Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files 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
d4eb8fb9-684e-4870-a1ee-f2b5b15a1f1b
low-latency-sequence-to-sequence-speech
2005.11185
null
https://arxiv.org/abs/2005.11185v2
https://arxiv.org/pdf/2005.11185v2.pdf
Low-Latency Sequence-to-Sequence Speech Recognition and Translation by Partial Hypothesis Selection
Encoder-decoder models provide a generic architecture for sequence-to-sequence tasks such as speech recognition and translation. While offline systems are often evaluated on quality metrics like word error rates (WER) and BLEU, latency is also a crucial factor in many practical use-cases. We propose three latency reduction techniques for chunk-based incremental inference and evaluate their efficiency in terms of accuracy-latency trade-off. On the 300-hour How2 dataset, we reduce latency by 83% to 0.8 second by sacrificing 1% WER (6% rel.) compared to offline transcription. Although our experiments use the Transformer, the hypothesis selection strategies are applicable to other encoder-decoder models. To avoid expensive re-computation, we use a unidirectionally-attending encoder. After an adaptation procedure to partial sequences, the unidirectional model performs on-par with the original model. We further show that our approach is also applicable to low-latency speech translation. On How2 English-Portuguese speech translation, we reduce latency to 0.7 second (-84% rel.) while incurring a loss of 2.4 BLEU points (5% rel.) compared to the offline system.
['Jan Niehues', 'Gerasimos Spanakis', 'Danni Liu']
2020-05-22
null
null
null
null
['sequence-to-sequence-speech-recognition']
['speech']
[ 3.78959388e-01 2.06882611e-01 -8.53425562e-02 -3.55515689e-01 -1.63990963e+00 -6.72785938e-01 4.73495871e-01 3.43345255e-01 -7.86949039e-01 7.76058137e-01 1.40196338e-01 -8.84348094e-01 4.06631261e-01 -2.56365716e-01 -8.41362357e-01 -3.15246314e-01 8.30408260e-02 6.35066330e-01 3.85730416e-01 -2.99836714e-02 1.07068583e-01 3.35181683e-01 -1.33114970e+00 4.60054815e-01 7.64454067e-01 7.46033072e-01 2.72511303e-01 1.13395691e+00 -8.69188085e-02 6.74201608e-01 -8.03442955e-01 -4.94740367e-01 1.74723938e-01 -5.13526201e-01 -1.01110244e+00 -1.59824684e-01 1.31406963e-01 -4.22514826e-01 -2.90060371e-01 7.89977431e-01 7.50307441e-01 -4.65459116e-02 3.08407962e-01 -7.91188240e-01 8.19968656e-02 7.42970526e-01 -7.50152022e-02 2.55839974e-01 5.29448569e-01 -2.74932589e-02 8.52688193e-01 -1.03533208e+00 4.89833176e-01 9.14422452e-01 5.57903171e-01 5.05952477e-01 -1.15677333e+00 -3.10537398e-01 -1.37593105e-01 2.69760489e-01 -1.52958179e+00 -1.30964792e+00 -4.73527089e-02 -2.19958425e-02 1.67460525e+00 6.07926548e-01 1.67806789e-01 8.82001698e-01 2.78431326e-01 7.32365549e-01 9.70838249e-01 -6.34643316e-01 2.66503960e-01 1.24024972e-01 -1.50336355e-01 4.34887260e-01 -2.79802948e-01 -7.72553980e-02 -7.02206194e-01 3.41134295e-02 2.92932898e-01 -4.99581218e-01 -2.40144029e-01 4.20774758e-01 -1.35949624e+00 4.22269285e-01 -2.16388598e-01 3.91365699e-02 -2.06009150e-01 1.57111362e-01 7.54832685e-01 7.50866294e-01 5.35103381e-01 3.36529404e-01 -6.06163144e-01 -1.13457334e+00 -1.36647522e+00 1.26862908e-02 1.15510762e+00 1.35479808e+00 6.43757224e-01 2.33501829e-02 -1.88984305e-01 9.80100989e-01 -1.30966216e-01 6.33774400e-01 5.51965058e-01 -6.83047235e-01 9.21434283e-01 -7.52827674e-02 4.22609970e-02 -2.60478705e-01 -1.89406872e-01 -3.48869950e-01 -4.84420657e-01 -4.57315177e-01 2.72255957e-01 -2.33196825e-01 -8.87093127e-01 1.43179095e+00 8.55312720e-02 4.14338149e-02 -1.39129907e-01 6.71102941e-01 1.89340308e-01 9.62937176e-01 -2.82551318e-01 -7.82751977e-01 1.37094712e+00 -1.16522264e+00 -7.08199441e-01 -3.21755409e-01 9.52981234e-01 -1.27374816e+00 1.22698569e+00 1.67934477e-01 -1.51453352e+00 -2.27951109e-01 -8.98922980e-01 -1.25341654e-01 2.15328351e-01 6.05222173e-02 -5.13943285e-02 7.32836008e-01 -1.21711278e+00 6.13507926e-01 -1.12001550e+00 -4.41201210e-01 -2.14081824e-01 5.41308105e-01 -1.45974219e-01 1.42556265e-01 -9.98539329e-01 1.00175881e+00 1.74631789e-01 -1.51741832e-01 -7.62978554e-01 -7.08407700e-01 -4.95332897e-01 2.95589894e-01 2.52027929e-01 -2.60835528e-01 1.93878222e+00 -8.43881667e-01 -2.08022285e+00 4.91467595e-01 -7.59617746e-01 -7.31557310e-01 6.13710761e-01 -3.50952685e-01 -6.10763192e-01 1.39230922e-01 -1.91160485e-01 3.50338310e-01 5.85961163e-01 -5.53734660e-01 -7.06257999e-01 1.96742192e-01 -7.94432312e-02 3.13183665e-01 -3.35029602e-01 4.17393625e-01 -8.27983856e-01 -6.36335731e-01 -1.11657351e-01 -1.24700725e+00 8.30606595e-02 -4.69237387e-01 -1.39325485e-01 1.76681310e-01 6.37451589e-01 -1.05085504e+00 1.67165887e+00 -1.94628394e+00 -5.97598478e-02 -6.14698157e-02 -2.18238667e-01 4.53421772e-01 -2.05586880e-01 7.29361236e-01 2.13794768e-01 7.41409045e-03 -4.76387531e-01 -4.99854475e-01 -1.84278920e-01 1.25120908e-01 -2.96782345e-01 4.20783132e-01 1.11528426e-01 9.35441554e-01 -6.98724985e-01 -3.35177243e-01 6.20156117e-02 2.92175323e-01 -5.13605058e-01 3.69564891e-01 -1.35884658e-01 1.00004941e-01 1.05364889e-01 5.75714648e-01 2.78443784e-01 8.23903754e-02 4.08079058e-01 1.03539817e-01 -2.08745897e-01 1.28408504e+00 -7.44128644e-01 1.78730035e+00 -1.08652651e+00 1.00978589e+00 9.64001268e-02 -6.37621820e-01 7.63642013e-01 6.52107000e-01 -1.73677579e-02 -1.05570459e+00 -1.28112644e-01 6.12414300e-01 1.33069530e-01 -2.18320131e-01 6.15200758e-01 -1.26409931e-02 -2.67388634e-02 6.99440658e-01 -1.05970122e-01 7.12501211e-03 1.73140436e-01 1.11747392e-01 1.40397203e+00 5.68628460e-02 3.04663450e-01 -2.38349646e-01 4.02606577e-01 -9.46986303e-02 3.94210756e-01 4.66419518e-01 -7.81769902e-02 6.27856553e-01 5.14994562e-01 -1.92698568e-01 -1.46647751e+00 -8.23867798e-01 -1.84394363e-02 1.06729710e+00 -2.08346680e-01 -7.49847174e-01 -1.07807136e+00 -5.73613226e-01 -4.26093429e-01 8.87340665e-01 8.34668986e-03 -3.47167859e-03 -9.62277114e-01 -4.09831077e-01 9.54918742e-01 4.75673050e-01 1.68670714e-01 -7.52435327e-01 -4.70577508e-01 6.64568305e-01 -3.95878345e-01 -1.38689172e+00 -9.04171228e-01 1.65564314e-01 -8.87863398e-01 -3.58872354e-01 -6.14620030e-01 -5.63599527e-01 4.06466156e-01 1.66721255e-01 1.28270328e+00 -1.04748746e-02 4.75295372e-02 -4.76163849e-02 -4.95038629e-01 3.43038701e-02 -8.82388711e-01 5.37469506e-01 2.16941074e-01 -2.22168162e-01 3.02771538e-01 -5.29559731e-01 -5.00914156e-01 5.21951497e-01 -3.97990227e-01 9.48461443e-02 6.96016550e-01 8.69163156e-01 4.71611589e-01 -6.17337942e-01 5.55727959e-01 -7.10600019e-01 4.61684793e-01 -2.52257615e-01 -6.21595144e-01 3.18087399e-01 -9.06459868e-01 2.57976443e-01 8.67587686e-01 -4.24061120e-01 -8.40208352e-01 -6.77089244e-02 -5.13206482e-01 -2.08299324e-01 3.31079885e-02 3.11516672e-01 -2.09533013e-02 1.73707277e-01 4.82111901e-01 4.07147586e-01 3.00318878e-02 -4.91928637e-01 2.51436830e-01 1.28843260e+00 3.48246008e-01 -3.77300650e-01 4.20826524e-01 -1.74755067e-01 -4.66149777e-01 -8.72015238e-01 -4.00389165e-01 -5.06962419e-01 -3.11585218e-01 7.50315934e-02 3.82520616e-01 -1.11781871e+00 -6.88669443e-01 3.90449613e-02 -1.30108464e+00 -5.48895240e-01 -1.09428264e-01 5.60475647e-01 -7.48266459e-01 4.01012421e-01 -9.55064237e-01 -7.50994086e-01 -6.72586203e-01 -1.30199957e+00 1.27874577e+00 -4.51082677e-01 -4.51712728e-01 -5.75091422e-01 -1.15627192e-01 2.83012122e-01 4.68792915e-01 -6.75846815e-01 6.32028818e-01 -6.67329669e-01 -4.53869969e-01 -9.22456533e-02 -2.80444443e-01 3.04624557e-01 -1.50799423e-01 -2.59062558e-01 -1.04329813e+00 -4.79848415e-01 -2.12156236e-01 -1.21210799e-01 5.56750894e-01 -1.15370132e-01 1.00199759e+00 -6.29790604e-01 -3.13758589e-02 4.03858066e-01 1.26478398e+00 5.08696735e-01 7.15840936e-01 6.36069998e-02 4.57702577e-01 3.75689059e-01 6.31811798e-01 3.76094908e-01 2.69244373e-01 1.13995433e+00 -6.93051592e-02 3.04548413e-01 -3.49314243e-01 -2.58853078e-01 1.04032576e+00 1.86420786e+00 1.28332153e-01 -4.49822873e-01 -1.09013319e+00 7.01359391e-01 -1.78819501e+00 -7.93524444e-01 -2.12808661e-02 2.55747271e+00 1.06700194e+00 3.30957353e-01 2.96314895e-01 2.07245260e-01 5.60765266e-01 -3.66431698e-02 -1.15022801e-01 -1.06128633e+00 2.57907927e-01 4.47216362e-01 7.93890357e-01 1.00765586e+00 -5.10745883e-01 1.09982705e+00 6.36295462e+00 1.08588743e+00 -1.23655391e+00 4.94197488e-01 5.09215176e-01 -4.47899461e-01 -3.23431462e-01 2.08676666e-01 -9.09800410e-01 7.17257798e-01 2.12412786e+00 -2.08450124e-01 7.72503197e-01 5.40138900e-01 3.82529467e-01 9.59977359e-02 -1.31025541e+00 1.04170120e+00 -3.53096277e-02 -1.24090040e+00 -2.62369275e-01 9.80003104e-02 5.45831323e-01 2.69765437e-01 -3.84972215e-01 4.17915344e-01 -1.73932970e-01 -9.14560497e-01 9.21281576e-01 7.85074085e-02 1.44713950e+00 -9.09940362e-01 7.91766763e-01 4.31813687e-01 -1.21488416e+00 3.31135213e-01 -2.80012935e-01 -1.03259161e-01 5.42064846e-01 7.78364897e-01 -1.35757196e+00 4.79519963e-01 2.75876969e-01 9.80841219e-02 -8.59918594e-02 6.74747825e-01 -1.45052984e-01 9.89462256e-01 -4.75329727e-01 -2.24628344e-01 3.26526374e-01 8.04624632e-02 4.21537459e-01 1.74877775e+00 6.32813513e-01 -1.43598840e-01 -2.19861656e-01 5.13518825e-02 -2.86454767e-01 3.39976639e-01 -1.82779476e-01 -1.04112171e-01 9.87196028e-01 7.80952454e-01 -4.74935889e-01 -5.79275668e-01 -4.01573628e-01 1.52339911e+00 3.61280888e-01 1.80067375e-01 -1.02962410e+00 -7.33159482e-01 7.33508289e-01 1.87676415e-01 3.90379488e-01 -4.56436038e-01 -2.53785074e-01 -1.10816216e+00 3.35410178e-01 -1.24215949e+00 -4.16758567e-01 -2.49247566e-01 -3.99866790e-01 8.73665392e-01 -3.29706073e-01 -1.20305836e+00 -6.84641063e-01 -2.26447091e-01 -1.98811740e-01 1.01867414e+00 -1.23681235e+00 -5.58333397e-01 1.23514652e-01 1.42166808e-01 1.11477244e+00 -1.89190786e-02 1.05132318e+00 7.42601395e-01 -5.47139466e-01 1.02039862e+00 3.17465007e-01 -2.18095869e-01 7.88589835e-01 -9.88479078e-01 1.14874649e+00 9.02444482e-01 3.56189758e-01 7.03326643e-01 6.97569549e-01 -4.07946527e-01 -1.62142527e+00 -9.81450737e-01 1.73019302e+00 -2.43536457e-01 6.25995517e-01 -6.81254208e-01 -8.04041088e-01 5.91358781e-01 4.24690992e-01 -2.12169170e-01 4.97374028e-01 3.11785638e-01 -3.29243779e-01 -1.96956649e-01 -7.11545050e-01 8.33458245e-01 1.11490238e+00 -9.56436932e-01 -2.44449079e-01 3.36693168e-01 1.03509843e+00 -6.57310545e-01 -9.73990023e-01 3.69319953e-02 7.71068394e-01 -7.67704666e-01 4.29491371e-01 -3.21720183e-01 1.39266431e-01 -2.87210733e-01 -3.21047813e-01 -1.26662016e+00 6.63106050e-03 -1.23079467e+00 -4.44294095e-01 1.02272749e+00 9.02685344e-01 -4.67357993e-01 5.46834886e-01 3.29178125e-01 -4.03463840e-01 -8.51867616e-01 -1.20163858e+00 -1.06386101e+00 -1.76716536e-01 -8.35867822e-01 6.20700955e-01 4.41706598e-01 1.34815425e-01 5.19087970e-01 -5.14090359e-01 4.90264110e-02 9.07522440e-02 -1.60945073e-01 7.29154289e-01 -3.78655821e-01 -5.47727466e-01 -2.90923595e-01 -1.52336538e-01 -1.60416663e+00 -6.71918094e-02 -8.05085957e-01 3.80035967e-01 -1.18262279e+00 -2.26794444e-02 -3.50868791e-01 -1.02281585e-01 3.86828303e-01 5.48435841e-03 8.64007920e-02 3.43068272e-01 2.42631927e-01 -4.48401511e-01 3.97957683e-01 5.79417288e-01 1.00102410e-01 -1.52239814e-01 -9.87253040e-02 -1.80271477e-01 1.65093288e-01 8.56535017e-01 -6.65186465e-01 -3.64115596e-01 -8.86611938e-01 1.60819769e-01 4.57954735e-01 -2.26771340e-01 -9.09539878e-01 3.84585261e-01 1.42874554e-01 -4.08264965e-01 -3.92962545e-01 3.81938875e-01 -5.08433640e-01 1.22625560e-01 5.23278296e-01 -2.50703365e-01 4.63539541e-01 3.10345829e-01 2.63017952e-01 -3.39807481e-01 -2.16089353e-01 6.38792157e-01 2.06579462e-01 -3.22992325e-01 6.97674453e-02 -7.15920389e-01 1.03451423e-01 7.45520771e-01 -1.63953260e-01 -7.71139115e-02 -5.25558889e-01 -5.61789870e-01 -1.52134985e-01 4.48724121e-01 2.33685568e-01 4.52030867e-01 -9.57833350e-01 -8.33818436e-01 2.06881613e-01 5.84464073e-02 -4.43124324e-01 -1.03889398e-01 1.12072372e+00 -9.03847575e-01 7.78702378e-01 3.29908311e-01 -6.24582946e-01 -1.38323665e+00 1.74350545e-01 6.99509531e-02 -2.90442944e-01 -4.87053305e-01 8.49596798e-01 -2.45639130e-01 -2.23531693e-01 2.09931463e-01 -2.41038591e-01 7.11341739e-01 -1.15688391e-01 6.10717118e-01 6.17583215e-01 7.93057621e-01 -3.54674041e-01 -6.54242337e-01 8.93809274e-02 -2.64299691e-01 -7.20994174e-01 9.23916757e-01 -3.00212651e-01 4.36277054e-02 4.22246188e-01 1.47670257e+00 2.77896285e-01 -1.00095332e+00 -7.62674138e-02 3.36673737e-01 -2.10016653e-01 -7.30550587e-02 -8.24885428e-01 -6.17372751e-01 1.00286543e+00 4.00305718e-01 -4.82098311e-02 1.12024260e+00 -3.99430811e-01 1.24987078e+00 4.54600990e-01 4.67246860e-01 -1.19173157e+00 -5.36685050e-01 9.29655790e-01 6.60757422e-01 -9.17306125e-01 -2.61380345e-01 -2.59167016e-01 -4.76780057e-01 1.02433360e+00 8.65162537e-02 2.86770374e-01 1.57919213e-01 7.91975141e-01 1.02766179e-01 5.80475569e-01 -1.34763145e+00 -1.01334173e-02 8.31115320e-02 1.10409617e-01 8.37624371e-01 3.37304175e-01 -6.69697583e-01 1.23998329e-01 -3.28296840e-01 -1.83522046e-01 5.15111804e-01 8.15261960e-01 -3.92245561e-01 -1.46836627e+00 -1.57917365e-01 2.76749581e-01 -5.73471606e-01 -5.98508000e-01 -1.56078145e-01 4.68294948e-01 -4.56657588e-01 1.15194595e+00 3.41038108e-01 -5.70260465e-01 3.55178505e-01 3.62099588e-01 4.20733333e-01 -7.21251905e-01 -8.93651724e-01 3.93993944e-01 7.74717450e-01 -8.46474946e-01 -2.27037109e-02 -7.30908215e-01 -1.11074853e+00 -7.06347883e-01 -4.17450249e-01 3.93683702e-01 9.30230737e-01 8.50185931e-01 6.93171561e-01 4.66862679e-01 7.17927396e-01 -3.27771008e-01 -6.81135237e-01 -1.04657698e+00 -1.28508985e-01 -1.95171759e-01 3.71494710e-01 -3.90102044e-02 -2.74070382e-01 9.29246768e-02]
[14.464800834655762, 7.032690525054932]
15107676-c39f-433e-b8d1-e3de88d40e01
st-mfnet-mini-knowledge-distillation-driven
2302.08455
null
https://arxiv.org/abs/2302.08455v2
https://arxiv.org/pdf/2302.08455v2.pdf
ST-MFNet Mini: Knowledge Distillation-Driven Frame Interpolation
Currently, one of the major challenges in deep learning-based video frame interpolation (VFI) is the large model sizes and high computational complexity associated with many high performance VFI approaches. In this paper, we present a distillation-based two-stage workflow for obtaining compressed VFI models which perform competitively to the state of the arts, at a greatly reduced model size and complexity. Specifically, an optimisation-based network pruning method is first applied to a recently proposed frame interpolation model, ST-MFNet, which outperforms many other VFI methods but suffers from large model size. The resulting new network architecture achieves a 91% reduction in parameters and 35% increase in speed. Secondly, the performance of the new network is further enhanced through a teacher-student knowledge distillation training process using a Laplacian distillation loss. The final low complexity model, ST-MFNet Mini, achieves a comparable performance to most existing high-complex VFI methods, only outperformed by the original ST-MFNet. Our source code is available at https://github.com/crispianm/ST-MFNet-Mini
['David R. Bull', 'Nantheera Anantrasirichai', 'Fan Zhang', 'Duolikun Danier', 'Crispian Morris']
2023-02-16
null
null
null
null
['video-frame-interpolation']
['computer-vision']
[ 0.3056803 0.18613818 -0.3226866 -0.08599804 -0.5854744 0.12763354 0.50958943 -0.16702968 -0.539411 0.9283939 -0.01948026 -0.5329595 -0.01200223 -0.5231773 -1.0744643 -0.32698914 -0.07097666 0.33525616 0.20377061 0.03461755 0.12624504 0.2734691 -1.4386399 0.31767377 1.0485402 1.2848457 0.2631092 0.80969775 0.0826455 1.1692653 -0.38056192 -0.42753518 0.22497723 -0.25674522 -1.0435566 -0.65243065 0.860149 -0.59573483 -0.63126576 0.63886315 0.5533594 0.21604168 0.29083988 -1.163414 -0.38474053 0.7177912 -0.25622913 0.14292277 0.08853737 0.17955908 0.8301688 -1.1017046 0.5396103 1.2025073 1.0465378 0.68772 -1.346124 -0.8565532 0.16433765 0.5441124 -1.5588095 -0.547181 0.69707423 -0.302263 1.2151313 0.20146976 0.88816017 1.0017163 0.01330307 1.0132049 0.71871966 -0.30452105 0.09879306 -0.24530718 -0.24104768 0.82028735 0.02962975 0.18251537 -0.61400515 0.2181187 1.0743487 -0.09289041 -0.41024327 -0.04407057 -1.2153271 0.66668254 0.6667345 0.16572268 -0.1323016 0.6398451 0.43249616 0.29010826 0.63622344 0.24041446 -0.7190645 -0.32245702 -1.5326144 0.6426939 0.88499254 0.9943516 0.7803454 0.4554348 -0.38142356 0.71047515 0.41720706 0.13993375 0.30922166 -1.4265597 0.557193 0.2953007 -0.14025293 -0.75331235 -0.32383257 -0.56230026 -1.3293549 0.22984807 0.35495475 -0.05539814 -0.97427934 1.6661083 0.08554448 0.8860916 -0.28952405 0.65351933 1.1590178 0.52600294 0.15164344 0.12332745 0.9091846 -1.4233881 -0.33166385 -0.12144731 0.55853033 -0.6810241 1.0983205 0.49163213 -1.6162814 -0.9124468 -1.0433836 -0.54235446 -0.02641973 0.22625203 0.77239347 0.28478578 -1.3832325 1.0583907 -0.6995572 0.13852012 1.1228801 0.76659375 0.01152051 -0.13684724 -1.1669692 0.6944973 0.6017044 0.21292983 -1.0026325 -1.405404 -0.9468081 0.35935766 0.34838006 -1.1052792 1.4263399 -0.95272595 -1.6294303 0.43314913 -0.5614442 -1.0059974 0.79300225 -0.29100904 -0.10362956 0.0099919 -0.29422703 1.3399096 0.9468708 -1.0519882 -0.6591853 0.20954342 0.24030447 -0.05736408 -0.05566588 -0.30009803 -0.56519777 -0.81537634 -0.39330155 -0.934016 -0.33121732 0.26666895 -0.13158564 -0.380759 0.7693948 -0.780899 1.5869572 -1.8576136 0.09089918 -0.08773624 0.6374448 0.9182551 -0.11074407 -0.10557152 -0.10909469 0.26589975 -0.36944273 -0.6796329 -0.2240279 0.285164 0.09944312 0.04917651 0.38556775 1.1604683 -0.8924238 -0.5726625 0.38883844 0.8525505 -1.0600065 -0.04464083 -0.3486599 0.6521574 0.21600924 0.49220848 0.81394804 -0.4604045 -0.19266102 -0.58160293 -0.04249943 0.3986301 -0.96340036 2.1327817 -0.65855175 0.8403831 -0.05934306 -1.1557493 0.6331837 0.4950282 0.459492 -0.496344 0.15298891 0.3714497 0.01419372 -0.19041206 0.5342769 0.35619578 0.5476061 -0.09237825 0.38275567 0.0409744 0.4663501 0.2549505 0.6778311 0.64811885 -0.15639608 -0.3466244 0.8032466 -0.18645886 0.773086 0.88336027 -0.17867944 0.6770163 0.30269057 -0.7951318 -1.2670759 -0.7653256 -0.04304068 0.91618794 -0.04908866 -0.6065132 -0.85422987 -0.39225313 -0.09944718 0.6292504 -0.4810877 -0.01147491 -1.0170262 -0.2992187 0.72341007 0.6259779 0.8802173 -0.9826925 -0.3782463 0.262953 -0.36748257 -1.0513623 -0.35569757 -0.03033771 -1.1591936 -0.654627 -0.872247 -0.9379703 0.3480006 0.04367443 1.5084091 0.4972456 -0.13047734 -0.3506641 0.06562381 -0.4697025 -0.41597694 0.60470843 -0.2766283 -0.43809223 0.2183761 -0.7003332 -0.98782814 0.018224 -0.8048096 0.67444867 0.35744172 0.9107689 0.69683367 -0.35849628 0.52108485 -0.70853674 0.21572334 -0.30351406 -0.4677076 -0.01991088 -0.7720678 0.05921543 0.72457796 -0.5371259 -1.0573409 -0.04508875 -0.5958845 -0.7105243 -0.03007457 0.44177228 0.16638252 -0.2987152 0.5951139 -0.07439863 0.03973003 -0.46410748 0.31910726 0.25637814 0.636948 -0.42608905 0.61620724 0.2729364 0.20272481 -0.6783504 -0.88621795 -0.11601125 -0.7285713 -0.1907078 0.6288658 -1.2451462 -0.75194776 0.56333804 -1.3504541 -0.82880914 -0.47778693 0.32618433 -0.6781137 0.26221335 -0.7693132 -0.2965497 -0.73426425 -1.3542091 0.7600607 0.1604217 -0.08266607 -1.1010413 -0.267362 0.46617755 0.8429731 0.10762539 0.74569535 -0.2714459 -0.7533277 0.18897931 -0.49807313 0.5914872 -0.23554456 0.04981072 -0.93755186 -0.36918685 -0.29363066 -0.23652862 1.202058 0.7621532 1.6288618 -0.41388947 -0.05877507 1.2398721 1.4783633 0.02083488 0.9080657 0.3826921 1.1042018 -0.07973833 0.13776101 0.41498402 0.48779264 0.5420989 0.4549529 -0.37473583 -0.52205807 -0.24204876 0.05435142 0.94527805 -0.62897635 -0.1250985 -0.79075176 0.51600766 -1.9700487 -0.8836577 -0.2187554 2.0204186 0.8864791 0.17483893 0.12613316 0.3626157 0.31459063 0.18759479 -0.6687037 -0.43482506 0.05489003 0.55537266 0.6557471 0.6408676 -1.2029482 1.0770326 6.2500844 1.181438 -1.0656862 0.20631103 0.93308496 -0.31601328 -0.08334754 -0.02202068 -0.8271332 0.49753 1.3324411 -0.03947338 0.75451684 0.71147364 0.44100982 0.07497786 -1.0107356 1.0712799 -0.14296243 -2.0635571 0.17415476 -0.2331311 0.77541333 0.4295512 0.02261819 0.40752298 0.16771396 -1.2042581 0.7571028 0.49009496 1.2025628 -0.922254 0.58255017 0.12870464 -1.492017 -0.11472528 -0.55135494 -0.4355119 0.27098218 0.5282383 -0.46920016 0.63235974 0.8572104 0.9162672 -0.4858945 1.2381998 -0.14769995 1.0061738 -0.3695549 0.3922819 0.2337382 -0.06755146 0.29171067 1.3648278 0.2908374 -0.1918745 0.01481978 0.83609146 -0.38069013 -0.28545702 -0.31116092 0.4666248 0.46755412 1.0721035 -0.2636278 -0.80489653 -0.52798265 0.8501818 0.29325387 0.4574299 -1.0572462 -0.33620828 0.66402227 0.27321082 0.5271115 -0.13258919 -0.39591888 -1.3094577 -0.10622253 -0.8750668 0.02262266 -0.62284476 -0.7383899 0.70613635 0.06023904 -1.1703994 -0.27965203 -0.35185045 -0.49152967 0.94878423 -1.9656564 -1.2857916 -0.54990727 0.65105665 0.7547062 -0.05758893 0.5459042 0.8047359 -0.65332884 1.0456432 0.12752105 0.02825633 0.43813708 -0.94875103 0.84987575 0.7611396 -0.05209498 0.3885994 0.32564977 -0.36634785 -1.1314516 -1.4780079 1.1237851 0.02275899 0.4083956 -0.27933684 -1.1038549 0.7700261 0.22693314 0.22141223 0.23354113 -0.23811096 -0.15592644 -0.14323981 -1.1740594 0.6254522 1.3497162 -0.17041704 -0.04791807 0.23323752 1.1203669 -0.7214695 -1.1121738 0.6068866 0.6807672 -0.9740951 1.3562801 -0.3561016 0.82205844 -0.24700138 0.21290381 -0.91660285 -0.5800697 -0.8537274 -0.74176365 0.9687882 0.4117505 -0.46524462 1.0044849 0.5434711 -0.16903406 -1.1733172 -1.1101995 -0.74663776 0.22000974 -0.49986386 0.74161375 0.49651888 -0.58589447 0.2128512 -0.6536447 -0.39899403 0.6335736 -0.37721595 0.7290725 -1.2654166 -0.34977418 -0.54914427 -0.15771887 -1.5323026 -0.05783636 -1.0238074 -0.21188864 -1.6647036 -0.09258422 -0.4071725 -0.25000873 0.34933156 -0.21762948 0.50278527 0.5963035 0.3522068 -0.47338122 0.5505772 1.3248538 -0.06726179 -0.30839452 -0.04802004 -0.36344135 0.9804419 0.9270123 -0.26608336 -0.44525832 -0.8177875 0.06058733 -0.02237639 0.45886052 -1.4422076 0.27418882 0.1544475 0.47842926 -0.73003083 0.3036454 -0.5527056 0.2408174 0.67547977 -0.36239833 0.14459512 0.3400406 0.28861937 -0.16688794 -0.2587652 0.8475514 -0.22517632 -0.6549662 0.65607375 -0.28341708 0.08048822 0.52925545 -0.36102262 -0.04320239 -0.15556003 -0.5177873 -0.06504535 0.17792264 0.15249413 0.6290187 -1.286223 -0.88332546 0.26391923 -0.50203246 0.67275566 0.24488138 0.939222 -0.8254165 0.4534938 -0.05155481 -0.5923471 -1.1759735 0.33976707 0.20153512 -0.59835064 -0.8085739 1.0584164 0.06343754 -0.3059384 0.41803992 -0.665558 -0.00860445 -0.39965102 0.4550397 0.80066246 0.02724609 -0.5500629 -0.06842463 0.34534705 -0.11140375 0.30349422 1.3540885 0.02734235 0.03559947 -0.06668535 1.2678785 -0.66856086 -1.7318611 -0.3466787 -0.29851028 -0.4361212 0.16487837 -0.64377004 -1.4535979 0.9615123 0.5932382 -0.4020296 1.2629883 -0.2951349 1.1455941 0.305918 0.2112557 -0.82645845 -0.29551402 0.73333246 0.69248354 -1.1687568 0.14751965 -0.24740076 -0.29475173 0.97333086 0.56691855 -0.31455925 0.83360356 0.25400075 -0.34000304 0.20441926 -0.812005 0.19385885 0.48498875 0.5119493 0.5467783 -0.20219012 -0.3065093 0.2559965 -0.30138785 0.6985296 0.22886644 0.6980546 -0.18247937 -0.9944306 0.16826376 0.7518446 -0.58774126 -0.5011376 0.471847 0.6283501 0.30196422 0.74931824 0.04210802 -0.44467947 0.15846585 -0.18622789 0.47866452 -0.0356456 -0.79718506 -0.08558155 -0.07222369 -0.8002315 -0.64043134 -0.2784207 -1.0786098 -0.7205322 -0.28907678 -0.32021275 0.636781 0.8029406 0.6780416 0.77686507 0.14252919 -1.5188302 -0.18440838 -0.8221283 0.16543888 0.2225605 0.39831015 -0.50868875 -0.2657678 0.28850466]
[10.67928409576416, -1.3576750755310059]
4c6f75df-2f59-434b-94c5-cf475f039f6d
a-novel-bi-hemispheric-discrepancy-model-for-1
1906.01704
null
http://arxiv.org/abs/1906.01704v1
http://arxiv.org/pdf/1906.01704v1.pdf
A Novel Bi-hemispheric Discrepancy Model for EEG Emotion Recognition
The neuroscience study has revealed the discrepancy of emotion expression between left and right hemispheres of human brain. Inspired by this study, in this paper, we propose a novel bi-hemispheric discrepancy model (BiHDM) to learn the asymmetric differences between two hemispheres for electroencephalograph (EEG) emotion recognition. Concretely, we first employ four directed recurrent neural networks (RNNs) based on two spatial orientations to traverse electrode signals on two separate brain regions, which enables the model to obtain the deep representations of all the EEG electrodes' signals while keeping the intrinsic spatial dependence. Then we design a pairwise subnetwork to capture the discrepancy information between two hemispheres and extract higher-level features for final classification. Besides, in order to reduce the domain shift between training and testing data, we use a domain discriminator that adversarially induces the overall feature learning module to generate emotion-related but domain-invariant feature, which can further promote EEG emotion recognition. We conduct experiments on three public EEG emotional datasets, and the experiments show that the new state-of-the-art results can be achieved.
[]
2019-05-11
a-novel-bi-hemispheric-discrepancy-model-for
https://arxiv.org/abs/1906.01704
https://arxiv.org/pdf/1906.01704
arxiv190601704-search-help-advanced-search
['eeg-emotion-recognition']
['miscellaneous']
[-5.31941056e-02 -1.59118310e-01 4.46813852e-01 -7.18735933e-01 -1.98072061e-01 -3.85985136e-01 2.26849273e-01 -6.61805332e-01 -2.09079117e-01 7.92400599e-01 2.89881319e-01 5.05173318e-02 -1.64549068e-01 -5.46134889e-01 -5.19569099e-01 -9.76597369e-01 -1.55329242e-01 -3.05838495e-01 -4.24972147e-01 -2.40836769e-01 1.76337868e-01 5.15815675e-01 -1.22915149e+00 4.04280156e-01 9.81718004e-01 1.35176253e+00 -1.70429364e-01 -4.72879596e-02 5.54479882e-02 5.33928335e-01 -6.57514095e-01 -2.79830605e-01 2.06185788e-01 -8.06667626e-01 -6.95760667e-01 -1.13332503e-01 -4.02018905e-01 -3.69931944e-02 -7.06321359e-01 1.14447343e+00 9.91063058e-01 -9.03297123e-03 9.50728118e-01 -1.46620953e+00 -8.15324545e-01 2.68009365e-01 -7.58030772e-01 2.49019727e-01 1.88392922e-01 8.23025554e-02 6.25421584e-01 -7.74644613e-01 3.81349385e-01 6.81930482e-01 3.61502767e-01 6.82648003e-01 -9.84662414e-01 -1.41940212e+00 3.40372264e-01 3.50889444e-01 -1.68215537e+00 -1.87024266e-01 1.34375501e+00 -2.54512042e-01 6.57832503e-01 1.28455341e-01 1.01750851e+00 1.70046461e+00 6.59088135e-01 6.10286176e-01 1.51189077e+00 2.10585207e-01 1.62438512e-01 1.41321927e-01 -1.04329407e-01 1.52934968e-01 -2.40024596e-01 6.46609664e-02 -5.53648591e-01 -1.64949126e-03 6.90379024e-01 1.42343551e-01 -7.27200210e-01 -3.69066000e-01 -1.20789516e+00 4.37550247e-01 7.12308705e-01 5.69154143e-01 -6.40873671e-01 -4.68064547e-01 7.57007480e-01 5.81007659e-01 4.60814655e-01 4.75794464e-01 -4.60174769e-01 -6.68076277e-02 -7.43328750e-01 -1.16094552e-01 6.11300111e-01 7.62552619e-01 4.18424934e-01 6.51017949e-02 -3.73993129e-01 8.63660216e-01 1.14938982e-01 1.57113597e-01 9.59003806e-01 -3.59940141e-01 4.77034748e-01 6.82516277e-01 -3.33947957e-01 -1.31876493e+00 -6.25843704e-01 -5.81254601e-01 -1.56826305e+00 -3.76493894e-02 -1.65173113e-01 -4.10755605e-01 -4.55164045e-01 2.06481409e+00 -1.23258173e-01 3.34466696e-01 2.48699963e-01 1.21086049e+00 7.62475193e-01 4.86174852e-01 9.96082462e-03 -1.71498746e-01 1.34702182e+00 -6.09442890e-01 -6.25479341e-01 -6.78301528e-02 4.71496731e-01 -1.33095190e-01 8.03194821e-01 4.61414278e-01 -9.26789820e-01 -3.39593977e-01 -1.30067599e+00 1.64763615e-01 -6.26035333e-01 1.12421505e-01 5.19150019e-01 4.10874844e-01 -7.60632813e-01 4.58534658e-01 -5.01304150e-01 1.70330390e-01 7.70328343e-01 4.57452089e-01 -6.50481701e-01 2.48972446e-01 -1.66386843e+00 7.31518090e-01 1.95767671e-01 5.75839520e-01 -5.69382131e-01 -6.58664346e-01 -7.27840245e-01 1.73059121e-01 -4.54042882e-01 -5.42339623e-01 3.84341180e-01 -1.46095467e+00 -1.59593070e+00 9.00023580e-01 9.94878858e-02 7.61354342e-03 3.80761802e-01 3.90540034e-01 -8.23599160e-01 -4.00045626e-02 -1.53278261e-01 6.93723321e-01 6.08115911e-01 -9.20945346e-01 -4.77685072e-02 -7.09215283e-01 -4.22722220e-01 2.91503429e-01 -6.08286440e-01 1.40489519e-01 1.06158420e-01 -8.67914975e-01 2.02171937e-01 -5.99599302e-01 2.24771932e-01 -2.95216620e-01 -4.87056494e-01 -1.34716153e-01 4.46326822e-01 -8.34591925e-01 1.02065575e+00 -2.44385123e+00 5.52513480e-01 5.47532082e-01 3.20890844e-01 -2.39175394e-01 -3.18207353e-01 -1.11742718e-02 -7.95105100e-01 -8.47227126e-02 -4.68750238e-01 6.49914294e-02 7.32508600e-02 -1.82696190e-02 -4.68075216e-01 6.51395440e-01 4.03162956e-01 9.73254442e-01 -4.35726434e-01 -1.11756928e-01 -1.71401381e-01 5.88578224e-01 -3.53434235e-01 4.36341703e-01 5.93424678e-01 9.20095742e-01 -5.42358458e-01 2.59323984e-01 1.18193781e+00 1.68878391e-01 4.72910814e-02 -3.71241361e-01 3.99570912e-02 2.71513045e-01 -9.57157075e-01 1.91547215e+00 -4.26955819e-01 5.37325919e-01 -4.92093749e-02 -1.41370857e+00 1.23760819e+00 3.98051947e-01 4.71369952e-01 -1.10431445e+00 5.06528378e-01 1.64143205e-01 2.61674643e-01 -4.74242121e-01 -2.83985317e-01 -3.31743538e-01 -2.03258201e-01 4.72764492e-01 7.43145570e-02 1.15609914e-01 -5.87709606e-01 -3.33660096e-01 7.73949564e-01 -3.95015143e-02 -2.56184656e-02 -6.34644032e-01 7.23431647e-01 -8.10347795e-01 8.10670078e-01 1.99532464e-01 -3.73396993e-01 5.91266572e-01 8.57756436e-01 -4.33046371e-01 -6.24231458e-01 -8.96880269e-01 -3.68393004e-01 7.46677220e-01 4.28636521e-01 2.20742792e-01 -8.78475428e-01 -5.88112175e-01 -2.92552173e-01 6.57046616e-01 -9.60909784e-01 -8.74086022e-01 -2.00679392e-01 -8.72225761e-01 8.09552372e-01 8.51704955e-01 9.03893590e-01 -1.20489180e+00 -4.85657007e-01 -7.23700151e-02 -8.31784159e-02 -6.28590584e-01 -3.70497614e-01 3.42791498e-01 -3.96977395e-01 -8.18146944e-01 -1.02813530e+00 -9.26731050e-01 6.67306960e-01 -2.83418179e-01 8.51669729e-01 -3.16789001e-01 -2.62665510e-01 4.08742465e-02 -1.73091859e-01 -3.20656657e-01 4.42316681e-01 8.88673291e-02 1.55485362e-01 3.95565271e-01 6.75684810e-01 -1.12962520e+00 -8.93430293e-01 5.15531301e-01 -8.93861949e-01 1.51091386e-02 4.88806695e-01 9.99669135e-01 4.15083766e-01 -1.12212539e-01 1.04353356e+00 -2.81584024e-01 1.04742289e+00 -6.90581739e-01 -3.55683640e-02 3.02137733e-01 -2.92297751e-01 4.35009599e-02 8.49734366e-01 -5.36765754e-01 -1.13278270e+00 -3.38653773e-01 -1.52695626e-01 -4.52872097e-01 -2.24607483e-01 3.61637324e-01 -6.11613512e-01 -9.82289463e-02 3.11855853e-01 6.77484989e-01 -1.92865968e-01 -1.31729752e-01 -3.89676243e-02 9.19535518e-01 3.70248944e-01 -4.93852943e-01 2.07778469e-01 2.22230166e-01 -1.79110155e-01 -2.32239395e-01 -3.41119736e-01 1.66487526e-02 -6.41862810e-01 7.41261765e-02 9.32867825e-01 -9.25430596e-01 -8.63731146e-01 6.71226025e-01 -1.30687845e+00 -4.24684994e-02 6.63336068e-02 7.21063077e-01 -5.72218716e-01 2.80608777e-02 -5.10527849e-01 -3.84780735e-01 -4.37639296e-01 -1.21336591e+00 8.14643204e-01 2.44747952e-01 -5.83159477e-02 -6.44494593e-01 5.58521971e-02 -3.91517878e-01 3.93445998e-01 8.88214782e-02 9.70355392e-01 -9.18336928e-01 1.36033431e-01 -2.37982627e-02 -4.47254300e-01 4.88739073e-01 -2.54451763e-02 -4.36715901e-01 -1.04590797e+00 -1.07973129e-01 5.29331148e-01 -4.87989604e-01 6.91430867e-01 9.16821659e-02 1.84946096e+00 -3.87574062e-02 -1.75932407e-01 1.12780952e+00 8.96435797e-01 3.48417342e-01 1.00587070e+00 2.13265806e-01 4.02036399e-01 8.52417946e-01 6.58016652e-02 4.54223216e-01 3.46948534e-01 1.78958327e-01 1.42628640e-01 -2.82585710e-01 4.79401082e-01 -1.59025922e-01 5.94779626e-02 9.82734621e-01 -1.55355141e-01 4.90981638e-02 -6.16689026e-01 2.90785372e-01 -1.43602085e+00 -8.16735029e-01 4.89576668e-01 1.96524978e+00 6.51831806e-01 -3.48417521e-01 -1.54113248e-01 -5.57432622e-02 7.75274873e-01 1.87904254e-01 -9.47186828e-01 -4.18676883e-01 -2.81150550e-01 2.87091315e-01 7.50277638e-02 -3.56834650e-01 -7.88144290e-01 3.95835400e-01 5.92568493e+00 6.63284421e-01 -1.55077899e+00 -2.11884342e-02 9.56683099e-01 -3.25416982e-01 -4.67723638e-01 -5.93154252e-01 -6.43491372e-02 7.70784140e-01 6.99329436e-01 -3.03340524e-01 7.95585871e-01 5.68256021e-01 2.41980255e-02 6.53871596e-01 -1.07148659e+00 1.56206489e+00 1.88659340e-01 -7.39584744e-01 -1.54173272e-02 5.18026091e-02 6.10798836e-01 -8.95568822e-03 1.55908003e-01 5.55164278e-01 -3.49120736e-01 -1.42912638e+00 3.02472740e-01 8.68378878e-01 8.99465084e-01 -1.18102837e+00 8.99008632e-01 3.18690330e-01 -1.00808287e+00 -9.96098146e-02 -6.56522274e-01 -4.92396280e-02 -2.73082703e-01 4.00833756e-01 1.31374538e-01 6.77693367e-01 9.47382390e-01 9.51135099e-01 -3.65932494e-01 6.74432039e-01 -1.56571656e-01 1.56436384e-01 4.45172116e-02 -1.48379818e-01 3.99508979e-03 -5.41807413e-01 2.11076304e-01 9.97268915e-01 3.84133846e-01 3.17451954e-01 -3.68351728e-01 1.37361491e+00 -4.59032327e-01 1.63794562e-01 -8.16764176e-01 1.29782289e-01 3.02649409e-01 1.32459700e+00 -3.69072616e-01 -1.00475155e-01 -3.06624413e-01 1.61396551e+00 5.19390345e-01 6.55197561e-01 -1.05194604e+00 -9.81001914e-01 7.68705308e-01 -6.43401146e-01 -5.80302328e-02 2.48182997e-01 -3.37425500e-01 -1.43450928e+00 3.43545347e-01 -9.15567636e-01 3.34899761e-02 -9.62977350e-01 -1.66585851e+00 9.87842023e-01 -3.94642621e-01 -1.30746937e+00 1.26607111e-02 -7.27148414e-01 -9.77936268e-01 1.32982337e+00 -1.42585838e+00 -9.41343844e-01 -3.46795499e-01 9.35862422e-01 -6.23351634e-02 -2.02391967e-01 9.03204679e-01 3.85721564e-01 -8.09024215e-01 7.84590125e-01 8.06880295e-02 4.51784402e-01 8.49304676e-01 -7.83677340e-01 -1.59716338e-03 2.89878786e-01 -3.07330459e-01 7.66846120e-01 1.22569673e-01 -1.10279240e-01 -1.19251215e+00 -9.42767620e-01 5.21529257e-01 -1.24275424e-02 5.13781846e-01 -7.12059677e-01 -1.05007517e+00 5.49474120e-01 5.35899401e-01 2.61068523e-01 1.01626408e+00 1.15674272e-01 -5.72083473e-01 -3.96127462e-01 -1.12601888e+00 5.59769809e-01 1.15289080e+00 -8.82813454e-01 -9.30192888e-01 -2.39990354e-02 3.06386411e-01 -1.31074265e-01 -1.00387466e+00 6.05420887e-01 8.32359672e-01 -1.07279062e+00 5.50213397e-01 -9.29485440e-01 5.63025713e-01 -1.63698196e-03 -1.32339234e-02 -1.89485407e+00 -2.91210592e-01 -2.07572654e-01 4.63100970e-01 1.16940212e+00 2.97861427e-01 -1.18784773e+00 3.12719792e-01 6.86969161e-01 5.07353805e-03 -1.23855138e+00 -1.05693996e+00 -5.51601171e-01 5.30067682e-01 -1.92023769e-01 1.08330595e+00 1.13240874e+00 6.21704638e-01 2.69494891e-01 -1.83747739e-01 -1.42284846e-02 -9.75289643e-02 3.38788360e-01 1.98096454e-01 -1.03399265e+00 4.74770628e-02 -7.14140892e-01 -7.47742116e-01 -9.24996614e-01 8.84527802e-01 -1.14343822e+00 -7.60115497e-03 -9.40104008e-01 3.69787782e-01 -1.69376340e-02 -1.04153097e+00 3.74297172e-01 1.99092194e-01 1.65227070e-01 -3.92804027e-01 -1.39281511e-01 -3.80911916e-01 1.21251512e+00 1.33242905e+00 -2.20967427e-01 -8.58515874e-02 -2.80188203e-01 -1.01964176e+00 5.86766303e-01 8.53079557e-01 -2.73065895e-01 -4.26168621e-01 -3.44748825e-01 1.70919582e-01 -5.51327579e-02 4.12315756e-01 -9.31013763e-01 1.61346212e-01 2.30879799e-01 1.06650901e+00 -7.81380907e-02 7.43068904e-02 -9.62396860e-01 -1.80371806e-01 1.35570303e-01 -5.30608535e-01 2.08981827e-01 3.10710549e-01 3.63573760e-01 -5.27528644e-01 2.35236794e-01 6.41730309e-01 1.69144273e-01 -2.59445816e-01 5.72920918e-01 -2.21360773e-01 1.15094334e-01 1.18152344e+00 -1.29020423e-01 -1.57613590e-01 -1.38851538e-01 -6.24311090e-01 2.81809479e-01 1.38422772e-01 4.50666308e-01 7.02014744e-01 -1.65950882e+00 -6.99306548e-01 8.95402551e-01 1.98174715e-01 -4.79752362e-01 6.94398224e-01 1.07561052e+00 -5.76801524e-02 2.20911145e-01 -8.70639920e-01 -3.19786280e-01 -7.13288069e-01 4.69961673e-01 7.69895375e-01 1.32394284e-01 -5.18601716e-01 8.37325096e-01 6.77619040e-01 -7.71660626e-01 -3.79642006e-03 -5.87861054e-03 -3.66706491e-01 1.46443650e-01 5.10309637e-01 1.59818992e-01 1.41747907e-01 -5.06000638e-01 -6.43613219e-01 4.66807157e-01 1.57765541e-02 7.93389790e-03 1.49311340e+00 1.48642678e-02 -4.17944521e-01 3.22650790e-01 1.74733186e+00 -2.10065573e-01 -1.15936637e+00 2.94833183e-01 -5.53048074e-01 -2.27052361e-01 1.86738558e-02 -8.43164563e-01 -1.49540138e+00 1.34378672e+00 8.39369178e-01 -3.21179666e-02 1.67308414e+00 -1.82433173e-01 5.69025517e-01 2.16102079e-01 2.78234959e-01 -1.09160590e+00 -2.47887641e-01 2.68641412e-01 1.15785670e+00 -7.89162576e-01 -5.52075863e-01 1.60424545e-01 -9.51652288e-01 1.11856282e+00 7.45176673e-01 -3.19105208e-01 8.19696546e-01 2.07603112e-01 -7.48509020e-02 -2.68946350e-01 -5.23402750e-01 3.89345258e-01 4.48747993e-01 5.72708309e-01 4.66073900e-01 6.14418238e-02 -4.39461917e-01 1.48967886e+00 -1.78130776e-01 -8.41450393e-02 -1.49112314e-01 4.11363661e-01 8.41277987e-02 -8.06212962e-01 1.18536226e-01 5.46443403e-01 -2.83571571e-01 -1.97623283e-01 -5.41271627e-01 5.46711504e-01 1.08324699e-01 4.85253930e-01 2.74536163e-01 -7.54814863e-01 5.28613865e-01 4.05634224e-01 4.12930071e-01 3.93340876e-03 -5.90209067e-01 -3.15049082e-01 -5.08078575e-01 -6.31379724e-01 -1.79694191e-01 -3.28853428e-01 -1.29994452e+00 6.13025688e-02 -2.88968123e-02 2.11008742e-01 4.13007259e-01 9.79483783e-01 1.00507307e+00 7.89022446e-01 1.12957013e+00 -8.37031364e-01 -4.92288172e-01 -1.12110806e+00 -1.11226189e+00 6.39958680e-01 1.84702188e-01 -7.50895798e-01 -4.95855212e-01 -3.79957110e-01]
[13.12378215789795, 3.499345064163208]
c64a4cba-e27f-445c-a748-ae3d01024ae1
exploiting-neighborhood-structural-features
2302.05114
null
https://arxiv.org/abs/2302.05114v1
https://arxiv.org/pdf/2302.05114v1.pdf
Exploiting Neighborhood Structural Features for Change Detection
In this letter, a novel method for change detection is proposed using neighborhood structure correlation. Because structure features are insensitive to the intensity differences between bi-temporal images, we perform the correlation analysis on structure features rather than intensity information. First, we extract the structure feature maps by using multi-orientated gradient information. Then, the structure feature maps are used to obtain the Neighborhood Structural Correlation Image (NSCI), which can represent the context structure information. In addition, we introduce a measure named matching error which can be used to improve neighborhood information. Subsequently, a change detection model based on the random forest is constructed. The NSCI feature and matching error are used as the model inputs for training and prediction. Finally, the decision tree voting is used to produce the change detection result. To evaluate the performance of the proposed method, it was compared with three state-of-the-art change detection methods. The experimental results on two datasets demonstrated the effectiveness and robustness of the proposed method.
['Yuanxin Ye', 'Jianwei Fan', 'Ming Hao', 'Bai Zhu', 'Peizhen Yang', 'Zhiqiang Han', 'Mengmeng Wang']
2023-02-10
null
null
null
null
['change-detection']
['computer-vision']
[ 5.91772854e-01 -8.40877950e-01 -1.28045022e-01 -3.34106207e-01 -1.70359910e-01 -1.33061051e-01 3.85258228e-01 1.58904612e-01 -4.23178732e-01 5.53062141e-01 1.17470130e-01 -1.82234938e-03 -2.06749424e-01 -1.00653815e+00 -1.05389848e-01 -1.02152252e+00 1.28796488e-01 -4.35919821e-01 7.02382326e-01 -1.03530914e-01 5.74782968e-01 4.21585619e-01 -1.69608414e+00 1.80750042e-01 1.09406900e+00 1.22702849e+00 3.60156536e-01 2.09127650e-01 -8.52022246e-02 5.70905209e-01 -3.19913954e-01 4.77364898e-01 1.50360525e-01 -5.26388824e-01 -3.24444860e-01 1.32428870e-01 8.41689408e-02 -1.26530573e-01 9.48986933e-02 1.12821388e+00 4.31972802e-01 3.33355516e-01 5.44921577e-01 -7.22898841e-01 -2.26471841e-01 8.57940465e-02 -7.40650833e-01 6.97003901e-01 3.64283741e-01 1.64168701e-02 7.77400315e-01 -9.17471290e-01 8.20382118e-01 1.18835700e+00 4.11227584e-01 -1.58471540e-01 -9.07442093e-01 -6.70664549e-01 2.81787902e-01 8.06271970e-01 -1.35110092e+00 -1.02282017e-01 1.15702093e+00 -4.24403280e-01 4.54268485e-01 3.73050779e-01 8.88687968e-01 2.09708631e-01 3.83402616e-01 6.76509857e-01 1.60016572e+00 -5.15242279e-01 -3.07998713e-02 -3.99900787e-02 2.76134640e-01 8.25724840e-01 1.56450957e-01 2.06437543e-01 -1.16462946e-01 2.97268368e-02 5.36764562e-01 2.89893180e-01 -5.03030896e-01 -2.19275311e-01 -1.15822625e+00 4.99620289e-01 1.02665937e+00 7.43214846e-01 -4.65435803e-01 -3.99502397e-01 2.15152487e-01 1.71517834e-01 3.22427601e-01 -1.56234056e-01 -1.18573919e-01 9.76516679e-02 -7.68015802e-01 -2.04330295e-01 1.48920432e-01 1.23029411e-01 9.75746036e-01 -2.55213112e-01 -1.94043159e-01 8.04682672e-01 3.61899912e-01 5.92118621e-01 5.84426343e-01 -4.53987241e-01 3.58820081e-01 9.91982698e-01 -1.17389046e-01 -1.49418283e+00 -5.29435217e-01 -5.83449066e-01 -1.02295899e+00 2.72262484e-01 4.35941704e-02 2.64484644e-01 -8.83807003e-01 1.23431063e+00 6.09207153e-01 1.47928149e-01 -6.75854385e-02 9.64672744e-01 6.68085217e-01 7.64201343e-01 -2.56574959e-01 -6.38384163e-01 1.16888618e+00 -6.89995468e-01 -7.85962522e-01 1.83683515e-01 3.22550863e-01 -8.91944945e-01 9.00534451e-01 1.32627413e-01 -5.74760795e-01 -7.56474376e-01 -1.20814610e+00 3.77004504e-01 -3.11591804e-01 2.35410482e-01 2.75146514e-01 2.12455153e-01 -5.83288252e-01 4.71419841e-01 -9.01724517e-01 -5.27638316e-01 1.83499798e-01 -2.13650763e-02 -1.18294664e-01 -8.99318308e-02 -1.07372499e+00 7.65593171e-01 6.22548223e-01 2.97246337e-01 -3.01273316e-01 -2.17595488e-01 -6.36037827e-01 6.83435723e-02 1.94821730e-01 -2.77464122e-01 6.48975134e-01 -1.03460705e+00 -1.28716946e+00 5.42353809e-01 -4.10096139e-01 3.63116302e-02 2.90366799e-01 3.12779576e-01 -7.35315382e-01 2.72263825e-01 1.33330971e-01 1.91631675e-01 7.13769674e-01 -1.20198870e+00 -1.06631052e+00 -3.38887274e-01 -2.83838123e-01 3.44415069e-01 -1.86421871e-01 -2.93985512e-02 -2.65850216e-01 -5.73483467e-01 8.61167789e-01 -7.37373054e-01 1.95058305e-02 -1.85875818e-01 -2.23680735e-01 4.37674252e-03 1.24351406e+00 -9.03275967e-01 1.67456913e+00 -2.26840401e+00 -1.34928524e-01 8.63312542e-01 -2.07471609e-01 2.15541616e-01 1.48715470e-02 1.03390321e-01 -4.50679325e-02 4.77373004e-02 -4.92566377e-01 5.68554163e-01 -6.46922052e-01 -1.41593575e-01 2.66120464e-01 2.84531981e-01 8.67797658e-02 3.72719109e-01 -6.83949828e-01 -7.20257580e-01 3.78992230e-01 9.69540104e-02 -2.09092841e-01 -5.36039732e-02 1.77673936e-01 5.12033641e-01 -6.10274434e-01 6.76472366e-01 8.51802647e-01 1.18504249e-01 8.10986981e-02 -5.26285589e-01 -5.30816853e-01 5.30129261e-02 -1.36568856e+00 9.97796595e-01 -2.64382482e-01 3.43503445e-01 -2.92070985e-01 -9.71970439e-01 1.41860819e+00 -9.23299566e-02 5.26058853e-01 -8.40615511e-01 5.35754152e-02 2.49889418e-01 1.93835005e-01 -5.68442523e-01 1.88530773e-01 1.07566185e-01 2.20909536e-01 2.07265377e-01 -6.49801433e-01 3.76836769e-02 2.79920608e-01 -2.56923437e-01 9.69316006e-01 9.62102488e-02 5.67792058e-01 -2.02772096e-01 1.10511422e+00 7.10361227e-02 9.22691941e-01 2.71902472e-01 -2.75739968e-01 1.47585303e-01 4.01591845e-02 -5.74152589e-01 -6.30903244e-01 -9.53425109e-01 -2.73248613e-01 3.74337822e-01 6.26495719e-01 -1.55329838e-01 -5.44001579e-01 -7.22512126e-01 -2.18918055e-01 3.17516237e-01 -4.52075541e-01 -1.99020609e-01 -7.26575255e-01 -7.87045717e-01 -3.07479769e-01 3.72542411e-01 1.27230120e+00 -1.00967264e+00 -5.73767006e-01 2.99146712e-01 -4.07365054e-01 -5.41913092e-01 -4.20629174e-01 -3.57580632e-01 -1.13910949e+00 -1.04712999e+00 -3.21761727e-01 -1.08114135e+00 7.07097650e-01 7.18014181e-01 4.21167970e-01 3.51799786e-01 -3.54080588e-01 -3.77044715e-02 -4.38392609e-01 7.98701197e-02 -1.42302573e-01 -9.02577396e-03 -1.43187761e-01 3.03054392e-01 1.09326191e-01 -7.64920175e-01 -8.83910596e-01 4.33197588e-01 -7.81220734e-01 2.80020591e-02 9.03499186e-01 8.51483524e-01 9.82829094e-01 6.01063192e-01 4.06167001e-01 -5.41601837e-01 4.37814981e-01 3.30913253e-03 -6.41915798e-01 3.87468338e-01 -8.67003500e-01 5.17086014e-02 3.69962424e-01 -3.11353624e-01 -1.56397808e+00 5.64380661e-02 2.23323196e-01 9.54533294e-02 1.09142810e-01 9.43425238e-01 -1.16534412e-01 -7.96181560e-02 3.75053912e-01 4.48306084e-01 -9.30485353e-02 -4.78284448e-01 1.89678892e-02 8.17438960e-01 4.25065249e-01 -9.63472649e-02 8.18223655e-01 4.65438128e-01 1.27617851e-01 -7.23973691e-01 -5.31374753e-01 -5.29099107e-01 -7.54654467e-01 -4.69557017e-01 8.76219928e-01 -6.21047378e-01 -3.20128024e-01 7.87664533e-01 -8.70429575e-01 2.75034964e-01 1.27835602e-01 7.08111644e-01 -2.17761591e-01 5.05939722e-01 -4.16686982e-01 -7.54768074e-01 -4.83803689e-01 -7.79952466e-01 5.53869367e-01 5.94710588e-01 2.16646537e-01 -7.53413498e-01 9.01380405e-02 1.30155727e-01 3.39545548e-01 4.04652357e-01 1.12542331e+00 -2.10873649e-01 -7.89609313e-01 -1.77429199e-01 -3.17757785e-01 1.41037345e-01 5.70484281e-01 3.34147960e-01 -4.44025159e-01 -8.45827684e-02 1.63305998e-01 4.10219520e-01 9.60604250e-01 4.09230292e-01 9.16326940e-01 -6.32557720e-02 -6.08966827e-01 4.91985083e-01 1.54199922e+00 6.87627912e-01 7.32620835e-01 7.13265061e-01 4.95109677e-01 3.13276201e-01 1.20703554e+00 3.69971424e-01 1.74775451e-01 7.14364827e-01 1.28663972e-01 -2.25733459e-01 -1.71991721e-01 -2.47876376e-01 2.96294779e-01 1.15782857e+00 -2.84575522e-01 3.35519075e-01 -7.40353584e-01 2.54005790e-01 -1.82260895e+00 -1.27520239e+00 -3.93568039e-01 2.14740610e+00 7.32979774e-01 3.12988520e-01 -2.58679926e-01 3.97285491e-01 1.12548041e+00 2.55007744e-01 -4.64042008e-01 -2.38725431e-02 -9.72885415e-02 -4.60399725e-02 2.47286826e-01 3.26975733e-01 -1.19288766e+00 6.43692791e-01 5.65381336e+00 8.14246297e-01 -1.39728522e+00 -4.03985679e-02 3.20443153e-01 3.61717731e-01 -2.05174178e-01 1.65639624e-01 -5.28551579e-01 5.64665496e-01 7.65900612e-02 -6.92171231e-02 2.99299568e-01 5.40532231e-01 5.52031040e-01 -5.72803140e-01 -3.44189584e-01 8.55461717e-01 -8.01245496e-02 -8.32276821e-01 -8.00540745e-02 -2.24887654e-01 9.25223649e-01 -3.86887431e-01 -1.28561124e-01 -4.41723242e-02 -1.94335088e-01 -2.75552630e-01 2.69339412e-01 1.02974176e+00 4.45085227e-01 -6.62868202e-01 9.21515703e-01 2.07210124e-01 -1.66801333e+00 -2.11679772e-01 -1.95045024e-01 -1.38129324e-01 -1.00502953e-01 8.34994495e-01 -5.86803615e-01 8.25705051e-01 7.03545570e-01 9.23146307e-01 -7.12515771e-01 1.30927420e+00 -4.18097645e-01 5.77860951e-01 -4.29795533e-01 -1.38429254e-01 -5.94289973e-02 -5.35005629e-01 5.40901542e-01 9.29408491e-01 5.45010984e-01 5.32593206e-02 3.45861554e-01 5.94447017e-01 2.53387481e-01 5.76162517e-01 -2.84103692e-01 3.65188062e-01 6.06181145e-01 1.32257116e+00 -9.35699642e-01 -3.92218292e-01 -4.13129687e-01 9.48727727e-01 4.19249088e-02 2.87940413e-01 -5.97321153e-01 -6.47852361e-01 5.39098680e-02 3.23097920e-03 5.26543558e-01 -7.96684846e-02 -1.51555672e-01 -1.14525259e+00 3.55302811e-01 -6.16295755e-01 4.24033016e-01 -6.92922592e-01 -9.89004374e-01 3.23334426e-01 9.28575397e-02 -1.60836053e+00 9.15699899e-02 -2.29875445e-01 -1.07506776e+00 8.51822674e-01 -1.49280787e+00 -9.95975256e-01 -5.89979827e-01 4.95436251e-01 2.49790549e-01 1.06456280e-01 3.78736347e-01 7.37616345e-02 -7.98852384e-01 2.16462433e-01 5.44383466e-01 -2.82484964e-02 5.20311356e-01 -7.78774500e-01 -1.57445699e-01 1.20093799e+00 -2.08386555e-01 2.67319173e-01 2.75823742e-01 -8.25108886e-01 -7.94671774e-01 -9.43639994e-01 7.04235911e-01 4.34621811e-01 4.41055208e-01 2.91919917e-01 -9.88699496e-01 1.90111279e-01 -2.60428548e-01 6.85765222e-02 3.44220221e-01 -2.04800874e-01 -5.86989410e-02 -6.44312799e-01 -1.25395918e+00 4.85815734e-01 9.98094976e-01 -2.94239193e-01 -6.60501301e-01 -2.21664205e-01 4.87862825e-01 -1.13694318e-01 -8.95979822e-01 7.30399251e-01 7.01030731e-01 -9.78429914e-01 6.31717384e-01 1.86910197e-01 1.85556963e-01 -9.30579722e-01 -8.20547342e-02 -1.14827085e+00 -7.37690210e-01 9.16379169e-02 4.18138444e-01 1.47751510e+00 2.57321835e-01 -8.11848402e-01 2.89000869e-01 7.07232067e-03 9.55165997e-02 -6.40195429e-01 -8.49581540e-01 -7.01001942e-01 -5.41668832e-01 3.61568443e-02 6.50425196e-01 7.43173599e-01 -1.13130867e-01 2.47154757e-01 1.68162584e-02 1.64919317e-01 3.68880987e-01 4.94829774e-01 3.58683318e-01 -1.28639269e+00 5.98043017e-02 -3.58399421e-01 -5.82194865e-01 -6.11733317e-01 -2.07125962e-01 -8.30368638e-01 3.50803696e-02 -1.62430084e+00 3.72481316e-01 -4.71957535e-01 -7.89730310e-01 2.65093446e-01 -6.79434001e-01 9.14322361e-02 1.43102035e-01 5.64852715e-01 -1.85902312e-01 6.59110308e-01 1.54419863e+00 -3.50482643e-01 -5.73456526e-01 2.02900887e-01 -2.01934770e-01 6.14059389e-01 9.95981753e-01 -3.51013899e-01 -2.96708345e-01 1.62597597e-01 -3.30377460e-01 -1.06573462e-01 1.95772186e-01 -1.40595555e+00 -7.25440308e-02 -3.74202877e-01 5.90855658e-01 -7.61253417e-01 -7.64687583e-02 -9.27788138e-01 1.59310341e-01 8.94231260e-01 -1.15467668e-01 -4.80477884e-02 -1.28105044e-01 4.58846122e-01 -4.46725398e-01 -3.05640370e-01 7.96327651e-01 1.23547807e-01 -1.00694454e+00 1.42084703e-01 -2.58545756e-01 -5.16332328e-01 1.08448565e+00 -3.02246213e-01 -3.82130802e-01 -1.67121306e-01 -5.39233923e-01 2.89710034e-02 5.35803616e-01 1.84079081e-01 7.51322687e-01 -1.68187308e+00 -5.40266931e-01 2.74579912e-01 2.62731612e-01 -3.92707705e-01 4.06796545e-01 1.05324316e+00 -4.86304283e-01 -1.04806460e-02 -2.99391329e-01 -7.44000435e-01 -1.69117546e+00 6.50468707e-01 3.13639194e-01 -4.79090661e-01 -3.78771901e-01 8.42977762e-02 -9.11122039e-02 -9.95700583e-02 -2.81015933e-01 -5.14330447e-01 -5.76703072e-01 4.99765724e-02 4.16195124e-01 4.34838325e-01 3.41786910e-03 -5.94780087e-01 -5.49663067e-01 1.14427209e+00 4.17453572e-02 -1.79093435e-01 1.04495132e+00 -3.53867412e-01 -4.35615659e-01 3.99938792e-01 1.13554931e+00 2.75166053e-02 -9.93572652e-01 -6.01552784e-01 9.04142857e-02 -6.23280048e-01 1.67352125e-01 -7.97500968e-01 -1.03223777e+00 7.30171561e-01 1.37249076e+00 1.09024219e-01 1.65143216e+00 -3.66885006e-01 7.33616352e-01 2.90984213e-01 2.25076228e-01 -1.09040642e+00 -8.73087421e-02 2.61772186e-01 8.01011562e-01 -1.12146699e+00 1.93654999e-01 -6.12285614e-01 -3.73095721e-01 1.18972468e+00 6.46961212e-01 5.90313412e-02 7.51643181e-01 -1.73476756e-01 7.64724165e-02 1.08523250e-01 -2.51468658e-01 -5.49701393e-01 3.60875338e-01 5.60114563e-01 1.82241455e-01 6.01056740e-02 -9.95857716e-01 1.56716526e-01 4.95606363e-02 2.05158904e-01 1.09383740e-01 1.04051781e+00 -9.42784131e-01 -1.01750457e+00 -4.97120231e-01 5.05427539e-01 -5.46723641e-02 2.20989987e-01 -1.18150480e-01 5.17500401e-01 3.29877764e-01 1.14683819e+00 4.04277705e-02 -8.02402079e-01 2.95132667e-01 -6.63869902e-02 3.08091283e-01 -1.25335738e-01 -3.20535660e-01 5.37987612e-02 -1.66437387e-01 -3.87338340e-01 -8.05009067e-01 -7.22084105e-01 -1.41156399e+00 3.01253106e-02 -4.44332331e-01 1.43347949e-01 5.67802489e-01 8.71780634e-01 1.68336615e-01 5.04137218e-01 1.23719692e+00 -5.04895985e-01 -1.45509675e-01 -1.06079268e+00 -3.80981058e-01 5.79163074e-01 9.23786089e-02 -6.37946784e-01 -3.29134107e-01 -1.02637485e-02]
[9.991923332214355, -1.031110167503357]
620db457-de38-4a52-bf7d-9920336d67f7
transferable-deep-metric-learning-for
2302.06523
null
https://arxiv.org/abs/2302.06523v1
https://arxiv.org/pdf/2302.06523v1.pdf
Transferable Deep Metric Learning for Clustering
Clustering in high dimension spaces is a difficult task; the usual distance metrics may no longer be appropriate under the curse of dimensionality. Indeed, the choice of the metric is crucial, and it is highly dependent on the dataset characteristics. However a single metric could be used to correctly perform clustering on multiple datasets of different domains. We propose to do so, providing a framework for learning a transferable metric. We show that we can learn a metric on a labelled dataset, then apply it to cluster a different dataset, using an embedding space that characterises a desired clustering in the generic sense. We learn and test such metrics on several datasets of variable complexity (synthetic, MNIST, SVHN, omniglot) and achieve results competitive with the state-of-the-art while using only a small number of labelled training datasets and shallow networks.
['Jesse Read', 'Rim Kaddah', 'Simo Alami. C']
2023-02-13
null
null
null
null
['metric-learning', 'metric-learning']
['computer-vision', 'methodology']
[ 4.06255201e-02 -2.40113467e-01 3.35558467e-02 -5.79987884e-01 -6.66864693e-01 -8.60737026e-01 7.27418840e-01 4.69642341e-01 -7.83770561e-01 4.31162924e-01 4.41657426e-03 -7.59226270e-03 -6.60414577e-01 -6.66902721e-01 -2.81036139e-01 -8.29094410e-01 -1.48739889e-01 9.55244362e-01 3.75097215e-01 -5.20205460e-02 2.91076392e-01 4.34618503e-01 -1.50629592e+00 1.57527640e-01 5.42393446e-01 7.65959084e-01 2.31092304e-01 6.55096412e-01 6.44686669e-02 3.20676118e-01 -5.26476383e-01 -1.66811541e-01 3.94806027e-01 -4.69150662e-01 -9.42525029e-01 1.88602507e-01 1.04182877e-01 4.33376431e-01 -1.31912127e-01 8.45757365e-01 5.78085840e-01 4.95982468e-02 1.13649869e+00 -1.20340979e+00 -4.79603380e-01 5.98959684e-01 -1.76978692e-01 1.70421273e-01 1.58024967e-01 -8.66091549e-02 1.22521353e+00 -6.90990627e-01 6.61819398e-01 9.95726645e-01 7.05018759e-01 5.21027863e-01 -1.55178368e+00 -2.76872844e-01 -1.43439338e-01 1.32304296e-01 -1.44473767e+00 -2.25313872e-01 6.35990798e-01 -7.01906919e-01 6.04792297e-01 9.76762623e-02 3.70577097e-01 1.03521204e+00 -4.11636561e-01 4.17092830e-01 1.13297367e+00 -3.94495636e-01 5.63879371e-01 2.64643222e-01 2.34848708e-01 3.15164626e-01 2.37751693e-01 -2.92427628e-03 -1.21721521e-01 -5.23510025e-05 2.94364244e-01 -1.00601375e-01 -1.04673930e-01 -1.05920219e+00 -1.44650984e+00 1.04573107e+00 4.41757411e-01 7.79769123e-01 -1.10975243e-01 -4.21638004e-02 5.01419306e-01 7.37503529e-01 2.88414717e-01 9.46850181e-01 -6.84041798e-01 -3.16955745e-01 -8.60760093e-01 2.66893834e-01 1.03848255e+00 5.54087341e-01 7.82220840e-01 -4.85684365e-01 8.37726817e-02 8.72814357e-01 1.22515291e-01 7.52628967e-02 6.50353432e-01 -9.45061564e-01 2.51787305e-01 6.41302705e-01 1.14593888e-02 -9.73463178e-01 -8.27547133e-01 -3.36447716e-01 -9.10913467e-01 3.24829221e-01 7.48578727e-01 -2.62512654e-01 -7.30861485e-01 1.77841973e+00 3.21181089e-01 3.40978727e-02 8.27618986e-02 7.72747040e-01 4.01798189e-01 2.96643853e-01 -2.03971133e-01 1.48043960e-01 8.39627206e-01 -6.66782737e-01 -3.18999708e-01 -3.36309254e-01 1.05135465e+00 -5.24605572e-01 1.09754312e+00 5.72852254e-01 -6.73225105e-01 -3.68742615e-01 -1.17207289e+00 1.26338229e-01 -8.03376675e-01 -4.28045355e-02 5.01571238e-01 8.00782859e-01 -1.11425209e+00 9.36995149e-01 -6.82881236e-01 -7.53856301e-01 3.38231802e-01 6.49959087e-01 -5.10383606e-01 -9.68832970e-02 -9.60920215e-01 9.68784988e-01 8.24679732e-01 -2.22536147e-01 -7.65517592e-01 -3.94827425e-01 -5.96305668e-01 -7.63254985e-02 3.32689993e-02 -4.61334527e-01 8.74243855e-01 -6.58814549e-01 -1.21057975e+00 1.14356041e+00 4.97778922e-01 -5.14815152e-01 6.54876947e-01 2.02008963e-01 -3.32158685e-01 7.48679191e-02 -6.38059676e-02 6.73380613e-01 6.88977182e-01 -1.20433640e+00 -6.26025915e-01 -4.53848779e-01 1.28545061e-01 6.77288547e-02 -6.23753667e-01 -9.99638960e-02 -8.08260739e-02 -3.05787653e-01 7.66310766e-02 -1.04099798e+00 -3.68306220e-01 -1.08008645e-01 -3.38842779e-01 -2.68311441e-01 8.76980722e-01 -1.86585307e-01 1.12296247e+00 -2.31933522e+00 6.40137434e-01 2.35320583e-01 3.46491933e-01 2.25848198e-01 -1.11102320e-01 4.42693084e-01 -1.81434304e-01 1.92931458e-01 -6.14705145e-01 -3.39995742e-01 3.50749671e-01 2.60593295e-01 2.70792544e-01 7.89034307e-01 1.98291734e-01 6.38468444e-01 -1.02218997e+00 -3.67509335e-01 3.65994871e-01 4.26528633e-01 -5.63206911e-01 1.34981275e-01 2.56570522e-02 3.67944092e-01 -2.35240743e-01 1.03416644e-01 3.18645090e-01 -3.26430470e-01 2.67231822e-01 1.80181898e-02 1.56107008e-01 1.29091769e-01 -1.67508054e+00 1.95212150e+00 -5.10013044e-01 6.24036551e-01 -5.59677929e-02 -1.63230491e+00 1.04567719e+00 1.47828817e-01 6.86288476e-01 -3.76645267e-01 2.57272542e-01 3.49539697e-01 2.94914603e-01 -4.51586366e-01 5.74426167e-02 -1.50188997e-01 -3.54316086e-01 6.04722500e-01 2.72825152e-01 -4.97818273e-03 4.70185310e-01 -2.52799019e-02 1.39760613e+00 -3.01945210e-01 7.56112188e-02 -4.96492922e-01 5.45352399e-01 -1.45721972e-01 2.56375790e-01 4.94876891e-01 -1.31876215e-01 7.73969531e-01 7.45466530e-01 -5.08927345e-01 -1.20341516e+00 -1.03396308e+00 -3.98890883e-01 1.14371383e+00 -1.22353762e-01 -4.32389468e-01 -7.66895056e-01 -9.00362670e-01 2.36426573e-02 3.20581585e-01 -7.64491677e-01 -1.66916087e-01 -3.94331694e-01 -1.04363322e+00 4.47221547e-01 1.63435549e-01 1.75460905e-01 -9.78022695e-01 -5.60756743e-01 1.61538780e-01 2.56666839e-01 -1.09217608e+00 -5.80461696e-02 6.95858836e-01 -7.18428135e-01 -1.29499614e+00 -4.24990326e-01 -8.01852286e-01 5.56344271e-01 8.78909901e-02 1.25476158e+00 2.64335368e-02 -2.55265087e-01 1.46365911e-01 -3.27193677e-01 2.62274239e-02 -5.26600301e-01 5.96435606e-01 2.31387377e-01 -3.65850478e-02 7.58895934e-01 -7.49580145e-01 -4.18493271e-01 4.26224113e-01 -1.08864784e+00 -4.55714405e-01 4.42411095e-01 7.71075130e-01 2.25097030e-01 4.99936968e-01 6.29918218e-01 -1.15121090e+00 7.09499478e-01 -5.82098424e-01 -4.14613962e-01 -1.87626258e-02 -5.60667694e-01 2.54939467e-01 9.36800063e-01 -3.77485216e-01 -1.45664409e-01 2.40516126e-01 7.89667740e-02 -1.53482899e-01 -6.00889325e-01 4.13688004e-01 -2.35930264e-01 -1.19925752e-01 9.98283446e-01 -1.17483906e-01 -1.03238069e-01 -7.47375727e-01 6.66651785e-01 8.73987377e-01 3.06177109e-01 -4.67847586e-01 9.06404972e-01 4.62609529e-01 1.10961482e-01 -6.42864585e-01 -6.79840863e-01 -5.69727302e-01 -1.41612697e+00 1.95179626e-01 9.17317629e-01 -5.33979535e-01 -4.40268010e-01 1.30675718e-01 -8.43729317e-01 -3.70624334e-01 -5.11851311e-01 4.43863004e-01 -6.40763402e-01 1.79954797e-01 -3.09755743e-01 -2.19444990e-01 2.41697475e-01 -1.14049077e+00 8.68355930e-01 -2.08676234e-01 -3.51254553e-01 -1.36483085e+00 3.31441879e-01 2.95815002e-02 3.43758494e-01 6.33110464e-01 1.02590406e+00 -1.13631582e+00 -1.21106736e-01 -4.40292656e-01 -1.70941800e-01 4.76316631e-01 3.57156038e-01 3.96825634e-02 -9.68136370e-01 -3.68825644e-01 -8.40386972e-02 -3.75866592e-01 8.29913139e-01 6.35019094e-02 1.15326083e+00 -1.44120544e-01 -3.45281959e-01 7.13277817e-01 1.49464142e+00 -1.42867535e-01 3.86954606e-01 5.19746184e-01 7.86912680e-01 8.28185320e-01 2.97588319e-01 2.43480697e-01 4.38231707e-01 7.35229433e-01 3.52647305e-01 -2.93707661e-02 1.02453321e-01 2.25772709e-01 -2.09578276e-02 8.85337591e-01 2.90002912e-01 -4.10044752e-02 -1.27377856e+00 7.36584604e-01 -1.84843683e+00 -7.32585549e-01 -1.30520031e-01 2.33501339e+00 8.26086223e-01 3.11043918e-01 5.29972553e-01 7.05459237e-01 6.11540973e-01 2.10911289e-01 -4.05170649e-01 -3.40491354e-01 -7.28017092e-02 7.87026957e-02 3.39440107e-01 4.19455290e-01 -1.31055987e+00 6.92933261e-01 6.90210438e+00 6.31565809e-01 -1.04252279e+00 1.04328625e-01 4.24858570e-01 4.81853224e-02 -7.47490348e-03 -1.26762763e-01 -5.40740788e-01 5.98703206e-01 1.28104174e+00 -1.55321002e-01 5.03424644e-01 7.29488373e-01 -1.13253810e-01 2.46556416e-01 -1.60658824e+00 1.02918851e+00 -4.87011820e-02 -1.08686066e+00 -3.30452263e-01 1.34939641e-01 6.54598236e-01 2.76981413e-01 6.87773377e-02 2.18102485e-01 5.07480860e-01 -1.35250843e+00 1.41710356e-01 2.36767441e-01 5.89895666e-01 -8.87449026e-01 5.84013104e-01 4.01830435e-01 -8.20934296e-01 -1.61931202e-01 -4.85097885e-01 1.68485995e-02 -3.21616143e-01 6.23867035e-01 -7.63416827e-01 3.48134488e-01 5.84693849e-01 1.03434563e+00 -1.02850449e+00 1.14507699e+00 1.48540502e-02 3.19006503e-01 -4.77298886e-01 -1.21955648e-02 3.74546081e-01 -1.85067356e-01 1.75227702e-01 1.34478295e+00 3.31646979e-01 -3.76040548e-01 1.98362187e-01 5.44338465e-01 -1.52519420e-01 1.36469170e-01 -8.18450749e-01 -1.40539721e-01 3.95281732e-01 1.41653180e+00 -9.61350977e-01 -9.57719423e-03 -3.10556054e-01 9.40739155e-01 4.80007231e-01 1.33854151e-01 -5.37197292e-01 -4.95083511e-01 7.44353652e-01 4.77665961e-02 4.72248942e-01 -3.57252359e-01 -2.73985267e-01 -1.08685958e+00 5.67451958e-03 -9.04226661e-01 5.73036373e-01 -2.72754520e-01 -1.57175446e+00 6.19706988e-01 -1.25604689e-01 -1.39006054e+00 -3.89618129e-01 -7.94395208e-01 -3.30439329e-01 4.13594067e-01 -1.32152998e+00 -5.72032571e-01 -3.05037796e-01 5.74303806e-01 3.84992510e-02 -3.00464511e-01 1.01526570e+00 4.87263769e-01 -4.03527915e-01 3.03407699e-01 5.30466378e-01 1.82979167e-01 8.01684976e-01 -1.75588977e+00 2.61725664e-01 4.91089135e-01 4.82450843e-01 3.27182204e-01 7.62513876e-01 8.15747976e-02 -1.21435332e+00 -1.05712628e+00 8.27602983e-01 -9.17766035e-01 9.69001830e-01 -7.14116693e-01 -9.11348939e-01 4.54708338e-01 -1.26844626e-02 2.77411550e-01 1.00052476e+00 2.35175505e-01 -5.02506971e-01 -1.94319069e-01 -1.26392245e+00 2.98668116e-01 1.16146803e+00 -4.81417298e-01 -6.75031602e-01 6.65148795e-01 6.98114872e-01 -6.57735169e-02 -1.21467066e+00 1.45642847e-01 2.46637762e-01 -1.13769281e+00 9.53033984e-01 -7.44515777e-01 3.13352913e-01 -3.46458256e-01 -3.90730023e-01 -1.60066020e+00 -4.48310852e-01 -5.59685886e-01 4.02794406e-03 1.24832988e+00 4.69848335e-01 -4.54794049e-01 7.69866586e-01 3.88632178e-01 3.02919030e-01 -7.03216910e-01 -9.47529554e-01 -9.94900882e-01 4.31763440e-01 -4.34027255e-01 8.43792498e-01 1.24779379e+00 4.52694669e-02 6.17816329e-01 -9.06802192e-02 -1.82638038e-02 7.44903624e-01 -9.64045823e-02 8.88828099e-01 -1.89183056e+00 -1.31608501e-01 -6.85458422e-01 -9.37196970e-01 -5.48601925e-01 2.24462941e-01 -1.15036917e+00 -1.98442712e-01 -1.40111446e+00 -1.12684406e-01 -7.74492562e-01 -5.83519578e-01 2.32075155e-01 1.61963344e-01 3.36196601e-01 2.38171786e-01 1.43647015e-01 -8.65265787e-01 2.80479491e-01 6.82778120e-01 -9.89040583e-02 -1.93066850e-01 -1.24656722e-01 -7.88643181e-01 6.44100249e-01 8.15897703e-01 -7.81130433e-01 -3.75670403e-01 -4.88798350e-01 2.60702908e-01 -4.59752619e-01 1.82055920e-01 -1.38001668e+00 3.67029727e-01 1.32461399e-01 3.11573058e-01 -1.07423015e-01 2.04571411e-01 -1.14803863e+00 -1.44958152e-02 2.66890466e-01 -6.25346303e-01 5.81359975e-02 -1.70558065e-01 5.55092096e-01 -2.15322435e-01 -3.74708295e-01 1.00820923e+00 -2.34904103e-02 -5.63645184e-01 3.00977290e-01 6.36254996e-02 4.04564649e-01 1.18478394e+00 -1.80651039e-01 -1.03081249e-01 -3.17204185e-02 -8.13423038e-01 2.74081707e-01 8.58410776e-01 5.29117823e-01 2.53618121e-01 -1.43987966e+00 -6.88518047e-01 1.50130644e-01 2.64195055e-01 5.66757331e-03 -3.52062881e-01 5.49629867e-01 -5.05747497e-01 3.65722626e-01 -1.53686747e-01 -8.24327409e-01 -9.81082141e-01 7.51886368e-01 4.68675613e-01 -3.40207070e-01 -5.34717560e-01 5.07112682e-01 -1.87259942e-01 -9.86928821e-01 2.95326024e-01 -1.71691030e-01 -2.19061509e-01 4.03068930e-01 3.36728692e-01 2.42897481e-01 2.26168379e-01 -7.15111673e-01 -4.71278012e-01 7.53304482e-01 1.17880970e-01 -1.33033946e-01 1.69519615e+00 -1.49749666e-01 4.86562178e-02 7.06794500e-01 1.70742476e+00 -5.12860596e-01 -1.18701708e+00 -3.32613349e-01 5.46199024e-01 -4.21566606e-01 -1.36575818e-01 -4.40751463e-01 -1.04579270e+00 1.06008554e+00 6.30369782e-01 7.78503418e-01 9.10867333e-01 7.78218433e-02 4.45866674e-01 6.01113558e-01 3.26772511e-01 -1.29303801e+00 1.21889338e-01 4.34940398e-01 4.77025270e-01 -1.36103976e+00 -6.76565692e-02 1.63730070e-01 -4.90495473e-01 1.25097215e+00 1.40606940e-01 -5.12985706e-01 9.60981846e-01 1.16082892e-01 3.96719798e-02 -4.37442362e-01 -6.10742807e-01 -3.48790795e-01 1.79392323e-01 8.67108822e-01 4.63911235e-01 -3.80124599e-02 -1.45823345e-01 1.84148192e-01 -3.85729164e-01 -1.91766530e-01 4.82838720e-01 7.11616993e-01 -3.96326691e-01 -1.35386109e+00 -3.70435156e-02 5.32527924e-01 -4.77276593e-01 2.78174341e-01 -6.42275333e-01 8.83423805e-01 2.64935166e-01 9.49394345e-01 -1.39935806e-01 -5.28038800e-01 2.68714786e-01 1.85685992e-01 3.96205485e-01 -8.34946394e-01 -5.45550883e-01 -3.49179387e-01 -1.03704154e-01 -4.47960526e-01 -6.19628549e-01 -7.82785177e-01 -9.57462430e-01 -2.98929155e-01 7.88516402e-02 1.29195362e-01 7.65904665e-01 9.85023797e-01 2.80105233e-01 2.68904030e-01 9.58054900e-01 -7.87325740e-01 -4.71057296e-01 -8.48544836e-01 -8.07233691e-01 8.45119357e-01 4.19249594e-01 -6.58345878e-01 -7.94632554e-01 -9.09071490e-02]
[9.134471893310547, 3.0916271209716797]
d611fcf0-98da-41e6-8bc6-3d9a4e489068
batch-prompting-efficient-inference-with
2301.08721
null
https://arxiv.org/abs/2301.08721v1
https://arxiv.org/pdf/2301.08721v1.pdf
Batch Prompting: Efficient Inference with Large Language Model APIs
Performing inference on hundreds of thousands of samples with large language models (LLMs) can be computationally and financially costly. We propose batch prompting, a simple alternative prompting approach that enables the LLM to run inference in batches, instead of one sample at a time. Our method reduces both token and time costs while retaining downstream performance. We theoretically demonstrate that under a few-shot in-context learning setting, the inference costs decrease almost inverse linearly with the number of samples in each batch. We extensively validate the effectiveness of batch prompting on ten datasets across commonsense QA, arithmetic reasoning, and NLI/NLU: batch prompting significantly~(up to $5\times$ with six samples in batch) reduces the LLM (Codex) inference token and time costs while achieving better or comparable performance. Our analysis shows that the number of samples in each batch and the complexity of tasks affect its performance. Further, batch prompting can be applied across different LLMs and reasoning methods.
['Tao Yu', 'Jungo Kasai', 'Zhoujun Cheng']
2023-01-19
null
null
null
null
['arithmetic-reasoning']
['reasoning']
[ 2.04578191e-01 1.50512293e-01 -1.59198772e-02 -5.65687358e-01 -1.14691341e+00 -7.29798198e-01 7.04552889e-01 3.08885485e-01 -8.63251805e-01 6.36741579e-01 2.29829594e-01 -7.41031170e-01 -1.56183749e-01 -7.77439296e-01 -8.28371584e-01 -1.38791770e-01 2.51843363e-01 6.94562852e-01 -1.01025030e-03 5.14499843e-02 5.43022096e-01 1.92184106e-01 -1.26714599e+00 3.94679755e-01 1.30581057e+00 7.39986062e-01 4.78308052e-01 9.32968497e-01 -5.05004823e-01 1.52469277e+00 -8.11620951e-01 -6.40499890e-01 1.50201291e-01 -3.10078025e-01 -1.05616879e+00 -4.09646928e-01 7.46866405e-01 -5.43097794e-01 -1.51762635e-01 9.72926199e-01 6.33717477e-01 4.98279601e-01 5.61099946e-01 -1.06527913e+00 -4.27183837e-01 1.15407884e+00 -3.21751595e-01 4.40215826e-01 4.20253724e-01 3.94531548e-01 1.11123931e+00 -7.44503796e-01 5.89036942e-01 1.47547615e+00 6.95562541e-01 3.46083075e-01 -1.39917874e+00 -6.82490110e-01 1.88774765e-02 3.24430496e-01 -1.20003200e+00 -7.12693214e-01 1.75186232e-01 -3.73669416e-01 1.36155474e+00 1.67939916e-01 2.75159538e-01 9.62659836e-01 1.66048765e-01 7.16936946e-01 1.03406942e+00 -6.99601114e-01 5.91161311e-01 -1.30252495e-01 5.31798542e-01 8.48601937e-01 1.77859917e-01 -2.55536675e-01 -7.23586857e-01 -2.90953219e-01 3.93576056e-01 -1.38654292e-01 1.69368789e-01 4.94829535e-01 -1.13327587e+00 8.73656392e-01 6.67923912e-02 8.63714293e-02 -1.98676080e-01 5.00357091e-01 5.51549733e-01 3.82023573e-01 3.80411655e-01 7.18212724e-01 -5.29311657e-01 -5.28529406e-01 -1.05935323e+00 5.02150893e-01 1.06745601e+00 1.15009952e+00 5.62681973e-01 -1.46731853e-01 -7.31191456e-01 7.52482295e-01 -2.02204794e-01 4.80400562e-01 4.56400126e-01 -1.31068766e+00 8.64243627e-01 6.01638019e-01 4.28460300e-01 -6.54794097e-01 -4.19627547e-01 -4.98679131e-02 -5.41323721e-01 -3.78350198e-01 8.12029064e-01 -4.19052988e-01 -6.44696295e-01 1.81107450e+00 8.19707438e-02 1.24219812e-01 -1.25774324e-01 3.94795835e-01 6.06892049e-01 5.10332167e-01 4.53051358e-01 -1.16442710e-01 1.48294520e+00 -7.72048175e-01 -6.56260848e-01 -6.45901740e-01 1.05008709e+00 -7.26588130e-01 1.88268828e+00 3.93459976e-01 -1.34818649e+00 -2.95375347e-01 -5.41588366e-01 -5.36707044e-01 -1.81260899e-01 9.27360803e-02 9.26844835e-01 3.99180651e-01 -8.32697749e-01 7.43870795e-01 -6.81423426e-01 -9.72101018e-02 4.63004678e-01 6.04618751e-02 6.68766350e-02 -5.55611312e-01 -1.12580323e+00 1.07965076e+00 4.03778404e-01 -1.25369012e-01 -7.95188427e-01 -1.15393138e+00 -8.43195081e-01 5.89906871e-01 6.63024604e-01 -5.57373524e-01 1.62896669e+00 -2.86735654e-01 -1.32779646e+00 4.82243687e-01 -5.66314280e-01 -5.21372378e-01 6.37580872e-01 -5.26379406e-01 -5.00710234e-02 -5.37353419e-02 3.51358742e-01 6.15914226e-01 5.09556770e-01 -4.46728915e-01 -4.57158148e-01 -1.38774812e-01 4.14613485e-01 7.25202635e-02 -2.36786291e-01 2.16629222e-01 -8.61448124e-02 -2.54860520e-01 -2.53073603e-01 -6.35416925e-01 -1.71717212e-01 -3.09296221e-01 -3.14647108e-01 -5.67544997e-01 1.15411580e-01 -7.33421087e-01 1.16432309e+00 -2.05068588e+00 2.15946743e-03 -2.61831164e-01 1.24618702e-01 5.80391176e-02 -2.28960514e-01 2.23182693e-01 3.50321412e-01 2.86085814e-01 -1.26414031e-01 -3.14859182e-01 3.82821143e-01 3.62282246e-01 -4.16202873e-01 1.66721623e-02 1.68957323e-01 1.22839057e+00 -1.13298607e+00 -7.02280700e-01 4.51338030e-02 -1.72587588e-01 -9.01240289e-01 2.48647079e-01 -5.98476887e-01 2.84811333e-02 2.49833688e-02 4.90867078e-01 4.20722872e-01 -6.34469211e-01 3.08476150e-01 1.50646210e-01 2.09267423e-01 7.41104901e-01 -1.14173245e+00 1.71487439e+00 -7.25380003e-01 7.04781175e-01 -8.49405825e-02 -9.09364402e-01 4.33347374e-01 1.42247275e-01 -3.27403903e-01 -8.40802848e-01 -1.68241709e-01 1.26075044e-01 1.68395206e-01 -7.50352621e-01 5.41907787e-01 -3.06542754e-01 -3.43535483e-01 6.24283731e-01 9.49899480e-02 -2.46096462e-01 7.21148074e-01 5.63273072e-01 1.24612498e+00 -1.54264659e-01 3.43843341e-01 -2.81702280e-01 5.76789007e-02 -7.70225748e-03 6.82550132e-01 1.37097037e+00 1.49303069e-02 -2.29093686e-01 7.51204073e-01 -3.05542439e-01 -1.02784121e+00 -8.88731301e-01 1.39572084e-01 1.75876701e+00 -3.40364695e-01 -4.31409478e-01 -7.03445673e-01 -3.38076085e-01 2.30571389e-01 1.44959188e+00 -3.00527424e-01 2.03873757e-02 -5.10078847e-01 -4.91015255e-01 9.01038051e-01 8.09024751e-01 5.04157186e-01 -1.14240849e+00 -7.48827934e-01 2.13166833e-01 -4.32724833e-01 -1.32486117e+00 -2.61009574e-01 2.46016487e-01 -9.35119033e-01 -8.57543468e-01 -7.45298490e-02 -1.88036576e-01 5.82547307e-01 -1.93121508e-01 1.43752778e+00 -1.34367635e-02 -4.84230906e-01 1.88872799e-01 -1.75925985e-01 -6.91996038e-01 -4.10483360e-01 2.42521003e-01 7.62430504e-02 -7.61914551e-01 5.68707526e-01 -4.29342449e-01 -1.80401146e-01 -3.39149863e-01 -7.48213947e-01 6.01671673e-02 4.32145059e-01 7.84048319e-01 1.35759622e-01 -2.14440301e-01 6.48396730e-01 -1.17587245e+00 9.02200758e-01 -4.36255664e-01 -7.58623302e-01 5.28277695e-01 -6.07190430e-01 2.19337374e-01 7.20923424e-01 -4.40945089e-01 -1.19321418e+00 -4.34045672e-01 1.06061786e-01 -6.03951477e-02 -1.54726952e-02 6.70500636e-01 1.56314880e-01 2.93748170e-01 7.45721936e-01 7.48952851e-02 -2.41183400e-01 -4.44075882e-01 6.39011800e-01 4.91647691e-01 3.54991108e-01 -9.94753718e-01 4.25456822e-01 4.64924052e-02 -1.65589079e-01 -5.72193861e-01 -1.24587512e+00 -8.08161348e-02 -4.32348341e-01 1.13074295e-01 6.51615202e-01 -9.15518224e-01 -1.10263813e+00 1.53892040e-01 -1.15199101e+00 -9.56920385e-01 -3.17473233e-01 3.96115720e-01 -4.43170071e-01 2.78771400e-01 -9.01025176e-01 -8.00934434e-01 -6.73835933e-01 -9.70825016e-01 7.40400136e-01 2.40248203e-01 -7.01148927e-01 -9.42919433e-01 -4.01157707e-01 5.16293585e-01 4.38709676e-01 -3.60852540e-01 1.38229096e+00 -8.21738660e-01 -5.33672810e-01 -1.95532545e-01 -3.72029960e-01 1.49435878e-01 -2.14802474e-01 -3.64752322e-01 -8.69267881e-01 1.80344470e-02 -4.34784070e-02 -6.83349788e-01 6.60900772e-01 2.93824375e-01 1.33293140e+00 -5.81989229e-01 -5.26448674e-02 4.03146565e-01 1.37657154e+00 -5.62472492e-02 5.53308904e-01 2.65679695e-02 5.10235369e-01 2.40372226e-01 7.93462217e-01 5.54051518e-01 2.27823883e-01 1.03381924e-01 -1.94728166e-01 4.34333056e-01 2.31978580e-01 -2.55803376e-01 4.76801217e-01 6.25610828e-01 4.85281907e-02 -9.25602764e-02 -1.19897127e+00 5.08668005e-01 -1.60751760e+00 -1.12469316e+00 1.74404562e-01 1.97464156e+00 1.26617134e+00 2.18579769e-01 -1.81206957e-01 -7.04175383e-02 1.94139123e-01 -8.91048983e-02 -7.03967571e-01 -6.35165513e-01 2.41814226e-01 5.10512710e-01 2.90269643e-01 8.60998631e-01 -4.03463989e-01 1.12280297e+00 7.16110039e+00 9.56636608e-01 -8.08961570e-01 3.13377619e-01 6.03951693e-01 -7.00120211e-01 -4.97635782e-01 3.94730791e-02 -9.15475667e-01 5.66271126e-01 1.44692504e+00 -3.39658707e-01 8.29849243e-01 9.86966431e-01 1.24852784e-01 -5.53538263e-01 -1.52587509e+00 8.39702964e-01 -9.51207429e-02 -1.66260111e+00 -3.83274667e-02 -2.03242570e-01 6.02540255e-01 2.46155441e-01 -5.00631273e-01 8.32976103e-01 7.73818791e-01 -1.15457642e+00 6.66185558e-01 5.04855454e-01 8.29872668e-01 -6.61733031e-01 5.79910994e-01 9.19678092e-01 -8.14521253e-01 -3.69178861e-01 -4.50703502e-01 -7.64770687e-01 9.94881913e-02 7.77654886e-01 -1.02211571e+00 1.87608317e-01 4.83872920e-01 2.00778991e-02 -5.91292262e-01 4.56824958e-01 -3.22138041e-01 8.32715988e-01 -3.17675442e-01 -3.86208981e-01 1.00744322e-01 -1.13517575e-01 1.53105687e-02 1.61756158e+00 1.04840295e-02 4.83527571e-01 3.61247540e-01 1.10303879e+00 -2.14910179e-01 -1.46552533e-01 -3.22609484e-01 -4.15254235e-01 1.12075269e+00 9.89180744e-01 -5.49048781e-01 -1.03937244e+00 -1.51428670e-01 8.32991183e-01 7.96842396e-01 3.67069662e-01 -7.83724010e-01 -5.46105802e-01 3.55141729e-01 -4.44549248e-02 6.59484118e-02 -3.40257317e-01 -6.71521306e-01 -1.12745273e+00 3.37566598e-03 -8.76383364e-01 4.06273037e-01 -8.38286877e-01 -1.41997087e+00 -9.07333642e-02 1.94663659e-01 -2.19637260e-01 -4.57779586e-01 -5.76037407e-01 -6.16579473e-01 1.03149140e+00 -1.30234146e+00 -5.96672714e-01 -3.50995004e-01 2.40324080e-01 5.94679058e-01 1.33435935e-01 8.65484536e-01 2.12464944e-01 -7.05394864e-01 7.02302516e-01 5.95862605e-02 2.52274752e-01 6.13655984e-01 -1.38172901e+00 6.16809845e-01 7.24891424e-01 -1.20163642e-01 1.23448873e+00 5.65504253e-01 -6.76318526e-01 -1.41129339e+00 -9.09702778e-01 1.33919501e+00 -6.45007312e-01 6.71867132e-01 -5.54705620e-01 -7.43443131e-01 1.10503471e+00 1.18139327e-01 -1.79389060e-01 8.15317571e-01 6.17649496e-01 -4.40659046e-01 -8.03486109e-02 -1.24832714e+00 7.70486355e-01 1.12583399e+00 -9.23623264e-01 -7.80336678e-01 7.09533393e-01 9.65897799e-01 -5.68958819e-01 -9.57363605e-01 1.49656050e-02 3.84807557e-01 -6.93240166e-01 8.33691657e-01 -8.25644791e-01 7.30773330e-01 1.61757484e-01 -2.51988292e-01 -9.26471353e-01 -2.25558043e-01 -5.56412578e-01 -3.44707936e-01 1.14092410e+00 3.20657074e-01 -7.11909771e-01 3.19997281e-01 1.11048222e+00 1.08507365e-01 -6.89069211e-01 -8.19806695e-01 -8.32311809e-01 2.38052636e-01 -7.54970908e-01 6.03737652e-01 8.69143665e-01 2.78823584e-01 6.36301041e-01 5.43852225e-02 -2.02679455e-01 4.90065008e-01 1.78278312e-01 7.25180209e-01 -1.08866644e+00 -5.57312191e-01 -1.60066545e-01 2.57406652e-01 -9.40295517e-01 2.28451043e-01 -1.11336613e+00 6.51084036e-02 -1.54714978e+00 4.56484199e-01 -5.65199196e-01 -3.84434201e-02 7.72050321e-01 -6.44141197e-01 -3.09295744e-01 3.37035894e-01 -1.09148122e-01 -5.02098322e-01 4.91743088e-02 6.66522384e-01 5.94254807e-02 -6.70146942e-02 -3.65272820e-01 -7.87495255e-01 8.59980941e-01 6.38634920e-01 -4.22287524e-01 -5.00593424e-01 -7.05586374e-01 5.00202000e-01 1.19214445e-01 3.99227440e-01 -7.69461751e-01 2.99160749e-01 -4.00853068e-01 4.11959797e-01 -5.04548609e-01 1.33512616e-01 -1.57234535e-01 -3.45905334e-01 5.29134452e-01 -7.73121119e-01 1.46063134e-01 3.45278382e-01 4.33315694e-01 4.27713215e-01 -6.67394638e-01 6.68099582e-01 -6.50917888e-01 -6.71557307e-01 -2.58965194e-01 -3.19625258e-01 5.63844562e-01 7.08873808e-01 1.98358178e-01 -4.31514174e-01 -6.72914013e-02 -4.69286323e-01 2.45516896e-01 9.92696658e-02 -1.31459042e-01 2.73018807e-01 -8.50169957e-01 -5.52210212e-01 -1.35897294e-01 -1.31094053e-01 3.47185194e-01 2.04990759e-01 7.94068038e-01 -4.02921885e-01 6.43890440e-01 1.63563296e-01 -4.82973546e-01 -1.00690341e+00 4.05950099e-01 8.88040289e-03 -4.21868265e-01 -6.14086747e-01 1.25897658e+00 -2.84784019e-01 -6.06515050e-01 3.33328635e-01 -6.68089330e-01 3.57108504e-01 1.51750371e-01 6.83436334e-01 7.14890003e-01 -1.13755148e-02 4.84674126e-01 -2.60097176e-01 -3.39519605e-02 -3.28020304e-01 -3.15000057e-01 1.19266784e+00 2.25887984e-01 -3.51696044e-01 7.13563740e-01 8.01121473e-01 -7.46742636e-02 -9.50065613e-01 -3.46989363e-01 2.63019681e-01 -5.16708434e-01 -7.67393336e-02 -1.12992787e+00 -2.20958471e-01 7.96357095e-01 -3.21540125e-02 -1.99411258e-01 7.25071192e-01 -9.73723158e-02 7.43366957e-01 9.18614924e-01 4.62827563e-01 -1.22372186e+00 1.30220577e-01 7.34985292e-01 5.03612936e-01 -1.18259883e+00 1.95763692e-01 -8.54147375e-02 -5.22994518e-01 7.36575723e-01 6.13607824e-01 1.11371078e-01 9.35327262e-02 4.70047265e-01 -3.07228178e-01 -2.15236992e-01 -1.03704274e+00 3.41639906e-01 -2.45493054e-01 2.65255064e-01 6.82320714e-01 3.18258941e-01 -2.61377901e-01 6.43177509e-01 -4.40426111e-01 3.36698174e-01 4.86183316e-01 9.66697514e-01 -5.98557115e-01 -7.49814928e-01 -1.93661302e-01 9.09048617e-01 -3.08574051e-01 -6.37654960e-01 -9.53679606e-02 5.01690507e-01 7.05444366e-02 1.05264938e+00 3.80683601e-01 2.24176496e-02 7.03939795e-02 6.79378748e-01 6.74665630e-01 -8.87642741e-01 -6.15383565e-01 -5.31156778e-01 3.49571675e-01 -6.29121721e-01 1.96600854e-01 -6.09883845e-01 -1.47368300e+00 -8.27816963e-01 -3.43694627e-01 -2.55847629e-02 4.22824413e-01 1.39562094e+00 5.50611258e-01 4.88546312e-01 2.55992785e-02 -5.03528833e-01 -1.27744591e+00 -1.45075321e+00 -2.75420696e-01 2.42328599e-01 4.07785662e-02 -3.46136034e-01 -2.52392739e-01 1.33415803e-01]
[9.77253246307373, 7.4462714195251465]
9c198a98-2454-4b2b-8c14-49466992d6bc
feature-extraction-of-text-for-deep-learning
2010.05496
null
https://arxiv.org/abs/2010.05496v2
https://arxiv.org/pdf/2010.05496v2.pdf
Feature Extraction of Text for Deep Learning Algorithms: Application on Fake News Detection
Feature extraction is an important process of machine learning and deep learning, as the process make algorithms function more efficiently, and also accurate. In natural language processing used in deception detection such as fake news detection, several ways of feature extraction in statistical aspect had been introduced (e.g. N-gram). In this research, it will be shown that by using deep learning algorithms and alphabet frequencies of the original text of a news without any information about the sequence of the alphabet can actually be used to classify fake news and trustworthy ones in high accuracy (85\%). As this pre-processing method makes the data notably compact but also include the feature that is needed for the classifier, it seems that alphabet frequencies contains some useful features for understanding complex context or meaning of the original text.
['HyeonJun Kim']
2020-10-12
null
null
null
null
['deception-detection']
['miscellaneous']
[-2.29942888e-01 -1.05023518e-01 -2.40909934e-01 -4.61932868e-01 -2.34359317e-02 -5.06202638e-01 8.61175179e-01 5.11546969e-01 -4.10284877e-01 7.54876018e-01 3.39315236e-01 -4.65026051e-01 1.76955894e-01 -1.18645215e+00 -6.66593075e-01 -6.05272055e-01 -2.94162072e-02 2.04395384e-01 -1.67448252e-01 -5.89304209e-01 8.66286397e-01 6.21456683e-01 -1.63647485e+00 9.83984709e-01 4.91590708e-01 1.04957998e+00 -3.84920090e-01 3.76384884e-01 -5.07197857e-01 1.07338500e+00 -1.21086597e+00 -6.16989434e-01 3.94884832e-02 -4.59355086e-01 -8.87760282e-01 -3.11444908e-01 1.50054082e-01 -5.75704932e-01 -6.85195088e-01 1.16895092e+00 -4.03949134e-02 -3.23638022e-01 9.49344218e-01 -9.60138023e-01 -8.63367736e-01 8.69587839e-01 -2.51075178e-01 4.36312765e-01 4.34794605e-01 -1.09823249e-01 6.87401593e-01 -7.20851660e-01 4.63433951e-01 1.22140765e+00 7.15051472e-01 9.65548083e-02 -3.57862175e-01 -7.60983348e-01 -4.97449905e-01 5.45953870e-01 -9.75219250e-01 -2.73817837e-01 8.22587013e-01 -6.12312973e-01 7.81491518e-01 3.52677464e-01 9.27305341e-01 1.35153830e+00 9.56194162e-01 9.40663278e-01 1.31573880e+00 -6.12463713e-01 7.03902682e-03 4.59936976e-01 6.88059211e-01 7.55157113e-01 5.20417750e-01 3.75410616e-01 -4.41925466e-01 -3.55960965e-01 2.40660936e-01 1.18964918e-01 6.85604895e-03 3.93001705e-01 -9.50083792e-01 1.44233882e+00 3.32762897e-01 9.03836966e-01 -1.85221329e-01 6.06342629e-02 8.18220615e-01 6.79113686e-01 4.96730536e-01 7.86774874e-01 -3.82080555e-01 -2.95209587e-01 -1.03238595e+00 1.76278666e-01 9.32936490e-01 4.84902322e-01 4.86203045e-01 2.69432306e-01 -1.86749980e-01 2.63326406e-01 -5.22372946e-02 6.26639664e-01 1.00379646e+00 -1.57898322e-01 1.32941350e-01 6.31964684e-01 1.03967153e-02 -1.73733938e+00 -5.98576069e-01 -5.60386777e-01 -8.86756361e-01 2.61056125e-02 5.47007203e-01 -7.92208835e-02 -8.43208909e-01 9.26542521e-01 1.38505530e-02 -2.92819917e-01 -2.13557668e-03 8.22402954e-01 1.06703627e+00 9.31181073e-01 -4.06770378e-01 -1.69798359e-01 1.57535076e+00 -4.85428691e-01 -1.18288958e+00 -1.12511314e-01 8.52344811e-01 -8.67478848e-01 8.25974703e-01 6.91510320e-01 -2.73486644e-01 -3.56868595e-01 -1.32411695e+00 -1.67212486e-01 -1.01075530e+00 3.26260418e-01 1.10612381e+00 8.62966716e-01 -3.98968279e-01 8.47700596e-01 -2.03627512e-01 7.25634396e-02 6.13742828e-01 7.54958987e-02 -4.56577152e-01 7.67026618e-02 -1.61600649e+00 1.32448888e+00 7.51227975e-01 6.29936233e-02 -6.03347659e-01 3.57472226e-02 -7.58719087e-01 2.09559664e-01 8.21602792e-02 -2.09093392e-01 8.99236679e-01 -1.45897913e+00 -1.31265736e+00 7.68970251e-01 8.11883807e-02 -7.92247534e-01 5.15287876e-01 -4.19714510e-01 -5.80973625e-01 1.99403495e-01 -2.67029613e-01 -2.70540565e-01 1.48844898e+00 -7.86484778e-01 -4.35941696e-01 -4.99170125e-01 -2.46150807e-01 -3.94079953e-01 -4.69364196e-01 7.78964236e-02 6.04553699e-01 -8.66663933e-01 2.06476554e-01 -4.78088319e-01 4.90842521e-01 -2.20982999e-01 -3.70456427e-01 -3.74682277e-01 1.11578822e+00 -1.15798891e+00 1.18212557e+00 -2.16468906e+00 -4.10312593e-01 3.13102990e-01 4.28167373e-01 5.80610991e-01 3.68648678e-01 6.26057863e-01 6.67809770e-02 2.44311988e-01 -4.86290678e-02 3.01857919e-01 -1.58263147e-01 7.77822956e-02 -5.65067291e-01 9.95718718e-01 -9.45050269e-03 8.56774747e-01 -8.04659128e-01 -3.57661724e-01 1.33196503e-01 2.52968550e-01 -7.88790956e-02 -1.42166108e-01 -5.30337403e-03 1.46343589e-01 -5.69950998e-01 4.65337634e-01 5.45080483e-01 2.02847630e-01 -1.69541657e-01 -1.60069942e-01 1.22740902e-01 5.01531303e-01 -5.33682168e-01 8.35927367e-01 -3.27316642e-01 1.36344218e+00 -4.65865344e-01 -1.41011822e+00 1.07630467e+00 1.83777332e-01 1.94150489e-02 -7.39954591e-01 6.66100979e-01 2.72852421e-01 1.32472709e-01 -8.58966112e-01 7.23897815e-01 -1.56679928e-01 -2.18032926e-01 5.15797853e-01 2.50359923e-02 -1.87808573e-01 -1.20140634e-01 2.55745705e-02 6.81772530e-01 -4.46158469e-01 6.26359463e-01 -2.65010715e-01 4.77040023e-01 1.54074028e-01 1.57692984e-01 8.49490106e-01 1.47115383e-02 6.68198168e-02 6.54753327e-01 -1.05501246e+00 -1.30498052e+00 -1.60811096e-01 -3.50789398e-01 5.91652870e-01 -1.24419719e-01 -1.43763959e-01 -6.22290611e-01 -7.75634587e-01 1.62286520e-01 1.03584087e+00 -8.07733178e-01 -5.99023104e-01 -4.19367701e-01 -8.38971853e-01 8.52213085e-01 -8.87913853e-02 7.25841522e-01 -1.11369097e+00 -4.79559362e-01 1.81577533e-01 3.44978198e-02 -9.24114943e-01 8.35690051e-02 1.82968020e-01 -9.14687634e-01 -1.12922883e+00 -1.21701010e-01 -6.62463725e-01 3.80490392e-01 1.12834975e-01 8.10583591e-01 4.37444210e-01 -2.14149103e-01 -5.26933253e-01 -9.87483740e-01 -5.89795470e-01 -9.91483092e-01 -1.31832659e-01 6.82747141e-02 1.04331588e-02 6.63115501e-01 -2.92821258e-01 6.12652395e-03 -3.70812625e-01 -1.01072752e+00 -1.51571468e-01 7.59374917e-01 1.11340487e+00 -2.31095254e-01 3.28676611e-01 5.75593293e-01 -1.03088176e+00 9.27832127e-01 -4.92032379e-01 -4.12660390e-01 6.80900272e-03 -5.01253486e-01 2.07939282e-01 1.11120832e+00 -2.74402440e-01 -4.89347398e-01 -4.80331004e-01 -2.94812828e-01 1.52620733e-01 -1.96160704e-01 5.45191646e-01 8.07432756e-02 -1.28291309e-01 8.70935619e-01 8.98057699e-01 1.81738943e-01 -3.36045325e-01 1.24275662e-01 1.22886264e+00 -1.53446011e-02 1.35266045e-02 6.47310257e-01 3.37716192e-01 -1.11466251e-01 -1.11768222e+00 -8.72838140e-01 -7.78115392e-02 -5.84269345e-01 4.03503664e-02 4.66479719e-01 -3.46316159e-01 -7.26604939e-01 7.48055696e-01 -1.44649839e+00 5.11482298e-01 1.79895945e-02 5.75197160e-01 -6.05671704e-02 5.99328279e-01 -7.47303963e-01 -9.80602264e-01 -3.76722425e-01 -8.26110065e-01 7.60446727e-01 -3.11636156e-03 -1.14548616e-01 -8.03479552e-01 -4.06833678e-01 4.28584933e-01 3.17728281e-01 3.74372244e-01 1.15962934e+00 -1.32622659e+00 3.36091742e-02 -8.83978844e-01 -1.19289070e-01 5.23396492e-01 7.35132769e-02 1.05359636e-01 -1.00033283e+00 -7.40774050e-02 6.41507387e-01 -1.66354731e-01 9.99033272e-01 9.02071074e-02 1.23083484e+00 -1.06365585e+00 -1.60006851e-01 4.37101722e-01 1.08706605e+00 2.36076027e-01 8.44779372e-01 5.57194233e-01 4.54935223e-01 4.29845899e-01 6.06795490e-01 4.60880339e-01 -1.71013042e-01 4.07337397e-01 3.70778054e-01 3.23429972e-01 2.84483314e-01 -1.05863266e-01 5.24417996e-01 8.50657463e-01 2.49331415e-01 -3.16701740e-01 -7.97067344e-01 2.05373660e-01 -1.36455464e+00 -1.36859047e+00 -5.09525597e-01 1.86910808e+00 8.16517115e-01 3.53245080e-01 -5.77243045e-02 7.41386056e-01 5.81587791e-01 2.28212327e-01 -4.05069143e-02 -1.01927710e+00 -3.25165927e-01 8.01977143e-02 4.89965439e-01 4.15617853e-01 -1.24548709e+00 9.88179207e-01 6.39631224e+00 1.19237077e+00 -1.39885747e+00 4.52283099e-02 6.99906468e-01 2.68151939e-01 -2.26764321e-01 -3.24151874e-01 -4.51247454e-01 7.27569938e-01 1.02392948e+00 1.46005610e-02 2.98997283e-01 9.14423168e-01 2.65401185e-01 -3.56532604e-01 -8.47545803e-01 1.00827980e+00 3.41515571e-01 -1.56402123e+00 4.17705327e-01 1.71534158e-02 2.98918545e-01 -3.47987324e-01 -9.41647217e-02 3.80080491e-01 -2.39915222e-01 -1.34730554e+00 8.22546661e-01 5.46092212e-01 3.22188914e-01 -1.03724658e+00 1.60064924e+00 8.52332950e-01 -7.41248354e-02 -2.38812029e-01 -5.84824443e-01 -5.92442095e-01 -2.83497959e-01 1.18806171e+00 -1.12592614e+00 5.14271319e-01 4.10607845e-01 6.37113333e-01 -6.17755473e-01 7.53140807e-01 -2.24475786e-01 7.76298285e-01 -1.08649001e-01 -8.84459794e-01 4.45926785e-01 -1.22889981e-01 6.30509853e-01 1.33470690e+00 1.46447405e-01 -4.71880846e-02 -2.10971802e-01 6.59843683e-01 1.15369000e-02 2.42909297e-01 -9.83153760e-01 -4.83279467e-01 2.48885214e-01 8.32038462e-01 -7.69231915e-01 -5.74551821e-01 -7.73758953e-03 9.12757754e-01 5.15026301e-02 -1.17255606e-01 -8.85848641e-01 -5.98021507e-01 3.35693270e-01 1.56862289e-01 5.24163991e-02 -3.07349652e-01 -6.71309650e-01 -1.06285107e+00 2.76639294e-02 -1.11406326e+00 -6.80385381e-02 -4.37572449e-01 -1.19633305e+00 8.28519881e-01 -3.01247746e-01 -1.19303644e+00 -1.65303543e-01 -8.75263155e-01 -3.49314183e-01 4.68128651e-01 -1.08440185e+00 -1.01070142e+00 8.37456658e-02 4.02831107e-01 4.70045745e-01 -6.12916172e-01 7.82860100e-01 5.06383218e-02 -2.06459180e-01 6.51950479e-01 5.93968093e-01 7.34020948e-01 3.00352305e-01 -8.49396944e-01 1.80147588e-01 6.59498036e-01 4.59743977e-01 5.78491211e-01 9.56471622e-01 -7.27466226e-01 -1.50289273e+00 -4.71870482e-01 1.18547702e+00 -3.18004638e-01 6.54942989e-01 -3.80639076e-01 -7.58967221e-01 2.68766761e-01 8.92438963e-02 -2.38076895e-01 5.65712988e-01 -1.22213162e-01 -4.55396324e-01 7.74907768e-02 -1.40450597e+00 2.34050125e-01 3.59778821e-01 -6.16018474e-01 -1.17220235e+00 7.10319281e-01 5.36737442e-01 -2.99150914e-01 -3.36072296e-01 -7.81902596e-02 6.85520053e-01 -1.14031196e+00 5.67302763e-01 -9.85141754e-01 6.35849118e-01 1.12530433e-01 1.67879507e-01 -1.27962935e+00 -2.11008951e-01 -2.90955096e-01 -3.67170006e-01 7.38196731e-01 3.67153674e-01 -8.69218707e-01 4.45700288e-01 -9.53901932e-02 -1.07204013e-01 -5.44303775e-01 -1.00866413e+00 -7.44293094e-01 2.02299669e-01 -3.44808370e-01 4.66516078e-01 1.39179778e+00 2.24853605e-01 1.11734241e-01 -7.12547004e-01 -1.38591677e-01 1.52526245e-01 2.04773769e-01 5.34457743e-01 -1.35563982e+00 7.88773298e-02 -4.55082893e-01 -8.34033787e-01 -6.68519735e-01 2.75934994e-01 -8.44652355e-01 -3.09298426e-01 -1.05086052e+00 2.85117663e-02 -8.85736868e-02 2.29134500e-01 3.32003921e-01 1.04087852e-01 -5.91659860e-04 -1.74121141e-01 2.43338346e-01 1.22671798e-01 5.77388287e-01 1.31809068e+00 -4.01196152e-01 2.19965547e-01 1.93369433e-01 -4.98421133e-01 8.49494815e-01 9.16931629e-01 -9.33699787e-01 1.48076639e-01 -1.52228057e-01 3.49931926e-01 -1.00300796e-01 3.50342631e-01 -8.07604373e-01 -3.72383334e-02 -7.29454160e-02 7.27563560e-01 -5.02342403e-01 1.80454463e-01 -8.97742927e-01 -4.18209553e-01 8.54606569e-01 -2.56922334e-01 -3.65621485e-02 9.42329839e-02 3.97046387e-01 -4.56091791e-01 -7.58381128e-01 6.51465297e-01 -3.12044889e-01 -5.03657997e-01 -1.50913447e-01 -8.81229043e-01 -2.15696365e-01 8.44963849e-01 -3.35905820e-01 -5.11387944e-01 -6.84044659e-01 -3.90284091e-01 -5.15355706e-01 1.67699963e-01 4.35717732e-01 7.61945903e-01 -1.07866633e+00 -8.25562656e-01 2.00101793e-01 -6.63872659e-02 -5.68773508e-01 -7.48606101e-02 6.32144511e-01 -1.15319574e+00 5.69936633e-01 -3.94001126e-01 -1.54192701e-01 -1.12662947e+00 6.50855124e-01 1.37572378e-01 -1.71285030e-02 -7.10329056e-01 5.99302471e-01 -5.54295659e-01 -2.67552976e-02 2.57368619e-03 -4.96912211e-01 -5.46550632e-01 2.83175230e-01 9.01383221e-01 2.33053803e-01 2.90941954e-01 -7.63471067e-01 -2.48649687e-01 9.93716270e-02 -1.69775113e-01 1.82177976e-01 1.47536409e+00 3.44546556e-01 -6.04255438e-01 5.28231859e-01 1.40950787e+00 2.64460206e-01 -2.89242268e-01 6.00604191e-02 3.16854268e-01 -7.56959856e-01 3.66601616e-01 -7.45390594e-01 -8.66727114e-01 9.24140811e-01 3.79695833e-01 7.90550172e-01 6.45260513e-01 -3.57239574e-01 9.25709367e-01 7.58664370e-01 4.85672385e-01 -1.12296939e+00 -1.41609192e-01 9.27382886e-01 1.01337910e+00 -1.32467926e+00 2.82471210e-01 -1.27978653e-01 -5.56384683e-01 1.75409162e+00 1.74724907e-01 -3.74768317e-01 6.24944508e-01 1.31640658e-01 -2.00892329e-01 -4.23921824e-01 -9.45057571e-02 2.66256779e-01 3.27874601e-01 4.74861145e-01 4.11881328e-01 1.40043527e-01 -9.56545293e-01 8.02294374e-01 -7.62101948e-01 -2.72214770e-01 9.01213586e-01 7.46724308e-01 -9.80486631e-01 -6.59099579e-01 -6.25156939e-01 7.96586096e-01 -8.05678666e-01 -3.62950623e-01 -8.04345727e-01 8.73691499e-01 2.61448979e-01 9.77487087e-01 -2.96939686e-02 -6.91733599e-01 -2.04090372e-01 2.64229476e-01 2.37378269e-01 -4.43243384e-01 -9.02653515e-01 -6.62118375e-01 3.43936533e-01 -3.71468067e-01 -5.69036417e-03 -3.52889597e-01 -1.13400793e+00 -7.39232063e-01 -5.88403404e-01 4.53191459e-01 8.88288736e-01 1.46797812e+00 -2.26755925e-02 1.96213454e-01 7.48001218e-01 -4.65665758e-01 -6.94373131e-01 -1.37892818e+00 -6.41100466e-01 3.63640487e-01 6.72601581e-01 -4.87439483e-01 -6.78642333e-01 -1.98857307e-01]
[8.143367767333984, 10.244889259338379]
75811b2f-08da-4e94-ae1a-aa8258044781
multi-hop-reading-comprehension-across-2
2006.06478
null
https://arxiv.org/abs/2006.06478v2
https://arxiv.org/pdf/2006.06478v2.pdf
Multi-hop Reading Comprehension across Documents with Path-based Graph Convolutional Network
Multi-hop reading comprehension across multiple documents attracts much attention recently. In this paper, we propose a novel approach to tackle this multi-hop reading comprehension problem. Inspired by human reasoning processing, we construct a path-based reasoning graph from supporting documents. This graph can combine both the idea of the graph-based and path-based approaches, so it is better for multi-hop reasoning. Meanwhile, we propose Gated-RGCN to accumulate evidence on the path-based reasoning graph, which contains a new question-aware gating mechanism to regulate the usefulness of information propagating across documents and add question information during reasoning. We evaluate our approach on WikiHop dataset, and our approach achieves state-of-the-art accuracy against previously published approaches. Especially, our ensemble model surpasses human performance by 4.2%.
['Wei Xu', 'Yongliang Shen', 'Zeyun Tang', 'Weiming Lu', 'Xinyin Ma', 'Jiale Yu']
2020-06-11
null
null
null
null
['multi-hop-reading-comprehension']
['natural-language-processing']
[ 2.67278165e-01 4.76121008e-01 -1.50784850e-01 -3.42005014e-01 -7.90970385e-01 -4.27394181e-01 4.64896441e-01 9.41116869e-01 -5.38772106e-01 6.05089068e-01 6.58302784e-01 -6.35447383e-01 -4.98584151e-01 -1.19730270e+00 -6.84718609e-01 -1.82717845e-01 5.94212651e-01 3.16964805e-01 9.81429160e-01 -6.25191808e-01 7.04197109e-01 -2.79216260e-01 -1.22337794e+00 5.91356993e-01 1.44582283e+00 4.86535221e-01 2.60903031e-01 7.07855880e-01 -6.17540717e-01 1.00819111e+00 -4.50612843e-01 -6.18760049e-01 -4.42751169e-01 -6.81001961e-01 -1.22195113e+00 -2.93379933e-01 3.79981488e-01 -2.33129501e-01 -2.82696664e-01 1.05832732e+00 3.85486066e-01 3.70240122e-01 2.96936661e-01 -7.03722298e-01 -1.12415183e+00 1.01279688e+00 -6.22210622e-01 5.55128217e-01 9.62250292e-01 -1.04113169e-01 1.21422899e+00 -3.26458424e-01 5.26305676e-01 1.14084661e+00 3.10969949e-01 5.27858138e-01 -6.89560235e-01 -1.36508569e-01 8.80005240e-01 6.20791554e-01 -7.45616257e-01 -8.43653746e-04 6.55686498e-01 -2.55530570e-02 1.13532293e+00 2.58436799e-01 8.05681527e-01 9.01816308e-01 2.61916280e-01 1.12785578e+00 1.06825066e+00 -6.96005404e-01 2.01526597e-01 -4.25596505e-01 8.42246294e-01 1.01984489e+00 2.78105110e-01 -4.82422769e-01 -7.68666863e-01 3.88319939e-02 3.47466856e-01 -1.59686953e-01 -6.82358503e-01 1.38282701e-01 -1.07406938e+00 7.52787292e-01 5.40711462e-01 1.93331629e-01 -4.00782168e-01 -1.29025849e-02 1.09494120e-01 3.57552588e-01 2.00994894e-01 5.20007610e-01 -4.37084198e-01 -1.52696684e-01 -5.76983333e-01 3.76649618e-01 8.08536232e-01 8.51764679e-01 1.92587808e-01 -6.64321542e-01 -8.48381400e-01 8.13569009e-01 6.86114252e-01 2.85935998e-02 4.84800786e-01 -5.77754319e-01 9.49711382e-01 1.01558709e+00 -2.27581099e-01 -9.89535213e-01 -7.19909608e-01 -5.82908273e-01 -5.99132657e-01 -4.67607409e-01 4.12944347e-01 6.75417930e-02 -7.87406027e-01 1.67799473e+00 3.44750494e-01 -1.94324836e-01 9.05031934e-02 9.60932076e-01 1.09825528e+00 5.92180729e-01 2.71575868e-01 -1.86823323e-01 1.76440573e+00 -1.51253688e+00 -8.57092738e-01 -1.73864633e-01 4.50100571e-01 -4.13932353e-01 1.19412804e+00 6.35137260e-01 -1.26721740e+00 -3.59197289e-01 -9.34832275e-01 -3.85634691e-01 -2.74521112e-01 -3.41215461e-01 5.31730533e-01 2.08866253e-01 -1.13326120e+00 1.69100598e-01 -4.75160003e-01 -5.44093013e-01 3.69937807e-01 -1.74022600e-01 6.85556382e-02 -5.55881977e-01 -1.49167347e+00 8.17624569e-01 4.02293622e-01 1.42368123e-01 -4.25423294e-01 -3.89377356e-01 -5.54782987e-01 4.60773826e-01 8.25666308e-01 -1.06565356e+00 1.36414886e+00 -5.49455639e-03 -1.34203100e+00 5.97330570e-01 -3.64559025e-01 -1.92939639e-01 3.93644154e-01 -4.04715002e-01 -4.43640649e-01 3.12384099e-01 6.17335737e-02 3.96966934e-01 5.67558289e-01 -9.60838795e-01 -6.16773665e-01 -4.11034167e-01 5.60189128e-01 5.06275713e-01 -2.06240207e-01 -3.29988122e-01 -9.05669451e-01 -4.67815667e-01 4.58976567e-01 -3.04358244e-01 -7.10189901e-03 -2.02281982e-01 -5.15363574e-01 -7.39317179e-01 2.55929857e-01 -9.09049511e-01 1.63039470e+00 -1.59067261e+00 3.29520375e-01 -4.92747128e-02 6.20524168e-01 1.36166617e-01 -2.98026025e-01 7.50109553e-01 3.60048383e-01 1.51108399e-01 -2.21080869e-01 1.66416273e-01 4.84648868e-02 4.87581193e-02 -1.87720403e-01 -2.63140172e-01 1.58873513e-01 1.16406572e+00 -1.21258867e+00 -4.98348922e-01 -3.36875588e-01 6.92165270e-02 -6.16311371e-01 2.31661275e-01 -7.41669357e-01 1.37026340e-01 -8.33923876e-01 5.80492258e-01 6.65658772e-01 -5.61546862e-01 2.91589886e-01 9.23902467e-02 2.06447244e-01 5.64182162e-01 -6.35915041e-01 1.96248317e+00 -2.64351100e-01 1.76066235e-01 -2.35567182e-01 -8.08624268e-01 8.50767434e-01 -5.55725908e-03 -2.93010205e-01 -1.19028711e+00 -1.46672130e-03 -1.84800446e-01 1.84960023e-01 -7.50546515e-01 5.08984745e-01 1.92991808e-01 4.95770164e-02 6.42519355e-01 -1.99974924e-02 1.68681458e-01 3.61022741e-01 6.97672188e-01 1.49363232e+00 2.52110422e-01 2.86284387e-01 -2.37146109e-01 9.26446199e-01 -6.51353002e-02 1.08320236e-01 8.94492030e-01 -3.31278257e-02 2.64433950e-01 8.85312557e-01 -6.54679816e-03 -2.16080800e-01 -1.05501425e+00 3.28648746e-01 1.38682592e+00 4.17922884e-01 -6.42462075e-01 -8.93497825e-01 -9.51199234e-01 -9.82752070e-02 9.94201481e-01 -5.51306486e-01 -4.16944355e-01 -5.41940570e-01 -6.49683833e-01 2.75864065e-01 7.29314625e-01 9.01154459e-01 -1.08245718e+00 -4.06330913e-01 4.15842921e-01 -6.28711224e-01 -1.04668903e+00 -2.77724147e-01 -3.99051495e-02 -9.11802590e-01 -1.28297842e+00 -4.76820886e-01 -7.55483091e-01 7.28748441e-01 3.58050287e-01 1.45729697e+00 8.26548815e-01 1.80063426e-01 4.99914885e-01 -8.64711404e-01 -3.68315130e-01 7.53381988e-03 3.73648107e-01 -7.18899429e-01 -3.23298216e-01 5.42595983e-01 -1.59849718e-01 -7.32794464e-01 3.91181856e-02 -7.34777331e-01 2.87782729e-01 6.44563496e-01 6.99133992e-01 4.45099205e-01 1.09233968e-01 9.39041317e-01 -9.96999860e-01 1.44865203e+00 -5.78643620e-01 -4.14436847e-01 8.05702746e-01 -7.76170254e-01 2.17599958e-01 6.77699506e-01 2.63889320e-02 -1.44774914e+00 -7.83988655e-01 -3.39975268e-01 4.43976760e-01 1.33287877e-01 7.66449332e-01 -6.60317466e-02 2.47523442e-01 3.71155858e-01 2.12257728e-01 -4.97659951e-01 -3.66905749e-01 5.96026361e-01 5.10701954e-01 4.44275409e-01 -6.98971868e-01 3.47587645e-01 9.20550227e-02 -2.64076829e-01 -3.57213795e-01 -1.44181705e+00 -4.70915645e-01 -3.07403028e-01 -2.71418750e-01 1.19771779e+00 -6.55114293e-01 -9.42017317e-01 4.71162260e-01 -1.36670899e+00 -1.92493662e-01 -1.91294160e-02 2.46317253e-01 -3.11542511e-01 3.41733992e-01 -7.78262079e-01 -7.56859541e-01 -4.64956343e-01 -8.25795949e-01 8.62236142e-01 6.73359275e-01 -1.22888245e-01 -1.10588551e+00 -8.73463321e-03 1.00768304e+00 3.79775703e-01 -1.41076952e-01 1.28945386e+00 -7.86178589e-01 -9.32836235e-01 5.79229407e-02 -5.93077183e-01 -1.40049458e-01 -2.88150847e-01 -5.30913293e-01 -7.26354718e-01 -2.82491054e-02 -1.02185987e-01 -4.55823183e-01 1.30077028e+00 7.14766756e-02 1.39667261e+00 1.83180068e-02 -2.17175409e-01 7.76919276e-02 1.25988114e+00 -4.44407165e-02 6.30793393e-01 4.10706580e-01 5.43554842e-01 5.68821430e-01 6.03838980e-01 2.22444668e-01 1.04878259e+00 2.91009039e-01 4.12883937e-01 2.31136590e-01 -1.53800562e-01 -6.34460747e-01 -3.84782478e-02 1.14597225e+00 -1.14967607e-01 -1.01412654e+00 -8.95147443e-01 4.00375813e-01 -2.08187580e+00 -7.90500998e-01 -4.56485659e-01 1.69273341e+00 7.54666865e-01 3.76602024e-01 -2.81683117e-01 5.76244891e-02 4.19043213e-01 3.77241701e-01 -5.18949866e-01 -5.14581203e-01 4.56445739e-02 2.79974312e-01 -1.05766490e-01 5.79332352e-01 -4.83328402e-01 9.66940939e-01 6.06041479e+00 5.22992969e-01 -3.08660775e-01 1.54773876e-01 2.27688134e-01 3.48064572e-01 -6.87030137e-01 -5.25417551e-02 -8.94044220e-01 2.49021262e-01 6.35325253e-01 -9.79130119e-02 2.97081351e-01 1.30251572e-01 -1.24292029e-02 -5.71285248e-01 -7.98045635e-01 4.54792202e-01 4.38134521e-01 -1.18435979e+00 2.90762097e-01 -4.00589436e-01 4.49481815e-01 -3.48243117e-01 -3.30296665e-01 6.03222191e-01 4.36048746e-01 -7.48110294e-01 3.12702000e-01 9.38014984e-01 6.46115616e-02 -5.25025845e-01 9.50916290e-01 6.28147602e-01 -1.11919999e+00 -9.45122167e-02 -2.71421760e-01 -1.24764167e-01 4.25548196e-01 5.94301641e-01 -2.34381720e-01 9.96383190e-01 6.20271385e-01 5.23248374e-01 -8.44067872e-01 1.02722502e+00 -1.04136968e+00 7.36108482e-01 2.11576104e-01 -5.34121990e-01 3.27903271e-01 -1.28875136e-01 2.03635082e-01 9.12730813e-01 1.29641309e-01 6.57316387e-01 4.53019813e-02 7.61100292e-01 -4.12898570e-01 1.28438473e-01 -5.30068986e-02 -5.00943772e-02 2.92442232e-01 1.14700985e+00 -7.91064501e-01 -3.97803366e-01 -6.00613832e-01 1.01443410e+00 9.07334805e-01 3.01549226e-01 -8.56216609e-01 -4.81671333e-01 -7.09942579e-02 1.08790481e-02 2.51025289e-01 -3.59569266e-02 1.23886415e-03 -1.12216747e+00 3.81705374e-01 -7.41390765e-01 8.20978284e-01 -1.20764434e+00 -1.27285147e+00 4.98000383e-01 -1.27484247e-01 -5.92857242e-01 2.61828661e-01 -5.23943245e-01 -8.82187068e-01 7.51315892e-01 -1.94527471e+00 -8.97282898e-01 -8.05039704e-01 3.48628193e-01 6.77292585e-01 1.25870258e-01 5.99813163e-01 -1.29756585e-01 -4.91911888e-01 3.53125870e-01 -4.41138685e-01 -7.02005476e-02 4.08132404e-01 -1.39664173e+00 4.78976190e-01 9.03392792e-01 9.20836329e-02 7.77924597e-01 2.83070832e-01 -7.47083545e-01 -1.36283040e+00 -8.56728852e-01 1.00639760e+00 -5.32834172e-01 5.54880738e-01 5.98615706e-02 -1.44949210e+00 6.40037537e-01 6.78023458e-01 -6.78750157e-01 6.67761743e-01 4.94320065e-01 -5.52753329e-01 1.20177314e-01 -9.60151672e-01 7.57812560e-01 1.44894958e+00 -1.80967540e-01 -1.44400120e+00 3.97526234e-01 1.14839983e+00 -4.29729819e-01 -6.29029095e-01 1.84230413e-02 2.37348557e-01 -1.07419097e+00 6.77605927e-01 -6.63908362e-01 6.04670465e-01 -2.67845392e-01 6.62439242e-02 -1.45910013e+00 -5.68027496e-01 -3.90083581e-01 -3.68083507e-01 1.09759212e+00 5.61715245e-01 -7.68144131e-01 5.09277999e-01 2.12414354e-01 -2.90050089e-01 -1.06307292e+00 -5.05238235e-01 -3.68199974e-01 8.91443063e-03 -2.28865758e-01 6.83207631e-01 6.53899133e-01 5.16262174e-01 7.10088789e-01 2.23063082e-01 7.78322518e-02 4.49095398e-01 1.41359672e-01 2.43954092e-01 -1.23417723e+00 -4.12306935e-01 -5.75352907e-01 1.07644923e-01 -1.60237563e+00 2.03018367e-01 -1.09787667e+00 6.05026148e-02 -2.54158068e+00 4.01261061e-01 4.43315171e-02 -4.84492391e-01 3.72490019e-01 -8.46838474e-01 -4.62886482e-01 2.30577722e-01 -2.79769063e-01 -1.21617329e+00 5.13183236e-01 1.77894318e+00 -1.12607226e-01 -5.25754280e-02 -2.45612070e-01 -1.19714952e+00 6.09992683e-01 8.85382712e-01 -1.53238565e-01 -8.42192173e-01 -8.33512425e-01 7.25880742e-01 2.04016209e-01 2.77600318e-01 -7.92677760e-01 6.48344159e-01 1.95331089e-02 1.87824890e-01 -7.99682081e-01 -5.63677140e-02 -3.49323720e-01 -7.10509419e-01 4.00923818e-01 -7.65260994e-01 2.90120125e-01 6.70178756e-02 9.96704102e-01 -2.76307553e-01 -3.28054160e-01 3.63089025e-01 -2.52533019e-01 -7.84048557e-01 -1.65149849e-03 -2.31090590e-01 4.91850257e-01 7.24132478e-01 3.03790629e-01 -1.02015173e+00 -4.40803975e-01 -5.23490250e-01 9.19976652e-01 -1.07489176e-01 6.56164646e-01 8.62261832e-01 -8.91082227e-01 -6.64940715e-01 5.73319383e-02 3.76519620e-01 1.84178635e-01 4.19684350e-01 9.15140808e-01 -4.29788530e-01 5.46624959e-01 4.62476015e-02 -2.86465466e-01 -1.06196845e+00 4.58230168e-01 -3.45063629e-03 -7.26679325e-01 -7.26960599e-01 1.09434962e+00 -2.18486950e-01 -4.62612420e-01 3.22587043e-02 -4.84488338e-01 -6.91059828e-01 8.63870829e-02 6.79836154e-01 3.61659616e-01 1.39738783e-01 1.65861130e-01 -1.97974280e-01 6.63092315e-01 -4.01823848e-01 -2.30058823e-02 1.04550934e+00 -2.59602726e-01 -2.82947332e-01 3.41784418e-01 5.61218917e-01 7.97308795e-03 -7.24235117e-01 -1.62215754e-01 1.94034889e-01 -2.85792083e-01 5.33231609e-02 -1.29975820e+00 -6.99766517e-01 9.24090862e-01 -1.64232701e-01 4.57088619e-01 1.16898274e+00 2.33507112e-01 9.63650107e-01 4.43198144e-01 3.37979198e-01 -1.04290032e+00 3.14839751e-01 6.55537724e-01 9.16699529e-01 -1.15577137e+00 1.11590259e-01 -6.27635121e-01 -5.63125968e-01 1.03754425e+00 1.00827384e+00 2.53707379e-01 3.20877850e-01 -2.68022984e-01 -1.14222899e-01 -5.83311617e-01 -1.16223359e+00 -3.13605160e-01 3.32491487e-01 4.07463700e-01 6.38738096e-01 -5.99978715e-02 -8.67538989e-01 7.98950553e-01 -7.19776154e-02 1.47559077e-01 5.18408358e-01 9.97272670e-01 -8.20968747e-01 -9.07669187e-01 -1.32738799e-01 6.45411968e-01 3.69202462e-03 -3.38685572e-01 -3.19802016e-01 6.81328654e-01 -2.26385340e-01 1.32667994e+00 -4.68989648e-02 -1.71317443e-01 7.17339754e-01 3.35732162e-01 6.02512956e-01 -5.02086937e-01 -7.34696627e-01 -3.34540248e-01 1.97816372e-01 -6.49845958e-01 -3.87804180e-01 -3.69701266e-01 -1.54582393e+00 -1.47389099e-01 -4.92659897e-01 2.43237853e-01 2.04420969e-01 1.03677773e+00 3.49639922e-01 1.14287591e+00 1.43985339e-02 3.19054276e-01 -4.89706218e-01 -8.72506201e-01 -2.92394638e-01 2.44671270e-01 8.38696808e-02 -4.18142945e-01 -2.13637620e-01 -3.70289207e-01]
[10.849823951721191, 7.939654350280762]
14768aa6-40c4-4fc7-b51f-792d7b2888f1
inference-of-a-rumor-s-source-in-the
2205.12125
null
https://arxiv.org/abs/2205.12125v1
https://arxiv.org/pdf/2205.12125v1.pdf
Inference of a Rumor's Source in the Independent Cascade Model
We consider the so-called Independent Cascade Model for rumor spreading or epidemic processes popularized by Kempe et al.\ [2003]. In this model, a small subset of nodes from a network are the source of a rumor. In discrete time steps, each informed node "infects" each of its uninformed neighbors with probability $p$. While many facets of this process are studied in the literature, less is known about the inference problem: given a number of infected nodes in a network, can we learn the source of the rumor? In the context of epidemiology this problem is often referred to as patient zero problem. It belongs to a broader class of problems where the goal is to infer parameters of the underlying spreading model, see, e.g., Lokhov [NeurIPS'16] or Mastakouri et al. [NeurIPS'20]. In this work we present a maximum likelihood estimator for the rumor's source, given a snapshot of the process in terms of a set of active nodes $X$ after $t$ steps. Our results show that, for cycle-free graphs, the likelihood estimator undergoes a non-trivial phase transition as a function $t$. We provide a rigorous analysis for two prominent classes of acyclic network, namely $d$-regular trees and Galton-Watson trees, and verify empirically that our heuristics work well in various general networks.
['Malin Rau', 'Lena Krieg', 'Dominik Kaaser', 'Max Hahn-Klimroth', 'Petra Berenbrink']
2022-05-24
null
null
null
null
['epidemiology']
['medical']
[ 1.55174816e-02 4.76608604e-01 -3.86584669e-01 7.01232627e-03 1.34100184e-01 -5.37554801e-01 4.45320189e-01 2.46431544e-01 -3.30089539e-01 8.40720475e-01 -2.12633461e-01 -4.50102985e-01 -3.69852304e-01 -1.17700005e+00 -5.96652985e-01 -7.90237129e-01 -8.30838621e-01 1.06992924e+00 1.31235734e-01 -3.23860496e-01 2.30976135e-01 3.17189932e-01 -6.20423377e-01 -4.53058392e-01 4.76754695e-01 5.53564131e-01 -1.25472814e-01 7.62869060e-01 4.06869315e-02 8.27910960e-01 -6.44176006e-01 -4.37104553e-01 1.01249302e-02 -7.14479029e-01 -1.00355065e+00 8.74915421e-02 -8.31451714e-01 -1.86342284e-01 -3.70534033e-01 1.15939808e+00 1.54748768e-01 -3.73477817e-01 7.25880325e-01 -1.35292149e+00 -1.78348318e-01 9.47812855e-01 -1.07861435e+00 6.75273597e-01 3.25887859e-01 -1.57327101e-01 8.12533140e-01 -3.69097531e-01 9.56973076e-01 1.16755986e+00 8.90360415e-01 2.22943589e-01 -1.35508692e+00 -8.73602748e-01 -1.20924190e-01 -1.05512880e-01 -1.39243793e+00 -1.86830238e-01 6.99385583e-01 -4.34766740e-01 3.79297495e-01 8.47527534e-02 6.32307827e-01 9.37998176e-01 6.53746367e-01 4.81535405e-01 1.29877400e+00 -1.82864591e-01 4.75631714e-01 8.90264586e-02 2.71897942e-01 8.82001579e-01 7.21193612e-01 2.94953212e-02 -5.70032775e-01 -8.46793890e-01 9.62159276e-01 3.45956534e-02 -2.04087809e-01 -1.58924818e-01 -7.96776175e-01 1.37766647e+00 2.91782647e-01 3.19758713e-01 -6.53271556e-01 1.74364492e-01 -5.54024801e-02 5.45690000e-01 9.21250761e-01 2.48687398e-02 -6.94331974e-02 1.02691084e-01 -7.92967916e-01 2.27455348e-01 1.48579967e+00 7.27934003e-01 7.74024069e-01 -2.38604099e-01 5.80366910e-01 3.43626410e-01 5.22304237e-01 8.44071925e-01 -3.92354578e-01 -7.69476831e-01 2.18181938e-01 3.17463398e-01 3.81039679e-01 -9.44067776e-01 -5.27012885e-01 -5.12053072e-01 -1.20842206e+00 -5.96148409e-02 4.85850096e-01 -6.63307846e-01 -6.56062543e-01 1.83080673e+00 3.17431599e-01 2.88438112e-01 -2.24824473e-01 5.33494651e-01 7.24897683e-02 7.25850344e-01 -1.37121469e-01 -9.44601536e-01 1.30046248e+00 -3.33848566e-01 -6.84636950e-01 -3.51433784e-01 1.61930531e-01 -6.13720894e-01 -1.61856174e-01 3.17877948e-01 -1.24580848e+00 3.90114039e-01 -4.60991472e-01 8.88379753e-01 9.26454961e-02 -7.73133457e-01 6.83941424e-01 7.44426489e-01 -1.29702950e+00 3.65462780e-01 -8.06799531e-01 -8.12331617e-01 3.35367024e-01 2.46544361e-01 1.83866367e-01 -3.67301732e-01 -1.28651822e+00 6.07875109e-01 -1.54841453e-01 2.60030180e-02 -1.27723897e+00 -1.41868979e-01 -2.09313989e-01 -9.95329767e-02 6.59914315e-01 -9.36224580e-01 1.23339200e+00 -7.97973037e-01 -9.10430610e-01 7.81467855e-01 -5.46889603e-01 -6.55170143e-01 5.49539387e-01 1.79903805e-01 -1.19581304e-01 4.75573510e-01 3.08024347e-01 -2.65166581e-01 9.06145394e-01 -1.33478940e+00 -4.48324054e-01 -6.28573298e-01 -3.37392883e-03 1.88555643e-02 2.52388328e-01 4.06718493e-01 -2.14254320e-01 -3.78709197e-01 2.33508453e-01 -1.13426161e+00 -7.32730508e-01 -2.81992227e-01 -4.91850853e-01 -1.53932273e-01 5.67909479e-01 -4.43560064e-01 1.01494420e+00 -1.59532285e+00 2.19240282e-02 6.17833078e-01 6.31119609e-01 -1.38732374e-01 2.62049317e-01 1.04856241e+00 1.90673187e-01 5.15708745e-01 -5.09779274e-01 -1.61311686e-01 -5.58952987e-01 1.93808645e-01 -1.60367683e-01 9.87858832e-01 -2.27493182e-01 4.43066210e-01 -1.00236320e+00 -3.41477543e-01 -3.86620849e-01 2.89615691e-01 -2.95695275e-01 6.37735799e-02 -1.61115974e-01 3.96976620e-01 -7.15182126e-01 3.59888494e-01 4.98122334e-01 -8.60448599e-01 4.62810904e-01 8.37845087e-01 1.09967351e-01 -1.00623280e-01 -9.86457169e-01 7.79330254e-01 -3.97132561e-02 4.21814740e-01 7.44382858e-01 -9.25728619e-01 6.64789855e-01 7.13660359e-01 6.63317740e-01 2.44820908e-01 4.09742773e-01 6.77632689e-02 -3.72734107e-02 -3.47252905e-01 2.91024417e-01 -6.78828239e-01 -7.15263113e-02 1.02638161e+00 -4.58559960e-01 2.48779222e-01 -6.85384795e-02 5.20344734e-01 1.64356792e+00 -8.36269379e-01 6.67489827e-01 -3.69572699e-01 -1.03034578e-01 3.17803651e-01 3.52617294e-01 1.27721310e+00 -2.41202727e-01 -1.64574347e-02 8.60682786e-01 5.46102338e-02 -9.21013355e-01 -1.31284893e+00 -1.16293645e-02 7.02702224e-01 2.53777623e-01 -1.41459733e-01 -8.28213274e-01 -2.56682485e-01 -9.52538997e-02 4.40112323e-01 -8.57120335e-01 2.67458647e-01 -4.42390800e-01 -1.41518879e+00 4.71616447e-01 -2.17570439e-01 5.55721641e-01 -9.69285131e-01 -3.68340552e-01 4.41136628e-01 -3.77758861e-01 -7.73992240e-01 -8.98318514e-02 4.54408210e-03 -1.14762747e+00 -1.36310136e+00 -8.11196208e-01 -5.46488404e-01 8.33399713e-01 3.54453862e-01 1.06329656e+00 2.26295859e-01 -5.78760207e-02 3.85777384e-01 -9.16534439e-02 -3.00055027e-01 -8.20125818e-01 -5.14971791e-03 4.68523875e-02 -4.70137186e-02 1.42109588e-01 -6.18964791e-01 -7.06562459e-01 3.67079705e-01 -8.73641968e-01 -4.01157260e-01 4.26172346e-01 3.38597417e-01 1.12816952e-01 4.85965818e-01 5.59813559e-01 -1.23020732e+00 9.29719567e-01 -1.34502685e+00 -4.53700304e-01 -6.83408156e-02 -6.44979775e-01 -1.22626998e-01 4.04825568e-01 -1.36198089e-01 -7.90276706e-01 -3.62634182e-01 2.59306341e-01 -2.86725145e-02 9.02301893e-02 8.56522262e-01 4.49844211e-01 1.33254886e-01 4.46738183e-01 7.11624697e-02 2.52551705e-01 -3.40093523e-01 1.98637292e-01 5.43119669e-01 3.80983129e-02 -1.57427117e-01 8.30495536e-01 9.08658981e-01 1.68059036e-01 -1.17494249e+00 -4.14094329e-01 -6.09195292e-01 -1.61238119e-01 -3.38927001e-01 6.57656252e-01 -6.07829571e-01 -9.65510905e-01 6.86276555e-01 -1.32084537e+00 -2.23798081e-01 3.09401155e-02 3.59998405e-01 -5.21555305e-01 2.29228631e-01 -1.11262703e+00 -1.35808694e+00 -2.10781202e-01 -5.22706568e-01 1.78248495e-01 1.50489181e-01 -1.79309741e-01 -1.40275061e+00 5.58756411e-01 1.68144986e-01 5.19812644e-01 1.48659781e-01 8.71138394e-01 -7.59032130e-01 -7.30072320e-01 -3.64206821e-01 -1.68076530e-01 -2.74673134e-01 1.38776049e-01 -2.23719895e-01 -3.87673587e-01 -4.56644833e-01 4.41299975e-01 3.35064232e-01 7.55645454e-01 7.25711584e-01 2.54778773e-01 -7.09538043e-01 -8.74158740e-01 8.86216015e-02 1.43550968e+00 2.10016891e-01 3.18749845e-01 -1.97869509e-01 1.69667915e-01 7.23077416e-01 1.44970581e-01 5.87970257e-01 4.04517919e-01 4.61674444e-02 5.29412985e-01 1.31846607e-01 5.75357974e-01 -1.54863477e-01 2.90551335e-01 8.05280089e-01 -4.28967297e-01 -7.68289804e-01 -1.09875536e+00 6.84728265e-01 -1.64493203e+00 -1.18147004e+00 -2.56494790e-01 2.17787457e+00 7.25164533e-01 1.93661705e-01 4.19366866e-01 -1.58622667e-01 1.22298944e+00 3.66605610e-01 -3.87689471e-01 -1.10524096e-01 1.27604067e-01 1.33896545e-01 9.14213300e-01 8.49120378e-01 -5.13849914e-01 5.70803225e-01 6.82886982e+00 2.28449196e-01 -7.05655515e-01 3.77328545e-01 7.47578025e-01 9.16075110e-02 -2.50514060e-01 2.92569101e-01 -7.61145949e-01 2.97154397e-01 1.30179870e+00 -5.43134570e-01 5.56930542e-01 4.58903342e-01 6.11646473e-01 -3.92943084e-01 -5.14407635e-01 3.72692853e-01 -5.81285208e-02 -1.07925010e+00 -4.88892376e-01 6.15788937e-01 7.32063532e-01 4.14873362e-01 -2.81884700e-01 -1.61373079e-01 1.16729522e+00 -1.12281704e+00 1.63947210e-01 4.35144275e-01 3.61420542e-01 -7.47586668e-01 6.29814506e-01 7.68430233e-01 -9.91055906e-01 1.16952702e-01 -1.37896150e-01 -1.88327953e-01 7.56824493e-01 9.72408533e-01 -1.16479492e+00 4.00167793e-01 5.41369259e-01 4.63397652e-01 7.44671971e-02 1.07837164e+00 -1.23837329e-01 1.09927595e+00 -5.63532174e-01 -1.54573798e-01 1.80813178e-01 -3.78927767e-01 8.90896201e-01 8.05272222e-01 2.04254866e-01 5.67968249e-01 -9.72110480e-02 7.64155567e-01 -1.13643780e-01 -1.46471083e-01 -7.87890494e-01 -2.21973687e-01 6.34839058e-01 8.34203243e-01 -1.15289164e+00 -2.21038267e-01 1.14579208e-01 6.90261424e-01 -3.51434685e-02 4.97390121e-01 -4.53898937e-01 -2.67243292e-02 3.29131961e-01 5.60599148e-01 4.74027336e-01 -2.49859616e-01 -9.62329358e-02 -8.58989596e-01 -4.65966791e-01 -5.88884771e-01 3.99110943e-01 -4.03492153e-01 -1.44644129e+00 5.25475204e-01 1.33953601e-01 -5.30299723e-01 -5.47996461e-01 7.13213757e-02 -7.54776359e-01 8.71555328e-01 -1.26650679e+00 -4.58363712e-01 2.24852279e-01 7.03727782e-01 1.92791820e-01 2.39739805e-01 5.11903942e-01 -3.15254629e-01 -5.03187418e-01 -2.26885274e-01 4.85957503e-01 1.88140109e-01 4.54084650e-02 -9.34285104e-01 4.42506403e-01 7.13324249e-01 -1.30157724e-01 7.01400101e-01 1.22046030e+00 -1.21413183e+00 -1.24961984e+00 -8.18625867e-01 1.14400077e+00 -1.98369801e-01 8.88990283e-01 -2.88675517e-01 -7.59144723e-01 9.40467596e-01 3.81877214e-01 -1.30991936e-01 4.05264825e-01 8.16223994e-02 7.87800997e-02 2.71963835e-01 -1.29650557e+00 4.14299130e-01 9.15472388e-01 -2.07833752e-01 -2.08562911e-01 4.74402338e-01 3.15416694e-01 2.70394474e-01 -8.07697177e-01 3.26843299e-02 2.30327368e-01 -1.14607799e+00 6.79282010e-01 -5.79603970e-01 3.02204967e-01 2.26371899e-01 3.36499661e-01 -1.44854677e+00 -1.34350479e-01 -9.67464149e-01 7.15058148e-02 7.90035903e-01 3.10643524e-01 -9.91201460e-01 9.34382796e-01 -1.68148741e-01 8.09758902e-01 -7.08886027e-01 -1.13863921e+00 -5.80727100e-01 2.46474594e-01 -7.21925572e-02 1.86782360e-01 1.01621413e+00 1.25235677e-01 5.13462365e-01 -4.69843477e-01 4.32365388e-01 1.25697112e+00 -4.29143757e-02 4.64888066e-01 -1.47847629e+00 -4.18640852e-01 -1.39018193e-01 1.97543234e-01 -1.04315889e+00 -2.02121362e-01 -6.06375515e-01 3.28122765e-01 -1.55152225e+00 4.19001073e-01 -7.44363666e-01 1.33588329e-01 -1.19534656e-01 4.97829206e-02 -1.82069112e-02 -2.21524164e-02 7.33384550e-01 -3.60559672e-01 4.88812067e-02 8.94139707e-01 2.79079080e-01 -9.83008370e-02 6.36089444e-01 -5.71165919e-01 9.83998179e-01 1.02039504e+00 -1.02902126e+00 -3.96923870e-01 4.06921543e-02 5.74776351e-01 1.26060688e+00 5.04396737e-01 -2.09639341e-01 2.77532339e-01 -2.49226719e-01 -2.60991096e-01 -5.29173851e-01 2.07892761e-01 -5.10002434e-01 4.86116171e-01 9.42285180e-01 -1.25275418e-01 2.78810352e-01 -5.35319030e-01 1.18813276e+00 3.18806827e-01 -3.57866287e-01 7.36886799e-01 -4.75679636e-01 1.37178376e-01 4.22065347e-01 -9.05092001e-01 3.18405867e-01 9.33574080e-01 2.26088449e-01 -5.76632798e-01 -1.11522579e+00 -8.34212124e-01 2.08637908e-01 5.75741172e-01 -1.42607599e-01 4.07213241e-01 -5.04658222e-01 -8.16192210e-01 -2.18923137e-01 -4.01793391e-01 -1.44795969e-01 -3.15952464e-03 1.24323726e+00 -3.23600918e-01 1.59583539e-01 1.46856695e-01 -4.22105402e-01 -1.05650270e+00 5.19878447e-01 3.10203791e-01 -5.52357614e-01 -2.84646660e-01 5.19259095e-01 1.19133919e-01 2.02770472e-01 -1.97971269e-01 2.57482320e-01 1.51922815e-02 1.19126417e-01 4.37262356e-01 5.62817395e-01 -6.05430782e-01 -5.63621700e-01 -3.88749331e-01 8.63769203e-02 -3.01292151e-01 -4.83836800e-01 1.38465762e+00 -5.23516953e-01 -6.35598958e-01 6.69790745e-01 8.98730040e-01 1.18377268e-01 -7.79086113e-01 -5.93198895e-01 8.46583694e-02 -6.98019862e-02 -1.52990460e-01 -4.49781537e-01 -1.16157186e+00 3.80197018e-01 -1.79643333e-02 9.89323258e-01 7.72024870e-01 4.70224530e-01 4.35207754e-01 1.34845152e-01 6.93175793e-01 -4.45061088e-01 -1.67534158e-01 3.68296593e-01 2.83102274e-01 -8.16722274e-01 -3.40887643e-02 -6.98696315e-01 -2.67942607e-01 6.84738398e-01 -2.51670569e-01 -4.30428565e-01 1.17182064e+00 3.00854385e-01 -2.55051136e-01 -6.88240111e-01 -7.94003844e-01 8.40251520e-02 -6.53941989e-01 3.84716362e-01 8.38788226e-02 8.49973187e-02 -4.04424936e-01 3.27556342e-01 -1.17740102e-01 2.49331459e-01 9.35753345e-01 1.01190162e+00 -9.42585468e-01 -8.37077916e-01 -6.62830949e-01 6.45755291e-01 -6.68954492e-01 -1.62064135e-01 -5.00388980e-01 7.98746347e-01 -3.40672731e-01 1.19685614e+00 -5.38050160e-02 -6.64504990e-02 -2.09360749e-01 -9.93226394e-02 2.72596896e-01 -6.69973314e-01 -5.18948317e-01 1.01885423e-01 1.96414530e-01 -1.44420490e-01 -6.14624500e-01 -9.40137863e-01 -8.86207342e-01 -1.12517154e+00 -3.98860157e-01 5.91619670e-01 3.46181840e-01 1.11976016e+00 -1.60688370e-01 -1.41095771e-02 9.93935227e-01 -1.83856860e-01 -5.99550366e-01 -8.96413028e-01 -1.18724704e+00 -1.22036599e-01 3.73318285e-01 -3.34206939e-01 -8.72263968e-01 -1.25175536e-01]
[6.677347660064697, 5.073212146759033]
94b8ea10-3d13-43c7-9e9e-2372312982b0
unsupervised-low-light-image-enhancement
2306.02082
null
https://arxiv.org/abs/2306.02082v1
https://arxiv.org/pdf/2306.02082v1.pdf
Unsupervised Low Light Image Enhancement Using SNR-Aware Swin Transformer
Image captured under low-light conditions presents unpleasing artifacts, which debilitate the performance of feature extraction for many upstream visual tasks. Low-light image enhancement aims at improving brightness and contrast, and further reducing noise that corrupts the visual quality. Recently, many image restoration methods based on Swin Transformer have been proposed and achieve impressive performance. However, On one hand, trivially employing Swin Transformer for low-light image enhancement would expose some artifacts, including over-exposure, brightness imbalance and noise corruption, etc. On the other hand, it is impractical to capture image pairs of low-light images and corresponding ground-truth, i.e. well-exposed image in same visual scene. In this paper, we propose a dual-branch network based on Swin Transformer, guided by a signal-to-noise ratio prior map which provides the spatial-varying information for low-light image enhancement. Moreover, we leverage unsupervised learning to construct the optimization objective based on Retinex model, to guide the training of proposed network. Experimental results demonstrate that the proposed model is competitive with the baseline models.
['Yanzeng Gao', 'Zihan Huang', 'Yueen Hou', 'Jiahui Tang', 'Zhijian Luo']
2023-06-03
null
null
null
null
['image-enhancement', 'low-light-image-enhancement', 'image-restoration']
['computer-vision', 'computer-vision', 'computer-vision']
[ 6.31698966e-01 -5.58884501e-01 2.18855917e-01 -3.00276279e-01 -5.26183665e-01 -2.69432366e-01 3.27939272e-01 -2.00988382e-01 -2.34080359e-01 8.17030191e-01 1.70116037e-01 -9.10464525e-02 -1.37269303e-01 -7.55411923e-01 -5.07031262e-01 -1.23485386e+00 4.95989978e-01 -7.61566699e-01 -2.96833925e-02 -2.17649356e-01 2.59548753e-01 2.44819567e-01 -1.56094837e+00 9.74951237e-02 1.07464504e+00 1.09227109e+00 5.36073923e-01 3.77097338e-01 6.61958978e-02 8.24636042e-01 -1.82433248e-01 -1.22402996e-01 4.84336585e-01 -4.27551091e-01 -1.83424264e-01 2.79958844e-01 5.38276553e-01 -6.41743958e-01 -4.92416859e-01 1.50669324e+00 5.71962237e-01 2.51471668e-01 3.86557698e-01 -9.26894784e-01 -8.01476479e-01 -1.14925772e-01 -8.66777539e-01 4.63004023e-01 9.64119062e-02 6.35796010e-01 6.08787358e-01 -1.00552833e+00 2.31278867e-01 8.78034651e-01 5.94243884e-01 2.22568065e-01 -1.14870584e+00 -6.60376370e-01 -1.03990301e-01 2.31855944e-01 -1.01616251e+00 -7.35702097e-01 1.11854315e+00 -9.01051089e-02 3.82517755e-01 1.90822005e-01 6.49233401e-01 5.56149781e-01 2.72331059e-01 4.90266591e-01 1.80430579e+00 -3.44382465e-01 -1.68537423e-01 -5.93265928e-02 -2.18687236e-01 6.20843887e-01 2.84405321e-01 5.84805429e-01 -3.55015099e-01 3.81861925e-01 7.49060929e-01 2.13275537e-01 -7.86962211e-01 -9.75587443e-02 -9.67099488e-01 8.26951191e-02 6.51041806e-01 1.36180148e-01 -2.98459113e-01 -1.58429682e-01 4.69336547e-02 1.05102524e-01 4.79165167e-01 1.90292239e-01 -2.32689232e-01 2.76028335e-01 -7.15790808e-01 -2.16912150e-01 -8.98068398e-02 5.83567441e-01 1.02642095e+00 1.23338513e-01 -3.58704031e-01 9.50480521e-01 5.25272906e-01 6.89058781e-01 2.33159781e-01 -8.07677209e-01 4.95698899e-01 4.11404341e-01 2.79240698e-01 -1.17319393e+00 -3.21963489e-01 -7.00192332e-01 -1.34520233e+00 5.30599833e-01 2.96539485e-01 3.39069404e-02 -8.47130537e-01 1.57258940e+00 3.45363349e-01 3.95015448e-01 8.85128453e-02 1.43259788e+00 8.37393522e-01 8.08829606e-01 -6.02373704e-02 -8.20383310e-01 1.21323454e+00 -8.76433194e-01 -9.08848286e-01 -2.03209534e-01 -7.60974437e-02 -1.08856249e+00 1.26170576e+00 4.92665887e-01 -1.35329449e+00 -7.90999055e-01 -1.12525439e+00 -2.69058526e-01 4.58042659e-02 2.75419891e-01 4.33138520e-01 6.59461737e-01 -8.75171721e-01 4.85023916e-01 -5.79072356e-01 7.02340752e-02 4.79687661e-01 6.28054962e-02 -1.89610779e-01 -4.65239882e-01 -1.02006471e+00 6.73198998e-01 1.17679864e-01 8.08883488e-01 -1.04104209e+00 -6.18988574e-01 -7.04185545e-01 -7.66558498e-02 4.34442878e-01 -5.78653991e-01 7.08029926e-01 -6.51364744e-01 -1.69102538e+00 7.09248126e-01 -2.43838936e-01 3.07403132e-03 2.88173258e-01 -3.32664289e-02 -5.31755567e-01 1.32779598e-01 4.39890809e-02 6.23866096e-02 1.06879914e+00 -1.63245428e+00 -6.05777204e-01 -4.21429843e-01 -4.71735839e-03 4.73051816e-01 -3.04221570e-01 3.76590490e-02 -4.57657814e-01 -5.30653477e-01 2.53569514e-01 -2.78657138e-01 -1.36219531e-01 8.29546750e-02 -3.81246120e-01 3.14656258e-01 8.83568108e-01 -7.46133745e-01 1.09616244e+00 -2.23909426e+00 -3.85067582e-01 -3.64418677e-03 3.57550859e-01 4.51389134e-01 -2.41895989e-01 6.65283352e-02 -1.51129007e-01 -2.47715145e-01 -3.58821154e-01 -1.26771048e-01 -4.38631803e-01 1.75266508e-02 -2.40789518e-01 8.47190320e-01 9.25007835e-02 7.67542899e-01 -1.02080536e+00 -5.59664249e-01 5.92870533e-01 6.53867066e-01 -2.69032270e-01 4.37228292e-01 1.85867548e-01 7.32778311e-01 -1.76379502e-01 6.44978166e-01 1.14593887e+00 -8.28231722e-02 -2.00731173e-01 -7.75984049e-01 -4.10433471e-01 -1.32442549e-01 -1.00322270e+00 1.48936844e+00 -6.84269071e-01 6.83196247e-01 1.92625090e-01 -8.63887787e-01 9.87841547e-01 -2.93781748e-03 2.71378666e-01 -1.10733557e+00 2.43753567e-01 1.65749729e-01 -3.59326690e-01 -7.94679880e-01 1.75355241e-01 -4.02403116e-01 6.30898118e-01 2.59419948e-01 -3.32764834e-01 3.18387784e-02 3.99261601e-02 -2.41670921e-01 5.73137820e-01 2.69963384e-01 2.08257958e-01 3.38724330e-02 8.60048532e-01 -7.00659037e-01 6.00502133e-01 5.29003441e-01 -2.43539125e-01 6.10046268e-01 -2.36674063e-02 -2.94207960e-01 -9.23720539e-01 -1.13543046e+00 -3.03283960e-01 7.92838156e-01 6.87982678e-01 5.25402501e-02 -5.49944699e-01 -2.05727711e-01 -6.38684630e-01 3.81493360e-01 -1.16110437e-01 -2.68627346e-01 -5.27228951e-01 -1.05354118e+00 6.29476756e-02 3.68353836e-02 1.11819303e+00 -7.90458620e-01 -5.77372849e-01 7.57704228e-02 -5.08345187e-01 -1.12943959e+00 -4.77132976e-01 -7.17075765e-02 -7.20581532e-01 -1.11384559e+00 -7.85819769e-01 -8.27573359e-01 7.72965848e-01 9.10505652e-01 9.35273051e-01 2.52324730e-01 -4.73184973e-01 9.00202431e-03 -1.02466382e-01 -2.44619533e-01 -4.30530719e-02 -5.50484657e-01 -1.69222236e-01 5.44266403e-01 3.70266498e-03 -7.55772293e-01 -1.21566939e+00 4.20580387e-01 -9.45017993e-01 2.59951621e-01 8.23652864e-01 1.15401721e+00 6.92755401e-01 6.07486188e-01 3.73087287e-01 -4.81101573e-01 4.86485094e-01 -1.86413303e-02 -7.93591976e-01 2.20555618e-01 -8.41101646e-01 -3.15826684e-01 8.65823150e-01 -1.26013115e-01 -1.70083904e+00 -2.15783149e-01 8.79596621e-02 -3.94667357e-01 -1.91415265e-01 2.71580845e-01 -4.08213764e-01 -3.53027821e-01 5.40598869e-01 7.02348650e-01 -1.00917004e-01 -4.49995995e-01 2.31808737e-01 6.87492013e-01 8.54215980e-01 -1.48337334e-01 9.95112062e-01 7.65241265e-01 2.72485763e-01 -9.16506469e-01 -9.67827559e-01 -4.59261745e-01 -1.41621247e-01 -4.39098775e-01 6.65806293e-01 -1.05259478e+00 -8.82041574e-01 7.40748405e-01 -9.36075687e-01 -3.03797871e-01 -7.64197931e-02 4.86572564e-01 -4.00445849e-01 6.03725731e-01 -5.99802136e-01 -8.98280680e-01 -4.11737084e-01 -1.06515384e+00 9.71441984e-01 5.91362834e-01 7.94813037e-01 -7.86461055e-01 -1.60344347e-01 5.85394859e-01 4.46454465e-01 1.06561944e-01 8.66409123e-01 6.08627856e-01 -8.74888659e-01 1.62979394e-01 -8.58171821e-01 7.27250457e-01 3.90526414e-01 -3.89263719e-01 -1.19740367e+00 -4.69700664e-01 4.21116322e-01 -2.50541836e-01 7.76518583e-01 6.33727491e-01 1.36556482e+00 -1.55695617e-01 -4.80337106e-02 1.14270806e+00 1.74371016e+00 1.51688367e-01 9.98887062e-01 3.28520685e-01 6.31832123e-01 5.24371684e-01 7.58214593e-01 4.00996745e-01 8.57673064e-02 5.45976818e-01 5.94609201e-01 -9.00373220e-01 -5.50640464e-01 -1.62236318e-01 1.35586306e-01 5.93891442e-01 -2.40853816e-01 -1.50049508e-01 -4.71878499e-01 4.13459539e-01 -1.39808822e+00 -8.58776450e-01 -3.54951948e-01 2.13130450e+00 9.89168704e-01 -1.58609331e-01 -3.98135543e-01 1.08172409e-01 7.26855874e-01 2.97764421e-01 -6.02217853e-01 3.69600683e-01 -4.15862978e-01 1.31261259e-01 5.12010396e-01 5.43090761e-01 -9.05072570e-01 6.12876654e-01 5.55606127e+00 7.59035170e-01 -1.29102361e+00 1.52590334e-01 8.41624737e-01 -5.50929643e-02 -2.78348923e-01 9.20376740e-03 -4.79582012e-01 6.81388617e-01 1.10771686e-01 -8.02095532e-02 7.46217489e-01 2.23556802e-01 8.85096073e-01 -2.63760209e-01 -5.51300049e-01 1.36814404e+00 1.73769966e-01 -8.19290519e-01 -1.59042463e-01 -8.24004635e-02 8.37698638e-01 -2.10094318e-01 3.40253443e-01 -2.23994136e-01 -9.32865068e-02 -8.59829366e-01 2.93477267e-01 6.26450419e-01 1.07214952e+00 -7.01554775e-01 7.01606512e-01 2.30428800e-01 -1.07949436e+00 -1.64547294e-01 -5.06376326e-01 5.94618209e-02 2.31905863e-01 8.79860163e-01 -2.63358325e-01 6.15495622e-01 8.18748355e-01 1.04648709e+00 -4.83166516e-01 1.16269350e+00 -5.19251466e-01 4.80368584e-01 -1.57245789e-02 3.57644200e-01 1.78497657e-02 -6.75890982e-01 5.95986485e-01 9.01621342e-01 1.65188670e-01 3.87661666e-01 1.36046022e-01 8.61573458e-01 2.34531742e-02 4.70377831e-03 -5.29720545e-01 3.94275963e-01 -7.44711235e-03 1.49358666e+00 -4.78576869e-01 -6.76310509e-02 -4.92908299e-01 8.43413472e-01 -1.97658211e-01 8.30235183e-01 -7.12517619e-01 -6.22870624e-01 3.74284953e-01 2.25808814e-01 -1.87502787e-01 -4.81075086e-02 -2.97494173e-01 -1.25946617e+00 3.84399891e-01 -7.63601184e-01 4.26377077e-03 -1.29283202e+00 -1.26788199e+00 3.85178298e-01 -3.19866329e-01 -1.41846132e+00 4.76699173e-01 -5.15300512e-01 -6.42845035e-01 1.11780512e+00 -2.37985277e+00 -1.16698754e+00 -8.51877809e-01 8.84178042e-01 4.15750772e-01 4.32808511e-02 1.45228267e-01 6.17260277e-01 -7.09148228e-01 3.08089942e-01 3.19794685e-01 -1.00890316e-01 8.95602882e-01 -7.68392444e-01 -3.35136741e-01 1.32139230e+00 -2.96480685e-01 4.91343260e-01 5.55846393e-01 -3.62412959e-01 -1.43257248e+00 -1.12126291e+00 3.68338585e-01 1.27037629e-01 3.23814005e-01 -5.75108267e-02 -8.07819486e-01 2.55395889e-01 3.29277962e-01 2.90378213e-01 2.90003300e-01 -2.35573322e-01 2.48601865e-02 -6.41848266e-01 -1.06370497e+00 6.76900208e-01 9.85308290e-01 -5.43089688e-01 -2.61043012e-01 3.39498848e-01 3.18139017e-01 -3.09338868e-01 -5.53211927e-01 4.52492177e-01 4.94933397e-01 -1.34350669e+00 1.17335558e+00 -1.59082338e-01 6.77802622e-01 -7.07592249e-01 -9.42871422e-02 -1.30530739e+00 -4.41694409e-01 -6.92442894e-01 2.58972973e-01 1.40975559e+00 -2.92717833e-02 -7.78678596e-01 5.72161674e-01 1.38807237e-01 -2.22545207e-01 -6.74351215e-01 -5.55030406e-01 -5.72345674e-01 -4.65719074e-01 -1.71648756e-01 4.60151017e-01 8.46911013e-01 -4.22081262e-01 3.72858524e-01 -6.29121900e-01 4.70055461e-01 1.18770111e+00 2.44209990e-01 7.18202710e-01 -8.57382894e-01 -1.09791502e-01 -2.35192478e-01 -6.59210980e-02 -1.21615124e+00 -7.52574131e-02 -7.43163407e-01 3.80383372e-01 -1.58669865e+00 4.46620256e-01 -3.73603314e-01 -4.82082784e-01 3.09659004e-01 -3.84367734e-01 7.10934103e-01 -1.28394654e-02 2.61364698e-01 -4.67898518e-01 7.34374464e-01 1.61596215e+00 -2.39805847e-01 -1.93691790e-01 -2.53262669e-01 -7.69920230e-01 7.96129704e-01 5.81954777e-01 -1.46916211e-01 -5.46603143e-01 -5.82384050e-01 3.25778097e-01 4.30603959e-02 6.28370106e-01 -9.57035303e-01 2.68944383e-01 -4.31396157e-01 6.62242711e-01 -4.75040048e-01 2.28236049e-01 -9.45946038e-01 -1.06743976e-01 1.56835765e-01 -1.29455835e-01 -5.26916921e-01 -8.19988996e-02 7.10862398e-01 -3.83735389e-01 -1.56021759e-01 1.20442379e+00 -9.92249697e-02 -5.68507850e-01 4.16615307e-01 -6.75822124e-02 -2.44791117e-02 7.57000327e-01 -3.66099745e-01 -5.45053720e-01 -2.62142181e-01 -2.27569416e-01 1.97399165e-02 3.80758613e-01 5.22202626e-02 9.47953403e-01 -1.27609575e+00 -7.06543386e-01 5.16880572e-01 9.95893553e-02 -4.25720103e-02 7.42404997e-01 1.08440268e+00 -4.03140664e-01 -7.60456994e-02 -3.96468431e-01 -6.18839085e-01 -1.24330223e+00 5.17083347e-01 4.74379212e-01 -1.06629893e-01 -6.28330946e-01 5.42263389e-01 5.52498162e-01 1.61579568e-02 -1.45855814e-01 -1.63739815e-01 -2.68395901e-01 -3.36100221e-01 7.70468533e-01 5.27610958e-01 6.01952858e-02 -4.90273118e-01 4.31284346e-02 9.59108889e-01 8.63873884e-02 1.79785922e-01 1.40235686e+00 -7.83864617e-01 -3.91676098e-01 1.77727520e-01 1.19303739e+00 7.95897469e-02 -1.37632406e+00 -3.85853261e-01 -7.11931407e-01 -1.11441171e+00 7.09892631e-01 -7.70160258e-01 -1.33288598e+00 1.03963280e+00 9.37979877e-01 6.67711422e-02 1.91079104e+00 -5.64610660e-01 9.05744374e-01 1.40038684e-01 2.32653335e-01 -9.81532872e-01 2.42435083e-01 -3.64087075e-02 6.29788399e-01 -1.54390955e+00 1.43557519e-01 -5.12196660e-01 -1.09437980e-01 1.11803770e+00 6.02767050e-01 2.19902620e-01 4.00628388e-01 1.10929936e-01 1.88207269e-01 -1.00752190e-01 -3.53545189e-01 -3.48845154e-01 1.77744046e-01 7.73723304e-01 2.45914951e-01 -3.73703182e-01 -1.92958310e-01 1.99049279e-01 2.53024340e-01 2.41161823e-01 4.73044842e-01 4.86936837e-01 -4.81869191e-01 -6.37904465e-01 -5.44947505e-01 3.31533283e-01 -4.43409383e-01 -3.41527671e-01 2.14046568e-01 3.13693970e-01 3.52856606e-01 1.33089530e+00 -2.23440930e-01 -1.10415958e-01 3.99836659e-01 -5.21128595e-01 4.45416957e-01 -2.26026610e-01 -1.62725255e-01 3.92433256e-01 -3.51197779e-01 -4.33027208e-01 -6.75364554e-01 -2.88734049e-01 -7.15122223e-01 -1.08923279e-01 -4.48760569e-01 -2.93626368e-01 4.34297562e-01 7.15898037e-01 7.38306344e-02 4.93576795e-01 9.92487669e-01 -8.54411662e-01 -2.54699141e-01 -8.22182059e-01 -9.15832162e-01 5.67429006e-01 6.98052585e-01 -5.44138312e-01 -5.33465803e-01 1.95431247e-01]
[10.817492485046387, -2.504446268081665]
8cbd803e-8b26-4355-adb5-7c1441ef10ca
asvspoof-2019-spoofing-countermeasures-for
2102.05889
null
https://arxiv.org/abs/2102.05889v1
https://arxiv.org/pdf/2102.05889v1.pdf
ASVspoof 2019: spoofing countermeasures for the detection of synthesized, converted and replayed speech
The ASVspoof initiative was conceived to spearhead research in anti-spoofing for automatic speaker verification (ASV). This paper describes the third in a series of bi-annual challenges: ASVspoof 2019. With the challenge database and protocols being described elsewhere, the focus of this paper is on results and the top performing single and ensemble system submissions from 62 teams, all of which out-perform the two baseline systems, often by a substantial margin. Deeper analyses shows that performance is dominated by specific conditions involving either specific spoofing attacks or specific acoustic environments. While fusion is shown to be particularly effective for the logical access scenario involving speech synthesis and voice conversion attacks, participants largely struggled to apply fusion successfully for the physical access scenario involving simulated replay attacks. This is likely the result of a lack of system complementarity, while oracle fusion experiments show clear potential to improve performance. Furthermore, while results for simulated data are promising, experiments with real replay data show a substantial gap, most likely due to the presence of additive noise in the latter. This finding, among others, leads to a number of ideas for further research and directions for future editions of the ASVspoof challenge.
['Kong Aik Lee', 'Junichi Yamagishi', 'Md Sahidullah', 'Héctor Delgado', 'Massimiliano Todisco', 'Ville Vestman', 'Tomi Kinnunen', 'Nicholas Evans', 'Xin Wang', 'Andreas Nautsch']
2021-02-11
null
null
null
null
['voice-anti-spoofing']
['audio']
[ 1.35911867e-01 -9.23002884e-02 2.46494696e-01 -3.35508287e-01 -1.20505166e+00 -7.80908823e-01 9.10560966e-01 3.72305103e-02 -3.51217479e-01 3.39546621e-01 6.72183633e-01 -6.62073314e-01 1.15881953e-02 1.23735242e-01 -5.02507031e-01 -5.89000523e-01 -3.14826593e-02 1.22130595e-01 -1.03724107e-01 -6.27050400e-01 7.09899515e-02 4.56735164e-01 -1.51066446e+00 2.67872006e-01 3.33192080e-01 4.98269975e-01 -1.28558949e-01 9.71235812e-01 1.10356130e-01 4.35559124e-01 -1.19063711e+00 -4.68595922e-01 7.84853473e-02 -2.96189547e-01 -6.78784311e-01 -3.56862783e-01 7.11584032e-01 -2.20399454e-01 -3.05216193e-01 9.19508755e-01 1.22527230e+00 1.17478929e-02 7.99387917e-02 -1.36799073e+00 4.83661406e-02 7.90270746e-01 -5.67633891e-03 3.63134801e-01 9.22947347e-01 3.74107242e-01 9.57763076e-01 -7.22078323e-01 2.91581720e-01 1.42018843e+00 1.03478587e+00 4.29466397e-01 -1.33612621e+00 -8.97472978e-01 -1.80693358e-01 4.30983268e-02 -1.42384171e+00 -1.34646881e+00 3.72595757e-01 -2.77903289e-01 1.19069457e+00 6.39228582e-01 3.54249060e-01 1.44950247e+00 -2.92807668e-01 6.25639617e-01 1.19212008e+00 -3.81853342e-01 1.32119536e-01 5.24316609e-01 3.62785161e-01 1.07286774e-01 4.97406758e-02 4.76525456e-01 -9.20717657e-01 -6.90743625e-01 -1.77613944e-01 -7.19226480e-01 -6.97300315e-01 4.09453288e-02 -1.29975641e+00 6.35662794e-01 -1.27541646e-01 4.53680545e-01 -1.68148786e-01 -2.44050309e-01 6.13013387e-01 6.52808785e-01 7.45644867e-02 3.46392304e-01 -3.91156852e-01 -5.59819043e-01 -1.31681347e+00 6.38430178e-01 1.11693108e+00 5.05358577e-01 1.40339404e-01 3.53177339e-01 -1.18173689e-01 8.63640845e-01 5.31547725e-01 7.79592574e-01 3.08419287e-01 -6.89753890e-01 6.47512734e-01 -3.70358258e-01 2.03482836e-01 -8.30461919e-01 -3.47664684e-01 -4.38644350e-01 -6.22488298e-02 3.01083356e-01 5.57061493e-01 -2.90244550e-01 -6.71526253e-01 1.84439504e+00 8.89951363e-02 1.76308528e-01 1.86214998e-01 8.32661211e-01 7.55873263e-01 4.46490675e-01 -2.25188926e-01 -3.20669234e-01 1.42992651e+00 -4.89994675e-01 -1.12647665e+00 -2.18437284e-01 4.98146206e-01 -1.30361176e+00 7.97228277e-01 4.65019852e-01 -1.06101871e+00 -3.24846327e-01 -1.20738816e+00 5.33346832e-01 -2.26420939e-01 -4.48867857e-01 2.68359751e-01 1.62981105e+00 -1.38690066e+00 1.53999597e-01 -6.47174478e-01 -4.73012388e-01 6.93230852e-02 3.80124271e-01 -4.59504128e-01 -8.58581290e-02 -1.42133749e+00 1.02887928e+00 -7.86042660e-02 3.94580513e-02 -8.13792765e-01 -6.46157444e-01 -8.03843081e-01 -1.75522223e-01 -1.16179839e-01 -2.69453615e-01 1.42804122e+00 -2.99531132e-01 -1.47356939e+00 6.33969963e-01 -4.14886504e-01 -7.51580775e-01 6.15550697e-01 -1.40008152e-01 -1.04895663e+00 -1.04635827e-01 -6.21545166e-02 1.81121975e-01 8.29839110e-01 -1.18544042e+00 -3.66353899e-01 -1.78031772e-01 -3.34790826e-01 5.13741747e-02 -1.24052726e-01 8.53355110e-01 9.59134195e-03 -4.76446956e-01 5.79207726e-02 -9.57223952e-01 1.95485264e-01 -9.04156804e-01 -4.43718761e-01 -6.83910176e-02 9.11976099e-01 -1.08565342e+00 1.13270164e+00 -2.42538500e+00 -2.99089313e-01 2.46321887e-01 -2.34663442e-01 6.18332326e-01 1.18914038e-01 6.82784915e-01 -3.28561366e-01 4.90372688e-01 -7.25206584e-02 -6.57070518e-01 2.12776154e-01 -3.16446811e-01 -6.48265898e-01 6.19736075e-01 -3.91636901e-02 2.40890160e-01 -6.17633939e-01 -2.33704582e-01 2.51785338e-01 7.41071641e-01 -2.65107185e-01 -2.86664627e-02 4.42661375e-01 2.85521358e-01 2.27488533e-01 5.83724260e-01 8.82358015e-01 5.27533948e-01 -1.41287759e-01 -1.73931867e-02 -2.11121187e-01 8.97009730e-01 -1.49343312e+00 1.23949015e+00 -2.09811226e-01 8.53804827e-01 9.29724455e-01 -6.53975010e-01 5.79665899e-01 9.33828294e-01 2.32107848e-01 -3.64390463e-01 9.91468877e-02 5.61366916e-01 2.82724798e-01 -4.36329871e-01 5.10285258e-01 -1.70703456e-01 8.05800185e-02 6.03670180e-01 1.05736606e-01 -3.43297929e-01 -2.67529160e-01 3.65783274e-01 1.06844163e+00 -4.59739000e-01 -1.59265086e-01 -2.31260657e-01 5.26913226e-01 -1.90923288e-01 3.98284078e-01 1.04868615e+00 -9.07435536e-01 7.57977784e-01 1.21319492e-03 3.01081002e-01 -6.22999966e-01 -1.00427270e+00 -3.90529424e-01 8.76258314e-01 -4.40542996e-02 -7.52578259e-01 -9.25317466e-01 -4.44944739e-01 -9.10083428e-02 1.00390613e+00 -1.65761143e-01 1.67448893e-01 -4.43571001e-01 -5.46780229e-01 1.31206882e+00 8.89001489e-02 3.19959283e-01 -8.90077710e-01 -1.87883750e-01 2.24877819e-01 -5.75926363e-01 -1.40305769e+00 -4.84970957e-01 1.71803462e-03 -3.43611032e-01 -5.85715413e-01 -6.86919332e-01 -5.89419305e-01 -2.13652864e-01 6.34439051e-01 6.39644206e-01 9.46365669e-02 1.80006716e-02 6.41859114e-01 -3.08897883e-01 -7.78233290e-01 -8.11605453e-01 -1.40323818e-01 5.68430245e-01 -2.91379206e-02 4.84184891e-01 -2.19851002e-01 -2.90138125e-01 5.90934157e-01 -4.80979025e-01 -5.73344052e-01 3.88827473e-02 8.95459294e-01 -3.38248104e-01 9.81922597e-02 7.47023165e-01 -1.60747647e-01 8.98791254e-01 -4.11918938e-01 -1.75712466e-01 3.93329784e-02 -5.50458312e-01 -2.90468335e-01 1.44889429e-02 -1.55231267e-01 -9.68336165e-01 -3.65730733e-01 -5.26807368e-01 -1.27729043e-01 -3.56838197e-01 3.21124852e-01 -2.65974492e-01 -1.63042709e-01 8.99299979e-01 1.36716202e-01 3.97946864e-01 -4.19580340e-01 -5.29650450e-02 1.30172682e+00 4.50625718e-01 -2.35937521e-01 8.23674619e-01 2.98625052e-01 -7.96101511e-01 -1.21679163e+00 4.13400494e-02 -4.68618691e-01 4.18720692e-02 -9.46126040e-03 6.61677420e-01 -1.07092106e+00 -6.95364773e-01 7.66307473e-01 -1.07546556e+00 -2.95366734e-01 1.21096551e-01 7.83376753e-01 -6.13989197e-02 5.19151926e-01 -5.17142534e-01 -1.19504917e+00 -1.81321472e-01 -1.61522412e+00 9.41508889e-01 -1.74359173e-01 -6.92142844e-01 -9.10908580e-01 6.37255535e-02 9.90089178e-01 7.66837955e-01 -2.59755373e-01 2.18238026e-01 -1.18646562e+00 -1.41711812e-02 -6.24562681e-01 2.35129297e-01 5.80249369e-01 7.38391057e-02 4.05371599e-02 -1.48631263e+00 -7.74022579e-01 1.87512219e-01 -2.71481544e-01 6.49799585e-01 1.92079142e-01 3.10981393e-01 -6.46725297e-02 -3.39745134e-01 2.47836947e-01 6.36249065e-01 -1.07652079e-02 3.89759988e-01 1.93503559e-01 3.78787875e-01 8.71162057e-01 3.22996080e-01 2.56535530e-01 3.77395988e-01 1.15375423e+00 1.76442221e-01 2.82802135e-01 -3.60954642e-01 -7.79206976e-02 1.00880885e+00 8.04448366e-01 3.64765972e-01 -2.79140055e-01 -1.14024746e+00 5.94876170e-01 -1.14862072e+00 -1.24254537e+00 -3.44436795e-01 2.52055120e+00 4.36443329e-01 1.23583376e-01 4.64266866e-01 7.01714814e-01 9.90011692e-01 3.81023377e-01 4.88635339e-02 -7.47151554e-01 -4.50175107e-01 1.54497653e-01 3.60958040e-01 9.53039944e-01 -7.89048553e-01 6.13351524e-01 7.29406166e+00 6.24402463e-01 -1.13777840e+00 2.48833805e-01 1.80692956e-01 -1.96234435e-01 -2.41910726e-01 3.06011718e-02 -8.99183929e-01 5.65450370e-01 1.43429410e+00 -3.65746929e-03 6.87427223e-01 1.37762189e-01 1.33375287e-01 1.53768897e-01 -7.85624743e-01 9.83138621e-01 2.85052329e-01 -8.96913230e-01 -5.80128372e-01 2.96576083e-01 3.77391309e-01 5.92297316e-01 1.36935070e-01 2.71621197e-01 3.47151726e-01 -1.14364207e+00 1.01307905e+00 -2.77281314e-01 4.87929523e-01 -3.92081529e-01 8.46837163e-01 3.18268359e-01 -8.88156831e-01 -2.04618484e-01 2.52476603e-01 -2.02024598e-02 4.13627893e-01 2.53511757e-01 -1.23343146e+00 5.06120741e-01 7.61644661e-01 -8.64164010e-02 -3.34956557e-01 9.84578669e-01 -8.35014209e-02 1.24261153e+00 -6.62147760e-01 6.22380525e-02 -1.13218874e-01 6.37569904e-01 1.03824532e+00 1.69116318e+00 2.50640720e-01 -1.75431803e-01 -2.18320504e-01 1.49029121e-01 3.39775324e-01 -1.57836284e-02 -6.33668423e-01 -1.12992510e-01 8.90606940e-01 1.00730956e+00 -2.29774624e-01 -1.02079391e-01 -2.82854527e-01 7.11374462e-01 -3.27724159e-01 4.62704062e-01 -6.55421495e-01 -3.29312533e-01 9.34152663e-01 3.09997294e-02 1.32213727e-01 -2.22382486e-01 -4.15184259e-01 -9.13894951e-01 -9.14142206e-02 -1.46039581e+00 3.56618971e-01 -4.14785504e-01 -7.57829487e-01 5.75887978e-01 -1.10393211e-01 -8.52119565e-01 -4.05645907e-01 -2.16474190e-01 -6.87211215e-01 1.17912376e+00 -1.08728468e+00 -8.79046321e-01 -3.97878736e-02 6.02561533e-01 2.40908787e-01 -4.98895705e-01 1.08549082e+00 4.70097452e-01 -4.67797071e-01 1.18607402e+00 -7.43676648e-02 -1.44937202e-01 1.04367781e+00 -9.17137027e-01 7.82243252e-01 1.01582944e+00 3.18495810e-01 9.37740207e-01 1.32544756e+00 -4.54629511e-01 -1.42061961e+00 -2.77579963e-01 1.16753781e+00 -7.34713495e-01 5.80759823e-01 -5.04288137e-01 -8.44814301e-01 5.32544971e-01 5.41241825e-01 -8.79253522e-02 8.71681094e-01 2.80358404e-01 -5.62750161e-01 8.10030848e-02 -1.40982556e+00 3.36226523e-01 8.23180079e-01 -9.64395940e-01 -6.04552388e-01 1.90268859e-01 5.88775396e-01 -3.69712323e-01 -5.65781295e-01 3.45535517e-01 6.46540582e-01 -1.13085985e+00 9.07061040e-01 -3.20723772e-01 -4.75978106e-01 -2.87742168e-01 -5.74310720e-01 -1.43308282e+00 7.10120350e-02 -1.33985209e+00 1.74638972e-01 1.77677774e+00 5.74301004e-01 -9.86490011e-01 5.01175761e-01 6.21936083e-01 -1.41484082e-01 3.55118215e-02 -1.15728116e+00 -7.87748873e-01 -9.23859030e-02 -9.92663980e-01 7.88089871e-01 1.02864254e+00 1.92216918e-01 1.72783971e-01 -3.62884343e-01 4.68518376e-01 5.84088147e-01 -7.40169287e-01 1.06571531e+00 -7.65975177e-01 -4.47384149e-01 -6.48123205e-01 -5.90908706e-01 -5.51385999e-01 6.87277168e-02 -7.35214472e-01 9.11764801e-02 -1.11555815e+00 -4.13307279e-01 -3.73556286e-01 -3.81369174e-01 7.16847852e-02 -1.78635627e-01 1.19643733e-01 6.15564942e-01 1.91124771e-02 9.90633890e-02 -3.94277647e-02 4.68079895e-01 -2.15813026e-01 -2.77878702e-01 4.23098326e-01 -9.06249225e-01 2.61576474e-01 8.44185889e-01 -2.56268740e-01 -3.32473218e-01 -3.35870743e-01 -2.47554749e-01 1.63247943e-01 4.51414347e-01 -1.21342754e+00 2.78094739e-01 3.15899223e-01 -1.42017081e-01 -4.14723933e-01 7.48615503e-01 -5.62433898e-01 2.77278543e-01 4.19383168e-01 -3.07517886e-01 -3.24465111e-02 5.53485870e-01 2.16700017e-01 -2.33523071e-01 -2.97913663e-02 5.68171442e-01 4.46155041e-01 -1.78247929e-01 -3.71477365e-01 -6.56795442e-01 -5.75060472e-02 5.48075557e-01 -2.68443227e-01 -3.48596811e-01 -8.11042190e-01 -5.83970666e-01 -1.77783575e-02 3.15485001e-01 5.96850336e-01 2.87115186e-01 -9.66751993e-01 -1.10631073e+00 4.95700330e-01 -3.57736088e-02 -7.54837751e-01 3.99778932e-01 9.60395277e-01 -2.16891810e-01 4.64616865e-01 1.74987018e-01 -6.46124184e-01 -1.76762855e+00 1.35268271e-01 3.99065375e-01 1.21233091e-01 -3.41947943e-01 1.24293733e+00 -2.72157609e-01 -6.39390528e-01 6.11710429e-01 2.35530898e-01 1.00550093e-01 7.90894926e-02 1.02379262e+00 4.32469487e-01 4.91564184e-01 -1.24405694e+00 -7.87821054e-01 -1.32244825e-02 -1.45963162e-01 -8.48259807e-01 8.17510068e-01 -1.21733151e-01 2.58495659e-01 2.55405247e-01 1.25176525e+00 5.94257832e-01 -7.94175923e-01 -1.82633102e-01 -1.27734691e-01 -3.66485476e-01 1.60115987e-01 -1.03736782e+00 -5.28649390e-01 9.46615934e-01 9.54900503e-01 3.68549943e-01 6.49331689e-01 -1.85808703e-01 7.58633375e-01 -1.22500574e-02 3.88833582e-01 -9.01794851e-01 -4.96350884e-01 3.82362187e-01 8.56733561e-01 -1.12898672e+00 -1.39857948e-01 -1.63350090e-01 -6.59864247e-01 7.01625288e-01 8.63728225e-02 5.14661491e-01 7.15768158e-01 4.28208143e-01 4.58308488e-01 -2.47393874e-03 -3.59663755e-01 1.79242387e-01 -2.28413641e-01 8.30800951e-01 5.50031245e-01 3.06494057e-01 2.42113136e-02 3.90641183e-01 -8.89314950e-01 -4.31125224e-01 3.07420164e-01 8.53049695e-01 -2.08265796e-01 -1.23386288e+00 -1.06167185e+00 5.44244498e-02 -7.87344515e-01 -7.89293200e-02 -5.84303439e-01 5.75215518e-01 -1.37761250e-01 1.68662131e+00 -2.38291770e-01 -7.84175813e-01 4.09387529e-01 3.44805032e-01 1.68240085e-01 -3.03606927e-01 -1.13876438e+00 1.74353208e-05 6.00710571e-01 -4.53982860e-01 -1.29762158e-01 -1.33763027e+00 -9.61613119e-01 -7.70972848e-01 -6.98811412e-01 4.19247895e-01 1.32270503e+00 8.94044220e-01 3.29518050e-01 2.95922428e-01 6.40687406e-01 -8.25923145e-01 -9.23530459e-01 -1.15697503e+00 -3.64308387e-01 3.54353577e-01 8.21009696e-01 -3.89762819e-01 -6.37971580e-01 -3.24181706e-01]
[14.188016891479492, 5.991428852081299]
9605487c-caf9-4547-b2c9-b650c6811dc7
lexical-complexity-prediction-an-overview
2303.04851
null
https://arxiv.org/abs/2303.04851v1
https://arxiv.org/pdf/2303.04851v1.pdf
Lexical Complexity Prediction: An Overview
The occurrence of unknown words in texts significantly hinders reading comprehension. To improve accessibility for specific target populations, computational modelling has been applied to identify complex words in texts and substitute them for simpler alternatives. In this paper, we present an overview of computational approaches to lexical complexity prediction focusing on the work carried out on English data. We survey relevant approaches to this problem which include traditional machine learning classifiers (e.g. SVMs, logistic regression) and deep neural networks as well as a variety of features, such as those inspired by literature in psycholinguistics as well as word frequency, word length, and many others. Furthermore, we introduce readers to past competitions and available datasets created on this topic. Finally, we include brief sections on applications of lexical complexity prediction, such as readability and text simplification, together with related studies on languages other than English.
['Matthew Shardlow', 'Marcos Zampieri', 'Kai North']
2023-03-08
null
null
null
null
['lexical-complexity-prediction', 'reading-comprehension']
['natural-language-processing', 'natural-language-processing']
[ 2.26058602e-01 2.29197755e-01 -3.45053613e-01 -2.70969093e-01 -4.69447047e-01 -5.28392076e-01 3.78364861e-01 8.76301646e-01 -1.02367473e+00 7.96218455e-01 5.68960369e-01 -7.48047233e-01 -2.36010760e-01 -6.64445221e-01 -4.45625961e-01 -1.02727693e-02 4.01103824e-01 4.91623670e-01 -2.16224208e-01 -5.29661894e-01 7.60706365e-01 2.44495243e-01 -1.94462276e+00 8.32497999e-02 1.52700889e+00 3.00622106e-01 6.59724474e-01 6.51048720e-01 -3.07742447e-01 4.05269086e-01 -7.91674674e-01 -8.53737175e-01 -2.41795704e-01 -2.40569830e-01 -8.97876561e-01 -4.38866407e-01 8.94769013e-01 -2.95999378e-01 -4.95538980e-01 8.04726303e-01 6.34804189e-01 3.45751107e-01 1.00096202e+00 -3.28521192e-01 -1.16129565e+00 1.00673568e+00 1.35122091e-01 7.79325604e-01 9.00441051e-01 1.51192740e-01 1.14833486e+00 -8.48031521e-01 4.23425257e-01 1.30204785e+00 7.65677035e-01 4.54089820e-01 -1.03628039e+00 -2.71701962e-01 3.20359498e-01 7.01967597e-01 -9.56243992e-01 -5.32599866e-01 3.35564971e-01 -3.77193421e-01 1.67117536e+00 5.65645576e-01 9.76432502e-01 1.16196263e+00 3.36300015e-01 8.95184994e-01 1.38404155e+00 -1.23906291e+00 -2.76228875e-01 2.46535689e-01 7.64863193e-01 6.03181005e-01 6.25414133e-01 1.22628473e-01 -6.96014881e-01 2.80281186e-01 2.38139749e-01 -7.77271509e-01 -1.18458897e-01 2.54879981e-01 -9.65172291e-01 1.15973806e+00 -1.32531419e-01 2.54694402e-01 4.81425934e-02 -7.02002227e-01 4.33707833e-01 6.29572451e-01 4.79450464e-01 9.83534634e-01 -9.35321271e-01 -1.89904809e-01 -7.44839072e-01 4.47541177e-01 1.06234694e+00 8.31160605e-01 2.82093376e-01 1.83327556e-01 -2.95220166e-01 1.34662676e+00 7.80838430e-02 5.64000368e-01 8.38618100e-01 -3.00119668e-01 7.48684287e-01 1.36423036e-01 -3.91626626e-01 -8.52976382e-01 -1.02410662e+00 -7.69728050e-02 -4.97993678e-01 -2.77211756e-01 6.68946385e-01 -5.14071211e-02 -4.71084416e-01 1.60926354e+00 -1.38843074e-01 -8.91349316e-01 -4.12028939e-01 2.73299724e-01 1.11623967e+00 5.01853287e-01 4.34969068e-01 -4.07098234e-01 1.48263192e+00 -9.94790375e-01 -9.95966017e-01 -4.31397021e-01 9.94461477e-01 -8.11173975e-01 1.47016776e+00 5.33739150e-01 -1.42703569e+00 -6.77462816e-01 -9.52756822e-01 -7.01009393e-01 -9.72753525e-01 3.81064005e-02 8.43836069e-01 1.19560325e+00 -9.19210255e-01 6.61888123e-01 -4.54115242e-01 -5.54237902e-01 2.56276757e-01 4.47218776e-01 -2.25791395e-01 3.37330177e-02 -1.62424076e+00 1.90104508e+00 6.78799033e-01 -2.78067738e-01 1.87589531e-03 -5.28136313e-01 -1.25561035e+00 8.45526978e-02 5.79949655e-02 -6.82928085e-01 1.15123034e+00 -1.08171022e+00 -1.64465857e+00 1.13536263e+00 -1.66790158e-01 -2.14954272e-01 3.34341750e-02 -5.49565732e-01 -4.16093200e-01 -3.98977548e-01 8.36518258e-02 4.99473780e-01 5.04083991e-01 -1.99373826e-01 -7.48703361e-01 -2.46087164e-01 1.28067032e-01 7.67708480e-01 -5.03016412e-01 4.57366586e-01 2.07751229e-01 -9.77660239e-01 -3.85819793e-01 -6.28525734e-01 2.12463990e-01 -2.93204069e-01 -3.06141615e-01 -8.44753027e-01 -3.55464458e-01 -1.47567451e+00 1.78837800e+00 -1.45715141e+00 4.42631721e-01 2.79031564e-02 3.97768855e-01 3.83856744e-01 -2.36181319e-01 5.38177252e-01 -1.19832054e-01 4.36247557e-01 1.72500849e-01 -8.68449584e-02 1.77392170e-01 -9.50856730e-02 4.34876323e-01 3.48953784e-01 1.35656342e-01 1.16760433e+00 -6.06397867e-01 -3.42026144e-01 2.61815846e-01 5.85101843e-02 -5.73752761e-01 -1.67197034e-01 -8.65403190e-02 -1.39370158e-01 2.47587740e-01 6.00794792e-01 5.11533856e-01 2.90851086e-01 -2.40151640e-02 4.12464917e-01 -4.62878108e-01 1.16831684e+00 -6.60618782e-01 1.14508486e+00 -7.60607600e-01 1.22400880e+00 -4.31985527e-01 -9.11821306e-01 3.46039832e-01 -1.56672791e-01 -5.73312283e-01 -9.40551400e-01 4.24207091e-01 1.32792175e-01 7.69787908e-01 -6.93947732e-01 7.17925847e-01 2.50756055e-01 4.96675782e-02 3.74508381e-01 -5.47720566e-02 -1.44278690e-01 4.91301000e-01 -3.05071741e-01 5.59383631e-01 -4.43416461e-02 8.72718573e-01 -5.42615891e-01 6.48383021e-01 -1.36006728e-01 -1.44356787e-01 1.02921331e+00 -7.74196237e-02 3.41231115e-02 1.86319694e-01 -4.08287078e-01 -1.39909148e+00 -7.03202307e-01 -5.88821173e-01 1.73416328e+00 -4.58352357e-01 -6.99775696e-01 -1.05964351e+00 -1.14859780e-02 -1.34653956e-01 1.26726711e+00 -5.62016368e-01 -4.35356617e-01 -8.26235890e-01 -7.51962483e-01 3.67839664e-01 5.35419643e-01 1.53032215e-02 -1.34025300e+00 -5.46528399e-01 2.25754976e-01 -3.98199975e-01 -9.12332892e-01 -1.33606836e-01 2.59134114e-01 -8.68616641e-01 -6.45304799e-01 -6.19065702e-01 -1.18697190e+00 1.85887247e-01 1.29671708e-01 1.49729395e+00 4.41505998e-01 -3.66358727e-01 1.19715743e-01 -6.32490456e-01 -9.60822105e-01 -4.85116154e-01 7.68411815e-01 2.35227749e-01 -1.08886552e+00 9.23018694e-01 2.32822858e-02 -3.90918776e-02 -4.32912737e-01 -5.84802687e-01 1.41136706e-01 7.68132448e-01 1.11206138e+00 4.86011468e-02 -6.11386932e-02 3.78134221e-01 -9.27160084e-01 1.20979726e+00 -3.70006710e-01 -2.97401994e-01 2.78715104e-01 -7.06265390e-01 -1.77274019e-01 4.98060018e-01 -8.75561297e-01 -7.95863688e-01 -3.28958333e-01 -5.57120442e-01 7.19365001e-01 -3.21637243e-01 9.14668262e-01 -2.79708833e-01 -2.98685998e-01 9.47383225e-01 3.35900337e-01 1.29973873e-01 -4.79192793e-01 4.11587656e-01 6.53486609e-01 -1.21727146e-01 -3.20340693e-01 6.24933600e-01 -6.01014912e-01 -4.70299602e-01 -1.65080941e+00 -9.01471257e-01 1.00599319e-01 -9.58269656e-01 1.21844880e-01 6.65479600e-01 -6.97677791e-01 -4.48751986e-01 7.66696632e-01 -1.10553551e+00 -2.82834411e-01 1.90041158e-02 7.09210634e-01 -5.18582165e-01 3.98524910e-01 -6.53075933e-01 -5.51759243e-01 -2.82745987e-01 -8.45553875e-01 6.67867422e-01 4.42759931e-01 -8.93099070e-01 -1.23772120e+00 -2.25890234e-01 5.41091979e-01 5.87543130e-01 -3.19059998e-01 1.75949788e+00 -1.00831246e+00 -2.50977427e-02 -5.46776429e-02 1.49828762e-01 2.51342684e-01 -2.90570557e-01 -2.53609791e-02 -5.60868204e-01 6.49500117e-02 -2.98895277e-02 -4.44926351e-01 7.56400943e-01 8.25861573e-01 1.01960647e+00 -4.05661702e-01 -1.10191390e-01 4.70709682e-01 8.99605572e-01 -7.29280384e-03 5.92127442e-01 8.08395803e-01 7.59558797e-01 9.95786369e-01 4.52350318e-01 1.48541495e-01 7.97811985e-01 4.97524053e-01 -3.69601786e-01 7.10964156e-03 -2.69436270e-01 -1.89233944e-01 2.32344881e-01 1.21900749e+00 -8.35650042e-02 -3.92157227e-01 -8.66224706e-01 2.33756319e-01 -1.01466358e+00 -8.68345380e-01 -4.09091473e-01 2.06977057e+00 1.09678972e+00 2.35371605e-01 3.56081694e-01 5.68063200e-01 7.00933456e-01 7.53291622e-02 -3.18898141e-01 -1.16569090e+00 -7.53697276e-01 6.71570241e-01 3.98631334e-01 8.68537486e-01 -1.15188444e+00 1.61251938e+00 7.46463442e+00 1.06234515e+00 -5.86675584e-01 -1.57301445e-02 5.76056182e-01 5.18893497e-03 -2.41742060e-01 -3.42359662e-01 -1.01645494e+00 2.09486410e-01 1.20974994e+00 -4.36411232e-01 7.84520924e-01 5.22205353e-01 5.88495769e-02 -6.13640070e-01 -1.04261291e+00 8.45994830e-01 5.24433672e-01 -7.79206097e-01 2.51962423e-01 -2.38794178e-01 4.59287226e-01 -2.07276672e-01 1.81388736e-01 6.42165899e-01 -3.44501972e-01 -1.36192036e+00 7.46921241e-01 5.25607407e-01 7.89097071e-01 -6.93023443e-01 6.38305426e-01 4.56590235e-01 -5.56159139e-01 -3.99038941e-02 -7.09114492e-01 -1.01575220e+00 -2.31100559e-01 2.14044303e-01 -1.75126925e-01 -6.01070486e-02 6.85998559e-01 4.51834828e-01 -1.14310265e+00 1.11453259e+00 -3.86529654e-01 5.24777472e-01 -2.41451859e-01 -1.05574894e+00 -2.31362000e-01 7.66384602e-02 4.08192188e-01 1.33846295e+00 -1.15095392e-01 2.19572261e-01 -3.39766771e-01 4.06734765e-01 2.95828462e-01 1.02419221e+00 -5.23936033e-01 -1.62188023e-01 5.88130951e-01 7.97116220e-01 -4.87791389e-01 -1.62372440e-01 -8.87892187e-01 8.48381102e-01 8.86154532e-01 3.39505990e-04 -6.40772462e-01 -8.63899469e-01 4.36960995e-01 7.48296827e-02 -1.28506392e-01 -4.00104851e-01 -9.39860880e-01 -1.27290452e+00 6.19443767e-02 -1.30764377e+00 2.21108139e-01 -4.45210636e-01 -1.52409124e+00 3.92943114e-01 1.44363582e-01 -5.41746974e-01 1.23415090e-01 -1.20828331e+00 -3.68931562e-01 1.28858662e+00 -1.27035618e+00 -7.31631577e-01 4.20464054e-02 2.34653413e-01 1.05355585e+00 -4.02706683e-01 8.77551734e-01 4.14947141e-03 -5.74742019e-01 1.05536342e+00 3.61163974e-01 -2.46469110e-01 5.97733080e-01 -1.38051748e+00 8.80799413e-01 4.10557330e-01 -4.53165323e-02 6.15557909e-01 5.72200060e-01 -8.03852737e-01 -9.04860675e-01 -3.09387654e-01 1.77060628e+00 -7.37577140e-01 6.57125771e-01 -5.53960919e-01 -8.16188037e-01 4.76204365e-01 5.10005951e-01 -1.32936502e+00 9.30718422e-01 6.25955403e-01 -1.88076440e-02 3.92711937e-01 -7.32718527e-01 1.16120374e+00 1.22726893e+00 -4.72758532e-01 -1.13950396e+00 7.60902166e-01 6.44204676e-01 -5.41625619e-01 -6.54732049e-01 1.01663776e-01 6.85944915e-01 -8.27855825e-01 9.72118914e-01 -9.02906835e-01 6.50460720e-01 7.14834690e-01 2.78270334e-01 -1.74000239e+00 -9.65331674e-01 -3.47982824e-01 6.63688080e-03 7.57932544e-01 7.81493783e-01 -5.68909764e-01 4.92436320e-01 8.10951293e-01 -2.42166415e-01 -6.45809829e-01 -8.27203989e-01 -6.23584747e-01 9.18851972e-01 -2.79688656e-01 2.36775309e-01 8.23725879e-01 2.71823764e-01 7.48902857e-01 -7.61785433e-02 -2.62244254e-01 1.88959554e-01 -6.18386388e-01 2.57719249e-01 -1.56012118e+00 -3.82107608e-02 -1.10379326e+00 -2.56344050e-01 -9.67790544e-01 7.90637434e-01 -9.71766055e-01 -2.48900041e-01 -1.44031906e+00 -7.73900794e-03 8.27850029e-02 4.37742203e-01 4.66233166e-03 -8.70148122e-01 -2.64386445e-01 1.52360663e-01 -2.04240695e-01 -6.40053824e-02 4.78614837e-01 1.22553444e+00 -5.83642609e-02 -3.53045344e-01 8.90213698e-02 -1.02896821e+00 8.10436606e-01 1.30087864e+00 -2.06040174e-01 -2.02490479e-01 -8.68454516e-01 5.32406569e-01 -3.87744188e-01 -2.68086612e-01 -8.74844968e-01 -3.17665040e-02 -1.53468177e-01 8.76351953e-01 -4.33176726e-01 2.15412036e-01 -2.50345796e-01 -6.38792157e-01 5.69898069e-01 -5.09285092e-01 6.77159429e-01 5.18700123e-01 -4.75665042e-03 1.53624147e-01 -9.78849173e-01 7.01251328e-01 -1.09906316e-01 -3.57011497e-01 -3.44989479e-01 -1.16361761e+00 4.54719365e-01 7.58365989e-01 -3.52115422e-01 -4.76991504e-01 -5.63829064e-01 -7.88510561e-01 2.32006628e-02 1.00024007e-01 6.30252302e-01 4.25909668e-01 -9.36683536e-01 -9.70995069e-01 1.35318875e-01 7.01037124e-02 -7.90885270e-01 1.03518561e-01 6.57144964e-01 -9.37283993e-01 9.92098093e-01 -4.33777034e-01 2.19854668e-01 -1.32574606e+00 8.13229620e-01 1.91305280e-01 4.37405743e-02 -2.90994912e-01 9.52477753e-01 -1.64012849e-01 -5.42105794e-01 4.14559782e-01 -3.64790231e-01 -8.71201694e-01 2.48198047e-01 7.16912329e-01 8.02812338e-01 3.05432588e-01 -6.73030853e-01 -9.60023701e-02 6.33370459e-01 -4.70300436e-01 1.28774680e-02 1.00735962e+00 -4.55564022e-01 -2.77895033e-01 5.94404101e-01 9.26427424e-01 1.40030473e-01 -2.10458264e-01 -2.36235842e-01 3.30388486e-01 -2.22197995e-01 2.58709732e-02 -1.04700541e+00 -2.08318859e-01 8.39826822e-01 3.47215861e-01 1.95672333e-01 8.90337110e-01 -1.42405972e-01 5.47470510e-01 8.92586172e-01 -7.18159750e-02 -1.79341853e+00 -3.64908278e-01 1.20388603e+00 8.74366701e-01 -1.19506884e+00 1.95078894e-01 -6.08439744e-01 -5.18590391e-01 1.29461217e+00 7.62176335e-01 6.83271661e-02 6.44566774e-01 -1.68252692e-01 -3.58153582e-01 2.25400940e-01 -7.01731324e-01 -4.50579435e-01 5.24546862e-01 1.04360271e+00 9.15430307e-01 1.58524334e-01 -1.30309451e+00 9.53737319e-01 -1.16217434e+00 -5.13916135e-01 6.71194613e-01 7.53853083e-01 -5.80318451e-01 -1.04861629e+00 -4.79228914e-01 1.27711570e+00 -4.54499424e-01 -1.17403984e+00 -5.78877866e-01 1.01339996e+00 6.02296554e-02 9.65321958e-01 1.17341965e-01 -1.58180565e-01 5.68360865e-01 3.18692058e-01 7.64235377e-01 -8.80837679e-01 -8.49294305e-01 -2.74938494e-01 5.34930170e-01 -1.26075307e-02 1.08205087e-01 -1.10489893e+00 -4.94878083e-01 -7.18447924e-01 -6.84092581e-01 -3.69154781e-01 6.18718743e-01 1.00487077e+00 -3.77192646e-01 3.50393474e-01 1.51281673e-02 -8.56573761e-01 -4.54383194e-01 -1.53122711e+00 -5.91268957e-01 -4.10974734e-02 -9.26955119e-02 -5.46549499e-01 -4.43401009e-01 -1.13670103e-01]
[10.888261795043945, 10.337769508361816]
3e0648a8-bfbe-4896-a5df-d18c55f20c64
construction-of-segmentation-and-part-of
null
null
https://aclanthology.org/2022.lt4hala-1.23
https://aclanthology.org/2022.lt4hala-1.23.pdf
Construction of Segmentation and Part of Speech Annotation Model in Ancient Chinese
Among the four civilizations in the world with the longest history, only Chinese civilization has been inherited and never interrupted for 5000 years. An important factor is that the Chinese nation has the fine tradition of sorting out classics. Recording history with words, inheriting culture through continuous collation of indigenous accounts, and maintaining the spread of Chinese civilization. In this competition, the siku-roberta model was introduced into the part-of-speech tagging task of ancient Chinese by using the Zuozhuan data set, and good prediction results were obtained.
['Zhuying Z. Xia', 'Huyin H. Xie', 'Qinyu C. Chang', 'Longjie Jiang']
null
null
null
null
lt4hala-lrec-2022-6
['culture']
['speech']
[-3.55324298e-01 -2.32256711e-01 -2.14151844e-01 -3.51384670e-01 -3.54955345e-01 -5.97595334e-01 6.21655107e-01 -3.53372276e-01 -8.35875034e-01 1.07589042e+00 6.77607894e-01 -4.79672283e-01 2.03299612e-01 -8.18614900e-01 1.61766317e-02 -5.89311242e-01 1.81761961e-02 6.52401984e-01 -2.87843291e-02 -5.36603510e-01 3.58173639e-01 1.50195956e-01 -9.71421421e-01 -2.28731886e-01 8.48812044e-01 4.71076161e-01 2.52645046e-01 2.99262077e-01 -1.74916446e-01 6.25896215e-01 -8.19694757e-01 -5.85195780e-01 -7.95547217e-02 -8.28546345e-01 -9.27888513e-01 -1.98201239e-01 -2.32176691e-01 -2.75175065e-01 -4.78579670e-01 8.70318055e-01 2.28249699e-01 -1.15783811e-01 4.33543861e-01 -2.04027116e-01 -1.22135198e+00 1.52318430e+00 -3.01768363e-01 8.64109248e-02 2.03900829e-01 -7.19948351e-01 1.31396437e+00 -6.99190974e-01 7.13693023e-01 1.03128397e+00 3.04568231e-01 4.10206944e-01 -9.36975002e-01 -7.10083604e-01 -3.20746116e-02 1.30008250e-01 -1.79048789e+00 -3.88569385e-01 5.44659853e-01 -2.48505414e-01 1.07242060e+00 2.88895458e-01 1.21545613e+00 9.68694270e-01 1.54554337e-01 7.91215956e-01 9.53005493e-01 -4.57552761e-01 -7.83152655e-02 -2.33073413e-01 -8.82955045e-02 5.72066188e-01 1.10885352e-01 -5.96383661e-02 -4.00228858e-01 -5.55073172e-02 3.95538688e-01 -1.05973314e-02 -1.99236020e-01 8.17816079e-01 -1.37629294e+00 7.04024792e-01 2.53669947e-01 9.67561007e-01 4.71205637e-02 1.29191601e-03 2.75886536e-01 4.00191128e-01 4.78743672e-01 5.31457007e-01 -6.37278378e-01 -6.28122449e-01 -1.17102945e+00 8.43835101e-02 7.73430109e-01 9.18978751e-01 4.04994577e-01 2.05933124e-01 4.83634174e-01 1.02382004e+00 -9.76745784e-02 9.41842079e-01 7.95209825e-01 -7.48457432e-01 7.09762285e-03 3.78645241e-01 -2.61507601e-01 -4.45420086e-01 -2.16507152e-01 -5.67258120e-01 -7.46103644e-01 -2.89125472e-01 2.68966019e-01 -2.34512985e-01 -9.13489878e-01 1.46328866e+00 -2.09253475e-01 -5.19213378e-01 4.75254320e-02 6.62324190e-01 3.14485759e-01 9.06741500e-01 -9.77337286e-02 -2.21072614e-01 1.36283505e+00 -7.13399231e-01 -6.41578734e-01 -9.41575170e-02 3.06960404e-01 -7.15095282e-01 8.62935364e-01 4.77335721e-01 -8.24401498e-01 -3.44061494e-01 -1.07057106e+00 -1.28838643e-01 -2.73486018e-01 -2.01111883e-02 7.25053966e-01 6.32483482e-01 -7.07223594e-01 6.86238348e-01 -9.54003096e-01 -4.86380786e-01 3.19200963e-01 -2.14000374e-01 -2.64852345e-01 8.23125765e-02 -1.48630667e+00 1.12027788e+00 8.02399755e-01 4.99464385e-02 -7.37877548e-01 -4.48845893e-01 -4.36997443e-01 -7.81966075e-02 3.09286296e-01 -1.56406537e-01 9.66733217e-01 -9.46285427e-01 -1.43542445e+00 8.89251351e-01 9.27971303e-02 -2.49470294e-01 3.17130446e-01 -5.39531827e-01 -9.99205768e-01 -2.00628847e-01 1.37154669e-01 2.98183769e-01 2.63378590e-01 -7.88126051e-01 -8.91601384e-01 -3.67133588e-01 -3.61140758e-01 -1.22727312e-01 -3.75010312e-01 4.09103960e-01 -6.52451754e-01 -9.71117854e-01 1.38594583e-01 -9.73321617e-01 -1.39944151e-01 -8.24477077e-01 -3.46899927e-01 -1.26146287e-01 3.33845049e-01 -1.18783522e+00 1.71428561e+00 -2.40924096e+00 4.58323687e-01 2.68002540e-01 -6.01331480e-02 1.27530769e-02 2.02199996e-01 7.70693243e-01 3.02452177e-01 4.61303145e-01 -5.58684289e-01 8.09584837e-03 -1.49473384e-01 6.99061036e-01 -2.20441774e-01 3.29100758e-01 -1.46715194e-01 9.23143327e-01 -1.12358975e+00 -4.81077701e-01 -3.13927472e-01 4.49625045e-01 -3.68471086e-01 3.98802711e-03 4.92539778e-02 4.08562303e-01 -5.43276072e-01 9.00863886e-01 2.38669381e-01 -4.57956158e-02 5.48166871e-01 3.69926661e-01 -7.28677690e-01 4.28350151e-01 -3.04107159e-01 1.88395584e+00 -6.39020205e-02 5.04161060e-01 -2.50191003e-01 -5.62781632e-01 9.16398525e-01 3.99436831e-01 5.60683668e-01 -8.10491502e-01 3.59513253e-01 6.04927599e-01 2.97860771e-01 -3.12251478e-01 8.34903121e-01 -4.37655091e-01 -5.59744895e-01 3.12159866e-01 -2.11012438e-01 -2.94423699e-01 3.29395562e-01 2.02330470e-01 8.87681007e-01 3.68305862e-01 7.16661513e-01 -5.24492383e-01 2.95319766e-01 3.45318347e-01 9.30293143e-01 2.67687112e-01 1.17749989e-01 7.32346833e-01 1.27848893e-01 -6.01253986e-01 -1.37838173e+00 -1.24098825e+00 -4.78919655e-01 9.76011157e-01 -2.79808491e-01 -4.94800091e-01 -5.09887099e-01 -1.49363533e-01 -2.18779653e-01 8.55979800e-01 -5.31132519e-01 2.37326205e-01 -8.50414217e-01 -5.96827984e-01 1.00671744e+00 3.00563753e-01 7.50330985e-01 -1.12715101e+00 -5.22024214e-01 5.60491323e-01 -3.45940351e-01 -8.02374005e-01 -4.53046620e-01 9.55887809e-02 -5.02233863e-01 -7.17308640e-01 -1.27995253e+00 -8.70750785e-01 1.87072113e-01 -1.73622563e-01 1.04014957e+00 2.26932630e-01 -4.28636558e-02 -2.49538958e-01 -8.25108647e-01 -3.00609350e-01 -5.90786994e-01 4.84283149e-01 -5.28786965e-02 -4.42374974e-01 2.28461146e-01 -6.71326876e-01 -1.00551490e-02 8.39116350e-02 -6.22704685e-01 1.29714562e-03 5.84552765e-01 6.15270495e-01 7.40423054e-02 1.86756223e-01 6.14156127e-01 -1.05423117e+00 -9.68085676e-02 -4.60978091e-01 -2.66329050e-01 3.23074877e-01 -5.46981871e-01 -5.62300012e-02 6.11922562e-01 -3.10325742e-01 -1.23738420e+00 -2.12080330e-01 -5.78005612e-01 3.55648756e-01 1.82743177e-01 9.59529459e-01 -1.43967316e-01 6.45774722e-01 2.76328325e-01 3.37629437e-01 -3.49997878e-01 -7.13748872e-01 5.37381947e-01 8.42737973e-01 8.26995730e-01 -6.15080118e-01 6.76702976e-01 3.09521109e-01 -5.00681937e-01 -1.61492014e+00 -7.66143799e-01 -1.93519399e-01 -1.06700814e+00 -1.50248885e-01 1.14975166e+00 -1.03299975e+00 -3.27222317e-01 7.12633550e-01 -9.51723754e-01 -7.25484639e-02 -7.88323432e-02 6.17744207e-01 1.11101516e-01 4.51293617e-01 -9.03625369e-01 -8.35088491e-01 -1.37260109e-01 -4.54021335e-01 4.27177042e-01 2.03112215e-01 -5.14712155e-01 -7.30088830e-01 4.47083712e-01 -8.55865330e-02 3.78398299e-01 2.48610049e-01 1.19503593e+00 -6.20613754e-01 -2.25892514e-01 -3.75770405e-02 9.99825969e-02 5.04337668e-01 4.58266467e-01 1.45322740e-01 -6.04206979e-01 -2.83475667e-01 -1.62961259e-01 -2.88260505e-02 9.25125718e-01 4.71314453e-02 3.85729462e-01 -1.49451062e-01 -1.77069619e-01 4.34585065e-01 1.39472806e+00 7.57152736e-01 6.93949997e-01 4.17520642e-01 5.87878942e-01 3.72593284e-01 7.80594051e-02 3.37637454e-01 4.89441574e-01 1.40335754e-01 -3.03240001e-01 3.26584466e-02 -1.28716258e-02 -2.30438128e-01 4.55180854e-01 1.79606438e+00 -6.18761718e-01 -1.79970972e-02 -1.11742926e+00 9.14373279e-01 -1.47059095e+00 -1.00376773e+00 -4.11883406e-02 1.96044338e+00 1.01051033e+00 2.86044151e-01 1.74524523e-02 -1.12656236e-01 3.67143542e-01 2.01347068e-01 -8.12004581e-02 -8.20219144e-02 -5.92339575e-01 3.76352668e-01 5.46210647e-01 6.05849981e-01 -8.65392566e-01 1.32570446e+00 7.64835548e+00 5.99384665e-01 -9.76825535e-01 1.22811925e-03 2.31908411e-01 5.95898787e-03 -3.79148841e-01 3.94349277e-01 -7.01794267e-01 5.43521166e-01 8.61937404e-01 -4.44841713e-01 5.49796700e-01 4.57251579e-01 -1.58949569e-01 4.79038153e-03 -6.49573863e-01 5.10942101e-01 1.49088204e-01 -1.09147763e+00 -2.06584662e-01 2.13078052e-01 7.62295365e-01 5.37596107e-01 -4.01797920e-01 2.60360241e-01 5.01970589e-01 -8.46877992e-01 1.26389277e+00 6.27194166e-01 6.32549882e-01 -8.18508923e-01 5.25321901e-01 2.57045001e-01 -1.14422762e+00 1.57581363e-02 -4.17676866e-01 -2.39485249e-01 6.69916749e-01 5.71546793e-01 -2.76362836e-01 8.30278516e-01 7.23665476e-01 9.93121743e-01 -4.94765133e-01 5.86980999e-01 -4.85607624e-01 1.14414895e+00 -2.62300164e-01 -1.88251197e-01 3.21718693e-01 -6.18184924e-01 6.54367030e-01 1.27139282e+00 6.15043879e-01 4.14230555e-01 1.02116734e-01 5.77388167e-01 1.50291905e-01 1.36010453e-01 -3.54473054e-01 -5.00923097e-01 5.54238141e-01 7.86970198e-01 -7.81894445e-01 -2.73418903e-01 -4.56298500e-01 9.72310364e-01 2.08215714e-01 3.29098672e-01 -8.49218547e-01 -5.82997501e-01 5.12369335e-01 2.20687389e-02 3.02149951e-01 -8.02645922e-01 -4.58881222e-02 -1.25068295e+00 -2.67023057e-01 -6.06305897e-01 3.21595609e-01 -3.12327474e-01 -1.25181878e+00 9.03615057e-01 -9.67777818e-02 -6.79420948e-01 1.56625155e-02 -4.03456837e-01 -4.63331819e-01 7.37811446e-01 -1.12002599e+00 -1.36448765e+00 4.57938492e-01 1.46664709e-01 4.38131571e-01 -4.44663793e-01 1.01622105e+00 6.03807151e-01 -6.00958824e-01 1.95447683e-01 3.88637483e-01 6.04354441e-01 5.27449250e-01 -1.12251270e+00 5.84583640e-01 7.47307897e-01 2.35050365e-01 7.20815778e-01 4.81851339e-01 -9.70931351e-01 -9.52928782e-01 -5.84934890e-01 1.53087175e+00 -2.71844864e-01 1.02617145e+00 -5.63385487e-01 -9.51760650e-01 8.42949569e-01 4.52452749e-01 -1.09373581e+00 7.46744931e-01 3.21351230e-01 -3.91376853e-01 5.40838838e-02 -3.48970711e-01 6.25686467e-01 9.14509356e-01 -5.05072951e-01 -9.45261359e-01 -1.89226985e-01 9.89298105e-01 1.71533227e-01 -1.02216065e+00 -3.67279612e-02 1.16948891e+00 -3.98021400e-01 6.61569953e-01 -4.35457110e-01 3.74787360e-01 -1.96497157e-01 -4.30720836e-01 -1.26758552e+00 -7.65565693e-01 -7.46476531e-01 6.27152264e-01 1.66696310e+00 8.51668298e-01 -5.76080561e-01 3.68966967e-01 1.44367233e-01 -3.85776311e-01 -1.75669983e-01 -1.06357932e+00 -8.85518253e-01 4.71660405e-01 -1.53919756e-01 6.14653766e-01 1.12117779e+00 1.70354068e-01 6.46519303e-01 -7.45908678e-01 -3.13166022e-01 4.56827879e-01 4.05296832e-02 2.64516294e-01 -1.17194116e+00 -1.31296560e-01 -5.90848267e-01 -7.00561628e-02 -8.65383625e-01 3.04424632e-02 -1.17696607e+00 1.93898529e-01 -1.31695294e+00 1.46398842e-01 -3.80542964e-01 -1.93834439e-01 5.94526708e-01 -8.89181271e-02 1.31295681e-01 2.55019397e-01 5.30286074e-01 -2.58464757e-02 4.71774697e-01 1.06497645e+00 -9.83215496e-02 -1.80481002e-01 -1.72914609e-01 -7.03359783e-01 5.24053812e-01 7.49716103e-01 -5.78514636e-01 1.50481001e-01 -6.70746863e-01 6.65089190e-01 -1.36295617e-01 -3.97381425e-01 -9.80004847e-01 5.67398891e-02 -2.07185745e-01 3.36360693e-01 -6.12241089e-01 1.26390522e-02 -7.24636793e-01 6.33328199e-01 7.54937828e-01 1.31628543e-01 7.56085292e-02 -1.31147474e-01 1.17364898e-01 -1.97624445e-01 3.53600979e-02 8.83066833e-01 -2.94862986e-01 -8.99033546e-01 -4.63595651e-02 -9.29937720e-01 1.90711357e-02 6.94378912e-01 -5.94004989e-03 -1.20105714e-01 9.67008621e-03 -6.69407964e-01 1.70844391e-01 4.46920484e-01 4.76550311e-01 2.89469361e-01 -1.32535481e+00 -1.19860268e+00 3.53435993e-01 8.11221451e-02 -4.79499102e-01 -5.94057888e-02 4.44674790e-01 -9.56185997e-01 1.36278525e-01 -3.15712363e-01 -2.28947505e-01 -7.67091572e-01 4.84441042e-01 -1.19503155e-01 1.76285226e-02 -9.19720411e-01 6.08810246e-01 -1.38201207e-01 -2.81707168e-01 -5.12603782e-02 -7.05826730e-02 -3.20297956e-01 2.25097194e-01 5.10926962e-01 5.18430114e-01 -4.24259394e-01 -9.20757651e-01 -6.23169601e-01 4.76670504e-01 4.91057569e-03 -5.87954998e-01 1.50906396e+00 -3.52263570e-01 -7.11036265e-01 9.32470500e-01 1.01141691e+00 5.26036263e-01 -5.36787271e-01 -4.72678207e-02 2.97610492e-01 -3.21589142e-01 -1.78546950e-01 -1.04066837e+00 -8.19157600e-01 5.87157667e-01 -1.35922700e-01 1.14176966e-01 9.09603834e-01 1.36085778e-01 1.09714890e+00 3.07299256e-01 6.83023572e-01 -1.12309372e+00 -5.51011384e-01 1.00944161e+00 5.07776678e-01 -6.05314493e-01 2.11126596e-01 2.99492359e-01 -8.87289762e-01 9.18007016e-01 -5.21828420e-02 2.67141443e-02 1.11342061e+00 2.10224181e-01 2.37649724e-01 -1.78098097e-01 -3.88884366e-01 -2.50719130e-01 6.52148873e-02 1.03171967e-01 8.25904250e-01 5.46639144e-01 -8.13145280e-01 7.86532581e-01 -8.50760877e-01 -2.61159867e-01 3.43931109e-01 9.58070278e-01 -8.51973474e-01 -1.25517356e+00 -1.21167541e-01 7.29076043e-02 -8.43243062e-01 -4.19995964e-01 -6.36982143e-01 1.03996813e+00 4.09956336e-01 8.12511802e-01 1.71393082e-01 -3.87128502e-01 1.77464873e-01 3.11261296e-01 3.72134954e-01 -4.89946336e-01 -6.46231592e-01 6.08615696e-01 3.08802933e-01 -4.09583300e-02 -5.19605100e-01 -6.87205791e-01 -1.49635744e+00 -7.20244288e-01 -2.05573797e-01 6.99030936e-01 5.39471447e-01 9.49688494e-01 -2.91572690e-01 3.50061119e-01 5.85489154e-01 -3.45724612e-01 -3.00951879e-02 -1.13489676e+00 -1.24251533e+00 -1.09368645e-01 -2.69240271e-02 -5.57033904e-02 -1.58355609e-01 1.99773386e-01]
[10.45710277557373, 10.114514350891113]
89765475-942c-4b66-a3e0-5eea461cfb79
a-hypergraph-based-machine-learning-ensemble
2211.03933
null
https://arxiv.org/abs/2211.03933v2
https://arxiv.org/pdf/2211.03933v2.pdf
A Hypergraph-Based Machine Learning Ensemble Network Intrusion Detection System
Network intrusion detection systems (NIDS) to detect malicious attacks continue to meet challenges. NIDS are often developed offline while they face auto-generated port scan infiltration attempts, resulting in a significant time lag from adversarial adaption to NIDS response. To address these challenges, we use hypergraphs focused on internet protocol addresses and destination ports to capture evolving patterns of port scan attacks. The derived set of hypergraph-based metrics are then used to train an ensemble machine learning (ML) based NIDS that allows for real-time adaption in monitoring and detecting port scanning activities, other types of attacks, and adversarial intrusions at high accuracy, precision and recall performances. This ML adapting NIDS was developed through the combination of (1) intrusion examples, (2) NIDS update rules, (3) attack threshold choices to trigger NIDS retraining requests, and (4) a production environment with no prior knowledge of the nature of network traffic. 40 scenarios were auto-generated to evaluate the ML ensemble NIDS comprising three tree-based models. The resulting ML Ensemble NIDS was extended and evaluated with the CIC-IDS2017 dataset. Results show that under the model settings of an Update-ALL-NIDS rule (specifically retrain and update all the three models upon the same NIDS retraining request) the proposed ML ensemble NIDS evolved intelligently and produced the best results with nearly 100% detection performance throughout the simulation.
['Nathaniel D. Bastian', 'Mark M. Bailey', 'Thomas D. Pike', 'Zong-Zhi Lin']
2022-11-08
null
null
null
null
['network-intrusion-detection']
['miscellaneous']
[ 2.78003335e-01 -2.73441195e-01 3.78424451e-02 -2.13077486e-01 4.44314368e-02 -7.90515125e-01 7.97841132e-01 -2.07601666e-01 -4.00238395e-01 5.38499713e-01 -6.28567517e-01 -8.51686478e-01 -3.96633774e-01 -1.09561098e+00 -1.15699582e-01 -3.43457639e-01 -4.54740644e-01 9.79192853e-01 8.40661585e-01 -1.29232377e-01 4.40815389e-01 1.09560215e+00 -1.03205645e+00 1.28732234e-01 4.31685656e-01 6.95956409e-01 -6.77027166e-01 1.07001305e+00 -3.46802354e-01 3.47260416e-01 -1.29977560e+00 -5.10757983e-01 8.33132207e-01 -2.80059159e-01 -2.88858712e-01 -1.36316285e-01 -8.97676498e-02 -1.74505815e-01 -5.51382422e-01 8.29667926e-01 5.02177715e-01 1.22167572e-01 3.59944552e-01 -1.70095813e+00 -4.25566472e-02 5.08264601e-01 -4.44614261e-01 8.40146959e-01 1.86106429e-01 9.92790937e-01 3.79105091e-01 -5.14998995e-02 6.37601972e-01 1.24295378e+00 4.68940169e-01 7.88496673e-01 -1.44718337e+00 -9.09732759e-01 1.88637048e-01 1.74592942e-01 -1.04152143e+00 -1.47743776e-01 6.56174898e-01 -3.04823611e-02 1.11927247e+00 3.18318814e-01 3.06580096e-01 1.81037772e+00 2.31425583e-01 -8.05349927e-03 1.06054091e+00 -2.03212678e-01 4.87283677e-01 5.53340316e-01 1.05655089e-01 4.20439422e-01 5.16195536e-01 6.28180623e-01 -1.67598762e-02 -6.48277342e-01 5.30828595e-01 1.44692674e-01 2.69505501e-01 2.96691835e-01 -6.16786718e-01 8.47242057e-01 2.29811758e-01 3.47489804e-01 -7.70722687e-01 -5.18370569e-01 4.72001731e-01 6.44319355e-01 1.88693833e-02 7.25126505e-01 -8.56075466e-01 -1.55514553e-01 -3.27682376e-01 -2.10258752e-01 1.53102970e+00 4.62368757e-01 5.83330214e-01 7.63275862e-01 -4.13502231e-02 5.78600824e-01 8.09461251e-02 6.87812150e-01 3.92578274e-01 -2.74151891e-01 2.41217494e-01 9.43473816e-01 -3.90297145e-01 -8.44358742e-01 -4.64684635e-01 -6.24713182e-01 -5.21120787e-01 4.02134478e-01 2.13175714e-01 -5.81305385e-01 -1.29098976e+00 1.63046563e+00 4.74719316e-01 7.37415552e-01 4.44960371e-02 8.68540332e-02 -3.77132259e-02 5.66427648e-01 4.82741356e-01 -1.88117608e-01 1.04109514e+00 -3.04513186e-01 -2.31531590e-01 -3.70743088e-02 1.78513885e-01 -3.37526917e-01 6.33521914e-01 1.35090739e-01 -4.67681825e-01 -3.11409444e-01 -1.04464853e+00 1.56520331e+00 -8.88206780e-01 -9.91935074e-01 4.77169812e-01 1.30229354e+00 -8.42869818e-01 4.93719339e-01 -7.05622435e-01 -6.68884814e-01 3.25972021e-01 6.77518368e-01 6.36301860e-02 2.11925641e-01 -1.31564665e+00 7.68073916e-01 4.23640847e-01 -4.65200543e-01 -9.74349976e-01 -8.78087282e-01 -2.38347933e-01 -1.00832134e-01 4.69585747e-01 -5.04303217e-01 8.64635527e-01 -7.03406215e-01 -1.54053986e+00 2.25838467e-01 3.65276277e-01 -6.25965774e-01 5.41438401e-01 5.27602375e-01 -1.33725643e+00 1.04719745e-02 -2.53663719e-01 -7.16670305e-02 9.40138638e-01 -1.24987650e+00 -6.64877772e-01 -4.33686078e-01 -5.56036532e-02 -3.37218374e-01 -3.36700082e-01 1.50372237e-01 -3.75901908e-02 -3.61408502e-01 -2.38150343e-01 -8.01683426e-01 -4.01716948e-01 -8.02828252e-01 -6.94955409e-01 1.21879421e-01 1.43266892e+00 -1.28176183e-01 1.29274368e+00 -1.62285066e+00 -4.06000257e-01 1.03577733e+00 6.94676265e-02 9.21680272e-01 -3.70229483e-01 5.94359756e-01 -1.73352361e-01 6.05772436e-01 -7.72821978e-02 2.21527413e-01 -1.25217602e-01 3.86619002e-01 -2.91879773e-01 -1.09211892e-01 3.68459761e-01 4.30661619e-01 -6.09377503e-01 -1.26381710e-01 4.40969229e-01 3.00183326e-01 -5.57128251e-01 5.83639562e-01 -3.00075859e-01 7.06067026e-01 -6.42069757e-01 9.66770530e-01 3.85829657e-01 -7.08710253e-02 2.70135701e-01 1.39442578e-01 1.82467759e-01 -4.87222560e-02 -1.33251560e+00 3.62243831e-01 -3.80562305e-01 4.60376330e-02 -1.70311004e-01 -5.80045640e-01 1.25485790e+00 2.50168949e-01 6.79745078e-01 -7.82630861e-01 2.24379763e-01 5.98091222e-02 6.23510480e-01 -3.24696898e-01 -4.07087356e-01 2.62365431e-01 1.31594986e-01 9.61232245e-01 1.37717336e-01 3.85493070e-01 3.60913455e-01 1.37879938e-01 2.03638124e+00 -7.99358189e-01 2.44346753e-01 3.20466280e-01 9.06036079e-01 5.28325848e-02 5.86758852e-01 1.08702302e+00 -5.53094745e-01 -3.50280285e-01 5.76810241e-01 -8.58027279e-01 -9.28550959e-01 -1.47042799e+00 8.45131427e-02 8.90726268e-01 -1.65048629e-01 -4.56060357e-02 -5.38457870e-01 -1.35674906e+00 4.58926111e-02 1.10676897e+00 -5.19111931e-01 -4.80803818e-01 -8.72699797e-01 -1.08474016e+00 9.70037818e-01 -3.43193226e-02 6.80093348e-01 -1.43942785e+00 -2.92837650e-01 4.96288985e-01 5.46562076e-01 -1.15101290e+00 -1.22296371e-01 3.66997838e-01 -5.71624517e-01 -1.75852191e+00 2.69457191e-01 -1.29459321e-01 6.25504076e-01 -3.66809189e-01 8.47752392e-01 3.05103242e-01 -6.13036215e-01 6.79037213e-01 -3.01388681e-01 -5.84165156e-01 -1.09847665e+00 1.12960033e-01 3.54590923e-01 3.80437106e-01 7.65754580e-01 -1.12689888e+00 -3.44217896e-01 6.09994471e-01 -1.06012833e+00 -8.73026848e-01 8.52786362e-01 5.50822079e-01 6.57029822e-02 3.95510525e-01 1.01364172e+00 -1.43668568e+00 7.88048565e-01 -9.21886265e-01 -8.38462591e-01 2.70870149e-01 -1.12428927e+00 -3.99673218e-03 1.04094517e+00 -1.19682527e+00 -9.30662990e-01 -3.99179012e-01 -1.50950372e-01 -4.91662681e-01 -5.91793656e-01 -4.14446592e-02 -2.35143170e-01 -3.36281240e-01 9.91517365e-01 2.55906552e-01 -3.69940288e-02 -2.20701292e-01 -2.26985693e-01 5.70673645e-01 6.39607236e-02 -6.08237267e-01 1.42626679e+00 1.17248692e-01 1.71767667e-01 -7.27781892e-01 -9.94001999e-02 -1.86366931e-01 -5.10138810e-01 -4.43535894e-01 2.42735609e-01 1.65249497e-01 -8.25963557e-01 4.22006339e-01 -8.35612357e-01 -1.86983377e-01 -2.08201095e-01 1.72692135e-01 5.64456061e-02 1.05550392e-02 -5.32854378e-01 -8.53326380e-01 -4.23627287e-01 -1.06644738e+00 5.92670031e-02 5.98476589e-01 -6.02367669e-02 -1.21394706e+00 4.69682187e-01 -4.49465327e-02 1.03309095e+00 6.25496209e-01 1.25519979e+00 -1.96758008e+00 -4.24023002e-01 -7.23569155e-01 -1.69052780e-02 2.14257151e-01 2.35832155e-01 3.42472851e-01 -6.44400716e-01 -2.58219302e-01 -1.13327250e-01 3.03886265e-01 2.88072415e-02 -2.14187130e-01 7.46854305e-01 -6.18745148e-01 -3.52408469e-01 7.52383709e-01 1.48218584e+00 1.17128241e+00 4.78710115e-01 6.89010918e-01 3.05989742e-01 3.29559326e-01 -1.26921877e-01 6.16516769e-01 -2.59638131e-01 5.03879189e-01 7.44976819e-01 7.84693584e-02 1.24025106e-01 -3.14953238e-01 2.96007037e-01 1.66534260e-01 1.31030798e-01 -4.74153638e-01 -9.68682885e-01 -9.35092345e-02 -1.13392508e+00 -1.12546480e+00 1.19596049e-01 2.25619984e+00 2.98939705e-01 6.76029682e-01 4.84847605e-01 8.83006677e-02 9.51427519e-01 4.53525446e-02 -1.22779500e+00 -8.35153401e-01 6.17177114e-02 7.36587822e-01 7.29083538e-01 2.67832369e-01 -9.26747799e-01 9.44262683e-01 5.89464235e+00 2.64924735e-01 -1.08071041e+00 -6.99815378e-02 3.78635943e-01 2.14662865e-01 1.55840024e-01 4.13861461e-02 -7.88108528e-01 5.49879909e-01 1.64078999e+00 -3.10769856e-01 8.20188940e-01 6.60794079e-01 2.06454806e-02 5.40208220e-01 -6.86853886e-01 3.15317482e-01 -2.07729608e-01 -1.06460452e+00 4.56316918e-01 3.23258430e-01 5.48210025e-01 8.97492543e-02 4.52073589e-02 7.77218044e-01 9.16131616e-01 -7.22534239e-01 -3.11230779e-01 2.92518854e-01 4.97936606e-01 -7.44462609e-01 7.69936979e-01 3.51297349e-01 -9.54714954e-01 -5.29786825e-01 1.30292967e-01 5.31605363e-01 1.04877390e-01 7.98105448e-02 -1.43714571e+00 4.68068838e-01 2.77874082e-01 3.10665481e-02 -7.82685995e-01 8.43672216e-01 4.36313301e-02 1.08000016e+00 -6.19296908e-01 -1.03448838e-01 3.05994600e-01 1.10265762e-01 1.23615086e+00 1.13998580e+00 -9.99719650e-02 5.58637455e-02 3.10438961e-01 7.62463093e-01 1.60990328e-01 -1.77668899e-01 -6.19390905e-01 -1.89042151e-01 8.81418943e-01 1.22600484e+00 -7.40715325e-01 -5.60135394e-02 -1.73885241e-01 5.28194547e-01 -6.13235570e-02 7.05688894e-01 -8.85231435e-01 -4.74707931e-01 9.93812859e-01 3.89490932e-01 -1.13467664e-01 9.92233083e-02 -1.14055894e-01 -6.76465929e-01 -6.07025385e-01 -1.04156625e+00 9.46958959e-01 7.28406757e-02 -1.63456178e+00 1.19699037e+00 2.20462270e-02 -9.13832903e-01 -5.23498297e-01 -6.73971713e-01 -1.28413606e+00 8.15975130e-01 -1.15826166e+00 -7.89300978e-01 -1.31272152e-01 8.76503289e-01 4.81346339e-01 -9.15196121e-01 9.31746185e-01 2.85242379e-01 -1.19244695e+00 9.27467585e-01 -4.93950516e-01 2.93102741e-01 3.66456360e-01 -1.03796363e+00 5.38656354e-01 1.01842880e+00 -4.92840670e-02 5.58114648e-01 6.72878981e-01 -1.04568481e+00 -1.20519602e+00 -1.37325406e+00 6.91485703e-02 -4.32790488e-01 8.88260841e-01 -8.11493248e-02 -9.79008913e-01 7.31621087e-01 -1.55065805e-01 -3.82534489e-02 6.65690958e-01 -2.20078245e-01 -5.19467056e-01 -1.99976772e-01 -1.85814679e+00 6.72075272e-01 8.07763040e-01 -1.82759762e-02 -1.51099890e-01 2.78293043e-01 5.56534052e-01 8.98539424e-02 -7.21160293e-01 5.28243244e-01 2.07945764e-01 -8.93381596e-01 9.50037301e-01 -1.25507474e+00 -5.89799702e-01 -1.06482752e-01 -1.42192766e-01 -9.47666943e-01 -2.25163713e-01 -8.46121907e-01 -3.14458758e-01 1.22289622e+00 6.87398016e-01 -1.39486372e+00 6.82712078e-01 6.38384044e-01 2.96257704e-01 -7.21329391e-01 -8.99037838e-01 -8.28595817e-01 -5.27259052e-01 -3.45788509e-01 8.61968756e-01 7.72226810e-01 -6.03128076e-01 2.41682604e-01 1.20633831e-02 6.03678346e-01 8.75695527e-01 -6.67414308e-01 9.46570277e-01 -1.49935591e+00 -3.10370773e-01 -6.44740164e-01 -6.54813051e-01 1.35803476e-01 1.70460090e-01 -6.83317304e-01 -7.46523321e-01 -7.68308103e-01 -3.06027293e-01 -6.73426449e-01 -6.27744496e-01 5.07926106e-01 1.24875888e-01 -1.29022419e-01 9.82902348e-02 2.56572604e-01 -1.74712226e-01 -8.01930726e-02 5.67556739e-01 -6.71665370e-03 -7.38720953e-01 6.41419888e-01 -4.21557963e-01 5.14077902e-01 1.13639355e+00 -6.56739354e-01 -4.46455657e-01 3.50097448e-01 -4.67306048e-01 6.96871281e-02 1.43704608e-01 -1.06936753e+00 3.42793971e-01 -3.75502646e-01 5.17604589e-01 -1.99889123e-01 4.06277664e-02 -1.00813997e+00 4.16216224e-01 8.44132423e-01 -1.00487418e-01 5.48968196e-01 4.62078929e-01 8.31393600e-01 3.46336752e-01 1.37114272e-01 1.11454928e+00 -6.96322396e-02 -6.44622087e-01 8.23104143e-01 -5.44762373e-01 9.83053818e-02 1.46970069e+00 -3.97245049e-01 -4.87187952e-01 -1.92941085e-01 -6.95803463e-01 3.44007052e-02 2.07779348e-01 6.00627184e-01 5.58966994e-01 -6.52858019e-01 -6.72140241e-01 8.44677389e-01 -1.62884235e-01 -8.23135972e-01 8.41836557e-02 2.60293216e-01 -3.07532519e-01 2.35416248e-01 -5.20069599e-01 -4.76180613e-01 -9.41813827e-01 5.48049688e-01 7.20927179e-01 -8.17749023e-01 -5.26356816e-01 3.60317290e-01 -5.84616125e-01 -7.00174868e-01 2.42412537e-01 8.84899378e-01 -2.65513778e-01 -4.79220361e-01 6.11935735e-01 5.45108974e-01 3.90505642e-02 -2.72556990e-01 -4.76770699e-01 -7.56171793e-02 -3.12362611e-01 -1.71527311e-01 1.12437630e+00 2.96519160e-01 2.07119569e-01 -3.83896218e-03 7.54849136e-01 -1.30870655e-01 -8.20689857e-01 -2.96957672e-01 4.91354525e-01 -5.36520779e-01 -3.85755181e-01 -1.48756409e+00 -1.35251307e+00 2.21755192e-01 6.97402239e-01 5.51107705e-01 1.12833631e+00 -5.42940557e-01 7.06718564e-01 3.23520184e-01 3.94846320e-01 -4.62821633e-01 1.64792076e-01 4.35055882e-01 1.58179909e-01 -9.62105095e-01 -3.80472481e-01 1.07890874e-01 -5.15397191e-01 1.29133856e+00 1.21735799e+00 -1.47484928e-01 8.44865203e-01 4.65838194e-01 7.30912164e-02 -2.75103986e-01 -1.15780985e+00 1.58547297e-01 -1.45346358e-01 1.17612255e+00 -6.34237289e-01 -2.71911249e-02 1.82912320e-01 2.98154116e-01 1.74383163e-01 -6.85801327e-01 4.53096002e-01 7.55968094e-01 -2.23747015e-01 -1.27731156e+00 -2.79415846e-01 8.44196379e-01 -1.98953941e-01 2.47290626e-01 -7.53747284e-01 1.19475007e+00 -9.05711651e-02 1.08490622e+00 -1.65903363e-02 -9.99114573e-01 7.06425488e-01 4.29106265e-01 -1.41782925e-01 -4.38575119e-01 -1.15464997e+00 -6.36503041e-01 -1.21441253e-01 -5.20919561e-01 3.49376112e-01 -4.43973005e-01 -8.99750590e-01 -7.88856268e-01 -9.71703231e-02 2.34413847e-01 8.58248591e-01 7.61649668e-01 4.71820474e-01 7.94795513e-01 1.30757809e+00 -2.81711459e-01 -9.78197515e-01 -1.05924082e+00 -3.32647830e-01 2.94847727e-01 -2.08580926e-01 -6.87577188e-01 -9.84950662e-01 -6.52437687e-01]
[5.353024482727051, 7.335224151611328]
da45c2ca-8078-416b-9f45-8104d0323c1d
werewolf-among-us-a-multimodal-dataset-for
2212.08279
null
https://arxiv.org/abs/2212.08279v1
https://arxiv.org/pdf/2212.08279v1.pdf
Werewolf Among Us: A Multimodal Dataset for Modeling Persuasion Behaviors in Social Deduction Games
Persuasion modeling is a key building block for conversational agents. Existing works in this direction are limited to analyzing textual dialogue corpus. We argue that visual signals also play an important role in understanding human persuasive behaviors. In this paper, we introduce the first multimodal dataset for modeling persuasion behaviors. Our dataset includes 199 dialogue transcriptions and videos captured in a multi-player social deduction game setting, 26,647 utterance level annotations of persuasion strategy, and game level annotations of deduction game outcomes. We provide extensive experiments to show how dialogue context and visual signals benefit persuasion strategy prediction. We also explore the generalization ability of language models for persuasion modeling and the role of persuasion strategies in predicting social deduction game outcomes. Our dataset, code, and models can be found at https://persuasion-deductiongame.socialai-data.org.
['Diyi Yang', 'James M. Rehg', 'Shirley Anugrah Hayati', 'Wenqi Jia', 'Fiona Ryan', 'Aryan Pariani', 'Miao Liu', 'Hongxin Zhang', 'Bolin Lai']
2022-12-16
null
null
null
null
['persuasion-strategies']
['computer-vision']
[ 3.57833683e-01 4.87257749e-01 -1.83172315e-01 -3.66491318e-01 -5.26327610e-01 -6.65945828e-01 1.17908454e+00 2.44462751e-02 -4.13199872e-01 7.89858937e-01 8.78056347e-01 -7.40774632e-01 7.35727474e-02 -6.41942859e-01 -2.30870172e-01 -3.31426144e-01 1.87664315e-01 3.07267487e-01 5.88871771e-03 -7.00750709e-01 3.25940818e-01 -2.37902597e-01 -1.25549483e+00 7.18861163e-01 7.33720481e-01 5.34604788e-01 1.32463142e-01 1.45356965e+00 -5.49765415e-02 1.82305110e+00 -9.60254490e-01 -6.19550228e-01 -1.66408971e-01 -9.75595415e-01 -1.09374893e+00 1.72665581e-01 1.34917676e-01 -5.52390397e-01 -3.93486887e-01 6.16031170e-01 5.36909282e-01 3.98550868e-01 9.09991384e-01 -1.50631666e+00 -7.93383718e-01 1.06621683e+00 -4.03970808e-01 3.22035372e-01 9.22956526e-01 4.51980770e-01 1.06980908e+00 -3.56319755e-01 7.23443747e-01 1.73574436e+00 3.86844665e-01 1.08513343e+00 -1.02654052e+00 -3.44385803e-01 1.73376560e-01 6.20589972e-01 -5.31776965e-01 -4.34932142e-01 9.85657752e-01 -4.44865555e-01 8.17187607e-01 5.07065237e-01 8.90228689e-01 1.82003379e+00 -7.85737783e-02 1.44786918e+00 1.23646474e+00 -3.84347916e-01 5.20902462e-02 2.73891054e-02 5.09839356e-01 1.01172519e+00 -5.23967445e-01 -5.96620478e-02 -7.99402237e-01 -2.89010316e-01 4.34061408e-01 -3.46231669e-01 -3.08643073e-01 7.14399805e-03 -1.08917856e+00 1.35914981e+00 1.94239244e-01 -1.98042333e-01 -3.59440684e-01 1.29037514e-01 5.95837891e-01 4.45537329e-01 4.06981140e-01 3.56604397e-01 -9.73688066e-02 -1.02187049e+00 1.60439029e-01 2.39176184e-01 9.74551082e-01 5.09544075e-01 4.98053193e-01 1.15307622e-01 -4.24363881e-01 1.19153655e+00 3.86518121e-01 5.83533406e-01 3.22938919e-01 -1.21361828e+00 3.33044589e-01 6.27228439e-01 1.15882322e-01 -1.29357779e+00 -6.94062114e-01 4.83961225e-01 -2.42276415e-01 7.65998438e-02 6.48912370e-01 -6.75004840e-01 -2.09912345e-01 1.63739026e+00 2.69891620e-01 -2.05213189e-01 3.56869519e-01 9.54235315e-01 1.63605416e+00 6.68070436e-01 3.58938843e-01 -1.66288808e-01 1.78162611e+00 -1.10656250e+00 -9.24435437e-01 -4.04855490e-01 9.04030204e-01 -4.51494604e-01 1.65552545e+00 5.51678538e-02 -9.58958447e-01 -2.66741484e-01 -7.37789392e-01 -2.40382791e-01 -8.18741024e-02 -5.82041889e-02 8.67266357e-01 6.16840184e-01 -6.88117325e-01 -2.71258146e-01 -3.84185374e-01 -4.67672735e-01 5.42228460e-01 -2.78503776e-01 6.52709752e-02 4.19702232e-01 -1.45179939e+00 9.23337996e-01 -9.53980163e-02 -2.91564167e-01 -9.41902995e-01 -2.52418935e-01 -1.06329894e+00 -1.48177817e-01 9.30254638e-01 -5.45809090e-01 1.74768233e+00 -1.15563440e+00 -1.75976217e+00 8.72660041e-01 6.21077903e-02 -6.17164075e-01 6.91178799e-01 -1.12042855e-02 -1.86884314e-01 3.59312773e-01 4.71112765e-02 8.28524888e-01 6.94089651e-01 -1.14076638e+00 -7.83403456e-01 -2.59174317e-01 7.07485378e-01 8.17597449e-01 -2.08311006e-01 3.81186679e-02 -1.20443724e-01 -2.22517371e-01 -9.27816808e-01 -7.94059634e-01 1.92806609e-02 -4.37539965e-01 -4.80472088e-01 -7.14172006e-01 7.21167862e-01 -6.71218872e-01 8.99327636e-01 -2.14012718e+00 1.82713598e-01 -3.65392327e-01 7.00224280e-01 -4.57362160e-02 -1.87628403e-01 3.17442179e-01 4.84337181e-01 5.98381422e-02 4.52922851e-01 -2.83040404e-01 3.93461525e-01 -7.06215622e-03 -1.22337297e-01 2.34141648e-01 -3.70284855e-01 1.35068965e+00 -9.79254007e-01 -5.69884419e-01 4.40710902e-01 1.91482648e-01 -3.54345620e-01 2.18592495e-01 -4.52317208e-01 2.84232110e-01 -6.56684399e-01 2.02380002e-01 7.52323642e-02 -3.77461612e-01 4.20852035e-01 1.35296896e-01 1.17267862e-01 7.38218606e-01 -2.01409414e-01 1.29608977e+00 -2.48710975e-01 1.24743187e+00 1.37949288e-01 -6.16960645e-01 6.43079817e-01 1.57857463e-01 2.56978601e-01 -8.19740176e-01 6.78676009e-01 -5.78324616e-01 3.95918936e-01 -4.29142535e-01 7.92662740e-01 6.71177581e-02 -4.95182991e-01 9.50262427e-01 -3.90812099e-01 -1.98066846e-01 2.76026428e-01 7.14859784e-01 1.02647889e+00 -1.91523805e-01 5.55526078e-01 -4.09482121e-02 4.26015198e-01 5.14040709e-01 3.82022597e-02 1.09842682e+00 -6.04504883e-01 -3.80316749e-02 9.73932207e-01 -3.73664260e-01 -5.49011588e-01 -8.17944407e-01 5.67388415e-01 1.57201672e+00 2.54892349e-01 -6.76694930e-01 -9.73220468e-01 -8.14919770e-01 -3.45311105e-01 9.56572711e-01 -7.89940596e-01 -1.86087415e-01 -2.94051528e-01 -4.37788308e-01 7.45616436e-01 3.56590629e-01 7.63899803e-01 -1.31667256e+00 -8.45390618e-01 -1.61422074e-01 -6.97066128e-01 -1.00886786e+00 -3.82415503e-01 -2.97870308e-01 -1.65854305e-01 -1.59177852e+00 8.74117762e-02 -4.65118498e-01 1.73574872e-02 6.06460154e-01 1.20066321e+00 3.71288717e-01 -5.05557097e-02 1.16198921e+00 -5.18811941e-01 -7.32470095e-01 -9.91653025e-01 -6.24530241e-02 -1.07970506e-01 -3.41801107e-01 7.71283984e-01 -6.00111205e-03 -4.34216112e-01 4.63988990e-01 -2.30020300e-01 8.50846112e-01 -1.21510744e-01 7.61777818e-01 -2.25499094e-01 -4.57107544e-01 5.81513584e-01 -8.86581957e-01 1.70058692e+00 -3.79271001e-01 -1.53475627e-01 -3.39729711e-02 -2.74103820e-01 -5.49155712e-01 4.84894991e-01 -5.74011981e-01 -1.24643862e+00 -4.64379847e-01 -8.07409957e-02 1.64403066e-01 -4.53724384e-01 4.49249893e-01 8.09369981e-02 5.01852930e-01 8.60670030e-01 8.52710754e-02 5.35599947e-01 -6.67886436e-02 6.62640631e-01 7.87953794e-01 2.73394734e-01 -5.68000495e-01 1.84694603e-01 3.64091426e-01 -5.18550515e-01 -1.27042377e+00 -5.92899263e-01 -1.12286381e-01 -6.14266321e-02 -9.68465805e-01 1.11791837e+00 -7.12487698e-01 -1.70614851e+00 4.83860791e-01 -8.93098056e-01 -8.93517494e-01 1.57186836e-02 1.91141263e-01 -8.28128219e-01 6.37467504e-01 -8.06872487e-01 -1.10400224e+00 -2.85118818e-01 -9.83473718e-01 6.87934279e-01 4.10218209e-01 -7.39966869e-01 -1.30202568e+00 -1.04487613e-01 1.28111219e+00 9.61743817e-02 -1.17848501e-01 7.89963722e-01 -8.39620590e-01 1.81546826e-02 2.97582984e-01 -1.15608387e-01 -1.90555274e-01 2.31813014e-01 -2.47368306e-01 -6.75238073e-01 2.43882746e-01 -2.49943808e-02 -9.56221879e-01 5.24407387e-01 6.28234446e-01 6.84458554e-01 -5.12685537e-01 -1.27821311e-01 -2.80036658e-01 2.95142144e-01 4.85137761e-01 6.52941763e-01 3.31398547e-01 7.73349464e-01 8.26619565e-01 1.01058888e+00 4.53923523e-01 1.00244367e+00 6.60609901e-01 2.80104190e-01 -2.06098855e-02 3.12147886e-02 -5.60613573e-01 6.17097437e-01 3.17387819e-01 -2.78101832e-01 -7.68351376e-01 -9.80722427e-01 2.68662989e-01 -2.11009097e+00 -1.51893759e+00 -5.29873192e-01 1.24056566e+00 8.65580797e-01 -5.70482127e-02 7.11340189e-01 -1.76530983e-02 5.14384210e-01 5.81118405e-01 -2.53612846e-01 -7.36851454e-01 -1.30305260e-01 -6.39525652e-01 -4.28902730e-02 1.22646666e+00 -1.01212263e+00 1.28909910e+00 5.86580992e+00 8.74155819e-01 -7.24911511e-01 9.50794742e-02 7.89798677e-01 -1.68028712e-01 -3.76372010e-01 -4.44795698e-01 -4.76925850e-01 2.97661304e-01 7.35970438e-01 -1.36816099e-01 5.74857175e-01 7.39296436e-01 7.21285164e-01 -4.13320184e-01 -9.48201776e-01 9.22187090e-01 3.84297639e-01 -1.46170235e+00 -1.89252242e-01 8.99191797e-02 1.71812162e-01 -3.67496163e-01 -4.40601371e-02 5.70500135e-01 7.96246469e-01 -9.74916279e-01 5.12207508e-01 3.25848818e-01 4.04499918e-01 -4.96258438e-01 4.46843833e-01 4.34491932e-01 -6.24778628e-01 5.15526496e-02 -2.29293518e-02 -7.04742789e-01 4.01383340e-01 -5.40445805e-01 -1.50130630e+00 -6.10199980e-02 5.77351928e-01 9.07793522e-01 -5.30917287e-01 -2.04198673e-01 -5.26432753e-01 8.13257635e-01 7.84527361e-02 -1.00753117e+00 2.61053443e-01 -3.72652650e-01 6.99375153e-01 1.10868335e+00 -4.78219569e-01 4.66032743e-01 2.04131141e-01 5.39634764e-01 1.21090030e-02 1.18127696e-01 -7.84477890e-01 -5.05878747e-01 2.88056403e-01 1.15280330e+00 -6.49898946e-01 -3.42356384e-01 -5.10272920e-01 8.88821542e-01 3.23028594e-01 3.14940631e-01 -1.07691526e+00 2.21769750e-01 9.44366992e-01 -2.25829080e-01 -1.23684928e-01 -1.15158446e-01 -4.06980753e-01 -9.96290982e-01 -6.08997166e-01 -1.41361737e+00 5.12396991e-01 -1.04003537e+00 -1.09438539e+00 2.77313232e-01 -5.71311079e-02 -6.90368891e-01 -7.66125977e-01 -7.71369457e-01 -5.44855952e-01 3.99276227e-01 -8.19761932e-01 -1.05999279e+00 -6.37928426e-01 6.67792380e-01 1.12655175e+00 -4.38024521e-01 7.97103822e-01 -5.18661141e-01 -2.89969027e-01 2.43191943e-01 -5.41093647e-01 3.27407807e-01 5.83613992e-01 -1.23695922e+00 5.26963733e-02 3.74017388e-01 -5.43132704e-03 3.20988387e-01 1.01406395e+00 -5.84514141e-01 -1.44193816e+00 -3.05951029e-01 2.17747986e-01 -7.94316530e-01 9.27427649e-01 -5.25554538e-01 -5.17850459e-01 6.56582296e-01 8.80890429e-01 -1.23505974e+00 1.24673569e+00 3.81324649e-01 -2.47918051e-02 6.75602555e-01 -8.44792008e-01 1.43995678e+00 1.25227618e+00 -5.84623337e-01 -7.90081501e-01 2.66953021e-01 4.02734011e-01 -4.77615267e-01 -4.66168582e-01 -1.99440435e-01 5.53828418e-01 -1.04495072e+00 1.04014051e+00 -9.38311517e-01 1.04815233e+00 9.88428444e-02 -3.22397947e-02 -1.43165255e+00 -6.72892109e-03 -6.63116753e-01 -2.39677429e-01 9.22840416e-01 4.47095126e-01 -3.16995919e-01 8.47971499e-01 6.94921911e-01 1.02314532e-01 -4.13622737e-01 -5.17397463e-01 5.34485392e-02 -1.17467590e-01 -5.62150478e-01 2.92646050e-01 1.00903189e+00 6.28556192e-01 1.49471605e+00 -7.95249403e-01 -4.09344673e-01 2.85349965e-01 6.82136565e-02 1.43704748e+00 -9.25975084e-01 -3.31599921e-01 -6.25662923e-01 -5.72002977e-02 -1.54534316e+00 5.92098892e-01 -3.98149282e-01 3.17852572e-02 -1.90319312e+00 2.91338861e-01 1.34765401e-01 6.22650325e-01 6.71176612e-01 -3.52827698e-01 1.82607830e-01 2.68516719e-01 -1.34861797e-01 -1.01624763e+00 7.61050582e-01 1.49002540e+00 -4.44249630e-01 -2.95378476e-01 -1.13529906e-01 -1.05827546e+00 7.79747844e-01 1.07522869e+00 1.96720406e-01 -9.11513329e-01 -2.09682614e-01 3.43466774e-02 3.68447423e-01 4.31556076e-01 -1.81741547e-02 8.23423937e-02 -5.59277952e-01 -1.56569928e-02 -2.72919804e-01 8.30874145e-01 -4.26696569e-01 -4.61882561e-01 3.42089981e-01 -6.90157831e-01 -1.33385301e-01 1.73773527e-01 7.04159021e-01 1.79974556e-01 -7.19702020e-02 3.84214908e-01 4.07922789e-02 -8.99892092e-01 -3.35873961e-01 -1.50788903e+00 1.48513839e-01 9.41204906e-01 -4.11835343e-01 -9.69306290e-01 -1.59614861e+00 -6.76906765e-01 6.85377777e-01 4.49708700e-01 8.69912267e-01 6.30579352e-01 -1.02501905e+00 -8.38106215e-01 -4.03183788e-01 4.34027649e-02 -6.47591174e-01 4.39058095e-01 8.28227639e-01 -3.32305044e-01 1.70976490e-01 -2.89741904e-01 -5.43199420e-01 -2.03869963e+00 2.11573184e-01 3.78787667e-01 2.26556540e-01 -5.48759222e-01 7.18548238e-01 5.12141287e-01 -4.12324607e-01 9.40506682e-02 1.78086400e-01 -7.73615062e-01 -5.03811287e-03 8.02210271e-01 5.57881773e-01 -6.35003686e-01 -7.34430730e-01 -2.53569901e-01 -2.26837203e-01 -8.54740217e-02 -4.34814960e-01 9.63041246e-01 -5.07513583e-01 1.45974115e-01 6.11339092e-01 6.51625335e-01 3.12050402e-01 -1.22236800e+00 -1.47815812e-02 -3.90389055e-01 -3.96774262e-01 -2.25531653e-01 -1.14677715e+00 -3.98794979e-01 6.55921280e-01 1.65653899e-01 8.87615681e-01 6.79811299e-01 3.15528661e-01 1.78216219e-01 4.42546666e-01 1.98982537e-01 -1.21515989e+00 4.19724792e-01 7.79213905e-01 9.81768489e-01 -1.61633456e+00 -2.35101029e-01 -6.48044407e-01 -1.44339073e+00 7.32160866e-01 9.65843678e-01 3.28004956e-01 1.80487558e-01 1.54879481e-01 5.71871340e-01 -6.02000535e-01 -1.04433906e+00 2.50662910e-03 7.84837175e-03 8.65854859e-01 6.24201953e-01 4.62393552e-01 -5.99329710e-01 8.65560591e-01 -6.91117764e-01 -4.06022817e-01 8.34505916e-01 6.78343952e-01 -2.96201795e-01 -6.00609362e-01 -2.30267376e-01 3.98175448e-01 -1.08254276e-01 -4.98148315e-02 -1.29396582e+00 7.95151889e-01 -6.65996253e-01 1.70364022e+00 1.42741635e-01 -6.71356022e-01 1.15688652e-01 -4.45761159e-03 4.22225028e-01 -4.22043443e-01 -6.49029255e-01 -1.17686510e-01 1.09157300e+00 -3.63746166e-01 -5.82253933e-01 -8.00028861e-01 -1.40846765e+00 -1.04038429e+00 5.77548034e-02 3.81699800e-01 1.09853998e-01 1.03465605e+00 2.44822845e-01 5.26361287e-01 2.51170427e-01 -4.84042704e-01 -1.58931479e-01 -1.36926031e+00 -1.55601397e-01 5.36924601e-01 1.24094775e-02 -6.67522967e-01 -3.34550917e-01 -1.89426709e-02]
[12.841485023498535, 7.906250476837158]
fb7775a1-5ffc-48ca-88e5-6c8881a8f5eb
enhancing-mapless-trajectory-prediction
2306.14177
null
https://arxiv.org/abs/2306.14177v1
https://arxiv.org/pdf/2306.14177v1.pdf
Enhancing Mapless Trajectory Prediction through Knowledge Distillation
Scene information plays a crucial role in trajectory forecasting systems for autonomous driving by providing semantic clues and constraints on potential future paths of traffic agents. Prevalent trajectory prediction techniques often take high-definition maps (HD maps) as part of the inputs to provide scene knowledge. Although HD maps offer accurate road information, they may suffer from the high cost of annotation or restrictions of law that limits their widespread use. Therefore, those methods are still expected to generate reliable prediction results in mapless scenarios. In this paper, we tackle the problem of improving the consistency of multi-modal prediction trajectories and the real road topology when map information is unavailable during the test phase. Specifically, we achieve this by training a map-based prediction teacher network on the annotated samples and transferring the knowledge to a student mapless prediction network using a two-fold knowledge distillation framework. Our solution is generalizable for common trajectory prediction networks and does not bring extra computation burden. Experimental results show that our method stably improves prediction performance in mapless mode on many widely used state-of-the-art trajectory prediction baselines, compensating for the gaps caused by the absence of HD maps. Qualitative visualization results demonstrate that our approach helps infer unseen map information.
['Jianru Xue', 'Lei Bai', 'Pu Zhang', 'Yuning Wang']
2023-06-25
null
null
null
null
['trajectory-prediction', 'trajectory-forecasting']
['computer-vision', 'computer-vision']
[-5.20227961e-02 2.36312836e-01 -5.01853168e-01 -6.37675107e-01 -5.60540736e-01 -4.64960277e-01 6.46293879e-01 9.06106904e-02 -2.36003056e-01 8.35640550e-01 5.38149998e-02 -7.26587296e-01 -2.69660503e-01 -1.19755185e+00 -9.05924499e-01 -4.69700724e-01 1.59068301e-01 7.41729200e-01 9.83470559e-01 -3.61577421e-01 1.86685875e-01 5.36543369e-01 -1.93098497e+00 2.42452756e-01 1.23962855e+00 8.71199608e-01 3.68760467e-01 3.60680163e-01 -2.46882364e-01 9.28561091e-01 -1.78098649e-01 -4.02936488e-01 2.24134699e-01 2.16065168e-01 -7.12022901e-01 -3.63006890e-01 4.00037855e-01 -5.06320953e-01 -6.87143803e-01 6.53406262e-01 1.00214794e-01 3.69148254e-01 4.51559573e-01 -1.71166170e+00 -7.56327435e-02 3.94825101e-01 -1.29070550e-01 2.21886024e-01 -2.29598349e-03 2.04074070e-01 5.41001499e-01 -6.76684558e-01 7.03295112e-01 7.99986780e-01 9.38956141e-01 2.48236462e-01 -8.55850041e-01 -6.84653997e-01 4.04569298e-01 9.29722309e-01 -1.52505815e+00 -4.16313916e-01 7.64871538e-01 -6.16004705e-01 7.87499964e-01 3.67062390e-01 6.03469610e-01 8.63627911e-01 3.03472206e-03 7.73906291e-01 6.96695030e-01 1.10479593e-01 1.21496566e-01 4.97962326e-01 6.51908144e-02 6.08329654e-01 -2.25807145e-01 1.91441402e-01 -5.12022436e-01 1.86604157e-01 3.52010995e-01 7.61523470e-02 -1.06046215e-01 -4.29112941e-01 -1.15134037e+00 5.48800349e-01 6.51053727e-01 -9.61625725e-02 -4.55532521e-01 -4.47213054e-02 2.25961506e-01 7.78744444e-02 7.01704919e-01 9.82544944e-02 -3.20080549e-01 -4.68841076e-01 -9.80380833e-01 5.45580029e-01 5.81795156e-01 1.22585177e+00 1.03904676e+00 -1.58541530e-01 -1.46969274e-01 5.82199931e-01 -1.21407613e-01 4.75878209e-01 6.39478043e-02 -9.42568064e-01 7.70754278e-01 7.41396308e-01 4.09451276e-01 -1.21277106e+00 -4.64076519e-01 -2.07895711e-01 -5.76597333e-01 1.14117853e-01 6.33056045e-01 5.21158287e-03 -7.93982804e-01 1.37509811e+00 5.47539294e-01 7.05848038e-01 -1.63934261e-01 9.43266690e-01 6.72451913e-01 8.13561618e-01 1.30613789e-01 2.21776322e-01 7.80873656e-01 -1.21238613e+00 -4.43173289e-01 -7.36168772e-02 8.70863497e-01 -3.78601998e-01 1.01281989e+00 1.57298341e-01 -6.44493997e-01 -6.71455979e-01 -8.01509798e-01 1.04739740e-02 -8.75807226e-01 1.03323877e-01 4.36829895e-01 5.10460973e-01 -1.06499422e+00 6.63543880e-01 -8.39291811e-01 -2.58974314e-01 5.81331611e-01 1.78134069e-01 -2.90574789e-01 -2.75233686e-01 -1.18983603e+00 1.24900305e+00 5.17393112e-01 1.87234014e-01 -5.64305007e-01 -1.23039377e+00 -6.24752998e-01 -1.17874816e-01 3.97473752e-01 -2.13878796e-01 1.17739117e+00 -4.29858595e-01 -1.35987496e+00 3.17161024e-01 -3.78497005e-01 -4.72835243e-01 9.44463134e-01 -1.64076865e-01 -6.46656394e-01 -5.58684841e-02 2.18363956e-01 8.16169441e-01 3.89689535e-01 -1.10222149e+00 -1.14921379e+00 -4.79249209e-02 9.12535042e-02 2.01166943e-01 -8.71492177e-02 -4.54163969e-01 -6.22308850e-01 -8.33492130e-02 1.44699857e-01 -9.32474554e-01 -3.99205863e-01 1.06510967e-01 -4.65421140e-01 -3.99778783e-01 1.24416029e+00 -7.61881948e-01 1.19463289e+00 -2.02140236e+00 -4.61686164e-01 4.47613150e-01 1.78704888e-01 3.06329340e-01 -2.43554786e-02 4.67135668e-01 2.79658347e-01 -1.81568280e-01 -6.44705072e-02 -6.76887631e-02 7.58132413e-02 3.33631873e-01 -7.57773042e-01 2.23639175e-01 3.02069895e-02 1.01004231e+00 -1.04676330e+00 -5.57456374e-01 6.06859624e-01 5.01512289e-01 -3.03192377e-01 1.84431314e-01 -3.91859770e-01 7.88131416e-01 -5.20191789e-01 2.86369801e-01 6.79751933e-01 -8.80196840e-02 -4.87268046e-02 -1.11971751e-01 -5.40314198e-01 6.34000659e-01 -1.01444244e+00 1.51435745e+00 -4.77179527e-01 9.61655557e-01 -5.15439510e-01 -9.95802760e-01 9.89391029e-01 8.97445381e-02 5.18057764e-01 -9.10857737e-01 -3.74444962e-01 1.00764029e-01 -2.38212928e-01 -5.92681944e-01 8.09273362e-01 2.80607939e-01 1.59119397e-01 2.78116792e-01 -4.54313546e-01 1.89099442e-02 1.20308749e-01 8.92218351e-02 9.43112969e-01 4.17434961e-01 -2.54343301e-01 -1.81801133e-02 3.28038037e-01 7.51508594e-01 6.95713103e-01 5.66622138e-01 -2.04326555e-01 4.29815263e-01 3.71290356e-01 -9.08975601e-01 -1.23662806e+00 -1.01970041e+00 -1.72593623e-01 9.79672909e-01 5.03216028e-01 -4.07974929e-01 -5.63390374e-01 -8.06487978e-01 -9.04582143e-02 1.05948889e+00 -3.22085619e-01 5.67617007e-02 -7.54952550e-01 -5.23885310e-01 5.29623866e-01 7.41840184e-01 6.05129063e-01 -6.89155519e-01 -6.61953211e-01 2.92058110e-01 -5.02219498e-01 -1.23054934e+00 -6.11916743e-03 -2.65603751e-01 -5.68719208e-01 -1.18954527e+00 -2.66050369e-01 -5.08734763e-01 6.64396584e-01 5.43534458e-01 8.94800186e-01 1.78024061e-02 2.85976589e-01 3.93394306e-02 -2.09084213e-01 -3.79504055e-01 -2.81424314e-01 3.75944108e-01 -6.26582503e-02 -1.55101821e-01 5.42317390e-01 -7.94355035e-01 -6.06608152e-01 8.35725665e-01 -2.56695211e-01 7.38945425e-01 2.87682682e-01 4.23310965e-01 6.33541107e-01 4.08992022e-01 8.03662539e-01 -9.23323035e-01 1.93528309e-01 -7.13023186e-01 -7.12691307e-01 1.72385544e-01 -8.08436573e-01 -1.44098967e-01 7.20650375e-01 -2.52239794e-01 -1.19448805e+00 2.53654659e-01 -1.79506421e-01 -4.61625397e-01 -4.45016354e-01 3.98387849e-01 -1.13265492e-01 7.93600269e-03 6.00624442e-01 3.05952609e-01 -1.65752083e-01 -2.97376156e-01 5.32943726e-01 6.48390055e-01 7.17252374e-01 -3.37760955e-01 9.67252910e-01 4.80172187e-01 4.90351096e-02 -5.90944827e-01 -6.45517409e-01 -5.19445300e-01 -9.66841698e-01 -5.89455903e-01 6.72429383e-01 -1.01479805e+00 -8.06209147e-01 1.61617756e-01 -1.02728522e+00 -6.22774363e-01 -1.26258314e-01 4.35997337e-01 -5.80900669e-01 -4.22323346e-02 -1.59911558e-01 -6.27884328e-01 1.64525062e-01 -1.06574440e+00 7.41306007e-01 6.46164343e-02 2.81771296e-03 -8.72868001e-01 -8.42459965e-03 4.11127359e-01 5.67475975e-01 1.31487161e-01 9.46319461e-01 -4.60581452e-01 -1.04762030e+00 -2.59724885e-01 -4.72332388e-01 -2.42695123e-01 -1.17044985e-01 -2.14356959e-01 -1.10299218e+00 2.52782553e-01 -7.33615577e-01 4.13682163e-02 7.66191363e-01 6.81194514e-02 1.34813058e+00 -3.74967009e-01 -8.10369372e-01 4.27468419e-01 1.07869625e+00 1.62128344e-01 6.44719064e-01 4.09452736e-01 1.03782606e+00 1.10834241e+00 1.02668500e+00 1.11409664e-01 1.00406611e+00 9.43759799e-01 5.49604535e-01 4.17752750e-02 -3.72364596e-02 -7.86259770e-01 3.33586857e-02 6.08333707e-01 -3.30654383e-02 -2.54193366e-01 -1.33284676e+00 9.63176489e-01 -2.44524980e+00 -1.08661294e+00 -5.89085996e-01 2.10948110e+00 4.26225096e-01 1.04882158e-01 3.73158514e-01 2.04428956e-02 6.42122626e-01 -8.48337263e-03 -6.96595669e-01 4.85448986e-02 1.51193365e-01 -4.05534327e-01 7.57966995e-01 5.04274666e-01 -1.06819367e+00 1.18551028e+00 5.59115505e+00 9.48570013e-01 -1.15089869e+00 1.91320986e-01 4.86183971e-01 7.74070213e-04 -3.22196037e-01 -6.06004931e-02 -7.90960193e-01 5.88713408e-01 1.16790891e+00 -4.70571853e-02 1.55400142e-01 1.06881547e+00 5.85187018e-01 -1.65364385e-01 -9.71570551e-01 8.81032526e-01 -3.88105005e-01 -1.79532599e+00 -2.26136848e-01 -3.87284905e-02 5.86484551e-01 3.48391831e-01 3.35033680e-03 3.92052472e-01 3.34164947e-01 -1.01515973e+00 7.62751400e-01 7.76968956e-01 6.70881927e-01 -8.45430315e-01 6.56667531e-01 8.24921906e-01 -1.31797695e+00 5.09062083e-03 -3.65589172e-01 -2.15695739e-01 4.33725148e-01 4.59831148e-01 -1.50534070e+00 6.41087770e-01 6.30120337e-01 9.51869369e-01 -5.76022446e-01 1.30725884e+00 -1.92011625e-01 7.89796174e-01 -4.62036997e-01 1.45324051e-01 2.97531635e-01 -2.29185417e-01 4.13036555e-01 1.00060225e+00 3.56793582e-01 2.35326346e-02 3.26650620e-01 8.81629169e-01 2.46298715e-01 -1.29531041e-01 -8.53927433e-01 3.82214040e-01 7.73636281e-01 1.01178646e+00 -7.04335093e-01 -4.29324657e-01 -3.21481317e-01 5.02347410e-01 4.41137880e-01 3.69572520e-01 -9.14126635e-01 -1.44305140e-01 7.86463022e-01 6.06943786e-01 1.61700621e-01 -2.32001200e-01 -5.05924702e-01 -6.93015993e-01 1.28024220e-01 -1.17863402e-01 7.18318019e-03 -8.38140368e-01 -9.16237533e-01 5.37164688e-01 1.59566268e-01 -1.55525315e+00 -4.58003104e-01 -2.86540926e-01 -7.57411301e-01 7.83225775e-01 -1.83463204e+00 -1.43956280e+00 -5.72991669e-01 4.43571419e-01 5.00314057e-01 -3.81452516e-02 5.21417677e-01 5.96824825e-01 -5.94118834e-01 3.63492608e-01 8.04831535e-02 -1.51425049e-01 4.94304061e-01 -1.10448301e+00 7.20232964e-01 8.24967086e-01 1.01728179e-02 -1.05523281e-02 7.57792473e-01 -7.53752649e-01 -1.09721112e+00 -1.73777473e+00 1.07530713e+00 -7.39233375e-01 4.72702831e-01 -2.14105889e-01 -1.11278486e+00 7.49702036e-01 -3.25749725e-01 3.68662272e-03 3.33345681e-01 1.61403894e-01 -1.58333048e-01 -2.81867564e-01 -8.36676776e-01 5.76965272e-01 1.17726183e+00 -4.44444627e-01 -1.43523335e-01 3.43720794e-01 7.11684704e-01 -5.82695603e-01 -7.29493380e-01 4.59487617e-01 5.92537344e-01 -9.23352897e-01 9.48497653e-01 -4.19495851e-01 3.18164319e-01 -5.98269165e-01 5.75487502e-02 -1.31115842e+00 -2.06901789e-01 -2.25677401e-01 -2.73186658e-02 1.01978123e+00 6.46170259e-01 -5.33862054e-01 1.22298038e+00 9.43764925e-01 -5.63964188e-01 -7.75253952e-01 -9.56584156e-01 -6.91448689e-01 -5.04911579e-02 -9.62273836e-01 1.06748116e+00 9.46111917e-01 8.08239058e-02 2.34546456e-02 -4.62857544e-01 4.36214983e-01 4.36800957e-01 1.42959297e-01 1.22412479e+00 -1.32439160e+00 3.00175726e-01 -3.36198449e-01 -5.37620187e-01 -1.29547668e+00 3.28761220e-01 -8.33770156e-01 1.57985047e-01 -1.65048802e+00 -1.89239129e-01 -1.07937121e+00 5.67699708e-02 6.20903909e-01 -1.93700977e-02 1.76651016e-01 4.01574187e-02 2.88977891e-01 -6.60822332e-01 5.91702223e-01 1.06606507e+00 -1.03887193e-01 -4.04422909e-01 3.40083718e-01 -1.02446295e-01 6.51780427e-01 9.76890206e-01 -4.27535117e-01 -8.90087247e-01 -3.36532682e-01 2.85068572e-01 8.50753933e-02 6.51491404e-01 -1.13495171e+00 6.55011356e-01 -4.44343358e-01 1.02100909e-01 -1.19459474e+00 4.37534958e-01 -9.41852033e-01 3.81931335e-01 1.28470808e-02 -1.58226341e-01 -6.36721356e-03 2.79976159e-01 6.88799858e-01 -1.49798214e-01 1.90619364e-01 2.53971815e-01 3.54606539e-01 -1.17947876e+00 4.83870059e-01 -4.30007696e-01 -2.23042354e-01 1.09457338e+00 -5.28258622e-01 -5.90852737e-01 -3.71421933e-01 -3.94812077e-01 7.21094012e-01 4.04827356e-01 6.26551211e-01 5.51824093e-01 -1.27011597e+00 -4.53401476e-01 2.02488288e-01 2.41706386e-01 3.16597819e-01 4.11627203e-01 1.03473985e+00 -5.58264554e-01 5.13067663e-01 -2.21539110e-01 -7.25174367e-01 -9.35860217e-01 3.68591815e-01 2.42959112e-01 -6.21136241e-02 -1.15450335e+00 4.73783165e-01 1.03696644e-01 -7.66240478e-01 1.22847237e-01 -2.92352706e-01 -3.28903079e-01 -1.45813107e-01 5.28276503e-01 7.01767325e-01 1.31759301e-01 -8.61826897e-01 -1.93298221e-01 2.51536757e-01 8.30764174e-02 2.17939727e-02 1.25781572e+00 -3.57959270e-01 3.79038811e-01 4.67456996e-01 7.24724948e-01 -3.67157370e-01 -1.60607934e+00 -1.61762655e-01 1.21339478e-01 -5.62628150e-01 6.79730624e-02 -9.25142527e-01 -9.40536439e-01 9.56369162e-01 4.41316873e-01 -8.76096776e-04 7.50136137e-01 -1.32727504e-01 1.13865960e+00 5.00637174e-01 5.66893697e-01 -1.15702188e+00 -5.32879889e-01 3.74855697e-01 4.40380186e-01 -1.42019284e+00 -3.48773003e-01 -7.80253589e-01 -6.85039759e-01 9.23220038e-01 9.09377635e-01 2.61825383e-01 7.04780281e-01 -6.67600408e-02 5.35624288e-02 -1.43043742e-01 -7.98210025e-01 -2.67374635e-01 4.51106101e-01 9.39691663e-01 -2.02854171e-01 2.60971844e-01 2.76075929e-01 3.78394753e-01 -6.45923853e-01 3.65341194e-02 3.17651927e-01 5.18681943e-01 -6.44787252e-01 -9.17525947e-01 -2.39513442e-02 5.36701024e-01 2.80063927e-01 2.52752930e-01 5.66754825e-02 8.16594064e-01 3.60337377e-01 9.38378334e-01 2.99833715e-01 -7.34235644e-01 4.25021678e-01 1.29083395e-01 -1.91910565e-02 -1.77538350e-01 -2.35048532e-01 -6.60564899e-01 3.74256670e-01 -8.08411896e-01 -1.99521959e-01 -5.35571337e-01 -1.25320029e+00 -7.01780260e-01 -1.97111052e-02 1.81771457e-01 7.73500800e-01 1.15648794e+00 7.29909778e-01 4.30303454e-01 3.68332505e-01 -8.76284957e-01 6.59273341e-02 -7.19823360e-01 -1.10875286e-01 2.40692049e-01 2.66776085e-01 -8.57187808e-01 2.15549543e-01 6.53127730e-02]
[5.933268070220947, 0.9448404908180237]
dab6bb65-7655-48b3-ac3b-a5b15a0bdef1
a-proposal-for-multimodal-emotion-recognition
null
null
https://www.mdpi.com/2076-3417/12/1/327
https://www.mdpi.com/2076-3417/12/1/327/pdf
A proposal for Multimodal Emotion Recognition using aural transformers and Action Units on RAVDESS dataset
Emotion recognition is attracting the attention of the research community due to its multiple applications in different fields, such as medicine or autonomous driving. In this paper, we proposed an automatic emotion recognizer system that consisted of a speech emotion recognizer (SER) and a facial emotion recognizer (FER). For the SER, we evaluated a pre-trained xlsr-Wav2Vec2.0 transformer using two transfer-learning techniques: embedding extraction and fine-tuning. The best accuracy results were achieved when we fine-tuned the whole model by appending a multilayer perceptron on top of it, confirming that the training was more robust when it did not start from scratch and the previous knowledge of the network was similar to the task to adapt. Regarding the facial emotion recognizer, we extracted the Action Units of the videos and compared the performance between employing static models against sequential models. Results showed that sequential models beat static models by a narrow difference. Error analysis reported that the visual systems could improve with a detector of high-emotional load frames, which opened a new line of research to discover new ways to learn from videos. Finally, combining these two modalities with a late fusion strategy, we achieved 86.70% accuracy on the RAVDESS dataset on a subject-wise 5-CV evaluation, classifying eight emotions. Results demonstrated that these modalities carried relevant information to detect users’ emotional state and their combination allowed to improve the final system performance.
['Fernando Fernández-Martínez', 'Juan M. Montero', 'Zoraida Callejas', 'David Griol', 'Ricardo Kleinlein', 'Cristina Luna-Jiménez']
2021-12-30
null
null
null
applied-sciences-journal-2021-12
['facial-emotion-recognition', 'multimodal-emotion-recognition', 'multimodal-emotion-recognition']
['computer-vision', 'computer-vision', 'speech']
[-3.25390883e-03 1.85496897e-01 2.00053304e-01 -4.03983384e-01 -1.38569757e-01 -1.29623339e-01 4.10063535e-01 -1.00512661e-01 -7.70233572e-01 5.63804030e-01 4.22783606e-02 2.43774414e-01 2.16816261e-01 -3.50226820e-01 -6.12511039e-01 -7.15189517e-01 -1.44563481e-01 -1.79473519e-01 9.97396111e-02 -4.95472550e-01 4.59093079e-02 6.53764069e-01 -2.15500569e+00 6.16759837e-01 5.27062237e-01 1.34429979e+00 9.85230599e-03 7.41191745e-01 4.71226536e-02 7.84207642e-01 -6.88563287e-01 -4.39394236e-01 1.65093374e-02 -2.62986392e-01 -4.95797724e-01 -1.45459352e-02 3.53492387e-02 6.29392639e-03 -1.55858710e-01 7.12843716e-01 7.31167853e-01 2.81241149e-01 4.80516106e-01 -1.34428275e+00 -3.41850340e-01 1.84729427e-01 -2.95299411e-01 1.13537043e-01 4.58048761e-01 -4.95814011e-02 5.95863163e-01 -1.00946820e+00 6.86964869e-01 1.04493737e+00 6.78617060e-01 6.90479696e-01 -8.89249921e-01 -7.70671010e-01 4.05002274e-02 8.05538833e-01 -1.37671661e+00 -5.91090322e-01 8.39235961e-01 -5.55240512e-01 1.20276988e+00 1.35416120e-01 8.03202629e-01 1.37261891e+00 2.48114407e-01 5.79906404e-01 1.26020980e+00 -5.03893852e-01 2.88654268e-01 6.97972298e-01 2.91906446e-02 6.46922588e-01 -2.96431661e-01 2.53470123e-01 -5.38536429e-01 1.14900678e-01 2.22680613e-01 -1.90965086e-01 -2.22879708e-01 -9.80515704e-02 -6.37934327e-01 8.55912089e-01 3.64420027e-01 8.85962427e-01 -6.55178308e-01 -3.10288548e-01 8.26803565e-01 4.59077775e-01 4.66634929e-01 2.37835750e-01 -5.30153155e-01 -3.03979129e-01 -9.17106867e-01 -5.62033653e-01 8.00414503e-01 1.80628121e-01 5.11552036e-01 3.40490729e-01 -1.59280583e-01 7.46836245e-01 1.12183712e-01 3.60597283e-01 7.92020857e-01 -4.77748901e-01 -1.42934069e-01 6.32560134e-01 -3.11554313e-01 -1.33088887e+00 -7.19959199e-01 -3.43864769e-01 -8.56874406e-01 5.24394095e-01 7.14935735e-02 -3.28650385e-01 -8.29355299e-01 1.57958519e+00 2.20452309e-01 3.64523798e-01 3.03210765e-01 9.92914438e-01 8.47137928e-01 8.44851792e-01 2.84102559e-01 -3.50954354e-01 1.45995295e+00 -7.84208119e-01 -1.08136749e+00 3.00358832e-02 5.59782982e-01 -6.03855193e-01 7.05159128e-01 8.34063709e-01 -7.15226948e-01 -1.06823015e+00 -1.20304501e+00 3.95105451e-01 -7.58715868e-01 5.70870936e-01 4.19312567e-01 8.94728959e-01 -1.27404547e+00 6.87758029e-01 -5.61932564e-01 -5.69174409e-01 2.81176597e-01 3.89732897e-01 -7.54533291e-01 4.37638789e-01 -1.43158007e+00 1.23430955e+00 4.23034608e-01 3.62956494e-01 -6.25494897e-01 -8.41054544e-02 -7.67291367e-01 2.51859576e-01 1.58442035e-01 -2.99983203e-01 7.10416615e-01 -1.71113873e+00 -1.95678461e+00 8.43763351e-01 -7.26827532e-02 -4.71981615e-01 2.89516896e-01 -2.81254321e-01 -1.03568220e+00 3.92849386e-01 -4.52295661e-01 4.44679677e-01 1.19536674e+00 -1.03113556e+00 -5.16580105e-01 -3.80533904e-01 -3.73255074e-01 -1.34731997e-02 -6.48704708e-01 1.21425353e-01 -3.20933491e-01 -3.77852619e-01 -4.73458439e-01 -7.23894119e-01 1.81488559e-01 -3.11990738e-01 1.12928629e-01 -1.31703407e-01 8.81509423e-01 -9.51028526e-01 1.30743062e+00 -2.57525539e+00 1.89173982e-01 4.03748244e-01 3.40551846e-02 7.67106593e-01 -1.79631770e-01 2.46990576e-01 -5.53847373e-01 -5.92208877e-02 3.03993076e-01 -2.97330648e-01 -1.17271304e-01 -5.34068141e-03 -7.18157217e-02 4.85817790e-01 3.35821152e-01 5.93230486e-01 -5.07464826e-01 -4.37067181e-01 4.32580084e-01 9.73241150e-01 -3.52605432e-01 4.08729345e-01 2.75267243e-01 1.94694474e-01 -5.37396185e-02 4.15463179e-01 3.83513093e-01 1.93562850e-01 -8.47273041e-04 -6.13354802e-01 -2.21108384e-02 -6.61248863e-01 -1.27570975e+00 1.36546206e+00 -5.76084495e-01 9.32982743e-01 1.55473322e-01 -1.07977664e+00 1.35274827e+00 6.02226853e-01 5.95189035e-01 -9.49796259e-01 7.20734954e-01 -4.07438129e-02 -1.84407219e-01 -1.05112183e+00 2.84924120e-01 -1.35325536e-01 1.05873510e-01 -1.30406260e-01 4.45042789e-01 3.47170681e-01 2.86567416e-02 -2.14783445e-01 8.80457044e-01 1.27636418e-01 3.27202111e-01 1.40927523e-01 9.39353585e-01 -3.20337623e-01 2.41394415e-01 3.37195277e-01 -3.28397512e-01 2.58015066e-01 2.75916874e-01 -4.46474969e-01 -5.02348304e-01 -6.42520130e-01 -5.14808409e-02 1.11157453e+00 -1.84223920e-01 -4.05209988e-01 -7.70371556e-01 -7.51769364e-01 -3.89932245e-01 6.30959988e-01 -8.53887916e-01 -5.93712628e-01 -3.75327729e-02 -4.86196429e-01 5.75432479e-01 2.50523001e-01 3.59853983e-01 -1.32435691e+00 -9.61570382e-01 1.72339678e-01 -6.06090091e-02 -1.17835414e+00 4.07767519e-02 3.24174076e-01 -4.70653564e-01 -1.06254816e+00 -6.65366232e-01 -6.51357412e-01 3.93828332e-01 -3.86278212e-01 7.64992416e-01 -6.59888834e-02 -5.09130538e-01 6.28211200e-01 -7.21753359e-01 -5.39219439e-01 -3.84820074e-01 -2.32714161e-01 1.35632858e-01 7.07020104e-01 4.64322001e-01 -3.42678189e-01 -3.60976845e-01 1.82894215e-01 -8.05557430e-01 -2.23326564e-01 7.64895678e-01 7.06829607e-01 2.54400581e-01 1.06400689e-02 5.52224338e-01 -3.20675522e-01 7.59282172e-01 -2.95174181e-01 -2.69829571e-01 2.22476482e-01 -4.46365893e-01 -1.96639579e-02 6.29921138e-01 -6.65908515e-01 -1.12220764e+00 2.66914219e-01 -4.77152765e-01 -7.03731596e-01 -4.55397308e-01 1.94422156e-01 -1.46754041e-01 -5.84922656e-02 6.39325261e-01 2.05431227e-02 1.15569331e-01 -1.32024422e-01 2.30162516e-01 9.48359609e-01 1.92242980e-01 -9.08298641e-02 1.82084203e-01 2.38065884e-01 -2.45087087e-01 -1.12358570e+00 -3.10365051e-01 -4.14753973e-01 -4.63219225e-01 -8.65948856e-01 1.17226267e+00 -8.11745465e-01 -1.16402447e+00 5.65376222e-01 -1.12836671e+00 -1.24418996e-01 -3.55946511e-01 8.31148684e-01 -3.50212127e-01 1.69413075e-01 -3.74846756e-01 -1.02287161e+00 -4.57019687e-01 -9.73295033e-01 8.29996645e-01 4.28099334e-01 -2.03480273e-01 -8.83889139e-01 3.17934871e-01 1.00040950e-01 6.47407174e-01 2.90663183e-01 4.45721328e-01 -9.21229720e-01 4.38865602e-01 -2.48940930e-01 -5.99389561e-02 8.15483034e-01 -1.09949239e-01 1.60419613e-01 -1.40134001e+00 -6.59882128e-02 8.92751217e-02 -2.92249441e-01 8.83443296e-01 6.61664680e-02 1.00906432e+00 -9.04389247e-02 -8.16803128e-02 4.05379564e-01 1.24618673e+00 4.92101997e-01 9.21693027e-01 1.89001620e-01 3.27159524e-01 6.85427427e-01 4.33198690e-01 6.22026742e-01 2.16735844e-02 8.17709386e-01 4.68394637e-01 -4.48009193e-01 -7.32208192e-02 1.91725358e-01 8.63583207e-01 7.08006024e-01 -3.76666456e-01 -6.54553398e-02 -4.91494954e-01 1.61799029e-01 -1.69490182e+00 -1.17165792e+00 1.32544950e-01 1.92194653e+00 5.39328098e-01 3.58586460e-02 1.47112966e-01 2.92932510e-01 6.01011515e-01 -2.37723738e-02 -3.23248804e-01 -1.08474314e+00 -2.01476261e-01 4.42632645e-01 4.17105965e-02 2.39337757e-01 -1.11150301e+00 8.60903442e-01 5.72153234e+00 8.15492988e-01 -1.66404963e+00 1.58279836e-01 4.98825133e-01 -1.38921022e-01 4.25397605e-01 -5.74137390e-01 -6.36174619e-01 5.18194497e-01 1.46305609e+00 3.03721517e-01 2.94733644e-01 9.46729660e-01 2.94478506e-01 -1.17216147e-01 -7.35427976e-01 1.29184818e+00 5.87255359e-01 -7.67164052e-01 -2.20069498e-01 -2.37875476e-01 2.43875891e-01 -1.46099597e-01 -2.00457454e-01 7.82860339e-01 -4.33458775e-01 -9.24681544e-01 4.69623387e-01 1.09791815e+00 5.61281800e-01 -6.79720044e-01 1.06174779e+00 7.26808906e-02 -1.03355956e+00 -1.27774268e-01 -5.14613651e-02 -6.15408979e-02 2.87699476e-02 5.24710357e-01 -8.30527604e-01 4.74625677e-01 9.12394345e-01 5.98311305e-01 -7.67224610e-01 7.96661675e-01 7.16658728e-03 5.00473142e-01 -1.83709785e-01 -1.41059130e-01 -1.03164606e-01 -1.23389408e-01 3.63524169e-01 1.55526805e+00 3.84898633e-01 -2.02373434e-02 -2.61679500e-01 3.63750339e-01 1.73754618e-01 5.66922963e-01 -4.51809376e-01 -4.99675013e-02 -2.26098955e-01 1.63283265e+00 -6.51907027e-01 -3.47532541e-01 -3.57461393e-01 1.18069589e+00 2.04987571e-01 2.47222319e-01 -1.16549492e+00 -6.06882691e-01 4.42701459e-01 -1.98691010e-01 4.96345192e-01 1.49523437e-01 1.81966379e-01 -1.20773780e+00 -1.95532933e-01 -7.25046277e-01 3.75958949e-01 -1.12098479e+00 -9.91991580e-01 1.04858541e+00 -2.01318681e-01 -1.03288341e+00 -2.36232415e-01 -9.23841178e-01 -3.89607221e-01 4.98573303e-01 -1.30725241e+00 -8.12326074e-01 -4.32679474e-01 7.98118353e-01 3.47770572e-01 -2.72288501e-01 1.00052536e+00 5.00094354e-01 -6.78957105e-01 6.67416692e-01 -2.32811496e-01 4.69644479e-02 9.13911402e-01 -8.72925878e-01 -6.85520232e-01 8.19777250e-01 1.89344406e-01 6.25057518e-02 6.51290655e-01 -2.23734304e-01 -1.22861242e+00 -7.48524308e-01 8.45416069e-01 1.11243688e-01 3.74318749e-01 -2.46157914e-01 -9.81617570e-01 1.97683051e-01 5.21806180e-01 -2.03727335e-02 7.63682187e-01 -1.22567430e-01 -1.46652326e-01 -3.18826765e-01 -1.24216640e+00 3.23303133e-01 5.38681805e-01 -4.23818767e-01 -6.43815339e-01 -2.10600883e-01 2.84195215e-01 -1.78864479e-01 -9.85452354e-01 4.90723670e-01 7.78798282e-01 -1.24352360e+00 6.85451031e-01 -4.83314663e-01 2.55895376e-01 -8.85635912e-02 -1.14704669e-01 -1.44369829e+00 -1.58728287e-01 -1.25207052e-01 -2.80837238e-01 1.33025491e+00 1.44381896e-01 -5.85082352e-01 3.19728136e-01 4.64414388e-01 -8.06635991e-02 -7.46027887e-01 -8.17729056e-01 -4.89801019e-01 -5.81329107e-01 -4.93985713e-01 1.50579885e-01 8.09804440e-01 8.09436217e-02 4.54188079e-01 -4.73751932e-01 2.17429493e-02 -1.14057727e-01 -3.13139141e-01 5.88500381e-01 -1.21158814e+00 -2.64132679e-01 -6.23807907e-01 -8.67762208e-01 -1.33739918e-01 5.32169938e-01 -8.01568389e-01 -1.91690698e-01 -1.21061099e+00 -1.86460823e-01 2.65862435e-01 -6.23575568e-01 7.61662126e-01 -2.96614511e-04 3.91978920e-01 2.60366321e-01 -3.70524019e-01 -6.00894272e-01 7.67784715e-01 7.77247846e-01 -2.59900764e-02 -3.78876418e-01 -8.21188837e-02 -4.67112035e-01 6.91236198e-01 7.50484943e-01 -3.13765168e-01 -1.57755464e-01 2.58605629e-01 3.04571241e-01 -1.35638177e-01 3.56237352e-01 -1.32020259e+00 2.02548370e-01 3.41480434e-01 7.40647316e-01 -1.47280276e-01 5.60661376e-01 -1.31024361e+00 3.24084908e-01 4.10810858e-01 -1.13904700e-01 -2.88666546e-01 7.08365262e-01 2.14599639e-01 -3.99765491e-01 -1.93735734e-01 7.59913504e-01 3.67047846e-01 -1.29763484e+00 -1.31429374e-01 -6.35078669e-01 -4.52821195e-01 1.43921483e+00 -3.57018501e-01 4.51488681e-02 -2.99778730e-01 -1.30274212e+00 -1.68265954e-01 3.22824754e-02 6.77842736e-01 6.69095039e-01 -1.02980638e+00 -5.39245069e-01 4.44027930e-01 1.01605952e-01 -8.92144978e-01 6.64144397e-01 1.01093245e+00 2.12371759e-02 3.44895124e-02 -6.54684126e-01 -5.29864907e-01 -1.64763546e+00 7.31997669e-01 5.65002978e-01 -1.17349386e-01 -2.31082410e-01 5.67281604e-01 -3.39306891e-01 -1.50792241e-01 4.36038315e-01 -3.06668971e-02 -9.26823676e-01 7.22476482e-01 6.27825558e-01 3.77962410e-01 3.99274290e-01 -9.83510315e-01 -4.56339777e-01 6.75125718e-01 2.47737437e-01 -2.23377664e-02 1.38054121e+00 -1.33675903e-01 1.38584554e-01 6.39935315e-01 1.41534698e+00 -8.69429633e-02 -8.83135140e-01 1.37911454e-01 -2.87861019e-01 -4.56151441e-02 2.62661368e-01 -9.74078596e-01 -1.32717001e+00 9.17581618e-01 1.27065551e+00 1.87353805e-01 1.65398502e+00 -2.11441219e-01 2.61077344e-01 2.82240093e-01 1.71079502e-01 -1.33478165e+00 9.21886340e-02 4.94069844e-01 9.79009807e-01 -1.26144457e+00 -3.68543416e-01 8.76913592e-02 -1.03141212e+00 1.54902172e+00 4.08783942e-01 -5.17137162e-02 8.35507154e-01 1.72677040e-01 1.44898579e-01 -2.97876596e-01 -6.83092535e-01 -4.19748455e-01 6.72061026e-01 4.60436165e-01 2.75174946e-01 -5.91440536e-02 -2.56011307e-01 1.13031030e+00 -9.53207090e-02 4.25527573e-01 2.69464225e-01 5.75893223e-01 -2.95567960e-01 -7.41335213e-01 -3.75091493e-01 2.64224648e-01 -3.32672030e-01 3.00490439e-01 -1.85591429e-01 8.50040615e-01 4.89228547e-01 1.08791101e+00 5.18567786e-02 -8.58007789e-01 7.09783494e-01 5.68861544e-01 3.53003055e-01 -1.65368736e-01 -8.98215055e-01 -6.39852956e-02 5.56284152e-02 -1.01083159e+00 -6.18002415e-01 -5.39026737e-01 -1.10790443e+00 9.56034288e-02 -2.74750322e-01 1.80839613e-01 9.91120517e-01 9.83108222e-01 6.25319064e-01 8.44833374e-01 7.58798838e-01 -1.01690722e+00 -8.40549171e-02 -9.42668855e-01 -5.81343234e-01 4.92889196e-01 1.91777796e-01 -7.20383942e-01 -5.06292522e-01 3.60999495e-01]
[13.364147186279297, 5.017551898956299]
86fc2b08-194d-43a6-860a-90d5d4554cea
tgif-a-new-dataset-and-benchmark-on-animated
1604.02748
null
http://arxiv.org/abs/1604.02748v2
http://arxiv.org/pdf/1604.02748v2.pdf
TGIF: A New Dataset and Benchmark on Animated GIF Description
With the recent popularity of animated GIFs on social media, there is need for ways to index them with rich metadata. To advance research on animated GIF understanding, we collected a new dataset, Tumblr GIF (TGIF), with 100K animated GIFs from Tumblr and 120K natural language descriptions obtained via crowdsourcing. The motivation for this work is to develop a testbed for image sequence description systems, where the task is to generate natural language descriptions for animated GIFs or video clips. To ensure a high quality dataset, we developed a series of novel quality controls to validate free-form text input from crowdworkers. We show that there is unambiguous association between visual content and natural language descriptions in our dataset, making it an ideal benchmark for the visual content captioning task. We perform extensive statistical analyses to compare our dataset to existing image and video description datasets. Next, we provide baseline results on the animated GIF description task, using three representative techniques: nearest neighbor, statistical machine translation, and recurrent neural networks. Finally, we show that models fine-tuned from our animated GIF description dataset can be helpful for automatic movie description.
['Liangliang Cao', 'Yuncheng Li', 'Jiebo Luo', 'Joel Tetreault', 'Alejandro Jaimes', 'Yale Song', 'Larry Goldberg']
2016-04-10
tgif-a-new-dataset-and-benchmark-on-animated-1
http://openaccess.thecvf.com/content_cvpr_2016/html/Li_TGIF_A_New_CVPR_2016_paper.html
http://openaccess.thecvf.com/content_cvpr_2016/papers/Li_TGIF_A_New_CVPR_2016_paper.pdf
cvpr-2016-6
['video-description']
['computer-vision']
[ 2.43773967e-01 -1.72189698e-01 -2.98185647e-01 -4.83109087e-01 -1.06605673e+00 -9.32524562e-01 8.89745772e-01 -2.12297007e-01 -3.53549540e-01 7.69551754e-01 6.90767109e-01 1.85458601e-04 4.54352677e-01 -5.27068138e-01 -8.53289187e-01 -2.37735555e-01 -9.24712121e-02 4.90406781e-01 2.83118188e-01 -4.04716671e-01 2.07324624e-01 1.38069212e-01 -1.56553364e+00 8.82501543e-01 6.75569773e-02 8.78548026e-01 3.91771287e-01 1.10537910e+00 -6.49997443e-02 1.44630647e+00 -6.48484230e-01 -6.60622954e-01 3.28543842e-01 -6.87765419e-01 -1.16189027e+00 3.38743746e-01 7.84564614e-01 -6.65356040e-01 -7.26761103e-01 7.78959513e-01 5.46719313e-01 3.65658671e-01 7.27269709e-01 -1.67559445e+00 -1.31608510e+00 7.55385458e-01 -1.49208412e-01 2.53708690e-01 7.90603101e-01 4.84604448e-01 1.15070212e+00 -8.11809361e-01 1.32507968e+00 1.31782818e+00 4.92828101e-01 8.12311530e-01 -9.98936832e-01 -4.12170529e-01 -1.37915149e-01 1.67822957e-01 -1.61880982e+00 -7.31677055e-01 2.81317860e-01 -6.13040268e-01 8.81458044e-01 2.29483038e-01 5.67509770e-01 1.49617445e+00 -2.09031686e-01 6.85792446e-01 7.67430127e-01 -3.67233098e-01 3.42277959e-02 1.77317008e-01 -3.27476591e-01 7.12005854e-01 -2.30204344e-01 -2.85696477e-01 -7.69850373e-01 -9.69662741e-02 8.99972618e-01 -5.24024069e-01 -2.06805736e-01 -2.52837270e-01 -1.74850094e+00 9.18612361e-01 2.35507593e-01 1.45974740e-01 -2.40861047e-02 5.34867465e-01 6.49158359e-01 3.27670902e-01 5.00067651e-01 6.19989753e-01 1.15013219e-01 -4.42579597e-01 -9.19345737e-01 7.09918439e-01 7.75654912e-01 1.53130305e+00 5.80439270e-01 7.18284771e-02 -5.19616842e-01 8.91800284e-01 1.36779591e-01 8.83040786e-01 5.30950487e-01 -1.61385548e+00 6.10281050e-01 -1.04953423e-01 2.95408219e-01 -1.27536297e+00 9.11140367e-02 4.91078854e-01 -5.12544513e-01 -2.85100400e-01 5.09101093e-01 1.10956185e-01 -5.82875013e-01 1.56206226e+00 -2.04194427e-01 -2.64113933e-01 -1.23801650e-02 1.20999050e+00 1.31222975e+00 8.07092369e-01 9.59957093e-02 9.85006541e-02 1.27185607e+00 -1.07359433e+00 -6.80456460e-01 9.93606970e-02 6.46590054e-01 -1.00301421e+00 1.40499818e+00 -5.00612333e-02 -1.18271613e+00 -5.48035026e-01 -8.09369922e-01 -4.40892130e-01 -2.21381024e-01 -2.36454159e-01 3.42946023e-01 5.32803908e-02 -1.42043519e+00 4.52891856e-01 -2.19726190e-01 -5.73928535e-01 1.54893681e-01 -8.08598846e-02 -5.73835731e-01 -1.99980244e-01 -1.20901823e+00 5.89744806e-01 4.04207498e-01 -5.63776612e-01 -9.13751662e-01 -2.33293518e-01 -1.01587522e+00 -4.42036867e-01 1.56245261e-01 -6.92039549e-01 1.64555037e+00 -1.18283594e+00 -1.11979949e+00 1.32462883e+00 -2.89919060e-02 -5.28786659e-01 5.06646276e-01 2.35521987e-01 -3.68474364e-01 6.33748949e-01 3.78208429e-01 1.63209987e+00 8.39777946e-01 -1.17176700e+00 -4.89287972e-01 3.85335743e-01 4.15389240e-01 2.34728292e-01 -2.13858411e-01 3.71935219e-01 -6.29807353e-01 -9.33847249e-01 -4.63575661e-01 -1.24825823e+00 1.64366439e-01 1.27428353e-01 -4.62600201e-01 1.10814348e-02 5.61091185e-01 -7.78619468e-01 1.01411998e+00 -2.06519985e+00 -5.40519878e-02 -2.82291263e-01 3.68304521e-01 -1.45566255e-01 -5.15038013e-01 5.85893512e-01 8.40044394e-02 5.25351465e-01 1.01247706e-01 -3.81946534e-01 1.64848179e-01 4.72791940e-02 -5.94457507e-01 2.28763387e-01 1.51218802e-01 1.23977613e+00 -1.00270963e+00 -9.30092692e-01 -5.30556552e-02 3.96940291e-01 -5.93154609e-01 3.97113383e-01 -6.18119657e-01 3.73909831e-01 -1.58526510e-01 6.39365375e-01 1.09296300e-01 -6.33937776e-01 -2.19614789e-01 -2.55529314e-01 -2.79399697e-02 -2.08522682e-03 -6.41879976e-01 1.77786207e+00 -2.95253783e-01 1.28811932e+00 -3.35383922e-01 -2.79246867e-01 8.09897423e-01 4.71915543e-01 4.96900916e-01 -8.42803240e-01 1.09686032e-01 5.22941947e-02 -5.39461792e-01 -8.71629119e-01 1.20079350e+00 -3.66795026e-02 -4.36824471e-01 7.02048779e-01 4.28126529e-02 -4.43450689e-01 6.08073771e-01 5.92025638e-01 1.04534042e+00 2.33700052e-01 2.26604596e-01 3.63752544e-02 4.08752561e-02 6.42131031e-01 1.16018485e-03 9.67843592e-01 -2.48827040e-01 1.40287375e+00 3.36959392e-01 -6.48558557e-01 -1.86992550e+00 -8.09120893e-01 1.20958537e-01 1.23333192e+00 2.54640669e-01 -4.95751292e-01 -8.18728924e-01 -4.33741480e-01 -1.99504852e-01 4.01467592e-01 -4.45258439e-01 3.15821022e-01 -5.33532262e-01 -3.48255247e-01 8.15402389e-01 3.25510710e-01 7.04578817e-01 -1.28781688e+00 -2.82316804e-01 1.05621159e-01 -7.98258424e-01 -1.65200913e+00 -9.33936238e-01 -5.86024284e-01 -2.55561739e-01 -8.43496323e-01 -1.24835467e+00 -9.01692569e-01 3.11909586e-01 6.23178542e-01 1.65760922e+00 6.11414872e-02 -1.17258087e-01 7.78489530e-01 -8.26532960e-01 7.48244748e-02 -9.83531177e-01 4.99562733e-02 1.04233928e-01 -3.82147461e-01 1.80702180e-01 -2.25287989e-01 -4.13994372e-01 5.72912753e-01 -1.16583872e+00 5.64622939e-01 1.48679271e-01 4.73718047e-01 5.61747551e-01 -6.00673139e-01 3.58360469e-01 -5.97817481e-01 6.67327523e-01 -6.94682300e-01 -2.66327351e-01 2.71983504e-01 -6.93611801e-02 -8.20117828e-04 6.05673909e-01 -6.30909204e-01 -5.56661129e-01 6.14778064e-02 1.37601465e-01 -6.23806536e-01 -1.14503376e-01 1.83716729e-01 2.71209896e-01 -6.04514666e-02 9.72750962e-01 1.29679576e-01 7.18321232e-03 -2.70617783e-01 6.90797031e-01 9.55163121e-01 8.79815578e-01 -3.99332643e-01 8.55250835e-01 4.12307680e-01 -4.65992212e-01 -8.09213936e-01 -6.69514239e-01 -4.38314945e-01 -3.68271530e-01 -4.71770883e-01 1.14085889e+00 -1.32892382e+00 -6.59667075e-01 2.59246111e-01 -1.47570491e+00 -6.13189876e-01 -1.77096069e-01 1.80088878e-01 -1.02434671e+00 4.16448057e-01 -9.32577908e-01 -2.93385267e-01 -2.27537319e-01 -1.12256610e+00 1.27955770e+00 -1.62618622e-01 -6.37993932e-01 -8.47432494e-01 7.65061229e-02 6.08117342e-01 5.59066176e-01 3.22464645e-01 4.80915546e-01 -5.17499089e-01 -7.00142384e-01 5.14945090e-02 -5.54971576e-01 4.76982370e-02 -8.26024264e-02 2.52772540e-01 -7.02186346e-01 -1.46859720e-01 -5.59080899e-01 -9.05020595e-01 5.13683856e-01 3.06721199e-02 1.02115738e+00 -7.28371620e-01 2.43205965e-01 5.12500703e-01 1.30008733e+00 3.17641124e-02 6.63664818e-01 6.86564624e-01 8.31765056e-01 4.12489176e-01 6.56341612e-01 5.52924633e-01 6.65024161e-01 1.07600451e+00 7.56844133e-02 7.88605120e-03 -4.96015072e-01 -4.86739010e-01 4.63756472e-01 9.06965375e-01 7.25593418e-02 -8.68090808e-01 -9.72234011e-01 4.68585551e-01 -1.80637825e+00 -1.27502954e+00 9.87217501e-02 1.89565992e+00 7.57658124e-01 -2.41101801e-01 4.90718544e-01 -3.55474651e-01 9.06662762e-01 2.94505030e-01 -2.41237968e-01 -3.72502893e-01 -5.29342592e-01 -5.49726963e-01 5.41923523e-01 3.24204057e-01 -1.21092927e+00 1.03489733e+00 7.59247494e+00 7.97427356e-01 -1.02095139e+00 6.86930642e-02 7.01965332e-01 -2.17213452e-01 -4.17725861e-01 -2.91227698e-01 -7.50613570e-01 5.33338070e-01 1.04982257e+00 -3.99691343e-01 7.96455443e-01 7.09666014e-01 4.69310343e-01 9.47379172e-02 -1.10569060e+00 1.27122176e+00 5.29917657e-01 -1.68529975e+00 4.78033423e-01 -2.52873003e-01 9.62894619e-01 1.41832769e-01 8.62433836e-02 1.42468080e-01 5.28149426e-01 -1.07961249e+00 1.15835738e+00 4.55221474e-01 1.31547666e+00 -2.12886944e-01 5.53850472e-01 -1.88814834e-01 -1.12056279e+00 2.80851483e-01 -4.54367489e-01 3.58467326e-02 5.22130251e-01 1.13032207e-01 -1.03794074e+00 1.44596072e-02 6.52513444e-01 9.39126253e-01 -8.27301919e-01 7.90133893e-01 1.57789275e-01 2.95859337e-01 1.40015721e-01 -1.98001996e-01 3.33088845e-01 1.65137798e-01 6.02984905e-01 1.32442355e+00 2.62848824e-01 -1.78242251e-02 5.44636026e-02 8.97365212e-01 -3.66071522e-01 2.17050567e-01 -1.07609725e+00 -5.89871526e-01 3.77489179e-01 1.00105381e+00 -8.49849224e-01 -5.43138802e-01 -4.05413330e-01 1.13014257e+00 1.74469724e-01 4.29067761e-01 -1.08604038e+00 -2.37501055e-01 4.21282142e-01 3.07128012e-01 2.27251127e-02 -4.74729776e-01 3.36954385e-01 -1.41099155e+00 -9.04821008e-02 -1.36876237e+00 1.37132123e-01 -1.61185431e+00 -1.31138468e+00 1.01789033e+00 4.02798086e-01 -1.52960551e+00 -9.22082961e-01 -2.62472689e-01 5.00663789e-03 3.84607524e-01 -9.78967369e-01 -1.06872416e+00 -5.78552485e-01 5.21330714e-01 8.43763053e-01 -3.37578654e-01 6.74865484e-01 4.10570681e-01 -5.97671717e-02 4.41543847e-01 3.53206694e-02 3.67856383e-01 1.02048266e+00 -9.05737638e-01 9.79822874e-01 5.02237797e-01 4.44591135e-01 2.66403288e-01 9.70112503e-01 -6.89405918e-01 -1.20374954e+00 -1.15400040e+00 8.54175210e-01 -6.39934838e-01 7.74428844e-01 -4.91123587e-01 -6.43767178e-01 7.25378335e-01 2.56307840e-01 7.19075799e-02 4.88076150e-01 -5.73144376e-01 -4.57359582e-01 4.29683268e-01 -9.00003433e-01 8.40547562e-01 1.36559331e+00 -8.58895540e-01 -3.20805073e-01 8.14255118e-01 1.20111310e+00 -4.37955260e-01 -9.10510719e-01 2.48808805e-02 5.58870852e-01 -7.62133539e-01 9.42651272e-01 -5.98038733e-01 1.07261527e+00 -2.49835968e-01 -5.05056381e-01 -9.74029183e-01 -2.49036357e-01 -6.73776686e-01 2.48049229e-01 1.39640355e+00 4.46030766e-01 2.18831554e-01 7.45044470e-01 8.72811198e-01 1.36269614e-01 -5.09897098e-02 -5.19562840e-01 -1.00693631e+00 -8.63568410e-02 -3.28459322e-01 4.17658776e-01 1.00208127e+00 -6.01150319e-02 3.81368518e-01 -8.38882506e-01 -4.48156089e-01 2.67332852e-01 -5.42419180e-02 1.02384853e+00 -4.02804732e-01 -2.78218746e-01 -1.09243982e-01 -7.49016523e-01 -1.19673359e+00 3.20219606e-01 -9.74398434e-01 2.17358083e-01 -1.50202882e+00 6.31752729e-01 -2.48319596e-01 5.24513900e-01 4.68177080e-01 1.41906962e-01 1.06756020e+00 5.79402149e-01 6.35267675e-01 -1.43340659e+00 3.67451817e-01 1.37225461e+00 -2.44989231e-01 2.75741629e-02 -7.13883579e-01 -4.38215047e-01 4.39241171e-01 5.68454087e-01 -4.77708697e-01 -3.46851945e-01 -7.77540386e-01 3.29683185e-01 3.29112411e-01 5.35099983e-01 -1.02844977e+00 -1.65546988e-03 -2.27948427e-01 8.43911842e-02 -1.36232987e-01 4.26874846e-01 -3.63858372e-01 3.91180724e-01 -1.92931682e-01 -9.85350013e-01 5.70095658e-01 -3.50285880e-03 2.84187257e-01 -3.55664402e-01 -2.00660050e-01 6.26379967e-01 -3.76033217e-01 -1.14439487e+00 4.07019168e-01 -7.04990685e-01 6.66138589e-01 7.64201820e-01 -1.90487325e-01 -6.43550336e-01 -1.20037818e+00 -5.80017865e-01 -1.02968648e-01 1.06596863e+00 7.85865545e-01 6.05806649e-01 -1.69844174e+00 -1.05268550e+00 -3.73443604e-01 4.67844784e-01 -5.07443190e-01 5.19391149e-02 2.13767365e-01 -1.04245520e+00 4.29066837e-01 -3.96546543e-01 -4.92367625e-01 -1.36787534e+00 7.15697885e-01 1.69590185e-03 3.16188514e-01 -6.26054168e-01 5.53721726e-01 9.83704403e-02 5.91900852e-03 4.55402993e-02 7.87208453e-02 -1.26394659e-01 -6.25011995e-02 8.88267457e-01 1.03532091e-01 -3.91867757e-01 -1.32090259e+00 -3.33441459e-02 4.42318112e-01 1.47426993e-01 -5.86019993e-01 1.03288078e+00 -6.62461996e-01 1.62375093e-01 5.47903419e-01 1.69787395e+00 -1.75625324e-01 -1.14471638e+00 -1.31728277e-01 -2.01156288e-01 -5.34859419e-01 -4.11721796e-01 -3.83968264e-01 -8.64489973e-01 4.46024984e-01 3.05755049e-01 3.25239152e-01 6.89218879e-01 1.61464185e-01 1.02779174e+00 6.82011425e-01 5.68758905e-01 -8.76637399e-01 4.98980165e-01 5.84607542e-01 1.16158617e+00 -1.51386964e+00 -9.52899531e-02 -2.28369057e-01 -1.06643629e+00 1.14979076e+00 2.80760288e-01 7.65293837e-02 -2.01421995e-02 7.16094002e-02 3.06659788e-01 1.08292587e-02 -8.39157462e-01 -1.08821198e-01 3.40860426e-01 6.13550723e-01 3.95626366e-01 -1.94350332e-01 1.46254420e-01 2.91723788e-01 -5.56191206e-01 1.18582025e-01 9.04976487e-01 5.70783138e-01 -3.75484884e-01 -7.94353604e-01 -3.74178261e-01 2.22844139e-01 -5.61022580e-01 -2.35769257e-01 -6.01443887e-01 9.27638948e-01 -2.71401674e-01 1.05254173e+00 1.36790812e-01 -4.94857669e-01 3.39255603e-05 -1.74773276e-01 3.57920706e-01 -5.18001258e-01 -3.82598996e-01 -3.75274599e-01 4.79960889e-01 -6.64887667e-01 -6.01499259e-01 -5.69366157e-01 -1.01960242e+00 -6.82761967e-01 1.81889400e-01 1.10152140e-01 6.18158042e-01 6.97237611e-01 2.40095794e-01 9.04952548e-03 4.87412900e-01 -9.53468680e-01 -3.86589020e-02 -6.96647942e-01 -4.44918156e-01 1.04485989e+00 1.35775805e-01 -2.70860761e-01 -5.58302045e-01 8.89811516e-01]
[10.692038536071777, 0.9362819790840149]
d7c936fd-c336-416e-85a9-014dd22dd74c
sport-task-fine-grained-action-detection-and
2301.13576
null
https://arxiv.org/abs/2301.13576v1
https://arxiv.org/pdf/2301.13576v1.pdf
Sport Task: Fine Grained Action Detection and Classification of Table Tennis Strokes from Videos for MediaEval 2022
Sports video analysis is a widespread research topic. Its applications are very diverse, like events detection during a match, video summary, or fine-grained movement analysis of athletes. As part of the MediaEval 2022 benchmarking initiative, this task aims at detecting and classifying subtle movements from sport videos. We focus on recordings of table tennis matches. Conducted since 2019, this task provides a classification challenge from untrimmed videos recorded under natural conditions with known temporal boundaries for each stroke. Since 2021, the task also provides a stroke detection challenge from unannotated, untrimmed videos. This year, the training, validation, and test sets are enhanced to ensure that all strokes are represented in each dataset. The dataset is now similar to the one used in [1, 2]. This research is intended to build tools for coaches and athletes who want to further evaluate their sport performances.
['Julien Morlier', 'Laurent Mascarilla', 'Renaud Péteri', 'Jenny Benois-Pineau', 'Boris Mansencal', 'Jordan Calandre', 'Pierre-Etienne Martin']
2023-01-31
null
null
null
null
['fine-grained-action-detection']
['computer-vision']
[ 4.78246570e-01 -3.71226609e-01 -5.17736018e-01 8.58116373e-02 -5.55353761e-01 -7.01686263e-01 6.05170429e-01 7.48093240e-03 -6.41835272e-01 5.24760842e-01 5.62536955e-01 2.65823483e-01 1.95566863e-01 -3.70366633e-01 -6.14543974e-01 -2.99669266e-01 -2.46428683e-01 -9.30524915e-02 8.24036777e-01 -1.56177253e-01 6.65055871e-01 6.27060294e-01 -2.10090780e+00 8.90973032e-01 1.06294610e-01 7.88542569e-01 -2.05696210e-01 1.16857183e+00 3.67793322e-01 1.13070750e+00 -8.88114929e-01 -6.68697178e-01 4.17302758e-01 -7.00464427e-01 -5.71818709e-01 -2.27712244e-02 1.10363674e+00 -2.56258190e-01 -8.65727365e-01 7.52391160e-01 6.66166067e-01 5.40761411e-01 3.72386366e-01 -1.26721203e+00 1.64734751e-01 4.69720721e-01 -3.80217582e-01 9.28125203e-01 7.81576216e-01 2.64562219e-01 6.13614261e-01 -5.99021196e-01 1.07616150e+00 7.40914166e-01 8.55479777e-01 4.10623729e-01 -7.43834615e-01 -7.95445979e-01 -1.76257625e-01 7.62164056e-01 -1.15732098e+00 -3.73384893e-01 7.24766612e-01 -8.31940114e-01 5.06082714e-01 4.99785841e-01 1.04118991e+00 1.44861865e+00 1.59861222e-01 1.09724498e+00 8.09293926e-01 -2.29853347e-01 2.85725016e-02 -3.71787578e-01 1.92154974e-01 4.07613926e-02 1.98616281e-01 2.31009111e-01 -1.29761910e+00 1.69364437e-01 6.98847413e-01 -6.40351698e-02 -8.83221403e-02 -2.45731637e-01 -1.33401990e+00 4.76140648e-01 -8.31029639e-02 3.98304284e-01 -3.39955002e-01 -1.60237458e-02 9.19309020e-01 3.00005347e-01 2.02045187e-01 3.71674210e-01 1.77339926e-01 -1.15724218e+00 -1.33891451e+00 8.48290145e-01 6.74164474e-01 6.87040687e-01 1.11755721e-01 1.97727904e-01 -5.65843523e-01 6.92918837e-01 -5.07590234e-01 1.09299392e-01 6.03560984e-01 -1.11895001e+00 5.35857677e-01 5.17736077e-01 -5.97598311e-03 -1.24081421e+00 -4.16003346e-01 -2.93995976e-01 -3.02736074e-01 2.82826364e-01 7.62302041e-01 4.49165665e-02 -7.21915841e-01 1.18296504e+00 1.36326686e-01 5.86274683e-01 -2.63991117e-01 1.19096911e+00 1.07046294e+00 3.32427531e-01 1.70342978e-02 -3.60622862e-03 1.38589191e+00 -8.01896274e-01 -7.49168158e-01 -2.29299873e-01 3.91315639e-01 -7.86358893e-01 9.44059789e-01 9.24276531e-01 -1.22779965e+00 -8.16487670e-01 -1.10923231e+00 1.01496592e-01 -1.33896917e-01 1.21991180e-01 3.57488394e-01 7.63565540e-01 -2.66131580e-01 7.54354656e-01 -6.88312352e-01 -2.34104365e-01 5.00123799e-01 -1.47330314e-01 -6.25365376e-01 6.43092170e-02 -1.10588658e+00 8.58975232e-01 3.97169650e-01 6.97069094e-02 -8.95726562e-01 -7.12521911e-01 -8.10445964e-01 -4.19597507e-01 6.00166082e-01 1.70550480e-01 1.28057706e+00 -9.12636042e-01 -1.20375264e+00 1.27672756e+00 2.51508951e-01 -6.48226798e-01 9.94505525e-01 -6.67236090e-01 -5.93795300e-01 3.63540202e-01 1.16878264e-01 4.46952842e-02 5.89282095e-01 -6.00051761e-01 -8.81497324e-01 -2.42813230e-01 -7.36739933e-02 1.79516464e-01 -6.17978796e-02 5.75699687e-01 -6.64540589e-01 -1.26379979e+00 -1.92400500e-01 -9.31398869e-01 1.92582130e-01 -3.61138999e-01 -2.72967845e-01 1.87583059e-01 7.64265656e-01 -1.04721785e+00 1.50988543e+00 -2.30760741e+00 1.61299542e-01 -1.25865331e-02 1.31163508e-01 3.68620694e-01 6.27463460e-02 5.70619524e-01 -4.03075926e-02 -2.09089965e-01 1.46165475e-01 2.36275345e-01 -9.67103317e-02 -1.96296163e-02 -1.71851531e-01 5.35362422e-01 -1.58358410e-01 7.01189518e-01 -8.96122634e-01 -5.71099758e-01 3.21453512e-01 7.20637152e-03 -3.99750918e-01 4.53194678e-02 3.86883020e-01 4.13215935e-01 -2.30315000e-01 7.33318150e-01 3.18148375e-01 5.94373524e-01 4.42545190e-02 -3.68187189e-01 -3.00110817e-01 -2.53179222e-01 -1.55280972e+00 1.81164300e+00 1.70553938e-01 1.26487279e+00 -1.41515106e-01 -1.09022880e+00 7.44076610e-01 2.13914201e-01 7.21601307e-01 -7.83371329e-01 5.11219092e-02 -7.63332844e-02 -2.10159458e-02 -9.98085022e-01 9.73847449e-01 1.95081476e-02 -3.94139051e-01 2.73201447e-02 1.11318380e-01 2.12080225e-01 1.08499098e+00 1.56834021e-01 1.31557286e+00 5.01329124e-01 1.52077079e-01 1.32783905e-01 2.27867007e-01 3.64354849e-01 6.63033843e-01 8.87447119e-01 -6.26326382e-01 9.01838899e-01 4.44717407e-01 -6.05710626e-01 -9.74379838e-01 -1.21285427e+00 2.01036572e-01 1.19178534e+00 3.10672373e-01 -6.06073499e-01 -7.93930113e-01 -3.64591777e-01 -1.10873468e-01 1.95922151e-01 -6.35130763e-01 -1.84803963e-01 -1.05787790e+00 -2.84130067e-01 1.00790632e+00 8.01561773e-01 4.54607546e-01 -1.29118407e+00 -1.08219373e+00 2.78521717e-01 -4.52906191e-01 -1.19697416e+00 -6.37361705e-01 -3.63040686e-01 -5.58567047e-01 -1.59613574e+00 -8.96761298e-01 -6.64980114e-01 -1.97480395e-01 1.12318374e-01 1.11637843e+00 -1.67717487e-01 -1.10562301e+00 3.67525071e-01 -6.83082581e-01 -4.93795156e-01 -2.12899119e-01 -3.28092501e-02 1.43700555e-01 1.22852452e-01 5.22793055e-01 -2.64485985e-01 -5.40511191e-01 5.91210067e-01 -9.01483953e-01 -1.51283860e-01 4.01198685e-01 4.98869687e-01 5.84919989e-01 -2.71903694e-01 2.93190509e-01 -6.83372498e-01 4.36675727e-01 -1.44277036e-01 -2.11363897e-01 -6.91149011e-03 1.59686550e-01 -6.09382451e-01 1.77678779e-01 -8.42616856e-01 -7.34697938e-01 6.77568838e-02 1.42540604e-01 -4.97058630e-01 -3.31715405e-01 2.42930681e-01 1.18440673e-01 2.86051445e-02 1.05325925e+00 1.90856442e-01 -1.05211817e-01 -6.09993994e-01 1.24677710e-01 6.17103040e-01 1.41887414e+00 -3.72590601e-01 5.44245958e-01 2.84847349e-01 -1.19875267e-01 -1.01810062e+00 -7.29603708e-01 -7.82442689e-01 -8.51644337e-01 -1.21187007e+00 9.76639688e-01 -8.96050811e-01 -7.21447647e-01 6.95052087e-01 -5.63971817e-01 -2.62116581e-01 -5.37935257e-01 8.78524959e-01 -7.01025903e-01 3.84163886e-01 -6.18860424e-01 -6.47476614e-01 1.19360738e-01 -5.76232553e-01 8.05923939e-01 5.39336383e-01 -5.66789746e-01 -4.06270415e-01 3.70914757e-01 8.76487315e-01 2.06064130e-03 9.90181148e-01 9.43023260e-05 -6.97539210e-01 -9.81337950e-02 -7.83178270e-01 2.52657920e-01 3.33379090e-01 -1.36241093e-01 1.14736006e-01 -7.18020260e-01 -2.14282840e-01 -4.77208793e-01 -3.57026517e-01 8.62767935e-01 3.45339954e-01 9.79132771e-01 1.33235261e-01 -3.40325944e-02 3.26663017e-01 6.77565217e-01 1.03816956e-01 8.32541585e-01 6.17976665e-01 5.23935676e-01 7.66283691e-01 9.82924581e-01 4.16265309e-01 -8.31278488e-02 9.52342629e-01 2.13964790e-01 2.48102933e-01 -6.09795511e-01 -2.49487117e-01 5.00896096e-01 4.62459981e-01 -8.27603221e-01 3.15459855e-02 -8.63159657e-01 5.73557258e-01 -1.88228154e+00 -1.88736629e+00 -4.50660884e-01 2.31676245e+00 4.67221439e-01 4.15900201e-01 8.67490351e-01 5.92608750e-01 9.04426217e-01 2.36822486e-01 -4.92276847e-01 -3.90730888e-01 -3.25363725e-01 1.45122141e-01 5.06134570e-01 -1.58704951e-01 -1.41511202e+00 6.65476024e-01 6.80455923e+00 1.06677258e+00 -9.88736868e-01 -1.31189674e-01 6.33092374e-02 -6.96340084e-01 5.82777917e-01 -3.44588727e-01 -5.94846547e-01 6.17605805e-01 8.74749362e-01 -2.83395052e-01 6.43710122e-02 5.83124220e-01 3.93026352e-01 -3.12590629e-01 -8.13913584e-01 1.12094998e+00 4.72789854e-01 -1.55139709e+00 -3.14189881e-01 -1.24642767e-01 6.18879497e-01 -2.13160023e-01 -3.23329687e-01 5.90387881e-01 -2.07199246e-01 -9.52361882e-01 1.08837891e+00 8.81686330e-01 6.96983695e-01 -7.91838169e-01 5.75090468e-01 3.18843454e-01 -1.22054791e+00 -2.20201910e-01 6.08169883e-02 -5.13601124e-01 3.52335393e-01 3.96581925e-02 -6.69664517e-02 5.31576335e-01 9.95391011e-01 1.06266606e+00 -7.58043349e-01 1.55675280e+00 -4.88707647e-02 8.84319544e-01 -2.88791973e-02 1.53399915e-01 -1.41643882e-01 -1.99914891e-02 7.25718677e-01 1.43129611e+00 1.75932020e-01 1.46616653e-01 3.17607224e-01 4.82888333e-02 7.84007236e-02 1.44862115e-01 -5.16640723e-01 -2.18076542e-01 1.42207414e-01 1.06906164e+00 -9.11195815e-01 -4.20705616e-01 -3.17010224e-01 8.37731063e-01 -2.29474440e-01 2.87765153e-02 -8.91032338e-01 -6.13444209e-01 6.51034713e-01 5.14955342e-01 1.85055777e-01 -1.30257934e-01 -4.70931493e-02 -1.14875209e+00 1.08029768e-01 -1.14043772e+00 6.44644499e-01 -7.85714209e-01 -1.03258932e+00 2.54835904e-01 3.77455384e-01 -1.56616342e+00 -4.84549314e-01 -5.06995857e-01 -7.50716507e-01 3.94416928e-01 -5.88726163e-01 -8.62984538e-01 -6.43068910e-01 4.59665477e-01 8.46722782e-01 -2.30827913e-01 2.63610482e-01 6.32533431e-01 -6.47261739e-01 4.53410536e-01 -2.61297196e-01 5.74487209e-01 9.03154790e-01 -9.16391313e-01 1.47553459e-01 1.07787156e+00 2.60464221e-01 -1.68905437e-01 1.09008563e+00 -8.98271143e-01 -1.27098811e+00 -8.04954886e-01 5.24826407e-01 -5.74721932e-01 7.23453760e-01 -1.77557081e-01 -7.04463720e-01 4.96604085e-01 -1.48219407e-01 3.83442603e-02 7.85071671e-01 -2.23234758e-01 8.46753921e-03 6.60322011e-02 -6.17500782e-01 4.97970194e-01 1.28553343e+00 -4.39907402e-01 -9.15026963e-01 1.17986634e-01 -1.39467239e-01 -8.44784796e-01 -9.50634360e-01 4.07691926e-01 1.10779941e+00 -1.00014317e+00 1.02204943e+00 -9.19555724e-01 5.74778736e-01 -2.49917075e-01 2.12767851e-02 -8.30674529e-01 9.81948450e-02 -5.63806593e-01 -2.29224578e-01 9.99546230e-01 -1.24595188e-01 2.29850873e-01 1.26754940e+00 3.13232005e-01 -3.07085335e-01 -2.09383234e-01 -9.89726365e-01 -1.13267493e+00 -2.68716782e-01 -7.24595964e-01 -3.52414250e-02 6.92844391e-01 2.92562041e-03 -1.75221354e-01 -8.12835455e-01 -2.95901328e-01 7.28619933e-01 1.43431276e-01 1.25077605e+00 -1.21949029e+00 -1.94414020e-01 -6.27116024e-01 -1.23496175e+00 -7.18399525e-01 -3.14000815e-01 -5.66476762e-01 -3.42648402e-02 -1.16590929e+00 3.04700226e-01 5.24804831e-01 -1.21889926e-01 2.55375445e-01 -3.14948410e-02 1.00980771e+00 5.46539664e-01 2.99093872e-01 -6.03782296e-01 -1.05219655e-01 1.21648502e+00 -1.09765837e-02 -4.42587845e-02 3.44945461e-01 -7.97733217e-02 7.74277091e-01 6.03971720e-01 -3.69765788e-01 -7.34126121e-02 1.72229767e-01 1.82573311e-02 1.37829900e-01 7.07437754e-01 -1.53489912e+00 1.46108896e-01 -3.44771504e-01 3.67775410e-01 -7.88635433e-01 4.38033015e-01 -1.22817703e-01 5.05811870e-01 6.30015492e-01 -4.52734023e-01 -4.39137928e-02 2.57245302e-01 4.89773452e-01 -6.26657486e-01 -2.26279795e-01 5.05275786e-01 1.44123128e-02 -1.23563266e+00 1.62699297e-01 -7.13566899e-01 5.14095485e-01 1.35856402e+00 -9.35368657e-01 -2.66433269e-01 -3.05769056e-01 -1.23325229e+00 4.16960008e-02 3.73488247e-01 7.28767872e-01 4.45914924e-01 -1.45358813e+00 -1.04488373e+00 -1.05584972e-01 3.26440901e-01 -6.51895583e-01 7.22103953e-01 8.17931890e-01 -9.07315314e-01 -9.57958475e-02 -8.42646658e-01 -5.43436408e-01 -1.72176015e+00 4.24113393e-01 4.14376855e-02 -1.26150757e-01 -1.04322970e+00 5.60942590e-01 -3.49956185e-01 2.70866364e-01 3.73156846e-01 -5.87845743e-02 -6.56355381e-01 4.66884941e-01 1.02319407e+00 9.52456594e-01 -3.87114361e-02 -8.61055553e-01 -2.88039833e-01 5.55795491e-01 2.92019874e-01 1.26761366e-02 1.16010654e+00 2.24682376e-01 7.13714361e-01 7.89983511e-01 7.44638681e-01 1.95198864e-01 -1.57613122e+00 1.19603677e-02 7.84419477e-02 -7.64223456e-01 -1.88024879e-01 -4.81183618e-01 -1.01218295e+00 9.57451880e-01 7.68753231e-01 2.50112146e-01 9.78448927e-01 -2.66364783e-01 6.69188797e-01 -5.24424165e-02 2.00959161e-01 -1.44451177e+00 2.13022813e-01 2.62374371e-01 9.43409801e-01 -9.55808401e-01 1.28000244e-01 -2.35719472e-01 -9.03452694e-01 1.18577266e+00 4.80837435e-01 -3.29582125e-01 3.00726920e-01 2.81674922e-01 2.29045879e-02 -1.15274116e-01 -3.32597554e-01 -2.32242376e-01 6.06164455e-01 6.73961580e-01 2.44783938e-01 -3.56862321e-02 -5.07195830e-01 8.12471449e-01 -4.38618451e-01 3.48731875e-01 8.13163817e-01 1.22200990e+00 -5.03994465e-01 -7.86682129e-01 -5.42496204e-01 4.76568997e-01 -6.23251140e-01 4.14221942e-01 -5.39780557e-01 9.99772251e-01 2.44887099e-01 7.74198353e-01 5.22310520e-03 -6.04978561e-01 1.02340448e+00 -6.66210130e-02 6.76117718e-01 -4.45534796e-01 -9.48993683e-01 -2.85236031e-01 4.89984930e-01 -9.36352253e-01 -5.09585857e-01 -1.08718562e+00 -8.58659744e-01 -4.47570682e-01 1.38507292e-01 1.25925481e-01 4.68117714e-01 6.19983792e-01 4.22305912e-02 6.17842913e-01 1.47097901e-01 -1.32950211e+00 -1.75099626e-01 -8.86497319e-01 -8.59740138e-01 7.84633696e-01 -7.08040372e-02 -9.67095256e-01 -3.98000292e-02 5.62916934e-01]
[7.767632007598877, 0.16224876046180725]
43608be2-3306-4a0e-8730-fc5ea97e7bc0
chatvideo-a-tracklet-centric-multimodal-and
2304.14407
null
https://arxiv.org/abs/2304.14407v2
https://arxiv.org/pdf/2304.14407v2.pdf
ChatVideo: A Tracklet-centric Multimodal and Versatile Video Understanding System
Existing deep video models are limited by specific tasks, fixed input-output spaces, and poor generalization capabilities, making it difficult to deploy them in real-world scenarios. In this paper, we present our vision for multimodal and versatile video understanding and propose a prototype system, \system. Our system is built upon a tracklet-centric paradigm, which treats tracklets as the basic video unit and employs various Video Foundation Models (ViFMs) to annotate their properties e.g., appearance, motion, \etc. All the detected tracklets are stored in a database and interact with the user through a database manager. We have conducted extensive case studies on different types of in-the-wild videos, which demonstrates the effectiveness of our method in answering various video-related problems. Our project is available at https://www.wangjunke.info/ChatVideo/
['Yu-Gang Jiang', 'Zuxuan Wu', 'Lu Yuan', 'Xiyang Dai', 'Chong Luo', 'Dongdong Chen', 'Junke Wang']
2023-04-27
null
null
null
null
['video-understanding']
['computer-vision']
[-2.48034313e-01 -4.90404427e-01 -3.04512322e-01 -1.41321316e-01 -2.95611531e-01 -8.09928954e-01 4.81535435e-01 -3.81132215e-01 -1.65207401e-01 3.91495734e-01 1.91587120e-01 -1.24815702e-01 2.04220593e-01 -4.45333481e-01 -7.16401160e-01 -4.20124948e-01 7.29191750e-02 -9.29317810e-03 5.81640780e-01 -2.53866732e-01 1.51455551e-01 5.41266277e-02 -1.83572102e+00 7.44944274e-01 3.59010905e-01 1.01318598e+00 5.68995118e-01 9.78870511e-01 -1.51121378e-01 1.28603315e+00 -3.94077182e-01 -3.26310813e-01 1.29187956e-01 -2.43123010e-01 -5.95932841e-01 2.34295264e-01 4.59043801e-01 -6.54677510e-01 -9.70805228e-01 9.41744208e-01 4.59800929e-01 2.01805994e-01 3.46724927e-01 -1.80299628e+00 -5.23728907e-01 3.58255893e-01 -3.30708176e-01 5.57506502e-01 7.98248291e-01 2.92480469e-01 8.22694719e-01 -8.84310544e-01 8.21465433e-01 1.03877175e+00 6.66664481e-01 6.31719470e-01 -6.86823487e-01 -6.06749594e-01 3.02499831e-01 7.60566592e-01 -1.41995728e+00 -7.88704634e-01 5.44736564e-01 -5.93290985e-01 8.92208934e-01 4.20045555e-01 8.60450506e-01 1.39629030e+00 -5.02418280e-02 1.25928032e+00 3.52014840e-01 -1.75979063e-01 -7.19869211e-02 2.03085363e-01 6.07631728e-02 8.51573825e-01 -2.27219895e-01 -2.36815482e-01 -8.80689561e-01 2.11836174e-01 1.05224192e+00 1.07382737e-01 -6.17972493e-01 -4.14638460e-01 -1.39305413e+00 5.49387455e-01 5.38233481e-02 2.43144631e-01 -2.34640256e-01 2.54885346e-01 6.23234034e-01 5.22984564e-01 1.90257967e-01 -2.43464440e-01 -4.55054432e-01 -4.51396257e-01 -7.61939466e-01 1.55602962e-01 7.63665915e-01 1.34439909e+00 4.04649079e-01 -1.51460811e-01 -2.29597688e-01 7.31105387e-01 3.35436225e-01 2.12892517e-01 3.52731317e-01 -1.31918979e+00 5.59366107e-01 3.75917464e-01 1.08068392e-01 -1.07839799e+00 -2.25032553e-01 2.09430709e-01 -5.21117389e-01 -2.24020168e-01 2.15476274e-01 -7.97874704e-02 -8.45982015e-01 1.61650836e+00 3.08029562e-01 4.37141210e-01 -9.51387286e-02 9.46531713e-01 1.41340315e+00 8.29451680e-01 1.92755893e-01 -1.84779167e-01 1.34229350e+00 -1.15465772e+00 -8.68032038e-01 1.31955519e-01 5.80836952e-01 -7.91222572e-01 1.13117719e+00 4.65676844e-01 -1.14683843e+00 -8.52313161e-01 -7.48404562e-01 -1.61335155e-01 -3.52868468e-01 2.39999756e-01 4.71900314e-01 3.50590229e-01 -1.32502592e+00 2.61173993e-01 -7.38578439e-01 -7.82734036e-01 1.81037962e-01 1.57169446e-01 -4.82080907e-01 -6.96193501e-02 -1.00865448e+00 3.44729394e-01 5.35796642e-01 4.77655008e-02 -1.09519804e+00 -4.24958766e-01 -8.71805787e-01 -1.04406029e-01 6.53516650e-01 -8.58219147e-01 1.43335974e+00 -1.01976693e+00 -1.23764789e+00 8.15028012e-01 -2.17377767e-01 -2.86072516e-03 5.23104787e-01 -4.17170703e-01 -6.08324409e-01 4.75079358e-01 -1.10840268e-01 8.57815385e-01 8.29761147e-01 -1.15598154e+00 -7.57910371e-01 -2.57122144e-03 6.14537835e-01 3.25039983e-01 -5.29496789e-01 2.88441718e-01 -1.33607209e+00 -7.27787912e-01 -2.07201332e-01 -7.74249494e-01 2.07313627e-01 1.94475949e-01 -3.03706795e-01 -2.15004072e-01 1.21881270e+00 -6.93416178e-01 1.53235757e+00 -2.40598869e+00 2.63964862e-01 -1.96090043e-01 3.13833892e-01 3.24675411e-01 -1.77463785e-01 6.99737489e-01 5.38459681e-02 4.69192490e-02 3.09448957e-01 -3.50493342e-01 -9.60929394e-02 2.17955336e-01 -2.77338903e-02 1.40816182e-01 -1.15822420e-01 7.81996608e-01 -8.68644059e-01 -7.17112243e-01 3.65810513e-01 3.70688289e-01 -4.15084928e-01 4.17532891e-01 -3.91015738e-01 3.35802495e-01 -3.56814563e-01 9.22060788e-01 3.49737078e-01 -3.22421581e-01 2.04876691e-01 -5.18875122e-01 -9.05448124e-02 -3.51674885e-01 -1.26173878e+00 1.85380161e+00 -9.17929560e-02 1.01052022e+00 1.85927287e-01 -9.56839859e-01 3.65990609e-01 6.94994390e-01 5.33631444e-01 -5.89403927e-01 1.80045724e-01 -2.92780280e-01 -3.54695827e-01 -1.38988793e+00 6.73511803e-01 5.88192999e-01 2.01161057e-01 3.37921172e-01 1.15494378e-01 4.63840842e-01 5.43419838e-01 4.84318107e-01 9.84448493e-01 4.11218554e-01 5.11528105e-02 1.31494090e-01 4.40726221e-01 1.10124260e-01 4.26040888e-01 7.00793743e-01 -3.11883628e-01 7.06915796e-01 3.06016475e-01 -5.28627276e-01 -9.81377542e-01 -1.06858492e+00 3.36665660e-01 1.36080587e+00 4.79842007e-01 -8.76788378e-01 -6.52695060e-01 -5.83696544e-01 -2.37254247e-01 2.55177796e-01 -5.07794261e-01 1.27442777e-01 -3.87689710e-01 -1.52427614e-01 4.71977651e-01 7.06929564e-01 5.14647901e-01 -1.21361971e+00 -5.52285194e-01 7.71175846e-02 -6.24783993e-01 -1.38910520e+00 -6.19736791e-01 -5.32178044e-01 -6.42588317e-01 -1.34870529e+00 -5.52375555e-01 -8.24726701e-01 4.54047680e-01 7.65467107e-01 1.08225787e+00 3.04642409e-01 -2.45257452e-01 7.68704772e-01 -6.84876561e-01 -2.64386058e-01 -1.13979541e-01 -3.16997200e-01 8.68264735e-02 8.90249908e-02 3.48695815e-01 -1.77737877e-01 -6.47730350e-01 5.36894500e-01 -9.98798847e-01 2.54612148e-01 1.44384906e-01 4.13002491e-01 3.45621079e-01 9.00943279e-02 1.82724044e-01 -6.38142526e-01 5.16267240e-01 -6.78225040e-01 -2.66866267e-01 4.14500296e-01 2.49285623e-01 -6.02409482e-01 3.01702887e-01 -7.97112584e-01 -9.31383669e-01 -1.48992508e-03 4.02900316e-02 -8.75603080e-01 -4.09369379e-01 6.32498443e-01 -3.05058032e-01 7.72868171e-02 2.46220663e-01 3.59763563e-01 3.26265171e-02 -4.06340301e-01 2.94141322e-01 7.39983499e-01 6.13765299e-01 -3.22173327e-01 5.04777193e-01 4.58140165e-01 -5.71083963e-01 -1.03613925e+00 -5.96130967e-01 -6.50171459e-01 -5.57983160e-01 -8.73491466e-01 9.53073084e-01 -1.25017262e+00 -9.13812459e-01 5.38481593e-01 -9.62396920e-01 -3.56217951e-01 1.99706212e-01 5.42812288e-01 -6.58169150e-01 5.33366680e-01 -7.94383526e-01 -6.33910358e-01 -1.07511170e-01 -1.14275813e+00 9.19344246e-01 3.69019151e-01 -9.39672068e-02 -9.97458458e-01 -7.23274052e-02 5.77027142e-01 2.68752217e-01 1.23821102e-01 4.81459230e-01 -4.08934861e-01 -8.19608212e-01 -1.19185232e-01 -3.18246126e-01 1.62868842e-01 -2.61982501e-01 4.07983422e-01 -9.73022044e-01 -3.32224995e-01 -1.83988944e-01 -4.26708579e-01 7.53367662e-01 4.45746064e-01 1.38164055e+00 -4.32378352e-02 -4.17623371e-01 6.88420415e-01 1.34323967e+00 4.29655552e-01 7.08510399e-01 3.11539561e-01 8.67561102e-01 3.38879108e-01 7.69661903e-01 5.41468918e-01 6.80229545e-01 8.22922707e-01 3.79187822e-01 6.42974004e-02 -1.64799079e-01 -1.49068311e-01 7.67141700e-01 1.03610981e+00 -4.09897566e-01 -7.34237731e-01 -6.97550535e-01 5.62861145e-01 -2.37390327e+00 -1.34883094e+00 -1.99833155e-01 1.68417954e+00 3.45001340e-01 -1.03572316e-01 3.87618691e-01 -2.18327537e-01 8.12521338e-01 5.63975014e-02 -3.00137311e-01 1.34487497e-02 3.21414396e-02 -3.99787277e-01 -1.95706089e-04 2.33397931e-02 -1.26589012e+00 1.03945768e+00 6.57208776e+00 7.25102901e-01 -1.03959394e+00 2.40891427e-01 2.29779407e-01 -2.02926010e-01 5.09959273e-02 -2.44358867e-01 -5.45318365e-01 5.16464591e-01 7.05599785e-01 -1.21370926e-01 2.91063815e-01 5.48175633e-01 4.00156260e-01 -9.07733664e-02 -1.25788152e+00 1.27939343e+00 1.96054697e-01 -1.47827554e+00 1.84684694e-01 -5.07854447e-02 3.03326339e-01 5.27203083e-02 -9.58074704e-02 2.85926402e-01 -8.00071135e-02 -8.35661352e-01 1.00475109e+00 7.34694064e-01 8.78382206e-01 -2.96420902e-01 5.77730536e-01 3.52571696e-01 -1.50219452e+00 -1.89772546e-01 -2.17830345e-01 -1.49134025e-02 3.31680536e-01 -1.56295910e-01 -4.78409141e-01 6.56976700e-01 1.10483253e+00 1.09669280e+00 -4.02260542e-01 1.26714897e+00 9.44082960e-02 4.70973819e-01 -3.07590634e-01 9.55940485e-02 -1.27290096e-02 -7.43441060e-02 4.40629780e-01 1.30126274e+00 4.36470002e-01 3.04770738e-01 4.64328200e-01 3.58452529e-01 1.56548955e-02 -4.00320180e-02 -7.23912954e-01 -8.48980471e-02 4.83262092e-01 1.17366660e+00 -7.29190707e-01 -5.12029588e-01 -8.68339777e-01 8.91501904e-01 4.97833230e-02 5.72300375e-01 -1.13822293e+00 -1.36326402e-01 6.29009187e-01 9.02882777e-03 5.30889034e-01 -3.34796757e-01 4.63918537e-01 -1.62114942e+00 1.23405918e-01 -1.01764035e+00 6.38467371e-01 -1.13309014e+00 -9.67797399e-01 5.99231720e-01 6.91112727e-02 -1.45417774e+00 -3.84824425e-02 -5.71714342e-01 -7.20056176e-01 1.56215265e-01 -1.02881706e+00 -1.10431552e+00 -6.86301827e-01 1.17391264e+00 9.63258088e-01 -3.04947019e-01 6.64186060e-01 6.70710087e-01 -9.25944209e-01 4.48460400e-01 -9.65971798e-02 3.13672304e-01 6.27324164e-01 -7.95247138e-01 2.47686148e-01 7.75049627e-01 3.80867183e-01 3.66143256e-01 6.84964955e-01 -4.59530830e-01 -1.73322213e+00 -8.88261318e-01 4.40740019e-01 -6.00109756e-01 7.21719503e-01 -4.73060638e-01 -7.61920512e-01 7.98833251e-01 5.17743170e-01 -8.05186033e-02 7.92030871e-01 -2.88715899e-01 -2.15229094e-01 3.85791156e-03 -7.76291966e-01 5.96811175e-01 1.48548472e+00 -6.39589190e-01 -3.64703268e-01 4.43257272e-01 6.49976552e-01 -3.70455831e-01 -8.26471329e-01 2.76917756e-01 7.83788502e-01 -9.98643935e-01 1.02607143e+00 -8.14341187e-01 3.34224254e-01 -5.31386375e-01 -3.67301434e-01 -8.49218369e-01 -4.06039029e-01 -4.99653131e-01 -4.80590701e-01 1.29946399e+00 6.28263578e-02 -7.64307231e-02 7.87759542e-01 4.95337725e-01 -1.59087524e-01 -4.86075640e-01 -5.35485148e-01 -5.61145663e-01 -7.97571182e-01 -6.66943848e-01 2.10736305e-01 1.03155077e+00 6.88427081e-03 1.65770173e-01 -6.10337555e-01 3.14954519e-01 4.21046466e-01 -6.27916539e-03 1.00307894e+00 -8.90047550e-01 -3.00144225e-01 -2.82150984e-01 -5.64807594e-01 -1.06143022e+00 -3.36315334e-02 -4.79230434e-01 -1.96772546e-01 -1.52736259e+00 3.59789640e-01 -7.88092539e-02 -2.44494990e-01 4.35173541e-01 3.14365402e-02 3.95159483e-01 5.69068551e-01 4.30939913e-01 -1.12309313e+00 3.45183134e-01 1.10220838e+00 -1.12271756e-01 -9.53737125e-02 -7.55492076e-02 -3.50685954e-01 8.51904869e-01 7.33555853e-01 5.02183009e-03 -5.99922657e-01 -8.22000086e-01 4.49537523e-02 2.60602981e-01 3.86127025e-01 -1.07196200e+00 4.69700098e-01 -1.43382326e-01 4.50496107e-01 -6.56584024e-01 6.80368662e-01 -7.89295673e-01 5.56260824e-01 1.79356143e-01 -1.89532578e-01 3.62383008e-01 2.18213618e-01 4.95779634e-01 -3.94494534e-01 8.68215133e-03 2.17812255e-01 -2.30508387e-01 -1.58686364e+00 5.71599722e-01 -4.49486166e-01 -7.00518489e-02 1.35183036e+00 -4.37471688e-01 -5.26768744e-01 -6.77259088e-01 -1.00702679e+00 4.66350228e-01 5.76885462e-01 8.06065738e-01 8.78886044e-01 -1.29628980e+00 -5.38818717e-01 6.70861602e-02 4.79766130e-01 -5.64414501e-01 5.45798719e-01 7.59697199e-01 -7.79478312e-01 2.04236791e-01 -3.69561553e-01 -8.46076071e-01 -1.81439829e+00 8.22625101e-01 1.16892472e-01 4.44109470e-01 -7.27913260e-01 8.43379080e-01 3.30670655e-01 6.94869012e-02 7.22114503e-01 -3.24942805e-02 -4.89752471e-01 7.15325475e-02 7.29349375e-01 1.58632338e-01 -3.76534045e-01 -7.73266852e-01 -3.37458313e-01 6.51986480e-01 -9.33563262e-02 6.94521889e-02 1.13713300e+00 -4.74098176e-01 1.25946343e-01 5.30651510e-01 1.06179237e+00 -3.77538115e-01 -1.20903897e+00 -2.85625428e-01 -2.94449329e-01 -6.19967461e-01 -6.15266077e-02 -3.99031997e-01 -1.27639842e+00 7.40310907e-01 5.48564255e-01 4.19848621e-01 1.18508148e+00 1.93541422e-01 7.77047932e-01 4.09319341e-01 3.82145673e-01 -1.07094061e+00 1.30072623e-01 4.17718023e-01 6.96232855e-01 -1.21014619e+00 -2.89186150e-01 -4.44450349e-01 -8.59178841e-01 1.18273902e+00 8.41920555e-01 3.10452849e-01 6.71739340e-01 1.61041245e-01 4.50953454e-01 -3.25902164e-01 -1.30564344e+00 -2.89968729e-01 1.26068681e-01 5.71353734e-01 4.62674379e-01 -1.37147054e-01 1.17275372e-01 2.24662408e-01 1.08879954e-01 2.71936625e-01 5.67071140e-01 1.06553626e+00 -3.78430039e-01 -8.85594666e-01 -3.03764552e-01 2.44591787e-01 -3.13467443e-01 1.65999487e-01 -2.76666462e-01 8.51460099e-01 1.60042867e-01 1.07368433e+00 1.13114543e-01 -7.24640250e-01 3.07748020e-01 -4.73283492e-02 5.01946449e-01 -4.60368276e-01 -3.59768808e-01 1.92928359e-01 2.38081038e-01 -9.44162130e-01 -7.88276613e-01 -7.30313301e-01 -9.13484752e-01 -5.36756575e-01 -6.53352439e-02 -5.65988645e-02 4.19483066e-01 8.00578058e-01 5.00143945e-01 4.12243366e-01 3.00170451e-01 -7.85318851e-01 9.73164439e-02 -7.72118151e-01 -2.66632259e-01 6.31776273e-01 3.85592043e-01 -6.91751659e-01 1.04264855e-01 5.97555876e-01]
[9.942973136901855, 0.757461667060852]
15591718-2a1b-4492-93fa-404387798eeb
pacanet-a-study-on-cyclegan-with-transfer
2301.13082
null
https://arxiv.org/abs/2301.13082v5
https://arxiv.org/pdf/2301.13082v5.pdf
PaCaNet: A Study on CycleGAN with Transfer Learning for Diversifying Fused Chinese Painting and Calligraphy
AI-Generated Content (AIGC) has recently gained a surge in popularity, powered by its high efficiency and consistency in production, and its capability of being customized and diversified. The cross-modality nature of the representation learning mechanism in most AIGC technology allows for more freedom and flexibility in exploring new types of art that would be impossible in the past. Inspired by the pictogram subset of Chinese characters, we proposed PaCaNet, a CycleGAN-based pipeline for producing novel artworks that fuse two different art types, traditional Chinese painting and calligraphy. In an effort to produce stable and diversified output, we adopted three main technical innovations: 1. Using one-shot learning to increase the creativity of pre-trained models and diversify the content of the fused images. 2. Controlling the preference over generated Chinese calligraphy by freezing randomly sampled parameters in pre-trained models. 3. Using a regularization method to encourage the models to produce images similar to Chinese paintings. Furthermore, we conducted a systematic study to explore the performance of PaCaNet in diversifying fused Chinese painting and calligraphy, which showed satisfying results. In conclusion, we provide a new direction of creating arts by fusing the visual information in paintings and the stroke features in Chinese calligraphy. Our approach creates a unique aesthetic experience rooted in the origination of Chinese hieroglyph characters. It is also a unique opportunity to delve deeper into traditional artwork and, in doing so, to create a meaningful impact on preserving and revitalizing traditional heritage.
['Yingfang Yuan', 'Yisheng Yuan', 'Yue Wang', 'Wei Pang', 'Yang Xu', 'Zhang Luo', 'Huajun Bai', 'Zuhao Yang']
2023-01-30
null
null
null
null
['one-shot-learning']
['methodology']
[ 3.81185025e-01 -5.12785651e-02 1.03892468e-01 -8.53614509e-02 -2.95981258e-01 -7.05520630e-01 7.90553331e-01 -6.47919774e-01 8.00213218e-02 8.50930870e-01 6.31966054e-01 1.84747070e-01 -3.86720560e-02 -1.02065110e+00 -5.97580969e-01 -8.67778480e-01 2.77985632e-01 1.77756950e-01 -2.78557509e-01 -5.09903729e-01 4.59385097e-01 8.16531181e-01 -1.51850975e+00 5.56409180e-01 1.02593970e+00 6.40165865e-01 5.39613366e-01 5.07977784e-01 -5.00447810e-01 6.43021405e-01 -5.56226730e-01 -8.00515771e-01 3.09946805e-01 -1.00096405e+00 -4.41904068e-01 2.14306504e-01 3.43165129e-01 -1.53206274e-01 -2.23135799e-01 5.70155203e-01 5.94450057e-01 1.28802639e-02 5.70075452e-01 -7.48841524e-01 -1.30107033e+00 8.56724858e-01 -6.07630789e-01 -4.09034789e-01 2.24640995e-01 5.35495043e-01 8.17509413e-01 -6.04570448e-01 9.44466770e-01 1.30642998e+00 3.83079708e-01 8.55259955e-01 -1.38501191e+00 -8.74426126e-01 1.43264327e-02 -5.56277744e-02 -9.86784160e-01 -1.68164074e-01 1.10428321e+00 -3.40159923e-01 4.13907319e-01 3.27914029e-01 1.44065523e+00 1.33781552e+00 2.25111336e-01 8.25514913e-01 1.28778303e+00 -6.28970861e-01 2.27603480e-01 1.46360025e-01 -9.28180337e-01 4.63925481e-01 -3.31444405e-02 1.50542006e-01 -5.13237059e-01 2.62927026e-01 1.17745125e+00 1.95407514e-02 -2.53823012e-01 -1.58753052e-01 -9.96415913e-01 7.78906763e-01 5.69054365e-01 7.07772434e-01 -2.39143655e-01 4.57915694e-01 -3.33015993e-02 -1.29829999e-02 2.17731446e-01 1.13085222e+00 -1.46077767e-01 -2.36732721e-01 -1.05376458e+00 -1.57331955e-02 4.87070709e-01 8.12068641e-01 6.07318580e-01 6.84026301e-01 -9.19694975e-02 8.40875864e-01 -6.95158616e-02 5.08898318e-01 4.92246985e-01 -1.19696116e+00 -5.10098189e-02 6.41643345e-01 -3.82843405e-01 -9.12666678e-01 9.65603888e-02 -4.13632989e-01 -7.16412842e-01 7.16222167e-01 5.09967320e-02 -2.11018428e-01 -1.17945659e+00 1.61062813e+00 -1.75681114e-01 -2.67592579e-01 -1.02777563e-01 7.77513683e-01 5.09271085e-01 8.78156841e-01 2.16172844e-01 3.11067015e-01 1.02420497e+00 -8.47697735e-01 -6.08941615e-01 1.51401646e-02 2.27924302e-01 -1.01753294e+00 1.52017617e+00 3.62946838e-01 -1.20777941e+00 -5.79554260e-01 -1.08616400e+00 8.87723342e-02 -4.36587483e-01 -2.15630494e-02 9.13515866e-01 7.50010192e-01 -9.10622656e-01 7.73705661e-01 -4.52572674e-01 -3.00253332e-01 8.28876436e-01 3.93330753e-02 -1.63210407e-01 2.21430082e-02 -7.94469416e-01 8.56768131e-01 4.83360052e-01 4.82036285e-02 -8.12483609e-01 -8.43538105e-01 -4.60092038e-01 1.52441382e-01 2.72909582e-01 -6.72217011e-01 6.46842897e-01 -1.56137407e+00 -2.03086424e+00 6.09522700e-01 4.23852295e-01 6.78262860e-02 5.38862765e-01 -1.71371579e-01 -5.54088354e-01 2.11377561e-01 -2.14432269e-01 1.07626998e+00 9.06466484e-01 -1.59314191e+00 -4.36465472e-01 -6.27383366e-02 -1.61822394e-01 7.12120384e-02 -5.14375210e-01 -4.37000215e-01 -5.74893475e-01 -1.12345815e+00 -2.15370104e-01 -9.14452910e-01 -1.69002056e-01 1.32444829e-01 -1.98980972e-01 4.59991634e-01 9.61438656e-01 -7.77639151e-01 9.12103415e-01 -2.30604553e+00 4.44371879e-01 2.02448502e-01 -2.71446593e-02 3.19048703e-01 -3.03611755e-01 7.73283958e-01 2.64054835e-01 5.26847959e-01 -4.27653432e-01 -2.02243835e-01 -1.64196566e-01 3.21106642e-01 -4.31004375e-01 -2.15256751e-01 3.27993035e-01 1.11044931e+00 -9.45322454e-01 -5.37333667e-01 2.78926104e-01 8.10143888e-01 -4.96174067e-01 -1.33226246e-01 -5.24012506e-01 9.27406728e-01 -4.00211841e-01 8.11072588e-01 5.89301705e-01 -1.51695367e-02 1.81835845e-01 -4.26393412e-02 -3.30090612e-01 -3.32067698e-01 -8.36719632e-01 1.89957201e+00 -7.00311363e-01 8.86680186e-01 -2.17214018e-01 -1.87325314e-01 1.13723886e+00 7.48897642e-02 3.38827223e-01 -1.04080391e+00 1.21222295e-01 2.14564532e-01 -2.98212826e-01 -5.60767591e-01 6.61654770e-01 -4.77066427e-01 -1.71663500e-02 4.49792743e-01 -1.54681876e-01 -8.77823472e-01 1.27064854e-01 1.37044996e-01 4.95740443e-01 7.51154900e-01 -2.36781806e-01 -1.23974733e-01 2.70556360e-01 2.47982773e-03 3.60353380e-01 3.37612122e-01 5.20395696e-01 9.14818645e-01 3.27256829e-01 -4.55623984e-01 -1.29033160e+00 -1.29616845e+00 1.61815688e-01 8.15115869e-01 -1.62860081e-01 -1.48324236e-01 -5.89733481e-01 -1.96098417e-01 -2.11797178e-01 1.04200733e+00 -7.33315408e-01 -2.62983322e-01 -7.33046710e-01 -2.98378468e-01 6.09417498e-01 4.00309294e-01 8.83206367e-01 -1.37830126e+00 -6.05527461e-01 1.59121081e-01 4.31843512e-02 -6.34914517e-01 -4.29927886e-01 -2.37380892e-01 -8.16654027e-01 -6.02621853e-01 -1.06160545e+00 -9.65709388e-01 5.93266964e-01 1.29925050e-02 9.06607866e-01 -9.97861922e-02 -4.76594746e-01 3.63486439e-01 -4.84181345e-01 -4.61930871e-01 -5.15541971e-01 -5.19191623e-02 -5.26562333e-01 1.11758433e-01 -2.85574228e-01 -7.55694687e-01 -5.63552022e-01 -8.72381702e-02 -1.32141447e+00 4.26902145e-01 8.82684469e-01 5.58406413e-01 4.36057985e-01 1.43714622e-01 4.78761107e-01 -7.89223969e-01 4.98494804e-01 2.78867558e-02 -4.48883921e-01 3.49475741e-01 -4.76093262e-01 1.04853205e-01 6.41770959e-01 -7.64296174e-01 -1.79051733e+00 -5.87486699e-02 1.88776493e-01 -1.58284128e-01 3.09145331e-01 2.86197633e-01 -1.31820142e-01 -2.18231827e-01 5.73297441e-01 1.25367478e-01 2.81120767e-03 -4.94369924e-01 8.54726076e-01 4.43454772e-01 6.44844949e-01 -6.19270742e-01 7.67399430e-01 7.12227583e-01 -1.48399994e-01 -9.86097813e-01 -2.16295615e-01 3.64020795e-01 -5.15251756e-01 -4.71144199e-01 1.07271552e+00 -7.05775261e-01 -5.23537815e-01 4.41334575e-01 -6.92325413e-01 -6.32878363e-01 -7.22814202e-01 3.15675616e-01 -4.04622734e-01 1.77288696e-01 -3.87654036e-01 -6.30084991e-01 -2.14977190e-01 -8.10360312e-01 8.05362165e-01 6.30794525e-01 -2.39540577e-01 -8.90220046e-01 2.37782672e-01 3.70069116e-01 6.30980015e-01 7.12744057e-01 1.17367327e+00 3.16670507e-01 -7.63203800e-01 1.16029596e-02 -9.36075151e-02 3.94632578e-01 3.48833352e-01 3.86923015e-01 -8.51218760e-01 -7.22895563e-02 -3.24692309e-01 -2.30062574e-01 9.02923882e-01 2.92899348e-02 1.00368547e+00 -2.85268545e-01 -5.75467944e-02 7.86703825e-01 1.63907337e+00 5.57040811e-01 1.10732412e+00 4.90615666e-01 7.25494027e-01 4.76681501e-01 5.36134392e-02 3.99338096e-01 -8.31982344e-02 3.38459522e-01 2.05027580e-01 -3.17682147e-01 -7.02348530e-01 -7.37237096e-01 2.96755731e-01 5.83583355e-01 -6.19854987e-01 -2.20423833e-01 -5.47562718e-01 5.26167095e-01 -1.34206343e+00 -1.00690353e+00 1.78278923e-01 1.93953800e+00 8.76264870e-01 -1.78955227e-01 -6.78907558e-02 -2.27083266e-01 7.13140070e-01 3.14531662e-02 -3.01906466e-01 -6.36888623e-01 -6.59630477e-01 7.20858276e-01 2.57088631e-01 4.90688086e-02 -3.95032257e-01 1.36840820e+00 6.32084131e+00 6.79623961e-01 -1.53282726e+00 -1.97041348e-01 4.72380787e-01 -3.56393576e-01 -8.27381074e-01 2.16377899e-01 -3.33453685e-01 5.61770678e-01 3.14879566e-01 2.02454720e-02 7.48841822e-01 5.32990932e-01 1.08340615e-02 -3.43336053e-02 -5.76631367e-01 7.94027865e-01 2.21717998e-01 -1.87738478e+00 4.91811216e-01 1.15786202e-01 1.34105361e+00 -5.06896973e-01 4.28897411e-01 8.01871344e-02 5.09106159e-01 -1.05304623e+00 8.52629960e-01 8.19163382e-01 9.67477858e-01 -7.79147923e-01 1.78500459e-01 -3.04656416e-01 -9.06298161e-01 -3.07830721e-01 -1.80100039e-01 -2.42318716e-02 3.82026762e-01 3.26809227e-01 -5.66481352e-01 4.09901410e-01 6.73791468e-01 5.42095423e-01 -5.97842455e-01 9.38187242e-01 -5.44555902e-01 3.38380694e-01 -5.88720944e-03 -1.68303952e-01 3.72991204e-01 -3.90283108e-01 3.56461763e-01 1.16971099e+00 5.91425955e-01 2.19660122e-02 -3.45108360e-01 1.11745942e+00 -1.34386525e-01 -2.40171812e-02 -4.79977965e-01 -6.33162618e-01 2.52374619e-01 1.20499778e+00 -8.66199911e-01 -1.47855058e-01 -7.57683814e-02 1.07581532e+00 6.09932020e-02 5.14917612e-01 -7.94449985e-01 -3.81174564e-01 1.24899872e-01 1.41194373e-01 6.89873874e-01 -3.48166972e-01 -7.14239359e-01 -9.94559646e-01 -1.75901398e-01 -9.28385973e-01 8.78963023e-02 -1.10929644e+00 -1.18549204e+00 6.06016219e-01 -1.67312995e-01 -1.05998874e+00 1.71100929e-01 -3.01306248e-01 -8.08525324e-01 6.75797462e-01 -1.16574597e+00 -1.76078939e+00 -4.74537313e-01 3.83463264e-01 4.80253696e-01 -2.45723918e-01 8.51899922e-01 4.78355549e-02 -3.97426337e-01 3.77954543e-01 2.70659834e-01 -1.57775670e-01 7.35337734e-01 -7.69195676e-01 7.59606808e-02 6.88988864e-01 2.50843942e-01 3.62199903e-01 5.15549302e-01 -8.75899136e-01 -1.35794055e+00 -7.72658348e-01 3.75755101e-01 -1.00502901e-01 4.07696903e-01 -3.30312520e-01 -5.55699110e-01 6.01267517e-01 4.20491487e-01 -9.60891128e-01 7.50786185e-01 -1.94564953e-01 -2.65163481e-01 -2.04409227e-01 -1.09438479e+00 1.09122431e+00 1.01341391e+00 -2.59982377e-01 -4.95414317e-01 -7.38307536e-02 5.82945347e-01 1.65284872e-02 -7.69234419e-01 8.89875516e-02 1.00411940e+00 -7.87945747e-01 8.60688031e-01 -2.77433097e-01 8.99822891e-01 -1.23124585e-01 -8.58024657e-02 -1.37713027e+00 -6.26014113e-01 -9.46792603e-01 2.70056725e-01 1.48119926e+00 4.50840294e-01 -4.25212979e-01 8.30757082e-01 6.21438265e-01 -3.51206332e-01 -4.02154982e-01 -5.71144342e-01 -6.84521794e-01 2.33948931e-01 -6.40889630e-02 8.79182041e-01 7.99986124e-01 -2.68547088e-01 7.21206069e-02 -6.32312119e-01 -4.81135428e-01 3.87122095e-01 2.64687449e-01 8.41206849e-01 -1.03360677e+00 -3.74522477e-01 -8.55781317e-01 -1.68644324e-01 -2.62271255e-01 -2.34888852e-01 -8.72938216e-01 -1.95693552e-01 -1.52376676e+00 1.44831939e-02 -5.30875981e-01 8.23605284e-02 3.76200706e-01 2.16171980e-01 7.05534875e-01 7.69387186e-01 3.02382022e-01 1.23985864e-01 6.58607543e-01 2.01494384e+00 -2.17432320e-01 -7.45006323e-01 -4.08796728e-01 -1.15954709e+00 4.24375147e-01 9.23795700e-01 -2.14035124e-01 -3.30589145e-01 -7.77341008e-01 4.43203032e-01 -3.92718911e-01 2.04162806e-01 -1.04266584e+00 -1.16556920e-01 -2.15189621e-01 8.37412715e-01 -2.26360843e-01 5.51276922e-01 -7.68922508e-01 7.05413938e-01 5.35663426e-01 -2.27289170e-01 -4.19144720e-01 2.68828839e-01 4.45003003e-01 7.00401962e-02 -1.07403666e-01 9.28734064e-01 -5.39618850e-01 -9.48022783e-01 -1.53493077e-01 -4.07481015e-01 -2.93348312e-01 1.07674468e+00 -6.84068024e-01 -2.05264091e-01 -4.28686857e-01 -6.02026820e-01 -9.89704952e-02 8.73842895e-01 5.55359244e-01 6.41684055e-01 -1.35235846e+00 -5.93976557e-01 3.21874410e-01 -9.50095803e-02 -2.21736014e-01 4.79109168e-01 2.76689261e-01 -9.45013940e-01 -2.12994423e-02 -1.00847971e+00 -2.69272417e-01 -7.25736916e-01 6.01988137e-01 -1.44620061e-01 1.95918581e-03 -8.09863031e-01 6.59805536e-01 1.12626404e-01 -4.95675430e-02 -2.25707386e-02 1.19531401e-01 -1.29392996e-01 1.15077190e-01 4.22477722e-01 4.44704652e-01 -4.30975586e-01 -3.61290693e-01 1.45724416e-01 8.09652150e-01 3.87235284e-02 -3.43772471e-01 1.64955699e+00 1.46903303e-02 -1.98975831e-01 3.37178677e-01 8.31841886e-01 3.55849534e-01 -1.52980030e+00 3.06364745e-01 -3.95371586e-01 -7.65217721e-01 -7.88846686e-02 -1.36067581e+00 -1.34997535e+00 7.46566594e-01 5.64520121e-01 -1.69595778e-01 1.38695514e+00 2.47054696e-02 9.36754882e-01 3.04211732e-02 3.71492893e-01 -1.38517153e+00 6.32847607e-01 8.16709176e-02 1.16895592e+00 -6.32902503e-01 6.19323850e-02 -1.09991394e-01 -1.02758813e+00 1.35831034e+00 4.65608269e-01 -1.00867003e-01 1.28113627e-01 3.66622657e-02 1.11600310e-01 -2.10132197e-01 -1.87242165e-01 -1.45822152e-01 1.17578186e-01 7.46463060e-01 3.51653159e-01 6.98885322e-02 -2.54337013e-01 2.59662241e-01 -5.35916090e-01 1.60344556e-01 5.59128225e-01 1.13243055e+00 -2.07278684e-01 -1.26590919e+00 -4.40185070e-01 1.44392043e-01 -1.46930620e-01 3.85083668e-02 -7.63733864e-01 1.01333332e+00 4.74662453e-01 6.88436270e-01 1.04487618e-03 -3.82265329e-01 5.94441630e-02 -4.27860506e-02 7.78902173e-01 -3.20963800e-01 -6.58689559e-01 2.89172560e-01 -9.42117199e-02 -2.00227782e-01 -3.65864098e-01 -3.91665220e-01 -1.02544999e+00 -4.71566826e-01 -8.74618627e-03 -1.76294908e-01 9.72082734e-01 6.45992756e-01 4.36888218e-01 8.25585902e-01 5.53413630e-01 -1.10466206e+00 2.75501817e-01 -6.69329822e-01 -6.02653384e-01 2.94849932e-01 -2.11409345e-01 -3.44131768e-01 -2.43875086e-01 2.71732032e-01]
[11.763496398925781, -0.4837616980075836]
e5817af7-f89b-4952-a8a1-cd440bd036b2
vision-transformer-with-super-token-sampling
2211.11167
null
https://arxiv.org/abs/2211.11167v1
https://arxiv.org/pdf/2211.11167v1.pdf
Vision Transformer with Super Token Sampling
Vision transformer has achieved impressive performance for many vision tasks. However, it may suffer from high redundancy in capturing local features for shallow layers. Local self-attention or early-stage convolutions are thus utilized, which sacrifice the capacity to capture long-range dependency. A challenge then arises: can we access efficient and effective global context modeling at the early stages of a neural network? To address this issue, we draw inspiration from the design of superpixels, which reduces the number of image primitives in subsequent processing, and introduce super tokens into vision transformer. Super tokens attempt to provide a semantically meaningful tessellation of visual content, thus reducing the token number in self-attention as well as preserving global modeling. Specifically, we propose a simple yet strong super token attention (STA) mechanism with three steps: the first samples super tokens from visual tokens via sparse association learning, the second performs self-attention on super tokens, and the last maps them back to the original token space. STA decomposes vanilla global attention into multiplications of a sparse association map and a low-dimensional attention, leading to high efficiency in capturing global dependencies. Based on STA, we develop a hierarchical vision transformer. Extensive experiments demonstrate its strong performance on various vision tasks. In particular, without any extra training data or label, it achieves 86.4% top-1 accuracy on ImageNet-1K with less than 100M parameters. It also achieves 53.9 box AP and 46.8 mask AP on the COCO detection task, and 51.9 mIOU on the ADE20K semantic segmentation task. Code will be released at https://github.com/hhb072/SViT.
['Tieniu Tan', 'Ran He', 'Jie Cao', 'Xiaoqiang Zhou', 'Huaibo Huang']
2022-11-21
null
http://openaccess.thecvf.com//content/CVPR2023/html/Huang_Vision_Transformer_With_Super_Token_Sampling_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Huang_Vision_Transformer_With_Super_Token_Sampling_CVPR_2023_paper.pdf
cvpr-2023-1
['superpixels']
['computer-vision']
[ 8.02224502e-02 -9.24513564e-02 -1.73570752e-01 -3.92575413e-01 -6.26967609e-01 -2.42155552e-01 4.49675620e-01 -5.68604730e-02 -6.07510388e-01 3.61790121e-01 1.78542435e-02 -1.68729141e-01 3.66144121e-01 -6.67362213e-01 -8.81658137e-01 -7.78017938e-01 3.23468715e-01 1.19801529e-01 4.64542061e-01 1.78856567e-01 7.82985613e-02 8.95953625e-02 -1.29949892e+00 4.03658450e-01 9.17376101e-01 1.33603883e+00 7.64256060e-01 4.59098786e-01 -3.25353175e-01 8.98234427e-01 -5.27523518e-01 -2.97057480e-01 1.35590225e-01 -1.39569372e-01 -8.77407789e-01 4.11175154e-02 7.20851123e-01 -4.12155628e-01 -2.48413876e-01 1.22742748e+00 2.74974734e-01 -1.02011397e-01 4.74079162e-01 -9.79731560e-01 -9.16785121e-01 7.63953984e-01 -9.00383830e-01 2.48627976e-01 -2.81132251e-01 2.76518315e-01 1.14140940e+00 -1.24026072e+00 2.47819230e-01 1.23409522e+00 5.05840957e-01 3.78470600e-01 -1.18933153e+00 -8.59768212e-01 4.24709678e-01 3.02020967e-01 -1.38772213e+00 -4.33023602e-01 4.72661018e-01 -3.11559051e-01 1.07263136e+00 3.96393873e-02 7.75094390e-01 8.62554669e-01 1.49779897e-02 1.07672262e+00 1.03643394e+00 -1.86065495e-01 8.02008212e-02 -2.09000513e-01 3.16719115e-01 9.39045310e-01 2.89832354e-01 -2.39665821e-01 -5.87917447e-01 2.74554640e-01 1.02970588e+00 1.53159261e-01 -9.90765542e-02 3.54217514e-02 -1.10321164e+00 6.43054008e-01 9.90605116e-01 2.94582695e-01 -3.64267290e-01 5.37574768e-01 2.89461255e-01 -8.61251056e-02 2.33340144e-01 1.11025706e-01 -4.48777974e-01 1.81312576e-01 -8.37731600e-01 -1.40030965e-01 3.23936433e-01 9.52010453e-01 1.00889623e+00 1.32767946e-01 -3.68605226e-01 1.02598429e+00 3.47405970e-01 5.27573586e-01 4.69215423e-01 -9.02638376e-01 4.54866081e-01 6.24479592e-01 -2.85946846e-01 -7.15868413e-01 -1.21950500e-01 -6.47488177e-01 -9.49193299e-01 6.36145696e-02 3.81050497e-01 6.75506294e-02 -1.52084470e+00 1.61451983e+00 1.82193547e-01 3.30980062e-01 -1.74352869e-01 9.11072552e-01 8.49135518e-01 6.83934212e-01 3.40461493e-01 1.76350787e-01 1.59694397e+00 -1.36814618e+00 -3.79408598e-01 -7.22100854e-01 3.01512599e-01 -7.67306626e-01 1.05014086e+00 1.48445576e-01 -1.14862680e+00 -7.93740153e-01 -8.47102702e-01 -5.51137209e-01 -1.74422756e-01 3.22279721e-01 7.22526073e-01 1.32819891e-01 -1.13427448e+00 1.92165479e-01 -9.77468014e-01 -7.23846555e-02 8.93721759e-01 3.78648132e-01 -8.80330335e-03 -3.60920995e-01 -9.28686142e-01 5.65249562e-01 3.03489834e-01 3.62149388e-01 -1.06113589e+00 -8.36470127e-01 -9.07206953e-01 2.90303946e-01 4.20690924e-01 -7.99267054e-01 1.18962359e+00 -9.93106186e-01 -1.25943637e+00 7.24258184e-01 -4.22378421e-01 -4.89688069e-01 2.73602068e-01 -3.24099720e-01 5.70930578e-02 1.36280343e-01 5.01727819e-01 1.01744187e+00 1.04099321e+00 -1.04339707e+00 -9.40685868e-01 -2.25862756e-01 6.52351081e-02 1.38922244e-01 -4.28553522e-01 -9.09761935e-02 -1.18849719e+00 -7.59613454e-01 1.77269787e-01 -8.46975744e-01 -4.25309122e-01 -1.36606200e-02 -5.17094374e-01 -2.81385154e-01 7.39601970e-01 -5.95995486e-01 8.93485010e-01 -2.22239566e+00 1.97518412e-02 1.27362147e-01 4.08612400e-01 4.88681078e-01 -2.63785928e-01 -2.37630203e-01 9.27637294e-02 1.41842961e-01 -4.24553812e-01 -6.92348599e-01 -2.03297958e-01 3.18704367e-01 -2.61640370e-01 7.12008700e-02 4.11530167e-01 1.29295897e+00 -8.52621675e-01 -5.59383869e-01 3.08419079e-01 5.81203103e-01 -5.72049856e-01 -8.17042142e-02 -1.60954088e-01 4.59626913e-02 -4.54514265e-01 8.82129610e-01 5.43060064e-01 -4.92095768e-01 3.12410612e-02 -3.58794004e-01 -1.77389935e-01 3.06183070e-01 -8.21821392e-01 1.82909274e+00 -5.45240581e-01 6.70119762e-01 2.12109417e-01 -1.00396407e+00 6.66716576e-01 -8.77094716e-02 2.47284949e-01 -9.91985261e-01 1.42286003e-01 1.18424565e-01 -8.64410922e-02 -2.54957080e-01 4.24230725e-01 1.32693917e-01 4.05315543e-03 1.16754517e-01 1.03321493e-01 1.92190811e-01 1.54705748e-01 2.64861226e-01 8.90970349e-01 -6.62106723e-02 2.14654077e-02 -2.09362239e-01 3.25538427e-01 -9.06406902e-03 7.92885542e-01 7.10750699e-01 -1.41440719e-01 6.30176425e-01 4.39874619e-01 -3.43708098e-01 -8.27752590e-01 -1.04753685e+00 -2.72946004e-02 1.03659630e+00 3.23604643e-01 -5.24411738e-01 -7.55940735e-01 -6.16914392e-01 1.44641725e-02 3.82015586e-01 -5.65741301e-01 -1.62743434e-01 -5.59734821e-01 -8.27897668e-01 4.18890744e-01 9.38763320e-01 9.52172875e-01 -1.17105556e+00 -5.30514836e-01 2.46391207e-01 -2.64671445e-01 -1.36514008e+00 -7.23549128e-01 2.56528139e-01 -7.32695639e-01 -8.77491832e-01 -6.88461721e-01 -1.09920979e+00 8.30585182e-01 6.68897212e-01 9.17754173e-01 5.61754256e-02 -4.55358535e-01 1.23874187e-01 -1.64637610e-01 -2.64853328e-01 1.66313037e-01 1.36546344e-01 -2.52120703e-01 1.19385868e-01 3.34593594e-01 -4.70625192e-01 -8.20656598e-01 2.32084572e-01 -7.32530773e-01 2.19599098e-01 9.50133741e-01 9.82057095e-01 8.97720098e-01 -1.66899413e-01 3.86518717e-01 -7.49233603e-01 1.03403442e-01 -3.14788699e-01 -6.18010402e-01 1.72970176e-01 -4.10711437e-01 -2.56582890e-02 6.52702689e-01 -3.57291460e-01 -1.04769397e+00 1.83174118e-01 -5.33735342e-02 -7.20354915e-01 4.26820964e-02 4.20462132e-01 -1.92751899e-01 -2.73245778e-02 3.61192912e-01 3.81187558e-01 -2.89242148e-01 -4.64782745e-01 4.44324493e-01 4.08776581e-01 6.46247566e-01 -5.68467140e-01 7.15485215e-01 6.56246722e-01 -3.87124270e-01 -9.22035635e-01 -1.08855605e+00 -4.67045248e-01 -4.13691252e-01 1.19705908e-01 1.19452369e+00 -1.14686859e+00 -6.36179328e-01 8.62163663e-01 -9.37840402e-01 -7.22658217e-01 -2.71607757e-01 2.31110588e-01 -1.14042446e-01 2.57379025e-01 -7.37742126e-01 -3.83985162e-01 -5.08059144e-01 -1.23388290e+00 9.38725770e-01 4.93106425e-01 1.29176795e-01 -6.87899470e-01 -5.39034307e-01 6.47101879e-01 3.76456618e-01 -1.55234009e-01 6.59532070e-01 -1.58878654e-01 -1.03171539e+00 1.67387053e-01 -8.16758811e-01 6.04049861e-01 1.71652913e-01 -1.14856817e-01 -1.16840553e+00 -2.40450785e-01 -1.95006579e-01 -4.31826383e-01 1.51385212e+00 6.26299262e-01 1.48777056e+00 -1.56623259e-01 -3.32948267e-01 9.78659570e-01 1.47673285e+00 6.74623474e-02 4.17990118e-01 1.86200917e-01 1.33316505e+00 3.04491669e-01 4.02518541e-01 1.74928874e-01 6.84860528e-01 4.72063988e-01 5.22632003e-01 -3.53041828e-01 -5.16104817e-01 -2.16094375e-01 4.15738791e-01 8.12853813e-01 -4.20989189e-03 1.72692671e-01 -9.76127803e-01 8.62873554e-01 -1.86019516e+00 -7.39314914e-01 -2.26863176e-02 1.90508401e+00 8.12099993e-01 3.19422930e-01 -2.02899814e-01 -2.43559107e-01 5.91982067e-01 2.10266620e-01 -7.57989407e-01 -1.18754633e-01 -6.77987337e-02 3.30975175e-01 7.08468080e-01 5.23740351e-01 -1.15879393e+00 1.40728676e+00 5.12539291e+00 1.05129886e+00 -1.19821036e+00 2.16163173e-01 8.98943186e-01 -1.57807082e-01 -8.15897658e-02 1.14126236e-03 -9.49302852e-01 5.27048588e-01 4.45004940e-01 2.04410538e-01 2.45415598e-01 8.03083599e-01 4.37371433e-03 -1.95630372e-01 -8.18039000e-01 1.03035688e+00 -1.29391000e-01 -1.49078441e+00 1.74813628e-01 6.51423559e-02 8.29848111e-01 4.92027253e-01 1.59382254e-01 2.85574526e-01 3.08420122e-01 -1.06476057e+00 8.60392988e-01 2.23136082e-01 8.58381867e-01 -6.26003265e-01 4.64208901e-01 1.76216856e-01 -1.52245688e+00 -1.60840005e-01 -6.17776096e-01 8.45330283e-02 -1.56940043e-01 7.00644970e-01 -6.40408695e-01 2.54110247e-01 1.04468477e+00 8.55476379e-01 -5.12825787e-01 9.89845157e-01 -5.20387650e-01 6.78271770e-01 -5.16556799e-01 1.82179108e-01 5.66568017e-01 -1.05967782e-01 2.25652084e-01 1.34628856e+00 -4.51722741e-02 1.24971516e-01 2.75190234e-01 9.89165664e-01 -3.48124295e-01 -1.57740921e-01 -1.15262464e-01 2.67502189e-01 5.35126626e-01 1.43180430e+00 -8.21002305e-01 -5.59643924e-01 -5.44726253e-01 1.15477228e+00 5.46122789e-01 5.55377901e-01 -8.24903965e-01 -3.13344508e-01 8.81492913e-01 -1.79075360e-01 7.35130072e-01 -2.83998162e-01 -5.38633049e-01 -1.18715346e+00 1.63315609e-01 -5.90584636e-01 2.85354316e-01 -6.08304679e-01 -1.04083562e+00 5.46769619e-01 -3.17209303e-01 -7.37395525e-01 2.06344232e-01 -5.27951479e-01 -7.41073847e-01 9.30961609e-01 -1.89258027e+00 -1.44209146e+00 -4.85161990e-01 7.29750633e-01 7.86895990e-01 1.10801689e-01 4.13238704e-01 3.74541342e-01 -8.98811758e-01 7.85295665e-01 -1.19471185e-01 5.01531065e-01 6.02044404e-01 -1.28748143e+00 6.60825670e-01 1.01442909e+00 2.18923107e-01 6.19287491e-01 1.22808702e-01 -4.67983067e-01 -1.12935829e+00 -1.50042677e+00 7.76449621e-01 -2.35516399e-01 6.11529827e-01 -3.64718199e-01 -1.00254738e+00 6.55282617e-01 4.98742796e-02 3.88626844e-01 2.43837565e-01 -1.08458204e-02 -6.44684792e-01 -3.18312764e-01 -7.60377944e-01 6.31529212e-01 1.05911255e+00 -6.22688115e-01 -4.50231135e-01 1.80969134e-01 8.03474963e-01 -3.73739749e-01 -6.37653887e-01 1.74385801e-01 4.45858091e-01 -8.37221742e-01 1.14232051e+00 -1.41351074e-01 6.03700757e-01 -4.44073230e-01 -1.35943055e-01 -9.34198260e-01 -5.80076694e-01 -2.90773690e-01 -7.59825259e-02 1.37397969e+00 4.18228716e-01 -5.73026478e-01 7.88127005e-01 4.25001472e-01 -3.41470391e-01 -1.01575875e+00 -7.39064217e-01 -6.86399996e-01 -3.10934484e-02 -4.41548526e-01 3.21010679e-01 9.07980382e-01 -4.45039839e-01 4.39822197e-01 -1.43444836e-01 2.79352367e-01 8.95423532e-01 2.59055018e-01 4.49222028e-01 -7.63607681e-01 -2.51633853e-01 -5.76441109e-01 -2.00537726e-01 -1.41675174e+00 7.84875825e-02 -9.20451641e-01 6.12324439e-02 -1.66826296e+00 4.05166745e-01 -7.07226038e-01 -5.64815283e-01 9.96667862e-01 -4.49373364e-01 5.71676254e-01 3.65200788e-01 3.00618440e-01 -6.22529984e-01 5.89455664e-01 1.18979156e+00 -3.65140527e-01 -1.18603028e-01 -2.95805991e-01 -8.53761196e-01 8.93617570e-01 9.54950035e-01 -3.72627676e-01 -4.25601691e-01 -9.67529476e-01 -1.76889196e-01 -3.91445041e-01 7.08308697e-01 -9.58002508e-01 3.84135097e-01 -6.83374181e-02 5.40718257e-01 -4.66457099e-01 3.36513609e-01 -5.41528940e-01 -4.05886978e-01 5.49995482e-01 -5.19253500e-02 -2.18594000e-01 2.29351252e-01 5.26259422e-01 -3.67893964e-01 1.28481798e-02 9.05789554e-01 -2.39573255e-01 -1.12026668e+00 6.48231566e-01 -1.63279206e-01 1.23147674e-01 8.02995384e-01 -2.33479932e-01 -3.11413497e-01 1.39173657e-01 -5.85722923e-01 6.21311545e-01 4.09551769e-01 4.13080394e-01 6.73787594e-01 -1.17174625e+00 -5.44190764e-01 2.06461638e-01 -3.45242359e-02 6.63289070e-01 4.07982826e-01 7.12367773e-01 -4.24517244e-01 3.65030646e-01 -1.54038608e-01 -8.35709572e-01 -1.13750076e+00 3.24054122e-01 2.63717532e-01 -6.64233342e-02 -8.11210811e-01 1.33685338e+00 8.16345811e-01 1.92605495e-01 1.82591885e-01 -7.13029563e-01 -3.46800797e-02 5.96014149e-02 5.69255888e-01 3.66632976e-02 -1.30073637e-01 -4.59458649e-01 -4.75746125e-01 7.79632509e-01 -4.73126441e-01 1.05333298e-01 1.24073946e+00 -1.29065886e-01 -1.70461357e-01 1.01285368e-01 1.12288046e+00 -2.28599265e-01 -1.68585634e+00 -5.19104362e-01 -1.07679971e-01 -3.25026363e-01 2.90716588e-01 -6.42834425e-01 -1.45420182e+00 1.02550161e+00 3.44355077e-01 -5.13748378e-02 1.25762618e+00 2.19802275e-01 1.05342829e+00 2.48445973e-01 7.89652318e-02 -9.53776777e-01 1.85932189e-01 7.00036705e-01 7.36392081e-01 -1.32052743e+00 -2.25997254e-01 -5.11566162e-01 -6.00064933e-01 7.06193566e-01 9.34027016e-01 -2.59189993e-01 4.56380725e-01 2.34553888e-01 5.37520349e-02 -1.60219908e-01 -7.47852981e-01 -3.38179201e-01 2.12930888e-01 4.86210525e-01 1.42165318e-01 1.38243154e-01 2.23410100e-01 6.32539749e-01 -4.68866853e-03 -2.43598074e-01 2.47047409e-01 5.96595585e-01 -5.12110710e-01 -8.27140093e-01 -1.48059815e-01 4.43569809e-01 -5.01357198e-01 -4.79372144e-01 -2.34333873e-01 5.77619672e-01 3.20818186e-01 7.32109308e-01 1.74517781e-01 -2.16162115e-01 9.85175446e-02 -2.25208670e-01 2.98414201e-01 -7.01914132e-01 -5.65629661e-01 2.48869210e-01 -1.88635632e-01 -6.54650688e-01 -2.94958055e-01 -6.29476547e-01 -1.43574405e+00 -8.64764303e-02 -1.66415885e-01 -9.95193347e-02 5.39536893e-01 9.43247318e-01 3.78150433e-01 7.79620647e-01 3.59695703e-01 -8.06121767e-01 -1.98088452e-01 -7.83550024e-01 -2.38214001e-01 7.28475824e-02 4.72516656e-01 -3.76763552e-01 -6.55290559e-02 8.61855969e-02]
[9.594328880310059, 0.4926674962043762]
de5ece4c-8d1e-49f9-826b-df829aa6e39f
foc-osod-focus-on-classification-one-shot
null
null
https://openreview.net/forum?id=r7qgus1bZ2
https://openreview.net/pdf?id=r7qgus1bZ2
FOC OSOD: Focus on Classification One-Shot Object Detection
One-shot object detection (OSOD) aims at detecting all instances that are consistent with the category of the single reference image. OSOD achieves object detection by comparing the query image and the reference image. We observe that the essential problem behind the limited performance of OSOD is that OSOD generates a lot of false positives due to its poor classification ability. This paper analyzes the serious false positive problem in OSOD and proposes a Focus on Classification One-Shot Object Detection (FOC OSOD) framework, which is improved in two important aspects: (1) classification cascade head with the fixed IoU threshold can enhance the robustness of classification by comparing multiple close regions; (2) classification region deformation on the query feature and the reference feature to obtain a more effective comparison region. Without bells and whistles, a single FOC obtains 1.8% AP and 1.3% AP improvement on the seen classes and the unseen classes over a Siamese Faster R-CNN baseline on the MS-COCO dataset in the one-shot setting. The code will be available.
['Yu Zhang', 'SABA GHORBANI BARZEGAR', 'Huaijin Pi', 'Hanqing Yang']
2021-01-01
null
null
null
null
['one-shot-object-detection']
['computer-vision']
[-8.76877755e-02 -9.21793580e-02 -2.35218361e-01 -1.57265827e-01 -1.21251798e+00 -2.41850644e-01 6.27332091e-01 -5.86314872e-02 -6.63279235e-01 3.01641226e-01 -1.90742224e-01 4.03515875e-01 1.18456826e-01 -5.94247103e-01 -7.22347736e-01 -8.07161808e-01 1.64630666e-01 1.52300596e-01 9.35879409e-01 -1.21134438e-01 1.47402778e-01 4.35406893e-01 -1.73927081e+00 2.93712556e-01 4.28882837e-01 1.34938395e+00 2.47755349e-01 7.44402528e-01 3.63595109e-03 7.15292275e-01 -9.33303595e-01 -4.73447353e-01 3.72042656e-01 -4.67196763e-01 -5.27222514e-01 -9.32511687e-02 7.13123441e-01 -4.94708270e-01 -3.76460701e-01 1.21160829e+00 7.53059745e-01 3.59912246e-01 6.81722701e-01 -1.48614883e+00 -6.84204817e-01 3.61210942e-01 -9.04725075e-01 7.95712948e-01 -1.50218651e-01 1.68906599e-01 1.19029164e+00 -1.47177792e+00 6.08667612e-01 1.28123105e+00 3.55286807e-01 5.88125527e-01 -9.37633336e-01 -6.67490244e-01 3.09376102e-02 4.68416750e-01 -1.84854150e+00 -3.79567623e-01 2.66880572e-01 -1.95805833e-01 8.46015394e-01 2.71266073e-01 2.45664611e-01 8.69138122e-01 2.38923863e-01 9.58979070e-01 5.76157808e-01 -3.08223546e-01 2.91688889e-01 3.53904605e-01 6.29793108e-01 3.76208276e-01 3.08940142e-01 2.19509318e-01 -5.00154436e-01 1.65946826e-01 3.19786549e-01 3.28578085e-01 -1.35452852e-01 -4.13812965e-01 -1.00297403e+00 7.94060588e-01 1.10683513e+00 3.36911798e-01 -8.57893471e-03 -1.31984812e-03 4.05987769e-01 1.26002595e-01 2.44377062e-01 3.82209986e-01 -1.22724466e-01 5.70635498e-02 -8.36677790e-01 1.58990800e-01 4.74859536e-01 1.08583105e+00 5.56307435e-01 -1.56426594e-01 -5.53954661e-01 9.80263114e-01 -1.64740253e-02 5.57568908e-01 5.28626621e-01 -5.59501231e-01 3.80718559e-01 6.33451283e-01 1.37272596e-01 -8.99197340e-01 -1.89968571e-01 -8.59106898e-01 -5.00221670e-01 1.86780527e-01 4.96134639e-01 6.09383583e-02 -1.02828753e+00 1.41770041e+00 3.08385581e-01 5.18745594e-02 7.06064478e-02 1.21795511e+00 1.30748677e+00 6.67322338e-01 -1.57316744e-01 6.94031566e-02 1.51436448e+00 -1.37296557e+00 -6.32744312e-01 -2.96269119e-01 7.22771049e-01 -6.17744505e-01 9.26273525e-01 4.29245271e-02 -8.04380238e-01 -7.38412619e-01 -1.30550838e+00 -7.83427507e-02 -6.82564557e-01 3.66687804e-01 1.87449187e-01 4.51128900e-01 -6.57644689e-01 2.05899999e-01 -6.55104876e-01 -5.41968584e-01 7.30599761e-01 2.28471324e-01 -2.61807650e-01 -2.16121852e-01 -9.66293037e-01 8.65493238e-01 4.80211020e-01 -6.84569478e-02 -1.10035741e+00 -5.64371467e-01 -7.39090621e-01 2.24385396e-01 7.72028387e-01 9.09910873e-02 1.19769621e+00 -9.95764017e-01 -7.63445675e-01 9.99969602e-01 -8.04202035e-02 -6.52699172e-01 7.10459411e-01 -3.96416247e-01 -3.44572127e-01 2.60635912e-01 3.00142616e-01 7.93281138e-01 7.01675653e-01 -8.60907435e-01 -8.30187440e-01 -5.70598066e-01 -1.84015110e-01 1.21407248e-01 -1.90490529e-01 2.49029383e-01 -8.80651653e-01 -6.62465096e-01 1.45234928e-01 -8.90311420e-01 5.59670292e-02 3.03201735e-01 -3.19366515e-01 -4.45695013e-01 9.96747494e-01 -1.88389137e-01 1.06701410e+00 -2.59895349e+00 -2.58187622e-01 -2.73775786e-01 3.27428520e-01 5.49301803e-01 -3.50317240e-01 1.26754016e-01 -8.36636797e-02 -7.11796507e-02 1.77365586e-01 -7.61790201e-02 -2.56688476e-01 -1.03515409e-01 -3.68019134e-01 5.86859465e-01 5.20543635e-01 1.08598506e+00 -8.28277469e-01 -4.95563984e-01 5.81702478e-02 3.07925850e-01 -2.28872463e-01 1.99659169e-01 1.70582801e-01 -1.26492381e-01 -1.30138800e-01 7.27790475e-01 7.98954546e-01 -4.83656019e-01 -4.51442033e-01 -2.03649223e-01 5.42204492e-02 -1.68333631e-02 -1.34291935e+00 1.27558386e+00 5.82106672e-02 7.36442626e-01 -2.65279740e-01 -7.67694056e-01 9.38373268e-01 -1.35113939e-03 9.36298296e-02 -8.21095943e-01 2.49817804e-01 1.97301984e-01 1.36117831e-01 -3.11327815e-01 4.65373397e-01 1.96582451e-01 1.48327217e-01 2.66337872e-01 3.64570349e-01 4.38253641e-01 1.27071947e-01 6.28270209e-01 9.33335185e-01 -1.76845923e-01 3.99793386e-01 -2.94437557e-01 3.37750912e-01 -3.83313335e-02 6.33046567e-01 1.24533594e+00 -7.21474886e-01 8.74601722e-01 4.27936912e-01 -3.23076904e-01 -7.57015288e-01 -1.12077034e+00 -4.20731604e-01 1.37279308e+00 6.48495018e-01 -1.27324253e-01 -5.55438697e-01 -8.24484050e-01 -4.01669256e-02 6.52884364e-01 -9.86354232e-01 -4.23756719e-01 -3.02419037e-01 -9.67162371e-01 4.13950026e-01 6.92551076e-01 6.20230317e-01 -8.25913668e-01 -7.80198693e-01 -5.93487099e-02 4.86534685e-02 -1.02262318e+00 -5.78477502e-01 3.85381162e-01 -4.76511627e-01 -1.11409652e+00 -9.24458146e-01 -7.80449688e-01 5.10118544e-01 9.24117327e-01 8.01292539e-01 7.03246817e-02 -6.66758001e-01 -9.03426781e-02 -5.01863718e-01 -5.48200667e-01 -1.36258364e-01 -6.30385578e-02 -1.09773576e-01 1.85659975e-01 7.61463761e-01 3.57944936e-01 -6.88121617e-01 7.13611543e-01 -6.35255277e-01 -2.93411791e-01 5.76267719e-01 9.10884321e-01 4.61196899e-01 -2.75117517e-01 7.24002123e-01 -6.30907118e-01 -5.38639426e-02 -4.98205036e-01 -5.67485869e-01 3.01139474e-01 -5.74108958e-01 -2.17190996e-01 1.40942842e-01 -5.64452767e-01 -8.53558302e-01 -1.29326895e-01 1.42686322e-01 -6.76694810e-01 7.74234459e-02 -2.03753948e-01 1.05352245e-01 4.56792153e-02 9.87872601e-01 1.90711245e-01 -3.65247801e-02 -3.90673935e-01 2.16277376e-01 8.79621327e-01 5.71225464e-01 9.48645920e-02 6.18620574e-01 5.81081748e-01 -4.10050780e-01 -7.72635639e-01 -1.24738741e+00 -9.63340282e-01 -4.96406823e-01 -2.60007590e-01 8.57510507e-01 -1.22002304e+00 -4.02755886e-01 3.72433573e-01 -1.07998407e+00 1.23696409e-01 -3.28968734e-01 4.93241996e-01 -1.80655420e-01 1.75977666e-02 -5.86163402e-01 -8.89404595e-01 -4.47825819e-01 -1.16482985e+00 1.20251322e+00 5.09875178e-01 2.11770684e-02 -1.17655523e-01 -2.94968992e-01 2.02483296e-01 2.67200351e-01 -6.31468073e-02 3.86338204e-01 -1.06965315e+00 -5.85542738e-01 -6.31471694e-01 -6.60252512e-01 3.11575592e-01 -1.35386050e-01 1.47755638e-01 -1.23460889e+00 -5.04522145e-01 -4.82779592e-02 -4.12567943e-01 1.02536178e+00 2.58039564e-01 9.01916385e-01 4.08146977e-02 -4.11875814e-01 3.81701618e-01 1.48226380e+00 2.52761781e-01 6.96054161e-01 1.17211372e-01 3.58691841e-01 3.54706079e-01 1.00468743e+00 1.76238492e-01 -1.06923290e-01 8.57276559e-01 3.94518375e-01 -1.73612818e-01 -4.38498646e-01 -2.03609079e-01 3.50639760e-01 1.81464940e-01 2.71140635e-01 -2.59851575e-01 -8.12373757e-01 8.10208797e-01 -1.92798305e+00 -1.02688503e+00 -2.40707517e-01 2.29473424e+00 2.61637717e-01 3.00258875e-01 3.54717284e-01 4.53375094e-02 1.18094063e+00 2.70310622e-02 -6.58251166e-01 6.30254149e-02 -1.26099646e-01 -1.68902934e-01 4.18001413e-01 1.06644467e-01 -1.14822555e+00 8.34399521e-01 6.02610636e+00 1.13790762e+00 -1.01712775e+00 3.89454633e-01 7.39412487e-01 -3.79579246e-01 6.47146046e-01 -4.28027868e-01 -1.33608985e+00 4.83848631e-01 6.21045291e-01 -1.63360521e-01 -4.30035777e-02 1.20157492e+00 -3.11081558e-02 -4.15933520e-01 -1.04797864e+00 1.22097158e+00 4.29776520e-01 -1.20892406e+00 4.07973193e-02 -1.59672663e-01 5.94314933e-01 4.57451224e-01 -5.15859164e-02 7.07219839e-01 -8.28537345e-02 -7.52241910e-01 8.87645423e-01 9.64006931e-02 6.16901338e-01 -6.33270562e-01 8.99104238e-01 4.00687337e-01 -1.10522282e+00 -3.11630070e-01 -6.58713222e-01 1.58795208e-01 -3.17847639e-01 3.52120250e-01 -7.80243754e-01 2.81810880e-01 1.10417449e+00 6.26218557e-01 -9.39663529e-01 1.29450643e+00 3.47481295e-02 4.67350185e-01 -1.71332747e-01 -3.85927826e-01 3.00205439e-01 1.99224770e-01 6.95967376e-01 1.21526563e+00 5.45377657e-02 -7.37828091e-02 -5.81951998e-02 1.00513232e+00 -2.12280959e-01 6.22411221e-02 -3.96510303e-01 1.16248026e-01 3.79920632e-01 1.35397482e+00 -9.41428781e-01 -5.95564604e-01 -5.13738155e-01 8.90361071e-01 4.21433836e-01 2.77607083e-01 -1.04806828e+00 -7.54780889e-01 3.27089161e-01 -3.72132286e-02 7.25471437e-01 3.37657809e-01 -1.28920928e-01 -1.15776885e+00 1.12285450e-01 -5.68320215e-01 6.89182818e-01 -7.95865774e-01 -1.28171265e+00 5.89336932e-01 -1.81859687e-01 -1.37724924e+00 3.37059021e-01 -5.27240515e-01 -6.58966482e-01 5.41341186e-01 -1.18979442e+00 -8.25697899e-01 -5.60482383e-01 2.58766234e-01 8.26120138e-01 -7.59476274e-02 4.11137700e-01 4.60050732e-01 -9.27990258e-01 8.56829762e-01 1.58256234e-03 5.30220509e-01 7.61169434e-01 -1.06343663e+00 3.79805535e-01 1.04450488e+00 3.15345109e-01 3.69636685e-01 5.47579348e-01 -4.46486026e-01 -1.10510063e+00 -1.19470513e+00 7.04979599e-01 -5.61131418e-01 5.02159178e-01 -4.61034745e-01 -1.05793428e+00 4.19326127e-01 -3.37094039e-01 5.48375428e-01 3.44326645e-01 -5.85812330e-02 -6.09438539e-01 -3.04626882e-01 -1.08954787e+00 4.45075572e-01 8.89348030e-01 -3.49192977e-01 -7.19978273e-01 3.05301487e-01 7.58335412e-01 -2.52678573e-01 -5.02021849e-01 3.75340343e-01 4.83084232e-01 -9.00797129e-01 8.19629610e-01 -5.95624983e-01 3.32999408e-01 -4.93224114e-01 -2.28753462e-01 -6.29119396e-01 -2.52992809e-01 -2.00387627e-01 -2.19367802e-01 1.19813204e+00 3.41618001e-01 -3.98162305e-01 5.07211506e-01 3.88192654e-01 1.93736017e-01 -7.09545016e-01 -1.07570696e+00 -1.12948239e+00 -1.88746110e-01 -1.71193108e-01 2.76937902e-01 6.69936001e-01 -3.54871482e-01 5.40137231e-01 -2.63767242e-01 1.54093534e-01 6.34974182e-01 1.52245253e-01 8.64079177e-01 -9.43344951e-01 -2.02140883e-01 -2.45627895e-01 -7.04355776e-01 -9.97241259e-01 -5.87630510e-01 -7.74688005e-01 3.19651872e-01 -1.08580339e+00 7.28405654e-01 -1.39397994e-01 -6.66694701e-01 3.93105328e-01 -4.30635929e-01 7.41208971e-01 3.21782619e-01 4.02289778e-01 -1.22630358e+00 5.37158906e-01 1.09834683e+00 -1.58752739e-01 -1.19179696e-01 -4.08858396e-02 -5.77775896e-01 3.69979829e-01 4.02121872e-01 -7.68190444e-01 -3.74886356e-02 -1.43826827e-01 -1.52233154e-01 -1.59865305e-01 6.06068313e-01 -1.05566335e+00 2.31087297e-01 2.49145180e-01 5.01784444e-01 -9.17215407e-01 3.10444474e-01 -5.04433632e-01 -3.20798457e-01 8.21750879e-01 -4.64856237e-01 -2.40622371e-01 8.31479654e-02 8.99984956e-01 -1.43985137e-01 -3.70849878e-01 1.40906990e+00 -7.13841850e-03 -1.01748681e+00 1.10684752e-01 -3.95541219e-03 2.32517034e-01 1.44101655e+00 -3.18632007e-01 -6.46488547e-01 -1.26024529e-01 -5.13309598e-01 2.89585650e-01 2.56005734e-01 7.85185993e-01 4.98758614e-01 -1.21831393e+00 -6.75764740e-01 6.09260015e-02 6.73287809e-01 -1.96322232e-01 1.92056924e-01 9.30755854e-01 -3.35060894e-01 3.17833483e-01 1.66732296e-02 -8.72975647e-01 -1.32095337e+00 8.05540919e-01 6.37559652e-01 7.41296411e-02 -7.19555080e-01 1.18818879e+00 4.80001301e-01 2.34949514e-02 5.91485143e-01 3.34937163e-02 -9.44061130e-02 1.55807927e-01 1.16370666e+00 6.05893493e-01 2.04759106e-01 -6.88205779e-01 -6.13672495e-01 3.39474261e-01 -5.64959347e-01 2.39094511e-01 1.16971076e+00 2.34755836e-02 1.80455059e-01 5.17163396e-01 1.38000572e+00 -4.10398424e-01 -1.27969754e+00 -4.63218063e-01 1.77252255e-02 -6.36534691e-01 2.69491434e-01 -7.22396374e-01 -1.12355793e+00 8.89828324e-01 1.17689514e+00 1.16289854e-01 6.90250337e-01 4.71850544e-01 5.50272644e-01 4.21143115e-01 1.05764255e-01 -1.11250556e+00 3.30146998e-01 2.81213611e-01 7.96669483e-01 -1.57138681e+00 1.04293125e-02 -4.20215130e-01 -7.19251156e-01 8.80923986e-01 9.08828855e-01 -3.17817658e-01 3.77265513e-01 -1.69559922e-02 -4.60763760e-02 -3.91879171e-01 -6.68640971e-01 -4.59458023e-01 5.37260592e-01 2.87780881e-01 3.81557383e-02 -1.67523757e-01 -1.90253437e-01 5.86273491e-01 3.84671062e-01 -1.85217127e-01 3.40841442e-01 6.89433575e-01 -6.78610206e-01 -2.33092442e-01 -4.09322828e-01 6.55060351e-01 -5.01993001e-01 9.66275558e-02 -3.87968987e-01 7.81487644e-01 9.55017358e-02 1.06286168e+00 3.95371228e-01 -2.06441000e-01 3.38384688e-01 -1.69685230e-01 7.87822306e-02 -8.07730436e-01 -5.79741299e-01 1.35089159e-02 -3.13092709e-01 -6.97967947e-01 -1.35810509e-01 -3.78190100e-01 -1.20608330e+00 1.16231218e-01 -9.64735627e-01 4.59230766e-02 4.13369209e-01 7.18239307e-01 3.30674887e-01 4.38783616e-01 5.39457619e-01 -6.07711077e-01 -8.53391051e-01 -8.92935216e-01 -6.55067325e-01 4.99311954e-01 4.67961729e-01 -8.85561883e-01 -6.06311083e-01 -4.00104880e-01]
[9.37417984008789, 1.2021145820617676]
49027efe-0ed5-4608-b023-1d275f159f4f
fast-and-effective-adaptation-of-facial
1909.12158
null
https://arxiv.org/abs/1909.12158v2
https://arxiv.org/pdf/1909.12158v2.pdf
Fast and Effective Adaptation of Facial Action Unit Detection Deep Model
Detecting facial action units (AU) is one of the fundamental steps in automatic recognition of facial expression of emotions and cognitive states. Though there have been a variety of approaches proposed for this task, most of these models are trained only for the specific target AUs, and as such they fail to easily adapt to the task of recognition of new AUs (i.e., those not initially used to train the target models). In this paper, we propose a deep learning approach for facial AU detection that can easily and in a fast manner adapt to a new AU or target subject by leveraging only a few labeled samples from the new task (either an AU or subject). To this end, we propose a modeling approach based on the notion of the model-agnostic meta-learning, originally proposed for the general image recognition/detection tasks (e.g., the character recognition from the Omniglot dataset). Specifically, each subject and/or AU is treated as a new learning task and the model learns to adapt based on the knowledge of the previous tasks (the AUs and subjects used to pre-train the target models). Thus, given a new subject or AU, this meta-knowledge (that is shared among training and test tasks) is used to adapt the model to the new task using the notion of deep learning and model-agnostic meta-learning. We show on two benchmark datasets (BP4D and DISFA) for facial AU detection that the proposed approach can be easily adapted to new tasks (AUs/subjects). Using only a few labeled examples from these tasks, the model achieves large improvements over the baselines (i.e., non-adapted models).
['Vladimir Pavlovic', 'Ognjen Rudovic', 'Maja Pantic', 'Mihee Lee']
2019-09-26
null
null
null
null
['action-unit-detection', 'facial-action-unit-detection']
['computer-vision', 'computer-vision']
[ 4.52627540e-01 7.23666921e-02 6.57995492e-02 -5.52749395e-01 -4.92272228e-01 -3.44490141e-01 5.98511636e-01 -2.60624915e-01 -3.78382206e-01 2.90695369e-01 -4.72756058e-01 3.40916574e-01 4.83709514e-01 -7.10251391e-01 -7.86023140e-01 -8.31122994e-01 4.41176966e-02 3.56193125e-01 1.12141944e-01 -1.97040334e-01 -1.35924280e-01 6.49837375e-01 -1.77421820e+00 5.05894244e-01 4.76098269e-01 1.29612756e+00 -1.69422049e-02 7.26361692e-01 1.64738640e-01 6.45801127e-01 -6.70370877e-01 -4.66277897e-01 2.07556397e-01 -6.81332171e-01 -9.34229374e-01 4.98870403e-01 5.32798052e-01 -4.26276803e-01 -3.08269799e-01 8.33854675e-01 3.34584028e-01 2.86942363e-01 7.33816624e-01 -1.30264616e+00 -4.51690763e-01 1.39385104e-01 -5.83030164e-01 -2.54472997e-02 4.03105840e-02 8.89667794e-02 4.94581103e-01 -9.95444477e-01 4.35651600e-01 1.31112158e+00 2.36517578e-01 1.34086418e+00 -1.15935504e+00 -8.08474720e-01 3.69133741e-01 2.59706616e-01 -1.26116753e+00 -7.37061441e-01 7.32098162e-01 -2.73135841e-01 6.26412570e-01 6.05154485e-02 5.08402467e-01 1.44662106e+00 -2.08259344e-01 1.05886734e+00 1.10625958e+00 -3.75136167e-01 5.39048791e-01 4.15437222e-01 1.04442127e-01 7.00354218e-01 -3.41239393e-01 1.28964230e-01 -3.99268478e-01 -2.30064079e-01 4.09938425e-01 -2.66230196e-01 2.84887268e-03 -2.18270287e-01 -6.21597052e-01 6.29501164e-01 3.07974935e-01 2.40739688e-01 -3.59689236e-01 7.41006900e-03 5.64322710e-01 3.17581147e-01 5.43897450e-01 1.43623427e-01 -4.87711310e-01 4.62869667e-02 -6.30635977e-01 -4.90102619e-02 7.06560671e-01 6.41208708e-01 1.06977260e+00 2.11138010e-01 -1.82744831e-01 9.06143725e-01 4.91112061e-02 3.54550272e-01 6.72328472e-01 -6.83575928e-01 -2.29282882e-02 7.55792320e-01 -7.52390843e-05 -7.70387352e-01 -4.72513855e-01 -3.13117504e-01 -5.78130484e-01 6.57938659e-01 3.45500827e-01 -3.57336491e-01 -1.26633561e+00 2.18721461e+00 6.65567517e-01 4.32151884e-01 3.69102120e-01 5.41251302e-01 8.06621611e-01 8.04193020e-01 1.08728699e-01 -2.93160826e-01 1.21413493e+00 -9.94570732e-01 -2.77304143e-01 -5.37681341e-01 5.32782376e-01 -4.80354339e-01 8.44818890e-01 5.94462752e-01 -9.13689256e-01 -8.53813887e-01 -9.53496337e-01 4.96531099e-01 -2.97952354e-01 2.42101714e-01 3.53678733e-01 6.12613320e-01 -8.85603607e-01 3.45848203e-01 -6.81422591e-01 -9.17570770e-01 5.78597248e-01 4.46044654e-01 -3.59208792e-01 -1.12591565e-01 -9.15780902e-01 7.49977291e-01 4.08653617e-01 2.27444410e-01 -1.51916242e+00 -2.00345322e-01 -6.21410191e-01 1.67527534e-02 6.01867080e-01 -3.67538661e-01 1.41785765e+00 -2.07622623e+00 -1.91144192e+00 1.02523661e+00 -9.75496396e-02 -3.14796269e-01 3.75565290e-01 -1.33870751e-01 -4.77755785e-01 2.79984176e-01 -2.89780259e-01 6.90628052e-01 1.48056221e+00 -1.38958204e+00 -7.34476507e-01 -4.03596580e-01 3.76894921e-01 8.88583884e-02 -6.52999520e-01 2.10232005e-01 -3.41719389e-01 -3.35217685e-01 -3.17621440e-01 -1.14236295e+00 1.47726536e-01 1.62111595e-02 -1.61328077e-01 -3.84927630e-01 1.02459168e+00 -3.04626673e-01 9.27236259e-01 -2.27490735e+00 2.98924208e-01 2.28331700e-01 1.14731878e-01 6.31230533e-01 -5.74721694e-01 4.24472615e-02 -2.14421138e-01 -1.84034243e-01 -1.33601114e-01 -4.76597309e-01 -2.39788726e-01 3.62601638e-01 3.62730809e-02 2.54366487e-01 3.92847985e-01 5.78847468e-01 -9.42504466e-01 -1.33407235e-01 -6.09608814e-02 1.85005650e-01 -2.81482726e-01 6.24276340e-01 -3.09623837e-01 4.37795758e-01 -4.81536150e-01 6.57360613e-01 5.70312381e-01 6.32364601e-02 7.56432042e-02 -2.86073506e-01 2.53584832e-01 -4.02130038e-01 -9.86624241e-01 1.24416900e+00 -7.33704209e-01 2.11446777e-01 9.68759209e-02 -1.07271743e+00 1.17480969e+00 5.07859409e-01 3.69099319e-01 -3.48174542e-01 2.17978343e-01 1.45173207e-01 1.09666869e-01 -7.05676198e-01 -1.66263521e-01 -2.40228742e-01 -2.57312413e-03 7.59953499e-01 3.91450852e-01 3.61236244e-01 1.32985920e-01 -1.27986386e-01 1.17241824e+00 2.56822467e-01 4.73180562e-01 1.74477398e-01 8.70874822e-01 -2.69404888e-01 7.35916018e-01 6.46698654e-01 -3.88625562e-01 3.03969264e-01 3.96848947e-01 -8.21884096e-01 -7.50225663e-01 -8.57335031e-01 1.94181576e-01 1.68233895e+00 -2.98831105e-01 -2.43631542e-01 -1.03562534e+00 -1.14249933e+00 -1.95173666e-01 3.88051689e-01 -1.19605470e+00 -7.31775701e-01 -4.02782708e-01 -8.32664907e-01 5.11481106e-01 4.96348351e-01 6.03021383e-01 -1.39455748e+00 -7.81150401e-01 2.28452310e-01 2.13790387e-01 -1.05836713e+00 -1.86716482e-01 9.52599123e-02 -6.05263174e-01 -1.08030665e+00 -6.81915104e-01 -7.92681575e-01 8.59434068e-01 2.50870287e-02 7.37901747e-01 1.09038599e-01 -1.34555310e-01 7.00858653e-01 -4.56319779e-01 -4.51636106e-01 -8.28523755e-01 -6.03281446e-02 2.30407134e-01 1.04274535e+00 5.47047794e-01 -3.39548558e-01 -4.05136198e-01 3.92331451e-01 -1.00895882e+00 -3.24976817e-02 7.56551087e-01 1.04542673e+00 3.31456989e-01 -3.46554458e-01 7.79156148e-01 -1.00477886e+00 3.94821197e-01 -4.36307907e-01 -3.84461999e-01 4.51178253e-01 -2.84169137e-01 -1.94872186e-01 6.98194802e-01 -1.02499676e+00 -1.14272082e+00 4.95493323e-01 -6.58953935e-02 -6.55425787e-01 -5.31646788e-01 4.67644811e-01 -3.96940976e-01 -3.33029449e-01 7.90918291e-01 2.21741125e-01 1.70718804e-01 -3.50226104e-01 6.32993653e-02 7.43301868e-01 4.45201516e-01 -6.18737102e-01 7.62819350e-01 4.64277238e-01 2.58165076e-02 -7.21355379e-01 -9.89002824e-01 -8.67570788e-02 -9.54023838e-01 -4.15124148e-01 8.67071450e-01 -7.47410774e-01 -6.05010092e-01 9.06510293e-01 -1.13720334e+00 -5.32599449e-01 -2.45013461e-01 1.28660604e-01 -6.04316711e-01 2.03901663e-01 -3.21434796e-01 -9.68780696e-01 -3.43601257e-01 -9.99241233e-01 9.90469217e-01 4.89035308e-01 -2.82586128e-01 -9.75263000e-01 2.41954491e-01 -2.63651554e-03 2.05925629e-01 4.26886082e-01 9.13974822e-01 -7.60668218e-01 -8.84415861e-03 -5.13080955e-01 6.05771802e-02 9.47705030e-01 3.45577598e-01 1.49125010e-01 -1.47158170e+00 -4.12353069e-01 1.73078179e-01 -8.94279122e-01 7.18554914e-01 -1.76649630e-01 1.11402869e+00 -4.02810127e-01 -2.49847144e-01 4.85655844e-01 9.76830244e-01 3.93363953e-01 5.85575879e-01 2.14383349e-01 4.99436498e-01 6.18945956e-01 5.88218927e-01 5.20418227e-01 1.44205451e-01 8.58249247e-01 5.25915205e-01 -1.92110687e-01 1.45354658e-01 1.18118279e-01 9.00730491e-01 1.73883349e-01 -1.34302258e-01 -4.43118550e-02 -7.05919683e-01 3.20910841e-01 -1.90025973e+00 -8.57162893e-01 5.80071926e-01 2.25447798e+00 7.44546235e-01 -1.03936441e-01 4.70099151e-01 -1.43240958e-01 6.23921514e-01 7.66757056e-02 -9.12298083e-01 -7.99403071e-01 5.75472079e-02 4.60069478e-01 -3.14681977e-01 1.93975046e-01 -1.17104697e+00 1.23307300e+00 5.18143702e+00 5.98449230e-01 -1.37615430e+00 2.75807410e-01 8.30676079e-01 8.52465536e-03 4.35312182e-01 -2.64440626e-01 -8.12861979e-01 1.28569871e-01 9.39809740e-01 -1.18010446e-01 4.05836135e-01 1.08563483e+00 -3.99750136e-02 1.07902428e-02 -1.53373754e+00 9.22098398e-01 5.73055565e-01 -6.69450641e-01 2.29748353e-01 -3.06262851e-01 7.09298670e-01 -3.42045903e-01 2.40371481e-01 7.41338789e-01 -4.68368866e-02 -8.38916838e-01 5.77791214e-01 5.08335054e-01 7.17864931e-01 -6.93923712e-01 8.07980895e-01 3.43513459e-01 -9.36071157e-01 -3.31496090e-01 -4.52030629e-01 1.16164938e-01 -2.95590043e-01 -8.17520097e-02 -8.96543920e-01 2.03187123e-01 7.20232189e-01 6.47429287e-01 -6.73478425e-01 6.39753699e-01 -3.56486529e-01 6.63853943e-01 -1.92895338e-01 2.31297836e-02 2.60379702e-01 4.68592867e-02 4.64004904e-01 9.44972217e-01 2.08766401e-01 3.55753675e-02 2.15327725e-01 6.33158982e-01 -1.12909630e-01 2.43251890e-01 -3.43090981e-01 -3.27908099e-02 -6.95123374e-02 1.65701663e+00 -1.79489955e-01 -4.26967740e-01 -6.13263845e-01 1.27453434e+00 6.84246600e-01 4.61304516e-01 -6.01638556e-01 2.63374345e-03 8.19735885e-01 -9.96187553e-02 2.40420938e-01 1.94953650e-01 3.67022127e-01 -9.23787773e-01 -1.81698129e-01 -1.09370422e+00 6.89249635e-01 -6.76817358e-01 -1.46202207e+00 1.04583549e+00 2.89031956e-02 -1.13214684e+00 -5.10010898e-01 -8.06080103e-01 -9.33274925e-01 7.61205733e-01 -1.34371340e+00 -1.45411754e+00 -5.24185002e-01 8.78055513e-01 4.82180178e-01 -3.96550059e-01 1.01544106e+00 -1.35533782e-02 -7.59249508e-01 9.14524496e-01 -1.97578296e-01 4.00823295e-01 7.90095329e-01 -8.83893728e-01 5.63267544e-02 8.87665331e-01 1.13168605e-01 2.78840601e-01 3.82567525e-01 -3.99519980e-01 -1.14453375e+00 -1.38170207e+00 4.98712331e-01 -3.27021450e-01 5.25481582e-01 -6.53087974e-01 -1.05854595e+00 8.53375256e-01 -7.17187747e-02 2.30794236e-01 7.23361254e-01 2.85276249e-02 -3.33304822e-01 -4.75502014e-01 -1.21658731e+00 6.22903049e-01 8.71896446e-01 -3.49749058e-01 -3.31276625e-01 3.12546223e-01 4.99621592e-02 -2.73750424e-01 -5.33081532e-01 5.03325641e-01 7.08291173e-01 -8.44298244e-01 5.53708017e-01 -1.05767870e+00 1.02394767e-01 -6.28271624e-02 -6.08001538e-02 -1.32369781e+00 -3.20298404e-01 -6.38506353e-01 -4.35960501e-01 1.27295351e+00 2.54644215e-01 -6.08998239e-01 7.94413388e-01 5.99322319e-01 -1.19356522e-02 -8.65112066e-01 -1.09878314e+00 -7.26415873e-01 -9.11970064e-02 -3.11089724e-01 5.72944641e-01 6.54481947e-01 -2.12490439e-01 4.03758913e-01 -5.23827136e-01 1.89626694e-01 3.21984768e-01 2.08512768e-01 1.20367455e+00 -1.13025677e+00 -4.56656337e-01 -2.30834410e-01 -6.00463331e-01 -7.33361959e-01 5.02935529e-01 -8.73063743e-01 1.60555616e-01 -9.84458685e-01 3.42147678e-01 -3.47567409e-01 -4.91807371e-01 1.03961480e+00 -3.34925234e-01 2.72244245e-01 2.87557781e-01 1.32422030e-01 -4.42060769e-01 6.52515173e-01 8.90490890e-01 -2.35806659e-01 -3.84458333e-01 4.90816236e-01 -5.52998722e-01 9.26239192e-01 7.94452727e-01 -4.87898797e-01 -2.29948506e-01 -1.35443196e-01 9.21448916e-02 -3.60060900e-01 4.87032056e-01 -1.13198900e+00 -7.36405477e-02 -1.36565551e-01 4.73734170e-01 1.05206102e-01 4.39499736e-01 -8.42271328e-01 -1.23952381e-01 3.90237242e-01 -1.52964607e-01 -2.13866085e-01 3.76113772e-01 4.88553345e-01 -3.35801020e-02 -4.38118905e-01 1.15261912e+00 -2.30789050e-01 -9.67393219e-01 6.60563827e-01 -5.17300427e-01 -2.53596157e-01 1.34688139e+00 -1.85608983e-01 -1.74885795e-01 -4.78877902e-01 -1.07703280e+00 1.92484193e-04 3.90593588e-01 5.18237412e-01 6.60028040e-01 -1.20682883e+00 -7.76095748e-01 2.21232608e-01 4.75327015e-01 -1.96696281e-01 2.44867325e-01 7.48777688e-01 7.33180717e-02 -2.49814138e-01 -4.82032686e-01 -6.96659267e-01 -1.49875653e+00 7.92588532e-01 7.25327313e-01 -1.23276398e-01 -9.23777148e-02 9.18651342e-01 7.61597931e-01 -3.17111880e-01 1.50126144e-01 8.94496962e-02 -3.74757379e-01 8.42652246e-02 7.42572546e-01 2.15180013e-02 -2.92757191e-02 -8.57949078e-01 -1.51542842e-01 5.28288007e-01 -3.95201474e-01 1.56248897e-01 1.32981372e+00 3.28271352e-02 -1.15610780e-02 6.89503133e-01 1.20900321e+00 -3.16058666e-01 -1.47212684e+00 -5.46767771e-01 -1.01336963e-01 -2.73587048e-01 -4.54148859e-01 -7.34993458e-01 -1.18892670e+00 1.04820132e+00 9.18768883e-01 -3.29983473e-01 1.43879223e+00 1.39273852e-01 5.49521029e-01 4.91837919e-01 3.19133461e-01 -1.18431258e+00 6.39086187e-01 3.38041842e-01 8.97347629e-01 -1.17405629e+00 -2.78808326e-01 -4.77466024e-02 -7.40247667e-01 1.56175637e+00 1.18304372e+00 7.62172267e-02 4.44258273e-01 2.04990748e-02 2.33723581e-01 -3.17533724e-02 -8.29981267e-01 -3.17964524e-01 3.05941969e-01 7.17069209e-01 -4.56378004e-03 -2.03492165e-01 3.11872184e-01 6.24680758e-01 3.52059782e-01 6.29315004e-02 3.62290770e-01 9.70279157e-01 -4.73012358e-01 -1.12408459e+00 -1.90635547e-01 3.83986771e-01 -1.10076860e-01 2.86655128e-01 -8.00902605e-01 7.40761578e-01 3.21801633e-01 8.14351439e-01 -5.93613461e-03 -6.53986514e-01 3.63206148e-01 3.66577119e-01 6.26602709e-01 -9.97379065e-01 -5.69385290e-01 -1.33650377e-01 -1.07711837e-01 -5.57541609e-01 -6.47624373e-01 -7.38904536e-01 -9.67739701e-01 1.34560958e-01 -9.95419770e-02 -6.60178363e-02 2.30257496e-01 1.01826799e+00 2.46810138e-01 1.03783786e-01 1.08748853e+00 -8.59359741e-01 -5.48909545e-01 -1.07746530e+00 -4.72979605e-01 5.92745662e-01 1.27098039e-01 -7.17193246e-01 -4.19142932e-01 1.56042695e-01]
[13.601908683776855, 1.66022789478302]
7df860fe-3e88-49ce-934e-9ad2951d44da
reducing-crowdsourcing-to-graphon-estimation
1703.08085
null
https://arxiv.org/abs/1703.08085v4
https://arxiv.org/pdf/1703.08085v4.pdf
Reducing Crowdsourcing to Graphon Estimation, Statistically
Inferring the correct answers to binary tasks based on multiple noisy answers in an unsupervised manner has emerged as the canonical question for micro-task crowdsourcing or more generally aggregating opinions. In graphon estimation, one is interested in estimating edge intensities or probabilities between nodes using a single snapshot of a graph realization. In the recent literature, there has been exciting development within both of these topics. In the context of crowdsourcing, the key intellectual challenge is to understand whether a given task can be more accurately denoised by aggregating answers collected from other different tasks. In the context of graphon estimation, precise information limits and estimation algorithms remain of interest. In this paper, we utilize a statistical reduction from crowdsourcing to graphon estimation to advance the state-of-art for both of these challenges. We use concepts from graphon estimation to design an algorithm that achieves better performance than the {\em majority voting} scheme for a setup that goes beyond the {\em rank one} models considered in the literature. We use known explicit lower bounds for crowdsourcing to provide refined lower bounds for graphon estimation.
['Christina Lee Yu', 'Devavrat Shah']
2017-03-23
null
null
null
null
['graphon-estimation']
['graphs']
[ 2.92030841e-01 5.50593913e-01 1.38890639e-01 -2.37464860e-01 -1.15819228e+00 -9.37439263e-01 5.81718385e-01 5.26300848e-01 -4.87157762e-01 8.46938550e-01 1.00788042e-01 -1.39433280e-01 -6.49618581e-02 -6.98752999e-01 -6.91172183e-01 -6.97881639e-01 2.42262006e-01 8.09391737e-01 5.10602653e-01 -3.20710897e-01 3.70481052e-02 2.54297078e-01 -1.20278406e+00 2.67321877e-02 8.77975345e-01 8.05275559e-01 -1.77404925e-01 1.04598105e+00 -1.76572338e-01 9.22129452e-01 -8.17542195e-01 -1.00568509e+00 4.11634684e-01 -4.53541011e-01 -7.15594947e-01 2.42919832e-01 5.16224504e-01 5.08432202e-02 -1.80667654e-01 1.30495024e+00 6.78536773e-01 2.43919000e-01 5.06349385e-01 -1.58771360e+00 -6.53680384e-01 6.44825637e-01 -7.71777928e-01 2.66480237e-01 6.10201061e-01 -9.12929028e-02 1.11653078e+00 -7.12406099e-01 5.60770452e-01 1.29710460e+00 6.15861297e-01 2.40339607e-01 -1.22033572e+00 -2.25458950e-01 1.98110163e-01 2.49202684e-01 -1.56566060e+00 -2.02588931e-01 4.92156625e-01 -6.29749775e-01 4.46894556e-01 5.47108948e-01 3.53186071e-01 6.99205875e-01 1.19795073e-02 7.50821292e-01 1.28405952e+00 -5.61661065e-01 5.87438345e-01 2.23113298e-01 9.91750062e-02 6.57821119e-01 7.12353170e-01 -6.94897830e-01 -8.72662544e-01 -6.63754582e-01 3.32116604e-01 -2.71957099e-01 -2.39608124e-01 -1.67068869e-01 -8.19813132e-01 9.14272428e-01 9.91133153e-02 1.47176236e-01 -2.34423652e-01 3.15467358e-01 -2.34618643e-03 2.32385889e-01 1.07539701e+00 2.41773367e-01 7.45892450e-02 9.86953303e-02 -1.01651132e+00 5.15695810e-01 1.44127488e+00 1.01696634e+00 9.90237951e-01 -2.17496797e-01 -4.08810586e-01 3.37724507e-01 1.45039439e-01 5.18743336e-01 -1.67114437e-01 -1.16132700e+00 5.83115041e-01 5.37474930e-01 6.00274205e-01 -1.15329039e+00 -1.35612980e-01 -1.39697909e-01 -6.37598753e-01 -1.47209717e-02 9.80792403e-01 -3.71814072e-01 -4.85866368e-01 1.47573817e+00 7.69493043e-01 3.59065272e-02 -4.87821460e-01 1.04410219e+00 5.96195459e-01 5.17961457e-02 -9.35772210e-02 -6.80553243e-02 1.38679171e+00 -7.46393025e-01 -6.83409214e-01 -2.40558550e-01 5.48404276e-01 -7.41765797e-01 5.99474907e-01 5.92632592e-01 -8.21951687e-01 -2.01736972e-01 -7.86826789e-01 -3.53126794e-01 -3.31552207e-01 -5.02377152e-02 4.88006502e-01 9.75715756e-01 -1.47804487e+00 4.92030412e-01 -4.30633873e-01 -4.84663844e-01 2.05065176e-01 4.16686088e-01 -2.61248171e-01 -2.90137410e-01 -1.27285504e+00 8.62044156e-01 -1.21911056e-01 1.92276537e-01 -6.49657369e-01 -7.98065215e-02 -7.18328238e-01 -2.92517573e-01 9.08128023e-01 -6.32089019e-01 1.20409036e+00 -6.66085303e-01 -1.10765421e+00 9.44252729e-01 -5.31120300e-01 -5.22954822e-01 1.07155681e+00 8.34254324e-02 7.04485327e-02 1.56358629e-01 4.58285511e-01 4.24122304e-01 8.61998498e-01 -1.29702437e+00 -5.37569404e-01 -6.53260112e-01 4.25516725e-01 1.50162905e-01 -5.18378615e-02 5.04302792e-02 -4.95836973e-01 -3.99099201e-01 -1.07844464e-01 -1.33357954e+00 -5.60841680e-01 -5.96024320e-02 -6.55924976e-01 -5.72535694e-01 1.58784911e-01 -6.39753222e-01 9.82046127e-01 -1.51650012e+00 2.10498571e-01 2.49768659e-01 1.08906674e+00 3.55952084e-02 8.93357545e-02 5.47187269e-01 5.24935722e-01 2.84092456e-01 -2.27988988e-01 -7.83282518e-01 2.90682912e-01 2.30714381e-01 -2.04752296e-01 1.06020093e+00 1.61977887e-01 1.00315785e+00 -1.21124136e+00 -6.41650021e-01 -1.13624722e-01 2.13564172e-01 -1.45151407e-01 3.14449668e-02 -3.49431813e-01 4.76352751e-01 -4.58163857e-01 4.30100441e-01 8.47279310e-01 -5.77519655e-01 4.97704357e-01 2.58798927e-01 3.45896006e-01 -3.51576686e-01 -1.56516719e+00 1.11755276e+00 -9.12080780e-02 7.19854116e-01 5.32228708e-01 -7.94696629e-01 6.46895826e-01 2.45559469e-01 2.79353648e-01 -6.01284727e-02 2.02993006e-01 2.52647787e-01 -4.21721846e-01 -5.80718458e-01 8.96477699e-01 -8.42882991e-02 -1.82907000e-01 5.66161931e-01 -1.10589005e-01 -2.37560198e-01 5.48569560e-01 5.04533887e-01 1.44791937e+00 -2.27395549e-01 5.65158308e-01 -2.97509044e-01 3.68804395e-01 1.28557652e-01 1.63750097e-01 1.30932486e+00 -2.81613052e-01 6.94720507e-01 8.26934755e-01 -1.16133392e-01 -7.29005814e-01 -9.55395639e-01 2.33561322e-01 1.35340214e+00 3.22065502e-01 -3.01420301e-01 -1.07628441e+00 -6.35937631e-01 2.25343183e-01 2.89442748e-01 -8.41273189e-01 4.60742027e-01 -3.10381860e-01 -8.02173376e-01 5.51974356e-01 2.63638526e-01 6.51506856e-02 -3.68156612e-01 1.55871063e-02 6.70348331e-02 -3.83253157e-01 -1.58850014e+00 -6.32855654e-01 -1.33573666e-01 -3.97428900e-01 -1.19638240e+00 -1.08604109e+00 -2.87465423e-01 6.53142095e-01 5.86244226e-01 1.40608656e+00 1.91593587e-01 2.12731659e-01 8.52507472e-01 -4.20824647e-01 -6.80796266e-01 -2.65309244e-01 3.43902946e-01 2.66656745e-02 4.70422447e-01 7.18118250e-01 -4.91959274e-01 -5.44460893e-01 4.70954061e-01 -1.10215151e+00 -5.34504116e-01 1.87440872e-01 1.07695855e-01 3.99329871e-01 -2.35868230e-01 4.83950645e-01 -1.19557583e+00 1.13260949e+00 -9.37361538e-01 -6.77659988e-01 3.18201512e-01 -3.93506855e-01 9.38872695e-02 5.54671526e-01 -2.34650791e-01 -6.21688366e-01 1.21735409e-01 3.34971160e-01 -4.13366966e-03 6.46704808e-02 2.72265792e-01 5.49717210e-02 -4.44457531e-01 9.27096605e-01 -1.18882254e-01 -8.94862339e-02 -2.21687958e-01 5.99586964e-01 7.16666460e-01 3.74090761e-01 -4.91584927e-01 8.46261203e-01 1.02364898e+00 2.46441245e-01 -9.15816307e-01 -1.13293660e+00 -8.86134505e-01 -4.63546723e-01 -4.21293139e-01 9.72978354e-01 -1.16167319e+00 -9.83262718e-01 2.15897113e-01 -1.42358017e+00 -2.28271320e-01 -1.83625326e-01 3.12384125e-02 -1.87099323e-01 8.82168651e-01 -3.61119658e-01 -1.45811534e+00 3.22857015e-02 -1.04338562e+00 1.56930530e+00 1.68912366e-01 -4.39979136e-01 -1.18303955e+00 2.37785101e-01 6.11779392e-01 2.23502353e-01 3.10575634e-01 4.00406905e-02 -9.31709468e-01 -8.32613945e-01 -6.74443364e-01 -1.53541714e-01 1.14366263e-01 -2.52580136e-01 -2.84511000e-01 -1.15679026e+00 -1.19980104e-01 1.83120459e-01 -1.41393140e-01 8.21181118e-01 3.20473790e-01 7.53682852e-01 -2.14593217e-01 -2.81928927e-01 -2.45490089e-01 1.22794795e+00 -7.06168175e-01 4.06837970e-01 -1.10345349e-01 8.47222924e-01 8.94645214e-01 2.21512601e-01 5.68170011e-01 9.29454744e-01 6.40468478e-01 3.28285754e-01 2.20267579e-01 2.07937390e-01 -1.29522830e-01 1.89166352e-01 6.85930073e-01 -1.70451477e-01 -8.59591365e-01 -8.25216770e-01 8.60083342e-01 -2.10553265e+00 -6.33873582e-01 -6.68210030e-01 1.99124062e+00 5.86215556e-01 -1.32677525e-01 3.70620549e-01 7.24234432e-02 1.25597191e+00 3.99988353e-01 -2.24977389e-01 6.31705970e-02 -2.45029598e-01 -6.84431121e-02 7.70763636e-01 8.92188668e-01 -9.10922289e-01 5.67871392e-01 5.97157717e+00 1.11394572e+00 -3.93916488e-01 5.56468487e-01 6.04594529e-01 2.06984039e-02 -6.05178535e-01 1.80398956e-01 -8.83753002e-01 4.97578710e-01 5.91714561e-01 -2.42169037e-01 6.40511572e-01 6.95401609e-01 1.50239155e-01 -7.61179149e-01 -1.04392183e+00 9.56952989e-01 3.92783970e-01 -9.84958827e-01 -2.64695734e-01 3.49146426e-01 1.24961841e+00 -8.78380146e-03 -3.46800387e-01 -5.45574315e-02 9.25952196e-01 -9.56017196e-01 6.71442330e-01 7.46865749e-01 4.04050261e-01 -2.10292101e-01 8.37237954e-01 5.97628176e-01 -1.30375540e+00 3.46492469e-01 -3.58747303e-01 -3.98733318e-01 4.43324327e-01 1.41131520e+00 -9.39703941e-01 5.60496986e-01 3.24044019e-01 6.16869479e-02 -6.85898960e-01 9.27516162e-01 -5.21763921e-01 5.59126854e-01 -5.31647205e-01 -5.55547655e-01 -1.19551480e-01 -2.29036674e-01 7.68113136e-01 9.52964723e-01 6.31930083e-02 7.94536322e-02 4.20292467e-01 5.43617666e-01 -5.18306255e-01 -2.04013847e-02 -7.15974629e-01 1.57046467e-01 2.88055599e-01 1.52245760e+00 -9.83847558e-01 -3.91242772e-01 -2.15996712e-01 9.19058859e-01 5.00878513e-01 4.40809458e-01 -6.66673660e-01 -1.15398072e-01 1.45643249e-01 3.88354599e-01 1.58781379e-01 -3.91523123e-01 -3.95846337e-01 -1.07342315e+00 4.50747371e-01 -6.61787212e-01 2.65968680e-01 -5.49460709e-01 -1.65686655e+00 5.28534830e-01 -1.87785570e-02 -8.97302687e-01 -3.78947973e-01 -6.17804945e-01 -4.98031527e-01 9.55269992e-01 -1.20857632e+00 -7.58512020e-01 -6.45255744e-01 2.33048096e-01 1.81498826e-01 3.21049988e-01 3.39401543e-01 4.98831533e-02 -1.91123486e-01 2.86411256e-01 -2.26565800e-03 2.21691504e-02 5.97725630e-01 -1.66890645e+00 6.32041872e-01 1.06185293e+00 3.76950324e-01 1.98093072e-01 9.96672750e-01 -8.89967442e-01 -1.35042489e+00 -9.87049460e-01 1.19552994e+00 -1.36035502e+00 8.70369613e-01 -7.07637012e-01 -6.29042566e-01 4.74911034e-01 3.76517028e-02 3.52283508e-01 5.41985452e-01 -1.53947234e-01 2.16922183e-02 2.23114654e-01 -1.02890003e+00 5.50606489e-01 1.04039383e+00 -8.76737654e-01 3.30955051e-02 7.45956838e-01 4.29872572e-01 -4.76900965e-01 -7.28154123e-01 -2.18328759e-01 2.19380438e-01 -9.76803303e-01 4.47896212e-01 -4.93440032e-01 3.32989134e-02 -5.18409252e-01 -2.37440377e-01 -1.21824765e+00 1.61082283e-01 -1.19903731e+00 -2.22293928e-01 1.07953763e+00 3.32865119e-01 -6.63366258e-01 9.44628954e-01 9.01543379e-01 5.20922661e-01 -4.69380170e-01 -1.16465783e+00 -5.14423072e-01 -1.76454112e-01 -5.56272388e-01 3.30999076e-01 5.55229247e-01 -1.85634583e-01 5.03514826e-01 -4.78987426e-01 3.73075724e-01 8.81400704e-01 -2.42213219e-01 1.17099535e+00 -1.36727941e+00 -4.74071681e-01 1.06190614e-01 -4.93508935e-01 -1.22453845e+00 3.02979112e-01 -7.97140718e-01 2.95115322e-01 -1.74520743e+00 2.31230304e-01 6.97309598e-02 4.15026575e-01 -4.73359637e-02 -6.29608631e-01 5.91622412e-01 1.63760364e-01 1.02376133e-01 -1.32139051e+00 2.48665735e-01 1.11068952e+00 -1.07486285e-01 3.19777369e-01 1.49638399e-01 -9.15501118e-01 7.22033858e-01 4.72448379e-01 -8.43742669e-01 -2.73433626e-01 -4.62047011e-01 1.06852615e+00 1.09670006e-01 6.01445198e-01 -7.00173855e-01 7.82205880e-01 9.25415680e-02 -3.36854495e-02 -1.38552487e-01 2.11330637e-01 -5.63075244e-01 5.76281501e-03 -1.17672645e-01 -3.02385420e-01 1.16920389e-01 -3.72779846e-01 1.37883627e+00 -9.38556064e-03 -5.06570220e-01 2.71609187e-01 -3.15959215e-01 -3.28299910e-01 1.98266685e-01 -4.39751834e-01 7.34639585e-01 8.24318945e-01 -1.93993688e-01 -4.02632862e-01 -9.79829431e-01 -8.63641381e-01 2.71901488e-01 4.33007717e-01 2.81963609e-02 4.98701371e-02 -1.04310584e+00 -1.00633562e+00 -6.27441108e-01 3.21486406e-02 1.07348420e-01 1.89682618e-01 1.17671633e+00 -3.21963280e-01 1.77742317e-02 5.73206782e-01 -5.71861982e-01 -9.74113524e-01 2.21952662e-01 2.13379264e-01 -5.78524172e-01 1.83990851e-01 8.61192882e-01 -2.34448418e-01 -3.87370884e-01 -3.87409609e-03 -3.35690603e-02 2.46579293e-02 3.36089730e-01 4.43543404e-01 9.92476583e-01 3.44146669e-01 -6.19314909e-01 -3.60576391e-01 2.84927636e-01 2.61298448e-01 -3.83971125e-01 9.02229786e-01 -4.25469488e-01 -3.63966674e-01 4.40221936e-01 9.03126538e-01 6.49618149e-01 -1.06851172e+00 -3.64129215e-01 1.56622723e-01 -3.55179787e-01 -3.67540151e-01 -1.69070199e-01 -6.97553456e-01 5.07082939e-01 1.77088156e-01 1.13716900e+00 5.04913449e-01 3.27257603e-01 5.15063405e-01 3.67923021e-01 6.06684983e-01 -1.15093899e+00 -1.20329231e-01 2.05039695e-01 7.31885254e-01 -1.51425481e+00 2.36233234e-01 -7.06835449e-01 -6.01983964e-01 7.79334426e-01 1.51370898e-01 -4.19303209e-01 6.47259951e-01 2.33440667e-01 -1.17509976e-01 -4.38007712e-01 -3.73853236e-01 -6.02017581e-01 4.76021349e-01 6.36214793e-01 2.50268012e-01 2.57742733e-01 -3.15581620e-01 6.24224961e-01 -9.81560126e-02 3.17637697e-02 8.64402056e-01 6.27371609e-01 -4.72219676e-01 -9.49073255e-01 -5.29774070e-01 6.55951977e-01 -5.84950149e-01 -6.04920276e-02 -9.83048439e-01 5.11450827e-01 1.18607273e-02 1.73228562e+00 -3.56995583e-01 -2.11136326e-01 2.46959612e-01 -1.94995850e-01 4.44572538e-01 -6.82238817e-01 -4.50834960e-01 -4.00602907e-01 4.97726589e-01 -3.30932409e-01 -7.70426273e-01 -5.30847788e-01 -6.69687927e-01 -5.51423311e-01 -6.07977688e-01 3.22834104e-01 3.74769747e-01 1.33947420e+00 2.22748473e-01 1.98061243e-01 3.30188721e-01 -1.06477988e+00 -8.71553481e-01 -9.33092177e-01 -8.78138363e-01 2.52510279e-01 1.27193421e-01 -5.31689167e-01 -6.98700786e-01 -1.93199322e-01]
[9.632292747497559, 4.688546657562256]
fbd4b0d7-7ac8-4a01-85ea-4a80b2e7449f
a-symmetric-local-search-network-for-emotion
null
null
https://aclanthology.org/2020.coling-main.12
https://aclanthology.org/2020.coling-main.12.pdf
A Symmetric Local Search Network for Emotion-Cause Pair Extraction
Emotion-cause pair extraction (ECPE) is a new task which aims at extracting the potential clause pairs of emotions and corresponding causes in a document. To tackle this task, a two-step method was proposed by previous study which first extracted emotion clauses and cause clauses individually, then paired the emotion and cause clauses, and filtered out the pairs without causality. Different from this method that separated the detection and the matching of emotion and cause into two steps, we propose a Symmetric Local Search Network (SLSN) model to perform the detection and matching simultaneously by local search. SLSN consists of two symmetric subnetworks, namely the emotion subnetwork and the cause subnetwork. Each subnetwork is composed of a clause representation learner and a local pair searcher. The local pair searcher is a specially-designed cross-subnetwork component which can extract the local emotion-cause pairs. Experimental results on the ECPE corpus demonstrate the superiority of our SLSN over existing state-of-the-art methods.
['Qing Gu', 'Hua Yu', 'Yafeng Yin', 'Zhiwei Jiang', 'Zifeng Cheng']
2020-12-01
null
null
null
coling-2020-8
['emotion-cause-pair-extraction']
['natural-language-processing']
[ 3.43638033e-01 3.80154550e-01 -2.03183591e-01 -4.34772998e-01 -7.70963430e-01 -4.53386158e-01 5.12559235e-01 4.05824445e-02 -2.02040270e-01 6.24846399e-01 1.26909971e-01 2.01565381e-02 -2.30098516e-01 -8.11159670e-01 -2.74491996e-01 -5.69876909e-01 -9.59380791e-02 3.89473587e-01 2.27640092e-01 3.00767012e-02 -2.66704168e-02 1.38554350e-01 -1.66202927e+00 5.12452245e-01 9.79056954e-01 1.07540464e+00 -1.74319278e-02 3.51788759e-01 -5.12860715e-01 1.24716723e+00 -6.25065565e-01 -3.54885638e-01 -3.37989986e-01 -8.65069747e-01 -9.44085956e-01 -1.94744378e-01 -4.88646924e-01 2.31533453e-01 2.46534556e-01 1.15177071e+00 4.49858069e-01 -2.62420196e-02 4.75408494e-01 -1.78258169e+00 -1.91616938e-01 9.14462447e-01 -6.48514092e-01 -7.83592165e-02 6.30465508e-01 -4.28100854e-01 1.49706924e+00 -6.87160909e-01 6.79258585e-01 1.37474418e+00 4.20424879e-01 5.94998598e-01 -7.46183753e-01 -1.23154318e+00 6.20336115e-01 3.61983180e-01 -1.30241251e+00 -7.48721063e-02 1.21381056e+00 -2.54831426e-02 1.20282173e+00 3.88275772e-01 6.47267044e-01 9.28169012e-01 9.58090723e-02 1.24346387e+00 8.48063588e-01 -3.37600350e-01 1.73896089e-01 -2.47795358e-02 4.74249482e-01 4.88237828e-01 -2.27472439e-01 -3.03574771e-01 -4.54513669e-01 -3.44087839e-01 4.19374943e-01 -3.37949574e-01 -5.38761199e-01 -9.94769782e-02 -1.02752924e+00 8.51576090e-01 2.61772633e-01 5.34648061e-01 -4.66289103e-01 -2.03222744e-02 5.71210861e-01 2.27960631e-01 5.80508709e-01 3.50691825e-01 -8.93920600e-01 1.51932123e-03 -8.13601077e-01 2.25905403e-01 1.13781238e+00 9.13641095e-01 5.52086174e-01 -6.09113157e-01 -1.60142958e-01 7.39451170e-01 1.60052702e-01 7.37239346e-02 4.67234880e-01 -3.33289564e-01 5.26930392e-01 1.14474821e+00 -2.19218284e-01 -1.46305096e+00 -5.04549265e-01 -4.25026596e-01 -9.71479297e-01 -5.66127598e-01 -1.70327231e-01 -5.03295600e-01 -5.35937965e-01 1.94632995e+00 6.11136317e-01 3.69539827e-01 2.18112752e-01 9.62251306e-01 1.34531569e+00 7.91647851e-01 1.90839246e-02 -5.78009605e-01 1.48972201e+00 -1.06596434e+00 -1.13344765e+00 -3.70534956e-01 5.90557277e-01 -7.53330112e-01 6.08600318e-01 3.99844617e-01 -9.81544435e-01 -7.80177861e-02 -9.67390120e-01 -9.83326808e-02 -3.65133137e-01 1.30583718e-01 8.53615344e-01 -7.22874999e-02 -5.84544837e-01 3.40936840e-01 -3.92857254e-01 -7.75423348e-02 1.78252786e-01 4.79005128e-01 -2.26921558e-01 1.50904939e-01 -1.68410945e+00 4.99948055e-01 6.33225620e-01 2.44099766e-01 -3.14504772e-01 -5.41501820e-01 -1.06267381e+00 4.48408633e-01 8.24892521e-01 -5.83408654e-01 1.27794158e+00 -1.38872123e+00 -1.32290280e+00 7.09088564e-01 -5.85627794e-01 -1.33335823e-02 -1.46219924e-01 -1.52603358e-01 -7.36545265e-01 6.76208138e-02 2.60377020e-01 4.67738718e-01 5.72530687e-01 -1.31104624e+00 -8.89398158e-01 -1.54996321e-01 -1.98033586e-01 2.08842427e-01 -1.87149972e-01 4.78838146e-01 -9.44817662e-01 -4.64189172e-01 3.34546030e-01 -4.34561849e-01 -5.53536869e-04 -5.65571427e-01 -6.80158854e-01 -9.00072217e-01 9.46331441e-01 -4.64202076e-01 1.55079639e+00 -2.08534265e+00 2.38956988e-01 4.38930809e-01 4.93747979e-01 -7.51357153e-02 -3.49331081e-01 2.92522073e-01 -7.51839757e-01 1.33214802e-01 -4.31200378e-02 -3.24910402e-01 2.35536844e-01 2.55927473e-01 -2.86257267e-01 9.75776836e-02 5.62119365e-01 7.81015038e-01 -1.18028724e+00 -9.46471095e-01 -3.09406847e-01 3.58227849e-01 -4.53440398e-01 4.96090204e-01 -4.41344678e-01 -1.53620943e-01 -5.41379154e-01 7.16273069e-01 7.63656974e-01 -2.61229783e-01 4.25554305e-01 -2.42892429e-01 3.09915319e-02 6.92544878e-01 -1.40314448e+00 1.08488321e+00 -2.04250023e-01 1.48919940e-01 3.68717223e-01 -8.73990476e-01 1.08720839e+00 5.81762552e-01 4.40442562e-01 -6.72073960e-01 4.84515965e-01 1.28905833e-01 -8.29093829e-02 -6.73198402e-01 9.23723131e-02 -3.10532957e-01 -4.77412373e-01 4.14803237e-01 2.13549048e-01 5.73317111e-02 3.87127459e-01 3.17164570e-01 1.25020730e+00 -1.41552612e-01 4.36420619e-01 -3.68472487e-02 8.74537289e-01 -1.15725901e-02 1.31741405e+00 2.10927486e-01 -1.99098498e-01 3.19676369e-01 1.29484940e+00 -2.55054981e-01 -1.90094694e-01 -6.00408018e-01 3.90003711e-01 8.49941432e-01 3.78491253e-01 -9.06050086e-01 -5.01896560e-01 -1.06985569e+00 -3.49296719e-01 6.89223766e-01 -6.00173950e-01 -1.57172054e-01 -6.85053051e-01 -6.09317601e-01 4.61617351e-01 4.85940188e-01 6.78540289e-01 -1.32563639e+00 -3.79620880e-01 2.34576464e-01 -7.47025847e-01 -1.04899037e+00 -3.36321503e-01 6.36967301e-01 -2.38713190e-01 -1.40264964e+00 -9.00740549e-02 -1.08519602e+00 5.88469982e-01 -2.17348367e-01 1.30981755e+00 1.86006173e-01 -5.51663600e-02 -6.65229484e-02 -5.33835113e-01 -5.61776876e-01 -2.62809591e-03 1.68338269e-01 -4.59362000e-01 9.39819142e-02 7.28954434e-01 -5.41859269e-01 -2.71346778e-01 1.30153894e-01 -7.33595431e-01 2.45251358e-01 7.61770248e-01 7.17050493e-01 7.54088879e-01 5.91160238e-01 4.90850896e-01 -9.16255474e-01 8.68486047e-01 -7.15036333e-01 -4.15473163e-01 3.12054038e-01 -4.63198572e-01 -5.54423779e-02 6.64111614e-01 -3.59287113e-01 -1.01624215e+00 1.65328860e-01 -7.20835179e-02 -3.23929489e-01 -2.97484845e-01 8.37941945e-01 -7.52747059e-01 6.09606743e-01 -2.75881797e-01 1.51219591e-01 -3.34250182e-01 -3.60735416e-01 1.74559623e-01 4.97938603e-01 6.99713111e-01 -3.94931376e-01 6.18929446e-01 4.30771187e-02 -1.51202679e-01 -1.83095887e-01 -9.47632611e-01 -7.63123691e-01 -4.20428425e-01 -1.82369187e-01 1.02833664e+00 -8.45056951e-01 -8.93531024e-01 1.35558471e-01 -1.52231789e+00 6.73395917e-02 8.15022811e-02 2.47707129e-01 -1.18375637e-01 7.45884851e-02 -6.27451122e-01 -8.77468050e-01 -5.06508172e-01 -6.72293603e-01 1.17615390e+00 3.89378935e-01 -6.26027048e-01 -8.07982445e-01 9.24335569e-02 4.67999512e-03 -1.38351753e-01 3.55035037e-01 1.17570543e+00 -9.81610715e-01 -2.73256898e-01 -2.30311647e-01 -3.19628805e-01 -6.70872778e-02 1.81709945e-01 9.26148742e-02 -7.70094275e-01 2.04388216e-01 2.52562910e-01 -3.10238391e-01 7.24568665e-01 -1.81695335e-02 8.88862550e-01 -5.58775306e-01 -6.21990740e-01 3.36702436e-01 1.32263756e+00 3.52820694e-01 4.54835981e-01 2.31600553e-01 5.21415532e-01 7.34725177e-01 7.99452066e-01 2.26940691e-01 4.63531256e-01 3.00499052e-01 4.26099479e-01 -5.42486072e-01 2.17700422e-01 -2.00689331e-01 3.12002808e-01 1.01688814e+00 1.83118433e-01 -3.80667090e-01 -6.81333840e-01 6.27429724e-01 -2.12933636e+00 -1.06982696e+00 -3.80404770e-01 1.34096777e+00 1.26177037e+00 1.90192789e-01 -4.15385328e-02 4.90685821e-01 7.32996643e-01 1.97819605e-01 -2.44589448e-01 -4.90571380e-01 -1.31667107e-01 8.52669477e-02 -2.05878168e-01 3.23750973e-01 -1.19881320e+00 8.65343511e-01 5.20239449e+00 1.13813531e+00 -8.59782994e-01 -1.24860629e-01 5.43013632e-01 -4.01929356e-02 -3.89694124e-01 1.12818591e-02 -6.88059032e-01 2.06139520e-01 3.78253967e-01 3.94352078e-02 1.22742228e-01 1.03001964e+00 8.00015107e-02 -8.74739960e-02 -1.14333725e+00 8.53653491e-01 1.43297270e-01 -1.00085247e+00 -1.45608455e-01 -4.46026653e-01 5.02437174e-01 -3.89169335e-01 -5.36215663e-01 4.63930309e-01 1.55898809e-01 -8.84916544e-01 6.13911390e-01 3.93550515e-01 3.57230604e-01 -1.27976453e+00 1.26651311e+00 3.81465733e-01 -1.72214448e+00 1.63758695e-02 -2.66564358e-02 -1.83140516e-01 8.50336701e-02 1.01479661e+00 -3.24113250e-01 8.49218667e-01 8.32267523e-01 8.36199820e-01 -2.59433627e-01 8.03764284e-01 -8.56065929e-01 6.00167036e-01 -2.77649343e-01 -3.68374914e-01 2.70337313e-01 -9.59950238e-02 5.63362241e-01 1.44100535e+00 -8.61667842e-02 4.14225727e-01 1.84028581e-01 1.23493564e+00 -1.29486427e-01 2.52218336e-01 -1.51059940e-01 9.50294361e-02 4.98218268e-01 1.77178848e+00 -9.52876866e-01 -3.59923631e-01 -2.89420813e-01 1.07796121e+00 4.24648196e-01 1.96887583e-01 -1.02105868e+00 -1.06903303e+00 5.49190223e-01 -4.92513746e-01 5.39623082e-01 3.93462628e-01 -2.26971284e-01 -1.12841189e+00 2.60586858e-01 -9.97997224e-01 6.33250892e-01 -8.74544024e-01 -1.35494316e+00 7.43089795e-01 -1.52698001e-02 -8.62630427e-01 -2.54196346e-01 -3.04455876e-01 -1.25334811e+00 8.44785452e-01 -1.58200216e+00 -1.11076844e+00 -1.69288158e-01 7.94756174e-01 3.67023885e-01 4.11771476e-01 7.80398965e-01 1.72999993e-01 -1.04236090e+00 4.38810259e-01 -7.07731903e-01 4.40530598e-01 5.35488665e-01 -1.26622021e+00 -2.97688335e-01 8.72337699e-01 -2.10245132e-01 7.82475352e-01 4.85652655e-01 -8.73886645e-01 -1.11742866e+00 -9.71097052e-01 1.85957599e+00 7.37623572e-02 6.91347539e-01 -7.22761035e-01 -7.81960845e-01 4.91534442e-01 4.90811497e-01 -3.02323431e-01 6.84030116e-01 3.25629562e-01 -2.39103690e-01 -1.37194663e-01 -9.24122155e-01 6.05318069e-01 8.71702075e-01 -3.27534348e-01 -9.46773350e-01 2.21368462e-01 9.56924260e-01 -2.60312706e-01 -4.20241773e-01 5.65363646e-01 1.20478235e-01 -9.28568959e-01 6.57636166e-01 -3.28226417e-01 8.72511446e-01 -3.18225324e-01 2.34335229e-01 -1.06521976e+00 -2.35662550e-01 -7.90492952e-01 -2.50306353e-02 2.06483817e+00 6.11366987e-01 -3.48887295e-01 4.55670506e-01 6.38425946e-01 -1.17360316e-01 -1.02877748e+00 -7.87018776e-01 -5.75113535e-01 -3.54188681e-01 -5.76892555e-01 9.46799874e-01 1.00169325e+00 5.25644004e-01 9.68111515e-01 1.03893066e-02 2.59181440e-01 2.29778141e-01 7.64383793e-01 3.54363948e-01 -1.24182320e+00 -3.09923768e-01 -6.29833937e-01 1.36264160e-01 -7.24665642e-01 5.58194578e-01 -9.06858027e-01 4.51054722e-01 -1.59922409e+00 4.34720576e-01 -1.84507295e-02 -3.10167491e-01 8.37060869e-01 -5.85132957e-01 -4.00006890e-01 5.22139110e-02 -1.66202281e-02 -9.19369340e-01 7.28799820e-01 1.10188186e+00 1.17285689e-02 -3.18411618e-01 -1.85465112e-01 -8.93391907e-01 1.03407836e+00 6.80165470e-01 -7.88042307e-01 -4.73922640e-01 5.53575940e-02 6.11701369e-01 -4.47927490e-02 2.94817388e-01 -6.39485121e-01 5.24412632e-01 -2.46306568e-01 2.67336100e-01 -9.78774250e-01 -1.69490770e-01 -8.41897428e-01 -2.72511452e-01 2.77269810e-01 -2.31203824e-01 1.30735949e-01 1.66601241e-01 2.46670470e-01 -6.79055572e-01 5.94553798e-02 3.50933313e-01 3.41090560e-02 -5.64443409e-01 -3.68677080e-02 -3.15028667e-01 8.51067826e-02 9.66040730e-01 2.85566837e-01 -2.81232208e-01 -2.95483530e-01 -4.58835453e-01 6.96033537e-01 -2.99250454e-01 2.94262201e-01 7.16279387e-01 -1.32991409e+00 -4.28606272e-01 9.64673534e-02 -8.49680975e-02 3.71563137e-01 -1.97611317e-01 9.47370768e-01 2.09971875e-01 3.99402231e-01 4.15585995e-01 -4.24699672e-02 -1.68102407e+00 6.36887372e-01 2.88947552e-01 -8.49540949e-01 -2.95233786e-01 1.25410163e+00 1.78504437e-01 -5.78540087e-01 3.38619649e-01 -4.79337603e-01 -6.08136833e-01 2.52458990e-01 4.37691569e-01 7.24521792e-03 -1.94187626e-01 -4.24938947e-01 -7.96037138e-01 5.84989846e-01 9.53130648e-02 3.90366502e-02 1.46577704e+00 -4.17430326e-02 -9.71399426e-01 3.85554373e-01 1.36203063e+00 -5.37371635e-03 -4.87615645e-01 -1.05754659e-01 8.71068388e-02 1.69723630e-01 2.44886130e-02 -1.01334012e+00 -1.28721762e+00 4.41481084e-01 -1.98732883e-01 2.45508209e-01 1.70071375e+00 3.86967450e-01 9.49486315e-01 7.12787434e-02 -6.71178550e-02 -1.17995441e+00 -3.91454697e-02 8.18908989e-01 1.01499975e+00 -9.40114677e-01 -1.76080987e-01 -1.10699022e+00 -3.80751789e-01 1.06247318e+00 9.80747402e-01 -5.27515784e-02 6.93177581e-01 6.30099595e-01 1.87892523e-02 -6.85127199e-01 -1.19757748e+00 -3.41026455e-01 5.77012002e-01 2.60665417e-01 5.33087015e-01 -1.88751727e-01 -7.54527450e-01 1.87088346e+00 -2.65500665e-01 1.40561368e-02 -1.38913080e-01 1.02144682e+00 -1.22525349e-01 -1.18454504e+00 -3.17939222e-01 1.54539853e-01 -3.75029832e-01 -3.51244479e-01 -1.17071700e+00 9.68730450e-01 6.89259827e-01 1.31378174e+00 1.89167604e-01 -7.53458440e-01 2.56577551e-01 2.54187793e-01 -1.23700827e-01 -4.62914824e-01 -1.04547894e+00 5.37976801e-01 4.09270883e-01 -8.92659485e-01 -5.91896832e-01 -5.46211064e-01 -1.82763886e+00 9.80810523e-02 -6.39211953e-01 4.24408704e-01 2.15788797e-01 1.27931571e+00 2.96210766e-01 9.76775050e-01 6.68448329e-01 -4.34819937e-01 9.43035483e-02 -7.04411805e-01 -6.02998197e-01 4.10978466e-01 1.40480682e-01 -3.28534991e-01 -5.57585061e-01 -1.56635523e-01]
[12.626951217651367, 6.211202144622803]
f04509c8-fcc6-4217-a94a-e51aa088c49e
clone-seeker-effective-code-clone-search
2106.03042
null
https://arxiv.org/abs/2106.03042v1
https://arxiv.org/pdf/2106.03042v1.pdf
Clone-Seeker: Effective Code Clone Search Using Annotations
Source code search plays an important role in software development, e.g. for exploratory development or opportunistic reuse of existing code from a code base. Often, exploration of different implementations with the same functionality is needed for tasks like automated software transplantation, software diversification, and software repair. Code clones, which are syntactically or semantically similar code fragments, are perfect candidates for such tasks. Searching for code clones involves a given search query to retrieve the relevant code fragments. We propose a novel approach called Clone-Seeker that focuses on utilizing clone class features in retrieving code clones. For this purpose, we generate metadata for each code clone in the form of a natural language document. The metadata includes a pre-processed list of identifiers from the code clones augmented with a list of keywords indicating the semantics of the code clone. This keyword list can be extracted from a manually annotated general description of the clone class, or automatically generated from the source code of the entire clone class. This approach helps developers to perform code clone search based on a search query written either as source code terms, or as natural language. In our quantitative evaluation, we show that (1) Clone-Seeker has a higher recall when searching for semantic code clones (i.e., Type-4) in BigCloneBench than the state-of-the-art; and (2) Clone-Seeker can accurately search for relevant code clones by applying natural language queries.
['Mark van den Brand', 'Hamid Abdul Basit', 'Önder Babur', 'Muhammad Hammad']
2021-06-06
null
null
null
null
['code-search', 'code-search']
['computer-code', 'computer-vision']
[-1.38894394e-01 -1.15835935e-01 -6.06806576e-01 -7.22127780e-02 -8.82235587e-01 -7.87250638e-01 3.15099746e-01 6.31871104e-01 1.26322821e-01 2.62155980e-02 1.55773699e-01 -4.93377745e-01 -2.08029114e-02 -6.53406799e-01 -5.24746001e-01 -4.05038744e-02 1.58645764e-01 3.72786298e-02 6.93666518e-01 -1.24303401e-01 9.05315876e-01 3.63424681e-02 -1.95738518e+00 3.11525077e-01 1.22763133e+00 7.08097279e-01 1.05281806e+00 1.89041540e-01 -9.82991278e-01 6.64360702e-01 -6.24064922e-01 -4.96573120e-01 -1.69363871e-01 -6.87170625e-01 -1.02309453e+00 -1.69621214e-01 8.67381766e-02 3.85577619e-01 3.81865114e-01 1.49536467e+00 -4.74536531e-02 -4.20024395e-01 2.08296046e-01 -1.26102340e+00 -3.57345432e-01 8.55541289e-01 -5.38932979e-01 1.25642315e-01 8.12418044e-01 -3.05337399e-01 1.00278831e+00 -1.09834480e+00 1.06824958e+00 8.69481564e-01 4.84253466e-01 3.86533320e-01 -9.48794901e-01 -4.97311562e-01 -1.83766186e-01 -4.53790575e-02 -1.64923763e+00 -7.76623935e-02 6.29864573e-01 -8.82668138e-01 1.12655222e+00 4.18845266e-01 4.98451144e-01 5.59556186e-01 4.99538213e-01 5.35025656e-01 1.95038721e-01 -6.87002361e-01 3.72277230e-01 5.02648652e-01 3.02498907e-01 8.10156822e-01 4.15200561e-01 -2.30203882e-01 -2.89922416e-01 -9.29720879e-01 3.43507938e-02 7.38710687e-02 -5.09796679e-01 -5.34837663e-01 -8.77744555e-01 7.61002600e-01 1.70867547e-01 6.86969399e-01 -1.20784221e-02 2.34841213e-01 4.73587722e-01 5.29502153e-01 5.16707264e-02 8.07932794e-01 -5.37625849e-01 -3.47733051e-01 -8.31913471e-01 1.79114804e-01 1.03264785e+00 1.74361074e+00 1.26600826e+00 -4.81104881e-01 1.41276732e-01 8.77436161e-01 5.35427690e-01 2.20924094e-01 9.24639761e-01 -6.10359430e-01 4.31322843e-01 1.40525436e+00 -3.48930955e-02 -8.49684119e-01 -2.53389310e-02 -5.13245821e-01 -1.59785561e-02 1.33289218e-01 -2.95329601e-01 4.00466174e-01 -2.53515899e-01 1.36436558e+00 2.89189994e-01 -5.85642338e-01 1.20120347e-02 2.11291939e-01 6.54865384e-01 3.05478573e-01 -3.22371870e-01 2.16019619e-02 1.43571770e+00 -9.11020160e-01 -2.42508203e-01 -4.36453700e-01 1.10876167e+00 -1.00333536e+00 8.50547791e-01 1.34213939e-01 -6.26644015e-01 -2.78728604e-01 -9.25909460e-01 1.75413534e-01 -4.86968935e-01 3.58984232e-01 2.34870493e-01 6.62699163e-01 -9.64007020e-01 3.04479480e-01 -5.77135682e-01 -6.80677235e-01 4.04505357e-02 -1.52322561e-01 -2.37450361e-01 -4.67201602e-03 -4.70312297e-01 5.09884834e-01 5.30589163e-01 -7.06581593e-01 -7.84561515e-01 -6.27448201e-01 -1.08397198e+00 4.15562361e-01 6.12715781e-01 -4.63561863e-01 1.32090783e+00 -1.27361000e+00 -5.98995149e-01 7.67417848e-01 -1.83868527e-01 -8.39879084e-03 -2.32745290e-01 -7.73358718e-02 -5.87724388e-01 -1.06723189e-01 6.42971992e-01 1.64235026e-01 7.17595100e-01 -1.37862480e+00 -1.05551958e+00 -1.01794086e-01 1.57106653e-01 -3.89953285e-01 -2.20883936e-01 5.73360980e-01 -7.17631638e-01 -8.66576731e-01 2.47205663e-02 -9.27115381e-01 -4.31662910e-02 9.24488157e-02 -2.76023179e-01 -1.55831248e-01 7.05824196e-01 -5.99299490e-01 2.01840019e+00 -2.41665101e+00 9.63323787e-02 2.92822301e-01 2.51167685e-01 -7.79941380e-02 -1.46843508e-01 8.22798908e-01 -1.79961368e-01 4.98887330e-01 -4.75918055e-01 2.83318639e-01 -1.19861484e-01 -2.89534420e-01 -2.03223258e-01 4.38605361e-02 -6.21422753e-03 7.47970700e-01 -1.00285625e+00 -3.87357920e-01 -6.23694897e-01 -1.22509070e-01 -7.29752362e-01 6.67937025e-02 -6.58623338e-01 -5.21087833e-02 -8.36835861e-01 8.76402736e-01 2.05586582e-01 -3.55773687e-01 6.76795393e-02 3.67810577e-01 -3.89741093e-01 3.30780596e-01 -9.51417685e-01 1.93769383e+00 -7.21806228e-01 5.25649250e-01 -1.64543182e-01 -3.83115411e-01 1.14875996e+00 2.16475442e-01 9.34742317e-02 -4.36397076e-01 -2.95236647e-01 7.92341530e-01 -2.15762109e-01 -9.10446227e-01 4.50946331e-01 4.54427212e-01 -5.40312469e-01 9.70794261e-01 -1.12910487e-01 -1.69126049e-01 4.30742353e-01 2.72464991e-01 1.47192240e+00 5.26023246e-02 7.61469126e-01 -5.89064419e-01 8.24690461e-01 4.24853295e-01 3.96592051e-01 8.12504053e-01 5.69443285e-01 2.02417418e-01 6.08959198e-01 -2.29980260e-01 -7.76686311e-01 -6.42169714e-01 -6.17039874e-02 1.01177156e+00 3.52914631e-01 -1.22902632e+00 -6.99404180e-01 -8.46632242e-01 3.00439652e-02 8.53492737e-01 -4.12727237e-01 -4.98782873e-01 -6.46682382e-01 4.76951934e-02 5.21988392e-01 5.37548900e-01 5.44781573e-02 -9.93521273e-01 -1.23302376e+00 1.90537453e-01 -9.06089321e-02 -2.59160727e-01 -8.90597761e-01 6.60398602e-02 -6.67826593e-01 -1.42977571e+00 -4.42427367e-01 -9.73411679e-01 9.13770020e-01 2.50955164e-01 1.37568724e+00 8.04959476e-01 -4.51398164e-01 3.49007040e-01 -7.85933554e-01 5.55530749e-02 -1.14143836e+00 2.36058325e-01 -7.52791762e-01 -5.36512434e-01 4.62283432e-01 -3.99445415e-01 -2.32362613e-01 5.53495944e-01 -1.16803110e+00 -2.00895816e-01 5.35628140e-01 7.11621344e-01 3.56434286e-01 2.86344141e-02 2.96499074e-01 -7.15184808e-01 7.64464736e-01 -8.82265747e-01 -1.02155268e+00 7.28156388e-01 -7.33221889e-01 5.21920502e-01 3.73273343e-01 -3.31360966e-01 -9.75222290e-01 1.20718956e-01 -1.95014309e-02 -1.99687295e-02 1.86819509e-01 1.27432430e+00 1.88017450e-02 -1.47982389e-01 9.43302572e-01 5.61086535e-01 -2.86556393e-01 -9.49163616e-01 1.62027538e-01 1.00568628e+00 3.19190890e-01 -7.77199149e-01 6.75125599e-01 5.18721715e-03 -5.41015327e-01 -3.10916454e-01 -3.90743166e-02 -9.33770955e-01 -4.81014580e-01 3.26213278e-02 4.13084358e-01 -5.92317343e-01 2.25414678e-01 1.05268434e-01 -1.22506940e+00 2.25248083e-01 -1.46118775e-01 4.18002233e-02 -2.82741606e-01 5.13294518e-01 -8.89195278e-02 -3.91450703e-01 -3.22095424e-01 -1.51140392e+00 1.32455623e+00 1.09601945e-01 -5.28154194e-01 -4.94719177e-01 4.21054959e-01 -1.85617343e-01 5.82532823e-01 -1.82089552e-01 1.74355662e+00 -8.39856446e-01 -8.75726223e-01 -4.46387053e-01 -1.38096148e-02 -3.06087673e-01 4.83098537e-01 1.87956318e-01 -4.13689047e-01 -3.26460689e-01 -8.56974944e-02 1.41018972e-01 3.54825914e-01 -4.48072165e-01 7.20167398e-01 -2.94461042e-01 -9.77880538e-01 2.86789000e-01 1.70827985e+00 6.59656405e-01 4.32932079e-01 5.47058046e-01 2.46214926e-01 7.08414018e-01 7.37860680e-01 6.78462803e-01 2.41522893e-01 8.00301194e-01 3.29717904e-01 7.03407347e-01 1.13883168e-01 -1.38922468e-01 2.13153869e-01 8.54467034e-01 6.71044111e-01 -4.62996922e-02 -1.49447727e+00 8.41764390e-01 -1.63964522e+00 -6.13985062e-01 -9.33761373e-02 2.44731069e+00 1.02141023e+00 -1.89463198e-01 -1.04895346e-01 -2.08825693e-01 7.77586520e-01 -4.46992010e-01 -3.59637409e-01 -1.42443866e-01 4.01087582e-01 -1.02370024e-01 2.80832965e-02 2.30047524e-01 -4.45551097e-01 5.74220955e-01 5.10048056e+00 8.63257170e-01 -8.14502537e-01 2.27447450e-01 -2.78021485e-01 3.37713867e-01 -8.27908635e-01 6.75561965e-01 -7.47524500e-01 5.49192309e-01 5.93385994e-01 -1.02883768e+00 2.25469619e-01 1.48522890e+00 -5.21117330e-01 -4.94980961e-01 -1.19570732e+00 6.55825078e-01 2.42079124e-01 -1.43455577e+00 -9.66125429e-02 -1.26847133e-01 7.64251888e-01 7.91110769e-02 -4.85976815e-01 2.41109356e-01 4.28456254e-02 -4.73898917e-01 1.09470582e+00 3.86278361e-01 6.96885049e-01 -5.32685935e-01 5.85433364e-01 4.64484721e-01 -1.63755810e+00 -2.62762398e-01 -3.56914282e-01 5.70455909e-01 -2.38126874e-01 4.24194098e-01 -5.05164504e-01 5.56974232e-01 8.68063450e-01 5.71397960e-01 -9.96372759e-01 1.70918310e+00 -5.54694086e-02 4.47452404e-02 3.91140161e-03 -2.68532634e-01 -4.74291295e-02 1.72649845e-01 6.18254006e-01 1.37656713e+00 8.51379633e-01 -5.75244248e-01 8.51607248e-02 1.37306440e+00 -4.15862836e-02 3.27586651e-01 -6.39072597e-01 -2.75806665e-01 7.96118736e-01 8.52186024e-01 -7.96230555e-01 -2.75532484e-01 -7.95848250e-01 6.72189176e-01 4.61454503e-02 2.39643976e-01 -5.01523376e-01 -1.00315821e+00 6.86905682e-01 9.27377492e-02 5.70993543e-01 1.35766298e-01 2.46417820e-01 -9.43332314e-01 6.32443368e-01 -1.01372576e+00 3.33395630e-01 -8.06316733e-01 -6.77698612e-01 1.00174296e+00 2.06904873e-01 -1.49450469e+00 -5.21414757e-01 -1.39048221e-02 -7.47827888e-01 6.74244523e-01 -9.67503011e-01 -4.46589440e-01 -4.94317144e-01 1.11407243e-01 7.08661199e-01 -4.03367996e-01 7.10089207e-01 3.42567712e-01 -2.67460555e-01 3.67500275e-01 1.82175428e-01 -9.65959281e-02 3.39845449e-01 -1.00420570e+00 7.58186340e-01 9.75508213e-01 1.33554891e-01 1.33659542e+00 3.79939735e-01 -1.08562744e+00 -1.20782137e+00 -1.04995418e+00 1.22554874e+00 -2.79479057e-01 7.59122431e-01 -4.04971272e-01 -1.15146017e+00 5.27389288e-01 8.31829682e-02 -2.41210565e-01 6.45643294e-01 -3.41198176e-01 -6.69647992e-01 2.07753554e-01 -8.95322919e-01 4.34264392e-01 9.31992054e-01 -7.43840754e-01 -6.01500630e-01 9.23515409e-02 8.27147722e-01 -2.17195973e-01 -7.67977476e-01 -6.86767325e-02 4.13134307e-01 -8.94514978e-01 4.71873671e-01 -2.08069414e-01 4.08657968e-01 -6.23112738e-01 -1.60821214e-01 -1.07731354e+00 5.86776733e-02 -5.68620086e-01 1.38251156e-01 1.41251493e+00 5.93826532e-01 -4.90292877e-01 3.24757427e-01 3.47818911e-01 -9.09346640e-02 -6.90371811e-01 -8.06941807e-01 -1.10898340e+00 -2.40182102e-01 -2.26228893e-01 1.02663493e+00 8.14040184e-01 5.51728606e-01 -5.88734692e-04 4.42939192e-01 -4.59960438e-02 7.92260617e-02 7.04727471e-01 6.07035577e-01 -1.33073831e+00 -4.09209549e-01 -8.21773231e-01 -5.81925154e-01 -9.38300014e-01 2.69466847e-01 -1.09091139e+00 3.58165741e-01 -1.22055888e+00 4.14445013e-01 -6.25727713e-01 3.68064165e-01 4.67494875e-01 -1.27620026e-01 -6.06197774e-01 -2.55741954e-01 6.76347435e-01 -3.54254395e-01 3.04427177e-01 4.08369422e-01 -3.56983185e-01 -1.56791344e-01 2.67054290e-01 -8.53400528e-01 5.31193316e-01 5.47839582e-01 -1.23142147e+00 -4.95308638e-01 -3.57504249e-01 8.71332407e-01 3.23371083e-01 1.68725058e-01 -8.60934854e-01 5.56504488e-01 -1.41678318e-01 -7.12072313e-01 -2.14677602e-01 -4.58622962e-01 -9.32621181e-01 6.95823789e-01 7.43390858e-01 -2.54501402e-01 4.54571992e-01 1.49351314e-01 4.94557530e-01 -4.39689070e-01 -1.42416704e+00 4.06318545e-01 -3.96423012e-01 -8.54654074e-01 -2.50912189e-01 -6.21343017e-01 3.27259719e-01 9.52759027e-01 -2.69393086e-01 -5.14935374e-01 2.67426223e-01 2.65171845e-02 6.95530549e-02 1.16443753e+00 1.03621185e+00 5.38372099e-01 -1.21790600e+00 -3.43990564e-01 2.68782318e-01 1.13570452e+00 -4.20030624e-01 -5.56147099e-01 5.53924978e-01 -6.58707440e-01 3.42707157e-01 1.11122213e-01 -4.68956023e-01 -1.30415356e+00 7.67336547e-01 2.34681949e-01 1.61811486e-01 -6.35558844e-01 7.47271776e-01 3.13510746e-01 -4.32501256e-01 1.12361066e-01 -6.19493186e-01 -6.14450388e-02 -2.83394217e-01 6.55789912e-01 2.32186452e-01 2.82561570e-01 -5.43455899e-01 -7.74044812e-01 8.69839191e-01 -1.03832418e-02 1.28064424e-01 9.73546565e-01 1.87456317e-03 -6.67154908e-01 2.94811487e-01 1.31957686e+00 4.75341380e-01 -5.27989328e-01 -2.47201219e-01 9.67868924e-01 -7.36469448e-01 -3.40730548e-01 -7.66221464e-01 -1.12539518e+00 2.19170973e-01 2.78498471e-01 1.79022342e-01 9.35401142e-01 7.66768694e-01 2.80013084e-01 5.41878760e-01 1.17113662e+00 -4.42427486e-01 -2.43616011e-02 1.88319385e-01 1.04541767e+00 -7.27021575e-01 -1.09375685e-01 -5.76852322e-01 -2.36509487e-01 1.34566414e+00 5.47364116e-01 5.05949676e-01 5.99073172e-01 4.45758015e-01 -5.03456220e-02 -6.72817886e-01 -7.92208493e-01 1.92901772e-02 2.92835802e-01 3.81591439e-01 5.03105640e-01 -5.45375526e-01 -5.55617690e-01 5.97523451e-01 6.74534738e-02 -1.33900359e-01 5.54932296e-01 1.54931343e+00 -8.17568779e-01 -1.53827405e+00 -2.54727840e-01 5.35715520e-01 -1.78699031e-01 -4.05095875e-01 -6.46772325e-01 5.29226959e-01 1.21051483e-01 8.87547135e-01 -1.59882352e-01 -4.32919502e-01 2.98000038e-01 1.91436820e-02 9.77415591e-02 -1.15372980e+00 -9.70147073e-01 -2.32866809e-01 -3.67640145e-02 -6.39908552e-01 1.34883272e-02 -6.45438194e-01 -1.18112206e+00 4.00209725e-01 -9.66180801e-01 6.11620963e-01 7.30827153e-01 7.93429077e-01 7.44638085e-01 2.02512771e-01 4.57505912e-01 -2.20979843e-02 -2.81561047e-01 -5.00365138e-01 -2.49638125e-01 1.53116569e-01 6.14135787e-02 -6.69691026e-01 -2.47292325e-01 3.08406800e-01]
[7.525604248046875, 8.140523910522461]
8146e6bf-12e0-47ef-a970-d84b759273e8
raat-relation-augmented-attention-transformer
2206.03377
null
https://arxiv.org/abs/2206.03377v1
https://arxiv.org/pdf/2206.03377v1.pdf
RAAT: Relation-Augmented Attention Transformer for Relation Modeling in Document-Level Event Extraction
In document-level event extraction (DEE) task, event arguments always scatter across sentences (across-sentence issue) and multiple events may lie in one document (multi-event issue). In this paper, we argue that the relation information of event arguments is of great significance for addressing the above two issues, and propose a new DEE framework which can model the relation dependencies, called Relation-augmented Document-level Event Extraction (ReDEE). More specifically, this framework features a novel and tailored transformer, named as Relation-augmented Attention Transformer (RAAT). RAAT is scalable to capture multi-scale and multi-amount argument relations. To further leverage relation information, we introduce a separate event relation prediction task and adopt multi-task learning method to explicitly enhance event extraction performance. Extensive experiments demonstrate the effectiveness of the proposed method, which can achieve state-of-the-art performance on two public datasets. Our code is available at https://github. com/TencentYoutuResearch/RAAT.
['Bo Ren', 'Di Yin', 'Zhuoxuan Jiang', 'Yuan Liang']
2022-06-07
null
https://aclanthology.org/2022.naacl-main.367
https://aclanthology.org/2022.naacl-main.367.pdf
naacl-2022-7
['document-level-event-extraction']
['natural-language-processing']
[ 1.29200518e-01 1.12319402e-01 -4.77080345e-01 -4.96155292e-01 -1.32321095e+00 -4.50414509e-01 8.25627744e-01 4.29534048e-01 -4.19103324e-01 8.39657903e-01 7.10078359e-01 -3.87863904e-01 -1.85924530e-01 -8.95447254e-01 -8.68819833e-01 -3.69321316e-01 1.81809384e-02 2.12329760e-01 3.66777837e-01 -1.31112352e-01 -1.72835514e-01 2.24899068e-01 -9.30214107e-01 8.97469521e-01 6.25596762e-01 1.03954649e+00 -1.07165821e-01 3.89578909e-01 -2.00072736e-01 1.44173360e+00 -5.47086835e-01 -7.05301166e-01 -2.42680773e-01 -2.93302119e-01 -1.09030950e+00 -3.16562593e-01 -4.08668816e-01 -2.85903633e-01 -4.83237028e-01 5.56085944e-01 4.46260631e-01 1.47696855e-02 4.50726509e-01 -1.35139012e+00 -5.52293777e-01 1.30458593e+00 -6.84931338e-01 7.89562523e-01 3.32802445e-01 -3.91747415e-01 1.54444218e+00 -1.18173635e+00 6.02733195e-01 1.23314095e+00 5.66928566e-01 -1.49097492e-03 -6.95605338e-01 -8.67782235e-01 5.79856634e-01 5.54159045e-01 -1.03735256e+00 -3.01646888e-01 9.33778822e-01 -9.19072330e-02 1.39295447e+00 1.81007892e-01 4.36580598e-01 1.45986748e+00 3.82794350e-01 1.29711735e+00 7.93172538e-01 -2.65666336e-01 -2.51898915e-01 -4.53667521e-01 4.06663567e-01 3.67801279e-01 1.29669160e-01 -2.41020814e-01 -6.22417986e-01 -2.16240838e-01 4.08385485e-01 2.60502547e-01 -2.57281154e-01 4.65348095e-01 -1.40371346e+00 7.23025799e-01 3.59167606e-01 4.40955281e-01 -6.76149607e-01 2.17341378e-01 8.43664646e-01 4.25649993e-02 7.71235406e-01 -1.67679772e-01 -9.06821728e-01 -1.97597623e-01 -3.09850037e-01 4.08721864e-01 5.60445428e-01 8.79813373e-01 1.93530187e-01 -3.60230058e-01 -6.39001429e-01 9.33900595e-01 3.28820616e-01 4.60810103e-02 4.33629811e-01 -4.19065416e-01 1.01444614e+00 8.12642694e-01 -1.23398922e-01 -7.59385109e-01 -5.49948990e-01 -3.41627896e-01 -7.41658270e-01 -5.94112813e-01 5.52669913e-02 -3.61709654e-01 -4.99410987e-01 1.76079369e+00 7.33437717e-01 3.94919604e-01 1.85259312e-01 6.14709795e-01 1.23422575e+00 9.33485270e-01 4.49013144e-01 -3.90656054e-01 1.94654131e+00 -1.10776329e+00 -1.26830757e+00 -4.80002791e-01 5.59136569e-01 -6.89191699e-01 7.96758473e-01 9.13356915e-02 -1.03703296e+00 -1.65664792e-01 -8.89464200e-01 -4.56306487e-01 -3.46671939e-01 1.55586421e-01 8.87086332e-01 -5.07526807e-02 5.27361687e-03 2.67800927e-01 -1.01726699e+00 4.32364941e-02 6.06213808e-01 9.61018540e-03 -1.20404392e-01 2.36332044e-01 -1.87237835e+00 7.35310912e-01 7.13788509e-01 1.66526467e-01 -6.78382754e-01 -6.58223927e-01 -9.39036012e-01 2.63396353e-01 9.58281159e-01 -4.70685691e-01 1.51580465e+00 -2.32521802e-01 -1.05878091e+00 6.68648660e-01 -3.74065697e-01 -4.25384015e-01 1.30242631e-01 -6.84330940e-01 -7.51534343e-01 3.63665186e-02 3.10238361e-01 -5.38964383e-02 3.65763366e-01 -7.96992540e-01 -8.03178906e-01 -2.11244732e-01 9.83956680e-02 1.17111422e-01 -2.51587421e-01 7.59754479e-01 -4.40469682e-01 -1.06725657e+00 -1.10614069e-01 -4.87422884e-01 2.08629910e-02 -5.28283060e-01 -7.04935372e-01 -9.97285247e-01 7.00267792e-01 -6.40228868e-01 1.63001502e+00 -2.02445364e+00 -5.50161973e-02 -4.22404140e-01 3.04391265e-01 4.68849428e-02 1.61305204e-01 6.86438560e-01 -4.82507944e-01 3.29661854e-02 -1.46411225e-01 -4.20187742e-01 4.88551520e-02 2.86845416e-01 -6.21947944e-01 8.96302983e-02 6.54950917e-01 1.25404334e+00 -9.92260516e-01 -8.32980633e-01 -1.85251966e-01 4.98160601e-01 -1.60106212e-01 1.61776826e-01 -2.04776749e-01 8.43227953e-02 -1.05461371e+00 6.40708447e-01 4.77534086e-01 -7.58836091e-01 2.01515779e-01 -5.30714333e-01 -5.84653728e-02 1.06006002e+00 -9.53770697e-01 1.40606713e+00 -3.38074297e-01 2.97754049e-01 -3.29371810e-01 -1.06906211e+00 5.74995697e-01 6.92813575e-01 6.21442080e-01 -5.55549085e-01 3.61344337e-01 1.24817509e-02 -4.99874763e-02 -4.13102567e-01 3.98392975e-01 -1.01384535e-01 -3.24922889e-01 5.61159968e-01 1.77558273e-01 5.93894184e-01 2.46941060e-01 4.62583184e-01 1.32815015e+00 2.87649967e-02 8.31685722e-01 7.67593756e-02 4.24694270e-01 -3.22170228e-01 1.07659650e+00 5.80791175e-01 -7.42091313e-02 3.21802735e-01 7.63620257e-01 -4.02025014e-01 -5.97019196e-01 -8.98107588e-01 -2.58215815e-01 1.02946186e+00 -1.37226505e-03 -8.67347896e-01 -3.08546424e-01 -1.11957800e+00 -2.90820777e-01 6.84145331e-01 -7.12756276e-01 1.60743013e-01 -9.63901043e-01 -1.35218775e+00 5.54120600e-01 9.31734800e-01 5.15787184e-01 -1.29785490e+00 -4.58110213e-01 6.68145359e-01 -7.58489370e-01 -1.40091157e+00 -4.86522973e-01 3.64483446e-01 -4.40337270e-01 -1.23519480e+00 -3.27192187e-01 -5.25794029e-01 -6.88019469e-02 -9.23661795e-03 1.35335398e+00 -2.04201624e-01 2.12674979e-02 -8.92687440e-02 -8.34134519e-01 -7.51852989e-01 -2.18868256e-03 8.91405791e-02 -4.81363654e-01 7.79708400e-02 5.22950292e-01 -5.94085097e-01 -5.95117390e-01 -5.88343153e-03 -8.41754794e-01 5.80416955e-02 6.25949442e-01 8.84068251e-01 9.30241168e-01 1.78934216e-01 1.01266015e+00 -1.15407252e+00 6.73860013e-01 -8.94629657e-01 -2.38177523e-01 3.54264975e-01 -4.46718186e-01 -1.38002634e-01 5.79544008e-01 -5.34687579e-01 -1.41573155e+00 -3.68335098e-01 -2.94952363e-01 -1.28106400e-01 5.18026389e-02 8.98443997e-01 -4.18869972e-01 8.26235235e-01 1.39272243e-01 5.44414595e-02 -7.52760887e-01 -4.10016149e-01 2.45511532e-01 4.52934951e-01 4.94782358e-01 -6.08697236e-01 3.92139584e-01 4.66738433e-01 -2.25638866e-01 -1.10790739e-02 -1.61442077e+00 -2.30821848e-01 -1.80237323e-01 8.94232690e-02 8.68000507e-01 -1.02107370e+00 -7.05243170e-01 3.88945520e-01 -1.62702298e+00 -3.11577916e-01 -2.72983074e-01 6.14652932e-01 -1.16361164e-01 1.71527594e-01 -1.16548514e+00 -7.49811530e-01 -6.33171797e-01 -7.74431825e-01 1.21854532e+00 1.28961772e-01 -5.50065786e-02 -9.42111492e-01 1.13418691e-01 2.31544465e-01 -2.30218932e-01 3.86655539e-01 8.58058929e-01 -1.04592395e+00 -4.90943074e-01 8.42885487e-03 -4.91296411e-01 -1.74070463e-01 2.10020304e-01 -2.25831017e-01 -7.65420616e-01 2.69030631e-01 1.28790259e-01 -2.81314880e-01 1.02755880e+00 3.50571752e-01 1.29716337e+00 -3.35581958e-01 -6.16807401e-01 2.16640711e-01 9.88825202e-01 2.60312468e-01 5.36969781e-01 4.33567494e-01 8.70278180e-01 4.13835198e-01 8.92476857e-01 6.78982556e-01 1.00748897e+00 7.57343411e-01 6.39346093e-02 -1.05336465e-01 -2.29296058e-01 -2.47530460e-01 3.76364678e-01 9.02283311e-01 4.00014408e-02 -7.33029902e-01 -7.09053814e-01 7.29378283e-01 -2.16032338e+00 -1.05596399e+00 -4.96222138e-01 1.45938051e+00 1.12571287e+00 4.29960608e-01 2.91909324e-03 2.86426902e-01 6.90653205e-01 4.35241550e-01 -4.43634689e-01 -1.53524354e-01 -3.01870435e-01 2.19310492e-01 1.54823080e-01 1.72355548e-01 -1.50721753e+00 8.07963431e-01 5.22006321e+00 1.14801419e+00 -5.19936740e-01 5.34039974e-01 6.05818033e-01 2.94750673e-03 -3.22698087e-01 1.09266549e-01 -1.08025336e+00 6.57290757e-01 1.08064914e+00 -3.26022148e-01 -1.67678088e-01 5.68081141e-01 1.25654265e-01 1.61900640e-01 -1.08535588e+00 6.84702933e-01 -9.30286348e-02 -1.39219630e+00 -1.31707445e-01 2.86547486e-02 1.65408239e-01 5.80061525e-02 -2.58182228e-01 5.63217103e-01 3.84992540e-01 -4.74064529e-01 6.92650378e-01 1.86030671e-01 4.37033623e-01 -7.84084737e-01 8.19200873e-01 7.77638853e-02 -1.82755828e+00 -7.18255639e-02 1.16946921e-01 6.34503439e-02 8.39007199e-01 9.72786009e-01 -4.22124118e-01 1.07489467e+00 8.94081593e-01 1.06530428e+00 -2.98808724e-01 5.23524582e-01 -5.45758724e-01 8.71827185e-01 -4.00364488e-01 1.42389387e-02 8.50597918e-02 2.99897701e-01 5.54139555e-01 1.33532226e+00 2.47193292e-01 6.67676508e-01 1.01997085e-01 7.46169567e-01 -3.67262274e-01 7.13127628e-02 -2.16114998e-01 -1.06152773e-01 7.20457077e-01 1.33874869e+00 -7.20184624e-01 -4.55265105e-01 -8.53247881e-01 8.66512775e-01 5.79064786e-01 1.26549140e-01 -1.35002279e+00 -3.58735621e-01 2.77205974e-01 -3.52568895e-01 5.83904982e-01 2.54338682e-01 -9.01422724e-02 -1.45290792e+00 2.82013655e-01 -6.30214870e-01 1.08301282e+00 -6.60635948e-01 -1.62336195e+00 6.84265852e-01 2.30349496e-01 -9.73792613e-01 -1.28548235e-01 -4.32670057e-01 -7.99321592e-01 4.06092793e-01 -1.64811528e+00 -1.57023704e+00 -3.74913290e-02 5.86997271e-01 6.93976104e-01 1.70332342e-01 5.98987699e-01 7.78116524e-01 -9.30073500e-01 5.40193439e-01 -3.97342056e-01 4.79048669e-01 6.56041145e-01 -1.24506640e+00 5.14456213e-01 8.61400485e-01 3.39722127e-01 3.93393785e-01 2.00920835e-01 -8.09571087e-01 -1.16290784e+00 -1.25644374e+00 1.41519105e+00 -5.54416478e-01 9.83084440e-01 -3.15845072e-01 -1.04105508e+00 1.27896821e+00 2.33926237e-01 9.59402621e-02 9.19186711e-01 4.43610132e-01 -4.33855385e-01 -1.74968950e-02 -6.16310060e-01 4.65549856e-01 1.03411281e+00 -5.69348216e-01 -1.05372798e+00 5.94808996e-01 1.15703380e+00 -6.01402164e-01 -1.11824071e+00 7.08579719e-01 2.09662437e-01 -5.55701911e-01 1.15837312e+00 -5.80144346e-01 8.31323683e-01 -1.54285356e-01 -1.31203914e-02 -8.30227971e-01 -3.40914190e-01 -4.30605859e-01 -9.52065706e-01 1.77239192e+00 6.59885049e-01 -6.36213720e-01 1.68174088e-01 3.36035997e-01 -3.26508462e-01 -1.16466618e+00 -8.94901812e-01 -5.96161902e-01 -9.33763664e-03 -5.90425611e-01 8.08143079e-01 1.11713409e+00 5.49594052e-02 9.72034633e-01 -4.46631134e-01 4.08703208e-01 3.53186816e-01 5.75515628e-01 1.62464738e-01 -1.02801633e+00 -4.82549608e-01 -2.21840739e-01 2.47140840e-01 -9.71955955e-01 4.43630576e-01 -9.16292906e-01 -1.38971344e-01 -1.67961299e+00 5.37354648e-01 -1.95342541e-01 -7.22007215e-01 8.15054655e-01 -7.80924141e-01 -1.75099492e-01 -1.65777162e-01 1.19618729e-01 -9.29747522e-01 8.77924025e-01 1.00428796e+00 8.06926191e-02 2.59599872e-02 -1.45171940e-01 -7.99290836e-01 7.11456537e-01 6.75138295e-01 -8.13713789e-01 -1.35817990e-01 -4.63247299e-01 3.41487229e-01 2.08514541e-01 3.80207807e-01 -4.81478065e-01 2.22145513e-01 -4.12556576e-03 3.45384598e-01 -9.94622409e-01 2.56780207e-01 -5.49339831e-01 -1.78546421e-02 9.62126479e-02 -4.62277234e-01 2.40586996e-01 2.75561482e-01 7.93381631e-01 -4.51093256e-01 -6.67461008e-02 1.15051650e-01 1.61630929e-01 -5.17101049e-01 4.61220622e-01 -1.22508407e-01 4.75551039e-01 8.99115264e-01 6.02137625e-01 -6.51783347e-01 -1.32562399e-01 -5.24872124e-01 4.85697597e-01 -4.43135172e-01 4.35467541e-01 5.11474729e-01 -1.59711325e+00 -1.00120056e+00 -3.89729291e-01 2.53604323e-01 2.04877317e-01 4.07483190e-01 8.26488197e-01 2.20444456e-01 2.63791591e-01 5.90608358e-01 -1.02067977e-01 -1.17878687e+00 6.54552937e-01 -4.74638492e-02 -9.92585182e-01 -9.43668723e-01 9.60756719e-01 2.78772175e-01 -2.04551861e-01 -1.11429423e-01 -5.03269732e-01 -3.47086906e-01 2.43370682e-01 7.44001210e-01 5.73522188e-02 2.94907223e-02 -2.94642538e-01 -5.91827750e-01 1.48276240e-01 -5.01720130e-01 -5.48600480e-02 1.47518933e+00 -1.49942841e-02 -2.23352447e-01 5.55251837e-01 9.79109049e-01 1.34245649e-01 -1.21162152e+00 -4.33482260e-01 1.37322396e-01 -8.66634697e-02 1.17766306e-01 -8.27694118e-01 -8.98298919e-01 5.86534381e-01 -1.34235874e-01 3.91914546e-01 1.26639211e+00 5.21044850e-01 1.26601291e+00 2.24283680e-01 -2.07308587e-02 -7.63236821e-01 6.06728233e-02 5.14875948e-01 9.32991564e-01 -1.15723729e+00 1.21283546e-01 -8.21792483e-01 -7.35591948e-01 7.07425833e-01 5.33700049e-01 1.22012515e-02 7.67010570e-01 6.23110950e-01 -3.96196127e-01 -4.33167487e-01 -1.16157317e+00 -2.24568143e-01 3.00420672e-01 4.83167777e-03 6.69506073e-01 -2.63635982e-02 -6.29533589e-01 1.39096284e+00 2.25045085e-01 3.05606164e-02 1.32399932e-01 9.76121247e-01 3.47084664e-02 -1.42430687e+00 3.93117554e-02 4.89323735e-01 -9.96723175e-01 -5.09163499e-01 -2.03911304e-01 7.45666981e-01 8.41573030e-02 1.02629113e+00 -1.43131213e-02 -1.53582871e-01 3.69217783e-01 1.84241205e-01 2.55217522e-01 -4.01015162e-01 -8.05273712e-01 3.33910644e-01 5.73631585e-01 -6.69323504e-01 -5.80097854e-01 -8.56652558e-01 -1.55052698e+00 8.97116512e-02 -3.25536728e-01 2.04855874e-01 9.00229886e-02 1.19888186e+00 5.99228024e-01 1.20254672e+00 4.56174016e-01 -2.86285192e-01 -6.77578971e-02 -1.00708258e+00 -3.00459146e-01 3.49812359e-01 1.94307595e-01 -9.11719143e-01 -9.06385109e-02 8.30869749e-02]
[9.04732608795166, 9.142507553100586]
c0225991-4273-40fe-b5ae-21f634f1dd8f
learning-binary-features-online-from-motion
1601.03821
null
http://arxiv.org/abs/1601.03821v2
http://arxiv.org/pdf/1601.03821v2.pdf
Learning Binary Features Online from Motion Dynamics for Incremental Loop-Closure Detection and Place Recognition
This paper proposes a simple yet effective approach to learn visual features online for improving loop-closure detection and place recognition, based on bag-of-words frameworks. The approach learns a codeword in bag-of-words model from a pair of matched features from two consecutive frames, such that the codeword has temporally-derived perspective invariance to camera motion. The learning algorithm is efficient: the binary descriptor is generated from the mean image patch, and the mask is learned based on discriminative projection by minimizing the intra-class distances among the learned feature and the two original features. A codeword for bag-of-words models is generated by packaging the learned descriptor and mask, with a masked Hamming distance defined to measure the distance between two codewords. The geometric properties of the learned codewords are then mathematically justified. In addition, hypothesis constraints are imposed through temporal consistency in matched codewords, which improves precision. The approach, integrated in an incremental bag-of-words system, is validated on multiple benchmark data sets and compared to state-of-the-art methods. Experiments demonstrate improved precision/recall outperforming state of the art with little loss in runtime.
['Mason J. Lilly', 'Guangcong Zhang', 'Patricio A. Vela']
2016-01-15
null
null
null
null
['loop-closure-detection']
['computer-vision']
[ 2.45467588e-01 -3.18551868e-01 -5.29579282e-01 -4.03935343e-01 -9.45979774e-01 -5.19104123e-01 6.82078838e-01 5.13853848e-01 -3.35297555e-01 2.07278356e-01 3.26137602e-01 2.13278066e-02 -2.73492157e-01 -4.18935418e-01 -8.97197843e-01 -7.59115100e-01 -4.52450067e-01 -1.11364886e-01 4.10444945e-01 5.28312773e-02 7.49886632e-01 5.86097956e-01 -2.00343537e+00 4.22749221e-01 2.13867649e-01 1.17621136e+00 3.17849636e-01 6.14056766e-01 1.98207855e-01 3.72135967e-01 -6.14887774e-02 2.30464265e-01 6.34672761e-01 -3.11962783e-01 -3.67091060e-01 3.46378267e-01 1.01569653e+00 -2.48338073e-01 -5.36409497e-01 8.96388352e-01 1.14272036e-01 3.72632831e-01 6.64467871e-01 -8.23357105e-01 -7.38316000e-01 -4.62389320e-01 -5.36631584e-01 7.02838674e-02 7.65052736e-01 -1.65311828e-01 1.25559568e+00 -1.25860274e+00 8.31687093e-01 1.02156138e+00 5.86097479e-01 2.48010114e-01 -1.25759959e+00 -2.25428119e-01 2.00811833e-01 4.37377065e-01 -1.65972149e+00 -4.76470679e-01 4.80883062e-01 -6.35622501e-01 1.17793262e+00 4.00293142e-01 9.82516527e-01 5.72320461e-01 6.45547926e-01 5.64093530e-01 7.07987666e-01 -9.25808907e-01 2.43544772e-01 8.30722526e-02 2.21534688e-02 1.14893746e+00 1.55224934e-01 5.74810207e-01 -9.00388122e-01 -2.57740736e-01 6.49006367e-01 4.16136384e-01 -2.96297878e-01 -1.23044157e+00 -1.34616947e+00 8.03792894e-01 5.68617046e-01 3.47937703e-01 -9.50071588e-02 1.65752813e-01 1.54291019e-01 2.01272424e-02 1.75870344e-01 -2.46409532e-02 -5.81918433e-02 1.17785700e-01 -1.10579085e+00 3.48483026e-01 3.31381202e-01 1.20460987e+00 1.00956929e+00 -5.09888470e-01 -2.96042919e-01 6.13138676e-01 5.78589261e-01 5.18823862e-01 5.74354589e-01 -5.63042223e-01 1.23557098e-01 6.07127488e-01 -6.16116300e-02 -1.49818349e+00 -1.63056672e-01 -5.20192496e-02 -4.63028222e-01 2.63468444e-01 -2.62253545e-02 6.66607261e-01 -9.11060572e-01 1.30063224e+00 3.57120574e-01 4.91195470e-01 -2.09026054e-01 8.91814351e-01 4.42222297e-01 6.88908637e-01 -6.32276416e-01 -1.98101863e-01 1.25913966e+00 -1.17789531e+00 -4.73103642e-01 -3.16077679e-01 6.80373430e-01 -8.76633823e-01 5.94959795e-01 1.82480887e-01 -8.26384783e-01 -7.16310859e-01 -1.42956674e+00 -1.09835923e-01 -4.98285115e-01 1.99834689e-01 1.95926115e-01 4.93664593e-01 -1.08431756e+00 5.15434563e-01 -7.26649702e-01 -5.60055494e-01 1.66606437e-02 2.82512248e-01 -7.50206888e-01 -4.53582443e-02 -6.74660206e-01 7.97871947e-01 3.95385534e-01 -1.92237973e-01 -8.45454872e-01 -5.05416036e-01 -1.33831263e+00 -1.79374650e-01 -8.70902687e-02 -3.95555109e-01 8.40033710e-01 -8.12458098e-01 -1.10044682e+00 1.14926553e+00 -5.42110562e-01 -5.48545182e-01 1.86544210e-01 -1.75870761e-01 -3.26389194e-01 4.15245235e-01 2.27356285e-01 8.65733624e-01 1.30374146e+00 -1.22876668e+00 -8.85649681e-01 -3.44402045e-01 -2.11863533e-01 2.14213654e-01 -3.17653179e-01 -1.68719366e-01 -6.63656890e-01 -6.05755866e-01 5.27692199e-01 -1.04776978e+00 -5.23683392e-02 7.03672647e-01 7.42287263e-02 1.09693632e-01 1.06561208e+00 -5.17035007e-01 1.37451458e+00 -2.58351326e+00 7.31824711e-02 4.93138313e-01 1.25857936e-02 8.07382986e-02 -1.76283330e-01 5.55650473e-01 -2.21256405e-01 -4.98854399e-01 -1.92378238e-01 -4.39000934e-01 -3.20731431e-01 1.85198680e-01 -5.43323934e-01 1.14607203e+00 -2.84959599e-02 5.42783320e-01 -1.03958905e+00 -5.59898555e-01 6.87676787e-01 4.31112051e-01 -5.19436598e-01 2.94503123e-01 1.17344640e-01 1.77049011e-01 -2.17620403e-01 5.73601425e-01 6.18116915e-01 -1.37183648e-02 -9.56882685e-02 -5.84314093e-02 -3.65241170e-01 7.37677291e-02 -1.27767694e+00 2.11337805e+00 -2.08796501e-01 7.28066206e-01 -5.25826991e-01 -9.95887280e-01 1.17904162e+00 5.65227792e-02 4.05742556e-01 -7.36140132e-01 -1.17043450e-01 2.11068884e-01 -3.89055669e-01 -5.45203149e-01 5.66865027e-01 2.65572131e-01 -4.81908966e-04 1.20699614e-01 8.05585682e-02 -9.96307954e-02 2.28597328e-01 3.72938775e-02 9.24990058e-01 3.50643545e-01 7.82630563e-01 -3.72449279e-01 6.81866825e-01 -1.37051612e-01 2.05837369e-01 7.27537334e-01 -2.93729067e-01 9.32534099e-01 -8.51519313e-03 -7.18943834e-01 -1.06240833e+00 -1.06424975e+00 -2.62444347e-01 8.77037406e-01 5.42918444e-01 -8.01866114e-01 -5.55507123e-01 -6.90677404e-01 2.22042590e-01 3.81542146e-01 -8.47075105e-01 -1.63779348e-01 -3.76322687e-01 1.71881542e-01 8.06311369e-02 4.11895573e-01 1.17474444e-01 -5.55465102e-01 -1.17220902e+00 9.61723328e-02 1.57006904e-01 -9.13077533e-01 -7.71693468e-01 -5.45984413e-03 -8.50273907e-01 -1.15281141e+00 -4.63844538e-01 -1.13054550e+00 9.15608168e-01 1.00886869e+00 6.75367475e-01 2.78658092e-01 -7.35119820e-01 6.68641329e-01 -4.96744990e-01 8.77799019e-02 1.04230396e-01 -5.56783020e-01 7.93000013e-02 3.49869400e-01 3.47289532e-01 -2.47712001e-01 -7.40222573e-01 2.38793522e-01 -7.70842314e-01 1.96865220e-02 4.41637039e-01 1.10946012e+00 9.94228721e-01 -5.81847072e-01 -3.87991369e-01 -4.31370549e-02 3.08151618e-02 -1.64195076e-01 -7.67055631e-01 3.98889512e-01 -5.31155527e-01 3.18802744e-01 3.44136357e-01 -3.61464649e-01 -4.36910689e-01 4.53789145e-01 5.35920680e-01 -6.71197653e-01 -8.21767151e-02 2.31468558e-01 2.40379602e-01 -4.31493908e-01 5.83555281e-01 4.60859656e-01 -1.70476083e-02 -1.63941979e-01 6.44772768e-01 6.48160219e-01 5.16511798e-01 -3.54852110e-01 6.46829784e-01 8.60837936e-01 1.55846864e-01 -9.43167686e-01 -7.07500219e-01 -1.22028542e+00 -1.01770520e+00 -2.80932873e-01 7.53210187e-01 -9.73260224e-01 -2.82724559e-01 8.09112266e-02 -1.27364671e+00 3.91381204e-01 -1.93940863e-01 7.08028972e-01 -9.46206748e-01 5.07471442e-01 -1.92840584e-02 -7.31365383e-01 -1.39571093e-02 -1.04846144e+00 1.44328403e+00 1.51635945e-01 -2.69002050e-01 -9.66934800e-01 5.28155327e-01 -4.87802960e-02 -8.95262435e-02 1.50036067e-01 7.39932418e-01 -5.09188354e-01 -8.42333138e-01 -6.02162421e-01 -1.92401692e-01 2.52469063e-01 1.99412964e-02 7.86823109e-02 -8.74557257e-01 -6.99459195e-01 -1.37897626e-01 -1.16116712e-02 7.92075455e-01 4.00953412e-01 9.84613717e-01 -7.55359828e-02 -6.89865947e-01 6.45303309e-01 1.60013080e+00 2.19522536e-01 5.52252650e-01 4.54107612e-01 4.29299235e-01 3.79810274e-01 8.44509780e-01 5.94519436e-01 1.39396802e-01 9.94008720e-01 4.05857444e-01 1.80104256e-01 -6.15230799e-02 -4.36196119e-01 3.25134873e-01 6.72878921e-01 3.51668507e-01 2.11103201e-01 -8.42829466e-01 8.12507391e-01 -2.09485126e+00 -9.82865155e-01 -7.17332885e-02 2.67503524e+00 3.21443975e-01 -5.75885288e-02 -1.40221134e-01 1.52087599e-01 7.32315063e-01 5.01757324e-01 -1.59370944e-01 -4.12731946e-01 3.10102142e-02 7.64685273e-02 4.74054694e-01 6.48974419e-01 -1.35281670e+00 5.89464128e-01 6.33646297e+00 7.02367723e-01 -1.07514739e+00 1.14968466e-02 7.52941296e-02 -1.01770472e-03 5.61794452e-03 4.06019419e-01 -9.08806980e-01 2.27896750e-01 5.42056024e-01 -4.06219922e-02 2.58241385e-01 8.72888327e-01 7.03314543e-02 -3.54148567e-01 -1.41488838e+00 1.21582019e+00 8.01138103e-01 -1.57574487e+00 -6.19172119e-02 1.57503828e-01 9.05938148e-01 -1.81366667e-01 1.68854490e-01 -7.66496435e-02 -6.36362553e-01 -7.30216861e-01 1.08013320e+00 7.99790680e-01 6.79534674e-01 -5.24625123e-01 3.26603323e-01 2.02537134e-01 -1.54579484e+00 -9.59706455e-02 -4.69980031e-01 -1.18979722e-01 -2.11955622e-01 2.82982677e-01 -6.25491381e-01 3.91437173e-01 7.12394416e-01 1.18241489e+00 -7.51402020e-01 1.37558711e+00 9.47787762e-02 -6.11233078e-02 -1.60162717e-01 6.74123615e-02 4.40036207e-01 -2.83472300e-01 5.65340936e-01 1.39480817e+00 5.51563025e-01 -1.94098160e-01 4.28630024e-01 5.14261305e-01 4.49843973e-01 3.81897956e-01 -9.98366177e-01 2.99322993e-01 4.84300554e-01 1.14768946e+00 -7.07742274e-01 -2.93168753e-01 -5.63912272e-01 1.06276941e+00 3.12673002e-01 8.04905370e-02 -6.68452382e-01 -3.31683993e-01 6.49655700e-01 2.42351487e-01 7.51817822e-01 -3.57008159e-01 -5.05520888e-02 -8.62533450e-01 3.16347569e-01 -6.13118768e-01 2.52189636e-01 -6.97306395e-01 -8.44350338e-01 2.62836963e-01 1.21702217e-01 -2.03988981e+00 -2.81649888e-01 -7.02674985e-01 -3.55011344e-01 5.86141884e-01 -1.28158581e+00 -1.13160729e+00 -3.19388330e-01 5.80458999e-01 5.58543563e-01 -2.07150370e-01 1.10734272e+00 1.09499730e-01 -3.76478173e-02 5.52689373e-01 6.51632190e-01 -9.81085151e-02 8.21674287e-01 -8.24602902e-01 3.34014654e-01 8.35568964e-01 5.81335008e-01 7.41729319e-01 6.21426225e-01 -4.58359808e-01 -1.45939946e+00 -1.08741188e+00 1.10446978e+00 -3.00751001e-01 6.54046118e-01 -4.51806337e-01 -6.91635311e-01 5.74074209e-01 4.44491860e-03 6.69111490e-01 5.85597456e-01 -2.79843837e-01 -8.63408089e-01 -1.26203120e-01 -9.70961988e-01 5.29714823e-01 1.00700283e+00 -7.73567975e-01 -7.18045592e-01 4.28263158e-01 3.70525002e-01 -6.92812622e-01 -4.60719228e-01 1.78603470e-01 8.86769056e-01 -1.31162775e+00 1.13084233e+00 -4.20444548e-01 1.55582607e-01 -5.97345769e-01 -6.26509607e-01 -9.61249769e-01 -4.92285252e-01 -4.45531994e-01 -2.32513562e-01 5.88665843e-01 4.22394164e-02 -2.20651567e-01 4.33188379e-01 -1.12319268e-01 -1.19472437e-01 -9.86530542e-01 -1.19204271e+00 -1.01157439e+00 -5.57912767e-01 -2.93156624e-01 2.18119502e-01 4.42885369e-01 2.98450023e-01 -5.43537997e-02 -3.82506281e-01 2.70623744e-01 6.35976315e-01 2.93395728e-01 6.62386000e-01 -8.88662517e-01 -2.39714459e-01 -1.91615328e-01 -1.26608264e+00 -1.24610102e+00 1.38007000e-01 -8.41652632e-01 3.24739277e-01 -1.16896653e+00 2.60116220e-01 -1.51515201e-01 -2.10585460e-01 3.35783452e-01 1.92736641e-01 2.92954445e-01 3.43933791e-01 1.32331192e-01 -7.74475515e-01 5.35234272e-01 7.99500823e-01 -1.59370020e-01 -3.35118696e-02 -3.46082777e-01 4.84428816e-02 6.41295135e-01 2.94664204e-01 -5.40639102e-01 -2.12792501e-01 -1.22839123e-01 -1.81690797e-01 -1.28671139e-01 5.23315251e-01 -1.11910629e+00 3.56308907e-01 -4.88690063e-02 4.04445440e-01 -8.53749752e-01 4.73730654e-01 -1.10377729e+00 3.87425311e-02 5.70853949e-01 -4.14720744e-01 2.34649405e-01 1.17276892e-01 9.81212676e-01 -1.56373233e-01 -4.15916473e-01 8.27275932e-01 2.60700405e-01 -1.00239992e+00 1.42856434e-01 -1.39990076e-01 -3.24359983e-01 1.56893992e+00 -7.88569808e-01 2.69194189e-02 -2.24864021e-01 -3.33386421e-01 -1.86250612e-01 6.35683954e-01 8.48370850e-01 1.15610063e+00 -1.71868944e+00 -4.22062397e-01 9.53863919e-01 9.25566256e-01 -4.05573457e-01 9.55434218e-02 7.03965366e-01 -6.21480465e-01 7.50470817e-01 -5.70513569e-02 -1.12229013e+00 -1.62095726e+00 8.48947287e-01 3.15374762e-01 3.46563309e-02 -8.02206337e-01 7.06276476e-01 2.26288408e-01 -2.42371187e-02 5.65845191e-01 -5.36167204e-01 7.39382133e-02 -6.97406335e-03 8.03735256e-01 2.23478988e-01 1.60965234e-01 -1.02417552e+00 -5.76220453e-01 1.18273234e+00 -1.48609310e-01 -1.85336903e-01 1.02833140e+00 -2.35251129e-01 -1.32946685e-01 5.41013896e-01 1.62873006e+00 1.39047140e-02 -1.33604765e+00 -3.65580887e-01 5.94880953e-02 -1.06283319e+00 4.34533060e-02 -2.50821084e-01 -6.43639505e-01 6.89159691e-01 1.12957346e+00 -1.09720446e-01 8.70192647e-01 -1.08082347e-01 5.22431791e-01 2.67231107e-01 5.46220362e-01 -9.56350505e-01 2.55248517e-01 5.57871521e-01 9.54727888e-01 -1.21628535e+00 1.47976622e-01 -1.40183210e-01 -2.12700441e-01 1.30961335e+00 1.83560133e-01 -5.54325223e-01 8.36597919e-01 -2.41062231e-02 -1.41352132e-01 -1.10105537e-01 -4.99856830e-01 -1.06819205e-01 8.57299805e-01 6.12526476e-01 2.18578592e-01 -3.43169048e-02 -2.82381833e-01 -1.45619303e-01 2.02564850e-01 -2.11506069e-01 6.80241510e-02 1.13297832e+00 -6.73853874e-01 -8.93047750e-01 -6.68079078e-01 1.02633372e-01 2.72424221e-01 -1.55238599e-01 -8.61226842e-02 6.28188968e-01 3.67021382e-01 6.32562697e-01 3.38987708e-01 -4.30503815e-01 2.56857425e-01 -1.57197326e-01 6.82245433e-01 -6.58825457e-01 -3.17303210e-01 1.73920333e-01 -3.78598958e-01 -8.77086043e-01 -4.35437679e-01 -8.63847435e-01 -1.18212366e+00 7.59146139e-02 -2.93862075e-01 -1.30297109e-01 6.94745600e-01 9.36265409e-01 5.87253690e-01 -1.11932881e-01 8.64047348e-01 -1.04991472e+00 -4.87964123e-01 -4.40936059e-01 -5.69039762e-01 4.27063316e-01 7.42663920e-01 -7.87123144e-01 -3.52891177e-01 3.94164681e-01]
[7.846816539764404, -1.9453561305999756]
82e06137-a6bb-4468-ab63-2d5edf23a0dd
reformulating-zero-shot-action-recognition
null
null
http://proceedings.neurips.cc/paper/2021/hash/d6539d3b57159babf6a72e106beb45bd-Abstract.html
http://proceedings.neurips.cc/paper/2021/file/d6539d3b57159babf6a72e106beb45bd-Paper.pdf
Reformulating Zero-shot Action Recognition for Multi-label Actions
The goal of zero-shot action recognition (ZSAR) is to classify action classes which were not previously seen during training. Traditionally, this is achieved by training a network to map, or regress, visual inputs to a semantic space where a nearest neighbor classifier is used to select the closest target class. We argue that this approach is sub-optimal due to the use of nearest neighbor on static semantic space and is ineffective when faced with multi-label videos - where two semantically distinct co-occurring action categories cannot be predicted with high confidence. To overcome these limitations, we propose a ZSAR framework which does not rely on nearest neighbor classification, but rather consists of a pairwise scoring function. Given a video and a set of action classes, our method predicts a set of confidence scores for each class independently. This allows for the prediction of several semantically distinct classes within one video input. Our evaluations show that our method not only achieves strong performance on three single-label action classification datasets (UCF-101, HMDB, and RareAct), but also outperforms previous ZSAR approaches on a challenging multi-label dataset (AVA) and a real-world surprise activity detection dataset (MEVA).
['Mubarak Shah', 'Yogesh Rawat', 'Kevin Duarte', 'Alec Kerrigan']
2021-12-01
null
https://openreview.net/forum?id=mHHU6KWQ1ci
https://openreview.net/pdf?id=mHHU6KWQ1ci
neurips-2021-12
['zero-shot-action-recognition']
['computer-vision']
[ 6.86023533e-01 -2.57398814e-01 -3.07513028e-01 -6.09767556e-01 -1.10385370e+00 -4.78218436e-01 6.09530509e-01 3.32068279e-02 -3.92314106e-01 5.16103566e-01 3.54520500e-01 2.07359612e-01 -1.25487685e-01 -5.10006309e-01 -6.12809062e-01 -5.73669553e-01 -1.54601038e-02 4.58807051e-01 5.62335372e-01 7.70072117e-02 3.37557018e-01 3.57846558e-01 -1.96057189e+00 9.41395521e-01 2.97451943e-01 1.34522188e+00 -1.53348828e-02 5.48040390e-01 9.06723067e-02 1.28873265e+00 -5.19531906e-01 -2.60840714e-01 2.43483573e-01 -8.41607571e-01 -1.03799427e+00 2.36473247e-01 7.28285551e-01 -2.22036451e-01 -1.26627758e-01 8.84285450e-01 1.70580789e-01 5.68172455e-01 8.59743118e-01 -1.53148222e+00 -3.69757503e-01 1.52573481e-01 -2.57553577e-01 1.54594168e-01 7.23082840e-01 -8.35910160e-03 1.21931851e+00 -7.63261616e-01 7.85700619e-01 1.09748018e+00 7.40190566e-01 8.68414223e-01 -1.19664872e+00 -4.81614918e-01 3.04547101e-01 6.08805954e-01 -1.32994783e+00 -5.71791172e-01 5.50192654e-01 -4.61293578e-01 1.09168136e+00 2.18606621e-01 6.04264677e-01 1.42003453e+00 4.03283611e-02 9.31031942e-01 1.03458309e+00 -2.93551058e-01 7.51876175e-01 -2.27641016e-01 -4.53043729e-02 6.71176612e-01 -3.68829370e-01 -2.30868943e-02 -9.81990457e-01 -1.31082460e-01 2.34117419e-01 1.85555816e-01 -2.74382204e-01 -8.27203929e-01 -1.29623508e+00 6.79040551e-01 2.68129528e-01 3.78828049e-01 -1.56394988e-01 1.49492353e-01 6.15179539e-01 2.96538681e-01 2.95587569e-01 3.92084211e-01 -4.68644559e-01 -4.70000416e-01 -8.79655778e-01 -1.02081718e-02 5.80756366e-01 7.26084113e-01 5.44812858e-01 -2.43356258e-01 -4.18217897e-01 7.43484437e-01 1.83469743e-01 2.21584201e-01 7.30450153e-01 -1.35168707e+00 2.73889959e-01 4.84269023e-01 3.35386209e-02 -8.76314998e-01 -1.90263867e-01 1.61134601e-01 -3.10909480e-01 6.37620866e-01 6.07709587e-01 4.97282684e-01 -1.00803769e+00 1.71408045e+00 1.38048828e-01 5.16434133e-01 2.91710496e-01 9.82209742e-01 6.48798645e-01 4.22115535e-01 2.57517904e-01 -1.94683701e-01 8.91929507e-01 -1.03648591e+00 -3.47593635e-01 -2.93027699e-01 8.08880568e-01 -1.91274211e-01 1.17478442e+00 5.71682274e-01 -6.33171976e-01 -4.79752719e-01 -1.03929198e+00 1.13290191e-01 -3.86732429e-01 -6.88391849e-02 2.15563595e-01 2.96393275e-01 -7.93092132e-01 9.17420685e-01 -8.73655796e-01 -5.84508002e-01 6.34787381e-01 5.69871776e-02 -7.56034255e-01 -2.44684935e-01 -8.22839260e-01 8.64012539e-01 3.50759566e-01 -4.99595761e-01 -1.03418279e+00 -2.59234637e-01 -9.27222490e-01 -3.47439311e-02 6.05854571e-01 -1.60383493e-01 1.19474936e+00 -1.49512935e+00 -1.37461686e+00 9.00975764e-01 -9.26696658e-02 -5.57713568e-01 3.72576624e-01 -2.05504239e-01 -4.93650377e-01 3.79610062e-01 2.97880143e-01 7.29119718e-01 8.25973451e-01 -1.01610601e+00 -8.18285048e-01 -4.16002035e-01 1.47841707e-01 4.19716924e-01 -6.99705929e-02 1.03061013e-01 -1.12458922e-01 -5.20502925e-01 1.81645960e-01 -9.89185750e-01 1.30396085e-02 1.54103920e-01 -2.79040784e-02 -2.75380582e-01 7.84293890e-01 -3.21615130e-01 7.88411140e-01 -2.38499045e+00 2.46935517e-01 4.31966521e-02 -2.66815275e-01 1.05882891e-01 -2.14802712e-01 1.66817948e-01 -3.22916806e-01 -2.74751157e-01 -2.16476813e-01 -1.66968569e-01 -1.10844098e-01 3.20989400e-01 -2.88004786e-01 5.12962103e-01 5.69687821e-02 6.42415047e-01 -1.10491335e+00 -6.40459597e-01 3.33821326e-01 2.55913317e-01 -4.69758153e-01 1.92318037e-01 -2.79876143e-01 3.07680905e-01 -1.87685311e-01 8.53074729e-01 1.97279919e-03 -3.68473768e-01 2.24247932e-01 -1.13746665e-01 2.80574054e-01 -6.50396273e-02 -1.28778577e+00 1.93459344e+00 -1.66782305e-01 4.60127950e-01 -6.37629032e-01 -1.34135199e+00 7.24963188e-01 3.15716714e-01 8.01538348e-01 -6.52210951e-01 3.93370651e-02 4.69974168e-02 -4.10965055e-01 -5.37004828e-01 -4.82994616e-02 -2.89984912e-01 -2.79786706e-01 3.97503585e-01 4.40143853e-01 2.65244365e-01 1.90538734e-01 8.84553269e-02 1.52403772e+00 4.33447421e-01 4.96005207e-01 1.56290263e-01 5.59961915e-01 1.52029648e-01 8.79835606e-01 7.85781682e-01 -6.61898434e-01 7.91446567e-01 4.95938808e-01 -5.84896684e-01 -6.70562863e-01 -1.31371105e+00 9.81507823e-02 1.30172718e+00 2.99766600e-01 -5.83084404e-01 -5.80924153e-01 -1.33786368e+00 -1.33645341e-01 7.73836315e-01 -7.54880726e-01 -4.09810930e-01 -2.92881489e-01 -1.26218706e-01 3.24617505e-01 7.07701921e-01 2.85867929e-01 -1.29886508e+00 -9.14484024e-01 -7.82058537e-02 -3.79683524e-01 -1.06785238e+00 -3.13312829e-01 3.01212132e-01 -7.28568137e-01 -1.48705018e+00 -2.79723257e-01 -6.00468218e-01 3.36909354e-01 2.62018442e-01 1.06128752e+00 -2.11265877e-01 -2.97226965e-01 6.87996447e-01 -6.51053905e-01 1.03119180e-01 -3.05991232e-01 -5.75066805e-01 2.92969704e-01 3.66016567e-01 6.95685923e-01 -2.96236217e-01 -3.91541362e-01 5.44585168e-01 -6.64424598e-01 -1.31374478e-01 5.20231366e-01 8.21991324e-01 9.30891156e-01 -1.52412488e-03 5.75615644e-01 -5.31095743e-01 -5.75835183e-02 -5.01970768e-01 -1.68797821e-01 6.08369529e-01 -3.18616867e-01 -1.13775566e-01 6.01303339e-01 -4.67149258e-01 -8.84964764e-01 6.11981332e-01 2.27518499e-01 -8.29147160e-01 -6.61109388e-01 1.76808491e-01 -2.05262810e-01 9.47977304e-02 7.29927003e-01 1.34148702e-01 3.36518176e-02 -3.19562405e-01 3.18426490e-01 5.37038505e-01 6.19154036e-01 -1.41264766e-01 2.88862497e-01 6.55494630e-01 3.41488905e-02 -4.62512523e-01 -1.25404477e+00 -7.29549766e-01 -9.69158888e-01 -5.58543086e-01 1.35873544e+00 -9.51087177e-01 -4.44125235e-01 3.43383342e-01 -7.43677139e-01 -3.45565647e-01 -5.67644835e-01 7.31669426e-01 -1.17045259e+00 2.77339637e-01 -2.21598372e-01 -6.17378771e-01 2.07294419e-01 -9.70707774e-01 1.18694639e+00 -1.80922486e-02 -5.38229227e-01 -5.92253208e-01 1.00259706e-01 6.57886922e-01 -4.80816625e-02 2.78588712e-01 5.30469120e-01 -9.87178564e-01 -3.15923870e-01 -2.37888217e-01 3.39547917e-02 4.97411251e-01 2.68550575e-01 -2.40674376e-01 -9.10390913e-01 -3.00567985e-01 -1.91177532e-01 -9.93939579e-01 9.31257665e-01 1.70792297e-01 1.31546390e+00 -8.40565711e-02 -3.35204393e-01 3.22898388e-01 1.29310536e+00 3.57522011e-01 4.90713477e-01 2.77256072e-01 5.77460468e-01 4.02439922e-01 1.15816665e+00 2.99546808e-01 8.00605416e-02 9.30363595e-01 6.07740879e-01 3.43881518e-01 -8.55653062e-02 -2.19284758e-01 5.50150752e-01 1.51139110e-01 2.51167510e-02 -1.91819623e-01 -7.93230534e-01 4.02917951e-01 -2.13632298e+00 -1.47013962e+00 2.71606266e-01 2.30444741e+00 5.44998586e-01 -5.84791298e-04 2.67583430e-01 8.08968171e-02 5.68139672e-01 1.20641001e-01 -7.05175877e-01 -8.54703635e-02 2.96559036e-02 1.57621264e-01 1.40324101e-01 1.44655943e-01 -1.59194529e+00 8.91098976e-01 6.41970205e+00 8.07393193e-01 -7.88521230e-01 1.86915368e-01 5.71967542e-01 -4.67810363e-01 4.50325489e-01 -8.55325460e-02 -4.92038459e-01 4.76769090e-01 9.61741745e-01 1.81224138e-01 3.86312574e-01 1.12443197e+00 -1.11190066e-01 -3.36971819e-01 -1.43907559e+00 1.13536954e+00 6.28039598e-01 -1.16459072e+00 -2.71245912e-02 -3.09332430e-01 5.91105223e-01 4.98366430e-02 -2.45673269e-01 4.68921334e-01 2.98787415e-01 -1.01335585e+00 7.02288926e-01 7.16596186e-01 7.17230856e-01 -6.86008692e-01 4.70980763e-01 2.21716344e-01 -9.94115591e-01 -4.49819654e-01 -2.55582958e-01 4.72969041e-05 -4.48630080e-02 -3.71396467e-02 -6.59931898e-01 2.06401676e-01 1.12172377e+00 1.13102376e+00 -5.69434941e-01 9.49664593e-01 -5.56949005e-02 3.68323237e-01 8.12915489e-02 2.63376683e-01 2.08770871e-01 1.89769287e-02 2.74899751e-01 8.59596968e-01 3.36399227e-01 2.24919051e-01 4.66509908e-01 4.38115865e-01 2.25120574e-01 8.63085389e-02 -8.14088345e-01 1.18319616e-01 1.87312230e-01 9.73590255e-01 -7.92907000e-01 -4.39265549e-01 -7.04486668e-01 1.36917365e+00 4.26655203e-01 1.47730559e-01 -8.94837022e-01 -9.66914669e-02 8.82168710e-01 -1.42838567e-01 4.54062819e-01 9.53846872e-02 2.06068963e-01 -1.26773894e+00 7.13081062e-02 -9.28094268e-01 8.98244500e-01 -9.48231399e-01 -1.14045620e+00 4.52866673e-01 -9.67719033e-02 -1.76704144e+00 -4.79742527e-01 -6.08365238e-01 -3.65990341e-01 3.10291559e-01 -1.09633040e+00 -9.30452585e-01 -3.14802378e-01 9.13142860e-01 7.74633765e-01 -3.58689874e-01 1.11801362e+00 1.09356791e-01 -1.03771649e-01 3.15385997e-01 9.39749405e-02 1.46624699e-01 8.96999121e-01 -1.21317661e+00 -1.51508436e-01 6.02361381e-01 6.16210878e-01 -2.03303739e-01 5.79198182e-01 -6.11837447e-01 -9.17674601e-01 -1.17982984e+00 7.82135963e-01 -8.23692977e-01 5.73403537e-01 -1.74227729e-01 -8.52581084e-01 7.83314228e-01 -3.11162353e-01 5.87116897e-01 9.47624564e-01 -5.44558391e-02 -5.24948835e-01 -5.42914346e-02 -1.12551010e+00 3.12690914e-01 1.44292331e+00 -6.73471749e-01 -7.16486633e-01 4.86144572e-01 2.99939841e-01 -1.30961955e-01 -7.93569326e-01 4.83356148e-01 6.35052919e-01 -1.29919815e+00 9.55148160e-01 -1.14837253e+00 4.77845609e-01 -4.85074520e-01 -6.95204735e-01 -1.32899296e+00 -2.14219153e-01 2.42518052e-01 -1.17262036e-01 8.69445860e-01 2.04964921e-01 -4.99829538e-02 8.03487837e-01 5.19694209e-01 -6.85477033e-02 -6.66461885e-01 -1.25107133e+00 -1.09792268e+00 -4.83862609e-01 -5.20799696e-01 2.57933468e-01 1.05362022e+00 1.57534078e-01 1.86889380e-01 -5.07459283e-01 4.95247617e-02 5.08674085e-01 2.97131002e-01 6.29788756e-01 -1.22389901e+00 -4.10781652e-01 -2.42060080e-01 -1.06506348e+00 -5.04085422e-01 4.71649498e-01 -8.92392039e-01 3.32003772e-01 -1.28066647e+00 3.76349419e-01 -1.50671735e-01 -6.89706028e-01 8.82754624e-01 1.74518228e-01 5.95111668e-01 2.01312736e-01 3.38792741e-01 -1.34928071e+00 5.85493386e-01 6.48913026e-01 -1.65461957e-01 7.89731219e-02 5.47088236e-02 -2.53266990e-01 9.43306565e-01 5.27670145e-01 -5.12782395e-01 -5.95411420e-01 -3.72111276e-02 -1.12917304e-01 2.81772584e-01 5.93427956e-01 -1.49260390e+00 -5.44144399e-02 -4.16006386e-01 5.05017459e-01 -3.16795319e-01 5.72595358e-01 -9.96742487e-01 2.49271579e-02 2.57535309e-01 -7.38726914e-01 -4.17649418e-01 -3.61734390e-01 9.01775837e-01 -2.31387645e-01 -2.21774250e-01 9.50704336e-01 -1.77389845e-01 -1.37718558e+00 1.53111756e-01 -2.77042091e-01 7.07864463e-02 1.50899506e+00 -4.50248569e-01 -1.42344519e-01 -3.46571475e-01 -1.10301113e+00 1.59978010e-02 6.52131855e-01 7.63402283e-01 7.15997100e-01 -1.66251409e+00 -3.79763782e-01 1.64877042e-01 5.89150429e-01 -4.48987335e-01 2.23270461e-01 6.34223104e-01 -5.59562221e-02 1.94263011e-01 -5.60160279e-01 -7.42813051e-01 -1.61030602e+00 6.89267099e-01 4.65638965e-01 -2.40870863e-02 -7.28713334e-01 8.49413991e-01 9.21290070e-02 -1.98713452e-01 4.15883392e-01 -3.73150744e-02 -3.83280724e-01 1.46271707e-03 6.72067523e-01 2.43016258e-01 4.19454612e-02 -1.10037017e+00 -6.65901124e-01 4.48512554e-01 1.32616311e-01 -3.72488871e-02 1.23417842e+00 -1.45883542e-02 3.85938227e-01 9.11282361e-01 1.26139748e+00 -5.42980611e-01 -1.48914409e+00 -2.94010788e-01 1.81872964e-01 -8.64038169e-01 -1.37367725e-01 -9.68066990e-01 -9.05781507e-01 5.33336341e-01 8.76106381e-01 -6.80087954e-02 1.11258125e+00 3.12491328e-01 4.91130441e-01 5.41988373e-01 4.85339314e-01 -1.46414709e+00 6.89777017e-01 4.09329444e-01 6.96603596e-01 -1.51389527e+00 -2.70563126e-01 -4.08254080e-02 -1.12693429e+00 9.83551383e-01 9.30346489e-01 -2.12932322e-02 4.64282721e-01 -1.87822938e-01 5.70985451e-02 -3.11502665e-01 -8.26635659e-01 -3.39294076e-01 2.86626488e-01 5.34561753e-01 2.31142417e-01 -9.82829276e-03 1.56837881e-01 3.59150112e-01 2.42695868e-01 1.05519369e-01 3.02039504e-01 1.23126352e+00 -6.30879641e-01 -7.66464055e-01 -1.00371473e-01 6.13418281e-01 -3.24512035e-01 2.74907500e-01 -5.22888124e-01 5.17127573e-01 2.58821994e-01 9.66277957e-01 3.65435600e-01 -6.63368940e-01 2.87462026e-01 5.15888274e-01 4.55767542e-01 -6.62810564e-01 -5.85961714e-02 -2.81283200e-01 6.57689348e-02 -1.32889366e+00 -7.55536318e-01 -1.07147217e+00 -1.24979508e+00 2.49953747e-01 -2.80420016e-02 3.11820563e-02 3.74294847e-01 1.19494879e+00 2.51919866e-01 2.70126104e-01 5.63190341e-01 -6.73851669e-01 -4.77051765e-01 -5.75589418e-01 -7.38188982e-01 9.58684742e-01 8.94059464e-02 -1.07513773e+00 -3.93932492e-01 2.43036807e-01]
[8.491657257080078, 0.89389967918396]
ca566611-74bc-4fa1-95f9-638d7f9eb965
som-ncscm-an-efficient-neural-chinese
null
null
https://aclanthology.org/2021.emnlp-main.33
https://aclanthology.org/2021.emnlp-main.33.pdf
SOM-NCSCM : An Efficient Neural Chinese Sentence Compression Model Enhanced with Self-Organizing Map
Sentence Compression (SC), which aims to shorten sentences while retaining important words that express the essential meanings, has been studied for many years in many languages, especially in English. However, improvements on Chinese SC task are still quite few due to several difficulties: scarce of parallel corpora, different segmentation granularity of Chinese sentences, and imperfect performance of syntactic analyses. Furthermore, entire neural Chinese SC models have been under-investigated so far. In this work, we construct an SC dataset of Chinese colloquial sentences from a real-life question answering system in the telecommunication domain, and then, we propose a neural Chinese SC model enhanced with a Self-Organizing Map (SOM-NCSCM), to gain a valuable insight from the data and improve the performance of the whole neural Chinese SC model in a valid manner. Experimental results show that our SOM-NCSCM can significantly benefit from the deep investigation of similarity among data, and achieve a promising F1 score of 89.655 and BLEU4 score of 70.116, which also provides a baseline for further research on Chinese SC task.
['Cungen Cao', 'Yanan Cao', 'Jicun Li', 'Yu Liu', 'Shi Wang', 'Kangli Zi']
null
null
null
null
emnlp-2021-11
['sentence-compression']
['natural-language-processing']
[ 3.06341887e-01 -2.38588318e-01 2.57262170e-01 -6.34626746e-01 -6.72233999e-01 -6.16637319e-02 1.35571510e-01 1.30926594e-01 -6.41105831e-01 7.56659508e-01 5.67847192e-01 -2.62317747e-01 4.71196324e-03 -8.23495507e-01 -2.18761444e-01 -7.48668849e-01 3.06993634e-01 3.05809081e-01 3.40215385e-01 -4.53853756e-01 4.29021448e-01 -2.23799393e-01 -1.34470212e+00 5.97461164e-01 1.71667707e+00 8.21453154e-01 8.32618296e-01 3.69558573e-01 -5.01883984e-01 9.03641701e-01 -1.05986094e+00 -5.28380573e-01 -5.17415047e-01 -8.23166370e-01 -1.04236341e+00 -1.09499171e-01 9.72642936e-03 7.96429440e-02 -1.74911827e-01 1.44243479e+00 6.23585165e-01 9.93993878e-02 2.48216107e-01 -4.01291728e-01 -9.49002981e-01 1.14054120e+00 -1.03706785e-01 1.95379496e-01 1.78165048e-01 -3.17240775e-01 1.01277328e+00 -7.48620629e-01 4.82941031e-01 1.16127074e+00 3.03844899e-01 7.33860493e-01 -5.53052008e-01 -7.19882488e-01 1.76764950e-01 5.33728182e-01 -1.23804700e+00 -2.99382895e-01 6.63203835e-01 3.02448422e-01 1.10203791e+00 4.55281705e-01 5.42739868e-01 8.48322392e-01 1.98280096e-01 1.17139363e+00 7.44869709e-01 -5.20631075e-01 1.96195722e-01 9.90871638e-02 3.58607411e-01 1.45777777e-01 -2.21593566e-02 -6.40624285e-01 -2.74125248e-01 3.68286163e-01 2.43241012e-01 3.38300169e-02 -4.34825122e-01 3.88604879e-01 -1.09771693e+00 6.57043993e-01 4.40382838e-01 8.01911116e-01 -5.99945486e-02 -2.37042621e-01 5.98784566e-01 4.28533316e-01 5.24608791e-01 4.96535629e-01 -6.07117295e-01 -3.81318748e-01 -6.86973870e-01 3.14205766e-01 6.30185008e-01 1.44826949e+00 3.12757403e-01 6.85108826e-02 -1.99452654e-01 1.07310390e+00 -1.98340610e-01 4.51522797e-01 9.64006603e-01 -7.83337951e-01 9.13591921e-01 6.23999000e-01 -4.73390639e-01 -1.04741156e+00 -3.69005173e-01 -5.76342583e-01 -1.33245444e+00 -8.58652413e-01 -1.43676370e-01 -2.48982415e-01 -4.22508627e-01 1.47937965e+00 -2.29427934e-01 -8.39303210e-02 3.70728105e-01 8.06833804e-01 9.92795467e-01 9.28343773e-01 -1.42642438e-01 -3.70263278e-01 1.25298488e+00 -1.28173161e+00 -1.08904123e+00 -2.16066241e-01 7.71080315e-01 -7.56756246e-01 1.39358425e+00 4.93792802e-01 -1.12388349e+00 -7.32931376e-01 -9.47439015e-01 -2.35002816e-01 -3.28906983e-01 1.62015080e-01 5.23462117e-01 6.97674394e-01 -8.76172960e-01 5.43259978e-01 -6.67174280e-01 -4.06229317e-01 3.58835191e-01 1.20848127e-01 1.52707711e-01 -3.81770074e-01 -1.49517345e+00 7.89587975e-01 8.83822381e-01 3.34968209e-01 -1.46141052e-01 -3.17888498e-01 -6.95538282e-01 4.22171950e-01 5.85001469e-01 -3.62715632e-01 1.18066287e+00 -7.45938480e-01 -1.42917895e+00 3.63972276e-01 -3.36730480e-01 -5.15406847e-01 9.77234915e-02 -3.72602880e-01 -8.09722424e-01 2.56083876e-01 -2.46764868e-02 5.56239188e-01 3.71845454e-01 -7.42923319e-01 -6.76413774e-01 -4.01666611e-01 -2.89660096e-01 4.31754351e-01 -8.56440961e-01 1.83721125e-01 -9.59893227e-01 -8.82345557e-01 2.06100807e-01 -6.47283375e-01 -2.55533725e-01 -8.81757200e-01 -3.84019762e-01 -4.06893492e-01 7.80312777e-01 -8.64934266e-01 1.82395089e+00 -2.01634169e+00 1.47597328e-01 -3.03005636e-01 -5.57126701e-02 6.57309055e-01 -1.86444476e-01 4.44846302e-01 2.00142354e-01 3.16088855e-01 -6.63349211e-01 -3.92524630e-01 -3.47792536e-01 2.37214074e-01 -3.78932685e-01 -1.32765636e-01 3.36370736e-01 1.13849497e+00 -8.82907808e-01 -7.01543748e-01 -1.17859960e-01 -9.63138640e-02 -5.35151303e-01 1.01917572e-01 -4.11813676e-01 2.20346227e-01 -6.86755002e-01 5.60433030e-01 8.01409662e-01 3.57610397e-02 2.08944917e-01 3.90171409e-01 -5.32614104e-02 4.87966448e-01 -7.04652250e-01 1.98364341e+00 -2.44831160e-01 5.45982778e-01 -2.02986673e-01 -1.02415097e+00 1.11254096e+00 7.14705437e-02 1.24544576e-01 -1.25429165e+00 6.68902025e-02 1.51361078e-01 4.52552550e-02 -7.17648327e-01 8.68607163e-01 -1.07530288e-01 -1.96653351e-01 3.18957180e-01 -1.54725522e-01 -4.13899451e-01 4.48522598e-01 2.78305054e-01 8.76860559e-01 -1.76977023e-01 -6.23478889e-02 -4.96105641e-01 9.56566215e-01 9.89730656e-02 7.69647300e-01 5.46257734e-01 -1.92358196e-01 9.57591832e-01 4.33524460e-01 -1.70809682e-03 -8.88313830e-01 -6.58844411e-01 -2.57487178e-01 8.09969962e-01 3.31225961e-01 -6.54548824e-01 -1.19859195e+00 -6.00852251e-01 -5.24326742e-01 9.69002187e-01 -2.03854173e-01 -4.47531670e-01 -1.03195667e+00 -8.88545215e-01 7.90309429e-01 6.29943967e-01 9.47974145e-01 -1.65411150e+00 -1.45340189e-01 4.39381361e-01 -6.52483702e-01 -1.28904283e+00 -5.67676902e-01 1.28965089e-02 -1.07787192e+00 -6.64720416e-01 -6.98901415e-01 -1.44318318e+00 5.71813762e-01 5.68145096e-01 1.09859908e+00 4.75510955e-01 1.34976238e-01 -3.94946456e-01 -9.02092457e-01 -6.68299675e-01 -3.75622481e-01 5.37677109e-01 -9.37322825e-02 -2.96869248e-01 8.16596687e-01 -2.12090790e-01 -2.57745653e-01 2.94179231e-01 -1.22121847e+00 1.61544800e-01 5.79922616e-01 8.19090009e-01 3.93396229e-01 3.54049891e-01 1.06689799e+00 -1.14871156e+00 1.02635813e+00 -2.05036089e-01 -3.45051438e-01 4.32755232e-01 -6.60859585e-01 -3.63967493e-02 1.03634703e+00 -9.12568197e-02 -1.37818336e+00 -5.11032104e-01 -6.62455618e-01 1.91898793e-01 -9.80107710e-02 9.99054253e-01 -6.22531831e-01 5.24653792e-01 3.13199550e-01 7.51843631e-01 -1.74506411e-01 -7.66854167e-01 8.44315588e-02 1.11097026e+00 7.27440715e-01 -4.49083775e-01 2.38420278e-01 1.42443050e-02 -6.51364684e-01 -9.46496069e-01 -1.01492143e+00 -5.89476705e-01 -7.14919686e-01 2.42792338e-01 7.82218218e-01 -7.95692146e-01 -5.58045268e-01 6.20191395e-01 -1.33991170e+00 6.37414604e-02 4.95823584e-02 4.14933294e-01 -3.05897500e-02 8.80265951e-01 -8.42526317e-01 -4.95346487e-01 -7.94747770e-01 -1.00833917e+00 7.84622192e-01 4.03859019e-01 1.56504333e-01 -7.20940769e-01 -1.87179804e-01 4.41155344e-01 5.95900357e-01 -3.83181930e-01 1.22083783e+00 -6.25995457e-01 -3.67604136e-01 -1.89221740e-01 -3.09838384e-01 8.56072903e-01 1.46426465e-02 -3.38329881e-01 -7.82257497e-01 -2.26118743e-01 6.25491917e-01 -2.76538283e-01 1.25495028e+00 1.89992651e-01 1.68137884e+00 -1.49365947e-01 2.44827382e-03 3.46387684e-01 1.20954108e+00 5.29081047e-01 1.04046047e+00 3.24485540e-01 5.06226063e-01 6.73391521e-01 6.54578686e-01 1.68337315e-01 5.92610300e-01 2.58350194e-01 1.94613054e-01 1.61197737e-01 6.34754300e-02 -2.26462990e-01 2.70170242e-01 2.04720616e+00 -2.70632119e-03 -3.24933976e-01 -7.36488760e-01 5.22011578e-01 -1.79029512e+00 -9.09033120e-01 -3.71771336e-01 1.88191903e+00 1.12373376e+00 4.98784214e-01 -3.82544696e-01 8.93528387e-02 7.73738921e-01 1.49601161e-01 -4.13943142e-01 -2.30443463e-01 -5.82800686e-01 5.81426173e-02 -2.41176337e-02 1.24740064e-01 -1.11074126e+00 1.23128021e+00 5.45859146e+00 1.34611607e+00 -8.72413218e-01 -1.08501501e-01 6.57863557e-01 2.46266142e-01 -3.66172105e-01 -7.57441297e-02 -9.41172004e-01 7.68483639e-01 9.10433531e-01 -3.65601778e-01 3.09698552e-01 6.84157431e-01 1.36915803e-01 -2.67419852e-02 -5.97768188e-01 8.88353944e-01 4.92727250e-01 -1.32657385e+00 1.66100934e-01 -2.61653870e-01 9.32551146e-01 2.26411559e-02 -7.81060681e-02 7.91914523e-01 -1.06276929e-01 -9.43531930e-01 4.70793784e-01 4.57712948e-01 4.17346656e-01 -1.00372350e+00 1.19849038e+00 9.36850727e-01 -1.05983305e+00 6.03393763e-02 -1.24273956e+00 -1.83068171e-01 1.86597839e-01 5.91698945e-01 -2.24211484e-01 9.44790721e-01 8.67410719e-01 1.07319641e+00 -8.87977779e-01 1.03665268e+00 -2.81596899e-01 9.88684833e-01 -2.80517209e-02 -7.47170329e-01 4.39564109e-01 -4.79661733e-01 1.35318339e-01 1.45492351e+00 2.21279755e-01 2.67191231e-01 -2.91196734e-01 7.46486187e-01 -1.53923064e-01 4.50879991e-01 -1.50644600e-01 -1.05358832e-01 4.69119787e-01 1.07593322e+00 -5.84506929e-01 -4.96925116e-01 -3.85138392e-01 9.19808805e-01 3.98766816e-01 2.81716585e-01 -5.84295571e-01 -9.10190940e-01 2.10848272e-01 -6.10335469e-01 3.14204395e-01 -2.08678812e-01 -7.18552649e-01 -1.69216120e+00 3.09104562e-01 -1.02841187e+00 2.33366400e-01 -6.34746969e-01 -1.13532674e+00 9.94638503e-01 -3.50759685e-01 -1.49772203e+00 -2.76000821e-03 -2.41541222e-01 -8.44846785e-01 8.94404411e-01 -1.67077243e+00 -8.00020576e-01 -2.47756049e-01 4.72467571e-01 1.07172680e+00 -3.78907681e-01 8.27211022e-01 4.35238153e-01 -8.84666979e-01 6.87638164e-01 6.67043090e-01 2.56179333e-01 6.01425052e-01 -1.15893447e+00 6.95226490e-01 8.97524416e-01 1.29847705e-01 9.15183902e-01 2.41068825e-01 -6.22117281e-01 -1.26493669e+00 -1.22170019e+00 1.33373523e+00 -2.32832804e-01 4.51664805e-01 -5.32471657e-01 -1.23902500e+00 2.05204532e-01 4.21040833e-01 -5.96051872e-01 4.62621868e-01 1.10915132e-01 3.42831016e-01 -2.08266601e-01 -6.16183460e-01 7.43610740e-01 8.76683474e-01 -4.18501407e-01 -9.31881726e-01 3.22646648e-01 1.34158909e+00 -2.01894209e-01 -6.43986106e-01 3.22251409e-01 2.88190451e-02 -9.70012546e-01 6.78083718e-01 -5.78381062e-01 7.15926290e-01 -9.11166817e-02 -4.14901339e-02 -1.20855737e+00 -1.73338905e-01 -2.82177657e-01 2.32482731e-01 1.42415619e+00 4.91725862e-01 -2.98529953e-01 8.72972131e-01 2.84895778e-01 -6.34182453e-01 -8.97647798e-01 -7.50829220e-01 -7.00993478e-01 4.68352228e-01 -7.23359048e-01 9.07276213e-01 7.68079042e-01 3.00010175e-01 7.22819686e-01 -3.16267043e-01 -2.35623971e-01 1.34530261e-01 2.21883565e-01 2.77710021e-01 -9.87489998e-01 2.41918154e-02 -5.81508577e-01 6.83044121e-02 -1.57900345e+00 1.39332920e-01 -8.64589751e-01 1.61047801e-01 -1.40679872e+00 4.01005238e-01 -2.91718304e-01 -2.51715362e-01 -4.79925387e-02 -7.03306854e-01 -2.66062915e-01 2.86383390e-01 2.57238865e-01 -9.04831469e-01 1.06735623e+00 1.56602728e+00 -1.85918495e-01 -7.43529666e-03 1.08779609e-01 -9.09962296e-01 6.15193307e-01 8.01557064e-01 -3.05961549e-01 -3.10644329e-01 -9.46289480e-01 3.06071676e-02 1.47783250e-01 -3.67716879e-01 -9.23730254e-01 6.48580074e-01 3.36416773e-02 2.65967876e-01 -1.07108295e+00 -8.42608104e-04 -6.21800005e-01 -6.59725785e-01 4.78508800e-01 -4.73069698e-01 1.13668911e-01 1.26481531e-02 4.88300383e-01 -8.04736674e-01 -5.66810727e-01 4.27415073e-01 -2.92459756e-01 -7.52282023e-01 1.28406480e-01 -2.82769471e-01 3.62376392e-01 4.95439172e-01 -7.86602348e-02 -3.08102131e-01 -3.56308043e-01 -2.61811286e-01 7.08056629e-01 -2.27041412e-02 6.89207792e-01 8.29716027e-01 -1.27243400e+00 -8.86241078e-01 2.77591377e-01 1.58488423e-01 3.78292233e-01 4.36152726e-01 5.80175221e-01 -8.32768798e-01 9.02206957e-01 1.25955552e-01 -3.38699192e-01 -9.89991605e-01 5.95568776e-01 -8.71725976e-02 -3.28324348e-01 -7.72642612e-01 7.29619324e-01 -5.95799228e-03 -6.74370766e-01 4.05150115e-01 -5.83695054e-01 -7.45806217e-01 -1.54434234e-01 8.10937226e-01 2.98499852e-01 3.71135890e-01 -2.92639852e-01 -1.22900255e-01 4.08653527e-01 -4.28342730e-01 2.18641952e-01 1.39018178e+00 -4.10633594e-01 -4.49213326e-01 7.75763243e-02 1.30710065e+00 -1.54183134e-01 -5.75395405e-01 -4.84847367e-01 3.76339942e-01 -2.62442976e-01 -2.56567478e-01 -6.72474861e-01 -9.56325889e-01 1.23498905e+00 -7.82081783e-02 2.34100848e-01 1.42376423e+00 -1.97087064e-01 1.42234707e+00 9.52953696e-01 2.37843052e-01 -1.52801394e+00 2.31427979e-02 1.08176231e+00 8.10325503e-01 -1.14658380e+00 -2.15690687e-01 -4.15648729e-01 -1.01443446e+00 1.26012826e+00 6.95839107e-01 -5.36350720e-03 4.52675551e-01 2.16932520e-02 8.39870647e-02 1.12354487e-01 -6.96785510e-01 -9.70461443e-02 1.02802597e-01 3.69830012e-01 5.11247277e-01 -5.94907068e-02 -6.59996033e-01 1.18151748e+00 -5.09996593e-01 -2.01690331e-01 6.36202455e-01 9.19677138e-01 -7.52952695e-01 -1.17432368e+00 -6.78106851e-04 5.85516095e-01 -5.51868081e-01 -6.15793586e-01 -3.26586992e-01 3.19233388e-01 -1.19651549e-01 1.22437394e+00 5.32840565e-03 -5.40683270e-01 4.22482103e-01 -2.27721885e-01 -1.48601877e-02 -7.64455318e-01 -5.99920511e-01 3.21699440e-01 -9.82734337e-02 -1.00076713e-01 -3.52771789e-01 -6.55890882e-01 -1.46783471e+00 -1.63039088e-01 -5.71703672e-01 5.02888083e-01 3.90396804e-01 1.18540394e+00 2.03892782e-01 5.52128077e-01 7.55210578e-01 -1.75989226e-01 -5.46980619e-01 -1.37158716e+00 -6.62701130e-01 2.59166896e-01 -7.54795298e-02 1.65512696e-01 -3.85560282e-03 -9.19045229e-03]
[10.887847900390625, 9.310729026794434]
fef47d9e-1a36-40e7-b442-7f9e37703787
an-off-the-grid-approach-to-multi-compartment
2011.11193
null
https://arxiv.org/abs/2011.11193v1
https://arxiv.org/pdf/2011.11193v1.pdf
An off-the-grid approach to multi-compartment magnetic resonance fingerprinting
We propose a novel numerical approach to separate multiple tissue compartments in image voxels and to estimate quantitatively their nuclear magnetic resonance (NMR) properties and mixture fractions, given magnetic resonance fingerprinting (MRF) measurements. The number of tissues, their types or quantitative properties are not a-priori known, but the image is assumed to be composed of sparse compartments with linearly mixed Bloch magnetisation responses within voxels. Fine-grid discretisation of the multi-dimensional NMR properties creates large and highly coherent MRF dictionaries that can challenge scalability and precision of the numerical methods for (discrete) sparse approximation. To overcome these issues, we propose an off-the-grid approach equipped with an extended notion of the sparse group lasso regularisation for sparse approximation using continuous (non-discretised) Bloch response models. Further, the nonlinear and non-analytical Bloch responses are approximated by a neural network, enabling efficient back-propagation of the gradients through the proposed algorithm. Tested on simulated and in-vivo healthy brain MRF data, we demonstrate effectiveness of the proposed scheme compared to the baseline multicompartment MRF methods.
['Clarice Poon', 'Mohammad Golbabaee']
2020-11-23
null
null
null
null
['magnetic-resonance-fingerprinting']
['medical']
[ 3.71067405e-01 -2.40352601e-02 8.86816382e-02 -2.66314447e-01 -7.63396561e-01 -2.79388368e-01 3.10364425e-01 4.00466844e-02 -2.30014294e-01 9.90488112e-01 1.92398369e-01 5.65282181e-02 -3.80213439e-01 -2.46866569e-01 -6.72155797e-01 -1.18663204e+00 -6.04642332e-01 8.89448345e-01 -2.85910107e-02 2.51336128e-01 4.12510335e-02 7.27235019e-01 -1.16848648e+00 4.60700780e-01 9.09303725e-01 1.10550559e+00 2.83494025e-01 5.41803360e-01 1.15154497e-01 1.17718375e+00 -1.65218022e-02 5.58600247e-01 2.89514720e-01 -2.86812991e-01 -9.41505492e-01 1.64675981e-01 5.47837675e-01 -2.18217865e-01 -3.72215688e-01 1.27503228e+00 5.59656560e-01 4.81980175e-01 1.07693911e+00 -5.18746078e-01 -4.06209022e-01 7.12004602e-01 -8.00589919e-01 4.07313764e-01 -2.20385827e-02 -1.42017365e-01 7.82200545e-02 -9.71886158e-01 6.73243403e-01 8.76922786e-01 1.00986969e+00 3.00047547e-01 -1.76126897e+00 -4.86935496e-01 -1.26644429e-02 -1.14803366e-01 -1.60790396e+00 -5.62012851e-01 6.86533034e-01 -9.13562298e-01 7.93416321e-01 4.75053221e-01 6.05802774e-01 6.04966402e-01 5.24682999e-01 2.86242187e-01 1.87653565e+00 -2.49285176e-02 4.10162270e-01 1.06526941e-01 2.98967600e-01 6.28181577e-01 6.31431714e-02 -2.95424797e-02 -2.47242879e-02 -9.48357821e-01 1.00531662e+00 3.29289921e-02 -6.81802511e-01 -5.14324486e-01 -1.35881412e+00 6.95338547e-01 3.63736659e-01 6.11836672e-01 -9.42667603e-01 2.31573004e-02 4.79175240e-01 -1.85155243e-01 5.62031209e-01 2.75948346e-01 4.41948809e-02 6.99497819e-01 -1.62110996e+00 2.63565660e-01 6.62756145e-01 6.38288975e-01 6.58494592e-01 3.60934496e-01 -1.92105711e-01 8.69670630e-01 3.21344972e-01 6.62146628e-01 6.90443456e-01 -1.10182488e+00 -1.57828823e-01 -1.95033178e-01 2.14421764e-01 -1.16785991e+00 -5.89152396e-01 -6.12950265e-01 -1.71368897e+00 -1.62154540e-01 4.68591720e-01 1.12001888e-01 -9.46408570e-01 1.38690579e+00 8.11692178e-01 6.22853100e-01 -2.78568953e-01 1.42109561e+00 6.29513562e-01 6.38410330e-01 1.59610078e-01 -8.41638207e-01 1.23017490e+00 -7.82131910e-01 -7.38769650e-01 3.92658561e-01 4.39770758e-01 -5.45476377e-01 3.04450333e-01 4.15640086e-01 -1.22672033e+00 -2.94198871e-01 -6.38964236e-01 8.21872279e-02 9.82977077e-02 -8.77207965e-02 7.61130810e-01 3.12373757e-01 -9.49169755e-01 7.51128733e-01 -1.04496980e+00 2.49386072e-01 2.60114163e-01 5.25268018e-01 -4.16756421e-01 -2.36723021e-01 -1.01979303e+00 7.40381837e-01 1.72967657e-01 5.95050156e-01 -1.20998740e+00 -1.21227038e+00 -6.41854644e-01 -3.72131199e-01 -1.20645307e-01 -6.88321471e-01 4.79546547e-01 -6.75700068e-01 -1.32335603e+00 6.84049189e-01 -4.42229696e-02 -4.05203283e-01 5.28197408e-01 3.96029621e-01 -1.72598809e-01 6.40070438e-01 8.24592859e-02 3.18822801e-01 1.02141178e+00 -1.48665845e+00 2.56561786e-01 -4.42228973e-01 -5.41972876e-01 9.81859863e-02 2.93172091e-01 -3.14255767e-02 4.50719863e-01 -7.22423732e-01 5.68302274e-01 -8.15784574e-01 -7.57346392e-01 -2.40859568e-01 -4.10769135e-01 5.63558877e-01 3.20399791e-01 -1.21071148e+00 8.57855022e-01 -1.90204751e+00 3.69647264e-01 5.43085933e-01 7.11283743e-01 -1.88545763e-01 1.26681045e-01 -1.41580880e-01 -2.90327936e-01 -4.20137018e-01 -6.49150372e-01 9.77675430e-04 -3.04754913e-01 1.27454668e-01 8.54259133e-02 1.23649716e+00 -2.57174134e-01 4.90001202e-01 -1.01883328e+00 -6.07680976e-01 7.74216130e-02 9.53909814e-01 -4.59497422e-01 6.38223290e-02 2.70763665e-01 1.22634149e+00 -7.42034793e-01 4.64586854e-01 1.19551265e+00 -4.48386937e-01 4.68281716e-01 -6.90068662e-01 -1.31123304e-01 -1.85910627e-01 -1.31687701e+00 1.54581928e+00 -2.24808082e-01 -2.59933263e-01 9.09808815e-01 -1.46848857e+00 5.52207291e-01 5.85434675e-01 1.20233345e+00 -4.37981337e-01 3.74427997e-02 5.62932253e-01 4.36614715e-02 -2.94385582e-01 1.73194051e-01 -7.65996873e-01 4.76170003e-01 3.45811188e-01 1.45232812e-01 -6.58791065e-02 -7.25103468e-02 1.18059143e-01 9.05451119e-01 2.43762205e-03 7.68562183e-02 -1.15096128e+00 9.31038320e-01 1.77886635e-02 3.00924927e-01 7.22051978e-01 -1.03351459e-01 4.82170045e-01 2.62010042e-02 -4.30216372e-01 -1.16832447e+00 -6.39402688e-01 -8.98723304e-01 5.74878514e-01 -1.47572368e-01 3.01519305e-01 -9.77800608e-01 -1.57480836e-01 1.47561133e-01 2.90985834e-02 -8.53612959e-01 3.33211958e-01 -6.13294482e-01 -1.33054972e+00 3.13144118e-01 5.68683185e-02 -5.39411269e-02 -6.57685816e-01 -2.66043812e-01 5.69153786e-01 -4.22358930e-01 -1.18896103e+00 -4.14703280e-01 4.82042760e-01 -1.09742093e+00 -7.85847783e-01 -1.24286842e+00 -8.05445015e-01 9.05562043e-01 6.10887818e-02 9.41349685e-01 -5.30611128e-02 -3.32289308e-01 3.71458799e-01 6.24807663e-02 4.22305733e-01 -4.31102425e-01 -4.33494359e-01 3.18938494e-01 4.08126682e-01 -2.80450404e-01 -9.45061684e-01 -8.76544893e-01 1.97217062e-01 -8.07559669e-01 2.74467133e-02 3.93640041e-01 1.03230202e+00 1.35982871e+00 4.06612493e-02 4.08939868e-01 -9.93579984e-01 3.87734324e-01 -8.69087815e-01 -3.30276281e-01 1.53506085e-01 -3.15597534e-01 -5.08294329e-02 6.60917342e-01 -8.35368752e-01 -9.11595166e-01 1.25493824e-01 2.89180189e-01 -7.13170290e-01 -1.10229008e-01 6.42913282e-01 4.06373560e-01 -7.52857864e-01 5.99680364e-01 6.68592393e-01 1.28642738e-01 -5.07137597e-01 2.37839684e-01 4.62509900e-01 7.34909952e-01 -1.13551402e+00 3.04610670e-01 6.88884377e-01 4.31959450e-01 -1.01748443e+00 -5.30501544e-01 -6.98301911e-01 -8.10077250e-01 -1.91393122e-01 5.92205882e-01 -8.52113903e-01 -6.62400663e-01 4.40085679e-01 -8.40653598e-01 -4.75344896e-01 -2.26477593e-01 8.39996755e-01 -7.90002763e-01 8.70361686e-01 -1.20642400e+00 -9.54814315e-01 -6.43994629e-01 -1.34920943e+00 9.02616680e-01 -4.45528984e-01 3.34364697e-02 -9.64521825e-01 1.10265419e-01 4.41468716e-01 7.24615574e-01 6.35144711e-01 1.01354480e+00 -2.56159872e-01 -2.84272194e-01 6.15916550e-02 -1.50101572e-01 2.37277836e-01 -2.28364959e-01 -7.64462650e-01 -7.61583447e-01 -5.64591408e-01 7.50702858e-01 -2.83644199e-01 6.31564915e-01 1.07815063e+00 1.18003154e+00 -5.50605834e-01 -3.13593954e-01 7.55045533e-01 1.54732597e+00 -1.96670100e-01 3.38461280e-01 -2.10765764e-01 8.95922661e-01 6.77579641e-01 1.27612799e-01 6.27218246e-01 2.28606895e-01 4.39053357e-01 -4.01162542e-02 -1.16111942e-01 -1.70713574e-01 2.77453899e-01 8.65496248e-02 1.31626010e+00 -4.33126003e-01 4.14812565e-01 -8.06004047e-01 3.89110982e-01 -1.67946923e+00 -9.46955740e-01 -6.34049058e-01 2.17711329e+00 1.01982558e+00 -6.39341593e-01 1.29385486e-01 -1.04704387e-01 7.23978519e-01 -2.11254224e-01 -6.56702876e-01 1.75546914e-01 -5.02485819e-02 1.93628907e-01 6.59359694e-01 7.92830050e-01 -8.56197238e-01 3.12821150e-01 6.82915211e+00 7.54499435e-01 -1.38019907e+00 5.94428182e-01 9.32330906e-01 1.38913244e-01 -2.98312873e-01 -2.41900891e-01 -3.59219432e-01 4.80516106e-01 1.10151935e+00 2.36287490e-01 9.09159005e-01 2.97125429e-01 3.44700843e-01 -1.75425991e-01 -7.16759324e-01 9.28433299e-01 -6.68328777e-02 -1.11863208e+00 -1.02500811e-01 1.89348713e-01 8.05011094e-01 2.60658592e-01 -7.81262442e-02 -9.73887593e-02 -2.73085088e-02 -1.15447688e+00 7.08464682e-01 8.84497046e-01 9.15587246e-01 -3.84724677e-01 5.04183173e-01 3.78174692e-01 -9.97661710e-01 2.26608396e-01 -6.93907976e-01 3.29658300e-01 2.71350116e-01 1.17562568e+00 -5.30609310e-01 5.59592128e-01 5.75132310e-01 5.01533031e-01 7.66754299e-02 8.82519662e-01 6.80432498e-01 5.38542092e-01 -5.47933638e-01 5.17140090e-01 2.07723901e-01 -6.41159356e-01 5.59684455e-01 1.31546342e+00 3.23794127e-01 6.07455552e-01 4.48137939e-01 1.18976462e+00 4.07970905e-01 4.61098015e-01 -1.17950894e-01 -6.32033199e-02 -4.30264920e-02 1.62514687e+00 -1.05795372e+00 -6.20862722e-01 -2.09081605e-01 6.61178410e-01 2.25413784e-01 7.26614714e-01 -5.89315295e-01 3.81323606e-01 4.88407426e-02 7.70881593e-01 1.38799757e-01 -2.51915246e-01 4.49512713e-02 -1.38702989e+00 -2.78056502e-01 -9.83068407e-01 1.60806745e-01 -6.20925069e-01 -1.68900752e+00 7.67645121e-01 1.14227399e-01 -6.42068565e-01 6.97375685e-02 -4.86786574e-01 1.50214642e-01 1.19458902e+00 -1.54879296e+00 -1.17744851e+00 -3.54321264e-02 7.97465563e-01 -2.78490275e-01 3.08976173e-01 8.23516369e-01 5.89109600e-01 -1.55240372e-02 8.62944424e-02 5.53615451e-01 -1.71060175e-01 3.06546330e-01 -1.12516999e+00 -5.82260787e-01 2.10616961e-01 -6.38537407e-01 8.54449987e-01 8.16565752e-01 -8.81056905e-01 -1.57914877e+00 -1.06054258e+00 1.84176669e-01 1.11315459e-01 7.70424604e-01 -3.59403074e-01 -1.39082825e+00 5.21754146e-01 -1.17052771e-01 8.95522654e-01 7.64766574e-01 -2.83988893e-01 -2.78388290e-03 2.38936856e-01 -1.75121093e+00 -1.90314874e-01 5.25557220e-01 -4.19979960e-01 -1.48746312e-01 8.73480082e-01 9.37173888e-02 -6.03667676e-01 -1.72780836e+00 4.10214841e-01 5.48804641e-01 -6.40864193e-01 1.17361832e+00 -6.29119337e-01 3.51918042e-02 -4.40334022e-01 -4.24249530e-01 -1.14115620e+00 -8.66342664e-01 -5.31985223e-01 -3.85847718e-01 5.01833200e-01 -1.30614489e-01 -5.72100282e-01 6.79337978e-01 5.44946611e-01 -2.32949480e-01 -8.35230350e-01 -1.01974070e+00 -5.06543577e-01 4.24397707e-01 9.83258784e-02 3.45055610e-01 1.38466227e+00 4.66758274e-02 5.30697703e-02 -7.15179622e-01 2.73436934e-01 1.39790058e+00 2.63611376e-02 -8.40481818e-02 -1.08976722e+00 -5.41317284e-01 -6.03726618e-02 -3.20563495e-01 -8.96739721e-01 3.09330553e-01 -1.30895054e+00 1.15509570e-01 -1.04715335e+00 6.78404927e-01 -8.80585730e-01 -5.59892952e-01 2.01194972e-01 2.78803349e-01 4.13943052e-01 -3.11173081e-01 4.78021681e-01 -3.86161745e-01 3.37944061e-01 1.78028142e+00 -2.45098040e-01 1.05113052e-01 -4.63000149e-01 -1.99868664e-01 4.39552873e-01 1.49763584e-01 -5.28489947e-01 -3.03253710e-01 2.68611070e-02 -1.48250699e-01 8.41712475e-01 5.19061744e-01 -1.07554960e+00 2.13129193e-01 -7.49815851e-02 5.75552881e-01 -3.12987745e-01 2.91898012e-01 -8.31751049e-01 7.63900340e-01 4.20010090e-01 -3.28493774e-01 -3.95088941e-01 -9.46658911e-05 4.69659269e-01 -1.28502429e-01 -1.70199543e-01 1.12490213e+00 -6.12891257e-01 1.97640404e-01 5.45833468e-01 -6.18211448e-01 -8.21375754e-03 5.20955622e-01 -1.55747235e-01 3.06439728e-01 -3.99603732e-02 -1.42775667e+00 -1.57765463e-01 1.90954298e-01 -6.40901387e-01 4.62346226e-01 -1.43135464e+00 -7.68697023e-01 1.35978550e-01 -5.42874753e-01 3.40635143e-02 9.72096205e-01 1.76574874e+00 -5.24492741e-01 2.79202253e-01 -3.66925508e-01 -7.13262618e-01 -6.16880476e-01 8.27765107e-01 8.88457298e-01 -5.19518077e-01 -8.55724394e-01 5.31684756e-01 4.55051750e-01 -7.97024190e-01 -1.10494569e-01 -4.61214691e-01 -1.26434952e-01 -1.54854491e-01 6.31908596e-01 5.16734779e-01 1.38516903e-01 -1.28863466e+00 -2.86740720e-01 5.38666844e-01 -4.05208617e-02 1.94298550e-01 1.61464751e+00 -2.50979155e-01 -7.99379647e-01 3.30794364e-01 1.18666291e+00 -2.57246822e-01 -9.94498909e-01 -3.57680798e-01 -4.15455580e-01 -1.39789701e-01 5.82617342e-01 -7.99733877e-01 -1.14101243e+00 7.54499972e-01 6.72941387e-01 9.73602831e-02 8.01258445e-01 -3.12566817e-01 8.27289522e-01 -1.58684831e-02 5.92392504e-01 -6.82923853e-01 -6.28639817e-01 1.82486609e-01 1.00233567e+00 -8.32855880e-01 2.55282909e-01 -6.97763026e-01 -2.26248533e-01 9.98094738e-01 -4.39545512e-02 -2.66610503e-01 1.02185452e+00 5.65091848e-01 -1.60997167e-01 -4.54432875e-01 -1.57081738e-01 5.96337378e-01 3.99131715e-01 6.23364151e-01 5.96322119e-01 2.25272909e-01 -3.60495299e-01 5.51402569e-01 2.23874554e-01 1.13910154e-01 3.23355824e-01 6.92864835e-01 -2.22295210e-01 -4.52880591e-01 -7.47387350e-01 6.45696998e-01 -6.50521040e-01 -2.00911224e-01 4.07680273e-01 9.14492831e-02 -2.65872478e-03 3.35160047e-01 -2.69403040e-01 2.92916566e-01 -9.59307104e-02 -1.24516338e-01 9.55324829e-01 -4.03470039e-01 -5.02072692e-01 6.56756163e-01 -3.26392829e-01 -4.96236682e-01 -6.29501343e-01 -7.53080368e-01 -1.41896284e+00 -1.80888355e-01 -3.33143681e-01 5.41446805e-01 4.31743026e-01 8.29240322e-01 1.84155792e-01 4.17809606e-01 4.96747375e-01 -1.38098574e+00 -8.08702588e-01 -1.04979765e+00 -1.35895634e+00 4.62861151e-01 4.04244006e-01 -7.97929823e-01 -3.00225317e-01 6.77001402e-02]
[13.442456245422363, -2.418820381164551]
9afe063b-b907-4a6c-b250-cb41ad94771c
adaptive-re-ranking-with-a-corpus-graph
2208.08942
null
https://arxiv.org/abs/2208.08942v1
https://arxiv.org/pdf/2208.08942v1.pdf
Adaptive Re-Ranking with a Corpus Graph
Search systems often employ a re-ranking pipeline, wherein documents (or passages) from an initial pool of candidates are assigned new ranking scores. The process enables the use of highly-effective but expensive scoring functions that are not suitable for use directly in structures like inverted indices or approximate nearest neighbour indices. However, re-ranking pipelines are inherently limited by the recall of the initial candidate pool; documents that are not identified as candidates for re-ranking by the initial retrieval function cannot be identified. We propose a novel approach for overcoming the recall limitation based on the well-established clustering hypothesis. Throughout the re-ranking process, our approach adds documents to the pool that are most similar to the highest-scoring documents up to that point. This feedback process adapts the pool of candidates to those that may also yield high ranking scores, even if they were not present in the initial pool. It can also increase the score of documents that appear deeper in the pool that would have otherwise been skipped due to a limited re-ranking budget. We find that our Graph-based Adaptive Re-ranking (GAR) approach significantly improves the performance of re-ranking pipelines in terms of precision- and recall-oriented measures, is complementary to a variety of existing techniques (e.g., dense retrieval), is robust to its hyperparameters, and contributes minimally to computational and storage costs. For instance, on the MS MARCO passage ranking dataset, GAR can improve the nDCG of a BM25 candidate pool by up to 8% when applying a monoT5 ranker.
['Craig Macdonald', 'Nicola Tonellotto', 'Sean MacAvaney']
2022-08-18
null
null
null
null
['passage-ranking']
['natural-language-processing']
[ 1.04368180e-01 -1.88288346e-01 -1.89513221e-01 -5.86708030e-03 -1.25301588e+00 -1.07284737e+00 6.72500670e-01 7.95621276e-01 -5.93194366e-01 7.22495377e-01 4.62021410e-01 -2.57385880e-01 -8.44959676e-01 -9.60262239e-01 -3.72421414e-01 -2.48888239e-01 -1.21532403e-01 1.06503129e+00 9.01909411e-01 -4.53019202e-01 7.53258348e-01 4.34457630e-01 -1.79876053e+00 5.54252207e-01 9.22654331e-01 5.27493060e-01 2.93187410e-01 5.95542073e-01 -7.04826936e-02 2.36489236e-01 -7.41239429e-01 -1.15209199e-01 3.03277969e-02 -2.96314269e-01 -1.05909467e+00 -5.90062618e-01 5.37121177e-01 -3.31375569e-01 -1.52828783e-01 7.50371516e-01 4.00140375e-01 5.27370632e-01 5.43486953e-01 -4.76863742e-01 -4.55383688e-01 4.98100072e-01 -3.53614807e-01 4.23723876e-01 6.53600872e-01 -5.53778648e-01 1.47741914e+00 -9.90755320e-01 9.11274910e-01 9.06596661e-01 4.81505364e-01 1.48263067e-01 -1.06464446e+00 -3.61468434e-01 1.16864713e-02 1.26469567e-01 -1.43452609e+00 -2.27060035e-01 1.59975961e-01 -1.30129278e-01 1.14589906e+00 7.37166405e-01 5.41858971e-01 9.07657370e-02 -2.50542521e-01 2.37126037e-01 6.27180398e-01 -6.06605172e-01 1.84839278e-01 -8.77805576e-02 3.77332181e-01 3.88964176e-01 4.48845088e-01 -1.71544984e-01 -6.55642390e-01 -5.51028907e-01 2.68607110e-01 -1.49798289e-01 -1.10000730e-01 -1.41250923e-01 -1.11950946e+00 7.07060575e-01 4.65442270e-01 5.60365081e-01 -2.60916620e-01 -2.06824243e-01 1.71799839e-01 2.76367813e-01 3.14867854e-01 1.08449972e+00 -4.98828411e-01 -7.77056962e-02 -1.36292064e+00 2.29744419e-01 7.07598686e-01 6.75604105e-01 8.95789862e-01 -9.79723334e-01 -6.99081421e-01 1.05408156e+00 2.85711974e-01 9.27268341e-02 6.22161508e-01 -9.09975350e-01 2.35735312e-01 1.16715848e+00 1.93540931e-01 -9.45817947e-01 -3.92910212e-01 -6.46920264e-01 -1.58158928e-01 -1.46025419e-01 2.45632589e-01 4.34721202e-01 -9.70311701e-01 1.28254414e+00 2.44614884e-01 -1.53471157e-01 -2.48474747e-01 7.17547238e-01 8.54238629e-01 5.56794941e-01 -2.08448276e-01 -2.39788339e-01 1.28895593e+00 -7.94785917e-01 -1.71199501e-01 -2.68647745e-02 7.77894139e-01 -1.10897875e+00 1.19077909e+00 2.45732233e-01 -1.14502633e+00 -4.01882648e-01 -1.00725329e+00 -5.01479134e-02 -4.00207549e-01 -7.64071718e-02 4.71054792e-01 4.53377575e-01 -1.71721232e+00 7.54615545e-01 -3.83470327e-01 -4.65925246e-01 -2.05777943e-01 6.24864161e-01 -1.35332689e-01 -2.44619682e-01 -1.29295552e+00 8.57430816e-01 3.68632972e-01 -2.58137017e-01 -3.55318576e-01 -6.20117247e-01 -1.18962653e-01 4.22296286e-01 3.30872416e-01 -7.09200382e-01 1.05776227e+00 -4.35462892e-01 -1.02644765e+00 5.33453405e-01 -5.33806145e-01 -6.14084750e-02 1.38376296e-01 -1.31543040e-01 -3.29906702e-01 4.10289079e-01 3.66641730e-01 4.57965761e-01 2.78463423e-01 -1.12609577e+00 -8.71995330e-01 -2.37518415e-01 9.48179215e-02 6.64585650e-01 -5.78551531e-01 2.64476061e-01 -1.06449747e+00 -2.98348576e-01 3.76074731e-01 -9.26784575e-01 -2.49113277e-01 -7.83623040e-01 -1.72269404e-01 -3.09792399e-01 3.91297430e-01 -3.81113291e-01 1.85856700e+00 -1.64746904e+00 -1.53913215e-01 9.31221247e-01 2.63264984e-01 2.89408863e-01 -2.69937992e-01 7.30719864e-01 1.93988129e-01 6.28352642e-01 3.71977389e-01 1.91205993e-01 -1.55926466e-01 -1.31558791e-01 -1.30236477e-01 -1.04225107e-01 -3.09472620e-01 6.95347309e-01 -1.35456681e+00 -5.48803091e-01 -9.15634483e-02 2.22081497e-01 -5.75284839e-01 -1.19809419e-01 -3.92961502e-02 2.69525368e-02 -3.61510098e-01 4.45466727e-01 3.18228900e-01 -5.71163476e-01 2.69586623e-01 3.77836451e-02 -1.65004835e-01 8.22575867e-01 -1.07785535e+00 1.11198008e+00 -2.66228795e-01 4.10792947e-01 -4.70577955e-01 -4.03415442e-01 6.98506594e-01 1.03595868e-01 5.77223241e-01 -6.78758621e-01 -5.05443037e-01 6.73091650e-01 -1.37145966e-02 -1.08286003e-02 1.21191752e+00 3.84538621e-01 1.38442501e-01 5.95992684e-01 -2.90091097e-01 -5.49688796e-03 9.58615065e-01 7.62188554e-01 1.59189296e+00 -2.32424542e-01 1.46312686e-02 -1.71549827e-01 5.38961649e-01 2.61495531e-01 1.24908157e-01 1.16643846e+00 4.07636374e-01 7.50777781e-01 3.35678935e-01 -1.17581985e-04 -9.62641239e-01 -8.48964751e-01 -2.62171417e-01 1.50677669e+00 9.54077318e-02 -9.12379086e-01 -3.66743714e-01 -6.05734706e-01 3.90341016e-03 4.76323217e-01 -4.05387431e-01 -2.10111097e-01 -4.14031982e-01 -6.22884989e-01 4.57716435e-01 2.61216015e-01 -2.99016293e-02 -8.81548762e-01 -1.42752513e-01 2.76621163e-01 -4.21130359e-01 -4.51205581e-01 -5.23036778e-01 1.48213312e-01 -1.09689748e+00 -1.13800907e+00 -8.45631063e-01 -5.99661767e-01 9.33326483e-01 5.62318444e-01 1.44864118e+00 7.39879906e-01 1.43442243e-01 4.27219748e-01 -7.46797085e-01 1.63187206e-01 -3.58686417e-01 5.06381691e-01 -8.04750621e-02 -6.23950660e-01 6.26173735e-01 -2.58731358e-02 -9.17261899e-01 6.79038763e-01 -9.28592920e-01 -6.73515201e-01 5.10619462e-01 8.39476347e-01 7.47297108e-01 3.01642478e-01 5.70289493e-01 -8.15351307e-01 1.03058541e+00 -4.82775390e-01 -5.80622613e-01 5.14196932e-01 -1.22185659e+00 2.00855806e-01 3.26208919e-01 -3.04812968e-01 -7.41113961e-01 -2.10147142e-01 8.56631547e-02 2.51763612e-01 3.81464988e-01 1.04508531e+00 4.20074761e-01 -4.99544218e-02 9.30969834e-01 -1.48932248e-01 -3.69211942e-01 -6.53453290e-01 2.64638424e-01 6.82573378e-01 8.26891810e-02 -3.16309988e-01 8.02395463e-01 1.79713458e-01 4.14629206e-02 -3.47085148e-01 -6.72956884e-01 -1.11708736e+00 -5.74664593e-01 -1.31177932e-01 2.64071256e-01 -7.17994511e-01 -3.86968791e-01 -1.53162226e-01 -8.14748883e-01 4.72726300e-02 -2.67394751e-01 4.62041855e-01 9.20616686e-02 6.53205097e-01 -5.44769883e-01 -4.80358779e-01 -4.33063775e-01 -9.76274490e-01 9.71941471e-01 2.58024484e-01 -6.80045784e-01 -7.34129131e-01 3.52183193e-01 3.17872941e-01 3.89760345e-01 -3.98749590e-01 1.18943834e+00 -9.07103300e-01 -5.75563729e-01 -6.87174976e-01 -1.16067871e-01 -1.89719439e-01 8.21977556e-02 1.86998427e-01 -5.42997241e-01 -4.32388067e-01 -7.01519847e-01 -3.99093190e-03 1.16129863e+00 3.52257580e-01 7.19863176e-01 -4.15484309e-02 -7.40609229e-01 3.37797366e-02 1.19555306e+00 2.40759909e-01 7.02274799e-01 8.88554871e-01 2.09214151e-01 6.24603868e-01 6.56465828e-01 1.13452539e-01 2.40829423e-01 9.32293117e-01 1.39106557e-01 3.80730699e-03 -2.70977318e-02 -2.12863371e-01 2.62085587e-01 7.68565834e-01 -1.81900963e-01 -5.05551994e-01 -9.06705141e-01 6.98538959e-01 -1.72407150e+00 -1.06875992e+00 -4.09253031e-01 2.96500731e+00 9.43959355e-01 2.29396716e-01 2.04060674e-01 2.75488436e-01 8.14836681e-01 -3.03837270e-01 -1.80771217e-01 -2.61415005e-01 -2.53068041e-02 4.19380218e-01 4.50625360e-01 7.42693961e-01 -6.94935381e-01 9.97032642e-01 7.14810944e+00 8.80984962e-01 -5.52110434e-01 -2.67591238e-01 3.25676858e-01 -3.73043716e-01 -6.07531369e-01 3.27481598e-01 -1.07227778e+00 3.61289561e-01 1.07469511e+00 -4.80308115e-01 5.54718554e-01 6.04579329e-01 7.06288442e-02 -3.88870835e-01 -1.04706812e+00 3.78226787e-01 3.10449321e-02 -1.31142330e+00 8.20108280e-02 2.13573098e-01 7.52776623e-01 2.19493121e-01 1.35892168e-01 2.60626316e-01 5.08460760e-01 -8.89042377e-01 1.99499458e-01 2.92299509e-01 7.98088074e-01 -7.58106411e-01 8.63913417e-01 6.61282167e-02 -1.12918460e+00 -7.76256621e-02 -5.35887480e-01 2.20161319e-01 -7.08124936e-02 8.37899625e-01 -1.12625504e+00 5.36526680e-01 8.47584665e-01 7.72000849e-02 -9.29709613e-01 1.48773324e+00 -8.15655217e-02 6.00329876e-01 -5.93907535e-01 -2.91637927e-01 1.81827724e-01 3.48163135e-02 5.05202532e-01 1.07687604e+00 5.39363146e-01 -8.77874419e-02 -1.17961846e-01 1.56879410e-01 -1.49152622e-01 4.67108458e-01 -2.86369264e-01 1.13999143e-01 8.20208073e-01 1.20003974e+00 -8.89801621e-01 -4.96041328e-01 -1.38178751e-01 6.39165163e-01 2.80939579e-01 3.23691875e-01 -3.25493127e-01 -7.13368952e-01 3.14511843e-02 4.63731468e-01 1.21389084e-01 6.45426214e-02 -1.59702636e-03 -7.04444826e-01 1.15970477e-01 -7.99230039e-01 7.69544780e-01 -6.68581128e-01 -9.42393064e-01 7.63155878e-01 1.89143382e-02 -1.41401303e+00 -5.10888696e-01 -1.13554820e-01 -2.78254241e-01 9.98113155e-01 -1.43936861e+00 -4.38388199e-01 -1.78617418e-01 2.78614193e-01 1.32841751e-01 -5.93259223e-02 8.54797244e-01 4.99158204e-01 6.05418570e-02 6.26798093e-01 6.21847510e-01 -3.38256419e-01 8.95185173e-01 -1.36056626e+00 1.72631577e-01 6.82124376e-01 3.64794612e-01 1.26261485e+00 3.79771680e-01 -8.97996664e-01 -7.20553756e-01 -7.21118212e-01 1.42121351e+00 -6.69694662e-01 4.96837616e-01 2.03698590e-01 -9.77043927e-01 1.71878785e-01 -2.61828959e-01 -5.23622632e-01 8.52467477e-01 6.95753038e-01 -2.49999002e-01 -1.07576646e-01 -8.53454411e-01 7.00440884e-01 9.77075577e-01 -4.86180276e-01 -6.23737276e-01 5.52900493e-01 3.54796618e-01 -2.03825429e-01 -7.98494875e-01 3.70199502e-01 7.34486282e-01 -7.71301568e-01 1.21160972e+00 -4.65110719e-01 2.73787111e-01 -5.65630615e-01 1.00702956e-01 -1.26353097e+00 -8.78145337e-01 -6.23875856e-01 -1.55360624e-01 1.13112473e+00 9.74778533e-01 -3.68588001e-01 8.48375857e-01 6.37289166e-01 -4.07520458e-02 -6.35261059e-01 -8.26552927e-01 -8.43238294e-01 -4.08194140e-02 8.25070217e-02 7.25262880e-01 5.94029725e-01 2.94112682e-01 3.96880090e-01 1.04540147e-01 -6.24071136e-02 2.06719488e-01 1.16680957e-01 4.42234963e-01 -1.61267245e+00 -3.40182930e-01 -6.79102063e-01 -8.50757062e-02 -1.25068438e+00 -4.05742705e-01 -1.18214309e+00 7.42368922e-02 -2.10270810e+00 5.44832647e-01 -7.88943470e-01 -8.15385044e-01 4.99537438e-01 -5.55280268e-01 5.40302992e-01 -2.16895808e-02 7.83819139e-01 -9.91967797e-01 -1.10576287e-01 8.76923263e-01 4.88367490e-02 -7.08300948e-01 2.65724659e-02 -9.15276587e-01 5.00223100e-01 5.39001703e-01 -7.05207348e-01 -3.28059465e-01 -3.79940212e-01 8.01582098e-01 -3.02224487e-01 -1.36946276e-01 -6.48380518e-01 5.36823869e-01 6.86556101e-02 5.30777216e-01 -8.53651047e-01 5.55236228e-02 -4.97229815e-01 2.36055747e-01 3.55009139e-01 -5.44392228e-01 4.33428466e-01 5.93523160e-02 4.66674805e-01 -2.48765051e-01 -7.08467901e-01 2.49452233e-01 -1.59863904e-01 -5.95047653e-01 -1.50430948e-01 -4.81687427e-01 2.44390890e-02 4.75577742e-01 -5.25686085e-01 -6.45966530e-01 -4.29545671e-01 -4.90927547e-01 3.19906026e-01 6.84274733e-01 2.53690243e-01 4.99288112e-01 -1.11959004e+00 -5.40524602e-01 -2.19144240e-01 3.98660094e-01 -2.34743729e-01 -1.59663856e-01 5.69963455e-01 -4.88341421e-01 6.86896265e-01 3.79171014e-01 -3.70915234e-01 -1.51691532e+00 2.63234109e-01 -4.24408428e-02 -1.01550198e+00 -2.82785505e-01 8.67971241e-01 -2.25153580e-01 -3.44153017e-01 5.01795746e-02 2.62593441e-02 -5.31921089e-01 2.65157014e-01 5.94325483e-01 6.62386477e-01 6.06819868e-01 -5.23614883e-01 -6.67944610e-01 6.33811712e-01 -4.78004187e-01 -4.68113214e-01 1.10348380e+00 -6.25565574e-02 -3.86470914e-01 -1.12814158e-02 1.04109788e+00 6.08238578e-01 -3.42518747e-01 -7.02762082e-02 5.77885032e-01 -6.62785530e-01 2.42486566e-01 -1.18244553e+00 -6.24875605e-01 2.21006453e-01 3.64223480e-01 5.00280499e-01 1.12387562e+00 1.38616905e-01 5.85865736e-01 6.46689832e-01 3.22161198e-01 -1.39185500e+00 2.90350765e-01 7.24560916e-01 5.19167781e-01 -8.48784328e-01 3.79330873e-01 -2.49354571e-01 -3.05410296e-01 1.02749395e+00 2.80678093e-01 3.04306060e-01 3.63038659e-01 -1.44542471e-01 -1.26354232e-01 -3.69307727e-01 -8.20078850e-01 -3.73417348e-01 8.68685782e-01 3.62933576e-01 7.56897271e-01 -2.56488383e-01 -9.38231230e-01 1.79312661e-01 -2.29879931e-01 -7.73433521e-02 3.13290954e-01 7.31068850e-01 -7.90412247e-01 -1.44224250e+00 -4.38277990e-01 1.03685296e+00 -6.04103386e-01 -5.57249725e-01 -6.60762370e-01 3.36290002e-01 -1.55084565e-01 1.38558626e+00 1.10034794e-01 -4.72105056e-01 2.50226736e-01 -1.25259683e-02 2.39159524e-01 -9.32696521e-01 -8.63531291e-01 3.10368866e-01 3.24362904e-01 -3.67334068e-01 -1.53799564e-01 -5.44414163e-01 -1.24978256e+00 -2.06096634e-01 -8.63144457e-01 7.07228422e-01 4.86668766e-01 5.56854784e-01 6.17243946e-01 3.57132226e-01 4.82827038e-01 -4.11866337e-01 -4.65829313e-01 -8.68340552e-01 -3.50482732e-01 3.20608079e-01 -6.67277351e-02 -5.03829241e-01 -4.99917299e-01 -3.31548423e-01]
[11.459677696228027, 7.534677982330322]
78a75cd7-855f-4a39-8bcb-8169a687643a
from-dictations-to-clinical-reports-using
null
null
https://aclanthology.org/N18-3015
https://aclanthology.org/N18-3015.pdf
From dictations to clinical reports using machine translation
A typical workflow to document clinical encounters entails dictating a summary, running speech recognition, and post-processing the resulting text into a formatted letter. Post-processing entails a host of transformations including punctuation restoration, truecasing, marking sections and headers, converting dates and numerical expressions, parsing lists, etc. In conventional implementations, most of these tasks are accomplished by individual modules. We introduce a novel holistic approach to post-processing that relies on machine callytranslation. We show how this technique outperforms an alternative conventional system{---}even learning to correct speech recognition errors during post-processing{---}while being much simpler to maintain.
['David Suendermann-Oeft', 'a', 'Wael Salloum', 'Najmeh Sadoughi', 'Mark Miller', 'Gregory Finley', 'Nico Axtmann', 'Michael Brenndoerfer', 'Erik Edwards', 'Am Robinson']
2018-06-01
null
null
null
naacl-2018-6
['punctuation-restoration']
['natural-language-processing']
[ 8.97620857e-01 2.89092392e-01 -5.34303449e-02 -7.03689039e-01 -1.25487387e+00 -7.09002018e-01 4.76914108e-01 1.17712319e+00 -7.41590798e-01 8.52995396e-01 4.57221329e-01 -1.11613584e+00 -1.00649670e-01 -1.36876956e-01 -4.80511636e-01 -1.99262083e-01 1.46484897e-01 4.61566716e-01 -6.79748505e-02 1.38663307e-01 7.80607343e-01 7.03273773e-01 -9.04312849e-01 8.42878759e-01 5.45929015e-01 7.08926320e-01 -8.89009386e-02 1.30883861e+00 -6.63996875e-01 1.25342989e+00 -9.92188811e-01 -6.55544937e-01 -1.14649475e-01 -4.95666116e-01 -1.25257635e+00 -3.87794971e-02 2.61968404e-01 -2.65524000e-01 -2.29164392e-01 9.92241323e-01 4.25818056e-01 -4.89917770e-02 4.31116104e-01 -4.88588661e-01 -2.76340544e-01 8.30785513e-01 1.88528270e-01 2.31813118e-01 1.14140236e+00 -7.52480030e-02 4.14719850e-01 -5.29143512e-01 4.21228021e-01 1.00282717e+00 9.91890788e-01 6.47167087e-01 -9.42411900e-01 -2.25952521e-01 -2.50161946e-01 -1.41760305e-01 -9.58482981e-01 -7.25846648e-01 6.05645441e-02 -4.36805516e-01 1.46309912e+00 7.37839103e-01 4.41026598e-01 6.72812402e-01 6.73466027e-01 5.34275651e-01 1.11415660e+00 -7.63685644e-01 2.45886058e-01 3.27126905e-02 3.55446041e-01 7.21569002e-01 5.04837604e-03 -2.40562156e-01 -5.50842702e-01 -3.41453671e-01 6.71505928e-01 1.72008067e-01 -3.24302763e-02 6.12632751e-01 -1.54143083e+00 1.77781567e-01 -3.35675061e-01 5.63637652e-02 -4.32101399e-01 -2.11231798e-01 8.63599420e-01 3.66826981e-01 -1.60595551e-01 4.19143885e-01 -4.20737594e-01 -5.65950274e-01 -1.44214320e+00 -1.91764105e-02 1.13081729e+00 1.24000072e+00 3.54994625e-01 -1.90689355e-01 -3.50165933e-01 4.70392942e-01 7.62999281e-02 1.58162281e-01 8.22587729e-01 -6.64124966e-01 6.05305612e-01 2.95263290e-01 1.02658987e-01 -2.55682170e-01 -6.15728676e-01 -1.12303510e-01 -6.58264816e-01 -7.01771677e-02 5.86103976e-01 -4.04866487e-01 -1.36602092e+00 9.94148433e-01 -8.93346295e-02 -9.31061432e-02 1.82851568e-01 5.40677719e-02 9.64990795e-01 5.36028802e-01 4.28624272e-01 -6.32465124e-01 1.50383365e+00 -7.51426220e-01 -1.28534091e+00 -1.81398347e-01 7.08572388e-01 -1.48146713e+00 7.15043306e-01 5.56421757e-01 -1.65052056e+00 -2.57935464e-01 -7.84793258e-01 -3.76429498e-01 -4.65936214e-01 2.74428368e-01 3.86779100e-01 6.12600327e-01 -1.21959686e+00 9.37557936e-01 -9.13026929e-01 -2.80410886e-01 1.82203636e-01 5.47880173e-01 -5.25711954e-01 3.70032489e-01 -7.11904287e-01 1.06647086e+00 4.09678042e-01 9.48262066e-02 -2.35293865e-01 -8.40279222e-01 -1.04216838e+00 -1.93804651e-02 7.81816691e-02 -7.92647779e-01 2.06437111e+00 -5.88361084e-01 -1.72014356e+00 1.08519888e+00 -5.12614608e-01 -5.29673338e-01 7.06470788e-01 -1.47496387e-01 -6.87556624e-01 1.18374594e-01 -1.13326855e-01 3.22409011e-02 8.58247459e-01 -5.81692219e-01 -9.64259028e-01 -1.67253256e-01 -3.73134553e-01 1.08651422e-01 2.37047687e-01 4.76104409e-01 -3.34567994e-01 -8.79682481e-01 3.29473466e-01 -5.26900828e-01 -5.04400671e-01 -3.64171118e-01 -6.70817852e-01 8.54276121e-02 1.26694843e-01 -1.24550974e+00 2.04662156e+00 -2.21138906e+00 -3.83777022e-01 1.65627271e-01 1.50127172e-01 4.45565224e-01 2.68652767e-01 8.31367791e-01 -6.08191729e-01 -2.74516754e-02 -4.66880739e-01 -4.90774006e-01 -1.64229050e-01 1.49693236e-01 -3.86384994e-01 2.88215816e-01 2.01741874e-01 7.62511194e-01 -9.45435464e-01 -7.65572131e-01 3.72478962e-01 2.80683964e-01 -4.61545199e-01 1.93965137e-01 2.40054667e-01 8.71090889e-02 -1.49517721e-02 7.02366412e-01 4.43713069e-01 -9.70925763e-02 3.68294656e-01 1.65943637e-01 -3.35602313e-01 1.12510920e+00 -1.19659173e+00 1.46981442e+00 -3.92566472e-01 5.58986485e-01 9.23025459e-02 -8.65838408e-01 6.75286174e-01 3.01739335e-01 2.63108343e-01 -8.64074603e-02 2.90160477e-01 5.02629936e-01 -3.37888300e-01 -7.99402058e-01 7.97429085e-01 -1.95930257e-01 -2.28487357e-01 2.20119178e-01 -5.83824590e-02 -3.68008643e-01 2.77378827e-01 3.51079762e-01 1.30294967e+00 -2.29320645e-01 1.00444996e+00 2.46706218e-01 7.77490675e-01 1.99363679e-01 1.63856789e-01 8.27329636e-01 -2.95627892e-01 6.71404004e-01 6.25510395e-01 -3.66021723e-01 -1.03240204e+00 -1.12548268e+00 -9.95518416e-02 7.08508790e-01 -7.15180218e-01 -7.61862457e-01 -1.24835479e+00 -5.35373688e-01 -3.11800867e-01 8.65541220e-01 -3.77546042e-01 9.19495896e-02 -6.91514373e-01 -3.23241830e-01 7.73108959e-01 5.76659441e-01 -8.71791020e-02 -1.24374497e+00 -7.33429849e-01 6.30321860e-01 6.35514259e-02 -9.78015661e-01 -8.42452586e-01 8.65091264e-01 -1.01890421e+00 -7.13464081e-01 -5.61883628e-01 -1.23721290e+00 1.00728881e+00 -2.57698953e-01 8.55700672e-01 4.80156988e-02 -5.39075971e-01 2.43196368e-01 -2.15136841e-01 -5.17544627e-01 -9.83734787e-01 -7.67696798e-02 -1.81401670e-01 -4.07454818e-01 4.62977737e-01 -2.62305737e-01 -3.65581781e-01 -4.62337345e-01 -1.06812298e+00 9.21156816e-03 9.04046595e-01 7.56181419e-01 6.87419116e-01 -3.97156715e-01 1.12268567e-01 -1.40393054e+00 9.55117345e-01 1.41701158e-02 -4.42552119e-02 5.54657876e-01 -4.23209101e-01 1.11603089e-01 7.94443846e-01 -1.22309901e-01 -8.64479363e-01 3.91279697e-01 -4.37225968e-01 9.30393338e-02 -6.14462435e-01 3.60224366e-01 2.85970151e-01 3.74670148e-01 5.56594193e-01 5.36947310e-01 2.23505706e-01 -3.22921485e-01 1.91433445e-01 1.26531863e+00 1.21711695e+00 1.77679844e-02 5.14304936e-01 3.66819464e-02 -3.75095695e-01 -7.64277756e-01 -5.12816131e-01 -5.60588002e-01 -7.24441648e-01 6.71040639e-02 7.12174237e-01 -6.74671829e-01 -7.69981205e-01 3.73521745e-01 -1.35232198e+00 -1.26848325e-01 -4.88421589e-01 6.34538889e-01 -6.65984094e-01 7.85778761e-01 -9.08353567e-01 -6.55042648e-01 -5.77257752e-01 -9.22157347e-01 9.11394894e-01 3.77631038e-01 -9.92374599e-01 -8.70920658e-01 -4.72080447e-02 3.32154959e-01 2.50804693e-01 -1.79903522e-01 9.80752766e-01 -1.01268601e+00 1.47230715e-01 -4.48695958e-01 -2.87852399e-02 1.60736889e-01 4.52996910e-01 2.32911825e-01 -6.16746366e-01 1.51693806e-01 7.27104694e-02 2.85105526e-01 4.59441781e-01 3.13767552e-01 1.38399088e+00 -6.50267303e-01 -2.71551818e-01 6.54519558e-01 1.01779687e+00 7.90469408e-01 6.44191742e-01 3.15996170e-01 1.69130445e-01 5.13464630e-01 4.19815689e-01 6.91072226e-01 4.62123722e-01 -1.49480790e-01 -3.14054936e-01 -1.31954346e-02 1.23305134e-01 -2.40797997e-01 3.55621666e-01 1.16149163e+00 3.28079373e-01 1.70336470e-01 -9.59164381e-01 4.36566532e-01 -1.59008622e+00 -9.12627578e-01 -1.07388027e-01 2.09479403e+00 1.28158033e+00 3.11951578e-01 1.53852897e-02 3.64290088e-01 6.88233495e-01 -1.59489423e-01 -1.90349042e-01 -1.16743553e+00 3.85972351e-01 8.40644419e-01 6.82700694e-01 6.35527790e-01 -1.03793383e+00 9.75405514e-01 8.05169964e+00 4.75767553e-01 -1.25240576e+00 -3.74206543e-01 5.33873856e-01 1.69570073e-02 -3.82879488e-02 -2.39003763e-01 -7.69391358e-01 3.94632071e-01 1.34294522e+00 -1.75476924e-01 3.70017737e-01 7.02819228e-01 4.25130695e-01 -1.59845293e-01 -1.33636665e+00 1.30343282e+00 2.13640139e-01 -1.68937397e+00 2.28243172e-01 -5.42959929e-01 1.82919309e-01 -4.65818256e-01 -1.73934847e-01 8.39155540e-02 1.15605265e-01 -9.68758106e-01 7.79374957e-01 5.45005381e-01 9.93341386e-01 -4.07587528e-01 4.85987186e-01 -3.19304168e-02 -6.73315704e-01 5.42876534e-02 5.15380735e-03 -1.02471551e-02 3.36883724e-01 4.32580978e-01 -1.17404604e+00 5.62058866e-01 2.62879074e-01 2.96869934e-01 -3.93418342e-01 1.29925227e+00 -1.53724417e-01 3.35905075e-01 -5.95020093e-02 -4.27373089e-02 1.66497439e-01 2.24063061e-02 1.91074088e-01 2.07310104e+00 4.07823950e-01 4.65794653e-01 6.61420543e-03 -1.23742655e-01 1.29344523e-01 2.85879493e-01 -1.21306397e-01 -4.11268026e-01 4.69628602e-01 1.09293377e+00 -9.88209069e-01 -9.94212806e-01 -2.85082906e-01 1.22049713e+00 -8.20420086e-02 8.42848346e-02 -6.58279300e-01 -1.18006289e+00 4.05209184e-01 1.81229383e-01 2.11685479e-01 -4.26163822e-01 -7.19876885e-01 -8.09518158e-01 1.14569172e-01 -1.24935293e+00 6.95800364e-01 -6.33260489e-01 -7.34445155e-01 8.51986587e-01 -2.89920181e-01 -1.23893690e+00 -3.97940516e-01 -7.51527011e-01 -5.60310423e-01 8.85596931e-01 -1.41315830e+00 -3.53635758e-01 -1.30365118e-01 7.26784825e-01 6.93582594e-01 4.48264517e-02 1.00114429e+00 4.44926918e-01 -5.59171557e-01 6.62608922e-01 -2.51994822e-02 2.21851602e-01 9.87071157e-01 -1.46262527e+00 3.57173890e-01 7.64265597e-01 -1.27426803e-01 1.16120076e+00 6.59693480e-01 -7.87334263e-01 -1.28019440e+00 -7.28407562e-01 1.63293707e+00 -1.97260126e-01 7.10715413e-01 7.58138523e-02 -8.86886060e-01 8.62754107e-01 3.52922529e-01 -5.00084400e-01 1.04937840e+00 -4.23028976e-01 1.20746465e-02 -4.74099219e-02 -1.03221846e+00 5.85723877e-01 6.60027385e-01 -7.99913526e-01 -1.17464638e+00 5.61296880e-01 3.81807178e-01 -1.04820597e+00 -7.90636718e-01 -1.40653118e-01 3.71252239e-01 -3.79434615e-01 5.24131477e-01 -1.03102672e+00 5.35323322e-01 -2.78711736e-01 2.13256642e-01 -1.01785314e+00 -4.18606699e-02 -1.59538877e+00 2.92226374e-01 9.99133229e-01 7.21254170e-01 -5.25606036e-01 6.06469512e-01 8.54875445e-01 -6.65635407e-01 -3.18026483e-01 -7.02295840e-01 -2.84247935e-01 -3.43085796e-01 -4.40651923e-01 4.14670080e-01 9.08366323e-01 8.12604010e-01 -9.76044536e-02 -1.12976499e-01 4.76893745e-02 2.34768018e-01 -3.09425771e-01 3.24694097e-01 -5.17046511e-01 -6.31008819e-02 -5.88821948e-01 -1.78109810e-01 -8.90305817e-01 -3.21477562e-01 -8.29026043e-01 2.56194830e-01 -1.90367162e+00 -3.45938027e-01 -1.37123078e-01 -2.30850548e-01 6.91080332e-01 -1.97961271e-01 -2.94392020e-01 -1.02439299e-01 -6.25353754e-02 -4.71654177e-01 -5.63598216e-01 7.14256048e-01 1.65349498e-01 -4.80506510e-01 8.98742378e-02 -7.10449517e-01 5.92532218e-01 7.21453249e-01 -6.15431368e-01 1.04800627e-01 -1.64766446e-01 6.46973550e-02 4.84246910e-01 -1.81344584e-01 -8.94981444e-01 8.34121048e-01 -3.59381884e-01 4.29410487e-01 -8.35739195e-01 -1.66232869e-01 -5.39045095e-01 1.20896928e-01 6.93201423e-01 -5.59166491e-01 7.29052246e-01 4.48156208e-01 1.05569519e-01 -2.76198119e-01 -6.38201773e-01 5.14282346e-01 -3.51764262e-01 -5.91883063e-01 -2.70534545e-01 -1.13987541e+00 -1.51382610e-01 9.69612718e-01 -3.82053494e-01 -1.59614459e-01 -1.03608929e-01 -1.24753630e+00 -1.03201494e-01 5.08762151e-02 1.47620589e-01 4.73654360e-01 -6.17595494e-01 -3.80664736e-01 4.32543188e-01 -3.50089759e-01 -2.72826554e-04 2.45546266e-01 1.11229551e+00 -1.39956200e+00 6.28559709e-01 -1.48362771e-01 -3.52744281e-01 -1.69914067e+00 3.83222848e-01 1.49169609e-01 -4.31753933e-01 -6.91189289e-01 6.16331041e-01 -5.20231843e-01 -1.72622174e-01 6.31325066e-01 -9.99943614e-01 -1.49892032e-01 3.34765613e-01 9.88871574e-01 1.91731632e-01 7.93799400e-01 -1.04664095e-01 -6.96317911e-01 3.13536555e-01 -4.76087034e-01 -2.07041278e-01 1.08075321e+00 -1.16249003e-01 -2.07370535e-01 2.12145701e-01 8.45381141e-01 1.28981531e-01 -6.72316253e-01 -1.87532559e-01 4.29705709e-01 -9.28619469e-04 -3.52943927e-01 -1.08164155e+00 -2.75111973e-01 6.66861773e-01 1.29761115e-01 1.21749096e-01 1.16858053e+00 -4.14696336e-01 7.56717980e-01 6.01417422e-01 -1.85333759e-01 -1.36094904e+00 -2.78104722e-01 5.84923804e-01 4.86251622e-01 -6.29764318e-01 4.84463982e-02 -3.49177629e-01 -5.76706052e-01 1.65188015e+00 -9.89091620e-02 3.03416461e-01 6.57185733e-01 8.68635833e-01 5.95975757e-01 1.85852364e-01 -6.81543827e-01 2.18823880e-01 -5.27540594e-02 3.52439821e-01 7.35626578e-01 1.05060026e-01 -6.21036708e-01 5.83922565e-01 -6.94417119e-01 2.33552620e-01 8.35920811e-01 1.51254332e+00 -2.97776610e-01 -1.23720717e+00 -6.72763109e-01 6.76900625e-01 -1.01423728e+00 -4.92022097e-01 -4.22988862e-01 4.22114134e-01 -2.99410701e-01 8.93873990e-01 1.67425647e-01 -1.77286029e-01 6.02794647e-01 4.63916600e-01 4.59560066e-01 -9.16023791e-01 -1.37772942e+00 3.13174367e-01 2.67578244e-01 -3.93970519e-01 1.67112909e-02 -1.05629575e+00 -1.61484110e+00 -3.61184478e-01 3.12966406e-01 2.05709830e-01 8.25388551e-01 9.82002974e-01 2.45617941e-01 1.01950514e+00 1.98694527e-01 -3.31283897e-01 -8.67572188e-01 -6.25852704e-01 2.41488945e-02 3.02801311e-01 7.04429030e-01 2.09787309e-01 -2.26108328e-01 4.66789275e-01]
[8.617256164550781, 8.649327278137207]
0b7a9405-c7ed-4efe-a120-91b64b00083a
defending-against-poisoning-attacks-in-open
2212.10002
null
https://arxiv.org/abs/2212.10002v2
https://arxiv.org/pdf/2212.10002v2.pdf
Defending Against Misinformation Attacks in Open-Domain Question Answering
Recent work in open-domain question answering (ODQA) has shown that adversarial poisoning of the search collection can cause large drops in accuracy for production systems. However, little to no work has proposed methods to defend against these attacks. To do so, we rely on the intuition that redundant information often exists in large corpora. To find it, we introduce a method that uses query augmentation to search for a diverse set of passages that could answer the original question but are less likely to have been poisoned. We integrate these new passages into the model through the design of a novel confidence method, comparing the predicted answer to its appearance in the retrieved contexts (what we call \textit{Confidence from Answer Redundancy}, i.e. CAR). Together these methods allow for a simple but effective way to defend against poisoning attacks that provides gains of nearly 20\% exact match across varying levels of data poisoning/knowledge conflicts.
['Benjamin Van Durme', 'Dawn Lawrie', 'Nathaniel Weir', 'Aleem Khan', 'Orion Weller']
2022-12-20
null
null
null
null
['data-poisoning', 'open-domain-question-answering']
['adversarial', 'natural-language-processing']
[-1.31190792e-01 2.52560794e-01 1.39550030e-01 -7.87040964e-02 -1.51425648e+00 -1.19472039e+00 5.47503114e-01 4.12458360e-01 -4.86860782e-01 8.62612486e-01 3.13190490e-01 -5.07415175e-01 -1.85448378e-02 -8.55498612e-01 -9.39792156e-01 -3.28401357e-01 2.48019293e-01 7.95655429e-01 9.35736656e-01 -6.70730472e-01 4.32125479e-01 3.80923957e-01 -1.32783377e+00 5.29178798e-01 9.76568699e-01 5.62810898e-01 -3.85189742e-01 7.62095988e-01 -1.77824661e-01 1.24383986e+00 -1.45682383e+00 -1.01162970e+00 4.03734297e-01 -4.47526485e-01 -1.28178775e+00 -5.47419131e-01 7.29991555e-01 -5.17569363e-01 -3.74566644e-01 9.49596405e-01 7.06495702e-01 7.88150802e-02 4.27707821e-01 -1.12244833e+00 -7.33764350e-01 7.17919171e-01 -1.54063329e-01 7.70610809e-01 7.93723524e-01 3.20412546e-01 1.20755982e+00 -5.86038709e-01 7.00718164e-01 1.32349443e+00 4.40079272e-01 8.93551409e-01 -1.29586411e+00 -7.68915653e-01 -2.83886611e-01 2.34142482e-01 -1.14780223e+00 -4.51599270e-01 4.38550085e-01 5.61619885e-02 8.76104534e-01 5.87869465e-01 -3.37446034e-02 1.24608982e+00 -3.87493260e-02 7.45732605e-01 8.60353410e-01 -5.03013492e-01 2.07895160e-01 3.25413376e-01 2.45816946e-01 5.55321038e-01 2.84200490e-01 -3.33836824e-02 -6.86627567e-01 -8.40447903e-01 -4.11118045e-02 -6.98566020e-01 -2.26714462e-01 -2.72309393e-01 -6.37876034e-01 1.30273926e+00 3.24909210e-01 2.69424796e-01 -4.86014448e-02 -1.96222831e-02 4.05724674e-01 7.57062614e-01 3.25385392e-01 1.22777200e+00 -6.20663345e-01 5.00341319e-02 -5.11835396e-01 9.65626597e-01 1.42905676e+00 5.64886570e-01 4.20493186e-01 -4.06318367e-01 -4.13917333e-01 6.04117990e-01 6.49916530e-02 5.23742378e-01 2.98497051e-01 -1.18938267e+00 7.01312065e-01 4.47701931e-01 5.35228968e-01 -9.69910324e-01 -8.81552398e-02 -1.24909282e-01 1.71565861e-01 -7.99734071e-02 8.03157032e-01 -2.01703981e-01 -5.04851639e-01 1.91072202e+00 6.57269716e-01 -1.14439756e-01 3.11144203e-01 7.80775130e-01 5.31801581e-01 3.87205034e-01 2.50973552e-01 1.70698673e-01 1.40327752e+00 -7.23842502e-01 -5.75226545e-01 -2.12708414e-01 9.08902586e-01 -6.77347004e-01 1.23301435e+00 2.53636092e-01 -1.07556629e+00 -1.15307663e-02 -1.27641773e+00 -1.13454454e-01 -4.56109196e-01 -9.30100560e-01 3.14635545e-01 1.01851666e+00 -7.33823895e-01 4.25175071e-01 -4.05158460e-01 -1.99543625e-01 3.23560178e-01 1.76169246e-01 -6.79356903e-02 -2.17843801e-01 -1.63071990e+00 1.26587856e+00 2.39941150e-01 -7.07421064e-01 -8.76894593e-01 -7.38587916e-01 -5.78329384e-01 -7.86566287e-02 8.05457294e-01 -8.19603920e-01 1.47488213e+00 -5.87326288e-01 -9.45028126e-01 7.60123193e-01 2.22979840e-02 -8.80271852e-01 3.29670459e-01 -6.43755436e-01 -4.24696475e-01 4.92316008e-01 2.57247627e-01 5.06564677e-01 8.87150705e-01 -1.20431769e+00 -4.40149635e-01 -4.88384634e-01 4.54505682e-01 1.89798549e-01 -4.23156828e-01 4.82766062e-01 -1.13779768e-01 -6.53226793e-01 -2.82620788e-01 -7.48714030e-01 -5.85282817e-02 -3.11815321e-01 -5.21177888e-01 -4.58738685e-01 8.13293755e-01 -8.09088469e-01 1.37902927e+00 -1.96717978e+00 -1.89187855e-01 2.75277495e-01 2.26447001e-01 6.29376829e-01 -1.88752398e-01 6.69330776e-01 2.54298657e-01 6.39974654e-01 -1.43058211e-01 6.51358664e-02 1.14944272e-01 2.80620217e-01 -1.02724481e+00 2.20779583e-01 3.42078745e-01 9.41922247e-01 -8.85273576e-01 -5.18896818e-01 -3.85393322e-01 -7.12737814e-02 -8.49620938e-01 1.79251924e-01 -9.14252162e-01 3.42693686e-01 -7.26305485e-01 6.02946162e-01 2.96055466e-01 5.78310713e-03 -1.59117207e-01 3.59091192e-01 4.02382880e-01 7.97224104e-01 -7.63781607e-01 1.39402139e+00 3.97586823e-02 1.46297053e-01 1.69235244e-01 -4.73885208e-01 6.76978588e-01 2.78544754e-01 -4.13908847e-02 -8.21457982e-01 1.45146638e-01 4.89200443e-01 6.16438091e-02 -6.25743508e-01 6.61961913e-01 -1.91562727e-01 -3.20523560e-01 7.39734113e-01 -1.09199524e-01 -4.03792530e-01 2.82880455e-01 6.61712170e-01 1.72792089e+00 -3.22181791e-01 -2.60493398e-01 -1.07456736e-01 5.92314124e-01 5.63262701e-01 2.42059469e-01 1.25307131e+00 -3.33499521e-01 2.80537993e-01 5.75363457e-01 -2.23293453e-01 -1.11640477e+00 -1.10179853e+00 4.15971130e-02 1.18799639e+00 1.76449165e-01 -2.99712449e-01 -1.00399089e+00 -9.56340849e-01 1.94268614e-01 1.10233307e+00 -3.86662245e-01 -5.76377571e-01 -7.48621702e-01 -3.34263593e-01 1.41728806e+00 2.20815077e-01 3.82639199e-01 -1.26643205e+00 -6.11070335e-01 9.85263810e-02 -5.14952838e-01 -8.68232250e-01 -4.31675464e-01 1.74025595e-01 -6.12540603e-01 -1.23771477e+00 -2.62729287e-01 -4.28001106e-01 5.71508929e-02 -4.69007380e-02 1.58452034e+00 4.57621217e-01 -1.78193748e-01 4.92869258e-01 -5.59997082e-01 -6.75757289e-01 -1.00757945e+00 1.50299594e-01 -3.44890589e-03 -4.88601267e-01 7.32269764e-01 -2.79962420e-01 -4.37186718e-01 4.02410060e-01 -1.24651444e+00 -9.99248505e-01 1.12276815e-01 7.30655193e-01 2.92545587e-01 -1.76481977e-01 1.01009119e+00 -1.10923445e+00 1.22834146e+00 -7.72164762e-01 -4.29648697e-01 4.66197819e-01 -4.19502378e-01 3.59787226e-01 6.19266868e-01 -5.35803378e-01 -6.34403884e-01 -4.03189093e-01 -3.99639398e-01 -4.74206597e-01 -9.25834775e-02 3.16620976e-01 -7.65321329e-02 -7.18025863e-02 1.41929615e+00 -1.93820581e-01 4.49103303e-02 -5.27953804e-01 7.28722632e-01 4.38850105e-01 6.23335719e-01 -7.97200084e-01 1.08769286e+00 1.85428336e-01 -5.04681349e-01 -3.48328561e-01 -1.13931632e+00 -5.02095461e-01 4.69551757e-02 3.77106965e-01 8.47510815e-01 -5.41886568e-01 -5.74109435e-01 2.20020320e-02 -1.20793080e+00 -1.08101390e-01 -5.20823121e-01 -1.51360616e-01 -4.22955751e-01 5.58230877e-01 -7.32822359e-01 -8.71114254e-01 -4.39368278e-01 -6.67839646e-01 6.78567171e-01 2.59599206e-03 -5.50382257e-01 -5.92735112e-01 3.42954576e-01 8.48128796e-01 3.02469760e-01 6.77284077e-02 1.35004234e+00 -1.57409346e+00 -4.63352948e-01 -4.63956863e-01 2.51766473e-01 4.23030794e-01 -3.58903378e-01 -3.89767736e-01 -9.45054471e-01 -1.72113433e-01 3.72834563e-01 -9.02935326e-01 7.07507610e-01 -4.44551438e-01 7.79031456e-01 -6.18091524e-01 -1.11780830e-01 -1.05975471e-01 1.08300638e+00 1.06004514e-01 8.91575634e-01 4.84677643e-01 1.92851618e-01 6.77363098e-01 6.68328822e-01 7.80464709e-02 1.63967591e-02 5.57457447e-01 4.44833934e-01 4.03837919e-01 7.74138123e-02 -5.26623309e-01 2.80841351e-01 1.60145983e-01 7.74270296e-01 -5.38507402e-01 -7.78409243e-01 6.87811077e-01 -1.54388750e+00 -1.27475846e+00 6.27876595e-02 2.16397166e+00 1.08485377e+00 3.87601644e-01 2.30882496e-01 1.76603287e-01 4.98229861e-01 -1.09692465e-03 -7.19876766e-01 -7.16785133e-01 -2.58873403e-01 5.30624807e-01 3.70563924e-01 5.29242635e-01 -1.13091779e+00 1.01625204e+00 7.21908426e+00 9.13313270e-01 -2.83077925e-01 1.37789175e-01 4.20193404e-01 -2.09984872e-02 -5.75057566e-01 -2.00978909e-02 -6.85369551e-01 2.27788135e-01 1.23500359e+00 -1.39916748e-01 3.56463015e-01 7.68352926e-01 -4.27609622e-01 -2.26908967e-01 -1.02877843e+00 2.69706160e-01 3.08012664e-01 -1.24249017e+00 3.49677712e-01 -1.76346645e-01 5.66497803e-01 -8.60269889e-02 1.46017611e-01 5.82891762e-01 9.07511353e-01 -9.20543313e-01 6.45916045e-01 1.17156394e-01 1.71443403e-01 -8.44983935e-01 7.03079939e-01 7.67058909e-01 -3.08820575e-01 -4.37073559e-01 -2.77463496e-01 8.37817043e-02 -4.80019003e-02 4.15911712e-02 -9.33315277e-01 4.11377937e-01 7.19204962e-01 -3.29906285e-01 -7.74981022e-01 7.82421112e-01 -2.72333026e-01 7.60782957e-01 -4.69264030e-01 -2.17991993e-01 1.82640031e-01 3.13479662e-01 7.89725959e-01 6.56445324e-01 -1.74897358e-01 2.44235948e-01 2.03658305e-02 8.35881293e-01 -3.85329843e-01 3.54655981e-02 -9.21602130e-01 -3.55472714e-02 7.81230390e-01 5.45267999e-01 -1.56783968e-01 -3.07944387e-01 -3.08991253e-01 8.61718297e-01 5.22486925e-01 1.28498018e-01 -8.58737171e-01 -3.89785111e-01 5.25974751e-01 9.96911451e-02 3.65494013e-01 1.66317984e-01 -1.77842408e-01 -1.00871599e+00 2.89605886e-01 -1.65129721e+00 8.95783305e-01 -6.53844237e-01 -1.63768661e+00 5.12063980e-01 -5.37401177e-02 -5.59333205e-01 -3.95145178e-01 -1.98065862e-01 -3.68723333e-01 8.54672611e-01 -1.03852379e+00 -5.14620125e-01 4.77654040e-01 8.52337003e-01 4.36387122e-01 -1.21763363e-01 1.07593477e+00 2.32181534e-01 -2.30433658e-01 8.90839994e-01 -1.36898935e-01 2.41612405e-01 9.05433953e-01 -1.22499931e+00 4.24488038e-01 9.51046407e-01 4.74735171e-01 8.91337693e-01 1.01191854e+00 -8.26400995e-01 -1.14285672e+00 -7.35978246e-01 1.03080523e+00 -1.51973259e+00 8.99001002e-01 -1.53779462e-01 -1.16736150e+00 4.76457149e-01 7.06714690e-02 -3.83458972e-01 5.71724653e-01 -4.16092165e-02 -1.10795712e+00 2.19746202e-01 -1.54922807e+00 8.31461072e-01 7.05374956e-01 -8.47185910e-01 -1.18712854e+00 3.71681750e-01 1.27923548e+00 -2.71230817e-01 -5.11326373e-01 3.21441114e-01 1.34010240e-01 -7.88604259e-01 1.13312304e+00 -1.06038344e+00 3.06689531e-01 -2.11066872e-01 -3.63478541e-01 -8.50413501e-01 1.07709698e-01 -7.02221692e-01 -4.64357227e-01 1.02968633e+00 7.60355413e-01 -6.62697017e-01 7.70465374e-01 1.01845074e+00 -7.71912932e-03 -6.07552886e-01 -1.18663371e+00 -8.22014630e-01 5.54841399e-01 -2.01453432e-01 5.30646026e-01 7.08748341e-01 -1.74135119e-02 4.68926549e-01 -1.10237509e-01 4.04296964e-01 3.68703395e-01 -2.44200006e-01 8.20415795e-01 -9.86689210e-01 -3.57667565e-01 -7.31251985e-02 -1.52302876e-01 -1.00859475e+00 -2.90955249e-02 -6.16178453e-01 5.08504435e-02 -1.15541923e+00 -4.41207699e-02 -3.70843828e-01 -1.10249363e-01 3.75925899e-01 -4.55491513e-01 2.77610362e-01 1.01090610e-01 3.21481705e-01 -7.92165339e-01 2.98436165e-01 9.39058661e-01 -1.71654686e-01 -6.29155561e-02 -1.28124833e-01 -1.13901770e+00 5.33916950e-01 8.62031639e-01 -8.91917348e-01 -3.60661715e-01 -4.80869889e-01 2.75867224e-01 3.53836119e-02 4.60382134e-01 -9.51698899e-01 2.18341604e-01 2.16172040e-01 -5.67933880e-02 -3.62874150e-01 2.88881361e-01 -6.49187863e-01 -3.83411318e-01 6.33294642e-01 -7.53571451e-01 2.26158604e-01 2.82209605e-01 9.23709154e-01 -1.51354685e-01 -5.24463296e-01 5.42071819e-01 -2.77672112e-01 -2.97547638e-01 -1.10277653e-01 -4.63087201e-01 7.56961763e-01 7.80397773e-01 2.00412080e-01 -8.20374250e-01 -5.46027422e-01 -5.63825488e-01 4.83638465e-01 3.66375774e-01 5.21260858e-01 3.37997437e-01 -9.07158852e-01 -7.48694122e-01 -1.83227137e-01 2.18113407e-01 -3.04977655e-01 -1.07587405e-01 3.47015947e-01 -5.43880701e-01 3.88522059e-01 3.48208845e-01 1.95969790e-02 -1.15002429e+00 1.02219260e+00 4.39045787e-01 -5.63336015e-01 -3.76494944e-01 1.06079626e+00 -2.04339519e-01 -5.32126606e-01 4.17625278e-01 2.28274137e-01 -8.46662670e-02 -2.28539184e-01 6.08761787e-01 2.25254729e-01 1.87773809e-01 -2.80835330e-01 -3.89928818e-01 -1.28032416e-01 -3.52297425e-01 -4.52437788e-01 7.42975712e-01 1.64751351e-01 -6.48277327e-02 1.76072150e-01 9.25609350e-01 5.45994580e-01 -6.56044364e-01 -3.22845757e-01 4.45159167e-01 -5.44907153e-01 -5.51540196e-01 -1.25716329e+00 -3.94578040e-01 4.55159307e-01 3.77721190e-01 6.69878900e-01 7.23442674e-01 3.00227195e-01 1.12679684e+00 9.46184099e-01 5.73992729e-01 -8.87802541e-01 4.37966198e-01 5.81677139e-01 8.29701662e-01 -9.90829527e-01 -1.89116612e-01 -3.72794330e-01 -6.19542956e-01 4.11202103e-01 7.77525544e-01 -2.01446131e-01 2.31735766e-01 -6.83469847e-02 3.70937169e-01 -4.30112064e-01 -8.30843985e-01 -2.46812757e-02 -1.42574847e-01 5.70132613e-01 -4.52395268e-02 -3.11795086e-01 -3.27449292e-01 6.06137693e-01 -3.00975442e-01 -5.95459104e-01 3.68152797e-01 1.07771289e+00 -6.40774190e-01 -1.19396889e+00 -6.39712393e-01 4.09100890e-01 -1.10558546e+00 -2.97965676e-01 -9.83031392e-01 6.95188582e-01 -1.38662100e-01 1.59301722e+00 -3.02729845e-01 -5.07420599e-01 5.34000158e-01 5.59099555e-01 2.65416831e-01 -6.81635141e-01 -1.21525872e+00 -4.42875415e-01 5.70107520e-01 -6.13283753e-01 4.51691002e-02 -3.85774583e-01 -9.94135380e-01 -3.95336956e-01 -6.78508401e-01 7.41411805e-01 1.71243966e-01 8.81290674e-01 4.00928974e-01 7.22784773e-02 4.03300703e-01 1.52846903e-01 -1.32769728e+00 -6.94284201e-01 -2.41713941e-01 8.03545058e-01 2.64243007e-01 -4.15960312e-01 -6.15073681e-01 -3.70753646e-01]
[11.118722915649414, 7.987025737762451]
4e77a6a7-6263-42d2-a92d-9cee25217736
advancing-from-predictive-maintenance-to
2009.00351
null
https://arxiv.org/abs/2009.00351v1
https://arxiv.org/pdf/2009.00351v1.pdf
Advancing from Predictive Maintenance to Intelligent Maintenance with AI and IIoT
As Artificial Intelligent (AI) technology advances and increasingly large amounts of data become readily available via various Industrial Internet of Things (IIoT) projects, we evaluate the state of the art of predictive maintenance approaches and propose our innovative framework to improve the current practice. The paper first reviews the evolution of reliability modelling technology in the past 90 years and discusses major technologies developed in industry and academia. We then introduce the next generation maintenance framework - Intelligent Maintenance, and discuss its key components. This AI and IIoT based Intelligent Maintenance framework is composed of (1) latest machine learning algorithms including probabilistic reliability modelling with deep learning, (2) real-time data collection, transfer, and storage through wireless smart sensors, (3) Big Data technologies, (4) continuously integration and deployment of machine learning models, (5) mobile device and AR/VR applications for fast and better decision-making in the field. Particularly, we proposed a novel probabilistic deep learning reliability modelling approach and demonstrate it in the Turbofan Engine Degradation Dataset.
['Antonio R. Paiva', 'Haining Zheng', 'Chris S. Gurciullo']
2020-09-01
null
null
null
null
['probabilistic-deep-learning']
['computer-vision']
[-5.06266296e-01 -7.24264141e-03 -1.63539976e-01 -1.76893681e-01 -3.56888235e-01 4.27556515e-01 1.75653338e-01 -3.53854559e-02 4.31306452e-01 8.45386267e-01 -2.17776462e-01 -4.40798372e-01 -7.61761963e-01 -1.23646593e+00 -6.67219102e-01 -7.47259259e-01 -3.90942305e-01 1.35987854e+00 2.08325684e-01 -2.11974531e-01 2.64507025e-01 3.34667355e-01 -2.14965749e+00 -1.14943888e-02 7.63613403e-01 1.95688927e+00 3.86173517e-01 6.80888295e-01 2.42340446e-01 1.33323705e+00 -1.01723838e+00 9.74504352e-02 -3.40452820e-01 2.47299775e-01 -1.07077718e+00 -1.38124749e-01 -5.60455978e-01 -2.94348925e-01 -5.88371873e-01 3.44895244e-01 6.39353871e-01 1.39461150e-02 6.88924730e-01 -1.57710230e+00 -1.13877654e+00 6.65524065e-01 9.29339081e-02 2.05257460e-01 9.38009545e-02 -5.89249544e-02 3.39344144e-01 -3.97798240e-01 -1.27409652e-01 1.05113077e+00 6.98447704e-01 1.71773702e-01 -3.96887243e-01 -1.28971145e-01 -4.61522013e-01 9.37137961e-01 -1.27606845e+00 3.31080034e-02 9.01078761e-01 -4.99852419e-01 1.49582005e+00 -1.79872289e-01 4.16557461e-01 1.21878374e+00 1.28132451e+00 7.51376629e-01 7.06647635e-01 -4.70540732e-01 6.43939614e-01 -1.73892468e-01 2.02703893e-01 7.56115794e-01 -1.25979241e-02 5.44946015e-01 -5.52667558e-01 3.62043530e-02 2.97999412e-01 2.26932883e-01 7.50034809e-01 -3.17937136e-01 -6.59651339e-01 6.51522458e-01 9.46427435e-02 3.75625700e-01 -8.17434847e-01 4.85584140e-01 5.37175834e-01 7.43332386e-01 6.20913625e-01 -3.89803834e-02 -1.33808613e+00 -4.11584675e-01 -6.90372825e-01 -8.44988078e-02 5.89473665e-01 1.09201574e+00 3.82765740e-01 8.43435645e-01 -1.09847434e-01 9.83562768e-01 5.64025342e-01 6.42128766e-01 6.49911523e-01 -1.10465753e+00 -5.08060791e-02 3.42982233e-01 1.51340708e-01 -5.55313587e-01 -6.15805149e-01 -4.01363850e-01 -9.31915224e-01 2.39281997e-01 -7.28505373e-01 7.90420920e-02 -8.20004046e-01 6.79137349e-01 -5.34027405e-02 2.62913257e-02 2.93379482e-02 5.50091118e-02 6.45749629e-01 6.58736885e-01 -5.63372895e-02 -3.01823765e-02 1.24702024e+00 -8.18348110e-01 -9.74010468e-01 1.74251236e-02 5.12234092e-01 -2.60376960e-01 8.80833030e-01 1.01429272e+00 -1.08017004e+00 -1.02169633e+00 -1.30443144e+00 1.58882186e-01 -7.16245174e-01 -6.88998727e-03 5.01694262e-01 8.15170527e-01 -1.25353169e+00 1.12846088e+00 -1.03325820e+00 -1.38469075e-03 5.33802986e-01 6.17831051e-01 -1.89674571e-01 -2.61843920e-01 -1.34923720e+00 1.30504835e+00 4.19623882e-01 -1.52035609e-01 -1.55914092e+00 -2.51747817e-01 -9.16740417e-01 6.05922565e-03 3.18761826e-01 -1.01443994e+00 1.41510546e+00 -1.05475016e-01 -1.78555942e+00 1.14899389e-01 3.98125976e-01 -6.38665140e-01 -2.92784065e-01 -7.25410819e-01 -1.01986921e+00 -1.07079051e-01 -2.81398654e-01 -1.00752607e-01 9.35726225e-01 -1.19105160e+00 -6.31785214e-01 -7.31987894e-01 -3.38807791e-01 -8.07419896e-01 -1.97965056e-01 -2.17672754e-02 -3.15592922e-02 -3.42961431e-01 -3.45152766e-01 -6.98083699e-01 2.01246113e-01 -5.88789642e-01 -2.14357719e-01 -8.93706501e-01 1.24037123e+00 -1.03029108e+00 1.15634954e+00 -1.73768723e+00 1.29361272e-01 -1.64467543e-01 -2.15267330e-01 2.05997959e-01 1.80692971e-01 7.25612998e-01 1.10076509e-01 -3.57919425e-01 -3.21440026e-02 -5.16852438e-01 1.79079488e-01 6.72165513e-01 3.08015067e-02 2.38368094e-01 1.52471766e-01 8.45787823e-01 -6.35433733e-01 -5.19315481e-01 9.74618971e-01 6.41622126e-01 1.02144606e-01 2.04306662e-01 -4.20782357e-01 3.02619487e-01 -4.41481620e-01 1.22560084e+00 5.01887619e-01 -1.17875785e-02 -2.11931810e-01 -2.26638496e-01 2.63763279e-01 1.68132186e-01 -7.00243115e-01 1.43982387e+00 -1.14348674e+00 5.11905476e-02 -3.76988798e-01 -1.63396442e+00 1.21830785e+00 5.51056087e-01 9.80545878e-01 -9.65670347e-01 4.95947957e-01 -8.69686827e-02 -6.50543988e-01 -9.11545873e-01 4.29983228e-01 -8.22117031e-02 -1.81996122e-01 2.76896060e-01 8.43418658e-01 -3.57443750e-01 -3.00732493e-01 -2.92945415e-01 1.44501626e+00 2.85868734e-01 -5.36003001e-02 -1.84849218e-01 2.49150708e-01 -5.15517950e-01 3.82460028e-01 3.64714891e-01 -3.04666251e-01 1.09985597e-01 -2.60631498e-02 -7.79542983e-01 -8.24288070e-01 -1.11255705e+00 -1.85218468e-01 8.84524465e-01 -1.49722412e-01 -4.12686884e-01 -3.75319093e-01 -5.84410369e-01 2.18375787e-01 1.10314667e+00 -5.22668540e-01 -7.83390880e-01 -1.08410165e-01 -7.22656071e-01 1.26823559e-01 9.47354853e-01 5.28408706e-01 -1.47221911e+00 -3.61731589e-01 2.53969967e-01 -6.17782548e-02 -7.48005271e-01 6.48450315e-01 4.71015185e-01 -1.27702320e+00 -9.19636548e-01 1.02847211e-01 -6.67937458e-01 -2.41614401e-01 -1.12089865e-01 1.66385221e+00 2.99426466e-01 -3.73379201e-01 6.66024804e-01 -4.19916600e-01 -6.99980021e-01 -4.91729349e-01 -2.25691125e-01 6.38535738e-01 -7.17722297e-01 3.77430797e-01 -7.88597286e-01 -3.57422709e-01 4.53846127e-01 -5.76116741e-01 -7.26778805e-01 6.84077322e-01 7.54445553e-01 4.32323128e-01 1.37433434e+00 1.05227447e+00 -2.39943862e-01 5.43032825e-01 -8.51782382e-01 -4.33828235e-01 4.10723805e-01 -1.41150391e+00 2.41051316e-01 3.61093611e-01 1.99596331e-01 -1.01415706e+00 -5.25540352e-01 -3.96239042e-01 -3.59942406e-01 -1.05915427e-01 4.81351823e-01 -3.41094673e-01 3.09897095e-01 4.78800595e-01 1.87758029e-01 1.00229807e-01 -7.05570042e-01 3.65029752e-01 1.44183421e+00 5.49658477e-01 -8.00851047e-01 3.08382332e-01 1.50066569e-01 -1.42397135e-01 -5.62469661e-01 -2.18128547e-01 6.06744811e-02 -6.92502499e-01 -4.46921438e-01 6.92451298e-01 -8.26474428e-01 -9.90864158e-01 8.64252925e-01 -6.84663355e-01 -4.42854702e-01 -4.95005965e-01 2.69696027e-01 -1.02804470e+00 2.47346222e-01 -1.06260765e+00 -1.10724795e+00 -9.08545196e-01 -1.22044146e+00 1.35863745e+00 -8.17918554e-02 2.61635691e-01 -8.91920030e-01 7.13478103e-02 7.47867644e-01 7.19539881e-01 -1.01414614e-03 1.30667484e+00 -2.34479144e-01 -3.51615816e-01 -8.13013852e-01 2.60508299e-01 1.14962804e+00 1.71774253e-01 1.37886122e-01 -7.91855752e-01 5.58030754e-02 2.90830940e-01 -3.81236583e-01 4.14345413e-01 5.91283441e-01 1.75006282e+00 4.70155403e-02 -4.71700102e-01 -3.29196215e-01 1.01827443e+00 4.79189336e-01 9.85085726e-01 3.40132833e-01 3.86159033e-01 4.94263232e-01 9.43261921e-01 5.10256767e-01 7.79543340e-01 4.87629622e-01 8.20625007e-01 7.32415140e-01 7.89777339e-02 1.59422774e-02 5.49102783e-01 1.39685953e+00 -3.55801105e-01 -2.70888805e-01 -8.40167820e-01 4.69244272e-01 -1.84837580e+00 -8.23582590e-01 1.48345605e-01 2.05888128e+00 3.75431240e-01 4.45736766e-01 2.97676102e-04 8.74579608e-01 4.01989818e-01 -2.68870711e-01 -5.41271865e-01 -3.94559473e-01 2.35812932e-01 4.50678438e-01 2.40287170e-01 -3.08250338e-02 -9.37983036e-01 5.01453459e-01 7.43764591e+00 9.36724544e-01 -5.05262136e-01 6.99047565e-01 5.73955774e-01 1.10536277e-01 6.89202966e-03 -4.20393556e-01 -1.79011211e-01 8.14801395e-01 1.75949609e+00 3.71772856e-01 4.83633429e-01 1.42034435e+00 -2.16787562e-01 -1.46813855e-01 -8.07545245e-01 8.55664849e-01 9.34331641e-02 -1.31003523e+00 -3.89833182e-01 -2.52824631e-02 4.74696577e-01 4.19834852e-01 -3.55655439e-02 8.24367583e-01 7.19723880e-01 -8.73900592e-01 6.37610614e-01 9.24413919e-01 3.94721836e-01 -1.02064085e+00 1.39203990e+00 1.02144375e-01 -7.97783673e-01 -1.04057765e+00 -5.12741148e-01 -1.63312291e-03 4.53106821e-01 1.23859656e+00 -2.93108165e-01 1.27175915e+00 1.61650038e+00 6.37940049e-01 -4.63271797e-01 6.70410156e-01 1.15253648e-03 5.22968471e-01 -4.13689613e-02 1.70186877e-01 -7.53441989e-01 1.50033563e-01 -8.63397494e-03 2.09798545e-01 6.45487964e-01 -7.43364394e-01 -1.38310641e-01 4.95028973e-01 2.03538939e-01 -1.03017497e+00 -4.58401620e-01 -7.80792013e-02 5.28100193e-01 9.72106934e-01 -3.14455271e-01 -2.21254915e-01 -3.78194511e-01 6.52093053e-01 1.04107670e-02 -2.83020645e-01 -9.15297389e-01 -1.37631834e-01 4.33748066e-01 -9.22586676e-03 2.34056666e-01 -4.63846534e-01 -3.71663719e-01 -3.84867191e-01 -1.74363792e-01 -5.18737972e-01 4.82597351e-01 -1.36442757e+00 -1.55536640e+00 6.82335734e-01 2.09934950e-01 -9.53144073e-01 -2.77498782e-01 -6.93798304e-01 -3.76832932e-01 2.95154124e-01 -1.15760636e+00 -1.29717720e+00 -1.19380035e-01 5.39043546e-01 7.81405210e-01 -7.10149705e-01 9.82757270e-01 5.36199868e-01 -5.36119938e-01 7.74094090e-02 4.96693432e-01 -6.03077471e-01 1.32388353e-01 -1.20606124e+00 4.49389964e-01 4.55909193e-01 -2.33017281e-01 1.01109430e-01 7.19405591e-01 -8.30747843e-01 -1.57731414e+00 -1.00539207e+00 4.01097804e-01 -9.72160101e-01 5.61943054e-01 -7.01161474e-02 -4.31737870e-01 7.08589435e-01 1.77675053e-01 -8.59954208e-02 5.23345053e-01 3.12478393e-01 1.93566605e-01 -4.57942367e-01 -1.45264673e+00 -2.52695709e-01 6.29270673e-01 -6.78683996e-01 -5.78414857e-01 7.83138216e-01 1.21645236e+00 1.80879906e-01 -1.49379921e+00 9.19656217e-01 3.34799796e-01 -7.76697934e-01 1.18923557e+00 -4.03548151e-01 2.55500793e-01 -2.87178010e-01 -5.38052857e-01 -1.17889214e+00 -3.70009333e-01 -3.31743121e-01 -1.40449464e+00 1.09005952e+00 -1.10053144e-01 -3.98698688e-01 2.95903623e-01 3.83642942e-01 -1.01049745e+00 -1.13718128e+00 -1.29361892e+00 -1.03598106e+00 -8.07203948e-02 -8.99247050e-01 1.01080978e+00 1.65866897e-01 -3.92964929e-02 2.06626609e-01 -8.08334053e-01 1.77136362e-01 2.33060151e-01 -3.10383976e-01 3.63703400e-01 -1.79910350e+00 -1.83803216e-01 2.83249170e-01 -7.90330350e-01 -3.26036781e-01 -1.85480770e-02 -1.70410350e-01 2.31848359e-01 -2.09900689e+00 -2.86464274e-01 -5.10703564e-01 -8.96782815e-01 3.43712151e-01 4.83964205e-01 -2.53148615e-01 -5.84341824e-01 2.42990982e-02 -8.10529649e-01 1.30464649e+00 5.92394948e-01 -3.99515629e-01 3.24480236e-01 5.24890721e-01 -4.71013248e-01 2.05001682e-01 9.88758564e-01 -3.83136749e-01 -5.09697795e-01 -3.70788336e-01 2.08786428e-01 8.14331099e-02 3.31642330e-01 -1.76748621e+00 -1.89799741e-01 4.25465316e-01 5.49832284e-01 -8.96781921e-01 3.44660550e-01 -1.26932514e+00 3.13565135e-01 5.98597646e-01 6.22320771e-01 4.54436570e-01 2.40680873e-02 7.51004040e-01 5.58851995e-02 -1.46028697e-01 5.82086444e-01 1.58159316e-01 -6.82734013e-01 2.00817481e-01 -6.44073665e-01 -7.98750281e-01 1.19264960e+00 -2.80091935e-03 -5.68767607e-01 -1.29317164e-01 -8.42702150e-01 1.39517710e-01 6.82669729e-02 1.13365805e+00 6.95696831e-01 -1.32087600e+00 -1.14143267e-01 2.05204204e-01 -4.02255543e-03 -2.71913800e-02 8.13563704e-01 6.63461268e-01 -4.16380465e-01 5.93540967e-01 -3.30846250e-01 -5.83856463e-01 -6.85027182e-01 1.41801059e+00 3.61768454e-01 -2.05623999e-01 -5.65462470e-01 6.43755317e-01 -9.26270723e-01 -6.50797427e-01 2.53495723e-01 -2.34206215e-01 -2.79018968e-01 -4.78216887e-01 5.10150611e-01 9.58631754e-01 9.37203646e-01 -3.26255172e-01 -3.01292032e-01 3.28734070e-01 2.35714778e-01 3.73605996e-01 1.58079982e+00 -3.82639498e-01 -2.05152705e-01 9.58440006e-01 6.63747370e-01 -8.88782263e-01 -7.76790440e-01 1.45838395e-01 2.68613309e-01 2.34531090e-02 4.91313815e-01 -1.09650815e+00 -1.39621329e+00 8.77198339e-01 1.28089905e+00 6.20081067e-01 1.08821929e+00 3.90640259e-01 1.10148537e+00 5.22580326e-01 1.14345396e+00 -1.74361539e+00 3.79493237e-01 3.26608241e-01 8.57359469e-01 -1.10755122e+00 2.27371175e-02 2.84941912e-01 -5.91477573e-01 6.84434533e-01 5.38339376e-01 6.88585863e-02 1.61999571e+00 3.17507565e-01 -2.67863482e-01 -7.75358856e-01 -8.42519283e-01 1.70868754e-01 -2.07658246e-01 1.30622268e+00 1.71317998e-02 1.21978857e-01 2.73466885e-01 1.17429221e+00 -1.77300513e-01 4.91273314e-01 1.54711977e-02 1.27386844e+00 -5.62231779e-01 -1.18276405e+00 -2.53358364e-01 8.27409565e-01 -4.32141513e-01 3.25464934e-01 2.58307546e-01 2.46146113e-01 2.59957880e-01 1.73963261e+00 -3.27690989e-02 -1.27003324e+00 3.50074857e-01 1.85892493e-01 4.15837824e-01 -2.00446874e-01 -4.81381901e-02 -8.34576011e-01 1.10346466e-01 -8.50866973e-01 -1.37154296e-01 -1.18524529e-01 -1.01016366e+00 -4.61055666e-01 -7.29591072e-01 1.47763520e-01 1.09683311e+00 1.27112412e+00 9.02500510e-01 1.26109910e+00 7.25273788e-01 -9.58250582e-01 -4.74468142e-01 -1.60868144e+00 -1.03661942e+00 -3.44360888e-01 -2.78376639e-02 -1.42930186e+00 -7.92099833e-02 -5.61869703e-02]
[6.781292915344238, 2.454103946685791]
dc8e9de2-e441-4200-95dc-b1373e456b0a
mpsa-densenet-a-novel-deep-learning-model-for
2306.08798
null
https://arxiv.org/abs/2306.08798v1
https://arxiv.org/pdf/2306.08798v1.pdf
MPSA-DenseNet: A novel deep learning model for English accent classification
This paper presents three innovative deep learning models for English accent classification: Multi-DenseNet, PSA-DenseNet, and MPSE-DenseNet, that combine multi-task learning and the PSA module attention mechanism with DenseNet. We applied these models to data collected from six dialects of English across native English speaking regions (Britain, the United States, Scotland) and nonnative English speaking regions (China, Germany, India). Our experimental results show a significant improvement in classification accuracy, particularly with MPSA-DenseNet, which outperforms all other models, including DenseNet and EPSA models previously used for accent identification. Our findings indicate that MPSA-DenseNet is a highly promising model for accurately identifying English accents.
['Ton Viet Ta', 'Linh Thi Hoai Nguyen', 'Tianyu Song']
2023-06-15
null
null
null
null
['multi-task-learning']
['methodology']
[-6.06775463e-01 -2.60788739e-01 -1.21037802e-02 -6.49322331e-01 -8.01509559e-01 -7.12348163e-01 2.82957852e-01 -1.35802031e-01 -8.70494127e-01 1.02387202e+00 5.78567922e-01 -5.93834400e-01 -7.22856745e-02 -6.50654554e-01 -3.08964252e-01 -4.03623641e-01 -1.15750648e-01 7.56071806e-01 -3.28072459e-01 -6.04207754e-01 -1.44063160e-01 4.67814982e-01 -7.42065549e-01 -5.05190082e-02 1.18203604e+00 7.15870738e-01 4.35843796e-01 5.43829381e-01 -2.30837747e-01 9.21046376e-01 -6.75276399e-01 -8.41701806e-01 -3.71516794e-02 8.71334597e-02 -1.14469254e+00 -5.23976862e-01 5.44995248e-01 -2.03107804e-01 -7.76457563e-02 1.00314236e+00 9.75115776e-01 1.23725481e-01 1.67538345e-01 -8.02235842e-01 -8.34661603e-01 1.02980292e+00 -3.69253486e-01 4.45587069e-01 1.68877184e-01 -1.83059320e-01 9.55789924e-01 -9.18828368e-01 3.96107912e-01 1.45246923e+00 1.31965828e+00 3.29997927e-01 -7.83295870e-01 -9.97830391e-01 3.40559363e-01 1.35902554e-01 -1.59966767e+00 -4.84792054e-01 6.22223437e-01 8.38843733e-03 1.16098166e+00 3.32896680e-01 4.03486788e-01 1.37800288e+00 3.87746453e-01 8.18831086e-01 1.71709073e+00 -3.46922219e-01 6.98225945e-02 3.45822918e-04 3.65904987e-01 4.62043345e-01 -4.13416117e-01 1.22965299e-01 -7.22921729e-01 -1.67531013e-01 6.49578512e-01 -3.14816952e-01 7.10817277e-02 5.39832950e-01 -1.12450051e+00 9.34488654e-01 4.44937617e-01 3.79010260e-01 -8.65318298e-01 -4.55942810e-01 3.70861411e-01 5.82699478e-01 8.10270309e-01 4.17966932e-01 -1.26351583e+00 -5.34470558e-01 -6.31682217e-01 1.97100118e-01 8.22141469e-01 8.80389333e-01 6.79303944e-01 6.25532389e-01 4.97095257e-01 1.41634560e+00 1.62331432e-01 6.65875971e-01 6.37073398e-01 -4.35823023e-01 3.71875972e-01 1.23680525e-01 -8.47157463e-03 -4.83132452e-01 -8.03690195e-01 -6.71374679e-01 -1.09564245e+00 8.74617975e-03 1.83638468e-01 -1.02039373e+00 -1.00219071e+00 1.94221210e+00 -1.44151464e-01 1.22498177e-01 3.70826453e-01 7.87975907e-01 7.84533978e-01 4.51054394e-01 7.48178661e-01 2.68043399e-01 1.45520282e+00 -9.86435533e-01 -7.76403785e-01 -4.97886747e-01 2.87445068e-01 -1.07169962e+00 1.00620079e+00 1.99638575e-01 -1.07574570e+00 -1.07022154e+00 -5.96576989e-01 2.50882030e-01 -8.19682360e-01 2.72235069e-02 8.28288019e-01 7.64328182e-01 -1.44928992e+00 2.72590630e-02 -6.41316473e-01 -3.68057370e-01 2.89238662e-01 6.37475729e-01 -6.18475616e-01 1.60119966e-01 -1.66095102e+00 1.05605078e+00 5.90939879e-01 2.75150031e-01 -6.90032780e-01 -7.88006186e-01 -1.18675375e+00 1.74754307e-01 -2.64980495e-01 -4.32348162e-01 1.36483145e+00 -1.10198176e+00 -1.52367139e+00 7.81124115e-01 -7.68793225e-02 -5.56518316e-01 5.53631522e-02 -6.02556825e-01 -1.02045238e+00 -4.63080317e-01 1.51343703e-01 7.05888748e-01 8.76251608e-02 -9.91711199e-01 -8.65540802e-01 -3.91488224e-01 1.11612059e-01 4.23473895e-01 -3.16691309e-01 6.25579536e-01 3.28632325e-01 -1.02539289e+00 -2.56468803e-01 -7.45455146e-01 -5.51262498e-01 -1.15531814e+00 -5.15515804e-01 -4.42043155e-01 7.85835385e-01 -1.12048674e+00 1.01380706e+00 -2.32185745e+00 -9.25943553e-02 2.12594450e-01 -2.27948129e-02 3.22739035e-01 -2.70425558e-01 1.44402370e-01 -1.68142378e-01 1.31354287e-01 9.23367515e-02 -3.58834267e-01 1.51091367e-01 2.89615154e-01 -1.49977446e-01 6.83639646e-02 3.66092741e-01 8.19419980e-01 -7.64814258e-01 -2.10602507e-01 4.01813596e-01 4.94818717e-01 -1.59120038e-01 2.34007344e-01 2.31058300e-01 2.37642437e-01 -1.61869183e-01 9.40779388e-01 1.03547812e+00 2.71298736e-01 2.32612118e-01 1.82466790e-01 -4.91031229e-01 5.54393232e-01 -1.15232658e+00 1.55553353e+00 -9.31147516e-01 4.20528620e-01 6.26454711e-01 -7.11452484e-01 1.03099799e+00 4.64027196e-01 1.07146159e-01 -7.56513953e-01 -1.12294108e-01 5.24825096e-01 2.36191317e-01 1.78438351e-01 7.20533431e-01 -3.66738886e-01 -6.47519708e-01 2.16935948e-02 5.29609382e-01 3.12713712e-01 -1.94594800e-01 -3.89477424e-02 6.89820290e-01 -1.48441449e-01 2.31619403e-01 -8.40237260e-01 3.93456668e-01 -2.31313422e-01 1.07980061e+00 9.24756229e-01 -5.35576463e-01 4.44283366e-01 1.51612967e-01 -7.84572423e-01 -7.79107451e-01 -1.43743336e+00 -4.44800198e-01 1.78928566e+00 -3.12059909e-01 -3.68627131e-01 -5.93966722e-01 -7.01094627e-01 -1.46775290e-01 5.42226732e-01 -3.75944555e-01 2.80451983e-01 -7.49685943e-01 -1.17747784e+00 7.22925603e-01 9.18857276e-01 1.24398327e+00 -1.57748806e+00 1.37318864e-01 4.32882786e-01 -2.78306693e-01 -1.09280360e+00 -5.23426473e-01 7.91324437e-01 -1.94099858e-01 -4.45574075e-01 -8.01828206e-01 -1.48067021e+00 6.03297539e-02 -5.37122905e-01 1.48782992e+00 -5.12020826e-01 3.87870461e-01 1.15695968e-01 -2.80335516e-01 -8.66579115e-01 -4.02764976e-01 7.60278106e-01 2.40243867e-01 -1.24699377e-01 9.21423316e-01 -3.87487262e-01 -1.02271222e-01 2.35714838e-01 -7.70997331e-02 -2.23684877e-01 7.12144852e-01 9.82322514e-01 3.52238566e-01 -6.29360303e-02 1.33745050e+00 -9.95170772e-01 5.65236866e-01 -6.52648926e-01 -2.74990559e-01 2.74907565e-03 -2.05098584e-01 -4.99649256e-01 6.51297569e-01 -9.53861922e-02 -1.32535839e+00 1.08537465e-01 -1.31465745e+00 4.08242717e-02 -6.36242390e-01 8.39354992e-01 -4.26104993e-01 4.14264202e-02 2.58645773e-01 4.57591414e-02 -3.50685090e-01 -6.35270894e-01 1.29864216e-01 1.10137713e+00 8.81288588e-01 -5.02856493e-01 2.03315541e-01 -3.93147618e-01 -6.95467949e-01 -1.07137275e+00 -8.57892752e-01 -5.36398590e-01 -8.85407925e-01 3.69493030e-02 1.15402651e+00 -1.51051211e+00 -4.25602108e-01 1.18589675e+00 -9.24380720e-01 -5.92246056e-01 -3.94539945e-02 6.96715176e-01 -1.98261306e-01 -2.83875614e-01 -1.16252446e+00 -6.20168865e-01 -5.19918144e-01 -9.64735270e-01 1.02187693e+00 4.57408696e-01 -3.82441759e-01 -1.70810509e+00 7.61331543e-02 8.78717601e-02 8.50400209e-01 2.08546109e-02 8.13001156e-01 -9.12983239e-01 3.89961414e-02 3.45667899e-01 -6.95277797e-03 5.54345429e-01 2.40117103e-01 -4.53571618e-01 -1.25968075e+00 -3.76570433e-01 -2.30335861e-01 -6.89581156e-01 8.43416810e-01 7.85838485e-01 1.02523828e+00 3.65704261e-02 1.49693310e-01 8.02390873e-01 1.21068764e+00 4.71870720e-01 6.10482275e-01 7.09826767e-01 6.98417485e-01 -2.63184756e-02 3.81484210e-01 1.12381458e-01 7.88594604e-01 3.01088840e-01 -3.17667946e-02 -7.37156808e-01 -4.82790731e-02 8.57942849e-02 3.22633028e-01 1.34661984e+00 -1.75778031e-01 -2.44096756e-01 -1.15803158e+00 8.49420965e-01 -1.37442243e+00 -8.13083053e-01 1.78212360e-01 1.77158952e+00 8.67582142e-01 1.96891382e-01 3.83309662e-01 -1.61647558e-01 7.07615793e-01 4.44879889e-01 -2.20205829e-01 -1.09311914e+00 -7.24806488e-01 1.04687440e+00 3.83009851e-01 7.21009254e-01 -1.69543266e+00 1.31989884e+00 7.49958086e+00 4.82294589e-01 -1.06458318e+00 1.55600980e-01 1.03214431e+00 5.92764437e-01 7.22869635e-02 -3.65378320e-01 -1.03434980e+00 3.94472957e-01 1.57044590e+00 -8.81635249e-02 5.17916307e-02 8.58080745e-01 -2.47694686e-01 4.04365838e-01 -3.44711244e-01 5.38067579e-01 -3.11523139e-01 -1.09861684e+00 -2.67140239e-01 1.70720667e-02 7.74628460e-01 6.99309886e-01 2.23874658e-01 8.98514807e-01 1.14387619e+00 -9.95974481e-01 5.86093605e-01 1.77275226e-01 7.66941071e-01 -1.24178112e+00 1.42396891e+00 -5.66598773e-02 -1.06110573e+00 -8.87080133e-02 -3.70827585e-01 -2.13836268e-01 3.20914745e-01 3.47496122e-01 -5.08837342e-01 6.79893792e-01 1.10402489e+00 6.96840525e-01 -4.69059706e-01 5.66950798e-01 -1.32695166e-02 1.04254246e+00 -2.14674026e-01 1.57412499e-01 6.36511981e-01 -8.65866616e-02 2.28029221e-01 1.71800804e+00 2.45095387e-01 -3.39663029e-01 5.54940581e-01 2.87605166e-01 -3.10444325e-01 1.05523981e-01 -3.96471381e-01 2.77842194e-01 5.86314976e-01 1.32603574e+00 -4.47275728e-01 -1.82343856e-01 -8.30414772e-01 9.01903808e-01 5.75408101e-01 3.88491184e-01 -5.64542413e-01 -7.85871267e-01 1.01949370e+00 -5.51216125e-01 3.62370670e-01 -2.33624741e-01 -2.99644977e-01 -8.29362810e-01 -4.24498260e-01 -1.14631402e+00 5.35713494e-01 -5.87666214e-01 -1.65807772e+00 1.01197445e+00 -3.01039100e-01 -4.02520239e-01 -4.12144184e-01 -8.31177473e-01 -7.57849336e-01 1.47153246e+00 -1.53347993e+00 -1.59028506e+00 9.17585045e-02 7.75955558e-01 6.55143023e-01 -8.32660973e-01 1.47653162e+00 6.09423816e-01 -5.43838084e-01 8.98760438e-01 2.60266423e-01 6.79827988e-01 8.16701710e-01 -2.00417066e+00 1.04318655e+00 6.51430130e-01 8.25373307e-02 6.17520690e-01 5.71846962e-02 -4.07851249e-01 -6.30785167e-01 -1.16654778e+00 1.40879846e+00 -2.80400813e-01 6.37306511e-01 -7.31920898e-01 -7.77685761e-01 1.14111578e+00 1.02814138e+00 -4.11693037e-01 1.03644288e+00 1.16021168e+00 -7.60869831e-02 -8.47612843e-02 -1.03283226e+00 1.97107628e-01 7.86217570e-01 -8.26210082e-01 -7.64030159e-01 1.60411313e-01 5.87538064e-01 -5.81767917e-01 -1.34059405e+00 2.41164029e-01 3.69091332e-01 -7.41690099e-01 8.49576235e-01 -6.82075202e-01 -3.10659222e-02 3.79172236e-01 -2.65769809e-01 -2.18062377e+00 -7.37382293e-01 -5.73217392e-01 3.23065817e-01 1.55530131e+00 6.63631737e-01 -9.68675435e-01 3.75100464e-01 -5.01734540e-02 -8.40850711e-01 -3.92430484e-01 -1.05744290e+00 -5.48748195e-01 6.84702814e-01 -2.47858018e-01 8.68678927e-01 1.46875727e+00 -4.25895154e-01 5.66377997e-01 -3.14953357e-01 2.66336679e-01 1.51266724e-01 -1.06925294e-01 3.34928453e-01 -1.29762530e+00 7.12108836e-02 -4.20502394e-01 -5.65770507e-01 -7.54775226e-01 8.10066819e-01 -7.29498863e-01 -2.27153152e-01 -1.31537819e+00 -5.62765785e-02 -7.18772888e-01 -7.27722704e-01 8.13133538e-01 -4.47572649e-01 2.03703180e-01 1.25069574e-01 -4.40757394e-01 -3.40442181e-01 2.81716436e-01 6.75169170e-01 7.13248039e-03 -9.57448408e-02 1.98979899e-01 -1.06793308e+00 7.71804333e-01 1.05590701e+00 -2.19749734e-02 -2.81224418e-02 -6.99648976e-01 -2.29937226e-01 -4.03149843e-01 -4.90697064e-02 -9.95539606e-01 -3.69498692e-02 1.03660291e-02 7.53002524e-01 -5.46616435e-01 6.56271055e-02 -3.66134942e-01 -2.24626020e-01 1.41300157e-01 -1.70357078e-01 8.67377818e-01 8.44260573e-01 -1.12245061e-01 -7.89475083e-01 2.27251589e-01 7.95871437e-01 -1.37606442e-01 -1.40558589e+00 2.11963505e-01 -7.17868567e-01 8.68323743e-02 5.41393876e-01 3.45754474e-01 -4.03286159e-01 -2.08737433e-01 -1.12563407e+00 4.68405068e-01 -2.32424811e-01 7.21247315e-01 3.95792574e-01 -1.47966850e+00 -1.12282395e+00 7.52947032e-01 -3.34065765e-01 -1.80433303e-01 4.12183255e-01 4.39782023e-01 -4.84821975e-01 3.90769392e-01 -6.85733914e-01 -2.30049163e-01 -1.01483762e+00 -5.41853271e-02 4.65332955e-01 -3.37109745e-01 -3.12322110e-01 9.94531572e-01 1.25866115e-01 -1.48179817e+00 1.77689344e-01 -6.98323473e-02 -1.18530437e-01 -1.57098994e-01 2.58519650e-01 2.42961735e-01 1.03983618e-01 -8.96485090e-01 -6.20755613e-01 1.50327995e-01 -5.72934926e-01 -8.91520530e-02 1.27448690e+00 -7.23256171e-02 -5.62900715e-02 4.66076404e-01 8.27735364e-01 3.09260845e-01 -7.09183693e-01 4.55434211e-02 1.94875330e-01 7.70355836e-02 4.50513333e-01 -1.33982074e+00 -9.03379679e-01 6.47898197e-01 7.78349638e-01 8.67059678e-02 1.30238664e+00 -7.40578100e-02 7.51928568e-01 5.08013740e-02 3.51586998e-01 -1.28655982e+00 -5.97415209e-01 1.18229139e+00 5.56733787e-01 -1.34903240e+00 -7.18597591e-01 6.47203326e-02 -8.30925167e-01 8.78493667e-01 8.82363617e-01 8.07813331e-02 1.11626935e+00 6.58428967e-01 7.59419620e-01 -5.24582677e-02 -4.44474518e-01 -2.83590645e-01 -8.96917954e-02 7.54161119e-01 8.75982523e-01 4.27035749e-01 9.68360752e-02 8.57419133e-01 -8.41663778e-01 -3.33539754e-01 2.07818940e-01 9.55576539e-01 -4.69788462e-02 -1.01226521e+00 -2.49123290e-01 4.82601076e-01 -8.61131549e-01 -4.04291838e-01 -4.03670967e-01 1.15686071e+00 1.67282596e-01 7.90692985e-01 4.75812763e-01 -4.98941779e-01 5.27019382e-01 5.39142668e-01 -1.03596807e-01 -2.88926840e-01 -1.15858972e+00 2.05324322e-01 5.40330946e-01 3.49174850e-02 -4.34785783e-01 -7.79006422e-01 -9.79394436e-01 -6.54065311e-01 -2.43319228e-01 4.57537562e-01 5.42911410e-01 7.33802378e-01 3.52450997e-01 5.78731894e-01 7.90337920e-01 -4.02456790e-01 -3.98341268e-01 -1.44554138e+00 -1.06508505e+00 1.42938763e-01 4.85604584e-01 -2.86763012e-01 -1.55210644e-01 -2.67953247e-01]
[14.296006202697754, 6.754585266113281]
1e36dd42-e12b-4605-823b-3b9c8c521973
probing-representations-learned-by-multimodal
1908.11125
null
https://arxiv.org/abs/1908.11125v1
https://arxiv.org/pdf/1908.11125v1.pdf
Probing Representations Learned by Multimodal Recurrent and Transformer Models
Recent literature shows that large-scale language modeling provides excellent reusable sentence representations with both recurrent and self-attentive architectures. However, there has been less clarity on the commonalities and differences in the representational properties induced by the two architectures. It also has been shown that visual information serves as one of the means for grounding sentence representations. In this paper, we present a meta-study assessing the representational quality of models where the training signal is obtained from different modalities, in particular, language modeling, image features prediction, and both textual and multimodal machine translation. We evaluate textual and visual features of sentence representations obtained using predominant approaches on image retrieval and semantic textual similarity. Our experiments reveal that on moderate-sized datasets, a sentence counterpart in a target language or visual modality provides much stronger training signal for sentence representation than language modeling. Importantly, we observe that while the Transformer models achieve superior machine translation quality, representations from the recurrent neural network based models perform significantly better over tasks focused on semantic relevance.
['Jindřich Libovický', 'Pranava Madhyastha']
2019-08-29
null
null
null
null
['multimodal-machine-translation']
['natural-language-processing']
[ 4.40094948e-01 1.66633263e-01 -4.27077115e-01 -2.85052627e-01 -1.12671173e+00 -4.37589973e-01 1.14981484e+00 3.14435333e-01 -6.01118617e-02 3.43674541e-01 8.69613469e-01 -3.08438152e-01 1.63547501e-01 -5.50665498e-01 -6.57002211e-01 -3.25374663e-01 4.96498287e-01 2.53740162e-01 -4.27836239e-01 -2.91077405e-01 3.94396722e-01 2.76433080e-01 -1.44195962e+00 1.10896719e+00 4.41482306e-01 9.72324193e-01 3.54422450e-01 4.73552465e-01 -4.45564300e-01 1.22215152e+00 -4.97376144e-01 -5.02080560e-01 -2.50755191e-01 -5.87933719e-01 -8.88358474e-01 1.20390594e-01 9.81872797e-01 1.08623460e-01 -6.61430776e-01 7.06236362e-01 6.21344209e-01 -7.67947361e-02 7.56632328e-01 -9.23839033e-01 -1.46550226e+00 5.46047032e-01 -4.83224601e-01 3.60220939e-01 7.48696148e-01 6.89586699e-02 1.01946735e+00 -1.16930056e+00 9.65407014e-01 1.55087543e+00 4.54859674e-01 6.26130641e-01 -1.52701092e+00 -2.93467611e-01 1.65573820e-01 1.47063047e-01 -1.08399975e+00 -7.30625272e-01 7.39154696e-01 -5.27582169e-01 1.34818578e+00 4.56469566e-01 3.83332282e-01 1.52593112e+00 3.55913162e-01 8.99171591e-01 1.29846239e+00 -6.64106965e-01 -2.56798714e-01 7.22631037e-01 1.78751573e-01 7.24050581e-01 -1.37631476e-01 -9.77816731e-02 -8.64938140e-01 -1.34641662e-01 6.25958562e-01 2.44344071e-01 -1.78752080e-01 -3.34594309e-01 -1.26280534e+00 8.62108648e-01 4.87244576e-01 5.98164141e-01 -3.96902502e-01 2.94712871e-01 5.97602785e-01 5.90280890e-01 7.60204732e-01 6.02711022e-01 -4.12187912e-02 8.89959261e-02 -9.31051672e-01 -4.02833849e-01 3.74181271e-01 7.51113415e-01 7.79172242e-01 2.87458539e-01 -6.48575723e-01 8.90084326e-01 4.01519179e-01 6.20262146e-01 7.38178551e-01 -7.27130651e-01 7.90626049e-01 7.06845582e-01 -2.82134205e-01 -1.37202048e+00 -3.01908642e-01 -5.00725567e-01 -6.78128541e-01 -2.56818861e-01 -8.26428160e-02 3.05216551e-01 -6.44412577e-01 1.80769253e+00 -4.81982559e-01 -3.45429689e-01 4.27964002e-01 7.19863236e-01 1.36626899e+00 7.49878526e-01 5.29237747e-01 -1.63407221e-01 1.49386489e+00 -9.81806159e-01 -6.89343810e-01 -4.10938323e-01 8.38993490e-01 -9.96148944e-01 1.15507269e+00 -2.45335549e-01 -1.44583178e+00 -7.14494228e-01 -8.34266424e-01 -4.05681908e-01 -3.10022175e-01 4.45497125e-01 5.36441088e-01 4.26629454e-01 -1.31399739e+00 3.12877864e-01 -3.14909101e-01 -9.08554137e-01 2.46116638e-01 -6.82812035e-02 -5.13513505e-01 -1.34725019e-01 -9.79195952e-01 1.50748420e+00 -1.93364114e-01 -4.39604186e-02 -7.72809207e-01 -4.60518450e-01 -9.93658185e-01 6.46946952e-02 -2.40124151e-01 -9.89659250e-01 9.66809332e-01 -1.56789947e+00 -1.02415812e+00 1.35541952e+00 -5.73959529e-01 -5.93758941e-01 3.83869670e-02 5.90905733e-02 -3.54327530e-01 5.98712802e-01 -3.81371453e-02 8.83594394e-01 8.11803102e-01 -1.34416521e+00 1.25816643e-01 -5.05191565e-01 2.15908676e-01 3.77996713e-01 -6.64065897e-01 4.16426808e-01 -2.10962325e-01 -6.90247416e-01 -9.77262929e-02 -7.89641023e-01 1.67083833e-02 3.23172808e-02 -1.41951099e-01 -1.72585279e-01 5.17136693e-01 -6.82029486e-01 1.12836361e+00 -2.14805365e+00 3.77033293e-01 -2.07757697e-01 2.52734840e-01 -2.78448105e-01 -5.99569023e-01 8.50640595e-01 -3.56525451e-01 2.48186365e-01 1.66937247e-01 -5.33136249e-01 -1.72303803e-03 2.97869444e-02 -8.65505397e-01 2.69957423e-01 3.25406879e-01 1.45354760e+00 -5.62986910e-01 -5.60303271e-01 2.27959111e-01 7.06906736e-01 -6.94936141e-02 -2.96255294e-02 1.33580137e-02 5.38687743e-02 -3.83808047e-01 4.99328703e-01 1.56445473e-01 -7.73855507e-01 2.47509673e-01 -4.99457896e-01 1.48468941e-01 4.35321927e-01 -1.75621405e-01 2.02697158e+00 -7.97708750e-01 1.31995749e+00 -3.27852964e-01 -9.35860336e-01 9.91519690e-01 6.18921876e-01 2.34723300e-01 -1.31273174e+00 2.25742795e-02 -1.79217622e-01 -2.27552533e-01 -5.89589536e-01 8.30933809e-01 -3.41697156e-01 4.52575013e-02 7.56921768e-01 1.18749522e-01 2.03296337e-02 -9.84176993e-02 4.90685433e-01 7.64000714e-01 4.60883975e-02 -6.63364977e-02 1.38802910e-02 2.23399907e-01 1.49284914e-01 -2.44087845e-01 7.59989560e-01 1.75847217e-01 7.88012743e-01 2.42407650e-01 -2.80994326e-01 -9.63993967e-01 -9.22514975e-01 -2.53638420e-02 1.31892049e+00 4.09166627e-02 -6.67937636e-01 -4.32575971e-01 -2.22094715e-01 -2.35629871e-01 9.10833418e-01 -8.92005265e-01 -5.50921321e-01 -2.83834726e-01 -5.67960799e-01 4.04243708e-01 7.44710505e-01 1.88282470e-03 -1.16781056e+00 -6.23571157e-01 -1.36333928e-01 -4.17199880e-01 -1.12061417e+00 -2.53375560e-01 -1.48092985e-01 -1.17323995e+00 -5.50432861e-01 -8.50005329e-01 -7.55747259e-01 7.75400341e-01 6.86416984e-01 1.54522192e+00 9.50640589e-02 -2.57149398e-01 1.15854156e+00 -2.91395396e-01 -1.19829208e-01 -4.00157869e-01 -9.15650129e-02 -2.79452801e-01 -1.46724105e-01 3.58429253e-01 -1.54082447e-01 -4.88329768e-01 -7.39727216e-03 -8.13641429e-01 3.70471627e-01 7.91761994e-01 6.01847947e-01 3.14598799e-01 -8.93429339e-01 4.13612545e-01 -5.60857952e-01 9.06394482e-01 -5.59616864e-01 2.91544527e-01 6.45883501e-01 -4.52571899e-01 1.60822138e-01 4.57723625e-02 -4.66232538e-01 -1.03267312e+00 -2.50101089e-01 3.34993035e-01 -4.72367704e-01 -7.06051812e-02 7.54978001e-01 5.77382565e-01 1.38663068e-01 8.52564216e-01 4.76470947e-01 2.28553087e-01 -4.25871283e-01 4.74773526e-01 4.32574511e-01 5.64601421e-02 -6.37061417e-01 5.65599322e-01 4.83827382e-01 -2.61795908e-01 -8.93865466e-01 -7.19117045e-01 -1.16256587e-01 -3.68434012e-01 -2.23680303e-01 9.67205346e-01 -1.11794865e+00 -3.67833972e-01 -1.93731323e-01 -1.41340327e+00 1.27248317e-01 -3.91854703e-01 3.36217850e-01 -6.06284320e-01 3.46568942e-01 -7.42885292e-01 -7.55566359e-01 -5.59426367e-01 -1.11694586e+00 1.33171153e+00 -1.53782355e-04 -4.75922704e-01 -1.12112093e+00 -4.93927225e-02 5.67608356e-01 5.86994529e-01 2.58304714e-03 1.21020520e+00 -5.98882854e-01 -4.30707306e-01 -2.36536071e-01 -4.47894096e-01 -1.29851937e-01 -4.64741606e-03 -1.40165254e-01 -1.25020278e+00 -3.70216489e-01 -5.20483665e-02 -7.61735737e-01 1.08647037e+00 3.16546947e-01 8.38819444e-01 -2.12189898e-01 -1.94647908e-01 9.25601572e-02 1.35455918e+00 -4.89161350e-02 7.71433353e-01 3.28876436e-01 5.17803490e-01 9.96993661e-01 2.45170131e-01 4.45304550e-02 4.28707600e-01 7.60646403e-01 1.00190081e-01 -4.47370231e-01 -3.54397446e-01 -3.74374986e-01 8.02564204e-01 7.94854164e-01 -1.77642927e-02 -1.25271305e-01 -8.04093540e-01 3.07453960e-01 -1.93195093e+00 -1.31896758e+00 1.23826154e-01 1.89694369e+00 6.46892667e-01 -2.85944313e-01 -1.75990805e-01 -4.54768121e-01 5.93993783e-01 3.14329356e-01 -2.08977997e-01 -6.59962535e-01 -5.42954028e-01 -1.36364043e-01 -6.93758205e-02 2.82399774e-01 -7.37758636e-01 8.67819488e-01 7.35325098e+00 6.66493952e-01 -1.27891827e+00 1.95954666e-01 7.34864116e-01 -2.89032936e-01 -7.98548162e-01 -1.89820439e-01 -3.42281431e-01 -7.83810616e-02 1.17454267e+00 -1.00085877e-01 9.91999879e-02 6.14755034e-01 1.99219733e-01 1.12501815e-01 -1.31890142e+00 1.09944248e+00 7.66667366e-01 -1.81246328e+00 6.75066769e-01 -1.20039452e-02 5.62945485e-01 5.89160733e-02 6.09113157e-01 2.84686178e-01 -1.97463870e-01 -1.32110965e+00 9.42718208e-01 9.05841291e-01 7.71601975e-01 -2.46846095e-01 4.99723226e-01 4.27064039e-02 -9.65324461e-01 -1.13204867e-02 -4.04001117e-01 -4.44830395e-02 -7.07242340e-02 -1.71644360e-01 -5.92745006e-01 4.21297938e-01 3.48603010e-01 8.81987095e-01 -1.08886349e+00 4.83989626e-01 1.24554433e-01 2.38411650e-01 3.93021613e-01 -3.70016955e-02 1.76850051e-01 -1.42390076e-02 3.23436558e-01 1.43250692e+00 1.64084405e-01 -3.61153156e-01 7.49877915e-02 8.73795867e-01 3.61876469e-03 3.92345369e-01 -1.12937200e+00 -4.56658959e-01 2.09197998e-01 1.15834904e+00 -5.73332012e-01 -4.79357392e-01 -8.16895366e-01 8.56602132e-01 4.08851862e-01 6.97791874e-01 -5.34759820e-01 2.15668902e-01 3.06276262e-01 1.07983910e-01 -2.85689980e-02 -1.97139338e-01 -3.64987701e-01 -1.39971995e+00 -1.33273065e-01 -9.11980033e-01 3.30277085e-01 -1.55997026e+00 -1.35214520e+00 7.92708158e-01 -1.03080079e-01 -1.18862855e+00 -5.42287588e-01 -6.58900976e-01 -4.99733418e-01 1.04063487e+00 -1.32133889e+00 -1.58417153e+00 -1.26811061e-02 5.45741618e-01 8.82611036e-01 -3.91179621e-01 1.13191795e+00 -3.87174822e-02 -4.07592177e-01 4.67564821e-01 -8.93834159e-02 -7.46059716e-02 7.38243580e-01 -7.67376363e-01 -4.16624062e-02 3.93562585e-01 6.52943492e-01 9.44316566e-01 7.66162634e-01 -3.96396846e-01 -1.80191088e+00 -7.52315402e-01 9.96816635e-01 -7.19425023e-01 6.94857419e-01 -1.75471559e-01 -8.30251753e-01 8.98689866e-01 9.32422101e-01 -5.01377642e-01 9.77565646e-01 2.92341232e-01 -6.82075918e-01 1.72057778e-01 -8.21339488e-01 7.16273665e-01 7.45150387e-01 -1.33765662e+00 -8.33223045e-01 4.30353045e-01 6.36713505e-01 8.73067081e-02 -7.82541513e-01 1.96890980e-02 4.37918752e-01 -6.94836140e-01 1.17159057e+00 -1.05190158e+00 1.06221545e+00 2.98569705e-02 -5.84256828e-01 -9.20415938e-01 -3.26829672e-01 -1.12932727e-01 1.10573128e-01 1.20918095e+00 7.37709761e-01 -4.28983927e-01 4.26637024e-01 6.09214008e-01 -3.07553001e-02 -6.61411643e-01 -7.63792276e-01 -3.42331052e-01 1.23262964e-01 -3.46017897e-01 -3.09210680e-02 1.04498947e+00 1.66521296e-01 8.87318194e-01 -2.37245470e-01 -2.42444396e-01 1.62699461e-01 5.59461534e-01 6.06675029e-01 -9.18190241e-01 -8.24718252e-02 -5.88219166e-01 -3.31231266e-01 -8.86287987e-01 4.35522497e-01 -1.34669578e+00 -4.21126813e-01 -1.93319476e+00 8.40918720e-01 3.04381579e-01 -4.29395735e-01 5.16021252e-01 2.50350267e-01 4.86790568e-01 5.10177255e-01 4.90470707e-01 -7.06123769e-01 5.20368755e-01 1.27305949e+00 -4.79502141e-01 2.59736270e-01 -5.52054167e-01 -9.75504398e-01 6.55057132e-01 5.48145711e-01 -1.25985578e-01 -5.26690781e-01 -9.40417469e-01 5.49083114e-01 3.33753437e-01 6.06177151e-01 -5.07596195e-01 8.27999413e-02 -1.66509207e-02 7.47627854e-01 -4.65799600e-01 7.86706567e-01 -6.67015851e-01 -2.83926539e-02 4.44317728e-01 -1.06945491e+00 5.52560449e-01 3.67187887e-01 5.08319020e-01 -3.60040605e-01 -1.77749828e-01 5.10763347e-01 -3.02440107e-01 -5.90061426e-01 -1.78071961e-01 -6.15262151e-01 -1.07682966e-01 4.81747955e-01 -4.95457977e-01 -7.12453604e-01 -7.33150125e-01 -9.20328259e-01 -1.78680226e-01 4.57377642e-01 8.03276300e-01 1.03674757e+00 -1.55833519e+00 -6.82304740e-01 -2.00890854e-01 4.28033292e-01 -1.12398696e+00 2.32485324e-01 1.12097120e+00 1.66808516e-01 7.50774145e-01 -2.99562484e-01 -7.49061942e-01 -1.40122843e+00 6.31785989e-01 1.59511656e-01 -3.98396403e-02 -4.98727947e-01 5.15431941e-01 5.03156841e-01 -1.70890450e-01 -4.44924794e-02 -2.34279353e-02 -4.97788310e-01 4.92122859e-01 4.49833363e-01 8.52441490e-02 -1.56963542e-01 -1.11996615e+00 -2.98472136e-01 8.33853781e-01 -1.49477094e-01 -2.60513276e-01 1.19802701e+00 -3.78136098e-01 -1.76090568e-01 8.79317164e-01 1.25053227e+00 -2.16244161e-01 -5.30978918e-01 -4.23306048e-01 3.00810877e-02 -2.06343219e-01 -5.76478764e-02 -8.86047840e-01 -8.20259988e-01 1.39528215e+00 6.71627760e-01 2.37070560e-01 9.15907383e-01 3.53188068e-01 3.84157538e-01 5.00535786e-01 1.98902279e-01 -8.52251887e-01 4.89116192e-01 4.51703489e-01 1.41446483e+00 -1.25851429e+00 -6.93507493e-02 -1.06589952e-02 -1.08912039e+00 1.23098826e+00 4.89581704e-01 5.59602603e-02 -5.68018854e-03 -3.08224320e-01 1.10811412e-01 -3.87020499e-01 -1.18288302e+00 -5.72118983e-02 7.03438163e-01 3.88691902e-01 9.71876621e-01 -1.17843345e-01 8.35127980e-02 3.61860961e-01 2.68553998e-02 -4.22766924e-01 2.84069747e-01 7.83814132e-01 -3.59576404e-01 -7.53064811e-01 -3.68226320e-01 3.46537739e-01 -2.96531171e-01 -4.93277669e-01 -8.15924525e-01 5.14262259e-01 -5.20609379e-01 9.85642135e-01 2.82584280e-01 -3.30839932e-01 1.31230280e-01 3.69659722e-01 7.62210846e-01 -7.84777343e-01 -9.07284021e-01 -5.54098003e-02 2.91230470e-01 -5.74094415e-01 -6.67491317e-01 -4.66693431e-01 -9.88982558e-01 2.99287215e-02 -8.78151730e-02 1.67197630e-01 7.47761011e-01 9.65816677e-01 7.09887207e-01 4.22506183e-01 3.64404708e-01 -8.04608464e-01 -2.84634799e-01 -1.02147067e+00 -4.89972495e-02 5.04374504e-01 1.97064057e-01 -3.78916800e-01 -1.13362461e-01 1.32243723e-01]
[11.173471450805664, 1.564185619354248]
5324c612-d2db-4670-b48a-834d05c8b334
imenet-joint-3d-semantic-scene-completion-and
2106.15413
null
https://arxiv.org/abs/2106.15413v1
https://arxiv.org/pdf/2106.15413v1.pdf
IMENet: Joint 3D Semantic Scene Completion and 2D Semantic Segmentation through Iterative Mutual Enhancement
3D semantic scene completion and 2D semantic segmentation are two tightly correlated tasks that are both essential for indoor scene understanding, because they predict the same semantic classes, using positively correlated high-level features. Current methods use 2D features extracted from early-fused RGB-D images for 2D segmentation to improve 3D scene completion. We argue that this sequential scheme does not ensure these two tasks fully benefit each other, and present an Iterative Mutual Enhancement Network (IMENet) to solve them jointly, which interactively refines the two tasks at the late prediction stage. Specifically, two refinement modules are developed under a unified framework for the two tasks. The first is a 2D Deformable Context Pyramid (DCP) module, which receives the projection from the current 3D predictions to refine the 2D predictions. In turn, a 3D Deformable Depth Attention (DDA) module is proposed to leverage the reprojected results from 2D predictions to update the coarse 3D predictions. This iterative fusion happens to the stable high-level features of both tasks at a late stage. Extensive experiments on NYU and NYUCAD datasets verify the effectiveness of the proposed iterative late fusion scheme, and our approach outperforms the state of the art on both 3D semantic scene completion and 2D semantic segmentation.
['Rui Huang', 'Laiyan Ding', 'Jie Li']
2021-06-29
null
null
null
null
['3d-semantic-scene-completion', '2d-semantic-segmentation']
['computer-vision', 'computer-vision']
[ 2.76601195e-01 1.89669743e-01 1.63486451e-01 -6.41771734e-01 -5.24077535e-01 -2.57951736e-01 4.60499704e-01 -4.88750972e-02 -3.09463680e-01 3.60447377e-01 2.53571182e-01 8.03348143e-03 6.68554381e-02 -8.17165256e-01 -5.82059562e-01 -5.12829840e-01 3.11912358e-01 3.52257520e-01 8.15437615e-01 -1.83867827e-01 2.88707495e-01 3.98380458e-01 -1.82601392e+00 3.17105949e-01 1.21279526e+00 1.32072663e+00 8.16865742e-01 6.06161058e-01 -4.09733027e-01 6.08432531e-01 7.68606886e-02 -9.33088139e-02 5.04971266e-01 -3.81975584e-02 -7.46945322e-01 3.11716080e-01 3.70483607e-01 -5.43765962e-01 -2.05240771e-01 1.10295796e+00 3.93271774e-01 3.89869004e-01 4.39948291e-01 -1.06726658e+00 -3.03792089e-01 1.42952418e-02 -7.23505974e-01 -1.42818198e-01 6.03792846e-01 -1.36571348e-01 7.72446752e-01 -1.08765721e+00 4.48759079e-01 1.40694892e+00 6.18450820e-01 5.30243695e-01 -8.36156666e-01 -5.03895521e-01 6.80754483e-01 -1.25392959e-01 -1.21978784e+00 -1.72437295e-01 1.08255887e+00 -4.71472353e-01 9.61136699e-01 5.60587309e-02 8.85881603e-01 6.84228122e-01 -3.77783664e-02 9.55709517e-01 1.19523585e+00 -7.65263140e-02 2.32834265e-01 2.52757724e-02 2.24609047e-01 8.23293447e-01 -1.02259275e-02 2.28151575e-01 -4.35926169e-01 2.50997514e-01 1.09469998e+00 3.42200994e-01 -1.70536757e-01 -5.59159815e-01 -9.68603015e-01 4.95467603e-01 7.05587685e-01 1.91756770e-01 -4.38283533e-01 5.55036850e-02 -2.23336488e-01 -1.18005857e-01 7.43241727e-01 1.44584358e-01 -7.60587394e-01 1.37715936e-01 -6.97234452e-01 4.06005740e-01 4.11082596e-01 9.58854556e-01 1.32164800e+00 -5.25364041e-01 -8.52830634e-02 6.95762813e-01 8.17441523e-01 3.36448371e-01 8.05200785e-02 -1.16457963e+00 4.74787414e-01 9.56486464e-01 1.05708770e-01 -9.69343245e-01 -5.11887610e-01 -1.80282727e-01 -5.35221696e-01 2.44855016e-01 2.22063512e-01 2.65607834e-01 -1.48386335e+00 1.52083647e+00 7.72369564e-01 4.84550953e-01 1.66302826e-02 1.25737524e+00 1.07403553e+00 5.34989476e-01 2.42123216e-01 3.45860153e-01 9.42243993e-01 -1.27106476e+00 -4.01503950e-01 -4.89969254e-01 3.74324232e-01 -7.84790635e-01 7.53018737e-01 -9.66544151e-02 -1.04317904e+00 -9.03606415e-01 -1.01031101e+00 -7.15254009e-01 -2.74149358e-01 -1.39460161e-01 8.72992635e-01 3.97666574e-01 -9.56004620e-01 5.41730583e-01 -1.02554047e+00 -3.12939703e-01 7.79056489e-01 2.54424006e-01 -2.66460925e-01 -3.05008113e-01 -9.73341405e-01 5.34541905e-01 4.34308290e-01 2.18720198e-01 -7.95999348e-01 -7.18372762e-01 -1.09785724e+00 -2.84874827e-01 2.42650881e-01 -1.06299829e+00 1.02057910e+00 -6.23948753e-01 -1.53109860e+00 1.15043068e+00 -2.60197252e-01 -5.78726046e-02 4.76324320e-01 -5.46532989e-01 2.79129088e-01 -2.77349818e-02 3.90297800e-01 1.11719525e+00 7.53457129e-01 -1.66534567e+00 -1.05324554e+00 -8.46031606e-01 1.79671034e-01 8.85359824e-01 2.07494020e-01 -6.96681023e-01 -9.68570828e-01 -4.71926600e-01 9.17908847e-01 -7.04935431e-01 -7.48743296e-01 1.31279444e-02 -4.76154536e-01 -1.01969883e-01 8.00868213e-01 -6.15523100e-01 6.66092753e-01 -1.99711478e+00 4.71357554e-01 2.69376934e-01 2.96083480e-01 -5.36948107e-02 -7.82763734e-02 -3.22204709e-01 1.77554905e-01 -2.19046380e-02 -5.64660251e-01 -9.95868266e-01 -8.26460123e-03 3.41154724e-01 -1.99013129e-01 2.30440333e-01 2.41984934e-01 1.06486487e+00 -9.86064076e-01 -3.60146761e-01 7.85151005e-01 5.81444263e-01 -7.26734281e-01 5.68974912e-01 -3.14130843e-01 8.74740541e-01 -8.76831889e-01 6.84472263e-01 9.81982291e-01 -2.27134466e-01 -1.65841103e-01 -3.28505248e-01 -1.23495817e-01 2.82820553e-01 -1.26363409e+00 2.59713173e+00 -2.47112915e-01 -2.54274476e-02 3.19155119e-02 -8.25892448e-01 8.62066031e-01 -6.99092597e-02 6.26114428e-01 -9.13682222e-01 2.20943093e-01 2.44847655e-01 -6.98402524e-01 -3.51515502e-01 4.87988412e-01 -1.09512396e-02 -7.38593936e-02 9.11188945e-02 1.61764339e-01 -9.34203982e-01 -3.85117173e-01 1.41307577e-01 7.94503033e-01 7.24943578e-01 3.90931629e-02 -3.14996466e-02 6.52266860e-01 -6.59259036e-02 7.74855793e-01 6.18487537e-01 -2.69709915e-01 8.60202551e-01 8.54930729e-02 -4.14884716e-01 -9.76903558e-01 -1.15038025e+00 7.71808252e-02 6.26277626e-01 1.07581019e+00 -2.79728502e-01 -7.47771502e-01 -9.37431514e-01 -1.07816961e-02 3.85844648e-01 -6.14659190e-01 -5.87351446e-04 -3.71981263e-01 -4.65757102e-01 -1.15783922e-01 6.21433735e-01 8.52929056e-01 -6.90488279e-01 -7.50397921e-01 1.37741461e-01 -2.01896027e-01 -1.53769863e+00 -1.92279965e-01 4.13625807e-01 -8.53131294e-01 -1.14444017e+00 -6.67110085e-01 -7.83411622e-01 7.33070552e-01 6.17155015e-01 9.29599345e-01 2.86420822e-01 -5.34996502e-02 5.20259976e-01 -4.54171091e-01 -2.30582640e-01 -5.91858216e-02 -4.87098321e-02 -1.28364444e-01 -1.24226592e-01 1.18622467e-01 -7.61066616e-01 -6.08769774e-01 3.14686269e-01 -6.71005011e-01 8.17480743e-01 4.40191537e-01 4.59692389e-01 1.03972077e+00 4.10720445e-02 -1.75049324e-02 -7.45631576e-01 -2.12780628e-02 -1.70694456e-01 -5.05954325e-01 6.36696890e-02 -3.13364834e-01 -2.10227948e-02 1.28993526e-01 1.50946200e-01 -1.42019415e+00 4.66089755e-01 -5.64090610e-01 -6.27538621e-01 -5.93344688e-01 2.59347018e-02 -4.90447521e-01 1.36211831e-02 1.40715450e-01 3.88622805e-02 -4.31831837e-01 -5.35515130e-01 4.16473210e-01 3.16357374e-01 5.94140053e-01 -5.53242147e-01 1.02560449e+00 6.61695004e-01 -2.19575658e-01 -4.70482081e-01 -1.42920339e+00 -6.45831168e-01 -1.02429318e+00 -1.96143374e-01 1.41588581e+00 -1.31063211e+00 -3.36104125e-01 9.24399912e-01 -1.20571613e+00 -5.83155215e-01 -4.99506712e-01 3.89010876e-01 -7.68690050e-01 2.07185984e-01 -3.66954714e-01 -6.18437648e-01 -6.44785017e-02 -1.30012763e+00 1.75322187e+00 6.69361174e-01 1.23057954e-01 -9.48377967e-01 -2.03568339e-02 6.91277742e-01 -5.17237335e-02 3.41039002e-01 7.44760513e-01 -2.09133923e-01 -8.96260560e-01 2.08897486e-01 -5.69983661e-01 3.91702890e-01 2.81897157e-01 -3.91312957e-01 -1.38868725e+00 1.46605268e-01 1.23679414e-01 -2.62606919e-01 1.15889883e+00 5.15747249e-01 1.30077910e+00 3.59625608e-01 -4.77510363e-01 1.01638615e+00 1.23569000e+00 3.84509116e-02 5.29121995e-01 3.57983261e-02 1.13488042e+00 6.10324204e-01 8.57753038e-01 2.72821784e-01 8.99488330e-01 5.55896878e-01 6.83355451e-01 -4.39535320e-01 -5.06694794e-01 -5.16269207e-01 2.99445284e-03 8.00644219e-01 -1.74045607e-01 1.35456890e-01 -8.23185146e-01 1.27608508e-01 -1.97648287e+00 -5.04314959e-01 -1.89772382e-01 1.95714939e+00 5.92057586e-01 3.74258488e-01 -4.12662774e-01 4.97307032e-02 4.88530040e-01 2.55930781e-01 -6.83757246e-01 1.17650211e-01 -1.21037476e-01 4.07845080e-01 2.60080397e-01 8.10961962e-01 -1.26187825e+00 1.24712455e+00 4.86179399e+00 6.59113944e-01 -8.20519805e-01 9.23351124e-02 7.66853511e-01 3.40443790e-01 -5.66203773e-01 2.62361854e-01 -7.43083477e-01 2.52277046e-01 1.38189360e-01 3.41579437e-01 2.41394356e-01 8.77411842e-01 1.34331673e-01 -5.05856216e-01 -1.00907493e+00 1.08590376e+00 -1.36909947e-01 -1.33126855e+00 3.09796948e-02 -5.88936768e-02 1.06479239e+00 1.28889214e-02 -8.67488459e-02 1.27261713e-01 4.21513140e-01 -7.39345849e-01 8.05208087e-01 7.43577421e-01 5.31492710e-01 -6.82155728e-01 5.30790627e-01 3.55147511e-01 -1.60214710e+00 -3.91542427e-02 -2.36261368e-01 -7.19550475e-02 3.64224792e-01 7.89277554e-01 -4.96362686e-01 8.49161148e-01 1.01027656e+00 1.21273696e+00 -5.66684425e-01 8.28988254e-01 -5.72154760e-01 -8.70430991e-02 -3.44098151e-01 4.73134220e-01 3.45919073e-01 -1.64599061e-01 5.01707137e-01 6.48501277e-01 2.63601005e-01 4.99834359e-01 3.58720750e-01 1.07644582e+00 8.65063667e-02 -3.23847711e-01 -2.77028143e-01 4.86987799e-01 2.25060001e-01 1.25128472e+00 -1.02668226e+00 -3.91471267e-01 -3.31559867e-01 1.40536642e+00 2.98270643e-01 3.97412151e-01 -7.67251313e-01 -3.62402052e-02 9.83892918e-01 3.85885946e-02 2.78445989e-01 -4.93327379e-01 -7.17053115e-01 -1.35547936e+00 -1.25968859e-01 -1.43017307e-01 1.05018280e-01 -9.07112539e-01 -1.18642354e+00 3.16204339e-01 -2.22248390e-01 -9.14513886e-01 2.37829953e-01 -3.64543706e-01 -3.64474863e-01 9.95611310e-01 -1.95973539e+00 -1.26277184e+00 -9.00227666e-01 7.81232297e-01 8.22756290e-01 6.11535728e-01 6.78690314e-01 3.37766767e-01 -5.25248826e-01 -1.37783393e-01 -6.81599855e-01 -1.28343895e-01 2.37445831e-01 -1.32725275e+00 4.93978143e-01 9.54677403e-01 -1.38600588e-01 1.09807633e-01 1.82952479e-01 -9.94658709e-01 -9.92705584e-01 -1.32647383e+00 6.51166320e-01 -6.08788133e-01 -2.57879216e-02 -4.29162592e-01 -7.57566392e-01 4.18454856e-01 -3.92858475e-01 1.38498083e-01 3.48823071e-01 -2.45658413e-01 -7.38540962e-02 1.70044944e-01 -9.90025878e-01 4.22606349e-01 1.70441842e+00 -4.70385879e-01 -6.43306673e-01 1.42934442e-01 1.30731833e+00 -8.60928595e-01 -7.94380486e-01 8.70320439e-01 3.11141074e-01 -1.31559622e+00 1.30016899e+00 -9.04974490e-02 4.26254094e-01 -5.98408341e-01 -6.67617142e-01 -8.80866945e-01 -1.53211489e-01 -2.55226433e-01 -9.26807448e-02 9.50489163e-01 1.10563435e-01 -2.60739475e-01 9.65523124e-01 9.07576442e-01 -6.52067959e-01 -7.73338377e-01 -8.91303420e-01 -2.46642366e-01 -3.02727044e-01 -9.33509350e-01 7.33695447e-01 8.33215475e-01 -6.48216367e-01 3.53662848e-01 1.76750682e-02 4.06982303e-01 6.95986390e-01 4.80507106e-01 1.01725495e+00 -1.39851809e+00 2.16123551e-01 -3.16699952e-01 -4.04984206e-01 -1.79610813e+00 1.62494913e-01 -1.00360298e+00 1.68845668e-01 -1.89352405e+00 7.13559538e-02 -8.56917500e-01 -1.89852715e-01 4.50261176e-01 -5.27564824e-01 1.85000360e-01 1.15951613e-01 5.90098649e-02 -6.87302172e-01 1.01457059e+00 1.69997191e+00 -6.03356538e-03 -7.61983037e-01 -1.29311368e-01 -6.37743890e-01 1.06036699e+00 2.70343572e-01 -2.78108686e-01 -5.95994771e-01 -5.92017472e-01 -8.03511217e-02 -9.84947383e-02 6.15867138e-01 -1.23447657e+00 2.71253109e-01 -2.81745195e-01 8.29765201e-01 -9.25430179e-01 7.15374887e-01 -9.82851446e-01 -2.16702789e-01 1.81473866e-01 3.30621824e-02 -5.08074284e-01 6.47221580e-02 6.41566396e-01 -1.26911297e-01 3.35671812e-01 9.19785380e-01 -3.68766725e-01 -1.30859160e+00 8.55621934e-01 1.82899773e-01 -1.85097814e-01 1.15364134e+00 -6.94517076e-01 2.67755300e-01 1.54581871e-02 -9.96264040e-01 4.72321510e-01 7.17529535e-01 6.61216617e-01 1.01574814e+00 -1.21043599e+00 -3.37980837e-01 5.66572666e-01 -5.25801182e-02 1.04492092e+00 5.68022728e-01 7.94461071e-01 -3.01409602e-01 1.30525112e-01 -8.78165588e-02 -1.04231513e+00 -8.77049804e-01 1.28751650e-01 4.77433860e-01 -2.77620465e-01 -6.40183687e-01 1.29188943e+00 6.17928445e-01 -7.37416267e-01 2.72878617e-01 -6.53136015e-01 -1.53320506e-01 -2.99273152e-02 1.16628453e-01 7.46887624e-02 5.49583137e-02 -6.14288449e-01 -3.85935724e-01 1.05883396e+00 2.99126860e-02 2.36446075e-02 1.57531571e+00 -6.70355976e-01 -1.18418343e-01 3.01642567e-01 1.12643325e+00 -3.86139065e-01 -1.83353925e+00 -2.39211127e-01 -3.38369876e-01 -7.62008250e-01 1.66010618e-01 -5.62041104e-01 -1.10165954e+00 9.28261817e-01 5.89378417e-01 -1.37836635e-01 1.33528221e+00 2.06569836e-01 1.12987053e+00 -8.15458875e-03 5.52516520e-01 -1.08403313e+00 3.24214607e-01 7.13598490e-01 5.99755108e-01 -1.28352833e+00 2.02926174e-02 -8.96168351e-01 -4.40447271e-01 8.81959736e-01 9.96612787e-01 -6.36383295e-02 8.20886850e-01 6.65069073e-02 -2.14129418e-01 -1.81521252e-01 -2.92598397e-01 -5.59678197e-01 4.16314781e-01 6.34925544e-01 -5.69883920e-02 -1.05544545e-01 1.46336883e-01 7.61660039e-01 -1.26086637e-01 -1.13378994e-01 -5.18187545e-02 9.39272821e-01 -5.55577993e-01 -1.00335634e+00 -2.81974912e-01 2.68391848e-01 2.26852208e-01 1.19103901e-01 -2.58193821e-01 4.76508796e-01 5.71941376e-01 9.23023939e-01 1.84534147e-01 -7.28336751e-01 3.97634625e-01 -2.04360202e-01 5.62663317e-01 -8.38801801e-01 -1.83684960e-01 1.31152347e-01 -3.56637418e-01 -1.14879191e+00 -7.73933828e-01 -5.81036448e-01 -1.70371985e+00 3.36569315e-03 -1.38993189e-01 -1.99493557e-01 7.42572606e-01 1.05429590e+00 3.88548076e-01 6.36723280e-01 7.61373281e-01 -1.39089799e+00 3.57130319e-01 -4.63878721e-01 -3.36813003e-01 4.28106397e-01 2.83998847e-01 -9.61772561e-01 -2.33241737e-01 5.13891838e-02]
[8.459839820861816, -2.860056161880493]
af1934c0-c35d-4cf4-b629-0f69d05dc431
diffsrl-learning-dynamic-aware-state
2110.12352
null
https://arxiv.org/abs/2110.12352v2
https://arxiv.org/pdf/2110.12352v2.pdf
DiffSRL: Learning Dynamical State Representation for Deformable Object Manipulation with Differentiable Simulator
Dynamic state representation learning is an important task in robot learning. Latent space that can capture dynamics related information has wide application in areas such as accelerating model free reinforcement learning, closing the simulation to reality gap, as well as reducing the motion planning complexity. However, current dynamic state representation learning methods scale poorly on complex dynamic systems such as deformable objects, and cannot directly embed well defined simulation function into the training pipeline. We propose DiffSRL, a dynamic state representation learning pipeline utilizing differentiable simulation that can embed complex dynamics models as part of the end-to-end training. We also integrate differentiable dynamic constraints as part of the pipeline which provide incentives for the latent state to be aware of dynamical constraints. We further establish a state representation learning benchmark on a soft-body simulation system, PlasticineLab, and our model demonstrates superior performance in terms of capturing long-term dynamics as well as reward prediction.
['Jia Pan', 'Tingxiang Fan', 'Shang Wen Yao', 'Jialong Li', 'Yunhao Liu', 'Sirui Chen']
2021-10-24
null
null
null
null
['deformable-object-manipulation']
['robots']
[-2.52046406e-01 8.11814144e-02 -4.49164212e-01 -3.56100611e-02 -6.06179595e-01 -5.86868703e-01 8.06857228e-01 -1.67912543e-01 -5.48370779e-01 6.29610479e-01 3.76763582e-01 -3.31998646e-01 -1.43404588e-01 -5.38391948e-01 -9.89140451e-01 -5.34510732e-01 -5.59770584e-01 8.66182923e-01 3.45212132e-01 -6.42381251e-01 -8.25743601e-02 5.95712721e-01 -1.09508514e+00 -2.37388775e-01 5.66015899e-01 2.94018745e-01 3.22454393e-01 7.95175195e-01 6.17833197e-01 7.45574117e-01 -1.08234510e-01 5.83896935e-01 2.15684518e-01 -1.10288337e-01 -6.89394474e-01 -3.17468435e-01 -6.42006425e-03 -4.69919264e-01 -9.29803848e-01 6.16959989e-01 5.56127131e-01 6.52240455e-01 4.91287559e-01 -1.30455768e+00 -2.21914165e-02 4.24341619e-01 -1.11721136e-01 1.63167506e-01 2.99601972e-01 9.96158481e-01 4.73286420e-01 -3.84421676e-01 9.71686423e-01 1.51023436e+00 6.68929696e-01 9.26766753e-01 -1.23415017e+00 -5.85421085e-01 4.37729210e-01 -3.92483957e-02 -7.18302667e-01 -4.31479514e-01 6.02037370e-01 -5.31419218e-01 1.56272733e+00 -2.01763138e-01 8.57552230e-01 1.59204388e+00 9.47520316e-01 9.86084998e-01 8.51398051e-01 2.36334577e-01 4.99215573e-01 -6.34445667e-01 2.33896691e-02 9.64575589e-01 -1.20661249e-02 8.80908012e-01 -3.45231980e-01 -1.60507709e-01 1.20044184e+00 8.25982913e-02 4.89830822e-02 -6.89244688e-01 -1.42306614e+00 6.19070292e-01 4.73276556e-01 -4.78653133e-01 -3.38231832e-01 1.22848237e+00 6.43648386e-01 2.32565597e-01 1.32830115e-02 5.93446314e-01 -8.12601864e-01 -6.51765287e-01 -4.26670969e-01 8.42404008e-01 8.04813385e-01 7.45362341e-01 3.90690684e-01 5.45718729e-01 -1.83861196e-01 2.88056612e-01 6.24906182e-01 5.57042718e-01 5.55880189e-01 -1.28993917e+00 3.50835681e-01 4.39669371e-01 1.82929039e-01 -3.97860944e-01 -7.08975494e-01 -3.50886315e-01 -4.40651983e-01 8.66178334e-01 -8.25813785e-02 -3.58572870e-01 -1.25265431e+00 1.91183794e+00 4.64637458e-01 7.06628621e-01 4.70212787e-01 1.11360586e+00 4.51264411e-01 7.95156658e-01 1.40922308e-01 7.53560960e-02 9.34706688e-01 -1.13353217e+00 -4.53578055e-01 -6.00979149e-01 8.59087527e-01 -1.19684570e-01 1.03731418e+00 3.28981042e-01 -1.09853923e+00 -3.92619461e-01 -1.12975073e+00 -2.06643343e-02 -1.85666904e-02 -1.27263397e-01 1.04619598e+00 -8.06894749e-02 -8.46407235e-01 1.22353804e+00 -2.05404925e+00 -1.25489801e-01 9.40678865e-02 6.84656203e-01 -2.72490084e-01 1.07600987e-01 -1.06663203e+00 1.25835001e+00 3.59435439e-01 -1.25871927e-01 -1.86142981e+00 -6.55507982e-01 -1.32418716e+00 -1.71938971e-01 4.39036846e-01 -9.77247894e-01 1.59923458e+00 -2.75429964e-01 -2.17617083e+00 1.94381505e-01 3.06911975e-01 -6.03082657e-01 6.06522560e-01 -4.95352387e-01 -2.46785164e-01 -1.91811383e-01 -3.50883119e-02 8.20728958e-01 6.77743673e-01 -8.41660976e-01 -1.35142326e-01 8.54763016e-02 2.15718478e-01 4.94950533e-01 3.86421710e-01 -3.94363582e-01 -3.63760173e-01 -4.94816482e-01 2.96216756e-02 -1.62104452e+00 -9.25358653e-01 5.53002805e-02 7.44310841e-02 -5.27362823e-02 1.09592366e+00 -4.30344969e-01 6.79474354e-01 -1.80464065e+00 5.90630114e-01 -8.23978484e-02 -9.92429107e-02 7.51087889e-02 -2.74975389e-01 5.13182998e-01 -3.55180576e-02 -3.35451782e-01 -2.04969794e-01 -2.28467017e-01 3.32851976e-01 5.46754718e-01 -4.82250333e-01 5.82029402e-01 3.34215790e-01 1.25434732e+00 -1.29144764e+00 -1.57354429e-01 2.70688981e-01 3.26010555e-01 -7.92270541e-01 1.36215806e-01 -8.01524460e-01 7.99638808e-01 -5.82468092e-01 3.89185369e-01 1.67579710e-01 -3.51342589e-01 1.10650733e-01 1.73445702e-01 -2.20028684e-01 7.08658636e-01 -1.09849358e+00 2.23285437e+00 -4.73135769e-01 3.01798522e-01 3.87234651e-02 -7.08561003e-01 5.00934303e-01 1.92235917e-01 7.51067936e-01 -4.44780290e-01 6.38641566e-02 3.80621292e-03 2.53000528e-01 -3.53588074e-01 7.58176327e-01 -1.72739580e-01 -6.04716659e-01 4.28607672e-01 5.43232122e-03 -6.15898907e-01 -3.92591432e-02 2.89052069e-01 1.42805898e+00 9.60163534e-01 7.83371553e-02 -3.43411155e-02 -9.31507349e-02 3.49836320e-01 6.87647164e-01 4.15239632e-01 -1.44159004e-01 8.19032565e-02 4.01802808e-01 -1.05154805e-01 -9.35931981e-01 -1.22547781e+00 2.97286600e-01 9.97713029e-01 3.50736260e-01 -4.31663632e-01 -2.18615666e-01 -4.70426112e-01 3.35291415e-01 5.91353416e-01 -4.44345713e-01 -7.00470924e-01 -9.36070621e-01 -5.22364557e-01 4.83620077e-01 9.02318299e-01 5.59150092e-02 -1.28516853e+00 -1.08151591e+00 5.95140934e-01 2.99496591e-01 -7.79892206e-01 -4.62766290e-01 5.68788826e-01 -1.08662832e+00 -8.61203551e-01 -3.24609876e-01 -6.02001607e-01 3.25159490e-01 -1.49071619e-01 8.90623987e-01 -1.07568353e-01 -2.59254068e-01 6.71437085e-01 3.14703472e-02 1.18632913e-01 -7.02749014e-01 1.95677243e-02 4.89997864e-01 -1.11730707e+00 -5.93980730e-01 -5.37424862e-01 -5.05333245e-01 3.85694712e-01 -5.70866764e-01 3.55952322e-01 3.90644550e-01 1.09761894e+00 6.52562737e-01 -4.95868415e-01 3.96893740e-01 -4.13384676e-01 4.15875673e-01 -5.43203712e-01 -8.39448214e-01 -1.94364935e-01 -4.84975308e-01 5.33346534e-01 4.25986558e-01 -7.88873911e-01 -9.54220951e-01 4.32517290e-01 -3.31372738e-01 -5.53334892e-01 2.75060713e-01 8.97397280e-01 1.46001950e-01 -2.02545170e-02 5.71136177e-01 -2.62548830e-02 4.31916028e-01 -2.51546443e-01 4.10859585e-01 -2.78776735e-01 6.20935321e-01 -8.96291077e-01 7.90326715e-01 2.61694312e-01 4.08272296e-01 -2.74281412e-01 -3.36616993e-01 -3.50939214e-01 -2.55294830e-01 -4.70769890e-02 5.50182045e-01 -1.01913774e+00 -1.22124803e+00 3.82505119e-01 -7.55671442e-01 -1.31597936e+00 -4.87575173e-01 7.97027349e-01 -1.22670972e+00 2.10090891e-01 -1.01204395e+00 -5.54586112e-01 -1.34254664e-01 -1.57096362e+00 1.17866695e+00 1.49912432e-01 -2.41788492e-01 -9.93100226e-01 6.30841613e-01 -4.77520615e-01 2.90072560e-01 6.24496758e-01 6.21091247e-01 -4.79147881e-02 -7.47206867e-01 -4.60429378e-02 5.01216769e-01 -9.77095068e-02 -2.09342241e-01 -1.15508288e-01 -4.70172614e-01 -7.66261160e-01 -2.29944855e-01 -7.40578651e-01 9.47640181e-01 5.06301522e-01 7.19983518e-01 -1.24411210e-01 -7.82379389e-01 8.02426517e-01 1.10237312e+00 -8.45834389e-02 5.07837892e-01 1.82577938e-01 6.91797316e-01 2.45358124e-01 8.82577121e-01 3.70480269e-01 4.86677676e-01 7.41115987e-01 6.83826804e-01 1.75791621e-01 -1.17236882e-01 -5.11299491e-01 1.10840690e+00 1.03938079e+00 3.55581135e-01 1.75691471e-02 -1.12425673e+00 2.71498591e-01 -2.31343198e+00 -9.23981369e-01 2.15321213e-01 1.76006258e+00 7.59298086e-01 4.18718249e-01 -8.56637396e-03 -3.69516373e-01 4.28874306e-02 1.68968111e-01 -1.30038941e+00 -2.53266215e-01 4.32796210e-01 8.30276459e-02 6.70130849e-01 6.38191462e-01 -1.06282341e+00 1.48510730e+00 6.47273970e+00 4.73317087e-01 -1.48373163e+00 -6.69835657e-02 9.93655920e-02 -2.06456468e-01 -1.94952488e-01 2.22813472e-01 -8.11098039e-01 2.50519186e-01 1.18759429e+00 -2.41446182e-01 4.84063596e-01 1.17892456e+00 4.56031978e-01 -3.46763507e-02 -1.33866012e+00 6.18756950e-01 -5.20924032e-01 -1.50095534e+00 -3.40094119e-01 1.20499901e-01 6.80901468e-01 6.63090050e-01 2.25603774e-01 9.87285614e-01 1.08227146e+00 -1.15186596e+00 8.12340677e-01 6.42909348e-01 8.08058679e-01 -6.65651202e-01 1.86896726e-01 7.06668973e-01 -1.34662628e+00 -1.02212660e-01 -1.65546149e-01 -3.64002466e-01 7.28305817e-01 -1.97726697e-01 -9.72735941e-01 3.01096022e-01 2.02178240e-01 1.09801960e+00 -2.07533926e-01 8.68422508e-01 -1.75845116e-01 6.42331183e-01 -5.75002372e-01 9.35273170e-02 4.00599688e-01 -1.54309779e-01 7.21536458e-01 8.03360581e-01 1.04749277e-01 -7.78231248e-02 6.84058666e-01 7.80853271e-01 3.27980965e-01 -7.41716266e-01 -7.06163168e-01 -3.39733571e-01 -4.36213687e-02 1.07745266e+00 -7.81090140e-01 -2.71129042e-01 7.87886009e-02 8.99519265e-01 3.43065321e-01 3.04672450e-01 -1.13305068e+00 2.24639937e-01 1.20487595e+00 -1.58495054e-01 9.14637521e-02 -1.08100843e+00 6.29247203e-02 -1.31069183e+00 -4.02441829e-01 -8.01812172e-01 8.56025144e-02 -6.84646547e-01 -8.30383658e-01 6.28181547e-02 3.12815696e-01 -1.44512033e+00 -9.49394703e-01 -7.10288048e-01 -5.46790421e-01 7.18488157e-01 -1.33239162e+00 -1.10499024e+00 -8.70705992e-02 3.90040874e-01 5.97630322e-01 -7.73041621e-02 8.16492319e-01 -2.61000898e-02 -4.72688287e-01 2.60839015e-01 1.04739219e-01 -1.65623829e-01 5.67720234e-01 -1.22126198e+00 1.01156533e+00 6.04560614e-01 -2.44985446e-01 7.30923831e-01 9.56635296e-01 -1.15536606e+00 -2.04758883e+00 -1.11163199e+00 -1.92880243e-01 -6.99833751e-01 8.01943958e-01 -3.59313279e-01 -8.69006097e-01 8.97223771e-01 -3.48196328e-01 3.42997730e-01 6.19417951e-02 -1.30114153e-01 5.98637871e-02 5.19498110e-01 -7.74869978e-01 7.82299221e-01 1.38910675e+00 -4.02296126e-01 -4.72364008e-01 2.68576533e-01 1.11797798e+00 -1.25516033e+00 -7.79050231e-01 5.65518856e-01 5.98399520e-01 -2.86201276e-02 1.08140469e+00 -1.00293458e+00 5.30298412e-01 -3.68624091e-01 1.29182830e-01 -1.62561309e+00 -4.08519894e-01 -9.68508363e-01 -9.61144865e-01 4.82444376e-01 1.12071171e-01 -4.77914751e-01 1.05790114e+00 4.30959672e-01 -7.14248180e-01 -8.96954417e-01 -7.72403955e-01 -1.08710158e+00 2.67584413e-01 -4.54858005e-01 3.54265720e-01 6.92607641e-01 1.54152259e-01 2.73123741e-01 -3.85513812e-01 2.30963811e-01 2.41751149e-01 1.43472133e-02 1.06182492e+00 -8.42679381e-01 -8.10496032e-01 -3.68742764e-01 -2.66152114e-01 -1.52472901e+00 5.62610269e-01 -1.04833305e+00 5.04567206e-01 -1.76630342e+00 -2.76834872e-02 -6.31854296e-01 -1.53935656e-01 5.73137045e-01 -7.87570700e-02 -4.86957937e-01 2.24141687e-01 2.40724638e-01 -6.72438562e-01 1.11640108e+00 1.51859772e+00 1.61921587e-02 -2.99638659e-01 -6.09523617e-02 -8.61435384e-02 8.23690832e-01 6.32423222e-01 -5.23831129e-01 -6.50579393e-01 -2.75669634e-01 2.66070873e-01 5.85001290e-01 6.42557025e-01 -1.20248914e+00 1.88033208e-01 -6.40297174e-01 1.48119956e-01 -6.56681180e-01 6.97703600e-01 -5.90414107e-01 3.34578782e-01 1.05205166e+00 -4.18678671e-01 4.30097520e-01 5.31801820e-01 9.46642458e-01 2.25477144e-01 8.15949738e-02 6.94143951e-01 -3.23799439e-03 -1.09471548e+00 6.67838931e-01 -5.37299454e-01 6.49499223e-02 8.26955378e-01 -6.43475272e-04 -4.05636787e-01 -1.57731220e-01 -7.84481466e-01 6.18966043e-01 8.31959784e-01 5.57447314e-01 5.29581726e-01 -1.31942272e+00 -2.57061303e-01 -5.29565327e-02 4.33683358e-02 2.43061870e-01 1.37515455e-01 7.05084741e-01 -6.44655168e-01 1.18267931e-01 -5.25615036e-01 -5.37150919e-01 -7.75941968e-01 4.17037874e-01 4.17304963e-01 -6.57619596e-01 -1.12709141e+00 6.20574772e-01 8.17740411e-02 -7.33642638e-01 8.98816958e-02 -8.27947319e-01 1.16655089e-01 -5.59039772e-01 -1.30732298e-01 3.12664717e-01 -3.55130315e-01 -4.50780421e-01 -5.19309282e-01 1.48501322e-01 8.01130384e-02 -4.74928975e-01 1.45442390e+00 2.14371473e-01 4.28558111e-01 4.87429231e-01 1.02296746e+00 -5.78664601e-01 -2.28557682e+00 6.89148381e-02 1.20087638e-01 4.93787602e-02 -2.85263872e-04 -7.99136817e-01 -6.01461291e-01 7.18893826e-01 7.36258686e-01 -4.96967643e-01 3.35434705e-01 -1.30488187e-01 1.10192025e+00 8.54415119e-01 7.38662302e-01 -1.24830151e+00 7.69595265e-01 1.29631269e+00 9.82988894e-01 -1.33538294e+00 2.41176292e-01 2.19282042e-02 -7.86267519e-01 9.31916893e-01 9.41799581e-01 -7.86274612e-01 5.87628543e-01 6.61403418e-01 -2.89424807e-01 -2.64678270e-01 -1.37466705e+00 -7.81023353e-02 2.31152460e-01 6.03505135e-01 2.46372163e-01 1.74041130e-02 9.75882187e-02 5.28072715e-01 -1.29343346e-01 9.56924409e-02 5.99505663e-01 1.44484687e+00 -3.18528026e-01 -1.02600825e+00 6.09559752e-02 2.27960497e-01 1.21073127e-01 2.56181419e-01 2.61841655e-01 7.72225082e-01 -4.12662834e-01 1.70375988e-01 4.86576706e-02 -3.44063342e-01 2.06297606e-01 -1.68774888e-01 7.01094627e-01 -1.08008111e+00 -5.72748661e-01 -6.21503778e-03 7.87841305e-02 -9.83525038e-01 1.72457024e-01 -7.54608750e-01 -2.03013277e+00 -1.99592993e-01 -3.81651968e-02 -1.82251319e-01 7.20751286e-01 8.67249191e-01 4.72564697e-01 9.86326277e-01 1.37698397e-01 -1.55375135e+00 -1.03587937e+00 -7.60659635e-01 -5.76951988e-02 2.42866173e-01 5.86202145e-01 -1.13151765e+00 7.95333311e-02 5.12369582e-03]
[4.495150566101074, 1.1038001775741577]
c904e994-c7ee-467d-909f-f5724ede563a
thinkminers-disorder-recognition-using
null
null
https://aclanthology.org/S14-2116
https://aclanthology.org/S14-2116.pdf
ThinkMiners: Disorder Recognition using Conditional Random Fields and Distributional Semantics
null
['Avinesh PVS', 'Joy Mustafi', 'Ashish Mungi', 'Ankur Parikh', 'Lalit Agarwalla']
2014-08-01
null
null
null
semeval-2014-8
['clinical-concept-extraction']
['medical']
[-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]
[-7.241811275482178, 3.688123941421509]
a3f91c1f-82b5-4869-b06e-f982f6453858
rethinking-the-aligned-and-misaligned
2108.12176
null
https://arxiv.org/abs/2108.12176v5
https://arxiv.org/pdf/2108.12176v5.pdf
Rethinking the Misalignment Problem in Dense Object Detection
Object detection aims to localize and classify the objects in a given image, and these two tasks are sensitive to different object regions. Therefore, some locations predict high-quality bounding boxes but low classification scores, and some locations are quite the opposite. A misalignment exists between the two tasks, and their features are spatially entangled. In order to solve the misalignment problem, we propose a plug-in Spatial-disentangled and Task-aligned operator (SALT). By predicting two task-aware point sets that are located in each task's sensitive regions, SALT can reassign features from those regions and align them to the corresponding anchor point. Therefore, features for the two tasks are spatially aligned and disentangled. To minimize the difference between the two regression stages, we propose a Self-distillation regression (SDR) loss that can transfer knowledge from the refined regression results to the coarse regression results. On the basis of SALT and SDR loss, we propose SALT-Net, which explicitly exploits task-aligned point-set features for accurate detection results. Extensive experiments on the MS-COCO dataset show that our proposed methods can consistently boost different state-of-the-art dense detectors by $\sim$2 AP. Notably, SALT-Net with Res2Net-101-DCN backbone achieves 53.8 AP on the MS-COCO test-dev.
['Zihao Huang', 'Degang Sun', 'Junxing Ren', 'Bo Meng', 'Min Li', 'Yang Yang']
2021-08-27
null
null
null
null
['dense-object-detection']
['computer-vision']
[-1.06516711e-01 -2.06752375e-01 -1.64166778e-01 -4.61088747e-01 -9.23688829e-01 -3.48579794e-01 5.69680393e-01 -1.62211120e-01 -7.26031244e-01 4.84289825e-01 -1.14604183e-01 1.23489596e-01 1.11709282e-01 -5.81890762e-01 -8.53537381e-01 -8.38098943e-01 -6.34285761e-03 3.47293824e-01 6.83212519e-01 -1.67357642e-02 1.47846580e-01 4.64801490e-01 -1.23713911e+00 4.76374090e-01 8.11043382e-01 1.36581254e+00 5.10594785e-01 3.90544981e-01 1.14017889e-01 6.03780448e-01 -4.82061505e-01 -4.72561032e-01 6.09404862e-01 1.40046105e-01 -3.24843615e-01 -2.76286066e-01 9.04562950e-01 -2.76111037e-01 -6.05632842e-01 1.18965292e+00 3.34639877e-01 4.72091213e-02 7.80625343e-01 -1.29558158e+00 -8.25909674e-01 3.92088175e-01 -1.12021887e+00 6.14607215e-01 -3.04077893e-01 4.50499833e-01 1.18062449e+00 -1.47423840e+00 1.93975165e-01 1.17131400e+00 6.83050990e-01 3.40766668e-01 -1.31350422e+00 -1.20273221e+00 4.42051530e-01 1.59612939e-01 -1.73867822e+00 -3.25667918e-01 5.34188807e-01 -5.28168142e-01 7.18945920e-01 1.49882689e-01 4.10672843e-01 1.01229668e+00 1.63792998e-01 8.65663409e-01 9.53702867e-01 8.24632272e-02 -3.74542885e-02 1.63906291e-01 2.31466249e-01 7.09541917e-01 6.22801244e-01 2.08769098e-01 -7.51951694e-01 2.46474490e-01 8.66825819e-01 4.20151860e-01 -1.79460004e-01 -6.18281364e-01 -1.37914407e+00 7.10798264e-01 1.27470934e+00 1.24731578e-01 -1.99854746e-01 1.43573761e-01 2.25565866e-01 -2.17136919e-01 4.51110780e-01 6.18038177e-01 -3.68837178e-01 4.13243443e-01 -9.98398244e-01 3.91659826e-01 1.43508881e-01 1.07079554e+00 8.87229741e-01 -3.54158431e-01 -5.88849127e-01 5.79770446e-01 3.96482766e-01 5.44767618e-01 2.82344937e-01 -4.12400573e-01 1.04615819e+00 8.31608653e-01 1.05584070e-01 -1.10847533e+00 -2.82178074e-01 -9.16508436e-01 -9.74507213e-01 2.60021448e-01 6.13218129e-01 1.59344524e-01 -1.24539959e+00 1.65595520e+00 2.64506608e-01 3.40623558e-01 -2.31676057e-01 1.34025991e+00 7.03693032e-01 5.26840568e-01 1.55509681e-01 3.54396909e-01 1.45611453e+00 -1.24514484e+00 -2.43946075e-01 -7.45654881e-01 4.54797238e-01 -5.11576414e-01 1.03295553e+00 4.73592132e-02 -1.08967853e+00 -7.08871543e-01 -1.28521645e+00 -4.04429078e-01 -2.26466045e-01 4.06314760e-01 4.03801411e-01 2.43573889e-01 -5.74678540e-01 4.65460598e-01 -8.89090598e-01 2.10906610e-01 9.77477252e-01 2.76356936e-01 -3.75671864e-01 -1.13454185e-01 -9.23749328e-01 9.32844222e-01 2.23777398e-01 3.53268087e-01 -8.02891374e-01 -9.57839727e-01 -6.99336350e-01 2.28619620e-01 4.43296283e-01 -5.62167525e-01 9.97030139e-01 -5.37454486e-01 -6.95833206e-01 8.88399243e-01 -1.84371173e-01 -4.36539322e-01 6.43507361e-01 -3.44737113e-01 -2.70603538e-01 -8.98244455e-02 6.80127025e-01 9.04941797e-01 8.90810192e-01 -1.15520287e+00 -1.17811859e+00 -7.23621666e-01 -3.29750091e-01 3.20676178e-01 -1.36029914e-01 -1.86159089e-01 -6.77577794e-01 -7.18137324e-01 4.64757651e-01 -7.00555682e-01 -2.25331008e-01 4.93368477e-01 -6.19900942e-01 -2.19843745e-01 7.35739172e-01 -4.00398523e-01 9.86311436e-01 -2.35694218e+00 1.17073990e-01 9.61379632e-02 7.00763285e-01 2.08421662e-01 -2.31704652e-01 -3.61141413e-01 -1.17064184e-02 1.69529721e-01 -8.16093981e-02 -5.91335297e-01 -1.10458240e-01 -3.37465433e-03 -5.38387895e-01 6.56539917e-01 5.69606066e-01 1.01486123e+00 -8.28530192e-01 -4.48393553e-01 1.55063555e-01 3.55045378e-01 -4.51195925e-01 5.00536710e-02 1.21525936e-01 3.86292517e-01 -6.58466935e-01 5.53612113e-01 9.11876917e-01 -3.26945007e-01 -4.08362955e-01 -5.62165499e-01 -2.27822974e-01 2.18645364e-01 -1.06233203e+00 1.69928861e+00 -3.18477035e-01 7.82688141e-01 -1.50345489e-02 -8.03771675e-01 8.77715886e-01 -2.51290828e-01 9.32864174e-02 -9.38181996e-01 -1.04394928e-01 2.37141460e-01 1.24147698e-01 -2.75219202e-01 4.55854386e-01 2.43446920e-02 -4.29051444e-02 -6.41575530e-02 2.56913714e-02 1.00075796e-01 -1.40228197e-01 1.46085620e-01 9.70641792e-01 -5.91151677e-02 2.05342144e-01 -1.57624021e-01 2.48402432e-01 -2.51167506e-01 7.18073130e-01 9.68845427e-01 -5.32340705e-01 9.68191803e-01 5.91164052e-01 -5.19283891e-01 -9.62698877e-01 -1.34527433e+00 -4.07962322e-01 1.08967626e+00 5.74196458e-01 -5.06703779e-02 -3.75134706e-01 -1.00135195e+00 2.47883499e-01 4.37284559e-01 -8.93247783e-01 -1.38559029e-01 -7.20064223e-01 -6.17635131e-01 4.91452485e-01 8.39237750e-01 6.25644743e-01 -7.77337670e-01 -5.12935340e-01 4.76348288e-02 5.26443450e-03 -1.16430628e+00 -6.81431770e-01 3.12074453e-01 -5.38150489e-01 -1.16763961e+00 -7.97165692e-01 -7.54069746e-01 7.90530801e-01 7.68768311e-01 1.04774618e+00 -1.03532523e-02 -3.24556530e-01 -4.45172518e-01 -2.48353511e-01 -3.22948009e-01 2.48538315e-01 1.69544071e-01 -6.74525425e-02 7.99114779e-02 4.43733960e-01 -3.78931582e-01 -1.06759751e+00 7.48053908e-01 -5.63862681e-01 1.67091846e-01 8.27631176e-01 8.91626596e-01 7.33404934e-01 -4.07302260e-01 4.12473381e-01 -6.36501610e-01 -4.57509374e-03 -5.06187320e-01 -7.15109766e-01 2.31011629e-01 -4.76041049e-01 1.38794377e-01 4.05483335e-01 -4.59184200e-01 -6.99673235e-01 3.42164606e-01 1.22819848e-01 -7.30991840e-01 8.02567527e-02 3.47188450e-02 -2.79543459e-01 -1.64356753e-01 8.03460300e-01 1.30085453e-01 -2.33927488e-01 -3.03810954e-01 4.48838294e-01 4.94767308e-01 7.63124228e-01 -3.08558404e-01 1.02480567e+00 6.35652840e-01 -1.28451496e-01 -2.52260804e-01 -1.33695650e+00 -6.06249511e-01 -5.86295903e-01 9.75704193e-02 9.65517521e-01 -1.19082403e+00 -4.71386105e-01 4.03270364e-01 -1.20811737e+00 -4.95419279e-02 -3.87176901e-01 3.73619676e-01 -1.86673239e-01 -4.15650383e-02 -1.90857410e-01 -4.89758432e-01 -2.03793317e-01 -1.45414591e+00 1.38327432e+00 4.63721484e-01 1.05134733e-01 -3.37747157e-01 -2.04238415e-01 2.81537294e-01 3.27294052e-01 -2.62027197e-02 5.18311620e-01 -7.48064220e-01 -9.67741609e-01 -4.08505797e-01 -1.04844069e+00 3.41996908e-01 -1.33512884e-01 -2.41169274e-01 -1.08387101e+00 -3.48261386e-01 -5.02199940e-02 -2.79419541e-01 1.37224817e+00 3.90461236e-01 1.29699528e+00 -2.02871069e-01 -6.52428448e-01 9.51105833e-01 1.12874091e+00 -1.59789503e-01 5.55263340e-01 2.12698832e-01 9.86571729e-01 3.31832498e-01 6.36944890e-01 1.49737552e-01 3.80675912e-01 9.83409286e-01 6.34170353e-01 -2.44264916e-01 -3.43475193e-01 -3.54755193e-01 1.91059187e-01 2.04641864e-01 1.04021639e-01 2.32851040e-02 -7.69935310e-01 4.42735970e-01 -1.90793478e+00 -8.40183318e-01 -2.00054199e-01 2.16928124e+00 6.34424746e-01 3.79282087e-01 3.61539200e-02 -2.69062221e-01 8.43250692e-01 3.67504567e-01 -8.26916575e-01 3.96054417e-01 -4.84752059e-02 1.72470823e-01 8.17278028e-01 2.11084217e-01 -1.39379466e+00 8.12035441e-01 5.24185181e+00 1.08672571e+00 -1.02255094e+00 4.19205874e-01 7.53345251e-01 -5.67679524e-01 2.28488848e-01 -1.72789581e-02 -1.14145327e+00 6.47333622e-01 2.86360562e-01 7.60461241e-02 1.12126052e-01 9.82061505e-01 3.49178202e-02 -2.44683847e-02 -1.29674006e+00 1.01746058e+00 -8.12764019e-02 -1.47355533e+00 -5.40685318e-02 -7.48245418e-03 6.88256145e-01 3.18851978e-01 5.35940409e-01 5.12152076e-01 5.56981042e-02 -1.13882470e+00 1.09689438e+00 3.32020491e-01 8.11463714e-01 -4.88709033e-01 6.38776124e-01 2.97862172e-01 -1.41141856e+00 -1.87290981e-01 -5.24992585e-01 1.07724838e-01 -8.22009370e-02 6.29515111e-01 -6.78801179e-01 2.40705833e-01 9.74695504e-01 7.60643899e-01 -7.73513734e-01 1.30091822e+00 -3.70088518e-01 2.00679019e-01 -1.51033506e-01 -1.63203292e-02 2.38147438e-01 9.03740153e-02 6.41494989e-01 1.12581396e+00 1.56782448e-01 -1.86223313e-02 1.00646719e-01 1.49703896e+00 -2.47294381e-01 -3.83330107e-01 -2.36994371e-01 5.16619563e-01 6.01665437e-01 1.43320704e+00 -6.99259102e-01 -1.86326042e-01 -3.84403765e-01 9.42339182e-01 7.25095093e-01 3.95414531e-01 -1.10567343e+00 -3.72877896e-01 1.00228667e+00 2.58241743e-01 4.45867926e-01 -1.06358722e-01 -6.43040836e-01 -1.19804192e+00 2.71926194e-01 -5.01113892e-01 1.43578336e-01 -6.66689456e-01 -1.53903139e+00 5.14971435e-01 -2.38059536e-01 -1.41332126e+00 3.95777643e-01 -6.97993398e-01 -5.95631540e-01 1.16887128e+00 -1.73288155e+00 -1.14309299e+00 -4.96697605e-01 3.37471485e-01 5.47524750e-01 -7.00571835e-02 2.37485364e-01 4.60918337e-01 -8.93600643e-01 8.66151750e-01 7.47104059e-04 3.62442940e-01 8.11222792e-01 -1.14441693e+00 6.57300830e-01 9.81153965e-01 3.55896860e-01 5.12778342e-01 2.13990405e-01 -5.61162829e-01 -9.18520331e-01 -1.45508122e+00 5.49371302e-01 -7.45888770e-01 5.48220396e-01 -6.44080222e-01 -1.06536925e+00 4.72500741e-01 -3.42302620e-01 7.51969576e-01 9.86481085e-02 1.88201666e-02 -9.12702620e-01 -2.92841375e-01 -8.60642254e-01 4.71137524e-01 1.14104998e+00 -5.24414480e-01 -5.39687693e-01 3.80803049e-01 8.35240662e-01 -5.84550083e-01 -3.84557545e-01 4.20214772e-01 4.15034324e-01 -1.06930721e+00 1.24605846e+00 -5.77010691e-01 5.05121708e-01 -5.19592285e-01 -1.22753724e-01 -1.11513364e+00 -7.14006543e-01 -1.22935139e-01 -1.25877649e-01 9.80823517e-01 5.89372039e-01 -5.55802941e-01 8.75120342e-01 5.29495120e-01 -2.15077206e-01 -9.59338486e-01 -1.23612463e+00 -7.89012671e-01 6.52537346e-02 -3.26162964e-01 5.76267481e-01 7.24040806e-01 -4.26434249e-01 5.12216985e-01 -1.26861989e-01 5.31125128e-01 6.17518008e-01 1.29375709e-02 6.40963733e-01 -1.14819443e+00 -1.91828027e-01 -7.49384046e-01 -5.82616687e-01 -1.30271125e+00 -9.30160284e-02 -8.73441398e-01 2.86235243e-01 -1.23979282e+00 4.91699427e-01 -9.11510408e-01 -5.21886408e-01 6.47095382e-01 -5.25727808e-01 4.75667745e-01 3.55932891e-01 5.32340765e-01 -7.66462207e-01 5.49438953e-01 1.28198004e+00 -2.33035877e-01 -8.70944634e-02 7.76224062e-02 -7.90478170e-01 6.88043535e-01 3.56334329e-01 -7.71219134e-01 3.81808495e-03 -5.69788098e-01 1.31031603e-01 -3.44775409e-01 9.05246079e-01 -1.17533326e+00 5.30025423e-01 7.63453394e-02 8.32621336e-01 -8.05474639e-01 3.39715660e-01 -8.45877290e-01 -3.41247439e-01 3.60001206e-01 -4.63266671e-01 -1.10770769e-01 2.80560609e-02 8.13330293e-01 -8.47255662e-02 -1.35394216e-01 1.00881672e+00 2.16590270e-01 -6.95077002e-01 5.48317313e-01 3.38895082e-01 1.23342186e-01 1.20278656e+00 -3.47212046e-01 -6.07749045e-01 2.20793903e-01 -4.91397679e-01 4.85883832e-01 2.47246459e-01 5.06910086e-01 7.61721253e-01 -1.22981703e+00 -7.32811034e-01 3.84444505e-01 2.22295225e-01 6.31678462e-01 3.61439586e-01 8.62501740e-01 -2.06311405e-01 3.76170993e-01 -1.49489209e-01 -8.03932250e-01 -9.50379074e-01 3.56639564e-01 4.83633608e-01 -3.58641624e-01 -6.12211168e-01 1.21310771e+00 8.01630199e-01 -1.25300989e-01 3.91430050e-01 -4.72912729e-01 -1.49974283e-02 2.42117085e-02 6.41033649e-01 1.59329101e-01 5.79702966e-02 -5.53521872e-01 -5.84802806e-01 4.89974767e-01 -4.96563405e-01 5.32683730e-02 1.14979625e+00 1.41225517e-01 1.90583006e-01 9.74961519e-02 1.23443604e+00 -1.30202442e-01 -1.86149347e+00 -4.48083878e-01 -2.79368103e-01 -7.61581123e-01 1.73072904e-01 -7.40996361e-01 -1.32720423e+00 8.91649365e-01 7.90492296e-01 -5.07503450e-02 8.14446807e-01 2.36663491e-01 5.62103808e-01 9.43197012e-02 6.29844517e-02 -7.06045151e-01 3.25036824e-01 4.51348960e-01 9.00326312e-01 -1.35664415e+00 8.02700073e-02 -5.99088907e-01 -6.95679605e-01 7.43612528e-01 1.17231810e+00 -3.92490685e-01 3.47506553e-01 1.81876868e-01 -3.59198391e-01 -2.73397267e-01 -4.56960917e-01 -2.53354132e-01 7.38245964e-01 5.74970961e-01 8.13554972e-02 1.39837518e-01 8.36547390e-02 9.02772129e-01 9.03068408e-02 -3.57504487e-01 -3.35808359e-02 5.44548154e-01 -4.55346614e-01 -4.32339609e-01 -3.70154023e-01 6.97713435e-01 -1.52914658e-01 -1.85882404e-01 -2.27134570e-01 8.91891301e-01 3.52426529e-01 5.92693806e-01 4.18151259e-01 -5.21169662e-01 5.56242585e-01 -3.38011175e-01 2.54338324e-01 -6.61271632e-01 -5.71854293e-01 -1.49212778e-01 -4.21396255e-01 -7.74468958e-01 1.01043634e-01 -4.99303192e-01 -1.26966000e+00 -2.30511036e-02 -6.05069458e-01 -2.14267597e-01 5.63325167e-01 7.82491267e-01 5.82694292e-01 6.44642413e-01 6.22876525e-01 -1.13326943e+00 -8.27766538e-01 -7.86009312e-01 -4.79817271e-01 3.10499638e-01 4.90715235e-01 -9.02125120e-01 -3.29797447e-01 -4.63197738e-01]
[8.945087432861328, 0.2533133029937744]
8e73a27e-f298-4269-8753-d271298e11f4
social-cost-of-carbon-what-do-the-numbers
2001.08935
null
https://arxiv.org/abs/2001.08935v3
https://arxiv.org/pdf/2001.08935v3.pdf
Social Cost of Carbon: What Do the Numbers Really Mean?
Social cost of carbon (SCC) is estimated by integrated assessment models (IAM) and is widely used by government agencies to value climate policy impacts. While there is an ongoing debate about obtained numerical estimates and related uncertainties, little attention has been paid so far to the SCC calculation method itself. This work attempts to fill the gap by providing theoretical background and economic interpretation of the SCC calculation approach implemented in the open-source IAM DICE (Dynamic Integrated model of Climate and the Economy). Our analysis indicates that the present calculation method provides an approximation that might work pretty well in some cases, while in the other cases the estimated value substantially (by the factor of four) deviates from the "true" value. This deviation stems from the inability of the present calculation method to catch the linkages between two key IAM's components -- complex interconnected systems -- climate and economy, both influenced by emission abatement policies. Within the modeling framework of DICE, the presently estimated SCC valuates policy-uncontrolled emissions against economically unjustified consumption, which makes it irrelevant for application in climate-economic policies and, therefore, calls for a replacement by a more appropriate indicator. An apparent SCC alternative, which can be employed for policy formulation is the direct output of the DICE model -- the socially optimal marginal abatement cost (SMAC), which corresponds to technological possibilities at optimal level of carbon emissions abatement. In policy making, because of the previously employed implicit approximation, great attention needs to be paid to the use of SCC estimates obtained earlier.
['Michael Obersteiner', 'Alexey Smirnov', 'Nikolay Khabarov']
2020-01-24
null
null
null
null
['smac-1', 'smac']
['playing-games', 'playing-games']
[ 1.26200825e-01 2.38917217e-01 -1.61478356e-01 1.67510077e-01 -9.24472958e-02 -6.29064798e-01 9.54019547e-01 2.69658417e-01 -5.15588582e-01 8.51425886e-01 2.52586305e-01 -1.07976758e+00 -5.71089149e-01 -1.07379472e+00 -2.77279824e-01 -9.01236832e-01 3.68685126e-01 2.38153592e-01 -1.04535222e-01 -2.44494706e-01 2.77714461e-01 5.68010986e-01 -1.41812181e+00 -4.84882981e-01 1.01094103e+00 6.41781926e-01 2.68974423e-01 3.45607251e-01 -2.87841558e-01 3.13547015e-01 -3.53222519e-01 -2.17688695e-01 2.67694503e-01 -8.24925661e-01 -6.01498127e-01 -6.28201723e-01 -5.68786919e-01 -1.99554995e-01 5.37857592e-01 1.12692833e+00 1.15603954e-01 4.32015583e-03 9.84413624e-01 -1.15671444e+00 -1.73368141e-01 5.44666529e-01 -1.53178900e-01 -1.17747694e-01 -1.32714346e-01 3.53115946e-01 6.21531785e-01 -3.79890501e-01 3.10170352e-01 1.28450274e+00 4.04790074e-01 6.30841628e-02 -1.12622106e+00 -3.88249874e-01 1.55250235e-02 -1.43636435e-01 -1.16126001e+00 -4.94619496e-02 4.46976334e-01 -5.30702889e-01 1.03698218e+00 7.49309719e-01 1.00477171e+00 4.57215905e-01 4.69622731e-01 -2.69757241e-01 1.79463041e+00 -7.25140035e-01 6.50240660e-01 2.92985320e-01 -1.02907978e-01 -2.65456736e-01 9.17900562e-01 7.38007963e-01 4.66658443e-01 -3.31228375e-02 3.90533626e-01 -3.95640910e-01 -5.78770489e-02 -1.12037830e-01 -9.22606647e-01 5.32790124e-01 4.30389017e-01 6.17763996e-01 -4.39644933e-01 3.50104630e-01 1.67044967e-01 2.26344645e-01 6.20282471e-01 2.57288575e-01 -3.06546181e-01 -1.62768826e-01 -7.76943147e-01 3.13354820e-01 7.64530122e-01 1.71522051e-01 4.99271810e-01 1.38481691e-01 3.21448207e-01 2.70490974e-01 7.58660674e-01 1.10679173e+00 -2.16298103e-01 -1.09906447e+00 9.37580839e-02 5.69568217e-01 5.71152329e-01 -9.26641524e-01 -3.49640459e-01 -4.21188265e-01 -6.41168356e-01 8.10852945e-01 6.57415032e-01 -3.19168657e-01 -3.32630157e-01 1.58086836e+00 3.01963300e-01 -6.27293169e-01 9.58218575e-02 7.16728449e-01 -1.78618520e-01 8.65336597e-01 5.83581686e-01 -5.23785889e-01 1.09765112e+00 -2.74115950e-01 -9.36305761e-01 2.29881987e-01 4.98757780e-01 -4.96378601e-01 7.38402069e-01 7.32677802e-02 -9.31784928e-01 2.23712102e-02 -6.68303370e-01 7.69095898e-01 -1.06537330e+00 -5.53258836e-01 2.88700908e-01 1.00301194e+00 -9.45631742e-01 6.35430455e-01 -5.73381484e-01 -4.73133177e-01 -2.36709639e-01 8.69539753e-02 1.79946750e-01 3.17300707e-01 -1.22054660e+00 1.64464736e+00 3.54827642e-01 4.20188159e-01 -5.17440021e-01 -5.11525273e-01 -4.17786598e-01 2.12601915e-01 1.84752643e-01 -5.92003644e-01 1.00109839e+00 -9.62554157e-01 -1.31636095e+00 3.23215395e-01 3.06786358e-01 -3.86580735e-01 9.14272070e-01 1.77199543e-01 -6.44279540e-01 -2.23166626e-02 -2.40862399e-01 3.45567375e-01 1.58931509e-01 -1.42489183e+00 -3.50351721e-01 -5.67879200e-01 6.63264692e-02 2.76624411e-01 -2.73014754e-02 3.90688092e-01 7.08652258e-01 -3.32711637e-01 -1.92932308e-01 -1.03593409e+00 -4.84419972e-01 -1.07585050e-01 -6.68626279e-02 -1.20861016e-01 5.85670829e-01 -6.26090229e-01 1.22193038e+00 -1.75652635e+00 -1.81885600e-01 4.15315449e-01 -4.38287586e-01 4.00915861e-01 2.97584236e-01 8.32321882e-01 -2.07741529e-01 4.88768101e-01 -8.09353530e-01 1.40964285e-01 3.91062677e-01 3.04152340e-01 -9.88450348e-02 5.37777424e-01 -1.12890983e-02 5.96401870e-01 -1.10977376e+00 -1.06805339e-01 8.19121420e-01 3.44430327e-01 -7.02193081e-02 -6.08539954e-02 -3.43131691e-01 4.94993806e-01 -3.24342608e-01 2.29570702e-01 9.48234499e-01 5.80828667e-01 3.58444601e-01 1.58786982e-01 -1.09192860e+00 9.51039344e-02 -1.13905847e+00 1.03926682e+00 -5.71345568e-01 2.29731172e-01 1.34497151e-01 -8.78292620e-01 8.59788477e-01 4.03557003e-01 3.23317349e-01 -8.42687547e-01 4.32116278e-02 7.50801325e-01 3.29795659e-01 -3.76116782e-01 3.94209713e-01 -5.08495808e-01 1.48918509e-01 4.70894754e-01 -4.62276846e-01 -6.35069668e-01 -1.32068336e-01 -2.16223702e-01 3.72110784e-01 4.85826761e-01 4.55619752e-01 -1.14933646e+00 8.00711930e-01 1.06300235e-01 1.99275956e-01 3.80278192e-02 -2.98236459e-01 -6.81505874e-02 7.01940358e-01 -1.79707393e-01 -1.12796843e+00 -7.95801222e-01 -4.15563941e-01 5.36578834e-01 1.15073487e-01 2.25350901e-01 -9.51007664e-01 -3.90946507e-01 1.04390264e-01 1.47450948e+00 -3.87794912e-01 7.16641918e-02 -4.14886713e-01 -1.14282691e+00 2.59196907e-01 -2.56919805e-02 4.07194883e-01 -8.67885172e-01 -1.16574335e+00 2.53280610e-01 1.22487113e-01 -3.21033239e-01 4.38043058e-01 2.40724877e-01 -8.16204667e-01 -1.13620424e+00 -9.19329047e-01 2.95012236e-01 6.26153767e-01 1.34591669e-01 9.03569341e-01 -1.48393065e-01 3.37568045e-01 3.85817379e-01 -1.50138453e-01 -8.51626515e-01 -1.01642072e+00 -1.92250386e-01 -9.36052576e-02 -3.87434632e-01 2.74848431e-01 -5.41069508e-01 -7.76630461e-01 2.93876946e-01 -1.00459552e+00 -6.38244972e-02 2.36419022e-01 6.37796074e-02 1.53469592e-01 8.64912421e-02 1.17558563e+00 -5.21654189e-01 5.62551558e-01 -6.30383253e-01 -9.65004802e-01 1.09119713e-01 -1.41536653e+00 -2.43709721e-02 4.59517330e-01 2.69367784e-01 -1.60862541e+00 -3.72836232e-01 -2.81299591e-01 2.81920612e-01 -4.33122903e-01 6.05103970e-01 -3.49070162e-01 1.08238943e-01 1.25687078e-01 -2.17257813e-01 1.54030129e-01 -6.66605353e-01 4.24059749e-01 6.34743392e-01 4.14627790e-01 -6.64798021e-01 7.78763413e-01 2.28129297e-01 4.30370867e-01 -5.96740246e-01 -1.31279036e-01 -1.61393479e-01 -3.63338828e-01 -8.33704412e-01 8.14926386e-01 -8.23828876e-01 -4.85243559e-01 4.31114495e-01 -9.54886377e-01 -2.58621126e-01 -5.29340029e-01 4.83832628e-01 -4.07229364e-01 2.12806463e-01 2.10496485e-01 -1.63307512e+00 -1.44398585e-01 -1.14394057e+00 1.72967419e-01 -3.74487825e-02 -2.03847066e-01 -1.18243587e+00 3.68439257e-01 6.37642741e-02 9.88547504e-01 9.23689902e-01 1.14103401e+00 2.29850244e-02 -2.87036985e-01 1.03082195e-01 -1.90349013e-01 6.83761537e-01 2.86942691e-01 4.10610557e-01 -9.83007371e-01 -1.16487138e-01 1.41109213e-01 2.20019892e-01 6.98464215e-01 2.77802080e-01 6.34469390e-01 -3.78376007e-01 -1.94053743e-02 -7.41574019e-02 2.22085404e+00 5.16404510e-01 9.21070397e-01 5.59198439e-01 5.32222306e-03 1.20209312e+00 4.97926623e-01 3.09919029e-01 1.36626005e-01 4.45696622e-01 1.12738264e+00 6.26894832e-02 -7.62550086e-02 -1.25728175e-02 4.31531161e-01 7.17354298e-01 -7.01042593e-01 -1.20857000e-01 -1.02094936e+00 6.33871734e-01 -1.53566957e+00 -9.13635910e-01 -7.08697557e-01 2.61160398e+00 4.68135417e-01 2.66078919e-01 2.10679755e-01 2.73130208e-01 6.06328964e-01 2.39151910e-01 -2.27078319e-01 -1.10250962e+00 -1.16270646e-01 -8.31104517e-02 7.90371597e-01 5.74871659e-01 -3.89379561e-01 1.80271298e-01 6.36588526e+00 5.50263584e-01 -8.39265287e-01 1.09283075e-01 6.67779744e-01 4.30147082e-01 -8.93355608e-01 5.59803188e-01 -1.58476159e-01 6.52205348e-01 1.66643190e+00 -5.65994620e-01 2.70102710e-01 2.77740717e-01 1.06140208e+00 -7.43734002e-01 -3.74308348e-01 -1.93316564e-01 -8.19100380e-01 -8.62412333e-01 -1.24095373e-01 5.09564161e-01 7.64046013e-01 -8.44699368e-02 -1.38021156e-03 -7.17625991e-02 4.01053816e-01 -8.06073189e-01 8.87791872e-01 6.33762598e-01 8.69529545e-01 -1.02442765e+00 9.85455930e-01 5.17364860e-01 -1.07676446e+00 -2.27566302e-01 -3.52440238e-01 -3.98562372e-01 2.04054654e-01 8.51853907e-01 -1.33768156e-01 8.97246540e-01 4.07861561e-01 -9.98110250e-02 -1.11909367e-01 7.21608460e-01 -6.89273551e-02 7.62919843e-01 -5.09813488e-01 2.78676748e-02 5.43597281e-01 -1.02954066e+00 4.91622120e-01 1.31490350e+00 4.76789743e-01 8.20772424e-02 -8.95318627e-01 1.31431556e+00 5.43568432e-01 1.85350612e-01 -8.69441867e-01 -3.95837389e-02 4.97882336e-01 1.03754163e+00 -6.31722391e-01 -3.07361465e-02 -5.84270418e-01 2.12994397e-01 -4.37389672e-01 3.26081872e-01 -7.84373045e-01 -1.82307780e-01 5.91762602e-01 1.41224608e-01 -2.16482118e-01 -8.44276845e-02 -4.63774085e-01 -6.44618213e-01 -2.22546577e-01 -3.78525257e-01 -1.25628084e-01 -5.87284982e-01 -7.63904333e-01 7.42725432e-02 6.01799488e-01 -8.97067785e-01 -3.60428482e-01 -4.62404907e-01 -9.80439603e-01 1.38707602e+00 -1.73261058e+00 -8.60691786e-01 1.28286660e-01 -5.18726446e-02 6.70354366e-02 5.39598227e-01 9.33867335e-01 -6.15626983e-02 -4.77059543e-01 -2.53354520e-01 7.81469762e-01 -6.36309564e-01 2.65901297e-01 -1.47372329e+00 1.36228681e-01 8.36135507e-01 -8.99333000e-01 2.97771692e-01 1.05948400e+00 -6.64674103e-01 -8.03062916e-01 -1.00554085e+00 1.01454771e+00 -2.28165492e-01 8.56794417e-01 6.30081818e-02 -6.46852612e-01 9.55663770e-02 8.05518866e-01 -5.88144898e-01 4.25462782e-01 -3.78760099e-01 8.90236497e-02 -9.82407853e-02 -1.53848064e+00 4.63642240e-01 6.22225761e-01 -1.92144290e-01 -5.96589148e-01 9.78493541e-02 7.21028805e-01 5.02152264e-01 -1.13815379e+00 3.24071616e-01 8.45652103e-01 -1.13957465e+00 7.01129436e-01 -1.98432758e-01 2.18657866e-01 -3.32405657e-01 -2.52846628e-01 -1.51434267e+00 -7.73981512e-02 -3.05867583e-01 2.17671633e-01 1.53345573e+00 4.38013613e-01 -1.15949118e+00 9.37764421e-02 9.66354430e-01 -6.38645515e-02 -5.62426269e-01 -1.27576041e+00 -9.66767251e-01 8.81463945e-01 -4.43564236e-01 1.01919079e+00 1.00330389e+00 4.86427844e-02 -4.63662088e-01 1.51930153e-01 1.69908218e-02 8.20496559e-01 -8.02982599e-02 2.67758042e-01 -1.28064716e+00 3.65270168e-01 -7.10422337e-01 3.88391912e-01 7.28603303e-02 -7.45532662e-02 -5.13892651e-01 -4.14470553e-01 -1.60204864e+00 -4.46872786e-02 -5.44066131e-01 -5.77979445e-01 6.91127330e-02 2.07020521e-01 -2.97165543e-01 7.85937250e-01 1.27461001e-01 4.41040605e-01 6.97453558e-01 1.12058461e+00 8.70358124e-02 -2.65813451e-02 2.81222127e-02 -7.62532830e-01 8.35447967e-01 1.04550958e+00 -5.98620474e-01 -3.02832752e-01 -2.72658262e-02 5.57363987e-01 9.97793600e-02 6.64647460e-01 -1.02813256e+00 -2.81512082e-01 -7.91399181e-01 -1.44776329e-01 -6.16190851e-01 -2.67846048e-01 -1.49910045e+00 1.13517463e+00 9.87162411e-01 -1.45283595e-01 5.81961125e-02 1.52605623e-01 3.75541270e-01 1.97105259e-01 -5.33263505e-01 7.26787031e-01 -3.61509651e-01 -2.32711703e-01 -3.89413238e-01 -1.01039815e+00 -7.09214330e-01 1.05120432e+00 -2.45220676e-01 -6.26450002e-01 -1.80259496e-02 -5.62580407e-01 1.13920964e-01 9.32645082e-01 -5.52321896e-02 -1.62347645e-01 -1.18384337e+00 -6.31630301e-01 -4.34538931e-01 -2.04758793e-01 -2.20634431e-01 1.23459347e-01 7.18239546e-01 -8.00513148e-01 8.20262730e-01 -1.51828662e-01 2.12206133e-02 -3.44405800e-01 7.59509325e-01 8.14829171e-01 -1.62907496e-01 -2.61150092e-01 -1.91088602e-01 7.60388300e-02 -4.66483235e-01 -2.77425438e-01 -4.59244013e-01 -5.74669950e-02 1.93268925e-01 2.15562567e-01 9.49642420e-01 -1.24599420e-01 -7.57311523e-01 -3.71318191e-01 4.36553389e-01 9.88138795e-01 -3.10105383e-01 1.15572393e+00 -6.32693112e-01 -5.78136623e-01 6.64340138e-01 7.63427496e-01 -7.02421591e-02 -1.12336016e+00 4.78188992e-01 1.05444729e-01 -2.86666304e-01 2.74978071e-01 -1.40331662e+00 -7.53152072e-01 7.92345345e-01 7.17145681e-01 5.36876857e-01 1.25492203e+00 -6.36519194e-01 1.54870257e-01 1.95622578e-01 3.56267840e-01 -1.44448578e+00 -8.55011642e-01 -1.31280780e-01 1.14864087e+00 -8.55318785e-01 1.08954996e-01 -2.82192111e-01 -5.05599864e-02 1.00236917e+00 2.20105797e-01 4.77272011e-02 7.71944106e-01 2.41996333e-01 -1.68588355e-01 2.05786631e-01 -4.44144458e-01 -2.02637732e-01 -3.12237978e-01 5.66804588e-01 2.22930819e-01 7.61329174e-01 -1.08212399e+00 1.81650072e-01 3.13238442e-01 8.09412170e-03 6.81528568e-01 6.58123851e-01 -6.99728072e-01 -9.88145590e-01 -8.62728059e-01 1.42613173e-01 -3.71450037e-01 -1.82712860e-02 -4.67520148e-01 1.17754424e+00 5.44510245e-01 1.14494288e+00 5.91238551e-02 1.94889024e-01 2.12549970e-01 1.54785216e-01 -2.66964361e-02 1.88636377e-01 -7.56570339e-01 5.92456125e-02 2.62530237e-01 -2.82454282e-01 -1.03893542e+00 -7.71716654e-01 -9.99253631e-01 -6.53525531e-01 -5.53402007e-01 5.14497459e-01 1.32922304e+00 1.01994526e+00 2.00650934e-02 4.37557757e-01 7.50263095e-01 -8.74447107e-01 -7.92865932e-01 -8.75828862e-01 -9.15598571e-01 -2.02811912e-01 1.39997602e-01 -4.92580563e-01 -7.83745825e-01 -4.74394411e-01]
[5.615076065063477, 3.782487154006958]
2852743c-306a-4e58-b90a-83da16c0a532
learning-cross-image-object-semantic-relation
2207.00784
null
https://arxiv.org/abs/2207.00784v1
https://arxiv.org/pdf/2207.00784v1.pdf
Learning Cross-Image Object Semantic Relation in Transformer for Few-Shot Fine-Grained Image Classification
Few-shot fine-grained learning aims to classify a query image into one of a set of support categories with fine-grained differences. Although learning different objects' local differences via Deep Neural Networks has achieved success, how to exploit the query-support cross-image object semantic relations in Transformer-based architecture remains under-explored in the few-shot fine-grained scenario. In this work, we propose a Transformer-based double-helix model, namely HelixFormer, to achieve the cross-image object semantic relation mining in a bidirectional and symmetrical manner. The HelixFormer consists of two steps: 1) Relation Mining Process (RMP) across different branches, and 2) Representation Enhancement Process (REP) within each individual branch. By the designed RMP, each branch can extract fine-grained object-level Cross-image Semantic Relation Maps (CSRMs) using information from the other branch, ensuring better cross-image interaction in semantically related local object regions. Further, with the aid of CSRMs, the developed REP can strengthen the extracted features for those discovered semantically-related local regions in each branch, boosting the model's ability to distinguish subtle feature differences of fine-grained objects. Extensive experiments conducted on five public fine-grained benchmarks demonstrate that HelixFormer can effectively enhance the cross-image object semantic relation matching for recognizing fine-grained objects, achieving much better performance over most state-of-the-art methods under 1-shot and 5-shot scenarios. Our code is available at: https://github.com/JiakangYuan/HelixFormer
['Botian Shi', 'Jiayuan Fan', 'Tao Chen', 'Baopu Li', 'Jiakang Yuan', 'Bo Zhang']
2022-07-02
null
null
null
null
['fine-grained-image-classification']
['computer-vision']
[ 2.27765039e-01 -9.66542140e-02 -4.26699311e-01 -5.49803853e-01 -7.85014749e-01 -1.58249378e-01 5.34105480e-01 1.82146534e-01 3.96867143e-03 2.75256753e-01 1.64324135e-01 1.65846363e-01 -5.94646633e-01 -1.23499608e+00 -6.80685103e-01 -5.96680939e-01 1.13138497e-01 3.81976932e-01 6.76997244e-01 -3.62255961e-01 1.62982106e-01 5.80220044e-01 -1.97398698e+00 8.31020296e-01 7.12767243e-01 1.55270660e+00 2.45407343e-01 1.78499073e-01 -5.84930003e-01 9.31809008e-01 -2.58286715e-01 -3.71836573e-01 2.95278043e-01 -2.67147064e-01 -9.18461740e-01 6.08977973e-02 4.97397512e-01 -5.34568429e-02 -3.23306054e-01 1.20103943e+00 4.07805175e-01 6.97793439e-02 4.64660257e-01 -1.21175420e+00 -7.75923491e-01 5.65438151e-01 -8.56587410e-01 5.83536446e-01 4.99534700e-03 1.62233293e-01 1.07039869e+00 -1.01466310e+00 4.46400046e-01 1.42426419e+00 4.89733040e-01 1.02493450e-01 -9.85017359e-01 -1.05248499e+00 4.83740047e-02 5.14605939e-01 -1.71161973e+00 -3.27335060e-01 7.04752982e-01 -4.50539649e-01 9.45470810e-01 1.60167277e-01 4.75170583e-01 6.46392643e-01 1.05605893e-01 5.95246196e-01 1.07179439e+00 -2.02182099e-01 -2.24043112e-02 -9.65984911e-03 4.00194645e-01 8.37910056e-01 1.10497154e-01 1.32859349e-01 -6.57638133e-01 5.21796569e-03 6.72644317e-01 2.99894422e-01 -1.68968439e-01 -4.81402725e-01 -1.09296203e+00 7.83712804e-01 9.65035141e-01 6.38763249e-01 -5.20781040e-01 -8.65450427e-02 4.81051415e-01 3.51916671e-01 3.89914393e-01 1.86099425e-01 -4.06515449e-01 2.88648725e-01 -6.41733050e-01 5.68901896e-02 2.84306198e-01 1.03900385e+00 1.18082058e+00 -4.09244657e-01 -6.48268223e-01 1.08127224e+00 1.72904447e-01 8.20774361e-02 4.69698161e-01 -4.79662478e-01 5.04791141e-01 1.15938330e+00 -5.43944180e-01 -1.15589154e+00 -1.65012375e-01 -6.83850169e-01 -9.63456750e-01 2.39162557e-02 5.54163754e-02 4.56590205e-01 -1.00544643e+00 1.35088599e+00 5.44643164e-01 3.00675929e-01 -1.29868448e-01 9.77395773e-01 1.33297837e+00 4.61542219e-01 3.72736633e-01 3.86195742e-02 1.91889954e+00 -1.01142001e+00 -3.48346382e-01 -6.42538592e-02 5.51215529e-01 -6.56317532e-01 1.14514148e+00 -1.65409684e-01 -7.44898856e-01 -8.90820384e-01 -9.98563349e-01 -8.93888399e-02 -6.63451612e-01 -2.37949595e-01 6.58834755e-01 3.38045597e-01 -6.00410640e-01 3.98536861e-01 -2.47185156e-01 -2.72477031e-01 1.02640247e+00 1.05062395e-01 -4.28694636e-01 -3.81961048e-01 -1.46574044e+00 5.57949662e-01 5.98412752e-01 -2.06432894e-01 -8.15769792e-01 -1.15208948e+00 -8.66982222e-01 3.89658451e-01 7.08890736e-01 -8.57298136e-01 8.75769973e-01 -8.27933371e-01 -6.81674123e-01 1.07058871e+00 -7.20760897e-02 -2.49204382e-01 6.30839244e-02 6.88961074e-02 -5.08371294e-01 1.51485011e-01 5.78092635e-01 7.32035279e-01 7.92491734e-01 -1.05777407e+00 -1.04613507e+00 -7.26958215e-01 1.39059633e-01 2.10282594e-01 -2.44921774e-01 5.53464107e-02 -7.64807463e-01 -8.40046108e-01 4.27586772e-02 -4.18558776e-01 2.45615877e-02 7.62081798e-03 -2.71306962e-01 -6.15981638e-01 8.43663335e-01 -2.03030214e-01 1.20181537e+00 -2.20414591e+00 -2.98531353e-01 1.81309775e-01 3.39175612e-01 2.30451584e-01 -2.92829782e-01 1.69916615e-01 -3.13201845e-01 -7.97034521e-03 -1.44559508e-02 1.45535931e-01 -1.05181724e-01 1.14903174e-01 -9.55954120e-02 6.98060617e-02 4.71535116e-01 1.38023925e+00 -8.16813946e-01 -7.20669389e-01 2.47078747e-01 2.81829804e-01 -1.61411300e-01 1.01141572e-01 1.84651427e-02 -3.58534604e-02 -7.64459312e-01 9.70801532e-01 7.65658975e-01 -5.03317475e-01 -1.73372552e-01 -9.25951779e-01 1.85490549e-01 -1.35014713e-01 -1.21402895e+00 1.66700923e+00 -3.48766208e-01 2.14677989e-01 -3.39788139e-01 -1.01366985e+00 1.14559031e+00 -1.44626528e-01 5.07154346e-01 -1.49728405e+00 1.97317719e-01 8.41823146e-02 -1.15730450e-01 -5.69363177e-01 3.13558519e-01 -3.23534697e-01 -1.54298842e-01 4.17734832e-01 2.27132693e-01 3.94057423e-01 1.95398238e-02 2.22961619e-01 9.51093674e-01 -9.36683267e-02 5.00208497e-01 -3.26435059e-01 5.62213540e-01 2.71656620e-03 5.04646361e-01 7.96629429e-01 -2.91962266e-01 4.17287260e-01 4.95783873e-02 -6.12390995e-01 -6.27899706e-01 -1.08353567e+00 -1.92296520e-01 1.52910829e+00 8.09297979e-01 -3.40830088e-01 -4.49139774e-01 -9.24773335e-01 2.19798863e-01 4.06219572e-01 -8.92110109e-01 -4.36377227e-01 -3.92552197e-01 -6.49711192e-01 5.88350892e-01 5.90117872e-01 9.64979172e-01 -1.19203293e+00 -7.78750181e-01 3.26826721e-02 -2.87803113e-01 -8.83803606e-01 -4.84737128e-01 3.14327627e-01 -4.80676800e-01 -1.23063612e+00 -6.27541006e-01 -7.80101478e-01 3.44623595e-01 8.55678082e-01 1.39682472e+00 9.21550542e-02 -6.06743336e-01 1.19338565e-01 -5.13397753e-01 -2.22689301e-01 1.07201025e-01 -1.90284476e-01 -4.42715973e-01 3.66810262e-01 7.18444943e-01 -3.35458606e-01 -7.62444854e-01 6.36253476e-01 -9.08685386e-01 9.49729681e-02 7.70916462e-01 8.02464604e-01 1.04321969e+00 4.64285880e-01 5.15891135e-01 -8.97769332e-01 3.86065573e-01 -6.75304353e-01 -2.54088968e-01 5.21553755e-01 -6.06834054e-01 1.31000668e-01 2.24018976e-01 -2.74262756e-01 -1.25891495e+00 -2.11170495e-01 -3.95356938e-02 -5.91497838e-01 -4.49973702e-01 3.18116307e-01 -3.14591497e-01 -1.48044869e-01 6.18105948e-01 2.65947819e-01 -3.00684154e-01 -4.35924262e-01 4.38974977e-01 5.01377344e-01 4.93596852e-01 -5.91144621e-01 6.19108200e-01 4.84390110e-01 -1.33997291e-01 -2.99235314e-01 -1.10281742e+00 -9.22266841e-01 -4.35204595e-01 -1.89696342e-01 8.74892294e-01 -1.09516799e+00 -5.42376041e-01 3.46677959e-01 -7.26741850e-01 4.30966588e-03 -4.70324337e-01 5.99755272e-02 -3.24547499e-01 3.51471384e-03 -3.02348703e-01 -3.29698503e-01 -5.21595478e-01 -1.05061078e+00 1.50813890e+00 5.92891097e-01 2.79011745e-02 -5.22770822e-01 -1.58097759e-01 6.34755790e-01 2.86707848e-01 -2.72597801e-02 1.04733562e+00 -5.53592324e-01 -7.50261724e-01 2.09540188e-01 -8.80539536e-01 -4.84227249e-03 2.82712668e-01 -4.42809790e-01 -9.72939730e-01 8.85003898e-03 -1.74404413e-01 -3.27436030e-01 1.06732035e+00 2.73139507e-01 1.42237806e+00 -8.21308568e-02 -4.87383574e-01 6.68027699e-01 1.60541916e+00 1.43875480e-01 6.97162807e-01 5.45018137e-01 8.15823674e-01 6.47183299e-01 1.17443073e+00 3.32663834e-01 4.30668801e-01 8.33843708e-01 4.47822541e-01 -2.06394941e-01 -4.99169320e-01 -1.47378981e-01 -2.14387298e-01 1.61345452e-01 5.92607819e-02 1.20009534e-01 -8.19419146e-01 7.11099029e-01 -1.85240519e+00 -1.08895230e+00 -9.59266424e-02 1.72209084e+00 7.26305187e-01 7.87242651e-02 2.43038516e-02 -4.21430655e-02 9.05417025e-01 2.09675014e-01 -6.74625039e-01 -1.08419158e-01 -2.81217963e-01 3.59922111e-01 3.11715692e-01 -1.28698274e-01 -1.15976262e+00 9.83699203e-01 3.78114104e+00 1.55003810e+00 -1.03050995e+00 4.14580911e-01 8.92220914e-01 -5.92350867e-03 -2.43962631e-01 -1.78709045e-01 -7.85314202e-01 3.65036339e-01 3.30323547e-01 3.55203338e-02 7.65554905e-02 8.67996156e-01 -1.98084861e-01 -1.06728844e-01 -8.11659992e-01 1.30168808e+00 -4.18186672e-02 -1.71636903e+00 2.67722309e-01 -1.53427050e-01 6.82267189e-01 7.03006461e-02 -1.11130454e-01 4.58728254e-01 -4.44030836e-02 -9.19972003e-01 6.43144190e-01 5.61919212e-01 9.17983949e-01 -8.72994006e-01 7.40421772e-01 2.05853507e-01 -1.88014650e+00 -1.81666255e-01 -4.32555020e-01 3.87719840e-01 -7.14697763e-02 7.48502493e-01 -4.82318431e-01 1.02686465e+00 1.32064402e+00 6.57614291e-01 -7.21735299e-01 8.04442823e-01 2.18858689e-01 3.08095872e-01 5.73695712e-02 3.27854395e-01 3.37168753e-01 5.90941198e-02 2.49359220e-01 1.31795847e+00 8.22385699e-02 3.97730708e-01 5.01351207e-02 8.44865322e-01 3.72748934e-02 5.08203655e-02 -3.17978173e-01 1.43394619e-01 5.24772823e-01 1.47872615e+00 -9.17159617e-01 -4.46876884e-01 -4.11081076e-01 8.12738299e-01 3.68166834e-01 1.43702537e-01 -9.16010201e-01 -5.36203325e-01 9.01043117e-01 2.18515009e-01 6.75547719e-01 4.69868362e-01 -4.21055615e-01 -9.19472873e-01 1.32524982e-01 -8.40797961e-01 9.81521189e-01 -6.92909300e-01 -1.57184899e+00 7.09674835e-01 -4.85281534e-02 -1.06427383e+00 2.92972028e-01 -2.24644303e-01 -4.83411998e-01 8.99961948e-01 -1.81546128e+00 -1.53829253e+00 -8.49932969e-01 1.12394440e+00 8.53688419e-01 -7.34729245e-02 5.89261711e-01 5.67346156e-01 -2.86062151e-01 7.70024955e-01 -4.49046820e-01 -3.98408137e-02 5.21272242e-01 -9.21053350e-01 1.94902077e-01 5.59176385e-01 2.04071075e-01 4.62022036e-01 2.42511690e-01 -6.44491136e-01 -1.09345627e+00 -1.57059801e+00 6.10362411e-01 -4.23595682e-02 4.84450132e-01 -9.53906924e-02 -1.27659440e+00 1.25643387e-01 -1.90188304e-01 4.44409072e-01 5.60892999e-01 1.61663890e-01 -7.61380553e-01 -4.61786181e-01 -1.10121381e+00 2.58898646e-01 1.45072591e+00 -6.65567517e-01 -6.58177912e-01 1.11560285e-01 8.05071175e-01 -5.96692674e-02 -1.14413381e+00 9.38550293e-01 6.40585840e-01 -1.15924597e+00 1.29585540e+00 -5.60161233e-01 4.44521755e-01 -5.55416524e-01 -3.83531153e-01 -9.50623691e-01 -8.21976244e-01 6.46343678e-02 -1.24449559e-01 1.47847557e+00 -4.85769883e-02 -5.69295585e-01 6.40260756e-01 -4.82377671e-02 -1.14196777e-01 -1.11709321e+00 -6.92082047e-01 -7.82275081e-01 -2.79706717e-01 -4.15996104e-01 9.79831994e-01 1.00924182e+00 -5.73980689e-01 4.44279164e-01 9.08485204e-02 1.75577775e-01 5.55815399e-01 7.57367194e-01 5.57746232e-01 -1.13770676e+00 -3.10910374e-01 -6.55857861e-01 -6.11861467e-01 -8.34960580e-01 -4.53497469e-03 -1.00490201e+00 -3.66659239e-02 -1.50297117e+00 7.00052381e-01 -8.69252920e-01 -7.59261370e-01 6.27800405e-01 -3.83000970e-01 7.03478634e-01 2.60947078e-01 2.69334882e-01 -8.24544609e-01 6.42738342e-01 1.27796006e+00 -5.69773436e-01 1.90844402e-01 -2.64034837e-01 -1.01489592e+00 5.14223933e-01 5.79455256e-01 -5.30182183e-01 -5.69804430e-01 -4.21594046e-02 -1.23507142e-01 -1.98756665e-01 6.48123384e-01 -9.64581847e-01 2.56894141e-01 -2.30578095e-01 5.40701628e-01 -6.99765205e-01 4.99474145e-02 -6.37349308e-01 4.79410321e-01 3.54111463e-01 -1.90931320e-01 -4.20058936e-01 2.31018543e-01 5.50357878e-01 -4.76706594e-01 1.47550151e-01 9.42637026e-01 -1.89969838e-01 -1.52424145e+00 6.53854430e-01 3.53849828e-01 1.46865919e-01 1.28179002e+00 -4.78929222e-01 -5.98835707e-01 2.09545329e-01 -4.13884640e-01 1.46045938e-01 1.56132087e-01 7.53416955e-01 6.95698261e-01 -1.54498315e+00 -5.85921049e-01 3.25633675e-01 8.25294018e-01 -3.37228440e-02 9.23217237e-01 8.14538181e-01 1.15864210e-01 3.79837990e-01 -4.76365685e-01 -7.92207599e-01 -1.44484174e+00 8.25967848e-01 3.76384407e-01 -4.14869666e-01 -6.19383991e-01 1.14998579e+00 1.07034075e+00 -2.65202135e-01 -1.16636761e-01 4.12324443e-02 -3.42869073e-01 2.93598175e-01 6.35051370e-01 1.63914368e-01 3.78283173e-01 -8.55966806e-01 -5.06822765e-01 8.74476790e-01 -2.71251112e-01 6.01786256e-01 1.16872406e+00 -2.53039122e-01 -3.57475244e-02 2.11562783e-01 1.33545899e+00 -3.08079988e-01 -1.03335929e+00 -7.45168090e-01 -5.54681979e-02 -7.39895761e-01 8.17022026e-02 -8.60253334e-01 -1.28089333e+00 6.00278258e-01 9.12215650e-01 2.40618773e-02 1.58536828e+00 7.24770784e-01 8.65357459e-01 -1.24409482e-01 5.65997899e-01 -7.09863722e-01 1.47763819e-01 2.47101307e-01 7.80843973e-01 -1.29686284e+00 -2.55053133e-01 -6.92303956e-01 -6.14470780e-01 8.39844286e-01 7.85512805e-01 5.51061966e-02 7.46103346e-01 7.16252625e-02 -1.48865715e-01 -7.93671310e-01 -6.43144608e-01 -6.67195737e-01 7.87533402e-01 5.87191820e-01 9.10688639e-02 2.31520101e-01 2.42082272e-02 7.52907872e-01 1.47273853e-01 -7.36889336e-03 -3.30671817e-01 7.57870734e-01 -6.47747397e-01 -7.75368571e-01 -2.92018801e-01 7.98949182e-01 -1.25421524e-01 -3.16728115e-01 -2.19633773e-01 6.78547442e-01 7.28321373e-01 8.44245613e-01 3.21338862e-01 -5.65943778e-01 5.01422346e-01 -2.11866692e-01 4.08131331e-01 -6.33819759e-01 -5.98265409e-01 -1.70819417e-01 -3.43195386e-02 -1.12151909e+00 -4.12343293e-01 -3.53722095e-01 -1.25597799e+00 -1.62934318e-01 -2.91076183e-01 -1.25991032e-01 7.41791502e-02 1.04775012e+00 6.65953815e-01 8.00779939e-01 5.24317801e-01 -5.11420548e-01 -1.25342667e-01 -6.73594296e-01 -6.57417297e-01 6.19864225e-01 1.01528063e-01 -1.11026382e+00 1.81916907e-01 -1.23923071e-01]
[9.692957878112793, 1.98551607131958]
e7060067-b543-4d93-8d8b-5c57cace8955
layoutdiffusion-controllable-diffusion-model
2303.17189
null
https://arxiv.org/abs/2303.17189v1
https://arxiv.org/pdf/2303.17189v1.pdf
LayoutDiffusion: Controllable Diffusion Model for Layout-to-image Generation
Recently, diffusion models have achieved great success in image synthesis. However, when it comes to the layout-to-image generation where an image often has a complex scene of multiple objects, how to make strong control over both the global layout map and each detailed object remains a challenging task. In this paper, we propose a diffusion model named LayoutDiffusion that can obtain higher generation quality and greater controllability than the previous works. To overcome the difficult multimodal fusion of image and layout, we propose to construct a structural image patch with region information and transform the patched image into a special layout to fuse with the normal layout in a unified form. Moreover, Layout Fusion Module (LFM) and Object-aware Cross Attention (OaCA) are proposed to model the relationship among multiple objects and designed to be object-aware and position-sensitive, allowing for precisely controlling the spatial related information. Extensive experiments show that our LayoutDiffusion outperforms the previous SOTA methods on FID, CAS by relatively 46.35%, 26.70% on COCO-stuff and 44.29%, 41.82% on VG. Code is available at https://github.com/ZGCTroy/LayoutDiffusion.
['Xi Li', 'Ying Shan', 'Zhongang Qi', 'XueWei Li', 'Xianpan Zhou', 'Guangcong Zheng']
2023-03-30
null
http://openaccess.thecvf.com//content/CVPR2023/html/Zheng_LayoutDiffusion_Controllable_Diffusion_Model_for_Layout-to-Image_Generation_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Zheng_LayoutDiffusion_Controllable_Diffusion_Model_for_Layout-to-Image_Generation_CVPR_2023_paper.pdf
cvpr-2023-1
['layout-to-image-generation']
['computer-vision']
[-6.81567043e-02 -1.65248200e-01 7.45379627e-02 -8.50656256e-02 -4.14653271e-01 -4.85920221e-01 5.04217565e-01 -3.96778714e-03 -1.25032971e-02 5.45646906e-01 3.00346971e-01 -4.06277515e-02 -2.30862796e-01 -9.34380829e-01 -7.21608877e-01 -8.73962998e-01 2.84541577e-01 1.02140918e-01 3.72377455e-01 -2.31413350e-01 3.22852701e-01 5.38191080e-01 -1.24678004e+00 1.86781451e-01 1.22517133e+00 9.08541083e-01 9.58986461e-01 4.95650262e-01 -2.23945543e-01 6.57975733e-01 -7.64544487e-01 -1.31375343e-01 1.92936510e-01 -5.42900443e-01 -3.98312837e-01 4.32573825e-01 4.73058194e-01 -2.90377110e-01 -2.74919838e-01 1.07560408e+00 7.01490045e-01 9.81574878e-02 7.24366307e-01 -1.17731547e+00 -1.34158456e+00 7.11981475e-01 -1.20333529e+00 2.07718208e-01 -1.54971965e-02 3.63262534e-01 7.06049979e-01 -8.31754863e-01 6.05880678e-01 1.34589672e+00 -1.09697860e-02 3.23680520e-01 -1.25089884e+00 -6.40948951e-01 5.40002882e-01 1.71026826e-01 -1.51134098e+00 -3.15519810e-01 9.79644299e-01 -4.51991081e-01 5.16130805e-01 2.52167672e-01 5.73610306e-01 9.22185719e-01 1.11270316e-01 8.86471033e-01 1.03200185e+00 -3.31017196e-01 -1.62027124e-02 1.41592562e-01 -1.73239484e-01 6.12341285e-01 3.54167759e-01 -2.85544455e-01 -3.21019739e-01 4.16697383e-01 1.06442034e+00 5.25562372e-03 -4.41494256e-01 -4.42902356e-01 -1.34871495e+00 4.74167824e-01 1.00970221e+00 4.69119638e-01 -4.03665364e-01 1.38911813e-01 -1.31185576e-02 -2.21995026e-01 3.00099820e-01 6.88511550e-01 -1.40028009e-02 3.56839627e-01 -6.80613875e-01 4.35415000e-01 2.55563796e-01 1.30229950e+00 7.84139156e-01 1.80822760e-01 -7.24176228e-01 9.89400029e-01 2.78154999e-01 7.39002168e-01 2.75854945e-01 -9.69386697e-01 6.17377996e-01 7.17779577e-01 -1.31099597e-02 -1.31242847e+00 -1.77767575e-01 -6.87762141e-01 -1.22979128e+00 -8.83540662e-04 8.51177722e-02 -1.51534811e-01 -9.37157333e-01 1.59859669e+00 2.73149222e-01 -4.62729819e-02 -2.24309415e-01 9.62061942e-01 8.54458511e-01 9.45591986e-01 -1.23946369e-02 -6.43080100e-02 1.27308023e+00 -1.43234670e+00 -8.78475428e-01 -2.87580103e-01 2.76401013e-01 -9.42057490e-01 1.17953563e+00 3.09588969e-01 -1.32199037e+00 -7.74965107e-01 -1.08145666e+00 -2.07564756e-01 -3.06797743e-01 4.92391437e-01 4.61756527e-01 2.75837392e-01 -1.03935826e+00 1.65723749e-02 -2.50407517e-01 -1.90274402e-01 5.97140193e-01 1.19005285e-01 2.24804189e-02 -3.23454142e-01 -8.59800756e-01 5.58384836e-01 4.73687559e-01 1.33870825e-01 -8.34084272e-01 -7.08784282e-01 -6.88129127e-01 1.52634040e-01 5.40604055e-01 -7.01791584e-01 1.00371599e+00 -8.15421879e-01 -1.28367496e+00 3.61099124e-01 2.46251728e-02 2.01581158e-02 3.56868476e-01 -4.48796265e-02 -1.76951975e-01 3.47811123e-03 2.30323449e-01 1.03070998e+00 6.77320123e-01 -1.58584905e+00 -5.96463382e-01 -2.55839169e-01 1.77797034e-01 4.93136376e-01 -3.25877190e-01 -3.40943605e-01 -9.11653399e-01 -8.56067002e-01 -2.40277685e-02 -6.19944572e-01 -2.60620296e-01 1.18730990e-02 -6.30732715e-01 -6.09189942e-02 9.43531156e-01 -5.28261900e-01 1.59618270e+00 -2.22193813e+00 4.65002239e-01 -4.47724760e-02 3.46146643e-01 4.43295062e-01 -3.75397265e-01 4.56071854e-01 1.70530409e-01 2.06521168e-01 -2.35687509e-01 -3.74339908e-01 -1.18926562e-01 3.82869169e-02 -7.76074082e-02 1.04653068e-01 2.38534570e-01 1.19667685e+00 -8.36381733e-01 -5.69969773e-01 3.82284522e-01 3.61506015e-01 -6.10420406e-01 2.68334389e-01 -3.04441988e-01 4.98441964e-01 -6.26751304e-01 6.58709586e-01 1.02558756e+00 -2.97803253e-01 1.60136148e-02 -4.26873654e-01 -2.23067656e-01 -4.18670535e-01 -1.06836700e+00 1.94907022e+00 -4.90021348e-01 5.18463612e-01 1.96787983e-01 -5.99334598e-01 9.61259246e-01 -1.21647567e-01 2.31474459e-01 -1.18355203e+00 2.61933893e-01 -2.03206837e-02 1.27305165e-01 -5.75163186e-01 6.11641884e-01 3.22891176e-01 1.50501290e-02 1.18250430e-01 -1.64538026e-01 -1.64139047e-01 3.43494177e-01 4.04732525e-01 7.07088232e-01 1.62030496e-02 6.57253042e-02 -3.70450854e-01 5.65027714e-01 -2.96391964e-01 4.63489085e-01 5.06042540e-01 5.46784140e-02 8.78880203e-01 5.57656348e-01 3.08415610e-02 -1.06055295e+00 -1.06847942e+00 5.65533750e-02 6.24886930e-01 6.57609582e-01 -3.62549007e-01 -1.00161874e+00 -3.77133578e-01 -9.65334550e-02 6.54029787e-01 -6.13166988e-01 -1.24568671e-01 -5.61758459e-01 -5.22461414e-01 1.86081752e-01 5.10793328e-01 9.55470443e-01 -1.13894570e+00 -2.26176307e-01 1.26149669e-01 -2.04184830e-01 -9.23118949e-01 -9.53012466e-01 -3.75357807e-01 -4.83748317e-01 -7.11807072e-01 -1.09108150e+00 -8.76345456e-01 9.29750264e-01 6.45525396e-01 8.93815398e-01 1.31739840e-01 -3.63548815e-01 1.31868487e-02 -3.92636001e-01 -2.14396298e-01 1.22128632e-02 2.75229782e-01 -4.40019995e-01 3.50768119e-01 -3.66786808e-01 -5.67502081e-01 -9.10636365e-01 3.52033466e-01 -1.22557104e+00 6.13365889e-01 9.95223939e-01 7.81699479e-01 6.39989078e-01 1.91369921e-01 4.02345330e-01 -7.34908879e-01 7.94306219e-01 -5.88615239e-01 -5.39827108e-01 3.76889795e-01 -3.78147393e-01 -6.82517439e-02 4.52429295e-01 -4.84211415e-01 -1.27448821e+00 -3.86849754e-02 3.81544568e-02 -5.95619142e-01 -7.67059177e-02 2.45840400e-01 -7.36372113e-01 1.10127926e-01 4.13128108e-01 3.29531521e-01 -1.99693188e-01 -5.67357004e-01 6.00399196e-01 5.52589834e-01 3.86905998e-01 -7.33596385e-01 6.39056325e-01 2.75266528e-01 -1.52172178e-01 -7.24881768e-01 -5.78420401e-01 -1.54533505e-01 -6.52795970e-01 -2.88558990e-01 1.11827230e+00 -7.33688831e-01 -5.44385850e-01 6.22075737e-01 -1.25356758e+00 -6.14197731e-01 -1.38529599e-01 -3.44520696e-02 -2.76642323e-01 2.32846081e-01 -3.41830879e-01 -5.00366747e-01 -1.68960527e-01 -1.45125735e+00 1.11212826e+00 5.44333339e-01 1.34908453e-01 -8.50356877e-01 -3.17490280e-01 1.84319541e-01 5.49473464e-01 2.28089303e-01 7.98569143e-01 2.57951766e-01 -1.21385753e+00 2.22640768e-01 -7.73887753e-01 2.64599204e-01 2.36709222e-01 1.61350444e-01 -5.72769403e-01 -1.35005325e-01 -3.13616306e-01 7.15626627e-02 8.23947012e-01 5.11659205e-01 1.24043822e+00 -2.61364639e-01 -4.17229950e-01 7.68897116e-01 1.50959313e+00 5.07786453e-01 8.11144650e-01 1.95575178e-01 1.04988635e+00 5.61925769e-01 5.63878000e-01 3.59087378e-01 5.50049841e-01 8.38607132e-01 3.52989525e-01 -3.68940890e-01 -6.52951360e-01 -5.22299826e-01 4.77250777e-02 7.48811007e-01 2.48691916e-01 -8.05585980e-01 -7.25534081e-01 5.38547873e-01 -1.96952438e+00 -8.13221991e-01 -1.39820516e-01 1.72883832e+00 4.39177960e-01 -6.95500225e-02 -1.61244199e-01 -2.32036859e-01 7.82227457e-01 2.96245515e-01 -5.19248188e-01 -5.50664999e-02 -4.29358035e-01 -3.56080830e-01 4.40273762e-01 4.62194175e-01 -9.20883656e-01 1.04439163e+00 5.28447390e+00 1.29491639e+00 -1.00022829e+00 1.38564408e-01 9.41416621e-01 -1.11138880e-01 -5.16551912e-01 -1.38399392e-01 -7.39921510e-01 7.07879543e-01 1.32849604e-01 -1.38700023e-01 5.78444242e-01 4.56374079e-01 3.07083726e-01 -3.76394987e-01 -4.55107063e-01 1.20361221e+00 1.12962574e-01 -1.40340602e+00 3.90255392e-01 2.72567958e-01 1.10404718e+00 -4.57936496e-01 4.72720563e-01 4.85476442e-02 8.47861096e-02 -8.62936318e-01 9.21133399e-01 7.25814998e-01 6.45789564e-01 -7.46703625e-01 2.62789309e-01 4.84762341e-01 -1.38081467e+00 -1.86625719e-01 -3.46923292e-01 1.93147421e-01 2.47292429e-01 6.36845529e-01 -2.35462442e-01 6.56801105e-01 6.08404517e-01 8.31821799e-01 -8.03884149e-01 1.03064144e+00 -3.24588478e-01 3.87183577e-02 4.63404953e-02 -2.07306538e-03 3.74346584e-01 -3.46265227e-01 4.18362707e-01 9.56470430e-01 4.65289026e-01 1.13075487e-01 2.15180188e-01 1.22673881e+00 1.14703216e-02 2.38138095e-01 -5.64736843e-01 1.50498776e-02 4.75463361e-01 1.41845930e+00 -8.45941544e-01 -4.08661187e-01 -1.49316981e-01 1.02150631e+00 4.97835040e-01 6.30472124e-01 -9.97804642e-01 -3.23683918e-01 5.02845049e-01 2.89674759e-01 4.67179894e-01 -3.14649940e-01 -2.97922492e-01 -1.08484900e+00 1.41994178e-01 -7.41369963e-01 -8.15438330e-02 -1.17262399e+00 -1.05836368e+00 7.04051256e-01 8.32770839e-02 -1.09608781e+00 3.98145437e-01 -3.59659582e-01 -6.36566281e-01 1.03348684e+00 -1.32114530e+00 -1.29278004e+00 -5.31668127e-01 3.66569757e-01 7.38385081e-01 -7.65390247e-02 2.35467762e-01 5.17872751e-01 -9.15084422e-01 4.79998797e-01 -4.64527309e-02 -1.44503385e-01 6.69839144e-01 -1.07437563e+00 2.83102185e-01 8.72244179e-01 -8.05595145e-02 5.67862332e-01 2.81776786e-01 -7.62483358e-01 -1.40781116e+00 -1.21531475e+00 4.04353678e-01 -2.54288435e-01 2.19455913e-01 -4.40121859e-01 -9.41935301e-01 2.93654472e-01 6.76550031e-01 -2.54133642e-01 2.00739503e-02 -3.93153906e-01 -1.11037284e-01 -3.84627461e-01 -8.66231143e-01 9.22005236e-01 1.31282079e+00 -1.53650090e-01 7.81390220e-02 1.49289653e-01 9.17991102e-01 -5.59813440e-01 -5.46578944e-01 2.43620872e-01 1.93958014e-01 -1.07806540e+00 8.78392279e-01 1.55432150e-01 4.92470801e-01 -8.09094667e-01 -9.56021696e-02 -1.39395487e+00 -7.35239387e-01 -4.86118287e-01 7.04759955e-02 1.56008840e+00 2.85384864e-01 -4.62406486e-01 3.95106316e-01 2.29638562e-01 -2.89532304e-01 -8.39210570e-01 -3.67031962e-01 -6.08252943e-01 -1.21050261e-01 -2.68200301e-02 8.82592022e-01 6.91249311e-01 -4.30302083e-01 5.80915034e-01 -4.46372688e-01 2.03031570e-01 5.67433357e-01 1.26158908e-01 8.36517572e-01 -5.72821796e-01 -2.97220021e-01 -8.65930617e-01 -1.42391443e-01 -1.40241432e+00 -2.58307338e-01 -6.39341295e-01 -3.09927855e-02 -1.92689025e+00 1.54598311e-01 -5.82954705e-01 -2.61184990e-01 2.16184676e-01 -3.35902482e-01 1.97324798e-01 4.21816498e-01 6.45180717e-02 -6.42496288e-01 7.69270003e-01 2.12047052e+00 -3.57706815e-01 -1.60669878e-01 -3.92735898e-01 -1.00504518e+00 2.69867986e-01 8.34779024e-01 -1.01801254e-01 -8.18654478e-01 -9.29838657e-01 -5.39687350e-02 1.07678466e-01 2.04452768e-01 -1.02492368e+00 3.66825759e-01 -1.91303611e-01 4.33160067e-01 -7.88059354e-01 3.05996060e-01 -6.82214081e-01 3.38611871e-01 3.55040282e-01 -2.12099865e-01 2.23222062e-01 2.20211133e-01 5.44919074e-01 -2.47661427e-01 2.55214591e-02 7.66693830e-01 -4.21992801e-02 -9.06132996e-01 5.46827614e-01 5.79935126e-03 -1.07515253e-01 1.37163842e+00 -3.78874838e-02 -4.46213752e-01 -2.80722111e-01 -3.79993647e-01 4.30253446e-01 5.19784033e-01 6.70017958e-01 6.83885574e-01 -1.53543568e+00 -5.73892891e-01 3.18474412e-01 1.13874413e-01 2.56193459e-01 9.77422655e-01 8.06425512e-01 -5.98109543e-01 4.39368963e-01 -2.73600042e-01 -5.41816473e-01 -1.04373360e+00 8.76646876e-01 1.40122786e-01 -6.75236657e-02 -4.48515803e-01 8.77820253e-01 9.39569294e-01 -4.16143946e-02 1.90366209e-01 -4.11964685e-01 -1.80431753e-01 8.89108032e-02 6.41494513e-01 2.47672141e-01 -3.42380583e-01 -7.32260883e-01 -5.60672674e-03 8.12757254e-01 -1.50042981e-01 4.20830213e-02 1.00967312e+00 -4.10316586e-01 -2.36725286e-01 8.49657133e-02 1.10858178e+00 5.96463755e-02 -1.53547478e+00 -1.96743026e-01 -5.51562250e-01 -6.59694672e-01 1.13952331e-01 -8.62622619e-01 -1.43716407e+00 9.77301717e-01 4.50832427e-01 1.56722635e-01 1.36747384e+00 -8.75813812e-02 6.54472947e-01 -1.70991778e-01 2.34504342e-01 -7.42441952e-01 4.63650823e-01 1.89368710e-01 1.33750689e+00 -8.74737620e-01 9.48865432e-04 -5.55001259e-01 -8.24238777e-01 7.32035160e-01 1.07074213e+00 -1.41853109e-01 5.25206208e-01 2.51737595e-01 -3.42687778e-02 -1.20627090e-01 -6.70800865e-01 -2.40658045e-01 4.94975030e-01 6.93188071e-01 2.23708600e-01 1.72398001e-01 -1.42389238e-01 4.37043309e-01 1.66142508e-01 -3.02851319e-01 2.95699507e-01 6.65002108e-01 -3.99565488e-01 -1.18307459e+00 -3.53863418e-01 2.38330826e-01 -1.93396863e-02 -1.15373559e-01 -1.91413134e-01 6.65808082e-01 4.97094601e-01 7.69140720e-01 1.75619528e-01 -3.80727768e-01 4.49119955e-01 -5.28647482e-01 4.09834772e-01 -6.06099725e-01 -2.07142398e-01 5.46212375e-01 -2.16704428e-01 -3.62630039e-01 -2.54561603e-01 -4.22428221e-01 -1.00408876e+00 -3.21217269e-01 -2.57536799e-01 -1.44188330e-01 5.23958147e-01 4.87285465e-01 6.72573805e-01 1.12624860e+00 5.68561435e-01 -9.91522610e-01 7.05685169e-02 -8.58859777e-01 -7.51829803e-01 2.48049051e-01 1.52103320e-01 -7.64860570e-01 1.07203387e-01 2.77963630e-03]
[11.387269973754883, -0.7685655355453491]
79687516-2469-44c0-b687-ded41e9a45ed
improving-diversity-and-reducing-redundancy
null
null
https://ieeexplore.ieee.org/abstract/document/9206644
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9206644
Improving Diversity and Reducing Redundancy in Paragraph Captions
The purpose of an image paragraph captioning model is to produce detailed descriptions of the source images. Generally, paragraph captioning models use encoder-decoder based architectures similar to the standard image captioning models. The encoder is a CNN based model, and the decoder is a LSTM or GRU. The standard image captioning models produce unsatisfactory results for the paragraph captioning task due to the lack of diversity in the generated outputs [9]. The paragraphs generated from standard image captioning models lack in language diversity and contain redundant information. In this work, we have proposed an approach with language discriminator for increasing the diversity in language, and dissimilarity score using word mover’s distance [4] for reducing redundant information. Using this approach with a state-of-the-art model at testing time, we have improved the METEOR score from 13.63 to 19.01 for the Visual Genome dataset
['and Pushpak Bhattacharyya', 'Sriparna Saha', 'Chandresh S.', 'Kanani']
2020-07-19
null
null
null
international-joint-conference-on-neural-2
['dense-captioning']
['computer-vision']
[ 5.56800902e-01 4.61035877e-01 -7.04349130e-02 -3.79181325e-01 -7.39462256e-01 -4.33213651e-01 7.30801225e-01 -1.44227035e-03 -1.82993263e-01 1.03830767e+00 6.90720916e-01 -8.48777816e-02 5.40272117e-01 -4.13832814e-01 -1.22003627e+00 -5.89371443e-01 2.27057621e-01 4.06755507e-01 9.84295458e-02 -2.91427493e-01 6.14854932e-01 -5.27433716e-02 -1.48145568e+00 1.15421450e+00 6.93222761e-01 4.90932733e-01 6.50088251e-01 1.09051228e+00 -6.57203734e-01 9.50573146e-01 -1.00941026e+00 -3.89290959e-01 -1.49032816e-01 -1.06891835e+00 -7.46872008e-01 1.20322101e-01 7.02527225e-01 -1.64507896e-01 -2.45953903e-01 9.78455126e-01 6.53135777e-01 -2.90252358e-01 9.28933859e-01 -1.19529116e+00 -1.71992421e+00 7.17392206e-01 -5.22855341e-01 7.21260831e-02 3.82660151e-01 -2.94598062e-02 6.41632617e-01 -7.36201704e-01 8.47005844e-01 1.27584624e+00 3.70586246e-01 9.74611461e-01 -9.82750118e-01 -3.21723700e-01 -7.73340836e-02 2.30705708e-01 -1.15148854e+00 -1.44233435e-01 4.54237550e-01 -5.30709803e-01 1.26259696e+00 3.25872123e-01 3.65994155e-01 1.47763562e+00 4.48191643e-01 6.33469760e-01 6.43365741e-01 -6.19921744e-01 -1.16758374e-03 3.38650733e-01 -1.79766774e-01 4.42109048e-01 3.61706197e-01 -1.73147529e-01 -4.33667123e-01 1.40298188e-01 6.61377788e-01 -3.99210453e-01 -1.51982248e-01 -1.99768931e-01 -1.45688462e+00 9.62736070e-01 5.74759483e-01 3.88015151e-01 -3.02431345e-01 2.62783825e-01 6.26939356e-01 9.26413983e-02 4.45957482e-01 6.21721685e-01 -6.92909434e-02 1.71977073e-01 -1.06355941e+00 1.37063041e-01 4.34533268e-01 1.33706343e+00 4.62915599e-01 -1.90779399e-02 -8.26758146e-01 8.20197463e-01 4.24410939e-01 5.82624257e-01 7.96893001e-01 -6.38094962e-01 8.50097954e-01 3.14886898e-01 8.51865299e-03 -9.43443418e-01 -1.09030847e-02 -3.46546322e-01 -1.00229442e+00 6.64745690e-03 -9.26618427e-02 -8.36408958e-02 -1.48221695e+00 1.64335155e+00 -5.33101141e-01 -7.16629922e-02 6.22430980e-01 8.98775280e-01 1.33176184e+00 1.40830255e+00 1.84926555e-01 1.24807410e-01 1.40459299e+00 -1.43622053e+00 -9.09246147e-01 -3.35531175e-01 5.96329033e-01 -9.85329449e-01 8.78207266e-01 -7.63596967e-02 -1.00820005e+00 -8.70429933e-01 -1.25991106e+00 -2.26138517e-01 -5.89938760e-01 1.96519315e-01 -4.62227985e-02 4.23284173e-01 -1.42061293e+00 8.76319855e-02 -1.82900146e-01 -7.41149664e-01 4.75412868e-02 1.79468736e-01 -5.44241071e-01 6.31051287e-02 -1.22816145e+00 1.00401497e+00 1.04216659e+00 -1.10581435e-01 -8.74649107e-01 -2.72856653e-01 -9.90507841e-01 1.04867451e-01 -4.15594518e-01 -9.27454054e-01 1.19432831e+00 -1.47605455e+00 -1.15100789e+00 9.22173738e-01 -8.64473283e-02 -6.74935520e-01 3.91647995e-01 2.18922123e-01 -3.83033544e-01 2.62390584e-01 1.80204198e-01 1.69987631e+00 6.89777732e-01 -1.60766792e+00 -4.72677767e-01 1.43335342e-01 -2.54160613e-01 4.22657669e-01 -3.96955237e-02 -1.66600257e-01 -3.45127463e-01 -7.38687754e-01 -3.96452159e-01 -8.86831045e-01 2.73169260e-02 -2.29244739e-01 -5.42609870e-01 2.84384519e-01 9.62732971e-01 -1.02508843e+00 9.91476476e-01 -2.04751039e+00 6.75421581e-02 -3.82483840e-01 -3.14570785e-01 5.62599599e-01 -6.16416872e-01 6.45944774e-01 -3.64394128e-01 5.38523197e-01 -5.16578972e-01 -3.28919888e-01 -2.34805226e-01 2.36958161e-01 -2.54680455e-01 -1.83264744e-02 5.54220378e-01 1.03391421e+00 -7.35918462e-01 -6.59591317e-01 1.09551594e-01 6.48911178e-01 -2.68119335e-01 4.92657453e-01 -5.26531339e-01 2.45781168e-01 -5.04913852e-02 1.82511345e-01 7.67309427e-01 -2.16549054e-01 -3.08058649e-01 -1.27813190e-01 -7.28039742e-02 -6.97320476e-02 -4.86983985e-01 2.05717754e+00 -1.80189610e-01 1.34859836e+00 -5.98055542e-01 -7.40715861e-01 1.15572858e+00 5.97157300e-01 -6.31261021e-02 -8.48165691e-01 -1.18351489e-01 2.13243067e-01 5.94425537e-02 -9.11575854e-01 8.75922978e-01 1.19507082e-01 -1.41042992e-02 8.34042504e-02 4.27993760e-02 -8.74264464e-02 3.31259370e-01 1.09917201e-01 6.52703762e-01 3.30270410e-01 6.44796491e-02 -1.04499333e-01 6.83827162e-01 4.40385401e-01 -8.08768719e-02 6.72888696e-01 -3.35077047e-02 1.47951186e+00 5.02833843e-01 -3.39685589e-01 -1.93446207e+00 -9.61899698e-01 1.40742958e-01 5.80258727e-01 3.51697989e-02 2.74775773e-02 -1.05095887e+00 -4.37698603e-01 -4.43811357e-01 1.03553748e+00 -7.88530290e-01 -3.46138537e-01 -4.85152483e-01 -5.66374421e-01 8.28104913e-01 4.18500394e-01 7.55070448e-01 -1.40809524e+00 -2.52198189e-01 2.35442430e-01 -5.01463294e-01 -1.12043095e+00 -5.57093322e-01 -1.51227564e-02 -5.99249125e-01 -3.87678444e-01 -1.54887748e+00 -1.45007420e+00 1.02447331e+00 1.11494035e-01 1.19533432e+00 -1.04218297e-01 -1.84269130e-01 -1.98051870e-01 -5.52257180e-01 -5.24269283e-01 -9.45631266e-01 1.99300811e-01 -6.01094484e-01 -2.96606958e-01 2.31312066e-01 1.51286781e-01 -5.08024454e-01 -2.66807489e-02 -1.27855253e+00 7.04754233e-01 9.99875009e-01 8.25730562e-01 4.43735719e-01 -6.62779212e-01 6.06288493e-01 -5.97050309e-01 8.79044294e-01 -5.47351956e-01 -2.18304589e-01 3.92118007e-01 -3.02602500e-01 4.61105436e-01 7.53023505e-01 -4.55032349e-01 -1.00607312e+00 3.23689729e-02 -4.68162857e-02 -2.11842284e-01 -1.75689086e-01 3.31685781e-01 1.02663517e-01 2.41756454e-01 7.77279317e-01 6.61321342e-01 2.18687579e-02 -2.61132449e-01 3.99439335e-01 1.09406149e+00 6.16284966e-01 3.40630002e-02 4.20230478e-01 4.82140407e-02 -3.06029558e-01 -6.06095433e-01 -5.66610992e-01 -2.37040609e-01 -4.37561274e-01 -1.66973114e-01 1.50390184e+00 -1.11030889e+00 -2.32189931e-02 2.66117573e-01 -1.79603767e+00 7.56953657e-02 -1.33031365e-02 4.27831471e-01 -6.75375342e-01 3.60839188e-01 -4.69036132e-01 -3.76386136e-01 -5.89094341e-01 -1.40064573e+00 1.29535699e+00 3.96678448e-01 -1.01442046e-01 -8.21224868e-01 2.96481252e-01 4.18842703e-01 4.92800206e-01 5.00456393e-01 8.78940821e-01 -6.68563545e-01 -3.40631932e-01 1.61813527e-01 -5.63860238e-01 4.84560370e-01 1.21172994e-01 -4.89400816e-04 -8.07562113e-01 -5.74332960e-02 -4.10794586e-01 -5.12838662e-02 9.15713191e-01 3.47757280e-01 8.26290905e-01 -4.60858911e-01 -7.36162513e-02 3.86130720e-01 1.62722313e+00 4.31277871e-01 1.14832866e+00 6.48255765e-01 6.92215025e-01 5.82021654e-01 3.79996002e-01 -2.01630518e-02 1.42645970e-01 5.03662527e-01 4.75653470e-01 -4.46257085e-01 -5.41272044e-01 -5.12763798e-01 5.50403059e-01 8.31530392e-01 3.69953185e-01 -1.08543229e+00 -7.52451897e-01 7.47538745e-01 -2.07164145e+00 -1.12828052e+00 -4.44233239e-01 1.78594089e+00 6.64013386e-01 -5.61845675e-02 -1.50910720e-01 -4.51512367e-01 1.22997761e+00 8.17779750e-02 -4.79017317e-01 -1.37787533e+00 -4.91048187e-01 -5.48793077e-01 8.81676018e-01 3.77958834e-01 -8.41636062e-01 8.96677971e-01 6.96055126e+00 4.51570213e-01 -1.09811962e+00 -7.09273964e-02 8.17789078e-01 1.50363818e-01 -3.78041625e-01 -2.62212187e-01 -6.81757510e-01 9.39451039e-01 1.26711714e+00 -1.32938564e-01 1.67371649e-02 7.48172343e-01 2.42123142e-01 -8.63942876e-02 -9.59421456e-01 1.12445712e+00 9.03841138e-01 -1.41717970e+00 8.18434715e-01 -9.54606608e-02 9.51575696e-01 -6.52410313e-02 1.72005892e-01 5.44352010e-02 -2.51534969e-01 -1.28808653e+00 8.49337161e-01 6.23579383e-01 6.52861536e-01 -6.95475578e-01 1.17415714e+00 8.22623596e-02 -6.68055892e-01 3.42522591e-01 -7.39105701e-01 1.39385924e-01 2.72096157e-01 1.71576440e-01 -1.28414822e+00 4.98238295e-01 3.81890059e-01 3.44560534e-01 -8.17570269e-01 1.18162775e+00 -8.22203681e-02 2.13880554e-01 2.60217160e-01 -3.69958788e-01 7.65817821e-01 -8.57955683e-03 5.06561935e-01 1.64880693e+00 7.10068166e-01 -5.07684112e-01 -3.25133920e-01 7.94974089e-01 -1.65845558e-01 2.68104196e-01 -1.02303648e+00 -2.83863723e-01 8.66988581e-03 8.00885558e-01 -5.18602908e-01 -7.11488903e-01 -3.42283428e-01 1.49755275e+00 -4.02742485e-03 3.34940195e-01 -1.05484235e+00 -6.00076556e-01 2.92070538e-01 -6.37424067e-02 4.84886944e-01 -9.06206146e-02 -1.65264145e-01 -1.00632644e+00 8.16323683e-02 -9.00578737e-01 1.02983974e-02 -1.56912279e+00 -1.05548954e+00 1.03841770e+00 1.49481118e-01 -1.31626046e+00 -5.26141107e-01 -5.56765616e-01 -4.96754229e-01 1.00969851e+00 -1.35956311e+00 -1.31026840e+00 -3.21086854e-01 9.63525549e-02 1.05471087e+00 -5.01464784e-01 6.44469261e-01 2.52783924e-01 -2.93301731e-01 3.78406018e-01 3.36372733e-01 1.27473846e-01 8.58767152e-01 -1.10272765e+00 7.62475491e-01 8.67880762e-01 5.82965687e-02 2.13929459e-01 1.17003012e+00 -7.36449003e-01 -8.62227499e-01 -1.31200123e+00 1.21764600e+00 -2.00836420e-01 1.25170410e-01 -4.65614200e-01 -9.00215328e-01 5.56418121e-01 1.09951496e+00 -5.92033982e-01 6.84497237e-01 -8.40310693e-01 -2.83435881e-01 3.64447981e-01 -1.19346678e+00 8.10072005e-01 7.13633239e-01 -2.45782301e-01 -8.62734199e-01 4.10816401e-01 1.08152032e+00 -2.37782568e-01 -4.89002705e-01 -8.13154969e-03 3.71462464e-01 -7.88306117e-01 8.30458045e-01 -5.15022814e-01 9.60248291e-01 -6.08975351e-01 -6.98037222e-02 -1.34037530e+00 -5.11484325e-01 -2.10544407e-01 3.64166737e-01 1.22647810e+00 9.36556458e-01 -1.64771408e-01 5.62421024e-01 1.53870553e-01 -3.44423741e-01 -1.68566272e-01 -6.42443955e-01 -8.20511103e-01 7.12756217e-02 3.67397964e-01 5.66157937e-01 6.63570285e-01 -2.26606011e-01 6.88678324e-01 -7.15305686e-01 -2.49718465e-02 3.58388335e-01 -2.60830969e-01 4.08867896e-01 -5.61324060e-01 4.79244851e-02 -2.21257433e-01 -5.26485324e-01 -8.05516958e-01 8.05056542e-02 -9.86241281e-01 4.29097682e-01 -2.29775620e+00 6.03711724e-01 4.69171017e-01 1.93384737e-02 3.39751393e-01 -8.67223963e-02 4.64288384e-01 3.50585610e-01 2.28366718e-01 -5.06299317e-01 4.81204957e-01 1.35541809e+00 -5.35336733e-01 6.43698797e-02 -8.16748738e-01 -6.51310205e-01 2.23594084e-01 1.10667312e+00 -6.03272200e-01 -5.85629225e-01 -1.04016447e+00 2.23005041e-01 -1.27671331e-01 7.28455111e-02 -1.15661895e+00 6.65408000e-02 8.37547034e-02 6.61955714e-01 -8.44892740e-01 2.81571478e-01 -5.61586797e-01 3.78117889e-01 6.62418246e-01 -8.60118389e-01 5.24804413e-01 3.35943669e-01 3.01870197e-01 -4.69045162e-01 -6.13235831e-01 7.49826610e-01 -3.90906572e-01 -8.27600777e-01 1.14683120e-04 -6.09297991e-01 -2.49775738e-01 1.13059914e+00 -3.70563895e-01 -6.81405306e-01 -5.36078453e-01 -4.23720270e-01 7.44086877e-02 6.77269518e-01 9.73423898e-01 7.23061383e-01 -1.50379002e+00 -1.15554368e+00 -2.99239159e-02 4.35635358e-01 -3.36938113e-01 4.81674969e-02 1.12013169e-01 -1.25359929e+00 9.16422665e-01 -8.99451673e-01 -6.76568151e-01 -1.27407324e+00 8.00972283e-01 1.83733404e-01 -3.93781476e-02 -2.22897410e-01 6.29953861e-01 3.97809625e-01 -5.49958907e-02 4.60013337e-02 -3.11482430e-01 -5.25929391e-01 -4.36822511e-02 7.21598327e-01 -1.55363292e-01 -3.72405313e-02 -9.46282268e-01 -1.75347522e-01 5.61225235e-01 -1.85159147e-01 -3.87822300e-01 9.42111135e-01 -2.74449170e-01 -9.19607561e-03 2.95671701e-01 1.72356534e+00 -6.20841622e-01 -8.21416676e-01 3.98656547e-01 -2.39187125e-02 -1.70613825e-01 -2.72813022e-01 -8.77037764e-01 -7.06982255e-01 9.73250210e-01 9.04192269e-01 1.97382927e-01 9.72356975e-01 -1.49884075e-01 8.62008870e-01 3.30621570e-01 -2.22389221e-01 -9.86726224e-01 1.16796978e-01 5.70907593e-01 1.24990475e+00 -1.43479311e+00 -3.08196247e-01 4.05922383e-02 -9.80904043e-01 1.20239496e+00 7.28388548e-01 3.69804017e-02 -1.63517207e-01 6.80995313e-03 3.22248697e-01 1.24823898e-01 -9.25579727e-01 -6.24595210e-02 3.02376032e-01 9.18578625e-01 6.82920575e-01 -2.27818683e-01 -6.05554402e-01 -8.94109998e-03 -2.15515941e-01 -7.08791539e-02 8.62587273e-01 6.89760566e-01 -6.59309089e-01 -1.05310571e+00 -6.06135428e-01 1.19258516e-01 -4.86819327e-01 -3.91775310e-01 -7.10113823e-01 5.36832929e-01 9.47849229e-02 9.33495760e-01 4.42017585e-01 -2.00839311e-01 9.26078185e-02 -3.99799533e-02 2.31993496e-01 -6.94958866e-01 -6.13923490e-01 -2.32286260e-01 2.53575414e-01 -2.73514148e-02 -4.21456426e-01 -1.72640145e-01 -1.11171472e+00 6.83176098e-03 7.91091025e-02 3.38965207e-01 1.20312035e+00 5.23535848e-01 6.87861502e-01 5.89352787e-01 3.96335572e-01 -5.38018525e-01 7.76434764e-02 -1.23168886e+00 -7.69573916e-03 4.99616832e-01 3.11530381e-01 3.20299827e-02 -1.63530380e-01 6.39853120e-01]
[11.027884483337402, 1.0619678497314453]
4bf73653-dbcd-44d3-b613-cd1463cbc0c6
graph-convolutional-networks-based-word
1809.04283
null
https://arxiv.org/abs/1809.04283v4
https://arxiv.org/pdf/1809.04283v4.pdf
Incorporating Syntactic and Semantic Information in Word Embeddings using Graph Convolutional Networks
Word embeddings have been widely adopted across several NLP applications. Most existing word embedding methods utilize sequential context of a word to learn its embedding. While there have been some attempts at utilizing syntactic context of a word, such methods result in an explosion of the vocabulary size. In this paper, we overcome this problem by proposing SynGCN, a flexible Graph Convolution based method for learning word embeddings. SynGCN utilizes the dependency context of a word without increasing the vocabulary size. Word embeddings learned by SynGCN outperform existing methods on various intrinsic and extrinsic tasks and provide an advantage when used with ELMo. We also propose SemGCN, an effective framework for incorporating diverse semantic knowledge for further enhancing learned word representations. We make the source code of both models available to encourage reproducible research.
['Shikhar Vashishth', 'Partha Talukdar', 'Manik Bhandari', 'Chiranjib Bhattacharyya', 'Prateek Yadav', 'Piyush Rai']
2018-09-12
incorporating-syntactic-and-semantic
https://aclanthology.org/P19-1320
https://aclanthology.org/P19-1320.pdf
acl-2019-7
['learning-word-embeddings']
['methodology']
[-2.62618452e-01 -1.36257052e-01 -5.74223459e-01 -3.44743937e-01 -1.88246638e-01 -3.56026381e-01 4.96609747e-01 2.60591567e-01 -8.87018204e-01 4.95327622e-01 6.12262368e-01 -4.88168150e-01 6.15680665e-02 -1.00852716e+00 -1.61637470e-01 -3.99852842e-01 1.59369126e-01 5.52238198e-03 1.12882473e-01 -3.25825691e-01 2.11885765e-01 2.14754894e-01 -1.01574218e+00 1.08441459e-02 7.07632959e-01 5.26225805e-01 3.73845607e-01 3.07704836e-01 -7.42250979e-01 9.58119035e-02 -4.50746179e-01 -4.26191539e-01 7.95650184e-02 -1.05970964e-01 -6.27657652e-01 -3.63914400e-01 8.60167965e-02 -4.76634037e-03 -7.09413052e-01 9.79466498e-01 5.70444345e-01 4.76982504e-01 5.47095716e-01 -1.15492654e+00 -1.44214559e+00 7.29688287e-01 -5.41210055e-01 4.48290944e-01 7.38923922e-02 -8.05395320e-02 1.59182525e+00 -1.13178170e+00 4.50752169e-01 1.32771993e+00 5.01224637e-01 5.90652287e-01 -1.09672880e+00 -7.90985882e-01 4.55238909e-01 2.84093797e-01 -1.35595238e+00 9.59862545e-02 8.21037948e-01 -1.03782870e-01 1.53165162e+00 -1.79418147e-01 6.73025608e-01 1.25428915e+00 8.92443731e-02 5.56564152e-01 6.96505666e-01 -6.44422591e-01 6.01756983e-02 2.17454787e-02 6.79632425e-01 6.37623250e-01 6.36073172e-01 2.79864091e-02 -3.98312002e-01 -1.80142596e-01 6.06068671e-01 4.16771472e-01 -7.73890540e-02 -2.26127654e-01 -9.15838480e-01 1.29916489e+00 6.15774989e-01 5.27947128e-01 -7.95131847e-02 6.46689892e-01 5.34441948e-01 3.36048275e-01 7.55091786e-01 5.87624133e-01 -5.23218930e-01 -1.11083746e-01 -3.97826821e-01 2.74212454e-02 5.80507874e-01 8.83223593e-01 9.39041317e-01 8.97589922e-02 -1.19493559e-01 1.09574556e+00 5.60809731e-01 2.42944211e-01 9.82787311e-01 -2.60465324e-01 3.37193489e-01 9.14309919e-01 -3.03014159e-01 -1.22489083e+00 -4.06143963e-01 -2.14461997e-01 -2.96753407e-01 -1.64140120e-01 -7.91206062e-02 -2.35735252e-01 -1.17860055e+00 1.75789630e+00 1.37612134e-01 3.96586806e-01 1.56218350e-01 7.66792357e-01 9.83869553e-01 8.88613343e-01 3.90862197e-01 3.32423449e-01 1.51727927e+00 -1.20107758e+00 -8.74070883e-01 -5.23199141e-01 9.41552579e-01 -5.91811955e-01 1.30603707e+00 -4.41211946e-02 -3.86926085e-01 -3.95178288e-01 -1.01774323e+00 -3.70366842e-01 -9.57320213e-01 -1.99173138e-01 9.91418362e-01 7.25477397e-01 -9.71205413e-01 2.20408842e-01 -7.73878574e-01 -6.16064906e-01 3.76254082e-01 2.61767119e-01 -5.17049015e-01 -4.05430615e-01 -1.52534914e+00 9.89575267e-01 7.43286133e-01 -2.00441703e-01 -3.77091706e-01 -6.31346703e-01 -1.33337712e+00 2.62955874e-01 2.80391544e-01 -5.70172548e-01 8.16575468e-01 -5.87865770e-01 -9.68620896e-01 4.81199235e-01 -2.66195059e-01 -3.40007216e-01 -2.99025029e-01 -2.72615492e-01 -7.47110546e-01 -1.10014491e-01 -1.11454397e-01 6.97584987e-01 7.89146304e-01 -7.76051581e-01 -3.11350584e-01 -2.24715218e-01 4.23984468e-01 1.03242896e-01 -1.17408621e+00 -4.89839241e-02 -5.27932703e-01 -1.01055622e+00 -1.06552295e-01 -7.66112447e-01 -4.59474117e-01 7.45427310e-02 -2.33356990e-02 -7.40199983e-01 9.74144459e-01 -3.93137574e-01 1.58016169e+00 -2.22982693e+00 -4.88996580e-02 -8.26627389e-02 3.01049232e-01 7.17970610e-01 -7.68437386e-01 8.29855800e-01 -4.81491946e-02 4.60221916e-01 -1.13595024e-01 -3.47645819e-01 1.34215176e-01 6.32283807e-01 -3.04840088e-01 1.85965165e-01 3.77444774e-01 1.21760690e+00 -1.06302392e+00 -3.11220467e-01 2.40930557e-01 7.86998093e-01 -7.40951657e-01 5.12396581e-02 -2.03122154e-01 -3.33152831e-01 -6.74036026e-01 2.44111702e-01 4.72154409e-01 -3.19994539e-01 3.64580512e-01 -1.16611950e-01 2.76220143e-01 4.62891787e-01 -9.49654102e-01 1.84000289e+00 -9.28498447e-01 7.42747903e-01 -5.73538780e-01 -1.02920246e+00 9.91838574e-01 3.87730360e-01 1.63176134e-01 -4.98413652e-01 5.49931824e-02 1.88744366e-02 5.89101464e-02 -5.72392344e-01 8.17583203e-01 -2.95937508e-01 -3.59634832e-02 5.88000596e-01 3.30737233e-01 1.37331054e-01 2.59930223e-01 3.96996617e-01 1.10179770e+00 -3.14421877e-02 5.13538837e-01 -1.66235253e-01 3.06051344e-01 -1.86045229e-01 4.08763170e-01 4.66803640e-01 -1.42260432e-01 3.92077893e-01 3.14812541e-01 -4.91288006e-01 -8.82887065e-01 -8.76390755e-01 -5.37187699e-03 1.21757293e+00 9.54219475e-02 -9.85395908e-01 -2.76809484e-01 -8.06547999e-01 2.52636254e-01 8.51333797e-01 -7.76041448e-01 -3.36087137e-01 -4.82247353e-01 -7.10617661e-01 4.55388993e-01 9.62235093e-01 1.30902648e-01 -1.10400772e+00 -1.15059756e-01 2.93251634e-01 1.58898190e-01 -1.09826934e+00 -7.96573758e-01 1.43077180e-01 -8.55166554e-01 -9.13799524e-01 -5.08733988e-01 -9.62312639e-01 5.85849404e-01 6.18708074e-01 1.04221439e+00 3.70566428e-01 -5.45343161e-01 4.86538231e-01 -8.70472610e-01 -3.71513426e-01 1.05637021e-01 3.83133531e-01 1.05418690e-01 -3.72586697e-01 1.12641656e+00 -5.50989568e-01 -5.96594810e-01 -5.40176220e-02 -1.15668273e+00 -3.60668331e-01 3.17559719e-01 9.54487741e-01 2.39030808e-01 -3.59926879e-01 8.23569775e-01 -9.63838875e-01 1.00823700e+00 -6.91990614e-01 -2.67452717e-01 3.93568911e-02 -1.05667400e+00 2.64020473e-01 5.81707895e-01 -4.93899584e-01 -6.80332601e-01 -5.14080763e-01 -2.69918233e-01 -3.38822871e-01 1.04487263e-01 9.16533530e-01 2.15572521e-01 1.20421179e-01 3.39192271e-01 1.04865544e-01 -1.37092441e-01 -5.54773450e-01 8.11118782e-01 6.57138407e-01 -2.50965685e-01 -4.51409847e-01 7.45128572e-01 2.42094189e-01 -3.67056489e-01 -9.43980336e-01 -8.53967607e-01 -8.22838902e-01 -4.23316091e-01 3.77718538e-01 9.40876126e-01 -9.90123212e-01 -2.62028426e-01 -1.63514420e-01 -1.24352765e+00 9.67732593e-02 -1.26758799e-01 6.86886966e-01 8.87869447e-02 4.65009481e-01 -4.68952030e-01 -5.30927718e-01 -2.92286515e-01 -9.43702817e-01 7.82977700e-01 1.83428049e-01 -3.55005890e-01 -1.75712931e+00 3.26308370e-01 1.58512279e-01 6.92612350e-01 -1.84639171e-01 9.67396915e-01 -9.81683075e-01 -6.93927631e-02 -3.70910347e-01 -4.25655514e-01 4.80509877e-01 4.61497039e-01 -2.28166997e-01 -8.08899820e-01 -4.77085680e-01 -4.94100690e-01 -2.53538191e-01 1.11236942e+00 1.02624364e-01 1.08444393e+00 -2.04274163e-01 -5.69503129e-01 6.08678818e-01 1.82858121e+00 -1.05195714e-03 4.73961741e-01 3.60379726e-01 9.69565690e-01 2.81660289e-01 2.97704875e-01 2.21837491e-01 3.78184021e-01 5.75538933e-01 2.34035149e-01 -7.13737682e-02 -2.53076553e-01 -3.54780734e-01 3.65167558e-01 1.15098011e+00 1.16628058e-01 -5.49402058e-01 -9.80173826e-01 8.19820762e-01 -1.68866003e+00 -5.50748110e-01 1.76463407e-02 1.55892432e+00 6.40719235e-01 -1.31597564e-01 -2.83945620e-01 -1.92015558e-01 5.40210783e-01 7.09090471e-01 -2.80624688e-01 -7.83925593e-01 2.03334112e-02 7.60441959e-01 5.36394298e-01 5.98465145e-01 -7.55927145e-01 1.39704120e+00 6.58549929e+00 8.14540207e-01 -1.12020445e+00 3.75509232e-01 6.90407604e-02 -1.03195235e-01 -7.99933910e-01 1.49846360e-01 -7.88550198e-01 4.62543726e-01 8.70689094e-01 -4.14625108e-01 1.98148951e-01 8.74642074e-01 1.21070025e-02 2.87478447e-01 -9.13431466e-01 9.66368556e-01 2.78447628e-01 -1.29630446e+00 3.84809464e-01 -3.22883539e-02 5.69190025e-01 1.97973296e-01 -2.03685686e-02 5.71995139e-01 4.25831288e-01 -1.34216630e+00 -9.45457444e-02 3.96326184e-02 7.08024025e-01 -7.72965372e-01 7.72138715e-01 -5.98091483e-02 -1.37317824e+00 4.33858372e-02 -7.92155504e-01 -3.15776080e-01 2.12335810e-01 5.84403932e-01 -7.43008316e-01 5.53365886e-01 3.57523829e-01 9.77485836e-01 -6.65741324e-01 5.84088385e-01 -6.17795408e-01 8.93199503e-01 -3.50723863e-02 -4.57016438e-01 6.89485192e-01 -3.15965831e-01 3.91798437e-01 1.37359083e+00 2.56790102e-01 -4.24781367e-02 2.70535022e-01 7.36007988e-01 -2.89099008e-01 3.78758401e-01 -8.67502749e-01 -6.69818878e-01 5.78219473e-01 1.23264492e+00 -5.27885616e-01 -1.02638163e-01 -9.96691763e-01 9.85073805e-01 6.05541110e-01 5.37126243e-01 -5.84855258e-01 -7.51125991e-01 1.24478424e+00 -1.91451997e-01 5.22021294e-01 -6.44021034e-01 -7.90688843e-02 -1.21575391e+00 -1.19691342e-01 -3.47165793e-01 4.88512844e-01 -4.90187138e-01 -1.52631330e+00 5.37447095e-01 -1.64806023e-01 -8.22172940e-01 6.40576333e-02 -1.12985969e+00 -6.86420798e-01 9.35599804e-01 -1.74303460e+00 -1.12906075e+00 1.82822142e-02 2.79473424e-01 7.12261558e-01 -3.22320193e-01 1.04230881e+00 3.79292697e-01 -4.89155591e-01 7.24391878e-01 4.09309305e-02 3.49887311e-01 7.83514619e-01 -1.16807067e+00 7.80736804e-01 7.42834568e-01 6.44035339e-01 1.05196953e+00 4.38004076e-01 -5.57010531e-01 -1.45725632e+00 -1.18231452e+00 1.09525752e+00 -4.54234242e-01 9.83224988e-01 -4.75478411e-01 -9.66420054e-01 8.87265384e-01 4.51386511e-01 2.77855456e-01 1.11325598e+00 5.16988158e-01 -7.08781719e-01 8.27132240e-02 -7.04672217e-01 7.40508318e-01 1.12666380e+00 -6.31032228e-01 -9.24943924e-01 1.59292921e-01 1.43049037e+00 1.51969939e-01 -7.50631511e-01 4.15638685e-02 3.18715990e-01 -3.63782465e-01 9.17529941e-01 -9.58865404e-01 4.75488544e-01 -1.82987288e-01 -2.73292720e-01 -1.64676595e+00 -4.03775305e-01 -1.68904364e-01 -2.55518973e-01 1.04141772e+00 5.38558304e-01 -1.09808934e+00 6.56049907e-01 3.39524716e-01 -4.85502332e-02 -9.11336184e-01 -7.52523780e-01 -9.22933638e-01 4.79077220e-01 -6.47499263e-01 6.26390100e-01 1.15553188e+00 2.81655312e-01 4.53182757e-01 -3.03412378e-01 -5.96979223e-02 2.07392216e-01 -5.75981736e-02 3.65520924e-01 -9.55405891e-01 -6.91911578e-02 -2.89006114e-01 -7.45178103e-01 -1.03941226e+00 5.15575171e-01 -1.47953582e+00 -3.49392235e-01 -1.77155232e+00 2.18107194e-01 -3.22746456e-01 -7.75189221e-01 7.34039307e-01 -4.96581852e-01 3.91458064e-01 1.03500605e-01 -1.82149142e-01 -3.80977809e-01 7.94990480e-01 1.08092976e+00 -1.61546260e-01 1.78594664e-01 -6.22351944e-01 -9.26703930e-01 3.69727850e-01 9.34072971e-01 -5.98768055e-01 -7.33815491e-01 -8.44948471e-01 4.28169489e-01 -7.07404017e-01 1.01316027e-01 -5.42558908e-01 5.64024709e-02 -7.96731263e-02 6.26926869e-02 -1.99480772e-01 3.39156747e-01 -8.27357054e-01 -4.04319048e-01 3.74497980e-01 -2.68313587e-01 3.40841562e-01 2.80204445e-01 8.35208356e-01 -3.47254783e-01 -4.75114942e-01 4.03235614e-01 -1.46189823e-05 -1.03271043e+00 5.85243225e-01 -3.03227484e-01 1.57426670e-01 8.85029495e-01 -2.56545454e-01 -1.72049344e-01 -1.68264195e-01 -3.85550737e-01 2.63991237e-01 3.59573066e-01 9.96045530e-01 8.59192967e-01 -1.51923549e+00 -4.78078336e-01 1.48589164e-01 4.81049985e-01 -3.97820473e-01 -1.42274778e-02 2.91856825e-01 -4.28606361e-01 5.11041284e-01 9.36860889e-02 -9.36362427e-03 -1.09735358e+00 7.96465874e-01 -9.73130837e-02 -1.53203845e-01 -8.49566102e-01 8.95925283e-01 2.67042220e-01 -5.68452179e-01 2.05033645e-03 -3.11283857e-01 -4.24435228e-01 1.12767592e-01 5.44844925e-01 1.15975223e-01 -2.11388454e-01 -3.84346068e-01 -3.83038402e-01 4.95710969e-01 -3.68424386e-01 5.79982400e-02 1.56305921e+00 -8.97847638e-02 -1.69278294e-01 5.40142179e-01 1.55731082e+00 -6.22258596e-02 -7.15605795e-01 -5.05069196e-01 2.39802271e-01 -6.29573703e-01 2.30429783e-01 -4.36499834e-01 -1.10727608e+00 1.01994824e+00 5.40067136e-01 -2.88972072e-03 6.79184616e-01 -8.30894522e-03 1.14460301e+00 4.08857137e-01 3.69929016e-01 -1.03884566e+00 1.93666235e-01 8.30014408e-01 5.43301642e-01 -1.31528795e+00 6.65834993e-02 -3.76101613e-01 -5.22777438e-01 1.26811004e+00 6.97957277e-01 -3.49036694e-01 9.57803071e-01 9.70749557e-02 1.98448867e-01 -3.37365597e-01 -8.31625044e-01 -3.61605108e-01 3.64237964e-01 7.12469459e-01 8.19318831e-01 7.62691945e-02 -8.79376948e-01 6.69842303e-01 2.10135475e-01 -2.48931453e-01 3.41352195e-01 8.73584628e-01 -3.94875050e-01 -1.69979739e+00 9.55118239e-02 4.82505798e-01 -3.58657688e-01 -6.94533169e-01 -1.74809828e-01 6.92210317e-01 -4.15893309e-02 7.87813902e-01 7.62408823e-02 -2.99825251e-01 7.56196529e-02 2.73061812e-01 2.27086663e-01 -1.27851605e+00 -4.20075148e-01 -3.93349856e-01 1.11975782e-01 -3.45431030e-01 -2.25411162e-01 -1.65861860e-01 -1.30770493e+00 -8.32946450e-02 -4.91865367e-01 1.57382071e-01 6.58950210e-01 9.75948453e-01 5.01024961e-01 6.01687431e-01 3.29417109e-01 -3.92301500e-01 -4.12319273e-01 -1.14013982e+00 -6.38357937e-01 4.11914587e-01 1.13306632e-02 -8.04781973e-01 -3.22509915e-01 -2.99830824e-01]
[10.496020317077637, 8.636876106262207]
bd52b4a6-d298-4977-99b1-381238b7e9c5
optimizing-video-prediction-via-video-frame-1
2206.13454
null
https://arxiv.org/abs/2206.13454v1
https://arxiv.org/pdf/2206.13454v1.pdf
Optimizing Video Prediction via Video Frame Interpolation
Video prediction is an extrapolation task that predicts future frames given past frames, and video frame interpolation is an interpolation task that estimates intermediate frames between two frames. We have witnessed the tremendous advancement of video frame interpolation, but the general video prediction in the wild is still an open question. Inspired by the photo-realistic results of video frame interpolation, we present a new optimization framework for video prediction via video frame interpolation, in which we solve an extrapolation problem based on an interpolation model. Our video prediction framework is based on optimization with a pretrained differentiable video frame interpolation module without the need for a training dataset, and thus there is no domain gap issue between training and test data. Also, our approach does not need any additional information such as semantic or instance maps, which makes our framework applicable to any video. Extensive experiments on the Cityscapes, KITTI, DAVIS, Middlebury, and Vimeo90K datasets show that our video prediction results are robust in general scenarios, and our approach outperforms other video prediction methods that require a large amount of training data or extra semantic information.
['Qifeng Chen', 'Qiang Wen', 'Yue Wu']
2022-06-27
optimizing-video-prediction-via-video-frame
http://openaccess.thecvf.com//content/CVPR2022/html/Wu_Optimizing_Video_Prediction_via_Video_Frame_Interpolation_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Wu_Optimizing_Video_Prediction_via_Video_Frame_Interpolation_CVPR_2022_paper.pdf
cvpr-2022-1
['video-prediction']
['computer-vision']
[ 1.51741549e-01 -3.84832285e-02 -3.21153134e-01 -3.66984218e-01 -5.47380567e-01 -7.43588358e-02 4.27238077e-01 -4.05166209e-01 -2.05697969e-01 7.77307987e-01 1.78046316e-01 -1.75096631e-01 4.02744621e-01 -5.73496699e-01 -1.22604620e+00 -5.48656046e-01 2.58758874e-03 -1.76508084e-01 6.70332134e-01 -1.63655967e-01 1.12820849e-01 8.50081891e-02 -1.64400840e+00 5.72454214e-01 7.97853649e-01 1.33642876e+00 4.81908292e-01 7.62003243e-01 7.71817863e-02 1.38811481e+00 -1.29910663e-01 -4.81493175e-01 5.12927949e-01 -5.10152102e-01 -7.26996064e-01 3.97826552e-01 5.44475853e-01 -6.69463396e-01 -7.14554846e-01 8.25835407e-01 1.86701685e-01 3.22326779e-01 3.60228986e-01 -1.50354064e+00 -6.20211542e-01 1.12889431e-01 -3.73517156e-01 2.91142128e-02 6.95154905e-01 1.82814002e-02 5.53903461e-01 -1.09047437e+00 7.28043795e-01 1.02659118e+00 9.94419932e-01 5.35711527e-01 -1.03326631e+00 -4.83646125e-01 3.26147556e-01 7.57754683e-01 -1.45329034e+00 -4.64431405e-01 6.90302014e-01 -6.30288005e-01 6.11633956e-01 2.30507150e-01 9.98018026e-01 1.01324260e+00 -7.52496868e-02 1.10834146e+00 4.98021066e-01 -3.32132518e-01 1.96209207e-01 7.97952414e-02 -4.18219745e-01 5.48231184e-01 -4.11542326e-01 1.21307969e-01 -4.02910858e-01 9.39164162e-02 1.13579643e+00 2.92467594e-01 -6.42822564e-01 -3.48810822e-01 -1.35947406e+00 5.72191775e-01 2.41637826e-01 -1.32095084e-01 -3.31003159e-01 2.15416133e-01 4.43913192e-01 2.71712422e-01 7.06970155e-01 -2.69591480e-01 -3.25060159e-01 -5.55574857e-02 -1.32824838e+00 4.28914785e-01 6.25655890e-01 1.34512031e+00 1.02179480e+00 2.31212676e-01 -3.82070839e-02 5.47644258e-01 2.28177115e-01 2.96650231e-01 3.98741424e-01 -1.55409741e+00 7.03892589e-01 1.47620559e-01 5.66230655e-01 -1.17147160e+00 5.55613823e-02 3.32904994e-01 -7.36343086e-01 -1.15014920e-02 6.01398408e-01 -2.46376634e-01 -5.20789266e-01 1.39349568e+00 3.10953319e-01 1.14307046e+00 -2.00499687e-03 1.03644824e+00 7.58951187e-01 1.08326352e+00 -7.56277069e-02 -5.37062585e-01 7.98839509e-01 -1.35626614e+00 -5.98975301e-01 9.24964920e-02 5.77381551e-01 -6.46602809e-01 8.67671967e-01 2.33560309e-01 -1.15913332e+00 -9.56713021e-01 -7.99918771e-01 -3.45316917e-01 4.91386019e-02 -3.57146896e-02 4.85614091e-01 3.00767690e-01 -1.20281458e+00 6.40559316e-01 -6.64406061e-01 -2.88452983e-01 1.87323958e-01 1.77691951e-01 -4.36116695e-01 -1.44666895e-01 -1.13721550e+00 6.22186363e-01 7.10018814e-01 3.15526575e-01 -6.73762560e-01 -9.33798492e-01 -1.02838755e+00 -9.27514136e-02 3.30907851e-01 -6.94387496e-01 1.11272514e+00 -1.74002564e+00 -1.59769988e+00 3.81751418e-01 -5.12404978e-01 -6.14747107e-01 9.81937408e-01 -2.38398507e-01 -4.57914621e-01 2.10188374e-01 -8.82985368e-02 1.07378066e+00 9.36659694e-01 -1.13627732e+00 -9.00385439e-01 1.32161453e-01 3.87625843e-01 2.67089099e-01 -1.59201205e-01 1.14705665e-02 -7.55150139e-01 -8.56633842e-01 -2.68135130e-01 -8.43966126e-01 -4.06996191e-01 2.24797755e-01 3.78972143e-02 -7.65838549e-02 9.24319208e-01 -1.09531248e+00 1.24735212e+00 -2.12102938e+00 1.43289700e-01 -1.26821354e-01 -1.39725015e-01 1.45081714e-01 -5.37563004e-02 1.34416610e-01 -1.58547625e-01 -1.12836942e-01 -8.98169950e-02 -2.33859465e-01 -2.95260847e-01 2.22707912e-01 -5.04999697e-01 2.73458242e-01 6.75704330e-02 7.09795833e-01 -1.12238157e+00 -7.17106700e-01 4.72485036e-01 6.95877731e-01 -9.22154307e-01 3.57277572e-01 -4.66598302e-01 8.14788520e-01 -2.76744276e-01 5.08605361e-01 6.76282108e-01 -2.53193080e-01 -2.80726757e-02 -2.85501122e-01 -2.71606952e-01 -1.74506009e-01 -1.25122440e+00 1.96832871e+00 -3.12387258e-01 9.18762326e-01 -3.96035105e-01 -1.20071459e+00 7.59994030e-01 5.66077590e-01 7.12937474e-01 -3.78347158e-01 -2.57297784e-01 1.32006750e-01 -5.72182596e-01 -8.73611093e-01 6.99297667e-01 1.92986906e-01 4.39994693e-01 -6.43375143e-02 -1.56191617e-01 1.77662507e-01 3.91866654e-01 -5.98788187e-02 6.58004761e-01 7.89066732e-01 1.59344926e-01 -8.57742131e-02 7.23748386e-01 1.73207581e-01 9.62298751e-01 4.20290977e-01 -7.42041916e-02 1.02236331e+00 3.09440643e-01 -9.85484481e-01 -1.30247462e+00 -7.95258045e-01 -1.32829277e-02 9.10314560e-01 5.71529746e-01 -5.34574211e-01 -8.97311628e-01 -5.75045168e-01 -3.33795160e-01 4.87718552e-01 -3.99723709e-01 3.99598151e-01 -8.86179149e-01 -3.36488843e-01 3.05646639e-02 6.90640211e-01 6.32528841e-01 -8.68806839e-01 -2.57535726e-01 2.61679173e-01 -6.07391655e-01 -1.55677116e+00 -7.84856558e-01 -6.10239506e-01 -1.03382063e+00 -9.84758317e-01 -1.07980645e+00 -1.12947690e+00 6.62849188e-01 4.16841924e-01 1.13840163e+00 1.85339138e-01 2.13676140e-01 4.65848207e-01 -3.65314245e-01 -4.32136543e-02 -3.86492252e-01 -3.37373823e-01 -1.22059630e-02 2.81180590e-01 2.62657374e-01 -4.36561406e-01 -8.27810347e-01 5.52063048e-01 -7.04936206e-01 7.00304747e-01 -2.37168670e-02 8.53079140e-01 8.56603503e-01 -1.15178324e-01 3.69136244e-01 -5.78418672e-01 -1.94984898e-01 -5.83593369e-01 -7.12302506e-01 3.71677160e-01 -6.01792671e-02 -2.76707709e-01 8.77480209e-01 -5.04364669e-01 -1.25646043e+00 4.86848652e-01 -8.56654942e-02 -8.47269356e-01 -3.13815176e-02 1.97935045e-01 -1.71896905e-01 -7.60121867e-02 4.12696302e-01 3.31722140e-01 -8.28929245e-02 -3.50198776e-01 2.57184118e-01 4.58224565e-01 6.73951626e-01 -3.20946753e-01 6.48671508e-01 6.17475927e-01 -1.31641820e-01 -1.02539849e+00 -6.01682127e-01 -2.21688181e-01 -7.95960188e-01 -4.99379665e-01 9.89520490e-01 -1.36950934e+00 -6.68395936e-01 2.30650127e-01 -1.38506091e+00 -5.51886916e-01 -2.01823682e-01 7.44280159e-01 -9.54625547e-01 7.94609427e-01 -7.40343213e-01 -4.45482373e-01 2.28207652e-02 -1.38663018e+00 8.84536326e-01 1.66265950e-01 7.84369409e-02 -1.22605693e+00 -2.99856931e-01 3.41922224e-01 9.02112946e-03 2.84796327e-01 3.39741617e-01 6.73201978e-02 -1.09513795e+00 -5.66886067e-02 -2.10430577e-01 5.50983965e-01 9.33367759e-02 1.84980541e-01 -6.88349724e-01 -1.58635080e-01 -2.19853744e-02 -3.21186408e-02 6.35724723e-01 5.16907632e-01 1.77703094e+00 -4.11872685e-01 -1.72265634e-01 9.50195193e-01 1.42055666e+00 1.82930276e-01 9.76204276e-01 3.43712658e-01 9.89866138e-01 3.79856020e-01 7.84115851e-01 4.60649878e-01 8.19436133e-01 9.16034877e-01 3.73709470e-01 6.97655678e-02 -4.69223149e-02 -4.63124156e-01 6.20128989e-01 8.58774424e-01 -7.12952793e-01 -1.57867253e-01 -5.92995524e-01 3.90936702e-01 -2.28217793e+00 -1.40820062e+00 -2.45642379e-01 2.21035361e+00 6.69381857e-01 -3.36415946e-01 2.52357215e-01 2.17442624e-02 9.25140321e-01 4.60798964e-02 -4.89365458e-01 -3.85505706e-02 2.11120825e-02 -3.61895651e-01 4.76914108e-01 4.55809623e-01 -1.24831212e+00 8.87159646e-01 6.67449141e+00 9.99848127e-01 -1.08842170e+00 -2.20079012e-02 9.17746365e-01 5.39955907e-02 -2.53349721e-01 7.06880465e-02 -6.95032656e-01 9.13158953e-01 7.14432776e-01 -1.17261231e-01 6.19558096e-01 8.59942555e-01 6.54859006e-01 1.51864858e-02 -1.25925219e+00 1.36548424e+00 7.53296390e-02 -1.78219163e+00 6.50535747e-02 -3.08520913e-01 9.29926455e-01 -1.97170287e-01 -1.35252461e-01 1.19846188e-01 -1.26341805e-01 -8.38169336e-01 9.17488277e-01 7.08957911e-01 6.99633777e-01 -5.04216850e-01 5.56478500e-01 4.57929432e-01 -1.34207797e+00 1.39634060e-02 -6.91967010e-01 -2.46693864e-01 5.06530881e-01 2.69411743e-01 -4.52168852e-01 4.95376587e-01 8.09851468e-01 1.23410726e+00 -3.78763795e-01 1.18099320e+00 -1.28609955e-01 5.76122582e-01 -6.70047328e-02 4.62642759e-01 -5.86448163e-02 -4.64832783e-01 1.26170114e-01 1.07768261e+00 6.62722111e-01 2.16966495e-01 4.97899145e-01 3.91695321e-01 1.49712358e-02 2.34799802e-01 -4.81019974e-01 5.29662967e-01 2.15063408e-01 8.39281023e-01 -5.23231685e-01 -6.21343315e-01 -1.04216111e+00 1.13065946e+00 -4.86869970e-03 6.85719550e-01 -1.32572103e+00 1.42417088e-01 6.87145054e-01 2.16730595e-01 4.75343198e-01 -1.16724983e-01 -6.43028319e-02 -1.58473527e+00 1.81240514e-01 -5.18345773e-01 2.24404275e-01 -1.02376914e+00 -9.41268206e-01 5.26084423e-01 -2.27389317e-02 -1.87243795e+00 -6.76145971e-01 -4.89206344e-01 -5.44507205e-01 5.04005611e-01 -1.61413336e+00 -1.03042769e+00 -5.60909033e-01 8.66920531e-01 9.81922328e-01 -6.62089437e-02 3.98318678e-01 5.56491435e-01 -3.87080878e-01 3.99243057e-01 2.13691995e-01 1.99497893e-01 5.63565195e-01 -8.89992595e-01 3.24903280e-01 9.19607222e-01 -2.05686279e-02 -6.37443066e-02 7.91945279e-01 -4.68421191e-01 -1.29357004e+00 -1.41499901e+00 7.77035773e-01 -3.60830694e-01 5.21928847e-01 -1.05145276e-01 -9.75504935e-01 9.96037483e-01 1.68196663e-01 5.28843105e-01 4.51230437e-01 -5.32187462e-01 3.80994119e-02 -1.69741601e-01 -8.66972625e-01 7.04776466e-01 1.37917960e+00 -2.72964269e-01 -2.61221230e-01 6.23525798e-01 8.86286557e-01 -7.15571165e-01 -1.00142944e+00 3.00978273e-01 5.37788451e-01 -1.05723011e+00 1.29075432e+00 -4.51118499e-01 7.65360773e-01 -5.04215360e-01 -2.95083731e-01 -8.92098129e-01 -3.07365865e-01 -5.78369439e-01 -3.55897039e-01 1.06135595e+00 1.60655111e-01 -1.25774130e-01 9.99731064e-01 9.84073997e-01 -2.17290014e-01 -5.75260162e-01 -6.64687037e-01 -8.83109987e-01 -1.90521672e-01 -6.07260048e-01 6.36823535e-01 8.66879225e-01 -1.57352909e-01 -1.54050514e-01 -1.03346550e+00 1.61814079e-01 4.32638854e-01 1.33782148e-01 1.00226510e+00 -7.46594906e-01 -3.87111694e-01 8.08514729e-02 -6.09103024e-01 -1.61725354e+00 4.15341824e-01 -6.67279363e-01 -9.17442068e-02 -1.33789933e+00 -1.51948025e-02 -3.13684434e-01 -3.71114947e-02 1.73954263e-01 -2.86938190e-01 4.13259327e-01 4.51455742e-01 4.00206506e-01 -7.84411073e-01 6.58013701e-01 1.31657219e+00 -1.25885770e-01 -1.85419500e-01 7.77056394e-03 -1.43655106e-01 1.09432960e+00 3.90780926e-01 1.35101853e-02 -6.16786420e-01 -7.86551178e-01 -1.13657884e-01 5.59045255e-01 5.30819714e-01 -1.04525542e+00 2.58701980e-01 -5.02541125e-01 5.99417984e-01 -4.26020324e-01 5.32275796e-01 -1.04303539e+00 6.40867889e-01 2.49044448e-01 -1.03273787e-01 1.54703647e-01 -1.48441732e-01 6.51942492e-01 -3.53192270e-01 -2.40665272e-01 6.25248194e-01 -1.30254716e-01 -1.51747608e+00 7.05174506e-01 -2.81853139e-01 8.03285837e-03 1.25972736e+00 -6.73223257e-01 1.31517902e-01 -4.96853828e-01 -8.75104427e-01 2.80123979e-01 7.17983246e-01 4.65419292e-01 6.96737707e-01 -1.67726326e+00 -7.19782233e-01 7.68357962e-02 -1.28818840e-01 1.54459968e-01 4.24500287e-01 8.03461432e-01 -8.80739391e-01 -2.12135632e-03 -1.20471515e-01 -8.87344301e-01 -1.03116798e+00 9.09049749e-01 1.60640433e-01 2.04964355e-01 -9.41354990e-01 5.61882198e-01 4.49579895e-01 4.24751677e-02 1.87834457e-01 -4.09621686e-01 -1.83422521e-01 -1.28545925e-01 8.59995365e-01 5.31415164e-01 -2.89750785e-01 -8.63997340e-01 -2.13746764e-02 5.02485931e-01 2.56471246e-01 2.82197088e-01 1.27597153e+00 -3.88233483e-01 2.07743227e-01 4.08939719e-01 1.09209442e+00 -4.50792879e-01 -1.93311536e+00 -1.66823000e-01 -1.13268748e-01 -9.08378720e-01 -2.65916109e-01 5.75535633e-02 -1.28500736e+00 6.93948627e-01 3.30821633e-01 9.81138721e-02 1.28848410e+00 -3.40536475e-01 1.05452716e+00 1.54879630e-01 6.17319405e-01 -1.14809275e+00 -1.14425361e-01 4.26791281e-01 8.53590369e-01 -1.50793672e+00 -1.48482278e-01 -6.69350505e-01 -6.54017925e-01 1.35316241e+00 6.67260826e-01 -2.74102658e-01 6.48533583e-01 -9.45590138e-02 -1.64726272e-01 6.96621358e-01 -7.34599590e-01 5.29695936e-02 1.26446709e-01 7.13573396e-01 4.64660615e-01 -2.03172639e-01 -2.38109186e-01 3.20832789e-01 5.17855138e-02 6.61541343e-01 5.59555590e-01 4.87485319e-01 -2.17617616e-01 -9.73765969e-01 -3.46089810e-01 2.12839499e-01 -3.08193237e-01 -8.78502429e-02 4.97369498e-01 6.93669617e-01 1.96894437e-01 7.11231828e-01 2.33668670e-01 -3.43741924e-01 1.24554165e-01 -8.76434892e-02 4.40247893e-01 -1.66882113e-01 -2.54155267e-02 6.32085726e-02 -1.65280044e-01 -7.44927406e-01 -8.94620836e-01 -7.68349528e-01 -1.02806103e+00 -6.25692129e-01 -2.10258350e-01 1.21600769e-01 2.20540553e-01 9.41609323e-01 3.79788399e-01 1.53173238e-01 6.30894661e-01 -1.37752831e+00 7.18563655e-03 -4.97814059e-01 -3.81019920e-01 6.36627316e-01 3.20934474e-01 -4.90496665e-01 1.21171575e-03 9.27201509e-01]
[10.537529945373535, -1.0658124685287476]
d7234b4a-7e4f-49d5-a81f-9c6856a67265
efficient-bayesian-travel-time-tomography
2307.04228
null
https://arxiv.org/abs/2307.04228v1
https://arxiv.org/pdf/2307.04228v1.pdf
Efficient Bayesian travel-time tomography with geologically-complex priors using sensitivity-informed polynomial chaos expansion and deep generative networks
Monte Carlo Markov Chain (MCMC) methods commonly confront two fundamental challenges: the accurate characterization of the prior distribution and the efficient evaluation of the likelihood. In the context of Bayesian studies on tomography, principal component analysis (PCA) can in some cases facilitate the straightforward definition of the prior distribution, while simultaneously enabling the implementation of accurate surrogate models based on polynomial chaos expansion (PCE) to replace computationally intensive full-physics forward solvers. When faced with scenarios where PCA does not offer a direct means of easily defining the prior distribution alternative methods like deep generative models (e.g., variational autoencoders (VAEs)), can be employed as viable options. However, accurately producing a surrogate capable of capturing the intricate non-linear relationship between the latent parameters of a VAE and the outputs of forward modeling presents a notable challenge. Indeed, while PCE models provide high accuracy when the input-output relationship can be effectively approximated by relatively low-degree multivariate polynomials, this condition is typically unmet when utilizing latent variables derived from deep generative models. In this contribution, we present a strategy that combines the excellent reconstruction performances of VAE in terms of prio representation with the accuracy of PCA-PCE surrogate modeling in the context of Bayesian ground penetrating radar (GPR) travel-time tomography. Within the MCMC process, the parametrization of the VAE is leveraged for prior exploration and sample proposal. Concurrently, modeling is conducted using PCE, which operates on either globally or locally defined principal components of the VAE samples under examination.
['Niklas Linde', 'Stefano Marelli', 'Shiran Levy', 'Macarena Amaya', 'Giovanni Angelo Meles']
2023-07-09
null
null
null
null
['gpr', 'gpr']
['computer-vision', 'miscellaneous']
[ 1.20638765e-01 -3.91559511e-01 4.29281622e-01 3.06947790e-02 -8.92603397e-01 -4.99357820e-01 8.53632033e-01 -2.62137856e-02 -3.04316819e-01 7.78501749e-01 -5.90920486e-02 -4.22713548e-01 -6.35515630e-01 -9.27037597e-01 -4.48131889e-01 -1.14387810e+00 1.13333061e-01 1.01381743e+00 -3.92242223e-01 1.60934553e-01 -1.70128457e-02 1.11046505e+00 -1.31858277e+00 -3.50113064e-01 7.87973583e-01 9.09707844e-01 -2.90324306e-03 5.19778311e-01 7.45231006e-03 2.69620836e-01 -2.75731623e-01 -1.52925715e-01 -3.31074819e-02 -2.03702509e-01 -1.20237298e-01 -1.51169449e-01 -3.78502935e-01 -4.18673098e-01 -1.09454088e-01 8.42689276e-01 3.20884109e-01 2.81752825e-01 1.17720056e+00 -8.96393418e-01 1.01830222e-01 1.36791885e-01 -1.67806298e-01 4.56419028e-02 2.19510674e-01 2.23438695e-01 6.13691211e-01 -9.51588213e-01 3.92433822e-01 9.40923393e-01 7.57852495e-01 -5.47811352e-02 -1.83514559e+00 -4.44295883e-01 -4.34438437e-01 -3.33613455e-02 -1.66820157e+00 -3.68195206e-01 8.42497230e-01 -7.99354613e-01 7.79786348e-01 2.44378015e-01 1.00458384e+00 1.24581611e+00 5.03188193e-01 2.54130960e-01 1.22141123e+00 -2.87612975e-01 6.55737758e-01 4.98715118e-02 1.54862165e-01 2.72149265e-01 3.34108442e-01 4.94577050e-01 -2.70259678e-01 -6.67831838e-01 8.95597994e-01 -8.46463814e-02 -3.12595874e-01 -4.82159764e-01 -7.01203585e-01 1.12087822e+00 -5.25731444e-02 1.17208153e-01 -1.00229621e+00 2.67566085e-01 1.22169629e-01 -2.80182600e-01 2.66118765e-01 3.92597914e-01 -9.98540819e-02 -2.33506218e-01 -1.36828530e+00 4.34964210e-01 9.15582180e-01 4.84140515e-01 9.60472763e-01 3.61561328e-01 -1.21762261e-01 2.18472913e-01 7.26759911e-01 6.79017961e-01 -3.44212539e-02 -8.86271656e-01 -7.12191612e-02 7.58092627e-02 3.10108900e-01 -8.31313431e-01 -6.99725524e-02 -7.49156713e-01 -8.80433679e-01 2.73770928e-01 3.20541710e-01 -1.47270650e-01 -8.72225821e-01 1.49618530e+00 5.22649109e-01 -5.44539541e-02 1.96907818e-01 5.84111154e-01 2.83604413e-01 7.88635314e-01 2.56512791e-01 -4.36161309e-01 1.32962584e+00 -4.17768732e-02 -4.83358890e-01 -1.81240737e-01 3.77011895e-02 -7.02863276e-01 5.71168303e-01 3.77583712e-01 -8.64671350e-01 -2.83748686e-01 -8.90682459e-01 3.05216402e-01 6.58277869e-02 9.04499069e-02 5.83299339e-01 6.81104839e-01 -8.19790006e-01 6.38436079e-01 -1.33989072e+00 4.49550636e-02 2.26506323e-01 1.13074698e-01 -6.16608933e-02 -1.86381042e-01 -9.07115757e-01 1.02299154e+00 2.11134121e-01 5.30176520e-01 -1.29520309e+00 -8.26883793e-01 -5.30810595e-01 4.25129980e-01 6.37203604e-02 -9.29777861e-01 9.82872546e-01 -5.19956231e-01 -1.57401121e+00 8.31410587e-02 -9.97551307e-02 -3.57967407e-01 7.18491733e-01 -1.15171792e-02 -2.03906670e-01 3.90526950e-01 -4.48933803e-02 3.25337589e-01 1.10232890e+00 -1.22306693e+00 2.33449608e-01 2.91815232e-02 -3.54986846e-01 -2.20759604e-02 3.57637018e-01 -2.49584094e-01 -1.12586126e-01 -3.27181518e-01 3.61700356e-01 -9.12121892e-01 -4.54978257e-01 -1.11530863e-01 -2.25911662e-01 2.91628182e-01 5.96865356e-01 -7.98686564e-01 9.18267787e-01 -1.99809837e+00 1.14135034e-01 6.44164264e-01 1.54641613e-01 -4.33277376e-02 4.88280386e-01 8.44618499e-01 -8.88748318e-02 -3.24002802e-01 -5.57038009e-01 -3.32586914e-01 5.34387529e-02 3.21914881e-01 -5.86245716e-01 5.82283735e-01 3.67649287e-01 7.98865378e-01 -5.45207202e-01 -2.50282586e-01 3.56042087e-01 1.00617421e+00 -5.57144701e-01 8.61778632e-02 -2.34359160e-01 8.35669279e-01 -6.54091537e-01 3.70428443e-01 6.23641193e-01 -1.73191383e-01 2.33854741e-01 -1.86872914e-01 -1.89970180e-01 1.05849564e-01 -1.30970466e+00 1.08938074e+00 -5.41015625e-01 5.50702333e-01 2.39726126e-01 -9.16784406e-01 9.46021676e-01 4.30265486e-01 4.61904138e-01 -2.13788956e-01 8.95941183e-02 3.17581415e-01 1.43724874e-01 -1.13756083e-01 5.28614819e-01 -7.25286663e-01 2.04873353e-01 5.70384026e-01 2.63023539e-03 -3.98481965e-01 -9.95746925e-02 5.32306321e-02 7.78466761e-01 3.99786502e-01 3.78813416e-01 -4.62209135e-01 1.81332901e-01 1.34002745e-01 2.35212415e-01 6.48415804e-01 2.52144188e-01 4.78357404e-01 3.40022445e-01 -1.06396958e-01 -1.18003595e+00 -1.15518081e+00 -6.57317162e-01 -1.08731023e-04 -4.66524065e-01 -1.26264289e-01 -6.04900777e-01 2.15767339e-01 -9.26705897e-02 1.12176049e+00 -3.01993817e-01 2.28862539e-02 -4.67151135e-01 -1.17484868e+00 4.88872468e-01 3.18601191e-01 2.28275761e-01 -5.89383662e-01 -9.38493073e-01 6.40576422e-01 2.93651217e-04 -9.02926683e-01 4.69075531e-01 3.91600579e-01 -1.05920064e+00 -6.89472795e-01 -5.96985340e-01 2.74339676e-01 5.16147316e-01 -2.87642062e-01 8.34865510e-01 -4.78204519e-01 -1.73219606e-01 6.70652807e-01 -1.26507571e-02 -1.22381538e-01 -6.71958029e-01 -3.16999704e-01 8.60184897e-03 9.35154036e-02 1.14309371e-01 -1.13202608e+00 -4.85651791e-01 -5.24712503e-02 -9.38768327e-01 1.18154243e-01 5.86698592e-01 7.43513584e-01 7.59151220e-01 1.79027587e-01 2.53262937e-01 -4.04602975e-01 4.59372938e-01 -6.32414043e-01 -8.88459325e-01 9.94030908e-02 -5.91025889e-01 2.91637748e-01 5.32688200e-01 -4.93555456e-01 -1.17050374e+00 -1.85822725e-01 -4.03858304e-01 -6.65212274e-01 -2.47703597e-01 1.02533150e+00 -5.79662211e-02 -5.11825420e-02 4.97140825e-01 3.61384720e-01 7.28834188e-04 -5.22062957e-01 4.93755080e-02 1.23538531e-01 4.59649384e-01 -9.63979363e-01 7.97146678e-01 6.49036407e-01 4.25738424e-01 -9.12967563e-01 -2.45801702e-01 -2.42278397e-01 -3.73401523e-01 -4.00189310e-01 7.70172596e-01 -7.96297669e-01 -4.71233934e-01 1.98883355e-01 -1.03251624e+00 7.32578523e-03 -3.96566600e-01 9.00268137e-01 -5.42035818e-01 4.30719405e-01 -3.15701336e-01 -1.32556093e+00 -2.36328423e-01 -1.40230751e+00 8.91915262e-01 1.35888048e-02 -3.97927225e-01 -9.84407663e-01 6.33187741e-02 -7.24061579e-02 5.95413387e-01 5.60106099e-01 1.15345049e+00 -3.45760882e-01 -8.99807990e-01 -5.32363474e-01 1.28252551e-01 1.07242949e-01 -4.45356369e-01 3.43512267e-01 -1.02616525e+00 -2.71089166e-01 4.87703562e-01 2.68643528e-01 5.58892727e-01 7.87439167e-01 5.40256262e-01 -1.23044364e-02 -3.06630820e-01 7.60399044e-01 1.67825222e+00 1.75029561e-01 6.84639394e-01 9.10571497e-03 2.88208067e-01 4.54217613e-01 1.73223391e-01 6.63729250e-01 5.51575003e-03 5.80970943e-01 2.07463667e-01 3.47363412e-01 2.68548995e-01 -1.84697092e-01 1.58373579e-01 6.35057151e-01 -2.85305619e-01 -9.87808406e-02 -1.21179748e+00 1.48495421e-01 -1.37651944e+00 -9.43678737e-01 -3.59607041e-01 2.07527924e+00 3.80938232e-01 1.90408722e-01 -3.50124687e-01 2.26915076e-01 4.77886021e-01 -1.66573077e-01 -3.93079191e-01 -1.82412744e-01 8.01595747e-02 4.86824095e-01 2.95538992e-01 5.40500879e-01 -6.78258598e-01 1.61769822e-01 6.07582808e+00 8.79479647e-01 -1.06904852e+00 2.22644582e-01 2.60472983e-01 1.99637294e-01 -6.86307013e-01 5.24653792e-01 -8.22025955e-01 4.09381419e-01 1.21455526e+00 -1.12634450e-01 4.17568713e-01 6.96014822e-01 2.71556944e-01 -4.95607257e-01 -8.51555705e-01 7.73028076e-01 -4.49908704e-01 -1.22295380e+00 -1.03399165e-01 4.11567092e-01 4.42233175e-01 -9.75561962e-02 1.02896698e-01 2.07109675e-01 5.94173893e-02 -9.79078352e-01 9.00140643e-01 9.93581176e-01 8.41650307e-01 -6.60907149e-01 4.64911580e-01 6.62119389e-01 -7.76923120e-01 9.14853215e-02 -2.14133397e-01 -3.38708200e-02 6.13740444e-01 1.06769538e+00 -8.24547231e-01 6.90276623e-01 2.91437477e-01 1.22739904e-01 4.14444925e-03 9.51708674e-01 -2.17777520e-01 9.74350452e-01 -7.47320175e-01 2.29573876e-01 2.88663715e-01 -8.39828670e-01 9.49819982e-01 1.04419219e+00 6.94180429e-01 1.22234441e-01 -2.79045831e-02 1.43969703e+00 6.72683716e-01 -2.09631726e-01 -2.90745199e-01 -2.92507499e-01 4.98786986e-01 1.17907000e+00 -8.28659534e-01 -5.98772429e-02 1.28407195e-01 1.56613588e-01 -1.05632015e-01 7.19767451e-01 -8.27393651e-01 3.09990376e-01 5.36055744e-01 2.76520252e-01 5.93808770e-01 -7.32748210e-01 -2.15839386e-01 -9.49606359e-01 -1.36955783e-01 -7.10077167e-01 3.05376388e-02 -8.04851294e-01 -1.01084304e+00 4.91439372e-01 5.78479230e-01 -1.12777400e+00 -7.01744914e-01 -5.07486045e-01 -6.92807317e-01 1.32382762e+00 -1.38011253e+00 -1.06905866e+00 -3.54773998e-02 2.84126371e-01 -9.34642330e-02 8.85325596e-02 1.12126696e+00 1.95292935e-01 -4.85254288e-01 -9.97664258e-02 6.16077602e-01 -4.40308392e-01 -4.11172807e-02 -7.04852462e-01 1.06716547e-02 9.94932532e-01 -6.30608201e-02 8.77751708e-01 1.22031558e+00 -9.04775321e-01 -1.50763702e+00 -6.30337358e-01 3.46347481e-01 -1.46576986e-01 6.82504714e-01 -1.68966398e-01 -9.66574609e-01 5.84942400e-01 -4.10280228e-01 -1.74082741e-01 7.02166080e-01 -8.05797800e-02 1.54126421e-01 1.80171341e-01 -8.72956097e-01 2.55591959e-01 1.52467251e-01 -7.28425503e-01 -3.76865655e-01 -3.03750038e-02 -3.74390073e-02 -3.04857343e-01 -8.83734226e-01 3.23650390e-01 6.25502229e-01 -8.65191817e-01 1.14269960e+00 -1.22644484e-01 3.02815735e-01 -3.60265464e-01 -3.86177599e-01 -9.98212278e-01 -1.60969809e-01 -6.14218354e-01 -2.62485802e-01 1.20723307e+00 -1.29022403e-02 -7.11888850e-01 5.68838060e-01 7.77383924e-01 -1.20618902e-01 -5.98671913e-01 -1.32078350e+00 -5.68922877e-01 4.08942364e-02 -8.47803593e-01 4.38969165e-01 4.93240058e-01 -7.52920628e-01 -3.31430212e-02 -1.02761835e-01 5.13742387e-01 8.65214169e-01 1.81834757e-01 5.70876956e-01 -1.43784213e+00 -8.17919135e-01 -2.56414562e-01 -1.28400490e-01 -8.83277059e-01 -2.04061698e-02 -7.68772662e-01 6.55388832e-02 -1.35098505e+00 -3.57956961e-02 -6.12590551e-01 2.01456651e-01 -2.87097365e-01 1.73620135e-01 -2.20039859e-01 -6.70797750e-02 5.37354171e-01 4.53154594e-01 7.36392975e-01 8.54571879e-01 2.91673362e-01 -4.44718637e-02 1.42106995e-01 -1.17778301e-01 7.43551135e-01 4.25008595e-01 -7.46695578e-01 -1.98753789e-01 -2.42862701e-01 4.74455923e-01 6.63765907e-01 9.20282423e-01 -1.04291725e+00 4.05519783e-01 -8.69742706e-02 4.89721030e-01 -8.80759358e-01 7.65776157e-01 -6.63607538e-01 1.23807573e+00 4.02707845e-01 3.01190585e-01 -2.92489119e-02 2.10200578e-01 7.30808496e-01 -2.02826694e-01 -5.64683735e-01 7.82575428e-01 -9.72925425e-02 -3.06082755e-01 1.62209362e-01 -8.59071553e-01 -4.81264830e-01 5.50307810e-01 -3.76739353e-01 1.96884915e-01 -4.42128092e-01 -8.95023227e-01 -2.69974887e-01 4.20325696e-01 -7.58740246e-01 5.03053606e-01 -1.00402391e+00 -6.28566504e-01 2.86426932e-01 -4.00350958e-01 3.01856864e-02 5.42643428e-01 1.13271630e+00 -6.41665518e-01 3.58128041e-01 -4.44265641e-02 -6.86393797e-01 -4.95239735e-01 2.66697347e-01 4.87435937e-01 -5.52990854e-01 -7.20092654e-01 5.48743010e-01 1.95734903e-01 -8.85475948e-02 -3.43514800e-01 -8.36250558e-02 1.34939710e-02 3.10345054e-01 8.78908709e-02 4.51992244e-01 1.11722209e-01 -6.72733188e-01 -2.33012229e-01 2.63686776e-01 3.72727603e-01 -4.90617990e-01 1.42421317e+00 5.34525104e-02 -1.10002272e-01 3.27113658e-01 9.22540843e-01 4.62296791e-02 -1.36051095e+00 6.98487684e-02 -2.55649775e-01 -7.12587237e-02 5.73414803e-01 -5.54370880e-01 -6.06236815e-01 1.14038050e+00 3.13835353e-01 2.57629473e-02 8.53728056e-01 -3.44111055e-01 4.31013167e-01 1.79714426e-01 3.61352235e-01 -7.03801274e-01 -3.41943473e-01 3.05206120e-01 7.92014539e-01 -5.57740629e-01 3.27369481e-01 -1.43096149e-01 -2.96211034e-01 1.31845856e+00 -2.42332950e-01 -2.35497952e-01 8.52528155e-01 3.34152848e-01 -3.90432328e-01 -3.67821872e-01 -4.35452104e-01 -3.05794757e-02 2.61831194e-01 3.15946460e-01 5.65529242e-02 1.61487505e-01 -1.75818130e-01 6.48244202e-01 -2.88315147e-01 -1.87604800e-01 4.53631848e-01 8.01284671e-01 -1.65914431e-01 -9.62253451e-01 -7.13803113e-01 3.16362739e-01 -2.48017177e-01 -2.48860449e-01 4.03173476e-01 8.21654320e-01 -2.01496944e-01 6.69717312e-01 1.00521527e-01 2.06786796e-01 -8.77076536e-02 3.38348061e-01 4.44017529e-01 -3.65604460e-01 -2.33538568e-01 6.43774569e-01 9.83418524e-02 -5.09468354e-02 -2.36702502e-01 -1.18594801e+00 -8.38153899e-01 -2.11339980e-01 -5.30219913e-01 3.24487686e-01 9.68860686e-01 1.11171210e+00 1.89273525e-02 4.01508987e-01 1.77816331e-01 -1.14211965e+00 -9.74267662e-01 -7.55411983e-01 -6.91044688e-01 -1.84093252e-01 2.05129325e-01 -8.84018838e-01 -4.66099918e-01 -2.99542010e-01]
[6.812584400177002, 3.6402971744537354]
1cf9aca9-1960-490d-a7e4-e8770f288dc0
medical-federated-model-with-mixture-of
2306.14483
null
https://arxiv.org/abs/2306.14483v1
https://arxiv.org/pdf/2306.14483v1.pdf
Medical Federated Model with Mixture of Personalized and Sharing Components
Although data-driven methods usually have noticeable performance on disease diagnosis and treatment, they are suspected of leakage of privacy due to collecting data for model training. Recently, federated learning provides a secure and trustable alternative to collaboratively train model without any exchange of medical data among multiple institutes. Therefore, it has draw much attention due to its natural merit on privacy protection. However, when heterogenous medical data exists between different hospitals, federated learning usually has to face with degradation of performance. In the paper, we propose a new personalized framework of federated learning to handle the problem. It successfully yields personalized models based on awareness of similarity between local data, and achieves better tradeoff between generalization and personalization than existing methods. After that, we further design a differentially sparse regularizer to improve communication efficiency during procedure of model training. Additionally, we propose an effective method to reduce the computational cost, which improves computation efficiency significantly. Furthermore, we collect 5 real medical datasets, including 2 public medical image datasets and 3 private multi-center clinical diagnosis datasets, and evaluate its performance by conducting nodule classification, tumor segmentation, and clinical risk prediction tasks. Comparing with 13 existing related methods, the proposed method successfully achieves the best model performance, and meanwhile up to 60% improvement of communication efficiency. Source code is public, and can be accessed at: https://github.com/ApplicationTechnologyOfMedicalBigData/pFedNet-code.
['Kunlun He', 'Xinwang Liu', 'Qinghe Liu', 'Yawei Zhao']
2023-06-26
null
null
null
null
['tumor-segmentation']
['computer-vision']
[-7.26531893e-02 8.98894221e-02 -5.61831772e-01 -5.88919342e-01 -9.67107296e-01 -3.34951729e-01 6.14866614e-02 1.67413279e-01 -1.97619066e-01 6.43436372e-01 2.93695360e-01 -4.81529385e-01 -1.92786857e-01 -8.34532857e-01 -4.73670065e-01 -9.84506369e-01 3.60005647e-02 2.19034135e-01 -9.34585780e-02 3.27666014e-01 -2.65401304e-01 2.18321189e-01 -7.52669752e-01 4.86087322e-01 1.10425615e+00 9.96959746e-01 -2.53391918e-03 2.11374626e-01 4.41101752e-02 7.68025517e-01 -2.87105769e-01 -7.78911769e-01 3.60593736e-01 -3.59954000e-01 -9.79229033e-01 -1.69833079e-01 -1.93275616e-01 -5.52904189e-01 -5.50029516e-01 1.22860909e+00 7.43485212e-01 -4.74812746e-01 1.15547836e-01 -1.11709642e+00 -7.03981698e-01 8.16048384e-01 -3.68891150e-01 -2.52367765e-01 -2.63221823e-02 1.20899357e-01 6.29267752e-01 -3.29800427e-01 5.17008245e-01 4.51734364e-01 6.38931155e-01 8.27697933e-01 -6.99870586e-01 -9.81788397e-01 -1.67405307e-01 1.52072743e-01 -1.38967633e+00 -2.49607429e-01 6.21264994e-01 -1.85719162e-01 1.43737104e-02 7.00423777e-01 5.15468419e-01 1.23580551e+00 2.92417616e-01 9.62414861e-01 1.10637593e+00 7.24538341e-02 1.23777598e-01 5.71495056e-01 -4.79634404e-02 7.67385662e-01 5.43813944e-01 8.48620608e-02 -1.54640958e-01 -8.47947240e-01 4.79477704e-01 7.35530972e-01 -6.27549589e-01 -2.79039830e-01 -1.31377923e+00 5.56639850e-01 5.94496548e-01 3.04629534e-01 -2.36209869e-01 -1.60949662e-01 5.15421510e-01 4.21305858e-02 3.51806492e-01 -2.04993486e-01 -5.32020569e-01 2.31579244e-01 -5.33820629e-01 -1.05981022e-01 8.13525915e-01 1.02224648e+00 4.40454304e-01 -4.89541143e-01 -1.54053837e-01 4.74795818e-01 2.39419699e-01 4.09776568e-01 9.52201843e-01 -7.48193800e-01 3.26747477e-01 6.03601694e-01 -4.02468108e-02 -1.19526410e+00 -1.81930512e-01 -5.67849696e-01 -1.40082979e+00 -5.77612102e-01 1.16349444e-01 -3.97282451e-01 -3.30008537e-01 1.52739561e+00 7.82047093e-01 4.20277625e-01 3.09504300e-01 9.04974520e-01 8.77166450e-01 3.51688027e-01 -2.58107074e-02 -4.03265238e-01 1.50474942e+00 -8.76829147e-01 -8.25289905e-01 3.42130005e-01 1.06550729e+00 -5.74942887e-01 6.67454243e-01 2.34500229e-01 -7.61000454e-01 2.89616678e-02 -6.69579268e-01 2.01193511e-01 1.49097247e-02 1.25632867e-01 9.97686505e-01 8.23975801e-01 -7.04441249e-01 5.27748764e-01 -1.11795318e+00 -2.89321214e-01 1.01649225e+00 4.03632879e-01 -5.84377944e-01 -2.51433492e-01 -1.15055537e+00 2.68571943e-01 1.67972893e-01 -3.36663611e-02 -9.13917840e-01 -8.91271591e-01 -4.90486264e-01 -6.83352724e-02 2.71898031e-01 -1.01233876e+00 1.23656344e+00 -6.99227393e-01 -1.18330681e+00 8.82223248e-01 2.13954449e-01 -5.42958677e-01 7.11370885e-01 1.56316549e-01 -6.90341055e-01 2.23928671e-02 -1.87392861e-01 5.15355356e-02 5.21105409e-01 -1.02967334e+00 -6.49906158e-01 -6.28977597e-01 -3.40822279e-01 -1.21991016e-01 -1.01232314e+00 -1.11541696e-01 -3.82490784e-01 -6.19309306e-01 4.31117080e-02 -8.74244153e-01 -6.35967195e-01 2.13717222e-01 -4.95941848e-01 2.07582295e-01 9.52747703e-01 -7.52700269e-01 1.24728286e+00 -2.44518948e+00 -4.85878885e-01 2.85118669e-01 5.10871172e-01 2.66330093e-01 2.22190961e-01 2.44121104e-01 1.51317447e-01 2.73419261e-01 -3.64308774e-01 1.86540408e-03 -3.09293419e-01 2.16141254e-01 -2.11336389e-02 7.43382394e-01 -4.39570069e-01 8.54646623e-01 -7.08117545e-01 -8.95654321e-01 -2.39746168e-01 4.95411038e-01 -5.84263206e-01 2.30129078e-01 1.21194527e-01 7.16293037e-01 -1.22181010e+00 8.10487926e-01 9.79899049e-01 -6.81894243e-01 4.43919659e-01 -3.35868299e-01 3.63997310e-01 -2.20906064e-01 -9.92382109e-01 1.74748898e+00 -2.49072343e-01 -2.30046213e-01 3.48308235e-01 -9.04450953e-01 7.14523792e-01 7.77878761e-01 9.43014085e-01 -1.61258698e-01 5.68850815e-01 2.97800630e-01 -2.71895438e-01 -6.63271427e-01 -2.82054674e-02 3.98458391e-02 2.24302486e-02 5.33987880e-01 -2.32384533e-01 4.17173505e-01 -7.67126083e-01 3.79576862e-01 1.29086518e+00 -4.69778180e-01 1.83165297e-01 -1.36348590e-01 6.61453366e-01 -1.62882991e-02 8.67788196e-01 2.42808536e-01 -6.10107303e-01 2.91531324e-01 1.02048628e-01 -4.24689263e-01 -4.23935711e-01 -7.74193585e-01 -3.39879245e-01 4.29304451e-01 3.14809769e-01 -4.89736885e-01 -6.58858418e-01 -1.04850185e+00 1.28923059e-01 3.20435226e-01 -4.68081683e-01 -3.93932909e-01 -2.73765683e-01 -9.74555850e-01 7.06229448e-01 2.13808030e-01 8.02482665e-01 -6.08563900e-01 -3.29463571e-01 -9.23621655e-02 -3.96520436e-01 -7.88255453e-01 -6.84272647e-01 -2.26729333e-01 -1.06573439e+00 -1.10630774e+00 -6.74478412e-01 -8.06159556e-01 7.95112431e-01 2.66234100e-01 6.80710137e-01 2.31840625e-01 -4.95363116e-01 3.74547333e-01 -2.39029720e-01 -4.62501466e-01 -3.16311508e-01 6.57313913e-02 -1.41024798e-01 4.38770175e-01 2.77872205e-01 -3.59680921e-01 -1.11971867e+00 3.82770538e-01 -8.38034034e-01 -1.43124416e-01 5.19665360e-01 1.03142619e+00 7.26961613e-01 5.31225614e-02 4.02455062e-01 -1.40527761e+00 3.54706228e-01 -8.05392504e-01 -2.81963378e-01 2.77874619e-01 -8.90982628e-01 -4.04444277e-01 7.22709894e-01 -3.55768874e-02 -1.27661514e+00 3.29414487e-01 -9.13763884e-03 -3.94386441e-01 -1.33330032e-01 4.06328022e-01 -3.34221542e-01 -2.74579287e-01 5.03154993e-01 1.28396720e-01 4.19816315e-01 -4.74320233e-01 1.22146793e-01 1.12576175e+00 2.52849072e-01 -3.49544525e-01 5.99067271e-01 7.11144984e-01 -2.37279922e-01 -7.64528960e-02 -6.29605889e-01 -2.78199285e-01 1.07579514e-01 1.43180519e-01 6.08124614e-01 -1.09882784e+00 -9.36973870e-01 3.76265973e-01 -7.89030790e-01 2.22927690e-01 -3.00147951e-01 8.92601967e-01 -2.52516896e-01 3.68703514e-01 -8.83284867e-01 -3.31639856e-01 -8.96455944e-01 -9.38359916e-01 4.90815043e-01 1.77035496e-01 2.87402272e-01 -9.13944602e-01 -1.36978015e-01 6.06991887e-01 6.11858368e-01 4.61226791e-01 7.40636647e-01 -8.12859654e-01 -7.65300810e-01 -4.82807904e-01 4.14030850e-02 2.62002915e-01 4.76551801e-01 -2.02255130e-01 -9.72673178e-01 -6.63480461e-01 4.05642062e-01 -2.31434599e-01 4.10413474e-01 1.99223578e-01 1.88522339e+00 -7.24265814e-01 -8.47961724e-01 1.05784738e+00 1.36764193e+00 1.68323871e-02 3.88599873e-01 4.73571904e-02 6.21592283e-01 5.38892627e-01 7.07234919e-01 8.70699286e-01 4.49442923e-01 2.90732905e-02 6.07387900e-01 -1.81892917e-01 3.97664845e-01 -2.72969157e-01 -9.00300071e-02 1.02379596e+00 1.03722893e-01 6.72569498e-03 -8.15657735e-01 5.22145033e-01 -1.85539556e+00 -8.16970646e-01 -1.14313364e-01 2.31400895e+00 1.03562033e+00 -4.53594416e-01 -1.59000158e-01 -3.41971040e-01 7.36465335e-01 -1.81785300e-02 -8.10751021e-01 6.56517893e-02 8.86586234e-02 -2.06191182e-01 8.32983911e-01 -1.65825740e-01 -8.73402476e-01 4.98128146e-01 5.45028543e+00 1.03580022e+00 -1.30443263e+00 6.09155178e-01 1.14475679e+00 -2.17220172e-01 -4.92876917e-01 -1.58136100e-01 -3.85485709e-01 6.04974210e-01 8.00627530e-01 -6.27852440e-01 5.85061610e-02 1.11572826e+00 -8.13408047e-02 4.35555369e-01 -8.58876050e-01 1.07396817e+00 -4.10676420e-01 -1.55888438e+00 -2.48675551e-02 3.44007373e-01 6.82248652e-01 5.03906012e-02 1.99800074e-01 -2.10637808e-01 3.62953991e-01 -8.58853400e-01 -1.73571989e-01 5.30077159e-01 8.65354717e-01 -8.03552628e-01 9.36468244e-01 6.48057818e-01 -8.87162745e-01 -1.58588573e-01 -4.44369584e-01 5.04856288e-01 -1.92704111e-01 6.86187387e-01 -7.27757096e-01 1.05335164e+00 9.08778369e-01 7.82164037e-01 -3.83984983e-01 9.66119409e-01 2.27190360e-01 8.04945290e-01 -1.39561117e-01 6.82937130e-02 -1.33814722e-01 -1.30629241e-01 1.66517809e-01 8.00650060e-01 4.94275957e-01 5.35731494e-01 2.02190623e-01 4.05820459e-01 -2.35981256e-01 6.34580433e-01 -6.17853403e-01 5.75407743e-02 6.79580152e-01 1.58542252e+00 -1.63701728e-01 -1.74214929e-01 -3.83478492e-01 9.00193930e-01 1.36623010e-01 -7.96323121e-02 -9.45977509e-01 -2.77744144e-01 5.19400656e-01 3.86160761e-02 -8.24590996e-02 3.89023572e-01 -1.24470308e-01 -1.34697008e+00 5.88782653e-02 -1.07823682e+00 9.71231341e-01 -2.01979026e-01 -1.51731050e+00 7.16055989e-01 -4.87202168e-01 -1.59801519e+00 1.74653515e-01 -5.45602776e-02 -5.07744431e-01 6.21772408e-01 -1.16683698e+00 -1.40795183e+00 -5.38925469e-01 1.29153252e+00 -4.15470719e-01 -3.72488052e-01 1.18512213e+00 5.11323452e-01 -6.94818974e-01 1.20905852e+00 4.75046456e-01 2.49679491e-01 9.60990191e-01 -5.83144426e-01 -2.42215931e-01 5.49533486e-01 -2.41343066e-01 6.97127461e-01 8.83683339e-02 -5.40240705e-01 -1.66697395e+00 -1.47118115e+00 7.12592959e-01 -2.41072983e-01 4.05241340e-01 -1.47092029e-01 -9.11807120e-01 7.61716008e-01 1.06740080e-01 6.32161677e-01 1.30299067e+00 -1.36882752e-01 -2.24790752e-01 -5.94779551e-01 -1.75143850e+00 3.78837466e-01 9.04819667e-01 -3.90384316e-01 6.44232333e-02 8.77978384e-01 1.08820379e+00 -4.60624486e-01 -1.42422557e+00 5.16174197e-01 2.73555905e-01 -8.32970083e-01 6.88038886e-01 -5.64514399e-01 1.44161299e-01 -1.36151865e-01 -9.11719576e-02 -8.66649151e-01 -4.05574173e-01 -7.89360583e-01 -2.34069228e-02 1.29595029e+00 3.15738946e-01 -1.24758565e+00 1.25814545e+00 9.22202766e-01 6.32828176e-02 -1.13833928e+00 -9.39470351e-01 -4.95531470e-01 -7.56439641e-02 -1.50436237e-01 1.11543906e+00 1.45061564e+00 1.77225053e-01 -3.66056323e-01 -5.26347220e-01 4.03811693e-01 8.22052002e-01 2.78557211e-01 6.82568431e-01 -8.50244224e-01 -4.40610379e-01 7.33914748e-02 -4.31758225e-01 -5.22224963e-01 -1.25531256e-01 -1.17905676e+00 -4.97221172e-01 -1.16966772e+00 3.40242594e-01 -8.83129358e-01 -7.22988725e-01 7.66660690e-01 -6.64501116e-02 -6.17839806e-02 -1.41219348e-01 4.94000852e-01 -4.78153080e-01 4.72051710e-01 1.56509650e+00 -1.59287214e-01 1.58791378e-01 3.08845043e-01 -1.01851356e+00 4.17902887e-01 1.02170312e+00 -5.18854558e-01 -4.24547464e-01 -5.08363962e-01 -3.27442497e-01 4.26005542e-01 3.91057104e-01 -8.68167460e-01 5.26053905e-01 -2.15900406e-01 3.14399689e-01 -2.56181121e-01 -1.18718289e-01 -1.36405051e+00 6.84847295e-01 9.82700109e-01 -2.28043601e-01 -3.64172369e-01 -2.21056059e-01 7.32188582e-01 -2.69797355e-01 1.14195220e-01 5.89565635e-01 -1.66514099e-01 -1.86497837e-01 1.11321020e+00 1.27630994e-01 -1.79170117e-01 1.41434455e+00 1.21507145e-01 -3.92360151e-01 -7.01527596e-02 -6.71083033e-01 4.58964199e-01 4.54277873e-01 1.33632973e-01 4.99562681e-01 -1.31134140e+00 -8.02867532e-01 3.82832229e-01 8.56955051e-02 1.12413369e-01 8.92035067e-01 1.10214019e+00 -3.93626302e-01 1.70506909e-01 7.27803707e-02 -5.97857654e-01 -1.32219315e+00 8.58553171e-01 3.93598408e-01 -1.35777101e-01 -7.77079046e-01 8.02327037e-01 2.26899445e-01 -7.25717783e-01 3.58237267e-01 2.50874460e-02 2.68426657e-01 -3.87492925e-01 5.73815525e-01 3.34189594e-01 1.51754677e-01 -3.30964386e-01 -5.05330622e-01 9.02660117e-02 -4.39741880e-01 5.72691083e-01 1.15840316e+00 -9.64322388e-02 -3.92803967e-01 -1.22153856e-01 1.44798613e+00 1.67571113e-01 -8.43697965e-01 -3.99060696e-01 -4.03857648e-01 -7.03438044e-01 3.65354083e-02 -7.02745438e-01 -1.70609236e+00 4.54313666e-01 8.11766386e-01 -8.66393372e-02 1.37939751e+00 -1.51072279e-01 1.25447249e+00 8.08615610e-02 7.88186252e-01 -5.51473141e-01 -4.53893423e-01 -1.52625576e-01 2.81582415e-01 -1.36415100e+00 1.25080526e-01 -5.75745761e-01 -7.23115265e-01 6.67339683e-01 5.46431661e-01 1.31853372e-01 1.23130667e+00 3.25509340e-01 2.22595632e-01 -7.98562020e-02 -7.39564598e-01 5.71312785e-01 -2.64167935e-01 4.53321904e-01 2.02916265e-01 3.92849714e-01 -4.59318042e-01 1.29594636e+00 -4.87798341e-02 2.75509447e-01 1.95223644e-01 1.09379351e+00 -3.88853364e-02 -1.26327634e+00 -2.46065378e-01 6.30549729e-01 -8.31628442e-01 -4.78365347e-02 -8.10737610e-02 2.78570920e-01 1.14391588e-01 7.33816981e-01 -4.80295211e-01 -5.92156410e-01 1.32923231e-01 -2.77931213e-01 -9.09929723e-02 -2.58109570e-01 -8.88508439e-01 -6.28317222e-02 -2.82245725e-01 -7.45626867e-01 -2.10229486e-01 -5.54150641e-01 -1.30062485e+00 -4.93701607e-01 -3.70858818e-01 5.68769157e-01 6.22512758e-01 4.47044998e-01 8.38950574e-01 2.98741281e-01 1.23409259e+00 1.27561942e-01 -9.54136074e-01 -2.94564873e-01 -7.86751151e-01 3.18882108e-01 1.34693637e-01 1.80829525e-01 -1.96687549e-01 -1.54032903e-02]
[6.062697887420654, 6.486255645751953]
a7434222-cd61-4f81-b373-0f5f55f2a859
samo-speaker-attractor-multi-center-one-class
2211.02718
null
https://arxiv.org/abs/2211.02718v1
https://arxiv.org/pdf/2211.02718v1.pdf
SAMO: Speaker Attractor Multi-Center One-Class Learning for Voice Anti-Spoofing
Voice anti-spoofing systems are crucial auxiliaries for automatic speaker verification (ASV) systems. A major challenge is caused by unseen attacks empowered by advanced speech synthesis technologies. Our previous research on one-class learning has improved the generalization ability to unseen attacks by compacting the bona fide speech in the embedding space. However, such compactness lacks consideration of the diversity of speakers. In this work, we propose speaker attractor multi-center one-class learning (SAMO), which clusters bona fide speech around a number of speaker attractors and pushes away spoofing attacks from all the attractors in a high-dimensional embedding space. For training, we propose an algorithm for the co-optimization of bona fide speech clustering and bona fide/spoof classification. For inference, we propose strategies to enable anti-spoofing for speakers without enrollment. Our proposed system outperforms existing state-of-the-art single systems with a relative improvement of 38% on equal error rate (EER) on the ASVspoof2019 LA evaluation set.
['Zhiyao Duan', 'You Zhang', 'Siwen Ding']
2022-11-04
null
null
null
null
['voice-anti-spoofing', 'speaker-verification']
['audio', 'speech']
[-2.12726370e-01 1.63982704e-01 -2.85011560e-01 -1.11239217e-01 -7.66169727e-01 -7.04230607e-01 4.96958792e-01 -1.50856301e-01 -3.81090008e-02 2.15821147e-01 5.90149760e-01 -6.30578935e-01 -1.05225988e-01 -1.03041679e-01 -3.48706275e-01 -8.39658082e-01 -1.58071369e-01 3.34048629e-01 -5.25568314e-02 -4.48751360e-01 -3.15981030e-01 7.41776526e-01 -1.50199342e+00 3.30872446e-01 7.70053625e-01 5.07042587e-01 -9.10640359e-02 9.48820293e-01 1.61150858e-01 2.56308228e-01 -1.04300201e+00 -3.61108214e-01 -1.26979336e-01 -4.08779770e-01 -6.37821734e-01 -2.88618863e-01 4.57729399e-01 1.25560313e-01 -5.76527119e-01 1.22008252e+00 8.46580327e-01 4.59972173e-02 5.58320880e-01 -1.44139528e+00 -5.39207637e-01 1.27586854e+00 1.21483672e-02 5.10284960e-01 2.88139760e-01 -3.61282118e-02 1.01548171e+00 -1.01341271e+00 1.71699047e-01 1.73048902e+00 4.78218228e-01 1.19608629e+00 -1.07180405e+00 -1.10894477e+00 1.22682557e-01 1.66063786e-01 -1.41178799e+00 -1.23528278e+00 1.32353163e+00 -2.56552875e-01 6.49018288e-01 7.53287375e-01 4.80293036e-01 1.54180050e+00 -2.77628243e-01 1.04726505e+00 6.54049754e-01 -5.33981502e-01 1.46531895e-01 6.05281472e-01 3.42469096e-01 3.91078532e-01 2.43135784e-02 3.86900306e-01 -7.61261463e-01 -3.08917373e-01 -5.60296141e-02 -3.22638363e-01 -4.96024936e-01 -6.49783760e-02 -1.03433323e+00 9.98000503e-01 2.75820255e-01 5.96014619e-01 6.49632700e-03 -3.96135747e-01 3.69454116e-01 2.88430125e-01 2.16341838e-01 5.43063819e-01 -3.12824935e-01 -4.24550697e-02 -1.00903881e+00 -2.27612764e-01 8.08089495e-01 5.55223346e-01 1.55621871e-01 6.62539721e-01 1.09825306e-01 9.53441978e-01 6.43296838e-01 7.65681326e-01 9.47850704e-01 -3.75914901e-01 4.69330519e-01 2.21472904e-01 -2.78059959e-01 -8.52531791e-01 -1.89113572e-01 -6.54623747e-01 -8.58253956e-01 -1.48399144e-01 1.59045741e-01 -2.50608057e-01 -6.68396413e-01 1.83920896e+00 3.56705219e-01 4.57319051e-01 4.86412048e-01 5.25815189e-01 6.45507514e-01 7.84701765e-01 -3.04241776e-01 -4.69420224e-01 1.23084629e+00 -9.61802483e-01 -1.16840732e+00 -9.38541070e-02 6.72265947e-01 -9.37602282e-01 1.19834208e+00 3.39377582e-01 -5.61682761e-01 -5.57389498e-01 -1.10603869e+00 5.86787581e-01 -5.20353079e-01 1.82422400e-01 1.47835299e-01 1.63772392e+00 -9.96665478e-01 2.90860891e-01 -6.27434492e-01 -7.41804913e-02 3.18594366e-01 5.42563558e-01 -2.02075005e-01 5.58513701e-01 -1.56069803e+00 5.96109152e-01 3.99184644e-01 -1.13408297e-01 -1.08431590e+00 -7.75836289e-01 -7.92525232e-01 -4.76545683e-05 -1.18053682e-01 -2.93310103e-03 9.45076048e-01 -5.33834100e-01 -1.87752211e+00 6.35487616e-01 -1.34605259e-01 -7.81152129e-01 2.20489025e-01 -1.92781731e-01 -1.24531925e+00 -2.70377677e-02 -3.53380680e-01 4.97376174e-01 1.38259709e+00 -1.19970942e+00 -1.88197196e-01 -2.55526960e-01 -6.51426494e-01 -8.57544988e-02 -1.17434680e+00 2.97678083e-01 3.11745852e-01 -9.17904496e-01 1.52663007e-01 -1.19860840e+00 3.59066993e-01 -7.19081819e-01 -6.88964903e-01 -5.98057389e-01 1.65810299e+00 -7.35718250e-01 1.77087438e+00 -2.55795765e+00 7.69144520e-02 9.72419083e-02 -1.23918273e-01 9.43604589e-01 9.75720063e-02 6.81616515e-02 -5.06500840e-01 2.60924935e-01 5.98579012e-02 -6.54244840e-01 1.19538300e-01 -8.28875508e-03 -8.11489105e-01 6.43251538e-01 -2.10070461e-02 3.08674425e-01 -9.38495278e-01 -6.15132093e-01 4.60693896e-01 8.19027543e-01 -7.12387562e-01 8.53865668e-02 3.73272896e-01 4.08646703e-01 8.75570849e-02 7.82216787e-01 6.23209655e-01 3.17281783e-01 -1.13559263e-02 1.96742997e-01 1.48667797e-01 7.05042660e-01 -1.29699469e+00 1.17482436e+00 -3.82038713e-01 7.57825613e-01 3.95666093e-01 -7.77061880e-01 1.09103489e+00 7.35773206e-01 4.54743564e-01 2.00153098e-01 2.53066272e-01 4.50083137e-01 3.04005921e-01 -8.10627565e-02 2.70935953e-01 -1.80931255e-01 -7.47650042e-02 2.81334996e-01 3.15016210e-01 1.59461722e-02 -4.51242387e-01 1.47999555e-01 6.44545376e-01 -8.04826796e-01 -5.84552735e-02 -5.06355226e-01 1.00307953e+00 -8.52504969e-01 3.89728934e-01 5.05384803e-01 -9.38073099e-01 2.09975719e-01 3.82916257e-02 9.21596028e-03 -7.05764592e-01 -1.15597701e+00 -4.33047384e-01 1.02307749e+00 -2.20993146e-01 -6.03897154e-01 -1.10263443e+00 -6.84577823e-01 -8.29571709e-02 8.87099087e-01 -2.04804778e-01 -5.08351862e-01 -8.99191678e-01 -5.88702202e-01 1.18884242e+00 6.52468726e-02 4.44948561e-02 -1.12406528e+00 2.88409051e-02 1.50015950e-01 -4.36655283e-01 -1.10037851e+00 -7.93176651e-01 1.52752325e-01 -4.89117682e-01 -5.08273304e-01 -6.96140230e-01 -1.01137578e+00 1.99728265e-01 2.13449538e-01 6.18116319e-01 -2.43636578e-01 1.46219015e-01 1.56658888e-01 -1.35788366e-01 -6.50649011e-01 -1.21102250e+00 3.49668592e-01 1.11254144e+00 5.69069326e-01 3.17985684e-01 -3.35736454e-01 -5.56399003e-02 4.39596564e-01 -4.99076128e-01 -5.93853116e-01 4.61953180e-03 9.49020863e-01 2.56583896e-02 2.68434435e-01 9.33079660e-01 -2.54797727e-01 6.57075465e-01 -3.24425370e-01 -3.09504658e-01 1.26233697e-01 -4.96096104e-01 1.51041463e-01 6.20611787e-01 -9.78683114e-01 -6.14152730e-01 1.80978596e-01 -3.03053170e-01 -8.16865206e-01 -5.22124730e-02 -3.22319508e-01 -6.78041458e-01 1.18260197e-01 7.88810670e-01 2.98654377e-01 -2.79673971e-02 -4.72834259e-01 4.71637607e-01 1.42275178e+00 5.34959078e-01 -1.59538344e-01 9.48203385e-01 1.65842518e-01 -5.25731564e-01 -1.55905080e+00 -4.48025584e-01 -7.76325583e-01 -5.61288178e-01 -2.66398311e-01 6.50481224e-01 -8.09249163e-01 -1.08629918e+00 6.19794130e-01 -1.17819190e+00 1.78689837e-01 -1.98926285e-01 6.49340928e-01 -2.93315977e-01 4.36887324e-01 -4.03778344e-01 -1.37847781e+00 -4.88399655e-01 -1.16844237e+00 7.47702539e-01 1.02043226e-01 -4.94653374e-01 -8.63783538e-01 2.02314973e-01 4.23519522e-01 4.50590223e-01 -2.87194610e-01 4.90726948e-01 -1.26361525e+00 5.54113090e-02 -2.72105068e-01 6.71495974e-01 6.02913022e-01 3.13431799e-01 -9.29149538e-02 -1.53046679e+00 -7.02933550e-01 3.12328637e-01 3.03179137e-02 7.77982295e-01 3.38254839e-01 9.37050045e-01 -8.81365001e-01 -5.06464243e-01 6.26902759e-01 5.35627544e-01 1.74794704e-01 2.59787261e-01 -1.94690730e-02 7.08769143e-01 6.76261067e-01 1.52857155e-01 2.12279096e-01 -9.85828191e-02 7.43331254e-01 3.41981381e-01 1.98729381e-01 -4.42919433e-01 -5.42631567e-01 1.01364863e+00 1.52193260e+00 6.48267388e-01 -3.43924373e-01 -7.82558203e-01 8.58053505e-01 -1.18977177e+00 -1.18203449e+00 2.79946059e-01 2.25356650e+00 8.12119484e-01 2.16821551e-01 4.29840535e-01 9.57801461e-01 1.20443416e+00 4.48303074e-01 -3.11760038e-01 -4.67245281e-01 -3.43598485e-01 -6.85622618e-02 2.24959388e-01 1.05640829e+00 -1.43655705e+00 1.16063058e+00 5.74511147e+00 9.86594558e-01 -1.44078612e+00 3.17611903e-01 3.41997176e-01 -4.71079983e-02 -1.77007109e-01 -3.13715279e-01 -1.39663804e+00 7.22070932e-01 1.61222482e+00 3.03602889e-02 5.86290359e-01 9.97500241e-01 -1.63370967e-02 1.06788814e+00 -9.25313115e-01 1.15017581e+00 3.67766023e-01 -1.13513744e+00 4.54926267e-02 1.54439852e-01 5.48426330e-01 -4.80391979e-02 4.84464556e-01 2.73668885e-01 1.88524529e-01 -1.00649714e+00 7.60644972e-01 -1.58148646e-01 6.88169241e-01 -7.69911647e-01 3.12968105e-01 3.90915781e-01 -1.07090878e+00 -5.79610705e-01 -1.20482527e-01 6.01840317e-01 4.81843986e-02 2.93353379e-01 -1.42649245e+00 5.84134497e-02 5.15950084e-01 3.60086799e-01 -3.96191627e-01 4.90708411e-01 -1.26131177e-01 1.05202448e+00 -4.79852021e-01 -2.62934923e-01 -8.84533394e-03 3.84420633e-01 1.19160807e+00 1.43057024e+00 2.23156527e-01 -3.17004919e-01 -8.31195414e-02 4.55601841e-01 -1.52313262e-01 6.11586459e-02 -7.40301073e-01 -2.85506129e-01 1.01741111e+00 8.44506919e-01 -2.41922349e-01 -2.49171287e-01 3.64628017e-01 7.75699615e-01 1.20534850e-02 1.59964636e-01 -7.56262243e-01 -5.81819892e-01 9.34355497e-01 2.05188125e-01 1.52219340e-01 -6.25218600e-02 -2.42037803e-01 -1.20312083e+00 -2.95562059e-01 -1.26081347e+00 4.96641725e-01 2.42256131e-02 -9.66351092e-01 9.15671408e-01 -4.44518209e-01 -1.18127584e+00 -2.63853431e-01 -5.27939439e-01 -4.51862454e-01 7.20746100e-01 -1.32950270e+00 -1.05139434e+00 4.00098562e-01 8.95349979e-01 7.01529682e-01 -9.52611327e-01 9.90797758e-01 3.99204671e-01 -7.56905258e-01 1.27617228e+00 6.27709255e-02 3.50644946e-01 6.79094315e-01 -1.03184652e+00 4.95492339e-01 1.07498777e+00 6.76638007e-01 7.46176004e-01 7.33729839e-01 -3.45071673e-01 -1.20998633e+00 -8.95528376e-01 1.31785440e+00 -7.09684074e-01 7.95939326e-01 -6.84639394e-01 -8.26683819e-01 4.92256254e-01 1.59649458e-02 1.67817339e-01 8.30646038e-01 1.45151898e-01 -4.59958643e-01 -2.84158319e-01 -1.04445291e+00 5.48154294e-01 6.88592613e-01 -1.02198803e+00 -5.92084169e-01 4.29206550e-01 1.07205999e+00 -7.94506967e-02 -4.59251404e-01 9.56170559e-02 2.66222119e-01 -6.79454625e-01 1.35454726e+00 -8.96657586e-01 -4.89661276e-01 -2.55396545e-01 -3.77842844e-01 -1.38772976e+00 -1.80739447e-01 -1.50802302e+00 -6.21492207e-01 1.45950925e+00 3.80686074e-01 -6.22124910e-01 5.92690885e-01 -1.68853313e-01 -3.29517394e-01 -2.69737035e-01 -1.44508636e+00 -1.09372067e+00 2.23235279e-01 -3.88209879e-01 7.09011793e-01 1.24392152e+00 2.82191843e-01 4.82091695e-01 -5.84949195e-01 7.70821214e-01 7.26988792e-01 -4.99374539e-01 6.11997843e-01 -1.20156682e+00 -1.66328058e-01 -6.57862365e-01 -3.62209648e-01 -8.86187851e-01 5.54036260e-01 -1.15662026e+00 -1.60688490e-01 -4.43747520e-01 -4.13371414e-01 -4.61409569e-01 -5.75222850e-01 1.69869483e-01 -1.21066868e-02 -2.25393638e-01 3.97697538e-01 1.93693876e-01 -1.90037191e-01 6.03367746e-01 7.95738041e-01 -3.74547809e-01 -6.40884101e-01 4.60120380e-01 -4.75457966e-01 7.17009664e-01 8.54858458e-01 -6.92113638e-01 -2.61448234e-01 1.01221241e-01 -6.61060631e-01 2.95128543e-02 5.80293611e-02 -1.19377732e+00 2.15383053e-01 2.39280224e-01 -6.05109893e-02 -6.47805393e-01 6.49833858e-01 -7.11967647e-01 -2.05618590e-01 9.86382008e-01 -4.99515414e-01 -2.91556776e-01 9.05761570e-02 4.80864972e-01 -9.18294564e-02 -1.02537081e-01 1.31349468e+00 4.38298434e-01 3.66625860e-02 -1.38851432e-02 -2.83441484e-01 -1.69922318e-02 9.39380825e-01 1.42669380e-01 -2.95939773e-01 -3.87991756e-01 -8.90334249e-01 -2.76306868e-01 -5.07330522e-02 8.32754016e-01 5.90654671e-01 -1.41386569e+00 -8.52569461e-01 8.18871200e-01 -1.51998445e-01 -6.98709607e-01 1.45473465e-01 4.16220218e-01 -1.12270154e-02 8.28829706e-01 1.38536409e-01 -8.16519976e-01 -1.66487086e+00 8.15972328e-01 4.98557270e-01 -7.24973306e-02 -3.47294182e-01 1.14163649e+00 -1.60670146e-01 -7.39692032e-01 6.12630546e-01 -1.02534860e-01 -2.52267271e-01 2.23671168e-01 5.54506898e-01 5.56036770e-01 7.38804787e-02 -1.30084980e+00 -8.18327963e-01 1.47323430e-01 -1.24884315e-01 -2.00547978e-01 9.55059290e-01 -2.60295011e-02 2.70011872e-01 4.44805354e-01 1.42424607e+00 6.13032281e-01 -7.92768180e-01 -1.26015633e-01 -5.64522110e-02 -2.89765269e-01 1.39695585e-01 -4.79874194e-01 -6.73895597e-01 1.28647542e+00 9.26915050e-01 5.67622900e-01 6.13785386e-01 1.25990063e-01 1.00358605e+00 1.78074777e-01 -1.42643884e-01 -1.17097592e+00 3.77657712e-01 5.10738194e-01 9.05409515e-01 -1.14506078e+00 -5.67865908e-01 -2.47578040e-01 -6.20862007e-01 8.81412208e-01 2.59939581e-01 1.98675752e-01 1.11547291e+00 1.79464236e-01 3.33019495e-01 1.32871717e-01 -3.73035938e-01 2.60443747e-01 3.72608602e-01 6.99224114e-01 5.38759865e-02 6.17975354e-01 3.55170101e-01 6.46608472e-01 -9.33574378e-01 -1.09089756e+00 3.06927055e-01 3.61105591e-01 -5.79775095e-01 -9.06486988e-01 -9.02051687e-01 -1.92142352e-01 -5.19178331e-01 -1.71436500e-02 -5.26032984e-01 2.36667812e-01 -5.37785590e-02 1.44736350e+00 -1.11073568e-01 -7.41783082e-01 8.12258795e-02 4.81881052e-01 -7.23607391e-02 -3.34504962e-01 -4.96740937e-01 1.98553815e-01 -1.49025410e-01 8.71499404e-02 -4.93380167e-02 -8.20218682e-01 -9.46826398e-01 -2.91520447e-01 -7.85393476e-01 3.66943210e-01 9.90934253e-01 6.18788779e-01 3.59614164e-01 4.89129752e-01 1.30161321e+00 -4.67437118e-01 -1.11200857e+00 -1.13154995e+00 -3.31858844e-01 3.85963082e-01 1.09101510e+00 -4.69929397e-01 -9.24885750e-01 -1.74855422e-02]
[14.085243225097656, 5.900129318237305]
33537b04-e104-445e-b3c3-66b6fef81bfc
hawkes-process-based-on-controlled
2305.07031
null
https://arxiv.org/abs/2305.07031v2
https://arxiv.org/pdf/2305.07031v2.pdf
Hawkes Process Based on Controlled Differential Equations
Hawkes processes are a popular framework to model the occurrence of sequential events, i.e., occurrence dynamics, in several fields such as social diffusion. In real-world scenarios, the inter-arrival time among events is irregular. However, existing neural network-based Hawkes process models not only i) fail to capture such complicated irregular dynamics, but also ii) resort to heuristics to calculate the log-likelihood of events since they are mostly based on neural networks designed for regular discrete inputs. To this end, we present the concept of Hawkes process based on controlled differential equations (HP-CDE), by adopting the neural controlled differential equation (neural CDE) technology which is an analogue to continuous RNNs. Since HP-CDE continuously reads data, i) irregular time-series datasets can be properly treated preserving their uneven temporal spaces, and ii) the log-likelihood can be exactly computed. Moreover, as both Hawkes processes and neural CDEs are first developed to model complicated human behavioral dynamics, neural CDE-based Hawkes processes are successful in modeling such occurrence dynamics. In our experiments with 4 real-world datasets, our method outperforms existing methods by non-trivial margins.
['Noseong Park', 'Seungji Kook', 'Minju Jo']
2023-05-09
null
null
null
null
['irregular-time-series']
['time-series']
[-9.35879201e-02 -1.93113878e-01 9.63642001e-02 2.70110399e-01 -3.41023654e-02 -1.68442726e-01 8.56761873e-01 3.32874715e-01 -4.21838671e-01 6.39075637e-01 4.17351127e-02 -3.02809715e-01 -3.88614684e-01 -1.22044718e+00 -6.74918354e-01 -7.91294336e-01 -4.56133068e-01 6.22751594e-01 2.95975924e-01 -1.55070171e-01 -2.18638688e-01 1.65058598e-01 -1.37949502e+00 -4.83784825e-01 7.83593476e-01 7.68026173e-01 -3.41243558e-02 8.46558750e-01 -2.01740950e-01 9.66666043e-01 -4.49587882e-01 -1.97756961e-01 1.38888314e-01 -3.70416939e-01 -3.48111540e-02 -3.93094420e-01 -5.72558105e-01 -1.78638041e-01 -6.20699286e-01 8.97316337e-01 2.94660389e-01 4.31917638e-01 1.02048647e+00 -1.47081757e+00 -1.09626901e+00 4.42976892e-01 -5.79447567e-01 2.53368944e-01 -1.44897267e-01 2.15003118e-01 7.27355480e-01 -7.08220243e-01 2.10728496e-01 1.09774256e+00 1.21725881e+00 4.40481246e-01 -1.19190228e+00 -3.10313374e-01 1.96623772e-01 4.82566506e-02 -1.36254883e+00 2.75343746e-01 6.58272564e-01 -4.94498432e-01 8.83514941e-01 6.97464496e-02 9.81486499e-01 1.57423437e+00 7.57590652e-01 9.07756567e-01 8.19243371e-01 -1.00050652e-02 5.82428098e-01 -4.23698902e-01 2.33366519e-01 1.17266782e-01 1.25356838e-01 1.74382403e-01 -2.44695514e-01 -3.63194078e-01 1.11940634e+00 8.53738844e-01 6.28644899e-02 1.16794132e-01 -1.23909891e+00 8.22840512e-01 2.12497354e-01 4.36380059e-01 -1.12826037e+00 4.43042427e-01 3.20082754e-01 1.51285306e-01 7.62044549e-01 -9.05077755e-02 -2.25902185e-01 -3.37315381e-01 -8.42605650e-01 3.98244768e-01 1.03796470e+00 8.42855453e-01 3.03853214e-01 4.73221689e-02 -5.24737060e-01 6.18985295e-01 1.99788451e-01 8.66837621e-01 4.31858122e-01 -5.38742244e-01 2.09293976e-01 3.36226404e-01 3.41001451e-01 -1.32429779e+00 -5.14186382e-01 -2.03779325e-01 -1.79632866e+00 -4.58314359e-01 5.58406234e-01 -4.63252813e-01 -9.25273418e-01 1.84482026e+00 1.89659640e-01 6.56721830e-01 -9.89422277e-02 2.75249541e-01 2.66102284e-01 1.43148136e+00 1.73685327e-01 -5.64965010e-01 9.70979691e-01 -6.95378006e-01 -1.05643022e+00 3.04756880e-01 1.25850067e-01 -2.10215434e-01 8.22160780e-01 2.79309243e-01 -1.19928658e+00 -3.76285434e-01 -3.29799026e-01 2.71905750e-01 -4.88107741e-01 -3.62021178e-01 2.74323672e-01 2.30452359e-01 -1.28502405e+00 5.95105231e-01 -1.11468697e+00 -3.54128927e-01 9.64495838e-02 1.00611605e-01 2.28531599e-01 3.18440348e-01 -1.55671048e+00 5.69542944e-01 2.04495311e-01 3.43748987e-01 -1.08419466e+00 -7.40477026e-01 -4.84228969e-01 3.62191916e-01 4.92383748e-01 -5.07147431e-01 1.15241909e+00 -7.13651538e-01 -1.38870478e+00 2.31762946e-01 -1.62436351e-01 -8.23326826e-01 6.92799389e-01 -2.56621074e-02 -5.86532295e-01 2.94618253e-02 -6.19333386e-02 1.43718183e-01 1.07366288e+00 -9.92037296e-01 -4.66327846e-01 5.38777672e-02 -4.23873067e-01 -1.67105585e-01 -4.62000787e-01 -1.81749821e-01 -2.51078010e-01 -9.43855345e-01 -3.53065401e-01 -1.17982626e+00 -3.55086356e-01 5.76417223e-02 -3.26967090e-01 -7.47985899e-01 5.81579149e-01 -6.41521454e-01 1.53648472e+00 -2.02393484e+00 -6.54559210e-02 3.01572174e-01 5.34454048e-01 4.51400429e-01 1.03727225e-02 1.07372749e+00 2.80991644e-01 1.22753128e-01 -4.54107881e-01 -4.93986070e-01 6.16989844e-02 5.18474162e-01 -5.66383421e-01 4.57846880e-01 1.60202593e-01 9.87554431e-01 -1.03447580e+00 -7.66151473e-02 3.36391702e-02 5.98738968e-01 -2.21367106e-01 5.38306125e-02 -2.24996313e-01 2.60220975e-01 -2.83837497e-01 3.52433890e-01 4.75457072e-01 -4.96174365e-01 -2.66153812e-01 7.22831190e-01 -2.58468300e-01 -4.12918031e-01 -1.11248493e+00 8.26746464e-01 -5.83204627e-01 6.61764741e-01 -2.90309608e-01 -8.32609773e-01 8.92811060e-01 5.63226640e-01 7.26222932e-01 -3.51550817e-01 1.13161154e-01 7.77280182e-02 -1.58965662e-01 -2.33303532e-01 6.65602922e-01 -2.30746388e-01 -6.23485707e-02 6.65466964e-01 -2.55893826e-01 3.06485474e-01 1.95917740e-01 -4.77964766e-02 1.30869925e+00 -2.06320524e-01 3.34252059e-01 -1.76141724e-01 1.91706106e-01 -2.71905005e-01 6.47817016e-01 1.05475843e+00 -2.82040685e-01 5.77024221e-01 6.81216896e-01 -5.94693005e-01 -1.24819708e+00 -1.31871295e+00 3.33120972e-02 7.43224323e-01 1.19733222e-01 -9.71161872e-02 -6.87157750e-01 -5.92470765e-02 -5.69017753e-02 7.16313481e-01 -9.54261422e-01 -1.85924873e-01 -4.68172938e-01 -1.08523297e+00 6.39389455e-01 5.59063554e-01 4.16949481e-01 -1.47995877e+00 -5.91882825e-01 7.74966538e-01 -9.10324007e-02 -6.70292914e-01 -4.82386321e-01 2.63280459e-02 -5.58387637e-01 -8.19793940e-01 -1.23825860e+00 -5.31227648e-01 3.93729627e-01 1.59985766e-01 9.35661495e-01 -1.05469432e-02 5.77570684e-02 5.14434993e-01 -2.08924130e-01 -7.47220337e-01 -4.55718815e-01 5.39338998e-02 2.83513069e-01 3.97169083e-01 5.90337873e-01 -8.54258180e-01 -7.74214506e-01 2.42055103e-01 -1.46331155e+00 -2.10866362e-01 3.63994956e-01 8.05158675e-01 6.26041353e-01 4.41124231e-01 8.09089541e-01 -4.75892127e-01 1.20281482e+00 -9.93165672e-01 -4.65531588e-01 3.01330954e-01 -5.67121863e-01 -2.34009475e-01 8.70414376e-01 -1.19614434e+00 -9.40411627e-01 -4.24489558e-01 6.94625601e-02 -6.18476570e-01 -1.77098036e-01 7.19190776e-01 5.71783006e-01 7.64457643e-01 4.09453422e-01 6.44651711e-01 -1.78364068e-02 1.13397576e-02 1.71423420e-01 5.41768312e-01 4.03170139e-01 -4.11244094e-01 6.80986464e-01 7.00807393e-01 1.28183678e-01 -1.02871931e+00 -2.86595166e-01 -5.92201591e-01 -3.16035777e-01 -3.43589932e-01 1.01468158e+00 -7.66849399e-01 -1.10245121e+00 1.19081700e+00 -1.26869738e+00 -6.05360091e-01 -5.17867565e-01 4.44466203e-01 -6.57445729e-01 5.05173206e-02 -1.23420131e+00 -1.48964930e+00 -2.21338078e-01 -5.17573655e-01 9.53824162e-01 3.21408242e-01 -3.16750348e-01 -1.60631192e+00 5.21237731e-01 -6.56860113e-01 6.03992641e-01 4.91736770e-01 8.17951620e-01 -6.66922987e-01 -2.58985460e-01 -5.06983757e-01 -3.64578702e-02 1.91628873e-01 7.49889389e-02 3.48635882e-01 -5.38080156e-01 9.86706540e-02 1.56018823e-01 3.72513771e-01 4.98562962e-01 9.58463907e-01 9.52175438e-01 -3.01679730e-01 -2.90375739e-01 2.66088009e-01 1.32509816e+00 3.33351076e-01 7.42388368e-01 1.31588936e-01 6.25053644e-01 8.08290839e-01 8.09264481e-02 7.25420773e-01 6.13645256e-01 1.30724937e-01 3.65218252e-01 -2.94702034e-02 6.48977041e-01 -8.26853096e-01 4.01274472e-01 1.18260288e+00 -2.70096213e-01 -6.24853134e-01 -1.09496772e+00 8.41786981e-01 -2.29865336e+00 -1.34004557e+00 -7.71826446e-01 2.13361168e+00 6.47232771e-01 6.97679594e-02 4.92227674e-01 -7.38359243e-02 9.47737336e-01 2.17774749e-01 -7.35574841e-01 -1.71210200e-01 -3.56823772e-01 -1.11485250e-01 6.26486599e-01 1.78756833e-01 -1.03335762e+00 6.37148261e-01 6.23293543e+00 8.53218496e-01 -8.64982784e-01 1.88019842e-01 5.57103574e-01 -1.08110951e-03 -1.06817819e-02 -3.74058843e-01 -7.06941187e-01 9.81396437e-01 1.32572627e+00 -4.16187674e-01 3.85510415e-01 5.06015956e-01 6.01189017e-01 1.31840110e-01 -7.75611520e-01 9.93101060e-01 -3.50972414e-01 -9.45967317e-01 -1.12328105e-01 2.59443879e-01 9.03687954e-01 -1.02142885e-01 3.31221253e-01 4.89073783e-01 7.34764218e-01 -9.65997696e-01 5.35127223e-01 1.09237909e+00 2.05716833e-01 -6.87453151e-01 6.24873459e-01 8.77609670e-01 -1.13292694e+00 -1.26787096e-01 -4.52105105e-01 -2.98471451e-01 7.90786684e-01 8.97719502e-01 -2.41407916e-01 1.36116147e-01 7.19540000e-01 8.98060262e-01 3.99206653e-02 1.03821909e+00 -1.50337920e-01 9.46499944e-01 -6.87227666e-01 -4.44559991e-01 4.82701480e-01 -4.46278185e-01 7.34825969e-01 9.52016294e-01 7.29527116e-01 -1.25321701e-01 -4.08850238e-02 9.66614962e-01 2.76656926e-01 -9.49911550e-02 -7.39876509e-01 -1.88680679e-01 2.73804307e-01 9.34668839e-01 -9.56035554e-01 -3.50067139e-01 -3.58460605e-01 9.10023510e-01 -2.94558946e-02 7.45559454e-01 -1.04858613e+00 -6.83653280e-02 6.28538787e-01 1.55830637e-01 2.43211105e-01 -5.53144813e-01 2.38182358e-02 -1.16367674e+00 -8.27477351e-02 -3.67871076e-01 1.57651201e-01 -6.60012066e-01 -1.96913779e+00 5.94441593e-01 1.29766539e-01 -1.23614454e+00 -4.39562261e-01 -3.97213459e-01 -9.74662364e-01 8.05870414e-01 -1.19023907e+00 -9.04160619e-01 1.06087752e-01 8.91734183e-01 4.63352293e-01 2.48448685e-01 5.31912863e-01 2.09586456e-01 -6.74223721e-01 -1.07444212e-01 5.65377593e-01 -9.36144888e-02 3.04926932e-01 -1.35783124e+00 8.16014767e-01 7.86691904e-01 5.27890632e-03 7.29972780e-01 7.96523988e-01 -1.05067098e+00 -1.02364969e+00 -1.19568777e+00 8.54200780e-01 -5.15208781e-01 1.13609564e+00 -4.29839820e-01 -1.39608121e+00 5.37128806e-01 1.45859927e-01 -2.01715037e-01 5.72593570e-01 -6.91180900e-02 2.05516204e-01 2.68873304e-01 -7.85315871e-01 8.05163324e-01 9.35646176e-01 -3.26778591e-01 -4.57858413e-01 3.95467639e-01 6.84150457e-01 7.99820498e-02 -7.70064771e-01 9.49546471e-02 2.66668111e-01 -6.55837595e-01 8.26908171e-01 -6.05885744e-01 5.21498799e-01 -2.42063984e-01 2.11141810e-01 -1.58280230e+00 -3.92791003e-01 -9.14426982e-01 -5.78184664e-01 9.92966771e-01 1.95893213e-01 -9.08576429e-01 2.65725911e-01 6.75816834e-01 3.64204526e-01 -6.53016508e-01 -9.38660324e-01 -1.03292036e+00 1.93108678e-01 -5.57595730e-01 6.41798735e-01 8.09895933e-01 -4.09103423e-01 -1.68298054e-02 -1.04497302e+00 1.11412570e-01 6.32614791e-01 -4.30745035e-01 5.40247142e-01 -1.57844472e+00 -2.94782847e-01 -6.12201691e-01 -6.16740175e-02 -1.10934556e+00 8.82116333e-02 -2.48233005e-01 3.99303854e-01 -1.51797163e+00 1.79046858e-02 -2.45306358e-01 -2.55935133e-01 5.33352699e-03 -3.48157644e-01 -1.95960596e-01 5.94007000e-02 6.36993825e-01 -4.30110782e-01 9.48400974e-01 8.57560039e-01 2.27659762e-01 -5.81803203e-01 4.08516735e-01 -1.13383070e-01 7.70808816e-01 8.31190109e-01 -5.00515461e-01 -5.01796305e-01 -9.68659893e-02 6.46572471e-01 2.59923130e-01 5.89844823e-01 -8.34419787e-01 6.34080410e-01 -2.75723696e-01 -1.82445906e-02 -7.33629763e-01 2.96585828e-01 -8.47932041e-01 3.51425886e-01 4.90185976e-01 -3.50426495e-01 5.53419352e-01 -5.78928972e-03 1.28279221e+00 -3.28589857e-01 1.25060342e-02 2.70654142e-01 -9.77902710e-02 -3.15307498e-01 8.76231432e-01 -1.09977996e+00 6.30984306e-02 1.23042846e+00 4.27453109e-04 -1.18730500e-01 -1.01798475e+00 -7.73749709e-01 3.22166860e-01 1.31391317e-01 3.71836275e-01 4.06168431e-01 -1.33088326e+00 -4.00607437e-01 -4.76507656e-02 -2.32391894e-01 2.56740928e-01 6.25094771e-01 1.06700408e+00 -4.94338304e-01 2.12188900e-01 2.01324597e-01 -5.71808338e-01 -6.76085889e-01 8.61802578e-01 2.15517193e-01 -5.93481600e-01 -8.01796019e-01 3.17911029e-01 3.67993385e-01 -4.46518332e-01 2.07904428e-01 -5.93857527e-01 -1.23697333e-01 3.27555329e-01 6.90158129e-01 5.97074151e-01 -5.10883689e-01 -5.03032923e-01 1.55886993e-01 3.09956938e-01 3.98576200e-01 -2.92626023e-01 1.54949796e+00 -3.34060550e-01 -1.02465391e-01 1.22452533e+00 8.23046923e-01 -5.58743596e-01 -1.57090175e+00 -4.70629215e-01 8.06350857e-02 1.85976654e-01 -3.22329909e-01 -8.27828124e-02 -6.11901581e-01 1.05871534e+00 2.18101770e-01 1.26326203e+00 1.02261364e+00 -3.15673113e-01 1.21720254e+00 1.37875289e-01 2.95504093e-01 -1.16785121e+00 -9.57602188e-02 7.66291440e-01 6.36583269e-01 -7.45184898e-01 -8.07289660e-01 -6.67709634e-02 -7.06812620e-01 9.83183801e-01 1.77272260e-01 -5.12009919e-01 1.29558492e+00 2.38244459e-01 -3.27329278e-01 -2.72317827e-02 -9.57707882e-01 -1.03836499e-01 -3.91876064e-02 5.54603338e-01 -7.21967518e-02 2.64978349e-01 -1.76150605e-01 5.78900576e-01 4.05401662e-02 2.54618734e-01 6.14829421e-01 7.69260168e-01 -1.38708234e-01 -4.68631953e-01 -4.90358531e-01 3.90681893e-01 -4.87305820e-01 -2.64714271e-01 -5.79711720e-02 7.11713076e-01 -1.95105970e-01 8.30223024e-01 4.53199267e-01 -2.53365375e-03 2.71359682e-01 -4.03194390e-02 -1.67959735e-01 -2.04418004e-01 -5.16316533e-01 -8.66796151e-02 -6.08968079e-01 -4.54926267e-02 -3.54259849e-01 -7.22505629e-01 -9.44307745e-01 -7.30013669e-01 3.05428114e-02 -1.84569776e-01 3.71299326e-01 9.49669063e-01 2.43298993e-01 5.88376224e-01 4.12944496e-01 -7.60874331e-01 -6.92427874e-01 -1.03407884e+00 -9.71714973e-01 3.57135415e-01 3.83203089e-01 -6.27430499e-01 -5.30854702e-01 -3.60382684e-02]
[6.906551837921143, 3.450338840484619]
fe942554-777c-4980-8704-812d3840eb07
simulating-liquids-with-graph-networks
2203.07895
null
https://arxiv.org/abs/2203.07895v1
https://arxiv.org/pdf/2203.07895v1.pdf
Simulating Liquids with Graph Networks
Simulating complex dynamics like fluids with traditional simulators is computationally challenging. Deep learning models have been proposed as an efficient alternative, extending or replacing parts of traditional simulators. We investigate graph neural networks (GNNs) for learning fluid dynamics and find that their generalization capability is more limited than previous works would suggest. We also challenge the current practice of adding random noise to the network inputs in order to improve its generalization capability and simulation stability. We find that inserting the real data distribution, e.g. by unrolling multiple simulation steps, improves accuracy and that hiding all domain-specific features from the learning model improves generalization. Our results indicate that learning models, such as GNNs, fail to learn the exact underlying dynamics unless the training set is devoid of any other problem-specific correlations that could be used as shortcuts.
['Nils Thuerey', 'Philipp Holl', 'Jonathan Klimesch']
2022-03-14
null
null
null
null
['liquid-simulation', 'physical-simulations']
['miscellaneous', 'miscellaneous']
[-2.92661458e-01 -9.61473659e-02 2.59807289e-01 -6.48521408e-02 3.47405434e-01 -6.96484745e-01 6.96993649e-01 7.81887099e-02 -4.84082639e-01 9.41148579e-01 -1.90423265e-01 -8.33565533e-01 -2.08846867e-01 -9.64898467e-01 -9.54628050e-01 -8.02815378e-01 -6.94084287e-01 7.16174603e-01 4.63470638e-01 -7.10204720e-01 1.19880192e-01 8.10034275e-01 -1.43077576e+00 -9.44167972e-02 9.50948060e-01 3.29789788e-01 -2.37155594e-02 1.10644710e+00 -2.54196435e-01 7.35690653e-01 -5.77407420e-01 5.94992377e-02 3.04726809e-01 -5.00096262e-01 -5.68700552e-01 -4.81474817e-01 3.72085482e-01 -2.70836532e-01 -7.73793995e-01 9.50021982e-01 3.57633322e-01 4.09241736e-01 6.08202159e-01 -1.31608653e+00 -2.69621402e-01 5.22574663e-01 -1.68853313e-01 3.55711907e-01 -2.11158469e-01 7.15467095e-01 4.47572827e-01 -2.85422534e-01 6.08329892e-01 1.20318210e+00 1.10770428e+00 7.74495900e-01 -1.25719786e+00 -8.00253570e-01 1.56586915e-01 -1.14580721e-01 -1.10059810e+00 1.14594989e-01 6.71246052e-01 -4.23091590e-01 1.16534305e+00 -1.65443197e-02 9.16173577e-01 1.32709789e+00 6.52592719e-01 2.67299235e-01 1.00032592e+00 -1.92629859e-01 4.09182847e-01 1.60690203e-01 1.87697083e-01 8.19657743e-01 8.24128687e-01 6.64171576e-01 -1.44046694e-01 -2.96100467e-01 1.03520405e+00 -1.26101449e-01 -1.84180796e-01 -6.33057714e-01 -6.12607300e-01 7.64549017e-01 5.74163914e-01 3.63045275e-01 -4.07698676e-02 6.53302252e-01 5.38187504e-01 6.98577940e-01 1.37652650e-01 9.34351206e-01 -6.87175214e-01 -7.45055452e-02 -6.85558975e-01 6.09514832e-01 1.26229918e+00 5.55623829e-01 1.03442729e+00 6.36458874e-01 3.22457045e-01 1.28885508e-01 -9.26307663e-02 3.45917404e-01 3.07807475e-01 -9.14663494e-01 7.30740428e-02 4.92411494e-01 2.95550730e-02 -9.76885974e-01 -8.50196064e-01 -6.97826624e-01 -1.03698087e+00 5.15193403e-01 7.07912564e-01 -7.70264328e-01 -1.14338899e+00 1.69881630e+00 1.83889821e-01 6.08774066e-01 6.17222637e-02 8.19582105e-01 5.06712615e-01 6.20024085e-01 9.51118842e-02 3.54894340e-01 5.78209817e-01 -6.71789050e-01 -2.38015682e-01 -9.59343165e-02 9.25321937e-01 -2.65475154e-01 1.12921882e+00 4.46001470e-01 -8.87127042e-01 -5.00258327e-01 -1.22331429e+00 2.75829077e-01 -6.32755637e-01 -7.06326604e-01 1.00295913e+00 8.84129941e-01 -1.30643213e+00 1.56494546e+00 -1.15711415e+00 -2.31861651e-01 1.59798205e-01 7.81387806e-01 -1.17962256e-01 2.21806630e-01 -1.38473022e+00 9.74472523e-01 3.27290297e-01 -1.10553280e-01 -1.05645049e+00 -1.01242304e+00 -7.16356218e-01 1.05117187e-02 2.71139205e-01 -9.38228726e-01 1.15242147e+00 -9.53649879e-01 -1.74882734e+00 2.01947644e-01 1.48280054e-01 -7.37983584e-01 5.82890093e-01 -6.49406984e-02 -2.00192451e-01 5.53013496e-02 -6.52014136e-01 3.88594300e-01 7.56985247e-01 -1.40402889e+00 -4.46284609e-03 1.51177317e-01 3.00693572e-01 -6.86785355e-02 -1.68925002e-01 -6.02544606e-01 1.06750466e-01 -4.55103099e-01 -4.29713398e-01 -1.18207967e+00 -8.38487089e-01 -2.65914440e-01 -1.68892533e-01 4.67573181e-02 9.68783855e-01 -3.19897056e-01 8.96781623e-01 -1.51227272e+00 -5.60574606e-02 6.61833882e-01 5.32089353e-01 5.80969512e-01 -2.35595405e-01 7.09374785e-01 -8.36479738e-02 2.88026899e-01 -1.24597631e-01 2.39490401e-02 -2.05442607e-02 5.76625645e-01 -1.39490008e-01 5.94366670e-01 3.02272528e-01 9.26202774e-01 -1.03912008e+00 9.56914797e-02 4.00716841e-01 5.80492258e-01 -6.72019243e-01 -3.97227183e-02 -5.48680246e-01 6.79514825e-01 -3.43847066e-01 -1.11508235e-01 6.97245777e-01 -4.05761302e-01 2.00977162e-01 4.31167334e-01 1.43689573e-01 4.70427066e-01 -1.20376170e+00 1.00789833e+00 -5.49812973e-01 6.64815545e-01 7.92632103e-02 -1.24689424e+00 8.31798732e-01 -8.41512717e-03 3.73905957e-01 -6.06908798e-01 2.32780546e-01 4.00580987e-02 7.25667953e-01 -3.69228303e-01 5.11272609e-01 -3.21923256e-01 1.10452361e-01 5.37115753e-01 1.29379570e-01 -2.49165580e-01 6.06214404e-02 1.51276842e-01 1.37377763e+00 2.41881795e-02 1.58426762e-02 -6.81656897e-01 2.01812699e-01 1.58044070e-01 2.08921626e-01 1.26029384e+00 5.21507896e-02 2.41374418e-01 6.97459459e-01 -6.08211339e-01 -1.42260814e+00 -9.66529429e-01 2.48263001e-01 8.93062234e-01 5.84831759e-02 -4.87078011e-01 -6.85896218e-01 -4.52221572e-01 2.57870078e-01 6.23260379e-01 -8.40919793e-01 -4.82277423e-01 -9.33257878e-01 -8.73957038e-01 6.93946302e-01 5.52925766e-01 1.84768245e-01 -9.26850379e-01 -5.53685427e-01 3.49703044e-01 7.98881650e-01 -6.25293314e-01 9.37776193e-02 5.80519140e-01 -1.10758650e+00 -1.02526736e+00 -4.08481658e-01 -4.06229794e-01 5.35774589e-01 -3.08188982e-02 1.21341872e+00 7.52694428e-01 -5.40083461e-02 3.49910170e-01 -6.85383901e-02 -2.19058141e-01 -7.72711337e-01 3.20068181e-01 2.82125264e-01 -7.23266125e-01 8.21753219e-02 -9.99730885e-01 -5.78870356e-01 -2.32635532e-02 -8.57747197e-01 -1.67214066e-01 3.78440261e-01 9.68372881e-01 -2.18443513e-01 2.53358841e-01 5.10526419e-01 -1.25964141e+00 9.05104935e-01 -6.70631111e-01 -6.28956676e-01 -1.91505000e-01 -7.23149359e-01 4.32555050e-01 1.33505416e+00 -7.42369652e-01 -5.17428517e-01 -2.56217062e-01 -1.72933504e-01 -5.21530271e-01 -2.09012836e-01 5.15428603e-01 2.61706114e-01 -5.11576474e-01 8.90151799e-01 3.81544419e-02 2.78305709e-01 -2.01893821e-01 2.80292127e-02 -1.94737479e-01 2.33553529e-01 -8.72771680e-01 9.50089514e-01 1.98297143e-01 4.84282166e-01 -8.08242500e-01 -2.12765619e-01 6.99366350e-03 -4.51622993e-01 -1.15455821e-01 3.28113258e-01 -4.87703085e-01 -1.09488428e+00 5.39196610e-01 -9.39880550e-01 -1.04827440e+00 -2.81966388e-01 4.11161810e-01 -3.34442168e-01 1.11481063e-01 -9.03268993e-01 -7.27095604e-01 5.77653125e-02 -1.00284159e+00 3.72360915e-01 4.82723892e-01 -3.54221493e-01 -1.62533593e+00 2.36440390e-01 -7.16395915e-01 8.08876574e-01 3.84384066e-01 9.95060146e-01 -9.10329342e-01 -5.66802502e-01 2.01190431e-02 4.70257252e-02 5.68386875e-02 -5.76611273e-02 2.93298900e-01 -7.67363191e-01 -5.58384299e-01 -8.32765549e-02 -1.31028473e-01 1.01749218e+00 3.62075955e-01 1.16019785e+00 -3.58865857e-01 -3.45581323e-01 7.17961371e-01 1.40039408e+00 1.75229117e-01 6.09747231e-01 2.19166398e-01 8.26924443e-01 3.95772845e-01 -4.27117735e-01 2.20342785e-01 1.78158268e-01 6.49854168e-02 2.43666694e-01 -2.02792615e-01 -3.06385178e-02 -3.82708520e-01 3.57442796e-01 8.54587138e-01 -1.64538696e-01 -3.26174200e-01 -1.39628792e+00 1.05145335e-01 -1.62775397e+00 -8.85029674e-01 -1.32996351e-01 1.97601044e+00 5.72649658e-01 4.65210050e-01 2.11548910e-01 -5.51877096e-02 3.45970303e-01 2.09421497e-02 -7.73887217e-01 -5.77726126e-01 -2.38914625e-03 7.58882403e-01 7.24210799e-01 7.97343373e-01 -8.76378655e-01 9.66569483e-01 7.39107180e+00 4.92567569e-01 -1.46516168e+00 -3.38862360e-01 4.66309875e-01 1.42842144e-01 -3.67433548e-01 1.26611739e-01 -7.18774438e-01 3.14193398e-01 1.42551053e+00 -2.58664668e-01 6.00776434e-01 6.35858715e-01 3.04006815e-01 -1.61913902e-01 -1.13334513e+00 2.78096378e-01 -3.30886006e-01 -1.55264246e+00 1.67468682e-01 1.27275020e-01 8.67622912e-01 1.20449856e-01 5.34917675e-02 6.45861983e-01 1.09647000e+00 -1.63516581e+00 1.17937632e-01 6.56212449e-01 2.54643619e-01 -8.26184809e-01 8.40840220e-01 4.35331851e-01 -9.79024649e-01 2.55005985e-01 -3.26647222e-01 -6.67252719e-01 -2.00092897e-01 3.59964669e-01 -8.97737980e-01 4.12423044e-01 5.36363661e-01 6.60985291e-01 -7.87683308e-01 1.11916614e+00 1.54961005e-01 1.03553128e+00 -7.14988530e-01 -5.89019060e-01 6.44158542e-01 -3.80080640e-01 5.41782200e-01 1.33019960e+00 2.07451925e-01 -1.54035285e-01 2.99171627e-01 7.56559730e-01 1.76013649e-01 -3.25690567e-01 -1.02313089e+00 -1.47743091e-01 7.07694292e-02 7.70859361e-01 -9.20346558e-01 -2.82219380e-01 -2.45615289e-01 4.91824627e-01 4.84107316e-01 6.70220196e-01 -5.43323338e-01 -3.51990491e-01 8.66314352e-01 1.76209494e-01 5.33449113e-01 -6.39997661e-01 -4.53431934e-01 -8.78556967e-01 -2.97641903e-01 -8.17857563e-01 9.89421010e-02 -4.76864815e-01 -1.37047780e+00 2.57177711e-01 -5.16299624e-03 -1.05073929e+00 -5.37162066e-01 -9.04636264e-01 -9.38547611e-01 9.92640018e-01 -1.20579422e+00 -5.28208256e-01 -3.59405912e-02 4.04688895e-01 -2.05634773e-01 -1.28017455e-01 6.72248304e-01 7.35626370e-02 -3.14914912e-01 5.65623403e-01 2.88746417e-01 1.82457179e-01 2.26098880e-01 -1.40008247e+00 6.32523298e-01 6.69957757e-01 -1.55453429e-01 1.01193070e+00 1.43932343e+00 -8.68859947e-01 -1.51879478e+00 -8.79336059e-01 1.98294252e-01 -4.61756766e-01 1.09796178e+00 -6.35940075e-01 -1.45702446e+00 4.74195361e-01 9.72559154e-02 4.07870039e-02 2.11439162e-01 2.17227817e-01 -2.88404226e-01 1.59411758e-01 -9.09987092e-01 8.83728743e-01 1.17125499e+00 -4.24555421e-01 -1.95017546e-01 5.69327967e-03 6.87193871e-01 -5.62347293e-01 -5.28379500e-01 4.08323497e-01 4.32805240e-01 -8.95980000e-01 9.08500731e-01 -1.12525713e+00 4.39588785e-01 -1.53767332e-01 3.25787991e-01 -1.71824622e+00 6.34106155e-03 -7.69763649e-01 -3.71230036e-01 6.30476773e-01 2.66289920e-01 -9.18540418e-01 1.21541405e+00 5.08333683e-01 -6.55422881e-02 -7.17485428e-01 -5.66658258e-01 -9.75620508e-01 9.31897521e-01 -4.56258863e-01 5.66215754e-01 9.99569952e-01 -2.09452674e-01 2.88470238e-01 -3.06291491e-01 2.83347934e-01 3.66963416e-01 -1.68365583e-01 9.81334865e-01 -1.23795390e+00 -5.41571617e-01 -7.31376112e-01 -4.13867325e-01 -8.85021210e-01 3.42677802e-01 -8.62924576e-01 -2.46243283e-01 -1.16821551e+00 -2.39445642e-01 -6.75061107e-01 -1.47395626e-01 1.18921295e-01 -1.58096492e-01 -2.08549872e-01 9.05092135e-02 -2.26860687e-01 -2.74045944e-01 4.45621550e-01 1.52057421e+00 2.07967773e-01 -2.61326253e-01 1.37567192e-01 -4.62430239e-01 7.11798668e-01 1.12751341e+00 -5.67171097e-01 -6.33223355e-01 -4.15401533e-02 2.36966729e-01 -3.17819305e-02 6.68868661e-01 -1.31531572e+00 4.37123060e-01 -1.90859824e-01 5.83501756e-01 -3.49036306e-01 8.64869803e-02 -6.23016059e-01 1.95720047e-01 8.94836128e-01 -2.51115978e-01 3.23163778e-01 8.04484487e-01 6.22448981e-01 1.56239942e-01 -4.41142142e-01 5.68631709e-01 -3.26917350e-01 -4.54268008e-01 3.08646590e-01 -8.54843616e-01 2.69952148e-01 4.75712031e-01 -3.79797593e-02 -3.88570338e-01 -5.61998844e-01 -6.83412790e-01 3.22743207e-01 6.98452055e-01 6.27796873e-02 4.45809215e-02 -8.27062786e-01 -2.99927741e-01 1.39727235e-01 -4.26215440e-01 -2.29936428e-02 1.47569522e-01 4.10306841e-01 -9.27918196e-01 2.31103003e-01 -3.69311899e-01 -4.91738975e-01 -7.32302606e-01 5.30976474e-01 8.32990408e-01 -6.24960124e-01 -8.61198485e-01 4.92036015e-01 -2.70541850e-02 -7.58690417e-01 1.08788900e-01 -5.51434815e-01 1.39049470e-01 -5.07758975e-01 2.31576994e-01 1.64100111e-01 1.15594126e-01 1.27686590e-01 -7.37047493e-02 3.68994266e-01 -2.41580624e-02 8.45986828e-02 1.41966414e+00 2.94972867e-01 2.29896843e-01 6.59373820e-01 1.01596582e+00 -2.41511866e-01 -1.52811217e+00 -8.53417255e-03 7.62594417e-02 -1.56728504e-03 -1.27833992e-01 -5.82456946e-01 -1.05976486e+00 1.14442301e+00 3.24009895e-01 4.09737289e-01 5.87428510e-01 -3.25772822e-01 6.68729842e-01 9.60508704e-01 4.41073984e-01 -9.75613892e-01 2.13706642e-02 1.18838751e+00 4.46426213e-01 -1.08970952e+00 7.74471834e-02 -7.69556463e-02 -8.79571885e-02 1.33894122e+00 9.65335190e-01 -9.67231154e-01 9.09628749e-01 8.74942780e-01 -1.69922188e-01 -1.69974238e-01 -9.58406568e-01 -1.12107210e-01 3.53208594e-02 8.20417345e-01 1.91206247e-01 -4.49926481e-02 1.13284685e-01 3.43487650e-01 -5.54312348e-01 2.78979365e-04 7.55459428e-01 8.44261706e-01 -4.54114407e-01 -9.26591992e-01 -1.77433137e-02 6.92285717e-01 5.41893691e-02 2.78814193e-02 -3.19130808e-01 1.32593942e+00 -4.37747650e-02 3.33691508e-01 2.33239427e-01 -6.24699235e-01 1.26721531e-01 1.18940987e-01 6.03597105e-01 -4.15245742e-01 -9.86420035e-01 -5.71185291e-01 7.59441629e-02 -4.17109787e-01 -1.25817239e-01 -4.46491778e-01 -1.47531438e+00 -1.05235052e+00 1.04218341e-01 2.32803479e-01 2.57722527e-01 9.91913497e-01 2.71345526e-01 7.04409540e-01 1.35579571e-01 -1.18244410e+00 -4.94207889e-01 -6.26736104e-01 -4.28088009e-01 2.54146308e-01 7.69986868e-01 -8.30975831e-01 -5.67762673e-01 -2.85943478e-01]
[6.454824447631836, 3.460477113723755]
bc13a816-1712-4dde-8632-344a5ff2d0ed
embracing-compact-and-robust-architectures
2305.12236
null
https://arxiv.org/abs/2305.12236v1
https://arxiv.org/pdf/2305.12236v1.pdf
Embracing Compact and Robust Architectures for Multi-Exposure Image Fusion
In recent years, deep learning-based methods have achieved remarkable progress in multi-exposure image fusion. However, existing methods rely on aligned image pairs, inevitably generating artifacts when faced with device shaking in real-world scenarios. Moreover, these learning-based methods are built on handcrafted architectures and operations by increasing network depth or width, neglecting different exposure characteristics. As a result, these direct cascaded architectures with redundant parameters fail to achieve highly effective inference time and lead to massive computation. To alleviate these issues, in this paper, we propose a search-based paradigm, involving self-alignment and detail repletion modules for robust multi-exposure image fusion. By utilizing scene relighting and deformable convolutions, the self-alignment module can accurately align images despite camera movement. Furthermore, by imposing a hardware-sensitive constraint, we introduce neural architecture search to discover compact and efficient networks, investigating effective feature representation for fusion. We realize the state-of-the-art performance in comparison to various competitive schemes, yielding a 4.02% and 29.34% improvement in PSNR for general and misaligned scenarios, respectively, while reducing inference time by 68.1%. The source code will be available at https://github.com/LiuZhu-CV/CRMEF.
['Risheng Liu', 'Xin Fan', 'Guanyao Wu', 'JinYuan Liu', 'Zhu Liu']
2023-05-20
null
null
null
null
['multi-exposure-image-fusion', 'architecture-search']
['computer-vision', 'methodology']
[ 3.34848195e-01 -6.46469653e-01 1.73172474e-01 -2.51938432e-01 -8.45742881e-01 -3.66474986e-01 3.45875710e-01 -1.78328544e-01 -4.43275034e-01 5.17969787e-01 4.62719519e-03 -8.02870691e-02 -2.19199926e-01 -7.07897365e-01 -8.09769630e-01 -9.20351684e-01 3.35917503e-01 -3.14257741e-01 3.79383117e-02 -2.93948621e-01 3.15821975e-01 4.42009568e-01 -1.50863552e+00 5.91646358e-02 9.29475546e-01 1.22324038e+00 4.73522604e-01 5.00469029e-01 2.75377035e-01 4.89299268e-01 -4.72926378e-01 -7.05882549e-01 3.70414197e-01 -2.92613864e-01 -3.59811872e-01 1.41663685e-01 7.35599875e-01 -5.85333586e-01 -7.18485475e-01 1.24397361e+00 9.76439059e-01 3.01020071e-02 2.39927620e-01 -1.01068234e+00 -6.23413682e-01 4.10822541e-01 -7.81487346e-01 2.05781668e-01 2.07845718e-01 3.39056581e-01 5.09681761e-01 -1.04740524e+00 1.46487147e-01 9.56000149e-01 7.85755157e-01 2.97667146e-01 -9.85151887e-01 -8.20746303e-01 -3.21675502e-02 4.40614074e-01 -1.46374094e+00 -7.20418692e-01 1.12496698e+00 -1.01165220e-01 6.87415004e-01 3.54996026e-01 5.02750516e-01 9.76412892e-01 2.34955415e-01 5.86039007e-01 9.66770172e-01 -2.52608448e-01 -5.52357845e-02 -2.42997915e-01 -1.46783650e-01 8.29066694e-01 4.54822749e-01 1.51388243e-01 -6.00377917e-01 2.02582166e-01 9.56393540e-01 3.34402651e-01 -6.84162974e-01 -4.39567193e-02 -1.28627372e+00 5.08456945e-01 6.57908678e-01 3.68108630e-01 -4.79444921e-01 2.54662424e-01 3.29929799e-01 1.54872760e-01 2.04435378e-01 3.64992410e-01 -3.27385753e-01 2.99910069e-01 -1.13550293e+00 1.87953591e-01 2.83203781e-01 8.05370033e-01 7.98697889e-01 2.45468721e-01 -1.95318893e-01 6.23107314e-01 2.93770671e-01 4.99449819e-01 4.64511365e-01 -9.45046365e-01 5.64539254e-01 3.76563191e-01 5.07718958e-02 -1.48360729e+00 -5.00384867e-01 -6.78982139e-01 -1.43766999e+00 -1.11286491e-02 2.24704325e-01 -9.37440246e-02 -7.82061219e-01 1.67956984e+00 3.32166135e-01 4.10600871e-01 8.82090554e-02 1.14076531e+00 8.29854846e-01 5.60134828e-01 -3.13660234e-01 -5.07926285e-01 1.48486876e+00 -1.14688516e+00 -9.54046190e-01 -1.01174422e-01 1.79516464e-01 -1.05737841e+00 1.00854874e+00 5.88328481e-01 -1.44153988e+00 -9.33684051e-01 -1.39649904e+00 -1.62754446e-01 -1.22959442e-01 2.45343044e-01 5.39841175e-01 5.47713399e-01 -1.03070438e+00 6.34234011e-01 -7.03553081e-01 1.68921240e-02 4.57605451e-01 3.54556859e-01 -1.19466223e-01 -2.45103374e-01 -1.15014684e+00 7.61778712e-01 2.63778925e-01 5.20443320e-01 -7.62323439e-01 -5.92508912e-01 -7.33441591e-01 5.35103865e-02 4.14695710e-01 -9.59990323e-01 1.13467109e+00 -6.31762803e-01 -1.48280346e+00 4.15189922e-01 -4.31419872e-02 -4.10308123e-01 4.75361228e-01 -5.80360115e-01 -5.03026307e-01 2.09059149e-01 -1.97613895e-01 4.84057724e-01 1.16899621e+00 -1.25779080e+00 -4.37321335e-01 -2.68001884e-01 1.29926931e-02 2.89790481e-01 -5.80643535e-01 -2.21406724e-02 -6.94048285e-01 -9.30033863e-01 2.73533404e-01 -7.31159925e-01 -3.07097465e-01 8.41287300e-02 -2.58143306e-01 2.17999145e-01 8.51163387e-01 -8.14667523e-01 1.41771674e+00 -2.08884287e+00 1.08679347e-01 -2.07009152e-01 3.83778125e-01 5.06878138e-01 5.80045627e-03 1.96878806e-01 1.40735641e-01 -1.17348291e-01 -4.60126877e-01 -5.61526835e-01 -1.86389759e-01 -6.16382807e-02 -3.77932936e-01 7.06392944e-01 1.09806120e-01 8.10682535e-01 -6.42232239e-01 -6.07472897e-01 6.37326121e-01 6.85535908e-01 -3.99322987e-01 3.05909187e-01 2.83724368e-01 6.49382293e-01 -2.89425999e-01 8.05625618e-01 1.03863156e+00 -2.38232955e-01 4.64846641e-02 -9.83318150e-01 -2.13312015e-01 -2.35357866e-01 -1.23627257e+00 2.22319865e+00 -6.14298046e-01 5.89495957e-01 1.44014776e-01 -9.36957538e-01 8.23578000e-01 2.27917358e-01 4.57824022e-01 -7.82985926e-01 5.15365422e-01 2.22105294e-01 -2.70801932e-01 -5.51451743e-01 5.86287916e-01 2.13355720e-01 3.26773711e-02 1.32301688e-01 -3.73830870e-02 -1.47823542e-01 8.38665590e-02 -6.50907233e-02 8.13382447e-01 -9.83043481e-03 1.15035601e-01 1.88280568e-02 7.36387491e-01 -5.24502099e-01 5.38535833e-01 5.84964991e-01 -1.44374460e-01 8.59542072e-01 -2.67934144e-01 -6.40807033e-01 -9.54002857e-01 -9.36601162e-01 -1.19368367e-01 5.88595271e-01 6.97978914e-01 -3.19187999e-01 -8.33294094e-01 -1.42650709e-01 -4.98707443e-01 3.17235023e-01 -8.05657953e-02 -2.41793588e-01 -9.78251100e-01 -1.04852021e+00 5.96913934e-01 4.02448714e-01 1.17734289e+00 -6.57852948e-01 -9.11448598e-01 3.39868784e-01 -4.27782804e-01 -1.23842478e+00 -5.03074646e-01 -1.53880015e-01 -8.12066972e-01 -9.61221755e-01 -8.44468713e-01 -7.35689521e-01 5.23810983e-01 5.98122418e-01 9.99830127e-01 3.25007975e-01 -4.38930571e-01 -1.59078747e-01 -1.53152943e-01 -9.26860273e-02 7.40359426e-02 -5.17787784e-02 1.89611718e-01 1.26212582e-01 -1.49136350e-01 -8.23013604e-01 -1.14315987e+00 3.91603053e-01 -1.03675508e+00 2.96365529e-01 8.55103910e-01 8.60438526e-01 5.01206279e-01 2.07015097e-01 3.71600240e-01 -2.87843496e-02 3.48909944e-01 -5.84388636e-02 -6.88797534e-01 3.84141117e-01 -6.11098409e-01 -4.11089696e-02 7.07487166e-01 -4.36315149e-01 -1.17653251e+00 1.09441377e-01 -2.12345645e-01 -7.35515654e-01 -3.84211801e-02 1.59407243e-01 -2.43516922e-01 -4.44048136e-01 3.45348597e-01 4.08836275e-01 -1.52841210e-01 -3.48132968e-01 4.17871445e-01 5.73845863e-01 8.88961971e-01 -3.40978205e-01 8.81232262e-01 5.97638667e-01 1.02982648e-01 -5.69489896e-01 -7.51016736e-01 -9.75247100e-02 -3.96201253e-01 -3.89690310e-01 8.78083885e-01 -1.19444776e+00 -8.96403074e-01 8.89210165e-01 -1.21414411e+00 -3.82785709e-03 2.18595996e-01 6.31530404e-01 -3.86107564e-01 6.73521817e-01 -6.97473586e-01 -4.86199826e-01 -7.21268654e-01 -1.45876837e+00 9.66580510e-01 5.61518192e-01 1.54785067e-01 -5.05684912e-01 -3.76281738e-01 4.63829637e-01 7.10033715e-01 3.05711061e-01 3.27755898e-01 -1.03093997e-01 -8.49352658e-01 -1.71646737e-02 -3.87411654e-01 2.72504985e-01 5.33481203e-02 -1.32851079e-01 -9.06241119e-01 -5.76281965e-01 4.42173094e-01 -1.83601841e-01 7.66188502e-01 4.43609983e-01 1.39342117e+00 -3.03712845e-01 -1.92727357e-01 1.12124157e+00 1.60039818e+00 2.00230807e-01 7.03795552e-01 3.28442395e-01 7.80696332e-01 2.84405768e-01 4.25120741e-01 5.42377532e-01 3.86664093e-01 7.21337378e-01 5.42843699e-01 -3.67215842e-01 -2.73093134e-01 -6.34621009e-02 7.69554079e-02 9.00485277e-01 -5.04777618e-02 -4.63665605e-01 -6.81991935e-01 3.35940957e-01 -1.85820198e+00 -7.77783334e-01 -2.09401205e-01 2.06128216e+00 7.55449116e-01 6.49713501e-02 -3.83123249e-01 2.12973014e-01 8.76682699e-01 4.51039284e-01 -6.97189689e-01 1.14862263e-01 -3.69742036e-01 9.15916264e-02 7.13447452e-01 2.93569595e-01 -1.15822971e+00 6.18520975e-01 5.21108770e+00 1.02149355e+00 -1.23857927e+00 2.28673056e-01 6.67635918e-01 -1.90223321e-01 -3.38272788e-02 -1.80730462e-01 -6.92609012e-01 5.82397461e-01 6.34562373e-01 1.59557536e-01 4.50600207e-01 3.14200401e-01 4.23997715e-02 3.77402604e-02 -6.90501690e-01 1.57284343e+00 2.76691824e-01 -1.47157013e+00 -1.19314283e-01 -1.90294430e-01 8.29986095e-01 -2.20784828e-01 2.61933327e-01 -1.03622250e-01 -1.59588426e-01 -7.94808269e-01 6.49958014e-01 6.52066350e-01 7.72539437e-01 -8.25782418e-01 7.35708177e-01 1.08632520e-01 -1.35809994e+00 -1.81115702e-01 -2.17351228e-01 -1.46362707e-02 5.70889771e-01 8.38426232e-01 -5.76700375e-04 9.44228411e-01 8.87309492e-01 4.42058921e-01 -5.36794066e-01 9.93452311e-01 -1.13199249e-01 2.96870202e-01 -2.50902116e-01 3.77160937e-01 1.56947464e-01 -1.10623054e-01 4.58201408e-01 1.06264186e+00 6.15475714e-01 1.71474263e-01 1.67188440e-02 5.11265159e-01 -1.66592166e-01 -2.94186771e-01 -3.54636073e-01 4.34686422e-01 4.81432408e-01 1.42705798e+00 -7.02359498e-01 -3.78651470e-01 -4.58682030e-01 1.15237653e+00 4.72821221e-02 3.06782454e-01 -1.29487073e+00 -4.89238918e-01 5.78022599e-01 -2.16505229e-01 2.65555948e-01 -3.50062132e-01 -3.95326704e-01 -1.26640630e+00 2.49658719e-01 -8.98010254e-01 8.71971026e-02 -6.98719084e-01 -1.03295946e+00 8.24334145e-01 -1.27300307e-01 -1.36927760e+00 1.26090080e-01 -3.13053846e-01 -5.06192148e-01 6.78548574e-01 -1.55449998e+00 -1.17982697e+00 -7.69759178e-01 6.44089639e-01 6.82366967e-01 -1.47758454e-01 3.54118615e-01 7.60716856e-01 -8.43028963e-01 8.48526120e-01 1.13519683e-01 -9.71665159e-02 8.72047186e-01 -6.72997177e-01 4.47667837e-01 1.18407965e+00 1.03422031e-01 5.67829311e-01 6.30492628e-01 -3.36437643e-01 -1.74146891e+00 -1.11996651e+00 3.98277998e-01 -4.36518826e-02 3.76653880e-01 -1.18059583e-01 -1.01519930e+00 1.18528731e-01 5.87201536e-01 7.91806579e-02 4.04031307e-01 -5.42520463e-01 -2.06268147e-01 -4.04217809e-01 -1.07015073e+00 7.30398774e-01 1.21401668e+00 -3.76881719e-01 -2.11146340e-01 2.77904153e-01 9.50372219e-01 -6.93211615e-01 -9.48119938e-01 7.44232774e-01 4.76583093e-01 -1.12051153e+00 1.27203357e+00 1.75237909e-01 4.03169096e-01 -6.95439637e-01 -3.36395949e-01 -7.88546443e-01 -3.63839656e-01 -8.65467548e-01 -4.14394051e-01 1.18675745e+00 1.50944637e-02 -5.81816614e-01 3.88743669e-01 3.05066735e-01 -3.58062804e-01 -1.05467331e+00 -8.27911854e-01 -5.45047224e-01 -4.19670522e-01 -1.04215860e-01 7.20943034e-01 8.48337770e-01 -6.18895292e-01 1.98255181e-01 -7.75792778e-01 5.04466355e-01 8.47672939e-01 2.23666877e-01 6.38833046e-01 -6.05930090e-01 -2.46847734e-01 -5.35784721e-01 -2.44818032e-01 -1.18851638e+00 -8.36122558e-02 -3.11699599e-01 -4.89081405e-02 -1.26631093e+00 4.36759777e-02 -1.71730250e-01 -2.83136845e-01 2.21696243e-01 -3.65131855e-01 6.29472733e-01 2.76055962e-01 3.35740507e-01 -5.25058985e-01 6.90106750e-01 1.24111533e+00 -5.97480983e-02 1.87697145e-03 -3.46108407e-01 -6.74218178e-01 7.53963053e-01 1.09838653e+00 -1.90777123e-01 -3.84183317e-01 -9.17485654e-01 9.17187408e-02 1.28707021e-01 6.36053801e-01 -1.36811316e+00 6.58044577e-01 3.07910703e-02 6.08365834e-01 -7.02362716e-01 5.60333133e-01 -7.22613275e-01 4.51735824e-01 6.64866328e-01 -7.55766109e-02 3.41171145e-01 2.79647946e-01 4.53078836e-01 -3.53080273e-01 -5.94697297e-02 8.09634984e-01 3.22582833e-02 -7.58134425e-01 4.45086241e-01 7.82185793e-02 -4.34678406e-01 9.12602246e-01 -2.31657505e-01 -3.70816499e-01 -1.93101853e-01 -1.79393128e-01 6.38253689e-02 4.34708238e-01 2.20628515e-01 8.00875187e-01 -1.28777909e+00 -7.06682324e-01 3.01418155e-01 -3.53372157e-01 2.93075383e-01 9.43955421e-01 9.83027875e-01 -5.77760279e-01 1.13630511e-01 -2.18005881e-01 -5.39689779e-01 -1.21783221e+00 6.72223091e-01 1.57171085e-01 -1.40167847e-01 -5.36958396e-01 8.86171520e-01 1.93781525e-01 2.29453612e-02 1.52715638e-01 -1.58917606e-01 1.83819935e-01 -5.45241311e-02 5.93763113e-01 4.09079105e-01 1.88664094e-01 -5.04525602e-01 -2.95033514e-01 8.26657832e-01 -1.42745391e-01 6.40987456e-02 1.08098662e+00 -3.46143663e-01 9.29970369e-02 -1.75162211e-01 1.33880150e+00 -2.61232197e-01 -1.44312525e+00 -3.37804466e-01 -5.13082743e-01 -6.66389108e-01 3.07134300e-01 -5.87492943e-01 -1.39911628e+00 8.27196538e-01 9.19156075e-01 -1.07833117e-01 1.84133923e+00 -2.66495705e-01 1.26163077e+00 3.40161294e-01 2.36262858e-01 -8.11682165e-01 2.05352962e-01 1.37541324e-01 9.22535121e-01 -1.28096199e+00 2.78471857e-01 -2.70379603e-01 -3.51609707e-01 1.09875834e+00 7.05715418e-01 -7.37023652e-02 3.70231211e-01 3.49390000e-01 -7.84605369e-02 -1.59778237e-01 -4.13428545e-01 3.83188874e-02 1.49458036e-01 2.50617832e-01 1.73841685e-01 -1.21280327e-01 -3.29861999e-01 3.87258679e-01 -1.76976278e-01 -1.82514518e-01 2.22711131e-01 9.08934176e-01 -2.89980620e-01 -7.09208250e-01 -5.82116842e-01 1.53642431e-01 -4.96277273e-01 -2.11934313e-01 2.23661795e-01 5.89741945e-01 2.71342337e-01 9.83661592e-01 2.77553406e-03 -5.76771855e-01 2.72752821e-01 -5.60635865e-01 5.66071868e-01 2.05070674e-01 -5.66790998e-01 2.45511666e-01 -2.27713719e-01 -6.81385040e-01 -6.86563909e-01 -3.73986125e-01 -8.39147389e-01 -4.36113179e-01 -5.68541586e-01 -2.34914884e-01 6.69046223e-01 6.88598633e-01 4.96147066e-01 6.62644625e-01 8.09762061e-01 -1.10405898e+00 -5.06874263e-01 -8.46940756e-01 -3.40404436e-02 1.09605730e-01 4.88966912e-01 -5.42628288e-01 -2.73883224e-01 1.96894243e-01]
[10.864858627319336, -1.9377655982971191]
14578a2d-9656-4ce5-8899-ce4ce69c847b
efficient-global-2d-3d-matching-for-camera
null
null
http://openaccess.thecvf.com/content_iccv_2017/html/Liu_Efficient_Global_2D-3D_ICCV_2017_paper.html
http://openaccess.thecvf.com/content_ICCV_2017/papers/Liu_Efficient_Global_2D-3D_ICCV_2017_paper.pdf
Efficient Global 2D-3D Matching for Camera Localization in a Large-Scale 3D Map
Given an image of a street scene in a city, this paper develops a new method that can quickly and precisely pinpoint at which location (as well as viewing direction) the image was taken, against a pre-stored large-scale 3D point-cloud map of the city. We adopt the recently developed 2D-3D direct feature matching framework for this task [23,31,32,42-44]. This is a challenging task especially for large-scale problems. As the map size grows bigger, many 3D points in the wider geographical area can be visually very similar-or even identical-causing severe ambiguities in 2D-3D feature matching. The key is to quickly and unambiguously find the correct matches between a query image and the large 3D map. Existing methods solve this problem mainly via comparing individual features' visual similarities in a local and per feature manner, thus only local solutions can be found, inadequate for large-scale applications. In this paper, we introduce a global method which harnesses global contextual information exhibited both within the query image and among all the 3D points in the map. This is achieved by a novel global ranking algorithm, applied to a Markov network built upon the 3D map, which takes account of not only visual similarities between individual 2D-3D matches, but also their global compatibilities (as measured by co-visibility) among all matching pairs found in the scene. Tests on standard benchmark datasets show that our method achieved both higher precision and comparable recall, compared with the state-of-the-art.
['Yuchao Dai', 'Liu Liu', 'Hongdong Li']
2017-10-01
null
null
null
iccv-2017-10
['3d-feature-matching', 'camera-localization']
['computer-vision', 'computer-vision']
[-2.89887041e-02 -5.15127122e-01 1.52653560e-01 -2.06544027e-01 -9.05796707e-01 -7.96512127e-01 7.67479599e-01 5.81565738e-01 -1.97592795e-01 1.12380862e-01 -1.58217594e-01 -1.01236783e-01 -4.38580900e-01 -1.14002669e+00 -4.36714679e-01 -3.54181409e-01 -1.87189147e-01 7.52622485e-01 8.54317427e-01 -2.23368153e-01 5.06606638e-01 1.02700341e+00 -1.81181431e+00 -5.67031316e-02 6.10031545e-01 9.13746059e-01 6.09116316e-01 5.22516549e-01 -1.29595295e-01 2.16937557e-01 -3.75202894e-01 -3.38970780e-01 5.44764578e-01 6.99727312e-02 -6.47443414e-01 1.91431075e-01 9.10979927e-01 -2.44106621e-01 -3.08004409e-01 1.26970017e+00 2.30116099e-01 1.95323005e-01 5.10370612e-01 -1.24048603e+00 -2.43465215e-01 -4.89774317e-01 -7.21258640e-01 1.70920268e-01 8.10622752e-01 1.66501105e-02 9.82909799e-01 -1.00854158e+00 8.00079465e-01 1.06972075e+00 5.56885123e-01 -2.73037046e-01 -1.06783915e+00 -5.11585772e-01 1.60593970e-03 3.17488015e-01 -1.85253739e+00 -1.15130097e-01 7.36218989e-01 -5.88978708e-01 1.00895023e+00 4.52855319e-01 7.79307067e-01 2.74481803e-01 1.94405671e-02 2.56304026e-01 1.22755265e+00 -4.73669678e-01 6.68626875e-02 -7.48580620e-02 -7.42695630e-02 6.49596214e-01 1.61822606e-02 1.55125707e-01 -4.38199222e-01 -3.04458708e-01 9.40001845e-01 2.01009050e-01 -6.55903248e-03 -8.38277102e-01 -1.19017243e+00 5.82832217e-01 7.16344953e-01 4.47094291e-01 -4.62976545e-01 -2.25077286e-01 -1.65477842e-01 2.56031483e-01 2.96243936e-01 2.44357958e-01 -1.50530994e-01 -2.55559999e-02 -9.37227905e-01 3.65233570e-01 4.14708167e-01 8.79953682e-01 1.43474269e+00 -6.56440198e-01 2.57351667e-01 4.79554564e-01 2.39417046e-01 8.36895883e-01 -1.68514445e-01 -6.37487948e-01 7.88325965e-01 1.03279209e+00 2.44795084e-01 -1.98304522e+00 -4.43028748e-01 -7.01205209e-02 -7.74399757e-01 5.09351671e-01 4.58970517e-01 7.25450754e-01 -7.61897445e-01 1.40600872e+00 7.51820743e-01 2.78524756e-01 -2.18994290e-01 7.77634919e-01 6.62127435e-01 5.99202573e-01 -3.16894859e-01 9.83668715e-02 1.24748957e+00 -4.69641119e-01 -1.05564952e-01 -3.51981372e-01 2.97425181e-01 -1.10802901e+00 6.61906362e-01 -1.01508230e-01 -8.43949378e-01 -5.98735213e-01 -9.26062703e-01 1.44279286e-01 -7.07418025e-01 -8.11437666e-02 2.64946938e-01 4.05901402e-01 -1.13179886e+00 3.68229508e-01 -4.43387598e-01 -7.71690726e-01 1.92126967e-02 2.88796753e-01 -6.71428978e-01 -4.03510273e-01 -9.16469097e-01 1.07059968e+00 2.99967349e-01 5.06954715e-02 -3.83283168e-01 -4.30090338e-01 -7.36236751e-01 -1.40087724e-01 3.50588322e-01 -6.43309295e-01 8.02653134e-01 -5.77095985e-01 -9.37014043e-01 1.22638249e+00 -3.71344119e-01 7.16964304e-02 5.74890912e-01 -7.00242072e-03 -5.64778924e-01 2.97145337e-01 6.01534843e-01 4.89767224e-01 6.38483226e-01 -1.45607650e+00 -9.81481373e-01 -7.14199305e-01 1.09786019e-01 4.44800407e-01 2.25489840e-01 1.71564952e-01 -9.07884538e-01 -2.82296270e-01 7.22893655e-01 -8.96303773e-01 -4.13615167e-01 6.97516948e-02 -3.00131708e-01 -9.14678872e-02 7.97405779e-01 -6.04466856e-01 9.95194495e-01 -2.12413597e+00 -2.30935991e-01 1.07605326e+00 1.41683862e-01 2.41084293e-01 -7.31842145e-02 6.82477415e-01 -8.38361308e-03 -5.30035608e-02 -1.17121503e-01 -3.35546136e-02 -1.01868466e-01 1.27574608e-01 -9.55314040e-02 5.86210608e-01 -5.55850901e-02 8.63647282e-01 -1.16722834e+00 -5.33736348e-01 7.45146573e-01 2.54687309e-01 -8.54383223e-03 1.47874847e-01 3.69909585e-01 3.81919146e-01 -6.65668845e-01 6.54812157e-01 1.06625032e+00 -2.28765413e-01 -1.55711979e-01 -1.20500728e-01 -3.86492252e-01 6.45117238e-02 -1.65923822e+00 1.68781996e+00 -3.68705869e-01 5.09914935e-01 -3.08124453e-01 -7.31319666e-01 1.31436801e+00 2.76664533e-02 4.83787894e-01 -1.08163404e+00 -4.14533049e-01 4.37402576e-01 -6.14324033e-01 -3.69839251e-01 6.94895208e-01 2.85561562e-01 -6.57575279e-02 2.86218852e-01 -4.58322614e-01 -3.82777393e-01 -1.88940093e-02 9.61835235e-02 1.11423826e+00 -2.22743973e-01 5.25884926e-01 -1.55214235e-01 6.46950781e-01 2.35344008e-01 1.23956181e-01 1.01955163e+00 -6.77210093e-02 8.44110966e-01 1.53538212e-01 -6.22638643e-01 -1.12526131e+00 -1.19623649e+00 -2.70267110e-02 3.60199898e-01 7.02830851e-01 -5.28237641e-01 -2.46777758e-01 -6.14753008e-01 1.63573831e-01 1.80876940e-01 -4.60958511e-01 3.36853176e-01 -4.63660926e-01 -2.79391289e-01 5.55546023e-02 1.05316140e-01 6.72912776e-01 -5.34947455e-01 -8.31782997e-01 1.56280518e-01 -1.81925714e-01 -1.17795539e+00 -1.80744231e-01 -2.38858789e-01 -8.36551666e-01 -1.11488497e+00 -4.96474177e-01 -7.09175706e-01 7.49700487e-01 1.03519917e+00 1.18838811e+00 1.44747332e-01 -2.43418634e-01 4.25582826e-01 -3.18405211e-01 1.33369207e-01 -1.40772194e-01 -2.56375611e-01 -1.45666003e-01 1.01212814e-01 5.16302824e-01 -6.28011167e-01 -5.02556741e-01 8.23827267e-01 -5.34829378e-01 5.34145869e-02 5.72551608e-01 4.05588955e-01 8.37909281e-01 3.18388343e-01 -1.69376820e-01 -2.74797916e-01 5.58893979e-02 -2.46050790e-01 -1.01467359e+00 4.95258868e-01 -3.49509656e-01 -2.30722547e-01 5.04648201e-02 6.02559187e-02 -5.43983400e-01 4.23734903e-01 4.12923209e-02 -3.77441347e-01 -2.54304230e-01 5.26452780e-01 -2.53674030e-01 -3.78249913e-01 4.83802140e-01 2.96305925e-01 -2.41716251e-01 -4.90127236e-01 2.40655303e-01 5.83142102e-01 5.99128842e-01 -2.62536615e-01 1.35372138e+00 8.59182298e-01 4.74785775e-01 -7.96538949e-01 -5.27740002e-01 -1.13861084e+00 -1.20413232e+00 -3.74772012e-01 6.94215357e-01 -9.35889661e-01 -4.62194830e-01 4.68924642e-01 -1.19039178e+00 2.18447983e-01 4.15157191e-02 3.94569755e-01 -5.10317028e-01 5.04449129e-01 2.07320973e-01 -5.62961757e-01 1.53995827e-01 -1.06729329e+00 1.22562802e+00 9.57925916e-02 -1.97039042e-02 -8.07074368e-01 4.59217787e-01 7.68470690e-02 3.09382170e-01 4.42332447e-01 7.78676033e-01 -3.50230575e-01 -9.85129595e-01 -5.40387690e-01 -7.59409785e-01 -1.57960221e-01 1.89235121e-01 9.40218866e-02 -7.65916884e-01 -4.10010330e-02 -3.08000922e-01 3.27195078e-01 4.63358998e-01 1.97063729e-01 4.89816636e-01 1.82772338e-01 -4.91747350e-01 3.50698292e-01 1.83180070e+00 -3.29472832e-02 6.56590641e-01 7.77623773e-01 6.97942197e-01 8.10813129e-01 8.63258362e-01 3.64248008e-01 4.70605105e-01 1.27066410e+00 6.19114220e-01 -1.39061838e-01 -6.96219644e-03 -4.52564716e-01 -3.18331532e-02 6.02835417e-01 -3.04572761e-01 2.38209851e-02 -1.25326419e+00 6.63452029e-01 -2.17545390e+00 -1.03870642e+00 -5.00574648e-01 2.68852019e+00 -2.38881819e-02 -1.11617576e-02 4.00807671e-02 2.75797173e-02 1.02695084e+00 2.56698042e-01 -1.96229458e-01 4.39990833e-02 -3.13553154e-01 5.45063205e-02 7.25704193e-01 4.90903854e-01 -1.24750412e+00 6.63698375e-01 5.79571390e+00 9.94793653e-01 -7.86505461e-01 9.33897495e-03 2.74879098e-01 2.56556898e-01 -9.50220004e-02 2.71527439e-01 -6.80085123e-01 1.75490692e-01 4.51080143e-01 1.35791719e-01 9.63246822e-02 7.66548395e-01 2.38368541e-01 -5.29992461e-01 -8.20987165e-01 1.34147334e+00 6.48132637e-02 -1.41317892e+00 -2.92749610e-02 3.42583537e-01 8.89615893e-01 2.96030581e-01 -8.79710838e-02 -3.82820666e-01 1.23669565e-01 -8.20801258e-01 7.76550770e-01 4.24501330e-01 7.98739016e-01 -8.28775465e-01 5.98482013e-01 3.23305845e-01 -1.75631928e+00 1.88504755e-01 -3.53704780e-01 6.62498996e-02 1.73396617e-01 6.38879955e-01 -7.23907351e-01 1.00295401e+00 9.93727565e-01 8.13048661e-01 -8.83634448e-01 1.47372735e+00 -3.05789948e-01 -2.69242287e-01 -5.86660624e-01 1.79843560e-01 4.83159840e-01 -2.80662358e-01 6.95800006e-01 1.08263755e+00 6.97550118e-01 -5.38729243e-02 3.48872304e-01 5.31527340e-01 4.12452757e-01 2.42995262e-01 -1.02722156e+00 6.83752835e-01 5.15519083e-01 1.28455329e+00 -9.53198850e-01 -2.16389835e-01 -5.26850760e-01 8.21912050e-01 2.99774587e-01 2.20669061e-01 -4.88036007e-01 -2.91561127e-01 6.59993589e-01 2.12922916e-01 4.40883279e-01 -6.24910116e-01 -6.60379380e-02 -8.98773491e-01 2.88291425e-01 -4.35945719e-01 3.68258446e-01 -1.01278448e+00 -1.18262637e+00 4.91740257e-01 1.84329972e-01 -1.72009468e+00 -1.98468864e-01 -3.84245664e-01 -5.69071889e-01 1.03874576e+00 -1.68876410e+00 -1.25135994e+00 -5.47800303e-01 9.69196379e-01 1.21459961e-01 1.31543159e-01 6.69672072e-01 3.07518721e-01 -2.46219449e-02 6.37634844e-02 3.97020340e-01 -1.58737600e-02 3.71337473e-01 -9.98796225e-01 6.54484093e-01 9.81894732e-01 4.18581307e-01 2.88667411e-01 5.04190803e-01 -5.99535942e-01 -1.08343291e+00 -1.00050163e+00 1.44226134e+00 -6.70962572e-01 6.98019743e-01 -2.90348172e-01 -7.77811170e-01 2.06608400e-01 -2.66530663e-01 3.77799310e-02 1.47325158e-01 -3.74693908e-02 -4.74418133e-01 -1.73096016e-01 -1.15405166e+00 4.17306930e-01 1.23476899e+00 -8.65119934e-01 -5.85072339e-01 3.29100192e-01 2.15493083e-01 -4.57351357e-01 -7.76206553e-01 2.40066767e-01 4.23268378e-01 -1.34294355e+00 1.32813859e+00 -4.21811081e-02 -2.95193642e-02 -7.43711948e-01 -5.96682549e-01 -1.01976740e+00 -4.00353312e-01 -4.36476856e-01 4.49834228e-01 1.14699924e+00 1.31289557e-01 -5.62482476e-01 6.77937686e-01 4.33280259e-01 6.96844906e-02 -2.82298982e-01 -1.40739405e+00 -1.09046304e+00 -5.43436766e-01 -5.61650574e-01 9.76040721e-01 7.93937862e-01 -3.24259877e-01 -8.30836743e-02 -1.16990834e-01 7.58089840e-01 5.45937061e-01 5.29761970e-01 1.14354157e+00 -1.46143734e+00 1.78482443e-01 -4.19232607e-01 -1.23684871e+00 -1.11444020e+00 -2.23216981e-01 -7.50530779e-01 -1.05727814e-01 -1.60790026e+00 2.45764852e-01 -8.28587234e-01 -1.30676299e-01 2.58445323e-01 3.21337059e-02 3.86555314e-01 2.81165898e-01 6.69491410e-01 -6.44761503e-01 1.32527471e-01 9.52972710e-01 -6.60640076e-02 -2.35658005e-01 1.60047382e-01 -1.75482556e-02 5.87428033e-01 4.87867028e-01 -5.52608907e-01 -7.86460042e-02 -3.26223761e-01 4.45904881e-01 -3.12854312e-02 7.66010523e-01 -1.23751056e+00 4.74655479e-01 -2.90964544e-01 2.54892200e-01 -1.17113721e+00 5.39158463e-01 -1.18215477e+00 5.46326041e-01 3.03217351e-01 2.53874838e-01 6.62010789e-01 -1.40399886e-02 7.04428375e-01 -4.44923669e-01 -2.95569330e-01 4.69519407e-01 -2.87723571e-01 -1.27074802e+00 4.18859869e-01 4.38315757e-02 -2.90056258e-01 1.21194124e+00 -6.47721887e-01 -2.65754193e-01 -3.99081349e-01 -4.08813477e-01 1.62845165e-01 9.41847861e-01 4.21812087e-01 9.04140949e-01 -1.45079362e+00 -6.71373546e-01 3.37465852e-01 6.34042680e-01 2.07411349e-02 3.47996950e-01 7.65592635e-01 -6.59437299e-01 4.94021028e-01 -2.49254808e-01 -1.07670105e+00 -1.55466974e+00 6.58494949e-01 1.87543288e-01 -3.66547287e-01 -6.94960177e-01 4.15404320e-01 1.82203859e-01 -4.80822057e-01 -1.24494538e-01 -1.86109841e-01 -7.39820227e-02 1.77907243e-01 4.35666800e-01 2.38520771e-01 3.70687962e-01 -1.26080978e+00 -6.34080827e-01 1.53956771e+00 2.05920488e-01 -1.34522840e-01 1.22062194e+00 -5.15521109e-01 -9.49496552e-02 2.28072166e-01 1.35122573e+00 7.78000876e-02 -1.03283823e+00 -5.50001919e-01 8.06594044e-02 -1.28683197e+00 2.80293729e-02 -3.84749889e-01 -9.31686163e-01 7.29024589e-01 8.57017994e-01 4.02711391e-01 1.00537467e+00 2.65568793e-01 4.35065478e-01 2.86515683e-01 7.61558771e-01 -8.61881196e-01 -4.19302732e-02 4.75073636e-01 5.88916242e-01 -1.40774739e+00 2.20035180e-01 -4.55236733e-01 -3.82495284e-01 1.18343365e+00 1.33216560e-01 -1.40048176e-01 7.33305752e-01 -4.29703951e-01 -8.95661954e-03 -6.19821191e-01 -9.48283672e-02 -5.94056487e-01 3.93188566e-01 8.42039704e-01 -5.02666116e-01 -6.50401553e-03 3.31458002e-01 -4.30848211e-01 -3.25502791e-02 -3.88980895e-01 1.50382921e-01 9.81330693e-01 -6.62583470e-01 -1.12595129e+00 -6.84644043e-01 8.97658691e-02 1.70090258e-01 1.38259726e-02 -2.46815145e-01 9.93120909e-01 1.58518434e-01 9.86785591e-01 3.58881682e-01 -4.34233338e-01 5.78445196e-01 -4.32265043e-01 3.87460321e-01 -3.15325737e-01 -4.41741645e-01 -7.46458396e-02 -1.01426698e-01 -8.47262561e-01 -7.88932025e-01 -9.58986700e-01 -7.80320227e-01 -3.49262536e-01 -2.53155559e-01 -1.16754167e-01 9.20735121e-01 8.35843563e-01 5.08107841e-01 -3.75569224e-01 1.04240465e+00 -1.01383746e+00 -7.77489468e-02 -2.58975476e-01 -6.78892732e-01 5.06331801e-01 3.00226629e-01 -8.63173068e-01 -3.22086543e-01 -2.69272953e-01]
[7.625156402587891, -2.3459184169769287]
88e48a45-1281-4716-99cd-d804e0c06894
self-supervised-learning-framework-for-remote
2107.07695
null
https://arxiv.org/abs/2107.07695v2
https://arxiv.org/pdf/2107.07695v2.pdf
Self-supervised Representation Learning Framework for Remote Physiological Measurement Using Spatiotemporal Augmentation Loss
Recent advances in supervised deep learning methods are enabling remote measurements of photoplethysmography-based physiological signals using facial videos. The performance of these supervised methods, however, are dependent on the availability of large labelled data. Contrastive learning as a self-supervised method has recently achieved state-of-the-art performances in learning representative data features by maximising mutual information between different augmented views. However, existing data augmentation techniques for contrastive learning are not designed to learn physiological signals from videos and often fail when there are complicated noise and subtle and periodic colour or shape variations between video frames. To address these problems, we present a novel self-supervised spatiotemporal learning framework for remote physiological signal representation learning, where there is a lack of labelled training data. Firstly, we propose a landmark-based spatial augmentation that splits the face into several informative parts based on the Shafer dichromatic reflection model to characterise subtle skin colour fluctuations. We also formulate a sparsity-based temporal augmentation exploiting Nyquist-Shannon sampling theorem to effectively capture periodic temporal changes by modelling physiological signal features. Furthermore, we introduce a constrained spatiotemporal loss which generates pseudo-labels for augmented video clips. It is used to regulate the training process and handle complicated noise. We evaluated our framework on 3 public datasets and demonstrated superior performances than other self-supervised methods and achieved competitive accuracy compared to the state-of-the-art supervised methods.
['Jinman Kim', 'Euijoon Ahn', 'Hao Wang']
2021-07-16
null
null
null
null
['heart-rate-estimation']
['medical']
[ 7.06437171e-01 -3.50533426e-01 -1.23275593e-01 -4.50757295e-01 -8.86500835e-01 -2.16198951e-01 4.75375235e-01 -4.09075260e-01 -3.33494842e-01 8.10842633e-01 2.50552326e-01 4.25700277e-01 -2.27896526e-01 -2.00368986e-01 -6.15181863e-01 -1.17208064e+00 -4.16653186e-01 -2.46221080e-01 -2.95250535e-01 -2.20606960e-02 1.04273655e-01 4.82896686e-01 -1.84668148e+00 4.47347462e-01 6.09832227e-01 1.35867274e+00 -2.21931234e-01 4.89810497e-01 7.00712427e-02 7.25501776e-01 -3.95493597e-01 1.26436606e-01 2.22268701e-01 -9.39909935e-01 -5.26869893e-01 1.62208334e-01 4.78433073e-01 -6.71521798e-02 -3.90279979e-01 6.92704797e-01 9.81794715e-01 8.77618045e-02 5.43028355e-01 -1.21932924e+00 -4.67227221e-01 -4.47284207e-02 -7.27883220e-01 5.12360394e-01 4.92875874e-01 1.83976367e-01 6.49638534e-01 -8.29248786e-01 5.63105643e-01 7.14310229e-01 8.58175695e-01 9.50480282e-01 -1.50865567e+00 -7.21001983e-01 -8.35851058e-02 5.06146252e-01 -1.18313825e+00 -7.73134351e-01 1.18070745e+00 -2.03754306e-01 6.82933927e-01 3.43923390e-01 9.24977124e-01 1.50800061e+00 1.30690001e-02 7.23815918e-01 1.56840026e+00 -3.05794120e-01 2.72555500e-01 -4.26395983e-02 -5.13629794e-01 5.68665624e-01 -5.54604590e-01 3.87447685e-01 -1.09083152e+00 -1.68315604e-01 7.47727096e-01 -6.14583790e-02 -5.78961134e-01 -4.91664022e-01 -1.03181946e+00 6.16272390e-01 2.23977208e-01 4.14591402e-01 -5.70424914e-01 9.55203772e-02 6.01451159e-01 2.77821064e-01 8.29101861e-01 2.45542318e-01 -5.08108795e-01 -2.66330928e-01 -1.14890730e+00 -2.83795297e-01 6.08625948e-01 4.04306084e-01 5.25860786e-01 1.12646341e-01 -1.72116473e-01 9.52888250e-01 1.06033899e-01 3.90572041e-01 8.01281869e-01 -1.03135395e+00 -1.78095579e-01 2.15714514e-01 -2.23195896e-01 -8.69237483e-01 -5.77410877e-01 -2.44268999e-01 -1.02320325e+00 -5.26242936e-03 2.61347979e-01 -9.18031484e-02 -7.85768628e-01 1.80574548e+00 3.85435164e-01 1.09287250e+00 -3.50457360e-03 1.04943335e+00 8.80973756e-01 3.97620678e-01 -6.22852892e-02 -8.90390217e-01 9.55939353e-01 -4.58598733e-01 -9.27970350e-01 1.63634732e-01 5.57695031e-01 -3.92633528e-01 9.16157246e-01 4.90148187e-01 -9.68253255e-01 -5.34657955e-01 -8.01736593e-01 3.07195038e-01 -1.63222536e-01 3.96955982e-02 6.15600765e-01 7.65062869e-01 -1.03978574e+00 7.27500141e-01 -7.61567116e-01 -1.87622428e-01 7.78882682e-01 3.85951728e-01 -6.16978586e-01 -5.38012162e-02 -1.35040295e+00 6.05220377e-01 -2.18133807e-01 4.48571950e-01 -7.92529464e-01 -1.17531264e+00 -9.22992527e-01 -2.98336923e-01 2.28443131e-01 -3.53593588e-01 7.02006757e-01 -1.30066526e+00 -1.94759142e+00 1.16448081e+00 -8.34709257e-02 -2.46319488e-01 4.70177591e-01 -6.11136556e-02 -6.04065061e-01 7.50509799e-01 -2.52002686e-01 6.26404583e-01 1.17399168e+00 -1.07824159e+00 4.95973267e-02 -4.30767775e-01 -4.60767657e-01 3.30565214e-01 -4.21370685e-01 -1.24105051e-01 -2.68825561e-01 -5.01997769e-01 8.43851492e-02 -8.36062372e-01 -8.30059275e-02 3.18521410e-01 8.20100605e-02 1.36119202e-01 8.86907756e-01 -4.93890524e-01 7.97330379e-01 -2.26575708e+00 -1.20848526e-04 1.65018484e-01 1.97569519e-01 4.35752481e-01 -3.92908126e-01 1.74769521e-01 -3.86963874e-01 -2.64024556e-01 -3.16136926e-01 -3.96259308e-01 -3.25975060e-01 8.86987671e-02 -1.05439089e-01 1.05073655e+00 3.35975587e-01 9.19738531e-01 -9.77794588e-01 -5.14362574e-01 5.96313357e-01 1.04077101e+00 -3.71671766e-01 2.40731016e-01 1.73014507e-01 1.07389390e+00 -2.99259238e-02 6.73538923e-01 6.24531507e-01 -9.73313600e-02 6.28186390e-02 -3.26801926e-01 7.94621259e-02 1.03823775e-02 -8.02974284e-01 2.11716175e+00 -4.67690021e-01 7.11179852e-01 1.07383005e-01 -1.38325644e+00 1.03159344e+00 6.10336959e-01 1.25822949e+00 -9.83703732e-01 1.96405333e-02 1.03693433e-01 -3.12557489e-01 -8.45546842e-01 -3.78111333e-01 -6.40393019e-01 4.98333275e-01 2.14575857e-01 2.68602461e-01 3.88343656e-03 -3.19510818e-01 -1.95582762e-01 1.12259209e+00 7.28704929e-01 3.10420003e-02 -8.88717249e-02 8.53438079e-01 -4.78364557e-01 6.05811477e-01 4.50011492e-01 -5.58356404e-01 8.43782485e-01 4.28283334e-01 -5.11376083e-01 -6.69144273e-01 -6.55538440e-01 -4.02268827e-01 8.71593118e-01 -5.50714992e-02 -2.00178549e-01 -5.58177352e-01 -6.22420013e-01 -1.63041577e-01 7.54316524e-03 -9.64672267e-01 -4.02753562e-01 -3.08843225e-01 -1.00317514e+00 5.35239756e-01 4.18292016e-01 5.22561491e-01 -1.27274227e+00 -7.39853203e-01 1.04289062e-01 -2.26582915e-01 -1.21191382e+00 -2.97620445e-01 1.99592590e-01 -8.51605833e-01 -1.17179286e+00 -9.07003224e-01 -5.39898455e-01 4.98127669e-01 1.18375540e-01 8.09769869e-01 -1.87466219e-01 -8.55773270e-01 9.10139322e-01 -3.07936847e-01 -1.86705247e-01 3.29915658e-02 -4.05555576e-01 2.87933350e-01 5.73314190e-01 6.00217760e-01 -8.90664220e-01 -9.17067468e-01 2.58977175e-01 -8.67930651e-01 -3.15010041e-01 4.57669616e-01 1.07272077e+00 7.50320315e-01 -3.27961653e-01 7.80816019e-01 -6.59639120e-01 3.11712712e-01 -3.98857862e-01 -2.26772889e-01 -4.80669215e-02 -3.57037097e-01 -1.09324209e-01 5.52151203e-01 -7.93371320e-01 -9.72058952e-01 3.66947740e-01 5.50368205e-02 -8.29314530e-01 -3.12992185e-01 5.48977256e-01 1.16770275e-01 -5.39600968e-01 8.32886755e-01 6.10955238e-01 5.60340047e-01 -9.74727497e-02 2.17543036e-01 5.55134535e-01 5.38897753e-01 -3.71755809e-01 3.96627039e-01 7.57728934e-01 3.77293766e-01 -1.19533837e+00 -8.00935984e-01 -7.80176699e-01 -9.16604817e-01 -5.64154267e-01 5.36497593e-01 -8.68743420e-01 -9.21605349e-01 6.50155306e-01 -5.75737894e-01 -4.14117247e-01 -4.34214950e-01 6.26936972e-01 -1.00089622e+00 6.00531757e-01 -3.34376156e-01 -1.01212895e+00 -3.13107461e-01 -5.95174313e-01 1.13418376e+00 7.63247442e-03 8.71547312e-02 -1.13354254e+00 3.58800769e-01 3.44366491e-01 5.52925646e-01 6.58457458e-01 2.10019216e-01 -3.65908056e-01 -6.78732768e-02 -2.28903234e-01 8.31881538e-02 6.24184370e-01 3.76573771e-01 -3.43997389e-01 -1.44765961e+00 -2.24726215e-01 4.06489491e-01 -6.73206449e-01 8.94309998e-01 7.10594058e-01 1.39965296e+00 -1.61625639e-01 -3.97492424e-02 9.67042327e-01 1.15998793e+00 -7.93523490e-02 8.43353152e-01 -3.58516835e-02 5.75534642e-01 8.85363042e-01 4.39129502e-01 6.38899148e-01 -1.04383692e-01 6.14553154e-01 2.93576330e-01 -3.72581244e-01 -3.88010032e-02 6.43462837e-02 3.70766103e-01 4.91564006e-01 -1.72593459e-01 3.53382319e-01 -4.56730485e-01 4.58888143e-01 -1.63590920e+00 -1.13651371e+00 9.13923606e-02 2.31778240e+00 1.00822461e+00 -3.52276742e-01 2.10913256e-01 4.18582797e-01 4.87385273e-01 3.23520273e-01 -8.36112678e-01 4.68330570e-02 -2.15479910e-01 4.64182347e-01 2.26053730e-01 1.11585576e-03 -1.21286929e+00 6.59085929e-01 6.07184696e+00 4.36680198e-01 -1.58474791e+00 8.88657868e-02 6.69266343e-01 -5.05295694e-01 1.29241601e-01 -4.68134880e-01 -3.38260651e-01 6.03756070e-01 1.17186880e+00 2.78106958e-01 4.72208142e-01 3.34438324e-01 5.59416890e-01 -9.36485231e-02 -9.90612328e-01 1.55300522e+00 5.11108100e-01 -1.14447153e+00 -5.47563672e-01 -4.92715836e-03 6.42055750e-01 -1.42747760e-01 2.82420158e-01 -1.70090271e-03 -4.72978979e-01 -1.14483297e+00 -2.95076910e-02 7.99298763e-01 1.11960745e+00 -5.71772695e-01 4.77776587e-01 6.02494925e-02 -8.90909433e-01 -5.42729869e-02 -1.96026921e-01 7.19461888e-02 -8.33150744e-03 5.96271932e-01 -2.44820759e-01 2.82412738e-01 8.09245169e-01 1.16099536e+00 -2.13542193e-01 1.01859832e+00 -1.98493302e-02 6.29747152e-01 -3.10118705e-01 2.39108160e-01 6.22425489e-02 -1.57942995e-01 3.70271355e-01 1.00080526e+00 7.71530345e-02 3.05322111e-01 -1.18917659e-01 7.05560088e-01 -3.96246947e-02 2.52178460e-01 -6.76783502e-01 5.35267107e-02 1.52584612e-01 1.36814559e+00 -4.19917792e-01 4.07629237e-02 -5.83989501e-01 1.20199442e+00 -7.66286030e-02 3.77955168e-01 -6.71873450e-01 -1.91115260e-01 5.99435508e-01 1.95519716e-01 1.28417253e-01 4.75274883e-02 1.30808011e-01 -1.21196198e+00 1.39693350e-01 -7.80084133e-01 3.71156603e-01 -8.03086817e-01 -1.47705841e+00 3.81581366e-01 -9.78776962e-02 -1.27143574e+00 -1.28817573e-01 -5.09176850e-01 -4.48620081e-01 6.98840737e-01 -1.90701914e+00 -1.14162624e+00 -5.81508934e-01 1.03058290e+00 1.49739146e-01 -1.59142122e-01 1.03732073e+00 3.31610620e-01 -4.06987309e-01 5.28495073e-01 1.64135829e-01 -5.62799796e-02 1.00281775e+00 -1.06107938e+00 -2.49462470e-01 4.76090521e-01 1.09393530e-01 3.85144442e-01 3.92228246e-01 -2.48364300e-01 -1.48106515e+00 -9.05884683e-01 4.06160384e-01 -1.88066125e-01 5.46870768e-01 -4.04252172e-01 -9.70728993e-01 2.75514901e-01 -4.56353612e-02 9.37792599e-01 1.22096097e+00 -9.27381292e-02 -2.86661446e-01 -4.37503368e-01 -1.35074365e+00 1.24134921e-01 1.03319204e+00 -7.64084995e-01 -2.17512116e-01 4.50514555e-01 -6.24928251e-02 -2.93198764e-01 -9.51070845e-01 4.52661693e-01 7.66397417e-01 -1.04102373e+00 8.06821883e-01 -4.82693642e-01 2.84828246e-01 2.02212427e-02 1.43414110e-01 -1.33484864e+00 -4.96373419e-03 -1.03125215e+00 -2.69770026e-01 9.72223222e-01 -8.19555148e-02 -5.74207664e-01 1.03667903e+00 4.85837013e-01 7.73959979e-02 -7.83500671e-01 -1.11634707e+00 -6.43178999e-01 -2.73541719e-01 -3.43965203e-01 7.68471137e-02 1.17239082e+00 3.82103086e-01 1.14843294e-01 -7.44422019e-01 -1.36857390e-01 7.85409808e-01 -1.58589214e-01 3.40992272e-01 -1.20845854e+00 -3.39981169e-02 -7.49931857e-02 -8.04962993e-01 -7.19554722e-01 4.62829709e-01 -6.47566974e-01 1.54070467e-01 -9.90402699e-01 1.03049070e-01 -1.76125228e-01 -6.48327291e-01 4.55447227e-01 1.42549230e-02 8.50776136e-01 -2.41545707e-01 2.32616380e-01 -4.93501931e-01 7.74332643e-01 1.26466489e+00 1.75879598e-01 -5.70621133e-01 -1.06010817e-01 -4.86732334e-01 3.90688211e-01 6.36398196e-01 -1.84834525e-01 -5.33624947e-01 2.62311995e-01 -7.96275621e-04 2.18559429e-01 5.59344769e-01 -1.11999691e+00 2.16174319e-01 1.10086285e-01 6.32808924e-01 -2.56365985e-01 5.49550474e-01 -7.89603353e-01 -7.20253512e-02 2.13272363e-01 -5.45959294e-01 -4.67614561e-01 2.17177629e-01 6.91153407e-01 -2.39757985e-01 2.30557621e-01 9.98474896e-01 -2.29579121e-01 -4.99344409e-01 4.93858784e-01 -2.83716887e-01 -1.70932598e-02 9.16859925e-01 -4.39238101e-01 1.53059021e-01 -5.64582884e-01 -9.52600896e-01 -1.30981147e-01 7.44314417e-02 2.29643390e-01 8.74792814e-01 -1.35572982e+00 -6.70001924e-01 5.93070149e-01 1.96572199e-01 -4.82273519e-01 6.26086295e-01 1.45111823e+00 -2.44044751e-01 2.71719068e-01 -4.96727437e-01 -8.89922619e-01 -1.30128884e+00 4.77617472e-01 6.44771159e-01 2.01679453e-01 -8.66921008e-01 7.78292537e-01 9.69578419e-03 -2.97653049e-01 2.59941846e-01 4.71107487e-04 -4.98083025e-01 2.12967530e-01 6.70928597e-01 1.64018810e-01 1.79814547e-01 -7.98665166e-01 -3.88334543e-01 7.37193584e-01 1.22818105e-01 -1.00822086e-02 1.62349761e+00 -2.33930290e-01 -5.34490775e-03 5.51106513e-01 1.54759991e+00 -3.89467120e-01 -1.73113382e+00 -4.91660625e-01 -2.18730852e-01 -5.39186478e-01 3.76922458e-01 -7.50156283e-01 -1.34551477e+00 9.74083960e-01 1.12092996e+00 -1.06746159e-01 1.57992852e+00 -3.42326880e-01 5.93016267e-01 4.33568805e-02 6.65789992e-02 -1.11107361e+00 4.37688440e-01 3.26832011e-02 7.81179845e-01 -1.38696158e+00 3.01634725e-02 -2.86773622e-01 -8.18516433e-01 1.14649761e+00 3.09069216e-01 -1.83116086e-02 8.04319203e-01 1.15561314e-01 2.54584372e-01 -3.46579731e-01 -6.87218428e-01 -2.05003321e-01 4.83201534e-01 9.52439606e-01 5.53637624e-01 -5.44757724e-01 -2.54777908e-01 2.41583094e-01 2.91696995e-01 4.17037249e-01 2.15294972e-01 7.98931420e-01 1.30630791e-01 -7.51092792e-01 -8.06644410e-02 4.54272568e-01 -7.13623285e-01 -8.82370397e-02 -1.95660546e-01 5.81492245e-01 -4.70913686e-02 7.17910707e-01 6.03868859e-04 -2.37873659e-01 7.65936971e-02 3.47717404e-01 8.01303625e-01 -4.26418602e-01 -3.67166519e-01 3.99322629e-01 -2.46000573e-01 -8.13960850e-01 -1.16450131e+00 -8.97829771e-01 -8.99547398e-01 3.95280644e-02 -1.63723379e-01 -1.41817108e-02 6.51959717e-01 9.39966619e-01 4.42731768e-01 2.92318255e-01 1.04487443e+00 -1.11036694e+00 -3.45536262e-01 -9.19768751e-01 -8.78820896e-01 6.06032431e-01 6.67940259e-01 -7.42151022e-01 -6.48720860e-01 2.62785256e-01]
[13.854339599609375, 2.629767656326294]
807dc55b-e69e-48f8-aa13-7b9c1f319cf7
efficient-passive-membership-inference-attack
2111.0043
null
https://arxiv.org/abs/2111.00430v1
https://arxiv.org/pdf/2111.00430v1.pdf
Efficient passive membership inference attack in federated learning
In cross-device federated learning (FL) setting, clients such as mobiles cooperate with the server to train a global machine learning model, while maintaining their data locally. However, recent work shows that client's private information can still be disclosed to an adversary who just eavesdrops the messages exchanged between the client and the server. For example, the adversary can infer whether the client owns a specific data instance, which is called a passive membership inference attack. In this paper, we propose a new passive inference attack that requires much less computation power and memory than existing methods. Our empirical results show that our attack achieves a higher accuracy on CIFAR100 dataset (more than $4$ percentage points) with three orders of magnitude less memory space and five orders of magnitude less calculations.
['Giovanni Neglia', 'Chuan Xu', 'Oualid Zari']
2021-10-31
null
null
null
null
['membership-inference-attack']
['computer-vision']
[-1.24643348e-01 1.58242464e-01 -3.92446756e-01 -3.75679910e-01 -1.25043130e+00 -1.22459614e+00 1.72863439e-01 -2.77334489e-02 -4.98239696e-01 8.58231664e-01 -2.85660744e-01 -9.19630706e-01 2.20031232e-01 -8.67334723e-01 -1.13259375e+00 -9.54356372e-01 -1.32413790e-01 5.61268151e-01 4.14858162e-01 3.98817301e-01 -2.07520679e-01 4.43054855e-01 -1.02734852e+00 4.97845352e-01 1.19124614e-01 1.32040751e+00 -5.61389446e-01 7.03752041e-01 1.72295179e-02 9.06458855e-01 -8.07301164e-01 -8.18253040e-01 2.84805238e-01 9.96562615e-02 -1.21666181e+00 -4.11761761e-01 3.24827611e-01 -6.44091010e-01 -4.26970184e-01 1.26993299e+00 2.41964772e-01 -3.70820463e-01 -8.44700709e-02 -1.49316525e+00 -8.09838064e-03 1.05185270e+00 -1.26155227e-01 -6.30088896e-02 9.87437591e-02 1.28057241e-01 7.62111306e-01 -7.68575668e-02 3.01780164e-01 8.29316914e-01 5.17512739e-01 6.78206325e-01 -1.02713990e+00 -1.25562048e+00 4.09268513e-02 3.45488712e-02 -1.51049304e+00 -6.39410496e-01 5.80233157e-01 1.36530846e-01 6.34557068e-01 9.41355705e-01 -1.39762148e-01 1.10678697e+00 -1.75489306e-01 8.12561989e-01 1.08867347e+00 -1.89615898e-02 5.74346662e-01 4.82432067e-01 2.13988394e-01 6.99535489e-01 5.04497826e-01 -1.04386434e-01 -4.31802273e-01 -1.13957226e+00 2.40473464e-01 1.53798610e-01 -2.70241201e-01 -2.47245029e-01 -8.46974373e-01 8.66708279e-01 -1.14040207e-02 -5.31219356e-02 8.09310079e-02 4.08394903e-01 6.06759429e-01 6.67308688e-01 2.09709495e-01 -1.81600839e-01 -1.02283537e+00 -3.28180611e-01 -3.85998547e-01 -1.54040230e-03 1.44849765e+00 7.96696961e-01 6.21415734e-01 -2.86641687e-01 5.19483685e-01 -1.88973900e-02 3.56487095e-01 6.84117496e-01 3.82697016e-01 -1.07777369e+00 7.41392195e-01 3.51891488e-01 2.93640673e-01 -7.39001751e-01 1.97583199e-01 -1.71415523e-01 -8.06514740e-01 1.44621536e-01 6.03056133e-01 -6.22003853e-01 -5.97181134e-02 1.90407884e+00 6.89144015e-01 5.61697900e-01 2.08433419e-01 7.99396634e-01 2.24995092e-01 1.87420800e-01 2.63196621e-02 -1.72086462e-01 1.33340144e+00 -8.17583859e-01 -4.17348862e-01 5.36915138e-02 9.53398347e-01 -4.07772630e-01 7.57973194e-01 4.90190804e-01 -1.05899596e+00 1.29932314e-01 -1.06425846e+00 1.42743140e-01 -4.71901745e-01 -3.88688594e-01 1.04588926e+00 1.17735159e+00 -7.21630871e-01 3.84967268e-01 -1.21336114e+00 2.84510165e-01 8.17963779e-01 9.29599404e-01 -5.85707486e-01 1.42503276e-01 -1.27375245e+00 -2.26321463e-02 1.36412829e-01 -5.70867062e-01 -9.33287799e-01 -6.63364530e-01 -4.19398606e-01 1.36076421e-01 3.08106989e-01 -5.60316980e-01 1.42158735e+00 -6.84163034e-01 -1.60993731e+00 9.20912147e-01 -1.44707382e-01 -7.92178094e-01 7.04852879e-01 1.48523357e-02 -5.48436344e-01 5.04038036e-02 -4.59494919e-01 -3.40653598e-01 6.75048828e-01 -1.01685631e+00 -9.50669289e-01 -8.63391280e-01 4.79062080e-01 -2.64570713e-01 -5.20041585e-01 3.34674641e-02 -3.16201150e-01 -2.01375723e-01 4.90163863e-02 -9.46612477e-01 2.18346920e-02 1.30907774e-01 -4.23752695e-01 -2.11063534e-01 1.56776512e+00 -1.71709225e-01 1.06539166e+00 -2.25624967e+00 -4.91800636e-01 3.12528580e-01 4.72895294e-01 5.21826923e-01 4.00489181e-01 1.09660983e-01 4.11619425e-01 3.28175366e-01 1.20442189e-01 -5.08170247e-01 1.21865869e-01 3.13298047e-01 -6.13442421e-01 7.63341069e-01 -7.73077369e-01 8.31682384e-01 -6.17175162e-01 -3.58584076e-01 -2.87893027e-01 3.14900100e-01 -6.00014031e-01 1.46585017e-01 -1.96876973e-01 3.42770308e-01 -6.97576225e-01 4.73708242e-01 1.02851140e+00 -7.94322968e-01 7.12970853e-01 9.83857960e-02 5.46767056e-01 6.98044360e-01 -1.24567771e+00 1.45281410e+00 -4.96054083e-01 2.87437797e-01 5.68467259e-01 -6.83626056e-01 2.75525361e-01 7.26762176e-01 2.95216143e-01 -3.08205664e-01 3.79303247e-01 1.60829231e-01 -4.83357668e-01 -1.36024624e-01 -3.22609156e-01 1.13204181e-01 -1.78941175e-01 1.15665686e+00 -2.90138572e-01 9.06820893e-01 -9.04048085e-01 1.61868781e-01 1.51308203e+00 -6.17313266e-01 -1.97993428e-01 6.60133809e-02 6.02322340e-01 -3.49288762e-01 4.34631675e-01 1.04300439e+00 -3.89505476e-01 -1.36723161e-01 3.61086309e-01 -6.17766261e-01 -4.61368471e-01 -1.19969726e+00 2.18072504e-01 1.22427309e+00 2.15383723e-01 -6.00842237e-01 -9.21585917e-01 -1.46434629e+00 2.76560009e-01 3.62640679e-01 -1.97103813e-01 -1.58362776e-01 -6.63208127e-01 -3.64726245e-01 1.27330959e+00 3.95965457e-01 1.03613019e+00 -2.53587216e-01 -4.57117379e-01 -1.55473635e-01 -4.26707268e-01 -1.19887900e+00 -5.57214022e-01 7.73312673e-02 -8.73228073e-01 -1.20334685e+00 1.74828291e-01 -4.74109948e-01 5.67460299e-01 3.93093228e-01 8.69334102e-01 2.26647213e-01 1.94141686e-01 1.91854745e-01 2.35888198e-01 -3.10078532e-01 -3.59919459e-01 2.70854414e-01 2.07430989e-01 2.49013394e-01 7.37835824e-01 -7.01495230e-01 -7.23717809e-01 4.03318584e-01 -8.64416301e-01 -4.23797548e-01 9.34058428e-02 4.33242410e-01 4.17636544e-01 3.33133072e-01 3.79188955e-01 -1.53117859e+00 2.56436527e-01 -6.93292260e-01 -8.48178983e-01 2.35992193e-01 -6.70992911e-01 1.02301165e-01 9.14052784e-01 -7.37809002e-01 -9.68051434e-01 3.28658372e-01 -1.23613156e-01 -5.08946955e-01 -3.22122395e-01 -1.09873377e-01 -7.53301203e-01 -5.06487370e-01 4.82677072e-01 2.08067968e-01 -3.31875607e-02 -8.01685691e-01 1.08097650e-01 1.01282811e+00 5.59210479e-01 -7.73229063e-01 9.41522479e-01 8.77117336e-01 -4.67757732e-02 -7.21117035e-02 -5.75209975e-01 -1.49358600e-01 -1.01384327e-01 3.95817935e-01 3.33524764e-01 -8.22589874e-01 -1.73757648e+00 5.76940060e-01 -8.79061580e-01 -3.28844666e-01 1.24983070e-02 2.82996893e-01 -1.39124110e-01 3.96088064e-01 -7.38432288e-01 -6.39887869e-01 -6.37700617e-01 -9.74334240e-01 6.96053326e-01 5.06389923e-02 -2.32220744e-03 -1.07885110e+00 -2.03355789e-01 8.81735206e-01 5.46086431e-01 3.24823298e-02 7.62612462e-01 -1.15642858e+00 -8.92340958e-01 -5.93208909e-01 3.62654775e-03 1.22470751e-01 8.31478834e-02 -5.67075729e-01 -1.34130466e+00 -5.14186680e-01 4.81411308e-01 -2.49181911e-01 1.82481915e-01 -4.04065698e-01 1.81826663e+00 -1.37952137e+00 -4.66436297e-01 9.88491237e-01 1.38356507e+00 7.25947842e-02 5.57598472e-01 -3.17383036e-02 4.54141706e-01 -4.91467826e-02 9.19855461e-02 4.24614847e-01 2.73553908e-01 5.39414346e-01 3.87779862e-01 2.25362629e-01 3.08927685e-01 -5.14142573e-01 3.12008709e-01 3.14522624e-01 4.48017508e-01 -3.04541826e-01 -4.79728758e-01 9.34869051e-02 -1.77459598e+00 -7.84339368e-01 2.82097280e-01 2.41911411e+00 1.12609005e+00 2.65690953e-01 2.12150831e-02 2.83157736e-01 2.99395621e-01 -4.30998802e-02 -8.87897849e-01 -3.39452744e-01 1.02009870e-01 2.89253712e-01 1.17820334e+00 6.36773169e-01 -1.14747214e+00 8.31383109e-01 6.08943081e+00 9.97982323e-01 -1.30547047e+00 6.87963545e-01 8.93857777e-01 -3.03308278e-01 -1.89625919e-01 3.10042296e-02 -6.55976713e-01 8.19145501e-01 1.34811807e+00 -2.97985315e-01 7.17144608e-01 1.31921601e+00 -4.99900758e-01 1.89649910e-01 -1.13490450e+00 1.14722526e+00 -3.33175719e-01 -1.61749780e+00 -2.76355326e-01 5.24793565e-01 3.56561095e-01 3.26511443e-01 1.20821498e-01 2.62197349e-02 8.12063992e-01 -8.97474587e-01 2.08554864e-01 1.91457346e-01 9.17733967e-01 -1.16907585e+00 6.76737905e-01 8.30839038e-01 -9.96224999e-01 -1.22646928e-01 -1.74890488e-01 2.47493926e-02 -3.84064376e-01 1.47683516e-01 -6.43697441e-01 2.39948913e-01 7.55041242e-01 -4.59241927e-01 -3.80924046e-01 3.44917148e-01 8.71672034e-02 1.17492366e+00 -7.59824336e-01 6.58368915e-02 4.21237350e-02 9.24482793e-02 3.51218611e-01 9.11377549e-01 -1.48284778e-01 4.82251018e-01 -7.67593097e-04 6.08532250e-01 -8.26462448e-01 -2.53572285e-01 -5.92239499e-01 2.35258371e-01 8.31598103e-01 1.02066100e+00 -1.46141902e-01 -4.78700548e-01 -3.80109966e-01 1.33960390e+00 3.71463478e-01 1.72331214e-01 -9.98434722e-01 -4.37486231e-01 1.16394436e+00 2.92921066e-02 2.02525184e-01 -8.28132853e-02 -2.25741953e-01 -1.28172433e+00 4.20206785e-02 -9.68340755e-01 7.96646595e-01 -9.93592888e-02 -1.22825134e+00 3.56142670e-01 -6.19802475e-01 -6.93434834e-01 -1.50311843e-01 -2.18340725e-01 -6.55358255e-01 5.86349607e-01 -1.05226111e+00 -1.17497730e+00 2.10589141e-01 1.33487189e+00 -2.38148361e-01 -2.17509747e-01 1.45167291e+00 4.80883211e-01 -4.47129428e-01 1.64546168e+00 2.97495902e-01 6.53444409e-01 2.86847502e-01 -8.69507134e-01 2.73213416e-01 6.47781610e-01 3.13555181e-01 9.51660037e-01 3.53738070e-01 -3.13241422e-01 -1.99044991e+00 -1.19218063e+00 8.17580879e-01 -5.40880084e-01 5.43880045e-01 -4.90389109e-01 -7.84491837e-01 1.09459174e+00 -2.01567262e-01 6.55291915e-01 1.16374779e+00 3.24963816e-02 -1.01230502e+00 -4.77177233e-01 -1.82570171e+00 3.65282357e-01 7.53762126e-01 -1.15421486e+00 1.12737119e-01 4.62381124e-01 9.11860764e-01 -4.44530249e-01 -8.52156281e-01 4.52067479e-02 6.71734691e-01 -7.26789594e-01 8.93110931e-01 -1.20157361e+00 -7.69634366e-01 -1.68875545e-01 -4.35181379e-01 -4.34417337e-01 1.69465035e-01 -1.22813356e+00 -1.03930974e+00 1.15240169e+00 3.23919058e-01 -1.20396435e+00 1.41565073e+00 1.07540679e+00 8.42518151e-01 -4.38173473e-01 -1.54380488e+00 -7.01032996e-01 1.61464527e-01 -5.93201995e-01 1.22904420e+00 1.06282294e+00 7.54478946e-02 7.82345608e-02 -2.11446479e-01 7.17056930e-01 9.49394107e-01 5.09251952e-02 8.99212658e-01 -1.04854131e+00 -5.49310684e-01 9.25340205e-02 -2.72790551e-01 -1.08067131e+00 4.15056139e-01 -7.79630065e-01 -6.31007135e-01 -4.47565317e-01 5.67942187e-02 -8.61796498e-01 -6.44120991e-01 8.27846706e-01 3.82636964e-01 3.70105863e-01 2.12063864e-02 2.05086872e-01 -9.96493042e-01 -2.83868220e-02 4.83206928e-01 -1.50864229e-01 1.68325618e-01 7.99945831e-01 -9.38092887e-01 7.22196937e-01 1.17186844e+00 -8.04687142e-01 -5.01099467e-01 -4.08559084e-01 1.55037433e-01 2.26531744e-01 4.07123327e-01 -9.75109398e-01 6.35272801e-01 -3.87713015e-02 1.08773611e-01 -3.15379947e-01 2.52574593e-01 -1.39007187e+00 3.44443560e-01 5.70059001e-01 -2.13660091e-01 -1.57942742e-01 -1.89154968e-02 7.22205758e-01 1.97709903e-01 2.58908987e-01 7.20943153e-01 2.18320154e-02 1.73587412e-01 5.70363104e-01 -1.59571543e-01 2.25285254e-02 9.98896420e-01 3.09159458e-01 -6.39739156e-01 -5.51446557e-01 -5.84458470e-01 4.52345721e-02 5.57533622e-01 1.59436539e-01 3.06284726e-01 -1.13161922e+00 -1.82928041e-01 4.52888191e-01 -3.00800413e-01 -1.89567178e-01 1.08205631e-01 6.05552852e-01 -1.67037517e-01 5.94878197e-01 3.96196365e-01 -2.53652275e-01 -1.75318182e+00 7.63154864e-01 4.38596159e-01 -2.07334906e-01 -6.32181525e-01 7.96541393e-01 -2.26907253e-01 -3.72917712e-01 7.51910686e-01 3.99409011e-02 6.45752788e-01 -5.26639700e-01 9.97427940e-01 5.14302313e-01 2.97570229e-01 -4.06612784e-01 -4.40697193e-01 -1.17351197e-01 -1.00963973e-01 -1.49198458e-01 8.46474826e-01 -2.71897525e-01 -6.71556592e-02 4.49368656e-02 1.90960276e+00 5.10058761e-01 -1.07375073e+00 -7.60085225e-01 -6.05114587e-02 -6.89816117e-01 1.24393716e-01 -8.45512748e-01 -1.47539592e+00 5.80809712e-01 7.18194783e-01 2.10172936e-01 9.99771118e-01 1.82537884e-01 1.32062364e+00 6.47985756e-01 1.12365508e+00 -6.83436513e-01 -5.17756104e-01 3.34757380e-02 -8.54688883e-02 -1.11816812e+00 -3.43391865e-01 -2.54435241e-01 -3.78384173e-01 7.04219222e-01 5.26174247e-01 2.81174898e-01 1.09080315e+00 7.24048615e-01 1.18286446e-01 3.47162299e-02 -1.10553145e+00 5.78345299e-01 -4.16546345e-01 4.20231909e-01 -3.72463942e-01 3.47605199e-01 2.69779086e-01 1.23249781e+00 -8.86110216e-02 9.00797099e-02 -4.95072939e-02 1.10614133e+00 -2.04197913e-01 -1.40491796e+00 -2.22198486e-01 1.19602822e-01 -1.19732845e+00 3.00883591e-01 -2.99465299e-01 3.63947958e-01 9.72652156e-03 1.04594421e+00 -6.13128059e-02 -6.93397045e-01 -2.40435988e-01 2.84908295e-01 -2.11111102e-02 -2.01771215e-01 -7.72775531e-01 -3.04472566e-01 1.60903262e-03 -9.57357526e-01 5.45242019e-02 -2.65762031e-01 -1.52401519e+00 -1.04926038e+00 -3.08075100e-01 6.37570441e-01 7.82414436e-01 7.84587562e-01 7.67844677e-01 -2.94857055e-01 1.29289567e+00 2.52803802e-01 -7.83385277e-01 -3.90787542e-01 -5.40901899e-01 1.63264379e-01 3.96558940e-01 -6.27319664e-02 -8.38325024e-01 -2.06135869e-01]
[5.805443286895752, 6.824004173278809]
82d9b3c4-db22-4aea-b5bf-0e6dc61f9eba
multilingual-distributed-representations
1312.6173
null
http://arxiv.org/abs/1312.6173v4
http://arxiv.org/pdf/1312.6173v4.pdf
Multilingual Distributed Representations without Word Alignment
Distributed representations of meaning are a natural way to encode covariance relationships between words and phrases in NLP. By overcoming data sparsity problems, as well as providing information about semantic relatedness which is not available in discrete representations, distributed representations have proven useful in many NLP tasks. Recent work has shown how compositional semantic representations can successfully be applied to a number of monolingual applications such as sentiment analysis. At the same time, there has been some initial success in work on learning shared word-level representations across languages. We combine these two approaches by proposing a method for learning distributed representations in a multilingual setup. Our model learns to assign similar embeddings to aligned sentences and dissimilar ones to sentence which are not aligned while not requiring word alignments. We show that our representations are semantically informative and apply them to a cross-lingual document classification task where we outperform the previous state of the art. Further, by employing parallel corpora of multiple language pairs we find that our model learns representations that capture semantic relationships across languages for which no parallel data was used.
['Karl Moritz Hermann', 'Phil Blunsom']
2013-12-20
null
null
null
null
['cross-lingual-document-classification']
['natural-language-processing']
[ 1.12216242e-01 1.34497866e-01 -4.18314099e-01 -7.64492154e-01 -1.09191716e+00 -8.04276466e-01 7.30363727e-01 6.15315080e-01 -4.44037944e-01 6.35639787e-01 8.64292324e-01 -1.65204138e-01 -1.52719617e-01 -8.13744724e-01 -6.34444475e-01 -5.60138583e-01 1.51482418e-01 7.10551381e-01 -2.62245446e-01 -4.89329219e-01 1.46248877e-01 2.36604869e-01 -1.36129868e+00 5.90117276e-01 4.97580916e-01 4.23545361e-01 2.72078782e-01 3.54695112e-01 -7.43718445e-01 8.24302912e-01 -4.25628841e-01 -4.69592005e-01 2.40257010e-01 -5.20611823e-01 -1.11481726e+00 -3.53408217e-01 7.94232666e-01 1.04114480e-01 4.17226255e-02 1.03927577e+00 4.27791476e-01 3.05396765e-01 8.55683923e-01 -9.05625701e-01 -1.30070841e+00 7.98618495e-01 -6.54222548e-01 3.02111149e-01 3.71518761e-01 -6.98619902e-01 1.84408295e+00 -8.13734949e-01 7.73691833e-01 1.55595565e+00 7.35661089e-01 4.32776213e-01 -1.41543126e+00 -4.72124159e-01 2.57027864e-01 1.06814101e-01 -1.16067898e+00 -3.10494393e-01 6.01029217e-01 -3.14556479e-01 1.52962506e+00 -1.73993200e-01 6.12706095e-02 1.05251837e+00 1.39703527e-01 6.70817733e-01 8.69318426e-01 -7.89978027e-01 2.74248626e-02 4.35583778e-02 2.99416721e-01 4.30756658e-01 1.95729598e-01 -3.87918770e-01 -6.04565680e-01 -3.19700718e-01 4.26865429e-01 1.20130308e-01 -5.78262955e-02 -5.03654540e-01 -1.33251023e+00 1.37274241e+00 5.59475064e-01 8.83166492e-01 -2.08545595e-01 2.93401837e-01 6.27427816e-01 6.37292862e-01 8.48620594e-01 6.03828192e-01 -9.08492506e-01 3.44733894e-02 -5.26851833e-01 2.34945744e-01 8.48615468e-01 7.81930804e-01 9.75948811e-01 -2.79821634e-01 2.24261895e-01 1.37950921e+00 1.24669023e-01 4.11153018e-01 1.02927816e+00 -7.44295299e-01 5.83614171e-01 4.50478017e-01 -2.68900633e-01 -1.19847095e+00 -2.28722155e-01 5.44857047e-02 -5.51792681e-01 -2.95532197e-01 7.46562183e-02 -4.65647615e-02 -3.63912493e-01 2.22136855e+00 -3.01543288e-02 -1.03826322e-01 5.59692740e-01 6.81107461e-01 4.72292542e-01 7.76051819e-01 2.38159850e-01 1.23368286e-01 1.42456675e+00 -9.41463172e-01 -5.78205347e-01 -4.28684115e-01 1.18535161e+00 -9.00895298e-01 8.97594094e-01 8.72758403e-02 -7.66114295e-01 -4.48193818e-01 -9.12145734e-01 -6.63880706e-01 -7.62492478e-01 -2.54834861e-01 9.46440697e-01 5.60939610e-01 -1.27102637e+00 5.14564991e-01 -5.29040039e-01 -9.39286411e-01 1.91496253e-01 2.18975469e-01 -6.38584316e-01 -3.54386777e-01 -1.13644540e+00 1.19951069e+00 2.64607251e-01 -4.24050957e-01 -3.64653468e-01 -6.22720897e-01 -1.23076272e+00 2.34604757e-02 -5.39020374e-02 -7.32597768e-01 1.19151294e+00 -1.29187191e+00 -1.11932194e+00 1.07171941e+00 -3.98476273e-01 -5.39292276e-01 -5.36195755e-01 -2.82173544e-01 -2.68635392e-01 1.56854019e-02 5.17939568e-01 8.86103988e-01 4.19048697e-01 -7.94712365e-01 -3.76980841e-01 -4.10293877e-01 1.33038789e-01 5.52336872e-01 -8.18639934e-01 3.29291224e-01 1.68909013e-01 -7.40999401e-01 -2.50286888e-02 -7.88616657e-01 -2.47207776e-01 -2.15672851e-01 1.71302766e-01 -4.91205662e-01 6.24662101e-01 -5.72480440e-01 6.06698215e-01 -2.08312559e+00 6.25857294e-01 -3.34948674e-03 -1.01605318e-01 -1.48990765e-01 -5.24938345e-01 8.24504614e-01 -1.92153722e-01 2.80900955e-01 -4.09053147e-01 -5.11309564e-01 1.74879670e-01 7.56953537e-01 -5.79113960e-01 3.65037262e-01 3.47515762e-01 9.77044582e-01 -9.84255373e-01 -2.68387586e-01 -5.88064343e-02 5.69634020e-01 -6.84957385e-01 -2.98173656e-03 -1.76781967e-01 2.13056773e-01 -3.68164986e-01 2.21524909e-01 2.29770362e-01 -1.96698502e-01 6.67570531e-01 4.82289679e-02 1.63235202e-01 9.07348394e-01 -8.39659274e-01 2.40660214e+00 -1.05393255e+00 7.09593475e-01 -1.77846536e-01 -1.48835886e+00 1.00705647e+00 6.41223192e-01 4.47923660e-01 -6.99066997e-01 -8.74923840e-02 3.63931328e-01 -9.83229503e-02 -2.32951269e-01 7.17381060e-01 -7.36793756e-01 -5.05040288e-01 8.28218460e-01 5.77150643e-01 -3.52315485e-01 -8.37589726e-02 2.82159477e-01 9.69067574e-01 1.98444515e-01 5.71164131e-01 -6.19490445e-01 3.30919057e-01 8.74500722e-02 3.64833593e-01 3.62000972e-01 1.03200957e-01 4.67226624e-01 4.40543950e-01 -5.12305021e-01 -1.06943345e+00 -1.01996052e+00 -2.07845017e-01 1.46296453e+00 -1.38472930e-01 -6.74533248e-01 -2.10924566e-01 -6.01695657e-01 1.42139480e-01 7.63533950e-01 -6.42838061e-01 -2.11592969e-02 -5.50231278e-01 -5.44145346e-01 4.27113771e-01 8.79551768e-01 -2.50048131e-01 -1.03558624e+00 5.82429208e-02 1.52282283e-01 -7.62084723e-02 -1.11304998e+00 -2.79224455e-01 4.46973622e-01 -8.67992461e-01 -8.41618419e-01 -5.91259181e-01 -1.27225268e+00 5.03836274e-01 3.86177987e-01 1.57700253e+00 -2.27108389e-01 -2.89022207e-01 6.94155931e-01 -3.55577171e-01 -4.09915596e-01 -2.02890217e-01 1.86615840e-01 7.19388649e-02 -1.66867137e-01 9.64457273e-01 -6.78157926e-01 -1.74557671e-01 -1.57735467e-01 -9.18404400e-01 -2.46495605e-01 3.58726084e-01 9.09996867e-01 4.46094841e-01 -5.17996371e-01 9.35259819e-01 -1.19159508e+00 8.60679448e-01 -1.02083182e+00 -7.01926127e-02 9.44050029e-02 -1.74905494e-01 3.84570271e-01 6.61182642e-01 -9.95736048e-02 -9.54347312e-01 -2.48126894e-01 -1.11425355e-01 -9.56409648e-02 -2.21709013e-01 7.57166922e-01 1.34822667e-01 2.63013333e-01 6.73925400e-01 -2.20940873e-01 6.38234019e-02 -4.65338886e-01 9.22464848e-01 6.74267828e-01 9.53994319e-02 -1.04415715e+00 2.82804370e-01 3.81962955e-01 -1.81066781e-01 -9.10451412e-01 -1.06259739e+00 -7.61733174e-01 -8.03200960e-01 7.36787796e-01 1.02106631e+00 -1.18896377e+00 1.05912603e-01 -3.47370416e-01 -1.25719631e+00 1.03749871e-01 -5.89203298e-01 5.61935723e-01 -6.55540109e-01 3.11702639e-01 -5.40089071e-01 -1.80326894e-01 -1.94485173e-01 -8.08409929e-01 1.29445159e+00 -1.45624518e-01 -6.75713003e-01 -1.67179775e+00 6.19806528e-01 2.52652079e-01 5.42864203e-01 -1.36616960e-01 1.24411178e+00 -1.11447537e+00 -9.15845260e-02 -3.62858735e-02 -7.36273676e-02 5.34221828e-01 6.09696627e-01 -5.43647051e-01 -8.51076126e-01 -3.57281595e-01 -1.01903923e-01 -7.91419566e-01 8.03718030e-01 1.96690589e-01 7.25568831e-01 -5.31866774e-02 -1.82703823e-01 3.25034022e-01 1.56470466e+00 -3.38099688e-01 2.74735183e-01 2.16317683e-01 7.31503785e-01 9.54346836e-01 6.63208291e-02 -7.65741691e-02 6.24766827e-01 5.14250875e-01 -1.21812329e-01 5.72424270e-02 -1.60697386e-01 -1.84796795e-01 4.14915562e-01 1.26309681e+00 2.51365155e-01 -1.79165341e-02 -9.36792493e-01 1.10768533e+00 -1.87342012e+00 -8.99841905e-01 1.59520388e-01 1.94305980e+00 9.27833974e-01 -5.73707640e-01 -2.61019349e-01 -4.48238045e-01 5.18492639e-01 3.67035598e-01 -8.75478387e-02 -1.04081833e+00 -2.66319394e-01 8.78663719e-01 3.49219948e-01 5.22928298e-01 -9.79057252e-01 1.17783940e+00 6.61754847e+00 4.38630998e-01 -7.59593487e-01 4.91934627e-01 4.10553455e-01 -3.92977968e-02 -9.08823192e-01 1.65762216e-01 -6.78976536e-01 -1.26222715e-01 1.17763591e+00 -3.02407235e-01 2.33433962e-01 7.10274100e-01 -3.89435858e-01 2.77962863e-01 -1.46007097e+00 9.55369174e-01 5.10413468e-01 -1.28266144e+00 3.20919096e-01 -1.18569829e-01 1.06248271e+00 4.26382989e-01 -1.01992197e-01 4.18392748e-01 7.33618975e-01 -1.33558691e+00 9.26967636e-02 9.11927521e-02 4.51537281e-01 -8.20945919e-01 7.27995336e-01 1.65799648e-01 -1.24381423e+00 6.25608414e-02 -8.22578967e-01 -3.13108742e-01 9.46693346e-02 2.72058606e-01 -6.08993769e-01 7.62705624e-01 5.30712724e-01 1.22535443e+00 -3.57022375e-01 3.30459803e-01 -4.72984284e-01 2.67745793e-01 -2.80663073e-01 -4.75155655e-03 4.65846419e-01 -2.13358417e-01 1.40282542e-01 1.30565798e+00 3.06907147e-01 -3.72329026e-01 3.16983461e-01 7.08722532e-01 -3.17876756e-01 7.04458475e-01 -1.28814614e+00 -3.31504852e-01 2.14385316e-01 1.08464134e+00 -3.82060558e-01 -4.10216033e-01 -1.03739631e+00 1.09289932e+00 7.94366002e-01 1.94269612e-01 -1.19089171e-01 -2.42237404e-01 1.12900281e+00 -4.42227989e-01 4.05532777e-01 -4.18602765e-01 -1.05218783e-01 -1.39310026e+00 -1.19154170e-01 -8.60129416e-01 3.94084722e-01 -6.31722271e-01 -1.99958289e+00 4.55214709e-01 6.72158152e-02 -7.94417679e-01 -6.69497669e-01 -8.35415065e-01 -3.57122749e-01 1.27965808e+00 -1.62000251e+00 -1.35702503e+00 4.47055370e-01 6.27440214e-01 7.19974339e-01 -4.76518005e-01 1.60097218e+00 1.03737839e-01 -2.98231728e-02 2.75404692e-01 1.97292268e-01 1.68524355e-01 9.76122260e-01 -1.30838275e+00 5.34920573e-01 4.81681138e-01 1.03012156e+00 1.07177711e+00 3.16896349e-01 -2.97802031e-01 -1.34254503e+00 -7.43920445e-01 1.59894872e+00 -4.91948903e-01 1.18631756e+00 -4.63179320e-01 -8.26058090e-01 1.17390478e+00 7.15230465e-01 2.31262237e-01 1.42030990e+00 9.32505369e-01 -9.50452566e-01 1.84993431e-01 -7.48714149e-01 3.33766580e-01 8.29334915e-01 -1.02748430e+00 -1.32691109e+00 5.99070668e-01 7.95251012e-01 5.48119023e-02 -8.21377814e-01 -1.25877202e-01 3.06120932e-01 -5.18592060e-01 9.18572307e-01 -1.06530094e+00 6.23616338e-01 3.46514843e-02 -8.51552486e-01 -1.63375401e+00 -2.11859852e-01 -1.29275352e-01 3.51278543e-01 1.36561871e+00 4.85682368e-01 -8.29251945e-01 3.12177539e-01 2.80966908e-01 -6.03489466e-02 -4.79394972e-01 -8.74694943e-01 -7.54603446e-01 9.19567704e-01 -3.14758152e-01 4.23777699e-01 1.36608505e+00 2.19697267e-01 8.36787939e-01 -3.48582044e-02 8.03290009e-02 3.28815997e-01 4.33220237e-01 4.55226362e-01 -1.23892033e+00 -4.50707704e-01 -3.67975146e-01 -8.19394231e-01 -1.03422201e+00 9.69308615e-01 -1.61210859e+00 -1.46569878e-01 -1.58984733e+00 3.88231605e-01 -3.15029979e-01 -5.67880213e-01 6.67864203e-01 2.37244666e-02 2.00854734e-01 5.36434725e-02 2.51072109e-01 -3.48764420e-01 5.68087101e-01 6.63900018e-01 -3.42254847e-01 2.46278957e-01 -8.40850532e-01 -1.12557721e+00 6.60017073e-01 8.25261891e-01 -7.51568973e-01 -7.24296212e-01 -9.65494633e-01 5.55639684e-01 -4.68178272e-01 1.11226477e-01 -4.68426645e-01 1.98404561e-03 4.24622856e-02 1.62883654e-01 -1.73894111e-02 3.05654585e-01 -7.52529621e-01 -4.48274404e-01 -5.32195233e-02 -6.90737247e-01 3.53009224e-01 1.86489522e-01 5.75589538e-01 -7.43554056e-01 -4.75810558e-01 3.33428025e-01 -3.84658515e-01 -9.28147495e-01 -1.32642323e-02 -1.76933810e-01 3.01471800e-01 9.55144167e-01 2.17282727e-01 -1.65122032e-01 -1.97084948e-01 -5.30942142e-01 1.30335167e-01 5.50306320e-01 8.04691792e-01 4.13174510e-01 -1.44989550e+00 -7.22699583e-01 2.67543674e-01 3.47537458e-01 -2.66670495e-01 -5.06332479e-02 4.40887153e-01 -2.99319625e-01 6.53445184e-01 -1.86564624e-01 -4.30947095e-01 -1.11292076e+00 5.69426119e-01 8.84483606e-02 -6.13089204e-02 -8.15463603e-01 9.73047674e-01 3.65665972e-01 -9.28193390e-01 -2.20547289e-01 -2.46011823e-01 -2.05819190e-01 3.58029783e-01 4.62448746e-01 -3.42169195e-01 1.94286630e-01 -8.57996106e-01 -5.35587132e-01 9.17517304e-01 -3.68910789e-01 -1.82923391e-01 1.70853090e+00 -1.31595865e-01 -5.47892749e-01 9.53242779e-01 1.72107995e+00 1.48662180e-01 -5.30329287e-01 -3.70883882e-01 3.00823092e-01 -2.12813407e-01 -1.49676338e-01 -3.44897270e-01 -8.56710374e-01 1.00874197e+00 1.20304935e-01 7.07738996e-02 7.09492862e-01 4.67276633e-01 8.08714807e-01 6.27526343e-01 4.08313662e-01 -8.69016767e-01 9.04495344e-02 7.94746816e-01 7.95075595e-01 -1.31920087e+00 -4.66600396e-02 -1.04443103e-01 -7.82751322e-01 1.07046092e+00 1.35339964e-02 -5.81037521e-01 5.17122328e-01 1.56758621e-01 1.19209997e-01 -2.73677498e-01 -9.28875923e-01 -1.21349037e-01 1.70199588e-01 7.20285296e-01 1.09174478e+00 8.75701457e-02 -4.01394427e-01 4.24645096e-01 -2.53256232e-01 -4.52611864e-01 3.43168169e-01 1.10678530e+00 -2.07121998e-01 -1.76526690e+00 -4.56819216e-05 2.27023438e-01 -6.57447636e-01 -5.57392359e-01 -5.66875935e-01 5.93849063e-01 1.66671611e-02 7.51111925e-01 4.00718331e-01 1.73765779e-01 -8.12691376e-02 5.51155150e-01 6.06803179e-01 -1.26237583e+00 -5.91971338e-01 -2.76388884e-01 2.23851442e-01 -5.79869986e-01 -8.96645129e-01 -6.87307656e-01 -1.24541616e+00 5.85734174e-02 1.33299395e-01 3.56430262e-01 7.76939273e-01 1.07484818e+00 4.76405650e-01 4.64523286e-01 2.86713690e-01 -7.85538971e-01 -6.15220249e-01 -8.73355150e-01 -6.71058238e-01 7.01479793e-01 2.81080186e-01 -4.61264640e-01 -2.91610509e-01 1.37973145e-01]
[10.884676933288574, 9.663629531860352]
a828a2df-7434-43ac-a92c-149505503a68
open-domain-suggestion-mining-leveraging-fine
2007.04297
null
https://arxiv.org/abs/2007.04297v2
https://arxiv.org/pdf/2007.04297v2.pdf
Open Domain Suggestion Mining Leveraging Fine-Grained Analysis
Suggestion mining tasks are often semantically complex and lack sophisticated methodologies that can be applied to real-world data. The presence of suggestions across a large diversity of domains and the absence of large labelled and balanced datasets render this task particularly challenging to deal with. In an attempt to overcome these challenges, we propose a two-tier pipeline that leverages Discourse Marker based oversampling and fine-grained suggestion mining techniques to retrieve suggestions from online forums. Through extensive comparison on a real-world open-domain suggestion dataset, we demonstrate how the oversampling technique combined with transformer based fine-grained analysis can beat the state of the art. Additionally, we perform extensive qualitative and qualitative analysis to give construct validity to our proposed pipeline. Finally, we discuss the practical, computational and reproducibility aspects of the deployment of our pipeline across the web.
['Tanishq Goel', 'Sonika Dahiya', 'Shivang Chopra', 'Shreya Singal']
2020-06-27
null
null
null
null
['suggestion-mining']
['natural-language-processing']
[ 4.03000824e-02 4.50274289e-01 -2.80543774e-01 -3.04030627e-01 -9.68381882e-01 -7.98460245e-01 1.06497908e+00 4.74197835e-01 -2.43855849e-01 7.62770593e-01 8.53573084e-01 -4.20120001e-01 -3.75880718e-01 -5.99602163e-01 -2.64879495e-01 9.95620489e-02 5.63138165e-02 5.58819294e-01 4.01973695e-01 -3.59472156e-01 7.20669866e-01 -9.45188403e-02 -1.80147934e+00 7.84690440e-01 8.71885359e-01 7.84307778e-01 4.10667509e-02 1.40046015e-01 -5.82813740e-01 6.18659198e-01 -6.02875829e-01 -6.22052252e-01 3.03244770e-01 -9.14041698e-02 -1.15311182e+00 1.77563801e-01 6.33311689e-01 -1.67288974e-01 4.65992302e-01 7.66685963e-01 5.56555927e-01 3.24185550e-01 5.29649019e-01 -8.23405206e-01 -3.54140908e-01 9.79182065e-01 -2.91453928e-01 1.01222664e-01 6.46811962e-01 -1.36537582e-01 1.33500540e+00 -9.27725673e-01 1.11304259e+00 1.27902901e+00 6.73781872e-01 3.72019932e-02 -1.15919435e+00 -3.91376466e-01 2.35592961e-01 1.64398402e-01 -9.59989965e-01 -4.67616200e-01 9.86180186e-01 -4.91293162e-01 9.48345244e-01 4.10306156e-01 3.45636368e-01 1.16679204e+00 -7.45008469e-01 4.21263039e-01 1.53418994e+00 -6.88105941e-01 2.68226624e-01 4.28262025e-01 9.97050628e-02 6.05777919e-01 9.81275216e-02 -3.33472282e-01 -8.29777539e-01 -5.56852579e-01 2.86865830e-01 -2.25953817e-01 1.07961064e-02 -9.49612036e-02 -1.23956418e+00 8.39380145e-01 9.25698206e-02 3.86271775e-01 -5.15742660e-01 -2.95982510e-01 7.37568736e-01 2.75247812e-01 9.20574605e-01 8.67735147e-01 -5.29393733e-01 -3.72602642e-01 -1.27767456e+00 7.16801524e-01 1.17265916e+00 7.55816460e-01 7.80912578e-01 -5.95383227e-01 -3.92521560e-01 1.32532752e+00 2.77151585e-01 -3.72047663e-01 5.38840652e-01 -1.33473337e+00 5.96572042e-01 1.00806737e+00 3.02778929e-01 -1.28772736e+00 -3.32528532e-01 -5.38379967e-01 -1.51316747e-01 -1.10221669e-01 7.27932334e-01 4.34850045e-02 -2.55157743e-02 1.43550789e+00 7.12295234e-01 -2.96937495e-01 -4.69826967e-01 8.93628597e-01 9.18934345e-01 8.89268145e-03 7.55217150e-02 -3.13039392e-01 1.64697063e+00 -9.96086419e-01 -7.14741051e-01 -2.34272927e-02 6.97611988e-01 -8.99868548e-01 1.56232607e+00 3.74432236e-01 -9.27142322e-01 -3.33234727e-01 -7.25655794e-01 -4.19556022e-01 -3.32077771e-01 1.87382028e-01 6.13290250e-01 4.09333050e-01 -7.78706789e-01 4.57720101e-01 -3.42775553e-01 -6.80727839e-01 3.42312783e-01 -3.34214978e-02 -2.32967943e-01 5.70738092e-02 -9.54161882e-01 7.28736997e-01 1.65715858e-01 -5.38219512e-01 -2.26061776e-01 -9.06417966e-01 -3.99285108e-01 8.74229670e-02 7.52849162e-01 -5.92680991e-01 1.55582428e+00 -8.07949960e-01 -1.36808693e+00 1.02789688e+00 -1.22857645e-01 -3.73918205e-01 7.30085075e-01 -3.16666275e-01 -1.60952061e-01 1.13257572e-01 5.11600673e-01 4.92326468e-01 4.41712260e-01 -9.47917461e-01 -6.88558221e-01 -2.88978219e-01 2.58487374e-01 1.80177748e-01 -6.43306494e-01 3.61684412e-01 -1.08983412e-01 -7.24559784e-01 -8.97384286e-02 -5.28728485e-01 -1.47223011e-01 -1.91992313e-01 -4.33343023e-01 -6.96365356e-01 6.74239933e-01 -6.17401659e-01 1.35633075e+00 -2.04581594e+00 -3.35848153e-01 2.49092616e-02 3.41756225e-01 -7.42874742e-02 1.30299449e-01 8.59509230e-01 3.64347667e-01 3.27170402e-01 1.77229062e-01 -3.13981891e-01 3.42325032e-01 -1.44719571e-01 -3.71857733e-01 2.20786527e-01 1.12960450e-01 7.21447170e-01 -1.13420963e+00 -7.05942869e-01 -2.35221554e-02 -1.48682892e-01 -6.07827425e-01 2.13836581e-01 -7.25511670e-01 1.73055485e-01 -5.36650777e-01 4.59514380e-01 3.43227506e-01 -4.12814647e-01 5.28581381e-01 -1.32362634e-01 -2.66690731e-01 1.12846768e+00 -1.06754267e+00 1.96012580e+00 -3.18121880e-01 4.14171427e-01 1.23964459e-01 -8.52529943e-01 1.00427687e+00 2.58220732e-01 3.76835406e-01 -8.10821712e-01 -2.44176760e-01 4.23261344e-01 -2.20619962e-01 -6.08696401e-01 7.94562221e-01 -1.40645996e-01 -1.01850204e-01 1.05764747e+00 1.85429156e-02 1.28372014e-01 4.42786247e-01 5.07226110e-01 1.25800145e+00 2.27743044e-01 5.29125273e-01 -4.76530671e-01 4.23265457e-01 4.98742759e-01 5.65347672e-01 6.48455262e-01 -9.43027437e-02 3.02501917e-01 5.32483518e-01 -4.55340594e-01 -1.04256105e+00 -5.26799738e-01 4.30494435e-02 1.53811407e+00 -2.19513670e-01 -9.58660364e-01 -5.81825614e-01 -1.07684851e+00 1.59580261e-03 5.89181960e-01 -5.08026421e-01 5.35419405e-01 -3.86445016e-01 -4.01755244e-01 5.49325645e-01 -6.80254400e-02 5.42276084e-01 -8.14285457e-01 -7.05845237e-01 4.79243100e-01 -5.76075733e-01 -1.28529966e+00 -4.73552160e-02 -2.51326978e-01 -7.52210259e-01 -1.41408026e+00 -2.71925598e-01 -4.34786260e-01 2.50471264e-01 3.40407133e-01 1.55652821e+00 6.36744946e-02 -7.46341273e-02 3.17428440e-01 -6.14759505e-01 -3.04113686e-01 -4.23963636e-01 3.08294982e-01 -1.02950618e-01 -4.12924290e-01 5.10889769e-01 -6.48012102e-01 -7.56499887e-01 4.39978957e-01 -6.89365089e-01 1.34740040e-01 2.62849331e-01 7.21271336e-01 3.31093311e-01 -1.03750899e-01 1.08198917e+00 -1.35411704e+00 1.34662068e+00 -7.60205090e-01 -3.14664364e-01 1.97761655e-01 -8.66558611e-01 -1.18156105e-01 4.30005819e-01 -2.58846372e-01 -1.19667101e+00 -4.00135159e-01 -4.53956006e-03 3.02553803e-01 -2.28166267e-01 7.07876325e-01 2.64439851e-01 4.23376083e-01 1.22986448e+00 -4.77164626e-01 -3.07634547e-02 -9.97086942e-01 7.41782665e-01 9.09707010e-01 3.42798889e-01 -8.36823404e-01 5.24557412e-01 6.80892467e-01 -3.90062571e-01 -4.69589680e-01 -1.08988035e+00 -5.53833306e-01 -3.69479626e-01 -1.38925940e-01 1.32495195e-01 -6.94756031e-01 -4.32759345e-01 -2.30120152e-01 -8.78031492e-01 -3.22287500e-01 -5.34239650e-01 4.96743955e-02 -4.23869699e-01 5.95451534e-01 -3.41082007e-01 -8.35018694e-01 -6.08488500e-01 -5.50607264e-01 7.92092621e-01 8.72701406e-03 -8.59315693e-01 -1.10327208e+00 1.21645696e-01 9.67131317e-01 5.68950057e-01 1.58393517e-01 7.57713795e-01 -1.22991455e+00 -3.18127841e-01 9.22244042e-02 -3.14836830e-01 -1.59120068e-01 1.83830172e-01 3.60062271e-02 -8.71129036e-01 2.30497196e-01 -5.36597297e-02 -6.57788932e-01 5.99143326e-01 -1.23329230e-01 1.09104669e+00 -7.66262293e-01 -1.49853751e-01 -1.29440531e-01 1.00819933e+00 -7.57262826e-01 5.83380274e-02 9.63680685e-01 1.97685108e-01 1.09019172e+00 1.25579798e+00 7.61612535e-01 7.66730368e-01 7.79594898e-01 2.59459674e-01 2.26712987e-01 -3.22227180e-01 -4.24517691e-01 5.99128343e-02 7.40763664e-01 1.67441815e-01 1.02891661e-01 -8.84552717e-01 7.47251689e-01 -1.92786252e+00 -1.04746783e+00 -1.28049836e-01 1.86190593e+00 1.12827873e+00 1.63429305e-01 5.11485279e-01 1.78112596e-01 4.98335600e-01 6.24282705e-03 -4.92078811e-02 -6.34425163e-01 -1.20824233e-01 1.10924944e-01 -1.04813419e-01 1.61278099e-01 -8.82114232e-01 8.70105326e-01 6.78123474e+00 8.82487297e-01 -9.32567298e-01 2.99236506e-01 5.37514865e-01 1.74220018e-02 -7.99296200e-01 2.77773619e-01 -4.64818627e-01 4.25820678e-01 1.04498625e+00 -8.38863179e-02 3.10864031e-01 7.41826773e-01 5.06344497e-01 -2.46416792e-01 -1.15415764e+00 4.72026348e-01 -1.93469062e-01 -1.93866765e+00 -2.66913593e-01 7.06494078e-02 6.81465089e-01 2.62109369e-01 -1.71886757e-01 2.97332823e-01 4.92428809e-01 -8.78112257e-01 7.26085484e-01 7.75548741e-02 6.08172476e-01 -1.32429570e-01 4.63976145e-01 6.59959078e-01 -6.61614060e-01 -1.70735002e-01 -3.06682643e-02 -5.51839292e-01 -2.11429745e-02 9.30123687e-01 -1.56798255e+00 6.21813178e-01 6.61212504e-01 7.62079835e-01 -5.57399273e-01 9.41450298e-01 -3.24983746e-01 8.13603580e-01 -3.46244365e-01 -1.41867816e-01 7.22906590e-02 -1.17732108e-01 4.76075262e-01 1.51434565e+00 6.64691702e-02 -1.07286364e-01 3.15467387e-01 9.85546529e-01 -3.23078394e-01 4.84892428e-01 -4.12481695e-01 -8.12711641e-02 9.07537878e-01 1.57243776e+00 -5.50480068e-01 -3.87583256e-01 -4.44787085e-01 3.36849958e-01 8.81457865e-01 1.74288005e-01 -2.47627214e-01 1.64752662e-01 4.18055147e-01 4.31518108e-01 8.92118812e-02 -1.43559441e-01 -4.78124738e-01 -1.06006980e+00 1.70624956e-01 -1.29059362e+00 6.88621581e-01 -3.91222477e-01 -1.62262857e+00 2.32078239e-01 9.15504023e-02 -1.09111810e+00 -5.07820666e-01 -1.19373880e-01 -7.09193289e-01 7.02913523e-01 -1.60659683e+00 -1.17968225e+00 -4.16167974e-01 3.01052928e-01 7.51672149e-01 -5.91062903e-02 8.97152662e-01 2.43036851e-01 -2.90947884e-01 3.95676285e-01 -1.29590273e-01 -2.19296619e-01 1.08780587e+00 -1.33799016e+00 2.56708145e-01 5.62145174e-01 2.31176704e-01 8.66368294e-01 8.40157866e-01 -5.62896371e-01 -1.07441485e+00 -9.52052891e-01 1.03027582e+00 -3.52623314e-01 9.73895252e-01 -3.41627330e-01 -9.51058090e-01 3.30797344e-01 3.20151359e-01 -2.40176603e-01 1.06750464e+00 7.62831628e-01 -6.04463696e-01 1.05003335e-01 -1.29397714e+00 4.20703828e-01 1.09169185e+00 -5.65252781e-01 -1.02129126e+00 6.04358256e-01 5.91090381e-01 -2.32795209e-01 -1.03988981e+00 1.38238102e-01 4.88921195e-01 -1.07372439e+00 7.31943846e-01 -5.71523428e-01 6.71718240e-01 -1.59596547e-01 -1.20186560e-01 -1.08696270e+00 -7.54306614e-02 -7.79226601e-01 -8.71860906e-02 1.81632400e+00 4.86006141e-01 -3.84089410e-01 6.24833405e-01 6.95819318e-01 -3.68324034e-02 -7.26313531e-01 -8.60599935e-01 -3.85612994e-01 -3.77350952e-04 -5.20302117e-01 5.26318371e-01 1.17771447e+00 6.47821128e-01 4.13594812e-01 -5.74645251e-02 -5.11090875e-01 3.94986629e-01 4.94343936e-01 9.88178253e-01 -1.38031864e+00 -2.49997258e-01 -4.43049908e-01 -3.67842913e-02 -7.12140322e-01 1.37549713e-01 -8.70119810e-01 -3.27612162e-01 -1.77561033e+00 2.55944431e-01 -9.04900074e-01 -4.01138850e-02 3.16855162e-01 -2.04342499e-01 2.04030573e-01 -1.37777030e-01 3.67220223e-01 -9.23455536e-01 4.96039659e-01 9.11429107e-01 6.23813234e-02 -1.71842992e-01 -2.92533547e-01 -1.11640096e+00 6.46090090e-01 7.34417737e-01 -4.47667241e-01 -4.53767747e-01 -2.34656647e-01 6.14261270e-01 -4.78649527e-01 2.04359502e-01 -4.85621184e-01 1.24243043e-01 -2.10120469e-01 -1.83396086e-01 -4.05888140e-01 1.60062283e-01 -4.94309664e-01 -1.87865198e-01 -6.74042776e-02 -6.85260952e-01 -1.75165460e-01 -2.19106637e-02 6.48575842e-01 -1.71748653e-01 -5.22257052e-02 2.31628582e-01 -4.17518288e-01 -4.87050414e-01 -2.58166641e-01 -2.11330533e-01 3.67801517e-01 6.31473064e-01 -9.92612019e-02 -7.39529431e-01 -2.65522063e-01 -2.82040477e-01 2.18181580e-01 6.57605827e-01 4.28260863e-01 3.53779823e-01 -1.09186149e+00 -8.28918159e-01 -3.28442872e-01 3.73559833e-01 -2.44266260e-02 -1.12656109e-01 9.78069007e-01 -1.44823298e-01 3.28364760e-01 -1.44256622e-01 -3.27690959e-01 -1.13129210e+00 3.37892920e-01 -1.34504229e-01 -5.67929208e-01 -6.31367505e-01 5.36964536e-01 -3.14016134e-01 -8.26404393e-01 1.38419554e-01 -1.05892293e-01 -4.45366323e-01 2.03778431e-01 4.03086126e-01 5.14165521e-01 2.74814218e-01 -1.97937101e-01 -3.25306863e-01 -1.75223917e-01 4.07114252e-02 -4.83783931e-01 1.55264413e+00 -5.12795687e-01 -3.66044998e-01 4.73908216e-01 6.87339127e-01 4.72837299e-01 -8.61035943e-01 -5.25014937e-01 7.12104380e-01 -6.10243201e-01 -1.27436519e-01 -1.07770741e+00 -3.08747649e-01 5.05237043e-01 -7.42040649e-02 7.05882907e-01 7.82623410e-01 2.18674302e-01 4.86849993e-01 2.28403151e-01 1.33146808e-01 -1.47250664e+00 1.62493199e-01 2.60377198e-01 8.75967145e-01 -1.32830119e+00 2.37632036e-01 -6.62779748e-01 -6.03745282e-01 1.11332178e+00 4.25393969e-01 -1.01265509e-03 4.33081269e-01 -1.21709101e-01 9.56553966e-02 -5.44338942e-01 -1.23362231e+00 3.54470164e-02 4.10269380e-01 3.78849268e-01 9.75130379e-01 -4.81353141e-03 -8.92456234e-01 8.00329685e-01 -4.21819627e-01 5.52293137e-02 5.59009016e-01 7.95400858e-01 -5.02090812e-01 -1.37022448e+00 -2.30078369e-01 6.75241768e-01 -6.28480792e-01 -9.70945973e-03 -8.03961813e-01 4.78638113e-01 -6.82418421e-02 1.28994560e+00 -2.16693357e-01 -8.08299035e-02 1.39060110e-01 1.18643627e-01 9.69985425e-02 -9.16366220e-01 -1.11882651e+00 2.29506101e-02 9.83288229e-01 -6.37045860e-01 -8.12590063e-01 -8.43087196e-01 -1.02948940e+00 -3.36590469e-01 -2.79495388e-01 3.31916451e-01 8.44749928e-01 1.24882483e+00 8.70181799e-01 2.06896231e-01 3.49054277e-01 -3.18452448e-01 -8.16813767e-01 -1.32253861e+00 -1.73631132e-01 8.26501906e-01 -7.14097396e-02 -7.17718959e-01 -8.36900696e-02 -5.41458540e-02]
[10.469626426696777, 8.402549743652344]
0aac38d5-37fa-4fec-bc61-e161a68b1261
codes-a-distribution-shift-benchmark-dataset
2206.0548
null
https://arxiv.org/abs/2206.05480v2
https://arxiv.org/pdf/2206.05480v2.pdf
CodeS: Towards Code Model Generalization Under Distribution Shift
Distribution shift has been a longstanding challenge for the reliable deployment of deep learning (DL) models due to unexpected accuracy degradation. Although DL has been becoming a driving force for large-scale source code analysis in the big code era, limited progress has been made on distribution shift analysis and benchmarking for source code tasks. To fill this gap, this paper initiates to propose CodeS, a distribution shift benchmark dataset, for source code learning. Specifically, CodeS supports two programming languages (Java and Python) and five shift types (task, programmer, time-stamp, token, and concrete syntax tree). Extensive experiments based on CodeS reveal that 1) out-of-distribution detectors from other domains (e.g., computer vision) do not generalize to source code, 2) all code classification models suffer from distribution shifts, 3) representation-based shifts have a higher impact on the model than others, and 4) pre-trained bimodal models are relatively more resistant to distribution shifts.
['Yves Le Traon', 'Mike Papadakis', 'Lei Ma', 'Maxime Cordy', 'Xiaofei Xie', 'Yuejun Guo', 'Qiang Hu']
2022-06-11
null
null
null
null
['code-classification']
['computer-code']
[-3.66407990e-01 -3.18155885e-01 -4.35764134e-01 -3.33711296e-01 -7.81213582e-01 -7.84645796e-01 5.00962019e-01 4.24799532e-01 1.21072540e-02 2.23647207e-01 1.76184550e-01 -7.38218307e-01 2.03831419e-01 -4.38455969e-01 -7.94357359e-01 -4.36689198e-01 -2.70171195e-01 1.94896445e-01 3.18186194e-01 4.10728939e-02 4.84342694e-01 2.46274993e-01 -1.59839499e+00 7.30776429e-01 9.85420585e-01 6.24860942e-01 1.36522204e-01 6.29537880e-01 -5.15314162e-01 1.37934446e+00 -6.42748356e-01 -3.51302177e-01 -6.68021813e-02 -3.42124313e-01 -7.34253526e-01 -4.34076458e-01 5.58858395e-01 -3.22270572e-01 -4.19967234e-01 1.24255824e+00 5.23152590e-01 -4.93059993e-01 6.80388868e-01 -1.52559352e+00 -9.79570806e-01 9.90029335e-01 -9.51570749e-01 3.17116737e-01 1.97276101e-01 3.17919999e-02 7.84973383e-01 -1.05832255e+00 6.37771666e-01 1.21427393e+00 1.16902435e+00 4.13711846e-01 -1.32793534e+00 -6.15402162e-01 -2.99950153e-01 4.57388870e-02 -1.39505124e+00 -3.46844971e-01 4.26656425e-01 -1.15749359e+00 1.34974110e+00 -1.93317696e-01 1.56280249e-01 1.28420186e+00 6.98517323e-01 1.03280485e+00 7.20531940e-01 -4.36873049e-01 3.28767836e-01 2.70358354e-01 2.03265756e-01 5.64043641e-01 4.24849898e-01 -5.24798259e-02 -4.62402165e-01 -4.84797359e-01 7.18533993e-02 1.78423226e-01 -2.15773489e-02 -6.66917920e-01 -1.11819470e+00 8.37606311e-01 2.23823041e-01 4.94766891e-01 -3.93312834e-02 4.19767439e-01 8.71454835e-01 5.45314670e-01 3.18047762e-01 1.44313872e-01 -6.53482378e-01 -8.74558330e-01 -1.10875106e+00 4.37000036e-01 9.88732636e-01 1.29846573e+00 9.94628847e-01 2.55648226e-01 -2.97560215e-01 9.68005657e-01 2.45224342e-01 6.41308486e-01 7.27167487e-01 -4.73573685e-01 6.66769087e-01 8.92288685e-01 -4.30090815e-01 -9.03134406e-01 -4.87365305e-01 -3.11543733e-01 -5.66963732e-01 2.04940870e-01 4.78247553e-01 1.63396493e-01 -6.03403687e-01 1.51963520e+00 -3.57665062e-01 -4.12757754e-01 -2.64002502e-01 1.68716699e-01 8.04870129e-01 3.97551030e-01 1.46847010e-01 4.82106388e-01 1.00909781e+00 -6.01410806e-01 -9.55747440e-02 -4.87099499e-01 1.07422256e+00 -9.78514135e-01 1.11803770e+00 3.73052806e-01 -7.33960152e-01 -5.20341575e-01 -9.13852692e-01 -8.62870067e-02 -4.39437717e-01 2.00820640e-01 7.13439941e-01 8.66113245e-01 -1.09650970e+00 2.68737227e-01 -9.36301053e-01 -5.01806021e-01 5.45433044e-01 5.26481755e-02 -1.94464818e-01 -4.08090264e-01 -6.55159950e-01 5.95945179e-01 3.05891514e-01 -6.33945763e-01 -1.25772369e+00 -9.90405560e-01 -8.79824698e-01 3.20852280e-01 -1.45206213e-01 -9.84806493e-02 1.46887100e+00 -1.01879013e+00 -8.04805100e-01 1.10340631e+00 3.04965395e-02 -2.18540385e-01 1.88302055e-01 -5.97643405e-02 -3.88698012e-01 -5.53691328e-01 2.38846600e-01 3.42116743e-01 8.66299927e-01 -1.29198360e+00 -7.04981565e-01 -3.31134170e-01 -1.28945604e-01 -5.70895791e-01 -6.16859555e-01 2.41982326e-01 -1.44795388e-01 -6.23853862e-01 -3.17903548e-01 -9.66088712e-01 4.28827673e-01 -3.62985134e-01 -2.87828505e-01 -5.42805493e-01 7.32559800e-01 -5.64482808e-01 1.72044206e+00 -2.64616299e+00 1.23457871e-02 -8.61769915e-02 5.95058680e-01 -1.07150599e-01 -2.23754451e-01 6.09358191e-01 -2.88986951e-01 1.27159193e-01 -3.19894731e-01 -1.97807387e-01 1.63823411e-01 6.09552376e-02 -4.72172856e-01 6.48685455e-01 -1.25896767e-01 5.48299849e-01 -7.57245541e-01 -3.36383998e-01 -2.61602014e-01 1.34063289e-01 -1.05123377e+00 1.63462684e-01 -1.78174555e-01 -1.56184167e-01 -6.30877912e-02 1.06869733e+00 8.40596437e-01 -3.98348033e-01 -8.47581849e-02 3.76101464e-01 -4.07998204e-01 2.53272265e-01 -6.46735787e-01 1.91067076e+00 -6.43704772e-01 1.05175042e+00 4.56721522e-03 -6.97637022e-01 8.95549357e-01 -2.52661966e-02 3.06087643e-01 -7.50355959e-01 -8.47589895e-02 5.59739530e-01 3.89806211e-01 -7.13503897e-01 6.74108207e-01 3.65067601e-01 -4.73236173e-01 6.07069075e-01 3.15383106e-01 5.21003418e-02 3.69368374e-01 4.06185955e-01 1.50938475e+00 -3.00702415e-02 5.00689857e-02 -5.71446538e-01 6.21365793e-02 1.62367329e-01 4.49245483e-01 8.77964318e-01 -2.61781752e-01 4.92274016e-01 1.06639636e+00 -3.52505863e-01 -1.14016867e+00 -1.08609700e+00 -2.38074630e-01 1.76470578e+00 -1.96720690e-01 -5.39100766e-01 -6.97314024e-01 -8.43835652e-01 4.45687860e-01 6.93526983e-01 -4.89444882e-01 -4.08382058e-01 -4.46243405e-01 -7.41351128e-01 1.11461043e+00 6.95830464e-01 2.23317012e-01 -6.81110620e-01 -6.72288120e-01 1.34954646e-01 7.93761015e-02 -5.19338667e-01 -5.02074242e-01 5.30688345e-01 -7.10664153e-01 -1.09452581e+00 -6.20777249e-01 -7.50873387e-01 4.38840181e-01 1.67373419e-01 1.65982354e+00 1.93683907e-01 -4.33778971e-01 5.78664482e-01 -3.83391261e-01 -4.58974183e-01 -9.71490383e-01 4.51103389e-01 -2.89702088e-01 -6.32849395e-01 8.42611551e-01 -3.36720586e-01 -3.54704231e-01 1.46225691e-01 -8.90723228e-01 -4.76545215e-01 8.25043201e-01 8.94955575e-01 -1.69866845e-01 -3.05619892e-02 5.64315498e-01 -9.40186918e-01 9.10386384e-01 -1.22749209e+00 -5.03919959e-01 1.76095366e-01 -9.07894254e-01 9.61131155e-02 5.85188270e-01 -3.44200104e-01 -1.24313629e+00 -1.54283702e-01 -2.62776136e-01 -1.81393430e-01 -1.00958571e-01 7.43561745e-01 1.06542580e-01 5.52015081e-02 1.40806246e+00 3.49687219e-01 -6.46353662e-02 -6.74707413e-01 9.31050926e-02 9.30046201e-01 3.53423268e-01 -1.04848039e+00 4.06655014e-01 1.65707067e-01 -6.19259238e-01 -5.16012609e-01 -2.37399712e-01 -4.09092546e-01 -6.38725460e-01 4.28918116e-02 4.87193763e-01 -1.10461497e+00 -1.53370351e-01 7.67473042e-01 -1.24651480e+00 -5.31890035e-01 1.23193458e-01 2.36806780e-01 -1.84420973e-01 4.12712604e-01 -7.51859725e-01 -3.35297376e-01 7.65106753e-02 -1.47820437e+00 9.70399022e-01 5.26387431e-02 -4.03231770e-01 -1.12836468e+00 3.99428070e-01 -1.87099651e-01 8.72559667e-01 1.02636650e-01 1.53099871e+00 -6.84158862e-01 -4.80340183e-01 -5.31390458e-02 -4.68302548e-01 3.47182959e-01 -6.06324151e-02 5.49728036e-01 -1.22508943e+00 -5.66570759e-01 -2.03976110e-01 -4.50777650e-01 9.61998045e-01 6.78101033e-02 1.38037574e+00 -7.31996372e-02 -6.24091864e-01 6.29493415e-01 1.47959089e+00 2.49902293e-01 5.78904629e-01 3.34013939e-01 7.36663759e-01 2.21770257e-01 3.53098921e-02 8.02291512e-01 7.09266186e-01 2.82873273e-01 5.38113296e-01 1.81141421e-01 -3.46974909e-01 -2.87526399e-01 8.12032461e-01 9.14365649e-01 4.72229004e-01 6.74455892e-04 -1.78957963e+00 8.63792539e-01 -1.48817980e+00 -8.06185484e-01 -4.72067833e-01 2.20457554e+00 1.08643186e+00 4.77554575e-02 1.67886931e-02 -1.79246560e-01 4.75729704e-01 -1.32999852e-01 -6.15784645e-01 -4.32935417e-01 1.04222141e-01 -1.02754503e-01 4.36634302e-01 -2.02724457e-01 -9.08639610e-01 5.73624134e-01 6.58122015e+00 1.02989209e+00 -1.42634106e+00 2.49247089e-01 4.47463959e-01 3.43237549e-01 -5.50920248e-01 2.23345216e-02 -7.70759463e-01 9.46810424e-01 9.67753947e-01 -3.04965943e-01 3.62695754e-01 1.87368941e+00 -7.52827883e-01 -1.99469298e-01 -1.67663884e+00 1.05276799e+00 1.19638339e-01 -1.20863628e+00 -1.49994329e-01 1.28936052e-01 8.68549645e-01 5.82743406e-01 1.71587214e-01 1.03022289e+00 4.31559294e-01 -1.04071248e+00 1.20726430e+00 8.25585797e-02 1.03912401e+00 -6.37022018e-01 7.97206044e-01 4.02345926e-01 -1.00320327e+00 -4.17484730e-01 -5.96181870e-01 -5.92675284e-02 -6.39125705e-01 9.08320606e-01 -7.99900115e-01 -4.36677709e-02 1.08424747e+00 7.97193944e-01 -1.11790895e+00 1.02206969e+00 1.40600994e-01 7.89671302e-01 6.80433214e-02 9.02388915e-02 -1.20042644e-01 6.26361549e-01 1.24422647e-01 1.87210906e+00 6.14424706e-01 -9.39316809e-01 5.36396690e-02 1.26300681e+00 -3.02504241e-01 -3.33488733e-01 -7.96914577e-01 -3.13089937e-01 6.55258715e-01 1.10065579e+00 -6.96464241e-01 -1.27341568e-01 -9.70175207e-01 5.70847571e-01 3.72177988e-01 3.36526692e-01 -8.93811226e-01 -6.60086334e-01 7.35323548e-01 1.20628916e-01 2.04247221e-01 -2.04527318e-01 -3.45475972e-01 -1.20492208e+00 -2.17051595e-01 -1.14727676e+00 4.39177960e-01 -4.21980917e-01 -1.42212057e+00 4.82055128e-01 -3.10923774e-02 -1.04339743e+00 -3.60163569e-01 -7.22345173e-01 -4.91360515e-01 7.25919902e-01 -1.37251043e+00 -8.74011815e-01 -4.29960459e-01 6.39797091e-01 3.29775184e-01 -6.19995177e-01 7.03762770e-01 8.34782958e-01 -3.23896736e-01 9.75420713e-01 5.40307999e-01 3.22427183e-01 9.50930536e-01 -1.46509254e+00 4.41989183e-01 8.05862486e-01 -1.37439921e-01 9.33327138e-01 5.94153285e-01 -4.56501693e-01 -1.65464330e+00 -9.17625487e-01 6.80617750e-01 -8.36664975e-01 8.03460002e-01 -6.42244995e-01 -1.03526402e+00 7.87005723e-01 2.26253912e-01 7.64103830e-02 8.25988948e-01 3.62336755e-01 -1.07406259e+00 -1.65842935e-01 -1.01898444e+00 -3.76860872e-02 7.62304842e-01 -8.85509610e-01 -4.07163352e-01 -4.52862084e-02 3.26563418e-01 -6.09832048e-01 -8.76020730e-01 1.64169585e-03 4.01054561e-01 -1.40949321e+00 6.81698501e-01 -2.09529430e-01 9.88460124e-01 -6.09836951e-02 -2.32241556e-01 -1.43045878e+00 -3.95709902e-01 -2.01238528e-01 -3.51970524e-01 1.56527495e+00 2.46412516e-01 -3.28238159e-01 2.93900102e-01 4.88132954e-01 -4.76542354e-01 -5.11297107e-01 -7.99097002e-01 -8.86590302e-01 6.21595800e-01 -4.80343491e-01 6.04198575e-01 1.09334588e+00 2.35656977e-01 1.13213964e-01 -2.19578575e-02 -2.88912416e-01 2.95420587e-01 1.65060863e-01 8.72796714e-01 -1.30459213e+00 -3.68730724e-01 -8.91957521e-01 -2.15890586e-01 -1.03748238e+00 3.05095881e-01 -1.27179742e+00 1.82458371e-01 -1.30898535e+00 6.12009943e-01 -7.33819008e-01 -1.09744184e-01 7.46025741e-01 1.68520853e-01 -3.16090763e-01 -9.91936773e-02 4.24452841e-01 -5.53836823e-01 2.30780214e-01 5.41831553e-01 -3.32971185e-01 2.61750340e-01 -1.26536161e-01 -8.05493534e-01 6.48121536e-01 5.18706918e-01 -8.02307606e-01 -3.28250915e-01 -7.73333013e-01 8.03715765e-01 -5.93621992e-02 1.14920892e-01 -1.05770683e+00 2.04945251e-01 1.60572529e-02 2.79495716e-01 -3.64522427e-01 -6.83688879e-01 -6.59634948e-01 -2.53864765e-01 5.87669969e-01 -1.60382107e-01 3.60984832e-01 6.42492771e-01 2.99557477e-01 -3.16435248e-01 -3.88359308e-01 8.30544174e-01 -3.98735143e-02 -1.04011548e+00 -1.10775419e-02 -6.55479014e-01 5.84770381e-01 8.17987204e-01 6.30645603e-02 -8.29613030e-01 1.86117798e-01 2.48773731e-02 -5.66487834e-02 7.30388641e-01 9.53589737e-01 2.06464455e-01 -1.27648163e+00 -5.46962142e-01 3.38295937e-01 7.53387213e-01 -2.50116974e-01 3.20589036e-01 8.06352615e-01 -6.67980850e-01 3.11607778e-01 -2.13787943e-01 -9.13303912e-01 -9.56325948e-01 5.89088976e-01 2.34201208e-01 -1.58631578e-01 -3.57309222e-01 1.10405505e+00 2.54761249e-01 -6.87757492e-01 3.47138464e-01 -7.67528653e-01 3.69983524e-01 1.92966089e-01 5.90443552e-01 4.67603564e-01 3.62197399e-01 -2.24099696e-01 -5.98989367e-01 2.09252492e-01 -2.38692835e-01 6.19508684e-01 1.33796549e+00 2.22838074e-01 -6.33655608e-01 7.07041919e-01 1.44375598e+00 2.44701475e-01 -1.14946663e+00 -6.59589171e-02 4.21712726e-01 -5.38121223e-01 -1.90626428e-01 -9.24809635e-01 -9.35074568e-01 1.22069073e+00 5.89257360e-01 5.62420487e-01 7.50119746e-01 2.52648592e-01 5.04710138e-01 7.17509314e-02 6.47468209e-01 -9.95207489e-01 2.27786630e-01 9.54191267e-01 7.27496326e-01 -1.10414505e+00 -3.01984429e-01 3.94429386e-01 -3.01842004e-01 1.31908989e+00 1.00376952e+00 1.80658564e-01 7.76854396e-01 8.45189154e-01 -1.68431818e-01 -2.20148459e-01 -7.92502224e-01 3.32206249e-01 -1.62133742e-02 6.11285269e-01 1.09630430e+00 -2.29816154e-01 1.63907513e-01 6.32713497e-01 2.65039932e-02 1.12815395e-01 8.17594349e-01 1.09219170e+00 -4.38716173e-01 -1.03073180e+00 -4.15615112e-01 5.79429567e-01 -4.59192783e-01 -3.72694463e-01 -9.72325206e-02 7.30124831e-01 3.91699493e-01 4.97942090e-01 1.16237119e-01 -3.97084415e-01 1.23174541e-01 1.85675532e-01 4.23394978e-01 -9.33728516e-01 -7.00134516e-01 -4.49884802e-01 -4.88852978e-01 -4.65899318e-01 1.73850432e-01 -6.26998007e-01 -1.31108224e+00 -8.05388868e-01 1.20215617e-01 -1.79449826e-01 6.36272967e-01 4.59939331e-01 7.96631813e-01 6.77602232e-01 1.07497171e-01 -5.21750510e-01 -7.38440514e-01 -1.09048355e+00 -7.14896560e-01 4.98704165e-01 4.95733351e-01 -7.04989970e-01 -6.58715546e-01 2.37996951e-01]
[7.624568939208984, 7.933331489562988]
8a95f1e5-d2ee-4d7f-b820-044580aaed0e
motion-r3-fast-and-accurate-motion-annotation
2304.01672
null
https://arxiv.org/abs/2304.01672v1
https://arxiv.org/pdf/2304.01672v1.pdf
Motion-R3: Fast and Accurate Motion Annotation via Representation-based Representativeness Ranking
In this paper, we follow a data-centric philosophy and propose a novel motion annotation method based on the inherent representativeness of motion data in a given dataset. Specifically, we propose a Representation-based Representativeness Ranking R3 method that ranks all motion data in a given dataset according to their representativeness in a learned motion representation space. We further propose a novel dual-level motion constrastive learning method to learn the motion representation space in a more informative way. Thanks to its high efficiency, our method is particularly responsive to frequent requirements change and enables agile development of motion annotation models. Experimental results on the HDM05 dataset against state-of-the-art methods demonstrate the superiority of our method.
['Yipeng Qin', 'Andreas Aristidou', 'Yazhan Zhang', 'Zijiao Zeng', 'Kai Wang', 'Fengyi Fang', 'Shihui Guo', 'Tianxiang Ren', 'Jubo Yu']
2023-04-04
null
null
null
null
['philosophy']
['miscellaneous']
[ 7.37446884e-04 -3.69706929e-01 -7.37836540e-01 -3.28681141e-01 -6.47028148e-01 -2.82023728e-01 5.33834755e-01 9.61550977e-03 -2.59131581e-01 2.70337254e-01 9.49591756e-01 2.66613543e-01 -5.49033523e-01 -4.32278514e-01 -3.77435803e-01 -5.12331247e-01 -8.65732133e-02 3.11341345e-01 4.68516290e-01 -2.14378655e-01 4.94042605e-01 2.95047611e-01 -1.39100826e+00 4.63394910e-01 5.21003306e-01 4.75925773e-01 5.91586769e-01 6.01208150e-01 6.15875721e-02 1.30229890e+00 -4.06145781e-01 4.44490433e-01 7.69096911e-02 -4.88454878e-01 -1.23312521e+00 2.26957381e-01 1.57291055e-01 -2.70780981e-01 -5.34290254e-01 4.04322177e-01 4.95462298e-01 6.39390707e-01 5.64734221e-01 -1.21718514e+00 -4.38768744e-01 5.64258575e-01 -7.57738233e-01 2.44210139e-01 6.06744647e-01 1.04095116e-01 1.11686087e+00 -9.05287981e-01 9.21757817e-01 1.11928260e+00 4.33049083e-01 8.45968485e-01 -9.96890187e-01 5.87660400e-03 4.82170105e-01 4.57600266e-01 -1.37797952e+00 -1.89882413e-01 1.00398493e+00 -6.04571104e-01 8.41305137e-01 3.20160985e-01 6.61823034e-01 1.01704621e+00 -8.80965889e-02 1.15404952e+00 3.65610540e-01 -3.52703422e-01 4.65469360e-01 -5.43990672e-01 8.95069763e-02 6.31281495e-01 2.56397605e-01 -4.81141627e-01 -6.52114630e-01 -8.05949047e-02 6.98294699e-01 -8.60530809e-02 -3.86251241e-01 -1.00456238e+00 -1.67939568e+00 6.98626280e-01 1.45837203e-01 5.59322655e-01 -4.59431380e-01 7.50623226e-01 7.55431294e-01 -3.78700405e-01 2.50091434e-01 2.63182580e-01 -3.56747270e-01 -6.32453680e-01 -9.72701728e-01 2.20052272e-01 3.95661861e-01 1.00385940e+00 6.16080523e-01 1.73697263e-01 -4.21040803e-01 6.13242149e-01 3.71356577e-01 -1.44983321e-01 9.17797744e-01 -1.20471740e+00 5.05203247e-01 7.07812607e-01 2.41570577e-01 -1.04748070e+00 -5.88846445e-01 1.55203998e-01 -6.28756166e-01 -1.28864422e-01 -8.71136934e-02 2.77326047e-01 -4.13593084e-01 1.67804444e+00 3.99724692e-01 3.74629945e-01 1.95972770e-01 9.25188065e-01 5.82122684e-01 7.23040462e-01 7.39302263e-02 -2.36053944e-01 8.84527326e-01 -1.33620691e+00 -5.80266058e-01 1.31103294e-02 1.27669954e+00 -3.42671871e-01 1.44045246e+00 4.03543383e-01 -8.47682238e-01 -8.32167864e-01 -1.06287408e+00 8.55215788e-02 2.22675353e-01 2.76232749e-01 6.43302262e-01 3.91863972e-01 -8.54765773e-01 8.20239186e-01 -9.52342987e-01 -6.42442405e-01 3.76388073e-01 7.97759593e-02 -3.66371632e-01 1.97103946e-03 -6.51981890e-01 5.98499715e-01 8.08054626e-01 -1.53714448e-01 -9.19641733e-01 -5.83261847e-01 -8.75265539e-01 -3.89929295e-01 3.08898091e-01 -7.54664958e-01 1.32879579e+00 -5.53743958e-01 -1.43479812e+00 3.80317569e-01 -2.81326305e-02 -2.11186305e-01 4.05732274e-01 -6.37684822e-01 -5.46635985e-01 2.83944249e-01 2.03452751e-01 6.63045764e-01 6.26143336e-01 -1.21318197e+00 -8.74909818e-01 1.19978219e-01 2.48679277e-02 2.36522466e-01 -7.33512402e-01 -3.34239602e-01 -7.65782475e-01 -7.53695071e-01 -8.36155564e-02 -1.02814972e+00 -5.75613976e-01 -2.66448826e-01 -7.92423189e-02 -4.38849449e-01 1.13347471e+00 -3.14040869e-01 2.10293937e+00 -2.01449370e+00 6.17639780e-01 1.48707196e-01 2.80697942e-02 2.95559675e-01 -2.57161885e-01 6.62735701e-01 -1.21119265e-02 7.47836456e-02 -2.20195353e-01 -1.40056312e-01 3.91477160e-02 3.76948655e-01 -2.52236307e-01 4.65003133e-01 8.10788125e-02 7.61717916e-01 -1.38683021e+00 -6.39622569e-01 3.05351645e-01 2.46383607e-01 -6.57239914e-01 2.35659257e-01 -4.39726561e-02 6.91552997e-01 -7.93597341e-01 5.53265333e-01 1.81270450e-01 -1.32521853e-01 3.61600995e-01 -2.64973015e-01 -1.13416463e-01 -1.47977278e-01 -1.23774087e+00 2.54607344e+00 -5.98886728e-01 5.28714120e-01 -8.62756848e-01 -7.67077267e-01 1.05564070e+00 2.54499972e-01 9.42193151e-01 -3.69173646e-01 -3.12175035e-01 4.73822691e-02 -2.16008410e-01 -8.48596334e-01 9.69518900e-01 4.59861815e-01 -4.38116997e-01 5.48577011e-01 -4.11485940e-01 5.00236899e-02 3.06938857e-01 1.12988845e-01 1.17824090e+00 8.78857493e-01 6.81123793e-01 -3.79817873e-01 7.07237363e-01 2.03371391e-01 9.41670299e-01 4.07242745e-01 -4.76487488e-01 6.27316356e-01 9.77032781e-02 -9.91057932e-01 -1.00443530e+00 -4.64458525e-01 3.98324639e-01 1.20474625e+00 3.82006764e-01 -9.98606563e-01 -5.71240783e-01 -9.39556360e-01 -2.41271898e-01 7.48543620e-01 -7.24755049e-01 -3.26217383e-01 -1.24143791e+00 -4.00110245e-01 3.51640910e-01 1.15594471e+00 3.16285133e-01 -1.09229779e+00 -1.39031851e+00 3.58729482e-01 -3.80707741e-01 -8.87095928e-01 -8.09590816e-01 -3.12346876e-01 -1.01463592e+00 -8.25345993e-01 -5.58463812e-01 -8.24318767e-01 6.19365871e-01 7.74407446e-01 1.10675979e+00 -6.61621392e-02 -2.07931891e-01 7.52444983e-01 -9.10809755e-01 1.68866217e-01 -4.65515971e-01 1.35200381e-01 2.67283380e-01 -5.44351079e-02 3.32494348e-01 -3.23574603e-01 -9.49635804e-01 5.11607230e-01 -1.07946050e+00 -1.93993244e-02 4.69482541e-01 5.07670283e-01 8.89041662e-01 1.29246414e-01 4.83789116e-01 -7.18632042e-01 4.70190376e-01 -6.45055056e-01 -2.72072256e-01 3.43387395e-01 -5.61419904e-01 1.21068373e-01 4.59144056e-01 -6.41720057e-01 -8.89136195e-01 5.24290919e-01 4.50975537e-01 -5.99378884e-01 -8.03507715e-02 5.54084003e-01 -1.23779573e-01 3.21029961e-01 6.33287787e-01 1.45815447e-01 -5.71950674e-01 -3.86217386e-01 9.57630575e-01 1.84808314e-01 7.44782865e-01 -7.52747953e-01 8.05355608e-01 7.52653062e-01 1.14717297e-01 -7.70477533e-01 -3.98724735e-01 -1.00497878e+00 -9.80198026e-01 -4.15204167e-01 9.60633695e-01 -8.48522186e-01 -5.87755144e-01 -4.81680259e-02 -1.19076514e+00 -2.86872894e-01 -5.57962000e-01 7.26338565e-01 -1.10121942e+00 8.24036360e-01 -1.88175395e-01 -7.49355853e-01 -3.36081237e-01 -1.04825735e+00 1.06632197e+00 4.80818897e-02 -8.64770293e-01 -1.04157734e+00 6.04163766e-01 1.70485094e-01 1.18488751e-01 5.44251144e-01 8.23839009e-01 -3.96408319e-01 -4.30562228e-01 -1.09369017e-01 2.77283788e-01 -8.61224607e-02 3.91233832e-01 2.11895540e-01 -5.47951937e-01 -4.46562082e-01 -3.09722334e-01 -2.11678058e-01 7.48003840e-01 3.58358055e-01 1.44407797e+00 -3.60472411e-01 -3.41697842e-01 4.09412622e-01 1.46951222e+00 1.90469235e-01 7.31106400e-01 7.24631429e-01 1.19964182e+00 7.52837658e-01 1.48416662e+00 1.01467347e+00 5.05267799e-01 9.87927139e-01 5.82320273e-01 2.38938600e-01 -2.50269435e-02 -3.63159031e-01 4.53988403e-01 1.17000175e+00 -2.98424006e-01 -2.09538773e-01 -9.88638461e-01 8.38231742e-01 -2.61513257e+00 -1.14702892e+00 -3.42225611e-01 2.09903812e+00 3.65292847e-01 -1.63511097e-01 5.81896782e-01 2.29946345e-01 5.87751567e-01 4.84421104e-01 -2.96328992e-01 -2.77298868e-01 1.35282099e-01 -3.41810107e-01 1.96325094e-01 6.66790083e-02 -1.21310616e+00 9.55374002e-01 6.14366245e+00 8.00514460e-01 -6.06946647e-01 -4.77180891e-02 1.57962903e-01 -1.53618669e-02 -4.05038029e-01 1.22537389e-02 -5.13487995e-01 3.52473706e-01 6.60105824e-01 -2.92959690e-01 -3.61172557e-02 1.19096446e+00 5.10923147e-01 3.61043215e-02 -1.30349803e+00 1.20482171e+00 7.17106983e-02 -1.57453978e+00 4.76263046e-01 -1.64081395e-01 8.11958432e-01 -3.74899626e-01 -4.86839972e-02 1.41325504e-01 1.14993148e-01 -7.47063875e-01 9.02772486e-01 6.40175343e-01 4.59629059e-01 -8.95313680e-01 3.44620556e-01 3.27698976e-01 -1.83283150e+00 -1.95764631e-01 -4.49070513e-01 1.09618485e-01 2.73851514e-01 -2.12253276e-02 -8.01113427e-01 9.84266996e-01 5.14782190e-01 1.07946575e+00 -4.04912323e-01 9.67886269e-01 -8.62072855e-02 4.22076344e-01 4.09557581e-01 1.26629770e-01 4.11843628e-01 5.64148091e-02 5.95361948e-01 1.50636554e+00 3.24274570e-01 -8.26268047e-02 5.15130341e-01 2.94376910e-01 1.85174048e-01 2.23585293e-01 -8.16410363e-01 7.64709860e-02 6.42629027e-01 1.36783516e+00 -7.63575077e-01 -7.98021108e-02 -3.91191065e-01 1.09074163e+00 9.58009660e-02 1.82125121e-01 -9.65919733e-01 6.29182383e-02 5.12226284e-01 -9.71825123e-02 4.84362006e-01 -4.87617135e-01 2.28319407e-01 -9.13820267e-01 4.06771302e-02 -6.34028733e-01 6.93795204e-01 -5.38234890e-01 -8.64998102e-01 5.06668270e-01 2.93381840e-01 -2.22501159e+00 -5.26942849e-01 -2.43185580e-01 -7.22407103e-01 2.70462304e-01 -1.38007402e+00 -1.19151163e+00 -6.26997113e-01 5.62417984e-01 1.02955222e+00 -3.36895853e-01 8.18628371e-01 1.18630528e-01 -6.41295314e-01 2.64922023e-01 -8.85010511e-02 -1.17147475e-01 5.34883738e-01 -1.15040672e+00 6.34801030e-01 9.30940866e-01 2.81557739e-01 6.65642262e-01 4.67330098e-01 -6.76722288e-01 -1.66079354e+00 -1.46856225e+00 3.87169391e-01 -6.06195569e-01 6.50927007e-01 2.35824004e-01 -9.18329120e-01 5.98696947e-01 -6.32625520e-02 1.73257247e-01 9.35480773e-01 -4.46054131e-01 -4.12945002e-01 -7.15291351e-02 -6.74444795e-01 5.69612145e-01 1.44195032e+00 -3.42241168e-01 -7.38196373e-01 2.13858262e-01 8.52373540e-01 -2.05249265e-01 -1.02358723e+00 6.12242699e-01 5.95145643e-01 -6.25231683e-01 9.69760656e-01 -9.67096448e-01 3.05026948e-01 -8.23288321e-01 -6.27405763e-01 -1.08871639e+00 -8.61403584e-01 -1.00778604e+00 -6.13858640e-01 1.29947782e+00 -9.75473300e-02 3.05886745e-01 9.94131088e-01 2.01250091e-01 -2.94389009e-01 -7.51428962e-01 -7.29882479e-01 -1.25111866e+00 -3.52454722e-01 -4.17320430e-01 5.85565269e-01 1.12375975e+00 2.43202314e-01 2.19256639e-01 -8.18070352e-01 2.85672117e-02 2.95009613e-01 2.55575538e-01 1.11217260e+00 -1.06529939e+00 -4.71707553e-01 -3.79780561e-01 -6.77388608e-01 -1.04392815e+00 1.91933006e-01 -7.27467299e-01 2.15907097e-01 -1.82587957e+00 2.65140355e-01 -1.68700397e-01 -5.00136912e-01 3.15416306e-01 -1.58524215e-01 -6.61545321e-02 2.29875222e-01 7.60381520e-01 -1.24662697e+00 6.41288877e-01 9.69153643e-01 -1.06250405e-01 -5.48862994e-01 -1.77050456e-01 -5.25917053e-01 9.59088326e-01 7.57568002e-01 -2.79278874e-01 -9.35321093e-01 -7.15370476e-01 -5.11781648e-02 -1.40485317e-01 -3.76748629e-02 -1.11964464e+00 1.75835893e-01 -4.58259046e-01 1.08527727e-01 -7.93503761e-01 9.44517776e-02 -7.65510798e-01 2.54475951e-01 7.45364845e-01 -4.08452183e-01 3.96745712e-01 1.70627870e-02 7.61523485e-01 -9.09093693e-02 -1.90332115e-01 4.47053909e-01 1.37597635e-01 -1.91275108e+00 3.68934631e-01 -2.43998632e-01 7.06778765e-02 1.29415536e+00 -4.33291793e-01 -3.04646045e-01 -3.06624711e-01 -3.61757696e-01 1.88171864e-01 6.24407351e-01 7.33225524e-01 1.00119078e+00 -1.88942039e+00 -7.26762116e-01 -5.08931160e-01 8.12337637e-01 -1.68529153e-03 3.04068267e-01 6.03081703e-01 -5.92121840e-01 2.12671816e-01 -2.72136718e-01 -8.27200472e-01 -1.02941871e+00 6.02319777e-01 -2.51119196e-01 -2.61834264e-01 -8.98987055e-01 4.93450582e-01 2.42741138e-01 1.27222121e-01 1.20152026e-01 -2.93196559e-01 -5.55543661e-01 -5.57171972e-03 6.78163230e-01 9.97436941e-01 -2.08555967e-01 -9.45534825e-01 -6.27871573e-01 9.08379972e-01 5.17687052e-02 -1.61672663e-02 1.27694345e+00 -2.11945713e-01 2.49043882e-01 7.15173602e-01 1.10418272e+00 -1.07322514e-01 -1.52831411e+00 -4.14140448e-02 7.11847186e-01 -7.68155754e-01 -2.90844858e-01 -1.46128371e-01 -8.84069562e-01 5.10959387e-01 4.57720637e-01 -8.29796195e-02 1.20666015e+00 5.57933189e-02 6.22376680e-01 3.46162677e-01 4.78409499e-01 -1.38964760e+00 8.47967505e-01 2.20690906e-01 8.40393186e-01 -9.01042938e-01 1.91456959e-01 -1.43464044e-01 -1.15812635e+00 1.21848226e+00 7.64178693e-01 1.17929272e-01 -4.01592143e-02 -2.72551596e-01 -3.95856276e-02 -9.01191905e-02 -6.87554121e-01 -1.05053693e-01 4.70615268e-01 8.86281252e-01 4.11604404e-01 -1.04166949e-02 -4.04062748e-01 3.66970688e-01 4.66793597e-01 -2.37964876e-02 7.73654878e-01 1.30921495e+00 -5.06777227e-01 -1.12717414e+00 -4.29399461e-02 -2.94213414e-01 1.55415475e-01 4.72797662e-01 -1.29346699e-01 9.47352290e-01 -1.89153850e-02 8.46394002e-01 -2.41155460e-01 -7.33438909e-01 6.75979853e-01 -1.97223023e-01 3.57405365e-01 -7.15256810e-01 -1.45995662e-01 1.28767222e-01 3.24065983e-02 -1.15133619e+00 -9.27130282e-01 -7.73504555e-01 -1.66470027e+00 9.85232089e-03 1.16887465e-01 5.27171828e-02 3.25459093e-01 6.24417543e-01 4.43744332e-01 6.56041145e-01 8.94204259e-01 -1.01422155e+00 -4.92859155e-01 -5.50166547e-01 -4.06482279e-01 7.49970794e-01 1.96805038e-02 -5.74159622e-01 2.62323260e-01 2.44845331e-01]
[8.440022468566895, 0.5560885071754456]
3b23c515-4119-4565-bbe9-1f7fceec3da9
barlow-twins-self-supervised-learning-via
2103.0323
null
https://arxiv.org/abs/2103.03230v3
https://arxiv.org/pdf/2103.03230v3.pdf
Barlow Twins: Self-Supervised Learning via Redundancy Reduction
Self-supervised learning (SSL) is rapidly closing the gap with supervised methods on large computer vision benchmarks. A successful approach to SSL is to learn embeddings which are invariant to distortions of the input sample. However, a recurring issue with this approach is the existence of trivial constant solutions. Most current methods avoid such solutions by careful implementation details. We propose an objective function that naturally avoids collapse by measuring the cross-correlation matrix between the outputs of two identical networks fed with distorted versions of a sample, and making it as close to the identity matrix as possible. This causes the embedding vectors of distorted versions of a sample to be similar, while minimizing the redundancy between the components of these vectors. The method is called Barlow Twins, owing to neuroscientist H. Barlow's redundancy-reduction principle applied to a pair of identical networks. Barlow Twins does not require large batches nor asymmetry between the network twins such as a predictor network, gradient stopping, or a moving average on the weight updates. Intriguingly it benefits from very high-dimensional output vectors. Barlow Twins outperforms previous methods on ImageNet for semi-supervised classification in the low-data regime, and is on par with current state of the art for ImageNet classification with a linear classifier head, and for transfer tasks of classification and object detection.
['Stéphane Deny', 'Yann Lecun', 'Ishan Misra', 'Li Jing', 'Jure Zbontar']
2021-03-04
null
null
null
null
['self-supervised-image-classification']
['computer-vision']
[ 2.26122797e-01 1.51371136e-01 -4.94079888e-02 -3.25569451e-01 -1.82424650e-01 -4.45048809e-01 7.04374850e-01 2.13064328e-01 -7.32128263e-01 4.01722014e-01 3.51150110e-02 3.82806882e-02 -4.14836437e-01 -5.49796402e-01 -7.95591950e-01 -8.29695761e-01 -2.78466791e-01 4.47606444e-01 2.76257247e-01 -3.14937025e-01 8.28481838e-02 4.32375222e-01 -1.69236708e+00 2.40730852e-01 4.27167654e-01 1.25919449e+00 9.89562497e-02 6.46966040e-01 1.35005044e-03 6.42306685e-01 -2.47910857e-01 -5.04523516e-01 6.96861923e-01 -3.22426736e-01 -6.85288906e-01 5.53280711e-02 9.59241331e-01 -1.16863519e-01 -4.12667215e-01 1.20252562e+00 4.95454133e-01 -1.22741899e-02 7.63315439e-01 -1.46201003e+00 -7.55989432e-01 6.00327194e-01 -4.70719814e-01 2.44111314e-01 -1.40643746e-01 -1.45231606e-02 1.21596873e+00 -8.93900573e-01 7.52998412e-01 1.24321258e+00 9.70795453e-01 5.20961463e-01 -1.79960597e+00 -5.35882771e-01 7.49530122e-02 2.74327219e-01 -1.11533415e+00 -7.47401297e-01 5.42507768e-01 -5.78451216e-01 9.78145063e-01 2.06553608e-01 6.23115778e-01 1.07447171e+00 2.40435794e-01 5.27636588e-01 8.87976706e-01 -3.74053985e-01 3.49927396e-01 4.45382029e-01 2.82109529e-01 5.42443812e-01 3.46212178e-01 2.49731494e-03 -4.90977496e-01 -1.01872288e-01 4.23653603e-01 1.88812017e-01 -2.22322375e-01 -9.05904651e-01 -1.24782836e+00 8.71744037e-01 6.72976553e-01 3.61582279e-01 -2.12064818e-01 2.31137071e-02 5.88425875e-01 7.49950051e-01 4.73052979e-01 5.86214423e-01 -3.45770776e-01 2.30255723e-01 -7.88865864e-01 2.10441679e-01 8.07030797e-01 6.86971009e-01 7.16987610e-01 2.85234414e-02 2.28112578e-01 9.40283597e-01 8.62718895e-02 1.37891039e-01 9.14327264e-01 -9.30100918e-01 3.49251151e-01 5.67267716e-01 -4.66520995e-01 -1.08939981e+00 -3.60645205e-01 -5.14809668e-01 -1.11848271e+00 5.98085940e-01 6.05440617e-01 1.57333463e-01 -6.00761056e-01 1.88393152e+00 1.38612213e-02 1.58420175e-01 1.30252121e-02 7.79715300e-01 5.95259249e-01 2.48258322e-01 -3.74349684e-01 -5.37949204e-02 9.89839256e-01 -8.49267840e-01 -3.13302249e-01 -4.26285416e-01 6.28209352e-01 -5.56167066e-01 8.03916335e-01 2.79170007e-01 -9.34294939e-01 -5.41486979e-01 -1.28381777e+00 -1.00097828e-01 -4.33657318e-01 -1.13621458e-01 4.03186291e-01 5.14273942e-01 -1.17859983e+00 1.11713982e+00 -5.31975448e-01 -4.90572572e-01 5.44283390e-01 5.14296174e-01 -8.46435189e-01 -1.25156626e-01 -8.53715777e-01 1.11728346e+00 1.70810759e-01 -5.89398034e-02 -5.80486298e-01 -8.69679391e-01 -9.50937986e-01 -8.29963014e-02 -4.81989700e-03 -3.77437383e-01 9.33762729e-01 -1.24603415e+00 -1.08435583e+00 9.02151167e-01 1.22043379e-01 -8.78761470e-01 7.06951857e-01 7.13957027e-02 -1.28630340e-01 -3.75185162e-02 1.12245522e-01 9.26597714e-01 1.28010547e+00 -1.00020707e+00 -2.26772621e-01 -5.88576198e-01 -2.82326102e-01 1.52985841e-01 -7.39706993e-01 -2.62990773e-01 -4.99795415e-02 -4.98997837e-01 3.17334324e-01 -9.79698062e-01 -3.75529557e-01 4.98245925e-01 -1.64365172e-01 -1.17101796e-01 7.75772810e-01 -1.70696154e-01 5.71959913e-01 -2.45879149e+00 1.89471811e-01 1.35547221e-01 6.87709272e-01 3.07527184e-01 -5.43142855e-01 2.93252736e-01 -5.85318565e-01 -1.46006688e-01 -2.93603390e-01 -4.72633123e-01 -6.94315135e-02 3.31022054e-01 -3.73838514e-01 8.18530560e-01 3.96318406e-01 7.11231053e-01 -9.70514774e-01 -2.18271658e-01 7.64689818e-02 3.66074681e-01 -5.37766457e-01 2.05928758e-02 1.52927190e-01 -2.11506099e-01 1.96255833e-01 8.26013088e-02 6.20817363e-01 -1.74760252e-01 2.15360865e-01 -4.35940474e-01 -7.93840066e-02 2.71003395e-01 -1.27349961e+00 1.52909255e+00 -8.59073400e-02 9.30939317e-01 -2.99268253e-02 -1.56851494e+00 8.28812301e-01 1.19974770e-01 6.12227559e-01 -7.27250457e-01 2.10319668e-01 2.13409454e-01 3.62786800e-01 -3.24048102e-01 4.12634611e-02 -2.88010873e-02 3.85390997e-01 6.28087819e-01 5.84226131e-01 -5.09370305e-02 2.87567407e-01 3.67781460e-01 1.20422900e+00 -2.42055282e-01 1.65357471e-01 -3.90376806e-01 1.54967740e-01 -3.27087939e-01 4.49109495e-01 7.74363637e-01 -3.95961791e-01 7.59216845e-01 7.42020369e-01 -4.96387482e-01 -1.34771407e+00 -1.17468607e+00 -3.20597112e-01 9.05181408e-01 -1.42146468e-01 -3.87605369e-01 -6.89488113e-01 -6.94597006e-01 3.71667296e-01 2.58045048e-01 -9.97615159e-01 -4.17385787e-01 -2.08452225e-01 -6.61313951e-01 5.73975325e-01 3.39368910e-01 4.04792368e-01 -6.92672133e-01 -4.10468906e-01 1.25368759e-01 3.69412959e-01 -1.10956955e+00 -3.65376592e-01 6.83672786e-01 -9.76533592e-01 -1.15742910e+00 -6.29333854e-01 -9.28941071e-01 8.18095028e-01 4.30746317e-01 8.31579745e-01 -7.92809427e-02 -5.94143152e-01 1.16421148e-01 2.27982830e-02 -1.45163149e-01 -3.19772720e-01 -5.96043058e-02 4.24771488e-01 2.40663484e-01 1.91000432e-01 -8.58885586e-01 -4.29570884e-01 2.55050063e-01 -8.84539843e-01 -1.28797323e-01 5.47590256e-01 1.09526396e+00 3.54530752e-01 -1.45466551e-01 3.68735373e-01 -7.60363638e-01 5.96655428e-01 -5.30402601e-01 -3.24757218e-01 5.75294979e-02 -8.09029162e-01 3.63194644e-01 7.97561109e-01 -7.62942195e-01 -3.23022574e-01 2.44474083e-01 1.62539646e-01 -6.51797533e-01 1.02478400e-01 -7.11703557e-04 3.34057391e-01 -3.29392672e-01 1.12520528e+00 1.82271525e-01 6.38338685e-01 -3.42593014e-01 4.35356230e-01 4.32155401e-01 5.09666443e-01 -9.06466991e-02 8.57212961e-01 5.98356783e-01 -4.85563390e-02 -9.92048502e-01 -6.95929587e-01 -4.18381691e-01 -8.72837961e-01 -6.20541386e-02 5.29642820e-01 -7.38357604e-01 -6.76477730e-01 4.97878045e-01 -9.54427660e-01 -9.26355496e-02 -6.31188631e-01 4.79155660e-01 -4.59168553e-01 2.66208500e-01 -5.13516545e-01 -3.68967175e-01 -1.03370063e-01 -9.32044625e-01 5.47076166e-01 -1.70709163e-01 -3.59587580e-01 -1.00768018e+00 1.66580305e-01 -1.15484618e-01 4.51027870e-01 -3.16084251e-02 7.65332580e-01 -7.44579613e-01 -8.15176368e-02 -4.05416995e-01 -3.37337911e-01 1.01803887e+00 1.93834022e-01 -9.21027437e-02 -1.04192698e+00 -4.22428966e-01 1.82715863e-01 -5.02628863e-01 1.10954964e+00 1.58786431e-01 8.38098943e-01 -4.39233929e-01 -1.55597448e-01 6.30536377e-01 1.26718938e+00 -2.32515916e-01 3.92988682e-01 2.96377957e-01 6.24247074e-01 7.75696278e-01 -1.84444815e-01 1.47366017e-01 1.58175617e-01 4.68569398e-01 3.56610298e-01 -5.65689616e-02 -2.79465675e-01 -1.17572173e-01 3.57960671e-01 8.97224426e-01 1.54551759e-01 3.40769142e-01 -7.56451786e-01 5.95407724e-01 -1.83745921e+00 -1.10254228e+00 -6.03704117e-02 2.26251507e+00 6.80088818e-01 4.58431393e-01 -1.54462345e-02 4.15236562e-01 5.85456669e-01 3.19676131e-01 -7.43175924e-01 -3.59518260e-01 -3.29468846e-01 2.11347491e-01 6.39822781e-01 2.46292740e-01 -1.12114513e+00 4.53503132e-01 7.05782986e+00 5.61177671e-01 -1.14646649e+00 1.79602832e-01 4.38203573e-01 -3.50562334e-01 -2.70939004e-02 -1.21566013e-01 -6.39445662e-01 4.38895971e-01 7.94226766e-01 -2.36922652e-01 5.89445174e-01 9.24398124e-01 -1.43025652e-01 7.72570958e-03 -1.51012266e+00 1.06729257e+00 3.19952101e-01 -1.51189244e+00 -4.28622402e-02 -7.68327191e-02 7.39903927e-01 5.72166443e-01 1.82159111e-01 3.85447145e-02 2.10250601e-01 -9.64136004e-01 6.36687636e-01 2.09370837e-01 5.08169174e-01 -3.90028834e-01 5.29147625e-01 3.48845184e-01 -8.71870160e-01 -2.19962403e-01 -6.03317380e-01 -3.12547714e-01 -4.74802405e-01 7.63174236e-01 -6.69507682e-01 2.02981476e-02 7.46756315e-01 1.11800790e+00 -7.00548589e-01 8.62607718e-01 2.88168132e-01 2.89203316e-01 -4.39664841e-01 7.83316419e-02 2.22898051e-01 -3.07439506e-01 3.86957854e-01 1.10242140e+00 -1.83352157e-01 -3.12434733e-01 -9.22165215e-02 7.59011209e-01 -1.71076596e-01 3.09350062e-02 -1.00180721e+00 1.31585803e-02 1.64594904e-01 1.20031834e+00 -5.79982519e-01 -3.13201040e-01 -3.45211178e-01 1.02668166e+00 7.94142306e-01 1.66937038e-01 -2.93558359e-01 -3.60041082e-01 1.00083625e+00 1.33218542e-01 3.35917026e-01 -2.34594196e-01 -4.66159046e-01 -9.98114169e-01 2.31785715e-01 -7.68070757e-01 2.13111848e-01 -4.87044245e-01 -1.48616624e+00 6.40874863e-01 -2.60558039e-01 -1.37772048e+00 -1.07135445e-01 -8.26247454e-01 -6.05840802e-01 5.11701345e-01 -1.33151639e+00 -6.53470278e-01 -7.99138993e-02 6.77809834e-01 3.19397956e-01 -3.70527416e-01 8.14116716e-01 3.65031898e-01 -5.67256570e-01 6.83717787e-01 4.14660960e-01 1.55614734e-01 8.08510423e-01 -1.15541017e+00 4.42994088e-01 6.41862512e-01 4.37535852e-01 6.11768365e-01 5.94779968e-01 -1.67284161e-01 -1.31970286e+00 -8.63890171e-01 8.92726064e-01 -2.99076736e-01 9.97825503e-01 -6.15272224e-01 -9.79143143e-01 5.71431816e-01 7.87520260e-02 5.39606214e-01 3.64065915e-01 -2.50494350e-02 -9.04454648e-01 -4.73568767e-01 -9.90028620e-01 5.54826260e-01 1.10071135e+00 -7.22301364e-01 -5.68300724e-01 5.26320457e-01 4.62333322e-01 -5.93189336e-02 -6.27594709e-01 1.85210288e-01 7.32056677e-01 -1.24569178e+00 9.28622127e-01 -7.78542757e-01 5.55509269e-01 -8.14747140e-02 -9.98918563e-02 -1.54284811e+00 -3.71490866e-01 -3.61942023e-01 1.39874279e-01 7.55131304e-01 3.27987313e-01 -8.29824805e-01 6.33001447e-01 3.55718434e-01 9.76832807e-02 -7.69901633e-01 -1.05507815e+00 -1.10086441e+00 8.06020424e-02 -4.13004845e-01 2.58013476e-02 1.11144841e+00 7.77169317e-02 4.44450319e-01 -8.16716403e-02 -1.52119741e-01 1.04668045e+00 -2.07638919e-01 5.37866831e-01 -1.37965083e+00 -1.78070724e-01 -7.65162528e-01 -1.03038907e+00 -6.28180206e-01 2.21166685e-01 -1.16423273e+00 -6.48025647e-02 -1.13555789e+00 6.50851279e-02 -4.18046713e-01 -3.67243201e-01 5.10026693e-01 2.76900887e-01 4.81630832e-01 2.40367532e-01 3.00312907e-01 -3.78576189e-01 5.70670903e-01 8.98102224e-01 -3.70119423e-01 2.22876072e-02 -6.10362403e-02 -5.71392179e-01 7.68730938e-01 7.45857894e-01 -6.95350945e-01 -4.00548309e-01 -5.32098413e-01 1.54549986e-01 -5.36635280e-01 5.16750932e-01 -1.07565045e+00 5.47172308e-01 3.23160172e-01 4.44994003e-01 -1.54931070e-02 3.00636381e-01 -7.67725766e-01 -2.52662860e-02 6.45312369e-01 -5.36661267e-01 1.24015167e-01 3.44402865e-02 4.13718909e-01 -1.74574718e-01 -3.86248410e-01 1.13300514e+00 -5.09716123e-02 -4.52490807e-01 2.70063698e-01 -2.57626832e-01 1.22295909e-01 1.03876817e+00 -3.71562719e-01 -3.33510071e-01 -1.13746904e-01 -5.96447766e-01 1.14400283e-01 2.81906337e-01 6.64779842e-01 5.87727308e-01 -1.44925773e+00 -6.76116824e-01 6.23543739e-01 1.24742486e-01 -1.85703561e-01 -2.02594995e-02 9.10525739e-01 -2.20197648e-01 1.37259036e-01 -7.01500833e-01 -6.62935436e-01 -1.28040290e+00 4.12103385e-01 5.04698992e-01 3.22210193e-02 -6.11168921e-01 8.58362436e-01 1.00646250e-01 -5.73360264e-01 3.97664785e-01 -1.65087178e-01 -5.15515730e-02 4.21596020e-01 6.50387883e-01 3.01845133e-01 3.67474049e-01 -4.34595466e-01 -2.89644450e-01 5.08368313e-01 -4.30087745e-01 -7.30063394e-02 1.66949964e+00 2.15504110e-01 -2.82806933e-01 6.92839444e-01 1.68016016e+00 -4.66151536e-01 -1.40275490e+00 -5.82706273e-01 -2.87015401e-02 -3.27677757e-01 1.55979827e-01 -1.89136609e-01 -1.16207707e+00 8.04560542e-01 7.16491342e-01 4.34198618e-01 6.52006209e-01 -1.95253249e-02 3.61101806e-01 7.54124224e-01 8.48147348e-02 -1.08568966e+00 2.64562786e-01 5.84874988e-01 9.64130044e-01 -1.35268521e+00 1.55607224e-01 6.02338761e-02 -4.95474488e-01 1.11597991e+00 5.43465555e-01 -6.43561780e-01 8.96154165e-01 3.30938518e-01 5.48111573e-02 -1.08262382e-01 -9.36252713e-01 -1.55846104e-01 2.51776546e-01 6.53822780e-01 8.50679651e-02 -8.81381407e-02 -3.31792496e-02 -2.58137584e-02 -2.61014670e-01 -3.56239557e-01 3.94188643e-01 7.00583100e-01 -3.55767310e-01 -9.14143085e-01 -2.08372712e-01 7.86177278e-01 -1.27485748e-02 -1.24738343e-01 -3.25431138e-01 5.21802664e-01 6.75345436e-02 5.88535726e-01 3.81849021e-01 -4.81530964e-01 4.13965762e-01 -1.10232094e-02 6.01873755e-01 -5.60965836e-01 -5.34002185e-01 -4.12169546e-01 -3.26637924e-01 -6.33503675e-01 -2.97531009e-01 -7.64646351e-01 -8.86478722e-01 -3.82597923e-01 -2.19205618e-01 -1.79339021e-01 8.05201292e-01 9.62568998e-01 2.63377786e-01 1.81445405e-01 8.39653015e-01 -1.17361581e+00 -9.46860552e-01 -8.61970603e-01 -6.00213289e-01 7.28367090e-01 6.12399995e-01 -5.02904296e-01 -6.56755030e-01 -6.20515533e-02]
[9.298491477966309, 2.845961332321167]
5856e751-c04c-4d3d-ae30-57096931ba76
bilingual-topic-models-for-comparable-corpora
2111.15278
null
https://arxiv.org/abs/2111.15278v1
https://arxiv.org/pdf/2111.15278v1.pdf
Bilingual Topic Models for Comparable Corpora
Probabilistic topic models like Latent Dirichlet Allocation (LDA) have been previously extended to the bilingual setting. A fundamental modeling assumption in several of these extensions is that the input corpora are in the form of document pairs whose constituent documents share a single topic distribution. However, this assumption is strong for comparable corpora that consist of documents thematically similar to an extent only, which are, in turn, the most commonly available or easy to obtain. In this paper we relax this assumption by proposing for the paired documents to have separate, yet bound topic distributions. % a binding mechanism between the distributions of the paired documents. We suggest that the strength of the bound should depend on each pair's semantic similarity. To estimate the similarity of documents that are written in different languages we use cross-lingual word embeddings that are learned with shallow neural networks. We evaluate the proposed binding mechanism by extending two topic models: a bilingual adaptation of LDA that assumes bag-of-words inputs and a model that incorporates part of the text structure in the form of boundaries of semantically coherent segments. To assess the performance of the novel topic models we conduct intrinsic and extrinsic experiments on five bilingual, comparable corpora of English documents with French, German, Italian, Spanish and Portuguese documents. The results demonstrate the efficiency of our approach in terms of both topic coherence measured by the normalized point-wise mutual information, and generalization performance measured by perplexity and in terms of Mean Reciprocal Rank in a cross-lingual document retrieval task for each of the language pairs.
['Marianne Clausel', 'Massih-Reza Amini', 'Georgios Balikas']
2021-11-30
null
null
null
null
['topic-models']
['natural-language-processing']
[-4.43841100e-01 1.91533878e-01 -2.73247451e-01 -5.69410622e-01 -9.94898200e-01 -7.57861376e-01 1.16579771e+00 3.96374643e-01 -4.44682151e-01 5.85185528e-01 5.96874893e-01 -1.06162347e-01 -2.29705602e-01 -7.69606531e-01 -5.21545827e-01 -7.87494719e-01 7.61452988e-02 1.03758109e+00 1.73635036e-01 -1.30886346e-01 8.91203061e-02 1.51493385e-01 -1.35134315e+00 3.85435402e-01 8.32380533e-01 6.17732465e-01 3.02753985e-01 5.03147878e-02 -4.39738244e-01 -1.69189379e-01 -3.91990423e-01 -4.92261618e-01 2.28937879e-01 -2.83616394e-01 -8.66400480e-01 1.43436655e-01 3.46086293e-01 -1.48956239e-01 -1.10682473e-01 9.69257593e-01 2.69548208e-01 4.92083356e-02 1.06480241e+00 -9.80288863e-01 -5.27852774e-01 8.65217745e-01 -5.37638843e-01 1.26648620e-01 2.12634176e-01 -3.61331195e-01 1.33145940e+00 -9.79961932e-01 7.70073175e-01 1.32549560e+00 5.62305987e-01 3.03110242e-01 -1.57236660e+00 -3.58152002e-01 1.81193396e-01 -1.30501792e-01 -1.39241803e+00 -2.36821741e-01 6.76893055e-01 -6.29765451e-01 9.63685513e-01 -2.71184146e-01 4.42063540e-01 1.23849189e+00 2.51173466e-01 5.30644894e-01 1.15166926e+00 -7.28372693e-01 2.73825467e-01 8.12127292e-01 4.38143581e-01 1.41505271e-01 3.07437539e-01 -3.39692771e-01 -5.98534822e-01 -4.12945569e-01 1.72803938e-01 -2.49010950e-01 -2.38063321e-01 -8.71407568e-01 -1.09893095e+00 1.23949265e+00 7.51451552e-02 7.90487170e-01 -3.12567472e-01 -2.33024389e-01 6.29863620e-01 9.88106653e-02 7.66925037e-01 2.01990694e-01 -3.46082926e-01 1.59130141e-01 -1.10321438e+00 3.00456882e-01 1.06107199e+00 9.48444545e-01 7.82623112e-01 -5.17393291e-01 7.65820444e-02 8.84733796e-01 7.03842998e-01 2.87061095e-01 9.72887814e-01 -4.09479022e-01 3.52258563e-01 3.27995628e-01 1.46610349e-01 -9.38940465e-01 -1.20299257e-01 -2.42557004e-01 -3.24112177e-01 -3.90726477e-01 4.77761149e-01 7.01908171e-02 -5.00371218e-01 2.09017181e+00 2.19162196e-01 -4.04649436e-01 3.65932107e-01 7.03189135e-01 4.62952882e-01 8.53380442e-01 2.11175069e-01 -2.65038967e-01 1.61210048e+00 -8.41423571e-01 -6.82564795e-01 -1.06681570e-01 6.48175240e-01 -9.73673284e-01 1.18415618e+00 2.02815995e-01 -1.02594006e+00 -4.45080608e-01 -9.78922248e-01 -2.15627864e-01 -6.51799440e-01 6.44554198e-02 4.60693657e-01 7.64452815e-01 -1.11526060e+00 4.27187681e-01 -8.45124185e-01 -9.91899431e-01 -9.23816636e-02 2.35595986e-01 -3.88606876e-01 -5.31142391e-02 -1.39832890e+00 8.36423278e-01 5.95621169e-01 -2.49935135e-01 -7.61381447e-01 -4.17855531e-01 -7.48745263e-01 3.31757158e-01 -2.69462436e-01 -4.96595412e-01 8.98911178e-01 -8.69333327e-01 -1.15623152e+00 1.16684031e+00 -1.00781307e-01 -3.95810425e-01 3.09813142e-01 -2.25997776e-01 -2.15227559e-01 1.31373361e-01 4.74764794e-01 9.34289694e-01 4.44990844e-01 -1.17931211e+00 -5.33950269e-01 -5.33779025e-01 -7.55895078e-02 4.17880833e-01 -7.42784381e-01 1.53064445e-01 -4.54824328e-01 -5.18500626e-01 2.73825496e-01 -9.31486547e-01 2.11426288e-01 -2.20486104e-01 -1.32184938e-01 -6.22555315e-01 6.95832253e-01 -6.46711528e-01 8.49015951e-01 -2.30128932e+00 2.29303464e-01 1.33992761e-01 -2.16020614e-01 -3.33254069e-01 -9.34732705e-02 7.89988518e-01 -1.23000875e-01 -5.43027893e-02 -1.32574245e-01 -5.30629456e-01 2.65953153e-01 2.39318162e-01 -7.41292655e-01 6.42869949e-01 -4.41614203e-02 3.88530731e-01 -7.42200673e-01 -4.82705086e-01 -2.41838306e-01 5.38804710e-01 -5.29572010e-01 1.32963806e-01 -2.27222100e-01 1.66729525e-01 -1.59932584e-01 -2.79002078e-02 5.27412593e-01 1.26050398e-01 6.54687583e-01 -1.34903759e-01 -6.04047999e-02 8.90125930e-01 -9.78393555e-01 2.06009030e+00 -4.14108515e-01 7.35635459e-01 -3.18276435e-01 -9.73463178e-01 1.01402128e+00 6.71784282e-01 3.83347929e-01 -4.12071794e-01 -8.83582085e-02 3.58783722e-01 -4.83464263e-02 -1.08435526e-01 6.75359845e-01 -3.65728050e-01 -2.55756140e-01 8.37983012e-01 3.74758393e-01 -2.25337014e-01 4.66575593e-01 3.63295019e-01 4.14404035e-01 1.64631858e-01 3.36471498e-01 -9.25631404e-01 2.64258474e-01 -9.14571807e-02 2.47282878e-01 4.57289755e-01 2.87023243e-02 4.41568017e-01 8.31984639e-01 -1.06398277e-01 -1.36640394e+00 -1.37693238e+00 -6.58425331e-01 1.19218135e+00 3.41065787e-02 -4.30090874e-01 -6.88367367e-01 -6.44951820e-01 -1.36688441e-01 1.03973734e+00 -5.28279066e-01 1.41537879e-02 -3.59184861e-01 -8.44815910e-01 3.60363483e-01 3.03637475e-01 -5.74175036e-03 -7.94280529e-01 -3.26650381e-01 4.27973233e-02 -2.58175969e-01 -1.04580986e+00 -3.46852720e-01 3.41039985e-01 -9.43153322e-01 -5.51586390e-01 -7.75039315e-01 -8.47969472e-01 5.26225686e-01 9.39447358e-02 1.13998139e+00 -5.51553547e-01 2.17013925e-01 5.03069282e-01 -2.80841798e-01 -1.29744276e-01 -3.41549754e-01 2.93056726e-01 3.59422982e-01 -2.14762196e-01 8.32470953e-01 -6.56205714e-01 -3.47327292e-01 3.26251656e-01 -1.13829720e+00 -1.31770700e-01 2.96447337e-01 8.86638999e-01 2.38388613e-01 -5.34237288e-02 3.94785821e-01 -7.12593913e-01 7.91785121e-01 -7.28797555e-01 -4.39019978e-01 2.01860100e-01 -5.71518779e-01 1.53177172e-01 1.89454958e-01 -4.10977662e-01 -1.19519150e+00 -3.39295864e-01 2.05997854e-01 -2.37465519e-02 -3.66737932e-01 6.02936804e-01 -2.52299517e-01 6.43970728e-01 6.09776378e-01 2.36858785e-01 -2.22806066e-01 -4.65777606e-01 6.40627742e-01 8.89437556e-01 9.66778472e-02 -9.32024479e-01 3.44893336e-01 5.98015904e-01 -5.41692972e-01 -8.42252433e-01 -6.57264948e-01 -8.37339461e-01 -8.62378836e-01 2.39347756e-01 9.29619730e-01 -1.04847205e+00 -1.51031598e-01 9.25647765e-02 -1.54897273e+00 7.87080899e-02 -2.17204228e-01 9.39691246e-01 -5.60398459e-01 4.60643619e-01 -6.88329101e-01 -5.21970868e-01 -7.72731528e-02 -1.14197683e+00 1.20375454e+00 -1.52257919e-01 -4.74378139e-01 -1.38681197e+00 5.31877160e-01 -8.74996036e-02 2.26614043e-01 -4.04266179e-01 1.44647896e+00 -1.28600907e+00 -5.85588440e-02 -1.13949060e-01 -2.99633920e-01 3.67234260e-01 1.90405294e-01 -2.67884582e-01 -1.07468975e+00 -5.57426929e-01 2.17180893e-01 -2.54435271e-01 9.13249910e-01 3.56713802e-01 2.36888766e-01 -9.99844447e-03 -5.22313237e-01 1.16818823e-01 1.52219021e+00 1.36966854e-01 4.96782720e-01 2.15243876e-01 1.96239650e-01 1.16506660e+00 1.96765214e-01 1.46368816e-01 2.74634272e-01 9.16059852e-01 -1.48575649e-01 2.33534753e-01 6.22148104e-02 -1.33373603e-01 6.09528303e-01 1.34396994e+00 3.52688283e-01 -3.87721866e-01 -1.10428572e+00 1.03930604e+00 -1.58836126e+00 -7.84992218e-01 1.91509008e-01 2.23578477e+00 9.66789007e-01 2.26566732e-01 -1.05108286e-03 -2.28992626e-01 6.93576455e-01 1.60806801e-03 1.24697998e-01 -5.65463841e-01 -1.28046766e-01 -5.62574044e-02 1.49076544e-02 5.56834280e-01 -1.06703222e+00 9.46991861e-01 5.93292522e+00 8.44384313e-01 -1.04315257e+00 3.71274233e-01 3.55887294e-01 8.96566287e-02 -5.02161205e-01 2.44528741e-01 -1.07455909e+00 4.28876877e-01 1.05874991e+00 -2.30861977e-01 -7.61049315e-02 9.30113435e-01 -4.42920662e-02 -1.62602887e-01 -1.44883001e+00 3.95177662e-01 2.14782715e-01 -7.91814923e-01 1.62105039e-01 2.31347561e-01 8.46925259e-01 9.67936218e-02 1.65610760e-01 2.83053339e-01 1.90012679e-01 -6.58242702e-01 6.12813592e-01 2.52334028e-01 4.78565484e-01 -3.98905009e-01 8.67163718e-01 2.46545643e-01 -8.32195401e-01 3.59656900e-01 -6.32939756e-01 2.54809082e-01 1.63154483e-01 5.97203434e-01 -8.83135617e-01 4.47160780e-01 6.46140397e-01 4.24865067e-01 -2.25436822e-01 7.42397070e-01 -1.68452531e-01 6.06874228e-01 -4.24402028e-01 5.32749593e-02 4.55990076e-01 -3.83988410e-01 6.29133999e-01 1.37621713e+00 3.71958137e-01 -5.06428659e-01 1.19871259e-01 9.09695864e-01 2.05264926e-01 5.26738107e-01 -6.40641093e-01 -1.27338201e-01 3.89132977e-01 1.03999853e+00 -9.57018137e-01 -3.88700694e-01 -5.72635949e-01 8.57229173e-01 3.53717953e-01 3.39310944e-01 -5.65654516e-01 -1.99635789e-01 4.58214700e-01 -6.50009587e-02 5.04805267e-01 -3.49055618e-01 -1.17138745e-02 -1.17832255e+00 1.38026252e-01 -5.80023944e-01 2.76705325e-01 -5.56368709e-01 -1.60388863e+00 8.43163311e-01 4.03036445e-01 -9.47678268e-01 -4.40513879e-01 -5.42099535e-01 -3.66793901e-01 1.21179175e+00 -1.40557277e+00 -1.30184996e+00 1.53745294e-01 3.49474669e-01 6.62886024e-01 -2.87540793e-01 1.01268864e+00 3.11088741e-01 -2.17549190e-01 2.86661685e-01 5.67529678e-01 -8.51275250e-02 1.07532740e+00 -1.29195833e+00 2.38776952e-01 4.99254286e-01 4.69448090e-01 1.07770658e+00 8.09679568e-01 -6.99659884e-01 -5.79749882e-01 -6.44544661e-01 1.49422884e+00 -3.36549103e-01 8.27369690e-01 -7.66777515e-01 -9.71210063e-01 7.65231371e-01 6.87684536e-01 -4.80217546e-01 9.53891277e-01 5.58978677e-01 -6.09414279e-01 8.79398286e-02 -7.29674399e-01 2.82930613e-01 4.04897094e-01 -7.79493093e-01 -1.12095320e+00 5.83572030e-01 6.75846040e-01 1.25642329e-01 -7.80688167e-01 -8.05090293e-02 5.54939270e-01 -9.59840357e-01 5.78268230e-01 -7.61148512e-01 4.65624303e-01 2.88258278e-04 -5.28101563e-01 -1.37418365e+00 -1.00099146e-01 -9.38234180e-02 5.59207916e-01 1.48576176e+00 4.26087946e-01 -5.36032915e-01 5.04331708e-01 3.82270187e-01 -6.67195767e-02 -5.08647978e-01 -1.01317549e+00 -8.11991215e-01 6.34332716e-01 -3.29268634e-01 3.05297643e-01 1.06277359e+00 3.31966460e-01 6.35517120e-01 1.80772245e-02 1.45298550e-02 3.25616926e-01 1.61929563e-01 4.70558196e-01 -1.43108296e+00 -3.45992148e-01 -4.12014812e-01 -3.32350552e-01 -9.62724328e-01 5.14310718e-01 -9.75034833e-01 1.08235374e-01 -1.32492423e+00 5.11252880e-01 -4.68709826e-01 -8.20603520e-02 1.76803887e-01 3.22861314e-01 -1.21011967e-02 -1.78315639e-01 6.01192951e-01 -3.92699689e-01 7.15031862e-01 3.45215648e-01 8.72657150e-02 -1.53284714e-01 -3.27932328e-01 -3.55499506e-01 8.24825823e-01 4.70978200e-01 -6.88747883e-01 -2.93773443e-01 -5.14007092e-01 3.55324209e-01 -1.63575202e-01 -4.74209897e-02 -6.85842752e-01 1.25423044e-01 2.92447776e-01 -6.41228957e-03 -5.19264460e-01 3.80558491e-01 -8.57087493e-01 -1.67807415e-01 2.36766174e-01 -5.44594765e-01 -1.66291241e-02 2.52707064e-01 5.91203272e-01 -4.14844245e-01 -4.06590164e-01 6.17582321e-01 4.78866547e-02 -2.42386296e-01 -1.43306091e-01 -4.37060058e-01 -8.96467194e-02 9.45132732e-01 -2.68338174e-02 -1.34812042e-01 -3.11638921e-01 -7.06640363e-01 -1.99953288e-01 4.83110219e-01 3.99926603e-01 1.40371278e-01 -1.32653069e+00 -7.04086959e-01 2.45643239e-02 3.27540874e-01 -5.01846671e-01 -2.43727509e-02 8.95470142e-01 -1.32601127e-01 1.03564370e+00 -2.83819079e-01 -5.93957067e-01 -1.01094890e+00 6.61906123e-01 1.08712360e-01 -4.33450639e-01 -1.81968734e-01 4.72679108e-01 8.37445676e-01 -5.54404259e-01 1.06394604e-01 -4.15789187e-02 -2.49910384e-01 5.16355872e-01 1.14149943e-01 9.41821337e-02 8.55058879e-02 -8.03994954e-01 -2.54955649e-01 4.98348713e-01 -3.82296711e-01 -5.97539604e-01 1.19831896e+00 -3.40664923e-01 -4.70854402e-01 7.42442489e-01 1.32800448e+00 1.85851842e-01 -6.03508174e-01 -4.08714563e-01 2.62277007e-01 -1.68896362e-01 -5.11524454e-02 -4.35263753e-01 -5.48612297e-01 1.08192837e+00 7.06445575e-01 3.38047445e-01 6.09558225e-01 4.93744999e-01 4.00153577e-01 2.61688381e-01 1.44661605e-01 -9.23544824e-01 -1.59547567e-01 3.14557999e-01 7.22816110e-01 -1.08920717e+00 5.21299653e-02 -3.55267197e-01 -3.46332252e-01 1.17283261e+00 2.31702492e-01 -1.03197992e-01 7.24055588e-01 -9.73913893e-02 -8.48833099e-03 -2.62051374e-01 -7.52117515e-01 -9.30039212e-03 4.66200888e-01 4.19421911e-01 9.97148752e-01 -7.27510154e-02 -9.35444653e-01 6.22488797e-01 -2.37872362e-01 -6.10424876e-01 2.63897091e-01 5.29156446e-01 -2.83786029e-01 -1.30687106e+00 -3.14889073e-01 3.30073647e-02 -4.40344691e-01 -3.65461051e-01 -2.04238132e-01 9.26501870e-01 8.94693751e-03 6.56330884e-01 5.04351020e-01 3.49069595e-01 -1.77577123e-01 4.42499459e-01 3.02840948e-01 -7.44180679e-01 -3.42142254e-01 4.37359542e-01 -4.54303548e-02 3.11417598e-02 -5.01437783e-01 -9.24796343e-01 -8.09222043e-01 1.15456007e-01 -4.93467212e-01 6.62115932e-01 1.01056552e+00 1.11075139e+00 1.90150782e-01 -2.23592389e-02 2.63486743e-01 -6.31993651e-01 -6.46407545e-01 -1.32759154e+00 -8.54479671e-01 5.38473725e-01 -2.99471021e-01 -6.66075945e-01 -3.41787100e-01 1.50385574e-01]
[10.474815368652344, 6.998488426208496]
75d6d5f9-247f-44e9-a415-603054ade746
layered-rgbd-scene-flow-estimation
null
null
http://openaccess.thecvf.com/content_cvpr_2015/html/Sun_Layered_RGBD_Scene_2015_CVPR_paper.html
http://openaccess.thecvf.com/content_cvpr_2015/papers/Sun_Layered_RGBD_Scene_2015_CVPR_paper.pdf
Layered RGBD Scene Flow Estimation
As consumer depth sensors become widely available, estimating scene flow from RGBD sequences has received increasing attention. Although the depth information allows the recovery of 3D motion from a single view, it poses new challenges. In particular, depth boundaries are not well-aligned with RGB image edges and therefore not reliable cues to localize 2D motion boundaries. In addition, methods that extend the 2D optical flow formulation to 3D still produce large errors in occlusion regions. To better use depth for occlusion reasoning, we propose a layered RGBD scene flow method that jointly solves for the scene segmentation and the motion. Our key observation is that the noisy depth is sufficient to decide the depth ordering of layers, thereby avoiding a computational bottleneck for RGB layered methods. Furthermore, the depth enables us to estimate a per-layer 3D rigid motion to constrain the motion of each layer. Experimental results on both the Middlebury and real-world sequences demonstrate the effectiveness of the layered approach for RGBD scene flow estimation.
['Deqing Sun', 'Erik B. Sudderth', 'Hanspeter Pfister']
2015-06-01
null
null
null
cvpr-2015-6
['scene-flow-estimation']
['computer-vision']
[ 1.36076346e-01 -1.62563235e-01 -3.01440328e-01 -3.74515027e-01 -4.00605828e-01 -6.32596850e-01 7.52657875e-02 -3.60298634e-01 -3.86316329e-01 5.29243529e-01 2.03814402e-01 -2.29969442e-01 2.78121680e-01 -6.79957509e-01 -3.91430706e-01 -7.08187163e-01 2.04789415e-01 1.10860631e-01 6.12065554e-01 1.66840523e-01 2.81097323e-01 6.36748254e-01 -1.43931782e+00 6.10290915e-02 8.62474740e-01 9.48716879e-01 4.01532888e-01 9.08946931e-01 -3.98885906e-01 8.45295787e-01 -2.53554016e-01 -2.98001878e-02 5.12217462e-01 -4.15658444e-01 -8.00821304e-01 6.26863778e-01 7.34950185e-01 -1.02960515e+00 -6.04664862e-01 8.84589195e-01 2.74794370e-01 2.83785701e-01 2.80608147e-01 -1.21266901e+00 3.62069249e-01 -1.72286794e-01 -9.00827587e-01 2.23424166e-01 8.08884919e-01 1.38789922e-01 7.03464568e-01 -6.67206764e-01 8.10355425e-01 1.12739813e+00 3.64562213e-01 5.10490239e-01 -8.19443524e-01 -3.48529786e-01 5.82787395e-01 9.84912664e-02 -1.09232056e+00 -3.58077914e-01 1.02800310e+00 -5.79566956e-01 6.53105140e-01 1.63134605e-01 9.23719168e-01 6.97373450e-01 -1.33369848e-01 9.61097717e-01 1.06635547e+00 -1.56899378e-01 2.85831064e-01 -2.79199898e-01 6.26848219e-03 8.92550528e-01 3.95519525e-01 -8.57536271e-02 -7.30711818e-01 2.26902738e-01 1.19447291e+00 1.10001475e-01 -7.51119614e-01 -6.36297226e-01 -1.12908614e+00 4.48359907e-01 4.36014771e-01 -6.87965192e-03 -1.46629542e-01 2.73573726e-01 1.40320241e-01 -2.10916504e-01 5.02662241e-01 1.68389678e-02 -4.72634941e-01 -3.23115855e-01 -9.10224915e-01 1.11529373e-01 6.37106180e-01 9.37318504e-01 1.11326969e+00 -1.03451565e-01 4.75911766e-01 2.43518576e-01 5.60149431e-01 6.09392166e-01 6.98825568e-02 -1.58932686e+00 7.58623362e-01 7.11458921e-01 3.78879726e-01 -1.13359892e+00 -4.59751040e-01 1.70728311e-01 -6.79823339e-01 3.81882757e-01 1.05476427e+00 1.13662444e-02 -1.01858699e+00 1.25389957e+00 7.75404036e-01 4.56606656e-01 -9.89065468e-02 1.42426646e+00 6.63626909e-01 6.19512796e-01 -4.40618873e-01 -1.97017848e-01 1.02473915e+00 -8.99872422e-01 -7.62198567e-01 -6.08819604e-01 5.65853000e-01 -6.17059410e-01 7.37982988e-01 3.14341873e-01 -1.06005752e+00 -4.46504414e-01 -1.02811229e+00 -4.36069846e-01 1.27488285e-01 -3.98572743e-01 9.18155730e-01 5.98275959e-01 -8.59611571e-01 2.65949547e-01 -1.39817238e+00 -1.43920809e-01 1.94406167e-01 2.14109704e-01 -5.11236548e-01 -6.68567240e-01 -8.73336732e-01 5.87787926e-01 2.26495147e-01 5.88846505e-01 -6.35230541e-01 -5.84101140e-01 -1.15954971e+00 -3.84088218e-01 3.53583992e-01 -7.35286057e-01 9.94172931e-01 -5.87177038e-01 -1.56969631e+00 7.72713721e-01 -6.25128031e-01 6.65778890e-02 7.95011580e-01 -2.88362652e-01 4.18631047e-01 7.30219603e-01 7.97796324e-02 5.85912943e-01 5.90725422e-01 -1.43288434e+00 -8.30809176e-01 -4.73668993e-01 3.50796163e-01 5.88127553e-01 7.74077252e-02 -5.88966370e-01 -9.47033823e-01 -6.37680888e-02 8.60925794e-01 -8.34530473e-01 -4.48554158e-01 4.42239940e-01 -4.49785411e-01 1.69082761e-01 7.49873161e-01 -6.45305395e-01 9.72007573e-01 -1.84762454e+00 2.72732079e-01 -1.26395831e-02 3.21200222e-01 -1.61581948e-01 3.00606042e-01 -1.87788919e-01 4.35129493e-01 -6.68813363e-02 -4.90331948e-01 -5.65338373e-01 -4.11581725e-01 5.70982814e-01 -1.20850958e-01 8.03807020e-01 -5.80222979e-02 6.19932473e-01 -1.05791545e+00 -6.42873049e-01 7.10438967e-01 5.98746836e-01 -7.71014571e-01 3.38070005e-01 -6.55108243e-02 8.76847863e-01 -7.30367780e-01 6.53840899e-01 9.66437995e-01 -2.67871767e-01 1.85991246e-02 -1.55214384e-01 -3.10782313e-01 4.01962578e-01 -1.40974784e+00 2.22867703e+00 -2.56907344e-01 7.93074489e-01 2.62510449e-01 -5.16208708e-01 4.06663597e-01 1.00514732e-01 8.13712716e-01 -6.51797950e-01 6.85270131e-02 1.66987777e-01 -3.83884043e-01 -6.27080321e-01 5.32853782e-01 -2.23387517e-02 5.15676364e-02 3.77518773e-01 -3.33427131e-01 -6.13908768e-01 2.15189174e-01 1.38086185e-01 1.02471495e+00 6.18104637e-01 -1.40791265e-02 1.55914858e-01 4.18307215e-01 9.37973633e-02 9.66446877e-01 5.13334572e-01 -3.27978224e-01 1.14710045e+00 4.45162237e-01 -4.33617800e-01 -8.05638552e-01 -8.77322257e-01 1.05057552e-01 1.96858257e-01 7.89850116e-01 -1.56483069e-01 -6.51626766e-01 -5.51601708e-01 -1.35140508e-01 1.03302665e-01 -3.87897134e-01 3.98123085e-01 -8.87722373e-01 -5.35459697e-01 9.63467136e-02 3.93880874e-01 7.04472065e-01 -4.32089984e-01 -1.23937345e+00 8.40375274e-02 -8.65712345e-01 -1.78072655e+00 -5.45984566e-01 -2.62515284e-02 -1.20981348e+00 -1.35832167e+00 -7.71728218e-01 -4.38929349e-01 7.72367001e-01 7.28467405e-01 9.42104638e-01 5.18014468e-02 -2.43885547e-01 3.81616145e-01 -2.47815877e-01 2.75807649e-01 -3.54447328e-02 -8.52704123e-02 -1.98316559e-01 -2.54912786e-02 -1.68028250e-02 -4.24034804e-01 -1.03414774e+00 4.07383472e-01 -8.04285169e-01 3.71116132e-01 7.55866840e-02 3.36833209e-01 4.95500594e-01 -2.94550546e-02 -2.95818508e-01 -6.62079990e-01 -3.59540969e-01 -5.69519361e-05 -8.02652359e-01 -1.22076519e-01 -1.39586076e-01 -9.27457139e-02 6.23580329e-02 -1.89413130e-02 -1.33434069e+00 5.53059340e-01 -5.42619824e-02 -5.80015182e-01 -2.64635950e-01 6.65809307e-03 -3.16431999e-01 -4.32753079e-02 1.24162287e-01 4.82761498e-05 -2.33766139e-01 -3.52128953e-01 4.03767377e-01 3.80438864e-01 4.66833949e-01 -3.13473105e-01 6.69009566e-01 1.12550449e+00 1.28653452e-01 -1.03953505e+00 -8.44808400e-01 -9.22650099e-01 -1.04757726e+00 -4.66997981e-01 1.31070888e+00 -1.05878997e+00 -8.98899436e-01 6.28628254e-01 -1.43719172e+00 -5.64802170e-01 -1.32139139e-02 6.97377980e-01 -5.67715287e-01 7.99295604e-01 -9.02758181e-01 -8.93981874e-01 2.57524192e-01 -1.41364717e+00 1.25416684e+00 2.45577320e-01 -3.60652693e-02 -1.12451446e+00 -1.35531515e-01 5.70831239e-01 -2.47937635e-01 5.46670377e-01 5.42967081e-01 7.23301411e-01 -1.23316765e+00 2.97130067e-02 -2.11366564e-01 1.07809315e-02 2.46097758e-01 -5.76784648e-02 -1.24323666e+00 9.56334174e-02 2.86725700e-01 -1.57988425e-02 8.05344462e-01 5.58012307e-01 6.06068611e-01 1.54554367e-01 -2.06688195e-01 1.05658400e+00 1.48287034e+00 1.36774868e-01 6.13617182e-01 3.68168294e-01 1.33270276e+00 7.20778942e-01 7.35027909e-01 4.29725349e-01 6.24138832e-01 6.34549141e-01 4.86495733e-01 -2.25158021e-01 -3.25029075e-01 -2.13080436e-01 1.94531843e-01 8.99533510e-01 -2.65495330e-01 -2.95147747e-01 -9.82347131e-01 3.41078639e-01 -1.85098970e+00 -5.94910264e-01 -5.91079712e-01 2.25041342e+00 6.98563993e-01 8.77553001e-02 -1.73054978e-01 4.03616846e-01 3.77617925e-01 2.24870041e-01 -6.29840374e-01 1.55719846e-01 -2.08751410e-01 -1.86948851e-01 7.73617148e-01 1.06012332e+00 -8.24266195e-01 9.72320259e-01 5.89935017e+00 1.07995450e-01 -1.08283138e+00 -9.03469622e-02 4.62983280e-01 -7.58173019e-02 -3.96985292e-01 2.69663692e-01 -8.44285667e-01 3.08730870e-01 1.12732768e-01 4.26223367e-01 1.45222425e-01 4.14951980e-01 5.81039011e-01 -8.98703396e-01 -9.98916388e-01 1.30755556e+00 -4.96228524e-02 -8.70897770e-01 -2.21079707e-01 2.55092442e-01 9.41917717e-01 -6.31609038e-02 -4.00390387e-01 -5.19407928e-01 5.24611101e-02 -5.06072640e-01 7.46029913e-01 4.04239953e-01 6.16447031e-01 -5.18582582e-01 4.52992111e-01 4.21513915e-01 -1.22423315e+00 2.83178180e-01 -3.60832453e-01 -3.53351325e-01 7.41525769e-01 8.12374294e-01 -5.38069665e-01 6.04065657e-01 7.15881050e-01 1.03540838e+00 -2.62326837e-01 1.08632684e+00 -5.76543868e-01 1.02895580e-01 -5.04517198e-01 5.02219975e-01 2.20935404e-01 -4.47709739e-01 4.56918925e-01 8.34656596e-01 1.80475056e-01 3.94883215e-01 3.25978130e-01 6.72215223e-01 2.63370574e-01 -4.65144724e-01 -5.65228403e-01 2.76266605e-01 4.93413061e-02 9.74901319e-01 -1.22735155e+00 -7.05737025e-02 -6.08436584e-01 1.49229765e+00 4.92105484e-02 8.03779066e-01 -5.39160073e-01 -8.97873566e-02 9.63613808e-01 1.43464178e-01 2.61895269e-01 -8.65916371e-01 -4.53370452e-01 -1.58533180e+00 8.51619616e-02 -2.02387601e-01 9.82430205e-02 -7.96720862e-01 -8.39328051e-01 2.79293299e-01 -5.57938293e-02 -1.35448813e+00 -2.87676603e-01 -6.43962801e-01 2.73435051e-03 7.67882824e-01 -1.98331118e+00 -5.90337038e-01 -6.98842227e-01 6.12761319e-01 5.74391484e-01 8.63416910e-01 3.42908382e-01 4.23912197e-01 -5.17812192e-01 -1.31553918e-01 -3.22088450e-01 2.90640980e-01 4.08184618e-01 -1.12979519e+00 3.35007459e-01 1.18076944e+00 -4.19877693e-02 3.17811102e-01 4.99705583e-01 -6.22840583e-01 -1.68189907e+00 -6.55450940e-01 5.96020460e-01 -5.96358657e-01 1.99198514e-01 -4.48626369e-01 -8.15225899e-01 4.29571599e-01 -2.00938642e-01 3.07221413e-01 2.13899434e-01 -4.73978221e-01 -1.44359633e-01 -1.61644276e-02 -9.48810756e-01 3.39154601e-01 1.30190516e+00 -6.21282995e-01 -8.68127793e-02 -1.19804129e-01 6.52951002e-01 -9.12640870e-01 -6.51728332e-01 3.46786588e-01 5.34137666e-01 -1.38534486e+00 1.06321049e+00 1.80171922e-01 4.19908941e-01 -7.43109167e-01 -1.33826613e-01 -8.26888561e-01 4.37075078e-01 -5.38700938e-01 -2.51063228e-01 8.97810340e-01 1.03997774e-02 -3.52400273e-01 1.29713738e+00 1.16343856e+00 -9.09556001e-02 -3.43866706e-01 -9.53912735e-01 -2.29297981e-01 -4.07656789e-01 -8.19304347e-01 4.95102443e-02 1.00603688e+00 -2.58738935e-01 2.09712595e-01 -2.50288844e-01 3.29392046e-01 9.65103686e-01 3.38506848e-01 9.30206895e-01 -1.01302314e+00 -2.61568695e-01 -2.69730687e-01 -6.22301102e-01 -2.05660820e+00 3.10311560e-02 -4.18004662e-01 4.88070637e-01 -1.86817873e+00 -1.52569592e-01 -6.64021790e-01 1.70749843e-01 5.03599085e-02 -3.15853775e-01 3.30893725e-01 9.04750079e-02 2.28407323e-01 -4.54121649e-01 4.58661050e-01 1.68749630e+00 1.11491732e-01 -3.77309829e-01 1.00781699e-03 -5.03309816e-02 1.03214145e+00 4.68393773e-01 -2.94619292e-01 -6.28495753e-01 -8.65481615e-01 2.73231655e-01 5.17857850e-01 2.24251240e-01 -7.53101707e-01 2.95323312e-01 -2.17810512e-01 4.80022609e-01 -8.37691128e-01 5.53053141e-01 -9.37243998e-01 -2.11587116e-01 3.37465733e-01 1.75586969e-01 -1.72462642e-01 3.37851420e-02 7.65802383e-01 -2.04498425e-01 -1.44419372e-01 5.45905888e-01 -4.39246595e-01 -9.88581955e-01 4.62767273e-01 -2.73573399e-01 2.04030618e-01 7.69142330e-01 -6.59505546e-01 5.25537692e-02 -4.01835501e-01 -4.89837646e-01 3.18774343e-01 8.22008550e-01 2.46356592e-01 9.25608993e-01 -8.54577243e-01 -2.56790996e-01 3.12304646e-01 -1.84140757e-01 8.50528061e-01 2.97675461e-01 8.49255860e-01 -1.12882161e+00 2.62677401e-01 1.41222209e-01 -1.11357641e+00 -1.03155613e+00 2.08881572e-01 4.77720410e-01 1.13036215e-01 -9.72854853e-01 8.16842198e-01 6.47262573e-01 -1.73726246e-01 3.28316510e-01 -8.08047831e-01 1.52700111e-01 -7.23471567e-02 6.30135775e-01 5.54669142e-01 -8.13202411e-02 -7.61028409e-01 -4.48483855e-01 1.16534472e+00 3.14845860e-01 -4.97805774e-01 1.00927806e+00 -8.42471063e-01 -2.99552623e-02 7.11596251e-01 1.23606360e+00 -1.12894706e-01 -2.02458453e+00 8.16611648e-02 -1.13817699e-01 -9.25420105e-01 1.99707240e-01 -5.42564839e-02 -1.36056554e+00 1.25424314e+00 2.92597413e-01 -3.87367345e-02 1.10694218e+00 -2.74627268e-01 9.61153924e-01 1.25382900e-01 5.93395948e-01 -7.43313849e-01 1.36489123e-02 4.49760050e-01 1.66609019e-01 -1.27503538e+00 2.16438904e-01 -9.38071072e-01 -3.21565241e-01 1.38072562e+00 7.06852138e-01 7.01561421e-02 4.66732532e-01 3.51472795e-01 3.01077634e-01 6.10837340e-02 -1.98852405e-01 -3.78959119e-01 8.71366542e-03 5.07151961e-01 3.36742312e-01 -3.25556368e-01 1.80191904e-01 -4.04906213e-01 1.63915560e-01 1.96056589e-02 7.72619069e-01 1.03474498e+00 -3.13355237e-01 -9.48290408e-01 -4.23827440e-01 -2.07465157e-01 -3.34111571e-01 2.18560711e-01 -1.00374363e-01 8.15718591e-01 -7.96629954e-03 1.06570637e+00 1.53389230e-01 -7.83314779e-02 3.77761275e-01 -2.45215863e-01 8.16741168e-01 -4.66664791e-01 4.52091247e-02 3.70521396e-01 -1.19028576e-01 -1.00049031e+00 -8.88367295e-01 -6.59703493e-01 -1.64706802e+00 -1.64580107e-01 -2.40190789e-01 -1.62895203e-01 7.68758595e-01 8.75032127e-01 -5.93207888e-02 2.95483172e-01 5.85722566e-01 -1.21364093e+00 3.74764115e-01 -4.42336142e-01 -5.21383286e-01 3.96546036e-01 7.94296026e-01 -5.32933712e-01 -6.96996987e-01 2.82038301e-01]
[8.551708221435547, -2.084460973739624]
a78f175e-99e0-4b2d-904a-8a4e7b9c3565
how-effective-are-neural-networks-for-fixing
2305.18607
null
https://arxiv.org/abs/2305.18607v1
https://arxiv.org/pdf/2305.18607v1.pdf
How Effective Are Neural Networks for Fixing Security Vulnerabilities
Security vulnerability repair is a difficult task that is in dire need of automation. Two groups of techniques have shown promise: (1) large code language models (LLMs) that have been pre-trained on source code for tasks such as code completion, and (2) automated program repair (APR) techniques that use deep learning (DL) models to automatically fix software bugs. This paper is the first to study and compare Java vulnerability repair capabilities of LLMs and DL-based APR models. The contributions include that we (1) apply and evaluate five LLMs (Codex, CodeGen, CodeT5, PLBART and InCoder), four fine-tuned LLMs, and four DL-based APR techniques on two real-world Java vulnerability benchmarks (Vul4J and VJBench), (2) design code transformations to address the training and test data overlapping threat to Codex, (3) create a new Java vulnerability repair benchmark VJBench, and its transformed version VJBench-trans and (4) evaluate LLMs and APR techniques on the transformed vulnerabilities in VJBench-trans. Our findings include that (1) existing LLMs and APR models fix very few Java vulnerabilities. Codex fixes 10.2 (20.4%), the most number of vulnerabilities. (2) Fine-tuning with general APR data improves LLMs' vulnerability-fixing capabilities. (3) Our new VJBench reveals that LLMs and APR models fail to fix many Common Weakness Enumeration (CWE) types, such as CWE-325 Missing cryptographic step and CWE-444 HTTP request smuggling. (4) Codex still fixes 8.3 transformed vulnerabilities, outperforming all the other LLMs and APR models on transformed vulnerabilities. The results call for innovations to enhance automated Java vulnerability repair such as creating larger vulnerability repair training data, tuning LLMs with such data, and applying code simplification transformation to facilitate vulnerability repair.
['Sameena Shah', 'Petr Babkin', 'Lin Tan', 'Jordan Davis', 'Thibaud Lutellier', 'Hung Viet Pham', 'Nan Jiang', 'Yi Wu']
2023-05-29
null
null
null
null
['program-repair', 'program-repair']
['computer-code', 'reasoning']
[-2.55161911e-01 4.48996164e-02 -4.22979504e-01 5.84078440e-03 -1.23337567e+00 -1.13703871e+00 1.52929798e-01 3.62950414e-01 1.45469323e-01 1.89017966e-01 1.71952188e-01 -1.48824263e+00 1.04847923e-01 -7.74116099e-01 -1.03583598e+00 2.12181926e-01 -5.85386634e-01 -4.35563564e-01 5.33508122e-01 -5.09636343e-01 6.64618015e-01 2.58285422e-02 -1.17981946e+00 7.02164054e-01 8.41587961e-01 2.02626050e-01 -2.59309381e-01 1.13160574e+00 -2.99244523e-01 9.46791708e-01 -8.74555051e-01 -6.40759110e-01 3.18728805e-01 1.74616054e-01 -1.30598021e+00 -1.05077660e+00 4.37995434e-01 -4.51873630e-01 2.50709295e-01 1.04702830e+00 1.52952582e-01 -5.58856130e-01 1.24186940e-01 -1.39444351e+00 -7.46153533e-01 1.09897625e+00 -9.54722404e-01 1.79317564e-01 5.97177804e-01 3.91174078e-01 7.15334475e-01 -3.84734064e-01 4.74749088e-01 1.25703406e+00 1.33907127e+00 8.91071022e-01 -1.25895238e+00 -6.53843462e-01 -1.87560260e-01 -4.28956479e-01 -1.24967277e+00 -1.00097887e-01 2.86719203e-01 -7.36433148e-01 2.16192389e+00 4.21909779e-01 -9.66559872e-02 1.16202819e+00 8.37952733e-01 -8.45560953e-02 9.99147356e-01 -7.33683884e-01 7.14356899e-02 2.16628686e-01 7.49247313e-01 7.77274132e-01 4.34853315e-01 2.60546595e-01 4.00262684e-01 -1.14545441e+00 -6.51421100e-02 -1.82840317e-01 -4.65921387e-02 4.82231289e-01 -6.86672390e-01 1.14674008e+00 2.47563571e-01 3.69194090e-01 1.42206132e-01 5.56208432e-01 1.03605545e+00 7.29127109e-01 -1.24144077e-01 8.12142491e-01 -1.10047209e+00 -4.45851147e-01 -5.64660847e-01 3.50297064e-01 1.11017680e+00 8.28241527e-01 1.00118089e+00 4.08942193e-01 3.57754260e-01 5.52632034e-01 3.08954597e-01 3.73981088e-01 3.93109202e-01 -7.87542403e-01 9.15900409e-01 8.53741825e-01 -2.57207006e-01 -9.56613898e-01 -4.35447752e-01 5.05686402e-02 9.94398445e-02 7.55155921e-01 7.23089799e-02 -2.67618507e-01 -7.41754830e-01 1.71500111e+00 -1.88673869e-01 -2.79253930e-01 2.49399886e-01 2.99254116e-02 6.42135918e-01 6.38326883e-01 3.56384754e-01 3.72461915e-01 1.13423622e+00 -5.23549855e-01 6.47053272e-02 -1.97965249e-01 1.57294607e+00 -8.44756782e-01 1.51316130e+00 3.76784414e-01 -7.95517981e-01 -4.09804016e-01 -1.11547911e+00 1.94484279e-01 -7.25075424e-01 -3.02334994e-01 8.22914898e-01 1.49530816e+00 -1.50264418e+00 2.59095848e-01 -8.11991692e-01 -2.25974903e-01 -4.28103395e-02 4.86117542e-01 -4.40133750e-01 1.63344339e-01 -9.92728412e-01 7.06640840e-01 4.68604833e-01 -5.23494542e-01 -1.24383080e+00 -1.14307582e+00 -1.13027811e+00 1.76615924e-01 3.94885302e-01 -5.12803383e-02 1.34414196e+00 -8.58207226e-01 -8.06100607e-01 6.32131517e-01 2.90959686e-01 -2.39808872e-01 -2.74683326e-01 -2.49848917e-01 -6.20463967e-01 -3.05512756e-01 9.55591351e-02 2.58143067e-01 3.80180418e-01 -1.44192791e+00 -4.25840735e-01 2.93818593e-01 7.58753836e-01 -1.09884906e+00 -5.13396382e-01 7.28328049e-01 1.71728522e-01 -6.02142513e-01 -8.13080966e-01 -8.67519617e-01 -1.34615570e-01 -7.68176436e-01 -5.69384992e-01 7.39266798e-02 6.97603405e-01 -1.29749823e+00 1.69116282e+00 -2.06005096e+00 -2.06148088e-01 5.17646015e-01 2.50863850e-01 7.47884810e-01 -8.29310000e-01 7.35181749e-01 -6.44438684e-01 9.59479511e-01 -5.45457840e-01 1.69599727e-01 1.53195620e-01 3.85507271e-02 -8.41815412e-01 -2.07950491e-02 1.88135579e-01 9.45817947e-01 -5.17857790e-01 -1.11976720e-01 -2.04626113e-01 2.53058434e-01 -1.26409626e+00 -1.86518990e-02 -5.76062381e-01 -4.48821396e-01 -1.65481046e-01 1.01358998e+00 7.32243896e-01 2.44643956e-01 -3.17825973e-02 3.17548990e-01 -4.11049515e-01 3.25981438e-01 -8.26889336e-01 1.32765305e+00 -8.69532824e-01 3.64290059e-01 -4.85383756e-02 -1.96367010e-01 8.32047164e-01 3.37267876e-01 -1.46152094e-01 -6.17292941e-01 -2.62315542e-01 3.80745113e-01 1.04655281e-01 -8.14501166e-01 3.37021440e-01 7.12123513e-01 -7.61896551e-01 9.70686853e-01 -2.55540684e-02 1.23387568e-01 2.21889801e-02 4.39581811e-01 1.93423593e+00 7.76598305e-02 2.65421003e-01 -3.48889560e-01 7.21052289e-01 8.29077065e-02 5.75923383e-01 8.12822163e-01 1.33246332e-01 9.50595737e-02 1.17436528e+00 -5.89745045e-01 -1.12659371e+00 -9.02096033e-01 2.04614624e-01 9.96945322e-01 -6.09809816e-01 -1.10401285e+00 -1.19922185e+00 -1.24494350e+00 -1.35515898e-01 1.04867864e+00 -6.69356942e-01 -6.02166176e-01 -9.75508392e-01 -6.56761587e-01 1.40849626e+00 7.65983939e-01 2.23693058e-01 -1.09410942e+00 -5.62306821e-01 1.05034627e-01 -4.24136892e-02 -4.28489268e-01 -5.28430045e-01 1.57227859e-01 -3.36154521e-01 -1.36890316e+00 3.09753388e-01 -6.76693380e-01 4.46767062e-01 2.54452769e-02 1.13660800e+00 9.96871293e-01 -7.20297456e-01 6.21349812e-01 -6.33792520e-01 -8.44593197e-02 -1.29411137e+00 1.00371540e-01 -3.27651232e-01 -9.56914604e-01 4.54855114e-01 -4.65835571e-01 1.72812372e-01 7.29175135e-02 -1.01141286e+00 -8.60356033e-01 3.71451765e-01 7.05367982e-01 -7.34055415e-02 7.73195550e-02 5.65355599e-01 -9.51095939e-01 8.36159229e-01 -7.57169724e-01 -1.01538694e+00 5.07029951e-01 -6.83876276e-01 5.55225275e-02 7.49820054e-01 -4.67089444e-01 -1.24934876e+00 -4.52234358e-01 -8.62651885e-01 4.65056151e-02 6.61142468e-02 8.38706493e-01 -4.45741974e-02 -7.68535316e-01 1.39566064e+00 -3.10664445e-01 -9.27661359e-02 -4.28344458e-01 1.07095860e-01 4.76918370e-01 4.32975590e-01 -1.21422911e+00 1.17996156e+00 -3.23215991e-01 -4.35122728e-01 -2.64435738e-01 1.49907202e-01 2.61651188e-01 -1.85791627e-01 3.81211102e-01 6.78486824e-01 -4.33038861e-01 -8.41724038e-01 3.94432485e-01 -1.18411446e+00 -8.92535806e-01 9.37848836e-02 -2.69613683e-01 -6.08306192e-02 7.70117402e-01 -1.04316914e+00 -5.73272884e-01 -7.33094335e-01 -1.60627115e+00 8.54966879e-01 -7.88146853e-02 -6.66658401e-01 -9.08279479e-01 6.83290720e-01 7.28136450e-02 7.35402346e-01 3.98241401e-01 1.92558360e+00 -5.94622195e-01 -4.60161358e-01 -2.39102870e-01 -9.16170701e-02 4.66986924e-01 -6.11575842e-02 6.92499757e-01 -6.93962336e-01 -4.43575472e-01 -2.34872818e-01 -4.59762067e-01 3.73291790e-01 -9.37643498e-02 8.62492502e-01 -5.31464577e-01 -3.63942981e-01 7.29699612e-01 1.75614488e+00 4.28957671e-01 9.45288837e-01 8.13166916e-01 7.19334185e-01 5.17090678e-01 2.44155169e-01 1.28943995e-02 3.99363726e-01 4.17850316e-01 6.41132534e-01 2.11860806e-01 -1.00577265e-01 -1.56432912e-02 9.77627277e-01 5.42554021e-01 2.62435049e-01 2.28773713e-01 -1.50564516e+00 4.39774901e-01 -1.37565005e+00 -7.78570235e-01 -5.00686586e-01 2.05184031e+00 9.74882126e-01 2.61169136e-01 1.40948802e-01 2.09019482e-01 1.46024480e-01 -3.35813582e-01 2.15664655e-02 -1.43859100e+00 2.80941814e-01 6.62257612e-01 3.72361243e-01 8.53876412e-01 -7.84107387e-01 9.56313252e-01 5.99780560e+00 5.25646567e-01 -9.96587574e-01 3.60287130e-01 7.78433532e-02 5.06258786e-01 -7.43181348e-01 6.42841995e-01 -9.58681285e-01 3.27186346e-01 1.38043141e+00 -8.93091112e-02 6.03260577e-01 1.12858737e+00 -5.25636554e-01 4.01495285e-02 -9.05134082e-01 1.10255294e-01 -1.85013935e-03 -1.38695550e+00 -1.68026581e-01 3.00348066e-02 6.53698504e-01 -1.14077806e-01 1.23036966e-01 1.02849877e+00 5.70539236e-01 -1.02959788e+00 7.64049649e-01 2.67167330e-01 7.02016473e-01 -9.38344419e-01 8.71715367e-01 2.75868867e-02 -1.23177147e+00 -7.81879187e-01 -1.07302450e-01 1.66522950e-01 -3.59045297e-01 9.82405618e-02 -8.17633748e-01 4.37202752e-01 1.22862661e+00 5.16017005e-02 -1.38568270e+00 7.39847004e-01 -1.84299052e-01 9.25871491e-01 1.23836912e-01 2.88722903e-01 1.17546566e-01 4.87740040e-01 3.27062994e-01 1.53682208e+00 2.21504733e-01 -3.85457665e-01 2.02314407e-02 1.23460865e+00 3.03145081e-01 -2.50483066e-01 -4.29068029e-01 -1.82567030e-01 5.32857955e-01 1.10612345e+00 -4.32751864e-01 -1.34983333e-02 -8.24600399e-01 1.51014000e-01 3.34436566e-01 3.91132146e-01 -8.65357757e-01 -9.03991103e-01 9.45027411e-01 -1.69850308e-02 -1.72773551e-03 -1.31530881e-01 -2.79465735e-01 -9.29325461e-01 2.39388570e-01 -1.58037114e+00 5.06432354e-01 -5.86052060e-01 -9.57708180e-01 8.81492376e-01 3.54527563e-01 -6.30462408e-01 -4.26904023e-01 -5.53835928e-01 -1.19106054e+00 1.04550254e+00 -1.12545192e+00 -1.30901408e+00 1.22194417e-01 4.02434140e-01 2.07976401e-01 -5.39813280e-01 1.12359416e+00 4.99256760e-01 -5.28434992e-01 1.31215322e+00 -2.95326740e-01 1.98030889e-01 4.84093815e-01 -1.19080293e+00 1.29790068e+00 1.30987561e+00 -6.69911265e-01 1.51569915e+00 3.87641758e-01 -1.14378858e+00 -1.46700919e+00 -1.13352966e+00 7.53418982e-01 -9.17748690e-01 8.89248312e-01 -7.05986679e-01 -1.54136920e+00 1.11513829e+00 1.74653515e-01 -3.16229165e-01 5.97065449e-01 -1.59706976e-02 -1.23329186e+00 1.47900671e-01 -1.41724122e+00 7.20098078e-01 5.02111614e-01 -8.45975697e-01 -5.13624489e-01 1.15829378e-01 1.28606641e+00 -9.33987871e-02 -1.09654105e+00 2.49457747e-01 4.61002320e-01 -9.35757637e-01 1.08546662e+00 -4.71616656e-01 3.69598478e-01 -2.69389510e-01 -3.73835742e-01 -9.41179872e-01 -2.92069286e-01 -7.63163865e-01 -8.61018151e-02 1.82969868e+00 5.65389276e-01 -1.10984468e+00 1.87386915e-01 6.85168624e-01 -3.80026698e-01 -4.26812351e-01 -5.16558647e-01 -6.94280028e-01 5.19655228e-01 -5.73746681e-01 8.93799365e-01 1.14211833e+00 1.98600397e-01 -4.45544153e-01 1.57667175e-02 3.62658143e-01 3.28718752e-01 -3.33681196e-01 7.85406113e-01 -8.02474499e-01 -7.54147530e-01 -5.64338386e-01 -1.29696488e-01 2.31932163e-01 4.64643806e-01 -9.61671293e-01 -3.12884778e-01 -9.18798745e-01 2.83898115e-01 -5.40190160e-01 1.97480291e-01 1.40301192e+00 -2.33349741e-01 -1.24913491e-01 9.69193410e-03 -1.39409631e-01 1.07892096e-01 -3.05363685e-01 -9.51734036e-02 -2.91775197e-01 -1.59707829e-01 -1.48824289e-01 -1.00009012e+00 6.23988926e-01 8.33743930e-01 -7.88382471e-01 -4.25724417e-01 -4.38640654e-01 8.35595012e-01 1.62872985e-01 4.93427098e-01 -8.47774982e-01 -2.35980883e-01 -2.96938065e-02 -3.97892036e-02 -3.64957899e-02 -5.07452786e-01 -4.64414567e-01 2.40397274e-01 9.15469110e-01 -2.15221629e-01 7.74273694e-01 1.05706382e+00 -1.12206481e-01 2.34752819e-01 -8.43886614e-01 4.44708019e-01 -1.38895944e-01 -9.47811902e-01 -1.29020050e-01 -6.55316174e-01 8.91184807e-02 1.16683030e+00 -1.90759167e-01 -1.12075245e+00 3.83144021e-01 -2.05995932e-01 -4.53597400e-03 8.42652202e-01 9.31920767e-01 5.17974257e-01 -8.91552329e-01 -3.47635776e-01 4.13259774e-01 2.34749705e-01 -7.01343715e-01 2.34109446e-01 3.24171960e-01 -8.76908064e-01 2.20473856e-01 -3.94450277e-01 -5.43936193e-02 -1.50051033e+00 1.17077148e+00 3.01938415e-01 -1.68047503e-01 -5.30445814e-01 7.87305057e-01 2.58340668e-02 -1.14647019e+00 2.30150647e-03 -4.25823629e-01 9.39567164e-02 -8.09846282e-01 8.28383505e-01 4.39436257e-01 1.87269211e-01 -3.40718061e-01 -6.57575488e-01 6.81617320e-01 -4.10393447e-01 1.72314167e-01 1.19376683e+00 6.35673821e-01 -7.78323889e-01 -2.91507065e-01 1.50206482e+00 5.03911078e-01 -5.85233986e-01 4.24927264e-01 3.84335905e-01 -3.73858094e-01 -4.90513116e-01 -1.14071357e+00 -9.25543427e-01 8.35068166e-01 5.13392389e-01 2.07460433e-01 1.10575783e+00 -9.39030126e-02 7.44187593e-01 3.93172979e-01 7.20557868e-01 -3.34768027e-01 8.12570676e-02 5.70659518e-01 7.66118467e-01 -6.75379515e-01 -2.02214688e-01 -9.13361534e-02 -1.04401544e-01 1.20720077e+00 1.05202901e+00 -2.41527911e-02 5.08704484e-01 1.09183979e+00 -2.38184053e-02 -1.31512582e-01 -8.64478946e-01 6.36844635e-01 -1.57681376e-01 8.92276585e-01 3.82508248e-01 -9.42600965e-02 -1.67928338e-01 7.80811965e-01 -4.51771319e-02 -3.94179732e-01 9.76526022e-01 1.44564259e+00 -1.62056774e-01 -1.65344954e+00 -8.23586404e-01 4.04651195e-01 -7.02341616e-01 -4.19924468e-01 -4.82750148e-01 1.03216720e+00 1.79710731e-01 1.06611466e+00 -6.48745894e-01 -8.80903900e-01 3.19546044e-01 -3.35251428e-02 2.34451964e-01 -8.73232603e-01 -1.62714231e+00 -4.21993971e-01 2.82440752e-01 -8.48428249e-01 7.32589662e-01 -3.89325738e-01 -1.23730743e+00 -6.00396395e-01 -1.88518777e-01 1.64500162e-01 4.69744265e-01 5.02717793e-01 5.34851789e-01 6.25504732e-01 2.34033152e-01 -6.20203376e-01 -7.57711768e-01 -4.69055802e-01 1.08288243e-01 1.28994986e-01 4.46908891e-01 -2.81447828e-01 -4.57301497e-01 1.37010977e-01]
[7.106083393096924, 7.781507968902588]
74888ab7-012c-4e24-93fc-ec4ebe4ceb77
spoken-language-understanding-for
2212.10728
null
https://arxiv.org/abs/2212.10728v1
https://arxiv.org/pdf/2212.10728v1.pdf
Spoken Language Understanding for Conversational AI: Recent Advances and Future Direction
When a human communicates with a machine using natural language on the web and online, how can it understand the human's intention and semantic context of their talk? This is an important AI task as it enables the machine to construct a sensible answer or perform a useful action for the human. Meaning is represented at the sentence level, identification of which is known as intent detection, and at the word level, a labelling task called slot filling. This dual-level joint task requires innovative thinking about natural language and deep learning network design, and as a result, many approaches and models have been proposed and applied. This tutorial will discuss how the joint task is set up and introduce Spoken Language Understanding/Natural Language Understanding (SLU/NLU) with Deep Learning techniques. We will cover the datasets, experiments and metrics used in the field. We will describe how the machine uses the latest NLP and Deep Learning techniques to address the joint task, including recurrent and attention-based Transformer networks and pre-trained models (e.g. BERT). We will then look in detail at a network that allows the two levels of the task, intent classification and slot filling, to interact to boost performance explicitly. We will do a code demonstration of a Python notebook for this model and attendees will have an opportunity to watch coding demo tasks on this joint NLU to further their understanding.
['Josiah Poon', 'Henry Weld', 'Siqu Long', 'Soyeon Caren Han']
2022-12-21
null
null
null
null
['spoken-language-understanding', 'intent-detection', 'intent-classification', 'slot-filling', 'spoken-language-understanding']
['natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'speech']
[ 4.26478118e-01 6.02728486e-01 -3.02964598e-01 -7.10988998e-01 -6.88048303e-01 -4.79971021e-01 5.97429335e-01 6.28624931e-02 -2.51638502e-01 4.99969751e-01 5.99871755e-01 -7.69990861e-01 2.36914024e-01 -8.14521253e-01 -3.58551919e-01 -1.52755797e-01 1.59891888e-01 7.04804420e-01 -3.23847570e-02 -5.28764904e-01 2.93279171e-01 -9.95626375e-02 -1.52387702e+00 6.86128318e-01 3.94604534e-01 1.16220927e+00 3.31253052e-01 8.39066565e-01 -9.21454310e-01 1.40204358e+00 -6.97899401e-01 -3.21465939e-01 -1.92658603e-01 -4.29412663e-01 -1.63983667e+00 -1.19798705e-01 2.02400447e-03 -3.41719985e-01 6.46159127e-02 5.49292624e-01 3.41168910e-01 2.00858027e-01 2.47034654e-01 -1.20359445e+00 -2.52554387e-01 8.23815107e-01 1.37435213e-01 1.07130341e-01 9.63458955e-01 -2.83645689e-02 1.18409348e+00 -6.79730833e-01 5.18027663e-01 1.53219199e+00 6.80236995e-01 5.96906364e-01 -8.93350720e-01 -4.88414198e-01 7.94175938e-02 3.61984074e-01 -8.73404980e-01 -4.49046761e-01 6.17702782e-01 -3.25213313e-01 1.49858809e+00 1.75491765e-01 4.74389732e-01 1.20414197e+00 1.45000279e-01 1.26908565e+00 8.29949677e-01 -6.98272824e-01 1.73633203e-01 4.22867209e-01 6.79862499e-01 7.02660322e-01 -6.85859680e-01 -1.54858246e-01 -6.80284798e-01 -6.34392649e-02 4.17328507e-01 5.02504148e-02 -6.40702024e-02 -7.77040329e-03 -1.19581628e+00 1.16386020e+00 4.66025740e-01 5.94258010e-01 -3.05550843e-01 7.66346678e-02 7.37169266e-01 4.63302344e-01 4.62473065e-01 5.20976067e-01 -6.89487040e-01 -5.34057736e-01 -7.02169359e-01 5.19066080e-02 1.42098641e+00 7.03096330e-01 7.25069523e-01 -2.10452631e-01 -2.93310851e-01 1.16942263e+00 5.07638156e-01 -3.87511365e-02 5.42021036e-01 -9.54455853e-01 3.78336638e-01 7.50786483e-01 -1.21941052e-01 -8.67821455e-01 -5.57185769e-01 -1.32801309e-01 -5.93847334e-01 -5.57565270e-03 1.58824638e-01 -3.40420753e-01 -9.40117717e-01 1.57972181e+00 9.68864039e-02 -7.60714114e-02 2.57628202e-01 7.10140407e-01 1.27375746e+00 9.40896869e-01 4.79142100e-01 1.29747987e-01 2.13080978e+00 -1.09792948e+00 -1.10960722e+00 -7.86727548e-01 9.10721958e-01 -6.13161683e-01 1.11032724e+00 1.95417732e-01 -9.05164242e-01 -5.67160666e-01 -1.06040442e+00 -5.86259484e-01 -7.93807387e-01 -1.11760713e-01 8.65701735e-01 3.76226872e-01 -1.17293918e+00 2.64717281e-01 -5.25837421e-01 -8.55881572e-01 1.73541009e-01 3.05906713e-01 -2.20266089e-01 1.42155103e-02 -1.81142569e+00 1.07989407e+00 3.80502939e-01 -8.49271119e-02 -5.57920277e-01 -4.50218529e-01 -1.18744922e+00 1.26873821e-01 3.52020979e-01 -7.06592977e-01 1.94140708e+00 -9.33511913e-01 -1.61935854e+00 1.17977059e+00 -5.12405992e-01 -8.93331468e-01 5.47548616e-03 -2.48081774e-01 -2.23515987e-01 -9.76136047e-03 3.17141593e-01 1.04190314e+00 5.39525926e-01 -6.86257064e-01 -7.70550489e-01 -1.71944603e-01 5.14831901e-01 3.19753110e-01 -3.05132363e-02 2.56273419e-01 -8.15851167e-02 -3.59853148e-01 -4.03757617e-02 -4.71321791e-01 8.05708617e-02 -1.46994725e-01 -2.26879850e-01 -8.94912899e-01 9.26551521e-01 -7.81309187e-01 1.11885607e+00 -2.00901699e+00 -2.49816567e-01 -2.58959621e-01 1.82617575e-01 2.18924835e-01 3.12622562e-02 8.75855744e-01 -2.96454608e-01 1.84589401e-01 -1.82821944e-01 -4.98244077e-01 1.89008519e-01 2.80749738e-01 -6.15647495e-01 -1.47298530e-01 9.91029516e-02 9.98087108e-01 -8.05252910e-01 -2.28661001e-01 3.20469737e-01 3.99099976e-01 -4.99409169e-01 5.99445760e-01 -5.49371779e-01 1.67212456e-01 -3.69753152e-01 4.00452852e-01 1.63558766e-01 -4.76145744e-01 1.65144309e-01 -7.80161917e-02 -1.09667733e-01 1.17230809e+00 -7.84969866e-01 1.76070118e+00 -9.71688867e-01 9.92577791e-01 2.88174272e-01 -1.35347438e+00 8.47258747e-01 7.06042528e-01 1.54483784e-02 -7.47814059e-01 1.34660557e-01 1.17560916e-01 -2.51312762e-01 -5.43274283e-01 3.58555287e-01 -2.80906737e-01 -2.77490288e-01 8.81960988e-01 1.79673493e-01 -1.44350812e-01 1.02368463e-02 3.39140683e-01 1.29118323e+00 -1.56510636e-01 6.12666190e-01 -1.45428926e-01 5.96072853e-01 9.11418870e-02 5.74801601e-02 8.46855462e-01 -2.46910945e-01 2.31006086e-01 6.19304180e-01 -8.82627189e-01 -7.51255929e-01 -5.58934033e-01 -2.91220583e-02 1.66513157e+00 -1.60660461e-01 -3.84506017e-01 -7.06663489e-01 -6.67636395e-01 -4.95518506e-01 1.13745296e+00 -5.50946712e-01 -6.59738928e-02 -3.38900983e-01 -9.30789635e-02 3.01282018e-01 4.57049698e-01 9.59819198e-01 -1.71245062e+00 -6.17535889e-01 3.02667975e-01 -6.14071548e-01 -1.22922885e+00 -1.19395263e-01 5.82962275e-01 -4.19119030e-01 -7.56103873e-01 -3.70509744e-01 -1.12533987e+00 1.63444817e-01 7.33848214e-02 1.43069530e+00 2.01447457e-01 -1.75675556e-01 5.33079684e-01 -5.73953152e-01 -4.82569873e-01 -4.83188033e-01 2.23199561e-01 -2.06677333e-01 -1.35882422e-01 9.87729430e-01 -5.22990108e-01 -3.46747667e-01 2.23958224e-01 -9.16043401e-01 5.15808761e-01 4.16846186e-01 7.26050735e-01 -1.34462088e-01 -1.14025280e-01 3.76004726e-01 -7.52066135e-01 8.96811724e-01 -5.28622210e-01 -1.26084745e-01 2.87892997e-01 -5.93707711e-02 2.33349383e-01 2.71758914e-01 1.02645643e-01 -9.43968952e-01 4.83449660e-02 -7.85793185e-01 2.59312898e-01 -4.68411505e-01 5.47621727e-01 -8.54021460e-02 2.54209787e-01 5.29723227e-01 1.91717118e-01 2.71407943e-02 -3.35179359e-01 3.65433067e-01 1.25136185e+00 2.23510578e-01 -2.52268255e-01 1.86277509e-01 2.06043258e-01 -6.82244658e-01 -1.04756141e+00 -1.26310992e+00 -6.65596485e-01 -4.70246881e-01 -9.78478491e-02 1.16611707e+00 -6.98532939e-01 -9.57346380e-01 2.53039688e-01 -1.60972989e+00 -5.89278758e-01 -2.86604792e-01 9.69517976e-02 -6.60509646e-01 1.04889460e-01 -6.90476418e-01 -9.13581252e-01 -5.76196373e-01 -1.14471507e+00 1.14604855e+00 1.85526699e-01 -8.17645848e-01 -1.27797186e+00 -1.22547299e-01 8.24952483e-01 6.15556180e-01 -3.63142431e-01 9.06942129e-01 -1.20291293e+00 -4.22718614e-01 -1.72404632e-01 -4.33978707e-01 2.18546748e-01 5.25729358e-02 -7.28476465e-01 -1.26420605e+00 2.48453990e-01 3.26375663e-01 -7.36156821e-01 7.60556340e-01 2.90969670e-01 1.09154308e+00 -5.05910158e-01 -3.23154181e-01 1.49016470e-01 8.03464592e-01 2.92754441e-01 6.43445253e-01 3.34012836e-01 2.08764553e-01 9.45047021e-01 3.73492658e-01 1.51267618e-01 6.14638448e-01 7.08864033e-01 1.86910585e-01 -9.22559053e-02 -1.17532229e-02 -3.48788619e-01 4.26122099e-01 6.02562010e-01 5.78581989e-01 -2.36863926e-01 -1.05207658e+00 4.11304712e-01 -1.91257048e+00 -7.99737334e-01 1.27891287e-01 1.66565359e+00 8.77023995e-01 2.11505264e-01 -9.38553214e-02 2.27230489e-01 6.12365723e-01 3.60511094e-01 -2.45295584e-01 -1.10621631e+00 3.00357521e-01 1.36602208e-01 -2.46684570e-02 9.51165318e-01 -1.21125972e+00 1.26270759e+00 6.65012455e+00 7.54245520e-01 -9.20298398e-01 1.04643323e-01 8.26305091e-01 5.36983490e-01 -2.25403681e-01 -3.63011360e-02 -9.58090305e-01 6.61368743e-02 1.26961637e+00 1.23901747e-01 5.30190945e-01 9.07166541e-01 1.43852785e-01 -3.11626822e-01 -1.27389407e+00 9.33581948e-01 1.21284403e-01 -1.35136473e+00 -8.33938196e-02 -2.50511646e-01 -2.62164790e-02 6.01559654e-02 -2.66046822e-01 7.27399170e-01 4.91280168e-01 -1.12273908e+00 3.38873655e-01 3.21204752e-01 4.92335439e-01 -2.67962098e-01 8.11733425e-01 5.47962129e-01 -1.06244445e+00 -1.61064059e-01 2.30361167e-02 -5.98211050e-01 4.07923937e-01 4.96259451e-01 -1.23080504e+00 8.29483494e-02 5.34074485e-01 6.26618862e-01 -3.29383984e-02 4.47428912e-01 -4.47809458e-01 4.50258642e-01 -4.87804949e-01 -4.32858169e-01 5.14096379e-01 1.16928868e-01 4.16427940e-01 1.33456874e+00 -3.92916910e-02 2.69408971e-01 1.65390968e-01 1.00125098e+00 4.27640887e-04 3.81120406e-02 -6.42454624e-01 -3.27828944e-01 1.47685826e-01 1.20067656e+00 -6.56514406e-01 -5.02657175e-01 -4.72029269e-01 1.08863509e+00 2.38512039e-01 2.82155454e-01 -4.53685254e-01 -4.90641922e-01 7.75457680e-01 8.27440619e-02 -3.98049690e-02 -9.24314484e-02 -3.00259888e-01 -9.50803041e-01 -1.85659081e-01 -9.08818185e-01 4.38952804e-01 -1.02192771e+00 -1.04463553e+00 6.60949469e-01 -8.02052245e-02 -7.03790903e-01 -7.72795558e-01 -7.77424455e-01 -8.00722718e-01 8.25079560e-01 -1.38477743e+00 -9.17435944e-01 -1.56047747e-01 3.14253747e-01 1.14410162e+00 -2.44893413e-02 1.13272595e+00 1.31148443e-01 -2.62605608e-01 1.29650027e-01 -5.91773450e-01 4.30791646e-01 3.27556521e-01 -1.21302998e+00 7.28330374e-01 2.58825630e-01 2.41598457e-01 6.99600518e-01 7.90140033e-01 -4.41593468e-01 -1.08795142e+00 -5.24367213e-01 1.54832733e+00 -4.50782597e-01 7.36512542e-01 -7.25475132e-01 -6.70271337e-01 8.95090282e-01 5.92055857e-01 -4.87360477e-01 9.13366079e-01 2.52684772e-01 -8.17241222e-02 2.12306082e-01 -1.08131063e+00 5.11724472e-01 8.25324476e-01 -9.16386127e-01 -1.04343784e+00 7.09742904e-01 1.19007242e+00 -2.91981936e-01 -3.72394562e-01 2.38125160e-01 5.61632514e-01 -1.13680685e+00 7.96207666e-01 -6.04363918e-01 4.55933005e-01 6.34156242e-02 -2.42104791e-02 -1.02971101e+00 -4.14630286e-02 -5.98745346e-01 7.32659101e-02 1.14711165e+00 5.74898243e-01 -7.05889642e-01 5.93302310e-01 8.90866280e-01 -4.61986884e-02 -6.08262777e-01 -8.31989586e-01 -4.30985950e-02 -2.51773238e-01 -1.05205488e+00 3.35253000e-01 7.66548812e-01 5.04628479e-01 9.44436550e-01 -2.21397430e-01 -3.66750807e-01 8.08361843e-02 8.78239423e-02 5.92767656e-01 -1.14321041e+00 -4.59553450e-02 -3.44580054e-01 -1.60387859e-01 -1.66690207e+00 2.15819046e-01 -8.86424899e-01 1.24332167e-01 -1.92923641e+00 -1.33132935e-01 -9.68806520e-02 2.21640557e-01 8.41948867e-01 2.36710578e-01 -6.91910312e-02 6.00752421e-02 1.00938931e-01 -7.36843884e-01 4.79161024e-01 7.48578072e-01 -2.37397686e-01 -1.65823817e-01 1.84494600e-01 -8.63185108e-01 8.38867784e-01 8.18141758e-01 -1.78827435e-01 -2.77702302e-01 -2.59322166e-01 3.86674941e-01 1.58173844e-01 3.15245062e-01 -1.06074774e+00 4.26027060e-01 2.37840325e-01 8.95804688e-02 -5.43184042e-01 5.32132506e-01 -8.60893011e-01 -4.68018442e-01 4.05519396e-01 -8.65807414e-01 -3.87463868e-01 2.83236474e-01 1.34103671e-01 -4.07081693e-01 -3.21062982e-01 4.43026692e-01 -3.56387168e-01 -9.83501911e-01 -1.14467949e-01 -8.66928697e-01 -1.74577422e-02 7.15540707e-01 -9.31316391e-02 -5.82814217e-02 -1.17611659e+00 -9.33405399e-01 5.57626128e-01 -1.98242038e-01 7.95798481e-01 5.36507785e-01 -1.04464579e+00 -3.66825104e-01 4.13896620e-01 6.53833523e-02 -2.50568032e-01 1.29046008e-01 6.12676382e-01 -4.66447294e-01 1.05498767e+00 1.46383405e-01 -4.81934369e-01 -1.07426381e+00 2.82694668e-01 5.65528214e-01 -5.46748340e-01 -4.79254007e-01 1.06299949e+00 3.77214223e-01 -9.65680063e-01 5.72311878e-01 -4.22243416e-01 -5.64276338e-01 2.11939543e-01 8.82204354e-01 -2.02153131e-01 9.41120535e-02 -3.59476030e-01 -4.30813760e-01 2.58678377e-01 -2.15328131e-02 -2.56835282e-01 1.09257483e+00 -3.89099121e-01 -2.65476495e-01 6.24854386e-01 1.52239776e+00 -7.20497489e-01 -6.38061821e-01 -3.63111168e-01 1.73405260e-01 1.71180889e-01 6.86666444e-02 -1.03358603e+00 -4.81274188e-01 1.33980346e+00 4.21680957e-01 7.80998707e-01 7.99237132e-01 3.06672513e-01 1.14220536e+00 7.07241237e-01 1.94061115e-01 -1.15751231e+00 1.40581369e-01 1.12282526e+00 9.82224882e-01 -1.40422034e+00 -4.57329035e-01 -2.89199919e-01 -7.18350410e-01 1.29093957e+00 4.46544975e-01 2.17409819e-01 7.81528950e-01 2.69653857e-01 3.69634688e-01 -4.41075802e-01 -8.53267789e-01 -2.89371789e-01 1.13118999e-01 4.00734037e-01 8.15934896e-01 -1.30797938e-01 -6.94338754e-02 5.21942735e-01 -5.43246865e-01 1.22535735e-01 5.50429299e-02 9.68646288e-01 -6.76019788e-01 -1.18368399e+00 -2.86513418e-01 4.33541447e-01 -4.10587192e-01 -2.44833037e-01 -7.56433666e-01 3.57835829e-01 -1.59869328e-01 1.36961734e+00 2.17092708e-01 -4.52149540e-01 3.87852937e-02 7.23055840e-01 -2.48937249e-01 -9.95727122e-01 -6.94907904e-01 -3.88013661e-01 5.24312139e-01 -7.04430163e-01 -3.42703670e-01 -1.48356676e-01 -1.31914830e+00 -2.07919925e-01 1.24129979e-02 3.88280958e-01 7.23491907e-01 1.41146100e+00 2.52529621e-01 5.08238971e-01 2.94667602e-01 -8.52163494e-01 -1.02422759e-01 -1.14577341e+00 -2.03323498e-01 1.53288636e-02 4.09574777e-01 -3.49697471e-01 -5.55661500e-01 -1.23815186e-01]
[12.535622596740723, 7.459218502044678]
190d7bd8-7622-4a12-94b9-38050a497597
informing-the-design-of-spoken-conversational
null
null
https://openreview.net/forum?id=rJgGxq1_z4
https://openreview.net/pdf?id=rJgGxq1_z4
Informing the Design of Spoken Conversational Search
We conducted a laboratory-based observational study where pairs of people performed search tasks communicating verbally. Examination of the discourse allowed commonly used interactions to be identified for Spoken Conversational Search (SCS). We compared the interactions to existing models of search behaviour. We find that SCS is more complex and interactive than traditional search. This work enhances our understanding of different search behaviours and proposes research opportunities for an audio-only search system. Future work will focus on creating models of search behaviour for SCS and evaluating these against actual SCS systems.
['Mark Sanderson', 'Hideo Joho', 'Lawrence Cavedon', 'Damiano Spina', 'Johanne R. Trippas']
2019-01-12
null
null
null
null
['conversational-search']
['natural-language-processing']
[ 1.56351298e-01 2.64253736e-01 -1.72831044e-01 -4.32696998e-01 -5.62246561e-01 -6.30434275e-01 9.96288419e-01 2.04826161e-01 -5.69472253e-01 4.86052603e-01 8.65814447e-01 -5.85297942e-01 -5.27161419e-01 -1.25351325e-02 2.34465584e-01 -3.93036529e-02 -2.91096449e-01 6.99212730e-01 3.64382893e-01 -2.81633317e-01 7.31061518e-01 3.80276561e-01 -1.68145835e+00 4.02908742e-01 4.36458737e-01 3.15175682e-01 5.06824374e-01 1.23802483e+00 -5.11566214e-02 8.49227786e-01 -1.14903414e+00 1.59469947e-01 -2.72066116e-01 -4.97797638e-01 -1.55956948e+00 -1.79705799e-01 2.10375011e-01 -6.53519988e-01 -3.29133302e-01 1.65543437e-01 9.12765741e-01 6.18755281e-01 3.84611785e-01 -1.28609693e+00 -4.27654207e-01 4.07051653e-01 5.21951795e-01 6.87800109e-01 1.72902048e+00 3.18132281e-01 6.06329381e-01 -2.17942834e-01 6.23442650e-01 1.62302136e+00 5.63757181e-01 6.78821921e-01 -1.19584572e+00 -8.59003663e-01 -2.62758255e-01 3.60414565e-01 -1.32496989e+00 -9.79427814e-01 2.80126542e-01 -3.13627571e-01 1.58529317e+00 8.96055818e-01 9.82996643e-01 1.24899852e+00 -1.98896810e-01 4.73160654e-01 1.23431230e+00 -7.51762509e-01 1.18623391e-01 6.27771139e-01 4.20972496e-01 4.63134557e-01 -5.04943073e-01 1.71096444e-01 -1.19509113e+00 -9.26674664e-01 5.26519418e-01 -6.73452139e-01 -5.22766054e-01 3.60058725e-01 -1.02395856e+00 4.91659403e-01 -2.22448856e-01 6.68465376e-01 -3.14890265e-01 6.09829277e-02 4.35119927e-01 5.43500125e-01 3.89037371e-01 8.97770882e-01 -1.29735842e-01 -1.33871055e+00 -6.75589383e-01 4.18290347e-01 1.58663845e+00 6.33749604e-01 -3.12022660e-02 -5.15219271e-01 -4.03377742e-01 1.15151763e+00 5.08579195e-01 2.16744423e-01 4.88994181e-01 -1.24549866e+00 -1.16143785e-01 4.27606255e-01 3.68656099e-01 -1.05822957e+00 -5.63598692e-01 3.85101080e-01 3.98880482e-01 -2.50307202e-01 1.78267106e-01 -5.08417264e-02 -2.56485105e-01 1.15981901e+00 -7.00595975e-02 -1.73119009e-01 -1.11331016e-01 7.85460830e-01 1.16529894e+00 3.02521557e-01 1.14081293e-01 -3.69284898e-01 1.25081587e+00 -7.05067754e-01 -1.09759080e+00 -5.31723425e-02 8.44492078e-01 -9.46422458e-01 1.30728662e+00 3.10069740e-01 -1.19068921e+00 -8.54085833e-02 -5.80627918e-01 6.79450557e-02 -1.02531210e-01 -4.21795249e-01 6.31289363e-01 1.05235958e+00 -1.65624118e+00 1.91454947e-01 -6.55528545e-01 -1.06030142e+00 -4.71266985e-01 4.83463347e-01 -5.99812865e-02 7.08603680e-01 -1.46314919e+00 1.22330570e+00 -1.37330666e-01 -3.73793207e-02 -3.19525868e-01 -1.60424829e-01 -7.25355327e-01 -1.53608248e-01 2.27124453e-01 -4.19475973e-01 2.12816644e+00 -4.18128222e-01 -1.84508991e+00 8.14615488e-01 -5.50035179e-01 -7.32793957e-02 -7.04446137e-02 -2.49136657e-01 -5.00823319e-01 7.17832863e-01 1.02323137e-01 2.78581470e-01 5.25210937e-03 -1.31606925e+00 -2.92913467e-01 6.67254329e-02 1.95333943e-01 7.00809479e-01 -1.23920694e-01 9.70823407e-01 -1.27796754e-01 -1.70566872e-01 -3.37196916e-01 -7.44818985e-01 2.75275767e-01 -1.29754543e-01 -1.40721679e-01 -5.59173644e-01 6.11487031e-01 -6.32971466e-01 1.82088184e+00 -1.73901367e+00 -2.78757393e-01 3.16475987e-01 1.29529968e-01 2.73543626e-01 1.54714048e-01 1.43196130e+00 4.41102475e-01 5.15335441e-01 4.91716325e-01 -3.50386411e-01 3.73207241e-01 3.46343517e-02 2.55862512e-02 2.44127303e-01 -6.77806020e-01 7.84614146e-01 -9.88729656e-01 -8.34313750e-01 2.49266386e-01 2.18711182e-01 -2.91011691e-01 3.19021553e-01 1.52719766e-01 1.69083536e-01 -3.40186059e-01 5.02908111e-01 -1.39356300e-01 -1.28856584e-01 -5.78648746e-02 5.78159630e-01 -4.90574598e-01 8.48040879e-01 -6.67120874e-01 1.14340866e+00 -4.31824058e-01 9.54219103e-01 5.53505778e-01 -3.80357862e-01 5.37894964e-01 7.00964749e-01 3.88111942e-03 -6.94265902e-01 6.92107305e-02 2.00982675e-01 1.28124923e-01 -1.11481166e+00 3.96642357e-01 1.92428082e-01 5.96238188e-02 1.08251452e+00 -2.46746093e-01 -6.98691308e-01 -3.12737584e-01 3.12747151e-01 1.25150657e+00 -4.65751112e-01 2.53393799e-01 -3.54463786e-01 4.87214655e-01 1.40222535e-01 -4.93596703e-01 1.18630564e+00 -5.19831777e-01 1.37185499e-01 1.14728317e-01 -1.01822160e-01 -1.36879310e-01 -5.50738692e-01 1.45750225e-01 1.34405231e+00 3.10183823e-01 -8.68003249e-01 -9.43516433e-01 7.75833847e-03 -4.95564431e-01 1.03883564e+00 -1.12098917e-01 -1.93984896e-01 -1.84410244e-01 3.34371090e-01 1.13121796e+00 1.43717956e-02 3.52339774e-01 -1.55194950e+00 -1.36604476e+00 4.66965772e-02 -3.71524245e-01 -7.69739389e-01 -8.86220992e-01 -4.55822647e-02 -4.33918923e-01 -9.91250575e-01 -3.88378829e-01 -1.00555503e+00 -4.95162047e-02 4.13954288e-01 7.99645483e-01 7.64923155e-01 -3.81389052e-01 1.36160326e+00 -6.25891030e-01 -3.86513650e-01 -3.16245228e-01 -1.46354184e-01 1.52997077e-01 -7.74981797e-01 5.84159076e-01 -2.37336144e-01 -6.88933969e-01 6.49759114e-01 -5.09656966e-01 -1.52137056e-01 -5.91266006e-02 6.82763815e-01 -5.64117849e-01 -2.20387980e-01 2.48880193e-01 -1.77167013e-01 2.05298305e+00 -2.47047305e-01 1.84270710e-01 3.40027422e-01 -6.82676792e-01 -4.60375458e-01 -1.71602488e-01 -7.56503403e-01 -1.21765125e+00 -5.74186206e-01 1.51963294e-01 1.94541097e-01 -5.02877116e-01 5.28141677e-01 7.65579462e-01 -4.80039477e-01 6.28317773e-01 2.35583052e-01 4.72433805e-01 -1.81089610e-01 -3.68933558e-01 1.34411597e+00 1.63839892e-01 -6.07561767e-01 -5.58218174e-02 -3.37756984e-02 -7.90748298e-01 -1.58340132e+00 4.03515160e-01 -8.02055001e-01 -2.59265117e-02 -8.06448102e-01 6.30024731e-01 -3.32462281e-01 -1.47078347e+00 3.05728108e-01 -1.06719339e+00 -1.00069630e+00 2.95888901e-01 7.39695251e-01 -5.20634830e-01 4.04087424e-01 -6.26203597e-01 -1.82043684e+00 -2.65505075e-01 -1.15860176e+00 1.40148759e+00 3.84544522e-01 -1.66690505e+00 -9.92391467e-01 4.42204177e-01 7.80596197e-01 6.39687598e-01 -5.85048735e-01 5.80555558e-01 -1.12775266e+00 -5.97258005e-03 -2.24434901e-02 1.50862053e-01 -4.63966727e-01 1.61842421e-01 -1.60820074e-02 -7.12453425e-01 1.58238292e-01 2.28311524e-01 -6.51729941e-01 -9.69279334e-02 1.18440613e-01 6.36565208e-01 -6.25363350e-01 -7.05136657e-01 -2.95935869e-01 4.17782307e-01 1.06762540e+00 2.77647346e-01 5.25294423e-01 -2.28876397e-01 1.14221430e+00 4.01318341e-01 3.13441157e-01 4.57603812e-01 1.18457937e+00 -4.56279188e-01 4.01603192e-01 2.86842376e-01 -7.73014650e-02 2.51730561e-01 6.66056514e-01 9.01733786e-02 -5.80723464e-01 -1.28921306e+00 5.19598424e-01 -1.60594332e+00 -1.21786058e+00 7.11809844e-02 1.63448000e+00 9.06499565e-01 -2.51009822e-01 3.11739802e-01 8.90915021e-02 4.42910463e-01 1.90005936e-02 6.11169636e-02 -1.06659961e+00 5.73805571e-01 5.03800213e-01 -9.63858142e-02 1.08970177e+00 -4.35067654e-01 5.08640289e-01 8.32702732e+00 3.63547981e-01 -6.58248484e-01 -1.04995653e-01 2.20788553e-01 -3.17026138e-01 -2.87778497e-01 -1.83810398e-01 -2.40421012e-01 3.80124301e-01 1.32242227e+00 -3.70449483e-01 7.42113650e-01 1.85726359e-01 6.65944397e-01 -9.57603753e-01 -1.14803708e+00 9.09032702e-01 2.12777227e-01 -9.41595674e-01 -2.85978436e-01 -4.20201086e-02 -3.57810616e-01 -5.19903362e-01 -3.77076209e-01 3.41620147e-01 7.17524961e-02 -1.11253726e+00 3.87553751e-01 6.70383334e-01 4.08078998e-01 -5.63835017e-02 5.54208517e-01 6.93304837e-01 -7.53649175e-01 1.00371450e-01 7.62294292e-01 -6.34056747e-01 6.45841122e-01 -7.71230996e-01 -1.21705902e+00 -3.65351737e-02 1.00569761e+00 -2.61107355e-01 -3.71156931e-01 9.51240122e-01 4.71169055e-01 5.62633753e-01 -6.79401398e-01 -8.91874492e-01 1.77728042e-01 -9.91518274e-02 7.55949914e-01 1.29771864e+00 1.41698518e-03 6.27253711e-01 -1.67370518e-03 5.09551406e-01 7.55143583e-01 -3.65348458e-02 -8.29055905e-01 -3.91532332e-01 1.05772495e+00 5.05963504e-01 -8.40864956e-01 -1.22189879e-01 -1.33376405e-01 9.97000217e-01 -3.62079233e-01 3.74132693e-01 -1.55604690e-01 -2.44578540e-01 3.67267489e-01 1.44842684e-01 -4.96363729e-01 -1.75031230e-01 -3.51202823e-02 -2.13792592e-01 -1.34907350e-01 -1.28783751e+00 -8.08009952e-02 -9.94664311e-01 -9.13531899e-01 5.67887366e-01 9.88526642e-01 -4.26488042e-01 -8.34243894e-01 1.84161216e-01 -6.98600888e-01 1.15294480e+00 -3.15276384e-01 -3.08036625e-01 -4.13742125e-01 2.70903915e-01 9.91192520e-01 -5.62004671e-02 1.29104042e+00 -4.60750572e-02 -4.91710491e-02 4.22478765e-01 -4.15316641e-01 -4.60749090e-01 7.42690921e-01 -7.55146801e-01 4.09366414e-02 -3.05469126e-01 -2.03452855e-01 1.28305256e+00 9.50112402e-01 -8.06774080e-01 -1.28463614e+00 4.41490382e-01 1.22528088e+00 -4.11443532e-01 8.09682190e-01 -3.68716478e-01 -8.06034923e-01 3.55418533e-01 9.68799651e-01 -1.02577829e+00 9.61455286e-01 7.73518756e-02 3.73537183e-01 6.93854690e-01 -1.06523597e+00 9.06712294e-01 1.32682467e+00 -1.29513896e+00 -9.47437763e-01 4.76702988e-01 6.75783396e-01 -4.39793378e-01 -4.73180354e-01 1.35762105e-02 1.21253192e+00 -1.02092147e+00 9.53714252e-01 -5.73134422e-02 -1.77711412e-01 4.50152338e-01 1.76300660e-01 -1.29394448e+00 3.80763620e-01 -1.33213782e+00 2.29596645e-01 1.01274323e+00 2.80441225e-01 -1.05453038e+00 3.60986531e-01 1.47830141e+00 -3.69748250e-02 -6.47104502e-01 -1.07548308e+00 -6.42535865e-01 -5.81819534e-01 -1.89917922e-01 1.29499853e-01 6.20231986e-01 1.13646567e+00 3.09796393e-01 2.96493266e-02 -3.16186935e-01 -3.05963635e-01 -6.84897244e-01 5.43916225e-01 -1.25656676e+00 -2.53484577e-01 -9.47590530e-01 -1.87863231e-01 -9.31416392e-01 1.39588565e-01 -2.52811015e-01 2.77437329e-01 -1.60482121e+00 -1.33794010e-01 -1.33153647e-01 6.72318637e-01 1.65434808e-01 3.07296455e-01 -3.92399251e-01 1.32436767e-01 3.24160993e-01 -4.58161682e-01 4.33294237e-01 8.21128249e-01 1.44398659e-01 -6.97665215e-01 -4.91994619e-03 -6.93763673e-01 3.98119152e-01 1.03023291e+00 -1.44215703e-01 -8.37722063e-01 8.70454609e-02 8.79617222e-03 4.06657189e-01 2.75970876e-01 -6.76519692e-01 7.73428619e-01 -2.29552716e-01 -4.29290712e-01 -3.52296442e-01 7.49703288e-01 -6.07297957e-01 4.31550443e-01 3.59282017e-01 -9.18936074e-01 2.47433782e-01 5.33469081e-01 2.08182395e-01 -2.17433676e-01 -5.80371797e-01 -3.58031720e-01 -1.00413196e-01 -2.59122640e-01 -8.35711718e-01 -1.45097053e+00 -1.87635422e-01 1.02963924e+00 -8.09718251e-01 -4.09351170e-01 -1.36076641e+00 -6.27309144e-01 6.75442994e-01 2.81274825e-01 1.93179652e-01 6.87554717e-01 -7.88718462e-01 8.62668380e-02 -2.90583335e-02 -3.88466157e-02 -4.46549743e-01 -3.00012141e-01 8.81728768e-01 -8.63204956e-01 9.70374286e-01 2.19396561e-01 -1.59955144e-01 -2.05425429e+00 1.95976477e-02 2.25205183e-01 4.16106820e-01 -5.29604495e-01 1.02820933e+00 -3.33466321e-01 -2.87198156e-01 9.87669885e-01 -6.09929711e-02 -4.38819319e-01 7.66833648e-02 5.40984690e-01 7.87799001e-01 -4.15162712e-01 -5.81412137e-01 -7.53671050e-01 1.32533923e-01 1.79111585e-01 -9.98471916e-01 7.17840254e-01 -5.34669220e-01 -1.20812684e-01 9.60784912e-01 9.55253601e-01 2.20892131e-02 -2.34459996e-01 1.78619310e-01 1.19039178e-01 -3.65932077e-01 -5.61387390e-02 -1.05595005e+00 1.36900827e-01 1.28408492e-01 4.25809801e-01 8.72528493e-01 7.83971608e-01 4.34252501e-01 3.98170173e-01 8.78446639e-01 5.08279145e-01 -1.34565651e+00 2.07199499e-01 3.05633754e-01 1.36642885e+00 -8.38469386e-01 2.26271879e-02 -3.94695431e-01 -7.75405169e-01 9.53624785e-01 5.50073028e-01 3.99910539e-01 7.53704727e-01 2.09926590e-01 1.90074921e-01 -9.25117791e-01 -9.78156924e-01 1.83963701e-02 1.73764676e-01 6.62070990e-01 8.11411560e-01 -7.12527409e-02 -1.03319073e+00 2.94792056e-01 -4.62541282e-01 2.29415998e-01 3.41215253e-01 1.36273110e+00 -1.18178405e-01 -1.05884230e+00 -5.45711637e-01 6.18357539e-01 -2.33933270e-01 -7.30200410e-02 -1.47976816e+00 9.56041574e-01 -6.68429077e-01 1.96504211e+00 9.37216803e-02 -6.33690000e-01 4.77865934e-01 5.76356173e-01 1.80976942e-01 -6.58326924e-01 -1.34150863e+00 -3.17286998e-02 1.18447137e+00 -8.52597356e-01 -8.51989746e-01 -8.83369207e-01 -1.00230539e+00 -2.97188789e-01 -5.66915810e-01 7.35170901e-01 3.22730005e-01 8.20784092e-01 4.02672887e-01 1.49438381e-01 1.18140966e-01 -8.35442841e-01 -3.39805722e-01 -1.23464108e+00 -2.68652260e-01 1.55564342e-02 4.85247910e-01 -8.25729370e-01 -7.24971473e-01 -2.77206302e-01]
[12.291970252990723, 7.782953262329102]
1e825799-6dbf-4a77-a496-e24f9d950b02
image-difference-captioning-with-pre-training
2202.04298
null
https://arxiv.org/abs/2202.04298v1
https://arxiv.org/pdf/2202.04298v1.pdf
Image Difference Captioning with Pre-training and Contrastive Learning
The Image Difference Captioning (IDC) task aims to describe the visual differences between two similar images with natural language. The major challenges of this task lie in two aspects: 1) fine-grained visual differences that require learning stronger vision and language association and 2) high-cost of manual annotations that leads to limited supervised data. To address these challenges, we propose a new modeling framework following the pre-training-finetuning paradigm. Specifically, we design three self-supervised tasks and contrastive learning strategies to align visual differences and text descriptions at a fine-grained level. Moreover, we propose a data expansion strategy to utilize extra cross-task supervision information, such as data for fine-grained image classification, to alleviate the limitation of available supervised IDC data. Extensive experiments on two IDC benchmark datasets, CLEVR-Change and Birds-to-Words, demonstrate the effectiveness of the proposed modeling framework. The codes and models will be released at https://github.com/yaolinli/IDC.
['Qin Jin', 'Weiying Wang', 'Linli Yao']
2022-02-09
null
null
null
null
['fine-grained-image-classification']
['computer-vision']
[ 7.74428919e-02 -2.92232126e-01 -2.38674089e-01 -6.06155753e-01 -6.51961803e-01 -5.47784567e-01 6.46216691e-01 2.72866767e-02 -5.25550604e-01 5.19071162e-01 2.70642966e-01 -9.16341245e-02 2.40302578e-01 -3.33260983e-01 -6.33595824e-01 -4.03963834e-01 4.20361280e-01 1.69040114e-01 3.34518068e-02 -1.63391218e-01 3.57654363e-01 1.58027813e-01 -1.55031478e+00 5.90925872e-01 1.10599864e+00 1.00528526e+00 4.72595602e-01 3.59217554e-01 -3.85137618e-01 6.52288020e-01 -3.01538259e-01 -3.15947264e-01 2.42592648e-01 -4.36616302e-01 -8.13364744e-01 3.29681396e-01 7.11099505e-01 -4.02239710e-01 -1.31027311e-01 1.22764575e+00 4.15984958e-01 -4.17214585e-03 6.57044113e-01 -1.47175515e+00 -1.10932815e+00 3.63482177e-01 -8.59870434e-01 2.22886413e-01 1.48801893e-01 4.01872158e-01 8.94371271e-01 -1.07425046e+00 3.01490843e-01 1.44073653e+00 4.31944907e-01 7.44142473e-01 -1.08851326e+00 -7.13343978e-01 4.87525225e-01 4.14022714e-01 -1.53486025e+00 -4.53704774e-01 7.92361557e-01 -6.54391706e-01 6.24464393e-01 1.77084044e-01 2.46333972e-01 1.07824409e+00 -1.41957328e-01 7.32564330e-01 1.34449899e+00 -3.76211017e-01 -1.20151930e-01 2.76200473e-01 4.48449999e-02 8.40578377e-01 1.26590997e-01 7.00370371e-02 -3.23455036e-01 2.42130309e-01 6.18907154e-01 1.65316805e-01 -4.71373469e-01 -3.76116961e-01 -1.32699132e+00 5.98138988e-01 4.94953483e-01 3.14921439e-01 -6.04214072e-02 6.11115322e-02 5.57006121e-01 2.22265795e-01 5.84953725e-01 2.74572343e-01 -4.64523017e-01 1.34718031e-01 -6.80053592e-01 5.31580634e-02 3.00637186e-01 1.15708077e+00 9.36762989e-01 -1.61809772e-01 -3.52167100e-01 9.85418320e-01 4.64170724e-01 5.67252696e-01 7.28084743e-01 -7.53531456e-01 7.90089488e-01 5.62708914e-01 3.68745811e-02 -1.03117883e+00 -5.01642413e-02 -2.53018260e-01 -9.44497108e-01 6.28920197e-02 2.34336615e-01 1.42139062e-01 -9.32595670e-01 1.68240333e+00 1.60646603e-01 -5.91877475e-03 -6.25464274e-03 1.04016328e+00 1.09854853e+00 6.29153073e-01 2.84311175e-01 1.40814269e-02 1.43681860e+00 -1.41124892e+00 -7.26368427e-01 -3.66277903e-01 6.13364816e-01 -7.88674951e-01 1.63996017e+00 -8.17787200e-02 -8.86087954e-01 -9.60743606e-01 -9.34687018e-01 -3.87160778e-01 -4.88614202e-01 4.77048606e-01 2.76801139e-01 1.70335323e-01 -9.48725641e-01 1.91158921e-01 -4.11484987e-01 -3.35548699e-01 4.36774552e-01 -6.76864535e-02 -3.36744338e-01 -1.62496403e-01 -1.09058642e+00 6.83808804e-01 6.87809885e-01 8.84195864e-02 -8.28467131e-01 -6.54949665e-01 -1.02737844e+00 -2.29337392e-03 3.43917817e-01 -5.24933338e-01 1.14225566e+00 -1.29365778e+00 -1.11085737e+00 1.28681517e+00 -9.97579321e-02 -1.77421689e-01 6.07563317e-01 -8.17764476e-02 -3.88386160e-01 -4.90533449e-02 3.82057965e-01 1.05183828e+00 8.92461777e-01 -1.46236980e+00 -7.24300325e-01 -3.99118096e-01 1.98697314e-01 4.71487373e-01 -2.60028869e-01 -1.37673691e-01 -7.54037619e-01 -8.16829085e-01 -2.76610106e-01 -8.43548059e-01 -3.24527323e-02 1.75450936e-01 -3.42069149e-01 -3.60970050e-01 7.70619631e-01 -5.75738072e-01 1.03705549e+00 -2.28027081e+00 -3.34769338e-02 -3.27713609e-01 4.76238132e-01 3.68217170e-01 -5.77084363e-01 1.70936048e-01 -9.44575295e-02 1.95989460e-01 -1.36548832e-01 -4.26519126e-01 -8.85628909e-03 3.51517722e-02 -1.17181778e-01 2.23968685e-01 2.80468255e-01 9.19725478e-01 -8.57346714e-01 -8.15042019e-01 2.82044888e-01 1.98316306e-01 -3.63943934e-01 4.64780360e-01 -2.51676887e-01 6.09795988e-01 -3.88280421e-01 5.20443916e-01 9.73754883e-01 -4.26057577e-01 -1.49523586e-01 -7.11995542e-01 -2.73870587e-01 -2.41538435e-02 -9.22311544e-01 1.82783663e+00 -5.82441509e-01 4.34106767e-01 -6.14778139e-02 -1.03362727e+00 8.72786045e-01 -4.09483910e-02 -1.36865722e-02 -1.00151730e+00 1.22873984e-01 1.66041076e-01 -1.13544188e-01 -6.50094151e-01 3.43709618e-01 2.34406088e-02 -1.46502793e-01 2.99186766e-01 2.13810802e-02 -9.41153616e-02 3.55680525e-01 3.10030207e-02 4.05160069e-01 8.53005201e-02 4.61751699e-01 -3.61147076e-01 9.08581734e-01 -2.63737515e-02 7.58427560e-01 5.56351125e-01 -6.37249947e-01 6.50038958e-01 2.21075147e-01 -4.29384321e-01 -8.78573835e-01 -7.89716423e-01 6.44432530e-02 1.24550247e+00 5.43787956e-01 -4.35878694e-01 -7.42727280e-01 -9.15000141e-01 -3.73959579e-02 4.94891942e-01 -7.38851547e-01 -5.06737502e-04 -2.03299120e-01 -3.31155479e-01 3.64566833e-01 4.98747706e-01 8.58739614e-01 -9.61356163e-01 -2.68235743e-01 -3.83242846e-01 -4.24717993e-01 -1.27214897e+00 -1.11765480e+00 -1.38757676e-01 -5.34821630e-01 -1.00079906e+00 -7.36629725e-01 -1.17756069e+00 7.30605423e-01 6.21562362e-01 1.17412114e+00 1.24105439e-01 -3.17877412e-01 2.43849784e-01 -3.80770385e-01 -1.93296075e-01 -2.85223484e-01 -1.75491616e-01 -7.09923059e-02 2.79211104e-02 4.29083347e-01 -1.57471061e-01 -7.51637757e-01 3.77721310e-01 -8.77370119e-01 6.55707657e-01 7.97480643e-01 9.09638166e-01 8.42902660e-01 -1.08026750e-01 4.59135294e-01 -8.30700338e-01 6.42847657e-01 -3.18253070e-01 -5.28359115e-01 6.21479511e-01 -7.06055284e-01 1.95948303e-01 7.38180876e-01 -4.43236202e-01 -1.22721946e+00 1.90511290e-02 7.92813599e-02 -5.44512868e-01 -5.08841634e-01 2.89318860e-01 -3.95570338e-01 2.31911410e-02 3.67163718e-01 3.91266197e-01 -7.88057297e-02 -5.36362290e-01 6.52966380e-01 8.78245056e-01 6.92993999e-01 -6.15684986e-01 7.30780363e-01 3.31472307e-01 -5.54602027e-01 -4.78809237e-01 -1.12808096e+00 -5.05370617e-01 -7.97043979e-01 -1.21420696e-02 1.15614820e+00 -1.22464645e+00 -4.89581823e-01 5.26590884e-01 -1.24345171e+00 -2.97888219e-01 2.54164333e-03 4.37673897e-01 -4.40731972e-01 4.59884495e-01 -4.60701704e-01 -2.83004105e-01 -4.39198613e-01 -1.18861485e+00 1.08507073e+00 4.78783369e-01 1.89905226e-01 -1.01900613e+00 1.12059884e-01 6.79791152e-01 2.98388124e-01 -4.34000697e-03 8.68096471e-01 -4.47304636e-01 -4.15679008e-01 2.57859051e-01 -7.51319230e-01 5.98369181e-01 4.26254570e-01 -1.39652908e-01 -7.29325056e-01 -3.58332992e-01 -2.11921394e-01 -7.19769657e-01 8.35226595e-01 8.26037116e-03 1.52846205e+00 -3.46494436e-01 -2.02965692e-01 7.02083826e-01 1.47938418e+00 4.34449799e-02 2.48941898e-01 2.26721227e-01 1.07063448e+00 5.91166914e-01 8.23597729e-01 3.88985246e-01 7.24161446e-01 6.58206999e-01 2.57280439e-01 -4.22237247e-01 -4.42331463e-01 -4.37560380e-01 -2.19794130e-03 8.97095859e-01 1.55817911e-01 -1.84912786e-01 -9.39165890e-01 6.44920766e-01 -1.87851369e+00 -7.71259606e-01 -7.04462752e-02 1.89607704e+00 9.82201397e-01 -9.19829160e-02 -1.12908445e-01 -3.55644733e-01 1.00587475e+00 3.61036897e-01 -7.09841788e-01 -2.62767434e-01 -6.22192174e-02 -3.84408265e-01 3.12762260e-01 3.33745092e-01 -1.25153983e+00 1.12825406e+00 5.16866827e+00 8.81276548e-01 -1.17650402e+00 1.61002010e-01 8.71330917e-01 1.25500247e-01 -3.44546288e-01 -1.49208009e-01 -6.89758658e-01 7.31715024e-01 4.04434800e-01 -1.04751356e-01 3.91663671e-01 7.49626398e-01 3.16179365e-01 1.68288454e-01 -1.18127489e+00 1.58054173e+00 3.99415344e-01 -1.16354299e+00 3.30483556e-01 -3.30358058e-01 7.30637908e-01 5.48785552e-02 6.78706691e-02 4.04781938e-01 4.36779484e-02 -9.33023036e-01 7.45907784e-01 4.01580751e-01 1.07715940e+00 -4.83763069e-01 5.98168731e-01 1.99788809e-01 -1.30627954e+00 1.26345158e-01 -4.67190623e-01 8.15232545e-02 -1.40627965e-01 3.17847550e-01 -4.73960102e-01 4.16304260e-01 9.70774710e-01 9.76506770e-01 -9.00736272e-01 7.06415892e-01 -1.86122507e-01 1.96380720e-01 2.80776411e-01 1.57760829e-01 2.57097661e-01 -1.64648384e-01 1.02722913e-01 1.25514150e+00 2.56083254e-02 -5.94992712e-02 3.33556712e-01 1.03762734e+00 -2.13178217e-01 1.93010211e-01 -3.70521754e-01 -9.95946154e-02 6.17986917e-01 1.36317861e+00 -3.71347815e-01 -3.45139265e-01 -8.50663185e-01 1.36111736e+00 5.00324547e-01 3.38282794e-01 -8.78523767e-01 -3.56289029e-01 6.59983635e-01 5.66220470e-02 2.15940848e-01 -2.34764181e-02 -1.33516163e-01 -1.39015102e+00 4.77708094e-02 -1.08678818e+00 3.25525165e-01 -8.89092684e-01 -1.59314179e+00 7.10608125e-01 -2.10360866e-02 -1.35304213e+00 -9.17742699e-02 -6.31151438e-01 -6.00523055e-01 9.01124179e-01 -1.80594730e+00 -1.36287892e+00 -9.03388321e-01 8.46523464e-01 8.50702465e-01 -7.82210976e-02 5.24204612e-01 4.22080308e-01 -6.48080111e-01 7.73738265e-01 1.15912907e-01 2.36623421e-01 1.11085737e+00 -1.17639124e+00 4.49205667e-01 7.65641093e-01 -2.56732069e-02 3.40425998e-01 4.19612169e-01 -3.88674915e-01 -9.91841972e-01 -1.34616911e+00 8.23475778e-01 -1.84596911e-01 5.22440910e-01 -4.55171734e-01 -1.01445198e+00 4.89373773e-01 3.81013572e-01 3.25659126e-01 5.97862542e-01 -1.43158123e-01 -7.00541556e-01 -2.86597222e-01 -9.39459920e-01 5.70978522e-01 1.02262664e+00 -7.46569037e-01 -6.31166279e-01 2.53196627e-01 8.18958938e-01 -2.92119801e-01 -6.09029830e-01 3.69790196e-01 3.83873522e-01 -8.92582536e-01 8.12521100e-01 -5.51404417e-01 5.12720168e-01 -5.52055180e-01 -2.02664226e-01 -1.38408673e+00 -3.45126271e-01 -1.48953497e-01 3.11060339e-01 1.58299470e+00 1.70539528e-01 -4.62471753e-01 3.78769487e-01 4.48201865e-01 -7.61314994e-03 -5.84906459e-01 -4.24863935e-01 -7.57266462e-01 1.36835128e-01 -5.42359706e-03 5.59867620e-01 1.12471330e+00 -3.01028728e-01 6.17972255e-01 -4.18794513e-01 1.01611868e-01 5.69094300e-01 3.49067539e-01 7.27452219e-01 -9.32810724e-01 -3.16095978e-01 -3.64911646e-01 -3.41402024e-01 -1.19427919e+00 3.20257246e-01 -8.22797716e-01 2.79482663e-01 -1.51147950e+00 6.67307198e-01 -4.56891745e-01 -2.41007999e-01 6.06605113e-01 -4.40111965e-01 3.03383350e-01 3.04741561e-01 4.84682947e-01 -1.01626337e+00 7.45552301e-01 1.35509181e+00 -3.37983400e-01 2.61867158e-02 -4.51116472e-01 -8.24790180e-01 6.93519175e-01 9.70381916e-01 -2.06879288e-01 -6.41785383e-01 -9.02765989e-01 -2.81552166e-01 -1.33710206e-01 4.28850323e-01 -7.50070751e-01 5.33035062e-02 -3.93846393e-01 2.93658048e-01 -6.76059544e-01 3.70739005e-03 -6.00921035e-01 -2.44560167e-01 3.05228531e-01 -5.92356086e-01 2.37622112e-01 2.65190095e-01 4.85278755e-01 -5.15090883e-01 -1.16240337e-01 1.02513421e+00 -1.65402427e-01 -1.17923200e+00 4.73455369e-01 1.00692883e-01 4.85883802e-01 1.08154798e+00 1.04208298e-01 -5.96146584e-01 -2.78776139e-01 -3.57004821e-01 5.38913369e-01 6.89151883e-01 7.92389214e-01 5.82404613e-01 -1.54712403e+00 -9.80474710e-01 1.33158296e-01 8.10368478e-01 -1.29363433e-01 5.56071579e-01 6.89068437e-01 -4.77708966e-01 5.72680593e-01 -3.82618487e-01 -6.16023540e-01 -1.41309190e+00 8.67392361e-01 3.92983884e-01 -7.69679844e-02 -3.83342057e-01 7.50480950e-01 8.52317035e-01 -4.87280041e-01 2.09025726e-01 -3.24351877e-01 -1.71065003e-01 -1.65771186e-01 6.69125795e-01 -3.99788609e-03 -1.14027075e-01 -7.70141721e-01 -5.65765619e-01 7.03289986e-01 -5.09886503e-01 2.36352906e-01 8.88178051e-01 -7.13383257e-01 -8.28995407e-02 3.13890398e-01 1.41085815e+00 -2.24484205e-01 -1.49435449e+00 -5.13521910e-01 -6.16932623e-02 -6.13891721e-01 5.81038091e-03 -8.91471028e-01 -1.03754115e+00 9.90567386e-01 9.65344608e-01 -1.30495280e-01 1.26973772e+00 1.74470991e-01 6.35761082e-01 2.30671644e-01 -1.35201350e-01 -1.09880209e+00 2.31064737e-01 3.65635306e-01 1.08760047e+00 -1.73545325e+00 -2.59076029e-01 -3.09121341e-01 -1.03151035e+00 7.53573775e-01 1.05586541e+00 8.93272609e-02 4.29302245e-01 -4.30326425e-02 4.13934767e-01 9.20880064e-02 -7.75512576e-01 -5.11086524e-01 5.88731945e-01 6.93264544e-01 6.73479021e-01 1.66097712e-02 -4.41894591e-01 7.15613902e-01 2.16217890e-01 -1.09506212e-01 2.57613719e-01 4.90442008e-01 -2.68709302e-01 -1.01009047e+00 -1.24761373e-01 2.34474033e-01 -2.17033669e-01 -3.94062787e-01 -4.87324059e-01 7.05980539e-01 3.81016493e-01 1.02841258e+00 2.51976311e-01 -2.34314248e-01 1.84508458e-01 -2.83363670e-01 4.33136165e-01 -5.94462693e-01 -1.07868247e-01 -4.60822955e-02 -1.36121690e-01 -5.56423068e-01 -4.80928719e-01 -2.94473767e-01 -1.13611662e+00 -2.09406182e-01 -8.17272365e-02 1.23509988e-01 4.32982326e-01 8.49506676e-01 5.78215301e-01 3.93276334e-01 6.47489369e-01 -6.06343269e-01 -4.73025501e-01 -9.72926915e-01 -4.68018293e-01 7.60907471e-01 3.21621507e-01 -5.22148132e-01 -2.97677964e-01 2.86853760e-01]
[10.795832633972168, 1.4321283102035522]
05d91110-5322-46c9-89e3-8c84aa6d9e20
espnet-onnx-bridging-a-gap-between-research
2209.09756
null
https://arxiv.org/abs/2209.09756v2
https://arxiv.org/pdf/2209.09756v2.pdf
ESPnet-ONNX: Bridging a Gap Between Research and Production
In the field of deep learning, researchers often focus on inventing novel neural network models and improving benchmarks. In contrast, application developers are interested in making models suitable for actual products, which involves optimizing a model for faster inference and adapting a model to various platforms (e.g., C++ and Python). In this work, to fill the gap between the two, we establish an effective procedure for optimizing a PyTorch-based research-oriented model for deployment, taking ESPnet, a widely used toolkit for speech processing, as an instance. We introduce different techniques to ESPnet, including converting a model into an ONNX format, fusing nodes in a graph, and quantizing parameters, which lead to approximately 1.3-2$\times$ speedup in various tasks (i.e., ASR, TTS, speech translation, and spoken language understanding) while keeping its performance without any additional training. Our ESPnet-ONNX will be publicly available at https://github.com/espnet/espnet_onnx
['Shinji Watanabe', 'Tomoki Hayashi', 'Yosuke Higuchi', 'Masao Someki']
2022-09-20
null
null
null
null
['spoken-language-understanding', 'spoken-language-understanding']
['natural-language-processing', 'speech']
[-2.70760804e-01 6.59673065e-02 -3.18035930e-01 -7.21534252e-01 -4.23679113e-01 -4.37303871e-01 2.33694669e-02 -1.95005193e-01 -2.86836088e-01 2.14885548e-01 -3.70920002e-02 -8.01472306e-01 4.06299531e-01 -7.21299410e-01 -7.32547700e-01 -2.10608140e-01 2.21081987e-01 4.02164608e-01 -3.37809995e-02 -1.63372651e-01 -1.09264940e-01 4.59501445e-01 -1.25577581e+00 2.34557703e-01 5.95741987e-01 1.01874340e+00 2.38301232e-01 8.44135225e-01 -4.77153093e-01 7.99383759e-01 -7.39016056e-01 -5.95422566e-01 -2.04249155e-02 -8.57230723e-02 -8.21181178e-01 -5.06935477e-01 1.24619886e-01 -3.39831650e-01 -4.88390714e-01 1.09639370e+00 6.15506947e-01 -2.30567530e-02 3.36302444e-02 -1.48331523e+00 -5.28985620e-01 1.24909365e+00 -2.59672135e-01 5.79212643e-02 1.42207518e-01 2.79045045e-01 9.84381020e-01 -6.95409894e-01 1.94647595e-01 1.33473766e+00 6.26293302e-01 6.56809390e-01 -8.09658885e-01 -1.01023221e+00 1.17025033e-01 3.10215920e-01 -1.55678225e+00 -9.19468343e-01 7.45712757e-01 -5.99009432e-02 1.23643601e+00 4.92186725e-01 4.35013652e-01 1.31608093e+00 -1.01853535e-02 1.03209889e+00 5.37690222e-01 -1.34158060e-01 3.19781452e-01 5.83235100e-02 2.51754552e-01 6.26789451e-01 -1.47744834e-01 -1.85248718e-01 -6.22779667e-01 -3.47343199e-02 6.38544619e-01 -2.75243610e-01 -1.14576653e-01 3.85336041e-01 -9.58887219e-01 5.58137894e-01 4.40548837e-01 1.75846681e-01 -3.08958709e-01 4.01221275e-01 7.59598434e-01 3.71988237e-01 5.89067876e-01 2.10666656e-01 -8.98115993e-01 -6.39945090e-01 -8.79303515e-01 8.68630931e-02 1.13374984e+00 1.23603415e+00 7.31481135e-01 4.10949081e-01 1.71245843e-01 1.10265553e+00 4.30408418e-01 3.13608199e-01 6.24757528e-01 -9.75716293e-01 5.49291313e-01 3.18792582e-01 -5.76168478e-01 -6.15027785e-01 -3.25869918e-01 -4.06893879e-01 -1.05402243e+00 -3.64042461e-01 1.20863309e-02 -4.88232791e-01 -9.40171063e-01 1.73246336e+00 1.75344497e-01 5.37675738e-01 5.32486960e-02 7.17172444e-01 1.10690629e+00 9.72520411e-01 1.54216170e-01 1.95991695e-01 1.21855998e+00 -1.41812825e+00 -6.19392216e-01 -5.30239344e-01 7.59257853e-01 -7.53126383e-01 1.44610238e+00 4.86665964e-01 -1.13055182e+00 -6.78416193e-01 -9.25230801e-01 -3.24377924e-01 -4.72342640e-01 1.39154628e-01 8.12853277e-01 5.80207586e-01 -1.36568570e+00 7.31567264e-01 -1.10526145e+00 -3.54289949e-01 2.08684236e-01 4.99803245e-01 -1.25635222e-01 1.47126764e-01 -1.32747841e+00 5.96749663e-01 4.26137209e-01 2.63446897e-01 -6.68512464e-01 -8.21791947e-01 -8.73260379e-01 3.33941042e-01 2.39288330e-01 -7.51812816e-01 1.76428723e+00 -9.23850834e-01 -2.05597615e+00 5.12232900e-01 -1.17499344e-01 -4.87132579e-01 4.35333028e-02 -1.78918958e-01 -7.10730314e-01 -4.40669715e-01 -2.76716769e-01 6.95538104e-01 7.47662425e-01 -5.68136156e-01 -3.30288410e-01 -1.84667394e-01 2.33457550e-01 8.96371529e-02 -4.78921473e-01 3.83845717e-01 -8.56772780e-01 -4.56326395e-01 -2.41498351e-01 -9.29610789e-01 -1.86143339e-01 -1.13055348e-01 -6.89032137e-01 -4.73772079e-01 8.76031756e-01 -8.74234080e-01 1.40360761e+00 -2.23110652e+00 -9.10898019e-03 9.99726281e-02 3.17008108e-01 6.12654150e-01 -2.91443765e-01 2.48490080e-01 -1.51532879e-02 3.35764229e-01 -1.89132467e-02 -5.92861116e-01 2.56027162e-01 3.74199927e-01 -1.40350580e-01 5.76751046e-02 7.40986019e-02 9.10517454e-01 -6.14458084e-01 -2.62095243e-01 2.09231645e-01 5.66767037e-01 -5.73242664e-01 2.48479486e-01 -3.38832289e-01 1.41712517e-01 -4.74952370e-01 6.88245058e-01 6.51590407e-01 -4.17076766e-01 2.16651887e-01 -1.29074499e-01 -5.98249100e-02 8.52849245e-01 -9.89825428e-01 1.86617970e+00 -1.00708163e+00 8.28575611e-01 4.81328756e-01 -9.46278393e-01 9.36528921e-01 3.75135928e-01 6.67817295e-02 -4.54087317e-01 2.93654501e-01 8.15769881e-02 1.14582494e-01 -4.18981940e-01 4.89427567e-01 3.66832703e-01 3.85778770e-02 5.25176108e-01 2.76592344e-01 -3.36795226e-02 -5.54324826e-03 2.22972147e-02 1.09772313e+00 -1.38825372e-01 9.04512256e-02 -5.59459366e-02 2.87548959e-01 -2.64615685e-01 3.93499583e-01 4.76594180e-01 -1.30632892e-01 4.40790802e-01 4.95422572e-01 -4.03409153e-01 -8.73067498e-01 -8.92584741e-01 1.45007774e-01 1.51710379e+00 -3.37588400e-01 -7.95447648e-01 -1.21196330e+00 -3.51527303e-01 -2.25776300e-01 8.73734176e-01 -8.00145492e-02 -2.37808630e-01 -6.98616207e-01 -6.93778396e-01 1.07256722e+00 6.44547284e-01 6.72451973e-01 -1.11860096e+00 -4.25724573e-02 1.95439830e-01 -1.87373370e-01 -1.20823061e+00 -5.85356832e-01 2.88068652e-01 -7.21470714e-01 -5.10909200e-01 -4.06688422e-01 -8.16537321e-01 1.80472597e-01 4.52940166e-02 1.33148170e+00 2.05800533e-01 6.55026734e-03 -1.08145624e-01 -2.39559516e-01 -5.27079642e-01 -5.67734063e-01 7.48717785e-01 5.03865369e-02 -2.47593313e-01 4.74911839e-01 -6.53350949e-01 -3.58111590e-01 1.90408275e-01 -8.10172677e-01 2.20170006e-01 4.29172069e-01 3.83675694e-01 3.18751097e-01 -1.38141707e-01 4.03920829e-01 -8.28474462e-01 8.97638202e-01 -5.57017803e-01 -6.60380721e-01 2.32396305e-01 -5.97983658e-01 -5.65136559e-02 9.03169274e-01 -4.96978968e-01 -7.53487349e-01 -7.32470304e-02 -9.09587979e-01 -5.03636122e-01 -1.20990179e-01 8.91987681e-01 -1.53733477e-01 1.05258472e-01 5.97928822e-01 1.01810060e-01 8.83762389e-02 -5.57366133e-01 4.25464958e-01 1.07587671e+00 2.43793920e-01 -6.02570713e-01 4.41027105e-01 -1.44123167e-01 -6.95810795e-01 -9.66054142e-01 -6.61300302e-01 -2.81467557e-01 -2.20177591e-01 1.31965190e-01 6.83499694e-01 -9.17169034e-01 -8.88401687e-01 5.49300492e-01 -1.34892201e+00 -8.51401567e-01 5.61058521e-02 4.78432089e-01 -2.90416032e-01 1.03713050e-01 -1.00161266e+00 -3.27157915e-01 -7.67566144e-01 -1.38938129e+00 1.02075052e+00 3.28300089e-01 -3.71844262e-01 -9.72362638e-01 -7.47446017e-03 2.91354448e-01 9.24656510e-01 -6.43467844e-01 8.47009301e-01 -7.38527179e-01 -4.11329240e-01 -6.15801401e-02 -3.30663949e-01 7.55810738e-01 -1.26331672e-01 4.19683665e-01 -1.18358290e+00 -1.84048444e-01 -9.89004672e-02 -2.13441193e-01 4.89825010e-01 3.65545124e-01 1.95184052e+00 -4.20647413e-01 -1.84927955e-01 9.96371388e-01 9.32693899e-01 1.28300682e-01 6.08395219e-01 1.38167381e-01 8.11469734e-01 3.25004250e-01 7.20620975e-02 2.52793252e-01 5.04901409e-01 6.96896911e-01 3.10683101e-01 -1.96422547e-01 -8.91659856e-02 -2.15479121e-01 5.49546897e-01 1.50490153e+00 8.02273676e-02 -4.56124604e-01 -1.02408123e+00 1.97305173e-01 -1.64876568e+00 -4.94058430e-01 1.46743119e-01 1.80243313e+00 9.55584824e-01 1.15166523e-01 -1.29241019e-01 -3.19371969e-01 7.41077721e-01 3.08337927e-01 -7.61643946e-01 -6.74039543e-01 2.33703479e-01 6.01559460e-01 4.36914593e-01 4.32498068e-01 -7.85563707e-01 1.36189914e+00 6.18022871e+00 1.00513029e+00 -1.71203005e+00 1.42946631e-01 7.78511226e-01 1.06243109e-02 -1.51375771e-01 -1.34486526e-01 -8.04987013e-01 4.68120098e-01 1.60257673e+00 -4.64080602e-01 8.64560366e-01 1.31623173e+00 3.07779610e-01 5.82330287e-01 -1.12444794e+00 1.20181453e+00 -2.01688737e-01 -1.31046081e+00 -9.55567062e-02 -1.91102996e-01 1.14132307e-01 6.37532294e-01 2.68798991e-04 7.11339533e-01 5.45847416e-01 -1.17655230e+00 7.24339306e-01 -3.36109698e-02 8.05554211e-01 -6.15670741e-01 7.42887437e-01 2.49398351e-01 -1.11663997e+00 3.82978469e-01 -4.35731143e-01 -1.31207198e-01 1.70454204e-01 4.93779361e-01 -1.10999846e+00 3.07047904e-01 9.41174626e-01 6.79078400e-01 -3.23040932e-01 6.52589023e-01 -2.68753707e-01 1.08006966e+00 -4.87655222e-01 -1.75692916e-01 9.13158953e-02 -1.82633594e-01 1.14370286e-01 1.34585750e+00 3.53313059e-01 -9.55084860e-02 -1.57855928e-01 1.00725174e+00 -6.20231450e-01 -1.45805608e-02 -4.03444648e-01 -3.29465359e-01 7.08579421e-01 1.42100990e+00 -3.55007082e-01 -4.72143441e-01 -4.77354079e-01 8.62930954e-01 3.43918115e-01 4.48476851e-01 -1.18249404e+00 -5.45286000e-01 1.06104219e+00 6.23646826e-02 1.69833899e-02 -3.50555688e-01 -2.09801733e-01 -1.21058249e+00 1.06594689e-01 -1.13986015e+00 -2.73937955e-02 -8.90712678e-01 -1.07008827e+00 9.40477192e-01 -7.43176565e-02 -7.55893707e-01 -4.22211289e-01 -6.97386801e-01 -7.16429949e-01 9.44514036e-01 -1.35468793e+00 -9.01524365e-01 -3.50280970e-01 4.45821881e-01 6.30214632e-01 -1.79240584e-01 8.07652235e-01 7.62154937e-01 -1.02258098e+00 8.61438930e-01 -1.76024646e-01 2.98536569e-01 5.53059697e-01 -1.02420700e+00 1.27464581e+00 6.27074778e-01 2.01652631e-01 8.97557437e-01 5.39684474e-01 -1.71362296e-01 -1.55561721e+00 -1.25126350e+00 8.56027961e-01 4.29873168e-02 1.03483558e+00 -7.39031553e-01 -1.03096449e+00 1.01848614e+00 4.07071948e-01 -3.15611698e-02 6.61004066e-01 2.96047598e-01 -2.87624627e-01 -3.06062639e-01 -7.62150168e-01 8.79851043e-01 1.15680242e+00 -7.90669322e-01 1.52652964e-01 4.10604179e-01 1.22445285e+00 -8.12881827e-01 -9.28514302e-01 1.65704787e-01 4.07581002e-01 -6.53145552e-01 8.15163791e-01 -6.02481008e-01 3.39558840e-01 -7.26545751e-02 -1.62087649e-01 -1.48616540e+00 3.80557869e-03 -9.08936679e-01 -1.68052703e-01 1.26107061e+00 7.80977070e-01 -8.08643579e-01 9.41153169e-01 5.65689802e-01 -5.88863015e-01 -7.88648605e-01 -6.82137191e-01 -6.07887149e-01 -6.57032058e-02 -9.09260869e-01 1.07017875e+00 8.78739655e-01 -2.92565405e-01 5.52590489e-01 -2.40238354e-01 2.18449771e-01 1.12563796e-01 -2.69354463e-01 9.54901040e-01 -9.42847133e-01 -5.34993112e-01 -5.13832331e-01 -8.42378885e-02 -1.45558429e+00 4.65773791e-01 -8.88186514e-01 8.17472786e-02 -1.33723903e+00 -2.21264631e-01 -6.19633913e-01 -2.80867428e-01 7.63739467e-01 1.95328698e-01 -1.41784087e-01 1.89503357e-01 -4.14390154e-02 -4.73320395e-01 4.27184254e-01 9.56265509e-01 -2.23402053e-01 -2.79633909e-01 1.57347888e-01 -8.89972329e-01 7.76678562e-01 1.26813078e+00 -4.02042240e-01 -4.21094447e-01 -9.71273899e-01 1.95669085e-01 1.11944815e-02 1.97853953e-01 -9.22193110e-01 3.93587410e-01 2.81721838e-02 -2.15888605e-01 -3.29989016e-01 4.90305066e-01 -3.83394003e-01 1.82728931e-01 9.91636068e-02 -3.27060729e-01 4.18974012e-01 4.32337314e-01 -1.37489792e-02 -4.97958004e-01 -3.20850104e-01 4.69412029e-01 -5.52683882e-02 -7.91627645e-01 5.88446677e-01 -1.31950870e-01 -1.02645300e-01 4.03247595e-01 2.63696045e-01 -6.19167268e-01 -6.15551472e-01 -7.00132489e-01 9.84138846e-02 1.50529772e-01 6.42534077e-01 5.22132099e-01 -1.07649875e+00 -5.98569393e-01 3.00668180e-01 -1.00572005e-01 1.35392100e-01 2.71378785e-01 6.60799444e-01 -8.32485795e-01 4.65862304e-01 8.31344500e-02 -5.08923054e-01 -1.27839983e+00 2.05698192e-01 4.01489675e-01 -1.76029682e-01 -2.33595073e-01 9.36120033e-01 7.14117959e-02 -8.97731125e-01 4.77296114e-01 -6.43666565e-01 1.57037660e-01 -4.61300164e-01 4.47762132e-01 2.20453769e-01 5.58061302e-01 -1.49099961e-01 -3.53766114e-01 8.41835588e-02 -2.00379476e-01 5.40974662e-02 1.27143419e+00 1.14812203e-01 -3.41552734e-01 4.40531969e-01 1.41484869e+00 -2.28641778e-01 -9.12808657e-01 -2.20787734e-01 -2.15926602e-01 1.67704821e-01 2.74696112e-01 -6.28125727e-01 -1.49409330e+00 1.04786098e+00 3.13762546e-01 2.54415482e-01 1.12626636e+00 1.98567137e-01 1.13768816e+00 5.34091532e-01 2.78379291e-01 -1.01324713e+00 -3.33626986e-01 7.91837633e-01 7.76119709e-01 -9.93037701e-01 -2.81002164e-01 -4.69506472e-01 -5.09480536e-01 1.14055514e+00 5.74408531e-01 1.05305590e-01 8.01049352e-01 6.14960074e-01 1.51535898e-01 -7.43462369e-02 -1.02649784e+00 1.34596661e-01 -4.62415852e-02 4.06549245e-01 7.12903976e-01 3.45984668e-01 1.72300801e-01 7.16122031e-01 -6.88505292e-01 1.96699798e-01 3.22904140e-01 6.82203174e-01 -8.59827250e-02 -1.24917114e+00 6.78519951e-03 5.37230611e-01 -5.19059002e-01 -5.48863828e-01 -3.02666366e-01 6.93077683e-01 -1.94023669e-01 8.30828249e-01 2.33545601e-01 -7.64388144e-01 4.09968823e-01 -3.86389084e-02 1.02564767e-01 -7.34448910e-01 -7.44270265e-01 -8.67223367e-02 3.99197310e-01 -6.90176070e-01 5.25888428e-02 -2.27581427e-01 -1.13546896e+00 -8.71933699e-01 -1.15988046e-01 1.74730755e-02 1.08939290e+00 7.44740367e-01 6.97039485e-01 6.43593907e-01 2.84354717e-01 -6.51786327e-01 -5.46854079e-01 -1.12326658e+00 -3.66091102e-01 -1.44674957e-01 7.21737668e-02 -1.21583275e-01 -3.43314290e-01 -5.78732369e-03]
[14.096797943115234, 6.405102729797363]
caa1cede-39bb-4406-976c-83dcdb1c193b
better-context-makes-better-code-language
2306.00381
null
https://arxiv.org/abs/2306.00381v1
https://arxiv.org/pdf/2306.00381v1.pdf
Better Context Makes Better Code Language Models: A Case Study on Function Call Argument Completion
Pretrained code language models have enabled great progress towards program synthesis. However, common approaches only consider in-file local context and thus miss information and constraints imposed by other parts of the codebase and its external dependencies. Existing code completion benchmarks also lack such context. To resolve these restrictions we curate a new dataset of permissively licensed Python packages that includes full projects and their dependencies and provide tools to extract non-local information with the help of program analyzers. We then focus on the task of function call argument completion which requires predicting the arguments to function calls. We show that existing code completion models do not yield good results on our completion task. To better solve this task, we query a program analyzer for information relevant to a given function call, and consider ways to provide the analyzer results to different code completion models during inference and training. Our experiments show that providing access to the function implementation and function usages greatly improves the argument completion performance. Our ablation study provides further insights on how different types of information available from the program analyzer and different ways of incorporating the information affect the model performance.
['George Karypis', 'Sheng Zha', 'Leonard Lausen', 'Jinman Zhao', 'Hengzhi Pei']
2023-06-01
null
null
null
null
['program-synthesis']
['computer-code']
[-3.26359011e-02 8.18329584e-03 -6.02759480e-01 -7.02039659e-01 -7.83135176e-01 -8.86371195e-01 2.86166906e-01 2.77726620e-01 -6.28646761e-02 3.76205921e-01 2.83964306e-01 -9.34310377e-01 3.61040980e-01 -9.25449550e-01 -8.98597836e-01 8.22474808e-02 6.45472258e-02 -6.67202845e-03 2.21772105e-01 -3.99741158e-02 3.72802228e-01 2.29281887e-01 -1.57057679e+00 7.81252861e-01 8.57718885e-01 -1.52748276e-03 3.00034881e-01 9.13914084e-01 -4.60550249e-01 8.84596407e-01 -6.88814521e-01 -4.76188302e-01 -3.78766358e-02 -1.22659110e-01 -1.04796588e+00 -5.31822801e-01 4.18468535e-01 -5.58341205e-01 1.73741564e-01 9.27900016e-01 2.23981425e-01 -2.67660320e-01 1.95471376e-01 -1.30971134e+00 -2.92910904e-01 1.38829517e+00 -3.59433413e-01 1.75911024e-01 5.63463867e-01 3.31678331e-01 1.23529136e+00 -7.56840348e-01 6.65739775e-01 9.65408504e-01 8.92089963e-01 4.99282002e-01 -1.50740874e+00 -4.07131225e-01 6.90762699e-02 -1.03417993e-01 -1.08838904e+00 -6.40188873e-01 4.89205122e-01 -7.69108415e-01 1.70089996e+00 5.82382917e-01 7.57016465e-02 8.76249135e-01 -9.96721387e-02 5.25725603e-01 3.84639174e-01 -5.94350040e-01 -1.45553038e-01 3.37446719e-01 7.57559717e-01 9.30936635e-01 1.89284682e-01 -1.99709907e-01 1.61256641e-02 -7.80322850e-01 1.35855004e-01 -1.18050873e-01 -1.20805494e-01 -1.77198440e-01 -1.15869248e+00 6.58474743e-01 1.53777629e-01 3.71217489e-01 3.06771308e-01 3.74077618e-01 6.86302781e-01 5.10719240e-01 -1.44446671e-01 7.70230353e-01 -1.06360340e+00 -5.10170281e-01 -7.84408033e-01 2.79316545e-01 1.43707049e+00 1.38795280e+00 1.41875219e+00 -1.90150380e-01 -1.99022532e-01 6.95660889e-01 2.16023266e-01 1.61647767e-01 2.10066840e-01 -1.05220044e+00 8.97309124e-01 9.95962918e-01 4.37301360e-02 -8.19327891e-01 -3.08599025e-01 -2.54780382e-01 2.80028939e-01 1.74618080e-01 5.67965686e-01 -1.54866368e-01 -2.52717733e-01 1.73723149e+00 1.15595318e-01 -3.56839001e-01 -1.84321970e-01 2.81666279e-01 8.71629477e-01 2.20899135e-01 1.54898539e-01 3.00905257e-01 1.09770107e+00 -9.42196071e-01 -2.35215172e-01 -5.63963056e-01 1.71497452e+00 -8.50407898e-01 1.49937034e+00 3.60671133e-02 -8.52383792e-01 -5.29218197e-01 -9.84228492e-01 -2.34403402e-01 -3.72043341e-01 4.95337754e-01 1.03098309e+00 7.74657488e-01 -8.31829488e-01 6.79088712e-01 -1.07639277e+00 -2.33958662e-01 1.91977173e-01 3.87657106e-01 -2.96592951e-01 -1.23200431e-01 -6.21768177e-01 7.76668191e-01 2.22581983e-01 -2.76268065e-01 -7.38117397e-01 -1.07506275e+00 -1.14162517e+00 2.63434619e-01 4.41336185e-01 -6.78559065e-01 1.49357796e+00 -1.08747792e+00 -8.13764930e-01 8.68783712e-01 -3.53518665e-01 -5.13975136e-02 2.26766869e-01 -2.64823973e-01 -2.44179014e-02 -6.46281004e-01 2.85697103e-01 1.46423548e-01 4.29825604e-01 -8.95896077e-01 -4.85175610e-01 -1.95153013e-01 7.01392293e-01 -3.99394423e-01 -2.21244708e-01 5.58219433e-01 -6.60990655e-01 -2.69016922e-01 -4.55090284e-01 -8.78200531e-01 1.15561061e-01 -2.85209179e-01 -5.95375896e-01 -1.15560867e-01 3.56910408e-01 -8.54407549e-01 1.79389513e+00 -2.13603091e+00 1.12961993e-01 6.73832148e-02 3.08591634e-01 -2.26209722e-02 -2.53505379e-01 5.07789016e-01 -3.23075593e-01 4.97620672e-01 -4.44718659e-01 -2.22706124e-01 3.93667996e-01 3.16580892e-01 -2.61978924e-01 2.89655149e-01 1.22094285e-02 8.22772205e-01 -8.23662698e-01 -4.22347337e-01 -6.05688542e-02 6.69234917e-02 -1.37812150e+00 2.60903239e-01 -6.30802453e-01 -9.15459022e-02 -6.86029911e-01 8.20381463e-01 5.74597418e-01 -1.68044344e-01 3.14408183e-01 -6.19395860e-02 -2.41808295e-01 9.76581752e-01 -9.41766858e-01 1.92593372e+00 -9.66754973e-01 5.93793988e-01 1.96299508e-01 -8.36034715e-01 5.77840090e-01 1.94455795e-02 1.76530480e-02 -2.67009079e-01 -5.06142676e-01 1.59888774e-01 1.93164915e-01 -8.91625702e-01 4.02485460e-01 3.52284968e-01 -3.57207268e-01 8.09423268e-01 -9.73752141e-02 2.13785917e-02 3.21482748e-01 3.12857032e-01 1.65437496e+00 8.29013228e-01 3.23580027e-01 -1.82760864e-01 6.13829136e-01 2.19784990e-01 5.74857593e-01 9.94557619e-01 2.87738025e-01 1.64925680e-01 1.05714488e+00 -5.23925960e-01 -9.78243649e-01 -6.47510052e-01 -3.59485634e-02 1.65764093e+00 -5.44206142e-01 -1.37136233e+00 -8.34420383e-01 -1.05017853e+00 9.97299626e-02 1.10331428e+00 -6.46259785e-01 -1.35949150e-01 -1.12374628e+00 -7.61270404e-01 8.32901359e-01 7.75602579e-01 -6.82466701e-02 -8.94005775e-01 -6.43378615e-01 7.22624213e-02 -2.71963328e-01 -8.43743086e-01 -4.01773363e-01 1.85334951e-01 -8.59556854e-01 -1.36557627e+00 2.18825459e-01 -5.91024458e-01 8.48414361e-01 -1.64187759e-01 1.86851501e+00 7.52735734e-01 -4.39735383e-01 3.70002329e-01 -1.28057733e-01 -4.70087826e-02 -1.00823700e+00 5.24284899e-01 -7.54608154e-01 -7.83601046e-01 7.15825975e-01 -7.90124953e-01 5.14256656e-02 3.80154461e-01 -7.83941090e-01 1.12536758e-01 2.95411885e-01 6.38417721e-01 -2.60403268e-02 -2.33245462e-01 1.65201694e-01 -1.49217892e+00 4.13835436e-01 -5.83736241e-01 -1.00860023e+00 3.26344520e-01 -3.72556835e-01 6.08568966e-01 8.93628836e-01 -3.28739770e-02 -1.06086361e+00 2.61448771e-01 -3.52289200e-01 1.56867668e-01 -1.61981463e-01 8.48679483e-01 -3.95228535e-01 1.54050410e-01 1.09877789e+00 -2.50236150e-02 -2.31388554e-01 -7.89387941e-01 3.13255459e-01 4.24100190e-01 3.38847876e-01 -1.42352653e+00 8.74965727e-01 1.78175010e-02 -4.34153527e-01 -3.78578097e-01 -3.92165661e-01 -2.85937965e-01 -6.02955341e-01 5.63327730e-01 3.66453767e-01 -7.21233904e-01 -6.69372559e-01 -1.58604595e-03 -1.44331074e+00 -6.21444821e-01 -7.41836578e-02 -9.22954269e-03 -4.37957346e-01 5.15343666e-01 -5.95398247e-01 -3.80707890e-01 -1.41026944e-01 -1.68520892e+00 8.97752881e-01 -3.94677192e-01 -5.22591233e-01 -9.43833351e-01 1.19794272e-01 1.16107628e-01 6.57993376e-01 -9.01852101e-02 1.64609516e+00 -6.31542802e-01 -9.26692486e-01 -1.43942878e-01 -1.44536525e-01 2.85656542e-01 5.48639633e-02 3.30064803e-01 -9.72762108e-01 1.03675714e-02 -1.76159605e-01 -4.38335054e-02 5.57901919e-01 -1.73399910e-01 1.25868392e+00 -4.07950580e-01 -5.58801830e-01 9.98047411e-01 1.41626012e+00 -4.36710358e-01 5.82521319e-01 4.97727454e-01 8.17990959e-01 5.27971089e-01 2.05336243e-01 3.63649756e-01 3.92902672e-01 6.80638790e-01 4.39363211e-01 1.14456549e-01 6.07616641e-02 -3.32329780e-01 5.90162456e-01 5.07359743e-01 2.99255371e-01 2.88535208e-01 -1.12240744e+00 5.62421203e-01 -1.64712000e+00 -6.94406033e-01 -2.57103443e-01 2.34255815e+00 1.25866687e+00 5.79987280e-02 9.01061818e-02 -2.61251032e-01 3.20968062e-01 -1.67416066e-01 -3.42377186e-01 -5.25227904e-01 3.34520996e-01 2.10292786e-01 4.82047439e-01 8.45358372e-01 -8.56914937e-01 7.98908770e-01 6.70913267e+00 3.00508052e-01 -7.88661480e-01 5.06904125e-02 3.87808904e-02 1.21640399e-01 -8.47675741e-01 5.97175300e-01 -1.18137455e+00 3.12186837e-01 1.09332693e+00 -1.96282580e-01 7.59314060e-01 1.29518235e+00 -7.19296783e-02 -9.87469181e-02 -2.02313614e+00 3.67182583e-01 -2.31557429e-01 -1.04155993e+00 -3.39339852e-01 -1.10736139e-01 2.79910505e-01 3.14497501e-01 -3.66695404e-01 8.84045243e-01 5.68016887e-01 -1.06964755e+00 5.21761715e-01 4.99270618e-01 5.39695442e-01 -3.74845684e-01 4.72449154e-01 5.36371112e-01 -1.08334148e+00 -1.24556944e-01 -3.05336624e-01 -3.33647609e-01 -5.23805439e-01 4.33741003e-01 -1.07096505e+00 1.45940185e-01 3.53060991e-01 7.84090757e-01 -1.21115303e+00 7.72607267e-01 -4.05864358e-01 7.82379985e-01 -4.45651025e-01 3.07489187e-02 -3.20581049e-01 1.65010959e-01 3.53531718e-01 1.76077962e+00 6.47274852e-02 -5.30901134e-01 1.77680418e-01 1.62172520e+00 -1.40303716e-01 1.48556709e-01 -7.84515738e-01 -1.16297252e-01 3.84274781e-01 1.13136101e+00 -2.66292602e-01 -4.56561059e-01 -1.03414512e+00 4.38282192e-01 6.88243687e-01 4.80756581e-01 -1.00586045e+00 -5.57306767e-01 8.76840055e-01 2.24570468e-01 2.77904779e-01 -2.11627230e-01 -3.10360789e-01 -1.46251822e+00 5.52164197e-01 -1.18404746e+00 5.01903534e-01 -5.97571313e-01 -8.10064018e-01 1.26239553e-01 2.35926226e-01 -5.82826734e-01 -4.65396583e-01 -4.61341679e-01 -6.51490092e-01 1.23813176e+00 -1.33527970e+00 -8.50288570e-01 -1.16700068e-01 2.54386514e-01 2.68261850e-01 4.30191457e-02 1.20235789e+00 5.57522714e-01 -5.70118010e-01 9.35356855e-01 -1.23856820e-01 4.70083714e-01 7.27806509e-01 -1.30568373e+00 8.62637579e-01 1.09993207e+00 -1.85614675e-01 1.52869701e+00 6.37003422e-01 -5.56774318e-01 -1.86964095e+00 -8.61907423e-01 9.25116360e-01 -1.14072752e+00 8.05824280e-01 -5.01274586e-01 -9.76472914e-01 1.37977624e+00 -1.61353290e-01 1.84841052e-01 7.55077004e-01 5.93755245e-01 -8.76409292e-01 6.72326013e-02 -6.13143623e-01 4.26093072e-01 9.03519690e-01 -9.06573057e-01 -5.77379942e-01 2.08101809e-01 7.16172278e-01 -5.25071979e-01 -9.16007996e-01 5.81513271e-02 5.67219138e-01 -9.67137694e-01 1.00097454e+00 -1.04417050e+00 7.55313039e-01 -2.95768321e-01 -3.65402639e-01 -9.19761121e-01 4.76424657e-02 -5.30981600e-01 -1.27442718e-01 1.32451713e+00 8.82776976e-01 -6.03982568e-01 6.11730814e-01 1.15666807e+00 -4.44568992e-01 -3.18842709e-01 -1.34463280e-01 -3.82904679e-01 2.00631231e-01 -7.46816814e-01 8.11213493e-01 1.00973189e+00 3.65420640e-01 3.74146700e-01 5.28027155e-02 5.34145720e-02 2.13012502e-01 5.16132355e-01 1.29561853e+00 -9.45395172e-01 -1.07629812e+00 -4.94582683e-01 5.82497753e-02 -1.01475251e+00 4.62974370e-01 -1.40238929e+00 -5.95835634e-02 -1.24329090e+00 3.78084630e-01 -7.47937143e-01 2.52795994e-01 8.82692456e-01 -3.55774969e-01 -4.08032149e-01 -2.38714442e-02 1.54516459e-01 -4.25420344e-01 -2.50257701e-01 4.67338800e-01 -8.23399946e-02 -1.20882243e-01 2.36923501e-01 -8.18064213e-01 8.07703257e-01 5.77262104e-01 -8.76802683e-01 -2.68859953e-01 -9.42255735e-01 7.58758247e-01 2.44909793e-01 3.25249612e-01 -7.65014112e-01 8.06060955e-02 -1.34698907e-02 -7.04413056e-02 -2.11417943e-01 -1.72416449e-01 -7.43792593e-01 1.64535910e-01 2.67728955e-01 -6.32454872e-01 1.37203395e-01 5.51204979e-01 1.51069328e-01 9.39171761e-02 -9.12845016e-01 3.44806105e-01 -5.37031889e-01 -4.47321802e-01 -1.17048912e-01 -3.41928303e-01 3.20193738e-01 4.93331641e-01 7.94911608e-02 -5.51602542e-01 2.16238692e-01 -7.29393423e-01 3.63759696e-02 9.26029623e-01 4.94924456e-01 -1.45792633e-01 -7.27631032e-01 -5.14945805e-01 4.41115141e-01 3.35506171e-01 -2.54387438e-01 -3.72201443e-01 8.74936342e-01 -5.99900961e-01 3.23057562e-01 7.33049884e-02 -5.76195121e-01 -1.41891026e+00 6.63882077e-01 3.76083523e-01 -2.87636936e-01 -5.25984108e-01 5.69521546e-01 2.49738768e-01 -1.10676992e+00 2.82656052e-04 -7.57461667e-01 2.03549668e-01 -3.72048527e-01 5.45528412e-01 1.37848333e-01 4.97125149e-01 -1.88265607e-01 -4.41242903e-01 2.20562235e-01 -1.61619321e-01 2.94540137e-01 1.39183903e+00 3.86613786e-01 -6.72037840e-01 2.65846550e-01 1.43016124e+00 6.01976931e-01 -9.13536370e-01 -2.51430184e-01 2.88001120e-01 -5.92566729e-01 -3.55834752e-01 -9.34229612e-01 -9.54832613e-01 9.66163337e-01 -1.56400159e-01 7.48492554e-02 7.67364681e-01 7.30823399e-03 2.34614581e-01 8.46764147e-01 4.71773863e-01 -6.02936208e-01 -5.12917101e-01 5.74299037e-01 4.88303542e-01 -1.15984416e+00 2.20135018e-01 -5.50981641e-01 1.30707130e-01 1.39319050e+00 8.28189433e-01 1.88363805e-01 4.02354985e-01 8.10632646e-01 -2.92957306e-01 -6.06882460e-02 -8.71676505e-01 2.12623566e-01 -5.34244813e-02 5.96581340e-01 1.11381137e+00 -9.58513692e-02 -2.16018274e-01 9.06861782e-01 -1.62147805e-01 9.90063846e-02 7.00188935e-01 1.33045983e+00 -1.38024673e-01 -1.68063712e+00 -3.79264891e-01 5.96592247e-01 -6.90007389e-01 -5.32077730e-01 -2.83485115e-01 9.22117591e-01 -1.29318535e-01 4.84312773e-01 -3.32569122e-01 -2.43520588e-01 3.78825247e-01 2.83557028e-01 5.57819664e-01 -1.12709498e+00 -9.88261700e-01 -3.41857910e-01 6.02623999e-01 -7.85547733e-01 5.73052205e-02 -5.96035361e-01 -1.33028936e+00 -3.27174544e-01 -3.00189912e-01 2.51526207e-01 7.73406446e-01 8.46629918e-01 6.08108521e-01 5.35292625e-01 4.88533899e-02 -4.77822870e-01 -7.83537984e-01 -7.02244222e-01 1.72766492e-01 3.74131560e-01 2.96764970e-01 -2.44597048e-01 -4.80899543e-01 4.13054794e-01]
[7.74192476272583, 7.806286334991455]
c606824f-d82b-49ee-ab22-ced964e48bed
risks-and-benefits-of-using-a-commercially
1903.04907
null
http://arxiv.org/abs/1903.04907v2
http://arxiv.org/pdf/1903.04907v2.pdf
Risks and Benefits of Using a Commercially Available Ventricular Assist Device for Failing Fontan Cavopulmonary Support: A Modeling Investigation
Fontan patients often develop circulatory failure and are in desperate need of a therapeutic solution. A blood pump surgically placed in the cavopulmonary pathway can substitute the function of the absent sub-pulmonary ventricle by generating a mild pressure boost. However, there is currently no commercially available device designed for the cavopulmonary application; and the risks and benefits of implanting a ventricular assist device (VAD) originally designed for the left ventricular application on the right circulation of failing Fontan patients is not yet clear. Moreover, further research is needed to compare the hemodynamics between the two clinically-considered surgical configurations (Full Support and IVC Support) for cavopulmonary assist, with Full and IVC Support corresponding to the entire venous return or only the inferior venous return, respectively, being routed through the VAD. In this study, we used a numerical model of the failing Fontan physiology to evaluate the Fontan hemodynamic response to a left VAD during the IVC and Full supports. We observed that during the Full support the VAD improved the cardiac output while maintaining blood pressures within safe ranges, and lowered the IVC pressure to <15mmHg; however, we found a potential risk of lung damage at higher pump speeds due to the excessive pulmonary pressure elevation. IVC support the other hand, did not benefit the hemodynamics of the example failing Fontan patients, resulting in the superior vena cava pressure increasing to an unsafe level of >20 mmHg. The findings in this study may be helpful to surgeons for recognizing the risks of a cavopulmonary VAD and developing coherent clinical strategies for the implementation of cavopulmonary support.
[]
2019-04-18
null
null
null
null
['circulatory-failure']
['medical']
[-3.78834635e-01 2.57585078e-01 9.70133319e-02 4.61228579e-01 5.01221836e-01 -9.87392068e-01 7.73019493e-02 1.13366798e-01 -3.56379926e-01 5.70855260e-01 1.99649036e-01 -1.21297550e+00 -1.33758932e-01 -5.65777123e-01 -7.33602792e-02 -7.48710692e-01 -2.62252567e-03 5.88411927e-01 3.53945851e-01 1.67562827e-01 -1.66251312e-03 1.05417264e+00 -9.50632274e-01 -2.73040622e-01 9.55854595e-01 6.74660683e-01 1.76214263e-01 5.31978011e-01 -4.94917184e-01 4.20386821e-01 -6.67088032e-01 -3.96135710e-02 8.06104183e-01 -8.80503595e-01 -5.32879718e-02 -3.90275687e-01 2.44113654e-01 -8.08718622e-01 -1.88920707e-01 -7.04807788e-02 7.91207135e-01 -3.18546265e-01 5.04065454e-01 -8.94279242e-01 3.59498203e-01 5.28791130e-01 1.34883583e-01 4.25925910e-01 -2.30242953e-01 4.77921605e-01 3.92526895e-01 -7.57055163e-01 6.02365211e-02 4.93379265e-01 3.59008729e-01 7.90822446e-01 -1.30920362e+00 -2.48928457e-01 -3.56573164e-01 -6.64995015e-01 -9.35158730e-01 -3.44256341e-01 5.87835491e-01 -7.43005335e-01 5.08508086e-01 3.15756530e-01 1.49725628e+00 1.91795811e-01 2.93625355e-01 -4.83946264e-01 1.04095197e+00 -3.67892861e-01 6.31433278e-02 8.63198936e-01 -8.58736187e-02 3.88812810e-01 9.81055379e-01 3.87336671e-01 -1.88357875e-01 -3.08861047e-01 1.13735104e+00 -7.76795521e-02 -8.15899134e-01 -6.24814987e-01 -8.66227150e-01 2.58192480e-01 -1.23451419e-01 6.81276679e-01 -7.80091524e-01 1.88013576e-02 6.48201287e-01 2.92738736e-01 -6.83569908e-02 6.61354899e-01 -4.45472121e-01 -2.66659111e-01 -8.47204089e-01 -2.52013654e-02 1.24283969e+00 5.56791127e-01 -2.26112291e-01 1.82362899e-01 -2.65823036e-01 6.49915576e-01 3.57342839e-01 4.61333841e-01 2.97669798e-01 -1.13255739e+00 3.44053030e-01 2.87438959e-01 2.53889978e-01 -3.70547920e-01 -2.42509574e-01 -6.10676229e-01 -6.04986250e-01 5.36107183e-01 9.36385334e-01 -5.05166054e-01 -2.24991262e-01 1.16046381e+00 3.85559410e-01 -4.03748810e-01 1.24201126e-01 1.03248823e+00 3.68063480e-01 1.94536615e-02 3.13368589e-01 -7.74139583e-01 1.11192894e+00 -6.52059734e-01 -4.59794313e-01 -4.63081151e-03 7.95650959e-01 -5.11566281e-01 9.14595187e-01 -3.40718813e-02 -1.19762254e+00 -2.46901631e-01 -8.95703971e-01 5.62599659e-01 2.18962505e-01 -1.01928383e-01 1.58918306e-01 9.94228184e-01 -1.08452177e+00 8.45054686e-01 -8.20967674e-01 -4.17362511e-01 3.25007915e-01 -9.90317091e-02 -3.19491804e-01 4.06421721e-01 -6.56266868e-01 1.16678154e+00 -7.10264826e-03 9.70242023e-02 -3.37852687e-01 -1.31854606e+00 -5.22525072e-01 7.65162110e-01 9.99245346e-02 -1.25520384e+00 4.56077933e-01 -2.71865934e-01 -1.45515633e+00 6.62535429e-01 9.14821550e-02 -4.50145662e-01 9.76956308e-01 -5.08332923e-02 1.77258372e-01 4.89823610e-01 -1.34620026e-01 -1.39584199e-01 3.64890277e-01 -9.76283431e-01 5.82951903e-02 -3.42421800e-01 -3.63163322e-01 2.28579700e-01 -1.43245354e-01 -1.95674911e-01 3.67201626e-01 -4.21966314e-01 3.31817567e-01 -7.74194598e-01 -1.11529320e-01 5.05081117e-01 1.39674574e-01 3.45131129e-01 5.13271272e-01 -4.54332650e-01 1.06954443e+00 -2.18804216e+00 -2.07589626e-01 9.37149525e-02 3.83243740e-01 5.68587959e-01 4.86881167e-01 7.86611676e-01 -3.28540243e-02 5.64280570e-01 -3.47128272e-01 2.09076300e-01 -3.37803096e-01 2.51585066e-01 -2.88780302e-01 5.43420374e-01 -3.35585326e-01 5.45985222e-01 -5.81605852e-01 -5.87924659e-01 3.95930529e-01 4.17599380e-01 -4.63349849e-01 4.38967556e-01 3.67387205e-01 7.64681458e-01 -1.57115623e-01 2.04176500e-01 5.87547958e-01 2.74792403e-01 5.63314497e-01 2.39401579e-01 -6.73948228e-01 3.11049581e-01 -4.52852964e-01 8.85827005e-01 -3.56892645e-01 3.23249221e-01 4.55924958e-01 -7.52086878e-01 1.19846678e+00 9.30497110e-01 8.50043356e-01 -3.52251709e-01 2.55236059e-01 5.50773621e-01 6.10763431e-01 -7.29868710e-01 -2.33348668e-01 -1.13905299e+00 9.26019788e-01 4.33375090e-01 -3.68978202e-01 -6.15141451e-01 -3.95332314e-02 2.36038134e-01 7.76880085e-01 4.77343891e-03 1.62896246e-01 -8.91753197e-01 7.56246090e-01 -8.38610306e-02 6.34253144e-01 4.52574193e-01 -5.81771433e-01 4.21156496e-01 1.07887995e+00 -4.92498517e-01 -1.09189439e+00 -1.10517669e+00 -7.01942384e-01 -1.99015476e-02 1.12766489e-01 -1.43349439e-01 -2.61484355e-01 -4.11646008e-01 6.52316749e-01 5.32729149e-01 2.23016024e-01 1.16904877e-01 -8.20492804e-01 -3.26603428e-02 5.30097246e-01 4.40804720e-01 3.35484114e-03 -6.18903756e-01 -1.40116966e+00 6.28754437e-01 3.28567237e-01 -9.46209848e-01 -8.90891626e-02 -2.61867106e-01 -1.26896524e+00 -1.21364391e+00 -1.10305989e+00 -2.82217741e-01 5.76781034e-01 3.87146510e-02 8.51826727e-01 6.91074967e-01 -1.95727140e-01 3.77467692e-01 -5.63771278e-03 -4.03968453e-01 -8.57395291e-01 -3.67601484e-01 2.80303568e-01 -5.74577153e-01 -3.08137208e-01 -9.98424590e-01 -8.51076066e-01 5.63519776e-01 -1.84511289e-01 1.00700416e-01 1.03975333e-01 4.46366698e-01 3.71198654e-01 -5.99033177e-01 1.12321424e+00 -7.00070977e-01 4.50514853e-01 -5.43220520e-01 -5.94819725e-01 -3.28055292e-01 -1.14394009e+00 -5.79231679e-01 1.23583865e+00 -8.93065110e-02 -8.33639622e-01 -7.44124770e-01 8.88614282e-02 -4.59526688e-01 -2.33398397e-02 2.23649815e-01 2.27089420e-01 3.17764014e-01 5.94617307e-01 9.84238312e-02 5.32737732e-01 -4.39988911e-01 -6.56173453e-02 3.47364753e-01 3.45662892e-01 -5.49214780e-01 8.66012990e-01 3.01207364e-01 6.19623303e-01 -8.07598352e-01 2.31227368e-01 -5.91116399e-02 -8.20646703e-01 -3.95737708e-01 7.03785300e-01 -3.89572084e-01 -9.44909453e-01 -5.30475080e-02 -6.82464302e-01 -5.01209140e-01 -9.00628626e-01 1.00310123e+00 -5.36040485e-01 6.35317504e-01 -3.83400261e-01 -1.02673161e+00 -5.80337405e-01 -7.04490364e-01 -1.02576651e-01 6.05099536e-02 -3.15162778e-01 -1.14358938e+00 1.49870485e-01 1.61377415e-01 1.08042347e+00 7.82104492e-01 1.46557248e+00 -2.72037745e-01 -4.60658550e-01 -2.35023633e-01 3.06649715e-01 6.86624527e-01 1.13729298e-01 8.47270340e-02 -3.73955041e-01 -2.92904019e-01 3.03663701e-01 4.19814259e-01 3.89171958e-01 4.87926483e-01 5.40965021e-01 -2.34991983e-01 -2.31742367e-01 5.47495723e-01 1.35185993e+00 5.30633509e-01 2.37298667e-01 -2.80276597e-01 2.01421097e-01 7.98442483e-01 3.53334725e-01 5.85961938e-01 -1.08489227e-02 4.75302339e-02 1.64202332e-01 -1.38101012e-01 -3.94951135e-01 -1.85982853e-01 -6.73135696e-03 6.27563536e-01 -3.16347390e-01 -4.85689119e-02 -1.02595603e+00 3.95633906e-01 -9.31646407e-01 -5.33207595e-01 -4.32642192e-01 2.81394696e+00 4.69174922e-01 1.05222151e-01 1.31661564e-01 -1.91292882e-01 4.23491597e-01 -1.15292765e-01 -3.93875450e-01 -6.74895287e-01 1.19483314e-01 2.48025551e-01 3.27994347e-01 5.42692065e-01 -2.90325284e-02 2.78359890e-01 6.88571072e+00 -7.64197290e-01 -1.44794393e+00 -2.08230957e-01 2.66773254e-01 -2.68090695e-01 -3.47715408e-01 7.70915270e-01 -2.07077682e-01 5.30075848e-01 8.35159183e-01 -8.54276597e-01 1.57875884e-02 5.21652937e-01 5.45422137e-01 -3.50496829e-01 -1.04556620e+00 4.31269795e-01 -4.07096475e-01 -1.29137230e+00 -4.51959372e-01 2.14279413e-01 -3.02645534e-01 -2.63866067e-01 -7.03083158e-01 1.61959585e-02 -1.02171707e+00 -5.88285327e-01 3.81578743e-01 3.85601997e-01 1.33208036e+00 6.11976124e-02 6.29132032e-01 6.71265841e-01 -9.54217851e-01 -3.51243690e-02 -1.85637936e-01 -2.39860833e-01 5.58799028e-01 4.87117380e-01 -1.15207887e+00 4.81874049e-02 1.81839645e-01 1.31744482e-02 1.08264670e-01 1.01595783e+00 7.18946084e-02 7.55944073e-01 -5.08559287e-01 1.24539889e-01 -6.52225196e-01 -4.01330888e-01 1.07335830e+00 1.88332275e-01 2.66600996e-01 3.49585921e-01 -3.90833914e-01 1.44091868e+00 4.65341993e-02 5.42702854e-01 -8.81418765e-01 1.23276293e-01 7.02280879e-01 9.87477064e-01 -4.86323744e-01 -3.25129807e-01 -4.13040012e-01 2.87590176e-01 -2.38896176e-01 4.33371544e-01 -3.47361714e-01 2.23964192e-02 4.17881668e-01 1.26141083e+00 -2.72713244e-01 -5.76876476e-02 -9.90431845e-01 -9.95957792e-01 3.21298510e-01 3.44212204e-02 1.64199203e-01 -3.33227843e-01 -4.62217480e-01 3.36465538e-01 1.43355042e-01 -1.08282828e+00 -1.36749804e-01 -3.15677643e-01 -1.12023282e+00 1.75093865e+00 -1.42103362e+00 -3.97890806e-01 -3.55130136e-01 -1.48758888e-01 9.08930451e-02 -8.16711038e-02 7.87933350e-01 1.30594045e-01 -1.63580239e-01 1.49244800e-01 -4.96063888e-01 -3.94194424e-01 6.02777600e-01 -1.17158461e+00 -3.86501014e-01 7.16641128e-01 -1.13775480e+00 7.32218444e-01 6.68806851e-01 -7.92982399e-01 -1.36277354e+00 -3.42762053e-01 1.00885463e+00 -4.46939349e-01 -1.07966430e-01 3.28572057e-02 -8.84230673e-01 1.36302024e-01 1.66891530e-01 1.38897434e-01 1.01097584e+00 -5.27879834e-01 1.69525936e-01 -2.66221076e-01 -1.47550142e+00 2.65613526e-01 6.96033657e-01 2.05446944e-01 -2.61658609e-01 8.36147964e-02 2.49131784e-01 -1.60508737e-01 -1.48717308e+00 5.64045966e-01 8.71794999e-01 -9.02261257e-01 6.21136725e-01 -4.97466564e-01 1.66103393e-01 -2.10465178e-01 5.50642371e-01 -1.10879779e+00 -1.74942985e-01 -8.43873024e-01 1.28727928e-01 1.26596773e+00 2.68956602e-01 -1.35854197e+00 4.24469382e-01 1.25220048e+00 -3.00788403e-01 -8.93014312e-01 -1.08739388e+00 -4.48759973e-01 7.30245769e-01 3.51783693e-01 2.56109923e-01 6.60309494e-01 4.73139673e-01 1.49708465e-01 -1.33886263e-01 -1.67849123e-01 2.81991661e-01 1.07621461e-01 6.96504712e-01 -1.28277695e+00 -3.82219285e-01 -5.31896114e-01 4.75950278e-02 -3.04091662e-01 -3.07611912e-01 -9.86372054e-01 -4.72104281e-01 -1.76614273e+00 -4.46117431e-01 -9.39883351e-01 1.52209148e-01 9.58742052e-02 5.32264933e-02 -8.25925887e-01 6.19172513e-01 2.48828009e-01 1.15462816e+00 3.22048843e-01 1.66942072e+00 7.43530095e-01 -6.29593015e-01 9.44505483e-02 -6.68756962e-01 2.59883076e-01 7.79601038e-01 -5.96328855e-01 -6.47919655e-01 2.00459093e-01 -3.00842047e-01 9.28692758e-01 -3.87222096e-02 -6.83791995e-01 2.31877923e-01 -1.42090634e-01 3.75874639e-01 -3.60949069e-01 -9.19688120e-02 -1.19821799e+00 3.34566534e-01 1.12297654e+00 1.29523814e-01 -1.37282372e-01 4.27061826e-01 -1.73542559e-01 6.35541677e-02 -1.34159774e-01 1.16998553e+00 -2.85872847e-01 5.20197570e-01 1.57151043e-01 -1.00067222e+00 3.39249462e-01 1.29828417e+00 -3.59933913e-01 -3.90592843e-01 -2.66798168e-01 -6.40199780e-01 2.99374580e-01 5.54182231e-01 1.40952552e-02 5.64446151e-01 -5.82427025e-01 -8.58803928e-01 6.54318511e-01 -3.59979630e-01 -5.48455399e-03 5.80721021e-01 1.40867567e+00 -1.40695739e+00 4.77293611e-01 -1.12218104e-01 -5.86088359e-01 -1.06151640e+00 6.00691140e-01 1.21584105e+00 1.95100620e-01 -1.28717017e+00 4.92412508e-01 3.15918565e-01 8.64211395e-02 5.29326797e-02 9.67315398e-03 -4.88600135e-02 -3.25041085e-01 9.20228064e-02 4.13143516e-01 -4.78113621e-01 -6.23006560e-02 -3.76664102e-01 4.14346129e-01 6.28898025e-01 1.58013344e-01 8.18913102e-01 -7.24636093e-02 -2.15597481e-01 1.28003536e-02 3.00795943e-01 4.95001137e-01 -7.77693152e-01 2.79576749e-01 -6.60384595e-01 -9.75411236e-01 -3.12532574e-01 -9.36479390e-01 -1.15584874e+00 8.05295706e-01 2.79945642e-01 4.19856548e-01 1.07934964e+00 4.91171144e-02 4.24154431e-01 -3.08547288e-01 2.38492116e-01 -6.53123796e-01 -4.65375900e-01 -3.14616621e-01 9.42930818e-01 -3.70915890e-01 9.35798213e-02 -7.38042891e-01 -5.67797482e-01 1.21058488e+00 7.62276530e-01 -4.87219021e-02 9.12974894e-01 4.06441599e-01 5.73908448e-01 -2.29102224e-02 -7.18882799e-01 4.52810407e-01 -2.89364457e-01 2.05762222e-01 8.59642029e-01 1.97484434e-01 -1.41452265e+00 5.62901974e-01 -1.40845299e-01 1.15648108e-02 9.73500013e-01 8.18693101e-01 -5.13685524e-01 -9.74367082e-01 -1.13169983e-01 6.01264954e-01 -6.06096506e-01 8.69875103e-02 -6.13922104e-02 7.56973028e-01 4.55146767e-02 7.12475359e-01 -1.62406117e-01 5.94667137e-01 7.90534914e-01 6.61070585e-01 4.10925150e-01 -3.65285814e-01 -8.10113668e-01 3.62216085e-01 -3.18644829e-02 -2.72187758e-02 -3.57115865e-02 -6.51102126e-01 -1.43224895e+00 -1.33834168e-01 -2.46603817e-01 4.09607023e-01 7.59486496e-01 2.99646437e-01 3.74373138e-01 4.88090813e-01 6.62574708e-01 -1.32066488e-01 -9.23138440e-01 -9.83677268e-01 -1.21442533e+00 -6.83698133e-02 4.26966548e-01 -5.41076422e-01 -9.11680937e-01 -4.28503275e-01]
[14.101117134094238, 3.0230000019073486]
9b799b9a-5e04-4b8d-8ece-11565b5208ce
improving-sample-diversity-of-a-pre-trained
1910.0476
null
https://arxiv.org/abs/1910.04760v4
https://arxiv.org/pdf/1910.04760v4.pdf
A cost-effective method for improving and re-purposing large, pre-trained GANs by fine-tuning their class-embeddings
Large, pre-trained generative models have been increasingly popular and useful to both the research and wider communities. Specifically, BigGANs a class-conditional Generative Adversarial Networks trained on ImageNet---achieved excellent, state-of-the-art capability in generating realistic photos. However, fine-tuning or training BigGANs from scratch is practically impossible for most researchers and engineers because (1) GAN training is often unstable and suffering from mode-collapse; and (2) the training requires a significant amount of computation, 256 Google TPUs for 2 days or 8xV100 GPUs for 15 days. Importantly, many pre-trained generative models both in NLP and image domains were found to contain biases that are harmful to society. Thus, we need computationally-feasible methods for modifying and re-purposing these huge, pre-trained models for downstream tasks. In this paper, we propose a cost-effective optimization method for improving and re-purposing BigGANs by fine-tuning only the class-embedding layer. We show the effectiveness of our model-editing approach in three tasks: (1) significantly improving the realism and diversity of samples of complete mode-collapse classes; (2) re-purposing ImageNet BigGANs for generating images for Places365; and (3) de-biasing or improving the sample diversity for selected ImageNet classes.
['Michael A. Alcorn', 'Qi Li', 'Long Mai', 'Anh Nguyen']
2019-10-10
null
null
null
null
['model-editing']
['natural-language-processing']
[ 3.93625498e-01 2.64064699e-01 2.67422229e-01 -1.04668066e-01 -7.95431376e-01 -5.24977624e-01 7.10341990e-01 -6.85318947e-01 -1.28616795e-01 9.56888318e-01 4.79163863e-02 -2.45434597e-01 4.00322378e-01 -1.12584150e+00 -1.03844976e+00 -8.57162237e-01 2.42402226e-01 4.03442115e-01 -1.41351342e-01 -2.43498951e-01 9.67791975e-02 6.10081851e-01 -1.54587114e+00 1.01588935e-01 1.08067012e+00 6.54269099e-01 2.54732490e-01 7.80166626e-01 1.95872843e-01 6.66324258e-01 -8.04888606e-01 -7.08270490e-01 5.83958745e-01 -1.00510490e+00 -3.63317490e-01 8.76938403e-02 4.62117553e-01 -4.38470542e-01 -3.31857383e-01 1.18886828e+00 6.78591669e-01 -5.46182245e-02 5.91862321e-01 -1.35340178e+00 -1.06932271e+00 3.51500452e-01 -5.74495137e-01 -2.41926581e-01 -1.33102730e-01 6.85365677e-01 3.93124044e-01 -5.67375600e-01 6.24089777e-01 1.22231066e+00 5.48195362e-01 9.59905505e-01 -1.19235790e+00 -1.05235803e+00 -1.84018269e-01 -3.12743895e-02 -1.35482013e+00 -4.43628848e-01 7.34025776e-01 -5.30502975e-01 8.43434930e-01 4.53607082e-01 9.73337173e-01 1.50284088e+00 2.21943125e-01 2.70432860e-01 1.12716031e+00 -1.93497509e-01 1.60612792e-01 1.70711353e-01 -7.78690934e-01 5.41758716e-01 2.60475457e-01 3.41391087e-01 -9.81500298e-02 2.68398840e-02 1.08402026e+00 -6.92496225e-02 -6.08304888e-02 4.94217612e-02 -1.06094444e+00 9.92261827e-01 5.62293291e-01 -1.77836549e-02 -4.55260962e-01 5.76446176e-01 5.54980189e-02 1.65645540e-01 5.65596342e-01 7.84892976e-01 -5.78199625e-02 -1.61132544e-01 -9.73658979e-01 2.96796858e-01 4.52288032e-01 1.14993215e+00 9.50266421e-01 5.53218126e-01 -8.93370584e-02 6.86094880e-01 1.72945987e-02 7.16939211e-01 4.54107732e-01 -9.59737360e-01 1.67253688e-01 4.99748886e-01 -2.89086141e-02 -9.49331522e-01 1.01139121e-01 -4.13527578e-01 -1.25518739e+00 3.53518546e-01 2.28926808e-01 -4.15664524e-01 -1.34963965e+00 1.80987656e+00 2.59506971e-01 2.36896321e-01 -6.02193922e-02 6.95367396e-01 8.69698226e-01 9.96245265e-01 1.91568717e-01 1.98976368e-01 1.14168262e+00 -1.09497833e+00 -3.57291013e-01 -4.58100021e-01 2.08045095e-01 -8.65089834e-01 1.28756928e+00 1.38415754e-01 -1.17311347e+00 -6.04302168e-01 -1.11502624e+00 -4.96648997e-02 -3.65101725e-01 1.10933766e-01 6.35380208e-01 9.20457423e-01 -1.09848654e+00 6.38127983e-01 -6.73547268e-01 -2.21231565e-01 8.13835025e-01 2.09196597e-01 -1.65575892e-01 -3.72951418e-01 -1.02365112e+00 7.03320444e-01 2.33253315e-01 -5.27137984e-03 -1.33862853e+00 -9.84921694e-01 -8.41056645e-01 1.15622744e-01 1.05494775e-01 -1.05047858e+00 6.71830416e-01 -1.24420428e+00 -1.65842903e+00 9.05331194e-01 8.44347104e-02 -2.13757217e-01 7.41155505e-01 -1.49652129e-02 -4.36186433e-01 -1.46684736e-01 8.89923796e-02 1.13833499e+00 1.27536798e+00 -1.30135071e+00 -2.71106422e-01 -1.49017707e-01 -4.60254438e-02 1.57931075e-01 -3.15927118e-01 -1.64816841e-01 -4.79875922e-01 -9.17303622e-01 -2.76819438e-01 -1.22751284e+00 -5.00275195e-01 -2.28377864e-01 -6.44913793e-01 3.08487833e-01 8.84339571e-01 -7.62057722e-01 8.02668750e-01 -2.06651616e+00 -1.38716055e-02 -4.28536870e-02 2.43221909e-01 5.84119439e-01 -3.53787214e-01 2.38463163e-01 -2.88641974e-02 4.69286770e-01 -1.81843713e-01 -2.05610052e-01 -1.65298149e-01 2.24299312e-01 -5.28803766e-01 2.35850722e-01 4.22652185e-01 1.22038710e+00 -8.97365451e-01 -3.10540438e-01 3.28645706e-01 8.47559810e-01 -7.43520558e-01 2.36661300e-01 -1.71161324e-01 7.15310037e-01 -1.64147183e-01 6.03470087e-01 7.74653673e-01 -1.29314154e-01 -1.17522962e-01 -9.02580470e-02 2.29098544e-01 -1.39792655e-02 -8.40532541e-01 1.48716354e+00 -5.88383377e-01 8.01174104e-01 -2.43943229e-01 -6.57259226e-01 7.89010584e-01 -3.78525183e-02 7.25913048e-02 -6.22675657e-01 1.73000112e-01 1.99099794e-01 -1.43672749e-01 -2.95835614e-01 4.96175826e-01 -2.62694240e-01 -4.60548997e-02 3.36043298e-01 1.53604075e-01 -7.08507776e-01 2.36982003e-01 6.45124093e-02 9.11726475e-01 1.07911378e-01 -5.94922528e-02 -2.15225115e-01 1.28328532e-01 -1.50126606e-01 5.10176241e-01 5.69018006e-01 8.45706537e-02 9.38148022e-01 5.18573225e-01 -4.25948918e-01 -1.52923930e+00 -9.61916745e-01 2.61784494e-01 8.37874830e-01 -1.20641723e-01 -1.92154288e-01 -1.15092564e+00 -3.88605714e-01 -2.41250649e-01 9.82365608e-01 -7.65055001e-01 -4.41041887e-01 -5.08095682e-01 -9.07018125e-01 8.08215737e-01 5.47614932e-01 8.28529954e-01 -1.09197807e+00 -3.21560353e-01 1.82908028e-02 -1.20782524e-01 -9.79525924e-01 -6.39332533e-01 -1.75976738e-01 -6.89712286e-01 -6.84975028e-01 -1.03778696e+00 -7.04742730e-01 1.18510211e+00 7.13350251e-02 1.20796025e+00 1.00601219e-01 -3.04878086e-01 3.63510326e-02 -1.00026950e-01 -5.97639799e-01 -7.31484950e-01 4.36614901e-02 -2.75880903e-01 -9.42272171e-02 -7.53180757e-02 -7.32839108e-01 -6.29343212e-01 4.48243171e-01 -9.22567487e-01 3.50076675e-01 6.59870327e-01 8.27798724e-01 6.63126290e-01 1.12932451e-01 6.20437682e-01 -9.98551905e-01 4.42069381e-01 -1.07285842e-01 -8.08124006e-01 1.29457176e-01 -5.96175909e-01 -7.53799304e-02 7.51531959e-01 -6.05613232e-01 -1.04108262e+00 -9.37798098e-02 -2.54373342e-01 -5.48513114e-01 5.37185520e-02 2.40041316e-02 -4.65611666e-01 -1.97593898e-01 8.83369207e-01 2.95298040e-01 -1.70812249e-01 3.34420241e-02 5.16426861e-01 3.79689693e-01 6.12616658e-01 -4.77313250e-01 1.12209773e+00 3.79394919e-01 8.52784440e-02 -8.96128237e-01 -4.46128786e-01 2.30146617e-01 -1.01116791e-01 -1.24588974e-01 1.02201748e+00 -1.18125951e+00 -2.33214796e-01 8.71021032e-01 -9.63400126e-01 -7.94915795e-01 -5.00898421e-01 1.33234009e-01 -3.70978802e-01 -1.14679318e-02 -4.27905858e-01 -4.29565966e-01 -4.20183480e-01 -1.31192863e+00 9.67153549e-01 5.45027494e-01 1.94601726e-03 -8.59367430e-01 3.75626571e-02 4.29529190e-01 6.98123217e-01 5.34397542e-01 6.80762410e-01 -2.14010440e-02 -7.47267187e-01 -2.52306581e-01 -3.44716549e-01 5.55550516e-01 -7.22527178e-03 1.83707207e-01 -1.13385987e+00 -3.44422698e-01 -2.47886986e-01 -4.28321540e-01 7.30996907e-01 4.53414589e-01 1.45604181e+00 -4.95047063e-01 -1.71389267e-01 1.13951290e+00 1.33397079e+00 2.59516835e-01 1.36772192e+00 4.64210175e-02 1.05774593e+00 3.27902019e-01 2.65896469e-01 2.91319758e-01 7.54901916e-02 3.63301843e-01 4.42021459e-01 -3.94791633e-01 -4.25808579e-01 -6.29078150e-01 2.76519567e-01 5.81311584e-01 -4.46791470e-01 -4.37492460e-01 -6.01043820e-01 6.61645651e-01 -1.46101952e+00 -1.06978488e+00 6.67738020e-02 2.13034439e+00 9.14725900e-01 -4.62696366e-02 -1.01566285e-01 -3.63439977e-01 8.62033904e-01 1.31475255e-01 -6.93021417e-01 -2.94835567e-01 -1.32637143e-01 4.70081300e-01 6.66706383e-01 2.45832667e-01 -8.73978376e-01 1.09014153e+00 6.11572456e+00 1.09053743e+00 -1.42344618e+00 1.37576520e-01 1.01049030e+00 -2.48471364e-01 -5.96745133e-01 3.92987691e-02 -7.93977320e-01 7.10453570e-01 8.04590642e-01 -6.25270158e-02 8.94073308e-01 1.11663246e+00 -9.64900106e-02 1.25223711e-01 -8.54799867e-01 1.20318079e+00 1.72974899e-01 -1.53642821e+00 2.21272334e-01 3.76289755e-01 1.38480592e+00 1.96670853e-02 3.30793977e-01 4.39824283e-01 6.18117988e-01 -1.39198744e+00 6.95194304e-01 2.48225003e-01 1.36634278e+00 -9.30441558e-01 5.25743127e-01 1.33661747e-01 -7.50092626e-01 2.17481092e-01 -6.49048209e-01 1.96634799e-01 1.43544659e-01 7.49221683e-01 -6.37652278e-01 1.43039510e-01 6.88176394e-01 3.82455647e-01 -6.99721813e-01 6.07845008e-01 -7.00979173e-01 7.11043119e-01 -2.56240100e-01 3.43674496e-02 2.36353263e-01 -3.57764870e-01 3.63224804e-01 1.00778162e+00 7.62005150e-01 1.17885701e-01 -3.27566743e-01 1.22282577e+00 -3.34782213e-01 -4.54764724e-01 -7.09215224e-01 -4.27885860e-01 3.67194265e-01 1.40844238e+00 -7.95840025e-01 -3.79835427e-01 5.49765192e-02 1.08433151e+00 -1.95309967e-02 3.56241047e-01 -1.17559803e+00 -5.40860236e-01 8.04551899e-01 2.51879364e-01 4.01678503e-01 -4.88014221e-02 -4.66916293e-01 -1.18983924e+00 -3.34610701e-01 -1.02527356e+00 -4.35242020e-02 -9.99322474e-01 -1.03229773e+00 7.23936856e-01 -2.67526597e-01 -1.06370759e+00 -4.90754426e-01 -3.96892637e-01 -7.86265135e-01 9.76292610e-01 -1.30729294e+00 -1.55774558e+00 -7.22749650e-01 6.03058159e-01 4.11888599e-01 -3.49680156e-01 8.13310146e-01 4.36530858e-01 -6.96261048e-01 8.48351479e-01 5.08184731e-02 5.38535975e-03 6.26706839e-01 -9.63074684e-01 6.70545042e-01 1.11969280e+00 6.38173148e-03 4.97296154e-01 6.85600042e-01 -6.12178504e-01 -1.26582992e+00 -1.53453279e+00 6.90908313e-01 -4.42607105e-01 2.15834752e-01 -6.67565107e-01 -5.17673552e-01 8.65466714e-01 2.88476288e-01 -4.88789678e-02 5.16504288e-01 -2.38512844e-01 -1.34241655e-01 -1.47009283e-01 -1.36364591e+00 9.52678084e-01 1.09636736e+00 -4.84504163e-01 1.48615107e-01 3.15965652e-01 6.71539009e-01 -5.21449506e-01 -6.24418736e-01 1.60493046e-01 4.38162059e-01 -9.03793991e-01 1.17599678e+00 -4.30134505e-01 9.17889535e-01 -1.30707324e-01 -4.20027897e-02 -1.61992896e+00 -4.00170952e-01 -9.39516664e-01 2.13430956e-01 1.32950354e+00 2.92685449e-01 -7.14772761e-01 7.93330252e-01 5.92124701e-01 -2.82602608e-01 -6.99660361e-01 -4.90039200e-01 -7.11197615e-01 1.60598248e-01 -2.81708300e-01 8.76636744e-01 9.14937139e-01 -7.99006462e-01 2.74236321e-01 -6.95905447e-01 -9.15121064e-02 5.52541196e-01 -7.74552673e-02 1.23946249e+00 -6.09019816e-01 -3.27405423e-01 -4.42575097e-01 -3.95849079e-01 -7.33134687e-01 -7.98808038e-02 -5.46790719e-01 2.90896948e-02 -1.40836453e+00 1.60847932e-01 -4.81088042e-01 2.09732652e-01 5.00660181e-01 -2.71219254e-01 7.31220603e-01 2.02752084e-01 1.09547026e-01 -2.81498041e-02 6.18401170e-01 1.65629876e+00 -2.27681801e-01 -6.29373966e-03 -2.05648497e-01 -1.01555049e+00 5.22120357e-01 7.43990541e-01 -5.03780484e-01 -7.84212232e-01 -4.50214952e-01 4.66815382e-01 -4.23622549e-01 6.32115483e-01 -1.10711586e+00 -3.31606060e-01 -3.57960373e-01 6.22563779e-01 -3.45583916e-01 3.43542993e-01 -4.99998063e-01 6.86261594e-01 5.42119145e-01 9.64073166e-02 -1.73751995e-01 3.35842997e-01 2.99932599e-01 -3.71929742e-02 -2.10904256e-01 1.06567228e+00 -2.93506056e-01 -6.48331046e-01 3.82495582e-01 -2.10842937e-01 3.33452135e-01 1.17573953e+00 -1.36725098e-01 -3.54631126e-01 -6.17907226e-01 -2.12015301e-01 -1.03178523e-01 8.92271280e-01 1.89948469e-01 4.78755116e-01 -1.45392728e+00 -8.41826260e-01 5.16386390e-01 -1.09799057e-01 2.14534208e-01 4.97325927e-01 3.68620753e-01 -9.16119218e-01 4.63466160e-02 -4.80836898e-01 -4.05403286e-01 -9.58454967e-01 5.43432891e-01 1.37762144e-01 -3.68290126e-01 -3.62006217e-01 1.13399470e+00 6.76726282e-01 -5.51786184e-01 -2.61043757e-01 -1.18080437e-01 2.56648570e-01 -1.54998019e-01 4.50574517e-01 3.79578918e-01 -3.03069279e-02 -4.53072667e-01 -2.56381333e-02 3.88966054e-01 1.65044725e-01 -3.26136053e-02 1.46880603e+00 1.77141502e-01 -4.08046209e-04 -1.19671352e-01 1.11630988e+00 -5.44512793e-02 -1.61546683e+00 3.22358578e-01 -1.02857757e+00 -6.03735924e-01 2.48302165e-02 -8.14125538e-01 -1.41442418e+00 9.90595579e-01 4.07982200e-01 -6.77742958e-02 1.25136864e+00 -2.48774648e-01 1.03741765e+00 1.25384614e-01 3.65890950e-01 -1.07230985e+00 1.62716508e-01 2.25121230e-01 1.05206144e+00 -1.03547573e+00 4.70494106e-02 -2.58798212e-01 -7.58833349e-01 6.64534211e-01 6.67491615e-01 -3.04465532e-01 4.85751092e-01 1.68321952e-01 1.07418895e-02 -1.00262314e-01 -3.87273699e-01 2.59383351e-01 2.44476810e-01 8.90327096e-01 1.02462046e-01 2.01188803e-01 1.62445679e-02 3.60595822e-01 -5.34853637e-01 4.08268021e-03 4.52177972e-01 6.32155478e-01 4.75376435e-02 -9.12810087e-01 -3.37453842e-01 5.07683158e-01 -2.04105303e-01 -2.31431544e-01 -3.17062140e-01 7.81805873e-01 3.54334980e-01 7.03643918e-01 3.92057486e-02 -5.82352757e-01 5.96523611e-03 -2.33638793e-01 6.59685791e-01 -5.07112086e-01 -4.69509959e-01 2.07517333e-02 3.67667601e-02 -5.09598076e-01 -2.79652085e-02 -4.67878193e-01 -7.32966542e-01 -7.82218635e-01 -3.42719018e-01 -3.19358230e-01 8.04358840e-01 6.79035723e-01 6.83336914e-01 6.42464817e-01 6.25795186e-01 -1.21375597e+00 -4.56087917e-01 -1.04001617e+00 -5.02565444e-01 4.22260195e-01 -2.25942031e-01 -4.56331551e-01 -2.95601368e-01 2.61316836e-01]
[11.606107711791992, -0.4059349298477173]
703a19ac-1514-4e90-add9-4254b115f78b
leveraging-dependency-forest-for-neural-1
1911.04123
null
https://arxiv.org/abs/1911.04123v2
https://arxiv.org/pdf/1911.04123v2.pdf
Leveraging Dependency Forest for Neural Medical Relation Extraction
Medical relation extraction discovers relations between entity mentions in text, such as research articles. For this task, dependency syntax has been recognized as a crucial source of features. Yet in the medical domain, 1-best parse trees suffer from relatively low accuracies, diminishing their usefulness. We investigate a method to alleviate this problem by utilizing dependency forests. Forests contain many possible decisions and therefore have higher recall but more noise compared with 1-best outputs. A graph neural network is used to represent the forests, automatically distinguishing the useful syntactic information from parsing noise. Results on two biomedical benchmarks show that our method outperforms the standard tree-based methods, giving the state-of-the-art results in the literature.
['Yue Zhang', 'Mo Yu', 'Jinsong Su', 'Zhiguo Wang', 'Linfeng Song', 'Daniel Gildea']
2019-11-11
leveraging-dependency-forest-for-neural
https://aclanthology.org/D19-1020
https://aclanthology.org/D19-1020.pdf
ijcnlp-2019-11
['medical-relation-extraction']
['medical']
[ 1.99467734e-01 6.04273200e-01 -6.11871719e-01 -4.16290849e-01 -8.53301227e-01 -2.17588052e-01 3.25971693e-01 5.47922969e-01 -4.97862101e-01 1.08326054e+00 3.82239074e-01 -6.71596587e-01 1.18599003e-02 -1.01951361e+00 -3.54406327e-01 -6.25153124e-01 -3.16689134e-01 5.42111993e-01 5.59133172e-01 3.93687896e-02 -6.09586462e-02 2.58613259e-01 -1.04963589e+00 5.21983802e-01 8.49630892e-01 7.91828096e-01 -7.78627396e-02 3.65342379e-01 -7.56753623e-01 8.34850729e-01 -6.06060863e-01 -7.12071478e-01 -1.75904855e-01 -3.14012110e-01 -1.17779565e+00 -3.27590495e-01 -2.73252092e-02 1.53197840e-01 3.84519622e-02 1.05071640e+00 3.12990427e-01 -3.48937511e-01 6.90220952e-01 -7.93338954e-01 -1.89277992e-01 9.68556702e-01 -5.77828765e-01 4.39238608e-01 5.78664660e-01 -3.38325649e-01 1.29780889e+00 -6.33815944e-01 1.11210358e+00 1.13082159e+00 7.02653706e-01 5.64014792e-01 -1.24169171e+00 -5.45733035e-01 1.21173687e-01 2.01155171e-01 -1.21391881e+00 -3.65126967e-01 4.82456625e-01 -3.78838211e-01 1.37511635e+00 2.97860086e-01 3.84881735e-01 1.09900606e+00 7.64136553e-01 7.01867700e-01 1.31291282e+00 -6.24408960e-01 1.25888124e-01 3.58386114e-02 5.63892841e-01 9.67732370e-01 6.68717086e-01 -2.20851019e-01 -5.13916552e-01 -3.78426909e-01 3.64295900e-01 -2.44721636e-01 -2.05103427e-01 -6.70100749e-02 -7.78812230e-01 9.58459139e-01 5.56667507e-01 5.45679331e-01 -5.57843089e-01 -3.81530285e-01 4.62742925e-01 1.50524631e-01 6.94806039e-01 5.05790770e-01 -5.89511812e-01 2.31682017e-01 -8.04068208e-01 -8.07663277e-02 1.01484382e+00 6.41446471e-01 5.12488782e-01 -4.85862851e-01 -1.80206090e-01 6.84451044e-01 2.18868345e-01 2.29090415e-02 2.85524875e-01 -2.74754912e-01 5.50892889e-01 1.00714612e+00 -3.85683209e-01 -9.14776683e-01 -9.66890156e-01 -6.85043216e-01 -1.00859451e+00 -1.23057619e-01 4.79627699e-01 -5.83959781e-02 -1.20907915e+00 1.45367122e+00 5.17248333e-01 -3.61039519e-01 1.89743146e-01 5.41801035e-01 1.19508278e+00 1.64915785e-01 5.45751452e-01 -5.33400834e-01 1.68346739e+00 -6.96025908e-01 -9.72778320e-01 -4.54362631e-01 8.02838862e-01 -5.69204628e-01 3.77564579e-01 3.13774109e-01 -6.26694739e-01 1.30294159e-01 -7.41081476e-01 6.67994795e-03 -5.66786945e-01 -2.04527721e-01 1.01371145e+00 6.88692212e-01 -6.91967189e-01 6.83765888e-01 -1.00196099e+00 -4.49762642e-01 7.05970645e-01 4.38165814e-01 -7.08274722e-01 -1.01451293e-01 -1.33926868e+00 1.36934686e+00 3.96729648e-01 5.22684380e-02 -4.05298211e-02 -3.15171301e-01 -8.66416216e-01 3.21956873e-02 7.29448318e-01 -8.00907433e-01 9.57215548e-01 -3.03140074e-01 -9.40539300e-01 9.50389028e-01 -3.13311607e-01 -7.32821941e-01 2.53352553e-01 -1.03603095e-01 -4.38230753e-01 2.15426639e-01 2.66405076e-01 2.73181856e-01 3.92812878e-01 -8.81077111e-01 -6.09425962e-01 -5.75386941e-01 -1.10708863e-01 -1.31155163e-01 -5.94427772e-02 1.47885397e-01 -3.28225672e-01 -3.36943835e-01 3.93033236e-01 -7.69583046e-01 -7.28927433e-01 -6.39993131e-01 -9.62584376e-01 -4.76991028e-01 2.88829118e-01 -8.42209935e-01 1.56688809e+00 -1.53204453e+00 -1.57316282e-01 2.69110978e-01 6.03598237e-01 1.23150535e-02 3.29181880e-01 2.15268612e-01 1.35045638e-02 4.17854697e-01 -3.39834899e-01 1.16097555e-02 -6.19876802e-01 5.22324800e-01 2.81280726e-01 1.89002484e-01 3.79369974e-01 8.36531997e-01 -8.74724567e-01 -9.75882113e-01 -1.85202688e-01 3.64857405e-01 -2.29804933e-01 -7.57004395e-02 -6.98254928e-02 1.89199671e-01 -8.56594265e-01 8.34235847e-01 3.75337154e-01 -3.94491911e-01 7.16501474e-01 -1.61970351e-02 2.55064845e-01 8.45098257e-01 -7.56591737e-01 1.29952073e+00 -3.04154098e-01 2.16561496e-01 -5.01382500e-02 -9.22310710e-01 8.64814818e-01 2.26866543e-01 3.99635106e-01 -4.84087020e-01 1.48436368e-01 1.65187433e-01 2.93124735e-01 -6.24077082e-01 1.11292675e-01 -3.25061381e-01 -2.58399136e-02 -7.08894432e-02 4.70657907e-02 3.62579495e-01 2.44969010e-01 2.42881566e-01 1.62925184e+00 -5.16528788e-04 1.02387905e+00 -4.64577049e-01 3.52891773e-01 2.45531276e-01 8.21124732e-01 5.67816198e-01 -6.56256918e-03 3.59109521e-01 9.49849963e-01 -5.75161159e-01 -3.74185532e-01 -7.13169396e-01 -2.95435309e-01 7.04200625e-01 -3.68932217e-01 -7.18703389e-01 -7.75012553e-01 -1.39859200e+00 -1.10370636e-01 7.21649408e-01 -8.24528635e-01 1.88943118e-01 -5.64159691e-01 -1.25020778e+00 3.76682609e-01 5.12148798e-01 7.16662109e-02 -1.09615815e+00 -8.57538879e-01 5.22131503e-01 -4.19307381e-01 -1.51330185e+00 1.40429065e-01 1.00253105e+00 -1.15216196e+00 -1.44914877e+00 -3.73493195e-01 -5.63715875e-01 7.24015594e-01 -1.91163018e-01 1.65187633e+00 1.00938149e-01 -4.38847065e-01 -3.96658361e-01 -4.05661255e-01 -4.60538715e-01 -5.54036319e-01 5.75398862e-01 -2.98341334e-01 -5.17537534e-01 6.45762026e-01 -4.02440250e-01 -3.06719124e-01 -9.92308743e-03 -5.31857431e-01 -4.22536395e-02 9.36910927e-01 1.13511491e+00 6.51680708e-01 1.28871081e-02 6.14566863e-01 -1.81671965e+00 4.25547689e-01 -4.64835793e-01 -2.13636264e-01 2.72105962e-01 -1.04379153e+00 4.21441048e-01 4.79799092e-01 2.32216883e-02 -8.22090566e-01 3.26320261e-01 -2.95483589e-01 2.15812936e-01 -3.43508720e-01 7.89021552e-01 -1.82852894e-01 2.76276588e-01 8.13050687e-01 -1.89439446e-01 -3.00507426e-01 -6.65279865e-01 1.07140169e-01 5.22689283e-01 2.65607864e-01 -2.40884408e-01 2.79391050e-01 5.08242399e-02 3.82446229e-01 -7.60660231e-01 -1.14294147e+00 -6.70493782e-01 -7.68397987e-01 1.76632628e-01 1.05757320e+00 -4.69233394e-01 -2.80666649e-01 -2.09967554e-01 -1.13520837e+00 1.93935215e-01 -4.23761420e-02 4.32781011e-01 -6.46796823e-02 2.20747203e-01 -8.24955463e-01 -8.71285141e-01 -6.12432837e-01 -9.54572022e-01 8.34113121e-01 7.68989474e-02 -5.73539436e-01 -7.55185544e-01 3.12045999e-02 3.16366911e-01 2.08648331e-02 5.96993804e-01 1.40680420e+00 -1.10317087e+00 -2.22788006e-01 -2.52906382e-01 -8.46326724e-02 -1.84826225e-01 1.50435537e-01 -1.77394226e-01 -1.01805198e+00 3.08068931e-01 -2.62014747e-01 1.01622812e-01 1.27099669e+00 4.95858967e-01 8.04044604e-01 -5.01628280e-01 -1.01962650e+00 1.77312106e-01 1.43432999e+00 7.61130676e-02 6.02559209e-01 5.10643780e-01 5.58954954e-01 8.83559406e-01 4.14424509e-01 -1.95004009e-02 4.15883124e-01 4.58203167e-01 3.62999111e-01 -2.86248654e-01 -9.56451148e-02 -1.67307556e-01 -3.63045745e-02 6.36080146e-01 -2.12734863e-01 -8.13990980e-02 -1.17419398e+00 5.26595712e-01 -1.68057251e+00 -4.96530116e-01 -6.76446617e-01 1.99098253e+00 1.12584484e+00 7.48125851e-01 1.85762212e-01 3.26011032e-01 5.93150556e-01 9.51277744e-03 -1.67448446e-01 -4.75903034e-01 -7.72199482e-02 5.11887968e-01 5.98865807e-01 3.23174119e-01 -1.32616520e+00 1.06712222e+00 6.82288980e+00 5.33246219e-01 -6.95767879e-01 5.18073477e-02 7.50733733e-01 2.60302007e-01 -7.52333403e-02 -4.54222113e-02 -8.90030384e-01 1.12845555e-01 1.08729661e+00 1.89938262e-01 -3.03525746e-01 9.17015195e-01 -8.42226744e-02 -3.75097454e-01 -1.03991711e+00 4.94003713e-01 -3.14005911e-01 -1.28568244e+00 -1.13952368e-01 2.39173010e-01 3.24880153e-01 1.94230601e-02 -5.62678039e-01 1.25342652e-01 4.91046309e-01 -1.17966032e+00 1.40064910e-01 2.49165788e-01 6.51814878e-01 -6.18201971e-01 1.29380667e+00 2.38933891e-01 -1.03328872e+00 2.20951349e-01 -2.75184810e-01 2.91385017e-02 6.98749870e-02 1.40100694e+00 -1.36725676e+00 8.67307007e-01 7.77129292e-01 4.88984227e-01 -7.18416572e-01 9.48605955e-01 -5.67282379e-01 8.26384425e-01 -2.89566100e-01 -2.25396231e-01 -3.14966924e-02 1.13352977e-01 4.34721947e-01 1.65726626e+00 -6.91366345e-02 1.97437420e-01 1.23786911e-01 4.22870725e-01 -9.02241245e-02 4.78026122e-01 -8.22748363e-01 1.15970962e-01 2.12445468e-01 1.27800918e+00 -1.36825800e+00 -2.43685439e-01 -5.39123774e-01 6.00563049e-01 5.60163498e-01 -8.27373341e-02 -2.78496712e-01 -1.53955996e-01 3.54995728e-01 1.59778535e-01 2.41905153e-01 8.37462023e-02 -6.43247366e-01 -9.03519988e-01 1.83866322e-02 -7.57749438e-01 9.43279862e-01 -9.54111442e-02 -1.23700941e+00 9.77697790e-01 -1.70817763e-01 -8.29799235e-01 -3.74569356e-01 -7.00785220e-01 -1.26859695e-01 7.21537650e-01 -1.57962155e+00 -1.02633667e+00 2.32601225e-01 1.98915377e-01 3.97617221e-01 2.08645061e-01 1.30144966e+00 -6.38846681e-02 -6.83592677e-01 4.31722462e-01 -3.34474742e-01 4.12591040e-01 4.60609227e-01 -1.57112551e+00 4.52789992e-01 7.62907386e-01 3.88194293e-01 6.89787984e-01 6.82188213e-01 -9.11017835e-01 -1.15632284e+00 -7.51559317e-01 1.63178289e+00 -5.27901113e-01 5.57367265e-01 -2.36579433e-01 -1.06726229e+00 4.18183655e-01 2.18940191e-02 -2.15565562e-02 9.78653371e-01 8.19815755e-01 -5.47572374e-01 2.84155011e-01 -1.19247591e+00 3.42519611e-01 1.10499120e+00 -2.30133459e-01 -7.64546812e-01 2.05238223e-01 4.94825661e-01 -4.48651433e-01 -9.39137399e-01 4.30029482e-01 6.21674776e-01 -9.40110326e-01 7.82295585e-01 -7.79333830e-01 7.70525515e-01 1.00261003e-01 1.41619697e-01 -1.22581077e+00 -3.68023098e-01 -3.12671185e-01 -2.33321741e-01 1.08089566e+00 1.08327723e+00 -6.01221323e-01 9.84391749e-01 6.53329849e-01 2.03439862e-01 -1.21917892e+00 -9.39225078e-01 -6.63444102e-01 7.41102472e-02 -3.21324795e-01 3.04565400e-01 9.81108010e-01 2.53261775e-01 1.04048002e+00 1.65249668e-02 2.21738461e-02 5.00659764e-01 1.77488804e-01 2.02077299e-01 -1.70983589e+00 -1.16471266e-02 -4.91328388e-01 -5.12738466e-01 -3.81366372e-01 2.55711794e-01 -1.06326163e+00 4.04578373e-02 -2.04031420e+00 4.79972303e-01 -4.75927055e-01 -2.62647778e-01 9.69291925e-01 -5.63510597e-01 4.34056073e-02 -8.79351497e-02 -1.56335421e-02 -4.42531914e-01 -1.83611900e-01 1.05036151e+00 -7.68581480e-02 -1.64980888e-01 1.97941288e-01 -9.54268217e-01 9.25012946e-01 7.00451672e-01 -1.04333794e+00 8.75995681e-02 2.25048382e-02 2.11050287e-01 2.88278341e-01 -1.79311961e-01 -6.32505059e-01 1.19134635e-01 -9.05135460e-03 5.48955500e-01 -5.22338212e-01 -7.84352273e-02 -7.40896225e-01 -6.08316325e-02 6.62259936e-01 -4.62031841e-01 -6.85675964e-02 -5.07771075e-02 6.19866192e-01 -1.38688341e-01 -3.18473071e-01 4.92427886e-01 -3.01081061e-01 -3.02361429e-01 -9.03622359e-02 -3.79644841e-01 1.06277764e-01 5.72361469e-01 -1.67235627e-03 -2.29621783e-01 -1.20719557e-03 -9.82508719e-01 1.18233509e-01 -4.27946635e-02 1.45028770e-01 4.22635376e-01 -7.20441759e-01 -8.41256559e-01 -1.05802037e-01 4.46876362e-02 1.66512415e-01 -1.88574865e-01 8.41110766e-01 -1.88246414e-01 4.73237365e-01 3.98462825e-02 -4.47902232e-01 -1.64090025e+00 5.97728729e-01 9.03408416e-03 -1.24085999e+00 -9.87860262e-01 7.71154106e-01 2.55970825e-02 -1.34981498e-01 5.81641011e-02 -5.86828172e-01 -7.72995293e-01 3.72527659e-01 3.60969275e-01 2.61837810e-01 5.59753418e-01 -3.91645759e-01 -8.81941855e-01 1.72022298e-01 -2.18263209e-01 2.93028384e-01 1.42895317e+00 1.70459479e-01 -3.25351596e-01 3.29163611e-01 8.38894963e-01 3.60722914e-02 -3.73765796e-01 -2.17254713e-01 8.94671559e-01 -1.46874681e-01 2.70152707e-02 -1.09307146e+00 -1.01877797e+00 7.02196121e-01 1.45987585e-01 5.28490663e-01 1.08053195e+00 2.76328325e-01 5.37414968e-01 3.88114274e-01 5.17362654e-01 -7.75507092e-01 -6.08663201e-01 4.23701018e-01 5.86951911e-01 -1.34941769e+00 3.79013211e-01 -1.12779450e+00 -5.98783135e-01 1.28501081e+00 2.74352312e-01 1.11121431e-01 7.50477135e-01 7.76936114e-01 1.75077096e-01 -3.62784415e-01 -7.78214037e-01 -5.48748314e-01 3.48748535e-01 6.54926360e-01 7.96101153e-01 2.93188453e-01 -8.47603142e-01 1.07182169e+00 -3.40207905e-01 -9.93800312e-02 2.60460228e-01 8.46657932e-01 -2.46844903e-01 -1.38381267e+00 -1.95252374e-01 1.01189828e+00 -1.26822007e+00 -3.93124849e-01 -6.77994251e-01 8.75374079e-01 -5.66589721e-02 1.06891477e+00 -5.06645143e-01 -2.55215228e-01 3.74610215e-01 5.17754078e-01 4.08454776e-01 -7.58595824e-01 -9.01988089e-01 1.75341263e-01 7.97444105e-01 -6.73570633e-01 -6.06806815e-01 -4.14801687e-01 -1.41065776e+00 1.89450249e-01 -6.50845706e-01 3.03747773e-01 4.52468395e-01 1.03324246e+00 1.80807218e-01 7.99712837e-01 2.08538875e-01 -5.67140654e-02 -3.00114453e-01 -9.79447782e-01 -4.81788367e-01 2.22787544e-01 2.09768549e-01 -6.54025972e-01 -8.76884982e-02 -1.15402035e-01]
[8.77941608428955, 8.754480361938477]
79912db1-71f9-4936-a1e7-5cff98f2a022
end-to-end-measure-for-text-recognition
1908.09584
null
https://arxiv.org/abs/1908.09584v1
https://arxiv.org/pdf/1908.09584v1.pdf
End-To-End Measure for Text Recognition
Measuring the performance of text recognition and text line detection engines is an important step to objectively compare systems and their configuration. There exist well-established measures for both tasks separately. However, there is no sophisticated evaluation scheme to measure the quality of a combined text line detection and text recognition system. The F-measure on word level is a well-known methodology, which is sometimes used in this context. Nevertheless, it does not take into account the alignment of hypothesis and ground truth text and can lead to deceptive results. Since users of automatic information retrieval pipelines in the context of text recognition are mainly interested in the end-to-end performance of a given system, there is a strong need for such a measure. Hence, we present a measure to evaluate the quality of an end-to-end text recognition system. The basis for this measure is the well established and widely used character error rate, which is limited -- in its original form -- to aligned hypothesis and ground truth texts. The proposed measure is flexible in a way that it can be configured to penalize different reading orders between the hypothesis and ground truth and can take into account the geometric position of the text lines. Additionally, it can ignore over- and under- segmentation of text lines. With these parameters it is possible to get a measure fitting best to its own needs.
['Svenja Leifert', 'Tobias Grüning', 'Roger Labahn', 'Gundram Leifert']
2019-08-26
null
null
null
null
['line-detection']
['computer-vision']
[ 1.80312201e-01 -3.10203582e-01 6.21871464e-02 -3.83739501e-01 -5.41300595e-01 -5.82234561e-01 7.67508447e-01 4.49877024e-01 -7.09350944e-01 3.96175027e-01 -2.57930726e-01 -3.08811188e-01 -3.69308703e-02 -6.58961117e-01 -3.24806422e-01 -6.07755899e-01 6.14965737e-01 5.37913322e-01 4.31561768e-01 -1.33960545e-01 7.94987679e-01 7.64278710e-01 -1.34803355e+00 7.65262172e-02 1.01376104e+00 6.17266774e-01 1.20647147e-01 8.56545568e-01 -3.31277043e-01 4.21319038e-01 -9.60734010e-01 -7.27493107e-01 -4.28983755e-02 -7.00940251e-01 -6.96374893e-01 3.51903081e-01 5.78187048e-01 -1.17600374e-01 -9.42956284e-02 1.17851472e+00 5.00344157e-01 -3.18099350e-01 8.88765216e-01 -9.48047400e-01 -2.38029733e-02 5.65074742e-01 -2.36220688e-01 1.82198301e-01 4.77977335e-01 1.91108838e-01 9.00809348e-01 -7.24794388e-01 3.92248631e-01 8.83858740e-01 5.21105468e-01 4.88872454e-02 -1.04110003e+00 -1.02213562e-01 -2.00257450e-01 1.24454610e-01 -1.31044352e+00 -5.30319333e-01 4.35899436e-01 -6.51204705e-01 7.40257263e-01 4.88424391e-01 3.05246800e-01 9.10415947e-01 2.88984686e-01 7.69371927e-01 1.03286421e+00 -9.70998645e-01 9.87290069e-02 4.87566173e-01 6.23077512e-01 4.80817288e-01 5.35673380e-01 -3.42149854e-01 -3.69980574e-01 1.01292238e-01 2.86719501e-01 -4.75600451e-01 -5.26755273e-01 -3.71595204e-01 -1.00129223e+00 5.20724595e-01 -3.23521867e-02 9.63043332e-01 -2.55617481e-02 -2.87005424e-01 5.95848858e-01 1.91486180e-01 2.50534385e-01 5.79063058e-01 2.52524205e-02 -3.77297044e-01 -1.50651085e+00 8.50718468e-02 9.99513268e-01 5.41494668e-01 3.44830573e-01 -2.42044758e-02 -2.88930297e-01 6.81130588e-01 2.20892087e-01 4.05772328e-01 7.93070197e-01 -5.08512370e-02 4.47734922e-01 6.99957490e-01 9.29001272e-02 -1.11496532e+00 -4.19085354e-01 -4.55608577e-01 -5.91790318e-01 4.35719341e-01 9.01379347e-01 1.09539472e-01 -4.23276931e-01 1.22573972e+00 1.24299740e-02 -3.85987997e-01 2.38639079e-02 8.16944897e-01 4.70355868e-01 3.98943573e-01 -3.96418005e-01 -3.29392999e-01 1.21140766e+00 -9.21525240e-01 -9.28854108e-01 -1.85414299e-01 8.13186705e-01 -1.16615069e+00 1.21007383e+00 5.67586184e-01 -8.65372002e-01 -3.75588953e-01 -1.30845344e+00 2.24841181e-02 -5.28899312e-01 4.96918917e-01 -1.69127211e-01 1.08954287e+00 -7.09081471e-01 6.25175357e-01 -5.12130797e-01 -5.83255529e-01 -1.55534029e-01 2.63190746e-01 -3.16698313e-01 6.49451762e-02 -8.11982155e-01 1.28244865e+00 4.47520286e-01 1.83777466e-01 -3.46395262e-02 4.67883870e-02 -2.99304247e-01 2.52288193e-01 3.76663446e-01 -3.23871613e-01 1.06491125e+00 -1.02659798e+00 -1.51160514e+00 8.87104034e-01 -4.19214256e-02 -2.31664851e-01 1.16791046e+00 -2.78996199e-01 -6.05340540e-01 2.84551717e-02 -2.10045770e-01 -1.62686203e-02 1.03914118e+00 -1.06182015e+00 -4.02842134e-01 -4.22188848e-01 -4.93005574e-01 2.34560773e-01 -3.49443823e-01 7.65668973e-02 -5.25528789e-01 -5.32220244e-01 9.11366343e-02 -6.24516129e-01 4.24498945e-01 -1.59220248e-01 -6.97247148e-01 -1.07046850e-01 6.67433500e-01 -6.98868155e-01 1.39730895e+00 -2.14833522e+00 -2.03721076e-01 2.71384031e-01 -1.19777665e-01 6.30132377e-01 1.02417678e-01 5.17064631e-01 9.78478566e-02 1.97385386e-01 -2.08655789e-01 -2.44485974e-01 2.32896969e-01 -1.43436894e-01 -1.91246361e-01 6.63275898e-01 6.80137426e-02 4.58711743e-01 -6.10587358e-01 -5.48951566e-01 3.79909039e-01 3.90877515e-01 -3.97437662e-02 7.53674135e-02 -9.18540359e-02 -7.43215010e-02 -2.97874719e-01 8.10711384e-02 5.89171231e-01 1.54793710e-01 5.78950683e-04 -9.39917639e-02 -3.73622924e-01 1.65407896e-01 -1.28501081e+00 1.15946364e+00 -9.28372145e-02 1.04002500e+00 -4.03108269e-01 -7.02677429e-01 1.23959458e+00 2.61444062e-01 1.47588067e-02 -6.57983780e-01 4.06324655e-01 5.95667899e-01 5.67921773e-02 -4.55950886e-01 9.25119877e-01 2.03841090e-01 1.62864372e-01 3.17340314e-01 -2.20926821e-01 -1.34606689e-01 4.76462841e-01 -1.25937060e-01 9.02052283e-01 1.61092691e-02 3.00883979e-01 -2.09747687e-01 8.85345638e-01 2.70179119e-02 -4.58193906e-02 7.97810912e-01 4.39378917e-02 7.37455785e-01 6.65020466e-01 -9.25108716e-02 -1.18886495e+00 -5.33104897e-01 -2.56678492e-01 7.38233984e-01 -2.08157208e-02 -2.44219035e-01 -1.07060349e+00 -6.97650969e-01 -2.23612845e-01 9.12580848e-01 -2.33831853e-01 4.55726124e-03 -3.32737982e-01 -7.55178034e-01 8.78976822e-01 9.34136333e-04 5.05034685e-01 -8.30766857e-01 -6.83791280e-01 4.24377657e-02 -1.25953048e-01 -1.07992709e+00 -4.04284269e-01 8.54555145e-02 -8.79140079e-01 -1.04899371e+00 -8.93624067e-01 -4.61125851e-01 6.55335724e-01 8.23810846e-02 8.63403559e-01 1.68608561e-01 -1.45968497e-01 1.19365864e-01 -5.74026465e-01 -1.39434472e-01 -9.72985327e-01 2.58969516e-01 -1.11584723e-01 1.29465923e-01 3.22010309e-01 2.96941727e-01 -1.80815831e-01 6.11622930e-01 -1.08687055e+00 -1.74399406e-01 5.48705399e-01 5.86214900e-01 1.49410054e-01 1.72469124e-01 1.60447553e-01 -8.30445230e-01 7.59296954e-01 2.22822845e-01 -6.97584450e-01 6.16834998e-01 -9.32627082e-01 2.36790255e-01 6.85823858e-01 -2.86997676e-01 -7.46789873e-01 -4.88763042e-02 -2.93977648e-01 5.44764511e-02 -4.63202715e-01 4.86215055e-01 -3.74923825e-01 -1.19659789e-01 8.25799108e-01 4.09973711e-01 -2.13990003e-01 -5.34084022e-01 1.73584491e-01 1.04736543e+00 5.73364258e-01 -8.79713148e-02 6.50049090e-01 -3.44019346e-02 -1.13495700e-01 -1.23391616e+00 -5.08843422e-01 -6.90326869e-01 -7.70301640e-01 -3.07921976e-01 5.24670005e-01 -2.31530562e-01 -5.30505657e-01 9.37039733e-01 -1.20909119e+00 -6.78399876e-02 5.61179705e-02 4.19554919e-01 -4.10955608e-01 1.07614911e+00 -2.51003981e-01 -1.03559899e+00 -3.50491256e-01 -1.24139345e+00 7.27181435e-01 9.24035534e-02 -2.61468261e-01 -9.52687621e-01 -2.06548292e-02 2.71801025e-01 2.16925308e-01 -5.16156182e-02 8.61984193e-01 -1.03949332e+00 1.29691483e-02 -8.42823684e-01 -1.29011095e-01 5.34238815e-01 -1.85041025e-01 5.00216544e-01 -9.45555091e-01 -2.39255250e-01 1.35262981e-01 -3.81759964e-02 6.28014684e-01 4.65369895e-02 5.55795491e-01 -1.63416833e-01 -4.20110077e-02 1.28680795e-01 1.48622251e+00 3.72711532e-02 1.06074870e+00 5.53423464e-01 5.26566684e-01 6.62077367e-01 6.04418516e-01 2.81637430e-01 -2.90248185e-01 1.02492619e+00 1.65623903e-01 3.26139224e-03 4.08351384e-02 -3.89452204e-02 4.89410013e-01 6.53633714e-01 3.47611636e-01 -8.85507405e-01 -9.01282191e-01 -2.10434408e-03 -1.60066569e+00 -1.03512430e+00 -7.10025132e-01 2.82271576e+00 4.35738653e-01 6.86228514e-01 2.68199921e-01 8.64734173e-01 1.05575848e+00 -1.04373738e-01 -1.43287078e-01 -5.93939960e-01 -3.24385643e-01 -2.59900361e-01 5.70061803e-01 5.40794015e-01 -8.57497454e-01 8.10746849e-01 5.98272705e+00 1.08614111e+00 -1.31210005e+00 -2.90080965e-01 4.51191187e-01 2.88586825e-01 1.19335234e-01 -1.46207407e-01 -9.14245963e-01 4.99709070e-01 8.26115727e-01 -1.96251512e-01 -7.87754431e-02 5.97127795e-01 3.10131788e-01 -4.56758976e-01 -1.12198210e+00 1.01401985e+00 4.28172350e-01 -8.27343583e-01 5.54611953e-03 -2.50372244e-03 2.38635018e-01 -2.31512994e-01 -6.01455420e-02 -1.25875711e-01 -4.59268272e-01 -8.89565945e-01 9.51670766e-01 4.75252151e-01 5.86861074e-01 -5.90639770e-01 9.77628052e-01 5.95189333e-01 -5.85493445e-01 4.37503695e-01 -2.81873316e-01 1.79304451e-01 6.57951534e-02 7.74622083e-01 -9.68236268e-01 4.86063480e-01 4.05160040e-02 1.05555035e-01 -7.95604706e-01 1.61630940e+00 -1.19050987e-01 5.05719066e-01 -4.51283008e-01 -5.52459180e-01 2.17986792e-01 -2.56645232e-01 5.77708006e-01 1.57650161e+00 3.65957379e-01 -5.52466691e-01 -2.46868744e-01 7.41369486e-01 2.68120199e-01 7.14152515e-01 -4.08473313e-01 -2.77138621e-01 1.60898611e-01 1.18500924e+00 -1.06932950e+00 -1.22394465e-01 -2.31012404e-01 1.37706661e+00 7.36685470e-02 -3.99244763e-02 -5.67893922e-01 -6.65231168e-01 3.84232751e-03 1.40314847e-01 7.71089420e-02 -2.02355251e-01 -4.22462374e-01 -9.74623322e-01 2.40549684e-01 -1.01918471e+00 8.03052932e-02 -6.45794630e-01 -1.01285267e+00 8.86246264e-01 -3.51418823e-01 -1.15658402e+00 -2.48579413e-01 -8.76811445e-01 -3.79661530e-01 8.49344313e-01 -1.13465238e+00 -7.26619005e-01 -3.68189663e-01 3.12180012e-01 4.46073949e-01 -9.87154916e-02 6.21634305e-01 1.34642690e-01 -8.19957793e-01 9.16266501e-01 4.90293384e-01 2.93797225e-01 8.23090971e-01 -1.16037977e+00 2.24217013e-01 1.27306485e+00 3.67776781e-01 1.42753378e-01 1.12153482e+00 -5.76080322e-01 -9.34459865e-01 -4.92420495e-01 1.16458440e+00 -4.80144590e-01 5.63704312e-01 -1.12770654e-01 -1.17657197e+00 7.68167246e-03 -2.67684106e-02 -4.42666531e-01 5.02016068e-01 -3.71216871e-02 -2.63125241e-01 2.07804926e-02 -9.01197791e-01 5.63551843e-01 2.80744880e-01 -3.84249121e-01 -6.38444602e-01 2.84817398e-01 -4.44253441e-03 -3.86048198e-01 -6.06273115e-01 2.64972113e-02 5.08335531e-01 -1.45938826e+00 3.77392024e-01 2.14446299e-02 3.27322632e-01 -2.60348111e-01 3.00855666e-01 -1.04254115e+00 -5.27841598e-02 -3.72124165e-01 4.08969045e-01 1.32375133e+00 5.17138720e-01 -5.01184523e-01 6.80343926e-01 4.80022699e-01 -8.84850174e-02 -2.59207368e-01 -8.35234821e-01 -8.77010584e-01 1.60782076e-02 -4.34100896e-01 1.53363138e-01 8.39124501e-01 3.14933419e-01 4.00425136e-01 -2.52720088e-01 1.37674883e-01 3.30180764e-01 -1.44998327e-01 7.10998535e-01 -1.22254026e+00 -2.31046990e-01 -1.14388275e+00 -7.64363825e-01 -9.47591662e-01 -5.64684793e-02 -5.96862197e-01 1.85740426e-01 -1.25753260e+00 -9.57030877e-02 -1.57289281e-01 7.37639740e-02 5.08018360e-02 -1.24123827e-01 1.90677047e-01 3.13909262e-01 4.81069177e-01 -4.21764344e-01 2.23317325e-01 9.02533174e-01 5.12746200e-02 -1.22831367e-01 1.79452524e-01 -1.04078136e-01 6.37058020e-01 8.06980848e-01 -4.62807477e-01 -1.35158505e-02 -3.31511885e-01 2.92132556e-01 -1.51202336e-01 2.51384616e-01 -1.22213078e+00 4.01667893e-01 2.44025081e-01 1.97790056e-01 -5.90796769e-01 -9.15189162e-02 -7.90500581e-01 -7.78486952e-02 5.58780432e-01 -3.65590960e-01 1.59472585e-01 -7.43089095e-02 2.67956823e-01 -2.22822398e-01 -1.15847468e+00 9.46033657e-01 1.11536987e-01 -4.95580792e-01 -2.97832251e-01 -5.09058654e-01 -1.49050504e-01 7.98257709e-01 -7.40921557e-01 -4.62502420e-01 -2.29838535e-01 -2.89883703e-01 -1.70600295e-01 8.02998602e-01 1.40010521e-01 3.16206306e-01 -7.18607366e-01 -7.30633855e-01 1.35057159e-02 4.33104575e-01 -5.53359151e-01 -1.85623989e-01 8.65915358e-01 -1.03621089e+00 4.87267524e-01 -1.54571921e-01 -5.56086719e-01 -1.47252023e+00 5.06528556e-01 5.16729057e-01 -3.33663493e-01 -2.43519768e-01 3.35679889e-01 -3.76802891e-01 6.93089329e-03 3.08259696e-01 -1.50615349e-01 -2.38412440e-01 1.47828028e-01 5.99893272e-01 5.12776494e-01 5.06667078e-01 -7.57938981e-01 -1.51178882e-01 5.60922563e-01 -1.32839410e-02 -1.92116931e-01 6.90226555e-01 -2.29884639e-01 -3.43920110e-04 7.30083168e-01 8.52131546e-01 3.46130073e-01 -6.81838512e-01 2.16713939e-02 3.37734729e-01 -4.66020197e-01 2.10764915e-01 -8.37736487e-01 -5.41214108e-01 1.05227184e+00 6.18856072e-01 6.12723112e-01 7.85164893e-01 -6.27471030e-01 2.90685773e-01 4.91431773e-01 4.48589735e-02 -1.24356616e+00 -8.91574658e-04 5.64607263e-01 6.71196997e-01 -1.08614290e+00 -1.98146608e-03 -4.19154406e-01 -5.24793267e-01 1.63484466e+00 2.80737489e-01 1.78063706e-01 6.89860713e-03 2.49992207e-01 1.95364386e-01 1.95590973e-01 -1.84427798e-01 -2.25916326e-01 4.38023299e-01 2.81850427e-01 7.49628246e-01 -1.86509579e-01 -7.49111474e-01 1.78088188e-01 -1.75232634e-01 3.64554711e-02 5.82746208e-01 7.00773001e-01 -7.78963089e-01 -1.18609560e+00 -7.44514465e-01 4.20989573e-01 -5.02602994e-01 2.68135536e-02 -6.54836416e-01 7.63467193e-01 -1.55913293e-01 1.02096975e+00 -9.91222039e-02 -3.79532695e-01 4.35031950e-01 2.88628787e-01 4.18233722e-01 -3.31011951e-01 -7.85722494e-01 1.57483041e-01 3.38265859e-02 -7.24929869e-02 -1.48406506e-01 -6.30660176e-01 -9.07100201e-01 -5.70849955e-01 -9.70567763e-01 1.77870408e-01 8.96553397e-01 1.11331701e+00 -1.70598477e-01 2.14483410e-01 4.24450129e-01 -5.83386302e-01 -7.07802296e-01 -1.06062722e+00 -6.92890882e-01 5.15834332e-01 9.83511377e-03 -2.10104853e-01 -5.50672352e-01 -7.72093311e-02]
[11.815350532531738, 2.6262943744659424]
2411c5ca-c9a5-4248-923c-1d0c1a8c624a
context-aware-document-embedding
1707.01521
null
http://arxiv.org/abs/1707.01521v1
http://arxiv.org/pdf/1707.01521v1.pdf
Context Aware Document Embedding
Recently, doc2vec has achieved excellent results in different tasks. In this paper, we present a context aware variant of doc2vec. We introduce a novel weight estimating mechanism that generates weights for each word occurrence according to its contribution in the context, using deep neural networks. Our context aware model can achieve similar results compared to doc2vec initialized byWikipedia trained vectors, while being much more efficient and free from heavy external corpus. Analysis of context aware weights shows they are a kind of enhanced IDF weights that capture sub-topic level keywords in documents. They might result from deep neural networks that learn hidden representations with the least entropy.
['Junfeng Hu', 'Zhaocheng Zhu']
2017-07-05
null
null
null
null
['document-embedding']
['methodology']
[-4.70638752e-01 1.79544643e-01 -2.42262781e-01 -3.09172511e-01 -3.83871883e-01 -3.37751210e-01 1.07527578e+00 1.93877652e-01 -7.13307798e-01 7.70701766e-01 1.06179976e+00 -2.04259411e-01 -1.81364641e-01 -1.01305592e+00 -2.35251307e-01 -5.49105167e-01 -7.16912076e-02 5.29102564e-01 1.26898944e-01 -4.90865767e-01 4.74974781e-01 -2.62970850e-02 -1.44863868e+00 4.99146253e-01 5.28985918e-01 8.62202585e-01 4.39832687e-01 7.80970216e-01 -8.89618635e-01 6.09327912e-01 -8.88161242e-01 -5.72978139e-01 -1.83022961e-01 1.88578263e-01 -8.94433439e-01 -4.34973210e-01 3.67172241e-01 -1.12434916e-01 -5.24963915e-01 8.12029421e-01 4.91653919e-01 3.60418528e-01 1.05411410e+00 -7.32392311e-01 -1.29298103e+00 1.27637339e+00 -2.23870561e-01 2.75459975e-01 -8.06420520e-02 -3.88590336e-01 1.40241611e+00 -1.22033036e+00 8.93574715e-01 1.16711438e+00 5.09294033e-01 4.98682529e-01 -8.27611387e-01 -3.60995829e-01 3.89700681e-01 4.54724967e-01 -1.15897727e+00 2.33859718e-01 7.24886119e-01 -3.70937824e-01 1.66608310e+00 -2.82668844e-02 5.17390847e-01 1.57520139e+00 4.95972186e-01 8.90981793e-01 3.55271518e-01 -6.35620356e-01 1.07573956e-01 4.06091273e-01 6.87064052e-01 9.55334604e-02 5.16298175e-01 -2.93535590e-01 -1.29262313e-01 -3.18871021e-01 1.45828277e-01 3.70283604e-01 -2.47967049e-01 7.20867813e-02 -9.18753862e-01 1.15008259e+00 2.43747279e-01 8.61978829e-01 -4.54569906e-01 4.91023034e-01 5.32382250e-01 9.31539237e-02 5.52489638e-01 7.64899731e-01 -8.96221280e-01 -5.55965602e-02 -7.48382330e-01 3.09350580e-01 7.07098305e-01 1.05329251e+00 5.99109769e-01 2.50104040e-01 -6.75398409e-01 1.07873070e+00 4.46710289e-01 4.48226243e-01 1.26177287e+00 -5.26517808e-01 2.38799959e-01 4.07988966e-01 -8.72724131e-02 -1.11462486e+00 -4.21952456e-01 -7.82216191e-01 -5.20179689e-01 -3.14504266e-01 -2.57065296e-01 -3.40940177e-01 -1.04360223e+00 1.46711564e+00 -1.96095422e-01 1.10866064e-02 1.22312948e-01 5.66420674e-01 9.27284062e-01 1.16913188e+00 3.67472827e-01 2.29031101e-01 1.61950409e+00 -1.11080980e+00 -1.14215648e+00 -2.73518503e-01 4.63229477e-01 -7.51147389e-01 9.32373703e-01 4.30904865e-01 -9.37366784e-01 -4.72464979e-01 -1.04335225e+00 -2.98859030e-01 -1.35684812e+00 -6.20507356e-03 8.21012318e-01 5.43871045e-01 -1.11560869e+00 5.95332146e-01 -1.46666020e-01 -3.44514698e-01 7.85858408e-02 2.05288734e-02 -2.14894786e-01 1.75088286e-01 -1.79186940e+00 1.04298997e+00 7.02251554e-01 -4.35053736e-01 -8.16644847e-01 -4.66754049e-01 -9.21365321e-01 7.15660572e-01 1.68659404e-01 -3.91553611e-01 1.20630229e+00 -5.88536263e-01 -9.10572708e-01 3.48437130e-01 -2.48422652e-01 -5.37016451e-01 -2.53291160e-01 -4.57672834e-01 -7.72423625e-01 -1.62103042e-01 -1.44475132e-01 5.29567838e-01 7.29363024e-01 -1.27246284e+00 -7.47011244e-01 -6.90879077e-02 -6.65295944e-02 -2.34235562e-02 -1.12774897e+00 -9.90829393e-02 -7.58839250e-01 -8.69130909e-01 -5.63667655e-01 -3.68771225e-01 -3.12949330e-01 -5.37066638e-01 -3.26663405e-01 -7.15096891e-01 7.33392179e-01 -4.12166715e-01 1.83533669e+00 -1.70503187e+00 -1.43095791e-01 1.31972596e-01 2.55272388e-01 5.85982740e-01 -3.35352749e-01 8.94735754e-01 2.06436366e-01 3.85892689e-01 1.23664767e-01 -1.92544386e-01 2.72763163e-01 2.21370786e-01 -4.89293069e-01 -3.80601645e-01 1.85200572e-01 8.76123726e-01 -1.06853640e+00 -5.10232508e-01 -1.15718335e-01 9.14053380e-01 -5.56270361e-01 1.11748636e-01 -3.05811912e-01 -8.19417894e-01 -4.16151136e-01 3.27197552e-01 5.67846477e-01 -1.97604775e-01 5.05496502e-01 -8.65628570e-02 1.71997711e-01 3.52911174e-01 -8.96749318e-01 1.43363607e+00 -7.91826725e-01 1.05409801e+00 -3.97124261e-01 -9.76897180e-01 9.01114583e-01 5.40833533e-01 1.06631599e-01 -4.67783421e-01 4.36617136e-01 -2.23765656e-01 -2.81365871e-01 -5.33568501e-01 1.30441725e+00 3.09521228e-01 -1.53092414e-01 4.31076914e-01 8.42821896e-01 2.11774513e-01 3.23941410e-01 5.84807396e-01 9.18622553e-01 -1.59916654e-01 4.13337469e-01 -3.86419177e-01 4.91023719e-01 -2.23542854e-01 2.14414746e-01 8.47641349e-01 -2.62006428e-02 5.80169082e-01 2.76939869e-01 -4.42797393e-01 -1.03392673e+00 -7.05924690e-01 -2.72959441e-01 1.26647782e+00 -3.21385235e-01 -9.05865073e-01 -4.13552225e-01 -9.10676181e-01 -7.51082832e-03 1.05197537e+00 -1.10762513e+00 -1.51633933e-01 -4.85544384e-01 -6.47121847e-01 3.67663771e-01 8.42001617e-01 1.70414075e-02 -1.11092830e+00 -2.01762214e-01 6.63588822e-01 -8.41295645e-02 -6.85453832e-01 -5.10826409e-01 6.73684001e-01 -8.17564309e-01 -6.56801760e-01 -9.23432648e-01 -8.70126724e-01 3.08517694e-01 4.25044268e-01 1.47519374e+00 2.08577383e-02 -2.24028140e-01 3.71540010e-01 -7.45217562e-01 -7.75171340e-01 2.08164856e-01 3.61712039e-01 -9.62014645e-02 -4.25120324e-01 1.26200330e+00 -3.84709686e-01 -6.98683679e-01 -2.21007317e-01 -1.05236793e+00 -6.64304554e-01 2.73384064e-01 1.10241044e+00 3.89337361e-01 1.26447633e-01 4.53384727e-01 -9.53673482e-01 1.28925180e+00 -9.70651448e-01 -7.15551227e-02 2.85748214e-01 -1.19278145e+00 4.41538781e-01 6.10765815e-01 -4.29916203e-01 -1.19339645e+00 -5.33678174e-01 -2.52929002e-01 -6.32113218e-02 2.64546555e-03 7.24689782e-01 2.61078686e-01 6.57714665e-01 7.56655812e-01 2.17649460e-01 -7.05952346e-01 -6.45383477e-01 7.89210320e-01 7.13336110e-01 -1.03368729e-01 -4.62449282e-01 2.74924904e-01 2.22090110e-01 -6.15744293e-01 -5.99882662e-01 -7.70333707e-01 -6.15824699e-01 -2.50719339e-01 5.26588061e-04 9.81031954e-01 -7.89709508e-01 -1.67120188e-01 -9.21855867e-02 -1.22457397e+00 2.00127289e-01 -3.18485826e-01 6.85889482e-01 1.28006235e-01 1.32981479e-01 -5.93750000e-01 -6.43365264e-01 -9.37386394e-01 -7.76142180e-01 6.48812592e-01 2.86670476e-01 -1.39150158e-01 -1.46203911e+00 4.31379288e-01 -4.60100174e-01 1.01990390e+00 -2.42958099e-01 8.92415464e-01 -1.05049706e+00 1.12934567e-01 -6.25762284e-01 -2.13322788e-01 4.05132294e-01 1.15681015e-01 1.92336887e-01 -1.24693537e+00 1.48939252e-01 -4.48044479e-01 -1.10587612e-01 1.42994225e+00 5.80664217e-01 1.41746151e+00 -4.18454081e-01 -5.56449115e-01 2.71475047e-01 1.70342255e+00 3.55544716e-01 5.31480730e-01 4.35558021e-01 4.05659050e-01 4.15493339e-01 2.00493574e-01 5.87171733e-01 5.48785269e-01 4.70978707e-01 2.84260243e-01 2.56826282e-01 -2.87651181e-01 -2.43313819e-01 1.30079553e-01 1.30196059e+00 -1.44968644e-01 -8.63672614e-01 -7.90371060e-01 9.90789473e-01 -1.69665480e+00 -1.13667071e+00 -1.99100137e-01 1.49831295e+00 9.48666513e-01 3.22102219e-01 -4.39161986e-01 -5.94664663e-02 5.30578673e-01 3.82341176e-01 -5.09757251e-02 -1.01726246e+00 -1.90428525e-01 4.43553984e-01 3.71663809e-01 5.69177806e-01 -1.03738987e+00 1.11863911e+00 7.28937531e+00 9.61874485e-01 -8.53060722e-01 2.75589198e-01 2.19725981e-01 -1.39515162e-01 -8.17045152e-01 -2.10593373e-01 -1.04494286e+00 4.93117005e-01 1.11043036e+00 -3.77159595e-01 -1.74447551e-01 1.44615591e+00 -2.77092278e-01 3.52674663e-01 -5.84973752e-01 5.09507716e-01 4.10105258e-01 -1.52517450e+00 5.19954681e-01 1.29389778e-01 8.35191965e-01 2.18514800e-01 8.12803954e-02 8.17583501e-01 8.45133245e-01 -7.19580650e-01 2.40874559e-01 6.43951178e-01 4.02386665e-01 -9.64819491e-01 1.12358463e+00 -1.75150171e-01 -1.04856086e+00 -2.22862586e-01 -1.14959276e+00 7.09232017e-02 1.46096751e-01 9.68472421e-01 -8.63719344e-01 3.57544661e-01 6.00279272e-01 5.85271895e-01 -4.98209298e-01 7.76593924e-01 -2.75714576e-01 8.69099677e-01 2.35430032e-01 -5.89231074e-01 7.77562797e-01 1.93935379e-01 3.37683320e-01 2.02061915e+00 5.04970491e-01 -2.61860304e-02 -1.82688415e-01 5.33207715e-01 -2.54247159e-01 4.67467129e-01 -6.99239850e-01 -1.19421162e-01 4.61334050e-01 1.45520043e+00 -4.42885935e-01 -8.64606082e-01 -4.41683054e-01 6.71659946e-01 3.89284939e-01 2.92992502e-01 -5.29976189e-01 -9.69796181e-01 1.00371277e+00 -3.52207333e-01 6.80044591e-01 -1.40383169e-01 -2.14848280e-01 -1.27100945e+00 -3.98973942e-01 -2.43048459e-01 4.64012414e-01 -8.43683302e-01 -1.30415249e+00 1.05438972e+00 -7.01980619e-03 -1.14450788e+00 -2.83935487e-01 -8.89315546e-01 -8.76521349e-01 7.60268867e-01 -1.75117743e+00 -6.95701838e-01 5.58714159e-02 4.35250014e-01 8.11766863e-01 -4.42497730e-01 1.15952373e+00 3.31623077e-01 -3.80631655e-01 8.38008761e-01 5.28338909e-01 1.30185977e-01 7.95303941e-01 -1.72120857e+00 5.97189307e-01 5.24203539e-01 2.43193075e-01 1.08157039e+00 6.12566769e-01 -5.37786543e-01 -1.14452147e+00 -1.06505883e+00 1.68139398e+00 -4.25268263e-01 6.59166753e-01 -2.66068250e-01 -7.86530495e-01 5.09838283e-01 1.08235860e+00 -3.12701464e-01 1.09117103e+00 5.12197793e-01 -6.63006425e-01 8.76649395e-02 -9.70995605e-01 6.71414971e-01 7.20403254e-01 -3.00058752e-01 -1.18120968e+00 3.78466100e-01 1.19252384e+00 4.12505120e-01 -7.67617285e-01 -8.79199281e-02 6.16541207e-01 -4.59320426e-01 1.13901007e+00 -8.83700490e-01 8.77245128e-01 8.03079754e-02 -3.76782387e-01 -1.75114405e+00 -9.92480516e-01 6.18503466e-02 -6.89677835e-01 1.34880924e+00 7.31661201e-01 -4.36287105e-01 4.26382780e-01 3.42554331e-01 -2.19934404e-01 -5.09687781e-01 -3.77986878e-01 -8.25306416e-01 4.70318198e-01 -6.73108697e-01 9.41123724e-01 1.10100722e+00 3.32271099e-01 2.81662375e-01 -4.54690039e-01 -3.98893565e-01 1.24585718e-01 -3.37103248e-01 1.35376796e-01 -1.27072322e+00 -8.02043453e-02 -7.06658125e-01 -2.90494859e-01 -7.60584831e-01 1.19499594e-01 -1.15559137e+00 -9.88561139e-02 -1.86570692e+00 1.27423987e-01 1.13813125e-01 -1.03361583e+00 5.50752223e-01 -4.74014282e-01 2.27225482e-01 3.83205488e-02 -1.88606113e-01 -4.93582636e-01 5.43408811e-01 9.22093451e-01 -6.04297876e-01 1.76546618e-01 -5.40110588e-01 -9.86406982e-01 5.67818165e-01 9.04804766e-01 -5.91380835e-01 -4.84852344e-01 -9.04654384e-01 5.99540055e-01 -5.57907879e-01 -4.94268775e-01 -8.26630473e-01 4.89920974e-02 -3.94831635e-02 7.21441269e-01 -7.03318536e-01 2.78731883e-01 -9.40691769e-01 -3.93323094e-01 2.14278564e-01 -6.59075737e-01 -2.05370486e-02 9.60828215e-02 5.85156620e-01 -3.73882055e-01 -8.17408562e-01 3.85494307e-02 -4.60162580e-01 -8.73773396e-01 1.39262050e-01 -1.07939315e+00 3.75826918e-02 4.90328819e-01 1.86519444e-01 -3.74935716e-01 -4.69698519e-01 -5.72501659e-01 9.75820795e-02 -2.51262248e-01 7.98344016e-01 5.70897162e-01 -1.59022558e+00 -6.47046208e-01 -1.18037365e-01 5.53235471e-01 -6.49498522e-01 8.30984116e-02 -7.76206329e-02 -5.37531152e-02 1.04436231e+00 1.06210761e-01 6.89719990e-02 -8.30364764e-01 8.78386736e-01 -9.25143287e-02 -6.89973950e-01 -6.27565920e-01 1.12512398e+00 1.36626273e-01 -3.82212371e-01 4.05683041e-01 -4.31900203e-01 -1.15557718e+00 6.48285389e-01 1.22839522e+00 1.64183587e-01 2.40664929e-01 -2.13610172e-01 -1.68935984e-01 4.32471991e-01 -6.49884284e-01 -7.72484466e-02 1.38711846e+00 4.55050990e-02 2.29133606e-01 1.61400571e-01 1.48260915e+00 -1.15488298e-01 -7.13301599e-01 -4.38993335e-01 1.30526200e-01 -2.75669396e-01 5.46948552e-01 -1.00787580e+00 -1.31212032e+00 9.73163486e-01 6.04860008e-01 4.92037803e-01 8.58071446e-01 -1.63581505e-01 1.03144491e+00 6.96456730e-01 1.37851864e-01 -1.59817708e+00 -5.47945201e-02 1.00960588e+00 7.77041435e-01 -9.71482277e-01 -2.99902027e-03 4.67284620e-01 -8.19082201e-01 1.42770004e+00 5.97270370e-01 -2.72480875e-01 1.30281162e+00 1.68248728e-01 1.30659536e-01 -4.38970357e-01 -1.19818723e+00 -4.60165709e-01 4.60996002e-01 7.74199069e-01 8.12607288e-01 1.37023583e-01 -9.54875112e-01 1.08297384e+00 -1.83824509e-01 -3.69839460e-01 6.07481956e-01 8.92057478e-01 -7.59285331e-01 -1.22943878e+00 1.06903063e-02 5.80023468e-01 -7.70839810e-01 -6.00407660e-01 -3.92228633e-01 3.84817809e-01 1.33323938e-01 8.98194551e-01 1.63878128e-01 -5.55635095e-01 1.79981664e-01 6.15628898e-01 -1.66836947e-01 -5.79782367e-01 -9.13741350e-01 -8.05160105e-02 9.07434803e-03 -3.64125907e-01 -1.14151366e-01 -3.17445129e-01 -9.76068556e-01 2.30470058e-02 -5.68664193e-01 4.30271596e-01 1.02798772e+00 6.80145264e-01 5.23050427e-01 8.22503865e-01 6.57579482e-01 -6.54493034e-01 -1.69074640e-01 -1.39668930e+00 -5.54273427e-01 2.81887889e-01 5.43073229e-02 -5.98082244e-01 -3.88307065e-01 -1.08986035e-01]
[10.5711669921875, 8.479058265686035]
f74dc857-d5cf-439d-9759-de683ad73799
softflow-probabilistic-framework-for
2006.04604
null
https://arxiv.org/abs/2006.04604v4
https://arxiv.org/pdf/2006.04604v4.pdf
SoftFlow: Probabilistic Framework for Normalizing Flow on Manifolds
Flow-based generative models are composed of invertible transformations between two random variables of the same dimension. Therefore, flow-based models cannot be adequately trained if the dimension of the data distribution does not match that of the underlying target distribution. In this paper, we propose SoftFlow, a probabilistic framework for training normalizing flows on manifolds. To sidestep the dimension mismatch problem, SoftFlow estimates a conditional distribution of the perturbed input data instead of learning the data distribution directly. We experimentally show that SoftFlow can capture the innate structure of the manifold data and generate high-quality samples unlike the conventional flow-based models. Furthermore, we apply the proposed framework to 3D point clouds to alleviate the difficulty of forming thin structures for flow-based models. The proposed model for 3D point clouds, namely SoftPointFlow, can estimate the distribution of various shapes more accurately and achieves state-of-the-art performance in point cloud generation.
['Joun Yeop Lee', 'Woo Hyun Kang', 'Nam Soo Kim', 'Hyeongju Kim', 'Hyeonseung Lee']
2020-06-08
null
http://proceedings.neurips.cc/paper/2020/hash/bdbca288fee7f92f2bfa9f7012727740-Abstract.html
http://proceedings.neurips.cc/paper/2020/file/bdbca288fee7f92f2bfa9f7012727740-Paper.pdf
neurips-2020-12
['point-cloud-generation']
['computer-vision']
[-3.39986831e-01 -2.35503484e-02 9.60380863e-03 -8.24259594e-02 -5.01892090e-01 -7.81999290e-01 7.20767915e-01 -4.28069413e-01 1.46600097e-01 7.43049145e-01 1.60551801e-01 -2.64311492e-01 1.06463172e-01 -1.16580367e+00 -9.47501302e-01 -6.98997736e-01 2.11919382e-01 8.09251845e-01 3.22562009e-02 1.68085247e-01 2.98194766e-01 7.94353902e-01 -1.48869646e+00 -2.24777877e-01 1.25337458e+00 5.33590257e-01 2.56647598e-02 9.05883372e-01 -6.38440013e-01 3.72148812e-01 -6.48490965e-01 -2.94216394e-01 4.00984168e-01 -3.97811204e-01 -4.25296694e-01 1.59992456e-01 8.13811779e-01 -3.62292767e-01 -5.65412164e-01 1.09495950e+00 2.03416497e-01 1.10045262e-01 1.33769417e+00 -1.64158678e+00 -9.40121114e-01 1.30841509e-01 -4.97633010e-01 -1.27976285e-02 7.32915774e-02 3.44695449e-01 6.86402559e-01 -1.09867477e+00 6.54269874e-01 1.52735150e+00 3.24407607e-01 5.87471485e-01 -1.38869655e+00 -7.57513463e-01 4.96041626e-02 -3.44908446e-01 -1.30203378e+00 -1.80640846e-01 8.81143391e-01 -8.57346237e-01 4.08488810e-01 -1.84255932e-02 7.72348106e-01 1.03365111e+00 7.70966932e-02 6.25692785e-01 6.79176033e-01 4.30779010e-02 2.98104674e-01 2.48637810e-01 -3.16886663e-01 5.88710487e-01 4.37943369e-01 2.55678833e-01 -3.92129779e-01 -2.86202848e-01 1.17469335e+00 6.49397522e-02 -3.22712272e-01 -6.44574523e-01 -1.22391915e+00 7.97091842e-01 3.32619071e-01 5.73507287e-02 -1.98315069e-01 3.47990334e-01 -1.53016880e-01 -9.61709842e-02 4.05611366e-01 2.59730726e-01 -1.15086250e-01 -3.05485994e-01 -1.04980552e+00 6.11983538e-01 7.93213844e-01 1.46826792e+00 1.02536690e+00 3.37796956e-01 -1.96341693e-01 2.74343461e-01 8.12922359e-01 1.09187710e+00 1.07804768e-01 -1.16082644e+00 6.56789482e-01 5.88941514e-01 2.87779450e-01 -1.00947940e+00 1.08352907e-01 -2.71624088e-01 -9.50741708e-01 4.19303089e-01 6.58066332e-01 -2.23387450e-01 -1.08313179e+00 1.64591753e+00 4.98312682e-01 6.97018087e-01 4.03233953e-02 8.32917213e-01 6.23923957e-01 9.58654046e-01 -7.82370344e-02 1.63355246e-01 6.31417751e-01 -5.90409636e-01 -3.95119607e-01 1.44705445e-01 2.17404708e-01 -8.58944237e-01 1.14632440e+00 -1.78451538e-02 -1.02221918e+00 -6.34969294e-01 -8.83224308e-01 4.65164892e-02 -1.21200923e-02 -6.26635551e-02 4.31434035e-01 7.07026839e-01 -8.21084678e-01 7.30825365e-01 -1.05494702e+00 -3.17820907e-02 6.38019979e-01 -3.05831432e-02 -2.62538522e-01 -8.45549279e-04 -7.35083222e-01 4.71409261e-01 -6.14822283e-02 -1.81040943e-01 -9.88945663e-01 -1.37459254e+00 -8.32588375e-01 8.39781314e-02 -3.39833677e-01 -1.07415187e+00 8.03762555e-01 -2.77615637e-01 -1.55644548e+00 5.94670475e-01 -3.78903598e-01 -1.02012083e-01 7.71268964e-01 -2.16809154e-01 3.00585460e-02 7.89012387e-02 1.14207007e-01 8.16707671e-01 1.31117058e+00 -1.39949620e+00 -2.86268622e-01 -1.45866334e-01 -3.19825232e-01 -3.03204544e-02 -6.63529858e-02 -7.28863418e-01 -1.38084114e-01 -5.38937330e-01 9.14907679e-02 -9.12283123e-01 -1.16459541e-01 3.58266264e-01 -6.22426391e-01 -3.81725319e-02 1.13271701e+00 -4.36157659e-02 8.17665577e-01 -2.10957837e+00 1.82351544e-01 2.26054743e-01 1.94085866e-01 1.93826109e-01 -6.74675256e-02 2.41110086e-01 2.48446111e-02 4.93737280e-01 -3.97204608e-01 -4.57443327e-01 2.02388555e-01 3.07458431e-01 -7.79496074e-01 5.59218526e-01 5.74241698e-01 7.45502710e-01 -9.44397151e-01 -4.53276664e-01 5.49882293e-01 6.63609266e-01 -7.18655586e-01 3.99139166e-01 -3.01933497e-01 7.73493826e-01 -3.74573559e-01 3.81600887e-01 1.20732963e+00 -1.29095957e-01 -3.94497156e-01 -1.45697802e-01 1.44566074e-01 -4.59340774e-02 -1.37342441e+00 1.71319985e+00 -3.87489557e-01 5.06639302e-01 -9.00526345e-02 -2.89617747e-01 1.24008954e+00 1.39670089e-01 7.00828373e-01 1.14796758e-01 5.40088490e-02 4.47809845e-02 -8.87458473e-02 -2.52395391e-01 4.67953771e-01 -4.37970906e-01 2.10365877e-01 5.09234965e-01 2.82765865e-01 -7.72533834e-01 6.05493598e-02 2.17380568e-01 7.56077528e-01 3.53714794e-01 -3.27189744e-01 -2.31822565e-01 5.33120096e-01 -2.33734503e-01 5.80281854e-01 4.50457066e-01 -8.66380259e-02 9.31259334e-01 4.57984477e-01 -2.51923263e-01 -1.35729265e+00 -1.73518610e+00 -2.66204268e-01 3.33926715e-02 2.47147903e-01 -4.23853159e-01 -6.66759193e-01 -8.48019540e-01 4.89095092e-01 7.55139768e-01 -4.70488042e-01 -9.64216441e-02 -4.35671419e-01 -6.49208009e-01 5.22130907e-01 3.78742963e-01 3.24670792e-01 -6.46029770e-01 -1.01690345e-01 1.82311013e-02 -9.52951685e-02 -1.08040452e+00 -7.26985753e-01 -5.63059449e-01 -1.17437601e+00 -1.14177191e+00 -7.60518849e-01 -4.35740322e-01 9.48679745e-01 1.67949215e-01 1.11240232e+00 -1.89346597e-01 -9.66374204e-03 3.01650137e-01 8.68564993e-02 -4.25745696e-01 -7.38763630e-01 9.51480642e-02 -1.33736311e-02 1.40816957e-01 2.27777481e-01 -8.85972798e-01 -5.86057901e-01 3.99373025e-01 -1.01603055e+00 -2.04985186e-01 1.95643321e-01 5.40391982e-01 6.40006125e-01 6.51630163e-02 5.06398559e-01 -6.43231988e-01 4.72393125e-01 -5.43615341e-01 -5.88138342e-01 -1.11737832e-01 -3.44025970e-01 4.77597326e-01 6.21257305e-01 -5.74818134e-01 -1.04468906e+00 7.26731345e-02 -7.50254840e-02 -1.12305439e+00 -3.62184376e-01 -1.58761635e-01 -2.08795264e-01 3.63072827e-02 6.86108470e-01 1.29143983e-01 1.89295873e-01 -3.74350697e-01 7.39852488e-01 3.39756221e-01 5.76340795e-01 -8.50251138e-01 1.58549595e+00 7.40196228e-01 4.20221925e-01 -8.47986937e-01 -3.88212800e-01 -2.95913994e-01 -7.62419522e-01 -2.15614989e-01 8.37014079e-01 -8.59018266e-01 -4.93768722e-01 5.88084757e-01 -1.28451538e+00 -1.91040114e-01 -6.66444123e-01 5.17175376e-01 -7.32729733e-01 2.22885713e-01 -4.37445849e-01 -7.78422713e-01 -1.52137816e-01 -1.07875824e+00 1.21309483e+00 4.42020833e-01 -3.52759287e-02 -1.16568410e+00 4.54536349e-01 -1.95508838e-01 3.11725497e-01 4.24322039e-01 8.75301301e-01 -4.74199327e-03 -1.02755201e+00 -1.68030500e-01 -8.85386690e-02 3.74792218e-01 3.91442686e-01 7.37953484e-01 -9.43224311e-01 -1.35386571e-01 -8.12295377e-02 1.51761547e-01 5.60137510e-01 5.24412572e-01 8.60098183e-01 -8.26348513e-02 -2.45774329e-01 9.64310348e-01 1.29830110e+00 -2.53925681e-01 8.38016212e-01 -2.52424568e-01 1.05616081e+00 3.70928675e-01 2.69906402e-01 4.78884786e-01 4.18053269e-01 1.63192481e-01 4.77557272e-01 9.51288715e-02 -2.58189738e-01 -8.87178600e-01 2.21452028e-01 8.12763453e-01 1.07226238e-01 -2.12787628e-01 -8.04473698e-01 5.90744197e-01 -1.51913190e+00 -9.33589220e-01 -3.88882935e-01 2.09468508e+00 5.25360286e-01 1.16720848e-01 9.60740447e-02 -8.83488283e-02 7.75082231e-01 9.98017639e-02 -5.99706531e-01 -6.57759234e-02 2.44933087e-02 1.72311783e-01 4.45193142e-01 6.93528354e-01 -9.09136236e-01 8.64946902e-01 6.60266638e+00 5.44766665e-01 -1.19287121e+00 -1.91787392e-01 3.52978528e-01 -3.79034393e-02 -8.20692778e-01 6.47648126e-02 -9.46143270e-01 6.77545846e-01 6.50482476e-01 -4.91312474e-01 1.38437316e-01 8.93926978e-01 2.39311501e-01 1.89033955e-01 -1.16129887e+00 1.15986168e+00 -1.30993798e-01 -1.51334918e+00 4.11550850e-01 3.68865967e-01 9.57386971e-01 -1.34366527e-01 2.06349224e-01 1.51734054e-01 4.00793940e-01 -1.10180926e+00 6.88856006e-01 9.03642535e-01 7.76914716e-01 -6.96857870e-01 4.11949962e-01 5.26133835e-01 -1.07714200e+00 3.65096956e-01 -5.89948595e-01 2.04291448e-01 4.27005529e-01 8.98060501e-01 -9.72303987e-01 4.55486655e-01 4.97791082e-01 1.02183044e+00 -2.66399950e-01 1.25519454e+00 -3.36316913e-01 6.72116458e-01 -5.40933430e-01 2.61505544e-01 -1.15343183e-02 -5.13681769e-01 1.04285932e+00 9.05636668e-01 7.73756802e-01 -3.51680100e-01 7.57968575e-02 1.68881869e+00 -1.77953780e-01 -1.09292224e-01 -8.97183657e-01 -9.93231833e-02 6.82772875e-01 1.16951609e+00 -4.02684063e-01 -1.53484195e-01 -1.31630495e-01 4.19421792e-01 1.07168488e-01 5.53042650e-01 -8.11678469e-01 -2.74350107e-01 1.18657243e+00 2.66495109e-01 3.02986443e-01 -5.41151702e-01 -5.88458717e-01 -1.42434692e+00 3.05908150e-03 -3.01690966e-01 -1.21126361e-01 -7.47806609e-01 -1.58043659e+00 4.00113404e-01 5.04937321e-02 -1.63324296e+00 -2.97108114e-01 -5.02825379e-01 -1.06899548e+00 1.26596916e+00 -1.41638684e+00 -9.50745463e-01 -4.89915729e-01 5.76518953e-01 1.38731152e-01 -1.41234785e-01 6.01584494e-01 1.45157233e-01 -2.89674610e-01 2.89788932e-01 -2.40232851e-02 1.19059525e-01 6.93373024e-01 -1.43185520e+00 7.26945698e-01 9.78096843e-01 1.88807264e-01 5.25878966e-01 6.77176476e-01 -7.14045107e-01 -1.24423420e+00 -1.32051301e+00 2.73427367e-01 -7.71445751e-01 5.11381507e-01 -3.21633965e-01 -1.04071712e+00 5.11917710e-01 -1.24948643e-01 2.24834651e-01 5.04853666e-01 -4.57059264e-01 -1.67050213e-01 -4.14159968e-02 -1.36033845e+00 3.27007562e-01 9.85762656e-01 -4.19532210e-01 -4.56260145e-01 8.59146416e-02 7.00810194e-01 -4.75668311e-01 -9.78208959e-01 1.66387707e-01 1.93000674e-01 -8.75175893e-01 1.10233235e+00 -6.13532782e-01 5.24492562e-01 -6.52667820e-01 -1.37110218e-01 -1.68102217e+00 -2.74329513e-01 -7.26347744e-01 -5.44537723e-01 1.39885950e+00 2.13569850e-01 -5.16041994e-01 1.28316033e+00 5.30950189e-01 -3.87485586e-02 -3.44023317e-01 -8.47262621e-01 -7.39046693e-01 6.73893869e-01 -4.08609420e-01 1.13322055e+00 7.41506398e-01 -4.85186994e-01 1.11941732e-01 -8.75589103e-02 3.67433131e-01 1.06423712e+00 1.89541236e-01 1.29731083e+00 -1.26570582e+00 -2.21150249e-01 -3.95438671e-01 -6.59738600e-01 -1.22715461e+00 3.19954425e-01 -1.06591856e+00 -2.19976846e-02 -1.42499995e+00 -2.85585016e-01 -7.96308696e-01 3.90909731e-01 -2.68684864e-01 -2.97497362e-01 -2.24738978e-02 3.71593058e-01 2.55819410e-01 1.94165438e-01 1.04166985e+00 1.79915297e+00 -1.77697793e-01 -3.48359495e-01 1.58782244e-01 -5.78390419e-01 6.13078415e-01 6.23897254e-01 -4.88937020e-01 -7.86015689e-01 -4.22934383e-01 2.83548962e-02 -7.09675997e-02 3.89080077e-01 -1.10517621e+00 7.10868314e-02 -4.34222847e-01 5.83818018e-01 -9.25712287e-01 2.00556487e-01 -8.66928041e-01 2.44946361e-01 5.47703281e-02 1.82338152e-02 -3.64372618e-02 1.42810374e-01 6.80874765e-01 -1.76497564e-01 -4.83267754e-02 8.82762015e-01 3.00126579e-02 1.01836085e-01 9.25845265e-01 -1.13242343e-01 4.74231482e-01 9.92457151e-01 -4.74446118e-02 -4.15469229e-01 -3.93804014e-01 -5.42192519e-01 1.45107433e-01 7.60871232e-01 4.24650967e-01 8.31925809e-01 -1.67998588e+00 -7.17707515e-01 6.48021102e-01 -1.18444756e-01 6.78670108e-01 1.82901785e-01 3.77610624e-01 -7.10193396e-01 6.29992411e-02 -1.89201429e-01 -1.03409350e+00 -5.35275221e-01 4.34266239e-01 4.91778016e-01 4.51486278e-03 -7.30247319e-01 6.28233373e-01 3.16667527e-01 -8.30757797e-01 -1.25770971e-01 -5.75542331e-01 2.22037539e-01 -2.38806769e-01 4.34358805e-01 4.21500742e-01 -2.73925543e-01 -6.33932173e-01 -1.51869670e-01 7.25556076e-01 3.25583667e-01 -1.46674648e-01 1.10213530e+00 3.36767919e-02 1.21151961e-01 4.41489428e-01 1.28013933e+00 6.10141046e-02 -1.79049003e+00 8.01288635e-02 -5.72781980e-01 -9.37494099e-01 -4.46274951e-02 -1.53883308e-01 -1.30547476e+00 1.22063887e+00 2.75295973e-01 9.79636908e-02 5.18524885e-01 -9.67454910e-02 7.93479919e-01 -1.65888861e-01 2.49773920e-01 -5.83528936e-01 1.75117534e-02 3.69800717e-01 7.97927856e-01 -9.36733544e-01 -1.72091663e-01 -6.67799413e-01 -4.14187610e-01 1.21935105e+00 6.69175327e-01 -6.56955898e-01 1.03865385e+00 2.59245843e-01 -3.98172252e-02 -3.91230620e-02 -6.40658140e-01 4.86505777e-02 3.82311612e-01 1.10664821e+00 3.33760157e-02 6.85892180e-02 5.21950722e-01 2.04975381e-02 -5.94670415e-01 3.82082090e-02 6.78281486e-01 5.67728758e-01 -1.99694201e-01 -1.06731546e+00 -7.14409053e-01 3.40616524e-01 5.73508851e-02 1.70259356e-01 -3.17030698e-02 6.09126985e-01 -7.08495229e-02 5.81051171e-01 5.36538482e-01 -3.26132357e-01 4.22748089e-01 8.69894400e-02 5.85110188e-01 -4.21402961e-01 1.07507847e-01 -2.14183889e-02 -5.00762343e-01 -3.89922112e-01 -2.65957654e-01 -7.84049988e-01 -1.21009064e+00 -6.20452046e-01 -2.75897211e-04 1.20430745e-01 6.12031996e-01 6.49177551e-01 5.25752366e-01 3.00354660e-01 8.02646220e-01 -1.06284666e+00 -5.61350822e-01 -8.83690596e-01 -6.42899036e-01 6.60291791e-01 4.35981840e-01 -9.23675835e-01 -7.70181775e-01 1.98237121e-01]
[8.883990287780762, -3.645747661590576]
adf78b9d-49c4-49fa-be99-22f3ea0acd11
tstnn-two-stage-transformer-based-neural
2103.09963
null
https://arxiv.org/abs/2103.09963v1
https://arxiv.org/pdf/2103.09963v1.pdf
TSTNN: Two-stage Transformer based Neural Network for Speech Enhancement in the Time Domain
In this paper, we propose a transformer-based architecture, called two-stage transformer neural network (TSTNN) for end-to-end speech denoising in the time domain. The proposed model is composed of an encoder, a two-stage transformer module (TSTM), a masking module and a decoder. The encoder maps input noisy speech into feature representation. The TSTM exploits four stacked two-stage transformer blocks to efficiently extract local and global information from the encoder output stage by stage. The masking module creates a mask which will be multiplied with the encoder output. Finally, the decoder uses the masked encoder feature to reconstruct the enhanced speech. Experimental results on the benchmark dataset show that the TSTNN outperforms most state-of-the-art models in time or frequency domain while having significantly lower model complexity.
['Wei-Ping Zhu', 'Bengbeng He', 'Kai Wang']
2021-03-18
null
null
null
null
['speech-denoising']
['speech']
[ 4.37343240e-01 4.98633943e-02 2.69716024e-01 -4.03529704e-01 -9.26363826e-01 1.54623250e-02 3.00407439e-01 -4.94019806e-01 -3.09964061e-01 1.57209173e-01 4.23165739e-01 -4.16064352e-01 2.72437632e-01 -5.31118929e-01 -4.83454704e-01 -8.15316975e-01 1.46863252e-01 -2.46394783e-01 2.82833338e-01 -2.93949783e-01 -5.79807907e-02 -7.86551982e-02 -1.30796278e+00 8.85308623e-01 7.99540341e-01 1.38198555e+00 6.22336507e-01 7.34270096e-01 -8.38615838e-03 1.15566993e+00 -6.44334614e-01 -2.60537654e-01 9.86148324e-03 -5.71268559e-01 -2.61916965e-01 5.29721975e-02 -2.23046660e-01 -2.46305212e-01 -6.93910122e-01 1.11976480e+00 7.22676039e-01 1.30333573e-01 7.17761964e-02 -6.91234350e-01 -6.52229428e-01 8.26488972e-01 -1.30849883e-01 3.40128601e-01 2.85384394e-02 -1.95976540e-01 5.84776878e-01 -1.24303722e+00 2.10083336e-01 1.37893856e+00 8.58456790e-01 4.51480508e-01 -9.36586201e-01 -6.80538118e-01 6.78582117e-02 4.82633293e-01 -1.06599772e+00 -8.28683257e-01 1.00937951e+00 -9.20096599e-03 1.42145419e+00 3.82894754e-01 5.86416423e-01 1.02346516e+00 5.20872355e-01 7.90576220e-01 1.07818317e+00 -2.80825645e-01 9.56413373e-02 -1.19475700e-01 -1.42857373e-01 4.31681633e-01 -6.78043246e-01 4.52630788e-01 -8.31785262e-01 1.25371054e-01 6.02600873e-01 -1.86152041e-01 -2.24996194e-01 3.45122963e-01 -8.23681593e-01 5.44154406e-01 4.79172200e-01 5.19482017e-01 -7.30738103e-01 3.84700508e-03 6.91121578e-01 8.35717261e-01 9.15382087e-01 -3.94065440e-01 -5.57657063e-01 -1.21838033e-01 -1.23678005e+00 -2.69944400e-01 5.37664473e-01 6.32645488e-01 1.77451953e-01 6.98023617e-01 -1.32605016e-01 9.72235262e-01 4.61081207e-01 3.60940814e-01 1.01667309e+00 -5.93750834e-01 7.47090697e-01 1.11744121e-01 -4.15461898e-01 -5.37630677e-01 -2.53989473e-02 -8.59805644e-01 -1.22419846e+00 1.36780635e-01 -3.71229291e-01 -1.39737114e-01 -1.33666348e+00 1.59451962e+00 5.27256243e-02 4.70310062e-01 3.57903272e-01 9.26068425e-01 1.04346645e+00 1.15457571e+00 -1.24940433e-01 -4.04025435e-01 1.30382562e+00 -1.34751582e+00 -1.35567009e+00 -5.14130354e-01 2.24942312e-01 -1.16845167e+00 5.28032899e-01 5.15430331e-01 -1.51923180e+00 -9.00262296e-01 -1.19201970e+00 -2.95561343e-01 -3.05403233e-01 5.14669955e-01 -7.88498521e-02 4.52306002e-01 -1.31070793e+00 7.50328302e-01 -9.46817815e-01 -2.71805618e-02 -7.11191967e-02 4.90691781e-01 -2.90973783e-01 2.91038066e-01 -1.43378103e+00 9.97912109e-01 3.07515889e-01 5.77194333e-01 -1.23667037e+00 -5.46608329e-01 -1.14947104e+00 2.68448323e-01 -1.16223253e-01 -5.34671843e-01 1.65770674e+00 -1.06511617e+00 -2.04257250e+00 3.45198482e-01 -6.74656868e-01 -8.30809236e-01 3.32934082e-01 -1.59578323e-02 -8.27986360e-01 -2.27769483e-02 -8.76596197e-02 1.94003418e-01 1.55732024e+00 -8.81347954e-01 -7.75578141e-01 -2.41440192e-01 -5.68161726e-01 2.21150503e-01 -2.29281962e-01 1.06273562e-01 -3.58902991e-01 -1.19894552e+00 6.18349850e-01 -2.43002862e-01 -2.24560663e-01 -4.88535583e-01 -4.44656938e-01 6.13176525e-02 1.57328558e+00 -1.36706150e+00 1.51567662e+00 -2.44585180e+00 3.81560653e-01 -3.37127335e-02 1.95572916e-02 5.24170697e-01 -2.14605808e-01 6.02519393e-01 -5.45158744e-01 -3.58005941e-01 -2.23527625e-01 -9.71757293e-01 -1.18440285e-01 1.60680473e-01 -6.52176619e-01 3.95372450e-01 2.46479586e-01 6.10392034e-01 -6.16084814e-01 -1.56450301e-01 5.03282726e-01 9.59314466e-01 -2.38293275e-01 3.06402653e-01 1.58955052e-01 3.94616097e-01 -5.90039343e-02 4.37008560e-01 8.59404385e-01 4.31020796e-01 1.08160913e-01 -3.80847484e-01 -4.55948889e-01 9.80242252e-01 -9.08450127e-01 1.59978032e+00 -6.85828447e-01 7.62484610e-01 6.03255033e-01 -1.04493642e+00 1.04667366e+00 1.02268672e+00 6.41885549e-02 -9.60745215e-01 4.13371980e-01 2.75019735e-01 -9.26486924e-02 -4.53635603e-01 3.42292041e-01 -4.55504030e-01 2.11472541e-01 1.67451277e-01 3.60962272e-01 -2.98302323e-02 -1.32993117e-01 -2.66151369e-01 1.05098367e+00 -9.92507041e-02 -2.18603984e-02 8.16598535e-02 8.15842032e-01 -5.28519392e-01 8.57277930e-01 1.04185268e-01 -1.88636795e-01 5.34924090e-01 1.72115609e-01 -2.98661202e-01 -1.15449679e+00 -1.10578084e+00 2.43396342e-01 9.55641568e-01 -2.62152404e-01 -4.71589416e-01 -9.98166621e-01 -2.84302205e-01 -4.50170666e-01 5.13888836e-01 -4.37475920e-01 -4.47500467e-01 -7.79676855e-01 -2.96842188e-01 5.93680561e-01 5.31873703e-01 9.54060495e-01 -1.02699018e+00 -3.10716301e-01 5.22618592e-01 -6.06831670e-01 -1.11633706e+00 -8.75518322e-01 7.02702165e-01 -1.07997513e+00 -2.71310210e-01 -5.84576786e-01 -1.36780250e+00 3.99666578e-01 2.10650072e-01 7.04404950e-01 -1.97501794e-01 3.14065129e-01 -5.08361340e-01 -2.93802887e-01 -1.70997649e-01 -6.12349689e-01 -1.32973149e-01 -8.68509412e-02 3.54496062e-01 8.40319321e-02 -8.83626103e-01 -4.17962492e-01 2.07402751e-01 -7.55360186e-01 3.24277729e-02 7.26590097e-01 1.06466210e+00 6.15637004e-01 4.77665782e-01 5.72420895e-01 -4.21520382e-01 6.60484076e-01 -1.73174888e-01 -5.60386240e-01 -8.65957960e-02 -2.01515660e-01 6.53301924e-02 9.99311507e-01 -4.32657361e-01 -1.27744293e+00 1.83047414e-01 -6.72605395e-01 -6.57898307e-01 4.07724261e-01 6.50724411e-01 -3.08915734e-01 1.03816226e-01 1.95886478e-01 7.00067937e-01 6.80411234e-02 -1.03198922e+00 1.45231381e-01 1.05871892e+00 7.98845291e-01 2.64854252e-01 8.32435489e-01 2.56943196e-01 -5.22292435e-01 -8.18422139e-01 -5.76961279e-01 -2.67518550e-01 -4.77546513e-01 -1.11289106e-01 8.50440145e-01 -1.27198780e+00 -2.75366008e-01 8.10895622e-01 -1.32956445e+00 -1.73562974e-01 -3.82725507e-01 4.97246712e-01 -5.16177893e-01 1.82831243e-01 -1.07077324e+00 -8.15831304e-01 -7.48032212e-01 -1.27090228e+00 1.00489378e+00 2.04754457e-01 2.04616919e-01 -7.84501672e-01 -1.55414090e-01 3.46347123e-01 7.52286673e-01 -2.76863694e-01 6.11893237e-01 -3.64256859e-01 -4.30691540e-01 -6.13957383e-02 2.27454886e-01 8.56761336e-01 1.98966607e-01 -3.40425164e-01 -1.44703138e+00 -3.60433221e-01 8.95614624e-01 2.31485069e-01 1.21102929e+00 5.90206921e-01 1.02871013e+00 -3.56842875e-01 -1.31940171e-01 8.30984294e-01 1.07686388e+00 5.87594807e-01 9.71881926e-01 1.37469009e-01 3.15995753e-01 1.87577605e-01 4.20457214e-01 1.87134460e-01 2.82440454e-01 3.94396961e-01 2.77252853e-01 -4.96619225e-01 -4.54967171e-01 -2.43653670e-01 9.51162517e-01 1.77958369e+00 2.89074540e-01 -2.63297886e-01 -4.12462592e-01 6.65017664e-01 -1.61329186e+00 -1.14426160e+00 7.38315284e-02 1.88741052e+00 8.77763510e-01 2.89851665e-01 -1.97334841e-01 8.49978924e-01 8.44149649e-01 4.16966856e-01 -3.30174088e-01 -8.48629773e-01 -7.84039944e-02 5.40071249e-01 2.55952895e-01 6.88062549e-01 -1.16885900e+00 8.66742909e-01 6.08101892e+00 9.90064859e-01 -1.37767017e+00 3.94544125e-01 4.92309242e-01 1.65863022e-01 4.37535681e-02 -2.32357711e-01 -2.86973983e-01 3.40806752e-01 1.39617753e+00 -8.86577368e-02 6.42172933e-01 5.05884171e-01 7.38090158e-01 4.05346513e-01 -6.60402536e-01 7.55521178e-01 -1.87035844e-01 -9.57966328e-01 -1.20064817e-01 -2.93165326e-01 2.33368054e-01 1.46089599e-01 3.89521033e-01 3.45803708e-01 -9.91797149e-02 -8.51495147e-01 1.04039991e+00 3.56704831e-01 8.91076505e-01 -9.31264937e-01 7.13051856e-01 3.86615604e-01 -1.77101433e+00 -3.47491831e-01 -2.20717743e-01 -7.05993921e-02 3.02541345e-01 6.39827847e-01 -6.61367834e-01 7.12623596e-01 8.79712403e-01 9.20828879e-01 4.24475148e-02 8.69866550e-01 -3.56416672e-01 7.37411737e-01 -1.46382272e-01 5.58865666e-01 3.46490651e-01 -1.38241276e-01 5.83295763e-01 1.24847507e+00 4.98240560e-01 1.99184828e-02 -2.40644544e-01 4.08315539e-01 -7.61855990e-02 -4.08931553e-01 -3.42348605e-01 4.03124280e-02 4.72949654e-01 1.06975055e+00 -1.80412680e-01 -3.68205875e-01 -3.82834435e-01 1.30189419e+00 -5.79173528e-02 4.47411507e-01 -8.08937073e-01 -7.77615666e-01 6.65818572e-01 -2.47161821e-01 9.16078508e-01 -2.95176413e-02 -4.25164729e-01 -9.22338665e-01 3.92426133e-01 -1.02724850e+00 1.21180497e-01 -9.72743273e-01 -9.15940046e-01 1.23480153e+00 -6.85346007e-01 -1.46655893e+00 -3.60985100e-01 -3.52032781e-01 -7.68564999e-01 1.25944340e+00 -1.69876635e+00 -1.25168312e+00 -8.40232521e-02 6.75967693e-01 1.00251615e+00 -1.86218128e-01 7.89559543e-01 6.58719540e-01 -6.53444350e-01 4.76916164e-01 2.14340881e-01 2.23309830e-01 3.40489924e-01 -1.10156608e+00 9.35351074e-01 1.24106121e+00 -2.01853395e-01 3.56019169e-01 6.59806609e-01 -6.41540766e-01 -1.28501189e+00 -1.35897255e+00 1.39891839e+00 3.28477055e-01 4.32942986e-01 -6.10293329e-01 -8.91527355e-01 7.50621259e-01 6.68746889e-01 -5.05483970e-02 2.15914145e-01 -4.46877182e-01 -3.86362046e-01 -4.51816410e-01 -1.17388272e+00 3.44673902e-01 6.58365428e-01 -1.03166306e+00 -7.63225496e-01 -3.70205902e-02 1.02274203e+00 -7.21500099e-01 -8.12969446e-01 2.74782449e-01 3.74075323e-01 -9.91729856e-01 9.74929273e-01 4.76579592e-02 2.09479555e-01 -4.24674541e-01 -3.56855214e-01 -1.64902675e+00 -3.75150681e-01 -9.12436962e-01 -1.02497399e-01 1.25356233e+00 4.73707050e-01 -5.08970797e-01 3.33076060e-01 -1.97121859e-01 -7.87700355e-01 -7.79348195e-01 -1.48285830e+00 -7.55025864e-01 -3.01402360e-01 -4.41761762e-01 5.14575481e-01 5.00464976e-01 -2.79071815e-02 3.63107473e-01 -5.34278035e-01 3.91265512e-01 5.10667622e-01 -4.07674760e-01 -7.52726197e-02 -7.11971104e-01 -7.75774643e-02 -1.87578991e-01 -2.13695556e-01 -1.23253131e+00 1.74451526e-02 -6.55133605e-01 4.19193506e-01 -1.57754540e+00 -1.96696803e-01 2.30050743e-01 -4.54014331e-01 4.40613210e-01 -2.91097201e-02 1.37636513e-01 1.28628975e-02 -2.13621870e-01 2.28891112e-02 8.96012008e-01 1.03703618e+00 -2.63652980e-01 -1.23766728e-01 9.64063853e-02 -3.28621179e-01 6.34051919e-01 8.12586784e-01 -5.21421850e-01 -4.67791915e-01 -8.11090946e-01 -6.04733646e-01 3.68382484e-01 2.83184797e-01 -1.13614094e+00 4.86527085e-01 5.41370690e-01 3.78290474e-01 -1.12608886e+00 8.81767631e-01 -9.76571679e-01 1.25651002e-01 7.81537890e-01 -1.42749131e-01 3.85960132e-01 2.76362211e-01 2.46176839e-01 -9.17221785e-01 2.16508791e-01 1.03011179e+00 1.20652989e-01 -4.06139046e-01 1.89087857e-02 -7.55411208e-01 -4.23536927e-01 5.86331785e-01 -8.49190056e-02 -1.35606170e-01 -3.98266524e-01 -8.86180639e-01 -5.95514588e-02 -3.17373753e-01 4.93612915e-01 1.03425562e+00 -1.41920853e+00 -7.81075537e-01 7.82265544e-01 -6.15626812e-01 -3.70782971e-01 4.07977492e-01 8.36286962e-01 -9.48420167e-02 4.78315592e-01 1.03828616e-01 -3.87916565e-01 -1.52631390e+00 3.84391129e-01 7.25658536e-01 -3.95590544e-01 -6.38312697e-01 1.05881917e+00 -2.90197805e-02 -3.40205997e-01 5.93476534e-01 -7.51342297e-01 -4.84653786e-02 -1.00463420e-01 6.33324206e-01 2.40315408e-01 5.56618750e-01 -8.12193692e-01 -4.01088804e-01 5.11901736e-01 -9.10388157e-02 -5.31453729e-01 1.55959129e+00 -2.24157304e-01 -2.14509964e-01 2.64511466e-01 1.23972905e+00 -3.54719073e-01 -1.05543518e+00 -3.66732687e-01 -2.03934610e-01 -1.02153391e-01 6.31122172e-01 -8.77224684e-01 -1.44856012e+00 1.02623403e+00 7.84696758e-01 2.98900276e-01 2.01459408e+00 -4.61756140e-01 1.43151379e+00 -6.04222529e-02 -1.12378225e-01 -1.18349636e+00 -1.07678816e-01 8.62931848e-01 1.04573154e+00 -6.71988964e-01 -5.12207031e-01 -3.47487301e-01 -3.95377010e-01 1.02492058e+00 2.36663058e-01 -1.17575772e-01 8.71160865e-01 7.83969760e-01 4.81953323e-01 1.29079595e-01 -1.27556908e+00 -1.47491887e-01 1.86783984e-01 6.27629280e-01 3.14937830e-01 -2.15753898e-01 -8.36086273e-02 9.00278091e-01 -4.22933429e-01 1.01726845e-01 4.74840403e-02 8.96376848e-01 -5.09755611e-01 -1.23650944e+00 -5.24425805e-01 8.61245319e-02 -6.19089127e-01 -4.07738596e-01 -2.20985338e-01 1.74282894e-01 2.49050274e-01 1.48346913e+00 7.72673562e-02 -1.02239132e+00 5.96045971e-01 -2.71955528e-03 1.15474641e-01 -2.16497496e-01 -1.34754491e+00 6.88049555e-01 2.03268722e-01 -5.83512247e-01 -9.47918445e-02 -4.65137333e-01 -1.12709498e+00 -1.28128812e-01 -3.70332748e-01 2.83611089e-01 8.00350428e-01 9.15558040e-01 3.75867248e-01 1.16954625e+00 9.97083724e-01 -7.78397918e-01 -5.12683570e-01 -1.31206894e+00 -3.70370835e-01 -1.54171124e-01 1.10377800e+00 -1.32740214e-01 -3.64664733e-01 2.65196413e-01]
[14.855613708496094, 5.9887776374816895]
594ad5da-a29f-48f8-b15e-b39b7ed390a7
a-cnn-toolbox-for-skin-cancer-classification
1908.08187
null
https://arxiv.org/abs/1908.08187v1
https://arxiv.org/pdf/1908.08187v1.pdf
A CNN toolbox for skin cancer classification
We describe a software toolbox for the configuration of deep neural networks in the domain of skin cancer classification. The implemented software architecture allows developers to quickly set up new convolutional neural network (CNN) architectures and hyper-parameter configurations. At the same time, the user interface, manageable as a simple spreadsheet, allows non-technical users to explore different configuration settings that need to be explored when switching to different data sets. In future versions, meta leaning frameworks can be added, or AutoML systems that continuously improve over time. Preliminary results, conducted with two CNNs in the context melanoma detection on dermoscopic images, quantify the impact of image augmentation, image resolution, and rescaling filter on the overall detection performance and training time.
['Fabrizio Nunnari', 'Daniel Sonntag']
2019-08-21
null
null
null
null
['skin-cancer-classification']
['medical']
[ 2.35137537e-01 8.88088569e-02 2.08853520e-02 -3.98342013e-01 -9.08184350e-02 -6.07113302e-01 1.56021476e-01 8.08026195e-02 -7.44282901e-01 1.77283198e-01 -4.19471771e-01 -6.27893984e-01 -1.30014688e-01 -8.33375454e-01 -1.30169451e-01 -4.92172867e-01 3.02991197e-02 5.16876802e-02 3.48134607e-01 -2.74585932e-01 -4.75848513e-03 6.21708810e-01 -1.39703047e+00 3.84244770e-01 6.61563516e-01 7.53303349e-01 1.47620365e-01 9.38939869e-01 -1.79774985e-01 2.96807826e-01 -6.52770460e-01 -3.26868206e-01 3.92595679e-01 1.52403444e-01 -6.43994093e-01 1.21730879e-01 8.43216538e-01 -4.18487608e-01 2.05942169e-02 8.26168835e-01 7.85504103e-01 -4.02054101e-01 2.48701889e-02 -8.72562468e-01 -3.08732063e-01 2.55264580e-01 -2.02214390e-01 3.92793655e-01 -1.21275805e-01 7.17097461e-01 2.17244267e-01 -2.71282554e-01 6.14144444e-01 8.41674030e-01 9.92853761e-01 7.97007561e-01 -1.11769199e+00 -5.38350224e-01 -8.06182697e-02 -7.34955296e-02 -1.16439569e+00 -2.63645560e-01 1.26497388e-01 -4.84945714e-01 1.08528781e+00 6.51819289e-01 8.34083378e-01 1.22826803e+00 3.00388694e-01 1.11462846e-01 9.77514684e-01 -5.70688426e-01 2.71288812e-01 5.78134596e-01 1.85455456e-01 8.09872687e-01 5.08494794e-01 -4.44693454e-02 -4.06017490e-02 -3.80065129e-03 1.09139872e+00 -2.89387643e-01 -4.52100076e-02 -3.03659439e-02 -6.36092305e-01 6.68328762e-01 3.46051097e-01 3.76145333e-01 -3.10394377e-01 2.47434005e-02 6.02178276e-01 3.68028462e-01 3.27825367e-01 7.23910928e-01 -4.95484591e-01 3.82517874e-02 -6.59027636e-01 1.29284799e-01 8.60333323e-01 7.04544902e-01 3.71811450e-01 -1.21711984e-01 -1.47909284e-01 8.15850079e-01 -1.57150343e-01 -2.50545979e-01 4.78935450e-01 -7.64618993e-01 2.00841092e-02 9.46565747e-01 -1.87924027e-01 -4.34263080e-01 -8.92593265e-01 -7.00427771e-01 -6.91236198e-01 6.63667142e-01 4.66303021e-01 -5.81823230e-01 -1.28104115e+00 9.94252980e-01 2.05815926e-01 -4.02575135e-01 -3.07911098e-01 7.81818449e-01 1.07735884e+00 -2.32925728e-01 2.77049959e-01 4.08535659e-01 1.49428904e+00 -6.66427076e-01 -3.59396040e-01 -1.92674726e-01 7.38933802e-01 -6.69923902e-01 1.08656645e+00 5.64721763e-01 -1.02840936e+00 -4.97319818e-01 -1.11070812e+00 1.07792784e-02 -7.38041997e-01 6.35048151e-01 7.89998114e-01 1.09208894e+00 -1.34511924e+00 5.90564072e-01 -1.08089721e+00 -9.61731374e-01 3.95422161e-01 7.33903289e-01 -3.75222236e-01 1.40400320e-01 -1.05929291e+00 9.88763392e-01 6.06694400e-01 1.54731914e-01 -4.23758239e-01 -9.23509836e-01 -4.88662302e-01 -3.22152711e-02 2.34415680e-01 -9.24518585e-01 1.15755975e+00 -1.26613152e+00 -1.33041930e+00 9.79437709e-01 4.15670931e-01 -3.83427113e-01 6.24172986e-01 1.10737190e-01 -5.38854301e-01 -3.35317776e-02 -5.13529480e-01 8.36328626e-01 5.77106833e-01 -5.84677100e-01 -5.63806236e-01 -2.32995033e-01 4.04958546e-01 4.51350696e-02 -6.17519319e-01 1.05294667e-01 -7.06991613e-01 -3.35564107e-01 -5.47093391e-01 -8.59759450e-01 -4.53300178e-01 3.95668328e-01 -2.70603985e-01 3.26668143e-01 5.39212644e-01 -6.08271420e-01 1.07584095e+00 -2.15972257e+00 -1.90656155e-01 2.97985882e-01 1.89951032e-01 7.09761560e-01 -3.56426150e-01 2.12607101e-01 -4.92889702e-01 3.46322447e-01 2.24634051e-01 4.16925997e-02 -4.55444962e-01 -1.46373967e-03 7.66300321e-01 9.29834470e-02 3.19723874e-01 6.52972877e-01 -4.86612678e-01 -3.43740851e-01 5.86322784e-01 5.93317568e-01 -2.14220002e-01 -1.29930228e-01 -7.99802989e-02 1.79980740e-01 -5.78227974e-02 6.36237502e-01 7.68520892e-01 -4.05382425e-01 3.46754968e-01 -3.28023821e-01 -2.02193663e-01 -1.13789491e-01 -1.12851298e+00 1.52448130e+00 -6.03138268e-01 8.67195308e-01 2.16215193e-01 -3.49947780e-01 6.30077600e-01 2.03536779e-01 1.52793482e-01 -2.91840822e-01 3.95894289e-01 -1.42730117e-01 2.03348354e-01 -8.55767548e-01 2.78509349e-01 3.05431604e-01 6.35778785e-01 1.02972172e-01 3.27355355e-01 3.68468732e-01 3.71615738e-01 -3.05255353e-01 1.19187427e+00 -2.07849547e-01 3.24862540e-01 -1.27870440e-01 2.02042803e-01 3.01028281e-01 -4.03623581e-02 5.01444697e-01 4.90749534e-03 4.95677918e-01 7.43331432e-01 -7.57693589e-01 -1.17366862e+00 -7.77546287e-01 -4.21318501e-01 1.10700858e+00 -4.96638298e-01 -3.21862280e-01 -1.14787948e+00 -5.32191277e-01 -2.43724182e-01 2.60087997e-01 -1.12445867e+00 1.36563525e-01 -2.76012331e-01 -9.56750989e-01 6.84704185e-01 5.09857357e-01 5.28559566e-01 -9.18998361e-01 -9.24020052e-01 -5.76535799e-02 6.28363192e-01 -8.07558477e-01 -7.09277987e-02 2.22004160e-01 -9.52778637e-01 -1.22184598e+00 -6.40114009e-01 -7.86032021e-01 9.38038766e-01 -1.08169518e-01 7.25295305e-01 4.67089504e-01 -1.05211353e+00 3.84667635e-01 -2.16779515e-01 -5.15467107e-01 -4.84816015e-01 5.88597775e-01 -5.65430582e-01 -3.91191006e-01 3.73239160e-01 -1.41945153e-01 -6.36005282e-01 2.61935834e-02 -1.16930783e+00 2.43052334e-01 5.37244499e-01 6.79823935e-01 3.29731584e-01 -3.55790742e-02 6.48965761e-02 -1.17122793e+00 9.16909039e-01 -2.35911116e-01 -7.28862584e-01 4.55688268e-01 -6.99903131e-01 -2.25942165e-01 4.10596818e-01 -4.88943309e-01 -7.72472084e-01 2.80083150e-01 -2.88808703e-01 -3.73416156e-01 -6.40952110e-01 5.13143241e-01 3.81731391e-01 -5.95346034e-01 1.15486324e+00 -2.95913845e-01 5.97821236e-01 -4.50286418e-01 1.22852236e-01 6.57696605e-01 2.80270994e-01 1.70965746e-01 3.15319031e-01 2.41032898e-01 -8.54096413e-02 -8.62145960e-01 -1.63122505e-01 -2.48515338e-01 -7.41766512e-01 -2.32035846e-01 7.30413973e-01 -7.22316444e-01 -6.10437512e-01 7.08250105e-01 -8.75519335e-01 -4.78311837e-01 3.73914428e-02 7.05250651e-02 1.50279209e-01 2.81248987e-02 -4.08932179e-01 -3.42907965e-01 -5.59228182e-01 -1.23662126e+00 4.42263871e-01 6.30529165e-01 -5.61740458e-01 -1.38940763e+00 -3.74133550e-02 9.49151665e-02 7.99174368e-01 5.73421121e-01 8.39338660e-01 -6.33567631e-01 -3.17360997e-01 -7.26443589e-01 -1.48132369e-01 4.48384434e-01 3.22261006e-01 6.33060336e-01 -1.14372063e+00 -4.55811679e-01 -6.54978573e-01 -7.01153837e-03 7.55851567e-01 4.28295612e-01 1.51953959e+00 -3.04527193e-01 -4.39632237e-01 7.76160240e-01 1.35519731e+00 6.13492168e-02 7.74212837e-01 5.29215336e-01 3.52362007e-01 6.20812654e-01 8.28169659e-02 1.17759421e-01 -8.66504386e-02 3.85406584e-01 6.47340477e-01 -6.71136260e-01 -1.42040759e-01 3.44794810e-01 -1.79129362e-01 -9.34812799e-02 -3.05364668e-01 5.15422635e-02 -9.38490093e-01 2.18081072e-01 -1.16426587e+00 -5.39031923e-01 -7.82539099e-02 2.09885812e+00 4.78912562e-01 1.69244215e-01 4.01728839e-01 -2.31189400e-01 7.13219941e-01 -3.46490681e-01 -5.57017684e-01 -5.82112849e-01 2.44949490e-01 6.20580435e-01 7.98468828e-01 4.22742486e-01 -1.04619908e+00 7.08344340e-01 7.34792233e+00 5.80130935e-01 -1.74176061e+00 1.19350776e-01 6.02996707e-01 -5.17387092e-01 2.97880266e-02 -2.83910483e-01 -6.68659925e-01 2.27207154e-01 8.62600327e-01 1.77609801e-01 2.94050395e-01 8.68168771e-01 1.18011370e-01 -1.67490095e-01 -8.07511270e-01 6.69295251e-01 -2.41464421e-01 -1.82604158e+00 -1.63890347e-01 2.71548420e-01 1.32722393e-01 3.03583235e-01 2.71462500e-01 1.68582459e-03 -7.28076473e-02 -1.21581399e+00 1.04934521e-01 4.28715557e-01 1.37281752e+00 -6.39707506e-01 9.84836578e-01 -1.48832694e-01 -5.71790397e-01 -2.48863429e-01 -1.72720015e-01 7.13654160e-02 -3.94680470e-01 2.51518875e-01 -1.39589977e+00 1.82852894e-01 8.14142406e-01 1.17053948e-01 -1.31100714e+00 1.39298320e+00 2.47180611e-01 3.62204254e-01 -2.25963697e-01 -4.16282028e-01 5.02268821e-02 1.45762861e-01 2.08785191e-01 1.57169199e+00 1.30585935e-02 -4.29901183e-01 -5.07744074e-01 7.16195822e-01 4.66797173e-01 1.06826954e-01 -2.58308560e-01 1.60154002e-03 3.14216942e-01 1.81697214e+00 -6.85708880e-01 2.20135882e-01 -2.26157427e-01 7.97280490e-01 2.13604316e-01 4.71160173e-01 -5.67356110e-01 -6.81816638e-01 7.82350659e-01 5.52493811e-01 3.16213630e-02 2.41007172e-02 -4.42705095e-01 -4.64749336e-01 -1.72701970e-01 -9.21753466e-01 4.48297858e-01 -6.38993204e-01 -8.37410450e-01 7.55634844e-01 -6.53831586e-02 -9.91084337e-01 -9.94201228e-02 -1.16658270e+00 -8.71852994e-01 9.52996194e-01 -1.02622032e+00 -1.12337232e+00 -7.99330771e-01 2.44700849e-01 1.74065784e-01 -4.30540055e-01 1.21109414e+00 9.82444808e-02 -9.03726757e-01 8.92360091e-01 -3.19901675e-01 2.52200514e-01 7.26201475e-01 -1.17969942e+00 5.74155152e-01 5.31002045e-01 -4.56703246e-01 7.94125080e-01 5.11359155e-01 -3.18775058e-01 -8.47559452e-01 -9.93296802e-01 1.18141696e-01 -3.04272383e-01 4.57344472e-01 -3.02979559e-01 -6.12090945e-01 4.27174479e-01 4.72906172e-01 -1.89609140e-01 1.10430908e+00 4.33933645e-01 -1.44166395e-01 -2.28351161e-01 -1.31658757e+00 6.97846174e-01 4.43064183e-01 -1.09381877e-01 4.20891464e-01 3.72219354e-01 3.95025253e-01 -8.05745602e-01 -9.96370971e-01 2.94612646e-01 7.51339495e-01 -1.02966940e+00 7.39160538e-01 -8.02903354e-01 2.40479022e-01 4.42981385e-02 4.88633126e-01 -1.17999828e+00 -4.21721548e-01 -6.95523858e-01 3.27810526e-01 6.82024240e-01 7.12003589e-01 -7.15883374e-01 9.04569507e-01 7.31029034e-01 6.60360381e-02 -1.07885742e+00 -8.59719217e-01 -2.33208925e-01 -6.77166283e-02 -2.86214590e-01 5.85788012e-01 6.96011424e-01 -8.30813125e-02 -5.84805124e-02 1.32160589e-01 2.62888283e-01 8.68955776e-02 -7.77364254e-01 7.43869781e-01 -1.14083123e+00 -3.87359560e-01 -6.36262596e-01 -6.88445330e-01 -8.24495479e-02 -6.17131829e-01 -6.91274226e-01 -5.52124441e-01 -1.39929307e+00 -2.96265204e-02 -4.10672307e-01 -2.03158095e-01 9.64277327e-01 -1.24590751e-02 2.67379045e-01 1.45600364e-01 -2.59907097e-01 2.34078299e-02 -6.74528062e-01 1.12924230e+00 1.08572863e-01 -4.10219461e-01 5.35950363e-02 -8.24470460e-01 5.84101617e-01 1.05016541e+00 -1.82119310e-01 -2.38901272e-01 -5.34656405e-01 3.09851795e-01 -4.16046232e-01 7.06627071e-01 -1.36116278e+00 3.27408403e-01 -3.67682949e-02 7.33113766e-01 -9.55356210e-02 1.85860008e-01 -7.54946232e-01 4.12610471e-01 7.17052460e-01 -3.90160620e-01 -3.18837725e-02 7.15016425e-01 1.82331335e-02 2.83379793e-01 -4.99761075e-01 1.01465511e+00 -3.22707415e-01 -5.46198726e-01 2.42084578e-01 -5.51196337e-01 -5.34329116e-01 1.24239433e+00 -6.92757905e-01 -5.79464078e-01 -8.51470605e-02 -9.78560090e-01 1.13520987e-01 6.48398280e-01 5.66872537e-01 3.01308423e-01 -7.89336979e-01 -3.96865278e-01 3.94556642e-01 2.96161234e-01 -1.21540636e-01 5.81686437e-01 7.87507713e-01 -1.02651417e+00 2.06457272e-01 -6.33534670e-01 -4.64196026e-01 -1.60183644e+00 1.73802629e-01 1.06808829e+00 -1.53003082e-01 -4.11658257e-01 9.03019428e-01 -3.94945174e-01 -3.99930775e-01 3.24667424e-01 -3.78508747e-01 -2.41569474e-01 4.22476716e-02 7.21412480e-01 4.31617588e-01 6.71335220e-01 3.79708141e-01 -1.95015296e-01 1.53179497e-01 -3.71211976e-01 1.25838026e-01 1.14781082e+00 2.92326391e-01 8.98308400e-03 4.66558449e-02 8.59721541e-01 -4.15455341e-01 -1.10713375e+00 3.02921027e-01 -4.33441967e-01 -1.91244572e-01 1.82338536e-01 -1.38027155e+00 -1.20673692e+00 7.51984715e-01 1.36859894e+00 1.64156899e-01 1.35124242e+00 -3.05859357e-01 1.02733649e-01 3.47875923e-01 -1.01495326e-01 -1.14405417e+00 -2.37768754e-01 1.79544926e-01 7.39785969e-01 -9.25601423e-01 1.28842965e-01 -4.44752634e-01 -6.07276261e-01 1.69783413e+00 9.55361187e-01 -3.42594460e-02 6.45200074e-01 7.00341225e-01 4.83370841e-01 -3.23303521e-01 -7.53355324e-01 -2.11936593e-01 1.41179949e-01 9.62845385e-01 7.30327010e-01 2.54005492e-02 -2.34442070e-01 2.10831121e-01 -1.43192604e-01 4.15686339e-01 7.01869488e-01 7.26405382e-01 -1.64693400e-01 -1.27068603e+00 -2.94116497e-01 9.26361620e-01 -5.12242854e-01 -7.82093331e-02 -4.59600925e-01 1.14456856e+00 5.43243170e-01 5.38137317e-01 3.63976806e-01 -3.56280625e-01 3.25413853e-01 -5.15744276e-02 6.12578452e-01 -8.35329354e-01 -1.10504198e+00 -2.09329516e-01 3.30722272e-01 -4.14321542e-01 -3.00951228e-02 -2.20558211e-01 -7.16960788e-01 -2.13324308e-01 -4.02773470e-01 -5.27373075e-01 9.81915355e-01 5.49678922e-01 5.54916501e-01 7.77813852e-01 -1.80251654e-02 -4.08769727e-01 -2.99109042e-01 -1.25594413e+00 -3.79400820e-01 -2.11221561e-01 1.86546922e-01 -2.44133815e-01 -9.51265730e-03 -7.27852248e-03]
[15.669193267822266, -2.979501724243164]
b22092db-782b-4d85-ac41-d58f7614b28a
bagging-regional-classification-activation
2207.07818
null
https://arxiv.org/abs/2207.07818v1
https://arxiv.org/pdf/2207.07818v1.pdf
Bagging Regional Classification Activation Maps for Weakly Supervised Object Localization
Classification activation map (CAM), utilizing the classification structure to generate pixel-wise localization maps, is a crucial mechanism for weakly supervised object localization (WSOL). However, CAM directly uses the classifier trained on image-level features to locate objects, making it prefers to discern global discriminative factors rather than regional object cues. Thus only the discriminative locations are activated when feeding pixel-level features into this classifier. To solve this issue, this paper elaborates a plug-and-play mechanism called BagCAMs to better project a well-trained classifier for the localization task without refining or re-training the baseline structure. Our BagCAMs adopts a proposed regional localizer generation (RLG) strategy to define a set of regional localizers and then derive them from a well-trained classifier. These regional localizers can be viewed as the base learner that only discerns region-wise object factors for localization tasks, and their results can be effectively weighted by our BagCAMs to form the final localization map. Experiments indicate that adopting our proposed BagCAMs can improve the performance of baseline WSOL methods to a great extent and obtains state-of-the-art performance on three WSOL benchmarks. Code are released at https://github.com/zh460045050/BagCAMs.
['Yanye Lu', 'Yunfei You', 'Lujia Jin', 'Qian Chen', 'Lei Zhu']
2022-07-16
null
null
null
null
['weakly-supervised-object-localization']
['computer-vision']
[-8.19050614e-03 -1.31150082e-01 -5.55350780e-01 -4.53901559e-01 -1.28795624e+00 -7.26162136e-01 6.56659484e-01 7.61608481e-02 -4.85034734e-01 4.55312192e-01 1.62351150e-02 -2.46728450e-01 7.52825961e-02 -6.26220644e-01 -9.66264546e-01 -9.91858244e-01 1.09935179e-01 1.76623662e-03 6.47433281e-01 2.00426262e-02 3.13392609e-01 6.23116672e-01 -1.49594069e+00 5.42610526e-01 7.69766688e-01 1.21361530e+00 5.85864723e-01 5.48556328e-01 -1.75493449e-01 7.26762831e-01 -4.89835232e-01 5.10663688e-02 2.81154841e-01 -3.52185935e-01 -6.20268285e-01 -2.01932609e-01 5.28067708e-01 -1.75449159e-02 4.93954048e-02 1.08259618e+00 1.95953146e-01 9.94781256e-02 8.25401783e-01 -1.03386533e+00 -6.19394481e-01 6.10163748e-01 -8.26175034e-01 4.41343099e-01 2.47640625e-01 3.93029448e-04 1.10506952e+00 -1.30943811e+00 3.05122256e-01 9.52123225e-01 4.88740742e-01 3.14534664e-01 -1.08699751e+00 -6.87688470e-01 5.68672597e-01 1.16546787e-01 -1.76269281e+00 -3.99539977e-01 6.76635921e-01 -2.84785062e-01 6.46959901e-01 4.08795297e-01 4.35306847e-01 7.25624859e-01 1.69561133e-01 1.04111695e+00 1.27381909e+00 -6.10292375e-01 3.73079389e-01 2.06738502e-01 1.91052780e-01 9.86255944e-01 7.08669499e-02 -8.30152482e-02 -9.12995517e-01 -1.49559215e-01 8.10901701e-01 1.14819169e-01 -3.97041112e-01 -4.21627790e-01 -1.40156865e+00 7.16512144e-01 1.13270783e+00 3.48966509e-01 -2.74240822e-01 2.95945883e-01 -1.86880678e-01 -1.40321612e-01 6.38736606e-01 5.44064939e-01 -5.18790483e-01 1.72851548e-01 -8.74872565e-01 1.06544398e-01 1.96508661e-01 9.08251524e-01 1.12161410e+00 -4.39290047e-01 -4.50192720e-01 9.02055800e-01 3.12800020e-01 4.05087769e-01 4.36998159e-01 -6.40463710e-01 4.23437148e-01 7.69561648e-01 7.72169679e-02 -8.16159904e-01 -2.18951628e-01 -6.60223842e-01 -2.70999312e-01 2.38174662e-01 3.69266838e-01 2.47714624e-01 -1.19199622e+00 1.67935467e+00 2.46870801e-01 3.20098698e-01 -3.18843633e-01 1.05807924e+00 6.61759138e-01 7.59924293e-01 1.89036235e-01 3.07277888e-01 1.27328849e+00 -1.36518824e+00 -2.77236074e-01 -4.73255575e-01 7.08045185e-01 -5.09322405e-01 1.28711104e+00 2.97029316e-01 -8.50006878e-01 -8.02387536e-01 -1.22385585e+00 -3.20318155e-02 -8.25040579e-01 5.36374509e-01 7.40613103e-01 3.17971766e-01 -1.24355114e+00 1.09429948e-01 -9.00416076e-01 -1.90444067e-01 7.49719679e-01 3.22495699e-01 -3.43590230e-01 -1.70840666e-01 -8.52478325e-01 7.58171797e-01 3.72381687e-01 1.04929104e-01 -1.08447683e+00 -6.59852982e-01 -7.63490736e-01 -4.52730656e-02 3.90998453e-01 -2.14383006e-01 1.00006628e+00 -8.67043674e-01 -9.82519150e-01 6.83004975e-01 -4.66332167e-01 -9.53281000e-02 1.72341391e-01 -1.51171416e-01 -1.16561145e-01 2.48863339e-01 5.22602916e-01 1.06993413e+00 8.71767461e-01 -1.61996353e+00 -1.23049247e+00 -2.12584779e-01 2.66468465e-01 2.89473325e-01 -2.42107943e-01 4.34254445e-02 -9.55716133e-01 -6.94430113e-01 4.88468438e-01 -7.31611311e-01 -1.85337439e-01 5.87712228e-02 -2.11461067e-01 -3.77432972e-01 5.84509015e-01 -2.33080029e-01 1.10143244e+00 -2.05929875e+00 1.34888394e-02 2.70341754e-01 2.83362657e-01 2.55694240e-03 -3.05348337e-01 1.65482759e-02 -2.19250116e-02 2.50826403e-02 -2.27979213e-01 -4.41705644e-01 -3.01163465e-01 -1.00758448e-01 -3.65133286e-01 5.46972036e-01 4.73770559e-01 1.14059949e+00 -1.00971150e+00 -5.25859296e-01 4.08878326e-01 3.00007254e-01 -3.69648576e-01 1.88213244e-01 -5.23006171e-02 4.31785613e-01 -5.95528603e-01 1.04656732e+00 7.46974766e-01 -2.80415118e-01 -2.96098739e-01 -3.60747188e-01 -2.05878332e-01 -7.85365421e-03 -9.31221485e-01 1.89056146e+00 -5.73459566e-01 5.01354933e-01 -2.15833768e-01 -9.38007355e-01 8.83995891e-01 -1.87472343e-01 3.73897940e-01 -6.38547242e-01 9.94094461e-02 2.21943527e-01 -2.53550231e-01 -1.60211369e-01 1.58080190e-01 2.37532616e-01 -2.34802946e-01 1.84662431e-01 4.40655738e-01 2.40112826e-01 5.37442788e-02 2.63705790e-01 1.16904044e+00 4.07228649e-01 2.46278405e-01 -4.67115968e-01 6.53932750e-01 1.51603237e-01 4.36960489e-01 9.95660305e-01 -1.50217071e-01 8.46962631e-01 1.93815961e-01 -3.57614756e-01 -5.26026189e-01 -1.19888413e+00 -2.98277706e-01 1.54549408e+00 5.62290490e-01 -4.43161845e-01 -7.45525956e-01 -1.22925043e+00 -3.25515680e-02 4.76604313e-01 -8.22274804e-01 -3.32402915e-01 -3.90995413e-01 -8.65047157e-01 3.62157673e-01 8.40762973e-01 4.92014080e-01 -9.53705847e-01 -4.70141649e-01 7.83503875e-02 -1.04759857e-01 -9.51679230e-01 -4.69390184e-01 7.62228072e-01 -4.92219388e-01 -9.87546086e-01 -7.31487632e-01 -1.00881660e+00 1.31399894e+00 6.98106945e-01 8.29146147e-01 1.89855158e-01 -2.17750728e-01 5.63303381e-02 -6.32176459e-01 -3.75464350e-01 7.02157840e-02 2.19224751e-01 -4.07877825e-02 2.64787763e-01 4.89183873e-01 -2.73250967e-01 -8.31326723e-01 5.41524231e-01 -6.96545839e-01 1.42953321e-01 8.50109041e-01 7.76429951e-01 8.50060701e-01 -5.78245595e-02 5.67823052e-01 -7.36730814e-01 2.95715153e-01 -3.73955488e-01 -6.79895401e-01 3.29698652e-01 -4.20244783e-01 6.13618717e-02 6.20133162e-01 -4.93953377e-01 -9.37522590e-01 5.13181567e-01 -8.09213519e-03 -3.27818394e-01 -1.90801904e-01 3.85595500e-01 -3.19132805e-01 -4.48493570e-01 7.91091800e-01 3.18249911e-01 -4.88590211e-01 -4.51699078e-01 5.70960581e-01 7.53329575e-01 6.10977948e-01 -6.81726456e-01 8.54534149e-01 5.36728263e-01 -3.30914468e-01 -2.46082425e-01 -1.32700348e+00 -9.30449486e-01 -8.84444535e-01 -2.99732387e-01 8.58345389e-01 -1.12874055e+00 -3.16348702e-01 2.35860363e-01 -8.60884011e-01 -4.30127352e-01 -1.12710863e-01 3.93857419e-01 -2.41891161e-01 -2.61541367e-01 -3.20523471e-01 -6.10503197e-01 1.66305229e-01 -1.23123062e+00 1.50114286e+00 4.65222687e-01 4.21927646e-02 -8.13347638e-01 -1.55437678e-01 2.57381827e-01 3.11402529e-01 5.56651913e-02 4.95921493e-01 -6.12729132e-01 -8.00987422e-01 -4.51527178e-01 -4.49122310e-01 2.97212064e-01 2.31446579e-01 -4.24773306e-01 -1.46741748e+00 -2.12658986e-01 -1.62287742e-01 -2.55525500e-01 1.24766827e+00 3.50934207e-01 1.59240854e+00 -7.47173578e-02 -6.79817379e-01 8.09041500e-01 1.48475194e+00 1.75850481e-01 3.23027432e-01 2.45925814e-01 8.58790457e-01 2.40859210e-01 8.69233310e-01 -3.89304645e-02 2.57424980e-01 6.57198429e-01 3.57142985e-01 -4.72348899e-01 -2.40166962e-01 -5.26857018e-01 4.13708746e-01 5.80454230e-01 -2.86960322e-02 -1.79604664e-01 -8.84136856e-01 5.60054541e-01 -2.02341676e+00 -4.14796233e-01 1.43392503e-01 2.00762701e+00 7.97129691e-01 1.50039673e-01 -2.07404345e-01 -8.20357576e-02 5.48339903e-01 2.40961522e-01 -3.55907023e-01 1.91389501e-01 -7.67912194e-02 2.22065315e-01 8.71243000e-01 5.77794671e-01 -1.40539956e+00 1.21802104e+00 5.78593731e+00 1.19098520e+00 -1.22319305e+00 3.92300874e-01 7.79674470e-01 5.87210432e-02 3.53281274e-02 5.53160198e-02 -1.08225155e+00 5.16392529e-01 5.15358031e-01 4.83083844e-01 3.59682620e-01 1.07875919e+00 5.77272251e-02 -4.26270902e-01 -1.09938383e+00 8.90999854e-01 2.90310413e-01 -1.32089794e+00 -6.47998452e-02 -2.85130236e-02 7.59996057e-01 1.25201806e-01 3.99386287e-02 3.45406473e-01 1.55907601e-01 -1.05780208e+00 9.73024070e-01 4.01756048e-01 6.40541494e-01 -4.89161223e-01 7.61465192e-01 3.91930670e-01 -1.47573876e+00 -1.86454102e-01 -6.21483743e-01 1.66698083e-01 -2.46078670e-01 3.56914967e-01 -8.46765637e-01 3.37259710e-01 8.61844182e-01 6.49609387e-01 -9.99379873e-01 1.09356236e+00 -6.07677281e-01 7.81260908e-01 -2.80013382e-01 -1.68858620e-03 3.67902011e-01 1.41924411e-01 1.58948272e-01 1.33005369e+00 1.23772673e-01 -1.98380183e-02 3.76211822e-01 9.97938812e-01 -6.46687020e-03 -2.56458800e-02 -2.02858865e-01 4.11274523e-01 5.53690910e-01 1.61966813e+00 -1.25730979e+00 -2.69797146e-01 -3.80812258e-01 9.63785827e-01 5.63413680e-01 5.28145254e-01 -9.91345644e-01 -3.28536123e-01 4.46207494e-01 2.56038696e-01 4.11394835e-01 -7.20506981e-02 -4.19058174e-01 -1.05979979e+00 2.73606721e-02 -5.66649854e-01 2.79057831e-01 -9.87619042e-01 -1.08132589e+00 6.48050904e-01 1.34022176e-01 -1.22426569e+00 2.50941396e-01 -8.06147337e-01 -6.81370258e-01 9.05341685e-01 -1.74092209e+00 -1.50267386e+00 -6.65249109e-01 5.42142570e-01 6.82170212e-01 -1.26185164e-01 5.92928350e-01 1.32698312e-01 -3.94346952e-01 5.86922228e-01 -5.18396311e-02 1.88423857e-01 7.92165339e-01 -1.40398693e+00 8.12844634e-02 1.05170822e+00 5.05443811e-01 8.13957632e-01 2.28599995e-01 -5.42044342e-01 -1.20968843e+00 -1.31144273e+00 6.06791794e-01 -8.21077049e-01 5.29370368e-01 -9.26453352e-01 -7.13634074e-01 4.56780493e-01 -1.93191484e-01 5.40182412e-01 3.65767866e-01 9.56907272e-02 -4.23099458e-01 -4.90942240e-01 -9.33225989e-01 4.38393950e-01 1.09264696e+00 -6.40687644e-01 -3.90190065e-01 2.54164219e-01 6.76913321e-01 -3.86774898e-01 -4.56607848e-01 3.05951744e-01 3.18589121e-01 -7.71948695e-01 9.75434482e-01 -1.54219210e-01 2.16261953e-01 -9.85149860e-01 -1.00097321e-01 -1.26655471e+00 -5.57744801e-01 -4.72452268e-02 5.41867614e-02 1.31601119e+00 7.62434840e-01 -6.49497032e-01 7.32591152e-01 1.23532690e-01 -2.62005627e-01 -1.04716790e+00 -7.48884380e-01 -6.40905023e-01 -2.83519506e-01 -4.84833896e-01 6.61378860e-01 7.74361968e-01 -2.64415652e-01 1.08390898e-01 1.40012875e-01 4.87492502e-01 5.06897926e-01 1.12139665e-01 5.66016138e-01 -7.27857471e-01 -1.20589122e-01 -3.29010010e-01 -4.72981334e-01 -1.37478602e+00 5.41593768e-02 -1.16371477e+00 6.45690858e-01 -1.54650903e+00 3.05115014e-01 -8.79277647e-01 -9.27501976e-01 9.25471604e-01 -4.80079085e-01 7.58620203e-01 1.07309073e-01 4.00922954e-01 -9.45258558e-01 3.38005871e-01 1.08615685e+00 -2.30016291e-01 -8.27763751e-02 -8.52566659e-02 -1.04693079e+00 6.27989292e-01 7.46566653e-01 -6.00055158e-01 -4.22301143e-01 -3.06830853e-01 1.76288665e-03 -5.28123081e-01 4.92187500e-01 -1.11765325e+00 4.43905592e-01 -1.25300318e-01 8.03885937e-01 -5.76614738e-01 3.15407783e-01 -7.49114871e-01 -3.81257743e-01 8.15393552e-02 -4.32886034e-01 -2.68826663e-01 3.55491936e-02 3.99995565e-01 -3.32059503e-01 -1.44444466e-01 6.40277743e-01 -4.60998341e-02 -9.86860693e-01 2.82292515e-01 -2.23329037e-01 -2.47257262e-01 1.00436425e+00 -3.45472634e-01 -2.83482641e-01 -1.66074950e-02 -3.66371781e-01 2.62311161e-01 3.81072760e-01 5.06597340e-01 5.48726261e-01 -1.33019841e+00 -4.10863578e-01 3.79970551e-01 6.12621665e-01 4.88582194e-01 -2.30967641e-01 7.41465509e-01 -4.88113761e-01 4.21723336e-01 4.70043980e-02 -9.29660797e-01 -8.48031044e-01 4.61890846e-01 3.25380027e-01 -4.90220562e-02 -3.84268105e-01 1.35897899e+00 7.92253852e-01 -3.44291598e-01 3.00464720e-01 -6.40666783e-01 -9.83432308e-02 -1.00432649e-01 6.39168143e-01 -6.06471114e-02 1.34065464e-01 -5.83794951e-01 -6.16547883e-01 7.32728243e-01 -8.07003397e-03 -4.53197062e-02 1.24805260e+00 -1.50203824e-01 -1.92293134e-02 4.38776791e-01 1.16036534e+00 1.64932534e-01 -1.48978984e+00 -1.17973655e-01 1.03754893e-01 -5.59795201e-01 3.83725733e-01 -1.04421818e+00 -1.13113165e+00 7.67757237e-01 7.20155180e-01 -1.27430320e-01 1.18978977e+00 5.56234002e-01 3.55112851e-01 9.00286213e-02 5.07134557e-01 -1.00578821e+00 1.91382796e-01 2.02248365e-01 8.65667939e-01 -1.35351288e+00 -1.17902294e-01 -5.07692695e-01 -5.04507244e-01 8.76161158e-01 8.59440386e-01 -3.17728430e-01 7.09558666e-01 5.04196048e-01 1.66568413e-01 -1.33841082e-01 -4.63535219e-01 -3.66175681e-01 6.86238766e-01 5.85957468e-01 2.66436994e-01 7.94880614e-02 1.96070061e-03 9.13905442e-01 1.32812038e-01 -1.88681990e-01 -1.71891838e-01 9.53915477e-01 -7.12755561e-01 -9.52913880e-01 -5.45304418e-01 4.45206314e-01 -2.94607162e-01 -1.80309385e-01 -3.89677316e-01 4.80729520e-01 6.36320651e-01 7.52913177e-01 8.79223794e-02 -4.23394769e-01 1.11175440e-01 -1.64505035e-01 2.63862312e-01 -7.45156407e-01 -4.34875220e-01 1.93068787e-01 -4.07100827e-01 -7.39749849e-01 -4.42228496e-01 -2.80269712e-01 -1.34048510e+00 3.48203093e-01 -6.05349302e-01 1.09563656e-01 7.29524255e-01 8.75723064e-01 2.26943299e-01 5.86795211e-01 7.36406624e-01 -9.83305514e-01 -6.67639002e-02 -9.76718068e-01 -2.58266926e-01 1.36514381e-01 3.70908380e-01 -8.89231443e-01 -3.82982880e-01 8.16923007e-02]
[9.556374549865723, 0.9453558325767517]
e137f196-23b6-44df-b535-adae7d66dae5
boosting-the-generalization-capability-in
2108.05028
null
https://arxiv.org/abs/2108.05028v2
https://arxiv.org/pdf/2108.05028v2.pdf
Boosting the Generalization Capability in Cross-Domain Few-shot Learning via Noise-enhanced Supervised Autoencoder
State of the art (SOTA) few-shot learning (FSL) methods suffer significant performance drop in the presence of domain differences between source and target datasets. The strong discrimination ability on the source dataset does not necessarily translate to high classification accuracy on the target dataset. In this work, we address this cross-domain few-shot learning (CDFSL) problem by boosting the generalization capability of the model. Specifically, we teach the model to capture broader variations of the feature distributions with a novel noise-enhanced supervised autoencoder (NSAE). NSAE trains the model by jointly reconstructing inputs and predicting the labels of inputs as well as their reconstructed pairs. Theoretical analysis based on intra-class correlation (ICC) shows that the feature embeddings learned from NSAE have stronger discrimination and generalization abilities in the target domain. We also take advantage of NSAE structure and propose a two-step fine-tuning procedure that achieves better adaption and improves classification performance in the target domain. Extensive experiments and ablation studies are conducted to demonstrate the effectiveness of the proposed method. Experimental results show that our proposed method consistently outperforms SOTA methods under various conditions.
['Juwei Lu', 'Peng Dai', 'Qiong Zhang', 'Hanwen Liang']
2021-08-11
null
http://openaccess.thecvf.com//content/ICCV2021/html/Liang_Boosting_the_Generalization_Capability_in_Cross-Domain_Few-Shot_Learning_via_Noise-Enhanced_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Liang_Boosting_the_Generalization_Capability_in_Cross-Domain_Few-Shot_Learning_via_Noise-Enhanced_ICCV_2021_paper.pdf
iccv-2021-1
['cross-domain-few-shot', 'cross-domain-few-shot-learning']
['computer-vision', 'computer-vision']
[ 2.57763416e-01 -3.24469239e-01 -1.07839689e-01 -4.80069309e-01 -6.05233073e-01 -3.25096875e-01 4.74150211e-01 -3.22792888e-01 -3.17006499e-01 7.02821791e-01 2.65062749e-01 3.91919136e-01 -3.73034000e-01 -8.26103568e-01 -5.18017709e-01 -6.90058172e-01 2.96462089e-01 2.36200407e-01 3.65569741e-01 -2.89714187e-01 -1.12547532e-01 2.77382851e-01 -1.80869019e+00 3.06808561e-01 9.99128342e-01 1.16802108e+00 3.11669350e-01 3.57332557e-01 -1.26717359e-01 5.97204328e-01 -5.63130856e-01 -3.08274418e-01 4.29751366e-01 -6.76132739e-01 -4.97032404e-01 6.20789789e-02 2.60562092e-01 -1.36764392e-01 -3.04507643e-01 9.98491704e-01 7.04013467e-01 5.13617575e-01 7.93022513e-01 -1.19690073e+00 -9.45740283e-01 5.20113349e-01 -3.40391606e-01 3.06917787e-01 -9.30181295e-02 1.15643159e-01 1.01799905e+00 -1.18207979e+00 5.31576455e-01 1.01697612e+00 6.65456295e-01 8.00958037e-01 -1.23720825e+00 -8.59376848e-01 -1.53783828e-01 3.40846807e-01 -1.45634449e+00 -4.21776503e-01 9.15397227e-01 -3.48639041e-01 6.96211994e-01 -8.09784606e-02 3.89746308e-01 1.47251809e+00 5.92381135e-02 7.75651157e-01 1.24737394e+00 -5.80578804e-01 6.45136952e-01 5.68188071e-01 2.65322536e-01 4.63680893e-01 2.07961112e-01 3.51175815e-01 -6.82501137e-01 -6.55612303e-03 4.62083101e-01 2.54581571e-01 -2.05682352e-01 -6.48924470e-01 -8.75907540e-01 9.72829700e-01 4.91701722e-01 5.39649844e-01 -3.23930144e-01 -3.31551522e-01 3.69962275e-01 4.27766711e-01 3.75063866e-01 4.94198471e-01 -3.85828197e-01 -5.50838448e-02 -6.68830216e-01 -1.71233699e-01 4.82658744e-01 9.61473882e-01 8.09435964e-01 4.14846241e-01 -2.90806353e-01 1.32418656e+00 -2.36536548e-01 3.60891223e-01 1.03411615e+00 -4.30917591e-01 1.66740179e-01 5.75823843e-01 -3.42158705e-01 -7.20081270e-01 -1.20106302e-01 -7.51544595e-01 -7.95623004e-01 3.13731730e-02 9.02185887e-02 -3.03494185e-01 -8.48871112e-01 1.88714552e+00 2.15647608e-01 5.56632936e-01 4.00561035e-01 7.12421358e-01 7.15953171e-01 6.60068512e-01 2.05381975e-01 -2.25794777e-01 1.02405107e+00 -7.71578968e-01 -6.64863706e-01 -3.38422567e-01 5.91245830e-01 -3.95767868e-01 1.21426141e+00 3.89026888e-02 -4.75245357e-01 -9.06663895e-01 -1.20559108e+00 3.29481691e-01 -6.22227311e-01 5.16644903e-02 4.79007393e-01 6.46626949e-01 -4.22287077e-01 6.95409715e-01 -4.17449355e-01 -4.28769201e-01 7.58181393e-01 2.11522967e-01 -4.19915140e-01 -2.93897003e-01 -1.43807936e+00 7.61544049e-01 6.08862340e-01 -4.69846785e-01 -7.52628028e-01 -9.92938221e-01 -1.03655863e+00 3.29638988e-01 1.49916366e-01 -3.42061430e-01 1.06426036e+00 -1.03126800e+00 -1.42201102e+00 6.03479207e-01 7.88291991e-02 -4.34093535e-01 -1.02501619e-03 -1.08467184e-01 -7.13970482e-01 -1.15597256e-01 4.94690388e-02 5.16101837e-01 9.05061543e-01 -1.01175749e+00 -6.79660439e-01 -3.90919209e-01 -3.20799261e-01 8.96288827e-02 -1.09466016e+00 -3.59834075e-01 -3.10111810e-02 -7.74669945e-01 -1.16357207e-01 -7.56075382e-01 9.20306966e-02 -1.37138247e-01 3.89665551e-03 -2.16736481e-01 9.28778172e-01 -2.21762121e-01 1.39098001e+00 -2.51637697e+00 4.10917252e-02 1.61723912e-01 -6.41046688e-02 6.94930851e-01 -3.14362109e-01 3.96981716e-01 -1.72214851e-01 -4.22902703e-01 -2.54908234e-01 7.25906566e-02 2.44163107e-02 4.06592429e-01 -4.21907753e-01 2.89728522e-01 2.73613334e-01 8.42047632e-01 -8.01486969e-01 -2.78186560e-01 1.95378363e-01 3.63853484e-01 -3.68415266e-01 3.51774156e-01 1.88520521e-01 1.10823438e-01 -3.30349624e-01 6.15976453e-01 6.26462221e-01 -3.01498443e-01 -9.94734094e-02 -2.70921171e-01 3.21984828e-01 -2.01633140e-01 -1.15534127e+00 1.74269402e+00 -6.10786319e-01 5.20883262e-01 -5.71339548e-01 -1.04030275e+00 1.17544305e+00 1.31372705e-01 1.75021261e-01 -9.87249017e-01 2.76325196e-01 1.94291681e-01 2.15651348e-01 -4.35098231e-01 -2.44555082e-02 -5.87693393e-01 -1.87521040e-01 3.47470343e-01 6.91595852e-01 2.86757559e-01 -8.58057737e-02 -4.28623855e-02 9.87533927e-01 -1.48392335e-01 7.68453598e-01 -7.31767789e-02 2.77473211e-01 -2.05404416e-01 7.96065569e-01 7.81475306e-01 -5.52331090e-01 4.67912227e-01 4.26853336e-02 -3.88847530e-01 -1.06055832e+00 -1.21252632e+00 -3.42003226e-01 1.42446041e+00 1.27717689e-01 -9.04507563e-02 -6.10067010e-01 -8.71654749e-01 1.69959694e-01 1.05767536e+00 -1.00499499e+00 -8.31965446e-01 -1.75639302e-01 -6.68722987e-01 4.13732678e-01 8.78053069e-01 5.95098257e-01 -9.34389412e-01 -6.80598795e-01 1.14992797e-01 2.06645966e-01 -1.08326626e+00 -2.86600828e-01 4.58997250e-01 -7.48977184e-01 -7.86068082e-01 -7.64351964e-01 -9.77745175e-01 4.78275746e-01 3.87149245e-01 6.23662233e-01 -5.18332481e-01 -3.42849135e-01 3.25115472e-01 -6.18081808e-01 -3.86987597e-01 -1.70192704e-01 -1.44732982e-01 3.99131119e-01 3.29440176e-01 7.97646940e-01 -7.20091045e-01 -2.38082990e-01 3.66675168e-01 -9.53547239e-01 -3.36198539e-01 7.83476353e-01 1.35294831e+00 4.32860076e-01 3.46296430e-01 9.07695532e-01 -1.16643393e+00 7.48245180e-01 -6.42679453e-01 -1.72437161e-01 4.22136217e-01 -6.55045986e-01 1.22659132e-01 8.15516949e-01 -7.76407063e-01 -1.23634207e+00 7.37039093e-03 3.86549123e-02 -8.03434908e-01 -3.79616767e-01 2.78671831e-01 -2.81452209e-01 -4.47918400e-02 1.13513100e+00 4.36990470e-01 -8.73474330e-02 -4.36827749e-01 2.21943095e-01 8.51370752e-01 3.66375029e-01 -3.79152864e-01 8.38490546e-01 2.53109306e-01 -3.06572318e-01 -8.04559231e-01 -1.14134753e+00 -5.49025714e-01 -7.53263950e-01 -7.94440433e-02 5.02690375e-01 -9.09838498e-01 -2.45125331e-02 3.33423853e-01 -7.02522457e-01 -7.89807811e-02 -5.86480141e-01 6.31261051e-01 -5.00896573e-01 -1.46829635e-01 -2.33352751e-01 -6.54020488e-01 -2.35239565e-01 -8.11992168e-01 6.05893552e-01 4.76544112e-01 -1.89569160e-01 -1.07550812e+00 2.93667406e-01 -1.50035456e-01 5.14358044e-01 -4.29047756e-02 9.47345972e-01 -1.19388688e+00 8.01567510e-02 -4.07622486e-01 -1.80489212e-01 5.51780462e-01 4.42124009e-01 -4.84914035e-01 -1.36970580e+00 -2.96792567e-01 7.79103041e-02 -5.58865786e-01 9.47601318e-01 2.72566408e-01 1.11538100e+00 -1.61413979e-02 -2.37621680e-01 6.17510200e-01 1.59998834e+00 1.89710870e-01 4.34194356e-01 1.18314534e-01 4.92294043e-01 4.50199336e-01 6.93613768e-01 5.88560641e-01 -1.26865402e-01 4.75821942e-01 -4.71284799e-03 2.84335554e-01 -3.77090573e-01 -4.41288859e-01 3.64136398e-01 9.24983025e-01 3.18891823e-01 3.31254154e-02 -8.71520281e-01 7.08105326e-01 -1.63390183e+00 -9.83498454e-01 5.47298193e-01 2.00623679e+00 7.78881907e-01 1.35692641e-01 -4.99131009e-02 1.72570631e-01 7.30719745e-01 3.53910355e-03 -7.53101945e-01 -2.17630833e-01 -1.57537445e-01 5.23927152e-01 1.97207555e-01 9.69532412e-03 -1.12862432e+00 8.66094589e-01 5.94052649e+00 1.23570788e+00 -1.14133751e+00 2.04012096e-01 3.35862905e-01 -3.53333652e-02 -1.68602809e-03 -2.70806491e-01 -8.78225088e-01 5.01675606e-01 9.66509879e-01 -3.24666739e-01 2.98937678e-01 1.13308489e+00 -3.69172782e-01 1.85530379e-01 -1.04529095e+00 1.00059330e+00 2.50327736e-01 -1.29536700e+00 1.87238604e-01 -1.84748709e-01 9.36899364e-01 -1.58115834e-01 2.79653847e-01 8.05856347e-01 2.44908452e-01 -7.21634626e-01 2.72122920e-01 4.53386575e-01 8.71288478e-01 -8.53617728e-01 8.06896508e-01 4.81252968e-01 -9.22897279e-01 -6.88357294e-01 -7.92197645e-01 -1.51451714e-02 -2.99350768e-01 3.38542581e-01 -8.23644340e-01 3.38037729e-01 6.32816315e-01 9.94511962e-01 -6.97384894e-01 1.08016610e+00 -4.56568785e-02 6.62898600e-01 5.25636710e-02 -2.38229424e-01 1.11214675e-01 6.12421818e-02 5.12018085e-01 1.04851329e+00 4.23949599e-01 3.08997542e-01 -4.53612730e-02 8.28835845e-01 -7.51949176e-02 5.80003262e-02 -9.24469948e-01 -2.22274080e-01 6.83834493e-01 9.93308127e-01 -3.41634035e-01 -4.05067921e-01 -5.12154520e-01 9.91975605e-01 6.01370931e-01 3.59231025e-01 -5.62427461e-01 -6.82692230e-01 6.67032719e-01 -1.55770138e-01 7.15026319e-01 2.71103501e-01 -3.04979801e-01 -1.17220974e+00 -2.30415747e-01 -8.23256433e-01 6.28214657e-01 -5.62830985e-01 -1.85910118e+00 6.41250968e-01 -1.07915103e-01 -1.66699421e+00 -6.57501742e-02 -5.32947004e-01 -6.57944500e-01 6.53341174e-01 -1.33280206e+00 -1.13135171e+00 -2.31315479e-01 7.97059953e-01 7.81407595e-01 -7.37603843e-01 1.18751299e+00 3.02735627e-01 -6.57491386e-01 1.02789629e+00 4.24345732e-01 1.87979519e-01 9.02280927e-01 -9.46011901e-01 -1.28294617e-01 8.26084554e-01 2.46377528e-01 6.57045066e-01 5.70314229e-01 -4.99255985e-01 -1.14726079e+00 -1.24893022e+00 4.68637496e-01 -1.79740578e-01 6.49031878e-01 -2.86515862e-01 -1.06005394e+00 5.52590787e-01 7.07876384e-02 4.06611085e-01 1.27846301e+00 2.03335777e-01 -7.16399431e-01 -4.20261145e-01 -1.12735546e+00 3.18619519e-01 8.74152362e-01 -6.67089999e-01 -8.30410659e-01 -1.00415714e-01 6.41844213e-01 1.34373024e-01 -7.82795489e-01 4.85091329e-01 5.34775913e-01 -9.09071684e-01 8.30660164e-01 -9.40381587e-01 4.65790331e-01 5.96157797e-02 -6.94215715e-01 -1.78603244e+00 -6.11910343e-01 1.00827947e-01 -2.80408978e-01 1.24286866e+00 2.17697918e-01 -4.53227937e-01 5.06001472e-01 2.70303100e-01 -2.85704806e-02 -6.76669121e-01 -8.29648912e-01 -1.11166131e+00 1.39644474e-01 -2.84307063e-01 5.14935374e-01 1.14419580e+00 1.66167691e-01 5.47391891e-01 -5.41809440e-01 8.88249651e-02 6.78722978e-01 2.84034014e-01 5.96541345e-01 -1.41296041e+00 -3.84638786e-01 -1.43368974e-01 -5.42046309e-01 -5.80614209e-01 2.56167352e-01 -9.36172068e-01 5.64732738e-02 -1.02142453e+00 3.41614306e-01 -2.63895929e-01 -8.06127310e-01 4.73612607e-01 -3.16548586e-01 1.94036350e-01 9.22747701e-02 1.12137102e-01 -7.26392627e-01 9.64419723e-01 1.05400443e+00 -9.29118767e-02 -2.00332701e-01 7.18941987e-02 -7.36733973e-01 5.34899056e-01 7.72466660e-01 -5.54767549e-01 -7.03568637e-01 -1.90928146e-01 -3.74497861e-01 -2.33661771e-01 1.64472252e-01 -1.43489838e+00 1.67305753e-01 -2.18849152e-01 8.55570853e-01 -1.43187404e-01 4.23781097e-01 -9.38531458e-01 -1.84169158e-01 4.66932952e-01 -6.07606649e-01 -5.30851543e-01 2.27956235e-01 9.17376637e-01 -2.86618114e-01 -3.23712885e-01 1.18599200e+00 5.25862724e-02 -1.16067886e+00 3.05157214e-01 -2.84561515e-02 1.44323662e-01 1.30397511e+00 -2.56935418e-01 -1.74475998e-01 -1.42183825e-01 -7.05026746e-01 -9.32919830e-02 1.86173394e-01 6.57971144e-01 8.11940253e-01 -1.75488651e+00 -4.79174793e-01 6.79036379e-01 6.94898784e-01 -6.00082636e-01 4.85642910e-01 5.56882441e-01 2.76668102e-01 3.28029424e-01 -8.06618690e-01 -3.69654059e-01 -9.60826159e-01 6.83643103e-01 1.72828972e-01 3.54072191e-02 -5.12841225e-01 1.12170196e+00 2.83751160e-01 -5.74307680e-01 2.23806277e-01 1.79760307e-01 -2.51300186e-01 2.96755642e-01 7.64551580e-01 4.09725428e-01 -8.83646086e-02 -4.57507998e-01 -2.62510568e-01 4.11198348e-01 -3.32720906e-01 1.70682326e-01 1.58010030e+00 -2.05798130e-02 5.32134473e-01 6.60815656e-01 1.33213985e+00 -3.16166461e-01 -1.28579986e+00 -8.15731287e-01 -1.09004445e-01 -5.93477607e-01 1.05621189e-01 -8.87017608e-01 -8.23333979e-01 1.12757432e+00 9.08682823e-01 -7.73605257e-02 1.16797531e+00 -1.10718377e-01 8.61070812e-01 2.29789495e-01 2.59929150e-01 -1.39179552e+00 3.47852707e-01 4.23912793e-01 4.34514940e-01 -1.36802506e+00 -2.95603931e-01 -1.29201308e-01 -9.02641177e-01 1.02662492e+00 9.19874310e-01 -3.31095904e-01 7.97431707e-01 -3.72233056e-02 -1.30472139e-01 6.23851921e-03 -8.20388556e-01 -3.05889696e-01 4.22695220e-01 8.70957196e-01 5.18875271e-02 2.95374636e-02 8.82735774e-02 1.11117756e+00 1.23466961e-02 -1.12085305e-02 2.57452756e-01 9.02279019e-01 -7.29617834e-01 -9.35188711e-01 -9.25626829e-02 5.03714502e-01 7.64039308e-02 6.04604650e-03 -3.59558821e-01 5.61814725e-01 1.71463877e-01 7.24916637e-01 -1.13376305e-02 -7.23790526e-01 4.40979272e-01 4.58301276e-01 3.31632942e-01 -8.06832492e-01 -2.98867375e-01 -1.19540311e-01 -3.32234234e-01 -3.49861652e-01 -2.02877849e-01 -3.85448515e-01 -1.05776393e+00 -3.11871953e-02 -4.29219097e-01 8.26685652e-02 3.30045253e-01 1.07865393e+00 4.96415585e-01 5.99839389e-01 8.88364434e-01 -3.28055322e-01 -9.43451762e-01 -1.03593361e+00 -8.62822115e-01 6.16496325e-01 3.00372332e-01 -1.11825073e+00 -4.36945289e-01 -1.34834453e-01]
[10.012916564941406, 2.986598491668701]
b7ff9cdf-bf3b-4b1c-b4b6-16e959c6cc58
towards-lightweight-cross-domain-sequential
2302.03221
null
https://arxiv.org/abs/2302.03221v1
https://arxiv.org/pdf/2302.03221v1.pdf
Towards Lightweight Cross-domain Sequential Recommendation via External Attention-enhanced Graph Convolution Network
Cross-domain Sequential Recommendation (CSR) is an emerging yet challenging task that depicts the evolution of behavior patterns for overlapped users by modeling their interactions from multiple domains. Existing studies on CSR mainly focus on using composite or in-depth structures that achieve significant improvement in accuracy but bring a huge burden to the model training. Moreover, to learn the user-specific sequence representations, existing works usually adopt the global relevance weighting strategy (e.g., self-attention mechanism), which has quadratic computational complexity. In this work, we introduce a lightweight external attention-enhanced GCN-based framework to solve the above challenges, namely LEA-GCN. Specifically, by only keeping the neighborhood aggregation component and using the Single-Layer Aggregating Protocol (SLAP), our lightweight GCN encoder performs more efficiently to capture the collaborative filtering signals of the items from both domains. To further alleviate the framework structure and aggregate the user-specific sequential pattern, we devise a novel dual-channel External Attention (EA) component, which calculates the correlation among all items via a lightweight linear structure. Extensive experiments are conducted on two real-world datasets, demonstrating that LEA-GCN requires a smaller volume and less training time without affecting the accuracy compared with several state-of-the-art methods.
['Xinhua Wang', 'Liancheng Xu', 'Lei Guo', 'Huichuan Duan', 'Jinyu Zhang']
2023-02-07
null
null
null
null
['collaborative-filtering']
['miscellaneous']
[ 4.78577055e-02 -5.72350860e-01 -2.20088229e-01 -3.43355149e-01 -4.95452017e-01 -2.30425522e-01 2.11149961e-01 -2.73978319e-02 -2.36066595e-01 3.94021392e-01 4.86169904e-01 -2.68918425e-01 -3.71738702e-01 -8.20693970e-01 -7.93139219e-01 -5.13677299e-01 -1.81679130e-01 -2.35666987e-02 2.23569110e-01 -4.83085990e-01 2.72369310e-02 -1.19957745e-01 -1.38012815e+00 4.37040478e-01 1.10601175e+00 1.22761536e+00 4.93747234e-01 1.93648800e-01 -1.26025409e-01 7.57967532e-01 -3.27944010e-01 -4.77044642e-01 2.95605063e-01 -3.12518507e-01 -4.20799553e-01 -5.99428138e-04 8.43968689e-02 -5.15000761e-01 -5.13095558e-01 1.01562750e+00 5.57867885e-01 4.94693965e-01 1.25901788e-01 -7.88801432e-01 -1.27376366e+00 9.49879766e-01 -5.19554973e-01 2.21429199e-01 2.88993686e-01 -1.26464561e-01 1.33595181e+00 -8.24728847e-01 2.83368170e-01 1.19209027e+00 7.29459226e-01 2.78622031e-01 -1.10020471e+00 -8.15800548e-01 8.48931432e-01 4.34024751e-01 -1.35321760e+00 -1.60987023e-02 7.76227474e-01 5.28073125e-02 8.97405088e-01 1.78433776e-01 4.67781037e-01 1.13110888e+00 -2.13726386e-01 9.93062079e-01 5.38706899e-01 1.07482281e-02 2.10971553e-02 -3.31452936e-02 3.71449381e-01 2.92295575e-01 3.34635288e-01 -1.07029445e-01 -3.07660669e-01 -2.03353271e-01 8.06982398e-01 5.94623208e-01 -4.77766961e-01 -1.67366236e-01 -9.47027683e-01 7.67668784e-01 5.29504299e-01 3.44713092e-01 -5.31601012e-01 -8.84834230e-02 3.93587619e-01 4.06650186e-01 5.33380985e-01 1.51852548e-01 -6.80184066e-01 1.18276998e-01 -4.88302499e-01 2.69284815e-01 5.74010253e-01 1.16594923e+00 4.24644530e-01 -1.89960644e-01 -2.42389053e-01 9.26240623e-01 4.17993575e-01 1.29384503e-01 6.98709071e-01 -4.59835321e-01 6.17267430e-01 8.39500666e-01 2.36731723e-01 -1.30430913e+00 -2.64186293e-01 -1.13005972e+00 -8.96255672e-01 -6.66762888e-01 1.07739896e-01 -2.06436530e-01 -3.72805148e-01 1.87657177e+00 2.72441715e-01 5.76559067e-01 -4.22151461e-02 1.13810241e+00 6.97854102e-01 7.55282879e-01 1.28212377e-01 -2.34665036e-01 1.35779464e+00 -1.30618227e+00 -7.90604293e-01 1.45762442e-02 6.27699077e-01 -6.22676015e-01 1.09794450e+00 4.69979703e-01 -1.05813420e+00 -8.92461896e-01 -1.04120338e+00 -1.13407999e-01 -3.35591882e-01 3.84691954e-01 7.41313338e-01 3.56122702e-01 -6.82227969e-01 6.15637779e-01 -4.88866419e-01 -1.12893932e-01 3.43445599e-01 4.92413878e-01 -1.03855673e-02 -2.21050099e-01 -1.57939994e+00 2.19888598e-01 2.60547012e-01 2.38156229e-01 -3.91998321e-01 -7.73873270e-01 -5.68456769e-01 5.52671313e-01 5.71518242e-01 -4.60520893e-01 1.10774231e+00 -1.16315567e+00 -1.48788941e+00 -7.63746276e-02 -8.59247670e-02 -3.91317129e-01 1.39319524e-01 -5.94214022e-01 -9.60046768e-01 -3.41244280e-01 -2.07028672e-01 -2.48628650e-02 4.25687730e-01 -1.11665404e+00 -8.40810597e-01 -4.10242081e-01 3.11444342e-01 3.27031523e-01 -7.94304550e-01 1.84208110e-01 -8.60936344e-01 -1.04996812e+00 -1.31173030e-01 -6.43180370e-01 -3.87078345e-01 -4.03112888e-01 -5.63962646e-02 -4.13743168e-01 6.08820260e-01 -8.04033935e-01 2.00592923e+00 -2.21023250e+00 5.56836165e-02 3.15698117e-01 1.29045948e-01 5.82722545e-01 -3.11540842e-01 3.73600304e-01 1.17106795e-01 -7.76364580e-02 1.07887194e-01 -2.94968754e-01 -2.49148849e-02 9.37645063e-02 -2.70275146e-01 1.65193796e-01 -6.28372878e-02 8.50354314e-01 -1.01385283e+00 -8.34079534e-02 -7.07155541e-02 5.76330960e-01 -9.49450433e-01 4.04588789e-01 -2.26046219e-01 2.34107956e-01 -5.34259081e-01 3.41345966e-01 8.83499205e-01 -7.92121649e-01 6.39819741e-01 -3.44469458e-01 -2.88309194e-02 5.30512094e-01 -1.50457442e+00 1.71960330e+00 -6.01066053e-01 -1.60585776e-01 1.30489543e-01 -1.10035789e+00 8.12929213e-01 2.58664429e-01 3.32629621e-01 -9.09402847e-01 2.18177274e-01 1.52035072e-01 4.33063507e-02 -3.90042841e-01 5.16589999e-01 3.73538792e-01 2.86694262e-02 6.36469066e-01 7.16308802e-02 1.35299635e+00 -3.93992253e-02 1.54010579e-01 9.82677460e-01 1.25143990e-01 3.51281941e-01 -1.56290874e-01 8.05889726e-01 -5.10666311e-01 8.41903269e-01 6.94643378e-01 3.73026617e-02 3.51797223e-01 1.34017304e-01 -4.66143787e-01 -7.08259344e-01 -6.37026429e-01 2.86457002e-01 1.38886952e+00 5.37499905e-01 -6.24936700e-01 -5.62218964e-01 -7.47915447e-01 -6.77302256e-02 3.31902325e-01 -5.89078486e-01 -1.82421014e-01 -6.71155572e-01 -9.90492642e-01 -6.64963294e-03 7.84050703e-01 5.33649087e-01 -8.06228220e-01 -7.04930574e-02 7.00817943e-01 -3.36264312e-01 -9.98855174e-01 -9.56786990e-01 -1.79582700e-01 -7.67081618e-01 -7.84579217e-01 -7.26070940e-01 -7.69450247e-01 4.95310068e-01 8.66201222e-01 9.23817575e-01 2.59309858e-01 2.24943504e-01 4.82563600e-02 -9.37341928e-01 5.02396636e-02 4.58721966e-01 2.36453250e-01 -1.99420750e-02 5.84748507e-01 7.41643965e-01 -6.78167045e-01 -9.35668886e-01 4.92252290e-01 -7.98651338e-01 -1.44465119e-01 6.77257299e-01 9.84503984e-01 5.78273594e-01 2.09653124e-01 9.14115012e-01 -1.06991875e+00 8.85020733e-01 -1.00305879e+00 -3.30830395e-01 3.15107912e-01 -6.50511861e-01 -1.54678494e-01 1.28564084e+00 -6.79391563e-01 -1.25405240e+00 -2.98625559e-01 -1.76245704e-01 -4.29747134e-01 -5.34202857e-03 6.87549949e-01 -3.31892461e-01 2.45052978e-01 2.37789914e-01 5.39431751e-01 -3.59081060e-01 -9.76532459e-01 2.98078060e-01 8.95493925e-01 3.23237032e-01 -5.04136145e-01 5.34815431e-01 2.67407000e-01 -7.12310135e-01 -3.39749902e-01 -1.05689812e+00 -7.45346725e-01 -1.22769862e-01 1.77193835e-01 4.93787974e-01 -1.27063501e+00 -8.87383461e-01 1.95375055e-01 -9.73675191e-01 -9.09472704e-02 5.48815951e-02 6.78350806e-01 -3.71832550e-02 5.52958667e-01 -7.33476102e-01 -7.82409787e-01 -7.03164816e-01 -9.12984967e-01 7.82035112e-01 2.18370616e-01 8.41571316e-02 -8.08884919e-01 -1.53158322e-01 3.64156276e-01 6.42074764e-01 -3.85019302e-01 7.81653345e-01 -8.28539968e-01 -5.96710265e-01 5.67272455e-02 -6.23200655e-01 3.18881392e-01 2.11830989e-01 -6.94713533e-01 -6.41215503e-01 -4.07189548e-01 -1.01771258e-01 1.97071388e-01 6.31288648e-01 9.59612504e-02 1.57315099e+00 -4.63820189e-01 -2.68470019e-01 6.08623624e-01 1.53384745e+00 4.42753077e-01 4.03157622e-01 2.36595124e-01 7.69630194e-01 3.22950691e-01 6.10836446e-01 7.11327434e-01 6.39184833e-01 9.30387855e-01 2.82288402e-01 -3.91927399e-02 2.38496345e-02 -3.49876314e-01 4.21820760e-01 1.20011175e+00 -3.17681670e-01 -6.04219437e-01 -8.17950740e-02 4.98672068e-01 -2.36741114e+00 -1.05033064e+00 -1.33786559e-01 2.18905187e+00 5.73602438e-01 -5.69000505e-02 1.55640736e-01 -6.42541870e-02 7.69828737e-01 -2.60015968e-02 -5.48104227e-01 -7.65366927e-02 -1.63920131e-02 1.16911903e-01 3.60389143e-01 2.57455766e-01 -1.11568773e+00 6.86010242e-01 5.15615940e+00 1.06223845e+00 -9.63745177e-01 2.06171736e-01 2.33615011e-01 -1.95444301e-01 -3.98065686e-01 -2.81178504e-01 -9.17149544e-01 1.01644742e+00 8.49831462e-01 -1.53791606e-02 7.09949136e-01 7.57827401e-01 2.93525428e-01 5.75283110e-01 -6.51379049e-01 9.06168997e-01 8.40196684e-02 -1.14807010e+00 1.90064579e-01 4.30109464e-02 6.66152656e-01 3.03378031e-02 2.47628838e-02 4.85265553e-01 4.06580538e-01 -4.66852307e-01 4.85510260e-01 4.48909432e-01 4.60019410e-01 -8.83379936e-01 1.01483715e+00 2.03863174e-01 -1.63252068e+00 -6.17785931e-01 -5.25994897e-01 -1.47845775e-01 2.75662541e-01 5.43088496e-01 -1.57087191e-03 1.12886953e+00 9.98843133e-01 1.14150798e+00 -3.42207670e-01 9.19891000e-01 1.36515200e-01 7.89269686e-01 -2.53422022e-01 -2.14941688e-02 2.98046321e-01 -3.94679755e-01 3.95384908e-01 1.22253013e+00 6.71098948e-01 3.13911051e-01 2.12228864e-01 5.82324147e-01 -1.87232688e-01 4.51665878e-01 -6.56183437e-02 1.27838597e-01 4.22010005e-01 1.20186770e+00 -2.65773475e-01 -2.95827180e-01 -1.10936117e+00 9.80520487e-01 5.25587201e-01 3.88616443e-01 -1.01447701e+00 -3.85035425e-01 8.49024653e-01 1.92160338e-01 9.74918783e-01 -3.45087387e-02 1.13675641e-02 -1.45462799e+00 1.68050215e-01 -9.68323112e-01 4.87347156e-01 -3.82099807e-01 -1.64886105e+00 6.15600646e-01 -5.09141803e-01 -1.52637064e+00 3.42006534e-01 -3.48184407e-01 -5.68480968e-01 8.17541838e-01 -1.63478565e+00 -1.11755109e+00 -1.96967632e-01 8.39516699e-01 4.98853356e-01 -2.31489465e-01 6.02863729e-01 1.10235929e+00 -9.66604054e-01 1.05294538e+00 3.84156168e-01 -4.27606814e-02 7.00015604e-01 -8.75798941e-01 3.94605726e-01 7.58145809e-01 -1.37261614e-01 1.23859024e+00 2.79608488e-01 -5.61815143e-01 -1.44784141e+00 -1.44767976e+00 7.98628747e-01 -2.42088409e-03 5.33474863e-01 -4.19295609e-01 -1.08260977e+00 7.47204542e-01 3.41673009e-02 4.89893444e-02 1.01479912e+00 4.92617369e-01 -3.97777110e-01 -2.68046290e-01 -7.16903865e-01 6.56397998e-01 1.56328046e+00 -2.71169960e-01 -4.85167742e-01 -5.94594255e-02 1.09259379e+00 4.72502373e-02 -9.91130948e-01 2.80844361e-01 7.54916131e-01 -8.19412410e-01 1.13954961e+00 -7.41227746e-01 2.89295405e-01 -5.57556033e-01 -3.49441111e-01 -1.16942513e+00 -9.59442616e-01 -6.16032541e-01 -6.52652085e-01 1.29131675e+00 2.21478194e-01 -7.32224762e-01 5.41478992e-01 3.02850097e-01 -2.32168958e-01 -9.88452435e-01 -3.96240503e-01 -6.77689672e-01 -4.14550602e-01 -3.66732270e-01 1.05167997e+00 9.73801553e-01 4.67567854e-02 5.09635150e-01 -8.62412930e-01 2.20986053e-01 4.70253974e-01 4.36465442e-01 5.65822184e-01 -1.23885870e+00 -6.65097535e-01 -2.39830941e-01 2.64165420e-02 -1.62466776e+00 -1.80556312e-01 -8.38142693e-01 -2.99794614e-01 -1.32598054e+00 1.18317522e-01 -4.84889299e-01 -9.62639213e-01 2.46837661e-01 -4.43900317e-01 5.88887790e-03 2.33656392e-02 1.83697999e-01 -1.13511050e+00 6.84835136e-01 1.25064683e+00 1.28402591e-01 -3.43635529e-01 1.74918398e-02 -1.32267463e+00 4.71124053e-01 6.53627038e-01 -3.13376427e-01 -5.83228886e-01 -6.48931384e-01 3.97113085e-01 -1.99360669e-01 3.63123752e-02 -9.03203726e-01 4.35482591e-01 -9.68656130e-03 3.84866536e-01 -5.02908647e-01 1.05101205e-01 -1.09800410e+00 1.31698966e-01 1.74170524e-01 -4.52615678e-01 3.62223685e-02 -6.50992841e-02 1.06261683e+00 -2.32063785e-01 1.40153646e-01 3.83218825e-01 -1.42494544e-01 -6.87435806e-01 6.50821805e-01 -9.88474861e-02 -2.78744936e-01 7.48909175e-01 1.58421516e-01 -1.83816165e-01 -3.31132025e-01 -5.39289773e-01 4.75304663e-01 7.57877827e-02 6.42588913e-01 4.21763599e-01 -1.63655579e+00 -4.96540099e-01 3.99987578e-01 1.35667682e-01 -2.57314473e-01 9.51017439e-01 7.67328203e-01 9.96189415e-02 5.37178814e-01 1.17559973e-02 -5.68945659e-03 -9.23806012e-01 8.43561411e-01 5.42347096e-02 -7.61611938e-01 -6.15162849e-01 7.76118159e-01 3.71395290e-01 -5.68718612e-01 2.33532339e-01 -1.05057187e-01 -6.76730037e-01 1.02772206e-01 9.69338477e-01 4.66322958e-01 -2.09280439e-02 -4.87427682e-01 -1.23853952e-01 4.61811364e-01 -5.82993209e-01 5.64476967e-01 1.44022787e+00 -4.88127142e-01 1.34314239e-01 -1.09038226e-01 1.17397153e+00 -5.91807589e-02 -1.17498183e+00 -8.44697535e-01 -3.02238703e-01 -4.73957270e-01 1.83255479e-01 -6.54073656e-01 -1.27819347e+00 6.66701019e-01 5.41373014e-01 3.07367355e-01 1.33579636e+00 -3.92863840e-01 1.42242086e+00 1.95940301e-01 4.53970671e-01 -1.12439179e+00 -2.61757404e-01 4.56439227e-01 5.93165636e-01 -1.03621745e+00 -1.01102881e-01 -3.48689407e-01 -5.56866050e-01 7.84149051e-01 8.11378539e-01 -2.77985960e-01 1.03570294e+00 -1.18763924e-01 -2.76593626e-01 8.95767957e-02 -8.89299154e-01 -2.76212096e-01 3.13722968e-01 2.35131398e-01 5.48924565e-01 2.05040127e-01 -6.37900710e-01 1.60574722e+00 3.78509641e-01 3.23599935e-01 6.09177724e-02 6.66317940e-01 -2.23565817e-01 -1.38752782e+00 1.55048251e-01 6.52934253e-01 -6.47393346e-01 -2.96074927e-01 1.95239782e-01 3.22056383e-01 2.98286289e-01 9.85179901e-01 -2.30668969e-02 -8.13561797e-01 5.29835463e-01 -1.93965659e-01 7.36393966e-03 -4.75084484e-01 -9.17628765e-01 3.78330648e-01 -3.83832939e-02 -7.19107568e-01 -4.67821538e-01 -6.45784795e-01 -9.82864082e-01 -3.51394802e-01 -5.00333369e-01 1.69652045e-01 8.44776258e-02 8.23983490e-01 1.00411320e+00 7.85983443e-01 6.74614072e-01 -6.73464298e-01 -6.02343082e-01 -9.61792409e-01 -6.51991069e-01 4.99800920e-01 1.81016997e-01 -6.63535416e-01 -6.88620508e-02 -2.94379324e-01]
[10.144558906555176, 5.5530900955200195]
5e2e5da5-8e6b-49d2-b3e6-2a0bec18e3f5
discovering-transferable-forensic-features
2208.11342
null
https://arxiv.org/abs/2208.11342v1
https://arxiv.org/pdf/2208.11342v1.pdf
Discovering Transferable Forensic Features for CNN-generated Images Detection
Visual counterfeits are increasingly causing an existential conundrum in mainstream media with rapid evolution in neural image synthesis methods. Though detection of such counterfeits has been a taxing problem in the image forensics community, a recent class of forensic detectors -- universal detectors -- are able to surprisingly spot counterfeit images regardless of generator architectures, loss functions, training datasets, and resolutions. This intriguing property suggests the possible existence of transferable forensic features (T-FF) in universal detectors. In this work, we conduct the first analytical study to discover and understand T-FF in universal detectors. Our contributions are 2-fold: 1) We propose a novel forensic feature relevance statistic (FF-RS) to quantify and discover T-FF in universal detectors and, 2) Our qualitative and quantitative investigations uncover an unexpected finding: color is a critical T-FF in universal detectors. Code and models are available at https://keshik6.github.io/transferable-forensic-features/
['Ngai-Man Cheung', 'Alexander Binder', 'Ngoc-Trung Tran', 'Keshigeyan Chandrasegaran']
2022-08-24
null
null
null
null
['image-forensics']
['computer-vision']
[ 1.62880182e-01 -2.99482077e-01 -2.08377823e-01 5.19479886e-02 -8.39920342e-01 -6.33079648e-01 6.83552682e-01 1.68940198e-04 -2.29329333e-01 4.13600415e-01 8.80358219e-02 -4.61677194e-01 -5.27244732e-02 -5.08037925e-01 -8.57885957e-01 -5.19986510e-01 -1.65767163e-01 -1.28235206e-01 5.35483181e-01 -1.15750641e-01 7.07609832e-01 5.48138082e-01 -1.52411413e+00 4.80677962e-01 6.04695618e-01 1.01664734e+00 6.64309412e-02 6.01836443e-01 7.69802779e-02 9.34715390e-01 -5.95665693e-01 -1.02936482e+00 4.96460319e-01 -5.49422979e-01 -6.37684762e-01 -1.00359693e-01 8.74707103e-01 -5.55157423e-01 -7.20673800e-01 1.41859293e+00 4.69841748e-01 -2.81807989e-01 7.76888430e-01 -1.46764493e+00 -1.16786778e+00 3.52384031e-01 -9.63013113e-01 9.45448518e-01 1.83749378e-01 5.28872252e-01 1.06678784e+00 -8.82726669e-01 8.74258339e-01 1.25591195e+00 6.99498832e-01 6.05694830e-01 -9.21437025e-01 -9.98621523e-01 -5.15671730e-01 7.64585614e-01 -1.28473151e+00 -6.20993733e-01 8.65234971e-01 -6.49660110e-01 5.02131343e-01 2.14932203e-01 3.81805629e-01 1.35792160e+00 1.07144229e-01 9.20437872e-01 1.38541818e+00 -3.38501453e-01 -7.82145336e-02 2.50894338e-01 1.96635321e-01 7.69165635e-01 5.25880098e-01 7.44162947e-02 -9.39552724e-01 -1.28470674e-01 8.22260916e-01 -1.85090065e-01 -1.92609847e-01 5.83482422e-02 -8.23181093e-01 8.99451911e-01 1.88759074e-01 2.06026480e-01 -1.45871297e-01 3.49825799e-01 6.88472331e-01 3.93737197e-01 3.43949467e-01 2.76457012e-01 -3.93843092e-03 -2.52962530e-01 -1.01616144e+00 2.05085352e-01 3.69951099e-01 5.87603509e-01 6.67989314e-01 -8.38781893e-02 -4.84460518e-02 4.87656742e-01 -1.69999570e-01 5.37627578e-01 4.11952525e-01 -9.73778069e-01 5.40401757e-01 3.93266678e-01 -1.16207056e-01 -1.19733405e+00 1.18991956e-01 -1.84432834e-01 -4.02886569e-01 2.15028733e-01 5.98940432e-01 2.76524752e-01 -3.52558643e-01 1.42354405e+00 2.31739312e-01 1.26923189e-01 -5.76884151e-01 8.33513260e-01 7.43748009e-01 1.20378107e-01 4.03246246e-02 5.55239916e-02 1.55369377e+00 -4.20912802e-01 -3.95034760e-01 2.76932251e-02 2.91292489e-01 -6.86646044e-01 1.30489075e+00 3.47798347e-01 -1.00526023e+00 -2.32467562e-01 -9.91296172e-01 -1.29266039e-01 -2.78185427e-01 1.59672201e-01 5.89652121e-01 1.09768820e+00 -6.12076819e-01 8.12672675e-01 -4.09455299e-01 -4.06244636e-01 1.02113903e+00 -8.49503204e-02 -5.40216506e-01 -9.95571613e-02 -7.99413383e-01 8.53909254e-01 1.41211003e-01 -2.99532831e-01 -9.95990872e-01 -5.20654500e-01 -3.83606970e-01 -1.38584435e-01 4.01645631e-01 -2.24343553e-01 8.69022727e-01 -1.00960243e+00 -1.03681993e+00 1.36535907e+00 1.23996474e-02 -3.42986256e-01 7.32835531e-01 -1.47106498e-01 -4.37711120e-01 6.80164576e-01 3.66329551e-01 2.52174318e-01 1.17838478e+00 -1.21779203e+00 -2.84762263e-01 -3.21930230e-01 -2.18839198e-01 -5.16431570e-01 -6.59257293e-01 6.01619422e-01 -8.45479071e-02 -7.49096930e-01 -2.09369570e-01 -5.26289523e-01 3.51590306e-01 3.27746838e-01 -7.18665481e-01 -3.80232707e-02 8.83882523e-01 -7.43179739e-01 1.22774053e+00 -2.24820375e+00 -4.03727829e-01 -6.24662898e-02 5.46411157e-01 3.79189700e-01 -3.17524299e-02 2.64762521e-01 -1.04735062e-01 2.95802236e-01 -2.89942920e-01 -1.08538471e-01 1.05910540e-01 -2.05951214e-01 -6.78111434e-01 1.06475508e+00 5.02763450e-01 1.05040848e+00 -8.11587095e-01 -7.00851679e-01 2.58900523e-01 2.72401363e-01 -4.06477183e-01 -8.06245059e-02 5.65930158e-02 1.52001753e-01 -5.17724514e-01 1.08848011e+00 8.42403173e-01 -3.28976482e-01 -1.53375000e-01 -1.97006524e-01 -2.21132785e-01 5.21642976e-02 -7.97155738e-01 1.02701092e+00 3.24638307e-01 1.07343864e+00 -8.19190443e-02 -9.84481692e-01 7.05831826e-01 -7.65510201e-02 3.31886500e-01 -1.11546481e+00 4.21130627e-01 3.49447757e-01 -6.27993569e-02 -7.12608755e-01 7.23638475e-01 -2.79165596e-01 2.50433851e-02 5.70614278e-01 2.34003082e-01 4.41995829e-01 2.53970802e-01 3.22029859e-01 1.44611549e+00 3.82578894e-02 1.54949039e-01 -2.82099515e-01 2.64465958e-01 -1.58739030e-01 4.18062061e-01 7.25151420e-01 -7.16741979e-01 7.87678003e-01 8.61280859e-01 -3.70508939e-01 -1.32934308e+00 -1.15569079e+00 -3.47527564e-01 8.30075026e-01 1.56369463e-01 -4.53015268e-01 -6.93339348e-01 -4.72640365e-01 1.89416513e-01 4.03392583e-01 -7.69522905e-01 -1.34780243e-01 -4.55381840e-01 -8.28532338e-01 1.19778299e+00 3.73702586e-01 4.77865845e-01 -9.86427009e-01 -1.03855884e+00 -3.27765763e-01 -7.36106932e-02 -1.23128915e+00 -3.54885280e-01 -2.51439214e-03 -7.62256920e-01 -1.33526206e+00 -7.16151536e-01 -1.04294546e-01 4.75576699e-01 5.03474236e-01 7.86059797e-01 2.19479635e-01 -8.08417261e-01 4.43193495e-01 -2.89952755e-01 -2.60709763e-01 -4.16683644e-01 -2.85174400e-01 -5.28095514e-02 1.07950561e-01 3.60511124e-01 -6.39997482e-01 -7.71127462e-01 3.04005653e-01 -9.73100781e-01 -3.82536292e-01 4.31510180e-01 4.41681504e-01 2.38021597e-01 -2.51056850e-01 4.25806046e-01 -6.74450457e-01 8.14359426e-01 -8.15912187e-01 -3.00163388e-01 1.59113154e-01 -3.97915244e-01 -4.18993421e-02 3.97454202e-01 -5.91075420e-01 -6.75527453e-01 -4.96676147e-01 2.20102012e-01 -8.32836449e-01 -2.57353988e-02 -4.24949303e-02 -1.48204625e-01 -6.18217997e-02 9.03751493e-01 1.80376828e-01 -2.18736321e-01 -5.31609297e-01 3.09615940e-01 4.09580708e-01 7.84378111e-01 -7.22229421e-01 9.57317233e-01 7.77003169e-01 1.19168468e-01 -9.95551348e-01 -4.82638299e-01 -5.02120554e-01 -4.08916920e-01 -4.51089501e-01 6.30656898e-01 -5.93648076e-01 -8.81065130e-01 4.98140037e-01 -1.14118874e+00 -5.91354556e-02 -1.70929760e-01 8.47105086e-02 -3.03554565e-01 9.66569602e-01 -6.25710428e-01 -9.97826934e-01 -2.43554026e-01 -1.02341926e+00 8.07397842e-01 3.06735039e-01 -2.24731863e-01 -6.70165002e-01 -2.09783446e-02 5.66737711e-01 2.86598533e-01 4.90470588e-01 7.23871827e-01 -4.18992490e-01 -6.28664374e-01 -1.89913258e-01 -8.47396851e-01 4.30513442e-01 -2.64576405e-01 2.46689767e-01 -1.31093395e+00 -4.77413014e-02 -1.48570031e-01 -3.62061322e-01 1.09615350e+00 3.68429780e-01 1.20821142e+00 -2.27206811e-01 -1.70366213e-01 7.54589736e-01 1.30650985e+00 -1.11979358e-01 9.17213857e-01 7.65671968e-01 5.50715506e-01 5.12821257e-01 9.91673544e-02 6.13921463e-01 6.62714392e-02 4.47906822e-01 4.88853335e-01 2.49854654e-01 -4.36485291e-01 -3.97351384e-01 7.41000712e-01 3.81444037e-01 -3.36421996e-01 -1.80968061e-01 -8.84406149e-01 6.49413526e-01 -1.55817425e+00 -1.39343584e+00 -4.63371426e-01 2.13495660e+00 5.18746912e-01 1.54888019e-01 4.01067227e-01 1.91833749e-01 1.03565121e+00 1.51111737e-01 -5.36713481e-01 -3.09648812e-01 -6.22692704e-01 4.05100405e-01 7.78054476e-01 -6.85255229e-02 -9.76611495e-01 8.21311414e-01 6.14462376e+00 1.19954896e+00 -1.28154993e+00 5.78200758e-01 6.50035620e-01 -1.73297241e-01 -2.68901348e-01 1.28565952e-01 -5.79895854e-01 6.84665799e-01 9.20138597e-01 -1.98788166e-01 1.81158960e-01 7.28793204e-01 -9.36425850e-02 -1.56232357e-01 -7.61909425e-01 1.11425233e+00 1.40883878e-01 -1.59487355e+00 7.03744441e-02 4.76644158e-01 2.56468892e-01 5.52876443e-02 5.56117892e-01 -1.30209416e-01 -8.40438157e-02 -9.49663401e-01 1.26872325e+00 3.95979285e-01 1.08067441e+00 -6.38095140e-01 1.64392173e-01 -5.37429601e-02 -8.19254637e-01 -3.26720059e-01 -4.90748256e-01 1.52205288e-01 6.93032593e-02 7.46344030e-01 -4.52367038e-01 1.86371118e-01 8.46057296e-01 6.89030886e-01 -8.60170305e-01 1.03403640e+00 -1.17965683e-01 7.83089697e-01 8.09491985e-03 2.89892137e-01 5.94387315e-02 -1.55727584e-02 7.62605250e-01 1.35490549e+00 4.41029817e-01 -1.93776160e-01 -5.94963193e-01 1.28748059e+00 -1.93910062e-01 -7.25793615e-02 -5.55370331e-01 -5.13105810e-01 3.60858738e-01 1.29498100e+00 -1.10710812e+00 4.30745818e-02 -2.33743444e-01 1.11345398e+00 2.17968971e-01 -6.58131614e-02 -9.04414237e-01 -3.26829702e-01 5.61388612e-01 7.15123177e-01 3.81195813e-01 -1.59133911e-01 -3.35413009e-01 -1.25124717e+00 2.08886623e-01 -7.16777742e-01 3.57496202e-01 -6.99446023e-01 -1.53698790e+00 2.87844628e-01 -2.14891341e-02 -1.26425874e+00 3.29805523e-01 -9.00959432e-01 -7.87671745e-01 3.91631126e-01 -1.46698296e+00 -1.25667214e+00 -1.05491549e-01 6.94627225e-01 1.26220286e-02 -4.16487426e-01 2.54832357e-01 6.01135015e-01 -6.15012705e-01 8.02127600e-01 2.33452082e-01 2.71056384e-01 7.03004241e-01 -8.52695584e-01 4.92221683e-01 1.08012938e+00 1.67613506e-01 7.92703629e-01 8.16160560e-01 -6.63993478e-01 -1.56577241e+00 -6.10000730e-01 4.28083003e-01 -7.26052165e-01 1.18286169e+00 -2.63867795e-01 -8.62631917e-01 4.23783094e-01 -2.05223821e-02 2.40463257e-01 7.09877133e-01 -2.08235502e-01 -1.12426782e+00 1.74835056e-01 -1.24566734e+00 3.30950737e-01 1.05988693e+00 -9.75064039e-01 -2.84393787e-01 1.27163157e-01 2.48570353e-01 1.82391331e-01 -5.61286211e-01 -6.17834590e-02 7.47144401e-01 -1.38013792e+00 1.12287235e+00 -3.82620156e-01 8.68498564e-01 -4.49593328e-02 -3.44748288e-01 -4.85459924e-01 -2.22582147e-02 -7.95879245e-01 -2.41519600e-01 1.24866569e+00 -3.06909382e-02 -5.86974859e-01 6.57761633e-01 2.97870100e-01 -1.40902242e-02 -4.52981114e-01 -1.42266226e+00 -1.05544293e+00 3.19007277e-01 -6.99783027e-01 3.29757690e-01 1.09019113e+00 7.87619408e-03 -1.17242180e-01 -5.39875388e-01 -2.27048740e-01 1.03318453e+00 -1.17235914e-01 5.82618952e-01 -1.13327742e+00 -3.43102872e-01 -7.82252371e-01 -5.68071246e-01 -5.37822783e-01 2.28201561e-02 -9.91592705e-01 -5.43805957e-01 -8.60826790e-01 7.54590094e-01 -2.07688347e-01 -3.14834654e-01 3.58367413e-01 1.20151388e-02 4.94804531e-01 4.32700455e-01 5.13714731e-01 -6.75505877e-01 2.42055669e-01 1.15297318e+00 -3.35583985e-02 4.45680916e-01 -4.06660974e-01 -8.45564127e-01 5.66969097e-01 7.89953887e-01 -6.74177885e-01 -4.48920168e-02 -2.05775440e-01 4.83313620e-01 -1.68666869e-01 1.03729606e+00 -8.88285518e-01 -2.72936895e-02 -2.22714886e-01 2.51873493e-01 -3.30983102e-01 2.10294351e-01 -1.88319683e-01 -8.94787684e-02 4.27377343e-01 1.17738500e-01 1.11309759e-01 1.89192474e-01 5.36918283e-01 1.05554201e-02 -4.35830176e-01 7.66327500e-01 -2.15225205e-01 -8.22920918e-01 1.71751007e-01 -2.11688459e-01 2.96159297e-01 8.67740333e-01 -5.69274962e-01 -9.42731202e-01 -1.15388878e-01 1.14143100e-02 -4.47850913e-01 6.44195855e-01 2.94772983e-01 7.73510873e-01 -1.10546947e+00 -7.50509322e-01 -1.42089635e-01 1.41130030e-01 -8.39959979e-01 3.57820779e-01 8.60300660e-01 -6.67129219e-01 1.25039324e-01 -5.74906230e-01 -2.74599195e-01 -1.21923065e+00 5.53440690e-01 4.25101370e-02 1.01939119e-01 -7.91689634e-01 8.81739557e-01 1.58412486e-01 1.62289470e-01 -3.25606048e-01 1.75761044e-01 3.49770695e-01 3.07416707e-01 6.86243713e-01 7.23083556e-01 -2.28032261e-01 -1.01746678e+00 -3.75382483e-01 2.43726730e-01 -5.15491813e-02 -4.92985584e-02 1.46377265e+00 1.52572170e-01 -6.12627529e-02 2.89829105e-01 1.20867109e+00 3.67211513e-02 -1.12564230e+00 -3.94204482e-02 1.78639010e-01 -9.67308164e-01 -5.20496294e-02 -3.67127150e-01 -1.19880712e+00 1.11309135e+00 6.03087783e-01 2.96140343e-01 8.81432593e-01 3.35593432e-01 9.73991275e-01 -5.10423258e-02 2.69366741e-01 -1.09964359e+00 4.78908896e-01 2.07127199e-01 7.50332892e-01 -1.06524384e+00 2.33806446e-01 -6.50135204e-02 -7.12621093e-01 1.11464286e+00 3.32769662e-01 -3.16163093e-01 2.90779799e-01 2.09221110e-01 -3.09867352e-01 -4.73314971e-01 -4.62587059e-01 -1.57526121e-01 1.30780950e-01 6.88106120e-01 3.63204479e-01 5.57689108e-02 -3.20363104e-01 6.69898272e-01 -2.13366523e-01 -2.06707403e-01 6.26282930e-01 7.39264786e-01 -3.89298856e-01 -1.01166606e+00 -5.33440173e-01 3.87434125e-01 -7.96377540e-01 -9.15934145e-02 -5.09674430e-01 7.29825079e-01 3.48322421e-01 7.30420947e-01 -4.99742180e-02 -5.99075258e-01 3.64816678e-03 -8.79902765e-02 8.39644134e-01 -1.47359788e-01 -6.29295468e-01 -3.28293145e-01 -2.06874594e-01 -5.62090993e-01 -2.60161132e-01 -9.58730102e-01 -8.60004306e-01 -6.93231285e-01 -4.38052833e-01 -3.46659958e-01 7.50204623e-01 7.01289713e-01 2.64342904e-01 5.72209693e-02 3.06604922e-01 -7.33569801e-01 -4.90833223e-01 -6.27074599e-01 -7.92465627e-01 6.76342249e-01 3.60342234e-01 -7.93829679e-01 -5.00333726e-01 1.25451148e-01]
[12.37857437133789, 1.0274885892868042]
17ee1fd4-a3fa-4298-8bdb-887e059224b8
unsupervised-hierarchical-semantic
2204.11432
null
https://arxiv.org/abs/2204.11432v1
https://arxiv.org/pdf/2204.11432v1.pdf
Unsupervised Hierarchical Semantic Segmentation with Multiview Cosegmentation and Clustering Transformers
Unsupervised semantic segmentation aims to discover groupings within and across images that capture object and view-invariance of a category without external supervision. Grouping naturally has levels of granularity, creating ambiguity in unsupervised segmentation. Existing methods avoid this ambiguity and treat it as a factor outside modeling, whereas we embrace it and desire hierarchical grouping consistency for unsupervised segmentation. We approach unsupervised segmentation as a pixel-wise feature learning problem. Our idea is that a good representation shall reveal not just a particular level of grouping, but any level of grouping in a consistent and predictable manner. We enforce spatial consistency of grouping and bootstrap feature learning with co-segmentation among multiple views of the same image, and enforce semantic consistency across the grouping hierarchy with clustering transformers between coarse- and fine-grained features. We deliver the first data-driven unsupervised hierarchical semantic segmentation method called Hierarchical Segment Grouping (HSG). Capturing visual similarity and statistical co-occurrences, HSG also outperforms existing unsupervised segmentation methods by a large margin on five major object- and scene-centric benchmarks. Our code is publicly available at https://github.com/twke18/HSG .
['Stella X. Yu', 'Xudong Wang', 'Yunhui Guo', 'Jyh-Jing Hwang', 'Tsung-Wei Ke']
2022-04-25
null
http://openaccess.thecvf.com//content/CVPR2022/html/Ke_Unsupervised_Hierarchical_Semantic_Segmentation_With_Multiview_Cosegmentation_and_Clustering_Transformers_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Ke_Unsupervised_Hierarchical_Semantic_Segmentation_With_Multiview_Cosegmentation_and_Clustering_Transformers_CVPR_2022_paper.pdf
cvpr-2022-1
['unsupervised-semantic-segmentation']
['computer-vision']
[ 2.22691372e-01 1.30601630e-01 -3.38130802e-01 -7.73611009e-01 -6.69767916e-01 -9.43081498e-01 5.76822937e-01 3.56854737e-01 -2.82879751e-02 -1.40962541e-01 5.19923031e-01 8.67263079e-02 -1.26183197e-01 -6.34738386e-01 -5.71745813e-01 -5.21842718e-01 -1.80747211e-02 4.71506923e-01 5.47410309e-01 2.49219850e-01 3.35339695e-01 4.64502305e-01 -1.66996717e+00 4.60350275e-01 8.24770987e-01 9.30142820e-01 2.94073373e-01 4.92983878e-01 -1.32083818e-02 6.18405044e-01 -2.06862941e-01 1.49860889e-01 5.01033306e-01 -4.71183538e-01 -1.29170096e+00 1.10821664e+00 7.68471181e-01 6.14364408e-02 6.57942668e-02 1.19926560e+00 -4.55272719e-02 2.06828862e-01 8.32561433e-01 -1.49748433e+00 -4.69987601e-01 4.94293213e-01 -7.98104942e-01 -1.14345014e-01 1.19932868e-01 1.70509845e-01 1.64499199e+00 -7.73124158e-01 8.44470143e-01 1.23546863e+00 5.47327518e-01 2.54639238e-01 -1.64868462e+00 -6.81516081e-02 4.92626727e-01 -1.40333131e-01 -1.44074214e+00 -2.36604095e-01 8.16257954e-01 -1.03343880e+00 6.63547814e-01 4.06693220e-01 6.56241119e-01 4.67765838e-01 -9.28119794e-02 8.69281352e-01 1.25639784e+00 -2.75137246e-01 4.74776506e-01 -1.85938150e-01 6.18503451e-01 7.24250674e-01 1.13413721e-01 -3.51266980e-01 -5.11483669e-01 1.98262334e-01 7.70341098e-01 2.40033671e-01 1.53609604e-01 -9.66226697e-01 -1.44873440e+00 8.25122774e-01 7.03467846e-01 2.36304775e-01 -2.27992550e-01 2.64762253e-01 2.94486582e-01 8.01011547e-02 3.48155886e-01 6.10041261e-01 -5.86324036e-01 3.45385015e-01 -1.15766335e+00 1.13066919e-02 5.38885236e-01 1.15509391e+00 1.31629992e+00 -2.87238121e-01 -4.92037926e-03 9.05825555e-01 3.51652414e-01 8.39818269e-02 4.51908648e-01 -1.49622238e+00 -8.04334134e-02 8.95341992e-01 -3.16970140e-01 -1.17886448e+00 -6.32904589e-01 -1.97981521e-01 -6.76916718e-01 2.71613717e-01 3.46440315e-01 2.72183746e-01 -1.44308960e+00 1.73548508e+00 4.86204743e-01 -2.29876995e-01 -2.90934563e-01 7.87349284e-01 9.84615564e-01 3.18115920e-01 2.58264780e-01 3.66731659e-02 1.51223052e+00 -1.22135711e+00 -3.43346000e-01 -4.00914341e-01 5.43048620e-01 -6.79452717e-01 1.19237196e+00 1.69356033e-01 -9.52378690e-01 -7.74477541e-01 -7.95380116e-01 -3.11238676e-01 -4.93273854e-01 -3.43731999e-01 8.01555336e-01 4.37133998e-01 -1.26712549e+00 4.13022786e-01 -7.73114502e-01 -5.75955749e-01 6.33857608e-01 2.67098129e-01 -5.40603459e-01 1.46789148e-01 -5.11241376e-01 2.23177820e-01 4.63045478e-01 -2.77735382e-01 -7.92300761e-01 -6.64159954e-01 -1.03558433e+00 -2.07146272e-01 4.17146683e-01 -6.58959687e-01 9.12454724e-01 -1.31115878e+00 -8.06648552e-01 1.41899574e+00 -3.72467577e-01 -2.43163466e-01 2.46280223e-01 6.86860234e-02 -2.94664688e-02 2.74857730e-01 7.61296272e-01 1.32318246e+00 8.57267976e-01 -1.84292436e+00 -7.90347040e-01 -5.14186561e-01 3.45504284e-02 3.75652820e-01 8.09092727e-03 -6.04446046e-02 -6.66667998e-01 -7.01090872e-01 9.04198229e-01 -9.64599729e-01 -5.31020999e-01 -1.70050249e-01 -8.49309087e-01 -3.20568800e-01 8.45654190e-01 -4.79309976e-01 9.50551033e-01 -2.25549412e+00 2.18318567e-01 3.38678032e-01 6.42657042e-01 -5.86934090e-01 2.71000881e-02 3.22241038e-01 -1.04488023e-01 4.09986019e-01 -6.11216068e-01 -4.15757746e-01 1.06547333e-01 3.36676866e-01 -5.54776005e-02 6.28241837e-01 4.69984673e-02 1.06209385e+00 -8.23825479e-01 -8.87822628e-01 3.49366188e-01 -1.53944343e-01 -7.83576548e-01 1.58001930e-01 -3.61104399e-01 6.00696385e-01 -3.60826552e-01 9.30924714e-01 5.86771905e-01 -7.14114368e-01 1.13627836e-01 -3.33726764e-01 -1.82353675e-01 7.66941234e-02 -1.27159059e+00 1.81600046e+00 6.64315298e-02 4.67652559e-01 -1.42081961e-01 -1.19011223e+00 7.58554459e-01 -2.42971450e-01 9.23187613e-01 -5.05334854e-01 -6.41985014e-02 -9.36070308e-02 -2.42188960e-01 -2.96108514e-01 5.88728547e-01 -1.12312101e-02 -5.91565430e-01 4.83137846e-01 2.14245424e-01 -4.77893025e-01 2.18649194e-01 4.23067361e-01 8.38635981e-01 -8.97081420e-02 3.94349009e-01 -6.86878562e-01 -7.79878348e-02 3.18855792e-01 7.14533567e-01 9.86887336e-01 -1.84512511e-01 1.11295569e+00 4.58150893e-01 -3.13366592e-01 -1.01388931e+00 -1.34389246e+00 -2.79922783e-01 1.37474370e+00 5.98868608e-01 -4.68641996e-01 -8.54513764e-01 -8.12890112e-01 1.65812701e-01 1.77984327e-01 -7.99357474e-01 3.91560078e-01 -2.34456986e-01 -3.72335255e-01 5.49488328e-02 6.17756248e-01 5.20437121e-01 -1.09755480e+00 -4.44365054e-01 -7.65352100e-02 -1.62178367e-01 -1.22141385e+00 -6.62880957e-01 4.99667108e-01 -8.21764886e-01 -1.19733691e+00 -1.75573468e-01 -1.09407401e+00 1.00811958e+00 5.92858672e-01 1.41258705e+00 2.38504723e-01 -4.52621192e-01 6.40287638e-01 -4.10127133e-01 -8.25429335e-03 -8.14741179e-02 1.16312638e-01 -1.59923106e-01 8.20163414e-02 2.80320168e-01 -5.20202041e-01 -6.89239502e-01 5.85040867e-01 -9.30013597e-01 2.94642657e-01 2.45821819e-01 4.91477281e-01 1.12819922e+00 7.63654336e-02 2.00354472e-01 -1.24638891e+00 -3.65391024e-03 -4.00248140e-01 -4.69039679e-01 4.79921438e-02 -3.64107668e-01 -3.54260743e-01 3.00921977e-01 1.24213085e-01 -6.82077408e-01 3.99568707e-01 3.10125649e-01 -2.66271919e-01 -7.54438937e-01 2.17420831e-01 -5.06165028e-01 2.09979698e-01 5.28507113e-01 1.55514374e-01 -1.57685101e-01 -4.31391597e-01 8.13680589e-01 3.52189034e-01 5.41704297e-01 -6.73890769e-01 8.52180839e-01 9.64875937e-01 -1.43525884e-01 -8.16086531e-01 -1.14642704e+00 -1.11208272e+00 -1.38401103e+00 -1.90605432e-01 1.31594896e+00 -1.01374269e+00 -9.71095189e-02 4.95545417e-01 -6.97704494e-01 -6.80192053e-01 -7.28865802e-01 1.12942224e-02 -8.49546492e-01 5.03004670e-01 -3.05281460e-01 -1.51228070e-01 2.84626484e-01 -1.11867559e+00 1.37713063e+00 1.51701719e-01 -5.74557900e-01 -1.19467390e+00 -1.44458562e-01 6.15461886e-01 -1.50490075e-01 5.54854274e-01 6.77956998e-01 -5.58295369e-01 -7.74680376e-01 2.04028577e-01 -4.40484911e-01 3.97704154e-01 4.78025377e-01 1.10806175e-01 -8.84083688e-01 -1.58810690e-01 -5.19105569e-02 -2.78372556e-01 1.03247607e+00 6.96963251e-01 1.47219241e+00 -2.83724576e-01 -2.27119088e-01 9.00514960e-01 1.56370437e+00 -6.71963841e-02 3.40275288e-01 2.82773972e-01 1.39294827e+00 9.20925975e-01 5.48717022e-01 2.03188673e-01 6.31737113e-01 6.09563231e-01 3.39559436e-01 -4.85261321e-01 -2.63928026e-01 -2.40178585e-01 -7.42333829e-02 8.26157987e-01 2.13685796e-01 1.56478405e-01 -1.12999535e+00 8.18382084e-01 -1.86528909e+00 -8.22936654e-01 -4.99701828e-01 1.84908974e+00 7.93475270e-01 1.95201598e-02 5.32433093e-01 1.17955105e-02 8.42769623e-01 3.15763950e-01 -5.19309878e-01 -2.07065746e-01 -1.59605652e-01 -1.17815964e-01 5.37182212e-01 6.06438637e-01 -1.54025638e+00 1.31339216e+00 6.24519873e+00 7.31309652e-01 -7.04717398e-01 4.10726666e-02 1.04447401e+00 3.42821866e-01 -4.67159420e-01 2.64517754e-01 -6.61283493e-01 3.88644785e-01 1.59331411e-01 1.03928737e-01 1.10082351e-01 9.23016250e-01 1.95154548e-01 -2.32345432e-01 -1.20929062e+00 6.95216060e-01 -5.90190552e-02 -1.23956299e+00 4.26690504e-02 2.31517449e-01 1.31301391e+00 3.64675894e-02 5.19984923e-02 -3.17829609e-01 7.22955227e-01 -1.26500845e+00 9.54657733e-01 2.87986994e-01 5.07991433e-01 -4.80716586e-01 2.11749941e-01 6.63391314e-03 -1.50659323e+00 1.50137499e-01 -1.74256608e-01 9.52094421e-02 4.11769422e-03 5.87554395e-01 -2.96606421e-01 2.74564385e-01 8.24216306e-01 1.15707862e+00 -9.81230497e-01 8.53147984e-01 -1.45381272e-01 6.15201890e-01 -1.91339687e-01 5.69635808e-01 5.63543320e-01 -2.76282728e-01 3.19865346e-01 1.41934466e+00 -2.57920593e-01 -7.93801025e-02 7.61406302e-01 7.35236168e-01 7.42605105e-02 7.12718964e-02 -4.59793150e-01 1.44947544e-01 5.27252913e-01 1.21543610e+00 -1.55164635e+00 -4.52475607e-01 -3.73372763e-01 1.00507557e+00 1.74027875e-01 3.89985591e-01 -6.69378579e-01 5.07628694e-02 7.80571997e-01 3.48914504e-01 4.92954135e-01 -4.61000234e-01 -9.64066267e-01 -1.04574335e+00 -9.16911736e-02 -7.45485723e-01 7.13836730e-01 -5.59038341e-01 -1.40148032e+00 1.06760681e-01 7.33461529e-02 -1.37448895e+00 5.06622531e-02 -3.61580014e-01 -3.84643823e-01 3.31727237e-01 -1.18940270e+00 -1.41942191e+00 -5.25112689e-01 6.79852784e-01 9.18201804e-01 1.98714256e-01 6.31538689e-01 -1.98492408e-01 -3.43407840e-01 2.33690590e-01 -9.81442183e-02 3.33345860e-01 4.58657324e-01 -1.73753631e+00 4.51505929e-01 8.48491967e-01 4.90696073e-01 7.03133166e-01 4.88138497e-01 -5.85520267e-01 -9.76121128e-01 -1.34993124e+00 7.83009052e-01 -7.33242571e-01 7.34205782e-01 -6.19283974e-01 -9.77926016e-01 8.42562437e-01 1.02118790e-01 1.94411382e-01 7.71137357e-01 2.76195198e-01 -7.10449159e-01 2.05430046e-01 -9.68845248e-01 4.96259868e-01 1.37559867e+00 -5.85212529e-01 -5.43528497e-01 4.71906781e-01 8.32195878e-01 -1.65550098e-01 -1.04347348e+00 3.29768598e-01 2.01668844e-01 -1.22431350e+00 1.05224121e+00 -4.73167717e-01 5.33345461e-01 -6.73846424e-01 -4.00516927e-01 -1.05661845e+00 -6.19242191e-01 -3.25924993e-01 3.54104251e-01 1.26808727e+00 2.67497480e-01 -4.38083678e-01 9.21909988e-01 5.99331796e-01 -2.67339915e-01 -5.24651289e-01 -5.62565327e-01 -9.01798785e-01 1.54501855e-01 -4.43667084e-01 3.67973655e-01 1.28927600e+00 -1.38569787e-01 1.44303337e-01 1.14257604e-01 3.78503740e-01 1.08498096e+00 5.21628320e-01 9.24347937e-01 -1.27324545e+00 -1.31943777e-01 -6.22922182e-01 -8.04223239e-01 -1.04985392e+00 7.42690340e-02 -1.02570593e+00 2.71529973e-01 -1.93920696e+00 4.95387405e-01 -5.65533936e-01 -1.84554532e-01 6.09738171e-01 -1.13593593e-01 6.02394223e-01 1.51909247e-01 5.75153232e-01 -1.21525013e+00 1.19908348e-01 1.14482594e+00 -1.13646835e-01 -2.14356929e-01 -3.82527679e-01 -1.02151191e+00 9.57043886e-01 9.25234199e-01 -4.21854854e-01 -4.21932995e-01 -3.84491175e-01 -3.51775736e-01 -5.39377093e-01 6.38721526e-01 -9.50701475e-01 2.55306870e-01 -3.96610439e-01 3.90163809e-01 -6.24406159e-01 -2.73401085e-02 -7.83234417e-01 1.65594459e-01 1.25836000e-01 -4.41230863e-01 -2.83506066e-01 -1.82032958e-01 4.48461831e-01 -3.44956845e-01 9.38301608e-02 1.01847005e+00 -3.24487597e-01 -1.33874238e+00 3.76181811e-01 -2.96804816e-01 4.93716002e-01 1.02653599e+00 -7.36266911e-01 -1.64941162e-01 -1.70569897e-01 -1.09921098e+00 3.73694211e-01 1.08914161e+00 4.14197505e-01 4.73085821e-01 -1.25232017e+00 -3.97976071e-01 1.87431589e-01 4.84688044e-01 3.95713985e-01 3.24829757e-01 6.37428522e-01 -3.47822517e-01 6.75513223e-02 -1.85087070e-01 -1.24840117e+00 -1.39893317e+00 3.89993578e-01 1.28673434e-01 4.20543998e-02 -8.44935298e-01 9.44371581e-01 9.13352311e-01 -5.09240866e-01 1.68299749e-01 -3.01966876e-01 -5.19039705e-02 2.14988723e-01 -7.40366150e-03 5.36661148e-02 -2.74252295e-01 -1.02962661e+00 -3.44794273e-01 1.01800787e+00 -3.26380581e-02 2.93976627e-02 1.17853153e+00 -4.13871557e-01 -3.91541451e-01 7.33440459e-01 1.45905650e+00 -1.09117135e-01 -1.59889197e+00 -3.11841637e-01 3.90272468e-01 -5.70214987e-01 -9.76211652e-02 -4.31304932e-01 -1.09989369e+00 5.05520761e-01 2.29194835e-01 6.35778129e-01 9.77993309e-01 7.54276574e-01 3.54245424e-01 -8.23718309e-02 3.51986051e-01 -1.30378509e+00 4.13778275e-01 4.12426472e-01 6.57804966e-01 -1.49812162e+00 1.76715225e-01 -8.42641830e-01 -8.36558998e-01 6.14724278e-01 6.90820396e-01 -3.70073676e-01 7.14195490e-01 1.25545338e-01 1.42214805e-01 -7.10412204e-01 -2.54797965e-01 -7.09739149e-01 6.58693671e-01 7.39821136e-01 2.74542511e-01 4.37436789e-01 6.19745888e-02 3.83105129e-01 -3.81934553e-01 -5.60531139e-01 1.92848787e-01 8.58296037e-01 -7.42569447e-01 -8.90893042e-01 -1.35021836e-01 5.49612045e-01 -1.58683285e-01 -5.35491444e-02 -5.96521676e-01 9.49872315e-01 3.30873638e-01 8.96086216e-01 4.43121165e-01 -2.90021926e-01 1.85241178e-01 -2.13316023e-01 2.50506938e-01 -1.10239232e+00 -3.48616302e-01 3.41622204e-01 -7.91984573e-02 -9.30599451e-01 -6.52265131e-01 -9.37842548e-01 -1.48259735e+00 7.69376382e-02 9.21906456e-02 -1.21737085e-01 3.65304917e-01 8.03573072e-01 2.39981025e-01 3.84133130e-01 8.77918184e-01 -7.92310953e-01 1.85911074e-01 -5.45802653e-01 -8.12812984e-01 1.00463724e+00 2.47132719e-01 -5.21025181e-01 -5.69660008e-01 8.04628968e-01]
[9.550002098083496, 0.7656238675117493]
a9870123-079b-4f22-9d2b-8d38517c752d
deep-patch-visual-odometry
2208.04726
null
https://arxiv.org/abs/2208.04726v2
https://arxiv.org/pdf/2208.04726v2.pdf
Deep Patch Visual Odometry
We propose Deep Patch Visual Odometry (DPVO), a new deep learning system for monocular Visual Odometry (VO). DPVO uses a novel recurrent network architecture designed for tracking image patches across time. Recent approaches to VO have significantly improved the state-of-the-art accuracy by using deep networks to predict dense flow between video frames. However, using dense flow incurs a large computational cost, making these previous methods impractical for many use cases. Despite this, it has been assumed that dense flow is important as it provides additional redundancy against incorrect matches. DPVO disproves this assumption, showing that it is possible to get the best accuracy and efficiency by exploiting the advantages of sparse patch-based matching over dense flow. DPVO introduces a novel recurrent update operator for patch based correspondence coupled with differentiable bundle adjustment. On Standard benchmarks, DPVO outperforms all prior work, including the learning-based state-of-the-art VO-system (DROID) using a third of the memory while running 3x faster on average. Code is available at https://github.com/princeton-vl/DPVO
['Jia Deng', 'Lahav Lipson', 'Zachary Teed']
2022-08-08
null
null
null
null
['monocular-visual-odometry']
['robots']
[-5.15850902e-01 -1.61450192e-01 -4.63517547e-01 3.76955755e-02 -2.75625050e-01 -2.69439965e-01 4.13922459e-01 -2.29562998e-01 -3.00546944e-01 4.72965330e-01 3.09176385e-01 -1.52483672e-01 1.77340254e-01 -7.38568842e-01 -9.84715044e-01 -3.12578857e-01 -2.60053277e-01 3.42937797e-01 4.24774885e-01 -1.38602048e-01 1.76641747e-01 7.03289866e-01 -1.70916188e+00 -5.60156554e-02 2.99624234e-01 8.45460355e-01 1.93456903e-01 8.55674982e-01 1.76384002e-02 1.07293904e+00 -1.97211266e-01 -3.02116603e-01 8.89796793e-01 -3.02644134e-01 -8.17277372e-01 -5.32589294e-02 1.28777087e+00 -6.64519250e-01 -9.26185250e-01 8.81439328e-01 6.34890735e-01 1.24238886e-01 1.60116836e-01 -1.52940381e+00 -5.93091428e-01 1.00705465e-02 -5.19047916e-01 3.37876707e-01 3.61545175e-01 2.88105279e-01 1.33353758e+00 -9.88533437e-01 1.26140535e+00 1.14572656e+00 9.91885364e-01 3.42561632e-01 -1.32050049e+00 -4.68702406e-01 -2.09314004e-02 3.35696995e-01 -1.43358481e+00 -6.17423773e-01 5.76532602e-01 -6.01524293e-01 1.38727367e+00 -5.14337160e-02 1.05946970e+00 8.28127086e-01 2.97568858e-01 9.09591615e-01 6.56843066e-01 -1.44685313e-01 -2.37966225e-01 -1.81967318e-01 -1.29449606e-01 1.09555256e+00 3.46668363e-01 5.57474136e-01 -6.95586085e-01 3.55933644e-02 1.15690327e+00 1.67001337e-01 -5.40419459e-01 -1.00626338e+00 -1.20665574e+00 8.45413387e-01 8.22849631e-01 3.13210748e-02 -3.26756388e-01 5.44950366e-01 3.23518902e-01 5.94097078e-01 4.59461391e-01 3.53652239e-01 -2.96577781e-01 -3.02775800e-01 -9.89188910e-01 4.81061876e-01 8.86011422e-01 9.18984711e-01 1.17622149e+00 1.91116825e-01 1.36348292e-01 4.45129782e-01 3.59007388e-01 4.56160098e-01 2.85658747e-01 -1.56959748e+00 1.99505523e-01 4.13117439e-01 9.26731825e-02 -1.38475370e+00 -4.49974120e-01 -5.47034562e-01 -7.07811952e-01 5.94579399e-01 4.01144981e-01 6.69707805e-02 -8.33867610e-01 1.60729933e+00 3.92484337e-01 5.73670506e-01 -1.53251380e-01 1.24714136e+00 8.88300240e-01 4.84017998e-01 -4.63512152e-01 1.27705932e-01 8.54151845e-01 -1.21356428e+00 -4.56331581e-01 -1.29736587e-01 8.06961119e-01 -9.03058469e-01 8.09202313e-01 2.36820996e-01 -9.77744758e-01 -4.97084439e-01 -1.19742942e+00 -5.19786119e-01 -2.08449308e-02 -2.71074623e-01 8.18027794e-01 3.96456957e-01 -1.46658075e+00 8.88785541e-01 -1.00523055e+00 -5.31375945e-01 4.77466106e-01 3.20394546e-01 -5.83692551e-01 -1.90290377e-01 -8.89432311e-01 8.42634916e-01 -2.62202084e-01 -8.16143602e-02 -8.45832467e-01 -1.05691838e+00 -1.21236742e+00 -7.19703808e-02 4.62935716e-02 -1.21928263e+00 1.23029411e+00 -8.31377029e-01 -1.40903306e+00 9.36141253e-01 -4.62579519e-01 -7.91177869e-01 8.96863282e-01 -4.74784106e-01 2.74894759e-02 2.64095962e-01 1.83484793e-01 1.07991576e+00 7.49287367e-01 -9.95526910e-01 -5.84254980e-01 -1.96463130e-02 2.47669131e-01 2.95978934e-01 1.08923979e-01 -4.75220352e-01 -5.01203001e-01 -3.60906243e-01 2.07407281e-01 -1.11247802e+00 -1.71756297e-01 6.99675500e-01 -3.64556722e-02 1.30875281e-03 8.19569588e-01 -5.61802149e-01 8.34631145e-01 -1.98272014e+00 1.61792502e-01 -1.51216447e-01 6.52100921e-01 2.50974923e-01 -2.64746696e-02 3.15918565e-01 1.20913669e-01 -3.13159049e-01 8.00390020e-02 -8.97048116e-01 -2.05017277e-04 3.74325991e-01 -2.61437982e-01 9.89946783e-01 7.13430569e-02 9.74452972e-01 -9.28918362e-01 -2.99352378e-01 5.50230384e-01 7.34379888e-01 -8.93041909e-01 2.39287421e-01 -1.14299774e-01 3.79204094e-01 2.08489090e-01 5.76484859e-01 6.39293671e-01 -5.07400572e-01 -1.85304120e-01 -2.95673043e-01 -2.62115210e-01 4.82227236e-01 -1.45122087e+00 2.03077841e+00 -3.73393834e-01 1.25464702e+00 -1.34527326e-01 -7.21798897e-01 6.23684824e-01 1.18585333e-01 8.05073142e-01 -7.18351841e-01 1.19345844e-01 2.52542138e-01 -2.77207553e-01 -2.41624773e-01 8.39127660e-01 1.79136097e-01 5.13912737e-01 3.23056251e-01 1.40755951e-01 -7.45180175e-02 1.28424868e-01 3.08294266e-01 1.07079506e+00 5.89189649e-01 2.44446471e-01 -1.92082465e-01 2.00813949e-01 3.77074569e-01 8.67694676e-01 5.94008148e-01 -4.77439046e-01 8.50574911e-01 2.21368089e-01 -8.40811670e-01 -1.26750708e+00 -1.02788961e+00 -5.49817905e-02 6.23212874e-01 5.34151018e-01 -4.68430966e-01 -3.20404619e-01 -2.03413725e-01 3.44920456e-01 -1.31236181e-01 -5.27339697e-01 1.76177844e-01 -6.69683397e-01 -1.53511241e-01 5.05482912e-01 4.61268038e-01 5.98902047e-01 -7.76366651e-01 -8.23715448e-01 2.76449919e-01 -1.30742058e-01 -1.24600708e+00 -6.46853983e-01 -2.08749592e-01 -8.72522235e-01 -1.21580231e+00 -8.00867915e-01 -7.10903704e-01 3.15216452e-01 8.45827162e-01 1.38489270e+00 3.26172471e-01 -3.78210872e-01 4.73030776e-01 -5.71830384e-02 -1.40561825e-02 -1.60659552e-01 1.63994491e-01 1.43038109e-01 -2.66955703e-01 3.05714339e-01 -7.35656857e-01 -8.69754016e-01 2.18793273e-01 -5.37198663e-01 6.74141720e-02 1.84279382e-01 9.38029528e-01 6.26975596e-01 -7.12683499e-01 -8.38642195e-02 -5.04598618e-01 9.01343580e-03 -2.88974047e-01 -9.11065459e-01 -2.40428150e-01 -6.02518499e-01 1.52703404e-01 3.14150661e-01 -2.15982705e-01 -4.77813631e-01 1.44939572e-01 -1.94071025e-01 -1.12212551e+00 1.49348870e-01 9.44578722e-02 4.27498490e-01 -6.43638194e-01 7.25066543e-01 -1.22218931e-04 4.17705983e-01 -3.35199863e-01 3.89890432e-01 1.67003453e-01 6.45338178e-01 -1.86140705e-02 7.47208714e-01 1.03023350e+00 2.51578420e-01 -8.27712834e-01 -8.45735908e-01 -8.76820326e-01 -6.45626605e-01 -1.48748323e-01 7.93559968e-01 -1.33959007e+00 -9.39347446e-01 4.30415303e-01 -1.18756056e+00 -4.81137663e-01 -3.68270248e-01 6.85757995e-01 -7.27190137e-01 5.14585018e-01 -7.76081324e-01 -2.13510156e-01 -3.47279996e-01 -1.17584932e+00 1.05695367e+00 -4.55695111e-03 -2.13187650e-01 -1.10700250e+00 4.74744916e-01 1.47967990e-02 4.21866745e-01 2.78103054e-01 -8.84914473e-02 1.66873649e-01 -1.23961031e+00 1.71345651e-01 -2.92998582e-01 1.40678003e-01 -3.06832250e-02 2.20572464e-02 -9.79966223e-01 -5.79213500e-01 -2.95285761e-01 -9.91652012e-02 1.00358915e+00 5.21004856e-01 4.64220285e-01 -7.67056271e-02 -2.30554953e-01 1.35059893e+00 1.60173750e+00 -4.46004152e-01 6.71569705e-01 6.46976948e-01 1.17308891e+00 1.82267323e-01 3.71838868e-01 2.35720053e-01 7.67568469e-01 9.01563823e-01 7.00919807e-01 -2.51871973e-01 -4.64204133e-01 -2.61288822e-01 4.26250547e-01 6.46992624e-01 -5.65622980e-03 8.87277946e-02 -7.55256891e-01 9.12089169e-01 -2.06314397e+00 -1.07506669e+00 -1.81424141e-01 2.16269493e+00 4.18492854e-01 5.93527518e-02 2.56362259e-01 -2.43628770e-02 3.36867899e-01 4.90926623e-01 -4.81906056e-01 -2.29027212e-01 -2.38057718e-01 1.46536618e-01 9.42249417e-01 9.31106091e-01 -1.16215479e+00 1.07990861e+00 5.89575577e+00 1.09322637e-01 -1.49148631e+00 2.19564959e-01 8.44132751e-02 -2.82384962e-01 -1.41601324e-01 1.67084441e-01 -9.92069602e-01 1.22991689e-01 4.85720098e-01 -1.49958506e-01 4.61672693e-01 9.03696179e-01 1.49754539e-01 -5.12780510e-02 -9.56594467e-01 1.29573858e+00 2.31142357e-01 -1.90407872e+00 -1.02530606e-01 2.49241769e-01 8.23443711e-01 8.61114264e-01 -1.72128752e-01 -1.11462682e-01 5.03827691e-01 -7.70047545e-01 7.37242579e-01 3.83565903e-01 5.87444007e-01 -3.94933879e-01 5.56415141e-01 1.01443781e-02 -1.43096113e+00 2.24592596e-01 -6.96343899e-01 -4.29465324e-01 3.98844808e-01 6.18021786e-01 -7.38202572e-01 5.71038723e-01 9.72412884e-01 1.48402870e+00 -3.82994294e-01 1.34731483e+00 -1.09366432e-01 1.46041811e-01 -4.96188551e-01 4.42878336e-01 4.30971086e-01 -4.77359332e-02 8.36982369e-01 1.03557730e+00 2.88874775e-01 -3.95062774e-01 3.27650458e-01 7.04616845e-01 -1.77960709e-01 -9.73776802e-02 -1.02754033e+00 4.45089430e-01 4.08094823e-01 9.78015721e-01 -3.32808763e-01 -3.91957879e-01 -5.21327972e-01 1.00478995e+00 5.13084173e-01 9.62818116e-02 -5.44064283e-01 -2.31609687e-01 1.42836297e+00 1.52297243e-01 6.78723633e-01 -5.69917619e-01 -1.38240561e-01 -1.50182843e+00 5.81716038e-02 -6.57921731e-01 3.57032984e-01 -7.56367743e-01 -9.81775582e-01 4.26360220e-01 -4.81208593e-01 -1.62107813e+00 -4.36520338e-01 -5.26278496e-01 -1.65309995e-01 7.51364291e-01 -1.90807283e+00 -9.68283117e-01 -6.98594332e-01 7.09539890e-01 3.92499417e-01 1.18964180e-01 5.24496019e-01 6.09362066e-01 -1.89844996e-01 4.17579710e-01 -9.57924351e-02 2.33898073e-01 9.47898328e-01 -1.17216754e+00 8.20656240e-01 1.13106787e+00 4.30981666e-01 3.79749388e-01 7.48841405e-01 -3.99157435e-01 -1.57369351e+00 -1.06878471e+00 1.01239669e+00 -5.97671747e-01 5.56648195e-01 -1.62276357e-01 -8.92327070e-01 1.00413549e+00 2.81361490e-01 5.55458128e-01 1.75706167e-02 -2.24021509e-01 -5.81821263e-01 -1.69933751e-01 -9.30213451e-01 6.42013371e-01 1.52575564e+00 -6.47156954e-01 -3.79409790e-01 1.97739154e-01 9.29474413e-01 -9.24611032e-01 -9.95647311e-01 2.22826570e-01 6.48074746e-01 -1.34529459e+00 1.15608144e+00 -2.70168513e-01 3.14307362e-01 -6.62024140e-01 -1.13162287e-01 -1.23120534e+00 -4.15847957e-01 -8.37517917e-01 -3.88833910e-01 5.74978888e-01 -1.01947777e-01 -8.14496398e-01 9.09325540e-01 1.22475311e-01 -2.58282185e-01 -5.87714791e-01 -8.73068094e-01 -8.75690043e-01 -2.62984037e-01 -4.18973446e-01 4.73040223e-01 1.16615832e+00 -3.07514310e-01 8.69193524e-02 -6.60907090e-01 3.10637027e-01 8.44298244e-01 1.41071230e-01 1.31424391e+00 -1.07049406e+00 -3.54641795e-01 -4.55667496e-01 -1.10073245e+00 -1.59102142e+00 1.15245417e-01 -8.79533648e-01 -1.68627337e-01 -1.55546248e+00 -3.94426823e-01 -3.84654045e-01 1.37805626e-01 3.51256698e-01 1.11287437e-01 7.50413120e-01 4.95500743e-01 5.01812994e-01 -4.38647240e-01 3.94988775e-01 1.25007474e+00 -4.39352309e-03 -2.75387585e-01 -3.31187725e-01 -2.16722116e-01 5.00873804e-01 6.20483935e-01 -4.23428625e-01 -1.76735982e-01 -7.93488801e-01 2.70946831e-01 -5.89640923e-02 6.70869589e-01 -1.16719496e+00 4.89648879e-01 2.26478755e-01 1.68149397e-01 -6.41129255e-01 5.08186579e-01 -6.57830060e-01 4.43257511e-01 7.92162359e-01 1.32983744e-01 4.15662974e-01 1.45873219e-01 4.43547755e-01 -2.38154113e-01 2.64696300e-01 8.12902331e-01 -1.30392462e-01 -1.11333406e+00 7.26656377e-01 5.30874589e-03 1.49569124e-01 6.70001328e-01 -3.17127049e-01 -4.79501307e-01 -5.00828147e-01 -5.31905353e-01 1.41676575e-01 9.51206625e-01 6.01859987e-01 5.20300627e-01 -1.35449636e+00 -5.71956098e-01 3.60196322e-01 9.76250693e-02 4.51290086e-02 1.52478695e-01 1.07763875e+00 -1.10804045e+00 3.07370126e-01 -4.27305222e-01 -1.19665134e+00 -1.31151021e+00 4.10970032e-01 7.41338432e-01 -7.43148103e-02 -1.26893997e+00 7.88558304e-01 6.33729398e-02 -3.67004961e-01 2.94999868e-01 -2.78752327e-01 1.58661991e-01 -2.52643734e-01 3.17434400e-01 4.31096524e-01 1.02108829e-01 -7.44573951e-01 -3.01544994e-01 8.91678214e-01 1.58754721e-01 9.69890505e-02 1.28025150e+00 -2.61638820e-01 -4.43091318e-02 3.07806581e-01 1.45612276e+00 -1.65700570e-01 -1.70464444e+00 -5.00892699e-01 -2.16523498e-01 -8.71647418e-01 1.80876136e-01 4.22133207e-02 -1.36000228e+00 8.92433286e-01 5.62987566e-01 -1.41417444e-01 7.48869896e-01 -3.03336352e-01 1.04899406e+00 3.86808664e-01 5.23795009e-01 -5.75093210e-01 -7.94141367e-02 6.80710971e-01 6.41199291e-01 -1.37413657e+00 2.59339839e-01 -2.99005687e-01 -3.07645887e-01 9.22247052e-01 4.73401159e-01 -6.14518046e-01 8.22423637e-01 1.15052544e-01 3.88773829e-01 -2.76414871e-01 -7.27468491e-01 -3.34803522e-01 2.09004328e-01 5.02726853e-01 2.85896212e-01 -2.10132301e-01 4.74099033e-02 -6.53034270e-01 -2.69686103e-01 1.70495361e-01 6.43070698e-01 9.87133145e-01 -2.10217744e-01 -9.30409431e-01 -2.10691765e-01 2.03808039e-01 -3.05186480e-01 -1.28146052e-01 1.25086578e-02 1.03356004e+00 -2.02348784e-01 4.97787863e-01 3.84106398e-01 -3.24682176e-01 3.01461220e-01 -4.11699474e-01 5.94185054e-01 -2.88830280e-01 -5.48695505e-01 -1.82407528e-01 5.17841764e-02 -1.27341402e+00 -8.34645748e-01 -6.00139201e-01 -9.73222315e-01 -9.16063070e-01 3.06994747e-02 -2.78680235e-01 4.33474422e-01 7.36108422e-01 8.43824029e-01 1.99664399e-01 4.54429150e-01 -1.20265210e+00 -2.12883085e-01 -5.17907500e-01 -2.60215193e-01 4.66558188e-01 9.18775439e-01 -9.18729722e-01 -4.17885005e-01 3.18048447e-02]
[8.481270790100098, -2.0686161518096924]
7d4860d7-da2f-44eb-931a-ece06c41050d
safe-exploration-by-solving-early-terminated
2107.042
null
https://arxiv.org/abs/2107.04200v1
https://arxiv.org/pdf/2107.04200v1.pdf
Safe Exploration by Solving Early Terminated MDP
Safe exploration is crucial for the real-world application of reinforcement learning (RL). Previous works consider the safe exploration problem as Constrained Markov Decision Process (CMDP), where the policies are being optimized under constraints. However, when encountering any potential dangers, human tends to stop immediately and rarely learns to behave safely in danger. Motivated by human learning, we introduce a new approach to address safe RL problems under the framework of Early Terminated MDP (ET-MDP). We first define the ET-MDP as an unconstrained MDP with the same optimal value function as its corresponding CMDP. An off-policy algorithm based on context models is then proposed to solve the ET-MDP, which thereby solves the corresponding CMDP with better asymptotic performance and improved learning efficiency. Experiments on various CMDP tasks show a substantial improvement over previous methods that directly solve CMDP.
['Bolei Zhou', 'Bo Dai', 'Jiadong Guo', 'Zhenghao Peng', 'Meng Fang', 'Ziping Xu', 'Hao Sun']
2021-07-09
null
null
null
null
['safe-exploration']
['robots']
[-9.51460078e-02 5.63153148e-01 -5.98478854e-01 2.26001382e-01 -8.84556949e-01 -5.02364576e-01 4.90724802e-01 4.87429984e-02 -8.39140892e-01 1.22336066e+00 2.86476631e-02 -6.01345062e-01 -1.48597956e-01 -7.29305148e-01 -5.53960085e-01 -1.02751529e+00 -4.69617665e-01 3.04869145e-01 2.01330945e-01 -4.20923457e-02 3.89094472e-01 2.84886748e-01 -1.13296151e+00 -2.94030398e-01 9.95553374e-01 8.65658700e-01 3.36641908e-01 5.78379452e-01 1.04780972e-01 9.92559671e-01 -4.40706134e-01 6.12212382e-02 2.84116536e-01 -4.07441199e-01 -1.11876953e+00 -1.65662453e-01 -5.23907721e-01 -7.49340117e-01 -1.17364638e-01 1.19842458e+00 4.66837108e-01 6.09375358e-01 4.55328226e-01 -1.37789369e+00 -9.27468762e-02 6.63141906e-01 -5.78044057e-01 -4.04260345e-02 5.89429140e-01 4.72687870e-01 7.91557610e-01 -2.07708433e-01 5.16936064e-01 1.47064435e+00 -8.52756053e-02 1.05499578e+00 -1.33865905e+00 -4.83321846e-01 7.62023091e-01 1.79925159e-01 -9.74947035e-01 1.94693431e-01 4.45275128e-01 -7.31193051e-02 9.17551756e-01 -1.13223605e-01 8.80581141e-01 1.47890627e+00 5.52070677e-01 1.30529201e+00 1.47132707e+00 -3.43275756e-01 9.98281360e-01 -1.63103268e-01 -2.66885012e-01 4.75794882e-01 1.99156955e-01 8.26752067e-01 -4.17942971e-01 -2.99852252e-01 6.32968664e-01 -3.32288325e-01 -7.99465477e-02 -7.80537546e-01 -1.05102229e+00 9.82817829e-01 -9.53214243e-02 -6.88814297e-02 -5.32253742e-01 2.64401615e-01 4.29803312e-01 4.68195796e-01 -7.24820644e-02 4.39025074e-01 -2.07052007e-01 -4.50732470e-01 -5.29089212e-01 8.70777667e-01 1.05334961e+00 6.91212773e-01 2.73557544e-01 1.45925805e-01 -2.88716704e-01 2.50669032e-01 3.49726468e-01 5.06018758e-01 1.48903400e-01 -1.31850970e+00 4.62818772e-01 3.81050631e-03 9.00428116e-01 -4.13452059e-01 -2.10180640e-01 -2.64589399e-01 -3.04835051e-01 8.88669550e-01 3.88390750e-01 -4.21938390e-01 -5.80066085e-01 1.89478397e+00 5.10619402e-01 1.26557186e-01 6.33833230e-01 6.86052144e-01 -1.59521088e-01 8.29376996e-01 3.17547023e-01 -8.06743920e-01 7.79824853e-01 -6.76973462e-01 -8.79119635e-01 -2.23120019e-01 5.10382771e-01 4.71876971e-02 1.25515938e+00 1.08774865e+00 -1.04380786e+00 -3.05530131e-02 -1.22694623e+00 5.67320883e-01 9.13258791e-02 -6.84087157e-01 3.31543624e-01 3.88105512e-01 -7.59219885e-01 7.66752779e-01 -9.34364438e-01 -6.25578836e-02 3.89851153e-01 3.17594439e-01 1.96556732e-01 9.08025950e-02 -1.27599835e+00 1.11255193e+00 9.10934210e-01 -7.73704946e-02 -2.00551867e+00 -1.35243297e-01 -8.01611722e-01 -2.14994103e-01 1.26573253e+00 -1.19032376e-01 1.71804821e+00 -5.70112586e-01 -2.04812598e+00 4.04029638e-01 2.27420345e-01 -5.87924123e-01 8.82067978e-01 -7.66893864e-01 -1.95854977e-01 1.32825866e-01 3.87584022e-03 4.76688564e-01 1.01683939e+00 -1.43498719e+00 -8.79075050e-01 -6.61398917e-02 4.40780759e-01 6.06058836e-01 5.82151562e-02 -1.76146761e-01 1.30780205e-01 -3.81264895e-01 -3.46690983e-01 -1.07498217e+00 -6.69996142e-01 -3.19099486e-01 -3.48428965e-01 -4.25650686e-01 7.41222680e-01 -2.45034501e-01 1.30784559e+00 -1.97390044e+00 4.49328125e-01 1.13611467e-01 -3.24529186e-02 2.47419968e-01 -1.16850078e-01 4.70338076e-01 2.95874625e-01 -4.38624807e-02 -4.00479823e-01 2.94859614e-02 1.91979393e-01 6.85049713e-01 -7.33091295e-01 3.73794377e-01 -1.58559740e-01 5.68314254e-01 -1.40839219e+00 -6.40926301e-01 3.58637199e-02 -1.15294650e-01 -4.94684130e-01 7.09533632e-01 -6.26699209e-01 6.37723684e-01 -8.85683537e-01 6.63108170e-01 3.70218843e-01 4.71906602e-01 4.61246908e-01 7.18200147e-01 -2.47705594e-01 -1.30360872e-01 -1.32065332e+00 1.40160036e+00 -4.18613285e-01 6.97233453e-02 1.98528081e-01 -8.98977935e-01 6.41575575e-01 3.16221714e-01 3.53456914e-01 -7.25817025e-01 7.80867860e-02 9.25983128e-04 -1.83897749e-01 -4.34296757e-01 4.04425085e-01 -4.55528855e-01 -3.78882885e-01 6.99575365e-01 -1.80670321e-01 -2.55961359e-01 -8.56904462e-02 9.75449830e-02 1.00627875e+00 8.85783434e-01 7.11756706e-01 -4.05835658e-01 4.86852378e-01 -4.01382484e-02 9.58375812e-01 1.22812116e+00 -7.33841538e-01 -2.63768315e-01 1.10293078e+00 -3.98389548e-01 -3.91617835e-01 -1.21161115e+00 1.50034860e-01 1.08055305e+00 4.24458176e-01 -1.71437338e-01 -8.18240345e-01 -1.17228723e+00 -8.18946585e-02 1.36200595e+00 -4.98138309e-01 -2.52729923e-01 -6.25157773e-01 -4.45875674e-01 2.63630837e-01 2.61829942e-01 5.92777073e-01 -1.49233544e+00 -1.30948508e+00 3.49225342e-01 -8.22350308e-02 -7.05958128e-01 -1.92271009e-01 4.77028191e-01 -8.59919727e-01 -9.25110400e-01 -5.21667719e-01 -4.25674140e-01 3.68587226e-01 -2.22560152e-01 8.12418818e-01 -2.37012997e-01 1.94965795e-01 4.90822732e-01 -1.78043351e-01 -4.87265378e-01 -4.96044606e-01 -2.53354400e-01 4.12222624e-01 -2.59144306e-01 2.49384537e-01 -3.89706492e-01 -4.84531760e-01 1.44609988e-01 -8.05095375e-01 -4.82706465e-02 3.59910160e-01 8.53947639e-01 7.56742060e-01 3.29519570e-01 6.74725771e-01 -6.29487813e-01 1.00724804e+00 -5.50095916e-01 -9.66160119e-01 3.09505105e-01 -8.35400105e-01 5.84285796e-01 5.44876039e-01 -7.74458826e-01 -1.42365205e+00 8.27472433e-02 9.58173349e-02 -3.19870174e-01 -2.51886696e-01 1.64157316e-01 -4.05957282e-01 4.70081680e-02 3.81659001e-01 3.74201268e-01 -1.69774115e-01 -2.72113681e-01 2.64635295e-01 3.94977838e-01 3.61662954e-01 -1.16203058e+00 6.51048243e-01 6.13830760e-02 1.12464011e-01 -4.79318976e-01 -8.71744990e-01 1.50866717e-01 -4.10075098e-01 -4.14596379e-01 7.69730568e-01 -6.21609986e-01 -1.28795588e+00 1.74957708e-01 -6.85600281e-01 -9.32931244e-01 -4.36135709e-01 4.49752629e-01 -1.19643390e+00 5.14288545e-01 -3.14173341e-01 -1.49475849e+00 -4.62126918e-02 -1.06642079e+00 5.46184361e-01 3.22498769e-01 -2.08902344e-01 -8.49355161e-01 3.45968276e-01 -6.46896735e-02 5.28410897e-02 6.56921744e-01 9.34709430e-01 -4.15614218e-01 -5.89547753e-01 3.15240920e-01 4.51898843e-01 1.74484715e-01 -2.59010434e-01 -3.48006636e-01 -5.52098632e-01 -6.63957298e-01 3.40389550e-01 -7.74240673e-01 5.94387233e-01 1.20641753e-01 1.10781741e+00 -7.95199692e-01 -2.28953660e-01 4.56240058e-01 1.46397865e+00 8.20058703e-01 4.00726646e-01 7.82288730e-01 8.06930512e-02 6.09318972e-01 1.53210461e+00 9.14485812e-01 2.37433195e-01 3.33208382e-01 9.07509327e-01 4.91668731e-01 8.24223518e-01 -7.09491551e-01 9.13860559e-01 -2.79176012e-02 9.67548415e-03 -1.45575449e-01 -8.17567050e-01 3.98362130e-01 -2.18894649e+00 -8.97761226e-01 6.58043146e-01 2.38794565e+00 1.04100060e+00 5.37094891e-01 3.58054936e-01 -5.86865917e-02 3.45488489e-01 1.47909373e-01 -1.14069736e+00 -8.54296207e-01 3.01577836e-01 4.20647711e-02 3.27218175e-01 8.21501255e-01 -1.01176953e+00 1.08850193e+00 6.77518940e+00 8.61916661e-01 -7.69446492e-01 -5.11116460e-02 4.57494974e-01 -3.84379089e-01 -1.22792378e-01 2.29642004e-01 -1.01492918e+00 4.30345982e-01 7.74483860e-01 -3.13474476e-01 6.63270354e-01 1.15553224e+00 2.93358594e-01 -7.39722550e-01 -1.25859606e+00 6.12990379e-01 -5.89122772e-01 -5.98101556e-01 -2.23704338e-01 1.31622866e-01 6.33471072e-01 -5.66645086e-01 8.60697180e-02 8.07906151e-01 9.92615402e-01 -9.46493626e-01 8.98609936e-01 3.27901304e-01 6.34368002e-01 -1.39533293e+00 4.37022448e-01 9.32887912e-01 -7.86950827e-01 -6.66037261e-01 -3.46135587e-01 -2.01135606e-01 5.16601205e-02 -1.68563910e-02 -5.52782238e-01 2.75331467e-01 6.04002953e-01 4.69525546e-01 1.17523365e-01 5.60559988e-01 -7.96199620e-01 5.29966831e-01 -1.68860286e-01 -2.54221827e-01 6.98471785e-01 -3.08090359e-01 7.83531249e-01 8.61740887e-01 1.07813023e-01 2.49221817e-01 4.87893045e-01 8.35978270e-01 5.70368409e-01 -8.01543370e-02 -7.09589183e-01 1.42623726e-02 4.64696050e-01 8.49707067e-01 -6.15135431e-01 -1.58930257e-01 9.02745426e-02 8.31172824e-01 5.70559502e-01 5.20852983e-01 -8.75864387e-01 -1.82276294e-01 6.90191448e-01 -3.63524467e-01 1.89907268e-01 -2.32953593e-01 7.91504234e-02 -9.88195658e-01 -2.64398575e-01 -1.13800085e+00 7.15004265e-01 -2.56875187e-01 -7.90600717e-01 3.97245616e-01 4.29322869e-01 -1.17241645e+00 -5.18185735e-01 -1.61795437e-01 -6.54283464e-01 4.47773427e-01 -1.51839209e+00 -5.28646946e-01 3.60705197e-01 7.02406406e-01 6.34550631e-01 -7.03734979e-02 7.27735639e-01 -7.06282735e-01 -7.40149975e-01 2.57579476e-01 2.05574796e-01 -5.11063635e-01 2.87139177e-01 -1.50731075e+00 -6.16386719e-02 8.56866241e-01 -5.44001281e-01 2.82955617e-01 8.51115227e-01 -9.90314841e-01 -1.35180092e+00 -9.22711611e-01 3.21828097e-01 -1.43636167e-01 4.94180769e-01 -3.48107040e-01 -8.23429048e-01 6.73232555e-01 1.80644035e-01 -3.23357761e-01 3.08862031e-01 -1.79808125e-01 1.08288214e-01 9.88927111e-02 -1.38001645e+00 1.00869584e+00 1.01195192e+00 -2.10894436e-01 -9.97990608e-01 1.32561922e-01 7.88109779e-01 -4.89401102e-01 -5.70738614e-01 1.07090510e-01 3.65158588e-01 -8.78074348e-01 7.97835886e-01 -7.79084980e-01 1.89336807e-01 -1.13662146e-01 2.94704325e-02 -1.37522900e+00 -1.62307769e-01 -1.37043357e+00 -7.29712784e-01 7.62079000e-01 -1.17477573e-01 -5.12353361e-01 6.42067790e-01 6.97744250e-01 3.61575812e-01 -1.07499695e+00 -1.08221114e+00 -1.22677636e+00 3.92817080e-01 -3.61520916e-01 4.11354274e-01 5.17381191e-01 4.09916401e-01 -2.38098443e-01 -7.19470143e-01 1.02442607e-01 1.13573861e+00 2.22255349e-01 3.90491068e-01 -6.90786004e-01 -4.17970955e-01 -1.98254377e-01 5.48927188e-01 -1.02683949e+00 4.76539820e-01 -3.86087656e-01 6.14088058e-01 -1.17918634e+00 -1.30518451e-02 -4.37240869e-01 -5.80338418e-01 4.65599030e-01 -2.11947560e-01 -7.73120880e-01 4.56455946e-01 -6.05102926e-02 -1.06886840e+00 8.75050783e-01 1.42838621e+00 2.83061326e-01 -6.02367818e-01 2.38890707e-01 -5.51381350e-01 9.04429018e-01 1.04013371e+00 -6.55343235e-01 -6.86708391e-01 2.20763087e-01 1.35631129e-01 8.19722950e-01 -8.39358568e-03 -9.01897252e-01 1.65054128e-02 -1.14757383e+00 1.97223667e-02 -5.34731448e-01 5.26671903e-03 -6.96564972e-01 -1.81865945e-01 1.13507915e+00 -7.87956536e-01 -1.23813137e-01 5.17991446e-02 9.22379196e-01 2.21297115e-01 -4.84484464e-01 1.02271783e+00 -2.93503433e-01 -8.25979888e-01 2.11164713e-01 -1.13872993e+00 3.57487947e-01 1.63667834e+00 -6.31746650e-03 2.33150840e-01 -5.37156165e-01 -9.80151713e-01 7.25660384e-01 1.78093106e-01 2.10939586e-01 8.68524730e-01 -1.08759403e+00 -2.64205486e-01 1.52424797e-02 -2.96912551e-01 2.03160718e-01 -6.07631132e-02 3.60463947e-01 -1.06792346e-01 2.50878334e-01 -4.34243947e-01 -1.13127641e-01 -8.69314671e-01 8.97092283e-01 4.79600400e-01 -8.01466465e-01 -7.62338579e-01 5.07078290e-01 2.29893282e-01 -8.58290046e-02 8.11122775e-01 -2.00483084e-01 -3.04156095e-01 -5.51424734e-02 5.40926456e-01 5.56988955e-01 -6.69666350e-01 8.62801969e-02 -1.05136596e-01 2.07300857e-01 -9.54735726e-02 -6.35762632e-01 1.26729763e+00 -2.81341285e-01 3.07710975e-01 4.12115693e-01 4.52394485e-01 -5.01046181e-01 -2.01949024e+00 -2.74901569e-01 4.19373840e-01 -5.48511028e-01 7.26138726e-02 -9.36548591e-01 -5.53568840e-01 7.43130982e-01 4.47552651e-01 -1.40235916e-01 9.70257401e-01 -4.38068330e-01 7.25931406e-01 7.17051268e-01 8.83361638e-01 -1.52639329e+00 4.53076750e-01 5.84757566e-01 9.10251737e-01 -1.07517874e+00 -4.56334017e-02 2.85087526e-01 -1.27269948e+00 9.66234624e-01 9.48005795e-01 -1.10878095e-01 3.38126212e-01 4.32673246e-01 -1.49167880e-01 2.74496317e-01 -1.05863619e+00 -1.21940590e-01 -5.33675909e-01 8.54354084e-01 -5.29107690e-01 9.46679115e-02 -3.67312819e-01 5.01516581e-01 1.87673181e-01 2.23678127e-01 6.09121799e-01 1.44466496e+00 -7.84944952e-01 -1.19541037e+00 -3.71853530e-01 -4.87718172e-03 -5.14744103e-01 3.37693155e-01 -9.09924433e-02 7.49055803e-01 -8.19385797e-02 7.80105233e-01 -2.82982498e-01 -2.00102568e-01 3.00318934e-02 3.02107129e-02 4.90890503e-01 -3.61636549e-01 -3.79111618e-01 4.15139273e-02 -6.49288967e-02 -1.12196457e+00 2.03964356e-02 -8.32351923e-01 -1.61776328e+00 -7.27427825e-02 2.18569815e-01 2.66888738e-01 2.31454521e-01 9.76550043e-01 -1.96801826e-01 1.08802520e-01 8.70933294e-01 -5.31667948e-01 -1.35714138e+00 -3.06603938e-01 -7.91831493e-01 5.35129346e-02 7.72318900e-01 -8.27857137e-01 -3.46234679e-01 -4.83251274e-01]
[4.498925685882568, 2.142439842224121]
214e26e7-d2aa-49e8-99da-dd53535d0b02
investigation-of-applying-quantum-neural
2210.03882
null
https://arxiv.org/abs/2210.03882v2
https://arxiv.org/pdf/2210.03882v2.pdf
Investigation of Applying Quantum Neural Network of Early-Stage Breast Cancer Detection
Due to the heavy burden on medical institutes and computer-aided image diagnostics (CAD) have been gaining importance in diagnostic medicine to aid the medical staff to attain better service for the patients. Breast cancer is a fatal disease that can be treated successfully if it is detected early. Quantum neural network (QNN) has been introduced by many researchers around the world and presented recently by research corporations such as Microsoft, Google, and IBM. In this paper, we are trying to answer the question of: whether can the QNN be an effective method for mass-scale early breast cancer detection. This paper is dedicated to drawing a baseline for examining QNN, and the results showed a promising opportunity to use it for mass-scale screening using a fully functional quantum computer.
['Muazez Al Ali', 'Amjad Y. Sahib', 'Musaddiq Al Ali']
2022-10-08
null
null
null
null
['breast-cancer-detection', 'breast-cancer-detection']
['knowledge-base', 'medical']
[ 4.92651999e-01 1.98629186e-01 -4.02634352e-01 -6.76843822e-02 -6.16779327e-01 1.80795521e-01 1.71068102e-01 4.67782915e-01 -5.88086188e-01 8.29733670e-01 -2.52334446e-01 -6.89445972e-01 1.22776376e-02 -1.19434607e+00 -2.11492896e-01 -7.22775578e-01 -2.04195991e-01 4.99736905e-01 2.87957966e-01 -3.97798777e-01 3.10114473e-01 7.25050211e-01 -1.11741388e+00 8.83232355e-02 7.99906135e-01 8.85549128e-01 1.36870435e-02 7.02281773e-01 1.72567129e-01 5.80610633e-01 -1.11013301e-01 -5.42264879e-01 -6.81752618e-03 -8.29631448e-01 -7.23271608e-01 -5.55505872e-01 -2.23482996e-01 -2.76214153e-01 -3.52622032e-01 1.44663608e+00 6.50054872e-01 -2.06463978e-01 6.05341017e-01 -5.88314176e-01 -2.35162452e-01 1.81934401e-01 -3.53753179e-01 1.61184758e-01 1.94427118e-01 -1.72623638e-02 7.54658401e-01 -3.15496862e-01 4.12165940e-01 1.09168053e+00 6.50395989e-01 9.30503905e-01 -9.97503757e-01 -6.22564554e-01 -1.20464146e+00 4.93661731e-01 -1.06516027e+00 -2.38237143e-01 4.23998922e-01 5.65350838e-02 9.18956697e-01 2.14578763e-01 8.57448876e-01 7.21653700e-01 6.92205012e-01 2.92031169e-01 9.91110384e-01 -7.40379214e-01 4.55337286e-01 3.75752151e-02 1.71347082e-01 9.67233181e-01 5.30207634e-01 6.17361546e-01 -2.23759666e-01 -2.57978231e-01 8.91293585e-01 4.29678410e-02 1.19445652e-01 -1.96022391e-02 -9.24689174e-01 1.32016027e+00 7.34976828e-01 5.20164430e-01 -3.19913864e-01 4.92519289e-01 4.93419319e-01 -1.73924416e-02 1.37754053e-01 6.08927071e-01 -1.32819340e-02 -3.61137480e-01 -6.73243523e-01 4.54633683e-02 5.98660946e-01 1.72515348e-01 3.01978528e-01 -2.71799088e-01 5.72019331e-02 2.84523278e-01 3.20163488e-01 6.09995604e-01 3.83049518e-01 -1.19163620e+00 -2.76930869e-01 6.80492699e-01 -4.93484885e-02 -7.11622417e-01 -6.33558810e-01 -2.57849187e-01 -1.27277446e+00 1.25538662e-01 2.70497143e-01 -1.01136327e-01 -7.73684859e-01 1.23370516e+00 3.37784350e-01 -1.60519123e-01 2.54623652e-01 6.55403018e-01 8.54108214e-01 6.20169520e-01 -7.26391934e-03 -2.01542154e-01 1.42083788e+00 -4.45330262e-01 -6.85037076e-01 2.33067811e-01 1.04052889e+00 -4.99917120e-01 2.51061261e-01 5.68906903e-01 -8.37498009e-01 -1.91197217e-01 -9.98188674e-01 1.94747131e-02 -3.47527742e-01 -1.38399214e-01 1.06646192e+00 1.16417634e+00 -9.05798197e-01 1.03902435e+00 -1.33465981e+00 -7.77828813e-01 5.10745943e-01 7.45310843e-01 -2.71954238e-01 -1.74042016e-01 -1.35387433e+00 9.60113406e-01 4.09103930e-01 2.32696231e-03 -6.67178810e-01 -1.67703986e-01 -4.06396806e-01 -1.78490520e-01 2.20318362e-01 -6.98264003e-01 1.26069593e+00 -5.81559360e-01 -1.51400423e+00 9.99419391e-01 -1.44635201e-01 -7.86009967e-01 1.98855698e-01 2.14224443e-01 -3.88985038e-01 3.50101858e-01 1.31582022e-01 8.09526145e-01 1.37564972e-01 -2.58269131e-01 -6.58446491e-01 -6.64914310e-01 -1.40938699e-01 -8.52666199e-02 -3.15385491e-01 2.82103300e-01 -7.61530772e-02 -4.77296393e-03 3.31390321e-01 -1.08655059e+00 -7.33341694e-01 -1.21892765e-01 -3.59028816e-01 -3.19265425e-01 4.30345833e-01 -6.37229621e-01 9.50472951e-01 -1.98521590e+00 -1.06474884e-01 1.45212233e-01 1.29759997e-01 4.09917563e-01 3.37922484e-01 4.40865219e-01 -9.67674926e-02 2.48139605e-01 -1.26562282e-01 5.07544398e-01 -5.62910497e-01 1.80264324e-01 4.38287675e-01 5.21339118e-01 3.15428197e-01 1.02466738e+00 -9.13136184e-01 -5.60538888e-01 -7.69048929e-02 4.15359497e-01 -5.06084681e-01 -1.98426127e-01 1.03060856e-01 5.18735528e-01 -6.88165843e-01 8.12962770e-01 3.19421619e-01 -6.74810708e-01 7.10218772e-02 -1.18099339e-01 4.29027081e-02 -4.97420169e-02 -7.80954063e-01 1.11662817e+00 2.48193130e-01 6.15579724e-01 -2.44244322e-01 -1.35114121e+00 5.85267425e-01 3.50954205e-01 5.46087384e-01 -1.04397345e+00 2.88291872e-01 6.25780463e-01 6.10371232e-01 -5.42362094e-01 3.27914715e-01 -6.76880479e-01 6.74321502e-02 -1.44303650e-01 -3.12233657e-01 -1.85879260e-01 1.58177823e-01 2.03756347e-01 1.34089172e+00 -5.07685244e-01 5.35990179e-01 -2.73303866e-01 4.43057775e-01 4.47858274e-01 2.32520163e-01 7.36671269e-01 -5.31295836e-01 1.54650241e-01 5.20495951e-01 -4.59059864e-01 -1.06767368e+00 -8.18217874e-01 -5.57637095e-01 3.89773697e-01 1.87917128e-01 -1.86738476e-01 -6.86790645e-01 -2.50491947e-01 -2.18352824e-01 4.58179355e-01 -4.65147436e-01 -4.81563687e-01 -4.68294233e-01 -1.25278592e+00 5.90002239e-01 3.44252855e-01 7.20093966e-01 -8.66196096e-01 -6.89807594e-01 3.55031282e-01 9.08623263e-02 -8.33229840e-01 4.05284107e-01 5.69339216e-01 -1.36049163e+00 -1.05537939e+00 -1.01876187e+00 -6.58363581e-01 4.57321972e-01 -1.06069393e-01 6.31186008e-01 1.93909988e-01 -7.85355747e-01 1.22726902e-01 -2.68244386e-01 -5.90716422e-01 -1.03724062e+00 1.55932624e-02 -8.01269989e-03 -5.00725865e-01 4.25378352e-01 -6.04983233e-02 -7.06888020e-01 -1.69559699e-02 -7.10258663e-01 -3.72238420e-02 8.39671195e-01 9.10711527e-01 5.39957166e-01 1.47536203e-01 4.30623710e-01 -8.46254408e-01 4.24570709e-01 -1.39623135e-01 -6.24473631e-01 2.17359252e-02 -6.00141287e-01 4.70337719e-02 2.53059000e-01 3.33517641e-02 -5.30050457e-01 -8.77607241e-02 -4.76865292e-01 3.64032030e-01 -5.50868288e-02 6.80381596e-01 3.14585328e-01 -5.22541642e-01 8.49393547e-01 -1.12321720e-01 2.61596262e-01 -9.80152413e-02 -2.15051487e-01 7.26643085e-01 2.83362508e-01 -3.91382026e-04 2.88483411e-01 5.39439678e-01 1.04665422e+00 -1.24775743e+00 -6.17886782e-01 -3.43380004e-01 -1.71029910e-01 -7.08547905e-02 1.29082203e+00 -6.09662950e-01 -9.08579528e-01 3.77175152e-01 -1.09622037e+00 -4.64095213e-02 -1.60412639e-01 6.06314778e-01 -2.77425319e-01 4.08035636e-01 -8.30334306e-01 -8.45987380e-01 -5.05312800e-01 -1.23614240e+00 8.04873228e-01 7.10134685e-01 1.02260813e-01 -9.16571319e-01 -3.83740887e-02 2.63223022e-01 6.01997137e-01 1.99010625e-01 1.00284612e+00 -4.99918908e-01 -5.95594287e-01 -7.27389216e-01 -4.97183710e-01 2.86061466e-01 -1.39792180e-02 4.12814654e-02 -6.62920356e-01 -2.09526658e-01 2.05279957e-03 -3.41216594e-01 7.99589396e-01 9.05046225e-01 9.90934491e-01 3.53860795e-01 -7.83405364e-01 3.62330556e-01 1.37062705e+00 6.48576498e-01 6.87630892e-01 1.88057289e-01 1.86444804e-01 1.82709366e-01 4.10803586e-01 7.93412253e-02 -1.56734005e-01 4.00519222e-01 5.39144754e-01 -1.72829732e-01 -3.26121859e-02 4.23023812e-02 1.39908325e-02 7.41596818e-01 -4.60603505e-01 4.79086265e-02 -1.09774983e+00 -3.77979092e-02 -1.47709131e+00 -7.99235106e-01 -5.81198514e-01 2.23316264e+00 7.31572270e-01 4.29473341e-01 -1.27304614e-01 2.05435921e-02 6.27447486e-01 -6.69212282e-01 -5.50850689e-01 -4.11257237e-01 1.53285161e-01 6.14655137e-01 7.09786534e-01 2.08573654e-01 -8.27054262e-01 6.29725933e-01 6.94829321e+00 1.07905674e+00 -1.39736724e+00 6.94516376e-02 1.20283222e+00 3.94952118e-01 3.17049742e-01 2.16232523e-01 -8.03153217e-01 5.24960943e-02 1.35952604e+00 -2.23530993e-01 1.74862370e-01 6.91162229e-01 2.04034075e-01 -7.43175209e-01 -9.80282962e-01 9.01504099e-01 -3.77011597e-01 -1.46594322e+00 -1.74340174e-01 3.83781791e-01 6.23469412e-01 1.03128605e-01 -4.77469452e-02 2.45167747e-01 -1.93283141e-01 -1.19681144e+00 -1.96411401e-01 6.83110058e-01 8.60478997e-01 -7.74804413e-01 1.10636449e+00 5.56731820e-01 -5.76915741e-01 -3.27473432e-02 -4.39857423e-01 -2.83149064e-01 4.84758206e-02 6.83454216e-01 -9.34251189e-01 3.66844088e-01 5.98002195e-01 2.09119290e-01 -5.92393339e-01 1.29052830e+00 2.73558080e-01 7.54872143e-01 -4.12417948e-01 -8.92287493e-01 3.15364301e-01 -3.31842810e-01 3.90729159e-01 6.80797756e-01 5.36361217e-01 1.96851268e-01 -1.72326311e-01 7.69846022e-01 1.99023798e-01 1.29697416e-02 -4.96349961e-01 -4.85065728e-01 -1.95224270e-01 1.09479988e+00 -1.23959255e+00 -3.49989235e-01 -2.03349575e-01 9.70738828e-01 -1.46247521e-01 -1.45846277e-01 -7.02644408e-01 -4.77123827e-01 4.21514586e-02 3.06426167e-01 -1.32528901e-01 6.00550547e-02 -1.43267065e-01 -8.33311081e-01 -6.32734835e-01 -6.19981408e-01 5.15101701e-02 -5.49013913e-01 -8.26708317e-01 2.04553843e-01 -2.18074843e-01 -6.77980721e-01 -5.25955632e-02 -1.24757349e+00 -3.83699059e-01 6.95378244e-01 -1.20497477e+00 -5.39257646e-01 1.62765592e-01 1.55553803e-01 -1.90607309e-01 -8.14525485e-02 1.20468402e+00 2.80799806e-01 -5.09214520e-01 2.43868902e-01 5.10628700e-01 2.53650755e-01 3.34839255e-01 -1.13954496e+00 -9.78355035e-02 4.68910158e-01 -1.10600501e-01 4.90925670e-01 8.10717404e-01 -6.64190412e-01 -1.50779355e+00 -5.93698263e-01 1.04705417e+00 -2.34677702e-01 4.64664876e-01 1.79973364e-01 -3.79942983e-01 1.28639368e-02 6.29226714e-02 -5.30894957e-02 5.52368939e-01 -2.14863241e-01 6.50488496e-01 -1.85118228e-01 -1.24000907e+00 4.90604669e-01 5.64227700e-01 -3.58518004e-01 -1.07914522e-01 8.19440961e-01 3.53350461e-01 -3.15949172e-01 -6.45138621e-01 6.80107713e-01 4.27836359e-01 -1.30281115e+00 7.29179084e-01 -4.58250254e-01 4.66150224e-01 3.39158803e-01 4.51811440e-02 -7.69240558e-01 -1.86744273e-01 -5.43856382e-01 2.91381150e-01 3.74757171e-01 3.03589314e-01 -6.49831593e-01 1.25613666e+00 6.30548537e-01 1.57978550e-01 -1.05223417e+00 -1.33338439e+00 -6.35531485e-01 2.57654667e-01 -4.69792873e-01 4.54330035e-02 3.09803277e-01 1.04693845e-01 2.91660935e-01 -5.51100224e-02 -1.88075319e-01 5.36634028e-01 -3.30810994e-01 2.64008462e-01 -1.38392746e+00 -2.89061964e-01 -5.25042593e-01 -9.34848845e-01 -5.54881036e-01 -6.51399136e-01 -9.16663766e-01 -3.31232190e-01 -1.35475421e+00 6.23059630e-01 -7.59779587e-02 -2.98298419e-01 5.92336468e-02 2.44048666e-02 2.94693917e-01 -3.32301080e-01 -8.90001878e-02 -4.57911879e-01 1.96278483e-01 1.48954070e+00 -8.33391547e-02 1.39400102e-02 3.02822024e-01 -6.05014205e-01 6.97155654e-01 7.62252271e-01 -7.23269224e-01 2.96925694e-01 4.20336366e-01 7.00626850e-01 6.13248944e-01 4.85409498e-01 -1.35974979e+00 2.20665395e-01 1.39861420e-01 4.05835956e-01 -4.22116041e-01 3.68189335e-01 -7.27096140e-01 8.43962654e-02 1.36113071e+00 -5.79582863e-02 -2.66827703e-01 -6.69273436e-02 4.37253952e-01 -2.48368979e-01 -7.56013274e-01 1.03304601e+00 -3.25225443e-01 -3.79431725e-01 1.61067560e-01 -4.04541522e-01 -2.69637078e-01 1.12954926e+00 -1.91296577e-01 -5.10030463e-02 -3.81800622e-01 -5.96164346e-01 -6.88667744e-02 2.97466338e-01 -3.97611409e-01 4.49506164e-01 -1.10627878e+00 -4.74466175e-01 1.60842299e-01 -8.24356303e-02 -3.59397501e-01 1.35843351e-01 1.17075920e+00 -1.13324976e+00 8.67383182e-01 9.42294523e-02 -6.26191199e-01 -1.19627583e+00 3.83841813e-01 6.57732069e-01 -5.78192651e-01 -4.20273513e-01 1.01425731e+00 -2.88149416e-01 -1.76488385e-01 7.09627941e-02 -3.94962579e-01 -2.13954523e-02 -4.02447313e-01 3.23506296e-01 3.99496675e-01 3.30261327e-02 -4.41671073e-01 -2.36409873e-01 4.40824419e-01 1.43507523e-02 -1.14702664e-01 1.22443736e+00 3.62019420e-01 -4.29947644e-01 4.02196199e-01 1.19924641e+00 -4.44544762e-01 -2.65957028e-01 3.41793090e-01 4.45938222e-02 2.69844592e-01 3.58781248e-01 -8.08645785e-01 -6.87627256e-01 1.25490654e+00 1.21082079e+00 6.38503134e-01 9.51565981e-01 1.32590592e-01 8.39672506e-01 8.98753464e-01 5.32757163e-01 -1.04594541e+00 1.42053172e-01 3.28832299e-01 1.43825039e-01 -1.53608119e+00 2.60747299e-02 -2.54451811e-01 -1.87792391e-01 1.49736667e+00 7.11744130e-02 -1.66596875e-01 7.99189925e-01 -1.71320051e-01 -3.25016022e-01 -3.84767622e-01 -2.47509226e-01 -1.42507121e-01 5.29481992e-02 1.50463343e-01 8.67078781e-01 2.77319610e-01 -6.76914096e-01 2.78549343e-01 2.33291730e-01 4.96874154e-01 4.30847079e-01 1.08072400e+00 -8.97209823e-01 -1.36173260e+00 -4.64626044e-01 1.00324965e+00 -7.06221581e-01 -6.61111474e-02 -3.02337885e-01 7.00391293e-01 2.18533441e-01 7.31467068e-01 -2.68647134e-01 1.69832222e-02 -2.85529066e-02 -2.27579903e-02 6.99675739e-01 -4.71769243e-01 -1.59677505e-01 -1.50477374e-02 4.82285731e-02 -4.35055703e-01 -4.19235438e-01 -4.70137745e-01 -1.60534585e+00 -1.34181857e-01 -8.37266505e-01 3.09285939e-01 8.46616566e-01 9.12053406e-01 6.88911751e-02 5.70544958e-01 2.55545437e-01 -5.30378819e-01 -8.10999513e-01 -8.18682194e-01 -9.13364172e-01 -8.21752325e-02 8.66826400e-02 -4.04694229e-01 -1.29228204e-01 -4.07245308e-01]
[15.262072563171387, -2.6707282066345215]
3e88144e-4354-482e-a588-3bf1fd749058
easy-things-first-installments-improve
null
null
https://aclanthology.org/P16-1058
https://aclanthology.org/P16-1058.pdf
Easy Things First: Installments Improve Referring Expression Generation for Objects in Photographs
null
['Sina Zarrie{\\ss}', 'David Schlangen']
2016-08-01
null
null
null
acl-2016-8
['referring-expression-generation']
['computer-vision']
[-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]
[-7.443027019500732, 3.5593693256378174]
e44bc3c3-b3bc-4ee7-b002-9e8f61b7d857
trajectory-user-linking-is-easier-than-you
2212.07081
null
https://arxiv.org/abs/2212.07081v1
https://arxiv.org/pdf/2212.07081v1.pdf
Trajectory-User Linking Is Easier Than You Think
Trajectory-User Linking (TUL) is a relatively new mobility classification task in which anonymous trajectories are linked to the users who generated them. With applications ranging from personalized recommendations to criminal activity detection, TUL has received increasing attention over the past five years. While research has focused mainly on learning deep representations that capture complex spatio-temporal mobility patterns unique to individual users, we demonstrate that visit patterns are highly unique among users and thus simple heuristics applied directly to the raw data are sufficient to solve TUL. More specifically, we demonstrate that a single check-in per trajectory is enough to correctly predict the identity of the user up to 85% of the time. Moreover, by using a non-parametric classifier, we scale up TUL to over 100k users which is an increase over state-of-the-art by three orders of magnitude. Extensive empirical analysis on four real-world datasets (Brightkite, Foursquare, Gowalla and Weeplaces) compares our findings to state-of-the-art results, and more importantly validates our claim that TUL is easier than commonly believed.
['Kyle Mede', 'Alameen Najjar']
2022-12-14
null
null
null
null
['activity-detection']
['computer-vision']
[-2.95848638e-01 -9.68808010e-02 -6.71105981e-01 -2.92337716e-01 -5.55854499e-01 -7.97335386e-01 8.93733859e-01 5.63189447e-01 -4.15786594e-01 9.32356477e-01 4.21841323e-01 -5.77691972e-01 -2.91555107e-01 -9.51871514e-01 -6.50945783e-01 -1.92211196e-01 -6.19090676e-01 6.21125221e-01 1.95147917e-01 -2.07148269e-01 -3.65348645e-02 4.59288150e-01 -1.39224386e+00 -1.10614952e-03 8.13835859e-01 6.95954740e-01 -5.58939993e-01 4.27379817e-01 -1.01152929e-02 5.90585470e-01 -1.04623653e-01 -6.40385449e-01 2.02028409e-01 -2.70383544e-02 -9.09760475e-01 -3.75478923e-01 4.21272039e-01 -3.92645270e-01 -7.98273683e-01 5.85184157e-01 2.94481456e-01 3.89648855e-01 6.73567891e-01 -1.65450716e+00 -1.01357615e+00 7.26587594e-01 -5.62290072e-01 4.74723786e-01 4.97059643e-01 -7.43069053e-02 1.20201468e+00 -2.29964569e-01 8.06385815e-01 9.25096393e-01 1.36006439e+00 2.27381036e-01 -1.16441071e+00 -7.58517683e-01 2.03681707e-01 1.03794791e-01 -1.45587146e+00 -3.66804391e-01 3.50580186e-01 -5.65582991e-01 8.97856832e-01 4.01913106e-01 4.87867177e-01 1.34216511e+00 -1.32456005e-01 8.21396410e-01 5.74455619e-01 1.92982346e-01 -7.10917041e-02 4.66738362e-03 4.42085177e-01 6.79895699e-01 5.37159681e-01 -1.94690600e-01 -1.04951754e-01 -6.01068139e-01 4.21635747e-01 3.23651046e-01 3.27173844e-02 -1.81454062e-01 -1.01547849e+00 9.96379137e-01 5.07646084e-01 3.95703822e-01 -4.08024848e-01 1.25352457e-01 4.94337559e-01 1.35682598e-01 5.29810131e-01 3.40792209e-01 -3.36210936e-01 -4.86689210e-01 -1.15425193e+00 6.75991476e-01 8.02876830e-01 7.70268857e-01 7.87441730e-01 -1.61444783e-01 -1.79052711e-01 4.98714298e-01 3.27171572e-02 4.89746869e-01 5.42127848e-01 -7.60112822e-01 3.99040073e-01 8.99324059e-01 3.54872972e-01 -1.34695935e+00 -6.67858303e-01 -2.95548946e-01 -6.77863121e-01 -5.14838874e-01 7.07385480e-01 -3.18073004e-01 -5.89112997e-01 1.85619891e+00 2.05600560e-01 6.58591926e-01 -3.82982284e-01 5.27550936e-01 7.11680532e-01 5.30729532e-01 1.15867980e-01 1.59895256e-01 1.11936390e+00 -6.61670864e-01 -3.17393899e-01 -1.85680091e-01 8.81682634e-01 2.10887846e-02 9.86843705e-01 -1.94815725e-01 -8.61418664e-01 -1.63455620e-01 -7.18218446e-01 -1.20472483e-01 -8.09531629e-01 -1.61518902e-01 1.06500149e+00 9.18960094e-01 -9.08532560e-01 7.24265218e-01 -7.62051761e-01 -8.45597625e-01 7.58121014e-01 4.67148185e-01 -5.49008608e-01 1.15865275e-01 -1.32428145e+00 5.41547358e-01 -1.27195008e-03 -3.78806829e-01 -3.85105640e-01 -8.99929523e-01 -7.00456858e-01 9.58966091e-02 1.32615477e-01 -4.21322644e-01 9.78054523e-01 -5.14770806e-01 -8.20320964e-01 9.44548309e-01 -5.38215995e-01 -7.42496610e-01 7.46712744e-01 -1.57656167e-02 -8.25634003e-01 -3.11052144e-01 5.77062964e-01 2.95982778e-01 3.40518117e-01 -9.25047994e-01 -8.42706621e-01 -2.87165016e-01 2.76662201e-01 -3.96007955e-01 -5.64362407e-01 3.85619961e-02 -5.82197249e-01 -4.47317958e-01 -1.64493948e-01 -1.22017944e+00 -6.62939847e-02 -4.09099787e-01 -3.51021498e-01 -6.32471204e-01 7.99806893e-01 -7.87458241e-01 1.68559706e+00 -1.87981439e+00 -1.86119184e-01 4.36511934e-01 4.32951570e-01 2.58157641e-01 2.63122898e-02 8.10214460e-01 2.30182678e-01 3.31565917e-01 -2.57291019e-01 -4.31276083e-01 2.80092359e-01 8.19882154e-02 -3.98707867e-01 7.88389981e-01 -1.95492283e-01 1.34309018e+00 -1.16203105e+00 -9.56273600e-02 -6.42270269e-03 2.90443003e-01 -4.37534899e-01 -1.73876852e-01 5.83222024e-02 2.67927766e-01 -3.23970526e-01 6.47967994e-01 4.64173913e-01 -6.25379801e-01 3.79632086e-01 3.68199080e-01 6.68989643e-02 3.89586389e-01 -7.40403354e-01 1.25495517e+00 -1.63154900e-01 9.62586701e-01 -3.63375217e-01 -7.85841465e-01 4.80672121e-01 9.96437594e-02 8.56104612e-01 -8.42617869e-01 -1.28927277e-02 3.42161022e-02 -2.49682546e-01 -4.29448992e-01 8.19451332e-01 4.94286150e-01 -3.77520263e-01 7.95114160e-01 -3.06860358e-01 1.11871862e+00 2.90929615e-01 3.02745402e-01 1.25414348e+00 -9.03566554e-02 2.32260242e-01 -2.19822377e-01 3.09314251e-01 -9.33036283e-02 5.13222516e-01 9.08477068e-01 -4.25845206e-01 2.28711933e-01 4.92298722e-01 -7.07177043e-01 -9.95109260e-01 -1.15114129e+00 -1.23894475e-02 1.40810680e+00 1.21828064e-01 -5.73498428e-01 -7.49403119e-01 -8.92581105e-01 5.46401978e-01 5.34199238e-01 -9.20317650e-01 1.47500142e-01 -6.39313996e-01 -9.08691585e-01 9.76891041e-01 5.26766956e-01 4.68315095e-01 -8.14768732e-01 -1.61807731e-01 3.01420420e-01 -4.46012259e-01 -1.16729939e+00 -4.61780101e-01 -6.47857666e-01 -5.12316823e-01 -1.18942082e+00 -6.69041991e-01 -3.76346946e-01 4.76747900e-01 4.36588109e-01 1.09511054e+00 2.60224819e-01 -3.40177827e-02 2.82308638e-01 -1.95788085e-01 4.99872118e-02 7.48246983e-02 7.55594492e-01 4.51543897e-01 1.96038425e-01 7.82592773e-01 -8.40138733e-01 -5.57484806e-01 2.08307162e-01 -3.82375032e-01 -4.27948147e-01 1.63381860e-01 1.66137069e-01 8.50662217e-02 -1.70816611e-02 8.07130039e-01 -1.20156062e+00 7.00717986e-01 -1.37886071e+00 -1.57813087e-01 -1.33270267e-02 -6.37754798e-01 -2.72002459e-01 4.65329111e-01 -5.79428911e-01 -5.57767570e-01 -1.53077379e-01 -7.85608441e-02 -1.62267551e-01 -2.62155116e-01 3.84485662e-01 2.73992598e-01 1.21635851e-02 6.13756061e-01 1.06511816e-01 -3.80818635e-01 -3.66102815e-01 4.66064185e-01 6.53182983e-01 4.87261534e-01 -4.58282471e-01 8.39313030e-01 8.73729348e-01 -5.02102338e-02 -1.02616513e+00 -5.43349445e-01 -5.57685554e-01 -6.16617680e-01 -3.31910662e-02 7.86099732e-01 -7.97238588e-01 -1.37281513e+00 2.75354564e-01 -7.69205689e-01 -3.51170629e-01 1.75381199e-01 1.18196540e-01 -2.43966356e-01 4.12137598e-01 -6.79964006e-01 -9.19117868e-01 -2.73213923e-01 -7.24379182e-01 8.00148129e-01 1.73949361e-01 -6.56867683e-01 -1.30779302e+00 1.12107120e-01 2.95774668e-01 4.14072484e-01 5.96274018e-01 7.99685240e-01 -8.90327394e-01 -5.31565785e-01 -6.17513835e-01 -4.87685800e-01 -5.76436639e-01 2.57870197e-01 -3.15787792e-01 -9.00584161e-01 -3.32741618e-01 -8.92737627e-01 3.40065323e-02 8.28975141e-01 2.27223516e-01 1.13247776e+00 -5.52375913e-01 -1.06676185e+00 4.52541322e-01 1.04914165e+00 -1.72623649e-01 5.23256898e-01 4.69314426e-01 1.01847863e+00 4.21189427e-01 1.86846107e-01 2.73804605e-01 1.13458204e+00 8.32876980e-01 1.58522487e-01 1.66408017e-01 2.84274220e-01 -6.05241597e-01 3.79886702e-02 1.91288009e-01 -3.04690123e-01 -4.87559289e-01 -1.10740948e+00 9.43692684e-01 -2.28451252e+00 -1.42904329e+00 -3.98337483e-01 2.18762684e+00 1.80192068e-01 4.30844016e-02 9.23377275e-01 1.19885644e-02 6.01785123e-01 3.20483744e-01 -6.41026735e-01 -1.20161995e-01 1.28484562e-01 -2.86080331e-01 9.11839843e-01 4.43483800e-01 -1.27914953e+00 9.69353199e-01 6.84348822e+00 5.84084511e-01 -1.05192399e+00 1.97640434e-01 4.30184662e-01 -7.50220418e-02 -4.35785592e-01 -3.57337534e-01 -7.47534871e-01 9.29625213e-01 1.27622414e+00 -3.01704705e-01 8.21450412e-01 6.07164919e-01 2.54629403e-01 2.88168192e-01 -1.13882983e+00 9.94734883e-01 -1.46243230e-01 -1.55083346e+00 -1.05539098e-01 6.20571673e-01 9.34117198e-01 3.62812608e-01 1.76541254e-01 5.28047264e-01 4.71832365e-01 -1.14563882e+00 4.68139350e-01 8.29703033e-01 6.49318159e-01 -9.83089089e-01 5.27226150e-01 4.63277638e-01 -1.38647437e+00 -4.33867544e-01 -1.15470819e-01 -7.43714795e-02 3.50247204e-01 3.45177919e-01 -8.34754348e-01 5.61238945e-01 8.18425119e-01 9.18822527e-01 -7.08749413e-01 9.30216312e-01 2.53560781e-01 9.70013916e-01 -4.18605655e-01 1.06326245e-01 6.83122158e-01 3.23165325e-03 3.11427236e-01 1.35999858e+00 4.66280907e-01 5.80964908e-02 7.63909742e-02 6.95508599e-01 -6.33448541e-01 -1.08989835e-01 -9.31183994e-01 -3.23589534e-01 6.32273316e-01 1.08497918e+00 -5.30185521e-01 -2.79662013e-01 -5.14174879e-01 9.66388345e-01 7.57505476e-01 3.55785102e-01 -9.75085795e-01 -2.77186751e-01 1.15010607e+00 5.76423287e-01 3.31134319e-01 -3.62973362e-01 -1.07750632e-01 -9.41668987e-01 -7.49654397e-02 -3.67774814e-01 3.68306011e-01 -1.78447023e-01 -1.62302387e+00 2.35419706e-01 -1.88728303e-01 -1.04179966e+00 -5.14701903e-01 -2.53692955e-01 -8.44782054e-01 7.01356232e-01 -1.29738557e+00 -1.27425647e+00 -9.82931107e-02 4.76260513e-01 -2.63797622e-02 -1.57074332e-01 8.88722658e-01 8.94916117e-01 -6.43542647e-01 9.83925402e-01 5.18691599e-01 5.63226402e-01 2.58709311e-01 -1.03874981e+00 1.22680628e+00 6.48618817e-01 1.89048246e-01 9.12013173e-01 4.72066075e-01 -8.62885416e-01 -1.23417270e+00 -1.23297560e+00 1.39490521e+00 -1.00039816e+00 8.93298209e-01 -3.95474464e-01 -9.93501544e-01 1.14158607e+00 -1.45464703e-01 1.02817468e-01 9.24894869e-01 4.90209103e-01 -3.47205490e-01 1.05509341e-01 -1.20347571e+00 6.97790682e-01 1.44645774e+00 -7.65409291e-01 -1.48759082e-01 3.08894515e-01 5.38583577e-01 -2.53182024e-01 -8.94923627e-01 1.84653103e-02 9.17938828e-01 -8.98516774e-01 1.18148136e+00 -1.08976853e+00 1.10099815e-01 -5.94399078e-03 -5.48416302e-02 -1.02436495e+00 -6.13644838e-01 -6.92031860e-01 -6.03986144e-01 1.36570394e+00 4.58878696e-01 -6.73455119e-01 1.05695105e+00 8.82748365e-01 2.33760849e-01 -7.77287006e-01 -8.20513070e-01 -7.46343374e-01 8.81889760e-02 -5.40793955e-01 1.10797560e+00 1.53536153e+00 4.29934487e-02 -5.08253323e-03 -7.28955746e-01 2.97601700e-01 6.54778540e-01 3.77967139e-04 8.38482916e-01 -1.76813996e+00 4.72529139e-03 -5.31578124e-01 -4.78915900e-01 -9.71808434e-01 4.74771410e-01 -1.15180838e+00 -6.51492178e-01 -1.72388518e+00 3.03255599e-02 -7.28545427e-01 -2.59038776e-01 5.56452096e-01 -5.36829382e-02 5.09292424e-01 3.17011103e-02 2.86309004e-01 -7.47115552e-01 8.32789764e-02 5.31678200e-01 -1.78758100e-01 -3.44427466e-01 3.69194120e-01 -9.84996676e-01 5.13401091e-01 7.67137170e-01 -2.86176533e-01 -3.09579700e-01 -3.84308666e-01 4.58162755e-01 -2.50275016e-01 4.54192221e-01 -1.01704371e+00 1.85879633e-01 5.98557629e-02 3.35077524e-01 -4.88785267e-01 4.70533878e-01 -7.17720985e-01 8.34276378e-02 4.10514735e-02 -2.85339773e-01 1.34976059e-01 1.04488060e-01 1.00893390e+00 4.14683044e-01 1.75564200e-01 2.30502635e-01 -3.60862762e-02 -8.30712616e-01 7.93792188e-01 -4.30385202e-01 1.56419829e-01 1.04201257e+00 -2.85577714e-01 -4.35662627e-01 -7.03251243e-01 -5.24451673e-01 3.97785872e-01 4.40337360e-01 6.83282614e-01 1.08515099e-01 -1.58243787e+00 -4.73672956e-01 5.32889999e-02 2.06466809e-01 -7.35968828e-01 2.17656940e-01 6.54462576e-01 -4.10934463e-02 6.42975628e-01 -1.03415713e-01 -2.96754390e-01 -1.04420435e+00 4.81163472e-01 1.97484970e-01 -6.37046695e-02 -8.05966735e-01 5.04407346e-01 -4.97603923e-01 -6.62081659e-01 1.59473732e-01 -4.62097004e-02 -2.98889428e-01 2.90119648e-01 6.36166751e-01 1.01361573e+00 -1.53257966e-01 -1.10793030e+00 -5.45518756e-01 1.87636465e-01 -1.34444103e-01 7.36439824e-02 1.27041793e+00 -2.81840384e-01 1.65844828e-01 4.55793470e-01 1.32431972e+00 1.46953449e-01 -1.02176189e+00 -2.90492833e-01 2.91828007e-01 -5.18469095e-01 -1.96491033e-01 -4.59124088e-01 -8.26761961e-01 6.65412009e-01 4.12957042e-01 8.71201515e-01 3.24654311e-01 -5.87738343e-02 1.40569127e+00 5.83983004e-01 5.36094785e-01 -8.09438646e-01 -5.00435770e-01 4.75606441e-01 1.57381967e-01 -1.40139270e+00 -2.06458405e-01 3.29416022e-02 -5.98382592e-01 4.27451164e-01 3.25336039e-01 -2.14289516e-01 8.11191499e-01 -4.07234609e-01 -4.07854259e-01 -7.01619387e-02 -2.44152620e-01 -2.84605473e-01 3.30352217e-01 6.85281634e-01 4.79263812e-01 3.58996421e-01 -4.75857891e-02 6.89378500e-01 -3.18422168e-01 8.21838230e-02 3.99325341e-01 5.68565905e-01 -3.46317142e-01 -9.99072671e-01 3.04358564e-02 8.90770257e-01 -5.80613136e-01 1.28719509e-01 -2.90563107e-01 8.74305189e-01 8.71948972e-02 1.00995195e+00 2.59628683e-01 -4.75624740e-01 1.88156754e-01 1.11270018e-01 1.45188794e-01 -2.44895697e-01 -6.49882793e-01 -8.80374312e-01 2.72379190e-01 -7.37403274e-01 -1.29345551e-01 -8.72381032e-01 -1.31641495e+00 -1.28808558e+00 3.25735927e-01 5.11088632e-02 2.85532087e-01 1.03815484e+00 7.17736483e-01 1.06770366e-01 4.16412205e-01 -8.96160245e-01 3.82567793e-02 -6.31739140e-01 -5.13343573e-01 6.22178972e-01 5.00284791e-01 -8.12169313e-01 -1.32409662e-01 -4.24088210e-01]
[6.560232162475586, 2.1096320152282715]
34e5d0e1-04ae-4ceb-9937-da1b7cd670c0
the-singularity-controversy-part-i-lessons
1601.05977
null
http://arxiv.org/abs/1601.05977v2
http://arxiv.org/pdf/1601.05977v2.pdf
The Singularity Controversy, Part I: Lessons Learned and Open Questions: Conclusions from the Battle on the Legitimacy of the Debate
This report seeks to inform policy makers on the nature and the merit of the arguments for and against the concerns associated with a potential technological singularity. Part I describes the lessons learned from our investigation of the subject, separating the argu-ments of merit from the fallacies and misconceptions that confuse the debate and undermine its rational resolution.
['Amnon H. Eden']
2016-01-22
null
null
null
null
['misconceptions']
['miscellaneous']
[ 5.27597725e-01 3.70110244e-01 -3.14934105e-01 -1.36505082e-01 -1.57758340e-01 -9.13895726e-01 7.85148978e-01 4.26708758e-01 -2.22520083e-01 5.83829582e-01 6.15967274e-01 -1.52428532e+00 -4.12024051e-01 -2.88848519e-01 -7.06818342e-01 -5.69599926e-01 1.98005661e-01 -4.27306771e-01 7.97321945e-02 -1.36733428e-01 9.96227026e-01 4.71584857e-01 -1.26145339e+00 -1.30483285e-01 6.81041062e-01 8.80989790e-01 -3.77208352e-01 2.96401769e-01 1.08634280e-02 8.54649901e-01 -9.43729818e-01 -5.12152255e-01 4.94611673e-02 -2.88241327e-01 -7.13490188e-01 -7.68956840e-02 3.31796557e-02 -3.57242644e-01 -1.07474111e-01 1.11998785e+00 2.20343515e-01 -4.36129451e-01 6.20728254e-01 -8.89614940e-01 -2.62480736e-01 2.83576131e-01 -3.17847878e-01 5.79376221e-01 1.09540038e-01 1.92272559e-01 5.36212862e-01 -8.15028399e-02 3.80005628e-01 1.18660402e+00 5.16456842e-01 2.18058422e-01 -5.91258168e-01 -5.94469488e-01 2.02783555e-01 -4.23675269e-01 -1.26710391e+00 -9.51076746e-01 3.59721541e-01 -7.14317024e-01 5.66474199e-01 3.84701908e-01 8.26565802e-01 1.04812646e+00 9.52283978e-01 7.42675876e-03 1.00837791e+00 -4.54014152e-01 2.27247551e-01 3.84028345e-01 1.44302174e-01 1.52574673e-01 1.39286017e+00 3.24390680e-01 -1.19519539e-01 -6.50048494e-01 9.28709686e-01 -7.86625326e-01 -2.82680869e-01 -9.44812000e-02 -5.82010746e-01 5.35640776e-01 -2.90012658e-01 7.60703027e-01 -3.11908364e-01 -1.19987831e-01 5.66707730e-01 1.47272423e-01 2.96170521e-03 5.20379782e-01 -3.14709276e-01 -4.32224184e-01 -7.73736611e-02 -1.10026769e-01 6.67993963e-01 2.14368656e-01 -7.64386207e-02 7.96989128e-02 7.10239708e-01 -1.96765527e-01 5.02276659e-01 8.13230798e-02 -1.18106470e-01 -1.10995436e+00 1.00762501e-01 1.18585080e-02 6.25804186e-01 -1.09520566e+00 -9.56546888e-02 -3.75947028e-01 -2.38274448e-02 2.45332778e-01 8.38060454e-02 -7.36689150e-01 -1.82240084e-01 1.26282907e+00 1.75444949e-02 -4.07079160e-01 1.93429738e-01 5.85901439e-01 5.88037968e-01 4.66793269e-01 4.23747271e-01 -4.97915596e-01 1.01424873e+00 3.67614813e-02 -8.06599677e-01 -1.02812484e-01 7.46829450e-01 -9.10778880e-01 5.00896633e-01 3.45890999e-01 -1.04030526e+00 1.03921637e-01 -1.29260838e+00 4.03772920e-01 4.60166996e-03 -5.12670875e-01 5.35776556e-01 1.24400318e+00 -5.93051195e-01 3.79968703e-01 -6.61464334e-01 -6.02749586e-01 -8.22545439e-02 1.67550012e-01 1.02546230e-01 4.62081105e-01 -5.20019591e-01 1.04772079e+00 5.42601883e-01 3.04275632e-01 1.04545280e-01 -2.76519507e-01 -2.67239451e-01 -1.27299756e-01 6.64279640e-01 -2.58641213e-01 8.16143990e-01 -6.88453257e-01 -1.06326318e+00 4.53483582e-01 3.34800571e-01 -2.91290343e-01 2.11590976e-01 4.77316529e-02 -9.79142427e-01 3.46488133e-02 9.47633684e-02 -1.91912994e-01 2.71224052e-01 -1.21756697e+00 -5.88563383e-01 -1.21261865e-01 1.59902461e-02 -2.96848297e-01 8.21672752e-02 4.97461677e-01 2.60135025e-01 -4.95244086e-01 1.95544779e-01 -8.84006619e-01 9.84762013e-02 -5.59108853e-01 1.21674233e-03 -1.46801367e-01 6.85571551e-01 -3.05767864e-01 1.06860101e+00 -2.57608652e+00 -7.60154009e-01 6.16982765e-03 -2.11257134e-02 7.45072588e-02 3.33859086e-01 9.26605105e-01 -3.38275507e-02 6.34850323e-01 4.37531501e-01 9.50321436e-01 -1.56761035e-01 -1.59713924e-02 -5.57130694e-01 8.62556338e-01 3.97959575e-02 4.81832922e-01 -8.36257279e-01 -1.00113116e-01 1.94428205e-01 1.75572455e-01 -5.59483953e-02 -5.53938150e-02 3.19512486e-01 2.39989758e-01 -9.11091268e-01 7.43220031e-01 4.37583447e-01 -1.24045461e-01 6.73410773e-01 -4.36492294e-01 -8.42170417e-01 1.18144417e+00 -5.89974165e-01 7.09060431e-01 2.62233377e-01 5.30317068e-01 4.46852893e-01 -6.93441391e-01 5.93576670e-01 4.06514406e-01 3.31096798e-01 -6.36197388e-01 6.03434741e-01 5.79084635e-01 3.97661120e-01 -9.75867808e-01 3.92845094e-01 -6.12761319e-01 -1.25239819e-01 5.64433038e-01 -4.82309967e-01 -2.58623570e-01 -5.77801526e-01 -6.99432790e-02 7.81895399e-01 3.13671380e-01 2.42484152e-01 -8.21443379e-01 -1.01281814e-01 -1.09384552e-01 6.33974075e-01 3.25973779e-01 -3.12878251e-01 1.02777081e-02 6.05941653e-01 -4.05699253e-01 -6.05000198e-01 -7.19565272e-01 -2.67439991e-01 3.16817552e-01 5.70061684e-01 -5.04520357e-01 -3.29915971e-01 -6.76395059e-01 6.59828484e-02 1.10698271e+00 -4.51023221e-01 -2.33928472e-01 -1.14171401e-01 -7.34487116e-01 2.16816843e-01 3.55780065e-01 8.70937929e-02 -1.16945706e-01 -1.62530303e+00 -1.59612298e-01 -7.52318278e-02 -1.12274873e+00 3.30028802e-01 1.52014986e-01 -8.64875734e-01 -1.45525646e+00 -1.06358767e-01 -1.47627592e-01 7.53377020e-01 4.74568725e-01 4.24537659e-01 1.18814811e-01 1.20958179e-01 7.05832243e-01 -1.51339769e-01 -7.53045917e-01 -6.44646525e-01 -2.89240092e-01 1.01064242e-01 -6.36653662e-01 4.36163753e-01 -5.77343106e-01 -2.66691625e-01 1.75827205e-01 -5.00621080e-01 -4.67895657e-01 6.11656010e-01 1.12487763e-01 -3.58656347e-01 3.18966061e-01 8.86370897e-01 -7.07488358e-01 8.11781764e-01 -6.36414170e-01 -4.41845506e-01 1.82798043e-01 -7.78037906e-01 -1.99313849e-01 -4.38654959e-01 -2.79068470e-01 -1.16324091e+00 -9.92256403e-01 -5.03669202e-04 3.21054071e-01 -1.40204057e-01 1.96298048e-01 -2.33669445e-01 -5.65778434e-01 4.94578153e-01 -2.92184472e-01 1.21949829e-01 -4.37065333e-01 -2.98784338e-02 6.74822986e-01 4.60350096e-01 -8.03735793e-01 5.22019684e-01 4.19345379e-01 -1.33707315e-01 -1.11636543e+00 -5.55737019e-01 -1.18730068e-01 2.19327688e-01 -3.81934166e-01 4.46113050e-01 -8.15690160e-01 -6.19267464e-01 -1.12724274e-01 -9.89455521e-01 9.69581082e-02 -1.52398109e-01 5.60671151e-01 -4.41988200e-01 5.57931960e-01 -3.79696190e-01 -1.17477405e+00 6.74410164e-02 -8.24034035e-01 1.73363239e-01 2.35827133e-01 -7.64799893e-01 -7.27338076e-01 -3.70729566e-01 6.75417304e-01 6.96559310e-01 7.28338659e-01 1.18008816e+00 -3.74242425e-01 -3.39957654e-01 -3.57843876e-01 -1.66488197e-02 -3.03442013e-02 5.60455918e-01 3.64963949e-01 -8.62943888e-01 -1.42333642e-01 6.82970703e-01 -2.92341888e-01 -7.80063570e-02 5.14765792e-02 3.62961739e-01 -5.85230589e-01 -3.17599714e-01 -1.31508291e-01 1.52554989e+00 6.84131205e-01 5.16554594e-01 8.78440619e-01 -1.15447350e-01 1.05728936e+00 6.86275542e-01 3.94222647e-01 4.16831784e-02 1.81287467e-01 3.56417209e-01 2.85376102e-01 3.37620407e-01 -1.76520720e-01 1.67649165e-01 4.77174431e-01 6.19535297e-02 -2.41419390e-01 -7.50211835e-01 5.40986300e-01 -1.50795412e+00 -2.26809189e-01 -3.19115698e-01 2.21493125e+00 5.03274016e-02 8.28457117e-01 4.96978074e-01 1.31477445e-01 6.65652454e-01 4.14616644e-01 -1.20953768e-01 -6.84784770e-01 5.32809831e-02 -6.56486809e-01 4.79326040e-01 2.60006964e-01 -5.75782359e-01 4.81012344e-01 8.66925049e+00 1.94419622e-01 -1.12534249e+00 -2.78906047e-01 6.85922086e-01 2.12114379e-01 -6.05434716e-01 3.02195400e-01 -2.23336741e-02 2.14734823e-01 9.91931081e-01 -6.11552238e-01 -3.06132168e-01 2.98788190e-01 2.55532205e-01 -5.07632375e-01 -5.81553996e-01 -3.02508902e-02 -1.05675243e-01 -1.17975569e+00 -6.80688620e-02 2.75576651e-01 8.00108016e-02 -1.22663394e-01 3.66592348e-01 -3.34776253e-01 9.72641706e-02 -5.93843758e-01 1.29772592e+00 2.20430419e-02 3.84945333e-01 -5.37298560e-01 8.87702644e-01 2.46612623e-01 -3.58825535e-01 -1.96562603e-01 -1.65521815e-01 -7.60062337e-01 1.99158221e-01 1.79435760e-01 -6.92615986e-01 2.49129981e-01 3.90477985e-01 -3.65416378e-01 -4.28126305e-02 8.03120971e-01 2.34681834e-02 9.24025416e-01 -2.75075048e-01 4.80528921e-02 5.41191161e-01 -4.41672802e-01 5.73095262e-01 7.19799161e-01 1.27268896e-01 4.32695746e-01 -8.11469078e-01 5.62850833e-01 8.61601412e-01 -4.43385653e-02 -7.81613410e-01 -7.76978731e-01 5.41014433e-01 6.64082766e-01 -9.33291137e-01 1.20785581e-02 -5.63305855e-01 1.98006541e-01 -5.28926134e-01 3.44991714e-01 -3.57800901e-01 -1.29535660e-01 5.27874112e-01 7.63919950e-01 2.11186767e-01 -3.43261838e-01 -6.66849911e-01 -5.90535879e-01 9.52925310e-02 -8.32949460e-01 2.19318494e-01 -2.50400722e-01 -5.18372774e-01 1.40248746e-01 2.37322494e-01 -7.20057726e-01 -9.35400873e-02 -4.63720471e-01 -9.49726522e-01 6.52678728e-01 -9.91789281e-01 -6.77763224e-01 6.49799466e-01 -5.82496762e-01 -1.76941007e-01 1.83456048e-01 3.69976580e-01 -3.45673889e-01 -1.90266684e-01 1.01915263e-01 6.80292249e-02 -4.90959704e-01 3.78226876e-01 -3.89948308e-01 5.81596613e-01 6.10950887e-01 -7.28694677e-01 9.37547147e-01 1.25145841e+00 -8.73416185e-01 -1.50379002e+00 -1.25768885e-01 9.35113251e-01 -4.60497230e-01 1.07681954e+00 2.59619713e-01 -6.09242260e-01 5.78301847e-01 3.73170525e-01 -9.61821258e-01 7.77000010e-01 3.79958123e-01 4.01386023e-02 2.86261499e-01 -1.28046107e+00 7.69972026e-01 5.40339410e-01 -4.22309756e-01 -7.62555957e-01 3.36572602e-02 6.20501220e-01 2.73157120e-01 -7.26339757e-01 4.34577256e-01 8.70504558e-01 -8.84328723e-01 6.47218466e-01 -4.60885674e-01 -2.22098947e-01 1.40121400e-01 -2.40158617e-01 -2.22916812e-01 -2.78597653e-01 -9.21897352e-01 6.16117597e-01 1.10516691e+00 4.25277293e-01 -8.24405670e-01 6.29927397e-01 1.20967269e+00 1.63529571e-02 -7.61259079e-01 -8.18286777e-01 -7.20972717e-01 3.07174474e-01 -3.39389920e-01 3.60245496e-01 1.08682609e+00 6.19904101e-01 2.96998173e-01 6.98406547e-02 3.64578873e-01 5.22915661e-01 -3.79123203e-02 4.55829263e-01 -1.04874754e+00 7.01004937e-02 -4.25029486e-01 -4.02286530e-01 -4.55513835e-01 -5.01872599e-01 8.13498124e-02 -4.04108584e-01 -1.15017533e+00 -4.11469489e-02 -3.73003185e-01 -6.69630989e-02 -3.14713687e-01 2.70113826e-01 -6.01499498e-01 4.40534472e-01 1.72644034e-01 -2.29221553e-01 1.30563483e-01 7.31933355e-01 1.92138493e-01 -7.72742927e-02 -1.22624040e-02 -1.62551522e+00 1.02117300e+00 9.21094477e-01 -1.23642549e-01 -4.39862400e-01 -1.67724609e-01 6.00786865e-01 2.70596653e-01 1.09807231e-01 -8.70124698e-01 -6.27852008e-02 -7.61929750e-01 -9.55733210e-02 -2.42467880e-01 6.86155707e-02 -1.10523915e+00 5.75808525e-01 7.08066821e-01 -2.98794031e-01 2.65016079e-01 5.79349101e-01 3.22331011e-01 4.27496016e-01 -4.33022231e-01 4.92876053e-01 1.02244198e-01 -1.65741276e-02 -5.09962499e-01 -7.43629336e-01 -3.15280914e-01 8.82714808e-01 -2.75625020e-01 -6.55090809e-01 -3.97057116e-01 -4.11479026e-01 -6.09097816e-03 1.14667952e+00 3.78267765e-01 1.57885149e-01 -8.26825857e-01 -2.32412368e-01 -1.06650986e-01 -2.26532966e-01 -6.24345064e-01 -1.03586659e-01 8.54372025e-01 -5.22566497e-01 6.20365560e-01 -1.73797205e-01 1.91141665e-01 -1.10239363e+00 7.27075756e-01 4.08078015e-01 6.26246691e-01 -9.37153697e-01 2.03620508e-01 3.39914769e-01 4.84166563e-01 7.61374906e-02 -2.03695431e-01 2.78322101e-02 -3.90509129e-01 3.04948181e-01 7.70211279e-01 -1.87034696e-01 -4.64420050e-01 -4.12827581e-01 1.45474955e-01 -1.95063323e-01 -4.83126312e-01 9.86410975e-01 -4.45234448e-01 -1.26087174e-01 7.07366526e-01 5.83431244e-01 2.86656916e-01 -8.58419180e-01 3.03631634e-01 1.63116485e-01 -6.98322833e-01 5.66636659e-02 -1.02348208e+00 -2.70747274e-01 5.93941391e-01 2.50418305e-01 6.42100334e-01 5.99827588e-01 2.03251820e-02 3.13909888e-01 3.48337650e-01 5.94507456e-01 -1.29458761e+00 -3.85823458e-01 -2.15512201e-01 8.58318686e-01 -4.24906999e-01 1.81376263e-01 -7.81462431e-01 -4.70008731e-01 9.26203191e-01 2.17900842e-01 3.41744781e-01 5.40497959e-01 4.84913945e-01 2.66389549e-01 -4.41474438e-01 -8.11234474e-01 3.81105363e-01 -5.91385365e-01 4.95368809e-01 5.93298852e-01 1.15566358e-01 -1.38598835e+00 6.02916420e-01 8.50430951e-02 1.99611783e-01 7.37272978e-01 1.40043712e+00 -7.80240774e-01 -7.86242843e-01 -9.56474423e-01 4.14544404e-01 -7.10577726e-01 4.97089535e-01 -8.72949481e-01 8.83278191e-01 4.35556144e-01 1.23599541e+00 -2.69449204e-01 -5.12418628e-01 8.96443892e-03 -1.00529574e-01 3.40011269e-01 -2.51756385e-02 -3.86338770e-01 5.09610891e-01 9.65624511e-01 -3.02328438e-01 -1.88398629e-01 -9.92925107e-01 -8.36211562e-01 -2.81114131e-01 -5.79815567e-01 6.09627426e-01 8.29950154e-01 1.11933279e+00 3.71566087e-01 2.07798213e-01 3.41856152e-01 4.42082249e-02 -7.46794879e-01 -4.59330499e-01 -5.16707778e-01 -2.78041720e-01 4.26156163e-01 -5.38574934e-01 -4.71103728e-01 -6.35092497e-01]
[8.910475730895996, 6.597665309906006]
34ce236b-8670-49b4-9b80-081b7f5c65e8
probability-calibration-for-knowledge-graph-1
1912.1
null
https://arxiv.org/abs/1912.10000v2
https://arxiv.org/pdf/1912.10000v2.pdf
Probability Calibration for Knowledge Graph Embedding Models
Knowledge graph embedding research has overlooked the problem of probability calibration. We show popular embedding models are indeed uncalibrated. That means probability estimates associated to predicted triples are unreliable. We present a novel method to calibrate a model when ground truth negatives are not available, which is the usual case in knowledge graphs. We propose to use Platt scaling and isotonic regression alongside our method. Experiments on three datasets with ground truth negatives show our contribution leads to well-calibrated models when compared to the gold standard of using negatives. We get significantly better results than the uncalibrated models from all calibration methods. We show isotonic regression offers the best the performance overall, not without trade-offs. We also show that calibrated models reach state-of-the-art accuracy without the need to define relation-specific decision thresholds.
['Luca Costabello', 'Pedro Tabacof']
2019-12-20
null
https://openreview.net/forum?id=S1g8K1BFwS
https://openreview.net/pdf?id=S1g8K1BFwS
iclr-2020-1
['calibration-for-link-prediction']
['graphs']
[-1.47197738e-01 7.03910589e-01 -5.95448613e-01 -2.02295005e-01 -6.43137813e-01 -5.73071718e-01 8.56083274e-01 2.22605929e-01 -3.44206661e-01 1.05987382e+00 5.50696962e-02 -3.57590944e-01 -3.84650826e-01 -1.12731338e+00 -1.02325761e+00 -4.01090175e-01 1.02253985e-02 1.15332007e+00 4.80547220e-01 -1.94210008e-01 -1.68775216e-01 1.06064521e-01 -1.30849349e+00 -3.04715969e-02 7.96251416e-01 3.32899988e-01 -6.00531697e-01 6.30849004e-01 -4.65589855e-03 6.63925350e-01 -2.87377924e-01 -1.17401528e+00 2.66401798e-01 -1.81439862e-01 -7.84505367e-01 -5.45519054e-01 7.22975075e-01 -1.35951176e-01 -6.10826969e-01 1.10523713e+00 2.11223051e-01 -3.80685687e-01 7.92865753e-01 -1.89272022e+00 -9.28640485e-01 1.31891143e+00 -6.00157678e-01 9.18570310e-02 4.34846729e-01 -3.39304984e-01 1.22877026e+00 -6.86991394e-01 9.02548552e-01 1.42003536e+00 1.11176085e+00 1.59410745e-01 -1.70229518e+00 -4.53799844e-01 -7.51482248e-02 7.19160736e-01 -1.63190639e+00 -1.70215964e-01 5.27020812e-01 -3.11079204e-01 8.80147457e-01 2.71703899e-01 6.62959158e-01 1.23163009e+00 -9.50816646e-02 1.50444508e-01 1.24016547e+00 -7.39986718e-01 1.09551132e-01 7.40952551e-01 4.34948832e-01 7.38480210e-01 1.16012084e+00 1.91820145e-01 -6.56523466e-01 -4.15194690e-01 5.33552706e-01 -5.25921106e-01 -3.03152919e-01 -9.08650339e-01 -1.45066309e+00 8.06848705e-01 5.22381246e-01 2.81603903e-01 -1.00361235e-01 4.99241054e-01 1.77181557e-01 3.11667919e-01 3.45141232e-01 3.99331957e-01 -6.30144596e-01 -1.24637879e-01 -9.69543636e-01 1.58996254e-01 1.27397501e+00 1.22926629e+00 9.45208848e-01 -2.86540091e-01 -1.68410227e-01 5.19119442e-01 2.14342430e-01 4.30188030e-01 -3.67786200e-03 -8.24289203e-01 2.82158792e-01 7.13506341e-01 3.63729239e-01 -1.16041684e+00 -2.15578943e-01 -1.67856187e-01 -4.93246347e-01 5.97717427e-02 7.22171903e-01 -6.98393732e-02 -9.87826407e-01 1.66728914e+00 3.92537892e-01 3.62417847e-01 1.59274146e-01 5.59330702e-01 6.38035655e-01 2.53021955e-01 2.87397057e-01 -1.36835054e-01 1.24729192e+00 -6.63671792e-01 -8.32559109e-01 -5.56442067e-02 8.57102692e-01 -4.37243044e-01 1.06829178e+00 1.96933657e-01 -5.72002232e-01 4.56961319e-02 -1.25593746e+00 -2.90214512e-02 -8.91541898e-01 -8.30510706e-02 9.39266741e-01 1.27429414e+00 -1.13592267e+00 7.68763661e-01 -6.95363283e-01 -5.83041489e-01 2.28378698e-01 4.27223742e-01 -7.21216083e-01 1.10924758e-01 -1.51472199e+00 1.57733428e+00 6.81456804e-01 -2.08370119e-01 -3.17270666e-01 -1.02661073e+00 -7.34418988e-01 1.35801835e-02 6.48456275e-01 -7.08315074e-01 7.46472538e-01 -5.66704452e-01 -1.02569652e+00 7.79261291e-01 5.54188825e-02 -5.36867797e-01 8.22523475e-01 -8.17417875e-02 -4.17269349e-01 -1.85416922e-01 -5.49609326e-02 3.54705870e-01 3.26674551e-01 -1.59625173e+00 -1.84299633e-01 -9.85935181e-02 2.11709589e-01 -1.49453938e-01 -4.91657108e-01 -3.31685215e-01 -5.25543869e-01 -6.52811453e-02 1.42020062e-01 -9.59006369e-01 1.79674059e-01 -1.45380780e-01 -6.86672688e-01 -1.58696547e-01 7.19371498e-01 -5.91377437e-01 1.28551447e+00 -1.46338272e+00 2.80949831e-01 5.59478998e-01 3.87727022e-01 9.69574675e-02 2.03368738e-01 4.85087335e-01 -1.82061285e-01 5.22282898e-01 1.27847474e-02 -1.11189477e-01 5.03740966e-01 5.16943872e-01 -2.64166743e-01 4.75122452e-01 2.34219819e-01 9.35019612e-01 -1.08169639e+00 -9.49511647e-01 3.14671546e-01 6.52572155e-01 -2.78036565e-01 -2.58428991e-01 -2.21929103e-01 -2.92426258e-01 -9.70086642e-03 6.14378750e-01 5.07149637e-01 -4.22524989e-01 7.64723301e-01 -5.57854533e-01 5.28400898e-01 7.34070465e-02 -1.40240121e+00 1.07884121e+00 -1.07883155e-01 5.18193543e-01 -5.22869766e-01 -6.40062809e-01 7.18834341e-01 1.32640332e-01 9.07893553e-02 -1.25430554e-01 3.75354812e-02 2.19859287e-01 7.35113248e-02 -8.41911808e-02 6.47545218e-01 -1.27999902e-01 3.44269499e-02 2.20356286e-01 3.79818141e-01 -1.33226633e-01 2.05125421e-01 6.44551754e-01 1.12018323e+00 4.74560678e-01 3.44168127e-01 -2.64251858e-01 1.86235815e-01 2.10319877e-01 5.30110121e-01 7.64602959e-01 -3.79961729e-02 2.84068435e-01 1.00231647e+00 -1.54384926e-01 -1.26797247e+00 -9.87480819e-01 -3.41980338e-01 7.41958320e-01 1.21892676e-01 -8.33792627e-01 -6.63451254e-01 -1.00417328e+00 4.19865757e-01 1.00839555e+00 -1.06818628e+00 -2.24881649e-01 -8.94302428e-02 -1.12179971e+00 7.42632806e-01 5.67714214e-01 1.23994827e-01 -3.77127349e-01 2.05101103e-01 -8.18655193e-02 -3.45542505e-02 -1.25900340e+00 1.25163823e-01 4.57458533e-02 -6.66748583e-01 -1.45085621e+00 -3.25993687e-01 -1.95998698e-01 6.70730293e-01 -2.16298327e-01 1.42184162e+00 1.66146487e-01 1.95794016e-01 6.50489151e-01 -2.37892896e-01 -3.17695588e-01 -4.79435652e-01 2.00523242e-01 2.68191278e-01 -3.60090584e-01 8.95712435e-01 -7.32398331e-01 -1.30893648e-01 2.59614348e-01 -5.25708854e-01 -1.93376884e-01 4.82549518e-01 7.76751280e-01 5.67527115e-01 -1.83711827e-01 4.82412755e-01 -1.51533031e+00 5.98874748e-01 -4.62641448e-01 -7.67661691e-01 7.95406997e-01 -1.36316168e+00 3.29236895e-01 2.29078650e-01 -4.10732687e-01 -5.91439486e-01 -4.18610200e-02 5.62853456e-01 -4.04634833e-01 3.50634992e-01 4.05945182e-01 -1.43331259e-01 -5.27752221e-01 1.15213048e+00 -4.35401827e-01 -2.23038986e-01 -2.36149743e-01 9.20634925e-01 4.20608371e-01 3.98018032e-01 -7.36303329e-01 1.05619323e+00 2.72933632e-01 1.64809972e-01 -3.57820958e-01 -7.15309322e-01 -6.57525212e-02 -8.08031201e-01 -1.23959079e-01 3.35904658e-01 -6.98096991e-01 -7.03488052e-01 -1.67181209e-01 -9.94159520e-01 -1.88574359e-01 -5.00058055e-01 3.90194714e-01 -4.33549076e-01 6.39319777e-01 -3.07695329e-01 -6.83227658e-01 -1.46210462e-01 -6.91034853e-01 6.07994556e-01 1.85898229e-01 -3.35305154e-01 -1.22517276e+00 3.23171765e-01 1.64080277e-01 2.82817632e-01 4.24586028e-01 9.61249471e-01 -8.00999403e-01 -5.60156226e-01 -3.87848616e-01 -4.66485977e-01 5.07044010e-02 -6.73017129e-02 4.95659590e-01 -1.03771663e+00 1.13560565e-01 -9.14061606e-01 -2.03754112e-01 7.99618661e-01 -1.73222959e-01 7.72461653e-01 -1.48866668e-01 -7.74273813e-01 2.77883470e-01 1.61727214e+00 -5.73095798e-01 8.40800524e-01 4.54229712e-01 8.37804317e-01 3.36027741e-01 4.92763788e-01 -4.43133116e-02 7.63649285e-01 6.85658813e-01 1.95255727e-01 2.22732231e-01 -3.53222996e-01 -6.88335955e-01 1.05223268e-01 6.62132382e-01 -4.05828387e-01 -2.35816553e-01 -1.21999753e+00 8.86531293e-01 -2.11834526e+00 -9.76464391e-01 -5.71555197e-01 2.36577535e+00 1.23117125e+00 2.33304322e-01 4.23079282e-02 3.31933975e-01 7.78033316e-01 -2.55453527e-01 -1.63159460e-01 -2.26969540e-01 -2.63345569e-01 1.37806103e-01 1.07023537e+00 9.48194325e-01 -9.31704938e-01 1.04315984e+00 7.42648697e+00 7.19515562e-01 -4.92747158e-01 4.27464128e-01 9.91653726e-02 1.53631449e-01 -7.48571932e-01 5.37801921e-01 -8.63200784e-01 2.60049164e-01 1.17016077e+00 -4.28248674e-01 3.47066700e-01 8.69624019e-01 -5.12157023e-01 -1.79397613e-01 -1.34858155e+00 8.35441411e-01 -3.21959779e-02 -1.42319417e+00 -8.72810110e-02 7.28543028e-02 9.09923196e-01 -4.33457755e-02 -3.98888826e-01 5.39219856e-01 9.57415104e-01 -9.95300114e-01 4.54057068e-01 8.10574114e-01 7.71058738e-01 -5.96117675e-01 9.70182419e-01 -2.05924392e-01 -7.16309428e-01 3.83190066e-01 -4.52878684e-01 2.32138932e-01 5.66996150e-02 8.93651128e-01 -1.23073721e+00 8.58344674e-01 4.61258471e-01 5.68963110e-01 -9.05519605e-01 9.72192883e-01 -5.45754433e-01 7.98152447e-01 -7.31521487e-01 -1.90699056e-01 -3.47745776e-01 -8.55395570e-02 4.10313189e-01 1.28533375e+00 9.08351783e-03 -3.15063536e-01 -3.08121175e-01 8.24084938e-01 -2.61603892e-01 -4.89905179e-02 -6.61192775e-01 -2.37261772e-01 7.06394196e-01 1.16918576e+00 -6.60673201e-01 -5.98382771e-01 -2.21438542e-01 6.02847159e-01 6.61096931e-01 3.37628037e-01 -9.72851694e-01 -3.08952063e-01 1.68811858e-01 1.88341498e-01 2.29160815e-01 -1.01505190e-01 -4.10147935e-01 -1.17907584e+00 -3.10519757e-03 -4.68157321e-01 5.67318916e-01 -8.43753994e-01 -1.65318120e+00 2.54863232e-01 5.06137908e-01 -5.77168047e-01 -1.83879733e-01 -7.50340283e-01 -2.16179803e-01 6.91837966e-01 -1.41251481e+00 -1.39955199e+00 -2.96763510e-01 2.39408791e-01 -6.99199378e-01 8.90837759e-02 1.07313049e+00 3.63268167e-01 -4.67517585e-01 8.37712169e-01 4.27053161e-02 1.02005109e-01 9.09739912e-01 -1.66490686e+00 1.43725514e-01 7.97979236e-01 3.24475884e-01 6.11225903e-01 1.13140917e+00 -9.09402966e-01 -1.42194331e+00 -8.44890594e-01 1.07610071e+00 -1.15234125e+00 1.12865496e+00 -2.28962421e-01 -9.23420668e-01 1.20903444e+00 2.31436342e-02 2.33925909e-01 7.04256892e-01 8.81554484e-01 -9.53426838e-01 -1.86692789e-01 -1.32175767e+00 4.98180032e-01 1.05616939e+00 -4.86013860e-01 -8.45779598e-01 6.63938999e-01 6.87105596e-01 -3.23521405e-01 -1.21729243e+00 4.05449539e-01 6.71267092e-01 -7.70010650e-01 9.90910828e-01 -7.86818385e-01 6.84256703e-02 -4.38024849e-01 -1.77897573e-01 -1.06328475e+00 -2.40297064e-01 -3.93528998e-01 -6.05806291e-01 1.37029147e+00 8.00962329e-01 -9.07277942e-01 9.92634177e-01 1.00426936e+00 4.48264629e-01 -3.60230297e-01 -9.18396175e-01 -9.59793985e-01 5.11835739e-02 -3.47874135e-01 7.36300409e-01 1.40953374e+00 3.14189434e-01 2.71019071e-01 -2.49135613e-01 4.39006001e-01 9.28134382e-01 -2.42116198e-01 8.52968335e-01 -1.38604951e+00 -1.46779552e-01 -4.86061685e-02 -1.00843406e+00 -7.78278410e-02 2.85402298e-01 -9.54932332e-01 -3.92117560e-01 -1.75815690e+00 4.48594809e-01 -6.95616484e-01 -2.77997613e-01 9.01901543e-01 -2.57166624e-01 5.49472034e-01 -1.44997193e-02 -9.96870548e-02 -5.70802569e-01 3.06809574e-01 7.58238554e-01 -1.65719241e-01 2.42920145e-02 -5.32878041e-01 -8.42567980e-01 5.86419165e-01 7.54029572e-01 -7.51333535e-01 -4.81360495e-01 -9.83080268e-02 8.15213501e-01 -4.64473993e-01 7.22878575e-01 -7.29949474e-01 2.78192163e-01 -5.60848638e-02 3.12277585e-01 -3.77339602e-01 4.82609242e-01 -9.19890702e-01 6.16792858e-01 1.75406665e-01 1.25874639e-01 -1.61355004e-01 2.74329484e-01 9.67300057e-01 1.82425022e-01 -2.43616238e-01 3.66286993e-01 1.82202965e-01 -6.61055624e-01 -1.77370459e-02 4.10463691e-01 9.93415192e-02 1.01297164e+00 -1.69607490e-01 -8.49785507e-01 -2.74701566e-01 -8.82536232e-01 1.73076272e-01 7.66078949e-01 2.08381340e-01 1.28841028e-01 -1.41800785e+00 -6.21504366e-01 -3.84251088e-01 2.15528369e-01 -5.34130275e-01 -2.42657706e-01 9.15298641e-01 -5.08125544e-01 2.38340557e-01 5.96623421e-02 -4.87918228e-01 -1.09085023e+00 6.65733457e-01 3.32961589e-01 -2.35501423e-01 -3.65022480e-01 5.77280819e-01 -7.40822315e-01 -6.13508046e-01 9.18845236e-02 -2.07741857e-01 -3.25984210e-02 1.22686565e-01 1.25464559e-01 5.61494827e-01 1.97950408e-01 -4.83804792e-01 -5.78465223e-01 4.61432725e-01 -5.43464087e-02 -5.53989746e-02 1.18189073e+00 2.21857369e-01 -3.19148213e-01 6.09260738e-01 6.85589314e-01 2.17980936e-01 -6.88286483e-01 -2.09545776e-01 1.13791034e-01 -5.70677876e-01 1.26066476e-01 -1.00420439e+00 -7.66383231e-01 5.01086354e-01 4.28835094e-01 4.13922071e-01 3.37837487e-01 5.13316840e-02 6.81540146e-02 5.29744744e-01 6.73399448e-01 -1.08982432e+00 -5.59156537e-01 1.02083787e-01 7.14483976e-01 -1.34071481e+00 6.81474328e-01 -7.46396005e-01 -5.43534160e-01 9.15411115e-01 5.56932986e-01 -1.15097463e-01 6.32157326e-01 9.11566317e-02 -1.78480223e-01 -2.26687342e-01 -6.83564782e-01 -2.65493870e-01 2.21709654e-01 9.60898459e-01 2.59524286e-01 3.78714710e-01 -6.14609540e-01 6.64559782e-01 -5.09008110e-01 -1.14196837e-02 7.42317855e-01 5.49693525e-01 -9.75795239e-02 -1.48105538e+00 -4.87296969e-01 6.71556354e-01 -1.09227911e-01 -7.35785365e-02 -6.90971613e-01 1.36843121e+00 -4.09393683e-02 7.56115377e-01 -4.80940044e-01 -6.72127843e-01 3.39006722e-01 4.72573876e-01 7.62463868e-01 -3.31580162e-01 -3.33807111e-01 -7.27136314e-01 6.52619064e-01 -3.83253366e-01 -5.38289130e-01 -4.89708722e-01 -9.84712899e-01 -8.71100187e-01 -8.36659551e-01 2.28250816e-01 5.49129009e-01 5.73773205e-01 4.29933012e-01 3.81597251e-01 -1.02334373e-01 -5.47090411e-01 -5.13599396e-01 -9.59625900e-01 -3.82647932e-01 2.21571669e-01 -5.14349714e-02 -1.12823665e+00 -5.84612072e-01 -6.06978079e-03]
[8.729150772094727, 7.678472995758057]
d4280bd1-4bcc-4ada-9c40-9ca5bec8cef6
contour-based-interactive-segmentation
2302.06353
null
https://arxiv.org/abs/2302.06353v1
https://arxiv.org/pdf/2302.06353v1.pdf
Contour-based Interactive Segmentation
Recent advances in interactive segmentation (IS) allow speeding up and simplifying image editing and labeling greatly. The majority of modern IS approaches accept user input in the form of clicks. However, using clicks may require too many user interactions, especially when selecting small objects, minor parts of an object, or a group of objects of the same type. In this paper, we consider such a natural form of user interaction as a loose contour, and introduce a contour-based IS method. We evaluate the proposed method on the standard segmentation benchmarks, our novel UserContours dataset, and its subset UserContours-G containing difficult segmentation cases. Through experiments, we demonstrate that a single contour provides the same accuracy as multiple clicks, thus reducing the required amount of user interactions.
['Anton Konushin', 'Anna Vorontsova', 'Polina Popenova', 'Danil Galeev']
2023-02-13
null
null
null
null
['interactive-segmentation']
['computer-vision']
[ 3.41949582e-01 -2.11373687e-01 -4.83458824e-02 -4.33910549e-01 -4.49957669e-01 -9.40850914e-01 3.11719626e-01 5.89860737e-01 -8.67508888e-01 4.28031296e-01 -5.00153184e-01 -5.76902330e-01 1.79626822e-01 -6.62473500e-01 -4.99341935e-01 -3.09946477e-01 3.14965606e-01 4.11582619e-01 9.82351899e-01 4.98873256e-02 4.23379391e-01 4.09941405e-01 -1.33539867e+00 3.75113748e-02 1.27838874e+00 1.04342127e+00 3.07211518e-01 4.29246932e-01 -4.95525926e-01 -1.43229574e-01 -7.04463124e-01 -4.83106136e-01 4.51087803e-01 -2.05310553e-01 -6.46715105e-01 4.43719089e-01 6.28388882e-01 -2.32035980e-01 3.70449960e-01 1.10880888e+00 3.23364973e-01 3.73461694e-01 4.92695302e-01 -1.02880669e+00 -1.94913208e-01 4.62590396e-01 -1.00291908e+00 -3.19827460e-02 2.09965795e-01 5.09577338e-03 7.55750060e-01 -8.58142436e-01 5.31884372e-01 9.67044115e-01 5.81256866e-01 3.27466369e-01 -1.45165336e+00 -4.68878329e-01 3.57974142e-01 -1.12709194e-01 -1.49239171e+00 -3.01444121e-02 5.45612812e-01 -6.08630538e-01 3.86863858e-01 5.70111394e-01 7.93076456e-01 1.60371155e-01 -1.38911113e-01 9.58768785e-01 9.69243348e-01 -3.87869865e-01 3.99658382e-01 1.40286878e-01 6.04444146e-01 6.71422482e-01 3.34712237e-01 -3.45315933e-01 -8.82920325e-02 -1.66573808e-01 7.97762752e-01 1.26094699e-01 -2.89372474e-01 -4.63115126e-01 -1.04242980e+00 7.44455516e-01 4.80218500e-01 1.01113610e-01 -1.41907468e-01 -7.10621476e-02 2.06455931e-01 -1.21569216e-01 4.06551957e-01 2.08369553e-01 -2.67568052e-01 -1.40472591e-01 -1.17874265e+00 3.88790220e-01 7.40806043e-01 1.12881196e+00 7.96122074e-01 -3.20497870e-01 -2.53418773e-01 9.27356899e-01 -8.29851702e-02 2.15305433e-01 2.72526871e-02 -7.62726009e-01 2.39191622e-01 7.76707053e-01 3.57774198e-01 -8.12486649e-01 -4.54133481e-01 -3.34704578e-01 -5.28298140e-01 3.97763163e-01 5.97083092e-01 -2.49385342e-01 -1.17417955e+00 1.36771178e+00 5.93430758e-01 -1.27323642e-01 -6.71755672e-01 8.44094515e-01 4.77998078e-01 3.31613541e-01 7.37517402e-02 -1.84269235e-01 1.38220561e+00 -1.08002257e+00 -6.43147469e-01 -1.28220230e-01 4.12150532e-01 -1.07229054e+00 1.54213488e+00 5.54875255e-01 -1.07734549e+00 -4.41626489e-01 -8.31923962e-01 -1.38860300e-01 -4.15533841e-01 1.96459636e-01 7.03983605e-01 7.21460640e-01 -6.86676145e-01 7.62366056e-01 -8.08794379e-01 -2.22943738e-01 5.02717257e-01 3.44088793e-01 5.39475083e-02 1.18455306e-01 -5.68957210e-01 2.76093334e-01 3.31079960e-01 -1.22019485e-01 -1.67361453e-01 -7.33015239e-01 -5.48026383e-01 7.95904845e-02 8.88471663e-01 -2.99648792e-01 1.41489887e+00 -8.11226904e-01 -1.36360168e+00 6.47589505e-01 -1.69508159e-01 -2.26222292e-01 1.02109683e+00 -6.57335222e-01 1.08519927e-01 2.04881281e-03 -4.93947277e-03 7.45231748e-01 7.35719323e-01 -1.27827311e+00 -7.39939868e-01 -3.88908833e-01 2.50597090e-01 5.21056913e-02 -2.05515921e-01 4.68259193e-02 -1.14657354e+00 -7.53883779e-01 3.35548848e-01 -1.16837966e+00 -3.88546705e-01 3.72015148e-01 -8.35851371e-01 -2.96586275e-01 9.61700618e-01 -4.29621696e-01 1.45860529e+00 -2.34814048e+00 -1.30326167e-01 4.59743321e-01 1.92350209e-01 3.19701135e-01 1.61995545e-01 1.00258119e-01 1.66896045e-01 4.26338881e-01 -5.50565720e-01 -4.21171159e-01 -2.87214331e-02 -1.31013365e-02 1.51106313e-01 1.09362222e-01 -1.95927709e-01 6.25542819e-01 -7.37154663e-01 -7.89079130e-01 2.57863075e-01 1.29106954e-01 -6.83690071e-01 9.56768319e-02 -3.79066676e-01 2.77940571e-01 -2.79655993e-01 4.94532585e-01 9.03635502e-01 -2.50825733e-01 5.92387244e-02 -1.77335903e-01 -3.03582311e-01 -4.71689925e-02 -1.49535215e+00 1.69079828e+00 -3.99740338e-01 3.79294485e-01 4.83560674e-02 -4.03059155e-01 5.91793835e-01 -1.20132759e-01 2.61732906e-01 -3.92259926e-01 7.90487453e-02 1.98019192e-01 2.93909814e-02 -1.44307464e-01 4.49701488e-01 3.35233063e-01 1.34315223e-01 6.41193748e-01 -3.46522540e-01 -3.41662258e-01 7.19123900e-01 3.67324680e-01 7.06431866e-01 1.58620432e-01 3.79305154e-01 -2.99701929e-01 1.97957814e-01 -9.28869843e-02 4.94145185e-01 8.77575159e-01 -1.38223872e-01 7.39552319e-01 5.18133759e-01 -1.30975083e-01 -7.99449563e-01 -9.36915576e-01 -2.78608650e-01 1.14065003e+00 5.55608749e-01 -5.73906004e-01 -1.08248723e+00 -9.49127018e-01 -1.10741317e-01 6.02778316e-01 -3.26259911e-01 4.20643806e-01 -4.64822948e-01 -5.28313756e-01 5.20352088e-02 5.78160882e-01 5.33306658e-01 -9.84490097e-01 -9.95416939e-01 2.07840979e-01 7.08157104e-03 -1.07544601e+00 -1.09102440e+00 -2.12815925e-02 -9.53162730e-01 -1.04605532e+00 -9.12272871e-01 -6.91790342e-01 1.03109550e+00 4.62177694e-01 8.43255281e-01 4.96180058e-01 -3.76566470e-01 -3.69411744e-02 -2.53259152e-01 -2.85299957e-01 8.67430419e-02 1.44732922e-01 -4.44871634e-01 7.01118633e-02 5.89550510e-02 -2.85578370e-01 -8.08687568e-01 5.41764259e-01 -1.02233708e+00 4.33396548e-01 3.46837968e-01 4.82614845e-01 7.49891222e-01 -7.67540932e-02 3.50242734e-01 -1.43946302e+00 6.22659504e-01 -4.91498038e-02 -6.95251822e-01 2.75818348e-01 -6.01504385e-01 -1.23705544e-01 4.00448769e-01 -5.16855597e-01 -1.02134633e+00 3.25328231e-01 -7.30638057e-02 -1.81506187e-01 -3.10106874e-01 5.41933358e-01 -1.12209514e-01 -2.12896213e-01 5.29918313e-01 -3.78687643e-02 -2.67771959e-01 -8.02548349e-01 5.30808866e-01 5.71541846e-01 5.37239015e-01 -5.23448050e-01 6.37293935e-01 4.93826210e-01 -3.79598975e-01 -7.12227762e-01 -6.28672957e-01 -5.88918865e-01 -8.76654744e-01 -1.61991104e-01 7.15166152e-01 -1.68924198e-01 -5.47796428e-01 3.74017745e-01 -1.06734741e+00 -3.57271671e-01 -1.89659089e-01 8.65317136e-02 -2.39853323e-01 4.77868825e-01 -4.74509209e-01 -7.04085350e-01 -1.70915395e-01 -1.36784410e+00 9.99340594e-01 5.73418319e-01 -3.34441125e-01 -5.81939220e-01 -3.19663793e-01 2.12015435e-01 2.53650606e-01 3.49640101e-01 8.68751466e-01 -5.58481634e-01 -6.18816137e-01 -5.25391519e-01 -3.52097929e-01 4.03332151e-02 3.03760022e-01 1.75477192e-01 -5.25576472e-01 -1.06524497e-01 -4.68028575e-01 -4.47439328e-02 6.92834914e-01 1.59617946e-01 1.52215242e+00 -5.59605397e-02 -4.28636879e-01 4.88898695e-01 1.30424798e+00 4.88982469e-01 5.37519515e-01 4.05748971e-02 6.64110601e-01 4.57481593e-01 7.45866895e-01 3.63945246e-01 4.26852554e-02 8.37095320e-01 7.56469294e-02 -4.39865738e-01 1.24352768e-01 -6.67025521e-02 -3.70656520e-01 5.66417515e-01 -1.71307951e-01 -8.37239176e-02 -7.91622579e-01 4.22150046e-01 -1.86563969e+00 -3.48745346e-01 -2.44080603e-01 2.45257926e+00 1.06860101e+00 5.21015763e-01 2.42010280e-01 2.52899557e-01 8.27255368e-01 -1.57163814e-01 -5.39970338e-01 -1.19593479e-01 2.95347661e-01 4.12330419e-01 2.91793227e-01 6.57902837e-01 -1.28105927e+00 1.04010057e+00 5.99192238e+00 1.05407631e+00 -1.00734603e+00 -2.98528243e-02 6.04822218e-01 -5.76705076e-02 -2.04696268e-01 1.04928397e-01 -6.61899805e-01 5.24382353e-01 7.73462355e-02 -8.37651417e-02 2.61054546e-01 8.39228332e-01 3.85484934e-01 -7.21841872e-01 -1.07844388e+00 9.51800883e-01 -1.22282535e-01 -1.09881473e+00 -4.80272528e-03 -1.62541375e-01 5.31850934e-01 -4.31419998e-01 -1.69953793e-01 4.93956963e-03 1.19072832e-02 -7.10385740e-01 7.40204811e-01 1.76985398e-01 7.32449114e-01 -6.06300950e-01 3.93549532e-01 3.95910800e-01 -1.13655329e+00 4.04688001e-01 8.55822787e-02 2.55858660e-01 3.06638986e-01 6.38765574e-01 -5.82149625e-01 2.91495293e-01 6.80223584e-01 1.63735360e-01 -8.28906238e-01 1.53506625e+00 -6.88941404e-02 5.11261761e-01 -7.37585783e-01 -7.11244792e-02 9.99136195e-02 -3.53420645e-01 3.84931773e-01 1.34058428e+00 -3.91091816e-02 1.87735856e-01 6.67872608e-01 7.97553301e-01 -1.35984691e-03 4.69700038e-01 -1.65669218e-01 1.56830400e-01 4.53971416e-01 1.48036373e+00 -1.51060796e+00 -5.06134987e-01 -3.84104162e-01 1.24636376e+00 2.08401620e-01 4.37027186e-01 -9.16514099e-01 -7.75309741e-01 7.99854696e-02 3.44362259e-01 3.33489239e-01 -4.78141189e-01 -6.04386985e-01 -8.34155619e-01 9.23506618e-02 -6.20790839e-01 2.42191166e-01 -5.00452518e-01 -6.64342403e-01 4.60485190e-01 9.15369242e-02 -1.06227028e+00 2.54598767e-01 -2.52267957e-01 -7.01681852e-01 5.68845630e-01 -1.01458073e+00 -7.41876125e-01 -6.42225385e-01 3.18183601e-01 7.82361984e-01 5.75174570e-01 3.83098960e-01 5.78442693e-01 -5.10980010e-01 6.19734526e-01 -1.93625912e-01 3.02250665e-02 6.24286830e-01 -1.51975262e+00 4.12014604e-01 7.37129211e-01 3.34787637e-01 8.12927544e-01 6.10481143e-01 -7.59741962e-01 -7.55239189e-01 -8.58042240e-01 4.96099770e-01 5.30808680e-02 1.98874041e-01 -5.01684129e-01 -9.55479324e-01 4.57943410e-01 1.78893104e-01 6.58670738e-02 5.47154367e-01 -3.67665030e-02 -6.04847446e-03 2.39931673e-01 -1.12537432e+00 9.50404942e-01 1.04917002e+00 -8.68419707e-02 -1.66538268e-01 3.26011360e-01 5.97422540e-01 -5.58570623e-01 -5.45653880e-01 3.61674309e-01 5.19778669e-01 -9.66139019e-01 7.10694909e-01 -3.09159964e-01 5.22566400e-02 -5.97569644e-01 2.49328837e-01 -1.24411523e+00 7.04900250e-02 -7.73199975e-01 1.76425666e-01 1.08221316e+00 5.01862943e-01 -3.30927372e-01 8.10976028e-01 8.12921226e-01 -9.36359242e-02 -9.26419318e-01 -5.68694830e-01 -6.72411263e-01 -2.13441953e-01 -4.03678834e-01 5.15470803e-01 6.53097928e-01 -1.13872856e-01 1.32005557e-01 4.64710183e-02 -4.67913151e-02 7.36886501e-01 2.83748835e-01 8.77568722e-01 -1.39670551e+00 -2.28136197e-01 -5.63844323e-01 -4.43641134e-02 -1.42345583e+00 -3.95756721e-01 -6.00732744e-01 1.61883757e-01 -1.47752118e+00 1.53955474e-01 -7.13882446e-01 6.72635958e-02 4.64292496e-01 -5.30157626e-01 3.75941068e-01 3.63050222e-01 7.93008879e-02 -6.20005131e-01 1.30103931e-01 1.36504316e+00 -3.56473029e-02 -4.30149823e-01 3.30655545e-01 -4.33703780e-01 1.10118604e+00 6.94387078e-01 -3.80090505e-01 -4.22447354e-01 -2.56806701e-01 1.10232919e-01 -1.42574564e-01 2.14216173e-01 -8.59766304e-01 1.66047856e-01 -3.46030176e-01 2.01048270e-01 -7.95525253e-01 6.10215142e-02 -9.15895998e-01 -2.51117405e-02 3.94480526e-01 -2.39902809e-01 -1.56059235e-01 2.72739142e-01 4.17190164e-01 -3.60127091e-02 -5.28474569e-01 9.65935171e-01 -9.84277204e-02 -4.60515589e-01 3.20435226e-01 -1.22516319e-01 3.03030983e-02 1.26099348e+00 -5.54444551e-01 9.88146439e-02 -1.70686364e-01 -1.11045384e+00 2.47828782e-01 4.91769582e-01 2.93850303e-01 4.74779755e-01 -9.80634093e-01 -1.76893637e-01 1.98765874e-01 5.02612591e-02 3.53779286e-01 7.05576092e-02 7.67817914e-01 -8.09079230e-01 2.91513857e-02 1.43512756e-01 -6.83995426e-01 -1.58784890e+00 2.90892184e-01 1.53474629e-01 9.52825397e-02 -8.38906229e-01 7.13435590e-01 4.56057906e-01 -1.25574753e-01 3.81269902e-01 -5.62260687e-01 -2.13803843e-01 4.92288098e-02 3.59331608e-01 5.10961711e-01 1.14153177e-01 -1.32041425e-01 -1.89810514e-01 6.68216884e-01 -3.35057706e-01 -2.60075003e-01 8.32704604e-01 -1.09130956e-01 1.36498779e-01 4.92220789e-01 7.35866964e-01 2.74294019e-01 -1.35386395e+00 -1.44828692e-01 2.75628150e-01 -8.09911430e-01 -2.09290966e-01 -8.01194131e-01 -1.15151680e+00 9.27739263e-01 6.30692005e-01 5.01027465e-01 9.38950837e-01 -1.91252064e-02 1.00454378e+00 1.99909553e-01 5.92269421e-01 -1.18812120e+00 -9.39309690e-03 2.04131857e-01 7.94995546e-01 -1.04342699e+00 2.84941755e-02 -1.04560935e+00 -6.79165483e-01 8.43074739e-01 8.22575867e-01 3.47625762e-02 8.66785228e-01 4.24566567e-01 8.35751146e-02 -8.84955674e-02 -1.81226164e-01 -1.55226052e-01 2.77748883e-01 2.21028179e-01 4.84383017e-01 2.14431643e-01 -8.91608179e-01 5.86965263e-01 -6.42991215e-02 4.70854640e-02 4.35537845e-01 1.20717549e+00 -5.56747258e-01 -1.11038744e+00 -2.36331969e-01 5.92967451e-01 -5.58660030e-01 -1.83029816e-01 -3.95905137e-01 7.60181189e-01 9.05863941e-02 7.31769264e-01 2.42507935e-01 -9.68400612e-02 4.42447841e-01 2.05979683e-03 3.03033203e-01 -9.03393507e-01 -7.07734764e-01 4.87995476e-01 -1.50967717e-01 -5.58036923e-01 -1.53629407e-01 -5.01507878e-01 -1.57502913e+00 -1.30124629e-01 -8.05804968e-01 8.22715983e-02 6.39296174e-01 8.06296289e-01 2.94468045e-01 3.12300891e-01 3.20208043e-01 -8.30916464e-01 -3.61424834e-01 -9.11673188e-01 -5.27156293e-01 5.87365925e-01 -3.72605249e-02 -7.60999143e-01 -7.57103786e-02 3.66638660e-01]
[9.421625137329102, -0.06974874436855316]
9e64ddf3-2f1b-4233-9159-752c66f7f148
multi-level-cross-modal-interaction-network
2007.14352
null
https://arxiv.org/abs/2007.14352v2
https://arxiv.org/pdf/2007.14352v2.pdf
Multi-level Cross-modal Interaction Network for RGB-D Salient Object Detection
Depth cues with affluent spatial information have been proven beneficial in boosting salient object detection (SOD), while the depth quality directly affects the subsequent SOD performance. However, it is inevitable to obtain some low-quality depth cues due to limitations of its acquisition devices, which can inhibit the SOD performance. Besides, existing methods tend to combine RGB images and depth cues in a direct fusion or a simple fusion module, which makes they can not effectively exploit the complex correlations between the two sources. Moreover, few methods design an appropriate module to fully fuse multi-level features, resulting in cross-level feature interaction insufficient. To address these issues, we propose a novel Multi-level Cross-modal Interaction Network (MCINet) for RGB-D based SOD. Our MCI-Net includes two key components: 1) a cross-modal feature learning network, which is used to learn the high-level features for the RGB images and depth cues, effectively enabling the correlations between the two sources to be exploited; and 2) a multi-level interactive integration network, which integrates multi-level cross-modal features to boost the SOD performance. Extensive experiments on six benchmark datasets demonstrate the superiority of our MCI-Net over 14 state-of-the-art methods, and validate the effectiveness of different components in our MCI-Net. More important, our MCI-Net significantly improves the SOD performance as well as has a higher FPS.
['Bi-Yuan Liu', 'Yun-Zhi Yang', 'Huai-Xin Chen', 'Zhou Huang', 'Tao Zhou']
2020-07-10
null
null
null
null
['rgb-d-salient-object-detection']
['computer-vision']
[-2.03129694e-01 -3.26797366e-01 -2.20508844e-01 -2.26288453e-01 -5.16153216e-01 2.22664341e-01 1.98624194e-01 -8.61345232e-02 -4.43624049e-01 2.87008941e-01 2.96793461e-01 1.08128101e-01 -1.55462369e-01 -1.03956032e+00 -4.65769857e-01 -8.61457288e-01 2.11384133e-01 -4.67674047e-01 1.05655479e+00 -4.30812359e-01 -1.16110705e-02 2.42941305e-01 -1.79201782e+00 2.11509228e-01 1.12180305e+00 1.48298323e+00 5.43292522e-01 -2.42361836e-02 -2.82206804e-01 6.82103515e-01 -2.86023051e-01 5.31817116e-02 1.83846816e-01 -2.27655575e-01 -2.92305648e-01 -7.06424192e-02 3.17302085e-02 -5.47249436e-01 -5.43835342e-01 1.09714687e+00 6.51872873e-01 -8.87119099e-02 2.35179842e-01 -1.24119377e+00 -3.84902686e-01 2.47793928e-01 -9.01668489e-01 2.21391499e-01 1.70965001e-01 3.34828049e-01 7.64236748e-01 -8.96699488e-01 5.51700108e-02 1.41674423e+00 4.06429678e-01 2.13149354e-01 -7.94745743e-01 -9.26445305e-01 4.19837266e-01 3.20046544e-01 -1.24573028e+00 -1.18560018e-02 1.19384074e+00 -1.71950564e-01 5.16740024e-01 9.13677097e-04 1.02758288e+00 7.06160843e-01 1.04294628e-01 1.14953780e+00 1.22745025e+00 -8.38818476e-02 -1.56051159e-01 1.00456178e-01 -1.10426567e-01 6.88137412e-01 2.08251491e-01 2.67703235e-01 -7.09678292e-01 3.22665304e-01 1.09255636e+00 3.09823841e-01 -3.65801543e-01 -4.56963658e-01 -1.25185001e+00 6.27932072e-01 1.27546537e+00 3.63190383e-01 -2.71440655e-01 -1.05508484e-01 2.96790302e-01 -1.50835678e-01 2.69869745e-01 1.98397860e-02 -2.63610572e-01 1.35667667e-01 -4.91802365e-01 1.26727186e-02 1.37707755e-01 6.98116839e-01 1.01429403e+00 -1.78824157e-01 -2.91954398e-01 7.63016522e-01 5.56786299e-01 4.10273284e-01 5.64378500e-01 -5.10035515e-01 6.05150044e-01 1.12073135e+00 -2.99701113e-02 -1.26884162e+00 -6.78372979e-01 -4.69690412e-01 -9.42129016e-01 2.76139617e-01 1.51913777e-01 1.18659735e-01 -6.25820100e-01 1.73005855e+00 4.77266490e-01 3.56016196e-02 -1.13323614e-01 1.43776619e+00 1.28071249e+00 5.57384849e-01 9.66103300e-02 -1.16224922e-01 1.32632828e+00 -1.04699469e+00 -7.45378256e-01 -3.76586795e-01 3.46877337e-01 -5.93795478e-01 1.32471073e+00 2.57615030e-01 -9.08543468e-01 -8.50870907e-01 -1.38600123e+00 -2.87280440e-01 -3.73870313e-01 2.34955937e-01 9.64556217e-01 1.93574980e-01 -5.56957066e-01 2.10642502e-01 -7.20529914e-01 1.27255186e-01 4.77934718e-01 3.50393742e-01 -2.10591629e-01 -2.16631368e-01 -1.43558419e+00 5.63205123e-01 4.75498140e-01 4.38978463e-01 -4.65463340e-01 -5.60283184e-01 -8.81159008e-01 -3.62160988e-02 4.90211487e-01 -5.25614977e-01 1.02158380e+00 -6.54441595e-01 -1.23492277e+00 3.58448744e-01 -1.89295728e-02 2.93950021e-01 5.59879124e-01 -2.18074158e-01 -4.90199447e-01 3.35852742e-01 2.34440818e-01 9.18372810e-01 5.03827274e-01 -1.27764511e+00 -1.09760773e+00 -4.91342992e-01 3.50780755e-01 5.36891043e-01 -7.93965220e-01 -4.49542046e-01 -8.31382453e-01 -4.59841728e-01 5.60365379e-01 -3.36256713e-01 -7.94929713e-02 2.88123935e-01 -2.95730144e-01 -2.30830714e-01 9.99604046e-01 -3.75282407e-01 1.21140730e+00 -2.32680035e+00 1.14380531e-01 -1.38563216e-01 4.27947313e-01 2.97252327e-01 9.95721743e-02 -8.61071050e-02 1.83370560e-01 -5.94063736e-02 -4.89154307e-04 -2.25403160e-01 -1.82051986e-01 1.30876675e-01 1.36741763e-02 2.15002656e-01 3.32833290e-01 9.39100027e-01 -1.07583642e+00 -7.87151039e-01 6.52715743e-01 5.85295320e-01 -2.97585547e-01 2.50421941e-01 -8.29165652e-02 4.00013924e-01 -8.21116149e-01 7.74169505e-01 9.06503677e-01 -1.62331924e-01 -4.44079161e-01 -7.07553148e-01 -3.03331912e-01 1.40009627e-01 -1.27338076e+00 1.75957417e+00 -3.74617904e-01 2.29272172e-01 -8.48268121e-02 -6.05552971e-01 1.02270699e+00 -8.43827799e-02 5.14569998e-01 -1.31949294e+00 2.14993551e-01 3.29374462e-01 -1.51991561e-01 -5.83182096e-01 3.42595905e-01 -1.70629740e-01 -8.03212821e-02 -8.56682435e-02 -1.78264812e-01 1.79931536e-01 -6.94709420e-02 1.66570857e-01 7.27164268e-01 1.09168790e-01 3.43050435e-02 -1.15091689e-01 7.99427986e-01 -3.37156862e-01 9.34413016e-01 3.78924638e-01 -3.28019142e-01 6.28850400e-01 4.83145386e-01 -3.56261611e-01 -4.71202910e-01 -1.02107501e+00 -2.64852583e-01 5.75865746e-01 1.09336042e+00 -3.17731112e-01 -3.80269408e-01 -5.71760178e-01 -2.95763090e-02 4.79055308e-02 -6.24088287e-01 -5.11363149e-01 -3.01350325e-01 -8.15930843e-01 1.44991666e-01 8.00343454e-01 1.22273839e+00 -9.79742050e-01 -7.71646976e-01 2.02507496e-01 -4.18484598e-01 -1.08765578e+00 -3.34652096e-01 1.63497888e-02 -9.47611749e-01 -1.01212478e+00 -6.40512288e-01 -6.21737361e-01 4.74077791e-01 8.08061838e-01 7.38555133e-01 3.18825364e-01 -2.06333235e-01 -9.47063640e-02 -3.98396015e-01 -3.95722687e-01 4.24070090e-01 1.45319700e-01 -1.00540332e-01 8.71524513e-02 3.09167683e-01 -6.54046059e-01 -1.03180313e+00 5.26805341e-01 -1.25245392e+00 4.55632269e-01 9.68536794e-01 8.02321792e-01 6.47880673e-01 2.04879761e-01 4.70147043e-01 -3.17167312e-01 1.75826058e-01 -2.60418802e-01 -3.99500430e-01 1.11627251e-01 -4.12869275e-01 -2.32886925e-01 3.58788431e-01 -3.66441131e-01 -1.12351477e+00 -9.71377734e-03 -2.68334389e-01 -4.88433629e-01 1.12698302e-01 4.90297079e-01 -8.57853830e-01 -2.15154439e-01 2.90037066e-01 2.53142744e-01 -2.00509261e-02 -4.25933152e-01 1.61724418e-01 6.35468364e-01 4.78165537e-01 -4.05905753e-01 7.85801291e-01 4.97828424e-01 -4.09759618e-02 -4.49380159e-01 -1.11919248e+00 -4.43525702e-01 -5.75060666e-01 -3.10278386e-01 8.43379021e-01 -1.24144053e+00 -7.70060480e-01 8.77345026e-01 -8.39846313e-01 -7.87403286e-02 -5.74504212e-02 5.84932804e-01 -1.93201140e-01 3.00309002e-01 -6.23020768e-01 -5.59127927e-01 -1.68977946e-01 -1.32680607e+00 1.20422971e+00 1.00884843e+00 4.98759627e-01 -6.01304710e-01 -5.49312234e-01 3.07656139e-01 3.97583127e-01 3.01147103e-01 7.63252795e-01 2.17815533e-01 -8.92455757e-01 4.42219619e-03 -8.67292881e-01 2.62042910e-01 3.75534713e-01 -1.86645567e-01 -1.03537190e+00 -3.33955102e-02 -8.98602456e-02 -3.43296051e-01 9.46861506e-01 3.66740257e-01 1.29383433e+00 2.78443336e-01 -4.53715414e-01 7.59819150e-01 1.49110389e+00 -2.96607986e-03 6.96195364e-01 6.53241217e-01 9.91366982e-01 5.04884183e-01 1.01146972e+00 3.28951329e-01 6.46739304e-01 8.12588632e-01 7.23984957e-01 -6.45737946e-01 -1.15105361e-01 -3.64510179e-01 1.96436495e-01 7.78531492e-01 6.90831244e-02 2.20775872e-01 -5.97155452e-01 4.91523951e-01 -1.96237850e+00 -5.93824387e-01 -2.19685182e-01 1.84245026e+00 8.07343125e-01 4.59372073e-01 2.75510937e-01 3.10619503e-01 4.51771438e-01 2.39541009e-01 -7.77822971e-01 4.06655490e-01 -3.34878772e-01 -2.19353318e-01 2.76892215e-01 6.87109400e-03 -1.18151653e+00 5.28003991e-01 4.61915588e+00 1.07974422e+00 -1.35530806e+00 1.34996206e-01 5.70562541e-01 -2.09608957e-01 -4.02812898e-01 -1.75917208e-01 -7.62935281e-01 6.88901424e-01 2.84143798e-02 2.62541436e-02 -1.16499528e-01 9.59483922e-01 1.18662797e-01 -4.41691130e-01 -6.23141170e-01 1.17269897e+00 -5.38118556e-02 -1.02827513e+00 -1.92125559e-01 1.39343098e-01 5.65026104e-01 -6.48195297e-02 4.66337204e-02 3.54657948e-01 -1.53419703e-01 -6.27597094e-01 8.15530777e-01 6.56388164e-01 6.11469865e-01 -9.66809571e-01 9.70858157e-01 4.01750952e-01 -1.67751145e+00 -2.35715359e-01 -5.22691667e-01 -5.10773659e-02 1.15373947e-01 9.21836078e-01 1.58515334e-01 1.03549480e+00 1.11481678e+00 1.09353662e+00 -7.61008143e-01 1.26168978e+00 -3.13718081e-01 1.20821171e-01 -3.60345513e-01 -8.16977862e-03 3.98985356e-01 -1.09305531e-01 2.32740209e-01 1.01309931e+00 1.97568193e-01 3.48749787e-01 3.77715915e-01 8.72382343e-01 2.50322789e-01 3.16930236e-03 -2.56733686e-01 4.28158402e-01 3.16821098e-01 1.52026987e+00 -7.63368070e-01 -2.52265751e-01 -6.07647777e-01 7.36688018e-01 2.36548841e-01 2.02696189e-01 -8.22164953e-01 -5.49999416e-01 7.48585999e-01 1.02245040e-01 2.95636237e-01 -1.00398682e-01 -3.87945324e-01 -1.21542060e+00 3.75159740e-01 -6.78903401e-01 4.49343890e-01 -9.71484244e-01 -1.27954078e+00 6.16561115e-01 -3.57623622e-02 -1.71677697e+00 4.06612337e-01 -5.37465572e-01 -5.50049424e-01 1.01466691e+00 -2.09477425e+00 -1.20079255e+00 -9.53229368e-01 6.37862742e-01 2.70022750e-01 2.24561751e-01 2.15314254e-01 5.54702878e-01 -8.40581775e-01 4.60495740e-01 -3.06242436e-01 1.10354647e-01 5.25538921e-01 -9.94130552e-01 -2.61453390e-01 8.74241412e-01 -3.24137688e-01 3.88604760e-01 8.09931904e-02 -3.67696792e-01 -1.29656386e+00 -9.72008288e-01 2.82539189e-01 -8.48841742e-02 2.51470447e-01 -2.12079257e-01 -9.81156230e-01 1.20417558e-01 -3.02945375e-01 3.68692100e-01 2.06812412e-01 -1.51080787e-01 -2.65222132e-01 -6.45838857e-01 -8.24446261e-01 5.82350671e-01 1.13863218e+00 -4.68063414e-01 -4.58909273e-01 1.07820861e-01 1.00832498e+00 -5.67035198e-01 -8.12806547e-01 8.73429000e-01 5.39803922e-01 -1.54714131e+00 1.12025476e+00 1.58897594e-01 7.92580068e-01 -7.23664939e-01 -6.46362230e-02 -9.33545172e-01 -3.23395908e-01 1.54269010e-01 -1.31372824e-01 1.48142028e+00 -1.17904179e-01 -6.25400960e-01 5.35657883e-01 3.48195612e-01 -2.55857974e-01 -1.28034687e+00 -8.58872116e-01 -5.84289491e-01 -3.13469350e-01 -5.32905459e-01 8.69016945e-01 6.24280989e-01 -1.80038348e-01 3.25866103e-01 -2.46001720e-01 2.48385146e-01 4.47088003e-01 2.99376965e-01 6.40204430e-01 -1.24359393e+00 -1.14321426e-01 -7.47529149e-01 -5.99908113e-01 -1.38503730e+00 -3.55736673e-01 -4.75497514e-01 9.81980935e-02 -1.74357378e+00 2.37160072e-01 -8.33103836e-01 -6.14081264e-01 5.05872369e-01 -6.28781080e-01 2.92245507e-01 6.53275400e-02 2.56402642e-01 -6.72919452e-01 1.04487312e+00 1.71233439e+00 -3.10246944e-02 -2.24343553e-01 -2.45974585e-01 -1.00255597e+00 7.75779605e-01 4.31521803e-01 -2.25790963e-01 -4.75226820e-01 -4.86052066e-01 2.73906868e-02 5.75355738e-02 5.98431647e-01 -1.38738990e+00 3.42729867e-01 -1.46638528e-01 8.85587215e-01 -9.90149081e-01 4.54274565e-01 -7.71339774e-01 -3.12831551e-01 3.80875856e-01 2.56186217e-01 -3.26087564e-01 3.00428718e-01 5.32151699e-01 -5.94426513e-01 2.44786650e-01 6.92865193e-01 -1.05843782e-01 -1.03951812e+00 6.01193726e-01 2.63064802e-01 -2.20961645e-01 9.63532448e-01 -4.84986156e-01 -3.13808113e-01 -1.68093801e-01 -1.54406294e-01 5.20882726e-01 4.71699774e-01 7.21384585e-01 6.91277564e-01 -1.55816972e+00 -2.38884971e-01 4.67632979e-01 2.91508108e-01 5.80849648e-01 6.66033685e-01 1.15568411e+00 -2.86351591e-01 2.03098893e-01 -4.97800261e-01 -9.23533499e-01 -7.76315272e-01 4.39381242e-01 4.11852568e-01 -1.80849895e-01 -5.76162398e-01 8.09643805e-01 6.33494616e-01 -2.37363741e-01 1.53167859e-01 -3.19005221e-01 -4.82790977e-01 1.11011118e-01 6.09647453e-01 9.11779851e-02 7.25865290e-02 -6.25405550e-01 -4.98585075e-01 9.24620509e-01 -9.15751699e-03 8.38985890e-02 1.23861098e+00 -3.38484585e-01 -8.59429911e-02 5.20927846e-01 1.09935391e+00 -2.79068559e-01 -1.56534636e+00 -5.75245142e-01 -5.36626816e-01 -7.58239388e-01 4.12507683e-01 -5.06832600e-01 -1.46386969e+00 1.16273642e+00 7.44695425e-01 9.41883586e-03 1.70239413e+00 -1.60809040e-01 1.05545044e+00 -3.29040475e-02 4.78260517e-01 -8.92558217e-01 5.19715965e-01 7.58147463e-02 5.59564233e-01 -1.45242369e+00 2.06724972e-01 -7.53992856e-01 -5.55335701e-01 1.05995369e+00 1.17435598e+00 4.91180941e-02 7.57455826e-01 1.23451233e-01 1.15248598e-01 -2.04656959e-01 -3.10729474e-01 -4.15919930e-01 4.33806688e-01 5.05774498e-01 1.27930194e-01 -2.05459759e-01 -2.71395624e-01 9.64144826e-01 5.55475093e-02 1.02721848e-01 -1.14788637e-02 8.77467096e-01 -4.57050592e-01 -8.97670925e-01 -2.06897214e-01 3.97157133e-01 -1.31169111e-01 6.16968907e-02 -4.74915877e-02 9.88419533e-01 6.40675128e-01 9.23221648e-01 -5.30632660e-02 -8.13525438e-01 6.25670850e-01 -5.69671512e-01 1.53209999e-01 -1.84155121e-01 -3.22459847e-01 2.63083637e-01 -3.78122658e-01 -7.21726000e-01 -6.54585004e-01 -4.73806351e-01 -1.28717732e+00 -1.43722609e-01 -4.29202318e-01 -1.74606755e-01 3.75191152e-01 1.03248954e+00 2.73504019e-01 8.73747468e-01 7.05461562e-01 -1.00185573e+00 -1.72972322e-01 -9.28011715e-01 -6.08862281e-01 2.39672706e-01 2.31391162e-01 -1.20739651e+00 -2.03288004e-01 -4.30031061e-01]
[9.683695793151855, -0.8703410029411316]
50101c63-c9a7-456e-bfc4-e0de707a801a
nqe-n-ary-query-embedding-for-complex-query
2211.13469
null
https://arxiv.org/abs/2211.13469v3
https://arxiv.org/pdf/2211.13469v3.pdf
NQE: N-ary Query Embedding for Complex Query Answering over Hyper-Relational Knowledge Graphs
Complex query answering (CQA) is an essential task for multi-hop and logical reasoning on knowledge graphs (KGs). Currently, most approaches are limited to queries among binary relational facts and pay less attention to n-ary facts (n>=2) containing more than two entities, which are more prevalent in the real world. Moreover, previous CQA methods can only make predictions for a few given types of queries and cannot be flexibly extended to more complex logical queries, which significantly limits their applications. To overcome these challenges, in this work, we propose a novel N-ary Query Embedding (NQE) model for CQA over hyper-relational knowledge graphs (HKGs), which include massive n-ary facts. The NQE utilizes a dual-heterogeneous Transformer encoder and fuzzy logic theory to satisfy all n-ary FOL queries, including existential quantifiers, conjunction, disjunction, and negation. We also propose a parallel processing algorithm that can train or predict arbitrary n-ary FOL queries in a single batch, regardless of the kind of each query, with good flexibility and extensibility. In addition, we generate a new CQA dataset WD50K-NFOL, including diverse n-ary FOL queries over WD50K. Experimental results on WD50K-NFOL and other standard CQA datasets show that NQE is the state-of-the-art CQA method over HKGs with good generalization capability. Our code and dataset are publicly available.
['Kaiyang Wan', 'Xueyuan Lin', 'Zichen Tang', 'Tianyu Yao', 'Yikai Guo', 'Gengxian Zhou', 'Yuhao Yang', 'Haihong E', 'Haoran Luo']
2022-11-24
null
null
null
null
['complex-query-answering', 'logical-reasoning']
['knowledge-base', 'reasoning']
[-4.01619822e-01 5.16353287e-02 -5.34463942e-01 -4.89057660e-01 -4.46587414e-01 -5.05433440e-01 2.49181405e-01 5.43715239e-01 -2.86393195e-01 7.42939055e-01 7.83897266e-02 -5.49937129e-01 -7.43941963e-01 -1.86408901e+00 -8.46167684e-01 -6.46387860e-02 -1.40583590e-01 9.10973728e-01 5.69414854e-01 -7.91612685e-01 -3.15252513e-01 3.57446611e-01 -1.56050766e+00 3.70322973e-01 7.85691679e-01 1.31269121e+00 -3.31861407e-01 2.61037618e-01 -2.54585266e-01 1.21724391e+00 -3.25890750e-01 -9.69652534e-01 -1.44365251e-01 2.23142013e-01 -1.28764129e+00 -6.00060344e-01 3.92766804e-01 -2.31067985e-01 -7.95770943e-01 9.72557247e-01 2.31835365e-01 1.71756193e-01 2.50223994e-01 -1.45156085e+00 -9.59180653e-01 9.20935512e-01 2.41550893e-01 1.81355551e-01 8.06207478e-01 -2.48288676e-01 1.75407445e+00 -7.67606378e-01 3.97822440e-01 1.46863627e+00 4.25522298e-01 2.58098632e-01 -9.01943147e-01 -7.25909293e-01 -5.99653181e-03 7.50907362e-01 -1.84883010e+00 -6.67631552e-02 4.77744102e-01 6.04968555e-02 1.30324435e+00 3.66817564e-01 6.13331795e-01 3.86236340e-01 1.20410938e-02 8.57268512e-01 6.18850231e-01 -3.80590379e-01 1.92677200e-01 2.23186128e-02 1.96282029e-01 9.55947399e-01 4.83576953e-01 -3.03089261e-01 -4.94175524e-01 -2.80685544e-01 4.45808321e-01 7.53515139e-02 -1.27677903e-01 -5.93912452e-02 -1.08023453e+00 1.02584422e+00 6.18394911e-01 1.81571081e-01 -1.75889537e-01 1.66099250e-01 4.58912015e-01 6.55965745e-01 -2.52918154e-01 4.89902884e-01 -6.71982527e-01 1.65001184e-01 -3.08404356e-01 5.47882795e-01 1.01441479e+00 1.37233734e+00 9.92082179e-01 -3.63451988e-01 -3.08373153e-01 5.42077065e-01 2.13027075e-01 6.55235946e-01 1.42808855e-01 -9.11408246e-01 5.25545835e-01 1.24432254e+00 -9.45639238e-02 -1.28611314e+00 -3.22431624e-01 -1.52467892e-01 -8.17243516e-01 -5.43458581e-01 2.45368946e-02 3.37656707e-01 -4.80953306e-01 1.61210823e+00 4.29263681e-01 -2.14737013e-01 4.62426990e-01 5.62328458e-01 1.20586562e+00 6.00430548e-01 -9.22856331e-02 -3.25087085e-02 1.64176202e+00 -6.09644473e-01 -7.45386481e-01 1.40995339e-01 8.79846513e-01 -2.87904382e-01 1.18345737e+00 2.97855109e-01 -9.58113313e-01 -3.39670420e-01 -8.68835270e-01 -7.00395107e-01 -9.62680101e-01 -1.54439583e-01 1.21245563e+00 3.91885906e-01 -8.32949102e-01 4.71942089e-02 -3.95316213e-01 -4.90431711e-02 3.81776661e-01 5.96179545e-01 -3.62388253e-01 -6.56858087e-01 -2.15115690e+00 7.00451732e-01 1.03242648e+00 1.29623255e-02 -4.62664723e-01 -7.00881541e-01 -1.20926702e+00 4.00743872e-01 1.26466906e+00 -8.52902055e-01 9.93052840e-01 2.03405350e-01 -1.16865385e+00 5.27801633e-01 -2.12422296e-01 -3.99122745e-01 -9.72800776e-02 -5.48035614e-02 -1.12711966e+00 1.25120804e-01 1.95107609e-01 4.26713437e-01 1.12517871e-01 -7.60588825e-01 -6.45497918e-01 -3.84171426e-01 1.08667052e+00 1.40588522e-01 -2.77427465e-01 -1.25678897e-01 -5.70431411e-01 -2.47341752e-01 1.67196970e-02 -3.61089796e-01 2.81681746e-01 -1.95185736e-01 -6.40916646e-01 -9.71125782e-01 7.28895605e-01 -3.29888351e-02 1.54039156e+00 -1.96533823e+00 -9.36800763e-02 4.85617399e-01 3.39944839e-01 2.66844094e-01 2.09918931e-01 7.31831253e-01 2.72142589e-01 7.01424256e-02 4.15167920e-02 3.12702864e-01 5.15885472e-01 7.67667055e-01 -5.87110996e-01 -2.45357558e-01 2.87092388e-01 1.19612277e+00 -1.13397980e+00 -9.54287887e-01 -9.36621353e-02 2.47495789e-02 -7.34602630e-01 3.24260183e-02 -7.43569970e-01 -4.17916596e-01 -7.21016526e-01 1.08729041e+00 4.96564686e-01 -6.07432663e-01 2.27434546e-01 -5.95097899e-01 3.86478275e-01 4.33521539e-01 -1.31365347e+00 1.41570890e+00 -4.62551087e-01 -2.41058059e-02 -4.22245443e-01 -8.01840663e-01 8.22892845e-01 2.33888671e-01 2.85768509e-01 -9.03159320e-01 -6.48285002e-02 3.47179592e-01 -1.74784213e-01 -5.59019208e-01 7.84583628e-01 -1.86206624e-01 -4.89246547e-01 5.87722100e-02 1.64773956e-01 -3.08717757e-01 7.09454477e-01 5.86409032e-01 1.10059118e+00 -5.61912000e-01 3.33368748e-01 6.60571158e-02 7.92066097e-01 1.17305294e-01 6.84687018e-01 7.45289743e-01 1.99553952e-01 -2.75855035e-01 6.67404592e-01 -5.03656328e-01 -3.78024578e-01 -1.00174725e+00 -1.84026197e-01 1.09353960e+00 4.31750774e-01 -1.06561732e+00 2.88821049e-02 -7.69543111e-01 4.74926233e-01 7.93027341e-01 -3.15769434e-01 -3.92591447e-01 -4.58783865e-01 -4.67689574e-01 8.94340575e-01 6.27931952e-01 7.41234183e-01 -1.08878195e+00 -2.43523642e-01 2.19093591e-01 -3.44897568e-01 -1.45973098e+00 -1.99537978e-01 -5.27381040e-02 -4.15216386e-01 -1.43440223e+00 2.47158453e-01 -7.49831796e-01 2.82958150e-01 -8.77197012e-02 1.42107391e+00 7.51839206e-02 6.15974367e-02 3.38767320e-01 -3.37385833e-01 -3.83761793e-01 -1.58692583e-01 7.20383674e-02 1.22590683e-01 -1.96890444e-01 8.23524296e-01 -3.11982840e-01 -3.57529283e-01 2.99549788e-01 -1.30924523e+00 -2.83525854e-01 4.07706529e-01 7.43023038e-01 9.55506682e-01 9.03397679e-01 6.12514019e-01 -1.21304846e+00 6.46688581e-01 -5.11911213e-01 -7.67602921e-01 8.94046485e-01 -7.19232023e-01 1.81115016e-01 1.13921344e+00 -1.04957595e-01 -7.50567496e-01 -3.66758317e-01 -1.23752721e-01 -4.52907681e-01 1.61349326e-01 1.14583182e+00 -5.38773894e-01 1.39095066e-02 4.06530082e-01 1.65550202e-01 -6.19261980e-01 -2.17082649e-02 6.44352138e-01 4.55460697e-01 7.08963454e-01 -1.00118911e+00 6.82903647e-01 3.07928443e-01 3.70254397e-01 -4.16043311e-01 -1.08825588e+00 -4.32321876e-01 -1.79447949e-01 2.39250660e-01 4.91752982e-01 -9.74040091e-01 -1.46464932e+00 1.21229760e-01 -8.05266082e-01 -2.58900207e-02 -4.42445636e-01 5.10649741e-01 -3.63193154e-01 2.67815173e-01 -6.36134267e-01 -5.81131876e-01 -3.79806131e-01 -8.68189871e-01 9.12261486e-01 2.29859471e-01 9.07193422e-02 -8.93561959e-01 -2.67021477e-01 4.40816492e-01 2.34408855e-01 -6.72716349e-02 1.56081414e+00 -7.14367628e-01 -7.87288785e-01 -2.21611217e-01 -3.29209685e-01 2.09727913e-01 1.29075766e-01 -1.67224929e-01 -5.08278966e-01 -1.52756542e-01 -4.22737956e-01 -1.01821077e+00 5.61674356e-01 -1.79401457e-01 1.36382115e+00 -4.80675101e-01 -1.34209722e-01 3.85353833e-01 1.57956946e+00 -7.17298593e-03 7.00834513e-01 3.31953727e-02 5.98760009e-01 9.27222073e-02 8.43531013e-01 3.89273226e-01 1.27037764e+00 5.36577046e-01 5.52322090e-01 4.09491509e-01 3.39743108e-01 -5.96403658e-01 -1.23845369e-01 9.32148039e-01 5.26254363e-02 -2.40145162e-01 -8.83157372e-01 6.09179974e-01 -1.67985427e+00 -9.17163432e-01 3.33027691e-02 1.86843932e+00 1.25121260e+00 2.37629086e-01 -1.29441202e-01 4.00659353e-01 2.54312843e-01 7.99313560e-02 -5.82466364e-01 -2.23090827e-01 -4.07490402e-01 4.38211590e-01 4.14731294e-01 3.95209610e-01 -8.94594908e-01 1.17072415e+00 5.43396616e+00 1.09951699e+00 -7.95327961e-01 -1.68721125e-01 -3.02004442e-02 1.48524672e-01 -7.85459042e-01 1.33361146e-01 -9.85320032e-01 2.56588012e-01 7.18397558e-01 -3.25365484e-01 6.58674717e-01 6.59840405e-01 -6.05691671e-01 -5.80986105e-02 -1.19691479e+00 1.01777732e+00 -2.33899489e-01 -1.43897438e+00 3.97435695e-01 -3.65509182e-01 4.91108507e-01 -1.90084413e-01 -2.09832013e-01 9.71787095e-01 5.64637542e-01 -1.15641320e+00 3.42515141e-01 7.06254542e-01 1.01400638e+00 -8.45612586e-01 8.51098180e-01 2.63050109e-01 -1.38435376e+00 -2.68207878e-01 -6.18943572e-01 1.44111678e-01 -5.64804561e-02 6.05507553e-01 -4.97843623e-01 1.29934359e+00 9.00308073e-01 5.79821944e-01 -5.41009009e-01 6.05408311e-01 -1.95378691e-01 2.79148191e-01 -6.29322767e-01 -2.44317740e-01 5.48859127e-02 2.11519599e-01 5.79303317e-03 8.69529545e-01 1.74197555e-01 5.88616788e-01 2.41996914e-01 7.95360684e-01 -6.04368925e-01 1.24274857e-01 -4.28992331e-01 -1.90516353e-01 9.37159598e-01 8.74800146e-01 -1.48270637e-01 -6.23754144e-01 -7.19467640e-01 4.09535021e-01 5.03961265e-01 3.17565143e-01 -7.34117091e-01 -8.41366410e-01 4.35506731e-01 -1.14236744e-02 4.07362700e-01 1.76054105e-01 2.80019879e-01 -1.47611606e+00 4.89924312e-01 -9.03128624e-01 1.21571088e+00 -7.71366417e-01 -1.47935593e+00 3.46718192e-01 2.53616869e-01 -7.65513718e-01 -1.18814334e-01 -5.68630993e-01 -1.93060324e-01 6.77688003e-01 -1.94609475e+00 -1.24268568e+00 -1.54176489e-01 1.22403288e+00 -2.47813255e-01 1.68085413e-03 1.07967198e+00 5.98056853e-01 -2.88836271e-01 8.72058749e-01 -3.00248981e-01 2.30605528e-01 4.92747575e-01 -1.31450629e+00 -1.36833489e-01 3.10818881e-01 7.74487257e-02 9.07895267e-01 2.82573819e-01 -4.71333534e-01 -1.94863033e+00 -1.23933554e+00 1.36406314e+00 -2.23673657e-01 7.50991285e-01 -2.37003401e-01 -1.10109603e+00 9.87880707e-01 -2.76603311e-01 5.59368074e-01 6.68902338e-01 4.69056010e-01 -7.07346797e-01 -6.88488543e-01 -1.13129520e+00 2.83693463e-01 9.69560981e-01 -9.43115056e-01 -8.22815537e-01 4.49014604e-01 1.04081047e+00 -5.80919325e-01 -1.50168061e+00 9.02715027e-01 1.98965237e-01 -6.91273928e-01 1.03211737e+00 -8.94828439e-01 1.50548175e-01 -7.11892545e-01 -5.67450702e-01 -7.92429686e-01 -2.39534259e-01 -2.82855570e-01 -7.59116173e-01 8.99809182e-01 3.08229983e-01 -7.64634669e-01 6.19668424e-01 5.36781371e-01 3.36933546e-02 -1.08340180e+00 -1.01938903e+00 -7.05109119e-01 -1.89899340e-01 -5.61841071e-01 1.34330773e+00 1.07729709e+00 1.83552787e-01 4.07152623e-01 -4.49485704e-02 4.58490402e-01 2.22278342e-01 7.09181666e-01 6.37806416e-01 -1.27905595e+00 -3.20143402e-01 -2.40412965e-01 -7.51211643e-01 -9.49639797e-01 1.38523400e-01 -1.08452761e+00 -5.13658524e-01 -1.62969255e+00 -5.07389195e-02 -6.33483529e-01 -4.02365506e-01 9.31440592e-01 1.45584177e-02 -2.21538872e-01 -1.26841351e-01 -1.24649331e-01 -1.01324618e+00 7.19546139e-01 1.40393615e+00 -4.38700855e-01 1.53110102e-01 -1.12451077e-01 -7.78802514e-01 2.84330606e-01 3.72411579e-01 -2.00563088e-01 -8.64110410e-01 -4.18683022e-01 1.17241204e+00 2.95996845e-01 4.74855363e-01 -8.97906721e-01 6.51919603e-01 -3.19859833e-01 -3.32182609e-02 -7.05644369e-01 3.76872987e-01 -8.19323957e-01 -2.63493825e-02 5.95036931e-02 -2.47804135e-01 1.85139820e-01 2.14227065e-02 5.77479243e-01 -8.17449510e-01 5.07253520e-02 1.26471549e-01 -2.30109673e-02 -9.12997782e-01 5.52045107e-01 3.64617050e-01 4.84801859e-01 7.66622186e-01 2.65238643e-01 -6.57420814e-01 -3.10043544e-01 -5.50632775e-01 8.54206979e-01 1.51802748e-02 1.54245242e-01 7.94907689e-01 -1.49460292e+00 -5.31427145e-01 1.59665689e-01 5.61639845e-01 7.94555783e-01 3.88542175e-01 5.87162197e-01 -3.84599179e-01 9.07142162e-01 1.41277581e-01 -8.63395035e-02 -7.43305445e-01 9.37725008e-01 2.66566932e-01 -5.09110093e-01 -3.58028501e-01 6.76121652e-01 -2.77159691e-01 -6.76045179e-01 2.90569335e-01 -6.55630410e-01 -3.23494464e-01 -2.14396164e-01 3.66748720e-01 2.59781569e-01 2.88033426e-01 -4.50475067e-01 -5.50456941e-01 3.33395779e-01 -1.55649170e-01 5.10994136e-01 9.35699284e-01 1.50516763e-01 -5.13984442e-01 4.00010347e-01 1.09179592e+00 1.96947940e-02 -1.10110216e-01 -8.19629312e-01 -1.39005125e-01 -3.82822812e-01 -1.95776641e-01 -7.84838498e-01 -1.08049774e+00 5.83860397e-01 -9.27398503e-02 5.10627747e-01 1.28923249e+00 4.22281712e-01 1.09107089e+00 1.20093393e+00 6.45070016e-01 -1.04684782e+00 -2.34656841e-01 7.81023502e-01 6.61985636e-01 -9.75662351e-01 -9.25392087e-04 -6.11153483e-01 -4.61623609e-01 1.06796789e+00 6.84617043e-01 2.36018568e-01 7.01043487e-01 -2.01382264e-01 -2.68787622e-01 -6.72552347e-01 -9.30091381e-01 -3.46570820e-01 1.76954389e-01 1.71701625e-01 4.66023348e-02 2.88094372e-01 -7.68033862e-02 9.06650543e-01 -5.30692816e-01 4.37696368e-01 2.97747105e-01 9.82610762e-01 -8.14410523e-02 -1.10378575e+00 2.07266286e-02 7.65293062e-01 -4.06339198e-01 -1.82335570e-01 -3.58637609e-02 9.30423498e-01 2.16513395e-01 1.00644100e+00 -3.10696542e-01 -5.59746265e-01 5.70442021e-01 4.87719215e-02 4.11863774e-01 -4.59098756e-01 -2.26551160e-01 -7.23055780e-01 2.44714975e-01 -5.56705415e-01 -3.12061489e-01 -2.58713126e-01 -1.69131970e+00 -6.26966119e-01 -3.84409904e-01 3.87739390e-01 4.07101624e-02 8.91050041e-01 4.08759326e-01 4.33878988e-01 3.62752438e-01 6.79533005e-01 -5.21228850e-01 -5.71682930e-01 -7.70057976e-01 4.51872885e-01 9.98236686e-02 -7.46343851e-01 1.06198877e-01 -3.94908190e-01]
[9.122954368591309, 7.666326999664307]
0a4ce2e3-3a4e-412d-8e1f-17fc41e61e09
refining-data-for-text-generation
null
null
https://aclanthology.org/2020.ccl-1.82
https://aclanthology.org/2020.ccl-1.82.pdf
Refining Data for Text Generation
Recent work on data-to-text generation has made progress under the neural encoder-decoder architectures. However, the data input size is often enormous, while not all data records are important for text generation and inappropriate input may bring noise into the final output. To solve this problem, we propose a two-step approach which first selects and orders the important data records and then generates text from the noise-reduced data. Here we propose a learning to rank model to rank the importance of each record which is supervised by a relation extractor. With the noise-reduced data as input, we implement a text generator which sequentially models the input data records and emits a summary. Experiments on the ROTOWIRE dataset verifies the effectiveness of our proposed method in both performance and efficiency.
['Sujian Li', 'Tianyi Li', 'Qianying Liu', 'Wenyu Guan']
null
null
null
null
ccl-2020-10
['data-to-text-generation']
['natural-language-processing']
[ 6.36255145e-01 3.75132322e-01 -3.77411813e-01 -3.83091629e-01 -7.58054733e-01 -2.48542055e-01 6.09892905e-01 4.45279777e-01 -4.80234683e-01 1.02863538e+00 7.30320811e-01 -2.34509438e-01 -3.80656049e-02 -1.06450284e+00 -5.86765110e-01 -3.59958410e-01 4.08636719e-01 6.26169622e-01 -5.01287431e-02 -1.65794566e-01 4.64838505e-01 -4.30878848e-02 -1.68051910e+00 6.62484467e-01 1.10276878e+00 7.91944981e-01 3.27339560e-01 8.21970642e-01 -4.82509017e-01 9.84692693e-01 -9.55776334e-01 -3.49829882e-01 1.48557276e-01 -7.91701853e-01 -9.13928568e-01 -7.84292594e-02 -2.16048453e-02 -3.29799980e-01 -4.44263369e-01 9.90024447e-01 7.00068235e-01 2.42462397e-01 7.47654378e-01 -1.03161800e+00 -7.88249493e-01 1.43398404e+00 -3.42824847e-01 1.81651413e-01 3.69637698e-01 -1.16603434e-01 1.14476562e+00 -9.53008890e-01 6.11567378e-01 1.06984138e+00 4.61705700e-02 7.62477219e-01 -1.20087957e+00 -5.80036581e-01 8.03638697e-02 -4.22041640e-02 -1.26275754e+00 -5.26213109e-01 7.57983863e-01 -1.82721704e-01 1.06181920e+00 3.17497998e-01 3.89025778e-01 1.10580695e+00 -5.46208117e-03 9.06406522e-01 5.46683908e-01 -6.17438674e-01 2.02111080e-01 4.52787653e-02 2.89079994e-01 2.39800796e-01 5.60285151e-01 -1.63332187e-02 -7.99798727e-01 -2.87008900e-02 4.53237385e-01 -5.76542020e-02 8.78027640e-03 4.54505920e-01 -1.15087950e+00 5.73467135e-01 1.52781993e-01 1.29479095e-01 -5.60017228e-01 3.20513956e-02 3.44883651e-01 4.27783519e-01 4.11197662e-01 4.87462252e-01 -3.84193689e-01 -2.61381835e-01 -9.77530479e-01 4.86297995e-01 9.02232170e-01 1.27807486e+00 4.01787817e-01 -6.85784370e-02 -7.04641581e-01 8.19375873e-01 1.79594696e-01 3.38940114e-01 7.38275945e-01 -1.99023843e-01 1.21104407e+00 9.56847072e-01 -4.37411554e-02 -7.89576828e-01 -2.42655456e-01 -2.57900476e-01 -1.26631188e+00 -1.31264897e-02 4.79438007e-02 -4.53470618e-01 -1.11244702e+00 1.50381088e+00 -4.26111370e-02 -2.83675253e-01 4.15210634e-01 6.65093005e-01 9.58965600e-01 8.41200709e-01 1.88336000e-02 -4.33830649e-01 1.12613392e+00 -8.35849106e-01 -9.98254716e-01 -2.40461588e-01 6.08837008e-01 -6.58193290e-01 1.15681863e+00 3.19238812e-01 -1.19894016e+00 -5.95031440e-01 -1.09743810e+00 -3.90516728e-01 -2.52706885e-01 4.66284931e-01 2.40817070e-01 2.96654046e-01 -6.48516536e-01 5.51340163e-01 -5.03832340e-01 -1.24694824e-01 4.47444320e-01 3.84930611e-01 -6.84509128e-02 2.91170895e-01 -1.30891502e+00 6.20905340e-01 1.00742066e+00 -1.36554455e-02 -5.53869903e-01 -3.86021048e-01 -7.74112284e-01 2.70134717e-01 4.72740471e-01 -7.09098756e-01 1.30157852e+00 -4.26725686e-01 -1.36662030e+00 1.63022965e-01 -4.04421329e-01 -4.66090471e-01 3.70492339e-01 -3.75345975e-01 -4.60296035e-01 -3.32608730e-01 -1.39579959e-02 5.06246626e-01 4.91557062e-01 -1.02421141e+00 -8.68424654e-01 -3.32213007e-02 -2.92945296e-01 3.51883680e-01 -5.07642984e-01 -2.59806439e-02 -4.24243033e-01 -9.66911495e-01 4.54500206e-02 -5.57036877e-01 -4.86224234e-01 -5.28338373e-01 -9.53044772e-01 -4.79932636e-01 4.81293172e-01 -4.73328263e-01 1.82402802e+00 -1.96917605e+00 -1.05761718e-02 2.91916221e-01 1.07250519e-01 1.33955851e-01 -1.12682767e-01 6.99314237e-01 -3.62078175e-02 3.97676498e-01 -1.73003301e-01 -2.43985295e-01 -9.00128335e-02 1.16962809e-02 -6.15953982e-01 -4.71803814e-01 3.99050832e-01 9.78914618e-01 -8.84018123e-01 -6.29675925e-01 -1.50990292e-01 1.11274697e-01 -5.18349886e-01 6.20583713e-01 -4.61382598e-01 -2.34425683e-02 -5.71310580e-01 2.13433608e-01 2.72571266e-01 -2.64114618e-01 1.67288650e-02 4.23510112e-02 4.77547571e-02 8.35947692e-01 -1.28135157e+00 1.45612872e+00 -2.07222283e-01 4.87780064e-01 -6.59631014e-01 -5.55954695e-01 1.20592690e+00 3.70927334e-01 6.91890568e-02 -6.03375733e-01 1.70384169e-01 1.33036762e-01 1.07504822e-01 -6.94508135e-01 1.04578686e+00 -3.51361632e-02 -1.25378460e-01 7.62461364e-01 -1.45829573e-01 -7.58733153e-02 5.98360419e-01 4.72174615e-01 1.14924908e+00 -7.72762895e-02 2.90908515e-01 2.40073100e-01 3.47058952e-01 2.82376856e-02 4.22144145e-01 8.18835735e-01 5.94873786e-01 8.38082612e-01 7.93418407e-01 -2.66971201e-01 -1.06999576e+00 -6.21439934e-01 2.52634436e-01 9.08853650e-01 3.83669361e-02 -7.73175776e-01 -6.65054798e-01 -9.74568486e-01 -1.31691054e-01 1.09908795e+00 -5.50039172e-01 -4.50806737e-01 -5.42323768e-01 -7.63694763e-01 6.96379662e-01 6.94007397e-01 1.54225796e-01 -1.52038050e+00 -3.44303429e-01 3.40410441e-01 -3.65271270e-01 -8.09379697e-01 -5.60130894e-01 5.36094010e-01 -8.28693688e-01 -6.06510282e-01 -3.25401366e-01 -7.87860632e-01 1.11275434e+00 -4.02516536e-02 1.06673682e+00 4.13843170e-02 3.72583084e-02 -7.85471320e-01 -4.87699270e-01 -5.54026961e-01 -8.05345178e-01 6.05252624e-01 -9.88990907e-03 -1.50066361e-01 5.31157494e-01 -3.73758227e-01 -3.25255603e-01 -1.98140159e-01 -1.16678095e+00 4.10049677e-01 1.03049135e+00 8.27196479e-01 5.97519338e-01 4.14442301e-01 9.79922354e-01 -1.32989693e+00 1.37930453e+00 -6.04190707e-01 -2.61391550e-01 1.97291881e-01 -1.01680791e+00 6.64406240e-01 1.10348403e+00 -3.14764738e-01 -1.08030915e+00 1.11423761e-01 -9.69966874e-02 6.05329238e-02 -1.40437424e-01 9.04569983e-01 -3.50197732e-01 9.53512728e-01 7.96243250e-01 3.76535177e-01 -3.75578344e-01 -5.71838677e-01 4.59000111e-01 1.09002411e+00 6.17746115e-01 -4.10244823e-01 8.35166752e-01 -3.02924424e-01 -3.10697794e-01 -3.33475083e-01 -7.43433654e-01 -1.03825778e-01 -5.75654030e-01 1.15497366e-01 4.92431045e-01 -7.57785022e-01 -3.70749533e-01 1.01086736e-01 -1.33703852e+00 -5.85108772e-02 -6.70739770e-01 3.51513624e-01 -2.18322754e-01 4.19022478e-02 -5.01466691e-01 -8.24649155e-01 -8.43181014e-01 -8.95429730e-01 8.73387039e-01 3.01888138e-01 -6.18701279e-01 -4.40845400e-01 -9.69844386e-02 -7.79745504e-02 1.32572562e-01 -5.89221381e-02 1.30719841e+00 -1.01146328e+00 -6.12823904e-01 -4.76398319e-01 -2.37583399e-01 1.55603215e-01 3.19432139e-01 -9.84062552e-02 -7.31362760e-01 2.30023876e-01 -3.09081644e-01 -3.78041893e-01 1.01430976e+00 -1.00106141e-02 1.37679768e+00 -6.67839468e-01 -2.71250069e-01 3.13544303e-01 1.16168606e+00 4.58894283e-01 5.69326162e-01 8.34478345e-03 7.85305440e-01 5.55420816e-01 6.46647990e-01 6.15041792e-01 3.99593383e-01 3.33135962e-01 1.70156807e-01 -5.10178171e-02 -1.29429504e-01 -8.17367554e-01 3.49304557e-01 1.11737990e+00 4.54261666e-03 -7.89848208e-01 -6.06523275e-01 5.68926811e-01 -1.85130620e+00 -9.31770921e-01 -2.07877949e-01 2.09316397e+00 1.47858679e+00 5.43198884e-01 -1.61431730e-01 4.22093868e-01 5.34072757e-01 6.07535057e-02 -5.22453368e-01 -2.98036963e-01 -1.09584048e-01 1.63218617e-01 3.41636717e-01 3.21686625e-01 -7.78076947e-01 8.24981391e-01 6.75170565e+00 7.82319784e-01 -8.07954371e-01 -4.59504634e-01 7.14394927e-01 -3.08150738e-01 -6.91824257e-01 -8.59684646e-02 -1.05699110e+00 5.76212883e-01 9.48188007e-01 -6.14409506e-01 3.88336331e-01 7.95389473e-01 3.99441659e-01 1.15633784e-02 -1.33781409e+00 8.18516910e-01 8.83767083e-02 -1.36555862e+00 7.65554786e-01 -1.32722169e-01 7.59599864e-01 -4.72964168e-01 -2.15119630e-01 2.32470289e-01 5.81880689e-01 -1.17948377e+00 6.31968319e-01 5.68508804e-01 9.92919505e-01 -9.30761695e-01 7.82182932e-01 6.00705981e-01 -8.26074660e-01 3.70294461e-03 -5.42415023e-01 -2.24948123e-01 7.23644421e-02 8.47728729e-01 -1.20898521e+00 4.63446975e-01 2.40724266e-01 5.73669434e-01 -6.69883907e-01 7.01667130e-01 -4.37938631e-01 6.28265083e-01 -1.12010360e-01 -4.39879626e-01 -1.99866518e-01 -1.59363057e-02 3.86451632e-01 1.32027137e+00 5.43777227e-01 2.32525557e-01 2.80706454e-02 9.63278472e-01 -6.30778611e-01 2.61139810e-01 -5.72388589e-01 -2.69808263e-01 6.75156713e-01 1.04341531e+00 -6.12104535e-01 -6.00497007e-01 -1.89070597e-01 7.71119356e-01 3.97377670e-01 3.39069664e-01 -3.63012165e-01 -9.54601109e-01 1.43747792e-01 -8.69373456e-02 4.53782193e-02 1.59877390e-01 -8.42163205e-01 -1.13956380e+00 3.62727135e-01 -9.23236191e-01 4.60903168e-01 -7.36700833e-01 -1.03861821e+00 9.46730375e-01 -1.91534624e-01 -1.29420376e+00 -6.53369367e-01 -9.89962816e-02 -5.51070452e-01 1.00231850e+00 -1.10548139e+00 -7.05616415e-01 -2.37683803e-01 1.47317916e-01 9.04171288e-01 -4.21412677e-01 8.18403900e-01 1.58921391e-01 -7.01728940e-01 8.08265805e-01 -2.00515296e-02 3.22355062e-01 4.57422704e-01 -1.47717154e+00 1.00146949e+00 1.06010592e+00 2.57382184e-01 7.69086540e-01 6.15364373e-01 -9.97756839e-01 -1.22072875e+00 -1.32437360e+00 1.43953490e+00 -3.40655953e-01 2.16441721e-01 -5.43392181e-01 -8.96716893e-01 5.27588606e-01 4.49990749e-01 -4.22284096e-01 6.17618442e-01 -9.27839987e-03 -9.27436799e-02 -1.04556307e-01 -8.10747504e-01 8.47860694e-01 1.05911160e+00 -2.43648797e-01 -8.12040746e-01 4.70194891e-02 1.09494770e+00 -3.88864517e-01 -4.84407872e-01 2.12249652e-01 3.21899474e-01 -4.05591071e-01 4.43298250e-01 -7.96556532e-01 9.10377502e-01 -4.07081753e-01 1.74514726e-01 -1.56748629e+00 -2.73484260e-01 -8.94262373e-01 -4.48653042e-01 1.56824231e+00 9.62452412e-01 -4.11786735e-02 8.21828783e-01 6.26122415e-01 -1.20735820e-02 -7.21631169e-01 -4.33355004e-01 -3.65038007e-01 -2.23840475e-01 -5.55865705e-01 1.01627755e+00 6.07471466e-01 2.61084765e-01 9.82972860e-01 -6.28253996e-01 -1.86836779e-01 2.32509911e-01 1.22599207e-01 6.79805100e-01 -1.05572748e+00 -4.89822775e-02 -3.33581626e-01 1.38435259e-01 -1.25315607e+00 -2.63467848e-01 -1.08626926e+00 4.42138761e-01 -1.85251868e+00 2.74121642e-01 -4.69797134e-01 -3.43642443e-01 5.24653673e-01 -7.55895078e-01 -2.60071099e-01 -1.99493207e-02 1.13809757e-01 -4.47677940e-01 5.54346144e-01 1.02206993e+00 -1.21956997e-01 -6.22833908e-01 1.76337823e-01 -1.12355256e+00 2.71846175e-01 7.69366503e-01 -7.08081782e-01 -7.35907257e-01 -5.09764433e-01 5.26138902e-01 9.16860476e-02 -2.07023352e-01 -9.11807716e-01 4.37512726e-01 -3.29246908e-01 5.93127549e-01 -1.03410733e+00 -2.09406018e-01 -7.12351978e-01 -5.27210049e-02 1.94969952e-01 -1.23635435e+00 1.48509040e-01 -1.56200632e-01 4.25817966e-01 -2.32745752e-01 -4.03321028e-01 3.47153038e-01 8.71237144e-02 -2.19761923e-01 2.96866387e-01 -3.80356044e-01 2.38045916e-01 4.86280590e-01 1.35108575e-01 -4.08185333e-01 -3.37147892e-01 -3.44400704e-01 2.06714526e-01 3.22416723e-02 6.51802480e-01 8.61542165e-01 -1.54219902e+00 -1.03896081e+00 6.28371119e-01 2.48407170e-01 4.08929139e-01 -2.84613043e-01 1.26197383e-01 -2.47962222e-01 3.23269516e-01 1.50490761e-01 1.43328169e-02 -1.09635770e+00 5.76403975e-01 -2.46658906e-01 -5.23271918e-01 -5.53192258e-01 6.95602000e-01 -2.16232836e-01 -1.48491981e-05 3.85788918e-01 -7.64324605e-01 -4.98119652e-01 1.69027939e-01 8.39323163e-01 2.29733154e-01 1.48143694e-01 -1.14597343e-01 1.33379191e-01 9.08468850e-03 -4.41993237e-01 -3.34559649e-01 1.26332247e+00 -1.96974557e-02 -6.45075664e-02 4.44132596e-01 9.18932796e-01 1.18245855e-01 -8.05381119e-01 -3.55312228e-01 4.17837590e-01 -1.62562817e-01 -3.04061115e-01 -8.21740150e-01 -8.89482081e-01 7.93857336e-01 7.43971439e-04 3.86505932e-01 1.27458966e+00 -2.14013264e-01 1.03918183e+00 6.30030751e-01 -1.22223191e-01 -1.25575507e+00 6.68614209e-02 7.24633455e-01 9.21877444e-01 -1.00537241e+00 -8.07070136e-02 -3.19563210e-01 -8.08265030e-01 1.08302867e+00 8.58266771e-01 1.90512329e-01 3.43651026e-01 6.12009346e-01 6.37944564e-02 5.36413938e-02 -1.30870020e+00 -1.85809106e-01 2.52549738e-01 4.60662454e-01 5.96310914e-01 -1.09172873e-01 -4.81359541e-01 1.05171990e+00 -6.27268374e-01 2.87953019e-01 7.06051469e-01 8.42406332e-01 -5.61131477e-01 -1.33268106e+00 -1.15215987e-01 1.07033670e+00 -5.21351218e-01 -4.27459121e-01 -6.55989110e-01 2.32071564e-01 7.11222216e-02 1.01012433e+00 -3.18190753e-02 -7.96412766e-01 5.60682952e-01 3.37516397e-01 -2.14550585e-01 -1.01487803e+00 -6.53016567e-01 6.80093595e-04 1.97426990e-01 3.00936326e-02 9.20384377e-02 -3.80310953e-01 -1.55632687e+00 -9.49505568e-02 -3.14392269e-01 4.75731790e-01 3.58320862e-01 8.02889526e-01 6.16684616e-01 8.21260870e-01 7.78281808e-01 -1.48733005e-01 -5.88905990e-01 -1.38941467e+00 -3.59385163e-01 4.79921728e-01 4.11263585e-01 1.86991319e-02 -7.71340951e-02 4.27661657e-01]
[11.767364501953125, 8.889494895935059]
ca4e8351-2b71-42a5-ab24-4c3a332a25ff
paraphrase-generation-via-adversarial
null
null
https://aclanthology.org/2020.wnut-1.32
https://aclanthology.org/2020.wnut-1.32.pdf
Paraphrase Generation via Adversarial Penalizations
Paraphrase generation is an important problem in Natural Language Processing that has been addressed with neural network-based approaches recently. This paper presents an adversarial framework to address the paraphrase generation problem in English. Unlike previous methods, we employ the discriminator output as penalization instead of using policy gradients, and we propose a global discriminator to avoid the Monte-Carlo search. In addition, this work use and compare different settings of input representation. We compare our methods to some baselines in the Quora question pairs dataset. The results show that our framework is competitive against the previous benchmarks.
['Jose Ochoa-Luna', 'Gerson Vizcarra']
null
null
null
null
emnlp-wnut-2020-11
['paraphrase-generation', 'paraphrase-generation']
['computer-code', 'natural-language-processing']
[ 3.31647277e-01 -5.14256358e-02 -2.16625690e-01 -2.64017969e-01 -1.06480706e+00 -7.12961674e-01 8.61388981e-01 -1.25086069e-01 -8.42474699e-01 1.04885554e+00 5.95513582e-01 -1.90739095e-01 3.96530002e-01 -8.72830212e-01 -8.23308408e-01 -3.09197366e-01 6.53068364e-01 2.20436886e-01 1.45273417e-01 -5.51098764e-01 6.42826915e-01 3.42850447e-01 -9.24623072e-01 5.41789889e-01 1.11764348e+00 6.32536411e-01 -9.86067429e-02 8.43380868e-01 -7.71496966e-02 1.04495096e+00 -1.15902567e+00 -7.83224463e-01 4.28353548e-01 -8.73227060e-01 -1.00291765e+00 -5.08480668e-01 7.30223179e-01 -3.59655172e-01 -6.39158309e-01 1.39520621e+00 8.27173173e-01 3.96219164e-01 8.01365852e-01 -1.14564776e+00 -1.27565885e+00 6.37405038e-01 -3.35991085e-01 6.40020072e-01 7.00107038e-01 1.26046196e-01 1.01372278e+00 -1.01951969e+00 5.65576553e-01 1.37923670e+00 5.69525361e-01 9.49870288e-01 -1.02662349e+00 -6.82415605e-01 -7.52175599e-02 6.44020498e-01 -1.02084064e+00 -4.20301527e-01 1.02667952e+00 -1.00946948e-01 1.00371015e+00 2.34606564e-01 3.92472953e-01 1.61312783e+00 4.81136352e-01 1.05557680e+00 1.26356876e+00 -6.08866692e-01 3.18792671e-01 -4.64754105e-02 2.90183630e-02 5.44011950e-01 -7.49705825e-03 2.91115552e-01 -4.00551081e-01 -3.45183015e-01 5.52330136e-01 -2.14857310e-01 -3.14305454e-01 1.85107619e-01 -1.00904608e+00 1.34553945e+00 6.63996935e-01 1.55166104e-01 -2.14642555e-01 4.76193577e-01 5.04798651e-01 7.26486206e-01 4.91743058e-01 9.03297722e-01 1.10312454e-01 -2.12512508e-01 -1.11316550e+00 6.41839802e-01 1.01335716e+00 8.09102893e-01 3.93621057e-01 1.32426515e-01 -7.18996227e-01 1.08510971e+00 -1.67087138e-01 3.73549998e-01 1.03165174e+00 -1.36466527e+00 8.74259889e-01 1.93527147e-01 1.69292822e-01 -9.70912576e-01 1.22791424e-01 -1.07691534e-01 -9.58813250e-01 2.21199483e-01 3.87426317e-01 -1.81302547e-01 -7.76451051e-01 1.67147434e+00 -2.25560084e-01 1.46586493e-01 2.56325096e-01 9.24818695e-01 7.94215679e-01 7.52075315e-01 -3.35407168e-01 2.70091444e-02 9.71219897e-01 -1.64186347e+00 -8.21630836e-01 -3.60778660e-01 -2.04621861e-03 -8.66971076e-01 1.30010641e+00 3.10719252e-01 -1.39876485e+00 -5.79998732e-01 -1.14637113e+00 -2.10003629e-01 -3.40500325e-01 -1.14070453e-01 2.91591436e-01 6.25706732e-01 -9.63604510e-01 1.03537810e+00 -4.46893930e-01 -2.60644585e-01 3.43379915e-01 -6.65713474e-02 -2.79968549e-02 9.40844119e-02 -1.73863685e+00 1.41667211e+00 5.26046693e-01 4.78574587e-03 -8.44338894e-01 -3.27739179e-01 -8.19824338e-01 4.29059267e-02 1.22166723e-01 -9.75382030e-01 1.58276105e+00 -1.08811104e+00 -2.05763531e+00 7.41941869e-01 -1.31765604e-01 -9.10660923e-01 8.47449660e-01 -4.60275203e-01 2.40061469e-02 2.37363622e-01 -3.10991220e-02 6.65894866e-01 8.65914285e-01 -8.20223629e-01 -9.86471251e-02 2.58989602e-01 4.55635160e-01 4.23962176e-01 -2.62730926e-01 8.93203989e-02 -1.80741735e-02 -1.08341718e+00 -3.55803370e-01 -8.61535907e-01 -4.23217326e-01 -4.40308750e-02 -4.04600918e-01 -2.92555034e-01 2.17847198e-01 -7.16699898e-01 1.17173016e+00 -1.63911688e+00 3.23135585e-01 -3.02057117e-01 -1.88005969e-01 4.62181240e-01 -2.46481046e-01 6.39353037e-01 1.00744575e-01 1.56207681e-01 -6.27526283e-01 -3.42491031e-01 1.55559808e-01 2.25622192e-01 -7.13258207e-01 1.87990487e-01 2.15114042e-01 1.23021340e+00 -1.09585905e+00 -6.36405766e-01 -1.16151072e-01 -1.25513956e-01 -7.49265194e-01 4.99560028e-01 -1.89220876e-01 -5.04758917e-02 -3.10195804e-01 4.63238537e-01 6.50543571e-01 3.87598164e-02 -2.53688037e-01 1.32839009e-01 2.93245167e-01 4.26596016e-01 -6.25739276e-01 2.13964343e+00 -6.95903242e-01 5.88135540e-01 -3.08387965e-01 -1.05572879e+00 8.80698204e-01 1.15800224e-01 -2.72848427e-01 -5.84254742e-01 1.01613730e-01 3.16938788e-01 -1.03500327e-02 -4.15894747e-01 8.46557260e-01 -2.91721433e-01 -2.10677251e-01 6.14944637e-01 -4.60384972e-02 -7.88598776e-01 2.44801715e-01 3.17242593e-01 1.11950004e+00 1.94960758e-01 4.18176770e-01 -1.87757790e-01 7.40362525e-01 8.01802278e-02 3.83446813e-01 1.23494911e+00 -5.34064531e-01 1.04933059e+00 5.07175803e-01 -2.40896717e-01 -1.11704123e+00 -1.32554686e+00 1.93092644e-01 7.83962429e-01 2.82495081e-01 -7.45558813e-02 -8.43194008e-01 -9.22776699e-01 1.04171440e-01 9.33405578e-01 -8.32343698e-01 -4.84054625e-01 -9.28264797e-01 -5.26196599e-01 1.06769526e+00 6.60567105e-01 8.81994009e-01 -1.43239796e+00 -3.80791903e-01 3.17674220e-01 -3.84521872e-01 -8.91524017e-01 -8.98937941e-01 -1.05837829e-01 -6.66846931e-01 -8.22911620e-01 -1.04721284e+00 -8.60550821e-01 1.61675438e-01 -4.39073555e-02 1.33981657e+00 -2.18608946e-01 -2.81054787e-02 5.16996300e-03 -2.92843908e-01 -4.24094409e-01 -9.11846697e-01 2.81342804e-01 -2.15176135e-01 -5.18852055e-01 3.19680840e-01 -6.44284070e-01 -8.24190676e-01 -9.39099938e-02 -8.86006892e-01 -7.95879066e-02 4.77525204e-01 1.34033203e+00 1.93381101e-01 -9.78774726e-01 1.08481395e+00 -8.21467757e-01 1.57733178e+00 -6.07229292e-01 -3.40033412e-01 3.95323992e-01 -4.50974733e-01 3.08548540e-01 1.19212496e+00 -4.80826706e-01 -9.67316449e-01 -3.49854589e-01 -2.70604402e-01 -6.20979071e-01 3.65956724e-02 2.70840406e-01 1.09551154e-01 -8.14387724e-02 1.00837839e+00 3.85689765e-01 3.20233591e-02 -3.03115785e-01 5.55503905e-01 7.89466262e-01 5.94882607e-01 -7.00148284e-01 7.58262038e-01 1.75865918e-01 -3.00798327e-01 -2.33083621e-01 -9.81277704e-01 -9.63462889e-02 -1.76682279e-01 1.55460581e-01 5.66431999e-01 -6.46007717e-01 -3.23641002e-01 3.50503206e-01 -1.74582136e+00 -1.22027040e-01 -4.69222069e-01 3.24187696e-01 -7.62629211e-01 7.47664690e-01 -1.08353448e+00 -3.68469417e-01 -8.55732679e-01 -1.07653034e+00 6.15477681e-01 3.48649383e-01 -2.70076036e-01 -8.64257216e-01 6.49214268e-01 3.08615565e-01 7.83982456e-01 2.57903486e-01 5.00571311e-01 -8.76180828e-01 -2.57926226e-01 -1.01840988e-01 -1.33887067e-01 7.17842937e-01 1.71329588e-01 -3.22894335e-01 -8.97216797e-01 -2.40703240e-01 2.36427143e-01 -7.82232881e-01 1.16982853e+00 -4.25506458e-02 1.26858652e+00 -5.53069472e-01 1.61855400e-01 6.43740296e-01 1.14535344e+00 8.11017379e-02 6.57039702e-01 4.88290757e-01 4.12453234e-01 3.85495692e-01 5.20631373e-01 9.01090428e-02 7.42958114e-02 5.60317338e-01 2.66498089e-01 1.99540883e-01 -1.55843914e-01 -5.30466676e-01 5.06508708e-01 8.69539380e-01 2.83489615e-01 -5.38080335e-01 -5.15775204e-01 6.04676664e-01 -1.96748960e+00 -1.37361765e+00 3.07338625e-01 1.64758658e+00 1.15006530e+00 4.88285981e-02 -1.75981745e-02 -7.55945295e-02 8.83783817e-01 5.42237878e-01 -7.06993163e-01 -1.20599747e+00 -2.05724269e-01 7.38690972e-01 3.05061638e-01 6.71094656e-01 -9.88785505e-01 1.17352593e+00 7.58514977e+00 9.81820107e-01 -8.00995648e-01 1.52623713e-01 3.93342614e-01 -2.37994164e-01 -3.65782678e-01 -1.15167335e-01 -2.79766470e-01 6.12354398e-01 7.54453003e-01 -6.84562981e-01 5.86084306e-01 9.17594492e-01 -2.71635270e-03 5.93677117e-03 -1.14118898e+00 8.91172409e-01 5.00031471e-01 -1.35024595e+00 3.21023643e-01 -5.30069292e-01 9.89381075e-01 -4.01921794e-02 1.69283506e-02 4.67928767e-01 6.85828745e-01 -1.22935319e+00 5.44772387e-01 3.96581441e-01 5.19064069e-01 -7.78950214e-01 6.81174397e-01 4.09223795e-01 -6.03967965e-01 -1.49224931e-02 -7.74360955e-01 -1.76946655e-01 3.04445505e-01 2.80568838e-01 -7.37021387e-01 5.10608375e-01 2.27290034e-01 4.60239232e-01 -5.36237061e-01 9.94637966e-01 -7.55711854e-01 6.78834379e-01 -5.30379377e-02 -5.00881672e-01 3.79632354e-01 -1.80318534e-01 6.35198772e-01 1.41440725e+00 1.07262522e-01 -2.18119696e-01 -8.49598050e-02 1.22286510e+00 -5.43660879e-01 1.52881652e-01 -7.13075995e-01 1.74531698e-01 4.80646253e-01 8.95142734e-01 -9.57778618e-02 -5.61834395e-01 -1.65388286e-01 1.58413112e+00 6.30751789e-01 3.07355523e-01 -9.03043866e-01 -9.19539630e-01 4.03414071e-01 -4.12545800e-01 5.50964512e-02 -1.71611644e-02 -2.14388758e-01 -1.55342317e+00 1.31575167e-01 -1.16217470e+00 2.89761901e-01 -6.79716885e-01 -1.71523297e+00 7.24068761e-01 1.12985499e-01 -1.10659492e+00 -7.17323005e-01 -4.08966392e-01 -1.10028529e+00 1.17909026e+00 -1.61118281e+00 -7.77657807e-01 -1.86065391e-01 5.23026407e-01 9.12374258e-01 -3.51560026e-01 7.66780496e-01 6.00597672e-02 -3.01673442e-01 9.66304600e-01 3.18917781e-01 2.31831774e-01 8.92233491e-01 -1.36235321e+00 1.08098483e+00 1.00721395e+00 2.21396118e-01 6.29646122e-01 7.91371644e-01 -4.02134120e-01 -8.39552701e-01 -8.70876849e-01 9.73762333e-01 -4.82630789e-01 8.20014596e-01 -2.73962408e-01 -9.76054907e-01 4.20783550e-01 1.01328158e+00 -2.05798939e-01 3.55640769e-01 -4.56323862e-01 -5.09637356e-01 1.84572563e-01 -1.44929421e+00 7.07741737e-01 1.06667566e+00 -7.26763666e-01 -1.35301471e+00 4.47194964e-01 8.84550095e-01 -5.89296818e-01 -3.02842706e-01 1.63656831e-01 3.83469969e-01 -8.59336913e-01 7.89762616e-01 -9.26822782e-01 1.15401185e+00 2.39378288e-02 3.15413065e-02 -1.81942320e+00 -2.50341088e-01 -7.34323323e-01 -1.50072753e-01 9.12882745e-01 2.64114827e-01 -6.87299848e-01 8.38275731e-01 3.10049355e-01 -1.21490777e-01 -7.05126941e-01 -1.12929690e+00 -9.26751912e-01 9.57289755e-01 1.96992770e-01 4.79787499e-01 8.16263616e-01 1.10788867e-01 4.53196615e-01 -6.26605153e-01 -4.41113979e-01 4.76990730e-01 3.08438271e-01 6.34498179e-01 -3.59092206e-01 -7.18131959e-01 -6.02821171e-01 -1.22999489e-01 -1.29325402e+00 7.26035833e-01 -1.18828964e+00 1.05996706e-01 -1.55556500e+00 1.84294328e-01 2.65050799e-01 -3.15199286e-01 1.82950228e-01 -5.73207676e-01 2.69007415e-01 4.75188106e-01 1.22903287e-01 -2.60752827e-01 8.52697194e-01 1.21994400e+00 -4.70881641e-01 6.24083206e-02 -4.61409315e-02 -5.48055708e-01 5.84527075e-01 1.12202978e+00 -6.38973892e-01 -2.41844654e-01 -6.16125047e-01 7.51856714e-02 7.18368590e-02 4.83081102e-01 -8.92673194e-01 2.92293847e-01 -2.75370657e-01 1.69095129e-01 -3.81043941e-01 4.63024288e-01 -2.97452658e-01 -3.71858567e-01 8.26185048e-01 -9.64876890e-01 5.45192897e-01 5.19030057e-02 5.88273287e-01 -4.76023197e-01 -8.62350702e-01 9.51187253e-01 -4.27360475e-01 -2.13009447e-01 -3.35403718e-02 -3.78324926e-01 6.92242503e-01 8.00551355e-01 5.82435317e-02 -4.68744248e-01 -6.45917654e-01 -4.15086299e-01 2.02263355e-01 4.23082829e-01 6.16223752e-01 8.15946460e-01 -1.47240305e+00 -1.07430160e+00 -3.03906709e-01 -5.52781783e-02 -4.92471725e-01 -1.59705102e-01 8.56830273e-03 -8.64503503e-01 1.22603223e-01 -3.42042863e-01 6.46682829e-02 -9.17790413e-01 6.11339271e-01 6.28123641e-01 -5.49513578e-01 -2.94465989e-01 9.18915808e-01 -1.29119471e-01 -4.22272056e-01 2.91845724e-02 -1.17330156e-01 -2.17943475e-01 -1.49515301e-01 4.74108666e-01 3.69879603e-01 -1.50616854e-01 -1.57000482e-01 -2.31708154e-01 2.33101726e-01 -3.89384717e-01 -4.82618004e-01 8.95385504e-01 2.70486444e-01 1.11360826e-01 2.90218949e-01 1.37171626e+00 4.13710065e-02 -7.70190537e-01 -2.52459019e-01 -2.44017579e-02 -5.74908316e-01 -5.32347143e-01 -7.05520988e-01 -7.70844936e-01 1.00265539e+00 3.34120393e-01 5.89386187e-03 7.76644647e-01 -4.58072513e-01 1.08763742e+00 7.60545313e-01 1.63483739e-01 -1.15600145e+00 4.66091931e-01 8.23677480e-01 1.41600776e+00 -1.27990901e+00 -6.06473070e-03 8.17883387e-02 -8.24455559e-01 1.08042216e+00 8.60111058e-01 -1.00540388e+00 1.94649979e-01 -4.35852967e-02 1.93378136e-01 3.96716416e-01 -8.73692274e-01 2.15410441e-01 1.20174535e-01 3.28956038e-01 3.37055057e-01 -2.68823504e-01 -1.01983023e+00 4.63827223e-01 -6.05513513e-01 6.67836741e-02 7.19967723e-01 1.00422454e+00 -3.02822888e-01 -1.43670380e+00 -2.10416079e-01 3.89526874e-01 -7.71832526e-01 -5.53695917e-01 -7.63652205e-01 5.06624043e-01 -3.71497661e-01 9.34033990e-01 -1.17075004e-01 -1.97981253e-01 5.45792401e-01 1.14063501e-01 7.02016234e-01 -5.13044477e-01 -1.29404533e+00 -7.11690903e-01 5.63715547e-02 -4.58576977e-01 -2.65918761e-01 -1.98796615e-01 -7.71358013e-01 -2.43301600e-01 -2.31481403e-01 4.63668376e-01 3.92776132e-01 6.55554771e-01 2.91802198e-01 1.79847449e-01 8.38758469e-01 -6.97676480e-01 -1.40368879e+00 -1.25746167e+00 -1.18050925e-01 7.53206491e-01 2.23364249e-01 -2.55968958e-01 -7.72312343e-01 -2.19247177e-01]
[11.756755828857422, 9.220707893371582]
99aa3c96-8076-4ad4-af22-f636fdeb080c
discriminative-nearest-neighbor-few-shot
2010.13009
null
https://arxiv.org/abs/2010.13009v1
https://arxiv.org/pdf/2010.13009v1.pdf
Discriminative Nearest Neighbor Few-Shot Intent Detection by Transferring Natural Language Inference
Intent detection is one of the core components of goal-oriented dialog systems, and detecting out-of-scope (OOS) intents is also a practically important skill. Few-shot learning is attracting much attention to mitigate data scarcity, but OOS detection becomes even more challenging. In this paper, we present a simple yet effective approach, discriminative nearest neighbor classification with deep self-attention. Unlike softmax classifiers, we leverage BERT-style pairwise encoding to train a binary classifier that estimates the best matched training example for a user input. We propose to boost the discriminative ability by transferring a natural language inference (NLI) model. Our extensive experiments on a large-scale multi-domain intent detection task show that our method achieves more stable and accurate in-domain and OOS detection accuracy than RoBERTa-based classifiers and embedding-based nearest neighbor approaches. More notably, the NLI transfer enables our 10-shot model to perform competitively with 50-shot or even full-shot classifiers, while we can keep the inference time constant by leveraging a faster embedding retrieval model.
['Caiming Xiong', 'Richard Socher', 'Philip S. Yu', 'Yao Wan', 'Chien-Sheng Wu', 'Wenhao Liu', 'Kazuma Hashimoto', 'Jian-Guo Zhang']
2020-10-25
null
https://aclanthology.org/2020.emnlp-main.411
https://aclanthology.org/2020.emnlp-main.411.pdf
emnlp-2020-11
['goal-oriented-dialog']
['natural-language-processing']
[-3.78632988e-03 2.51066275e-02 -6.24453843e-01 -6.44603550e-01 -9.11443591e-01 -4.14497584e-01 7.53174961e-01 2.08283484e-01 -6.01820588e-01 2.76987523e-01 6.84657216e-01 -3.68077978e-02 3.52027104e-03 -6.15955949e-01 -3.58952135e-02 -2.12387711e-01 1.40097708e-01 6.33492112e-01 2.25490659e-01 -4.69198644e-01 5.54591894e-01 -1.90310960e-03 -1.42462385e+00 2.35796854e-01 8.36138904e-01 1.13564813e+00 2.23787278e-01 7.27619767e-01 -3.67302358e-01 1.01276207e+00 -7.13812411e-01 -3.98285389e-01 1.07956439e-01 -2.40977526e-01 -7.91741908e-01 -2.05698758e-01 4.06556696e-01 -7.60989308e-01 -6.78020060e-01 7.35472977e-01 5.84217489e-01 7.31061518e-01 7.79314995e-01 -1.44002891e+00 -8.68982315e-01 1.93651944e-01 -6.42402396e-02 2.36289531e-01 7.05909789e-01 2.92722255e-01 1.47598279e+00 -7.91964471e-01 4.92058247e-01 1.31924093e+00 5.29173017e-01 7.03643262e-01 -1.04329443e+00 -4.30393040e-01 7.84611404e-02 3.36739033e-01 -1.18520522e+00 -6.70339942e-01 8.29524279e-01 -2.16021776e-01 1.11806214e+00 2.93083757e-01 3.17806214e-01 1.35169518e+00 8.02494735e-02 1.16022193e+00 6.59484565e-01 -4.64560717e-01 5.82180142e-01 3.70857000e-01 6.79427147e-01 8.81301463e-01 -2.26796404e-01 -2.50421703e-01 -7.86910236e-01 -5.23221970e-01 2.86592007e-01 7.02942729e-01 -2.19797403e-01 -1.87501535e-01 -9.14007783e-01 1.42583704e+00 5.17453551e-01 5.30031145e-01 -1.73133224e-01 -6.13988154e-02 4.14163113e-01 3.65232468e-01 5.36132872e-01 6.47603571e-01 -2.14113921e-01 -4.50278550e-01 -8.83012593e-01 1.77161336e-01 1.25474441e+00 1.02251434e+00 7.70663321e-01 -2.66584486e-01 -7.51591206e-01 1.14917588e+00 2.10971206e-01 7.73110613e-02 9.84583259e-01 -9.54945266e-01 2.53727019e-01 8.01126063e-01 1.16627440e-01 -7.76678860e-01 -3.34587693e-01 5.35390228e-02 -4.80781019e-01 -4.11519557e-02 3.23219538e-01 -3.62699665e-02 -6.66628540e-01 1.63755429e+00 2.40334392e-01 -2.81736292e-02 -4.88874018e-02 9.53926742e-01 8.08356702e-01 6.54000878e-01 1.59261584e-01 8.13487824e-03 1.76576293e+00 -1.13158548e+00 -7.96378791e-01 -6.49418235e-01 7.47451425e-01 -4.71202105e-01 1.25604224e+00 -1.25319630e-01 -5.46079874e-01 -3.84588301e-01 -1.04550636e+00 -5.19101262e-01 -6.01345539e-01 -1.40745044e-01 8.58371973e-01 5.12581587e-01 -6.34135425e-01 5.50521553e-01 -4.27679271e-01 -5.48631012e-01 2.88350731e-01 7.19983727e-02 -3.26337367e-01 -2.08190396e-01 -1.50393426e+00 8.95548344e-01 3.15627605e-01 -5.07068694e-01 -6.07489347e-01 -7.39494681e-01 -1.24481618e+00 3.68003458e-01 5.29507399e-01 -5.31432033e-01 1.32382691e+00 -2.68159151e-01 -1.64983857e+00 8.37545812e-01 -4.48435187e-01 -5.01366854e-01 1.91553786e-01 -2.00145394e-01 -1.29005536e-01 3.86774503e-02 2.75838286e-01 7.80044317e-01 9.86036420e-01 -5.14208317e-01 -5.02840281e-01 -4.36463445e-01 3.30973625e-01 2.95288444e-01 -9.65998709e-01 6.05846271e-02 -2.60677487e-01 -3.87009978e-01 -1.55001521e-01 -8.22895348e-01 -1.23345084e-01 1.71368748e-01 -2.81276286e-01 -9.87414777e-01 8.54932547e-01 -4.22800988e-01 1.38329899e+00 -2.27659512e+00 -1.14188850e-01 -5.07283688e-01 4.22599971e-01 2.90811628e-01 -7.65932575e-02 7.27476060e-01 4.88520712e-01 -2.01753333e-01 -1.50405988e-01 -6.86119854e-01 5.12747586e-01 1.49182007e-01 -3.89292270e-01 2.44013160e-01 2.35828936e-01 1.08866954e+00 -9.75456834e-01 -6.01161480e-01 4.57909465e-01 1.79499209e-01 -7.76945233e-01 8.30092490e-01 -3.02735597e-01 -8.60785469e-02 -4.30400044e-01 6.56860650e-01 1.50628567e-01 -4.60483432e-01 3.06041390e-02 -1.28329426e-01 1.24481879e-01 5.05374372e-01 -8.14648688e-01 1.91389167e+00 -9.02168334e-01 7.16653347e-01 -1.01543620e-01 -9.28751230e-01 9.26745832e-01 2.61689395e-01 1.66098431e-01 -7.03857541e-01 5.38330525e-02 -1.13021888e-01 -2.10664570e-01 -4.80808496e-01 6.88682258e-01 -1.54256761e-01 -6.65984988e-01 5.07519543e-01 5.03031433e-01 1.79238934e-02 8.63925293e-02 4.68796045e-01 1.24274719e+00 -2.82954574e-01 8.13550770e-01 -1.56062335e-01 4.10802543e-01 -3.57947908e-02 4.63603407e-01 1.04529750e+00 -7.22209573e-01 4.95094270e-01 4.60522860e-01 -4.83338594e-01 -5.65720856e-01 -8.99202406e-01 -1.71766169e-02 1.92778647e+00 1.65267125e-01 -5.16382694e-01 -4.25334662e-01 -9.63195801e-01 2.54545093e-01 1.08523417e+00 -5.64103901e-01 -5.25363445e-01 -3.54873091e-01 -3.45192313e-01 3.62494230e-01 5.58283389e-01 4.24831629e-01 -1.08237386e+00 -4.57197785e-01 2.56429851e-01 -2.02030063e-01 -1.00425863e+00 -8.86373401e-01 3.36822033e-01 -4.68482435e-01 -7.97337651e-01 -8.73236418e-01 -8.10392618e-01 8.09874982e-02 4.45865482e-01 9.57404137e-01 -1.14176579e-01 -5.20962656e-01 4.13159162e-01 -3.44449043e-01 -2.28350416e-01 -2.74744838e-01 1.70007259e-01 2.61372298e-01 -1.21643446e-01 1.05022335e+00 -4.12987709e-01 -5.85161686e-01 1.58094347e-01 -6.08687222e-01 -4.54384804e-01 4.02200878e-01 1.11269760e+00 -1.74758106e-01 -4.75761682e-01 7.07579017e-01 -6.36618912e-01 1.04872036e+00 -5.55885017e-01 -2.97209501e-01 1.79285854e-01 -5.76617360e-01 1.51999936e-01 7.11224973e-01 -4.88445371e-01 -1.08089685e+00 -1.03252657e-01 -2.54896283e-01 -5.87864995e-01 -4.90998507e-01 2.09792644e-01 -4.04724516e-02 1.36986658e-01 6.01834655e-01 3.29569697e-01 6.59501031e-02 -4.81858552e-01 5.08124352e-01 1.32180560e+00 2.16572523e-01 -1.02306485e-01 4.55218405e-01 4.13676083e-01 -6.71000838e-01 -1.17422485e+00 -1.23917222e+00 -1.19190967e+00 -4.78522092e-01 -2.47839466e-02 9.35422421e-01 -9.03742492e-01 -9.95799661e-01 1.43308192e-01 -1.27818239e+00 -1.57513976e-01 -1.76917180e-01 3.36490095e-01 -4.79532868e-01 3.47333878e-01 -8.10766339e-01 -1.10603023e+00 -5.04812181e-01 -9.94709611e-01 1.37069237e+00 3.49464566e-01 -6.54851556e-01 -9.91061687e-01 3.59541178e-01 6.96745634e-01 6.12868190e-01 -6.26647234e-01 5.89735746e-01 -1.43969405e+00 -2.66036719e-01 -5.91222286e-01 -2.86622822e-01 -3.40361670e-02 2.20175236e-01 -6.65060103e-01 -1.14941919e+00 -1.87641829e-01 2.68886536e-01 -7.44479775e-01 9.96044576e-01 -1.45327020e-02 7.69292414e-01 -4.44635421e-01 -3.20240140e-01 2.40250185e-01 1.08854818e+00 1.89067900e-01 3.45745981e-01 1.07577350e-02 4.87654597e-01 6.10226452e-01 8.77558410e-01 7.45979071e-01 4.33073342e-01 8.69236052e-01 1.67591527e-01 2.60537267e-01 5.41422665e-02 -3.28040034e-01 3.52420539e-01 4.86973524e-01 5.17697334e-01 -3.09018254e-01 -7.13114917e-01 4.89895821e-01 -1.92170727e+00 -1.13279140e+00 5.23356378e-01 1.99611330e+00 8.40038002e-01 3.68343629e-02 1.82757124e-01 -3.06924582e-01 6.37673557e-01 6.53160870e-01 -5.65950513e-01 -3.09058785e-01 5.15979707e-01 -2.45476495e-02 6.16478361e-02 5.89311659e-01 -1.34993994e+00 1.14962661e+00 5.88855600e+00 9.12915051e-01 -7.59549439e-01 3.62455398e-01 4.84108418e-01 -1.57045618e-01 3.68273035e-02 -2.53563583e-01 -1.10232985e+00 6.82519794e-01 9.01273429e-01 -1.46578431e-01 3.15073937e-01 1.49182880e+00 -2.02970102e-01 3.36537734e-02 -1.29257333e+00 8.84481788e-01 4.99869972e-01 -1.38281775e+00 -2.84680605e-01 -2.04845658e-03 4.10267860e-01 -5.00960015e-02 -2.22392201e-01 1.06977117e+00 3.34401667e-01 -6.99743211e-01 1.30610257e-01 3.65118980e-01 5.60586810e-01 -4.91255850e-01 7.11878598e-01 5.87250888e-01 -1.10613132e+00 -2.67757654e-01 -5.36993206e-01 -2.35291496e-01 1.11113355e-01 3.74448091e-01 -9.21638310e-01 -2.13831966e-03 5.02610207e-01 5.88255882e-01 -2.51838535e-01 6.27516627e-01 -4.36824672e-02 2.89872319e-01 -1.97938725e-01 -7.35070646e-01 3.77744883e-01 -6.77238107e-02 5.57901084e-01 1.22728479e+00 -1.22507259e-01 9.33190957e-02 4.43962753e-01 9.73799050e-01 -3.25570434e-01 5.52400993e-03 -9.76440847e-01 -3.42059731e-01 5.26933134e-01 1.36690986e+00 -2.39521772e-01 -5.88602841e-01 -5.50406277e-01 1.44907606e+00 6.87382817e-01 -5.94415180e-02 -7.41483927e-01 -8.01992953e-01 1.08721662e+00 -3.61129224e-01 3.28469932e-01 1.22535392e-03 5.74348383e-02 -1.32255960e+00 -2.23739624e-01 -5.35770833e-01 6.47662580e-01 -4.23703998e-01 -1.68441987e+00 3.46518308e-01 -2.47311950e-01 -1.22004235e+00 -6.35150254e-01 -6.91432178e-01 -1.08044744e+00 6.50977910e-01 -1.35462368e+00 -1.03557384e+00 -3.79631490e-01 3.29480201e-01 1.10146415e+00 -1.41642570e-01 1.06604493e+00 1.20395727e-01 -5.27976274e-01 7.37498879e-01 -3.03141847e-02 3.61653775e-01 1.06168640e+00 -1.30238235e+00 3.68815035e-01 3.65962833e-01 1.83494434e-01 9.83857572e-01 6.13758385e-01 -4.90603596e-01 -1.54222798e+00 -8.94606829e-01 1.09901726e+00 -5.77059507e-01 8.59258771e-01 -5.90387881e-01 -9.79845703e-01 6.57588065e-01 3.13108653e-01 -7.97981694e-02 9.06275094e-01 4.28187400e-01 -5.65523326e-01 1.89377725e-01 -1.17674375e+00 7.23307908e-01 1.07456744e+00 -8.66451442e-01 -1.16971421e+00 7.31072843e-01 1.08324409e+00 -6.99985102e-02 -8.00669491e-01 -4.33945330e-03 5.76499522e-01 -9.88588870e-01 1.15619552e+00 -8.84914815e-01 3.43441814e-01 2.58490711e-01 -2.48361319e-01 -1.05666792e+00 -4.25988734e-01 -6.19406760e-01 -5.26424289e-01 1.16241968e+00 6.19129539e-02 -6.32045686e-01 9.45653200e-01 8.82430255e-01 -1.40974447e-01 -5.42757213e-01 -1.05706215e+00 -8.34527433e-01 -2.20098585e-01 -2.52175957e-01 2.26322860e-01 9.63073015e-01 8.06447744e-01 8.66842389e-01 -6.03501081e-01 -1.58519208e-01 6.03332877e-01 4.77108806e-01 8.08525860e-01 -1.25830829e+00 -4.21309620e-01 -3.25730592e-01 -4.61422563e-01 -1.49437809e+00 5.62059999e-01 -7.97624588e-01 2.61628956e-01 -1.33257949e+00 2.84794331e-01 -5.44058047e-02 -2.35225052e-01 4.15405333e-01 -4.54885066e-01 1.79776505e-01 1.72875881e-01 2.49634966e-01 -1.12955379e+00 9.48436022e-01 6.12874091e-01 -3.34296435e-01 -2.91610330e-01 6.73249662e-02 -6.83222234e-01 7.37684488e-01 5.91490805e-01 -3.18780601e-01 -1.75565019e-01 1.03158683e-01 -3.56005400e-01 2.00869441e-01 1.84270456e-01 -8.72784913e-01 5.86616814e-01 -1.00316517e-01 7.01577812e-02 -4.45488036e-01 8.25756252e-01 -5.69969118e-01 -8.31811786e-01 5.14321089e-01 -6.70336723e-01 -6.33078456e-01 -1.78717554e-01 9.11119878e-01 -9.48696211e-02 -7.14891970e-01 7.01475084e-01 -2.04992458e-01 -1.13972032e+00 2.10625768e-01 -4.75904077e-01 3.29271823e-01 8.30508411e-01 7.76503161e-02 -3.76080573e-01 -6.82066202e-01 -3.18702251e-01 2.53307968e-01 3.03453147e-01 6.51733696e-01 6.64719105e-01 -1.12426734e+00 -3.30056727e-01 2.35329702e-01 7.82742918e-01 -6.09405816e-01 2.83340931e-01 5.26718557e-01 2.39608660e-01 8.02056313e-01 2.14874893e-01 -4.59386021e-01 -1.09576845e+00 6.41257346e-01 9.37310606e-03 -3.42651308e-01 -5.16328394e-01 9.73402143e-01 2.42364079e-01 -7.64575303e-01 5.23251414e-01 1.02142997e-01 -2.32448384e-01 1.95952147e-01 9.62798953e-01 4.25670028e-01 -1.43905610e-01 -2.43615851e-01 -5.22479355e-01 -4.95227240e-02 -4.53990817e-01 9.96199064e-03 1.05447733e+00 -4.27191742e-02 3.59360158e-01 6.06524348e-01 1.57962918e+00 -2.70465732e-01 -1.06994617e+00 -4.48705047e-01 2.14257315e-02 -5.96140683e-01 1.86748058e-01 -5.40147185e-01 -3.26881528e-01 9.26183641e-01 4.02285218e-01 4.51682389e-01 4.56688553e-01 3.04157764e-01 1.13165283e+00 8.55733514e-01 3.06840241e-01 -1.14967501e+00 5.13846576e-01 7.06689060e-01 7.15578198e-01 -1.97711039e+00 -3.16605836e-01 -1.01628013e-01 -8.89057457e-01 9.38586950e-01 7.70117223e-01 -1.49126753e-01 4.90903318e-01 2.50759032e-02 -8.18816870e-02 -2.60584980e-01 -8.79887640e-01 -3.97276402e-01 3.34815145e-01 4.68086839e-01 4.34868872e-01 -2.25003034e-01 -2.19510049e-02 6.56947017e-01 1.34124145e-01 -1.61472023e-01 1.42153636e-01 1.00769663e+00 -7.46981144e-01 -7.52782524e-01 -1.29607124e-02 7.06739008e-01 -8.43232572e-02 -3.63555282e-01 -3.51189256e-01 4.68817770e-01 -6.46454990e-01 1.14674568e+00 2.52978384e-01 -4.73201931e-01 1.30575538e-01 5.74425101e-01 -1.67678501e-02 -9.37266171e-01 -5.19267976e-01 -3.89894694e-01 3.93513106e-02 -8.11431348e-01 1.05869941e-01 -3.22367698e-01 -9.77429688e-01 -3.24883461e-01 -5.53609967e-01 7.70965964e-02 5.11048257e-01 1.09997201e+00 5.85419238e-01 3.65882903e-01 8.06100667e-01 -8.60131323e-01 -1.09380257e+00 -1.16454518e+00 -5.01250982e-01 5.88745832e-01 2.12369666e-01 -7.25679457e-01 -7.74434805e-01 -5.03871560e-01]
[12.188029289245605, 7.585257530212402]
28af6799-5174-48ae-a437-0369be6a7a92
technical-report-assisting-backdoor-federated
2207.12327
null
https://arxiv.org/abs/2207.12327v1
https://arxiv.org/pdf/2207.12327v1.pdf
Technical Report: Assisting Backdoor Federated Learning with Whole Population Knowledge Alignment
Due to the distributed nature of Federated Learning (FL), researchers have uncovered that FL is vulnerable to backdoor attacks, which aim at injecting a sub-task into the FL without corrupting the performance of the main task. Single-shot backdoor attack achieves high accuracy on both the main task and backdoor sub-task when injected at the FL model convergence. However, the early-injected single-shot backdoor attack is ineffective because: (1) the maximum backdoor effectiveness is not reached at injection because of the dilution effect from normal local updates; (2) the backdoor effect decreases quickly as the backdoor will be overwritten by the newcoming normal local updates. In this paper, we strengthen the early-injected single-shot backdoor attack utilizing FL model information leakage. We show that the FL convergence can be expedited if the client trains on a dataset that mimics the distribution and gradients of the whole population. Based on this observation, we proposed a two-phase backdoor attack, which includes a preliminary phase for the subsequent backdoor attack. In the preliminary phase, the attacker-controlled client first launches a whole population distribution inference attack and then trains on a locally crafted dataset that is aligned with both the gradient and inferred distribution. Benefiting from the preliminary phase, the later injected backdoor achieves better effectiveness as the backdoor effect will be less likely to be diluted by the normal model updates. Extensive experiments are conducted on MNIST dataset under various data heterogeneity settings to evaluate the effectiveness of the proposed backdoor attack. Results show that the proposed backdoor outperforms existing backdoor attacks in both success rate and longevity, even when defense mechanisms are in place.
['Tao Shu', 'Xueyang Hu', 'Tian Liu']
2022-07-25
null
null
null
null
['inference-attack']
['adversarial']
[-1.45196348e-01 -2.40527496e-01 -2.08471894e-01 1.83094397e-01 -7.78056681e-01 -9.73253429e-01 6.77400649e-01 -4.81768250e-02 -4.75134581e-01 7.77799547e-01 -1.41471848e-01 -5.14769971e-01 -7.75194913e-02 -8.01668584e-01 -1.05202496e+00 -1.04922080e+00 -1.30463436e-01 3.05127591e-01 2.65286535e-01 3.53692211e-02 1.50039777e-01 4.12233949e-01 -1.10259080e+00 2.58502424e-01 7.38285840e-01 7.29751825e-01 -3.73990908e-02 6.23358190e-01 -1.88294142e-01 8.38134587e-01 -8.91617060e-01 -4.03622061e-01 6.68139875e-01 -3.78074437e-01 -4.44072545e-01 -4.68919009e-01 2.45452568e-01 -5.64845324e-01 -4.32483107e-01 1.35293567e+00 5.16067445e-01 -2.54347604e-02 1.04241930e-01 -1.52834737e+00 -1.08808398e-01 8.97961318e-01 -6.33761644e-01 2.74979800e-01 1.23676270e-01 5.16607881e-01 7.08720684e-01 -5.23587048e-01 3.26503396e-01 1.06904912e+00 6.05701566e-01 5.83685875e-01 -1.20875168e+00 -1.24558794e+00 3.44931334e-01 -2.29067743e-01 -1.03266680e+00 -3.17485034e-01 7.06518590e-01 -2.77497441e-01 5.81080377e-01 3.81964266e-01 3.02660853e-01 1.51348686e+00 5.15221655e-01 7.11181343e-01 1.23026156e+00 -6.94256797e-02 5.68130851e-01 3.84855837e-01 3.68456215e-01 5.57157516e-01 3.81699055e-01 6.68723047e-01 -6.40830815e-01 -9.96774495e-01 2.37347811e-01 8.94881040e-02 -4.13049102e-01 -2.59619176e-01 -7.65433013e-01 9.34243143e-01 3.73128235e-01 2.04666093e-01 -4.20795232e-01 1.94675013e-01 5.53988636e-01 4.01425034e-01 3.50571334e-01 2.58230716e-01 -3.94918025e-01 -1.45416707e-01 -8.49947989e-01 4.51263517e-01 8.80054891e-01 3.30648392e-01 8.38794589e-01 1.12594124e-02 -3.18763226e-01 1.86671883e-01 8.38361010e-02 6.10835969e-01 6.99709654e-01 -5.11037171e-01 7.64303148e-01 4.95715350e-01 5.31426184e-02 -9.18297052e-01 -1.09820865e-01 -8.17391336e-01 -4.56603467e-01 2.49990553e-01 7.23624468e-01 -6.07470036e-01 -5.97491324e-01 2.21169519e+00 7.32844114e-01 3.12051147e-01 2.43549272e-02 8.03308427e-01 1.09210044e-01 5.57290018e-01 -1.01267241e-01 -8.13988522e-02 1.32007563e+00 -6.99518025e-01 -6.61862135e-01 -2.30402246e-01 7.02326298e-01 -4.04133797e-01 1.08029819e+00 3.26172799e-01 -4.90874082e-01 -1.88359648e-01 -1.13866818e+00 7.45128572e-01 -2.11135060e-01 -3.57948184e-01 5.31398118e-01 1.20261550e+00 -5.46119809e-01 2.72436291e-01 -9.16334033e-01 -2.76558958e-02 6.48260355e-01 2.86235899e-01 -3.78337622e-01 -1.33177742e-01 -1.29112613e+00 3.47951919e-01 2.41771370e-01 -6.07549697e-02 -1.56809390e+00 -1.09661663e+00 -3.12881440e-01 9.10526067e-02 6.07398927e-01 -5.89533389e-01 8.67689312e-01 -8.73516679e-01 -1.40611100e+00 2.61456519e-01 1.25535622e-01 -8.23461056e-01 9.87910151e-01 -1.66568652e-01 -1.13077991e-01 -1.01737179e-01 -4.32700962e-02 -1.01882257e-01 1.14706481e+00 -1.21722424e+00 -6.13246977e-01 -7.84050167e-01 9.20526609e-02 8.98322929e-03 -7.57511795e-01 -2.20974401e-01 8.12457968e-03 -4.80303854e-01 -5.16877770e-01 -1.03147924e+00 6.13720827e-02 -5.10364175e-01 -7.03801513e-01 2.09706828e-01 1.26714623e+00 -4.16934878e-01 1.24123478e+00 -2.29316759e+00 -2.92481035e-01 2.42455572e-01 3.42535079e-01 4.86235827e-01 -1.60285324e-01 5.37109315e-01 -4.57618311e-02 1.17522612e-01 -2.02646956e-01 -3.79971772e-01 -7.69858062e-02 -5.22905029e-02 -8.84948015e-01 7.49248743e-01 -4.29009587e-01 5.20548642e-01 -7.87813485e-01 5.45835868e-02 -2.83872753e-01 2.92258829e-01 -6.38943076e-01 3.54632854e-01 -3.01242918e-01 3.12445670e-01 -5.98428071e-01 4.04073864e-01 9.21529889e-01 1.08723275e-01 1.63486540e-01 -6.85217455e-02 1.29949301e-01 9.73990560e-02 -1.04625750e+00 1.07816768e+00 -3.48524451e-01 2.24140689e-01 1.99885771e-01 -6.77647173e-01 7.63509750e-01 3.05224061e-01 2.18038350e-01 -3.05176377e-01 2.82939374e-01 2.26584047e-01 -6.57753274e-02 -8.32800791e-02 1.68291241e-01 -1.09335698e-01 -4.14209776e-02 1.03487086e+00 -1.04423113e-01 5.92368066e-01 -2.85421699e-01 4.25721467e-01 1.42786050e+00 -7.18695670e-02 -1.06468379e-01 -2.15070590e-01 5.78714967e-01 -7.14752898e-02 6.12716436e-01 1.19322932e+00 -3.81511599e-01 -1.07009625e-02 6.87620580e-01 -4.22368526e-01 -4.89724994e-01 -1.21049869e+00 6.63057789e-02 1.07344699e+00 6.02616668e-02 -3.92629474e-01 -9.99360800e-01 -1.42277062e+00 3.30055624e-01 6.78648949e-01 -7.69345641e-01 -6.13930464e-01 -3.10114712e-01 -8.43913734e-01 1.08819818e+00 2.35062540e-02 8.99096787e-01 -7.37393200e-01 -7.06042469e-01 1.38990721e-02 -7.23446757e-02 -8.18953454e-01 -6.88307583e-01 1.49422184e-01 -8.06120515e-01 -1.20429850e+00 -4.27920520e-01 1.45697221e-02 6.46186769e-01 2.88127959e-01 3.25532705e-01 8.65297839e-02 4.27607335e-02 1.58370227e-01 -1.12419352e-01 -1.90759063e-01 -5.74604392e-01 1.73065141e-01 2.76482403e-01 4.93761659e-01 2.43265986e-01 -4.20721024e-01 -3.53390187e-01 2.48723328e-01 -1.00154400e+00 -5.14919996e-01 3.27146977e-01 1.03088713e+00 6.71337917e-02 4.31543559e-01 7.26303816e-01 -1.13698828e+00 9.19641435e-01 -6.93721414e-01 -8.22725713e-01 1.87518746e-01 -5.48709869e-01 2.62073338e-01 1.01818085e+00 -9.24247563e-01 -1.14283168e+00 -6.59184903e-02 6.11918829e-02 -8.28524113e-01 1.07395314e-01 3.15201223e-01 -3.54634047e-01 -7.44569823e-02 8.74984622e-01 3.56524885e-01 1.92148328e-01 -4.81080741e-01 3.26028317e-01 6.88952625e-01 4.15605813e-01 -8.35520208e-01 1.08141160e+00 7.12048650e-01 -2.92211592e-01 -4.84747350e-01 -5.51083386e-01 1.62173167e-01 2.28552744e-01 -2.72263378e-01 4.14383858e-01 -6.48521066e-01 -9.16947603e-01 8.01716447e-01 -9.45035875e-01 -3.26966077e-01 -1.30370378e-01 4.31725115e-01 -4.76409644e-02 4.22087371e-01 -4.08667952e-01 -1.05423558e+00 -5.64317048e-01 -1.24489284e+00 4.53277498e-01 1.57503918e-01 1.20733455e-01 -1.03123260e+00 2.97654182e-01 3.29343200e-01 4.36534464e-01 1.85109511e-01 9.25358295e-01 -1.24421620e+00 -4.69093353e-01 -5.64780474e-01 2.34747976e-01 2.82625645e-01 2.55274892e-01 -1.68799996e-01 -1.11974442e+00 -7.81872094e-01 5.68376064e-01 -3.44889045e-01 6.05677366e-01 5.54930121e-02 8.21926832e-01 -5.41648388e-01 -4.27906662e-01 7.48705268e-01 1.32303715e+00 5.12792841e-02 2.96337694e-01 5.62261641e-01 4.72473621e-01 3.85859370e-01 3.88640374e-01 5.33287644e-01 -3.73374596e-02 3.39295536e-01 6.75661564e-01 9.72470865e-02 2.24469334e-01 -6.67193592e-01 8.76680315e-01 1.78595260e-01 5.39745629e-01 -9.97025818e-02 -7.02938974e-01 3.02234113e-01 -1.69590580e+00 -9.09102082e-01 2.34734133e-01 2.75628304e+00 8.46994221e-01 3.28692168e-01 3.97707403e-01 -3.13178301e-02 8.23670506e-01 2.11502925e-01 -8.37949812e-01 -2.28443831e-01 -3.01594473e-02 -6.83554262e-02 7.29838490e-01 5.26573956e-01 -6.62092566e-01 8.08226585e-01 5.23171854e+00 1.03810155e+00 -1.29501796e+00 5.00901401e-01 6.02340341e-01 -2.98865974e-01 -3.04106206e-01 2.98675328e-01 -1.01056802e+00 7.18718529e-01 8.32034528e-01 -4.10445958e-01 5.58719039e-01 9.18921530e-01 -1.52490571e-01 -6.46396577e-02 -1.06554675e+00 5.66754460e-01 -1.00224808e-01 -1.16572928e+00 -8.21472555e-02 4.83813316e-01 4.50530291e-01 1.59924045e-01 3.18584353e-01 4.87790823e-01 5.63325524e-01 -5.96746445e-01 7.47750759e-01 8.69773105e-02 4.82427388e-01 -1.13093460e+00 6.21647894e-01 9.26522553e-01 -9.36326265e-01 -5.63732743e-01 -1.64153382e-01 1.63321689e-01 -2.14178100e-01 7.48249173e-01 -7.14359581e-01 5.78444123e-01 5.26098013e-01 -1.57670334e-01 -3.82592350e-01 5.52844703e-01 -1.44800082e-01 8.43615115e-01 -5.55963993e-01 1.56391308e-01 2.20443740e-01 -1.33290634e-01 1.05175996e+00 7.85609126e-01 1.95856905e-03 -4.56969768e-01 9.75119919e-02 1.11340451e+00 -2.02100992e-01 -1.68281361e-01 -8.38158727e-01 -7.68261999e-02 7.95657158e-01 1.19783688e+00 -3.79101753e-01 -2.14095488e-01 8.19517225e-02 7.21443474e-01 4.08456743e-01 4.62098330e-01 -1.08135307e+00 -4.16597873e-01 7.51782894e-01 -3.49202827e-02 1.21383332e-01 1.14522949e-01 -7.49295875e-02 -1.06695223e+00 -1.46262959e-01 -1.28003347e+00 5.61264992e-01 -5.53275757e-02 -1.39854431e+00 7.68123865e-01 -7.89998248e-02 -7.17976987e-01 -2.43160233e-01 4.48171189e-03 -8.04851413e-01 8.51341307e-01 -1.06377172e+00 -8.96523356e-01 4.98040579e-02 1.00258040e+00 1.80666775e-01 -4.46305335e-01 6.63609385e-01 1.65243059e-01 -1.05973196e+00 1.19523740e+00 5.81525452e-02 1.05192490e-01 5.78108311e-01 -7.79510140e-01 9.80183482e-02 1.17793226e+00 1.53831109e-01 8.93130362e-01 7.34540403e-01 -9.57542479e-01 -1.54303813e+00 -1.04573298e+00 3.06027859e-01 -3.76662463e-01 6.84629917e-01 -7.37039447e-01 -8.70973289e-01 7.52536774e-01 -1.77756786e-01 1.22177571e-01 4.80925351e-01 -1.01626195e-01 -6.18171096e-01 -4.12988693e-01 -1.59386635e+00 5.53727329e-01 4.10225928e-01 -6.11523151e-01 -3.09336543e-01 3.70876014e-01 6.50043964e-01 -2.15339988e-01 -4.24148679e-01 6.22238070e-02 4.56610054e-01 -9.74751711e-01 6.95042014e-01 -6.82186544e-01 -1.30975544e-01 -6.31964505e-02 -1.55326739e-01 -1.25404251e+00 1.27015591e-01 -1.11945915e+00 -5.01734138e-01 1.39643013e+00 2.01196566e-01 -1.38448584e+00 7.86208808e-01 4.37569678e-01 5.03651977e-01 -5.14360547e-01 -1.16361749e+00 -1.01022744e+00 1.53882325e-01 -4.05714244e-01 8.52213264e-01 8.41333687e-01 -2.10774228e-01 3.18821669e-02 -4.67734814e-01 4.68523383e-01 1.10080814e+00 -1.03770077e-01 1.14019752e+00 -6.53495669e-01 -6.41989112e-01 -6.34652749e-02 1.68080971e-01 -7.49876142e-01 1.63533822e-01 -8.46353412e-01 -1.50416687e-01 -4.80168015e-01 2.60367632e-01 -4.37651336e-01 -4.50410724e-01 6.23921216e-01 -3.98987085e-01 -1.30576327e-01 5.11903703e-01 4.06560600e-01 -2.58272719e-02 5.04136324e-01 7.95980275e-01 4.52930927e-02 -3.18461865e-01 2.32489675e-01 -6.54615700e-01 2.94105440e-01 7.46402025e-01 -1.11246657e+00 -6.26365840e-01 9.77403764e-03 -3.32060158e-02 7.40128234e-02 2.91313767e-01 -8.43638062e-01 3.67432654e-01 4.81875837e-02 3.79247069e-02 -3.34059536e-01 1.39410503e-03 -7.11762428e-01 2.76352108e-01 9.30754483e-01 -1.92864537e-01 -4.52358052e-02 2.26453021e-01 8.38772953e-01 3.08813810e-01 -1.05255663e-01 7.45730817e-01 1.02674723e-01 1.33609936e-01 2.73182690e-01 -3.99312645e-01 3.91109176e-02 1.16022861e+00 -1.04776829e-01 -6.19233966e-01 -2.13433623e-01 -2.66220808e-01 1.89709052e-01 3.95222127e-01 2.70711958e-01 3.68909150e-01 -9.95572090e-01 -5.70816278e-01 5.35599947e-01 -3.09703648e-01 -4.04819310e-01 2.07112879e-01 9.42765117e-01 -9.88743156e-02 1.58417910e-01 -2.58665886e-02 -5.52823722e-01 -1.12704694e+00 8.18998218e-01 6.93764806e-01 -6.71684444e-01 -7.06770658e-01 7.25785136e-01 3.74216229e-01 -5.94468117e-01 3.44295174e-01 2.33192384e-01 3.46297652e-01 -1.84358984e-01 7.40197778e-01 3.29484731e-01 1.10663414e-01 -2.01293677e-01 -5.22818446e-01 1.20860897e-02 -4.32516307e-01 -2.08902881e-01 1.03379750e+00 1.90713152e-01 3.71394008e-02 2.01797351e-01 1.17249799e+00 3.98602009e-01 -1.40895557e+00 -2.00588986e-01 -3.35121274e-01 -8.39566946e-01 2.52395004e-01 -8.27211618e-01 -1.32064903e+00 5.14215648e-01 6.92305565e-01 1.45885408e-01 9.50546026e-01 -4.45194960e-01 1.00447893e+00 1.69805124e-01 5.64459264e-01 -7.15460777e-01 1.09941997e-01 1.72586232e-01 3.45077664e-01 -5.60636997e-01 -7.46217147e-02 6.69412762e-02 -4.98192668e-01 7.32457221e-01 6.63483262e-01 -1.41125530e-01 6.22267902e-01 5.81778228e-01 -5.07530347e-02 -1.57195851e-01 -8.48019719e-01 5.01486003e-01 -3.67732733e-01 4.84197199e-01 -4.09453630e-01 -1.16757385e-01 -6.57337531e-02 9.64421511e-01 -7.84159750e-02 -7.84042478e-03 4.20780063e-01 9.66770351e-01 -2.49481693e-01 -1.09050500e+00 -8.40061724e-01 1.65549308e-01 -6.32656872e-01 1.60461485e-01 -4.02814060e-01 6.95603251e-01 -2.82252785e-02 8.72108579e-01 -1.13159955e-01 -6.06024742e-01 1.08450606e-01 1.43896982e-01 1.81163758e-01 -2.14277998e-01 -1.15871513e+00 -1.18201979e-01 -1.46317646e-01 -6.87151849e-01 4.57978100e-01 -3.40917706e-01 -1.14657795e+00 -5.27930558e-01 -5.97398698e-01 4.58897084e-01 8.07487845e-01 9.38435435e-01 4.20034260e-01 2.67623425e-01 1.15367734e+00 -3.94983888e-01 -1.37833858e+00 -7.47105122e-01 -6.23434603e-01 3.12622070e-01 4.03192818e-01 -5.84126651e-01 -1.04234970e+00 -5.42488098e-01]
[5.739367485046387, 7.348329067230225]
f2c6dc35-a269-4d2f-a471-b8ae63827a1a
density-map-distillation-for-incremental
2304.05255
null
https://arxiv.org/abs/2304.05255v1
https://arxiv.org/pdf/2304.05255v1.pdf
Density Map Distillation for Incremental Object Counting
We investigate the problem of incremental learning for object counting, where a method must learn to count a variety of object classes from a sequence of datasets. A na\"ive approach to incremental object counting would suffer from catastrophic forgetting, where it would suffer from a dramatic performance drop on previous tasks. In this paper, we propose a new exemplar-free functional regularization method, called Density Map Distillation (DMD). During training, we introduce a new counter head for each task and introduce a distillation loss to prevent forgetting of previous tasks. Additionally, we introduce a cross-task adaptor that projects the features of the current backbone to the previous backbone. This projector allows for the learning of new features while the backbone retains the relevant features for previous tasks. Finally, we set up experiments of incremental learning for counting new objects. Results confirm that our method greatly reduces catastrophic forgetting and outperforms existing methods.
['Joost Van de Weijer', 'Chenshen Wu']
2023-04-11
null
null
null
null
['object-counting']
['computer-vision']
[ 2.79156119e-01 -3.27011019e-01 -5.28005362e-02 -3.21322709e-01 -5.64624965e-01 -2.06521153e-01 5.24029970e-01 2.57538110e-01 -9.89458144e-01 1.18005097e+00 4.60327305e-02 -7.75446147e-02 1.23334512e-01 -6.37513220e-01 -1.02986395e+00 -6.07712626e-01 -5.06418087e-02 5.25992513e-01 5.45846045e-01 4.22013640e-01 5.69775462e-01 3.61915261e-01 -1.59006560e+00 2.64863908e-01 7.77374744e-01 6.34099901e-01 5.44174016e-01 7.49307930e-01 -4.11422819e-01 1.28272712e+00 -6.11235559e-01 -5.13672113e-01 7.29677454e-02 -2.42283568e-01 -7.73836672e-01 -9.16101933e-02 8.69858146e-01 -6.35486424e-01 -3.47233027e-01 9.18322027e-01 3.92829806e-01 3.54715109e-01 7.25635469e-01 -1.25030053e+00 -7.71830022e-01 5.92988312e-01 -7.69588828e-01 5.15145838e-01 -8.41292888e-02 1.29069671e-01 7.22555518e-01 -1.20267105e+00 4.82269108e-01 1.11627948e+00 9.16565597e-01 7.48821378e-01 -1.31275761e+00 -7.97651649e-01 6.09923422e-01 -2.55744029e-02 -1.12594628e+00 -3.13761234e-01 5.56726575e-01 -4.28714961e-01 8.53132486e-01 -1.94206357e-01 5.83614707e-01 1.00181043e+00 7.63606429e-02 9.56082582e-01 8.27584147e-01 -4.65436429e-01 4.08932090e-01 -2.63695642e-02 3.60956758e-01 1.14285064e+00 6.70223832e-01 -1.98619068e-01 -6.66561067e-01 -2.78291494e-01 7.36663997e-01 4.00624841e-01 2.20661044e-01 -5.68396568e-01 -8.94514859e-01 8.21572423e-01 3.14841866e-01 2.08977193e-01 2.23962571e-02 5.46321213e-01 4.72143859e-01 1.73213512e-01 6.00744784e-01 2.56825775e-01 -5.61624527e-01 -3.53351682e-02 -8.43160570e-01 3.37983519e-01 6.83293521e-01 9.58865106e-01 8.13333154e-01 5.44896796e-02 -7.51209795e-01 7.52108514e-01 -2.28877977e-01 2.88365841e-01 2.93297976e-01 -1.22194922e+00 5.69355965e-01 5.81892550e-01 2.73531616e-01 -4.46276993e-01 -4.75645334e-01 -3.13147515e-01 -7.19860256e-01 1.97989449e-01 3.68096471e-01 -1.41084660e-02 -9.88957226e-01 2.08918142e+00 2.64173061e-01 5.33922434e-01 -5.72052062e-01 3.18603754e-01 1.80214018e-01 3.33471209e-01 3.63813818e-01 -3.09075952e-01 9.10250247e-01 -1.01150775e+00 -3.70614499e-01 -3.71715218e-01 6.92541659e-01 -2.50667661e-01 1.46814048e+00 4.26139742e-01 -1.27664328e+00 -6.37318492e-01 -9.36509311e-01 -4.00700182e-01 -2.47224227e-01 1.66625664e-01 8.82271171e-01 2.98404783e-01 -8.01579118e-01 8.48670959e-01 -1.10760427e+00 1.13280006e-01 8.87817442e-01 3.50858450e-01 -2.16002408e-02 -2.56093353e-01 -6.11632049e-01 9.73999262e-01 5.23276567e-01 -4.11737293e-01 -1.11861634e+00 -9.55627918e-01 -7.17902899e-01 2.41986632e-01 3.55278909e-01 -9.74448442e-01 1.35403693e+00 -4.78586882e-01 -1.13202250e+00 7.64169693e-01 -3.62415105e-01 -5.78617454e-01 6.58173680e-01 -7.02998161e-01 2.33715221e-01 -2.40477577e-01 4.30535764e-01 7.38834739e-01 1.05575132e+00 -1.20505774e+00 -7.58980930e-01 -2.14282662e-01 -1.87106635e-02 1.40714407e-01 -6.70777202e-01 -4.09564793e-01 -3.70938122e-01 -6.67500496e-01 -2.62588918e-01 -7.52587736e-01 -1.59268737e-01 3.14915895e-01 1.56951770e-02 -4.37318057e-01 6.54238641e-01 -4.37449455e-01 1.32192636e+00 -2.19425154e+00 1.82295814e-01 -4.71256167e-01 5.25543749e-01 2.43492499e-01 -1.89859778e-01 -1.52282000e-01 1.24072269e-01 -1.37274802e-01 -5.39341688e-01 -8.18288803e-01 -3.54895234e-01 3.50967854e-01 -2.43533015e-01 2.75759250e-01 1.24084927e-01 7.80662119e-01 -1.24056816e+00 -5.44914961e-01 2.07596198e-02 3.40646505e-01 -8.43412161e-01 5.50927334e-02 -4.55354512e-01 2.92630255e-01 -5.92508838e-02 4.38732445e-01 7.44244814e-01 -5.08392155e-01 -1.23895913e-01 2.59479225e-01 -9.22917202e-02 1.30084485e-01 -9.12374198e-01 1.84828138e+00 -6.74769938e-01 1.60357818e-01 -3.51554304e-01 -7.67884314e-01 5.88126540e-01 -1.74814865e-01 2.62313038e-01 -5.77760279e-01 3.40242162e-02 1.60319984e-01 -3.12162399e-01 -1.74157456e-01 5.81057489e-01 -2.81694919e-01 -2.50563979e-01 4.69271868e-01 3.24920982e-01 -1.09844252e-01 5.12805045e-01 3.82092148e-01 1.31067681e+00 2.91804224e-03 2.49523863e-01 -1.42199732e-02 4.29293960e-01 -1.19338639e-01 7.15188622e-01 1.30637717e+00 -2.04847887e-01 4.40841913e-01 4.13967162e-01 -7.77492225e-01 -1.24286771e+00 -1.34782481e+00 1.18077472e-02 1.55872548e+00 -2.58981973e-01 -9.75817218e-02 -4.36258286e-01 -1.01430416e+00 5.38108885e-01 9.45844769e-01 -7.10722864e-01 -5.54935515e-01 -1.04048979e+00 -8.19723666e-01 2.92950302e-01 8.90805960e-01 5.22848487e-01 -9.56913114e-01 -5.88444114e-01 2.70240873e-01 -1.05356246e-01 -8.63769889e-01 -7.87375748e-01 5.75611472e-01 -1.37202752e+00 -9.98072565e-01 -7.19276488e-01 -8.45623374e-01 7.54613340e-01 4.67704087e-01 1.37654424e+00 1.60934895e-01 -6.11933887e-01 3.02999675e-01 9.34650898e-02 -4.83960062e-01 -2.03734681e-01 4.54501003e-01 2.43225247e-01 -4.79541808e-01 4.80805993e-01 -5.66276789e-01 -4.28396165e-01 -1.64922178e-01 -6.79751396e-01 -2.23088950e-01 5.04287481e-01 9.93484020e-01 4.98713911e-01 -1.34378120e-01 6.68695569e-01 -1.18018830e+00 6.51768088e-01 -4.65048850e-01 -7.81074584e-01 2.06074998e-01 -6.06802166e-01 4.56242144e-01 7.51623154e-01 -8.78407359e-01 -1.12850785e+00 2.20847428e-01 2.77758121e-01 -5.13077617e-01 5.24849057e-01 -7.81684648e-03 2.17534900e-01 7.73686692e-02 6.54327214e-01 2.36702010e-01 -3.05133700e-01 -6.21469378e-01 4.99359131e-01 5.30060520e-03 6.27554119e-01 -9.09027457e-01 6.89960003e-01 6.27872825e-01 -2.87749507e-02 -4.20283765e-01 -1.48961544e+00 -4.08890218e-01 -8.54640186e-01 1.07304111e-01 6.68295503e-01 -9.45511281e-01 -7.71714091e-01 5.40676892e-01 -1.38354146e+00 -4.83195573e-01 -8.39545131e-01 4.00987267e-01 -6.23265982e-01 2.32982844e-01 -7.49377251e-01 -7.84148693e-01 -2.99529672e-01 -4.76276278e-01 1.01909339e+00 1.11527413e-01 -9.83986780e-02 -7.97285020e-01 5.30148983e-01 -2.33781058e-02 3.57612073e-01 -2.29152098e-01 1.14539266e+00 -3.74865592e-01 -5.72022498e-01 4.60688537e-03 -4.66702253e-01 6.30092680e-01 1.73564050e-02 -5.13394892e-01 -8.01881194e-01 -6.58668935e-01 1.15256853e-01 -6.89964890e-01 1.65738690e+00 2.52248049e-01 1.48526752e+00 -2.21025199e-01 -3.77191573e-01 6.24861896e-01 1.44167054e+00 9.64461640e-02 3.61373872e-01 1.85723826e-01 7.53582716e-01 -1.50429970e-02 2.73657143e-01 6.29472613e-01 2.38444790e-01 8.36267248e-02 2.22622424e-01 3.60239089e-01 -3.24374408e-01 -5.74145019e-01 1.76602766e-01 7.79709876e-01 -8.04340541e-02 2.86155362e-02 -7.66760051e-01 7.20504999e-01 -1.79343569e+00 -1.00050187e+00 4.50980783e-01 2.37303925e+00 1.01731825e+00 4.56916690e-01 8.91355723e-02 3.33328359e-02 6.45704925e-01 2.48873718e-02 -9.62368131e-01 -1.72215000e-01 2.42080688e-01 4.87534136e-01 4.75535035e-01 5.19579709e-01 -1.15563214e+00 1.03133368e+00 6.91777658e+00 6.26180887e-01 -5.03017306e-01 4.76367742e-01 5.30334413e-01 -4.04129893e-01 1.14072695e-01 -3.54032069e-02 -1.32147217e+00 5.76425314e-01 8.34968626e-01 -2.36922562e-01 3.00259799e-01 1.13229060e+00 -4.10200804e-01 -3.56339246e-01 -1.28329897e+00 7.90393770e-01 2.25062221e-01 -1.20952237e+00 2.83838987e-01 -4.14490074e-01 8.47095609e-01 -7.61727542e-02 1.38422653e-01 9.27686691e-01 9.25881028e-01 -7.03209341e-01 5.33876181e-01 7.34868586e-01 8.53887916e-01 -7.29630589e-01 4.36460346e-01 5.89875579e-01 -1.19022524e+00 -3.78433675e-01 -7.19436407e-01 -3.18599820e-01 1.62485376e-01 7.44901776e-01 -7.08583057e-01 -2.06641316e-01 4.84243691e-01 4.29215819e-01 -7.57808745e-01 1.33166862e+00 -1.89605623e-01 3.82261693e-01 -2.52116591e-01 -2.73530707e-02 -7.32723624e-02 1.92652300e-01 2.41500571e-01 1.30849576e+00 3.34704667e-01 -1.37000009e-01 2.68009007e-01 8.06681573e-01 -6.12109900e-01 -3.16256166e-01 -7.81061649e-01 2.23481953e-01 5.85736513e-01 7.53009200e-01 -5.59374750e-01 -5.62272251e-01 -4.21550453e-01 1.38521171e+00 1.12098122e+00 3.36839780e-02 -9.40497279e-01 -2.79571950e-01 3.08610648e-01 2.26288274e-01 4.31131929e-01 -4.45413649e-01 -3.06536883e-01 -1.29124498e+00 9.31691676e-02 -2.14366332e-01 5.67593873e-01 -5.14916539e-01 -1.56200302e+00 3.24732326e-02 7.33573437e-02 -7.44394302e-01 1.50326252e-01 -5.47974467e-01 -6.37422025e-01 5.64288616e-01 -1.42936790e+00 -7.31583655e-01 -2.39447236e-01 4.35131758e-01 7.88242817e-01 -1.95038617e-01 4.89283651e-01 4.07483339e-01 -3.45887303e-01 6.52906120e-01 1.41412150e-02 4.79140952e-02 6.91036701e-01 -1.20070720e+00 4.77074832e-01 6.22517347e-01 6.56994432e-02 7.56359637e-01 2.43564233e-01 -9.02286351e-01 -8.68302524e-01 -1.33598197e+00 9.89217103e-01 -7.79985487e-01 4.80722249e-01 -6.39945507e-01 -1.02958536e+00 9.66071010e-01 -3.57204109e-01 3.10813397e-01 3.55660170e-01 2.65878916e-01 -6.62032902e-01 -5.40107191e-02 -1.12787163e+00 4.02281970e-01 1.39118838e+00 -4.22632277e-01 -7.40714848e-01 4.50463325e-01 9.27527726e-01 -9.61159021e-02 -1.30452141e-01 1.55761272e-01 6.21054530e-01 -6.75809979e-01 1.04745817e+00 -7.74082780e-01 3.27721536e-01 4.96707531e-03 9.19257849e-02 -9.81444001e-01 -7.00486362e-01 -1.88075975e-01 -6.84948087e-01 1.02986085e+00 2.37932384e-01 -3.00208151e-01 1.10565019e+00 2.80187309e-01 6.87739328e-02 -4.01539475e-01 -1.01616311e+00 -1.16717017e+00 5.41064620e-01 -2.04368263e-01 1.77859679e-01 7.09691048e-01 -2.80679017e-01 7.86975145e-01 -5.94378889e-01 -2.21957907e-01 9.67606962e-01 -1.94189891e-01 6.33778632e-01 -1.40392017e+00 -4.21711504e-01 -2.52597243e-01 4.53768624e-03 -1.21861625e+00 1.53713569e-01 -1.14824355e+00 -5.06862532e-03 -1.32949102e+00 9.03953612e-01 -4.78255063e-01 -4.45861846e-01 5.07570922e-01 -6.10835135e-01 -1.25444025e-01 2.79825568e-01 2.82563597e-01 -9.36189711e-01 6.20169222e-01 1.16186059e+00 -2.22306699e-01 -3.56132418e-01 2.39138752e-01 -5.21186173e-01 8.71887565e-01 7.90586412e-01 -1.11609542e+00 -3.85426134e-01 -8.38922977e-01 3.64158303e-01 -1.11019030e-01 2.48131007e-01 -1.24663210e+00 4.62305784e-01 3.19898427e-02 5.62048614e-01 -7.43071198e-01 2.72801787e-01 -4.93259013e-01 -4.80940372e-01 7.48438656e-01 -2.56922454e-01 1.56801656e-01 1.97931513e-01 9.22224700e-01 2.62050867e-01 -5.02073228e-01 1.04584789e+00 -6.03448033e-01 -3.13691139e-01 3.52393031e-01 -6.26980066e-02 4.52143013e-01 8.40651214e-01 3.32555472e-04 -2.33469233e-01 3.26282591e-01 -7.46567905e-01 3.21665943e-01 3.56916457e-01 8.07827264e-02 6.16016209e-01 -1.27180731e+00 -7.29331553e-01 8.66283625e-02 -1.26052842e-01 1.11755066e-01 8.21698904e-02 3.63190442e-01 -2.43035302e-01 2.00408190e-01 -1.69147313e-01 -3.43665302e-01 -1.00614464e+00 9.43299830e-01 -3.43377106e-02 -7.33129859e-01 -7.47183383e-01 1.32404447e+00 2.89206445e-01 -4.94909137e-01 5.06478488e-01 -3.85768056e-01 8.07358101e-02 1.47899538e-01 5.97567141e-01 7.90688276e-01 -1.27264529e-01 1.84230000e-01 -2.39326105e-01 3.53890777e-01 -7.52908707e-01 -2.83002518e-02 1.44440424e+00 5.68827391e-02 6.12116493e-02 9.52170849e-01 1.15149653e+00 -1.94228202e-01 -1.65703595e+00 -4.29680437e-01 2.73084998e-01 -6.10690892e-01 -2.88354933e-01 -6.63655996e-01 -7.54884481e-01 8.79050791e-01 5.75865805e-01 -4.95323926e-01 7.27115989e-01 -6.56301305e-02 7.90006220e-01 8.45071256e-01 6.62537754e-01 -1.21123886e+00 7.53598750e-01 9.33552265e-01 4.83807296e-01 -1.25807619e+00 3.74843836e-01 1.68853197e-02 -3.58655691e-01 8.09586048e-01 9.02442038e-01 -3.23904097e-01 5.35806417e-01 5.58265865e-01 -8.58162344e-01 -1.04246981e-01 -9.48359847e-01 -5.73485792e-02 -2.80787349e-01 4.46041971e-01 2.54515350e-01 -1.91862658e-01 -2.42291480e-01 4.99699861e-01 9.21295956e-02 5.18157780e-01 4.52328742e-01 1.31273818e+00 -8.62337530e-01 -7.61935711e-01 -1.28917947e-01 7.30300546e-01 -3.17137033e-01 -2.93187678e-01 -2.06314623e-02 6.58758581e-01 4.60882455e-01 1.73398152e-01 2.61309355e-01 -1.23676229e-02 3.05211574e-01 5.50043643e-01 9.15931642e-01 -1.09860003e+00 -2.98102945e-01 -5.66711366e-01 -2.92363137e-01 -1.96412832e-01 -6.66786209e-02 -6.83968425e-01 -1.03592074e+00 -4.07761782e-01 -4.74021345e-01 -7.77847245e-02 2.76763737e-01 6.95875108e-01 7.31103346e-02 6.68768466e-01 3.87009084e-01 -7.45503604e-01 -8.71206164e-01 -8.53971899e-01 -5.91168940e-01 2.14399025e-01 3.33194077e-01 -1.01683903e+00 -5.68032861e-01 7.71172047e-02]
[9.751349449157715, 3.279282569885254]
2059b2fb-2bee-4609-b863-b774e79b355c
high-fidelity-point-cloud-completion-with-low
2112.11271
null
https://arxiv.org/abs/2112.11271v2
https://arxiv.org/pdf/2112.11271v2.pdf
High-Fidelity Point Cloud Completion with Low-Resolution Recovery and Noise-Aware Upsampling
Completing an unordered partial point cloud is a challenging task. Existing approaches that rely on decoding a latent feature to recover the complete shape, often lead to the completed point cloud being over-smoothing, losing details, and noisy. Instead of decoding a whole shape, we propose to decode and refine a low-resolution (low-res) point cloud first, and then performs a patch-wise noise-aware upsampling rather than interpolating the whole sparse point cloud at once, which tends to lose details. Regarding the possibility of lacking details of the initially decoded low-res point cloud, we propose an iterative refinement to recover the geometric details and a symmetrization process to preserve the trustworthy information from the input partial point cloud. After obtaining a sparse and complete point cloud, we propose a patch-wise upsampling strategy. Patch-based upsampling allows to better recover fine details unlike decoding a whole shape, however, the existing upsampling methods are not applicable to completion task due to the data discrepancy (i.e., input sparse data here is not from ground-truth). Therefore, we propose a patch extraction approach to generate training patch pairs between the sparse and ground-truth point clouds, and an outlier removal step to suppress the noisy points from the sparse point cloud. Together with the low-res recovery, our whole method is able to achieve high-fidelity point cloud completion. Comprehensive evaluations are provided to demonstrate the effectiveness of the proposed method and its individual components.
['Lin Gao', 'Ling-Xiao Zhang', 'Chun-Peng Li', 'Bo wang', 'Ren-Wu Li']
2021-12-21
null
null
null
null
['point-cloud-completion']
['computer-vision']
[ 2.61680424e-01 6.61488622e-02 4.24215138e-01 -7.02341050e-02 -1.06613851e+00 -3.45034391e-01 4.37008798e-01 1.04808785e-01 6.37930483e-02 6.76511586e-01 -5.88598363e-02 2.47753039e-01 -1.97257451e-03 -1.13388240e+00 -9.67856288e-01 -6.91662490e-01 3.35112154e-01 7.00613141e-01 2.53659099e-01 3.11431382e-03 3.10516417e-01 8.13156486e-01 -1.82320762e+00 1.37273371e-01 1.20240676e+00 8.47967505e-01 4.57086653e-01 2.36529812e-01 -3.52284998e-01 -8.45079571e-02 -2.48922408e-01 -1.43709704e-01 4.55039769e-01 -1.37685642e-01 -4.17297632e-01 4.57962245e-01 5.86440325e-01 -4.61257011e-01 -7.84629360e-02 1.27657604e+00 4.31328975e-02 -2.26922855e-02 5.02684593e-01 -9.46425498e-01 -2.15850547e-01 -9.03566033e-02 -8.57861936e-01 -7.47432053e-01 3.99892062e-01 -5.40937521e-02 6.61766469e-01 -1.34806359e+00 5.98111987e-01 1.10320270e+00 1.05095017e+00 8.15356374e-02 -1.24082279e+00 -7.07675993e-01 -8.71053636e-02 -2.77974635e-01 -1.83862388e+00 -4.67146039e-01 1.16598308e+00 -4.05989349e-01 2.97622681e-01 3.47502083e-01 8.02438498e-01 5.17771661e-01 -1.35483012e-01 2.03685805e-01 9.85468328e-01 -1.28669083e-01 3.42888176e-01 -1.47778004e-01 -2.75310606e-01 3.62978339e-01 4.94148821e-01 1.19396232e-01 -2.67802089e-01 -5.93152523e-01 1.10088456e+00 5.99038422e-01 -5.29130340e-01 -4.76624519e-01 -1.26142251e+00 4.39069569e-01 5.52080035e-01 2.12575465e-01 -7.99153864e-01 -1.07233696e-01 -9.77863148e-02 -3.31259407e-02 6.76377475e-01 4.91997041e-02 -2.06617072e-01 -5.87239338e-04 -1.47588253e+00 4.91733938e-01 4.74689752e-01 1.20500541e+00 1.50179255e+00 -5.12838997e-02 1.29529446e-01 6.29867554e-01 4.02144611e-01 7.13686228e-01 -1.08141698e-01 -9.78156805e-01 5.39369822e-01 7.33345985e-01 3.65037501e-01 -1.31700039e+00 -7.14702606e-02 -3.27561498e-01 -1.21013975e+00 4.48108584e-01 1.12248451e-01 1.71241254e-01 -8.70649695e-01 1.05631185e+00 5.38319468e-01 6.12530112e-01 -2.54137907e-02 1.01049483e+00 6.89979255e-01 6.80254459e-01 -3.46635133e-01 -4.33877766e-01 1.10449505e+00 -4.35578257e-01 -6.24288440e-01 4.39242870e-02 1.38761565e-01 -1.00769353e+00 7.80251920e-01 5.00171661e-01 -1.07657409e+00 -5.66870868e-01 -1.02733552e+00 -1.43170893e-01 3.34352553e-01 1.33155495e-01 2.03367934e-01 4.69885841e-02 -6.84458375e-01 8.79228652e-01 -9.84356165e-01 1.26268461e-01 5.83007991e-01 1.07509598e-01 -5.40570199e-01 -6.10419691e-01 -5.41241765e-01 2.65519708e-01 1.64016008e-01 1.62819803e-01 -6.00944161e-01 -1.01350522e+00 -6.64402902e-01 1.27106413e-01 3.07633281e-01 -9.20787036e-01 6.70290053e-01 -8.04010689e-01 -1.11663330e+00 4.66092199e-01 -5.21838069e-01 9.07989161e-04 4.81025130e-01 3.64166778e-03 -1.55764623e-02 1.13884807e-01 2.33445317e-01 2.39710048e-01 1.14399302e+00 -1.95277965e+00 -6.05976105e-01 -5.22242427e-01 -4.08899099e-01 2.12166190e-01 1.24530256e-01 -7.01065660e-01 -5.44598877e-01 -7.35490441e-01 9.94549870e-01 -6.20594084e-01 -3.99584651e-01 1.54674321e-01 -3.29583764e-01 3.14262927e-01 1.04795563e+00 -8.02162349e-01 9.71992135e-01 -2.51481175e+00 3.70056517e-02 4.96422827e-01 3.61039221e-01 -6.68047816e-02 -2.40293033e-02 3.38235766e-01 -3.18861529e-02 -6.51357770e-02 -9.47599769e-01 -8.33092809e-01 -4.34037745e-01 3.75619888e-01 -4.09626693e-01 5.61597466e-01 2.28657231e-01 4.40838814e-01 -8.60178411e-01 -2.54187018e-01 3.18721592e-01 8.41057420e-01 -6.74386561e-01 9.70370770e-02 -2.34392047e-01 8.68390203e-01 -6.42223895e-01 8.02456260e-01 1.47804618e+00 -3.84619087e-02 -3.10238272e-01 -4.69667375e-01 -3.04667175e-01 -2.24737465e-01 -1.64706910e+00 2.09750628e+00 -3.18534762e-01 -1.36305705e-01 5.72378039e-01 -5.50129652e-01 1.21309626e+00 3.96847248e-01 7.36890197e-01 -1.75509632e-01 -1.11654371e-01 5.60806572e-01 -4.89753574e-01 -8.76802504e-02 6.50212586e-01 -4.30800378e-01 4.08560187e-01 -5.86715015e-03 -3.73135567e-01 -6.62878275e-01 -4.57130283e-01 1.15321137e-01 8.76621306e-01 1.99794486e-01 8.55614990e-02 1.00157904e-02 7.46856809e-01 2.53642231e-01 7.70972550e-01 1.67507395e-01 4.67998415e-01 1.48765182e+00 6.51023164e-02 -2.37112060e-01 -1.31880665e+00 -7.68890381e-01 -2.43196398e-01 -1.21508628e-01 4.29094225e-01 -5.16441584e-01 -6.43307388e-01 -2.88416654e-01 7.57815242e-02 5.63681781e-01 -3.28604102e-01 1.84620634e-01 -6.44438446e-01 -4.04507637e-01 -7.47807994e-02 9.77305695e-02 3.72085482e-01 -5.51111817e-01 -5.62759377e-02 2.69722432e-01 -2.44965628e-01 -9.75284159e-01 -2.88232356e-01 -3.25240672e-01 -1.43494785e+00 -1.10983622e+00 -7.62002289e-01 -5.96608400e-01 1.23801458e+00 5.18246055e-01 9.14509416e-01 6.24681413e-01 2.53782481e-01 -5.98066188e-02 -3.14070702e-01 2.18258545e-01 -2.41787389e-01 -2.89866805e-01 -1.13929875e-01 3.67176533e-01 -1.36589170e-01 -1.01465750e+00 -5.75493813e-01 1.87297419e-01 -1.02457643e+00 3.16949636e-01 7.17758358e-01 7.60957658e-01 1.27812421e+00 4.07721817e-01 9.33257118e-02 -7.83023655e-01 1.36634767e-01 -5.32534122e-01 -6.55320168e-01 -2.21975163e-01 -3.27499390e-01 -1.33136973e-01 7.47375906e-01 -1.10385612e-01 -9.10159409e-01 5.01845837e-01 -5.51593840e-01 -1.12219584e+00 -2.50552744e-01 3.51917177e-01 -3.78070325e-01 -3.23964894e-01 3.34605575e-01 4.97205198e-01 8.51239860e-02 -9.29383159e-01 1.84362859e-01 3.77242595e-01 6.35289848e-01 -6.21071517e-01 1.31600356e+00 9.20309484e-01 2.64800936e-01 -9.28646803e-01 -3.90961021e-01 -5.71579874e-01 -6.80621088e-01 2.81027220e-02 5.93718231e-01 -1.05455995e+00 -2.60203809e-01 2.63467252e-01 -1.36897469e+00 3.14633816e-01 -4.77060974e-01 4.48633373e-01 -4.97104406e-01 8.01073074e-01 -2.94777900e-01 -6.43094957e-01 -2.10204825e-01 -1.15740991e+00 1.56627107e+00 -1.66688040e-01 2.13680401e-01 -4.16845292e-01 1.01792604e-01 1.83974028e-01 1.39365897e-01 4.21520948e-01 6.79663539e-01 2.04800889e-02 -1.05606580e+00 -3.85405481e-01 -2.59442180e-01 2.43525416e-01 2.02208072e-01 4.29188982e-02 -7.80773580e-01 -3.58718067e-01 5.24902284e-01 1.80666134e-01 4.01869237e-01 3.10965717e-01 1.06854427e+00 -3.45440060e-01 -4.38550919e-01 9.84318674e-01 1.74685156e+00 -3.05160940e-01 8.51439893e-01 7.71292970e-02 1.02250290e+00 4.40241039e-01 8.74671280e-01 6.42183125e-01 3.12069327e-01 6.05066180e-01 7.55938590e-01 4.98158671e-02 -1.78574711e-01 -5.40178776e-01 -4.59022447e-02 1.00504339e+00 -4.36649591e-01 2.42539197e-01 -7.78848052e-01 4.88497138e-01 -1.64838755e+00 -7.08639085e-01 -6.84613705e-01 2.57119536e+00 6.57314539e-01 -2.62429476e-01 -2.35886231e-01 3.14389527e-01 8.07011247e-01 4.90321666e-02 -3.62852573e-01 3.53116810e-01 -5.19725829e-02 1.94255441e-01 4.09687102e-01 8.08906734e-01 -5.13251066e-01 7.33669639e-01 4.93077612e+00 1.23373008e+00 -1.04808450e+00 1.70261800e-01 2.09925547e-01 1.99454114e-01 -7.87905753e-01 4.51104760e-01 -6.46008968e-01 5.32808542e-01 3.18190962e-01 -4.10484942e-03 4.06405658e-01 8.21868360e-01 2.44366825e-01 -4.54351045e-02 -7.39186943e-01 1.32163882e+00 -8.18070117e-03 -1.34965038e+00 2.30841815e-01 2.37202689e-01 9.11725879e-01 -1.31058455e-01 -4.47938561e-01 -1.14903212e-01 -1.34472534e-01 -7.64088094e-01 7.40265012e-01 9.33910370e-01 9.35814977e-01 -8.65324616e-01 6.06860161e-01 8.42096508e-01 -1.31650007e+00 4.61107999e-01 -6.07473791e-01 -3.14034857e-02 4.07649815e-01 1.27322638e+00 -4.50332195e-01 1.11566401e+00 5.95030129e-01 9.11202550e-01 -3.05699944e-01 1.09070110e+00 -4.38979417e-02 2.30831355e-01 -5.17637610e-01 7.54095376e-01 -2.01850131e-01 -7.97942281e-01 9.58013535e-01 4.39703345e-01 8.94483924e-01 5.39342284e-01 3.29385072e-01 1.23721182e+00 9.97033492e-02 6.84110597e-02 -5.06162167e-01 4.04402763e-01 5.60776055e-01 1.26700950e+00 -6.35258436e-01 -3.51177305e-01 -4.67793882e-01 8.69922996e-01 4.11801189e-02 3.31242174e-01 -4.07077461e-01 -3.59009742e-03 5.92437923e-01 6.14384055e-01 3.62751186e-01 -5.31461477e-01 -7.84321189e-01 -1.25942254e+00 3.63178879e-01 -7.40273893e-01 -2.98187524e-01 -8.21621060e-01 -1.14918327e+00 6.31305158e-01 -1.81351140e-01 -2.01914191e+00 5.70840128e-02 1.00343227e-01 -4.91290301e-01 1.15500569e+00 -1.59003115e+00 -1.15506232e+00 -6.43278301e-01 6.24464273e-01 3.83364797e-01 2.94259310e-01 5.50196826e-01 3.38957161e-01 2.42798273e-02 -2.99804676e-02 1.50278240e-01 -3.18134904e-01 4.49993044e-01 -6.62059546e-01 1.46337703e-01 8.49027693e-01 -2.41208896e-01 7.27451205e-01 5.93875647e-01 -1.15871048e+00 -1.58792913e+00 -1.11946607e+00 4.98657644e-01 -8.25649500e-02 1.25651211e-01 -1.73499122e-01 -1.41556692e+00 5.03129661e-01 -3.91568094e-01 2.09516481e-01 9.20789465e-02 -3.56509149e-01 5.78868538e-02 -8.87905508e-02 -1.56270385e+00 4.08001184e-01 9.19191182e-01 -2.42589355e-01 -6.69321179e-01 1.18718728e-01 5.81790984e-01 -6.46007240e-01 -1.00185835e+00 7.33658552e-01 2.47613251e-01 -1.07596993e+00 1.07065296e+00 2.73149312e-01 4.96343076e-01 -9.28883910e-01 -2.64305919e-01 -1.16766679e+00 -4.81627196e-01 -4.57348615e-01 -7.06105027e-03 1.40581596e+00 -7.70837143e-02 -5.69678903e-01 1.04785836e+00 3.86646897e-01 -4.97888207e-01 -5.11301517e-01 -1.00554490e+00 -4.35050905e-01 -8.17861632e-02 -4.35291022e-01 1.03146362e+00 1.10367000e+00 -5.53002894e-01 -1.10447228e-01 -2.77803957e-01 8.42981696e-01 9.44874883e-01 4.46600854e-01 1.02600133e+00 -1.43020201e+00 -2.35732980e-02 -1.92539245e-02 -3.18858773e-01 -1.21497858e+00 -1.73932806e-01 -5.64404368e-01 2.05996156e-01 -1.45047605e+00 -5.98148815e-02 -1.01562178e+00 4.03065443e-01 1.70287475e-01 -1.38984680e-01 3.92057747e-01 2.85791270e-02 7.88588405e-01 1.27148136e-01 8.34599257e-01 1.46797478e+00 6.08783551e-02 -1.42693713e-01 -1.15389368e-02 -5.40182650e-01 7.57627130e-01 3.35717171e-01 -5.32746792e-01 -1.17716886e-01 -4.79183823e-01 8.25785249e-02 3.43182087e-01 4.38668728e-01 -1.29508305e+00 2.53583372e-01 -2.19471574e-01 2.81247377e-01 -1.11044741e+00 5.87285697e-01 -1.43276238e+00 9.03284490e-01 1.35573074e-01 4.99843508e-01 -2.66318291e-01 4.03957926e-02 6.24145329e-01 -4.78012860e-01 -3.28112960e-01 8.34570289e-01 -1.94769591e-01 -2.61416316e-01 8.34504008e-01 3.15828621e-01 -4.18457657e-01 8.80669534e-01 -5.87830722e-01 1.58834964e-01 -5.02844825e-02 -6.58912599e-01 3.37327737e-03 1.32856393e+00 3.93825322e-02 8.78002107e-01 -1.55053306e+00 -9.04177666e-01 8.77372384e-01 9.75335315e-02 9.88329411e-01 5.71980417e-01 7.16254652e-01 -8.22637796e-01 -6.06364198e-02 7.96316490e-02 -9.30433869e-01 -1.05381525e+00 5.32773137e-01 -1.26253888e-01 1.11751020e-01 -1.11590254e+00 3.68673027e-01 2.18297362e-01 -3.86749536e-01 -1.86320037e-01 -4.88747716e-01 2.54710242e-02 -2.15632498e-01 4.30033624e-01 2.47294948e-01 3.66066247e-01 -9.96232390e-01 -4.56476472e-02 1.26920938e+00 3.58056754e-01 6.37712404e-02 1.49875367e+00 -1.54769510e-01 -4.45727468e-01 2.06407279e-01 1.08441222e+00 5.53311229e-01 -1.34361827e+00 -1.46627158e-01 -5.10746181e-01 -1.02794647e+00 2.20765665e-01 4.99577932e-02 -1.18434155e+00 6.79839551e-01 1.64941490e-01 6.20594546e-02 1.09273863e+00 -2.30649367e-01 9.66737807e-01 -2.11885676e-01 7.18425274e-01 -5.11226058e-01 -5.84663212e-01 2.72643596e-01 1.16296613e+00 -1.01693773e+00 4.73129928e-01 -9.81419384e-01 -1.67667702e-01 1.16504359e+00 2.25468650e-01 -4.84197259e-01 6.75370157e-01 1.09110765e-01 -4.59992886e-01 -3.44724983e-01 -2.89406449e-01 -2.28242949e-02 2.57238328e-01 4.83353168e-01 -6.80980012e-02 -1.06558025e-01 -3.30159426e-01 5.12459815e-01 -3.11040938e-01 2.68624127e-01 3.52800667e-01 7.79728413e-01 -4.67522085e-01 -1.16449225e+00 -1.06355810e+00 4.49935615e-01 5.94110563e-02 -5.42447269e-02 1.19776735e-02 4.09768373e-01 3.11206073e-01 6.30805790e-01 1.98001578e-01 -3.06464851e-01 4.91274625e-01 -2.18137756e-01 1.32284537e-01 -7.85468519e-01 -1.13799766e-01 3.63285333e-01 -3.93598408e-01 -6.28827333e-01 -3.36153805e-01 -7.19496131e-01 -1.35713446e+00 -4.91422921e-01 -3.32781702e-01 3.27973843e-01 6.23850644e-01 6.36399865e-01 5.21047592e-01 1.39058843e-01 8.63615334e-01 -1.48839366e+00 -1.04836836e-01 -6.92939937e-01 -7.39659727e-01 5.65244198e-01 4.05455410e-01 -7.39769995e-01 -7.76512682e-01 -9.25370231e-02]
[8.413784980773926, -3.44325590133667]
6032c74d-ac58-4afc-9f30-4ef952b31b35
supervised-speech-representation-learning-for
2106.00531
null
https://arxiv.org/abs/2106.00531v2
https://arxiv.org/pdf/2106.00531v2.pdf
Supervised Speech Representation Learning for Parkinson's Disease Classification
Recently proposed automatic pathological speech classification techniques use unsupervised auto-encoders to obtain a high-level abstract representation of speech. Since these representations are learned based on reconstructing the input, there is no guarantee that they are robust to pathology-unrelated cues such as speaker identity information. Further, these representations are not necessarily discriminative for pathology detection. In this paper, we exploit supervised auto-encoders to extract robust and discriminative speech representations for Parkinson's disease classification. To reduce the influence of speaker variabilities unrelated to pathology, we propose to obtain speaker identity-invariant representations by adversarial training of an auto-encoder and a speaker identification task. To obtain a discriminative representation, we propose to jointly train an auto-encoder and a pathological speech classifier. Experimental results on a Spanish database show that the proposed supervised representation learning methods yield more robust and discriminative representations for automatically classifying Parkinson's disease speech, outperforming the baseline unsupervised representation learning system.
['Ina Kodrasi', 'Parvaneh Janbakhshi']
2021-06-01
null
null
null
null
['speaker-identification']
['speech']
[ 5.21476150e-01 5.98365247e-01 -8.83800685e-02 -6.09290898e-01 -1.13962758e+00 -2.22196475e-01 3.87636006e-01 -7.67298788e-02 -1.40717342e-01 5.59631586e-01 8.04514885e-01 1.79647899e-03 2.58647859e-01 -3.92459124e-01 -4.85053331e-01 -6.91059113e-01 -9.99800563e-02 3.17184955e-01 -7.46182650e-02 -3.11101880e-02 -1.68009609e-01 3.67915809e-01 -1.50745320e+00 5.85553586e-01 1.03666341e+00 5.88590324e-01 1.71085373e-01 7.07641780e-01 -1.09608993e-02 9.83058631e-01 -1.04229903e+00 -1.33677915e-01 -1.27255455e-01 -6.96716309e-01 -9.66012299e-01 5.63451052e-01 2.29012549e-01 -4.18866843e-01 -7.27199495e-01 1.20691431e+00 4.63371754e-01 -1.37139142e-01 9.55920696e-01 -9.21711445e-01 -8.60455215e-01 6.88924074e-01 1.76572427e-02 5.60065508e-01 5.13722718e-01 -1.41977578e-01 1.12112236e+00 -4.94963557e-01 4.87458497e-01 1.42966652e+00 5.17698348e-01 8.19828093e-01 -1.46227491e+00 -4.39395070e-01 1.19708963e-01 2.65834600e-01 -1.30163968e+00 -8.86578441e-01 1.08710408e+00 -5.78765810e-01 6.88965023e-01 2.79883921e-01 2.30691671e-01 1.76133609e+00 3.87657471e-02 9.27660346e-01 8.42855632e-01 -4.88203228e-01 3.96151423e-01 1.24646746e-01 4.74085473e-02 6.57744765e-01 -1.66878387e-01 3.01232040e-01 -4.53161418e-01 -4.53073740e-01 5.49494386e-01 -1.15920819e-01 -4.38228548e-01 -2.70602345e-01 -1.19673419e+00 9.37622547e-01 2.57757396e-01 6.42958939e-01 -5.14598250e-01 -1.67606235e-01 6.30668342e-01 4.52882141e-01 4.96078014e-01 4.02970798e-02 -6.62551373e-02 3.85827087e-02 -8.95479381e-01 -1.41356349e-01 5.44358492e-01 7.39982426e-01 1.94713727e-01 6.21341586e-01 -1.71299815e-01 1.17104733e+00 3.81793797e-01 6.60565436e-01 1.29197049e+00 -7.16938555e-01 4.08561170e-01 3.90939593e-01 -4.41885203e-01 -8.17291260e-01 -2.35366195e-01 -2.82323003e-01 -8.72902989e-01 -2.10184585e-02 -7.29518533e-02 2.97956437e-01 -8.91209602e-01 1.83306515e+00 -1.74587578e-01 1.12980716e-01 4.87830877e-01 5.60327590e-01 5.83657563e-01 3.83834273e-01 1.72173291e-01 -2.27418602e-01 1.12647903e+00 -6.04475379e-01 -8.81457686e-01 -3.16170335e-01 5.80593765e-01 -4.25860941e-01 8.21095645e-01 1.56421453e-01 -7.58871257e-01 -5.91773450e-01 -1.07099473e+00 1.12412229e-01 -2.05093503e-01 3.14846516e-01 2.20572099e-01 1.03330970e+00 -9.30905640e-01 4.00289476e-01 -1.05994332e+00 -4.65114951e-01 4.56141025e-01 3.66216093e-01 -5.80000222e-01 1.52407080e-01 -1.14179611e+00 8.55133712e-01 3.73693228e-01 -2.83626199e-01 -1.09006667e+00 -1.28881261e-01 -1.01862800e+00 -1.48633689e-01 -2.50216872e-01 -5.73109329e-01 1.09610033e+00 -1.35516548e+00 -1.54438066e+00 9.19249535e-01 -2.96621948e-01 -8.33186090e-01 2.03661442e-01 1.69432104e-01 -9.27411735e-01 6.78249836e-01 3.11219424e-01 2.49053851e-01 1.44751763e+00 -9.67527092e-01 -2.63833880e-01 -5.81096292e-01 -3.10146213e-01 -1.79313019e-01 -5.72749794e-01 1.14233904e-01 3.14793766e-01 -1.14417315e+00 2.71087140e-01 -7.13186443e-01 3.47332694e-02 -8.46061558e-02 -5.86043894e-01 -5.02219424e-02 8.16563189e-01 -1.13383365e+00 8.40799630e-01 -2.43722248e+00 2.90965945e-01 4.34421673e-02 7.39478394e-02 2.94218838e-01 -3.45405549e-01 2.72263974e-01 -5.67540824e-01 -9.67617556e-02 -6.16023779e-01 -4.98103380e-01 -8.17329884e-02 6.60229504e-01 -3.07616472e-01 7.59915173e-01 5.19199371e-01 5.25599957e-01 -7.60282874e-01 -3.75753015e-01 2.14135885e-01 6.49220884e-01 -6.58572376e-01 2.08527133e-01 2.41391882e-01 6.43467546e-01 -4.84855682e-01 7.48721838e-01 2.40376070e-01 1.22867964e-01 1.42998636e-01 -3.03347167e-02 4.34202701e-01 7.08058417e-01 -7.34639168e-01 1.69580412e+00 -5.35523653e-01 5.33334315e-01 1.59229897e-02 -1.72643125e+00 1.07961822e+00 7.26515532e-01 5.32588899e-01 -4.05322760e-01 1.06077202e-01 1.99326143e-01 9.51823071e-02 -5.42454243e-01 -6.57048076e-02 -4.39749002e-01 -3.43655124e-02 3.10874909e-01 5.40433288e-01 3.16682518e-01 -2.16112003e-01 7.16150403e-02 1.38651979e+00 -4.57602292e-01 2.69907147e-01 -6.89833984e-02 1.02661347e+00 -5.88149488e-01 6.63250983e-01 3.61195117e-01 -4.89408463e-01 8.49692285e-01 2.78376132e-01 3.33475508e-02 -8.15304875e-01 -1.28556037e+00 -2.18833506e-01 7.19693780e-01 -5.42358637e-01 -3.28162879e-01 -7.90421426e-01 -9.74671245e-01 -1.38404197e-04 8.68310213e-01 -6.37907863e-01 -8.48532438e-01 -3.33904415e-01 -4.61255342e-01 9.26331878e-01 7.48988390e-01 1.83297088e-03 -9.30615783e-01 -5.50071076e-02 4.27255541e-01 -3.72351438e-01 -1.15299296e+00 -5.44083238e-01 3.52871209e-01 -9.99466598e-01 -1.13210905e+00 -1.08405983e+00 -1.02427197e+00 9.31521773e-01 -1.23833559e-01 6.06649220e-01 -4.90621895e-01 -3.73472989e-01 7.97406018e-01 -5.09529710e-01 -1.02794267e-01 -1.30243719e+00 -1.06609024e-01 4.31686819e-01 4.87991899e-01 3.88580978e-01 -9.33960855e-01 -9.55333933e-02 1.61391571e-01 -8.71740997e-01 -6.16066992e-01 7.06557155e-01 1.05857658e+00 4.36991006e-01 3.53363268e-02 1.03509748e+00 -5.64946353e-01 5.86971104e-01 -3.23523432e-01 1.32367238e-01 7.36975810e-03 -2.00068176e-01 3.12075108e-01 6.96852565e-01 -4.78464186e-01 -9.31579530e-01 4.42277789e-01 -4.24504280e-01 -6.68131232e-01 -5.33648491e-01 2.51107067e-01 -3.85955334e-01 2.68563151e-01 8.05419624e-01 8.63595366e-01 5.05604982e-01 -6.40830457e-01 3.92786831e-01 1.16829550e+00 7.69309461e-01 -2.84758061e-01 7.87856758e-01 3.92193258e-01 -6.47193074e-01 -1.19614375e+00 -4.82157052e-01 -7.05737591e-01 -8.40689540e-01 1.44708991e-01 9.76897955e-01 -1.14260948e+00 -3.19734737e-02 2.23255157e-01 -1.20540595e+00 2.68380165e-01 -5.34511149e-01 6.28635228e-01 -1.00878263e+00 7.22790122e-01 -4.03522640e-01 -7.63369739e-01 -2.50201344e-01 -1.12364435e+00 1.21978617e+00 -4.74961489e-01 -5.44142723e-01 -8.81994665e-01 2.00169966e-01 5.89490116e-01 1.67743102e-01 -8.65444019e-02 9.18588936e-01 -1.06072438e+00 1.36903226e-01 -3.90593201e-01 2.98655242e-01 1.05991125e+00 8.38610232e-01 -4.73880231e-01 -1.16189766e+00 -1.95494518e-01 3.63325983e-01 -2.43211821e-01 9.20904756e-01 1.22667536e-01 1.07268107e+00 -5.23765147e-01 -1.53351471e-01 3.23506922e-01 6.85191572e-01 2.58904219e-01 7.49315679e-01 1.01181455e-01 5.00712574e-01 6.03448570e-01 -1.41184956e-01 2.96463042e-01 1.41641229e-01 4.69056398e-01 1.07564796e-02 2.43586943e-01 -4.71602947e-01 -2.37540349e-01 8.89357150e-01 1.13347518e+00 -2.67866068e-02 2.34877929e-01 -7.05469549e-01 7.52450287e-01 -1.34740102e+00 -1.15341187e+00 4.14737910e-01 2.07374930e+00 8.28256786e-01 -1.49247646e-01 4.46405321e-01 6.76967502e-01 9.90972281e-01 1.60616115e-01 -4.77101654e-01 -1.86126143e-01 -1.73483029e-01 2.92614222e-01 1.11448660e-01 3.11647683e-01 -1.24794114e+00 6.42453134e-01 6.87140751e+00 3.09925139e-01 -9.79326546e-01 3.25330675e-01 1.77732795e-01 5.93738295e-02 -1.05879381e-01 -5.70984960e-01 -1.86574280e-01 5.95362842e-01 1.28325427e+00 -1.72867715e-01 1.91428646e-01 9.60507035e-01 2.98057467e-01 5.99460185e-01 -1.23950911e+00 9.89368439e-01 5.14078557e-01 -7.24682748e-01 3.35152090e-01 1.64543629e-01 3.46340120e-01 -7.67614394e-02 8.08574408e-02 2.23098189e-01 1.49601400e-01 -9.11930859e-01 5.54290235e-01 4.97210830e-01 4.72898543e-01 -7.77312517e-01 6.40005767e-01 2.34446853e-01 -8.34662437e-01 -2.25935876e-01 -3.08189034e-01 4.84076679e-01 8.11542571e-03 3.42144996e-01 -9.83565152e-01 4.84799832e-01 4.03212458e-01 8.43590498e-01 -4.03277308e-01 5.70951819e-01 -2.73742229e-01 6.49988353e-01 -3.65140922e-02 2.91542143e-01 -1.34931132e-01 1.44249231e-01 8.61013532e-01 1.10573804e+00 2.10403517e-01 -9.90094393e-02 4.67684641e-02 1.06754732e+00 3.80149446e-02 1.26505390e-01 -7.87824452e-01 -6.27198637e-01 1.89769179e-01 5.07474840e-01 -1.07128911e-01 -2.75473267e-01 -4.89304960e-01 1.54834378e+00 2.47939110e-01 1.97088331e-01 -4.64987934e-01 -2.36201003e-01 8.48194838e-01 3.19508947e-02 4.85552281e-01 -1.02998845e-01 1.43979624e-01 -1.32030177e+00 6.41911402e-02 -1.29224300e+00 2.69793242e-01 -4.70392674e-01 -1.55726004e+00 7.92651296e-01 -3.60435456e-01 -1.47178030e+00 -6.34114683e-01 -4.84868437e-01 -6.14865363e-01 7.69399643e-01 -1.30779588e+00 -1.21942508e+00 2.72710174e-01 9.23061132e-01 7.39712179e-01 -8.45588803e-01 1.36976826e+00 1.87405989e-01 -4.99576628e-01 6.80870175e-01 1.54863268e-01 4.62737143e-01 6.52787387e-01 -1.30113637e+00 7.82388672e-02 7.37815082e-01 4.21151608e-01 6.25414550e-01 5.60771585e-01 -5.29904306e-01 -1.05628300e+00 -1.22579515e+00 9.08264875e-01 -2.65751272e-01 7.09098876e-01 -2.62782156e-01 -9.94062006e-01 7.85802782e-01 -1.00185618e-01 7.22572878e-02 1.09537184e+00 -1.32982116e-02 -6.61602676e-01 -1.38189033e-01 -1.24915588e+00 1.71904713e-01 9.15781021e-01 -1.26785696e+00 -1.43039632e+00 5.33486426e-01 4.77732658e-01 2.88395464e-01 -8.79281104e-01 -2.55459249e-02 1.48127407e-01 -6.40224397e-01 1.28582287e+00 -7.42200434e-01 2.65687287e-01 -4.91439141e-02 -5.12052000e-01 -1.40536964e+00 -2.30336025e-01 -3.56750578e-01 -1.33084416e-01 1.22236550e+00 2.19182923e-01 -7.51524985e-01 5.91606200e-01 1.04322039e-01 -3.21608603e-01 -4.65351269e-02 -1.29933524e+00 -1.18160641e+00 2.06909135e-01 -3.37468415e-01 2.28597879e-01 1.02636075e+00 3.36761683e-01 3.31988633e-01 -2.83038199e-01 5.93803108e-01 7.25062788e-01 -3.53410035e-01 2.72724360e-01 -1.29456258e+00 -3.91099989e-01 -1.51254699e-01 -1.12173188e+00 -5.87846637e-01 8.56335461e-01 -1.32181573e+00 1.09839663e-01 -1.25923145e+00 -2.95662377e-02 1.36866599e-01 -5.61165273e-01 5.21616042e-01 1.14463471e-01 -7.49162510e-02 -2.38760799e-01 3.45923215e-01 -4.05092426e-02 9.55340922e-01 7.80650079e-01 -8.05005610e-01 -1.26913711e-01 2.22521141e-01 -6.13835871e-01 8.12630951e-01 8.52735639e-01 -5.76112628e-01 -5.58791280e-01 -5.88198677e-02 -8.10789704e-01 -3.38214673e-02 5.59523046e-01 -1.11118376e+00 -1.38212115e-01 4.26189452e-01 2.22073108e-01 -3.06690246e-01 4.85716403e-01 -6.69282556e-01 -3.46991569e-01 5.82113445e-01 -6.11331046e-01 -5.45824409e-01 -5.79379611e-02 9.35149670e-01 -6.29114330e-01 -3.38288337e-01 8.50414455e-01 -1.44753650e-01 -3.63986313e-01 -3.49440575e-02 -9.62029219e-01 -4.31812964e-02 6.88610733e-01 -8.40656236e-02 1.58965588e-01 -5.11219919e-01 -1.27807581e+00 -5.09983659e-01 2.76841857e-02 6.57933533e-01 6.95131600e-01 -1.42211854e+00 -7.97254622e-01 4.92493242e-01 3.20331126e-01 -5.92569113e-01 1.34364322e-01 6.54502690e-01 -6.08462021e-02 5.03180981e-01 -2.78308451e-01 -5.78352332e-01 -1.33384216e+00 6.04508519e-01 3.56159747e-01 1.71807572e-01 -9.87042785e-01 6.04264915e-01 1.83513597e-01 -4.80186254e-01 4.58007812e-01 -5.33474982e-01 -2.25647256e-01 3.71870585e-02 7.61568904e-01 7.27442503e-02 1.90422505e-01 -1.22027266e+00 -6.13185525e-01 1.35222957e-01 -2.30054319e-01 -7.94466138e-02 1.25242960e+00 -8.17053095e-02 2.25164101e-01 4.69444275e-01 1.44055760e+00 -8.12627599e-02 -6.20122612e-01 -2.72912472e-01 7.13327974e-02 -1.36726677e-01 2.76276052e-01 -5.31616747e-01 -1.18198943e+00 8.81415606e-01 8.73074174e-01 3.07872802e-01 9.86498177e-01 2.40554571e-01 6.26612484e-01 4.01348203e-01 2.19823122e-01 -1.03035367e+00 1.49413362e-01 2.40495235e-01 1.36100185e+00 -1.01070857e+00 -3.94523621e-01 -4.97193396e-01 -5.89862883e-01 1.18713439e+00 2.68559828e-02 -2.46356010e-01 6.36657715e-01 5.18689584e-03 2.14167967e-01 1.03474580e-01 -4.10572022e-01 -3.81700695e-01 4.95420188e-01 1.31523049e+00 3.91736150e-01 6.79131746e-02 4.91976086e-03 9.06114638e-01 -3.98894459e-01 -3.20306540e-01 3.20080072e-01 7.71569371e-01 -3.47708553e-01 -1.26211500e+00 -5.84775031e-01 2.37973899e-01 -4.37913030e-01 8.81422162e-02 -7.38288164e-01 3.84250700e-01 -1.57854602e-01 9.86666739e-01 1.01027898e-01 -3.12627017e-01 4.36446041e-01 6.34311199e-01 3.92379910e-01 -9.32187200e-01 -1.61212936e-01 7.51754791e-02 3.11834902e-01 -5.34318566e-01 -6.17938638e-01 -9.66580689e-01 -1.12215197e+00 3.53549302e-01 -1.50218830e-01 -1.26316726e-01 4.05321300e-01 1.06094933e+00 3.21689457e-01 9.54986751e-01 7.74500370e-01 -5.97495794e-01 -1.03348470e+00 -1.12714791e+00 -9.20767963e-01 5.23892105e-01 5.29356718e-01 -8.21134031e-01 -5.83442748e-01 4.29985255e-01]
[14.475842475891113, 6.311915397644043]