-01 -1.0806718e+00 -6.7328088e-02 -9.3503505e-02 -1.9706973e-01 5.8740407e-01 7.4662083e-01 1.8373322e-01 4.3201932e-01 1.9065507e-02 -6.7873633e-01 -5.0287360e-01 -8.1959027e-01 -2.5973788e-01 8.6539906e-01 2.3151456e-01 -4.4041026e-01 1.1708549e-01 5.6138623e-01]
[10.8156156539917, 1.6325703859329224]
40e6ee3b-2ff9-411a-bbc9-7c166c6f06cf
variational-sequential-labelers-for-semi-1
1906.09535
null
https://arxiv.org/abs/1906.09535v1
https://arxiv.org/pdf/1906.09535v1.pdf
Variational Sequential Labelers for Semi-Supervised Learning
We introduce a family of multitask variational methods for semi-supervised sequence labeling. Our model family consists of a latent-variable generative model and a discriminative labeler. The generative models use latent variables to define the conditional probability of a word given its context, drawing inspiration from word prediction objectives commonly used in learning word embeddings. The labeler helps inject discriminative information into the latent space. We explore several latent variable configurations, including ones with hierarchical structure, which enables the model to account for both label-specific and word-specific information. Our models consistently outperform standard sequential baselines on 8 sequence labeling datasets, and improve further with unlabeled data.
['Karen Livescu', 'Mingda Chen', 'Kevin Gimpel', 'Qingming Tang']
2019-06-23
variational-sequential-labelers-for-semi
https://aclanthology.org/D18-1020
https://aclanthology.org/D18-1020.pdf
emnlp-2018-10
['learning-word-embeddings']
['methodology']
[ 3.97829980e-01 -1.16120661e-02 -9.37607110e-01 -5.65068185e-01 -9.36137974e-01 -8.46512973e-01 8.14058661e-01 -1.06457509e-01 -4.80151683e-01 7.36805260e-01 6.00946844e-01 -2.62513280e-01 4.89196658e-01 -3.55586737e-01 -3.90847027e-01 -9.63575006e-01 1.42345011e-01 8.10328543e-01 -1.08814411e-01 2.72832751e-01 9.72945914e-02 -1.90074235e-01 -8.88226569e-01 1.24668635e-01 5.80451787e-01 3.26981723e-01 6.87194228e-01 5.12912512e-01 -5.24125934e-01 5.93822837e-01 -4.07800049e-01 -1.38297677e-01 -2.67579317e-01 -3.99410933e-01 -9.54436779e-01 2.12058917e-01 1.32858947e-01 -1.04371063e-01 -1.38803408e-01 9.74016309e-01 2.13792250e-01 2.27977544e-01 1.23145270e+00 -1.34961510e+00 -1.03018832e+00 7.83613980e-01 -4.67678577e-01 1.55945390e-01 -1.11831212e-02 9.50865075e-02 1.95825934e+00 -9.74703252e-01 8.20119917e-01 1.46141863e+00 5.04606724e-01 6.70307934e-01 -1.75297272e+00 -3.63943845e-01 4.96105075e-01 1.94997594e-01 -1.10100329e+00 -2.32369617e-01 8.13894689e-01 -8.93778145e-01 1.30988932e+00 -1.86038613e-01 3.51746142e-01 1.63723278e+00 2.03235775e-01 1.08759689e+00 6.59346044e-01 -4.77298409e-01 2.69375175e-01 -1.18404351e-01 6.40974224e-01 6.86641514e-01 -1.95526049e-01 -2.08383366e-01 -6.77584171e-01 -5.77166438e-01 6.31970108e-01 1.22126654e-01 -4.49552983e-02 -7.08309650e-01 -1.00624669e+00 1.41337812e+00 -2.38033652e-01 2.13247225e-01 -2.20627487e-01 6.54774070e-01 4.01652843e-01 -8.96219462e-02 9.72014010e-01 1.61789462e-01 -6.73610389e-01 -2.33050436e-01 -1.08189094e+00 1.17510185e-01 6.14658773e-01 9.96888518e-01 1.14495349e+00 2.23102033e-01 -6.77944899e-01 1.02600467e+00 9.65982378e-01 3.93518537e-01 6.83641315e-01 -8.29952240e-01 2.68077761e-01 1.67418808e-01 6.24551028e-02 -1.72938928e-01 -1.68485060e-01 -1.43712014e-01 -1.97089151e-01 -2.41720244e-01 8.35615471e-02 -6.70732479e-05 -1.37265062e+00 2.09034991e+00 2.57080346e-02 4.92302954e-01 -2.38415971e-01 5.15479982e-01 5.87282896e-01 7.55897462e-01 5.93801916e-01 -4.82851774e-01 1.38814163e+00 -1.41854739e+00 -1.14573824e+00 -2.87716210e-01 8.67535532e-01 -6.34457946e-01 1.22863269e+00 3.74426204e-03 -7.67950296e-01 -4.05609906e-01 -7.69796848e-01 -3.41465414e-01 -1.14371784e-01 -2.17481628e-02 5.15053928e-01 6.31625772e-01 -1.26061821e+00 4.03850049e-01 -1.10316908e+00 1.63942277e-02 2.95815408e-01 -7.69983530e-02 -3.88450548e-02 1.37971178e-01 -1.38763881e+00 7.92718887e-01 3.31969261e-01 -4.50935602e-01 -1.29449093e+00 -6.19191587e-01 -1.22342443e+00 -5.04941940e-02 1.22232147e-01 -6.54516697e-01 1.47036576e+00 -6.92883372e-01 -1.37442255e+00 9.59808707e-01 -1.00726879e+00 -3.34790438e-01 1.23414278e-01 -2.62752742e-01 6.01772964e-02 -5.56553006e-02 2.79079705e-01 5.88976383e-01 1.20253491e+00 -1.23103333e+00 -2.19905168e-01 -4.31617200e-02 -2.58253366e-01 2.10386992e-01 -3.75763237e-01 -2.69228388e-02 -5.54140210e-01 -7.62293458e-01 -2.47738436e-01 -9.60612774e-01 -4.41953957e-01 -2.52257526e-01 -4.31279480e-01 -9.15666938e-01 8.02208602e-01 -6.01564050e-01 1.14316952e+00 -1.94371164e+00 5.81240237e-01 -1.09914057e-01 3.97023529e-01 -8.34083706e-02 -3.97420585e-01 5.36283910e-01 -1.23537064e-01 2.69394755e-01 -2.35439211e-01 -1.18375814e+00 3.12975913e-01 6.67369723e-01 -6.06677711e-01 3.92429471e-01 2.58964986e-01 1.37370360e+00 -1.14072418e+00 -7.00278938e-01 1.11780785e-01 4.92857933e-01 -6.02634132e-01 4.39314246e-01 -8.69216025e-01 3.18415076e-01 -5.49719036e-01 3.72184217e-01 2.26151109e-01 -7.22024262e-01 5.55366457e-01 4.37490940e-02 1.84702277e-01 5.58700919e-01 -7.20772326e-01 1.94327343e+00 -2.75567174e-01 6.35624290e-01 -1.62243664e-01 -9.20490921e-01 6.81281149e-01 6.83367670e-01 4.14478183e-01 5.78216910e-02 -9.96538475e-02 -1.71500981e-01 -4.74197686e-01 -5.65897465e-01 6.33667469e-01 -4.25903797e-01 -4.29574959e-02 8.07059407e-01 6.03541791e-01 4.45534550e-02 7.60134682e-02 5.55267811e-01 6.84433639e-01 6.25424385e-01 3.46000850e-01 -3.25408995e-01 1.73851669e-01 -3.37734282e-01 7.07734764e-01 7.17791677e-01 -2.69986540e-01 4.15655077e-01 6.93317473e-01 -1.83013201e-01 -1.11666453e+00 -1.25622511e+00 -6.07641451e-02 1.73948109e+00 -9.41146016e-02 -6.61992073e-01 -4.82914567e-01 -8.32490742e-01 -6.95492979e-03 9.28200841e-01 -8.03575099e-01 6.97414875e-02 -4.63292092e-01 -5.53595901e-01 4.23513234e-01 7.17889965e-01 -4.51832265e-01 -1.07814562e+00 -9.87613723e-02 2.18407974e-01 -4.38204855e-01 -8.95918429e-01 -8.52151513e-01 6.54185772e-01 -7.72591174e-01 -6.03653431e-01 -6.88935757e-01 -8.79967034e-01 3.76175255e-01 1.67109594e-02 1.44305921e+00 2.90942974e-02 -1.61632568e-01 6.79956138e-01 -3.58185172e-01 1.00641191e-01 -5.18341660e-01 3.27936620e-01 1.41958520e-01 -1.91273630e-01 5.90631962e-01 -3.42577904e-01 -2.43160233e-01 4.10457514e-02 -8.45822811e-01 -2.66655162e-02 1.18149906e-01 1.17266476e+00 6.25479400e-01 -8.87997985e-01 2.45799705e-01 -1.10161066e+00 3.83545578e-01 -7.35764503e-01 -3.62044096e-01 4.77726042e-01 -6.68426991e-01 6.15434110e-01 1.49245590e-01 -7.35755146e-01 -9.35457587e-01 5.09116799e-02 -2.03631476e-01 -5.42613089e-01 9.37043224e-03 5.44892192e-01 -1.65172875e-01 5.70459068e-01 2.30038300e-01 4.41134870e-01 -1.46583825e-01 -5.94135165e-01 9.83287394e-01 3.88360888e-01 2.05328818e-02 -6.40445590e-01 6.13813698e-01 2.58539140e-01 -2.73213893e-01 -7.42993355e-01 -9.68032718e-01 -6.30463362e-01 -9.97317314e-01 8.99397284e-02 1.32931507e+00 -9.89788711e-01 -1.83389813e-01 2.23379418e-01 -1.40608931e+00 -6.46693408e-01 -2.94813275e-01 3.56520087e-01 -5.43868661e-01 6.49498403e-01 -9.37634468e-01 -7.84674287e-01 -5.30200191e-02 -1.43457150e+00 1.32414782e+00 -1.94537580e-01 -6.18700147e-01 -1.86682773e+00 6.16513133e-01 1.56967759e-01 2.79267907e-01 -2.57741749e-01 1.08954012e+00 -8.48135471e-01 -3.93866479e-01 2.68298417e-01 1.18488073e-01 3.08044136e-01 2.62971312e-01 -6.34251833e-02 -1.07562625e+00 -4.48129654e-01 -1.22869238e-01 -6.58227146e-01 1.32021153e+00 5.66979229e-01 7.71353185e-01 -4.00827266e-02 -7.96866119e-01 5.19324005e-01 1.22376442e+00 -1.22790121e-01 2.67980099e-01 -1.97516039e-01 1.08272147e+00 3.20324749e-01 1.65788442e-01 3.21523696e-01 4.57250267e-01 7.63265193e-01 1.59264907e-01 6.86204955e-02 7.15145469e-02 -4.56963718e-01 3.82539749e-01 1.19292736e+00 3.60944420e-01 -5.54653645e-01 -8.55275810e-01 7.15764940e-01 -1.86098754e+00 -8.17388833e-01 -1.92535996e-01 1.45784843e+00 1.20877028e+00 -1.49170369e-01 6.98615387e-02 -3.30938339e-01 7.10122645e-01 9.56106663e-01 -6.43113375e-01 -3.29003602e-01 3.58740129e-02 2.61739433e-01 4.15233523e-01 1.05837560e+00 -1.16139138e+00 1.28722513e+00 7.85405731e+00 1.00617266e+00 -6.72143459e-01 6.57823563e-01 2.77136266e-01 -5.49972355e-02 -9.90927398e-01 2.27755651e-01 -1.06403816e+00 4.16220754e-01 9.11321878e-01 -6.00465871e-02 1.05450712e-01 8.78067791e-01 5.40374182e-02 3.67828727e-01 -1.16214907e+00 6.72299743e-01 2.64361620e-01 -1.24564445e+00 1.74254805e-01 2.81178266e-01 8.85197461e-01 2.08862931e-01 1.77723527e-01 3.79132539e-01 9.57791209e-01 -1.17256057e+00 6.42640173e-01 3.32051545e-01 8.06529760e-01 -3.41606528e-01 2.64847040e-01 2.61936396e-01 -1.24297631e+00 4.38844204e-01 -2.63230801e-01 -1.16758943e-01 7.52322614e-01 3.95910323e-01 -7.07956254e-01 4.32711802e-02 8.60595852e-02 1.03222477e+00 -2.21094295e-01 2.77799278e-01 -8.58124673e-01 1.24985576e+00 2.30002776e-01 -1.60196826e-01 4.90382880e-01 -3.95452261e-01 5.45222878e-01 1.54766273e+00 -2.86574692e-01 -2.26585239e-01 6.63766086e-01 9.57785308e-01 -7.97957852e-02 8.92091468e-02 -3.15170646e-01 -3.48527104e-01 4.87306863e-01 1.12812531e+00 -8.00316155e-01 -5.20389259e-01 -3.98378700e-01 1.23251033e+00 7.43539870e-01 8.42859507e-01 -8.31359148e-01 2.23642334e-01 1.01641774e+00 -4.44544822e-01 6.50125206e-01 -7.76922643e-01 -3.85794103e-01 -1.36055756e+00 -4.31677967e-01 -2.90280044e-01 3.85886669e-01 -4.84622449e-01 -1.36324573e+00 4.08072591e-01 3.76808904e-02 -6.30829215e-01 -6.64325237e-01 -6.41682506e-01 -5.15475988e-01 1.23499644e+00 -1.49875045e+00 -1.23328614e+00 3.46713573e-01 2.80085534e-01 9.92389321e-01 -1.94174156e-01 1.15572131e+00 -1.24122649e-01 -4.96234566e-01 5.99646688e-01 2.92444229e-01 7.46312141e-02 6.75425112e-01 -1.64481723e+00 8.45553279e-01 7.60155618e-01 7.03154981e-01 9.45359468e-01 8.74172211e-01 -8.44529212e-01 -1.07006967e+00 -9.11788881e-01 1.23635793e+00 -9.53065753e-01 8.40608716e-01 -8.32747281e-01 -1.00772178e+00 1.22039878e+00 4.19946820e-01 -1.71890780e-01 1.26814902e+00 4.45628166e-01 -7.88392067e-01 9.42516148e-01 -5.59821069e-01 5.08608699e-01 9.82565045e-01 -1.05486107e+00 -7.93239117e-01 3.75218958e-01 1.31644738e+00 -1.05942627e-02 -4.88566071e-01 -7.26475716e-02 5.10338604e-01 -2.00506821e-01 9.25840437e-01 -1.00148857e+00 3.82881582e-01 -4.38143127e-02 -2.65132934e-01 -1.41368580e+00 -7.12447405e-01 -6.28041327e-01 -4.01923418e-01 1.22283065e+00 5.97339511e-01 -4.20854151e-01 7.79000282e-01 5.32929957e-01 -4.00164835e-02 -5.43536603e-01 -9.03438568e-01 -8.29013586e-01 4.50191319e-01 -6.34037256e-01 1.36827603e-01 1.24536002e+00 5.17793857e-02 8.33386302e-01 -7.09298491e-01 -1.85310811e-01 6.87140763e-01 5.39312139e-02 1.78166345e-01 -1.06843388e+00 -7.05234349e-01 -3.06788355e-01 -1.41860144e-02 -1.70539224e+00 8.96078050e-01 -1.22019541e+00 3.12339127e-01 -1.70997202e+00 6.45704925e-01 -2.22774595e-01 -4.12141740e-01 7.38537431e-01 -7.66413212e-01 2.83625931e-01 -1.04802266e-01 3.70129257e-01 -7.67295182e-01 8.65710616e-01 7.83231020e-01 -1.48822665e-01 -7.42854327e-02 -3.03690732e-01 -2.53427088e-01 4.84216332e-01 6.25691354e-01 -7.67383695e-01 -5.97705841e-01 -5.59710920e-01 1.94961727e-01 -2.07506359e-01 5.00481948e-02 1.27184525e-01 -4.82973419e-02 -4.57610130e-01 -2.06598155e-02 -6.01893544e-01 4.83300984e-01 -2.50179201e-01 -1.07976176e-01 2.77220845e-01 -8.38155866e-01 -2.00333327e-01 -2.99681962e-01 7.49005735e-01 5.59749156e-02 -4.98064965e-01 6.38421535e-01 1.89935435e-02 -8.37566853e-01 4.24211234e-01 -7.69109666e-01 4.61235225e-01 6.98443413e-01 1.49547517e-01 -8.90095606e-02 -3.85485739e-01 -9.02203441e-01 4.19613749e-01 4.06660885e-01 6.64853871e-01 3.71402532e-01 -1.52522659e+00 -6.64479673e-01 4.81854975e-02 2.42199898e-01 -3.76150608e-01 -7.21161589e-02 4.04994398e-01 1.20534793e-01 4.13111657e-01 3.64350915e-01 -8.07811499e-01 -1.18079662e+00 7.31318653e-01 1.23079382e-02 -4.85233128e-01 -4.61756974e-01 1.22990382e+00 3.94718260e-01 -3.97361875e-01 3.20855230e-01 -2.44975671e-01 -3.26857507e-01 4.21128839e-01 3.72767836e-01 1.19453862e-01 -6.99625134e-01 -7.22055972e-01 -3.63046110e-01 3.90953958e-01 -1.68874606e-01 -6.20246470e-01 1.09546685e+00 -2.32391477e-01 -8.72791708e-02 1.19790065e+00 1.48810613e+00 -2.88082421e-01 -1.52740097e+00 -7.77845263e-01 3.20337236e-01 -1.11962758e-01 -2.91091502e-02 -2.55801946e-01 -5.68907857e-01 1.16071343e+00 1.39764830e-01 4.91157211e-02 2.33913407e-01 3.68776262e-01 6.82671010e-01 -7.85174370e-02 1.99994281e-01 -1.05043340e+00 5.35724282e-01 1.01808345e+00 3.25982779e-01 -1.05439103e+00 -1.04275234e-01 -2.10087329e-01 -9.35344934e-01 7.88878798e-01 1.83577031e-01 -1.06610380e-01 8.05510402e-01 3.30398083e-01 2.87819207e-01 -1.87110007e-01 -1.04135752e+00 -2.33441129e-01 4.31496739e-01 6.40882671e-01 8.13960195e-01 2.19450906e-01 -4.05936062e-01 5.28655469e-01 1.41474470e-01 -4.30061102e-01 9.34761986e-02 6.98204815e-01 -5.45926154e-01 -1.62208450e+00 9.32170227e-02 3.55801314e-01 -3.99212301e-01 -5.94820082e-01 -1.17052287e-01 1.11654915e-01 -8.29748660e-02 8.80819917e-01 2.72831529e-01 1.54393734e-02 -5.05440116e-01 8.64844263e-01 2.25437641e-01 -1.09173799e+00 -1.66164458e-01 4.20697242e-01 -5.36861271e-02 -4.46632028e-01 -4.54454929e-01 -9.93766963e-01 -1.21787870e+00 4.54087198e-01 -4.90831465e-01 3.22762549e-01 5.68704009e-01 1.42628849e+00 -1.24856889e-01 5.29371321e-01 4.44305837e-01 -8.06327462e-01 -7.13146567e-01 -1.24727190e+00 -6.65029466e-01 3.59244555e-01 3.95968378e-01 -7.13759780e-01 -3.56643111e-01 4.58675832e-01]
[11.614514350891113, 9.22871208190918]
3bad05d8-68a2-43e3-b289-db2a2f5c58ab
exploiting-context-information-for-generic
2207.01050
null
https://arxiv.org/abs/2207.01050v1
https://arxiv.org/pdf/2207.01050v1.pdf
Exploiting Context Information for Generic Event Boundary Captioning
Generic Event Boundary Captioning (GEBC) aims to generate three sentences describing the status change for a given time boundary. Previous methods only process the information of a single boundary at a time, which lacks utilization of video context information. To tackle this issue, we design a model that directly takes the whole video as input and generates captions for all boundaries parallelly. The model could learn the context information for each time boundary by modeling the boundary-boundary interactions. Experiments demonstrate the effectiveness of context information. The proposed method achieved a 72.84 score on the test set, and we reached the $2^{nd}$ place in this challenge. Our code is available at: \url{https://github.com/zjr2000/Context-GEBC}
['Ping Luo', 'Ran Cheng', 'Feng Zheng', 'Teng Wang', 'Jinrui Zhang']
2022-07-03
null
null
null
null
['boundary-captioning']
['computer-vision']
[ 2.75017262e-01 3.28176990e-02 -1.51703998e-01 -4.75960761e-01 -9.83876109e-01 -4.32927668e-01 3.98110986e-01 -1.20543800e-02 -2.64176637e-01 8.61204922e-01 5.19856811e-01 -1.97879791e-01 4.47280794e-01 -5.73209405e-01 -9.05068636e-01 -3.98271114e-01 3.53812799e-02 1.55714989e-01 4.75114703e-01 -1.47601053e-01 1.46388486e-01 -9.86986384e-02 -1.37120950e+00 7.88165450e-01 5.03554523e-01 8.40927422e-01 5.44128478e-01 8.03939939e-01 -1.30476519e-01 7.45077789e-01 -7.61774123e-01 -3.04696381e-01 -1.53365582e-01 -7.53472090e-01 -7.78626263e-01 -1.22336179e-01 2.38781333e-01 -3.26585948e-01 -6.46303475e-01 8.39151800e-01 4.82010275e-01 8.68965611e-02 3.38010192e-01 -1.38398623e+00 -5.01089454e-01 7.02274323e-01 -6.36508465e-01 6.84875965e-01 7.00575173e-01 1.04952015e-01 1.06468773e+00 -9.41396177e-01 8.09036136e-01 8.30068469e-01 1.69853240e-01 8.76393676e-01 -6.34600997e-01 -7.75707006e-01 6.65976286e-01 5.93393862e-01 -1.50995493e+00 -3.19539905e-01 8.00639629e-01 -3.43604177e-01 7.39017904e-01 4.04476941e-01 5.34490049e-01 1.61116886e+00 -1.17901064e-01 1.08979154e+00 6.08996809e-01 -2.47109950e-01 -1.39729129e-02 -4.56929088e-01 3.03393334e-01 3.73235852e-01 4.69749980e-02 -2.61197865e-01 -6.41180456e-01 1.55691907e-01 6.77076399e-01 -2.98468620e-01 -5.76704025e-01 3.46611887e-01 -1.39059162e+00 3.16895872e-01 2.46092808e-02 2.55163193e-01 -2.82560915e-01 3.31294984e-01 5.02170742e-01 4.26255800e-02 2.19423681e-01 1.60130430e-02 -3.34280819e-01 -6.97604477e-01 -6.81073964e-01 3.16262186e-01 7.18635678e-01 1.35159302e+00 5.23494542e-01 -1.71664163e-01 -7.06730425e-01 5.62690914e-01 1.22022092e-01 3.95997584e-01 3.56090039e-01 -1.02854323e+00 8.85862470e-01 2.99290210e-01 4.91446733e-01 -8.72387409e-01 -1.79110199e-01 -1.18576745e-02 -4.41364169e-01 -3.96104425e-01 3.24272215e-01 -5.29532850e-01 -9.52666283e-01 2.06555557e+00 3.16550493e-01 7.71785498e-01 -2.74600442e-02 1.02630460e+00 1.17261648e+00 1.12299025e+00 3.09545726e-01 -3.39008003e-01 1.53366673e+00 -1.07710195e+00 -9.98009324e-01 -3.08199048e-01 4.07916039e-01 -7.20955968e-01 1.12015343e+00 2.90807664e-01 -1.11646175e+00 -5.38739383e-01 -1.10948873e+00 -7.76276663e-02 7.97933638e-02 1.76191032e-01 1.59922108e-01 7.67809227e-02 -1.07875502e+00 2.38822222e-01 -7.47796416e-01 -1.89428955e-01 4.64024581e-02 7.26738423e-02 -1.42328203e-01 -4.31923382e-02 -1.61304426e+00 4.07523453e-01 3.69708180e-01 2.47093484e-01 -1.07866192e+00 -1.91815376e-01 -9.26370382e-01 7.83589631e-02 6.27001584e-01 -4.57073718e-01 1.67396557e+00 -1.10509813e+00 -1.23215365e+00 6.76142514e-01 -5.64992130e-01 -2.02428550e-01 5.42955995e-01 -4.74569947e-01 -5.59343874e-01 4.06888455e-01 1.33493319e-01 8.50219548e-01 4.61847007e-01 -1.18263960e+00 -7.04375148e-01 1.00207686e-01 2.30756387e-01 4.19065684e-01 1.65717423e-01 4.20240372e-01 -1.13623846e+00 -7.41706848e-01 -9.59791318e-02 -7.10058570e-01 9.92735289e-03 -5.48746407e-01 -3.59991401e-01 -2.24204853e-01 7.46228635e-01 -1.06083596e+00 1.74045897e+00 -2.20627761e+00 -9.03837383e-03 -3.31084698e-01 -3.21834058e-01 2.47065574e-01 -2.20375270e-01 6.09160185e-01 -4.13389117e-01 4.19667870e-01 -1.21022314e-01 -4.15517569e-01 -1.32150397e-01 3.80079783e-02 -3.27249944e-01 9.38569903e-02 2.73833603e-01 8.69204521e-01 -1.03741479e+00 -5.89630723e-01 -1.36797130e-01 5.50850809e-01 -3.96951944e-01 3.95987272e-01 -6.31652772e-01 4.64494795e-01 -5.17633975e-01 5.04170656e-01 4.48056191e-01 -3.63820523e-01 3.51123363e-01 -2.26186812e-02 3.06079201e-02 1.73179224e-01 -1.27859390e+00 1.74513483e+00 -4.23562229e-02 6.96363091e-01 -1.87396795e-01 -5.62200248e-01 6.99619830e-01 7.87384391e-01 3.46448153e-01 -7.02322721e-01 1.46525472e-01 -7.44353011e-02 -2.12484211e-01 -8.04112315e-01 4.24134105e-01 2.18332916e-01 -4.50413287e-01 3.46382797e-01 -2.91630626e-01 3.67957085e-01 2.85616964e-01 2.22713158e-01 1.17741251e+00 5.34287453e-01 1.53533846e-01 3.48510295e-01 4.03576821e-01 -2.31858250e-02 9.32389319e-01 6.59353673e-01 -5.05706429e-01 1.16679919e+00 8.60404551e-01 -3.11435819e-01 -9.14625108e-01 -8.52147818e-01 4.61537033e-01 8.71040225e-01 6.16794229e-01 -7.34167635e-01 -9.13960040e-01 -6.68970585e-01 -5.91147900e-01 8.86400878e-01 -6.63425565e-01 -7.51174167e-02 -9.34636354e-01 -3.72000098e-01 3.40168536e-01 5.49488068e-01 6.15552425e-01 -1.40494812e+00 -4.81604427e-01 1.35388330e-01 -1.02910185e+00 -1.30192316e+00 -8.70000541e-01 -4.82779264e-01 -3.99922013e-01 -9.70566809e-01 -6.68120444e-01 -8.71282816e-01 5.51415563e-01 1.21683680e-01 9.46783781e-01 1.55535534e-01 -1.99778974e-02 1.97008729e-01 -6.44723535e-01 -2.48549089e-01 -3.21249783e-01 -1.04558378e-01 -3.01743925e-01 6.70612529e-02 3.52898926e-01 -3.71007740e-01 -7.76182115e-01 5.47541380e-01 -8.36650193e-01 6.57225728e-01 1.81160092e-01 5.39287746e-01 6.05163753e-01 -2.00200528e-01 7.43656158e-01 -6.45201027e-01 5.00959158e-01 -6.62845194e-01 -3.02246898e-01 3.23056936e-01 6.25771657e-02 -3.91050935e-01 3.30740273e-01 -4.15697366e-01 -1.22607195e+00 -5.46595007e-02 -2.19901308e-01 -3.97539079e-01 -4.52799529e-01 5.24658322e-01 -4.36649293e-01 8.92427981e-01 2.61579156e-01 2.62919843e-01 -3.78291517e-01 -3.98752898e-01 2.67303772e-02 7.27404416e-01 6.77358449e-01 -7.66227603e-01 2.40595356e-01 2.80232430e-01 -5.48705935e-01 -5.90205431e-01 -9.34486270e-01 -4.93159622e-01 -2.26081476e-01 -4.72024292e-01 1.03589630e+00 -1.24644637e+00 -4.75628495e-01 5.15059590e-01 -1.36708176e+00 -5.77266812e-01 -2.43661888e-02 3.50745350e-01 -6.80633903e-01 4.56897229e-01 -6.92316771e-01 -5.59607208e-01 -3.00519079e-01 -1.01133478e+00 9.42395151e-01 5.66988289e-01 -2.79806018e-01 -5.59220195e-01 -1.47559956e-01 3.96705329e-01 -4.20681275e-02 3.88438582e-01 4.27379251e-01 -6.03927791e-01 -6.36053026e-01 -1.70529813e-01 -1.87818840e-01 1.61790431e-01 6.07694648e-02 2.38674939e-01 -7.99324751e-01 -5.85545264e-02 -1.21877432e-01 -1.48877623e-02 6.00176871e-01 2.21292719e-01 1.20763183e+00 -2.01467767e-01 -3.61144125e-01 1.27005309e-01 1.23628855e+00 6.73083782e-01 9.36053991e-01 1.86328307e-01 5.73029041e-01 2.38720715e-01 9.43994761e-01 6.17656529e-01 5.85498035e-01 8.30703437e-01 3.05439740e-01 2.11667106e-01 -2.42072731e-01 -3.61443758e-01 5.31532884e-01 5.70453227e-01 -6.43364489e-02 -7.15716839e-01 -1.05975688e+00 7.94623733e-01 -2.09837389e+00 -1.18050933e+00 -1.79847524e-01 1.93081057e+00 1.00525022e+00 4.06615734e-01 -4.18492965e-02 -1.42905265e-01 1.28124726e+00 2.85284340e-01 -3.68091881e-01 -2.51683116e-01 5.95269054e-02 -2.28087619e-01 7.52889290e-02 5.80915213e-01 -1.09806776e+00 1.04532361e+00 5.84927034e+00 6.64008379e-01 -9.95968282e-01 1.51592761e-01 8.18732500e-01 -2.78497338e-01 -2.25858033e-01 8.61414000e-02 -1.02138638e+00 8.46917272e-01 1.14720178e+00 -2.42359132e-01 1.24787137e-01 3.85275632e-01 6.82973444e-01 -2.98737198e-01 -1.08748436e+00 9.85142112e-01 1.26736462e-01 -1.17558336e+00 2.00814068e-01 -3.80380094e-01 4.82925802e-01 -9.84580293e-02 -2.29327187e-01 4.37454879e-01 -7.93699846e-02 -6.91524506e-01 9.74278986e-01 5.58650255e-01 9.23151731e-01 -4.16647881e-01 5.36329746e-01 2.89195716e-01 -1.27726567e+00 4.94150072e-02 8.64042640e-02 -3.13579142e-01 6.18051589e-01 3.32053274e-01 -8.08476031e-01 7.62508035e-01 6.83614254e-01 5.50536096e-01 -4.08903986e-01 1.20643604e+00 -6.15837157e-01 9.06037509e-01 -2.16120705e-01 1.31046534e-01 5.31523675e-02 -1.69775612e-03 7.04591095e-01 1.21993279e+00 5.53001881e-01 5.28991520e-01 1.24093242e-01 4.93937373e-01 -1.31148219e-01 -1.96911395e-02 -4.02830809e-01 7.47795030e-02 6.43354177e-01 9.72932816e-01 -8.02029788e-01 -4.74273503e-01 -4.65492070e-01 1.09337127e+00 1.26181275e-01 5.69639564e-01 -1.58236039e+00 -4.58855003e-01 5.17164052e-01 1.41987324e-01 4.86978173e-01 -3.53597790e-01 -3.65877263e-02 -1.25740671e+00 3.74492526e-01 -7.11472332e-01 7.00580657e-01 -1.36725056e+00 -7.19257772e-01 5.53465962e-01 3.20817828e-01 -1.23912740e+00 -6.17974214e-02 -2.12508559e-01 -8.68798375e-01 6.87965453e-01 -1.33840406e+00 -9.53883648e-01 -5.61103880e-01 3.18130493e-01 8.44884455e-01 3.74543041e-01 4.50301230e-01 4.82814759e-01 -8.48700106e-01 5.30534744e-01 -4.38429981e-01 3.90351683e-01 8.51239860e-01 -9.78756547e-01 6.83674276e-01 1.07043481e+00 -2.92475037e-02 3.37873340e-01 8.31653833e-01 -8.84660065e-01 -8.35452259e-01 -9.53547895e-01 1.22718382e+00 -3.92768562e-01 4.83263671e-01 -3.57934207e-01 -8.74046385e-01 9.19758797e-01 4.48735148e-01 -1.34544060e-01 6.60012782e-01 -2.81697094e-01 -5.72404452e-02 9.01758447e-02 -7.07116604e-01 9.94463801e-01 1.34379995e+00 -3.12165320e-01 -6.67498291e-01 1.16536446e-01 1.20784819e+00 -7.52441287e-01 -5.78022242e-01 3.81989449e-01 1.44483224e-01 -7.14201629e-01 6.90087616e-01 -3.95658404e-01 6.57818496e-01 -5.27442753e-01 -1.19852103e-01 -8.93238783e-01 -4.43819445e-03 -8.21447551e-01 -2.42119506e-01 1.50014699e+00 5.41918933e-01 -2.65062183e-01 6.56369865e-01 6.66103363e-01 -3.69110942e-01 -7.18455732e-01 -8.93922389e-01 -4.93685693e-01 -3.28023314e-01 -5.16866863e-01 6.30798519e-01 7.55391002e-01 1.88514158e-01 3.17086369e-01 -4.20639783e-01 3.97262275e-01 1.27434582e-01 1.62493721e-01 4.30904746e-01 -6.45810425e-01 -1.73453361e-01 -2.04526439e-01 -1.13205336e-01 -1.20154154e+00 2.21512336e-02 -5.91118157e-01 1.02672957e-01 -1.74820650e+00 2.72322118e-01 -1.40081719e-01 -2.78563023e-01 6.77301466e-01 -5.50240397e-01 2.13924557e-01 2.76893437e-01 3.19982134e-02 -9.45005417e-01 4.21319395e-01 1.49392557e+00 6.82960451e-02 -4.90490533e-02 -1.33734882e-01 -7.74049640e-01 4.58893567e-01 1.08824372e+00 -2.61061281e-01 -5.86136043e-01 -6.36278927e-01 2.34739929e-01 5.46100855e-01 2.34159961e-01 -1.16346669e+00 1.94789574e-01 -4.22548950e-01 2.48327971e-01 -5.54070115e-01 4.56789106e-01 -5.51721692e-01 4.62691128e-01 7.66017959e-02 -3.30926806e-01 1.96716443e-01 4.11019564e-01 5.33291996e-01 -3.64589930e-01 -3.06384057e-01 2.01040655e-01 -1.59200966e-01 -1.02749372e+00 3.22289407e-01 -3.36164147e-01 2.94508040e-01 1.43318987e+00 -6.67927042e-02 -4.79040951e-01 -5.99718869e-01 -1.11627758e+00 5.77465773e-01 3.98458928e-01 6.43329620e-01 5.97445309e-01 -1.31849670e+00 -8.70394289e-01 -6.55334070e-02 4.71886396e-02 2.34186471e-01 6.12896204e-01 4.91126686e-01 -5.34594357e-01 1.50975168e-01 -8.51785094e-02 -4.57929760e-01 -1.41302907e+00 4.20711637e-01 3.43814433e-01 -1.09703563e-01 -6.52690530e-01 8.70928764e-01 2.75865078e-01 2.80809492e-01 2.28620097e-01 -1.42427698e-01 -3.97339851e-01 1.36268124e-01 7.96163440e-01 -1.20439185e-02 -2.25554541e-01 -6.60420418e-01 -2.20669925e-01 4.03346479e-01 -2.05864325e-01 -3.14597219e-01 9.09013808e-01 -3.65019411e-01 2.84338295e-01 4.49115932e-01 9.94162321e-01 -2.79376864e-01 -1.62587988e+00 9.64386854e-04 7.72897154e-02 -3.46491605e-01 -5.29471695e-01 -8.94041359e-01 -7.46707976e-01 7.28439689e-01 1.80201769e-01 -6.57753600e-03 1.09347415e+00 -1.65255205e-03 8.82074296e-01 -1.89376120e-02 3.07641894e-01 -1.04441810e+00 2.07509294e-01 6.93831027e-01 1.05305111e+00 -9.80916560e-01 -3.12198460e-01 -5.48084497e-01 -9.69302237e-01 5.92819810e-01 9.68994737e-01 1.87184900e-01 4.55018312e-01 9.70809534e-02 2.38731533e-01 -1.00094303e-02 -1.08129013e+00 -4.42072339e-02 -1.31018594e-01 2.79172659e-01 4.19575840e-01 9.83816013e-02 -5.81245601e-01 9.26635146e-01 1.77864134e-01 3.20843816e-01 8.75632405e-01 1.18444526e+00 -2.71110177e-01 -1.15296543e+00 -1.73376575e-01 2.74707377e-02 -6.65228426e-01 -7.45438784e-02 -1.75071880e-01 5.88957489e-01 -4.94831540e-02 1.00900221e+00 1.72938779e-01 -3.96420449e-01 5.03297865e-01 4.23174828e-01 1.77022502e-01 -6.88135862e-01 -4.21743125e-01 5.90209328e-02 3.84994537e-01 -6.67090356e-01 -2.22840682e-01 -8.15123200e-01 -1.60855746e+00 -6.10271469e-02 1.93333045e-01 4.29464221e-01 2.95567542e-01 5.13456464e-01 6.26841843e-01 6.75363779e-01 5.66063762e-01 -8.36238623e-01 -1.52936324e-01 -9.33774412e-01 -2.34577745e-01 6.45507216e-01 1.23678423e-01 -4.75716263e-01 -2.98122585e-01 4.47745770e-01]
[10.386734962463379, 0.7050961256027222]
a45806f4-8254-4418-903b-1ea1d9695208
raidionics-an-open-software-for-pre-and
2305.14351
null
https://arxiv.org/abs/2305.14351v1
https://arxiv.org/pdf/2305.14351v1.pdf
Raidionics: an open software for pre- and postoperative central nervous system tumor segmentation and standardized reporting
For patients suffering from central nervous system tumors, prognosis estimation, treatment decisions, and postoperative assessments are made from the analysis of a set of magnetic resonance (MR) scans. Currently, the lack of open tools for standardized and automatic tumor segmentation and generation of clinical reports, incorporating relevant tumor characteristics, leads to potential risks from inherent decisions' subjectivity. To tackle this problem, the proposed Raidionics open-source software has been developed, offering both a user-friendly graphical user interface and stable processing backend. The software includes preoperative segmentation models for each of the most common tumor types (i.e., glioblastomas, lower grade gliomas, meningiomas, and metastases), together with one early postoperative glioblastoma segmentation model. Preoperative segmentation performances were quite homogeneous across the four different brain tumor types, with an average Dice around 85% and patient-wise recall and precision around 95%. Postoperatively, performances were lower with an average Dice of 41%. Overall, the generation of a standardized clinical report, including the tumor segmentation and features computation, requires about ten minutes on a regular laptop. The proposed Raidionics software is the first open solution enabling an easy use of state-of-the-art segmentation models for all major tumor types, including preoperative and postsurgical standardized reports.
['Ingerid Reinertsen', 'Ole Solheim', 'André Pedersen', 'Ragnhild Holden Helland', 'Valeria Gaitan', 'Demah Alsinan', 'David Bouget']
2023-04-28
null
null
null
null
['tumor-segmentation']
['computer-vision']
[-6.08306937e-02 3.69297087e-01 -1.28216997e-01 -2.61803597e-01 -1.01841879e+00 -2.58145422e-01 3.78843129e-01 8.36398125e-01 -7.42137432e-01 8.69594812e-01 4.55287099e-02 -5.13125896e-01 -2.12000847e-01 -6.60558939e-01 7.51851425e-02 -8.18947077e-01 -1.71852320e-01 9.02516127e-01 3.01845104e-01 1.40719548e-01 2.11745992e-01 5.64857721e-01 -1.12533581e+00 9.25275162e-02 1.23702860e+00 1.26751566e+00 8.47273648e-01 5.82568645e-01 -3.19603205e-01 4.41259563e-01 -4.20597821e-01 -2.99346358e-01 -9.82614011e-02 1.20189175e-01 -7.36774147e-01 7.53846392e-02 6.48598373e-02 -6.68807924e-02 1.52664140e-01 1.10625112e+00 4.35695291e-01 -2.72197604e-01 7.74516106e-01 -8.97272825e-01 -1.52737806e-02 5.36735415e-01 -3.26256901e-01 1.34210274e-01 2.60438114e-01 3.08584869e-01 3.99491698e-01 -6.30181849e-01 8.61382902e-01 2.34970436e-01 7.29010403e-01 3.33270520e-01 -7.97289193e-01 -5.31299829e-01 4.77645546e-02 -8.43401700e-02 -1.42228818e+00 1.87731683e-02 -5.67982979e-02 -1.00955009e+00 7.71090508e-01 4.17074442e-01 1.01412547e+00 6.64705515e-01 1.02363682e+00 3.39273512e-01 1.24405551e+00 -3.88416387e-02 4.04846966e-01 2.94030786e-01 5.55475891e-01 9.22379911e-01 6.41122878e-01 -2.25603774e-01 -3.86451147e-02 -1.17827334e-01 4.45972264e-01 4.70349193e-02 -4.81906176e-01 -2.48986393e-01 -1.43045497e+00 5.04948735e-01 3.78742158e-01 6.37076378e-01 -6.16258025e-01 -3.67041618e-01 7.23746777e-01 -3.98664773e-02 5.52276671e-01 3.63158099e-02 -2.94560224e-01 -1.01122195e-02 -1.14567637e+00 -3.51495929e-02 7.22460091e-01 9.51136053e-01 2.25699067e-01 -4.15660232e-01 -1.32551536e-01 5.30735433e-01 1.85409233e-01 1.74335346e-01 1.35039604e+00 -1.69345304e-01 3.78045112e-01 5.25497437e-01 -1.44586623e-01 -7.77332783e-01 -1.16289425e+00 -7.31944084e-01 -1.15842295e+00 1.71924710e-01 3.82585168e-01 -1.08317174e-01 -1.14162517e+00 1.33254409e+00 1.99673444e-01 -2.90109932e-01 3.91236395e-02 7.50747561e-01 1.02297080e+00 -1.07900217e-01 2.84900993e-01 -5.28234541e-01 1.65198851e+00 -8.71507943e-01 -8.19342613e-01 6.34198412e-02 1.26996601e+00 -6.33429170e-01 7.36207068e-01 3.70661974e-01 -9.96674061e-01 -4.81679849e-03 -8.05502713e-01 2.58026510e-01 -5.90453744e-01 2.92482853e-01 8.43656480e-01 7.89741755e-01 -1.29401886e+00 4.33323950e-01 -1.16339922e+00 -6.25606894e-01 5.54499745e-01 5.30301630e-01 -8.48203301e-01 1.07743479e-01 -7.32309163e-01 1.36016059e+00 4.98132765e-01 -3.27517465e-02 -7.47827709e-01 -1.03112912e+00 -5.83100080e-01 -1.93105832e-01 1.51172489e-01 -9.98552978e-01 9.97628093e-01 -6.19891882e-01 -1.40408778e+00 1.25322378e+00 2.40018647e-02 -5.58145344e-01 8.60996127e-01 3.08553368e-01 -2.98328072e-01 1.73409462e-01 2.57706970e-01 3.03759426e-01 2.79038459e-01 -7.84386933e-01 -7.01351285e-01 -6.74529076e-01 -5.93906403e-01 2.51101196e-01 -5.38801327e-02 1.03199678e-02 -1.92622676e-01 -4.79410052e-01 3.37475717e-01 -9.05099452e-01 -7.24058449e-01 -1.00729510e-01 -3.21803004e-01 1.82265520e-01 3.19380164e-01 -9.79406178e-01 1.20048666e+00 -1.83946013e+00 -4.29851227e-02 3.90626669e-01 5.63769221e-01 -8.79646912e-02 5.56410432e-01 -3.00238878e-01 -2.80534983e-01 4.19371985e-02 -4.24291492e-01 -2.65532762e-01 -3.77815872e-01 -2.10885495e-01 4.62874889e-01 5.93886375e-01 -3.74908030e-01 5.75816929e-01 -8.44254851e-01 -6.21497989e-01 3.30381185e-01 1.16980135e-01 -2.07982451e-01 4.35477570e-02 2.68374085e-01 5.48280180e-01 -2.52887964e-01 5.23504019e-01 5.25176048e-01 -2.04979002e-01 -3.48712988e-02 4.94074523e-02 -2.74157822e-01 -4.88475226e-02 -7.34897614e-01 2.02377772e+00 -6.42382503e-01 3.20299238e-01 2.17334121e-01 -7.17918992e-01 9.04710412e-01 6.05101943e-01 9.77566123e-01 -3.19620937e-01 4.94882941e-01 6.65531993e-01 7.99596384e-02 -5.38268447e-01 7.83042535e-02 -2.85216719e-01 2.34068722e-01 4.18069363e-01 1.00188896e-01 -5.32249451e-01 3.73919338e-01 1.47616625e-01 1.26362669e+00 -2.67406791e-01 6.99685395e-01 -5.75970829e-01 6.07681990e-01 3.24287117e-01 3.32502842e-01 2.17118502e-01 -4.86057907e-01 8.36296976e-01 3.97132665e-01 -5.12389660e-01 -5.38450480e-01 -1.02243018e+00 -5.20516932e-01 2.36698851e-01 -2.73710251e-01 -3.25859904e-01 -1.03906286e+00 -5.42977273e-01 -4.44485277e-01 5.79857469e-01 -6.81073368e-01 1.93693098e-02 -1.91274867e-03 -1.09219170e+00 1.57588750e-01 1.81356609e-01 4.54435676e-01 -7.04334021e-01 -1.00910258e+00 4.22206134e-01 -1.69268429e-01 -1.14721477e+00 -2.65653491e-01 3.38837653e-01 -1.24396026e+00 -1.25810957e+00 -1.22780633e+00 -6.83890641e-01 1.01306546e+00 -1.80577040e-01 9.37681615e-01 -4.86769825e-02 -4.51806456e-01 1.52655050e-01 -1.13354728e-01 -4.19813275e-01 -3.48439425e-01 2.23326966e-01 -9.03591979e-04 -2.13704810e-01 7.98735097e-02 -4.22772735e-01 -4.72308129e-01 1.57946572e-01 -7.05885470e-01 5.38310289e-01 6.88160539e-01 7.38842428e-01 8.17221642e-01 -2.15095013e-01 2.96038657e-01 -8.08535278e-01 6.78746760e-01 -5.42933404e-01 -4.77543920e-01 1.83583498e-01 -7.82129109e-01 -3.01795840e-01 5.03273487e-01 1.46437623e-02 -8.54786754e-01 1.91681609e-01 -1.70346737e-01 5.16738137e-03 -5.06782234e-01 7.76703715e-01 2.74328262e-01 -9.86655727e-02 5.61093748e-01 1.75569504e-01 3.61703277e-01 -2.57032979e-02 -1.46267682e-01 6.66836917e-01 5.46101451e-01 -1.69554919e-01 1.54273286e-01 3.16567093e-01 1.39567375e-01 -7.25851715e-01 -5.53577185e-01 -5.14117837e-01 -7.59626687e-01 -3.48401546e-01 1.02862966e+00 -7.21368134e-01 -2.41833836e-01 5.74252605e-01 -1.06291389e+00 -3.44423145e-01 -2.29082510e-01 8.72197032e-01 -8.89647126e-01 1.65655449e-01 -4.81455326e-01 -3.00855577e-01 -1.00666666e+00 -1.84545803e+00 6.87344015e-01 1.84000596e-01 -6.21640325e-01 -1.12262797e+00 6.69873208e-02 1.98973894e-01 6.31140411e-01 4.59100902e-01 9.57703888e-01 -8.49265695e-01 -8.16961825e-02 -5.61728597e-01 -2.69008875e-01 -3.73849533e-02 2.99627483e-01 8.66021886e-02 -6.70426071e-01 -1.40359595e-01 -1.02988612e-02 1.72332883e-01 5.57148218e-01 8.11038792e-01 1.12090600e+00 1.18049189e-01 -8.08933377e-01 6.36968195e-01 1.33137870e+00 4.11377251e-01 3.75480771e-01 3.47816110e-01 3.07359219e-01 5.80772519e-01 5.03789604e-01 3.74709755e-01 3.61638039e-01 3.84548306e-01 6.08390033e-01 -2.45704968e-03 -1.09784052e-01 4.75352019e-01 -4.40837704e-02 9.43603158e-01 -3.55656028e-01 1.58052534e-01 -1.39857662e+00 5.39456964e-01 -1.45916688e+00 -5.43901443e-01 -4.64327902e-01 2.49309230e+00 6.42576516e-01 1.66435361e-01 -1.96738586e-01 -5.28083555e-02 6.93266511e-01 -3.34401608e-01 -1.19150974e-01 -1.91488057e-01 2.05634683e-01 1.85783714e-01 5.74048758e-01 4.53027546e-01 -9.71967518e-01 5.76443732e-01 6.48522377e+00 7.37394333e-01 -1.44582295e+00 4.74153012e-01 8.35679054e-01 -1.11754999e-01 1.51516199e-01 -2.53489226e-01 -5.06209493e-01 5.15552640e-01 1.09511578e+00 -4.84766245e-01 -3.49301726e-01 8.14618766e-01 4.28202361e-01 -6.99594319e-01 -7.60518253e-01 9.42722678e-01 -2.02403702e-02 -1.48777580e+00 -1.74931303e-01 1.88158438e-01 6.26670480e-01 3.62204939e-01 -1.71532512e-01 -6.02008700e-02 -9.08516347e-03 -8.46238732e-01 5.92425525e-01 8.14326525e-01 9.33531046e-01 -5.85454524e-01 1.25104821e+00 2.12564975e-01 -9.06379282e-01 2.00271040e-01 5.12806438e-02 2.53661931e-01 9.27955061e-02 8.88488293e-01 -9.82940435e-01 7.45833158e-01 5.16048491e-01 5.50843418e-01 -6.12918496e-01 1.37223423e+00 2.20583007e-01 6.59168139e-02 -3.93908739e-01 6.26602769e-02 2.22909376e-01 -2.42322713e-01 3.49873692e-01 1.11402178e+00 6.52255416e-01 2.04035327e-01 9.02024582e-02 4.69009340e-01 4.60182101e-01 6.09728754e-01 -5.22478580e-01 3.15107703e-01 -2.07118839e-02 1.63610148e+00 -1.40901721e+00 -5.25232255e-01 -2.95476854e-01 6.22755468e-01 3.50920223e-02 -1.87937587e-01 -6.77542031e-01 -2.82193631e-01 3.09294552e-01 4.35392946e-01 -3.27316254e-01 -2.28758715e-02 -9.72565651e-01 -1.08749974e+00 -1.10067628e-01 -5.40442288e-01 3.95620108e-01 -7.39620507e-01 -7.18843102e-01 1.04930854e+00 -1.43165842e-01 -1.31856048e+00 -1.32533357e-01 -8.07887793e-01 -7.60714769e-01 8.11436713e-01 -1.08254623e+00 -7.32339561e-01 -7.35343933e-01 5.22915721e-01 3.48564625e-01 -3.00556928e-01 1.22349226e+00 6.72691613e-02 -6.67364299e-01 2.63937294e-01 -1.10114090e-01 -8.05211440e-02 5.15350699e-01 -1.17459941e+00 -3.53862017e-01 4.88505512e-01 -6.20560944e-01 3.81033421e-01 6.79408193e-01 -5.95089078e-01 -7.27693498e-01 -9.68349218e-01 6.42777443e-01 3.84160280e-02 8.76738369e-01 2.84566760e-01 -6.41287506e-01 4.66601104e-01 1.42687902e-01 1.23188511e-01 1.02011633e+00 -3.89310837e-01 3.98024231e-01 6.56779706e-02 -1.38004899e+00 5.78153014e-01 7.24800825e-01 -2.63703577e-02 -2.72768259e-01 6.36270344e-01 2.13674262e-01 -6.89785123e-01 -1.37179005e+00 3.87115955e-01 5.17331302e-01 -9.73208904e-01 5.32876909e-01 -1.34663925e-01 3.38345051e-01 6.44039288e-02 1.65349871e-01 -1.50698924e+00 -4.65312228e-02 -2.29449093e-01 3.69680166e-01 5.06889224e-01 7.67275095e-01 -8.68586540e-01 6.14640176e-01 1.04107785e+00 -6.69795990e-01 -1.14833736e+00 -9.95887756e-01 -4.75505143e-01 6.87789991e-02 -5.94622552e-01 3.14131707e-01 6.83504999e-01 4.81028646e-01 -2.70095259e-01 6.35252297e-01 8.60153660e-02 4.41077292e-01 -1.58603027e-01 4.42984968e-01 -1.39955974e+00 2.69229770e-01 -9.77478147e-01 -8.96238863e-01 -1.37128234e-01 -1.27812043e-01 -1.16334343e+00 -1.43086940e-01 -1.92609167e+00 2.28627726e-01 -5.99160552e-01 -2.54885525e-01 2.68925011e-01 6.07170202e-02 -2.93597486e-02 -5.57834953e-02 1.88581809e-01 -1.09670728e-01 1.25988960e-01 1.33382988e+00 -4.66712303e-02 -1.06037773e-01 1.04379013e-01 -3.99890840e-01 1.26061225e+00 7.52605975e-01 -5.30850410e-01 -9.11796615e-02 -1.44413665e-01 -1.32493168e-01 3.65444928e-01 7.72361979e-02 -1.40610087e+00 3.82504076e-01 -1.13242023e-01 1.85850218e-01 -7.30169296e-01 -1.18405707e-02 -8.21649373e-01 3.30165684e-01 1.05727386e+00 -1.21630030e-02 2.18778759e-01 2.36634836e-01 1.33496702e-01 -3.96556556e-01 -3.24609578e-01 9.08073485e-01 -3.85178953e-01 -4.15590078e-01 6.61937118e-01 -7.80186296e-01 -2.43276745e-01 1.59315562e+00 -3.85987192e-01 -1.39725536e-01 -4.20566201e-02 -1.33993328e+00 -1.45370886e-02 3.19759518e-01 -1.12011051e-02 5.03644645e-01 -1.00424898e+00 -6.92211151e-01 4.81858589e-02 2.01452687e-01 2.84603298e-01 4.93094116e-01 1.72718382e+00 -1.03712392e+00 6.59437716e-01 -5.15273213e-01 -5.93531251e-01 -1.18729281e+00 2.57841855e-01 6.22227550e-01 -8.21038067e-01 -7.52992213e-01 6.74978316e-01 1.49894968e-01 -2.31261164e-01 1.21280320e-01 -9.40420210e-01 -3.19901943e-01 2.59977639e-01 5.22212625e-01 1.83607027e-01 7.06060469e-01 -6.36835515e-01 -4.70436215e-01 4.56798643e-01 -2.67015457e-01 -9.22483131e-02 1.21351051e+00 3.35569531e-02 -2.34482110e-01 2.75323123e-01 8.56472313e-01 -2.95665801e-01 -4.94742721e-01 1.58481225e-01 9.49206725e-02 -1.96728203e-02 3.96570086e-01 -8.58025074e-01 -1.26789021e+00 6.09628737e-01 7.94944763e-01 2.41419002e-02 1.12364113e+00 -3.10845852e-01 5.21851838e-01 -3.36362422e-02 8.17927659e-01 -8.53196681e-01 -5.39217532e-01 4.02417898e-01 9.04999256e-01 -1.15237141e+00 7.47684389e-02 -4.20377016e-01 -6.95349753e-01 1.27983809e+00 4.05672878e-01 -4.02280241e-02 9.73986328e-01 5.59830546e-01 2.80841410e-01 -3.10032636e-01 -6.65494025e-01 -1.26685843e-01 5.02310395e-02 5.58333933e-01 6.74100459e-01 5.42558968e-01 -7.98798263e-01 9.10412192e-01 -5.36596835e-01 3.46895039e-01 6.81619346e-01 6.71582639e-01 -4.35129523e-01 -5.21409988e-01 -3.19436669e-01 9.01250422e-01 -5.74537933e-01 -1.13988660e-01 9.35369581e-02 7.78527558e-01 8.59638825e-02 4.11918670e-01 2.42787339e-02 -8.78981277e-02 2.59175569e-01 7.58664683e-02 2.66458452e-01 -7.29179502e-01 -7.34896600e-01 -2.70318627e-01 -2.43466683e-02 -6.02598548e-01 -1.17598958e-01 -6.86212122e-01 -1.50490069e+00 -3.64668071e-02 -4.03232962e-01 4.14071083e-02 1.12035251e+00 1.14801002e+00 1.55142605e-01 7.71195769e-01 1.44503862e-01 -7.75879741e-01 -1.77447319e-01 -1.11581254e+00 -7.07362890e-01 -3.25877368e-02 4.20211107e-02 -8.10301363e-01 -2.48281389e-01 -1.71282753e-01]
[14.68093490600586, -2.5449600219726562]
7e69260f-5feb-4a6e-aca0-f4e23d92fc3a
attack-on-practical-speaker-verification
2105.09022
null
https://arxiv.org/abs/2105.09022v1
https://arxiv.org/pdf/2105.09022v1.pdf
Attack on practical speaker verification system using universal adversarial perturbations
In authentication scenarios, applications of practical speaker verification systems usually require a person to read a dynamic authentication text. Previous studies played an audio adversarial example as a digital signal to perform physical attacks, which would be easily rejected by audio replay detection modules. This work shows that by playing our crafted adversarial perturbation as a separate source when the adversary is speaking, the practical speaker verification system will misjudge the adversary as a target speaker. A two-step algorithm is proposed to optimize the universal adversarial perturbation to be text-independent and has little effect on the authentication text recognition. We also estimated room impulse response (RIR) in the algorithm which allowed the perturbation to be effective after being played over the air. In the physical experiment, we achieved targeted attacks with success rate of 100%, while the word error rate (WER) on speech recognition was only increased by 3.55%. And recorded audios could pass replay detection for the live person speaking.
['Xiaolin Hu', 'Thomas Fang Zheng', 'Xingliang Cheng', 'Jianmin Li', 'Le Liu', 'Shuning Zhao', 'Weiyi Zhang']
2021-05-19
null
null
null
null
['real-world-adversarial-attack', 'room-impulse-response']
['adversarial', 'audio']
[ 5.31135142e-01 2.74427980e-01 5.37296653e-01 7.82058612e-02 -1.23412931e+00 -9.52686012e-01 2.99360335e-01 -1.27498120e-01 -2.88582146e-01 4.15463686e-01 2.07843930e-02 -6.35193884e-01 3.13775450e-01 -3.95816982e-01 -6.57620132e-01 -9.17669713e-01 -2.68542558e-01 -1.93562478e-01 1.13376349e-01 -1.83083817e-01 3.09621654e-02 5.95773339e-01 -1.31333995e+00 2.47233108e-01 4.29993719e-01 8.63597214e-01 -2.32932508e-01 1.35117960e+00 3.36068362e-01 5.33023179e-01 -1.62527478e+00 -2.30673313e-01 4.06966776e-01 -4.28347498e-01 -5.63168466e-01 -3.49817365e-01 6.22846223e-02 -5.47956765e-01 -6.02483034e-01 1.10537016e+00 9.65397298e-01 -2.55416855e-02 3.23468983e-01 -1.38089252e+00 -6.12530187e-02 1.10071385e+00 -2.02106461e-01 -1.14803575e-02 1.06719339e+00 2.56446838e-01 4.98581171e-01 -4.74128991e-01 -1.29293889e-01 1.09888840e+00 5.16907215e-01 7.31805205e-01 -1.00643325e+00 -1.21350300e+00 -1.42473474e-01 6.82084495e-03 -1.81436837e+00 -7.49620676e-01 8.96131754e-01 8.11386704e-02 6.64316058e-01 8.17236245e-01 3.02516371e-01 1.18439782e+00 1.76413685e-01 2.87245870e-01 6.87139332e-01 -7.75723219e-01 1.94847569e-01 3.38160425e-01 1.00787789e-01 2.35240430e-01 -5.27567640e-02 4.00481880e-01 -6.50389850e-01 -6.77719593e-01 3.13118875e-01 -5.19714952e-01 -7.97797203e-01 6.62383556e-01 -1.17089093e+00 3.20330620e-01 -2.66565651e-01 1.40869617e-01 -1.75385714e-01 -1.51269780e-02 4.98165518e-01 7.17082143e-01 -2.27673516e-01 1.89662203e-01 -1.88968524e-01 -3.23949188e-01 -6.04553580e-01 -1.65875573e-02 1.06618643e+00 5.87821603e-01 -5.79292215e-02 7.24186122e-01 -1.26787245e-01 3.92417520e-01 4.05739307e-01 1.24252975e+00 4.81132805e-01 -2.16049343e-01 4.28129107e-01 -1.45575330e-01 4.29361850e-01 -7.86427200e-01 -1.15786128e-01 -1.78367496e-01 -4.78003800e-01 4.48520929e-01 7.38139808e-01 -6.45252705e-01 -7.25187838e-01 1.53340566e+00 3.97819042e-01 3.44976187e-01 4.94802982e-01 7.58647621e-01 5.20633757e-01 7.24586487e-01 -3.64414692e-01 -2.55311400e-01 1.51445675e+00 -4.25030708e-01 -1.17478049e+00 -1.16210366e-02 2.87808210e-01 -1.32039547e+00 1.32612598e+00 8.65771353e-01 -8.78819704e-01 -5.45208037e-01 -1.65850270e+00 8.54983628e-01 1.07907325e-01 -2.23701626e-01 -7.45210517e-03 1.80871642e+00 -7.46531785e-01 -7.50378519e-02 -5.50085604e-01 1.97297797e-01 -4.67645884e-01 6.35293126e-01 -3.94506961e-01 3.62632900e-01 -1.65553606e+00 4.76099432e-01 -6.26465753e-02 1.86170176e-01 -1.15673411e+00 -4.10071492e-01 -6.98121667e-01 2.96893381e-02 -1.01262878e-03 -1.24234967e-01 1.35748482e+00 -1.02352905e+00 -2.01098728e+00 4.53653008e-01 -7.89330974e-02 -6.14206314e-01 4.99824673e-01 -2.18522936e-01 -1.43456459e+00 4.11329031e-01 -5.35312474e-01 -3.30268621e-01 1.57672083e+00 -1.04922235e+00 -1.15549684e-01 -4.58643511e-02 3.07444367e-04 5.62028447e-03 -3.37594241e-01 3.69435936e-01 2.47767791e-01 -7.06804216e-01 1.01257131e-01 -9.10089910e-01 2.79786795e-01 -6.77250266e-01 -6.11105800e-01 5.94234526e-01 9.24141645e-01 -7.80780435e-01 1.31898820e+00 -2.49867344e+00 -5.79090416e-01 6.12262666e-01 -4.00540948e-01 5.22197664e-01 1.75836951e-01 4.96918112e-01 -3.98951143e-01 1.52635336e-01 8.38297158e-02 -1.41465738e-01 -3.42773125e-02 -3.12579304e-01 -8.37250054e-01 1.01951396e+00 -2.34051198e-01 2.00022012e-01 -4.65159595e-01 7.37094134e-02 -5.02530597e-02 6.57534063e-01 -3.64712059e-01 4.37120527e-01 5.29446781e-01 2.20000297e-01 -2.92662024e-01 5.89997709e-01 8.08987021e-01 8.72652173e-01 -7.41356015e-02 -2.30767969e-02 1.34121299e-01 5.08739293e-01 -1.84114516e+00 8.68162811e-01 -4.40287054e-01 5.67343235e-01 7.46789277e-01 -6.19712770e-01 9.02795553e-01 1.06540036e+00 7.49864674e-04 -2.68254817e-01 4.03664023e-01 -5.29021844e-02 2.51106888e-01 -3.18081647e-01 2.68197924e-01 -1.29960701e-01 -2.04453230e-01 7.64843881e-01 -2.68247455e-01 -5.25510963e-03 -1.03797925e+00 2.11140752e-01 1.14370871e+00 -4.72338527e-01 7.27999732e-02 -1.95087388e-01 1.02003431e+00 -5.64081967e-01 3.43310893e-01 1.09165466e+00 -4.74737853e-01 3.41682345e-01 -6.87386468e-02 1.08603574e-01 -4.49667335e-01 -1.10845506e+00 -6.36446709e-03 7.78131485e-01 1.73298299e-01 -3.78205270e-01 -1.02833092e+00 -3.86437774e-01 -2.88210690e-01 7.94874609e-01 8.05549175e-02 -5.89585125e-01 -6.46518767e-01 -2.95853347e-01 1.77472782e+00 9.64652374e-02 2.37360477e-01 -7.45664954e-01 -8.25770795e-02 2.32939675e-01 -2.25540563e-01 -1.16205990e+00 -9.40661013e-01 2.05435932e-01 -1.60750672e-01 -7.65233755e-01 -2.56516933e-01 -6.11536026e-01 5.13837278e-01 3.25428575e-01 3.00381541e-01 2.51577348e-01 -1.67580977e-01 7.26435363e-01 -2.95884490e-01 -8.56351078e-01 -1.10152256e+00 -5.22854030e-01 7.56107986e-01 4.14841652e-01 9.22881588e-02 -3.89374465e-01 -3.80311519e-01 7.61264205e-01 -8.31294954e-01 -7.77054548e-01 -2.20944420e-01 6.87350273e-01 -3.55398268e-01 5.60256362e-01 8.06212425e-01 -1.27077252e-01 7.59582579e-01 1.34698227e-01 -3.44767451e-01 4.40689325e-02 -3.47631378e-03 -1.15421824e-01 8.60012233e-01 -1.10260439e+00 -1.03236401e+00 7.06387535e-02 -6.35835528e-01 9.04237032e-02 -3.52433473e-01 9.66419354e-02 -8.63555610e-01 -4.28006887e-01 8.72231662e-01 5.69698930e-01 9.50196385e-02 -7.11330324e-02 1.30467132e-01 1.16689789e+00 6.48365498e-01 -3.77766401e-01 1.41223729e+00 2.43065089e-01 -5.04445791e-01 -1.42759168e+00 -4.68151644e-02 -3.29268366e-01 -7.97132254e-02 -4.98363644e-01 4.44649816e-01 -1.01991010e+00 -1.36922789e+00 1.04855418e+00 -9.98512864e-01 -8.10507461e-02 3.61617565e-01 7.57475436e-01 -3.63120027e-02 9.10499513e-01 -5.04882753e-01 -1.56496561e+00 -5.98678768e-01 -1.01084340e+00 6.70972288e-01 -9.09422934e-02 -3.13925803e-01 -5.86903572e-01 -3.32593203e-01 4.43107367e-01 4.95982468e-01 -4.28917520e-02 4.35780913e-01 -1.05418146e+00 -8.53337273e-02 -9.75482106e-01 8.25100660e-01 3.84085208e-01 3.21083039e-01 1.32516488e-01 -1.58209598e+00 -4.74050939e-01 6.36912405e-01 -8.66139531e-02 3.24526466e-02 7.54824132e-02 5.64402640e-01 -6.60871625e-01 -6.25291467e-02 4.38822627e-01 7.43266702e-01 5.77207923e-01 9.20624733e-01 4.44086492e-02 2.99657315e-01 2.40000546e-01 4.87368464e-01 3.94377351e-01 -4.64667350e-01 6.85253859e-01 2.46282950e-01 1.10269338e-01 2.63429791e-01 -1.58557311e-01 1.17080629e+00 7.51911819e-01 2.31335342e-01 -5.08614123e-01 -6.69886112e-01 -3.24385278e-02 -1.00855744e+00 -1.22780442e+00 -1.66447341e-01 2.72043157e+00 7.77827561e-01 6.95020735e-01 1.44754693e-01 1.01221001e+00 1.01562595e+00 -1.39427498e-01 1.12871230e-02 -6.25712693e-01 -5.94781637e-02 2.54388064e-01 7.15284705e-01 1.10303354e+00 -9.29580927e-01 5.40192127e-01 6.56112623e+00 5.58904588e-01 -1.46286070e+00 -1.32714033e-01 5.29324748e-02 -9.18615311e-02 -1.36926383e-01 -1.90916896e-01 -8.19186330e-01 2.29373187e-01 1.46614981e+00 -3.60601783e-01 3.96449089e-01 5.71118116e-01 4.27359462e-01 1.41199902e-01 -9.79068577e-01 9.21638012e-01 2.30366498e-01 -4.42634404e-01 -1.05965674e-01 -5.46201952e-02 -2.34634787e-01 -6.15317464e-01 1.21926568e-01 3.86874765e-01 -6.02836795e-02 -9.74596024e-01 8.86101484e-01 9.43998098e-02 7.33475566e-01 -9.24242198e-01 5.95541894e-01 5.22713721e-01 -1.22387731e+00 1.46438321e-02 -2.77087335e-02 -2.93441951e-01 2.43699089e-01 2.48669475e-01 -1.40819931e+00 3.91012490e-01 1.11062177e-01 -6.86830521e-01 -2.63134148e-02 6.37496769e-01 -2.09360182e-01 1.37374341e+00 -6.55283988e-01 -5.73433284e-03 -1.75287440e-01 3.27570140e-01 1.05470932e+00 1.02576447e+00 2.41455331e-01 9.97038186e-02 -8.03155601e-02 2.15148613e-01 1.12496436e-01 1.95801370e-02 -6.10088170e-01 1.40576765e-01 7.52257526e-01 7.98805177e-01 -3.86433035e-01 5.70344441e-02 1.19715124e-01 1.14874625e+00 -7.63125539e-01 4.84321028e-01 -1.33050787e+00 -1.00546014e+00 6.63801730e-01 7.41587728e-02 -7.25285262e-02 -3.08039933e-01 -4.17559892e-02 -9.08450365e-01 1.16590574e-01 -1.30130434e+00 2.08760425e-01 -4.50531244e-01 -7.53747344e-01 6.32234454e-01 -5.13231695e-01 -1.38767385e+00 -1.99455380e-01 -1.87564999e-01 -8.64472806e-01 1.32308841e+00 -6.47254229e-01 -6.94449008e-01 9.62688625e-02 1.00883853e+00 2.73608595e-01 -3.33914787e-01 1.33417714e+00 3.64874899e-01 -4.31122124e-01 1.41709077e+00 -7.79843405e-02 5.23002326e-01 8.94108951e-01 -9.12558496e-01 6.34848595e-01 1.20308757e+00 5.23515567e-02 1.07758212e+00 1.13162041e+00 -4.92974222e-01 -1.73204374e+00 -5.75600386e-01 5.57152450e-01 -2.09901363e-01 6.32700920e-01 -9.16403890e-01 -9.35218751e-01 4.06312764e-01 2.31638476e-01 -4.69330102e-01 1.00485826e+00 -2.80045360e-01 -4.64363128e-01 -1.97551876e-01 -1.51750195e+00 6.76512957e-01 2.25749195e-01 -1.11212146e+00 -7.60420799e-01 1.18608318e-01 8.72236788e-01 -3.37445199e-01 -4.51596111e-01 -3.01119052e-02 8.09365928e-01 -4.92415816e-01 1.14665711e+00 -3.99458140e-01 -6.49858117e-01 -6.69036150e-01 -2.16860071e-01 -9.46194589e-01 2.78374791e-01 -1.46129203e+00 -2.90435199e-02 1.54758370e+00 6.21102989e-01 -9.27460790e-01 3.71531487e-01 6.18508875e-01 2.65178680e-01 3.15740585e-01 -1.11729991e+00 -9.07689273e-01 -3.17149252e-01 -8.12030852e-01 6.26357853e-01 7.87503660e-01 4.41519678e-01 2.44535059e-01 -7.83559918e-01 1.27744961e+00 4.87884372e-01 -8.34845781e-01 9.30216134e-01 -5.86843669e-01 -5.52821517e-01 8.73495936e-02 -6.49413764e-01 -9.52222168e-01 -9.15250853e-02 -4.96813178e-01 1.50557682e-01 -5.15796125e-01 -7.62241721e-01 -3.17669623e-02 -3.61729860e-01 2.48445459e-02 -2.50640124e-01 -4.18647528e-02 -1.14939332e-01 -3.27764332e-01 2.95134246e-01 2.43769288e-01 4.55531597e-01 -4.23798144e-01 -3.54026377e-01 7.89428949e-01 -4.27255720e-01 5.74720979e-01 8.35032165e-01 -5.42587757e-01 -5.39052129e-01 2.28085935e-01 -1.04217604e-01 4.58334804e-01 1.37945563e-01 -1.18286777e+00 2.14565337e-01 1.58836290e-01 1.47180229e-01 -2.87072212e-01 4.23005342e-01 -1.18940854e+00 1.19388849e-01 5.22947550e-01 -2.48100936e-01 -1.25700533e-01 5.39370000e-01 4.70916867e-01 -5.71294464e-02 -1.55680180e-01 7.29160190e-01 4.37543631e-01 2.59829670e-01 -3.66553575e-01 -9.78958368e-01 -4.04962718e-01 8.02405775e-01 -3.45564097e-01 -1.81079596e-01 -8.95735145e-01 -6.37983859e-01 -2.76640534e-01 8.55052099e-02 1.64664492e-01 6.86064184e-01 -8.33581626e-01 -6.26155496e-01 5.91871202e-01 -3.45472157e-01 -6.72616959e-01 3.66855323e-01 4.13054228e-01 -5.21562278e-01 1.98639184e-01 2.59200037e-01 -4.55213696e-01 -2.02789640e+00 5.96563697e-01 6.44062996e-01 4.00047302e-01 -6.57508314e-01 8.27581704e-01 -3.49473476e-01 -9.52578858e-02 8.53266418e-01 -1.88557401e-01 9.48269591e-02 -3.18090051e-01 1.24570322e+00 9.11250040e-02 4.62986797e-01 -4.39538270e-01 -5.85312903e-01 1.69142321e-01 8.68417397e-02 -6.38938069e-01 5.00528574e-01 -1.72243908e-01 3.91909361e-01 3.91509622e-01 1.07191801e+00 9.85992193e-01 -5.31426013e-01 -2.51318607e-02 -3.62190485e-01 -3.80384415e-01 -3.09046414e-02 -7.15352356e-01 -4.41202879e-01 4.31697845e-01 7.38936901e-01 7.11328387e-01 1.10071361e+00 -4.99237955e-01 7.50141621e-01 6.30820215e-01 3.88085723e-01 -7.60557055e-01 -1.78185761e-01 3.91932249e-01 7.58971334e-01 -7.62556612e-01 -2.05104679e-01 -3.39825571e-01 -6.81558311e-01 1.08659089e+00 1.04421012e-01 2.31053784e-01 8.83100748e-01 8.79248738e-01 6.49010003e-01 3.24183345e-01 -3.83901805e-01 6.39425933e-01 1.64961204e-01 8.73539448e-01 1.52513668e-01 1.33927777e-01 8.99952501e-02 9.89080131e-01 -6.90702796e-01 -5.97221971e-01 8.69911432e-01 9.48306441e-01 -4.45228070e-01 -9.00066972e-01 -1.27776349e+00 -1.72585785e-01 -1.05834651e+00 -1.47475451e-01 -5.69607913e-01 3.00936997e-01 -4.93676543e-01 1.57873762e+00 -4.12653655e-01 -6.73921704e-01 6.10954463e-01 4.86523867e-01 4.92221937e-02 -7.64118135e-02 -9.43594396e-01 1.91642046e-01 2.38109320e-01 -2.07115039e-01 -1.02311723e-01 -3.64237070e-01 -1.45474207e+00 -4.70912665e-01 -6.75717533e-01 4.45575833e-01 7.81418145e-01 8.10365438e-01 -1.98677450e-01 6.27668381e-01 1.21556628e+00 -3.86354953e-01 -8.10047805e-01 -9.43856418e-01 -8.58422577e-01 2.90946495e-02 7.99532950e-01 -1.28916934e-01 -1.07858026e+00 4.92721312e-02]
[13.987298965454102, 5.809381484985352]
effeb4fb-e6c5-4907-a495-7a6cac7a5a1b
machine-and-deep-learning-methods-with-manual
2210.10903
null
https://arxiv.org/abs/2210.10903v1
https://arxiv.org/pdf/2210.10903v1.pdf
Machine and Deep Learning Methods with Manual and Automatic Labelling for News Classification in Bangla Language
Research in Natural Language Processing (NLP) has increasingly become important due to applications such as text classification, text mining, sentiment analysis, POS tagging, named entity recognition, textual entailment, and many others. This paper introduces several machine and deep learning methods with manual and automatic labelling for news classification in the Bangla language. We implemented several machine (ML) and deep learning (DL) algorithms. The ML algorithms are Logistic Regression (LR), Stochastic Gradient Descent (SGD), Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbour (KNN), used with Bag of Words (BoW), Term Frequency-Inverse Document Frequency (TF-IDF), and Doc2Vec embedding models. The DL algorithms are Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network (CNN), used with Word2vec, Glove, and FastText word embedding models. We develop automatic labelling methods using Latent Dirichlet Allocation (LDA) and investigate the performance of single-label and multi-label article classification methods. To investigate performance, we developed from scratch Potrika, the largest and the most extensive dataset for news classification in the Bangla language, comprising 185.51 million words and 12.57 million sentences contained in 664,880 news articles in eight distinct categories, curated from six popular online news portals in Bangladesh for the period 2014-2020. GRU and Fasttext with 91.83% achieve the highest accuracy for manually-labelled data. For the automatic labelling case, KNN and Doc2Vec at 57.72% and 75% achieve the highest accuracy for single-label and multi-label data, respectively. The methods developed in this paper are expected to advance research in Bangla and other languages.
['Rashid Mehmood', 'Fahad AlQurashi', 'Istiak Ahmad']
2022-10-19
null
null
null
null
['news-classification']
['natural-language-processing']
[-4.10844058e-01 -7.95978904e-02 -4.82531697e-01 -3.89188796e-01 -5.98884583e-01 -6.30224705e-01 7.84983575e-01 4.67300057e-01 -9.07506764e-01 8.81047904e-01 7.49990165e-01 -7.48631299e-01 2.30098516e-01 -1.00166965e+00 -2.73075819e-01 -8.44165802e-01 9.59089771e-02 5.54784834e-01 -2.47167632e-01 -7.58868977e-02 4.09250528e-01 3.65512908e-01 -1.25242221e+00 4.75797147e-01 3.55931729e-01 9.79700685e-01 5.19531146e-02 6.32694185e-01 -5.32590210e-01 9.86787081e-01 -4.73044395e-01 -5.65162480e-01 -2.59632111e-01 -5.38603328e-02 -1.25147569e+00 -2.43703485e-01 -3.57590556e-01 -1.69512436e-01 -1.66210040e-01 6.10082269e-01 9.37502325e-01 2.33348861e-01 1.03597617e+00 -7.33997107e-01 -9.93608773e-01 6.10582411e-01 -6.71125650e-01 2.85365522e-01 1.83089405e-01 -4.28781778e-01 9.33628738e-01 -1.02962291e+00 7.03710854e-01 1.25353110e+00 1.08637345e+00 3.89249533e-01 -7.17065275e-01 -5.73476493e-01 -4.14056718e-01 1.21919103e-01 -1.30740643e+00 -1.00887291e-01 2.62150854e-01 -6.13711596e-01 1.68882322e+00 -1.76269576e-01 4.26139444e-01 1.12909210e+00 7.17309535e-01 6.97703302e-01 1.25105119e+00 -9.68807161e-01 1.02027766e-01 5.05290151e-01 5.72352350e-01 5.42856276e-01 8.61506760e-02 -2.68938005e-01 -2.53617644e-01 -4.03526485e-01 3.35889339e-01 6.44410476e-02 2.06016928e-01 3.21443826e-01 -1.03295612e+00 1.55375767e+00 -9.74551812e-02 6.69026315e-01 -4.97636765e-01 -1.97774991e-01 1.06993783e+00 4.54534322e-01 1.16835117e+00 2.87331581e-01 -1.00067997e+00 -1.51409581e-01 -8.55362535e-01 -6.42528981e-02 1.08959794e+00 7.49447465e-01 5.64574659e-01 1.75417528e-01 -2.57763505e-01 1.36131704e+00 6.18559778e-01 3.97243112e-01 1.23037493e+00 -2.69475013e-01 2.04810515e-01 3.05124015e-01 -9.11110118e-02 -1.23720574e+00 -8.09571087e-01 -3.45993191e-01 -9.36737597e-01 -4.82529014e-01 -6.34151474e-02 -7.17913985e-01 -9.91952240e-01 1.15540278e+00 1.80099145e-01 -2.70259917e-01 4.79636878e-01 3.61490846e-01 1.29318273e+00 1.34434080e+00 4.46751982e-01 -3.99843723e-01 1.79226470e+00 -1.06277573e+00 -8.91599417e-01 8.21091235e-02 1.49969053e+00 -1.04459047e+00 6.90758407e-01 2.69628078e-01 -5.76226711e-01 -4.73015219e-01 -6.75119460e-01 -2.05257908e-01 -1.06375265e+00 2.63892591e-01 6.21872723e-01 7.97552228e-01 -9.98094857e-01 1.51062667e-01 -4.95592505e-01 -7.93121874e-01 1.94728196e-01 2.35237777e-01 -4.89923477e-01 -7.65358051e-03 -1.78568876e+00 1.16529596e+00 4.76835787e-01 -1.20889917e-01 -6.36803448e-01 -4.13871603e-03 -1.09145117e+00 -1.83245108e-01 -3.51224333e-01 -2.04942971e-02 1.08327937e+00 -7.52545774e-01 -1.42093122e+00 1.23838007e+00 -9.17761698e-02 -6.42619789e-01 -2.78277189e-01 -1.62822396e-01 -7.24478483e-01 -1.98015273e-01 2.66901374e-01 6.01650596e-01 3.31149101e-01 -8.18075001e-01 -7.45901406e-01 -2.55840749e-01 -3.23740572e-01 1.32362127e-01 -5.08401394e-01 6.86561942e-01 2.78444558e-01 -7.39502609e-01 -1.11857489e-01 -6.79478943e-01 5.81135824e-02 -9.16640699e-01 -1.80801675e-01 -8.41548383e-01 5.78663051e-01 -1.25467217e+00 1.29934430e+00 -2.18874383e+00 -3.37607086e-01 -9.04433951e-02 -2.95398355e-01 2.82603472e-01 1.68479174e-01 7.76390731e-01 5.88899739e-02 1.99628651e-01 1.26699850e-01 -2.82309949e-01 9.73028317e-02 5.35422981e-01 -3.77811134e-01 6.64130986e-01 -1.98520824e-01 9.43507016e-01 -7.16155827e-01 -6.60480320e-01 1.43217128e-02 6.58195019e-01 -6.10379968e-03 -1.67515293e-01 2.51252741e-01 7.85479024e-02 -9.79957506e-02 5.13637781e-01 3.39761883e-01 1.33367658e-01 1.22858258e-02 -1.47770122e-02 -4.41961050e-01 5.30125678e-01 -9.32581365e-01 1.18763340e+00 -9.12863851e-01 8.65606189e-01 -3.17955077e-01 -1.13466167e+00 1.34553492e+00 8.67060602e-01 1.35690391e-01 -6.49556577e-01 4.36007440e-01 3.10202360e-01 -3.84899348e-01 -8.02477419e-01 5.64001799e-01 -2.92896658e-01 -3.40642065e-01 6.12169981e-01 6.13654435e-01 4.14332837e-01 -3.29216272e-02 -1.14935834e-03 6.59778476e-01 -2.53523022e-01 6.19756579e-01 -3.03267032e-01 2.76785284e-01 1.80053905e-01 2.91330844e-01 5.80884159e-01 -5.10202348e-02 3.43259394e-01 5.95955253e-01 -7.16814041e-01 -1.25864220e+00 -2.20519215e-01 -5.50804317e-01 1.70334029e+00 -6.33419275e-01 1.86256580e-02 -4.76688266e-01 -6.20516062e-01 -2.14804307e-01 1.19343555e+00 -5.49189985e-01 5.02494536e-02 -1.65388808e-01 -1.10635710e+00 8.96062434e-01 3.59162092e-01 5.63586056e-01 -1.65440834e+00 -3.37233655e-02 4.79625434e-01 -1.10061482e-01 -7.90957153e-01 1.00557432e-01 6.16178095e-01 -5.70523322e-01 -3.65524411e-01 -9.68628883e-01 -1.38607633e+00 3.65080655e-01 -1.74842536e-01 1.02545905e+00 -5.12802422e-01 4.11186852e-02 1.00750647e-01 -9.40120339e-01 -3.85551035e-01 -3.25560629e-01 3.08733612e-01 2.94513941e-01 -1.21819101e-01 1.09271228e+00 -1.27954185e-01 -1.65994138e-01 4.80347797e-02 -8.25489283e-01 -2.45788172e-01 5.72653353e-01 1.04547548e+00 2.09546044e-01 1.25293002e-01 9.03761506e-01 -1.03091633e+00 8.85534823e-01 -1.01581657e+00 -6.35147020e-02 1.19243629e-01 -7.98954785e-01 -2.13251546e-01 4.02321070e-01 -3.63124102e-01 -1.11477411e+00 -2.57737756e-01 -7.34030604e-01 4.39887255e-01 -6.26194835e-01 1.06009817e+00 3.50456625e-01 2.32146338e-01 8.07103276e-01 4.33383018e-01 -3.84794056e-01 -6.48544014e-01 4.43688899e-01 1.72651279e+00 -1.09138958e-01 -1.16561152e-01 -9.40817669e-02 1.61731213e-01 -6.83681071e-01 -8.58523548e-01 -1.23830342e+00 -9.76930141e-01 -8.28471184e-01 -1.13756113e-01 1.20792472e+00 -1.01466310e+00 -2.84125388e-01 7.43483961e-01 -1.19448662e+00 5.09305447e-02 -6.87494725e-02 9.38764572e-01 -7.50620812e-02 2.35421434e-01 -1.09969521e+00 -9.08526242e-01 -7.85033703e-01 -8.02912116e-01 7.89932072e-01 6.58350065e-02 -4.39354360e-01 -1.38772702e+00 3.14195991e-01 2.31724292e-01 4.43923116e-01 1.89399540e-01 1.10413241e+00 -1.44621730e+00 9.17545974e-01 -5.35807908e-01 -3.11041117e-01 7.57184625e-01 -4.14489117e-03 -2.00656638e-01 -9.14026558e-01 -2.80554265e-01 6.37384653e-02 -6.61385119e-01 8.86326909e-01 5.83865762e-01 4.78449255e-01 -5.34611642e-01 -1.40862912e-01 2.45825816e-02 1.47316003e+00 3.72174203e-01 6.17187679e-01 8.08233559e-01 7.54580617e-01 5.67407906e-01 4.11228299e-01 5.62627077e-01 4.85896945e-01 1.55844852e-01 -1.83882847e-01 4.87407483e-02 2.45951131e-01 1.20667145e-01 6.19178891e-01 1.34892786e+00 6.31088465e-02 -2.57842034e-01 -1.41188586e+00 8.35015714e-01 -1.46937764e+00 -7.85251558e-01 -3.66617829e-01 1.85770607e+00 1.10479343e+00 1.77072197e-01 -5.07668518e-02 3.18323255e-01 8.71578336e-01 1.04673043e-01 1.47822782e-01 -1.30331314e+00 -2.19967693e-01 2.35111073e-01 8.87720168e-01 3.15943122e-01 -1.40671134e+00 9.57783997e-01 4.99767447e+00 1.20021534e+00 -1.00598574e+00 6.62832022e-01 8.54929268e-01 5.18966734e-01 4.47568409e-02 -2.09543377e-01 -1.31188464e+00 5.21803558e-01 1.42160106e+00 3.08774501e-01 -9.74597707e-02 1.10157549e+00 1.58994123e-02 -8.43841657e-02 -3.51137817e-01 7.13749468e-01 2.42570728e-01 -1.16908514e+00 -6.27265424e-02 -6.09387010e-02 9.75541234e-01 4.11215663e-01 -5.71263172e-02 9.25039530e-01 3.01445305e-01 -9.36658978e-01 2.71850079e-01 2.74150968e-01 8.76289606e-01 -1.12891126e+00 1.66807032e+00 4.71505195e-01 -6.94708049e-01 -7.12149590e-02 -7.61582971e-01 -2.93723613e-01 1.02346107e-01 7.67171681e-01 -7.25443363e-01 2.48376116e-01 7.40386903e-01 6.47444248e-01 -1.70192733e-01 6.19714558e-01 -1.13582261e-01 1.16877508e+00 -1.56449541e-01 -5.62604427e-01 9.37885463e-01 1.29888490e-01 1.18112981e-01 1.85006917e+00 1.22322552e-01 -5.25687821e-02 -1.06876723e-01 -1.97287112e-01 -1.07792383e-02 7.48450279e-01 -4.78934020e-01 -2.20973678e-02 2.58035421e-01 1.33244741e+00 -1.07090425e+00 -4.66201097e-01 -6.73519909e-01 7.56483912e-01 1.86859414e-01 2.24636853e-01 -7.12734938e-01 -8.34591508e-01 -1.41230151e-01 -2.54983127e-01 3.44655663e-01 -1.98728070e-01 -2.26738051e-01 -8.50746870e-01 -5.38177788e-01 -5.19820631e-01 5.18454611e-01 -4.97671008e-01 -1.63274276e+00 6.80593431e-01 -9.29164961e-02 -8.19677711e-01 -1.27926737e-01 -6.53596818e-01 -2.34476298e-01 9.03928101e-01 -1.43951726e+00 -1.22594154e+00 3.52538586e-01 3.42048228e-01 6.74785137e-01 -5.35073221e-01 1.24985993e+00 6.18207693e-01 -3.88380408e-01 4.33640093e-01 1.04210532e+00 5.73676229e-01 5.98137558e-01 -1.17479527e+00 3.65778178e-01 2.16385666e-02 9.98022333e-02 2.40660787e-01 2.78530866e-01 -6.92020774e-01 -5.68273783e-01 -1.26359022e+00 1.85813618e+00 -1.32460251e-01 6.73438549e-01 -3.63798767e-01 -6.44461811e-01 7.09608257e-01 4.58487898e-01 -3.54591519e-01 1.18065834e+00 1.69678390e-01 -1.24181315e-01 3.24962735e-01 -1.19399726e+00 3.46532539e-02 -6.89394027e-02 -5.94619036e-01 -7.30813265e-01 8.69559467e-01 7.80296326e-01 1.34196961e-02 -1.07838309e+00 4.85476814e-02 4.81821984e-01 -2.59178251e-01 5.20454466e-01 -7.97520459e-01 4.97459263e-01 2.33655632e-01 -2.52954900e-01 -1.22833991e+00 -5.18735588e-01 3.00423000e-02 2.77085751e-02 1.72257745e+00 7.18818247e-01 -8.66669714e-01 2.08221391e-01 2.09399894e-01 -5.04743010e-02 -1.02634227e+00 -8.92132461e-01 -5.57063818e-01 3.20386648e-01 -4.53295767e-01 7.57831633e-02 1.57609081e+00 -8.98998871e-04 6.54284298e-01 -6.16864622e-01 -2.61884987e-01 2.61504911e-02 -3.01810920e-01 1.32905379e-01 -1.08662045e+00 1.69989571e-01 -1.46660702e-02 -5.10215044e-01 -7.88738251e-01 2.96565473e-01 -9.27418649e-01 -1.85424998e-01 -1.98501480e+00 9.87108052e-02 -2.39855587e-01 -3.55987549e-01 7.85018265e-01 4.05059159e-01 3.51322114e-01 -4.94482279e-01 2.69068927e-01 -5.07343411e-01 3.29861969e-01 6.65662646e-01 -4.78461795e-02 -3.09107229e-02 -9.54275429e-02 -3.23367149e-01 7.44934976e-01 1.03156590e+00 -9.85695481e-01 3.65251929e-01 -3.85161579e-01 5.21034837e-01 -2.68363982e-01 -1.66606307e-01 -5.14920175e-01 2.13332251e-01 3.31778824e-01 5.01811802e-01 -7.24200249e-01 -1.84385598e-01 -4.49393213e-01 -3.44548315e-01 4.83296126e-01 -5.77002764e-01 9.77935418e-02 4.18321490e-02 2.06831783e-01 -4.15364057e-01 -8.62017155e-01 6.90974176e-01 -2.38977313e-01 -8.08063269e-01 -5.58432229e-02 -1.15387106e+00 -1.08539313e-01 9.62916315e-01 -4.95950207e-02 -1.29565224e-01 -3.59280497e-01 -8.88979077e-01 1.13255501e-01 -3.16489518e-01 3.06499422e-01 3.99033278e-01 -1.39390731e+00 -9.68097687e-01 3.76850530e-03 6.56500086e-02 -2.17801675e-01 1.81272656e-01 7.57933140e-01 -8.56050730e-01 7.89828718e-01 -1.15140259e-01 -1.05154715e-01 -9.25330162e-01 4.22367454e-01 -1.25892192e-01 -5.27185023e-01 -2.38360807e-01 1.15724933e+00 -4.39542234e-01 -9.17311788e-01 2.51435429e-01 1.26811527e-02 -1.20859718e+00 8.56574297e-01 3.27926427e-01 5.15040934e-01 1.97189957e-01 -1.14079046e+00 -2.14075163e-01 5.66342592e-01 -2.97505766e-01 4.99836914e-03 1.42732298e+00 -2.69420385e-01 -5.28453290e-01 8.99261534e-01 1.71897280e+00 -3.48530442e-01 -5.41558191e-02 -3.48610133e-01 1.63637698e-01 2.84074932e-01 5.38942516e-01 -6.99396908e-01 -6.33805454e-01 8.71203184e-01 6.13979161e-01 7.42658138e-01 5.59320748e-01 -6.59386963e-02 1.09223747e+00 5.44102609e-01 5.62878959e-02 -1.52467585e+00 -5.95565140e-01 1.05581307e+00 4.49476421e-01 -1.13959658e+00 -1.03173845e-01 4.83871698e-01 -9.02542651e-01 1.29442620e+00 1.32808596e-01 -7.94615448e-02 1.19001555e+00 2.19590113e-01 3.11943680e-01 -3.06746811e-01 -5.55717230e-01 7.32345786e-03 1.06756717e-01 2.80877173e-01 8.79576266e-01 6.65469617e-02 -7.84471810e-01 7.20839620e-01 -1.89041376e-01 -2.57279932e-01 2.79418230e-01 1.00874543e+00 -6.93858206e-01 -6.50363982e-01 -2.52770126e-01 8.50496888e-01 -1.14051831e+00 -4.47474360e-01 -2.69027636e-03 5.28409660e-01 2.27823660e-01 9.57151473e-01 2.07451582e-01 -1.78968415e-01 -6.82989880e-02 5.33658385e-01 -3.07866156e-01 -8.74340057e-01 -7.23256648e-01 -1.15718879e-02 3.79565060e-01 3.24273348e-01 -6.95123255e-01 -5.82130492e-01 -1.15151739e+00 -2.97096699e-01 -8.42782915e-01 5.17377734e-01 1.11007369e+00 1.18530071e+00 6.47623837e-02 2.23276153e-01 6.88427567e-01 -6.38454497e-01 -2.50262469e-01 -1.61050379e+00 -8.72333288e-01 6.29190430e-02 -3.73748876e-02 -5.40882885e-01 -6.04307652e-01 1.78215727e-01]
[10.272985458374023, 9.699134826660156]
58a8fcd7-bb16-4214-b771-97a5924156e3
integrating-nearest-neighbors-on-neural
2305.06789
null
https://arxiv.org/abs/2305.06789v2
https://arxiv.org/pdf/2305.06789v2.pdf
Integrating Nearest Neighbors with Neural Network Models for Treatment Effect Estimation
Treatment effect estimation is of high-importance for both researchers and practitioners across many scientific and industrial domains. The abundance of observational data makes them increasingly used by researchers for the estimation of causal effects. However, these data suffer from biases, from several weaknesses, leading to inaccurate causal effect estimations, if not handled properly. Therefore, several machine learning techniques have been proposed, most of them focusing on leveraging the predictive power of neural network models to attain more precise estimation of causal effects. In this work, we propose a new methodology, named Nearest Neighboring Information for Causal Inference (NNCI), for integrating valuable nearest neighboring information on neural network-based models for estimating treatment effects. The proposed NNCI methodology is applied to some of the most well established neural network-based models for treatment effect estimation with the use of observational data. Numerical experiments and analysis provide empirical and statistical evidence that the integration of NNCI with state-of-the-art neural network models leads to considerably improved treatment effect estimations on a variety of well-known challenging benchmarks.
['Christos Diou', 'Niki Kiriakidou']
2023-05-11
null
null
null
null
['causal-inference', 'causal-inference']
['knowledge-base', 'miscellaneous']
[ 4.84799802e-01 -1.65691838e-01 -1.10313404e+00 -3.23772609e-01 -5.58983386e-01 -8.28276128e-02 6.15860105e-01 5.51788509e-01 -2.42553473e-01 1.18041646e+00 7.33787119e-01 -5.38587928e-01 -8.99100125e-01 -9.50049162e-01 -8.41553450e-01 -9.82814372e-01 -3.39892119e-01 2.22099945e-01 -4.00441855e-01 -1.24695860e-02 5.12543142e-01 3.04066151e-01 -1.37396359e+00 -9.71405022e-03 1.19519937e+00 7.60138869e-01 -1.80111200e-01 -5.75099513e-02 2.25285456e-01 5.91957688e-01 -1.58792868e-01 -2.28919178e-01 -1.44792972e-02 -2.07741141e-01 -4.73636359e-01 -6.71743393e-01 3.54711592e-01 -2.34885767e-01 -3.59198332e-01 8.97183895e-01 6.18115723e-01 1.82537094e-01 1.09470963e+00 -1.04650450e+00 -9.34589684e-01 1.04170454e+00 -7.90579855e-01 9.28493142e-02 7.15638325e-02 -6.97814599e-02 9.79055882e-01 -5.08310318e-01 2.51861721e-01 1.33980370e+00 8.46034586e-01 1.10629626e-01 -1.22431660e+00 -9.26666021e-01 7.04758540e-02 5.60106933e-01 -9.92699564e-01 -5.46857566e-02 8.67432415e-01 -5.19490778e-01 3.53631139e-01 1.13749065e-01 5.25479794e-01 1.47004330e+00 4.59903687e-01 5.70376098e-01 1.30330503e+00 -5.26696682e-01 3.00505310e-01 -1.64048567e-01 2.72911876e-01 2.72456765e-01 3.95226032e-01 8.33808005e-01 -3.39903772e-01 -3.07500303e-01 8.75123382e-01 1.95889547e-01 -3.76078367e-01 -2.32989132e-01 -1.04051387e+00 1.33905685e+00 9.74475622e-01 2.78572142e-01 -7.13993907e-01 2.92056829e-01 5.70976019e-01 -1.58093259e-01 9.13600266e-01 2.39481896e-01 -4.82158750e-01 1.41237289e-01 -6.04063511e-01 4.30832446e-01 3.17577332e-01 2.69356787e-01 5.65039292e-02 -3.10508609e-02 -4.52899545e-01 9.06515598e-01 1.37856200e-01 5.11661768e-01 2.45775253e-01 -7.39629567e-01 5.36021769e-01 6.62103832e-01 2.44929627e-01 -1.43599570e+00 -5.78892887e-01 -4.97568071e-01 -1.63797331e+00 4.10826094e-02 6.24154508e-01 -3.29887211e-01 -5.11412144e-01 1.78995907e+00 5.93364179e-01 6.43054128e-01 -2.54362822e-01 8.09477031e-01 7.30893016e-01 7.06512570e-01 4.32232082e-01 -3.49761695e-01 1.14356959e+00 -4.13974077e-01 -8.29959095e-01 4.56329361e-02 5.63886881e-01 -4.16872591e-01 7.87036240e-01 4.56478298e-01 -7.54423976e-01 -3.80190969e-01 -7.95430005e-01 3.87119383e-01 -4.57685620e-01 9.45967138e-02 8.26230943e-01 5.33968270e-01 -4.72846508e-01 1.02278566e+00 -2.49677479e-01 -9.11929458e-02 5.83627760e-01 4.77892488e-01 -3.30397874e-01 -3.26794207e-01 -1.69471955e+00 8.56173158e-01 4.13210332e-01 3.86385679e-01 -6.96865261e-01 -1.14398265e+00 -6.29599333e-01 2.31451079e-01 4.76407975e-01 -8.69721353e-01 1.01108479e+00 -6.08887434e-01 -1.22158945e+00 -1.27044693e-02 1.24604493e-01 -5.07497489e-01 3.35895270e-01 -3.52366030e-01 -3.10731381e-01 -3.65306228e-01 -1.73986122e-01 3.49979162e-01 7.11446285e-01 -1.14860034e+00 -6.31469965e-01 -7.88119674e-01 -8.92983302e-02 4.02976684e-02 -6.93181098e-01 -1.03234157e-01 2.93867052e-01 -8.92519653e-01 -2.65050083e-01 -7.04439163e-01 -4.23593879e-01 -4.45969939e-01 -6.08369350e-01 -6.50887609e-01 5.69995940e-01 -6.18445814e-01 1.46211123e+00 -1.72885776e+00 2.55674630e-01 5.65551184e-02 3.49829555e-01 3.52381378e-01 -1.01399347e-01 4.53506768e-01 -4.81337845e-01 9.38476101e-02 -3.19336861e-01 1.12883277e-01 -1.30686790e-01 -3.53714749e-02 -3.44575644e-01 6.60622895e-01 -5.39924353e-02 6.89418793e-01 -9.36364412e-01 -3.79080743e-01 5.93458056e-01 6.67698562e-01 -5.72320104e-01 1.37251124e-01 -5.95152639e-02 6.79895520e-01 -7.04049587e-01 1.66444793e-01 6.33832276e-01 -5.76151609e-02 1.57996699e-01 -7.11789802e-02 -1.86737090e-01 1.60862833e-01 -1.03774035e+00 1.10549283e+00 -7.12525964e-01 3.38285685e-01 -3.41256171e-01 -1.67279387e+00 5.81860244e-01 6.14638746e-01 4.91319478e-01 -5.34425259e-01 4.56770331e-01 1.29307210e-01 1.05131581e-01 -6.34037495e-01 3.91483344e-02 -3.38141710e-01 -1.44191757e-01 9.00110975e-03 -3.20930272e-01 4.05102015e-01 6.30940646e-02 -3.65538895e-01 7.36778140e-01 -1.43926337e-01 6.87395871e-01 -3.42007428e-01 4.02304173e-01 -2.99491584e-01 6.28377736e-01 8.74538600e-01 4.09633480e-02 2.10778669e-01 4.73464012e-01 -6.44227803e-01 -7.98431695e-01 -6.28418088e-01 -5.06307781e-01 8.70122671e-01 -2.42912769e-01 2.42755637e-01 -7.36177802e-01 -6.76051974e-01 3.04225981e-01 9.18723822e-01 -1.04898024e+00 -4.10643518e-01 -4.80626106e-01 -1.05934858e+00 5.59643805e-01 7.96242774e-01 3.58899504e-01 -1.10942948e+00 -1.85877204e-01 2.14878634e-01 -1.62963778e-01 -5.36774993e-01 2.62078876e-03 8.80829692e-02 -1.12667751e+00 -1.08425117e+00 -1.03193069e+00 -1.75988734e-01 1.63119674e-01 -4.99259457e-02 9.46121514e-01 -2.19052792e-01 2.48849131e-02 -2.99255580e-01 -1.54430807e-01 -5.34254551e-01 -3.53756845e-01 3.61620933e-02 2.06309617e-01 -3.62604647e-03 3.37010056e-01 -8.31026495e-01 -6.35650992e-01 1.99590996e-01 -7.50755727e-01 -2.26091534e-01 8.78480732e-01 1.17238081e+00 1.31433308e-01 2.21354142e-01 1.26761770e+00 -1.17846215e+00 6.82249248e-01 -9.18388784e-01 -8.72772098e-01 4.12447751e-02 -8.06490302e-01 1.44777268e-01 8.68863106e-01 -3.56653452e-01 -1.46019459e+00 -4.53903824e-01 -5.38557507e-02 -2.75349766e-01 -5.11210084e-01 9.87290084e-01 -6.34640232e-02 2.54185885e-01 7.55503237e-01 -2.66767532e-01 -4.02124137e-01 -5.71353257e-01 5.23145199e-01 7.24260271e-01 3.50156844e-01 -3.99915010e-01 3.58736426e-01 1.96926057e-01 4.86648530e-01 -4.60999817e-01 -1.04365456e+00 -3.75120223e-01 -4.73420203e-01 -1.16571307e-01 8.58782709e-01 -5.64385176e-01 -1.03094876e+00 4.99546021e-01 -1.18651116e+00 -2.23063365e-01 1.52757108e-01 8.64225626e-01 -3.50657880e-01 -4.36530150e-02 -4.72878486e-01 -7.83150554e-01 -2.26332992e-01 -1.21004009e+00 8.35775495e-01 2.55755305e-01 -7.28258416e-02 -1.43262994e+00 1.53350830e-01 3.51839423e-01 1.84384659e-01 3.62444431e-01 1.38792312e+00 -4.38892126e-01 -1.28631502e-01 -3.67576331e-01 -5.72053254e-01 -2.23229155e-02 2.77066201e-01 -6.49540722e-02 -8.61065090e-01 9.38463509e-02 -1.77914858e-01 -1.13466606e-02 1.01554728e+00 1.42475915e+00 1.46256006e+00 -5.15472531e-01 -7.09976137e-01 1.64590061e-01 1.42215681e+00 3.78901750e-01 4.81445670e-01 -1.30029889e-02 7.48742878e-01 8.21237206e-01 6.40716970e-01 6.15335882e-01 1.27197742e-01 7.81343877e-01 8.08547318e-01 -2.01977447e-01 3.54334086e-01 -1.67635605e-01 -1.74710795e-01 3.36190760e-01 -5.28257310e-01 -4.16818947e-01 -7.84033358e-01 6.51088059e-01 -2.03841567e+00 -1.00984120e+00 -7.97526538e-01 2.48745584e+00 9.04195130e-01 -5.74764013e-02 1.50399894e-01 2.00694546e-01 9.71437156e-01 2.76743522e-04 -4.38910455e-01 -3.68477702e-01 2.65779912e-01 2.30978459e-01 8.94841373e-01 1.81789935e-01 -1.37646210e+00 3.61375749e-01 6.07485056e+00 1.22563684e+00 -9.36415970e-01 1.52866721e-01 8.78932655e-01 2.03896001e-01 -9.33487117e-02 -1.51379660e-01 -6.58651412e-01 5.07095695e-01 1.05833876e+00 -9.31380764e-02 2.26966605e-01 6.30880117e-01 8.79678965e-01 -1.02799609e-01 -9.93509233e-01 5.31827807e-01 -3.86541277e-01 -1.55501664e+00 -1.17259465e-01 1.54499784e-01 1.19349790e+00 -2.93538511e-01 4.56265926e-01 1.92472935e-01 6.04441583e-01 -1.25053740e+00 1.40296504e-01 4.48816717e-01 5.06635249e-01 -1.05509269e+00 1.17444408e+00 3.23174804e-01 -6.85233951e-01 -5.81861436e-01 -4.62154627e-01 -4.82508093e-01 4.80210371e-02 1.25027347e+00 -5.69007516e-01 9.23071921e-01 7.12463677e-01 9.72766340e-01 -1.08111657e-01 1.14118659e+00 -2.48364151e-01 9.41411138e-01 -9.70444530e-02 -1.90287352e-01 2.93397218e-01 -1.63677350e-01 2.86066413e-01 7.37867594e-01 4.90119398e-01 6.25299066e-02 -2.45431677e-01 8.21223199e-01 -1.62482634e-01 2.57696271e-01 -1.03498185e+00 1.70710713e-01 5.28711796e-01 9.27488804e-01 -3.14421594e-01 -2.84869432e-01 -2.59098262e-01 9.96522829e-02 2.12490425e-01 1.62755683e-01 -1.00248373e+00 3.93687701e-03 3.68678093e-01 -5.46842888e-02 -4.11585867e-02 3.85508925e-01 -6.58799469e-01 -5.61110020e-01 -4.50827688e-01 -7.97739923e-01 7.04107344e-01 -4.07469213e-01 -1.60965335e+00 -1.29493311e-01 5.70474446e-01 -1.07180071e+00 -2.30979130e-01 -6.33727610e-01 -6.31183505e-01 7.31668770e-01 -1.35342336e+00 -9.76136267e-01 -4.48948443e-02 3.17993701e-01 4.31500822e-01 1.45689547e-01 6.91006005e-01 6.20955169e-01 -9.06093180e-01 3.54639739e-01 5.49417794e-01 -1.85459733e-01 9.23308611e-01 -1.29010117e+00 -9.58227515e-02 2.93205708e-01 -2.06878483e-01 5.61085045e-01 7.43052781e-01 -6.56328440e-01 -1.03394604e+00 -1.28845727e+00 9.20814753e-01 -2.01905429e-01 6.07701600e-01 1.41840234e-01 -7.76908755e-01 5.24048567e-01 3.18287402e-01 -1.95602998e-01 4.46315825e-01 7.75441587e-01 -1.60354361e-01 -2.35383034e-01 -9.21755612e-01 7.20804334e-01 8.10054362e-01 1.11928694e-01 -5.31285226e-01 3.80901426e-01 5.56544721e-01 -8.82020395e-04 -1.01308286e+00 6.81903720e-01 5.42868495e-01 -9.40453827e-01 1.30202091e+00 -6.67362571e-01 1.21135950e+00 1.37414992e-01 1.56805411e-01 -1.88143575e+00 -3.30876023e-01 1.19534872e-01 1.70555279e-01 1.12112236e+00 1.63667768e-01 -7.10186243e-01 5.06344378e-01 3.53487968e-01 1.22435629e-01 -7.10492015e-01 -1.05670679e+00 -5.60267389e-01 5.16441882e-01 -3.87540340e-01 6.28723562e-01 1.29985750e+00 1.00476041e-01 5.15669465e-01 -6.81965351e-01 1.55688807e-01 8.73996794e-01 5.49514778e-02 4.56961185e-01 -1.67778671e+00 -5.66124991e-02 -7.90200412e-01 -1.00672528e-01 -4.81355846e-01 2.70899296e-01 -5.88011622e-01 -7.57168531e-02 -1.40632284e+00 2.76709348e-01 -5.27281761e-01 -5.27099013e-01 4.11494762e-01 -5.51033378e-01 8.74571577e-02 -2.63302386e-01 -2.31519252e-01 2.37074032e-01 8.43479693e-01 1.19225287e+00 -2.47696832e-01 -1.58574879e-01 2.92313427e-01 -5.39323032e-01 9.75383401e-01 8.29555809e-01 -7.51086354e-01 -4.19063866e-01 -5.82267754e-02 1.58533230e-01 5.44410110e-01 6.25962973e-01 -8.37105215e-01 -3.31418216e-02 -5.45785129e-01 3.47915024e-01 -4.56973672e-01 -5.28682284e-02 -7.79409826e-01 1.05528489e-01 5.23167312e-01 -7.44067788e-01 -8.39613602e-02 1.53553501e-01 8.35799932e-01 -3.03557575e-01 -2.39440814e-01 5.42740941e-01 1.78042814e-01 -3.85293245e-01 1.56132787e-01 -2.26669610e-01 -3.15714836e-01 8.08051944e-01 3.49845767e-01 -3.51433486e-01 -4.01725084e-01 -4.79414493e-01 4.83104475e-02 -4.42327410e-01 3.48941594e-01 2.39287704e-01 -1.54991424e+00 -8.78247559e-01 -2.87490070e-01 9.68067124e-02 -3.95434797e-01 6.60519958e-01 1.17603600e+00 1.56425774e-01 8.65597785e-01 -1.39818797e-02 -4.49264884e-01 -1.04885781e+00 9.55083251e-01 4.01195772e-02 -7.23270118e-01 -3.59468967e-01 5.01259208e-01 6.74151242e-01 -6.64461493e-01 1.49395719e-01 -4.43762392e-01 -6.85472548e-01 9.68296826e-02 3.52717191e-01 8.38454008e-01 -1.38993626e-02 -2.21727252e-01 5.58355637e-02 3.62296194e-01 3.42340320e-01 2.71180600e-01 1.52783358e+00 2.13178813e-01 -1.41938001e-01 4.22888577e-01 1.06301928e+00 -3.42451036e-01 -8.99715006e-01 -1.45110413e-01 -3.32947411e-02 -4.31801349e-01 5.37934601e-01 -9.64910984e-01 -1.03915310e+00 1.11017287e+00 6.21020138e-01 3.06872040e-01 1.17572606e+00 -5.18711209e-01 3.56672794e-01 1.78732365e-01 2.54258633e-01 -9.26938295e-01 -3.14967334e-01 1.24237902e-01 8.73845220e-01 -1.36612153e+00 -4.53877412e-02 -4.49253947e-01 -8.57782140e-02 8.33438694e-01 3.36081743e-01 -1.92198321e-01 6.95295215e-01 -4.61286791e-02 -2.98578680e-01 -7.25768283e-02 -4.73189205e-01 2.00433612e-01 4.44195718e-01 5.05587518e-01 6.87103987e-01 3.71502459e-01 -7.40839660e-01 5.57080507e-01 2.81226128e-01 4.95458603e-01 2.14985549e-01 2.06344515e-01 2.91902293e-02 -1.09770811e+00 -7.01552570e-01 7.32540131e-01 -5.43212116e-01 -3.25410843e-01 -1.09101526e-01 1.06996322e+00 5.68913184e-02 1.09955752e+00 -1.13651328e-01 -5.76091371e-02 2.47582063e-01 -1.31557733e-01 2.77155250e-01 -4.48507108e-02 -4.78637695e-01 -1.20144039e-01 9.75191742e-02 -4.11364734e-01 -6.90684021e-01 -6.33961737e-01 -7.63937175e-01 -5.85969448e-01 -5.90035379e-01 4.97195609e-02 5.07062495e-01 1.21245694e+00 1.24301381e-01 9.16958749e-01 7.20522344e-01 -9.08958375e-01 -6.78028166e-01 -1.26921058e+00 -4.80865687e-01 2.94007689e-01 1.44233570e-01 -1.17588437e+00 -1.87925637e-01 -3.67592335e-01]
[8.045384407043457, 5.414320945739746]
e9215038-a20d-4c45-8d08-a4407f18f766
hitpr-hierarchical-transformer-for-place
2204.05481
null
https://arxiv.org/abs/2204.05481v1
https://arxiv.org/pdf/2204.05481v1.pdf
HiTPR: Hierarchical Transformer for Place Recognition in Point Cloud
Place recognition or loop closure detection is one of the core components in a full SLAM system. In this paper, aiming at strengthening the relevancy of local neighboring points and the contextual dependency among global points simultaneously, we investigate the exploitation of transformer-based network for feature extraction, and propose a Hierarchical Transformer for Place Recognition (HiTPR). The HiTPR consists of four major parts: point cell generation, short-range transformer (SRT), long-range transformer (LRT) and global descriptor aggregation. Specifically, the point cloud is initially divided into a sequence of small cells by downsampling and nearest neighbors searching. In the SRT, we extract the local feature for each point cell. While in the LRT, we build the global dependency among all of the point cells in the whole point cloud. Experiments on several standard benchmarks demonstrate the superiority of the HiTPR in terms of average recall rate, achieving 93.71% at top 1% and 86.63% at top 1 on the Oxford RobotCar dataset for example.
['Hui Kong', 'Chengzhong Xu', 'Yan Yan', 'Zhixing Hou']
2022-04-12
null
null
null
null
['loop-closure-detection']
['computer-vision']
[-1.11300563e-02 -1.76138371e-01 5.68163930e-04 -2.60547072e-01 -6.72033787e-01 -4.03066188e-01 7.53523529e-01 6.55235350e-01 -5.71774125e-01 5.79498410e-01 -2.55859308e-02 -3.26781720e-02 -4.21311945e-01 -1.07082748e+00 -9.18224871e-01 -7.34747112e-01 -1.94550663e-01 4.57971811e-01 6.19440019e-01 -1.79918692e-01 5.04664242e-01 9.63997662e-01 -1.54780543e+00 -1.69957832e-01 8.43643248e-01 1.39596403e+00 1.45650938e-01 -3.80684361e-02 -4.48633805e-02 6.41979039e-01 -4.51190472e-01 2.25891009e-01 2.26623476e-01 8.10734928e-02 -4.96034116e-01 -1.52053207e-01 2.28214100e-01 -8.17470625e-02 -8.65714997e-02 1.13692605e+00 3.75727862e-01 4.27115202e-01 3.19585204e-01 -1.41824090e+00 -1.28050074e-01 3.65184128e-01 -6.17216706e-01 -1.11501560e-01 2.98963368e-01 -1.23175472e-01 8.95248532e-01 -1.30351496e+00 6.81580007e-01 9.77231145e-01 7.99334109e-01 -1.02049477e-01 -9.90717888e-01 -7.06248522e-01 2.12137833e-01 1.83577150e-01 -1.92386127e+00 -4.10975486e-01 7.44541287e-01 -2.47973040e-01 9.95290220e-01 4.70439978e-02 7.15650678e-01 4.98911649e-01 3.61595511e-01 5.66412568e-01 9.15980637e-01 -1.75550431e-01 4.33027476e-01 -3.30486208e-01 1.30425021e-01 6.67355359e-01 3.68865401e-01 1.18587635e-01 -7.42323101e-01 -2.79631555e-01 1.06219530e+00 3.39317501e-01 -8.06011036e-02 -8.61958861e-01 -1.49595284e+00 6.45943761e-01 1.06620216e+00 2.18863815e-01 -5.81322134e-01 6.21289909e-02 2.52941817e-01 -2.06250735e-02 1.58782929e-01 3.05756122e-01 -1.72952116e-01 -3.06586851e-03 -8.35616708e-01 2.48915717e-01 6.97569430e-01 1.26075375e+00 1.23924196e+00 -4.94609118e-01 -2.20766351e-01 8.40990543e-01 2.07832113e-01 4.16638374e-01 2.67849624e-01 -7.16089487e-01 6.06238425e-01 9.49086428e-01 1.74466550e-01 -1.39508379e+00 -5.79808116e-01 -6.34446144e-01 -1.06234193e+00 -4.43121931e-03 2.74753384e-02 2.61328310e-01 -8.60962272e-01 1.51677501e+00 5.49129903e-01 2.85645962e-01 -1.25762550e-02 8.45763385e-01 6.36270940e-01 6.82021797e-01 -1.55898228e-01 1.22587718e-01 1.28631794e+00 -1.02312243e+00 -2.92376399e-01 -3.52037489e-01 7.12287784e-01 -5.74893117e-01 4.44688469e-01 1.30665705e-01 -7.11644173e-01 -5.58227301e-01 -1.23217595e+00 -3.65975380e-01 -4.93470043e-01 2.25261748e-01 4.94910121e-01 -1.94792539e-01 -9.28855479e-01 4.80682731e-01 -8.18462670e-01 -3.75235349e-01 2.29456604e-01 5.42373359e-01 -7.41780937e-01 -3.32526296e-01 -8.93806338e-01 6.93319082e-01 4.00861919e-01 4.50926930e-01 -6.53624058e-01 -4.39361572e-01 -1.08644974e+00 7.45166466e-02 3.63571167e-01 -5.30621290e-01 8.32288384e-01 -1.36843085e-01 -1.07289374e+00 4.57117677e-01 -3.81908029e-01 -5.67830622e-01 3.74990314e-01 -2.81205893e-01 -1.71512693e-01 1.16633356e-03 5.16518354e-01 7.19170094e-01 4.50036734e-01 -1.26715112e+00 -1.14044988e+00 -6.19942963e-01 -4.92206141e-02 3.50113630e-01 2.53356427e-01 -5.51247597e-01 -5.67044139e-01 -3.08681697e-01 8.45227003e-01 -1.01194537e+00 -4.28976506e-01 -9.43458155e-02 -3.79229248e-01 -4.58662421e-01 7.32103527e-01 -2.09755510e-01 9.81915653e-01 -2.41647911e+00 -9.03973952e-02 6.15254104e-01 2.93335140e-01 -1.03159763e-01 -1.04514353e-01 5.90770781e-01 6.24153279e-02 -3.52005690e-01 -6.24926537e-02 -3.23361784e-01 3.68584841e-02 3.15550685e-01 -2.79559970e-01 6.42982543e-01 1.41795695e-01 8.87021184e-01 -9.94826376e-01 -4.16187376e-01 4.61203694e-01 4.66320097e-01 -5.28212249e-01 -2.51567245e-01 2.55414456e-01 1.54407591e-01 -7.24778950e-01 8.15441251e-01 9.00302231e-01 -4.01268564e-02 -2.81086177e-01 -2.84904510e-01 -6.14670157e-01 4.65546042e-01 -1.34695375e+00 2.00475383e+00 -2.96770245e-01 1.75935313e-01 -6.37331456e-02 -4.09828424e-01 1.36730623e+00 -1.95028752e-01 5.79469144e-01 -1.02351272e+00 -1.24208229e-02 4.90950882e-01 -2.34665230e-01 1.73971877e-01 9.16819632e-01 1.99365556e-01 -3.81753474e-01 -2.15828605e-02 2.18992233e-02 1.25497445e-01 5.61583154e-02 -4.01472375e-02 1.17948341e+00 -1.83803197e-02 5.21073520e-01 -3.66440028e-01 7.78686941e-01 1.37930781e-01 7.01267064e-01 6.43801272e-01 -1.82695746e-01 5.95942855e-01 4.32118684e-01 -5.91813445e-01 -7.46870935e-01 -9.60022092e-01 3.94171774e-02 6.82105303e-01 8.60133171e-01 -4.83860075e-01 -3.33986551e-01 -4.93274629e-01 1.97270691e-01 4.31365907e-01 -6.21788979e-01 -3.36101621e-01 -6.70086741e-01 -3.61770779e-01 4.04886037e-01 6.38915300e-01 8.19607437e-01 -9.70795691e-01 -8.70630205e-01 2.13480860e-01 -2.37529382e-01 -1.11495018e+00 -2.65737534e-01 5.70695639e-01 -7.55057216e-01 -8.30287576e-01 -2.83580542e-01 -8.50204051e-01 8.11278224e-01 5.31426370e-01 6.57635272e-01 4.47936505e-02 2.24400267e-01 -3.34521770e-01 -4.70320076e-01 -1.24051273e-01 3.83845985e-01 3.75357836e-01 8.99270251e-02 9.14531127e-02 3.26514512e-01 -6.55886948e-01 -5.90319753e-01 4.52811003e-01 -4.33833838e-01 -1.58518136e-01 9.24098313e-01 7.99433291e-01 1.18850017e+00 -1.28142023e-02 5.19078113e-02 -4.53471750e-01 2.61094809e-01 -3.22737694e-01 -7.10775197e-01 -2.18980294e-02 -3.48162264e-01 6.41421899e-02 3.25059831e-01 -1.60044417e-01 -4.69928265e-01 4.98995483e-01 -1.45404786e-01 -3.76297355e-01 -1.27203777e-01 5.04118681e-01 -2.24849954e-01 -3.83948952e-01 4.28616494e-01 5.09035707e-01 -2.04953775e-01 -4.55754608e-01 1.32543683e-01 3.11659694e-01 6.18219912e-01 -4.77129310e-01 9.19929266e-01 6.62819207e-01 2.29133844e-01 -6.54916286e-01 -6.14527941e-01 -7.98297584e-01 -9.17009056e-01 -9.10853744e-02 6.30846202e-01 -9.74261940e-01 -6.74368501e-01 3.78514916e-01 -1.12717807e+00 8.98050815e-02 -5.60014725e-01 4.37399179e-01 -3.80885422e-01 6.28432333e-02 -3.22327197e-01 -5.22769570e-01 -2.87151158e-01 -1.26836741e+00 1.43014109e+00 2.48491660e-01 9.86074284e-02 -4.69584972e-01 1.67331100e-01 -1.10074617e-01 1.69039011e-01 3.64986509e-01 5.91895342e-01 -8.21123004e-01 -8.69155943e-01 -4.35437590e-01 -3.96026313e-01 -1.03736997e-01 1.00790940e-01 -3.94788176e-01 -7.35897899e-01 -2.89486647e-01 -9.19597521e-02 1.99095741e-01 8.04409504e-01 1.07159816e-01 7.64010489e-01 4.20710854e-02 -6.80981040e-01 7.50101805e-01 1.60110545e+00 2.72535086e-01 7.14380085e-01 6.42313898e-01 7.21066415e-01 3.45852971e-01 1.00237679e+00 3.78029913e-01 5.02607942e-01 5.34038424e-01 6.98845744e-01 -3.82692181e-02 1.85321108e-01 -6.46574259e-01 2.36862853e-01 7.15557396e-01 1.86441913e-02 3.58841754e-02 -9.73693728e-01 7.44412124e-01 -2.07957506e+00 -6.35051847e-01 -1.64496288e-01 2.42044497e+00 1.41568750e-01 1.04169153e-01 -1.85750082e-01 2.06499338e-01 7.66881645e-01 3.25185627e-01 -3.49955082e-01 6.93056658e-02 -1.42455682e-01 -8.87421891e-02 8.50685596e-01 4.74657834e-01 -1.18796766e+00 9.90117669e-01 4.78750277e+00 7.92238235e-01 -1.06447470e+00 -1.02474429e-01 1.90619737e-01 1.51229605e-01 2.98594713e-01 2.46262014e-01 -1.00008285e+00 1.20287329e-01 3.55520338e-01 1.51834562e-01 2.34642893e-01 8.83602798e-01 -9.74095762e-02 -4.00382757e-01 -9.96006250e-01 9.95437622e-01 -6.86236992e-02 -1.18286324e+00 1.46997750e-01 3.23662549e-01 5.25155067e-01 5.08474827e-01 -2.06402272e-01 2.60674953e-01 -9.76579860e-02 -6.83684587e-01 1.05348706e+00 4.04028475e-01 5.36254525e-01 -1.05444074e+00 9.84423041e-01 5.19207537e-01 -1.68529284e+00 -7.52711073e-02 -5.27634799e-01 2.67070774e-02 6.54007271e-02 7.39823043e-01 -7.93057561e-01 9.61402178e-01 7.50768840e-01 8.66149008e-01 -6.34143054e-01 1.30836463e+00 -1.23032920e-01 -2.61577308e-01 -8.63499641e-01 7.59928226e-02 5.21941185e-01 -2.43531793e-01 5.36770046e-01 9.91907358e-01 4.44863588e-01 -6.40658289e-02 2.83457667e-01 6.10657752e-01 9.47000384e-02 4.77060266e-02 -6.39211953e-01 5.14776230e-01 9.94256556e-01 1.42767107e+00 -9.32273805e-01 -1.48085624e-01 -9.78866890e-02 7.49382496e-01 4.78194803e-01 7.89418444e-02 -7.06997395e-01 -7.20090747e-01 5.75917125e-01 -4.52940874e-02 6.02399588e-01 -4.08875555e-01 -3.25977743e-01 -8.80338907e-01 3.04659933e-01 -4.29791838e-01 2.37471238e-01 -6.31305933e-01 -8.90375495e-01 6.65813744e-01 -1.90944925e-01 -1.62684035e+00 -8.45775530e-02 -4.44251269e-01 -3.16953033e-01 8.38054478e-01 -1.76336122e+00 -1.19305563e+00 -7.31573462e-01 7.29998529e-01 3.76576930e-01 3.70192915e-01 6.07132196e-01 2.33023763e-01 -4.43731576e-01 2.78043479e-01 1.98864695e-02 1.62944481e-01 4.53445584e-01 -9.13176179e-01 4.86259758e-01 8.56775582e-01 1.36545390e-01 9.54104304e-01 2.49837890e-01 -7.13822663e-01 -1.26199043e+00 -1.18595052e+00 1.25737250e+00 -1.62792608e-01 5.31255662e-01 -5.20886600e-01 -7.15348363e-01 6.25117600e-01 -4.35927033e-01 2.84334689e-01 2.81310733e-02 -1.60632674e-02 -2.81798452e-01 -4.09456879e-01 -1.07499313e+00 5.20592988e-01 1.11963642e+00 -4.98914748e-01 -5.13509214e-01 1.55870050e-01 5.05134463e-01 -7.07404673e-01 -9.45057929e-01 6.58135295e-01 4.39952880e-01 -1.06356275e+00 1.03228664e+00 3.48989576e-01 1.31513923e-01 -9.21424568e-01 -3.08822423e-01 -1.00861335e+00 -5.42384982e-01 -3.17079216e-01 1.52828947e-01 1.19285893e+00 3.10455203e-01 -8.14864337e-01 6.81093276e-01 1.14354819e-01 -3.80148858e-01 -8.66587281e-01 -1.39713466e+00 -8.32004249e-01 -3.06796432e-01 -3.70761007e-01 1.04733336e+00 7.42376804e-01 -1.67144805e-01 4.55075175e-01 -8.14924389e-03 5.78520477e-01 5.43362498e-01 3.91482085e-01 9.24957991e-01 -1.44014704e+00 4.31219459e-01 -2.84622252e-01 -6.17482841e-01 -1.34808612e+00 -2.59402871e-01 -7.05315471e-01 4.16968733e-01 -1.75717151e+00 -2.15745214e-02 -7.29827762e-01 -5.75689435e-01 5.08598089e-01 2.12256819e-01 4.22804579e-02 2.54626751e-01 6.04276836e-01 -9.16727245e-01 7.19588101e-01 1.03421378e+00 1.21585935e-01 -2.98415691e-01 -1.62916899e-01 -3.10360372e-01 7.24338710e-01 5.68630934e-01 -5.65286636e-01 -1.65580750e-01 -4.37092096e-01 8.87235478e-02 4.85788099e-03 6.20247841e-01 -1.54365110e+00 8.35073233e-01 1.17418757e-02 3.53919506e-01 -1.29156959e+00 5.74462831e-01 -1.01373661e+00 1.87241301e-01 5.28929830e-01 3.43365781e-02 4.03407663e-01 9.67703015e-02 6.76416457e-01 -5.40704429e-01 1.13903083e-01 5.84642768e-01 9.60184261e-02 -1.04002273e+00 3.46787333e-01 1.81764215e-01 -4.33644235e-01 1.06697476e+00 -4.10608083e-01 -3.55395228e-01 1.28002644e-01 -2.77014971e-01 4.25423831e-01 6.89434946e-01 5.03939927e-01 7.83818543e-01 -1.48497784e+00 -2.01612517e-01 4.92528796e-01 4.93011951e-01 6.30910695e-01 1.53505087e-01 1.21736467e+00 -6.70794547e-01 5.77328861e-01 -2.04152182e-01 -8.11268151e-01 -8.33249509e-01 4.66694504e-01 2.87754357e-01 -3.57459486e-01 -8.05363238e-01 7.91659415e-01 2.97416836e-01 -3.94319892e-01 1.30085364e-01 -7.02444911e-01 -2.79405177e-01 9.56625715e-02 1.86208189e-01 3.29482675e-01 2.59926170e-01 -1.02689910e+00 -7.30600059e-01 9.57294941e-01 2.84237396e-02 4.22220826e-02 1.33062053e+00 -1.63285881e-01 -5.02597809e-01 4.77319300e-01 1.06144559e+00 3.46882315e-03 -1.06956267e+00 -2.32284710e-01 1.69546306e-01 -2.80062348e-01 5.30119846e-03 -4.30371672e-01 -9.16810334e-01 4.76020068e-01 3.42929959e-01 -1.97280962e-02 1.06689167e+00 -9.21852365e-02 8.43168139e-01 6.69257879e-01 1.08389747e+00 -8.73681247e-01 -4.93148804e-01 9.66714561e-01 8.23530316e-01 -8.81074429e-01 4.87713851e-02 -5.10644615e-01 -3.25974971e-01 9.14953589e-01 4.27939415e-01 -5.99878550e-01 5.28813660e-01 2.18993388e-02 -2.47611657e-01 -4.37040776e-01 -5.03533483e-01 -2.01306462e-01 1.58288240e-01 3.35815996e-01 -1.89363599e-01 -1.07208848e-01 -1.87024951e-01 3.57061893e-01 -4.19159323e-01 -8.67413878e-02 -7.25922137e-02 1.21703696e+00 -6.69822991e-01 -7.46324360e-01 -3.60828847e-01 2.60910720e-01 1.45012617e-01 -5.98951168e-02 -4.28107500e-01 9.61714268e-01 4.26142663e-01 6.89859569e-01 2.87951529e-01 -7.36416042e-01 6.87754989e-01 -2.47510418e-01 1.04844354e-01 -3.63058805e-01 -6.60287857e-01 1.08057663e-01 -1.47029042e-01 -1.00382805e+00 -2.38273978e-01 -8.06172729e-01 -1.43632567e+00 -9.57513452e-02 -3.18946898e-01 3.85076582e-01 6.99699938e-01 9.17026460e-01 6.61257684e-01 2.95008242e-01 5.22550583e-01 -1.07262516e+00 -1.94117680e-01 -9.15219724e-01 -5.71889162e-01 1.02782054e-02 6.47521377e-01 -8.64592016e-01 -3.09544146e-01 -5.29922962e-01]
[7.588040828704834, -2.1698057651519775]
98374afd-5fa0-4c34-806f-eb2e41b24176
continuous-human-activity-recognition-using-a
2304.06173
null
https://arxiv.org/abs/2304.06173v1
https://arxiv.org/pdf/2304.06173v1.pdf
Continuous Human Activity Recognition using a MIMO Radar for Transitional Motion Analysis
The prompt and accurate recognition of Continuous Human Activity (CHAR) is critical in identifying and responding to health events, particularly fall risk assessment. In this paper, we examine a multi-antenna radar system that can process radar data returns for multiple individuals in an indoor setting, enabling CHAR for multiple subjects. This requires combining spatial and temporal signal processing techniques through micro-Doppler (MD) analysis and high-resolution receive beamforming. We employ delay and sum beamforming to capture MD signatures at three different directions of observation. As MD images may contain multiple activities, we segment the three MD signatures using an STA/LTA algorithm. MD segmentation ensures that each MD segment represents a single human motion activity. Finally, the segmented MD image is resized and processed through a convolutional neural network (CNN) to classify motion against each MD segment.
['Syed A. Hamza', 'LaJuan Washington Jr.', 'Bennett J. Richman', 'John Kobak']
2023-04-12
null
null
null
null
['human-activity-recognition', 'human-activity-recognition']
['computer-vision', 'time-series']
[ 7.59628832e-01 -6.25401199e-01 -3.98967750e-02 -1.85942322e-01 -1.00866592e+00 -3.73482019e-01 2.76015162e-01 -7.26807788e-02 -4.49291945e-01 5.18988013e-01 3.97199839e-01 -1.59267947e-01 -3.31863165e-01 -7.75629103e-01 -5.93228154e-02 -6.51622355e-01 -7.17097700e-01 2.08174065e-01 3.00380498e-01 1.72423482e-01 -2.71625519e-01 9.12991703e-01 -1.26995921e+00 4.64052916e-01 6.61587059e-01 1.23811650e+00 -2.30090976e-01 1.32880449e+00 5.94436944e-01 6.72177792e-01 -8.90481293e-01 3.05877179e-01 3.98510367e-01 -5.53567111e-01 3.65933962e-02 -4.39970642e-01 7.51572967e-01 -7.40727663e-01 -6.90559685e-01 5.07591546e-01 8.55116069e-01 2.10008860e-01 6.94446146e-01 -1.17102170e+00 -7.52960816e-02 3.88318710e-02 -5.73740661e-01 7.94148386e-01 7.96697736e-01 -3.04487720e-02 1.90355435e-01 -7.73933530e-01 -3.18442397e-02 9.68641222e-01 7.07412958e-01 2.87808657e-01 -8.48744690e-01 -7.38370955e-01 -2.69120127e-01 3.19631487e-01 -1.40073276e+00 -4.15633142e-01 6.49673462e-01 -4.14207786e-01 5.89318395e-01 6.06599569e-01 8.89241338e-01 1.17540765e+00 8.16896737e-01 5.79259992e-01 7.24061549e-01 -2.31334612e-01 2.03122452e-01 -9.93741512e-01 2.55722553e-01 6.05935872e-01 6.70514941e-01 4.76634085e-01 -8.71731937e-01 -3.62545907e-01 8.49939525e-01 1.84314549e-01 -2.76601225e-01 1.07220434e-01 -1.35803735e+00 4.11102235e-01 3.11888903e-01 1.30275294e-01 -8.57834816e-01 5.83268106e-01 -1.16879918e-01 -7.67512694e-02 -1.51708469e-01 5.89880794e-02 2.51746982e-01 -1.17974415e-01 -1.36339998e+00 4.12411183e-01 4.83697891e-01 5.22161007e-01 2.55141288e-01 3.48932564e-01 -5.82026958e-01 1.97325498e-01 6.24159753e-01 1.46157861e+00 8.34637731e-02 -1.01164722e+00 2.52464116e-01 -4.02774755e-03 2.43496820e-01 -1.03071535e+00 -9.29711163e-01 -7.75848210e-01 -1.02408934e+00 1.81083053e-01 4.50913876e-01 -6.42099738e-01 -9.62013125e-01 1.22690248e+00 2.33424544e-01 5.74632466e-01 6.09206446e-02 1.11448324e+00 5.49905479e-01 3.54577780e-01 1.94008842e-01 -3.22350740e-01 1.60131061e+00 -2.37938344e-01 -6.86218143e-01 -7.34573066e-01 -8.98140017e-03 -4.20995802e-01 -6.96926117e-02 5.57434618e-01 -8.03174555e-01 -5.89598060e-01 -1.38419628e+00 6.29382610e-01 2.50327140e-01 2.20766515e-01 3.44379634e-01 9.43365932e-01 -4.24087435e-01 -1.24540061e-01 -1.06773210e+00 -6.41886378e-03 4.07934606e-01 1.55927420e-01 2.15484753e-01 -4.14637119e-01 -1.34778035e+00 7.57179499e-01 -3.89456272e-01 4.76763636e-01 -8.35369289e-01 -6.96121693e-01 -9.30094004e-01 -2.77879506e-01 -2.14114591e-01 -6.29384398e-01 1.14716709e+00 -4.37924892e-01 -8.56213570e-01 1.01871446e-01 -3.10999751e-01 -7.37135470e-01 2.93590069e-01 -6.66919291e-01 -1.24732494e+00 6.63633764e-01 1.99104935e-01 1.78167999e-01 1.11760581e+00 -8.96029651e-01 -8.40169191e-01 -5.25640607e-01 -4.57557440e-01 5.59882149e-02 3.97556156e-01 6.51376620e-02 1.58590227e-02 -6.08095229e-01 4.93416399e-01 -6.50702298e-01 -3.25255871e-01 -9.57470611e-02 -1.27996668e-01 6.12117350e-01 6.59393609e-01 -7.69482613e-01 1.52421069e+00 -1.99075043e+00 -4.43972975e-01 6.44152880e-01 3.51829797e-01 2.87408736e-02 6.77159503e-02 3.46003622e-02 4.48548198e-01 -7.62625873e-01 -2.43616462e-01 2.53946602e-01 -2.79048979e-01 -1.53736889e-01 -3.10302675e-01 8.48833978e-01 1.12560824e-01 9.14435923e-01 -7.87348509e-01 -1.98907167e-01 1.71119019e-01 5.51779091e-01 -2.04302698e-01 -2.00870913e-02 2.82130986e-01 4.96643722e-01 -6.45025551e-01 1.01015961e+00 8.90331268e-01 2.04360917e-01 5.67796193e-02 -5.27655661e-01 -3.33043486e-01 -2.48275891e-01 -1.37337828e+00 1.16453397e+00 -1.85409755e-01 7.18063474e-01 2.46124640e-01 -7.78695762e-01 8.69697154e-01 2.44227991e-01 1.23076558e+00 -1.19940424e+00 4.01871875e-02 5.98825179e-02 6.10865951e-02 -5.63540459e-01 4.45003539e-01 -1.30088285e-01 -5.76471388e-01 6.76316738e-01 -4.60440069e-01 8.36854875e-01 -1.18909314e-01 -8.51540565e-02 1.79228520e+00 -4.59522396e-01 2.54366249e-01 5.20375669e-02 2.38413543e-01 1.52346149e-01 6.50566697e-01 1.10705280e+00 -5.76153517e-01 5.21394968e-01 -4.95815128e-01 -2.02971444e-01 -2.09068716e-01 -1.94595599e+00 -7.31194988e-02 9.27569211e-01 3.61979514e-01 1.64716646e-01 -1.32353529e-01 -3.24672490e-01 2.15050966e-01 4.84862268e-01 -1.74121603e-01 -2.63565451e-01 -1.17812431e+00 -8.92814696e-01 1.11429334e+00 7.49071062e-01 7.30591655e-01 -6.04035676e-01 -1.75823128e+00 7.21482337e-01 -5.19994915e-01 -9.74434972e-01 -4.08470839e-01 1.88899130e-01 -4.53401238e-01 -1.30335975e+00 -8.72258306e-01 -4.56261754e-01 2.89398670e-01 7.79665649e-01 7.25168765e-01 -2.71292955e-01 -8.09846342e-01 9.61201251e-01 -2.72985250e-01 -5.60834169e-01 3.43252599e-01 -5.04422188e-01 2.32305303e-01 2.35471874e-01 6.70032084e-01 -4.38023001e-01 -9.56488609e-01 2.16988042e-01 -3.83812606e-01 -4.05861974e-01 9.29348409e-01 2.54006892e-01 1.96112588e-01 -1.35328114e-01 6.31365478e-01 -2.59473566e-02 6.41814232e-01 -4.58089471e-01 -2.98150033e-01 1.29559398e-01 -1.26639068e-01 -2.95014024e-01 2.82746516e-02 -3.54430050e-01 -1.05640221e+00 2.83286482e-01 -5.44668213e-02 -8.49770531e-02 -4.00043011e-01 5.30116260e-01 -8.65799095e-03 2.67767045e-03 1.05148101e+00 3.81074488e-01 -6.33363649e-02 -1.77067090e-02 2.43047357e-01 6.72255516e-01 1.15780139e+00 -1.29183888e-01 8.90333593e-01 7.47670889e-01 2.77406365e-01 -1.21825039e+00 -5.24105549e-01 -7.72798955e-01 -6.85250938e-01 -8.10100138e-01 1.15397000e+00 -1.06297660e+00 -6.72758639e-01 3.87805283e-01 -9.44032073e-01 -7.11428300e-02 1.27906367e-01 1.06474304e+00 -2.94886410e-01 2.20125049e-01 -2.59388030e-01 -1.10857177e+00 -5.39769053e-01 -4.36531842e-01 1.10401773e+00 4.23705965e-01 -5.79801738e-01 -5.00155747e-01 3.23523611e-01 2.19992802e-01 5.75728536e-01 6.26232147e-01 1.90671355e-01 -1.26084149e-01 -7.68153608e-01 -4.90257114e-01 1.89140394e-01 -3.95763338e-01 5.99625222e-02 -5.59572279e-01 -7.87318170e-01 -1.67572096e-01 -7.69010335e-02 2.31444642e-01 7.89565563e-01 1.18491340e+00 3.96670997e-01 -1.00185433e-02 -7.19314456e-01 6.04853451e-01 1.04049277e+00 5.95238626e-01 7.64599204e-01 1.66748136e-01 4.63162333e-01 5.69611527e-02 9.77864504e-01 7.81765282e-01 1.99193642e-01 6.74882650e-01 1.44563705e-01 -3.31196301e-02 -3.03955197e-01 3.55829120e-01 6.80843532e-01 -1.01400569e-01 -1.53622717e-01 -1.72295615e-01 -8.12542915e-01 2.11147010e-01 -1.53426683e+00 -1.55044329e+00 -3.83814096e-01 2.13899446e+00 9.87508744e-02 -8.08596332e-03 3.22224617e-01 8.02923068e-02 5.03507137e-01 1.74385995e-01 -4.81126219e-01 4.88776639e-02 -2.89120376e-02 5.29146373e-01 7.59079099e-01 3.75426531e-01 -1.29780555e+00 -3.81572209e-02 6.90488291e+00 3.28132138e-02 -1.03574634e+00 -9.61439908e-02 -2.49247089e-01 -4.08110380e-01 -4.13409919e-02 -6.64304972e-01 -7.72191823e-01 2.36389801e-01 1.18094015e+00 -5.98333366e-02 -1.73247997e-02 2.92099625e-01 4.71978664e-01 -7.02204823e-01 -6.91772640e-01 9.97431993e-01 1.91929445e-01 -1.05154252e+00 -5.08161306e-01 1.86359078e-01 1.23867847e-01 -8.63271207e-02 -9.39734951e-02 1.48104310e-01 1.90541625e-01 -9.15321290e-01 6.93121076e-01 9.99873519e-01 6.65407240e-01 -7.33448207e-01 2.47503430e-01 3.41209561e-01 -1.82085776e+00 -3.55423123e-01 1.42369479e-01 4.49612252e-02 7.06659198e-01 8.16426039e-01 -5.66109836e-01 7.59531975e-01 5.51093817e-01 1.71478406e-01 -3.81523758e-01 1.25618410e+00 8.61325264e-02 5.45929253e-01 -3.89088213e-01 2.39219993e-01 -1.83261573e-01 3.76106310e-03 6.64089561e-01 1.44503045e+00 8.87231529e-01 5.79626441e-01 3.91932487e-01 3.97755474e-01 6.92230523e-01 -8.28079522e-01 -5.11739790e-01 3.88408214e-01 6.29916906e-01 1.02440798e+00 -4.40754563e-01 -1.32617712e-01 -4.19821352e-01 7.84621000e-01 -6.57270730e-01 3.91204327e-01 -1.03803051e+00 -5.14840662e-01 7.70421207e-01 3.20359886e-01 1.62516236e-01 -9.15696144e-01 -3.54104370e-01 -5.38677871e-01 -1.23154007e-01 -4.57272768e-01 6.93373919e-01 -4.97591525e-01 -9.48404789e-01 3.20921749e-01 1.76589400e-01 -1.60518730e+00 -4.50536311e-01 -3.40041339e-01 -7.57739007e-01 8.09017301e-01 -1.16661632e+00 -1.08202136e+00 -8.60562027e-01 6.64119840e-01 1.86629698e-01 -1.17473722e-01 6.72685921e-01 7.86484778e-01 -3.11132431e-01 3.93903375e-01 -2.66048878e-01 3.71605873e-01 6.09644473e-01 -8.28450680e-01 2.23303691e-01 1.35510457e+00 -1.74592942e-01 2.75623173e-01 6.87715769e-01 -1.31313527e+00 -1.55340910e+00 -1.43050802e+00 6.43438876e-01 -2.80027390e-01 1.34738952e-01 -7.60607142e-03 -4.41087842e-01 6.10372722e-01 -2.07239255e-01 6.29242212e-02 9.90846217e-01 -4.20712292e-01 1.83612406e-01 -1.78411558e-01 -1.07303584e+00 3.52473646e-01 1.06300867e+00 -2.67201334e-01 -7.14137435e-01 -1.35178611e-01 -1.16468698e-01 -3.45188916e-01 -7.13286996e-01 5.72665155e-01 1.19913125e+00 -6.28113329e-01 1.52385378e+00 -6.30985871e-02 -2.46011913e-01 -5.38545668e-01 -3.42678726e-01 -9.54923868e-01 -7.80726552e-01 -1.65204406e-01 -6.81583405e-01 6.42438114e-01 -1.82165682e-01 -2.26536036e-01 8.45988870e-01 2.43508309e-01 -2.84207851e-01 -9.80735421e-02 -1.19324374e+00 -8.89751256e-01 -6.27104640e-01 -8.31269324e-01 2.37494618e-01 3.50210249e-01 -1.52247846e-01 -5.75639941e-02 -5.39318919e-01 7.34554589e-01 1.20487416e+00 1.61316514e-01 4.00786072e-01 -1.32414675e+00 -1.59325644e-01 -1.34324268e-01 -4.01596725e-01 -1.27742684e+00 -4.48860109e-01 -6.69561982e-01 3.98887396e-01 -1.82740533e+00 -4.93288547e-01 -2.39652768e-01 -9.29552391e-02 6.16264045e-02 1.05951428e-01 3.77627730e-01 -1.55347124e-01 7.57455826e-02 -3.90975714e-01 3.10718000e-01 8.77929866e-01 -2.97021627e-01 -2.91321546e-01 6.40083849e-01 -4.70132440e-01 5.05792797e-01 7.73759663e-01 -3.07139456e-01 -1.06023952e-01 -1.35747477e-01 -2.47637659e-01 3.74461561e-01 9.55498338e-01 -1.89971662e+00 6.10892296e-01 -3.37062627e-01 1.21396267e+00 -1.22082663e+00 6.14075899e-01 -7.89392829e-01 3.64228874e-01 1.06721687e+00 4.24885228e-02 -2.96014529e-02 4.69340235e-02 7.23514140e-01 1.65377125e-01 3.18181366e-01 7.67112792e-01 9.18184891e-02 -8.27763498e-01 2.61480153e-01 -1.22199130e+00 -1.96283102e-01 1.02212512e+00 -4.41196471e-01 -4.77937281e-01 -3.45176429e-01 -6.82349682e-01 2.28117049e-01 -4.62552518e-01 2.54831225e-01 1.16085494e+00 -1.71778882e+00 -7.68925130e-01 4.97093618e-01 -1.60540063e-02 -3.74862134e-01 5.39921999e-01 1.15809131e+00 -4.39971000e-01 4.29156959e-01 -5.20889699e-01 -7.12423146e-01 -1.27873540e+00 -1.19093254e-01 6.07550681e-01 6.21669292e-02 -8.99783611e-01 4.35839832e-01 -3.86524796e-01 5.46618998e-01 4.55730818e-02 -1.91576868e-01 -3.50808054e-01 2.11769566e-01 1.00341165e+00 8.84118438e-01 -4.23601083e-02 -7.65565753e-01 -9.21735942e-01 5.73600769e-01 4.70293492e-01 -6.13635004e-01 8.01378071e-01 -1.80828527e-01 5.19444466e-01 -7.55927935e-02 6.01994574e-01 1.40786842e-01 -1.30448973e+00 9.05926432e-03 -2.68098060e-02 -3.26523304e-01 3.67204100e-02 -6.56863451e-01 -7.61694372e-01 6.73626542e-01 1.16462064e+00 3.78054641e-02 1.31992030e+00 -1.45122960e-01 9.92377996e-01 3.03561598e-01 4.38618898e-01 -8.48190427e-01 2.17830986e-01 4.72356886e-01 6.49816096e-01 -5.15096188e-01 1.01998992e-01 -1.66637868e-01 -4.15537149e-01 1.23329186e+00 4.86537546e-01 -1.96998775e-01 8.13874304e-01 6.86879456e-01 3.34475160e-01 -4.27781075e-01 -2.15360656e-01 -4.11193162e-01 5.14208257e-01 1.18617094e+00 1.06896788e-01 2.39967018e-01 -6.57741874e-02 8.22956264e-01 -7.55599737e-02 5.48830889e-02 3.27506840e-01 1.37266982e+00 -1.18714559e+00 -5.90177774e-01 -1.07903850e+00 8.85805786e-01 -1.04425661e-01 3.20250779e-01 -9.26181301e-02 3.45797896e-01 2.11534813e-01 1.14075887e+00 2.81814903e-01 -4.54327166e-01 6.48513258e-01 -7.57463649e-02 6.80549324e-01 -2.91562498e-01 -3.59359890e-01 2.63949633e-01 2.13031605e-01 -6.54007912e-01 -3.02918583e-01 -1.10775280e+00 -1.58749628e+00 9.74555612e-02 3.65040302e-01 -2.49796316e-01 2.75083750e-01 9.93595481e-01 1.78706273e-01 1.15256345e+00 4.62513328e-01 -7.87486494e-01 -3.50797743e-01 -5.16819596e-01 -7.32868731e-01 -2.16492578e-01 6.40123069e-01 -7.49928594e-01 -2.67817006e-02 8.39353129e-02]
[6.775967597961426, 0.516135036945343]
da883953-16ac-41a5-9f78-f3f65e7dce4f
active-sequential-two-sample-testing
2301.12616
null
https://arxiv.org/abs/2301.12616v3
https://arxiv.org/pdf/2301.12616v3.pdf
Active Sequential Two-Sample Testing
Two-sample testing tests whether the distributions generating two samples are identical. We pose the two-sample testing problem in a new scenario where the sample measurements (or sample features) are inexpensive to access, but their group memberships (or labels) are costly. We devise the first \emph{active sequential two-sample testing framework} that not only sequentially but also \emph{actively queries} sample labels to address the problem. Our test statistic is a likelihood ratio where one likelihood is found by maximization over all class priors, and the other is given by a classification model. The classification model is adaptively updated and then used to guide an active query scheme called bimodal query to label sample features in the regions with high dependency between the feature variables and the label variables. The theoretical contributions in the paper include proof that our framework produces an \emph{anytime-valid} $p$-value; and, under reachable conditions and a mild assumption, the framework asymptotically generates a minimum normalized log-likelihood ratio statistic that a passive query scheme can only achieve when the feature variable and the label variable have the highest dependence. Lastly, we provide a \emph{query-switching (QS)} algorithm to decide when to switch from passive query to active query and adapt bimodal query to increase the testing power of our test. Extensive experiments justify our theoretical contributions and the effectiveness of QS.
['Visar Berisha', 'Gautam Dasarathy', 'Pouria Saidi', 'Prad Kadambi', 'Karthikeyan Natesan Ramamurthy', 'Weizhi Li']
2023-01-30
null
null
null
null
['hypothesis-testing', 'hypothesis-testing']
['methodology', 'miscellaneous']
[ 7.08671272e-01 5.99764520e-03 -5.62788367e-01 -7.09625483e-01 -1.45187676e+00 -7.58093059e-01 2.77667612e-01 7.10715428e-02 -3.88308942e-01 1.00341487e+00 -6.40341759e-01 -5.16542435e-01 -6.03699803e-01 -9.39058244e-01 -6.05240703e-01 -9.20193791e-01 -7.79310167e-02 9.51894462e-01 4.91034925e-01 3.71752620e-01 3.98341924e-01 3.83244306e-01 -1.89336324e+00 -2.27118045e-01 9.46702838e-01 1.22335124e+00 -1.05963685e-01 9.98892903e-01 1.43697754e-01 2.99806654e-01 -8.21739674e-01 -1.33116186e-01 4.23190892e-01 -9.20628190e-01 -7.31561422e-01 1.58761278e-01 3.20827395e-01 -3.31129164e-01 4.11858857e-01 1.28772855e+00 6.02260590e-01 3.82667147e-02 7.65398204e-01 -1.52449143e+00 -1.72616839e-01 3.22793454e-01 -6.00353241e-01 2.60144383e-01 4.52338636e-01 2.16387138e-02 1.04103804e+00 -8.19683075e-01 2.92061359e-01 1.00592458e+00 2.72817284e-01 2.15828300e-01 -1.50880504e+00 -8.41398478e-01 1.01966029e-02 -1.12926900e-01 -1.66894901e+00 -4.40914392e-01 4.51552033e-01 -2.60553896e-01 3.62087458e-01 8.46332312e-01 5.45271456e-01 5.02533078e-01 1.54329538e-01 6.17692232e-01 1.13681459e+00 -4.48833138e-01 6.09200656e-01 3.80264878e-01 1.28897756e-01 7.46115208e-01 3.91671240e-01 1.95497870e-01 -5.30230582e-01 -8.07297409e-01 6.12822294e-01 -6.53673857e-02 -2.58538295e-02 -3.60493958e-01 -7.56647766e-01 9.18268383e-01 -4.09409106e-01 3.11554000e-02 -1.92725837e-01 2.26799697e-02 -1.73727259e-01 6.55934155e-01 3.25787216e-01 2.96028763e-01 -4.89104837e-01 -2.37745661e-02 -9.91709352e-01 2.00609177e-01 8.78279150e-01 8.74735415e-01 1.00570607e+00 -4.27441210e-01 -4.73717213e-01 6.30296886e-01 2.96955109e-01 1.12141430e+00 2.23409221e-01 -8.48291159e-01 2.51588702e-01 3.47414106e-01 3.08591187e-01 -4.61737812e-01 -1.95851382e-02 -4.34921890e-01 -3.59229714e-01 -5.56266606e-02 3.91095877e-01 -1.92158148e-01 -5.68473339e-01 1.83484685e+00 4.36376721e-01 3.32413465e-02 -3.50354642e-01 3.56388092e-01 2.35931948e-01 2.33427301e-01 -3.66047084e-01 -7.64149725e-01 1.12802982e+00 -1.33808360e-01 -4.09481049e-01 5.39475568e-02 6.89855456e-01 -5.88972390e-01 1.09944701e+00 4.33582246e-01 -1.18251717e+00 -4.38373923e-01 -1.02548647e+00 5.74769497e-01 -9.65523943e-02 -4.57373783e-02 3.11345190e-01 9.30182636e-01 -1.15124559e+00 1.53791249e-01 -5.24081886e-01 -2.29482293e-01 1.19558632e-01 8.26485097e-01 -8.55453089e-02 -6.42353669e-02 -9.14364517e-01 1.18976220e-01 1.45021081e-01 -1.56626984e-01 -7.14581668e-01 -3.25534821e-01 -5.58876991e-01 1.60344660e-01 6.40506804e-01 -3.18243235e-01 1.19779241e+00 -7.24363089e-01 -1.29388988e+00 7.12979734e-01 -5.48892438e-01 1.12203740e-01 4.86725062e-01 3.27131122e-01 -2.79694885e-01 1.36847004e-01 6.71442151e-01 4.79820549e-01 5.73952377e-01 -1.08156502e+00 -1.15553987e+00 -4.54420745e-01 -3.34284872e-01 2.88361334e-03 -1.50485501e-01 3.07461852e-03 -6.56836033e-01 -1.14861913e-01 4.00613338e-01 -7.35438287e-01 -1.23090789e-01 -1.66960046e-01 -7.43065596e-01 -4.08131748e-01 5.92851520e-01 3.62973474e-02 1.28912687e+00 -2.32278442e+00 -5.49545586e-01 1.04228175e+00 -6.03702627e-02 -3.87280703e-01 -1.59077033e-01 3.09438437e-01 5.22995405e-02 3.35394770e-01 -3.14424694e-01 6.72509372e-02 7.87894204e-02 8.11479054e-03 -3.40855837e-01 7.72585332e-01 7.84917399e-02 5.16498566e-01 -6.69289589e-01 -8.09448481e-01 -2.78993905e-01 -3.05683464e-01 -6.52577877e-01 4.26995248e-01 -4.79722284e-02 2.99214870e-01 -6.88955724e-01 9.61044371e-01 7.29549050e-01 -4.89636481e-01 2.29492381e-01 2.83405393e-01 -3.79953161e-02 -2.77569760e-02 -1.46511805e+00 6.06535196e-01 3.83654654e-01 5.58511950e-02 -5.48312478e-02 -1.08977783e+00 1.14676440e+00 6.48989081e-02 6.24109507e-01 -7.28587925e-01 3.79908457e-02 4.59101915e-01 -1.30970955e-01 -5.05179763e-01 1.40957758e-01 -1.14400156e-01 -3.31984997e-01 5.04160643e-01 -2.58146316e-01 -2.41625324e-01 1.41559973e-01 -1.87087432e-01 1.19562423e+00 -3.73551577e-01 4.42303002e-01 -4.35820371e-01 3.16394120e-01 -2.60412455e-01 6.81973696e-01 1.57168639e+00 -1.37118325e-01 2.94180542e-01 9.11598623e-01 3.30305010e-01 -3.38225991e-01 -1.43667531e+00 -4.25753444e-01 1.16081786e+00 1.68132901e-01 2.76529789e-01 -3.38166744e-01 -8.94141614e-01 1.77638695e-01 6.13657475e-01 -7.74275899e-01 -1.09028280e-01 -1.62438259e-01 -5.86622894e-01 3.75802159e-01 2.11299449e-01 4.46271211e-01 -5.88188529e-01 -5.49589097e-01 -1.18904665e-01 4.07440923e-02 -3.65506560e-01 -4.82009709e-01 5.12583554e-01 -7.10766077e-01 -1.29590237e+00 -2.74294823e-01 -7.84870923e-01 9.29476440e-01 -1.33157494e-02 7.51958311e-01 -6.05719797e-02 -4.04716879e-02 5.86492002e-01 -2.19877169e-01 -3.40655059e-01 -4.30142805e-02 -1.05699658e-01 -9.83598009e-02 1.30919784e-01 3.71227920e-01 -3.78229022e-01 -5.55791616e-01 7.19059169e-01 -1.05108595e+00 -5.79328954e-01 7.22053945e-01 9.16490376e-01 1.05724406e+00 2.06878021e-01 8.20930660e-01 -1.11578345e+00 4.72500026e-01 -6.29418731e-01 -9.44059193e-01 6.00158453e-01 -8.01543474e-01 -8.27943310e-02 3.41265231e-01 -5.41558444e-01 -6.07222021e-01 -1.85112551e-01 6.45332783e-02 1.71807501e-02 -5.51413782e-02 5.41753352e-01 -4.14116621e-01 2.30020080e-02 4.25602108e-01 3.09071034e-01 -1.93506368e-02 -2.49285311e-01 -1.63101748e-01 7.20377982e-01 4.47080046e-01 -7.12268353e-01 6.98693752e-01 3.71425211e-01 2.56854832e-01 -6.80566490e-01 -6.66676462e-01 -5.43143153e-01 -3.28249395e-01 -9.96192023e-02 4.86678243e-01 -3.47692847e-01 -9.82693553e-01 4.84180748e-02 -4.75717217e-01 -3.42732280e-01 -7.26696134e-01 7.32995749e-01 -6.00798726e-01 3.25581849e-01 1.67095363e-02 -1.19347525e+00 2.28647888e-01 -1.22257912e+00 1.15531373e+00 2.49109238e-01 -7.95691311e-02 -7.57245541e-01 3.68466824e-02 -5.45580611e-02 8.68857279e-02 2.99905357e-03 1.13239181e+00 -1.33915126e+00 -6.68233275e-01 -5.72304964e-01 -1.09210998e-01 -3.38900834e-02 1.64225146e-01 8.43080785e-03 -7.58591950e-01 -7.76007712e-01 8.28632116e-02 -3.84891123e-01 4.57171589e-01 6.04562819e-01 1.38297451e+00 -2.17888340e-01 -5.34325063e-01 2.67797440e-01 1.28078568e+00 5.90307832e-01 2.91513950e-01 -2.73654491e-01 -1.71420977e-01 4.48690146e-01 1.03315175e+00 8.22344840e-01 -5.20408386e-04 4.30709273e-01 1.79435149e-01 8.58177990e-03 5.58555603e-01 -1.24293908e-01 3.62295985e-01 5.26374876e-01 6.83462322e-01 -5.37419081e-01 -6.60769880e-01 4.11338717e-01 -1.49074566e+00 -8.05369139e-01 2.59078026e-01 2.83877850e+00 1.02844155e+00 2.71875978e-01 2.12142646e-01 9.38691646e-02 7.15767860e-01 -4.53784436e-01 -9.72354352e-01 1.56122651e-02 -9.71102417e-02 5.81012130e-01 5.82494318e-01 6.78867042e-01 -7.37120986e-01 3.37981522e-01 7.04380941e+00 9.48846579e-01 -9.10261691e-01 -2.11877197e-01 8.28417838e-01 6.83747651e-03 -7.47231066e-01 3.48067433e-01 -1.05451071e+00 3.27167779e-01 9.01720345e-01 -1.87797293e-01 2.09818054e-02 7.36361742e-01 1.44614354e-01 -7.47478306e-01 -1.51530612e+00 6.51968837e-01 7.44467229e-02 -7.24435151e-01 -1.25195831e-01 4.46364641e-01 4.55548882e-01 -4.13641065e-01 3.66471976e-01 3.83985579e-01 4.61198866e-01 -7.29307353e-01 5.57294369e-01 7.42850721e-01 1.30728495e+00 -7.01887965e-01 8.38033557e-01 4.60343301e-01 -1.06684780e+00 -1.18958689e-01 -2.11494714e-01 2.48700693e-01 -2.27418229e-01 8.04863453e-01 -1.07837462e+00 1.90153182e-01 5.71841598e-01 -1.87333286e-01 -7.53225029e-01 1.19194424e+00 2.17571706e-01 9.30886626e-01 -3.92507762e-01 -4.51677650e-01 -3.14688623e-01 -2.26406112e-01 3.91260445e-01 8.35928798e-01 5.25622785e-01 9.09694191e-03 6.75576746e-01 6.83652580e-01 2.34061927e-01 2.06110284e-01 -5.72306156e-01 8.64472240e-02 9.80749786e-01 6.58979237e-01 -9.59371388e-01 -4.28496271e-01 -2.93880314e-01 5.29458165e-01 3.29448730e-02 5.79548180e-01 -6.25097811e-01 -6.61236167e-01 1.70493454e-01 9.19489041e-02 3.63359451e-01 2.70799220e-01 -1.40407383e-01 -7.85901189e-01 -1.43061336e-02 -6.32815480e-01 7.00255334e-01 -3.46346080e-01 -1.40296137e+00 -1.76686689e-01 5.06930947e-01 -1.08251178e+00 -7.04696476e-01 -3.89664292e-01 -4.52725291e-01 8.36229920e-01 -9.99954581e-01 -4.79882568e-01 1.88207060e-01 7.73887157e-01 1.09597161e-01 -3.11927237e-02 6.86356664e-01 1.09611861e-01 -4.07291085e-01 1.18114519e+00 1.29174650e-01 3.73896658e-02 7.24780202e-01 -1.19125116e+00 -3.65213275e-01 7.14106023e-01 -3.18640918e-01 6.39287055e-01 5.63728392e-01 -8.95744801e-01 -1.30834651e+00 -7.08025694e-01 8.32372546e-01 -8.17866102e-02 4.02722687e-01 -3.99947673e-01 -6.29922688e-01 4.94647205e-01 -3.72264445e-01 4.24789302e-02 1.23872769e+00 1.43735141e-01 7.00840428e-02 -5.12486815e-01 -1.50944889e+00 3.62632006e-01 6.07363820e-01 -5.44291258e-01 -1.37553304e-01 3.12683284e-01 5.97262800e-01 2.18447283e-01 -7.98306227e-01 7.24352837e-01 6.45960927e-01 -1.06072843e+00 4.28128690e-01 -4.14124370e-01 -4.91913110e-01 -3.49431276e-01 -5.74560046e-01 -5.46142578e-01 -1.13808334e-01 -6.62722945e-01 3.60631198e-01 1.15296996e+00 7.01308727e-01 -1.17203474e+00 8.31003785e-01 5.33474684e-01 3.42077106e-01 -6.68673873e-01 -1.25198531e+00 -7.72082746e-01 -2.36217946e-01 -4.53896821e-01 5.43112636e-01 6.44916296e-01 -8.49945694e-02 1.02320537e-01 8.58384296e-02 2.86052287e-01 5.86337388e-01 3.33435029e-01 7.81669378e-01 -1.38922644e+00 -7.56467760e-01 -2.93460220e-01 -2.69360095e-01 -1.24310935e+00 -1.67655140e-01 -5.32655001e-01 3.70130807e-01 -8.92538905e-01 6.39717162e-01 -5.69498718e-01 -4.11805063e-01 4.78885919e-01 -2.07585886e-01 1.15750311e-02 -4.58456278e-01 1.19346619e-01 -7.03153372e-01 2.16145739e-01 9.94659066e-01 1.38454318e-01 -2.95077533e-01 6.46579146e-01 -6.68421328e-01 1.61657125e-01 4.91325200e-01 -5.15527248e-01 -6.64050162e-01 3.43191773e-01 3.17858934e-01 6.79960370e-01 2.48469114e-01 -6.91886723e-01 2.90627062e-01 -5.48076451e-01 4.82042760e-01 -9.79999661e-01 1.09439634e-03 -5.47407329e-01 4.79908520e-03 5.51135421e-01 -5.57870388e-01 1.60284325e-01 -1.58608526e-01 6.32426679e-01 -6.30561113e-02 -4.25239891e-01 7.94387460e-01 3.09661478e-01 3.97413336e-02 4.22298253e-01 -4.62987423e-01 1.80494949e-01 1.00993633e+00 -3.72230291e-01 -2.19483629e-01 -5.29707968e-01 -5.94765604e-01 3.81321818e-01 2.72837132e-01 -2.95470655e-01 6.57281220e-01 -1.15340877e+00 -5.63695908e-01 7.37153769e-01 1.73939019e-01 -3.26453485e-02 -1.11418694e-01 1.01516402e+00 5.86451851e-02 2.67676026e-01 4.00055826e-01 -8.70173275e-01 -1.18611228e+00 1.89083323e-01 1.42645240e-01 -1.69755831e-01 2.69069672e-01 9.98491526e-01 2.69210100e-01 -4.50926691e-01 2.17472032e-01 -1.57053486e-01 2.01926008e-01 -3.03238612e-02 3.58533829e-01 3.89709711e-01 -8.35482702e-02 -6.06074482e-02 -2.69522578e-01 3.28613758e-01 9.42170154e-03 -7.53464580e-01 8.59900355e-01 -1.94870770e-01 -2.86230326e-01 8.33213985e-01 1.23724008e+00 4.84589905e-01 -1.02497137e+00 -2.75482446e-01 2.16062982e-02 -7.20002651e-01 -5.66335618e-01 -7.08777845e-01 -5.47427416e-01 3.19941431e-01 8.44708622e-01 6.77159846e-01 1.26620507e+00 3.01960319e-01 -8.52747485e-02 5.73149025e-01 3.51709425e-01 -1.00083089e+00 1.95589885e-01 2.68393278e-01 3.52961004e-01 -1.02532887e+00 3.97889391e-02 -3.89340520e-01 -2.35564545e-01 7.76559889e-01 7.31858671e-01 8.67498815e-02 8.33207309e-01 3.77832949e-01 -2.35625565e-01 -2.13833317e-01 -7.11447001e-01 -2.00665280e-01 3.43421340e-01 4.53955173e-01 1.56243905e-01 -3.12771164e-02 -3.58859748e-01 4.86586154e-01 -2.37874299e-01 -1.28738269e-01 2.63482928e-02 1.02548993e+00 -7.53958225e-01 -1.13235259e+00 -4.54631984e-01 1.05445981e+00 -2.60558844e-01 2.21820101e-01 -5.61373591e-01 9.66488183e-01 1.82068944e-01 1.21138322e+00 4.26869124e-01 -3.86552751e-01 1.03831731e-01 1.27382278e-01 2.87822932e-01 -6.45775795e-01 4.66865078e-02 5.01594543e-01 -1.84876814e-01 -4.45144325e-01 -2.13478342e-01 -1.00800598e+00 -1.08572400e+00 -4.32822853e-02 -7.45674193e-01 6.62415564e-01 1.62416622e-01 1.03949881e+00 2.53401939e-02 3.77634726e-02 1.24186695e+00 -5.27918972e-02 -1.01915991e+00 -8.07532132e-01 -1.01637113e+00 -1.70793235e-01 2.14982033e-01 -5.86569011e-01 -5.74519753e-01 -3.42065692e-01]
[7.659017562866211, 4.383215427398682]
62671cbd-0880-4048-9ff3-adb3894241b2
crosel-cross-selection-of-confident-pseudo
2303.10365
null
https://arxiv.org/abs/2303.10365v2
https://arxiv.org/pdf/2303.10365v2.pdf
CroSel: Cross Selection of Confident Pseudo Labels for Partial-Label Learning
Partial-label learning (PLL) is an important weakly supervised learning problem, which allows each training example to have a candidate label set instead of a single ground-truth label. Identification-based methods have been widely explored to tackle label ambiguity issues in PLL, which regard the true label as a latent variable to be identified. However, identifying the true labels accurately and completely remains challenging, causing noise in pseudo labels during model training. In this paper, we propose a new method called CroSel, which leverages historical prediction information from models to identify true labels for most training examples. First, we introduce a cross selection strategy, which enables two deep models to select true labels of partially labeled data for each other. Besides, we propose a novel consistent regularization term called co-mix to avoid sample waste and tiny noise caused by false selection. In this way, CroSel can pick out the true labels of most examples with high precision. Extensive experiments demonstrate the superiority of CroSel, which consistently outperforms previous state-of-the-art methods on benchmark datasets. Additionally, our method achieves over 90\% accuracy and quantity for selecting true labels on CIFAR-type datasets under various settings.
['Lei Feng', 'Yiqun Wang', 'Hongxin Wei', 'Shiyu Tian']
2023-03-18
null
null
null
null
['partial-label-learning']
['methodology']
[ 1.81374043e-01 -2.61710547e-02 -6.68314755e-01 -7.22014844e-01 -1.31232488e+00 -7.20649958e-01 3.86025459e-01 -2.52251215e-02 -3.44536930e-01 9.81568694e-01 -4.02759850e-01 7.09213614e-02 -8.46960172e-02 -4.04494524e-01 -5.88307738e-01 -1.02158856e+00 3.27719808e-01 6.44465804e-01 -3.84583138e-02 5.86147130e-01 -4.49937023e-02 -1.29919842e-01 -1.34938884e+00 3.29786599e-01 8.49480212e-01 1.20962918e+00 6.30445257e-02 -1.06226675e-01 -2.17970237e-01 9.72214401e-01 -5.37065148e-01 -4.34928209e-01 2.48463929e-01 -3.42498720e-01 -7.72324622e-01 1.66198134e-01 7.21463382e-01 -1.21541798e-01 6.58189207e-02 1.19229496e+00 4.27937150e-01 -1.62977085e-01 6.60162687e-01 -1.46134293e+00 -4.00020242e-01 9.19507980e-01 -7.67709911e-01 -2.99068481e-01 -2.32189029e-01 8.31796005e-02 1.28155756e+00 -1.09548461e+00 3.38186085e-01 1.23253989e+00 9.28685308e-01 6.57000303e-01 -1.42678893e+00 -1.22161448e+00 4.05679405e-01 -6.25682622e-02 -1.60995281e+00 -4.37664390e-01 8.03012609e-01 -4.41573381e-01 7.19151795e-02 2.22752675e-01 -1.25966333e-02 1.11774683e+00 -3.14863086e-01 1.04458570e+00 1.38637674e+00 -4.53377634e-01 2.60859430e-01 4.78027999e-01 5.42258382e-01 6.69936121e-01 1.27248719e-01 1.34361252e-01 -6.46048725e-01 -4.21457916e-01 4.04642493e-01 3.82008255e-02 -3.48608822e-01 -2.28459895e-01 -1.37814176e+00 6.98496759e-01 3.47308189e-01 3.02183032e-02 -1.13359421e-01 1.48093849e-01 3.63342166e-01 3.47946621e-02 7.39179075e-01 3.36137921e-01 -8.13502014e-01 3.88613105e-01 -8.51008952e-01 1.05340205e-01 4.33046997e-01 9.92277324e-01 7.96319366e-01 -2.15693176e-01 -5.61978757e-01 1.21031737e+00 4.58303571e-01 4.42064494e-01 4.25257653e-01 -9.01723504e-01 5.05745888e-01 6.11280978e-01 2.75123596e-01 -7.03984857e-01 -3.69040757e-01 -1.03865552e+00 -8.31115723e-01 -1.52456865e-01 4.59191710e-01 -6.81922361e-02 -8.52875292e-01 2.12792373e+00 4.61337239e-01 6.70946419e-01 -5.66122308e-02 8.68382990e-01 6.87693834e-01 5.34896493e-01 3.38002682e-01 -4.96287823e-01 1.16356528e+00 -1.07573378e+00 -6.71880543e-01 -4.29522544e-01 7.38799274e-01 -5.40071785e-01 1.06489897e+00 4.43703681e-01 -4.29277629e-01 -5.01015604e-01 -6.99546754e-01 3.08943957e-01 1.19590042e-02 7.48668075e-01 7.47534633e-01 3.77793968e-01 -5.32135010e-01 4.77725446e-01 -4.86369222e-01 2.30436236e-01 6.49591804e-01 3.21000129e-01 -1.27059117e-01 -1.37801662e-01 -1.23228359e+00 2.48057693e-01 6.64497077e-01 1.64121747e-01 -9.54502523e-01 -7.83509970e-01 -6.06093705e-01 -3.95877101e-02 7.39373744e-01 -1.89787790e-01 1.39964235e+00 -1.08961964e+00 -1.04614162e+00 1.07045245e+00 -3.78792048e-01 -3.45362633e-01 5.44841409e-01 -1.43816680e-01 -4.11177993e-01 -2.66751945e-01 5.83455563e-01 7.75360346e-01 8.04459691e-01 -1.77850795e+00 -9.28821862e-01 -1.51258811e-01 -2.64628559e-01 2.91468110e-02 -2.60564744e-01 -4.59668517e-01 -3.46781701e-01 -7.55538583e-01 5.22997022e-01 -9.86380219e-01 -1.54611340e-03 -1.73837110e-01 -8.04206133e-01 -6.55159056e-01 5.73811650e-01 -1.60236299e-01 1.27837229e+00 -2.16346502e+00 -4.14589316e-01 8.02030638e-02 5.00390291e-01 2.51744032e-01 -1.65526010e-02 1.58776958e-02 -1.02052912e-01 2.89956778e-01 -2.46351659e-01 -6.86324239e-01 1.14405334e-01 2.05544025e-01 -7.33203650e-01 3.70960534e-01 2.57768154e-01 7.99607754e-01 -1.15764678e+00 -5.23592591e-01 -1.81546986e-01 1.81649595e-01 3.46371867e-02 1.47733927e-01 -3.63926470e-01 6.32398188e-01 -6.03825331e-01 7.79382408e-01 7.75682867e-01 -6.98429167e-01 2.28961587e-01 -2.01106295e-01 3.09446454e-01 2.49653324e-01 -1.23432803e+00 1.29542887e+00 -3.92703325e-01 5.41424192e-02 -2.12468266e-01 -9.10801589e-01 1.01776731e+00 3.39983135e-01 3.87973428e-01 -4.07208025e-01 1.46619380e-01 5.89457154e-01 -4.63448972e-01 -2.20889181e-01 -4.78424989e-02 -2.73398340e-01 -1.32063627e-01 5.97646654e-01 1.07547957e-02 4.46439624e-01 5.39041013e-02 6.35994598e-02 6.30533397e-01 1.87510058e-01 5.04875258e-02 -2.52284199e-01 4.80899721e-01 -9.58452523e-02 1.33105850e+00 8.96292448e-01 -3.57326537e-01 5.52015305e-01 3.63854617e-01 -5.77581048e-01 -6.48353815e-01 -8.57525587e-01 -4.61288571e-01 1.04510367e+00 3.98546129e-01 -1.23467587e-01 -5.18847883e-01 -1.20762253e+00 6.50146008e-02 8.00402701e-01 -4.41777647e-01 -2.27264747e-01 -3.66614580e-01 -9.91872311e-01 3.75170141e-01 2.80921668e-01 6.06615186e-01 -9.09506798e-01 1.26032159e-02 3.19165856e-01 -4.28323835e-01 -1.11389291e+00 -6.54338658e-01 5.45001090e-01 -7.01872289e-01 -1.19683468e+00 -4.20143157e-01 -9.90852058e-01 8.29109132e-01 2.20229596e-01 1.35448575e+00 5.33849075e-02 2.15254068e-01 -1.99258313e-01 -2.48084500e-01 -1.86654836e-01 -3.20665240e-01 1.89805418e-01 2.91736573e-01 4.20024961e-01 5.97905576e-01 -2.39591554e-01 -3.52674454e-01 6.60029054e-01 -5.85154891e-01 1.70649558e-01 4.99002576e-01 1.28678524e+00 1.20234048e+00 3.86908144e-01 1.22231436e+00 -1.52351654e+00 3.39846969e-01 -6.91825271e-01 -6.43408537e-01 6.14096045e-01 -9.36312675e-01 1.73915356e-01 7.42372155e-01 -7.14167118e-01 -9.79444802e-01 3.76801938e-01 1.91043124e-01 -5.54888785e-01 -1.25002205e-01 2.74451911e-01 -1.99102432e-01 1.90528169e-01 5.02099097e-01 1.29995704e-01 -4.36937988e-01 -7.07814991e-01 1.26555815e-01 7.20412493e-01 5.43836892e-01 -7.58920372e-01 6.00763142e-01 2.91091412e-01 -1.08719677e-01 1.16807170e-01 -1.90100908e+00 -5.07176876e-01 -5.99460483e-01 -9.97825339e-03 3.57036769e-01 -1.21419775e+00 -5.45818090e-01 7.41582215e-01 -9.13184047e-01 -3.34011436e-01 -2.14404970e-01 4.03931230e-01 -1.09543309e-01 2.72052903e-02 -5.61163962e-01 -8.96004558e-01 -2.93597758e-01 -1.16117966e+00 1.33379734e+00 2.90070593e-01 -2.11982317e-02 -9.75929379e-01 -1.56355262e-01 4.18281108e-01 -8.06206912e-02 1.06963307e-01 6.57935560e-01 -9.47402358e-01 -5.87128699e-01 -2.44564325e-01 -3.40759397e-01 4.15069222e-01 1.96886107e-01 -3.76288414e-01 -1.30861020e+00 -4.29879397e-01 6.25486895e-02 -8.94977450e-01 1.12128210e+00 2.85541028e-01 1.31484628e+00 -2.90656984e-01 -6.73404038e-01 4.81411308e-01 1.31702149e+00 9.01792049e-02 -1.55319661e-01 1.02752827e-01 8.27936351e-01 5.66960812e-01 9.21861529e-01 3.45531642e-01 3.20674002e-01 6.30036294e-01 2.59494364e-01 6.73561031e-03 4.79926774e-03 -4.17086333e-01 6.50702417e-03 6.90527439e-01 7.52308190e-01 -2.77465224e-01 -8.38396490e-01 4.31997061e-01 -1.96865582e+00 -6.44691288e-01 -1.96790278e-01 2.36733985e+00 1.28521168e+00 2.51124471e-01 -2.51693428e-01 9.89561677e-02 1.05298007e+00 -2.30905041e-02 -9.14380729e-01 5.83876014e-01 -1.92108274e-01 -1.38097331e-01 5.82914233e-01 4.40659344e-01 -1.59757471e+00 9.14251387e-01 5.65837955e+00 1.39524055e+00 -1.05262268e+00 3.42965037e-01 1.32240415e+00 1.56483892e-02 -3.15372765e-01 -2.71513015e-02 -1.26844239e+00 7.95040488e-01 4.69926655e-01 2.22323999e-01 5.58139980e-02 1.06434619e+00 6.87946528e-02 -5.56879118e-02 -1.28135335e+00 1.18786490e+00 -1.15927979e-01 -1.08018863e+00 -1.91567928e-01 -1.26347154e-01 1.08295500e+00 -1.71875387e-01 2.15007350e-01 5.28525233e-01 4.67215329e-01 -8.96730721e-01 7.53277361e-01 4.43199545e-01 1.11105323e+00 -6.92184210e-01 8.31085086e-01 7.21962452e-01 -1.12242246e+00 -2.35586792e-01 -4.06793237e-01 1.86622754e-01 1.05758514e-02 1.21352327e+00 -8.35688114e-01 3.33083302e-01 4.68142688e-01 8.22221518e-01 -5.31351686e-01 9.47376788e-01 -4.88622785e-01 1.14567089e+00 -3.30771476e-01 2.63571382e-01 2.02547923e-01 4.79831211e-02 6.06813058e-02 8.11746657e-01 1.20474249e-01 -6.87525719e-02 6.96413100e-01 9.70746994e-01 -4.77182090e-01 -5.78853600e-02 -2.50650883e-01 2.42412865e-01 1.01867735e+00 1.32191026e+00 -6.80824459e-01 -3.29785138e-01 -3.54497395e-02 7.15178549e-01 5.35841346e-01 3.19952458e-01 -8.74801040e-01 5.98358102e-02 2.70263940e-01 -2.63943911e-01 -1.78106368e-01 3.32811892e-01 -3.25396031e-01 -1.31989408e+00 1.32134899e-01 -7.08342075e-01 4.40727562e-01 -4.67711836e-01 -1.82309866e+00 5.93716681e-01 -2.24484682e-01 -1.55405819e+00 -2.05512047e-02 -3.28789920e-01 -3.59334141e-01 8.33715498e-01 -1.53836572e+00 -1.11162150e+00 -1.17094956e-01 7.70371184e-02 5.57738721e-01 -1.97588623e-01 5.71649373e-01 4.06175256e-01 -8.30262840e-01 9.02217448e-01 3.34049255e-01 2.08926648e-01 1.05763507e+00 -1.27525139e+00 1.99844003e-01 5.68660617e-01 4.36676979e-01 4.82352495e-01 2.72674173e-01 -6.05086684e-01 -7.42926657e-01 -1.53848577e+00 1.04019821e+00 -5.21311343e-01 3.41588050e-01 -3.42852950e-01 -1.10832560e+00 6.31690681e-01 -3.99597049e-01 4.27736402e-01 6.72501385e-01 6.52674586e-02 -5.69772005e-01 -4.10518229e-01 -1.08284009e+00 2.21702382e-01 8.71031404e-01 -4.02854115e-01 -1.61747694e-01 7.52668977e-01 9.77643847e-01 -4.22369123e-01 -5.13241291e-01 6.14895761e-01 3.00628096e-01 -6.88804090e-01 7.91030049e-01 -2.31217220e-01 8.17471370e-02 -5.42418242e-01 4.47026230e-02 -1.21214974e+00 -3.19927692e-01 -2.51826793e-01 -9.41136405e-02 1.82100105e+00 5.82390010e-01 -7.64731526e-01 8.74649823e-01 5.74576676e-01 8.55058953e-02 -9.99897122e-01 -7.81219840e-01 -9.11726892e-01 -1.44452050e-01 -4.21913475e-01 6.37867630e-01 1.34903705e+00 -4.63094503e-01 5.64860880e-01 -7.16894388e-01 2.03557506e-01 9.74811137e-01 3.39390606e-01 3.81797731e-01 -1.47428346e+00 -3.46539825e-01 -2.73793399e-01 1.36792958e-01 -1.16290593e+00 5.39905965e-01 -1.06565201e+00 3.43248904e-01 -1.11458695e+00 3.55187178e-01 -1.32488132e+00 -6.17270350e-01 9.39871907e-01 -6.08496249e-01 3.46056521e-01 -9.75985974e-02 7.27735102e-01 -8.88934970e-01 6.04960978e-01 1.07200146e+00 -1.57501385e-01 -5.08177131e-02 3.02297652e-01 -7.57899165e-01 6.28746867e-01 8.15574467e-01 -9.77803826e-01 -5.48451483e-01 -3.82873148e-01 1.51288271e-01 -2.00162739e-01 2.87708044e-01 -8.71630549e-01 7.08442628e-02 -2.25216627e-01 2.11126313e-01 -5.27772725e-01 5.18580433e-04 -8.18203211e-01 2.31325030e-01 2.88704216e-01 -8.45648110e-01 -5.98493278e-01 -2.63952404e-01 8.01511765e-01 -2.57607162e-01 -4.02345508e-01 8.39901209e-01 3.19024250e-02 -6.99706495e-01 5.64659953e-01 4.26947325e-01 2.17468590e-01 1.06011772e+00 4.34796810e-01 -3.66389692e-01 3.16702239e-02 -5.41410923e-01 7.15106189e-01 1.53215289e-01 2.90408045e-01 1.79977030e-01 -1.64766204e+00 -7.35794067e-01 2.44233403e-02 3.56642932e-01 4.54230100e-01 1.73132539e-01 6.14202678e-01 1.14919297e-01 3.49147737e-01 5.82489669e-01 -7.89992511e-01 -9.82210219e-01 6.38598204e-01 3.75189543e-01 -4.07647103e-01 -2.55159825e-01 1.18956649e+00 4.57563788e-01 -8.48357320e-01 3.59934717e-01 3.26776542e-02 -1.55565530e-01 1.06981903e-01 5.09385884e-01 1.06391720e-01 -6.51681200e-02 -6.26115680e-01 -3.67740750e-01 4.61689264e-01 -1.66821644e-01 2.56936014e-01 8.83056760e-01 -2.55049706e-01 -1.08960927e-01 8.01033199e-01 1.23333323e+00 -1.22799493e-01 -1.57947910e+00 -8.17560256e-01 2.40563720e-01 -4.21050012e-01 4.41159084e-02 -1.13028419e+00 -1.27978873e+00 7.78613865e-01 5.67293584e-01 9.95899141e-02 9.91839767e-01 4.18771952e-02 8.07359278e-01 2.63359487e-01 5.39108813e-01 -9.84880328e-01 5.14229983e-02 2.62474626e-01 3.60183269e-01 -1.78813863e+00 -2.36821339e-01 -5.84983170e-01 -8.13291550e-01 8.62112939e-01 9.38212931e-01 2.57950544e-01 5.03371537e-01 -4.90288436e-02 3.43684375e-01 1.28068740e-03 -9.10451531e-01 2.45764833e-02 2.86557823e-01 2.22697020e-01 3.09387118e-01 2.92434245e-01 -2.73820102e-01 9.41928685e-01 2.82011271e-01 -3.26315351e-02 8.99057835e-03 5.84413648e-01 -3.41573387e-01 -1.28972805e+00 -2.95043141e-01 6.56671941e-01 -5.61365187e-01 -1.88547567e-01 -1.97005033e-01 3.77764344e-01 5.79185426e-01 9.35270667e-01 -2.71402955e-01 -4.48320627e-01 -4.15125266e-02 3.16672027e-01 -1.10663086e-01 -8.43462110e-01 -3.99961114e-01 1.39270544e-01 -1.32641494e-01 -1.32401943e-01 -4.39744264e-01 -4.37569261e-01 -1.22061896e+00 2.02431172e-01 -7.04785526e-01 3.24150920e-01 1.95126191e-01 1.06327271e+00 3.28865051e-01 3.22604895e-01 9.34952021e-01 -5.04390955e-01 -9.40947950e-01 -8.58107984e-01 -8.17778766e-01 5.56209683e-01 3.02896827e-01 -9.33763266e-01 -5.12081563e-01 8.36929381e-02]
[9.446141242980957, 4.0109968185424805]
b8a68b55-8486-4073-bcb6-e277ff65b1f9
perceptual-loss-for-robust-unsupervised
2104.10011
null
https://arxiv.org/abs/2104.10011v1
https://arxiv.org/pdf/2104.10011v1.pdf
Perceptual Loss for Robust Unsupervised Homography Estimation
Homography estimation is often an indispensable step in many computer vision tasks. The existing approaches, however, are not robust to illumination and/or larger viewpoint changes. In this paper, we propose bidirectional implicit Homography Estimation (biHomE) loss for unsupervised homography estimation. biHomE minimizes the distance in the feature space between the warped image from the source viewpoint and the corresponding image from the target viewpoint. Since we use a fixed pre-trained feature extractor and the only learnable component of our framework is the homography network, we effectively decouple the homography estimation from representation learning. We use an additional photometric distortion step in the synthetic COCO dataset generation to better represent the illumination variation of the real-world scenarios. We show that biHomE achieves state-of-the-art performance on synthetic COCO dataset, which is also comparable or better compared to supervised approaches. Furthermore, the empirical results demonstrate the robustness of our approach to illumination variation compared to existing methods.
['Bahram Zonooz', 'Elahe Arani', 'Daniel Koguciuk']
2021-04-20
null
null
null
null
['homography-estimation']
['computer-vision']
[ 2.36734107e-01 -1.13077894e-01 8.49707276e-02 -1.40109658e-01 -5.85559487e-01 -5.59741676e-01 7.59071290e-01 -6.07300818e-01 -2.53783446e-02 6.08058691e-01 2.60723531e-02 3.40571851e-01 -3.48767964e-03 -5.53921998e-01 -9.95376587e-01 -8.06496024e-01 4.86250609e-01 3.84495348e-01 4.71206615e-04 -7.16894865e-02 2.76818067e-01 4.31634992e-01 -1.34226191e+00 -2.08351254e-01 6.87005937e-01 8.08879197e-01 1.42042503e-01 5.42770326e-01 6.32040560e-01 4.25235718e-01 -3.38131875e-01 -4.05092686e-01 8.44914734e-01 -5.17307460e-01 -4.26039994e-01 7.25241661e-01 8.36800933e-01 -2.82066435e-01 -7.23816752e-01 1.27137220e+00 3.83840322e-01 9.82001051e-02 6.00395441e-01 -1.31336534e+00 -6.14051342e-01 -5.48797660e-02 -6.92080975e-01 -3.41745406e-01 4.00187403e-01 4.05068602e-03 8.78143132e-01 -1.09944713e+00 1.11690485e+00 1.19200778e+00 3.97863626e-01 9.32918862e-02 -1.42808640e+00 -5.28447986e-01 -2.76714444e-01 2.26726219e-01 -1.44996679e+00 -5.54255486e-01 1.13320458e+00 -4.53617841e-01 4.95859832e-01 -1.66727323e-02 5.29385746e-01 1.00065923e+00 2.71781892e-01 5.35813391e-01 1.30380285e+00 -5.88580728e-01 -1.16241015e-02 1.68405250e-01 -3.11221600e-01 7.30062842e-01 2.44575649e-01 3.41981918e-01 -5.70161164e-01 5.92682585e-02 1.12284625e+00 1.10599749e-01 -5.78787267e-01 -1.09952259e+00 -1.32786548e+00 7.56222367e-01 4.22936589e-01 -9.39781591e-02 -1.20672882e-01 9.18063894e-02 -1.02243923e-01 3.86682600e-01 1.43432036e-01 4.73728329e-01 2.69365329e-02 -2.08528694e-02 -7.50337601e-01 -4.82420884e-02 9.10778403e-01 1.44467521e+00 1.03097868e+00 9.34399292e-02 3.53965223e-01 8.65669191e-01 2.04040214e-01 6.15627944e-01 1.72324166e-01 -1.32065439e+00 5.33649266e-01 4.57084119e-01 1.10229932e-01 -1.39510632e+00 4.96780910e-02 -4.86886144e-01 -8.96765053e-01 2.18032598e-01 3.94324958e-01 -1.14806341e-02 -7.08307445e-01 1.59534907e+00 4.21552420e-01 1.85144171e-01 -4.53488603e-02 9.39808786e-01 2.64409930e-01 5.16383111e-01 -8.50295544e-01 -3.09610993e-01 9.05180037e-01 -1.22307253e+00 -7.49136865e-01 -3.62117112e-01 1.74654186e-01 -1.15007246e+00 8.90762627e-01 3.91918451e-01 -9.73654211e-01 -3.89251947e-01 -1.40717351e+00 -4.12828505e-01 -8.66366327e-02 3.96378368e-01 2.03348681e-01 3.75701100e-01 -7.50712514e-01 5.46087921e-01 -7.38993883e-01 -4.00120586e-01 -4.46873717e-02 2.64967144e-01 -6.43021882e-01 -2.55051672e-01 -8.04823756e-01 8.65800619e-01 1.50551960e-01 -7.22969789e-03 -8.42160165e-01 -3.91302854e-01 -9.53584909e-01 -2.21281517e-02 5.73794961e-01 -7.86574543e-01 8.00077856e-01 -9.29630697e-01 -1.78389049e+00 9.17813540e-01 -1.02996819e-01 -3.12076569e-01 8.18376958e-01 -2.57904023e-01 -1.57359168e-02 2.33320504e-01 1.50622139e-02 5.44940710e-01 1.12676358e+00 -1.56612825e+00 -3.22194189e-01 -3.77796352e-01 -1.01907281e-02 4.26736534e-01 -9.13658738e-02 -4.65881884e-01 -8.49191427e-01 -6.31024122e-01 4.40304071e-01 -1.37967145e+00 8.04306492e-02 2.17883870e-01 -5.21242738e-01 3.25849026e-01 9.61270869e-01 -6.78143442e-01 8.31788182e-01 -2.18177462e+00 4.12112623e-01 2.02674747e-01 8.63951594e-02 1.38338357e-02 -3.75081040e-02 5.04542410e-01 -1.16448561e-02 -4.86468315e-01 -2.59972930e-01 -4.73329842e-01 1.08055577e-01 3.47355187e-01 -3.86200875e-01 9.50687885e-01 -2.23566234e-01 6.29432559e-01 -7.98756659e-01 -2.04234615e-01 4.83030021e-01 7.53338039e-01 -4.20553505e-01 3.18213522e-01 2.38873988e-01 6.44031823e-01 -5.29320538e-02 3.61388922e-01 8.16633046e-01 -2.44633421e-01 2.11299941e-01 -4.44861203e-01 -1.68837365e-02 8.40016976e-02 -1.25546157e+00 1.79653203e+00 -5.64374030e-01 8.11961830e-01 -3.63753363e-02 -7.71782517e-01 9.43549395e-01 2.20246553e-01 5.17055988e-01 -5.29288232e-01 2.50286758e-01 2.33519346e-01 4.35084030e-02 -2.02230215e-01 2.50538677e-01 1.44090503e-01 3.88051659e-01 8.92989933e-02 1.63750306e-01 -4.80729192e-01 1.93002716e-01 4.41113301e-02 6.22936070e-01 2.64814585e-01 5.96931994e-01 -2.01452047e-01 6.31805122e-01 -3.51654798e-01 7.18576670e-01 4.40180689e-01 -7.83868209e-02 1.09753633e+00 5.01176536e-01 -3.87609839e-01 -1.40840220e+00 -8.85539711e-01 -1.64041653e-01 2.49640569e-01 4.24869150e-01 -2.91248858e-01 -7.90340781e-01 -4.48518544e-01 -6.24194145e-02 4.29620057e-01 -5.01727104e-01 -1.61610678e-01 -6.11377776e-01 -4.37139839e-01 3.40968817e-02 3.72518182e-01 8.27280045e-01 -5.63176334e-01 -3.33463997e-01 4.70872130e-03 -2.49469101e-01 -1.66137910e+00 -8.50992084e-01 -3.25329930e-01 -7.45985150e-01 -1.09308183e+00 -6.44106805e-01 -6.12692356e-01 1.00618994e+00 5.81379831e-01 8.23638916e-01 -2.58529842e-01 -2.38018766e-01 5.52304447e-01 -1.35073245e-01 1.21330963e-02 -2.45840207e-01 -2.40660116e-01 -2.51556318e-02 4.41301435e-01 -2.50186771e-02 -7.21868515e-01 -7.34101892e-01 7.05872834e-01 -6.82886004e-01 2.13756621e-01 5.21480203e-01 1.15464878e+00 7.69523680e-01 -1.58946976e-01 -6.59583583e-02 -8.16129148e-01 3.27901170e-02 -1.18773796e-01 -1.06082523e+00 2.29277641e-01 -7.94318557e-01 -3.82740423e-02 6.93625093e-01 -4.32502508e-01 -9.68920767e-01 5.39028645e-01 4.13603932e-01 -8.98640275e-01 9.74075273e-02 8.70176926e-02 -5.23541868e-01 -5.09470761e-01 4.19470280e-01 3.57530922e-01 8.06935504e-02 -4.64454412e-01 4.13623303e-01 3.16573739e-01 7.40015328e-01 -3.40055674e-01 1.35433328e+00 9.50228810e-01 3.67206037e-01 -1.00033641e+00 -6.56105995e-01 -5.85990667e-01 -1.06586003e+00 -1.96434874e-02 7.02067196e-01 -8.61467123e-01 -5.78346193e-01 4.33690459e-01 -1.17163563e+00 -6.62765726e-02 -1.20713739e-02 5.61723351e-01 -8.84438038e-01 7.03547776e-01 -2.41442591e-01 -4.93115246e-01 -1.31764233e-01 -1.44272101e+00 1.14801729e+00 4.89470176e-02 2.29702234e-01 -1.04829502e+00 1.50066555e-01 5.64060092e-01 -7.72320703e-02 3.95830393e-01 5.54907620e-01 -6.09687157e-02 -1.01083374e+00 -2.69286990e-01 -2.55686879e-01 5.41384935e-01 2.39251018e-01 -1.01004262e-02 -9.18253660e-01 -5.71498275e-01 3.49910736e-01 -2.73518205e-01 5.70315838e-01 2.36850411e-01 7.82733500e-01 -3.47294152e-01 1.17733665e-01 1.22668314e+00 1.60460269e+00 1.21836290e-01 8.19465280e-01 4.06512201e-01 1.06361067e+00 7.33051956e-01 6.76946342e-01 3.45747501e-01 3.22364241e-01 9.67035294e-01 4.20662612e-01 -9.96745005e-02 -2.36571789e-01 -3.99802387e-01 3.78295928e-01 9.43094492e-01 -8.51615444e-02 -1.90015197e-01 -6.58471167e-01 5.29499531e-01 -1.89551508e+00 -7.49572456e-01 2.55821012e-02 2.60226345e+00 4.94759351e-01 -1.59662634e-01 -2.53673673e-01 6.59517646e-02 6.55995786e-01 3.58139843e-01 -4.86638784e-01 -7.16558844e-02 -2.71425456e-01 -3.08655471e-01 6.17273748e-01 7.38938332e-01 -1.04869103e+00 8.84243667e-01 6.13531208e+00 4.75816816e-01 -1.11128235e+00 -1.28145263e-01 2.38395363e-01 1.18918374e-01 5.22348359e-02 4.26019073e-01 -5.02534211e-01 2.82588303e-01 2.31570125e-01 -3.21422875e-01 8.02524269e-01 8.64643395e-01 -3.05852387e-02 6.23950586e-02 -1.30460608e+00 1.36488044e+00 6.10850751e-01 -1.04958105e+00 2.41175033e-02 3.56876105e-01 1.20916796e+00 -5.96578754e-02 2.04430342e-01 -2.30820969e-01 5.32907732e-02 -7.29591012e-01 4.35895294e-01 3.82859617e-01 7.31058776e-01 -5.72614491e-01 5.43741226e-01 3.12717259e-01 -1.08464110e+00 2.45614827e-01 -5.53485632e-01 2.31811270e-01 2.72590429e-01 4.27532762e-01 -8.19929004e-01 6.76951706e-01 4.19753492e-01 9.21423852e-01 -5.81338048e-01 1.04902840e+00 -4.44863915e-01 2.25752935e-01 -3.62923414e-01 5.95561028e-01 6.25795126e-02 -7.47005522e-01 8.51937950e-01 7.62497783e-01 3.40673864e-01 -9.01991948e-02 2.64033020e-01 7.47644603e-01 -3.16863209e-01 4.10355814e-02 -9.63637054e-01 1.02233529e-01 2.73427486e-01 1.26009572e+00 -4.16190803e-01 -2.84902930e-01 -5.73363543e-01 1.21159375e+00 1.33900762e-01 6.04464054e-01 -6.31576955e-01 -3.82029712e-01 4.41817194e-01 -2.03630030e-02 3.57006490e-01 -3.70090008e-01 -7.29259625e-02 -1.57006431e+00 2.42396891e-01 -1.00193107e+00 7.53323585e-02 -8.44430327e-01 -1.04558384e+00 3.65896136e-01 6.87800497e-02 -1.53693819e+00 -3.74762326e-01 -7.10532904e-01 -5.99500239e-01 6.99486852e-01 -1.41514158e+00 -1.29615152e+00 -5.56706905e-01 5.62174022e-01 6.46309793e-01 -7.51519278e-02 3.91185939e-01 2.32911780e-01 -3.28194320e-01 3.79462034e-01 4.83025432e-01 2.08621278e-01 1.02591562e+00 -1.27742493e+00 2.67439783e-01 1.14694881e+00 2.62424469e-01 7.65792370e-01 7.29471743e-01 -4.73771244e-01 -1.68453360e+00 -8.39023650e-01 6.25762880e-01 -4.63841051e-01 6.43929362e-01 -5.54867089e-01 -7.27359354e-01 9.89655018e-01 4.26264286e-01 1.78452507e-01 2.89778769e-01 -4.48117584e-01 -5.19511998e-01 -2.15187997e-01 -8.55369329e-01 6.62539244e-01 1.04165316e+00 -6.84723198e-01 -3.36285830e-01 1.94215402e-01 4.09393460e-01 -6.39913619e-01 -8.58639121e-01 3.63654941e-01 8.20524216e-01 -1.08782554e+00 1.04414594e+00 -1.76762700e-01 4.74340498e-01 -3.82860214e-01 -3.67921352e-01 -1.30504203e+00 -2.16013744e-01 -9.79638755e-01 -1.66028962e-01 1.07565999e+00 1.04918838e-01 -7.61614084e-01 6.42442822e-01 3.71309996e-01 1.26575589e-01 -4.92510855e-01 -7.19386041e-01 -1.00298500e+00 -1.69206530e-01 1.19077384e-01 3.44935745e-01 1.11813486e+00 -3.35386187e-01 4.10136521e-01 -9.13026512e-01 3.45792264e-01 1.22798479e+00 3.47123742e-01 1.32436943e+00 -8.73851418e-01 -4.80239302e-01 -7.69559108e-03 -7.05151260e-01 -1.25290322e+00 1.10505819e-01 -6.65439248e-01 4.29443717e-02 -1.14772928e+00 2.05998167e-01 9.87601057e-02 3.35905924e-02 -5.21729849e-02 1.29902437e-01 3.14461976e-01 3.82621646e-01 4.97763425e-01 -1.51251778e-01 8.77873361e-01 1.65380347e+00 -1.28044993e-01 -9.64087471e-02 -1.30870491e-01 -2.82641530e-01 8.62046838e-01 4.91416126e-01 -4.04464871e-01 -7.15801835e-01 -6.72583997e-01 2.05501989e-02 6.42325953e-02 2.66468465e-01 -8.77031744e-01 3.95489156e-01 -2.09526017e-01 3.10027897e-01 -5.80875695e-01 5.91033220e-01 -9.54296529e-01 3.32240164e-01 1.90711439e-01 -2.09328532e-02 3.91763486e-02 -3.01072180e-01 6.13141596e-01 -3.58512193e-01 -8.52538720e-02 9.28174555e-01 3.47365476e-02 -2.84079343e-01 4.50927436e-01 2.81581402e-01 2.50620209e-02 8.40958238e-01 -3.45311642e-01 -4.62170839e-01 -5.99802017e-01 -4.78608787e-01 3.05044651e-03 1.03610528e+00 3.48105490e-01 6.63866580e-01 -1.43430650e+00 -4.11643505e-01 3.89643341e-01 2.28561744e-01 -2.32376787e-03 -7.57334800e-03 1.08040369e+00 -8.25704336e-01 2.98257083e-01 -2.86030918e-01 -6.15323663e-01 -1.22208500e+00 6.14625573e-01 3.26930821e-01 -2.14394592e-02 -8.04227650e-01 2.12364748e-01 7.63128698e-01 -4.60040540e-01 2.01609448e-01 5.58953136e-02 4.27163988e-02 -3.43588710e-01 2.01271862e-01 6.23016834e-01 -2.60912955e-01 -1.13407350e+00 -8.70454833e-02 1.06441879e+00 2.01867651e-02 -3.73372465e-01 1.16648436e+00 -3.55250657e-01 -1.03273831e-01 2.98701137e-01 1.68275928e+00 7.76687339e-02 -1.49950337e+00 -3.10860187e-01 -3.32577109e-01 -1.01345003e+00 -4.44160812e-02 -3.54483098e-01 -1.20546830e+00 1.06507075e+00 6.07677341e-01 -2.31814966e-01 9.58444595e-01 -3.60337764e-01 7.03393519e-01 5.36034346e-01 3.14140201e-01 -1.11508298e+00 2.32676074e-01 3.60044152e-01 1.22076035e+00 -1.39167535e+00 3.78639698e-01 -7.81888783e-01 -6.32180572e-01 1.04781985e+00 5.75556755e-01 -5.45159459e-01 4.86375451e-01 -1.21899232e-01 2.34254107e-01 -9.79494452e-02 -5.42156994e-01 8.40623081e-02 5.75612068e-01 5.80361664e-01 1.19959511e-01 -2.34809473e-01 -1.34584710e-01 -3.29553753e-01 -3.00681621e-01 -3.81118089e-01 6.89626038e-01 7.40325451e-01 -1.30395353e-01 -1.26773846e+00 -5.28552055e-01 -2.34963953e-01 -5.18606566e-02 2.30644107e-01 -7.42230475e-01 9.74148810e-01 -2.50277340e-01 7.31802881e-01 -2.42811650e-01 -2.12759852e-01 3.17415744e-01 -2.08714694e-01 7.70339787e-01 -4.24940169e-01 5.12108766e-02 5.52634835e-01 -1.93737373e-01 -7.96852648e-01 -4.67962056e-01 -6.05663478e-01 -8.11205447e-01 -1.97166592e-01 -3.96596432e-01 -2.13912845e-01 6.29815280e-01 7.64706910e-01 2.63504505e-01 -2.64478493e-02 1.06368363e+00 -1.00399137e+00 -5.59213042e-01 -6.31423771e-01 -5.71417809e-01 7.86792934e-01 4.30893719e-01 -8.40634882e-01 -6.47166848e-01 3.73647243e-01]
[8.68472671508789, -2.333843946456909]
aa3024af-e302-48a5-9079-c06cf690935c
clothing-change-feature-augmentation-for
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Han_Clothing-Change_Feature_Augmentation_for_Person_Re-Identification_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Han_Clothing-Change_Feature_Augmentation_for_Person_Re-Identification_CVPR_2023_paper.pdf
Clothing-Change Feature Augmentation for Person Re-Identification
Clothing-change person re-identification (CC Re-ID) aims to match the same person who changes clothes across cameras. Current methods are usually limited by the insufficient number and variation of clothing in training data, e.g. each person only has 2 outfits in the PRCC dataset. In this work, we propose a novel Clothing-Change Feature Augmentation (CCFA) model for CC Re-ID to largely expand clothing-change data in the feature space rather than visual image space. It automatically models the feature distribution expansion that reflects a person's clothing colour and texture variations to augment model training. Specifically, to formulate meaningful clothing variations in the feature space, our method first estimates a clothing-change normal distribution with intra-ID cross-clothing variances. Then an augmentation generator learns to follow the estimated distribution to augment plausible clothing-change features. The augmented features are guaranteed to maximise the change of clothing and minimise the change of identity properties by adversarial learning to assure the effectiveness. Such augmentation is performed iteratively with an ID-correlated augmentation strategy to increase intra-ID clothing variations and reduce inter-ID clothing variations, enforcing the Re-ID model to learn clothing-independent features inherently. Extensive experiments demonstrate the effectiveness of our method with state-of-the-art results on CC Re-ID datasets.
['Tieniu Tan', 'Liang Wang', 'Yan Huang', 'Shaogang Gong', 'Ke Han']
2023-01-01
null
null
null
cvpr-2023-1
['person-re-identification']
['computer-vision']
[ 4.43715721e-01 -2.64980912e-01 2.51256585e-01 -6.07369363e-01 -3.02677333e-01 -7.13320255e-01 4.89239544e-01 -3.70684266e-01 -1.60110623e-01 5.20635009e-01 3.05685967e-01 4.76741225e-01 1.64125517e-01 -6.71511471e-01 -9.15546238e-01 -6.23574734e-01 1.14276223e-01 1.44549429e-01 -3.63304824e-01 -2.39874125e-01 -3.43203157e-01 4.77997273e-01 -1.45009422e+00 4.04745899e-02 7.68339276e-01 8.83158505e-01 -1.80913761e-01 7.88740098e-01 3.61304969e-01 2.63682604e-01 -4.22581375e-01 -8.36393774e-01 8.35002363e-01 -7.28866518e-01 -3.43389630e-01 7.16352582e-01 9.49503958e-01 -3.27861011e-01 -4.02599007e-01 1.11925924e+00 3.83454055e-01 2.45350718e-01 7.31348872e-01 -1.13634431e+00 -1.23051047e+00 -5.46180643e-03 -7.39883482e-01 -3.72477323e-01 5.20475090e-01 1.72382399e-01 5.54287672e-01 -8.34095240e-01 5.70622265e-01 1.05213869e+00 1.17135036e+00 9.98326659e-01 -1.59191883e+00 -3.78825426e-01 3.86965781e-01 -1.11411594e-01 -1.57960725e+00 -2.39580542e-01 1.20696139e+00 -3.77591401e-01 2.52383590e-01 9.14327621e-01 1.22632420e+00 1.13781083e+00 -8.95255879e-02 5.23017764e-01 1.36621535e+00 -5.14553070e-01 -1.49594083e-01 4.39678550e-01 -3.48005444e-01 5.97274423e-01 -5.32655083e-02 1.32908255e-01 -2.93374360e-01 4.22268249e-02 9.98931766e-01 2.02101946e-01 -1.00032978e-01 -3.22803706e-01 -1.20636714e+00 7.00136960e-01 4.57182527e-01 -8.41464251e-02 -1.69259727e-01 1.35577032e-02 2.78199732e-01 2.49675423e-01 7.12012291e-01 3.74003828e-01 -3.40576112e-01 3.78501207e-01 -5.89046657e-01 5.87216854e-01 3.91685307e-01 1.12534142e+00 6.79637074e-01 -1.91115931e-01 -4.34924543e-01 1.25050581e+00 1.36014298e-01 8.83476257e-01 1.21998616e-01 -8.08076441e-01 2.02047214e-01 5.37966013e-01 2.33032674e-01 -1.17613089e+00 -1.85237117e-02 -2.15377912e-01 -1.16780555e+00 -5.73721454e-02 4.57526058e-01 -2.09323224e-02 -9.32728827e-01 1.94597864e+00 5.41207552e-01 9.80702043e-02 -3.56027484e-01 1.14836967e+00 6.52965367e-01 2.38584325e-01 1.86791494e-01 -1.43960461e-01 1.37021112e+00 -8.71442020e-01 -6.58067107e-01 -3.25731874e-01 -2.38580301e-01 -9.12038803e-01 1.18351042e+00 8.89431983e-02 -1.02105165e+00 -1.06353426e+00 -9.56586897e-01 1.07115336e-01 -2.76160389e-01 2.01978952e-01 4.69904542e-01 8.16279531e-01 -7.68046558e-01 4.44572747e-01 -3.39499772e-01 -4.59606379e-01 2.31628641e-01 3.55958253e-01 -3.57797056e-01 -2.35762835e-01 -1.27867007e+00 7.10245311e-01 -9.58724692e-03 3.28851014e-01 -6.16913795e-01 -9.46974635e-01 -1.03229547e+00 -5.71850836e-01 1.42630011e-01 -9.73666847e-01 6.69090092e-01 -1.54212475e+00 -1.46408176e+00 1.23309398e+00 9.63105587e-04 1.88684817e-02 9.06372666e-01 -9.52879414e-02 -8.23007762e-01 -2.42992699e-01 -3.25488560e-02 5.26168585e-01 1.35212469e+00 -1.64873385e+00 -3.25691909e-01 -3.40375483e-01 -6.16361909e-02 2.85372436e-01 -3.14495146e-01 -3.90606299e-02 -7.39292085e-01 -1.31427777e+00 -1.48601428e-01 -1.26975393e+00 -2.56149322e-01 3.05508047e-01 -4.48082149e-01 2.89006710e-01 4.80134696e-01 -1.00831962e+00 1.02678442e+00 -2.12812710e+00 2.31762379e-01 4.27083403e-01 1.98394045e-01 -1.24744117e-01 -1.49881676e-01 5.28641865e-02 -1.27393072e-02 -1.84678212e-01 -3.26090276e-01 -6.49803936e-01 1.15754955e-01 2.70695597e-01 2.93086693e-02 8.35525990e-01 4.06411439e-01 9.35361266e-01 -7.18371212e-01 -3.92423958e-01 5.04675746e-01 7.25019813e-01 -4.87412065e-01 5.01373231e-01 -4.39619571e-02 7.86428511e-01 3.09258327e-02 7.57263064e-01 1.04499578e+00 2.44255573e-01 1.19929008e-01 -6.26911938e-01 1.45953655e-01 -6.74004197e-01 -1.16469777e+00 1.40328753e+00 -3.19812715e-01 2.46072710e-01 1.95165966e-02 -4.90581512e-01 1.05352890e+00 -1.48104131e-01 5.92547238e-01 -5.27960539e-01 9.18135345e-02 -1.63937330e-01 -4.21239793e-01 -2.44192109e-01 3.36524069e-01 -1.82018012e-01 -3.81088257e-01 7.78548196e-02 -2.22750872e-01 4.71687037e-03 -2.28436381e-01 -3.11355084e-01 3.54725659e-01 1.81023896e-01 9.40300301e-02 -2.44437575e-01 5.77523530e-01 -3.85620594e-01 4.36392426e-01 5.08485258e-01 -2.03564689e-01 8.26023877e-01 -3.47151548e-01 -6.12627685e-01 -1.50096881e+00 -1.27507138e+00 -2.35296518e-01 1.12861633e+00 3.22532594e-01 2.51016021e-02 -9.43299651e-01 -7.83359826e-01 2.46840492e-01 3.11149120e-01 -1.13129771e+00 -2.27716714e-01 -4.65225011e-01 -8.04684877e-01 3.92028123e-01 6.19206131e-01 6.65770888e-01 -6.86865151e-01 1.16405208e-02 -2.22210996e-02 -1.35493740e-01 -9.46207941e-01 -1.36451423e+00 -3.52102518e-01 -1.11769542e-01 -8.92267287e-01 -8.76227319e-01 -8.81377697e-01 1.33960366e+00 3.28693949e-02 1.03064418e+00 -3.20163928e-02 -6.55127943e-01 5.89684308e-01 -3.05675626e-01 -1.24992892e-01 -4.27678734e-01 -5.55779994e-01 3.71276647e-01 6.22367501e-01 4.07144040e-01 -2.55887449e-01 -8.42775941e-01 5.52518487e-01 -6.21665120e-01 2.12315798e-01 1.83898389e-01 9.63729680e-01 8.11118722e-01 2.02050343e-01 2.07865477e-01 -6.90923631e-01 4.47025657e-01 -2.29488403e-01 -2.25237399e-01 4.23166156e-01 -5.43210983e-01 -4.01755631e-01 6.84473515e-01 -1.04462075e+00 -9.12213206e-01 3.73672605e-01 -3.85230258e-02 -6.46962047e-01 -2.72177935e-01 -1.86031923e-01 -5.45578420e-01 -2.56989241e-01 3.58689517e-01 4.13733453e-01 -4.71047685e-02 -2.86426723e-01 7.34605193e-01 3.28831911e-01 8.70396674e-01 -6.99103832e-01 1.14251459e+00 4.16364461e-01 -1.11553147e-01 -5.36324978e-01 -7.07019746e-01 -3.52238655e-01 -9.89100099e-01 -6.07715249e-01 9.85193074e-01 -1.01933932e+00 -5.60130119e-01 7.62406111e-01 -7.16138363e-01 -2.97771066e-01 -6.31079316e-01 1.34029821e-01 -4.87755358e-01 2.52022982e-01 -4.47132796e-01 -8.14663649e-01 -2.86693871e-01 -6.48067236e-01 1.07363212e+00 2.54440606e-01 -2.66833663e-01 -1.15880227e+00 2.02638686e-01 2.31314361e-01 1.75318018e-01 8.60518217e-01 4.76517409e-01 5.57156047e-03 -1.07090017e-02 -2.81642228e-01 -1.14894323e-01 6.62724733e-01 6.55421913e-01 -1.44894227e-01 -1.00813198e+00 -5.40180147e-01 -3.07401985e-01 2.23429780e-02 5.44230521e-01 3.10826331e-01 1.32532108e+00 -7.58685946e-01 3.03179510e-02 8.90897870e-01 1.67595232e+00 -1.46418124e-01 4.39181924e-01 1.35382161e-01 1.08383596e+00 7.31945872e-01 4.34276551e-01 5.75285316e-01 5.02255797e-01 8.64780188e-01 9.80556756e-02 -5.94640136e-01 -4.37546432e-01 -5.50101101e-01 2.83421218e-01 6.21746957e-01 -3.96706969e-01 1.33459091e-01 -2.87830710e-01 5.96693397e-01 -1.56919074e+00 -8.86433065e-01 -6.16759621e-02 2.32532239e+00 1.09896183e+00 -2.58979321e-01 7.41668940e-01 -1.01644233e-01 9.69816864e-01 -5.03119789e-02 -7.48954833e-01 -3.98131013e-01 -2.24034339e-01 -4.74419817e-02 6.95604086e-01 4.26433563e-01 -1.26018083e+00 5.87751865e-01 6.08430767e+00 5.10204732e-01 -7.17514932e-01 3.63997184e-02 6.72162950e-01 -1.61249503e-01 -3.56258869e-01 -6.18628263e-01 -6.27220154e-01 7.76579320e-01 2.53210187e-01 6.17939532e-02 1.12757170e+00 7.70553350e-01 -1.00177065e-01 3.82837415e-01 -9.89536583e-01 1.25736272e+00 4.11070704e-01 -8.04684579e-01 -1.65059656e-01 1.08332105e-01 1.14918840e+00 -7.79521883e-01 3.40076447e-01 1.95865035e-01 3.26788574e-01 -9.22298074e-01 8.92986000e-01 9.42159653e-01 1.52440047e+00 -9.67365146e-01 5.65680265e-01 -2.54866302e-01 -1.48315752e+00 3.35368626e-02 -3.25996280e-01 1.76660299e-01 4.69525903e-02 3.41197610e-01 -5.59719861e-01 4.67796206e-01 7.34111190e-01 5.64281046e-01 -8.77337515e-01 4.70370293e-01 9.84166935e-02 2.72747695e-01 -2.00477600e-01 2.33039334e-01 -3.10623974e-01 -3.92746419e-01 3.98537815e-01 1.37544131e+00 8.08634833e-02 -9.55696404e-02 4.77017641e-01 9.25902724e-01 -1.12340301e-01 2.62592025e-02 -3.81281137e-01 3.68926376e-01 2.84035265e-01 1.24601638e+00 -3.19624871e-01 -1.82207957e-01 -3.83608788e-01 1.80256927e+00 4.91895787e-02 2.94804633e-01 -1.09127188e+00 9.87379551e-02 9.47007656e-01 4.14977908e-01 2.73213327e-01 -2.45506875e-02 -2.51879662e-01 -9.68860924e-01 2.38970965e-01 -8.67145061e-01 2.44600371e-01 -4.14833993e-01 -2.31055307e+00 5.71693599e-01 1.73087828e-02 -1.33019471e+00 -1.00349467e-02 -3.25508147e-01 -4.15804565e-01 1.01156104e+00 -1.03138483e+00 -2.16070676e+00 -6.73083007e-01 9.33804095e-01 5.16616881e-01 -8.59891176e-02 8.53910387e-01 3.45309675e-01 -4.27621633e-01 1.20453608e+00 7.74543136e-02 2.28056014e-01 8.67849946e-01 -1.50759280e+00 4.83125240e-01 7.77334869e-01 -2.43458748e-01 7.83958375e-01 7.25800455e-01 -8.38580310e-01 -1.46431935e+00 -1.64177179e+00 4.05458689e-01 -7.05945313e-01 4.79615510e-01 -7.37924159e-01 -5.48911273e-01 6.78914130e-01 -3.45498770e-02 2.95882374e-01 9.45577919e-01 6.46566972e-02 -4.91078943e-01 -4.94184405e-01 -1.55160832e+00 6.45495117e-01 1.25755727e+00 -6.44173980e-01 -2.80527174e-01 2.33232155e-01 5.26087523e-01 -4.64813173e-01 -1.39121902e+00 3.48766923e-01 9.39186394e-01 -4.37236309e-01 1.20419097e+00 -5.47033131e-01 1.51536256e-01 -6.18537486e-01 -4.45742071e-01 -1.36726606e+00 -8.03570509e-01 -5.55640996e-01 -1.79460630e-01 1.70431912e+00 -1.80285066e-01 -2.91756719e-01 5.55224478e-01 9.83227372e-01 2.74533510e-01 -6.21671975e-01 -5.99588394e-01 -7.59154499e-01 -1.82512149e-01 -2.05647275e-01 9.79863286e-01 1.14715087e+00 -4.92659003e-01 -1.77688986e-01 -1.02505553e+00 2.60577291e-01 9.94235575e-01 -4.38386835e-02 9.97978151e-01 -7.43346035e-01 -4.58559006e-01 -1.60837129e-01 -2.74185866e-01 -7.29407132e-01 1.98599666e-01 -6.85187101e-01 1.15924537e-01 -1.13022697e+00 6.55189574e-01 -6.30464375e-01 -2.60705650e-01 2.40522757e-01 -5.61204016e-01 8.32406282e-01 3.58075529e-01 3.65288891e-02 -4.42408741e-01 4.80339020e-01 1.54392588e+00 -4.20061022e-01 -1.24278381e-01 7.17178807e-02 -8.16707611e-01 5.20065188e-01 6.25807405e-01 1.93282589e-02 -3.12438399e-01 -1.56016260e-01 -6.51597530e-02 -5.85201800e-01 8.61148894e-01 -7.97595739e-01 -1.82778731e-01 -3.11261535e-01 1.24801719e+00 -2.67458856e-01 3.03944916e-01 -1.15866911e+00 6.72713399e-01 6.18917681e-02 -3.58859003e-01 -5.48124723e-02 1.80358917e-01 5.23793161e-01 3.05269271e-01 2.76128203e-01 9.41604197e-01 -1.82889894e-01 -6.01026833e-01 5.76100588e-01 -1.89686250e-02 -2.22423896e-01 1.06221044e+00 -4.18324858e-01 7.95987323e-02 -1.97009176e-01 -8.24920893e-01 -4.14907113e-02 9.40866470e-01 6.05980992e-01 6.40350580e-01 -1.97311687e+00 -9.66565311e-01 5.62468708e-01 3.95275772e-01 -2.15071753e-01 6.02589369e-01 3.50405246e-01 -2.51532733e-01 -3.38279575e-01 -4.27348942e-01 -5.19413829e-01 -1.44728327e+00 9.23093975e-01 4.75932509e-01 -1.66770771e-01 -5.48601151e-01 9.36257541e-01 3.01693857e-01 -7.93720782e-01 -9.33499262e-02 9.13402066e-02 1.14211626e-01 -2.99199551e-01 5.72297633e-01 3.39200348e-01 -3.22668046e-01 -9.36683893e-01 -3.68230790e-01 1.02906728e+00 3.91879231e-02 1.38546169e-01 1.09193885e+00 -5.38846135e-01 -1.68571413e-01 1.76448435e-01 1.24901950e+00 3.39083999e-01 -1.84060311e+00 -1.17586158e-01 -5.95738947e-01 -9.48278189e-01 -3.52928877e-01 -1.06251323e+00 -1.03308451e+00 8.03131089e-02 1.07564306e+00 1.03042774e-01 1.52425325e+00 1.12886513e-02 8.65814984e-01 -4.17469919e-01 3.03453475e-01 -1.16984248e+00 5.66384569e-02 -1.56579822e-01 1.16634464e+00 -1.16476488e+00 1.89738959e-01 -4.43010002e-01 -1.05497873e+00 7.46967494e-01 7.40776300e-01 -2.23765731e-01 5.31315207e-01 1.61537200e-01 -8.20198748e-03 6.83049187e-02 1.07767053e-01 -1.70742810e-01 9.16993976e-01 1.03734529e+00 6.81397766e-02 4.55552936e-01 1.76845565e-02 7.48781681e-01 -3.74466538e-01 -3.19360465e-01 -1.77794144e-01 3.38270158e-01 2.81912774e-01 -1.01768148e+00 -6.92606628e-01 3.14906776e-01 -2.01131850e-01 1.29950196e-01 -5.21010816e-01 7.39974916e-01 8.47306013e-01 9.16097343e-01 2.12190077e-01 -7.63822675e-01 5.97843230e-01 -3.79942536e-01 1.01972616e+00 -2.23926604e-01 -5.72854102e-01 1.88850865e-01 5.57233021e-02 -3.04045469e-01 -4.97139394e-01 -7.88579822e-01 -5.80478132e-01 -6.90923214e-01 -9.72704366e-02 -3.86256337e-01 5.28498530e-01 5.28201640e-01 -2.71439124e-02 5.44804811e-01 1.16414058e+00 -7.58209884e-01 -2.96504766e-01 -8.59003365e-01 -7.19210625e-01 1.15051317e+00 4.02812749e-01 -5.39799452e-01 -5.77993952e-02 7.57701337e-01]
[12.029805183410645, -0.837383508682251]
aa17a5eb-01c5-45c6-b2dc-b32af2f5431e
dynamic-portfolio-optimization-with-inverse
2112.15499
null
https://arxiv.org/abs/2112.15499v2
https://arxiv.org/pdf/2112.15499v2.pdf
Dynamic Portfolio Optimization with Inverse Covariance Clustering
Market conditions change continuously. However, in portfolio's investment strategies, it is hard to account for this intrinsic non-stationarity. In this paper, we propose to address this issue by using the Inverse Covariance Clustering (ICC) method to identify inherent market states and then integrate such states into a dynamic portfolio optimization process. Extensive experiments across three different markets, NASDAQ, FTSE and HS300, over a period of ten years, demonstrate the advantages of our proposed algorithm, termed Inverse Covariance Clustering-Portfolio Optimization (ICC-PO). The core of the ICC-PO methodology concerns the identification and clustering of market states from the analytics of past data and the forecasting of the future market state. It is therefore agnostic to the specific portfolio optimization method of choice. By applying the same portfolio optimization technique on a ICC temporal cluster, instead of the whole train period, we show that one can generate portfolios with substantially higher Sharpe Ratios, which are statistically more robust and resilient with great reductions in maximum loss in extreme situations. This is shown to be consistent across markets, periods, optimization methods and selection of portfolio assets.
['Tomaso Aste', 'Yuanrong Wang']
2021-12-31
null
null
null
null
['portfolio-optimization']
['time-series']
[-3.68615091e-01 -5.03973126e-01 1.64282937e-02 -2.56506111e-02 -4.14286137e-01 -1.16861534e+00 8.27960312e-01 -2.26498336e-01 2.27280855e-02 5.78945756e-01 1.15270950e-02 -6.40870929e-01 -9.14117873e-01 -1.00227058e+00 -2.26760492e-01 -7.44895995e-01 -2.24818960e-01 6.49702728e-01 1.74457416e-01 -1.15303427e-01 7.10261405e-01 6.22840106e-01 -1.24879360e+00 1.17808089e-01 6.52780414e-01 1.23449874e+00 -2.36274883e-01 4.23837274e-01 4.40469384e-02 6.58382475e-01 -4.20301348e-01 -6.08489692e-01 8.56932640e-01 -4.41343337e-01 -3.17098796e-01 2.27224514e-01 -3.84589940e-01 2.10257038e-01 2.94608831e-01 1.07469416e+00 1.68593213e-01 2.26655841e-01 7.60903656e-01 -1.03445816e+00 -3.64659429e-01 7.09083974e-01 -6.39118731e-01 4.64824617e-01 -6.91606179e-02 1.47896841e-01 1.25646353e+00 -7.38496363e-01 5.16789377e-01 9.42973316e-01 7.60750949e-01 -7.81118646e-02 -1.31757987e+00 -7.10257888e-01 -8.76311772e-03 -7.68288523e-02 -1.12236190e+00 -1.34948820e-01 9.26754773e-01 -8.38856578e-01 7.94311941e-01 3.34715188e-01 6.12426281e-01 4.37206239e-01 5.02246261e-01 1.50897577e-01 1.48801851e+00 -2.41747648e-01 3.99801224e-01 3.84254366e-01 1.69640750e-01 -1.72099426e-01 2.93889731e-01 6.59620941e-01 -2.82230794e-01 -3.40562195e-01 5.81261873e-01 1.65044535e-02 1.18521646e-01 -2.95206338e-01 -1.20431435e+00 1.00068116e+00 -1.41133830e-01 3.67935658e-01 -6.72151685e-01 -1.70995101e-01 1.74368739e-01 8.17318678e-01 7.94398308e-01 5.64904094e-01 -5.57054937e-01 -3.05509180e-01 -1.00722730e+00 3.42045456e-01 1.03673112e+00 6.31483316e-01 3.46286565e-01 8.86653960e-02 -8.98311660e-03 3.88566136e-01 3.01118046e-01 6.86964452e-01 4.21209395e-01 -1.03265846e+00 4.74011958e-01 3.83343875e-01 2.25754857e-01 -1.04490077e+00 -4.18610841e-01 -7.50763476e-01 -5.12126744e-01 4.49504465e-01 3.43435884e-01 -5.05824268e-01 -1.61027804e-01 1.46456313e+00 1.82142898e-01 2.38367230e-01 1.23556659e-01 3.67405683e-01 -4.24643964e-01 4.73421991e-01 -5.13741672e-01 -5.84830999e-01 1.01646376e+00 -6.60658538e-01 -6.15283847e-01 2.24219009e-01 2.07075745e-01 -9.27613676e-01 3.74397635e-01 5.36372066e-01 -1.04646134e+00 -4.03643638e-01 -9.19092059e-01 1.23717034e+00 -4.19195831e-01 -3.74968708e-01 6.94697499e-01 9.17132258e-01 -9.19867456e-01 1.18114269e+00 -8.18331897e-01 6.29233420e-02 -1.11423999e-01 4.48904514e-01 1.99948117e-01 8.03255975e-01 -9.41767037e-01 9.15038884e-01 5.58450162e-01 5.23804650e-02 -3.69234204e-01 -7.81067073e-01 -1.43795133e-01 1.67033389e-01 2.75748730e-01 -4.78204668e-01 8.08081388e-01 -1.08098090e+00 -1.51983345e+00 2.87824512e-01 3.63278955e-01 -7.11206257e-01 6.76636636e-01 7.94683844e-02 -6.99482501e-01 -1.01041093e-01 -1.04008198e-01 -3.12091708e-01 8.71648133e-01 -8.66902113e-01 -6.48207545e-01 -2.03947052e-01 -3.82152498e-01 -8.69046822e-02 -1.95483401e-01 3.74584913e-01 1.96039438e-01 -9.72329080e-01 1.16275050e-01 -1.11786783e+00 -2.90968984e-01 -9.08507347e-01 -1.20203465e-01 1.55923069e-01 3.33970189e-01 -5.95712781e-01 1.36809671e+00 -2.23460221e+00 6.40771687e-02 9.46651697e-01 -2.46396437e-01 -3.88736933e-01 2.34966576e-01 6.35328114e-01 -6.22734666e-01 1.14282198e-01 -3.16292048e-01 -1.35859951e-01 3.75307649e-01 1.08699471e-01 -6.95145369e-01 6.99756086e-01 2.65439868e-01 6.81625307e-01 -5.02472818e-01 1.16584592e-01 2.60307014e-01 -3.48724961e-01 -6.10211015e-01 2.02160012e-02 -8.97591189e-02 4.06353742e-01 -2.94775665e-01 5.46088099e-01 7.45331764e-01 -1.31080151e-01 1.86468348e-01 2.46043980e-01 -5.34457505e-01 -3.05823404e-02 -1.60844934e+00 9.08494174e-01 -4.36999351e-02 4.45952326e-01 -2.62837440e-01 -1.11981547e+00 1.08614087e+00 3.53519320e-01 9.58760560e-01 -6.82231963e-01 4.17705476e-02 4.08811539e-01 1.26688302e-01 -3.04464754e-02 3.22767794e-01 -7.06771135e-01 -1.58523634e-01 1.00900650e+00 -1.06321931e-01 8.48995671e-02 3.42153668e-01 -3.99855316e-01 1.00255859e+00 -8.27864930e-02 1.18620187e-01 -8.42940450e-01 4.19587493e-01 -4.37669978e-02 8.01781833e-01 5.34491599e-01 -7.01327622e-03 4.54959810e-01 7.06807673e-01 -9.43314806e-02 -1.11374247e+00 -1.41801333e+00 -3.57390255e-01 3.88832152e-01 -3.78385335e-01 1.46654040e-01 -3.81060451e-01 -2.51131445e-01 3.91206384e-01 8.04578602e-01 -5.73427200e-01 -7.48308972e-02 -4.65029776e-01 -1.35773349e+00 -4.28287201e-02 2.30381504e-01 3.19804817e-01 -9.90812063e-01 -5.49576461e-01 6.75460756e-01 4.48946387e-01 -6.91492319e-01 -3.62507313e-01 1.36003077e-01 -1.06487632e+00 -1.01867282e+00 -6.54916227e-01 -8.61812010e-02 3.20921689e-01 -1.81808904e-01 1.32383537e+00 -3.69339257e-01 9.55582112e-02 5.70875823e-01 -3.80228549e-01 -5.71484804e-01 -3.85839522e-01 -2.28505269e-01 1.38057411e-01 5.92468560e-01 2.41775587e-01 -7.19192803e-01 -5.32029569e-01 4.21932071e-01 -6.68128073e-01 -5.21088302e-01 2.90837228e-01 7.96429753e-01 3.91319841e-01 8.29335511e-01 8.80495369e-01 -8.93319666e-01 8.30149353e-01 -7.72441149e-01 -1.25540030e+00 5.42797744e-01 -1.19838536e+00 3.70515510e-02 2.69072950e-01 -3.93632799e-01 -1.24961948e+00 -2.50270545e-01 4.91277277e-01 -4.61710602e-01 3.25218439e-01 7.11229920e-01 4.17589277e-01 -9.09022288e-04 7.68579468e-02 9.33107361e-02 2.54915774e-01 -5.24269819e-01 1.74102500e-01 2.40888223e-01 4.70596224e-01 -5.92528880e-01 1.29938078e+00 4.59184349e-01 6.44922182e-02 -1.96075678e-01 -4.37653720e-01 -5.63029528e-01 -6.03470862e-01 -4.26030487e-01 7.30092883e-01 -6.72820389e-01 -5.23519397e-01 4.91340846e-01 -5.93889832e-01 -2.69151092e-01 -4.19138730e-01 7.53330171e-01 -6.41950309e-01 1.44204557e-01 -5.27807713e-01 -1.16361904e+00 -2.02984348e-01 -9.21415627e-01 3.99165392e-01 7.79739618e-02 -2.13925973e-01 -1.61333275e+00 6.90676808e-01 8.84601548e-02 4.96572226e-01 7.37454236e-01 8.12754631e-01 -9.22272325e-01 -6.31493986e-01 -1.39079496e-01 9.19972062e-02 5.41218877e-01 3.03707242e-01 5.26571870e-01 -6.50681019e-01 -3.62042159e-01 6.18847132e-01 7.11600244e-01 5.13947368e-01 4.14897442e-01 5.04949450e-01 -3.29526901e-01 1.92100286e-01 5.41848183e-01 1.64954662e+00 7.15579391e-01 3.95619333e-01 7.77371526e-01 -4.59321514e-02 9.08279479e-01 9.18493032e-01 8.45932901e-01 -3.87378037e-02 4.86673236e-01 1.30916387e-01 -8.29431117e-02 5.12894928e-01 3.99807900e-01 6.31748140e-01 1.17550111e+00 -2.21399203e-01 1.49796292e-01 -8.88374209e-01 5.01718044e-01 -1.88379765e+00 -1.42214048e+00 -7.64920861e-02 2.33882642e+00 5.16368687e-01 4.34055269e-01 4.16792452e-01 7.62237608e-02 5.74905813e-01 -7.20266700e-02 -3.79221141e-01 -4.65525687e-01 -3.64851952e-01 2.95334160e-01 8.86276186e-01 2.45711342e-01 -1.02683926e+00 3.01752180e-01 6.86051846e+00 6.08849585e-01 -7.57517278e-01 -2.78372923e-03 6.67196870e-01 -1.59114450e-01 -4.84728009e-01 2.80468434e-01 -5.61619103e-01 8.02934945e-01 1.31448889e+00 -8.13173115e-01 6.67473733e-01 6.13313019e-01 3.13857704e-01 9.23924521e-02 -9.42543149e-01 4.42261487e-01 -3.74765843e-01 -1.43519163e+00 -4.45002854e-01 4.45491433e-01 1.13658369e+00 -9.87142101e-02 4.48889166e-01 1.86580077e-01 3.43676180e-01 -6.34515822e-01 1.03556037e+00 1.07803321e+00 3.63455783e-03 -1.28865671e+00 8.68420660e-01 1.24301806e-01 -1.30295873e+00 -4.18773025e-01 -2.67299056e-01 -9.26220044e-02 2.62923956e-01 7.87146509e-01 -2.81420350e-01 8.23832333e-01 8.43454361e-01 7.76755333e-01 -5.52342772e-01 9.58193362e-01 4.01948899e-01 4.55382288e-01 -4.21036750e-01 2.90748119e-01 2.47337714e-01 -1.05141902e+00 6.21122658e-01 8.32421064e-01 7.42782474e-01 -9.71164331e-02 -3.20800066e-01 1.11637175e+00 4.50005263e-01 -5.82971871e-02 -7.03822851e-01 -1.88492924e-01 4.16062623e-01 8.56403053e-01 -9.55977380e-01 -1.89255103e-01 -6.05908096e-01 3.94552141e-01 -3.86524945e-01 3.22121471e-01 -7.67265081e-01 -1.57020971e-01 7.06336200e-01 -1.49773136e-01 5.40825725e-01 -3.29571277e-01 -6.72048748e-01 -1.09121406e+00 1.69144586e-01 -1.01231706e+00 6.04159296e-01 -2.05829069e-01 -1.64789200e+00 1.98158711e-01 2.42688894e-01 -1.65749061e+00 -4.99003708e-01 -7.07481027e-01 -1.24721003e+00 8.56303096e-01 -1.40382361e+00 -3.65152806e-01 4.51707929e-01 6.14516318e-01 1.34821683e-01 -6.15547597e-01 2.91992426e-01 1.93586856e-01 -6.53073907e-01 1.66854024e-01 8.53642225e-01 -8.32654834e-02 4.19670075e-01 -1.66744280e+00 3.86116475e-01 1.04604876e+00 9.36815888e-02 6.90612137e-01 6.67221725e-01 -7.33563781e-01 -1.10369718e+00 -8.21470380e-01 5.90885997e-01 -6.87905848e-01 1.54582453e+00 5.65914661e-02 -8.36549997e-01 5.94709098e-01 4.37682688e-01 -7.77496815e-01 8.98306429e-01 8.64422768e-02 8.00335556e-02 -4.28336650e-01 -1.02699399e+00 4.35603291e-01 3.44678432e-01 -4.96798217e-01 -8.37824881e-01 5.07219970e-01 6.72379971e-01 2.16999575e-01 -1.67008424e+00 3.58651042e-01 3.55701417e-01 -1.37889361e+00 1.11601508e+00 -2.83988982e-01 -2.52312601e-01 -2.05591217e-01 -8.29512551e-02 -1.12285733e+00 -4.29364502e-01 -9.30587351e-01 1.33078739e-01 1.60391438e+00 7.01741755e-01 -1.38208568e+00 6.26369715e-01 8.50804448e-01 9.25568640e-02 -1.79712132e-01 -1.14722109e+00 -1.32008612e+00 4.33829248e-01 -3.41793925e-01 9.84860063e-01 1.03907323e+00 -7.53763318e-02 -3.08105260e-01 -1.44371778e-01 1.82358965e-01 1.09984720e+00 7.40351140e-01 5.40946305e-01 -1.36895728e+00 -7.54667580e-01 -9.83797312e-01 2.28102263e-02 -9.52348392e-03 4.05072514e-03 -5.99060476e-01 -3.02117765e-01 -5.11001468e-01 6.08579302e-03 -5.08936703e-01 -6.82924867e-01 -2.65917987e-01 9.19962302e-02 -2.19400346e-01 4.78246272e-01 5.08088112e-01 -1.85208678e-01 3.90263438e-01 7.73740888e-01 1.08366258e-01 -2.17837453e-01 5.43678105e-01 -3.95429522e-01 4.11639750e-01 9.14366543e-01 -6.34979546e-01 -2.76913017e-01 3.61795664e-01 4.89912897e-01 2.59192407e-01 1.00800343e-01 -9.72424209e-01 2.10549265e-01 -4.69136536e-01 9.73461941e-03 -8.29488337e-01 -9.25030112e-02 -9.52771842e-01 1.15400541e+00 7.07890332e-01 4.16959124e-03 7.36966431e-01 1.87757283e-01 6.73569977e-01 -4.46561426e-01 -4.63486761e-01 5.49518943e-01 -1.04792960e-01 -4.19967830e-01 1.65329114e-01 -3.88588041e-01 -1.90125152e-01 1.40210795e+00 -6.86887875e-02 -7.32365549e-02 -9.56839398e-02 -9.05358791e-01 4.26277667e-01 5.57980239e-01 1.66091815e-01 1.76297694e-01 -1.43141603e+00 -8.44678998e-01 1.38881892e-01 -3.46790880e-01 -6.13900125e-01 4.39768508e-02 1.13644159e+00 -5.38999259e-01 4.90029693e-01 -1.06840640e-01 -4.06420499e-01 -7.79028058e-01 7.62547791e-01 5.50497472e-01 -6.36273026e-01 -3.59855682e-01 3.59564751e-01 1.84292160e-02 -2.26381987e-01 -2.34843910e-01 -2.05549538e-01 -2.37954989e-01 6.91238344e-01 2.58036733e-01 5.17019391e-01 9.62594710e-03 -3.48232925e-01 -2.08143130e-01 1.00032377e+00 4.61577743e-01 -4.43266094e-01 1.57897925e+00 -1.76291853e-01 -3.76091987e-01 6.17372632e-01 1.00906181e+00 -5.89226708e-02 -1.48901510e+00 -4.73592058e-02 9.13923740e-01 -4.22860354e-01 -2.02828661e-01 -2.48671308e-01 -1.50171244e+00 4.74956155e-01 5.33652902e-01 7.94569552e-01 1.27083719e+00 -4.82759774e-01 3.24113876e-01 1.24011286e-01 3.45494956e-01 -1.44317102e+00 -1.66131277e-02 2.20024765e-01 8.28368545e-01 -8.28596830e-01 -1.35115429e-03 7.26262629e-02 -7.32826829e-01 1.32657075e+00 -2.90234268e-01 -5.18708169e-01 1.30838406e+00 1.33464143e-01 -2.24143323e-02 -3.06555152e-01 -9.86611724e-01 -6.18630741e-03 2.63088197e-01 2.83634305e-01 -1.12126425e-01 3.19036365e-01 -1.60760075e-01 5.08576274e-01 -4.01080251e-01 -5.44811726e-01 5.28110683e-01 9.96617794e-01 -2.88148731e-01 -1.16881192e+00 -6.25119328e-01 4.45952266e-01 -7.56852269e-01 3.17114070e-02 -2.41468415e-01 9.90828097e-01 -2.53382146e-01 9.67587531e-01 4.49395448e-01 -2.58274108e-01 4.48715359e-01 1.41495720e-01 -2.63418388e-02 -4.03808773e-01 -1.08870578e+00 5.67186415e-01 2.22773440e-02 -4.28492814e-01 -7.28738666e-01 -1.54955554e+00 -7.30223596e-01 -6.73335493e-01 -5.72453856e-01 2.67447978e-01 5.23969769e-01 8.49562466e-01 8.65166262e-02 5.51915646e-01 1.52847946e+00 -5.48600614e-01 -1.12044120e+00 -6.22049689e-01 -1.21831226e+00 2.98804075e-01 -6.44728169e-03 -7.77990758e-01 -9.13385630e-01 -1.12728402e-01]
[4.927839279174805, 4.069913864135742]
54c32506-243c-4633-a626-ff932cc5838b
language-based-audio-retrieval-task-in-dcase
2209.09967
null
https://arxiv.org/abs/2209.09967v3
https://arxiv.org/pdf/2209.09967v3.pdf
Language-based Audio Retrieval Task in DCASE 2022 Challenge
Language-based audio retrieval is a task, where natural language textual captions are used as queries to retrieve audio signals from a dataset. It has been first introduced into DCASE 2022 Challenge as Subtask 6B of task 6, which aims at developing computational systems to model relationships between audio signals and free-form textual descriptions. Compared with audio captioning (Subtask 6A), which is about generating audio captions for audio signals, language-based audio retrieval (Subtask 6B) focuses on ranking audio signals according to their relevance to natural language textual captions. In DCASE 2022 Challenge, the provided baseline system for Subtask 6B was significantly outperformed, with top performance being 0.276 in mAP@10. This paper presents the outcome of Subtask 6B in terms of submitted systems' performance and analysis.
['Tuomas Virtanen', 'Samuel Lipping', 'Huang Xie']
2022-09-20
null
null
null
null
['audio-captioning']
['audio']
[ 5.26318312e-01 3.50779714e-03 3.26706320e-01 -1.52614221e-01 -2.22903609e+00 -7.69491971e-01 6.94716454e-01 3.31521034e-01 -1.24502957e-01 6.25048339e-01 7.93061554e-01 1.52375758e-01 -1.24251775e-01 -8.87225196e-02 -8.34304810e-01 5.88714750e-03 -5.26182234e-01 4.86790329e-01 2.84027904e-01 -3.71710271e-01 1.96967900e-01 -9.81240198e-02 -1.68885517e+00 1.02907479e+00 -1.23651095e-01 1.40490067e+00 8.01232308e-02 1.12962174e+00 -7.62639195e-03 8.51808488e-01 -9.08112288e-01 -3.57993357e-02 -2.80507833e-01 -5.28202057e-01 -1.20131516e+00 -4.98531967e-01 6.49591446e-01 2.05231551e-03 -4.55616981e-01 6.99238002e-01 8.73332977e-01 3.13365832e-02 7.33211160e-01 -1.59778821e+00 -4.17724162e-01 9.48990941e-01 2.36279126e-02 3.19822669e-01 1.41089511e+00 -1.81280330e-01 1.55241287e+00 -1.17101038e+00 4.10119921e-01 1.20745897e+00 5.08554876e-01 6.64393842e-01 -9.76290762e-01 -8.18277657e-01 -7.02079237e-02 4.17240053e-01 -1.97679102e+00 -7.83732593e-01 6.59484327e-01 -6.46821916e-01 8.74546826e-01 8.49762321e-01 1.99054807e-01 1.09470189e+00 -4.37400013e-01 1.06779528e+00 3.22510213e-01 -4.86049920e-01 7.32701868e-02 3.53226811e-02 -1.97758321e-02 -3.42067987e-01 -4.83964264e-01 -3.23841929e-01 -1.26400685e+00 -4.85896260e-01 1.09071225e-01 -9.93291974e-01 -4.01604593e-01 4.74537432e-01 -1.48356318e+00 4.75140542e-01 9.41673890e-02 2.05745026e-01 -4.85512227e-01 5.22103071e-01 8.74556780e-01 5.03017962e-01 2.49183983e-01 9.38253164e-01 -1.26738504e-01 -4.66654927e-01 -1.03859591e+00 6.93608940e-01 6.01137996e-01 1.16101158e+00 5.09195924e-02 2.32970379e-02 -8.98730099e-01 1.18845522e+00 3.45844448e-01 5.66711366e-01 5.99102855e-01 -8.55997562e-01 7.37483561e-01 -9.72164869e-02 4.00237679e-01 -8.72927010e-01 -6.86256513e-02 -3.16299736e-01 -3.93103659e-01 -7.91057169e-01 -8.34623799e-02 -1.25468671e-01 -4.97646332e-01 1.62642157e+00 -3.30589056e-01 2.06260458e-01 2.49767393e-01 9.68757391e-01 1.43311703e+00 1.34265780e+00 3.89900394e-02 -2.15694189e-01 1.42379510e+00 -7.23743021e-01 -7.14814246e-01 -2.17146978e-01 2.40866423e-01 -1.17793143e+00 1.06118953e+00 5.72357118e-01 -1.27813983e+00 -6.78751588e-01 -8.92031848e-01 1.18897922e-01 -1.39580071e-01 -4.97119687e-02 1.38169779e-02 2.50336975e-01 -1.41553307e+00 -2.32354268e-01 -1.12549834e-01 -1.95615128e-01 -1.84081450e-01 4.90896925e-02 -1.93584427e-01 2.50923246e-01 -1.89459693e+00 2.22015366e-01 4.32935923e-01 -7.89416954e-02 -1.52808523e+00 -1.06205523e+00 -7.17485368e-01 1.46207407e-01 3.00004393e-01 -2.35206783e-01 1.92092216e+00 -4.40759033e-01 -1.20787275e+00 7.44758368e-01 -8.71755779e-02 -8.87010276e-01 2.05495253e-01 -3.86289299e-01 -7.75727212e-01 5.60856819e-01 1.82601705e-01 1.23086596e+00 7.20813513e-01 -1.12503207e+00 -6.71051264e-01 3.66785169e-01 6.24952205e-02 3.40420991e-01 -3.05856496e-01 5.40940821e-01 -6.29485846e-01 -8.17319095e-01 -1.33899584e-01 -8.01057220e-01 2.08573148e-01 -3.00637364e-01 -4.64749098e-01 -6.19021595e-01 7.20667362e-01 -6.82341278e-01 1.64064419e+00 -2.35974979e+00 -1.84078082e-01 8.92703906e-02 -1.94279030e-01 1.35100260e-02 -4.73107010e-01 8.98684859e-01 -3.01679134e-01 2.78649569e-01 4.72479761e-02 -1.19255953e-01 5.16555548e-01 -5.82727194e-01 -1.07462990e+00 -3.23523998e-01 3.78594071e-01 7.22477615e-01 -1.08631325e+00 -5.86136162e-01 -2.39474669e-01 4.88947898e-01 -3.28749597e-01 4.29872602e-01 -4.56742495e-01 4.12915573e-02 -3.28906119e-01 5.10872126e-01 6.22659959e-02 1.35414705e-01 -4.16344434e-01 -1.62670743e-02 1.45844474e-01 7.80256331e-01 -9.54203606e-01 1.76770306e+00 -4.59375679e-01 1.19196057e+00 5.08797206e-02 -5.47142982e-01 1.10846937e+00 1.28972328e+00 7.34567404e-01 -6.01901591e-01 -5.06209970e-01 1.69287667e-01 -5.27349949e-01 -6.38189375e-01 6.50346994e-01 -2.17903219e-03 -8.77409458e-01 4.01904076e-01 4.70831990e-02 -8.12648833e-01 2.09609538e-01 4.20994610e-01 1.21294487e+00 -1.21135533e-01 -1.44425586e-01 7.80652538e-02 8.14787865e-01 1.88040594e-03 -2.06111595e-01 8.54659140e-01 -1.70658335e-01 1.27941072e+00 1.87688738e-01 6.33854568e-02 -6.76090956e-01 -1.08989501e+00 1.75084293e-01 1.42272902e+00 -2.03599200e-01 -8.19859684e-01 -4.94455040e-01 8.94480012e-03 -3.02374095e-01 6.11262500e-01 -2.99079269e-01 -1.41797662e-01 -3.23760420e-01 -6.81364164e-02 1.33291781e+00 2.70418823e-01 1.69589505e-01 -1.32802534e+00 -3.30254734e-01 4.12287146e-01 -9.24281180e-01 -1.27605176e+00 -9.59820688e-01 -3.86707508e-03 -3.39857548e-01 -7.70298600e-01 -1.12023020e+00 -1.06443202e+00 -1.36539340e-01 9.50850323e-02 1.34654582e+00 -3.49861979e-01 -2.72548616e-01 9.91032600e-01 -7.66609013e-01 -8.32154036e-01 -3.59231919e-01 3.69013190e-01 -4.84641045e-02 8.62188712e-02 2.96519428e-01 -4.76308733e-01 -3.85678977e-01 2.69871742e-01 -1.12750554e+00 -2.33555242e-01 3.33316833e-01 4.75384325e-01 5.13521791e-01 -3.18103522e-01 1.09358215e+00 -3.87942344e-02 1.34168839e+00 -4.40546423e-01 -1.46378800e-01 1.27567485e-01 1.49145373e-03 -3.30416828e-01 4.79224712e-01 -5.33015966e-01 -4.75230455e-01 2.89234877e-01 -1.54015511e-01 -3.67543101e-01 -1.91517934e-01 7.38398314e-01 1.68955103e-02 4.67221469e-01 8.34211648e-01 5.04647791e-01 -3.81755531e-01 -5.63702226e-01 6.62454888e-02 1.16697228e+00 9.71125245e-01 -7.78926253e-01 7.35840619e-01 -2.15505555e-01 -4.25871611e-01 -1.05146241e+00 -7.41657913e-01 -9.56766903e-01 -7.79611394e-02 -7.51365662e-01 7.74804413e-01 -1.29664505e+00 -4.84354734e-01 -1.74291376e-02 -1.40586734e+00 1.55371353e-01 -2.58226305e-01 5.21855474e-01 -8.35604310e-01 -1.05891027e-01 -3.94214809e-01 -1.09456933e+00 -5.67919374e-01 -8.43423605e-01 1.66555142e+00 -3.09981018e-01 -7.51204193e-01 -3.86260360e-01 2.68464208e-01 5.14315188e-01 2.97652900e-01 1.34145647e-01 6.60433888e-01 -1.05007160e+00 -1.41231254e-01 -6.04836285e-01 4.70238402e-02 3.41291696e-01 -1.44191921e-01 -3.44281882e-01 -1.45815086e+00 -1.97953552e-01 -5.84898710e-01 -9.64847624e-01 5.35859406e-01 1.51685784e-02 1.29634023e+00 -2.77206123e-01 6.71620890e-02 -2.83243328e-01 8.66174042e-01 5.08756936e-01 6.11278296e-01 1.40056729e-01 1.50911175e-02 8.07113707e-01 9.88935769e-01 4.19578850e-01 3.28847580e-02 1.17534244e+00 3.16491783e-01 3.11548263e-01 -3.17983925e-01 -6.31673157e-01 7.13930070e-01 1.32882738e+00 4.78364199e-01 -2.69255400e-01 -1.15697789e+00 9.28298771e-01 -1.62847972e+00 -1.00228393e+00 -8.77463892e-02 1.99874878e+00 1.10506940e+00 1.78079963e-01 4.40719157e-01 4.98480916e-01 8.19556952e-01 8.32589492e-02 -2.01307923e-01 -3.05870444e-01 1.11920089e-01 -8.14193918e-04 -3.46119672e-01 4.82165664e-01 -1.05389345e+00 4.21031475e-01 7.06842184e+00 1.11278725e+00 -9.58575785e-01 -7.74790533e-03 4.11186874e-01 -2.25286424e-01 -1.56313702e-01 -4.52800035e-01 -4.79926616e-01 3.61838222e-01 1.45261717e+00 -7.32250929e-01 2.37015575e-01 3.92085224e-01 4.06110317e-01 3.81603628e-01 -1.58021665e+00 1.22621667e+00 3.12597096e-01 -1.04281044e+00 6.15751207e-01 -5.57189941e-01 2.76912600e-01 -1.59880325e-01 3.18721861e-01 7.22262800e-01 -3.81628960e-01 -1.12112296e+00 1.18950856e+00 4.91287529e-01 9.67386842e-01 -6.59785986e-01 6.96046710e-01 3.23281437e-02 -1.54213035e+00 8.60719159e-02 -6.69211745e-02 -1.25396103e-02 3.79079193e-01 1.88901082e-01 -1.48525763e+00 5.16621172e-01 8.75553310e-01 2.73555905e-01 -4.76266831e-01 1.53510332e+00 1.19957551e-02 9.89173353e-01 -2.78014928e-01 -2.12160751e-01 3.82548273e-01 6.71666086e-01 8.20577204e-01 1.64245796e+00 4.79527622e-01 -1.01603210e-01 1.08826019e-01 4.18118507e-01 -2.83418924e-01 3.68282944e-01 -5.44908047e-01 -4.42479610e-01 8.11390281e-01 6.59263849e-01 -1.63483828e-01 -2.92329520e-01 6.73359483e-02 4.83757019e-01 -5.99387586e-01 3.47047806e-01 -7.70656645e-01 -8.53903294e-01 2.47333035e-01 3.55184257e-01 -6.01225272e-02 2.21939936e-01 4.43190455e-01 -5.47814071e-01 1.75570562e-01 -1.00978053e+00 5.31997859e-01 -1.57772541e+00 -1.23142326e+00 1.00704074e+00 4.38111871e-01 -1.81652391e+00 -7.64057159e-01 -1.39258146e-01 -1.77564979e-01 6.44778907e-01 -1.12957847e+00 -8.06909084e-01 -9.34102014e-02 5.52273452e-01 1.03979349e+00 -2.63960689e-01 1.12823486e+00 8.32831621e-01 2.93197930e-01 6.75887167e-01 -2.41930187e-01 1.11979105e-01 1.05046010e+00 -9.50337052e-01 3.09614033e-01 2.53691703e-01 5.61943293e-01 1.41101599e-01 1.11775994e+00 -1.70622259e-01 -1.16190088e+00 -1.27068758e+00 1.41002083e+00 -4.18913960e-01 1.06222892e+00 -5.85713267e-01 -6.80760562e-01 3.00389081e-01 3.08043778e-01 -3.46572220e-01 8.53231370e-01 -2.28465170e-01 -4.21725035e-01 -3.44034940e-01 -5.62289774e-01 2.64335692e-01 8.01710248e-01 -1.27908397e+00 -8.69742990e-01 6.45077705e-01 1.30443037e+00 -2.84476668e-01 -9.72753406e-01 3.64361018e-01 5.53535223e-01 -1.62036193e-03 1.18998981e+00 -5.77806890e-01 3.75044882e-01 -4.48766083e-01 -5.14582515e-01 -9.43260193e-01 1.43108383e-01 -1.19873893e+00 1.64635658e-01 1.53005981e+00 8.33230376e-01 1.93107113e-01 4.15824682e-01 2.73153167e-02 -2.59737998e-01 -5.17175049e-02 -1.29401731e+00 -8.90894055e-01 -2.07658142e-01 -9.06733692e-01 6.42567635e-01 5.88520408e-01 3.14605623e-01 6.90746307e-01 -4.87736821e-01 2.67663561e-02 1.29686529e-02 -1.42245963e-01 6.19734943e-01 -1.26201260e+00 -1.08038045e-01 -3.77393991e-01 -6.27443790e-01 -1.05758619e+00 2.94036176e-02 -9.20787752e-01 7.35467136e-01 -1.38936102e+00 2.30883304e-02 7.59181455e-02 -5.32546997e-01 4.89343822e-01 2.93939590e-01 9.43277478e-01 4.21877295e-01 1.90040529e-01 -1.10229349e+00 3.82049471e-01 7.24597454e-01 -7.88958073e-01 -3.21319848e-01 3.27327281e-01 -7.13618815e-01 1.09792687e-01 5.91363847e-01 -4.94057417e-01 -5.46710610e-01 -3.42649668e-01 4.40709263e-01 5.73338151e-01 2.59649396e-01 -1.24600923e+00 4.34495449e-01 1.77315325e-01 -3.30435187e-01 -8.83880973e-01 7.85911024e-01 -6.58941805e-01 1.95339695e-01 9.34475958e-02 -1.23061156e+00 1.57152951e-01 4.30779576e-01 3.76194447e-01 -1.07805860e+00 -3.50279838e-01 3.05087622e-02 2.41909131e-01 -4.16344136e-01 5.21486811e-03 -9.60709691e-01 4.14811075e-01 6.01545095e-01 1.77436069e-01 5.48370034e-02 -1.27600443e+00 -1.04874635e+00 3.46264184e-01 -6.14997804e-01 7.49037266e-01 9.72123981e-01 -1.79520571e+00 -1.29550898e+00 -4.57759023e-01 6.96657836e-01 -1.75222754e-01 1.32465698e-02 3.00435185e-01 -2.00720802e-01 1.19127977e+00 1.54825538e-01 -7.46428549e-01 -1.38422477e+00 7.71473870e-02 7.84217492e-02 -5.83010167e-02 -5.92155345e-02 9.89798069e-01 2.30038026e-03 5.86312376e-02 9.98001814e-01 -2.48749867e-01 -4.51244980e-01 3.09786797e-01 8.45514059e-01 7.67092332e-02 2.37678573e-01 -7.74904668e-01 -4.37534690e-01 2.60475188e-01 2.25573212e-01 -8.83439183e-01 1.09971607e+00 -5.17804995e-02 8.17322433e-02 9.68412519e-01 1.26331449e+00 -2.13173360e-01 -4.74518538e-01 -1.51311591e-01 4.62794155e-01 7.64918700e-02 1.55896749e-02 -1.03972578e+00 -3.37777734e-01 1.02228010e+00 5.55307269e-01 5.43318689e-01 1.26555884e+00 2.08267972e-01 8.89370203e-01 6.90546811e-01 3.94344866e-01 -1.00255728e+00 4.19801533e-01 6.82675004e-01 1.62758195e+00 -8.14351797e-01 -3.42530102e-01 -8.23354200e-02 -5.76914430e-01 9.64805603e-01 5.24723589e-01 2.77006626e-01 3.67073536e-01 -1.20913815e-02 1.70678005e-01 -5.64359240e-02 -1.12900722e+00 6.97290748e-02 8.94234598e-01 6.26696467e-01 5.84214568e-01 -4.44118157e-02 9.68095567e-03 9.08088446e-01 -5.48310876e-01 -1.06585458e-01 2.56684422e-01 7.54230678e-01 -4.38759029e-01 -8.59508157e-01 -7.04690337e-01 7.55482242e-02 -5.53149402e-01 -2.34516636e-01 -1.01249397e+00 4.09720093e-01 -4.38153774e-01 1.32983983e+00 -4.33579162e-02 -5.64798117e-01 5.52159429e-01 4.33329254e-01 -2.45698467e-02 -8.26422930e-01 -8.53032291e-01 2.06669658e-01 4.02120113e-01 -3.89054239e-01 -3.54580462e-01 -4.10532385e-01 -9.14302886e-01 6.70947790e-01 -5.34006208e-02 1.22155726e+00 5.80676258e-01 4.75608379e-01 2.52975583e-01 5.70906460e-01 7.07513034e-01 -7.63763666e-01 -4.46468711e-01 -1.20356357e+00 -2.97295988e-01 1.87507853e-01 5.26764333e-01 -1.08810633e-01 -4.01864350e-01 4.01100606e-01]
[15.279250144958496, 4.890787601470947]
ffbe97e9-6327-419f-ab91-38af4c775940
unsupervised-learning-of-compositional-energy
2111.03042
null
https://arxiv.org/abs/2111.03042v1
https://arxiv.org/pdf/2111.03042v1.pdf
Unsupervised Learning of Compositional Energy Concepts
Humans are able to rapidly understand scenes by utilizing concepts extracted from prior experience. Such concepts are diverse, and include global scene descriptors, such as the weather or lighting, as well as local scene descriptors, such as the color or size of a particular object. So far, unsupervised discovery of concepts has focused on either modeling the global scene-level or the local object-level factors of variation, but not both. In this work, we propose COMET, which discovers and represents concepts as separate energy functions, enabling us to represent both global concepts as well as objects under a unified framework. COMET discovers energy functions through recomposing the input image, which we find captures independent factors without additional supervision. Sample generation in COMET is formulated as an optimization process on underlying energy functions, enabling us to generate images with permuted and composed concepts. Finally, discovered visual concepts in COMET generalize well, enabling us to compose concepts between separate modalities of images as well as with other concepts discovered by a separate instance of COMET trained on a different dataset. Code and data available at https://energy-based-model.github.io/comet/.
['Igor Mordatch', 'Joshua B. Tenenbaum', 'Yash Sharma', 'Shuang Li', 'Yilun Du']
2021-11-04
null
http://proceedings.neurips.cc/paper/2021/hash/838aac83e00e8c5ca0f839c96d6cb3be-Abstract.html
http://proceedings.neurips.cc/paper/2021/file/838aac83e00e8c5ca0f839c96d6cb3be-Paper.pdf
neurips-2021-12
['unsupervised-image-decomposition']
['computer-vision']
[ 1.56713836e-02 -3.62818658e-01 1.84201941e-01 -6.01826251e-01 -3.70623171e-01 -8.40987980e-01 7.18355060e-01 5.30920684e-01 -2.47222424e-01 2.74793804e-01 1.61674023e-01 3.59303206e-01 -1.07213981e-01 -8.42195392e-01 -8.90393496e-01 -8.13239634e-01 -3.85507792e-02 3.22467059e-01 -3.42343710e-02 6.40351176e-02 1.68890849e-01 2.80035615e-01 -2.11233950e+00 2.56597519e-01 8.53722870e-01 9.34683919e-01 6.17352426e-01 6.76713645e-01 -2.82364964e-01 6.22336090e-01 -4.28862363e-01 -1.92633048e-01 1.70678213e-01 -4.70660061e-01 -4.90280420e-01 2.80898541e-01 6.95333719e-01 1.26562670e-01 -5.66921234e-02 9.16466296e-01 3.05346027e-02 5.35634696e-01 7.42191315e-01 -1.25769091e+00 -1.03614461e+00 4.23440605e-01 -3.23000282e-01 -1.05554074e-01 2.30569482e-01 2.78531283e-01 1.19370341e+00 -1.02641833e+00 6.18643701e-01 1.40784025e+00 1.75000787e-01 5.65934360e-01 -1.55916989e+00 -3.06624860e-01 4.90124673e-01 4.35469776e-01 -1.48192692e+00 -2.10256323e-01 9.09101009e-01 -5.49733877e-01 7.61601508e-01 4.21104491e-01 1.06304872e+00 1.14091074e+00 -5.51846512e-02 1.04911137e+00 1.03913903e+00 -3.27517837e-01 5.31078100e-01 2.46306628e-01 -1.04150616e-01 5.99934638e-01 6.73771575e-02 3.08497134e-03 -6.99116647e-01 3.09335571e-02 4.61596400e-01 2.67140001e-01 -1.87580511e-01 -6.86602712e-01 -1.32022083e+00 7.99258411e-01 8.24896812e-01 3.20041120e-01 -3.83413076e-01 3.22434753e-01 -5.64206727e-02 -5.67523725e-02 5.14388263e-01 6.43480897e-01 -6.59356713e-01 4.35311459e-02 -6.76087439e-01 2.35647991e-01 5.81735969e-01 8.73221815e-01 1.31743562e+00 -3.55595320e-01 -2.14123353e-01 7.61074126e-01 2.41586521e-01 6.89677954e-01 2.96343118e-01 -7.77081192e-01 5.41616604e-02 6.62757814e-01 -7.46868253e-02 -1.03539455e+00 -2.19312549e-01 -3.33802998e-01 -6.65541887e-01 1.57169402e-01 1.69594675e-01 1.42575711e-01 -1.22839928e+00 2.16252756e+00 3.93292964e-01 2.75718927e-01 -1.21735334e-01 8.97787631e-01 7.10635066e-01 7.77254522e-01 1.27446905e-01 1.12342224e-01 1.41902208e+00 -9.55079734e-01 -4.41076875e-01 -3.38005513e-01 1.32011622e-01 -5.35482228e-01 1.16264248e+00 5.78206003e-01 -8.07071328e-01 -8.62876594e-01 -8.23269129e-01 -1.87805369e-01 -8.86672676e-01 6.03892654e-02 8.14968169e-01 2.98059791e-01 -9.96372163e-01 4.64393944e-01 -9.22885656e-01 -5.35159945e-01 5.04590392e-01 -1.86214507e-01 -1.15480967e-01 -2.51232415e-01 -8.37349296e-01 6.19153261e-01 6.85616136e-01 -1.56790167e-01 -1.15304887e+00 -9.31307495e-01 -1.13044298e+00 1.76000476e-01 5.19901276e-01 -9.90350187e-01 8.79718959e-01 -1.07920933e+00 -1.26521254e+00 7.32976139e-01 -3.11719954e-01 -2.13899374e-01 2.14087680e-01 -3.89790744e-01 -2.39442885e-01 1.60318896e-01 7.87580460e-02 1.00111556e+00 1.11138368e+00 -1.59742332e+00 -4.14221287e-01 -3.02123576e-01 1.32136375e-01 3.03814352e-01 -3.71200830e-01 -4.01470333e-01 -7.32089043e-01 -6.26490474e-01 -4.01275232e-03 -7.74509013e-01 -2.74700880e-01 3.83116215e-01 -4.71875817e-01 -1.37424484e-01 7.55199373e-01 -4.65721279e-01 8.78496230e-01 -2.18169618e+00 4.82881069e-01 2.54948795e-01 3.19466412e-01 -4.15153980e-01 -3.17387640e-01 2.18542367e-01 -1.14428394e-01 2.09471017e-01 -5.18326938e-01 -5.00752211e-01 9.78178680e-02 4.03638065e-01 -2.83780321e-02 1.13487050e-01 5.80371261e-01 1.04553914e+00 -1.22430444e+00 -2.82273054e-01 5.67845225e-01 7.90993869e-01 -5.68252265e-01 2.30044514e-01 -5.87745488e-01 6.45416915e-01 -4.77634341e-01 5.96168101e-01 7.75398731e-01 -3.88747722e-01 -5.83994165e-02 -3.00316960e-01 7.99015090e-02 -7.02731237e-02 -1.25625563e+00 2.13704610e+00 -6.71754479e-01 5.87953150e-01 -3.52311283e-01 -9.78904486e-01 8.56628776e-01 -1.24412455e-01 4.48825151e-01 -6.02399230e-01 8.31812471e-02 -7.73414895e-02 -3.88724893e-01 -6.17214143e-01 4.28468466e-01 -1.16960078e-01 -1.77965090e-02 2.82173187e-01 4.15927947e-01 -4.40196872e-01 4.22378421e-01 3.07793796e-01 7.74427593e-01 4.50352341e-01 1.80213258e-01 -3.04862559e-01 2.41856575e-01 -1.71517551e-01 3.49802166e-01 7.07965314e-01 3.53624582e-01 7.83678651e-01 6.31698370e-02 -3.20948780e-01 -8.37954581e-01 -1.41677403e+00 -2.85126686e-01 9.64168489e-01 3.51513922e-01 -6.74164653e-01 -5.98754048e-01 -5.20425677e-01 2.73567170e-01 8.26256990e-01 -1.00679755e+00 -2.14177802e-01 -1.47558555e-01 -8.37474704e-01 -2.13452756e-01 4.15117830e-01 2.86581606e-01 -9.39140797e-01 -9.06716526e-01 1.03705391e-01 -3.41599733e-01 -1.06194293e+00 -3.99869025e-01 3.42865676e-01 -7.23146141e-01 -1.07439530e+00 -5.29561281e-01 -3.97629648e-01 9.76580918e-01 3.83127779e-01 1.36634362e+00 1.51100636e-01 -9.78447795e-01 6.84026062e-01 -5.52520871e-01 -4.46577549e-01 4.35516424e-02 -3.36145341e-01 -8.32897983e-03 5.48257947e-01 2.94730604e-01 -6.91112936e-01 -8.07783961e-01 7.81903118e-02 -1.06672168e+00 2.86081314e-01 4.38617200e-01 5.91988981e-01 9.30942774e-01 2.02789620e-01 -1.67970695e-02 -5.28394401e-01 3.22937727e-01 -6.84287906e-01 -4.11844701e-01 3.95989418e-01 -1.91146925e-01 3.83459628e-01 5.27544379e-01 -5.93921781e-01 -1.12804830e+00 1.02233626e-01 3.62088740e-01 -6.92039251e-01 -5.79139650e-01 4.58365083e-01 -4.67278600e-01 3.30231756e-01 8.10095668e-01 4.16270435e-01 -4.11042064e-01 -6.41108215e-01 8.52143228e-01 -1.87015850e-02 5.23611009e-01 -9.05362010e-01 1.05720711e+00 6.79416180e-01 -1.62036046e-01 -7.42086411e-01 -9.80693817e-01 -6.14149749e-01 -7.21977353e-01 -2.16703206e-01 1.13160729e+00 -8.68261278e-01 -4.38668668e-01 4.69437093e-01 -8.23445320e-01 -3.28505248e-01 -6.96093857e-01 4.64094311e-01 -4.52588618e-01 -9.02210921e-02 -8.16443264e-02 -6.71400607e-01 1.20900393e-01 -8.82181108e-01 1.30052459e+00 5.27778983e-01 9.07567590e-02 -1.23123944e+00 2.41981763e-02 3.77312116e-02 3.38149637e-01 6.55233264e-01 7.61934698e-01 -2.10559368e-01 -8.30452442e-01 7.18274266e-02 -3.85180227e-02 5.47288477e-01 5.03793180e-01 1.94800973e-01 -1.14621103e+00 -3.20505649e-01 1.23297134e-02 -3.91311646e-01 1.21665192e+00 4.67159331e-01 1.66055083e+00 -2.37781003e-01 -3.13491046e-01 9.30679500e-01 1.89337802e+00 1.85211357e-02 3.93958420e-01 4.07715030e-02 9.44570005e-01 5.76387048e-01 3.35184187e-01 4.44830090e-01 3.53447050e-01 5.57495773e-01 4.37867403e-01 -1.10816456e-01 -1.34223893e-01 -4.44566131e-01 1.34794578e-01 7.43657351e-01 -1.08935088e-01 -3.91675055e-01 -7.91317582e-01 5.79714954e-01 -1.81482029e+00 -1.02350450e+00 1.38690367e-01 2.17122865e+00 6.08097255e-01 -1.78459510e-01 -1.75399646e-01 -3.49773377e-01 5.08923233e-01 1.16966575e-01 -1.08145463e+00 -7.77616203e-02 -3.42905492e-01 2.73117691e-01 1.37114093e-01 4.11261708e-01 -1.12720001e+00 7.01673687e-01 5.53868341e+00 7.24278688e-01 -1.04092062e+00 -5.86793460e-02 6.89681411e-01 -3.33456337e-01 -7.38723040e-01 3.16263199e-01 -3.44695061e-01 3.39604437e-01 5.09491682e-01 -3.64445865e-01 6.46755934e-01 9.10048544e-01 -7.11559579e-02 -2.33669490e-01 -1.23292315e+00 1.15778852e+00 4.69443798e-01 -1.11947513e+00 4.93213832e-01 3.73263881e-02 8.40828419e-01 -9.20703337e-02 1.09755263e-01 1.65933937e-01 3.47758770e-01 -9.72952664e-01 1.02142549e+00 1.00706637e+00 5.62202275e-01 -4.65497166e-01 2.38681823e-01 2.13319525e-01 -1.56989276e+00 -5.94460480e-02 -3.29581290e-01 1.33240923e-01 -4.29710709e-02 7.96775639e-01 -4.56231475e-01 7.74474919e-01 1.03704727e+00 1.19705296e+00 -1.00009906e+00 1.00054550e+00 -4.28962737e-01 3.91340554e-01 -4.22109932e-01 7.36509934e-02 -7.43766949e-02 -3.05455923e-01 5.40532649e-01 1.20331144e+00 4.98581380e-01 5.29730739e-03 3.59534562e-01 1.45284176e+00 -1.32292211e-01 9.14083868e-02 -2.60374010e-01 9.08572525e-02 2.19227687e-01 1.54380083e+00 -7.60409176e-01 -5.38174272e-01 -2.16832966e-01 1.07046485e+00 3.98939610e-01 6.57373786e-01 -6.55537128e-01 -1.90826848e-01 9.80993867e-01 -5.22001609e-02 3.31883818e-01 -2.84963101e-01 -5.08303801e-03 -1.67240000e+00 2.22460851e-01 -5.52909553e-01 3.62873852e-01 -1.13482666e+00 -1.49306750e+00 4.18708712e-01 2.87632346e-01 -1.39731693e+00 1.63009554e-01 -7.77525604e-01 -7.17286766e-01 8.33897173e-01 -1.47917724e+00 -1.03734565e+00 -8.15428674e-01 7.90979564e-01 7.02961981e-01 2.54450768e-01 7.30829060e-01 -1.49250954e-01 -4.14585561e-01 1.94067970e-01 1.59186319e-01 -1.34018883e-01 6.22641802e-01 -1.68338203e+00 3.77505392e-01 8.52318048e-01 6.07960761e-01 7.35461712e-01 6.90783560e-01 -3.10059696e-01 -1.32544208e+00 -1.02944982e+00 3.51901084e-01 -7.92922914e-01 5.78296781e-01 -7.37564564e-01 -9.84960198e-01 3.29581410e-01 1.38016894e-01 1.81608945e-01 7.05306411e-01 1.71559095e-01 -7.11937785e-01 -1.69151112e-01 -9.99476552e-01 4.57053304e-01 1.22616458e+00 -6.99236751e-01 -3.33558977e-01 3.48260045e-01 8.15087438e-01 -3.09314638e-01 -7.74213314e-01 3.13242413e-02 2.82518983e-01 -9.35044587e-01 1.07620728e+00 -4.55128402e-01 5.04386842e-01 -4.93461072e-01 -3.39569151e-01 -1.64626265e+00 -4.48787242e-01 -1.83430940e-01 -3.60594988e-01 1.12271023e+00 3.78660828e-01 -3.86284232e-01 3.75758052e-01 5.24259269e-01 -1.45507492e-02 -4.88150686e-01 -6.87347174e-01 -8.30699801e-01 -1.09318141e-02 -6.21188581e-01 9.19706702e-01 9.01459396e-01 -3.08346272e-01 2.56137669e-01 -2.27475688e-01 1.69770062e-01 8.71048987e-01 3.62609804e-01 6.26187980e-01 -1.35509443e+00 -4.46901768e-01 -4.71596658e-01 -4.58831072e-01 -9.22668338e-01 -9.53925550e-02 -1.06274116e+00 2.33905137e-01 -1.57095504e+00 7.34470725e-01 -3.80598545e-01 -5.95351815e-01 5.39949775e-01 -5.55624545e-01 1.62026770e-02 4.72329140e-01 1.15439020e-01 -5.16341627e-01 9.36076343e-01 1.39223933e+00 -4.75306720e-01 -1.57342777e-01 -4.85140711e-01 -9.61660981e-01 5.70299029e-01 8.52364898e-01 -4.43813860e-01 -6.26685798e-01 -7.70457089e-01 3.83359641e-01 -4.59026664e-01 8.95423830e-01 -1.23795319e+00 -6.37402087e-02 -5.25474846e-01 7.42635190e-01 -3.15801114e-01 4.44509834e-01 -8.01787436e-01 2.15887338e-01 1.10859863e-01 -2.71515578e-01 -2.36130029e-01 3.14616203e-01 7.57583797e-01 -3.82008493e-01 -1.92687176e-02 4.74658370e-01 -3.49305630e-01 -9.86795425e-01 5.35156727e-01 1.57986894e-01 -1.89475548e-02 9.87973630e-01 -2.15141237e-01 -2.26343021e-01 -2.44937375e-01 -8.47421646e-01 3.34382832e-01 6.01887584e-01 9.01228011e-01 6.75477207e-01 -1.43337941e+00 -8.29983830e-01 1.90039963e-01 6.52286291e-01 2.45633990e-01 5.72567582e-01 3.38722259e-01 -1.09173402e-01 2.59791277e-02 -1.67685032e-01 -8.73028398e-01 -1.01026869e+00 8.73324811e-01 3.11813354e-01 1.17610656e-01 -6.01617038e-01 9.76400018e-01 6.42272949e-01 -4.99734998e-01 -6.18930832e-02 -6.41851783e-01 -7.16428012e-02 2.29918212e-01 5.97076297e-01 -2.26285771e-01 -2.30752781e-01 -5.20746887e-01 -4.06108022e-01 9.13479149e-01 1.67682976e-01 3.87912542e-02 1.36184251e+00 -2.55359948e-01 -8.53151754e-02 7.18127012e-01 1.40351641e+00 -1.72581092e-01 -1.48099673e+00 -3.70364040e-01 -2.76064783e-01 -7.61957228e-01 1.63753368e-02 -9.23561811e-01 -1.15111995e+00 8.98500383e-01 7.55338371e-01 1.06655091e-01 1.54139960e+00 4.31033045e-01 4.18883711e-01 3.27278107e-01 5.01201689e-01 -1.08426440e+00 5.63053489e-01 2.91489512e-01 1.02685714e+00 -1.38862085e+00 -1.43413711e-02 -2.06994697e-01 -6.16944969e-01 1.06412089e+00 5.96379280e-01 -9.44393408e-03 7.84073710e-01 -2.12259024e-01 -1.18833102e-01 -3.66403461e-01 -9.25834239e-01 -5.13267457e-01 7.27063000e-01 6.42350852e-01 9.11769792e-02 4.02026474e-01 3.08488965e-01 3.67470920e-01 -2.01279774e-01 -2.38886625e-01 1.62762180e-01 8.75633359e-01 -3.20110142e-01 -9.93538141e-01 -2.99343735e-01 4.74978477e-01 2.09359705e-01 -1.69211894e-01 -2.76595980e-01 4.04877275e-01 6.52410328e-01 8.15861702e-01 3.91301423e-01 -3.98330957e-01 3.52083743e-01 -7.10177124e-02 4.24548537e-01 -7.55100429e-01 -1.02690332e-01 -1.83465287e-01 -3.43374521e-01 -6.88549459e-01 -6.83625579e-01 -9.07102227e-01 -1.06579924e+00 6.98197559e-02 -1.68215454e-01 3.65506336e-02 7.96845317e-01 7.42468417e-01 2.73682296e-01 6.91725850e-01 6.92196667e-01 -1.21752858e+00 1.94883451e-01 -8.24992597e-01 -4.91344780e-01 9.31564271e-01 4.62180316e-01 -7.44165301e-01 -5.95384121e-01 5.75236857e-01]
[9.837468147277832, 1.0989406108856201]
24825579-01d8-40dc-9f9e-32bc5090ceab
describing-a-knowledge-base
1809.01797
null
http://arxiv.org/abs/1809.01797v2
http://arxiv.org/pdf/1809.01797v2.pdf
Describing a Knowledge Base
We aim to automatically generate natural language descriptions about an input structured knowledge base (KB). We build our generation framework based on a pointer network which can copy facts from the input KB, and add two attention mechanisms: (i) slot-aware attention to capture the association between a slot type and its corresponding slot value; and (ii) a new \emph{table position self-attention} to capture the inter-dependencies among related slots. For evaluation, besides standard metrics including BLEU, METEOR, and ROUGE, we propose a KB reconstruction based metric by extracting a KB from the generation output and comparing it with the input KB. We also create a new data set which includes 106,216 pairs of structured KBs and their corresponding natural language descriptions for two distinct entity types. Experiments show that our approach significantly outperforms state-of-the-art methods. The reconstructed KB achieves 68.8% - 72.6% F-score.
['Lifu Huang', 'Heng Ji', 'Boliang Zhang', 'Qingyun Wang', 'Zhiying Jiang', 'Kevin Knight', 'Xiaoman Pan']
2018-09-06
describing-a-knowledge-base-1
https://aclanthology.org/W18-6502
https://aclanthology.org/W18-6502.pdf
ws-2018-11
['kb-to-language-generation', 'table-to-text-generation']
['natural-language-processing', 'natural-language-processing']
[-1.82020620e-01 8.94967556e-01 -4.35505301e-01 -3.59943002e-01 -1.19803917e+00 -4.19960499e-01 5.20162046e-01 3.73840809e-01 -5.28878212e-01 1.50015557e+00 6.78470433e-01 -2.06032917e-01 2.38335773e-01 -1.30905461e+00 -1.06796157e+00 -1.30102649e-01 1.70303866e-01 1.01057434e+00 4.75423664e-01 -5.01077831e-01 5.10365255e-02 -6.65245727e-02 -1.23166788e+00 6.12164855e-01 1.08393157e+00 1.09780419e+00 2.82988280e-01 2.55540609e-01 -7.65304625e-01 1.05537701e+00 -7.05936730e-01 -1.02973521e+00 -6.11040592e-02 -2.64258057e-01 -1.27982473e+00 -4.37637806e-01 7.28454217e-02 -1.78264976e-01 -3.33370209e-01 8.35910320e-01 4.57639903e-01 2.50773113e-02 5.73385775e-01 -9.77986872e-01 -1.05986607e+00 1.35656703e+00 -3.04557115e-01 3.33051443e-01 6.41419828e-01 -1.88047290e-01 1.41066754e+00 -9.42160428e-01 1.06963491e+00 1.11539400e+00 4.58525270e-01 5.27096748e-01 -9.95865166e-01 -5.53214073e-01 -4.59454060e-02 4.98713493e-01 -1.49078417e+00 -3.78693581e-01 4.54629093e-01 2.15727207e-03 1.47745836e+00 5.65739088e-02 5.00405490e-01 1.00956595e+00 -1.86329067e-01 8.57278287e-01 6.11670673e-01 -6.16358817e-01 1.49533764e-01 4.52141434e-01 3.21204960e-01 6.57343924e-01 5.92150867e-01 -3.84151846e-01 -6.96802795e-01 -2.41699398e-01 5.69318652e-01 -7.69201279e-01 -3.47445816e-01 -1.91854596e-01 -1.36649799e+00 5.90423048e-01 4.87795740e-01 6.51786104e-02 -5.27642906e-01 -1.67338457e-02 3.67910594e-01 -1.30160645e-01 1.82021379e-01 7.11243629e-01 -6.76282227e-01 5.94618439e-04 -4.19744521e-01 6.69513643e-01 9.64812279e-01 1.60545671e+00 9.89554346e-01 -3.51372272e-01 -6.56363070e-01 7.73439348e-01 2.94109941e-01 4.93106604e-01 7.05249846e-01 -5.12124240e-01 9.87444580e-01 8.53707492e-01 5.70179939e-01 -8.89621913e-01 -3.13931376e-01 -2.50259846e-01 -5.86989462e-01 -9.47903395e-01 6.30224764e-04 -2.64530871e-02 -8.25052023e-01 1.78682148e+00 2.85166234e-01 -2.17259047e-03 5.88250816e-01 6.23887300e-01 1.49623168e+00 8.59404504e-01 2.55765855e-01 -9.86611471e-02 1.52330983e+00 -1.08589733e+00 -9.28973973e-01 -3.78775865e-01 6.38568819e-01 -1.65656745e-01 1.15783215e+00 -3.18377495e-01 -1.16527784e+00 -4.99938816e-01 -9.09093320e-01 -4.91133243e-01 -7.45148182e-01 1.23041444e-01 4.18625832e-01 3.14803809e-01 -1.03478444e+00 2.04128668e-01 -4.41831112e-01 -2.31778011e-01 1.03966929e-01 1.86380520e-01 -4.65656519e-01 -3.28390040e-02 -1.84634209e+00 8.31179380e-01 1.09743357e+00 -1.04987890e-01 -4.70760316e-01 -8.71997833e-01 -1.28481352e+00 4.03809071e-01 6.12967312e-01 -1.14597249e+00 1.28299773e+00 -3.73540640e-01 -1.20556843e+00 7.73739159e-01 -3.75202298e-01 -6.72706783e-01 1.31533936e-01 -2.34311819e-01 -5.81579387e-01 -3.45705338e-02 5.95222771e-01 8.86419475e-01 6.80785999e-02 -1.32063854e+00 -8.63213301e-01 -3.27503346e-02 2.05848560e-01 3.97756964e-01 -2.37798691e-01 -3.14459771e-01 -9.60972726e-01 -6.11582220e-01 -2.03028813e-01 -5.87144434e-01 1.73714329e-02 -5.94441891e-01 -1.05124772e+00 -2.06714079e-01 8.48272070e-02 -1.07715261e+00 1.69772816e+00 -1.76474905e+00 -4.91216369e-02 1.61039636e-01 6.88177496e-02 1.53142735e-01 -1.19724452e-01 5.38978815e-01 9.10161063e-02 3.34353209e-01 -1.86326385e-01 -2.43768677e-01 2.29318976e-01 1.26789376e-01 -6.44147635e-01 -4.90129679e-01 3.83434296e-01 1.29694164e+00 -1.10456192e+00 -6.39194787e-01 -4.67059523e-01 1.53005317e-01 -6.38368905e-01 2.65634507e-01 -6.10805571e-01 -2.76082933e-01 -4.02496189e-01 5.07912219e-01 3.48382711e-01 -5.02820194e-01 4.63213772e-01 -3.79694223e-01 4.87117022e-01 1.01956880e+00 -1.11701381e+00 1.67098868e+00 -4.01191086e-01 2.06426188e-01 -6.12441123e-01 -4.04482841e-01 1.08930218e+00 3.25667292e-01 -4.42998558e-02 -8.73971939e-01 -3.21798027e-01 1.34397388e-01 -3.15317929e-01 -2.55872279e-01 1.30309153e+00 9.37605742e-03 -5.64968109e-01 3.05715919e-01 4.38956648e-01 1.99987248e-01 5.12245595e-01 5.36700308e-01 1.28434658e+00 2.68141925e-02 6.82725906e-01 -2.48653129e-01 5.34646630e-01 1.00814186e-01 6.86813891e-01 7.75586188e-01 3.43845129e-01 1.80122957e-01 7.43058681e-01 -5.08221805e-01 -1.06999826e+00 -1.13090193e+00 2.04333365e-01 8.43779087e-01 3.12832952e-01 -7.90504754e-01 -6.36283576e-01 -8.76391649e-01 2.42112353e-01 9.72470522e-01 -8.84010315e-01 -2.95495093e-01 -4.58776623e-01 -5.79904497e-01 8.27903926e-01 9.10859764e-01 7.14018583e-01 -1.46680927e+00 -1.27723679e-01 4.60720390e-01 -8.33645761e-01 -1.18489087e+00 -2.57728428e-01 6.92775324e-02 -2.99990982e-01 -8.57968032e-01 -3.72890949e-01 -7.59458721e-01 4.78699833e-01 -1.41270623e-01 1.75128078e+00 -1.74976394e-01 1.86661765e-01 -7.83209950e-02 -3.88273120e-01 -3.91197085e-01 -5.15972853e-01 6.53420210e-01 -8.01429674e-02 -3.91746193e-01 5.03029525e-01 -1.35296598e-01 -2.61473894e-01 1.13811493e-01 -8.69246781e-01 4.69064504e-01 6.59069538e-01 9.80360508e-01 8.54468644e-01 2.42166221e-02 8.44985664e-01 -1.12236738e+00 6.64188147e-01 -7.61739433e-01 -4.08431798e-01 6.67776525e-01 -6.33841336e-01 5.44461071e-01 5.45404494e-01 9.47826877e-02 -1.37905264e+00 -1.81682900e-01 -7.52499476e-02 9.30383280e-02 -4.93796952e-02 9.88404751e-01 -5.77636719e-01 6.69964969e-01 6.66179001e-01 3.68399620e-01 -5.78129351e-01 -3.23009133e-01 6.58989489e-01 6.79801941e-01 8.45074356e-01 -8.14944029e-01 6.07288301e-01 8.01574811e-02 -6.04376197e-01 -3.44392151e-01 -1.18870413e+00 -3.34837288e-01 -4.83988643e-01 3.88633251e-01 6.21411920e-01 -1.08709097e+00 -2.80418456e-01 -2.23469418e-02 -1.30428183e+00 -4.57727671e-01 -4.34366941e-01 -9.68668144e-03 -6.27100646e-01 9.00286064e-02 -7.33702481e-01 -4.14512902e-01 -6.07230663e-01 -6.78184509e-01 1.05447435e+00 2.22146049e-01 -2.22962037e-01 -8.25875521e-01 3.29760551e-01 4.87345546e-01 1.94701269e-01 -1.06588840e-01 1.33404338e+00 -8.89696360e-01 -6.31205499e-01 -3.75311971e-02 -4.09052730e-01 -2.17020974e-01 1.70536041e-01 -2.86695480e-01 -6.44359410e-01 1.30857691e-01 -9.95845318e-01 -4.20939356e-01 8.68830442e-01 -7.32188448e-02 9.65197802e-01 -7.67872989e-01 -5.68524361e-01 4.03519839e-01 1.45960104e+00 3.27846497e-01 9.55446064e-01 6.54705286e-01 5.85400283e-01 4.46458846e-01 7.21847594e-01 3.19420069e-01 9.98315275e-01 8.39285791e-01 1.50010884e-01 5.86044192e-02 5.41266464e-02 -7.93525219e-01 4.16555479e-02 5.86434662e-01 1.02982968e-01 -6.83446705e-01 -1.09646118e+00 7.90468156e-01 -1.93664205e+00 -1.06104386e+00 3.41356367e-01 1.95579112e+00 1.30795205e+00 3.55539560e-01 3.35248150e-02 -9.43942741e-02 6.22328341e-01 -6.51658401e-02 -4.72249389e-01 -6.92243502e-02 -2.20302060e-01 2.30850056e-01 4.65086818e-01 5.17104566e-01 -9.18273151e-01 1.45386803e+00 5.94145966e+00 8.17287385e-01 -6.53486490e-01 -3.14414531e-01 6.83097661e-01 1.72939718e-01 -8.32429349e-01 -2.28820249e-01 -1.34792483e+00 5.17708361e-01 9.82046604e-01 -7.09686458e-01 2.21317530e-01 8.49755764e-01 -4.43990976e-01 7.37616792e-02 -9.12939727e-01 5.54852366e-01 3.03086620e-02 -1.69534934e+00 7.56643355e-01 -3.74585129e-02 5.56865990e-01 -2.21559897e-01 -2.45583519e-01 7.77978897e-01 5.48483551e-01 -9.45302963e-01 8.52857530e-01 5.95805943e-01 1.03921807e+00 -8.19050252e-01 1.06409109e+00 1.49318025e-01 -1.31902874e+00 1.15115881e-01 -6.10515058e-01 3.25637251e-01 2.97627568e-01 5.18696249e-01 -1.38207519e+00 9.58609521e-01 6.78178251e-01 4.69876081e-01 -5.88517308e-01 7.79877841e-01 -4.87751931e-01 3.87926310e-01 -1.53057778e-03 -2.60856956e-01 1.14078671e-01 2.72251278e-01 2.79924065e-01 1.45588398e+00 3.10971290e-01 1.24578707e-01 -3.21550936e-01 1.15394497e+00 -5.09781361e-01 1.95407376e-01 -4.78370458e-01 -8.29155520e-02 1.05644286e+00 1.15472066e+00 -5.36651075e-01 -8.09075832e-01 -3.31864268e-01 9.41271782e-01 8.28596175e-01 2.40092754e-01 -8.98726463e-01 -5.57615459e-01 6.34225667e-01 -5.34854867e-02 6.61245763e-01 3.30874056e-01 -7.73634315e-02 -1.20888984e+00 2.98555464e-01 -6.69918060e-01 5.85382581e-01 -1.16011524e+00 -1.09229529e+00 9.65371430e-01 7.25011453e-02 -6.96732342e-01 -7.36443520e-01 -2.00886771e-01 -1.06486879e-01 9.63031530e-01 -1.66678154e+00 -9.66267407e-01 -4.27099615e-01 4.06934023e-01 3.31877857e-01 -2.87815809e-01 1.13770986e+00 1.83332920e-01 -5.08377731e-01 8.81237030e-01 -2.11516261e-01 6.08442426e-01 5.95125139e-01 -1.60316336e+00 1.04152572e+00 5.44344306e-01 2.16061667e-01 8.88299763e-01 3.80762726e-01 -8.59731853e-01 -1.02589715e+00 -1.35790968e+00 1.33012593e+00 -5.32381475e-01 6.07974768e-01 -3.37767243e-01 -1.00197256e+00 9.59615886e-01 2.82068197e-02 -3.99594940e-02 8.41678679e-01 1.02544881e-01 -5.42077243e-01 -6.40660822e-02 -1.00991571e+00 6.01784289e-01 1.13351262e+00 -4.83216763e-01 -9.42800820e-01 5.45586534e-02 1.17294180e+00 -7.09737301e-01 -8.72780025e-01 4.60113794e-01 4.77637142e-01 -7.37786770e-01 1.03368890e+00 -8.93771827e-01 7.36275434e-01 -3.17096502e-01 -1.48109928e-01 -1.43088853e+00 -5.92240095e-01 -3.08317453e-01 -5.35399854e-01 1.45426130e+00 1.12590492e+00 -1.94521010e-01 9.36400115e-01 6.70320451e-01 -2.25631401e-01 -8.59589398e-01 -7.61628687e-01 -7.21667588e-01 -9.61336270e-02 -1.20659180e-01 1.10674131e+00 9.08478320e-01 2.95261264e-01 8.12750041e-01 -1.89582646e-01 1.84902504e-01 1.43783286e-01 7.54156485e-02 7.68343627e-01 -8.96243393e-01 -3.34685683e-01 -2.03348830e-01 -2.71516472e-01 -9.69934404e-01 4.33289170e-01 -9.58060086e-01 1.13673829e-01 -1.83438432e+00 4.62191790e-01 -6.22749150e-01 -3.07055712e-01 7.87207067e-01 -4.73706394e-01 -1.38053698e-02 -1.07923843e-01 1.88307911e-02 -8.21099102e-01 5.70306838e-01 8.30244243e-01 -2.08818555e-01 -2.27561474e-01 -5.33379078e-01 -1.19412816e+00 2.32987478e-01 6.39850378e-01 -3.56788546e-01 -4.25611258e-01 -3.53924274e-01 5.89641750e-01 3.96723598e-01 6.85434937e-02 -8.22877586e-01 9.98176262e-02 -1.85955361e-01 2.32068226e-01 -4.68545347e-01 1.83048725e-01 -3.58395338e-01 1.18369408e-01 1.81374587e-02 -6.03980541e-01 1.45919308e-01 3.66707116e-01 5.44761479e-01 -3.43393177e-01 -1.11015260e-01 1.42963767e-01 -2.90214568e-01 -1.00054884e+00 1.82302356e-01 -5.72106838e-02 4.51434642e-01 6.95192635e-01 1.77157536e-01 -7.78660357e-01 -4.69563276e-01 -5.27361691e-01 2.80254871e-01 3.75996113e-01 3.60635757e-01 4.92510706e-01 -1.56789529e+00 -5.89124143e-01 1.26909092e-01 6.95214450e-01 4.11124349e-01 6.61183745e-02 5.80689795e-02 -4.27596807e-01 8.22937012e-01 -2.54676074e-01 1.01416647e-01 -7.97415018e-01 5.72200477e-01 1.22729450e-01 -8.70322526e-01 -3.88782352e-01 8.03259313e-01 -9.91288666e-03 -5.52663624e-01 1.25943810e-01 -5.11149406e-01 -1.67644337e-01 -1.63869545e-01 7.37558901e-01 8.35111141e-02 3.19886953e-01 -4.48154479e-01 -3.74355435e-01 -1.29202634e-01 -3.48615259e-01 -2.57791191e-01 1.12590170e+00 2.77745072e-02 -7.65412822e-02 2.99889654e-01 7.57286549e-01 1.80109605e-01 -7.93585598e-01 -5.27983665e-01 3.48924726e-01 -8.51212740e-02 -3.40534091e-01 -1.27189958e+00 -7.64685810e-01 2.52628058e-01 -2.36163095e-01 1.76867381e-01 7.64892340e-01 2.80141234e-01 9.49456215e-01 8.25638115e-01 6.95460856e-01 -9.34638500e-01 -5.40278964e-02 8.01669598e-01 9.98777211e-01 -1.00051749e+00 -2.36855909e-01 -5.44217646e-01 -7.27982342e-01 7.08332717e-01 9.36623096e-01 3.24070752e-01 2.17343688e-01 2.66393542e-01 -1.10107124e-01 -6.56948015e-02 -1.10962093e+00 -3.55056554e-01 4.31768209e-01 7.67305195e-01 4.75631744e-01 6.24013357e-02 -1.97477043e-01 1.23030150e+00 -5.97164929e-01 8.56283531e-02 5.19459903e-01 7.30800450e-01 -4.20273751e-01 -9.99396801e-01 1.40631333e-01 6.24517202e-01 -2.59270519e-01 -5.15958309e-01 -5.67004681e-01 9.34364736e-01 -9.38758850e-02 3.71695757e-01 2.48289749e-01 -4.52680618e-01 3.73391360e-01 4.12277699e-01 1.05833776e-01 -9.24844563e-01 -2.86237925e-01 -5.51250994e-01 6.17783010e-01 -5.02480686e-01 -5.54510243e-02 -3.05935353e-01 -1.40161431e+00 -1.90595761e-01 -2.04581335e-01 5.27870357e-01 1.76804841e-01 7.11961210e-01 7.03813255e-01 4.58199769e-01 4.37572896e-02 -3.15854907e-01 -2.40768805e-01 -1.02104294e+00 -6.14133358e-01 4.63534683e-01 -4.17421311e-02 -6.58924460e-01 1.58047631e-01 9.24888626e-02]
[9.722877502441406, 8.58381462097168]
dc06250e-051d-46dc-8b33-0517043655a2
convolutional-attention-networks-for
1805.06606
null
http://arxiv.org/abs/1805.06606v2
http://arxiv.org/pdf/1805.06606v2.pdf
Convolutional Attention Networks for Multimodal Emotion Recognition from Speech and Text Data
Emotion recognition has become a popular topic of interest, especially in the field of human computer interaction. Previous works involve unimodal analysis of emotion, while recent efforts focus on multi-modal emotion recognition from vision and speech. In this paper, we propose a new method of learning about the hidden representations between just speech and text data using convolutional attention networks. Compared to the shallow model which employs simple concatenation of feature vectors, the proposed attention model performs much better in classifying emotion from speech and text data contained in the CMU-MOSEI dataset.
['Ji-Hoon Jeong', 'Chan Woo Lee', 'Woo Yong Choi', 'Kyu Ye Song']
2018-05-17
convolutional-attention-networks-for-1
https://aclanthology.org/W18-3304
https://aclanthology.org/W18-3304.pdf
ws-2018-7
['multimodal-emotion-recognition', 'multimodal-emotion-recognition']
['computer-vision', 'speech']
[-4.85744961e-02 -9.67679769e-02 1.27284825e-01 -5.86413205e-01 -6.12813115e-01 -2.10998327e-01 5.91934860e-01 1.75404012e-01 -5.91298163e-01 3.98867369e-01 4.48057264e-01 -3.29443328e-02 3.73470336e-01 -2.22090170e-01 -2.00780302e-01 -5.70731580e-01 1.22501716e-01 1.32692203e-01 -5.85963249e-01 -1.34120017e-01 1.89655930e-01 2.57620305e-01 -1.54881775e+00 4.47221816e-01 5.15401840e-01 1.49111271e+00 -7.56026506e-02 7.98719287e-01 -4.86871809e-01 1.05990934e+00 -6.75144553e-01 -4.35215592e-01 -4.08661276e-01 -5.81779063e-01 -9.92301106e-01 2.18145192e-01 9.28711817e-02 2.48254184e-02 -3.22142601e-01 9.92772639e-01 7.50587702e-01 5.34675062e-01 7.41630018e-01 -1.36171508e+00 -1.00793707e+00 3.36325049e-01 -3.23470324e-01 3.30214620e-01 5.42147219e-01 -3.89222950e-01 1.01678216e+00 -1.19912660e+00 2.27753788e-01 1.27733874e+00 4.46925461e-01 6.32897913e-01 -7.02263355e-01 -3.71039778e-01 1.55850142e-01 6.48092210e-01 -1.26334727e+00 -5.91336727e-01 9.54662979e-01 -2.73017466e-01 1.29734457e+00 2.28214249e-01 3.43427598e-01 1.38753092e+00 1.21463388e-01 1.32595086e+00 8.01686585e-01 -5.87840319e-01 2.15412945e-01 2.80925989e-01 3.78428966e-01 3.64647955e-01 -4.54617202e-01 -3.84689122e-01 -5.95667899e-01 -1.80723801e-01 2.29907259e-01 1.77281559e-01 -4.25865024e-01 -9.91208872e-05 -1.09062743e+00 9.11818206e-01 1.89791769e-01 6.74957573e-01 -7.61088073e-01 -3.48181389e-02 7.81665146e-01 4.24360633e-01 7.47068226e-01 2.35175088e-01 -4.55511212e-01 -6.31460071e-01 -5.84045529e-01 -3.67850989e-01 7.49982834e-01 6.78592980e-01 4.31319565e-01 4.51337695e-01 6.78992074e-04 9.66110170e-01 2.74887651e-01 4.10999656e-01 8.48853648e-01 -6.10255957e-01 2.10086927e-01 3.90135944e-01 -1.51769370e-01 -1.15021145e+00 -4.43611979e-01 5.44442609e-02 -9.40125883e-01 -8.16230923e-02 -1.00540146e-01 -4.38814729e-01 -7.96438158e-01 1.44704306e+00 -1.12075545e-01 -1.57569155e-01 6.86446905e-01 8.90536785e-01 1.31297708e+00 9.12021935e-01 1.93352789e-01 -2.56298900e-01 1.20750022e+00 -1.04195857e+00 -1.40799975e+00 -4.42240238e-01 5.98794341e-01 -6.12770736e-01 6.53981745e-01 4.70538765e-01 -8.88859749e-01 -4.64550823e-01 -6.31765544e-01 -1.70963258e-01 -8.46670091e-01 3.77526939e-01 7.08875477e-01 4.01130527e-01 -1.00032258e+00 1.46936238e-01 -4.48329896e-01 -7.49536872e-01 3.20294559e-01 1.60064906e-01 -6.25495970e-01 1.13118380e-01 -1.36549687e+00 1.09498644e+00 2.72450864e-01 4.13808763e-01 -4.79357719e-01 3.34549360e-02 -1.31492913e+00 3.65407318e-01 1.46935806e-01 -1.48803800e-01 1.27660310e+00 -1.58555603e+00 -1.87983751e+00 6.89307630e-01 -5.04090250e-01 -2.36705258e-01 -2.94698775e-01 -4.17946488e-01 -7.71646440e-01 3.13895822e-01 -4.08856481e-01 6.92402542e-01 1.03330564e+00 -8.81514847e-01 -2.89938092e-01 -5.52352667e-01 -1.73135877e-01 3.05431217e-01 -6.29845142e-01 3.77459109e-01 -2.92465448e-01 -5.31297624e-01 8.98576528e-02 -6.15748227e-01 -1.58256114e-01 -4.46825713e-01 -2.35451266e-01 -6.41416371e-01 1.14268112e+00 -5.63984632e-01 9.55569327e-01 -2.23373580e+00 4.08744872e-01 -2.42014796e-01 -3.24101262e-02 1.80068985e-01 -1.07730530e-01 4.85588938e-01 -3.58138204e-01 4.51178700e-02 3.94394100e-02 -6.78263247e-01 8.02712142e-02 2.46177807e-01 -3.56289297e-01 3.78483683e-01 4.20170009e-01 9.58433449e-01 -5.29752731e-01 -3.99246156e-01 3.38554382e-01 7.37737536e-01 -1.96276098e-01 3.15655202e-01 2.62337327e-01 3.51541042e-01 -3.60239565e-01 7.71159053e-01 2.08350509e-01 -1.20629564e-01 -5.19695804e-02 -2.06049755e-01 -4.81570996e-02 3.14838663e-02 -6.03974521e-01 1.81428957e+00 -5.44903517e-01 1.09602630e+00 2.31946617e-01 -1.48781002e+00 1.00431693e+00 1.03079283e+00 3.09222013e-01 -6.46535635e-01 7.54866838e-01 -3.78466755e-01 -8.24969560e-02 -8.97979021e-01 6.03641331e-01 -4.92239565e-01 -4.08797473e-01 3.47307622e-01 4.67844486e-01 -8.14818144e-02 -1.17571734e-01 3.04014608e-02 7.13163614e-01 -4.30798352e-01 2.92811930e-01 1.23797052e-01 7.15288758e-01 -2.98853487e-01 3.09016496e-01 3.54135603e-01 -6.52784228e-01 4.89125133e-01 4.25864607e-01 -4.37556803e-01 -4.21842039e-01 -5.19294977e-01 1.08454369e-01 1.29218626e+00 -6.82661012e-02 -4.24830735e-01 -4.63215262e-01 -6.95140362e-01 -3.72753620e-01 5.98290145e-01 -8.64117682e-01 -2.91195273e-01 1.10846587e-01 -3.79598647e-01 5.80857158e-01 7.19849944e-01 4.66830641e-01 -1.53966022e+00 -5.17297566e-01 -1.02572069e-01 -4.83708084e-01 -1.29588974e+00 -2.47626469e-01 3.50433081e-01 -5.06633997e-01 -6.53177500e-01 -9.41827834e-01 -9.58437324e-01 3.44827712e-01 2.76157632e-02 9.15790319e-01 -3.32718700e-01 -3.21327388e-01 8.81269813e-01 -7.32974112e-01 -6.86634302e-01 3.99628878e-02 -1.42782778e-01 2.35060509e-02 5.60326338e-01 6.96689367e-01 -1.10683195e-01 -9.14666429e-02 -9.88544077e-02 -8.60309303e-01 -2.96722263e-01 4.92368609e-01 8.46159935e-01 1.72630936e-01 -7.89542422e-02 7.51653135e-01 -3.64072174e-01 8.42357695e-01 -6.59569263e-01 2.46537179e-01 1.59790635e-01 4.42583114e-02 -5.59782833e-02 5.65886080e-01 -5.74408412e-01 -1.19386756e+00 1.21326804e-01 -4.19032305e-01 -7.21925855e-01 -7.83490062e-01 8.35837245e-01 -2.71786571e-01 8.85787308e-02 5.73556982e-02 1.69139892e-01 -1.01140793e-02 -3.35950315e-01 3.68384391e-01 1.16127837e+00 4.26724970e-01 -1.85940772e-01 -2.65015334e-01 3.50150377e-01 -6.31733119e-01 -1.30982828e+00 -5.14561296e-01 -6.22535944e-01 -3.54216844e-01 -3.17290574e-01 1.27817082e+00 -6.59774065e-01 -8.09957683e-01 5.77634454e-01 -1.42526662e+00 1.61085606e-01 2.24067941e-02 8.71858358e-01 -5.78283906e-01 3.69560719e-01 -7.85219371e-01 -1.20378268e+00 -3.40297461e-01 -1.12386513e+00 1.17036188e+00 3.00518870e-01 -5.09390712e-01 -1.17226756e+00 2.41624583e-02 1.20414443e-01 5.02443492e-01 -1.31696835e-01 6.51273191e-01 -9.70238626e-01 3.89697015e-01 -6.32354736e-01 -1.14132978e-01 4.59711045e-01 2.11354807e-01 -1.78333625e-01 -1.26793301e+00 -2.10654587e-02 3.28186184e-01 -8.67154717e-01 9.38467324e-01 4.70413178e-01 1.24069130e+00 -1.18924966e-02 -5.14413938e-02 2.01065302e-01 9.38797355e-01 5.75430095e-01 5.93286455e-01 -3.98276001e-02 5.44701159e-01 9.06215549e-01 3.70107681e-01 5.52695513e-01 4.83849823e-01 3.63494188e-01 4.71101314e-01 -1.77660272e-01 5.93676209e-01 4.45842147e-01 4.48814213e-01 1.01686692e+00 1.08354203e-01 -4.28876370e-01 -8.97758424e-01 6.75484359e-01 -1.89745629e+00 -1.12180436e+00 2.21083418e-01 1.53134739e+00 3.34871560e-01 -3.00790042e-01 -1.77401811e-01 2.00953871e-01 7.02992022e-01 3.96896213e-01 -4.00973737e-01 -1.08709741e+00 -1.92516834e-01 7.66488165e-02 -2.55652696e-01 2.96725661e-01 -1.29878473e+00 9.43181336e-01 7.29620266e+00 4.14119065e-01 -1.40800631e+00 8.17035362e-02 5.96522391e-01 -1.40611172e-01 9.04102623e-02 -5.62127829e-01 -2.91866422e-01 1.39026642e-01 1.30527663e+00 -4.15013023e-02 8.74405429e-02 9.82547581e-01 -3.45182121e-02 -1.23126872e-01 -1.02845740e+00 1.47113550e+00 7.13663578e-01 -8.38665664e-01 -5.33582531e-02 -2.73163676e-01 1.60472244e-01 -1.33245796e-01 1.32385164e-01 6.77820623e-01 -8.37541819e-02 -1.13176703e+00 4.42518204e-01 5.36156058e-01 3.38133514e-01 -9.58703697e-01 1.03453040e+00 3.35950792e-01 -1.00126874e+00 -1.06682062e-01 -1.40612468e-01 -3.89104784e-01 1.98517621e-01 2.58505732e-01 -3.47360253e-01 4.03477728e-01 1.02323294e+00 8.98085833e-01 -1.53633416e-01 4.16441202e-01 7.61450529e-02 4.64852542e-01 7.57734999e-02 -2.13769406e-01 2.23756462e-01 -2.44255248e-03 2.37825498e-01 1.55640292e+00 1.92939341e-01 3.20153654e-01 1.88590169e-01 4.86012280e-01 -1.16380133e-01 2.29289904e-01 -9.27336931e-01 -7.03734517e-01 -1.49760112e-01 1.46801889e+00 -4.88768131e-01 -6.38584971e-01 -7.58138239e-01 1.39865923e+00 4.63030845e-01 5.85660338e-01 -7.67784059e-01 -6.56258166e-01 7.70941377e-01 -9.90917504e-01 4.67355579e-01 -2.71656156e-01 -5.09362668e-02 -1.20165825e+00 -2.38914624e-01 -6.69605136e-01 3.40347767e-01 -1.04116917e+00 -1.34121120e+00 1.10976982e+00 -4.02314276e-01 -9.46667790e-01 -3.75838250e-01 -7.98355639e-01 -8.99605691e-01 8.28106046e-01 -1.17068005e+00 -7.63568282e-01 -1.18898474e-01 7.46818483e-01 8.86487067e-01 -1.04999714e-01 1.20100749e+00 2.16480434e-01 -7.63831258e-01 4.64067698e-01 -1.80926338e-01 2.83094257e-01 7.26557434e-01 -1.04256701e+00 -8.65168646e-02 4.57421184e-01 2.48685136e-01 4.38688308e-01 6.28679276e-01 -2.83802480e-01 -1.45291591e+00 -6.55207932e-01 1.15774345e+00 -2.13048160e-01 5.25168955e-01 -2.25702181e-01 -8.68545294e-01 6.99875712e-01 9.59232450e-01 -1.00507095e-01 1.04916954e+00 2.16276601e-01 -2.91575789e-01 2.91220754e-01 -9.40928280e-01 4.60750431e-01 2.73434997e-01 -9.60154951e-01 -7.70614922e-01 -9.63298082e-02 6.24165714e-01 -3.91161330e-02 -7.23848104e-01 3.86901855e-01 4.56926405e-01 -6.19783044e-01 6.92062914e-01 -8.82876337e-01 3.56463104e-01 1.76907256e-01 -5.06016135e-01 -1.57208467e+00 -1.36251822e-01 -2.52398431e-01 -1.03641108e-01 1.12975764e+00 2.56572932e-01 -3.52461427e-01 4.00137395e-01 6.94455981e-01 -2.29396641e-01 -5.88452637e-01 -8.53281260e-01 -2.73230255e-01 -1.55372441e-01 -6.72681630e-01 2.03819182e-02 1.19043672e+00 6.16610527e-01 7.61584342e-01 -5.61753809e-01 -2.28292737e-02 -1.02347573e-02 4.55325060e-02 3.56828123e-01 -1.11295867e+00 1.02299683e-01 -5.40964544e-01 -6.24343574e-01 -7.79328048e-01 7.65983224e-01 -7.04211652e-01 3.99304777e-01 -1.55365729e+00 3.16153243e-02 5.73127329e-01 -5.42731345e-01 5.05801558e-01 4.51540016e-02 2.66018271e-01 9.23647583e-02 -2.79109985e-01 -9.89809215e-01 1.16076040e+00 8.11323822e-01 -3.08591515e-01 3.85392755e-02 -3.45003188e-01 -6.18781090e-01 9.62051690e-01 9.06448364e-01 -7.18101189e-02 -8.23940337e-02 -3.67515296e-01 -1.12048397e-02 2.67500967e-01 1.66953072e-01 -7.46393561e-01 3.66415590e-01 7.90765211e-02 5.28994977e-01 -7.74721801e-01 8.09977829e-01 -9.34793472e-01 -4.16454077e-01 -2.24781428e-02 -5.16473353e-01 2.57299423e-01 4.84807998e-01 4.90164787e-01 -8.32662165e-01 -1.51368201e-01 3.63526642e-01 -7.88383931e-02 -1.05919230e+00 2.50040144e-01 -1.20323813e+00 -2.29971632e-01 8.82076442e-01 4.87306826e-02 -1.56213120e-01 -8.89577687e-01 -1.24521828e+00 2.23900884e-01 -1.47309676e-01 7.99418151e-01 1.00871050e+00 -1.42281401e+00 -3.50435287e-01 2.04265654e-01 2.45778963e-01 -5.64385355e-01 4.42539960e-01 8.52984786e-01 2.76997179e-01 6.46685481e-01 -2.20339432e-01 -4.35367793e-01 -1.43813777e+00 6.07809782e-01 2.42757991e-01 1.49903715e-01 -2.42911711e-01 7.68180192e-01 2.37830773e-01 -3.77536595e-01 6.29620612e-01 -1.50178522e-01 -6.43924236e-01 3.23826909e-01 7.14950979e-01 5.25888056e-02 -3.34338956e-02 -1.05000603e+00 -4.93003279e-01 3.59437793e-01 -1.08998150e-01 -2.94671029e-01 1.24916053e+00 -4.22327369e-01 -1.78800181e-01 1.13469791e+00 1.46454155e+00 -4.42344457e-01 -5.74058115e-01 -2.66644567e-01 1.69771418e-01 1.71234552e-02 2.45988026e-01 -6.59721315e-01 -1.02719557e+00 1.38251376e+00 6.61987603e-01 4.24540132e-01 1.22485340e+00 3.46778482e-01 5.97673237e-01 7.41044998e-01 -1.65278196e-01 -1.29326379e+00 2.70721227e-01 9.54203725e-01 1.13221633e+00 -1.70142472e+00 -5.96213222e-01 1.28986761e-01 -1.21211088e+00 1.12046063e+00 6.53222203e-01 1.85815722e-01 8.46960902e-01 8.66705552e-02 3.70916128e-01 -4.52921689e-01 -8.83069038e-01 -6.62678838e-01 3.66735905e-01 5.54476738e-01 7.83332348e-01 -1.59115568e-01 8.82765725e-02 9.85639572e-01 2.26761058e-01 -1.21688373e-01 3.09721082e-01 1.01102078e+00 -4.15212154e-01 -7.31620669e-01 -3.43186975e-01 1.73952326e-01 -6.99777305e-01 -2.72297412e-01 -8.11748266e-01 3.62883329e-01 -3.43155235e-01 1.29280186e+00 4.20495421e-01 -4.32093829e-01 2.47525290e-01 6.84948206e-01 2.25932553e-01 -4.64939803e-01 -4.67426062e-01 1.20573461e-01 1.58632576e-01 -5.45501649e-01 -7.97626436e-01 -4.23397183e-01 -1.25238931e+00 1.29936159e-01 -2.32588783e-01 3.62158328e-01 8.56161118e-01 1.18399072e+00 5.69720864e-01 5.92270970e-01 6.01328790e-01 -1.20034480e+00 3.41970997e-04 -1.22277129e+00 -6.32286131e-01 4.52373952e-01 4.42554712e-01 -6.13374531e-01 -4.26430941e-01 6.18589558e-02]
[13.297905921936035, 5.377673625946045]
a4920f9b-ac0a-414b-8695-d889dc62521c
evaluation-of-audio-visual-alignments-in
2108.02562
null
https://arxiv.org/abs/2108.02562v1
https://arxiv.org/pdf/2108.02562v1.pdf
Evaluation of Audio-Visual Alignments in Visually Grounded Speech Models
Systems that can find correspondences between multiple modalities, such as between speech and images, have great potential to solve different recognition and data analysis tasks in an unsupervised manner. This work studies multimodal learning in the context of visually grounded speech (VGS) models, and focuses on their recently demonstrated capability to extract spatiotemporal alignments between spoken words and the corresponding visual objects without ever been explicitly trained for object localization or word recognition. As the main contributions, we formalize the alignment problem in terms of an audiovisual alignment tensor that is based on earlier VGS work, introduce systematic metrics for evaluating model performance in aligning visual objects and spoken words, and propose a new VGS model variant for the alignment task utilizing cross-modal attention layer. We test our model and a previously proposed model in the alignment task using SPEECH-COCO captions coupled with MSCOCO images. We compare the alignment performance using our proposed evaluation metrics to the semantic retrieval task commonly used to evaluate VGS models. We show that cross-modal attention layer not only helps the model to achieve higher semantic cross-modal retrieval performance, but also leads to substantial improvements in the alignment performance between image object and spoken words.
['Okko Räsänen', 'Khazar Khorrami']
2021-07-05
null
null
null
null
['semantic-retrieval']
['natural-language-processing']
[ 1.72869354e-01 -6.08773753e-02 -8.91976729e-02 -4.09504026e-01 -1.35268188e+00 -3.91081184e-01 9.97315466e-01 9.30426195e-02 -5.28811991e-01 1.00195043e-01 3.91660213e-01 2.88751721e-02 -5.75457141e-03 -1.36857539e-01 -7.39256561e-01 -5.57270527e-01 2.64459729e-01 6.18398070e-01 7.20261261e-02 -1.87901139e-01 3.00670773e-01 2.95654744e-01 -1.63391149e+00 5.71926951e-01 4.90263462e-01 1.08921123e+00 3.72578144e-01 7.66363144e-01 -4.44465280e-01 6.40155673e-01 -4.70977277e-01 -2.81835645e-01 -1.08812796e-02 -5.61054647e-01 -1.10754347e+00 2.57693082e-01 1.03359890e+00 1.23048134e-01 -3.10407341e-01 9.08035338e-01 5.15994310e-01 1.60769746e-01 9.29644883e-01 -1.64090049e+00 -1.02807581e+00 3.23135316e-01 -2.90768445e-01 2.17040768e-03 4.38120633e-01 -2.55116206e-02 1.15047920e+00 -1.24387598e+00 6.55456960e-01 1.41897857e+00 2.71214217e-01 5.30735075e-01 -1.15960777e+00 -1.32393464e-01 2.60368317e-01 6.12893105e-01 -1.59084249e+00 -7.85866439e-01 6.64346874e-01 -4.65631247e-01 9.66449618e-01 4.16384101e-01 5.22954404e-01 1.23645449e+00 -1.75982147e-01 1.13320494e+00 8.17678511e-01 -6.59632564e-01 9.85023826e-02 1.83911577e-01 1.50844529e-01 6.04483485e-01 -3.81453365e-01 -2.88094968e-01 -1.01542008e+00 2.05727085e-01 2.84445733e-01 -2.63255209e-01 -2.28371546e-01 -6.27565682e-01 -1.71548188e+00 7.59469807e-01 4.89287108e-01 6.04319513e-01 -4.06676024e-01 3.44216853e-01 2.82792240e-01 7.02187717e-02 3.05524260e-01 3.69654089e-01 9.15604383e-02 1.89480960e-01 -1.02028072e+00 -4.31052782e-03 3.82193893e-01 9.40730214e-01 5.34865439e-01 8.08562562e-02 -5.60795903e-01 9.25576210e-01 7.07341731e-01 9.46779013e-01 6.39306426e-01 -7.80632317e-01 6.44470274e-01 4.99166250e-01 2.55824849e-02 -1.17485321e+00 -1.51096851e-01 -3.48100662e-02 -5.01853049e-01 -1.69821456e-01 8.54797065e-02 5.86310506e-01 -1.01482451e+00 1.83080220e+00 1.02920823e-01 7.73052648e-02 5.25804520e-01 1.22485518e+00 1.21639180e+00 7.35633373e-01 3.09895784e-01 -1.25024006e-01 1.45727825e+00 -1.32986104e+00 -9.00701463e-01 -3.71290088e-01 5.03575027e-01 -8.71156335e-01 1.27801454e+00 -1.42118588e-01 -1.21212888e+00 -5.72443426e-01 -1.03189278e+00 -1.73398256e-01 -6.17533565e-01 1.57957748e-01 2.02026337e-01 7.88427889e-02 -1.45781589e+00 -1.02145009e-01 -7.09918261e-01 -9.04563248e-01 1.58401355e-01 9.04634222e-02 -5.91138363e-01 1.67429194e-01 -1.01399171e+00 1.13607454e+00 2.93667853e-01 1.62686005e-01 -1.07297409e+00 -1.36422500e-01 -1.14795899e+00 9.57012400e-02 2.67140239e-01 -8.10475707e-01 1.16189075e+00 -1.16659713e+00 -1.21242070e+00 1.15599740e+00 -5.49381614e-01 -3.13841611e-01 3.20900902e-02 4.29051742e-02 -5.48200905e-01 4.87489402e-01 1.66732654e-01 1.17684853e+00 9.83731866e-01 -1.54182684e+00 -3.28535378e-01 -4.54621166e-01 -1.94117099e-01 6.28587961e-01 -4.72482353e-01 -1.19282454e-02 -7.12947369e-01 -5.48571467e-01 2.39023581e-01 -8.43936801e-01 2.17453524e-01 -2.27345392e-01 -4.81531829e-01 -2.47646555e-01 7.32766092e-01 -6.09282553e-01 7.74334311e-01 -2.01971722e+00 6.77264452e-01 1.09978057e-01 -4.54728715e-02 2.84095436e-01 -7.80259013e-01 6.41116917e-01 4.71734926e-02 -1.13724200e-02 -2.38028347e-01 -8.42643142e-01 1.49427727e-01 3.05994213e-01 -4.09038991e-01 2.23058626e-01 2.66827464e-01 1.39747596e+00 -7.28123963e-01 -7.73142040e-01 2.28529513e-01 4.99813199e-01 -5.67484759e-02 4.71454710e-01 -1.91062495e-01 6.07609570e-01 -1.03999667e-01 6.80423796e-01 2.34194756e-01 -3.62550586e-01 -9.57562029e-03 -6.14759684e-01 1.35340601e-01 -6.50430247e-02 -8.10202360e-01 2.06718516e+00 -5.63801229e-01 8.01901698e-01 -1.10612787e-01 -1.40069294e+00 7.47226000e-01 5.42800009e-01 3.38492393e-01 -1.14400744e+00 -5.51429018e-02 -3.82730109e-03 -3.48558575e-01 -9.28183913e-01 5.61679780e-01 1.85114920e-01 -8.19562078e-02 2.67565042e-01 5.35140216e-01 -1.63039088e-01 4.73081879e-02 3.79476428e-01 3.51791799e-01 -1.86365936e-02 5.48310801e-02 6.57639131e-02 8.33168983e-01 4.31712046e-02 -3.06615889e-01 7.54848778e-01 -7.25756958e-02 7.93153763e-01 9.36614573e-02 -9.59896073e-02 -1.00452054e+00 -1.28950965e+00 2.11243451e-01 1.26390731e+00 4.82799977e-01 -3.58033329e-01 -6.89272225e-01 -4.98467952e-01 -2.08757445e-01 6.63784087e-01 -5.93249917e-01 -1.43120617e-01 -1.80308595e-01 -3.35727036e-01 5.68933606e-01 5.78078806e-01 3.68937850e-01 -1.20427895e+00 -1.62258655e-01 -3.37386817e-01 -6.80240631e-01 -1.59128106e+00 -6.59057200e-01 -3.92930478e-01 -4.18144971e-01 -8.44549060e-01 -9.77684140e-01 -1.13731039e+00 5.84064364e-01 4.39891219e-01 1.07757187e+00 -9.87765193e-02 -2.17802241e-01 1.39112151e+00 -4.54496026e-01 -3.45405936e-01 -3.64626765e-01 7.10602105e-02 -2.73056515e-02 4.70697582e-01 2.48470902e-01 -8.50026906e-02 -5.55481255e-01 4.25634235e-01 -1.03182507e+00 2.49182180e-01 5.18693626e-01 8.26312006e-01 6.38066828e-01 -1.11299455e+00 5.49760282e-01 -1.09291952e-02 3.92794549e-01 -3.47452462e-01 -4.10656810e-01 7.83452451e-01 -4.15169895e-01 1.83764637e-01 -1.05368324e-01 -4.60716218e-01 -6.10088587e-01 1.15822680e-01 3.79857165e-03 -6.71467304e-01 -2.55652934e-01 6.51280761e-01 -1.47129133e-01 -2.01103598e-01 2.76392967e-01 4.49364513e-01 2.89467305e-01 -2.72508919e-01 7.77245343e-01 8.89412224e-01 7.79207230e-01 -3.78299326e-01 5.21423817e-01 4.76067364e-01 -1.27890587e-01 -9.94811773e-01 -7.30526030e-01 -7.29779601e-01 -7.93713808e-01 -3.97764951e-01 1.42208827e+00 -8.88467610e-01 -8.19108069e-01 2.77409434e-01 -1.50287235e+00 -1.22064225e-01 1.83579959e-02 5.97575605e-01 -8.37732852e-01 4.92649227e-01 -1.50146589e-01 -8.73149872e-01 -2.50708848e-01 -1.25924969e+00 1.69562232e+00 4.27580904e-03 -1.09966911e-01 -1.06821096e+00 1.51051268e-01 8.24281633e-01 4.63235199e-01 -6.97046071e-02 7.21703827e-01 -8.97646725e-01 -7.09937930e-01 -1.58391353e-02 -2.84179181e-01 2.19778448e-01 -1.01831831e-01 -2.70948857e-01 -1.05137968e+00 -2.47490585e-01 -2.72753865e-01 -4.38093930e-01 8.46630812e-01 1.79820970e-01 6.78822994e-01 -2.79400259e-01 -1.83327958e-01 2.95855254e-01 1.23199832e+00 2.33519852e-01 5.91049969e-01 2.55788088e-01 9.99239326e-01 8.76850128e-01 6.30580962e-01 -2.22203746e-01 6.19312823e-01 1.25840986e+00 6.43586516e-01 -3.31989139e-01 -3.81428540e-01 -1.00793488e-01 4.26032901e-01 1.20638835e+00 2.13092804e-01 -4.50697422e-01 -1.04459691e+00 9.08688188e-01 -2.20914245e+00 -8.91441286e-01 1.11711234e-01 2.11455798e+00 4.45111156e-01 -6.75979137e-01 1.73891529e-01 -3.23577583e-01 6.75488949e-01 1.72887459e-01 -1.20640740e-01 -2.56971389e-01 -3.55934292e-01 -9.16445814e-03 2.87845060e-02 7.94480860e-01 -1.13708651e+00 1.14083815e+00 6.37575817e+00 6.00412071e-01 -1.28074169e+00 3.18726033e-01 2.57746160e-01 -5.34513034e-02 -2.91147232e-01 -2.35455185e-01 -4.92770165e-01 1.50440916e-01 9.48182344e-01 1.05505876e-01 3.47798139e-01 4.80706841e-01 1.25229076e-01 1.02971092e-01 -1.42847633e+00 1.44063866e+00 7.70924270e-01 -1.30893898e+00 6.63522303e-01 -1.93259671e-01 5.37069559e-01 -8.11795816e-02 3.41439426e-01 -8.34714994e-03 -2.93118328e-01 -1.36122108e+00 1.05963600e+00 7.52535462e-01 4.50846761e-01 -2.90865183e-01 7.91501045e-01 -9.25362185e-02 -1.17543995e+00 2.31261969e-01 2.56143324e-02 5.17405510e-01 4.14573818e-01 -4.07566905e-01 -1.02237833e+00 7.21813500e-01 6.12934828e-01 6.60497785e-01 -7.72671461e-01 8.18432212e-01 1.50051981e-01 2.17624843e-01 5.28087802e-02 -6.79138070e-03 5.00211000e-01 -6.36011362e-02 7.12052107e-01 1.26842630e+00 3.08418751e-01 -3.73494208e-01 1.14485487e-01 8.23886812e-01 1.36036232e-01 5.32660246e-01 -7.90712297e-01 -3.95784616e-01 3.08718622e-01 1.02026474e+00 -4.17796344e-01 -4.38010186e-01 -5.47279537e-01 1.14478552e+00 2.39913449e-01 7.96955764e-01 -8.06608617e-01 -8.65889564e-02 5.52368820e-01 -1.43424600e-01 3.27430248e-01 -3.79791588e-01 6.93143457e-02 -1.11999977e+00 9.90454331e-02 -7.17190623e-01 2.51587749e-01 -1.36914647e+00 -1.17894018e+00 7.80529976e-01 1.70548856e-01 -1.24420297e+00 -4.07572299e-01 -4.79795545e-01 -1.90163270e-01 8.91000450e-01 -1.35323334e+00 -1.69630659e+00 -4.83533323e-01 8.96715403e-01 5.50558090e-01 -4.83109683e-01 7.99009919e-01 2.89009035e-01 -2.57586211e-01 6.73959434e-01 -1.44652754e-01 1.52170897e-01 8.46684217e-01 -8.91885459e-01 2.03037098e-01 6.43898487e-01 9.38987195e-01 5.28241038e-01 5.94612956e-01 -2.67718315e-01 -1.48601890e+00 -9.10224438e-01 9.63913679e-01 -5.95422506e-01 5.98608136e-01 -3.83439034e-01 -9.76387858e-01 6.42281532e-01 6.53528333e-01 -8.26452672e-02 5.94723821e-01 -1.68422669e-01 -5.93698561e-01 -5.57684600e-02 -6.88828707e-01 6.48363709e-01 9.42021489e-01 -1.02169132e+00 -7.03238547e-01 4.03397828e-01 7.80485094e-01 -3.02104324e-01 -7.48245060e-01 3.56851041e-01 4.10143763e-01 -5.46108961e-01 1.30989909e+00 -1.10407114e+00 2.68169284e-01 -4.28133756e-01 -7.15634584e-01 -1.09176958e+00 1.74850494e-01 -1.77139953e-01 1.03417583e-01 1.30920649e+00 4.64149922e-01 -3.76681834e-01 2.75422126e-01 1.89318672e-01 -1.97712898e-01 -3.15364808e-01 -1.19840693e+00 -7.10557222e-01 -2.76416272e-01 -4.42578316e-01 2.02587679e-01 1.00731933e+00 -3.41305695e-02 7.19785631e-01 -4.77996171e-01 5.22711396e-01 4.91812855e-01 2.16104478e-01 7.40876436e-01 -8.45988929e-01 -3.03074857e-03 -6.22665405e-01 -8.14872861e-01 -8.22241485e-01 5.14967322e-01 -1.20531774e+00 2.58756280e-01 -1.75605023e+00 2.92744875e-01 1.26966864e-01 -2.53646076e-01 6.42926693e-01 3.15266140e-02 6.38673306e-01 5.06073952e-01 3.64705384e-01 -1.09033203e+00 8.50697041e-01 9.86630261e-01 -6.00938261e-01 4.81834970e-02 -5.81666350e-01 -3.76175195e-01 3.46853763e-01 3.33108783e-01 -1.03767723e-01 -4.16391909e-01 -6.64055109e-01 3.18638161e-02 1.35995299e-01 8.72207820e-01 -7.90699065e-01 4.02630121e-01 -7.39325359e-02 -7.20225647e-02 -4.84481454e-01 6.61109507e-01 -8.12394321e-01 2.66023143e-03 7.72628635e-02 -6.81421697e-01 3.04441243e-01 2.89237112e-01 5.73433518e-01 -7.72051275e-01 -3.91699821e-02 5.90991378e-01 1.94842324e-01 -1.01817811e+00 9.43216011e-02 -2.87239462e-01 -1.08262487e-01 9.73305225e-01 -1.25129774e-01 -3.80160004e-01 -7.20980525e-01 -1.24143934e+00 3.80603343e-01 2.70471156e-01 9.53321218e-01 8.48488927e-01 -1.72497451e+00 -6.67722642e-01 -1.37244672e-01 7.41516113e-01 -4.77547199e-01 3.29648137e-01 1.20019579e+00 -2.47707263e-01 8.91839206e-01 -1.57601416e-01 -1.28211987e+00 -1.54896498e+00 7.23550022e-01 3.69415015e-01 3.40981513e-01 -2.46671602e-01 5.92698812e-01 3.19822282e-01 -3.25697333e-01 5.21793902e-01 -5.91279417e-02 -3.81422818e-01 1.83898151e-01 3.88145477e-01 -5.70124649e-02 1.55665129e-01 -1.36722207e+00 -6.40405059e-01 9.59540129e-01 1.53347492e-01 -5.35776854e-01 9.34262514e-01 -5.54272771e-01 -5.57452179e-02 7.67249823e-01 1.47646177e+00 -3.02973568e-01 -6.51436150e-01 -5.19692957e-01 7.92138278e-02 -1.93372875e-01 -1.28877580e-01 -6.87231779e-01 -1.06408370e+00 1.08310044e+00 9.84363675e-01 1.24017335e-01 8.97467077e-01 6.83767498e-01 5.74645340e-01 3.09978545e-01 -4.75364849e-02 -8.50279212e-01 7.49058604e-01 4.13278967e-01 1.20848560e+00 -1.49912000e+00 -5.48848033e-01 -1.50919810e-01 -1.13181329e+00 8.60037267e-01 4.12044227e-01 2.90347993e-01 1.47470847e-01 -3.22398335e-01 4.65912700e-01 -3.95767808e-01 -5.49148858e-01 -6.43126011e-01 9.78272438e-01 6.79538012e-01 2.47254670e-01 -2.98158675e-01 4.59560752e-02 1.75262421e-01 2.30301812e-01 -5.57319701e-01 -3.93196084e-02 5.76638103e-01 -2.06529006e-01 -8.96804929e-01 -4.78900790e-01 -1.94493815e-01 -7.75554031e-03 -2.97054023e-01 -7.00296283e-01 7.28565216e-01 -2.88015485e-01 8.90513837e-01 4.66719270e-01 -3.27701092e-01 3.80791903e-01 5.05663812e-01 5.68708420e-01 -3.31841737e-01 -2.22954050e-01 6.50698096e-02 -7.05375522e-02 -7.31269479e-01 -9.71776962e-01 -5.59959352e-01 -9.41423118e-01 2.82778502e-01 1.87397171e-02 1.98117793e-01 1.01813245e+00 1.20130873e+00 5.66907763e-01 3.32833529e-01 3.34058940e-01 -1.02199650e+00 -7.90738314e-02 -8.72991562e-01 -1.09508209e-01 7.05450237e-01 4.88540918e-01 -7.47916996e-01 -2.55222708e-01 2.31377885e-01]
[10.922431945800781, 1.4856189489364624]
d7a9670b-53b8-4e02-bb90-20b533ec3075
internvideo-general-video-foundation-models
2212.03191
null
https://arxiv.org/abs/2212.03191v2
https://arxiv.org/pdf/2212.03191v2.pdf
InternVideo: General Video Foundation Models via Generative and Discriminative Learning
The foundation models have recently shown excellent performance on a variety of downstream tasks in computer vision. However, most existing vision foundation models simply focus on image-level pretraining and adpation, which are limited for dynamic and complex video-level understanding tasks. To fill the gap, we present general video foundation models, InternVideo, by taking advantage of both generative and discriminative self-supervised video learning. Specifically, InternVideo efficiently explores masked video modeling and video-language contrastive learning as the pretraining objectives, and selectively coordinates video representations of these two complementary frameworks in a learnable manner to boost various video applications. Without bells and whistles, InternVideo achieves state-of-the-art performance on 39 video datasets from extensive tasks including video action recognition/detection, video-language alignment, and open-world video applications. Especially, our methods can obtain 91.1% and 77.2% top-1 accuracy on the challenging Kinetics-400 and Something-Something V2 benchmarks, respectively. All of these results effectively show the generality of our InternVideo for video understanding. The code will be released at https://github.com/OpenGVLab/InternVideo .
['Yu Qiao', 'LiMin Wang', 'Yali Wang', 'Jiashuo Yu', 'Junting Pan', 'Guo Chen', 'Sen Xing', 'Zun Wang', 'Yi Liu', 'Jilan Xu', 'Hongjie Zhang', 'Zhiyu Zhao', 'Bingkun Huang', 'Yinan He', 'Yizhuo Li', 'Kunchang Li', 'Yi Wang']
2022-12-06
null
null
null
null
['action-classification', 'video-question-answering', 'open-set-action-recognition', 'video-understanding', 'spatio-temporal-action-localization']
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
[ 1.56218819e-02 -3.43398362e-01 -6.88717604e-01 -2.81103313e-01 -7.38007367e-01 -3.71262223e-01 5.25517642e-01 -6.51868224e-01 -2.62152582e-01 3.33370119e-01 2.07209185e-01 -2.72870958e-01 5.48543155e-01 -2.95533270e-01 -1.11510539e+00 -5.95559597e-01 7.90674388e-02 -1.08969316e-01 3.20933759e-01 -8.97541791e-02 1.87284616e-03 -9.35993716e-02 -1.67602587e+00 8.24068606e-01 6.55476272e-01 8.93913746e-01 5.03795743e-01 9.34058309e-01 3.85681875e-02 1.19449914e+00 -1.12977885e-01 -3.71611863e-01 1.88476682e-01 -3.25868309e-01 -5.41935802e-01 3.42596233e-01 8.99232686e-01 -7.70179689e-01 -1.02076590e+00 1.08919895e+00 3.26789558e-01 -9.12947357e-02 3.92884046e-01 -1.62933493e+00 -8.57212842e-01 6.26185358e-01 -7.18381882e-01 5.94327271e-01 4.38390046e-01 6.84664488e-01 9.39401805e-01 -1.21208680e+00 6.21140361e-01 1.18073797e+00 4.66581225e-01 5.89392006e-01 -7.74445415e-01 -8.45733285e-01 4.12623018e-01 6.22092962e-01 -1.34094822e+00 -8.59661400e-01 4.85296071e-01 -6.16949558e-01 1.11745310e+00 -1.94800328e-02 8.06409061e-01 1.47332001e+00 1.44884020e-01 1.27499449e+00 6.75284922e-01 -7.81572685e-02 -3.15208167e-01 -2.16485977e-01 5.66055067e-02 1.02577090e+00 -1.32010328e-02 2.28881985e-01 -8.43136430e-01 4.25905615e-01 8.06686938e-01 1.09121524e-01 -5.21414995e-01 -2.44683072e-01 -1.32363558e+00 6.74126029e-01 1.13502681e-01 -8.10262486e-02 -6.44385293e-02 4.99808133e-01 5.80504656e-01 2.48589516e-01 2.64707804e-01 -1.79925159e-01 -3.94399613e-01 -3.41952384e-01 -8.90453935e-01 -2.34098434e-02 4.24792677e-01 1.37790155e+00 6.16887033e-01 5.04819334e-01 -2.48631671e-01 6.28806949e-01 5.15478313e-01 6.73628747e-01 3.41591984e-01 -1.23140585e+00 6.77618206e-01 2.83621818e-01 -3.32914025e-01 -8.49984586e-01 4.37129252e-02 -2.13874698e-01 -7.65172601e-01 -1.71390682e-01 1.30777299e-01 -6.99795187e-02 -1.18209374e+00 1.66420484e+00 1.88032210e-01 8.21633518e-01 1.05441079e-01 8.00505877e-01 1.45157850e+00 9.92837727e-01 1.20904453e-01 -1.79199472e-01 1.17141211e+00 -1.71270120e+00 -5.95845938e-01 -4.51379776e-01 6.12670898e-01 -6.46895111e-01 1.08516550e+00 2.53104717e-01 -1.14460409e+00 -1.00073183e+00 -8.84455800e-01 -3.08960527e-01 9.72697511e-02 3.44110221e-01 7.39979208e-01 3.28644037e-01 -1.21859789e+00 2.15720758e-01 -1.09495366e+00 -3.19423318e-01 6.79183125e-01 1.25566587e-01 -4.30881113e-01 -3.36040288e-01 -9.21038389e-01 1.70089141e-01 3.25054735e-01 1.32713869e-01 -1.47860277e+00 -7.37288415e-01 -1.05141425e+00 -1.19848929e-01 5.93963325e-01 -9.07974780e-01 1.15947115e+00 -1.09438848e+00 -1.31300652e+00 1.03494799e+00 -1.19172946e-01 -4.61072177e-01 5.46949327e-01 -5.82527757e-01 -4.62596178e-01 3.71034116e-01 8.32485929e-02 1.14782774e+00 1.00067580e+00 -8.56778741e-01 -6.75063312e-01 -6.03673346e-02 1.87673315e-01 2.59944856e-01 -3.86304349e-01 1.29018515e-01 -1.32682824e+00 -7.25273490e-01 -2.17719153e-01 -9.32500422e-01 8.43934566e-02 5.12329042e-02 -2.30635956e-01 -2.44628321e-02 1.06873918e+00 -7.88854957e-01 1.14195716e+00 -2.22094369e+00 2.99047440e-01 -5.56317091e-01 3.81500125e-01 4.76877779e-01 -4.83476847e-01 2.45995656e-01 -1.11136995e-01 7.22942408e-03 1.60869956e-01 -2.62023181e-01 -2.70070583e-01 4.14807498e-02 -2.91329622e-01 3.38123679e-01 1.62997186e-01 1.26297653e+00 -8.03898752e-01 -5.58670759e-01 4.46343124e-01 4.54993337e-01 -8.67854178e-01 4.13807333e-01 -3.13227206e-01 2.98985571e-01 -3.76637459e-01 9.75534439e-01 3.83503139e-01 -5.23677111e-01 8.55941102e-02 -5.40059149e-01 5.06724827e-02 -1.08723864e-01 -7.76409268e-01 2.04481745e+00 -3.13433185e-02 1.03092039e+00 1.81354675e-03 -1.09438241e+00 4.78863180e-01 1.60001576e-01 5.59852839e-01 -5.45349717e-01 1.81096531e-02 -9.38494429e-02 -1.39414698e-01 -8.05984378e-01 3.56741995e-01 6.58024549e-01 2.30795085e-01 3.26297358e-02 3.49108040e-01 2.28792727e-01 3.93771946e-01 5.61090887e-01 9.91659939e-01 5.48481047e-01 2.13501871e-01 1.82208307e-02 5.40483773e-01 -2.38129526e-01 5.54779112e-01 6.33242548e-01 -4.54761177e-01 7.86306858e-01 4.08871830e-01 -3.05430412e-01 -8.58217239e-01 -1.03768921e+00 2.58395284e-01 1.29183245e+00 2.33536899e-01 -8.61190140e-01 -7.18454301e-01 -5.67921400e-01 -1.96043774e-01 2.81831741e-01 -5.20031810e-01 -2.65651792e-01 -5.49729466e-01 -5.62471211e-01 5.18228590e-01 7.95538008e-01 7.53844500e-01 -9.19655323e-01 -1.39982313e-01 -8.22240040e-02 -5.07520974e-01 -1.68227732e+00 -7.85539031e-01 -3.78350675e-01 -7.40879059e-01 -1.14178538e+00 -4.99136657e-01 -8.71409655e-01 3.05534244e-01 8.95659208e-01 1.25336576e+00 3.06537330e-01 -3.11275095e-01 7.27596998e-01 -4.95275855e-01 7.86579028e-02 -3.11037689e-01 -1.11411482e-01 4.39792871e-02 2.98298206e-02 4.20141160e-01 -4.81439978e-01 -6.82705581e-01 3.49486798e-01 -7.28440166e-01 4.77677613e-01 5.55285633e-01 8.11422765e-01 7.01364696e-01 -4.49440062e-01 2.16508269e-01 -5.60757518e-01 -1.57546028e-01 -5.79128683e-01 -5.20863175e-01 2.91232049e-01 -3.26741993e-01 -2.75439978e-01 3.46599877e-01 -4.51756388e-01 -7.54754543e-01 1.63303524e-01 -1.19833283e-01 -1.08017218e+00 -1.05460957e-01 3.43465865e-01 -4.66750920e-01 -1.75929353e-01 3.08448821e-01 6.15831554e-01 2.18189005e-02 -1.57581940e-01 4.86753285e-01 6.10026836e-01 8.59256744e-01 -5.12938082e-01 6.57991886e-01 3.91941190e-01 -4.37225491e-01 -9.33351278e-01 -8.76646936e-01 -4.22819763e-01 -4.28412378e-01 -3.90324414e-01 1.18998647e+00 -1.56237054e+00 -5.07720649e-01 7.87256241e-01 -1.00727522e+00 -5.62574923e-01 7.15198666e-02 4.01261836e-01 -7.55650938e-01 6.57610655e-01 -7.43765652e-01 -3.24934781e-01 -2.29913801e-01 -1.54003239e+00 1.09141111e+00 3.56343538e-01 3.22976917e-01 -7.09844887e-01 -2.09189236e-01 8.12268078e-01 1.09915823e-01 -6.67911023e-02 3.09682548e-01 -2.83487260e-01 -1.05250847e+00 2.21802816e-01 -4.03041482e-01 5.84386945e-01 -4.39337231e-02 4.49282467e-01 -9.09024775e-01 -4.75746125e-01 -3.53567004e-01 -6.66969001e-01 1.28552628e+00 4.96521801e-01 1.45323670e+00 -1.09482221e-01 -3.07533562e-01 1.38293779e+00 1.14890289e+00 1.53552234e-01 8.31536233e-01 3.03509742e-01 1.20279789e+00 1.31489083e-01 6.20972455e-01 3.55147183e-01 6.93526506e-01 7.86007106e-01 6.00268126e-01 -2.03214493e-02 -4.90302682e-01 -4.02786374e-01 9.38357413e-01 8.43807042e-01 -2.97392666e-01 -3.50356758e-01 -8.91709745e-01 3.22048962e-01 -2.03564429e+00 -1.30406237e+00 -6.48753792e-02 1.73780382e+00 5.50182581e-01 5.64404465e-02 2.69117057e-01 -5.39588630e-01 7.81842828e-01 6.22664392e-01 -6.51959300e-01 3.11501473e-01 -3.43830466e-01 -3.70621413e-01 3.89267266e-01 2.92521030e-01 -1.43166518e+00 1.31881964e+00 6.12911701e+00 8.46866488e-01 -9.73219037e-01 1.29875332e-01 8.81126821e-01 -2.66662955e-01 1.11423191e-02 -8.13120008e-02 -1.02005744e+00 5.78062296e-01 8.82006705e-01 -1.20277748e-01 3.90082210e-01 9.21482086e-01 2.84410715e-01 1.69004977e-01 -1.24783456e+00 1.48470628e+00 3.98867249e-01 -1.65234852e+00 3.26291382e-01 -1.05264857e-01 7.97014713e-01 4.94700819e-01 2.10131049e-01 5.10888159e-01 -1.94390933e-03 -1.01411688e+00 7.49712527e-01 3.23907435e-01 9.69974160e-01 -3.32159966e-01 4.26197171e-01 -1.29252793e-02 -1.49505019e+00 -1.20404117e-01 -2.90617049e-01 1.21630691e-01 4.28498805e-01 2.20241584e-02 -2.16237560e-01 4.36571687e-01 9.33102250e-01 1.43961179e+00 -6.40622318e-01 8.03116679e-01 -2.23860443e-01 8.40740800e-01 -1.18462950e-01 4.38222945e-01 3.03408504e-01 -1.03156276e-01 4.03029650e-01 1.34769332e+00 2.05722362e-01 1.07202485e-01 5.65014899e-01 4.46070015e-01 -1.97540596e-01 -7.83211514e-02 -5.93134224e-01 -4.00395930e-01 3.22298974e-01 1.10690284e+00 -4.38347906e-01 -4.45551515e-01 -9.10263658e-01 1.03791273e+00 2.36479267e-01 5.51744759e-01 -1.42082191e+00 2.11712927e-01 1.03753293e+00 1.27259232e-02 5.89960277e-01 -3.88774753e-01 3.35346073e-01 -1.73843014e+00 2.67592669e-02 -1.21867156e+00 4.06204760e-01 -9.64707434e-01 -8.48892808e-01 4.95564789e-01 5.70463799e-02 -1.30152547e+00 -2.92955101e-01 -8.05598021e-01 -5.03311455e-01 4.37413417e-02 -1.42940724e+00 -1.37646043e+00 -6.43520474e-01 8.74359369e-01 1.19671166e+00 -5.60542941e-01 2.95195282e-01 5.15469968e-01 -8.67437065e-01 6.70853913e-01 3.78427878e-02 4.55163121e-01 8.33037734e-01 -8.20218980e-01 6.60405219e-01 1.21179342e+00 5.39538085e-01 2.17666104e-01 4.35111314e-01 -5.78842163e-01 -1.83807063e+00 -1.31749392e+00 2.08084553e-01 -4.41781074e-01 7.78545976e-01 -3.43857467e-01 -7.77116358e-01 1.02412510e+00 3.67067277e-01 4.07645106e-01 5.07298231e-01 -2.15016976e-01 -6.72313452e-01 -8.20334479e-02 -2.91338086e-01 7.04204559e-01 1.63484585e+00 -6.16054595e-01 -2.54745752e-01 6.44264936e-01 1.01784253e+00 -6.02568984e-01 -6.59598470e-01 3.97000849e-01 6.51219606e-01 -1.16039324e+00 1.25991917e+00 -9.45564508e-01 8.66922736e-01 -3.21583718e-01 -4.29742038e-01 -7.64916062e-01 -3.06756824e-01 -7.42332041e-01 -6.05318129e-01 1.25302470e+00 1.58342943e-01 -2.06991091e-01 8.43813419e-01 2.32466578e-01 -3.74241024e-01 -8.09488118e-01 -4.72280562e-01 -6.66179597e-01 -3.07718843e-01 -6.75412357e-01 9.75267962e-02 8.33464444e-01 -4.18424934e-01 3.21647286e-01 -8.67618144e-01 7.17158196e-03 6.01091802e-01 5.14082573e-02 1.16539943e+00 -4.73609358e-01 -6.38396800e-01 -3.69873255e-01 -6.59927845e-01 -1.58484280e+00 3.61456096e-01 -8.50138783e-01 -1.89914033e-01 -1.30036831e+00 5.77438951e-01 1.55325308e-01 -3.04256320e-01 4.55642372e-01 -3.44609112e-01 4.72076595e-01 4.81654674e-01 3.33280414e-01 -1.09207106e+00 6.82846189e-01 1.12831974e+00 -3.69757444e-01 3.07923518e-02 -2.32361570e-01 -7.06708968e-01 8.79546404e-01 6.75268829e-01 -4.91950586e-02 -7.47252584e-01 -8.77046824e-01 -1.40805274e-01 1.16720855e-01 6.09593689e-01 -8.91863942e-01 1.75147191e-01 -3.46677899e-01 2.97462821e-01 -5.74512005e-01 4.13511693e-01 -3.26515049e-01 1.30810246e-01 3.83067518e-01 -2.48256743e-01 2.70503074e-01 2.00218812e-01 8.01454544e-01 -3.90534133e-01 2.75171846e-01 6.73742652e-01 -1.01907596e-01 -1.60038137e+00 8.87824893e-01 -1.13838971e-01 2.87141681e-01 1.16450965e+00 -2.22204104e-01 -6.64355993e-01 -5.50777435e-01 -6.09749019e-01 3.82662416e-01 5.44533610e-01 7.56414115e-01 8.58090758e-01 -1.26302838e+00 -9.89290059e-01 2.24923566e-01 2.49264419e-01 -2.20571831e-01 6.93299711e-01 8.93278062e-01 -6.74040377e-01 4.42576617e-01 -3.59866172e-01 -1.04145467e+00 -1.48748434e+00 6.76296473e-01 2.56639868e-01 9.89165902e-02 -7.97456980e-01 1.18368876e+00 7.08292425e-01 1.25284746e-01 2.76880085e-01 -1.10342294e-01 -1.27368599e-01 -2.26652190e-01 8.19249451e-01 1.35998353e-01 -3.51306796e-01 -8.34143281e-01 -4.32286352e-01 5.96976519e-01 -4.10361350e-01 1.81198239e-01 1.10191071e+00 -3.07770133e-01 1.65833250e-01 8.58094692e-02 1.29906833e+00 -3.21732312e-01 -1.73532891e+00 -1.81122556e-01 -5.29344141e-01 -6.14946544e-01 -3.65225598e-02 -3.04922104e-01 -1.54963386e+00 9.94326055e-01 4.34021503e-01 -4.04929996e-01 1.26988983e+00 1.04955114e-01 7.70189047e-01 4.17468607e-01 2.07585335e-01 -9.12542462e-01 4.21425968e-01 4.86105382e-01 7.18397141e-01 -1.52378941e+00 3.31277326e-02 -5.88253379e-01 -8.21297944e-01 1.01179004e+00 1.13525856e+00 -4.04812135e-02 4.68825012e-01 3.59310776e-01 1.28496513e-01 9.22605619e-02 -1.21766198e+00 -1.29378840e-01 3.58074874e-01 6.67367280e-01 5.34643948e-01 -2.25656614e-01 -4.84288260e-02 4.05234635e-01 1.58243954e-01 9.80531499e-02 5.43650866e-01 6.15184784e-01 -3.86061877e-01 -7.07069635e-01 2.49278601e-02 2.02751443e-01 -4.31702763e-01 -3.82138848e-01 -4.51273471e-02 8.65304887e-01 -1.60507336e-01 8.03321362e-01 -6.22998504e-03 -7.42092371e-01 -1.58000946e-01 -1.48608416e-01 4.93303835e-01 -4.85642880e-01 -9.01876539e-02 2.23394126e-01 4.25898880e-02 -1.16549933e+00 -6.76775157e-01 -7.30451167e-01 -9.97403562e-01 -3.32864136e-01 3.43171172e-02 -2.16617167e-01 2.65407860e-01 9.56355214e-01 5.51965535e-01 5.09347022e-01 3.50425661e-01 -9.11685586e-01 -3.39350343e-01 -6.81352854e-01 -1.01620831e-01 3.11857432e-01 2.33525082e-01 -6.69632792e-01 -2.08219901e-01 5.24675786e-01]
[9.722247123718262, 0.8124778866767883]
438b4efa-24f8-4cfd-aa73-81d6652be46f
self-supervised-learning-for-robust-voice
2204.03421
null
https://arxiv.org/abs/2204.03421v2
https://arxiv.org/pdf/2204.03421v2.pdf
Self-supervised learning for robust voice cloning
Voice cloning is a difficult task which requires robust and informative features incorporated in a high quality TTS system in order to effectively copy an unseen speaker's voice. In our work, we utilize features learned in a self-supervised framework via the Bootstrap Your Own Latent (BYOL) method, which is shown to produce high quality speech representations when specific audio augmentations are applied to the vanilla algorithm. We further extend the augmentations in the training procedure to aid the resulting features to capture the speaker identity and to make them robust to noise and acoustic conditions. The learned features are used as pre-trained utterance-level embeddings and as inputs to a Non-Attentive Tacotron based architecture, aiming to achieve multispeaker speech synthesis without utilizing additional speaker features. This method enables us to train our model in an unlabeled multispeaker dataset as well as use unseen speaker embeddings to copy a speaker's voice. Subjective and objective evaluations are used to validate the proposed model, as well as the robustness to the acoustic conditions of the target utterance.
['Pirros Tsiakoulis', 'Aimilios Chalamandaris', 'Gunu Jho', 'June Sig Sung', 'Spyros Raptis', 'Konstantinos Markopoulos', 'Panos Kakoulidis', 'Georgios Vamvoukakis', 'Karolos Nikitaras', 'Nikolaos Ellinas', 'Konstantinos Klapsas']
2022-04-07
null
null
null
null
['voice-cloning']
['speech']
[ 3.69116187e-01 3.39930296e-01 1.92464262e-01 -3.02225173e-01 -1.07605910e+00 -6.52673721e-01 6.33200049e-01 -1.59462348e-01 -1.91022396e-01 5.92990160e-01 5.01600444e-01 -9.62185860e-02 1.88781768e-01 -2.66339391e-01 -5.87604403e-01 -8.07623327e-01 3.43994856e-01 2.84164995e-01 2.56751053e-04 -1.29144177e-01 -9.22672525e-02 6.06339335e-01 -1.93484080e+00 1.17932968e-01 6.09962702e-01 8.28102052e-01 3.79631728e-01 1.00523746e+00 2.56194975e-02 4.62020874e-01 -8.28021705e-01 -9.85359773e-02 2.18313590e-01 -4.37867790e-01 -5.38555861e-01 4.12892699e-01 3.37023824e-01 -2.59080708e-01 -5.43528795e-02 6.79743111e-01 6.71593487e-01 4.78569001e-01 6.80799723e-01 -7.79658377e-01 -5.24752378e-01 6.38663709e-01 2.96779513e-01 6.10964224e-02 3.38926077e-01 1.78450465e-01 1.13550746e+00 -1.04642928e+00 3.20055336e-01 1.29328966e+00 4.39446598e-01 8.09584498e-01 -1.27927291e+00 -6.06739938e-01 -1.56683698e-02 1.17443176e-02 -1.22246051e+00 -1.43190169e+00 1.08889472e+00 -3.14948529e-01 6.08656824e-01 4.03218836e-01 3.72362405e-01 1.39847469e+00 -2.45506451e-01 7.28211045e-01 1.00484502e+00 -8.10502887e-01 4.46391314e-01 7.47744441e-01 -1.67353615e-01 4.59695637e-01 -4.81430709e-01 1.30675718e-01 -5.36878049e-01 -1.50497615e-01 3.90842110e-01 -3.81074131e-01 -5.04711807e-01 -2.43685633e-01 -9.22290981e-01 9.02174413e-01 1.40148044e-01 6.20193124e-01 -5.23956597e-01 3.80000216e-03 2.30805859e-01 4.06118631e-01 4.76553828e-01 5.06570935e-01 -2.00924665e-01 -1.67217851e-01 -1.01670718e+00 -1.29225180e-01 7.83365428e-01 7.46317267e-01 6.16163015e-01 6.98764980e-01 -2.17549235e-01 1.15506494e+00 3.95168155e-01 3.99647534e-01 8.88871551e-01 -7.01516747e-01 2.18765765e-01 1.60685867e-01 2.09008947e-01 -6.02604032e-01 6.52762279e-02 -7.59040058e-01 -3.26547831e-01 1.23563267e-01 -6.29897043e-02 -9.33558717e-02 -9.08337176e-01 1.89194143e+00 3.99534643e-01 4.80311781e-01 5.06895602e-01 6.06594920e-01 6.45333111e-01 7.88429797e-01 -2.89331704e-01 -3.40628356e-01 1.09079540e+00 -1.08348191e+00 -9.72401738e-01 -1.66072190e-01 3.08556139e-01 -1.08118403e+00 1.38985789e+00 2.79723018e-01 -9.83179092e-01 -9.05058861e-01 -1.24162638e+00 1.97988287e-01 -2.19045535e-01 3.78806174e-01 7.16626197e-02 1.16460323e+00 -9.62521195e-01 3.72412056e-01 -6.60186887e-01 -1.81879118e-01 4.72628418e-03 3.94452602e-01 -3.65328282e-01 2.83593446e-01 -1.24318063e+00 8.80989313e-01 1.30856872e-01 2.48411551e-01 -1.31322956e+00 -4.95627970e-01 -9.08069253e-01 2.47897267e-01 1.37011871e-01 -3.38697612e-01 1.39495075e+00 -9.84793663e-01 -2.29176950e+00 3.98795575e-01 -4.09934968e-01 -4.16689157e-01 4.37509924e-01 -1.71583652e-01 -6.76083863e-01 3.54573131e-01 -6.33768886e-02 3.97910118e-01 1.45762289e+00 -1.35983801e+00 -3.38515311e-01 -5.29649854e-02 -2.53309637e-01 2.22534865e-01 -6.23528123e-01 -1.05447814e-01 -1.49976075e-01 -7.64474511e-01 1.41887041e-02 -9.55474377e-01 1.28919840e-01 -3.66778105e-01 -4.66814995e-01 -1.15433410e-02 9.03628469e-01 -6.54859543e-01 8.10678124e-01 -2.33155394e+00 1.25336334e-01 1.86283797e-01 -2.41941676e-01 3.59069318e-01 -3.02176178e-01 4.28398252e-01 -1.21572524e-01 -2.64964014e-01 -7.03131780e-02 -9.67335880e-01 7.99454898e-02 2.35223979e-01 -4.93845314e-01 3.49825948e-01 4.27122027e-01 4.45743829e-01 -5.50014675e-01 -6.20653518e-02 4.96506661e-01 6.20784044e-01 -5.07948279e-01 7.92066455e-01 -6.31940318e-03 6.70854032e-01 -1.93004236e-01 2.11093128e-01 3.39340866e-01 4.92894769e-01 -1.15740404e-01 -1.51742816e-01 -8.18616226e-02 5.73909760e-01 -1.26777017e+00 1.41591990e+00 -1.05126882e+00 5.27365506e-01 3.43135059e-01 -7.91650057e-01 1.20343494e+00 1.00954020e+00 -6.75574988e-02 -3.21753323e-01 2.87279278e-01 3.85145694e-01 9.12837610e-02 -4.44273144e-01 4.42305386e-01 -5.49368918e-01 1.41305894e-01 3.97226393e-01 4.30335343e-01 -3.15881133e-01 -3.95101696e-01 -1.39187932e-01 7.08051383e-01 -1.13195248e-01 3.57757211e-02 -1.80265993e-01 9.34540093e-01 -6.38494313e-01 2.94657648e-01 5.49471617e-01 4.81491685e-02 6.49463356e-01 -1.58648327e-01 1.97302237e-01 -9.61871505e-01 -9.96499658e-01 -8.28770548e-02 1.13591671e+00 -4.09989536e-01 1.31043136e-01 -8.61248910e-01 -1.93678305e-01 -2.97174096e-01 1.17661417e+00 -4.74740535e-01 -2.60697037e-01 -3.99263769e-01 -1.41428560e-01 4.94631916e-01 4.27452832e-01 -8.89711678e-02 -9.99351799e-01 -1.03641525e-01 6.79423809e-01 -1.07887127e-01 -1.15654385e+00 -5.68982184e-01 5.79899490e-01 -4.81315762e-01 -5.10870755e-01 -7.88052559e-01 -8.19531202e-01 3.90824288e-01 -9.94597301e-02 3.89034390e-01 -4.69725847e-01 -2.17068270e-02 4.53845769e-01 -3.97541374e-01 -3.46997142e-01 -1.18697703e+00 1.11525834e-01 6.15723789e-01 6.19953930e-01 -7.06609115e-02 -7.74227679e-01 -2.73541957e-01 3.92266989e-01 -7.60754466e-01 -2.61402547e-01 4.25129980e-01 1.07751405e+00 2.26300716e-01 -4.10347655e-02 1.10192204e+00 -5.72326779e-01 8.54369819e-01 -1.83893591e-01 -4.17193204e-01 6.40683100e-02 -3.62123579e-01 1.04415901e-01 9.88016546e-01 -7.48346984e-01 -1.08169365e+00 5.74349202e-02 -6.01664484e-01 -6.89792275e-01 -3.62861097e-01 8.44807178e-02 -4.98892844e-01 -1.32921152e-04 5.32095313e-01 4.36655611e-01 1.23097673e-01 -7.05137610e-01 6.62001610e-01 1.30261087e+00 6.42213464e-01 -3.59656036e-01 1.00058103e+00 1.50960954e-02 -4.32403147e-01 -1.12235463e+00 -6.51376486e-01 -6.00442469e-01 -7.72588611e-01 -3.42335217e-02 5.43703556e-01 -9.31026697e-01 -5.73599398e-01 2.78018653e-01 -1.04253447e+00 -1.84299201e-02 -4.26070333e-01 7.06799030e-01 -7.13155746e-01 1.56733423e-01 -3.17361653e-01 -1.23864460e+00 -3.18990976e-01 -1.42700195e+00 8.11307251e-01 1.00944601e-02 -2.20872864e-01 -9.30036247e-01 5.27158119e-02 5.24497092e-01 5.81187606e-01 -2.07018122e-01 8.53354096e-01 -1.05303335e+00 -3.61208946e-01 -2.03576282e-01 5.08094966e-01 9.86554921e-01 6.32322133e-01 -1.18217856e-01 -1.77523279e+00 -3.75537992e-01 4.54784721e-01 -4.78021771e-01 5.58451355e-01 1.60826340e-01 5.11228621e-01 -5.17661929e-01 8.42536837e-02 3.35211426e-01 9.11263525e-01 1.49862885e-01 1.97148442e-01 -1.77290872e-01 4.51159239e-01 6.93799436e-01 3.15715939e-01 1.89790517e-01 -1.09179109e-01 7.66112983e-01 2.21359536e-01 -2.07602214e-02 -2.90697277e-01 -3.40061039e-01 5.13420880e-01 1.38952136e+00 3.92595023e-01 -2.89135844e-01 -3.21326673e-01 6.49704039e-01 -1.26769650e+00 -8.58270764e-01 2.74614066e-01 2.24841714e+00 8.38884532e-01 1.35512888e-01 -1.18561290e-01 5.03470421e-01 8.65815759e-01 1.83817983e-01 -3.94779652e-01 -6.64201617e-01 1.02762341e-01 5.81819236e-01 -6.80736005e-02 7.46210814e-01 -9.61474895e-01 1.00156891e+00 5.80596113e+00 7.25232422e-01 -1.45143211e+00 2.02126846e-01 1.57351971e-01 3.80633436e-02 -4.23425764e-01 -2.48094007e-01 -5.94902694e-01 1.72407195e-01 1.32482493e+00 -4.05443944e-02 6.11675739e-01 8.80659342e-01 5.23597300e-01 3.85527730e-01 -1.32410181e+00 7.14141190e-01 3.19804937e-01 -1.04256797e+00 -6.85452670e-02 -1.56418234e-01 3.27552915e-01 -2.37589523e-01 2.95863956e-01 4.15456921e-01 -9.83214006e-02 -9.92030025e-01 8.27554107e-01 2.95093626e-01 6.57702923e-01 -7.59236813e-01 5.92519820e-01 4.54611123e-01 -1.04636133e+00 -1.14144139e-01 -2.74199009e-01 1.64360240e-01 2.19690531e-01 1.51300207e-01 -1.64839566e+00 4.12920684e-01 5.46947047e-02 5.43805659e-02 -2.45816633e-01 7.15593517e-01 -1.33434981e-01 1.01490843e+00 -2.77314365e-01 -8.59856233e-03 2.39216268e-01 1.84342459e-01 7.92452037e-01 9.32809114e-01 3.59198451e-01 -3.02396238e-01 -6.74300790e-02 9.11323667e-01 -2.30163157e-01 4.23033834e-01 -5.35446286e-01 -3.79820198e-01 6.13522530e-01 1.15394390e+00 -1.79688364e-01 -1.45595431e-01 -8.10255483e-02 1.17332840e+00 9.85924006e-02 4.23853546e-01 -4.45766926e-01 -4.61427629e-01 6.28915548e-01 8.03929521e-04 5.03889263e-01 -3.11768036e-02 3.20412368e-01 -1.08690929e+00 -1.00753173e-01 -9.12641346e-01 -2.42594466e-01 -7.53970444e-01 -1.15798461e+00 1.03812027e+00 -4.74900872e-01 -1.30697048e+00 -7.20057964e-01 -2.83309311e-01 -7.44719982e-01 1.34115303e+00 -1.60456753e+00 -1.23309922e+00 6.10607937e-02 6.02977633e-01 7.05478191e-01 -5.82176507e-01 1.33044934e+00 2.81867445e-01 -4.07435507e-01 9.35735226e-01 3.34033102e-01 -6.26134798e-02 7.00198770e-01 -1.07841480e+00 2.65234917e-01 7.50351012e-01 3.12099129e-01 7.73318887e-01 9.44678783e-01 -1.90700710e-01 -1.25646424e+00 -1.02784562e+00 7.11070299e-01 -1.96660534e-01 6.07559204e-01 -7.61298954e-01 -9.20372605e-01 5.48396409e-01 2.11283401e-01 4.00524847e-02 1.03941512e+00 9.53349173e-02 -1.00605890e-01 -2.34515592e-01 -1.12219906e+00 3.99159491e-01 3.79985124e-01 -9.11776423e-01 -1.09293413e+00 4.59696129e-02 9.09523547e-01 -5.40107675e-02 -6.05334997e-01 -8.31100275e-04 4.49879378e-01 -6.97707057e-01 6.97811484e-01 -4.64279234e-01 -1.72188491e-01 -1.50591537e-01 -3.00705731e-01 -1.47534871e+00 -6.60451576e-02 -9.91467535e-01 1.69753492e-01 1.81484222e+00 4.85859334e-01 -7.61118054e-01 5.84584296e-01 1.99130893e-01 -4.27379757e-01 -3.66860926e-01 -1.17470264e+00 -7.86463916e-01 -1.60274521e-01 -4.39235985e-01 6.43263936e-01 8.13081682e-01 -2.19072580e-01 4.70828950e-01 -5.69097221e-01 5.28831661e-01 3.52976084e-01 1.67229697e-02 8.70875120e-01 -1.03147876e+00 -5.21292984e-01 2.39750184e-02 -3.19702446e-01 -1.01564586e+00 4.02197331e-01 -9.05479193e-01 3.55015785e-01 -8.89938951e-01 -4.27725792e-01 -4.63896424e-01 -4.02610272e-01 2.30377287e-01 -4.69757058e-02 1.89971134e-01 1.76530644e-01 9.19229463e-02 9.73083451e-02 1.09400940e+00 8.77488017e-01 -8.29847381e-02 -4.30414617e-01 5.27140856e-01 -5.76550126e-01 6.27096772e-01 5.76863408e-01 -5.75681388e-01 -5.91544867e-01 -6.26506805e-02 -7.12439835e-01 2.54404783e-01 1.33963674e-01 -1.03385544e+00 7.70438986e-04 2.80005813e-01 1.93203300e-01 -2.30918735e-01 9.92599130e-01 -9.09313321e-01 -2.81103384e-02 2.14195892e-01 -5.99790573e-01 -4.05222863e-01 1.68200329e-01 4.59966034e-01 -4.23740268e-01 -6.84413135e-01 9.65174615e-01 1.77216128e-01 -1.21863119e-01 -9.09982249e-02 -5.01332223e-01 -2.95685560e-01 7.48966873e-01 -2.34730184e-01 2.41946086e-01 -6.17482960e-01 -1.09239638e+00 -1.85779855e-01 1.98751256e-01 5.01775384e-01 5.00213146e-01 -1.31689310e+00 -6.01382315e-01 5.81648946e-01 2.06054449e-01 -4.47324932e-01 3.08504373e-01 4.75399613e-01 -1.21524185e-01 4.62886423e-01 -8.68629515e-02 -5.83176434e-01 -1.26331627e+00 4.93308067e-01 3.13357353e-01 2.95974404e-01 -4.25092012e-01 8.98336768e-01 -1.66767109e-02 -5.75761557e-01 4.56475914e-01 -3.55051458e-01 -2.22679034e-01 1.28529400e-01 4.51200038e-01 -1.53100481e-02 2.63267308e-01 -8.72926474e-01 -2.80941457e-01 1.98620468e-01 -4.64567170e-02 -5.72545707e-01 1.30771518e+00 -3.44246328e-01 5.06794393e-01 7.52493799e-01 1.41596830e+00 6.41377389e-01 -1.02463460e+00 -3.82187128e-01 -1.82976276e-01 -3.96708667e-01 2.56747216e-01 -5.38085401e-01 -5.66491485e-01 1.19895136e+00 7.96856940e-01 2.38451838e-01 8.38286340e-01 -7.82514140e-02 7.69272566e-01 3.12345117e-01 1.40623420e-01 -9.98690486e-01 1.95129558e-01 1.32022038e-01 1.01332319e+00 -9.23187912e-01 -4.64638621e-01 -2.27528602e-01 -7.05927730e-01 1.18232501e+00 1.09559879e-01 9.32917744e-02 4.36326087e-01 1.50464416e-01 5.09029090e-01 3.88038963e-01 -7.50362694e-01 -2.25609601e-01 2.64062703e-01 5.36250889e-01 2.83183038e-01 -6.01094477e-02 2.69059420e-01 5.12751281e-01 -5.80327451e-01 -3.41163307e-01 5.26582897e-01 5.35090387e-01 -4.90874469e-01 -1.31000698e+00 -3.70627254e-01 3.06965727e-02 -3.50480497e-01 -5.85419475e-04 -2.14539036e-01 2.82445252e-01 -1.14263311e-01 1.22242415e+00 -1.05924293e-01 -3.43890339e-01 3.28247666e-01 6.53768599e-01 1.95270061e-01 -9.41351831e-01 -6.98949814e-01 4.45476741e-01 2.22669721e-01 2.30137501e-02 -2.61764675e-01 -5.67875445e-01 -1.01556134e+00 3.48786891e-01 -7.94997215e-01 3.79387230e-01 9.63851869e-01 1.06525850e+00 1.44863978e-01 7.27146804e-01 1.27328479e+00 -8.76231730e-01 -1.01377869e+00 -1.29634976e+00 -5.62331140e-01 3.14164877e-01 6.79402530e-01 -6.15394652e-01 -7.78937101e-01 9.58547816e-02]
[14.868396759033203, 6.540005207061768]
ae950962-070d-4233-b57d-e3c0e6a54b76
task-specific-normalization-for-continual
2107.13429
null
https://arxiv.org/abs/2107.13429v2
https://arxiv.org/pdf/2107.13429v2.pdf
Task-Specific Normalization for Continual Learning of Blind Image Quality Models
The computational vision community has recently paid attention to continual learning for blind image quality assessment (BIQA). The primary challenge is to combat catastrophic forgetting of previously-seen IQA datasets (i.e., tasks). In this paper, we present a simple yet effective continual learning method for BIQA with improved quality prediction accuracy, plasticity-stability trade-off, and task-order/-length robustness. The key step in our approach is to freeze all convolution filters of a pre-trained deep neural network (DNN) for an explicit promise of stability, and learn task-specific normalization parameters for plasticity. We assign each new task a prediction head, and load the corresponding normalization parameters to produce a quality score. The final quality estimate is computed by a weighted summation of predictions from all heads with a lightweight K-means gating mechanism, without leveraging the test-time oracle. Extensive experiments on six IQA datasets demonstrate the advantages of the proposed method in comparison to previous training techniques for BIQA.
['Xiaokang Yang', 'Guangtao Zhai', 'Kede Ma', 'Weixia Zhang']
2021-07-28
null
null
null
null
['blind-image-quality-assessment']
['computer-vision']
[ 3.05483013e-01 -2.73231447e-01 4.62096065e-01 -3.71429741e-01 -8.73786390e-01 -3.48280519e-01 3.69435549e-01 -1.81132659e-01 -6.75639987e-01 7.21945941e-01 2.27724880e-01 -3.07400644e-01 -5.02487898e-01 -4.23655778e-01 -7.23716855e-01 -1.00701141e+00 1.38716504e-01 2.50464827e-02 3.88849169e-01 1.06055714e-01 5.87140739e-01 4.13495094e-01 -1.82690430e+00 1.86581060e-01 1.31272304e+00 1.21817970e+00 4.23582137e-01 8.17107201e-01 2.97039181e-01 8.94912660e-01 -5.88397026e-01 -8.59106243e-01 3.07848215e-01 -3.56380522e-01 -9.53675032e-01 1.04285233e-01 7.25550234e-01 -4.33336645e-01 -3.63956690e-01 1.07376969e+00 9.68587399e-01 2.19869643e-01 4.41528916e-01 -1.02143717e+00 -8.68698657e-01 2.00876519e-01 -3.81545365e-01 6.29994392e-01 -1.42789245e-01 7.72965133e-01 7.78882980e-01 -9.20587242e-01 3.02187026e-01 7.89820075e-01 6.94467604e-01 8.77881527e-01 -1.08075595e+00 -4.59441483e-01 1.48886725e-01 6.74266338e-01 -1.04640543e+00 -8.87302399e-01 5.04894912e-01 -3.09918821e-01 1.22939765e+00 5.84228672e-02 8.68817687e-01 8.63442361e-01 4.23875362e-01 6.66212142e-01 1.29543030e+00 -3.42611253e-01 5.44826806e-01 -2.00014055e-01 1.27341300e-01 6.25099242e-01 5.13049215e-02 1.42068416e-01 -8.49424481e-01 1.30617321e-01 5.77476740e-01 -2.59805582e-02 -5.90792179e-01 -3.59660268e-01 -9.23762381e-01 2.24746302e-01 5.77434719e-01 -3.57269831e-02 -5.67947209e-01 2.56991178e-01 1.88544676e-01 5.68655968e-01 1.77021310e-01 4.56058860e-01 -5.33348560e-01 -2.10951045e-01 -1.06563878e+00 6.98775426e-02 2.51558572e-01 5.98746717e-01 6.32439077e-01 8.83548036e-02 -7.18791485e-01 8.43313038e-01 6.69609234e-02 3.46320331e-01 6.11499488e-01 -1.04838991e+00 1.52469501e-01 4.38531399e-01 1.93253517e-01 -2.75672942e-01 -2.53605455e-01 -7.56188750e-01 -7.65702844e-01 7.61773348e-01 4.39502746e-01 1.42993361e-01 -1.40403354e+00 1.74323654e+00 1.88923962e-02 1.84418052e-01 2.31189858e-02 9.19849694e-01 2.67417014e-01 3.43227476e-01 2.60204583e-01 -5.18275261e-01 1.03196311e+00 -1.14807951e+00 -5.05830824e-01 -2.19943970e-01 8.80841091e-02 -7.85193563e-01 1.51472425e+00 7.18267202e-01 -1.34005356e+00 -7.09055364e-01 -1.10400534e+00 -1.72528729e-01 5.10482863e-02 5.33706099e-02 4.60517049e-01 6.41398132e-01 -1.45927167e+00 7.99643159e-01 -7.75855184e-01 -1.88638479e-01 8.46475959e-01 6.26862824e-01 -7.86154121e-02 -1.88625395e-01 -7.25313246e-01 8.91410172e-01 1.00546315e-01 1.20119140e-01 -1.25672269e+00 -6.38669491e-01 -2.10554689e-01 8.74981508e-02 1.66005298e-01 -1.22320044e+00 1.35638165e+00 -1.18586326e+00 -1.68835258e+00 7.74943411e-01 -3.75920981e-01 -6.64113343e-01 3.94377381e-01 -5.79159617e-01 -2.67488539e-01 -8.15666020e-02 -2.00893819e-01 6.91749454e-01 1.41745329e+00 -1.22416508e+00 -6.65478706e-01 -4.92362171e-01 1.73754394e-01 4.25340742e-01 -6.04419410e-01 -9.56698433e-02 -5.64228714e-01 -5.64314365e-01 1.07731760e-01 -6.45814598e-01 1.17423274e-01 2.20485479e-01 -3.54856588e-02 -2.38898009e-01 3.14270705e-01 -1.03087676e+00 1.37231445e+00 -2.20541286e+00 2.39908010e-01 -1.15747146e-01 3.85121137e-01 4.96362060e-01 -2.69179314e-01 -1.51343390e-01 1.68014660e-01 -2.44351730e-01 -4.48901176e-01 -5.88312864e-01 -3.89880449e-01 1.80340305e-01 -3.34715366e-01 3.32205147e-01 4.09615010e-01 9.69512045e-01 -5.91492355e-01 -3.09873939e-01 -1.83663946e-02 4.72479641e-01 -6.22568667e-01 3.63346189e-01 -2.50801921e-01 3.92819494e-01 1.39935806e-01 7.80546248e-01 7.09620059e-01 -2.72477567e-01 -1.76105469e-01 -2.01467633e-01 -1.64347440e-01 3.54665428e-01 -6.80953860e-01 1.97293484e+00 -4.36814755e-01 3.97199035e-01 -2.74070948e-01 -6.04707479e-01 5.28294921e-01 1.72171324e-01 -6.91386536e-02 -1.20380676e+00 2.05781264e-03 1.50075972e-01 2.19083294e-01 -4.66098964e-01 2.99697101e-01 -2.04974845e-01 3.82121593e-01 3.18625838e-01 4.46324736e-01 1.42494112e-01 -1.35574728e-01 5.29463775e-02 1.15551186e+00 9.54048932e-02 -7.99270254e-03 -9.07434374e-02 4.71029907e-01 -3.87342513e-01 6.29564464e-01 8.44509125e-01 -6.51613712e-01 7.75849700e-01 2.41643041e-01 -5.12115359e-01 -1.34901071e+00 -1.24700928e+00 4.96323407e-02 1.12103593e+00 -2.04463303e-02 9.07661244e-02 -7.14411914e-01 -5.96901357e-01 -1.87696084e-01 5.40511250e-01 -6.07927918e-01 -5.14883518e-01 -5.61044395e-01 -7.89120674e-01 1.19219944e-01 4.00114864e-01 6.60970509e-01 -1.20819116e+00 -6.90162539e-01 1.48890600e-01 -4.29208465e-02 -5.74491858e-01 -4.15855229e-01 1.05955593e-01 -1.04205894e+00 -7.45443881e-01 -9.33024108e-01 -1.00521827e+00 6.11504316e-01 3.80246520e-01 9.68386889e-01 1.40539959e-01 -2.78855950e-01 2.08040535e-01 -4.57801595e-02 -1.99020207e-01 -3.29667889e-02 -1.18453942e-01 3.44511509e-01 1.37004331e-01 9.50418599e-03 -7.73977578e-01 -1.13847721e+00 7.87494704e-02 -7.50956416e-01 2.89873723e-02 9.47198570e-01 1.09093571e+00 9.28871572e-01 1.76098496e-01 8.04863513e-01 -3.55264038e-01 7.65511632e-01 1.75101742e-01 -5.47787845e-01 5.16926587e-01 -1.01943898e+00 1.31297752e-01 4.19585705e-01 -6.18686080e-01 -1.20100808e+00 -9.93394945e-03 -9.30718258e-02 -4.70841497e-01 2.89975524e-01 1.01120740e-01 -3.85841012e-01 -4.18868452e-01 8.06249797e-01 5.54151893e-01 -6.81909546e-02 -3.24464887e-01 6.12979770e-01 4.88837451e-01 9.03311491e-01 -3.14190358e-01 6.66481495e-01 2.45330900e-01 -3.64532292e-01 -2.00287864e-01 -9.99631703e-01 -2.94749320e-01 -5.27491748e-01 -4.06617582e-01 6.57015860e-01 -1.00413430e+00 -8.57726812e-01 1.21065474e+00 -1.02943432e+00 -5.32139778e-01 -5.13442397e-01 2.73031503e-01 -6.24917746e-01 2.43474171e-01 -4.99318361e-01 -8.55433226e-01 -8.57240856e-01 -9.59682584e-01 5.05303323e-01 5.28152585e-01 2.85298765e-01 -3.80145848e-01 -2.09837276e-02 5.31648517e-01 7.57021785e-01 -4.35145676e-01 9.49778378e-01 -2.03754306e-02 -7.30220139e-01 1.87467098e-01 -4.69815969e-01 9.10076320e-01 -6.39439076e-02 -3.32876086e-01 -1.32812607e+00 -6.36626065e-01 2.90774852e-01 -5.38795590e-01 1.29116952e+00 4.79416341e-01 1.21554732e+00 -4.91573989e-01 2.14004412e-01 6.73120856e-01 1.53427291e+00 2.05788732e-01 9.98690188e-01 3.93868834e-01 4.22020137e-01 1.59688205e-01 3.23865443e-01 3.53127182e-01 2.36832932e-01 1.89307407e-01 5.63019693e-01 2.69968778e-01 -7.05024064e-01 -5.37955575e-02 3.90002608e-01 9.17009354e-01 -2.04683527e-01 -1.46489203e-01 -8.88533592e-01 8.28697026e-01 -1.60294700e+00 -8.71066928e-01 2.19821870e-01 2.50771332e+00 1.01920760e+00 4.57498342e-01 -1.19063370e-01 4.22272354e-01 4.95466918e-01 -1.60883769e-01 -1.02657068e+00 -1.72642693e-01 -2.89412767e-01 5.13064981e-01 3.17181438e-01 4.32338834e-01 -9.66065943e-01 9.41830635e-01 6.09650040e+00 7.63583481e-01 -1.07632089e+00 3.88296008e-01 6.55130148e-01 -5.35746157e-01 -4.58609983e-02 3.28950025e-02 -4.96183455e-01 3.09885889e-01 1.04564452e+00 -7.90662244e-02 7.98944712e-01 5.92775345e-01 8.00774246e-02 -2.22973719e-01 -8.13364804e-01 9.76575911e-01 1.20385118e-01 -1.07398474e+00 2.14720964e-01 -2.66980499e-01 8.48912060e-01 9.57466513e-02 7.31894433e-01 2.77879685e-01 3.32836583e-02 -9.10364985e-01 7.43212283e-01 1.05964649e+00 1.00106585e+00 -6.98714256e-01 6.18680775e-01 3.93396690e-02 -5.74238062e-01 -6.44672513e-01 -5.35357118e-01 -7.02065453e-02 3.90849635e-02 7.54143894e-01 -5.73669851e-01 1.81164354e-01 1.00685441e+00 3.78388792e-01 -9.01494145e-01 1.77358317e+00 -2.30534703e-01 7.75389314e-01 1.43528283e-01 2.87034065e-01 -1.88983783e-01 2.13882551e-01 3.99037093e-01 6.22915506e-01 4.55776870e-01 -5.24111986e-02 -6.31723881e-01 5.68867803e-01 -3.74355555e-01 -7.71985352e-02 1.14368781e-01 1.02655657e-01 3.46739620e-01 9.51568604e-01 -2.88169980e-01 -2.69992173e-01 -2.61943609e-01 1.53615439e+00 5.81002772e-01 3.26570958e-01 -6.13718033e-01 -3.07518899e-01 6.71886265e-01 1.62963018e-01 4.48042423e-01 -5.20889871e-02 -5.75057507e-01 -9.24739659e-01 4.36206102e-01 -7.74783492e-01 2.66064972e-01 -1.07800734e+00 -1.44523728e+00 7.90170550e-01 -8.41154099e-01 -1.10789752e+00 3.31978172e-01 -4.82092977e-01 -5.49711406e-01 1.16570735e+00 -1.89885795e+00 -1.33830523e+00 -3.43600959e-01 8.66103292e-01 4.35083151e-01 -2.89858758e-01 6.17674470e-01 4.24213618e-01 -5.16904056e-01 8.38670135e-01 4.98014167e-02 -4.15970176e-01 9.63781238e-01 -1.17983747e+00 4.80980754e-01 1.20636201e+00 -1.96331963e-01 6.26520693e-01 5.81692457e-01 -5.63813686e-01 -1.26721919e+00 -1.09376764e+00 1.04549611e+00 -4.88281190e-01 4.96610165e-01 -5.66583015e-02 -1.13580096e+00 2.67769516e-01 1.42065704e-01 1.61450192e-01 4.15630162e-01 1.64969787e-01 -5.98927021e-01 -5.46598077e-01 -1.13048506e+00 3.02004248e-01 1.27244484e+00 -7.59130597e-01 -6.44975007e-01 1.55901462e-01 9.80705738e-01 -1.98832467e-01 -3.04071665e-01 3.82379383e-01 6.35349393e-01 -1.20264494e+00 8.35363388e-01 -5.27804434e-01 2.14387134e-01 -4.91216749e-01 5.29553294e-02 -1.16216743e+00 -6.48899853e-01 -6.59532309e-01 -3.66962016e-01 9.76914942e-01 1.58999249e-01 -3.96328181e-01 6.25536501e-01 5.85222244e-01 -3.88458550e-01 -7.35713780e-01 -1.31228065e+00 -6.87771559e-01 -7.01110363e-02 -4.62380230e-01 3.69870842e-01 3.38292837e-01 -4.83090192e-01 2.57085204e-01 -6.97502553e-01 4.22984481e-01 6.87382817e-01 -1.43908724e-01 2.90421933e-01 -1.18123579e+00 -5.12216926e-01 -4.62222546e-01 -4.42139953e-01 -8.54110479e-01 -4.63939637e-01 -6.02220893e-01 1.07455336e-01 -1.60726166e+00 3.53355974e-01 -3.38541716e-01 -8.34838331e-01 5.26720285e-01 -3.70117337e-01 2.95329869e-01 7.61525333e-02 5.04793465e-01 -1.01985097e+00 6.95774615e-01 1.33436191e+00 -1.21612974e-01 -4.21561897e-01 8.31075087e-02 -7.88893521e-01 1.57201126e-01 7.45303810e-01 -3.58769894e-01 -6.06878877e-01 -8.29123318e-01 3.41202050e-01 -1.27468482e-01 5.20413339e-01 -1.59330952e+00 5.59626579e-01 1.09895773e-01 4.21572179e-01 -2.90771365e-01 2.26722613e-01 -4.52003509e-01 1.99633613e-02 4.77075756e-01 -1.34741634e-01 -1.08101815e-01 2.71869928e-01 6.61613405e-01 -2.15133242e-02 2.80061122e-02 1.08386052e+00 9.13256109e-02 -6.70719922e-01 4.76855040e-01 3.81961418e-03 -1.27995461e-01 7.26660848e-01 -5.92279583e-02 -5.21087289e-01 -1.03061706e-01 -1.00302529e+00 1.25399217e-01 4.58602726e-01 2.79684216e-01 1.03560793e+00 -1.07929218e+00 -6.64646089e-01 1.88926771e-01 2.45615259e-01 -1.68589279e-01 8.76713395e-01 7.68626690e-01 -1.83178037e-01 1.81339025e-01 -6.30724847e-01 -2.93865323e-01 -1.17334783e+00 7.05998540e-01 4.99955326e-01 -1.83127135e-01 -5.09229600e-01 1.44566238e+00 1.09305844e-01 1.17051266e-01 4.91648197e-01 -3.07370853e-02 -7.20353872e-02 -4.87912893e-02 8.03652942e-01 1.76506519e-01 4.85840887e-01 -2.22182855e-01 -2.81509519e-01 3.21794748e-01 -3.89835626e-01 -1.93596408e-01 1.44964850e+00 -3.45366240e-01 -1.47461772e-01 2.56742954e-01 6.26814663e-01 -3.76968741e-01 -1.68673873e+00 -4.16791320e-01 -3.11375111e-01 -4.57083732e-01 3.09237003e-01 -1.54662013e+00 -1.07349670e+00 1.11438048e+00 1.36835754e+00 -4.52025861e-01 1.69121587e+00 -1.12416953e-01 8.00997794e-01 2.23315552e-01 3.83829445e-01 -1.01386511e+00 5.52800179e-01 3.78786504e-01 1.04515135e+00 -1.02363861e+00 -2.52092689e-01 2.72895724e-01 -6.35741770e-01 7.28814662e-01 7.70365417e-01 9.05033275e-02 3.48567069e-01 -1.86639309e-01 3.50168943e-02 5.09696901e-02 -8.78449798e-01 -2.14808404e-01 2.75804520e-01 7.88756669e-01 -2.98810657e-02 -9.07801092e-02 -1.72986686e-01 7.53693342e-01 -1.31615326e-01 3.88265550e-01 2.90612191e-01 6.97318971e-01 -6.81801498e-01 -8.83679807e-01 -6.05767476e-04 7.29411185e-01 -4.14378732e-01 -4.84081179e-01 1.34810537e-01 -1.18673043e-02 3.25623661e-01 8.15397859e-01 -2.08827592e-02 -6.35194778e-01 3.19760084e-01 4.24802959e-01 6.39138997e-01 -2.42643252e-01 -6.35894358e-01 -2.29763880e-01 -3.01872820e-01 -4.75888610e-01 -2.72911251e-01 -6.53348327e-01 -9.56828415e-01 -2.43581057e-01 -3.38220716e-01 -3.10842693e-01 5.83889961e-01 6.93592608e-01 5.92302680e-01 6.03148520e-01 5.71213961e-01 -3.09906691e-01 -5.55229604e-01 -9.92147386e-01 -3.58828783e-01 4.40321833e-01 5.76496422e-01 -6.22811139e-01 -3.56144845e-01 2.81209052e-01]
[11.87967300415039, -1.800777554512024]
2e195d75-7398-4df4-812e-15be28ab226e
stockemotions-discover-investor-emotions-for
2301.09279
null
https://arxiv.org/abs/2301.09279v2
https://arxiv.org/pdf/2301.09279v2.pdf
StockEmotions: Discover Investor Emotions for Financial Sentiment Analysis and Multivariate Time Series
There has been growing interest in applying NLP techniques in the financial domain, however, resources are extremely limited. This paper introduces StockEmotions, a new dataset for detecting emotions in the stock market that consists of 10,000 English comments collected from StockTwits, a financial social media platform. Inspired by behavioral finance, it proposes 12 fine-grained emotion classes that span the roller coaster of investor emotion. Unlike existing financial sentiment datasets, StockEmotions presents granular features such as investor sentiment classes, fine-grained emotions, emojis, and time series data. To demonstrate the usability of the dataset, we perform a dataset analysis and conduct experimental downstream tasks. For financial sentiment/emotion classification tasks, DistilBERT outperforms other baselines, and for multivariate time series forecasting, a Temporal Attention LSTM model combining price index, text, and emotion features achieves the best performance than using a single feature.
['Soyeon Caren Han', 'Josiah Poon', 'Hoyoul Luis Youn', 'Jean Lee']
2023-01-23
null
null
null
null
['emotion-classification', 'emotion-classification']
['computer-vision', 'natural-language-processing']
[-8.98605168e-01 -3.71899486e-01 -2.40420029e-01 -6.87439322e-01 -2.48602912e-01 -6.95546627e-01 6.74696982e-01 1.57488808e-02 -4.17151093e-01 5.33786654e-01 5.34205258e-01 -3.54821414e-01 3.52825582e-01 -7.43951857e-01 -1.78529024e-01 -2.21006572e-01 -2.60927707e-01 -1.28361508e-01 -7.46228695e-02 -5.33720434e-01 6.49098814e-01 2.32210591e-01 -1.44504011e+00 5.60325027e-01 3.65636230e-01 1.89434814e+00 -4.99698848e-01 4.58986223e-01 -7.44153261e-01 1.67163384e+00 -7.47171700e-01 -9.00336087e-01 1.77949622e-01 -1.33639783e-01 -4.54684108e-01 -1.74468562e-01 -1.95222750e-01 -3.19473177e-01 -7.54331648e-02 9.51865077e-01 2.71100104e-01 1.91377848e-01 3.80775034e-01 -1.32695401e+00 -1.24356341e+00 7.09286034e-01 -3.71670634e-01 6.20726466e-01 4.65548821e-02 -2.35433802e-01 1.51222932e+00 -1.13118708e+00 4.26031828e-01 1.01386976e+00 7.65195727e-01 1.02632798e-01 -5.96865475e-01 -7.13911951e-01 4.22813594e-01 5.99418730e-02 -3.85360301e-01 -8.26337561e-02 8.91279578e-01 -5.72997451e-01 1.55398417e+00 8.91048908e-02 8.77521813e-01 1.25101948e+00 5.40059566e-01 8.88330102e-01 1.21770024e+00 -7.65582845e-02 4.00800645e-01 5.57643890e-01 6.27718091e-01 3.47544193e-01 -2.40859732e-01 -3.65974121e-02 -8.57033670e-01 -3.14625502e-01 2.16719925e-01 4.33431268e-01 2.22154200e-01 5.39449275e-01 -9.56927121e-01 1.19763005e+00 1.61236152e-01 4.28843826e-01 -7.32135057e-01 4.70966846e-02 8.65180373e-01 9.13261652e-01 1.50512564e+00 6.97919726e-01 -1.14815402e+00 -6.88395917e-01 -7.06875384e-01 -2.56563188e-03 1.34324443e+00 4.92755443e-01 3.64826649e-01 5.66927135e-01 1.27802804e-01 8.60242844e-01 9.63813439e-02 3.44392687e-01 1.05991483e+00 -5.19298255e-01 1.98829561e-01 7.54233718e-01 1.31133854e-01 -1.20175946e+00 -6.00992680e-01 -2.33911037e-01 -2.40461886e-01 1.64414480e-01 -2.16039270e-01 -8.46338570e-01 -4.10617173e-01 1.17831945e+00 6.23761751e-02 1.99478060e-01 2.02561498e-01 7.56840348e-01 1.01107335e+00 8.03566635e-01 1.24681681e-01 -1.03820361e-01 1.39095676e+00 -1.07585347e+00 -8.57105851e-01 -6.41195104e-03 4.87680495e-01 -6.74046576e-01 8.47118020e-01 3.37338090e-01 -9.18680668e-01 -2.87872821e-01 -8.04642737e-01 -2.54200138e-02 -1.27769268e+00 1.19836051e-02 1.02642369e+00 5.52928209e-01 -9.63532150e-01 7.42481947e-01 -6.14530087e-01 3.87245491e-02 5.54439984e-02 -5.01477122e-02 -3.21642272e-02 8.92597497e-01 -1.46178341e+00 1.01244223e+00 4.13506031e-02 -1.58093441e-02 -7.87181128e-03 -6.49277747e-01 -8.83920789e-01 2.35918716e-01 -1.87944636e-01 -3.76508497e-02 1.39782786e+00 -1.28514731e+00 -1.93832099e+00 6.04367495e-01 1.06673971e-01 -7.85245657e-01 1.53265446e-01 -3.91858488e-01 -8.66248250e-01 -8.66114125e-02 -9.68909189e-02 1.77052811e-01 7.77926564e-01 -2.88699210e-01 -6.89702690e-01 -4.26973663e-02 -9.61707160e-02 -1.82257965e-01 -1.11584520e+00 7.60691643e-01 -1.00374836e-02 -9.08999920e-01 -5.80341399e-01 -5.17751813e-01 -1.86094761e-01 -5.90758562e-01 -1.81553707e-01 -4.61247593e-01 8.47669601e-01 -7.42715895e-01 1.18993235e+00 -2.25428152e+00 -4.88589466e-01 1.35844201e-01 1.21309169e-01 -1.64469391e-01 -1.44164935e-01 4.64749277e-01 -3.99627328e-01 3.25158060e-01 3.96718681e-01 -4.75529939e-01 6.59875214e-01 -1.68277487e-01 -1.02898705e+00 1.49051324e-01 2.96366304e-01 1.21526575e+00 -5.35262644e-01 1.67574361e-02 -1.10166676e-01 3.12941343e-01 -3.92712414e-01 1.16766267e-01 -3.10929567e-01 -2.69345164e-01 -5.83060622e-01 8.43509734e-01 1.66833103e-01 -4.54774141e-01 -4.15260226e-01 5.91730215e-02 -5.53157032e-01 3.83623093e-01 -4.53493685e-01 7.31998920e-01 -6.09265447e-01 9.52598631e-01 -5.63709140e-01 -7.22653151e-01 1.28807855e+00 5.49493909e-01 7.93481469e-01 -9.50457752e-01 5.44516921e-01 2.02898636e-01 -2.54749835e-01 -5.88807285e-01 8.19880724e-01 -4.52406436e-01 -5.74836791e-01 7.60695577e-01 -4.69678752e-02 1.37556881e-01 1.45069674e-01 -6.76208809e-02 1.08552301e+00 -2.28314418e-02 1.15941398e-01 -6.48117289e-02 1.28268618e-02 -6.52196333e-02 5.13857126e-01 4.43925917e-01 -2.68230468e-01 2.19978601e-01 7.73888230e-01 -8.62270534e-01 -6.40910685e-01 -4.31232542e-01 4.22342345e-02 1.55269694e+00 -3.97558242e-01 -4.04773265e-01 -2.57611424e-01 -5.63253224e-01 2.54510522e-01 6.43672585e-01 -1.03606403e+00 1.94185853e-01 2.84102894e-02 -9.90702629e-01 2.65173107e-01 6.98386550e-01 4.87949699e-01 -1.56915188e+00 -8.59510541e-01 4.34854031e-01 1.94282323e-01 -1.22787595e+00 -4.05576199e-01 2.61851549e-01 -6.25822663e-01 -8.31349254e-01 -7.23513305e-01 -3.37630004e-01 -6.96127415e-02 -3.53050441e-01 1.57366717e+00 -2.53906071e-01 -1.26993015e-01 5.22178173e-01 -6.75769210e-01 -9.72970843e-01 2.49982074e-01 -1.77898463e-02 -1.48426995e-01 2.13211998e-01 9.36761677e-01 -3.31093222e-01 -3.38771313e-01 -1.71271190e-01 -6.64029956e-01 -5.22517025e-01 1.66547626e-01 5.97919285e-01 -4.72654849e-02 -6.76555932e-02 1.13746047e+00 -8.29922318e-01 1.19181037e+00 -1.03861189e+00 -5.63012302e-01 8.36820006e-02 -5.57668447e-01 -3.51893932e-01 6.92260504e-01 -5.49173653e-01 -1.22702324e+00 -6.44237995e-01 -6.29710853e-02 -2.18522206e-01 1.33319944e-01 1.20020592e+00 6.83640540e-01 1.34908885e-01 1.16367444e-01 2.20119413e-02 -2.21795365e-01 -4.57371056e-01 1.80426523e-01 6.30310178e-01 2.62377458e-03 -1.80541381e-01 1.27446637e-01 2.23185122e-01 -6.38031721e-01 -7.02343345e-01 -1.04710519e+00 -3.63582283e-01 -3.46398242e-02 -3.14749241e-01 8.65876496e-01 -8.94702852e-01 -1.07714605e+00 7.23359227e-01 -8.77181530e-01 -2.39398122e-01 -2.82163471e-01 6.48055255e-01 -2.88369775e-01 -1.49901628e-01 -1.53541386e+00 -1.07125294e+00 -5.96285403e-01 -6.20855272e-01 6.18733048e-01 6.39393628e-02 -4.77733165e-01 -1.34093797e+00 1.67136639e-01 -1.09172776e-01 9.28013146e-01 5.21362245e-01 6.23018563e-01 -1.33559048e+00 1.44696742e-01 -5.51627219e-01 -3.77355702e-02 3.76885295e-01 6.68166354e-02 2.94627577e-01 -8.08844149e-01 2.41128817e-01 4.03955430e-01 -7.90299118e-01 1.09703362e+00 2.38761649e-01 9.56894517e-01 -4.10352796e-01 3.33752126e-01 2.87324101e-01 1.06672609e+00 6.51841104e-01 2.17273399e-01 8.45225930e-01 2.59926796e-01 5.75842261e-01 4.50742245e-01 1.15805089e+00 7.61762083e-01 -1.31401315e-01 1.76190734e-01 -2.37960368e-01 9.09596503e-01 2.01710194e-01 8.73377621e-01 1.15641141e+00 1.10193834e-01 -9.28404108e-02 -6.63527489e-01 3.22754353e-01 -1.80938995e+00 -1.06898010e+00 1.81868970e-01 1.41409659e+00 7.25245714e-01 2.06243232e-01 2.82002687e-01 -1.43157884e-01 3.85517418e-01 4.45161998e-01 -8.76909554e-01 -8.48537564e-01 -3.50887656e-01 2.33785316e-01 3.05285186e-01 5.27252071e-02 -1.27329171e+00 9.78976250e-01 6.45934868e+00 2.63300031e-01 -1.59163558e+00 2.18478739e-02 1.02134836e+00 -5.52169561e-01 -3.13147932e-01 -4.72966880e-01 -6.46970749e-01 7.91843116e-01 1.41598070e+00 -3.60338479e-01 1.24660626e-01 1.18564677e+00 1.45622894e-01 2.41206467e-01 -8.89628232e-01 5.36765695e-01 9.39297006e-02 -1.39338863e+00 -4.80071515e-01 -1.34921879e-01 6.25646234e-01 4.73211139e-01 5.00281274e-01 8.82481754e-01 5.90354025e-01 -7.09593236e-01 9.27383780e-01 7.04818368e-01 1.10432163e-01 -8.00362349e-01 9.27015722e-01 -1.85772888e-02 -1.02109933e+00 -3.03094447e-01 -1.76544517e-01 -4.75644708e-01 1.53194875e-01 6.24082148e-01 -2.46655673e-01 7.06216022e-02 1.28392303e+00 1.51992941e+00 -4.04021591e-01 5.94551325e-01 2.06347391e-01 6.63454950e-01 -3.34505066e-02 -5.97102225e-01 5.84422350e-01 -3.64545137e-01 -7.04476163e-02 1.38199162e+00 5.86461484e-01 3.14287275e-01 -1.07978053e-01 4.74552512e-01 -1.88043162e-01 4.24658567e-01 -3.75833809e-01 -7.40486085e-01 6.96762502e-02 1.38332570e+00 -6.94602311e-01 -7.22195029e-01 -1.00721574e+00 8.72059047e-01 3.64993393e-01 5.63162923e-01 -9.81835783e-01 -6.66565418e-01 8.63585293e-01 -7.11596727e-01 5.70106149e-01 1.61491662e-01 -4.00563717e-01 -1.58783484e+00 -7.59476349e-02 -7.97341108e-01 5.86374998e-01 -8.29305828e-01 -1.91665339e+00 9.72012043e-01 -6.21137321e-01 -1.08246911e+00 -4.36033010e-01 -1.01482809e+00 -1.00209057e+00 6.47176862e-01 -1.66055954e+00 -6.09844923e-01 1.97261095e-01 4.20252204e-01 5.28714359e-01 -6.63592100e-01 7.58635104e-01 1.98195651e-01 -7.48547733e-01 8.98637697e-02 4.37948816e-02 4.51894730e-01 6.48022950e-01 -1.38847625e+00 6.93482339e-01 3.78737211e-01 -4.53024954e-02 5.05776405e-01 5.51762819e-01 -5.70135295e-01 -1.08810461e+00 -9.11829293e-01 1.09527981e+00 -3.35493982e-01 1.56637371e+00 -1.25569940e-01 -7.71662414e-01 9.43131983e-01 8.22739542e-01 -1.01172194e-01 1.18813825e+00 2.18414724e-01 -5.01664221e-01 1.30690664e-01 -1.05738425e+00 5.61145604e-01 1.46847352e-01 -7.04632521e-01 -7.96226323e-01 1.37167230e-01 8.56059968e-01 5.09169176e-02 -1.29038668e+00 8.41007307e-02 5.31927764e-01 -1.16632199e+00 6.52769864e-01 -1.01602900e+00 9.28332806e-01 4.75246876e-01 -2.52360869e-02 -1.50871813e+00 -1.63384736e-01 -5.12388587e-01 -2.66317844e-01 1.23568189e+00 7.05622137e-01 -1.13432956e+00 5.59126258e-01 1.24525154e+00 -2.93970369e-02 -9.32909012e-01 -5.05216062e-01 -5.50187111e-01 2.21342579e-01 -5.76708496e-01 8.26574862e-01 1.35155892e+00 6.76530540e-01 1.75811589e-01 -4.80656534e-01 -3.67175967e-01 -4.19276655e-02 6.80064023e-01 2.32154101e-01 -1.17512465e+00 -3.50028098e-01 -9.50266182e-01 -4.99377511e-02 -5.02665758e-01 6.43121183e-01 -5.44208586e-01 -4.00097877e-01 -1.19531941e+00 -2.64243573e-01 5.61803579e-02 -8.26315105e-01 6.22275651e-01 7.96948150e-02 2.89403796e-01 3.09276015e-01 -4.52809595e-02 -6.60791278e-01 7.37887144e-01 6.94278002e-01 -6.63965195e-02 -1.13101989e-01 -1.68812603e-01 -6.37675107e-01 9.77179766e-01 9.13665771e-01 -1.81173161e-01 3.89187410e-02 -5.20174205e-02 7.83585250e-01 1.78836837e-01 -4.36245129e-02 -3.38999599e-01 2.19024122e-01 -6.05293997e-02 6.51802957e-01 -8.23929667e-01 6.15084648e-01 -6.93982542e-01 -3.14898700e-01 2.88205802e-01 -5.51288009e-01 6.91124797e-01 3.14359844e-01 2.12142006e-01 -8.60437274e-01 -9.02240798e-02 2.25886881e-01 -2.81455576e-01 -8.68280411e-01 4.22808081e-01 -7.63657749e-01 2.05306426e-01 8.21265280e-01 9.64528397e-02 -4.58753020e-01 -6.00632846e-01 -7.19834447e-01 2.49947533e-01 -1.92402139e-01 8.08338881e-01 5.60168386e-01 -1.23231280e+00 -4.89144921e-01 5.81741557e-02 -1.18779734e-01 -9.90116715e-01 4.28500213e-02 6.76189482e-01 -1.00230426e-01 2.93498695e-01 -7.58409575e-02 2.92106122e-01 -7.44274318e-01 5.89513063e-01 3.90229344e-01 -1.22737370e-01 -4.04524118e-01 9.21006739e-01 -9.07808095e-02 -3.08897108e-01 1.18793696e-01 -7.37940788e-01 -6.42061234e-01 9.23743486e-01 6.68256640e-01 2.57203758e-01 -3.70715037e-02 -3.15610051e-01 -1.65789783e-01 3.29043448e-01 -1.11399986e-01 -1.86273620e-01 2.00772071e+00 -2.05382220e-02 -3.74251157e-01 1.05921924e+00 1.34393346e+00 -2.81795375e-02 -1.10653031e+00 -2.61454135e-01 6.89397156e-01 -1.28397882e-01 1.07680924e-01 -7.67078280e-01 -1.36812162e+00 6.57510638e-01 -1.39236916e-02 1.07994533e+00 1.02241004e+00 -2.19853416e-01 1.20362079e+00 5.28819084e-01 -3.85998897e-02 -1.47880065e+00 2.90186346e-01 1.12046301e+00 8.97562504e-01 -1.29774201e+00 -4.29087251e-01 3.00774723e-01 -1.21736801e+00 1.33150446e+00 4.51691419e-01 -4.06683773e-01 1.31047213e+00 5.05229890e-01 6.39911175e-01 -4.72214967e-01 -1.41993093e+00 -1.05668172e-01 2.20698863e-01 -3.20511878e-01 7.41927624e-01 -1.84963182e-01 5.39502874e-02 1.32639432e+00 -5.88814557e-01 6.02196306e-02 5.42297065e-01 7.20454633e-01 -3.39531898e-01 -6.31912947e-01 -1.01606781e-02 7.51357555e-01 -1.27883923e+00 -4.30589497e-01 -5.15349507e-01 6.30913377e-01 -3.58382821e-01 9.75498378e-01 6.10644937e-01 -3.97664100e-01 3.07136506e-01 3.56802911e-01 -5.69603145e-01 -3.92524540e-01 -1.32467401e+00 3.42814475e-02 3.02649498e-01 -5.04721463e-01 -4.82158542e-01 -7.51791656e-01 -1.20390832e+00 -2.85369307e-01 1.03761427e-01 4.01151031e-01 7.33886123e-01 1.00212538e+00 5.52247584e-01 5.70289254e-01 1.09889436e+00 -7.85108268e-01 -4.56702858e-01 -9.65053260e-01 -8.41975570e-01 3.37219924e-01 5.30923665e-01 -3.57957572e-01 -6.85849369e-01 -4.10178266e-02]
[4.440478324890137, 4.2959771156311035]
d6d94415-74b2-4a57-99bf-09909af9da24
transition-based-dependency-parsing-with
null
null
https://aclanthology.org/P16-2001
https://aclanthology.org/P16-2001.pdf
Transition-based dependency parsing with topological fields
null
['Dani{\\"e}l de Kok', 'Erhard Hinrichs']
2016-08-01
null
null
null
acl-2016-8
['transition-based-dependency-parsing']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.355665683746338, 3.693446636199951]
21b77cf0-06e2-46a1-ad42-0bdf1c4a7fe3
assigning-confidence-to-molecular-property
2102.11439
null
https://arxiv.org/abs/2102.11439v1
https://arxiv.org/pdf/2102.11439v1.pdf
Assigning Confidence to Molecular Property Prediction
Introduction: Computational modeling has rapidly advanced over the last decades, especially to predict molecular properties for chemistry, material science and drug design. Recently, machine learning techniques have emerged as a powerful and cost-effective strategy to learn from existing datasets and perform predictions on unseen molecules. Accordingly, the explosive rise of data-driven techniques raises an important question: What confidence can be assigned to molecular property predictions and what techniques can be used for that purpose? Areas covered: In this work, we discuss popular strategies for predicting molecular properties relevant to drug design, their corresponding uncertainty sources and methods to quantify uncertainty and confidence. First, our considerations for assessing confidence begin with dataset bias and size, data-driven property prediction and feature design. Next, we discuss property simulation via molecular docking, and free-energy simulations of binding affinity in detail. Lastly, we investigate how these uncertainties propagate to generative models, as they are usually coupled with property predictors. Expert opinion: Computational techniques are paramount to reduce the prohibitive cost and timing of brute-force experimentation when exploring the enormous chemical space. We believe that assessing uncertainty in property prediction models is essential whenever closed-loop drug design campaigns relying on high-throughput virtual screening are deployed. Accordingly, considering sources of uncertainty leads to better-informed experimental validations, more reliable predictions and to more realistic expectations of the entire workflow. Overall, this increases confidence in the predictions and designs and, ultimately, accelerates drug design.
['Alán Aspuru-Guzik', 'Vincent A. Voelz', 'Seyone Chithrananda', 'Naruki Yoshikawa', 'Matteo Aldeghi', 'Riley J. Hickman', 'Matthew F. D. Hurley', 'Robert Pollice', 'AkshatKumar Nigam']
2021-02-23
null
null
null
null
['molecular-docking']
['medical']
[ 4.76557165e-01 -1.90584272e-01 -3.35737973e-01 -1.52918652e-01 -8.96854758e-01 -8.11027706e-01 4.71869290e-01 8.70468974e-01 -3.03449422e-01 1.52237856e+00 -1.32132590e-01 -7.03538537e-01 -5.30642092e-01 -5.82783043e-01 -7.36283779e-01 -9.10648406e-01 -5.24912365e-02 6.41219139e-01 -1.07125103e-01 7.75697827e-02 4.57621515e-01 9.35972214e-01 -1.29450011e+00 -1.30334243e-01 1.18768692e+00 8.37913156e-01 1.33741036e-01 2.54059821e-01 1.08199954e-01 5.75214624e-01 -3.56761068e-01 -4.27961349e-01 -2.70282030e-01 -3.62477720e-01 -6.01923287e-01 -3.31472844e-01 -3.82555053e-02 1.78570263e-02 4.21160519e-01 7.89476633e-01 6.82291567e-01 2.19220579e-01 9.51095760e-01 -5.91474414e-01 -2.62241364e-01 2.62130469e-01 -1.74564347e-01 -4.50373255e-02 5.67784131e-01 4.77025092e-01 6.76242709e-01 -8.30450714e-01 5.30993164e-01 7.56330371e-01 3.84316742e-01 2.79799223e-01 -1.55800831e+00 -5.77609777e-01 -3.18996571e-02 1.91486686e-01 -1.45770955e+00 -3.31687957e-01 4.46552187e-01 -7.43513525e-01 1.00813913e+00 4.62647945e-01 6.20320737e-01 1.04661870e+00 4.84996676e-01 2.43825465e-01 9.38072205e-01 -2.22295016e-01 8.28328311e-01 3.44928324e-01 -4.53117073e-01 3.67338657e-01 6.40095890e-01 4.26990271e-01 -5.52742064e-01 -6.45762622e-01 2.31948271e-01 -9.11188647e-02 -3.22779745e-01 -4.53452945e-01 -7.96050191e-01 8.63032818e-01 9.38712209e-02 -1.52765645e-03 -6.34025574e-01 7.38365203e-02 1.96589559e-01 -3.37280571e-01 3.43313694e-01 9.78401184e-01 -8.41727257e-01 -3.02497327e-01 -9.43764806e-01 3.43299598e-01 7.10281432e-01 5.90004265e-01 5.74244857e-01 -2.14938391e-02 3.92954975e-01 3.16953748e-01 4.89008009e-01 3.58297855e-01 -4.88105677e-02 -4.53194529e-01 -3.54202129e-02 4.02502835e-01 3.46175820e-01 -6.09772444e-01 -4.94311243e-01 -5.97849309e-01 -4.71205562e-01 2.44230315e-01 2.80897379e-01 -1.78115353e-01 -8.61324847e-01 1.52233160e+00 4.83248323e-01 -2.94866040e-02 8.53435770e-02 4.95564818e-01 3.95842701e-01 4.20004606e-01 7.83402681e-01 -6.40398741e-01 1.20819533e+00 3.83244790e-02 -3.03087682e-01 1.94417998e-01 7.31969774e-01 -9.11812603e-01 4.74754095e-01 7.65316010e-01 -8.07482302e-01 -1.84330344e-01 -1.05250096e+00 3.95216256e-01 -4.42104906e-01 -1.78971618e-01 9.14367914e-01 9.96158481e-01 -1.88221797e-01 1.25169230e+00 -9.86429334e-01 5.01988940e-02 6.07228756e-01 7.69684017e-01 -3.47813249e-01 8.04865062e-02 -1.14531481e+00 1.16539383e+00 5.66569448e-01 1.04018755e-01 -7.90851235e-01 -1.11974192e+00 -6.34306669e-01 -1.05492860e-01 3.46533388e-01 -8.45780909e-01 8.30105782e-01 -3.22139949e-01 -1.48984337e+00 3.01613629e-01 -1.71544224e-01 -3.86380911e-01 4.04904336e-01 -4.56557907e-02 -4.35762793e-01 -2.14358121e-01 -1.97809145e-01 3.86943638e-01 2.66263425e-01 -1.02643073e+00 -3.25492531e-01 -4.23649997e-01 -4.44259584e-01 -1.25130918e-02 4.27707464e-01 -1.69341013e-01 2.89818585e-01 -1.83191597e-01 -9.78340209e-02 -8.37579250e-01 -6.00989044e-01 -3.57259512e-01 -5.18216074e-01 -2.23651081e-02 2.76478469e-01 -3.07765514e-01 1.12386084e+00 -1.63050640e+00 1.63287863e-01 6.16137207e-01 1.74795046e-01 2.38269687e-01 2.53122419e-01 8.53858054e-01 -2.15177506e-01 3.41163754e-01 -2.01442480e-01 4.30889606e-01 -3.36162180e-01 -2.31102958e-01 -2.37138629e-01 4.89823163e-01 4.55874532e-01 8.44402730e-01 -8.79461467e-01 -4.92928959e-02 4.29359257e-01 7.96482801e-01 -6.65375292e-01 8.74640569e-02 -6.87668741e-01 9.21066225e-01 -7.87891507e-01 8.27478886e-01 5.29983401e-01 -4.09632534e-01 5.29786170e-01 -2.05301106e-01 -2.96925038e-01 4.04117286e-01 -8.22477877e-01 1.18970668e+00 -2.74598867e-01 1.07264044e-02 -7.76120067e-01 -5.99289119e-01 1.04240847e+00 1.88771293e-01 5.36131561e-01 -5.25201976e-01 2.01870427e-01 5.48234582e-01 3.75916779e-01 -2.40111783e-01 7.47919008e-02 -5.07271945e-01 2.96525240e-01 -4.43345169e-03 -5.30746020e-02 -3.35342944e-01 8.84750187e-02 -9.04606655e-02 7.26820171e-01 4.25587505e-01 5.91120541e-01 -2.32157648e-01 5.64050138e-01 2.00846121e-01 3.62209976e-01 1.95104674e-01 8.84401202e-02 2.39164159e-01 5.45146525e-01 -4.29477572e-01 -1.01846409e+00 -8.08738291e-01 -6.35353446e-01 5.05436242e-01 -2.17751861e-01 -4.37324464e-01 -4.09460396e-01 -4.10146624e-01 1.41064271e-01 8.29757333e-01 -7.15949595e-01 -3.39841247e-01 -7.89066181e-02 -1.07288122e+00 1.42046109e-01 2.62144566e-01 -3.41724306e-01 -7.80679524e-01 -2.02014223e-01 7.58760452e-01 4.77384329e-01 -6.34393334e-01 1.53740823e-01 6.24290884e-01 -7.98170149e-01 -1.44511652e+00 -6.03304744e-01 -6.82155713e-02 5.49176991e-01 -3.96730304e-01 8.35420489e-01 -1.45161241e-01 -5.41464388e-01 1.27269933e-02 -1.45260364e-01 -8.86762083e-01 -6.21716261e-01 -1.48217216e-01 1.80595428e-01 -4.19296801e-01 4.75425482e-01 -7.32735455e-01 -9.44634616e-01 3.27487737e-01 -6.75026953e-01 -3.46653014e-01 7.48747945e-01 6.89652562e-01 1.01002133e+00 4.30099107e-03 7.75349200e-01 -1.10356343e+00 6.54810429e-01 -4.73893106e-01 -9.33017910e-01 3.87827158e-01 -1.02053022e+00 2.86837667e-01 5.47061265e-01 -2.83953130e-01 -8.34837377e-01 2.66725272e-01 -4.96741652e-01 1.21484816e-01 -2.11547375e-01 8.62908483e-01 -3.25401276e-01 -2.59889781e-01 9.44565594e-01 1.30588278e-01 1.44526064e-02 -4.88113344e-01 1.38922542e-01 4.34056610e-01 -8.68021771e-02 -8.06031525e-01 4.56320524e-01 2.32669428e-01 5.18671334e-01 -7.79600382e-01 -5.44924855e-01 -1.79422989e-01 -5.07240713e-01 4.95604277e-02 6.34777069e-01 -7.40576982e-01 -1.25734866e+00 -1.51057780e-01 -9.23865438e-01 -4.70310897e-02 -1.74268022e-01 8.19930255e-01 -5.98301709e-01 3.62903982e-01 2.27349848e-02 -8.91543388e-01 -5.06236672e-01 -1.69900000e+00 8.59433174e-01 2.91674197e-01 -6.81183755e-01 -1.01494420e+00 2.83816159e-01 4.54582185e-01 3.46268326e-01 6.02234125e-01 1.20033538e+00 -7.47630537e-01 -7.33409107e-01 -2.96012789e-01 1.00432530e-01 7.71621540e-02 3.66571069e-01 2.13807270e-01 -9.35479939e-01 -8.09760317e-02 -3.79300833e-01 -4.31868970e-01 5.35637319e-01 7.01761961e-01 1.14702296e+00 -5.84607981e-02 -4.52162713e-01 3.13498318e-01 1.32117307e+00 6.62981689e-01 6.55319273e-01 1.30361572e-01 3.40729952e-01 5.33374906e-01 8.26913834e-01 6.85993969e-01 -1.56412259e-01 5.69074273e-01 4.86803979e-01 1.17379650e-01 3.71511817e-01 -4.93534952e-01 -1.39637381e-01 8.75389799e-02 -4.29697514e-01 -2.64549464e-01 -7.77339756e-01 -7.70161226e-02 -1.43769419e+00 -9.36654866e-01 -1.56811118e-01 2.66672444e+00 1.14078689e+00 2.17925012e-01 1.93416074e-01 -6.13737665e-02 2.99894392e-01 -4.16588902e-01 -9.04128015e-01 -3.16511005e-01 2.15256424e-03 3.95516664e-01 5.04803061e-01 4.94926423e-01 -7.13307500e-01 7.70399868e-01 6.97375345e+00 9.54217136e-01 -1.29036295e+00 -5.05206347e-01 1.01420712e+00 8.91437307e-02 -6.05095506e-01 2.39941493e-01 -9.43524659e-01 4.60654020e-01 1.18991899e+00 -4.30341512e-01 1.48146302e-01 7.29718387e-01 5.19361794e-01 -4.87915158e-01 -1.23451364e+00 6.69151068e-01 -4.99561310e-01 -1.87049329e+00 8.61172080e-02 3.81760150e-01 6.38363659e-01 -1.85259461e-01 1.42850161e-01 -1.82389542e-01 1.54761255e-01 -1.36720550e+00 1.90649897e-01 5.17297685e-01 9.07758772e-01 -1.07428360e+00 6.48507833e-01 2.54650950e-01 -7.02980578e-01 3.79083306e-01 -4.48753685e-01 1.60208985e-01 1.15875795e-01 9.12451684e-01 -1.20443857e+00 6.00332260e-01 -7.42364451e-02 3.54390889e-01 -2.44003847e-01 1.28104460e+00 -1.11908928e-01 4.28253651e-01 -4.17431235e-01 -4.24325377e-01 3.90779413e-02 -3.68322849e-01 6.97379187e-02 8.43903661e-01 1.61088035e-01 1.70682296e-01 -1.31317928e-01 9.81346607e-01 2.01604664e-01 2.37603992e-01 -2.88133562e-01 -5.05748451e-01 5.54493785e-01 9.02045906e-01 -6.11429036e-01 4.79371585e-02 -1.17724784e-01 4.52249169e-01 -1.66761130e-02 2.29663521e-01 -8.27629149e-01 -3.65231782e-02 8.80224407e-01 3.28673542e-01 -9.69169103e-03 -6.99843243e-02 -1.64050050e-02 -6.63752198e-01 -4.50652868e-01 -8.54542315e-01 2.16428205e-01 -4.14347589e-01 -1.01050806e+00 5.20804465e-01 7.17726946e-02 -1.05791771e+00 -1.73594400e-01 -8.21475148e-01 -2.92754441e-01 1.07574487e+00 -1.46680653e+00 -6.09101176e-01 3.42071742e-01 1.34752974e-01 1.55019954e-01 5.75580001e-02 1.07312882e+00 2.07204834e-01 -4.27372992e-01 1.90694466e-01 5.09286404e-01 -6.95443928e-01 6.27590656e-01 -8.92444432e-01 7.37721696e-02 1.84678629e-01 -8.37568119e-02 8.40903044e-01 1.04750454e+00 -9.64199722e-01 -1.43872738e+00 -8.66320550e-01 3.99886966e-01 -5.76385498e-01 8.19969475e-01 -5.80193438e-02 -8.13507080e-01 6.30393401e-02 -4.86157775e-01 -1.03457078e-01 1.44723058e+00 3.04232150e-01 -1.29569054e-01 3.13046962e-01 -1.21210217e+00 3.84100914e-01 5.40919900e-01 -2.93857932e-01 -8.75482261e-02 5.83544970e-01 4.28651959e-01 -3.40620637e-01 -1.24983466e+00 5.50225675e-01 7.51651108e-01 -7.60444522e-01 8.99523258e-01 -1.01895928e+00 1.46642596e-01 -3.94824803e-01 8.69134848e-04 -1.16398299e+00 -2.15512648e-01 -5.20411432e-01 1.80116609e-01 6.64441109e-01 1.09024835e+00 -5.93100727e-01 1.11775124e+00 1.16795301e+00 1.78162172e-01 -1.17828918e+00 -8.43905091e-01 -7.40315914e-01 1.89302415e-01 -4.07405198e-01 5.49526453e-01 5.98281920e-01 1.88411042e-01 2.27826148e-01 -3.82278234e-01 1.66483112e-02 4.79393870e-01 -5.76437190e-02 4.65718210e-01 -1.42875540e+00 -5.13716698e-01 -2.95655042e-01 -6.09576464e-01 -4.59847987e-01 -3.51730525e-01 -4.60034668e-01 -5.08134544e-01 -1.21257567e+00 2.31156752e-01 -4.42425817e-01 -9.20104831e-02 6.58203810e-02 -2.13732123e-01 -8.75278041e-02 -2.53794640e-01 2.58817710e-02 -3.18165690e-01 5.09650052e-01 1.16482973e+00 -9.25794914e-02 -4.63947833e-01 4.16103661e-01 -8.76343906e-01 4.18362260e-01 7.46743679e-01 -4.99345392e-01 -5.50584614e-01 5.54820478e-01 7.63919294e-01 1.37382289e-02 9.02222767e-02 -7.16309726e-01 1.76363543e-03 -5.67981899e-01 7.15960741e-01 -5.35856307e-01 2.76486427e-01 -8.58780384e-01 7.67414629e-01 6.20659292e-01 -7.34178498e-02 -6.30098224e-01 4.69257116e-01 7.28103101e-01 -5.50495572e-02 -1.86162189e-01 8.18912268e-01 -1.42916320e-02 -1.54984891e-01 5.53464532e-01 -3.73001993e-01 -3.96423131e-01 1.12586606e+00 -4.12625074e-01 -1.33862533e-03 -1.89700812e-01 -1.11533511e+00 -5.99155240e-02 2.96751469e-01 -1.50014507e-02 4.78190899e-01 -8.16736400e-01 -5.10674477e-01 -8.62874761e-02 2.65260935e-01 -1.18560873e-01 4.15839761e-01 8.32285583e-01 -5.67849100e-01 7.62519777e-01 1.37023747e-01 -4.30866480e-01 -1.25045979e+00 6.80600822e-01 3.26732248e-01 -2.13408500e-01 1.56860903e-01 7.44367540e-01 7.12388679e-02 1.55657642e-02 -6.89692199e-02 -8.23609233e-02 3.00950520e-02 6.58985376e-02 4.70325857e-01 3.48405242e-01 4.54748034e-01 -3.01193357e-01 -6.90884411e-01 4.22069937e-01 -3.43421817e-01 3.43291044e-01 1.47585404e+00 3.01077187e-01 2.16145232e-01 5.63383587e-02 9.40798402e-01 1.21271007e-01 -1.20789659e+00 2.99336761e-01 1.87727332e-01 -3.75298589e-01 9.02281329e-02 -1.30897939e+00 -2.40475208e-01 6.23741448e-01 4.40476716e-01 -8.20677355e-02 7.23487437e-01 1.95492711e-03 1.45898432e-01 3.92215729e-01 4.98709768e-01 -9.17397201e-01 -3.40179622e-01 -2.83358432e-03 8.67534459e-01 -1.27564323e+00 6.69366658e-01 -5.54026067e-01 -4.93511826e-01 1.19721735e+00 1.00845128e-01 4.93497491e-01 6.14538789e-01 7.73776844e-02 -4.61490601e-01 -3.39774668e-01 -7.31074393e-01 -9.11474973e-02 3.95186126e-01 6.30394876e-01 8.04498434e-01 1.27181798e-01 -4.26542133e-01 3.94859493e-01 8.03617388e-02 1.15791947e-01 1.89940542e-01 9.43444312e-01 -5.96624911e-01 -1.79436231e+00 -3.06321591e-01 4.67239231e-01 -4.19541359e-01 -2.91064113e-01 -6.57675922e-01 7.45332301e-01 8.89852867e-02 8.49906564e-01 -6.09282792e-01 -1.86561763e-01 2.98416764e-01 1.35114733e-02 6.72970355e-01 -5.80234408e-01 -2.19679862e-01 1.96819603e-01 2.49157101e-01 -2.59111881e-01 -2.06933066e-01 -5.30797899e-01 -1.15821075e+00 -3.01840276e-01 -9.58807230e-01 6.96637392e-01 1.02679276e+00 9.38003480e-01 7.48598516e-01 3.15927356e-01 5.95283031e-01 -5.91222703e-01 -4.95886415e-01 -5.12430489e-01 -5.55331707e-01 -2.33900085e-01 4.40338301e-03 -7.00296104e-01 -2.72430420e-01 -2.14526564e-01]
[5.126319885253906, 5.51729679107666]
944ff6d5-4dff-4372-8951-7734f55e0028
crackle-detection-in-lung-sounds-using
2104.14921
null
https://arxiv.org/abs/2104.14921v1
https://arxiv.org/pdf/2104.14921v1.pdf
Crackle Detection In Lung Sounds Using Transfer Learning And Multi-Input Convolitional Neural Networks
Large annotated lung sound databases are publicly available and might be used to train algorithms for diagnosis systems. However, it might be a challenge to develop a well-performing algorithm for small non-public data, which have only a few subjects and show differences in recording devices and setup. In this paper, we use transfer learning to tackle the mismatch of the recording setup. This allows us to transfer knowledge from one dataset to another dataset for crackle detection in lung sounds. In particular, a single input convolutional neural network (CNN) model is pre-trained on a source domain using ICBHI 2017, the largest publicly available database of lung sounds. We use log-mel spectrogram features of respiratory cycles of lung sounds. The pre-trained network is used to build a multi-input CNN model, which shares the same network architecture for respiratory cycles and their corresponding respiratory phases. The multi-input model is then fine-tuned on the target domain of our self-collected lung sound database for classifying crackles and normal lung sounds. Our experimental results show significant performance improvements of 9.84% (absolute) in F-score on the target domain using the multi-input CNN model based on transfer learning for crackle detection in adventitious lung sound classification task.
['Franz Pernkopf', 'Truc Nguyen']
2021-04-30
null
null
null
null
['sound-classification']
['audio']
[ 1.74179107e-01 -1.29703134e-01 4.75489907e-02 -1.42831355e-01 -1.21619153e+00 -5.17039001e-01 -1.46427780e-01 -1.22671381e-01 -1.53638110e-01 2.44415939e-01 9.45956483e-02 -4.28561091e-01 1.39065906e-01 -9.35108721e-01 -7.71423876e-01 -6.62016392e-01 1.52129248e-01 4.60287571e-01 7.60495305e-01 2.48122200e-01 -3.34086299e-01 3.62411439e-01 -1.26455235e+00 7.21113026e-01 2.84386426e-01 1.02920043e+00 4.24157143e-01 1.29384315e+00 3.80786397e-02 7.90904522e-01 -7.11833537e-01 3.60033638e-03 -1.18009284e-01 -5.60163260e-01 -7.82156408e-01 -3.10912669e-01 7.04269826e-01 -3.16432834e-01 -4.36489016e-01 5.11043787e-01 1.14096165e+00 -3.07063639e-01 5.55225015e-01 -8.51862490e-01 -3.63876343e-01 6.50171220e-01 2.16869503e-01 6.39559627e-01 5.08819558e-02 2.55608261e-01 9.97625530e-01 -8.09480488e-01 -2.84142385e-04 6.65500522e-01 1.09085310e+00 8.54191065e-01 -7.49795020e-01 -9.48690593e-01 -7.61797607e-01 1.28354177e-01 -1.10354519e+00 -1.78812087e-01 7.08819926e-01 -6.71929240e-01 9.00570512e-01 3.16483200e-01 5.98698795e-01 1.11621499e+00 1.99561529e-02 3.09385836e-01 6.17632329e-01 -2.30196729e-01 -5.21837547e-02 1.67947456e-01 -2.52838910e-01 6.71791255e-01 9.43794027e-02 6.00504577e-02 -3.97751749e-01 -1.50812104e-01 6.71260953e-01 -5.20472415e-02 -4.25152034e-01 1.30975470e-01 -1.40207732e+00 7.73579180e-01 5.60990095e-01 7.98794687e-01 -2.26657078e-01 2.34759092e-01 3.52159858e-01 2.06622645e-01 2.51592696e-01 5.42131543e-01 -6.04901314e-01 -1.38409063e-02 -9.08235431e-01 1.39957339e-01 9.73814011e-01 5.07254422e-01 1.62672117e-01 -1.08424919e-02 -4.69434470e-01 1.14800882e+00 2.85696745e-01 6.29600883e-01 8.03154588e-01 -6.77986026e-01 6.31316900e-01 1.11647509e-01 -4.19634521e-01 -6.71014309e-01 -6.17519498e-01 -6.23371422e-01 -8.22006702e-01 -3.52507412e-01 5.19698560e-01 -1.36713967e-01 -5.79416096e-01 1.43581402e+00 2.39832669e-01 6.62391007e-01 -6.77174851e-02 7.98330128e-01 1.32804203e+00 5.51246941e-01 -3.46475430e-02 1.10570617e-01 1.32543075e+00 -1.15005708e+00 -3.73127371e-01 -8.09734687e-03 5.51254869e-01 -7.66194105e-01 1.19387197e+00 2.73497522e-01 -8.63154709e-01 -8.27997446e-01 -1.05876720e+00 7.37728924e-02 -2.05670148e-01 4.02626336e-01 -1.48746639e-01 6.77433968e-01 -9.19974744e-01 7.20206320e-01 -8.34635317e-01 -2.98803419e-01 4.71388698e-01 5.03806829e-01 3.71421017e-02 1.97505474e-01 -1.34914255e+00 4.22791332e-01 1.37914270e-01 -1.53362051e-01 -1.17801380e+00 -1.04139793e+00 -3.09789002e-01 1.59796193e-01 1.76635772e-01 -7.31727958e-01 1.64867854e+00 -2.76257217e-01 -1.54103017e+00 9.08363521e-01 5.21388531e-01 -4.99991059e-01 3.83993715e-01 -4.66275923e-02 -5.48772514e-01 3.52638632e-01 -7.35166669e-02 2.89036512e-01 9.76060152e-01 -8.40189517e-01 -7.25680709e-01 8.80408734e-02 -4.74403292e-01 -3.12688500e-01 -5.23769140e-01 -5.66493906e-03 -4.28739280e-01 -8.05272639e-01 -2.81129241e-01 -8.68566215e-01 2.99188852e-01 -1.07896984e-01 -3.55634600e-01 -2.86027372e-01 8.25938761e-01 -7.85100758e-01 1.34613240e+00 -2.13504243e+00 -2.95290977e-01 -1.12712003e-01 3.93605739e-01 6.52381480e-01 -1.28057629e-01 1.84833527e-01 -2.05495238e-01 1.74264535e-01 -3.73995662e-01 -1.72126472e-01 -2.99145073e-01 1.27272001e-02 -1.27649218e-01 3.87871742e-01 2.76829630e-01 7.20035791e-01 -8.90192568e-01 -6.50408745e-01 2.39820495e-01 5.15327156e-01 -7.48448014e-01 8.36409092e-01 -7.43739828e-02 8.20239186e-01 -4.08555716e-01 5.01969635e-01 3.11993122e-01 -2.06527025e-01 -3.39380443e-01 -4.20965493e-01 1.73091531e-01 7.06257522e-01 -6.50191903e-01 1.48203337e+00 -9.18119907e-01 5.56636512e-01 -1.65259749e-01 -8.64911616e-01 7.08700359e-01 1.00962937e+00 8.61236453e-01 -1.04207389e-01 2.82508492e-01 5.01073182e-01 3.37282062e-01 -9.82629716e-01 -3.99352580e-01 -3.12997222e-01 1.39684692e-01 5.76799691e-01 3.50431353e-01 -5.80779970e-01 -2.99126446e-01 -4.13360685e-01 1.39017570e+00 -6.28574073e-01 -7.63387457e-02 4.60148267e-02 7.46833742e-01 -2.93975830e-01 1.94768980e-01 5.48885107e-01 -3.18754315e-01 1.10994411e+00 4.72028420e-04 -2.50014484e-01 -9.17759597e-01 -1.03288198e+00 -3.53009552e-01 1.17966342e+00 -7.34507620e-01 -1.46778151e-01 -8.62923026e-01 -7.52169549e-01 -5.27125858e-02 2.85379887e-01 -4.74683553e-01 -2.80824751e-01 -9.06134188e-01 -5.87160408e-01 1.30333006e+00 8.09657812e-01 4.82120156e-01 -1.54351377e+00 -6.15528882e-01 3.85993868e-01 -3.14939320e-01 -1.23995352e+00 -5.97087801e-01 4.26986009e-01 -6.99409008e-01 -1.24116051e+00 -7.54004419e-01 -9.78058100e-01 1.07598811e-01 -5.40109947e-02 1.15101063e+00 3.04479271e-01 -5.65249562e-01 4.59162325e-01 -2.43862838e-01 -6.65346265e-01 -8.56440842e-01 4.38997149e-01 -7.45348260e-02 4.41996641e-02 2.78356504e-02 -5.14244914e-01 -7.07414031e-01 2.68782616e-01 -6.39759481e-01 -3.75818759e-01 5.20593584e-01 8.36036861e-01 6.13934994e-01 1.17255794e-02 9.16702211e-01 -7.14637697e-01 6.32133842e-01 -7.55899847e-01 -1.82216093e-01 5.44847138e-02 -3.03690851e-01 -3.42336327e-01 8.64843667e-01 -6.03951514e-01 -5.23946822e-01 1.56109557e-01 -5.75059414e-01 -7.65973449e-01 -2.05240682e-01 1.33395821e-01 -1.57626290e-02 1.94299385e-01 7.59203970e-01 1.27216041e-01 -2.07668856e-01 -6.38793230e-01 4.15494060e-03 1.14368856e+00 1.03481925e+00 -3.69033664e-01 8.44959021e-01 1.86847746e-01 -2.15698153e-01 -6.27586126e-01 -9.74546432e-01 -7.01783359e-01 -7.50670850e-01 -9.49697793e-02 1.35316837e+00 -7.98921525e-01 -4.09914136e-01 7.42438138e-01 -9.10419285e-01 -5.69339275e-01 -3.94724756e-01 6.59339845e-01 -4.89669859e-01 2.93231048e-02 -5.80378115e-01 -4.61326897e-01 -8.14024210e-01 -9.93612051e-01 1.04308093e+00 -3.05841747e-03 -7.54262954e-02 -9.28920984e-01 5.26216388e-01 6.34193122e-01 6.91960871e-01 3.02857701e-02 9.34657454e-01 -1.19305444e+00 -2.57966101e-01 -3.25334877e-01 3.77404653e-02 7.86167026e-01 4.10271347e-01 4.38791774e-02 -1.53622532e+00 -2.59604871e-01 2.51341969e-01 -5.65696180e-01 9.30194676e-01 3.01578611e-01 1.72636294e+00 -8.33714753e-02 -1.99233994e-01 6.22195661e-01 8.99826109e-01 1.38196275e-01 1.72080114e-01 -1.40230909e-01 8.39269936e-01 3.69052202e-01 1.71536222e-01 1.52127519e-01 1.18485756e-01 5.41260004e-01 3.71485651e-01 -2.57881254e-01 -7.09651709e-01 -2.54333913e-01 2.79014647e-01 1.40667844e+00 1.16119869e-01 -3.25538993e-01 -1.21226597e+00 7.73038268e-01 -1.07065248e+00 -8.49057913e-01 -1.38468206e-01 2.09108901e+00 1.05331385e+00 -7.15886578e-02 4.24949348e-01 4.16346043e-01 7.02501237e-01 -6.94722310e-02 -3.78460199e-01 -2.94238962e-02 4.26369131e-01 7.07997084e-01 2.35866860e-01 9.96461287e-02 -1.36491501e+00 4.37474638e-01 6.29373503e+00 6.37612581e-01 -1.70732820e+00 5.73748350e-01 2.89153337e-01 -8.01526308e-02 1.75820753e-01 -7.42261589e-01 -6.22736394e-01 5.43972909e-01 1.41383374e+00 2.33873352e-01 3.23119223e-01 7.17508852e-01 -8.56863633e-02 5.50055206e-01 -1.31253684e+00 8.42180729e-01 8.45682621e-03 -1.02448738e+00 -4.63213354e-01 -2.41698518e-01 3.82535607e-01 6.90094948e-01 1.24343611e-01 3.73845786e-01 -3.13456416e-01 -1.17763221e+00 2.83195376e-01 2.17488259e-01 1.13081002e+00 -4.68000650e-01 7.62371004e-01 4.09659564e-01 -1.38872802e+00 -7.00322464e-02 -3.67793292e-01 5.92540801e-01 -3.26168150e-01 4.58529770e-01 -1.64958072e+00 2.46717528e-01 1.06328773e+00 5.16624391e-01 -6.31218672e-01 1.22886193e+00 1.42790928e-01 1.43986452e+00 -3.94137055e-01 5.04691042e-02 -1.97424829e-01 5.19071519e-01 3.07274640e-01 1.45551920e+00 6.20895743e-01 -2.49684408e-01 5.56712039e-02 9.26453948e-01 -4.26995039e-01 1.88716933e-01 -6.96874022e-01 -2.08730415e-01 4.48289394e-01 1.30661488e+00 -3.96535844e-01 -1.76026985e-01 -5.70920110e-01 4.87679124e-01 2.69234404e-02 -1.75521910e-01 -1.12717128e+00 -2.41458520e-01 1.95934474e-01 3.84934723e-01 4.24358219e-01 4.15862262e-01 2.53630728e-02 -8.40533853e-01 -2.42371276e-01 -1.00612772e+00 5.16472578e-01 -5.20572543e-01 -1.41988432e+00 8.85670960e-01 -1.35325029e-01 -1.39676642e+00 -2.78915405e-01 -5.57520807e-01 -1.17160273e+00 7.21698940e-01 -1.44047916e+00 -1.08837771e+00 -5.42824507e-01 7.10963607e-01 6.94000602e-01 -3.82014900e-01 9.55354095e-01 8.12093556e-01 -5.15969872e-01 7.75131464e-01 -1.45003080e-01 3.84436339e-01 9.46942866e-01 -1.44366610e+00 3.89120489e-01 3.81680757e-01 3.40295374e-01 8.76895711e-02 2.23878071e-01 -4.37223166e-01 -1.07897413e+00 -1.74718153e+00 8.44395101e-01 -7.10083306e-01 7.06155360e-01 -1.73133656e-01 -1.32038677e+00 3.70191514e-01 -3.55152190e-02 6.31666839e-01 9.88378465e-01 -5.18924117e-01 -2.37411872e-01 -2.31008947e-01 -9.99144793e-01 -1.18744738e-01 5.17703235e-01 -8.57743382e-01 -6.61309242e-01 6.45103037e-01 8.78846943e-01 -4.86801237e-01 -1.21119535e+00 6.27251685e-01 4.89422202e-01 -6.55951202e-01 1.19639194e+00 -3.49620879e-01 3.60880822e-01 -1.36731490e-01 -7.75762424e-02 -1.21612406e+00 -2.05024749e-01 -3.22872311e-01 -9.65480506e-02 1.24012864e+00 4.18060929e-01 -3.79813522e-01 6.02298677e-01 -1.73830971e-01 -3.09913993e-01 -1.07997394e+00 -8.65319848e-01 -6.01007462e-01 5.04655838e-01 -7.14569867e-01 6.28172159e-01 6.02052331e-01 -4.67527002e-01 3.93327892e-01 -1.38927609e-01 1.98770776e-01 1.28586233e-01 1.07775174e-01 5.47031164e-01 -1.23249900e+00 -6.94957018e-01 -2.14696050e-01 1.04274452e-01 -6.27884626e-01 -3.22625749e-02 -1.27685809e+00 4.03755814e-01 -1.28959322e+00 1.27969548e-01 -5.52347600e-01 -7.77887166e-01 5.52439094e-01 -1.64307505e-01 4.85337287e-01 7.69825419e-03 3.32077950e-01 -1.88118637e-01 1.36800572e-01 1.46273851e+00 -2.71390468e-01 -1.62538350e-01 6.07512236e-01 -1.63963795e-01 7.44178832e-01 9.67653215e-01 -8.90336752e-01 -3.66831928e-01 -3.03522706e-01 -2.68308610e-01 2.22400367e-01 6.20395541e-01 -1.55721438e+00 1.74698308e-01 3.19357336e-01 2.32085109e-01 -8.11219692e-01 1.58749461e-01 -7.97506154e-01 4.31987643e-02 8.30988884e-01 -5.50387502e-01 -3.51613641e-01 2.07260072e-01 5.16561449e-01 -1.70869097e-01 -4.00164992e-01 9.80944395e-01 2.11321469e-03 1.24119565e-01 4.23189610e-01 -3.89712334e-01 3.72170955e-01 6.98923767e-01 2.51451164e-01 -1.23405121e-01 -1.41039550e-01 -7.88113356e-01 -1.96298435e-01 -1.75143421e-01 4.21086103e-01 4.77455914e-01 -1.30453289e+00 -8.15652192e-01 3.08178186e-01 1.12294398e-01 2.08648786e-01 1.02827489e-01 8.68233621e-01 -6.98989630e-01 6.30677223e-01 -7.61907995e-02 -8.94743621e-01 -1.28009701e+00 3.93698215e-01 8.27363193e-01 -4.38016057e-01 -6.80445731e-01 1.10776174e+00 1.92378014e-01 -7.32862771e-01 4.12966341e-01 -1.00213063e+00 -2.28671014e-01 -1.79412246e-01 1.90939814e-01 2.31845737e-01 4.96252984e-01 -3.61053765e-01 -4.69394356e-01 6.85369790e-01 3.85248631e-01 2.34756649e-01 1.26390791e+00 4.07029510e-01 1.36848748e-01 5.00760972e-01 1.57819498e+00 1.57735690e-01 -7.81581581e-01 -2.48155639e-01 -2.67418683e-01 7.64191821e-02 6.48281351e-02 -8.44047487e-01 -1.48633075e+00 1.34999967e+00 9.91576076e-01 2.89519221e-01 1.23122025e+00 3.25636744e-01 1.42291415e+00 5.32429755e-01 -6.81058615e-02 -6.46525919e-01 5.80538154e-01 2.99210429e-01 1.07246900e+00 -1.10260749e+00 -5.11520982e-01 -3.37969691e-01 -3.71473670e-01 1.20761669e+00 6.70340002e-01 -1.75998673e-01 9.52045083e-01 3.77776563e-01 3.05542976e-01 -3.91481817e-01 -7.48291910e-01 -5.90029582e-02 5.46379268e-01 6.22207344e-01 6.72637343e-01 1.38031542e-01 2.49243885e-01 1.10529876e+00 -4.91337210e-01 1.00720733e-01 6.89563230e-02 5.83382726e-01 -4.73050624e-01 -9.80636716e-01 -4.52266276e-01 6.85921013e-01 -9.80584502e-01 1.44582801e-02 -4.14964497e-01 5.74609935e-01 6.08974814e-01 9.46272969e-01 -5.25336862e-02 -6.44505799e-01 3.58501524e-01 2.11414799e-01 4.06651825e-01 -1.07897449e+00 -1.19152379e+00 1.60154447e-01 -7.48190731e-02 -2.52860099e-01 -2.43964970e-01 -3.15264016e-01 -1.37160683e+00 4.32511508e-01 -4.92846817e-01 2.94313859e-02 3.93865258e-01 6.90255523e-01 1.40828742e-02 1.14230549e+00 7.63270140e-01 -5.39891779e-01 -8.13098669e-01 -1.08844423e+00 -2.82503635e-01 3.52149248e-01 5.35794258e-01 -4.45098639e-01 -5.10075629e-01 2.82666653e-01]
[14.5599365234375, 3.906963586807251]
320493e7-0ec5-4aea-9102-300da1addd4d
real-time-convolutional-neural-networks-for
1710.07557
null
http://arxiv.org/abs/1710.07557v1
http://arxiv.org/pdf/1710.07557v1.pdf
Real-time Convolutional Neural Networks for Emotion and Gender Classification
In this paper we propose an implement a general convolutional neural network (CNN) building framework for designing real-time CNNs. We validate our models by creating a real-time vision system which accomplishes the tasks of face detection, gender classification and emotion classification simultaneously in one blended step using our proposed CNN architecture. After presenting the details of the training procedure setup we proceed to evaluate on standard benchmark sets. We report accuracies of 96% in the IMDB gender dataset and 66% in the FER-2013 emotion dataset. Along with this we also introduced the very recent real-time enabled guided back-propagation visualization technique. Guided back-propagation uncovers the dynamics of the weight changes and evaluates the learned features. We argue that the careful implementation of modern CNN architectures, the use of the current regularization methods and the visualization of previously hidden features are necessary in order to reduce the gap between slow performances and real-time architectures. Our system has been validated by its deployment on a Care-O-bot 3 robot used during RoboCup@Home competitions. All our code, demos and pre-trained architectures have been released under an open-source license in our public repository.
['Paul Plöger', 'Matias Valdenegro-Toro', 'Octavio Arriaga']
2017-10-20
null
null
null
null
['gender-prediction']
['computer-vision']
[-2.89709747e-01 2.37874061e-01 4.23006296e-01 -7.61140764e-01 1.81270957e-01 -1.53613970e-01 6.62653565e-01 -1.37313500e-01 -7.99350023e-01 3.67107809e-01 -2.08914727e-01 -6.77581728e-02 -4.88556214e-02 -3.21068913e-01 -5.77798843e-01 -5.80057144e-01 -6.07731044e-01 5.41064024e-01 -1.90755352e-01 -5.99466980e-01 3.10183197e-01 6.22934699e-01 -2.00560522e+00 4.52226907e-01 1.40717894e-01 1.25968242e+00 -3.09903055e-01 1.04681838e+00 2.57850081e-01 8.51417363e-01 -6.12544715e-01 -6.59654081e-01 2.82364786e-01 1.20135501e-01 -7.26819515e-01 -2.48165846e-01 5.50619125e-01 -1.60649464e-01 -5.79037070e-02 8.11391175e-01 7.58223236e-01 4.09445800e-02 4.20637161e-01 -1.70993555e+00 -4.92101997e-01 5.08774698e-01 -3.33890319e-01 5.81531934e-02 4.35239337e-02 3.16894352e-01 5.82686007e-01 -1.00690806e+00 7.83815920e-01 1.31986654e+00 9.75143909e-01 8.65887582e-01 -1.05016017e+00 -7.07331002e-01 -1.01064995e-01 2.47029245e-01 -1.16153455e+00 -7.81821370e-01 6.95788503e-01 -4.21579063e-01 1.36441994e+00 1.59439221e-01 7.12408960e-01 1.51383138e+00 1.80817768e-02 4.94868010e-01 9.88742769e-01 -5.46016634e-01 1.43291816e-01 4.10339445e-01 1.31688341e-01 1.14012694e+00 -1.31757826e-01 1.67137668e-01 -7.90382743e-01 -1.28566101e-02 3.32678109e-01 -3.15125823e-01 1.17249355e-01 -3.62452000e-01 -6.92109704e-01 7.76982665e-01 4.58965898e-01 3.08167934e-01 -1.83163762e-01 5.59561193e-01 8.61548424e-01 4.23695445e-01 6.96025074e-01 3.36419851e-01 -5.96108735e-01 -3.14015716e-01 -8.50783348e-01 3.21165264e-01 9.37784553e-01 6.59363925e-01 4.58537161e-01 6.31122664e-02 -1.64004892e-01 8.12584639e-01 2.78253555e-01 -6.51780963e-02 3.56907576e-01 -1.17983758e+00 -1.24663096e-02 5.95299304e-01 -6.14523981e-03 -1.18281627e+00 -7.27642715e-01 -3.19923103e-01 -5.65650582e-01 7.68620014e-01 4.66151059e-01 -3.29555362e-01 -9.28539574e-01 1.59848452e+00 2.85201967e-01 -4.02903967e-02 -6.75249398e-02 9.26588476e-01 1.09126306e+00 1.25311956e-01 1.48768321e-01 3.48392069e-01 1.36114299e+00 -8.83087158e-01 -5.16354024e-01 7.70274922e-02 5.08017480e-01 -5.54357469e-01 1.05801272e+00 6.56325579e-01 -9.17222798e-01 -5.73276579e-01 -1.21418476e+00 -1.56090185e-01 -7.69311547e-01 7.37297475e-01 9.29777265e-01 8.90945733e-01 -1.36344850e+00 9.72202480e-01 -9.58030879e-01 -6.74515367e-01 7.37338781e-01 7.27690339e-01 -4.52760369e-01 4.67668414e-01 -6.26111150e-01 9.10733104e-01 1.15740150e-01 6.36206567e-01 -9.23561752e-01 -4.82265055e-01 -6.24210656e-01 -2.59478055e-02 -1.60984527e-02 -3.51573765e-01 1.19858468e+00 -1.23575175e+00 -1.58204269e+00 1.38520694e+00 3.48783523e-01 -5.34009337e-01 8.85057867e-01 -3.13206732e-01 -1.78045213e-01 -8.51648226e-02 -3.40148360e-01 1.12513912e+00 8.40070724e-01 -9.07599330e-01 -4.48553801e-01 -3.54965776e-01 1.08702749e-01 -2.60227025e-01 -6.31845832e-01 2.15966552e-01 -3.64844769e-01 -2.03925028e-01 -4.36911404e-01 -9.33224082e-01 -3.06533743e-02 2.41481096e-01 -3.68847936e-01 -1.52382985e-01 9.25680876e-01 -5.19556880e-01 5.94361246e-01 -2.33497691e+00 1.66901812e-01 2.35815812e-02 3.27807456e-01 3.04899454e-01 -6.72857650e-03 2.22224131e-01 -4.43349987e-01 -2.19817623e-01 -5.61737223e-03 -1.02331209e+00 4.04329509e-01 -1.33631332e-02 5.40193282e-02 6.13559902e-01 3.72778684e-01 7.27131128e-01 -4.09482211e-01 -1.82340145e-01 2.06597790e-01 9.52168643e-01 -5.40589154e-01 1.99427187e-01 -6.69252798e-02 3.09720635e-01 2.96876520e-01 7.61144340e-01 7.04815388e-01 2.66032904e-01 6.04972839e-02 -1.86874717e-01 -2.31944993e-01 -2.94299811e-01 -9.63366807e-01 1.71940541e+00 -3.48473310e-01 1.02774084e+00 3.07759225e-01 -1.05286098e+00 1.15857065e+00 1.42219082e-01 1.64226905e-01 -8.57029021e-01 5.38962662e-01 9.09026712e-02 -1.55132413e-01 -7.88108706e-01 6.68770194e-01 2.95171320e-01 1.57738298e-01 1.05796948e-01 7.12181747e-01 2.73905426e-01 3.40832412e-01 -4.20605503e-02 8.83896828e-01 5.32622814e-01 -2.15323955e-01 -4.70079005e-01 3.67367655e-01 8.38482101e-03 2.84463286e-01 4.65150476e-01 -4.29812223e-01 5.82218647e-01 9.27962422e-01 -9.23394084e-01 -1.00000048e+00 -3.36980551e-01 -1.08055405e-01 1.57363570e+00 -6.06512904e-01 -5.09829581e-01 -9.21056449e-01 -6.92154050e-01 -1.70613870e-01 4.24279988e-01 -1.39515650e+00 -1.41067058e-01 -4.59511727e-01 -8.64220738e-01 9.28716540e-01 3.40905994e-01 3.78998071e-01 -1.48967278e+00 -1.31300664e+00 -2.19711685e-03 4.17015374e-01 -8.56511772e-01 4.04537857e-01 6.23369038e-01 -4.83299404e-01 -1.24401724e+00 -5.71085215e-01 -6.87349617e-01 4.56495583e-01 -6.09740496e-01 1.24079728e+00 3.00453812e-01 -8.39554131e-01 3.50758225e-01 -2.31296420e-01 -7.11274743e-01 -5.97947501e-02 1.96286425e-01 -1.44362628e-01 -1.00952066e-01 5.69885254e-01 -7.79391527e-01 -5.62317193e-01 -1.41384885e-01 -4.00241196e-01 -6.53430494e-03 1.79062828e-01 9.21966434e-01 -3.08047440e-02 -3.25642079e-01 1.31543249e-01 -8.18320751e-01 7.51172364e-01 -3.13151330e-01 -8.26954365e-01 -1.36624733e-02 -4.67215955e-01 -5.24847768e-02 2.26945728e-01 -3.57302308e-01 -8.43232274e-01 3.16959292e-01 -4.28442925e-01 -4.81200069e-01 -3.59965265e-01 1.45103514e-01 2.60918736e-01 -1.59144893e-01 8.27969968e-01 -3.45854878e-01 5.42434119e-02 -4.20067817e-01 3.35119694e-01 5.39370358e-01 5.33940017e-01 -5.64763069e-01 2.24730194e-01 4.51230228e-01 -2.38323817e-03 -5.81522346e-01 -2.49803424e-01 3.57078947e-02 -7.23382592e-01 -4.89404142e-01 8.50160122e-01 -8.73998702e-01 -1.49263763e+00 7.02318132e-01 -1.13267255e+00 -6.56832874e-01 -9.12376493e-02 -1.08543225e-01 -5.61169624e-01 -2.60414243e-01 -5.66890001e-01 -9.12179589e-01 -5.94532490e-01 -1.19118679e+00 1.03803790e+00 3.58914733e-01 -2.24059343e-01 -9.45164382e-01 1.44824862e-01 4.09080163e-02 8.10263455e-01 4.88913178e-01 6.04025006e-01 -5.88898778e-01 6.75274134e-02 -5.95131516e-02 -3.92335594e-01 2.49131918e-01 -5.35630643e-01 5.53527832e-01 -1.61794031e+00 -1.27718031e-01 -2.06832021e-01 -8.24958146e-01 1.08618462e+00 2.69193798e-01 1.26518357e+00 2.36499757e-02 -6.32560328e-02 8.37595522e-01 1.24595881e+00 -6.64621443e-02 4.90863204e-01 6.33955956e-01 5.91489732e-01 9.96885121e-01 3.45016211e-01 6.29574060e-01 3.25736582e-01 6.11082137e-01 7.59469807e-01 -2.97041625e-01 1.08429581e-01 1.90326869e-01 9.54356715e-02 1.16036639e-01 -3.73279303e-01 2.88335174e-01 -1.10354936e+00 3.90793502e-01 -1.94611382e+00 -6.37848079e-01 -2.07763731e-01 1.60848308e+00 4.62474525e-01 1.04370125e-01 3.64677519e-01 1.42900690e-01 2.84732997e-01 -1.43833771e-01 -2.36881077e-01 -9.23992813e-01 1.97330907e-01 5.18880785e-01 2.18843296e-01 3.53943557e-01 -1.25748789e+00 9.54096913e-01 5.83850145e+00 4.27850515e-01 -1.65571845e+00 2.10236460e-01 8.65412831e-01 -3.83117706e-01 6.47462368e-01 -4.68348622e-01 -5.83552182e-01 1.28390014e-01 1.08906794e+00 4.26714033e-01 5.09201229e-01 1.25691211e+00 -1.78862121e-02 -1.54056475e-01 -1.18974805e+00 1.07728755e+00 1.90443099e-01 -1.28423810e+00 -6.16486251e-01 -1.84677929e-01 1.16449349e-01 2.07658738e-01 2.90753782e-01 5.95295310e-01 1.12995796e-01 -1.37772298e+00 1.05297577e+00 4.38976318e-01 8.07138026e-01 -9.78764415e-01 7.01924503e-01 -1.15606733e-01 -7.14300215e-01 -2.73555666e-01 -3.56042743e-01 -2.52353519e-01 -5.78478813e-01 3.81046176e-01 -7.81510770e-01 1.62086621e-01 1.31749070e+00 6.04626596e-01 -1.07497573e+00 7.46737063e-01 7.58193135e-02 3.15396667e-01 -2.65518099e-01 -1.26326740e-01 1.10465810e-01 2.30674133e-01 1.96496680e-01 1.65442348e+00 -3.42090577e-02 -4.62529391e-01 -3.21853310e-01 9.57840025e-01 4.35399823e-02 1.91419479e-02 -4.53447998e-01 1.37712546e-02 -1.14094190e-01 1.83539200e+00 -1.05115831e+00 -7.09640533e-02 -2.16148514e-02 9.83543694e-01 6.70920670e-01 2.07868814e-01 -9.56103027e-01 -6.15892649e-01 7.31865525e-01 -2.54919469e-01 3.92980367e-01 -1.67987034e-01 -3.17217678e-01 -9.09203708e-01 -2.44253501e-02 -7.53025532e-01 6.10756464e-02 -8.80432189e-01 -9.43817198e-01 1.12823415e+00 -4.97808214e-03 -6.89648569e-01 -2.23369777e-01 -1.01615930e+00 -5.83813131e-01 5.76961637e-01 -1.24465108e+00 -1.31027567e+00 -5.27763963e-01 5.72196543e-01 2.81391978e-01 -5.13348699e-01 1.17291462e+00 4.99757022e-01 -7.90878892e-01 7.05945253e-01 -4.31227297e-01 1.34956747e-01 5.57467997e-01 -1.29860377e+00 3.29082131e-01 3.94144833e-01 1.03746086e-01 5.29628575e-01 1.02191389e+00 -1.23620421e-01 -1.11607015e+00 -6.42802417e-01 6.83709979e-01 -2.53111273e-01 5.90667188e-01 -1.10109532e+00 -6.30913496e-01 7.03326285e-01 6.52614415e-01 1.04388453e-01 4.38151300e-01 3.04932296e-01 -4.25911754e-01 -1.41357437e-01 -1.30021417e+00 4.10529107e-01 9.49396193e-01 -2.82323480e-01 -2.82180011e-01 3.49907905e-01 2.53669083e-01 -5.49060762e-01 -3.58132392e-01 2.85008788e-01 8.06159794e-01 -1.56285024e+00 6.29297197e-01 -4.76640731e-01 7.39322841e-01 1.33726344e-01 3.66896056e-02 -1.13383317e+00 9.11979675e-02 -4.09607381e-01 1.09312870e-01 1.31811464e+00 3.85596067e-01 -3.99073213e-01 8.72519076e-01 4.91845638e-01 1.17999516e-01 -9.14224446e-01 -8.70971084e-01 -9.91119966e-02 -2.64445961e-01 -6.75808549e-01 2.60114521e-01 7.81017423e-01 -9.17375237e-02 5.02362289e-02 -4.16533470e-01 -2.25365654e-01 3.42207640e-01 -3.07059169e-01 8.70179594e-01 -1.18827546e+00 -1.78138480e-01 -4.96335447e-01 -5.95169783e-01 1.30241498e-01 2.69917756e-01 -6.11001849e-01 -1.43412743e-02 -9.20675039e-01 1.23975374e-01 -2.65290499e-01 -1.60514608e-01 9.13779974e-01 4.57264543e-01 7.72801161e-01 2.08166867e-01 -2.13469580e-01 -6.70142949e-01 4.67684209e-01 3.86195570e-01 -7.29584880e-03 -7.37812817e-02 -3.83943290e-01 -4.62053835e-01 8.06782067e-01 7.30507851e-01 -7.03963637e-01 -7.19885975e-02 -3.27699035e-01 3.80186349e-01 -5.54019213e-01 7.69833088e-01 -1.25300741e+00 1.53790265e-01 3.54669243e-01 7.25693524e-01 -2.21300542e-01 7.17896938e-01 -8.43480051e-01 -5.51880375e-02 5.36746681e-01 -3.49367410e-01 1.21016078e-01 6.43517673e-01 -6.67584166e-02 -2.66595534e-03 -3.91336754e-02 6.38217211e-01 -8.86761919e-02 -8.25791776e-01 -4.14352790e-02 -2.38914013e-01 -4.94693160e-01 9.17536795e-01 3.29985400e-03 -2.96902955e-01 -2.66571462e-01 -1.12815344e+00 8.65568817e-02 2.41871029e-01 6.80435002e-01 3.99146199e-01 -9.27090585e-01 -5.49753368e-01 4.48826969e-01 1.08112849e-01 -5.47991335e-01 1.57149807e-01 8.36094439e-01 -9.31919515e-01 2.25721374e-01 -9.26802635e-01 -6.95808053e-01 -1.74939895e+00 1.91000864e-01 7.72760808e-01 3.66918482e-02 -5.35827994e-01 1.28325152e+00 -4.25768048e-01 -6.79217339e-01 8.25294614e-01 -1.17789261e-01 -4.94879425e-01 1.35473132e-01 5.33294857e-01 3.26045096e-01 3.90377581e-01 -4.32847321e-01 -5.27584314e-01 2.25299045e-01 7.39247575e-02 -1.51536658e-01 1.92061234e+00 8.52684602e-02 -2.97005057e-01 4.90628779e-01 1.20200038e+00 -3.50459427e-01 -1.29312825e+00 5.96780360e-01 1.29165977e-01 -1.28258422e-01 4.60538454e-02 -1.03160608e+00 -1.43951547e+00 1.13841951e+00 1.27422249e+00 5.52212931e-02 1.17925739e+00 -3.21069658e-01 5.05718514e-02 5.80013216e-01 1.74821585e-01 -1.18460274e+00 1.14623651e-01 6.22912407e-01 1.02563751e+00 -1.32050622e+00 -6.22321665e-02 -1.07568167e-01 -6.37337923e-01 1.42062390e+00 7.65409470e-01 -4.32815045e-01 6.58560991e-01 5.34425259e-01 3.22423697e-01 -6.54303432e-01 -1.00787377e+00 -1.13255970e-01 4.64249961e-02 6.92063749e-01 6.76494598e-01 -9.43103284e-02 -1.50985166e-01 6.25115275e-01 -5.00534534e-01 3.89271677e-01 2.96548516e-01 8.92053723e-01 1.38439670e-01 -8.82093608e-01 -1.90717131e-01 -1.14647008e-01 -7.69787252e-01 1.54479206e-01 -6.24295115e-01 1.03742743e+00 6.50354028e-01 6.26980782e-01 1.23829834e-01 -5.45606554e-01 3.09313864e-01 3.84856671e-01 6.50516152e-01 -3.52066994e-01 -1.08903158e+00 -2.45255679e-01 3.63365173e-01 -8.06888282e-01 -3.97033453e-01 -4.45837557e-01 -9.86152232e-01 -3.98442447e-01 8.18149969e-02 -2.90770233e-01 1.39195013e+00 7.43726134e-01 4.76705313e-01 6.90460920e-01 5.25725223e-02 -1.48051071e+00 -8.50631297e-02 -1.19252706e+00 -3.97418439e-01 2.37774819e-01 1.66298300e-01 -8.26886833e-01 -3.25047374e-01 5.01597263e-02]
[13.522012710571289, 1.8426072597503662]
25657a42-8cda-4da9-a8d6-530660b44872
compressed-heterogeneous-graph-for
2303.06565
null
https://arxiv.org/abs/2303.06565v1
https://arxiv.org/pdf/2303.06565v1.pdf
Compressed Heterogeneous Graph for Abstractive Multi-Document Summarization
Multi-document summarization (MDS) aims to generate a summary for a number of related documents. We propose HGSUM, an MDS model that extends an encoder-decoder architecture, to incorporate a heterogeneous graph to represent different semantic units (e.g., words and sentences) of the documents. This contrasts with existing MDS models which do not consider different edge types of graphs and as such do not capture the diversity of relationships in the documents. To preserve only key information and relationships of the documents in the heterogeneous graph, HGSUM uses graph pooling to compress the input graph. And to guide HGSUM to learn compression, we introduce an additional objective that maximizes the similarity between the compressed graph and the graph constructed from the ground-truth summary during training. HGSUM is trained end-to-end with graph similarity and standard cross-entropy objectives. Experimental results over MULTI-NEWS, WCEP-100, and ARXIV show that HGSUM outperforms state-of-the-art MDS models. The code for our model and experiments is available at: https://github.com/oaimli/HGSum.
['Jey Han Lau', 'Jianzhong Qi', 'Miao Li']
2023-03-12
null
null
null
null
['graph-similarity', 'multi-document-summarization', 'document-summarization']
['graphs', 'natural-language-processing', 'natural-language-processing']
[ 1.82246447e-01 7.55575597e-01 -2.94219106e-01 -1.71419472e-01 -9.96653020e-01 -5.66582859e-01 7.92727530e-01 6.60441279e-01 1.56327069e-01 6.77898049e-01 1.24463296e+00 1.95398688e-01 -1.10316455e-01 -8.82467151e-01 -9.13238585e-01 -3.01793605e-01 -9.60880145e-02 5.55420160e-01 8.51992965e-02 -1.03919476e-01 5.57423830e-01 -1.62764534e-01 -1.04342592e+00 7.16695428e-01 1.02924287e+00 4.93039221e-01 5.49055576e-01 8.83465469e-01 -1.98803723e-01 8.55580926e-01 -8.15857053e-01 -8.19178104e-01 -1.15930311e-01 -9.84663606e-01 -9.12761748e-01 5.15972674e-02 7.54128635e-01 -3.19874704e-01 -1.19377029e+00 1.02515006e+00 7.04417050e-01 1.79339528e-01 7.12456822e-01 -1.09249651e+00 -1.17295194e+00 1.30943084e+00 -3.76244128e-01 1.73739657e-01 5.71137249e-01 -3.57443094e-01 1.41936457e+00 -4.63761121e-01 1.08442044e+00 1.36749995e+00 4.23793763e-01 5.06921172e-01 -1.08386123e+00 -1.06481984e-01 1.04240239e-01 3.15867364e-01 -1.23568320e+00 -4.93901968e-01 6.39293849e-01 1.48986027e-01 1.27883351e+00 4.37624782e-01 6.03168130e-01 1.04459143e+00 6.25803947e-01 9.23140526e-01 1.22612454e-01 -2.75411010e-01 2.24463046e-01 -3.68510127e-01 2.23591715e-01 8.36939096e-01 6.68585837e-01 -5.94307721e-01 -8.58650982e-01 -1.82734340e-01 1.82655141e-01 -1.45791709e-01 -5.14417589e-01 -2.98347831e-01 -1.15504086e+00 7.01429784e-01 7.21448004e-01 3.28619927e-01 -3.99214298e-01 3.53590846e-01 4.88494098e-01 7.71154389e-02 6.33583605e-01 4.48783308e-01 1.68163031e-01 1.36091605e-01 -1.28215015e+00 4.80036974e-01 1.04529095e+00 1.36551487e+00 4.55078453e-01 -1.96470499e-01 -7.05418766e-01 6.24724269e-01 1.69384301e-01 3.12223077e-01 5.61213255e-01 -1.00771964e+00 9.92478609e-01 6.42970324e-01 -4.48292822e-01 -1.21102178e+00 -1.53158069e-01 -7.32380927e-01 -1.03110027e+00 -8.01978827e-01 -3.38581741e-01 5.34819402e-02 -7.53667355e-01 1.50580668e+00 -5.62579297e-02 3.52067232e-01 2.62141407e-01 5.81382155e-01 1.37093925e+00 1.06311488e+00 -5.09561121e-01 -1.75370842e-01 9.63015258e-01 -1.33990180e+00 -8.46978247e-01 -4.93503451e-01 6.96855962e-01 -4.51942384e-01 6.70336068e-01 -1.17887914e-01 -1.63457692e+00 -2.70593137e-01 -1.28679609e+00 -4.75415707e-01 -1.49756014e-01 -1.10521950e-01 1.09225072e-01 -9.77001525e-03 -1.48394859e+00 1.01459515e+00 -7.36861289e-01 -3.63541007e-01 4.48297828e-01 -2.33244896e-02 -2.13003397e-01 -3.68981808e-01 -1.16030979e+00 8.56131554e-01 7.63815761e-01 -3.78219217e-01 -7.15452492e-01 -6.75272465e-01 -1.00657117e+00 5.34817517e-01 2.72317261e-01 -1.26817167e+00 1.28329599e+00 -2.35871688e-01 -9.82653677e-01 5.19328058e-01 -3.71318340e-01 -7.38719761e-01 2.09466919e-01 3.42070721e-02 -1.54734224e-01 5.03403962e-01 3.11299831e-01 6.79350078e-01 3.08234423e-01 -1.29468215e+00 -2.92954504e-01 -2.40678251e-01 -1.03850253e-01 5.46361327e-01 -2.46560469e-01 -2.82438099e-01 -8.45771611e-01 -8.24173987e-01 1.01964399e-01 -7.97591269e-01 7.12131783e-02 -5.79294324e-01 -1.04072928e+00 4.16570622e-03 6.50396049e-01 -1.12875247e+00 1.74988782e+00 -1.86692846e+00 7.69649863e-01 -4.53294143e-02 4.69151556e-01 -6.47434071e-02 -3.92078012e-01 1.09579432e+00 2.77042270e-01 1.05216615e-01 -6.72701001e-01 -8.69079173e-01 1.86283559e-01 1.69692129e-01 -2.28118792e-01 3.25760961e-01 -1.09891005e-01 1.13118780e+00 -1.07418883e+00 -6.00558639e-01 -2.68416852e-01 4.59775299e-01 -6.23228431e-01 8.08964297e-02 -4.87585843e-01 5.94558977e-02 -3.24036181e-01 2.38413528e-01 5.17550349e-01 -5.51131725e-01 2.84407854e-01 -2.04451308e-01 3.83521497e-01 7.13579595e-01 -8.26143622e-01 2.29288220e+00 -1.10794634e-01 7.04715788e-01 -2.76619196e-01 -9.24936950e-01 7.44375169e-01 2.54804730e-01 3.71401161e-01 -6.54685974e-01 -4.11491767e-02 1.51607275e-01 -4.35903579e-01 -1.04891278e-01 1.14850545e+00 4.14526761e-01 -2.29016796e-01 7.13352025e-01 5.20020306e-01 -4.47432697e-01 7.13807106e-01 1.25292969e+00 1.32303905e+00 -1.90683722e-01 2.47719094e-01 -3.49656753e-02 1.53027982e-01 -1.89833909e-01 2.02735782e-01 7.08319545e-01 4.82857257e-01 9.20755982e-01 8.30281556e-01 2.71982789e-01 -1.17502415e+00 -1.14449644e+00 3.31287384e-01 5.80236375e-01 2.40765959e-01 -1.16119301e+00 -8.90775979e-01 -6.73905909e-01 9.76494923e-02 1.46162891e+00 -4.38487470e-01 -6.82504058e-01 -5.77879906e-01 -5.11900663e-01 4.43904757e-01 2.58924544e-01 4.93454307e-01 -7.25701571e-01 1.53629463e-02 1.74588054e-01 -6.90916002e-01 -1.00399673e+00 -9.42122281e-01 -2.87947744e-01 -8.73894274e-01 -8.94403934e-01 -6.94784105e-01 -8.25810909e-01 6.72333658e-01 4.78285640e-01 1.43940473e+00 6.84475228e-02 4.57038209e-02 5.33516526e-01 -4.59748596e-01 -9.28979665e-02 -8.25155020e-01 4.27425772e-01 -5.52698016e-01 -3.69962722e-01 -8.21742043e-02 -6.79891229e-01 -5.50029874e-01 -3.43986779e-01 -1.04351926e+00 3.80355954e-01 5.30194402e-01 4.55836594e-01 6.20410264e-01 -5.63791618e-02 7.21685469e-01 -8.40030730e-01 1.17884028e+00 -8.09547365e-01 6.86538890e-02 6.22895241e-01 -6.35815442e-01 3.21094632e-01 4.58531976e-01 1.39541686e-01 -8.03083360e-01 -5.72196066e-01 1.59496680e-01 -1.55108631e-01 5.76411545e-01 8.67897034e-01 -2.58362629e-02 5.18509448e-01 5.04537463e-01 5.89218676e-01 -1.12618975e-01 -4.53488141e-01 7.69536853e-01 6.52232587e-01 8.01985979e-01 -1.87213734e-01 2.06058607e-01 1.65294871e-01 2.52186805e-02 -5.91559172e-01 -9.19044077e-01 -3.32806349e-01 -4.79351133e-01 -1.28480732e-01 6.71241879e-01 -1.00696635e+00 1.07106745e-01 -1.63772888e-02 -1.45844638e+00 2.88202129e-02 -4.82118905e-01 2.19536960e-01 -7.31355250e-01 7.15049207e-01 -6.34316742e-01 -2.30489731e-01 -9.04742122e-01 -6.47889853e-01 1.32563865e+00 1.78822905e-01 -4.33784992e-01 -1.07580864e+00 1.44885883e-01 1.24943033e-01 1.80592239e-01 3.54969800e-01 9.16001678e-01 -6.52264357e-01 -6.21627808e-01 -3.81398171e-01 -7.21050799e-02 2.11283639e-01 1.27874136e-01 -7.09365308e-02 -4.04134989e-01 -5.79584181e-01 -2.24496052e-01 -2.37662897e-01 1.41690505e+00 5.79455376e-01 9.16313529e-01 -8.94023359e-01 -5.03540456e-01 4.66709316e-01 1.41602671e+00 -1.89596474e-01 7.06679225e-01 1.52747696e-02 7.48719871e-01 4.75347072e-01 3.77993584e-02 5.87855935e-01 8.81669104e-01 4.24919993e-01 4.36637282e-01 5.04011393e-01 -6.74628437e-01 -8.41204345e-01 5.53747058e-01 1.37020338e+00 3.05002242e-01 -1.15338886e+00 -6.32089198e-01 6.18238747e-01 -2.02471757e+00 -1.19146621e+00 -1.03522383e-01 1.96538472e+00 7.21621454e-01 1.13274176e-02 -8.26655924e-02 -1.71936467e-01 9.61224198e-01 7.30027318e-01 -5.08242369e-01 -3.38041425e-01 -5.00778437e-01 -3.92952412e-02 4.35731262e-01 7.08720267e-01 -6.62693977e-01 7.19069660e-01 5.67458534e+00 7.10922003e-01 -5.00419140e-01 -1.13026835e-01 3.26401651e-01 -6.05880320e-01 -7.63836265e-01 6.00582659e-02 -5.66867650e-01 5.46476126e-01 1.19719112e+00 -1.02644658e+00 4.39101309e-01 3.93425107e-01 -1.59098238e-01 5.90849407e-02 -1.03297126e+00 6.80604637e-01 6.30106628e-01 -2.03810525e+00 4.30327326e-01 -1.76849335e-01 9.89514947e-01 2.32740313e-01 -9.78219286e-02 1.12761654e-01 5.25327325e-01 -6.97466016e-01 9.46315885e-01 7.21576929e-01 5.64404607e-01 -7.06141353e-01 5.39428711e-01 3.29692572e-01 -1.08366609e+00 8.91889781e-02 -6.44409060e-01 4.29450244e-01 4.67241913e-01 5.55284381e-01 -8.04632366e-01 1.21631932e+00 2.80043155e-01 1.24638975e+00 -7.74200976e-01 9.00760710e-01 -3.63528728e-01 3.41165841e-01 1.63797066e-01 -1.51308283e-01 2.24590972e-01 -1.71118051e-01 9.42518294e-01 1.34885478e+00 7.54288912e-01 -1.68534368e-01 -7.11429641e-02 7.83449352e-01 -6.10093117e-01 -1.10191368e-01 -6.66520655e-01 -5.00989079e-01 6.73271716e-01 8.52782309e-01 -4.14205551e-01 -7.21056998e-01 -3.16597849e-01 1.39116991e+00 4.09727752e-01 1.97882265e-01 -6.24313474e-01 -4.79693174e-01 1.59663692e-01 1.06488012e-01 4.01693463e-01 -2.12808743e-01 -2.66513586e-01 -1.30163467e+00 2.14964926e-01 -7.29003370e-01 6.45185769e-01 -9.39888537e-01 -8.67674112e-01 6.44332647e-01 1.72745660e-01 -1.02732277e+00 -4.90023255e-01 2.33663470e-01 -7.57076204e-01 6.96215868e-01 -1.13597906e+00 -1.06809509e+00 -1.85041398e-01 1.73638895e-01 5.59964955e-01 -1.96518093e-01 7.32207596e-01 -7.43565429e-03 -4.70923007e-01 4.17319000e-01 5.30527830e-01 -1.56461284e-01 5.41632175e-01 -1.37087715e+00 1.01048148e+00 1.02631855e+00 2.32538491e-01 4.24187988e-01 7.97434032e-01 -1.11608243e+00 -1.40362072e+00 -1.29749632e+00 1.37621510e+00 -1.02874503e-01 5.80298424e-01 -3.85620654e-01 -9.20231998e-01 8.52418482e-01 9.53693211e-01 -4.56280947e-01 6.13535166e-01 -3.29971075e-01 -2.17099145e-01 2.08193332e-01 -9.24508333e-01 6.31492496e-01 1.39820731e+00 -3.69802922e-01 -7.18627572e-01 6.86905146e-01 1.23389328e+00 -5.52848637e-01 -7.97126770e-01 -8.56678337e-02 1.26193583e-01 -7.99748182e-01 8.55731905e-01 -5.88708282e-01 1.15802753e+00 5.57431169e-02 -3.59189391e-01 -2.00881696e+00 -5.56526601e-01 -5.98371029e-01 -9.31265593e-01 1.38025200e+00 3.85883242e-01 -3.46260101e-01 6.36253238e-01 1.70811743e-01 -7.67367721e-01 -7.14462161e-01 -8.41357827e-01 -7.61874437e-01 4.80823293e-02 1.73694119e-01 8.31214130e-01 6.90259159e-01 3.91812712e-01 5.97692609e-01 -2.13376358e-01 4.47274484e-02 7.65543342e-01 2.81243414e-01 5.23949802e-01 -8.66650045e-01 -2.44970918e-01 -6.22968018e-01 -2.31490269e-01 -1.03456795e+00 3.78032297e-01 -1.64125216e+00 -2.52329558e-01 -2.71739459e+00 4.60725605e-01 2.95854062e-01 8.90682563e-02 6.96567744e-02 -2.80048460e-01 -3.67251426e-01 3.11851263e-01 1.87388033e-01 -9.41820741e-01 1.05876184e+00 1.38270617e+00 -4.58019525e-01 2.30524316e-02 -3.05836260e-01 -1.27773607e+00 8.17846581e-02 9.64582205e-01 -5.64071059e-01 -7.00430334e-01 -7.99379945e-01 2.57440120e-01 3.16869140e-01 1.36279196e-01 -8.93617451e-01 6.58570290e-01 2.33672827e-01 9.24050063e-02 -8.26795161e-01 2.18659848e-01 -1.79752409e-01 2.42909327e-01 5.36122620e-01 -7.22489059e-01 2.28277475e-01 -2.10590605e-02 7.79528022e-01 -3.83015841e-01 -3.13069403e-01 3.15988898e-01 -2.65943736e-01 -2.24733457e-01 3.52010697e-01 1.03802942e-01 4.39093024e-01 7.23123610e-01 -1.31116286e-02 -9.96136904e-01 -8.36910844e-01 -3.04320216e-01 5.35730064e-01 5.00968933e-01 5.14226854e-01 7.71517992e-01 -1.41568422e+00 -1.31457543e+00 -2.77582109e-01 1.66086815e-02 2.80069470e-01 4.85471636e-01 3.51329595e-01 -5.22138715e-01 5.79250932e-01 8.82403031e-02 -5.72255664e-02 -1.09637189e+00 5.03923476e-01 9.24374908e-02 -5.61899781e-01 -8.78029585e-01 7.71355152e-01 -1.18787229e-01 -1.20626323e-01 1.28003523e-01 -3.03262532e-01 7.21266400e-03 9.29429010e-02 6.59591019e-01 7.10762262e-01 1.19051144e-01 -6.17473960e-01 -1.51800975e-01 1.19766355e-01 -2.46860072e-01 -6.24599606e-02 1.54544580e+00 -2.95524538e-01 -4.41736460e-01 1.72500357e-01 1.51183331e+00 1.00956485e-02 -1.07298887e+00 -4.77337003e-01 2.32175272e-02 -2.86246181e-01 1.36079937e-01 -5.35473824e-01 -9.84567881e-01 5.11305630e-01 -3.62431228e-01 2.72939026e-01 1.00395250e+00 3.54136646e-01 1.05879033e+00 2.71832377e-01 2.52519362e-03 -1.06088030e+00 2.71902204e-01 4.06594783e-01 1.59629345e+00 -6.64652586e-01 4.53187048e-01 -3.64422321e-01 -8.82702887e-01 1.09058237e+00 2.48234734e-01 -1.47810116e-01 3.42404068e-01 7.18541369e-02 -7.89194882e-01 -2.94668227e-01 -1.09481585e+00 1.42994463e-01 5.22027850e-01 4.24097419e-01 4.21731830e-01 5.67001961e-02 -4.16930795e-01 5.55510283e-01 -6.32845163e-01 -2.17705339e-01 8.40331554e-01 9.40238178e-01 -6.26149833e-01 -8.85948122e-01 1.18242614e-01 8.83179307e-01 -2.78131962e-01 -2.92634785e-01 -8.69757175e-01 2.56230235e-01 -4.53591049e-01 9.85174477e-01 2.49302268e-01 -5.19767106e-01 4.26086217e-01 -2.33594868e-02 5.34095228e-01 -7.79258609e-01 -3.68751228e-01 -2.69227654e-01 3.83354574e-01 -3.97350043e-01 -1.37757897e-01 -7.69776702e-01 -1.45904684e+00 -7.91984081e-01 8.26576073e-03 2.69040197e-01 4.92168158e-01 4.53798026e-01 8.88910115e-01 9.30936754e-01 5.56421161e-01 -6.50081038e-01 -4.85633850e-01 -1.17210782e+00 -6.82486415e-01 2.71616876e-01 3.26813668e-01 -9.84691549e-03 -2.81043828e-01 -4.58447598e-02]
[12.437117576599121, 9.437346458435059]
dcd0e829-8568-40e8-af69-961860e1bce9
dreamidentity-improved-editability-for
2307.00300
null
https://arxiv.org/abs/2307.00300v1
https://arxiv.org/pdf/2307.00300v1.pdf
DreamIdentity: Improved Editability for Efficient Face-identity Preserved Image Generation
While large-scale pre-trained text-to-image models can synthesize diverse and high-quality human-centric images, an intractable problem is how to preserve the face identity for conditioned face images. Existing methods either require time-consuming optimization for each face-identity or learning an efficient encoder at the cost of harming the editability of models. In this work, we present an optimization-free method for each face identity, meanwhile keeping the editability for text-to-image models. Specifically, we propose a novel face-identity encoder to learn an accurate representation of human faces, which applies multi-scale face features followed by a multi-embedding projector to directly generate the pseudo words in the text embedding space. Besides, we propose self-augmented editability learning to enhance the editability of models, which is achieved by constructing paired generated face and edited face images using celebrity names, aiming at transferring mature ability of off-the-shelf text-to-image models in celebrity faces to unseen faces. Extensive experiments show that our methods can generate identity-preserved images under different scenes at a much faster speed.
['Zhendong Mao', 'Yongdong Zhang', 'Mengqi Huang', 'Qian He', 'Wei Liu', 'Shancheng Fang', 'Zhuowei Chen']
2023-07-01
null
null
null
null
['image-generation']
['computer-vision']
[ 2.34184310e-01 2.70168353e-02 2.57138252e-01 -7.98409581e-01 -6.86398029e-01 -4.34445590e-01 5.26082397e-01 -8.60094309e-01 -2.21157849e-01 5.28286994e-01 2.19212368e-01 3.39073449e-01 2.63249129e-01 -6.87390983e-01 -1.02449358e+00 -5.13508022e-01 4.94203746e-01 4.19412702e-01 -5.01686871e-01 -8.43934119e-02 2.52963975e-02 5.18862307e-01 -1.78900123e+00 3.22148830e-01 7.96063602e-01 9.20338094e-01 5.05492017e-02 4.64556545e-01 -3.84945497e-02 6.80424452e-01 -4.11071301e-01 -1.10079443e+00 4.34086800e-01 -7.11413920e-01 -3.91934216e-01 3.73653203e-01 1.05900967e+00 -6.08834445e-01 -5.06466746e-01 1.11053383e+00 5.91285229e-01 -1.76918954e-01 6.60587370e-01 -1.56945467e+00 -1.48186219e+00 2.19645455e-01 -5.79549909e-01 -4.15011406e-01 3.33182812e-01 1.89830363e-01 6.91466510e-01 -1.28720820e+00 8.37920487e-01 1.65868592e+00 4.57750052e-01 8.50729764e-01 -1.19480860e+00 -1.23282969e+00 -1.41118526e-01 2.23245636e-01 -1.76530147e+00 -1.08181632e+00 7.40691483e-01 -4.01724368e-01 3.56260061e-01 1.10123679e-01 3.79330158e-01 1.18052578e+00 1.16063811e-01 3.01016510e-01 9.52868760e-01 -1.77826867e-01 -2.51858801e-01 2.48468056e-01 -7.94977248e-01 9.79407668e-01 1.03175454e-01 2.47379512e-01 -6.75464690e-01 8.14892948e-02 9.93274748e-01 4.73553650e-02 -3.12451035e-01 -2.30405062e-01 -1.12999701e+00 7.79295743e-01 1.54855624e-01 -2.14659013e-02 -1.27499610e-01 -1.10253012e-02 1.74661845e-01 4.12514985e-01 5.50009310e-01 1.48913428e-01 -1.80577263e-01 3.74010772e-01 -9.63762760e-01 1.69477418e-01 4.93455231e-01 1.33852780e+00 1.11111605e+00 2.45216385e-01 -3.22608232e-01 8.11675966e-01 3.28392148e-01 8.47494483e-01 3.59388381e-01 -9.31454360e-01 2.27824107e-01 3.41147482e-01 -3.07503175e-02 -1.09100199e+00 2.38826558e-01 -1.17444530e-01 -9.69251275e-01 2.14439929e-01 4.94455062e-02 -5.06258495e-02 -7.53941596e-01 1.96850204e+00 4.55583841e-01 2.97876090e-01 -1.09060138e-01 6.37161911e-01 6.56339824e-01 6.03519320e-01 -3.65725309e-02 -2.45699912e-01 1.28693128e+00 -1.14012504e+00 -8.70367944e-01 1.03640817e-02 1.31832778e-01 -1.03865921e+00 1.07434356e+00 -8.12691897e-02 -1.16341329e+00 -8.51396501e-01 -8.20781887e-01 -4.13454682e-01 -1.66659534e-01 5.00115216e-01 1.56420603e-01 7.95405030e-01 -1.10406184e+00 4.39909011e-01 -3.16547185e-01 -1.19690038e-01 6.90555453e-01 4.05161768e-01 -9.50203657e-01 -2.37058491e-01 -9.18911994e-01 9.24819291e-01 1.91313073e-01 6.93866089e-02 -1.26516330e+00 -9.28698182e-01 -1.21082973e+00 4.09197919e-02 2.18963951e-01 -8.27922881e-01 8.51237833e-01 -1.42625475e+00 -1.78598607e+00 1.11991847e+00 -4.31527078e-01 1.08803079e-01 6.91677213e-01 -5.34835309e-02 -6.48893297e-01 2.13401794e-01 2.23288774e-01 9.39173400e-01 1.55640030e+00 -1.29731393e+00 -3.71035963e-01 -4.53233898e-01 -1.46390960e-01 8.36623311e-02 -7.45823681e-01 3.15626651e-01 -6.77746832e-01 -8.06123018e-01 -2.53015399e-01 -9.56114531e-01 2.79288858e-01 8.06961715e-01 -2.00281352e-01 1.40316123e-02 7.96813726e-01 -9.82486308e-01 7.89912343e-01 -2.25578141e+00 1.77011669e-01 -1.78100646e-01 1.28657535e-01 9.84216556e-02 -6.96406960e-01 2.37678558e-01 -1.80732831e-01 1.04644954e-01 -2.31734633e-01 -8.65814447e-01 2.89283209e-02 1.05740368e-01 -2.50430882e-01 5.63674688e-01 6.88469708e-01 9.55857396e-01 -6.53518260e-01 -8.52074087e-01 1.84132248e-01 9.55439448e-01 -7.63441205e-01 4.60599869e-01 1.25269555e-02 5.47070920e-01 -1.65494397e-01 4.89570409e-01 1.08881795e+00 -9.50412126e-04 6.05495414e-03 -3.76368374e-01 1.10610582e-01 -4.19130087e-01 -1.05203211e+00 1.75734568e+00 -7.03402996e-01 3.31778616e-01 2.92543143e-01 -6.72056317e-01 9.77887928e-01 3.16270500e-01 2.42391452e-01 -6.24758720e-01 1.15523867e-01 2.43646145e-01 -3.80356759e-01 -3.82874370e-01 4.02264416e-01 -4.56133038e-01 2.63303190e-01 5.51811397e-01 3.20419192e-01 -2.37772152e-01 1.00108437e-01 2.70433426e-02 3.74842167e-01 3.39545846e-01 -1.93285376e-01 -2.00523555e-01 7.73074985e-01 -6.85862780e-01 5.22291899e-01 8.02068636e-02 4.62600403e-02 9.65257525e-01 9.55348462e-02 -3.39016646e-01 -1.29794657e+00 -1.01172769e+00 -4.85652015e-02 1.10533750e+00 -1.03836015e-01 -2.48641193e-01 -9.21196461e-01 -6.10818148e-01 -1.27380230e-02 4.86917973e-01 -8.16463053e-01 -3.65937471e-01 -5.08819222e-01 -2.75091290e-01 5.52728474e-01 2.78845161e-01 6.10508680e-01 -9.03843403e-01 2.60257930e-01 -4.84450720e-02 -2.76101321e-01 -1.12171578e+00 -1.39143443e+00 -7.15137959e-01 -4.21928585e-01 -9.72196639e-01 -9.45845604e-01 -1.18861115e+00 1.17916834e+00 1.84719115e-01 9.10634398e-01 6.76312372e-02 -3.44076872e-01 3.20338905e-01 -5.88866360e-02 -2.62031496e-01 -4.90234703e-01 -3.44154030e-01 2.14090392e-01 7.52166212e-01 4.92832176e-02 -4.90499556e-01 -4.75597978e-01 4.45948869e-01 -1.05143297e+00 1.84740856e-01 5.04727125e-01 9.83400941e-01 5.20332754e-01 -9.39326510e-02 6.83268666e-01 -6.33884966e-01 3.07127923e-01 -7.15647489e-02 -5.51686764e-01 6.16596401e-01 -4.46883500e-01 2.06905633e-01 8.81317377e-01 -5.41392684e-01 -1.21830583e+00 3.83739978e-01 -1.37839496e-01 -9.11335647e-01 2.65786767e-01 -1.80121809e-01 -6.15365863e-01 -4.10468847e-01 4.28626895e-01 6.40686035e-01 3.58689427e-01 -2.54433155e-01 7.17839122e-01 6.35660172e-01 7.83427894e-01 -6.52925134e-01 1.24819851e+00 6.52990878e-01 -2.09344879e-01 -7.13115394e-01 -7.04284430e-01 7.90660381e-02 -6.54856026e-01 -1.48913637e-01 8.43973637e-01 -1.30475593e+00 -7.43168294e-01 6.34678066e-01 -1.29565883e+00 1.69911742e-01 -1.40828341e-01 1.67744398e-01 -6.15182519e-01 4.32676405e-01 -4.39681083e-01 -4.87098783e-01 -4.14119691e-01 -9.97425795e-01 1.39153337e+00 2.01454461e-01 2.31554776e-01 -6.85491681e-01 -9.75765064e-02 4.56014276e-01 4.79563445e-01 8.80614594e-02 5.58239162e-01 1.18202576e-02 -6.63453877e-01 -1.21657282e-01 -4.24417555e-01 6.80941641e-01 2.69736260e-01 1.64384872e-01 -9.76720929e-01 -5.14058709e-01 8.93300921e-02 -5.90974450e-01 6.12865686e-01 -2.98836995e-02 1.37458730e+00 -5.45204222e-01 -8.17244798e-02 9.51308787e-01 1.25211430e+00 -1.62892222e-01 7.89221942e-01 -3.26751977e-01 8.78695548e-01 7.11971641e-01 2.92822987e-01 5.23776054e-01 4.93880153e-01 6.36431694e-01 1.14942051e-01 -5.73584475e-02 -3.64206225e-01 -6.50008380e-01 4.53362435e-01 8.28373492e-01 4.86425497e-02 -3.63509147e-03 -2.05489382e-01 5.41334748e-01 -1.35387099e+00 -1.06978405e+00 4.87485975e-01 2.16222358e+00 1.16490519e+00 -4.87304956e-01 -2.14921519e-01 -3.13986748e-01 1.10742855e+00 6.29388466e-02 -5.55903137e-01 6.09776042e-02 -2.88666248e-01 2.77725309e-01 3.98769379e-02 5.89901984e-01 -8.64548683e-01 1.01152372e+00 6.04557180e+00 8.79596293e-01 -1.11954200e+00 3.01471293e-01 7.09642947e-01 -3.72821420e-01 -4.86700207e-01 -9.58469585e-02 -7.48666406e-01 5.83491564e-01 6.90548062e-01 -5.28569400e-01 6.76421463e-01 8.50883305e-01 1.71022434e-02 6.17340863e-01 -1.19943213e+00 1.42869651e+00 6.73264861e-01 -1.28304505e+00 5.73143840e-01 2.76985411e-02 1.02284288e+00 -6.20158494e-01 2.09172189e-01 2.31775731e-01 2.83846445e-02 -1.15226543e+00 9.13741648e-01 6.57028079e-01 1.49984694e+00 -8.29111159e-01 2.21661583e-01 -4.51843701e-02 -1.22611511e+00 2.31730156e-02 -6.10589802e-01 3.20900649e-01 1.04659900e-01 3.40814620e-01 -4.22357380e-01 5.41845798e-01 4.87427860e-01 7.48169780e-01 -6.24011219e-01 4.76737738e-01 -1.16494738e-01 -8.81327391e-02 -7.69605786e-02 3.03669542e-01 -2.69502968e-01 -3.43999237e-01 1.18192382e-01 8.31621945e-01 6.52159035e-01 1.99728161e-02 -5.41472025e-02 1.15037787e+00 -6.17765903e-01 2.28398755e-01 -6.35602713e-01 -2.61394829e-01 5.33092141e-01 1.43096876e+00 -3.27053703e-02 -2.94332623e-01 -4.04148936e-01 1.52093697e+00 3.53395343e-01 2.06340820e-01 -1.00626504e+00 -3.71899068e-01 7.98371792e-01 1.59300610e-01 2.59792000e-01 -3.32161859e-02 1.09760299e-01 -1.31729698e+00 2.80823022e-01 -1.07062972e+00 -1.09797962e-01 -8.62094223e-01 -1.41919219e+00 8.51157784e-01 -4.39021021e-01 -1.10513747e+00 -2.35566162e-02 -5.46650767e-01 -4.58814234e-01 9.94112194e-01 -1.59132671e+00 -1.72776997e+00 -4.82958287e-01 1.06013989e+00 4.49830681e-01 -4.43102688e-01 8.44526947e-01 6.01399302e-01 -5.13814569e-01 1.12029290e+00 2.32888848e-01 3.10167402e-01 1.25039518e+00 -7.01705337e-01 4.67898786e-01 6.56420887e-01 1.81413665e-01 6.49264455e-01 3.26964647e-01 -5.41884065e-01 -1.50526202e+00 -1.45319963e+00 1.01747692e+00 -4.57232207e-01 3.23080182e-01 -5.72986424e-01 -6.50380433e-01 8.16745996e-01 3.58056933e-01 2.31278300e-01 6.76046729e-01 -3.04016590e-01 -6.91786647e-01 -3.07915807e-01 -1.24359810e+00 7.04788089e-01 1.33758223e+00 -1.01574051e+00 -2.68418193e-01 3.70917708e-01 7.29118168e-01 -2.32945547e-01 -9.26847398e-01 2.20016599e-01 7.74191380e-01 -7.67970204e-01 9.82632995e-01 -4.90687460e-01 7.29880154e-01 -3.89130980e-01 -3.97305526e-02 -1.14146328e+00 -3.69615138e-01 -7.02877998e-01 2.42104918e-01 1.55533266e+00 1.16273500e-01 -5.30274034e-01 5.65902472e-01 5.64803660e-01 7.51918778e-02 -4.00483072e-01 -8.99837255e-01 -7.91543663e-01 1.01021178e-01 1.96838140e-01 9.62953091e-01 1.03854918e+00 -6.22645378e-01 3.29740465e-01 -9.24423635e-01 1.07400112e-01 8.05623651e-01 5.22863455e-02 9.41824615e-01 -1.01081502e+00 -1.26706949e-02 -1.50000304e-01 -2.51116455e-01 -8.00754488e-01 7.11677372e-01 -9.18362975e-01 8.34151730e-03 -9.76767242e-01 4.18921202e-01 -1.44020185e-01 8.15899372e-02 3.63177150e-01 -1.60987183e-01 6.00770295e-01 2.51205385e-01 1.30909204e-01 -4.32084709e-01 1.17772293e+00 1.69245672e+00 -2.74177849e-01 4.19083744e-01 -4.69442219e-01 -8.36451530e-01 4.26700920e-01 4.21180844e-01 -4.80977148e-01 -5.50559402e-01 -7.55607545e-01 -1.07997686e-01 -1.87339336e-01 4.28749293e-01 -9.06686187e-01 2.09966362e-01 -1.80242091e-01 6.04700744e-01 -1.79836944e-01 5.30411124e-01 -7.90880203e-01 4.71355617e-01 1.41729757e-01 -2.93010920e-01 -3.85272759e-03 1.40796199e-01 4.71586287e-01 -3.23216110e-01 -1.51172206e-01 1.14856303e+00 -1.19277641e-01 -3.19707841e-01 9.01775837e-01 3.00913334e-01 1.72269177e-02 1.14455318e+00 -9.96284038e-02 -1.93248853e-01 -5.04022598e-01 -4.86017555e-01 6.77905008e-02 8.49904478e-01 8.40569019e-01 7.95775712e-01 -1.72327411e+00 -1.31644285e+00 6.76319182e-01 2.68339872e-01 -3.86019945e-01 5.38717270e-01 2.45922327e-01 -4.66162503e-01 7.05837905e-02 -6.17133737e-01 -3.06665093e-01 -1.41404247e+00 7.52151608e-01 3.87268454e-01 2.27858528e-01 -2.61440814e-01 1.02564454e+00 4.98289973e-01 -7.08874166e-01 -3.22187990e-02 3.75564784e-01 1.95532128e-01 -3.18252407e-02 8.44308496e-01 -9.59093347e-02 -1.94842800e-01 -1.07010639e+00 -1.93881869e-01 9.70933318e-01 -3.01841408e-01 -1.70564383e-01 1.14373291e+00 -1.48257762e-01 -4.16501909e-01 -1.86289385e-01 1.54867887e+00 -4.45630290e-02 -1.46513224e+00 -2.18690187e-01 -5.75103998e-01 -9.10697281e-01 -2.09021300e-01 -3.97446722e-01 -1.37916899e+00 9.33847070e-01 6.95668399e-01 -5.41819751e-01 1.27981567e+00 -3.19227695e-01 8.19522142e-01 2.43168414e-01 3.85174483e-01 -9.85452414e-01 3.93327564e-01 2.11729810e-01 1.40249264e+00 -1.33556879e+00 -1.15208156e-01 -5.16501129e-01 -7.90763915e-01 1.00891340e+00 7.11869657e-01 2.11630818e-02 6.81648076e-01 1.74948236e-03 3.23289111e-02 1.37398541e-01 -5.79626083e-01 1.39518544e-01 1.08691521e-01 6.67833030e-01 2.41152808e-01 -1.72742486e-01 1.11596771e-02 3.90581161e-01 -4.06835467e-01 1.50960013e-01 3.08687568e-01 4.51060832e-01 -4.31502946e-02 -1.20912015e+00 -3.35698694e-01 1.80195764e-01 -3.11396658e-01 -2.81181991e-01 -3.22869688e-01 3.96762431e-01 3.53760868e-01 8.30575645e-01 1.07924983e-01 -4.34071362e-01 1.95032567e-01 1.43595055e-01 7.77625680e-01 -5.51254928e-01 -3.51498842e-01 -2.35283464e-01 -1.90900683e-01 -4.06679899e-01 -3.24258357e-01 -5.30537188e-01 -6.85523152e-01 -6.60464764e-01 -3.35805207e-01 -1.47434816e-01 5.58405340e-01 6.01614594e-01 7.36395121e-01 1.22625187e-01 1.05950320e+00 -8.70322585e-01 -5.99616289e-01 -8.94325614e-01 -7.26803780e-01 7.76076019e-01 1.51614502e-01 -6.14529312e-01 -3.40170473e-01 6.12023056e-01]
[12.569753646850586, -0.08234184980392456]
ed078b14-99ad-4d70-baea-48fe67df75cc
on-the-use-of-arxiv-as-a-dataset
1905.00075
null
http://arxiv.org/abs/1905.00075v1
http://arxiv.org/pdf/1905.00075v1.pdf
On the Use of ArXiv as a Dataset
The arXiv has collected 1.5 million pre-print articles over 28 years, hosting literature from scientific fields including Physics, Mathematics, and Computer Science. Each pre-print features text, figures, authors, citations, categories, and other metadata. These rich, multi-modal features, combined with the natural graph structure---created by citation, affiliation, and co-authorship---makes the arXiv an exciting candidate for benchmarking next-generation models. Here we take the first necessary steps toward this goal, by providing a pipeline which standardizes and simplifies access to the arXiv's publicly available data. We use this pipeline to extract and analyze a 6.7 million edge citation graph, with an 11 billion word corpus of full-text research articles. We present some baseline classification results, and motivate application of more exciting generative graph models.
["Kevin P. O'Keeffe", 'Alexander A. Alemi', 'Matthew Bierbaum', 'Colin B. Clement']
2019-04-30
null
null
null
null
['author-attribution', 'text-clustering']
['natural-language-processing', 'natural-language-processing']
[-2.35952288e-01 2.05023661e-01 -4.26758438e-01 1.65187642e-02 -6.55364990e-01 -1.13859808e+00 1.16942227e+00 6.22443080e-01 -1.68702111e-01 7.43597984e-01 5.20218968e-01 -7.63285995e-01 -2.69935399e-01 -1.04132581e+00 -7.12796330e-01 -8.66088420e-02 -2.38068312e-01 6.10452414e-01 -4.70370799e-02 2.29643002e-01 8.64211679e-01 3.80641907e-01 -1.12299430e+00 -2.03179032e-01 9.17987406e-01 3.79423350e-01 1.10207692e-01 6.26136720e-01 -6.61978543e-01 4.48963732e-01 -5.23754299e-01 -8.54795396e-01 -2.53054380e-01 -4.06104475e-01 -6.51998162e-01 -3.73369485e-01 7.79462934e-01 2.39500403e-01 -9.42073762e-01 9.56066847e-01 1.37078509e-01 -1.35633890e-02 9.91488457e-01 -1.41627121e+00 -1.03637922e+00 1.17172599e+00 -6.53025329e-01 6.36263728e-01 4.82074022e-01 1.00408465e-01 1.57714367e+00 -5.96765399e-01 1.31126881e+00 1.21252835e+00 6.37050688e-01 7.18449727e-02 -1.10431886e+00 -4.98621464e-01 -1.33629337e-01 -1.05630932e-02 -1.11502194e+00 -7.78633803e-02 9.16064858e-01 -8.09161544e-01 7.96569586e-01 8.74812379e-02 8.77854884e-01 1.49783731e+00 5.81886709e-01 1.46312401e-01 1.24865639e+00 -3.18539947e-01 2.43592840e-02 -1.57640904e-01 7.39294529e-01 1.01043236e+00 8.52110922e-01 -3.66277725e-01 -5.07321119e-01 -4.71065521e-01 6.32099092e-01 -3.93013619e-02 -1.74850062e-01 6.90526608e-03 -1.60599113e+00 5.35299540e-01 1.47526860e-01 3.71336758e-01 -1.18247367e-01 1.87087864e-01 2.00214565e-01 3.36869329e-01 5.48607588e-01 7.61990726e-01 2.56513990e-03 -3.44881535e-01 -1.09773397e+00 3.55014056e-01 1.24169624e+00 1.20571470e+00 7.46594608e-01 -1.90680146e-01 -1.98874280e-01 5.29542685e-01 4.44272876e-01 4.11464691e-01 2.54867803e-02 -9.21899021e-01 3.08656543e-01 7.06066072e-01 -5.03805220e-01 -1.32703280e+00 -2.31388286e-01 -8.70330811e-01 -7.25319862e-01 -4.64820415e-01 4.13580060e-01 -7.12268660e-03 -7.08334804e-01 1.35378528e+00 -5.45670390e-02 2.24090934e-01 -4.93466049e-01 3.01334113e-01 1.37394309e+00 5.03947794e-01 5.72413020e-02 2.09057163e-02 1.48999310e+00 -6.23012960e-01 -6.13728940e-01 1.07232695e-02 3.81424159e-01 -8.90630305e-01 7.39861250e-01 4.48554680e-02 -1.13359880e+00 -2.40453407e-01 -8.27869594e-01 -2.28867248e-01 -7.58732557e-01 -2.49117911e-01 7.48882353e-01 1.90497935e-01 -1.24297810e+00 8.97853076e-01 -3.43284041e-01 -4.38376486e-01 7.00016975e-01 -1.48800716e-01 -3.33272099e-01 -1.73315600e-01 -9.62143838e-01 6.85759246e-01 -2.04778537e-01 -7.35693455e-01 -3.89369249e-01 -1.13470781e+00 -5.05149305e-01 3.00282329e-01 1.77388296e-01 -1.14050722e+00 7.38201678e-01 1.63778558e-01 -8.51962984e-01 1.12782204e+00 -1.04915664e-01 -2.06298113e-01 2.10509434e-01 2.12292612e-01 -4.98087496e-01 8.54981914e-02 2.49115795e-01 1.00448519e-01 4.90750998e-01 -8.93692553e-01 -2.54632115e-01 -3.27054530e-01 -1.94665462e-01 -4.19646740e-01 -4.63143468e-01 -1.05395596e-02 -7.31662154e-01 -8.82258594e-01 -1.27796158e-01 -8.51001322e-01 -2.34814882e-02 -5.40238261e-01 -8.11032593e-01 -7.18983889e-01 4.23407495e-01 -6.14178181e-01 1.38847196e+00 -1.89973640e+00 2.87476778e-01 5.12781918e-01 9.92825150e-01 -6.64119899e-01 -1.35755643e-01 6.71953917e-01 1.04685619e-01 5.75077772e-01 1.65320024e-01 -3.22926134e-01 3.93051326e-01 -2.64238060e-01 -2.92840362e-01 5.59692740e-01 -2.22939420e-02 1.44686592e+00 -9.73768890e-01 -6.76928520e-01 -2.18467176e-01 1.95274398e-01 -2.65623719e-01 -1.71387941e-02 -3.48297477e-01 -2.74211187e-02 -5.63037813e-01 6.29101276e-01 4.10754174e-01 -8.00543308e-01 9.56154764e-02 -7.48076886e-02 -2.41592318e-01 6.32049501e-01 -6.93210423e-01 1.43043375e+00 1.32345289e-01 1.05732083e+00 1.01311237e-01 -6.88144326e-01 8.23582232e-01 -4.25025851e-01 5.06144941e-01 -2.96141684e-01 2.73520887e-01 1.73695073e-01 -3.35365497e-02 -5.26218563e-02 5.48857510e-01 5.73130250e-01 -2.78966516e-01 7.21369803e-01 5.26349962e-01 -2.36463875e-01 6.85974717e-01 1.21021843e+00 1.57609701e+00 -1.32557422e-01 -8.10750201e-02 -7.14265704e-01 -7.72298202e-02 1.39938861e-01 6.56767413e-02 9.21138942e-01 2.26347417e-01 4.59044695e-01 9.05297577e-01 -1.20232981e-02 -1.22393358e+00 -1.02286756e+00 -8.45430866e-02 7.65787542e-01 -2.67097175e-01 -8.69930804e-01 -4.73746091e-01 -5.32366395e-01 5.19332170e-01 6.17540419e-01 -5.13174355e-01 4.09680866e-02 -3.85325372e-01 -4.65407312e-01 4.80150759e-01 6.94312900e-02 -8.27138126e-02 -6.74645662e-01 4.66041923e-01 -6.31152606e-03 8.84985477e-02 -1.07378793e+00 -5.21278441e-01 -1.87910661e-01 -7.41439700e-01 -1.15641463e+00 -8.81304801e-01 -8.34556937e-01 6.11317396e-01 1.15529835e-01 1.80172646e+00 9.96447727e-02 -6.64534390e-01 7.49152243e-01 -1.48483794e-02 -3.44003618e-01 -3.25101197e-01 4.77425992e-01 -2.78820753e-01 -6.14540398e-01 2.91076571e-01 -6.91244364e-01 -2.43889302e-01 -5.36398888e-01 -4.88234133e-01 -1.82353169e-01 6.69018388e-01 4.39711958e-01 3.88198227e-01 -3.47951442e-01 5.34576237e-01 -1.12551510e+00 7.89366961e-01 -9.52146292e-01 -6.56395316e-01 2.02971682e-01 -9.61455405e-01 1.45490825e-01 4.57662880e-01 -7.61673376e-02 -4.13963974e-01 -6.35258615e-01 1.89939514e-01 -6.76604956e-02 -1.24015221e-02 9.33348835e-01 5.43790534e-02 -2.21727788e-01 3.15354556e-01 1.70440916e-02 2.45689042e-02 -7.82857299e-01 5.56412876e-01 5.05600452e-01 7.90228546e-01 -7.24605978e-01 1.12104654e+00 -3.10357078e-03 5.09637773e-01 -7.85004079e-01 -7.97596574e-01 -5.11980116e-01 -3.71114224e-01 -2.62649506e-01 4.93620485e-01 -8.93077254e-01 -7.64621973e-01 1.96486577e-01 -1.17933929e+00 -1.05762277e-02 -1.38103757e-02 4.81750935e-01 8.61702114e-02 6.48096919e-01 -8.46284807e-01 -2.24271014e-01 -3.04690242e-01 -6.18189812e-01 6.20996237e-01 3.99510860e-01 -4.29969609e-01 -1.36036348e+00 2.89887637e-01 3.44693065e-01 4.68563735e-01 3.48694831e-01 1.26126671e+00 -7.36028910e-01 -7.79754102e-01 -4.43747602e-02 -5.14607906e-01 -3.37093055e-01 -3.27421665e-01 6.49913192e-01 -3.16416919e-01 -1.00559011e-01 -7.54288852e-01 1.37852952e-01 1.28290594e+00 3.19821239e-01 1.39306355e+00 -3.07601959e-01 -7.86924124e-01 5.31171978e-01 1.25711024e+00 -5.46665430e-01 3.84629399e-01 3.44168097e-01 1.03089881e+00 2.97817856e-01 -1.82087615e-01 3.24333847e-01 7.65878916e-01 1.15607150e-01 1.82361618e-01 1.28630444e-01 -2.97797054e-01 -3.13627690e-01 1.99847966e-02 1.28717756e+00 -2.67198175e-01 -6.14847124e-01 -1.24356878e+00 6.21223688e-01 -1.36466646e+00 -1.26531160e+00 -8.00968111e-01 1.69768047e+00 8.99906695e-01 3.58352095e-01 1.62899479e-01 -4.83406425e-01 7.46044517e-01 2.72839695e-01 -1.75313577e-01 -5.42124920e-02 -4.09741521e-01 4.99177366e-01 7.03797102e-01 3.41799170e-01 -7.94934034e-01 6.87550664e-01 7.20333242e+00 6.51820898e-01 -7.26680696e-01 -2.37768263e-01 4.39533442e-01 8.93528163e-02 -1.03698051e+00 1.47154957e-01 -7.86177516e-01 7.40434706e-01 1.21794069e+00 -9.67715442e-01 5.22107482e-01 6.99380040e-01 -1.91456661e-01 1.38098091e-01 -1.02817392e+00 8.29008877e-01 9.59864184e-02 -2.02385688e+00 1.77401647e-01 4.76897269e-01 8.09375525e-01 4.41054016e-01 8.21604282e-02 2.92947888e-01 6.68191671e-01 -1.02757978e+00 3.50735247e-01 8.57805014e-01 8.17249238e-01 -5.39984167e-01 3.79370093e-01 7.32021779e-02 -7.45638132e-01 3.72717828e-01 -1.81081489e-01 -3.29947211e-02 -9.78912637e-02 1.28967464e+00 -4.40436959e-01 7.62304068e-01 5.81546545e-01 1.22404146e+00 -9.04677451e-01 1.10480237e+00 -1.23057533e-02 1.04352379e+00 -5.59364483e-02 -4.52437788e-01 -7.30084535e-03 -4.78016466e-01 7.78661668e-01 1.70517921e+00 5.13166666e-01 1.15748048e-01 6.85485303e-02 1.07839143e+00 -1.02631021e+00 -4.95924577e-02 -7.18922079e-01 -1.05324507e+00 8.90194714e-01 1.59231710e+00 -9.14878488e-01 -4.75618631e-01 -3.84158731e-01 4.77828264e-01 4.78725731e-01 3.76943111e-01 -5.85829735e-01 -8.91681433e-01 4.41743881e-01 2.86924958e-01 -9.64664221e-02 -7.24054277e-01 -4.67124581e-01 -1.22643983e+00 -3.26443344e-01 -6.33042693e-01 3.42986733e-01 -7.39689529e-01 -1.88025701e+00 2.51830578e-01 -1.90481186e-01 -4.29763913e-01 5.16186878e-02 -6.61549568e-01 -7.48724461e-01 9.93346095e-01 -1.19589376e+00 -9.42752600e-01 -3.68793279e-01 9.08161849e-02 -4.91032712e-02 -4.67332810e-01 4.69584167e-01 3.29984635e-01 -6.35305703e-01 3.85744542e-01 3.23820919e-01 3.65328401e-01 8.61129582e-01 -1.49089861e+00 8.07746828e-01 4.51756269e-01 4.94460672e-01 1.06897640e+00 7.05722630e-01 -1.02947772e+00 -2.03854084e+00 -9.11979496e-01 1.30940199e+00 -8.33770931e-01 1.48220932e+00 -4.29337353e-01 -8.89523327e-01 7.51472294e-01 5.84997773e-01 -3.19303662e-01 7.80125976e-01 5.97540200e-01 -5.59354365e-01 2.58584768e-01 -5.61133921e-01 6.41795754e-01 1.43447781e+00 -5.12368739e-01 -8.41271877e-01 6.01880193e-01 5.88435113e-01 -1.45430490e-01 -1.34098089e+00 -3.01491382e-04 3.79795521e-01 -3.21931005e-01 9.15960789e-01 -6.69797421e-01 8.91767085e-01 3.12162340e-01 4.16990638e-01 -1.42579174e+00 -1.01188815e+00 -9.26591635e-01 -3.02237988e-01 1.60842907e+00 4.86076057e-01 -7.40942955e-01 5.96332371e-01 3.10520887e-01 -4.18156594e-01 -6.29449904e-01 -5.41812658e-01 -5.18659294e-01 3.16903770e-01 -1.39543071e-01 4.27148432e-01 1.34010220e+00 2.38870487e-01 6.46698654e-01 2.28348047e-01 -5.21472931e-01 1.05886793e+00 4.38952595e-01 7.21249402e-01 -2.00992417e+00 -9.80502814e-02 -1.10821664e+00 -3.86785418e-01 -7.57478416e-01 5.04504085e-01 -1.35827029e+00 -5.77560008e-01 -2.16595006e+00 6.71009421e-01 -3.99191707e-01 -1.70158431e-01 3.31888974e-01 -2.76192367e-01 1.83605254e-01 -2.17459619e-01 3.63260746e-01 -7.03926206e-01 9.46371853e-02 1.31984460e+00 -1.31636739e-01 4.29193020e-01 -5.88367462e-01 -9.85476673e-01 3.79617035e-01 5.27773738e-01 -4.15214896e-01 1.14938229e-01 -2.26707280e-01 5.76936781e-01 -4.06667739e-01 5.48988760e-01 -6.59276903e-01 4.23479974e-01 -3.45386147e-01 4.27777141e-01 -6.37356758e-01 -1.93438902e-01 -9.36549231e-02 6.28407747e-02 2.60743145e-02 -4.14207011e-01 3.26442450e-01 5.45832627e-02 7.66782880e-01 4.82369289e-02 5.25413966e-03 2.60818154e-01 -2.82853186e-01 -2.39911184e-01 4.18253332e-01 -2.95312494e-01 6.00993156e-01 6.81880534e-01 2.94831932e-01 -1.12839007e+00 -2.86188245e-01 -3.77287835e-01 1.89047769e-01 7.95919657e-01 5.45223713e-01 1.71525314e-01 -1.02119458e+00 -1.00719929e+00 -1.31982386e-01 -6.43415749e-02 -6.04295909e-01 -7.44060799e-02 6.33993328e-01 -2.56475061e-01 4.28131908e-01 -1.73670307e-01 -1.64885387e-01 -9.51973855e-01 4.20564651e-01 -4.46935177e-01 -1.47918969e-01 -6.24148726e-01 6.13083959e-01 -4.26332265e-01 -3.94395180e-02 -3.81244943e-02 1.01033384e-02 -2.56080329e-01 1.16057865e-01 -1.18741505e-02 6.11298859e-01 -8.25877562e-02 -3.41717035e-01 -3.04990470e-01 2.89136678e-01 1.42653018e-01 -5.64161353e-02 1.55401611e+00 1.52179971e-01 -6.70222998e-01 6.41569555e-01 1.28097379e+00 6.06895328e-01 -1.45922929e-01 -1.33774534e-01 2.79066652e-01 -2.00736955e-01 2.45855272e-01 -6.14935219e-01 -7.98449337e-01 6.52779698e-01 -5.11001050e-01 7.03729331e-01 1.98149636e-01 4.82610315e-01 6.55410171e-01 4.57441844e-02 2.20742077e-01 -9.32416499e-01 8.72362256e-02 7.96989679e-01 6.05318189e-01 -8.03260982e-01 3.99501085e-01 -4.37960595e-01 -6.41382579e-03 1.16475487e+00 2.70300210e-01 -2.18137115e-01 8.04556012e-01 3.43158334e-01 -5.42717516e-01 -6.77157462e-01 -9.00176764e-01 1.19581707e-01 7.32531250e-01 8.38191137e-02 7.60228693e-01 -4.79519516e-02 -5.25232375e-01 6.42600238e-01 -4.83616143e-01 -2.19450757e-01 7.62548387e-01 5.29638588e-01 -4.38030958e-01 -1.03005314e+00 -2.12523878e-01 1.20479250e+00 -6.05063140e-01 -2.52017587e-01 -9.91330683e-01 5.92061818e-01 -4.43771929e-01 6.35245144e-01 3.22124183e-01 -2.22867250e-01 -9.16974470e-02 1.90779731e-01 5.45909703e-01 -5.85767627e-01 -5.09102821e-01 -6.32812008e-02 2.14808643e-01 -1.15710437e-01 -6.78406730e-02 -7.06128478e-01 -1.13905215e+00 -9.89091754e-01 9.90693644e-03 4.66464460e-01 8.49856973e-01 6.59614623e-01 9.81761992e-01 8.70056510e-01 2.85612077e-01 -5.63471258e-01 -2.83876985e-01 -1.03655529e+00 -5.33079624e-01 4.34908837e-01 -1.06114313e-01 -4.93041873e-01 -8.42484295e-01 7.92571530e-02]
[9.561223983764648, 8.159749031066895]
8f839d65-f2d8-44e2-94c5-34603db306f7
what-makes-a-good-dataset-for-symbol
2304.08352
null
https://arxiv.org/abs/2304.08352v1
https://arxiv.org/pdf/2304.08352v1.pdf
What Makes a Good Dataset for Symbol Description Reading?
The usage of mathematical formulas as concise representations of a document's key ideas is common practice. Correctly interpreting these formulas, by identifying mathematical symbols and extracting their descriptions, is an important task in document understanding. This paper makes the following contributions to the mathematical identifier description reading (MIDR) task: (i) introduces the Math Formula Question Answering Dataset (MFQuAD) with $7508$ annotated identifier occurrences; (ii) describes novel variations of the noun phrase ranking approach for the MIDR task; (iii) reports experimental results for the SOTA noun phrase ranking approach and our novel variations of the approach, providing problem insights and a performance baseline; (iv) provides a position on the features that make an effective dataset for the MIDR task.
['Bradley Eck', 'Joern Ploennigs', 'Karol Lynch']
2023-04-17
null
null
null
null
['phrase-ranking']
['natural-language-processing']
[ 3.75384331e-01 1.24300756e-01 -3.20010424e-01 -6.08002007e-01 -1.26733685e+00 -8.54209721e-01 9.13865566e-01 7.71036625e-01 -1.71571881e-01 5.17748415e-01 3.96347612e-01 -8.89676273e-01 -5.66494524e-01 -8.47607970e-01 -7.65549541e-01 3.33631933e-01 1.09910265e-01 8.50573778e-01 7.90416449e-02 -7.96047986e-01 1.14843333e+00 4.23942715e-01 -1.35971034e+00 9.17072296e-01 7.14832783e-01 1.03911293e+00 1.76027641e-02 1.14019489e+00 -8.04820955e-01 1.59193087e+00 -8.72746944e-01 -8.63595784e-01 -2.83959299e-01 -2.21720934e-02 -1.54765868e+00 -3.62252742e-01 1.09904408e+00 -5.66031039e-01 -2.01346442e-01 7.31572330e-01 2.20389124e-02 2.47467294e-01 9.88092542e-01 -9.89248574e-01 -1.10374582e+00 1.03675425e+00 -3.33214402e-01 4.22066242e-01 1.10287488e+00 -4.58392084e-01 1.68468511e+00 -7.62276769e-01 6.98727250e-01 1.33005869e+00 7.98023045e-01 5.18741786e-01 -8.44766140e-01 -3.19342405e-01 7.04634050e-03 2.15705365e-01 -1.56751704e+00 -3.19339842e-01 5.16129315e-01 -2.21145377e-01 1.84754539e+00 6.90042078e-01 2.92904913e-01 4.79577631e-01 -1.84132278e-01 8.37573826e-01 6.79819405e-01 -7.35112250e-01 -2.65466273e-01 1.92697957e-01 1.22034740e+00 8.38836610e-01 1.88390553e-01 -5.94508410e-01 -5.24608970e-01 -4.35731173e-01 5.04957378e-01 -3.38242888e-01 -2.50941873e-01 2.88718939e-01 -1.02384043e+00 8.67347300e-01 1.54703828e-02 3.75928581e-01 1.15799590e-03 7.67096937e-01 5.19256592e-01 2.98257560e-01 2.68883072e-02 1.11609781e+00 -5.62278330e-01 -6.16883576e-01 -7.55513728e-01 1.04111981e+00 9.68278170e-01 1.55350995e+00 2.94784307e-01 -4.47248161e-01 -2.64234424e-01 9.33369815e-01 5.89920171e-02 5.60453773e-01 1.57619759e-01 -1.19991648e+00 9.13613141e-01 5.63351810e-01 4.90453213e-01 -1.12318134e+00 -2.13612273e-01 -8.74693766e-02 -4.72585745e-02 -7.61935174e-01 6.02049589e-01 4.82614607e-01 -4.50868666e-01 1.27551472e+00 -5.73312819e-01 -3.92368406e-01 2.33280152e-01 4.34196085e-01 1.58458531e+00 8.72496545e-01 5.22051230e-02 4.82588261e-02 1.78867960e+00 -6.65424287e-01 -8.27962995e-01 1.45234719e-01 9.95916486e-01 -7.57567644e-01 1.25710666e+00 2.47497752e-01 -1.33631432e+00 -4.99546349e-01 -9.06476796e-01 -1.02106810e+00 -9.18715656e-01 2.13696674e-01 1.09404075e+00 5.89204550e-01 -9.06324863e-01 3.51972997e-01 -1.24808609e-01 -4.13882919e-02 2.48105749e-01 2.34878540e-01 -4.38197292e-02 -3.42927188e-01 -1.07155252e+00 1.06126380e+00 3.87460798e-01 -3.13862622e-01 -1.40928715e-01 -1.23074043e+00 -1.12495184e+00 1.91400588e-01 2.75006354e-01 -8.05953503e-01 1.58019876e+00 -4.95325297e-01 -9.61772799e-01 1.10007918e+00 -4.23070848e-01 -5.26930630e-01 6.49371604e-03 -2.90720284e-01 -2.86361665e-01 2.89214879e-01 1.10442704e-02 6.89782917e-01 5.78138053e-01 -1.08309364e+00 -6.25239313e-01 1.62096992e-02 7.52492547e-01 1.49017617e-01 1.53653501e-02 4.18586969e-01 -2.75091618e-01 -8.40048015e-01 9.54853296e-02 -2.34885365e-01 5.46237826e-01 -2.78110355e-01 -2.47751281e-01 -9.93040800e-01 1.97944358e-01 -8.77042472e-01 1.71038783e+00 -1.79621863e+00 -7.76966140e-02 1.65737987e-01 5.38340390e-01 -1.78253770e-01 5.16414037e-03 6.75245881e-01 -1.74226880e-01 5.32198250e-01 -1.45459101e-01 -1.06730811e-01 5.49137712e-01 -8.89392495e-02 -1.34092009e+00 -3.65049452e-01 1.17026351e-01 1.30162394e+00 -9.21976566e-01 -3.59281480e-01 -6.67307153e-02 1.75968185e-01 -2.42561191e-01 -1.61263514e-02 -7.71182716e-01 -5.81224859e-01 -3.91125798e-01 8.21776330e-01 3.54391843e-01 -3.34259152e-01 -1.68137357e-01 -2.96695352e-01 -2.19776686e-02 9.29951310e-01 -7.65967548e-01 1.18421531e+00 -4.86642540e-01 7.91338742e-01 -6.95488930e-01 -6.59106433e-01 8.55278432e-01 1.35005534e-01 -6.15169071e-02 -5.95722914e-01 9.02442038e-02 6.34972215e-01 -3.77766967e-01 -5.62021494e-01 9.81664360e-01 4.45730388e-02 -4.27687049e-01 5.03084898e-01 -2.88403243e-01 -8.18818808e-01 7.45622873e-01 6.32433832e-01 1.30049062e+00 -2.73148455e-02 4.58801061e-01 -3.29996526e-01 8.17627132e-01 2.74415582e-01 -3.64885241e-01 1.29425335e+00 4.16023821e-01 3.44410092e-01 4.40734059e-01 -8.47333729e-01 -9.32875991e-01 -1.08297169e+00 -5.73231518e-01 8.94013762e-01 2.76470594e-02 -1.01784277e+00 -2.50187069e-01 -6.19187772e-01 4.73804772e-01 1.32385731e+00 -5.09839177e-01 2.18021289e-01 -9.51677740e-01 -3.32129836e-01 9.44741070e-01 9.90862310e-01 5.80302954e-01 -5.55891752e-01 -4.92491931e-01 3.36371027e-02 -4.87038463e-01 -1.03522241e+00 -3.38423073e-01 -3.22259553e-02 -8.52300465e-01 -1.15717351e+00 -5.63024044e-01 -7.59319782e-01 6.59722865e-01 1.86529011e-01 1.90409327e+00 5.75636506e-01 -2.85071462e-01 1.06809878e+00 -5.30153096e-01 -6.46442890e-01 -5.12253523e-01 1.30399778e-01 -5.67132056e-01 -1.03331637e+00 9.75575268e-01 9.00599211e-02 2.70074666e-01 -4.05513376e-01 -8.32784832e-01 1.95982363e-02 2.04817817e-01 7.44068086e-01 3.60815912e-01 6.84889331e-02 5.07661998e-01 -1.10846210e+00 9.84146833e-01 -1.04637004e-01 -5.85534692e-01 8.71884584e-01 -3.32954496e-01 1.98574111e-01 5.86797535e-01 3.39286216e-02 -8.57631862e-01 -6.00234985e-01 -2.38684475e-01 3.64788562e-01 3.67350698e-01 5.92170238e-01 3.34288068e-02 -6.26694039e-02 4.81100023e-01 2.44301841e-01 -3.43810171e-01 -3.94660652e-01 5.88229656e-01 7.96311140e-01 5.23136497e-01 -1.19010746e+00 7.43527353e-01 5.57468310e-02 1.12063408e-01 -9.85117257e-01 -9.86333549e-01 -7.43568003e-01 -5.08163571e-01 8.42528865e-02 4.74082440e-01 -6.20153010e-01 -8.35787416e-01 2.34494254e-01 -1.60701525e+00 2.77232435e-02 -3.66149843e-01 -1.44229690e-02 -7.86489725e-01 4.82549876e-01 -5.72814167e-01 -8.55676770e-01 -3.47592264e-01 -7.62810886e-01 1.28123868e+00 2.86436211e-02 -7.64779866e-01 -1.16042995e+00 -3.37512583e-01 8.65078986e-01 3.49620223e-01 5.14335893e-02 1.82693577e+00 -6.25784338e-01 -8.04470062e-01 -1.95510507e-01 -6.63887799e-01 1.73089027e-01 -6.51710033e-02 3.38158682e-02 -6.47129238e-01 3.50310355e-01 -1.43628523e-01 -6.33543313e-01 8.62024009e-01 2.42500678e-02 1.62645948e+00 -3.75675291e-01 -3.24930437e-03 1.56568199e-01 1.46868694e+00 1.37447387e-01 7.16666400e-01 3.60094130e-01 4.91672128e-01 5.29925168e-01 5.68976223e-01 1.54327661e-01 8.52204144e-01 5.83703697e-01 -3.78986746e-02 6.03157282e-01 -2.70117164e-01 -2.60336369e-01 -3.24090607e-02 7.94973791e-01 -1.05157197e-01 -1.76644787e-01 -1.08692896e+00 4.60165471e-01 -1.52743387e+00 -8.69267523e-01 -2.23484516e-01 1.87749684e+00 1.24278796e+00 1.13223352e-01 -6.49650767e-02 2.77096540e-01 1.65770307e-01 4.63287607e-02 8.12248606e-03 -1.01501715e+00 -9.48064327e-02 6.80223346e-01 4.14828628e-01 6.64133549e-01 -1.03305662e+00 7.72307456e-01 7.33101606e+00 9.26811218e-01 -3.25277120e-01 -3.24956536e-01 3.66042197e-01 4.99049425e-01 -6.64667904e-01 -1.60907164e-01 -1.23217690e+00 -1.52830323e-02 9.18223619e-01 -4.06304270e-01 6.47832692e-01 7.49421597e-01 -5.60475528e-01 -2.53000781e-02 -1.54464734e+00 1.39116085e+00 6.19212091e-01 -1.85980189e+00 9.26719487e-01 -6.46240294e-01 2.16207653e-01 -7.74107099e-01 1.43118218e-01 5.95092773e-01 -1.90363992e-02 -1.50933743e+00 8.76911581e-01 8.17877114e-01 7.50675142e-01 -7.33117282e-01 8.97040308e-01 9.71574187e-02 -1.10628283e+00 -2.71467924e-01 -5.13919175e-01 -3.90289903e-01 -1.20185837e-01 9.22769904e-02 -5.56834102e-01 5.24616539e-01 4.06763077e-01 7.85232544e-01 -1.02092779e+00 5.27434468e-01 -1.62469909e-01 2.99650013e-01 -6.55201077e-02 -7.64071941e-01 1.42842487e-01 -1.13338284e-01 5.02433836e-01 1.44320512e+00 -1.22548118e-01 3.52163881e-01 -4.09647375e-01 1.20838642e+00 -5.12178838e-01 5.07097915e-02 -5.16916752e-01 -5.31855285e-01 5.35073102e-01 6.17738664e-01 -4.37317997e-01 -6.39590979e-01 -4.55314994e-01 5.89705110e-01 2.95107931e-01 7.74830356e-02 -7.41367340e-01 -1.11173642e+00 2.11881340e-01 -7.04153702e-02 9.68301669e-02 -2.48248756e-01 -5.19328415e-01 -1.10549843e+00 3.66129249e-01 -1.11365330e+00 6.42207086e-01 -1.23838341e+00 -1.31108761e+00 1.80311084e-01 4.90706295e-01 -6.77714825e-01 -2.83434629e-01 -1.07174325e+00 -1.96089625e-01 8.94363403e-01 -1.74580264e+00 -1.11415565e+00 -1.90641105e-01 -1.45355359e-01 6.10008955e-01 -1.28007874e-01 1.35559785e+00 2.69512832e-01 1.01043746e-01 8.57319653e-01 6.42289519e-02 3.56403559e-01 2.95758814e-01 -1.79694045e+00 3.80635560e-01 4.20420587e-01 5.55604994e-01 1.45451212e+00 6.23451650e-01 -3.41636151e-01 -1.66399765e+00 -7.38049388e-01 1.52789164e+00 -1.29582012e+00 7.49850988e-01 -3.40221286e-01 -9.76448953e-01 6.97402775e-01 -9.01023373e-02 -5.76217830e-01 7.73591220e-01 5.14332540e-02 -9.02719021e-01 -1.60284266e-01 -1.20144331e+00 6.78863049e-01 8.91523063e-01 -8.70572329e-01 -1.33834171e+00 6.85336769e-01 9.19634640e-01 -7.92920768e-01 -1.04376519e+00 1.28419921e-01 6.85931325e-01 -3.22236925e-01 1.09567237e+00 -8.54224563e-01 8.17674339e-01 -2.74993926e-01 -4.24877077e-01 -3.92777741e-01 4.45249081e-02 -5.79839587e-01 -5.95382273e-01 1.27714038e+00 5.08221745e-01 -4.04380172e-01 3.13225508e-01 1.06957853e+00 5.02166338e-02 -9.56642270e-01 -9.16222215e-01 -6.91516638e-01 5.01835465e-01 -6.66385174e-01 9.39494491e-01 6.41939461e-01 2.11108834e-01 3.86752874e-01 4.76432204e-01 -1.53265923e-01 2.54350334e-01 2.39416778e-01 5.62195718e-01 -8.41069758e-01 -2.70619452e-01 -6.81094766e-01 -5.53288639e-01 -1.69702268e+00 2.90000707e-01 -1.26428938e+00 -4.11093920e-01 -1.61201274e+00 4.54926401e-01 -1.94571942e-01 2.36718550e-01 2.92481214e-01 1.42311184e-02 3.48623283e-02 1.55288100e-01 1.13227546e-01 -8.08791041e-01 7.05625415e-02 7.80771732e-01 -6.49574280e-01 3.72828931e-01 -1.00364581e-01 -1.07171333e+00 6.29681408e-01 2.51278669e-01 -9.72412750e-02 -5.25499940e-01 -8.21027040e-01 8.20349276e-01 -7.41197765e-02 4.35883164e-01 -4.48473632e-01 4.18003276e-02 1.69645194e-02 2.00452998e-01 -9.51652110e-01 1.54734090e-01 -7.58013248e-01 -6.85360610e-01 1.77382663e-01 -9.06684995e-01 6.40963614e-01 2.84853429e-01 1.60874024e-01 -1.28637969e-01 -9.96527910e-01 4.52981234e-01 -3.35744888e-01 -8.41762722e-01 -3.80688667e-01 -2.35411048e-01 6.48436844e-01 4.23557132e-01 -2.40834087e-01 -7.32530773e-01 -3.23655128e-01 -6.43065339e-03 3.36338460e-01 1.63714588e-01 5.03895819e-01 8.15324247e-01 -1.04591227e+00 -5.46731353e-01 -3.83268923e-01 3.93563986e-01 3.09689101e-02 -4.00241286e-01 1.82373479e-01 -1.09565222e+00 1.09608197e+00 4.25490022e-01 -1.38063490e-01 -1.25273490e+00 2.47908235e-01 3.67720664e-01 -2.88981587e-01 -6.41010046e-01 1.09610593e+00 -4.68598753e-01 -4.38545287e-01 5.42961836e-01 -9.23418939e-01 -6.07478142e-01 -2.37401485e-01 1.05111814e+00 5.79713047e-01 2.15349898e-01 -4.10425901e-01 -3.73756886e-01 8.23191822e-01 -4.66857702e-01 8.44458118e-02 1.10571814e+00 -5.16334772e-02 -7.44683743e-01 6.72423005e-01 1.35300422e+00 1.95422918e-02 2.22404599e-02 -5.75137399e-02 5.41740894e-01 -4.76357013e-01 -1.12467386e-01 -1.16820562e+00 -1.25901671e-02 8.11667681e-01 -4.37965505e-02 2.76505828e-01 7.93328643e-01 3.56533565e-02 1.15118647e+00 1.19286227e+00 3.75428438e-01 -1.09299242e+00 1.16992109e-01 9.62540865e-01 1.19930685e+00 -9.78671432e-01 1.96170047e-01 -6.00018561e-01 -2.57025123e-01 1.90888453e+00 1.03806987e-01 1.31385833e-01 3.49370748e-01 6.95855245e-02 -1.66451514e-01 -7.41669536e-01 -6.04100764e-01 1.29230514e-01 4.93306249e-01 5.42149246e-01 7.13940084e-01 -3.40907365e-01 -3.34633917e-01 1.12890065e+00 -6.91479087e-01 -1.15716495e-01 7.12796450e-01 1.05387795e+00 -1.36590451e-01 -9.04087842e-01 -1.39408797e-01 9.68112230e-01 -4.94884580e-01 -5.40239990e-01 -6.85801387e-01 9.90439951e-01 -2.53206700e-01 8.71030688e-01 1.35112688e-01 1.22214176e-01 4.35869575e-01 3.56112838e-01 9.39014852e-01 -9.17896271e-01 -7.10057914e-01 -9.38738585e-01 4.70097244e-01 -1.39092267e-01 -2.31469557e-01 -5.27711570e-01 -1.51033318e+00 -6.02226436e-01 -3.64750952e-01 4.23995733e-01 3.78181905e-01 9.79882896e-01 1.80352237e-02 5.23659468e-01 1.67404190e-01 1.48027524e-01 -1.09105158e+00 -8.17720234e-01 -3.15121293e-01 4.92869437e-01 1.39389560e-01 -4.24643576e-01 -4.55870688e-01 1.22212149e-01]
[9.59323787689209, 7.50109338760376]
f3343604-3707-4b30-a4f7-26e2de041cbe
shuffle-and-learn-unsupervised-learning-using
1603.08561
null
http://arxiv.org/abs/1603.08561v2
http://arxiv.org/pdf/1603.08561v2.pdf
Shuffle and Learn: Unsupervised Learning using Temporal Order Verification
In this paper, we present an approach for learning a visual representation from the raw spatiotemporal signals in videos. Our representation is learned without supervision from semantic labels. We formulate our method as an unsupervised sequential verification task, i.e., we determine whether a sequence of frames from a video is in the correct temporal order. With this simple task and no semantic labels, we learn a powerful visual representation using a Convolutional Neural Network (CNN). The representation contains complementary information to that learned from supervised image datasets like ImageNet. Qualitative results show that our method captures information that is temporally varying, such as human pose. When used as pre-training for action recognition, our method gives significant gains over learning without external data on benchmark datasets like UCF101 and HMDB51. To demonstrate its sensitivity to human pose, we show results for pose estimation on the FLIC and MPII datasets that are competitive, or better than approaches using significantly more supervision. Our method can be combined with supervised representations to provide an additional boost in accuracy.
['Martial Hebert', 'C. Lawrence Zitnick', 'Ishan Misra']
2016-03-28
null
null
null
null
['self-supervised-action-recognition', 'video-alignment']
['computer-vision', 'computer-vision']
[ 4.44443256e-01 -1.06184445e-02 -3.51998180e-01 -6.45513892e-01 -7.52312005e-01 -7.10951447e-01 6.48951411e-01 -1.52623683e-01 -5.64091206e-01 6.69784307e-01 4.04469818e-01 2.62729019e-01 2.91727453e-01 -1.80008993e-01 -1.13040102e+00 -5.19798696e-01 -3.26484352e-01 2.95982361e-01 3.85119051e-01 -7.13272393e-02 -5.22904918e-02 3.31635535e-01 -1.61263621e+00 5.94487548e-01 -5.43096699e-02 1.25590885e+00 -9.58579965e-03 7.32695043e-01 6.40218079e-01 1.27481854e+00 -5.57691097e-01 9.34205130e-02 2.90266365e-01 -3.50323498e-01 -1.21952224e+00 3.94162506e-01 8.43035161e-01 -6.03959918e-01 -6.47882581e-01 6.39556348e-01 2.41038278e-01 1.86790168e-01 6.75794125e-01 -1.29853356e+00 -1.75703585e-01 8.44934955e-02 -3.60285938e-01 3.25132340e-01 7.67715096e-01 3.88125896e-01 1.03425729e+00 -7.73749888e-01 1.01768291e+00 1.34553039e+00 6.54958308e-01 4.96262670e-01 -1.34619462e+00 -4.20404017e-01 3.00248206e-01 3.81031901e-01 -1.21660447e+00 -5.98088920e-01 5.77995598e-01 -7.08308399e-01 9.81041789e-01 3.85211594e-02 4.88257170e-01 1.40372312e+00 -1.32401481e-01 1.05837929e+00 8.27053010e-01 -2.10616142e-01 8.03333148e-02 -3.16316515e-01 -8.40270743e-02 7.40725160e-01 -7.58463517e-02 3.84871483e-01 -7.98315823e-01 2.35852838e-01 7.76133955e-01 -1.09658111e-02 -4.19050038e-01 -7.68704474e-01 -1.45656872e+00 5.55318534e-01 7.15777338e-01 7.55517334e-02 -1.17693596e-01 6.80649400e-01 6.43956721e-01 3.23175043e-01 2.73338020e-01 4.54588681e-01 -5.45964062e-01 -2.40516916e-01 -9.47998643e-01 2.17087343e-01 4.96825457e-01 7.45030820e-01 6.15246654e-01 -7.18730241e-02 -2.99581438e-01 4.36308324e-01 1.70530528e-01 5.14893293e-01 3.74976069e-01 -1.28319097e+00 4.28640425e-01 1.93003222e-01 3.50794464e-01 -1.01875675e+00 -5.49158573e-01 -3.47513258e-02 -4.05378848e-01 2.74569422e-01 6.65899932e-01 -1.48699582e-01 -1.25437975e+00 1.97174048e+00 5.83143644e-02 3.34253937e-01 -3.70781086e-02 1.23969305e+00 8.12450051e-01 6.18785381e-01 1.39854074e-01 1.67493219e-03 1.12628007e+00 -9.01737869e-01 -5.43436050e-01 -3.15687746e-01 5.97533882e-01 -4.20332044e-01 6.04598165e-01 5.10024667e-01 -7.45411158e-01 -8.44223559e-01 -1.04986346e+00 2.04027668e-02 -2.07576171e-01 3.75845462e-01 6.02394700e-01 2.28932619e-01 -1.12431145e+00 7.31466055e-01 -1.08710527e+00 -6.47202253e-01 4.72070873e-01 3.63620549e-01 -7.76556969e-01 -4.14609984e-02 -9.79438126e-01 7.67393827e-01 3.93847972e-01 -2.94159516e-03 -1.63416326e+00 -4.07364696e-01 -1.26583731e+00 -2.59972960e-01 5.29480934e-01 -2.76701838e-01 1.45274580e+00 -1.33566773e+00 -1.20477736e+00 9.27008927e-01 -1.30536720e-01 -7.75236785e-01 5.63276708e-01 -4.48946238e-01 -8.93563405e-02 6.67875886e-01 1.35081649e-01 1.10850036e+00 1.01429331e+00 -1.19146407e+00 -5.37525892e-01 -2.26610795e-01 3.20177704e-01 5.93812354e-02 -4.30266708e-02 9.76618677e-02 -6.47151828e-01 -6.70656085e-01 -1.51535720e-01 -1.26205373e+00 -6.57367110e-02 2.65016884e-01 -3.03323567e-01 -7.93120041e-02 9.93819296e-01 -8.12425613e-01 7.02689469e-01 -2.14657354e+00 4.38244134e-01 2.81008519e-02 -6.87507614e-02 1.71506241e-01 -2.33073115e-01 2.74229825e-01 -2.33114958e-01 -2.01625854e-01 -1.24879263e-01 -2.77432829e-01 -1.91879645e-01 3.56562763e-01 -2.49879524e-01 7.00546563e-01 5.26155829e-01 1.09286392e+00 -1.00131142e+00 -3.56358439e-01 2.29366407e-01 3.64762843e-01 -6.01424515e-01 3.65492612e-01 -3.83888602e-01 8.24659050e-01 -2.75704209e-02 5.47399402e-01 3.76198366e-02 -6.27616048e-01 2.95858890e-01 -4.73572135e-01 3.59138429e-01 2.18522429e-01 -1.00456953e+00 2.15030289e+00 -2.33936206e-01 1.02801132e+00 -9.69880149e-02 -1.38167417e+00 4.93391544e-01 4.79538560e-01 6.41952932e-01 -7.05442131e-01 6.94679394e-02 -1.88238725e-01 -2.20106304e-01 -5.85590899e-01 1.57037601e-01 9.29916501e-02 -2.18101144e-01 2.80694306e-01 6.69507980e-01 1.55900523e-01 3.08550924e-01 3.40861529e-01 1.24349892e+00 5.94373882e-01 8.75864178e-02 -8.15525185e-03 2.93823540e-01 1.09095931e-01 5.43424547e-01 5.40277064e-01 -3.21945339e-01 7.92315304e-01 4.81786191e-01 -5.77093720e-01 -8.80927682e-01 -1.02613330e+00 3.46083902e-02 1.11021709e+00 2.73819685e-01 -6.73289478e-01 -4.34053749e-01 -9.05002773e-01 9.56097394e-02 -9.43842828e-02 -9.62701201e-01 -4.09870408e-02 -5.66956520e-01 -1.74481884e-01 4.41464275e-01 1.04137444e+00 4.32477325e-01 -1.06980872e+00 -9.83713627e-01 -7.49268383e-02 -3.59268069e-01 -1.52739155e+00 -3.18754941e-01 3.06117684e-01 -7.68533826e-01 -1.39471495e+00 -7.55153716e-01 -7.78677464e-01 6.57292783e-01 5.92436045e-02 1.23368728e+00 -7.46718124e-02 -5.37045181e-01 8.42312098e-01 -4.13353860e-01 -5.74885979e-02 3.00116763e-02 -1.87028572e-01 8.52206871e-02 1.39560446e-01 1.07082151e-01 -2.94331014e-01 -6.84421897e-01 4.58532453e-01 -6.55929446e-01 2.09523216e-02 2.65463412e-01 8.30456316e-01 4.25351769e-01 -2.60613680e-01 2.09143549e-01 -6.72873616e-01 -1.11486211e-01 -1.33850068e-01 -4.44857061e-01 4.70289625e-02 -7.75601808e-03 2.15479121e-01 3.69238734e-01 -3.68348271e-01 -8.82421374e-01 7.56288528e-01 2.81658679e-01 -7.27954805e-01 -4.19176787e-01 3.47851276e-01 1.99203677e-02 -1.08510628e-01 6.72462761e-01 3.81111689e-02 1.83367640e-01 -5.72216213e-01 3.54735434e-01 2.70949960e-01 8.98230791e-01 -4.65687960e-01 6.56566262e-01 7.08795071e-01 -2.39836685e-02 -5.68624973e-01 -1.00547314e+00 -6.57344937e-01 -9.40278947e-01 -3.47331345e-01 1.13251817e+00 -1.31721425e+00 -6.39619470e-01 3.56575906e-01 -1.03160071e+00 -6.15817189e-01 -1.27906904e-01 5.70131242e-01 -9.32062745e-01 2.61426270e-01 -6.87074006e-01 -5.28486848e-01 2.74443299e-01 -1.01218855e+00 1.44425857e+00 -1.97070763e-02 -4.25781757e-01 -8.13963056e-01 7.89481997e-02 4.36959505e-01 3.14924829e-02 6.88589692e-01 1.72412321e-01 -5.64640284e-01 -6.11678660e-01 -1.73629507e-01 -1.36795804e-01 5.40868640e-01 2.25811929e-01 -1.33674800e-01 -1.02893651e+00 -6.56260669e-01 -4.61209863e-01 -1.05486858e+00 1.15572035e+00 2.49556124e-01 1.18995178e+00 -2.68931806e-01 -3.33456308e-01 5.86905837e-01 1.23894405e+00 3.88017334e-02 5.67221761e-01 2.06680804e-01 8.58039320e-01 5.72187006e-01 8.22204351e-01 3.99812341e-01 1.93921953e-01 9.75441396e-01 5.13555229e-01 -2.20315650e-01 -2.56820589e-01 -2.72630513e-01 5.18685520e-01 1.44649893e-01 -2.28731692e-01 5.86486571e-02 -9.80067849e-01 5.49354136e-01 -1.99143136e+00 -1.15842795e+00 4.00882095e-01 2.12197447e+00 7.14098454e-01 9.04352665e-02 4.41892415e-01 2.03126609e-01 4.37234372e-01 2.75330335e-01 -4.81884152e-01 6.14473782e-03 1.00853600e-01 1.38732955e-01 5.31634331e-01 3.36722255e-01 -1.69974959e+00 9.01654065e-01 7.05059528e+00 2.34481201e-01 -1.33589721e+00 3.29070836e-02 6.26608789e-01 -2.82147110e-01 3.54945958e-01 -2.21415997e-01 -4.44363356e-01 3.59121889e-01 8.45049977e-01 3.08351249e-01 2.28168100e-01 6.55635655e-01 1.13371231e-01 -1.19101748e-01 -1.65275443e+00 1.17982078e+00 3.52003068e-01 -1.31050825e+00 -2.11465672e-01 -8.31785873e-02 7.40097463e-01 1.57080129e-01 -8.27811807e-02 3.27058911e-01 4.19767201e-01 -1.33392870e+00 7.65823066e-01 4.95783329e-01 9.52055395e-01 -5.61118186e-01 6.12363517e-01 7.39853978e-02 -1.26319802e+00 -6.42931834e-02 -8.14077538e-03 -2.14155301e-01 1.70802101e-01 -5.71474060e-02 -7.00153232e-01 4.55634862e-01 9.31362987e-01 1.42662132e+00 -8.53484511e-01 1.08544362e+00 -4.77141857e-01 5.90461493e-01 -3.02681655e-01 4.04038221e-01 2.44417742e-01 4.27646577e-01 2.54830450e-01 1.23071587e+00 -1.15935698e-01 3.26071009e-02 6.03187203e-01 3.69341612e-01 -7.20860437e-02 -2.74134308e-01 -8.67726147e-01 -1.55019969e-01 1.04407948e-02 9.76512849e-01 -5.77435613e-01 -5.45738399e-01 -3.80102813e-01 1.25859022e+00 3.26928049e-01 5.48215628e-01 -8.82144094e-01 -8.96372721e-02 5.90647221e-01 9.59309116e-02 6.41828656e-01 -3.69368017e-01 2.69425958e-01 -1.28144526e+00 -2.70702112e-02 -9.57502186e-01 5.49710453e-01 -9.56691325e-01 -8.92266333e-01 3.84833843e-01 1.84925050e-01 -1.51557493e+00 -6.30677164e-01 -9.70756531e-01 -3.18564862e-01 2.49422744e-01 -1.19798708e+00 -1.06494081e+00 -5.20576835e-01 7.64537632e-01 4.85648632e-01 -5.35224266e-02 6.18158162e-01 2.69228190e-01 -2.79702842e-01 4.59345490e-01 -2.89407820e-01 6.90008819e-01 8.19162011e-01 -1.16904342e+00 1.42695323e-01 8.63227427e-01 6.65367365e-01 2.15602338e-01 6.23174667e-01 -3.22684437e-01 -1.29426980e+00 -1.17931283e+00 4.91779596e-01 -8.23614299e-01 4.88558054e-01 -4.09607977e-01 -5.39345920e-01 1.03389478e+00 1.06119268e-01 5.19514024e-01 4.72101390e-01 8.10000896e-02 -5.93049586e-01 -7.30153695e-02 -7.80813992e-01 1.41310021e-01 1.26717961e+00 -7.76562154e-01 -6.52978301e-01 4.57970679e-01 6.48520887e-01 -6.10714436e-01 -7.66774893e-01 6.97233617e-01 5.63338459e-01 -6.81600869e-01 1.15880775e+00 -9.90656376e-01 5.46278059e-01 -4.73810434e-01 -2.37761557e-01 -1.17618370e+00 -1.84831679e-01 -3.49466890e-01 -1.24394566e-01 7.31058359e-01 1.91296846e-01 -1.72260322e-03 7.80490994e-01 4.19400632e-01 3.33176106e-01 -3.81288022e-01 -8.65959048e-01 -1.05733228e+00 -3.07978511e-01 -3.41327816e-01 9.22298431e-02 7.73196161e-01 1.19638424e-02 3.51469606e-01 -7.08223999e-01 1.61971524e-01 6.06893837e-01 9.28888395e-02 9.45696294e-01 -1.00458753e+00 -3.76996517e-01 8.08393210e-02 -1.06700695e+00 -1.20572448e+00 3.27047467e-01 -7.15182185e-01 3.44024837e-01 -1.33561385e+00 3.20286959e-01 8.71813670e-02 -5.67123711e-01 9.18197274e-01 2.01491937e-01 6.90411031e-01 4.56299275e-01 2.59596616e-01 -1.23153281e+00 4.85975802e-01 9.62249637e-01 -4.70127136e-01 2.00997233e-01 -4.61071670e-01 -1.51956856e-01 7.25675941e-01 5.52894592e-01 -3.01644921e-01 -5.55322766e-01 -4.77744490e-01 -1.65804312e-01 2.14240387e-01 8.11923206e-01 -1.37664235e+00 8.51016715e-02 9.55658704e-02 7.82926261e-01 -3.73223394e-01 4.98817235e-01 -8.22475255e-01 3.74217983e-03 5.16605139e-01 -6.27178133e-01 -8.31606686e-02 2.97872841e-01 8.64226639e-01 -4.76346493e-01 2.60021955e-01 6.06678545e-01 -8.26939568e-03 -1.26348579e+00 3.06600273e-01 -2.57361591e-01 2.89754778e-01 1.04750633e+00 2.63040550e-02 -1.45530224e-01 -7.39312232e-01 -1.15089250e+00 2.81822920e-01 5.19918501e-01 7.73843169e-01 5.43464482e-01 -1.53068852e+00 -4.72636372e-01 9.79491100e-02 4.33431387e-01 -2.33204514e-01 -8.97328183e-02 7.79240131e-01 -4.80865180e-01 6.25508189e-01 -4.52611238e-01 -1.11521292e+00 -1.42664754e+00 6.35453403e-01 3.45956057e-01 -3.32962349e-02 -6.18777573e-01 8.53390455e-01 3.35593134e-01 -8.85793790e-02 6.26941741e-01 -3.89402598e-01 -1.65916994e-01 6.47537485e-02 5.91366172e-01 -1.76948771e-01 -7.89437369e-02 -1.02262163e+00 -7.44657815e-01 4.25616652e-01 1.23647831e-01 -2.57004589e-01 1.19011605e+00 1.31789491e-01 3.12136829e-01 5.41837454e-01 1.56685996e+00 -5.05403697e-01 -1.80526292e+00 -2.80802250e-01 8.42415094e-02 -5.54250717e-01 -1.62085667e-01 -8.04307163e-01 -1.30640459e+00 7.53041685e-01 7.26381779e-01 -2.81024069e-01 9.80134845e-01 2.84479171e-01 3.98062915e-01 5.65126717e-01 6.44373477e-01 -1.16847658e+00 6.47509217e-01 4.32981312e-01 9.56738651e-01 -1.72387278e+00 -4.72541302e-02 -1.79246113e-01 -8.31612289e-01 9.62745607e-01 7.05981970e-01 -3.35018873e-01 5.38540602e-01 8.37049410e-02 9.45822373e-02 -3.31045985e-01 -8.65262866e-01 -5.30668616e-01 5.52562296e-01 5.91497898e-01 4.32318240e-01 -1.30803138e-01 1.45725384e-01 1.77512273e-01 1.35284141e-01 2.49203727e-01 1.45858288e-01 1.11408770e+00 -2.97091782e-01 -7.73468912e-01 -9.54520404e-02 1.85832232e-01 -3.66508663e-01 2.61363775e-01 -4.85301316e-01 7.84039557e-01 6.90834522e-02 9.23547089e-01 2.45622128e-01 -6.02544665e-01 2.15199262e-01 1.07552353e-02 7.73604035e-01 -7.08314478e-01 -2.29052797e-01 -1.33092374e-01 2.96694160e-01 -1.28960443e+00 -8.31227481e-01 -7.08063900e-01 -1.23513138e+00 1.79464728e-01 1.73308790e-01 -1.00728825e-01 2.62858927e-01 1.06373978e+00 1.58052921e-01 4.70613360e-01 4.50034797e-01 -1.28816092e+00 -2.72329360e-01 -6.61751449e-01 -2.98392922e-01 1.01798117e+00 5.89131117e-01 -9.88861978e-01 -2.77602941e-01 5.89645505e-01]
[8.298137664794922, 0.4740368723869324]
6d396de9-f603-4f08-9acc-a0f28932767c
a-sequence-to-sequence-model-for-user
1607.00070
null
http://arxiv.org/abs/1607.00070v1
http://arxiv.org/pdf/1607.00070v1.pdf
A Sequence-to-Sequence Model for User Simulation in Spoken Dialogue Systems
User simulation is essential for generating enough data to train a statistical spoken dialogue system. Previous models for user simulation suffer from several drawbacks, such as the inability to take dialogue history into account, the need of rigid structure to ensure coherent user behaviour, heavy dependence on a specific domain, the inability to output several user intentions during one dialogue turn, or the requirement of a summarized action space for tractability. This paper introduces a data-driven user simulator based on an encoder-decoder recurrent neural network. The model takes as input a sequence of dialogue contexts and outputs a sequence of dialogue acts corresponding to user intentions. The dialogue contexts include information about the machine acts and the status of the user goal. We show on the Dialogue State Tracking Challenge 2 (DSTC2) dataset that the sequence-to-sequence model outperforms an agenda-based simulator and an n-gram simulator, according to F-score. Furthermore, we show how this model can be used on the original action space and thereby models user behaviour with finer granularity.
['Kaheer Suleman', 'Jing He', 'Layla El Asri']
2016-06-30
null
null
null
null
['user-simulation']
['natural-language-processing']
[ 4.64426637e-01 5.79218447e-01 1.27702102e-01 -5.24475992e-01 -7.92025924e-01 -6.85685754e-01 1.03560007e+00 -9.08148661e-02 -4.34295654e-01 7.94314742e-01 8.47206950e-01 -5.74014425e-01 3.20717186e-01 -4.37044024e-01 6.09864946e-03 -1.56324625e-01 7.51494318e-02 8.71696353e-01 1.76216751e-01 -8.82862747e-01 1.66536510e-01 -1.74465943e-02 -1.35540116e+00 4.46382135e-01 6.60928965e-01 3.37444693e-01 3.88792902e-01 1.30064666e+00 -2.90970225e-02 9.83738601e-01 -9.31021810e-01 1.03901990e-01 -2.08145916e-01 -1.47592449e+00 -1.13123059e+00 3.49751592e-01 -1.61638543e-01 -5.94009519e-01 -3.14357489e-01 6.01125777e-01 7.23947227e-01 4.49894369e-01 6.13475621e-01 -9.47356880e-01 -2.86830105e-02 6.05511785e-01 4.88739669e-01 -1.13190554e-01 1.16059840e+00 3.99915069e-01 9.69268680e-01 -2.81411439e-01 6.78689122e-01 1.36511767e+00 4.21532005e-01 1.12112200e+00 -1.31147099e+00 8.63732845e-02 7.73214027e-02 -2.81005085e-01 -7.59772301e-01 -7.40886271e-01 2.57365972e-01 -2.49308959e-01 1.24040639e+00 5.04543602e-01 6.59808218e-01 1.39963353e+00 -1.52002394e-01 9.27243412e-01 1.00797653e+00 -7.59081125e-01 4.15475428e-01 2.73756772e-01 2.52374411e-01 6.84224367e-01 -7.55389869e-01 2.10455973e-02 -3.23120534e-01 -2.79379994e-01 6.37532294e-01 -5.73841751e-01 -3.83198589e-01 9.46811363e-02 -9.67442632e-01 9.21636879e-01 -1.65525928e-01 3.27258527e-01 -2.62562752e-01 -3.54052037e-01 7.61230469e-01 5.81093013e-01 8.73806253e-02 7.01545238e-01 -3.74636620e-01 -7.81960368e-01 -5.13614655e-01 6.41838253e-01 1.51085174e+00 7.56167233e-01 1.88052073e-01 1.48366526e-01 -5.91847539e-01 8.60935092e-01 2.17168853e-01 2.32613951e-01 7.40656137e-01 -1.12053061e+00 5.00071406e-01 4.58833188e-01 5.38517594e-01 -3.01180959e-01 -6.07576668e-01 3.15424740e-01 -5.31039596e-01 -1.23946890e-02 6.86065972e-01 -7.21247375e-01 -6.99948370e-01 1.93533456e+00 7.52786472e-02 -7.98553452e-02 4.93289411e-01 6.75137937e-01 8.37466121e-01 8.51056755e-01 7.21725896e-02 -4.04646635e-01 1.38539946e+00 -7.50204861e-01 -9.12082255e-01 -3.61107439e-01 1.17202473e+00 -4.44849759e-01 1.24589729e+00 3.23021561e-01 -1.30508137e+00 -3.99467260e-01 -7.15239584e-01 -1.23765143e-02 -1.69730745e-02 5.74741811e-02 2.46990636e-01 8.39271128e-01 -1.16290951e+00 4.40076888e-01 -6.30979896e-01 -4.34545338e-01 -4.55576271e-01 3.18787664e-01 1.35321571e-02 3.72447163e-01 -1.61086810e+00 1.38103807e+00 5.06184042e-01 1.07850702e-02 -6.32505119e-01 -1.20889336e-01 -1.18182957e+00 1.00832924e-01 2.61544824e-01 -4.96255785e-01 2.20822954e+00 -7.62536466e-01 -2.39849114e+00 4.65912461e-01 -2.30869681e-01 -5.41361511e-01 5.32387614e-01 -1.27180919e-01 -1.90024659e-01 -2.29806110e-01 -3.70645374e-01 2.55124122e-01 1.74058601e-01 -8.20904970e-01 -6.52099311e-01 -4.17976640e-02 2.73755610e-01 8.36572409e-01 2.07718819e-01 2.46443823e-01 -4.43815589e-01 -1.98270380e-02 -5.15700161e-01 -1.23374450e+00 -4.50034648e-01 -9.85533178e-01 -5.05756140e-01 -3.01978648e-01 4.32182312e-01 -7.99732804e-01 1.54839015e+00 -1.81657922e+00 4.60387737e-01 -8.63338411e-02 -1.97345987e-01 5.29721975e-01 -1.89142272e-01 8.55372488e-01 1.43904567e-01 -1.13852710e-01 -1.28161922e-01 -5.19151270e-01 3.12922031e-01 1.29295409e-01 -8.41443911e-02 9.03898999e-02 -2.35706531e-02 8.06082368e-01 -1.03298867e+00 -1.93341002e-01 4.48388308e-01 2.86817342e-01 -5.48798621e-01 8.68757069e-01 -6.66727006e-01 6.86594665e-01 -2.81338274e-01 -3.17082763e-01 -2.02374235e-01 -8.10863301e-02 6.23008072e-01 4.17086273e-01 -1.88314989e-01 1.16235459e+00 -9.95495617e-01 1.40438497e+00 -8.39200497e-01 3.81505311e-01 1.62774399e-01 -4.11238283e-01 6.71934009e-01 9.20034766e-01 -2.55215168e-03 -6.77440822e-01 1.93672925e-01 -2.40795743e-02 4.20269400e-01 -4.62328434e-01 7.47374296e-01 -1.99567661e-01 -6.62752450e-01 1.15310538e+00 4.57765423e-02 -2.92847008e-01 2.83207595e-01 3.83489698e-01 8.84402394e-01 7.39409924e-02 5.83769619e-01 1.03942037e-01 6.15500093e-01 -1.40921682e-01 1.79041252e-01 7.79204786e-01 4.00141376e-04 5.31952858e-01 8.39038670e-01 -9.45490599e-02 -1.07157958e+00 -3.85368496e-01 4.41491812e-01 1.48738086e+00 -3.51480395e-01 -5.31621814e-01 -1.10201955e+00 -4.45559800e-01 -5.34966290e-01 1.40219009e+00 -4.89376694e-01 -2.16324240e-01 -8.00167263e-01 -4.39383298e-01 8.05334270e-01 1.86125711e-01 3.29418540e-01 -1.52808690e+00 -7.56335616e-01 4.67968524e-01 -5.06847322e-01 -1.00153852e+00 -6.73574507e-01 -3.53887007e-02 -5.14484465e-01 -6.81683600e-01 -5.18025696e-01 -6.18831575e-01 2.89393604e-01 -2.17703655e-01 1.06012368e+00 -6.59782961e-02 1.61324918e-01 3.07939410e-01 -2.44284838e-01 -2.45957613e-01 -1.46978164e+00 2.81319976e-01 8.43563378e-02 -2.62149483e-01 2.01343507e-01 -7.80825168e-02 -2.54251987e-01 3.87229860e-01 -6.16752982e-01 5.59596300e-01 1.49453074e-01 9.81181383e-01 -5.37534773e-01 -3.72270137e-01 7.24878907e-01 -1.27962470e+00 1.48862183e+00 -2.87275255e-01 -2.26992726e-01 3.15427154e-01 -3.07637453e-01 3.35268706e-01 7.25231946e-01 -2.60314971e-01 -1.47968793e+00 6.12275042e-02 -5.39631546e-01 4.82578337e-01 -4.05863166e-01 3.85305107e-01 -2.82580167e-01 7.28679240e-01 8.50224912e-01 5.33311903e-01 4.11478043e-01 -2.97624022e-01 5.31970978e-01 9.51257348e-01 4.76539195e-01 -4.21709388e-01 1.30468056e-01 -4.05691683e-01 -6.15704656e-01 -9.89244103e-01 -5.90890408e-01 -5.45961201e-01 -6.37378097e-01 -2.30715305e-01 6.82531655e-01 -7.26586640e-01 -9.21773553e-01 5.33725142e-01 -1.26606679e+00 -9.51576471e-01 2.76205856e-02 2.53474265e-01 -1.05578470e+00 5.22316396e-01 -8.65066886e-01 -1.40864229e+00 -3.73510897e-01 -1.09233463e+00 7.61857212e-01 1.59786627e-01 -1.09695506e+00 -1.22007000e+00 1.86743543e-01 1.80976793e-01 3.23238134e-01 -4.79498804e-02 7.87424147e-01 -1.04882741e+00 -2.89937649e-02 -1.96912706e-01 3.14388603e-01 1.72336221e-01 3.11247706e-01 -1.91229954e-01 -8.21197271e-01 -1.30982220e-01 3.18870805e-02 -7.36454606e-01 1.62725091e-01 6.60259798e-02 4.68210906e-01 -6.67484105e-01 -2.45461948e-02 -1.69393927e-01 6.13274872e-01 3.10732961e-01 6.28817379e-01 -7.72473887e-02 3.39679152e-01 9.46887434e-01 4.44297433e-01 4.54632491e-01 5.47011018e-01 8.47478628e-01 -9.96711552e-02 8.30982700e-02 2.18032226e-01 -4.15770739e-01 6.41421676e-01 8.62628460e-01 4.44968604e-03 -5.70726335e-01 -8.49086583e-01 4.73667383e-01 -1.94269753e+00 -1.03283620e+00 3.06246686e-03 2.34401679e+00 1.25938928e+00 3.36108983e-01 7.31734812e-01 -3.54230125e-03 5.27411103e-01 2.69399554e-01 -2.44538203e-01 -9.82938170e-01 3.62256914e-01 -4.15253006e-02 4.09299992e-02 1.17831206e+00 -5.85545897e-01 1.00578475e+00 6.50048018e+00 1.22736759e-01 -7.87179112e-01 -1.85199365e-01 3.75403881e-01 1.14791013e-01 -1.24014832e-01 -1.71373844e-01 -9.47172880e-01 3.85073066e-01 1.77702904e+00 -3.38693529e-01 5.22659540e-01 3.22240949e-01 6.44546449e-01 -2.05682591e-01 -1.22638249e+00 3.49800289e-01 -3.37399840e-02 -9.57150519e-01 -2.30139434e-01 1.30889686e-02 3.20674032e-01 -2.31484413e-01 -2.44552583e-01 7.62025654e-01 9.21311915e-01 -1.17399883e+00 2.37664759e-01 4.03891593e-01 7.89098680e-01 -6.15141213e-01 5.05654216e-01 1.03760803e+00 -4.66206223e-01 1.20572329e-01 8.92440453e-02 -3.46545637e-01 5.28989255e-01 -4.08105314e-01 -1.74807751e+00 2.07864568e-01 -2.92781651e-01 7.08839390e-03 -4.48046736e-02 5.21641970e-01 -1.57285899e-01 8.28058660e-01 -2.82417238e-01 -5.35610139e-01 5.50521195e-01 -1.57266349e-01 3.98206383e-01 1.51241362e+00 -3.50169167e-02 4.35636967e-01 3.44087929e-01 2.78087705e-01 2.31303290e-01 9.27021205e-02 -6.95381880e-01 -2.02058002e-01 3.08201402e-01 9.39207077e-01 -1.82824329e-01 -4.49383408e-01 -3.33343804e-01 1.17034972e+00 6.16811812e-02 2.14082912e-01 -2.99979210e-01 -1.79895982e-01 5.29137373e-01 -2.92723943e-02 -1.23879649e-01 -2.44324848e-01 -5.67289814e-03 -9.31925535e-01 -5.79303980e-01 -1.44472718e+00 1.89453214e-01 -4.79046702e-01 -6.17008328e-01 7.86290467e-01 7.85286427e-02 -7.55118906e-01 -1.44255352e+00 -4.10282701e-01 -7.24050164e-01 1.40276778e+00 -6.59639776e-01 -5.15974402e-01 5.85565418e-02 1.63928285e-01 1.05356157e+00 -2.11432725e-01 1.52646649e+00 -1.58793524e-01 -4.90891159e-01 4.31639522e-01 -2.10624769e-01 2.62775660e-01 5.73399544e-01 -1.52518547e+00 1.00825167e+00 2.70376951e-01 -2.40382105e-01 6.04465306e-01 1.19672179e+00 -6.19014382e-01 -1.02474570e+00 -6.47066236e-01 1.36722708e+00 -7.73340046e-01 3.57654989e-01 -5.76020181e-01 -1.04518044e+00 7.82145143e-01 5.48411667e-01 -9.14464116e-01 7.96641707e-01 3.23140293e-01 1.97653800e-01 7.56041348e-01 -7.74740398e-01 1.04431677e+00 6.82779849e-01 -7.32937694e-01 -8.89363229e-01 2.79288113e-01 5.97667098e-01 -8.85429621e-01 -6.32209241e-01 -1.01260617e-01 5.19160330e-01 -1.04091823e+00 4.20383573e-01 -9.65526760e-01 2.56712496e-01 1.90300971e-01 2.28384733e-01 -1.60423553e+00 -3.92603427e-02 -1.29506779e+00 -9.44172367e-02 9.31620538e-01 6.34389877e-01 -2.83763766e-01 6.15991294e-01 1.16908395e+00 3.29141468e-02 -3.57082188e-01 -5.45295298e-01 -2.71623075e-01 -1.31271288e-01 -1.86724097e-01 4.27681595e-01 5.69731236e-01 6.12189293e-01 1.22368133e+00 -7.60961294e-01 -2.10574135e-01 -5.64851910e-02 -2.41832152e-01 9.81866837e-01 -1.09370136e+00 -4.73350704e-01 -5.01673222e-01 2.90591240e-01 -1.46159196e+00 3.36663090e-02 -3.57337832e-01 5.42708814e-01 -1.58226240e+00 -3.09412211e-01 -1.01509541e-01 2.25807086e-01 1.67986542e-01 -2.79988438e-01 -4.99872953e-01 2.20717624e-01 -1.20827541e-01 -6.88887894e-01 6.00245357e-01 1.05962932e+00 4.29231286e-01 -7.25147069e-01 5.37316322e-01 -5.77842236e-01 5.61053753e-01 8.52125525e-01 -8.86599813e-03 -6.76974535e-01 1.29244655e-01 -2.92772986e-02 9.37322199e-01 -1.66551754e-01 -5.73483646e-01 2.70120621e-01 -1.95473447e-01 -1.46655589e-01 -2.61961371e-01 5.90201676e-01 -1.72240987e-01 -8.05461034e-02 5.83818436e-01 -1.07211554e+00 1.82059072e-02 1.80797964e-01 3.76270831e-01 5.14765037e-03 -5.29604435e-01 5.38225651e-01 -4.67597812e-01 -5.33889472e-01 -2.68651903e-01 -1.06309164e+00 6.38444796e-02 6.55333757e-01 -2.23299235e-01 1.04723625e-01 -1.16929936e+00 -8.13294351e-01 3.84651244e-01 3.05254579e-01 4.84567821e-01 2.76666641e-01 -8.97511542e-01 -9.27255154e-01 1.80756271e-01 4.65965122e-02 -2.19660088e-01 1.23063996e-01 2.52533078e-01 -3.17380130e-01 7.98053741e-01 -1.70302510e-01 -3.30215633e-01 -1.57218766e+00 -4.81193997e-02 5.46500802e-01 -6.05871379e-01 -6.06954217e-01 5.19154549e-01 -1.54985185e-03 -1.00761759e+00 4.22933817e-01 -1.18644677e-01 -5.35962105e-01 6.66219220e-02 6.94619834e-01 1.15455739e-01 -1.24669418e-01 -6.90491557e-01 1.21748313e-01 -3.20407480e-01 -1.74027011e-01 -8.23574066e-01 9.29826558e-01 -3.04133356e-01 3.68272722e-01 8.89120042e-01 8.53556037e-01 -1.99286282e-01 -1.29480875e+00 -4.62992668e-01 2.41978899e-01 -1.87697951e-02 -2.19970778e-01 -1.04969430e+00 1.20603681e-01 6.10967219e-01 6.90429956e-02 7.18805671e-01 5.94043911e-01 -4.00305092e-01 8.35917354e-01 5.58371186e-01 2.19409257e-01 -1.20276701e+00 -2.96956524e-02 1.17575538e+00 9.11084712e-01 -1.05019164e+00 -3.67342651e-01 -8.45554918e-02 -1.40811360e+00 9.81345117e-01 6.69224381e-01 2.89189041e-01 8.96301195e-02 8.85409862e-02 3.63557607e-01 5.13731986e-02 -1.23311126e+00 -2.76922882e-01 1.63240567e-01 4.05607343e-01 8.52431595e-01 1.02904946e-01 -1.78151920e-01 1.68268979e-01 -5.47772527e-01 2.63912566e-02 8.15151691e-01 8.65978956e-01 -5.46276033e-01 -1.33175015e+00 -1.33286536e-01 3.64399046e-01 -3.41306895e-01 4.11232375e-02 -7.72983611e-01 6.20102167e-01 -6.58523917e-01 1.00713229e+00 -3.13225910e-02 -2.39839643e-01 5.83563745e-01 6.54101431e-01 3.28793347e-01 -1.25568402e+00 -1.05357802e+00 -5.54091558e-02 9.18533385e-01 -2.60833055e-01 2.23149676e-02 -7.75131166e-01 -1.26222575e+00 -2.24214762e-01 -2.56026417e-01 4.75421071e-01 5.06425500e-01 1.03530908e+00 1.48805723e-01 4.61351871e-01 7.22219348e-01 -6.59484148e-01 -9.27053392e-01 -1.51716220e+00 -2.57134676e-01 4.60173070e-01 4.95970666e-01 -2.62270384e-02 -5.54325245e-02 -9.60880369e-02]
[12.935517311096191, 7.996414661407471]
efda7080-252b-4a06-b43a-241a59cf01f8
perception-based-energy-functions-in-seam
1701.06141
null
http://arxiv.org/abs/1701.06141v1
http://arxiv.org/pdf/1701.06141v1.pdf
Perception-based energy functions in seam-cutting
Image stitching is challenging in consumer-level photography, due to alignment difficulties in unconstrained shooting environment. Recent studies show that seam-cutting approaches can effectively relieve artifacts generated by local misalignment. Normally, seam-cutting is described in terms of energy minimization, however, few of existing methods consider human perception in their energy functions, which sometimes causes that a seam with minimum energy is not most invisible in the overlapping region. In this paper, we propose a novel perception-based energy function in the seam-cutting framework, which considers the nonlinearity and the nonuniformity of human perception in energy minimization. Our perception-based approach adopts a sigmoid metric to characterize the perception of color discrimination, and a saliency weight to simulate that human eyes incline to pay more attention to salient objects. In addition, our seam-cutting composition can be easily implemented into other stitching pipelines. Experiments show that our method outperforms the seam-cutting method of the normal energy function, and a user study demonstrates that our composed results are more consistent with human perception.
['Tianli Liao', 'Chao Wang', 'Nan Li']
2017-01-22
null
null
null
null
['image-stitching']
['computer-vision']
[ 4.17873859e-01 -1.17455177e-01 1.65726170e-01 -3.57670844e-01 -2.14364395e-01 -4.02856499e-01 2.56610274e-01 -1.93157405e-01 -2.72911876e-01 2.91516721e-01 2.56831437e-01 -1.18688755e-01 2.96322078e-01 -4.76228625e-01 -7.39192963e-01 -3.85999590e-01 5.02928019e-01 -3.57448786e-01 5.43348789e-01 -4.33717668e-01 6.70090199e-01 4.49773930e-02 -1.59353423e+00 2.86375195e-01 1.42907977e+00 8.89831960e-01 5.34488678e-01 3.74895453e-01 2.23809332e-02 2.40916088e-01 -9.29217637e-01 -5.07306159e-01 6.54823363e-01 -5.68833709e-01 -3.42456251e-01 2.11134255e-01 9.85653818e-01 -5.19217908e-01 2.68308390e-02 1.57295001e+00 4.52177733e-01 2.47006536e-01 2.53060043e-01 -1.37571383e+00 -8.46380234e-01 1.20352045e-01 -1.06192327e+00 4.48787585e-02 4.85912621e-01 3.25040281e-01 6.71032608e-01 -8.75019073e-01 4.61598903e-01 1.32726192e+00 7.12723374e-01 3.73241216e-01 -1.06994069e+00 -7.12938488e-01 2.24244878e-01 1.00896440e-01 -1.09360921e+00 -1.72684968e-01 9.32875872e-01 -2.05969542e-01 5.81325412e-01 4.73192275e-01 8.49377871e-01 9.41979229e-01 4.46336687e-01 7.20438361e-01 1.23488903e+00 -3.94747555e-01 1.73009172e-01 2.20154330e-01 -2.29347676e-01 5.66595316e-01 2.95631379e-01 3.45383346e-01 -6.42879307e-01 7.59819373e-02 1.11191249e+00 -1.96962338e-02 -5.01949847e-01 -3.66316885e-01 -1.27275395e+00 4.45160747e-01 7.80245721e-01 -3.57515812e-02 -8.89440924e-02 2.59684920e-01 2.22618937e-01 -1.77892856e-02 5.93786657e-01 6.15552902e-01 -1.50982141e-01 -5.08095883e-02 -1.17219424e+00 2.32268766e-01 2.36965194e-01 9.45095956e-01 6.80747688e-01 5.91268763e-02 -2.34018981e-01 8.60745549e-01 8.68620053e-02 2.41230026e-01 4.55219507e-01 -1.17929387e+00 9.89691094e-02 5.99741340e-01 1.68584123e-01 -1.23759568e+00 -2.32009053e-01 -3.28338265e-01 -5.74605286e-01 9.36320722e-01 1.66647777e-01 -3.19707729e-02 -1.03608835e+00 1.59923196e+00 3.15817863e-01 2.35905930e-01 -2.73077965e-01 1.65676522e+00 6.35537922e-01 4.98512208e-01 -2.22500473e-01 -9.55920368e-02 1.40454388e+00 -1.38333583e+00 -1.04903400e+00 -5.10758936e-01 -1.04438901e-01 -1.06609523e+00 1.51991022e+00 5.07663071e-01 -1.28135443e+00 -7.55703092e-01 -1.27133226e+00 -3.56796503e-01 -2.27149129e-01 8.32508281e-02 7.26394892e-01 6.55824065e-01 -1.18883157e+00 6.40130699e-01 -5.01869619e-01 -4.12177533e-01 1.98492602e-01 -1.37507513e-01 4.26036753e-02 9.12773982e-02 -1.01603425e+00 1.12330508e+00 2.44768158e-01 7.82885700e-02 -4.71825600e-01 -7.24370837e-01 -8.68082106e-01 1.09856956e-01 4.09341395e-01 -8.76042068e-01 1.17630029e+00 -1.26287985e+00 -1.55513787e+00 8.00559044e-01 -7.62696937e-02 -2.32277811e-01 6.41804099e-01 -4.34538186e-01 -1.81938201e-01 5.25240861e-02 8.55220184e-02 8.92268062e-01 1.27685249e+00 -1.54054368e+00 -5.27012050e-01 -1.93031028e-01 6.45192042e-02 6.47952080e-01 -2.97332227e-01 -2.96503957e-02 -5.08826554e-01 -1.03339374e+00 1.38328120e-01 -5.80775142e-01 -2.80519545e-01 4.54627752e-01 -4.05621439e-01 2.09494680e-01 1.00680995e+00 -6.08645856e-01 1.37203288e+00 -2.30892301e+00 -1.51908171e-04 -2.63823494e-02 2.00452134e-01 1.36265785e-01 -7.96592534e-02 3.15108746e-01 1.03309408e-01 -2.35836916e-02 -5.52891552e-01 -4.41664368e-01 -8.59526545e-02 -3.31212441e-03 -3.83033276e-01 3.82672518e-01 6.43811449e-02 7.41784155e-01 -9.56461728e-01 -5.27898610e-01 4.48814780e-01 3.37898284e-01 -5.82787931e-01 1.97474957e-01 -2.34999917e-02 1.21257722e-01 1.45578772e-01 8.16382527e-01 1.19014359e+00 1.38715982e-01 -3.13226789e-01 -6.01193368e-01 -4.95471656e-01 -1.14108689e-01 -1.05909169e+00 2.24410129e+00 -2.93591022e-01 7.93781161e-01 2.94662803e-01 -2.74219155e-01 6.87265873e-01 -1.99245721e-01 1.25654280e-01 -8.18269968e-01 8.01961720e-02 1.34163290e-01 -1.19951576e-01 -4.65099007e-01 9.88452375e-01 2.49763113e-02 1.83758080e-01 3.20704728e-01 -2.72643089e-01 -6.43455386e-01 -5.23956940e-02 1.50157735e-01 6.31518126e-01 3.95419091e-01 8.08468182e-03 -3.42979044e-01 1.13024436e-01 -6.66011870e-02 6.85292006e-01 4.96952444e-01 -3.17005008e-01 1.26440489e+00 1.03260487e-01 -3.32867265e-01 -9.83226776e-01 -1.07067561e+00 -6.51386231e-02 9.99490321e-01 1.07292354e+00 -3.25123757e-01 -1.12149906e+00 -4.27905947e-01 -1.67637751e-01 8.28350186e-01 -5.43489218e-01 -4.20852810e-01 -3.23482931e-01 -4.02587503e-01 2.29497418e-01 3.39715004e-01 8.25756073e-01 -9.82258737e-01 -1.30577064e+00 -3.09931096e-02 -1.63674191e-01 -9.24952447e-01 -1.27110600e+00 -2.61908919e-01 -8.36129546e-01 -9.39439237e-01 -9.47834730e-01 -7.72364676e-01 8.09653401e-01 7.93886662e-01 8.80261481e-01 3.62752050e-01 -5.96170485e-01 1.72065228e-01 -2.79933155e-01 -4.78608251e-01 -1.61144137e-02 -3.72685462e-01 -1.48807764e-01 -6.65845498e-02 -2.71118414e-02 -3.71119827e-01 -1.32651579e+00 5.66683948e-01 -1.13325548e+00 4.43501025e-01 6.00506723e-01 6.99830294e-01 1.85445502e-01 7.07683489e-02 1.21330082e-01 -4.56358910e-01 9.85939264e-01 9.86161232e-02 -5.55292666e-01 2.09279403e-01 -7.99114943e-01 -2.44284451e-01 2.80960113e-01 -7.41673589e-01 -1.33485210e+00 -9.16140825e-02 2.24959284e-01 -6.90859795e-01 1.63422540e-01 8.40913057e-02 7.65685970e-03 -4.00679827e-01 6.09326839e-01 1.76491618e-01 9.79875252e-02 -2.20027938e-01 4.50107127e-01 3.62665236e-01 9.05017138e-01 -2.67539263e-01 7.86655188e-01 6.19983792e-01 -2.86693484e-01 -5.30899823e-01 -6.04060471e-01 -3.87098461e-01 -3.27617884e-01 -5.73671639e-01 7.53825128e-01 -5.77054858e-01 -7.52720773e-01 5.43336213e-01 -1.24721038e+00 -1.88673213e-01 -3.39477181e-01 3.28967601e-01 -4.11024153e-01 7.10262418e-01 -5.01759708e-01 -8.65276814e-01 -2.88746089e-01 -1.21131563e+00 1.30471659e+00 6.63885832e-01 -7.81535208e-02 -7.21909523e-01 3.82995084e-02 2.27731273e-01 6.59400165e-01 3.03708315e-01 5.68466842e-01 2.76374310e-01 -3.91980588e-01 1.22386970e-01 -4.55473572e-01 3.90528142e-01 2.03502998e-01 1.18758619e-01 -1.06918073e+00 -2.93422908e-01 2.37354696e-01 -6.74795210e-02 8.45999539e-01 7.72984624e-01 1.16526508e+00 -2.65507966e-01 -1.44390374e-01 9.15473700e-01 1.41427171e+00 1.69230640e-01 9.78200018e-01 3.65154415e-01 5.89751899e-01 7.16112852e-01 8.55538249e-01 2.24563926e-01 2.58373320e-01 7.30652988e-01 7.00387955e-01 -9.25972939e-01 -3.23895603e-01 -4.15232271e-01 3.98124307e-01 2.76831150e-01 -1.00495927e-01 -7.22201169e-02 -3.84974599e-01 2.57425874e-01 -1.85661721e+00 -9.53712523e-01 -1.29880771e-01 2.37599802e+00 8.84475946e-01 1.62747473e-01 9.27398875e-02 -1.26864433e-01 9.30341959e-01 1.58846885e-01 -5.50394058e-01 -6.05756879e-01 -2.20063031e-01 -5.17899767e-02 6.37299895e-01 5.84854901e-01 -8.94573808e-01 1.08122373e+00 7.26327372e+00 8.53822827e-01 -1.08781302e+00 4.11540596e-03 4.73655641e-01 -1.08105369e-01 -4.08239663e-01 2.32486024e-01 -2.42219821e-01 7.28112400e-01 6.19138926e-02 -1.39172673e-01 4.12081897e-01 5.70140481e-01 3.30179840e-01 -4.88615870e-01 -8.35572720e-01 1.21154439e+00 3.36886048e-01 -1.06172085e+00 -1.10118791e-01 7.11964909e-03 7.22815454e-01 -5.59154093e-01 5.50093591e-01 -7.62840062e-02 -2.21979152e-02 -8.73098671e-01 1.06657743e+00 4.18138534e-01 7.57666588e-01 -6.69769943e-01 4.65710998e-01 1.93429202e-01 -1.03831744e+00 3.68094482e-02 -4.63985741e-01 -3.71798314e-02 4.49506491e-01 6.00058556e-01 -5.54232657e-01 3.09696674e-01 8.30675006e-01 4.25977826e-01 -5.36132038e-01 1.35527277e+00 -1.71843663e-01 8.91385153e-02 -7.99786747e-02 2.63523787e-01 1.03683703e-01 -5.93588471e-01 6.76977575e-01 1.26509488e+00 3.82200599e-01 1.86639175e-01 9.77194905e-02 1.30412567e+00 2.07906708e-01 -4.42255996e-02 -3.35553080e-01 3.94228697e-01 3.81197929e-01 1.33459902e+00 -8.50815475e-01 -2.63393104e-01 -3.26576948e-01 1.68238306e+00 -1.00846589e-01 5.17331481e-01 -1.04164958e+00 -5.08065283e-01 7.02593744e-01 1.83087990e-01 -7.92063773e-02 -1.57526970e-01 -5.68114877e-01 -1.07450151e+00 9.42536816e-02 -7.78458714e-01 5.70324548e-02 -1.38227403e+00 -1.04876530e+00 5.37919641e-01 -4.05431576e-02 -1.50162220e+00 2.83795536e-01 -3.98339510e-01 -8.99174094e-01 7.04994738e-01 -1.48906434e+00 -1.03789282e+00 -7.12018847e-01 3.71427953e-01 1.00374269e+00 4.56717074e-01 2.29593262e-01 2.30128020e-01 -4.73879099e-01 6.03447855e-01 -5.26932597e-01 -4.87185270e-01 1.13722384e+00 -1.24547362e+00 4.76694643e-01 1.15492308e+00 -5.55348247e-02 6.74010038e-01 1.11755514e+00 -6.67512417e-01 -1.08018184e+00 -5.87625623e-01 2.25059330e-01 -2.02555865e-01 2.34624133e-01 -3.20745826e-01 -1.10392213e+00 2.17740804e-01 7.58685768e-01 -1.39919788e-01 2.85164267e-01 -3.87629062e-01 -2.38783672e-01 4.29325178e-02 -1.35141647e+00 8.15829575e-01 1.27705610e+00 -1.95007175e-01 -6.60761416e-01 -1.97779424e-02 7.79877722e-01 -7.35913634e-01 -3.34547907e-01 3.36260766e-01 6.81304693e-01 -1.43515563e+00 1.06794322e+00 -5.09401523e-02 4.55811977e-01 -6.18635356e-01 2.23289937e-01 -1.56917405e+00 -4.02832001e-01 -7.97496676e-01 1.22192256e-01 9.89919245e-01 7.01528490e-02 -5.42282939e-01 4.64279532e-01 5.77712595e-01 -4.29753840e-01 -5.38817942e-01 -7.88545966e-01 -6.42346740e-01 -3.79804969e-01 4.01539169e-02 4.45103288e-01 1.01584029e+00 1.42975450e-01 -7.05709606e-02 -5.92652738e-01 1.24703750e-01 7.48365283e-01 2.10195348e-01 6.63653672e-01 -9.33157027e-01 -2.46056646e-01 -7.37362742e-01 -3.18669021e-01 -1.21526408e+00 -3.31052423e-01 -3.64373922e-01 3.57698381e-01 -1.52073205e+00 2.49098942e-01 -1.03291593e-01 -6.11744188e-02 1.80765793e-01 -5.96076667e-01 4.21403080e-01 2.50202268e-01 1.35590300e-01 -2.30781794e-01 7.28049695e-01 1.62813568e+00 -4.42682728e-02 -3.01622123e-01 -3.13524961e-01 -6.74815297e-01 8.84008825e-01 6.03870690e-01 -1.02806523e-01 -5.33980131e-01 -6.17236376e-01 1.54530481e-01 -2.56873280e-01 4.50166881e-01 -9.75878239e-01 4.98935074e-01 -3.61587584e-01 4.28905278e-01 -6.06926203e-01 4.21820998e-01 -8.82202864e-01 -5.08222682e-03 4.87805873e-01 -1.64880469e-01 2.18456000e-01 2.85528183e-01 4.46887523e-01 -2.56796360e-01 -1.04618900e-01 8.55062783e-01 -4.09121178e-02 -7.85311580e-01 9.54554230e-03 -2.25066707e-01 -1.99382961e-01 9.92487788e-01 -8.08276415e-01 -3.44844908e-01 -4.26417828e-01 -4.82594483e-02 2.34194756e-01 9.85976815e-01 4.96591717e-01 1.07151425e+00 -1.32427633e+00 -4.64756489e-01 3.27446848e-01 -3.10079288e-02 -9.74287912e-02 3.44948202e-01 8.14447761e-01 -6.67803824e-01 -2.80024827e-01 -5.33763885e-01 -5.67753971e-01 -1.26243889e+00 6.52188897e-01 3.70515227e-01 4.32973951e-01 -6.77975357e-01 9.84064937e-01 5.78298628e-01 9.95580107e-02 2.43668377e-01 -5.89190423e-01 1.66312128e-01 -2.04974353e-01 4.64147359e-01 5.51788330e-01 -8.53851810e-02 -3.34678322e-01 -3.13722372e-01 8.61382186e-01 2.58218087e-02 -3.72405678e-01 7.53844798e-01 -4.22410101e-01 6.68966323e-02 1.44672319e-01 8.37547958e-01 2.14766279e-01 -1.65357649e+00 2.10112616e-01 -3.81371796e-01 -1.10385013e+00 1.04438737e-01 -8.21320772e-01 -1.12779260e+00 7.39274144e-01 7.30335772e-01 1.04838662e-01 1.51097512e+00 -2.91176885e-01 9.61944997e-01 -3.37350905e-01 3.03799301e-01 -1.37939334e+00 3.93506765e-01 -7.18507767e-02 1.08148992e+00 -1.28539598e+00 1.00514978e-01 -7.63601661e-01 -9.39073384e-01 9.80715275e-01 1.04617393e+00 -2.01619074e-01 1.84094116e-01 3.52388293e-01 3.25371653e-01 -2.09136099e-01 -5.08383393e-01 -1.59710810e-01 4.61595029e-01 5.55132687e-01 2.26086274e-01 -2.89529979e-01 -6.23258948e-01 2.17840239e-01 -3.56129497e-01 -3.32172662e-01 6.15329742e-01 8.47962737e-01 -6.29299283e-01 -7.64309108e-01 -5.72266221e-01 -2.15348173e-02 -1.52402267e-01 -2.08190143e-01 -4.97454971e-01 6.02628767e-01 3.25941801e-01 9.85451579e-01 1.68804273e-01 -3.71369064e-01 5.25946319e-01 -4.72781092e-01 4.96491551e-01 -4.77554262e-01 -6.91513836e-01 2.79856533e-01 -4.18897957e-01 -9.47551966e-01 -3.44369590e-01 -1.85432971e-01 -1.18519068e+00 -4.74790037e-02 -4.79212582e-01 -2.70933807e-01 7.92352676e-01 3.73032570e-01 2.29575813e-01 5.38919985e-01 5.40249109e-01 -1.15158975e+00 -2.50245184e-01 -8.47992361e-01 -4.32678878e-01 6.70474052e-01 4.21011209e-01 -6.98054850e-01 -5.63584864e-01 1.11897357e-01]
[11.165724754333496, -1.2078474760055542]
48d866ce-37d2-4a98-ac78-c4c319a203e3
attribute2font-creating-fonts-you-want-from
2005.07865
null
https://arxiv.org/abs/2005.07865v1
https://arxiv.org/pdf/2005.07865v1.pdf
Attribute2Font: Creating Fonts You Want From Attributes
Font design is now still considered as an exclusive privilege of professional designers, whose creativity is not possessed by existing software systems. Nevertheless, we also notice that most commercial font products are in fact manually designed by following specific requirements on some attributes of glyphs, such as italic, serif, cursive, width, angularity, etc. Inspired by this fact, we propose a novel model, Attribute2Font, to automatically create fonts by synthesizing visually-pleasing glyph images according to user-specified attributes and their corresponding values. To the best of our knowledge, our model is the first one in the literature which is capable of generating glyph images in new font styles, instead of retrieving existing fonts, according to given values of specified font attributes. Specifically, Attribute2Font is trained to perform font style transfer between any two fonts conditioned on their attribute values. After training, our model can generate glyph images in accordance with an arbitrary set of font attribute values. Furthermore, a novel unit named Attribute Attention Module is designed to make those generated glyph images better embody the prominent font attributes. Considering that the annotations of font attribute values are extremely expensive to obtain, a semi-supervised learning scheme is also introduced to exploit a large number of unlabeled fonts. Experimental results demonstrate that our model achieves impressive performance on many tasks, such as creating glyph images in new font styles, editing existing fonts, interpolation among different fonts, etc.
['Yue Gao', 'Zhouhui Lian', 'Yizhi Wang']
2020-05-16
null
null
null
null
['font-style-transfer']
['computer-vision']
[ 4.65327203e-01 -6.33349195e-02 1.27680972e-01 -5.70948303e-01 -2.89422646e-02 -9.36152637e-01 5.12495935e-01 -6.28269315e-02 -3.25051621e-02 6.88108027e-01 -1.66564226e-01 -3.86722118e-01 1.77526623e-01 -8.77995014e-01 -6.74208403e-01 -5.30343831e-01 6.61545098e-01 3.81750017e-01 -4.09007892e-02 -3.50399375e-01 4.50105339e-01 6.29674435e-01 -1.76242089e+00 4.45339113e-01 1.28229213e+00 9.65375483e-01 4.78094280e-01 4.93990391e-01 -4.76119697e-01 2.79060960e-01 -8.87918293e-01 -6.44312799e-01 4.10020769e-01 -3.94766539e-01 -3.75898033e-01 5.68354011e-01 6.68230772e-01 -3.26769054e-01 1.85231000e-01 1.23351431e+00 2.87758440e-01 -8.81540179e-02 1.03289700e+00 -1.00771117e+00 -1.61570621e+00 6.02464080e-01 -3.81714821e-01 -5.16203284e-01 3.75709116e-01 4.31672156e-01 1.03143394e+00 -9.33722317e-01 5.85074365e-01 9.52668667e-01 1.07914098e-01 5.84375024e-01 -1.14211607e+00 -6.65660679e-01 5.02749562e-01 -2.36257479e-01 -1.35875320e+00 5.81309237e-02 1.07470536e+00 -4.39016253e-01 3.65879506e-01 5.19591272e-01 8.31681430e-01 1.25970089e+00 -5.12994491e-02 7.38618791e-01 1.28414822e+00 -5.50187171e-01 2.59303033e-01 7.23895133e-01 -3.99729073e-01 4.06436265e-01 -6.55405596e-02 -2.72260517e-01 -1.79613858e-01 1.82301566e-01 1.07567561e+00 -1.81430086e-01 -4.17711675e-01 -6.35497153e-01 -1.35498607e+00 5.43439627e-01 1.84252486e-01 1.85778573e-01 -1.70479491e-01 -3.53341460e-01 2.78056443e-01 3.47588807e-01 1.83644801e-01 6.46105647e-01 -4.56507862e-01 3.38186473e-02 -5.59338510e-01 1.20899312e-01 5.25241375e-01 1.63711250e+00 6.02002859e-01 1.31533161e-01 -1.23392820e-01 9.95526910e-01 2.59918660e-01 7.32281506e-01 4.90692586e-01 -4.43868011e-01 6.38054907e-01 7.94980288e-01 3.19763452e-01 -7.50354052e-01 -8.38426203e-02 -1.95860997e-01 -9.11454678e-01 5.20330906e-01 3.65890443e-01 -8.91894009e-03 -8.95056963e-01 1.51889527e+00 1.43697619e-01 -5.16066611e-01 -2.21157782e-02 9.70882177e-01 7.51168191e-01 6.68054402e-01 -1.35993641e-02 -1.44382119e-01 1.45151079e+00 -8.41304004e-01 -8.15732300e-01 3.30433287e-02 2.43074134e-01 -1.02010131e+00 1.97971964e+00 7.15456843e-01 -8.28836441e-01 -9.12089050e-01 -1.23373246e+00 -1.27731249e-01 -8.00011277e-01 5.72538316e-01 7.16831923e-01 1.06327128e+00 -6.71675742e-01 2.85688519e-01 -9.23392177e-03 -1.95021629e-01 3.15200180e-01 4.14230451e-02 1.60265286e-02 3.34176391e-01 -1.09258795e+00 6.00092113e-01 3.91700655e-01 1.41765103e-01 -4.37754512e-01 -5.98110259e-01 -6.45401418e-01 7.86447749e-02 2.24609837e-01 -5.43966055e-01 9.24906909e-01 -1.39826381e+00 -1.67801487e+00 8.52209628e-01 3.42595428e-01 -6.45562634e-02 7.90414751e-01 -1.11866808e-02 -9.05182481e-01 -2.08828554e-01 -1.38558000e-01 9.72927868e-01 1.32307017e+00 -1.60159039e+00 -4.71857816e-01 -8.93494412e-02 3.55087727e-01 3.88765782e-01 -9.19685662e-01 -2.21238956e-01 -3.49400312e-01 -9.17501509e-01 -7.27947280e-02 -7.78502822e-01 6.46371245e-02 3.27529222e-01 -7.65812933e-01 -8.15646127e-02 7.91642845e-01 -4.16073412e-01 1.20460594e+00 -2.08485222e+00 8.81471038e-02 3.34950864e-01 -8.06478411e-03 2.77526766e-01 -2.72611585e-02 2.57917017e-01 -1.73266143e-01 6.12575859e-02 -4.03373361e-01 -1.82863455e-02 2.36998290e-01 -2.35368703e-02 -4.38481748e-01 -2.03636006e-01 3.54310036e-01 6.69597805e-01 -7.68202722e-01 -5.06929934e-01 3.80752563e-01 4.87669528e-01 -2.74164468e-01 1.45855680e-01 -5.88989794e-01 5.45824289e-01 -4.22257483e-01 7.52674878e-01 8.78141463e-01 -1.14100657e-01 2.60909587e-01 -4.11768258e-01 -1.53847009e-01 -3.90426427e-01 -1.13709509e+00 1.60632443e+00 -6.87210381e-01 3.83615196e-01 -3.59938204e-01 -3.86082828e-01 1.55935967e+00 1.79396078e-01 1.32134676e-01 -4.18978363e-01 3.24912727e-01 2.75095940e-01 -3.49574909e-02 -4.02282715e-01 8.53725970e-01 1.99162871e-01 -1.61178738e-01 4.88140047e-01 -4.84045923e-01 -5.74255824e-01 5.74222803e-01 -5.06094005e-03 2.13482872e-01 3.21605414e-01 1.40372723e-01 -2.24459335e-01 8.79544258e-01 -3.87036294e-01 2.27941703e-02 4.42315161e-01 1.98898137e-01 9.70659971e-01 4.55183864e-01 -6.38393044e-01 -1.54563069e+00 -1.12228012e+00 -3.14337760e-01 9.06844079e-01 -1.31423865e-02 -1.78251415e-01 -8.33366454e-01 -8.39346647e-01 -1.61516950e-01 9.06347632e-01 -5.70291340e-01 1.20606169e-01 -5.62302411e-01 -4.87295985e-01 1.16549045e-01 6.12288117e-01 5.31636834e-01 -1.47392416e+00 -7.17102110e-01 -2.75567304e-02 1.75009280e-01 -1.04794693e+00 -1.01380277e+00 2.04614829e-02 -7.30387449e-01 -9.42776680e-01 -1.24604809e+00 -1.05851865e+00 1.33980608e+00 2.78546270e-02 1.08747816e+00 1.51330292e-01 -9.48968902e-02 1.53769895e-01 -6.20951474e-01 -4.32277322e-01 -6.18621409e-01 1.06357507e-01 -1.77507013e-01 4.32362765e-01 1.06148385e-01 -6.52477920e-01 -5.40015280e-01 3.18860173e-01 -1.30843651e+00 5.40082037e-01 7.21639693e-01 7.58297324e-01 6.99672878e-01 2.78722215e-02 3.98972541e-01 -1.20509160e+00 1.02632058e+00 1.00654408e-01 -6.69670641e-01 4.31069016e-01 -5.89949667e-01 4.73066568e-02 1.31249201e+00 -7.61926353e-01 -1.19190133e+00 2.88551897e-01 1.31580085e-01 -2.78995067e-01 -4.54379439e-01 -1.84963606e-02 -6.72590852e-01 1.82103217e-01 6.13899410e-01 4.88428026e-01 -1.53871059e-01 -5.27929068e-01 8.65576565e-01 8.31627786e-01 5.29866099e-01 -7.61592388e-01 9.33940947e-01 2.97637165e-01 -1.91879958e-01 -5.10345280e-01 -4.99713302e-01 3.64502698e-01 -9.39669371e-01 -3.05961013e-01 7.83708632e-01 -4.73743379e-01 -6.17395163e-01 5.11190176e-01 -1.07364750e+00 -2.76031177e-02 -4.64575589e-01 1.10002910e-03 -7.22907543e-01 4.74420369e-01 -1.79439664e-01 -7.22042978e-01 -3.78845274e-01 -1.25775993e+00 1.07567775e+00 2.88249761e-01 -1.30376488e-01 -7.13483572e-01 -6.16778255e-01 1.79632083e-02 3.03877622e-01 2.68584102e-01 1.41904807e+00 -3.43124449e-01 -6.47040904e-01 -2.07825541e-01 -3.57430816e-01 4.70476627e-01 5.81318617e-01 1.66213885e-01 -7.31882095e-01 -1.98828533e-01 -4.23023999e-01 -1.31719798e-01 4.65621352e-01 -5.03078960e-02 1.57895482e+00 -2.55813181e-01 1.63071062e-02 6.50563836e-01 1.18382967e+00 6.27013326e-01 6.63049638e-01 4.03204709e-01 8.52346539e-01 5.74742496e-01 6.08676016e-01 6.02485061e-01 1.39388800e-01 6.71499491e-01 4.69712108e-01 -1.26966342e-01 -1.22175626e-01 -3.44700992e-01 6.81593418e-02 5.39326251e-01 -7.66073391e-02 -3.06414127e-01 -2.33320624e-01 2.21689507e-01 -1.37678468e+00 -5.23224294e-01 -7.58064017e-02 2.46945000e+00 9.72839355e-01 4.63569850e-01 9.04420614e-02 1.81808650e-01 9.28956091e-01 1.11200400e-01 -8.72181594e-01 -6.98115408e-01 -4.87767935e-01 -1.56404842e-02 2.33660638e-01 2.37691496e-02 -1.02972233e+00 1.00187099e+00 5.48937893e+00 7.24147320e-01 -1.06203222e+00 -4.85956222e-01 6.95926726e-01 1.17835671e-01 -7.16357529e-01 -3.20991576e-02 -7.07510352e-01 7.51449585e-01 1.98853448e-01 -1.04220159e-01 7.17452407e-01 8.41456413e-01 4.75472994e-02 3.14764440e-01 -1.03966606e+00 9.22628045e-01 3.12522143e-01 -8.76911700e-01 6.03358626e-01 -1.37988463e-01 8.16718817e-01 -1.19866633e+00 6.49089217e-01 3.63853276e-02 1.23164289e-01 -8.65502715e-01 1.03302622e+00 5.85500181e-01 1.23447239e+00 -7.80200124e-01 1.59338206e-01 -4.11015525e-02 -1.12634301e+00 -6.15696907e-02 -5.13075948e-01 2.41183668e-01 1.31289139e-01 4.46437478e-01 -8.14403653e-01 4.35863942e-01 5.27754664e-01 5.91781974e-01 -9.02081072e-01 8.46624494e-01 -5.28699994e-01 1.35907428e-02 4.66644652e-02 -3.95981789e-01 6.00429326e-02 -4.23694164e-01 3.06285501e-01 8.92306089e-01 6.25566900e-01 -2.60744959e-01 -7.17390776e-02 1.31924558e+00 -7.83108100e-02 4.91366208e-01 -5.04199326e-01 -1.95949689e-01 5.57881653e-01 1.37110937e+00 -7.52613485e-01 -5.21916389e-01 -3.12411666e-01 1.18581438e+00 -1.09898858e-01 4.66499239e-01 -1.01122952e+00 -8.07242215e-01 2.44517520e-01 1.29914805e-01 3.91457438e-01 -2.97790486e-02 -6.21807158e-01 -9.56709981e-01 3.73708010e-01 -9.89763796e-01 8.80482197e-02 -1.35188091e+00 -1.32682419e+00 1.01700115e+00 -1.46802008e-01 -1.88070071e+00 4.02633734e-02 -8.67890537e-01 -4.42962557e-01 9.64866459e-01 -1.21334875e+00 -1.33813572e+00 -3.84049296e-01 4.99076128e-01 7.65277624e-01 -4.72400934e-01 7.26942420e-01 8.51992220e-02 -2.25916803e-01 7.73443401e-01 4.21410576e-02 5.08725829e-02 8.38421345e-01 -1.50916886e+00 6.95213854e-01 4.37578321e-01 1.75909147e-01 7.28344321e-01 7.45227277e-01 -6.17620707e-01 -1.08096898e+00 -8.35791707e-01 6.87086940e-01 -2.91120261e-01 4.11647350e-01 -7.16547608e-01 -8.80692661e-01 4.78610545e-01 3.77613604e-01 -3.53806823e-01 4.38148856e-01 -2.55732030e-01 -6.46765709e-01 -2.41306871e-01 -9.18505669e-01 1.04522443e+00 1.07154334e+00 -2.09010869e-01 -3.19399804e-01 6.21797919e-01 6.21399581e-01 -5.43607295e-01 -8.48114014e-01 2.37621337e-01 5.22215605e-01 -9.79165018e-01 9.25744236e-01 -4.38167423e-01 4.45364177e-01 -4.90187466e-01 1.40750602e-01 -1.43202448e+00 -1.20202355e-01 -5.19404292e-01 2.05775574e-01 1.40835989e+00 5.98104835e-01 -5.85179627e-01 7.60676265e-01 7.00043619e-01 -2.18731135e-01 -5.71321309e-01 -3.79006386e-01 -9.12822485e-01 7.21893087e-02 -4.31785211e-02 1.41641200e+00 8.47629547e-01 -2.20840678e-01 1.36042267e-01 -5.59060693e-01 1.23161804e-02 3.81516635e-01 6.07242346e-01 8.03750515e-01 -1.27790940e+00 -3.47026318e-01 -6.90780461e-01 1.54784871e-02 -1.15459406e+00 1.45999277e-02 -7.06664801e-01 -2.48947725e-01 -1.28091693e+00 -1.79081276e-01 -5.12964606e-01 -1.43687343e-02 4.09243375e-01 -1.58474192e-01 2.03455687e-01 4.65187162e-01 1.34305000e-01 -3.06874335e-01 6.08981907e-01 1.98009360e+00 -4.40492958e-01 -3.61174405e-01 1.22983523e-01 -8.88375044e-01 5.44714272e-01 8.32329690e-01 4.46093408e-03 -6.44775569e-01 -3.90826315e-01 3.15478981e-01 -2.64566094e-01 1.31996283e-02 -8.81072998e-01 -1.18706897e-01 -1.77827969e-01 8.21071804e-01 -5.65972090e-01 3.53055477e-01 -1.04633689e+00 3.59802209e-02 -1.24707157e-02 -4.98404503e-01 3.57875437e-01 2.99699400e-02 2.60715604e-01 1.36745781e-01 -3.95758390e-01 6.07653558e-01 -1.71551570e-01 -6.90940440e-01 7.35090896e-02 -3.76135767e-01 -2.09465399e-01 9.73955095e-01 -6.84394956e-01 -1.23345412e-01 -3.32302898e-01 -6.09012663e-01 -1.54298559e-01 5.85104346e-01 7.66949654e-01 6.85333550e-01 -1.60192633e+00 -5.53817391e-01 5.49184859e-01 4.94912058e-01 -1.57231137e-01 2.87622780e-01 1.82027414e-01 -4.20964241e-01 1.80553302e-01 -5.51540673e-01 -1.90555453e-01 -1.11234045e+00 1.11999750e+00 -1.60495371e-01 -3.72175733e-03 -7.10859120e-01 5.57844222e-01 3.94186437e-01 -4.74040478e-01 1.62763521e-01 -4.55596715e-01 -4.02792007e-01 -4.00751680e-02 4.25096124e-01 -9.31726545e-02 -2.56040916e-02 -4.22569931e-01 2.18624339e-01 9.52489138e-01 -2.60219455e-01 5.71899749e-02 7.89310813e-01 -8.84702355e-02 1.76906019e-01 2.28985235e-01 7.45624483e-01 1.46148935e-01 -1.52932286e+00 3.05136871e-02 -2.77281523e-01 -7.39582539e-01 -7.03697622e-01 -1.16825759e+00 -1.16284037e+00 8.67736816e-01 6.05545998e-01 2.22208098e-01 1.44358051e+00 -4.80498761e-01 6.64568603e-01 2.85296142e-01 4.32055295e-01 -1.27063298e+00 1.86123788e-01 8.17392021e-02 1.23641229e+00 -1.01287127e+00 -1.79107666e-01 -6.48527563e-01 -1.02145898e+00 1.36083603e+00 9.78267431e-01 1.22563042e-01 2.18048930e-01 1.07649766e-01 2.66002119e-01 2.08144709e-01 -3.53602707e-01 4.56871875e-02 4.27906901e-01 7.95149088e-01 5.26584566e-01 1.40169472e-01 -4.24218506e-01 3.79776955e-01 -5.09318829e-01 -3.14966232e-01 5.65287948e-01 5.41361213e-01 -2.81749278e-01 -1.41606224e+00 -4.86354381e-01 3.64883095e-01 2.10123971e-01 -3.88057619e-01 -6.15578115e-01 7.50321031e-01 3.33473802e-01 6.53269172e-01 -3.01294890e-03 -2.53756523e-01 5.61843991e-01 2.59159021e-02 4.18307632e-01 -3.59223396e-01 -5.83770752e-01 -2.52057686e-02 -2.40363866e-01 1.64164916e-01 -3.14995907e-02 -4.46735114e-01 -1.01302576e+00 4.29486297e-03 -1.93484083e-01 4.99054193e-02 6.60173774e-01 4.00733143e-01 1.48285478e-01 6.88556790e-01 7.88675427e-01 -5.09233177e-01 -5.60543001e-01 -8.76536429e-01 -7.66516805e-01 7.66101718e-01 -3.15720826e-01 -5.58774412e-01 -7.12516904e-02 2.84556121e-01]
[11.67818546295166, -0.32473182678222656]
187cfff9-ea30-4041-972c-41869a365081
automatically-identifying-words-that-can
2010.13641
null
https://arxiv.org/abs/2010.13641v1
https://arxiv.org/pdf/2010.13641v1.pdf
Automatically Identifying Words That Can Serve as Labels for Few-Shot Text Classification
A recent approach for few-shot text classification is to convert textual inputs to cloze questions that contain some form of task description, process them with a pretrained language model and map the predicted words to labels. Manually defining this mapping between words and labels requires both domain expertise and an understanding of the language model's abilities. To mitigate this issue, we devise an approach that automatically finds such a mapping given small amounts of training data. For a number of tasks, the mapping found by our approach performs almost as well as hand-crafted label-to-word mappings.
['Hinrich Schütze', 'Helmut Schmid', 'Timo Schick']
2020-10-26
null
https://aclanthology.org/2020.coling-main.488
https://aclanthology.org/2020.coling-main.488.pdf
coling-2020-8
['few-shot-text-classification']
['natural-language-processing']
[ 3.95742863e-01 8.91340524e-02 -4.22181517e-01 -8.33871126e-01 -1.11868930e+00 -7.89099336e-01 7.31453955e-01 4.72443342e-01 -5.69820523e-01 5.17999232e-01 2.74668872e-01 -5.08995771e-01 4.55237664e-02 -6.74580276e-01 -1.75206453e-01 -3.26670446e-02 4.71508831e-01 6.48245454e-01 2.50216573e-01 -4.27600056e-01 5.17460465e-01 -1.13804683e-01 -1.32588494e+00 5.10124922e-01 7.30485559e-01 6.75608814e-01 1.83117911e-01 8.67603779e-01 -7.03267038e-01 1.27580154e+00 -4.85457063e-01 -6.67095304e-01 7.95209929e-02 -5.68643272e-01 -1.30250382e+00 6.61329075e-04 2.77439713e-01 1.58064608e-02 3.20385769e-02 9.77644086e-01 5.42476885e-02 4.20454204e-01 9.47609782e-01 -1.03808522e+00 -6.60177767e-01 7.60775268e-01 -2.11884558e-01 3.68918478e-01 4.78404313e-01 1.19167089e-01 1.38334382e+00 -1.06189251e+00 6.90162301e-01 1.03035164e+00 7.53995597e-01 7.10686028e-01 -1.41768336e+00 -5.78360200e-01 9.97811034e-02 -3.98476608e-02 -1.30713212e+00 -4.80073690e-01 5.65055668e-01 -8.61758173e-01 1.16499901e+00 1.02964893e-01 1.66127756e-01 9.63595390e-01 -4.07250673e-02 1.99206427e-01 1.11355412e+00 -9.17481780e-01 3.34834784e-01 7.62768388e-01 6.60082519e-01 6.69133008e-01 -1.06112249e-01 -3.10308397e-01 -3.62051040e-01 -3.20658118e-01 2.05160603e-01 -1.86051279e-01 -8.44897255e-02 -1.16516553e-01 -7.64332235e-01 1.00995600e+00 1.64806336e-01 5.58145583e-01 -8.19064900e-02 1.46885395e-01 5.83005488e-01 6.07022941e-01 9.95918751e-01 1.11822581e+00 -3.86720806e-01 -2.89791375e-02 -1.07444394e+00 4.94846292e-02 9.93909359e-01 8.92072916e-01 9.60163713e-01 -4.05312687e-01 -3.04353058e-01 1.05030155e+00 1.53440803e-01 -2.82650962e-02 6.59334958e-01 -6.22139573e-01 5.41271687e-01 5.40985346e-01 3.21568072e-01 -8.68315816e-01 -3.29075545e-01 -5.42993769e-02 9.44806561e-02 6.11108467e-02 5.47654569e-01 -5.16700670e-02 -8.78634930e-01 1.64310849e+00 -8.27795863e-02 -2.50963509e-01 5.36546856e-02 7.10413873e-01 6.00036502e-01 7.12834358e-01 6.21096849e-01 -2.22666431e-02 1.53724360e+00 -9.06847358e-01 -7.57708311e-01 -8.70950162e-01 1.10039747e+00 -5.85004807e-01 1.65264964e+00 3.22611332e-02 -7.23906815e-01 -3.17886412e-01 -1.12576342e+00 -2.86248475e-01 -5.90347886e-01 -1.33021295e-01 3.93447906e-01 6.65090919e-01 -8.89035642e-01 4.52410728e-01 -3.49141806e-01 -7.04762638e-01 2.79264390e-01 3.68152671e-02 -5.49647212e-02 -1.43481553e-01 -1.32240129e+00 1.27633405e+00 4.73644763e-01 -6.58085644e-01 -7.67735600e-01 -5.63665450e-01 -8.76434326e-01 1.14341415e-01 5.00861049e-01 -2.95551032e-01 1.62650406e+00 -1.22725034e+00 -1.01504147e+00 1.32135248e+00 -3.05421531e-01 -1.13730453e-01 1.73656970e-01 9.62552428e-02 -2.96163529e-01 -1.63354382e-01 2.05226257e-01 4.79375631e-01 7.25900054e-01 -1.08107400e+00 -6.97291076e-01 -9.88697112e-02 2.43992433e-01 1.14495926e-01 -7.57898986e-01 6.27993643e-01 -9.05039161e-02 -5.56569695e-01 -1.66725352e-01 -5.15185535e-01 -1.88999310e-01 2.03802511e-02 -3.71892862e-02 -6.32909894e-01 3.07708591e-01 -7.12604582e-01 1.41254127e+00 -1.86584485e+00 -2.29766518e-01 9.99567062e-02 3.32193196e-01 1.05092235e-01 -3.06577593e-01 5.35310030e-01 -1.07758030e-01 3.73409808e-01 -8.66349712e-02 -3.64920348e-01 1.48260504e-01 1.24276383e-02 -5.09195805e-01 2.19708055e-01 1.63724914e-01 9.03073847e-01 -1.35620224e+00 -4.87628341e-01 -1.40604466e-01 -7.94834830e-03 -3.88004392e-01 4.56784993e-01 -5.98218143e-01 -9.18114409e-02 -3.89109880e-01 2.95207173e-01 4.21367027e-02 -4.20202047e-01 1.39225453e-01 2.57598460e-01 8.48456696e-02 8.38781476e-01 -5.68430424e-01 1.44239402e+00 -8.59061837e-01 9.00553226e-01 -2.66764641e-01 -7.76651204e-01 1.13273680e+00 5.51860034e-01 2.60489937e-02 -3.18627298e-01 4.53688204e-01 2.55362004e-01 -2.72000909e-01 -8.81138146e-01 3.38791221e-01 -8.74791503e-01 -2.57123709e-01 1.08886385e+00 3.71727586e-01 -2.59196073e-01 2.67405927e-01 3.35987121e-01 1.26924658e+00 -9.71771404e-02 5.25178492e-01 -3.69707346e-01 3.25668216e-01 5.55184126e-01 -1.48111675e-02 9.45832074e-01 -1.87392477e-02 3.64302069e-01 4.24161375e-01 -5.52857220e-01 -1.34552121e+00 -3.20704371e-01 6.01817295e-02 1.85714316e+00 -3.30098778e-01 -5.94552279e-01 -9.85688627e-01 -8.24787199e-01 -1.56345636e-01 1.37309170e+00 -7.41353571e-01 -3.91670108e-01 -1.76830158e-01 -2.54652292e-01 5.64892054e-01 6.45267248e-01 -2.41789013e-01 -1.07259774e+00 -5.64535260e-01 3.03661168e-01 8.11348762e-03 -9.77804959e-01 -5.60905695e-01 5.38350880e-01 -5.44430435e-01 -8.85628581e-01 -3.03020447e-01 -1.04079282e+00 7.52164841e-01 2.41464376e-01 1.35146296e+00 3.33525509e-01 -1.39495388e-01 1.38661876e-01 -6.07163370e-01 -3.62599850e-01 -9.14317131e-01 1.67070687e-01 -1.48883551e-01 -2.08243683e-01 1.15466070e+00 -1.56304136e-01 1.25785649e-01 1.34935334e-01 -7.34330952e-01 -1.04048778e-03 1.89012602e-01 7.56860197e-01 2.67831814e-02 -1.38775438e-01 5.74413240e-01 -1.29367161e+00 1.31110108e+00 -5.57555377e-01 -3.79973710e-01 6.27213061e-01 -8.98575723e-01 2.76129782e-01 7.20246434e-01 -6.27866268e-01 -7.98606515e-01 1.52110353e-01 -3.22256833e-02 -2.90071815e-01 -1.98322296e-01 6.47629023e-01 1.94496498e-01 8.53931010e-02 1.27421451e+00 -2.12495312e-01 -1.47272632e-01 -5.37722528e-01 6.17891431e-01 1.13715124e+00 3.79131556e-01 -5.87779224e-01 8.20110857e-01 1.05102010e-01 -7.78019309e-01 -2.90983349e-01 -1.59070373e+00 -6.10439658e-01 -8.34279239e-01 -1.03820495e-01 9.35183287e-01 -7.10846603e-01 -4.96697463e-02 -2.83994973e-01 -1.37156951e+00 -5.08235872e-01 -4.15016562e-01 3.18950564e-01 -5.70156276e-01 -2.74488986e-01 -3.99592042e-01 -7.65271783e-01 -2.89998978e-01 -6.43583715e-01 6.97655022e-01 -1.26009420e-01 -1.00935590e+00 -1.32847786e+00 2.63385057e-01 2.93047637e-01 4.11842018e-01 -2.81077355e-01 1.27682519e+00 -1.33825552e+00 1.38293773e-01 -6.33010209e-01 -2.38211140e-01 1.96729928e-01 2.29544371e-01 -4.15036559e-01 -1.14100361e+00 -2.20947206e-01 3.50001872e-01 -8.47172976e-01 6.78347945e-01 -1.87054589e-01 9.02953327e-01 -3.30426127e-01 -2.91226864e-01 2.34199435e-01 1.38077652e+00 1.61217988e-01 4.21583466e-02 2.78692156e-01 5.85767448e-01 9.06328261e-01 5.10167420e-01 1.54743955e-01 3.61399382e-01 4.86987025e-01 -1.93862632e-01 1.33639663e-01 3.60551551e-02 -5.52283645e-01 1.66161314e-01 5.22587597e-01 5.90864956e-01 -3.38264972e-01 -1.31294465e+00 5.91614902e-01 -1.60712850e+00 -8.95846009e-01 2.80742556e-01 1.84459019e+00 1.17784405e+00 4.42331970e-01 -5.62157035e-02 -3.13716382e-03 7.84698963e-01 1.69707865e-01 -3.46260190e-01 -7.44673610e-01 4.72667158e-01 2.92387664e-01 4.51676637e-01 8.48495483e-01 -9.62048888e-01 1.34268701e+00 7.74444294e+00 7.30826974e-01 -9.47285473e-01 5.50194621e-01 5.73171496e-01 6.48663566e-02 -5.89293778e-01 2.89536536e-01 -6.27473593e-01 2.06320599e-01 1.25404775e+00 -5.36756396e-01 6.05839670e-01 8.87834013e-01 4.69939038e-02 -2.90867570e-03 -1.45131469e+00 7.39547968e-01 4.96882647e-01 -1.01126719e+00 -5.67777529e-02 -3.48637938e-01 6.19715452e-01 -2.86015034e-01 -2.77882665e-01 4.77143496e-01 7.34407723e-01 -1.31632948e+00 7.72554874e-01 3.69929373e-01 1.03407848e+00 -4.44284916e-01 5.53016067e-01 6.70431077e-01 -8.53594840e-01 -1.75832018e-01 -5.90893328e-01 -5.31611204e-01 -9.26380157e-02 1.86300978e-01 -1.17130196e+00 -2.78899759e-01 1.19214401e-01 4.37473446e-01 -7.98013270e-01 5.59642017e-01 -5.93426347e-01 8.66202593e-01 1.48111731e-01 -4.92994547e-01 3.65839243e-01 1.73959911e-01 9.98930559e-02 1.32377100e+00 8.91482830e-02 1.98067158e-01 2.94063628e-01 1.09987223e+00 -1.30164772e-01 4.18410569e-01 -7.21347332e-01 -4.58817810e-01 6.27943337e-01 1.30443668e+00 -8.70067060e-01 -7.67327249e-01 -5.93789876e-01 7.77351320e-01 6.43315911e-01 3.06034684e-01 -3.70955318e-01 -5.55284381e-01 4.33580101e-01 3.79518330e-01 -7.90849403e-02 4.22182381e-02 -4.95965213e-01 -1.05806315e+00 -1.99233174e-01 -8.56085002e-01 4.09417748e-01 -1.06825423e+00 -1.47145784e+00 6.42498672e-01 -8.50154459e-02 -9.32593942e-01 -5.20947874e-01 -7.13771522e-01 -8.56237710e-01 1.11265981e+00 -1.10040510e+00 -7.74077833e-01 -3.07155848e-01 1.58414155e-01 8.38189900e-01 -2.44180679e-01 9.61627483e-01 1.69972032e-02 -1.97829396e-01 5.01184821e-01 -2.77099997e-01 2.39798218e-01 9.76116002e-01 -1.27251208e+00 6.84116423e-01 6.27082646e-01 3.60226214e-01 6.75097406e-01 1.16773343e+00 -5.78018844e-01 -8.05052936e-01 -1.06452847e+00 1.52855957e+00 -1.01389897e+00 1.08397973e+00 -7.62510478e-01 -1.17511082e+00 8.40127826e-01 2.62394875e-01 3.93682346e-02 9.84321415e-01 3.18801314e-01 -8.02790165e-01 1.81339189e-01 -9.34461534e-01 3.44183445e-01 7.92719722e-01 -1.20744848e+00 -1.21959126e+00 7.80667961e-01 6.32417202e-01 -1.08098142e-01 -3.97205114e-01 -4.05657768e-01 1.69453591e-01 -2.91342288e-01 4.16968465e-01 -1.27666128e+00 5.61884344e-01 2.70250328e-02 -4.22943529e-04 -1.61906469e+00 -3.75662982e-01 -3.33493054e-01 4.44088340e-01 1.24870515e+00 8.22086096e-01 -3.01607430e-01 4.99267429e-01 1.12904620e+00 9.57869440e-02 -4.33898538e-01 -4.87836659e-01 -8.38812709e-01 2.72961676e-01 -4.80896652e-01 2.63778806e-01 1.24972939e+00 6.24183834e-01 9.38455403e-01 -2.33695120e-01 -3.74447018e-01 2.60273546e-01 -1.82219893e-01 4.38250393e-01 -1.41449153e+00 3.76795791e-02 -3.69869351e-01 -1.19240969e-01 -7.25915551e-01 5.89071333e-01 -1.38047493e+00 4.71253753e-01 -1.64569581e+00 4.46678728e-01 -3.89320701e-01 -3.57153982e-01 9.11736250e-01 -3.50037009e-01 2.37580389e-01 -7.92165175e-02 1.96666047e-01 -7.69978166e-01 -1.89937323e-01 5.85413158e-01 -1.31674618e-01 -1.03691511e-01 -3.05031836e-01 -9.86976802e-01 6.53081834e-01 7.44143367e-01 -1.00386345e+00 -5.64897120e-01 -4.71065938e-01 6.25965297e-01 -8.71194825e-02 1.17440186e-01 -7.04331160e-01 4.64090168e-01 -4.24606383e-01 -2.64283959e-02 9.82158855e-02 5.02691902e-02 -5.76220810e-01 -2.81521171e-01 2.91481853e-01 -1.26751244e+00 3.13428305e-02 -1.45548761e-01 3.14364374e-01 -9.80085656e-02 -9.63453829e-01 9.70148742e-01 -3.19598913e-01 -7.51659989e-01 -1.34250879e-01 -5.19869983e-01 4.86176640e-01 1.00735068e+00 -2.24496186e-01 -3.16234499e-01 -6.40455961e-01 -6.01321816e-01 2.75584430e-01 5.96534193e-01 6.34764791e-01 2.68063337e-01 -1.14546955e+00 -6.54403627e-01 8.45617875e-02 5.26736438e-01 -5.81333578e-01 -4.71219510e-01 3.42425913e-01 -2.84586549e-01 6.33343995e-01 -5.72369732e-02 1.30350944e-02 -9.73364294e-01 8.27011049e-01 3.44408602e-01 -1.26919657e-01 -5.83511353e-01 9.46845174e-01 -4.20907922e-02 -3.69950265e-01 1.09892853e-01 7.82684889e-03 -4.34019297e-01 4.03977901e-01 8.92769217e-01 -6.80116266e-02 9.29552764e-02 -6.12222135e-01 -2.56097943e-01 3.66769433e-01 -1.24997728e-01 -3.58545780e-01 1.19772673e+00 -1.17625467e-01 -5.56727573e-02 8.72165978e-01 1.33100593e+00 -2.13803530e-01 -1.03935719e+00 -6.11363947e-01 6.60793781e-01 -5.35145044e-01 1.05147809e-01 -7.30143964e-01 -4.04361814e-01 1.02768493e+00 2.08332121e-01 3.42334032e-01 5.91278791e-01 3.91546309e-01 4.89869893e-01 6.12861156e-01 2.78734803e-01 -1.27286637e+00 3.87733728e-01 7.61994421e-01 5.39287210e-01 -1.28248894e+00 -1.72545388e-01 -1.97463900e-01 -7.46472836e-01 1.14132380e+00 6.83458865e-01 -3.13962325e-02 5.20020723e-01 2.57952988e-01 5.37663579e-01 -4.10740256e-01 -1.23373520e+00 -2.47110218e-01 3.84988755e-01 4.13300961e-01 8.05050790e-01 -1.84950963e-01 -3.76582772e-01 7.88340151e-01 -3.10624719e-01 -1.93551872e-02 4.74736571e-01 8.99429202e-01 -9.93913531e-01 -9.24022496e-01 -1.08732432e-01 6.93384111e-01 -2.82901227e-01 -4.17481750e-01 -7.43660808e-01 2.85536826e-01 -2.60374188e-01 1.06907260e+00 1.89845383e-01 -4.73344475e-01 2.51934797e-01 8.67778480e-01 -8.04922730e-02 -1.53309822e+00 -7.85240650e-01 -2.59286374e-01 2.97253370e-01 -2.04787984e-01 1.30768968e-02 -3.24593872e-01 -1.16243315e+00 -4.13093250e-03 -3.86604041e-01 3.80100101e-01 6.60294235e-01 1.41012871e+00 -1.60639852e-01 2.73661047e-01 5.07530332e-01 -3.08942646e-01 -1.05608451e+00 -1.16241682e+00 -4.00880635e-01 7.48885453e-01 1.12658270e-01 -4.09329742e-01 -5.01694262e-01 7.31950998e-02]
[10.702798843383789, 7.943620681762695]
21416845-a1f0-4195-833d-18dbd7ddcd33
synbols-probing-learning-algorithms-with
2009.06415
null
https://arxiv.org/abs/2009.06415v2
https://arxiv.org/pdf/2009.06415v2.pdf
Synbols: Probing Learning Algorithms with Synthetic Datasets
Progress in the field of machine learning has been fueled by the introduction of benchmark datasets pushing the limits of existing algorithms. Enabling the design of datasets to test specific properties and failure modes of learning algorithms is thus a problem of high interest, as it has a direct impact on innovation in the field. In this sense, we introduce Synbols -- Synthetic Symbols -- a tool for rapidly generating new datasets with a rich composition of latent features rendered in low resolution images. Synbols leverages the large amount of symbols available in the Unicode standard and the wide range of artistic font provided by the open font community. Our tool's high-level interface provides a language for rapidly generating new distributions on the latent features, including various types of textures and occlusions. To showcase the versatility of Synbols, we use it to dissect the limitations and flaws in standard learning algorithms in various learning setups including supervised learning, active learning, out of distribution generalization, unsupervised representation learning, and object counting.
['David Vázquez', 'Matt Craddock', 'Frédéric Branchaud-Charron', 'Pau Rodríguez', 'Issam Laradji', 'Parmida Atighehchian', 'Massimo Caccia', 'Laurent Charlin', 'Alexandre Lacoste', 'Alexandre Drouin']
2020-09-14
null
http://proceedings.neurips.cc/paper/2020/hash/0169cf885f882efd795951253db5cdfb-Abstract.html
http://proceedings.neurips.cc/paper/2020/file/0169cf885f882efd795951253db5cdfb-Paper.pdf
neurips-2020-12
['object-counting']
['computer-vision']
[ 3.82118225e-01 -1.24268100e-01 -1.25476450e-01 -4.93300945e-01 -6.23845220e-01 -7.97950804e-01 9.28430498e-01 2.86474645e-01 -3.39263767e-01 8.03837538e-01 3.10164923e-03 -3.21038991e-01 -3.81706357e-01 -7.30860353e-01 -5.87585509e-01 -6.93148673e-01 -3.80132079e-01 4.40989554e-01 2.07434803e-01 -6.56804517e-02 5.03972769e-01 7.65141785e-01 -2.12175679e+00 4.63148922e-01 5.80261469e-01 7.93236852e-01 8.97260234e-02 5.92173696e-01 -5.36317408e-01 8.62840235e-01 -7.34034598e-01 -2.20674694e-01 2.33728781e-01 -8.22659656e-02 -3.82182807e-01 -9.87597629e-02 6.12583995e-01 -1.16982386e-01 8.44550058e-02 5.86955070e-01 4.32874084e-01 -3.33893597e-02 8.58219802e-01 -1.25793815e+00 -4.75699067e-01 3.71952355e-01 -4.21292931e-01 3.41414511e-01 4.72596169e-01 2.38112465e-01 7.76633143e-01 -8.62535357e-01 8.02910388e-01 1.20489824e+00 4.79366660e-01 3.54511201e-01 -1.58026063e+00 -6.86537087e-01 -1.65219232e-01 -6.92680925e-02 -1.07258153e+00 -6.23563826e-01 4.92807657e-01 -8.26925218e-01 8.98959696e-01 5.02916455e-01 4.56164956e-01 1.54351938e+00 -2.47723311e-01 8.06567252e-01 1.36447024e+00 -9.50702548e-01 3.77257943e-01 4.03913438e-01 1.28539070e-01 7.21149325e-01 4.00296658e-01 2.00675040e-01 -7.99914896e-01 -2.02652559e-01 9.55358088e-01 -2.27110565e-01 7.30703324e-02 -8.57394278e-01 -1.23263812e+00 6.52024269e-01 -1.32601887e-01 1.03292264e-01 2.29933992e-01 -8.74666274e-02 4.26893264e-01 4.66670632e-01 5.22482932e-01 7.61853158e-01 -4.46418822e-01 -5.34619272e-01 -7.29292572e-01 3.35117221e-01 8.83749306e-01 7.92184830e-01 7.75686920e-01 -4.57183868e-02 -9.35058072e-02 1.10522449e+00 1.71946585e-01 3.34292352e-01 3.84859592e-01 -8.25334311e-01 3.56385291e-01 8.18761230e-01 2.15454355e-01 -7.38864779e-01 -2.54454434e-01 -5.34526527e-01 -4.25316870e-01 8.22080374e-01 6.49455786e-01 8.68502110e-02 -7.03244567e-01 1.47897279e+00 3.62316906e-01 -2.70555764e-01 -1.41916409e-01 2.01058641e-01 5.21230102e-01 2.30339527e-01 -8.20187032e-02 -7.66234919e-02 6.39170885e-01 -4.42080528e-01 -2.44556129e-01 6.21123938e-04 7.66329527e-01 -8.73683214e-01 1.50443256e+00 6.28988862e-01 -8.33155572e-01 -5.95885038e-01 -1.11503279e+00 6.61750650e-03 -8.83977592e-01 -5.82690574e-02 9.31879997e-01 7.86798716e-01 -7.93199122e-01 6.10941470e-01 -6.98197782e-01 -2.58068681e-01 7.00935960e-01 3.21334064e-01 -4.18578953e-01 -1.57285511e-01 -5.97495914e-01 9.09658849e-01 5.16096652e-01 -5.31007171e-01 -4.08828110e-01 -8.74535620e-01 -7.11353600e-01 -1.35957152e-01 3.44307154e-01 -2.62526691e-01 1.09275866e+00 -8.67532611e-01 -1.23162377e+00 9.90375936e-01 4.03087556e-01 -1.92880630e-01 8.81673396e-01 -2.68393993e-01 -3.12415421e-01 -1.86149254e-01 -8.33972022e-02 6.58793390e-01 9.71401632e-01 -1.15696073e+00 -5.55306733e-01 -1.83958203e-01 -3.10730934e-02 -1.24309892e-02 -6.17908835e-01 -5.19900508e-02 -1.38495892e-01 -8.01041007e-01 -2.62994885e-01 -5.60813606e-01 1.41120881e-01 2.99191058e-01 -1.29078284e-01 -1.33215874e-01 1.16061211e+00 -1.02071026e-02 1.03939986e+00 -2.63739300e+00 -2.42417723e-01 2.20220715e-01 2.39492245e-02 1.32014692e-01 6.11697659e-02 5.67584693e-01 -3.34912658e-01 1.58934921e-01 -5.98184429e-02 -2.31815919e-01 1.52237087e-01 4.14903760e-01 -5.95378637e-01 2.00307772e-01 3.05457354e-01 6.34963393e-01 -8.78375471e-01 -6.24023438e-01 3.41049492e-01 3.34461957e-01 -2.67940789e-01 1.19286411e-01 -3.66313189e-01 4.29194868e-01 -1.36520147e-01 6.87497854e-01 3.37026745e-01 -3.83706212e-01 2.88999807e-02 2.71334499e-01 -4.13243324e-01 2.87177533e-01 -1.27707815e+00 1.71303988e+00 -3.50965023e-01 6.24607801e-01 -4.08369750e-01 -7.51958549e-01 1.13763916e+00 -1.12779260e-01 2.99516976e-01 -7.30851412e-01 -2.20228821e-01 3.01273584e-01 -4.81647067e-02 -4.77202773e-01 4.55582827e-01 2.91146874e-01 1.26618892e-01 7.27686763e-01 5.21580726e-02 -8.20508506e-03 7.04195440e-01 1.57164395e-01 1.14913237e+00 5.05934834e-01 2.82925576e-01 -3.88353348e-01 1.70573816e-01 -3.78287062e-02 -1.70049012e-01 1.01913357e+00 3.86010319e-01 5.95710218e-01 6.69701934e-01 -4.65032637e-01 -1.25162387e+00 -1.37815082e+00 -5.22717476e-01 1.27073777e+00 -5.55339098e-01 -4.21302617e-01 -3.20925921e-01 -6.56472087e-01 1.50552765e-01 6.36919916e-01 -6.96249247e-01 2.30382401e-02 -5.17042816e-01 -8.67657423e-01 4.54709560e-01 5.17276525e-01 2.77913213e-01 -1.18060660e+00 -1.01508987e+00 -1.97859585e-01 4.99525875e-01 -8.58861506e-01 3.31894875e-01 4.07918125e-01 -1.08898342e+00 -1.21070552e+00 -6.37006879e-01 -5.95027804e-01 7.11286426e-01 -1.04422159e-01 1.46588278e+00 -2.81964280e-02 -7.48178542e-01 5.91004789e-01 -5.75647593e-01 -5.53601921e-01 -5.56449234e-01 2.30016366e-01 -1.92708433e-01 -2.47454092e-01 1.63627326e-01 -7.12305725e-01 -1.58449024e-01 1.20693415e-01 -1.20727837e+00 2.27856457e-01 6.17145240e-01 7.29572356e-01 2.43990153e-01 -8.38788226e-03 3.49094748e-01 -1.22568345e+00 5.78335464e-01 -3.83193612e-01 -8.53138387e-01 2.59158581e-01 -5.55913329e-01 3.23970854e-01 5.53292155e-01 -6.08376324e-01 -8.77320945e-01 1.63357593e-02 1.10222943e-01 -1.30323982e-02 -4.59982276e-01 2.46119708e-01 -9.26405266e-02 3.77455354e-03 1.05807960e+00 7.39237517e-02 -1.78001113e-02 -7.19316185e-01 4.51049805e-01 6.82181656e-01 4.67424035e-01 -9.77527320e-01 7.86560595e-01 3.41409624e-01 6.76063672e-02 -7.80403078e-01 -6.15215957e-01 -1.40378118e-01 -7.33311594e-01 -1.37868077e-01 2.53205091e-01 -5.95514894e-01 -3.70677829e-01 4.94385868e-01 -7.20006526e-01 -6.31251037e-01 -8.28762412e-01 2.83456296e-01 -5.77614844e-01 2.23076165e-01 -2.36772858e-02 -7.92207599e-01 7.17621297e-02 -1.04424798e+00 8.44002187e-01 1.90889016e-01 -5.44585824e-01 -9.64685142e-01 2.25243166e-01 -9.19989273e-02 4.81809109e-01 5.46918333e-01 1.54263902e+00 -7.02288091e-01 -5.98509848e-01 -2.65626609e-01 -9.16673988e-02 3.18490803e-01 1.20970324e-01 2.94736415e-01 -1.13305831e+00 -2.77000099e-01 -5.87565780e-01 -6.85637772e-01 6.36072040e-01 -1.45044342e-01 1.47837722e+00 9.46921408e-02 -1.75735071e-01 5.42317450e-01 1.36713660e+00 6.13257699e-02 5.91816485e-01 6.43770933e-01 3.36465865e-01 6.57660782e-01 2.20594481e-01 5.92050493e-01 -3.20658416e-01 6.51924968e-01 2.52750337e-01 -1.20900542e-01 -4.64865565e-02 -1.13989867e-01 8.09610263e-02 3.68864983e-01 1.79767042e-01 3.63089293e-02 -1.12072229e+00 1.35595590e-01 -1.44387376e+00 -7.49014616e-01 -2.10036687e-03 2.73227668e+00 9.30171967e-01 6.02636516e-01 1.74480885e-01 6.20865166e-01 4.13737506e-01 1.29962236e-01 -4.38027740e-01 -4.15586829e-01 -1.16674252e-01 4.86368388e-01 1.91495761e-01 1.62702158e-01 -1.02227318e+00 3.99827749e-01 6.75046968e+00 1.03504324e+00 -1.08280897e+00 -2.39004642e-01 4.12076682e-01 -9.75610837e-02 -1.49696901e-01 -2.30206639e-01 -8.66675019e-01 4.83949184e-01 7.58737981e-01 1.20316103e-01 3.17473143e-01 1.01776481e+00 -1.02454931e-01 -3.21115226e-01 -1.44112158e+00 1.07985926e+00 7.23580271e-02 -1.54048800e+00 7.14796260e-02 1.17028013e-01 6.00381613e-01 -4.56750840e-02 3.43634158e-01 3.24418575e-01 4.80789483e-01 -1.16867769e+00 6.77050829e-01 7.28414297e-01 1.27310181e+00 -5.06481946e-01 1.35940164e-01 3.65001529e-01 -6.05017126e-01 -1.58441514e-01 -2.03517184e-01 -1.72093183e-01 -3.47373545e-01 6.44980073e-01 -9.52970624e-01 1.87783748e-01 5.90121269e-01 6.04478121e-01 -1.15229917e+00 1.23235297e+00 2.15485925e-03 6.07532382e-01 -4.99809951e-01 6.68729097e-02 -2.77504146e-01 -1.64006483e-02 1.75374895e-01 1.54071999e+00 6.48992360e-02 -5.98519385e-01 2.64686216e-02 7.26893425e-01 9.07282457e-02 1.07419342e-01 -7.54152298e-01 -3.12283516e-01 4.93448555e-01 1.29649138e+00 -1.00677431e+00 -2.13294268e-01 -3.90375793e-01 5.51301241e-01 4.58696693e-01 1.44092292e-01 -4.34329063e-01 -3.75583649e-01 2.12370783e-01 3.18747848e-01 1.35278970e-01 -5.56614935e-01 -4.39231515e-01 -8.95575464e-01 -3.82326879e-02 -1.17583156e+00 3.43320996e-01 -8.60966921e-01 -1.29256034e+00 1.63046494e-01 3.73531789e-01 -1.08347464e+00 -3.81728321e-01 -9.48373199e-01 -3.38426709e-01 7.53268182e-01 -1.00117993e+00 -9.44995999e-01 -5.00521302e-01 4.45457309e-01 4.91268843e-01 -5.05130529e-01 1.28952432e+00 1.30394310e-01 -4.45522666e-01 5.87841332e-01 5.06132364e-01 2.06244871e-01 7.15469778e-01 -1.40671539e+00 6.16939187e-01 3.12773228e-01 5.33716738e-01 6.34058952e-01 5.13306022e-01 -3.05020511e-01 -1.11021817e+00 -5.66271663e-01 4.37544346e-01 -7.48156607e-01 6.14750624e-01 -7.86539376e-01 -7.38940537e-01 4.87176061e-01 -2.42126882e-01 1.03688352e-01 9.16574180e-01 2.96008289e-01 -7.52477050e-01 -1.99401647e-01 -8.84907365e-01 5.16806662e-01 9.43202853e-01 -4.74946618e-01 -3.70613426e-01 3.93413097e-01 1.14906736e-01 -3.60730022e-01 -6.61290526e-01 4.76857245e-01 8.53262544e-01 -1.30304098e+00 1.09199643e+00 -5.91178894e-01 6.45127058e-01 1.53516322e-01 -1.83565319e-01 -8.72879446e-01 -1.27584562e-01 -5.51441789e-01 -2.31371850e-01 1.41584539e+00 4.92433965e-01 -5.74508429e-01 8.33580613e-01 3.13968927e-01 5.12100905e-02 -9.17931020e-01 -6.62530482e-01 -7.30608106e-01 3.07999421e-02 -5.13503015e-01 4.87142295e-01 9.86550093e-01 -2.65306443e-01 5.89945503e-02 4.66702543e-02 -4.79928464e-01 5.27067065e-01 -9.96443480e-02 1.20348907e+00 -1.63805604e+00 -5.57351470e-01 -4.05876666e-01 -4.17381436e-01 -8.86086345e-01 -1.61620542e-01 -8.48548949e-01 -5.87912917e-01 -1.10928094e+00 5.49194403e-02 -8.62078607e-01 -1.29528940e-01 4.89442468e-01 1.05434544e-01 3.51099342e-01 1.79934323e-01 5.36677182e-01 -4.29443240e-01 2.30235094e-03 9.46841598e-01 1.85729980e-01 -1.31409511e-01 -1.95235521e-01 -3.78071845e-01 8.10050547e-01 7.13893533e-01 -5.20549417e-01 -5.39968550e-01 -4.66045588e-01 6.01746440e-01 -5.12499750e-01 4.46146548e-01 -1.30824077e+00 -1.80887282e-01 -7.20662177e-02 9.53093469e-01 -4.68331635e-01 3.96937668e-01 -6.74314618e-01 -6.64309710e-02 1.06197685e-01 -7.60553777e-01 2.41972715e-01 2.54076153e-01 3.45743060e-01 1.82796881e-01 -3.35496157e-01 7.34000325e-01 -3.45525086e-01 -8.00131559e-01 -4.39475738e-02 -1.72763512e-01 4.97714996e-01 8.56229484e-01 -5.09578347e-01 -4.90717381e-01 7.61909410e-02 -5.07127523e-01 -4.56311077e-01 7.14191318e-01 5.97699463e-01 1.73794404e-01 -8.73666525e-01 -4.03943002e-01 7.90239334e-01 3.61176223e-01 -7.87241161e-02 -2.23075494e-01 4.80439603e-01 -6.01334572e-01 9.79779512e-02 -4.07148629e-01 -7.13080287e-01 -1.25246453e+00 1.91996828e-01 9.43032056e-02 -2.05536336e-01 -7.61851549e-01 7.80141830e-01 -7.75855035e-02 -3.37596744e-01 3.85440886e-01 3.60686041e-04 -1.39364481e-01 2.08213627e-01 5.25062799e-01 4.57005024e-01 2.81259507e-01 6.85320469e-03 1.85329597e-02 1.19920179e-01 -2.68597662e-01 -1.67258874e-01 1.35287356e+00 2.91551769e-01 1.30340591e-01 9.70966578e-01 1.03501272e+00 2.04267412e-01 -1.38303173e+00 -6.94216937e-02 4.43183690e-01 -7.55835474e-01 -3.03247750e-01 -1.12963629e+00 -4.15881515e-01 8.62212479e-01 9.34634864e-01 3.88782650e-01 6.78204000e-01 -1.21613517e-01 -6.35053366e-02 4.81801003e-01 6.03930593e-01 -1.09287047e+00 2.95312285e-01 3.60156655e-01 8.30780625e-01 -1.10294878e+00 2.78163701e-01 -2.68310130e-01 -1.97799116e-01 1.43544328e+00 3.80893975e-01 7.89462402e-02 2.74395943e-01 8.05975080e-01 3.18748027e-01 -8.41368884e-02 -7.04020739e-01 -1.05802387e-01 -4.74559851e-02 8.51361930e-01 6.85521066e-01 -2.72347271e-01 -4.25410792e-02 -1.50234967e-01 -4.35962230e-01 7.78401643e-02 4.68600720e-01 1.12699616e+00 -3.66029590e-01 -1.50867712e+00 -5.12889981e-01 6.92994058e-01 -1.83574975e-01 -5.35175577e-02 -4.06512678e-01 1.05512881e+00 3.60009551e-01 5.79121411e-01 1.33132130e-01 -6.06429502e-02 5.49662113e-02 3.46027821e-01 7.38574684e-01 -8.76601458e-01 -3.10672760e-01 -2.44661078e-01 9.23771113e-02 -9.65541042e-03 -1.89522266e-01 -5.73866725e-01 -7.54741788e-01 -8.47647805e-03 -4.17986587e-02 -7.93160275e-02 7.70840704e-01 9.13093150e-01 1.75479591e-01 3.67844790e-01 3.33206832e-01 -8.29769135e-01 -7.71218538e-01 -9.49674606e-01 -5.70000291e-01 6.08066082e-01 -3.27228531e-02 -9.31214213e-01 -4.81543869e-01 7.76175931e-02]
[9.655993461608887, 2.635838747024536]
36cc2597-1ca2-4a98-988f-927a5b1a9665
self-supervised-deep-subspace-clustering-with
2206.04958
null
https://arxiv.org/abs/2206.04958v1
https://arxiv.org/pdf/2206.04958v1.pdf
Self-Supervised Deep Subspace Clustering with Entropy-norm
Auto-Encoder based deep subspace clustering (DSC) is widely used in computer vision, motion segmentation and image processing. However, it suffers from the following three issues in the self-expressive matrix learning process: the first one is less useful information for learning self-expressive weights due to the simple reconstruction loss; the second one is that the construction of the self-expression layer associated with the sample size requires high-computational cost; and the last one is the limited connectivity of the existing regularization terms. In order to address these issues, in this paper we propose a novel model named Self-Supervised deep Subspace Clustering with Entropy-norm (S$^{3}$CE). Specifically, S$^{3}$CE exploits a self-supervised contrastive network to gain a more effetive feature vector. The local structure and dense connectivity of the original data benefit from the self-expressive layer and additional entropy-norm constraint. Moreover, a new module with data enhancement is designed to help S$^{3}$CE focus on the key information of data, and improve the clustering performance of positive and negative instances through spectral clustering. Extensive experimental results demonstrate the superior performance of S$^{3}$CE in comparison to the state-of-the-art approaches.
['Xuesong Yin', 'Simin Kou', 'Guangyi Zhao']
2022-06-10
null
null
null
null
['motion-segmentation']
['computer-vision']
[-8.94108638e-02 -1.94984570e-01 -5.75162172e-02 -2.42908344e-01 -3.67169559e-01 1.03870086e-01 9.51119885e-02 -2.24627405e-01 -4.19257015e-01 3.66017133e-01 2.47889206e-01 4.19838220e-01 -4.62062210e-01 -6.38467610e-01 -5.50781906e-01 -1.11315596e+00 -2.62997627e-01 1.24417275e-01 1.42202720e-01 -1.73343405e-01 -2.62061525e-02 1.53671652e-01 -1.48264921e+00 1.56190291e-01 1.26635253e+00 1.28267860e+00 2.66477048e-01 -8.85599703e-02 -2.23497123e-01 5.47761440e-01 -5.95140122e-02 2.97211693e-03 2.58776903e-01 -5.78402579e-01 -3.90039176e-01 3.53794843e-01 2.06751120e-03 -1.14571571e-01 -5.05820155e-01 1.40923977e+00 5.23273587e-01 2.36458406e-01 7.42932737e-01 -1.21493912e+00 -4.98201460e-01 5.19241750e-01 -9.10364270e-01 1.39813796e-01 -1.56376809e-01 1.68683782e-01 9.12125051e-01 -1.04078031e+00 6.59153402e-01 1.02296042e+00 5.93978465e-01 2.92111039e-01 -1.07451248e+00 -9.43934321e-01 1.44249693e-01 4.60670322e-01 -1.87702787e+00 -2.85873294e-01 1.35233843e+00 -5.65746427e-01 4.18120682e-01 1.31770045e-01 7.39465117e-01 9.38714325e-01 -3.07203442e-01 1.00091112e+00 9.82258737e-01 -1.81193110e-02 3.44591975e-01 1.33223876e-01 8.76942948e-02 8.73044550e-01 -5.32235466e-02 8.06812122e-02 -3.66019845e-01 2.36282274e-01 8.39665234e-01 -1.46636218e-02 -3.99048746e-01 -7.87673533e-01 -1.06959116e+00 8.69818926e-01 6.69073105e-01 5.64316928e-01 -1.10311516e-01 -1.72856197e-01 6.03512645e-01 -6.32026866e-02 4.05823082e-01 1.51467055e-01 -1.91695943e-01 -4.93087173e-02 -9.57317352e-01 -1.61491200e-01 2.60518104e-01 8.96297157e-01 9.76251662e-01 4.56901759e-01 4.62196805e-02 1.06647491e+00 3.27147335e-01 2.67167896e-01 7.12883234e-01 -9.40878808e-01 5.00709713e-01 1.02791345e+00 -4.25099403e-01 -1.26938546e+00 -4.43393111e-01 -9.11840439e-01 -1.65450943e+00 9.14411098e-02 -5.86654292e-03 -3.60494964e-02 -4.70971137e-01 1.82484055e+00 4.14215684e-01 2.92650193e-01 -1.04288924e-02 1.05714142e+00 8.96610796e-01 6.26668811e-01 -1.22129723e-01 -6.73616111e-01 8.75481844e-01 -7.49198735e-01 -7.82303274e-01 -7.87760839e-02 5.32645524e-01 -5.39481580e-01 1.04386866e+00 1.25374302e-01 -7.30548143e-01 -6.77437127e-01 -1.14804602e+00 3.72671522e-02 -1.63205236e-01 2.96608418e-01 6.39422178e-01 2.46577039e-01 -6.96274281e-01 6.35796785e-01 -6.91612184e-01 1.74841576e-03 7.02156186e-01 4.93675321e-01 -4.61760044e-01 5.25474995e-02 -1.12127519e+00 3.71622950e-01 5.59868991e-01 1.97143510e-01 -4.01021093e-01 -4.58530486e-01 -8.78153503e-01 9.94327888e-02 5.14571548e-01 -3.49775702e-01 3.28967243e-01 -9.65571105e-01 -1.36112285e+00 6.95343435e-01 -3.48941348e-02 -3.58793475e-02 4.87041116e-01 -6.62084296e-02 -4.05381769e-01 2.87280262e-01 2.96757340e-01 5.14672816e-01 8.17160070e-01 -1.29926586e+00 -3.98770899e-01 -7.53466368e-01 -5.93549669e-01 4.66752112e-01 -8.77035320e-01 -2.51600713e-01 -7.57795990e-01 -8.87817562e-01 4.36279833e-01 -7.04651892e-01 -2.76984930e-01 -5.82359210e-02 -3.55843484e-01 -2.02215910e-01 9.26025033e-01 -5.91925085e-01 1.50776124e+00 -2.51123595e+00 5.22044599e-01 3.62588525e-01 2.98659623e-01 3.51342499e-01 -2.09742296e-03 2.51090620e-03 -2.80179113e-01 -3.35554518e-02 -5.18792570e-01 -6.50049001e-02 -1.57599002e-01 1.56128639e-02 1.74289539e-01 5.29543817e-01 2.01793507e-01 6.43472970e-01 -8.23103786e-01 -9.53779697e-01 4.00239021e-01 6.41353190e-01 -5.96424937e-01 8.90928209e-02 2.12468266e-01 5.37447035e-01 -5.42340815e-01 3.81041378e-01 8.03729177e-01 -2.58416235e-01 -1.50637373e-01 -6.06575429e-01 -1.62635088e-01 -3.35475683e-01 -1.66901231e+00 1.89098620e+00 9.24179256e-02 1.52641445e-01 3.88325900e-01 -1.32863033e+00 7.91045845e-01 4.82919253e-02 1.02851307e+00 -6.12546563e-01 3.45814645e-01 2.56903201e-01 -3.67107317e-02 -5.50459564e-01 -4.36032042e-02 -1.73532456e-01 2.57867634e-01 1.51605070e-01 1.45367905e-01 1.55723184e-01 1.90487891e-01 3.03994298e-01 7.39785671e-01 1.57666914e-02 -3.48062720e-03 -5.42796135e-01 9.13786113e-01 -3.19585651e-01 1.17109811e+00 -1.81004463e-04 -3.23718816e-01 4.23522770e-01 3.00472856e-01 -1.57829255e-01 -9.52360332e-01 -8.17974806e-01 -2.41015121e-01 6.36711299e-01 3.34299892e-01 -4.45031226e-01 -9.44156587e-01 -6.60502791e-01 -2.94806868e-01 3.86979073e-01 -5.28089464e-01 -6.30634487e-01 -6.12629712e-01 -8.55936587e-01 2.41892412e-01 7.37286866e-01 1.01364005e+00 -8.26498151e-01 -1.46574408e-01 1.89386979e-01 -4.15708274e-01 -8.69384885e-01 -7.04513133e-01 1.54032260e-01 -9.48080480e-01 -9.17112768e-01 -7.76821017e-01 -1.01728225e+00 9.12014365e-01 3.47180277e-01 5.07230222e-01 -2.84255352e-02 -3.32662582e-01 4.81049344e-02 -3.55232596e-01 -1.70452818e-01 -2.15931162e-02 -1.07283071e-01 2.17555493e-01 3.17896932e-01 4.45377558e-01 -7.37745225e-01 -7.45372713e-01 2.66059160e-01 -1.04691362e+00 1.38987452e-01 7.05571175e-01 1.02871037e+00 8.59136224e-01 5.97285986e-01 4.86862451e-01 -6.60369098e-01 2.64658123e-01 -3.48327368e-01 -3.99851292e-01 -8.32780823e-02 -7.13910460e-01 6.37229830e-02 9.02026892e-01 -3.99681449e-01 -9.40392137e-01 4.61020917e-01 -2.27599040e-01 -9.19076622e-01 9.78813842e-02 5.10179996e-01 -5.39539456e-01 7.89281167e-03 4.53357220e-01 6.56766236e-01 2.56328642e-01 -3.93883705e-01 2.83103287e-01 5.60682237e-01 4.69647676e-01 -3.48492801e-01 9.10588145e-01 5.95902622e-01 1.35885775e-01 -8.64765048e-01 -7.30692446e-01 -6.54944658e-01 -7.49188423e-01 -2.08048522e-01 8.84879053e-01 -9.99106467e-01 -7.84804702e-01 5.10737717e-01 -6.49808347e-01 5.43186851e-02 -3.77417237e-01 6.80506110e-01 -6.23315334e-01 8.97482872e-01 -5.67557156e-01 -7.60554075e-01 -3.49110842e-01 -1.15322149e+00 8.37618053e-01 2.09693447e-01 1.73765138e-01 -7.38759220e-01 -3.16003233e-01 4.59814727e-01 1.16647579e-01 2.95877755e-01 9.40459907e-01 -1.72075272e-01 -3.86381954e-01 -5.66410385e-02 -3.07755649e-01 7.47318566e-01 1.51779987e-02 -1.82280347e-01 -7.58235455e-01 -4.60891485e-01 3.20475966e-01 -3.49879235e-01 9.79825377e-01 4.45442408e-01 1.50958788e+00 -4.00007874e-01 -2.15970159e-01 9.04859304e-01 1.38477099e+00 1.92035183e-01 5.33575177e-01 1.39101923e-01 1.04959190e+00 7.31421709e-01 5.23801446e-01 3.93921703e-01 3.25493813e-01 4.19375837e-01 3.31406057e-01 -2.21783638e-01 -1.24273868e-02 -2.70438939e-01 2.46025816e-01 1.44124842e+00 -1.92502901e-01 4.59383249e-01 -6.59676611e-01 4.18026686e-01 -1.98436701e+00 -9.98522520e-01 -1.60398692e-01 2.19617462e+00 8.31962049e-01 9.56848077e-03 1.46986474e-03 4.51112926e-01 6.69875205e-01 2.42364377e-01 -8.74508917e-01 4.16757643e-01 -3.81216049e-01 -6.63426071e-02 1.74038202e-01 1.53985858e-01 -1.36141634e+00 8.47058415e-01 4.70089960e+00 1.35946655e+00 -1.06341243e+00 -5.21715693e-02 6.14727139e-01 -7.15560019e-02 -1.10965706e-01 -2.12240025e-01 -4.75229591e-01 8.57314527e-01 2.84738272e-01 -1.25949541e-02 4.38489228e-01 8.47947717e-01 9.96289030e-02 1.75961807e-01 -9.52474296e-01 1.42164910e+00 2.07708359e-01 -1.32056558e+00 -5.15443161e-02 1.42836124e-01 6.81454241e-01 -1.44284308e-01 1.50221631e-01 2.94622898e-01 -1.15585111e-01 -7.70906985e-01 4.64831769e-01 3.09814095e-01 9.82091188e-01 -1.03983033e+00 6.86076224e-01 5.48211455e-01 -1.49406767e+00 -2.61751205e-01 -5.06467283e-01 2.99648851e-01 -1.25878572e-01 8.08304846e-01 -1.03376590e-01 7.06898332e-01 9.78283882e-01 1.10917866e+00 -4.53278065e-01 7.48238385e-01 1.41116336e-01 4.58296448e-01 -2.19613552e-01 1.55916125e-01 2.86357164e-01 -8.21374893e-01 6.77471399e-01 1.15250492e+00 2.92413980e-01 3.37975293e-01 1.93099573e-01 1.01454604e+00 -2.59812884e-02 3.92411083e-01 -5.41123867e-01 3.73952240e-02 2.41766423e-01 1.32530010e+00 -6.46285355e-01 -2.15482801e-01 -1.93707153e-01 1.15120244e+00 3.29166353e-01 3.70968819e-01 -5.75266600e-01 -4.62737769e-01 3.36675912e-01 1.01630919e-01 3.94559771e-01 -2.63076603e-01 -3.52601498e-01 -1.50220335e+00 2.71597087e-01 -8.12005281e-01 3.63536626e-01 -4.45167929e-01 -1.34412837e+00 4.31422234e-01 -2.90808201e-01 -1.59723294e+00 1.12812713e-01 -4.70202416e-01 -4.14035678e-01 5.45283735e-01 -1.27806032e+00 -1.05010796e+00 -4.12982821e-01 9.31088626e-01 3.16330135e-01 -5.25962651e-01 6.43256068e-01 6.85304046e-01 -1.08336401e+00 6.71016872e-01 5.37122011e-01 3.67454797e-01 5.59654891e-01 -1.18662584e+00 -4.27457541e-01 7.47061074e-01 -6.83675185e-02 6.49632156e-01 3.86123061e-01 -5.15810430e-01 -1.40686178e+00 -1.19734669e+00 3.03594440e-01 1.08753718e-01 4.98214006e-01 -2.99481958e-01 -9.77347672e-01 2.48792544e-01 -3.75778154e-02 1.18979469e-01 8.94680023e-01 -1.51990177e-02 -2.67936796e-01 -5.74211717e-01 -9.72840071e-01 5.21377027e-01 1.10171473e+00 -4.27587211e-01 -2.77021319e-01 2.80839801e-01 5.68829834e-01 -5.47192730e-02 -1.06106806e+00 6.65233135e-01 2.03627020e-01 -1.08320940e+00 9.92530107e-01 -1.54936686e-01 4.98255163e-01 -5.07344425e-01 -3.66985053e-02 -1.21329188e+00 -6.65805161e-01 -2.85185784e-01 -1.47231698e-01 1.42284524e+00 1.25383129e-02 -2.71417201e-01 1.00215113e+00 3.59404773e-01 -1.97460562e-01 -9.98541474e-01 -1.12143612e+00 -7.47586489e-01 1.59920439e-01 -2.76784778e-01 2.93201208e-01 1.29975522e+00 5.84961995e-02 6.51704729e-01 -3.78398687e-01 -5.56947105e-02 9.38590407e-01 8.40610936e-02 4.31895107e-01 -1.33876419e+00 -7.37084150e-02 -5.34395874e-01 -4.86619174e-01 -1.05334353e+00 9.71839055e-02 -9.94845688e-01 -1.49354100e-01 -1.42703748e+00 5.37563264e-01 -4.59944397e-01 -4.28095847e-01 1.67234138e-01 -2.17244491e-01 -2.01039463e-02 1.16259642e-01 4.34163362e-01 -8.44236732e-01 1.21204865e+00 1.41317785e+00 -1.75799489e-01 -1.71930194e-01 -2.59686649e-01 -7.58976877e-01 8.93923283e-01 5.41705787e-01 -1.30659372e-01 -5.21413982e-01 -2.17539310e-01 8.40164721e-02 -4.96816523e-02 1.73310757e-01 -1.10365200e+00 4.57132936e-01 -4.53128405e-02 6.73438728e-01 -6.90156579e-01 3.95729333e-01 -1.03139448e+00 -1.51217461e-01 4.82622474e-01 -7.50664398e-02 -4.72391903e-01 -1.64234847e-01 8.08362365e-01 -5.67502677e-01 -2.57317889e-02 1.02728260e+00 -2.26651385e-01 -6.68958902e-01 5.64878643e-01 -5.39259054e-02 9.41951722e-02 1.12189591e+00 -4.54924315e-01 1.24062926e-01 -2.31258541e-01 -5.41799247e-01 5.65800786e-01 3.51522088e-01 2.82016605e-01 8.46723735e-01 -1.70389211e+00 -5.95003426e-01 4.62299705e-01 1.50748760e-01 4.03380185e-01 7.03435183e-01 9.89070475e-01 -3.09359759e-01 1.40581876e-02 -2.39327550e-01 -7.23165572e-01 -8.58136892e-01 6.12111509e-01 1.85910732e-01 -2.64420301e-01 -6.12390578e-01 8.19692075e-01 4.13431108e-01 -3.86286169e-01 3.42406064e-01 1.36931121e-01 -5.37546873e-01 1.01855986e-01 2.83132195e-01 7.18862951e-01 -3.89579296e-01 -1.00788867e+00 -3.71155947e-01 8.01364720e-01 4.89548445e-02 6.03206307e-02 1.48712313e+00 -1.74420580e-01 -2.54557580e-01 4.65266585e-01 1.57809222e+00 -2.57627040e-01 -1.35037577e+00 -4.17331010e-01 -2.36216038e-01 -3.33187550e-01 1.80262566e-01 -3.26118022e-01 -1.39789736e+00 1.04111731e+00 8.32764208e-01 -2.73295969e-01 1.31028795e+00 -1.55333310e-01 9.26458538e-01 2.33684212e-01 2.07709655e-01 -1.74460685e+00 3.47423971e-01 3.04938436e-01 8.20850968e-01 -1.34712052e+00 8.47958103e-02 -5.93750238e-01 -7.98045218e-01 8.34766448e-01 8.39249671e-01 -1.56795114e-01 8.74757171e-01 -9.94821712e-02 -1.88747913e-01 -3.52088869e-01 -7.86440000e-02 -1.39689460e-01 4.87278014e-01 4.52062160e-01 2.93031484e-01 -5.84272109e-02 -5.34051180e-01 9.52642798e-01 -1.12962434e-02 -1.02261893e-01 -2.77263317e-02 5.15359700e-01 -2.81702876e-01 -8.23510647e-01 -1.50669903e-01 6.11403346e-01 -1.86773822e-01 7.70149380e-02 -3.84625733e-01 5.61776340e-01 4.99459326e-01 8.98747265e-01 -8.26269537e-02 -6.95806146e-01 1.03771240e-01 -2.14093223e-01 -3.07297762e-02 -3.81325722e-01 -2.43599221e-01 5.84847748e-01 -3.76264960e-01 -6.69353306e-01 -6.50484025e-01 -6.40176415e-01 -1.64383554e+00 8.75527635e-02 -4.44907129e-01 1.97779953e-01 3.86747777e-01 6.77362323e-01 4.30131346e-01 4.37100470e-01 9.80312645e-01 -5.07345676e-01 -5.40651858e-01 -8.89453888e-01 -9.78754938e-01 7.54759073e-01 -6.97548836e-02 -6.72116041e-01 -4.58230317e-01 -3.01657151e-03]
[8.483704566955566, 4.023707866668701]
9202691e-9d20-48ff-8393-9515bf21757c
non-intrusive-load-monitoring-with-fully
1812.03915
null
http://arxiv.org/abs/1812.03915v1
http://arxiv.org/pdf/1812.03915v1.pdf
Non-Intrusive Load Monitoring with Fully Convolutional Networks
Non-intrusive load monitoring or energy disaggregation involves estimating the power consumption of individual appliances from measurements of the total power consumption of a home. Deep neural networks have been shown to be effective for energy disaggregation. In this work, we present a deep neural network architecture which achieves state of the art disaggregation performance with substantially improved computational efficiency, reducing model training time by a factor of 32 and prediction time by a factor of 43. This improvement in efficiency could be especially useful for applications where disaggregation must be performed in home on lower power devices, or for research experiments which involve training a large number of models.
['Cillian Brewitt', 'Nigel Goddard']
2018-12-10
null
null
null
null
['non-intrusive-load-monitoring', 'non-intrusive-load-monitoring', 'non-intrusive-load-monitoring']
['knowledge-base', 'miscellaneous', 'time-series']
[-1.03061825e-01 6.80293664e-02 -1.30834971e-02 -4.83131230e-01 -4.82149422e-01 -5.11663795e-01 3.60560119e-01 -7.75902122e-02 -1.72002494e-01 9.98640120e-01 2.38959581e-01 -3.53669554e-01 1.67122439e-01 -1.12993240e+00 -2.94071257e-01 -1.05152178e+00 5.79999313e-02 4.56078321e-01 -5.32944322e-01 2.39765003e-01 -6.93508387e-01 7.62148678e-01 -1.34438920e+00 -4.96812016e-02 6.34135306e-01 1.52793241e+00 -4.52924222e-01 6.26769066e-01 4.59562361e-01 7.82470703e-01 -9.01135027e-01 -9.21383724e-02 1.19770646e-01 -2.32694626e-01 -5.91403663e-01 -3.19754392e-01 2.57936180e-01 -1.25411212e+00 -4.30272073e-01 1.03819144e+00 8.17148924e-01 3.19226444e-01 2.70870954e-01 -1.36257160e+00 -2.75446951e-01 8.24533582e-01 -1.27073944e-01 3.85866731e-01 1.23733729e-01 2.01876182e-02 7.61485994e-01 3.66299242e-01 -3.45781893e-01 8.30798090e-01 6.72028780e-01 4.85677928e-01 -1.69328403e+00 -1.06967854e+00 -1.02067776e-01 3.86796653e-01 -1.18280756e+00 -5.55647433e-01 9.43811774e-01 1.46115467e-01 1.87837791e+00 5.95810413e-01 6.37322426e-01 1.28537738e+00 1.98615685e-01 7.56681979e-01 7.93581069e-01 1.25382110e-01 6.32360935e-01 -1.47422761e-01 1.96573421e-01 3.70691091e-01 6.12248778e-01 9.26936716e-02 2.96154488e-02 -4.46101755e-01 1.55852258e-01 2.46513456e-01 -2.90056132e-03 3.19684744e-01 -6.55516446e-01 7.78003633e-01 4.52112496e-01 5.49849868e-01 -5.88350296e-01 5.74783206e-01 7.50721574e-01 9.65818316e-02 6.20009422e-01 3.54500115e-01 -7.99595475e-01 -4.51983303e-01 -1.14737034e+00 -4.50683534e-02 1.09524882e+00 5.81433475e-01 4.38781023e-01 4.44001526e-01 1.51470974e-01 4.20972764e-01 -2.07886510e-02 6.08511031e-01 7.06356227e-01 -1.27474558e+00 4.73460779e-02 6.48924232e-01 2.42771804e-01 -2.90527880e-01 -1.04124641e+00 1.85120121e-01 -1.37800026e+00 2.89899409e-01 -4.16869186e-02 -7.71824241e-01 -6.45457506e-01 1.79235351e+00 1.49734840e-01 -2.97336072e-01 -1.72791287e-01 2.60262549e-01 4.61754650e-01 7.36246645e-01 3.36389512e-01 -3.16624343e-01 1.36551535e+00 -5.94006062e-01 -1.04911232e+00 2.76770052e-02 5.34379721e-01 -1.20860651e-01 5.02876580e-01 2.86164373e-01 -9.73193765e-01 -2.94691354e-01 -1.21235967e+00 -2.87621111e-01 -8.70544672e-01 1.39984101e-01 6.41561866e-01 9.77517188e-01 -9.43648815e-01 8.96946847e-01 -1.43268704e+00 -5.26046753e-01 8.00872266e-01 1.02811837e+00 6.14854731e-02 6.55362785e-01 -9.43518460e-01 9.38729048e-01 5.45047462e-01 -2.07441449e-02 -6.18725479e-01 -8.29676092e-01 -7.82542288e-01 5.99228919e-01 -1.86799437e-01 -6.72133684e-01 1.47105610e+00 -5.58632255e-01 -1.63128424e+00 1.85515285e-01 8.10977146e-02 -7.91509449e-01 5.03944159e-01 -3.36361349e-01 -8.91082525e-01 -2.65007883e-01 -2.36825109e-01 3.12065929e-01 4.88158435e-01 -7.13864386e-01 -9.75301266e-01 -5.69878936e-01 3.57998051e-02 -2.93252677e-01 -4.53737229e-01 -3.26541603e-01 6.57279611e-01 -3.77311438e-01 -6.31584108e-01 -8.49840939e-01 1.42820194e-01 -4.64846104e-01 -6.30034804e-01 -6.61863029e-01 1.48549664e+00 -1.24389207e+00 1.17718947e+00 -1.83338213e+00 -5.85018635e-01 -6.81438521e-02 2.59564936e-01 3.33531380e-01 3.48517179e-01 1.93871811e-01 -7.25197569e-02 -2.22600647e-03 -1.44142762e-01 -6.75686777e-01 6.74397349e-01 1.75323620e-01 -5.37172556e-02 4.37335104e-01 -2.04768553e-01 1.25145257e+00 -5.84107518e-01 -2.99144220e-02 1.03093088e+00 7.69015372e-01 2.58789565e-02 4.10242587e-01 -2.26373315e-01 1.68495625e-01 -1.65828139e-01 6.34690166e-01 4.02912825e-01 -2.28082001e-01 3.45891774e-01 -6.76042855e-01 2.07001805e-01 7.65576899e-01 -7.13709176e-01 1.30060887e+00 -7.73736179e-01 9.60349262e-01 -3.83638851e-02 -8.96141350e-01 3.84922713e-01 5.50921500e-01 1.01024723e+00 -8.65334868e-01 8.46682250e-01 -1.76769361e-01 -3.40231270e-01 -2.05934823e-01 2.79826343e-01 -5.23813926e-02 -3.74051690e-01 8.41289043e-01 -7.68513605e-02 2.63456315e-01 2.87770152e-01 -6.48276389e-01 1.50366640e+00 -2.73236126e-01 4.44845289e-01 -3.76722902e-01 1.40327618e-01 -4.19940919e-01 3.26276720e-01 4.02267754e-01 -4.77933228e-01 -2.37397254e-01 6.50986359e-02 -8.96481454e-01 -9.38264906e-01 -9.45559442e-01 -2.40017086e-01 1.21819735e+00 -4.04573858e-01 -2.23224834e-02 -1.14692903e+00 -6.82260990e-01 -3.54955941e-02 1.40841746e+00 -7.19620824e-01 -3.68370771e-01 -7.94692993e-01 -1.19040155e+00 5.54409206e-01 1.10522294e+00 1.13427150e+00 -1.11349440e+00 -1.33744502e+00 4.57182974e-01 -2.58965999e-01 -1.28340816e+00 -2.33612731e-01 8.66067946e-01 -6.42279446e-01 -9.42968786e-01 -1.94310635e-01 -4.85117763e-01 2.18292817e-01 -3.73837322e-01 1.32228494e+00 -3.05748403e-01 -1.33095831e-01 1.44723505e-01 1.36978954e-01 -5.79839349e-01 -1.77660406e-01 6.58666492e-01 4.14768785e-01 -3.38307202e-01 1.24870169e+00 -1.26963782e+00 -8.05130363e-01 -3.47521454e-02 -7.18328953e-01 -2.51951486e-01 5.51736429e-02 3.80891383e-01 2.43665189e-01 7.98494160e-01 5.01573265e-01 -3.27360243e-01 7.96772480e-01 -5.10523498e-01 -9.45896089e-01 -2.64070779e-02 -1.07447708e+00 7.03197299e-03 1.20243394e+00 -3.81818980e-01 -9.57720876e-01 8.59983861e-02 -2.56053030e-01 -7.87706673e-02 -6.80600405e-01 -4.77604300e-01 -4.82457459e-01 3.21611077e-01 1.97175130e-01 -2.54353248e-02 -5.08012712e-01 -6.35365963e-01 6.97871447e-02 7.39639640e-01 5.39637268e-01 -1.66053891e-01 4.44773674e-01 5.10929823e-01 1.54169410e-01 -5.23535550e-01 -5.32632470e-01 2.08145991e-01 -6.28058970e-01 4.45922434e-01 1.07993114e+00 -9.20824707e-01 -1.39644873e+00 4.83566463e-01 -9.05869722e-01 -8.44112456e-01 -5.96343279e-01 -2.56317183e-02 -4.11460012e-01 -9.32466909e-02 -6.37023509e-01 -6.32196248e-01 -1.10662842e+00 -8.85965645e-01 9.76211309e-01 5.92659295e-01 -9.08236325e-01 -1.18511724e+00 8.97472352e-02 8.64484608e-02 6.14776969e-01 6.81415141e-01 8.85233045e-01 -8.14402342e-01 -1.83383122e-01 -5.07860601e-01 -1.47155792e-01 5.67379594e-01 8.46358895e-01 -3.44364375e-01 -1.23157489e+00 -6.95410788e-01 1.66700559e-03 -1.54081345e-01 2.75668114e-01 5.93642354e-01 1.47510672e+00 -7.48961806e-01 -4.77377802e-01 6.40598655e-01 1.51909375e+00 5.99992871e-01 2.59159267e-01 4.21720296e-02 6.76673174e-01 -1.14195101e-01 -5.46638787e-01 5.34050405e-01 5.11332929e-01 5.10439754e-01 3.02177727e-01 -3.74365151e-02 -8.82601365e-03 4.96427715e-02 2.32635930e-01 5.46218574e-01 1.18316542e-02 -4.04281557e-01 -5.99472821e-01 6.44144058e-01 -1.75896096e+00 -8.88935030e-01 4.54112917e-01 1.59531415e+00 5.82053006e-01 -1.36038467e-01 5.47228754e-01 4.78037596e-01 2.79553115e-01 3.05282414e-01 -1.20161080e+00 -6.39071286e-01 3.46218348e-01 3.91852349e-01 6.21486425e-01 3.75040501e-01 -1.11608183e+00 2.11256221e-01 7.09033394e+00 3.97656858e-01 -9.05847132e-01 3.80754203e-01 7.84095585e-01 -6.87211275e-01 4.21503186e-01 -8.89648497e-01 -4.87662584e-01 1.27746141e+00 1.57893860e+00 -2.90774316e-01 1.02033317e+00 9.55393314e-01 1.49954274e-01 -3.81182730e-01 -1.71289015e+00 1.26099753e+00 -4.28921640e-01 -8.22975814e-01 -3.14561665e-01 4.60346758e-01 8.10862720e-01 3.62541616e-01 -2.10803166e-01 2.99894869e-01 9.96272445e-01 -1.04705584e+00 1.63991064e-01 2.98517197e-01 4.29700553e-01 -1.14192796e+00 1.07365024e+00 1.64072856e-01 -1.33808553e+00 -3.82229835e-01 -2.98275258e-02 -3.59882832e-01 3.50959480e-01 7.54238605e-01 -2.96193510e-01 7.33122304e-02 1.42851734e+00 2.01564968e-01 -3.56629610e-01 1.60717949e-01 -8.02956820e-02 5.54103315e-01 -1.00789893e+00 -2.81619668e-01 -3.33552323e-02 -1.52544677e-01 -7.80929849e-02 1.18900514e+00 2.71954358e-01 2.81454414e-01 -1.76107228e-01 8.65082383e-01 -3.59310597e-01 -8.38686585e-01 -4.35027778e-01 1.16181239e-01 5.51985741e-01 1.40815294e+00 -4.86053586e-01 -5.61553359e-01 -2.02097327e-01 1.27589512e+00 5.63761815e-02 1.76519886e-01 -1.03604746e+00 -6.14519417e-01 1.30999100e+00 -2.15459436e-01 4.67561126e-01 -5.28793782e-02 -4.88961548e-01 -7.87061334e-01 -8.70944336e-02 -4.52955276e-01 1.88824683e-01 -5.08998036e-01 -1.22772849e+00 5.03815830e-01 1.28100634e-01 -3.46150219e-01 -7.54232883e-01 -4.60726947e-01 -7.82694936e-01 5.14125466e-01 -1.02983403e+00 -1.01669931e+00 -6.96374059e-01 6.21727645e-01 3.37186217e-01 5.75472750e-02 1.31521499e+00 5.15134037e-01 -1.11779976e+00 7.36413658e-01 5.37044048e-01 3.94627333e-01 -2.73195595e-01 -1.55140483e+00 7.88628399e-01 6.58735812e-01 -1.86640039e-01 -1.23239271e-01 8.73140097e-01 -1.84723511e-01 -1.13024437e+00 -1.35994780e+00 5.96344590e-01 -7.48961270e-01 5.12743235e-01 -4.37932193e-01 -4.03938979e-01 1.09018660e+00 6.59854054e-01 -6.02419302e-02 9.60843921e-01 -9.67714638e-02 1.47720492e-02 -6.79560781e-01 -1.79196680e+00 2.48945102e-01 8.22912216e-01 -7.38071561e-01 -3.20030481e-01 1.76421687e-01 3.30926180e-01 3.89747322e-02 -1.36544704e+00 4.15681213e-01 6.94436908e-01 -9.89454091e-01 6.28664672e-01 -9.83426645e-02 -2.67482191e-01 -1.71904229e-02 -4.42500740e-01 -1.56111240e+00 -7.20268548e-01 -5.76151729e-01 -1.38943779e+00 1.46703982e+00 -1.32499993e-01 -8.53239775e-01 7.67223775e-01 1.27612412e+00 4.43918526e-01 -5.56835055e-01 -1.25954533e+00 -6.17701650e-01 8.34485590e-02 -1.46846831e-01 1.46736860e+00 6.60855234e-01 -5.83541691e-02 3.72632563e-01 6.00400232e-02 2.46442571e-01 8.19526374e-01 -6.86754584e-02 2.43444800e-01 -1.17031109e+00 2.36247644e-01 -5.88783741e-01 -2.16725007e-01 -5.61769545e-01 3.39112699e-01 -1.97221741e-01 -2.51329124e-01 -1.65997994e+00 -5.88278137e-02 2.82238960e-01 -6.53671324e-01 9.24476564e-01 4.11822557e-01 4.17966008e-01 -5.39495014e-02 -4.48288828e-01 -3.95566225e-01 5.01294136e-01 1.38647065e-01 -6.20088637e-01 -2.10023150e-01 -5.84816486e-02 -6.42080605e-01 6.64960325e-01 1.55711102e+00 -2.05496058e-01 -3.29449236e-01 -4.00594741e-01 -2.06839815e-01 -6.50807321e-01 2.36230880e-01 -1.39327049e+00 -1.01379193e-01 2.24904433e-01 9.99689698e-01 -6.93958402e-01 4.26181287e-01 -1.55831516e+00 6.56204760e-01 5.17574847e-01 1.16226189e-01 1.88057452e-01 5.73892355e-01 1.32259890e-01 4.41618800e-01 3.47636104e-01 8.47628534e-01 -1.22928441e-01 -2.32924610e-01 2.04500377e-01 -6.02669179e-01 -6.74151361e-01 1.13902080e+00 4.60853912e-02 -4.14020926e-01 -2.67307043e-01 -6.95038497e-01 2.48164922e-01 2.76731551e-01 3.99154574e-01 -3.73026013e-01 -1.81455207e+00 -7.40239099e-02 1.78000972e-01 -4.72188056e-01 -7.36628994e-02 -3.75738442e-02 3.81098926e-01 -2.51381099e-01 7.12672234e-01 -1.53197035e-01 -2.25014910e-01 -1.12781203e+00 7.30756760e-01 8.70714843e-01 -4.52280849e-01 -7.10562885e-01 1.28821328e-01 -2.39861086e-01 -1.01849101e-01 4.09967542e-01 -9.41990793e-01 2.69486487e-01 1.71546876e-01 7.49393284e-01 1.38921058e+00 4.30740178e-01 -4.47612613e-01 -4.73453701e-01 2.62850702e-01 1.53325230e-01 5.22702754e-01 1.50116575e+00 -3.05757433e-01 -1.31305933e-01 4.13173139e-01 1.67600930e+00 -7.96726167e-01 -1.44417405e+00 1.44811705e-01 -1.98169708e-01 3.66396010e-01 6.62020862e-01 -1.31408489e+00 -1.35023189e+00 5.00369787e-01 1.27137268e+00 9.69102561e-01 1.75703692e+00 -1.02817155e-01 1.55652797e+00 6.07233644e-01 5.52760482e-01 -1.49700832e+00 -7.04196393e-01 -1.08440094e-01 3.05822253e-01 -1.25893366e+00 2.73336601e-02 5.85513949e-01 4.74899232e-01 9.33108032e-01 4.20320749e-01 -2.35836774e-01 1.14717209e+00 7.32294321e-01 -3.13929588e-01 -4.24852818e-02 -4.49248105e-01 5.42683015e-03 -4.14727032e-02 6.91309452e-01 2.84495540e-02 7.56001711e-01 2.67090857e-01 6.48857892e-01 -6.50278330e-01 1.80403814e-01 -4.60900329e-02 8.15679133e-01 -1.21693358e-01 -5.50139904e-01 -1.01845987e-01 8.99785757e-01 -6.21961415e-01 2.37544104e-01 -1.43609464e-01 4.57661241e-01 3.32937568e-01 1.36642778e+00 5.67089558e-01 -4.27479744e-01 4.37664002e-01 3.93377572e-01 3.77040893e-01 1.61001742e-01 -5.43368638e-01 -3.44954938e-01 2.14453712e-01 -9.12121475e-01 -2.39850923e-01 -5.12646735e-01 -9.64320779e-01 -1.27968979e+00 4.63464931e-02 -3.77402455e-01 5.28323948e-01 1.15319502e+00 4.47107166e-01 9.31334496e-01 8.54750395e-01 -1.30419064e+00 -3.11561942e-01 -1.26818335e+00 -8.09100151e-01 2.38510132e-01 8.62661898e-01 -2.78820366e-01 -5.21147847e-01 -3.23602818e-02]
[16.064964294433594, 7.5791754722595215]
9e8e2a77-6639-4833-a29e-2233eca20b60
genplot-increasing-the-scale-and-diversity-of
2306.11699
null
https://arxiv.org/abs/2306.11699v1
https://arxiv.org/pdf/2306.11699v1.pdf
GenPlot: Increasing the Scale and Diversity of Chart Derendering Data
Vertical bars, horizontal bars, dot, scatter, and line plots provide a diverse set of visualizations to represent data. To understand these plots, one must be able to recognize textual components, locate data points in a plot, and process diverse visual contexts to extract information. In recent works such as Pix2Struct, Matcha, and Deplot, OCR-free chart-to-text translation has achieved state-of-the-art results on visual language tasks. These results outline the importance of chart-derendering as a pre-training objective, yet existing datasets provide a fixed set of training examples. In this paper, we propose GenPlot; a plot generator that can generate billions of additional plots for chart-derendering using synthetic data.
['Brendan Artley']
2023-06-20
null
null
null
null
['optical-character-recognition']
['computer-vision']
[ 1.94628403e-01 -1.51201561e-01 -5.76275848e-02 -3.32964778e-01 -6.68570518e-01 -1.11310768e+00 8.66420627e-01 3.27394992e-01 3.96717817e-01 4.23806995e-01 2.71902859e-01 -8.42471659e-01 1.99290246e-01 -5.41404665e-01 -4.89560544e-01 -2.24314700e-03 3.35136987e-02 2.06883937e-01 -8.83542076e-02 -1.92430347e-01 4.03445274e-01 6.51031852e-01 -1.30510199e+00 8.73556972e-01 9.56689417e-01 4.31916744e-01 -5.42989895e-02 8.99486661e-01 -6.83864772e-01 5.65449357e-01 -1.25876963e+00 -4.78626579e-01 -1.77795328e-02 -6.92207754e-01 -3.69621724e-01 1.97717354e-01 5.68411887e-01 1.30939111e-01 1.64819002e-01 7.93565214e-01 3.21118236e-01 -3.11130017e-01 1.02479982e+00 -1.67301857e+00 -1.12084866e+00 7.70352900e-01 -1.06740057e+00 8.24622530e-03 6.35727644e-01 3.43632191e-01 9.93460834e-01 -1.35080934e+00 1.17206109e+00 1.30673933e+00 4.99101192e-01 1.74321055e-01 -1.67297947e+00 -5.51968753e-01 -1.53961912e-01 -2.88890123e-01 -1.08001316e+00 -5.30531332e-02 8.63694608e-01 -8.88483465e-01 9.83511388e-01 7.28241742e-01 9.71767724e-01 1.16738260e+00 5.90928867e-02 7.86307096e-01 1.20772433e+00 -7.20343769e-01 1.73308492e-01 -8.04778263e-02 -2.22750176e-02 5.50927341e-01 5.19285686e-02 -1.45051926e-01 -7.57691085e-01 -3.75388935e-02 8.41976047e-01 -7.56842792e-02 -2.54429411e-02 -4.05637860e-01 -1.46752095e+00 6.52892411e-01 5.05315483e-01 1.59096077e-01 2.81972941e-02 7.35254064e-02 5.67235887e-01 3.59869629e-01 5.85977137e-01 8.96127343e-01 2.18523949e-01 -3.35145384e-01 -1.20070946e+00 5.72743714e-01 4.98005003e-01 1.45400739e+00 5.46089470e-01 1.61410451e-01 -7.30506778e-01 5.43738067e-01 3.88139412e-02 6.53380096e-01 1.34711236e-01 -4.35142428e-01 1.16823375e+00 1.04705524e+00 2.69154273e-02 -9.66432214e-01 -3.51872534e-01 -1.95209354e-01 -6.59368575e-01 8.93017113e-01 3.45927864e-01 -1.11632138e-01 -1.44150949e+00 9.93502378e-01 9.37427878e-02 -6.74398839e-01 -1.94713518e-01 6.80382669e-01 9.18115258e-01 8.02084386e-01 4.19611484e-02 4.16667849e-01 1.33307636e+00 -9.81935859e-01 -9.03918147e-01 -3.10786217e-01 7.67271280e-01 -1.07739115e+00 1.94049108e+00 2.83843219e-01 -9.52028513e-01 -5.28681040e-01 -1.45398831e+00 -5.43636262e-01 -8.96738887e-01 5.51378369e-01 2.97508985e-01 4.88353729e-01 -7.81819880e-01 4.09378678e-01 -6.84324682e-01 -3.94220829e-01 5.75938880e-01 -5.87980807e-01 -2.59992003e-01 2.94255495e-01 -5.74916959e-01 6.70467496e-01 1.96698293e-01 -2.69654304e-01 -3.11660647e-01 -1.14147127e+00 -9.16873753e-01 1.16052553e-01 7.53413364e-02 -5.19607663e-01 1.20462108e+00 -4.46949929e-01 -9.50393021e-01 9.14844394e-01 -1.47290662e-01 -1.11120947e-01 8.62245917e-01 -3.57862055e-01 -4.53301907e-01 -2.52532009e-02 4.06907722e-02 5.88462591e-01 8.37973773e-01 -1.37501037e+00 -1.28379896e-01 -1.76135778e-01 -4.18633252e-01 -4.77702171e-02 1.50126833e-02 4.00365800e-01 -2.44573697e-01 -1.08993232e+00 -1.10363945e-01 -5.41771889e-01 3.63657437e-02 4.26540464e-01 -1.07780588e+00 1.30860016e-01 1.30103958e+00 -7.99850583e-01 1.65346861e+00 -2.33487058e+00 1.45195387e-04 4.09169376e-01 4.88927215e-01 -5.08446693e-02 -1.03394188e-01 1.14238918e+00 -4.75703031e-01 7.43485451e-01 -1.64439917e-01 -3.54242027e-01 2.10983440e-01 -3.49531561e-01 -5.33267021e-01 1.97823392e-03 4.47490603e-01 1.18404257e+00 -5.42199612e-01 -7.36712456e-01 2.32164532e-01 3.23802948e-01 -1.94847554e-01 3.14697057e-01 -5.03086388e-01 8.98005217e-02 -5.04042692e-02 6.39799416e-01 5.69518924e-01 -5.80136299e-01 2.09428743e-02 -1.41690582e-01 -4.61367399e-01 2.20849648e-01 -9.54778135e-01 1.64025581e+00 -1.16224654e-01 1.43747711e+00 -4.74783391e-01 -1.32887051e-01 1.34665084e+00 -3.04559618e-02 -1.12546213e-01 -8.42041612e-01 -1.48458436e-01 -3.95187736e-02 -2.65799046e-01 -5.05010307e-01 9.26266789e-01 3.82257164e-01 -2.13081941e-01 6.97234571e-01 -2.56626517e-01 -5.16647220e-01 6.71493173e-01 5.63279450e-01 9.15897071e-01 4.08302456e-01 3.76881063e-01 1.41013369e-01 -2.54677653e-01 4.77555901e-01 -3.26536804e-01 5.23528516e-01 6.50574207e-01 1.14225245e+00 1.14823174e+00 -3.91761839e-01 -1.69677341e+00 -1.18453515e+00 1.66661277e-01 8.60129118e-01 -3.09179306e-01 -1.08242106e+00 -7.26453900e-01 -4.58803087e-01 -2.20450368e-02 1.04009378e+00 -9.23184693e-01 3.73393357e-01 -4.96779919e-01 -7.79917315e-02 6.00733221e-01 7.03419626e-01 -9.58514027e-03 -1.16627753e+00 -1.12448943e+00 -1.70335367e-01 3.24080408e-01 -6.54947817e-01 -5.29747009e-01 2.55483478e-01 -6.00363433e-01 -9.40075576e-01 -8.64098966e-01 -5.65341353e-01 9.37900722e-01 2.00749084e-01 1.44778872e+00 -7.56460428e-02 -4.78855044e-01 -1.35945871e-01 -3.81509721e-01 -6.30172133e-01 -6.48235619e-01 1.28429309e-01 -9.68639195e-01 -5.70274413e-01 -9.33000967e-02 -4.74580705e-01 -4.10327911e-01 1.28272355e-01 -1.09800196e+00 8.48644853e-01 3.99634689e-01 4.99653876e-01 6.08961284e-01 -6.41812146e-01 -1.09536491e-01 -1.16287339e+00 1.19631958e+00 -2.01580212e-01 -4.21958447e-01 6.45129263e-01 -6.18140042e-01 2.58457601e-01 7.10315406e-01 -4.93468076e-01 -7.46230602e-01 -1.57710925e-01 4.03420597e-01 -4.44006711e-01 -7.31700286e-02 8.27005267e-01 -1.60672478e-02 6.50461733e-01 1.05758512e+00 -1.12079449e-01 -1.71343192e-01 -6.67727649e-01 1.11203337e+00 3.89529586e-01 5.78317523e-01 -4.16909188e-01 1.16044700e+00 1.41888246e-01 -1.81250110e-01 -4.63340908e-01 -1.76092222e-01 1.08269555e-02 -8.19624782e-01 -4.69517827e-01 8.57198060e-01 -5.10949552e-01 -2.72312790e-01 -4.88754548e-02 -1.34721959e+00 -5.28679252e-01 -4.89926606e-01 -2.10814103e-01 -2.52188712e-01 -8.92367885e-02 -1.57903329e-01 -6.29083753e-01 -4.93718326e-01 -9.12844718e-01 1.18922520e+00 3.75643700e-01 -9.23253000e-01 -7.06085861e-01 9.28244963e-02 -2.85105735e-01 2.33216599e-01 1.12261677e+00 1.32446265e+00 -4.57964689e-01 -5.10275602e-01 -6.93417862e-02 -6.07944369e-01 -2.66523540e-01 2.91349888e-01 9.80295241e-01 -5.37319005e-01 2.93739024e-03 -9.29634690e-01 -2.95755446e-01 4.26772445e-01 -9.63569060e-02 1.09894025e+00 -5.09909809e-01 -5.20040929e-01 7.13972330e-01 1.22614872e+00 3.91515523e-01 6.19441807e-01 4.02278423e-01 9.69709337e-01 3.04243773e-01 3.94147098e-01 4.79981631e-01 2.29550973e-01 4.56047297e-01 2.52410412e-01 -7.35587656e-01 -3.39612901e-01 -8.95030856e-01 -1.13906875e-01 4.12402272e-01 3.32919270e-01 -5.07635891e-01 -1.13650978e+00 3.77524167e-01 -1.64880133e+00 -9.78188157e-01 -5.58215797e-01 1.70410156e+00 7.44048297e-01 2.16352269e-01 3.35388243e-01 2.25985259e-01 5.00268579e-01 5.70053220e-01 -4.72406626e-01 -7.53056526e-01 -2.77136475e-01 2.57406175e-01 1.83609486e-01 6.14581676e-03 -7.86222696e-01 7.63859332e-01 6.80322266e+00 5.96790969e-01 -1.36655450e+00 -4.72500473e-01 6.47365093e-01 -4.71667498e-02 -8.19637477e-01 6.63214996e-02 -2.32534215e-01 2.83180475e-01 5.76161981e-01 -4.78898942e-01 2.55235761e-01 1.05250847e+00 4.29891258e-01 7.05032470e-03 -1.28626108e+00 1.16975021e+00 2.59343479e-02 -1.76575768e+00 3.26956034e-01 -1.12405434e-01 6.11596346e-01 -5.43656826e-01 3.66083384e-01 -3.50887887e-02 5.50573289e-01 -1.41354799e+00 1.24646628e+00 5.62893212e-01 1.36289895e+00 -5.65549374e-01 -1.83257550e-01 -2.48011932e-01 -1.13698387e+00 4.94012982e-01 -2.25157700e-02 1.17276788e-01 2.86542952e-01 4.06975120e-01 -1.19032347e+00 6.36069298e-01 4.49207217e-01 6.63132489e-01 -1.36319160e+00 1.15803611e+00 -4.78498608e-01 5.80378115e-01 -7.97576234e-02 -4.92115617e-01 6.23382255e-02 -1.63747951e-01 2.49406323e-01 1.72522867e+00 6.76732957e-01 -4.97837454e-01 -3.13962847e-02 1.30380070e+00 -4.67224419e-02 3.08640271e-01 -8.11960280e-01 -5.77741981e-01 4.85153496e-01 1.23510218e+00 -9.94069219e-01 -3.26240838e-01 -1.76198840e-01 8.59151602e-01 4.18542415e-01 6.50006235e-01 -9.24689829e-01 -9.76734757e-01 3.64029169e-01 3.83586109e-01 1.21988662e-01 -7.12935030e-01 -1.20381415e+00 -7.66519129e-01 2.47266352e-01 -1.05459571e+00 2.95077354e-01 -1.68176353e+00 -7.77059793e-01 7.06506670e-01 1.67043462e-01 -1.57542622e+00 -4.80094910e-01 -4.76226330e-01 -8.75189662e-01 9.92980838e-01 -6.17645085e-01 -1.33581698e+00 -6.32229745e-01 1.44148037e-01 5.70502579e-01 3.64861786e-02 6.80459857e-01 -9.91944224e-02 -1.85963288e-01 4.06435430e-01 1.61190435e-01 5.70197642e-01 7.73471892e-01 -1.64150143e+00 1.34193826e+00 8.96640956e-01 7.02194214e-01 5.71516573e-01 1.06181204e+00 -7.74812818e-01 -1.36488557e+00 -8.03978920e-01 5.79647124e-01 -5.42935908e-01 8.32085609e-01 -9.56934094e-01 -9.93042588e-01 6.25358164e-01 7.69367814e-01 -2.11579964e-01 6.47891104e-01 5.39286546e-02 -8.92290056e-01 2.69696712e-01 -6.98982537e-01 1.12734866e+00 9.42912579e-01 -5.55982232e-01 -5.93088746e-01 1.73981100e-01 7.70017266e-01 -9.66694534e-01 -4.99302745e-01 -2.89766163e-01 7.16360033e-01 -9.45258915e-01 7.60865927e-01 -7.67924786e-01 1.08891571e+00 -4.68803406e-01 4.87650365e-01 -1.69333673e+00 -4.17903215e-02 -9.30163145e-01 1.03590235e-01 1.53317630e+00 1.02887189e+00 -1.02181435e-01 5.31104803e-01 5.03339529e-01 -1.72938436e-01 -4.71773207e-01 -3.78264010e-01 -6.53623104e-01 -1.00108288e-01 -4.14683133e-01 8.99716079e-01 9.31192935e-01 3.68461311e-01 3.72449458e-01 -1.90920353e-01 -4.68764335e-01 2.32794881e-01 4.22718525e-01 1.25733137e+00 -9.86143827e-01 2.09154084e-01 -6.81861103e-01 -2.33308002e-02 -7.77880371e-01 -5.16024351e-01 -9.44334745e-01 -4.14495021e-01 -2.17355943e+00 -1.32131875e-01 -2.50849992e-01 3.94627094e-01 5.89209974e-01 -1.80435508e-01 3.57160438e-03 6.62958980e-01 1.41021833e-01 -2.52614886e-01 2.15921700e-01 1.53283668e+00 -3.08167070e-01 -4.45783913e-01 -4.43579644e-01 -5.31063259e-01 2.26463854e-01 6.63969696e-01 -4.47624892e-01 -5.66727281e-01 -4.84864652e-01 5.63817739e-01 9.10803452e-02 3.02761823e-01 -8.72757375e-01 -6.86532408e-02 -4.41775590e-01 9.39540744e-01 -1.22937763e+00 1.79981485e-01 -4.00605410e-01 3.11035693e-01 8.07327498e-03 -6.94361985e-01 1.01778507e+00 3.06228638e-01 6.22885413e-02 -6.51962683e-02 1.83277160e-01 3.59472185e-01 -5.45550417e-03 -2.17974827e-01 -2.89416522e-01 -3.22425842e-01 2.69598216e-01 7.50011742e-01 -4.52650458e-01 -1.11638737e+00 -3.63229960e-01 -4.11339402e-01 9.43859816e-02 7.00210035e-01 7.75307775e-01 7.48678327e-01 -1.53910005e+00 -5.97717881e-01 1.79172888e-01 4.65504289e-01 -7.64916688e-02 -1.37380555e-01 7.50436485e-02 -9.81069684e-01 8.95653516e-02 -5.20609736e-01 -6.11708403e-01 -1.46032071e+00 7.49276698e-01 -2.66831845e-01 -2.33824998e-01 -1.03556955e+00 3.46180171e-01 -7.95767158e-02 1.99037455e-02 -1.07245736e-01 -8.54634225e-01 6.53221756e-02 2.88898528e-01 6.77181661e-01 2.91941613e-01 -6.58815727e-02 -1.71303436e-01 -1.19275041e-01 2.44493589e-01 1.46324605e-01 -5.88146091e-01 1.27555788e+00 3.11338902e-01 3.50947171e-01 7.00318635e-01 1.15567541e+00 2.52147526e-01 -1.22388756e+00 2.90705532e-01 3.58683258e-01 -6.07573390e-01 -6.87950134e-01 -1.15121639e+00 -5.79101801e-01 1.15058446e+00 2.60481805e-01 8.75968516e-01 7.77228892e-01 6.90441206e-02 9.74688902e-02 -4.62766550e-02 -1.99339479e-01 -8.36216271e-01 3.37975472e-01 1.74854815e-01 1.70144415e+00 -7.91280210e-01 2.57640213e-01 -4.48569655e-01 -9.72964346e-01 1.52000165e+00 6.95176959e-01 7.51448199e-02 2.27711305e-01 7.89831400e-01 4.01786000e-01 -4.60361391e-01 -7.62926638e-01 -5.44822998e-02 6.51901007e-01 7.92226970e-01 7.77751327e-01 1.37912020e-01 -1.39059857e-01 3.48267481e-02 -8.93146634e-01 -2.90254295e-01 5.90115726e-01 1.08370602e+00 -9.29414928e-02 -9.10763383e-01 -6.38806939e-01 4.15178150e-01 -1.05091490e-01 -1.61034808e-01 -9.72905159e-01 1.19643152e+00 -5.12183487e-01 6.11091435e-01 3.07607710e-01 -2.70522296e-01 6.15913570e-01 2.58950293e-01 1.73325047e-01 -4.80787128e-01 -6.84351146e-01 2.80256957e-01 2.40633518e-01 -2.94789881e-01 1.43745184e-01 -4.12337184e-01 -1.31261396e+00 -2.54816681e-01 4.24705833e-01 -4.90316525e-02 8.97920847e-01 4.27587897e-01 6.18411005e-01 7.98538387e-01 1.38105094e-01 -6.37163997e-01 -2.18462683e-02 -1.13029635e+00 -2.43218303e-01 6.87165856e-01 2.61082917e-01 -3.35093826e-01 2.28615361e-03 5.09232759e-01]
[11.315768241882324, 1.962185263633728]
00675df2-596b-453c-9a89-2296d6c37f11
end-to-end-integration-of-speech-recognition
2204.00540
null
https://arxiv.org/abs/2204.00540v1
https://arxiv.org/pdf/2204.00540v1.pdf
End-to-End Integration of Speech Recognition, Speech Enhancement, and Self-Supervised Learning Representation
This work presents our end-to-end (E2E) automatic speech recognition (ASR) model targetting at robust speech recognition, called Integraded speech Recognition with enhanced speech Input for Self-supervised learning representation (IRIS). Compared with conventional E2E ASR models, the proposed E2E model integrates two important modules including a speech enhancement (SE) module and a self-supervised learning representation (SSLR) module. The SE module enhances the noisy speech. Then the SSLR module extracts features from enhanced speech to be used for speech recognition (ASR). To train the proposed model, we establish an efficient learning scheme. Evaluation results on the monaural CHiME-4 task show that the IRIS model achieves the best performance reported in the literature for the single-channel CHiME-4 benchmark (2.0% for the real development and 3.9% for the real test) thanks to the powerful pre-trained SSLR module and the fine-tuned SE module.
['Shinji Watanabe', 'Yuya Fujita', 'Takashi Maekaku', 'Xuankai Chang']
2022-04-01
null
null
null
null
['robust-speech-recognition']
['speech']
[ 4.61563855e-01 1.86967656e-01 3.12307596e-01 -4.31155324e-01 -1.53666246e+00 -2.00987890e-01 6.68185353e-01 -2.14848742e-01 -4.71586347e-01 2.32307598e-01 3.92166287e-01 -5.16633928e-01 1.26347184e-01 -7.82687217e-02 -6.00532413e-01 -7.07432866e-01 4.27296758e-02 3.64071578e-02 1.58990815e-01 -4.57070172e-01 -9.00916681e-02 3.86021674e-01 -1.76559401e+00 4.24418867e-01 7.11330056e-01 1.21296644e+00 5.38925707e-01 1.11966145e+00 1.20745171e-02 8.21285367e-01 -9.12406981e-01 1.19530618e-01 1.06500395e-01 -3.24997991e-01 -3.99111301e-01 2.05143332e-01 1.16146788e-01 4.74228449e-02 -5.59878170e-01 7.68462002e-01 1.14718556e+00 5.26326835e-01 3.38111401e-01 -6.88512743e-01 -3.18747878e-01 8.53732049e-01 -2.05053583e-01 2.57909298e-01 3.69053990e-01 -1.89449545e-02 7.32765436e-01 -1.40036917e+00 6.11642338e-02 1.08491886e+00 3.31012517e-01 7.36681700e-01 -8.24818552e-01 -6.85255408e-01 -1.91449195e-01 1.04269437e-01 -1.51594579e+00 -1.36228573e+00 6.55768871e-01 -6.82928190e-02 1.36966443e+00 5.08534074e-01 -2.46463958e-02 1.13616419e+00 -3.47736031e-01 9.85047936e-01 1.29748666e+00 -6.48324192e-01 5.27962148e-01 3.88668269e-01 3.55811566e-01 2.85606503e-01 -5.89298368e-01 5.62229156e-01 -7.13933587e-01 1.10764414e-01 2.49173611e-01 -5.79830587e-01 -2.91680604e-01 4.94488239e-01 -8.84424448e-01 3.61773312e-01 1.42606616e-01 4.64012623e-01 -5.47327816e-01 -1.74151793e-01 4.77503747e-01 5.80930293e-01 6.63098693e-01 2.78110690e-02 -4.83999491e-01 -4.54171568e-01 -1.20469868e+00 -3.86004120e-01 5.68208933e-01 7.78435528e-01 2.88064510e-01 7.10616827e-01 -3.38674843e-01 1.72835517e+00 5.97663462e-01 8.80801260e-01 8.58940303e-01 -2.46898443e-01 6.82424426e-01 -3.15963738e-02 -2.76498258e-01 -1.52372569e-01 -4.43296097e-02 -9.25912261e-01 -8.48143637e-01 2.17765838e-01 -2.63215125e-01 -1.79609716e-01 -1.28502440e+00 1.35720873e+00 -2.52644606e-02 4.56054211e-01 8.20059717e-01 7.33378589e-01 1.19756830e+00 1.26256132e+00 -1.31858131e-02 -2.78686076e-01 1.04735672e+00 -1.31059074e+00 -8.25316191e-01 -3.57958674e-01 4.09335852e-01 -1.02868152e+00 9.09356117e-01 4.91094649e-01 -1.22284973e+00 -8.14122796e-01 -1.17441905e+00 2.45091990e-01 -3.08250785e-01 8.52517128e-01 -2.02854291e-01 1.04887891e+00 -1.33517027e+00 5.88100515e-02 -5.46644390e-01 -1.85260385e-01 2.26747449e-02 2.20105499e-01 -4.52110559e-01 2.26776943e-01 -1.24843931e+00 6.56395316e-01 2.27617741e-01 2.76553601e-01 -1.16454732e+00 -5.25515854e-01 -8.65472257e-01 1.60227761e-01 2.32159957e-01 -6.90514862e-04 1.49248433e+00 -9.91799355e-01 -2.35426950e+00 6.69778645e-01 -2.34401569e-01 -7.13615239e-01 1.54905811e-01 -1.81843191e-01 -1.00442946e+00 2.99291313e-02 -4.77482170e-01 1.03209741e-01 1.15345979e+00 -1.21944332e+00 -5.51150560e-01 -3.36961597e-01 -6.60096705e-01 2.86147743e-01 -2.33264342e-01 4.91392523e-01 -4.42856133e-01 -1.09520066e+00 1.20823070e-01 -6.13037586e-01 7.30500557e-03 -8.63348067e-01 -3.36694121e-01 -2.20834419e-01 8.96802962e-01 -1.21054780e+00 1.40890729e+00 -2.57769084e+00 -1.72741085e-01 3.31098676e-01 -3.84688407e-01 1.10181320e+00 -3.64562839e-01 2.50702858e-01 -6.66780055e-01 -3.74900639e-01 -2.28269696e-01 -7.15704560e-01 4.52717580e-02 -8.35965201e-02 -3.81674707e-01 8.60686302e-02 2.21633613e-01 5.24074554e-01 -6.14323676e-01 1.19142860e-01 5.14012575e-01 8.46749067e-01 -1.04547121e-01 7.08114564e-01 2.74782091e-01 1.22364081e-01 1.26745561e-02 4.41987634e-01 6.92234993e-01 3.80937070e-01 -2.68384784e-01 4.68633063e-02 -2.49243617e-01 8.42356384e-01 -1.41110432e+00 1.39405203e+00 -8.92452538e-01 5.65856338e-01 4.44566667e-01 -8.30607772e-01 1.50229049e+00 8.53015602e-01 -5.29207215e-02 -8.02846372e-01 -7.51985312e-02 5.74244857e-01 -1.13747314e-01 -2.17527449e-01 3.40448439e-01 3.45924199e-02 1.95234776e-01 1.46250784e-01 4.60482776e-01 -5.88868298e-02 -3.76699001e-01 2.01569229e-01 1.20697629e+00 -3.86845052e-01 2.95616359e-01 8.39775652e-02 1.20410287e+00 -8.17865133e-01 1.68963283e-01 5.38134038e-01 -2.65475661e-01 6.67322218e-01 -3.38167727e-01 3.86378676e-01 -9.20504212e-01 -1.21939278e+00 -2.62961611e-02 1.11006093e+00 -4.10298496e-01 -4.02104378e-01 -8.28058243e-01 -4.82668400e-01 -5.48186004e-01 9.51975405e-01 5.84937399e-03 -3.07367623e-01 -2.72374094e-01 -4.81172502e-01 8.47704351e-01 5.36304593e-01 6.04203224e-01 -1.04968929e+00 2.61740535e-01 2.79165059e-01 -2.58990712e-02 -1.26379466e+00 -5.99372923e-01 4.39493686e-01 -3.38983297e-01 -1.86824322e-01 -9.73741651e-01 -8.19993496e-01 3.38362455e-01 3.73743266e-01 5.76755226e-01 -3.58710080e-01 1.15952656e-01 5.51965952e-01 -7.57092297e-01 -2.03695506e-01 -1.19785798e+00 -1.96077585e-01 1.22677572e-01 4.43922907e-01 1.26709908e-01 -4.60697651e-01 -1.66437477e-01 4.84349757e-01 -6.34646058e-01 -2.00036868e-01 8.15924346e-01 9.59956110e-01 5.75629950e-01 4.92097810e-02 1.15760136e+00 -3.88390064e-01 5.38160503e-01 -2.66070276e-01 -6.35075867e-01 4.16851401e-01 -5.68039954e-01 -1.20277315e-01 5.69262922e-01 -2.67349511e-01 -1.48694324e+00 2.20952928e-01 -1.04454386e+00 -3.75409752e-01 -4.14516389e-01 3.23506117e-01 -4.84618545e-01 -3.13888453e-02 6.92489326e-01 8.16229105e-01 -1.87364325e-01 -6.79835200e-01 2.60998935e-01 1.89166820e+00 6.41980529e-01 -2.59457886e-01 7.20281601e-01 -2.85911441e-01 -6.37199581e-01 -1.49931204e+00 -4.26281214e-01 -9.67874646e-01 -2.49193475e-01 -2.22886771e-01 4.64319289e-01 -1.36355364e+00 -1.52752161e-01 6.99848652e-01 -1.07700586e+00 -2.92744935e-01 -3.90712827e-01 7.11573005e-01 -6.29713953e-01 3.96948546e-01 -5.14970124e-01 -1.50736475e+00 -8.27466846e-01 -1.24804330e+00 1.37120211e+00 8.73837918e-02 1.12845391e-01 -4.51460898e-01 1.77884653e-01 7.84739554e-01 7.40121365e-01 -7.31589496e-01 2.42397711e-01 -1.15791368e+00 -2.99139619e-01 -1.10082395e-01 8.25103670e-02 1.29678833e+00 8.57432261e-02 -2.58761108e-01 -1.70996213e+00 -3.64285737e-01 4.65506501e-02 -2.86047816e-01 8.69848013e-01 9.85328481e-02 9.65588510e-01 -2.02683598e-01 2.55319595e-01 4.91235971e-01 8.88096452e-01 6.24112368e-01 9.33951259e-01 -4.18197736e-02 2.50719517e-01 3.65781993e-01 5.62564254e-01 2.60018975e-01 -3.19062732e-02 9.21061456e-01 -1.81263849e-01 -2.96371967e-01 -6.71112478e-01 -2.15189382e-01 1.11126244e+00 1.49851298e+00 3.81988674e-01 -2.14622602e-01 -8.77634704e-01 4.96102422e-01 -1.42249298e+00 -8.46769094e-01 1.31578460e-01 2.44536376e+00 8.60222936e-01 3.61866206e-02 1.33363396e-01 7.31306255e-01 7.83454776e-01 1.24870814e-01 -1.51404813e-01 -5.24080813e-01 -3.84793311e-01 6.49528682e-01 1.29239336e-01 7.54623413e-01 -1.07748294e+00 9.37328935e-01 5.64930820e+00 1.50927901e+00 -1.18790174e+00 3.72642845e-01 4.74378556e-01 2.35518411e-01 1.53246209e-01 -5.11547685e-01 -8.41970146e-01 2.88783789e-01 1.77028704e+00 -2.16003414e-02 5.11146605e-01 9.44057584e-01 4.76906478e-01 2.28986755e-01 -6.94651961e-01 1.17380512e+00 1.72250286e-01 -8.40400934e-01 -6.13180809e-02 -3.62217903e-01 3.69376510e-01 3.81968141e-01 2.09845871e-01 5.82716286e-01 -3.58480737e-02 -8.16267490e-01 5.90291440e-01 1.55329928e-01 1.09129786e+00 -7.82145977e-01 7.24193096e-01 3.13381463e-01 -1.30548990e+00 -1.67430967e-01 -4.53910381e-02 5.88024199e-01 8.55786204e-02 5.89050055e-01 -1.30788052e+00 7.24386215e-01 4.29532528e-01 3.25287014e-01 -3.81599098e-01 1.01080167e+00 -3.75266552e-01 1.28579187e+00 -3.31433952e-01 1.56586096e-01 1.03229865e-01 2.27843791e-01 1.02354050e+00 1.69870830e+00 3.63588035e-01 -3.96799520e-02 -3.09789449e-01 3.56169611e-01 -7.26181939e-02 3.80018353e-01 -2.48289824e-01 -2.34757513e-01 5.96611977e-01 1.09940231e+00 -1.62973240e-01 -3.15517247e-01 -2.70683020e-01 1.01824939e+00 -5.89102618e-02 4.54463929e-01 -5.29865980e-01 -6.78865075e-01 5.25374651e-01 -1.51794210e-01 4.73694503e-01 -1.36191115e-01 6.69802204e-02 -9.64589357e-01 1.52107794e-02 -1.31972444e+00 9.98973399e-02 -1.02627003e+00 -1.14404500e+00 1.24997687e+00 -5.17678916e-01 -1.11947119e+00 -3.83020729e-01 -5.61992586e-01 -4.84343439e-01 1.17444146e+00 -1.74713194e+00 -1.07347178e+00 2.56139990e-02 6.48576975e-01 1.09843290e+00 -8.48338962e-01 8.66720974e-01 6.32331431e-01 -6.84649289e-01 1.06343842e+00 2.92696893e-01 2.07803827e-02 4.72838402e-01 -1.13984406e+00 7.23427057e-01 1.10634470e+00 3.22376460e-01 3.48717660e-01 4.30728287e-01 -3.90326440e-01 -1.22017491e+00 -1.27355647e+00 9.93546486e-01 -1.05560377e-01 5.49767196e-01 -5.09928524e-01 -9.90724683e-01 4.73711699e-01 9.34927762e-02 -5.08529833e-03 7.12698579e-01 -2.61647344e-01 -4.40585762e-01 -4.10890192e-01 -9.37401056e-01 1.64475396e-01 6.10150158e-01 -9.94247675e-01 -6.97970808e-01 1.27706066e-01 1.01107001e+00 -3.25308889e-01 -8.10652435e-01 4.19225931e-01 8.82239342e-02 -5.75764596e-01 9.56093550e-01 -1.87529027e-01 -4.67127562e-02 -4.58588094e-01 -6.50090516e-01 -1.65294206e+00 1.65120795e-01 -1.09712970e+00 -1.12369575e-01 1.61547935e+00 8.01893950e-01 -5.74324131e-01 2.64377356e-01 -1.43732075e-02 -5.99769652e-01 -4.26095992e-01 -1.20534062e+00 -1.08022404e+00 -5.27654350e-01 -7.60629952e-01 3.58998448e-01 5.24053335e-01 -1.20916858e-01 5.21354139e-01 -4.48614031e-01 5.33430696e-01 5.35807550e-01 -5.94165921e-01 6.92494154e-01 -5.58230579e-01 -6.02636874e-01 -1.23174436e-01 -4.44513768e-01 -1.28150773e+00 1.89860016e-01 -9.26173270e-01 4.62988764e-01 -1.02982974e+00 -2.95374364e-01 -2.64797270e-01 -5.36485553e-01 3.14822972e-01 -1.62212849e-01 -2.29188442e-01 1.52328268e-01 -2.35597566e-01 -5.17817795e-01 9.30404902e-01 5.77240765e-01 -5.34242950e-02 -3.66153091e-01 5.90133786e-01 -4.71783549e-01 3.10679287e-01 5.65807164e-01 -2.47708499e-01 -3.61932367e-01 4.23988588e-02 -6.60324991e-01 2.17338711e-01 2.10465595e-01 -1.08791506e+00 3.87311548e-01 5.35302997e-01 -1.55216202e-01 -7.38220274e-01 5.61611772e-01 -6.90542579e-01 -1.97328702e-01 4.63293120e-02 -5.87781370e-01 -5.26990891e-01 3.36119235e-01 4.50850099e-01 -6.79647684e-01 -1.28381327e-01 1.11378384e+00 4.75103587e-01 -7.58063734e-01 -1.33161068e-01 -5.32238722e-01 -2.76810437e-01 6.23497427e-01 5.75422049e-02 -2.40764797e-01 -6.21207237e-01 -9.92809474e-01 -5.83522655e-02 -4.80840594e-01 5.52818358e-01 1.05080926e+00 -1.06770670e+00 -1.00542557e+00 6.02814436e-01 2.47708693e-01 -4.08808649e-01 4.73634452e-01 4.80996370e-01 1.70408651e-01 3.93035918e-01 3.23888779e-01 -5.11526704e-01 -1.58769143e+00 1.61593705e-01 5.73026359e-01 -3.66348699e-02 -3.29547048e-01 1.05457234e+00 -1.64808363e-01 -7.40265012e-01 7.61261284e-01 -1.16449460e-01 -2.97260791e-01 -3.13419849e-01 9.39412415e-01 5.70215523e-01 7.93353856e-01 -9.42112327e-01 -3.68705153e-01 1.56829327e-01 -6.77295476e-02 -5.01087368e-01 1.41641247e+00 -1.84147984e-01 2.85299838e-01 4.67587471e-01 1.17993665e+00 1.18471734e-01 -7.42214024e-01 -5.74197173e-01 -6.92640990e-03 -1.27722412e-01 6.65177822e-01 -1.27255511e+00 -7.82644570e-01 9.43268120e-01 1.09816194e+00 4.85046767e-02 1.42548752e+00 -1.16383135e-01 8.35085988e-01 5.81289351e-01 1.22206450e-01 -1.18294168e+00 -5.29152248e-03 7.11986780e-01 1.24484992e+00 -9.67470348e-01 -6.46324813e-01 -4.03522611e-01 -7.82780647e-01 8.51779103e-01 3.58084381e-01 2.44872347e-01 8.00836205e-01 6.86317682e-01 3.89319956e-01 2.32113436e-01 -8.58144343e-01 -3.81869942e-01 4.41160142e-01 6.12531483e-01 3.92810464e-01 6.45032972e-02 -1.36086503e-02 8.97767723e-01 -3.35844159e-01 -2.42126688e-01 2.57753223e-01 5.45864344e-01 -6.45889163e-01 -1.06001949e+00 -5.93861401e-01 1.93286291e-03 -3.31858039e-01 -3.29294652e-01 -2.85597652e-01 1.13813505e-01 -4.33442950e-01 1.53959262e+00 -1.42780378e-01 -6.62200809e-01 7.93376327e-01 2.62915343e-01 6.82435408e-02 -7.41817951e-01 -9.21881080e-01 5.21861255e-01 3.84352237e-01 -4.43699807e-01 -2.91206777e-01 -4.27785963e-01 -1.11899841e+00 4.12287980e-01 -5.54518044e-01 2.37028658e-01 1.22468293e+00 8.76588464e-01 4.92266685e-01 8.03915024e-01 1.38831103e+00 -5.12511194e-01 -9.88831580e-01 -1.17906189e+00 -6.97797954e-01 1.23228289e-01 6.36451840e-01 -2.47859597e-01 -5.88776886e-01 1.08200330e-02]
[14.688508987426758, 6.214816570281982]
6c20ff67-bd32-45dd-ab8b-652f47984cb3
a-framework-for-differentiable-discovery-of
null
null
https://openreview.net/forum?id=ueiBFzt7CiK
https://openreview.net/pdf?id=ueiBFzt7CiK
A Framework For Differentiable Discovery Of Graph Algorithms
Recently there is a surge of interests in using graph neural networks (GNNs) to learn algorithms. However, these works focus more on imitating existing algorithms, and are limited in two important aspects: the search space for algorithms is too small and the learned GNN models are not interpretable. To address these issues, we propose a novel framework which enlarge the search space using cheap global information from tree decomposition of the graphs, and can explain the structures of the graph leading to the decision of learned algorithms. We apply our framework to three NP-complete problems on graphs and show that the framework is able to discover effective and explainable algorithms.
['Le Song', 'Xin Gao', 'Yu Li', 'Xinshi Chen', 'Hanjun Dai']
2021-01-01
null
https://openreview.net/forum?id=5UvvKsBTDcR
https://openreview.net/pdf?id=5UvvKsBTDcR
neurips-workshop-lmca-2020-12
['tree-decomposition']
['graphs']
[ 3.04781139e-01 5.25871098e-01 -5.18219471e-01 4.69786488e-02 2.67181635e-01 -7.06813693e-01 3.50483626e-01 2.26697236e-01 1.86869964e-01 7.22447932e-01 -1.24757521e-01 -7.05114543e-01 -6.36391103e-01 -1.40632617e+00 -9.93756950e-01 -3.22719038e-01 -1.71580225e-01 7.84933329e-01 2.27394640e-01 -8.75157490e-02 3.28772396e-01 5.11810839e-01 -1.27647102e+00 2.01531366e-01 9.26441014e-01 4.78781134e-01 -1.25112191e-01 8.57137322e-01 -4.42928970e-01 1.02240503e+00 -3.57689828e-01 -3.50166768e-01 2.03308254e-01 -8.02391231e-01 -1.31468427e+00 7.45738670e-02 9.22447294e-02 8.04545283e-02 -6.88520253e-01 1.26995015e+00 -3.34347218e-01 9.06599779e-03 4.75794494e-01 -1.61102545e+00 -8.02761853e-01 1.28970909e+00 -2.68176585e-01 1.35083809e-01 1.89651862e-01 -9.28938687e-02 1.35468853e+00 7.80126173e-03 8.17692161e-01 1.28170693e+00 5.98807395e-01 8.27398598e-01 -1.17224562e+00 -3.12083781e-01 2.55006135e-01 6.52971268e-01 -9.02909815e-01 5.15216887e-02 8.50840747e-01 -1.82763100e-01 8.30164850e-01 4.14120525e-01 1.03612959e+00 8.32219541e-01 1.09379485e-01 7.44495869e-01 9.40469086e-01 -7.40458846e-01 2.36840129e-01 -1.86577350e-01 4.54071641e-01 1.11848152e+00 6.88999414e-01 3.34335744e-01 -3.45098585e-01 6.80116713e-02 1.04237437e+00 7.83853456e-02 -4.30917591e-01 -8.65325928e-01 -8.71487081e-01 1.12662077e+00 8.34793389e-01 3.88501763e-01 3.19734700e-02 4.56825852e-01 1.67610660e-01 3.92875165e-01 2.56077766e-01 9.02751982e-01 -4.09953535e-01 -3.58840887e-04 -6.01776481e-01 -2.68920767e-03 1.20536709e+00 7.41485357e-01 8.70809495e-01 6.22425824e-02 5.14738023e-01 1.30098965e-02 9.62726399e-02 7.74707645e-02 1.42570317e-01 -8.34062696e-01 4.13970023e-01 1.09824288e+00 -4.36320901e-01 -1.32299459e+00 -5.92678249e-01 -5.10367870e-01 -7.96209037e-01 -1.10339180e-01 4.91674930e-01 1.29832849e-01 -1.00774610e+00 1.69567943e+00 6.83849677e-02 1.63450539e-01 -1.73806250e-01 7.25048840e-01 7.54673839e-01 5.94494462e-01 -1.90327123e-01 -1.47926047e-01 6.89694583e-01 -1.44262695e+00 -6.27720833e-01 -4.13377136e-01 8.54388237e-01 1.85016930e-01 7.12135613e-01 3.43327314e-01 -1.11578000e+00 -3.63174140e-01 -1.11227000e+00 1.20072462e-01 -5.77750623e-01 -3.92622113e-01 1.19304514e+00 5.50352752e-01 -1.30377913e+00 9.55074847e-01 -8.04691136e-01 -4.82303470e-01 4.68122184e-01 8.18086267e-01 -1.40022159e-01 -2.22887784e-01 -1.06190670e+00 9.04385209e-01 8.19807708e-01 1.53106689e-01 -7.77689755e-01 -1.88797489e-01 -8.63038063e-01 3.25370759e-01 8.84628892e-01 -1.13570511e+00 8.96633923e-01 -1.26888895e+00 -1.26454270e+00 6.14521086e-01 -5.12751937e-02 -7.42999554e-01 2.65591055e-01 3.70187908e-01 -1.30067468e-01 1.41893610e-01 -4.38849866e-01 2.97385722e-01 4.92473215e-01 -1.17614508e+00 -3.88779104e-01 -5.78340888e-01 5.88464856e-01 -1.54228151e-01 -3.45725775e-01 -2.92835265e-01 -1.60404146e-01 -2.39456937e-01 2.10145876e-01 -9.77137804e-01 -5.69598734e-01 -3.62034231e-01 -4.38714325e-01 -3.52550417e-01 7.22498238e-01 -1.81828916e-01 1.30906498e+00 -1.54853666e+00 6.87516332e-01 4.20119792e-01 1.00258756e+00 1.49149209e-01 -2.90982574e-01 3.88591558e-01 -8.58737007e-02 6.81322515e-01 2.68580373e-02 3.61878425e-01 1.46408126e-01 8.17929089e-01 -1.89049587e-01 1.98149055e-01 -1.01354569e-01 1.35359240e+00 -9.04892683e-01 -2.67349660e-01 1.34664282e-01 -9.36563089e-02 -5.79945266e-01 1.37207434e-01 -5.31037092e-01 3.01373184e-01 -6.52794898e-01 3.89727712e-01 4.62151170e-01 -7.12719262e-01 5.83685935e-01 4.08450931e-01 3.56611341e-01 3.77008229e-01 -1.09091485e+00 1.26431596e+00 -2.55996466e-01 5.71273804e-01 -1.67912960e-01 -1.45110679e+00 5.70245087e-01 -1.38027787e-01 8.23340416e-02 -2.66359925e-01 1.20585971e-01 5.73025793e-02 1.89290226e-01 -2.89108872e-01 1.07475176e-01 5.04425131e-02 3.27274442e-01 6.84482396e-01 -1.20903581e-01 -7.06339628e-02 3.63815218e-01 2.49743253e-01 1.42282343e+00 -1.52719811e-01 4.06300962e-01 -1.58970371e-01 5.40944457e-01 2.04619169e-01 3.34445834e-01 1.03347206e+00 9.87863094e-02 8.65370035e-02 8.65315974e-01 -1.02513981e+00 -8.86037111e-01 -7.07607269e-01 5.62944531e-01 1.15134287e+00 3.08078796e-01 -7.51093745e-01 -9.58397150e-01 -1.11988151e+00 -2.63098389e-01 5.58298230e-01 -9.33065772e-01 -2.20437422e-01 -6.68762207e-01 -4.17141050e-01 3.21830004e-01 6.21121347e-01 1.28627688e-01 -1.27464437e+00 -3.89436096e-01 7.60253742e-02 -8.61837640e-02 -1.00040281e+00 -1.00867145e-01 4.73917633e-01 -1.44870543e+00 -1.53172314e+00 -5.82175367e-02 -8.86337757e-01 1.08333623e+00 2.77468234e-01 1.42135072e+00 7.52141654e-01 -1.69514403e-01 4.34176683e-01 -2.53216505e-01 -3.88725519e-01 -6.26132190e-01 4.84854311e-01 -2.25761175e-01 -4.78091627e-01 3.04110974e-01 -6.81421399e-01 -5.30163050e-02 1.51392817e-01 -9.39067125e-01 3.94599736e-01 4.11525160e-01 6.60503983e-01 3.70442271e-01 6.01680696e-01 1.46957472e-01 -1.23565090e+00 6.66251063e-01 -3.48240197e-01 -9.39139664e-01 4.85827029e-01 -1.01797843e+00 6.50413156e-01 1.07614291e+00 -2.59967715e-01 -2.88845152e-01 4.42907121e-03 4.49049741e-01 -2.48334840e-01 -1.39000341e-01 6.74549997e-01 -8.60759094e-02 -4.37024057e-01 5.01479089e-01 2.03099370e-01 -1.35211900e-01 -2.08478525e-01 4.39644784e-01 -2.86226533e-02 2.36524776e-01 -6.23412967e-01 1.14127052e+00 2.52106518e-01 6.15783632e-01 -3.85438502e-01 -8.91500354e-01 -2.93982513e-02 -6.36239529e-01 -2.82262941e-03 4.32775825e-01 -1.25791103e-01 -1.00209010e+00 -1.73412021e-02 -1.14360082e+00 -3.36356938e-01 -2.37641931e-01 1.90394029e-01 -7.42616117e-01 4.52311426e-01 -6.92338109e-01 -6.99256003e-01 -1.80284694e-01 -1.18390000e+00 4.20723528e-01 3.69353294e-01 -1.71991572e-01 -1.34755075e+00 -2.04188470e-02 3.15792143e-01 3.63107681e-01 9.16182101e-02 1.62641704e+00 -9.39456522e-01 -1.10361981e+00 -1.47225177e-02 -3.34623873e-01 -1.23012714e-01 -1.72259826e-02 -8.98809135e-02 -5.09659469e-01 -1.58525378e-01 -6.24430254e-02 -2.46918663e-01 7.67578483e-01 4.95562613e-01 1.68970573e+00 -7.42778897e-01 -5.86726487e-01 7.51328588e-01 1.53855181e+00 8.53905529e-02 5.65340757e-01 6.34001672e-01 8.49229991e-01 3.91909480e-01 -1.33441284e-01 -2.85126586e-02 2.26744413e-01 3.86507124e-01 8.31127346e-01 5.07548079e-02 9.57382992e-02 -7.12072849e-01 6.63631335e-02 7.26186633e-01 -6.91056192e-01 -6.15445077e-01 -1.00201976e+00 3.58443916e-01 -2.07741094e+00 -7.75102377e-01 -2.89025098e-01 1.82291436e+00 3.40508431e-01 2.39645928e-01 1.65863216e-01 3.57921690e-01 8.45418215e-01 2.35429049e-01 -5.74133575e-01 -7.30707347e-01 7.15999911e-03 4.52792317e-01 4.86317784e-01 5.24785817e-01 -7.96705544e-01 1.10284126e+00 7.44766235e+00 5.40089071e-01 -7.77042925e-01 -4.17366773e-01 4.40954804e-01 4.39618111e-01 -4.63786364e-01 3.94813359e-01 -3.23714852e-01 -9.98043194e-02 1.18969548e+00 -5.34725606e-01 1.15678346e+00 9.53257084e-01 -1.42509490e-01 3.22732717e-01 -1.40813482e+00 7.98410773e-01 6.52026385e-03 -1.65998018e+00 4.65476751e-01 2.77722389e-01 7.98539698e-01 -2.70653218e-01 3.99256833e-02 2.24528477e-01 5.71516573e-01 -1.56646454e+00 9.05225798e-02 2.06793532e-01 1.79978341e-01 -1.00270772e+00 7.13997722e-01 5.55038929e-01 -1.03957689e+00 -1.93590119e-01 -5.43299615e-01 -5.87016582e-01 -5.77036500e-01 2.07834005e-01 -1.06946194e+00 6.73523605e-01 2.75064647e-01 6.69630408e-01 -7.03655541e-01 1.05626738e+00 -7.87176073e-01 6.21648788e-01 -1.81052819e-01 -5.55003822e-01 4.01945978e-01 -2.98134297e-01 5.12023211e-01 6.79714203e-01 1.27223670e-01 2.34496713e-01 2.41152152e-01 1.22178912e+00 -4.13168788e-01 -1.23324588e-01 -8.61944497e-01 -5.88954508e-01 -1.02962263e-01 1.14707327e+00 -1.14554358e+00 -1.77241787e-02 -1.35456324e-01 8.03881466e-01 6.19858503e-01 2.56188065e-01 -8.27493072e-01 -1.23469606e-01 -7.30812252e-02 2.17502028e-01 3.32344621e-01 -3.82298261e-01 -3.95709366e-01 -1.18245018e+00 -1.81895599e-01 -1.12838984e+00 6.40920997e-01 -7.91194022e-01 -9.83294308e-01 6.18043482e-01 -2.82861680e-01 -4.94308829e-01 -2.39045143e-01 -8.49993467e-01 -5.27756751e-01 4.08558756e-01 -1.29321265e+00 -9.92029667e-01 -2.62649179e-01 4.15138841e-01 3.42490554e-01 1.82622876e-02 9.23551500e-01 -3.27264041e-01 -3.04284036e-01 4.01645839e-01 -6.28485978e-02 1.41686380e-01 -2.14922264e-01 -1.49514353e+00 3.84618342e-01 8.29505861e-01 6.99815094e-01 6.85249865e-01 7.86728024e-01 -4.27050531e-01 -1.90747321e+00 -8.32118094e-01 8.08371007e-01 -6.82844162e-01 9.74420309e-01 -1.88711092e-01 -7.00475097e-01 9.51965690e-01 5.66946864e-02 -8.34669620e-02 3.17517638e-01 6.13787949e-01 -3.09311539e-01 2.12413579e-01 -8.86987329e-01 6.01111829e-01 1.44416034e+00 -2.67959118e-01 -5.72214007e-01 5.21803617e-01 8.06628048e-01 -4.16360855e-01 -4.48791891e-01 3.51678073e-01 1.86703041e-01 -8.46218050e-01 6.68261468e-01 -1.27726960e+00 7.33101785e-01 -6.03368431e-02 2.52094895e-01 -1.45503080e+00 -4.22394633e-01 -8.08749139e-01 -6.36473358e-01 5.36515653e-01 4.99560297e-01 -6.57378078e-01 1.33298075e+00 3.92043829e-01 1.99022859e-01 -7.74520159e-01 -6.55892372e-01 -7.18683898e-01 -7.55005777e-02 -3.99933189e-01 6.72095656e-01 8.76360416e-01 1.32284880e-01 6.13834262e-01 -3.92530650e-01 1.55156597e-01 6.60872638e-01 6.63432717e-01 6.68859005e-01 -1.60592604e+00 -4.16299492e-01 -6.39899909e-01 -3.76996309e-01 -1.01794314e+00 4.97322500e-01 -1.18368149e+00 -2.91916877e-01 -1.88322484e+00 5.84471464e-01 -5.83465844e-02 -3.72714102e-01 6.85520411e-01 1.48635715e-01 -2.44871974e-01 1.65327534e-01 5.18750288e-02 -9.33842182e-01 2.90146649e-01 1.46567214e+00 -3.96701157e-01 -1.05971538e-01 2.43840829e-01 -8.05307984e-01 9.27062750e-01 6.53171659e-01 -6.52132452e-01 -4.48072076e-01 -5.33512592e-01 7.06545174e-01 7.38949254e-02 2.62566388e-01 -9.06725109e-01 3.84650916e-01 -4.65096086e-01 2.13094637e-01 -2.34218657e-01 -3.40645522e-01 -9.67805982e-01 1.76042393e-01 9.20173466e-01 -5.59054375e-01 1.53064832e-01 1.06549589e-02 8.90070319e-01 -6.39809221e-02 -4.41745758e-01 4.66620117e-01 -4.21182632e-01 -5.11680782e-01 5.50340831e-01 -2.63152987e-01 2.27685347e-02 8.05141449e-01 -3.61870468e-01 -2.70190865e-01 -5.51638901e-01 -8.15504253e-01 2.51346916e-01 5.81009030e-01 2.43049741e-01 4.67101425e-01 -9.43204343e-01 -2.91977048e-01 8.81221071e-02 -9.11678746e-02 1.17523164e-01 -1.47936746e-01 4.02342141e-01 -8.44272375e-01 8.23139906e-01 -1.89046934e-01 -2.56787658e-01 -1.22646701e+00 1.03647006e+00 4.87472862e-01 -8.24507475e-01 -4.87472802e-01 5.92130840e-01 1.81487694e-01 -5.89332163e-01 2.41473883e-01 -4.55018073e-01 -2.43298858e-01 -5.54128408e-01 2.35531732e-01 4.39180315e-01 -2.74776667e-01 -5.99551015e-02 -3.10142696e-01 5.27678847e-01 -4.40225676e-02 5.03826320e-01 1.85836542e+00 1.83758810e-01 -6.25578582e-01 1.36538923e-01 6.80770278e-01 -2.16519982e-01 -5.42201996e-01 -2.54551709e-01 3.11331421e-01 -2.08518922e-01 -8.34466144e-02 -6.95111275e-01 -1.19122434e+00 8.96586895e-01 1.27211399e-02 8.10524046e-01 1.26882350e+00 2.01850295e-01 6.74211562e-01 8.53576958e-01 5.00539541e-01 -9.57277536e-01 -2.17571780e-02 8.64301503e-01 4.52800155e-01 -8.97572696e-01 9.10386667e-02 -6.70701683e-01 2.36137547e-02 1.79088652e+00 6.86886847e-01 -8.08875114e-02 1.41477793e-01 -4.08317819e-02 -5.90081573e-01 -4.75018531e-01 -7.33968973e-01 -9.60636437e-02 2.67343372e-01 6.46077752e-01 -1.85182486e-02 4.66300175e-02 -2.47807220e-01 6.91610813e-01 -2.74165094e-01 1.71804592e-01 8.52652848e-01 6.73895538e-01 -5.66917598e-01 -1.17336237e+00 -1.82691947e-01 2.80864805e-01 -3.16260934e-01 -1.35111481e-01 -1.09195030e+00 9.59657073e-01 -3.73293720e-02 8.56196642e-01 -4.43667442e-01 -4.21422601e-01 -1.26827016e-01 -2.34241381e-01 8.77527177e-01 -6.04808211e-01 -1.93430498e-01 -4.93929207e-01 1.32784113e-01 -5.78850925e-01 -4.75615829e-01 1.30562142e-01 -1.13763797e+00 -6.36276007e-01 -4.63856786e-01 4.89878654e-01 3.19725186e-01 1.08817458e+00 1.24542385e-01 6.29802585e-01 5.40221512e-01 -3.20518583e-01 -3.69491875e-01 -4.01908189e-01 -5.49840987e-01 7.07233474e-02 8.14156309e-02 -3.02468330e-01 -6.97149158e-01 -3.36906493e-01]
[7.313413143157959, 6.139611721038818]
6bdfbce2-b6e0-482f-a1b0-dc4fd5f3af3d
game-plan-what-ai-can-do-for-football-and
2011.09192
null
https://arxiv.org/abs/2011.09192v1
https://arxiv.org/pdf/2011.09192v1.pdf
Game Plan: What AI can do for Football, and What Football can do for AI
The rapid progress in artificial intelligence (AI) and machine learning has opened unprecedented analytics possibilities in various team and individual sports, including baseball, basketball, and tennis. More recently, AI techniques have been applied to football, due to a huge increase in data collection by professional teams, increased computational power, and advances in machine learning, with the goal of better addressing new scientific challenges involved in the analysis of both individual players' and coordinated teams' behaviors. The research challenges associated with predictive and prescriptive football analytics require new developments and progress at the intersection of statistical learning, game theory, and computer vision. In this paper, we provide an overarching perspective highlighting how the combination of these fields, in particular, forms a unique microcosm for AI research, while offering mutual benefits for professional teams, spectators, and broadcasters in the years to come. We illustrate that this duality makes football analytics a game changer of tremendous value, in terms of not only changing the game of football itself, but also in terms of what this domain can mean for the field of AI. We review the state-of-the-art and exemplify the types of analysis enabled by combining the aforementioned fields, including illustrative examples of counterfactual analysis using predictive models, and the combination of game-theoretic analysis of penalty kicks with statistical learning of player attributes. We conclude by highlighting envisioned downstream impacts, including possibilities for extensions to other sports (real and virtual).
['Demis Hassabis', 'Thore Graepel', 'Jackson Broshear', 'Nathalie Beauguerlange', 'Simon Bouton', 'Razia Ahamed', 'Trevor Back', 'Remi Munos', 'Andrew Jaegle', 'Mark Rowland', 'Ali Eslami', 'Bart De Vylder', 'Julien Perolat', 'Alex Bridgland', 'Nicolas Heess', 'Michal Valko', 'Praneet Dutta', 'Marta Garnelo', 'Kris Cao', 'Pol Moreno', 'Pablo Sprechmann', 'Romuald Elie', 'Gregory Thornton', 'Alexandre Galashov', 'Adria Recasens', 'Pauline Luc', 'Dafydd Steele', 'Tim Waskett', 'William Spearman', 'Ian Graham', 'Daniel Hennes', 'Jerome Connor', 'Zhe Wang', 'Paul Muller', 'Shayegan Omidshafiei', 'Karl Tuyls']
2020-11-18
null
null
null
null
['game-of-football']
['playing-games']
[ 9.46837142e-02 -4.01947312e-02 -4.97688711e-01 1.67594045e-01 -5.65730035e-01 -5.29934108e-01 3.65139097e-01 5.35428882e-01 -6.55568957e-01 5.95799148e-01 3.97013932e-01 -2.55112857e-01 -7.96109021e-01 -8.51568580e-01 -4.36649382e-01 -5.71812987e-01 -5.37740767e-01 6.62352204e-01 5.37193790e-02 -7.56632090e-01 2.94946790e-01 2.45025381e-01 -1.88987267e+00 3.05363625e-01 5.44995844e-01 9.34740603e-01 -2.47860521e-01 4.50778097e-01 2.43517458e-01 1.03354609e+00 -2.97343433e-01 -9.36389267e-01 4.97719616e-01 -2.55058914e-01 -3.11173379e-01 -1.75598383e-01 -1.16331942e-01 1.65198565e-01 -4.55438137e-01 4.96698499e-01 4.27758992e-01 3.57051134e-01 2.98617989e-01 -1.42021894e+00 -2.30507404e-01 7.63251662e-01 -7.34873056e-01 3.78415585e-01 4.86140192e-01 5.49089849e-01 1.26659143e+00 -6.60536112e-03 3.30471843e-01 9.01330888e-01 7.89006412e-01 2.00256482e-01 -1.09884870e+00 -8.32513809e-01 1.92622185e-01 4.81499732e-01 -8.00625086e-01 -3.97163481e-02 5.73019743e-01 -8.67105722e-01 4.65768337e-01 2.92799950e-01 1.26130569e+00 7.73596585e-01 2.04826176e-01 1.01468611e+00 9.82504964e-01 -5.16953528e-01 3.16091925e-01 -3.46850276e-01 -9.34460759e-03 1.27176628e-01 1.67025656e-01 6.85888648e-01 -8.16694558e-01 6.26503900e-02 5.89990437e-01 6.48850109e-03 3.51297557e-01 -4.06133264e-01 -1.17362690e+00 1.16999364e+00 2.65044812e-02 -2.91783977e-02 -5.70797324e-01 1.51601970e-01 8.73818159e-01 1.30988702e-01 5.29500306e-01 6.70072377e-01 -1.57490358e-01 -9.33125615e-01 -8.94150794e-01 9.69119251e-01 7.51805842e-01 2.30154440e-01 3.10312092e-01 2.42002055e-01 -2.82338914e-02 6.36234820e-01 -9.87664834e-02 2.23972186e-01 3.32356811e-01 -9.36391175e-01 3.34254235e-01 6.06683254e-01 -2.67772675e-01 -1.16576076e+00 -5.17280877e-01 -6.55137122e-01 -5.60316265e-01 3.02468568e-01 7.47660518e-01 -9.94741917e-02 -3.48657727e-01 1.62015069e+00 3.22646171e-01 1.10426277e-01 -3.04124087e-01 9.53549981e-01 4.82830822e-01 1.22404426e-01 2.66257346e-01 -1.69031546e-01 1.20325196e+00 -2.78596222e-01 -4.34368223e-01 -4.19514298e-01 4.62347120e-01 -3.87412935e-01 1.05323601e+00 8.07268560e-01 -1.08562851e+00 -3.48113090e-01 -6.36486650e-01 3.29798430e-01 -5.93652166e-02 -5.20332754e-01 1.20911479e+00 7.20880926e-01 -3.68089199e-01 4.80776280e-01 -7.30222702e-01 -7.08442032e-02 5.39912999e-01 3.32757980e-01 -1.99770972e-01 9.94137228e-02 -1.42843246e+00 7.46419489e-01 4.42495525e-01 -1.81095794e-01 -5.31705558e-01 -1.06947327e+00 -8.77121150e-01 1.03628136e-01 8.92270923e-01 -3.63076121e-01 1.05328751e+00 -7.89069414e-01 -1.23489642e+00 9.69160557e-01 5.66849113e-01 -7.43814945e-01 6.25883698e-01 -2.40366161e-01 -2.52922446e-01 -5.18467844e-01 3.90734106e-01 -5.86033762e-02 -4.22500633e-03 -6.75363898e-01 -1.14032936e+00 -5.91489315e-01 1.65011257e-01 2.74730265e-01 8.16911235e-02 3.25426549e-01 -3.63680273e-02 -7.03629613e-01 -9.03023556e-02 -9.32538033e-01 -5.97729087e-01 -6.76762998e-01 2.44710296e-02 -1.85784534e-01 1.28879566e-02 -3.13386381e-01 1.55690527e+00 -2.09199142e+00 2.36276090e-01 3.22013348e-01 2.57939219e-01 3.34268250e-02 3.92782807e-01 7.86544979e-01 -8.83859172e-02 -1.43852979e-01 1.29675359e-01 2.08383858e-01 -1.85760632e-02 8.53432491e-02 -3.23650062e-01 3.39047700e-01 -3.57139945e-01 9.20221329e-01 -1.07137573e+00 -2.31889576e-01 4.44747835e-01 -2.45684564e-01 -7.04922676e-01 -3.47346306e-01 -1.20329909e-01 2.97224134e-01 -2.58010149e-01 5.03136516e-01 2.21552346e-02 4.31122333e-01 4.13253218e-01 4.28158373e-01 -4.76262301e-01 6.69092312e-02 -1.28411233e+00 1.39118588e+00 -4.03102301e-02 6.92424536e-01 -5.02609313e-02 -1.41630328e+00 8.22206855e-01 5.08828973e-03 8.44612420e-01 -7.19722986e-01 2.48108953e-01 4.05378491e-02 5.10648727e-01 -6.24399841e-01 5.18226445e-01 -6.81437671e-01 -4.72498894e-01 5.13794482e-01 -1.35701984e-01 4.94181775e-02 6.39393210e-01 8.34803842e-03 8.78392816e-01 1.54054716e-01 4.21073347e-01 -6.13960400e-02 -1.03160009e-01 3.68409276e-01 7.97435403e-01 8.58240545e-01 -3.12724292e-01 9.46271867e-02 7.19702840e-01 -7.17163622e-01 -8.91180634e-01 -1.16206241e+00 1.86651364e-01 1.59003556e+00 -1.22446038e-01 -4.07159209e-01 -3.18184853e-01 -5.96789680e-02 4.74815309e-01 4.64646846e-01 -7.17403471e-01 -3.57711524e-01 -5.23649693e-01 -8.75472486e-01 5.99911988e-01 4.34368402e-01 1.57752126e-01 -9.28294957e-01 -6.57888770e-01 3.07377458e-01 -2.43836060e-01 -9.95371282e-01 1.19387969e-01 -1.17948115e-01 -7.93329179e-01 -1.18739891e+00 -4.03242335e-02 -2.45236591e-01 -5.44031143e-01 8.64786357e-02 1.23119771e+00 -2.42292374e-01 -4.13216472e-01 3.87551367e-01 -5.24042130e-01 -1.09100437e+00 -3.52040082e-01 -6.02651993e-03 4.02475804e-01 6.14870824e-02 5.81823528e-01 -8.25884104e-01 -2.67696649e-01 1.67796761e-01 -6.42322123e-01 9.48177874e-02 5.89363515e-01 6.26177609e-01 2.89044052e-01 2.06988230e-02 6.81789637e-01 -6.92572355e-01 7.77083576e-01 -8.51939797e-01 -3.46283168e-01 -4.70267572e-02 -4.17461902e-01 -3.13675374e-01 2.60301620e-01 -4.71821696e-01 -7.22361267e-01 -4.33520406e-01 1.29094943e-01 -6.97947294e-02 -1.80061877e-01 9.97203350e-01 1.33405805e-01 2.92234957e-01 1.04880118e+00 5.14822751e-02 5.38578510e-01 -4.04741257e-01 3.70889217e-01 6.50131822e-01 5.96370161e-01 -7.71683514e-01 5.88176966e-01 4.60337818e-01 2.45266363e-01 -7.19174385e-01 -6.95197761e-01 -4.37054604e-01 -4.78130281e-01 -8.93267810e-01 7.69820213e-01 -8.04681599e-01 -1.47005701e+00 3.23252052e-01 -2.66746461e-01 -2.41018981e-01 -6.99791551e-01 9.56313491e-01 -8.19507122e-01 5.10036647e-02 -4.00297731e-01 -1.01709211e+00 2.26578578e-01 -9.30212200e-01 4.00306404e-01 2.69617647e-01 -3.14342797e-01 -9.04399216e-01 3.93420368e-01 9.66834009e-01 6.89285574e-03 6.83341801e-01 6.94901645e-01 -8.13035667e-01 -2.11793110e-01 -6.67883277e-01 4.24464673e-01 8.04553255e-02 -1.90224156e-01 -1.15942627e-01 -6.18265748e-01 6.44692257e-02 -3.34902138e-01 -2.30171606e-01 6.24168396e-01 8.35796237e-01 7.02874899e-01 1.71847895e-01 -9.32211429e-02 3.58208627e-01 1.02409852e+00 2.84323633e-01 3.87575924e-01 8.24952364e-01 3.13102156e-01 8.53284001e-01 1.12951684e+00 7.42061615e-01 4.67271566e-01 8.75215471e-01 4.79550838e-01 1.63923010e-01 2.93571025e-01 -3.93352777e-01 1.81147084e-01 3.19426924e-01 -8.83205175e-01 2.46858463e-01 -1.10610044e+00 5.81255674e-01 -2.16767597e+00 -1.57073939e+00 -2.35734865e-01 2.55203605e+00 5.12057364e-01 4.05637711e-01 1.03912270e+00 3.77391219e-01 6.01996183e-01 8.05568881e-03 -2.40799963e-01 -7.70545423e-01 -2.09144950e-02 2.60089099e-01 4.66283679e-01 1.03129782e-01 -1.09976017e+00 8.66890371e-01 6.49521589e+00 1.18738401e+00 -8.27836752e-01 -1.28677726e-01 5.48309445e-01 -7.50573397e-01 -2.96449438e-02 9.26575363e-02 -2.72752494e-01 5.04931986e-01 6.78345323e-01 -4.01018977e-01 5.86382866e-01 7.90670097e-01 5.91333628e-01 -3.77973437e-01 -8.22641850e-01 8.21949482e-01 -1.41122058e-01 -1.57171595e+00 -3.08732480e-01 5.23606598e-01 4.14989918e-01 1.41355479e-02 1.45198211e-01 5.82266569e-01 6.05563045e-01 -9.97940481e-01 9.15853858e-01 4.66784477e-01 4.33659375e-01 -9.58808601e-01 6.44603789e-01 5.56414306e-01 -8.18194807e-01 -5.96579969e-01 -3.56626138e-02 -1.35067403e+00 9.87157822e-02 5.71629524e-01 -4.20916408e-01 6.05249882e-01 9.24585044e-01 6.02966130e-01 1.95772517e-02 1.23642278e+00 3.10615543e-02 8.56231511e-01 -2.43621677e-01 -2.36248281e-02 2.02510789e-01 -3.25155169e-01 7.81356096e-01 7.88986564e-01 -1.82339609e-01 3.14281136e-01 3.58864278e-01 5.53313792e-01 4.68788356e-01 1.92780271e-01 -5.56767762e-01 -1.18102878e-01 4.44983214e-01 1.00681496e+00 -6.26615167e-01 1.45789236e-01 -4.15594131e-01 -1.29024982e-01 1.37659803e-01 -7.00319558e-02 -7.26833820e-01 2.30452549e-02 1.18321431e+00 4.97438967e-01 -3.16162258e-01 -2.14737803e-01 -7.39434004e-01 -8.61083388e-01 -2.12367848e-01 -1.21615469e+00 7.37837136e-01 -9.10311416e-02 -1.18091869e+00 -2.14143038e-01 2.35459551e-01 -1.22943985e+00 -4.60010082e-01 -4.54215229e-01 -5.49567044e-01 3.97484332e-01 -5.81006467e-01 -1.15031171e+00 6.52990043e-02 1.96319550e-01 2.95301050e-01 -4.06946898e-01 3.77280682e-01 2.07686067e-01 -4.63726103e-01 4.35577303e-01 2.90919095e-01 2.39814505e-01 3.27052891e-01 -9.84522700e-01 2.18273386e-01 5.17639458e-01 3.30414891e-01 9.75344144e-03 8.64280403e-01 -5.26211262e-01 -1.23057795e+00 -3.12498301e-01 2.35880807e-01 -5.98936498e-01 9.08117890e-01 -3.87804925e-01 -3.23583871e-01 7.30552018e-01 -4.73197520e-01 -6.72046483e-01 1.28840756e+00 9.95160222e-01 5.92533350e-02 -2.44579107e-01 -6.32495224e-01 8.11842978e-01 8.16807568e-01 -2.20189542e-01 -5.60168684e-01 1.47598898e-02 1.80935770e-01 -4.42124724e-01 -9.66502488e-01 3.44449550e-01 9.93455708e-01 -1.26348960e+00 1.10433388e+00 -1.20067489e+00 6.32121086e-01 1.13992840e-01 8.83354545e-02 -1.13191748e+00 -4.49167699e-01 -5.69668293e-01 4.05096501e-01 9.35660005e-01 1.18512496e-01 -2.92150557e-01 1.17202377e+00 1.00753546e+00 -2.86262155e-01 -1.00662959e+00 -1.02227175e+00 -7.30447650e-01 3.52961689e-01 -1.21683657e+00 3.74693245e-01 9.43508685e-01 7.32258797e-01 9.70255509e-02 -6.20770812e-01 -5.02512991e-01 6.58472776e-01 1.17990851e-01 1.17006946e+00 -1.67614079e+00 -5.48393905e-01 -7.36773789e-01 -1.24435019e+00 -4.21649098e-01 -2.17811331e-01 -8.60175371e-01 -3.91274333e-01 -1.04306388e+00 3.14786464e-01 -4.21633542e-01 -1.46073818e-01 2.73759305e-01 -2.26371005e-01 3.92550737e-01 5.98264694e-01 1.61284924e-01 -4.39266354e-01 6.64828420e-02 1.00708318e+00 2.99441576e-01 -4.61395860e-01 4.13187921e-01 -1.03320718e+00 8.36409807e-01 6.68267190e-01 -2.52019912e-01 -1.10904105e-01 9.84598398e-02 8.93821180e-01 -3.48650143e-02 4.79505211e-01 -1.10028386e+00 1.36422500e-01 -7.19204843e-01 -1.13600716e-01 1.11573197e-01 2.06230029e-01 -3.70459646e-01 1.76865682e-01 3.92812848e-01 -3.96934628e-01 -2.54599094e-01 3.04951161e-01 6.14015162e-01 -4.01846498e-01 1.09354861e-01 4.82347518e-01 -1.50366485e-01 -8.34227622e-01 1.41371205e-01 -6.77682519e-01 4.68471974e-01 1.46866298e+00 -6.86075568e-01 8.39179754e-02 -7.56028771e-01 -1.05335343e+00 4.51212376e-01 7.31966645e-02 3.94944698e-01 1.90933824e-01 -9.47327256e-01 -1.04928792e+00 5.53619266e-02 2.24285163e-02 -3.77392411e-01 7.57879436e-01 1.11071670e+00 -4.10085857e-01 1.59617260e-01 -5.18473864e-01 -3.41188788e-01 -1.11363590e+00 5.82319498e-01 6.09666556e-02 -4.37626660e-01 -4.24966425e-01 6.71843469e-01 1.30091578e-01 -2.95891106e-01 5.67080304e-02 1.24999620e-01 -2.55878359e-01 2.42427856e-01 3.13858122e-01 7.82762647e-01 -2.10524321e-01 -4.12569940e-01 -2.29387507e-01 1.64392248e-01 1.80486277e-01 7.08535826e-03 1.43494177e+00 1.37639001e-01 5.49550392e-02 7.20941246e-01 1.77563712e-01 9.19738486e-02 -9.29426670e-01 -1.43149316e-01 7.89684281e-02 -5.81759512e-01 -5.16774878e-02 -7.22190320e-01 -9.25097346e-01 7.21662581e-01 1.94749624e-01 4.25345033e-01 1.04528892e+00 1.72866434e-01 2.83371240e-01 -1.46761000e-01 5.74073017e-01 -1.32107675e+00 -1.94630459e-01 2.88790017e-01 4.73174483e-01 -1.16137981e+00 1.86915204e-01 -2.00408116e-01 -1.07047081e+00 9.11663830e-01 3.15086514e-01 -2.37553611e-01 6.56329632e-01 2.07912803e-01 -9.82104018e-02 -2.67109513e-01 -6.73865914e-01 -5.12178838e-01 1.21848740e-01 7.40489662e-01 4.95434463e-01 4.63893175e-01 -6.76412225e-01 1.13823903e+00 -7.48681307e-01 2.14915499e-01 5.51806033e-01 7.14485705e-01 -4.88429159e-01 -1.06748438e+00 -5.65625906e-01 9.14207757e-01 -7.05640793e-01 -8.48872662e-02 -2.76397198e-01 1.24628377e+00 4.07368273e-01 1.02597392e+00 2.43875220e-01 -7.77588427e-01 5.69953561e-01 -9.84970704e-02 2.56923199e-01 -4.54587549e-01 -8.73882234e-01 -1.31379068e-01 2.27458701e-01 -5.22391856e-01 -2.26382315e-01 -1.03509843e+00 -7.72417784e-01 -9.29232478e-01 -2.68321902e-01 3.81505132e-01 4.87183630e-01 1.08107257e+00 1.16151176e-01 3.56394231e-01 2.73043931e-01 -8.01372349e-01 -3.26548964e-01 -7.65935481e-01 -9.68317389e-01 2.12408289e-01 -3.25513244e-01 -1.05048156e+00 -1.21618904e-01 -3.25175077e-01]
[6.585975646972656, 0.3760640621185303]
e0106137-0c8c-43d5-926d-f40213f6b91b
on-the-sample-complexity-of-vanilla-model
2303.04268
null
https://arxiv.org/abs/2303.04268v1
https://arxiv.org/pdf/2303.04268v1.pdf
On the Sample Complexity of Vanilla Model-Based Offline Reinforcement Learning with Dependent Samples
Offline reinforcement learning (offline RL) considers problems where learning is performed using only previously collected samples and is helpful for the settings in which collecting new data is costly or risky. In model-based offline RL, the learner performs estimation (or optimization) using a model constructed according to the empirical transition frequencies. We analyze the sample complexity of vanilla model-based offline RL with dependent samples in the infinite-horizon discounted-reward setting. In our setting, the samples obey the dynamics of the Markov decision process and, consequently, may have interdependencies. Under no assumption of independent samples, we provide a high-probability, polynomial sample complexity bound for vanilla model-based off-policy evaluation that requires partial or uniform coverage. We extend this result to the off-policy optimization under uniform coverage. As a comparison to the model-based approach, we analyze the sample complexity of off-policy evaluation with vanilla importance sampling in the infinite-horizon setting. Finally, we provide an estimator that outperforms the sample-mean estimator for almost deterministic dynamics that are prevalent in reinforcement learning.
['Ufuk Topcu', 'Mustafa O. Karabag']
2023-03-07
null
null
null
null
['offline-rl']
['playing-games']
[-2.94026639e-02 4.34508651e-01 -8.51766706e-01 -7.83703849e-03 -1.28207898e+00 -6.04269624e-01 3.57231677e-01 3.57311398e-01 -9.08217669e-01 1.38793385e+00 -1.65966094e-01 -6.62923336e-01 -2.26572409e-01 -7.78957665e-01 -1.09404922e+00 -6.82775497e-01 -5.20783544e-01 6.74798906e-01 -4.55295891e-02 9.70708430e-02 1.19997986e-01 2.03493938e-01 -1.33575547e+00 -3.97013396e-01 8.08095932e-01 1.02144623e+00 2.40943626e-01 1.17077720e+00 1.55933812e-01 9.22020078e-01 -7.02632964e-01 -5.36891297e-02 4.38290477e-01 -4.98853296e-01 -4.88232583e-01 2.63990220e-02 -1.33102581e-01 -9.34466124e-01 -9.83868912e-02 1.15095031e+00 3.93367022e-01 3.77519310e-01 5.38337648e-01 -1.12780476e+00 2.71666199e-02 7.59188890e-01 -3.60548824e-01 1.49573788e-01 3.37737173e-01 3.66605669e-01 8.96353245e-01 -1.90015867e-01 3.83785963e-01 1.34130609e+00 3.56634468e-01 6.22749031e-01 -1.25324655e+00 -3.65293980e-01 6.00084603e-01 2.40375921e-01 -6.22583091e-01 -2.42686674e-01 2.61185706e-01 -2.53939003e-01 6.74434423e-01 -8.94822702e-02 1.05237985e+00 1.19583142e+00 3.65151614e-01 1.25462830e+00 1.34109807e+00 -5.12745619e-01 9.80888546e-01 1.47471696e-01 -6.83039799e-02 5.15908420e-01 4.15205777e-01 8.13589931e-01 -2.13250473e-01 -3.72211546e-01 6.92490339e-01 8.59641507e-02 1.25220064e-02 -4.56778735e-01 -7.77335048e-01 1.11980939e+00 -3.45690340e-01 -3.57777536e-01 -7.25751281e-01 6.38667166e-01 3.60266268e-01 7.10291803e-01 2.85532236e-01 2.97080666e-01 -5.56489348e-01 -7.05841064e-01 -9.44367468e-01 5.41861832e-01 1.24072874e+00 1.10812199e+00 5.11696339e-01 3.76307905e-01 -5.81209540e-01 2.59579927e-01 1.28700227e-01 7.23651350e-01 2.76787817e-01 -1.30922270e+00 5.75670242e-01 -3.82115811e-01 1.09568655e+00 6.19301498e-02 -1.71012685e-01 -5.59535921e-01 -1.52379856e-01 4.47885334e-01 7.26170480e-01 -6.73624337e-01 -4.95654106e-01 1.84808302e+00 5.56193352e-01 -8.39713737e-02 2.54160345e-01 4.32827115e-01 -4.51578170e-01 4.93752539e-01 -2.06231833e-01 -1.00727332e+00 8.15528512e-01 -8.97500038e-01 -9.21348035e-01 1.56413093e-01 5.57190180e-01 -1.61895722e-01 1.11486089e+00 9.84042585e-01 -1.20096040e+00 -3.27836312e-02 -1.09190500e+00 5.99305511e-01 2.41461739e-01 -1.64387152e-01 3.92389119e-01 7.06080556e-01 -8.07009816e-01 8.54857385e-01 -9.70720291e-01 3.99823263e-02 4.58639324e-01 7.41155446e-02 4.18318778e-01 -6.34063035e-02 -1.04596496e+00 8.55078936e-01 3.75068516e-01 -2.34154999e-01 -1.99655366e+00 -4.41525847e-01 -5.45010388e-01 -2.42421776e-02 1.27093196e+00 -1.71624780e-01 2.16909671e+00 -1.06924307e+00 -2.03850055e+00 -2.04873785e-01 -1.42280404e-02 -1.01616597e+00 1.08637929e+00 -2.78627634e-01 1.87358662e-01 2.37366423e-01 -1.32461175e-01 -5.97820096e-02 1.05966508e+00 -1.10660255e+00 -9.26735759e-01 -5.30666709e-02 5.35853744e-01 2.91861445e-01 9.85808894e-02 -5.90050399e-01 2.85821348e-01 -2.57462174e-01 -8.40442717e-01 -9.48668242e-01 -6.47425234e-01 -2.95603275e-01 -1.29687175e-01 -2.51972854e-01 3.68207693e-01 -5.51219165e-01 1.24013197e+00 -1.74943495e+00 -3.23038608e-01 1.95570856e-01 -2.62902707e-01 -1.74594253e-01 2.46820226e-02 7.06716657e-01 4.75094676e-01 -5.51596954e-02 -1.54408395e-01 -1.74415678e-01 2.72862554e-01 4.49665189e-01 -5.12800753e-01 5.98161399e-01 -3.65791678e-01 6.20268881e-01 -1.28098893e+00 -4.67270315e-01 4.43572253e-02 -4.59388345e-01 -7.24796891e-01 3.86934966e-01 -8.19209874e-01 2.20328018e-01 -6.57212198e-01 5.05487323e-01 1.97996393e-01 -5.19052893e-02 4.54778165e-01 6.29810393e-01 -5.50628826e-02 -8.95650387e-02 -1.21175468e+00 1.29357123e+00 -9.08343911e-01 2.94152737e-01 2.48594344e-01 -1.10828733e+00 3.03159803e-01 3.03014904e-01 5.30719221e-01 -2.62808412e-01 -2.07546894e-02 7.79913813e-02 -1.34824798e-01 -4.05785054e-01 5.98493397e-01 -3.81903291e-01 -1.59451976e-01 7.94148922e-01 2.01605767e-01 -2.83896253e-02 2.36204997e-01 6.61816774e-03 1.22366071e+00 4.43634272e-01 8.06542575e-01 -2.22187340e-01 -1.01038270e-01 -8.83390978e-02 3.79006386e-01 1.40281403e+00 -4.20185655e-01 -3.24360013e-01 9.24322009e-01 3.82326357e-02 -7.84900427e-01 -1.03628206e+00 -6.42452715e-03 9.18923914e-01 -1.58142954e-01 -1.37582168e-01 -7.20854163e-01 -9.09409583e-01 2.60666013e-01 1.12712908e+00 -7.68378913e-01 -4.55919802e-02 -6.29939809e-02 -4.08960313e-01 1.82044711e-02 1.79820195e-01 1.28847823e-01 -8.48431349e-01 -1.10734594e+00 7.62429655e-01 1.21587865e-01 -7.24900365e-01 -4.46035564e-01 4.06248242e-01 -1.02896273e+00 -1.10664642e+00 -8.11880112e-01 9.13096517e-02 3.96501064e-01 -1.61551356e-01 9.39190269e-01 -4.67790008e-01 2.00632155e-01 1.26151872e+00 -1.74891666e-01 -7.88681030e-01 -4.26195323e-01 -1.37275741e-01 1.07499838e-01 -5.38323559e-02 -1.10571057e-01 -2.52955914e-01 -5.69659412e-01 -1.55507132e-01 -7.83873379e-01 -3.81759137e-01 3.42164487e-01 1.15049076e+00 6.92168057e-01 1.16450177e-03 8.69578123e-01 -9.05293941e-01 9.48892474e-01 -6.74883604e-01 -1.32146430e+00 5.47526002e-01 -9.00371373e-01 4.65446055e-01 9.57834601e-01 -8.14580441e-01 -9.88113403e-01 -1.71042189e-01 3.43078285e-01 -4.40193057e-01 2.90019065e-01 3.35077733e-01 9.75366905e-02 2.34133810e-01 4.58534420e-01 3.38469535e-01 2.60048747e-01 -3.11661899e-01 2.62579292e-01 4.34675604e-01 -1.06877968e-01 -1.17333937e+00 3.68185639e-01 2.31303662e-01 1.81069717e-01 -7.10915267e-01 -1.03181934e+00 -3.85536999e-02 3.47720683e-02 -5.00544965e-01 4.35528547e-01 -8.44473302e-01 -1.23467636e+00 1.43098328e-02 -7.61874855e-01 -1.06658125e+00 -1.24156725e+00 7.44276345e-01 -1.44738984e+00 2.62578696e-01 -3.33501518e-01 -1.99294972e+00 9.08062309e-02 -1.11582041e+00 5.87189555e-01 2.62247145e-01 2.79307663e-01 -1.10618055e+00 2.45405376e-01 -2.86397934e-01 1.93608314e-01 2.53151715e-01 6.39842093e-01 -5.74567258e-01 -6.62916481e-01 -3.86709124e-02 5.47198653e-01 3.66422445e-01 -2.52784580e-01 -2.35004857e-01 -6.73287690e-01 -5.51491082e-01 2.62458511e-02 -7.65701354e-01 6.79939926e-01 7.80350149e-01 1.42680621e+00 -9.63965893e-01 7.34878927e-02 4.85360287e-02 1.60632122e+00 5.15519798e-01 9.36102942e-02 3.71820539e-01 -1.18054651e-01 5.13381720e-01 1.32010508e+00 1.47851062e+00 2.19965532e-01 3.47740293e-01 6.10922635e-01 6.10449433e-01 7.77933419e-01 -7.17280924e-01 8.48404586e-01 1.83410585e-01 -3.39020863e-02 -1.55601472e-01 -2.38894925e-01 5.25525093e-01 -1.99980283e+00 -1.22299719e+00 5.71304679e-01 3.00597668e+00 1.27270567e+00 2.19758645e-01 7.52659738e-01 -2.55873233e-01 4.52792048e-01 -7.26536140e-02 -1.33677793e+00 -6.49826705e-01 3.22692901e-01 3.99739772e-01 1.11665511e+00 8.87505233e-01 -6.56106532e-01 5.85141659e-01 6.65657949e+00 1.09460521e+00 -5.73789001e-01 4.06245142e-01 6.94799185e-01 -5.51541686e-01 -3.21680158e-01 1.77016959e-01 -1.00105929e+00 5.20161271e-01 1.52876568e+00 -4.73100662e-01 9.53429163e-01 1.26198804e+00 4.17427391e-01 -6.93960488e-01 -1.34900534e+00 5.90916395e-01 -6.87052011e-01 -9.43005919e-01 -5.50832272e-01 4.92818266e-01 8.09620023e-01 -1.62581220e-01 -2.64820773e-02 8.44197094e-01 1.13595998e+00 -6.45487428e-01 9.92153704e-01 7.91792512e-01 8.64836156e-01 -1.18855190e+00 3.89319330e-01 8.70813727e-01 -8.41725647e-01 -5.46912372e-01 -4.52419341e-01 -3.85875143e-02 1.47296518e-01 5.63943624e-01 -8.51320624e-01 2.47078493e-01 1.52234122e-01 4.19765294e-01 2.50844955e-01 8.96829367e-01 -2.73208678e-01 1.03746283e+00 -4.31288362e-01 -5.41866899e-01 4.36635762e-01 -4.60045666e-01 5.80991328e-01 8.28256488e-01 4.24976051e-01 -2.29051679e-01 5.03436685e-01 5.89150906e-01 1.75234586e-01 -6.51285276e-02 -5.95444918e-01 -1.43914148e-01 5.55019379e-01 8.07886660e-01 -2.03192815e-01 -5.17235935e-01 -7.59544745e-02 5.11694193e-01 3.66346896e-01 6.04294240e-01 -8.49259198e-01 -1.24828078e-01 4.30523723e-01 -1.89451978e-01 5.32214820e-01 -1.83310375e-01 3.13101619e-01 -9.64505315e-01 -1.17238499e-01 -9.18173492e-01 4.33485150e-01 2.21489556e-02 -9.89571989e-01 -2.14976668e-01 4.92049187e-01 -1.10539651e+00 -8.92708898e-01 -3.05592626e-01 -3.24654639e-01 5.27566016e-01 -1.59037340e+00 -1.28723204e-01 5.75895369e-01 4.79737818e-01 7.54544854e-01 5.87577745e-02 6.00745618e-01 -3.59308630e-01 -3.97624016e-01 5.48150420e-01 7.31504500e-01 -5.45383096e-01 1.95023179e-01 -1.58020985e+00 -2.32208818e-01 5.31423092e-01 -2.77514726e-01 2.52075829e-02 8.88640702e-01 -5.04910111e-01 -1.54011953e+00 -1.06181693e+00 -5.20270653e-02 3.71573344e-02 7.11996317e-01 -2.15900615e-01 -4.35417563e-01 6.77024484e-01 1.41124595e-02 -6.67741895e-03 4.82612193e-01 1.23599293e-02 2.68293291e-01 -1.76467687e-01 -1.50893760e+00 6.24382973e-01 7.13001490e-01 -1.57124057e-01 -2.73809195e-01 4.10202652e-01 7.84266651e-01 -2.40519181e-01 -9.32386220e-01 -1.56721249e-01 7.63992071e-01 -6.31498814e-01 5.09828627e-01 -9.15039599e-01 9.96893086e-03 3.06484520e-01 -1.57177448e-01 -1.49191415e+00 2.27159768e-01 -1.43801081e+00 -8.62296522e-01 5.99333823e-01 3.24347675e-01 -1.00633717e+00 6.37543440e-01 4.00208950e-01 3.01365525e-01 -1.02638495e+00 -1.21055102e+00 -1.30509663e+00 2.34457955e-01 -5.40354192e-01 3.99147213e-01 1.54506981e-01 9.01148021e-02 -1.48503110e-01 -5.70599198e-01 -1.59967333e-01 1.18863666e+00 8.27300176e-02 7.18450844e-01 -6.72559321e-01 -1.06648326e+00 -1.65226206e-01 4.04475212e-01 -1.37645328e+00 3.03570598e-01 -2.11481661e-01 3.30501586e-01 -1.15332830e+00 1.25869513e-01 -4.55095142e-01 -3.20821106e-01 3.16987112e-02 2.48452015e-02 -9.46915865e-01 3.48931611e-01 -9.95637402e-02 -9.98733044e-01 1.02989292e+00 1.47254014e+00 2.26507425e-01 -2.40300223e-01 6.51189566e-01 -4.11039263e-01 5.93355656e-01 8.72997046e-01 -5.55188954e-01 -7.62265384e-01 4.46141899e-01 1.40913039e-01 1.04576635e+00 -6.91739097e-02 -8.03205073e-01 -1.80087611e-01 -8.71315658e-01 -6.10279851e-02 -6.56817615e-01 2.35739067e-01 -7.06405401e-01 -2.35129222e-01 1.00618672e+00 -9.25284207e-01 -5.55007942e-02 -1.98628321e-01 1.36247706e+00 3.50310087e-01 -7.40307212e-01 7.64592767e-01 -4.38237339e-01 -1.32726297e-01 6.36485875e-01 -8.11307132e-01 5.56710780e-01 1.20557284e+00 9.52298492e-02 1.65236682e-01 -1.01081455e+00 -8.40252280e-01 5.28961182e-01 1.04065776e-01 -4.17727113e-01 5.91654301e-01 -9.89340901e-01 -3.66889358e-01 -2.56348848e-01 -2.30469108e-01 -8.53740126e-02 5.85656948e-02 8.01686347e-01 7.01274276e-02 2.52774775e-01 1.71015188e-01 -2.48014614e-01 -7.03993022e-01 7.06939042e-01 4.37474132e-01 -9.66593504e-01 -1.68236151e-01 3.45963508e-01 -1.10809639e-01 -9.59481150e-02 5.56753814e-01 -6.78199410e-01 -1.80736009e-03 7.05401897e-02 5.61525822e-01 7.05838799e-01 -3.36614162e-01 4.17514920e-01 3.82151812e-01 -2.35071912e-01 1.08540267e-01 -9.67256069e-01 1.19028032e+00 -2.35215589e-01 6.67523921e-01 1.15631580e+00 6.65685475e-01 -1.36275724e-01 -2.10963058e+00 -3.35445344e-01 2.22079102e-02 -4.10977066e-01 2.54065986e-03 -7.08603978e-01 -5.62967360e-01 8.47181916e-01 5.54502726e-01 2.97733396e-01 7.16266572e-01 -5.04823029e-01 4.22483772e-01 6.37408018e-01 1.01101398e+00 -1.66586220e+00 -7.32258111e-02 5.19054413e-01 6.69802427e-01 -1.06691122e+00 1.55806616e-01 6.44385159e-01 -6.99800253e-01 1.10976171e+00 3.63537908e-01 -3.34085137e-01 7.22874165e-01 1.95780322e-01 -6.67881727e-01 4.66529965e-01 -1.39509511e+00 -4.59590048e-01 -2.88366318e-01 6.75203264e-01 -1.64499938e-01 5.91760039e-01 -4.22782242e-01 5.42973876e-01 -2.76092114e-03 2.73675233e-01 8.79423022e-01 1.14122331e+00 -8.11606705e-01 -1.06699753e+00 -3.02951515e-01 7.64878392e-01 -6.36196733e-01 5.07042445e-02 1.55679241e-01 8.18012595e-01 -5.79448164e-01 1.07006657e+00 -1.01408452e-01 1.47222281e-01 7.03703612e-02 -1.11576281e-02 9.27447677e-01 -4.47231948e-01 -3.34505558e-01 1.53671131e-01 2.65584350e-01 -7.83126593e-01 -1.93388000e-01 -8.84631574e-01 -1.05558431e+00 -1.81400284e-01 -3.06560785e-01 3.55611116e-01 7.25993752e-01 1.05074787e+00 -3.85001511e-03 3.39819908e-01 1.17441237e+00 -6.72213912e-01 -2.00761008e+00 -1.02401590e+00 -1.00125468e+00 -3.72130990e-01 7.31145203e-01 -7.85125434e-01 -6.07984245e-01 -5.01317859e-01]
[4.28427267074585, 2.6892616748809814]
8bf81ae6-4e90-4d47-b925-86f0c22fa7b5
interpretable-amr-based-question
2206.08486
null
https://arxiv.org/abs/2206.08486v1
https://arxiv.org/pdf/2206.08486v1.pdf
Interpretable AMR-Based Question Decomposition for Multi-hop Question Answering
Effective multi-hop question answering (QA) requires reasoning over multiple scattered paragraphs and providing explanations for answers. Most existing approaches cannot provide an interpretable reasoning process to illustrate how these models arrive at an answer. In this paper, we propose a Question Decomposition method based on Abstract Meaning Representation (QDAMR) for multi-hop QA, which achieves interpretable reasoning by decomposing a multi-hop question into simpler sub-questions and answering them in order. Since annotating the decomposition is expensive, we first delegate the complexity of understanding the multi-hop question to an AMR parser. We then achieve the decomposition of a multi-hop question via segmentation of the corresponding AMR graph based on the required reasoning type. Finally, we generate sub-questions using an AMR-to-Text generation model and answer them with an off-the-shelf QA model. Experimental results on HotpotQA demonstrate that our approach is competitive for interpretable reasoning and that the sub-questions generated by QDAMR are well-formed, outperforming existing question-decomposition-based multi-hop QA approaches.
['Patricia Riddle', 'Michael Witbrock', 'Yang Chen', 'Yonghua Zhu', 'Zhenyun Deng']
2022-06-16
null
null
null
null
['multi-hop-question-answering']
['knowledge-base']
[ 1.85395315e-01 1.23674166e+00 2.80871063e-01 -5.82215607e-01 -1.71125340e+00 -1.01703358e+00 2.64120936e-01 3.11350375e-01 2.44176894e-01 6.59043610e-01 4.39121664e-01 -1.12820375e+00 -2.51471132e-01 -1.17984855e+00 -7.61278868e-01 2.63497084e-01 4.74092185e-01 1.14481068e+00 5.76829374e-01 -7.16373146e-01 1.01248868e-01 -3.36995795e-02 -1.13734365e+00 1.04180288e+00 1.43806350e+00 5.64826846e-01 -3.33868824e-02 1.24008811e+00 -9.21867847e-01 1.31799424e+00 -8.50320220e-01 -1.03360415e+00 -9.30853486e-02 -9.76292253e-01 -2.03718734e+00 9.39867795e-02 4.11945194e-01 -4.06774729e-01 8.63333344e-02 6.60973489e-01 -3.11407298e-01 -1.06724156e-02 4.27862704e-01 -1.10337591e+00 -9.05782044e-01 1.08599579e+00 -6.51369989e-02 -7.85457268e-02 1.11951911e+00 2.44419798e-02 1.66957378e+00 -4.70099300e-01 5.54091394e-01 1.71931493e+00 4.03277308e-01 7.36195624e-01 -9.14670289e-01 3.04241836e-01 2.81998336e-01 4.67961520e-01 -6.59806967e-01 3.89780826e-03 6.18493974e-01 -6.34013191e-02 1.10193980e+00 8.70748580e-01 4.24314082e-01 4.12704021e-01 6.49686232e-02 8.15329790e-01 9.84822571e-01 -6.43747270e-01 3.13394219e-01 -2.09554866e-01 8.92214119e-01 1.06404936e+00 1.65184334e-01 -1.04213536e+00 -1.87147737e-01 -3.27267379e-01 3.84013891e-01 -4.15296406e-01 -1.63583785e-01 2.94132024e-01 -8.42521727e-01 1.14262176e+00 3.34340692e-01 5.74228540e-02 -3.55196208e-01 1.14058584e-01 5.72168976e-02 5.76535881e-01 1.50960669e-01 8.12767923e-01 -6.50665581e-01 9.77118388e-02 -5.97655952e-01 5.26807964e-01 1.44665051e+00 8.49793315e-01 8.30251992e-01 -5.75601697e-01 -4.11375225e-01 4.85265911e-01 4.85492200e-01 3.61370683e-01 1.20560586e-01 -1.71177137e+00 9.88779724e-01 1.16827142e+00 3.15196812e-01 -7.82108366e-01 -1.61184803e-01 2.52223983e-02 -2.54538894e-01 -2.11876705e-01 6.72844410e-01 -2.39952400e-01 -7.48690605e-01 1.25209749e+00 6.51757598e-01 -4.47748661e-01 6.69641376e-01 8.67417336e-01 1.16542089e+00 9.92488801e-01 6.49890816e-03 9.77787822e-02 2.14137650e+00 -1.45814610e+00 -8.04129779e-01 -5.00679910e-01 7.19171882e-01 -4.24797952e-01 1.44748676e+00 2.56732851e-01 -1.32062829e+00 -4.14459199e-01 -7.25294352e-01 -8.20075572e-01 -9.00364444e-02 -9.85480100e-02 4.15085882e-01 4.11272138e-01 -9.45666909e-01 6.49750084e-02 -4.08636838e-01 -2.13856772e-01 7.46630654e-02 9.33473632e-02 -6.99681565e-02 -5.77226400e-01 -1.43843722e+00 8.35430920e-01 2.91700333e-01 8.98466185e-02 -5.98428011e-01 -6.17092192e-01 -1.06776798e+00 2.16628239e-01 7.76057065e-01 -1.37169182e+00 1.80519009e+00 -4.52947885e-01 -1.60800982e+00 5.09127557e-01 -6.26398146e-01 -5.42773247e-01 1.83043018e-01 -3.30847532e-01 -1.53418824e-01 8.46158087e-01 4.98723418e-01 7.01295316e-01 4.88115013e-01 -1.32897413e+00 -5.64936042e-01 -3.94556880e-01 1.28283632e+00 3.08298200e-01 2.98451811e-01 -9.67270136e-02 -3.92520040e-01 -6.09119870e-02 4.63661134e-01 -6.12382233e-01 -4.02907521e-01 -3.30647409e-01 -5.52609026e-01 -5.12899578e-01 6.21169567e-01 -1.15810585e+00 1.24936235e+00 -1.44108236e+00 1.40489221e-01 -1.38441220e-01 5.65576792e-01 -4.41543907e-02 -5.13794661e-01 7.31465638e-01 1.69171154e-01 3.69883627e-01 -5.44503748e-01 -1.11227594e-02 3.63711238e-01 7.12855339e-01 -7.12244689e-01 -4.88668621e-01 6.19661331e-01 1.30104756e+00 -9.81372714e-01 -7.02176213e-01 -1.01429105e-01 -7.25218952e-02 -8.16937685e-01 5.94630837e-01 -1.14613521e+00 2.86542088e-01 -7.58553147e-01 4.67407227e-01 5.24599493e-01 -5.03423870e-01 2.79287755e-01 -1.11704022e-02 4.07061160e-01 7.66859174e-01 -8.76908600e-01 1.68295038e+00 -8.06802809e-01 3.69441807e-01 -5.84354997e-02 -6.61892056e-01 9.25031781e-01 2.77140915e-01 -3.64796132e-01 -4.05877680e-01 -3.74088496e-01 2.73651063e-01 -5.50686903e-02 -9.05338883e-01 6.99168861e-01 -2.23173574e-01 -4.13150042e-01 7.55905747e-01 -4.57112305e-02 -6.48880363e-01 5.60012698e-01 7.38570094e-01 1.31374693e+00 1.32385969e-01 1.33333400e-01 9.93913338e-02 1.02207470e+00 7.14699090e-01 7.59292394e-02 8.90524864e-01 2.72479177e-01 6.06085360e-01 9.94801164e-01 -5.16646981e-01 -6.22344971e-01 -9.65825200e-01 5.02144277e-01 8.93227518e-01 3.34061459e-02 -8.12247813e-01 -1.26698542e+00 -1.20530152e+00 -4.37839776e-01 1.35786998e+00 -3.86958361e-01 2.98074663e-01 -8.52271736e-01 -1.41388044e-01 6.06893718e-01 3.76858860e-01 5.72623134e-01 -9.88543153e-01 -8.03744078e-01 3.52511913e-01 -1.20891488e+00 -1.30936956e+00 -3.72153446e-02 -1.55833960e-01 -8.50988686e-01 -1.25996816e+00 -3.09589237e-01 -6.32596254e-01 8.53174448e-01 2.01555133e-01 1.54805708e+00 5.44963300e-01 1.63719282e-01 6.66140854e-01 -7.58934796e-01 -1.39875874e-01 -9.09145713e-01 2.76370794e-01 -9.87133801e-01 -2.22120851e-01 2.32826024e-01 -2.55175352e-01 -5.84203660e-01 2.81475158e-03 -1.11205375e+00 3.25989842e-01 5.15031159e-01 3.65376949e-01 5.29194474e-01 1.96466153e-03 5.83087683e-01 -1.27554154e+00 1.15026748e+00 -4.65748280e-01 -2.31544420e-01 9.35122669e-01 -9.45413783e-02 6.75616205e-01 1.06899726e+00 2.01145202e-01 -1.47767723e+00 -3.50214034e-01 -5.28410554e-01 4.56160337e-01 -2.62722582e-01 6.50858700e-01 -3.04515451e-01 4.29202288e-01 6.47496879e-01 4.14599255e-02 -4.02985752e-01 -2.35907853e-01 1.06300139e+00 5.90235531e-01 4.95720744e-01 -1.02672005e+00 9.32387292e-01 2.59721696e-01 -8.37348029e-02 -3.75910372e-01 -1.64011991e+00 -2.66444027e-01 -4.29788738e-01 -2.19158959e-02 1.18806386e+00 -5.11219144e-01 -7.32713938e-01 -4.09089983e-01 -1.82062054e+00 -2.64738739e-01 -5.90292394e-01 -2.09162742e-01 -5.84386766e-01 7.03610957e-01 -7.05760241e-01 -8.41925383e-01 -6.39907897e-01 -9.73575056e-01 1.27320135e+00 2.51125932e-01 -6.36960030e-01 -1.01512158e+00 9.41209048e-02 1.49360216e+00 4.23137061e-02 2.44022682e-02 1.58854306e+00 -7.98626661e-01 -7.78936028e-01 -1.10851154e-01 -2.50992894e-01 6.84479401e-02 -1.57682002e-01 -2.98891813e-01 -7.02185214e-01 5.11177003e-01 1.12602636e-01 -5.41434765e-01 3.83162111e-01 -3.88352275e-02 1.04114878e+00 -9.08559978e-01 2.32237846e-01 -2.04105347e-01 1.22226965e+00 -2.82524168e-01 8.59235168e-01 2.81281114e-01 4.23301905e-01 9.79428351e-01 6.91137493e-01 -1.47646159e-01 1.33054519e+00 1.49515107e-01 4.49542552e-01 1.34715632e-01 -1.39071211e-01 -5.16631782e-01 1.55305460e-01 8.95818949e-01 1.57153323e-01 -2.09309414e-01 -1.02852035e+00 5.61381519e-01 -1.93908179e+00 -7.80116260e-01 -9.78535771e-01 1.45166457e+00 8.18901956e-01 -8.45428482e-02 -1.33927450e-01 1.04830727e-01 2.42716715e-01 3.61908227e-02 -2.67828315e-01 -9.57153320e-01 1.27580166e-01 4.14573848e-01 -3.83379579e-01 1.14343166e+00 -3.00670356e-01 1.03593230e+00 5.86291933e+00 4.62412238e-01 6.26472756e-03 1.35359526e-01 5.08722067e-01 6.71847641e-01 -1.11263621e+00 6.08850181e-01 -6.10377848e-01 -2.73994416e-01 1.14967716e+00 -3.43722776e-02 4.18626875e-01 5.85937381e-01 -2.07240731e-02 -2.19831958e-01 -1.10541284e+00 3.25700670e-01 7.05390647e-02 -1.34435844e+00 8.29865873e-01 -5.17471850e-01 3.33740443e-01 -7.48271644e-01 -4.81308252e-01 5.33591807e-01 5.70422351e-01 -9.48867917e-01 6.04466677e-01 4.83500123e-01 2.54033178e-01 -6.80520117e-01 8.29025626e-01 6.29375160e-01 -1.21725464e+00 -2.37183198e-01 -4.21766073e-01 -2.47779608e-01 3.07819009e-01 2.98935086e-01 -7.42597878e-01 1.25883770e+00 3.47780377e-01 -2.15162873e-01 -7.98715532e-01 4.34864879e-01 -1.05844092e+00 6.63368583e-01 -4.45171446e-02 -2.07196593e-01 3.81323129e-01 -3.87220681e-01 3.41579676e-01 8.56659472e-01 2.24891067e-01 7.56973684e-01 -6.41203448e-02 1.01554906e+00 -1.40951529e-01 -1.54144531e-02 -2.26536416e-03 -2.54769716e-02 1.82056010e-01 1.44523859e+00 -4.92954344e-01 -6.74296737e-01 -4.14466858e-01 1.22912478e+00 5.00739038e-01 3.57113957e-01 -7.06980109e-01 -6.41942263e-01 1.50564164e-01 4.02816720e-02 9.16330442e-02 -6.46888167e-02 -3.08342814e-01 -1.31014550e+00 4.16485906e-01 -1.35944831e+00 9.09339905e-01 -1.52487087e+00 -1.09256041e+00 9.61920917e-01 5.28955199e-02 -5.99668920e-01 -5.09256184e-01 -4.02296722e-01 -7.03101635e-01 9.85513628e-01 -1.76630235e+00 -1.32144165e+00 -3.35019439e-01 3.52917880e-01 8.92854393e-01 6.95943475e-01 1.08696330e+00 -3.33640844e-01 -1.93790674e-01 1.20679401e-01 -9.16094124e-01 2.93683074e-02 6.52377605e-02 -1.75487673e+00 6.15331769e-01 8.61596167e-01 3.63355070e-01 6.54683590e-01 7.91813195e-01 -4.41307366e-01 -1.56710649e+00 -1.01113737e+00 1.35031343e+00 -9.50192332e-01 8.54109228e-01 -1.48815632e-01 -1.25867546e+00 7.86602676e-01 6.35318816e-01 -5.46402454e-01 8.05829704e-01 4.73623574e-02 -4.58715647e-01 5.98654188e-02 -1.04377067e+00 7.57046998e-01 7.22538233e-01 -5.41673958e-01 -1.70710313e+00 6.12370968e-01 1.40588844e+00 -4.33008999e-01 -7.75471032e-01 5.55292182e-02 -1.68148857e-02 -7.76705086e-01 6.27090871e-01 -9.84717011e-01 9.78828788e-01 -6.28578126e-01 -3.06734517e-02 -1.01921642e+00 8.25634524e-02 -7.52247989e-01 -2.77542502e-01 1.06462312e+00 8.85149598e-01 -5.33073366e-01 6.27397299e-01 1.00899291e+00 -2.22375467e-01 -8.49695206e-01 -7.15354800e-01 -2.23193571e-01 1.05900034e-01 -6.31547689e-01 8.96402299e-01 4.31323647e-01 1.81421056e-01 9.89378273e-01 1.86135054e-01 6.31398439e-01 6.21234953e-01 7.07155824e-01 7.59765804e-01 -9.97578919e-01 -6.29928648e-01 1.36136830e-01 3.94607902e-01 -1.52756906e+00 2.39731535e-01 -7.61146724e-01 7.09414259e-02 -2.79919100e+00 -1.63758826e-03 9.29961503e-02 4.98503536e-01 5.37329316e-01 -5.19838750e-01 -2.17270717e-01 1.61673933e-01 8.93522985e-03 -1.11307967e+00 2.51747280e-01 1.56227386e+00 -2.01890081e-01 1.61475927e-01 -8.63440335e-02 -1.07198250e+00 7.35187888e-01 6.51597559e-01 -4.17731702e-01 -7.06573904e-01 -8.14716399e-01 6.18492723e-01 7.73034036e-01 4.02344376e-01 -6.65987968e-01 1.80736199e-01 -1.93583965e-01 -2.38369361e-01 -6.02148235e-01 2.05545679e-01 -5.97962141e-01 -2.64854044e-01 3.72090995e-01 -6.09592319e-01 2.74233729e-01 -1.95108969e-02 3.24082315e-01 -4.35856760e-01 -7.74845481e-01 1.42167270e-01 -4.39027011e-01 -3.32229853e-01 -1.89493820e-01 -6.06173813e-01 6.29793167e-01 6.50872052e-01 8.23184196e-03 -5.64909756e-01 -7.35546768e-01 -7.82164574e-01 6.43365443e-01 -9.02594030e-02 8.30949545e-02 6.78237557e-01 -8.19802105e-01 -7.95627356e-01 -4.09863651e-01 5.95976785e-02 4.24424529e-01 2.88433582e-01 3.14109504e-01 -8.67861629e-01 7.53228009e-01 2.28085101e-01 -1.83605433e-01 -1.17099464e+00 3.13152462e-01 4.29419041e-01 -9.17824268e-01 -3.82823408e-01 4.88977730e-01 6.73617944e-02 -9.09685969e-01 -3.51415008e-01 -7.12551892e-01 -4.55859393e-01 -1.84368759e-01 6.07250333e-01 1.02204815e-01 1.65057667e-02 -1.81066319e-01 -1.29702821e-01 6.28890634e-01 1.25236303e-01 -3.48364532e-01 1.13086379e+00 -3.61553609e-01 -6.50210261e-01 1.65763646e-01 6.54509127e-01 -4.19642404e-02 -6.38908207e-01 -1.73398349e-02 2.02710018e-01 -6.73753545e-02 -6.80681109e-01 -9.73324537e-01 -3.63413632e-01 9.97588575e-01 -4.73872066e-01 8.34980786e-01 1.11218739e+00 5.36483705e-01 1.30429518e+00 9.65741396e-01 2.21872658e-01 -7.00873852e-01 3.85736614e-01 8.07214499e-01 1.27516377e+00 -7.96885669e-01 -3.14763010e-01 -7.96803772e-01 -6.73371732e-01 1.44994044e+00 7.11702764e-01 2.30079457e-01 -2.41479993e-01 -1.76106587e-01 4.23147142e-01 -5.40904701e-01 -1.02492452e+00 -2.18029827e-01 2.94308364e-01 4.39969629e-01 1.18610486e-01 -9.42061171e-02 -3.24169964e-01 9.36172783e-01 -7.03523099e-01 -2.09027380e-01 8.31555843e-01 8.88868034e-01 -6.79615438e-01 -1.32185614e+00 -5.59676826e-01 1.16068967e-01 -4.84794348e-01 -2.42609873e-01 -8.66069734e-01 6.48414969e-01 -3.32788736e-01 1.57884228e+00 -2.98729450e-01 4.41403873e-03 5.68148792e-01 4.61862236e-01 5.36638677e-01 -9.48627353e-01 -7.71817029e-01 -6.38800681e-01 8.30721736e-01 -3.11348796e-01 -1.81471288e-01 -2.43976302e-02 -1.82915187e+00 -1.51635066e-01 -1.02985963e-01 8.19718599e-01 2.61703044e-01 1.45589924e+00 4.73995686e-01 6.17382824e-01 3.09226781e-01 2.06905052e-01 -7.85818398e-01 -6.78308547e-01 7.01298192e-02 4.27991062e-01 1.99349031e-01 2.95558959e-01 -3.03183466e-01 1.66021466e-01]
[10.984414100646973, 7.9216437339782715]