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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
7a5e8152-d046-4793-b20d-b48b965116be | multilingual-paraphrase-generation-for | null | null | https://aclanthology.org/2021.nlp4convai-1.4 | https://aclanthology.org/2021.nlp4convai-1.4.pdf | Multilingual Paraphrase Generation For Bootstrapping New Features in Task-Oriented Dialog Systems | The lack of labeled training data for new features is a common problem in rapidly changing real-world dialog systems. As a solution, we propose a multilingual paraphrase generation model that can be used to generate novel utterances for a target feature and target language. The generated utterances can be used to augment existing training data to improve intent classification and slot labeling models. We evaluate the quality of generated utterances using intrinsic evaluation metrics and by conducting downstream evaluation experiments with English as the source language and nine different target languages. Our method shows promise across languages, even in a zero-shot setting where no seed data is available. | ['Patrick Lehnen', 'Tobias Falke', 'Caglar Tirkaz', 'Subhadarshi Panda'] | null | null | null | null | emnlp-nlp4convai-2021-11 | ['paraphrase-generation', 'paraphrase-generation'] | ['computer-code', 'natural-language-processing'] | [ 8.71177763e-02 5.27070582e-01 -2.67107248e-01 -7.54712760e-01
-8.83334994e-01 -9.26534235e-01 9.59452987e-01 -4.80448678e-02
-6.78056955e-01 1.16600287e+00 6.65267110e-01 -3.45359623e-01
6.17741942e-01 -4.39741492e-01 -2.40116581e-01 4.97240238e-02
3.92220318e-01 1.06818092e+00 2.67149461e-03 -8.82888973e-01
-1.14815108e-01 6.25345111e-02 -1.03607738e+00 5.63592017e-01
9.20085073e-01 2.09950611e-01 1.91572711e-01 6.15542531e-01
-4.77995098e-01 7.35079229e-01 -9.00264621e-01 -7.06372857e-01
2.43855119e-01 -7.16063559e-01 -1.16858494e+00 1.91230491e-01
8.52843449e-02 -3.79858315e-01 -2.19532698e-01 6.20839596e-01
4.55370635e-01 4.00691211e-01 4.11875665e-01 -1.29784191e+00
-6.81876123e-01 1.03250420e+00 1.18819863e-01 1.12545013e-01
6.93495512e-01 3.55443537e-01 1.12121022e+00 -1.07514977e+00
9.54272807e-01 1.63675344e+00 4.38596725e-01 9.90603805e-01
-1.46174550e+00 -4.07741845e-01 1.99701767e-02 -5.39939851e-02
-9.68374491e-01 -9.06907737e-01 5.17326474e-01 -2.02088639e-01
1.29985023e+00 2.58834474e-02 2.16721043e-01 1.50409031e+00
-1.39646187e-01 7.78308570e-01 9.72891450e-01 -8.01269412e-01
2.01681912e-01 6.97685957e-01 3.38094294e-01 5.99209130e-01
-1.90889865e-01 1.47594541e-01 -6.64700568e-01 -4.52740401e-01
7.57518262e-02 -3.20373148e-01 -6.06485121e-02 -1.22228704e-01
-1.04396117e+00 1.38519526e+00 1.58410445e-01 3.67454410e-01
-5.18661849e-02 -3.41474891e-01 5.25232852e-01 7.94531882e-01
6.36546612e-01 1.14213860e+00 -5.79477549e-01 -2.35931173e-01
-5.30537128e-01 4.75804478e-01 1.20667589e+00 1.04014754e+00
8.75451386e-01 9.33439210e-02 -5.07791162e-01 1.35304487e+00
1.42331704e-01 4.14925218e-01 1.05844629e+00 -1.09992146e+00
5.23564219e-01 5.60242355e-01 2.73360997e-01 -1.35869518e-01
-4.05173361e-01 1.19470961e-01 -3.14346179e-02 -1.48294851e-01
3.96780729e-01 -5.73625982e-01 -6.85624003e-01 2.02185416e+00
2.34868169e-01 -2.70943969e-01 7.22625732e-01 5.31519055e-01
9.09303248e-01 6.60236359e-01 1.11484818e-01 -2.56103069e-01
1.15180206e+00 -1.38685584e+00 -6.34726942e-01 -5.99557459e-01
9.66122091e-01 -7.55378127e-01 1.52669799e+00 -1.93822637e-01
-7.48262227e-01 -4.58916038e-01 -7.02930808e-01 -2.03902096e-01
-3.62974107e-01 -5.57834432e-02 6.13819003e-01 7.56008625e-01
-1.15767753e+00 1.13527805e-01 -4.13519502e-01 -8.11024547e-01
-2.80811012e-01 2.40853149e-02 -3.02962691e-01 -1.90346032e-01
-1.55438018e+00 1.27021372e+00 4.47719067e-01 -3.85561109e-01
-7.58199155e-01 -5.55741191e-01 -1.17460072e+00 -4.80733998e-02
1.78392068e-01 -5.33349872e-01 1.83441722e+00 -8.08501661e-01
-1.83044362e+00 6.71726525e-01 -2.59458899e-01 -3.56660545e-01
2.75152624e-01 1.15590312e-01 -2.94027030e-01 -2.56142288e-01
5.19936979e-01 1.04724681e+00 5.58588922e-01 -9.95241463e-01
-6.44112706e-01 -1.01837300e-01 3.09224725e-01 7.02818990e-01
-3.64662468e-01 -8.44198465e-03 -4.81832810e-02 -3.60928506e-01
-3.59305650e-01 -1.11317503e+00 -3.45812410e-01 -4.80151772e-01
-1.89448699e-01 -3.64386529e-01 5.94788969e-01 -6.19092286e-01
8.42978776e-01 -1.88593066e+00 8.19740444e-02 -2.26552799e-01
-2.68280953e-01 2.08820656e-01 -4.24632043e-01 6.57091081e-01
2.74015337e-01 4.93388250e-02 -1.61899850e-01 -5.13588130e-01
-9.13789645e-02 1.95125446e-01 -4.78003532e-01 -3.93444389e-01
3.34500760e-01 1.00417209e+00 -1.13248360e+00 -2.32959136e-01
1.07714653e-01 -2.01764069e-02 -6.09086454e-01 5.05928338e-01
-5.01327276e-01 5.47154188e-01 -1.28249377e-01 2.92934656e-01
5.16457409e-02 -1.73118174e-01 3.46778035e-01 1.54671878e-01
2.13218942e-01 6.43812418e-01 -6.21392310e-01 1.95662153e+00
-9.46176231e-01 5.78475356e-01 -1.53019324e-01 -4.91768032e-01
9.92185712e-01 4.44845110e-01 1.15989372e-02 -5.90221286e-01
-8.07429384e-03 2.81477869e-01 1.54545486e-01 -3.53958070e-01
9.23364580e-01 -5.26554227e-01 -6.88133001e-01 8.64865601e-01
4.01092827e-01 -4.23743963e-01 3.84692848e-01 4.52202260e-01
1.05886793e+00 -3.38960946e-01 3.92120779e-01 -1.31979838e-01
3.43253255e-01 5.06280959e-01 2.77527034e-01 8.56459916e-01
-2.10493356e-01 2.77298898e-01 1.47799224e-01 -2.79067487e-01
-1.10073674e+00 -9.49672759e-01 2.12505404e-02 1.59864128e+00
-1.18589483e-01 -2.20408514e-01 -5.06453931e-01 -9.36009467e-01
2.65045743e-02 1.19964516e+00 -3.03828508e-01 -3.35467666e-01
-4.45660681e-01 -4.09308046e-01 8.56305897e-01 2.73449630e-01
3.34273934e-01 -1.27167642e+00 -3.02750379e-01 5.07612288e-01
-6.58865035e-01 -1.27478099e+00 -5.66710830e-01 2.38439023e-01
-4.27107781e-01 -6.19411647e-01 -5.66780329e-01 -9.04343784e-01
4.05718148e-01 2.62084335e-01 1.28005552e+00 -9.81565788e-02
-5.08891558e-03 4.03323859e-01 -4.90184605e-01 -2.10060552e-01
-1.25413001e+00 3.97735745e-01 3.20557803e-01 -4.32008564e-01
5.86604834e-01 -1.07301176e-01 6.38782531e-02 2.72507071e-01
-7.12511599e-01 1.92132190e-01 1.80310190e-01 1.39934337e+00
-3.21930677e-01 -5.74148238e-01 1.03235924e+00 -1.06558406e+00
1.33847189e+00 -7.01432467e-01 -3.47200215e-01 4.43616688e-01
-5.41545510e-01 3.25708419e-01 7.21177876e-01 -4.79952604e-01
-1.58365202e+00 -2.27972064e-02 -2.28222251e-01 -5.68552166e-02
-1.42113209e-01 6.34461284e-01 4.01043259e-02 3.44408602e-01
1.10796916e+00 -3.82081568e-02 1.81970745e-01 -5.50847471e-01
9.40716803e-01 1.10706365e+00 2.90168136e-01 -6.32417202e-01
4.61447537e-01 -4.65676263e-02 -8.87722611e-01 -8.67190182e-01
-8.28090489e-01 -4.19995040e-01 -5.96530735e-01 -9.12881084e-03
5.37253201e-01 -1.06417036e+00 -1.08711794e-01 1.95353210e-01
-1.35066736e+00 -7.15910852e-01 -5.46097517e-01 2.43398026e-01
-3.82670730e-01 1.70827494e-03 -7.40670621e-01 -5.77338338e-01
-4.18023497e-01 -1.18934774e+00 1.01994467e+00 3.15883130e-01
-7.66958535e-01 -1.25988722e+00 4.29914415e-01 4.81766790e-01
5.40907979e-01 -5.99059045e-01 9.39559698e-01 -1.25925815e+00
-7.46736899e-02 -3.50306742e-02 9.25239548e-02 2.21128151e-01
5.02430797e-01 -5.02227962e-01 -9.11910832e-01 -4.98298585e-01
5.21335229e-02 -1.16222262e+00 3.96311432e-01 -1.75258428e-01
1.75057158e-01 -4.05370206e-01 -3.54459405e-01 -7.91383237e-02
7.79916942e-01 3.70240420e-01 -6.71629757e-02 3.26561294e-02
2.61129946e-01 9.14656460e-01 7.43617117e-01 3.48209083e-01
5.29500067e-01 9.08451438e-01 -7.04799652e-01 1.03748500e-01
-2.24063203e-01 -6.44410789e-01 5.12645841e-01 9.42353785e-01
9.83717024e-01 -4.54627723e-01 -9.64113355e-01 7.10707903e-01
-1.72393239e+00 -7.15030313e-01 6.12821639e-01 1.84090698e+00
1.27326500e+00 -9.11057591e-02 6.52801096e-02 -7.09774256e-01
6.69919193e-01 4.60706055e-02 -7.14897871e-01 -5.26518404e-01
-3.59962508e-02 3.15522194e-01 5.56132682e-02 1.03540683e+00
-6.96574569e-01 1.60031414e+00 7.34525013e+00 4.06781435e-01
-1.08784163e+00 3.87959450e-01 5.93775153e-01 5.63558750e-02
-5.65358460e-01 2.71336764e-01 -9.57359791e-01 2.64868200e-01
1.17275012e+00 -8.06874752e-01 5.06690264e-01 9.34212863e-01
4.98414114e-02 3.04449443e-02 -1.21679771e+00 6.37692869e-01
2.62538165e-01 -1.27038658e+00 1.30495772e-01 -3.03431481e-01
5.97942591e-01 1.27140820e-01 -2.07027376e-01 1.02974069e+00
1.09053135e+00 -7.19096124e-01 2.03223348e-01 -9.84171852e-02
8.52715671e-01 -2.40340889e-01 5.44020593e-01 4.81880367e-01
-7.56306171e-01 3.99211701e-03 -3.01992238e-01 -2.23252311e-01
4.68769580e-01 -1.87826529e-01 -1.67949879e+00 1.40340328e-01
1.40433893e-01 3.92542750e-01 -6.59541726e-01 4.37656164e-01
-1.95343941e-01 4.77557629e-01 -3.77510577e-01 -2.37816632e-01
2.09307924e-01 6.37953952e-02 5.62517464e-01 1.12712216e+00
6.58897832e-02 -8.05994198e-02 5.57103336e-01 8.13379526e-01
-3.97182643e-01 3.23246270e-01 -1.12557447e+00 -1.87274069e-01
8.12287390e-01 1.15511322e+00 -3.97508711e-01 -5.91738760e-01
-3.96132320e-01 1.25760233e+00 5.87333202e-01 3.81354004e-01
-3.89077514e-01 -1.98847383e-01 5.85280716e-01 -4.28915262e-01
-3.12721282e-01 -5.69797643e-02 3.32703292e-02 -1.54733253e+00
-2.20008910e-01 -1.25092876e+00 4.09548432e-01 -7.05367446e-01
-1.65177536e+00 8.76581371e-01 2.17243999e-01 -8.91167879e-01
-1.26115620e+00 -3.50823641e-01 -5.74094892e-01 9.09739971e-01
-1.04904866e+00 -1.12098444e+00 -4.77081649e-02 4.07970250e-01
1.37552214e+00 -7.58278310e-01 1.20528495e+00 6.93694800e-02
-3.02632719e-01 8.84327650e-01 2.93211527e-02 1.16706513e-01
1.03806615e+00 -1.05970263e+00 8.85554910e-01 6.81265533e-01
3.80846471e-01 6.79319918e-01 8.36522639e-01 -7.33985305e-01
-9.68309104e-01 -9.53329265e-01 1.07130265e+00 -6.27239883e-01
7.06764221e-01 -5.56810200e-01 -8.83777380e-01 9.43347633e-01
5.67917109e-01 -4.78932589e-01 9.68534648e-01 4.57078516e-01
-4.13468748e-01 4.15689856e-01 -1.25817013e+00 8.13177407e-01
9.38831389e-01 -6.89075232e-01 -1.00151885e+00 5.31990469e-01
1.23990321e+00 -3.95736873e-01 -5.80729842e-01 1.39811844e-01
3.58793497e-01 -4.43644792e-01 5.45532823e-01 -1.04988587e+00
1.70891553e-01 1.96024060e-01 -2.51894623e-01 -1.84222436e+00
-4.57760356e-02 -5.49480200e-01 4.05188918e-01 1.30401266e+00
9.93760407e-01 -5.94263494e-01 7.15243697e-01 1.11745012e+00
-2.00563923e-01 -1.55328721e-01 -1.03693295e+00 -8.51820052e-01
2.42786139e-01 -1.61482811e-01 5.47881842e-01 1.05691850e+00
7.66433477e-01 1.41829407e+00 -3.99207026e-01 -4.30967301e-01
1.94707721e-01 8.50374475e-02 1.16762769e+00 -8.43188584e-01
-4.12345320e-01 -1.46217495e-01 7.62937590e-02 -1.11979747e+00
8.27725351e-01 -1.15259242e+00 2.97258943e-01 -1.28263521e+00
9.88589004e-02 -7.68634319e-01 2.70768255e-01 7.71861792e-01
-3.52281153e-01 2.96318308e-02 3.16065818e-01 2.02551991e-01
-4.65194196e-01 9.60775077e-01 6.05860054e-01 -2.72767186e-01
-6.63985968e-01 -8.54791924e-02 -7.64972210e-01 4.88783479e-01
7.30821252e-01 -4.03676808e-01 -7.98407376e-01 -4.25856620e-01
-4.29508954e-01 3.08923483e-01 -2.13383347e-01 -7.31096864e-01
1.84580591e-02 -2.81667590e-01 -2.11411998e-01 -1.13416888e-01
5.98052502e-01 -4.41689700e-01 -6.98924884e-02 3.22222561e-01
-9.45652664e-01 1.65899366e-01 3.70641291e-01 3.39540243e-01
-2.54843026e-01 -6.18788421e-01 7.42104471e-01 -3.07135373e-01
-9.12906408e-01 -2.20059887e-01 -5.78055143e-01 4.39175040e-01
8.03167045e-01 1.45697653e-01 -6.12656355e-01 -8.72290552e-01
-7.13937283e-01 4.32816952e-01 7.51251578e-01 9.99396563e-01
4.45904136e-01 -1.42292225e+00 -8.53638709e-01 2.55095184e-01
6.34270370e-01 -5.82486212e-01 -1.16019003e-01 9.88004357e-03
-1.92837253e-01 5.75372458e-01 -2.58217663e-01 -4.68751162e-01
-1.12685406e+00 2.81746715e-01 4.29640561e-01 -4.65093315e-01
-3.08063954e-01 7.49202907e-01 2.87360400e-02 -1.15127909e+00
3.32684703e-02 1.08619057e-01 -1.11692637e-01 -4.60493518e-03
4.76296335e-01 -1.66522592e-01 -2.58483030e-02 -6.37653589e-01
-6.98107257e-02 -5.11526227e-01 -4.93617177e-01 -9.71926630e-01
9.25267518e-01 -3.95608962e-01 2.27567717e-01 8.41813982e-01
1.07285666e+00 -4.29561287e-02 -7.14348555e-01 -6.35065734e-01
1.95192784e-01 -4.26496148e-01 -5.67821980e-01 -9.27996814e-01
-3.52087766e-01 5.22628844e-01 5.50505340e-01 6.95057362e-02
3.98909181e-01 1.88785836e-01 8.29000652e-01 9.63297606e-01
6.92933381e-01 -9.84198213e-01 4.04452950e-01 1.05135226e+00
8.78228247e-01 -1.58176279e+00 -5.60823798e-01 -1.42235324e-01
-1.17299032e+00 6.73308253e-01 1.02252364e+00 3.09714586e-01
4.17115778e-01 4.77706119e-02 6.32242739e-01 6.21070005e-02
-1.12535632e+00 -1.75789863e-01 -2.03524053e-01 6.19882762e-01
7.48632789e-01 -2.98104137e-02 -4.43916023e-01 4.01465476e-01
-5.39622664e-01 -2.31803313e-01 8.23049366e-01 1.01003313e+00
-3.27888548e-01 -1.65155232e+00 -2.67106686e-02 6.39110386e-01
2.00254265e-02 -4.57125396e-01 -7.39598155e-01 5.18037558e-01
-4.59744751e-01 1.27903056e+00 4.49155411e-03 -3.42824131e-01
3.04336607e-01 8.02224994e-01 6.06953241e-02 -1.42397034e+00
-6.86253846e-01 -2.96094507e-01 6.94823623e-01 -2.01831296e-01
-1.59413084e-01 -6.34444475e-01 -1.06725800e+00 3.14473026e-02
-3.93260449e-01 6.80724263e-01 4.10097331e-01 9.83210206e-01
5.00496328e-01 -3.80481197e-03 8.33114743e-01 -5.19911110e-01
-7.23111272e-01 -1.42011380e+00 -1.21632241e-01 7.87496507e-01
1.87725797e-02 -6.68301702e-01 -2.79142052e-01 1.72387600e-01] | [12.678009986877441, 8.090580940246582] |
fb212340-9e91-4a41-953e-9d4bf20b0483 | multiple-news-headlines-generation-using-page | null | null | https://aclanthology.org/W19-8612 | https://aclanthology.org/W19-8612.pdf | Multiple News Headlines Generation using Page Metadata | Multiple headlines of a newspaper article have an important role to express the content of the article accurately and concisely. A headline depends on the content and intent of their article. While a single headline expresses the whole corresponding article, each of multiple headlines expresses different information individually. We suggest automatic generation method of such a diverse multiple headlines in a newspaper. Our generation method is based on the Pointer-Generator Network, using page metadata on a newspaper which can change headline generation behavior. This page metadata includes headline location, headline size, article page number, etc. In a previous related work, ensemble of three different generation models was performed to obtain a single headline, where each generation model generates a single headline candidate. In contrast, we use a single model to generate multiple headlines. We conducted automatic evaluations for generated headlines. The results show that our method improved ROUGE-1 score by 4.32 points higher than baseline. These results suggest that our model using page metadata can generate various multiple headlines for an article In better performance. | ['Kango Iwama', 'Yoshinobu Kano'] | 2019-10-01 | null | null | null | ws-2019-10 | ['headline-generation'] | ['natural-language-processing'] | [ 1.23073123e-01 3.18724692e-01 -2.39377260e-01 -1.75685227e-01
-1.16613412e+00 -7.97721565e-01 1.11766887e+00 3.85676235e-01
-1.66400373e-01 1.17059517e+00 1.00428402e+00 -5.44623472e-02
1.96808502e-01 -8.44680786e-01 -9.15004909e-01 -1.55680865e-01
1.98736742e-01 4.24470961e-01 4.67751980e-01 -5.58222532e-01
8.72162998e-01 1.53832436e-01 -1.19574308e+00 6.36152625e-01
8.31568956e-01 4.94868219e-01 5.71282268e-01 1.03975487e+00
-4.18088228e-01 7.51059234e-01 -1.15758514e+00 -2.58403629e-01
1.49403261e-02 -6.33700430e-01 -5.26173413e-01 9.99759510e-02
8.96071613e-01 -2.40833983e-01 -6.56430945e-02 1.01798069e+00
4.74237293e-01 -1.33302301e-01 8.65112185e-01 -1.12454808e+00
-8.58827889e-01 1.30238008e+00 -7.74442196e-01 1.62423566e-01
8.74630451e-01 -1.63109973e-01 1.26384854e+00 -6.10697567e-01
1.03756738e+00 1.09669173e+00 5.00534594e-01 1.86629757e-01
-9.34029877e-01 -3.94058287e-01 1.29483547e-02 -9.04838070e-02
-8.50183785e-01 -2.31577337e-01 5.49953878e-01 -3.51812720e-01
7.26352632e-01 6.18143380e-01 4.54096109e-01 1.06402159e+00
8.07133198e-01 7.82464623e-01 1.05819857e+00 -6.48744226e-01
-1.93739444e-01 2.52910316e-01 5.61410785e-01 4.72648889e-01
3.49290997e-01 -3.30403209e-01 -3.64672095e-01 -1.13359258e-01
3.85409057e-01 -3.52591634e-01 -2.57388473e-01 2.64230043e-01
-1.47593665e+00 9.93516684e-01 1.98410660e-01 1.82999119e-01
-4.04987395e-01 -7.26919919e-02 3.23131502e-01 6.09525770e-04
1.44834280e-01 8.95484209e-01 -8.79368931e-02 -5.91756999e-02
-1.21517062e+00 8.50217760e-01 1.28628707e+00 1.33216572e+00
7.46125937e-01 -3.87810588e-01 -1.10228276e+00 8.15450728e-01
6.63761720e-02 7.28911102e-01 7.17240691e-01 -6.49631917e-01
9.15687501e-01 5.27368844e-01 4.56559986e-01 -1.45131230e+00
-4.12325203e-01 -7.34792471e-01 -5.98105371e-01 -2.03462675e-01
-4.18065712e-02 -4.54629987e-01 -6.20067060e-01 1.23453593e+00
-8.51132646e-02 -5.69085062e-01 -2.67615348e-01 4.91653383e-01
9.16295409e-01 1.25146675e+00 -2.70950615e-01 -2.69806117e-01
1.39760089e+00 -1.40138042e+00 -9.79491770e-01 -1.72815740e-01
5.38671672e-01 -1.47283459e+00 1.26314998e+00 1.94183692e-01
-1.02979517e+00 -4.86326158e-01 -1.30453658e+00 -6.19801320e-03
-3.54645550e-01 4.35167074e-01 -7.19573051e-02 2.23698288e-01
-1.03826046e+00 2.04184994e-01 1.53342672e-02 -4.93643880e-01
-3.18850577e-01 -3.01104665e-01 -1.65205821e-01 3.26735884e-01
-1.40336287e+00 6.97110474e-01 4.71081078e-01 -5.06254017e-01
-3.86516064e-01 -7.24893868e-01 -6.06566072e-01 -1.31255776e-01
2.07721159e-01 -7.21758962e-01 1.58049977e+00 -3.29674602e-01
-1.13142371e+00 2.09746554e-01 -3.96700501e-02 -6.35888696e-01
5.51601231e-01 -3.86779845e-01 -7.61276364e-01 3.77257168e-02
5.47929287e-01 6.13689899e-01 6.19740248e-01 -1.27328849e+00
-1.02257001e+00 1.35899842e-01 -6.43572435e-02 3.62670004e-01
-1.90228030e-01 1.56425461e-02 -5.38056433e-01 -1.04077399e+00
-4.94565189e-01 -9.09344673e-01 5.79456575e-02 -6.84866965e-01
-1.12901413e+00 -1.55210078e-01 7.71438777e-01 -7.56795168e-01
2.14761257e+00 -1.64410985e+00 -5.36875188e-01 9.13524106e-02
1.04714997e-01 -2.64552593e-01 -5.26084378e-02 9.27659750e-01
3.45029414e-01 5.24428308e-01 7.09648803e-02 2.20924050e-01
2.10128888e-01 -4.05684590e-01 -6.22172534e-01 -8.99149626e-02
-1.63839206e-01 8.66199970e-01 -6.89210117e-01 -6.47305012e-01
-4.05584127e-01 8.85954797e-02 -5.78048825e-01 1.14187814e-01
-5.79720140e-01 -3.73330623e-01 -3.67083639e-01 1.04351252e-01
6.39892578e-01 -5.23477197e-01 -2.69705474e-01 -5.02997816e-01
-4.56848770e-01 3.95945191e-01 -1.04616773e+00 1.06051230e+00
-4.09675330e-01 8.42645943e-01 -5.28423429e-01 2.90968955e-01
1.06742513e+00 6.92300722e-02 4.28544693e-02 -7.77753949e-01
-2.95056581e-01 6.97320467e-03 -2.52577484e-01 -3.81218165e-01
1.35915005e+00 3.71081710e-01 -6.06210828e-01 1.01684499e+00
-4.61626053e-01 -8.80906545e-03 9.37322974e-01 7.10836589e-01
1.00564015e+00 -9.11782030e-03 3.77514303e-01 -7.41395578e-02
2.95189321e-01 2.15276226e-01 -3.75106148e-02 1.32746923e+00
2.98326045e-01 9.89540517e-01 5.48894048e-01 -4.26976895e-03
-1.40846336e+00 -6.38120472e-01 7.77401775e-02 1.17374921e+00
-6.25996059e-03 -7.81632364e-01 -9.74202633e-01 -6.77139223e-01
-2.27559000e-01 1.40537930e+00 -6.37432873e-01 2.04445332e-01
-8.57240319e-01 -6.83592916e-01 5.75922310e-01 -7.68116340e-02
5.29394031e-01 -1.17977405e+00 -3.86410862e-01 4.31942970e-01
-6.05871022e-01 -9.49407399e-01 -1.20256412e+00 -2.60288119e-01
-3.12874407e-01 -7.44481444e-01 -9.88174438e-01 -7.85519481e-01
6.51349962e-01 2.11543217e-01 1.23040617e+00 -2.22810403e-01
1.89886302e-01 -5.56441061e-02 -5.85940480e-01 -4.67422754e-01
-7.67686546e-01 5.67622483e-01 -3.08639050e-01 -3.30420554e-01
5.59183657e-02 8.12762901e-02 -5.64076424e-01 1.32070497e-01
-1.11562276e+00 3.43924373e-01 7.84823000e-01 5.54884672e-01
1.75858796e-01 -2.53465865e-02 4.94185597e-01 -1.51981151e+00
1.57290232e+00 -7.10562885e-01 -2.76635706e-01 4.78419155e-01
-8.18594575e-01 4.58355635e-01 7.96833098e-01 -1.35295197e-01
-1.15633035e+00 -3.68001074e-01 -1.46447262e-02 6.20142937e-01
6.72248099e-03 7.51897216e-01 2.31529087e-01 5.47048032e-01
8.14835727e-01 4.24661785e-01 -1.48945808e-01 -5.99582732e-01
4.65145737e-01 7.43115008e-01 5.00515342e-01 -2.07222164e-01
7.56960392e-01 -1.58297524e-01 -4.33760703e-01 -5.13432801e-01
-9.38289523e-01 -4.75031912e-01 -3.16853881e-01 -3.55262309e-01
4.11166251e-01 -6.17356718e-01 9.53337550e-02 4.27102953e-01
-1.38251400e+00 2.19635472e-01 1.80035397e-01 8.35473165e-02
-2.03029990e-01 4.23719227e-01 -4.79782254e-01 -1.10817373e-01
-6.59907341e-01 -1.00464606e+00 9.85417843e-01 1.60994649e-01
-6.72668040e-01 -8.60257208e-01 4.45167959e-01 2.28492588e-01
4.21782881e-01 2.54808873e-01 8.57411206e-01 -9.45361733e-01
-2.10628927e-01 -6.18022740e-01 -1.64278641e-01 -1.40563250e-01
1.99105933e-01 4.25114125e-01 -2.40695149e-01 4.41806428e-02
-3.61051291e-01 1.79804951e-01 5.22092700e-01 2.29472384e-01
6.64971352e-01 -1.40076101e+00 -3.86031836e-01 1.48423806e-01
1.33748043e+00 1.02236448e-02 6.24182463e-01 1.03071773e+00
6.26052856e-01 3.73162717e-01 5.43017209e-01 7.30185151e-01
5.91219842e-01 7.27316141e-01 4.70784940e-02 -7.73663074e-02
-4.68145162e-01 -6.23069167e-01 6.34181380e-01 9.95691180e-01
5.96825898e-01 -9.21315551e-01 -6.70710623e-01 2.92447448e-01
-1.46886170e+00 -1.27674901e+00 -5.01731932e-01 1.86813021e+00
1.22347295e+00 5.49702287e-01 3.74098599e-01 -1.40475318e-01
1.06595552e+00 3.21568400e-01 -1.00075044e-01 -3.67257208e-01
-3.98596644e-01 -5.17255902e-01 7.75409758e-01 6.47436857e-01
-7.30542243e-01 7.04142213e-01 6.90178728e+00 7.23672569e-01
-1.14125848e+00 -4.50089067e-01 5.11724412e-01 7.60172531e-02
-6.88534141e-01 -1.03916645e-01 -1.65459895e+00 9.43984687e-01
8.75478566e-01 -8.83328021e-01 -2.71362960e-02 7.14533389e-01
5.84574878e-01 -1.62035286e-01 -8.17863882e-01 4.78287190e-01
4.64324892e-01 -1.66959095e+00 8.59599173e-01 -3.60338017e-02
1.08351529e+00 -2.62463421e-01 1.03615606e-02 2.03935668e-01
4.02919918e-01 -5.91077328e-01 9.26833928e-01 8.39372694e-01
5.16907513e-01 -6.76517010e-01 7.44100511e-01 4.13087547e-01
-7.09301174e-01 3.54460686e-01 -1.39069349e-01 2.00165868e-01
5.21726251e-01 7.88764119e-01 -1.29696500e+00 1.55462787e-01
2.24926099e-01 3.26004177e-01 -1.04232919e+00 1.27625179e+00
-1.51852682e-01 6.05054677e-01 -2.84538530e-02 -7.73894012e-01
5.51309884e-01 1.47938237e-01 7.66394317e-01 1.64246118e+00
6.30233288e-01 -4.67549592e-01 1.14716552e-02 6.52016163e-01
-2.97405809e-01 3.86387080e-01 -4.87439632e-01 -1.77470878e-01
7.76444793e-01 1.50886786e+00 -4.97122198e-01 -5.20569444e-01
-2.23031193e-01 7.55898476e-01 1.35373011e-01 2.73404360e-01
-1.00328124e+00 -9.92083669e-01 -2.80859441e-01 6.63432300e-01
2.30835587e-01 -3.32859792e-02 -3.27757806e-01 -7.80241251e-01
1.37965336e-01 -1.03865278e+00 2.03011692e-01 -1.09190989e+00
-1.06708086e+00 9.34135020e-01 8.50227848e-02 -1.39097714e+00
-6.17931545e-01 -2.94847228e-02 -7.73649812e-01 7.70164907e-01
-1.19998908e+00 -8.28138053e-01 -2.67386854e-01 1.71362199e-02
8.91078591e-01 -2.83905834e-01 4.53455895e-01 -7.66266882e-02
-3.93685937e-01 6.56649530e-01 2.28772670e-01 1.58714190e-01
1.11181056e+00 -1.39934707e+00 6.61034048e-01 9.44273233e-01
8.34225491e-02 9.10416782e-01 1.18893456e+00 -1.06741190e+00
-1.04711628e+00 -1.21703625e+00 1.48813117e+00 -2.00761378e-01
8.42708409e-01 -6.98134452e-02 -6.66673839e-01 4.89365727e-01
9.14565742e-01 -9.66656625e-01 7.02998936e-01 -2.07497403e-01
7.16804806e-03 -5.62578291e-02 -7.69264698e-01 8.58413339e-01
3.12955558e-01 7.73086548e-02 -8.70630324e-01 6.60478890e-01
1.01079798e+00 -3.41576040e-01 -6.40776753e-01 -6.80579543e-02
4.32306886e-01 -9.30851102e-01 3.78334850e-01 -1.17973715e-01
1.05263770e+00 -4.90994185e-01 3.07832062e-01 -1.49722219e+00
-7.81969726e-01 -7.10175276e-01 -2.23692581e-01 1.41489160e+00
1.05256093e+00 -3.28649580e-01 2.46916518e-01 4.77936208e-01
-2.32426837e-01 -4.50379670e-01 -1.10849559e-01 -4.71297324e-01
-2.96405852e-01 8.66757855e-02 8.71321082e-01 3.53032440e-01
2.28124425e-01 9.49750841e-01 -4.78991687e-01 -3.05776685e-01
4.77897704e-01 1.84698701e-01 9.12740827e-01 -8.96844685e-01
-7.98754692e-02 -6.47033274e-01 3.34445685e-01 -1.32134783e+00
-2.94296622e-01 -9.98495698e-01 1.66171372e-01 -1.95910895e+00
4.61949080e-01 -1.87616140e-01 2.10904017e-01 1.56583339e-01
-3.08329970e-01 1.16694033e-01 2.99435705e-01 5.41156590e-01
-8.19137275e-01 1.82644650e-01 1.51235163e+00 -1.44442186e-01
-1.56007409e-01 4.04043235e-02 -1.08480334e+00 3.81564021e-01
8.19695771e-01 -3.63566935e-01 -2.94621021e-01 -2.29180560e-01
6.64010763e-01 5.69842421e-02 -1.30357042e-01 -1.12836468e+00
5.80340385e-01 -1.51223376e-01 4.19088751e-01 -1.01522040e+00
-3.43128920e-01 -1.93556219e-01 6.10122196e-02 2.27194250e-01
-9.36850965e-01 6.92207754e-01 -8.99939463e-02 4.65702474e-01
-2.72237122e-01 -5.82952440e-01 3.19940090e-01 -3.98285270e-01
-6.90277636e-01 4.98252660e-02 -6.03175223e-01 3.39791000e-01
8.52430224e-01 -1.57952487e-01 -7.48794556e-01 -6.97905064e-01
-1.39211580e-01 2.47632116e-02 5.15226483e-01 7.13964283e-01
4.59214330e-01 -1.42660081e+00 -1.10692430e+00 -2.62722839e-02
2.25755259e-01 -5.45781136e-01 -3.03187966e-01 2.98769444e-01
-9.15421546e-01 7.65499175e-01 -2.96320438e-01 -1.10290937e-01
-1.08401465e+00 1.84602767e-01 -2.95431107e-01 -5.69198728e-01
-5.57882071e-01 4.22686547e-01 -2.22832095e-02 -1.06525198e-01
8.47447664e-02 -1.97492048e-01 -8.04429770e-01 3.61567408e-01
9.87975538e-01 6.14354193e-01 1.08552635e-01 -7.34092236e-01
2.10587144e-01 3.44498366e-01 -7.30428100e-01 -3.81640762e-01
1.07073307e+00 -4.01887625e-01 -2.23898441e-01 4.95320350e-01
1.34360111e+00 5.92526793e-01 -7.26193488e-01 1.59954280e-01
5.36301993e-02 -9.43251029e-02 1.00360606e-02 -1.15817106e+00
-5.16157687e-01 -8.18297938e-02 -1.15665659e-01 5.57597578e-01
6.69256687e-01 -1.41750216e-01 1.22190964e+00 2.83173531e-01
1.84476957e-01 -1.38857746e+00 8.03183839e-02 9.06486928e-01
1.45622027e+00 -9.95343983e-01 1.96742520e-01 -3.79802883e-02
-9.42826748e-01 1.40639818e+00 6.20561004e-01 5.92060275e-02
3.54803801e-01 2.77758747e-01 1.57887802e-01 -1.96992248e-01
-8.58050406e-01 4.95086938e-01 6.33203804e-01 1.53534904e-01
8.26516151e-01 -4.40978212e-03 -8.19956958e-01 5.16204774e-01
-1.06840575e+00 -1.71641871e-01 1.04879618e+00 7.66629279e-01
-8.85280728e-01 -8.53576839e-01 -5.18758893e-01 9.68158841e-01
-4.64051902e-01 -4.92239267e-01 -4.81197834e-01 6.05721772e-01
-4.04263020e-01 8.90232861e-01 7.24712685e-02 -3.86543512e-01
2.23046511e-01 -1.77068427e-01 -1.74141034e-01 -6.99122667e-01
-6.94602847e-01 2.52693355e-01 3.78641039e-01 -1.43523842e-01
-1.29573429e-02 -5.81023395e-01 -1.22127306e+00 -4.44203138e-01
-1.73289150e-01 5.30407727e-01 6.76698148e-01 4.21700627e-01
4.37396556e-01 4.18520689e-01 7.82471120e-01 -4.04807359e-01
-6.23632371e-01 -1.32847750e+00 -4.34206218e-01 4.78745967e-01
3.64300132e-01 -1.63882915e-02 -5.63431919e-01 3.92869979e-01] | [12.2688570022583, 9.310391426086426] |
ef9ab5ba-4369-4024-89a4-865ba42752d4 | sparsity-and-coefficient-permutation-based | 2305.12986 | null | https://arxiv.org/abs/2305.12986v1 | https://arxiv.org/pdf/2305.12986v1.pdf | Sparsity and Coefficient Permutation Based Two-Domain AMP for Image Block Compressed Sensing | The learned denoising-based approximate message passing (LDAMP) algorithm has attracted great attention for image compressed sensing (CS) tasks. However, it has two issues: first, its global measurement model severely restricts its applicability to high-dimensional images, and its block-based measurement method exhibits obvious block artifacts; second, the denoiser in the LDAMP is too simple, and existing denoisers have limited ability in detail recovery. In this paper, to overcome the issues and develop a high-performance LDAMP method for image block compressed sensing (BCS), we propose a novel sparsity and coefficient permutation-based AMP (SCP-AMP) method consisting of the block-based sampling and the two-domain reconstruction modules. In the sampling module, SCP-AMP adopts a discrete cosine transform (DCT) based sparsity strategy to reduce the impact of the high-frequency coefficient on the reconstruction, followed by a coefficient permutation strategy to avoid block artifacts. In the reconstruction module, a two-domain AMP method with DCT domain noise correction and pixel domain denoising is proposed for iterative reconstruction. Regarding the denoiser, we proposed a multi-level deep attention network (MDANet) to enhance the texture details by employing multi-level features and multiple attention mechanisms. Extensive experiments demonstrated that the proposed SCP-AMP method achieved better reconstruction accuracy than other state-of-the-art BCS algorithms in terms of both visual perception and objective metrics. | ['Shuhao Bi', 'Huake Wang', 'Xingsong Hou', 'Junhui Li'] | 2023-05-22 | null | null | null | null | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 5.27319670e-01 -4.57582474e-01 8.07004049e-02 -1.33091792e-01
-7.90816963e-01 4.07005489e-01 2.96525955e-01 -1.94207072e-01
-1.52537182e-01 4.04915452e-01 5.02133548e-01 -3.27877253e-02
-2.10876137e-01 -7.78144598e-01 -6.93105876e-01 -1.04460859e+00
2.10435212e-01 -3.82382959e-01 3.12383592e-01 -1.93364203e-01
2.38304988e-01 2.08093271e-01 -1.25830507e+00 5.03632486e-01
1.24229956e+00 1.41265428e+00 6.84204757e-01 3.49190921e-01
1.00449929e-02 9.98340309e-01 -2.16810867e-01 -7.54613057e-02
2.78203070e-01 -7.87755251e-01 -1.46400318e-01 3.65093760e-02
2.20107749e-01 -7.23283350e-01 -8.27654421e-01 1.35153437e+00
7.18933284e-01 1.94912460e-02 3.36404651e-01 -8.39132547e-01
-8.70620012e-01 4.11544442e-01 -8.48591149e-01 2.97681391e-01
2.06331834e-01 1.77575052e-01 3.41568679e-01 -1.02257311e+00
2.16868818e-01 1.26085198e+00 9.10997152e-01 1.62085935e-01
-1.15312326e+00 -7.38097847e-01 -2.16323748e-01 7.39348412e-01
-1.46202898e+00 -5.71930945e-01 1.12435961e+00 -1.09869130e-01
7.09013224e-01 1.86534882e-01 7.88053572e-01 7.48031378e-01
5.65568686e-01 7.22861290e-01 1.35926545e+00 -3.22631598e-01
2.84805417e-01 -4.64364529e-01 -2.80099928e-01 4.99476969e-01
1.51097268e-01 1.63131729e-01 -5.59323192e-01 4.78840731e-02
1.12938750e+00 2.46747121e-01 -5.82097411e-01 -2.76354492e-01
-1.24857783e+00 6.55307353e-01 6.40916288e-01 5.44532537e-01
-8.52774858e-01 2.67467707e-01 3.53737473e-01 1.41672075e-01
3.97645473e-01 2.73047443e-02 -4.11426090e-02 2.20507920e-01
-1.22897530e+00 1.21355213e-01 3.78791988e-01 4.10936713e-01
4.95165020e-01 3.31504881e-01 -3.97198319e-01 1.03296101e+00
3.30383986e-01 5.08092582e-01 8.53253603e-01 -1.14117074e+00
3.16598594e-01 1.65356830e-01 -6.52226061e-02 -1.69031656e+00
-5.34770079e-02 -5.93575895e-01 -1.80503953e+00 1.21101245e-01
-2.32706472e-01 3.64852726e-01 -7.64818430e-01 1.61463678e+00
1.64320201e-01 6.05821490e-01 -1.33203343e-01 1.27492332e+00
8.24302852e-01 8.63180280e-01 -5.59629463e-02 -6.24487996e-01
1.12831438e+00 -8.61136317e-01 -1.04442096e+00 -1.73551962e-01
-7.13951960e-02 -7.23418474e-01 6.27598047e-01 6.41582668e-01
-1.18659973e+00 -9.94849443e-01 -1.28850126e+00 -2.51011223e-01
3.06546479e-01 1.78326130e-01 3.28873903e-01 3.31636339e-01
-1.00973082e+00 6.42057002e-01 -9.26437140e-01 2.39729919e-02
6.84215665e-01 -6.76769856e-03 -2.49479458e-01 -6.63009286e-01
-1.14162004e+00 5.47843158e-01 9.14266407e-02 3.38972658e-01
-1.01121891e+00 -4.99680042e-01 -9.27319407e-01 3.27366740e-01
1.20020866e-01 -6.99524641e-01 7.47142553e-01 -1.00097466e+00
-1.60400581e+00 3.43306690e-01 -2.50831395e-01 -5.16426504e-01
3.32320124e-01 -1.60390034e-01 -5.53531766e-01 5.01469731e-01
2.36430377e-01 4.32407588e-01 1.46331096e+00 -1.03024721e+00
-1.65658012e-01 -3.18270057e-01 -4.58093017e-01 2.97118992e-01
-1.65974975e-01 -4.21877444e-01 -5.04206657e-01 -1.22351706e+00
7.67987907e-01 -3.15754533e-01 -2.73158640e-01 2.55641162e-01
-5.62830381e-02 3.12993646e-01 8.40269923e-01 -1.26425827e+00
1.55523086e+00 -2.54552746e+00 2.90755868e-01 4.52367924e-02
2.78725684e-01 5.48778951e-01 -3.44897687e-01 3.09123129e-01
-1.94123670e-01 -2.20387965e-01 -5.38817644e-01 -2.18372568e-01
-4.11004663e-01 -1.31332977e-02 -2.48281553e-01 5.65974772e-01
2.58744564e-02 7.32002914e-01 -8.59375834e-01 -3.76879483e-01
5.25707603e-01 7.43396223e-01 -6.18991137e-01 1.25438318e-01
1.41000509e-01 4.96280670e-01 -4.33342427e-01 5.63382447e-01
1.20032609e+00 -1.70905069e-01 1.22883454e-01 -8.20945740e-01
-7.68375024e-02 1.01227798e-02 -1.32548201e+00 1.87243521e+00
-5.39697647e-01 4.51555669e-01 5.50667882e-01 -1.26948404e+00
8.01813841e-01 1.93962425e-01 5.87313533e-01 -1.08025336e+00
7.91898295e-02 4.25218940e-01 -1.28148437e-01 -6.38797522e-01
2.05682933e-01 -2.78755456e-01 4.55723226e-01 2.46471837e-02
-2.16102228e-01 -1.57162309e-01 -1.89259157e-01 1.22507080e-01
1.26807606e+00 -9.33226570e-03 4.36462760e-01 -2.51608789e-01
7.79002845e-01 -3.22487682e-01 7.24876642e-01 6.00320220e-01
-3.31760943e-01 9.77052689e-01 2.66423374e-02 -3.37972283e-01
-9.40890074e-01 -7.54534543e-01 -1.79519951e-01 4.38298434e-01
6.03586197e-01 -3.15007955e-01 -6.47143245e-01 -6.03440404e-02
-1.48792222e-01 4.61235046e-01 -3.55812490e-01 -4.11457330e-01
-5.95785975e-01 -6.94837153e-01 1.33555010e-01 2.25601301e-01
1.30321360e+00 -1.00510406e+00 -2.85176605e-01 5.81725955e-01
-6.41342938e-01 -9.18388665e-01 -6.19939983e-01 -1.59529105e-01
-9.07019913e-01 -9.00836229e-01 -9.14621592e-01 -9.83226061e-01
5.53927183e-01 7.56253958e-01 6.11270308e-01 1.31106317e-01
-5.22590727e-02 2.06041429e-02 -4.63356644e-01 3.13366130e-02
-2.13718951e-01 -5.08700669e-01 -2.32068747e-01 4.23775554e-01
1.62572473e-01 -8.56939554e-01 -1.14116669e+00 2.55705088e-01
-1.24718142e+00 2.65159070e-01 1.11310387e+00 1.20340562e+00
8.10192347e-01 3.97139758e-01 5.79498291e-01 -3.49968672e-01
6.70632184e-01 -4.02037352e-01 -2.42987633e-01 -8.15574452e-02
-6.13843620e-01 -3.00072491e-01 7.87211895e-01 -2.74719298e-01
-1.00502479e+00 -5.29780798e-02 -4.63619858e-01 -6.37026906e-01
1.61342591e-01 6.55811965e-01 -2.82475114e-01 -3.03779393e-01
3.78895283e-01 9.89250600e-01 3.89660090e-01 -5.87516665e-01
-8.33341181e-02 7.19942570e-01 6.54799402e-01 -1.24323100e-01
4.62862909e-01 5.19417465e-01 4.57028076e-02 -8.65187764e-01
-6.52916253e-01 -2.22890854e-01 -3.84195074e-02 2.48253066e-02
8.29524517e-01 -1.34957325e+00 -5.73263824e-01 9.82257843e-01
-1.06113434e+00 -2.45266512e-01 -1.71569526e-01 6.87880754e-01
-5.06751716e-01 9.24248815e-01 -8.75258207e-01 -4.86679494e-01
-4.60910171e-01 -1.26949108e+00 8.43859196e-01 2.87011862e-02
2.40346536e-01 -3.10596585e-01 -2.71204591e-01 4.42726314e-01
8.98425102e-01 1.19105708e-02 8.17310452e-01 1.63944289e-01
-7.72088289e-01 -1.50397681e-02 -4.88623619e-01 8.50482106e-01
4.23168056e-02 -8.20289314e-01 -6.06928229e-01 -5.02319336e-01
7.81420350e-01 -1.71235070e-01 1.06669557e+00 9.74629939e-01
1.67676353e+00 -3.23144019e-01 -3.46361324e-02 9.94773626e-01
1.59275103e+00 3.78912948e-02 1.08871961e+00 2.95731902e-01
6.28132164e-01 -6.06614612e-02 4.57202703e-01 4.02440965e-01
1.81106284e-01 5.68302691e-01 4.18847561e-01 -7.47078434e-02
-6.60464942e-01 -2.84464717e-01 2.58327335e-01 1.16060364e+00
1.54432803e-01 -1.08439617e-01 -3.33135486e-01 5.73428512e-01
-1.72312117e+00 -9.54345942e-01 -9.10024792e-02 2.01399803e+00
8.71469259e-01 -1.59123078e-01 -4.91678029e-01 2.67650932e-01
6.82392716e-01 5.02060950e-01 -7.35066175e-01 -1.98218077e-02
-3.10492575e-01 3.10640693e-01 3.18861723e-01 3.26934010e-01
-9.53649282e-01 2.46480048e-01 5.19422150e+00 1.44645905e+00
-1.09527040e+00 3.61826271e-01 8.19045007e-01 1.84978351e-01
-1.81910276e-01 -1.05303831e-01 -1.87034547e-01 8.46976459e-01
3.88536245e-01 3.14589620e-01 7.75762737e-01 5.92200518e-01
4.13297981e-01 -1.57228649e-01 -5.25699794e-01 1.43127322e+00
2.11239159e-01 -1.29815102e+00 1.14412121e-01 -1.74894571e-01
7.50410974e-01 -1.90910995e-02 2.24106982e-02 8.02537501e-02
-3.49279344e-01 -9.05840039e-01 5.84999919e-01 6.15152299e-01
1.12473238e+00 -5.55794060e-01 7.26148069e-01 3.88722897e-01
-1.16653299e+00 -1.88992471e-01 -5.43796301e-01 1.25025045e-02
1.86248198e-01 1.17864215e+00 1.79048225e-01 6.12624705e-01
8.60633731e-01 1.05162179e+00 -7.36810938e-02 1.03715944e+00
-1.70479625e-01 5.85372686e-01 -1.94234192e-01 3.78120273e-01
9.91581455e-02 -4.21449244e-01 8.02594900e-01 8.34864736e-01
6.88714266e-01 4.97591436e-01 -1.13630772e-01 7.72161603e-01
-5.93038537e-02 -1.24121428e-01 -1.38170511e-01 3.29575896e-01
3.53545964e-01 1.05945694e+00 -4.36625332e-01 -3.46686244e-01
-3.79456460e-01 1.22960734e+00 -1.19227000e-01 3.96748662e-01
-5.37321806e-01 -3.72663230e-01 4.09753203e-01 2.82191128e-01
8.40218544e-01 -1.18850671e-01 -3.79403710e-01 -1.39991486e+00
1.85051173e-01 -1.27668154e+00 -2.23714579e-02 -9.94288146e-01
-1.27571154e+00 3.79963875e-01 -4.48080212e-01 -1.40646851e+00
3.22025567e-01 6.10657968e-03 -5.20392060e-01 9.29932475e-01
-1.87309647e+00 -9.94892836e-01 -6.29620671e-01 6.16631269e-01
6.87840700e-01 -1.01343885e-01 4.66191441e-01 7.28773117e-01
-3.80973667e-01 3.51696491e-01 3.87978762e-01 -9.99316424e-02
4.95821595e-01 -4.94975626e-01 -5.28786592e-02 1.11489415e+00
-3.57558668e-01 5.17199278e-01 5.30055225e-01 -7.97156453e-01
-1.59240961e+00 -1.15231681e+00 4.05961752e-01 6.43333316e-01
1.76571816e-01 1.18523568e-01 -1.06984472e+00 2.22149372e-01
1.34709314e-01 2.28040516e-01 1.50385842e-01 -7.35024214e-01
-1.14248477e-01 -5.29504895e-01 -1.38410950e+00 3.66530150e-01
8.78994405e-01 -4.06776875e-01 -3.35020304e-01 1.25675604e-01
5.34261942e-01 -4.09423500e-01 -7.01304197e-01 5.51303148e-01
5.00853300e-01 -1.40696633e+00 1.15430427e+00 4.55008030e-01
8.53217244e-01 -5.85822165e-01 -3.53656083e-01 -1.27877998e+00
-8.39315951e-01 -4.77113307e-01 -3.88991386e-01 8.78755927e-01
-4.72288758e-01 -6.09838247e-01 4.72595960e-01 -1.03385538e-01
-3.57816696e-01 -9.51205373e-01 -1.07770073e+00 -4.22511637e-01
-3.77686352e-01 -3.97218227e-01 5.11183858e-01 7.49321818e-01
-4.62588727e-01 1.24752838e-02 -6.24534905e-01 3.05551916e-01
8.43114078e-01 1.30115926e-01 3.67763251e-01 -5.42478859e-01
-5.77205420e-01 -3.20335388e-01 -4.33155686e-01 -1.60656822e+00
-3.17089826e-01 -6.36854827e-01 6.51166663e-02 -1.56665623e+00
3.03939193e-01 -3.04632694e-01 -3.92380416e-01 -6.64561838e-02
-1.44017994e-01 4.77355212e-01 6.37553632e-02 4.90482688e-01
-4.17211294e-01 9.78340566e-01 1.49102163e+00 -2.47355834e-01
8.19109753e-02 -3.16402107e-01 -6.96053863e-01 5.20077467e-01
3.88161987e-01 -4.42175865e-01 -2.30344877e-01 -6.86963618e-01
1.07712019e-02 3.04695070e-01 5.91203749e-01 -1.34717607e+00
4.45153296e-01 1.07827038e-01 7.65727580e-01 -7.09075511e-01
3.87568831e-01 -8.63041580e-01 3.19142550e-01 8.13440323e-01
-6.72895238e-02 -5.29696703e-01 1.61695648e-02 8.98346066e-01
-6.87967598e-01 9.13107097e-02 1.15769660e+00 -2.15845957e-01
-6.56883955e-01 3.05305809e-01 -2.83912539e-01 -3.83007407e-01
8.70651126e-01 -2.82993406e-01 2.37105992e-02 -4.77366120e-01
-3.64878148e-01 -3.16518396e-02 3.73014748e-01 -6.43358082e-02
1.15685630e+00 -1.44237041e+00 -9.66572642e-01 7.95146942e-01
-3.26724112e-01 5.49655557e-02 8.62651944e-01 1.16349852e+00
-8.07631075e-01 9.02945623e-02 -2.39673167e-01 -5.71304858e-01
-8.65730822e-01 5.40096998e-01 2.34190360e-01 -3.09673727e-01
-7.71633983e-01 7.35538006e-01 1.32806256e-01 -5.57612516e-02
3.95854563e-02 -2.05216542e-01 1.42483013e-02 -3.94220054e-01
8.02031219e-01 5.09422064e-01 1.01756424e-01 -5.60625494e-01
-1.29703522e-01 6.77711904e-01 1.20620452e-01 1.47138521e-01
1.64457357e+00 -4.90910649e-01 -4.47043270e-01 -1.25968277e-01
1.20303857e+00 -1.60108745e-01 -1.33981133e+00 -3.88292432e-01
-7.13433921e-01 -8.74511302e-01 7.69897223e-01 -6.89128816e-01
-1.35845232e+00 7.32786119e-01 8.76671076e-01 1.18355498e-01
1.82065070e+00 -4.79170918e-01 1.47902632e+00 -1.15991436e-01
2.32479274e-01 -7.26086915e-01 1.09858721e-01 1.87974215e-01
1.20553565e+00 -1.18924201e+00 3.95109683e-01 -3.89247745e-01
-2.05055267e-01 1.00016379e+00 3.64798069e-01 -4.07563448e-01
8.92237544e-01 -3.17528099e-02 -2.17093691e-01 -5.40970303e-02
-2.52915889e-01 3.34679633e-01 2.33607233e-01 5.19496262e-01
3.66033195e-03 -2.46143207e-01 -6.46098912e-01 6.08234525e-01
2.52261162e-01 4.47480828e-01 3.40382457e-01 6.95144415e-01
-4.58463997e-01 -6.68658972e-01 -5.71661711e-01 6.36532068e-01
-4.87765372e-01 -3.60112637e-01 3.95756841e-01 1.32553160e-01
3.63134593e-01 1.01793933e+00 -8.87689143e-02 -4.71736908e-01
1.48921192e-01 -5.15403390e-01 2.87351847e-01 -2.84003586e-01
-1.53321981e-01 3.41603816e-01 -4.36718315e-01 -8.96877527e-01
-5.53154707e-01 -4.70428824e-01 -8.19710076e-01 -3.98108691e-01
-2.74663508e-01 -1.83690861e-01 6.35588348e-01 7.49977648e-01
7.10162997e-01 6.26597285e-01 7.91149974e-01 -1.04781997e+00
-5.58704972e-01 -1.10611892e+00 -8.24995279e-01 4.88631666e-01
5.12057960e-01 -3.54403406e-01 -4.60177839e-01 -2.94041168e-02] | [11.224644660949707, -2.1292672157287598] |
0e3fbef7-37a1-4e95-abc4-3b4eaf02c48b | semantic-sentence-matching-via-interacting | null | null | https://aclanthology.org/2022.coling-1.78 | https://aclanthology.org/2022.coling-1.78.pdf | Semantic Sentence Matching via Interacting Syntax Graphs | Studies have shown that the sentence’s syntactic structures are important for semantic sentence matching. A typical approach is encoding each sentence’s syntactic structure into an embedding vector, which can be combined with other features to predict the final matching scores. Though successes have been observed, embedding the whole syntactic structures as one vector inevitably overlooks the fine-grained syntax matching patterns, e.g. the alignment of specific term dependencies relations in the two inputted sentences. In this paper, we formalize the task of semantic sentence matching as a problem of graph matching in which each sentence is represented as a directed graph according to its syntactic structures. The syntax matching patterns (i.e. similar syntactic structures) between two sentences, therefore, can be extracted as the sub-graph structure alignments. The proposed method, referred to as Interacted Syntax Graphs (ISG), represents two sentences’ syntactic alignments as well as their semantic matching signals into one association graph. After that, the neural quadratic assignment programming (QAP) is adapted to extract syntactic matching patterns from the association graph. In this way, the syntactic structures fully interact in a fine granularity during the matching process. Experimental results on three public datasets demonstrated that ISG can outperform the state-of-the-art baselines effectively and efficiently. The empirical analysis also showed that ISG can match sentences in an interpretable way. | ['Ji-Rong Wen', 'Zhenhua Dong', 'Jun Xu', 'Chen Xu'] | null | null | null | null | coling-2022-10 | ['graph-matching'] | ['graphs'] | [ 4.41387534e-01 2.73860157e-01 -1.46898478e-01 -9.19099689e-01
-5.02223432e-01 -3.90452296e-01 3.14983666e-01 6.02803171e-01
-2.27553084e-01 1.23724237e-01 6.10266089e-01 -2.15892747e-01
-1.22911252e-01 -8.92447770e-01 -6.63049459e-01 -4.36730832e-01
2.01940238e-01 3.53255451e-01 1.18070375e-02 -2.98516512e-01
2.65019327e-01 -1.69956520e-01 -1.44696331e+00 5.88928342e-01
9.98415649e-01 8.55568349e-01 5.26688695e-01 2.39008382e-01
-9.07502294e-01 4.97922719e-01 -4.97686297e-01 -6.32799387e-01
-4.21423391e-02 -4.75239158e-01 -9.33217168e-01 -1.63582563e-02
4.75491136e-01 3.62151533e-01 -1.44801915e-01 1.50749290e+00
2.07854077e-01 6.85632080e-02 3.84225279e-01 -1.11990941e+00
-7.88319588e-01 9.16161180e-01 -4.45604414e-01 1.54996634e-01
7.93895960e-01 -2.44153708e-01 1.81036663e+00 -8.82401943e-01
4.68512595e-01 1.65103233e+00 5.03345907e-01 2.02875867e-01
-1.14752507e+00 -4.48304564e-01 2.85286188e-01 3.16615224e-01
-9.93808568e-01 -2.03752872e-02 1.10972726e+00 -5.05896389e-01
1.29056489e+00 4.42719579e-01 5.14516234e-01 6.77867651e-01
4.79089230e-01 7.71146297e-01 7.17006207e-01 -5.62809467e-01
-1.58492222e-01 2.25994345e-02 9.59688365e-01 7.18449533e-01
8.72591063e-02 -3.19016039e-01 -2.48547941e-01 -1.19668528e-01
-1.18237242e-01 -5.00499792e-02 -3.74556631e-01 -5.38646355e-02
-1.06708109e+00 1.00110209e+00 8.04419339e-01 8.00810218e-01
-3.51642966e-01 -7.08991662e-02 5.97828805e-01 4.89827573e-01
4.12025511e-01 4.31290776e-01 -3.22454929e-01 4.69347954e-01
-3.69185477e-01 1.64748490e-01 6.80178106e-01 8.99194181e-01
1.07315910e+00 -5.13587117e-01 -5.35285950e-01 8.99301350e-01
7.01670229e-01 2.68128723e-01 7.38075316e-01 -3.54275763e-01
1.06552732e+00 1.31098950e+00 -4.37672287e-01 -1.81142998e+00
-4.60507929e-01 -2.01981038e-01 -8.43116045e-01 -5.63851893e-01
1.61288485e-01 2.17362031e-01 -3.81231785e-01 1.81523848e+00
3.02640051e-01 1.33690283e-01 2.24699944e-01 9.55699384e-01
1.24809921e+00 8.41225326e-01 2.67633945e-01 -4.60121557e-02
1.86584830e+00 -9.17075634e-01 -8.16600561e-01 -6.48925960e-01
8.40025842e-01 -8.23003411e-01 1.19719291e+00 -4.18159574e-01
-8.17718148e-01 -7.45300829e-01 -1.05620241e+00 -2.03054724e-03
-2.51558751e-01 -1.19144149e-01 5.12592196e-01 3.73215824e-01
-7.72357643e-01 7.73626626e-01 -3.09912533e-01 -3.08618098e-01
1.23240307e-01 4.28798139e-01 -4.06795442e-01 7.80032761e-03
-1.74414837e+00 8.72544289e-01 8.68953466e-01 1.70059845e-01
2.80499220e-01 -5.38765848e-01 -1.35940731e+00 4.34907824e-01
3.03425193e-01 -7.71108329e-01 9.73414481e-01 -1.17217755e+00
-9.36511755e-01 1.16977668e+00 -5.88972926e-01 -2.75513172e-01
-2.28922069e-01 2.12424487e-01 -3.98035914e-01 -1.47173718e-01
3.26613158e-01 2.94137686e-01 5.93973637e-01 -7.81867683e-01
-4.72180188e-01 -6.86739624e-01 1.81061670e-01 4.26193178e-01
-4.34021235e-01 4.30732340e-01 -5.10487735e-01 -5.02292216e-01
4.62892175e-01 -6.86677098e-01 -2.42943108e-01 -5.73172569e-01
-5.64149737e-01 -7.87058890e-01 4.74246264e-01 -8.66038084e-01
1.53630936e+00 -2.04879999e+00 5.58358967e-01 2.62437135e-01
3.68946403e-01 -3.59138288e-03 -2.29406714e-01 5.40118158e-01
-3.08759123e-01 1.76371545e-01 -3.84663880e-01 -4.43961799e-01
3.35436493e-01 1.88338712e-01 -1.07036345e-01 2.16105968e-01
1.44188151e-01 1.23280406e+00 -1.11067057e+00 -6.54921412e-01
6.52273446e-02 7.98061788e-02 -4.11115021e-01 4.07209933e-01
-1.24481507e-01 1.09585531e-01 -7.07379460e-01 2.11677611e-01
6.89658046e-01 -3.28791201e-01 5.48112094e-01 -5.32936513e-01
3.41415465e-01 5.45659006e-01 -1.01888680e+00 1.63492644e+00
-4.70729947e-01 3.91812712e-01 -2.79124171e-01 -1.29698741e+00
1.22869897e+00 1.05386354e-01 1.93829894e-01 -9.44573224e-01
1.81163266e-01 1.90153480e-01 5.37928566e-02 -5.96765459e-01
4.09294069e-01 -8.04745927e-02 -6.16674244e-01 4.57101673e-01
3.35026421e-02 1.23185247e-01 1.53032944e-01 2.58841932e-01
8.55661273e-01 -2.83072889e-01 5.55014312e-01 -3.52144212e-01
1.01149642e+00 -1.05360337e-01 6.69764578e-01 4.22621220e-01
1.04832575e-01 1.99720532e-01 5.52137494e-01 -4.71311599e-01
-7.35765338e-01 -7.94052303e-01 1.15387060e-01 8.27806175e-01
4.54941660e-01 -7.94674635e-01 -8.45360577e-01 -8.45306575e-01
9.13872868e-02 4.84177023e-01 -5.09864390e-01 -2.83584267e-01
-9.15333986e-01 -6.17351651e-01 1.96799785e-01 4.50424194e-01
2.93141723e-01 -1.40678418e+00 -1.30684063e-01 3.20020109e-01
-2.90112227e-01 -1.19195914e+00 -8.69082093e-01 -5.47437370e-02
-5.27827322e-01 -1.15437067e+00 -9.35048163e-02 -1.17409337e+00
7.48553574e-01 2.95691758e-01 1.31210935e+00 4.27227408e-01
-2.64826063e-02 -1.68993287e-02 -5.45530677e-01 -6.86246231e-02
-7.10786223e-01 -1.09356426e-01 -6.30768463e-02 3.79020214e-01
7.07634091e-01 -3.61322165e-01 -3.77204001e-01 6.10636268e-03
-7.30277479e-01 1.07307270e-01 4.63896275e-01 1.02411604e+00
6.97407603e-01 -3.89584452e-02 3.92407477e-01 -1.21728945e+00
9.50417221e-01 -5.20096302e-01 -5.41415453e-01 5.61147630e-01
-5.17210662e-01 4.06894535e-01 8.90602589e-01 2.89823618e-02
-8.30702662e-01 -2.84457253e-03 -1.16679721e-01 -2.30514035e-01
2.33979281e-02 9.67762887e-01 -5.59762418e-01 1.40055835e-01
2.80343384e-01 2.68035412e-01 -8.88826251e-02 -3.23021412e-01
3.79084438e-01 8.91193211e-01 2.67804146e-01 -4.64870632e-01
7.39631057e-01 -2.83716023e-01 -6.66286796e-03 -5.05598247e-01
-1.01905715e+00 -6.57204211e-01 -7.35331714e-01 3.37280594e-02
1.07838440e+00 -4.31299895e-01 -6.56166732e-01 2.40626454e-01
-1.57165551e+00 3.51045787e-01 9.88751203e-02 3.29581589e-01
-1.85537010e-01 8.87145579e-01 -4.86870289e-01 -4.07218069e-01
-5.53047895e-01 -1.26360941e+00 1.20716929e+00 2.93686450e-01
-4.64815766e-01 -1.23837268e+00 1.75457478e-01 5.43435335e-01
4.68895957e-02 -1.07578805e-03 1.30931008e+00 -9.35290933e-01
-4.00901794e-01 -2.13163361e-01 -3.64222169e-01 2.44782150e-01
2.15361550e-01 -3.33779126e-01 -7.62267590e-01 -2.63019979e-01
1.06353842e-01 1.70984507e-01 6.96566820e-01 1.09711736e-01
8.72688115e-01 -4.49259251e-01 -3.53645205e-01 4.47167337e-01
1.31259453e+00 1.19405970e-01 3.82840484e-01 4.01452899e-01
8.96280229e-01 1.03824937e+00 6.79237068e-01 -1.29139766e-01
5.31012535e-01 1.02833378e+00 3.18423629e-01 4.51241955e-02
-5.19473031e-02 -3.66226017e-01 4.10829723e-01 1.44391799e+00
5.45610905e-01 -1.97466522e-01 -7.46361554e-01 1.40321538e-01
-2.06003618e+00 -9.34324980e-01 -6.18007839e-01 1.95875394e+00
5.54202080e-01 5.95762469e-02 -4.50184464e-01 9.58377421e-02
1.01493609e+00 4.18586910e-01 -5.58509678e-02 -6.70730948e-01
-2.24733531e-01 1.69759154e-01 7.10154474e-02 6.34672105e-01
-1.18403423e+00 9.41000581e-01 4.72721910e+00 7.63857305e-01
-7.24742711e-01 -2.97722183e-02 3.88220638e-01 6.52534008e-01
-7.31983066e-01 2.52256095e-01 -8.57227147e-01 7.68078625e-01
5.74260116e-01 -4.41919774e-01 2.35771149e-01 5.13784528e-01
-3.54300961e-02 4.01832789e-01 -1.07474840e+00 9.39015329e-01
1.44597486e-01 -1.33003139e+00 1.78706378e-01 -3.23826045e-01
3.37295890e-01 -3.52920711e-01 -4.78872806e-01 5.17741561e-01
-1.49922520e-01 -9.11335230e-01 5.33279777e-01 4.68406439e-01
3.86097670e-01 -4.47424948e-01 1.14366639e+00 3.30929965e-01
-1.73246419e+00 -4.90592187e-03 -5.53841949e-01 -1.40208796e-01
2.53995031e-01 5.13438880e-01 -6.52844012e-01 1.24693763e+00
5.95409632e-01 1.07891130e+00 -6.37276351e-01 6.45422935e-01
-5.76379657e-01 4.04305547e-01 4.45719846e-02 -4.90830064e-01
3.35163116e-01 -5.70703149e-01 5.46153128e-01 1.27873147e+00
9.46938396e-02 4.43196185e-02 3.21165055e-01 1.02290511e+00
-1.16035052e-01 6.91999078e-01 -6.01053417e-01 8.40827525e-02
4.92421389e-01 1.21174061e+00 -4.12973851e-01 -2.79781997e-01
-7.13616848e-01 9.35556471e-01 7.47350931e-01 1.79624632e-02
-5.93525589e-01 -4.74137336e-01 5.26078999e-01 -3.36661279e-01
-4.48481627e-02 2.68507004e-01 -1.57782704e-01 -1.27393270e+00
5.56150377e-01 -7.50786602e-01 6.96783543e-01 -6.43401682e-01
-1.62321937e+00 7.91830182e-01 -2.03901172e-01 -1.09244084e+00
4.21625236e-03 -4.80823129e-01 -1.06620073e+00 1.17801332e+00
-1.34646344e+00 -1.05761361e+00 -3.90220851e-01 3.70299131e-01
5.86592019e-01 -1.42175555e-01 9.27750349e-01 2.34682873e-01
-7.63410211e-01 7.63325214e-01 -3.43291014e-01 3.05767775e-01
1.58065498e-01 -1.26867306e+00 6.01192296e-01 8.82000327e-01
4.04009700e-01 6.51305795e-01 5.72863281e-01 -6.38372779e-01
-1.50670218e+00 -1.07549751e+00 1.79704857e+00 -2.92130798e-01
8.74527156e-01 -3.79125357e-01 -1.36753762e+00 4.93462175e-01
2.64418185e-01 -4.17983830e-01 6.28009379e-01 2.07173154e-01
-2.96492308e-01 -1.21358164e-01 -7.07819343e-01 5.76257348e-01
1.15457046e+00 -7.35899746e-01 -1.29934311e+00 3.52977246e-01
1.16922188e+00 -2.80632973e-01 -7.66405165e-01 4.17367727e-01
1.23129889e-01 -6.77563846e-01 8.33395123e-01 -1.02984548e+00
4.56445575e-01 -2.91060954e-01 -4.04084265e-01 -1.23207998e+00
-4.43147451e-01 -3.03948492e-01 2.19484851e-01 1.53454018e+00
4.57593769e-01 -7.05521762e-01 4.81741279e-01 4.42182928e-01
-3.07626069e-01 -8.05783808e-01 -8.87141585e-01 -6.31656826e-01
-3.25734645e-01 -3.18101615e-01 8.81242573e-01 1.20523155e+00
4.00135517e-01 8.69449317e-01 -2.04305761e-02 2.61790693e-01
4.49554652e-01 6.53051913e-01 4.87049967e-01 -1.27143657e+00
-4.18350637e-01 -6.93705916e-01 -8.31726551e-01 -9.18076098e-01
1.00485301e+00 -1.75019979e+00 -7.07972273e-02 -1.56934452e+00
5.14657021e-01 -3.01684380e-01 -3.74377102e-01 3.12749207e-01
-8.49746943e-01 -3.58315051e-01 2.32004926e-01 1.60758346e-01
-4.44478214e-01 6.19756401e-01 1.12299228e+00 -6.10893130e-01
-1.95310952e-03 -8.47970843e-02 -6.60777867e-01 5.75991750e-01
7.43698418e-01 -6.12549722e-01 -3.22116852e-01 -4.75243509e-01
2.90515661e-01 1.51434198e-01 6.55561537e-02 -4.21620101e-01
3.75763714e-01 -3.08462000e-03 -2.21799836e-01 -2.83734560e-01
-5.07737249e-02 -8.68839443e-01 1.73618361e-01 5.41046023e-01
-4.37482089e-01 3.08646590e-01 -1.25358149e-01 6.14003956e-01
-7.15153873e-01 -5.12985051e-01 2.66191930e-01 3.11283357e-02
-8.35949600e-01 2.78730810e-01 7.87579566e-02 1.48137823e-01
9.47231591e-01 -3.08773458e-01 -2.61716276e-01 1.50120643e-03
-6.73535824e-01 5.15796542e-01 1.27753556e-01 7.04296231e-01
6.70580149e-01 -1.56330669e+00 -7.58868098e-01 2.07407445e-01
3.12123686e-01 -1.48156062e-01 2.98696488e-01 6.38609827e-01
-1.04090787e-01 4.17605728e-01 7.07098916e-02 -5.98266363e-01
-1.69379127e+00 6.05099261e-01 3.45843166e-01 -6.83691680e-01
-5.38911998e-01 7.79013634e-01 2.95756847e-01 -6.89654529e-01
7.47808591e-02 -3.61921072e-01 -7.48197913e-01 9.26447809e-02
3.50141793e-01 -2.32256725e-01 -2.15859758e-03 -9.81030166e-01
-5.81953704e-01 1.06520498e+00 4.91259284e-02 5.31778157e-01
1.03809166e+00 -8.84494409e-02 -6.98811352e-01 2.74204940e-01
1.65056503e+00 -2.09317133e-02 -3.77329648e-01 -5.15541077e-01
5.12416780e-01 -5.13368130e-01 -3.03749442e-01 -4.54395153e-02
-1.03324330e+00 7.63689756e-01 6.70551956e-02 3.25969785e-01
1.05383193e+00 3.95672679e-01 9.35306787e-01 3.49543780e-01
2.99068630e-01 -5.27294636e-01 -2.89327409e-02 6.77868605e-01
8.15953553e-01 -1.22052801e+00 -3.30097914e-01 -9.03574049e-01
-6.35091186e-01 1.29520011e+00 5.01380324e-01 -1.60264850e-01
4.81387466e-01 -1.54694557e-01 -5.84205724e-02 -5.07261753e-01
-5.35431683e-01 -1.73679143e-01 9.27046478e-01 1.07202142e-01
5.11239111e-01 2.26620883e-01 -6.59333348e-01 1.09758401e+00
-3.25457960e-01 -6.61142945e-01 1.94825102e-02 5.36807537e-01
-4.82778609e-01 -1.37570262e+00 -2.99066063e-02 5.16404510e-01
-2.00829402e-01 -2.72909164e-01 -5.41198134e-01 4.24777389e-01
-8.08998942e-02 9.98409092e-01 4.88997847e-02 -6.52755260e-01
6.96742177e-01 -3.78355756e-02 2.30310470e-01 -9.58109200e-01
-1.01543915e+00 -4.55879778e-01 2.03191295e-01 -8.23193669e-01
-2.20090061e-01 -4.76745397e-01 -1.44782388e+00 5.38881496e-02
-3.58371496e-01 4.24548090e-01 3.23119223e-01 1.10962427e+00
3.37355614e-01 6.73460960e-01 8.31236601e-01 -3.89501601e-01
-6.17133856e-01 -7.70370543e-01 -3.60513180e-01 1.06917274e+00
-9.01771113e-02 -3.68650794e-01 -4.38433379e-01 -2.93785006e-01] | [10.900505065917969, 8.58779525756836] |
17744caa-043e-42df-a213-e293b50634ce | collaborative-spatial-temporal-modeling-for | 2105.06818 | null | https://arxiv.org/abs/2105.06818v1 | https://arxiv.org/pdf/2105.06818v1.pdf | Collaborative Spatial-Temporal Modeling for Language-Queried Video Actor Segmentation | Language-queried video actor segmentation aims to predict the pixel-level mask of the actor which performs the actions described by a natural language query in the target frames. Existing methods adopt 3D CNNs over the video clip as a general encoder to extract a mixed spatio-temporal feature for the target frame. Though 3D convolutions are amenable to recognizing which actor is performing the queried actions, it also inevitably introduces misaligned spatial information from adjacent frames, which confuses features of the target frame and yields inaccurate segmentation. Therefore, we propose a collaborative spatial-temporal encoder-decoder framework which contains a 3D temporal encoder over the video clip to recognize the queried actions, and a 2D spatial encoder over the target frame to accurately segment the queried actors. In the decoder, a Language-Guided Feature Selection (LGFS) module is proposed to flexibly integrate spatial and temporal features from the two encoders. We also propose a Cross-Modal Adaptive Modulation (CMAM) module to dynamically recombine spatial- and temporal-relevant linguistic features for multimodal feature interaction in each stage of the two encoders. Our method achieves new state-of-the-art performance on two popular benchmarks with less computational overhead than previous approaches. | ['Fei Wang', 'Jizhong Han', 'Wenguan Wang', 'Guanbin Li', 'Zihan Ding', 'Si Liu', 'Shaofei Huang', 'Tianrui Hui'] | 2021-05-14 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Hui_Collaborative_Spatial-Temporal_Modeling_for_Language-Queried_Video_Actor_Segmentation_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Hui_Collaborative_Spatial-Temporal_Modeling_for_Language-Queried_Video_Actor_Segmentation_CVPR_2021_paper.pdf | cvpr-2021-1 | ['referring-expression-segmentation'] | ['computer-vision'] | [ 2.84324974e-01 -1.72024861e-01 -4.17277068e-01 -6.63289070e-01
-1.02485454e+00 -4.96821940e-01 5.17071247e-01 -1.83456942e-01
-5.70686281e-01 1.88813820e-01 2.81629086e-01 6.56828061e-02
3.41223538e-01 -5.16317844e-01 -6.06334925e-01 -6.49273872e-01
-6.35351613e-03 1.44118309e-01 4.41296846e-01 2.45787531e-01
1.23720586e-01 5.58098555e-01 -1.35448432e+00 8.12672198e-01
3.94981354e-01 1.44677794e+00 4.26604360e-01 8.33771229e-01
-2.18102813e-01 9.66811597e-01 -5.16015589e-01 -2.20678687e-01
1.42474085e-01 -5.86430192e-01 -1.00747561e+00 5.24412215e-01
5.24419188e-01 -4.99739826e-01 -7.49597788e-01 8.75356913e-01
4.42723900e-01 2.75516361e-01 4.35761958e-01 -1.12706482e+00
-3.86929184e-01 5.21165490e-01 -7.08420455e-01 3.83291692e-01
5.59940100e-01 3.97000849e-01 8.96510184e-01 -8.26310396e-01
7.51380026e-01 1.30919826e+00 1.95436344e-01 5.10119736e-01
-9.23252702e-01 -4.14617687e-01 4.46685851e-01 3.50650728e-01
-1.42384648e+00 -6.03162587e-01 8.58048499e-01 -5.20658076e-01
9.09479976e-01 1.46436200e-01 8.79171371e-01 8.89855444e-01
2.39302546e-01 1.12103379e+00 5.72504520e-01 -2.94259116e-02
-5.95201626e-02 -2.86551774e-01 -2.28124961e-01 1.00017321e+00
-9.43823040e-01 9.07974839e-02 -9.48083818e-01 1.30251199e-01
9.34803843e-01 1.12469561e-01 -2.01776803e-01 -7.89240822e-02
-1.45730388e+00 5.49276948e-01 2.93645948e-01 3.56548071e-01
-5.82270324e-01 3.32866818e-01 4.82311755e-01 1.15107201e-01
3.13829541e-01 8.88488814e-02 -5.97554386e-01 -3.59542906e-01
-1.25608504e+00 1.71123490e-01 3.18182290e-01 1.03713512e+00
8.33292544e-01 -2.48884767e-01 -8.39937031e-01 6.28767908e-01
3.32929313e-01 2.78236240e-01 2.50900388e-01 -1.50380456e+00
4.59749401e-01 8.44266653e-01 -4.47608903e-02 -9.53405559e-01
-1.35861292e-01 7.72003606e-02 -4.94973540e-01 -7.07437098e-02
3.21822017e-01 5.96686825e-03 -1.05223477e+00 1.59487414e+00
5.31191170e-01 3.40021342e-01 -7.52573609e-02 1.37581968e+00
8.53961349e-01 8.38178992e-01 1.02330290e-01 -1.66624472e-01
1.46161604e+00 -1.15953612e+00 -7.00476050e-01 -2.84647137e-01
3.99000227e-01 -7.20159590e-01 8.65811884e-01 1.02225803e-02
-1.41141713e+00 -6.60171330e-01 -7.21805930e-01 -3.69218677e-01
-3.76613326e-02 1.80315167e-01 3.56223226e-01 -3.30250524e-02
-8.29740822e-01 2.47905686e-01 -8.90711904e-01 -4.42824215e-02
6.00550056e-01 4.99252439e-01 -4.08028662e-01 -3.06481272e-02
-1.23251820e+00 6.76242709e-01 3.41193944e-01 1.58009350e-01
-1.31958914e+00 -3.92730713e-01 -1.12031960e+00 -8.22353512e-02
5.12627542e-01 -5.80722809e-01 1.31366777e+00 -1.36362636e+00
-1.60894608e+00 1.01442981e+00 -6.07548475e-01 -2.29823485e-01
2.52748877e-01 8.63905437e-03 -2.82100052e-01 6.25340939e-01
2.06973001e-01 1.21885550e+00 1.01055992e+00 -9.27309632e-01
-9.74973440e-01 -4.12705690e-01 3.97640973e-01 5.55124223e-01
4.53788787e-02 1.38150036e-01 -1.39383447e+00 -6.78336382e-01
2.04342484e-01 -5.46482861e-01 -1.05392925e-01 2.47427180e-01
-4.57455873e-01 -1.30521506e-01 9.89226460e-01 -4.93135393e-01
1.30715191e+00 -2.35631776e+00 4.77827400e-01 -6.08993992e-02
1.82258114e-01 1.09371752e-01 -2.70752996e-01 1.08990557e-01
2.52771854e-01 -2.47810543e-01 -1.50994286e-01 -5.39568484e-01
-1.17633082e-01 2.71053761e-01 -9.82879847e-02 5.49449503e-01
3.55698794e-01 1.17642307e+00 -8.66936147e-01 -8.96130681e-01
4.23024982e-01 5.89418411e-01 -5.19015849e-01 5.23004651e-01
-3.76399875e-01 6.99195981e-01 -6.47095442e-01 8.67873251e-01
2.59057343e-01 -8.91041979e-02 -1.65464506e-01 -3.70029658e-01
-2.58294851e-01 1.55339599e-01 -8.61052990e-01 2.28905535e+00
-2.45539859e-01 6.41968489e-01 1.78332135e-01 -1.01945710e+00
6.69088960e-01 4.25382286e-01 7.54912615e-01 -8.00571561e-01
3.27334374e-01 -3.57423685e-02 -4.37273264e-01 -9.62334812e-01
3.17606062e-01 1.86684966e-01 -3.35976481e-01 2.65722871e-01
2.24400371e-01 1.11978091e-01 1.30982891e-01 3.13974880e-02
1.05828440e+00 3.87067258e-01 5.60241155e-02 1.34968370e-01
7.75235116e-01 -9.35751945e-02 8.37494254e-01 5.36247313e-01
-4.87451315e-01 6.92421496e-01 6.36518598e-01 -6.23414934e-01
-6.72539413e-01 -8.48019719e-01 2.34374270e-01 1.32940435e+00
5.58582962e-01 -2.43469208e-01 -7.60583222e-01 -9.27754343e-01
-3.21544617e-01 3.53639007e-01 -6.42426372e-01 -1.28625184e-01
-6.29387319e-01 -1.04179038e-02 5.32252431e-01 3.95514429e-01
6.96501076e-01 -1.04546261e+00 -1.08260000e+00 1.21126734e-01
-4.18917984e-01 -1.43000042e+00 -1.24127984e+00 1.12743549e-01
-3.85292351e-01 -1.09987867e+00 -7.10174561e-01 -9.36360002e-01
5.45170605e-01 3.79519351e-02 9.30257797e-01 -1.21808879e-01
-4.60171886e-02 3.67925853e-01 -3.90972048e-01 2.44265586e-01
-1.20695531e-01 -1.11663483e-01 -4.07649398e-01 4.73349273e-01
4.17285740e-01 -1.71960562e-01 -8.50116313e-01 4.36114758e-01
-8.64526570e-01 4.37341571e-01 4.35858727e-01 6.34969115e-01
8.66003036e-01 -2.12106198e-01 8.89688656e-02 -4.55521137e-01
1.67857692e-01 -2.89114505e-01 -4.75818247e-01 3.19777131e-01
1.96931317e-01 -3.09586376e-02 4.62140799e-01 -6.00572944e-01
-8.82014632e-01 6.11014485e-01 -1.04196325e-01 -8.74853075e-01
-3.17088425e-01 2.96065718e-01 -4.25150394e-01 2.03551844e-01
7.39151686e-02 3.75314564e-01 -1.61995426e-01 -2.39250675e-01
3.51753086e-01 6.71393991e-01 8.08825970e-01 -4.63315099e-01
2.10464418e-01 4.91160005e-01 -2.61380106e-01 -4.94085342e-01
-9.40202534e-01 -3.56944203e-01 -9.10854697e-01 -4.98274982e-01
1.49814868e+00 -9.26960289e-01 -9.00493920e-01 4.58020836e-01
-1.44540203e+00 -3.53353858e-01 -2.61672050e-01 4.29402322e-01
-8.28973472e-01 2.02137366e-01 -5.19418120e-01 -5.18776655e-01
-6.60631135e-02 -1.84311557e+00 1.67085433e+00 2.39968047e-01
-1.69171587e-01 -7.19259024e-01 -3.16597819e-01 5.29403150e-01
8.52633175e-03 2.61684626e-01 7.42259085e-01 -3.49314094e-01
-8.01193655e-01 -6.29648790e-02 -1.83206946e-01 7.26018697e-02
-7.45080062e-04 9.07884724e-03 -8.03454518e-01 8.17910954e-02
-9.77384150e-02 -1.95194170e-01 7.95517266e-01 6.63800240e-01
1.24645901e+00 -1.98486388e-01 -3.46445233e-01 7.63494551e-01
9.55705106e-01 5.07608235e-01 5.04447997e-01 -1.03525288e-01
7.28360891e-01 5.11455476e-01 7.30175853e-01 3.65395427e-01
5.85358679e-01 9.46509063e-01 4.44718689e-01 -1.76837996e-01
2.47637834e-02 -2.04293206e-01 6.90844297e-01 3.91283810e-01
2.63006240e-01 -3.15320045e-01 -6.91865265e-01 4.75919455e-01
-2.00702548e+00 -1.00903869e+00 2.70082325e-01 1.88812256e+00
9.36694920e-01 -1.05687216e-01 1.99600056e-01 -1.20834716e-01
7.61376560e-01 4.60018277e-01 -8.43114436e-01 -2.03829616e-01
-9.86438096e-02 -1.44577950e-01 2.39836022e-01 6.50937915e-01
-1.39011168e+00 1.10060716e+00 5.98376608e+00 8.92435968e-01
-1.31775439e+00 1.91407695e-01 8.15589726e-01 -4.48179215e-01
-1.92018047e-01 -1.38946414e-01 -6.07508838e-01 4.35927093e-01
6.72215462e-01 3.40013564e-01 3.39606553e-01 4.70434368e-01
6.43325508e-01 -2.84086406e-01 -1.29448164e+00 1.17189670e+00
1.79580167e-01 -1.34758890e+00 -5.20107755e-03 -8.45048726e-02
5.41797340e-01 -1.66987598e-01 3.79793346e-02 -3.68494950e-02
-2.46365115e-01 -1.08481824e+00 1.03044403e+00 8.78500640e-01
9.88218367e-01 -7.13889003e-01 3.97121459e-01 3.27980250e-01
-1.42682993e+00 -1.18809290e-01 1.33043125e-01 2.32408464e-01
4.04489666e-01 1.92885220e-01 -4.92077500e-01 2.04038665e-01
6.14818394e-01 9.08441544e-01 -1.88648358e-01 6.88890636e-01
-1.37994200e-01 1.77230075e-01 -3.58673856e-02 2.53988326e-01
5.89788020e-01 3.72350253e-02 6.30394161e-01 1.15160942e+00
1.94409192e-01 2.77148157e-01 3.47983509e-01 8.97595227e-01
-2.16083825e-02 -6.47958741e-02 -3.63958746e-01 -2.30028063e-01
3.50876242e-01 1.00686157e+00 -6.07812703e-01 -3.83201450e-01
-6.95806921e-01 1.46567369e+00 1.64039209e-01 5.30987501e-01
-1.06073952e+00 -1.53660655e-01 7.22320318e-01 1.05930202e-01
5.25576770e-01 -3.25685948e-01 3.39855663e-02 -1.13244700e+00
1.90893334e-04 -8.43792081e-01 3.88336688e-01 -7.49013901e-01
-8.43469739e-01 6.26389861e-01 -1.42381206e-01 -1.23751819e+00
-1.79199293e-01 -3.73241007e-01 -2.82570958e-01 8.16478908e-01
-1.29071152e+00 -1.24866068e+00 -1.42729759e-01 8.33639681e-01
9.64921832e-01 -1.90712124e-01 5.50844669e-01 4.64426816e-01
-6.02122903e-01 4.57600296e-01 -5.63888490e-01 3.90257329e-01
5.19428909e-01 -9.46796834e-01 8.51060450e-03 7.13484406e-01
2.33319014e-01 1.44561023e-01 2.71347702e-01 -4.64924097e-01
-1.42019451e+00 -1.20427299e+00 1.01363921e+00 -1.82981372e-01
2.30186522e-01 -3.00760716e-01 -6.59198463e-01 6.10855222e-01
3.15011561e-01 2.56513029e-01 5.09466052e-01 -5.73186398e-01
-1.24671705e-01 -4.12638821e-02 -9.27742600e-01 5.29479980e-01
9.77960110e-01 -8.07316244e-01 -4.42551345e-01 4.66837846e-02
7.89749920e-01 -7.70249605e-01 -8.45521390e-01 3.54888946e-01
7.02163637e-01 -1.12218225e+00 8.58575881e-01 -4.01390523e-01
6.08483732e-01 -6.22029781e-01 -4.11663443e-01 -6.29018843e-01
8.87284949e-02 -6.31318867e-01 -2.56270736e-01 1.01168001e+00
2.78260678e-01 1.03080876e-01 6.63929880e-01 6.06683254e-01
-3.23776841e-01 -1.24303663e+00 -1.19780588e+00 -6.96405470e-02
-3.93620223e-01 -8.26102853e-01 4.94082361e-01 6.61284029e-01
-5.55398166e-02 3.24038655e-01 -3.83933127e-01 2.41638318e-01
1.04923710e-01 2.86676377e-01 4.58078474e-01 -5.19297302e-01
-1.14867114e-01 -5.81300676e-01 -5.19930363e-01 -1.79705966e+00
4.78416413e-01 -8.37317586e-01 2.36289114e-01 -1.23485732e+00
2.95912951e-01 -2.89689023e-02 -1.15677588e-01 5.99563301e-01
-8.39767531e-02 2.33950138e-01 1.28072307e-01 7.86300451e-02
-1.05518126e+00 5.57200134e-01 1.66066635e+00 -4.13052142e-01
-3.81841093e-01 -1.32808208e-01 -3.52460027e-01 6.61775768e-01
2.71073937e-01 -3.44638050e-01 -3.62051815e-01 -7.93049634e-01
-1.50527492e-01 7.52580464e-01 4.83805537e-01 -8.07654858e-01
4.55898315e-01 -3.90224159e-01 5.45277774e-01 -7.89865613e-01
7.26144135e-01 -8.96434903e-01 -2.50322726e-02 7.95875415e-02
-7.40322590e-01 4.89573665e-02 -2.04262938e-02 4.05235618e-01
-5.33610761e-01 7.35247508e-02 8.48320365e-01 -1.61898330e-01
-8.56494069e-01 7.98350751e-01 -5.42927325e-01 -2.77599636e-02
1.21921825e+00 -3.96266758e-01 2.17090532e-01 -4.28782880e-01
-9.56447065e-01 4.78758544e-01 4.79913920e-01 4.25302953e-01
9.78091300e-01 -1.26242268e+00 -4.17223543e-01 3.57092977e-01
-6.79123178e-02 1.81144223e-01 4.62878793e-01 1.08786929e+00
-3.95583659e-01 3.18148524e-01 1.11957811e-01 -7.86873698e-01
-1.32367849e+00 2.73436576e-01 7.68511057e-01 2.04789694e-02
-4.74124461e-01 1.08130419e+00 2.57908851e-01 -2.51839515e-02
4.91460711e-01 -2.21323445e-01 -1.69456974e-01 1.93983600e-01
4.61387008e-01 -1.79973692e-02 -4.65912849e-01 -1.34239542e+00
-3.93521011e-01 6.39827192e-01 1.20244585e-01 -2.70224631e-01
9.80361521e-01 -4.40482646e-01 -1.39296353e-01 6.48088574e-01
1.59360635e+00 -3.35398406e-01 -1.73102987e+00 -4.66045260e-01
-2.89628297e-01 -5.16131401e-01 1.22369483e-01 -6.06440187e-01
-1.42994595e+00 1.02841330e+00 4.52032268e-01 -1.21152692e-01
1.46599877e+00 3.61124814e-01 1.01894939e+00 -2.90782526e-02
3.52445543e-02 -1.12318659e+00 2.10870147e-01 6.73996627e-01
6.99720025e-01 -1.05857456e+00 -3.84643167e-01 -2.01988250e-01
-1.04946029e+00 1.17020941e+00 7.22620726e-01 2.17468560e-01
5.43806076e-01 2.72855401e-01 1.88030258e-01 -3.76771957e-01
-9.17757928e-01 -3.75355482e-01 4.95752335e-01 2.97179461e-01
4.07226473e-01 -2.21450552e-01 -5.70463808e-03 5.89526713e-01
2.72318959e-01 -1.26588047e-01 -7.76736736e-02 8.95654917e-01
-3.05762351e-01 -9.40448165e-01 -1.37169063e-01 3.37089092e-01
-4.71783310e-01 6.46500662e-03 -4.15924788e-01 3.70572865e-01
6.57132804e-01 8.73437166e-01 3.84130687e-01 -6.11737013e-01
2.98109144e-01 1.86836526e-01 3.50545168e-01 -5.59560657e-01
-7.10260570e-01 5.68112552e-01 -1.71087041e-01 -1.02020180e+00
-8.31313848e-01 -7.75836051e-01 -1.66250837e+00 1.41016513e-01
4.75942306e-02 -2.15927567e-02 2.65904546e-01 1.13265789e+00
5.68083465e-01 5.12817383e-01 6.98674321e-01 -1.11260688e+00
2.05615357e-01 -5.87113082e-01 -4.48687047e-01 2.96875477e-01
6.60797834e-01 -6.23649359e-01 -2.75412556e-02 3.91700715e-01] | [9.704131126403809, 0.5738435387611389] |
c32083dc-29a2-47aa-81c6-afe422d581d9 | prototypical-quadruplet-for-few-shot-class | 2211.02947 | null | https://arxiv.org/abs/2211.02947v3 | https://arxiv.org/pdf/2211.02947v3.pdf | Prototypical quadruplet for few-shot class incremental learning | Scarcity of data and incremental learning of new tasks pose two major bottlenecks for many modern computer vision algorithms. The phenomenon of catastrophic forgetting, i.e., the model's inability to classify previously learned data after training with new batches of data, is a major challenge. Conventional methods address catastrophic forgetting while compromising the current session's training. Generative replay-based approaches, such as generative adversarial networks (GANs), have been proposed to mitigate catastrophic forgetting, but training GANs with few samples may lead to instability. To address these challenges, we propose a novel method that improves classification robustness by identifying a better embedding space using an improved contrasting loss. Our approach retains previously acquired knowledge in the embedding space, even when trained with new classes, by updating previous session class prototypes to represent the true class mean, which is crucial for our nearest class mean classification strategy. We demonstrate the effectiveness of our method by showing that the embedding space remains intact after training the model with new classes and outperforms existing state-of-the-art algorithms in terms of accuracy across different sessions. | ['Subhasis Chaudhuri', 'Biplab Banerjee', 'Sanchar Palit'] | 2022-11-05 | null | null | null | null | ['few-shot-class-incremental-learning'] | ['methodology'] | [ 5.70093393e-01 2.18437403e-01 -2.35736594e-02 -1.29315332e-01
-5.25581419e-01 -6.28731310e-01 6.74364865e-01 -2.50240296e-01
-3.23732793e-01 1.11920762e+00 2.03289818e-02 4.64076698e-02
8.29768777e-02 -7.13322699e-01 -9.78304327e-01 -9.40102279e-01
4.32910979e-01 3.73648107e-01 1.82739034e-01 2.95408852e-02
2.28218436e-01 4.20711428e-01 -1.73310888e+00 1.71222031e-01
9.99898374e-01 7.76975751e-01 9.74781066e-02 7.68956363e-01
2.15915833e-02 5.77163815e-01 -9.39336896e-01 -5.79449713e-01
2.76268691e-01 -6.70141280e-01 -4.60704416e-01 1.47336379e-01
6.07353270e-01 -4.26899552e-01 -5.23438275e-01 8.31328809e-01
3.89443994e-01 4.28289995e-02 5.47083020e-01 -1.55398655e+00
-1.03326106e+00 2.81673431e-01 -2.30097994e-01 1.65837944e-01
-8.58087912e-02 2.86049247e-01 4.25023139e-01 -9.75354970e-01
5.27857125e-01 9.09214377e-01 1.07398248e+00 1.19732308e+00
-1.57575738e+00 -7.64909089e-01 1.17228590e-01 3.39639157e-01
-1.18815243e+00 -6.84621274e-01 9.95852172e-01 -3.51353705e-01
6.75058603e-01 2.16864839e-01 8.52563381e-01 1.68441451e+00
4.24533099e-01 7.19520032e-01 1.07089114e+00 -4.43630368e-01
5.92695713e-01 4.67115849e-01 -2.70745218e-01 3.78588915e-01
3.62061292e-01 2.16433510e-01 -7.44597554e-01 -3.14086080e-02
6.60602927e-01 3.19401473e-01 -1.67108700e-01 -7.55102992e-01
-1.06627536e+00 7.82199442e-01 3.67799640e-01 6.24442250e-02
-2.58957535e-01 1.34632573e-01 1.90441772e-01 5.02434790e-01
5.45773029e-01 4.21639085e-01 -5.03749371e-01 -1.39067039e-01
-9.00345206e-01 1.30610466e-02 6.62155747e-01 7.74201393e-01
6.59716785e-01 3.44119787e-01 -8.95610675e-02 7.62592971e-01
-2.69371629e-01 5.42501748e-01 7.83408344e-01 -8.26772928e-01
4.45627049e-02 3.32807243e-01 -1.19750410e-01 -6.92095160e-01
4.86657247e-02 -7.83128440e-01 -1.11293721e+00 3.97074431e-01
2.91477352e-01 5.68585433e-02 -1.26326001e+00 2.13622808e+00
3.39044422e-01 5.58705270e-01 3.02045465e-01 2.69182116e-01
2.20285758e-01 4.60872591e-01 -6.44766493e-03 -2.64362454e-01
6.81269288e-01 -9.11255181e-01 -5.88608325e-01 -5.64882278e-01
-1.00055840e-02 -5.38766384e-01 1.20950496e+00 3.77605498e-01
-8.63223374e-01 -7.95436025e-01 -1.13471866e+00 2.28749111e-01
-4.99986738e-01 -1.31258443e-01 5.41754186e-01 7.93917120e-01
-1.14482117e+00 6.76632941e-01 -8.60114038e-01 -4.80470777e-01
7.39278555e-01 2.29782283e-01 -4.70654100e-01 -2.11466312e-01
-9.55805004e-01 9.65864897e-01 2.59447962e-01 -4.50658165e-02
-1.07994223e+00 -9.92897570e-01 -7.08546937e-01 1.33107472e-02
9.90161002e-02 -9.62382853e-01 1.07287574e+00 -1.14788568e+00
-1.46347976e+00 5.02509713e-01 -1.65890098e-01 -6.42710388e-01
7.01728523e-01 -2.87890017e-01 -3.08451116e-01 -8.97604674e-02
-4.36720550e-02 8.33305240e-01 1.72528684e+00 -1.47905612e+00
-3.82429808e-01 -3.02607805e-01 -2.52209425e-01 1.54631454e-02
-7.27594316e-01 -9.05258954e-01 9.73744392e-02 -8.88563216e-01
1.68169409e-01 -9.66878653e-01 -1.67693980e-02 2.06225410e-01
-9.55873057e-02 9.16293114e-02 1.12663460e+00 -5.48501432e-01
9.26266372e-01 -2.24589729e+00 2.23506033e-01 -1.69887319e-01
3.58982056e-01 4.33551699e-01 -1.32990137e-01 2.27841303e-01
-9.59899500e-02 -6.18996751e-03 -3.66665810e-01 -7.92043447e-01
-2.81830758e-01 5.75382650e-01 -6.72985494e-01 2.80106097e-01
4.45588529e-01 1.05918586e+00 -1.02232325e+00 1.93403270e-02
2.37257853e-01 7.23553538e-01 -5.17548978e-01 2.52882391e-01
-7.70832226e-02 6.55779600e-01 1.37024820e-01 4.06634718e-01
7.40566254e-01 -1.82264492e-01 -8.08335841e-02 6.80822507e-03
3.78474802e-01 -1.11637585e-01 -7.40234077e-01 1.79037035e+00
-4.38445687e-01 6.38637543e-01 -4.76343423e-01 -1.11763728e+00
7.96193957e-01 6.90045953e-02 1.18091702e-01 -5.67312241e-01
-3.22127551e-01 5.87427579e-02 -2.15390012e-01 -2.04132870e-01
3.18481296e-01 -2.82593489e-01 -1.19951228e-02 4.80943978e-01
2.80434102e-01 -8.13690796e-02 -3.27471346e-01 1.99920833e-01
1.34086144e+00 1.94970235e-01 1.37141168e-01 1.85409620e-01
1.19596422e-01 -2.21759036e-01 7.52563298e-01 1.17775929e+00
-3.02623659e-01 7.53106177e-01 1.78606629e-01 -6.90988719e-01
-1.53493655e+00 -1.41293097e+00 2.55848113e-02 7.53520489e-01
-1.68115333e-01 6.18567578e-02 -6.46558821e-01 -9.69167471e-01
1.83840647e-01 9.48631167e-01 -1.09091079e+00 -1.07221079e+00
-4.68889475e-01 -5.79091907e-01 4.25103784e-01 5.22907913e-01
6.62680805e-01 -1.01951635e+00 -6.06539488e-01 3.66992027e-01
-1.40646929e-02 -9.14711773e-01 -2.94901580e-01 2.00471401e-01
-1.25863564e+00 -9.77606595e-01 -8.03099096e-01 -6.04721010e-01
1.08280659e+00 2.56138802e-01 1.12798047e+00 4.59874310e-02
-3.59240085e-01 5.58608949e-01 -3.58440489e-01 -3.51755708e-01
-6.44202173e-01 3.15030485e-01 3.60619456e-01 8.18654299e-02
2.05683783e-01 -9.55667555e-01 -6.92339599e-01 1.10712647e-01
-1.03648508e+00 2.12943822e-01 8.10955465e-01 1.32140315e+00
4.42437679e-01 6.37601539e-02 1.03832114e+00 -1.12902951e+00
4.22237903e-01 -4.17590082e-01 -1.64745525e-01 2.55713850e-01
-8.93734813e-01 1.32256761e-01 8.83623362e-01 -9.35878575e-01
-9.30914819e-01 3.23477834e-02 2.86532398e-02 -8.17695558e-01
1.30363172e-02 -2.98514348e-02 -7.64773265e-02 -2.49792218e-01
7.83007443e-01 7.78432012e-01 2.53776997e-01 -3.40585291e-01
4.84713167e-01 4.17886227e-01 8.22801113e-01 -3.23473006e-01
1.08244956e+00 6.01372421e-01 -1.73983917e-01 -5.84048748e-01
-1.02494705e+00 -1.88862905e-02 -7.36841083e-01 -2.39802405e-01
3.62210512e-01 -8.07939649e-01 -2.66078234e-01 8.80690336e-01
-1.03282595e+00 -2.67292976e-01 -9.11464036e-01 9.21012163e-02
-6.85235202e-01 9.09754410e-02 -2.59988546e-01 -8.13115001e-01
-3.72987866e-01 -6.53096557e-01 8.38455677e-01 3.64008904e-01
-6.69424385e-02 -9.52471137e-01 2.45584071e-01 1.09072655e-01
7.42856145e-01 2.80350119e-01 7.24701464e-01 -4.08766598e-01
-4.64857310e-01 -3.23423624e-01 1.91275582e-01 7.75019169e-01
4.29680139e-01 -3.91732484e-01 -1.16980779e+00 -8.26006174e-01
2.47042343e-01 -2.30585948e-01 1.19625092e+00 4.50952239e-02
1.20972478e+00 -5.13295352e-01 -3.46034527e-01 5.88701010e-01
1.28195643e+00 1.85418159e-01 9.79540229e-01 2.76366115e-01
4.72937733e-01 5.68483993e-02 1.40990645e-01 3.60540599e-01
9.56308544e-02 2.70344973e-01 3.33652258e-01 1.64616644e-01
-5.72655559e-01 -6.96932435e-01 3.45668882e-01 8.02487135e-01
3.53039682e-01 -1.02398910e-01 -6.66095078e-01 6.85204804e-01
-1.72791827e+00 -1.04535937e+00 8.51894319e-01 2.51528096e+00
8.60027909e-01 4.23164845e-01 -3.04791480e-01 3.66899073e-01
6.26254559e-01 -1.05216019e-01 -1.23170066e+00 -6.89626411e-02
-1.73200369e-01 3.34563553e-01 2.76759058e-01 3.82817894e-01
-9.49431062e-01 8.70925426e-01 6.93988848e+00 5.94371319e-01
-1.05358851e+00 3.48542094e-01 6.21541917e-01 -2.56668866e-01
-2.85798371e-01 5.98934181e-02 -7.39933610e-01 7.91114688e-01
8.88883591e-01 -1.04858518e-01 7.03318477e-01 8.96782696e-01
-3.27465951e-01 -1.11094356e-01 -1.14595389e+00 1.05523014e+00
7.05857992e-01 -1.43997872e+00 2.88869530e-01 -1.83449730e-01
9.62884128e-01 -3.58938754e-01 5.27326822e-01 5.90577424e-01
1.94352478e-01 -7.55775213e-01 6.32901251e-01 8.81775737e-01
9.49664354e-01 -7.27740109e-01 5.63392878e-01 4.13889438e-01
-6.85815334e-01 -2.45472133e-01 -4.40897882e-01 1.05188586e-01
-1.38225049e-01 9.13327157e-01 -9.22264338e-01 1.11072175e-01
7.17480183e-01 6.25932455e-01 -1.03114283e+00 9.04909670e-01
-2.69272745e-01 6.47318006e-01 -3.62165987e-01 3.18249583e-01
-3.80920887e-01 1.91822037e-01 6.60179853e-01 4.79255289e-01
3.90370399e-01 -2.00898707e-01 -1.75268531e-01 8.60930383e-01
-2.41442069e-01 -6.65033519e-01 -8.28854799e-01 2.62484234e-03
5.78286469e-01 7.93643475e-01 -5.45217037e-01 -3.78888249e-01
-1.16765551e-01 1.58143818e+00 5.78637242e-01 3.87570590e-01
-7.41859138e-01 -4.06200141e-01 6.05300307e-01 -1.78956296e-02
4.24492091e-01 -3.06006104e-01 -3.41018766e-01 -1.33607006e+00
2.45088458e-01 -7.46388972e-01 2.28559434e-01 -7.80452967e-01
-1.64273715e+00 5.44670820e-01 -4.42296326e-01 -1.26908123e+00
-2.77408987e-01 -1.76850364e-01 -6.28721356e-01 5.20129800e-01
-1.38500977e+00 -1.35811162e+00 -4.27585006e-01 5.67427278e-01
7.60944068e-01 -5.17619312e-01 9.36247349e-01 7.66685009e-02
-3.05358261e-01 1.04029036e+00 4.11498517e-01 -1.91312164e-01
8.11123669e-01 -1.12070262e+00 5.70842564e-01 9.79346514e-01
3.10238451e-01 5.04254282e-01 8.79566908e-01 -7.50051379e-01
-1.33076310e+00 -1.16281462e+00 6.99868083e-01 -7.84380078e-01
2.63024658e-01 -5.78810096e-01 -1.12833261e+00 7.83876240e-01
-8.32248032e-02 -2.81908941e-02 7.18776345e-01 -2.18040030e-02
-5.47906220e-01 -3.91216308e-01 -1.37845170e+00 5.59368134e-01
1.11204755e+00 -5.60950577e-01 -7.09482193e-01 1.72149554e-01
8.32926631e-01 -2.56309807e-01 -4.56180185e-01 2.39780933e-01
6.16736293e-01 -8.52842569e-01 9.46667969e-01 -7.16812670e-01
2.21611440e-01 -2.23438084e-01 7.56607428e-02 -1.68718290e+00
-2.64975727e-01 -5.81672132e-01 -7.01391459e-01 1.12532520e+00
1.51668698e-01 -9.47435200e-01 1.06791139e+00 4.06809688e-01
2.23585833e-02 -4.65229869e-01 -1.28378022e+00 -9.83537793e-01
1.31653130e-01 -1.68836787e-01 5.60647607e-01 8.98840010e-01
-5.37797868e-01 2.92509407e-01 -7.10111201e-01 1.63241681e-02
9.27007735e-01 -1.22889698e-01 9.06262755e-01 -1.25629044e+00
-1.61627054e-01 -1.18200138e-01 -6.16545022e-01 -7.71007717e-01
1.71787217e-01 -5.41029871e-01 -5.31042702e-02 -1.23129261e+00
2.54881769e-01 -4.38620031e-01 -5.51225424e-01 7.09944844e-01
-1.69418946e-01 4.74905461e-01 1.85949489e-01 4.60948616e-01
-3.97897631e-01 9.89009857e-01 1.12087488e+00 -2.63930976e-01
-1.59890160e-01 -1.39380777e-02 -8.63100231e-01 3.96419495e-01
8.09102356e-01 -7.45060623e-01 -6.14468336e-01 -4.02307838e-01
2.40342945e-01 -4.96738285e-01 5.59724391e-01 -1.54035854e+00
3.95725429e-01 4.40693237e-02 1.00010431e+00 -2.85351843e-01
5.12469471e-01 -8.58316839e-01 4.59836453e-01 6.72815204e-01
-2.44776413e-01 -6.23365343e-02 3.01937848e-01 9.96110439e-01
5.73168099e-02 -7.87117705e-02 1.01031804e+00 -3.96960787e-03
-5.41999221e-01 2.85678834e-01 -4.29151990e-02 -6.02362044e-02
1.31033218e+00 -4.14696604e-01 -3.40069801e-01 -3.28111470e-01
-9.63631511e-01 -4.15023277e-03 7.38390505e-01 6.42636120e-01
9.04990375e-01 -1.57164991e+00 -6.04184747e-01 8.30915987e-01
5.57195134e-02 -6.42279312e-02 4.42485482e-01 4.22489583e-01
-1.22920506e-01 -3.12207248e-02 -4.07490849e-01 -6.16122842e-01
-8.77479315e-01 7.87967205e-01 2.47687191e-01 -1.11692831e-01
-7.46724606e-01 7.99857557e-01 -4.13019806e-02 -2.59188235e-01
6.03944734e-02 1.57659397e-01 2.34172076e-01 -1.43364863e-02
5.65590620e-01 9.60666463e-02 1.72168523e-01 -9.97780189e-02
-1.10083647e-01 3.53156328e-01 -6.10844135e-01 -8.05815496e-03
1.37558484e+00 -3.17022413e-01 2.30745822e-01 6.57938898e-01
8.88304830e-01 -2.65911877e-01 -1.68105209e+00 -4.12511110e-01
-5.41860461e-01 -8.05933774e-01 -1.72110721e-01 -9.36791301e-01
-1.01517236e+00 8.38951707e-01 9.30786431e-01 1.67626008e-01
1.15021384e+00 -5.59023432e-02 8.83412421e-01 3.46637309e-01
5.31137884e-01 -1.12539768e+00 4.79416788e-01 3.87391508e-01
9.25431967e-01 -1.20798528e+00 -6.28064349e-02 5.35675287e-02
-4.77888077e-01 9.74566340e-01 7.97219455e-01 -1.24215856e-01
7.09918976e-01 -7.76224956e-02 -1.57192461e-02 2.95812696e-01
-8.31668973e-01 1.36261329e-01 -7.38218799e-02 9.16260183e-01
-3.69622111e-01 -2.21423492e-01 2.04001531e-01 3.61171037e-01
-2.53450692e-01 1.22230098e-01 4.93157148e-01 1.16184223e+00
-2.61166066e-01 -1.00037777e+00 -1.87785327e-01 5.99215269e-01
-4.98498464e-03 -2.70677041e-02 -1.91025436e-01 5.24151444e-01
2.72683889e-01 5.73988736e-01 2.46461704e-01 -5.35998106e-01
1.98170826e-01 4.95758891e-01 6.44984484e-01 -5.21458268e-01
-9.37705040e-02 -5.37829101e-01 -6.91703022e-01 -2.59680212e-01
-1.82521313e-01 -6.55959606e-01 -5.29411674e-01 -5.25370061e-01
-2.54784554e-01 -1.16526904e-02 6.03599012e-01 8.31493556e-01
6.37010992e-01 5.98706603e-01 8.64403486e-01 -8.18023562e-01
-7.57537663e-01 -9.14002001e-01 -4.16709900e-01 5.65624774e-01
6.14621937e-01 -8.81998897e-01 -7.46994317e-01 3.37971687e-01] | [9.854318618774414, 3.3774940967559814] |
12f4761b-9487-4d27-9cc0-3e11685050d2 | finding-directions-in-gan-s-latent-space-for | 2202.00046 | null | https://arxiv.org/abs/2202.00046v2 | https://arxiv.org/pdf/2202.00046v2.pdf | Finding Directions in GAN's Latent Space for Neural Face Reenactment | This paper is on face/head reenactment where the goal is to transfer the facial pose (3D head orientation and expression) of a target face to a source face. Previous methods focus on learning embedding networks for identity and pose disentanglement which proves to be a rather hard task, degrading the quality of the generated images. We take a different approach, bypassing the training of such networks, by using (fine-tuned) pre-trained GANs which have been shown capable of producing high-quality facial images. Because GANs are characterized by weak controllability, the core of our approach is a method to discover which directions in latent GAN space are responsible for controlling facial pose and expression variations. We present a simple pipeline to learn such directions with the aid of a 3D shape model which, by construction, already captures disentangled directions for facial pose, identity and expression. Moreover, we show that by embedding real images in the GAN latent space, our method can be successfully used for the reenactment of real-world faces. Our method features several favorable properties including using a single source image (one-shot) and enabling cross-person reenactment. Our qualitative and quantitative results show that our approach often produces reenacted faces of significantly higher quality than those produced by state-of-the-art methods for the standard benchmarks of VoxCeleb1 & 2. Source code is available at: https://github.com/StelaBou/stylegan_directions_face_reenactment | ['Georgios Tzimiropoulos', 'Vasileios Argyriou', 'Stella Bounareli'] | 2022-01-31 | null | null | null | null | ['face-reenactment'] | ['computer-vision'] | [ 1.79418668e-01 5.79120040e-01 2.03812122e-01 -4.09810603e-01
-6.38041139e-01 -7.79999912e-01 9.01045859e-01 -9.42839980e-01
2.81049479e-02 6.98562443e-01 3.31674337e-01 2.08308443e-01
3.48583311e-01 -7.00070143e-01 -8.43837798e-01 -9.40816224e-01
2.37068459e-01 4.72029209e-01 -4.84342426e-01 -5.56680799e-01
-2.86845535e-01 7.34457254e-01 -1.66701186e+00 1.07091978e-01
2.04212040e-01 7.54098892e-01 -4.94304776e-01 7.43206441e-01
3.78564119e-01 3.03953081e-01 -4.45826471e-01 -7.53599465e-01
6.18227601e-01 -8.40295553e-01 -5.74064374e-01 2.73987144e-01
8.93543303e-01 -4.84435946e-01 -2.29608431e-01 9.02711868e-01
6.76396489e-01 -2.35900879e-01 5.49903095e-01 -1.50119615e+00
-4.85826164e-01 2.22302452e-01 -6.13602281e-01 -4.92348075e-01
5.64369440e-01 2.12683544e-01 7.88661718e-01 -8.09598982e-01
8.62810075e-01 1.46010792e+00 5.55414796e-01 1.22833407e+00
-1.72978485e+00 -1.10004246e+00 -2.22937450e-01 -2.88320184e-01
-1.37308514e+00 -1.09059024e+00 8.89157832e-01 -4.39641923e-01
2.29725018e-01 3.96135330e-01 7.96920419e-01 1.74094713e+00
3.39807384e-02 3.78174514e-01 1.36721385e+00 -4.03201550e-01
-1.56106697e-02 2.55743504e-01 -6.34093642e-01 8.00926149e-01
-1.77211259e-02 3.66754711e-01 -5.45475483e-01 1.73238758e-02
1.06946993e+00 -1.46438330e-01 -3.78083616e-01 -6.74356461e-01
-1.09390247e+00 9.13198352e-01 3.99788976e-01 1.27706483e-01
-1.62745982e-01 2.28003040e-01 2.51715243e-01 4.85377163e-01
4.63585883e-01 4.25657243e-01 -3.06406349e-01 3.71328034e-02
-7.90355742e-01 4.50496912e-01 7.93384492e-01 8.96941483e-01
7.31053948e-01 1.84822738e-01 -2.18078837e-01 5.97562075e-01
1.60124585e-01 6.09169960e-01 2.13844076e-01 -1.11314857e+00
-4.04838920e-02 2.50735730e-01 -7.62567148e-02 -9.59724188e-01
-5.25346734e-02 -1.19171984e-01 -7.90144265e-01 6.90498888e-01
4.00190085e-01 -3.34670544e-01 -1.13079512e+00 2.24664950e+00
5.02569795e-01 1.22320689e-01 1.78110525e-02 7.20109761e-01
7.17543006e-01 3.97172421e-01 -2.50720918e-01 -1.05998069e-01
1.36320651e+00 -6.42751098e-01 -6.30862355e-01 -1.07208791e-03
2.39598975e-01 -7.70575106e-01 9.44678247e-01 2.35520378e-01
-1.16679716e+00 -4.27066565e-01 -8.87809694e-01 -4.64233011e-02
-2.33641237e-01 -4.97689843e-03 5.28268158e-01 8.97098780e-01
-1.43458605e+00 4.48456049e-01 -6.61006987e-01 -4.50985491e-01
6.34532034e-01 6.24689341e-01 -1.03533709e+00 -3.22921723e-02
-1.02095783e+00 7.51544178e-01 -6.12405911e-02 1.32297501e-01
-1.10475886e+00 -7.51440227e-01 -1.05694842e+00 -1.82208359e-01
2.39707857e-01 -8.81071746e-01 1.00421214e+00 -1.33778548e+00
-2.09767556e+00 1.30717468e+00 -2.69907098e-02 9.61007774e-02
8.06970716e-01 -5.05485944e-02 -1.38097823e-01 2.58980766e-02
-5.79524823e-02 1.06029177e+00 1.42185247e+00 -1.42968428e+00
-2.96648610e-02 -5.35956025e-01 7.52375424e-02 -1.52886510e-02
-2.91329652e-01 -4.08741795e-02 -4.16108251e-01 -6.68395936e-01
-2.20887914e-01 -1.22229815e+00 1.64302960e-01 2.99365878e-01
-5.02781987e-01 2.92904019e-01 8.43234777e-01 -6.87040865e-01
3.78325611e-01 -2.14471412e+00 6.04339778e-01 1.54122531e-01
3.58300060e-01 1.63882047e-01 -4.08474326e-01 3.33323777e-01
-4.96254146e-01 1.37332901e-01 -7.66524896e-02 -8.14611197e-01
2.33306102e-02 7.65149221e-02 -3.63573134e-01 6.70451283e-01
1.27094314e-01 1.06994689e+00 -6.48577690e-01 -1.58886299e-01
1.60080552e-01 9.69231129e-01 -7.18282342e-01 4.51606572e-01
-6.37708604e-02 1.04985499e+00 -4.76983301e-02 3.89870077e-01
8.30024242e-01 2.89578855e-01 2.98090935e-01 -3.27531993e-01
1.72738180e-01 -8.81837457e-02 -8.00402462e-01 1.74591696e+00
-5.36257744e-01 5.82052112e-01 3.66298050e-01 -4.93138373e-01
8.07053328e-01 6.22346818e-01 3.62417430e-01 -6.11235440e-01
4.40449625e-01 1.53327778e-01 -9.20155644e-02 -1.74741998e-01
-4.51743777e-04 -5.03652751e-01 8.42247449e-04 3.25434119e-01
3.83433610e-01 -5.02231956e-01 -2.37896934e-01 -5.12990616e-02
7.44580567e-01 3.10613513e-01 8.63995217e-03 -2.54775286e-01
3.47474307e-01 -6.36859477e-01 4.45455909e-01 1.53681621e-01
1.06932536e-01 8.66680920e-01 7.58081257e-01 -1.87316611e-01
-1.27087712e+00 -1.18370819e+00 2.31777765e-02 8.66558194e-01
-3.89279962e-01 -3.93486857e-01 -1.13298368e+00 -6.20362937e-01
-7.93595091e-02 4.75917876e-01 -1.17194426e+00 -2.24167153e-01
-5.53176045e-01 -4.53176022e-01 6.96425021e-01 1.97437376e-01
3.60714644e-01 -9.77616251e-01 -2.65431821e-01 -3.60278428e-01
1.26959309e-01 -1.11819386e+00 -4.74538565e-01 -3.63712430e-01
-4.99525666e-01 -9.14283037e-01 -8.19034338e-01 -5.18801689e-01
9.66533303e-01 -2.68810898e-01 9.97190833e-01 1.44453794e-01
-3.15540433e-01 4.45985317e-01 -1.29025698e-01 -2.44954824e-01
-4.73471045e-01 -3.09195444e-02 2.08485052e-01 4.58575428e-01
-2.33461127e-01 -1.05149174e+00 -6.82394505e-01 3.21498781e-01
-9.39787447e-01 3.84885728e-01 3.48081410e-01 9.05790985e-01
1.63560554e-01 -3.03291529e-01 4.30775940e-01 -1.09900439e+00
2.90220201e-01 -1.97852403e-01 -6.93403542e-01 -1.61088571e-01
-3.47909212e-01 -5.19153774e-02 5.16095042e-01 -2.93180406e-01
-9.33663726e-01 2.32379571e-01 -3.78129989e-01 -5.28783441e-01
-3.19360554e-01 -3.74627113e-01 -6.23267829e-01 -3.18825334e-01
5.21148920e-01 4.82486635e-02 5.19044578e-01 -3.62800270e-01
7.48064697e-01 2.52703756e-01 5.22357941e-01 -5.33361375e-01
1.25084341e+00 7.86154687e-01 2.25708291e-01 -7.19016314e-01
-5.37995458e-01 2.39217669e-01 -8.52143705e-01 -1.15224026e-01
8.50945294e-01 -9.55829680e-01 -8.97772074e-01 7.08734214e-01
-9.98118281e-01 -4.26097244e-01 -3.55136245e-01 -3.56881917e-02
-8.44365001e-01 -1.80597946e-01 -5.00310898e-01 -3.76149982e-01
-3.27317476e-01 -1.16216075e+00 1.39996862e+00 1.04318336e-01
-3.28340203e-01 -9.45510924e-01 1.90952957e-01 4.54528332e-01
4.65816170e-01 8.38355601e-01 7.79415965e-01 -2.10024640e-01
-4.21397120e-01 -1.37261286e-01 1.10229731e-01 3.20750207e-01
4.47182357e-01 1.06574304e-01 -1.34039342e+00 -6.44600034e-01
2.15851068e-02 -4.50909674e-01 5.89536846e-01 8.31481144e-02
7.45987773e-01 -6.18022919e-01 -1.79510206e-01 9.99460518e-01
1.07505453e+00 -2.75966883e-01 8.34609568e-01 -1.80006146e-01
1.00455725e+00 8.55104446e-01 7.52948821e-02 1.29058689e-01
1.29255995e-01 1.06718981e+00 4.39186603e-01 -3.58658642e-01
-3.35482299e-01 -5.22865176e-01 5.64053118e-01 3.97964805e-01
-2.51670450e-01 -7.68089220e-02 -4.06493515e-01 4.09959137e-01
-1.54205787e+00 -9.11173940e-01 3.17770034e-01 2.13522434e+00
9.27643001e-01 -3.01342398e-01 2.49028340e-01 7.65379518e-02
3.98975134e-01 1.38800010e-01 -4.52587515e-01 -3.21172327e-01
3.76427993e-02 5.67506492e-01 1.74672484e-01 6.77403569e-01
-7.82276392e-01 1.09900379e+00 5.87390757e+00 4.73270506e-01
-1.46073675e+00 2.54282326e-01 7.06687093e-01 -4.42881793e-01
-4.41225260e-01 -1.66918799e-01 -6.42677903e-01 1.97144359e-01
7.50254035e-01 1.36651888e-01 6.20796084e-01 5.58597326e-01
8.64682347e-03 3.06168497e-01 -1.43802452e+00 1.04754519e+00
2.86244303e-01 -1.25765514e+00 1.68412358e-01 2.80898303e-01
8.43555629e-01 -3.78365785e-01 4.53889012e-01 6.56998679e-02
3.14039379e-01 -1.51629829e+00 7.35928714e-01 4.27816600e-01
1.45307863e+00 -7.87974358e-01 3.82140636e-01 -7.11938664e-02
-6.84483111e-01 2.85261959e-01 2.21024919e-02 1.70841753e-01
1.06977589e-01 3.76328588e-01 -8.63962054e-01 4.30601627e-01
5.60000658e-01 6.23203874e-01 -5.03981173e-01 2.20740497e-01
-5.21836042e-01 2.95879722e-01 -2.20833912e-01 5.39528430e-01
-2.11631685e-01 -1.44029737e-01 7.29344845e-01 7.85551667e-01
3.73795509e-01 -1.05765626e-01 -2.88998097e-01 1.13419056e+00
-4.09424514e-01 -1.05961159e-01 -9.59691644e-01 9.90228355e-02
4.51212861e-02 1.46329045e+00 -4.10832107e-01 4.08061519e-02
-6.17939569e-02 1.32852316e+00 2.56654233e-01 3.04947555e-01
-7.89533913e-01 -7.39903562e-03 1.03736162e+00 3.01631898e-01
3.90423656e-01 -3.80191058e-02 -1.66848339e-02 -1.26694810e+00
-1.93680339e-02 -1.24754012e+00 -6.13449886e-02 -7.39403963e-01
-1.08904016e+00 9.60583150e-01 4.31654118e-02 -7.76261687e-01
-6.24665737e-01 -6.34524763e-01 -6.56711400e-01 8.63430977e-01
-1.09650242e+00 -1.71793520e+00 -4.03758198e-01 8.37117970e-01
2.33192801e-01 -1.71398073e-01 1.14550066e+00 2.29501814e-01
-4.87672925e-01 9.96654987e-01 -3.09232712e-01 2.34986991e-01
7.90783823e-01 -9.84561205e-01 4.82584387e-01 6.15364730e-01
2.49203011e-01 5.41079998e-01 8.10755372e-01 -2.76629031e-01
-1.56509864e+00 -9.71457303e-01 4.60380763e-01 -7.52478361e-01
2.59296298e-01 -1.16198134e+00 -5.09746552e-01 1.03480518e+00
4.10181463e-01 1.33568063e-01 5.54128468e-01 1.58402696e-02
-5.71917295e-01 -2.65263945e-01 -1.39105201e+00 8.45330179e-01
1.25970650e+00 -6.03316188e-01 -7.94027224e-02 1.30171701e-01
5.43369234e-01 -4.66751337e-01 -6.37844622e-01 3.58946443e-01
8.55477035e-01 -1.15077031e+00 8.88290048e-01 -6.02502644e-01
4.11405593e-01 -1.74636304e-01 -8.91009420e-02 -1.51905715e+00
-7.36677349e-02 -1.10054517e+00 1.66563898e-01 1.53450954e+00
1.98710874e-01 -8.42387795e-01 7.73685515e-01 5.20793557e-01
3.87389898e-01 -6.28561795e-01 -8.98625791e-01 -4.51432675e-01
1.90072015e-01 9.42481533e-02 7.74278462e-01 1.00629008e+00
-3.99210930e-01 4.79005069e-01 -6.46334112e-01 -4.54228139e-03
6.37460530e-01 3.02576702e-02 1.19532871e+00 -9.84713197e-01
-2.40198851e-01 -2.62956887e-01 -5.80055714e-01 -5.98855853e-01
6.14654064e-01 -8.49086821e-01 -1.50732845e-01 -7.70103276e-01
9.78017822e-02 -2.41309091e-01 1.74914613e-01 7.33612001e-01
3.07646781e-01 8.67197335e-01 3.03112507e-01 -1.34860069e-01
1.80922791e-01 7.21804559e-01 1.41849554e+00 1.16848424e-01
-7.91401230e-03 -9.67402384e-02 -8.65945160e-01 6.24778390e-01
6.32360995e-01 -4.18999702e-01 -3.84474933e-01 -3.08728129e-01
1.62901625e-01 -2.23551765e-02 7.07651615e-01 -7.66395330e-01
-1.21162668e-01 1.02208421e-01 5.46596408e-01 2.67480493e-01
8.00792754e-01 -8.60263765e-01 6.64879084e-01 2.80921251e-01
-1.66050598e-01 -3.86085920e-02 3.40337843e-01 1.63144931e-01
-2.47556567e-02 2.19578311e-01 1.08326483e+00 -7.13052377e-02
-1.86877444e-01 5.41985333e-01 1.23822978e-02 -8.67889822e-02
9.92114425e-01 -4.04968746e-02 -4.27595805e-03 -7.82061815e-01
-7.40847111e-01 -2.86860108e-01 9.32214320e-01 5.67865431e-01
4.35532540e-01 -1.57462347e+00 -9.99017835e-01 7.18546212e-01
-9.55113024e-03 -1.18531480e-01 7.11239576e-02 6.36028230e-01
-3.75837058e-01 3.79338190e-02 -6.83829963e-01 -5.56200862e-01
-1.53312683e+00 3.10511410e-01 5.81789494e-01 7.15560317e-02
-5.11920512e-01 9.30531621e-01 5.68021536e-01 -5.94648302e-01
-2.77614653e-01 2.31898174e-01 8.21771771e-02 4.33349647e-02
3.15005362e-01 -6.46722615e-02 -1.55302677e-02 -1.06714594e+00
-2.04618484e-01 8.62551332e-01 -2.45380457e-02 -3.37168902e-01
1.52581906e+00 5.96970655e-02 -3.69917184e-01 1.01522163e-01
1.51070392e+00 4.98392910e-01 -1.38021910e+00 1.80795535e-01
-6.64516211e-01 -6.98070645e-01 -1.91739112e-01 -6.05366588e-01
-1.59744310e+00 7.53537655e-01 6.08186305e-01 -2.68003106e-01
1.17129242e+00 1.04508169e-01 6.17167354e-01 -1.29454702e-01
4.36796159e-01 -4.21613812e-01 1.68784857e-01 1.88952237e-01
1.31638002e+00 -1.10818207e+00 -2.44250521e-01 -4.90936965e-01
-3.68558943e-01 8.59804392e-01 5.61685860e-01 -1.72436073e-01
5.96274376e-01 3.10731709e-01 3.05893779e-01 -2.91051298e-01
-5.42576194e-01 1.64113358e-01 3.27695310e-01 6.03871465e-01
3.64634365e-01 1.16864562e-01 3.02484363e-01 1.26642689e-01
-8.26268733e-01 -2.01820850e-01 4.08376545e-01 5.26411831e-01
3.86325181e-01 -1.37813377e+00 -3.04109424e-01 -3.00150253e-02
-4.92913902e-01 1.27261847e-01 -5.84178388e-01 8.75437558e-01
1.51717201e-01 5.88887036e-01 3.71746309e-02 -3.91342968e-01
3.15096676e-01 6.64836466e-02 1.04820824e+00 -6.97960973e-01
-3.68021548e-01 -1.56108022e-01 -5.29858917e-02 -8.02123964e-01
-2.85239279e-01 -5.68068266e-01 -7.47484863e-01 -6.58580780e-01
8.88395123e-03 -1.64179415e-01 5.55639148e-01 6.96163237e-01
4.90850419e-01 3.11025143e-01 8.08123171e-01 -1.28494406e+00
-1.68038353e-01 -7.91570723e-01 -6.05603218e-01 8.04851770e-01
6.10771656e-01 -7.11512923e-01 -4.24932003e-01 1.94538400e-01] | [12.648117065429688, -0.20455901324748993] |
e51da4df-4415-43bd-aeeb-ba0dd7754f16 | interactive-concept-learning-for-uncovering | 2305.05094 | null | https://arxiv.org/abs/2305.05094v1 | https://arxiv.org/pdf/2305.05094v1.pdf | Interactive Concept Learning for Uncovering Latent Themes in Large Text Collections | Experts across diverse disciplines are often interested in making sense of large text collections. Traditionally, this challenge is approached either by noisy unsupervised techniques such as topic models, or by following a manual theme discovery process. In this paper, we expand the definition of a theme to account for more than just a word distribution, and include generalized concepts deemed relevant by domain experts. Then, we propose an interactive framework that receives and encodes expert feedback at different levels of abstraction. Our framework strikes a balance between automation and manual coding, allowing experts to maintain control of their study while reducing the manual effort required. | ['Dan Goldwasser', 'Ming Yin', 'Lyle Ungar', 'Tunazzina Islam', 'Maria Leonor Pacheco'] | 2023-05-08 | null | null | null | null | ['topic-models'] | ['natural-language-processing'] | [ 8.86580944e-02 4.46417630e-01 -7.89451599e-02 -1.88459218e-01
-1.00659525e+00 -9.71799135e-01 5.48202753e-01 6.81488216e-01
-3.81865799e-01 5.56135178e-01 4.41760123e-01 -4.18769956e-01
-1.47273615e-01 -7.47916698e-01 -1.73396438e-01 -3.22457731e-01
2.93868631e-01 4.60881829e-01 2.23536059e-01 4.40263934e-02
6.11687183e-01 3.09063215e-02 -1.63721621e+00 1.76976278e-01
1.13990104e+00 5.79630077e-01 3.20098996e-01 2.51121819e-01
-7.39304900e-01 6.39983296e-01 -8.63655806e-01 -3.79676580e-01
-1.41029119e-01 -2.35743895e-01 -1.02893138e+00 2.80027121e-01
-6.69475943e-02 2.17105404e-01 4.73924339e-01 1.21081853e+00
7.02304468e-02 3.47507119e-01 6.13110125e-01 -1.10528922e+00
-5.01563847e-01 8.87440622e-01 -4.16326076e-01 8.07031915e-02
6.70312941e-01 -1.53527498e-01 1.15145075e+00 -9.61023033e-01
7.83664227e-01 1.30267823e+00 4.92075711e-01 8.96858424e-02
-1.33806479e+00 -5.55637598e-01 7.21875191e-01 3.18560824e-02
-1.64587128e+00 -1.56718910e-01 7.24936366e-01 -1.06902671e+00
7.74328232e-01 3.39826494e-01 5.32253921e-01 1.11307812e+00
1.13904297e-01 4.12211508e-01 9.28212166e-01 -5.06990314e-01
8.56525302e-01 5.01006663e-01 4.18457121e-01 1.20276116e-01
5.29866159e-01 -5.72384953e-01 -7.15719283e-01 -6.32871211e-01
4.75740910e-01 1.13188818e-01 -4.40430880e-01 -1.60560563e-01
-1.10644710e+00 9.95602071e-01 -1.75486848e-01 5.32246232e-01
-7.06655562e-01 -3.77908796e-01 2.12105170e-01 2.52350718e-01
8.77094328e-01 9.64933336e-01 -4.02374119e-01 -3.02060157e-01
-1.23800826e+00 4.95393306e-01 1.21634781e+00 1.08488965e+00
8.92266393e-01 -6.28043294e-01 -6.17311150e-02 6.72770381e-01
5.05138576e-01 1.11089647e-01 5.10337770e-01 -8.54693770e-01
1.99339204e-02 9.40183699e-01 3.96444052e-01 -1.16555572e+00
-1.40794575e-01 -3.64327073e-01 -4.29289907e-01 2.07375854e-01
1.41695037e-01 -2.49059588e-01 -8.33515227e-01 1.47473764e+00
3.72048438e-01 -3.11447710e-01 -2.99265593e-01 7.50036538e-01
4.92975116e-01 4.74486887e-01 3.76335800e-01 -4.01222795e-01
1.58190095e+00 -5.22782505e-01 -1.12334096e+00 -2.06748113e-01
3.88960958e-01 -7.14324176e-01 1.05583358e+00 9.09163773e-01
-1.02260506e+00 -2.04977036e-01 -8.01191926e-01 -1.87405959e-01
-6.88799202e-01 -2.83010334e-01 1.88782960e-01 3.89026612e-01
-1.14558685e+00 3.92966777e-01 -8.12604785e-01 -5.89683771e-01
2.49232829e-01 -8.50655138e-02 1.63622145e-02 6.83569834e-02
-1.18490314e+00 7.88619757e-01 4.23431158e-01 -2.42556334e-01
-6.42834783e-01 -8.88890028e-01 -6.93604827e-01 2.18770295e-01
7.60143638e-01 -7.62061894e-01 1.56836045e+00 -6.36645675e-01
-1.08454514e+00 7.00474739e-01 -4.18219566e-01 -1.61259070e-01
3.24281067e-01 -6.19194567e-01 -2.32114464e-01 1.77201495e-01
4.39088285e-01 7.75242150e-02 5.93574405e-01 -1.39551270e+00
-6.59306765e-01 -3.32881838e-01 1.64349154e-01 1.27871811e-01
-5.93747199e-01 3.27037394e-01 -3.81057739e-01 -8.04329634e-01
1.42715320e-01 -5.93502641e-01 -5.97632468e-01 2.01001689e-02
-5.49841166e-01 -5.14484882e-01 6.98484719e-01 -4.57370073e-01
1.72951221e+00 -1.96780181e+00 2.06352010e-01 5.13875067e-01
8.32751274e-01 -2.81411052e-01 4.74316210e-01 7.72564650e-01
1.36009097e-01 6.35449886e-01 -2.65807509e-01 -4.71355408e-01
8.16804618e-02 -1.49614243e-02 -5.22894502e-01 3.68738733e-02
4.66784164e-02 4.19140041e-01 -1.14682269e+00 -6.70572877e-01
-7.51240328e-02 3.29180270e-01 -3.79727840e-01 2.09747821e-01
-5.39594769e-01 1.56596765e-01 -8.29175055e-01 6.21464491e-01
5.15173137e-01 -7.58381546e-01 4.67849463e-01 2.07360417e-01
-5.25203109e-01 5.92522502e-01 -1.36621988e+00 1.89370799e+00
-4.40486491e-01 5.81853807e-01 3.63570452e-01 -8.06814790e-01
8.50649774e-01 5.96580029e-01 3.70764315e-01 2.24331111e-01
6.25988171e-02 5.47936792e-03 -5.25544107e-01 -4.97709602e-01
7.17437744e-01 -1.02331564e-01 -1.48469597e-01 8.69175673e-01
8.99167731e-02 -2.55763233e-01 1.69399738e-01 5.07788181e-01
1.06301403e+00 -1.56553522e-01 7.01479256e-01 -5.73690832e-01
3.47072328e-03 4.15277332e-01 3.51362705e-01 9.97744024e-01
1.17439657e-01 4.35594589e-01 4.62188154e-01 -1.27972826e-01
-7.76492059e-01 -9.02800143e-01 -1.71580568e-01 1.27346945e+00
1.37021318e-02 -9.78080809e-01 -6.12990856e-01 -4.45415735e-01
-6.74497113e-02 7.74475336e-01 -7.70952642e-01 1.02047205e-01
1.11440547e-01 -3.12264383e-01 1.13795497e-01 3.38145733e-01
2.90565044e-02 -8.32431674e-01 -8.74355972e-01 3.71558994e-01
-3.75464380e-01 -8.33603024e-01 -2.13543162e-01 2.20449865e-01
-7.70047724e-01 -8.61145377e-01 -6.48076952e-01 -4.46872681e-01
5.15818834e-01 3.59313637e-01 1.35519207e+00 -1.14280172e-02
-3.62856060e-01 5.38662195e-01 -7.61209190e-01 -6.89939618e-01
-1.47818893e-01 2.87516028e-01 -1.21260732e-02 -2.32451469e-01
8.06663752e-01 -8.28955114e-01 -4.11274433e-01 1.50347739e-01
-1.18715477e+00 -2.11444214e-01 3.52203727e-01 3.77987772e-01
3.82382751e-01 6.37210980e-02 4.27855045e-01 -1.12176847e+00
1.20648253e+00 -8.93858194e-01 -3.52738321e-01 3.48558545e-01
-8.77798617e-01 -1.80970386e-01 3.68497193e-01 -3.95785719e-01
-1.05898893e+00 -2.63514638e-01 2.54769593e-01 -3.43436658e-01
-4.83023643e-01 1.03662264e+00 -9.34757963e-02 3.33834678e-01
9.04788256e-01 -9.72128510e-02 -4.24583882e-01 -6.19425297e-01
4.86723840e-01 8.95008862e-01 2.04479352e-01 -7.81728625e-01
7.59692788e-01 3.36821258e-01 -8.00097942e-01 -7.91394413e-01
-9.98709142e-01 -7.72084355e-01 -6.46960557e-01 -2.89384723e-01
5.55561364e-01 -9.25180316e-01 -3.89266193e-01 -1.99375108e-01
-1.27638459e+00 3.13723050e-02 -4.72082198e-01 3.81293148e-01
-7.30954632e-02 2.57644832e-01 -2.05370501e-01 -9.88444984e-01
-2.97068864e-01 -8.10519397e-01 9.99559045e-01 2.41227806e-01
-9.45633292e-01 -1.07849300e+00 2.24967152e-01 -4.47660536e-02
5.25362015e-01 2.42760062e-01 6.70684159e-01 -1.02450800e+00
-3.67419660e-01 -2.99706727e-01 -1.38299521e-02 -2.19880402e-01
4.57870662e-01 -1.39207216e-02 -1.13235259e+00 -3.18284333e-02
2.14890420e-01 -2.48990774e-01 5.68728209e-01 1.47592098e-01
1.25592864e+00 -3.30734670e-01 -7.29695737e-01 4.67226282e-02
1.23909390e+00 1.52880535e-01 1.12380698e-01 5.32132626e-01
4.22971785e-01 8.29487145e-01 4.71375495e-01 8.09905887e-01
4.95339602e-01 3.16945136e-01 7.28443339e-02 4.39327024e-02
7.97249079e-01 -1.35133550e-01 -2.49534771e-01 5.75585783e-01
-5.19130193e-02 -4.77032274e-01 -1.28052044e+00 9.46621478e-01
-1.96116459e+00 -6.48942471e-01 1.19433060e-01 1.91029060e+00
9.79608893e-01 1.95204273e-01 1.18152045e-01 5.87154925e-02
6.88725173e-01 3.37365121e-02 -3.04591596e-01 -2.45869800e-01
3.62469167e-01 -7.50810979e-03 7.37795159e-02 5.63460588e-01
-8.21593106e-01 7.70561934e-01 7.27415657e+00 6.41300857e-01
-6.51455283e-01 6.79049361e-03 3.08877885e-01 -1.63941547e-01
-7.06038654e-01 2.54597366e-01 -5.66084743e-01 3.28790456e-01
7.17535138e-01 -6.79957628e-01 6.01305589e-02 9.42840099e-01
3.89374137e-01 -2.32775107e-01 -1.08111465e+00 7.36355186e-01
2.99046021e-02 -1.10256767e+00 1.72463655e-01 2.40080029e-01
5.93738616e-01 -3.53855133e-01 -2.34489083e-01 4.00132462e-02
1.01076937e+00 -1.02895260e+00 6.23094082e-01 7.35847950e-01
4.54321593e-01 -4.49261487e-01 5.71800470e-01 5.00428081e-01
-1.11587131e+00 7.90835619e-02 -2.74887607e-02 -2.44865224e-01
3.48656952e-01 1.06844151e+00 -7.75556266e-01 6.25801027e-01
8.94231260e-01 5.19605875e-01 -2.70493478e-01 1.04582560e+00
-2.26133943e-01 6.94393754e-01 -3.37310970e-01 -1.13337591e-01
6.32564500e-02 -7.40274116e-02 6.76429927e-01 1.57347143e+00
3.92655730e-01 3.12537730e-01 7.63784468e-01 1.12777984e+00
1.19689561e-01 2.42480561e-01 -6.92087352e-01 -2.47028768e-01
9.53752279e-01 1.23847544e+00 -8.92497301e-01 -6.03860378e-01
-1.13855183e-01 5.47118008e-01 1.66699335e-01 6.12694621e-01
-1.01052344e-01 -6.90035999e-01 7.19823658e-01 3.02415639e-01
2.16792151e-01 -3.06815118e-01 -5.74635148e-01 -1.10775304e+00
1.64860114e-01 -8.05750787e-01 4.18435454e-01 -8.36449385e-01
-1.26654279e+00 7.38313973e-01 3.72303098e-01 -1.08242464e+00
-3.87630433e-01 -3.86332273e-01 -7.04599440e-01 1.03729105e+00
-1.10565555e+00 -5.68969965e-01 -3.98988992e-01 3.20983231e-01
6.65243864e-01 2.35781789e-01 8.78889561e-01 -1.02950580e-01
-3.08966070e-01 -1.49908587e-02 -1.57667682e-01 -1.86559081e-01
6.86311722e-01 -1.40098107e+00 2.75008470e-01 7.32061684e-01
-8.87870118e-02 1.36168051e+00 1.12316537e+00 -8.90654922e-01
-9.24649298e-01 -8.03946078e-01 1.14834368e+00 -6.02762699e-01
1.00099838e+00 -5.72664797e-01 -1.27896774e+00 6.26885295e-01
4.12594825e-01 -6.03379607e-01 1.11888409e+00 6.49226725e-01
-3.57105672e-01 4.25203145e-01 -9.25577462e-01 4.32171196e-01
7.89680183e-01 -6.54690385e-01 -1.00441682e+00 3.14479351e-01
8.51006985e-01 -6.02239408e-02 -8.08765769e-01 -1.57690838e-01
2.21612632e-01 -3.90837520e-01 5.81580162e-01 -6.53887331e-01
3.97638410e-01 -3.92938048e-01 -5.37134428e-03 -1.45280290e+00
-2.95815080e-01 -8.31660211e-01 -1.99512225e-02 1.31530321e+00
5.03561974e-01 -2.88266152e-01 3.48010629e-01 8.43131065e-01
6.89710230e-02 -3.03072304e-01 -6.10268593e-01 -3.98983449e-01
3.34045589e-02 -7.70324349e-01 3.77625227e-01 1.18130624e+00
8.15439820e-01 3.05605710e-01 8.60243961e-02 1.62661031e-01
4.26515669e-01 1.66938499e-01 4.43191200e-01 -1.95232463e+00
6.31267428e-02 -4.50973451e-01 -3.61957140e-02 -1.01408291e+00
-2.31168374e-01 -5.48000574e-01 4.43506330e-01 -1.76268220e+00
2.12735608e-01 -1.40154324e-02 -1.71288431e-01 3.64760458e-01
-3.16493392e-01 -1.57403678e-01 -2.39378154e-01 3.47609013e-01
-1.03930140e+00 2.89526463e-01 7.06814349e-01 1.11126013e-01
-3.32850158e-01 -2.40031078e-01 -1.43274772e+00 9.98833477e-01
5.96288919e-01 -7.26517260e-01 -5.25131941e-01 -3.49794596e-01
6.73931897e-01 -3.60167742e-01 4.28445965e-01 -7.82261074e-01
6.48112953e-01 -3.65310997e-01 6.02707826e-02 -4.39609766e-01
-2.64906101e-02 -6.97773635e-01 -1.34676909e-02 -1.42657310e-01
-6.91460669e-01 -7.17011699e-03 1.95746511e-01 7.34207809e-01
-3.88402432e-01 -3.43278617e-01 2.23707110e-01 -5.15946627e-01
-2.48122811e-01 -5.60731180e-02 -9.21446979e-01 1.08238764e-01
7.38868654e-01 -1.35453239e-01 -9.81250033e-02 -5.12795866e-01
-1.10053349e+00 4.88114655e-01 5.01779735e-01 3.54046345e-01
3.61423403e-01 -8.61776710e-01 -4.87135828e-01 -1.77473053e-01
3.31738263e-01 3.56955349e-01 -6.05142899e-02 4.12569761e-01
-4.33340073e-02 3.96366686e-01 3.51259559e-01 -4.85609353e-01
-8.10614526e-01 4.99759793e-01 -8.93368647e-02 -3.11416060e-01
-7.79660165e-01 6.91586018e-01 1.59762219e-01 -2.29435056e-01
4.23735917e-01 -3.66113693e-01 -5.24065316e-01 5.17560542e-01
8.90912712e-01 2.63873160e-01 6.36838675e-02 -1.22310497e-01
-3.83531898e-01 3.87572467e-01 -2.64102757e-01 -4.51442719e-01
1.30492508e+00 -4.72454101e-01 -1.25640303e-01 9.79926050e-01
6.15534663e-01 5.56796743e-03 -8.45456779e-01 -5.07465601e-01
5.23369551e-01 -2.91636616e-01 2.32358441e-01 -9.80493724e-01
-3.01600099e-01 6.80378020e-01 1.66648358e-01 9.94772375e-01
9.60128009e-01 2.96346128e-01 2.48642817e-01 5.44014573e-01
2.75736570e-01 -1.22252738e+00 -1.80055663e-01 4.68831837e-01
8.49360645e-01 -1.11867452e+00 -1.04089808e-02 -5.21702230e-01
-5.74201465e-01 1.13579440e+00 4.88376677e-01 1.39752999e-01
9.48202670e-01 4.03381735e-01 2.87322819e-01 -5.94436109e-01
-9.88942266e-01 -2.24822745e-01 3.07191819e-01 4.52363998e-01
7.21848547e-01 -1.82721272e-01 -4.60167825e-01 9.53716457e-01
-2.53484070e-01 3.81123513e-01 4.61082608e-01 1.32490301e+00
-8.85856450e-01 -8.99616659e-01 -6.13109529e-01 4.73111778e-01
-6.76777840e-01 -1.50700003e-01 -7.15196252e-01 5.51077902e-01
-5.46376780e-02 1.31010246e+00 1.06512100e-01 -1.90944940e-01
3.09415102e-01 3.76254678e-01 -4.47329432e-01 -1.12581599e+00
-5.22177994e-01 4.55052823e-01 -9.46234241e-02 -2.50022173e-01
-5.20327270e-01 -6.71892583e-01 -7.94348836e-01 6.02268465e-02
-3.14231485e-01 8.77111018e-01 6.08268738e-01 1.19989216e+00
4.35025156e-01 6.10635817e-01 3.46554726e-01 -4.96807754e-01
-3.73734802e-01 -1.31756997e+00 -5.77163160e-01 1.42773256e-01
2.12439418e-01 -7.61472166e-01 -4.90878940e-01 4.42470700e-01] | [10.395886421203613, 7.121559143066406] |
00c3a6f6-3d85-40ce-b3a4-72afdafda803 | shape-aware-multi-person-pose-estimation-from-1 | 2110.02330 | null | https://arxiv.org/abs/2110.02330v1 | https://arxiv.org/pdf/2110.02330v1.pdf | Shape-aware Multi-Person Pose Estimation from Multi-View Images | In this paper we contribute a simple yet effective approach for estimating 3D poses of multiple people from multi-view images. Our proposed coarse-to-fine pipeline first aggregates noisy 2D observations from multiple camera views into 3D space and then associates them into individual instances based on a confidence-aware majority voting technique. The final pose estimates are attained from a novel optimization scheme which links high-confidence multi-view 2D observations and 3D joint candidates. Moreover, a statistical parametric body model such as SMPL is leveraged as a regularizing prior for these 3D joint candidates. Specifically, both 3D poses and SMPL parameters are optimized jointly in an alternating fashion. Here the parametric models help in correcting implausible 3D pose estimates and filling in missing joint detections while updated 3D poses in turn guide obtaining better SMPL estimations. By linking 2D and 3D observations, our method is both accurate and generalizes to different data sources because it better decouples the final 3D pose from the inter-person constellation and is more robust to noisy 2D detections. We systematically evaluate our method on public datasets and achieve state-of-the-art performance. The code and video will be available on the project page: https://ait.ethz.ch/projects/2021/multi-human-pose/. | ['Otmar Hilliges', 'Chen Guo', 'Xu Chen', 'Jie Song', 'Zijian Dong'] | 2021-10-05 | shape-aware-multi-person-pose-estimation-from | http://openaccess.thecvf.com//content/ICCV2021/html/Dong_Shape-Aware_Multi-Person_Pose_Estimation_From_Multi-View_Images_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Dong_Shape-Aware_Multi-Person_Pose_Estimation_From_Multi-View_Images_ICCV_2021_paper.pdf | iccv-2021-1 | ['3d-multi-person-pose-estimation'] | ['computer-vision'] | [-2.59535402e-01 -4.66150194e-02 -1.18425198e-01 -4.56712276e-01
-1.26082492e+00 -6.98579252e-01 6.57245219e-01 -1.52450249e-01
-4.29643899e-01 5.69819570e-01 3.52573216e-01 5.03938556e-01
1.18984558e-01 -3.77137214e-01 -7.98260629e-01 -3.66960287e-01
-1.64757166e-02 9.56091821e-01 2.93883413e-01 6.84403256e-02
-1.61052689e-01 4.13639456e-01 -1.39268744e+00 -9.73875299e-02
4.83903319e-01 7.33104646e-01 -1.13251403e-01 7.68933892e-01
5.12300968e-01 -1.07141577e-01 -4.94006246e-01 -7.70226240e-01
5.41531801e-01 -7.79436156e-02 -3.31711859e-01 5.49133658e-01
9.82311010e-01 -3.91426533e-01 5.71679324e-02 1.01119685e+00
6.26714408e-01 1.37359388e-02 4.87617463e-01 -1.17001510e+00
-9.61116627e-02 -5.24767414e-02 -1.00193501e+00 -1.31997302e-01
1.01020789e+00 1.52841836e-01 8.68243337e-01 -1.15709817e+00
5.49657345e-01 1.54405248e+00 1.01810884e+00 3.54303420e-01
-1.24757040e+00 -4.48785454e-01 3.34686786e-01 -5.21092527e-02
-1.69533372e+00 -3.95882070e-01 6.58414483e-01 -4.91642207e-01
5.33923566e-01 3.01687449e-01 9.81526017e-01 1.27018189e+00
3.05446200e-02 8.20027649e-01 9.51451778e-01 -1.56035393e-01
-1.63476527e-01 -7.10986461e-03 -6.74185529e-02 8.28410327e-01
5.92468739e-01 9.07396600e-02 -7.10007012e-01 -3.89076620e-01
9.18503404e-01 2.21118242e-01 4.60363850e-02 -6.46726847e-01
-1.37394643e+00 5.67218363e-01 2.91008264e-01 -2.96294659e-01
-4.25601512e-01 1.26810744e-01 6.76897243e-02 -2.55221993e-01
4.63186622e-01 4.99858148e-02 -5.26512086e-01 -6.24323897e-02
-8.43536735e-01 7.15215087e-01 4.10522580e-01 9.56957936e-01
6.98659718e-01 -4.33021754e-01 -4.88286316e-02 8.46270800e-01
6.68464601e-01 8.03777277e-01 -1.06507920e-01 -1.24948204e+00
6.08304739e-01 5.65662146e-01 5.41126728e-01 -1.04018390e+00
-6.59022987e-01 -5.00645876e-01 -6.33463979e-01 -4.72120643e-02
4.96532291e-01 -2.78366089e-01 -7.38276184e-01 1.74280798e+00
9.09907997e-01 -3.35427970e-02 -3.93875718e-01 1.30719459e+00
7.47809529e-01 1.38569683e-01 -2.86810875e-01 8.35101702e-04
1.66485310e+00 -8.42345357e-01 -4.42593783e-01 -6.60570621e-01
8.64998251e-02 -7.64323711e-01 2.98819602e-01 4.80035424e-01
-1.16067088e+00 -7.64431655e-01 -8.58559072e-01 7.29698862e-04
7.31124058e-02 6.29841924e-01 2.38319427e-01 7.88670599e-01
-9.21882510e-01 2.08509639e-01 -8.76424849e-01 -4.97832119e-01
1.42266169e-01 4.75301892e-01 -6.82007134e-01 3.06497235e-02
-8.77658904e-01 9.56800282e-01 1.78100139e-01 2.96578944e-01
-5.82449555e-01 -2.44172156e-01 -1.04921889e+00 -6.03856921e-01
8.41817081e-01 -1.05778360e+00 1.19744503e+00 -2.03545347e-01
-1.26434422e+00 1.10008872e+00 -3.49682361e-01 -2.81168997e-01
8.58220577e-01 -7.06310332e-01 -2.35998154e-01 2.05733687e-01
4.00213778e-01 7.79437721e-01 9.34677601e-01 -1.37952781e+00
-6.30469680e-01 -7.85096228e-01 -1.68653324e-01 4.41618472e-01
2.10160732e-01 -1.49641961e-01 -1.07648242e+00 -6.25819743e-01
5.73732376e-01 -1.24364090e+00 -3.47775847e-01 2.04924226e-01
-6.27999008e-01 -7.84952343e-02 2.58593768e-01 -7.68816590e-01
7.85215974e-01 -1.69714451e+00 5.27680695e-01 3.29034239e-01
3.73802423e-01 -7.73667172e-02 1.67242646e-01 2.64311910e-01
2.42031887e-01 -2.83799320e-01 -2.63120490e-03 -1.06068957e+00
5.27053997e-02 6.01484217e-02 2.53593206e-01 9.76257443e-01
1.12177730e-01 8.70816827e-01 -7.40077853e-01 -5.11079073e-01
4.49521869e-01 6.03651583e-01 -6.19755208e-01 1.92173347e-01
1.21974774e-01 7.97229588e-01 -4.30359006e-01 7.52857506e-01
8.32287014e-01 -2.43931979e-01 2.50558943e-01 -3.46210837e-01
-6.09738417e-02 -6.80851936e-02 -1.72327578e+00 1.63929677e+00
1.44419834e-01 -4.85468656e-02 1.55520827e-01 -5.91966987e-01
8.10664237e-01 2.93704808e-01 5.21577418e-01 -2.02938672e-02
1.23269454e-01 4.48152460e-02 -5.68797886e-01 -2.62096077e-01
6.02724314e-01 5.27801253e-02 -4.63270277e-01 6.56071752e-02
2.01733708e-01 -6.00590333e-02 4.92329933e-02 4.71584536e-02
5.49094439e-01 5.91853619e-01 6.71430886e-01 8.60312432e-02
4.86021280e-01 -3.32566440e-01 7.35023797e-01 7.83722043e-01
-3.04654926e-01 9.71200824e-01 1.73868433e-01 -4.73343909e-01
-9.92202461e-01 -1.39382970e+00 -1.05449952e-01 7.21527576e-01
2.11869135e-01 -5.82339227e-01 -5.47658920e-01 -8.11786413e-01
3.41095209e-01 1.36968404e-01 -6.88803792e-01 1.87519073e-01
-5.66115618e-01 -6.36239946e-01 3.62521917e-01 4.59276080e-01
3.85996521e-01 -1.56316057e-01 -3.69379580e-01 -5.29533289e-02
-4.98277664e-01 -1.31900120e+00 -8.13365996e-01 -2.06253141e-01
-7.08607793e-01 -1.21520007e+00 -9.17347848e-01 -2.49961272e-01
7.26303756e-01 4.21311706e-01 9.82591033e-01 -6.63202032e-02
-6.21286184e-02 6.50928855e-01 -2.71814764e-01 -1.91745862e-01
1.62586458e-02 -3.00081819e-01 7.24795461e-01 1.79911837e-01
3.64731967e-01 -3.22622985e-01 -6.19835079e-01 5.70072114e-01
-6.26200512e-02 -6.74235597e-02 4.07325447e-01 5.74658513e-01
7.59643078e-01 -3.97318810e-01 1.16089024e-01 -5.07234156e-01
2.80093569e-02 -2.93819785e-01 -7.07888722e-01 3.01083233e-02
-7.50946403e-02 -1.27722353e-01 3.42166275e-02 -3.05680484e-01
-8.73184681e-01 5.11404157e-01 -1.94585815e-01 -4.57097441e-01
-3.47610533e-01 1.78381130e-02 -2.56887704e-01 1.17732316e-01
5.17263353e-01 -5.25710359e-03 -4.09690440e-02 -7.27825701e-01
3.10191095e-01 4.16446239e-01 7.10633278e-01 -6.29894435e-01
1.03308916e+00 6.59938037e-01 2.70808619e-02 -7.46360779e-01
-1.03497910e+00 -8.19171429e-01 -9.43424225e-01 -4.53837723e-01
8.95722449e-01 -1.41071761e+00 -7.97788680e-01 5.34693480e-01
-1.13790679e+00 2.60366857e-01 3.09972931e-02 6.46999598e-01
-4.84287739e-01 7.85914361e-01 -4.11460578e-01 -1.21680617e+00
-7.29038566e-02 -1.21149123e+00 1.83125532e+00 1.43042147e-01
-5.49713731e-01 -6.11064196e-01 2.16291577e-01 7.27094829e-01
-4.07956541e-01 3.24557841e-01 -3.50786150e-01 -3.16332012e-01
-4.35031831e-01 -5.65152287e-01 1.00478455e-01 5.96888773e-02
-1.13327324e-01 -1.61935583e-01 -8.35121930e-01 -4.29919034e-01
-2.32183203e-01 -2.34688252e-01 7.65917540e-01 7.04405844e-01
5.51568210e-01 -5.43087311e-02 -4.51923281e-01 4.67061579e-01
8.72348011e-01 -6.65901244e-01 2.45794892e-01 2.96811223e-01
7.49857366e-01 6.00623310e-01 7.94550121e-01 7.22108603e-01
9.49521482e-01 1.32997310e+00 4.58403379e-01 2.33932719e-01
1.33051286e-02 -4.02905613e-01 4.55316812e-01 4.68802840e-01
-4.67912346e-01 9.43884999e-02 -6.80421114e-01 3.55750889e-01
-1.93889332e+00 -8.87598395e-01 -3.35852385e-01 2.51812911e+00
6.23835921e-01 1.20024234e-01 7.38591373e-01 -1.67411745e-01
9.43329632e-01 1.15174331e-01 -3.02273005e-01 5.62625110e-01
1.76269580e-02 -3.07797045e-01 4.60951239e-01 6.64072514e-01
-1.34737897e+00 6.52113497e-01 5.51384306e+00 4.78440911e-01
-3.06712121e-01 1.63045332e-01 2.37584606e-01 -4.04191554e-01
1.75222620e-01 -1.44615456e-01 -1.38878453e+00 4.78754014e-01
4.27480280e-01 4.90459800e-01 1.08989894e-01 6.47656739e-01
2.05037102e-01 -3.64159822e-01 -9.45285559e-01 1.32467127e+00
1.06275246e-01 -1.04840136e+00 -2.05340207e-01 3.32268775e-01
7.18968093e-01 -3.95499021e-02 -9.19617862e-02 -1.65935699e-02
2.33224809e-01 -4.99783069e-01 1.19171441e+00 6.40729606e-01
7.09002852e-01 -7.17621088e-01 7.51486301e-01 5.16652644e-01
-1.42495644e+00 5.29676042e-02 -1.55559629e-01 -1.00748532e-01
5.28753996e-01 7.53082335e-01 -6.94825411e-01 7.44217753e-01
9.35198903e-01 7.61812985e-01 -6.30550265e-01 1.03338265e+00
-3.95150036e-01 1.58912186e-02 -6.52756870e-01 2.21403614e-01
-2.39356458e-01 -1.81764990e-01 9.51810718e-01 1.04639661e+00
4.10778254e-01 -9.08623561e-02 6.18860245e-01 5.26547134e-01
1.51152268e-01 -3.69582742e-01 -3.16215605e-01 6.31758690e-01
6.11706018e-01 1.31479478e+00 -6.32328689e-01 -3.10517043e-01
-3.07611048e-01 9.71605361e-01 3.09530020e-01 1.44255355e-01
-1.00365162e+00 4.80946749e-01 7.99193084e-01 1.85235500e-01
4.70929056e-01 -4.69085008e-01 -1.78321421e-01 -1.60987377e+00
4.66558963e-01 -8.03091705e-01 4.72839117e-01 -6.94795787e-01
-1.36250281e+00 3.54010403e-01 3.47937763e-01 -1.53440571e+00
-5.58619082e-01 -5.80603302e-01 -1.22417867e-01 7.12566674e-01
-9.41423893e-01 -1.37419379e+00 -1.76692516e-01 4.47151423e-01
3.78798753e-01 5.40122613e-02 6.86345220e-01 2.07947284e-01
-5.22116840e-01 5.99370182e-01 -3.94892752e-01 1.64602906e-01
1.04061091e+00 -1.20444715e+00 5.94994605e-01 9.33110893e-01
1.34223744e-01 6.26839519e-01 9.36220527e-01 -9.82317448e-01
-1.37475324e+00 -9.79326308e-01 7.91924000e-01 -1.17508996e+00
3.01442534e-01 -5.44467688e-01 -3.00214410e-01 8.57969284e-01
-4.53089774e-01 1.21533640e-01 5.40795863e-01 3.79930943e-01
-2.89305806e-01 2.61867922e-02 -1.16772056e+00 4.56005603e-01
1.23222160e+00 -2.36878455e-01 -6.63570881e-01 2.60055304e-01
2.74827212e-01 -8.87988985e-01 -7.98670411e-01 3.31584483e-01
8.49144340e-01 -1.08449638e+00 1.47364914e+00 -1.62993923e-01
-1.66386947e-01 -6.79906011e-01 -3.23420376e-01 -1.04478884e+00
-2.61817575e-01 -4.79523033e-01 -4.63590294e-01 1.00264013e+00
1.04836129e-01 -5.07462859e-01 7.84442067e-01 3.72250289e-01
1.46513820e-01 -5.31067669e-01 -1.20004690e+00 -7.00571001e-01
-5.64893126e-01 -7.41079569e-01 2.69852221e-01 4.42660153e-01
-3.28482479e-01 3.29136223e-01 -9.13202763e-01 6.89271510e-01
1.26903665e+00 2.79074237e-02 1.28530431e+00 -1.30675399e+00
-5.22240043e-01 -2.94887982e-02 -5.96584797e-01 -1.45556164e+00
-2.10627630e-01 -4.34530139e-01 6.10793121e-02 -1.20103276e+00
5.29536664e-01 -1.58844143e-02 3.23642313e-01 2.88877934e-01
-4.70574349e-01 7.12614655e-01 4.00398225e-01 2.50455946e-01
-1.01022756e+00 4.69798654e-01 9.41561759e-01 1.70732468e-01
-2.64389049e-02 3.96484971e-01 -6.07368767e-01 1.01828635e+00
4.98517156e-01 -4.22085613e-01 1.62724867e-01 -2.57945418e-01
1.97392866e-01 8.32914189e-02 9.66213822e-01 -1.06149709e+00
1.27948657e-01 7.49423653e-02 1.00563598e+00 -1.11081707e+00
9.33867693e-01 -7.25350738e-01 3.92372012e-01 3.06479365e-01
1.28340587e-01 -2.48656217e-02 -1.20831527e-01 8.20636213e-01
1.84004962e-01 8.28077123e-02 7.27573931e-01 -4.13288265e-01
-4.60513711e-01 4.31575090e-01 -4.74922806e-02 -3.84918377e-02
8.23109269e-01 -5.56635082e-01 1.26290202e-01 -5.89387238e-01
-9.81849551e-01 4.55857158e-01 6.52229190e-01 5.51328838e-01
5.27887940e-01 -1.52365470e+00 -9.48644400e-01 2.91729152e-01
1.36947498e-01 7.64225572e-02 3.81911755e-01 1.10925364e+00
-1.21158317e-01 2.87432820e-01 6.00291602e-03 -1.12880051e+00
-1.47267413e+00 1.92263648e-01 3.01281422e-01 -3.34872782e-01
-4.42662627e-01 8.30174685e-01 1.15664434e-02 -8.15115094e-01
1.85410753e-01 -3.95448618e-02 -5.39225619e-03 1.20751195e-01
7.20223546e-01 5.63323855e-01 -1.24011785e-01 -1.11710203e+00
-7.84511447e-01 1.05042601e+00 9.54274014e-02 -2.68090904e-01
1.29172564e+00 -6.74603283e-01 1.71792537e-01 2.57961631e-01
1.01712406e+00 4.17594075e-01 -1.65297651e+00 -3.20993543e-01
-2.73070842e-01 -6.31494880e-01 -3.21155250e-01 -5.08427978e-01
-8.00322354e-01 3.74636799e-01 4.09447759e-01 -5.52602857e-02
6.67551994e-01 4.74337965e-01 5.19973457e-01 1.91168219e-01
6.14962637e-01 -1.08729172e+00 2.02413261e-01 4.96363819e-01
8.74896049e-01 -1.44492912e+00 4.91793901e-01 -6.00113690e-01
-5.58538318e-01 8.99149179e-01 6.11532032e-01 -1.52291581e-01
2.37471893e-01 1.58679247e-01 -1.04896635e-01 -1.29939139e-01
-2.80502737e-01 -3.85829598e-01 6.77140594e-01 7.63047814e-01
1.98505193e-01 2.53343016e-01 2.11312640e-02 7.56089628e-01
-2.56143987e-01 -1.84570909e-01 2.43362084e-01 6.53865278e-01
-3.54557842e-01 -1.07982969e+00 -1.10824454e+00 1.26966387e-01
-3.65466386e-01 3.37520242e-01 -4.20070082e-01 7.26904750e-01
3.79735738e-01 8.49490345e-01 -9.24220085e-02 -4.58182126e-01
5.29668272e-01 -7.92391077e-02 8.00530910e-01 -6.11539304e-01
-2.65646517e-01 5.01057148e-01 3.77050668e-01 -7.52542317e-01
-6.13078296e-01 -1.17268813e+00 -9.64752316e-01 -2.94942856e-01
-2.76534379e-01 -1.20122164e-01 3.39290798e-01 9.96479630e-01
3.60686570e-01 2.65695397e-02 3.01319689e-01 -1.57925856e+00
-6.09763026e-01 -8.28188002e-01 -3.78074408e-01 2.84784734e-01
4.91886586e-01 -1.11331749e+00 -1.40932828e-01 1.52905751e-02] | [7.0407209396362305, -0.9714621305465698] |
5ae2c6e5-adb7-44cb-912f-9669e38bb874 | small-signal-stability-analysis-of-numerical | 2201.09529 | null | https://arxiv.org/abs/2201.09529v1 | https://arxiv.org/pdf/2201.09529v1.pdf | Small-Signal Stability Analysis of Numerical Integration Methods | The paper provides a novel framework to study the accuracy and stability of numerical integration schemes when employed for the time domain simulation of power systems. A matrix pencil-based approach is adopted to evaluate the error between the dynamic modes of the power system and the modes of the approximated discrete-time system arising from the application of the numerical method. The proposed approach can provide meaningful insights on how different methods compare to each other when applied to a power system, while being general enough to be systematically utilized for, in principle, any numerical method. The framework is illustrated for a handful of well-known explicit and implicit methods, while simulation results are presented based on the WSCC 9-bus system, as well as on a 1, 479-bus dynamic model of the All-Island Irish Transmission System. | ['Federico Milano', 'Ioannis Dassios', 'Georgios Tzounas'] | 2022-01-24 | null | null | null | null | ['numerical-integration'] | ['miscellaneous'] | [-2.34275773e-01 -1.01705730e-01 3.67650181e-01 3.64567667e-01
-2.29422748e-01 -7.48672426e-01 5.63163877e-01 1.02647558e-01
-6.56043515e-02 1.14810705e+00 -5.96399665e-01 -6.41854465e-01
-7.68151939e-01 -6.03930771e-01 9.12757963e-02 -1.11883318e+00
-4.88917112e-01 3.40623438e-01 -4.40397449e-02 -6.90939844e-01
-1.04869366e-01 6.72450721e-01 -1.07000792e+00 -3.90671015e-01
8.41646135e-01 7.27725327e-01 -2.93704271e-01 3.79391909e-01
4.30693418e-01 5.86100698e-01 -8.94544840e-01 3.45193967e-02
6.56930283e-02 -6.61227345e-01 -7.47690678e-01 2.86835432e-01
-6.79450870e-01 -4.12562966e-01 -3.18671256e-01 1.05493701e+00
3.93932313e-01 4.63647187e-01 9.30589199e-01 -1.35911453e+00
3.04195672e-01 2.85665512e-01 -2.55747229e-01 2.81498730e-01
6.25937700e-01 1.04365662e-01 6.36194050e-01 -5.26678145e-01
1.74343064e-01 7.89271057e-01 9.90644693e-01 -2.15853170e-01
-1.47256291e+00 -6.51636496e-02 -4.18537438e-01 -9.63425711e-02
-1.61591136e+00 -5.75612560e-02 6.85135543e-01 -4.66329753e-01
1.14194155e+00 7.64858782e-01 1.16425157e+00 -7.80657157e-02
5.21433890e-01 2.18970716e-01 1.11906052e+00 -5.23589432e-01
4.56847489e-01 1.53110445e-01 1.34985715e-01 3.17412078e-01
2.65168548e-01 2.94411421e-01 3.74370188e-01 -6.78103864e-01
5.90800643e-01 -3.91367495e-01 -3.64441961e-01 -3.73555630e-01
-7.94141710e-01 9.36604857e-01 1.69199482e-01 8.73491824e-01
-3.11506152e-01 -1.95120946e-01 4.84107256e-01 3.71090889e-01
5.86485267e-01 2.60368943e-01 -2.60058977e-02 -2.65661687e-01
-1.12605977e+00 5.81635416e-01 1.05943942e+00 7.78670669e-01
2.46186152e-01 8.21273685e-01 4.18718457e-01 5.07886946e-01
3.26891065e-01 5.42952299e-01 1.17203988e-01 -8.42499137e-01
-1.26058936e-01 3.45189363e-01 4.09572631e-01 -8.93551588e-01
-5.39021313e-01 -4.07382995e-01 -7.09638655e-01 6.52812302e-01
5.15138507e-01 -5.53063869e-01 2.14867994e-01 1.17211711e+00
4.53672081e-01 -2.48372599e-01 3.99139076e-02 5.95547974e-01
3.94910388e-02 1.25657892e+00 -3.26607406e-01 -1.00405824e+00
8.58328998e-01 -3.76913160e-01 -7.69506931e-01 6.63732767e-01
1.03555977e+00 -9.66314614e-01 1.48305014e-01 4.35334951e-01
-1.60286951e+00 -1.93269923e-01 -1.19867814e+00 5.87741435e-01
-4.27946627e-01 5.98623231e-02 1.06466658e-01 5.24680078e-01
-1.34496522e+00 8.35058749e-01 -8.86507928e-01 -3.91931593e-01
-4.30652142e-01 1.78310558e-01 1.68134019e-01 6.58111036e-01
-1.27605772e+00 1.48648179e+00 3.65304738e-01 6.80945754e-01
-1.50719583e-01 -7.66548932e-01 -4.62257862e-01 8.05691481e-02
8.60466659e-02 -2.95945913e-01 1.38720846e+00 -9.45471108e-01
-1.66658783e+00 9.84557346e-02 1.01412892e-01 -2.55835623e-01
6.09234571e-01 6.98590815e-01 -6.48866892e-01 2.71271408e-01
-1.13267475e-03 -5.39008975e-01 4.16261643e-01 -8.48933101e-01
-4.94915158e-01 3.16670984e-01 -1.15328170e-02 1.74483195e-01
1.23097926e-01 6.97330758e-02 7.80862153e-01 -6.94369614e-01
-1.22460976e-01 -6.30912840e-01 -4.54097301e-01 -1.46304846e-01
-1.51222244e-01 -2.00068697e-01 7.08929121e-01 -8.97278488e-01
1.53113377e+00 -2.03097224e+00 4.36777413e-01 7.98750222e-01
-3.80745441e-01 3.89604241e-01 4.70566362e-01 1.56408548e+00
-4.79339600e-01 -2.31167093e-01 -5.41944265e-01 3.58003497e-01
2.59476244e-01 2.01050878e-01 -3.91207546e-01 1.03640354e+00
-3.07629295e-02 2.88652658e-01 -6.96328461e-01 -1.96050614e-01
5.34106731e-01 5.41587532e-01 -1.87471658e-01 -1.39759928e-01
4.49760675e-01 1.77181423e-01 -5.90655744e-01 -1.71780482e-01
7.02794015e-01 3.27617526e-01 3.61580253e-01 -2.08388388e-01
-8.56626749e-01 8.37382898e-02 -1.43741608e+00 1.12184548e+00
-6.55928850e-01 4.64752376e-01 6.92214608e-01 -1.39764643e+00
5.70192754e-01 8.38066638e-01 6.24345303e-01 -4.95477140e-01
1.88564762e-01 5.58744550e-01 3.30549687e-01 -1.71444386e-01
1.96155697e-01 -3.85967314e-01 1.01662688e-01 5.77284157e-01
-2.51029991e-02 -6.92840338e-01 8.23103905e-01 1.92064062e-01
6.72859073e-01 -1.27940997e-01 6.93667829e-01 -1.15865934e+00
1.02818966e+00 3.58014524e-01 2.52215147e-01 -1.46637112e-02
7.14187324e-02 -2.51301885e-01 9.25140023e-01 -2.92836756e-01
-1.22573912e+00 -6.95083261e-01 -8.82897139e-01 9.75065865e-03
-1.76715434e-01 -2.49408633e-01 -7.43839562e-01 -2.21125651e-02
2.74919439e-02 1.13542676e+00 -3.24272633e-01 -1.75277684e-02
-6.36911452e-01 -9.62353587e-01 1.98581010e-01 -7.38899270e-03
8.10531005e-02 -4.39488143e-01 -8.19175482e-01 7.26892054e-01
4.49956745e-01 -7.29724407e-01 4.53523174e-02 5.33212647e-02
-9.99952137e-01 -1.21213043e+00 -6.92595720e-01 -5.81363976e-01
6.97583616e-01 -3.55126977e-01 9.27729964e-01 6.13423109e-01
-1.79416910e-01 4.42496270e-01 3.47531475e-02 2.39477560e-01
-8.50903869e-01 -1.90159634e-01 1.25524804e-01 -2.82253653e-01
-3.90937209e-01 -6.60473049e-01 -1.05905429e-01 4.47522670e-01
-9.03005123e-01 -2.65090168e-02 -1.05956912e-01 1.03052866e+00
-1.73206910e-01 8.58015180e-01 9.28867161e-01 -3.47809315e-01
8.63037646e-01 -4.14695501e-01 -1.47891796e+00 2.17131227e-01
-7.15480983e-01 -3.96998972e-01 1.17306340e+00 7.83469621e-03
-1.09070134e+00 -1.89601004e-01 -2.79724896e-01 4.51888233e-01
-1.67761296e-02 7.77482390e-01 6.35563061e-02 -7.16156840e-01
-3.30449454e-02 2.38726124e-01 1.45173714e-01 -4.91526127e-01
5.59836179e-02 5.25334179e-01 2.09311262e-01 -5.75308800e-01
9.99209344e-01 1.96790069e-01 6.69845879e-01 -1.19676781e+00
3.10992092e-01 -1.13060534e-01 -7.49018729e-01 -1.97889641e-01
1.63365304e-01 -4.79042053e-01 -8.34462762e-01 6.88049018e-01
-1.13592148e+00 -3.55788738e-01 -5.36016107e-01 3.96582842e-01
-6.31042242e-01 4.24072444e-01 -8.04563463e-01 -1.19576454e+00
8.17572325e-02 -1.30284500e+00 4.66045588e-01 -4.10026051e-02
-2.08548963e-01 -1.83644128e+00 2.80262202e-01 -5.66925585e-01
5.50602674e-01 5.14388025e-01 1.23144376e+00 -3.17923665e-01
-1.72693154e-03 -3.56017768e-01 3.87811273e-01 2.52653390e-01
-2.10236773e-01 7.40339339e-01 -4.13982034e-01 -8.60273957e-01
4.53736573e-01 3.66979182e-01 -4.13143694e-01 1.56709239e-01
9.44695994e-02 -5.23069441e-01 -3.33757073e-01 1.42799795e-01
1.87935412e+00 5.77524960e-01 2.57139593e-01 1.77475527e-01
-4.46333438e-02 3.44430417e-01 5.17078340e-01 5.74232221e-01
-7.82687683e-03 9.33743238e-01 2.90048551e-02 -3.06511402e-01
5.39340079e-01 5.34193993e-01 2.25557029e-01 1.11260092e+00
-1.28660187e-01 1.28581420e-01 -7.68116474e-01 4.88912374e-01
-1.46137762e+00 -7.86304235e-01 -5.08021414e-01 2.06663966e+00
5.90837836e-01 4.77812178e-02 5.63012242e-01 9.68652666e-01
5.70889831e-01 -2.07781807e-01 1.24479659e-01 -1.02869236e+00
-2.59584654e-02 2.20446721e-01 2.23523617e-01 7.36736059e-01
-6.38978362e-01 -4.23425406e-01 6.69067669e+00 1.14289153e+00
-9.12082970e-01 -1.07432760e-01 3.11213970e-01 2.71222353e-01
8.18399563e-02 3.20797116e-01 -2.43115336e-01 4.78450090e-01
1.40089369e+00 -1.13518834e+00 4.89528179e-01 4.05008942e-01
7.20581114e-01 -5.55163264e-01 -8.72225881e-01 3.48821163e-01
-5.28290987e-01 -1.00203717e+00 -5.61406255e-01 -4.69848961e-02
9.52376664e-01 -5.82279980e-01 -2.45759130e-01 4.01852801e-02
-3.90286744e-01 -6.68497682e-01 7.16114998e-01 3.04206461e-01
2.27177128e-01 -1.13390589e+00 9.19827163e-01 7.21733630e-01
-1.18918025e+00 -4.07364033e-02 2.84685637e-03 -4.91782308e-01
6.51367068e-01 4.91642863e-01 -4.32091475e-01 1.17812192e+00
3.34363282e-01 6.96374178e-01 -1.29487261e-01 1.33348787e+00
3.20783585e-01 5.97118556e-01 -5.16242385e-01 -2.84161389e-01
4.85740721e-01 -8.90282869e-01 7.32352734e-01 9.21703696e-01
4.93080735e-01 2.46833339e-01 -1.54889375e-01 8.67604613e-01
1.03570449e+00 4.42451425e-02 -4.03507084e-01 2.42742196e-01
2.28832766e-01 1.25521553e+00 -6.02239907e-01 -4.27690506e-01
-6.10569656e-01 7.49767572e-02 -5.11393070e-01 9.47555006e-01
-8.38928223e-01 -9.03167903e-01 3.75873715e-01 1.59378305e-01
1.83973853e-02 -3.87838125e-01 -1.59067616e-01 -7.40696251e-01
-1.45194322e-01 -7.53744483e-01 1.50834188e-01 -6.82523251e-01
-1.08095372e+00 6.25803173e-01 8.12688351e-01 -1.48263097e+00
-1.06804025e+00 -6.16878569e-01 -8.69700432e-01 1.51003993e+00
-1.03735518e+00 -3.64215583e-01 6.50787055e-01 3.12125564e-01
1.26891345e-01 4.08057272e-02 9.64733124e-01 4.60763395e-01
-7.81553030e-01 -1.39758149e-02 9.89336610e-01 -1.34529695e-01
-4.37756598e-01 -1.12892735e+00 6.95178434e-02 9.22884345e-01
-6.70096815e-01 3.73911947e-01 1.23188758e+00 -7.56277218e-02
-1.54797792e+00 -2.86814690e-01 5.55161953e-01 3.21337998e-01
1.05963600e+00 -1.38109829e-02 -9.50120330e-01 5.09474516e-01
8.72960865e-01 -9.71681327e-02 8.24315846e-02 -5.91280162e-01
8.33939791e-01 -9.87744480e-02 -1.41978467e+00 4.62902993e-01
-1.46618158e-01 -3.52087557e-01 -6.38092577e-01 1.68787405e-01
-2.52168804e-01 -5.32181263e-01 -1.49570489e+00 1.64873987e-01
3.26941133e-01 -8.39882135e-01 8.83675814e-01 -3.54324095e-02
-3.62762839e-01 -5.77934265e-01 3.45176280e-01 -1.74214947e+00
-1.20450251e-01 -1.09171057e+00 2.73030251e-02 1.11375666e+00
2.84897447e-01 -1.41125882e+00 -1.00761019e-01 6.64209247e-01
5.36066890e-02 -8.93674493e-01 -1.60935330e+00 -8.79606068e-01
6.93398058e-01 1.47220835e-01 6.14975572e-01 8.36837590e-01
8.72216880e-01 -2.18163475e-01 3.38966697e-01 1.87814161e-01
6.95021629e-01 1.16321191e-01 3.01769227e-02 -8.73415887e-01
-2.57331789e-01 -6.98790669e-01 -2.73493588e-01 -4.56187427e-01
1.28106877e-01 -4.84199017e-01 -5.30726433e-01 -1.25001538e+00
-6.00085557e-01 -4.71175581e-01 1.98519230e-01 1.02819130e-02
4.99644428e-01 9.06656757e-02 8.17167461e-02 5.70573285e-02
2.95908242e-01 6.04439974e-01 8.92866194e-01 2.25980282e-01
1.55438915e-01 2.70961821e-01 3.77252787e-01 7.96151102e-01
5.73801577e-01 -2.52260715e-01 -4.66416895e-01 5.79167716e-02
2.76290774e-01 8.03592622e-01 5.13884723e-01 -1.11146116e+00
1.90648139e-01 5.16361594e-02 1.59478523e-02 -6.53910339e-01
1.34255826e-01 -1.28944278e+00 8.25610578e-01 9.40715969e-01
-3.63022499e-02 5.58738649e-01 6.32227361e-01 -1.58416200e-02
-4.64252770e-01 -6.73046350e-01 6.22696459e-01 2.15293676e-01
-2.76433736e-01 -3.99479389e-01 -1.18288159e+00 -1.68416157e-01
1.16972589e+00 -5.74833974e-02 -3.53198141e-01 -3.62113357e-01
-9.05767500e-01 2.95139104e-01 4.85284805e-01 -6.30897701e-01
1.25830173e-02 -1.41641235e+00 -6.15162551e-01 3.54692429e-01
-6.08653605e-01 -5.19192934e-01 2.09204271e-01 1.16623151e+00
-8.97597790e-01 6.97876096e-01 -2.08046690e-01 -6.21897638e-01
-7.59123921e-01 2.40855873e-01 9.94005263e-01 -2.48171747e-01
-4.44634378e-01 -1.53335646e-01 -3.01389039e-01 -5.94484396e-02
-5.75004697e-01 -4.75121796e-01 1.63931642e-02 -8.51851702e-02
1.61010593e-01 9.01066482e-01 2.97107667e-01 -9.22373414e-01
-2.24026635e-01 7.37862229e-01 8.26723456e-01 -3.37532431e-01
1.19021833e+00 -3.74336720e-01 -6.12883091e-01 6.69993043e-01
1.11005187e+00 -3.90941113e-01 -1.07781303e+00 3.02714527e-01
-3.19994867e-01 -5.51087707e-02 2.33622879e-01 -5.32209694e-01
-1.02632511e+00 6.54983222e-01 2.09770888e-01 1.35003364e+00
1.33193958e+00 -9.27956700e-01 2.23485917e-01 6.06696717e-02
5.00816584e-01 -1.01044929e+00 -1.05056715e+00 2.83566207e-01
9.44835246e-01 -2.18127266e-01 5.51703990e-01 -5.67700386e-01
4.83262129e-02 1.68088496e+00 -1.76343128e-01 -4.99043763e-01
9.44951177e-01 6.77419960e-01 -2.35967949e-01 3.60615969e-01
-6.94416881e-01 4.49786782e-01 1.30727589e-01 4.42691177e-01
6.75824821e-01 -1.93285123e-01 -1.10140836e+00 2.69916356e-01
4.49755825e-02 6.38057441e-02 1.07137167e+00 1.11972821e+00
1.97472632e-01 -1.03264356e+00 -6.10031784e-01 -1.12575488e-02
-4.35147941e-01 2.40541756e-01 3.87227684e-01 1.53422511e+00
-4.30295169e-01 7.91920006e-01 1.68981729e-04 5.61680496e-01
5.82845092e-01 -1.28694316e-02 3.92906100e-01 -4.32244083e-03
-8.56295168e-01 3.22555155e-01 3.52481931e-01 -1.43720388e-01
-3.16841930e-01 -9.28648949e-01 -1.25098717e+00 -7.06638455e-01
-5.89521587e-01 1.04236174e+00 3.74251485e-01 1.02738678e+00
-3.65000278e-01 5.84502518e-01 8.81992757e-01 -1.02853549e+00
-8.57466996e-01 -7.55592644e-01 -1.28376031e+00 -3.15179229e-01
2.86350876e-01 -6.73312485e-01 -7.50570834e-01 -5.26162922e-01] | [5.724118232727051, 2.7604146003723145] |
79f3975d-a80b-450a-ae9c-d74d2c7c58e8 | evaluation-of-various-open-set-medical | 2110.10888 | null | https://arxiv.org/abs/2110.10888v1 | https://arxiv.org/pdf/2110.10888v1.pdf | Evaluation of Various Open-Set Medical Imaging Tasks with Deep Neural Networks | The current generation of deep neural networks has achieved close-to-human results on "closed-set" image recognition; that is, the classes being evaluated overlap with the training classes. Many recent methods attempt to address the importance of the unknown, which are termed "open-set" recognition algorithms, try to reject unknown classes as well as maintain high recognition accuracy on known classes. However, it is still unclear how different general domain-trained open-set methods from ImageNet would perform on a different but more specific domain, such as the medical domain. Without principled and formal evaluations to measure the effectiveness of those general open-set methods, artificial intelligence (AI)-based medical diagnostics would experience ineffective adoption and increased risks of bad decision making. In this paper, we conduct rigorous evaluations amongst state-of-the-art open-set methods, exploring different open-set scenarios from "similar-domain" to "different-domain" scenarios and comparing them on various general and medical domain datasets. We summarise the results and core ideas and explain how the models react to various degrees of openness and different distributions of open classes. We show the main difference between general domain-trained and medical domain-trained open-set models with our quantitative and qualitative analysis of the results. We also identify aspects of model robustness in real clinical workflow usage according to confidence calibration and the inference efficiency. | ['Xin Wang', 'ZongYuan Ge'] | 2021-10-21 | null | null | null | null | ['open-set-learning'] | ['miscellaneous'] | [ 4.39270794e-01 7.67547607e-01 -6.61846399e-02 -5.37318647e-01
-4.94646668e-01 -6.12741828e-01 5.61168909e-01 2.30012298e-01
-4.22480464e-01 8.56209457e-01 6.67196959e-02 -4.58104998e-01
-7.10291624e-01 -6.19289041e-01 -5.03168643e-01 -6.27880692e-01
1.33183271e-01 9.85768974e-01 -1.16080977e-01 -1.68720067e-01
-4.48754542e-02 3.45656246e-01 -1.38029563e+00 5.29308915e-01
7.20968246e-01 1.04006529e+00 -4.67900872e-01 3.45429957e-01
-2.63057724e-02 4.19066578e-01 -7.28984416e-01 -6.41059697e-01
4.54681963e-01 -2.12564945e-01 -1.03399289e+00 -9.09083411e-02
5.09447157e-01 -9.94978622e-02 -2.27268130e-01 1.08965850e+00
7.95332491e-01 -6.96734488e-02 9.91534710e-01 -9.27231669e-01
-9.06193435e-01 4.33727741e-01 9.58735123e-02 2.77610928e-01
1.89060077e-01 5.07782757e-01 4.72803086e-01 -4.23087835e-01
8.45745444e-01 9.49816287e-01 1.07743180e+00 9.54083622e-01
-1.30458117e+00 -5.65923929e-01 7.24936426e-02 -5.02391644e-02
-1.32319582e+00 -4.25423443e-01 3.63804013e-01 -6.34175599e-01
9.74758983e-01 5.09880126e-01 3.82488489e-01 1.43060040e+00
3.02724123e-01 2.67280906e-01 1.30780613e+00 -4.33792830e-01
2.58571059e-01 6.59971833e-01 4.63919371e-01 4.27805245e-01
5.07518828e-01 5.72439313e-01 5.18931784e-02 -3.51688266e-01
6.52579606e-01 6.58650175e-02 -6.10190570e-01 -5.32722175e-01
-1.26190615e+00 7.59763062e-01 4.75125700e-01 5.53355992e-01
-2.72457659e-01 -5.67204595e-01 7.69483447e-01 4.28780764e-01
5.20354450e-01 6.94918633e-01 -8.06089997e-01 2.76055366e-01
-7.82359123e-01 -4.66809981e-02 1.22028792e+00 6.66607738e-01
3.14321548e-01 -2.61380315e-01 -2.49723926e-01 8.79435062e-01
-5.58931520e-03 2.35446781e-01 8.42576027e-01 -6.35480464e-01
-6.50769239e-03 6.78734899e-01 -2.55194068e-01 -1.01396704e+00
-5.81022263e-01 -8.61841917e-01 -1.29743338e+00 2.94297993e-01
6.49505138e-01 -4.63786796e-02 -1.28651953e+00 1.64547074e+00
1.16664218e-02 -6.49351999e-02 4.11676973e-01 7.73601770e-01
1.33798802e+00 -6.33986741e-02 -5.82693927e-02 -1.91500828e-01
1.38530731e+00 -4.98534620e-01 -5.97710311e-01 -1.94044039e-01
6.98255122e-01 -2.25174189e-01 8.12320173e-01 5.01796663e-01
-6.58718944e-01 -4.98514473e-01 -9.35198069e-01 3.21328193e-01
-6.50683701e-01 -8.95804688e-02 3.01634550e-01 9.15980756e-01
-9.24765646e-01 6.82923913e-01 -4.24738795e-01 -4.18487459e-01
8.33587229e-01 5.08981824e-01 -5.68409145e-01 -2.97213420e-02
-1.35229826e+00 1.26608598e+00 6.14602506e-01 -2.08292246e-01
-9.01639521e-01 -9.75856543e-01 -6.85351491e-01 8.88865814e-02
2.45061100e-01 -9.62865233e-01 9.95449066e-01 -1.40485692e+00
-9.17274892e-01 1.62942088e+00 5.39382696e-01 -8.59136701e-01
8.13399971e-01 8.07927623e-02 -5.74237764e-01 7.65071288e-02
-1.26486734e-01 7.54347265e-01 6.20608747e-01 -1.29193592e+00
-1.52511343e-01 -3.75556797e-01 -4.88058440e-02 -3.85374874e-02
-3.28349739e-01 -1.34451240e-01 1.26448035e-01 -5.23154855e-01
5.81390895e-02 -8.39322090e-01 -2.93895394e-01 3.05962175e-01
-3.96046311e-01 -9.69700962e-02 4.89942044e-01 -3.96521568e-01
1.14458764e+00 -2.20946002e+00 -4.03982818e-01 3.30558151e-01
6.56653523e-01 7.26361156e-01 1.10258341e-01 -1.21569782e-01
-6.85137868e-01 2.14778796e-01 -4.92985547e-01 3.33940089e-01
-6.87401071e-02 5.24221539e-01 -2.34369472e-01 6.13900185e-01
1.60974070e-01 7.57189691e-01 -7.79130042e-01 -5.04254282e-01
3.19503993e-01 4.15884882e-01 -4.89535362e-01 -9.03282985e-02
-6.88193366e-02 3.97031099e-01 -7.76545182e-02 5.23921251e-01
6.54369235e-01 -5.70513546e-01 2.63757527e-01 -1.74324736e-01
4.18715000e-01 -1.86985675e-02 -1.04884756e+00 1.29792321e+00
-8.92435834e-02 6.93226159e-01 -2.26805672e-01 -1.26954031e+00
1.00126576e+00 5.11150122e-01 4.17955190e-01 -5.98999381e-01
2.87954897e-01 4.30341691e-01 5.48442483e-01 -4.77965534e-01
-1.14330307e-01 -4.29351926e-01 1.98080853e-01 1.11746885e-01
3.29404384e-01 1.00698337e-01 -2.61600822e-01 -1.26770675e-01
1.13513255e+00 -4.34519127e-02 8.16242278e-01 -5.59432626e-01
4.68845606e-01 8.53568837e-02 5.14454067e-01 9.69335079e-01
-7.99860179e-01 9.41866994e-01 5.14731050e-01 -9.48564947e-01
-7.42382109e-01 -1.03226566e+00 -9.09592032e-01 4.91915256e-01
-1.29444867e-01 2.01502964e-01 -8.29424024e-01 -9.79175627e-01
6.49675652e-02 6.57108605e-01 -1.04049385e+00 -4.56260026e-01
-2.11302876e-01 -7.71919727e-01 9.48719740e-01 3.59519184e-01
2.29169339e-01 -1.08809948e+00 -8.92970800e-01 -8.80922005e-02
-1.10171258e-01 -7.69565105e-01 9.96402726e-02 3.74029726e-01
-8.25529456e-01 -1.41460288e+00 -9.14986968e-01 -4.66155678e-01
7.60695696e-01 -4.36032206e-01 1.52723157e+00 1.07161999e-02
-3.25614959e-01 4.25977141e-01 -2.49013871e-01 -8.14031243e-01
-8.47363114e-01 -1.78080216e-01 1.76523045e-01 -2.96325553e-02
6.64141953e-01 -1.18851826e-01 -7.29399681e-01 4.93996561e-01
-1.00677180e+00 -2.49322101e-01 7.03606188e-01 1.13068259e+00
5.01757264e-01 -2.32206672e-01 5.95361233e-01 -1.07320571e+00
6.27144039e-01 -5.57820737e-01 -3.43352556e-02 4.26870435e-01
-1.03940248e+00 -4.28537726e-02 4.66314077e-01 -7.24714041e-01
-8.29577088e-01 -1.75942078e-01 -1.01903826e-01 -8.76010537e-01
-6.25372529e-01 3.04041922e-01 5.85779175e-02 1.46332877e-02
1.48776186e+00 4.83317375e-02 3.36242080e-01 -3.12869638e-01
7.04158172e-02 9.49512303e-01 4.79077846e-01 -3.67367983e-01
2.89087027e-01 7.45093644e-01 -3.05106014e-01 -6.93835616e-01
-9.63556528e-01 -3.93810719e-01 -5.52416205e-01 -6.01133406e-02
7.23986030e-01 -5.86229026e-01 -4.05080169e-01 2.95950711e-01
-9.13223624e-01 -2.00305238e-01 -9.08713937e-01 3.61759275e-01
-5.56813061e-01 3.56224567e-01 -6.17137216e-02 -5.52375793e-01
-5.83192348e-01 -1.12697041e+00 8.75726938e-01 1.83945242e-02
-7.29275346e-01 -1.22910225e+00 1.12863347e-01 1.26055643e-01
3.00453156e-01 3.95685285e-01 9.59650815e-01 -1.56770933e+00
1.50267571e-01 -2.29084715e-01 -2.66284138e-01 7.69274771e-01
4.67978306e-02 -3.05181026e-01 -1.15204430e+00 -2.96043277e-01
2.56627321e-01 -3.37212950e-01 8.66715372e-01 5.65287948e-01
1.10109389e+00 -4.11880553e-01 -6.92048848e-01 6.32682562e-01
1.34049559e+00 6.84496164e-02 9.17713583e-01 3.52687925e-01
2.09501579e-01 7.42038310e-01 5.24411738e-01 2.58873433e-01
-3.17616731e-01 2.17516065e-01 2.92475224e-01 -3.96391958e-01
-1.12137921e-01 1.79196149e-01 -9.00409147e-02 1.82630837e-01
2.20428422e-01 -2.26518407e-01 -1.33254445e+00 7.11626291e-01
-1.51756823e+00 -8.47475469e-01 -9.14228708e-02 2.27196693e+00
9.89415228e-01 4.55332696e-01 -6.49848208e-02 1.98970050e-01
7.72331834e-01 -3.88959557e-01 -5.98810315e-01 -5.65352857e-01
-2.84012914e-01 2.71711767e-01 2.70233274e-01 9.66697857e-02
-1.12995255e+00 2.56176919e-01 7.09194803e+00 7.86119163e-01
-1.15336919e+00 1.78151727e-01 1.05868781e+00 -2.84134038e-02
-1.02687463e-01 -1.11705601e-01 -4.30495620e-01 3.03997040e-01
1.17717290e+00 2.06275985e-01 -2.78536171e-01 9.22704577e-01
-3.25918972e-01 1.80432379e-01 -1.61301315e+00 1.01824176e+00
2.26064414e-01 -1.59528375e+00 1.63036644e-01 1.12511091e-01
6.25559270e-01 2.93492943e-01 -9.01754200e-02 4.18933719e-01
1.61533073e-01 -1.14948332e+00 4.44021821e-01 8.42619658e-01
1.20735466e+00 -1.45117387e-01 9.67639565e-01 3.43077183e-01
-1.66264176e-01 -2.51189351e-01 -2.07001522e-01 8.12499039e-03
-3.96720976e-01 6.42578006e-01 -9.57725585e-01 4.42065626e-01
9.22137082e-01 3.60723317e-01 -4.33789849e-01 8.55001092e-01
3.58867407e-01 4.97152776e-01 -3.75371456e-01 2.15846226e-01
1.26329064e-01 1.89474449e-01 6.61509633e-01 1.28559196e+00
-1.85618997e-01 9.34277028e-02 -7.36081880e-03 1.07771230e+00
2.11877942e-01 -1.99638948e-01 -8.66641819e-01 -3.37295756e-02
4.74812761e-02 7.52010465e-01 -6.69143319e-01 -6.07192397e-01
-3.14186603e-01 5.49482644e-01 4.50103804e-02 1.00191288e-01
-7.23281503e-01 -2.88222492e-01 5.85849226e-01 3.51277888e-01
-1.12533666e-01 5.64439714e-01 -5.64230740e-01 -1.03908682e+00
-4.96700332e-02 -1.16011047e+00 9.54978347e-01 -6.62294149e-01
-1.71455538e+00 6.20650530e-01 8.29023570e-02 -1.46478117e+00
-5.11421338e-02 -1.00718474e+00 -1.11935385e-01 7.62453794e-01
-1.28124356e+00 -8.17157984e-01 -6.39858320e-02 5.65030396e-01
2.20697075e-01 -3.32238793e-01 1.25277317e+00 1.80383176e-01
-2.05419540e-01 7.22862899e-01 4.81052190e-01 5.26684523e-01
8.29415917e-01 -1.04193199e+00 -4.09762785e-02 3.01419407e-01
-1.70022041e-01 8.00590992e-01 6.64910495e-01 -5.35718381e-01
-5.64181983e-01 -9.99364138e-01 6.27287388e-01 -1.00410831e+00
1.94546223e-01 -4.92742332e-03 -1.12032616e+00 6.53661788e-01
-7.65498308e-03 3.45325440e-01 9.69695926e-01 2.14520127e-01
-4.60653335e-01 -6.40541390e-02 -1.63841891e+00 3.03295135e-01
9.90157127e-01 -2.49253199e-01 -1.14676833e+00 4.40423459e-01
5.48401117e-01 -4.52522367e-01 -1.01752520e+00 7.49406397e-01
5.68147600e-01 -1.18526995e+00 1.06178856e+00 -1.01867259e+00
5.25837719e-01 2.33256727e-01 -9.95398760e-02 -1.30238116e+00
-3.27234238e-01 -1.45581782e-01 1.53625654e-02 7.05842614e-01
4.04576153e-01 -1.11216521e+00 6.08238816e-01 7.70515800e-01
-1.70242533e-01 -9.86041784e-01 -1.01051295e+00 -7.61504114e-01
4.47454900e-01 -2.52952874e-01 2.33991399e-01 1.40826809e+00
1.01267854e-02 1.87787250e-01 1.94525063e-01 9.44274366e-02
3.73348892e-01 -5.61594367e-02 3.84559393e-01 -1.66128373e+00
-2.48338267e-01 -4.66070354e-01 -7.56009877e-01 -1.45846069e-01
2.92245969e-02 -9.65824485e-01 -2.53937125e-01 -1.36811507e+00
2.90386975e-01 -4.73044664e-01 -6.30894840e-01 5.76113522e-01
1.06428117e-01 1.54279292e-01 -2.55503714e-01 4.31947440e-01
-4.61689621e-01 -7.23109469e-02 1.10425663e+00 -3.96394461e-01
-1.22273147e-01 -1.03703417e-01 -9.21547949e-01 7.44707108e-01
6.05154157e-01 -5.58037519e-01 -4.03023601e-01 -1.56935617e-01
9.89701077e-02 -1.02002122e-01 7.30780482e-01 -1.08505285e+00
-1.11456193e-01 -5.73317185e-02 5.32132566e-01 -1.22895122e-01
-2.42339130e-02 -8.67167413e-01 1.70354813e-01 9.80769634e-01
-5.10162354e-01 -5.92266798e-01 4.41661119e-01 5.41796625e-01
-3.74563746e-02 -3.39706302e-01 1.05913210e+00 -5.14795423e-01
-5.60888231e-01 9.75109488e-02 -1.00953080e-01 4.30002362e-01
1.12384498e+00 -7.96920657e-01 -2.45766044e-01 -6.86910301e-02
-1.30005085e+00 -1.19937502e-01 2.23187074e-01 4.38140124e-01
5.62651634e-01 -7.78677464e-01 -8.94045532e-01 3.21709573e-01
4.89890158e-01 -3.59042315e-03 4.45590347e-01 9.12876010e-01
-4.17910129e-01 5.98464608e-01 -2.44686663e-01 -9.63489056e-01
-1.16612256e+00 7.61456072e-01 1.02565646e+00 -3.91117662e-01
-7.06642270e-01 8.14539135e-01 5.78894377e-01 -1.00032473e+00
2.51345187e-01 -2.91234285e-01 -1.29704088e-01 -9.60337892e-02
3.05807769e-01 1.19052954e-01 4.99059975e-01 -4.07440186e-01
-6.26647413e-01 5.39599024e-02 -2.20478043e-01 2.99823105e-01
1.05150020e+00 2.06416711e-01 8.44514072e-02 2.37564638e-01
1.08001423e+00 -7.10474074e-01 -8.17638457e-01 -2.96152025e-01
-2.78562963e-01 -2.09944755e-01 2.24716663e-02 -1.48369157e+00
-6.76565766e-01 8.70620131e-01 1.25512612e+00 4.99332398e-02
9.98888791e-01 -1.38860056e-02 1.11844450e-01 5.18440366e-01
1.69269383e-01 -1.16140151e+00 -1.94205701e-01 3.53431195e-01
9.57788825e-01 -1.57299173e+00 7.23065734e-02 -2.40461648e-01
-5.69204152e-01 8.78596008e-01 6.08753264e-01 5.41769750e-02
9.07208204e-01 1.13811746e-01 2.77593046e-01 -4.65105772e-01
-5.26052833e-01 7.35671222e-02 3.89995784e-01 1.06298733e+00
3.24243128e-01 2.35445961e-01 -2.77475685e-01 7.82306433e-01
2.14613397e-02 3.94828975e-01 3.36502671e-01 7.36658454e-01
-4.55944538e-02 -6.52567565e-01 -5.87959766e-01 1.02430308e+00
-5.04571199e-01 -6.88778013e-02 -5.91277242e-01 9.92715716e-01
4.80759978e-01 5.42368889e-01 2.48528153e-01 -2.14006051e-01
3.47306043e-01 2.75262475e-01 3.97005022e-01 -7.34773755e-01
-8.22882771e-01 -5.03692508e-01 2.76713431e-01 -3.61689776e-01
-2.92392403e-01 -5.30977070e-01 -1.11246455e+00 -4.78909053e-02
-5.34782708e-01 -1.77188199e-02 2.75234550e-01 9.16645229e-01
6.03699386e-01 6.16742074e-01 -5.29928841e-02 -3.38412762e-01
-1.14450192e+00 -9.89818394e-01 -2.00648293e-01 9.46207047e-01
4.49460983e-01 -7.11406291e-01 -3.94202113e-01 -1.42021507e-01] | [15.031068801879883, -2.370415210723877] |
d55f8374-50e5-4096-bfb3-66fc65a1baed | pivotal-tuning-for-latent-based-editing-of | 2106.05744 | null | https://arxiv.org/abs/2106.05744v1 | https://arxiv.org/pdf/2106.05744v1.pdf | Pivotal Tuning for Latent-based Editing of Real Images | Recently, a surge of advanced facial editing techniques have been proposed that leverage the generative power of a pre-trained StyleGAN. To successfully edit an image this way, one must first project (or invert) the image into the pre-trained generator's domain. As it turns out, however, StyleGAN's latent space induces an inherent tradeoff between distortion and editability, i.e. between maintaining the original appearance and convincingly altering some of its attributes. Practically, this means it is still challenging to apply ID-preserving facial latent-space editing to faces which are out of the generator's domain. In this paper, we present an approach to bridge this gap. Our technique slightly alters the generator, so that an out-of-domain image is faithfully mapped into an in-domain latent code. The key idea is pivotal tuning - a brief training process that preserves the editing quality of an in-domain latent region, while changing its portrayed identity and appearance. In Pivotal Tuning Inversion (PTI), an initial inverted latent code serves as a pivot, around which the generator is fined-tuned. At the same time, a regularization term keeps nearby identities intact, to locally contain the effect. This surgical training process ends up altering appearance features that represent mostly identity, without affecting editing capabilities. We validate our technique through inversion and editing metrics, and show preferable scores to state-of-the-art methods. We further qualitatively demonstrate our technique by applying advanced edits (such as pose, age, or expression) to numerous images of well-known and recognizable identities. Finally, we demonstrate resilience to harder cases, including heavy make-up, elaborate hairstyles and/or headwear, which otherwise could not have been successfully inverted and edited by state-of-the-art methods. | ['Daniel Cohen-Or', 'Amit H. Bermano', 'Ron Mokady', 'Daniel Roich'] | 2021-06-10 | null | null | null | null | ['facial-editing'] | ['computer-vision'] | [ 6.41086102e-01 5.10125756e-01 2.02712640e-01 -4.04315412e-01
-3.49443704e-01 -8.81183088e-01 7.06883729e-01 -4.66009974e-01
3.37163359e-02 5.59985638e-01 1.12084135e-01 1.25650391e-01
4.00994956e-01 -7.65180349e-01 -8.26213956e-01 -7.15147793e-01
2.11509869e-01 1.32622808e-01 -1.31938949e-01 -4.41090494e-01
-3.69204883e-03 5.53269088e-01 -1.45734072e+00 1.52340963e-01
8.57593000e-01 7.48271227e-01 -2.39558965e-01 4.28373843e-01
1.40398934e-01 2.63795078e-01 -6.39368892e-01 -1.00492823e+00
5.22094011e-01 -7.96639323e-01 -5.42822421e-01 3.46809357e-01
8.99842978e-01 -3.66908073e-01 -2.17295408e-01 1.11943042e+00
2.81605989e-01 -2.08117336e-01 7.35974550e-01 -1.30237091e+00
-1.04170561e+00 1.32827848e-01 -7.39029586e-01 -6.03429914e-01
4.34743732e-01 1.31651342e-01 7.13613868e-01 -9.08842027e-01
9.18916106e-01 1.36533344e+00 8.10566247e-01 7.82723188e-01
-1.55201423e+00 -1.02487373e+00 3.23228678e-03 -4.01904374e-01
-1.29208159e+00 -7.40169227e-01 9.66242313e-01 -4.43477422e-01
1.41431868e-01 3.27569485e-01 6.24759734e-01 1.27151334e+00
3.24641556e-01 2.98187256e-01 1.01005805e+00 -4.59966034e-01
1.96429482e-03 1.80117697e-01 -8.47239196e-01 8.94232452e-01
-5.49569074e-03 2.01594144e-01 -4.24419314e-01 -4.05932106e-02
9.25214708e-01 -4.50369418e-02 -1.93892404e-01 -6.20948911e-01
-9.81284618e-01 7.23073661e-01 3.40654552e-01 -1.73387453e-02
-1.41911954e-01 9.12814736e-02 5.00016324e-02 3.83787096e-01
6.89862311e-01 6.31330252e-01 -1.18034057e-01 9.85124931e-02
-1.16660082e+00 9.84876975e-02 5.54149210e-01 9.39688265e-01
1.00378263e+00 4.20497209e-02 -1.27364859e-01 1.00084698e+00
2.79437248e-02 4.96968091e-01 2.03846261e-01 -1.10346508e+00
8.70521739e-02 6.03569269e-01 -1.53813779e-01 -1.16829991e+00
1.70537442e-01 -2.52436936e-01 -1.06057990e+00 7.26678312e-01
3.50115538e-01 -5.81897907e-02 -1.10237336e+00 2.11869645e+00
3.91902477e-01 -7.34189674e-02 -2.28924006e-01 6.97140574e-01
5.08794546e-01 4.26681280e-01 -2.29908139e-01 -1.21899150e-01
1.19266915e+00 -9.27329659e-01 -6.45724237e-01 -1.63573086e-01
1.65316552e-01 -1.10621631e+00 1.11880755e+00 1.70944393e-01
-1.32833195e+00 -5.99536538e-01 -1.03867710e+00 -1.49790317e-01
-1.74339384e-01 1.01893887e-01 4.32655782e-01 7.14809895e-01
-1.22800696e+00 6.58755243e-01 -4.68027920e-01 -3.94845456e-01
4.68873233e-01 3.12889665e-01 -7.83676207e-01 2.23489404e-02
-1.06134737e+00 6.69828832e-01 -1.78956777e-01 -6.28399700e-02
-9.44739163e-01 -9.48214650e-01 -8.40463281e-01 -2.15216607e-01
2.07970932e-01 -8.54124069e-01 9.61943030e-01 -1.60442865e+00
-1.87558258e+00 1.34748828e+00 -1.38773471e-01 7.45802522e-02
9.46090877e-01 -2.52475888e-02 -4.94726270e-01 1.81730948e-02
2.67596900e-01 9.64566052e-01 1.59610522e+00 -1.41899097e+00
-2.76172876e-01 -2.69476593e-01 1.81390047e-01 8.48440826e-03
-4.91414458e-01 -9.08460543e-02 -6.88418806e-01 -1.12014484e+00
1.02552287e-01 -1.23856473e+00 1.50869787e-01 6.35075867e-01
-5.09136558e-01 5.17188668e-01 9.82628584e-01 -6.95380449e-01
9.35015500e-01 -2.41026092e+00 2.28187189e-01 3.83831322e-01
4.09272552e-01 2.36858800e-01 -3.46457154e-01 3.70149791e-01
-3.97937357e-01 2.71078587e-01 -4.32560265e-01 -6.68515503e-01
-1.06476896e-01 2.56946534e-01 -4.14423555e-01 5.25107622e-01
2.80189157e-01 9.79033828e-01 -7.29243100e-01 -3.77017081e-01
-2.88111009e-02 7.16860771e-01 -8.87209058e-01 1.20808817e-01
-8.85175914e-02 6.63433671e-01 5.89575395e-02 6.32742524e-01
9.13547158e-01 5.90853542e-02 1.14049368e-01 -3.03539693e-01
-5.35406806e-02 -2.95281112e-01 -9.81673181e-01 1.69120932e+00
-5.43395638e-01 6.89807713e-01 1.86866283e-01 -5.49087465e-01
9.45074737e-01 9.61699933e-02 2.93555021e-01 -6.24175429e-01
-4.70999032e-02 1.15082413e-01 -1.91082671e-01 -1.08971126e-01
4.43441987e-01 -3.21768522e-01 3.67404036e-02 4.68412340e-01
-1.74531452e-02 -2.77043790e-01 -1.68926537e-01 2.07690105e-01
8.17476511e-01 2.98573881e-01 1.21318266e-01 -1.78278118e-01
4.52747703e-01 -4.30756271e-01 4.29001540e-01 3.67034078e-01
6.32379055e-02 1.08532345e+00 6.34628296e-01 -2.49289304e-01
-1.40597522e+00 -1.22337139e+00 7.29386955e-02 1.07108450e+00
-6.76496103e-02 -3.28434616e-01 -1.11644781e+00 -5.80794871e-01
4.41600084e-02 3.37190121e-01 -1.11109042e+00 -5.69785893e-01
-4.04189080e-01 -1.83828086e-01 9.28384125e-01 2.00831980e-01
5.66161513e-01 -8.25388372e-01 -2.95170158e-01 -5.72796278e-02
-5.95416091e-02 -8.36174250e-01 -9.11670327e-01 -3.45556676e-01
-5.94156146e-01 -8.25376630e-01 -9.62805808e-01 -7.43043125e-01
1.29397309e+00 -1.94622148e-02 9.47512388e-01 2.63652980e-01
-3.09287578e-01 1.06575593e-01 -3.00002515e-01 -1.28801838e-01
-7.21813560e-01 -1.57347694e-01 9.45552066e-02 5.24561346e-01
-4.47453409e-01 -8.62963796e-01 -6.91234767e-01 5.25774539e-01
-1.11305249e+00 2.19396487e-01 5.22921920e-01 9.21660960e-01
4.19717968e-01 -3.58776674e-02 1.81359082e-01 -1.05770600e+00
4.01947260e-01 -2.84030288e-02 -4.30932164e-01 2.33311146e-01
-4.43417698e-01 2.15330943e-02 5.75744390e-01 -6.44515753e-01
-1.08697033e+00 1.12119161e-01 -4.52419259e-02 -6.33105516e-01
1.21218145e-01 -4.67628613e-02 -5.06810188e-01 -4.22217041e-01
6.50943696e-01 1.88379273e-01 4.43963498e-01 -2.51698345e-01
5.83907366e-01 4.45006132e-01 7.89919436e-01 -5.19431651e-01
1.36129558e+00 7.68454492e-01 -9.47080255e-02 -6.49528861e-01
-6.05770469e-01 2.36137807e-01 -7.53462374e-01 -8.44619945e-02
6.65653169e-01 -6.90619290e-01 -4.27581489e-01 8.02783191e-01
-1.07251489e+00 -4.07351822e-01 -4.92256731e-01 -1.55087709e-01
-4.48407501e-01 2.25467250e-01 -3.90563101e-01 -3.53972077e-01
-1.31216254e-02 -9.97359097e-01 1.19005203e+00 1.63376331e-01
-4.49304104e-01 -9.03477073e-01 3.79826091e-02 1.63814783e-01
4.73662674e-01 6.43494129e-01 8.95618737e-01 1.39156878e-01
-2.52468616e-01 -1.14486128e-01 -1.31187394e-01 4.88476366e-01
5.80833316e-01 3.71363401e-01 -1.07165384e+00 -4.60460126e-01
-2.18283787e-01 -1.24382652e-01 6.70060515e-01 -6.89438358e-02
1.02170491e+00 -5.65882444e-01 -1.84553578e-01 1.12274516e+00
9.81365085e-01 1.45546347e-01 9.38681662e-01 7.86812901e-02
7.72968471e-01 6.53415799e-01 3.40298057e-01 3.76537293e-01
7.68538117e-02 7.74976432e-01 3.31524193e-01 -5.73103666e-01
-4.11209196e-01 -7.26638436e-01 5.73945165e-01 3.70353490e-01
-1.89341962e-01 -1.84945334e-02 -4.85195518e-01 3.16662699e-01
-1.29858196e+00 -1.00282669e+00 2.70632535e-01 2.08096600e+00
1.24760926e+00 -1.05549902e-01 -9.83678363e-03 3.42087522e-02
7.50972152e-01 2.26728559e-01 -6.12538517e-01 -5.38361967e-01
-2.53423125e-01 4.18191314e-01 3.08717251e-01 6.00661516e-01
-9.57013845e-01 1.13558888e+00 6.13905716e+00 7.66620338e-01
-1.50674307e+00 4.98025231e-02 8.03178251e-01 -1.79658413e-01
-6.13433778e-01 5.54712117e-02 -4.28463995e-01 4.89646792e-01
2.28277966e-01 -1.08675271e-01 6.48757935e-01 6.92534983e-01
9.73834395e-02 1.58087358e-01 -1.11946535e+00 8.68055284e-01
3.39791209e-01 -1.31996799e+00 2.63588190e-01 3.66658866e-01
1.00789487e+00 -7.13927329e-01 6.13313854e-01 1.14615485e-01
1.30061835e-01 -1.14614177e+00 1.01203644e+00 5.11690259e-01
1.71410561e+00 -7.62494266e-01 2.60326147e-01 -7.18351007e-02
-1.00046360e+00 1.68765888e-01 -1.99867904e-01 4.68010493e-02
1.67907588e-02 6.85639441e-01 -5.65528393e-01 3.72961074e-01
5.05473495e-01 5.18180490e-01 -6.53843343e-01 4.22835559e-01
-5.24915278e-01 2.77232051e-01 -1.57112867e-01 7.22538114e-01
-5.95069192e-02 -3.68368417e-01 5.53955674e-01 8.89945388e-01
4.21737105e-01 8.76814947e-02 -1.45861492e-01 1.07870901e+00
-4.26505864e-01 -5.34463450e-02 -8.35188091e-01 7.54318535e-02
3.37176502e-01 1.29795825e+00 -4.38997835e-01 -1.29885554e-01
-1.00125141e-01 1.60367942e+00 9.47193429e-02 3.67484212e-01
-1.02457261e+00 -3.16487193e-01 9.50609267e-01 4.63779986e-01
2.66449869e-01 -2.82100346e-02 -2.79313207e-01 -1.02078605e+00
1.09140240e-01 -9.69914615e-01 -7.14908913e-02 -7.96141982e-01
-1.14768183e+00 6.97795570e-01 -2.73332536e-01 -1.20546520e+00
-1.24459863e-01 -1.10257693e-01 -7.22598493e-01 8.01085353e-01
-1.31980538e+00 -1.62699461e+00 -4.54913527e-01 5.44753432e-01
2.35611707e-01 -2.08672151e-01 8.30398381e-01 3.29332858e-01
-4.15278196e-01 1.32588971e+00 8.79424289e-02 1.76075861e-01
1.09800184e+00 -8.83643031e-01 4.79656756e-01 9.23081279e-01
8.44463259e-02 8.61433387e-01 6.57549560e-01 -6.33524537e-01
-1.13864696e+00 -1.31198919e+00 6.86333060e-01 -3.44729722e-01
4.13042963e-01 -7.01868713e-01 -7.16623127e-01 7.62835622e-01
1.17579110e-01 1.50454557e-02 5.05410671e-01 -1.29493177e-01
-6.89495265e-01 -2.80069292e-01 -1.31874108e+00 8.38282108e-01
1.35261595e+00 -6.96172476e-01 -2.99981564e-01 3.58922146e-02
4.85390216e-01 -5.36877811e-01 -6.81289315e-01 3.26288402e-01
1.04077137e+00 -1.12195778e+00 9.18966293e-01 -3.41928184e-01
7.06457675e-01 -3.25226903e-01 1.40077260e-03 -1.54935527e+00
-3.47403705e-01 -1.19795895e+00 2.41101578e-01 1.62972879e+00
2.72380650e-01 -6.17580116e-01 7.75337577e-01 4.98601645e-01
4.73540090e-03 -6.10835254e-01 -7.49629855e-01 -7.30267346e-01
1.01125903e-01 4.47937176e-02 7.65828550e-01 1.20519555e+00
-3.61277461e-01 2.82926001e-02 -5.46918869e-01 2.49326006e-02
5.45616329e-01 7.13660382e-03 1.11611652e+00 -9.50020015e-01
-6.41662255e-02 -4.35885936e-01 -4.24210012e-01 -7.93837368e-01
2.38411695e-01 -7.15241134e-01 -1.29570469e-01 -8.43280196e-01
9.19104517e-02 -5.02416611e-01 1.15808025e-01 7.52085626e-01
2.35735830e-02 9.10326719e-01 2.98272341e-01 2.29408100e-01
-5.03323302e-02 5.91280878e-01 1.61481762e+00 -8.93569887e-02
-1.32740632e-01 -2.34415889e-01 -9.44517672e-01 8.10144424e-01
6.46322846e-01 -4.23515230e-01 -4.97645974e-01 -3.89581770e-01
1.63873792e-01 -2.45368212e-01 5.45231342e-01 -8.67129564e-01
-5.89220189e-02 -1.33415595e-01 5.63545823e-01 7.15671480e-02
5.08573771e-01 -6.83160603e-01 5.56261778e-01 3.36103767e-01
-2.45737880e-01 6.10895976e-02 1.29593536e-01 1.96187213e-01
-2.36090288e-01 2.24928841e-01 1.24582696e+00 4.89309728e-02
-3.73587161e-01 4.27546054e-01 2.09069438e-02 8.73787403e-02
1.17041683e+00 -6.61779523e-01 -1.34011641e-01 -6.66246831e-01
-6.72940493e-01 -2.16937602e-01 1.24541366e+00 5.46916246e-01
6.09729469e-01 -1.49972999e+00 -8.01859796e-01 8.07326615e-01
1.84520304e-01 -1.63405508e-01 1.94167390e-01 6.70724392e-01
-5.71556985e-01 -3.08685482e-01 -6.39623582e-01 -5.33023894e-01
-1.37193573e+00 3.93705517e-01 3.54247153e-01 8.66206363e-02
-5.01021266e-01 9.33266282e-01 6.97862923e-01 -4.18841362e-01
-4.04416174e-02 -4.20938358e-02 2.12781310e-01 1.96753532e-01
3.94736081e-01 -1.87308434e-02 -6.81707710e-02 -7.65980303e-01
-2.43745089e-01 9.13539767e-01 -4.23144735e-02 -2.60705203e-01
1.24891818e+00 -1.75870061e-01 -2.91955262e-01 -2.23448761e-02
1.33449221e+00 5.47910094e-01 -1.60051191e+00 3.51136997e-02
-6.35596514e-01 -8.24957669e-01 -3.16976964e-01 -7.55169928e-01
-1.37506568e+00 7.69234300e-01 4.92773652e-01 -8.10075998e-02
1.28520739e+00 -1.45633519e-01 7.83314764e-01 -1.99840769e-01
2.94571549e-01 -8.90021384e-01 2.92862236e-01 2.99877346e-01
1.18223643e+00 -8.84573638e-01 -9.37105343e-03 -5.23667097e-01
-7.26255298e-01 8.87962997e-01 5.18138170e-01 -5.72353676e-02
5.18437564e-01 2.69634753e-01 1.54428497e-01 -5.09882122e-02
-3.08601946e-01 1.80650204e-01 3.21490824e-01 6.98429585e-01
1.96511775e-01 -2.89129484e-02 9.26564410e-02 1.73830688e-01
-7.31477618e-01 -9.08525959e-02 4.35038418e-01 6.11327350e-01
3.28684710e-02 -1.35645616e+00 -4.85602915e-01 1.28523186e-01
-2.77397782e-01 -1.25794252e-02 -6.87174261e-01 8.10762823e-01
5.30802071e-01 5.69930375e-01 7.89536983e-02 -5.03178358e-01
3.60990465e-01 -1.21847011e-01 5.93706131e-01 -5.72378516e-01
-4.32608813e-01 -9.59778950e-02 -1.54784322e-01 -6.86660826e-01
-1.63551599e-01 -5.13489306e-01 -8.67892504e-01 -6.74848020e-01
-8.55403543e-02 -1.63251728e-01 5.29920995e-01 5.47933936e-01
4.36487347e-01 2.99054176e-01 8.41035068e-01 -8.67529750e-01
-2.21206054e-01 -6.42914414e-01 -5.37698507e-01 8.26063573e-01
4.37119156e-01 -7.95764446e-01 -3.10849845e-01 5.11996448e-01] | [12.506314277648926, -0.21218331158161163] |
156b372a-fdb0-4bc7-b826-dba2750e1ad6 | rotation-equivariant-vector-field-networks | 1612.09346 | null | http://arxiv.org/abs/1612.09346v3 | http://arxiv.org/pdf/1612.09346v3.pdf | Rotation equivariant vector field networks | In many computer vision tasks, we expect a particular behavior of the output
with respect to rotations of the input image. If this relationship is
explicitly encoded, instead of treated as any other variation, the complexity
of the problem is decreased, leading to a reduction in the size of the required
model. In this paper, we propose the Rotation Equivariant Vector Field Networks
(RotEqNet), a Convolutional Neural Network (CNN) architecture encoding rotation
equivariance, invariance and covariance. Each convolutional filter is applied
at multiple orientations and returns a vector field representing magnitude and
angle of the highest scoring orientation at every spatial location. We develop
a modified convolution operator relying on this representation to obtain deep
architectures. We test RotEqNet on several problems requiring different
responses with respect to the inputs' rotation: image classification,
biomedical image segmentation, orientation estimation and patch matching. In
all cases, we show that RotEqNet offers extremely compact models in terms of
number of parameters and provides results in line to those of networks orders
of magnitude larger. | ['Devis Tuia', 'Nikos Komodakis', 'Michele Volpi', 'Diego Marcos'] | 2016-12-29 | rotation-equivariant-vector-field-networks-1 | http://openaccess.thecvf.com/content_iccv_2017/html/Marcos_Rotation_Equivariant_Vector_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Marcos_Rotation_Equivariant_Vector_ICCV_2017_paper.pdf | iccv-2017-10 | ['patch-matching', 'colorectal-gland-segmentation', 'breast-tumour-classification', 'multi-tissue-nucleus-segmentation'] | ['computer-vision', 'medical', 'medical', 'medical'] | [ 3.40581357e-01 1.63197562e-01 1.28516674e-01 -6.18483365e-01
2.93740015e-02 -6.65259898e-01 4.60679263e-01 -6.19479492e-02
-7.25351751e-01 4.35029179e-01 6.95498951e-04 -2.74887949e-01
-3.22314888e-01 -8.13528299e-01 -1.00307190e+00 -7.37223983e-01
-1.52925059e-01 1.01106830e-01 2.03671187e-01 -3.92743140e-01
3.96504313e-01 1.16671622e+00 -1.37485528e+00 2.93771327e-01
2.53001750e-01 1.09932387e+00 2.75220275e-02 8.96553278e-01
4.53485399e-01 5.70207119e-01 -3.91640931e-01 -6.65480569e-02
4.34825331e-01 -1.34764180e-01 -1.04174376e+00 -3.42529267e-02
6.61580741e-01 -9.83859003e-02 -4.35964882e-01 9.18584287e-01
4.54832762e-01 1.15275763e-01 8.87340844e-01 -7.44815111e-01
-6.23055041e-01 4.59578127e-01 -2.25046247e-01 2.86280155e-01
1.44935131e-01 -1.76573515e-01 1.04911554e+00 -7.68853903e-01
9.09522474e-01 1.15535128e+00 7.62775123e-01 2.53985733e-01
-1.47614229e+00 -3.05938590e-02 3.71998735e-02 4.24484909e-02
-1.23383832e+00 -1.64162725e-01 6.29029214e-01 -3.98121089e-01
1.04208195e+00 3.55638683e-01 5.49256027e-01 8.39962780e-01
4.40653801e-01 4.21917319e-01 8.71079743e-01 -4.23489302e-01
1.31877810e-01 1.52605604e-02 1.17117904e-01 6.18370771e-01
1.96882963e-01 -8.71972144e-02 -1.94193095e-01 3.07020187e-01
1.27370667e+00 -2.41590645e-02 -2.97382116e-01 -5.92007458e-01
-1.35702586e+00 7.00154662e-01 9.37542200e-01 5.22164702e-01
-3.58490616e-01 2.58501351e-01 1.56433210e-01 4.66107339e-01
1.24882355e-01 7.63200283e-01 -6.44313931e-01 4.52954471e-01
-5.93725085e-01 4.33801264e-01 6.15924418e-01 7.52355039e-01
9.12201643e-01 1.10817514e-01 -1.08082727e-01 7.50171483e-01
8.80872980e-02 3.85872960e-01 6.02290213e-01 -7.86445558e-01
8.10905099e-02 5.73153794e-01 -2.46813878e-01 -1.37684178e+00
-1.06828976e+00 -7.91626811e-01 -1.04601765e+00 2.18357012e-01
5.32140911e-01 2.64727008e-02 -1.11983633e+00 1.77758908e+00
1.38486058e-01 -2.62624741e-01 9.00588408e-02 9.51726317e-01
6.70018613e-01 3.68969351e-01 -3.47972840e-01 2.84272015e-01
1.47760129e+00 -3.89554948e-01 -3.55838537e-01 -3.12347189e-02
6.51731193e-01 -8.41915905e-01 6.72370493e-01 3.90661329e-01
-9.16599095e-01 -6.87531531e-01 -1.29721916e+00 -1.22605726e-01
-6.82995141e-01 3.33566308e-01 6.70265853e-01 4.93738383e-01
-1.27531779e+00 1.00852013e+00 -6.38792336e-01 -3.52572173e-01
1.66864410e-01 6.91874981e-01 -6.84713423e-01 1.07077509e-01
-1.01136255e+00 8.31723392e-01 2.84638315e-01 3.80819261e-01
-2.98601598e-01 -5.43603957e-01 -1.00574589e+00 2.32087359e-01
-3.24005932e-01 -5.80319762e-01 1.03260934e+00 -1.26532841e+00
-1.27088785e+00 7.13583052e-01 3.95393819e-02 -4.42152768e-01
5.97758293e-01 1.59422562e-01 -1.87340915e-01 4.84481640e-02
-1.44676670e-01 1.00395131e+00 9.71577704e-01 -7.03732967e-01
-2.48740032e-01 -5.09434223e-01 2.19187483e-01 -8.64421353e-02
-1.58029616e-01 -2.15889305e-01 -1.83609739e-01 -8.13307524e-01
5.16472459e-01 -1.03227258e+00 -4.71135348e-01 3.92057374e-02
-2.14296296e-01 -6.07872382e-02 7.02103615e-01 -3.74918342e-01
7.06149876e-01 -2.23114944e+00 2.57318348e-01 4.67179894e-01
1.17264278e-01 5.36577031e-02 -3.08960229e-01 1.37872651e-01
-7.31143415e-01 -6.71788491e-03 -1.51263475e-01 2.29464054e-01
-2.17842355e-01 2.66264319e-01 -1.51025444e-01 8.23506594e-01
6.42028928e-01 7.57543206e-01 -5.02072155e-01 -2.04442769e-01
2.48518303e-01 8.39700222e-01 -9.62701738e-01 -9.89810005e-02
1.88374370e-01 2.61461407e-01 -2.90048152e-01 2.20456496e-01
8.91720831e-01 -2.96734393e-01 3.15820873e-01 -6.20529830e-01
-7.80140832e-02 9.43430588e-02 -1.54724097e+00 1.43878114e+00
-4.70251977e-01 1.00220215e+00 -3.07889938e-01 -1.40953290e+00
1.12037289e+00 2.68227696e-01 6.31845415e-01 -7.99316406e-01
2.71434516e-01 1.49043635e-01 1.93855986e-01 -3.06392014e-01
5.60637474e-01 2.24665597e-01 9.34979320e-02 1.63145259e-01
3.97968054e-01 5.57225533e-02 1.23846851e-01 -1.39221027e-01
9.19009030e-01 7.39874598e-03 2.98201084e-01 -5.95694244e-01
6.32354319e-01 -3.81579667e-01 2.46433988e-01 7.14184642e-01
1.50777280e-01 8.48444402e-01 7.10128725e-01 -1.06894112e+00
-1.06825387e+00 -8.84656310e-01 -5.86743414e-01 7.84014463e-01
-1.12599470e-01 -9.06675383e-02 -9.25244868e-01 -3.74517888e-01
-8.95664990e-02 -2.51872651e-02 -9.32915151e-01 -5.66375591e-02
-1.10410619e+00 -8.83358538e-01 5.55895448e-01 7.32840061e-01
5.48812449e-01 -1.15871310e+00 -9.32850301e-01 1.33373156e-01
3.32895964e-01 -1.04318154e+00 -3.33903849e-01 5.86536229e-01
-9.96731520e-01 -1.11268818e+00 -8.12621951e-01 -9.23776031e-01
8.59349549e-01 1.31072002e-02 8.77931476e-01 -2.40887120e-01
-5.56152225e-01 7.94306472e-02 -2.05187544e-01 -2.00608134e-01
1.69258844e-02 2.95344919e-01 -9.40698013e-02 1.94399133e-01
2.54602171e-02 -6.86079025e-01 -7.85440564e-01 3.81920666e-01
-1.20675921e+00 -2.02108175e-01 7.03212976e-01 8.96066427e-01
4.47633624e-01 -2.87957162e-01 3.78612280e-01 -8.16989779e-01
4.44205284e-01 -4.45655026e-02 -7.05949187e-01 -1.95579063e-02
-3.93792868e-01 5.61713994e-01 7.80035853e-01 -5.33779919e-01
-4.92823631e-01 5.25833488e-01 -1.92665473e-01 -1.53902113e-01
-2.71156341e-01 4.04531926e-01 1.84189811e-01 -3.92201126e-01
8.71597588e-01 2.99227107e-02 -1.51129454e-01 -5.25080621e-01
3.61929327e-01 3.77357125e-01 5.68509936e-01 -2.23800287e-01
5.42949975e-01 4.38218474e-01 5.71569681e-01 -9.46663976e-01
-3.98412168e-01 -3.05840045e-01 -9.73363519e-01 -1.69378519e-01
1.01479959e+00 -5.18224120e-01 -9.14575577e-01 5.33491910e-01
-1.37924147e+00 7.01110438e-03 -2.90532857e-01 7.14275539e-01
-6.27744675e-01 1.40131088e-02 -3.37776065e-01 -2.63701290e-01
-1.22406878e-01 -1.54913771e+00 9.52695251e-01 2.04473883e-01
-7.33220764e-03 -1.04169202e+00 4.79958095e-02 -3.85687381e-01
5.58245242e-01 1.74096048e-01 9.88353610e-01 -5.68979681e-01
-5.27236819e-01 -4.22279209e-01 -4.00628358e-01 4.27864492e-01
-1.35749310e-01 -3.02251372e-02 -1.05643082e+00 -3.13937604e-01
-1.36017904e-01 9.99903772e-03 1.02774334e+00 6.21025920e-01
1.44374073e+00 -2.04233423e-01 4.69937474e-02 1.08864284e+00
1.50033498e+00 7.93204978e-02 7.73927093e-01 4.09733742e-01
5.90353787e-01 5.64389348e-01 -2.86251865e-03 1.97113842e-01
-7.34641105e-02 8.81556571e-01 4.70947891e-01 -4.94393766e-01
1.92200560e-02 1.26587197e-01 1.27846494e-01 4.92192894e-01
-5.26539147e-01 -1.68170482e-02 -6.01037860e-01 3.94564480e-01
-1.58057356e+00 -8.86433363e-01 -2.96563562e-02 2.09812069e+00
3.55204761e-01 -4.71731685e-02 -1.41397277e-02 3.24502230e-01
5.77328563e-01 2.69763976e-01 -4.06977057e-01 -7.90907383e-01
-2.39418387e-01 6.42307162e-01 9.65202510e-01 4.59103882e-01
-1.22620487e+00 6.65832520e-01 7.16875172e+00 4.77639109e-01
-1.77907145e+00 -3.80693316e-01 6.15944505e-01 4.32775348e-01
-1.48139820e-01 -1.60153940e-01 -4.41595018e-01 -2.38435477e-01
6.01381421e-01 4.28250372e-01 6.01157486e-01 8.44321191e-01
-2.20773265e-01 3.34036320e-01 -1.16254842e+00 8.54865074e-01
6.05439171e-02 -1.40467823e+00 1.25304550e-01 1.10228270e-01
5.28223336e-01 7.14225322e-02 3.14543784e-01 -8.90710428e-02
-7.11230338e-02 -1.28856659e+00 6.50741637e-01 5.28314412e-01
6.94759011e-01 -7.14317679e-01 6.54457808e-01 -1.01000994e-01
-1.02277219e+00 -2.49909479e-02 -5.77320874e-01 1.18572354e-01
-3.22992921e-01 1.86452299e-01 -9.36693132e-01 4.84690964e-01
7.46203601e-01 6.82064295e-01 -6.55140281e-01 7.91054010e-01
9.00741965e-02 1.94858715e-01 -2.71508455e-01 -6.18637502e-02
3.57966006e-01 -1.38451919e-01 5.11073649e-01 1.37829590e+00
2.77106375e-01 -2.50709981e-01 -3.16839278e-01 7.52041578e-01
-1.78160846e-01 1.82306126e-01 -6.93401039e-01 3.01442146e-01
-7.84875080e-02 1.37956989e+00 -9.06231701e-01 -1.13749675e-01
-1.94288746e-01 8.46100211e-01 3.62322688e-01 6.01134002e-01
-5.14411569e-01 -8.86589229e-01 6.33186579e-01 8.02257285e-02
7.34400749e-01 -1.79340646e-01 -1.14051037e-01 -1.06451309e+00
1.17635623e-01 -7.08918869e-01 1.66112036e-01 -7.43134975e-01
-7.48754263e-01 9.16717827e-01 -7.41137266e-02 -1.19125783e+00
-4.09208953e-01 -1.39741850e+00 -3.73482674e-01 9.04330194e-01
-1.33297026e+00 -8.46427977e-01 -1.76975533e-01 6.11553609e-01
2.17783406e-01 -8.38378444e-02 8.09371710e-01 3.29557240e-01
-1.81705236e-01 7.39589512e-01 -8.78672600e-02 4.44537342e-01
6.08326912e-01 -1.32119739e+00 3.82787108e-01 6.27985537e-01
1.88880473e-01 7.89189756e-01 7.11876273e-01 1.11550011e-01
-1.37494457e+00 -1.02158391e+00 8.46674323e-01 -2.16851264e-01
2.90625602e-01 -3.25681686e-01 -7.42668927e-01 7.17982292e-01
8.79264101e-02 3.95776510e-01 1.67288750e-01 -4.78306748e-02
-6.13759458e-01 -3.61026138e-01 -1.12279546e+00 5.05817831e-01
8.38293731e-01 -4.85659570e-01 -2.27201998e-01 2.02080965e-01
6.77917719e-01 -3.53758454e-01 -9.65896845e-01 4.94766444e-01
8.35949242e-01 -9.68848348e-01 1.19984090e+00 -9.02546227e-01
3.51910472e-01 -2.48543367e-01 -3.24579298e-01 -1.28373671e+00
-5.53995550e-01 -3.00142139e-01 5.64461291e-01 4.65200216e-01
5.75975657e-01 -8.31048310e-01 5.01436830e-01 -3.75609333e-03
-1.65877253e-01 -7.89998472e-01 -9.73321855e-01 -5.34750342e-01
1.99476555e-01 -2.80695289e-01 5.86821616e-01 9.11724091e-01
-3.00526917e-01 2.92926967e-01 -2.05513775e-01 2.13888139e-01
9.46871862e-02 -4.74587791e-02 5.24231791e-01 -1.10649490e+00
-2.79962599e-01 -5.56510031e-01 -1.07570350e+00 -1.03880632e+00
-1.00712337e-01 -1.10598779e+00 -1.54582307e-01 -1.13857067e+00
-1.96211874e-01 -2.96011597e-01 -3.86676401e-01 4.27805007e-01
3.25510621e-01 4.69752640e-01 5.33232279e-02 -1.09745264e-01
-1.57453254e-01 -4.02953401e-02 1.41689122e+00 -8.85081813e-02
-1.37170598e-01 4.78904247e-02 -4.31197762e-01 7.25859284e-01
7.73769021e-01 -3.67083192e-01 -9.87078622e-02 -4.95071411e-01
4.39758509e-01 5.94983902e-03 5.91998816e-01 -1.08427417e+00
3.61739397e-02 1.19788917e-02 9.01619375e-01 -3.67071062e-01
2.34506413e-01 -7.80651093e-01 1.58864573e-01 5.55344641e-01
-5.62995195e-01 5.67341626e-01 1.05227776e-01 1.49048090e-01
-2.55010009e-01 -2.72263527e-01 8.85335505e-01 1.70851219e-03
-3.86468142e-01 2.67154068e-01 -5.20667613e-01 -4.85944092e-01
4.65523481e-01 -2.43329212e-01 -2.26473480e-01 -1.65521309e-01
-8.91518056e-01 -4.57006246e-01 2.83244431e-01 2.40841046e-01
5.70813417e-01 -1.27766466e+00 -4.46716160e-01 7.05915630e-01
1.36308279e-02 6.51360815e-03 2.57580757e-01 8.43848407e-01
-9.21717107e-01 7.48405039e-01 -6.19718850e-01 -7.16300011e-01
-1.13949656e+00 5.82571328e-01 7.86897838e-01 -2.20514685e-01
-4.92849946e-01 4.77614850e-01 2.80587703e-01 -6.40653729e-01
8.36037248e-02 -6.88366055e-01 -6.12993777e-01 -3.44529003e-02
3.80237192e-01 1.19125456e-01 4.23351198e-01 -7.80641258e-01
-3.43895763e-01 9.38612163e-01 4.86793742e-02 -7.32921883e-02
1.39036274e+00 4.52418000e-01 -1.51845649e-01 -6.11897334e-02
1.60376656e+00 -1.74579844e-01 -9.53763723e-01 -1.44692466e-01
-2.18702480e-01 -2.91316688e-01 1.39421046e-01 -4.57294643e-01
-1.38776410e+00 9.83059525e-01 9.29451823e-01 2.07813933e-01
1.17514217e+00 -1.56759664e-01 1.05188712e-01 8.50724459e-01
8.56098998e-03 -8.88608634e-01 -6.47862330e-02 6.48518205e-01
1.06906796e+00 -1.01917708e+00 -7.91363716e-02 -1.30393356e-01
-2.26992160e-01 1.49981511e+00 3.75234157e-01 -7.32838690e-01
8.22061360e-01 1.38903528e-01 1.56757966e-01 -1.94889009e-01
-3.93372983e-01 -7.02358782e-02 6.53992116e-01 6.37216270e-01
7.84324825e-01 7.63181895e-02 -3.31522644e-01 7.60092214e-02
-3.33747357e-01 -3.64536256e-01 3.78598630e-01 7.63238728e-01
-3.23667973e-01 -1.08747411e+00 -3.66094142e-01 1.38726220e-01
-5.47989666e-01 1.20777205e-01 -3.26235145e-01 8.67649436e-01
1.13995805e-01 2.99016297e-01 4.80733722e-01 -3.39648098e-01
5.51612198e-01 -3.54214281e-01 6.99973702e-01 -2.33058646e-01
-5.05960166e-01 8.40620846e-02 -1.86847016e-01 -6.48829162e-01
-4.45434928e-01 -4.86176163e-01 -1.09779274e+00 -5.94396368e-02
-9.47917476e-02 -1.97553799e-01 9.61747706e-01 7.63743937e-01
1.92138970e-01 7.65305221e-01 5.57931423e-01 -9.69697356e-01
-4.65745658e-01 -9.43054080e-01 -5.46877205e-01 5.07450104e-01
6.13427222e-01 -5.04028082e-01 -1.39004946e-01 2.24200450e-02] | [9.146623611450195, 2.250411033630371] |
bbd397a9-feac-4271-95e2-c5c9d96cacf4 | distilling-causal-effect-from-miscellaneous | 2210.03980 | null | https://arxiv.org/abs/2210.03980v1 | https://arxiv.org/pdf/2210.03980v1.pdf | Distilling Causal Effect from Miscellaneous Other-Class for Continual Named Entity Recognition | Continual Learning for Named Entity Recognition (CL-NER) aims to learn a growing number of entity types over time from a stream of data. However, simply learning Other-Class in the same way as new entity types amplifies the catastrophic forgetting and leads to a substantial performance drop. The main cause behind this is that Other-Class samples usually contain old entity types, and the old knowledge in these Other-Class samples is not preserved properly. Thanks to the causal inference, we identify that the forgetting is caused by the missing causal effect from the old data. To this end, we propose a unified causal framework to retrieve the causality from both new entity types and Other-Class. Furthermore, we apply curriculum learning to mitigate the impact of label noise and introduce a self-adaptive weight for balancing the causal effects between new entity types and Other-Class. Experimental results on three benchmark datasets show that our method outperforms the state-of-the-art method by a large margin. Moreover, our method can be combined with the existing state-of-the-art methods to improve the performance in CL-NER | ['Qianli Ma', 'Haibin Chen', 'Zhanxian Liang', 'Junhao Zheng'] | 2022-10-08 | null | null | null | null | ['miscellaneous', 'continual-named-entity-recognition', 'fg-1-pg-1'] | ['miscellaneous', 'natural-language-processing', 'natural-language-processing'] | [-1.56430408e-01 1.10535182e-01 -4.20367867e-01 -3.51572990e-01
-3.80589783e-01 -2.47523561e-01 3.77908528e-01 4.74220812e-01
-6.45759284e-01 1.07565415e+00 3.55401486e-01 4.48166542e-02
-1.04022190e-01 -9.86902118e-01 -1.08706653e+00 -6.15097821e-01
-5.45277167e-03 3.55253965e-01 5.15434623e-01 5.07568941e-02
-1.46418318e-01 1.97821297e-02 -1.52539337e+00 2.13825032e-01
1.04357493e+00 5.98443031e-01 1.32543609e-01 6.85334504e-02
-5.78133345e-01 1.07053721e+00 -5.85424662e-01 -5.15763462e-01
-2.94758976e-01 -2.17744023e-01 -6.92967951e-01 -3.65505815e-01
3.35648417e-01 -2.28711307e-01 -5.11698544e-01 9.53654766e-01
3.67161423e-01 1.03720397e-01 5.11808634e-01 -1.32922602e+00
-7.21285701e-01 1.01454604e+00 -5.02409577e-01 1.83215499e-01
-1.45169184e-01 -8.86694193e-02 9.87641335e-01 -1.23304868e+00
6.80701017e-01 1.34318769e+00 8.60583782e-01 7.09422946e-01
-9.10336077e-01 -1.16516924e+00 7.83448398e-01 3.54340017e-01
-1.20642257e+00 -3.74217868e-01 6.46736026e-01 -2.34496415e-01
6.32469356e-01 -1.67999953e-01 2.79600590e-01 1.36442947e+00
7.57216737e-02 1.15502453e+00 8.14135611e-01 -2.74404854e-01
3.08710337e-01 1.31426677e-01 5.17637789e-01 5.60531437e-01
5.03088236e-01 1.05182894e-01 -7.93324471e-01 -1.22462176e-01
2.76787847e-01 3.13533664e-01 -1.44320175e-01 -2.61805415e-01
-9.64764714e-01 4.11720365e-01 4.52730507e-01 3.09627682e-01
-3.81953865e-01 1.34075314e-01 2.49558643e-01 2.01589599e-01
5.15487492e-01 1.01361476e-01 -1.21079457e+00 3.63508752e-03
-5.77808678e-01 1.77631706e-01 6.62580073e-01 9.99411166e-01
7.79617369e-01 2.72643822e-03 -2.20103249e-01 8.09216380e-01
3.03259790e-01 2.76044548e-01 5.99708319e-01 -2.57389337e-01
4.86026347e-01 9.52505708e-01 8.80813450e-02 -4.36976433e-01
-4.53472018e-01 -7.11503625e-01 -9.84444439e-01 -2.91883409e-01
4.85898107e-01 -5.07730663e-01 -1.02361512e+00 2.18275523e+00
5.77946663e-01 9.66476500e-01 1.74540848e-01 3.89384210e-01
7.49670565e-01 5.86841345e-01 5.91790080e-01 -4.24842119e-01
1.24442649e+00 -7.67650127e-01 -9.68124211e-01 -4.16444182e-01
4.13781703e-01 -3.78553540e-01 9.75345671e-01 2.41324976e-01
-4.21590358e-01 -4.89146560e-01 -9.70155180e-01 2.73001075e-01
-4.43124235e-01 -3.12573314e-02 6.64343238e-01 3.91632169e-01
-2.19919473e-01 7.82814801e-01 -7.56264865e-01 -9.43179652e-02
5.08733809e-01 7.56076723e-02 -9.31398869e-02 -1.85863107e-01
-1.72360241e+00 5.22565842e-01 7.30694056e-01 -1.07559629e-01
-8.74867737e-01 -1.46707392e+00 -4.79491115e-01 2.50660032e-01
7.06625998e-01 -7.31819928e-01 1.15316474e+00 -5.95894039e-01
-9.70744371e-01 1.87362164e-01 -2.99709648e-01 -3.60701799e-01
5.42034388e-01 -6.87323332e-01 -7.02612221e-01 -4.52655882e-01
4.80156764e-02 2.61722326e-01 8.20155501e-01 -1.44976866e+00
-1.16933632e+00 -3.54685724e-01 -2.88674980e-01 2.99951825e-02
-7.87791193e-01 -4.64990586e-01 -4.75511670e-01 -8.72161031e-01
-3.59796360e-02 -7.19613433e-01 -6.59691403e-03 -4.78121191e-01
-1.79582864e-01 -6.81762695e-01 8.52683425e-01 -3.85372192e-01
1.64439178e+00 -2.15717149e+00 -1.63393412e-02 -2.13188946e-01
3.78290564e-01 2.62537956e-01 -5.08202314e-02 2.00418383e-01
-2.71667093e-01 2.16446042e-01 -7.38657312e-03 -3.75934392e-01
-3.08794230e-01 3.80422354e-01 -6.57023251e-01 7.22488537e-02
2.93012410e-01 5.95672309e-01 -1.40706038e+00 -2.10992053e-01
-1.33608758e-01 3.36475968e-01 -4.31850255e-01 2.82300323e-01
-2.71601439e-01 3.47832412e-01 -4.45964038e-01 4.10114944e-01
6.77352965e-01 -1.53773531e-01 1.91450804e-01 -6.44673109e-02
-6.44371808e-02 3.80626231e-01 -1.42974961e+00 1.52208781e+00
-2.71075070e-01 1.52187916e-02 -3.89233917e-01 -6.03086829e-01
7.45584905e-01 3.38759929e-01 4.01352078e-01 -4.50155914e-01
-2.47399122e-01 2.93625265e-01 -1.87864408e-01 -4.31264669e-01
4.52649266e-01 -3.84630591e-01 -6.30158409e-02 2.39952534e-01
3.65958244e-01 7.56382763e-01 1.38805315e-01 3.09540123e-01
1.23362029e+00 -7.65585676e-02 3.22224349e-02 1.14474908e-01
2.25755334e-01 -3.11015904e-01 1.47800291e+00 1.01522672e+00
-1.51287928e-01 5.96366301e-02 3.16438287e-01 -4.72047269e-01
-8.49229455e-01 -1.08351588e+00 8.23724177e-03 1.26306033e+00
2.21097544e-01 -3.66421759e-01 -3.75441253e-01 -1.20811808e+00
2.39575416e-01 1.08690858e+00 -6.97754204e-01 -5.34563184e-01
-6.29745424e-01 -1.23998535e+00 4.91965532e-01 7.72565246e-01
4.51643884e-01 -1.10537827e+00 -2.00070608e-02 5.45939624e-01
-3.25437427e-01 -8.78686130e-01 -3.48389566e-01 3.93713653e-01
-1.09101510e+00 -1.09981680e+00 -5.26063204e-01 -5.77558041e-01
5.92150450e-01 9.42467600e-02 1.13432670e+00 -7.57489502e-02
1.53311919e-02 4.24261354e-02 -3.88526171e-01 -6.73297107e-01
-3.84027153e-01 4.06267762e-01 3.23464960e-01 2.25567803e-01
4.16661769e-01 -5.94919682e-01 -3.77499998e-01 1.28256842e-01
-8.25306237e-01 -1.91755444e-01 7.76440561e-01 1.06495059e+00
4.36650902e-01 6.60511911e-01 1.14935541e+00 -1.41166401e+00
2.06867993e-01 -9.70870793e-01 -2.74503350e-01 4.51076418e-01
-1.18223202e+00 3.98959637e-01 6.15997195e-01 -8.58895659e-01
-1.63134003e+00 -1.12415537e-01 1.46895617e-01 -4.46489960e-01
-1.05443999e-01 6.00937903e-01 -5.58182418e-01 7.30976999e-01
4.44167942e-01 1.82320811e-02 -7.58524656e-01 -8.29020143e-01
4.66130227e-01 3.70683908e-01 6.64399385e-01 -5.22469699e-01
8.52007926e-01 3.98341686e-01 -3.61197889e-01 -4.48037982e-01
-1.32864010e+00 -5.33697784e-01 -6.41609371e-01 -1.74349949e-01
4.22163695e-01 -1.20576847e+00 -3.00583839e-01 9.41073298e-01
-1.14829540e+00 -2.49188036e-01 -4.64845449e-01 4.76173162e-01
2.32581377e-01 9.65508446e-02 -5.74635684e-01 -8.28977704e-01
-2.35845208e-01 -5.36489666e-01 6.70012653e-01 4.73357171e-01
8.89720321e-02 -8.74076426e-01 1.64069682e-01 -4.07680757e-02
1.61310703e-01 -1.58593714e-01 1.15233755e+00 -7.40819275e-01
-5.88045359e-01 -7.88144097e-02 -6.50476515e-02 6.58233315e-02
2.01377302e-01 -1.69243544e-01 -9.40284550e-01 -2.94904083e-01
-1.51417106e-01 -3.50707211e-02 1.36302054e+00 9.75512806e-03
8.25097382e-01 -4.31130022e-01 -5.59360683e-01 2.03829572e-01
1.32129192e+00 9.13763419e-02 4.81989563e-01 3.51037830e-01
9.78554547e-01 4.94199097e-01 7.49535441e-01 5.92530906e-01
7.37186909e-01 2.44647235e-01 2.76058227e-01 -3.69712263e-02
-3.67885053e-01 -8.61522198e-01 3.87338936e-01 9.62877512e-01
3.49601030e-01 -2.78683871e-01 -9.46635306e-01 8.56202364e-01
-2.07756615e+00 -9.81373549e-01 -3.45802337e-01 2.31348014e+00
1.40763748e+00 2.87935257e-01 -1.72923788e-01 1.10271201e-01
9.02145207e-01 -9.20965746e-02 -9.47310746e-01 5.60976386e-01
-2.13289484e-01 -5.03382199e-02 6.54162049e-01 2.44864270e-01
-1.20936120e+00 1.08184505e+00 5.52393389e+00 8.92138958e-01
-8.43966365e-01 2.53222048e-01 4.80254382e-01 7.18407035e-02
-3.89180690e-01 1.96869031e-01 -1.35179687e+00 5.52477896e-01
9.26051855e-01 -3.41300279e-01 4.25920784e-02 7.27506757e-01
-1.25979066e-01 1.50703505e-01 -1.04102421e+00 5.68386555e-01
-1.75639600e-01 -1.00452065e+00 4.48791310e-02 -2.85983890e-01
8.77574384e-01 6.75818697e-02 -1.20089278e-01 8.46407950e-01
7.20821798e-01 -4.14849520e-01 7.76934564e-01 7.24674165e-01
5.24102688e-01 -7.96731174e-01 7.18465447e-01 5.71053147e-01
-1.13616240e+00 -3.88062477e-01 -4.17168826e-01 1.62311032e-01
-8.92700180e-02 1.22114730e+00 -7.05515802e-01 6.74817979e-01
8.47936273e-01 7.81655550e-01 -6.39966726e-01 1.13045537e+00
-6.85538352e-01 1.04880679e+00 -8.50625634e-02 1.40586555e-01
-2.89212614e-01 3.68869483e-01 6.44982934e-01 1.13093019e+00
1.48691833e-01 8.55894759e-02 -7.00804368e-02 6.70534551e-01
-5.67631066e-01 2.02559158e-02 -1.35546193e-01 1.34293452e-01
1.00607634e+00 1.04071367e+00 -3.97828937e-01 -5.82554281e-01
-4.36873972e-01 7.57789850e-01 5.43355703e-01 3.98783624e-01
-7.38276482e-01 -3.28593135e-01 5.76082349e-01 -7.27959350e-02
4.91139859e-01 1.81950152e-01 -2.29219764e-01 -1.50157630e+00
-1.45363221e-02 -5.61779022e-01 8.28828573e-01 -3.92603815e-01
-1.80107725e+00 1.80105269e-01 -2.06749916e-01 -8.89625490e-01
1.72760874e-01 -1.39809981e-01 -4.13337439e-01 4.18890029e-01
-1.90001917e+00 -9.01417911e-01 -2.94691503e-01 3.97000521e-01
5.41583538e-01 -1.29927725e-01 7.17629135e-01 6.65361047e-01
-8.79911721e-01 8.32859457e-01 2.30493397e-01 3.28769863e-01
1.23433876e+00 -1.44171715e+00 3.38070929e-01 9.78449047e-01
3.11874412e-02 7.45633662e-01 4.83610511e-01 -1.10141015e+00
-1.18115914e+00 -1.47722483e+00 1.28492856e+00 -6.02125764e-01
7.13340521e-01 -3.63982409e-01 -1.31271064e+00 6.58251703e-01
-8.86393264e-02 -9.09240246e-02 6.79032326e-01 6.99602067e-01
-8.31534147e-01 -4.32387829e-01 -8.47590387e-01 5.20950496e-01
1.00911784e+00 -1.25572771e-01 -9.70334113e-01 -7.99633265e-02
1.13339305e+00 -2.22351819e-01 -7.70525992e-01 6.16660953e-01
3.57188165e-01 -5.25586724e-01 8.03563893e-01 -8.43593359e-01
4.56463754e-01 -4.46369499e-01 3.13851923e-01 -1.55168784e+00
-3.04135740e-01 -4.75163847e-01 -6.44935787e-01 1.97063482e+00
3.96246165e-01 -5.49607158e-01 6.33927703e-01 6.12440169e-01
6.50081318e-03 -3.16081673e-01 -8.69883537e-01 -8.38435113e-01
2.44359002e-01 -3.62754732e-01 7.22060263e-01 1.30566525e+00
-3.42204779e-01 7.75576293e-01 -5.63680172e-01 4.58604425e-01
9.10180271e-01 -5.02692759e-02 5.27174592e-01 -1.63618195e+00
-2.39042249e-02 -1.02212064e-01 6.40655756e-02 -8.78424764e-01
1.58504158e-01 -7.42646754e-01 4.61433530e-02 -1.21123505e+00
4.30702239e-01 -1.06337261e+00 -9.36613202e-01 9.83641624e-01
-1.04011583e+00 -5.20749092e-01 7.09118322e-02 3.14200282e-01
-7.63530135e-01 8.01624537e-01 8.52633595e-01 -7.75476247e-02
-3.61666322e-01 -8.06908160e-02 -8.73699129e-01 7.06886649e-01
4.99714166e-01 -1.11442006e+00 -2.74855137e-01 -2.95898616e-01
4.03085202e-01 -2.92558193e-01 1.81350380e-01 -8.24020326e-01
5.74270129e-01 -2.61507463e-03 4.45045620e-01 -7.11696386e-01
-3.78377765e-01 -7.73261130e-01 7.56648108e-02 3.52132022e-01
-4.47481543e-01 -3.81493062e-01 1.84671625e-01 1.14751327e+00
-5.80308028e-02 -2.47856244e-01 6.13648653e-01 6.48150742e-02
-9.08543050e-01 3.46446276e-01 -9.23906788e-02 5.07581949e-01
7.34644353e-01 5.07488906e-01 -4.64116096e-01 1.10831805e-01
-5.63857079e-01 5.15520751e-01 1.42603982e-02 8.93959880e-01
4.42622066e-01 -1.43645322e+00 -7.91739881e-01 1.20232269e-01
2.02385426e-01 2.33943790e-01 5.03700256e-01 6.01475656e-01
5.48890650e-01 -3.29539105e-02 2.06723392e-01 -2.33544379e-01
-1.14010501e+00 6.73624337e-01 1.00253701e-01 -5.93458951e-01
-4.88265038e-01 9.38571513e-01 4.01527792e-01 -5.72707772e-01
5.76644301e-01 -1.99902520e-01 -3.54698479e-01 3.93630832e-01
6.78867340e-01 4.48718071e-01 7.49248564e-02 1.82826333e-02
-4.05686975e-01 4.04966157e-03 -5.38973927e-01 2.23125234e-01
1.70971358e+00 -1.97709575e-01 7.06348103e-03 9.36545014e-01
7.72866547e-01 8.87751952e-02 -1.36296093e+00 -5.99475205e-01
3.40605557e-01 -1.49740905e-01 -3.23397927e-02 -1.06742752e+00
-1.09578335e+00 5.89868069e-01 6.30067885e-01 -1.76037863e-01
1.03318918e+00 -3.40902098e-02 8.20123434e-01 2.93412387e-01
3.90754044e-01 -1.15907466e+00 1.38300881e-01 6.94337249e-01
4.36127245e-01 -1.05557930e+00 -6.74580187e-02 -4.15652156e-01
-4.87540781e-01 8.69560301e-01 7.57327735e-01 1.03758596e-01
8.04496050e-01 3.87013108e-01 -9.12531689e-02 4.74864021e-02
-1.18871498e+00 -8.87891278e-02 9.02317464e-03 3.83065194e-01
4.07856047e-01 -7.22983526e-03 -2.33656526e-01 1.10988963e+00
2.30474785e-01 2.33438388e-01 4.40398097e-01 7.71250963e-01
-2.77681500e-01 -1.33201957e+00 -1.53844565e-01 3.87982935e-01
-5.33736229e-01 -2.70200104e-01 -1.99319601e-01 6.65516198e-01
5.44941962e-01 7.88177073e-01 -1.12740345e-01 -3.16421121e-01
5.98949671e-01 4.46394652e-01 7.44326934e-02 -5.05130649e-01
-4.87585932e-01 -5.91993339e-05 -8.34694225e-03 -2.32888177e-01
-3.03391725e-01 -8.06468666e-01 -1.55168581e+00 -3.93454637e-03
-4.48460639e-01 1.67961031e-01 2.80329049e-01 9.18034256e-01
5.59411347e-01 9.47803795e-01 5.82803547e-01 8.89600888e-02
-6.30891800e-01 -9.57312822e-01 -5.47715604e-01 5.67842185e-01
3.66263866e-01 -9.43426669e-01 -5.50692439e-01 1.41566113e-01] | [9.53813362121582, 9.049711227416992] |
9b0410f1-62a0-4955-8516-aeccb8c7d4ee | associative-learning-for-network-embedding | 2208.14376 | null | https://arxiv.org/abs/2208.14376v1 | https://arxiv.org/pdf/2208.14376v1.pdf | Associative Learning for Network Embedding | The network embedding task is to represent the node in the network as a low-dimensional vector while incorporating the topological and structural information. Most existing approaches solve this problem by factorizing a proximity matrix, either directly or implicitly. In this work, we introduce a network embedding method from a new perspective, which leverages Modern Hopfield Networks (MHN) for associative learning. Our network learns associations between the content of each node and that node's neighbors. These associations serve as memories in the MHN. The recurrent dynamics of the network make it possible to recover the masked node, given that node's neighbors. Our proposed method is evaluated on different downstream tasks such as node classification and linkage prediction. The results show competitive performance compared to the common matrix factorization techniques and deep learning based methods. | ['Mohammed J. Zaki', 'Dmitry Krotov', 'Yuchen Liang'] | 2022-08-30 | null | null | null | null | ['network-embedding'] | ['methodology'] | [-1.79717019e-01 3.94788116e-01 -2.07504719e-01 -1.44968361e-01
3.73673499e-01 -3.69100273e-01 5.99454224e-01 1.48712829e-01
-2.64777064e-01 3.32452118e-01 5.36958277e-01 -1.10083684e-01
-4.87871498e-01 -1.10179615e+00 -7.54838347e-01 -7.87802875e-01
-5.38835287e-01 4.47651088e-01 5.97732179e-02 -2.83866465e-01
1.39919445e-01 6.14932835e-01 -9.64371562e-01 2.49452978e-01
4.77197558e-01 5.54900110e-01 1.97746441e-01 4.33620512e-01
-4.68531922e-02 9.05068517e-01 -3.05639356e-01 -3.16590846e-01
5.99698238e-02 -3.08575984e-02 -7.11359262e-01 -6.89301491e-01
2.77621478e-01 -2.49950260e-01 -1.38129234e+00 1.00412595e+00
1.52594417e-01 6.24676235e-02 6.87551022e-01 -1.40346813e+00
-1.26315498e+00 1.03363442e+00 -4.42998171e-01 4.09646124e-01
-2.73794448e-03 -1.82031468e-01 1.34995484e+00 -8.43424499e-01
8.42596829e-01 1.32789779e+00 9.42757428e-01 3.22007895e-01
-1.70635390e+00 -7.30907321e-01 3.34527910e-01 2.97011793e-01
-1.51055241e+00 -2.69404262e-01 9.21335042e-01 -5.49063683e-01
1.14065206e+00 -2.07792372e-01 9.59574580e-01 1.10103869e+00
5.50867677e-01 5.09584546e-01 5.78257263e-01 -1.79673761e-01
2.36938130e-02 -1.64362490e-01 3.15190583e-01 7.75730431e-01
1.95931748e-01 1.56496689e-01 -8.04937840e-01 -3.52650106e-01
9.93672013e-01 6.25555992e-01 -2.27093771e-01 -4.86576945e-01
-1.38443208e+00 8.56874824e-01 1.27981436e+00 5.44049919e-01
-5.06669521e-01 5.40686309e-01 -1.51011916e-02 4.07669127e-01
3.75333250e-01 2.66758502e-01 -5.52415475e-02 5.02475441e-01
-7.74447799e-01 9.56624597e-02 7.53509998e-01 5.25583982e-01
9.13084149e-01 -1.54187858e-01 2.60568447e-02 4.03787225e-01
7.67747045e-01 2.04574823e-01 4.65138912e-01 -7.22074687e-01
2.79544204e-01 8.59398246e-01 -1.99686944e-01 -1.82059979e+00
-5.19219220e-01 -4.62775618e-01 -1.27054858e+00 6.03052676e-02
-9.03908629e-03 6.73663765e-02 -7.86187947e-01 2.11781764e+00
6.86703175e-02 8.18745017e-01 1.39517158e-01 7.64147460e-01
8.40869129e-01 8.32780659e-01 -9.34003145e-02 1.98169082e-01
1.02481723e+00 -1.06287205e+00 -7.39575148e-01 -1.35028288e-01
5.07098258e-01 -2.43562371e-01 3.12577516e-01 -1.45989746e-01
-7.30370045e-01 -3.89230162e-01 -1.23433638e+00 -8.08151439e-02
-5.78343213e-01 -1.58391148e-02 8.46890092e-01 3.18558887e-02
-1.45359349e+00 9.36748803e-01 -1.01013839e+00 -5.07830083e-01
2.93229103e-01 7.38960564e-01 -7.85715759e-01 -5.69688678e-02
-1.48629105e+00 6.68337464e-01 1.75974920e-01 5.42003393e-01
-7.67232239e-01 -8.36133480e-01 -8.39166462e-01 4.86659288e-01
-2.44606107e-01 -7.01024055e-01 5.29412150e-01 -6.53713107e-01
-1.04246318e+00 4.19167042e-01 -2.81263500e-01 -6.69483066e-01
-1.06738091e-01 -1.06189251e-01 -4.46234852e-01 2.60708988e-01
5.38769353e-04 9.56454396e-01 7.30832934e-01 -7.96898127e-01
4.20644501e-04 -4.56250191e-01 5.18217608e-02 6.53287489e-03
-6.72063649e-01 -4.59514946e-01 -1.01696655e-01 -8.43715847e-01
5.90978980e-01 -1.02749491e+00 -2.72036016e-01 3.45541269e-01
-4.32663620e-01 -1.59590900e-01 9.43083405e-01 -6.56571567e-01
1.11263967e+00 -2.20773935e+00 7.34265506e-01 5.77292144e-01
9.18848991e-01 2.31207564e-01 -5.10460496e-01 8.21196198e-01
-4.99998987e-01 5.11099398e-02 4.90797944e-02 -8.89656097e-02
-2.12547421e-01 2.44234443e-01 -4.97694403e-01 4.58252817e-01
2.89295018e-01 1.34048223e+00 -9.56695378e-01 -1.10250548e-01
-1.15835451e-01 1.03867495e+00 -6.01380348e-01 9.46974233e-02
1.39837176e-01 1.27841026e-01 -2.57872671e-01 1.60344869e-01
5.26063383e-01 -4.50867087e-01 6.09002471e-01 -5.83943963e-01
3.07500690e-01 3.23628038e-01 -9.69922721e-01 1.87206054e+00
-2.03673571e-01 8.72172058e-01 -3.10760131e-03 -1.16937792e+00
1.02175117e+00 3.99593383e-01 3.16775262e-01 -4.03254420e-01
-1.68700382e-01 -1.78416029e-01 4.89381969e-01 -1.45387396e-01
1.17828988e-01 1.36125594e-01 3.29997331e-01 7.88975716e-01
4.89588886e-01 9.26600099e-01 -2.02391624e-01 8.83584023e-01
1.62925184e+00 -2.21957847e-01 -6.99698254e-02 -4.59113419e-01
1.07130051e-01 -5.02195835e-01 5.74103653e-01 6.96565092e-01
-5.46421390e-03 1.00598320e-01 7.84652352e-01 -6.27265811e-01
-9.24890518e-01 -1.28297865e+00 1.17184818e-01 7.21943319e-01
5.01296157e-03 -6.57453656e-01 -5.10873199e-01 -4.00575310e-01
4.63144541e-01 -1.12448908e-01 -1.29966676e+00 -5.81996143e-01
-4.48944956e-01 -7.07321644e-01 6.48662865e-01 7.07372069e-01
2.76914299e-01 -9.35318351e-01 1.19429603e-01 2.86881834e-01
-7.90684596e-02 -6.34890020e-01 -1.60894349e-01 2.90659845e-01
-1.11001945e+00 -8.58172596e-01 -3.75806749e-01 -1.08062005e+00
9.73170877e-01 3.46522242e-01 7.87482858e-01 4.39738750e-01
-2.74803370e-01 -6.85182959e-02 -5.69409356e-02 4.66595322e-01
-1.55018996e-02 3.50012183e-01 4.22537893e-01 1.66041166e-01
4.56887960e-01 -1.27292001e+00 -7.86062479e-01 1.90594807e-01
-8.92966449e-01 -5.00923097e-02 7.87464499e-01 1.03398383e+00
2.65387177e-01 9.07483846e-02 6.84622169e-01 -1.01265562e+00
5.60077369e-01 -6.89940393e-01 -1.81516722e-01 2.93073118e-01
-5.61592281e-01 4.32133496e-01 3.72244507e-01 -5.57322621e-01
-5.83136261e-01 8.76984373e-02 2.81522214e-01 -5.76763630e-01
3.87069494e-01 7.13616192e-01 -1.07211091e-01 -2.44138956e-01
4.55603302e-01 7.40928650e-02 2.08583474e-01 -6.10702157e-01
4.96768177e-01 2.38725856e-01 4.06152189e-01 -1.66261584e-01
9.19792652e-01 5.50305247e-01 1.06761828e-01 -4.78974819e-01
-5.18270791e-01 -1.18698046e-01 -1.16966510e+00 4.99150939e-02
5.61962008e-01 -1.06079841e+00 -7.85181940e-01 1.64263457e-01
-1.28412569e+00 -2.40612417e-01 1.25122592e-01 4.36081320e-01
-2.24207006e-02 5.71335554e-02 -1.12443900e+00 -2.01859325e-01
-1.43521100e-01 -6.23323083e-01 4.46648031e-01 1.08860992e-01
-2.27309987e-01 -1.27433848e+00 5.09877563e-01 -1.52254626e-01
4.17692006e-01 -9.49971229e-02 1.18629920e+00 -4.63695139e-01
-7.67792642e-01 -2.77395308e-01 -4.53499138e-01 -2.25763723e-01
-2.35539954e-02 7.32892305e-02 -7.29499400e-01 -4.52136427e-01
-4.80810940e-01 1.34897351e-01 1.40627694e+00 1.51947632e-01
6.34576023e-01 -4.13490593e-01 -9.72358108e-01 7.52331734e-01
1.32215762e+00 -3.28174680e-01 6.85175538e-01 1.11420237e-01
9.44851279e-01 6.28268421e-01 -9.86676142e-02 -2.07140185e-02
2.45105669e-01 4.58131135e-01 3.62886131e-01 5.96518926e-02
-1.69536397e-01 -7.03241348e-01 2.50558883e-01 1.10948682e+00
2.94608623e-02 -1.37423769e-01 -1.05313301e+00 5.52373886e-01
-2.34081173e+00 -1.00599754e+00 1.17537342e-01 1.51631582e+00
5.01868725e-01 5.80163002e-02 -2.45314151e-01 -2.72353925e-02
8.31785083e-01 2.94720590e-01 -6.75667942e-01 -7.89283514e-02
-2.44969636e-01 8.26448053e-02 1.33187860e-01 5.92598915e-01
-8.66065323e-01 9.20305252e-01 6.78886652e+00 1.59197778e-01
-1.07460153e+00 1.46186769e-01 3.83164078e-01 -1.26329243e-01
-4.85814959e-01 1.55711487e-01 -5.94261229e-01 1.60608098e-01
1.04735684e+00 6.62945658e-02 5.81037939e-01 3.54898274e-01
-1.27190113e-01 2.85858691e-01 -1.23646581e+00 7.52407372e-01
5.42369671e-02 -1.78002667e+00 2.31063694e-01 1.76050454e-01
5.34392536e-01 2.22949907e-01 3.45129222e-01 1.26452371e-01
4.86862421e-01 -1.41925299e+00 2.12907270e-01 8.31259847e-01
3.58786076e-01 -6.09738946e-01 6.30479157e-01 1.12239979e-01
-1.13675666e+00 -1.07389987e-01 -6.80427074e-01 -4.95215654e-01
-4.80191782e-02 7.45929897e-01 -5.65731406e-01 1.84107512e-01
8.35802197e-01 1.34622192e+00 -6.24431252e-01 7.50476360e-01
-3.53514493e-01 5.99402308e-01 -2.47725934e-01 1.92109928e-01
2.29013637e-01 -3.70147347e-01 4.37008202e-01 8.79951477e-01
9.10588354e-02 -6.71926420e-03 -1.73928544e-01 1.31783450e+00
-4.70974356e-01 -1.24295570e-01 -9.54878986e-01 -3.01878303e-01
7.43392229e-01 1.54705942e+00 -8.46422553e-01 -6.30158174e-04
-3.77398401e-01 1.12952638e+00 1.13859344e+00 7.39031971e-01
-4.98887926e-01 -5.47372103e-01 8.82862270e-01 1.32688716e-01
4.68837261e-01 -3.11407685e-01 1.91170741e-02 -1.07974076e+00
-7.84307625e-03 -2.18397573e-01 1.12973012e-01 -7.67382145e-01
-1.26761544e+00 6.39542401e-01 -5.73998094e-01 -7.76667237e-01
3.15482207e-02 -6.76937342e-01 -8.04505229e-01 9.14847374e-01
-1.19424820e+00 -1.20353305e+00 2.65745763e-02 4.50841516e-01
-4.20982987e-01 -4.01930869e-01 1.14378679e+00 4.75197405e-01
-7.82928944e-01 4.93023872e-01 4.66980368e-01 6.81668639e-01
5.14182448e-01 -1.06981242e+00 5.76169908e-01 5.76555252e-01
6.38000429e-01 1.28887665e+00 3.22728395e-01 -7.68011212e-01
-1.51084542e+00 -1.10498452e+00 1.06725061e+00 -2.55698502e-01
1.12185609e+00 -7.97846437e-01 -1.06154847e+00 1.14172041e+00
1.34012192e-01 4.21155542e-01 8.16682458e-01 4.63524044e-01
-9.29679513e-01 -2.05680370e-01 -8.57819915e-01 6.37839675e-01
1.31247306e+00 -9.96228218e-01 -5.25610566e-01 1.22179151e-01
9.47823167e-01 2.10233971e-01 -1.08834267e+00 1.24659322e-01
7.35169768e-01 -6.92143559e-01 1.18896520e+00 -1.03290653e+00
5.74204206e-01 -3.66088241e-01 -4.16737720e-02 -1.59567785e+00
-9.79642987e-01 -5.24965703e-01 -4.58729744e-01 1.16163957e+00
5.50108612e-01 -5.75660408e-01 9.48149681e-01 4.50249285e-01
3.36103380e-01 -7.47080147e-01 -8.48607063e-01 -4.11730945e-01
2.89021078e-02 1.72569826e-01 4.91761386e-01 1.19642043e+00
3.62332940e-01 7.80641019e-01 -4.61487114e-01 3.07101369e-01
5.08883357e-01 2.79941946e-01 1.04356088e-01 -1.54622185e+00
-2.89252460e-01 -1.98151261e-01 -9.93468583e-01 -8.88478458e-01
4.78320539e-01 -1.38216615e+00 -4.97473150e-01 -1.31507540e+00
3.41510564e-01 -5.75787425e-01 -7.28663445e-01 6.75459683e-01
2.64772233e-02 4.16287512e-01 1.53531387e-01 5.21569610e-01
-4.42063183e-01 5.69365680e-01 7.19396234e-01 -2.97428042e-01
-9.11871269e-02 -4.90502656e-01 -7.21612990e-01 5.57208240e-01
7.41403937e-01 -6.32301211e-01 -4.23956215e-01 -7.71098375e-01
6.73940361e-01 -4.32368442e-02 5.15080571e-01 -8.93691063e-01
6.85499132e-01 4.98156250e-01 7.20559418e-01 -4.51077163e-01
3.90073955e-01 -8.27034473e-01 2.89124489e-01 6.96174622e-01
-5.52876949e-01 3.32851350e-01 -3.73641662e-02 1.16785109e+00
-2.60059029e-01 5.95726855e-02 2.90219605e-01 1.63373500e-01
-5.12957871e-01 4.05493736e-01 -3.81794393e-01 -5.14617443e-01
8.02971721e-01 2.07138658e-01 -4.45092589e-01 -3.55726600e-01
-1.21072412e+00 2.18626827e-01 2.62684643e-01 5.18570840e-01
9.20026124e-01 -1.74349785e+00 -4.40967947e-01 3.79430413e-01
-1.16722360e-01 -4.34887886e-01 2.43681237e-01 1.00067735e+00
-3.44702482e-01 4.91728634e-01 -4.64497477e-01 -2.55422294e-01
-9.28989410e-01 8.22560012e-01 3.44494045e-01 -2.36327410e-01
-8.36103916e-01 8.89366984e-01 2.58896738e-01 -5.78317046e-01
3.50315720e-01 -7.28368312e-02 -4.00797188e-01 1.27618596e-01
6.23840690e-01 1.29335925e-01 -2.09046766e-01 -7.65759826e-01
-4.46354449e-01 4.27241087e-01 -3.96654785e-01 -3.80179435e-02
1.62326741e+00 -3.13391127e-02 -7.44495392e-01 5.72031498e-01
1.39483464e+00 -4.54638064e-01 -9.38287556e-01 -7.53220916e-01
1.08581521e-01 -2.24130541e-01 2.50079662e-01 -4.20690954e-01
-1.35524869e+00 1.13769853e+00 4.82293367e-01 -4.06351797e-02
4.75543559e-01 -1.31578939e-02 6.61265910e-01 7.71050155e-01
7.34547377e-02 -5.57410359e-01 2.21280530e-01 4.76296663e-01
6.53493762e-01 -9.82174873e-01 -1.00088179e-01 -3.90605628e-01
-1.82184580e-04 1.15053916e+00 5.02701938e-01 -6.67980552e-01
1.33018398e+00 1.94651738e-01 -1.90116405e-01 -5.27715564e-01
-1.11071301e+00 2.54520327e-01 2.53783643e-01 6.15517795e-01
4.43306506e-01 -4.28943373e-02 1.48568153e-01 5.81394315e-01
4.02076654e-02 -2.65245020e-01 2.34876275e-01 6.60984159e-01
-3.30644041e-01 -1.09269428e+00 4.61192317e-02 5.74651897e-01
1.09892609e-02 -4.05347914e-01 -5.90779960e-01 4.00342703e-01
-1.16822436e-01 5.03209174e-01 5.13173461e-01 -7.51605213e-01
-2.45543242e-01 1.68614030e-01 3.14717919e-01 -5.65640330e-01
-4.04913127e-01 -2.41233513e-01 -3.93574357e-01 -6.47374749e-01
-4.59196270e-01 -4.78074133e-01 -1.32445872e+00 -5.79318941e-01
-1.74868852e-01 2.33855397e-01 3.19626749e-01 6.26714349e-01
7.47048497e-01 5.04434049e-01 4.00719821e-01 -7.89604962e-01
-9.80111435e-02 -9.28601861e-01 -7.40689397e-01 3.63370031e-01
4.82476801e-01 -8.41477096e-01 -1.93809107e-01 -2.64437139e-01] | [7.1599345207214355, 6.200343132019043] |
d7c1e9e0-cf18-4884-968c-ef54f65e4385 | chartreader-a-unified-framework-for-chart | 2304.02173 | null | https://arxiv.org/abs/2304.02173v1 | https://arxiv.org/pdf/2304.02173v1.pdf | ChartReader: A Unified Framework for Chart Derendering and Comprehension without Heuristic Rules | Charts are a powerful tool for visually conveying complex data, but their comprehension poses a challenge due to the diverse chart types and intricate components. Existing chart comprehension methods suffer from either heuristic rules or an over-reliance on OCR systems, resulting in suboptimal performance. To address these issues, we present ChartReader, a unified framework that seamlessly integrates chart derendering and comprehension tasks. Our approach includes a transformer-based chart component detection module and an extended pre-trained vision-language model for chart-to-X tasks. By learning the rules of charts automatically from annotated datasets, our approach eliminates the need for manual rule-making, reducing effort and enhancing accuracy.~We also introduce a data variable replacement technique and extend the input and position embeddings of the pre-trained model for cross-task training. We evaluate ChartReader on Chart-to-Table, ChartQA, and Chart-to-Text tasks, demonstrating its superiority over existing methods. Our proposed framework can significantly reduce the manual effort involved in chart analysis, providing a step towards a universal chart understanding model. Moreover, our approach offers opportunities for plug-and-play integration with mainstream LLMs such as T5 and TaPas, extending their capability to chart comprehension tasks. The code is available at https://github.com/zhiqic/ChartReader. | ['Alexander G. Hauptmann', 'Teruko Mitamura', 'Jingdong Sun', 'SiYao Li', 'Qi Dai', 'Zhi-Qi Cheng'] | 2023-04-05 | null | null | null | null | ['optical-character-recognition'] | ['computer-vision'] | [ 1.59157664e-01 -2.45462388e-01 -3.53639238e-02 -3.91445190e-01
-1.06953990e+00 -9.49473619e-01 3.83032769e-01 5.41807294e-01
7.89319053e-02 2.48847321e-01 3.32009405e-01 -8.50943863e-01
2.99111992e-01 -5.20217299e-01 -6.53604627e-01 -2.87748128e-01
2.93055326e-01 1.38298482e-01 7.43184164e-02 -1.31359883e-02
3.34299266e-01 2.97543615e-01 -1.42233539e+00 5.77683508e-01
1.22791362e+00 1.08549213e+00 2.37597808e-01 1.17886567e+00
-7.32157826e-01 1.04672301e+00 -8.17913771e-01 -6.76312208e-01
-6.19029664e-02 -4.29462224e-01 -6.16454661e-01 9.88965929e-02
6.98682725e-01 -5.21332145e-01 -2.27621030e-02 8.55639160e-01
4.28413332e-01 -1.75031602e-01 6.27653420e-01 -1.30364585e+00
-1.17572486e+00 2.42242232e-01 -6.48431897e-01 -2.11690575e-01
9.07687783e-01 3.41111034e-01 1.21691191e+00 -8.35956216e-01
4.01814371e-01 1.24916148e+00 5.22749066e-01 4.10571367e-01
-1.19530416e+00 -5.46854317e-01 3.56501937e-01 2.74716854e-01
-1.04614007e+00 -4.82106298e-01 7.42359400e-01 -6.56809568e-01
1.04625571e+00 5.01173854e-01 5.22972345e-01 1.10186255e+00
-4.26538922e-02 1.19904542e+00 7.88991153e-01 -6.38029277e-01
2.09412038e-01 9.34941545e-02 2.46939287e-01 9.74460900e-01
1.83914974e-01 -3.98903668e-01 -6.01751387e-01 1.50654122e-01
9.41014290e-01 -2.32121628e-02 -4.29916501e-01 -5.83255172e-01
-1.10778987e+00 4.66661930e-01 3.72675836e-01 -1.39650077e-01
5.87473158e-03 1.03121258e-01 7.58436322e-01 3.56709868e-01
1.13803567e-02 4.28912342e-01 -3.38773876e-01 -4.47715014e-01
-6.98566675e-01 1.60396904e-01 8.34456921e-01 1.43029320e+00
6.49650693e-01 -4.41382937e-02 -3.91303301e-01 7.33415902e-01
3.40537190e-01 4.76395160e-01 1.55223429e-01 -5.68498015e-01
9.89492893e-01 1.17509127e+00 -5.63812107e-02 -7.58253515e-01
-3.40135664e-01 -1.55161858e-01 -6.55358374e-01 3.50772649e-01
4.67582375e-01 -3.12966369e-02 -1.17337227e+00 1.13397574e+00
1.09814785e-01 -6.64389193e-01 -1.75431609e-01 7.04753876e-01
8.31319451e-01 5.91668963e-01 1.04455590e-01 2.89056718e-01
1.48098421e+00 -1.19232070e+00 -8.92735839e-01 -1.48751467e-01
8.50284100e-01 -8.17905366e-01 1.87140775e+00 3.70364577e-01
-8.21236789e-01 -5.98363578e-01 -1.08533621e+00 -6.85972512e-01
-6.53444171e-01 5.79622686e-01 5.73750854e-01 7.05016196e-01
-9.65869188e-01 -2.59115286e-02 -7.58967102e-01 -4.29905117e-01
6.31515384e-01 7.10348133e-03 -2.08771184e-01 -1.20130815e-01
-4.53267217e-01 6.32008195e-01 2.17567071e-01 7.53011405e-02
-5.37198365e-01 -7.99627662e-01 -1.17244971e+00 3.40841472e-01
5.54044604e-01 -5.88936210e-01 1.63414979e+00 -6.04919016e-01
-1.26998389e+00 4.43047136e-01 -5.27265072e-01 -1.98853731e-01
7.56356418e-01 -7.11232960e-01 -3.32057267e-01 2.24109679e-01
3.38496529e-02 5.17634094e-01 8.57756138e-01 -1.37002146e+00
-3.86348397e-01 -2.25489199e-01 4.33611423e-02 1.44810483e-01
-1.71455324e-01 -1.53575949e-02 -9.96103287e-01 -6.82674885e-01
-2.60962218e-01 -3.92996252e-01 1.34524360e-01 5.15851438e-01
-7.13083863e-01 -1.79334097e-02 8.61215770e-01 -1.02642977e+00
1.64012945e+00 -2.18364692e+00 -1.29859477e-01 8.61701220e-02
3.57244492e-01 3.97698402e-01 -1.96800560e-01 7.84895420e-01
3.35605368e-02 2.02170998e-01 -4.10374641e-01 -4.10958558e-01
2.10874200e-01 -3.37324329e-02 -3.98593515e-01 -1.53078884e-01
5.13442576e-01 9.84700501e-01 -6.04352534e-01 -5.79558671e-01
6.56755745e-01 4.35763150e-01 -3.36250544e-01 4.26541597e-01
-5.56528866e-01 1.58464015e-01 -3.31699014e-01 1.06156588e+00
6.81010365e-01 -4.08945173e-01 -2.13869102e-02 5.53953201e-02
-1.62667617e-01 3.02017003e-01 -1.06031537e+00 1.89665067e+00
-6.00708723e-01 8.67916465e-01 -7.59600103e-02 -4.02752638e-01
1.10851681e+00 1.79273784e-01 -1.84106395e-01 -1.03740597e+00
-5.90044186e-02 1.74161755e-02 -3.33097309e-01 -7.20431924e-01
6.66986585e-01 5.31617820e-01 -3.17460741e-03 1.61745846e-01
-2.73413897e-01 -2.35397056e-01 3.87156874e-01 4.33609903e-01
1.01887894e+00 4.67648268e-01 5.35089254e-01 2.73596585e-01
5.15794635e-01 3.20231169e-01 -9.71492671e-04 6.48633122e-01
-9.72044393e-02 7.43760347e-01 9.69717264e-01 -3.19601208e-01
-9.30436969e-01 -1.10285962e+00 1.30539328e-01 1.24124062e+00
-1.55089304e-01 -7.41605282e-01 -6.24149382e-01 -7.27277815e-01
1.10040046e-01 1.01162076e+00 -6.23562813e-01 3.18064332e-01
-3.75973225e-01 -1.05551220e-01 4.48019743e-01 9.97695029e-01
5.64713836e-01 -8.62737000e-01 -7.46112108e-01 -1.18941236e-02
1.62979439e-01 -1.04364395e+00 -6.75046265e-01 5.94641156e-02
-7.93930590e-01 -1.20374203e+00 -6.07722819e-01 -8.63097727e-01
7.72575915e-01 2.36691460e-01 1.16164649e+00 1.08272552e-01
-3.03581476e-01 5.17894447e-01 -4.42236364e-01 -7.57977486e-01
-4.92478073e-01 1.43854558e-01 -7.13432431e-01 -3.00235629e-01
3.63646030e-01 -7.94571564e-02 -5.80905616e-01 1.04876079e-01
-9.60009158e-01 4.54167783e-01 6.36500418e-01 4.85070109e-01
3.52077663e-01 -5.13733268e-01 4.54314977e-01 -8.38306665e-01
9.26503360e-01 7.49467090e-02 -8.71417999e-01 6.55290842e-01
-6.91510856e-01 2.61231184e-01 7.49594629e-01 -7.13402852e-02
-1.23559737e+00 1.49992123e-01 -8.92477259e-02 -2.77624965e-01
-2.43182585e-01 3.74577314e-01 -4.02696997e-01 3.11793655e-01
4.43198472e-01 9.04047489e-02 -2.04012945e-01 -6.37970388e-01
7.96659946e-01 5.87737918e-01 7.74127364e-01 -5.06229520e-01
9.58751500e-01 2.25212812e-01 -4.54923749e-01 -6.57006681e-01
-3.93114656e-01 -5.24839342e-01 -6.89494550e-01 -1.80364281e-01
9.98045087e-01 -8.93253922e-01 -8.25790226e-01 2.75615752e-01
-1.28226554e+00 -2.38810271e-01 -9.07854140e-02 3.61792073e-02
-2.73506135e-01 3.38414669e-01 -3.51050317e-01 -9.10116374e-01
-4.41957384e-01 -1.19128931e+00 1.25797379e+00 3.64940077e-01
-3.54627967e-01 -9.99652565e-01 -1.06900506e-01 3.36923927e-01
3.24804842e-01 3.06777686e-01 1.30265987e+00 -4.68250006e-01
-9.18770254e-01 -1.96212888e-01 -8.42732430e-01 2.03100264e-01
2.08278164e-01 3.88367087e-01 -1.06336713e+00 -6.73271297e-03
-6.08760357e-01 -4.97199357e-01 8.35732043e-01 3.98222320e-02
1.32025313e+00 -1.48261473e-01 -1.55487970e-01 7.13201642e-01
1.50502133e+00 4.84504580e-01 4.53663647e-01 4.49241579e-01
1.29661644e+00 4.32326078e-01 3.27104241e-01 4.56870705e-01
8.05845141e-01 3.31715941e-01 3.99693251e-01 -3.25289726e-01
-1.77916646e-01 -7.32770741e-01 2.97265053e-01 7.78358757e-01
2.10632354e-01 -2.55719125e-01 -1.15225065e+00 6.44767880e-01
-1.72273207e+00 -4.43375945e-01 -2.31901973e-01 1.92508531e+00
6.54868066e-01 2.61410978e-02 1.12198360e-01 2.58900583e-01
3.09009612e-01 3.29688460e-01 -6.10494435e-01 -9.75196719e-01
1.96109176e-01 9.16346088e-02 3.25522810e-01 3.65837932e-01
-1.01874197e+00 7.45553732e-01 5.75608540e+00 3.93564910e-01
-9.64499891e-01 -3.91376317e-01 3.25870126e-01 9.33687612e-02
-4.62682873e-01 -5.34781702e-02 -5.15534639e-01 1.63367428e-02
6.04332268e-01 -1.47206023e-01 3.11792284e-01 8.69034171e-01
2.58716762e-01 -2.03603849e-01 -1.52455032e+00 1.15013456e+00
2.96645463e-01 -1.40699852e+00 4.46828961e-01 -4.04376417e-01
3.03867072e-01 -4.39732581e-01 -2.78472956e-02 4.20340121e-01
2.33836904e-01 -1.11713195e+00 8.18149745e-01 3.05414975e-01
1.06125736e+00 -4.92999077e-01 3.62470329e-01 -2.40545243e-01
-1.43386185e+00 1.90732125e-02 -1.62444636e-02 -1.48527309e-01
-3.18931066e-03 2.31937110e-01 -1.03122842e+00 6.61207139e-01
5.01761496e-01 5.88504136e-01 -1.20891953e+00 1.23479366e+00
-3.71547699e-01 5.01260936e-01 1.69902802e-01 -2.28530452e-01
4.64198142e-02 -8.73406380e-02 2.17227459e-01 1.53441954e+00
1.86666772e-01 -5.13238311e-01 6.27080575e-02 8.97847950e-01
-1.55022934e-01 2.50171185e-01 -5.76542735e-01 -2.21687526e-01
3.31270307e-01 8.95211279e-01 -5.68735659e-01 -3.80829513e-01
-8.55586529e-01 1.04404998e+00 3.43818724e-01 6.99563205e-01
-8.33202720e-01 -8.64627123e-01 5.16883194e-01 -1.56654641e-01
4.25188780e-01 -3.70185554e-01 -8.66997838e-01 -1.23188567e+00
5.39730430e-01 -1.11052918e+00 4.18482363e-01 -1.02684438e+00
-8.46137106e-01 4.74500895e-01 -1.10799760e-01 -1.35098064e+00
-1.05646521e-01 -8.66014898e-01 -7.61434734e-01 7.60253251e-01
-1.61424363e+00 -1.16323018e+00 -6.60985112e-01 3.87451053e-01
8.62174988e-01 1.34064257e-01 7.62990177e-01 1.97151050e-01
-7.62306094e-01 7.17372537e-01 4.90213335e-02 5.19515216e-01
8.43828142e-01 -1.69918799e+00 9.73553836e-01 1.01073885e+00
1.26208127e-01 5.72791398e-01 5.52949011e-01 -4.81515080e-01
-1.77128804e+00 -1.01728284e+00 5.25938928e-01 -7.50845969e-01
8.33938360e-01 -8.99052799e-01 -1.26300824e+00 8.19623113e-01
4.12209898e-01 -4.28884804e-01 5.87095439e-01 1.08535737e-01
-7.59580433e-01 -1.59048110e-01 -5.73341608e-01 8.83674979e-01
8.56982052e-01 -7.64658928e-01 -7.53333509e-01 1.03802413e-01
8.59450519e-01 -6.41104043e-01 -6.47578061e-01 -1.88147858e-01
7.15860963e-01 -9.61391270e-01 6.24716997e-01 -2.21416458e-01
7.35480547e-01 -4.50140566e-01 7.62142092e-02 -1.25888956e+00
9.84920412e-02 -7.35967875e-01 -1.43674165e-01 1.38656688e+00
6.91969395e-01 -4.16715652e-01 4.98441517e-01 7.90126383e-01
-2.79291093e-01 -5.16746700e-01 -5.22880673e-01 -4.87891257e-01
-2.57437378e-01 -7.63385653e-01 7.73327708e-01 6.33674562e-01
1.19595379e-01 3.44719589e-01 -6.47015646e-02 2.90188551e-01
3.55764091e-01 1.07994743e-01 1.21581900e+00 -8.69016588e-01
-1.75439969e-01 -5.79679072e-01 -9.60684046e-02 -1.25404954e+00
-4.71080273e-01 -7.61820495e-01 -1.71281219e-01 -2.18725181e+00
-7.89131373e-02 4.55779321e-02 7.52823353e-02 4.94195908e-01
-3.84823799e-01 -3.18378627e-01 6.41939223e-01 8.30867812e-02
-6.25180423e-01 4.71474290e-01 1.48261929e+00 -2.51873285e-01
-5.80936730e-01 -3.45077485e-01 -7.89688468e-01 4.98263717e-01
6.99800134e-01 2.68949829e-02 -6.55873954e-01 -9.53636408e-01
1.37913615e-01 6.05737977e-02 3.48460674e-01 -1.07116830e+00
1.97501928e-01 9.09723192e-02 6.31495833e-01 -9.00406957e-01
2.37200744e-02 -8.36186349e-01 -4.86279041e-01 1.65173516e-01
-4.74807292e-01 6.13987803e-01 6.79746747e-01 6.29976273e-01
-2.66757131e-01 5.49175404e-02 2.12749273e-01 4.46821302e-02
-8.01945865e-01 -2.13366970e-01 -2.25211382e-01 -6.52571991e-02
7.70103872e-01 -3.57494265e-01 -8.14433277e-01 -4.33010101e-01
-3.07513297e-01 6.05908811e-01 5.40577888e-01 6.33472383e-01
8.01396787e-01 -9.46404517e-01 -3.35534304e-01 4.85845387e-01
8.11100066e-01 3.40359300e-01 -2.43165102e-02 4.52725887e-01
-1.02908325e+00 5.58689177e-01 -1.84127107e-01 -6.65993750e-01
-1.30179238e+00 5.39427280e-01 -5.29590286e-02 -1.62952140e-01
-8.43420804e-01 5.55467248e-01 2.87877262e-01 -2.73653090e-01
6.65889084e-01 -1.10095477e+00 -4.24036905e-02 3.96261513e-02
7.24784315e-01 4.61194217e-01 1.15495220e-01 5.63583858e-02
-3.02016854e-01 5.58674037e-01 -4.08260435e-01 -2.29046233e-02
8.69948208e-01 -1.19266704e-01 2.08283290e-01 5.27160287e-01
1.02892530e+00 1.35488898e-01 -1.37625825e+00 1.18218087e-01
2.51309216e-01 -4.15689886e-01 -2.84726471e-01 -1.14481449e+00
-4.73072231e-01 1.18033254e+00 4.66025710e-01 1.36019945e-01
1.32149410e+00 -1.80029690e-01 6.40275359e-01 5.19773066e-01
-9.03341845e-02 -8.91698897e-01 -5.27278446e-02 2.36945614e-01
1.13234651e+00 -1.40930116e+00 3.31125036e-02 -3.95268768e-01
-1.05296707e+00 1.27173519e+00 7.90220857e-01 2.54791379e-01
8.86617042e-03 3.56328577e-01 5.65918207e-01 -1.36426598e-01
-9.56791997e-01 -2.01368436e-01 4.88104999e-01 8.48764360e-01
7.08542287e-01 1.84046790e-01 1.81458950e-01 2.11196095e-01
-2.70567358e-01 1.07998848e-01 6.17374063e-01 1.01255107e+00
-1.58435002e-01 -8.73993933e-01 -5.97382486e-01 6.05216742e-01
8.94443244e-02 -1.40462101e-01 -6.29896224e-01 1.08570707e+00
-4.14068282e-01 9.53436255e-01 1.01083681e-01 -1.62660122e-01
6.10267520e-01 4.26881760e-01 2.28098065e-01 -4.38014477e-01
-5.76540411e-01 9.16855186e-02 1.86605290e-01 -6.12655997e-01
2.77260765e-02 -4.68497455e-01 -1.15015531e+00 -5.71767315e-02
5.52542806e-02 -2.74422467e-01 6.65650010e-01 4.98695165e-01
4.96446699e-01 8.57044578e-01 4.67442274e-02 -2.66940296e-01
-3.14318389e-01 -7.72285581e-01 -1.14616208e-01 3.33334655e-01
7.76318967e-01 -3.06321144e-01 4.10487950e-02 3.41640145e-01] | [11.31597900390625, 2.1517281532287598] |
4fed22e3-aaa2-42d0-9b54-30b4659ec998 | cross-lingual-inference-with-a-chinese-1 | 2203.06264 | null | https://arxiv.org/abs/2203.06264v1 | https://arxiv.org/pdf/2203.06264v1.pdf | Cross-lingual Inference with A Chinese Entailment Graph | Predicate entailment detection is a crucial task for question-answering from text, where previous work has explored unsupervised learning of entailment graphs from typed open relation triples. In this paper, we present the first pipeline for building Chinese entailment graphs, which involves a novel high-recall open relation extraction (ORE) method and the first Chinese fine-grained entity typing dataset under the FIGER type ontology. Through experiments on the Levy-Holt dataset, we verify the strength of our Chinese entailment graph, and reveal the cross-lingual complementarity: on the parallel Levy-Holt dataset, an ensemble of Chinese and English entailment graphs outperforms both monolingual graphs, and raises unsupervised SOTA by 4.7 AUC points. | ['Mark Steedman', 'Liane Guillou', 'Mohammad Javad Hosseini', 'Sabine Weber', 'Tianyi Li'] | 2022-03-11 | null | https://aclanthology.org/2022.findings-acl.96 | https://aclanthology.org/2022.findings-acl.96.pdf | findings-acl-2022-5 | ['entity-typing'] | ['natural-language-processing'] | [ 1.13787934e-01 7.42083788e-01 -2.82577783e-01 -3.63099217e-01
-1.12659955e+00 -8.25882077e-01 6.10898018e-01 6.07623100e-01
-3.37956429e-01 9.17356730e-01 4.00576621e-01 -8.63639951e-01
-2.25597948e-01 -8.29470754e-01 -9.99532223e-01 2.44335175e-01
-1.12604193e-01 8.73978257e-01 4.65577245e-01 -5.28502941e-01
-5.35277613e-02 -2.34864131e-01 -1.33307302e+00 9.89284635e-01
1.25876606e+00 8.59599650e-01 -3.73017430e-01 6.36345744e-01
-2.15193853e-01 1.22126245e+00 -2.88882881e-01 -1.30519724e+00
-1.02541849e-01 -3.38951528e-01 -1.57205617e+00 -5.21956682e-01
6.17101789e-01 8.29442516e-02 -1.85033575e-01 7.09875166e-01
5.09175994e-02 -3.22526067e-01 5.21718204e-01 -1.03031361e+00
-8.22559237e-01 1.25928760e+00 -2.46104766e-02 8.65938738e-02
9.76266503e-01 -3.35927665e-01 1.95035410e+00 -7.41719842e-01
1.02916479e+00 9.49528337e-01 1.00386548e+00 1.64954320e-01
-1.02860522e+00 -1.68633416e-01 -1.63512483e-01 4.60180372e-01
-1.28875959e+00 -2.26901859e-01 -1.00203209e-01 -9.83928964e-02
1.96974635e+00 5.94873726e-01 6.44668579e-01 5.86493850e-01
7.87524357e-02 7.06579030e-01 1.17234004e+00 -8.91323507e-01
-2.81458199e-01 -2.08488807e-01 5.18092334e-01 1.02242589e+00
7.10190415e-01 -3.85229588e-01 -4.80435342e-01 -2.14860916e-01
-1.99518383e-01 -4.32176560e-01 -8.75391066e-03 1.46403283e-01
-1.00512958e+00 6.62547648e-01 2.83517778e-01 4.23256218e-01
-5.45276292e-02 -4.46263142e-02 5.29690802e-01 7.81052351e-01
7.20069170e-01 5.30429482e-01 -9.04017866e-01 -3.27410474e-02
-5.59668541e-01 3.93549025e-01 1.55612242e+00 1.40136456e+00
8.70024621e-01 -1.04040051e+00 9.79039352e-03 5.02647519e-01
1.95758849e-01 1.71592921e-01 2.34995529e-01 -4.16609108e-01
8.18492174e-01 1.20876563e+00 -1.91411167e-01 -5.16187668e-01
-6.29702270e-01 5.62438592e-02 -6.78010657e-02 -7.31777787e-01
7.03712463e-01 -2.95183390e-01 -3.69459718e-01 1.22680366e+00
6.05804920e-01 -2.09660381e-01 5.52801490e-01 4.54868823e-01
1.19156301e+00 1.00703694e-01 -1.69550832e-02 3.66731104e-03
2.08564639e+00 -7.71163106e-01 -7.05102801e-01 -1.75584927e-01
1.40641570e+00 -7.37861335e-01 7.36274123e-01 -1.15903273e-01
-7.71513104e-01 8.83570407e-03 -1.08058286e+00 -6.73109233e-01
-5.95882416e-01 -1.48338616e-01 8.91021132e-01 7.35059977e-01
-4.54184562e-01 4.71776813e-01 -5.69976628e-01 -6.97644830e-01
2.92011887e-01 1.47927657e-01 -6.85014367e-01 -4.72773582e-01
-1.57285154e+00 1.22625613e+00 8.72891188e-01 2.40302179e-02
-4.62180264e-02 -8.72135878e-01 -1.18785977e+00 -8.72797072e-02
8.99896443e-01 -4.84043211e-01 1.11645055e+00 -3.14206719e-01
-9.43345547e-01 1.35437584e+00 -2.45271146e-01 -8.43568146e-01
1.90726534e-01 -6.09061182e-01 -6.23649299e-01 1.38706341e-01
2.88896829e-01 -1.79835856e-02 9.61044151e-03 -6.73261106e-01
-6.91646159e-01 -5.40532589e-01 3.50849479e-01 1.53478473e-01
8.67074281e-02 2.94245988e-01 -1.45655170e-01 -1.22490972e-02
2.01341808e-01 -9.22701776e-01 2.02252433e-01 -8.02713871e-01
-7.71859050e-01 -7.18425274e-01 5.60762286e-01 -8.86344314e-01
1.51196599e+00 -1.67046225e+00 -2.62330770e-01 3.45927775e-01
4.60451692e-01 -1.70565583e-02 3.29913408e-01 8.55176747e-01
-9.78938118e-02 3.24904591e-01 -3.17354977e-01 1.10870577e-01
4.44837481e-01 5.88136852e-01 1.34357931e-02 3.07416707e-01
6.91782713e-01 1.68744254e+00 -9.01189208e-01 -7.52690077e-01
-5.76884270e-01 -2.07915455e-01 -7.56874919e-01 7.65473247e-02
-7.51532972e-01 -1.40114650e-01 -9.39251631e-02 7.25743711e-01
4.96106446e-01 -3.07336092e-01 1.10821033e+00 -3.67554158e-01
-1.83095500e-01 8.44948769e-01 -1.01179409e+00 1.63555598e+00
-2.23366305e-01 3.24404389e-01 -4.77919251e-01 -6.41375899e-01
5.73830783e-01 4.21728462e-01 1.97869942e-01 -6.83524847e-01
-3.95957790e-02 7.03224778e-01 5.42337000e-02 -1.07258964e+00
7.53651142e-01 -1.49790823e-01 -5.24780810e-01 5.72174370e-01
3.56010020e-01 -9.64294150e-02 6.39415324e-01 3.85005414e-01
1.26239383e+00 5.31581044e-01 6.30621552e-01 -4.55978662e-01
4.15961891e-01 4.97713447e-01 4.55685169e-01 4.87766922e-01
2.40500808e-01 3.43179405e-01 1.08299136e+00 -6.23108268e-01
-7.91459978e-01 -8.14477265e-01 -5.05252302e-01 9.02756155e-01
-2.06304833e-01 -1.06404281e+00 -7.45248735e-01 -1.37513316e+00
3.16792987e-02 4.80099797e-01 -6.78203821e-01 3.18973869e-01
-7.95328140e-01 -7.88625956e-01 1.25285923e+00 2.02451453e-01
2.67913878e-01 -5.74529767e-01 -3.40097517e-01 -9.74243972e-03
-7.94145405e-01 -1.72794449e+00 -3.74630481e-01 3.12537849e-01
-3.53401423e-01 -1.71368420e+00 8.11244026e-02 -1.09243226e+00
3.03747684e-01 -6.91419721e-01 1.71376944e+00 4.14877564e-01
-1.18732609e-01 3.45731270e-03 -6.07622921e-01 -3.80307227e-01
-5.02453804e-01 5.60177445e-01 -2.40352422e-01 -1.89147621e-01
1.01833308e+00 -1.57789052e-01 -2.15867609e-01 -2.94752084e-02
-7.66484141e-01 -1.05377205e-01 4.30529028e-01 8.64034414e-01
5.30002177e-01 -3.23045671e-01 5.14422297e-01 -1.77116024e+00
6.35013402e-01 -5.05586624e-01 -4.39521998e-01 6.32017314e-01
-6.90932453e-01 4.38070655e-01 1.96223542e-01 2.62333781e-01
-1.21339190e+00 -2.47383371e-01 -5.74470580e-01 5.54629683e-01
-1.95127308e-01 9.52120900e-01 -2.45378241e-01 2.71766722e-01
8.84047031e-01 -2.96304435e-01 -1.79645777e-01 -1.84440136e-01
8.51275802e-01 6.38027489e-01 4.94534403e-01 -8.83734822e-01
4.30037856e-01 2.73325682e-01 1.30471945e-01 -8.25913429e-01
-1.19840395e+00 -5.11219203e-01 -9.85610783e-01 4.31998700e-01
1.09463263e+00 -9.94264781e-01 -8.92211080e-01 3.25352177e-02
-1.21502137e+00 -2.94822663e-01 -4.47213262e-01 3.17328334e-01
-8.31011534e-02 6.66626155e-01 -1.10266912e+00 -4.67225254e-01
-3.98330241e-01 -6.57431662e-01 9.73195970e-01 -2.56024241e-01
-7.82601714e-01 -1.14490163e+00 3.48322958e-01 5.49749792e-01
-2.45606780e-01 4.16684955e-01 1.13224626e+00 -1.10669589e+00
-3.91126096e-01 -3.80589843e-01 -3.72288525e-01 -9.94279981e-02
-3.24618995e-01 -2.71069825e-01 -7.85031378e-01 2.42279649e-01
-4.06362832e-01 -7.34102190e-01 6.76870167e-01 -2.43157461e-01
1.32692516e-01 -4.54102814e-01 -2.15355158e-01 3.16625804e-01
1.48828888e+00 -5.72864354e-01 8.19858432e-01 3.39628130e-01
8.47783089e-01 9.18356836e-01 4.00103092e-01 -8.78805220e-02
1.10487282e+00 2.14736313e-01 4.79676127e-02 1.63061574e-01
3.09313443e-02 -5.41731358e-01 5.34534417e-02 9.40226078e-01
-2.99416006e-01 1.95866093e-01 -8.14112544e-01 5.16785204e-01
-1.85986900e+00 -8.90892863e-01 -1.03421879e+00 1.75422168e+00
1.17583656e+00 1.24965355e-01 3.70159335e-02 1.72035873e-01
1.72911882e-01 -3.13811570e-01 1.41686559e-01 -4.42403734e-01
-4.54062700e-01 6.53122902e-01 3.57166499e-01 7.20997810e-01
-9.23838317e-01 1.13716578e+00 5.99864531e+00 5.80899060e-01
-1.42145708e-01 3.52675170e-01 3.98847796e-02 3.60382855e-01
-6.20380819e-01 5.84842920e-01 -8.22879314e-01 3.01548727e-02
1.08336878e+00 5.35667650e-02 1.06497817e-01 4.11400259e-01
-6.86215341e-01 -2.96883285e-01 -1.36747777e+00 3.13468635e-01
9.85984132e-02 -1.48280537e+00 -4.18468148e-01 -3.97879966e-02
7.29636550e-01 3.54169071e-01 -7.19315410e-01 6.00961089e-01
5.62397778e-01 -8.09794545e-01 5.95183372e-01 1.91670001e-01
9.21093941e-01 -6.34465933e-01 1.01995194e+00 1.00845635e-01
-1.39812732e+00 1.34612784e-01 -1.11987382e-01 -3.51734877e-01
2.18987897e-01 6.84697151e-01 -7.75593758e-01 1.17292047e+00
6.39732182e-01 6.86194956e-01 -8.28966081e-01 4.14174855e-01
-7.57138371e-01 5.78408599e-01 -4.13959414e-01 -3.18887621e-01
3.23683247e-02 -1.00409985e-01 2.34820858e-01 1.54772973e+00
-7.67765567e-02 2.54802346e-01 -7.46526569e-03 6.08445287e-01
-5.51329434e-01 3.68885309e-01 -6.24136984e-01 -3.17229405e-02
2.48947740e-01 1.28963625e+00 -5.43235183e-01 -5.23381591e-01
-8.85429382e-01 9.05928016e-01 1.02258873e+00 1.77379940e-02
-5.98821640e-01 -5.97018600e-01 3.25158775e-01 -1.78897213e-02
3.64289045e-01 8.92910585e-02 -2.65705019e-01 -1.57245827e+00
3.01836699e-01 -9.44434404e-01 1.14091897e+00 -2.85799772e-01
-1.39215505e+00 6.28989577e-01 -1.93909220e-02 -6.88471735e-01
-1.94594398e-01 -6.66540802e-01 -3.68735552e-01 7.28842497e-01
-1.50652778e+00 -1.64817870e+00 1.25328347e-01 5.83516121e-01
-5.73991500e-02 3.73647034e-01 1.17153919e+00 4.76810545e-01
-5.85308075e-01 7.98791468e-01 -4.01252806e-01 3.99118125e-01
5.10764003e-01 -1.52666295e+00 7.29106188e-01 9.63279724e-01
5.41836023e-01 9.88545418e-01 4.56162542e-01 -8.89264405e-01
-1.68556213e+00 -9.31399167e-01 1.88669777e+00 -1.24018466e+00
1.07500291e+00 -5.39806128e-01 -9.56228316e-01 1.29189873e+00
5.39858341e-01 7.62089761e-03 1.02675939e+00 8.75070333e-01
-7.65801191e-01 4.04478371e-01 -9.95202243e-01 3.08401406e-01
1.29064715e+00 -9.89778697e-01 -1.07912445e+00 4.27956879e-01
9.70788062e-01 -6.72615886e-01 -1.57209229e+00 4.11784828e-01
7.02518165e-01 -7.96532273e-01 4.61983204e-01 -9.76974607e-01
6.16373539e-01 -2.46170461e-01 -3.28968048e-01 -9.67368245e-01
8.00849870e-02 -5.63867211e-01 -3.95890921e-01 1.14374304e+00
1.15474737e+00 -8.65140736e-01 5.94606221e-01 6.22062147e-01
-2.95211285e-01 -9.26351607e-01 -9.06123698e-01 -4.81628001e-01
2.41170838e-01 -8.35739017e-01 7.34554589e-01 9.55399990e-01
1.05387998e+00 1.01354945e+00 4.85110693e-02 1.89879313e-01
5.27385235e-01 3.91171366e-01 5.56666255e-01 -1.07212269e+00
-4.99722809e-01 -3.32319140e-02 -2.27267399e-01 -8.09153974e-01
2.65033752e-01 -1.51870978e+00 -1.46254465e-01 -1.76390493e+00
3.19938660e-01 -3.99164468e-01 3.98609281e-01 3.65729362e-01
-4.57982153e-01 3.76051456e-01 -1.74136788e-01 4.27837446e-02
-1.09380794e+00 -7.09558278e-02 8.64977837e-01 8.91461968e-02
3.01072419e-01 -4.22504634e-01 -8.47315073e-01 6.59036815e-01
3.97372544e-01 -5.03467739e-01 -4.58786590e-03 -3.92863750e-01
1.03986454e+00 -2.98842907e-01 5.65596260e-02 -3.22513521e-01
1.32113039e-01 3.22211862e-01 -6.77303225e-03 -4.89640474e-01
-3.34646583e-01 -5.38503647e-01 -9.55220610e-02 4.57385242e-01
-1.34740293e-01 1.72682300e-01 5.83998337e-02 5.21628797e-01
-1.00851506e-01 -2.74618119e-01 6.39903396e-02 -2.41511092e-01
-7.42649436e-01 7.82485753e-02 -1.59938589e-01 1.01844060e+00
7.48073339e-01 2.78935969e-01 -6.93902314e-01 1.38133451e-01
-5.99202931e-01 3.69891733e-01 1.92695409e-01 1.77722245e-01
3.06706160e-01 -1.07978094e+00 -9.72017229e-01 1.25369951e-01
5.91926396e-01 -2.92439070e-02 1.90957841e-02 1.23096585e+00
-7.95759678e-01 5.10362983e-01 3.86102021e-01 -3.76183271e-01
-1.25305057e+00 4.76356506e-01 1.41535580e-01 -8.62877965e-01
-5.23295164e-01 6.81928456e-01 -4.51314062e-01 -9.81844306e-01
-3.94817621e-01 -7.37798095e-01 -2.56842464e-01 1.36908377e-02
3.55401099e-01 2.92210042e-01 6.07798517e-01 -6.20463014e-01
-4.69082087e-01 1.70393586e-01 3.11680813e-03 1.09502628e-01
1.07237840e+00 -7.44761154e-02 -7.61779904e-01 3.15480411e-01
1.30363607e+00 2.95205414e-01 -2.98807740e-01 -2.75529385e-01
7.97830880e-01 -6.30132258e-02 -7.09376216e-01 -5.07200062e-01
-1.83465645e-01 2.39808157e-01 -2.78268486e-01 6.68621540e-01
4.19410646e-01 5.90635777e-01 9.22953844e-01 5.65031409e-01
3.89426291e-01 -1.00899613e+00 -3.79044563e-01 1.10309827e+00
4.89513189e-01 -1.31517696e+00 1.28461316e-01 -7.59421170e-01
-6.28822505e-01 7.98141479e-01 3.94297838e-01 1.07122205e-01
6.43365443e-01 5.39306879e-01 -4.34849933e-02 -8.79764020e-01
-6.79956853e-01 -7.86947787e-01 4.28255767e-01 4.36751187e-01
8.49058747e-01 2.74845123e-01 -6.93554699e-01 9.07064676e-01
-6.20615184e-01 -2.41528645e-01 1.42180189e-01 1.07702935e+00
3.16133574e-02 -1.31626356e+00 -7.42171034e-02 6.99011862e-01
-9.49762523e-01 -5.83453536e-01 -5.05707383e-01 9.92700517e-01
3.38126481e-01 1.15595794e+00 -1.18973278e-01 -3.34520668e-01
6.98171854e-01 4.06636268e-01 1.01927114e+00 -7.61566460e-01
-8.35428536e-01 -4.25599545e-01 1.15985692e+00 -3.50782841e-01
-4.18378919e-01 -7.94706106e-01 -1.42368042e+00 -4.32241529e-01
-7.27563620e-01 4.55804676e-01 1.95923378e-03 1.48819816e+00
2.42546111e-01 4.76835608e-01 -9.57722440e-02 1.32946089e-01
-7.53255263e-02 -1.13837087e+00 -4.44237262e-01 8.63716722e-01
-1.65464655e-01 -1.07915305e-01 8.29323195e-03 6.49336129e-02] | [9.65932846069336, 8.675687789916992] |
4ec25850-dc54-4d2f-9249-dfd395c963fa | monoscene-monocular-3d-semantic-scene | 2112.00726 | null | https://arxiv.org/abs/2112.00726v2 | https://arxiv.org/pdf/2112.00726v2.pdf | MonoScene: Monocular 3D Semantic Scene Completion | MonoScene proposes a 3D Semantic Scene Completion (SSC) framework, where the dense geometry and semantics of a scene are inferred from a single monocular RGB image. Different from the SSC literature, relying on 2.5 or 3D input, we solve the complex problem of 2D to 3D scene reconstruction while jointly inferring its semantics. Our framework relies on successive 2D and 3D UNets bridged by a novel 2D-3D features projection inspiring from optics and introduces a 3D context relation prior to enforce spatio-semantic consistency. Along with architectural contributions, we introduce novel global scene and local frustums losses. Experiments show we outperform the literature on all metrics and datasets while hallucinating plausible scenery even beyond the camera field of view. Our code and trained models are available at https://github.com/cv-rits/MonoScene. | ['Raoul de Charette', 'Anh-Quan Cao'] | 2021-12-01 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Cao_MonoScene_Monocular_3D_Semantic_Scene_Completion_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Cao_MonoScene_Monocular_3D_Semantic_Scene_Completion_CVPR_2022_paper.pdf | cvpr-2022-1 | ['3d-semantic-scene-completion', 'single-view-3d-reconstruction', '3d-semantic-scene-completion-from-a-single', '3d-scene-reconstruction'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 3.76984775e-01 2.20433682e-01 2.60676622e-01 -5.30877113e-01
-5.26928663e-01 -7.39499390e-01 6.77811027e-01 -2.43179649e-01
-2.95377746e-02 3.21493506e-01 3.50082815e-01 1.04218699e-01
-1.08615167e-01 -6.10451877e-01 -9.11092281e-01 -4.31474328e-01
4.74223226e-01 2.53125936e-01 1.97576806e-01 -1.53832197e-01
2.93188483e-01 6.15472317e-01 -1.50806463e+00 2.18757167e-01
4.29390073e-01 1.07239723e+00 6.92006826e-01 6.93906248e-01
-6.25714893e-03 7.99746335e-01 1.27937526e-01 -2.31489718e-01
6.63919926e-01 -2.22758636e-01 -7.67042458e-01 5.30694902e-01
6.51841819e-01 -4.66953367e-01 -7.70910025e-01 1.03551245e+00
1.84292212e-01 1.17221653e-01 9.65615809e-02 -1.17021441e+00
-7.00396419e-01 -3.13093752e-01 -3.91983539e-01 -2.70226032e-01
9.59779739e-01 6.96578622e-02 9.66078639e-01 -1.39189303e+00
7.79924214e-01 1.20807898e+00 5.40801525e-01 2.71565735e-01
-1.18826294e+00 -1.38491154e-01 2.36303523e-01 1.07544288e-01
-1.57166040e+00 -4.65118080e-01 1.02572453e+00 -2.41967037e-01
8.40458691e-01 2.14871541e-01 6.81829512e-01 1.17211461e+00
-2.89624095e-01 6.26477242e-01 1.23406804e+00 -3.79772097e-01
3.31217498e-01 6.68562157e-03 -1.13526665e-01 5.73951960e-01
1.52531907e-01 8.16521645e-02 -7.80751705e-01 9.96777043e-02
1.16488874e+00 3.29694748e-01 -4.13902164e-01 -9.23882067e-01
-1.09913838e+00 5.86913645e-01 6.17415488e-01 -2.20703959e-01
-1.57708243e-01 1.72929361e-01 -2.41698265e-01 -1.14850320e-01
4.95911568e-01 3.81522626e-01 -5.06085753e-01 1.34103268e-01
-6.39463246e-01 1.11921109e-01 3.70194107e-01 1.53260112e+00
9.70799685e-01 -4.77034412e-02 4.44024175e-01 5.89916766e-01
2.99946606e-01 7.01761484e-01 -2.57239193e-01 -1.64221656e+00
2.36343503e-01 5.49594939e-01 3.20311129e-01 -9.47141707e-01
-3.79103243e-01 -3.72082323e-01 -4.91519004e-01 1.60322681e-01
8.14190209e-02 4.17206734e-01 -8.58848095e-01 1.31141460e+00
5.14327765e-01 4.95021045e-01 -1.03529759e-01 1.20690680e+00
9.19035435e-01 3.39651823e-01 -4.88707960e-01 3.40448260e-01
1.09706080e+00 -9.90364432e-01 -4.18654978e-01 -3.79859716e-01
1.04318455e-01 -8.13049972e-01 1.24667227e+00 5.13696134e-01
-1.21696520e+00 -3.80077004e-01 -8.07230294e-01 -8.82945001e-01
-2.34248534e-01 -9.16806608e-02 6.33803248e-01 2.10314661e-01
-1.23772359e+00 2.19569936e-01 -7.47574031e-01 -5.04997909e-01
4.89431083e-01 -1.27765298e-01 -6.47349715e-01 -6.07917190e-01
-6.23138487e-01 8.05393636e-01 2.72824436e-01 1.01440430e-01
-1.08872664e+00 -8.45974028e-01 -1.29687917e+00 -1.85055420e-01
5.25842190e-01 -1.11340928e+00 1.14768195e+00 -4.92970288e-01
-1.42544961e+00 1.12519276e+00 -3.60074788e-01 -1.16449282e-01
5.32247484e-01 -5.00623345e-01 5.56454947e-03 5.29491603e-01
1.51325032e-01 5.49642920e-01 6.17132604e-01 -1.79131949e+00
-1.61400169e-01 -4.90396291e-01 5.19768298e-01 6.59807324e-01
3.99649709e-01 -3.26763451e-01 -6.98767662e-01 -2.90983826e-01
7.32654870e-01 -7.31570542e-01 -2.64352888e-01 4.83003885e-01
-6.80357039e-01 5.49495459e-01 7.42683589e-01 -6.26426935e-01
4.61447388e-01 -2.16018987e+00 2.91569263e-01 6.40602829e-03
3.70593637e-01 -2.71923780e-01 -1.84456557e-01 3.93934816e-01
-3.54613513e-02 -2.61967480e-01 -5.17552435e-01 -1.03980315e+00
9.61636230e-02 2.83612520e-01 -5.30610561e-01 7.70663142e-01
9.57916677e-02 9.21459258e-01 -1.05167150e+00 -5.52279502e-02
8.27382088e-01 7.64118791e-01 -8.01942229e-01 3.53596389e-01
-2.59338707e-01 8.65896344e-01 -3.78399789e-01 7.38205731e-01
1.12617064e+00 -5.67909181e-01 -6.08756579e-02 -2.54395247e-01
-1.06007569e-01 5.16957104e-01 -1.12774730e+00 2.71059203e+00
-5.24879098e-01 4.07176703e-01 2.24801153e-01 -7.82581747e-01
7.07781017e-01 3.94048020e-02 4.01284486e-01 -6.66559279e-01
-7.32596740e-02 2.55584717e-04 -9.17517185e-01 -3.14309925e-01
5.96765339e-01 -1.67995766e-01 -2.09967024e-03 2.46548112e-02
1.13418266e-01 -1.01406443e+00 -5.52644908e-01 4.68940973e-01
9.43345189e-01 7.80764580e-01 3.06610227e-01 -2.21567992e-02
2.36669108e-01 9.69096497e-02 4.84367818e-01 5.40654361e-01
6.91479966e-02 1.35195625e+00 2.16289565e-01 -2.82764792e-01
-1.20753276e+00 -1.63258469e+00 -2.03265324e-01 2.78061569e-01
8.48569274e-01 -5.69628537e-01 -3.93488973e-01 -2.55417287e-01
-1.17163770e-01 9.59628224e-01 -4.79910940e-01 1.91084966e-01
-1.42049059e-01 -9.80006754e-02 1.38124079e-01 2.15177074e-01
6.19676054e-01 -5.69904566e-01 -8.26168716e-01 -3.55508149e-01
-3.74148101e-01 -1.72771215e+00 -2.46724784e-01 -1.23927683e-01
-7.45872140e-01 -1.20137048e+00 -3.78192186e-01 -3.21518809e-01
6.27296984e-01 9.34548438e-01 1.20049953e+00 -1.56351402e-01
-4.96840507e-01 9.99189794e-01 -3.99820924e-01 -1.25077963e-01
1.36765987e-01 -5.18798113e-01 5.71651757e-02 7.26098055e-03
-2.10688468e-02 -9.97106910e-01 -7.83717513e-01 1.01252705e-01
-7.97410965e-01 7.57783473e-01 1.44705117e-01 2.68535584e-01
8.58827770e-01 -1.55476362e-01 -2.11632729e-01 -6.92591369e-01
-3.98324430e-01 -6.66267157e-01 -6.85439825e-01 -1.96032897e-02
-1.98774457e-01 -3.10148597e-01 4.07249331e-01 3.60976100e-01
-1.41100407e+00 3.47847641e-01 1.46657318e-01 -1.01655948e+00
-5.44663250e-01 -1.19929381e-01 -4.08097386e-01 4.11877520e-02
4.52735305e-01 3.85337800e-01 -2.38529772e-01 -7.49159694e-01
6.41055942e-01 9.70753357e-02 5.94042599e-01 -4.27064478e-01
9.51660752e-01 1.29819524e+00 1.76105276e-01 -8.29603434e-01
-1.38207018e+00 -7.10496426e-01 -1.01254177e+00 -1.20631680e-01
9.71578956e-01 -1.42573786e+00 -4.96152461e-01 2.42830917e-01
-1.40136218e+00 -3.52430791e-01 -5.24604261e-01 4.79500979e-01
-9.71760392e-01 6.48451269e-01 -2.89745986e-01 -7.11630225e-01
2.60516942e-01 -8.28242064e-01 1.75882626e+00 -8.14265460e-02
-2.62661241e-02 -9.14647162e-01 -1.32913157e-01 6.12515152e-01
5.55761270e-02 5.27055740e-01 3.97226304e-01 3.54892999e-01
-1.17075324e+00 1.54184222e-01 -5.05564153e-01 2.73878008e-01
1.16249420e-01 -3.96952301e-01 -1.45571911e+00 2.41044909e-02
2.88785040e-01 -3.45941275e-01 7.24980175e-01 3.76948118e-01
1.22684622e+00 -5.24845533e-02 6.42400309e-02 1.37256348e+00
1.79105616e+00 -3.15517515e-01 6.79506719e-01 1.38781369e-01
1.00835621e+00 7.40051329e-01 4.83823717e-01 6.64314508e-01
8.15304101e-01 6.67797208e-01 1.03737080e+00 -1.01839013e-01
-6.47076964e-01 -6.53339982e-01 2.28207842e-01 5.53710103e-01
-9.41665322e-02 -2.04039589e-02 -8.36648762e-01 4.15511578e-01
-1.75290155e+00 -7.78131187e-01 -3.58383477e-01 2.16528702e+00
3.73391956e-01 -3.06386739e-01 -3.03521633e-01 -1.07298180e-01
3.47954035e-01 2.84398258e-01 -7.25894690e-01 3.88601646e-02
-6.73243463e-01 1.70800000e-01 4.93207365e-01 9.98869836e-01
-7.71851242e-01 1.11526370e+00 5.69804096e+00 4.30669814e-01
-7.54945457e-01 2.64281392e-01 3.53595078e-01 -4.56356496e-01
-7.98612654e-01 4.11324710e-01 -4.73805130e-01 1.72048807e-02
2.51938999e-01 1.78542167e-01 7.81918883e-01 6.40874207e-01
2.42133588e-01 -4.42347139e-01 -1.06851351e+00 1.42788565e+00
3.20976168e-01 -1.42923760e+00 -1.26216998e-02 7.03646392e-02
9.28865135e-01 3.58549595e-01 6.20401688e-02 -3.35004866e-01
3.77259851e-01 -9.48885620e-01 1.20295477e+00 8.06806564e-01
1.07132339e+00 -1.60728112e-01 -2.26455461e-02 3.55498672e-01
-1.13631022e+00 1.00793265e-01 -3.47891301e-01 -2.44668573e-01
3.52937847e-01 7.21666873e-01 -6.01454735e-01 9.54618335e-01
8.74196231e-01 1.30549109e+00 -4.79562312e-01 8.11344981e-01
-4.78960305e-01 3.83234918e-02 -4.94553238e-01 6.62038147e-01
2.02370845e-02 -4.51300710e-01 7.18515635e-01 8.22011650e-01
4.49135423e-01 4.20697272e-01 -7.70240203e-02 1.32982159e+00
6.00905456e-02 -3.91671270e-01 -9.27285850e-01 5.92809856e-01
4.35426235e-01 1.12814736e+00 -5.49769342e-01 -5.87477721e-02
-4.65409070e-01 1.56664455e+00 2.61894643e-01 7.57201314e-01
-7.98826694e-01 2.11111289e-02 7.67436564e-01 2.32957765e-01
2.58428991e-01 -6.42215490e-01 -7.75940597e-01 -1.68802965e+00
3.80257875e-01 -1.70232520e-01 7.18107820e-02 -1.65217400e+00
-1.13447595e+00 3.12816948e-01 1.44005537e-01 -1.33668542e+00
2.72093922e-01 -7.32367158e-01 -3.39008629e-01 8.56967092e-01
-1.86593223e+00 -1.32309556e+00 -7.21697509e-01 1.00542092e+00
5.23554504e-01 4.42940891e-01 7.94646859e-01 -1.04642406e-01
-2.36678198e-02 -1.71108395e-01 -1.60582379e-01 -3.16689759e-01
3.82016867e-01 -1.10862219e+00 5.22105336e-01 9.68729913e-01
1.56823575e-01 5.54817557e-01 5.85293889e-01 -3.83525342e-01
-1.78648734e+00 -1.11583340e+00 7.44285047e-01 -1.06583202e+00
3.01860839e-01 -7.48085141e-01 -4.59654331e-01 9.36985850e-01
-5.02303615e-02 3.88165087e-01 3.51004332e-01 -2.23010704e-01
-7.28074610e-01 1.88986570e-01 -1.18778813e+00 6.42152548e-01
1.94881248e+00 -9.81185317e-01 -4.23896700e-01 4.39451873e-01
1.03758669e+00 -6.45954013e-01 -6.78677917e-01 3.13104212e-01
3.31023157e-01 -1.37298155e+00 1.41079748e+00 -1.98747098e-01
6.10984147e-01 -6.05776310e-01 -9.72857714e-01 -9.61698771e-01
-2.95596272e-02 -6.64674580e-01 -8.70265737e-02 4.55060869e-01
2.19892543e-02 -4.96704012e-01 6.24992490e-01 6.80553436e-01
-5.19939244e-01 -3.80779713e-01 -9.27687287e-01 -7.52101600e-01
-3.68109196e-01 -9.66101646e-01 4.66264486e-01 1.03385830e+00
-4.18706834e-01 2.00903341e-01 -3.30805629e-01 6.54138565e-01
1.05898643e+00 4.58200336e-01 8.52169335e-01 -9.27190423e-01
-3.98958713e-01 -1.40207753e-01 -3.99504393e-01 -1.41592741e+00
6.09311014e-02 -1.04450917e+00 -1.33813649e-01 -1.67803466e+00
1.12455063e-01 -3.10199350e-01 1.20133840e-01 2.69951284e-01
3.39729905e-01 4.01099831e-01 3.15308362e-01 1.62532061e-01
-8.25309873e-01 1.06560457e+00 1.39697063e+00 3.36943299e-01
7.46582448e-02 -4.09791708e-01 -6.96648538e-01 8.38415742e-01
6.09017909e-01 -2.15853348e-01 -6.40722632e-01 -8.46657097e-01
4.50696796e-01 4.58780341e-02 1.17805803e+00 -7.23750889e-01
1.43306270e-01 -2.58849353e-01 4.48934585e-01 -5.87555945e-01
1.17602944e+00 -8.59709382e-01 3.31633717e-01 -1.87475875e-01
5.12900203e-03 -3.41499656e-01 6.99286237e-02 7.77221441e-01
-2.78491974e-02 1.72784552e-01 6.19216800e-01 -3.86524051e-01
-9.11225200e-01 5.34814656e-01 3.31508130e-01 7.66137317e-02
8.12112212e-01 -4.76671576e-01 -1.64700180e-01 -5.72248161e-01
-9.04253423e-01 -5.43138422e-02 1.05465782e+00 3.32215428e-01
1.15220618e+00 -1.31402934e+00 -5.40537238e-01 3.78841877e-01
3.98003072e-01 6.36173904e-01 6.13312662e-01 7.11220980e-01
-7.51556754e-01 4.67156321e-01 -1.87698398e-02 -8.76565993e-01
-8.39768112e-01 4.21367198e-01 3.46582204e-01 4.78767365e-01
-1.05466521e+00 9.88935709e-01 8.83539736e-01 -9.14301217e-01
4.87683006e-02 -3.42217773e-01 5.29122531e-01 -6.48869753e-01
3.23523879e-01 3.30343813e-01 -6.18413799e-02 -7.86532938e-01
-3.73786330e-01 9.31544542e-01 4.64350104e-01 -3.49453986e-01
1.45680130e+00 -6.98416233e-01 -1.36836991e-01 5.76203704e-01
1.25178242e+00 8.27935562e-02 -1.80980742e+00 -5.59586048e-01
-3.68170232e-01 -9.93743539e-01 1.20321974e-01 -7.08675086e-01
-6.82045698e-01 9.63840783e-01 5.11738099e-02 -2.80277073e-01
1.32084310e+00 3.40189785e-01 5.56353390e-01 2.23102197e-01
7.40797520e-01 -6.33810341e-01 1.20977826e-01 6.65320873e-01
1.25297749e+00 -1.21823740e+00 1.45918980e-01 -8.24110031e-01
-6.78545594e-01 8.45745623e-01 4.24595952e-01 -3.30297470e-01
8.20590973e-01 -2.49413196e-02 -2.32130140e-01 -5.21588802e-01
-4.53131676e-01 -3.07720274e-01 3.17520291e-01 6.45968139e-01
-4.49440442e-02 1.18696287e-01 7.18521535e-01 3.43959242e-01
-3.37738365e-01 -2.77191103e-01 7.23504424e-01 6.96228802e-01
-2.06814125e-01 -7.54579544e-01 -3.43747318e-01 -2.63954967e-01
2.66057462e-01 -3.24806392e-01 -3.69735479e-01 6.61027670e-01
1.75596952e-01 1.03674722e+00 1.22786850e-01 -2.15741143e-01
3.39868903e-01 -2.13889405e-01 9.21664357e-01 -9.86385643e-01
2.93543428e-01 7.20517933e-02 -4.79926541e-02 -1.24984455e+00
-5.91361523e-01 -6.95577383e-01 -1.29005551e+00 -2.14324832e-01
1.21783279e-01 -4.98784959e-01 7.44424641e-01 6.05568945e-01
6.09499574e-01 5.33690723e-03 7.94311881e-01 -1.19412577e+00
4.12801616e-02 -5.48770368e-01 -7.81479478e-01 5.11454582e-01
7.08592236e-01 -7.29517937e-01 -5.13404727e-01 2.28996396e-01] | [8.696176528930664, -2.899186372756958] |
3c6136ca-d481-4bd2-be1e-6e1159741a43 | instance-aware-semantic-segmentation-via | 1512.04412 | null | http://arxiv.org/abs/1512.04412v1 | http://arxiv.org/pdf/1512.04412v1.pdf | Instance-aware Semantic Segmentation via Multi-task Network Cascades | Semantic segmentation research has recently witnessed rapid progress, but
many leading methods are unable to identify object instances. In this paper, we
present Multi-task Network Cascades for instance-aware semantic segmentation.
Our model consists of three networks, respectively differentiating instances,
estimating masks, and categorizing objects. These networks form a cascaded
structure, and are designed to share their convolutional features. We develop
an algorithm for the nontrivial end-to-end training of this causal, cascaded
structure. Our solution is a clean, single-step training framework and can be
generalized to cascades that have more stages. We demonstrate state-of-the-art
instance-aware semantic segmentation accuracy on PASCAL VOC. Meanwhile, our
method takes only 360ms testing an image using VGG-16, which is two orders of
magnitude faster than previous systems for this challenging problem. As a by
product, our method also achieves compelling object detection results which
surpass the competitive Fast/Faster R-CNN systems.
The method described in this paper is the foundation of our submissions to
the MS COCO 2015 segmentation competition, where we won the 1st place. | ['Jian Sun', 'Kaiming He', 'Jifeng Dai'] | 2015-12-14 | instance-aware-semantic-segmentation-via-1 | http://openaccess.thecvf.com/content_cvpr_2016/html/Dai_Instance-Aware_Semantic_Segmentation_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Dai_Instance-Aware_Semantic_Segmentation_CVPR_2016_paper.pdf | cvpr-2016-6 | ['multi-human-parsing'] | ['computer-vision'] | [ 4.89465475e-01 1.60950065e-01 -1.17150001e-01 -5.90695083e-01
-1.06054544e+00 -7.82244980e-01 4.04532373e-01 -1.55918822e-01
-7.33670294e-01 3.57495070e-01 -2.73277998e-01 -2.86066592e-01
3.87824655e-01 -5.32544315e-01 -1.02204168e+00 -5.11889815e-01
1.73429132e-01 5.16597509e-01 7.35851288e-01 -1.30035639e-01
2.33179599e-01 1.82356313e-01 -1.52451885e+00 6.96840286e-01
5.94507456e-01 1.12401986e+00 1.20984837e-01 9.70013618e-01
-1.61684334e-01 7.18161583e-01 -6.19677246e-01 -5.63167930e-01
3.47404391e-01 -3.36115927e-01 -1.28366101e+00 2.27198422e-01
9.61003482e-01 -3.56300801e-01 -4.01601829e-02 9.92645800e-01
4.49563980e-01 -6.98231086e-02 3.15649033e-01 -1.22904348e+00
-5.42390347e-01 8.28879178e-01 -7.88222909e-01 3.74209762e-01
-2.22557813e-01 2.70603970e-02 1.16224802e+00 -9.74989712e-01
4.65468764e-01 1.23583257e+00 8.09954405e-01 8.06319177e-01
-1.06081724e+00 -6.59854710e-01 6.85446858e-01 1.22705922e-01
-1.18449366e+00 -2.81362325e-01 3.45564544e-01 -3.93856913e-01
9.68185842e-01 6.06019348e-02 7.29173005e-01 1.08398879e+00
-3.82121325e-01 1.49626982e+00 9.08920944e-01 -7.43347406e-02
1.19163357e-01 -2.23583773e-01 2.56937802e-01 8.10047984e-01
2.69104630e-01 -2.91207910e-01 -3.38455141e-01 2.89579213e-01
8.97774577e-01 -4.37558489e-03 1.56849042e-01 -1.67063013e-01
-1.14809716e+00 6.12375140e-01 7.55064249e-01 3.22626270e-02
-3.20147350e-02 5.40768206e-01 3.46669167e-01 -1.26206055e-01
6.89234674e-01 2.77287930e-01 -6.04292095e-01 1.32939368e-01
-1.21462417e+00 4.34077233e-01 5.87457061e-01 1.14151096e+00
5.54445326e-01 -2.21236289e-01 -2.15107307e-01 7.67524421e-01
1.41131774e-01 1.23304762e-01 2.76015460e-01 -1.07599151e+00
3.90093565e-01 6.35161817e-01 9.85496342e-02 -3.20253819e-01
-4.77600425e-01 -6.99881375e-01 -6.22376502e-01 1.01472676e-01
7.75581539e-01 -1.62320510e-01 -1.36911666e+00 1.43327510e+00
5.12304604e-01 5.28290927e-01 -1.70312628e-01 1.12858033e+00
1.00068963e+00 5.00757277e-01 3.44793379e-01 4.84238982e-01
1.63858306e+00 -1.62502337e+00 -2.49483362e-01 -4.19991881e-01
5.20725131e-01 -5.93358755e-01 8.58864665e-01 4.82257128e-01
-1.43845797e+00 -6.66148424e-01 -1.09261799e+00 -4.98236984e-01
-5.13170242e-01 2.00549811e-01 8.75650406e-01 5.30851185e-01
-1.26773071e+00 6.53662205e-01 -9.46530640e-01 -2.43771359e-01
1.16927338e+00 3.12573910e-01 6.67001531e-02 -7.88193196e-02
-8.34419608e-01 5.26229143e-01 4.09275204e-01 3.35469723e-01
-1.11563170e+00 -9.45562661e-01 -6.57210290e-01 -7.47270286e-02
6.08041108e-01 -6.68379068e-01 1.67012417e+00 -1.05716860e+00
-1.23683071e+00 1.33472562e+00 -2.30915710e-01 -7.68425643e-01
8.35265279e-01 -4.91451144e-01 -7.30472058e-02 1.80740863e-01
3.88172835e-01 1.23673820e+00 8.42964411e-01 -1.21008921e+00
-1.12973952e+00 -3.99489284e-01 2.54071921e-01 1.44307941e-01
5.00857411e-03 1.35561675e-01 -8.15357566e-01 -5.33954978e-01
1.48641258e-01 -7.39144206e-01 -6.76490664e-01 1.09137885e-01
-5.70647120e-01 -4.06008482e-01 7.63422549e-01 -3.92873406e-01
7.25110769e-01 -2.04128313e+00 -7.65998438e-02 -2.91843832e-01
5.30015826e-01 4.02001947e-01 -2.20950827e-01 -1.04435213e-01
9.85117331e-02 3.32129955e-01 -6.62281930e-01 -7.54519403e-01
-3.23686302e-02 -1.31918579e-01 -3.97909045e-01 3.79357487e-01
6.40745163e-01 1.35904050e+00 -1.00869560e+00 -4.70766515e-01
3.76426950e-02 3.84509355e-01 -4.65658844e-01 6.72186911e-02
-4.48638678e-01 2.84434527e-01 -3.19643140e-01 7.86075652e-01
6.87379420e-01 -6.38552606e-01 -1.88514560e-01 -1.92428187e-01
-1.33026540e-01 2.06963167e-01 -9.37057614e-01 1.95926929e+00
-1.00852281e-01 6.41468883e-01 2.32802480e-01 -1.22987187e+00
5.02806365e-01 -3.50709558e-02 3.08998853e-01 -6.06045604e-01
1.64749101e-01 2.93058276e-01 -1.69708773e-01 -1.19923756e-01
5.56802928e-01 1.73953772e-01 -1.32958665e-01 1.94209740e-01
2.59219646e-01 -3.02732974e-01 5.37166774e-01 3.57665062e-01
9.37456906e-01 4.00442749e-01 -3.55437770e-02 -2.04264805e-01
2.49000534e-01 2.32387111e-01 5.35500526e-01 1.03475142e+00
-3.81976485e-01 9.82787549e-01 6.30086243e-01 -5.57442129e-01
-1.02127111e+00 -1.09411955e+00 -6.25345781e-02 1.29153538e+00
4.47385550e-01 -1.99755058e-01 -1.16620648e+00 -8.45405102e-01
6.92765741e-03 2.84260988e-01 -7.03255594e-01 3.07735294e-01
-6.91255450e-01 -9.63572741e-01 8.70510280e-01 9.30384159e-01
9.73754466e-01 -1.09283936e+00 -4.96680975e-01 3.01767707e-01
-1.10170864e-01 -1.54678178e+00 -3.68181378e-01 2.22365901e-01
-9.00001705e-01 -1.24181986e+00 -9.30440784e-01 -1.11127567e+00
6.33664966e-01 3.93181294e-01 1.32026207e+00 2.66439140e-01
-5.84448993e-01 8.73649269e-02 -2.21051797e-01 -5.77705681e-01
7.65725300e-02 3.15191448e-01 -6.36180162e-01 -2.07766145e-02
3.88718218e-01 -2.99635530e-02 -1.08235514e+00 3.24963778e-01
-8.95477951e-01 3.16392273e-01 6.16574109e-01 5.87466240e-01
9.65643644e-01 -3.62677693e-01 6.23257399e-01 -1.26324332e+00
1.27231553e-01 -3.92395079e-01 -6.83503151e-01 2.10897386e-01
-4.20335859e-01 -2.97244638e-01 3.38916153e-01 -1.18179366e-01
-1.08533180e+00 5.28461158e-01 -3.95679325e-01 -1.82038903e-01
-4.58600163e-01 -6.01844192e-02 -2.15001535e-02 -3.46027054e-02
4.40800011e-01 -5.37134223e-02 -3.41445237e-01 -4.91523057e-01
7.26393700e-01 4.89471018e-01 8.35164905e-01 -5.30866086e-01
6.03699446e-01 7.77014554e-01 -3.06308389e-01 -5.30627549e-01
-1.44417059e+00 -8.63619566e-01 -7.24539995e-01 -8.23636428e-02
1.33704960e+00 -1.25501251e+00 -5.82026839e-01 9.20666277e-01
-1.30486178e+00 -8.16581309e-01 -2.03798339e-01 -1.38218433e-01
-4.98795301e-01 7.32218027e-02 -9.08739448e-01 -5.10352671e-01
-2.68763214e-01 -1.09411514e+00 1.44253325e+00 3.88123006e-01
9.88210589e-02 -7.17921853e-01 -3.52794975e-01 9.16128278e-01
3.89340371e-01 1.89134419e-01 2.70602316e-01 -6.39633775e-01
-8.93141568e-01 2.71015167e-02 -7.66166091e-01 4.21565354e-01
-3.18539530e-01 -3.51201370e-02 -1.16950488e+00 -1.53490618e-01
-2.75938779e-01 -5.51858008e-01 1.61514413e+00 5.71079016e-01
1.77626586e+00 9.07177180e-02 -4.14126009e-01 8.04536879e-01
1.41337764e+00 -1.41455784e-01 5.37504077e-01 2.09082350e-01
1.03028405e+00 5.06935418e-01 5.17396688e-01 -2.94327782e-03
5.67885697e-01 3.96326095e-01 5.84391177e-01 -6.71181858e-01
-4.56800222e-01 -1.13925695e-01 -5.55376969e-02 1.63926974e-01
-5.42905815e-02 -3.26948434e-01 -8.93104136e-01 8.63814950e-01
-1.94433975e+00 -7.56616354e-01 -4.49061722e-01 1.72493124e+00
6.58547282e-01 3.40009540e-01 4.20039654e-01 -1.89011335e-01
9.44057226e-01 2.28310138e-01 -8.73089612e-01 -2.17955500e-01
-1.17809877e-01 4.87066269e-01 7.58137584e-01 3.48645121e-01
-1.63209295e+00 1.59714580e+00 6.66827106e+00 8.05205643e-01
-7.87699580e-01 2.66584307e-01 1.12229645e+00 -2.63714135e-01
3.75171797e-03 -1.78567201e-01 -1.00318921e+00 2.47187704e-01
7.07377434e-01 2.52138406e-01 1.53961927e-01 9.62127805e-01
-1.97521493e-01 -5.34038916e-02 -9.81128335e-01 7.92360663e-01
-3.86019121e-04 -1.61246288e+00 1.29448799e-02 -2.66709447e-01
9.42229271e-01 5.06506324e-01 1.49674803e-01 3.38507265e-01
6.41538262e-01 -1.09807110e+00 9.53350008e-01 -2.78655514e-02
7.00634241e-01 -5.42938352e-01 6.01779819e-01 6.48809373e-02
-1.22722423e+00 -4.84897159e-02 -3.17911416e-01 -5.18984012e-02
3.20711941e-01 6.32915080e-01 -5.09867609e-01 1.93633094e-01
1.12899923e+00 8.46884727e-01 -6.78686440e-01 1.27833009e+00
-3.99003714e-01 7.55314648e-01 -1.27551094e-01 -1.96963153e-03
6.01376951e-01 1.29017115e-01 1.19659416e-01 1.57710648e+00
-1.74464434e-01 1.98587522e-01 2.16335475e-01 1.07826042e+00
-5.70067525e-01 -3.04412991e-01 -2.40738429e-02 -5.26297465e-02
2.53439486e-01 1.61063242e+00 -1.41679621e+00 -7.16185451e-01
-4.45148021e-01 1.27012658e+00 4.21919256e-01 3.97802889e-01
-1.04353356e+00 -3.71743023e-01 6.60031617e-01 -1.43724784e-01
6.17251098e-01 -4.59312648e-02 -7.13727236e-01 -1.02609921e+00
1.38001088e-02 -5.53232014e-01 2.73727268e-01 -6.27477765e-01
-1.30968714e+00 4.75814313e-01 -1.65998012e-01 -6.37271941e-01
1.82476357e-01 -8.65781784e-01 -5.23579001e-01 5.39971352e-01
-1.76531231e+00 -1.26936615e+00 -4.75494832e-01 4.64087725e-01
9.21433747e-01 3.20960701e-01 3.32875192e-01 4.00432557e-01
-7.75756776e-01 5.63313305e-01 -3.49736810e-01 4.80593294e-01
4.78787184e-01 -1.55207181e+00 1.13351405e+00 1.00054431e+00
4.01259542e-01 1.88768938e-01 2.54399151e-01 -4.79704469e-01
-8.93704653e-01 -1.49690759e+00 6.35227978e-01 -6.78294718e-01
7.06923604e-01 -7.03530371e-01 -8.59902382e-01 8.46746445e-01
5.69788180e-02 3.21801662e-01 4.30024743e-01 6.30698055e-02
-5.50933778e-01 1.68235805e-02 -1.05044222e+00 4.50987369e-01
1.49604213e+00 -2.22028524e-01 -3.64570796e-01 6.14498079e-01
1.06360006e+00 -8.14927757e-01 -4.05233949e-01 3.91427189e-01
4.25316364e-01 -8.99260283e-01 1.06385946e+00 -6.90646887e-01
7.02375054e-01 -3.42771679e-01 1.65299565e-01 -9.16357994e-01
-2.59233505e-01 -4.35905904e-01 1.08083293e-01 1.12169886e+00
7.04862237e-01 -4.31245953e-01 1.04413998e+00 4.18635666e-01
-4.67754960e-01 -9.27344918e-01 -6.79965794e-01 -6.21839345e-01
1.86176077e-01 -5.16479075e-01 5.38773179e-01 5.94725907e-01
-6.25460446e-01 3.16083878e-01 1.48809463e-01 2.10740671e-01
9.35546398e-01 1.78292945e-01 6.30284250e-01 -1.06799996e+00
-2.04382017e-01 -7.89238334e-01 -2.28210673e-01 -1.46789443e+00
-2.49645044e-03 -8.80927265e-01 3.70433509e-01 -1.80730927e+00
3.87540787e-01 -4.09636319e-01 -5.22164226e-01 6.43059134e-01
-5.64719439e-01 9.49949026e-01 1.58554032e-01 5.52357174e-02
-1.21372116e+00 6.95415512e-02 1.34018087e+00 -3.09773386e-01
1.77757163e-02 -1.49906809e-02 -1.00514889e+00 7.78714418e-01
9.01370168e-01 -3.27731997e-01 -1.81332499e-01 -7.93847263e-01
-3.64557803e-02 -4.96956438e-01 7.78187573e-01 -8.93875182e-01
2.67128021e-01 1.76756307e-01 5.52841067e-01 -8.04529309e-01
2.65524566e-01 -3.42385709e-01 -4.35581833e-01 3.07594061e-01
-4.02819365e-01 -1.67037964e-01 3.49934071e-01 5.52270591e-01
-2.24538967e-01 5.44856228e-02 8.36048841e-01 -4.83224750e-01
-1.16862237e+00 6.06063664e-01 -3.47406082e-02 4.32473123e-01
9.87181723e-01 -3.17041069e-01 -4.91964936e-01 1.74829453e-01
-6.54456973e-01 6.01221919e-01 1.98056772e-01 5.28698981e-01
4.93355840e-01 -8.77886653e-01 -8.10168862e-01 -1.09599717e-01
-8.43915418e-02 6.48010135e-01 3.38301748e-01 7.14602947e-01
-6.37910128e-01 1.89049631e-01 -8.19860550e-04 -8.96284878e-01
-1.07619178e+00 3.39805514e-01 3.43817532e-01 -1.59274399e-01
-6.07238889e-01 1.53810263e+00 4.39649463e-01 -2.80725509e-01
2.31744274e-01 -3.31525922e-01 8.69781151e-03 8.11733454e-02
6.14022613e-01 3.18162501e-01 5.40364459e-02 -2.88937420e-01
-4.68974382e-01 5.38424790e-01 -4.12794203e-01 -1.01245239e-01
1.37300253e+00 6.91438317e-02 -9.65746045e-02 1.95660353e-01
1.13343441e+00 -6.17029786e-01 -1.77930295e+00 -9.19954926e-02
5.03006876e-02 -7.26265162e-02 -2.84582339e-02 -9.86272871e-01
-1.27816117e+00 1.01594365e+00 3.57575804e-01 2.23028690e-01
9.95906234e-01 3.72426838e-01 1.07886684e+00 1.03109963e-01
2.19541490e-01 -1.24799228e+00 8.48638564e-02 4.84040052e-01
4.92160738e-01 -1.53405547e+00 -2.32433096e-01 -7.27381587e-01
-4.85979795e-01 8.34660590e-01 9.61571634e-01 -4.78605807e-01
3.36574703e-01 3.14949483e-01 7.39044100e-02 -3.38673204e-01
-5.48667848e-01 -6.00414574e-01 3.28366667e-01 4.79282141e-01
3.42191190e-01 1.12911545e-01 -1.55248448e-01 5.85880458e-01
-9.78066474e-02 -1.06705435e-01 2.74000674e-01 5.83274901e-01
-5.16800523e-01 -8.68661225e-01 -2.46245731e-02 3.94880354e-01
-7.08937585e-01 -2.22316697e-01 -5.69699705e-01 8.11298966e-01
2.49781385e-01 1.05202699e+00 2.77019143e-01 -1.03073761e-01
3.52832407e-01 -1.62360724e-02 5.28630197e-01 -7.21346796e-01
-7.34135270e-01 -1.00078270e-01 -6.94471747e-02 -1.02084529e+00
-5.14215052e-01 -5.42068183e-01 -1.42977726e+00 -3.64128649e-02
-2.25799799e-01 -1.96601659e-01 7.96898007e-01 1.01229501e+00
3.59856099e-01 7.51979470e-01 2.12645814e-01 -1.15552962e+00
-2.38700226e-01 -7.19951391e-01 -2.97268480e-01 3.83811593e-01
3.61247301e-01 -4.08519268e-01 -2.14800119e-01 1.83438689e-01] | [9.524362564086914, 0.3577296733856201] |
b75a927b-f060-4ace-b202-c208e043c11c | deep-attributed-graph-clustering-with-self | null | null | https://www.sciencedirect.com/science/article/abs/pii/S0893608021002756 | https://www.sciencedirect.com/science/article/abs/pii/S0893608021002756 | Deep attributed graph clustering with self-separation regularization and parameter-free cluster estimation | Detecting clusters over attributed graphs is a fundamental task in the graph analysis field. The goal is
to partition nodes into dense clusters based on both their attributes and structures. Modern graph
neural networks provide facilitation to jointly capture the above information in attributed graphs
with a feature aggregation manner, and have achieved great success in attributed graph clustering.
However, existing methods mainly focus on capturing the proximity information in graphs and often
fail to learn cluster-friendly features during the training of models. Besides, similar to many deep
clustering frameworks, current methods based on graph neural networks require a preassigned cluster
number before estimating the clusters. To address these limitations, we propose in this paper a
deep attributed clustering method based on self-separated graph neural networks and parameterfree
cluster estimation. First, to learn cluster-friendly features, we jointly optimize a jumping graph
convolutional auto-encoder with a self-separation regularizer, which learns clusters with changing
sizes while keeping dense intra-cluster structures and sparse inter structures. Second, an additional
softmax auto-encoder is trained to determine the natural cluster number from the data. The hidden
units capture cluster structures and can be used to estimate the number of clusters. Extensive
experiments show the effectiveness of the proposed model. | ['Minglong Lei', 'Ye Liang', 'Junzhong Ji'] | 2021-07-15 | null | null | null | neural-networks-2021-7 | ['graph-clustering'] | ['graphs'] | [-1.40686855e-01 2.40606248e-01 -2.25804970e-01 -7.04386950e-01
-2.52188802e-01 -3.28898579e-01 4.27771568e-01 5.46272993e-01
-1.71965957e-01 -3.41543742e-02 2.70049311e-02 6.01804927e-02
-2.40864724e-01 -8.53284478e-01 -6.73893273e-01 -8.02046955e-01
-5.53032815e-01 7.11526930e-01 -1.02576226e-01 1.98106706e-01
-6.64723990e-03 1.82896137e-01 -1.37171209e+00 2.40676567e-01
1.17383122e+00 7.84836471e-01 3.14914644e-01 6.00359082e-01
-1.56098217e-01 1.05970514e+00 -4.84763861e-01 -2.87350953e-01
5.38881943e-02 -2.69439638e-01 -6.09954715e-01 4.50037479e-01
1.05307095e-01 -5.03578819e-02 -7.38331676e-01 1.22309422e+00
3.11186671e-01 8.72360766e-02 6.27492130e-01 -1.43522620e+00
-9.41054881e-01 1.17973876e+00 -6.06625736e-01 -5.83946817e-02
-1.45169541e-01 -8.78266320e-02 1.23427212e+00 -4.86436129e-01
2.27140561e-01 1.11186397e+00 5.79874754e-01 1.61597997e-01
-1.25098157e+00 -7.04349220e-01 4.72902983e-01 2.21573278e-01
-1.74811995e+00 -1.73299357e-01 1.11638093e+00 -5.71697891e-01
6.35461688e-01 2.04865891e-03 7.72006929e-01 6.76817060e-01
-1.78823635e-01 8.23787510e-01 4.61081296e-01 -2.03271329e-01
1.65129811e-01 -5.49948178e-02 2.60252029e-01 9.22409713e-01
1.74719110e-01 -3.66438836e-01 -1.37047265e-02 9.30385962e-02
6.10840857e-01 5.36729813e-01 -6.57868534e-02 -5.20350635e-01
-1.05003536e+00 1.07093120e+00 1.14591467e+00 4.04195428e-01
-3.05240452e-01 3.32816839e-01 2.67230481e-01 5.35751618e-02
5.75427353e-01 2.38186806e-01 -8.58087018e-02 4.66275871e-01
-9.73477125e-01 -2.14128479e-01 5.82466066e-01 1.13028800e+00
1.18503177e+00 -1.42507749e-02 -1.05257042e-01 7.74136543e-01
5.02502382e-01 1.10652313e-01 3.22319835e-01 -6.17448628e-01
4.34116960e-01 1.26407230e+00 -5.56243718e-01 -1.44010258e+00
-5.55154622e-01 -6.89628005e-01 -1.28698909e+00 -1.39198020e-01
2.92557105e-02 4.52306867e-02 -1.20197105e+00 1.70127845e+00
3.29777628e-01 3.21338832e-01 -1.04700439e-01 8.91868234e-01
9.19885278e-01 6.48633838e-01 9.60391108e-03 1.35741472e-01
1.07970822e+00 -9.69735801e-01 -6.98268414e-01 -2.57155418e-01
7.09455729e-01 -3.69732261e-01 8.18984151e-01 3.77381667e-02
-8.06231797e-01 -5.05212545e-01 -1.02693295e+00 2.22388897e-02
-3.46128792e-01 1.55566126e-01 8.78524303e-01 3.40081602e-01
-1.09255683e+00 6.97352707e-01 -1.08652079e+00 -1.18432574e-01
6.29157484e-01 7.82393098e-01 -3.65037680e-01 -3.58080468e-03
-9.93845165e-01 2.53735363e-01 7.90860415e-01 4.12852287e-01
-6.41687036e-01 -2.83527583e-01 -1.13253295e+00 5.73802114e-01
3.89008790e-01 -5.20347357e-01 5.54583132e-01 -1.23232925e+00
-1.17714071e+00 6.77089155e-01 -3.30849085e-04 -2.44198859e-01
-1.35327308e-02 2.21432105e-01 -3.85804504e-01 1.01210110e-01
3.19615901e-01 6.79972053e-01 7.59575009e-01 -1.12742376e+00
-5.88985622e-01 -5.53695142e-01 -1.39970958e-01 1.76284224e-01
-7.80613065e-01 -3.09918821e-01 -9.66338992e-01 -6.88639164e-01
2.70191699e-01 -9.63586509e-01 -5.75038970e-01 -5.69204807e-01
-9.47068810e-01 -4.15954977e-01 7.40477622e-01 -3.57463300e-01
1.43615460e+00 -2.23954439e+00 3.78853053e-01 5.30661702e-01
1.08080983e+00 -6.10515103e-02 -1.08531095e-01 2.24982575e-01
-9.75289941e-02 1.46855652e-01 -3.55240166e-01 -3.67285103e-01
2.07078457e-01 4.50330302e-02 2.64021754e-01 6.08613133e-01
4.13988382e-01 9.32686627e-01 -8.61635208e-01 -6.42378509e-01
3.24036449e-01 6.05330765e-01 -6.11737311e-01 5.59639812e-01
-5.96160702e-02 2.42012531e-01 -4.21678305e-01 4.45667535e-01
6.63839400e-01 -8.71227264e-01 4.61420983e-01 -2.81497806e-01
2.21813560e-01 -3.95742152e-03 -1.15195310e+00 1.47950220e+00
-1.07287146e-01 4.50638294e-01 2.72059351e-01 -1.45344055e+00
9.68937635e-01 1.12614237e-01 8.09038818e-01 -2.18807891e-01
3.05954158e-01 -3.86994854e-02 2.24846989e-01 -2.93281347e-01
1.73992276e-01 1.39812604e-01 -1.13196746e-01 2.89962590e-01
1.89825848e-01 4.84538108e-01 3.18306178e-01 4.88887489e-01
1.32266223e+00 -5.26945829e-01 -1.05358995e-01 -2.65075445e-01
4.13353890e-01 -1.90596700e-01 6.51506722e-01 2.87956774e-01
-3.16925682e-02 6.77073956e-01 7.56982148e-01 -5.31096160e-01
-7.71918595e-01 -8.17133307e-01 2.46898532e-01 1.10253572e+00
3.07807654e-01 -6.86672091e-01 -8.47407877e-01 -8.57674241e-01
-1.14424899e-02 8.54402035e-02 -8.17728817e-01 -3.47972125e-01
-5.54150879e-01 -9.43566620e-01 2.76921272e-01 6.18260384e-01
1.71947092e-01 -1.04698479e+00 7.54683912e-02 2.68871009e-01
-4.79929708e-02 -1.12971246e+00 -7.06543922e-01 6.60371244e-01
-5.98432422e-01 -1.33823907e+00 -4.42416131e-01 -1.26573980e+00
1.07863069e+00 2.92225003e-01 1.06895876e+00 4.69434261e-01
-1.17008865e-01 -8.38633627e-02 -3.97665381e-01 -8.93269032e-02
-1.36165872e-01 6.82831883e-01 -1.67596623e-01 3.65270883e-01
4.82551605e-01 -7.54121840e-01 -4.67213601e-01 1.21940985e-01
-8.55283856e-01 7.29818270e-02 8.28042090e-01 6.59466922e-01
5.41548312e-01 4.20174211e-01 3.82505089e-01 -1.24711537e+00
5.41896403e-01 -7.02509999e-01 -5.02412736e-01 1.16115399e-01
-6.23704314e-01 3.33490551e-01 9.79994595e-01 -3.54709476e-01
-4.98705298e-01 4.02700752e-01 -5.86077459e-02 -9.47269320e-01
-8.47144276e-02 7.54065871e-01 -5.94126940e-01 1.80379421e-01
2.09561184e-01 8.04155692e-03 3.31171006e-02 -1.77111492e-01
5.26448429e-01 4.82000321e-01 4.34616059e-01 -2.65029788e-01
8.29479754e-01 4.32607204e-01 -1.80273592e-01 -5.62578082e-01
-7.11883426e-01 -6.88344479e-01 -9.15561378e-01 -8.11984688e-02
9.95711744e-01 -9.42385197e-01 -8.18322480e-01 2.97362357e-01
-8.72574687e-01 -5.87867558e-01 1.04717433e-01 4.73721862e-01
-1.90382019e-01 4.57182854e-01 -7.68742561e-01 -3.80984634e-01
-2.19377488e-01 -1.24383605e+00 1.06463718e+00 1.05110668e-01
7.92208463e-02 -1.21744347e+00 -6.91248849e-02 3.00141554e-02
8.75468552e-02 4.83982146e-01 9.09618437e-01 -8.19083095e-01
-4.86808330e-01 -3.42438817e-01 -4.51301664e-01 -3.00163105e-02
4.93881226e-01 8.93381014e-02 -7.00166225e-01 -4.82026517e-01
-5.49478233e-01 -2.17362434e-01 1.20016491e+00 6.28745735e-01
1.41066742e+00 -5.60335279e-01 -7.74924159e-01 1.00078976e+00
1.26562321e+00 -6.31400272e-02 1.95019260e-01 -5.58628216e-02
1.54540896e+00 6.30891740e-01 -1.23145711e-02 2.21912310e-01
8.40540588e-01 3.38587582e-01 7.23388791e-01 -3.49177718e-01
5.33214305e-03 -1.91083312e-01 1.01217717e-01 1.23771524e+00
1.16524026e-01 -2.87087202e-01 -8.89847755e-01 5.06754041e-01
-1.97848356e+00 -8.13364565e-01 -2.77918816e-01 1.92983723e+00
5.48591554e-01 2.97667652e-01 1.77403644e-01 6.58594631e-03
1.33028734e+00 2.27709725e-01 -5.98398149e-01 1.09631985e-01
-3.98152694e-02 -1.81687430e-01 4.20554876e-01 5.98271668e-01
-1.40202808e+00 9.95382607e-01 5.12625551e+00 5.95498204e-01
-1.19785094e+00 -2.04232037e-01 8.17592561e-01 1.73677385e-01
-2.73777783e-01 8.51140842e-02 -3.69002104e-01 7.10462272e-01
8.07353079e-01 2.14249808e-02 4.55437601e-01 9.32916880e-01
-2.13200986e-01 7.05311060e-01 -1.13755274e+00 8.48085225e-01
-2.18015872e-02 -1.21327996e+00 -8.42311904e-02 2.91291803e-01
6.60520017e-01 1.84177816e-01 -2.76349969e-02 4.99464661e-01
7.87109077e-01 -1.24094498e+00 5.81173003e-01 3.07404071e-01
7.13346601e-01 -1.02821076e+00 7.12717474e-01 2.73481369e-01
-1.65325403e+00 -1.91675335e-01 -6.14054501e-01 -9.40612778e-02
-2.27642015e-01 7.98527837e-01 -8.64248097e-01 4.39621508e-01
6.99851871e-01 1.26894021e+00 -6.57259285e-01 7.43007600e-01
-1.21769607e-01 7.32185543e-01 -3.92985940e-01 9.01434422e-02
3.69373411e-01 -4.52437788e-01 9.83478427e-02 1.13839483e+00
5.59497513e-02 -4.12421003e-02 6.31292641e-01 1.05505836e+00
-3.83907437e-01 -8.63549262e-02 -4.59306300e-01 -2.52116948e-01
6.27607226e-01 1.64303970e+00 -1.25657678e+00 -2.49792933e-01
-3.34162384e-01 9.46248829e-01 9.47283030e-01 3.46051633e-01
-8.09580803e-01 -6.36522353e-01 3.33489686e-01 2.15627715e-01
5.09604812e-01 -1.65028811e-01 4.00866196e-02 -1.22927094e+00
-1.17847793e-01 -6.25357807e-01 6.83157325e-01 -2.07525626e-01
-1.44028866e+00 9.45134103e-01 -3.43096107e-01 -1.12959218e+00
2.29955930e-02 -3.54230285e-01 -9.37602997e-01 4.65395004e-01
-1.15411866e+00 -1.50786507e+00 -5.78873217e-01 5.69829106e-01
1.16121002e-01 -1.56656384e-01 6.04600489e-01 5.25721967e-01
-1.15345585e+00 6.38815641e-01 1.69314444e-01 7.67250836e-01
6.53165996e-01 -1.68489432e+00 4.61177707e-01 7.84900725e-01
1.61825344e-01 7.10321665e-01 2.83582270e-01 -6.74087882e-01
-1.36298633e+00 -1.71942544e+00 5.12556076e-01 -2.06279546e-01
4.61478353e-01 -8.42675567e-01 -1.06211436e+00 8.99000525e-01
1.29626513e-01 4.82120275e-01 6.31886721e-01 5.11417627e-01
-2.91237593e-01 -3.20517540e-01 -6.07692480e-01 2.23805130e-01
8.30181003e-01 -3.95068079e-01 2.95388773e-02 2.82685935e-01
8.79475117e-01 -8.65404382e-02 -1.11731386e+00 2.77549416e-01
1.17023751e-01 -8.72605324e-01 7.84655631e-01 -4.35409248e-01
3.53989959e-01 -4.41920578e-01 1.16501711e-01 -1.56622159e+00
-9.01518881e-01 -5.15125513e-01 -2.41901562e-01 1.32651198e+00
2.73859024e-01 -2.95427352e-01 1.19332445e+00 3.73096943e-01
-4.05058861e-01 -7.71970272e-01 -5.02363563e-01 -4.48942900e-01
-1.81284785e-01 -8.47694948e-02 8.30428898e-01 1.41391075e+00
1.11493364e-01 9.26651955e-01 -8.22705775e-02 5.88909030e-01
8.61915886e-01 2.75284141e-01 7.63069808e-01 -1.54226851e+00
-4.51841764e-03 -8.10250163e-01 -6.59634948e-01 -8.00188124e-01
5.05077600e-01 -1.36842465e+00 7.57631287e-02 -1.70067024e+00
3.56080413e-01 -7.22596288e-01 -4.61042196e-01 6.21618629e-01
-7.70322800e-01 1.87891982e-02 1.36175435e-02 3.33743781e-01
-1.19435716e+00 7.96121895e-01 7.80437112e-01 -5.37147045e-01
-3.35520327e-01 -1.69845700e-01 -8.21534216e-01 6.44598007e-01
6.71909153e-01 -6.27970457e-01 -4.57903177e-01 -6.28355861e-01
2.09652498e-01 -1.12627521e-01 2.59896100e-01 -9.85952199e-01
5.88102281e-01 8.80826414e-02 4.60756510e-01 -5.33032715e-01
1.83449239e-02 -9.36638355e-01 -4.40215394e-02 4.01263028e-01
-3.12904954e-01 -4.59186845e-02 -2.32977107e-01 8.12003434e-01
-4.18249190e-01 3.10991555e-01 5.55901170e-01 -1.00367926e-01
-5.15339196e-01 8.59259427e-01 -1.35844991e-01 1.26297856e-02
9.72799122e-01 -2.23046511e-01 1.68215577e-02 -3.30731750e-01
-8.47502708e-01 7.43519366e-01 5.34695268e-01 2.84907848e-01
5.60008049e-01 -1.57062662e+00 -8.28329682e-01 2.62675911e-01
2.39312232e-01 6.28422081e-01 2.28619874e-01 7.37770319e-01
-4.65033144e-01 1.01248562e-01 -4.65759337e-02 -8.04353416e-01
-9.09043610e-01 9.98614967e-01 2.38642842e-01 -3.14642519e-01
-5.60297489e-01 8.95764887e-01 4.51614708e-01 -7.61998415e-01
4.18555439e-01 -1.66110292e-01 -3.85445446e-01 -3.66235189e-02
1.83423728e-01 1.48470789e-01 5.96490577e-02 -5.87345183e-01
-2.83365995e-01 2.29646429e-01 -2.29807034e-01 6.68451309e-01
1.49928975e+00 -1.06647961e-01 -3.72863203e-01 3.75721514e-01
1.48828340e+00 -2.72789121e-01 -1.28130436e+00 -2.83706099e-01
1.98277503e-01 -6.41193688e-02 1.15466356e-01 1.75833255e-02
-1.62018776e+00 9.95103300e-01 2.04686001e-01 5.75437307e-01
1.04210603e+00 1.57625973e-01 7.19846129e-01 3.40131849e-01
-2.66051799e-01 -8.30067039e-01 2.20077068e-01 1.64559200e-01
2.98032552e-01 -1.45948994e+00 -4.00476418e-02 -5.28035343e-01
-5.89139163e-01 9.84096289e-01 8.40015709e-01 -3.86268854e-01
1.00679135e+00 2.99311996e-01 -1.49219722e-01 -6.07878387e-01
-6.09008729e-01 -3.36787373e-01 3.90772343e-01 5.47701716e-01
5.08137167e-01 2.60268688e-01 3.26989621e-01 8.58575344e-01
-7.47589245e-02 -5.73059857e-01 3.36018473e-01 4.49754357e-01
-4.25966650e-01 -8.70496154e-01 -9.31476355e-02 8.05050194e-01
-4.40002263e-01 -1.47617832e-01 -6.00621581e-01 6.73348904e-01
1.25796303e-01 9.03555751e-01 4.22113836e-01 -7.50625968e-01
7.30685443e-02 -5.47472954e-01 1.03254586e-01 -7.96862900e-01
-5.02192974e-01 1.57985300e-01 -3.79497230e-01 -3.48806739e-01
-4.11118746e-01 -2.20726237e-01 -1.42760420e+00 -4.02821958e-01
-7.01523483e-01 5.78697920e-01 1.40094995e-01 7.22520709e-01
5.46065986e-01 7.83130586e-01 9.56688821e-01 -8.31949651e-01
-8.35053343e-03 -9.87190843e-01 -7.96850979e-01 6.69900954e-01
3.56034935e-01 -4.83997077e-01 -2.33484209e-01 6.68941885e-02] | [7.386258602142334, 5.954362392425537] |
97f42ff6-c16f-4ada-a614-aa8b8fef9e99 | multi-predictor-fusion-combining-learning | 2307.01408 | null | https://arxiv.org/abs/2307.01408v1 | https://arxiv.org/pdf/2307.01408v1.pdf | Multi-Predictor Fusion: Combining Learning-based and Rule-based Trajectory Predictors | Trajectory prediction modules are key enablers for safe and efficient planning of autonomous vehicles (AVs), particularly in highly interactive traffic scenarios. Recently, learning-based trajectory predictors have experienced considerable success in providing state-of-the-art performance due to their ability to learn multimodal behaviors of other agents from data. In this paper, we present an algorithm called multi-predictor fusion (MPF) that augments the performance of learning-based predictors by imbuing them with motion planners that are tasked with satisfying logic-based rules. MPF probabilistically combines learning- and rule-based predictors by mixing trajectories from both standalone predictors in accordance with a belief distribution that reflects the online performance of each predictor. In our results, we show that MPF outperforms the two standalone predictors on various metrics and delivers the most consistent performance. | ['Marco Pavone', 'Apoorva Sharma', 'Sushant Veer'] | 2023-07-03 | null | null | null | null | ['trajectory-prediction', 'autonomous-vehicles'] | ['computer-vision', 'computer-vision'] | [-3.49325031e-01 2.09821954e-01 -7.45272338e-01 -3.94496709e-01
-7.66445577e-01 -3.91982734e-01 1.14786696e+00 1.36252940e-01
-1.99460432e-01 8.59555364e-01 2.03557551e-01 -5.00704825e-01
-4.16676164e-01 -7.81720817e-01 -8.47550392e-01 -6.61989331e-01
-1.90586999e-01 9.58371818e-01 7.82531321e-01 -2.70063847e-01
-1.13493465e-02 5.77093005e-01 -1.92379403e+00 3.42294067e-01
9.81629670e-01 7.58023024e-01 1.85328379e-01 5.70336998e-01
3.69650647e-02 1.17432177e+00 2.79825386e-02 -4.12845671e-01
8.41037109e-02 3.34399231e-02 -4.63045031e-01 -3.81438583e-01
-5.42567968e-02 -2.25964367e-01 -5.59560061e-01 5.95060587e-01
-1.56277061e-01 3.27596217e-01 8.24252486e-01 -1.93814564e+00
4.39530402e-01 5.73334336e-01 -5.58784558e-03 -2.53978670e-01
4.17458862e-01 4.45912063e-01 7.80478835e-01 -5.78835130e-01
8.24729323e-01 1.31508183e+00 4.75607365e-01 4.64411855e-01
-1.11055863e+00 -4.57028031e-01 2.62410402e-01 1.01936102e+00
-1.22322154e+00 -7.08526134e-01 4.32998747e-01 -6.23678565e-01
1.16449547e+00 2.22310469e-01 5.16195357e-01 9.56996620e-01
6.46730721e-01 1.05795693e+00 6.89995229e-01 1.58952475e-02
4.97292131e-01 2.35122308e-01 -4.58433144e-02 6.63897216e-01
1.51251152e-01 5.28267264e-01 -6.97923899e-01 -2.52294838e-01
4.14597569e-03 -1.48514539e-01 2.60879308e-01 -4.88313586e-01
-1.25475991e+00 5.97761631e-01 8.35077018e-02 -2.09981073e-02
-9.68473017e-01 3.44197094e-01 1.99492589e-01 6.16920665e-02
2.56883409e-02 1.99340042e-02 -3.52693021e-01 -5.19807577e-01
-7.56944716e-01 7.39092350e-01 8.18233132e-01 1.11227274e+00
9.20602083e-01 3.22572961e-02 -3.91593903e-01 1.60738960e-01
7.00308084e-01 5.75100482e-01 -5.55135049e-02 -1.40924716e+00
2.75077105e-01 6.45603180e-01 5.28134644e-01 -8.42607260e-01
-6.10708117e-01 5.13192341e-02 -4.53533053e-01 5.27889907e-01
1.63138628e-01 -3.25742930e-01 -7.34515250e-01 1.64034593e+00
4.30240393e-01 4.60635930e-01 3.90633106e-01 5.62874913e-01
5.08481920e-01 9.99753535e-01 3.79046887e-01 -2.65693784e-01
7.39785731e-01 -1.08130908e+00 -5.40589988e-01 -1.04229376e-01
3.71798277e-01 -1.40830353e-01 2.59517878e-01 5.75701952e-01
-8.82572114e-01 -5.41301489e-01 -9.16751802e-01 5.83405614e-01
-1.46877751e-01 -1.05065383e-01 3.91038150e-01 3.69248182e-01
-1.21666145e+00 5.10789931e-01 -1.10751033e+00 -3.60727221e-01
2.29433745e-01 5.33754408e-01 -4.07806188e-01 8.05590972e-02
-8.91232491e-01 1.60334432e+00 4.73943740e-01 -4.18248534e-01
-1.31010103e+00 -6.10949934e-01 -7.36774623e-01 -1.65403828e-01
6.02091908e-01 -5.26010811e-01 1.38596463e+00 -3.57241452e-01
-1.73786843e+00 -1.13104008e-01 -6.40930712e-01 -7.36145020e-01
4.36061114e-01 -1.67353734e-01 -5.76703787e-01 -7.27411406e-03
-2.11393479e-02 8.41957033e-01 6.65636182e-01 -1.60846281e+00
-1.34739494e+00 1.00967295e-01 -1.20848469e-01 4.31491919e-02
2.49702945e-01 -1.80759937e-01 -3.28157663e-01 2.60591924e-01
-1.87939778e-01 -8.94266784e-01 -5.89331925e-01 -3.84706229e-01
-2.42420822e-01 -7.27070987e-01 9.94760990e-01 -2.40751594e-01
1.20707166e+00 -1.84946156e+00 4.13873404e-01 4.41767901e-01
1.44160435e-01 3.09406877e-01 -1.14407457e-01 7.66638696e-01
6.06900692e-01 -4.03130323e-01 1.21031776e-01 -4.66939837e-01
3.69750947e-01 7.11409211e-01 -4.18475896e-01 3.66970271e-01
2.62437284e-01 7.98442185e-01 -1.16074395e+00 -6.73787951e-01
7.77667999e-01 2.25951344e-01 -3.67995739e-01 2.70907879e-01
-8.03017080e-01 6.98860407e-01 -5.53012669e-01 4.18416262e-01
3.66250336e-01 2.61793494e-01 4.30758208e-01 1.09233208e-01
-5.44333041e-01 1.26654819e-01 -9.56708431e-01 1.24524403e+00
-2.08088070e-01 6.33328915e-01 -5.21480888e-02 -5.00978768e-01
8.24370980e-01 3.81356031e-01 7.41932631e-01 -4.29897815e-01
-4.32431512e-02 2.01464698e-01 -1.12938307e-01 -6.66012406e-01
7.47470200e-01 5.20974323e-02 -4.68631983e-02 1.53510049e-01
7.52716735e-02 8.29232559e-02 3.74981701e-01 2.19411120e-01
1.33585846e+00 5.37243962e-01 3.67285311e-01 1.97964713e-01
7.87342250e-01 4.29411769e-01 8.27627897e-01 8.14273119e-01
-2.70952731e-01 -1.58364460e-01 4.32673305e-01 -3.96713048e-01
-8.63389552e-01 -1.02591372e+00 2.25091502e-01 9.57313776e-01
3.12195897e-01 -6.98388636e-01 -3.66169482e-01 -7.04656720e-01
1.16204545e-01 1.36540782e+00 -3.04765701e-01 5.80574162e-02
-5.44192910e-01 -2.12434262e-01 4.95741457e-01 4.52685356e-01
1.13155641e-01 -8.20090473e-01 -8.89730692e-01 5.01181602e-01
-1.47514835e-01 -1.10507238e+00 2.15091377e-01 2.57972870e-02
-3.96460295e-01 -1.05476391e+00 -2.25462057e-02 -1.54209152e-01
3.06771398e-01 1.81852445e-01 6.91670656e-01 -9.60903615e-02
6.74287021e-01 4.38726783e-01 -4.24999833e-01 -4.46632504e-01
-9.94623005e-01 3.15427743e-02 3.47897619e-01 1.56614468e-01
2.58658707e-01 -5.40441930e-01 -1.06268786e-01 5.28101325e-01
-4.34937149e-01 2.87773877e-01 6.78699672e-01 3.96180272e-01
5.23097098e-01 2.39707544e-01 5.87150991e-01 -3.74208480e-01
2.88071603e-01 -9.46147740e-01 -8.38983119e-01 2.56967902e-01
-8.22358429e-01 1.93615690e-01 5.47184885e-01 -1.41372785e-01
-1.21506846e+00 4.83838409e-01 -6.88957274e-02 -4.57468569e-01
-5.93837202e-01 4.91092175e-01 -1.43565774e-01 -2.52451450e-02
3.64479631e-01 2.12168291e-01 1.18453167e-01 -1.42889187e-01
6.13882482e-01 5.07769227e-01 6.80405915e-01 -4.98789459e-01
7.14797735e-01 2.84597784e-01 3.96572798e-01 -5.03538430e-01
-1.89173892e-01 -4.64500695e-01 -3.68585736e-01 -1.06197512e+00
6.72068119e-01 -5.97870767e-01 -1.34389782e+00 3.27946216e-01
-1.30449605e+00 -4.04682279e-01 -1.24799542e-01 6.01970196e-01
-1.02807832e+00 1.71589516e-02 -1.30668223e-01 -1.24540305e+00
2.46267512e-01 -1.37448347e+00 7.17904329e-01 2.74712682e-01
-2.87268758e-01 -7.69532144e-01 2.73156583e-01 4.17238250e-02
4.60911363e-01 3.09620559e-01 8.13637435e-01 -8.47620189e-01
-7.97930598e-01 -1.68893710e-01 1.82211474e-01 -1.97087064e-01
-3.35887164e-01 1.54592857e-01 -7.38580883e-01 1.11776561e-01
-4.95236039e-01 1.62277505e-01 6.55231774e-01 3.55792940e-01
3.46513122e-01 -5.20357430e-01 -9.16872263e-01 8.33311155e-02
1.20223010e+00 4.95575756e-01 5.13269544e-01 4.50227976e-01
4.46671188e-01 9.52287078e-01 9.35385704e-01 4.71424848e-01
1.09033740e+00 9.65176404e-01 7.44950831e-01 6.27127767e-01
3.97770666e-02 -4.31651652e-01 7.86008179e-01 2.74711251e-01
-1.58264652e-01 -3.39752495e-01 -1.12817276e+00 4.52859849e-01
-2.82958555e+00 -1.25931466e+00 -4.96951133e-01 2.07661080e+00
4.11981463e-01 1.34014413e-01 5.32936215e-01 -8.29656571e-02
3.56636763e-01 -4.60172296e-02 -5.03704429e-01 -2.40034804e-01
1.29974678e-01 -5.12238324e-01 6.10968590e-01 7.55085170e-01
-1.00764978e+00 1.07728171e+00 6.64542055e+00 9.35717404e-01
-7.05332339e-01 1.83663040e-01 8.48758668e-02 1.28582552e-01
-5.16743958e-01 1.96410462e-01 -8.89234781e-01 5.80861449e-01
1.42316222e+00 -2.14969531e-01 4.86726910e-01 8.45133662e-01
5.38555145e-01 -5.28835118e-01 -1.30829227e+00 5.04177213e-01
-3.29644233e-01 -1.69052041e+00 -7.53787756e-02 4.25440818e-02
8.20428848e-01 4.99819487e-01 -6.97567910e-02 3.19937199e-01
7.30761528e-01 -9.43935990e-01 1.14991593e+00 1.09342909e+00
1.33896783e-01 -8.96774471e-01 9.11213756e-01 7.68876433e-01
-1.09205866e+00 -4.30544049e-01 4.55244705e-02 -1.77291110e-02
7.33663082e-01 1.68510795e-01 -1.07051241e+00 9.48327482e-01
2.91655898e-01 7.56684780e-01 -6.98251799e-02 1.28146970e+00
-2.71288842e-01 6.52595937e-01 -4.44511712e-01 -1.28712207e-01
4.97025967e-01 -2.48786122e-01 1.08880329e+00 1.13013196e+00
2.67201662e-01 -3.67814675e-02 3.62343580e-01 6.78588092e-01
5.86490571e-01 -3.48803669e-01 -7.14535832e-01 3.73732358e-01
8.10601115e-01 1.07283318e+00 -3.00788492e-01 -4.44288880e-01
-3.74261290e-01 8.87840465e-02 5.48332743e-02 3.38421911e-01
-9.55005765e-01 3.65340739e-01 9.60925162e-01 -3.76206562e-02
3.79303724e-01 -5.06662488e-01 -2.72841632e-01 -4.44184184e-01
-3.28247875e-01 -5.50008833e-01 1.46557033e-01 -6.09621346e-01
-8.22335839e-01 6.71021104e-01 4.75523204e-01 -1.34371042e+00
-7.39864945e-01 -3.67417395e-01 -5.38196206e-01 4.52002048e-01
-1.50319755e+00 -1.29937983e+00 -1.05445929e-01 4.81813043e-01
3.88286740e-01 -5.76560915e-01 6.91869736e-01 6.65329471e-02
-3.74731153e-01 -1.22598512e-02 5.36206998e-02 -6.57529533e-01
2.43132919e-01 -9.45084035e-01 6.64015114e-02 8.84284496e-01
-1.27003536e-01 1.70704219e-02 1.15012968e+00 -9.45197940e-01
-1.79917467e+00 -1.31116438e+00 9.67785776e-01 -6.12078369e-01
5.99473119e-01 3.55926752e-01 -5.26089489e-01 7.46079862e-01
1.75502881e-01 -4.34839606e-01 3.19649547e-01 -1.63186416e-01
5.40750138e-02 -3.54352176e-01 -1.07724071e+00 6.18472576e-01
6.37435496e-01 9.33178067e-02 -4.29712623e-01 1.41838491e-01
4.82650131e-01 -4.05499905e-01 -6.42991185e-01 5.94096959e-01
5.56822121e-01 -1.05530298e+00 6.85267568e-01 -7.30842173e-01
1.28982693e-01 -7.84613192e-01 -3.57311308e-01 -1.10936880e+00
-3.90146703e-01 -7.23610282e-01 -5.81990659e-01 9.62528944e-01
4.76115882e-01 -7.45230556e-01 7.25023925e-01 7.96009600e-01
-5.53891957e-01 -7.37180829e-01 -1.30329597e+00 -7.87133634e-01
-3.11581820e-01 -1.01712942e+00 8.61688077e-01 4.60186712e-02
3.00042421e-01 -3.11678257e-02 -5.99133790e-01 3.46611887e-01
6.50202274e-01 -1.77649975e-01 1.09318697e+00 -1.09894633e+00
3.30249369e-02 -7.14006543e-01 -5.70796669e-01 -8.92655730e-01
5.25563717e-01 -6.98385239e-01 4.93879706e-01 -1.67141318e+00
-2.11885944e-02 -6.01507425e-01 -2.66051572e-02 6.76432669e-01
1.27780050e-01 -3.22896421e-01 3.69431734e-01 2.69842058e-01
-1.14197874e+00 7.06277311e-01 6.92193568e-01 -7.51063898e-02
-3.50063145e-01 4.18914974e-01 -1.70148998e-01 7.86733866e-01
7.53269732e-01 -4.87466186e-01 -4.25629914e-01 1.85214425e-03
1.29662141e-01 4.05843765e-01 3.90538275e-01 -1.19856596e+00
8.97984743e-01 -7.13580728e-01 -2.31197461e-01 -1.07320952e+00
5.33578038e-01 -1.05891395e+00 7.40478575e-01 4.74842995e-01
-1.10528946e-01 -3.99989933e-02 2.27008581e-01 8.48660827e-01
-3.63181420e-02 -5.84995300e-02 2.51591653e-01 4.36478645e-01
-1.23426294e+00 2.08592057e-01 -1.15722620e+00 -6.89532638e-01
1.53111506e+00 -1.57471120e-01 -3.77137125e-01 -6.41426861e-01
-4.50375736e-01 7.65536666e-01 3.45497131e-01 4.29849267e-01
8.06961298e-01 -1.30702412e+00 -7.72778273e-01 2.98584811e-02
-1.54007189e-02 -2.96962857e-01 1.16455704e-01 1.14464116e+00
-1.59387410e-01 7.03432024e-01 -2.44616300e-01 -7.00171173e-01
-1.00295937e+00 5.33905149e-01 1.99928299e-01 -1.73994139e-01
-4.73355353e-01 2.46133059e-01 -4.89655018e-01 -4.56884921e-01
4.27197099e-01 -5.61253773e-03 -3.77088398e-01 -2.10971057e-01
4.59352195e-01 1.03479505e+00 -1.29809543e-01 -9.08400893e-01
-6.38291538e-01 1.46661967e-01 3.83126765e-01 -4.13013697e-01
1.26391518e+00 -1.68366492e-01 1.15457602e-01 4.02864903e-01
3.72489214e-01 -3.38175446e-01 -1.65545762e+00 -5.77231571e-02
2.64357507e-01 -1.73659369e-01 -9.16007459e-02 -8.61635149e-01
-6.13948524e-01 2.81686246e-01 1.40070200e-01 4.02982906e-02
7.24295437e-01 1.18717559e-01 7.23868310e-01 4.42286938e-01
9.83286560e-01 -1.00803339e+00 -4.02907133e-01 7.23552167e-01
5.40504336e-01 -1.10432959e+00 -1.73810795e-01 -1.61530674e-01
-9.44735587e-01 1.08102190e+00 5.62138438e-01 1.59010112e-01
5.31855524e-01 2.41281539e-01 -2.21181482e-01 -2.39015128e-02
-1.41390443e+00 -3.84862900e-01 4.73403573e-01 9.62461233e-01
-4.11984980e-01 4.82535660e-01 -2.14108691e-01 7.26741612e-01
1.07735045e-01 1.45889699e-01 1.87981367e-01 7.61495113e-01
-8.10712576e-01 -1.35014391e+00 -3.48818541e-01 3.89376998e-01
5.01465797e-01 5.25022089e-01 1.46814091e-02 5.83611608e-01
3.80326241e-01 1.52877891e+00 -6.91731796e-02 -1.07748902e+00
3.12398940e-01 9.94687248e-03 1.52487591e-01 -1.07929721e-01
-4.47477728e-01 -3.96276772e-01 6.08738482e-01 -1.06139433e+00
-4.01818395e-01 -1.17027271e+00 -1.56333697e+00 -7.01418042e-01
1.63525924e-01 2.27927357e-01 6.20791674e-01 1.40596330e+00
5.20764053e-01 3.71640354e-01 5.22569060e-01 -1.08925390e+00
-4.54893053e-01 -5.47555804e-01 9.80226137e-03 -2.63183236e-01
3.41779888e-01 -9.62271333e-01 2.00572208e-01 -2.22264633e-01] | [5.725086212158203, 1.0311616659164429] |
b033c99e-1c92-48e3-8e1a-ddabe503e336 | domain-adaptive-semantic-segmentation-with-1 | 2110.05170 | null | https://arxiv.org/abs/2110.05170v3 | https://arxiv.org/pdf/2110.05170v3.pdf | Domain Adaptive Semantic Segmentation via Regional Contrastive Consistency Regularization | Unsupervised domain adaptation (UDA) for semantic segmentation has been well-studied in recent years. However, most existing works largely neglect the local regional consistency across different domains and are less robust to changes in outdoor environments. In this paper, we propose a novel and fully end-to-end trainable approach, called regional contrastive consistency regularization (RCCR) for domain adaptive semantic segmentation. Our core idea is to pull the similar regional features extracted from the same location of different images, i.e., the original image and augmented image, to be closer, and meanwhile push the features from the different locations of the two images to be separated. We innovatively propose a region-wise contrastive loss with two sampling strategies to realize effective regional consistency. Besides, we present momentum projection heads, where the teacher projection head is the exponential moving average of the student. Finally, a memory bank mechanism is designed to learn more robust and stable region-wise features under varying environments. Extensive experiments on two common UDA benchmarks, i.e., GTAV to Cityscapes and SYNTHIA to Cityscapes, demonstrate that our approach outperforms the state-of-the-art methods. | ['Lizhuang Ma', 'Ran Yi', 'Xuequan Lu', 'Chuyun Zhuang', 'Qianyu Zhou'] | 2021-10-11 | null | null | null | null | ['synthetic-to-real-translation'] | ['computer-vision'] | [-3.33931856e-02 -2.51074135e-01 -6.93248883e-02 -6.11304224e-01
-8.57422531e-01 -3.34150136e-01 2.47457013e-01 -2.38960698e-01
-6.02510214e-01 6.18404150e-01 -2.54215933e-02 1.79144755e-01
-1.60072371e-01 -4.49452370e-01 -7.53623426e-01 -9.71384466e-01
4.22580272e-01 3.39221954e-01 6.54932439e-01 -3.91885675e-02
-1.68365482e-02 1.01341248e-01 -1.06267536e+00 -5.04655503e-02
1.22662914e+00 9.73957479e-01 6.49817050e-01 1.24045126e-01
-8.99804011e-03 3.58348668e-01 -3.91969621e-01 -8.18406418e-02
2.99639761e-01 -4.20098841e-01 -8.86585534e-01 3.65837038e-01
3.69403422e-01 -1.71576217e-01 -2.45490402e-01 1.30661547e+00
5.37999749e-01 5.12766302e-01 4.24437076e-01 -9.73142803e-01
-8.57353806e-01 3.84467244e-01 -9.62933719e-01 2.56954551e-01
-2.00696707e-01 1.47381365e-01 8.23895454e-01 -8.55042577e-01
7.33037710e-01 1.23733675e+00 5.31481385e-01 5.12251318e-01
-1.06618786e+00 -6.47343218e-01 8.48247528e-01 1.88039467e-01
-1.28034413e+00 -4.31158692e-02 9.50333893e-01 -1.58449113e-01
4.66303557e-01 -2.71533597e-02 3.61811131e-01 9.39376771e-01
-1.38587505e-01 1.14510977e+00 1.21042132e+00 -1.84001714e-01
1.73378602e-01 3.20823252e-01 8.70895237e-02 5.22404671e-01
7.19685853e-02 -3.17631930e-01 -2.50315785e-01 1.78178340e-01
8.30278158e-01 2.22689882e-01 -2.91262001e-01 -6.94226861e-01
-1.04182732e+00 7.87993729e-01 5.08121848e-01 2.53154963e-01
-3.72978270e-01 -2.28587210e-01 4.77606088e-01 3.33986133e-02
6.78511918e-01 7.94026703e-02 -8.60333323e-01 8.61649290e-02
-7.23679483e-01 2.53738910e-01 2.77039230e-01 1.02766025e+00
7.30045319e-01 -1.10833243e-01 -2.35053256e-01 1.24697828e+00
4.08902913e-01 3.97688806e-01 6.88114047e-01 -8.64786863e-01
6.12796187e-01 5.22334993e-01 6.69127628e-02 -8.44219744e-01
-2.74521738e-01 -5.90247691e-01 -7.13209391e-01 -9.48532000e-02
5.11212349e-01 -2.16939211e-01 -1.14052951e+00 1.85225272e+00
5.93694329e-01 5.67252874e-01 2.31244460e-01 1.27318883e+00
8.17912698e-01 6.83076203e-01 1.89835712e-01 1.15275465e-01
1.16340411e+00 -1.41056669e+00 -4.54516411e-01 -5.26717901e-01
3.37837845e-01 -5.92230618e-01 1.25619805e+00 2.50118464e-01
-1.20058954e+00 -7.02561796e-01 -9.54758763e-01 -1.78128302e-01
-2.75828630e-01 1.07312039e-01 3.30677658e-01 2.13387549e-01
-5.51785827e-01 5.97465277e-01 -9.31894302e-01 -3.13291281e-01
6.10107124e-01 1.13103062e-01 -1.33548126e-01 -1.24256410e-01
-1.14269638e+00 4.47749257e-01 4.97208714e-01 3.92463990e-02
-6.83373213e-01 -7.24907696e-01 -8.44969451e-01 -1.48294598e-01
5.66416681e-01 -4.95127052e-01 1.14557564e+00 -1.24371970e+00
-1.62933862e+00 9.68110800e-01 -9.31629315e-02 -4.71329242e-01
7.22388446e-01 -3.89979839e-01 -2.77850211e-01 8.09608847e-02
4.27044034e-01 7.50096500e-01 7.70595968e-01 -1.34700501e+00
-8.42325568e-01 -5.67674696e-01 -2.57084250e-01 5.68075240e-01
-4.15194392e-01 2.79850326e-02 -9.61794257e-01 -8.48355651e-01
4.14097309e-01 -8.85837257e-01 -4.33985621e-01 -1.00984581e-01
-2.96547145e-01 -2.12827742e-01 9.48824048e-01 -7.97941327e-01
9.60161328e-01 -2.38775849e+00 2.04937547e-01 1.89896807e-01
1.26975793e-02 4.16834712e-01 -1.87846035e-01 -2.88957536e-01
-7.49541968e-02 -1.57839552e-01 -5.67520678e-01 -3.68818372e-01
-1.06470950e-01 3.00294310e-01 -2.22928658e-01 4.51883882e-01
2.43897408e-01 7.84581900e-01 -1.02851832e+00 -5.98925531e-01
1.65671691e-01 3.59567195e-01 -4.31932509e-01 5.20958975e-02
-1.86091244e-01 7.67821670e-01 -8.61881137e-01 3.96181792e-01
1.05217743e+00 -2.96027929e-01 -1.40729815e-01 2.90339757e-02
-6.09879009e-02 1.54514596e-01 -1.32483947e+00 1.97932100e+00
-2.56918877e-01 2.55996943e-01 1.46956623e-01 -1.17707431e+00
1.06094480e+00 -3.18730175e-02 3.12150180e-01 -1.11166406e+00
3.59286927e-02 2.71959424e-01 -3.01338583e-01 -3.04000467e-01
5.76419890e-01 1.25793358e-02 -1.37979731e-01 7.03858882e-02
-1.73543151e-02 -2.20753886e-02 -6.09562881e-02 2.27503711e-03
4.42030787e-01 3.10337335e-01 -3.04348152e-02 -2.25000113e-01
6.92175448e-01 -1.09921843e-01 1.14750314e+00 5.20165086e-01
-5.97977579e-01 9.36938941e-01 2.59104520e-01 -3.35227966e-01
-9.15076554e-01 -1.08024752e+00 -1.68615147e-01 1.13672888e+00
9.06088531e-01 8.14739838e-02 -8.94840837e-01 -9.88529861e-01
-1.53014883e-01 5.06128550e-01 -4.59767193e-01 -1.88018933e-01
-7.53751636e-01 -6.98243320e-01 2.38145530e-01 8.60462964e-01
1.07693589e+00 -1.04273236e+00 -4.13154900e-01 2.25632727e-01
-2.38394678e-01 -1.23507702e+00 -8.04166913e-01 6.68997914e-02
-9.51793432e-01 -5.74189782e-01 -1.25448203e+00 -1.20703375e+00
5.99280477e-01 4.81667012e-01 9.06375468e-01 -3.79260063e-01
5.99785941e-03 3.44952531e-02 -3.05171698e-01 -2.18842342e-01
1.55700237e-01 1.33688226e-01 -1.78419411e-01 1.12609334e-01
1.79142252e-01 -3.70030403e-01 -8.04026783e-01 5.96599877e-01
-8.01072836e-01 -4.34485003e-02 5.54570794e-01 9.88594353e-01
1.29599333e+00 1.30270064e-01 5.71604431e-01 -1.08912992e+00
4.12994474e-01 -5.13656676e-01 -4.70874280e-01 2.53447831e-01
-4.38338667e-01 -7.88442269e-02 8.34491909e-01 -5.84540665e-01
-1.37283719e+00 5.58377355e-02 -3.05568147e-02 -6.50351942e-01
-2.87208587e-01 2.14192837e-01 -4.80234563e-01 1.56069770e-01
3.13076794e-01 4.02372986e-01 -2.55807936e-01 -4.33148503e-01
4.43713546e-01 4.96162981e-01 6.62848830e-01 -6.52571976e-01
6.71192765e-01 7.40949810e-01 -5.53844690e-01 -6.28102660e-01
-9.51325178e-01 -8.00742984e-01 -5.27539551e-01 1.56994209e-01
1.00324023e+00 -1.06054986e+00 -3.72114987e-03 8.95890474e-01
-8.83585691e-01 -4.65426326e-01 -3.61634076e-01 4.79014128e-01
-5.04445553e-01 3.35804671e-01 -4.48887765e-01 -3.95101696e-01
-3.64472568e-01 -1.17301428e+00 1.17327654e+00 9.36122894e-01
2.10155323e-01 -1.07938766e+00 2.30618920e-02 3.98155481e-01
1.63298383e-01 1.07975096e-01 5.37289202e-01 -6.55000687e-01
-4.64506507e-01 9.99865755e-02 -5.69538951e-01 6.66525126e-01
1.34144917e-01 -1.84286371e-01 -8.23325932e-01 -2.96862215e-01
-1.20274816e-02 -2.32507661e-01 1.11541057e+00 7.04278886e-01
1.32823217e+00 -9.70589966e-02 -5.06432831e-01 7.68083632e-01
1.25112641e+00 2.21852720e-01 5.62231958e-01 5.36269367e-01
7.73917615e-01 4.83312726e-01 1.15603459e+00 3.28612715e-01
4.43054467e-01 6.44216776e-01 1.95748240e-01 -4.55973983e-01
-1.00042902e-01 -1.56308085e-01 1.02076061e-01 7.91135430e-01
8.56428668e-02 1.39029799e-02 -5.14430285e-01 8.16264153e-01
-2.03949499e+00 -7.43042231e-01 8.81262403e-03 2.03711653e+00
7.79368818e-01 3.50437671e-01 2.46408328e-01 -4.29397732e-01
9.78420198e-01 2.84810454e-01 -1.00236523e+00 -1.27757877e-01
-3.08339298e-01 7.60383606e-02 5.96071661e-01 1.34030476e-01
-1.37674642e+00 1.21850514e+00 4.85754490e+00 1.13246262e+00
-1.19160593e+00 2.06039771e-01 9.46031034e-01 6.14319229e-04
-1.81175366e-01 1.25000961e-02 -6.08695984e-01 7.69971728e-01
3.68006855e-01 4.10796516e-02 1.63001761e-01 1.15097737e+00
1.84725538e-01 7.06216181e-03 -5.48759639e-01 7.99672425e-01
-9.25775990e-02 -9.57746148e-01 -2.52833784e-01 -3.00553650e-01
1.19838381e+00 8.90864208e-02 4.45586354e-01 3.35675657e-01
4.23426360e-01 -6.81275129e-01 8.41068387e-01 4.46956992e-01
4.15806323e-01 -8.97519171e-01 7.13713348e-01 2.15845600e-01
-1.27224803e+00 -2.43873727e-02 -5.60148597e-01 3.83931667e-01
5.02009168e-02 5.09064317e-01 -3.99990052e-01 6.94819033e-01
1.13873875e+00 9.24687147e-01 -5.49351990e-01 9.37647521e-01
-2.24734440e-01 6.12988651e-01 -2.17267960e-01 7.67458528e-02
6.67553067e-01 -4.54269856e-01 5.80546439e-01 1.08827424e+00
3.98453698e-02 3.74778993e-02 3.03653181e-01 8.56879056e-01
-1.08016647e-01 1.02586538e-01 -1.17327899e-01 4.04426247e-01
4.84054118e-01 1.15305054e+00 -8.69978487e-01 -2.65991807e-01
-4.96237308e-01 1.22346497e+00 4.22566593e-01 5.69914639e-01
-1.03895640e+00 -4.82636958e-01 6.34725928e-01 -1.52260393e-01
6.80970609e-01 -1.44890979e-01 -4.45392013e-01 -1.13721907e+00
2.70951092e-01 -6.39504611e-01 5.61614871e-01 -6.27798617e-01
-1.42678976e+00 4.68488395e-01 -6.71466514e-02 -1.28642929e+00
2.95259416e-01 -2.75620759e-01 -8.34655106e-01 1.01894546e+00
-1.77799428e+00 -1.12207949e+00 -3.43481153e-01 6.88920677e-01
7.59771228e-01 4.18014564e-02 2.40920290e-01 3.57093006e-01
-9.70674753e-01 6.96373045e-01 5.28156459e-01 2.50250250e-01
8.93942893e-01 -1.28199744e+00 4.60909039e-01 9.12637830e-01
-4.88534234e-02 5.36676049e-01 4.33234602e-01 -5.72547734e-01
-7.62301683e-01 -1.45923376e+00 3.69295299e-01 -8.92374665e-02
4.66718256e-01 -2.19136611e-01 -1.33172357e+00 5.15055656e-01
3.10508963e-02 2.92603999e-01 1.56690687e-01 -1.51260840e-02
-3.09398323e-01 -3.52815241e-01 -1.18544638e+00 6.80452347e-01
9.35645938e-01 -2.58818775e-01 -5.54145455e-01 2.77686089e-01
9.50102746e-01 -6.92306876e-01 -5.73272407e-01 4.81660277e-01
2.91022182e-01 -8.01103830e-01 9.64866042e-01 -5.68370521e-01
3.54243904e-01 -5.09119570e-01 6.86369017e-02 -1.39704311e+00
-2.93803573e-01 -3.52747858e-01 3.05162042e-01 1.48023140e+00
3.68020713e-01 -8.51031363e-01 9.02777791e-01 4.30838853e-01
-2.83049315e-01 -7.63386130e-01 -8.32107186e-01 -9.45014179e-01
4.13964659e-01 -1.33747622e-01 4.92482275e-01 1.04354453e+00
-4.80474144e-01 2.34437555e-01 -1.41417503e-01 3.28015506e-01
4.32689548e-01 3.20179373e-01 6.12878144e-01 -1.04160941e+00
-2.07596481e-01 -4.75018829e-01 -1.76674888e-01 -1.33569086e+00
1.36228234e-01 -8.19635928e-01 3.26248795e-01 -1.46534026e+00
3.72497916e-01 -6.80601895e-01 -6.09769285e-01 3.04991543e-01
-6.02828324e-01 -2.15061381e-02 1.02423169e-02 2.48378694e-01
-1.00763571e+00 7.73047745e-01 1.58800018e+00 -1.57789573e-01
-5.55695117e-01 1.29176220e-02 -7.12376475e-01 9.78132486e-01
9.51792657e-01 -5.47932565e-01 -4.48290199e-01 -5.99229157e-01
-3.98729891e-01 -3.36465836e-01 3.71746331e-01 -8.83532941e-01
1.81541219e-01 -2.50109017e-01 4.37075108e-01 -5.71332276e-01
5.10633104e-02 -7.53257215e-01 -3.77550572e-01 1.46884590e-01
-2.78876364e-01 -2.27409109e-01 3.07886869e-01 7.70580828e-01
-2.92300642e-01 -1.23045653e-01 1.21253228e+00 -1.28476530e-01
-1.03172255e+00 4.83530790e-01 2.07643896e-01 5.15430093e-01
1.11799896e+00 -2.95750439e-01 -1.22930624e-01 -9.18801036e-03
-5.52870691e-01 6.71400845e-01 4.13531572e-01 4.61386710e-01
5.68853676e-01 -1.16695046e+00 -7.46301413e-01 1.62654016e-02
1.32079661e-01 6.10363007e-01 7.62960970e-01 7.88462579e-01
-3.29770505e-01 1.89948589e-01 -1.45937651e-01 -7.35034943e-01
-1.22502935e+00 4.36984003e-01 3.75476718e-01 -4.21596467e-01
-8.38077307e-01 1.17492187e+00 5.34829557e-01 -6.83636963e-01
2.33448014e-01 -3.29183578e-01 -7.03588501e-02 -1.47348180e-01
3.67365330e-01 2.80030698e-01 -6.36656731e-02 -5.94947815e-01
-4.20571327e-01 8.08958471e-01 -5.64056098e-01 2.33940370e-02
1.24301565e+00 -3.36680263e-01 2.34232694e-01 2.63104498e-01
1.25806999e+00 -1.27714172e-01 -1.85983825e+00 -6.16596043e-01
-1.28423303e-01 -5.25691271e-01 -4.48935889e-02 -7.46012986e-01
-1.44095862e+00 8.33730578e-01 7.57158995e-01 -9.58315060e-02
1.30260599e+00 1.95391372e-01 1.16612220e+00 3.96999381e-02
6.74281195e-02 -1.61164045e+00 1.04260296e-01 5.44693351e-01
4.81303722e-01 -1.34247410e+00 -9.14619416e-02 -3.58713835e-01
-1.14440870e+00 6.30721807e-01 8.11314046e-01 -2.30224609e-01
4.08172727e-01 -1.62140220e-01 1.29981682e-01 7.89112076e-02
-3.23161036e-01 -3.39805961e-01 2.06482589e-01 6.93177342e-01
7.58194551e-02 4.51846756e-02 -2.09125370e-01 1.02194393e+00
2.36724824e-01 -2.59063751e-01 8.14362019e-02 8.57359231e-01
-2.91508615e-01 -7.75292516e-01 -3.11638474e-01 1.98863000e-01
-4.40674454e-01 7.36290365e-02 -1.87483072e-01 8.20278406e-01
3.51170421e-01 7.50283360e-01 1.05020888e-01 -9.28046703e-02
5.03053248e-01 -2.08305389e-01 1.76003590e-01 -4.39892143e-01
-2.61763215e-01 2.89245605e-01 -3.80784154e-01 -3.89563501e-01
-4.71830517e-01 -9.48163569e-01 -1.69537497e+00 1.66396141e-01
-3.12871903e-01 7.40198418e-02 4.54911739e-01 9.35613513e-01
2.44693071e-01 6.52609825e-01 7.79091895e-01 -5.79041779e-01
-3.78004938e-01 -7.40326941e-01 -7.64585733e-01 5.86167455e-01
1.19851977e-01 -6.35184944e-01 -8.82053599e-02 8.86548012e-02] | [9.65636157989502, 1.3744609355926514] |
144953cb-429e-436a-99c3-8a968a9a9f45 | video-probabilistic-diffusion-models-in | 2302.07685 | null | https://arxiv.org/abs/2302.07685v2 | https://arxiv.org/pdf/2302.07685v2.pdf | Video Probabilistic Diffusion Models in Projected Latent Space | Despite the remarkable progress in deep generative models, synthesizing high-resolution and temporally coherent videos still remains a challenge due to their high-dimensionality and complex temporal dynamics along with large spatial variations. Recent works on diffusion models have shown their potential to solve this challenge, yet they suffer from severe computation- and memory-inefficiency that limit the scalability. To handle this issue, we propose a novel generative model for videos, coined projected latent video diffusion models (PVDM), a probabilistic diffusion model which learns a video distribution in a low-dimensional latent space and thus can be efficiently trained with high-resolution videos under limited resources. Specifically, PVDM is composed of two components: (a) an autoencoder that projects a given video as 2D-shaped latent vectors that factorize the complex cubic structure of video pixels and (b) a diffusion model architecture specialized for our new factorized latent space and the training/sampling procedure to synthesize videos of arbitrary length with a single model. Experiments on popular video generation datasets demonstrate the superiority of PVDM compared with previous video synthesis methods; e.g., PVDM obtains the FVD score of 639.7 on the UCF-101 long video (128 frames) generation benchmark, which improves 1773.4 of the prior state-of-the-art. | ['Jinwoo Shin', 'Subin Kim', 'Kihyuk Sohn', 'Sihyun Yu'] | 2023-02-15 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Yu_Video_Probabilistic_Diffusion_Models_in_Projected_Latent_Space_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Yu_Video_Probabilistic_Diffusion_Models_in_Projected_Latent_Space_CVPR_2023_paper.pdf | cvpr-2023-1 | ['video-generation'] | ['computer-vision'] | [-6.93147108e-02 -3.17758560e-01 -9.64249820e-02 1.71018466e-01
-6.53456926e-01 -3.76814216e-01 8.99748385e-01 -1.08468032e+00
8.89975727e-02 8.33046198e-01 6.32721186e-01 1.02335811e-01
1.54178709e-01 -8.15315545e-01 -9.44560051e-01 -1.20965922e+00
1.05062090e-01 2.98074961e-01 1.46584809e-01 1.58185214e-01
-1.28058478e-01 2.84918755e-01 -1.46539307e+00 3.52663130e-01
7.94488907e-01 9.10308838e-01 4.11638796e-01 7.73242295e-01
6.51434138e-02 9.92251277e-01 -5.91862261e-01 -2.18876079e-01
2.15117663e-01 -7.76719630e-01 -3.02991837e-01 4.84842509e-01
4.26794201e-01 -8.78944516e-01 -1.06439233e+00 8.49476457e-01
4.64523137e-01 1.61843985e-01 1.07478118e+00 -1.35137248e+00
-1.24676764e+00 4.94329095e-01 -5.69449782e-01 2.18245462e-01
1.87012002e-01 3.32363993e-01 6.69633567e-01 -1.08104467e+00
1.01691675e+00 1.33596122e+00 3.21543097e-01 8.69189858e-01
-1.36542749e+00 -5.22515416e-01 1.19703777e-01 1.73604429e-01
-1.53771973e+00 -3.86866897e-01 7.17456520e-01 -8.14648628e-01
1.01595533e+00 -1.39998466e-01 7.55531967e-01 1.83825755e+00
3.77771527e-01 8.27044725e-01 7.17145920e-01 1.25053376e-01
2.12777957e-01 -3.58173043e-01 -6.69757128e-01 4.97757077e-01
5.07813320e-02 1.27230868e-01 -6.05098426e-01 -9.93056446e-02
1.45220816e+00 1.23057894e-01 -4.30629402e-01 -4.20117229e-01
-1.33961248e+00 9.76151645e-01 1.65676117e-01 1.84128657e-01
-7.68227279e-01 3.32230985e-01 1.13091059e-01 7.66381398e-02
4.73499060e-01 3.06588430e-02 6.70309812e-02 -2.54446685e-01
-1.13977325e+00 5.77977657e-01 5.29565096e-01 1.13324022e+00
3.47466648e-01 6.22125983e-01 -5.24950504e-01 6.30867004e-01
3.62042427e-01 6.41235232e-01 6.21032178e-01 -1.17993033e+00
4.66500074e-01 1.68824583e-01 1.31668523e-01 -1.04803717e+00
1.97078899e-01 -2.89814115e-01 -1.33217895e+00 -1.39447674e-01
2.20669191e-02 -2.83845872e-01 -9.29652214e-01 1.77550638e+00
1.81471154e-01 4.42195266e-01 1.95193499e-01 9.08001959e-01
7.85898566e-01 1.38180006e+00 -1.51231149e-02 -4.48038757e-01
9.71823871e-01 -1.36783218e+00 -9.13207948e-01 1.33634865e-01
-1.05977297e-01 -7.91337311e-01 6.39127135e-01 3.32502276e-01
-1.37846303e+00 -8.13219011e-01 -9.93682206e-01 1.63732860e-02
2.15453222e-01 1.78394347e-01 4.72187489e-01 1.90220073e-01
-1.35084105e+00 4.00763541e-01 -9.35277641e-01 4.21409905e-02
4.10040438e-01 -5.72393760e-02 -1.65721983e-01 -3.02763462e-01
-1.09732687e+00 2.51107723e-01 2.68451571e-01 -5.13838753e-02
-1.52195048e+00 -8.05338264e-01 -8.45576584e-01 1.68329865e-01
3.60865206e-01 -1.11577213e+00 9.94778395e-01 -6.49201334e-01
-1.76159000e+00 2.30122045e-01 -1.58145651e-01 -4.64984447e-01
8.02936196e-01 -2.56648004e-01 -4.72923726e-01 3.14146250e-01
1.45669147e-01 9.21555996e-01 1.44142175e+00 -1.09502244e+00
-4.73126829e-01 1.71380833e-01 -2.71578074e-01 2.02966169e-01
-3.38023394e-01 -2.62083411e-01 -8.28787386e-01 -1.17685723e+00
-1.68500721e-01 -1.08258843e+00 -1.99013039e-01 -1.39102146e-01
-1.80652589e-01 -2.35083506e-01 1.02243161e+00 -7.68120050e-01
1.45509219e+00 -1.99611151e+00 8.45401466e-01 -3.59044731e-01
3.57977688e-01 2.94046998e-01 -2.95989245e-01 4.48776633e-01
1.34093541e-04 1.75014511e-02 -3.47226746e-02 -3.18033189e-01
8.52006525e-02 1.58838496e-01 -6.40841484e-01 1.66807458e-01
3.37204307e-01 1.05164766e+00 -8.88654232e-01 -2.51395583e-01
2.09960133e-01 1.01006186e+00 -8.04572403e-01 5.00899017e-01
-3.79413486e-01 5.28528392e-01 -2.98233718e-01 4.41070497e-01
5.62787414e-01 -5.51534951e-01 1.80145800e-01 -1.91156968e-01
1.22775316e-01 -1.46631286e-01 -1.06114852e+00 1.91536152e+00
-8.44989270e-02 5.76647520e-01 -1.87855050e-01 -6.29933417e-01
6.93458319e-01 5.70797920e-01 6.77188218e-01 -4.55185682e-01
-1.48244435e-02 6.60224184e-02 -3.81138802e-01 -5.32270312e-01
5.72698593e-01 -6.02580756e-02 1.58198938e-01 4.44288641e-01
2.87342250e-01 2.14160047e-02 4.58133966e-01 4.02625740e-01
1.01422286e+00 5.14098465e-01 -1.65587202e-01 -2.57887058e-02
3.21280450e-01 -4.85606462e-01 5.37476182e-01 4.31523949e-01
1.80551037e-01 7.97262967e-01 5.26207745e-01 -4.18468535e-01
-1.45616722e+00 -1.32229960e+00 2.59665489e-01 5.76238811e-01
-5.82379885e-02 -6.62728131e-01 -9.49403763e-01 -3.92623842e-01
-2.25120574e-01 5.24869442e-01 -4.72914696e-01 -2.06464872e-01
-7.46763229e-01 -8.54999185e-01 2.64364988e-01 6.10074878e-01
4.32971179e-01 -9.66338754e-01 -2.80907810e-01 3.69694382e-01
-5.05416691e-01 -1.38607085e+00 -7.31532216e-01 -5.39041817e-01
-7.75632143e-01 -6.22557521e-01 -1.39067900e+00 -7.26339340e-01
6.53644025e-01 4.27198708e-01 1.15152872e+00 -3.71014595e-01
-1.37283564e-01 2.60487914e-01 -3.32297385e-01 1.95367977e-01
-6.12699270e-01 -5.42469472e-02 2.44102016e-01 1.98244423e-01
4.28166017e-02 -6.63445234e-01 -8.27750206e-01 3.80783379e-01
-1.22119927e+00 4.38269228e-01 5.91674805e-01 9.87528801e-01
7.96859980e-01 2.60144770e-01 3.66143376e-01 -3.94643307e-01
6.64966166e-01 -8.07953119e-01 -6.46079898e-01 1.00127138e-01
-3.47646922e-01 6.83617890e-02 5.98369181e-01 -7.31217921e-01
-1.15847230e+00 -1.20257720e-01 2.57104971e-02 -1.07424879e+00
1.45472139e-01 1.91947937e-01 -1.27486750e-01 3.40510130e-01
4.49201733e-01 6.74294174e-01 -2.96808854e-02 -3.20112795e-01
5.30179977e-01 2.66661048e-01 5.37644029e-01 -5.83315849e-01
1.02753007e+00 3.67344379e-01 -6.49061352e-02 -7.81856120e-01
-5.08906245e-01 8.51613581e-02 -4.13259685e-01 -1.95806295e-01
1.17878926e+00 -1.33927512e+00 -2.92857617e-01 7.93391764e-01
-1.10548878e+00 -4.24753845e-01 -2.04852775e-01 5.32387257e-01
-8.51497829e-01 3.99397492e-01 -9.75257516e-01 -4.34302062e-01
-2.83397287e-01 -1.34224117e+00 1.11780238e+00 1.99580327e-01
-6.94742575e-02 -8.21324348e-01 1.39740720e-01 1.81653619e-01
5.59153736e-01 3.09755474e-01 7.99196541e-01 2.52075762e-01
-1.16279221e+00 1.51387200e-01 -5.58770038e-02 5.42796671e-01
4.83682603e-02 3.23593318e-01 -5.81116855e-01 -4.54275280e-01
1.18419945e-01 -1.84962690e-01 8.99298191e-01 6.88310862e-01
1.04489768e+00 -4.54594612e-01 -2.19166785e-01 8.37953210e-01
1.35516739e+00 2.68183291e-01 7.30726957e-01 -1.11227281e-01
8.38877797e-01 8.93687233e-02 2.79214472e-01 6.28665566e-01
3.16934943e-01 8.75473917e-01 1.86693951e-01 1.51905537e-01
-4.89153057e-01 -6.27413273e-01 6.93420291e-01 1.18856108e+00
-3.63252223e-01 -5.78188181e-01 -5.13219595e-01 4.03602958e-01
-2.02656507e+00 -1.27473044e+00 1.78722814e-01 1.78537190e+00
6.43838823e-01 -6.08446188e-02 1.85918689e-01 -1.21889681e-01
6.09059155e-01 4.37541783e-01 -6.26940727e-01 1.47251293e-01
-3.19331616e-01 -1.44763827e-01 5.23500592e-02 2.72099018e-01
-1.01639307e+00 1.00086772e+00 6.70855379e+00 1.14820492e+00
-1.11191177e+00 3.16648073e-02 6.20746315e-01 -3.44005287e-01
-3.88755471e-01 -5.15508711e-01 -9.70764756e-01 7.22994328e-01
9.40776408e-01 -1.88799694e-01 7.00828910e-01 8.06639671e-01
1.05082206e-01 3.40771526e-01 -8.50207090e-01 1.30962431e+00
2.85299361e-01 -1.75028789e+00 6.47284806e-01 2.47385398e-01
1.35800266e+00 -5.95120974e-02 4.74748641e-01 3.70990694e-01
1.57867149e-01 -1.10090768e+00 9.13791120e-01 7.48104632e-01
9.86040473e-01 -6.17059290e-01 3.35257709e-01 3.61675411e-01
-1.22066033e+00 -4.91626710e-02 -5.93355656e-01 2.42847785e-01
5.86731255e-01 3.78252447e-01 -3.55899245e-01 3.84271890e-01
6.81000233e-01 9.98178065e-01 -4.21687029e-02 6.77663028e-01
-3.64427149e-01 5.74053168e-01 -4.33933325e-02 3.43133658e-01
4.12206352e-01 -4.39355493e-01 6.03815675e-01 9.49096322e-01
9.06804860e-01 3.95364352e-02 4.40181531e-02 1.15622079e+00
-2.12473854e-01 -2.43172482e-01 -5.79351962e-01 -3.88803631e-01
2.32685879e-01 1.12807679e+00 -5.05771518e-01 -5.28004348e-01
-4.27093893e-01 1.24178112e+00 1.30889386e-01 7.39844799e-01
-1.35527015e+00 2.31748268e-01 7.85476446e-01 1.05717145e-01
8.68811071e-01 -5.28670728e-01 3.67468596e-01 -1.61627388e+00
1.81578025e-02 -1.01641965e+00 2.47136708e-02 -9.36865032e-01
-1.21848679e+00 9.95773792e-01 -1.44211308e-03 -1.33012736e+00
-7.06902564e-01 -3.04626524e-01 -1.25143617e-01 8.61730814e-01
-1.31149721e+00 -1.13918436e+00 -3.68695140e-01 7.51250327e-01
9.83803570e-01 -5.42072952e-01 6.97829187e-01 4.58965480e-01
-5.85115671e-01 3.71422440e-01 2.53526330e-01 -3.15673947e-02
4.86749738e-01 -9.41610336e-01 6.19866431e-01 9.65994060e-01
2.54220217e-01 4.16284889e-01 5.10534823e-01 -7.48747647e-01
-1.53948152e+00 -1.29497433e+00 6.48117363e-01 -4.30969447e-01
5.83941936e-01 -3.23079407e-01 -8.26260805e-01 6.27318382e-01
3.28700006e-01 -7.32988268e-02 5.00959396e-01 -6.57931924e-01
-2.26583987e-01 1.31086722e-01 -6.98738217e-01 8.05356145e-01
1.28214598e+00 -4.11544859e-01 -2.20282987e-01 2.72100419e-01
8.90225708e-01 -5.11217117e-01 -9.81773138e-01 2.13076547e-01
4.94216919e-01 -9.61312473e-01 1.12066579e+00 -3.99213821e-01
9.53658462e-01 -3.88440073e-01 -2.39511967e-01 -1.24848664e+00
-6.86830282e-01 -1.01597595e+00 -8.12817812e-01 1.16400301e+00
-9.68004297e-03 -3.74963842e-02 6.76113963e-01 3.74410480e-01
-2.07227692e-02 -8.13408852e-01 -6.65384173e-01 -7.23230839e-01
-6.84195831e-02 -9.17766914e-02 6.32707238e-01 7.15796828e-01
-6.40287936e-01 5.14566898e-01 -1.02580953e+00 -9.67676118e-02
6.73990309e-01 5.51800802e-02 8.51278484e-01 -5.97462714e-01
-5.03256142e-01 -4.90917385e-01 -2.05977544e-01 -1.65756643e+00
7.67040849e-02 -5.64489365e-01 -1.18960366e-01 -1.48677218e+00
3.50895226e-01 -3.06859128e-02 2.30952259e-02 -5.76984473e-02
-1.48243815e-01 2.24998936e-01 2.65561521e-01 4.96393472e-01
-4.84209657e-01 1.19222999e+00 1.56416976e+00 -1.06001101e-01
-5.01460023e-02 -2.52700865e-01 -4.40901697e-01 5.41677117e-01
2.89504498e-01 -2.71906614e-01 -8.41687262e-01 -6.83231235e-01
1.87895894e-01 5.11918187e-01 2.39182889e-01 -1.11772287e+00
1.19281784e-01 -2.63732195e-01 5.91097951e-01 -6.02260172e-01
5.13310194e-01 -6.34311080e-01 7.28536844e-01 3.91172886e-01
-2.01003283e-01 2.06813693e-01 -8.25441107e-02 8.71728718e-01
-4.12488967e-01 2.83649892e-01 5.94200790e-01 4.99574118e-04
-6.86391950e-01 8.42223763e-01 -5.54270208e-01 -1.20875686e-01
1.15746021e+00 -3.60404551e-02 -3.38800967e-01 -5.60085893e-01
-6.33537412e-01 -1.27284467e-01 5.55641830e-01 8.91245842e-01
7.62738585e-01 -1.93386269e+00 -9.69507337e-01 4.44069386e-01
-2.43755847e-01 4.04824093e-02 7.40896881e-01 5.53995311e-01
-6.13082767e-01 4.94372070e-01 -4.08019125e-01 -6.69443607e-01
-8.81776631e-01 8.79192770e-01 1.42465578e-02 -3.23381066e-01
-8.25375319e-01 8.98643494e-01 6.21689796e-01 2.68747211e-01
1.13874339e-01 -1.21137351e-01 -1.29006341e-01 1.00292843e-02
7.14742303e-01 3.50877345e-01 -5.55039465e-01 -9.34451699e-01
1.71632603e-01 5.90350807e-01 4.36326489e-02 -1.01085946e-01
1.39170063e+00 -7.39567727e-02 8.09583962e-02 1.55391872e-01
1.20306325e+00 -2.38997594e-01 -2.00844574e+00 -3.69982690e-01
-6.37336791e-01 -5.85305214e-01 -5.20381741e-02 -3.70693207e-01
-1.34806800e+00 8.69112253e-01 2.92105556e-01 3.56191881e-02
1.05563104e+00 -1.48862496e-01 1.17776453e+00 -9.51586589e-02
2.48622075e-01 -8.72314334e-01 5.14625907e-01 3.66399914e-01
1.12210262e+00 -8.59628797e-01 -5.24720214e-02 -2.01781049e-01
-8.82404983e-01 1.01386118e+00 4.53709602e-01 -3.12706321e-01
5.74615121e-01 1.70422330e-01 -2.52187431e-01 9.32256505e-02
-1.20976698e+00 4.38553095e-01 5.58550835e-01 5.22145927e-01
2.78489798e-01 2.59008128e-02 -6.16062358e-02 5.81546605e-01
-1.31373983e-02 2.59019703e-01 3.62581879e-01 5.29411614e-01
-1.32658690e-01 -8.82656515e-01 -1.26888752e-01 1.64366886e-01
-3.50077957e-01 -2.22461615e-02 2.15692475e-01 4.78227019e-01
1.05412319e-01 6.42213345e-01 1.18159585e-01 -4.61561978e-01
-1.05723962e-01 -1.80095166e-01 4.71687257e-01 -3.01054060e-01
3.78732160e-02 6.32965744e-01 -3.53019893e-01 -7.04858363e-01
-6.04632020e-01 -6.99516237e-01 -7.61539280e-01 -4.35667247e-01
2.22174749e-01 2.92826835e-02 4.30426866e-01 6.01944447e-01
5.93890607e-01 7.38343120e-01 5.19783199e-01 -1.09917867e+00
-6.54379904e-01 -9.39550281e-01 -6.00606740e-01 4.59740222e-01
2.39885867e-01 -6.87730134e-01 -2.27722764e-01 5.38367867e-01] | [10.839008331298828, -0.621537983417511] |
c1869020-14a9-40ff-9dbc-2c323ae55b57 | eocsa-predicting-prognosis-of-epithelial | 2210.05258 | null | https://arxiv.org/abs/2210.05258v1 | https://arxiv.org/pdf/2210.05258v1.pdf | EOCSA: Predicting Prognosis of Epithelial Ovarian Cancer with Whole Slide Histopathological Images | Ovarian cancer is one of the most serious cancers that threaten women around the world. Epithelial ovarian cancer (EOC), as the most commonly seen subtype of ovarian cancer, has rather high mortality rate and poor prognosis among various gynecological cancers. Survival analysis outcome is able to provide treatment advices to doctors. In recent years, with the development of medical imaging technology, survival prediction approaches based on pathological images have been proposed. In this study, we designed a deep framework named EOCSA which analyzes the prognosis of EOC patients based on pathological whole slide images (WSIs). Specifically, we first randomly extracted patches from WSIs and grouped them into multiple clusters. Next, we developed a survival prediction model, named DeepConvAttentionSurv (DCAS), which was able to extract patch-level features, removed less discriminative clusters and predicted the EOC survival precisely. Particularly, channel attention, spatial attention, and neuron attention mechanisms were used to improve the performance of feature extraction. Then patient-level features were generated from our weight calculation method and the survival time was finally estimated using LASSO-Cox model. The proposed EOCSA is efficient and effective in predicting prognosis of EOC and the DCAS ensures more informative and discriminative features can be extracted. As far as we know, our work is the first to analyze the survival of EOC based on WSIs and deep neural network technologies. The experimental results demonstrate that our proposed framework has achieved state-of-the-art performance of 0.980 C-index. The implementation of the approach can be found at https://github.com/RanSuLab/EOCprognosis. | ['Leyi Wei', 'Xiuting Li', 'Changming Sun', 'Ran Su', 'Tianling Liu'] | 2022-10-11 | null | null | null | null | ['survival-analysis'] | ['miscellaneous'] | [-1.35146797e-01 -7.40607232e-02 -4.04409856e-01 -1.51380897e-01
-7.28726506e-01 -1.92520488e-02 2.24692136e-01 4.55381960e-01
-4.17832166e-01 7.08441138e-01 8.46050233e-02 -2.68118888e-01
-3.19786370e-01 -7.14968979e-01 -3.29841554e-01 -1.13335323e+00
-3.46895367e-01 3.96455735e-01 -1.99376479e-01 -1.51666403e-01
3.98986608e-01 7.73649275e-01 -1.22648621e+00 1.07125618e-01
1.29788911e+00 1.07903993e+00 2.42700920e-01 5.38288057e-01
-1.14008181e-01 2.93901742e-01 -2.19243973e-01 -7.91791379e-02
-3.11684042e-01 -3.82593364e-01 -5.48465610e-01 -5.85445106e-01
-1.29652396e-01 6.00076467e-02 -3.48451763e-01 1.03386509e+00
6.80722535e-01 -2.75187105e-01 8.33024025e-01 -9.87554014e-01
-4.11410093e-01 2.80207396e-01 -8.29144597e-01 4.19923395e-01
-1.76458508e-01 7.72000775e-02 6.51200652e-01 -8.96587849e-01
5.79049170e-01 9.77508903e-01 7.04843819e-01 7.88366377e-01
-7.65287161e-01 -9.52995658e-01 -3.03114563e-01 2.89067358e-01
-1.40869749e+00 -2.94345826e-01 4.17625219e-01 -1.17415346e-01
6.56938970e-01 3.83709341e-01 9.05149221e-01 9.85204041e-01
8.58075261e-01 9.51094747e-01 8.32572758e-01 -2.39753082e-01
-2.40005013e-02 -6.91284100e-03 5.14285080e-02 7.90349603e-01
3.69722784e-01 2.04837099e-01 -2.02974886e-01 -6.52015507e-02
5.64995944e-01 3.14813495e-01 -3.78122479e-01 4.56431601e-03
-8.52972448e-01 8.94366980e-01 9.22474504e-01 6.12727463e-01
-2.70352624e-02 2.32374012e-01 6.71664953e-01 -2.17514843e-01
3.02983046e-01 2.51427919e-01 -1.83335885e-01 5.38769215e-02
-7.75268376e-01 -6.34760484e-02 4.38420922e-01 5.17071545e-01
1.50189817e-01 -3.34433645e-01 -2.92248338e-01 5.77276111e-01
2.77073264e-01 7.45511591e-01 1.12035906e+00 -5.35978913e-01
1.01149775e-01 7.66190708e-01 -2.88857430e-01 -1.33178139e+00
-9.80828404e-01 -7.17774212e-01 -1.35671318e+00 -2.85123140e-01
-1.30035244e-02 1.21539995e-01 -8.79294991e-01 1.30753040e+00
8.02174881e-02 4.76633459e-01 6.79960698e-02 8.55483174e-01
1.17455411e+00 4.49865192e-01 3.91794592e-01 -2.28852749e-01
1.47552419e+00 -6.34845257e-01 -8.73231053e-01 1.10491596e-01
1.13916659e+00 -2.41554067e-01 6.70221746e-01 -2.28961068e-03
-6.43601477e-01 -2.47191831e-01 -8.83178115e-01 -1.63549781e-01
-4.25283045e-01 4.25648868e-01 9.78798926e-01 4.53383118e-01
-9.27077413e-01 5.73887885e-01 -1.33351207e+00 -6.47279322e-01
9.40591514e-01 5.75575709e-01 -6.37813032e-01 6.17892109e-02
-1.04251814e+00 6.74158990e-01 3.40450883e-01 2.17848197e-01
-1.01466715e+00 -4.78140861e-01 -9.09212410e-01 2.22969696e-01
6.83721676e-02 -5.20632803e-01 6.45026743e-01 -9.62475300e-01
-1.19759524e+00 9.12371397e-01 -3.07510674e-01 -3.12025875e-01
-1.02733532e-02 3.59298824e-03 -3.60841721e-01 1.24889009e-01
2.55575895e-01 4.79374528e-01 1.91296414e-01 -8.95615041e-01
-8.01180184e-01 -7.47660100e-01 -1.04034960e+00 2.82099098e-01
-6.50868475e-01 -2.09251255e-01 -7.82312334e-01 -4.69818592e-01
9.86909121e-02 -8.80199730e-01 -3.96438420e-01 4.14719693e-02
-3.92506361e-01 -2.71836013e-01 6.81840479e-01 -6.96717381e-01
1.50240445e+00 -2.41870284e+00 2.67713636e-01 3.81442726e-01
3.16415757e-01 3.73944640e-01 -1.51509523e-01 6.95375353e-02
5.64872473e-02 4.99366164e-01 -9.76183489e-02 -1.84317067e-01
-4.24043447e-01 -3.66523787e-02 2.20002890e-01 8.54286671e-01
2.25777626e-01 1.16148424e+00 -9.43408847e-01 -8.21208119e-01
-8.04442167e-02 4.49893743e-01 -4.09290761e-01 3.34497094e-02
1.78960815e-01 4.23842996e-01 -8.39331806e-01 9.76109982e-01
6.39472246e-01 -4.88534391e-01 1.17304362e-01 9.60194692e-02
2.04053298e-01 -4.89143938e-01 -3.68576556e-01 1.66844463e+00
-1.97756842e-01 6.85174644e-01 -1.09908655e-01 -1.02580118e+00
8.50707293e-01 1.58392310e-01 5.43953598e-01 -7.48464882e-01
5.45669138e-01 5.49352467e-01 5.09387106e-02 -8.71273220e-01
8.49683508e-02 2.61455048e-02 -2.56864488e-01 -2.05472276e-01
-1.84989199e-01 3.00859362e-01 -4.14194167e-02 1.41550854e-01
1.14071357e+00 -1.65073052e-01 6.31688356e-01 -4.26476896e-01
6.58873200e-01 1.00662179e-01 8.84430528e-01 4.49120164e-01
-5.35437644e-01 4.83683139e-01 6.10068977e-01 -5.33473551e-01
-5.39602935e-01 -7.12016284e-01 -5.74959934e-01 4.82518643e-01
4.52128202e-01 2.25639921e-02 -4.07269239e-01 -7.30122268e-01
2.70194739e-01 3.60765755e-01 -1.06037819e+00 -2.92766213e-01
-4.40285414e-01 -1.01310265e+00 8.71423483e-01 5.85799158e-01
5.44452667e-01 -1.15070951e+00 -1.84384167e-01 1.39041796e-01
-4.76694033e-02 -5.40543199e-01 -3.02421361e-01 1.58571362e-01
-1.04751968e+00 -1.22445726e+00 -8.76057982e-01 -9.20512557e-01
7.74333537e-01 -9.69299525e-02 7.94941723e-01 5.92626393e-01
-8.42219830e-01 -3.64812911e-01 -1.63055182e-01 -4.56034482e-01
-4.29328680e-02 2.91525543e-01 -1.45503730e-01 2.47520185e-03
6.53789937e-01 -1.31012097e-01 -9.77742553e-01 -3.11110858e-02
-6.62385106e-01 -5.21836802e-03 1.04106009e+00 1.11021721e+00
8.10417891e-01 -1.05419993e-01 5.80084205e-01 -9.90460694e-01
2.99920827e-01 -8.70726585e-01 -3.75690997e-01 -1.41747983e-03
-3.97763968e-01 -1.51193529e-01 7.85054624e-01 -7.23992214e-02
-9.05138910e-01 -6.03061318e-02 -1.97548553e-01 -2.38205314e-01
-2.08104521e-01 7.86306202e-01 -3.72974053e-02 -1.16636649e-01
4.39825118e-01 2.82051146e-01 1.51801243e-01 -1.32112965e-01
-3.76489520e-01 8.76024961e-01 3.23478222e-01 -1.13338850e-01
2.05233067e-01 6.37276888e-01 2.65305102e-01 -7.52525330e-01
-5.67999125e-01 -6.23843372e-01 -2.55965292e-02 -8.10160711e-02
9.45170164e-01 -8.62331271e-01 -1.11432242e+00 6.07853770e-01
-7.21983016e-01 -1.98771879e-01 3.99430871e-01 5.62386990e-01
-1.49415314e-01 1.94568157e-01 -9.24508333e-01 -6.06877565e-01
-6.91366494e-01 -1.01895702e+00 1.18084764e+00 8.35800648e-01
1.14043452e-01 -1.16859746e+00 -2.93637021e-03 4.62253504e-02
6.00044668e-01 4.73976076e-01 9.36787665e-01 -7.47945428e-01
-2.25963250e-01 -5.46257257e-01 -4.64335889e-01 -3.82991612e-01
-4.22620913e-04 6.06905445e-02 -8.69329453e-01 -4.85353529e-01
-3.70855540e-01 -1.65106177e-01 1.09107196e+00 7.47374952e-01
1.76772976e+00 1.84075739e-02 -1.30587971e+00 1.20923889e+00
1.67363417e+00 2.79378057e-01 7.73979247e-01 4.21106637e-01
5.60769141e-01 2.57890940e-01 6.10269547e-01 3.91714036e-01
2.25390553e-01 3.64612103e-01 5.99791169e-01 -3.27047408e-01
1.65417686e-01 -1.12128988e-01 -7.24056289e-02 6.74436510e-01
-2.56526768e-01 -4.87353802e-01 -1.08807158e+00 6.59295797e-01
-1.66545475e+00 -7.97930300e-01 -2.08634496e-01 1.84054697e+00
7.16667175e-01 -1.04278520e-01 -6.44555211e-01 -4.23657596e-02
5.40284336e-01 -2.30955005e-01 -5.95960081e-01 -2.13992774e-01
5.29521145e-03 1.69795766e-01 6.37891114e-01 1.82167590e-01
-1.26268387e+00 7.65661478e-01 4.99117470e+00 1.29041779e+00
-1.34378791e+00 -1.46948382e-01 1.29991651e+00 -6.20080754e-02
-1.18409514e-01 -3.09391171e-01 -1.00294673e+00 6.28268301e-01
8.12027574e-01 -5.76291531e-02 -1.10687114e-01 6.58122122e-01
2.44066492e-01 -7.96128735e-02 -7.54730284e-01 8.79401207e-01
1.12082548e-01 -1.31923318e+00 -1.10698931e-01 2.05365911e-01
5.67120373e-01 -6.53594453e-03 5.46708889e-02 4.74042445e-01
-5.14999628e-02 -1.53279567e+00 -3.53550553e-01 9.49760795e-01
1.17198002e+00 -7.65328348e-01 1.47669291e+00 3.50747645e-01
-1.01560187e+00 -2.35593677e-01 -3.06725383e-01 4.04472202e-01
-4.55045015e-01 4.39827174e-01 -5.70977390e-01 5.61202884e-01
9.97886896e-01 1.05690706e+00 -7.30872512e-01 1.20524681e+00
2.61166155e-01 6.87473416e-01 -3.16677511e-01 -4.88784224e-01
5.48840016e-02 1.79974899e-01 2.48009324e-01 1.16384459e+00
8.25862408e-01 2.73337662e-01 -1.26259699e-01 4.92309839e-01
-2.82563921e-02 3.84354740e-01 -4.31139618e-01 -5.45514114e-02
5.69392681e-01 1.46106541e+00 -9.78731990e-01 -6.88536540e-02
-2.39414051e-01 7.67763376e-01 4.31993306e-01 3.12480062e-01
-7.86935091e-01 -6.58232033e-01 1.39390886e-01 -3.89542282e-02
-8.77657384e-02 5.30461550e-01 -6.00232065e-01 -1.06452620e+00
-4.58809197e-01 -5.46014786e-01 5.78563571e-01 -3.77658665e-01
-1.18666959e+00 4.82208788e-01 -4.56624478e-01 -1.27477193e+00
4.45677251e-01 -6.43378377e-01 -8.48130763e-01 7.83300221e-01
-1.63982296e+00 -9.53563273e-01 -7.02741861e-01 2.69073248e-01
3.27006161e-01 -3.24770540e-01 1.22375095e+00 9.80689973e-02
-1.05331326e+00 7.66234457e-01 4.75950360e-01 3.05272102e-01
7.07495093e-01 -1.16245532e+00 -3.78327042e-01 2.49871701e-01
-6.87000751e-01 5.18579245e-01 2.84585267e-01 -7.15484798e-01
-1.31148493e+00 -1.10616207e+00 8.98415089e-01 8.83385539e-02
4.10740077e-01 5.23719452e-02 -8.21739376e-01 3.56997252e-01
8.01807344e-02 4.93291110e-01 9.84633386e-01 -1.93315536e-01
4.65729415e-01 -2.54544854e-01 -9.72834527e-01 4.67009515e-01
5.62319934e-01 5.70632219e-02 1.60103604e-01 1.37611404e-01
2.14638665e-01 -5.20500779e-01 -1.03876865e+00 7.55932331e-01
5.43007076e-01 -7.98499346e-01 7.09102452e-01 -3.18773896e-01
7.37027466e-01 -1.71435848e-01 3.22015822e-01 -1.16658330e+00
-4.78526294e-01 4.60466444e-02 2.78889090e-02 8.67553651e-01
5.36061108e-01 -5.68037927e-01 9.65268850e-01 2.02836022e-01
-2.53982931e-01 -1.43372095e+00 -8.70538294e-01 -2.53843158e-01
2.46427178e-01 9.81708691e-02 4.30926353e-01 8.49949062e-01
8.38501230e-02 -1.26613677e-01 1.23842008e-01 2.17679217e-01
4.69398528e-01 4.10634130e-02 3.03513050e-01 -1.18884277e+00
-2.59083863e-02 -6.72634542e-01 -6.52230322e-01 -3.84450763e-01
1.07093379e-01 -1.11650109e+00 -1.09657034e-01 -1.48101723e+00
5.68239152e-01 -6.59628272e-01 -5.97844005e-01 6.05566680e-01
-6.28682554e-01 3.87578905e-01 -2.51624048e-01 3.49264801e-01
-6.23719394e-01 7.06798255e-01 1.50265551e+00 -2.57681787e-01
-2.32271284e-01 -1.02790788e-01 -6.74605370e-01 5.45476615e-01
9.31724966e-01 -4.38021034e-01 1.78251058e-01 -2.13131104e-02
1.00025022e-02 3.68179649e-01 1.74528509e-01 -9.02166367e-01
2.84782380e-01 -1.13487378e-01 9.60356593e-01 -4.57133740e-01
-3.27791534e-02 -7.11867094e-01 8.83146301e-02 9.89031911e-01
-2.09663242e-01 -5.11455119e-01 5.10278285e-01 5.63661397e-01
-3.99412543e-01 -1.21701747e-01 6.33138597e-01 1.14394516e-01
-7.17394233e-01 5.88868499e-01 -3.47326040e-01 -2.73286194e-01
1.43124866e+00 -1.65542379e-01 -3.21006835e-01 1.08360559e-01
-6.08695865e-01 6.87631607e-01 3.06906193e-01 1.23239636e-01
8.41792405e-01 -1.46081316e+00 -9.40828562e-01 3.69870365e-01
3.29791158e-01 1.25566497e-01 4.81576622e-01 1.20932961e+00
-9.92966890e-01 5.24836183e-01 -1.00143343e-01 -5.51029146e-01
-1.22896302e+00 3.94140095e-01 4.81302857e-01 -6.25739992e-01
-6.13024294e-01 1.02830505e+00 2.60645092e-01 -1.98322192e-01
3.06802928e-01 1.66522171e-02 -9.34900403e-01 -1.19522363e-01
3.82426530e-01 8.17210004e-02 -1.10343747e-01 -5.33078969e-01
-4.70185220e-01 6.16295576e-01 -2.63238847e-01 5.50495446e-01
1.31988406e+00 2.65104827e-02 -2.40644678e-01 5.27885109e-02
1.33092737e+00 -8.71497095e-02 -8.89870167e-01 1.46629706e-01
-1.41220331e-01 -2.94915110e-01 3.16957355e-01 -6.80307388e-01
-1.39799428e+00 8.52009594e-01 9.21422303e-01 -1.09892175e-01
1.35423839e+00 -5.19086532e-02 8.98807466e-01 -4.99345437e-02
7.33604133e-02 -7.18212068e-01 -1.35849983e-01 1.57939032e-01
6.26001298e-01 -1.51790929e+00 1.72909517e-02 -2.68296361e-01
-5.64200997e-01 1.17033100e+00 7.22819626e-01 -3.22358727e-01
7.53240585e-01 3.50688249e-01 1.18973427e-01 -3.18203926e-01
-6.77474022e-01 -4.20073196e-02 2.04819128e-01 3.18128258e-01
6.75915539e-01 3.49731743e-01 -6.14720941e-01 1.19295895e+00
1.58892363e-01 1.30968660e-01 1.16992585e-01 5.82952499e-01
-5.11643827e-01 -6.40666127e-01 -2.32812926e-01 9.34929073e-01
-8.46332133e-01 -2.01339990e-01 2.70022154e-02 8.69603813e-01
9.12338123e-02 5.12103021e-01 1.56654909e-01 -1.51474074e-01
-1.06262878e-01 -4.18179423e-01 -7.62588754e-02 -3.26949656e-01
-3.88982564e-01 6.92871362e-02 -3.33198518e-01 -6.10322893e-01
-2.04810053e-01 -4.28499401e-01 -1.66811073e+00 -9.07011852e-02
-5.62777162e-01 3.08173507e-01 5.67086637e-01 6.91365659e-01
4.53806877e-01 7.22710967e-01 5.74584067e-01 -7.28598833e-01
-2.68733911e-02 -1.05765879e+00 -7.85798788e-01 3.66436720e-01
2.68222898e-01 -4.88310307e-01 -3.97448748e-01 -1.81791350e-01] | [15.147439002990723, -2.944864273071289] |
df9d9da7-1f0b-40cd-80f2-ab5a75689c5b | fast-as-chita-neural-network-pruning-with | 2302.14623 | null | https://arxiv.org/abs/2302.14623v1 | https://arxiv.org/pdf/2302.14623v1.pdf | Fast as CHITA: Neural Network Pruning with Combinatorial Optimization | The sheer size of modern neural networks makes model serving a serious computational challenge. A popular class of compression techniques overcomes this challenge by pruning or sparsifying the weights of pretrained networks. While useful, these techniques often face serious tradeoffs between computational requirements and compression quality. In this work, we propose a novel optimization-based pruning framework that considers the combined effect of pruning (and updating) multiple weights subject to a sparsity constraint. Our approach, CHITA, extends the classical Optimal Brain Surgeon framework and results in significant improvements in speed, memory, and performance over existing optimization-based approaches for network pruning. CHITA's main workhorse performs combinatorial optimization updates on a memory-friendly representation of local quadratic approximation(s) of the loss function. On a standard benchmark of pretrained models and datasets, CHITA leads to significantly better sparsity-accuracy tradeoffs than competing methods. For example, for MLPNet with only 2% of the weights retained, our approach improves the accuracy by 63% relative to the state of the art. Furthermore, when used in conjunction with fine-tuning SGD steps, our method achieves significant accuracy gains over the state-of-the-art approaches. | ['Rahul Mazumder', 'Zhe Zhao', 'Natalia Ponomareva', 'Hussein Hazimeh', 'Xiang Meng', 'Wenyu Chen', 'Riade Benbaki'] | 2023-02-28 | null | null | null | null | ['combinatorial-optimization'] | ['methodology'] | [ 4.03315902e-01 3.30117308e-02 -3.11943591e-01 -5.73571682e-01
-5.16433358e-01 5.26218899e-02 6.58213198e-02 1.98816881e-01
-8.74549210e-01 6.38652503e-01 4.05072607e-03 -2.09933102e-01
-4.41454530e-01 -6.70278788e-01 -7.45913029e-01 -4.51009661e-01
-3.74744087e-02 4.74816352e-01 2.02857807e-01 -3.50487866e-02
1.54281318e-01 4.75000679e-01 -1.49201667e+00 2.41759926e-01
7.73372054e-01 1.56461322e+00 2.69971579e-01 2.03009441e-01
-2.36301194e-03 6.59197986e-01 -4.17101294e-01 -6.20439291e-01
5.75980961e-01 -7.36348554e-02 -6.66195571e-01 -2.28662640e-01
8.63897264e-01 -2.80580282e-01 -3.65998060e-01 1.09873807e+00
4.47406679e-01 2.23528087e-01 1.60487816e-01 -8.39963853e-01
-2.09012374e-01 9.27573323e-01 -5.38803995e-01 5.38764596e-01
-4.91968751e-01 -1.53617322e-01 1.14152062e+00 -8.21522951e-01
4.40135807e-01 1.12209165e+00 9.93892431e-01 5.42077780e-01
-1.55775583e+00 -8.44340503e-01 3.94678086e-01 2.89301097e-01
-1.34437978e+00 -8.42321217e-01 5.23781955e-01 2.13756524e-02
1.31950283e+00 1.92532524e-01 9.04463768e-01 6.28194392e-01
3.78566831e-02 7.28858888e-01 6.24247968e-01 -4.43585634e-01
3.18512499e-01 -1.20937318e-01 4.36584860e-01 8.89404535e-01
5.81149936e-01 -1.58154499e-02 -7.48276412e-01 -1.21556886e-01
5.72878003e-01 5.35142832e-02 -2.25486338e-01 -2.79621214e-01
-8.78803492e-01 8.19369853e-01 4.37344462e-01 1.51238844e-01
-3.18602353e-01 4.85643357e-01 5.24819732e-01 3.04400116e-01
5.75015724e-01 5.85775435e-01 -6.99167907e-01 -1.98009834e-01
-1.58265436e+00 3.78512889e-01 8.95149648e-01 7.58274376e-01
3.70827168e-01 3.36486578e-01 -7.48919994e-02 1.15707219e+00
1.15653418e-01 -2.42152624e-02 5.39262891e-01 -1.06573880e+00
8.80089045e-01 5.75334013e-01 -5.14543772e-01 -1.10893261e+00
-3.54020357e-01 -1.09654641e+00 -9.18124139e-01 1.47133037e-01
2.93161124e-01 -1.17590740e-01 -1.04343045e+00 1.82288992e+00
1.51756227e-01 8.21828395e-02 -1.44941822e-01 6.40189171e-01
5.76073766e-01 3.88196439e-01 -2.23870091e-02 -2.47180685e-01
1.03724825e+00 -1.20304465e+00 -4.91775990e-01 -5.65798461e-01
5.54173648e-01 -4.51595008e-01 8.98047686e-01 6.00798666e-01
-1.54369926e+00 -1.40133217e-01 -1.20566142e+00 -1.18256211e-01
-1.25299357e-02 1.98081851e-01 7.38026440e-01 6.39912069e-01
-1.10621715e+00 1.12814391e+00 -1.10876286e+00 1.12927422e-01
1.13956022e+00 7.15705752e-01 -2.48428285e-01 -2.23027721e-01
-6.61662221e-01 1.10189068e+00 6.35084331e-01 -5.24881184e-02
-6.67626500e-01 -1.28257799e+00 -6.51986241e-01 5.75476527e-01
6.03912592e-01 -6.35719717e-01 1.30748200e+00 -7.28347123e-01
-1.30676413e+00 5.57569385e-01 -1.48535967e-01 -1.18523359e+00
5.11243582e-01 -4.69197750e-01 -4.91714552e-02 4.65392917e-02
-4.04553145e-01 8.75695050e-01 7.72010922e-01 -6.52741373e-01
-5.07813811e-01 -3.20109725e-01 -1.01996608e-01 4.07582298e-02
-9.06692863e-01 -3.93120907e-02 -5.30280054e-01 -9.17976975e-01
3.25730085e-01 -5.90964079e-01 -4.97573107e-01 3.58341038e-01
-2.02907905e-01 2.23801248e-02 5.53363383e-01 -6.45022869e-01
1.54593575e+00 -1.96781301e+00 2.43713453e-01 2.88280487e-01
5.52140951e-01 5.44150054e-01 -1.95921212e-01 -1.69335604e-01
-2.10748632e-02 2.16721877e-01 -4.36952561e-01 -7.69827724e-01
3.76412719e-02 4.26041543e-01 -1.02074817e-01 3.01313013e-01
1.01171553e-01 7.15048492e-01 -4.70883310e-01 -4.60929930e-01
-1.68865293e-01 6.08899713e-01 -1.01638937e+00 -3.27413350e-01
-1.38627619e-01 -3.47298861e-01 -9.46468189e-02 6.20009601e-01
4.52282101e-01 -3.20376992e-01 1.17841393e-01 -2.62659669e-01
2.18050003e-01 5.94836175e-01 -1.06769192e+00 1.79087687e+00
-3.06445092e-01 6.24100387e-01 2.03131273e-01 -1.34739637e+00
7.86780000e-01 3.92521992e-02 3.88362169e-01 -5.53892910e-01
2.67758489e-01 3.33208948e-01 2.23924786e-01 2.98239794e-02
3.16885471e-01 3.24372724e-02 4.83866006e-01 3.70491087e-01
3.76950204e-01 -1.13636637e-02 5.09112298e-01 -1.84711237e-02
1.35665226e+00 -3.41977894e-01 2.35591024e-01 -3.99434477e-01
1.35457978e-01 -1.98220745e-01 7.73217380e-01 6.84033215e-01
-8.15664232e-02 4.05969858e-01 5.09332478e-01 -5.90243101e-01
-1.13204503e+00 -5.21534562e-01 -2.29949877e-01 1.04851401e+00
-3.96791428e-01 -5.91094673e-01 -9.26640034e-01 -5.11289358e-01
1.62523121e-01 5.39998710e-01 -5.62059641e-01 -2.74129570e-01
-7.32972085e-01 -1.00964963e+00 5.67108035e-01 6.35341406e-01
7.75421381e-01 -7.55720377e-01 -8.79911423e-01 2.38508925e-01
1.42272860e-01 -1.16057193e+00 -4.04877007e-01 5.33212304e-01
-1.69176030e+00 -7.28145897e-01 -5.27321041e-01 -6.53102040e-01
6.77415788e-01 1.99298516e-01 1.21288764e+00 3.77100617e-01
-3.86691034e-01 -2.37245768e-01 -4.64626923e-02 -5.35620809e-01
6.62163198e-02 5.24693012e-01 -4.56514284e-02 -3.64383698e-01
1.62753493e-01 -9.20466840e-01 -5.15346587e-01 -9.77421924e-02
-8.17573488e-01 1.33242905e-01 7.76387930e-01 9.59657729e-01
9.15307224e-01 1.70230120e-01 4.26267564e-01 -1.04674935e+00
7.24555254e-01 -2.94604868e-01 -6.32873178e-01 1.48552209e-01
-1.05186999e+00 3.92235696e-01 5.86284935e-01 -5.86568892e-01
-6.03861511e-01 -2.60220859e-02 -1.48862988e-01 -7.40364909e-01
2.78597683e-01 7.73345053e-01 1.42987847e-01 -5.16183436e-01
7.12796211e-01 6.41376302e-02 1.11656465e-01 -7.43970335e-01
5.89627260e-03 -3.97929363e-02 4.08251047e-01 -4.35247511e-01
4.72319424e-01 1.34908468e-01 8.64608511e-02 -5.61821997e-01
-1.02622998e+00 -2.05066264e-01 -4.00943160e-01 2.11311847e-01
1.60560220e-01 -7.20812201e-01 -2.71096915e-01 2.58491993e-01
-9.92882609e-01 -4.50895995e-01 -5.99806309e-01 5.01741409e-01
-2.72624761e-01 1.61116362e-01 -7.06288874e-01 -5.74138939e-01
-7.90246487e-01 -1.03723872e+00 5.32468617e-01 -7.61803985e-02
-1.09436974e-01 -7.47494757e-01 -2.22880453e-01 3.00671488e-01
7.95853436e-01 1.15727238e-01 1.08135581e+00 -6.82044268e-01
-4.19622809e-01 -3.43424201e-01 -4.15482014e-01 6.45799756e-01
-3.98410767e-01 -3.82867038e-01 -6.79165304e-01 -3.44367832e-01
2.42127463e-01 -3.03150654e-01 1.32208097e+00 5.82103491e-01
1.60173023e+00 -6.62889719e-01 -1.61926866e-01 1.00090289e+00
1.49376285e+00 2.47121248e-02 3.66329908e-01 3.60733598e-01
4.39143419e-01 3.00700188e-01 2.59341985e-01 5.90965509e-01
7.97893927e-02 5.44825971e-01 5.61010242e-01 8.82224590e-02
-1.43786654e-01 5.04909605e-02 1.35190308e-01 1.06632507e+00
-1.88385099e-01 7.28045264e-03 -8.24570894e-01 4.98076349e-01
-1.82952094e+00 -9.36423540e-01 4.49100643e-01 2.07818699e+00
1.12517154e+00 4.58191037e-01 -5.67942783e-02 5.66400170e-01
3.60791534e-01 2.42313027e-01 -8.51944089e-01 -4.49489295e-01
1.20412568e-02 6.59957111e-01 8.87839794e-01 2.96829969e-01
-9.87786770e-01 7.65492797e-01 7.27180290e+00 1.03732800e+00
-9.76466715e-01 3.41577441e-01 7.64668405e-01 -9.56646025e-01
2.05545388e-02 -3.00261706e-01 -1.18059492e+00 2.49974385e-01
1.05922771e+00 -1.09414935e-01 8.42884421e-01 1.07899678e+00
-4.21259962e-02 9.95868742e-02 -1.08067441e+00 1.17399549e+00
2.02745706e-01 -1.68817580e+00 -1.25175908e-01 1.81585580e-01
7.19815075e-01 3.82423162e-01 7.26160258e-02 3.15735698e-01
2.63455391e-01 -1.22674310e+00 9.72649693e-01 2.82374680e-01
5.71571946e-01 -8.30898583e-01 7.22902834e-01 4.12979335e-01
-8.93293023e-01 -3.66817504e-01 -6.54294729e-01 -1.39673322e-01
1.04236215e-01 9.69011307e-01 -3.80843431e-01 -9.49089974e-03
8.45858335e-01 7.12410390e-01 -5.43847561e-01 1.28070354e+00
1.55048808e-02 9.18867230e-01 -7.94981122e-01 4.88339365e-02
2.27510065e-01 9.89201106e-03 4.46158230e-01 1.23561478e+00
2.14348912e-01 8.63880590e-02 -9.15145800e-02 8.89563322e-01
-5.24510264e-01 1.38895482e-01 -3.08846474e-01 -1.13704100e-01
6.58852100e-01 1.01278174e+00 -6.08925462e-01 -3.10181499e-01
-6.15582205e-02 4.36562210e-01 8.95225823e-01 8.84995610e-02
-7.23013997e-01 -3.44318360e-01 6.32611811e-01 6.38680682e-02
6.40190423e-01 -2.07104996e-01 -7.63268232e-01 -9.34541702e-01
3.85524094e-01 -9.77608740e-01 2.58899629e-01 -1.36156976e-01
-7.91114688e-01 6.85002506e-01 7.94315040e-02 -6.46705568e-01
1.81796029e-01 -6.14886463e-01 -3.94784719e-01 5.08813798e-01
-1.55658948e+00 -7.42559552e-01 -1.44196287e-01 2.56957173e-01
6.18642867e-01 -3.65192890e-01 7.44617581e-01 6.81886911e-01
-8.45431149e-01 1.03415620e+00 7.77129307e-02 -2.77449429e-01
1.14284813e-01 -7.23340034e-01 3.53317499e-01 8.60288560e-01
2.85970330e-01 5.72806478e-01 5.41123867e-01 -2.82878518e-01
-1.30010903e+00 -1.10176671e+00 8.67867351e-01 1.68758586e-01
5.04133821e-01 -1.93837687e-01 -8.98676991e-01 6.26963854e-01
-1.91521049e-01 1.59604862e-01 6.51075721e-01 2.63713837e-01
-4.25540507e-01 -4.24984097e-01 -1.25181592e+00 4.61140931e-01
1.35668719e+00 -5.03348373e-02 -5.16964197e-01 3.61492187e-01
7.71923482e-01 -5.18859744e-01 -9.67013180e-01 5.23171425e-01
7.22313225e-01 -8.86892855e-01 1.05445278e+00 -6.77325547e-01
6.32754087e-01 3.05674374e-01 -3.35715801e-01 -1.14953446e+00
-2.18359739e-01 -5.39735675e-01 -6.91730917e-01 6.13044560e-01
6.03575706e-01 -6.62869513e-01 1.13090706e+00 8.38231623e-01
-3.20809871e-01 -1.44596159e+00 -1.12241590e+00 -8.85233402e-01
-5.72845303e-02 -6.20183170e-01 6.35973155e-01 6.25385761e-01
-1.13565281e-01 4.63500395e-02 -2.81566471e-01 -3.53302449e-01
7.21538961e-01 -2.37786278e-01 3.47852528e-01 -1.30395007e+00
-4.99048293e-01 -8.57392073e-01 -4.42812026e-01 -1.09568989e+00
-5.07196747e-02 -9.72050846e-01 -1.58879593e-01 -1.38238335e+00
7.10773170e-02 -7.18724847e-01 -4.75469470e-01 9.54083860e-01
1.98183134e-01 3.27185184e-01 4.71323311e-01 2.11714670e-01
-6.00124896e-01 5.02296269e-01 9.53901708e-01 -1.19151689e-01
-2.20894307e-01 -1.02576531e-01 -7.86605835e-01 8.14074576e-01
9.49365020e-01 -8.15493345e-01 -4.43505526e-01 -1.06696236e+00
2.72602051e-01 -2.77845353e-01 2.09183976e-01 -1.43313718e+00
5.64689279e-01 1.39579937e-01 2.07743138e-01 -4.94460553e-01
6.26131177e-01 -7.90567994e-01 4.78928499e-02 7.23611057e-01
-5.59579015e-01 2.27568701e-01 4.55409795e-01 4.92792845e-01
-2.32117042e-01 -5.03243327e-01 9.70248163e-01 -5.42433895e-02
-1.85999587e-01 4.70580608e-01 -2.68650451e-03 2.21273914e-01
6.32485747e-01 -4.39798415e-01 -2.29289502e-01 5.92710786e-02
-6.72474086e-01 1.50131479e-01 1.12948745e-01 -3.27081941e-02
8.17760885e-01 -1.08151495e+00 -5.46193123e-01 2.59799540e-01
-4.72769976e-01 2.56747305e-01 -6.92607239e-02 1.11358309e+00
-5.96371472e-01 4.74520773e-01 -1.75009742e-01 -3.07100207e-01
-1.47359014e+00 9.45271850e-02 2.65980929e-01 -7.11425006e-01
-7.67826021e-01 1.30795157e+00 -3.34760934e-01 -1.09538496e-01
7.49049306e-01 -6.99748576e-01 2.91476771e-02 -8.28056782e-02
5.43425620e-01 6.33022487e-01 4.52658564e-01 -1.54902562e-01
-1.36127248e-01 2.28443906e-01 -4.36714500e-01 3.73143069e-02
1.89830613e+00 4.45312858e-01 -3.95306051e-02 9.93398726e-02
1.09942997e+00 -4.86758560e-01 -1.29788160e+00 -4.96910512e-01
8.12011063e-02 -5.16755104e-01 5.79984903e-01 -7.78883398e-01
-1.79694021e+00 8.10371757e-01 5.00575423e-01 -1.29881561e-01
1.25897086e+00 -4.57925707e-01 1.06098843e+00 7.94075847e-01
3.65061641e-01 -1.23833191e+00 -1.07111126e-01 6.17289245e-01
7.42511511e-01 -1.01820767e+00 4.45021480e-01 -3.37334931e-01
-2.66321599e-01 9.92847681e-01 4.49976027e-01 -3.73482287e-01
8.37470531e-01 5.08674383e-01 -5.02186418e-01 -2.10660756e-01
-1.11387146e+00 2.98843145e-01 3.21832925e-01 2.14686885e-01
2.40715057e-01 -1.86140418e-01 -4.13205534e-01 6.76184058e-01
-5.24975538e-01 8.93711224e-02 -1.41206831e-01 9.99852836e-01
-6.78002954e-01 -9.14796412e-01 3.15044187e-02 1.13529229e+00
-7.07693875e-01 -5.65237999e-01 -3.83282937e-02 6.50986195e-01
5.65312207e-02 5.15127182e-01 4.88308705e-02 -3.68588477e-01
3.61570507e-01 8.96783024e-02 5.71103573e-01 -6.00966275e-01
-8.21832836e-01 -1.34902358e-01 1.39363855e-01 -6.84516490e-01
-3.38527828e-01 -5.46315432e-01 -9.83488500e-01 -5.73897839e-01
-6.67941988e-01 -2.02767536e-01 7.36978412e-01 9.89038765e-01
5.50927520e-01 5.90972543e-01 -5.21317944e-02 -9.80663478e-01
-1.04318225e+00 -9.02915001e-01 -3.36341411e-01 4.38942872e-02
1.36743769e-01 -7.44517684e-01 -2.62861431e-01 -1.76687226e-01] | [8.577067375183105, 3.2441022396087646] |
e4ac344b-418c-416a-b93d-e13924e1888d | a-large-scale-analysis-of-mixed-initiative-in | 2104.07096 | null | https://arxiv.org/abs/2104.07096v2 | https://arxiv.org/pdf/2104.07096v2.pdf | A Large-Scale Analysis of Mixed Initiative in Information-Seeking Dialogues for Conversational Search | Conversational search is a relatively young area of research that aims at automating an information-seeking dialogue. In this paper we help to position it with respect to other research areas within conversational Artificial Intelligence (AI) by analysing the structural properties of an information-seeking dialogue. To this end, we perform a large-scale dialogue analysis of more than 150K transcripts from 16 publicly available dialogue datasets. These datasets were collected to inform different dialogue-based tasks including conversational search. We extract different patterns of mixed initiative from these dialogue transcripts and use them to compare dialogues of different types. Moreover, we contrast the patterns found in information-seeking dialogues that are being used for research purposes with the patterns found in virtual reference interviews that were conducted by professional librarians. The insights we provide (1) establish close relations between conversational search and other conversational AI tasks; and (2) uncover limitations of existing conversational datasets to inform future data collection tasks. | ['Maarten de Rijke', 'Evangelos Kanoulas', 'Svitlana Vakulenko'] | 2021-04-14 | null | null | null | null | ['conversational-search'] | ['natural-language-processing'] | [ 4.69209641e-01 9.85586524e-01 -1.77274734e-01 -4.71691221e-01
-7.96126485e-01 -9.51637805e-01 1.44965851e+00 3.82947475e-01
-3.29872042e-01 6.80960178e-01 1.17461467e+00 -6.64065540e-01
-4.29007441e-01 -2.02830032e-01 3.33909333e-01 -2.73093402e-01
1.02447189e-01 1.02955794e+00 1.21714007e-02 -8.08270335e-01
8.88073206e-01 2.14064345e-01 -1.19591618e+00 7.64373004e-01
9.32706535e-01 5.32311916e-01 1.08692423e-01 8.63834023e-01
-6.10417068e-01 1.31532896e+00 -6.72478557e-01 -5.26366770e-01
-7.69820511e-02 -8.64007413e-01 -2.07547450e+00 1.58991829e-01
7.83816949e-02 -7.65219331e-02 -2.45011792e-01 6.54789209e-01
5.52979052e-01 2.40439903e-02 5.89630783e-01 -1.14231396e+00
-3.93986292e-02 7.97984362e-01 1.21376038e-01 3.86520118e-01
1.21960807e+00 1.72735825e-01 1.24355447e+00 -2.39232495e-01
1.37711966e+00 1.44280648e+00 4.70578521e-01 8.27196956e-01
-1.04717064e+00 -7.25086927e-02 -4.28344876e-01 3.79992276e-01
-7.65105546e-01 -9.09912884e-01 9.15525615e-01 -5.90141833e-01
1.01718581e+00 6.76379144e-01 7.52662838e-01 1.42786491e+00
-1.54801145e-01 1.22315550e+00 1.13717043e+00 -7.99789667e-01
6.16783909e-02 4.44435388e-01 6.25070155e-01 3.54228795e-01
-6.29564345e-01 -6.34577155e-01 -6.15857184e-01 -8.77907515e-01
2.51889855e-01 -7.17535734e-01 -5.39044976e-01 -5.57783954e-02
-1.32508814e+00 1.11081350e+00 -6.32835999e-02 8.95371318e-01
-5.33329964e-01 -6.18158042e-01 6.78674340e-01 7.62455881e-01
3.60976934e-01 1.01805305e+00 -4.39907372e-01 -9.61486995e-01
-3.11774790e-01 4.61174905e-01 2.10553145e+00 9.51534092e-01
6.98285520e-01 -9.13026810e-01 -3.53785634e-01 1.34243941e+00
2.90678978e-01 -2.96616405e-01 6.33309484e-01 -1.19585764e+00
5.15442431e-01 9.96456504e-01 3.43302101e-01 -1.01112151e+00
-5.60360014e-01 5.28635442e-01 -3.62733960e-01 -6.87050998e-01
8.53129327e-01 -2.47450650e-01 1.01395145e-01 1.11321366e+00
2.68086076e-01 -8.79292488e-01 4.49334860e-01 6.40161395e-01
1.35113478e+00 2.98853993e-01 -2.50729650e-01 -5.00060797e-01
1.71852791e+00 -1.06785047e+00 -1.08337307e+00 -2.11824194e-01
9.56393540e-01 -9.92949188e-01 1.09698248e+00 1.95833564e-01
-1.05447161e+00 1.51875436e-01 -4.92723823e-01 -3.31396788e-01
-3.89843017e-01 -4.01441604e-01 8.15023422e-01 6.63087487e-01
-1.14811742e+00 1.86475977e-01 -2.44479015e-01 -8.54885221e-01
-5.98027781e-02 -6.87706023e-02 -6.77606612e-02 2.41926908e-01
-1.31099880e+00 1.17057633e+00 -7.68602267e-02 -1.41742200e-01
-2.56233841e-01 -5.09890735e-01 -6.22652411e-01 -3.27144414e-01
7.01599360e-01 -1.93527117e-01 1.93785024e+00 -7.95215249e-01
-1.59567177e+00 1.29104102e+00 -3.04388553e-01 -2.59847254e-01
4.56195533e-01 1.21331699e-01 -1.39996365e-01 1.95363790e-01
1.91335037e-01 2.98120826e-01 -1.67120295e-03 -1.12243485e+00
-4.86696810e-01 -3.82233918e-01 2.31161699e-01 5.65307438e-01
-7.82569796e-02 4.09521490e-01 -4.15657014e-01 -1.19461253e-01
-1.74753979e-01 -1.05453730e+00 -1.82202429e-01 -4.23239142e-01
-5.94313860e-01 -8.57410789e-01 5.46757758e-01 -7.73679435e-01
1.20160854e+00 -1.63978088e+00 4.09997612e-01 -7.89244547e-02
4.75877821e-01 6.47328570e-02 1.50735052e-02 1.37131238e+00
4.51959848e-01 2.84344316e-01 -1.59229904e-01 -1.01701304e-01
1.30218998e-01 1.97894841e-01 -7.57698044e-02 6.23476878e-02
-8.09039474e-02 9.50772822e-01 -1.19447398e+00 -7.52043605e-01
-8.98883492e-02 -9.81755555e-02 -2.12174520e-01 3.87856394e-01
-4.56826001e-01 4.25867945e-01 -6.97322130e-01 4.76821989e-01
-6.95808679e-02 -1.74430862e-01 5.72583497e-01 3.42344679e-02
-3.09892833e-01 1.01089942e+00 -4.12096947e-01 1.78315020e+00
-4.62021589e-01 1.25971258e+00 5.50939560e-01 -7.17756987e-01
8.83725345e-01 5.42323112e-01 3.98831725e-01 -6.46701038e-01
1.74676284e-01 2.85004936e-02 4.59989160e-01 -9.00652587e-01
8.45701635e-01 2.89863080e-01 -2.70460725e-01 1.20782852e+00
-7.10979551e-02 -5.51617265e-01 2.68273741e-01 4.29427385e-01
1.33302796e+00 -6.29941702e-01 4.82876748e-01 -5.44247866e-01
8.20398271e-01 5.58166146e-01 3.42587382e-02 9.56287861e-01
-4.71986264e-01 3.84470113e-02 8.97681475e-01 -5.46554983e-01
-7.85080791e-01 -1.85559869e-01 -3.29678804e-01 1.14478445e+00
-1.69145435e-01 -8.47570658e-01 -7.22312033e-01 -6.37148857e-01
-2.25404948e-01 6.31268382e-01 -4.40414190e-01 3.29784602e-01
-4.92281556e-01 -4.74295586e-01 8.91351461e-01 -2.41368860e-01
6.22951984e-01 -1.53552294e+00 -8.02084327e-01 3.71075779e-01
-8.91437411e-01 -1.13743281e+00 -3.01450402e-01 -1.19980380e-01
-3.19366455e-01 -1.51702821e+00 -5.28826296e-01 -6.02762938e-01
4.43624072e-02 9.32818279e-02 1.46351182e+00 3.04263979e-01
-3.09913218e-01 9.68490064e-01 -6.30757272e-01 -4.26464140e-01
-1.18040335e+00 6.08183622e-01 -3.59303236e-01 -2.43893474e-01
9.20613587e-01 -3.20692390e-01 -2.80154884e-01 5.81702769e-01
-8.45746398e-01 7.26810247e-02 1.57288522e-01 9.48324025e-01
-6.89906955e-01 -3.63472819e-01 6.04034245e-01 -1.05380595e+00
1.96029460e+00 -7.07493305e-01 -7.06584454e-02 4.07477587e-01
-6.57607317e-01 -7.62105435e-02 1.88530888e-02 -1.66192830e-01
-1.33245766e+00 -4.45748657e-01 1.58453286e-02 6.84079945e-01
-2.56060362e-01 7.03984737e-01 3.20486635e-01 9.58957523e-03
8.63384724e-01 1.41226277e-01 4.19750363e-01 -4.08861279e-01
2.40920886e-01 1.51958072e+00 9.41476151e-02 -6.23675764e-01
1.15376309e-01 4.14363705e-02 -6.45655692e-01 -1.43955529e+00
-4.29221362e-01 -8.39290500e-01 -6.97735906e-01 -5.52399874e-01
5.99018753e-01 -3.12646598e-01 -1.27413797e+00 2.74286419e-01
-1.24090779e+00 -6.01728022e-01 2.43405662e-02 3.05022933e-02
-6.93436205e-01 6.66483104e-01 -9.15099323e-01 -1.31859469e+00
-4.85499859e-01 -1.06218290e+00 7.51082182e-01 2.61999041e-01
-1.30149221e+00 -1.24468529e+00 3.21891427e-01 8.93454313e-01
4.73004609e-01 -1.53728589e-01 9.23186302e-01 -1.26225770e+00
-2.13152021e-02 1.27469659e-01 -8.29273686e-02 -3.56889933e-01
2.60501534e-01 -1.09331153e-01 -8.73965859e-01 1.72444940e-01
2.46457070e-01 -6.55883074e-01 1.70398504e-01 -2.81627476e-01
3.06536555e-01 -7.73008049e-01 -3.20033580e-01 -5.11305094e-01
6.02518380e-01 6.53747320e-01 5.70437789e-01 6.60413563e-01
8.07395019e-03 1.57383120e+00 5.53952575e-01 4.69940394e-01
6.24206007e-01 7.59086549e-01 -2.49394253e-01 4.00672257e-01
2.62876421e-01 1.13050200e-01 4.87169027e-02 9.40348804e-01
5.94621710e-02 -9.26934630e-02 -1.52221823e+00 8.25339317e-01
-2.20497894e+00 -1.01769161e+00 -1.53106108e-01 1.47130537e+00
1.56028831e+00 -2.68637121e-01 3.79252434e-01 2.69501321e-02
4.08048421e-01 3.59190404e-01 -3.24555784e-01 -7.93078482e-01
1.41626626e-01 -1.14960134e-01 -2.71327496e-01 8.67376685e-01
-6.64941728e-01 8.80739331e-01 6.61229944e+00 3.93117070e-01
-3.61944854e-01 -3.34842533e-01 4.76624548e-01 2.61565387e-01
-4.40207541e-01 1.61210716e-01 -3.84861648e-01 2.08308190e-01
9.74462867e-01 -5.79669714e-01 7.30397403e-01 6.40509427e-01
5.69203049e-02 -4.92378026e-01 -1.57464612e+00 8.32872987e-01
7.67998621e-02 -1.38864744e+00 -3.22277814e-01 1.18489660e-01
1.43630549e-01 -1.63729191e-01 -6.44867778e-01 2.61845022e-01
6.53025210e-01 -9.67942119e-01 -5.45998663e-02 4.62891966e-01
-8.18213895e-02 -1.64160758e-01 7.51844287e-01 8.38838995e-01
-5.29324710e-01 -7.57184327e-02 7.18480870e-02 -2.75768638e-01
2.84711748e-01 1.37216747e-01 -1.56416833e+00 4.02236611e-01
5.78040779e-01 6.34780347e-01 -5.44085801e-01 5.74903905e-01
2.30126441e-01 2.92585820e-01 -1.25451565e-01 -6.17602825e-01
5.25253773e-01 -6.17747366e-01 8.49570096e-01 1.49957836e+00
-6.34255528e-01 3.40707630e-01 -3.09971813e-02 9.24036205e-01
1.68610848e-02 2.88330674e-01 -9.62797344e-01 -5.12745082e-01
6.57667994e-01 1.44011950e+00 -4.40690339e-01 -2.56408662e-01
-2.41245016e-01 8.51575673e-01 1.25796407e-01 1.21811971e-01
1.77073434e-01 -4.70390946e-01 5.63528597e-01 -3.27854127e-01
-5.08485794e-01 -3.52694131e-02 5.17814085e-02 -1.07030654e+00
-2.60302098e-03 -1.58759284e+00 4.38655376e-01 -6.68340921e-01
-1.51104832e+00 7.38092303e-01 1.58504561e-01 -6.30168438e-01
-1.21747375e+00 -2.23029647e-02 -5.72540581e-01 9.57247436e-01
-9.25344288e-01 -6.70865715e-01 -4.16644007e-01 4.50451344e-01
9.97841716e-01 -1.67840660e-01 1.18592501e+00 -5.53521179e-02
-3.37188393e-01 2.27249891e-01 -2.17317507e-01 4.48578477e-01
8.46849620e-01 -1.13516378e+00 4.08631653e-01 -1.92411125e-01
-6.32240623e-02 8.08612525e-01 7.17974126e-01 -4.40652609e-01
-1.70782506e+00 8.03145170e-02 1.36939120e+00 -9.38513637e-01
9.32002604e-01 -2.54400998e-01 -8.81618559e-01 6.53475642e-01
1.03633690e+00 -1.09529591e+00 1.05422044e+00 4.42854255e-01
7.09968284e-02 6.94715440e-01 -1.20609725e+00 8.37420046e-01
1.01305580e+00 -9.71458733e-01 -1.27003038e+00 6.05867922e-01
4.03162479e-01 -3.20039570e-01 -1.16038322e+00 3.76984663e-02
6.67288363e-01 -1.03822887e+00 7.16207623e-01 -8.62937212e-01
4.15901154e-01 5.69768488e-01 1.92895740e-01 -1.34445190e+00
2.45841920e-01 -1.33552778e+00 2.47281015e-01 1.35345614e+00
5.02686024e-01 -8.14409137e-01 6.28085554e-01 1.45141959e+00
5.89764044e-02 -5.87713063e-01 -7.81105578e-01 -1.27374828e-01
4.63729128e-02 1.01229064e-02 2.46635064e-01 1.53912222e+00
1.27609515e+00 1.02706003e+00 3.86477485e-02 -8.18083763e-01
1.91435352e-01 4.51542698e-02 1.04991591e+00 -1.60824037e+00
1.17326431e-01 -6.87264025e-01 8.56885388e-02 -1.30539405e+00
2.62098014e-01 -7.10930288e-01 1.44826651e-01 -1.55010450e+00
1.04113534e-01 -3.56646329e-01 7.93829918e-01 4.08082217e-01
8.50087851e-02 -5.03093243e-01 3.64810787e-02 5.90252757e-01
-4.66922849e-01 4.58716750e-01 1.07529604e+00 -2.01937824e-01
-6.84378982e-01 1.14433365e-02 -9.60437477e-01 6.27616465e-01
7.18974411e-01 -4.93920743e-02 -3.38809073e-01 6.09301925e-02
2.87953377e-01 6.47055566e-01 -1.07450791e-01 -1.38984010e-01
6.44505680e-01 -1.65315717e-01 -3.92902285e-01 -4.20252562e-01
2.54445106e-01 -4.92513150e-01 -1.93374202e-01 2.69311100e-01
-1.10627079e+00 2.52623726e-02 8.73575658e-02 2.47763216e-01
-2.96951383e-01 -5.98709583e-01 8.40940848e-02 -5.12447119e-01
-4.25503135e-01 -4.58287090e-01 -1.32648349e+00 3.97613198e-01
5.97068727e-01 -7.62864500e-02 -5.87254047e-01 -1.15186608e+00
-4.51684713e-01 5.96789062e-01 4.00073409e-01 5.66229820e-01
8.97138119e-02 -9.02788997e-01 -6.64578319e-01 -3.13722968e-01
3.06019038e-01 -1.64348796e-01 -2.85755485e-01 8.27093840e-01
-2.17733517e-01 8.79956663e-01 -1.14949211e-01 -4.93902892e-01
-1.51094735e+00 5.08942790e-02 9.38445851e-02 -2.51587123e-01
-5.82908750e-01 3.99712831e-01 -4.30488616e-01 -8.11503410e-01
3.64776075e-01 1.77020073e-01 -7.37718761e-01 6.13838613e-01
5.70024371e-01 3.56247902e-01 -1.15263447e-01 -4.68567759e-01
-2.24737823e-01 -1.01984590e-01 -3.49106103e-01 -6.61282480e-01
1.17312074e+00 -4.54303324e-01 -5.34802258e-01 8.51483166e-01
1.05335963e+00 -2.80779395e-02 -2.50136346e-01 -6.54888809e-01
8.52362990e-01 -3.42572242e-01 -2.92687953e-01 -9.60500479e-01
-2.16308117e-01 3.73223990e-01 -1.74811572e-01 1.33928299e+00
5.05960464e-01 3.55462879e-01 4.76525635e-01 1.11498332e+00
1.68855369e-01 -1.34337962e+00 8.48037750e-02 8.76783252e-01
1.23720324e+00 -1.41368246e+00 -1.44971743e-01 -4.91970837e-01
-9.93110061e-01 1.33092952e+00 3.62550586e-01 6.31165564e-01
4.69865143e-01 -9.88593549e-02 4.30172682e-01 -8.84661257e-01
-1.03125787e+00 -2.29274228e-01 1.87302530e-01 4.05896932e-01
8.93603384e-01 -4.25795734e-01 -8.76997709e-01 3.34843457e-01
-4.84294236e-01 -2.10021973e-01 7.63689220e-01 1.27781129e+00
-5.08964397e-02 -1.29582703e+00 -5.05758710e-02 4.71990287e-01
-1.30264819e-01 -1.81391865e-01 -1.86248565e+00 8.94896030e-01
-1.12517881e+00 1.60938466e+00 -6.67224452e-02 -2.67238021e-01
2.52000481e-01 6.15606427e-01 9.58780721e-02 -5.56858897e-01
-1.25743389e+00 -3.73867571e-01 1.25172961e+00 -4.49235022e-01
-9.47686315e-01 -8.26855183e-01 -6.33440912e-01 -3.73658329e-01
-3.11910212e-01 8.81198585e-01 6.04107559e-01 9.40361142e-01
5.19688129e-01 -6.05105758e-02 4.59604502e-01 -6.04183018e-01
-4.20263529e-01 -1.53464293e+00 -3.87411788e-02 5.58568537e-01
1.39715299e-01 -2.24622428e-01 -5.53533554e-01 -1.27451435e-01] | [12.383711814880371, 7.906962871551514] |
39f7c223-adf5-4d00-bf04-7d7f15e6e7af | a-generative-language-model-for-few-shot-1 | 2204.05356 | null | https://arxiv.org/abs/2204.05356v1 | https://arxiv.org/pdf/2204.05356v1.pdf | A Generative Language Model for Few-shot Aspect-Based Sentiment Analysis | Sentiment analysis is an important task in natural language processing. In recent works, pre-trained language models are often used to achieve state-of-the-art results, especially when training data is scarce. It is common to fine-tune on the downstream task, usually by adding task-specific layers on top of the model. In this paper, we focus on aspect-based sentiment analysis, which involves extracting aspect term, category, and predicting their corresponding polarities. In particular, we are interested in few-shot settings. We propose to reformulate the extraction and prediction tasks into the sequence generation task, using a generative language model with unidirectional attention (GPT2 is used unless stated otherwise). This way, the model learns to accomplish the tasks via language generation without the need of training task-specific layers. Our evaluation results on the single-task polarity prediction show that our approach outperforms the previous state-of-the-art (based on BERT) on average performance by a large margins in few-shot and full-shot settings. More importantly, our generative approach significantly reduces the model variance caused by low-resource data. We further demonstrate that the proposed generative language model can handle joint and multi-task settings, unlike previous work. We observe that the proposed sequence generation method achieves further improved performances on polarity prediction when the model is trained via joint and multi-task settings. Further evaluation on similar sentiment analysis datasets, SST-2, SST- and OOS intent detection validates the superiority and noise robustness of generative language model in few-shot settings. | ['Caiming Xiong', 'Wenhao Liu', 'Ehsan Hosseini-Asl'] | 2022-04-11 | null | https://aclanthology.org/2022.findings-naacl.58 | https://aclanthology.org/2022.findings-naacl.58.pdf | findings-naacl-2022-7 | ['aspect-based-sentiment-analysis'] | ['natural-language-processing'] | [ 5.16275644e-01 1.34545028e-01 -1.68392107e-01 -4.59625870e-01
-1.12622702e+00 -5.69730818e-01 8.98379922e-01 -7.26127401e-02
-3.28728050e-01 6.37411892e-01 3.41798037e-01 -1.85897678e-01
3.27652037e-01 -6.95236087e-01 -5.06664932e-01 -7.80047596e-01
4.23829854e-01 4.89251047e-01 5.82879118e-04 -6.24730527e-01
3.36214304e-01 -2.92872220e-01 -1.50276721e+00 4.31016862e-01
8.60718906e-01 9.33608115e-01 2.54911870e-01 6.33535206e-01
-2.88267881e-01 6.77925885e-01 -4.35113341e-01 -8.94423485e-01
1.05134374e-03 -3.46040219e-01 -5.27376413e-01 6.27580956e-02
-5.68244047e-02 1.76578999e-01 4.58202213e-01 8.58275592e-01
7.57422745e-01 9.56076086e-02 7.90684164e-01 -1.04566085e+00
-6.47308171e-01 6.18812263e-01 -6.59669220e-01 -1.32973596e-01
1.89625144e-01 4.64467295e-02 1.45869720e+00 -1.02727902e+00
5.64966142e-01 1.06046748e+00 5.46455383e-01 5.04936337e-01
-1.14820313e+00 -5.29538095e-01 5.12101293e-01 1.65305808e-01
-8.88473332e-01 -3.88234109e-01 1.00681365e+00 -4.05151069e-01
1.19619322e+00 -1.37311786e-01 3.09296191e-01 1.53261638e+00
4.40756917e-01 9.40509439e-01 1.19557524e+00 -4.15363550e-01
2.52508521e-01 3.16701680e-01 2.95758694e-01 3.79427582e-01
7.49598891e-02 -4.55540031e-01 -6.46136701e-01 -1.18709899e-01
-3.63270529e-02 -1.02892779e-01 -1.45886794e-01 -2.42817089e-01
-1.15646148e+00 1.01072204e+00 -2.30154884e-03 3.31111252e-01
-4.68280852e-01 -2.89912038e-02 7.24237680e-01 3.35287213e-01
1.01959264e+00 5.07667720e-01 -7.59826243e-01 -3.76400560e-01
-1.00911605e+00 1.25387430e-01 1.00062311e+00 9.36443865e-01
5.93743563e-01 1.47629172e-01 -6.37487531e-01 9.90911067e-01
1.63876474e-01 4.56897497e-01 8.47217023e-01 -1.36728764e-01
6.41147256e-01 5.23087680e-01 -3.09045166e-02 -6.03516698e-01
-4.64287370e-01 -6.04137719e-01 -7.97281802e-01 -2.61397690e-01
4.16894294e-02 -6.04838848e-01 -1.14419746e+00 1.83646667e+00
1.88265577e-01 4.99528833e-02 4.22722161e-01 7.03762591e-01
7.80346632e-01 7.93455184e-01 1.71104342e-01 -4.65383083e-01
1.80525923e+00 -1.14019799e+00 -7.85254538e-01 -7.31101394e-01
8.37656915e-01 -8.54508400e-01 1.43170047e+00 3.01272899e-01
-7.49766588e-01 -4.26669687e-01 -8.37310016e-01 -5.95843755e-02
-5.02662063e-01 4.05091494e-01 6.50888741e-01 7.10319698e-01
-8.63821268e-01 2.34905824e-01 -5.60520470e-01 -2.56902963e-01
3.22218448e-01 1.83658674e-01 -1.19600974e-01 2.29474008e-01
-1.38528168e+00 6.81642652e-01 1.16233237e-01 1.59507263e-02
-8.49691272e-01 -7.21302688e-01 -1.04119182e+00 3.67658943e-01
5.94496429e-01 -9.08497870e-01 1.43823206e+00 -8.19905341e-01
-1.89164400e+00 6.79539621e-01 -5.41350067e-01 -3.51146460e-01
2.78631538e-01 -4.05623376e-01 -1.21495090e-01 -2.51191646e-01
1.52185857e-01 4.20307308e-01 1.11872840e+00 -9.64789450e-01
-4.94991064e-01 -1.88814253e-01 7.82966390e-02 3.76774877e-01
-6.15304172e-01 4.73842816e-03 -4.94323045e-01 -7.23093569e-01
-3.96201104e-01 -1.12628376e+00 -2.53010362e-01 -8.10226738e-01
-5.54336190e-01 -3.35988492e-01 6.03161275e-01 -3.73832405e-01
1.13906741e+00 -2.03453565e+00 1.43919259e-01 -1.31144747e-01
-2.00549394e-01 2.62837768e-01 -2.20331267e-01 4.46964890e-01
-1.21852972e-01 7.76917636e-02 -4.00031984e-01 -9.10205960e-01
1.48659408e-01 -1.73354715e-01 -5.32642186e-01 9.47129056e-02
3.56023341e-01 1.08108950e+00 -7.58366287e-01 -2.23002151e-01
-1.50368601e-01 5.22626758e-01 -6.58609390e-01 8.82207677e-02
-3.71960104e-01 2.63569504e-01 -4.52281237e-01 5.98689139e-01
3.77553344e-01 -3.68540913e-01 3.65369469e-01 -1.78215131e-01
1.22335024e-01 5.48636794e-01 -8.64940882e-01 1.58857179e+00
-1.03589547e+00 3.71925354e-01 -1.79897279e-01 -1.06778824e+00
9.20579076e-01 5.25619030e-01 5.53592257e-02 -5.98551810e-01
2.68239081e-01 1.53513074e-01 1.53157443e-01 -3.74193847e-01
6.69251204e-01 -5.11848271e-01 -7.04661667e-01 5.91693223e-01
3.81350458e-01 -3.48475456e-01 4.76105690e-01 1.68870062e-01
7.00009108e-01 1.18187919e-01 5.42717218e-01 -1.71557605e-01
5.94885409e-01 -1.70650050e-01 4.97311950e-01 5.92071831e-01
1.59132421e-01 5.04919410e-01 9.71836746e-01 6.71272352e-02
-9.45380211e-01 -4.77933705e-01 6.31089136e-02 1.48074877e+00
-6.96211010e-02 -5.04408360e-01 -5.75061619e-01 -8.41490567e-01
-4.34951633e-01 9.65395570e-01 -7.89912581e-01 -7.87848383e-02
-2.92262346e-01 -1.39283383e+00 1.59359530e-01 4.85683620e-01
2.89259285e-01 -1.23272729e+00 -2.65304267e-01 3.02147865e-01
-2.26641431e-01 -1.39425790e+00 -3.13965172e-01 3.94033819e-01
-7.14733958e-01 -6.44926429e-01 -9.05647635e-01 -7.86062241e-01
4.43286598e-01 7.08114505e-02 1.01974785e+00 -5.67034364e-01
2.39378765e-01 1.35685086e-01 -5.56109548e-01 -5.57814896e-01
-2.13384047e-01 5.99034429e-01 -3.14787090e-01 3.66103113e-01
4.01462495e-01 -6.14271998e-01 -4.68205601e-01 1.01883635e-02
-7.21176744e-01 2.40756094e-01 8.04701388e-01 1.16945660e+00
6.00402355e-01 -3.55194271e-01 8.71887088e-01 -1.38732290e+00
1.09423637e+00 -4.81603742e-01 -3.96177351e-01 2.07636029e-01
-7.53586292e-01 2.41992369e-01 8.18169713e-01 -3.75080675e-01
-1.38757086e+00 -4.23824079e-02 -1.21375255e-01 -1.47109911e-01
-6.37372360e-02 7.17367291e-01 -1.76595882e-01 4.10381407e-01
3.27997774e-01 3.99129242e-01 -2.74251610e-01 -2.49516279e-01
4.21696275e-01 7.47300446e-01 -1.93119813e-02 -2.29558244e-01
4.76770341e-01 4.11356002e-01 -4.55996543e-02 -6.27141356e-01
-1.27103794e+00 -5.89677572e-01 -2.90629536e-01 9.62780640e-02
8.39249790e-01 -1.22771287e+00 -4.63540822e-01 5.73740184e-01
-1.07010829e+00 -3.88151109e-01 -3.93300988e-02 2.54653841e-01
-5.84939480e-01 9.89086181e-02 -6.35590076e-01 -9.20574307e-01
-9.12975371e-01 -1.33268929e+00 1.57333410e+00 2.96761785e-02
-3.11345607e-01 -1.12833965e+00 3.50963503e-01 3.89379680e-01
3.67010444e-01 -2.09544986e-01 8.58634830e-01 -1.00311494e+00
-1.42499954e-01 -1.15692258e-01 -8.26107711e-03 2.61550248e-01
2.36231405e-02 -3.17224473e-01 -1.32915425e+00 9.95634124e-03
1.30527779e-01 -4.38696414e-01 1.15059304e+00 4.01701689e-01
9.62319016e-01 -1.93805307e-01 -1.93795308e-01 3.72876644e-01
1.29412532e+00 -2.71027908e-03 5.24675727e-01 2.64319599e-01
7.08946884e-01 7.39237189e-01 8.57517302e-01 4.41331595e-01
6.32102668e-01 6.85681641e-01 1.59732014e-01 -6.35966361e-02
5.52191623e-02 -9.36316699e-02 6.17021382e-01 8.30217004e-01
2.24193651e-02 -5.28069973e-01 -7.50753641e-01 7.43787706e-01
-2.02921510e+00 -9.49101865e-01 -2.31225695e-02 1.87839675e+00
9.56602275e-01 3.65354389e-01 -3.49879451e-02 1.61231875e-01
4.66725022e-01 4.64386374e-01 -3.31759661e-01 -7.64526546e-01
-7.89453313e-02 2.23527357e-01 1.56989276e-01 4.48873699e-01
-1.25714123e+00 1.26748562e+00 5.12114143e+00 9.83487964e-01
-1.35145760e+00 4.38329488e-01 8.96305144e-01 -3.23120356e-01
-4.52946395e-01 -5.84318005e-02 -1.14027178e+00 4.77701247e-01
9.17974353e-01 -2.94201821e-01 3.01717259e-02 9.97961521e-01
2.73788989e-01 7.27582257e-03 -9.02697623e-01 6.86634004e-01
4.78076756e-01 -9.88428414e-01 2.05573998e-02 -1.75303802e-01
9.72837806e-01 -7.61504285e-03 1.99431822e-01 8.07786226e-01
9.24675763e-02 -6.27501249e-01 5.93774438e-01 1.98439255e-01
7.06053913e-01 -6.92608714e-01 1.05658293e+00 2.72427708e-01
-1.10452712e+00 -1.45289555e-01 -1.50162175e-01 -6.67891577e-02
6.20779574e-01 9.81829464e-01 -1.03614211e+00 5.39573550e-01
4.57193583e-01 7.86443949e-01 -3.09124619e-01 5.25682509e-01
-6.74801111e-01 7.62262583e-01 -1.16510328e-03 -4.49745864e-01
2.95032054e-01 -1.67801932e-01 5.70549667e-01 1.43845665e+00
4.62835670e-01 -2.08886355e-01 -3.00135370e-02 4.98693436e-01
-2.96033442e-01 3.60178500e-01 -5.28851688e-01 -1.14778727e-01
-8.27113464e-02 1.71503091e+00 -7.59191155e-01 -6.12138748e-01
-4.84467000e-01 9.30025101e-01 3.38297874e-01 4.00017470e-01
-8.32778335e-01 -4.12468910e-01 6.50379241e-01 -1.63716614e-01
7.88085938e-01 9.56778377e-02 -4.87888783e-01 -1.36372352e+00
-3.28034982e-02 -8.95106852e-01 1.44489363e-01 -7.29272842e-01
-1.24236882e+00 8.11960340e-01 -1.07011214e-01 -1.18328834e+00
-7.55865753e-01 -5.52343071e-01 -8.19734573e-01 8.46632183e-01
-1.75920892e+00 -1.30980051e+00 -4.54104226e-03 3.22680563e-01
9.38640177e-01 -5.04666045e-02 7.81629801e-01 1.08930163e-01
-5.68022549e-01 6.06655478e-01 -7.93784782e-02 -1.68858960e-01
9.67366099e-01 -1.19157791e+00 4.34234858e-01 9.48084116e-01
4.45269793e-02 4.91353184e-01 7.89630175e-01 -6.97630048e-01
-1.19044244e+00 -1.12770772e+00 1.29610860e+00 -2.79250413e-01
9.38304961e-01 -9.26666915e-01 -6.00237310e-01 5.47646224e-01
3.41589153e-01 -3.46047193e-01 8.84708762e-01 5.06936252e-01
-1.96905479e-01 1.18629783e-02 -6.69962645e-01 7.49086142e-01
8.61254752e-01 -5.09141684e-01 -7.13002920e-01 2.94229954e-01
9.47166741e-01 -2.47800902e-01 -4.52178657e-01 5.27523875e-01
4.68195289e-01 -6.17527187e-01 5.35521805e-01 -3.69872719e-01
8.12673628e-01 -4.25734930e-02 9.26885381e-02 -1.64820993e+00
-1.90245152e-01 -6.30265832e-01 5.81411235e-02 1.31733549e+00
9.71094728e-01 -6.09185398e-01 5.92413723e-01 2.86017746e-01
-2.40826264e-01 -9.20124173e-01 -6.17606580e-01 -3.87081742e-01
-2.08373159e-01 -5.81359982e-01 3.36095810e-01 7.06389248e-01
1.52093843e-01 1.14003897e+00 -5.72079718e-01 -1.41954616e-01
2.13856414e-01 6.71765924e-01 7.49240994e-01 -9.60592151e-01
-5.24851143e-01 -4.29589719e-01 4.32499796e-02 -9.22135353e-01
3.44721735e-01 -7.53372312e-01 2.67657846e-01 -1.55462646e+00
2.70579219e-01 -4.28639986e-02 -2.09292889e-01 6.67875767e-01
-6.07880354e-01 3.14998716e-01 3.21755946e-01 -3.16133760e-02
-7.15631902e-01 1.00177121e+00 1.22486651e+00 -5.45556769e-02
-3.54495376e-01 9.83279720e-02 -1.14759457e+00 5.78953564e-01
8.08302164e-01 -4.32504296e-01 -6.03901029e-01 -1.11354031e-01
6.90884471e-01 -2.63128549e-01 -7.49498010e-02 -5.77587128e-01
5.57163283e-02 1.23653553e-01 -1.05416119e-01 -3.59323174e-01
5.67135990e-01 -3.83783907e-01 -3.82204503e-01 1.30844772e-01
-3.15673977e-01 -2.03216791e-01 2.27980599e-01 5.34938514e-01
-3.33900273e-01 -2.38295659e-01 3.65340084e-01 7.21720755e-02
-5.98923087e-01 2.21615583e-01 -4.61167842e-01 1.15557939e-01
9.28152144e-01 1.34290114e-01 -3.66906792e-01 -5.32853186e-01
-6.30183399e-01 1.09451726e-01 1.01065747e-01 5.42258501e-01
3.00496161e-01 -9.14891899e-01 -6.39628768e-01 2.20242411e-01
3.54761481e-01 -9.41018313e-02 4.11644816e-01 1.20945466e+00
3.48558426e-01 5.75329065e-01 3.09928566e-01 -4.24923956e-01
-1.21786010e+00 4.40796971e-01 -3.76939401e-02 -9.28327739e-01
-1.34363398e-01 7.87666321e-01 6.09401345e-01 -4.10531253e-01
-2.31027484e-01 -3.17633033e-01 -6.41710520e-01 6.53468668e-01
3.04110140e-01 -1.49415106e-01 3.44413161e-01 -4.12796021e-01
-2.53029048e-01 5.68503976e-01 -3.39875579e-01 -2.99611688e-01
1.49174535e+00 -5.01913726e-02 1.27566531e-01 5.77527165e-01
1.03660095e+00 1.18468381e-01 -1.00281751e+00 -1.65566225e-02
-2.34725475e-01 1.18825644e-01 1.10672489e-01 -7.31106043e-01
-8.78068388e-01 1.03540468e+00 8.60449597e-02 2.17871487e-01
1.17887878e+00 8.79791006e-02 6.80389881e-01 2.85059065e-01
1.06006481e-01 -1.08773065e+00 8.71876255e-02 8.42252016e-01
6.67682707e-01 -1.38390112e+00 -2.29549438e-01 -4.14908260e-01
-1.20544338e+00 6.79900229e-01 5.07957399e-01 -1.13202304e-01
5.52892804e-01 2.62903482e-01 5.57962805e-02 -2.35537514e-01
-1.21094584e+00 -3.34906846e-01 4.22123432e-01 2.74929732e-01
8.23554635e-01 2.95639578e-02 -6.43508434e-01 1.05639625e+00
-4.34410661e-01 -7.98626691e-02 4.58235741e-01 7.79950142e-01
-1.55528754e-01 -1.25689781e+00 1.05700856e-02 5.52583337e-01
-7.72513688e-01 -7.02049017e-01 -2.71742553e-01 4.61177379e-01
-2.60897666e-01 9.17847812e-01 -9.64011177e-02 -2.46215895e-01
2.80909806e-01 4.07002896e-01 5.03823441e-03 -9.04045403e-01
-8.36786211e-01 3.03527534e-01 4.21377301e-01 -5.31394184e-01
-5.31033695e-01 -6.04351342e-01 -8.80758047e-01 2.97571748e-01
-4.46667165e-01 1.17917918e-01 7.59026766e-01 1.03029954e+00
7.18479574e-01 8.05239201e-01 5.98793089e-01 -9.13839042e-01
-5.43450952e-01 -1.23305786e+00 -3.56374860e-01 3.22106302e-01
8.87771696e-02 -5.51111937e-01 -4.46279198e-01 1.20272107e-01] | [11.51556396484375, 6.721602439880371] |
fe8a2f10-ecdb-4d25-a2a9-a7226c4d9bc6 | dynamic-spatio-temporal-specialization | 2209.01425 | null | https://arxiv.org/abs/2209.01425v1 | https://arxiv.org/pdf/2209.01425v1.pdf | Dynamic Spatio-Temporal Specialization Learning for Fine-Grained Action Recognition | The goal of fine-grained action recognition is to successfully discriminate between action categories with subtle differences. To tackle this, we derive inspiration from the human visual system which contains specialized regions in the brain that are dedicated towards handling specific tasks. We design a novel Dynamic Spatio-Temporal Specialization (DSTS) module, which consists of specialized neurons that are only activated for a subset of samples that are highly similar. During training, the loss forces the specialized neurons to learn discriminative fine-grained differences to distinguish between these similar samples, improving fine-grained recognition. Moreover, a spatio-temporal specialization method further optimizes the architectures of the specialized neurons to capture either more spatial or temporal fine-grained information, to better tackle the large range of spatio-temporal variations in the videos. Lastly, we design an Upstream-Downstream Learning algorithm to optimize our model's dynamic decisions during training, improving the performance of our DSTS module. We obtain state-of-the-art performance on two widely-used fine-grained action recognition datasets. | ['Jun Liu', 'Jinghua Wang', 'Anran Wang', 'Hossein Rahmani', 'Qiuhong Ke', 'Lin Geng Foo', 'Tianjiao Li'] | 2022-09-03 | null | null | null | null | ['fine-grained-action-recognition'] | ['computer-vision'] | [ 3.41209978e-01 -4.01570231e-01 -3.19027126e-01 -6.30084395e-01
-3.84874016e-01 -5.22290647e-01 6.31608129e-01 -2.36013159e-01
-5.63471258e-01 4.62696224e-01 5.04104435e-01 2.68051296e-01
-2.01328292e-01 -5.75751066e-01 -7.12442517e-01 -7.84914970e-01
-1.50952250e-01 1.56792730e-01 6.66593313e-01 1.78836230e-02
3.85036290e-01 8.01141560e-01 -1.86232567e+00 1.00733757e+00
8.00800204e-01 1.37855506e+00 2.88290322e-01 5.68117201e-01
2.40279827e-02 7.54112244e-01 -6.06588840e-01 1.34262457e-01
3.04218322e-01 -5.09565234e-01 -8.12891841e-01 3.08592528e-01
7.96741903e-01 -1.84200644e-01 -3.08821291e-01 8.73609781e-01
2.28462115e-01 3.89475256e-01 6.39754415e-01 -9.40614760e-01
-5.89516580e-01 2.74074495e-01 -3.17615896e-01 7.84039497e-01
1.11336976e-01 5.91319740e-01 1.08919311e+00 -7.26702690e-01
3.69837433e-01 1.44253898e+00 3.55536401e-01 6.33603156e-01
-1.08653724e+00 -4.19621527e-01 7.92521834e-01 6.43352270e-01
-1.18224716e+00 -5.43088675e-01 5.45720935e-01 -6.22679055e-01
1.17224514e+00 1.34273499e-01 7.12567866e-01 1.17196465e+00
3.11302841e-01 1.02392626e+00 1.02286100e+00 6.61931410e-02
3.28317285e-01 -5.42691827e-01 8.38123336e-02 5.43125451e-01
-2.85684884e-01 9.90243554e-02 -7.30252564e-01 2.30666146e-01
8.35022569e-01 3.26181084e-01 -1.99554116e-01 -4.27375257e-01
-1.31752622e+00 5.89160323e-01 5.36853671e-01 5.68713903e-01
-5.03773570e-01 5.28527386e-02 4.54813838e-01 2.22742274e-01
1.06172778e-01 5.65089405e-01 -7.74650633e-01 -3.63472283e-01
-7.97342002e-01 -1.95134114e-02 3.54904681e-01 4.92429703e-01
8.50940764e-01 -8.89896974e-02 -1.02908754e+00 8.49312961e-01
2.04707250e-01 3.83511037e-02 7.62660861e-01 -1.01899874e+00
5.40109456e-01 1.09281039e+00 -2.43453175e-01 -6.86358690e-01
-3.75630289e-01 -5.87662220e-01 -8.97547960e-01 3.29878777e-01
6.21365190e-01 3.47787797e-01 -1.06248391e+00 1.61546075e+00
2.12015942e-01 4.69927400e-01 -2.35059559e-01 1.04446602e+00
4.90248471e-01 2.86776125e-01 3.12026918e-01 4.72992249e-02
1.23093426e+00 -1.28771484e+00 -9.00204703e-02 -4.64456260e-01
6.29650474e-01 -2.17097312e-01 1.50708151e+00 1.36696920e-01
-8.15561593e-01 -8.66430223e-01 -7.70277262e-01 2.19341125e-02
-3.13555300e-01 2.75179386e-01 4.52703714e-01 1.94835082e-01
-6.71361446e-01 1.00484443e+00 -9.32638288e-01 -7.15710819e-02
1.01061380e+00 2.42940232e-01 -4.42240149e-01 6.92443475e-02
-1.00407922e+00 6.32845700e-01 4.17197376e-01 8.85780826e-02
-1.16360271e+00 -7.89616883e-01 -7.27455795e-01 2.26087138e-01
4.91115719e-01 -6.42165780e-01 9.80243266e-01 -1.28193367e+00
-1.49206948e+00 9.08807278e-01 -3.54478508e-02 -4.54758257e-01
3.82149100e-01 -1.06566556e-01 -2.53317297e-01 2.26771012e-01
2.32549027e-01 5.46341896e-01 1.19957995e+00 -7.00204730e-01
-9.22245026e-01 -5.34941494e-01 2.70021677e-01 3.73220518e-02
-2.09529445e-01 -1.29363745e-01 -5.44110179e-01 -1.06701243e+00
-1.46885723e-01 -8.04865301e-01 -2.42396265e-01 1.27845183e-01
-1.31972656e-01 -4.85250682e-01 7.95647264e-01 -2.86947846e-01
1.22735047e+00 -2.33245325e+00 5.34216106e-01 -1.49918675e-01
7.17814788e-02 5.71591794e-01 -6.27859056e-01 -1.38594583e-02
-5.89808300e-02 -1.72567815e-01 -1.27883852e-01 -2.39709824e-01
4.83427234e-02 4.58855093e-01 -1.37640148e-01 2.93954909e-01
5.63675761e-01 1.13205552e+00 -7.67988682e-01 -2.03792989e-01
3.10084939e-01 1.66252732e-01 -5.26954293e-01 2.02580214e-01
-3.90029699e-01 7.81262815e-01 -7.82174408e-01 7.09247649e-01
2.39478424e-01 -4.20629710e-01 -2.60120425e-02 -4.02718157e-01
2.32842322e-02 1.33392826e-01 -1.05825901e+00 1.78862143e+00
-4.31219906e-01 3.28428268e-01 -1.36926800e-01 -1.24454486e+00
6.55766606e-01 -2.64624536e-01 4.26061124e-01 -1.08849013e+00
1.07761525e-01 -4.75496380e-03 5.80016337e-02 -5.23386419e-01
5.57483360e-02 4.80656624e-02 -7.27215186e-02 4.76793051e-01
1.15340717e-01 3.18210363e-01 3.55770260e-01 -1.95982307e-01
1.25333822e+00 2.67047137e-01 1.98237389e-01 -1.69245958e-01
7.57085562e-01 -3.42581779e-01 7.38400877e-01 8.10380340e-01
-6.23369634e-01 4.81517226e-01 4.40485805e-01 -7.56378770e-01
-5.29855490e-01 -8.10481250e-01 1.38732076e-01 1.72174847e+00
1.08895727e-01 -3.10117751e-01 -6.27387643e-01 -1.17863023e+00
1.23753294e-01 3.55088264e-01 -1.08590984e+00 -5.55892825e-01
-6.72568560e-01 -3.26136976e-01 4.50368136e-01 8.76778364e-01
7.55972028e-01 -1.47014713e+00 -1.04959607e+00 1.09486163e-01
7.58969411e-02 -1.05197728e+00 -6.69311285e-01 3.38098764e-01
-8.95538449e-01 -1.17063403e+00 -7.73711741e-01 -5.02991378e-01
5.11725247e-01 2.41134465e-01 1.19483757e+00 3.48547846e-02
-3.43893021e-01 2.48184741e-01 -6.11974478e-01 1.88957065e-01
4.91439663e-02 1.00939814e-02 -1.23356439e-01 5.45930028e-01
4.55537111e-01 -6.22402549e-01 -6.38035476e-01 6.77386463e-01
-9.18486714e-01 -2.13602334e-01 7.67448902e-01 9.76171672e-01
8.60655427e-01 1.54271141e-01 2.28784278e-01 -4.61233884e-01
2.98260719e-01 -1.69715822e-01 -2.66080588e-01 3.87649536e-01
-6.48079365e-02 5.26954651e-01 9.03189659e-01 -6.58873379e-01
-7.12483108e-01 1.07849337e-01 2.30588503e-02 -5.97196281e-01
-5.60413778e-01 2.15997800e-01 -3.27332437e-01 -1.11310884e-01
6.23482049e-01 4.97150928e-01 -3.68320763e-01 -7.50974715e-01
1.80658370e-01 3.43582779e-01 3.64037424e-01 -4.82181013e-01
3.56126547e-01 5.19335985e-01 -1.28071159e-01 -5.58937430e-01
-1.22939920e+00 -4.67427820e-01 -1.03727317e+00 -6.28463104e-02
1.10884941e+00 -7.36439228e-01 -6.00307763e-01 8.28260124e-01
-7.07393169e-01 -9.40372825e-01 -6.64279699e-01 3.79376262e-01
-6.91803694e-01 2.61571556e-01 -2.44062364e-01 -1.71106011e-01
1.14160463e-01 -1.21610415e+00 1.48570967e+00 3.41367751e-01
-9.95463729e-02 -7.90791690e-01 1.61934227e-01 2.73618519e-01
2.40429729e-01 -3.91717441e-02 7.97562480e-01 -4.63597208e-01
-6.06871843e-01 2.14919612e-01 -3.59581530e-01 5.55579722e-01
2.70588130e-01 -1.18886046e-01 -7.15567231e-01 -3.60248685e-01
-3.39967847e-01 -4.48064417e-01 1.44829428e+00 3.72915655e-01
1.68095946e+00 -2.81299978e-01 -3.51855606e-01 9.11810696e-01
8.78285766e-01 6.86175302e-02 6.42683923e-01 4.66216147e-01
8.01060975e-01 4.53260362e-01 8.38548183e-01 6.14245117e-01
1.21236458e-01 9.02740061e-01 4.32601899e-01 2.61752754e-01
-2.98901469e-01 -9.81259421e-02 3.48328322e-01 3.56201492e-02
-3.16638023e-01 1.11726277e-01 -4.86777604e-01 4.87089396e-01
-2.05879331e+00 -1.17770636e+00 4.29241717e-01 1.92107642e+00
7.72962868e-01 1.44275293e-01 3.28811228e-01 -2.55663320e-03
4.55559015e-01 6.66875362e-01 -9.67015803e-01 -1.45124599e-01
-1.01787172e-01 2.70651877e-01 2.89824635e-01 1.91119555e-02
-1.44530356e+00 1.14795125e+00 5.96562099e+00 1.11624551e+00
-1.39460003e+00 -1.37351364e-01 5.22453249e-01 -2.10250348e-01
-1.12131469e-01 -4.08992976e-01 -7.40506649e-01 5.61120927e-01
6.67361081e-01 4.21135634e-01 4.22116220e-01 9.01552022e-01
2.69829601e-01 -1.69212893e-02 -1.29584169e+00 9.69898164e-01
-8.03657398e-02 -1.49235535e+00 2.11410239e-01 -1.35737807e-01
8.01954567e-01 6.76246583e-02 -3.56417149e-02 4.51931655e-01
5.51602170e-02 -1.12061441e+00 7.38185406e-01 7.58069336e-01
6.46115959e-01 -5.32308638e-01 2.95923293e-01 3.72904330e-01
-1.40031862e+00 -4.54639941e-01 -4.90248710e-01 -1.84181724e-02
-5.57671599e-02 4.11705017e-01 -2.14377895e-01 1.83436841e-01
9.73771513e-01 1.21916950e+00 -7.97313035e-01 9.77133274e-01
-3.05407792e-01 2.62151331e-01 1.04251370e-01 2.00916320e-01
4.96594995e-01 -9.44000036e-02 3.64740163e-01 1.21062112e+00
4.20113429e-02 3.58010568e-02 2.58132815e-01 5.40624321e-01
1.96293145e-01 -3.20313662e-01 -1.77558213e-01 -1.97978348e-01
1.58969965e-02 1.08169699e+00 -5.94863474e-01 -3.67575169e-01
-2.89501995e-01 1.29631507e+00 6.77377284e-01 3.84081125e-01
-7.41845846e-01 -1.25514403e-01 1.21486676e+00 1.18452003e-02
9.64996576e-01 -3.32416177e-01 -1.50798112e-01 -1.37918174e+00
7.69988894e-02 -1.15952563e+00 7.10788369e-01 -4.11068678e-01
-1.36117959e+00 6.72215998e-01 -3.41091573e-01 -1.21379602e+00
-1.95812419e-01 -7.81020105e-01 -6.67617917e-01 6.52582288e-01
-1.40905535e+00 -1.19459164e+00 -4.42025661e-01 1.15418088e+00
8.60243440e-01 -2.95776486e-01 6.70204461e-01 3.22940022e-01
-6.11639202e-01 7.13316441e-01 -1.78519487e-01 1.01367071e-01
5.85494161e-01 -1.10494435e+00 3.14842612e-01 8.89424443e-01
2.45596603e-01 3.50515544e-01 2.37388730e-01 -5.12005925e-01
-1.10768080e+00 -1.36550069e+00 5.53819478e-01 -3.16112339e-01
6.21943831e-01 -3.26511413e-01 -9.32710469e-01 4.07168686e-01
-4.55729276e-01 4.58188772e-01 5.25985658e-01 1.36017546e-01
-8.19398761e-01 -2.83864588e-01 -8.59791458e-01 4.97041941e-01
1.43202209e+00 -8.02471459e-01 -6.98124886e-01 9.34583172e-02
4.31424230e-01 -7.38270953e-02 -6.85380816e-01 4.00116026e-01
8.03707063e-01 -1.22678626e+00 1.05815947e+00 -1.09436023e+00
3.42687726e-01 -5.23055732e-01 -1.98511705e-01 -1.45680189e+00
-8.80981743e-01 -3.04122925e-01 -3.00007790e-01 6.86171472e-01
1.21414654e-01 -5.25489986e-01 8.90507042e-01 1.97553635e-01
-3.86691660e-01 -8.57808292e-01 -1.11472404e+00 -9.28053021e-01
-1.17200673e-01 -3.20372641e-01 5.73312044e-01 6.57079637e-01
-3.97628248e-01 1.44964427e-01 -2.78716862e-01 -4.90530655e-02
2.49504879e-01 6.10483646e-01 4.76840973e-01 -1.10866380e+00
-6.68411255e-01 -7.61449933e-01 -6.96327567e-01 -1.39560986e+00
3.14655811e-01 -6.24880075e-01 9.95503739e-02 -1.26584423e+00
2.06115514e-01 -2.23792359e-01 -6.04638278e-01 8.28477621e-01
-3.63231778e-01 4.95699912e-01 3.50482523e-01 1.96285561e-01
-1.04213393e+00 7.47531652e-01 1.64070761e+00 -4.09840852e-01
-2.21586585e-01 2.37107903e-01 -6.35232866e-01 6.48888648e-01
5.59077799e-01 -2.18489558e-01 -3.30871075e-01 -4.82750654e-01
-2.65021533e-01 -3.22222263e-01 6.12699986e-01 -1.19328523e+00
-5.99843673e-02 -6.18467212e-01 6.45695806e-01 -4.06082392e-01
2.22020343e-01 -7.16262102e-01 -1.61386058e-01 4.54331011e-01
-5.93909740e-01 -4.43008393e-01 1.11678071e-01 5.26404083e-01
-4.06080455e-01 8.56849775e-02 1.08305299e+00 -2.40503401e-01
-1.35598671e+00 8.10291529e-01 -3.52791667e-01 2.87496150e-01
1.26960051e+00 -4.40820724e-01 -2.52618998e-01 1.14523582e-01
-9.52245474e-01 1.15455270e-01 3.88433635e-01 6.58799529e-01
4.00141448e-01 -1.52328205e+00 -4.92901355e-01 4.30831194e-01
3.52734774e-01 -1.78154275e-01 7.71343708e-01 9.51271653e-01
6.51354790e-02 5.09366512e-01 -7.98474133e-01 -7.19982624e-01
-1.25366545e+00 6.07997537e-01 4.98034954e-01 -4.04813141e-01
-5.50631344e-01 1.11172891e+00 6.58064544e-01 -1.98952138e-01
1.82075903e-01 -6.75987720e-01 -4.26218003e-01 2.19051942e-01
8.03270161e-01 3.69883746e-01 -5.06606735e-02 -7.65226781e-01
-6.76765323e-01 8.30478311e-01 -1.08493194e-02 3.76203418e-01
1.38757396e+00 -9.08248723e-02 1.64780781e-01 3.39572430e-01
1.30083346e+00 -3.25609893e-01 -2.11502028e+00 -3.11879933e-01
-1.10973559e-01 -7.90155590e-01 -1.42995313e-01 -8.12726557e-01
-1.24572754e+00 1.02909064e+00 5.52151322e-01 7.15872422e-02
1.40715718e+00 3.30523252e-01 5.93922079e-01 2.50226438e-01
3.21930438e-01 -1.29480302e+00 5.34507096e-01 9.70437825e-01
1.00263786e+00 -1.17122114e+00 -2.74906725e-01 -3.45591679e-02
-7.95953572e-01 1.19535828e+00 6.79295957e-01 -2.87070036e-01
4.66058403e-01 -1.45231500e-01 -7.75182098e-02 -2.09083900e-01
-7.42922902e-01 -4.57440227e-01 7.98152089e-01 7.10928261e-01
1.00610211e-01 6.57399893e-02 -8.82961527e-02 6.65558636e-01
3.13389391e-01 -4.89632133e-03 -2.18304649e-01 8.55639338e-01
-6.29560649e-01 -1.08715963e+00 7.40784556e-02 5.79968572e-01
-9.94803160e-02 1.43235520e-01 -4.64685798e-01 2.84023136e-01
5.02500236e-01 6.12316728e-01 2.27227464e-01 -5.58180869e-01
5.79446673e-01 -9.61683542e-02 5.26877701e-01 -7.14285433e-01
-6.41682327e-01 -9.18076709e-02 -1.13749474e-01 -1.45714414e+00
-5.32343268e-01 -7.20015645e-01 -1.19449115e+00 1.05467528e-01
3.15788120e-01 -2.39644766e-01 2.88853254e-02 1.31579471e+00
6.18265092e-01 8.17864299e-01 6.35880113e-01 -1.03236377e+00
-6.90800130e-01 -7.57519841e-01 -6.99018180e-01 5.65357685e-01
4.70133722e-01 -8.96677077e-01 -2.03803003e-01 -3.86285968e-02] | [8.42962646484375, 0.6623556613922119] |
96ce757f-ba81-4fbc-9113-c5bc0bfbee49 | single-image-haze-removal-using-dark-channel | null | null | https://ieeexplore.ieee.org/abstract/document/5567108/authors#authors | http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.672.3815&rep=rep1&type=pdf | Single Image Haze Removal Using Dark Channel Prior | In this paper, we propose a simple but effective image prior-dark channel prior to remove haze from a single input image. The dark channel prior is a kind of statistics of outdoor haze-free images. It is based on a key observation-most local patches in outdoor haze-free images contain some pixels whose intensity is very low in at least one color channel. Using this prior with the haze imaging model, we can directly estimate the thickness of the haze and recover a high-quality haze-free image. Results on a variety of hazy images demonstrate the power of the proposed prior. Moreover, a high-quality depth map can also be obtained as a byproduct of haze removal. | ['Jian Sun', 'Xiaoou Tang', 'Kaiming He'] | 2010-09-09 | null | null | null | ieee-transactions-on-pattern-analysis-and-10 | ['single-image-haze-removal'] | ['computer-vision'] | [ 2.73165017e-01 -3.66558224e-01 7.28233933e-01 -3.15561146e-01
-3.63991559e-01 -1.20314680e-01 1.72693968e-01 -4.88499403e-01
-1.74020752e-01 8.63182664e-01 -1.22283511e-01 -5.61630800e-02
1.69083089e-01 -1.05642486e+00 -5.54066122e-01 -1.41768658e+00
9.11072269e-02 -1.47619233e-01 6.70836270e-01 -4.49287146e-01
1.20765090e-01 1.43446758e-01 -1.58434832e+00 2.16035679e-01
1.30858958e+00 8.51524055e-01 6.14463627e-01 6.32909119e-01
1.21584572e-01 6.38725460e-01 -7.00573385e-01 8.31649899e-02
5.64872921e-01 -4.04965162e-01 -3.87263522e-02 5.02442777e-01
6.51458561e-01 -8.88216496e-01 -5.84663570e-01 1.50277400e+00
1.35364637e-01 7.94413558e-04 6.80715859e-01 -7.21729398e-01
-5.81378460e-01 -4.25642371e-01 -5.19201040e-01 2.11851791e-01
-3.48061025e-02 1.08538970e-01 3.24960172e-01 -1.07983840e+00
2.34932929e-01 8.56570065e-01 2.57660806e-01 9.63859558e-02
-6.77429199e-01 -6.48114443e-01 -2.55945660e-02 1.46165624e-01
-1.66399539e+00 -3.41547608e-01 6.44212902e-01 -2.88189054e-01
4.93979931e-01 3.73435020e-01 7.68082261e-01 1.63995281e-01
8.72612178e-01 5.31753123e-01 1.54613817e+00 -5.25690854e-01
1.38572544e-01 1.09532867e-02 -7.74436593e-02 6.00298941e-01
7.75277674e-01 2.26353958e-01 -4.34783220e-01 2.43783191e-01
6.31768048e-01 6.77649975e-01 -7.52930880e-01 2.44054347e-01
-7.04222798e-01 3.52553129e-01 3.64546180e-01 -3.83858979e-02
-2.78524548e-01 3.95550430e-02 -8.55794191e-01 4.60031360e-01
7.96874583e-01 5.01489304e-02 1.50549769e-01 5.72515070e-01
-1.12291813e+00 7.97739476e-02 3.72266024e-01 1.03821087e+00
1.47981775e+00 2.79381514e-01 2.85823643e-01 4.68035132e-01
7.63426304e-01 1.47423708e+00 -1.94027185e-01 -1.04421031e+00
-3.75891384e-03 1.88195035e-01 6.63260579e-01 -7.51235485e-01
3.13218415e-01 -1.07363641e-01 -1.04520512e+00 8.70445192e-01
5.05490601e-02 2.28232425e-02 -1.57666624e+00 9.56498563e-01
3.71483147e-01 2.80640960e-01 1.65786475e-01 1.11595285e+00
8.31945837e-01 1.19733250e+00 -7.79905260e-01 -2.73276657e-01
9.83582973e-01 -9.44797218e-01 -1.31852925e+00 -5.41918993e-01
-2.15035349e-01 -9.71325338e-01 4.49782640e-01 8.23324382e-01
-1.11177063e+00 -3.68772984e-01 -1.42169559e+00 3.54573280e-02
-3.57191950e-01 -2.30184644e-01 3.66136670e-01 5.66069841e-01
-1.24525249e+00 2.58803993e-01 -7.79311717e-01 8.33064318e-03
-1.52549878e-01 9.94020551e-02 -2.96585798e-01 -9.39853609e-01
-1.06143999e+00 8.28167558e-01 1.97173700e-01 5.46351612e-01
-1.22959864e+00 -3.73694032e-01 -8.19139123e-01 -9.47687998e-02
3.12937260e-01 -6.09364808e-01 6.02788806e-01 -8.59263182e-01
-1.34501696e+00 5.14974594e-01 -4.90712643e-01 -4.91378568e-02
3.20087490e-03 -6.32228136e-01 -5.99741518e-01 6.21564448e-01
-7.57670626e-02 1.65540710e-01 1.39933968e+00 -1.96522844e+00
-5.50920367e-01 -1.50708407e-01 -2.00850770e-01 1.95266023e-01
2.39248112e-01 -3.74197990e-01 -6.01963818e-01 -2.92524934e-01
3.38607073e-01 -5.75008810e-01 -2.37286001e-01 3.25794891e-02
-4.31841820e-01 6.23071790e-01 9.71923649e-01 -7.52719760e-01
1.00194860e+00 -2.33017635e+00 -3.07320207e-01 1.17699914e-01
5.69236994e-01 2.26951286e-01 -8.24579075e-02 4.61047590e-01
2.16288909e-01 8.28251913e-02 -4.61043388e-01 -2.98294008e-01
-3.94149423e-01 2.99116731e-01 -2.84411669e-01 9.08347070e-01
2.06309304e-01 5.51292479e-01 -7.05345929e-01 -4.06843632e-01
5.65311253e-01 6.68384910e-01 -2.29760274e-01 7.24782825e-01
7.99555406e-02 5.12524426e-01 -2.07168400e-01 7.08589017e-01
1.64755213e+00 7.18080848e-02 -3.71776044e-01 1.30297109e-01
-5.26382148e-01 -1.11825280e-01 -1.08960474e+00 1.06800878e+00
-1.48781732e-01 8.41211319e-01 6.67209744e-01 -2.76787639e-01
8.44625235e-01 7.03625008e-02 2.94536967e-02 -6.38707817e-01
-4.96055372e-03 2.36303762e-01 -3.67675871e-01 -7.01445818e-01
6.74512923e-01 -5.53247571e-01 5.78505874e-01 -2.40772255e-02
-3.02777261e-01 -6.49031937e-01 -8.79990309e-02 1.59420252e-01
8.36552620e-01 -2.68390000e-01 -2.05834992e-02 -4.47254419e-01
3.74750048e-01 -2.01501712e-01 7.01988280e-01 9.29674327e-01
-3.22412476e-02 1.26362288e+00 -3.44466865e-01 -2.20272511e-01
-9.67729211e-01 -1.37737620e+00 -1.62838325e-01 2.48949319e-01
7.63784170e-01 -1.24521375e-01 -5.85537791e-01 1.83175534e-01
-3.34462613e-01 3.53696197e-01 -6.00582540e-01 -4.64149378e-02
-3.95135850e-01 -1.01253307e+00 -2.05950171e-01 -2.90615529e-01
1.15691304e+00 -4.43738133e-01 -2.86438763e-01 -1.90562546e-01
-3.29067498e-01 -1.25278389e+00 -1.41579524e-01 5.81035502e-02
-8.79727125e-01 -9.26086426e-01 -9.30722892e-01 -6.89751744e-01
8.79327655e-01 1.15206909e+00 1.17939723e+00 6.63865268e-01
-1.35244355e-01 3.06535661e-01 -5.24850428e-01 -6.00213468e-01
-1.51839629e-01 -8.08435678e-01 -2.39504963e-01 5.47523379e-01
2.58221440e-02 -6.07451320e-01 -1.10670877e+00 3.25315505e-01
-1.08353639e+00 1.84344351e-01 7.16935635e-01 5.72377503e-01
7.21391618e-01 8.47405136e-01 -3.70652735e-01 -5.67606091e-01
-2.32950062e-01 -6.28694654e-01 -1.07027972e+00 -7.47447908e-02
-6.11245513e-01 -5.34066617e-01 1.84441715e-01 9.91236120e-02
-1.45192087e+00 -2.57572502e-01 2.74826616e-01 -4.59886402e-01
-2.97261506e-01 3.67434353e-01 -4.25246298e-01 -2.86873758e-01
1.17596760e-01 6.77745819e-01 7.59124104e-03 -5.11862099e-01
6.51027262e-02 6.65563166e-01 7.79991806e-01 -1.49125010e-01
1.46304798e+00 1.19003820e+00 1.68854564e-01 -1.58563232e+00
-7.74792492e-01 -8.26400578e-01 -4.32423949e-01 -3.06722671e-01
1.15416944e+00 -1.42909968e+00 -4.12088156e-01 1.02108645e+00
-9.85699892e-01 -7.44994998e-01 2.33930260e-01 6.27855241e-01
2.30525345e-01 6.09667301e-01 -6.23019159e-01 -1.19958460e+00
-9.50849950e-02 -8.27876031e-01 9.99804437e-01 4.34825510e-01
1.06574810e+00 -1.15648544e+00 -1.47434190e-01 3.79417330e-01
6.57188833e-01 2.62390882e-01 5.93713343e-01 5.02674043e-01
-1.66614997e+00 5.40388487e-02 -2.84824759e-01 7.40671813e-01
3.15388381e-01 1.68832645e-01 -1.14336324e+00 -3.97980571e-01
4.70002264e-01 2.94330776e-01 1.34138930e+00 6.04245245e-01
4.46896702e-01 -8.45947191e-02 -6.26306236e-02 1.12021053e+00
2.10811448e+00 1.68208238e-02 1.30933881e+00 2.98240393e-01
6.39765978e-01 3.65311027e-01 8.56428504e-01 2.07970396e-01
1.16920523e-01 3.66334558e-01 9.45734143e-01 -7.02454567e-01
-4.66466695e-01 2.50097752e-01 3.86017948e-01 8.92555475e-01
-3.07416081e-01 -7.79999077e-01 -6.14341974e-01 7.50805676e-01
-1.39511776e+00 -6.26804292e-01 -8.00073862e-01 2.19406867e+00
5.51168144e-01 -1.05984330e-01 -7.35126555e-01 -1.47713259e-01
5.43470860e-01 2.41402447e-01 -3.39524746e-02 1.86167300e-01
-5.27832747e-01 2.58340001e-01 7.60775149e-01 1.09872544e+00
-8.74047101e-01 9.98062730e-01 7.11119366e+00 4.66864645e-01
-1.07522023e+00 1.27211750e-01 1.14380065e-02 2.15079501e-01
-5.41188657e-01 1.23549156e-01 -7.18392611e-01 7.25475490e-01
7.15375602e-01 1.11845575e-01 3.79077315e-01 2.08667085e-01
4.77652371e-01 -1.04672313e+00 -4.16972339e-01 1.04217756e+00
9.25338268e-02 -1.05132830e+00 1.62474886e-01 4.01350826e-01
1.10974717e+00 1.34325475e-01 2.10061878e-01 -3.41390342e-01
2.86725700e-01 -9.50253427e-01 6.09015107e-01 8.26758325e-01
7.38014817e-01 -3.92451197e-01 1.15081894e+00 3.92596900e-01
-9.92633224e-01 3.10070068e-01 -8.08121681e-01 -4.34132755e-01
2.48325169e-01 1.38457024e+00 -2.84820706e-01 6.66876495e-01
1.20810044e+00 7.64086127e-01 -5.35471320e-01 1.28666329e+00
-8.24124992e-01 7.89781749e-01 -3.57396066e-01 6.59518361e-01
1.56758830e-01 -7.28370190e-01 5.50668597e-01 1.14825976e+00
6.31851315e-01 7.95080066e-01 -2.33868603e-02 9.61425900e-01
3.07827532e-01 -4.94824827e-01 -7.46686757e-01 5.10879040e-01
1.38158143e-01 9.73803341e-01 -4.45647568e-01 -2.77808756e-01
-5.32484412e-01 1.20473433e+00 -6.01485252e-01 9.96054530e-01
-3.84810567e-01 -4.19525713e-01 8.64571512e-01 1.86372980e-01
5.80016851e-01 -5.26613712e-01 -1.08055644e-01 -1.32301617e+00
-1.80178717e-01 -6.79665506e-01 -2.82226086e-01 -1.28194737e+00
-1.10589075e+00 2.38215521e-01 2.13069804e-02 -1.22476220e+00
5.33462405e-01 -7.57515192e-01 -9.78954792e-01 1.21129143e+00
-2.30048490e+00 -9.22290087e-01 -1.14472437e+00 8.90651047e-01
2.50770926e-01 3.44287306e-01 4.75546807e-01 2.91588455e-01
-5.00451863e-01 -1.99656844e-01 6.72135711e-01 -6.64426014e-02
7.45523810e-01 -1.37940371e+00 2.24060770e-02 1.62363434e+00
-4.68391061e-01 7.59395063e-01 1.00106609e+00 -7.30277479e-01
-1.45665324e+00 -1.01885378e+00 1.93765476e-01 -4.27266508e-01
-5.89360856e-03 -2.65126497e-01 -1.21505618e+00 6.42131090e-01
6.91617966e-01 2.71746278e-01 4.76234496e-01 -6.80069685e-01
-1.23071764e-02 -2.92492270e-01 -1.01336312e+00 2.38192424e-01
3.15352887e-01 -3.14489365e-01 -6.58062041e-01 1.71079278e-01
5.60272872e-01 -4.23410118e-01 -5.84471643e-01 4.67726737e-01
2.47325972e-01 -1.51912832e+00 8.54371548e-01 2.85451263e-01
2.88643301e-01 -9.64573026e-01 -5.40428162e-01 -1.21286011e+00
-4.49125528e-01 -7.39612520e-01 5.44670448e-02 9.07280326e-01
9.23822001e-02 -7.37377703e-01 6.21282279e-01 1.17278934e-01
-5.10877848e-01 -8.99701044e-02 -5.32545447e-01 -8.88009012e-01
2.48354133e-02 9.41510797e-02 4.85881627e-01 6.73311532e-01
-8.20546865e-01 1.22531191e-01 -8.00863683e-01 1.18326283e+00
1.36271262e+00 1.92184642e-01 9.48793173e-01 -1.14706945e+00
9.12995189e-02 3.59610617e-01 -2.22246960e-01 -1.20694566e+00
-2.68059045e-01 -2.16519728e-01 6.47471964e-01 -1.91130924e+00
4.36359942e-01 -1.49917439e-01 -2.88563579e-01 1.54455647e-01
-4.30660039e-01 7.72598743e-01 3.01108621e-02 4.79950041e-01
-4.79023635e-01 6.58785999e-01 1.44390309e+00 -3.52650195e-01
-2.51538455e-02 -1.74405500e-01 -2.94326574e-01 7.04019427e-01
5.75885653e-01 -6.03509188e-01 -3.05498511e-01 -6.31779850e-01
3.07185203e-01 -2.52480090e-01 4.19386387e-01 -9.70416129e-01
7.56535679e-02 -5.34641922e-01 5.08764684e-01 -9.17528331e-01
7.10968137e-01 -8.99729013e-01 1.92160606e-01 2.49907538e-01
7.96451688e-01 -6.26926661e-01 1.17137365e-01 7.82703102e-01
-7.23810494e-01 -2.01996475e-01 1.08015692e+00 -3.80438626e-01
-9.97016549e-01 2.58503199e-01 -6.04994118e-01 -2.24032834e-01
6.40421510e-01 -5.09968996e-01 -7.46124446e-01 -7.09743500e-01
-4.20643449e-01 5.75694852e-02 1.06889892e+00 -3.74798119e-01
1.26030409e+00 -8.66998017e-01 -8.53694379e-01 4.47152287e-01
7.39845708e-02 3.18410993e-01 3.44389677e-01 9.28579628e-01
-1.11310232e+00 -2.57502720e-02 -3.14260088e-02 -6.37178838e-01
-1.18503177e+00 4.91880804e-01 4.92064595e-01 2.66972244e-01
-1.02507949e+00 7.30224013e-01 8.49951863e-01 1.03092425e-01
-2.95983195e-01 -1.68958575e-01 1.86623976e-01 -7.51215219e-01
9.53478217e-01 3.22410077e-01 -5.52517511e-02 -5.41447937e-01
-1.95668533e-01 8.91200066e-01 3.30907166e-01 -1.21897623e-01
1.33647537e+00 -8.27263355e-01 -6.94075108e-01 3.53979886e-01
8.60062897e-01 4.78723317e-01 -1.43933916e+00 -6.85127899e-02
-7.45382011e-01 -1.21440887e+00 5.65647244e-01 -5.23040712e-01
-1.31357324e+00 1.16745996e+00 7.16560185e-01 1.25795394e-01
1.42260015e+00 -2.63267666e-01 8.58451784e-01 3.76679659e-01
4.27874655e-01 -6.98238373e-01 2.27656960e-01 4.67813134e-01
5.78278542e-01 -1.31746268e+00 5.22972167e-01 -8.06234062e-01
5.61728589e-02 1.02050972e+00 4.06887740e-01 -6.51344508e-02
1.00711823e+00 2.25946277e-01 3.93623203e-01 -4.78176653e-01
-3.87128502e-01 -5.03843784e-01 1.92085505e-01 7.50257015e-01
-1.17445499e-01 -2.23544240e-01 1.93552002e-01 -8.73124748e-02
2.53927797e-01 -2.16986671e-01 1.24576724e+00 8.83435071e-01
-1.27120006e+00 -5.56557536e-01 -8.99562299e-01 -4.92047630e-02
-4.31652337e-01 -3.97224486e-01 -2.56892871e-02 6.77599251e-01
2.60532677e-01 1.61990023e+00 -1.02032132e-01 -2.15033785e-01
-1.61921814e-01 -3.95435810e-01 4.56287265e-01 -7.63028085e-01
3.52322072e-01 5.00821471e-01 -3.50949287e-01 -5.55747151e-01
-5.87259173e-01 7.74490274e-03 -9.91776764e-01 -3.53555113e-01
-5.63530922e-01 2.35121846e-01 3.97049040e-01 7.70650566e-01
-6.04820810e-03 2.39149109e-01 6.93139732e-01 -9.69143689e-01
3.42458487e-01 -8.71052861e-01 -1.57919109e+00 1.80948153e-01
1.14937007e+00 -6.66490674e-01 -1.13405597e+00 2.27173418e-01] | [10.864924430847168, -3.178379774093628] |
72f4b9f0-1e1d-4d2b-ab70-5599e967b29a | samu-xlsr-semantically-aligned-multimodal | 2205.08180 | null | https://arxiv.org/abs/2205.08180v1 | https://arxiv.org/pdf/2205.08180v1.pdf | SAMU-XLSR: Semantically-Aligned Multimodal Utterance-level Cross-Lingual Speech Representation | We propose the SAMU-XLSR: Semantically-Aligned Multimodal Utterance-level Cross-Lingual Speech Representation learning framework. Unlike previous works on speech representation learning, which learns multilingual contextual speech embedding at the resolution of an acoustic frame (10-20ms), this work focuses on learning multimodal (speech-text) multilingual speech embedding at the resolution of a sentence (5-10s) such that the embedding vector space is semantically aligned across different languages. We combine state-of-the-art multilingual acoustic frame-level speech representation learning model XLS-R with the Language Agnostic BERT Sentence Embedding (LaBSE) model to create an utterance-level multimodal multilingual speech encoder SAMU-XLSR. Although we train SAMU-XLSR with only multilingual transcribed speech data, cross-lingual speech-text and speech-speech associations emerge in its learned representation space. To substantiate our claims, we use SAMU-XLSR speech encoder in combination with a pre-trained LaBSE text sentence encoder for cross-lingual speech-to-text translation retrieval, and SAMU-XLSR alone for cross-lingual speech-to-speech translation retrieval. We highlight these applications by performing several cross-lingual text and speech translation retrieval tasks across several datasets. | ['James Glass', 'Antoine Laurent', 'Sameer Khurana'] | 2022-05-17 | null | null | null | null | ['speech-to-text-translation', 'speech-to-speech-translation'] | ['natural-language-processing', 'speech'] | [ 3.24702919e-01 1.48950994e-01 -3.15041274e-01 -4.60276991e-01
-2.01901817e+00 -5.62534451e-01 8.98683488e-01 6.91009983e-02
-4.84039307e-01 3.92643541e-01 6.29213274e-01 -7.73869634e-01
2.33664826e-01 -2.85664052e-01 -9.79194820e-01 -3.05429369e-01
2.91979045e-01 6.12192154e-01 -1.77678823e-01 -2.94955283e-01
-3.73623997e-01 5.81867807e-03 -1.15606475e+00 7.51554489e-01
4.32433218e-01 5.59637785e-01 5.76416731e-01 9.80971754e-01
-4.01574939e-01 6.30014539e-01 -6.46946967e-01 -2.67858863e-01
-1.93314239e-01 -5.03974199e-01 -8.15192461e-01 -1.68368116e-01
6.40127838e-01 -5.74434884e-02 -5.60062766e-01 7.39108324e-01
7.79166222e-01 2.09077165e-01 6.37358189e-01 -7.01811254e-01
-1.07288563e+00 9.07452583e-01 -1.27574906e-01 1.65733978e-01
7.79374719e-01 -2.46571288e-01 1.18693042e+00 -1.25337982e+00
7.95978963e-01 1.84513152e+00 2.01015547e-01 5.45273662e-01
-1.23309398e+00 -2.97486395e-01 2.32946035e-02 3.14135849e-01
-1.34100175e+00 -1.03778815e+00 4.91483301e-01 -1.67479739e-01
1.72568130e+00 3.84898722e-01 -8.62305705e-03 1.77871299e+00
3.39607805e-01 1.08360553e+00 9.58634317e-01 -8.46249521e-01
9.98402610e-02 3.16460371e-01 -6.47470504e-02 3.95952135e-01
-5.40838301e-01 1.11953594e-01 -8.95596564e-01 6.54896572e-02
3.62451524e-01 -3.42018098e-01 7.61377290e-02 1.82706207e-01
-1.47723603e+00 9.25414205e-01 -4.43948358e-02 6.63854063e-01
-1.73088610e-01 2.50170618e-01 1.14319754e+00 8.60243857e-01
5.40930510e-01 -4.14639376e-02 -5.80218315e-01 -9.82296169e-02
-8.98608029e-01 -4.19964135e-01 4.55677867e-01 9.73964274e-01
6.77115262e-01 6.31196320e-01 -1.09840125e-01 1.25214958e+00
5.93654394e-01 1.14710093e+00 1.10211456e+00 -6.25542998e-01
7.28592753e-01 -1.43372610e-01 -6.13009155e-01 -2.92487681e-01
1.26573682e-01 3.78599088e-03 -4.16578859e-01 -3.22968632e-01
-3.73619974e-01 -1.15151182e-01 -7.00885236e-01 1.70589590e+00
-3.45780514e-02 8.12202320e-02 1.04447365e+00 5.97520947e-01
1.14262557e+00 1.28782570e+00 2.65211612e-01 -4.12775815e-01
1.60262704e+00 -1.09795904e+00 -1.00911748e+00 -3.72702479e-01
9.28480625e-01 -1.11202741e+00 1.35433078e+00 -4.03586060e-01
-1.20638192e+00 -7.57845163e-01 -9.13767040e-01 -3.99518937e-01
-7.18435109e-01 4.38304096e-01 1.11290477e-02 3.56789798e-01
-1.26733041e+00 -3.57119203e-01 -8.02157581e-01 -6.10235453e-01
-4.72351849e-01 4.05758135e-02 -7.35650659e-01 -2.13803980e-03
-1.54997218e+00 1.18674505e+00 2.90572912e-01 -3.05201590e-01
-1.01662278e+00 -3.99972916e-01 -1.49678719e+00 1.05520589e-02
1.47694722e-01 -4.83642071e-01 1.31108582e+00 -8.31534505e-01
-1.63538611e+00 8.27550709e-01 -6.70426369e-01 -5.16208529e-01
-2.44291022e-01 6.46699071e-02 -8.43279839e-01 2.63919294e-01
2.66727477e-01 5.97035766e-01 9.55125332e-01 -1.21729898e+00
-3.31339598e-01 -3.66921842e-01 -2.98555851e-01 4.64742512e-01
-2.02054635e-01 5.67276597e-01 -1.14070438e-01 -7.66807675e-01
1.28754703e-02 -8.02032113e-01 2.13755488e-01 -8.04803729e-01
4.36764359e-02 -4.93693471e-01 1.00792086e+00 -9.11397934e-01
1.15120661e+00 -2.24918246e+00 4.24102813e-01 -4.13114965e-01
-7.25122869e-01 2.45902985e-01 -7.39031672e-01 1.03047955e+00
-3.26499045e-01 -2.93298401e-02 -4.43798937e-02 -1.08210623e+00
3.05396676e-01 6.18894279e-01 -8.39095891e-01 2.75516212e-01
3.11248302e-01 1.15733898e+00 -8.18782270e-01 -4.74876255e-01
5.53192914e-01 9.68540609e-01 -2.17477158e-02 4.47611153e-01
5.71661890e-02 3.37700129e-01 -1.92071334e-01 7.89303899e-01
-1.04417279e-02 4.01397079e-01 2.16086850e-01 -2.21296549e-01
-7.50948340e-02 9.60324168e-01 -5.89353621e-01 2.21411395e+00
-1.05294526e+00 8.15912783e-01 -3.64587605e-02 -1.00964499e+00
8.97266507e-01 1.08401012e+00 3.44569355e-01 -1.15812528e+00
-3.98431271e-02 5.51037610e-01 -3.84222925e-01 -5.19882381e-01
6.04225576e-01 -3.95264179e-01 -5.22869289e-01 5.36591709e-01
8.03146482e-01 -1.71974406e-01 -1.08975127e-01 1.95298195e-01
7.79743493e-01 -4.34228182e-02 2.81352013e-01 9.73528624e-02
7.86424041e-01 -5.99059284e-01 -9.36216936e-02 4.57459778e-01
-3.17994729e-02 4.23987925e-01 -3.35844159e-02 -1.15955047e-01
-1.08853948e+00 -1.28858972e+00 -2.07012683e-01 1.79140973e+00
-4.21828181e-01 -6.87093973e-01 -6.43928230e-01 -6.18353128e-01
-3.27174246e-01 9.72190797e-01 -2.74951220e-01 -9.03685167e-02
-7.36114621e-01 -1.42918229e-01 1.02027035e+00 4.91251707e-01
-2.62607664e-01 -1.36297357e+00 5.82532547e-02 3.20527285e-01
-3.58473569e-01 -1.52833438e+00 -8.45399916e-01 3.58387947e-01
-3.58086854e-01 -5.72263658e-01 -6.54493690e-01 -1.13221347e+00
3.15696031e-01 -4.63216193e-03 9.51900959e-01 -5.85331321e-01
-1.58686534e-01 1.08174825e+00 -6.20957851e-01 -6.70799240e-02
-1.04687750e+00 1.28809988e-01 3.96109730e-01 -1.61937267e-01
4.52884704e-01 -2.81141520e-01 1.64662123e-01 -3.19626071e-02
-9.26889181e-01 -3.30699503e-01 5.80889523e-01 9.21498060e-01
7.30543017e-01 -6.13548160e-01 1.08567715e+00 -2.67375141e-01
7.28382170e-01 -4.99517590e-01 -2.20098093e-01 4.84978855e-01
-1.30762041e-01 3.11599020e-02 4.70409483e-01 -5.80518961e-01
-8.04442465e-01 -1.35763407e-01 -5.01009166e-01 -7.05081820e-01
-2.24443749e-01 8.78563166e-01 -8.35007802e-02 4.62856859e-01
3.34233165e-01 7.86277115e-01 -4.77636456e-02 -3.61643314e-01
9.08351541e-01 1.14663255e+00 5.21350205e-01 -6.42560720e-01
3.47044587e-01 -1.92390785e-01 -5.08370578e-01 -1.23827338e+00
-3.78501385e-01 -4.14083242e-01 -7.65080333e-01 -2.42074467e-02
1.21459389e+00 -1.20975459e+00 -3.59450072e-01 -3.13507974e-01
-1.57324433e+00 -1.54525816e-01 -3.75669658e-01 7.20444500e-01
-9.30010200e-01 4.31976765e-01 -8.66860867e-01 -1.11649096e+00
-5.25823534e-01 -1.50437891e+00 1.83850503e+00 -5.41106582e-01
-3.78114104e-01 -1.21909475e+00 3.44608128e-01 5.77412426e-01
3.80389065e-01 -2.66618669e-01 8.70494425e-01 -9.35285389e-01
-2.70828098e-01 -1.50707513e-02 1.25580817e-01 6.06066525e-01
1.33195132e-01 -3.61745000e-01 -1.05958116e+00 -6.38672173e-01
-1.21405907e-01 -6.05811596e-01 6.14122450e-01 2.36663640e-01
4.02564406e-01 -3.64936620e-01 7.51235262e-02 2.30272755e-01
1.04227233e+00 3.79657984e-01 3.30262333e-01 2.06141219e-01
5.53336084e-01 6.45366669e-01 3.91028523e-01 -1.56468928e-01
6.67089343e-01 8.80967140e-01 4.81578261e-02 -4.07143161e-02
-5.46185553e-01 -3.66367400e-01 1.29025102e+00 2.09128428e+00
6.79195523e-01 -2.18169048e-01 -6.90996766e-01 7.54166186e-01
-1.83392942e+00 -7.90113449e-01 4.55077082e-01 2.11638141e+00
8.77383113e-01 -3.41252446e-01 -2.33510748e-01 -2.43142128e-01
2.85353512e-01 4.64043349e-01 -2.09125876e-01 -9.03822422e-01
-4.33752686e-01 2.35067874e-01 1.24859087e-01 1.06775188e+00
-8.98580909e-01 1.48726559e+00 6.21130419e+00 1.06566417e+00
-1.23011339e+00 7.00523198e-01 1.23730011e-01 5.78955077e-02
-6.41806185e-01 -1.16883233e-01 -7.03765035e-01 2.16962561e-01
1.83693254e+00 -1.30819082e-01 5.56465566e-01 6.25221908e-01
-1.12804037e-03 5.13425767e-01 -1.43413663e+00 1.21519804e+00
5.60776234e-01 -1.23044407e+00 4.17782843e-01 -2.26939723e-01
4.44633037e-01 4.73535269e-01 1.90019444e-01 8.31882298e-01
-2.61412635e-02 -1.23442197e+00 7.01641977e-01 2.11264119e-01
1.44004142e+00 -6.53414190e-01 6.52767420e-01 1.62485436e-01
-1.62318432e+00 3.37013662e-01 -3.37861776e-01 7.03557193e-01
5.44882238e-01 -2.95587033e-01 -1.04424417e+00 9.48646545e-01
3.66337508e-01 8.20229053e-01 -3.20899457e-01 -2.37936556e-01
-4.76627760e-02 6.26986265e-01 -1.02466896e-01 2.05599546e-01
4.44471359e-01 -1.81853786e-01 7.54818320e-01 1.84832990e+00
4.94882733e-01 -4.58597928e-01 3.29917431e-01 5.14613986e-01
-1.07623994e-01 4.96179998e-01 -9.84166682e-01 -5.65776646e-01
6.75218701e-01 6.92405701e-01 5.56103624e-02 -5.82544684e-01
-6.99065864e-01 1.15176225e+00 1.78913414e-01 5.84878683e-01
-3.43263507e-01 -1.04612350e-01 8.37314844e-01 -1.90101892e-01
1.31769836e-01 -5.18421412e-01 2.88905263e-01 -1.19404483e+00
-1.18846014e-01 -1.14886630e+00 2.01671332e-01 -9.12772715e-01
-1.40075040e+00 1.11837375e+00 -1.44315660e-01 -1.03079784e+00
-1.01941061e+00 -4.95824784e-01 -2.55602092e-01 1.02825689e+00
-1.74076521e+00 -1.72389853e+00 7.54499495e-01 7.69623816e-01
1.35396039e+00 -9.84494507e-01 1.58579457e+00 4.81331468e-01
-2.06295177e-01 8.03131163e-01 1.95939735e-01 1.53132945e-01
1.02309680e+00 -1.11117768e+00 4.55358356e-01 3.70510012e-01
6.12450719e-01 8.31245184e-01 4.52761382e-01 -3.79935652e-01
-1.88735783e+00 -1.09719455e+00 1.37566638e+00 -8.01308632e-01
1.12688863e+00 -6.19493484e-01 -1.01938605e+00 1.16552067e+00
9.13467228e-01 -2.61362828e-02 1.06765497e+00 1.00842230e-02
-4.81075823e-01 -3.84042151e-02 -5.95230162e-01 7.37137496e-01
4.51975375e-01 -1.81907845e+00 -1.05339813e+00 3.60830188e-01
1.46291506e+00 -2.37070084e-01 -1.17125416e+00 3.93581748e-01
3.18636924e-01 -1.85737893e-01 1.21531129e+00 -7.06187367e-01
2.14514378e-02 -2.61749446e-01 -1.07335389e+00 -1.38389552e+00
3.44492942e-01 -5.13590217e-01 1.38703296e-02 1.18310642e+00
4.46893185e-01 -6.21740580e-01 -9.79364514e-02 -4.59525347e-01
-6.71799660e-01 -4.69701380e-01 -1.75911963e+00 -8.33117008e-01
3.45084757e-01 -7.56714463e-01 2.56786317e-01 1.02480698e+00
3.27944934e-01 9.02534187e-01 -4.04160470e-01 4.12677675e-01
3.53333712e-01 -2.46311694e-01 6.13232017e-01 -4.70683932e-01
-2.54273146e-01 -2.71936178e-01 -4.26931769e-01 -1.14398384e+00
1.07285798e+00 -1.39189219e+00 3.51694897e-02 -1.38983321e+00
-1.09919235e-01 1.78914651e-01 -3.37822706e-01 4.96471137e-01
2.28059754e-01 -1.03739090e-01 9.27529484e-02 -4.44261245e-02
-5.78900039e-01 1.06364667e+00 7.92442918e-01 -3.59727442e-01
1.88545629e-01 -5.77410221e-01 -5.60813546e-02 2.09524155e-01
2.87470460e-01 -3.70405287e-01 -5.14712930e-01 -5.74416637e-01
-1.81987166e-01 7.27857709e-01 -4.82746623e-02 -5.17209888e-01
2.19201237e-01 1.63771436e-01 -1.16798103e-01 -6.85316861e-01
8.85678828e-01 -7.62068808e-01 -1.31077513e-01 -1.12441853e-02
-6.69940174e-01 1.20839015e-01 4.02196944e-01 3.07681739e-01
-7.89694846e-01 -1.72717169e-01 3.73082757e-01 3.15203778e-02
-5.05077600e-01 6.89124614e-02 -8.09522867e-01 -2.39924416e-01
5.72408199e-01 1.88728943e-01 -4.37557340e-01 -4.71947372e-01
-1.08950472e+00 -6.46277815e-02 -8.41554925e-02 1.06274855e+00
7.54859149e-01 -1.64189565e+00 -9.24359143e-01 5.15923738e-01
5.34317791e-01 -6.06637180e-01 1.62566423e-01 5.83679616e-01
3.53127837e-01 9.84849215e-01 2.49221921e-01 -8.27593386e-01
-1.35885096e+00 4.19801831e-01 2.10810587e-01 8.23992416e-02
-3.30842614e-01 5.89083374e-01 1.64358795e-01 -1.05590844e+00
2.66344577e-01 -1.99237153e-01 8.43351781e-02 6.75695017e-02
5.15951455e-01 -5.72670549e-02 1.25859991e-01 -1.45173228e+00
-4.47775155e-01 4.82203931e-01 -4.77089137e-02 -6.56700730e-01
1.09953582e+00 -6.63385153e-01 -1.56968981e-01 1.12551725e+00
1.69021654e+00 1.43577695e-01 -5.67667484e-01 -5.95107257e-01
6.51082620e-02 3.32860574e-02 4.17556018e-02 -6.02417886e-01
-3.29706401e-01 1.04509127e+00 6.68666184e-01 -4.47548069e-02
5.35085201e-01 5.52025676e-01 9.86831605e-01 5.93500257e-01
3.02382380e-01 -1.19815469e+00 2.68872082e-01 1.03072393e+00
1.22540867e+00 -1.57288373e+00 -6.89567864e-01 1.57932788e-01
-8.71703863e-01 1.18270957e+00 2.09137902e-01 1.98556378e-01
4.26531911e-01 2.88037479e-01 4.97104853e-01 -4.10169140e-02
-1.01652718e+00 -2.06861883e-01 6.62486434e-01 5.14632106e-01
9.38528240e-01 3.69916648e-01 5.30085601e-02 5.56329489e-01
-4.77830827e-01 -8.15863192e-01 7.75450170e-02 9.12629306e-01
-3.67826492e-01 -1.55468762e+00 -3.22708815e-01 -3.38714719e-01
-2.55094647e-01 -5.78669906e-01 -2.00854197e-01 3.29133123e-01
-3.94395739e-01 1.08543742e+00 1.34396017e-01 -2.65325487e-01
3.64961028e-01 7.65385091e-01 3.09758782e-01 -9.36810791e-01
-3.41029495e-01 6.73109055e-01 1.93300560e-01 -5.92550039e-01
-3.62888426e-01 -6.81261718e-01 -1.34971535e+00 1.49853393e-01
-1.07144210e-02 2.33013675e-01 1.21704507e+00 1.05164361e+00
3.01341265e-01 7.61480510e-01 5.86211562e-01 -8.13061655e-01
-4.20250744e-01 -1.25027823e+00 -2.94971466e-01 1.14886917e-01
6.91466391e-01 -1.35333598e-01 -3.61515433e-01 2.03846574e-01] | [14.420610427856445, 7.078566551208496] |
e02e1a77-a336-4db1-9ef8-0e1d168a724b | collective-relevance-labeling-for-passage | 2205.03273 | null | https://arxiv.org/abs/2205.03273v2 | https://arxiv.org/pdf/2205.03273v2.pdf | Collective Relevance Labeling for Passage Retrieval | Deep learning for Information Retrieval (IR) requires a large amount of high-quality query-document relevance labels, but such labels are inherently sparse. Label smoothing redistributes some observed probability mass over unobserved instances, often uniformly, uninformed of the true distribution. In contrast, we propose knowledge distillation for informed labeling, without incurring high computation overheads at evaluation time. Our contribution is designing a simple but efficient teacher model which utilizes collective knowledge, to outperform state-of-the-arts distilled from a more complex teacher model. Specifically, we train up to x8 faster than the state-of-the-art teacher, while distilling the rankings better. Our code is publicly available at https://github.com/jihyukkim-nlp/CollectiveKD | ['Minsoo Kim', 'Seung-won Hwang', 'Jihyuk Kim'] | 2022-05-06 | null | https://aclanthology.org/2022.naacl-main.305 | https://aclanthology.org/2022.naacl-main.305.pdf | naacl-2022-7 | ['passage-retrieval'] | ['natural-language-processing'] | [-1.48372129e-01 5.02703190e-01 -7.72169054e-01 -5.46071887e-01
-1.79575062e+00 -6.57390058e-01 6.46299481e-01 1.80428639e-01
-5.32745063e-01 9.04396296e-01 3.22774202e-01 -3.98190945e-01
-3.91407281e-01 -6.26663387e-01 -8.48554075e-01 -5.08196592e-01
-4.57067154e-02 1.17209244e+00 3.17018293e-02 1.87976956e-01
-2.79123001e-02 3.29689942e-02 -1.37900317e+00 2.62920201e-01
6.44716799e-01 9.02462423e-01 -2.37700809e-02 6.04195058e-01
-7.65503943e-02 1.09593666e+00 -5.70523083e-01 -5.08371413e-01
5.62980361e-02 2.87357327e-02 -1.28377140e+00 -3.71295571e-01
7.75496960e-01 -7.00843692e-01 -7.45735645e-01 9.39822912e-01
5.12512267e-01 3.99041116e-01 7.32385457e-01 -1.02290356e+00
-1.23668373e+00 1.09167969e+00 -3.66873085e-01 4.58667614e-02
-2.76279338e-02 -5.50033189e-02 1.50347507e+00 -8.55693281e-01
5.49798846e-01 1.32771122e+00 8.21826160e-02 5.09163439e-01
-1.33052409e+00 -7.99675226e-01 4.03765202e-01 2.24301144e-01
-1.47418582e+00 -2.45938346e-01 5.62585592e-01 -2.32894585e-01
6.88656747e-01 1.87852889e-01 1.30513594e-01 9.97332454e-01
-5.46831727e-01 1.38733923e+00 8.87352586e-01 -4.44766700e-01
1.53956741e-01 2.17680991e-01 6.32549286e-01 6.14572883e-01
2.14704335e-01 3.81177813e-02 -3.52139115e-01 -5.55337608e-01
3.42719227e-01 3.01790744e-01 -3.06016564e-01 -2.50404865e-01
-9.49010611e-01 9.09074187e-01 6.02769732e-01 8.58626813e-02
-3.24783623e-01 7.11134136e-01 5.82263544e-02 3.41622025e-01
9.24988151e-01 3.72138560e-01 -9.83303487e-01 9.42931138e-03
-9.63917792e-01 3.20393622e-01 7.68317819e-01 1.04052103e+00
1.14589369e+00 -4.76492882e-01 -5.58080852e-01 6.72916889e-01
5.30351639e-01 6.84631705e-01 1.70307934e-01 -1.40205216e+00
1.45452127e-01 3.28087002e-01 3.38516295e-01 -5.07837594e-01
-1.34441525e-01 -6.44443691e-01 -4.35936481e-01 -2.09131733e-01
3.62591624e-01 -2.69492447e-01 -1.25677454e+00 1.90920579e+00
4.92183715e-01 7.33366311e-02 -2.73543522e-02 1.01250231e+00
7.57683635e-01 7.26015687e-01 3.65754306e-01 2.64109015e-01
1.46959555e+00 -1.16529572e+00 -5.54592192e-01 -6.94560707e-02
7.04255164e-01 -6.73624039e-01 1.16661859e+00 4.34181720e-01
-1.21872437e+00 -1.03682011e-01 -5.11423528e-01 -7.77874112e-01
-2.64482230e-01 2.47124687e-01 1.21181667e+00 1.88669503e-01
-1.28986084e+00 4.02351439e-01 -8.32699418e-01 9.84492823e-02
6.46971166e-01 4.35586303e-01 -1.50561426e-02 -2.63255596e-01
-1.46477926e+00 6.56576574e-01 2.05199838e-01 -2.47189015e-01
-1.26914036e+00 -1.01511478e+00 -4.75414574e-01 4.27106768e-01
6.81427300e-01 -9.34561014e-01 2.00614738e+00 -6.71819210e-01
-1.41357589e+00 7.11372733e-01 -1.63387209e-01 -1.37954041e-01
4.26624715e-01 -6.76362216e-01 8.66808295e-02 1.13349818e-01
-7.38822892e-02 8.84675562e-01 3.76195639e-01 -1.44253981e+00
-4.78223056e-01 -4.35177445e-01 3.23632717e-01 3.28359812e-01
-4.09974575e-01 -3.39764804e-02 -8.04708183e-01 -3.68801326e-01
-1.26605764e-01 -8.45835924e-01 -1.18603818e-01 -1.28921308e-02
-6.75609052e-01 -9.43062544e-01 2.51286089e-01 -3.69999290e-01
1.09019268e+00 -1.90189302e+00 -2.48341650e-01 1.62497625e-01
5.79286277e-01 2.15138927e-01 -3.35294038e-01 5.35414875e-01
2.41257563e-01 1.79217890e-01 2.53690839e-01 -6.76866353e-01
4.85523015e-01 -5.31188771e-02 -5.76860130e-01 2.98360199e-01
4.62051947e-03 1.02068162e+00 -1.33243978e+00 -4.29381847e-01
-2.44556665e-01 6.66278720e-01 -5.30498743e-01 1.87519610e-01
-7.73403227e-01 1.27143338e-01 -1.00591505e+00 4.71370012e-01
4.16278422e-01 -1.00580490e+00 1.59120664e-01 -1.88280102e-02
3.11498106e-01 8.71344805e-01 -8.32794011e-01 1.87073362e+00
-3.65412176e-01 5.21265209e-01 -3.69629040e-02 -8.42382193e-01
5.23932517e-01 4.54363525e-01 4.73612338e-01 -4.02586371e-01
1.63682550e-01 1.59875005e-01 -5.08534312e-01 -1.84868306e-01
8.07790041e-01 3.88487130e-01 7.20564649e-02 8.93613279e-01
1.18468694e-01 1.08412668e-01 8.05315655e-03 8.26542854e-01
1.07740033e+00 -1.31322414e-01 -4.10568625e-01 -1.95183679e-01
-1.15765698e-01 9.59333777e-02 3.77615452e-01 1.13089335e+00
9.38045308e-02 3.75651538e-01 3.32632124e-01 -3.70533586e-01
-7.77361751e-01 -1.25696301e+00 -6.55239895e-02 1.59035873e+00
2.29292065e-02 -2.37592563e-01 -4.39522684e-01 -8.35987151e-01
2.99195677e-01 8.56417656e-01 -5.66146076e-01 -3.45191248e-02
-1.17998503e-01 -4.82235074e-01 3.92678052e-01 4.04428273e-01
9.63817686e-02 -8.83271992e-01 2.08272025e-01 -3.70055288e-02
-9.75267664e-02 -6.75182819e-01 -6.68350756e-01 1.36318624e-01
-6.84910357e-01 -1.00200748e+00 -7.88194716e-01 -6.56850874e-01
6.50210977e-01 2.28734359e-01 1.67830467e+00 2.47998580e-01
-2.93398738e-01 5.73504567e-01 -1.20413169e-01 -4.85398859e-01
-5.61771393e-02 3.63407612e-01 -1.37487501e-01 -5.78915894e-01
9.46909010e-01 -3.58581036e-01 -1.09291661e+00 5.86422021e-03
-9.66246724e-01 -1.86593711e-01 6.22390866e-01 8.17874968e-01
7.00787961e-01 1.29686864e-02 5.66789746e-01 -1.39035642e+00
5.95816076e-01 -5.86738646e-01 -7.56228924e-01 5.18472493e-01
-8.28984618e-01 3.70718330e-01 2.41686493e-01 -6.80202901e-01
-1.14435482e+00 -3.71644735e-01 2.61769682e-01 -3.70736212e-01
-9.22338963e-02 5.02847314e-01 2.83861846e-01 4.12908316e-01
7.76483595e-01 -2.44924396e-01 -4.12146330e-01 -7.94699609e-01
8.92872691e-01 8.86221170e-01 3.93539488e-01 -9.79654849e-01
6.01338565e-01 3.53684515e-01 -3.48626316e-01 -3.42806205e-02
-1.63357735e+00 -6.68339312e-01 -4.13119644e-02 2.36700922e-01
5.19208610e-01 -1.26858866e+00 -9.11020041e-01 2.06847191e-01
-1.04849887e+00 -7.59977996e-01 -4.86283571e-01 7.10221171e-01
-1.98961630e-01 3.86620574e-02 -1.03805125e+00 -6.61963165e-01
-6.51474595e-01 -8.78061593e-01 1.17319584e+00 2.79498070e-01
2.28299275e-02 -9.92042661e-01 3.24799091e-01 5.75213313e-01
5.74567974e-01 -5.03101051e-01 9.07157063e-01 -1.01937187e+00
-1.08535910e+00 -3.04804981e-01 -4.82478023e-01 2.29457363e-01
-3.52989770e-02 -5.03209755e-02 -1.18588543e+00 -3.13619733e-01
-5.55277407e-01 -9.56317186e-01 1.30531192e+00 4.67221618e-01
1.41582549e+00 -5.72604239e-01 -5.03671288e-01 2.08748579e-01
1.19906819e+00 -3.33321333e-01 2.70946056e-01 8.60273615e-02
5.95948100e-01 4.63467985e-01 5.32365978e-01 4.23118979e-01
9.58309650e-01 4.48973566e-01 1.77110091e-01 -3.19564015e-01
-2.34263003e-01 -4.02778208e-01 4.91129700e-03 5.81420541e-01
3.26616943e-01 -5.93791306e-01 -8.17684770e-01 8.07001889e-01
-1.84034204e+00 -8.04003239e-01 1.97517112e-01 2.12125540e+00
1.63505375e+00 -2.16297314e-01 -2.69791037e-01 -5.31212628e-01
4.90109563e-01 1.11974873e-01 -7.36800015e-01 1.46154076e-01
2.40132540e-01 1.86543077e-01 6.95481420e-01 9.60249603e-01
-1.09679568e+00 1.12814593e+00 5.92810059e+00 1.11499739e+00
-8.08037043e-01 2.31748551e-01 8.74389052e-01 -3.36217731e-01
-8.87618005e-01 2.10261330e-01 -1.01829982e+00 2.17777699e-01
8.91320169e-01 -1.17591031e-01 5.61865985e-01 1.05916786e+00
-2.49121889e-01 -9.87912416e-02 -1.15462971e+00 4.67838466e-01
-2.79852033e-01 -1.19234645e+00 -1.92956009e-03 7.42472634e-02
1.15689409e+00 5.92609584e-01 3.14923555e-01 5.49360096e-01
1.33222616e+00 -7.76973486e-01 2.77836353e-01 7.29395449e-01
6.71296895e-01 -4.90451187e-01 6.19031608e-01 2.87840873e-01
-5.59849381e-01 9.05477256e-02 -5.21725833e-01 2.12684035e-01
-3.41154225e-02 1.08159924e+00 -7.73569405e-01 1.58921525e-01
6.05050862e-01 2.84971923e-01 -5.30214846e-01 7.94968188e-01
-5.75943589e-01 9.42740798e-01 -4.77336198e-01 -1.08683757e-01
2.05886632e-01 2.26920187e-01 1.47554412e-01 1.14364350e+00
7.50435516e-02 3.42965811e-01 5.50713122e-01 9.08517957e-01
-8.02419543e-01 -1.82413876e-01 -2.01908022e-01 -2.71371901e-01
9.88977253e-01 1.26444936e+00 -8.00419524e-02 -8.92009020e-01
-3.52820456e-01 6.25728369e-01 6.81867003e-01 7.04478085e-01
-4.36238199e-01 -1.78048909e-01 5.48956573e-01 -2.22761258e-01
1.40040264e-01 2.12234780e-01 6.92626089e-02 -1.20362163e+00
-1.38087735e-01 -6.15866125e-01 6.88351274e-01 -7.76318073e-01
-1.79474902e+00 4.20719922e-01 1.59720004e-01 -4.98289108e-01
-4.44403738e-01 -4.67237294e-01 -1.44567788e-01 1.02792060e+00
-2.01186156e+00 -1.18458116e+00 -2.05258772e-01 6.05543196e-01
1.31083310e-01 3.36721353e-02 9.51193094e-01 3.29968840e-01
-2.41235480e-01 8.57702017e-01 4.94449347e-01 1.24915771e-01
1.04565465e+00 -1.66603434e+00 3.71831000e-01 1.92855746e-01
3.52227360e-01 9.44354057e-01 4.17412698e-01 -4.45954502e-01
-1.27335680e+00 -9.99835968e-01 1.30697012e+00 -8.24785709e-01
8.04721415e-01 -1.79002017e-01 -9.13905561e-01 9.70647752e-01
3.47008735e-01 1.16893068e-01 8.37726653e-01 7.52129853e-01
-6.51333749e-01 3.77263911e-02 -9.81945634e-01 3.65054458e-01
6.86558008e-01 -7.54170179e-01 -5.12800992e-01 8.86453688e-01
1.04402769e+00 -3.42737526e-01 -9.64034557e-01 2.66932577e-01
4.54920709e-01 -3.53003979e-01 1.11248326e+00 -7.98685014e-01
3.26277971e-01 -1.74338102e-01 -5.01784086e-02 -1.04977381e+00
-3.63841802e-01 -5.11143386e-01 -5.03374159e-01 1.16201973e+00
7.28409767e-01 -5.08108556e-01 1.15823436e+00 1.13828552e+00
-1.22701423e-02 -8.84134591e-01 -4.51297462e-01 -5.81043720e-01
2.65265375e-01 -3.79085720e-01 5.72802961e-01 1.05279505e+00
-1.72163099e-01 3.87980610e-01 -8.95539075e-02 3.24011773e-01
7.70056009e-01 1.27150074e-01 7.14357257e-01 -1.25228560e+00
-5.56877971e-01 -2.37112433e-01 2.22000971e-01 -1.51825607e+00
3.08521509e-01 -9.52471793e-01 7.14084357e-02 -1.83543277e+00
6.43757939e-01 -8.37377310e-01 -8.22365522e-01 9.15863037e-01
-4.68058527e-01 2.44829729e-01 -8.55230615e-02 2.93734819e-01
-1.20443308e+00 6.83499753e-01 1.29303694e+00 -2.82545447e-01
-3.25593315e-02 1.37305008e-02 -9.38161731e-01 4.70761120e-01
6.78477108e-01 -8.55073333e-01 -6.71815515e-01 -1.08084393e+00
4.37841386e-01 -3.02781854e-02 3.72867078e-01 -2.81190395e-01
5.29520571e-01 -1.54941212e-02 2.51807451e-01 -7.80594945e-01
4.26553577e-01 -5.19490123e-01 -1.78068087e-01 -4.88150083e-02
-1.18319535e+00 -2.96459258e-01 -1.08043008e-01 5.39425850e-01
-5.57089373e-02 -2.27536544e-01 1.90697148e-01 -1.29181340e-01
-2.09562734e-01 6.29517972e-01 7.51049519e-02 5.59646010e-01
2.67582238e-01 7.71661818e-01 -9.36810136e-01 -5.30597925e-01
-4.90362614e-01 6.09168649e-01 3.47893715e-01 1.66266307e-01
1.65316582e-01 -1.16533744e+00 -7.89463580e-01 -3.41355890e-01
9.78096724e-02 5.07945657e-01 3.20805758e-01 4.65651810e-01
-7.10878596e-02 7.52227485e-01 7.31713831e-01 -3.64104658e-01
-8.02698851e-01 6.68033063e-01 -5.14988750e-02 -6.14003003e-01
-4.08549160e-01 1.32090366e+00 3.43790799e-01 -7.45591998e-01
6.79331481e-01 -1.03436604e-01 1.63516626e-01 -1.68462828e-01
6.19806528e-01 3.03874522e-01 -8.69776681e-02 2.43268594e-01
-1.02177709e-01 1.52094036e-01 -5.94428897e-01 -2.39446267e-01
1.19592774e+00 -5.69784045e-02 -5.65517098e-02 2.29693726e-01
1.39205205e+00 -3.21023837e-02 -1.34551334e+00 -8.67014527e-01
1.46609947e-01 -4.93284225e-01 6.58883035e-01 -1.30802381e+00
-1.05788612e+00 7.83860981e-01 3.79693419e-01 8.28268602e-02
9.06583607e-01 4.93237048e-01 7.00148225e-01 1.01716733e+00
3.29408526e-01 -8.83120537e-01 2.52459571e-02 2.79700130e-01
4.47980046e-01 -1.47492802e+00 1.76430479e-01 -5.59725836e-02
-5.19930482e-01 5.42202413e-01 4.48017567e-01 -9.93696153e-02
9.15041745e-01 -1.36145363e-02 2.94800699e-01 -5.32303572e-01
-1.33166325e+00 -3.54300499e-01 5.30795634e-01 2.74955928e-01
7.04200864e-01 1.60576493e-01 -7.52522871e-02 5.08478761e-01
-7.88603574e-02 1.48927301e-01 9.06009674e-02 6.61097348e-01
-4.06846195e-01 -1.17840517e+00 8.74184668e-02 5.54526508e-01
-8.08474123e-01 -6.18145227e-01 -2.16317117e-01 6.64884210e-01
-3.75525683e-01 9.58820105e-01 1.33068800e-01 5.73935136e-02
-1.24912500e-01 -5.59085235e-02 2.24307016e-01 -7.81289995e-01
-3.44146252e-01 4.73162383e-02 1.00388922e-01 -4.95567173e-01
-2.78022200e-01 -3.56064349e-01 -1.23718488e+00 -3.55509043e-01
-5.13107777e-01 5.69439828e-01 7.01915383e-01 6.67001843e-01
6.29499853e-01 2.53647417e-01 6.64726794e-01 -4.91300792e-01
-1.13771343e+00 -1.13570940e+00 -5.86931646e-01 2.60391355e-01
4.68636274e-01 -4.91032392e-01 -5.44291258e-01 -1.57705799e-01] | [11.398509979248047, 7.669262409210205] |
64cf5a11-e625-48ff-aeea-d837a9a18e6f | radarformer-lightweight-and-accurate-real | 2304.08447 | null | https://arxiv.org/abs/2304.08447v1 | https://arxiv.org/pdf/2304.08447v1.pdf | RadarFormer: Lightweight and Accurate Real-Time Radar Object Detection Model | The performance of perception systems developed for autonomous driving vehicles has seen significant improvements over the last few years. This improvement was associated with the increasing use of LiDAR sensors and point cloud data to facilitate the task of object detection and recognition in autonomous driving. However, LiDAR and camera systems show deteriorating performances when used in unfavorable conditions like dusty and rainy weather. Radars on the other hand operate on relatively longer wavelengths which allows for much more robust measurements in these conditions. Despite that, radar-centric data sets do not get a lot of attention in the development of deep learning techniques for radar perception. In this work, we consider the radar object detection problem, in which the radar frequency data is the only input into the detection framework. We further investigate the challenges of using radar-only data in deep learning models. We propose a transformers-based model, named RadarFormer, that utilizes state-of-the-art developments in vision deep learning. Our model also introduces a channel-chirp-time merging module that reduces the size and complexity of our models by more than 10 times without compromising accuracy. Comprehensive experiments on the CRUW radar dataset demonstrate the advantages of the proposed method. Our RadarFormer performs favorably against the state-of-the-art methods while being 2x faster during inference and requiring only one-tenth of their model parameters. The code associated with this paper is available at https://github.com/YahiDar/RadarFormer. | ['Hisham Cholakkal', 'Jean Lahoud', 'Yahia Dalbah'] | 2023-04-17 | null | null | null | null | ['radar-object-detection'] | ['robots'] | [ 8.23619068e-02 -4.12587076e-01 9.27420631e-02 -7.06310153e-01
-6.87355220e-01 -3.28717470e-01 8.47386897e-01 -1.26136959e-01
-7.54385948e-01 4.16063279e-01 -5.14154673e-01 -4.68077093e-01
-1.10180311e-01 -1.02087057e+00 -6.30248904e-01 -7.41012752e-01
-4.09284653e-03 5.70205986e-01 4.52723652e-01 -3.98906648e-01
1.59047544e-01 8.18034232e-01 -2.03898335e+00 -6.25084341e-02
8.58598650e-01 1.11861837e+00 4.80373681e-01 7.25718498e-01
3.33192274e-02 3.37042123e-01 -3.79436821e-01 -1.26817569e-01
6.01319909e-01 1.53493583e-01 2.01440021e-01 -3.14850450e-01
1.01261508e+00 -5.19922972e-01 -5.27668357e-01 9.15079355e-01
6.34672642e-01 -2.49766596e-02 4.30186361e-01 -1.24224615e+00
-2.31748655e-01 1.31353199e-01 -6.95712507e-01 4.51062620e-01
-3.37433130e-01 1.46841377e-01 8.27985704e-01 -1.00384045e+00
9.49384198e-02 1.07031274e+00 4.73092258e-01 2.08048224e-01
-8.85654271e-01 -1.12099600e+00 -1.60612583e-01 8.96979570e-01
-1.44014871e+00 -5.59845567e-01 7.24620819e-01 -4.72203374e-01
1.01364827e+00 -6.03237227e-02 5.10920823e-01 8.62734199e-01
4.65264529e-01 4.49735910e-01 1.12634170e+00 -1.07474186e-01
6.16903380e-02 1.79087594e-01 3.66795510e-01 6.16838813e-01
6.36962652e-01 8.21771264e-01 -4.98487204e-01 2.91978210e-01
3.03926289e-01 6.32882714e-02 2.43633986e-02 -2.87843585e-01
-7.96673596e-01 1.19929326e+00 5.11121035e-01 -3.25096622e-02
-4.26918477e-01 2.80203909e-01 4.73743007e-02 2.87158877e-01
3.55982214e-01 8.63502026e-02 -1.67445451e-01 9.78303105e-02
-1.16811991e+00 4.71568286e-01 5.17820418e-01 8.10702384e-01
8.20795774e-01 4.15118665e-01 3.75280410e-01 6.98342204e-01
3.78413886e-01 1.19106948e+00 -3.83942015e-02 -8.05198312e-01
2.77954191e-01 2.68354803e-01 5.75329512e-02 -7.45106936e-01
-6.81923985e-01 -7.42896914e-01 -7.94793904e-01 7.20043838e-01
1.34261623e-01 -9.58088785e-03 -1.27036035e+00 1.26865923e+00
2.57872522e-01 2.24300489e-01 2.49685243e-01 9.71221209e-01
8.10017169e-01 7.02004671e-01 -2.35180199e-01 1.09494224e-01
1.45304692e+00 -6.11202180e-01 -5.05479813e-01 -5.06274402e-01
1.39865369e-01 -8.43292415e-01 4.03040200e-01 4.86068785e-01
-4.74344432e-01 -9.98952329e-01 -1.44251215e+00 -1.67306811e-02
-6.28516972e-01 7.17080235e-02 6.40216351e-01 6.98231697e-01
-7.77281642e-01 7.63829648e-02 -8.61242473e-01 -4.45952415e-01
4.66226101e-01 1.64449036e-01 -2.83675641e-02 -2.90838629e-01
-1.05653405e+00 1.29664373e+00 1.27570719e-01 3.21589947e-01
-1.04663253e+00 -7.95255303e-01 -7.96115339e-01 -1.00814134e-01
3.78914833e-01 -2.87799984e-01 1.21585536e+00 -3.02791893e-01
-1.17876256e+00 6.36455476e-01 -1.39388934e-01 -1.09868896e+00
5.21762729e-01 -5.08667767e-01 -5.84303796e-01 1.65303983e-02
-1.86512098e-01 1.00777888e+00 1.03597975e+00 -1.24113989e+00
-1.11182964e+00 -3.65753353e-01 -3.50951552e-02 2.86860410e-02
2.84095824e-01 -4.03317571e-01 -1.61193237e-01 9.92011465e-03
-4.58425423e-03 -9.92717266e-01 -7.06968084e-02 -5.33572882e-02
7.14189783e-02 -1.74000546e-01 1.19567299e+00 -1.73809588e-01
4.72632408e-01 -2.08395052e+00 -5.13998747e-01 -7.35830292e-02
1.95588320e-01 5.51326513e-01 -1.22765742e-01 4.52656358e-01
2.70596683e-01 -4.04049635e-01 -2.48252183e-01 -2.53117621e-01
-6.52717203e-02 3.94558400e-01 -6.20180070e-01 6.23805821e-01
3.21499795e-01 6.81222320e-01 -4.71726388e-01 -2.45116279e-01
6.38081253e-01 6.72336817e-01 -1.86285615e-01 9.88659188e-02
-9.98112373e-04 1.88143417e-01 -2.13534489e-01 7.53833592e-01
1.24107730e+00 4.56981868e-01 -3.66324425e-01 -2.35951617e-01
-6.60327971e-01 1.93902999e-01 -9.58592355e-01 1.12279081e+00
-4.63359684e-01 1.24826503e+00 1.97475210e-01 -1.17552590e+00
1.25129354e+00 -9.37368870e-02 1.81393236e-01 -1.22662246e+00
1.89080462e-01 2.83961892e-01 5.19585371e-01 -5.21086991e-01
6.11723185e-01 -4.20313030e-01 1.09317094e-01 -6.87790960e-02
-2.46495157e-01 -4.09499764e-01 1.38401493e-01 2.66012433e-03
8.26093316e-01 -1.16641834e-01 1.15042888e-01 -4.71992083e-02
4.77157503e-01 2.49213412e-01 5.17738223e-01 9.56811547e-01
-2.67388344e-01 3.97861302e-01 -2.00894147e-01 -4.05268162e-01
-7.80550420e-01 -1.28128600e+00 -6.64955437e-01 7.88763463e-01
1.92020640e-01 1.77113488e-01 -1.42441005e-01 -1.96452081e-01
5.07336617e-01 9.09113944e-01 -4.25022721e-01 -1.05095059e-01
-5.26757658e-01 -8.84576201e-01 6.71159804e-01 5.98811924e-01
7.99262881e-01 -6.87829792e-01 -1.30024898e+00 1.14327624e-01
1.10942878e-01 -1.33607304e+00 4.15739030e-01 3.92937422e-01
-7.71453679e-01 -8.96397829e-01 -2.43103325e-01 -3.84346068e-01
1.89426959e-01 8.75638843e-01 8.14691782e-01 -1.15075201e-01
-6.96581602e-01 2.24803314e-01 -3.00396919e-01 -1.13909221e+00
-7.40265986e-03 -1.65776059e-01 2.18940780e-01 -5.78158200e-02
7.32610166e-01 -4.95356858e-01 -4.87128079e-01 6.21070936e-02
-6.31062746e-01 -1.73034638e-01 1.07424176e+00 5.70915163e-01
2.75346637e-01 -1.93604399e-02 5.69812298e-01 -6.17970169e-01
1.81261390e-01 -2.50018418e-01 -1.16807318e+00 -4.76724297e-01
-7.09854007e-01 -1.96864437e-02 4.41949815e-01 2.03797389e-02
-8.82258892e-01 8.45721588e-02 -1.55860499e-01 -4.09952164e-01
-3.77358973e-01 3.80147666e-01 1.74279347e-01 -2.83053249e-01
4.99865413e-01 2.41338253e-01 9.59103275e-03 -4.09514964e-01
2.98825711e-01 8.04338813e-01 7.71027029e-01 -2.16518552e-03
1.34405863e+00 8.70947480e-01 2.47678280e-01 -1.45386612e+00
-7.01924026e-01 -7.20744848e-01 -6.03148460e-01 -3.60909075e-01
6.06567383e-01 -1.22056019e+00 -6.85217977e-01 5.21924436e-01
-1.03724492e+00 5.58819324e-02 1.41090984e-02 7.99564123e-01
-2.83850253e-01 1.34380296e-01 2.53052227e-02 -1.24746740e+00
-2.50468373e-01 -9.88745511e-01 9.84971523e-01 3.63783717e-01
3.62971574e-01 -5.07107079e-01 2.69659478e-02 6.06619596e-01
5.77229500e-01 5.71710840e-02 6.03235900e-01 -3.39143157e-01
-1.09977818e+00 -4.16489422e-01 -5.42577088e-01 4.21247393e-01
-1.91056982e-01 -1.08213961e-01 -1.26154947e+00 -2.80780822e-01
-4.06174958e-02 -1.23816438e-01 1.60062146e+00 4.71704602e-01
5.46825945e-01 5.63791275e-01 -3.81442487e-01 8.41391623e-01
1.56327581e+00 4.29514557e-01 6.29590154e-01 3.15504491e-01
5.21858394e-01 7.11417317e-01 9.58051443e-01 2.83108652e-01
5.26314497e-01 5.90580225e-01 8.10832500e-01 -1.39252812e-01
-1.56082839e-01 1.62354365e-01 3.31423104e-01 3.59474242e-01
-2.25837067e-01 -1.82565227e-01 -9.82470512e-01 6.54470503e-01
-1.62790263e+00 -1.08179247e+00 -3.52692634e-01 2.14015794e+00
2.32509315e-01 5.37822723e-01 -2.77016133e-01 5.62540926e-02
2.48973891e-01 2.94916749e-01 -5.92674971e-01 -4.48804379e-01
9.23655033e-02 3.83123934e-01 1.03921127e+00 7.13400185e-01
-1.21140444e+00 9.04748142e-01 5.20943642e+00 4.77263987e-01
-1.42158961e+00 -2.42514461e-02 -1.89142808e-01 -1.86655685e-01
1.41283184e-01 6.78488892e-03 -1.29901290e+00 2.16870725e-01
1.16208696e+00 3.18702608e-02 1.18558735e-01 7.51536906e-01
4.03761446e-01 -5.76938391e-01 -8.67018223e-01 1.02445984e+00
1.06392123e-01 -1.29674387e+00 -1.52743518e-01 2.27963403e-01
1.94040254e-01 7.45242715e-01 1.08467802e-01 4.59898978e-01
2.95980014e-02 -9.60176468e-01 7.38038838e-01 5.01347363e-01
4.96555328e-01 -9.79170322e-01 1.01642776e+00 4.61010039e-01
-1.15722191e+00 -3.85098085e-02 -7.88883686e-01 -3.41046840e-01
9.80346054e-02 8.42336297e-01 -1.02805221e+00 6.02950752e-01
8.35340798e-01 5.42226970e-01 -4.72061455e-01 1.16494513e+00
-1.69666305e-01 6.96411610e-01 -5.78826010e-01 5.67207336e-02
4.72898394e-01 -2.02730775e-01 6.75681174e-01 1.37572145e+00
3.72936606e-01 3.48768160e-02 2.28552684e-01 7.23920286e-01
2.40755811e-01 -5.78334749e-01 -9.02094901e-01 2.61958241e-01
5.84631860e-01 1.56231165e+00 -3.82136703e-01 -1.20546021e-01
-5.65459192e-01 2.18075708e-01 -4.52166125e-02 2.43838474e-01
-1.16401052e+00 -5.46771467e-01 1.01568222e+00 1.93463728e-01
6.95118189e-01 -7.33745933e-01 -3.01605374e-01 -5.03092647e-01
-1.20332070e-01 -3.96007478e-01 -8.82773176e-02 -5.73798656e-01
-9.94298518e-01 4.90028799e-01 1.68329880e-01 -1.33811569e+00
-5.16346283e-02 -7.80511022e-01 -5.10643184e-01 8.37054014e-01
-2.38293242e+00 -1.05905557e+00 -7.26916730e-01 2.42540970e-01
5.68663955e-01 -2.99963623e-01 4.55710143e-01 4.15068954e-01
-3.08698982e-01 2.75221676e-01 1.42401502e-01 -1.26184642e-01
8.09124529e-01 -9.57443237e-01 4.75903213e-01 1.03153729e+00
1.09569840e-01 1.96622461e-01 1.02099776e+00 -4.64692801e-01
-1.66955078e+00 -1.25869894e+00 7.44342566e-01 -2.10721686e-01
6.07195854e-01 -6.48765385e-01 -7.79929519e-01 3.81487489e-01
2.70311892e-01 -1.18620820e-01 3.35903376e-01 -6.81372061e-02
-4.49106187e-01 -7.20280945e-01 -8.53345215e-01 1.71344221e-01
6.81975663e-01 -3.34123433e-01 -7.50838101e-01 1.65304437e-01
3.25920612e-01 -2.70663500e-01 -1.95513338e-01 7.53303289e-01
6.88613236e-01 -1.17193890e+00 9.21146214e-01 2.54722182e-02
-3.71534117e-02 -5.89606941e-01 -3.99522275e-01 -1.11385942e+00
-4.25976366e-01 -5.43627851e-02 5.51404012e-03 7.90090144e-01
3.12656105e-01 -9.20878172e-01 7.26135969e-01 -4.08970833e-01
-4.05128032e-01 -4.77108777e-01 -1.14439642e+00 -1.01959574e+00
-1.67536363e-01 -7.43948460e-01 4.22564894e-02 4.15896058e-01
-7.31932759e-01 5.65037429e-01 -2.56221205e-01 7.20804095e-01
1.03217757e+00 5.28506994e-01 8.44719708e-01 -1.64039612e+00
3.05957086e-02 -1.81261107e-01 -6.72312915e-01 -9.22148108e-01
-7.51907229e-02 -7.28216350e-01 4.65378731e-01 -1.62186944e+00
-1.64524898e-01 -3.13505262e-01 -6.58721253e-02 2.79691875e-01
1.44728750e-01 5.19664586e-01 3.78715783e-01 9.96182263e-02
-7.53010735e-02 6.23364568e-01 6.75184309e-01 -3.30382735e-01
1.40234351e-01 4.29929011e-02 -4.41076249e-01 7.04207480e-01
9.92618382e-01 -3.76229018e-01 -3.13287646e-01 -6.11085713e-01
-9.17339996e-02 -3.46607119e-01 6.28938556e-01 -1.62200451e+00
5.01457393e-01 5.76084219e-02 4.92483050e-01 -1.29674363e+00
7.83611834e-01 -8.60794902e-01 -1.37061715e-01 6.13224745e-01
2.50601292e-01 -1.24126762e-01 4.67744946e-01 5.94184577e-01
-2.13183835e-01 3.98199223e-02 1.17187095e+00 1.43439248e-01
-1.14940691e+00 1.02386974e-01 -7.38248169e-01 -3.38967741e-01
9.74947989e-01 -3.68161023e-01 -5.61364353e-01 -3.27125698e-01
-1.20575942e-01 4.59961951e-01 -2.21278425e-02 6.37712121e-01
8.80229890e-01 -8.17236841e-01 -1.07465518e+00 4.49593008e-01
2.18814090e-01 -7.92399570e-02 1.77261859e-01 9.09112751e-01
-3.90892774e-01 8.61478388e-01 -3.13757628e-01 -8.80142212e-01
-1.24507189e+00 2.61694849e-01 3.44506651e-01 1.68343678e-01
-7.32951820e-01 7.44305670e-01 2.36486793e-01 -2.70095766e-01
1.99967265e-01 -4.00603890e-01 -2.08233178e-01 2.71211356e-01
5.84900260e-01 3.83118272e-01 3.82304996e-01 -5.06962955e-01
-6.02754474e-01 7.12091267e-01 -3.52626413e-01 -5.18202893e-02
1.29888856e+00 -2.71469150e-02 2.87236542e-01 4.68367398e-01
7.65060306e-01 -1.08429730e-01 -1.27044570e+00 -1.41062573e-01
-2.02183157e-01 -2.91521788e-01 6.83827281e-01 -7.40047216e-01
-1.00632608e+00 1.25541079e+00 1.09204650e+00 1.05883941e-01
9.82060671e-01 -9.47081596e-02 8.26389790e-01 7.87959218e-01
4.07613605e-01 -1.00082159e+00 -4.44589347e-01 9.32295203e-01
6.64675355e-01 -1.42245078e+00 1.83203503e-01 -2.11256474e-01
-3.65016013e-01 1.06159103e+00 4.92431372e-01 -3.42226893e-01
6.56948626e-01 6.15652263e-01 4.04954880e-01 -2.38284305e-01
-7.79340446e-01 -8.83145332e-01 1.02535158e-01 7.61352718e-01
1.28702626e-01 2.29991302e-01 -1.37954623e-01 8.36592242e-02
-3.34817171e-01 -3.34113747e-01 5.93646944e-01 8.95238936e-01
-1.11760116e+00 -7.77946055e-01 -5.53507984e-01 4.77000892e-01
1.66142240e-01 -1.94599941e-01 -1.85589284e-01 9.00283515e-01
4.44998056e-01 1.21758914e+00 2.86096632e-01 -4.95907187e-01
5.05260229e-01 3.02494299e-02 4.59225982e-01 -3.46084267e-01
-6.40547499e-02 -3.39978859e-02 8.21596310e-02 -5.65818965e-01
-4.09884185e-01 -7.38482416e-01 -1.25877333e+00 -3.34469825e-01
-2.55778909e-01 -9.24792066e-02 1.03776860e+00 8.15181911e-01
2.97769696e-01 6.29037559e-01 5.15356243e-01 -9.92705882e-01
-5.53463221e-01 -9.69554245e-01 -5.11998355e-01 -4.45064485e-01
7.39449143e-01 -1.04815912e+00 -2.57954061e-01 -1.06098048e-01] | [7.860193252563477, -1.4780597686767578] |
2b7a84df-6802-4939-a1ed-b07e57652e48 | transedrp-dual-transformer-model-with-edge | 2210.17401 | null | https://arxiv.org/abs/2210.17401v1 | https://arxiv.org/pdf/2210.17401v1.pdf | TransEDRP: Dual Transformer model with Edge Emdedded for Drug Respond Prediction | GNN-based methods have achieved excellent results as a mainstream task in drug response prediction tasks in recent years. Traditional GNN methods use only the atoms in a drug molecule as nodes to obtain the representation of the molecular graph through node information passing, whereas the method using the transformer can only extract information about the nodes. However, the covalent bonding and chirality of a drug molecule have a great influence on the pharmacological properties of the molecule, and these information are implied in the chemical bonds formed by the edges between the atoms. In addition, CNN methods for modelling cell lines genomics sequences can only perceive local rather than global information about the sequence. In order to solve the above problems, we propose the decoupled dual transformer structure with edge embedded for drug respond prediction (TransEDRP), which is used for the representation of cell line genomics and drug respectively. For the drug branch, we encoded the chemical bond information within the molecule as the embedding of the edge in the molecular graph, extracted the global structural and biochemical information of the drug molecule using graph transformer. For the branch of cell lines genomics, we use the multi-headed attention mechanism to globally represent the genomics sequence. Finally, the drug and genomics branches are fused to predict IC50 values through the transformer layer and the fully connected layer, which two branches are different modalities. Extensive experiments have shown that our method is better than the current mainstream approach in all evaluation indicators. | ['Hu Wenbin', 'Li Kun'] | 2022-10-23 | null | null | null | null | ['drug-response-prediction'] | ['medical'] | [ 1.13987893e-01 -1.11382738e-01 -4.53782141e-01 -6.15485348e-02
-9.39868316e-02 -4.27055508e-01 3.48350585e-01 3.50314856e-01
-1.49196789e-01 8.43633771e-01 2.86288559e-01 -3.28820229e-01
3.27487811e-02 -1.04620409e+00 -9.80019212e-01 -9.71391857e-01
1.15821369e-01 1.95487961e-01 -1.10575026e-02 -3.17615777e-01
2.26789936e-01 4.53273624e-01 -8.00738931e-01 4.57033604e-01
9.67545807e-01 9.89583552e-01 8.23090971e-02 1.70246884e-01
-1.97767869e-01 9.17887568e-01 -4.29800302e-01 -2.51937747e-01
-2.41336510e-01 -6.20546162e-01 -6.09555423e-01 -4.21385646e-01
-4.17948924e-02 -7.87250847e-02 -6.15617275e-01 1.11868858e+00
7.88441002e-01 -1.71414539e-01 6.33623302e-01 -9.33190465e-01
-9.19625282e-01 6.17280662e-01 -4.31417048e-01 -1.47563249e-01
6.24773622e-01 3.48114401e-01 9.29184437e-01 -7.98614383e-01
8.62203240e-01 1.33306694e+00 4.58208472e-01 5.14181137e-01
-8.80702734e-01 -7.14030802e-01 3.11999172e-01 4.34829473e-01
-1.37719858e+00 -7.32683837e-02 6.84455097e-01 -5.04691422e-01
1.21712995e+00 1.56811878e-01 7.02210188e-01 1.13302791e+00
7.50420094e-01 4.05321091e-01 5.93555033e-01 2.06530884e-01
1.51366517e-02 -3.21561605e-01 -2.10439153e-02 9.15270329e-01
5.55098765e-02 6.29835725e-02 -3.73738378e-01 -2.39664584e-01
7.35913396e-01 5.42687893e-01 -5.64720929e-01 -2.31408030e-02
-1.09646130e+00 8.69661152e-01 1.17270505e+00 2.48045161e-01
-4.08072114e-01 1.84603527e-01 4.60625947e-01 6.76686168e-02
2.88609833e-01 3.42767388e-01 -5.34895420e-01 4.18961465e-01
-2.33172342e-01 -5.62797002e-02 7.69733608e-01 6.78613067e-01
6.90869033e-01 -1.36975095e-01 -1.76416948e-01 5.04598260e-01
5.76338947e-01 -5.70312291e-02 5.79340935e-01 -5.77519685e-02
4.46316093e-01 1.08009648e+00 -2.22432286e-01 -1.29078543e+00
-7.80552506e-01 -5.70586979e-01 -1.15362740e+00 -2.94836611e-01
-1.92642007e-02 -7.63674155e-02 -1.09944642e+00 1.85872078e+00
4.23667610e-01 5.59296906e-01 -8.46941117e-03 8.06204200e-01
1.37794900e+00 7.66875386e-01 4.78273988e-01 -3.49578828e-01
1.31550884e+00 -9.73657012e-01 -6.83290482e-01 1.71549737e-01
9.77656841e-01 -3.31965476e-01 5.09018123e-01 1.18887641e-01
-6.65368319e-01 -4.36104655e-01 -1.17215860e+00 -2.52552301e-01
-6.95124030e-01 -1.52805552e-01 6.79284692e-01 1.57299399e-01
-7.61357367e-01 9.16898608e-01 -6.09767437e-01 5.31506576e-02
5.05393922e-01 6.83351696e-01 -4.90869254e-01 -2.52998233e-01
-1.72378147e+00 8.65194976e-01 6.51259661e-01 5.18389583e-01
-9.80726302e-01 -7.82976151e-01 -9.83855903e-01 2.89322555e-01
1.33194566e-01 -8.33916187e-01 5.17546237e-01 -6.98204279e-01
-1.36277640e+00 4.07595128e-01 1.15451775e-02 -1.52102128e-01
2.05077007e-01 3.32754582e-01 -4.17279691e-01 -4.16331440e-02
-2.36242965e-01 6.17726624e-01 1.79104969e-01 -6.67653739e-01
-1.90107822e-01 -7.21171796e-01 1.75512135e-01 3.40045512e-01
-1.40630156e-01 -3.38967413e-01 -4.54025865e-01 -4.47379678e-01
-5.03526651e-04 -5.95551312e-01 -4.33110386e-01 -2.42371842e-01
-5.85638106e-01 -3.00204247e-01 7.17665076e-01 -7.16541827e-01
1.42316353e+00 -2.00304365e+00 5.01923978e-01 1.60587624e-01
6.62256181e-01 1.60134315e-01 -3.22303474e-01 7.99312472e-01
-5.85585237e-01 4.62787837e-01 -5.58532812e-02 3.98667336e-01
-4.51225847e-01 -9.62666273e-02 -1.08324587e-02 5.01486421e-01
1.02105968e-01 1.20960057e+00 -1.09046042e+00 -1.22835457e-01
-3.94967720e-02 8.62113953e-01 -5.14996231e-01 2.52582692e-02
-6.54813230e-01 5.83942294e-01 -9.88909185e-01 4.94647980e-01
7.97856152e-01 -5.17503262e-01 4.96099323e-01 -6.26183987e-01
2.25952789e-01 2.98572212e-01 -5.68171501e-01 1.82475674e+00
-5.74253500e-02 -2.02304348e-01 -3.58829796e-01 -9.02358651e-01
9.47369576e-01 4.83355492e-01 5.58267951e-01 -8.00974965e-01
7.87975788e-02 2.08483502e-01 3.78233105e-01 -4.43507880e-01
-2.77690887e-01 -1.64206028e-01 3.61239612e-01 -4.50294614e-02
-6.76847249e-02 3.69724482e-01 -4.34688777e-02 2.25304112e-01
1.06932199e+00 2.31171161e-01 2.00090826e-01 -4.63750437e-02
8.64580452e-01 -3.16168100e-01 6.00191593e-01 8.40540603e-02
2.54266828e-01 2.57983059e-01 9.26023245e-01 -6.08440280e-01
-6.87653780e-01 -2.94187039e-01 -7.34843984e-02 6.68243289e-01
2.61417866e-01 -7.09003568e-01 -8.26409280e-01 -7.80593872e-01
-1.29760802e-01 1.47990093e-01 -9.06854272e-01 -6.64561152e-01
-4.23584104e-01 -1.11131525e+00 3.64197344e-01 3.58338952e-01
5.13432145e-01 -1.15556610e+00 2.56161213e-01 5.36124945e-01
1.99206993e-02 -7.32137382e-01 -6.04717374e-01 2.01106027e-01
-6.99920416e-01 -1.46584475e+00 -4.67122406e-01 -7.91272461e-01
5.55265725e-01 -2.18419254e-01 6.25815928e-01 4.56868052e-01
-2.60662735e-01 -4.63030994e-01 -1.10100307e-01 -1.62990868e-01
-1.18781589e-01 1.35102542e-02 -2.74721712e-01 2.08823442e-01
3.32170039e-01 -5.88990390e-01 -9.25990045e-01 1.53139904e-01
-8.85429919e-01 9.38884690e-02 3.41193318e-01 9.11208093e-01
7.18763590e-01 -1.28209531e-01 6.84045076e-01 -1.30830050e+00
6.94134474e-01 -7.78117955e-01 -4.31103051e-01 2.74527073e-01
-5.27357996e-01 1.76739067e-01 1.08158791e+00 -2.41302490e-01
-5.83528519e-01 1.63045332e-01 -5.00187933e-01 -1.96775392e-01
1.90798491e-01 1.20123816e+00 -7.01028407e-01 -2.04649240e-01
4.20949519e-01 3.56180698e-01 -1.55792296e-01 -5.03278971e-01
9.52551439e-02 4.82482046e-01 3.96182276e-02 -3.59273225e-01
5.83759462e-03 9.57843512e-02 5.37029684e-01 -4.07250553e-01
-4.97717172e-01 -1.23362467e-01 -3.50983679e-01 2.26476729e-01
9.73689377e-01 -8.94431174e-01 -1.25394464e+00 5.90372384e-01
-1.30035150e+00 -3.92659847e-03 3.32776517e-01 4.38566417e-01
-3.08799207e-01 5.98312974e-01 -9.60681558e-01 -2.44076699e-01
-4.90063906e-01 -1.49530411e+00 8.84373367e-01 1.47809297e-01
3.04543942e-01 -1.21181989e+00 8.42811465e-02 5.99097125e-02
7.14313462e-02 6.35623276e-01 1.57555962e+00 -7.52569914e-01
-6.30512655e-01 -2.62594759e-01 -3.08627069e-01 -6.53984323e-02
2.63096631e-01 -4.16051038e-02 -8.20553720e-01 -2.48643845e-01
-4.66336906e-02 -1.51496232e-01 1.05000174e+00 4.87460911e-01
1.39109564e+00 -3.48647207e-01 -7.64847279e-01 8.21055532e-01
1.57379007e+00 6.56077683e-01 9.84528542e-01 -6.59511387e-02
1.25591576e+00 2.75994301e-01 1.48876995e-01 1.76626176e-01
2.78837770e-01 4.64403540e-01 7.38485515e-01 -3.44077289e-01
-1.10230548e-02 -6.39742374e-01 3.52122217e-01 7.67939210e-01
-1.80893987e-01 -7.28287518e-01 -7.38196969e-01 -1.09248899e-01
-1.96539330e+00 -6.99172080e-01 -2.57305145e-01 2.20128393e+00
9.94349420e-01 -1.13726139e-01 -1.36877164e-01 -4.22105581e-01
7.37019658e-01 1.67786106e-02 -1.03940284e+00 -1.86285824e-01
-1.33669376e-01 1.03591561e-01 3.59102577e-01 5.49314916e-01
-6.30503774e-01 9.15229797e-01 6.15872574e+00 1.05082107e+00
-1.28922439e+00 -3.08518201e-01 9.76227999e-01 3.42414886e-01
-5.20674706e-01 9.31801870e-02 -6.72880054e-01 7.12136924e-01
9.23414350e-01 1.47297069e-01 2.79602766e-01 3.71330798e-01
2.63479322e-01 3.06922734e-01 -1.33788657e+00 9.97003257e-01
-2.10628703e-01 -1.67714190e+00 6.07222199e-01 3.09572130e-01
4.89278704e-01 4.27993648e-02 3.79801542e-03 5.19802943e-02
5.98594658e-02 -1.63941908e+00 -5.74233606e-02 7.17185915e-01
9.70981956e-01 -7.87236989e-01 7.76722491e-01 3.92888814e-01
-1.33099473e+00 4.91327047e-02 -5.84895313e-01 1.32912025e-01
-2.56833792e-01 2.81592309e-01 -6.79420471e-01 9.74930048e-01
8.24454054e-02 1.31516910e+00 -4.89861101e-01 1.01269376e+00
-2.17002764e-01 3.11583906e-01 -5.59402443e-03 -6.35669380e-02
4.02566016e-01 -6.07971430e-01 4.54509519e-02 9.85831916e-01
1.80446044e-01 1.80420786e-01 3.49004924e-01 8.79554391e-01
-5.98887742e-01 3.02344322e-01 -5.99722445e-01 -4.12035972e-01
2.20589340e-01 1.11581206e+00 -4.33262646e-01 -2.47661039e-01
-3.81366938e-01 8.37291718e-01 3.58489811e-01 6.79427624e-01
-8.79455209e-01 -5.53911626e-01 4.71904337e-01 -2.72439383e-02
6.13034964e-02 2.51196057e-01 1.70425549e-01 -9.74725664e-01
-2.84635067e-01 -1.11451173e+00 5.02229571e-01 -7.35206068e-01
-1.15456617e+00 5.89011431e-01 -6.56427920e-01 -9.90897417e-01
3.39610934e-01 -9.41549122e-01 -4.24581289e-01 1.31234705e+00
-1.50672889e+00 -1.08473802e+00 -9.57482085e-02 5.40549159e-01
2.27187142e-01 -3.58285941e-02 1.03882515e+00 4.38837409e-01
-1.03061819e+00 4.47779298e-01 1.60714209e-01 3.66349854e-02
4.74486023e-01 -8.97125125e-01 2.62688190e-01 8.84071887e-02
-2.39165723e-01 9.94558692e-01 1.87760368e-01 -9.72937465e-01
-1.56065893e+00 -1.07710481e+00 6.33123994e-01 -1.81501910e-01
5.98793745e-01 -3.33357871e-01 -9.58631456e-01 5.24294257e-01
4.44786660e-02 -6.35802299e-02 8.40301752e-01 -1.52309239e-01
-4.82824594e-01 1.91571921e-01 -9.33716059e-01 5.37794352e-01
1.07604325e+00 -6.29524529e-01 6.06653988e-02 8.01245570e-01
1.09519875e+00 -4.11313146e-01 -1.15052688e+00 4.71743613e-01
2.11022198e-01 -4.86311734e-01 9.95393753e-01 -1.15764928e+00
6.90431833e-01 -6.16541684e-01 1.30849317e-01 -1.35860217e+00
-6.43452764e-01 -3.49992543e-01 -1.27158314e-01 6.41433239e-01
7.41949260e-01 -7.94377029e-01 6.86472952e-01 3.20525825e-01
-2.06350312e-01 -1.24054897e+00 -7.33048260e-01 -2.11833954e-01
1.39691666e-01 1.97470680e-01 9.10232723e-01 1.13251805e+00
2.46844962e-01 1.02644682e+00 -3.96489680e-01 -1.57079682e-01
2.89274585e-02 2.37786733e-02 1.40269369e-01 -1.26768589e+00
-2.71528602e-01 -4.90631789e-01 -6.05210721e-01 -1.18851006e+00
5.58385998e-02 -1.31048417e+00 -5.65349460e-01 -1.75571811e+00
6.08723521e-01 -1.28461465e-01 -6.89447641e-01 5.38764179e-01
-2.12351725e-01 -4.07018736e-02 -1.19935624e-01 5.41237108e-02
-3.64244163e-01 7.03939795e-01 1.81624055e+00 -5.49452960e-01
-2.34094467e-02 -3.56624246e-01 -8.90982509e-01 3.67536306e-01
4.31154907e-01 -2.72188157e-01 -3.99426699e-01 -3.46997887e-01
7.96204567e-01 3.35866302e-01 1.13597773e-01 -4.94156897e-01
4.01707321e-01 -1.09975062e-01 6.80872202e-01 -3.06926250e-01
3.25898707e-01 -9.09908772e-01 4.30995762e-01 8.45506072e-01
-2.13546380e-01 -1.53460905e-01 6.43039942e-02 9.34551120e-01
-3.55668098e-01 -1.14052976e-02 5.64049125e-01 -3.48300695e-01
-1.88862786e-01 1.11389804e+00 -1.06238816e-02 -4.34224010e-01
7.06816316e-01 -2.09817544e-01 -5.37990570e-01 -1.78379372e-01
-8.79269302e-01 4.40399766e-01 2.70290911e-01 2.60681123e-01
7.90297687e-01 -1.34615862e+00 -6.87631249e-01 1.30024940e-01
4.95749824e-02 -2.65012626e-02 4.12654608e-01 8.51901233e-01
-7.13975132e-01 5.53859890e-01 -1.57231674e-01 -1.95265755e-01
-9.50573742e-01 1.01529908e+00 8.59352946e-01 -3.68203044e-01
-1.80112317e-01 8.29335809e-01 8.39319944e-01 -4.35500950e-01
6.93632662e-02 -1.95725828e-01 -7.00700343e-01 8.31116661e-02
4.58975166e-01 -1.99446455e-02 1.61006600e-01 -6.76211059e-01
-4.50033098e-01 6.73463941e-01 -5.04989862e-01 6.14735842e-01
1.42159510e+00 3.80394489e-01 -5.77966750e-01 1.41443476e-01
1.55361891e+00 -2.74408311e-01 -7.83726394e-01 -5.46121635e-02
-3.47423941e-01 -9.72612575e-02 1.74276456e-02 -1.07460594e+00
-1.14703143e+00 1.06367338e+00 4.18645501e-01 -9.38890427e-02
9.43691134e-01 -3.32765698e-01 8.44908357e-01 2.87132174e-01
2.09086314e-01 -5.34432828e-01 -1.36337087e-01 5.49009442e-01
8.62923682e-01 -8.14554870e-01 -6.42294646e-04 -7.00580359e-01
-4.07127023e-01 1.22789049e+00 6.74832821e-01 1.17435455e-01
6.47641897e-01 -2.18378723e-01 -3.42800796e-01 -6.86553538e-01
-7.30441809e-01 3.09807003e-01 3.86449695e-01 3.77171040e-01
8.84869039e-01 -3.55446190e-02 -4.85614657e-01 6.31933808e-01
3.00786972e-01 3.73819768e-02 1.23809978e-01 4.30969119e-01
-2.84501582e-01 -1.28889990e+00 1.75875634e-01 2.66950667e-01
-4.88073677e-01 -6.41227007e-01 -9.37869430e-01 3.85271013e-01
2.81718165e-01 6.92144156e-01 -3.77045959e-01 -7.10094988e-01
2.95301199e-01 2.62349173e-02 4.12098020e-01 -6.49952054e-01
-7.79980600e-01 2.88791716e-01 -1.84652314e-01 -6.18395686e-01
-1.22042522e-01 8.73033851e-02 -1.63956392e+00 -2.69106328e-01
-4.97255027e-01 4.26012099e-01 3.44745636e-01 8.20424497e-01
7.60532200e-01 7.80314445e-01 7.77975202e-01 -1.48845404e-01
-2.45352551e-01 -9.42941010e-01 -6.17024481e-01 2.72957146e-01
3.03748786e-01 -3.54633600e-01 2.82918364e-02 -2.83011883e-01] | [5.149643898010254, 5.903248310089111] |
389bd6e5-cdfc-4fa2-ab96-8b868be9d67b | bert4eth-a-pre-trained-transformer-for | 2303.18138 | null | https://arxiv.org/abs/2303.18138v1 | https://arxiv.org/pdf/2303.18138v1.pdf | BERT4ETH: A Pre-trained Transformer for Ethereum Fraud Detection | As various forms of fraud proliferate on Ethereum, it is imperative to safeguard against these malicious activities to protect susceptible users from being victimized. While current studies solely rely on graph-based fraud detection approaches, it is argued that they may not be well-suited for dealing with highly repetitive, skew-distributed and heterogeneous Ethereum transactions. To address these challenges, we propose BERT4ETH, a universal pre-trained Transformer encoder that serves as an account representation extractor for detecting various fraud behaviors on Ethereum. BERT4ETH features the superior modeling capability of Transformer to capture the dynamic sequential patterns inherent in Ethereum transactions, and addresses the challenges of pre-training a BERT model for Ethereum with three practical and effective strategies, namely repetitiveness reduction, skew alleviation and heterogeneity modeling. Our empirical evaluation demonstrates that BERT4ETH outperforms state-of-the-art methods with significant enhancements in terms of the phishing account detection and de-anonymization tasks. The code for BERT4ETH is available at: https://github.com/git-disl/BERT4ETH. | ['Ling Liu', 'Bingsheng He', 'Shengliang Lu', 'Bingqiao Luo', 'Zhen Zhang', 'Sihao Hu'] | 2023-03-29 | null | null | null | null | ['fraud-detection'] | ['miscellaneous'] | [-9.16881338e-02 -3.19944024e-02 -3.33733171e-01 -2.87284464e-01
-3.95284712e-01 -8.76370966e-01 5.31696439e-01 2.76597887e-01
-1.82505220e-01 5.79838634e-01 1.84391811e-01 -8.29856277e-01
5.52347414e-02 -8.77133965e-01 -4.31587994e-01 -1.18615776e-01
-3.16863954e-01 6.65023744e-01 -7.26154000e-02 -2.81992942e-01
4.47321922e-01 6.08539164e-01 -5.21962643e-01 2.43117958e-01
8.18445981e-01 8.18731785e-01 -9.26775694e-01 4.48080420e-01
3.76073390e-01 1.08757126e+00 -7.47024715e-01 -1.46771538e+00
6.78808093e-01 -3.24597359e-01 -6.53570235e-01 -3.54831189e-01
-6.42595738e-02 -9.24180567e-01 -8.83920431e-01 1.17365003e+00
3.24863166e-01 -5.17549157e-01 6.95802927e-01 -1.82125068e+00
-5.88998616e-01 8.82155359e-01 -8.63353074e-01 4.52291250e-01
2.04338998e-01 3.12463313e-01 1.35049272e+00 -4.79924500e-01
7.55044222e-01 9.04640675e-01 8.92232716e-01 5.19047201e-01
-1.04060531e+00 -1.11302996e+00 -3.48557919e-01 1.63229108e-01
-1.06602001e+00 -5.84002852e-01 7.48825192e-01 -1.88114509e-01
1.00405967e+00 1.98187038e-01 6.05428338e-01 1.30656695e+00
4.62629527e-01 1.01790333e+00 7.00406432e-01 6.00975007e-02
-3.57403457e-02 1.17876202e-01 1.37502894e-01 6.57351017e-01
1.13744628e+00 2.21458718e-01 -4.87425387e-01 -1.06689715e+00
5.53322911e-01 2.60777861e-01 -3.13557714e-01 -4.58163947e-01
-8.66496384e-01 1.28159583e+00 3.70620281e-01 1.82536215e-01
-6.11748517e-01 4.59045947e-01 1.19716370e+00 7.56920874e-01
3.39156121e-01 5.65775335e-01 -5.31912386e-01 -4.72809762e-01
-9.63980138e-01 3.93341452e-01 1.31908524e+00 1.17651880e+00
4.12031442e-01 -3.50308046e-02 2.34618455e-01 1.04837827e-01
4.75800067e-01 6.66320100e-02 1.97710797e-01 -9.26061511e-01
6.39527500e-01 8.30785155e-01 -1.07326373e-01 -9.31547880e-01
2.93322932e-02 -2.24946618e-01 -8.25122237e-01 -8.95543247e-02
4.62392598e-01 -1.86429083e-01 -3.29159915e-01 1.37934446e+00
3.76195125e-02 -5.33251762e-02 -1.30706847e-01 3.82861763e-01
-1.41454497e-02 1.59510106e-01 -2.30473638e-01 1.32245598e-02
1.44155741e+00 -5.21510482e-01 -6.35330617e-01 -3.01603395e-02
9.62078512e-01 -5.39109588e-01 3.73637706e-01 2.24762574e-01
-1.01375723e+00 6.45400882e-01 -9.20483291e-01 1.17996201e-01
-4.29892838e-01 -6.28549218e-01 1.01391828e+00 1.09942269e+00
-6.99851811e-01 6.77904904e-01 -6.75068438e-01 -1.10420257e-01
1.10016644e+00 4.51154977e-01 -3.11553687e-01 1.00908905e-01
-1.53577507e+00 7.64666677e-01 3.17820311e-01 5.97223965e-03
-8.15725923e-01 -5.23808241e-01 -7.14688480e-01 1.89069584e-01
3.36330473e-01 -6.48585737e-01 1.37565303e+00 -7.01822758e-01
-9.27272975e-01 8.93270135e-01 2.07032934e-01 -1.13051343e+00
1.19074559e+00 2.53036559e-01 -4.44645315e-01 8.78607631e-02
1.97589993e-01 -1.32194906e-01 6.75025165e-01 -1.13972628e+00
-4.62436944e-01 -4.90604192e-01 1.35869235e-01 -1.19581640e-01
-2.11877197e-01 2.59047806e-01 4.03260887e-02 -8.30962360e-01
-4.34719592e-01 -8.95051956e-01 2.39464901e-02 -2.62238115e-01
-6.19287431e-01 -8.49801376e-02 1.11775839e+00 -7.23117113e-01
1.39057684e+00 -1.74253321e+00 -5.12490511e-01 6.07300937e-01
5.18045485e-01 6.18738651e-01 2.89610028e-01 8.75554800e-01
1.33438289e-01 4.20938820e-01 -3.91954780e-01 -3.79178256e-01
4.44073975e-01 -1.39718384e-01 -4.44430172e-01 7.60218441e-01
-1.02695301e-01 1.25214803e+00 -9.13732052e-01 -2.91317761e-01
1.58389911e-01 2.43819699e-01 -6.54103994e-01 4.91790473e-02
-8.51178691e-02 2.02987775e-01 -4.05493170e-01 1.24368870e+00
9.25404012e-01 -4.69729513e-01 6.26703739e-01 1.48308426e-01
6.15856707e-01 4.48565692e-01 -7.73349881e-01 1.16864681e+00
-8.07830691e-02 5.00916302e-01 2.76953369e-01 -7.68217444e-01
6.19624078e-01 5.17953575e-01 7.25795031e-01 -6.59473836e-01
6.26077771e-01 8.11852992e-01 -2.62052026e-02 -1.64095491e-01
4.99982387e-01 -1.13198608e-01 -2.58696795e-01 1.11907113e+00
-2.28473172e-01 3.32279950e-01 1.95736066e-01 4.42658067e-01
1.44718575e+00 -4.78718460e-01 7.21406460e-01 -1.64557964e-01
3.09823722e-01 -2.78765440e-01 7.89343238e-01 7.13089406e-01
-7.30414629e-01 1.08621456e-01 1.05369592e+00 -5.18657327e-01
-1.00921321e+00 -8.89625132e-01 -8.73098373e-02 4.85746354e-01
-2.75944956e-02 -5.76087177e-01 -7.98686862e-01 -1.43512809e+00
8.69166017e-01 8.04007828e-01 -5.16061425e-01 -3.34056407e-01
-6.72482848e-01 -9.95420277e-01 1.14332926e+00 3.33174020e-01
7.39224851e-01 -9.46382105e-01 -5.27561426e-01 3.74679029e-01
-1.68350846e-01 -1.00049090e+00 -7.56073236e-01 -1.77953884e-01
-7.05160916e-01 -1.76399493e+00 -3.71196330e-01 -1.93121314e-01
5.00905812e-01 2.06703201e-01 1.08261716e+00 2.80695558e-01
-3.13157707e-01 5.25330082e-02 -7.26745874e-02 -5.42794347e-01
-6.99915528e-01 1.38183534e-01 -1.51944831e-01 2.34396905e-01
1.16886997e+00 -6.79811120e-01 -8.70718598e-01 2.68566340e-01
-9.51092124e-01 -5.77351034e-01 2.96344310e-01 6.90622568e-01
-2.95355231e-01 -2.84813270e-02 8.53651762e-01 -1.52704537e+00
1.06689322e+00 -7.48176396e-01 -5.02477825e-01 5.27646951e-02
-1.27847314e+00 -3.56267869e-01 6.88730299e-01 5.95131330e-02
-5.42225957e-01 -6.62053227e-01 1.21315971e-01 -5.87110400e-01
4.02285904e-01 2.33522296e-01 -9.55506973e-03 -2.14516595e-01
4.27397639e-01 4.08693030e-02 1.53457075e-01 -1.97021112e-01
4.54687774e-02 1.03985012e+00 7.86779374e-02 -2.32781634e-01
7.58972883e-01 6.13259912e-01 -2.45730773e-01 -1.74662888e-01
-1.49747089e-01 -5.11056364e-01 -6.81241378e-02 2.21497729e-01
1.42199725e-01 -8.17424417e-01 -1.27469528e+00 9.29925978e-01
-1.14673543e+00 -1.52489869e-02 -1.74728826e-01 -1.45366602e-02
-4.65736836e-01 8.31534088e-01 -1.27096474e+00 -7.40045965e-01
-8.06090653e-01 -9.72429752e-01 6.67459130e-01 -3.27232569e-01
-6.50437832e-01 -1.14833462e+00 1.90791339e-02 7.37415433e-01
5.08590341e-01 4.40387666e-01 8.99906754e-01 -1.20269239e+00
-8.33285630e-01 -8.39067876e-01 -3.76788169e-01 3.31391275e-01
1.86906680e-01 -2.28781804e-01 -7.29767263e-01 -8.31457496e-01
1.22802012e-01 -1.64990440e-01 6.67791009e-01 -1.12839043e-01
8.74293566e-01 -9.51726735e-01 -2.42749795e-01 7.85437763e-01
1.37234247e+00 3.24891329e-01 7.25618243e-01 5.55475295e-01
6.79882169e-01 3.96621197e-01 4.87987027e-02 8.10832083e-01
7.29825377e-01 3.75286341e-01 7.07494557e-01 2.36631572e-01
5.38522482e-01 -5.27662754e-01 1.66397527e-01 3.43884081e-01
4.43746969e-02 -4.56909627e-01 -8.47033918e-01 5.35291791e-01
-1.91321349e+00 -1.11774790e+00 -1.02370260e-02 2.06849813e+00
5.69255173e-01 1.30636305e-01 4.30934250e-01 2.33055562e-01
6.67156816e-01 3.69246364e-01 -7.33097196e-01 -7.16631353e-01
1.17391281e-01 -1.83195055e-01 1.19812298e+00 2.54467607e-01
-8.73253942e-01 6.46090984e-01 5.84745073e+00 3.42104852e-01
-6.95547163e-01 1.78031132e-01 5.72982728e-01 -8.87420774e-02
-5.41978180e-01 2.24668041e-01 -4.45849329e-01 1.01823950e+00
1.03909028e+00 -5.80109954e-01 7.58095682e-01 5.77044427e-01
2.18832139e-02 4.20284241e-01 -1.13341975e+00 9.78128791e-01
-1.49108395e-01 -1.41557097e+00 1.20044284e-01 6.47075891e-01
5.51616728e-01 1.90681130e-01 -1.70830607e-01 1.17321998e-01
7.23633826e-01 -9.71059799e-01 5.25715113e-01 1.39498323e-01
6.33246779e-01 -9.67629135e-01 1.10626590e+00 -7.61273317e-03
-8.37436438e-01 -3.63101184e-01 -1.28501691e-02 4.77586716e-01
2.33924001e-01 5.18202364e-01 -9.90138948e-01 7.18630850e-01
4.06198442e-01 9.77416575e-01 -2.65030950e-01 6.87369287e-01
-1.07423916e-01 5.86313605e-01 -1.76078975e-01 2.54899651e-01
3.11885267e-01 -3.47212434e-01 6.35295272e-01 1.28166974e+00
2.94201076e-01 -3.89449894e-01 -3.86605889e-01 9.88667846e-01
-8.25718999e-01 -2.57210404e-01 -9.88839388e-01 -4.95852202e-01
8.26990128e-01 8.38089168e-01 -3.55226457e-01 -3.51301134e-01
-5.47497988e-01 1.08184588e+00 -1.50911917e-03 3.32969218e-01
-7.28850067e-01 -5.53228736e-01 7.52576649e-01 4.04862314e-01
1.54124871e-01 1.41768113e-01 -5.35620332e-01 -1.72595000e+00
1.14222124e-01 -1.48886323e+00 8.21353793e-01 -1.45438075e-01
-1.44112849e+00 2.83402681e-01 -4.45859015e-01 -1.26647079e+00
-2.85093546e-01 -3.41552734e-01 -8.41282427e-01 7.21228778e-01
-1.55810332e+00 -1.17030966e+00 2.15498745e-01 8.12361896e-01
-1.43842679e-02 -4.80621338e-01 7.46398270e-01 3.74504745e-01
-5.17533422e-01 7.60934770e-01 2.84759641e-01 5.24707377e-01
4.92602557e-01 -1.26759887e+00 9.38841879e-01 9.09311354e-01
-7.94854686e-02 8.93515706e-01 5.35896003e-01 -8.38524461e-01
-1.49948668e+00 -8.90561402e-01 1.25784290e+00 -4.64189231e-01
9.95013833e-01 -5.99878132e-01 -8.61600041e-01 1.28563869e+00
1.63864896e-01 -1.50698453e-01 6.57540560e-01 -9.91606116e-02
-7.48736441e-01 -8.45205188e-02 -1.57552624e+00 4.46461707e-01
9.74828541e-01 -9.23998058e-01 -4.36495036e-01 2.58973539e-01
3.27363908e-02 -1.08227313e-01 -9.18510854e-01 1.11824699e-01
4.47229445e-01 -1.32720816e+00 7.23650217e-01 -5.05832672e-01
1.53304905e-01 9.63096097e-02 1.21856786e-01 -1.06149948e+00
-6.27676845e-02 -1.36263454e+00 -8.72914910e-01 1.02377677e+00
1.71780020e-01 -1.30884659e+00 1.15531278e+00 4.45062310e-01
6.69689238e-01 -6.66082919e-01 -1.04425788e+00 -6.75967097e-01
1.30599692e-01 -3.85951474e-02 1.08530521e+00 1.39916599e+00
3.79426837e-01 -9.51824412e-02 -5.22274315e-01 -1.31407276e-01
1.02729523e+00 1.69569612e-01 7.40642011e-01 -1.19968653e+00
-3.70378271e-02 -5.49728632e-01 -5.31077802e-01 -5.17973840e-01
9.21607316e-02 -1.06092918e+00 -8.44362140e-01 -1.34007132e+00
2.35765815e-01 -1.72145486e-01 -4.40534800e-01 4.44047034e-01
2.27642342e-01 6.14894368e-02 2.00598940e-01 6.46186233e-01
-3.76584023e-01 4.02237058e-01 7.81160831e-01 -2.73893744e-01
1.51005015e-01 3.95984322e-01 -1.15153480e+00 3.61168236e-01
9.28661466e-01 -6.62953079e-01 -4.00900334e-01 -9.90813673e-02
3.14871788e-01 1.64001957e-01 4.45473969e-01 -3.94915819e-01
1.46877572e-01 1.70477256e-01 -1.55599236e-01 -2.97072142e-01
-1.45596698e-01 -8.93181503e-01 1.77280486e-01 7.27024317e-01
5.78038543e-02 4.65783238e-01 -1.97529256e-01 7.76170969e-01
-1.86711773e-01 -1.69332966e-01 6.99450314e-01 -4.75580812e-01
-9.01390687e-02 7.73974657e-01 -3.20321649e-01 3.96212012e-01
1.12644064e+00 -2.73350447e-01 -5.11910379e-01 -6.91981196e-01
-9.98795629e-02 5.19615352e-01 9.59786177e-01 3.29704493e-01
3.97013336e-01 -1.08676040e+00 -7.61676371e-01 3.72453928e-01
1.64017141e-01 -2.32929200e-01 -6.54633939e-02 8.84414136e-01
-7.81123459e-01 4.91509795e-01 -2.38307685e-01 1.79854974e-01
-1.05553651e+00 4.61237937e-01 6.77870631e-01 -7.63040185e-01
-6.31015003e-01 4.03176606e-01 -2.84363568e-01 -4.48342919e-01
1.51199996e-01 8.72255191e-02 3.64532530e-01 1.78881928e-01
2.45338634e-01 7.08930194e-01 2.31497169e-01 -5.11705935e-01
-5.35734355e-01 -1.15948625e-01 -5.88687658e-01 2.35571295e-01
1.27945161e+00 -2.51264162e-02 -3.36218089e-01 -2.34856024e-01
1.44664752e+00 1.51575193e-01 -1.03715992e+00 -1.84075251e-01
4.67754126e-01 -8.33374381e-01 -2.44453996e-01 -6.45107806e-01
-1.38132298e+00 6.09956682e-01 -2.09508896e-01 6.06569111e-01
7.68270671e-01 -5.75399637e-01 1.38045073e+00 2.82143503e-01
7.69445658e-01 -8.56175661e-01 -3.43207091e-01 3.85062814e-01
4.75281894e-01 -1.04895520e+00 -1.16257168e-01 -1.92043692e-01
-8.12635422e-01 8.74832273e-01 2.00253293e-01 -2.79613376e-01
7.01519787e-01 2.56805122e-01 3.59537639e-02 -5.31677783e-01
-6.11601353e-01 7.12073088e-01 -5.10509670e-01 1.83393463e-01
7.02714846e-02 2.71630943e-01 -3.90680224e-01 2.19527841e-01
-1.84416935e-01 2.51170546e-01 8.95233929e-01 1.18092775e+00
4.37515795e-01 -1.35096741e+00 1.30772009e-01 8.15410137e-01
-1.03921556e+00 -1.59923181e-01 -5.63135147e-01 8.84768188e-01
-3.64627898e-01 6.54202819e-01 -1.97203487e-01 -3.83750707e-01
3.38909090e-01 5.65273240e-02 1.68872885e-02 -2.23072201e-01
-1.05782282e+00 -4.04458225e-01 4.07020122e-01 -6.13157034e-01
1.07291073e-01 -8.36533666e-01 -8.37172568e-01 -1.37609065e+00
-3.74648452e-01 3.00130904e-01 2.43688703e-01 3.62155080e-01
6.29874825e-01 1.11887425e-01 8.94381166e-01 -5.53990379e-02
-1.12371480e+00 -7.54453719e-01 -1.05543804e+00 7.70290136e-01
4.71892416e-01 -2.85223663e-01 -7.81291306e-01 -4.95089889e-01] | [6.748358726501465, 6.551349639892578] |
7e80f694-dac6-45fc-bdc9-860f564bd269 | adversary-aware-continual-learning | 2304.14483 | null | https://arxiv.org/abs/2304.14483v1 | https://arxiv.org/pdf/2304.14483v1.pdf | Adversary Aware Continual Learning | Class incremental learning approaches are useful as they help the model to learn new information (classes) sequentially, while also retaining the previously acquired information (classes). However, it has been shown that such approaches are extremely vulnerable to the adversarial backdoor attacks, where an intelligent adversary can introduce small amount of misinformation to the model in the form of imperceptible backdoor pattern during training to cause deliberate forgetting of a specific task or class at test time. In this work, we propose a novel defensive framework to counter such an insidious attack where, we use the attacker's primary strength-hiding the backdoor pattern by making it imperceptible to humans-against it, and propose to learn a perceptible (stronger) pattern (also during the training) that can overpower the attacker's imperceptible (weaker) pattern. We demonstrate the effectiveness of the proposed defensive mechanism through various commonly used Replay-based (both generative and exact replay-based) class incremental learning algorithms using continual learning benchmark variants of CIFAR-10, CIFAR-100, and MNIST datasets. Most noteworthy, our proposed defensive framework does not assume that the attacker's target task and target class is known to the defender. The defender is also unaware of the shape, size, and location of the attacker's pattern. We show that our proposed defensive framework considerably improves the performance of class incremental learning algorithms with no knowledge of the attacker's target task, attacker's target class, and attacker's imperceptible pattern. We term our defensive framework as Adversary Aware Continual Learning (AACL). | ['Robi Polikar', 'Muhammad Umer'] | 2023-04-27 | null | null | null | null | ['class-incremental-learning', 'incremental-learning', 'misinformation'] | ['computer-vision', 'methodology', 'miscellaneous'] | [ 5.06304920e-01 1.35746866e-01 -8.85466039e-02 7.06259459e-02
-4.54741955e-01 -1.44243383e+00 7.62943923e-01 3.62581521e-01
-6.05081856e-01 8.17163348e-01 -4.41460967e-01 -7.57572651e-01
-2.03684375e-01 -9.89313126e-01 -1.36713016e+00 -8.88695061e-01
-4.19632941e-01 1.48982301e-01 4.77617174e-01 -2.98314691e-01
4.16736603e-01 6.21220529e-01 -1.24709558e+00 4.26298290e-01
4.85030115e-01 7.56935298e-01 -3.88756454e-01 8.94553721e-01
3.46053243e-01 9.63343680e-01 -1.11987150e+00 -4.63596046e-01
7.24155724e-01 -4.32232797e-01 -7.07216978e-01 -3.10383171e-01
4.08366978e-01 -4.93158728e-01 -3.55275303e-01 1.29491711e+00
1.38420060e-01 -6.99886829e-02 4.24219161e-01 -1.47389221e+00
-4.42665875e-01 8.43700349e-01 -4.39253002e-01 3.36683661e-01
2.06682965e-01 3.26575190e-01 4.26282704e-01 -2.84301013e-01
3.86239737e-01 9.60411727e-01 3.70912224e-01 7.19276607e-01
-1.07276857e+00 -1.17321014e+00 3.94066751e-01 5.66279292e-02
-1.24119675e+00 -2.26888970e-01 1.19194353e+00 -2.61441201e-01
3.30752909e-01 6.41910195e-01 2.63671309e-01 1.59538221e+00
5.08577824e-01 6.25843227e-01 1.35033166e+00 -5.22237837e-01
6.34100378e-01 5.64531982e-01 4.43294168e-01 6.63421094e-01
4.00713891e-01 7.85110235e-01 -2.49104559e-01 -8.69147956e-01
8.18213299e-02 1.41420200e-01 -3.76308352e-01 -4.15060341e-01
-6.92539394e-01 7.66858280e-01 6.22208655e-01 9.14201885e-02
-1.94194511e-01 4.99254605e-03 3.68568152e-01 7.90421426e-01
1.21761893e-03 3.50401342e-01 -6.09921992e-01 2.65746057e-01
-3.00798178e-01 2.32604802e-01 8.58943403e-01 4.87966061e-01
5.32520533e-01 3.63742948e-01 2.09082812e-01 6.57069683e-02
-1.63601205e-01 5.31344891e-01 7.40007102e-01 -1.15037866e-01
5.05221903e-01 3.00071239e-01 1.37047112e-01 -1.04465234e+00
-5.67298234e-02 -9.46344912e-01 -5.21194756e-01 6.24951661e-01
3.30212176e-01 -3.17457795e-01 -7.02901065e-01 2.13196683e+00
5.63106179e-01 5.10738909e-01 4.29022044e-01 3.49928707e-01
6.79346323e-02 5.10609031e-01 -1.47306263e-01 -2.80913532e-01
1.09184456e+00 -6.36769891e-01 -1.99025020e-01 -2.39330634e-01
4.51883614e-01 -3.16205651e-01 1.03389561e+00 6.13837123e-01
-4.07313555e-01 -3.99127811e-01 -1.51716197e+00 8.95208359e-01
-5.71373403e-01 -5.54774046e-01 5.11542439e-01 1.37502313e+00
-3.34578037e-01 4.37491894e-01 -7.10946620e-01 1.64547920e-01
5.01846850e-01 4.62959230e-01 -4.14095670e-01 7.45971315e-03
-1.33311081e+00 3.98116231e-01 6.39968753e-01 -1.94588497e-01
-1.80677176e+00 -5.97103298e-01 -3.52403164e-01 -3.43555473e-02
6.15840495e-01 -4.53537315e-01 8.96827221e-01 -1.29070008e+00
-1.24571359e+00 3.63876104e-01 5.01710176e-01 -1.17559922e+00
8.21852088e-01 -1.68664992e-01 -3.05838704e-01 9.92493108e-02
-4.50868636e-01 6.99281916e-02 1.39686191e+00 -1.25572360e+00
-4.10937726e-01 -5.66759706e-01 4.65594590e-01 -7.01897964e-02
-6.71430349e-01 -4.51377332e-01 8.43836367e-01 -7.89221108e-01
-2.00770944e-01 -1.08024979e+00 1.95526406e-01 -1.98613614e-01
-7.70376444e-01 2.28605554e-01 1.43381345e+00 -3.70193683e-02
8.63267899e-01 -2.23637462e+00 -3.76441747e-01 1.99038967e-01
1.08117715e-01 7.55174756e-01 -8.20603967e-02 5.05960107e-01
-3.38782817e-01 1.55854523e-01 -3.37804168e-01 8.59797224e-02
-4.21370983e-01 2.27007627e-01 -1.13396358e+00 6.42442405e-01
-4.65240568e-01 6.14439368e-01 -7.98360169e-01 3.30714762e-01
2.17744559e-02 3.28197151e-01 -5.66362143e-01 1.99077174e-01
-1.53473243e-01 2.60586619e-01 -5.49488842e-01 3.45890313e-01
7.85875380e-01 3.85109812e-01 -1.18290791e-02 2.41847262e-01
1.97597861e-01 1.23544503e-02 -1.16926980e+00 8.22525620e-01
-2.00503498e-01 1.78100884e-01 -3.94651175e-01 -9.67847824e-01
7.39967644e-01 2.93565243e-01 -1.93527505e-01 -1.39454439e-01
7.35338181e-02 5.38775027e-02 2.71981388e-01 4.64232713e-02
-9.73281637e-02 -1.14239305e-01 -8.69770199e-02 7.62729347e-01
-8.35329220e-02 5.96367300e-01 -6.69956863e-01 2.63074338e-01
1.28165948e+00 -3.57999295e-01 4.93943244e-01 -1.56447396e-01
8.35057497e-01 -1.99122146e-01 4.69900399e-01 1.43859398e+00
-1.85427219e-01 -1.53015509e-01 4.81745243e-01 -8.01540375e-01
-5.19140184e-01 -1.36916280e+00 8.12805295e-02 9.92964625e-01
-2.39008293e-02 -4.29161265e-02 -7.06582010e-01 -1.38748515e+00
1.75326452e-01 9.95885968e-01 -1.00520909e+00 -9.51072335e-01
-4.51835096e-01 -6.10463917e-01 1.12193382e+00 1.18815735e-01
8.25577199e-01 -7.49519587e-01 -6.34831369e-01 -2.25887708e-02
2.40620732e-01 -6.57772064e-01 -3.22006404e-01 2.97886312e-01
-8.84427965e-01 -1.36558199e+00 -1.17564619e-01 -3.54956806e-01
8.12080145e-01 2.71569997e-01 2.95462281e-01 1.48899227e-01
-1.35780424e-01 4.39002037e-01 -1.92984417e-01 -7.19850779e-01
-8.04125130e-01 -1.63373739e-01 4.22404915e-01 3.99613172e-01
-7.33732805e-02 -5.71261048e-01 -4.27835166e-01 3.10083002e-01
-1.15517497e+00 -5.73151350e-01 2.79655308e-01 7.33947456e-01
1.45900130e-01 6.72401249e-01 6.28301024e-01 -1.52373934e+00
6.86416090e-01 -5.36819518e-01 -6.83145106e-01 4.10050094e-01
-6.65063977e-01 5.67447059e-02 1.17341816e+00 -1.20564532e+00
-8.31429243e-01 -2.68465746e-02 2.35297568e-02 -4.75654960e-01
-2.25618124e-01 1.73089147e-01 -2.76913822e-01 -7.57906258e-01
1.18095624e+00 7.38670945e-01 -2.87094176e-01 -4.10742253e-01
1.97687417e-01 3.53005737e-01 5.95477998e-01 -6.82893395e-01
1.54801750e+00 5.59043109e-01 2.26583749e-01 -4.97912467e-01
-5.93668938e-01 2.11601913e-01 -2.81630248e-01 8.01658630e-03
1.76080704e-01 -6.44182503e-01 -8.14097583e-01 9.12970662e-01
-8.29331815e-01 -5.49130030e-02 -1.06931388e-01 1.76842541e-01
-1.43035352e-01 4.13511425e-01 -3.43169093e-01 -9.32230949e-01
-3.80875319e-01 -9.55368996e-01 1.65790208e-02 6.10763133e-02
3.09662133e-01 -9.43361402e-01 5.62748648e-02 2.10867241e-01
5.26409805e-01 6.26618385e-01 1.24023402e+00 -1.59210467e+00
-6.10767126e-01 -7.67395139e-01 5.61292589e-01 5.75822473e-01
2.86585391e-01 -4.21785593e-01 -1.20844817e+00 -7.70940781e-01
6.85645163e-01 -4.81129438e-01 6.58789635e-01 -2.71814972e-01
1.03976989e+00 -1.09321332e+00 -9.05507803e-02 5.31008661e-01
1.27112043e+00 5.75571537e-01 4.08609688e-01 5.93274534e-01
4.37491268e-01 1.70572206e-01 2.80187190e-01 3.06879908e-01
-1.32755414e-01 3.44863176e-01 6.72323048e-01 2.76349276e-01
3.30662847e-01 -5.93813241e-01 6.44423485e-01 -2.75756717e-02
2.77149230e-01 -2.16131330e-01 -5.44938505e-01 -2.64859330e-02
-1.28746617e+00 -9.91823435e-01 4.72692043e-01 2.86460614e+00
8.92331600e-01 7.03301132e-01 -3.21952328e-02 4.90831375e-01
5.70995152e-01 1.10400073e-01 -7.15875208e-01 -5.60225487e-01
-4.19286378e-02 1.36860162e-01 6.40231609e-01 7.49732733e-01
-1.12462902e+00 8.19387674e-01 4.86719847e+00 9.48769033e-01
-1.30909014e+00 1.08621918e-01 6.75099790e-01 -4.07598391e-02
-1.71045467e-01 3.26113135e-01 -7.97296524e-01 3.14643145e-01
9.39529538e-01 -4.72554505e-01 6.17871404e-01 1.17618322e+00
-3.90615404e-01 1.09944768e-01 -1.25612152e+00 3.75706971e-01
2.58286208e-01 -1.19822371e+00 3.12355071e-01 6.13668598e-02
4.06752020e-01 -4.33737844e-01 5.34631610e-01 6.02097094e-01
3.78227443e-01 -5.90556443e-01 6.50916278e-01 1.05662175e-01
3.65812570e-01 -1.20137250e+00 4.94000524e-01 9.58933890e-01
-8.71489346e-01 -7.23055243e-01 -2.82097548e-01 -9.68997106e-02
-5.43147564e-01 1.21201806e-01 -9.96325076e-01 3.63081187e-01
1.60541996e-01 -2.89425880e-01 -9.11527693e-01 6.60244942e-01
-3.78922075e-01 1.05795944e+00 -2.36540034e-01 2.07970321e-01
3.48015219e-01 4.28060889e-01 1.07573485e+00 6.34028733e-01
-1.82949498e-01 3.81960757e-02 3.48915964e-01 5.19004941e-01
-1.65067911e-01 -1.43942744e-01 -1.01754582e+00 1.36589091e-02
7.44150758e-01 7.24284708e-01 -6.64769292e-01 -1.25353113e-01
9.44622681e-02 1.00579143e+00 1.08746469e-01 3.95559162e-01
-7.93650329e-01 -5.49903095e-01 4.08259600e-01 -1.67555474e-02
-3.46956216e-02 -6.67083412e-02 5.43927886e-02 -8.80004883e-01
-1.34787649e-01 -1.33939767e+00 8.52128625e-01 -2.57218510e-01
-1.05935395e+00 7.82637119e-01 -1.00453226e-02 -9.97544706e-01
-2.49266088e-01 -2.62481362e-01 -8.00010622e-01 5.29049516e-01
-1.08331740e+00 -1.26292670e+00 3.47578228e-01 1.10406470e+00
3.72855753e-01 -6.94009542e-01 8.78862262e-01 -2.40593746e-01
-3.35331738e-01 1.26704872e+00 -8.84545743e-02 2.85696417e-01
5.03241658e-01 -8.30747187e-01 1.76008180e-01 1.13780785e+00
4.63279366e-01 1.08276272e+00 7.96166718e-01 -7.83872306e-01
-1.48404777e+00 -1.08881927e+00 2.96338081e-01 -4.78041619e-01
7.48711109e-01 -6.38956785e-01 -9.75302637e-01 9.43799973e-01
-2.36547947e-01 1.03755936e-01 6.37716293e-01 -1.72689229e-01
-9.50982749e-01 -3.34866613e-01 -1.66057324e+00 5.42404473e-01
4.63207394e-01 -6.97773218e-01 -7.15105593e-01 3.68181914e-01
1.05659568e+00 -2.54906833e-01 -2.81006899e-02 3.40903103e-01
5.69149971e-01 -8.19936633e-01 1.09680605e+00 -9.34712946e-01
-2.46938780e-01 -1.36386946e-01 -2.60747552e-01 -1.01068771e+00
6.58678189e-02 -8.02352786e-01 -2.94354856e-01 9.85942781e-01
1.70806706e-01 -1.30393147e+00 7.24273980e-01 1.66807458e-01
2.77818859e-01 -4.97606784e-01 -1.22216678e+00 -1.06443536e+00
3.40443462e-01 -2.67931521e-01 5.66351652e-01 1.20406830e+00
-3.89025271e-01 -7.05131218e-02 -7.02860057e-01 7.79763997e-01
9.60146070e-01 -6.14334866e-02 9.96923029e-01 -9.97305751e-01
-6.12909913e-01 1.13653272e-01 -4.06426787e-01 -4.57984895e-01
1.47497237e-01 -8.18523169e-01 -4.91840452e-01 -6.05376177e-02
6.65512532e-02 -3.59953225e-01 -8.71958554e-01 7.28390098e-01
-1.18918002e-01 3.11630853e-02 3.66086364e-01 2.52225131e-01
-1.50872916e-01 2.41952911e-01 7.43881166e-01 -2.17490345e-01
-2.88429946e-01 6.43679738e-01 -8.59572768e-01 6.50395989e-01
7.60725498e-01 -1.17873025e+00 -9.70830023e-01 5.35845309e-02
4.93365005e-02 -2.05762796e-02 7.05985785e-01 -1.18148470e+00
4.45886165e-01 -1.80213466e-01 2.97903985e-01 -2.14220464e-01
-5.05886488e-02 -8.87137651e-01 5.22545278e-02 1.24341857e+00
-6.69855952e-01 7.34286476e-03 4.31946158e-01 1.09176719e+00
3.40795636e-01 -2.83212543e-01 8.75125587e-01 -1.40000120e-01
-1.96625635e-01 1.84267744e-01 -3.57275039e-01 -1.73529148e-01
1.29978502e+00 -1.38675142e-02 -7.54471779e-01 -2.38420486e-01
-6.06674731e-01 -1.98613957e-01 2.43977606e-01 3.79834652e-01
7.32993841e-01 -8.28171849e-01 -4.70621765e-01 5.40413141e-01
-1.00019611e-02 -6.48451805e-01 3.29508990e-01 2.11932108e-01
-1.49788484e-01 1.94609269e-01 -3.11959922e-01 -2.33172461e-01
-1.44220006e+00 1.21134853e+00 4.42186475e-01 -1.27103552e-01
-6.09664381e-01 8.87380958e-01 3.92069995e-01 -3.47385734e-01
2.92811960e-01 2.69298702e-01 7.89879113e-02 -2.88102388e-01
7.71820724e-01 2.62476861e-01 4.14338522e-02 -2.03443259e-01
-3.78708303e-01 5.32609411e-02 -6.32344067e-01 9.13894270e-03
8.78974617e-01 1.60248563e-01 1.25207692e-01 4.28429782e-01
1.03173554e+00 3.63708764e-01 -1.00256884e+00 -4.17690188e-01
-1.72661155e-01 -6.66155398e-01 -2.08824366e-01 -1.07955766e+00
-6.89172149e-01 7.18304455e-01 1.00396645e+00 1.53101653e-01
1.02560496e+00 -5.63650191e-01 6.30614519e-01 7.73392737e-01
7.66497970e-01 -5.13031185e-01 2.48219475e-01 1.61553845e-01
6.06331110e-01 -7.73197830e-01 2.10865736e-02 -7.34914094e-02
-4.73697782e-01 9.95905638e-01 5.19488394e-01 -2.76885927e-01
7.94277549e-01 2.47562423e-01 2.47267988e-02 1.69152483e-01
-9.73353684e-01 6.18283808e-01 1.17846064e-01 8.09121549e-01
-5.36601424e-01 1.02174751e-01 1.48059167e-02 7.56467998e-01
-2.70021260e-01 -3.34374607e-01 7.40095079e-01 1.18334317e+00
-5.95047057e-01 -1.06016111e+00 -6.59659028e-01 -4.35754545e-02
-5.80473721e-01 -2.79214792e-02 -5.98958790e-01 8.59471560e-01
2.22867385e-01 6.85635805e-01 -4.01638657e-01 -6.98261499e-01
9.00743937e-04 3.32337677e-01 2.36075863e-01 -4.60752577e-01
-7.47127771e-01 -4.14638907e-01 -3.28768611e-01 -3.13557625e-01
3.98745418e-01 -1.89052507e-01 -8.52757156e-01 -3.03248018e-01
-2.05344737e-01 3.97390455e-01 5.41278601e-01 9.71896827e-01
3.30182351e-02 1.60928980e-01 1.41522539e+00 -1.16454542e-01
-1.30261338e+00 -5.31037390e-01 -4.32658941e-01 2.53840029e-01
6.02616489e-01 -6.66696012e-01 -8.77390563e-01 -3.00307691e-01] | [5.7422099113464355, 7.525333404541016] |
ee5ea0ba-5778-441a-9889-f702ce85cd20 | design-based-conformal-prediction | 2303.01422 | null | https://arxiv.org/abs/2303.01422v1 | https://arxiv.org/pdf/2303.01422v1.pdf | Design-based conformal prediction | Conformal prediction is an assumption-lean approach to generating distribution-free prediction intervals or sets, for nearly arbitrary predictive models, with guaranteed finite-sample coverage. Conformal methods are an active research topic in statistics and machine learning, but only recently have they been extended to non-exchangeable data. In this paper, we invite survey methodologists to begin using and contributing to conformal methods. We introduce how conformal prediction can be applied to data from several common complex sample survey designs, under a framework of design-based inference for a finite population, and we point out gaps where survey methodologists could fruitfully apply their expertise. Our simulations empirically bear out the theoretical guarantees of finite-sample coverage, and our real-data example demonstrates how conformal prediction can be applied to complex sample survey data in practice. | ['Jerzy Wieczorek'] | 2023-03-02 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [ 5.30144095e-01 5.48117876e-01 -7.05511808e-01 -7.92781174e-01
-1.01903462e+00 -4.76742417e-01 5.31477749e-01 4.08729874e-02
-2.90537477e-02 1.14593637e+00 4.92957085e-01 -8.62481177e-01
-5.91153741e-01 -1.01470017e+00 -6.49098158e-01 -2.92700589e-01
-2.96948552e-01 7.69735277e-01 -1.76039979e-01 1.53068736e-01
2.39813432e-01 1.04808971e-01 -1.24433410e+00 -1.24868952e-01
9.87327218e-01 3.86929393e-01 -2.10946172e-01 4.64909017e-01
1.78600401e-01 4.02188599e-01 -1.11354768e-01 -8.76748383e-01
9.89582688e-02 -4.11419749e-01 -4.48294461e-01 -3.69776398e-01
6.59820020e-01 -3.46558154e-01 -6.37975382e-03 7.31894553e-01
4.82539862e-01 -1.15934551e-01 1.13666749e+00 -1.11393487e+00
-7.30881155e-01 7.15437710e-01 -3.56780380e-01 -1.52814999e-01
4.95305032e-01 -3.23661298e-01 8.07730377e-01 -8.15829992e-01
2.74222702e-01 1.22880757e+00 1.30525351e+00 3.49160254e-01
-1.55819964e+00 -5.90667069e-01 -1.46290764e-01 -2.43401811e-01
-1.27127528e+00 -3.30588669e-01 4.78006303e-01 -5.48362017e-01
2.70395279e-01 6.55301511e-01 1.01031792e+00 1.07247376e+00
6.89673066e-01 5.33978105e-01 1.33151054e+00 -5.71345925e-01
7.10627556e-01 2.66258210e-01 1.22677937e-01 2.95278281e-01
4.38634038e-01 7.78842628e-01 -1.37994736e-01 -8.86907756e-01
9.83394146e-01 3.20465595e-01 1.33303618e-02 -5.50328016e-01
-8.75462115e-01 1.39193249e+00 -1.22795440e-01 -3.36680114e-01
-9.18974578e-02 2.31812939e-01 -5.31699844e-02 1.60579637e-01
8.52417588e-01 -4.82374104e-03 -3.68865579e-01 3.98534499e-02
-9.93043482e-01 8.21986318e-01 1.01868248e+00 1.40597188e+00
3.50762278e-01 -1.69076025e-01 -3.92385125e-01 7.49498367e-01
2.26953015e-01 1.05482531e+00 -2.61979997e-01 -9.98028874e-01
1.24169096e-01 1.77023217e-01 4.49130505e-01 -8.26076150e-01
-4.90216553e-01 -1.42512724e-01 -1.01255417e+00 -9.64058563e-02
4.89641756e-01 -4.56669778e-01 -4.56354171e-01 1.60828280e+00
3.00357759e-01 -7.99631607e-03 -2.70215631e-01 3.98522764e-01
3.04704070e-01 2.02001587e-01 1.71119720e-01 -3.34973514e-01
1.40890467e+00 -6.78907484e-02 -4.84429955e-01 -8.61359388e-02
4.85954344e-01 -7.29938030e-01 1.05782473e+00 1.60860285e-01
-1.11216140e+00 -1.24681912e-01 -6.63973868e-01 2.85365611e-01
4.55613360e-02 -1.97988540e-01 1.02152932e+00 9.00601327e-01
-8.48907292e-01 5.67239285e-01 -6.44703567e-01 -5.48467994e-01
6.75304413e-01 1.02568418e-01 3.21177617e-02 -1.87926441e-01
-1.03437090e+00 7.40895808e-01 -2.96349078e-01 -3.08003485e-01
-4.32029635e-01 -1.32535374e+00 -8.07070136e-01 4.88982126e-02
1.62727907e-01 -1.00448954e+00 1.36512971e+00 -4.84993041e-01
-9.24570203e-01 5.58301747e-01 -2.89414287e-01 -3.69931042e-01
4.29647267e-01 1.52202934e-01 -4.52776790e-01 -4.27712202e-01
2.52706885e-01 3.04496765e-01 3.35671991e-01 -1.04019940e+00
-7.39475250e-01 -3.03501874e-01 -2.40822047e-01 -4.62435298e-02
-1.24323338e-01 -4.49357815e-02 2.43924424e-01 -8.46682072e-01
3.57805826e-02 -8.22783113e-01 -7.73532152e-01 -6.19780214e-04
-1.42406538e-01 -3.04257035e-01 -2.52152264e-01 -3.75363320e-01
1.14413428e+00 -1.79104495e+00 -4.05099422e-01 5.11931360e-01
-1.89536765e-01 -6.48080647e-01 1.32269397e-01 7.60312259e-01
2.69591331e-01 3.58559042e-02 -6.53790653e-01 -3.76095250e-02
2.13311478e-01 7.65865967e-02 -5.15078306e-01 6.91766679e-01
-2.13589326e-01 9.22134101e-01 -9.09551024e-01 -5.07121325e-01
5.83618991e-02 1.71024054e-01 -9.66507077e-01 -6.81273267e-02
-1.50041049e-02 4.22657095e-02 -2.97782183e-01 5.67943573e-01
7.92009890e-01 -2.69175231e-01 4.51030463e-01 3.85662764e-01
-2.88813431e-02 3.01431250e-02 -1.08486021e+00 1.19326818e+00
-4.84922647e-01 5.03408432e-01 -2.02118963e-01 -1.13425863e+00
9.60261762e-01 5.94333522e-02 3.34245473e-01 -4.72567290e-01
-2.46522993e-01 1.99124426e-01 -2.57262081e-01 -1.04893066e-01
3.97692949e-01 -6.50891721e-01 -4.65682328e-01 6.22283101e-01
-2.61771262e-01 -2.56322920e-01 -4.20545787e-01 -2.22230360e-01
9.61852729e-01 -1.06787249e-01 5.14696002e-01 -1.06647897e+00
-1.97677702e-01 9.54966098e-02 3.25410962e-01 1.27222717e+00
1.59355760e-01 6.07907593e-01 4.25507545e-01 -6.84587657e-01
-1.38969886e+00 -1.48632360e+00 -1.10965312e+00 7.99540520e-01
-2.78841943e-01 -1.04584403e-01 -6.50137126e-01 -5.30485213e-01
4.10512120e-01 9.72884178e-01 -1.03936613e+00 -4.69640903e-02
-1.00845724e-01 -8.27828288e-01 3.51988435e-01 5.07800460e-01
-9.73736867e-02 -7.32659221e-01 -4.45995510e-01 2.25563928e-01
1.14324205e-01 -3.62439036e-01 -4.98327255e-01 -1.11301988e-01
-9.59356248e-01 -1.05329561e+00 -9.74209428e-01 -6.93805397e-01
6.98274791e-01 6.15242086e-02 1.38657188e+00 -1.51660472e-01
-8.17093924e-02 8.09888542e-01 -4.30225693e-02 -8.08158159e-01
-4.70840454e-01 -2.11666133e-02 1.83830053e-01 -3.41100007e-01
7.06394315e-01 -7.25377083e-01 -7.24078059e-01 4.99571085e-01
-6.50424838e-01 4.98364903e-02 2.22428367e-01 1.07214379e+00
5.07249475e-01 -4.04061645e-01 1.06789136e+00 -1.27997363e+00
5.47417223e-01 -7.13998616e-01 -9.99743819e-01 3.58513951e-01
-8.31737459e-01 -1.45156145e-01 1.04703538e-01 -4.30663019e-01
-9.66877043e-01 -3.71231645e-01 -8.30437057e-03 2.34034419e-01
2.41105165e-02 6.69961751e-01 8.80623460e-02 1.07349232e-01
8.91969025e-01 4.09613848e-02 1.71118453e-01 -3.55143785e-01
1.76752582e-01 7.75585711e-01 2.15434596e-01 -6.80779994e-01
5.40383577e-01 5.86007953e-01 3.91090810e-01 -5.55608630e-01
-9.03634667e-01 1.43017098e-01 -2.83615619e-01 8.64374265e-02
4.98319387e-01 -9.07200038e-01 -8.21518779e-01 -2.39123523e-01
-5.92879593e-01 -6.88947618e-01 -7.36829221e-01 7.73238242e-01
-1.01783216e+00 -1.62038039e-02 8.74604061e-02 -1.15282786e+00
-1.69826359e-01 -5.79660535e-01 1.04324448e+00 5.01203053e-02
-5.45624614e-01 -1.65141928e+00 5.36499381e-01 -1.34889096e-01
5.04788876e-01 1.65511534e-01 1.03233194e+00 -5.46248198e-01
-2.34324202e-01 -3.67949069e-01 6.04412798e-03 -2.39466980e-01
7.08039328e-02 -2.14068159e-01 -7.31158972e-01 -3.36039245e-01
-1.10258825e-01 -3.55634354e-02 3.93408775e-01 1.08795404e+00
1.63362908e+00 -6.15674138e-01 -6.88151181e-01 5.81621528e-01
1.35683942e+00 -5.95138855e-02 5.64315617e-01 -2.23800987e-01
5.86085953e-02 7.92699039e-01 7.58850634e-01 7.47504354e-01
6.30006075e-01 6.51185989e-01 -2.69496053e-01 -1.20453604e-01
2.59793192e-01 -8.27720284e-01 -7.13758543e-02 3.23712587e-01
2.21025199e-01 1.91253915e-01 -7.84836113e-01 6.76575661e-01
-1.78353977e+00 -1.29937160e+00 -3.23354483e-01 2.75448132e+00
9.83763993e-01 -1.06973469e-01 3.50173205e-01 -3.74527127e-01
5.81425548e-01 -4.51890826e-01 -3.57712746e-01 -3.95230085e-01
-1.26178339e-01 3.38012606e-01 7.81884074e-01 6.57874107e-01
-7.56509602e-01 3.44370782e-01 8.40660667e+00 1.04947424e+00
-2.23233998e-01 1.52853280e-01 8.81033361e-01 -2.01428477e-02
-1.34720981e+00 4.93852854e-01 -7.78838277e-01 4.81026590e-01
1.10833359e+00 -4.96530175e-01 3.17318827e-01 8.15153599e-01
1.05638430e-01 -2.50884116e-01 -1.38973761e+00 5.67547739e-01
-2.13560849e-01 -1.59264052e+00 -2.27663696e-01 3.82718235e-01
9.81285691e-01 -3.50120038e-01 1.32206798e-01 2.78886527e-01
7.85633504e-01 -1.41441131e+00 3.84593278e-01 9.60853219e-01
1.27341235e+00 -8.68492961e-01 7.88050234e-01 3.74173939e-01
-7.26108670e-01 6.58372119e-02 -8.38957608e-01 -3.01216036e-01
2.12554410e-01 1.01862252e+00 -7.72307515e-01 3.15573692e-01
6.18181288e-01 3.47027600e-01 -2.60748953e-01 1.23406243e+00
3.52860928e-01 1.12714899e+00 -4.44444090e-01 -3.59805554e-01
-3.79765898e-01 -3.35224867e-01 2.42571503e-01 8.98629725e-01
8.11016083e-01 3.07257146e-01 -1.94046572e-01 9.73634124e-01
4.13818300e-01 2.59892065e-02 -1.09386432e+00 4.40296471e-01
8.73867631e-01 8.85446787e-01 -3.18481207e-01 -1.10939384e-01
-6.19284987e-01 1.05813682e-01 1.16143018e-01 3.81252915e-01
-7.49653876e-01 -1.00293204e-01 3.17708284e-01 5.52881956e-01
1.05203912e-02 2.09082067e-01 -8.29884410e-01 -8.39200199e-01
-2.84349620e-01 -6.57507360e-01 5.19290447e-01 -5.18455207e-01
-1.78808630e+00 -1.98119864e-01 6.74242079e-01 -1.07378387e+00
-4.68778849e-01 -3.23918939e-01 -5.84098756e-01 1.16896403e+00
-7.86221504e-01 -1.04757619e+00 1.18947476e-01 3.07386279e-01
9.19971094e-02 -3.28025818e-01 1.10068297e+00 2.04755999e-02
1.54481456e-01 9.95124578e-01 3.56306642e-01 -3.22178632e-01
5.54437220e-01 -1.32608771e+00 4.35352206e-01 2.64409065e-01
-8.42673481e-02 6.21280313e-01 7.92753756e-01 -1.01379716e+00
-1.26637328e+00 -9.76980746e-01 9.10887718e-01 -8.28659892e-01
4.40324098e-01 -5.99991679e-01 -5.86629629e-01 1.01858199e+00
3.60309109e-02 -2.45076194e-01 1.23040104e+00 7.68555403e-01
7.98423961e-02 -1.39425889e-01 -1.37400973e+00 9.47922766e-01
1.31410193e+00 -1.87097177e-01 -4.42311913e-01 5.27921140e-01
6.15900636e-01 -2.32231334e-01 -1.28370965e+00 5.57169259e-01
8.87985051e-01 -7.48733222e-01 8.36220384e-01 -5.95840454e-01
3.70327026e-01 2.42833391e-01 -3.95767808e-01 -1.13439763e+00
-5.73431075e-01 -6.86472178e-01 3.19657058e-01 1.20351183e+00
4.74931419e-01 -9.55750823e-01 7.35181987e-01 8.25093627e-01
-9.74494591e-03 -8.38542461e-01 -1.16431141e+00 -4.76572067e-01
8.31488371e-01 -6.24250889e-01 1.04803193e+00 7.14376390e-01
2.61200130e-01 1.75411310e-02 -5.80858231e-01 -1.51833192e-01
1.12497497e+00 3.14483732e-01 1.00328672e+00 -1.67260718e+00
-2.65436590e-01 -2.44115293e-01 -2.54124030e-02 -8.40457141e-01
1.91680957e-02 -7.34364927e-01 -1.74754813e-01 -1.29596329e+00
8.22816432e-01 -1.04451811e+00 3.65395218e-01 1.51398674e-01
-1.92781299e-01 1.10223927e-02 -3.69975179e-01 -1.81948319e-01
2.91471016e-02 5.75599849e-01 1.17158675e+00 9.08477530e-02
7.27588087e-02 7.07490027e-01 -1.11277294e+00 5.60035467e-01
6.93745673e-01 -5.33970356e-01 -7.06262708e-01 2.49398977e-01
5.77621341e-01 5.08592784e-01 6.27247989e-01 -6.44043982e-01
-7.03801215e-02 -8.04163575e-01 7.20660806e-01 -6.24919534e-01
-2.29564980e-01 -8.64737928e-01 7.35170841e-01 3.59376192e-01
-5.10713458e-01 -4.12811451e-02 3.67964596e-01 5.56691527e-01
3.94927800e-01 -2.97068298e-01 3.73379111e-01 2.14815512e-01
2.65559524e-01 3.94565135e-01 -2.49025479e-01 3.34805906e-01
8.84432971e-01 5.19821569e-02 -3.05069059e-01 -6.92448556e-01
-4.63876963e-01 2.15154380e-01 5.56641757e-01 3.53037789e-02
5.47999680e-01 -1.80190599e+00 -1.09231126e+00 3.82369846e-01
1.03108510e-01 -2.55476773e-01 3.70239377e-01 8.36296320e-01
-7.78268352e-02 5.46791255e-01 1.69630483e-01 -6.77589059e-01
-5.55191815e-01 5.54224789e-01 2.64980257e-01 6.16850704e-02
-5.67500234e-01 4.36833352e-01 4.28409994e-01 -8.87190819e-01
8.74720514e-03 -2.00144753e-01 1.75909534e-01 -3.28888357e-01
5.35664618e-01 4.45633978e-01 -4.54374701e-01 1.96537390e-01
-1.96559265e-01 5.33392966e-01 3.85153383e-01 -4.61331338e-01
1.23005366e+00 -2.24230275e-01 1.24302931e-01 6.03841722e-01
7.86100566e-01 2.00730607e-01 -1.26708281e+00 -2.44256929e-02
-2.32041538e-01 -5.08704305e-01 -3.97282630e-01 -7.34805346e-01
-3.21674049e-01 6.27634168e-01 5.18714011e-01 5.11973023e-01
8.84864569e-01 1.64824247e-01 3.25025544e-02 -3.17779481e-02
7.18287289e-01 -7.53406227e-01 -5.77399969e-01 -3.99820767e-02
1.03438997e+00 -1.28197563e+00 3.12017173e-01 -2.15441063e-01
-2.79419720e-01 6.76157594e-01 6.09302111e-02 -2.08347112e-01
1.12437272e+00 4.28432345e-01 -4.10342306e-01 -7.28469864e-02
-8.07573199e-01 3.19500566e-01 6.48764074e-01 1.07325542e+00
5.72195053e-01 5.54154932e-01 -5.78570724e-01 9.70802724e-01
-7.06935465e-01 1.62685499e-01 5.32361627e-01 4.31696773e-01
-4.09616113e-01 -1.14706075e+00 -4.64501143e-01 1.19807744e+00
-2.58598745e-01 -2.22580671e-01 7.17773885e-02 1.06087780e+00
-2.82361180e-01 8.18919182e-01 6.27191842e-01 -2.66565233e-02
1.88421711e-01 -3.08985654e-02 6.41696155e-01 -3.82738352e-01
6.82276636e-02 8.94953534e-02 1.31326601e-01 -1.86129570e-01
-2.82004327e-01 -1.33920538e+00 -5.75650632e-01 -8.91829789e-01
-8.92131999e-02 4.58560437e-01 1.24700665e-01 6.24234200e-01
2.66386718e-01 -2.49631237e-02 7.11102426e-01 -4.20456499e-01
-1.06777918e+00 -9.96976018e-01 -8.37320268e-01 -1.11350417e-01
1.74866527e-01 -6.54063165e-01 -2.96570003e-01 -3.20312023e-01] | [7.732021331787109, 4.550881862640381] |
6d45ea89-760a-49d8-8069-8d5428613a8d | camouflaged-instance-segmentation-dataset-and | 2103.17123 | null | https://arxiv.org/abs/2103.17123v4 | https://arxiv.org/pdf/2103.17123v4.pdf | Camouflaged Instance Segmentation In-The-Wild: Dataset, Method, and Benchmark Suite | This paper pushes the envelope on decomposing camouflaged regions in an image into meaningful components, namely, camouflaged instances. To promote the new task of camouflaged instance segmentation of in-the-wild images, we introduce a dataset, dubbed CAMO++, that extends our preliminary CAMO dataset (camouflaged object segmentation) in terms of quantity and diversity. The new dataset substantially increases the number of images with hierarchical pixel-wise ground truths. We also provide a benchmark suite for the task of camouflaged instance segmentation. In particular, we present an extensive evaluation of state-of-the-art instance segmentation methods on our newly constructed CAMO++ dataset in various scenarios. We also present a camouflage fusion learning (CFL) framework for camouflaged instance segmentation to further improve the performance of state-of-the-art methods. The dataset, model, evaluation suite, and benchmark will be made publicly available on our project page: https://sites.google.com/view/ltnghia/research/camo_plus_plus | ['Khanh-Duy Nguyen', 'Minh-Quan Le', 'Tam V. Nguyen', 'Minh-Triet Tran', 'Thanh-Toan Do', 'Tan-Cong Nguyen', 'Yubo Cao', 'Trung-Nghia Le'] | 2021-03-31 | null | null | null | null | ['camouflaged-object-segmentation'] | ['computer-vision'] | [ 5.86888969e-01 4.92099859e-02 -4.41303879e-01 5.85853541e-03
-8.11658144e-01 -7.75396883e-01 5.14917493e-01 -4.95115876e-01
-1.37470849e-02 6.61645591e-01 -2.81119972e-01 -3.81179154e-01
3.14581901e-01 -4.93323952e-01 -6.96097374e-01 -7.92427242e-01
2.55533457e-01 2.67154843e-01 5.15209734e-01 -9.18367431e-02
3.52497905e-01 3.35146576e-01 -1.53051531e+00 3.13050151e-01
1.12327588e+00 1.03937912e+00 1.39064074e-01 5.61622560e-01
1.10243605e-02 6.17662191e-01 -8.90786290e-01 -2.86345989e-01
7.16419697e-01 -4.65045929e-01 -8.54825616e-01 3.88621151e-01
1.03555679e+00 -1.55742541e-01 -2.84770340e-01 1.33876932e+00
1.72223538e-01 -4.52474058e-02 3.58021677e-01 -1.63402236e+00
-1.01210105e+00 3.24178964e-01 -7.27625310e-01 1.79178074e-01
-1.20781817e-01 5.65711260e-01 5.51199257e-01 -7.73095071e-01
9.57873464e-01 1.02072573e+00 5.94120800e-01 8.26437950e-01
-1.13741124e+00 -9.71625745e-01 2.71577924e-01 1.59321815e-01
-1.09688449e+00 -8.14853236e-02 8.60917032e-01 -5.94697237e-01
4.52948064e-01 6.40949607e-01 7.61592984e-01 1.04292047e+00
1.15715079e-01 1.25297034e+00 1.44668686e+00 -1.80551469e-01
2.09347576e-01 -2.72780478e-01 3.58599991e-01 7.18489885e-01
3.67281705e-01 4.41858381e-01 -1.60247698e-01 1.69458576e-02
5.99288046e-01 8.61774981e-02 -5.58501482e-01 -2.81118065e-01
-1.15897477e+00 6.42209530e-01 5.58064401e-01 -1.16695166e-02
2.03550816e-01 1.53088197e-01 4.13648427e-01 1.50234878e-01
7.79418468e-01 8.26928020e-01 -7.09999919e-01 3.85883674e-02
-9.30931270e-01 4.90124315e-01 5.13481319e-01 8.74645114e-01
7.27190435e-01 1.13139972e-02 -2.76561975e-01 7.57326961e-01
-1.36687383e-01 3.37659895e-01 2.71425664e-01 -1.22871590e+00
1.50560647e-01 5.09626150e-01 1.20316543e-01 -6.54967964e-01
6.75323606e-02 -9.74039286e-02 -6.27939701e-01 4.86627638e-01
4.14482117e-01 -1.16913714e-01 -1.60912001e+00 1.36220562e+00
5.68967819e-01 1.01332188e+00 -2.42170960e-01 1.04433620e+00
9.18050408e-01 6.10370994e-01 -6.86342418e-02 1.38692170e-01
1.21564269e+00 -1.66568458e+00 -7.87699640e-01 -3.25971901e-01
8.31528366e-01 -8.74505222e-01 1.14551079e+00 5.50967157e-01
-7.32478738e-01 -6.44892633e-01 -9.74752903e-01 -3.05048168e-01
-8.36248100e-01 -8.16138610e-02 9.64563429e-01 6.76256418e-01
-8.61020863e-01 3.55129659e-01 -6.75058067e-01 1.77960753e-01
1.04684532e+00 1.50394142e-01 -2.93868750e-01 -1.53118849e-01
-8.97043467e-01 6.96352065e-01 5.19263327e-01 -1.32037839e-02
-9.73361850e-01 -1.19676578e+00 -9.10728276e-01 -2.74336725e-01
8.72680545e-01 -4.60907727e-01 1.11827314e+00 -1.06243443e+00
-9.24020231e-01 1.09717846e+00 1.06035464e-01 -4.66274738e-01
4.49144751e-01 -2.31424838e-01 -4.27005649e-01 1.58552483e-01
2.28887931e-01 1.24995804e+00 1.14650154e+00 -1.87546194e+00
-8.97483528e-01 -8.60485956e-02 1.16815776e-01 -3.69174272e-01
3.94107074e-01 -3.41144323e-01 -5.60485899e-01 -1.11966777e+00
-3.89244020e-01 -1.08437383e+00 -2.67695993e-01 6.83752596e-02
-7.57622361e-01 -4.39869165e-02 1.78405881e+00 -5.80101967e-01
1.17027867e+00 -2.31686211e+00 -1.16047800e-01 -2.65708059e-01
6.87278569e-01 8.67403030e-01 -4.39729929e-01 -1.71058506e-01
-1.90485284e-01 3.88027310e-01 -5.99243581e-01 -5.33676088e-01
-2.30831176e-01 3.16163927e-01 -4.99633253e-01 5.13672650e-01
3.75859439e-01 1.23127103e+00 -1.02985930e+00 -4.65460688e-01
6.60414398e-01 2.32756555e-01 -5.22271454e-01 3.97112519e-02
-6.74320579e-01 9.39338386e-01 -2.01939896e-01 1.16281044e+00
1.17317593e+00 -7.71137103e-02 -4.28207040e-01 1.07666343e-01
3.85337919e-02 -2.55007178e-01 -7.01828659e-01 1.59878862e+00
-1.29840270e-01 7.47358441e-01 1.15272701e-01 -7.09026217e-01
4.89854693e-01 -4.73681912e-02 5.69874942e-01 -5.94331801e-01
2.98075169e-01 2.32079729e-01 -6.11595251e-02 -2.92173326e-01
6.23949111e-01 3.37105066e-01 -5.96422181e-02 7.96048939e-02
6.42000884e-02 -6.00515425e-01 4.62453693e-01 2.28891641e-01
8.75223398e-01 2.70318598e-01 2.86193967e-01 -3.72627944e-01
2.68646300e-01 4.04828966e-01 6.38579845e-01 7.99608290e-01
-6.13874912e-01 9.49889183e-01 5.35148740e-01 -3.49775225e-01
-9.24185991e-01 -9.40444827e-01 -3.77955765e-01 6.27980113e-01
6.83613002e-01 -2.29906261e-01 -1.16525376e+00 -1.11913824e+00
2.16213360e-01 5.47636628e-01 -1.04261661e+00 3.27730440e-02
-4.07291293e-01 -6.26652002e-01 5.65518260e-01 3.06212902e-01
8.20937991e-01 -1.18409586e+00 -4.02273327e-01 -4.32424843e-01
-9.34129581e-02 -1.10298765e+00 -8.67081583e-01 2.72512566e-02
-5.36571920e-01 -1.43680561e+00 -5.94989359e-01 -6.83525383e-01
6.98225856e-01 8.04103255e-01 1.13150573e+00 5.38395405e-01
-6.18828475e-01 1.95282042e-01 -5.71314752e-01 -6.27822340e-01
-2.07792446e-01 -1.55766517e-01 -5.02485275e-01 6.87176175e-03
1.06292702e-01 -1.28930166e-01 -7.45856106e-01 3.62226307e-01
-1.34574294e+00 2.22023681e-01 1.33121341e-01 9.90394354e-01
1.04700756e+00 -1.01167172e-01 2.60388017e-01 -1.32096493e+00
1.22617900e-01 -4.72363859e-01 -8.83641183e-01 8.47991705e-02
-4.92432833e-01 -4.31488574e-01 2.36507401e-01 -5.79353034e-01
-8.62113178e-01 1.06673548e-02 -3.37313823e-02 -8.46391201e-01
-3.81165385e-01 -5.82709499e-02 -2.59564251e-01 -3.87145340e-01
4.65971649e-01 -2.38878652e-01 -1.12272620e-01 -5.15146375e-01
8.79700124e-01 4.62745368e-01 9.30653572e-01 -2.93847769e-01
8.77770722e-01 7.91195869e-01 -9.68164653e-02 -6.69519305e-01
-1.16729200e+00 -6.69544399e-01 -6.15105808e-01 -1.45631194e-01
1.00752687e+00 -7.09975958e-01 -3.20867956e-01 8.57727826e-01
-8.87534142e-01 -8.83105993e-01 -5.18261850e-01 -2.63146043e-01
-3.91976774e-01 4.54666138e-01 -3.86872321e-01 -3.96182120e-01
-1.25487059e-01 -1.43767774e+00 1.62112212e+00 3.97951007e-01
5.89325726e-02 -1.10250592e+00 -2.17178091e-02 9.70176637e-01
1.09668404e-01 7.67138481e-01 6.06317759e-01 -2.85833120e-01
-8.69999170e-01 -4.18335125e-02 -4.50471848e-01 5.68565667e-01
1.22475371e-01 7.31335804e-02 -1.13810062e+00 -2.28818431e-01
-1.67220652e-01 -2.11350724e-01 1.33855593e+00 5.51639855e-01
1.84944165e+00 -1.21342659e-01 -4.92724270e-01 9.44894433e-01
1.39808345e+00 2.63464242e-01 8.68159771e-01 2.98215121e-01
1.19381881e+00 2.49861568e-01 1.04707050e+00 -3.79543044e-02
1.02546461e-01 6.54475331e-01 7.91745245e-01 -5.62450111e-01
-8.91624033e-01 -8.96076113e-02 -8.89203027e-02 2.68851846e-01
2.13920966e-01 -6.47611558e-01 -5.75984597e-01 7.04470515e-01
-1.87642384e+00 -7.96652257e-01 -3.81295621e-01 1.75812244e+00
6.08939171e-01 -9.14732963e-02 -3.29533184e-04 1.36095732e-01
7.69245207e-01 4.46060598e-01 -5.95827699e-01 -3.28847647e-01
-1.53197557e-01 5.40937304e-01 7.36851633e-01 6.92201555e-01
-1.58509135e+00 1.35547566e+00 5.96712828e+00 1.36230910e+00
-1.34431517e+00 4.56540167e-01 9.47734058e-01 2.82314401e-02
-7.39838481e-02 6.07882254e-02 -6.23989999e-01 9.49086905e-01
4.01839167e-01 2.27327332e-01 5.05857468e-01 1.00391209e+00
1.15219548e-01 -4.31559026e-01 -6.20248854e-01 8.11648965e-01
2.59272140e-02 -1.69660437e+00 -1.19960949e-01 2.08572522e-01
1.48718441e+00 1.53144255e-01 4.76396024e-01 2.26929560e-01
1.79030299e-01 -1.00871754e+00 6.64970338e-01 1.84295282e-01
1.05571675e+00 -4.02594030e-01 6.20707393e-01 -4.76828823e-03
-9.68531311e-01 -7.08027110e-02 -8.18131790e-02 4.23978791e-02
2.67563939e-01 8.20177078e-01 -3.10261220e-01 5.00951886e-01
7.46886611e-01 8.74288678e-01 -9.17359829e-01 1.16288590e+00
-1.85017779e-01 7.57206440e-01 -7.42519572e-02 6.44049466e-01
3.78054827e-01 -3.33094031e-01 3.95633996e-01 1.00401938e+00
-7.35612884e-02 -3.49096730e-02 5.62258601e-01 8.75798941e-01
-2.93231368e-01 -4.20966417e-01 -5.75652838e-01 7.34300539e-02
4.61890012e-01 1.35086894e+00 -1.03590298e+00 -6.00233316e-01
-4.22597587e-01 9.77605045e-01 -7.55703449e-02 2.87960708e-01
-1.09504342e+00 -2.97592223e-01 1.03357601e+00 9.85582247e-02
3.54164839e-01 2.28583980e-02 -7.36668706e-01 -1.31389153e+00
-1.97581530e-01 -1.04531491e+00 1.47034541e-01 -8.18182945e-01
-1.34099329e+00 2.32749671e-01 2.02347904e-01 -1.20161414e+00
2.08192348e-01 -8.84705782e-01 -6.24271333e-01 4.54689682e-01
-1.54418576e+00 -1.47783303e+00 -5.52612722e-01 3.34898591e-01
8.17279696e-01 1.10632226e-01 4.33651060e-01 3.48658502e-01
-5.95505655e-01 3.03682506e-01 -1.78643212e-01 1.55319586e-01
7.07077444e-01 -1.33861768e+00 7.86850750e-01 1.20617092e+00
1.19721003e-01 3.06456268e-01 4.48818862e-01 -9.27521348e-01
-1.04996002e+00 -1.49748552e+00 1.66419432e-01 -9.96624351e-01
6.81900620e-01 -4.01029527e-01 -8.79774451e-01 7.38329887e-01
7.16595203e-02 5.58328807e-01 2.98623204e-01 -4.61212605e-01
-3.44468653e-01 3.12223762e-01 -1.47987556e+00 7.62839675e-01
1.28833759e+00 -2.45774806e-01 -3.35880965e-01 5.04853308e-01
1.22095406e+00 -8.70138645e-01 -5.24275780e-01 6.08691216e-01
2.22641960e-01 -8.90704572e-01 9.70349669e-01 -3.89308184e-01
5.18336475e-01 -6.41403079e-01 7.28744417e-02 -1.14378345e+00
-7.35016763e-02 -8.36721420e-01 -4.03427422e-01 1.01169550e+00
3.16422507e-02 -7.16039181e-01 9.37739968e-01 2.25688949e-01
-5.26332021e-01 -1.03827953e+00 -9.24694002e-01 -7.25003242e-01
1.53040111e-01 -4.76778120e-01 7.67506659e-01 1.12815166e+00
-6.00411534e-01 -1.77666441e-01 -4.31852847e-01 2.20421273e-02
6.28356397e-01 1.85275197e-01 8.66816580e-01 -9.86185849e-01
-1.44883528e-01 -4.45777148e-01 -4.20851588e-01 -1.03498518e+00
1.93005159e-01 -8.98282170e-01 -6.38087317e-02 -1.08742571e+00
1.35412917e-01 -4.08588499e-01 -8.30445886e-02 7.43440688e-01
-5.41910410e-01 1.12864292e+00 5.50979078e-01 1.24897227e-01
-5.76302707e-01 1.15409359e-01 1.87419426e+00 -3.71131361e-01
-9.94444117e-02 -2.18407780e-01 -7.96730459e-01 6.79245651e-01
9.40780103e-01 -4.02460992e-01 -3.79433960e-01 -3.87508959e-01
-5.31486571e-01 -3.76961291e-01 7.11336553e-01 -7.22905338e-01
-4.30494368e-01 -3.17961365e-01 1.40286997e-01 -5.97045183e-01
3.43947411e-01 -6.06184244e-01 6.27460480e-02 4.96810466e-01
-2.02113073e-02 -1.92850754e-01 3.05636197e-01 4.23672855e-01
-2.07199723e-01 -2.92007753e-04 1.22103465e+00 -1.07920632e-01
-1.34723949e+00 5.72925270e-01 1.06487293e-02 2.26082265e-01
1.25836611e+00 -4.46020156e-01 -1.03252387e+00 1.09289363e-01
-4.99520272e-01 2.98423052e-01 8.55014920e-01 6.56477809e-01
4.26961452e-01 -1.07711411e+00 -2.42891967e-01 2.81145126e-01
2.95490533e-01 1.39323443e-01 4.13573772e-01 7.81023622e-01
-7.38116384e-01 3.02568018e-01 -2.11655885e-01 -4.96657401e-01
-1.37215376e+00 8.79745483e-01 2.72263110e-01 -1.70978680e-01
-7.15373218e-01 8.99183989e-01 5.31699300e-01 -5.25539756e-01
-1.09136641e-01 -5.75109899e-01 2.52186388e-01 -4.10752654e-01
4.87282723e-01 4.67026502e-01 -4.67064939e-02 -6.05085254e-01
-1.73190877e-01 4.62055683e-01 -6.39899448e-03 2.56317019e-01
8.02257240e-01 5.73193729e-02 -1.91064075e-01 -6.66869357e-02
1.15049469e+00 -5.88821433e-02 -1.58104098e+00 2.75378495e-01
-3.04219186e-01 -1.05728185e+00 5.82373440e-02 -1.18545556e+00
-1.53630209e+00 5.63669682e-01 6.65433645e-01 9.34997946e-02
1.21564686e+00 1.46923169e-01 1.13127947e+00 -1.53824300e-01
3.61665696e-01 -8.00358117e-01 -2.86426824e-02 2.83993959e-01
6.12519085e-01 -1.36025584e+00 -8.19198489e-02 -1.03789902e+00
-5.75565159e-01 2.80945927e-01 7.76621282e-01 -5.03075480e-01
8.00783873e-01 5.35560489e-01 3.96343291e-01 -1.93935022e-01
-4.39754218e-01 -4.41643476e-01 4.22309577e-01 1.05203545e+00
-4.96686287e-02 3.53677422e-01 -3.62316370e-01 3.00996453e-01
-2.68153161e-01 8.41182992e-02 5.72601497e-01 7.53019691e-01
-2.18414173e-01 -1.11867654e+00 -5.56261420e-01 6.06995523e-01
-4.76144731e-01 2.83019096e-02 -7.67976820e-01 1.02842760e+00
9.19003785e-01 9.99748051e-01 -1.06680142e-02 -2.96833396e-01
5.91758043e-02 -3.78007561e-01 5.62083900e-01 -7.28670061e-01
-5.17674387e-01 -5.47895320e-02 -1.26198202e-01 -8.74955416e-01
-2.91818529e-01 -4.15586054e-01 -9.73373413e-01 -3.55767787e-01
-3.97321582e-01 -1.91648334e-01 4.68930781e-01 9.65563118e-01
4.40266222e-01 5.33119559e-01 2.11674213e-01 -1.35139894e+00
1.35720357e-01 -7.33086169e-01 -5.79361916e-01 6.25753403e-01
4.51874465e-01 -8.17357659e-01 -3.73662889e-01 2.69400716e-01] | [9.654020309448242, -0.16972748935222626] |
526f8f0c-2e8f-44e2-a509-e991cec4bbe2 | representation-learning-via-invariant-causal-1 | 2010.07922 | null | https://arxiv.org/abs/2010.07922v1 | https://arxiv.org/pdf/2010.07922v1.pdf | Representation Learning via Invariant Causal Mechanisms | Self-supervised learning has emerged as a strategy to reduce the reliance on costly supervised signal by pretraining representations only using unlabeled data. These methods combine heuristic proxy classification tasks with data augmentations and have achieved significant success, but our theoretical understanding of this success remains limited. In this paper we analyze self-supervised representation learning using a causal framework. We show how data augmentations can be more effectively utilized through explicit invariance constraints on the proxy classifiers employed during pretraining. Based on this, we propose a novel self-supervised objective, Representation Learning via Invariant Causal Mechanisms (ReLIC), that enforces invariant prediction of proxy targets across augmentations through an invariance regularizer which yields improved generalization guarantees. Further, using causality we generalize contrastive learning, a particular kind of self-supervised method, and provide an alternative theoretical explanation for the success of these methods. Empirically, ReLIC significantly outperforms competing methods in terms of robustness and out-of-distribution generalization on ImageNet, while also significantly outperforming these methods on Atari achieving above human-level performance on $51$ out of $57$ games. | ['Charles Blundell', 'Lars Buesing', 'Jacob Walker', 'Brian McWilliams', 'Jovana Mitrovic'] | 2020-10-15 | representation-learning-via-invariant-causal | https://arxiv.org/abs/2010.07922 | https://arxiv.org/pdf/2010.07922.pdf | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 5.65406978e-01 6.24598324e-01 -7.42797434e-01 -6.68936372e-01
-6.60955727e-01 -3.27716917e-01 9.64985907e-01 4.02281471e-02
-4.90673512e-01 8.59687805e-01 5.06726384e-01 -9.20159295e-02
-2.67978549e-01 -5.38871229e-01 -9.67885911e-01 -5.57687521e-01
-1.21792063e-01 4.45310026e-01 -9.36558545e-02 -1.54132843e-01
1.43816814e-01 4.91363883e-01 -1.61874199e+00 9.63889658e-02
8.64458919e-01 9.68493104e-01 -1.39465243e-01 5.22761717e-02
1.93635404e-01 9.77487147e-01 -2.78705448e-01 -1.71084359e-01
4.01944757e-01 -5.46979368e-01 -7.74581552e-01 3.37978154e-01
6.02148652e-01 -1.18878372e-01 -3.92015666e-01 8.92803431e-01
1.81401715e-01 3.58897507e-01 8.63518834e-01 -1.24639249e+00
-8.88458371e-01 7.86840379e-01 -5.23599625e-01 1.53468370e-01
-1.57203451e-01 2.21619327e-02 1.24831772e+00 -6.59587979e-01
4.82443392e-01 1.18969917e+00 6.71783328e-01 7.33241439e-01
-1.75817943e+00 -7.30168402e-01 4.38588440e-01 -5.83727136e-02
-1.02716982e+00 -5.59844434e-01 7.73294747e-01 -3.40880513e-01
7.39086866e-01 -9.91021171e-02 3.30807328e-01 1.25632358e+00
-2.67773867e-01 1.17376113e+00 1.44126189e+00 -6.30642235e-01
3.50108236e-01 1.77793607e-01 3.25848281e-01 9.40660357e-01
2.50260383e-01 5.16812980e-01 -7.07985282e-01 -9.42602605e-02
7.50663936e-01 -3.34839821e-02 -1.02626890e-01 -7.11015761e-01
-8.59627187e-01 1.10773206e+00 8.27312171e-01 1.36876568e-01
-3.87635320e-01 4.44601566e-01 2.88475484e-01 2.91777223e-01
5.96725464e-01 8.46205473e-01 -6.27122164e-01 2.77452052e-01
-8.16069305e-01 -3.89914736e-02 3.53260905e-01 8.87483835e-01
8.68399739e-01 6.16389930e-01 -8.55223164e-02 8.95200849e-01
1.43074751e-01 3.62298101e-01 7.75001168e-01 -8.61134708e-01
8.27527866e-02 6.48457408e-01 -2.36536562e-01 -6.56837523e-01
-4.05349374e-01 -8.92560661e-01 -7.12992013e-01 1.62706256e-01
1.86566383e-01 7.35594332e-02 -1.14782643e+00 2.25949836e+00
-1.86789662e-01 1.70016199e-01 1.45833880e-01 6.45930707e-01
1.56891689e-01 3.10389280e-01 3.85333151e-01 -3.30779403e-01
8.65763962e-01 -8.72998476e-01 -5.13726234e-01 -3.59137237e-01
7.85407126e-01 -1.88184485e-01 1.26588333e+00 3.88924748e-01
-8.00034881e-01 -4.98169690e-01 -1.28498816e+00 2.58556068e-01
-3.00465524e-01 6.10322282e-02 1.17167056e+00 7.20779777e-01
-8.75821769e-01 1.01184833e+00 -9.67775822e-01 -3.05776030e-01
8.47431898e-01 5.14327705e-01 -4.64834034e-01 -1.50659680e-01
-9.55788136e-01 1.05654514e+00 3.24436635e-01 -4.61804241e-01
-1.10568321e+00 -8.25903654e-01 -9.29181457e-01 -1.32360280e-01
4.54729289e-01 -3.94378006e-01 1.16646194e+00 -1.35442054e+00
-1.28109670e+00 7.50610173e-01 8.53390992e-02 -9.71508622e-01
3.41563255e-01 -4.63371217e-01 -1.48508340e-01 1.37864100e-02
3.78490061e-01 1.02654123e+00 1.12965655e+00 -1.44864511e+00
-3.17998737e-01 -4.30410713e-01 -2.03196958e-01 1.48292989e-01
-7.54190326e-01 -3.29241633e-01 -7.25646168e-02 -9.11497653e-01
1.74993888e-01 -9.85570073e-01 -4.85213310e-01 -1.36413082e-01
-1.48318261e-01 -3.99365395e-01 9.07629073e-01 -1.35286868e-01
7.97886312e-01 -2.08921385e+00 9.22031030e-02 2.94299453e-01
1.01097830e-01 1.42217293e-01 -3.90270591e-01 -3.36613208e-02
-3.81646484e-01 3.28455597e-01 -4.83243734e-01 -3.18944633e-01
-1.16060019e-01 6.52039111e-01 -6.63706720e-01 5.79261601e-01
6.25960469e-01 8.52114856e-01 -9.75502014e-01 -2.80049205e-01
1.57362074e-01 3.10087949e-01 -5.87200761e-01 1.80712283e-01
-2.35677809e-01 2.56852180e-01 -2.67912745e-01 6.55795097e-01
2.15214312e-01 -2.49865726e-01 3.19461912e-01 -6.25819117e-02
4.80081886e-02 2.93879658e-01 -8.10701132e-01 1.68971860e+00
-5.41350961e-01 6.06539249e-01 -3.13318402e-01 -1.53202569e+00
8.92658532e-01 1.13147169e-01 5.54583013e-01 -7.54084527e-01
1.23253986e-02 1.58896774e-01 1.12775505e-01 -1.74128264e-02
1.85755238e-01 -2.75239557e-01 6.97276220e-02 5.44367909e-01
4.06930894e-01 -1.31231938e-02 1.71036765e-01 3.00277710e-01
1.16156209e+00 5.50200462e-01 3.19910675e-01 -4.74160552e-01
6.37959316e-02 1.23957373e-01 5.87339997e-01 1.07164240e+00
-7.67626613e-02 4.77929354e-01 2.16107875e-01 -3.03338677e-01
-1.02426410e+00 -1.07120025e+00 -2.72891432e-01 1.25865197e+00
-1.92860737e-01 -2.82078028e-01 -3.96948457e-01 -9.74878371e-01
1.46066964e-01 8.69837046e-01 -9.69551146e-01 -4.82871622e-01
-5.06011605e-01 -1.00464237e+00 6.08969450e-01 1.07930577e+00
6.43485665e-01 -9.45584953e-01 -5.04930496e-01 8.31126273e-02
1.46135822e-01 -9.83785450e-01 -4.73020747e-02 9.21348631e-01
-1.24861932e+00 -9.51570928e-01 -4.97227877e-01 -4.52559739e-01
8.91295552e-01 3.67392182e-01 1.01423848e+00 2.25654375e-02
-2.02440694e-01 6.34038448e-01 -3.88272643e-01 -4.73323435e-01
-3.91211152e-01 7.66924024e-02 5.36241531e-01 -2.44385786e-02
8.94329473e-02 -7.68455446e-01 -2.09024325e-01 4.04483497e-01
-8.32321286e-01 -1.78563654e-01 9.87808645e-01 1.29889512e+00
6.11501217e-01 -9.84307379e-02 8.03175211e-01 -1.08757901e+00
4.53941494e-01 -5.46550214e-01 -5.71336150e-01 2.66576884e-03
-1.26624358e+00 6.20080888e-01 5.48053265e-01 -5.49428999e-01
-1.18405843e+00 3.35051239e-01 3.58507276e-01 -7.04936147e-01
3.49575989e-02 5.52236080e-01 -7.18748197e-02 -8.40721130e-02
1.18829072e+00 7.44090378e-02 3.10534477e-01 -4.03510302e-01
5.87790430e-01 1.34330437e-01 5.31879067e-01 -8.43229413e-01
9.98527348e-01 5.94541252e-01 1.38797745e-01 -7.54171491e-01
-1.37423027e+00 -3.09795767e-01 -7.03527153e-01 -2.94099655e-02
5.93931317e-01 -1.04321051e+00 -2.60711133e-01 -1.36592716e-01
-7.15402842e-01 -5.97123444e-01 -6.88900173e-01 5.54821312e-01
-8.13291967e-01 9.63854417e-02 -3.85063320e-01 -9.00246680e-01
6.54485375e-02 -7.88581371e-01 6.08368039e-01 1.70533769e-02
-2.25853831e-01 -9.25975621e-01 -2.12683722e-01 2.67576128e-01
4.21032697e-01 1.05436660e-01 8.51754844e-01 -8.77827406e-01
-4.41270441e-01 -2.71152437e-01 -3.25782031e-01 7.15601861e-01
2.43685663e-01 -4.40734714e-01 -1.22370279e+00 -2.03532994e-01
-1.93237215e-01 -8.02678347e-01 1.35435200e+00 3.57042491e-01
1.32730854e+00 -3.54730457e-01 -2.31204838e-01 4.84893709e-01
1.36284208e+00 -9.46455225e-02 6.55723989e-01 3.55333269e-01
6.67116880e-01 6.84639335e-01 4.99857068e-01 2.21976861e-01
-9.91696045e-02 5.22494614e-01 5.70431530e-01 -2.01475069e-01
-4.02763784e-01 -5.26086807e-01 2.72167504e-01 2.95938522e-01
-3.27112705e-01 3.88411358e-02 -7.82681167e-01 5.18166304e-01
-2.07406688e+00 -9.90911126e-01 2.19627738e-01 2.25461411e+00
8.55084419e-01 3.76656622e-01 7.90077597e-02 2.50756502e-01
5.00270605e-01 7.00906515e-02 -7.61588931e-01 9.74527150e-02
-3.20113242e-01 4.15779471e-01 8.56982350e-01 3.04671705e-01
-1.33303475e+00 1.15456223e+00 6.96238232e+00 6.54069483e-01
-7.42588878e-01 1.27541572e-01 6.61886871e-01 1.70951292e-01
-1.29971221e-01 5.12174070e-02 -5.96879303e-01 1.54691255e-02
9.10211921e-01 1.87696233e-01 3.94606471e-01 1.18157136e+00
-9.23463050e-03 4.40817736e-02 -1.39184213e+00 7.52296984e-01
2.88382083e-01 -1.52384019e+00 1.17892548e-01 1.36196956e-01
9.13429677e-01 9.59791616e-02 3.98970515e-01 5.48078597e-01
8.08904231e-01 -1.04036450e+00 5.61810315e-01 2.88310140e-01
6.64586663e-01 -6.04111016e-01 5.28396308e-01 1.82145357e-01
-6.84590936e-01 -4.06168163e-01 -2.88757235e-01 -2.18490630e-01
-3.78495574e-01 3.61742228e-01 -7.59279966e-01 2.23028094e-01
4.87743884e-01 8.62357199e-01 -7.79778302e-01 6.15049303e-01
-6.39756501e-01 1.05817175e+00 -2.78474152e-01 2.45740741e-01
2.52317458e-01 2.27284685e-01 3.57506901e-01 1.16001701e+00
-1.63140371e-01 5.23766652e-02 3.90286684e-01 8.10229182e-01
-3.32718253e-01 -8.00363496e-02 -8.77581775e-01 -1.57496721e-01
3.11917901e-01 9.45319653e-01 -6.94861174e-01 -3.12323630e-01
-1.37420073e-01 7.78046131e-01 5.40328026e-01 3.27226847e-01
-6.97030902e-01 1.23920269e-01 4.20949221e-01 8.86948109e-02
5.17188683e-02 -6.30010590e-02 -4.79321629e-01 -1.13837874e+00
-2.78270215e-01 -7.72890747e-01 4.87910986e-01 -6.21189356e-01
-1.47168517e+00 3.42267275e-01 2.01542422e-01 -1.23499823e+00
-2.54769027e-01 -7.18584716e-01 -3.01265419e-01 4.65449482e-01
-1.56913209e+00 -1.26714551e+00 -3.82074378e-02 5.87539077e-01
6.13769829e-01 -4.88386482e-01 9.46176469e-01 -5.62954210e-02
-5.56139469e-01 7.38396168e-01 -1.14216641e-01 1.14836685e-01
7.13094234e-01 -1.38477850e+00 -4.88043092e-02 9.76251960e-01
3.79491478e-01 7.65124142e-01 5.70479572e-01 -6.55477285e-01
-1.19167066e+00 -1.17721725e+00 2.21077055e-01 -3.41003895e-01
7.35822856e-01 -7.60999247e-02 -8.32883835e-01 7.89748609e-01
4.49575037e-02 2.67798662e-01 7.04995096e-01 5.32503366e-01
-1.03275347e+00 -1.32654607e-01 -1.02331305e+00 5.67520738e-01
1.28205216e+00 -4.97839838e-01 -7.03758121e-01 4.83122200e-01
6.55295849e-01 4.45543565e-02 -5.92934012e-01 5.46399832e-01
2.09687293e-01 -6.90151572e-01 8.83362651e-01 -9.71077681e-01
5.27600229e-01 -1.05489835e-01 -5.12223661e-01 -1.39310074e+00
-4.40988481e-01 -5.40113866e-01 -1.18277997e-01 1.05197263e+00
4.64351565e-01 -6.90608025e-01 9.67951059e-01 4.34500933e-01
-4.99573827e-01 -5.76836169e-01 -7.17568874e-01 -1.19996274e+00
1.35449544e-01 -5.03523707e-01 7.55333081e-02 1.20808852e+00
-4.43964563e-02 5.10316372e-01 -4.69954610e-01 -1.53578874e-02
1.02070820e+00 -3.05049539e-01 5.62746882e-01 -1.42979169e+00
-2.96770960e-01 -4.46789235e-01 -5.54453194e-01 -9.02677953e-01
6.40043199e-01 -1.16058338e+00 1.65104777e-01 -1.10138178e+00
2.00886782e-02 -6.72875464e-01 -4.81051862e-01 1.16868043e+00
7.50445426e-02 4.09560323e-01 1.52363792e-01 5.53727031e-01
-4.27355021e-01 6.40009165e-01 8.01717401e-01 -2.46874854e-01
-1.91668540e-01 -1.86442882e-01 -9.63673174e-01 9.85475421e-01
8.93121421e-01 -5.77151060e-01 -6.14718437e-01 -3.14454228e-01
-6.68260157e-02 -3.36162537e-01 4.78219867e-01 -1.03035033e+00
-1.08756006e-01 -1.37571290e-01 6.26360774e-01 -1.31589562e-01
2.90520668e-01 -7.27688611e-01 -5.53103805e-01 5.82948625e-01
-9.37024713e-01 -1.52882084e-01 2.31312141e-01 9.82989192e-01
-1.76614653e-02 -2.27557003e-01 9.85661685e-01 -1.81362599e-01
-7.01495588e-01 8.82625729e-02 -3.22112232e-01 3.50207724e-02
7.46683478e-01 -4.75421734e-02 -2.22309560e-01 -2.53020287e-01
-8.10787976e-01 4.56284843e-02 1.17104709e-01 3.01079422e-01
6.07656240e-01 -1.37338912e+00 -5.73233545e-01 1.32085472e-01
2.49056950e-01 -3.23752373e-01 -2.42112517e-01 6.19433403e-01
1.93866998e-01 2.29496643e-01 -2.33622819e-01 -5.66148520e-01
-7.62478650e-01 5.97810090e-01 2.40869373e-01 -1.83625653e-01
-5.25921762e-01 7.92908013e-01 1.33233979e-01 -2.64964670e-01
3.85740131e-01 1.23165391e-01 -5.33396304e-02 6.70970380e-02
2.72225738e-01 1.83060244e-01 -4.46293131e-02 -3.31383586e-01
-1.82136670e-01 1.78909674e-01 -3.87650490e-01 -3.60547811e-01
1.45480955e+00 1.77870542e-01 2.11887255e-01 3.43641341e-01
1.06878495e+00 -2.39148244e-01 -1.57055414e+00 -5.31316817e-01
3.42797756e-01 -3.40191126e-01 2.45121077e-01 -8.62426877e-01
-1.01586616e+00 6.52077794e-01 6.38941586e-01 1.44120641e-02
9.72172797e-01 9.68887880e-02 9.46371555e-02 6.58118665e-01
2.19061673e-01 -1.18616819e+00 7.01452315e-01 4.11682546e-01
9.78986740e-01 -1.50935400e+00 3.12845409e-01 -2.83876956e-01
-6.78134620e-01 9.27274942e-01 7.52833903e-01 -5.03365636e-01
5.65741181e-01 -6.96974620e-03 -2.04800904e-01 -5.04887179e-02
-7.57060587e-01 -6.00044429e-01 3.16421360e-01 9.59705949e-01
4.37084794e-01 -1.11909490e-02 -9.52430964e-02 2.74592102e-01
-5.25203869e-02 -2.06727788e-01 4.46389884e-01 9.11893845e-01
-2.86783338e-01 -1.03772795e+00 -1.36637390e-01 5.45342684e-01
-2.10568383e-01 -2.50967652e-01 -3.77437681e-01 1.05992675e+00
-3.13939489e-02 8.89486432e-01 8.46319050e-02 -3.54716629e-01
1.47405773e-01 2.10260436e-01 5.72420120e-01 -8.72801721e-01
-2.82034278e-01 4.52909544e-02 7.50061944e-02 -3.91478837e-01
-7.36739397e-01 -7.56373227e-01 -1.22043395e+00 2.02709720e-01
-4.36633259e-01 -2.08913609e-02 5.51654637e-01 1.02107632e+00
2.10872874e-01 4.83715594e-01 7.68479705e-01 -8.21289480e-01
-9.30599034e-01 -9.47236478e-01 -2.11598530e-01 4.58080083e-01
1.24399684e-01 -1.09779465e+00 -4.84561235e-01 1.24618314e-01] | [9.371954917907715, 2.767833709716797] |
67ee3f0c-f1b9-417d-b65b-9e0497762004 | maser-multi-agent-reinforcement-learning-with | 2206.10607 | null | https://arxiv.org/abs/2206.10607v1 | https://arxiv.org/pdf/2206.10607v1.pdf | MASER: Multi-Agent Reinforcement Learning with Subgoals Generated from Experience Replay Buffer | In this paper, we consider cooperative multi-agent reinforcement learning (MARL) with sparse reward. To tackle this problem, we propose a novel method named MASER: MARL with subgoals generated from experience replay buffer. Under the widely-used assumption of centralized training with decentralized execution and consistent Q-value decomposition for MARL, MASER automatically generates proper subgoals for multiple agents from the experience replay buffer by considering both individual Q-value and total Q-value. Then, MASER designs individual intrinsic reward for each agent based on actionable representation relevant to Q-learning so that the agents reach their subgoals while maximizing the joint action value. Numerical results show that MASER significantly outperforms StarCraft II micromanagement benchmark compared to other state-of-the-art MARL algorithms. | ['Youngchul Sung', 'Whiyoung Jung', 'Woojun Kim', 'Jeewon Jeon'] | 2022-06-20 | null | null | null | null | ['starcraft-ii'] | ['playing-games'] | [-6.45452619e-01 2.62160063e-01 -4.57206100e-01 2.53157675e-01
-9.87696171e-01 -4.71344322e-01 4.21290725e-01 3.73652428e-01
-6.39158905e-01 1.42325032e+00 8.33027139e-02 1.98115796e-01
-5.25077164e-01 -7.51621068e-01 -5.56811035e-01 -8.76845479e-01
-4.32426631e-01 6.16651237e-01 1.63869411e-01 -5.64497352e-01
3.42285603e-01 8.78187940e-02 -1.47391224e+00 -6.96320906e-02
1.08691788e+00 8.57296467e-01 3.79802436e-01 7.34124660e-01
4.55350250e-01 1.51661170e+00 -9.40384030e-01 1.69990025e-02
4.38436687e-01 -6.30168736e-01 -7.11951911e-01 1.97623044e-01
-4.26045656e-01 -6.80526376e-01 -1.69485122e-01 9.51784015e-01
5.72075605e-01 5.97168028e-01 3.83964777e-01 -1.65297282e+00
-3.01284611e-01 1.04036009e+00 -7.21182942e-01 4.45941500e-02
4.50796366e-01 6.07565224e-01 1.11306870e+00 -2.35770881e-01
6.86688542e-01 1.26458216e+00 -2.70947954e-03 6.15081787e-01
-1.12442040e+00 -2.84470111e-01 2.83835948e-01 2.80640811e-01
-9.82696056e-01 2.79676057e-02 4.70259070e-01 3.32190320e-02
1.15427816e+00 -2.41718248e-01 1.02312338e+00 6.80894852e-01
7.32567310e-01 1.08111036e+00 1.11132443e+00 -7.52787441e-02
1.08135426e+00 -3.58106852e-01 -3.82751644e-01 8.00333142e-01
2.99670696e-01 4.52874005e-01 -4.78421837e-01 -3.65997940e-01
8.75827968e-01 -1.58199668e-01 3.65050852e-01 -6.32460892e-01
-1.31597805e+00 8.44205558e-01 3.25766444e-01 -1.51740331e-02
-1.25878108e+00 6.65723205e-01 3.46987784e-01 8.25275064e-01
-9.77987349e-02 7.15380967e-01 -3.53125393e-01 -3.19607407e-01
-3.80374342e-01 8.70195508e-01 7.08761632e-01 8.53840590e-01
8.51358294e-01 4.98099387e-01 -1.95252806e-01 4.02172178e-01
3.22777718e-01 4.61789459e-01 5.56559861e-01 -1.53042960e+00
5.01056790e-01 5.64562380e-01 5.74648082e-01 -4.26230580e-01
-5.33709288e-01 -3.73358876e-01 -3.16380352e-01 8.09293628e-01
1.06486604e-01 -8.44994307e-01 -3.74497473e-01 1.55851889e+00
4.82524693e-01 2.99490001e-02 9.98880029e-01 1.09752405e+00
4.11247075e-01 7.56440282e-01 1.28354520e-01 -6.71950877e-01
8.55065405e-01 -1.19006824e+00 -5.98685801e-01 -1.55160680e-01
6.73051715e-01 -1.65347889e-01 6.73454285e-01 7.68603802e-01
-1.44053650e+00 -2.89691329e-01 -1.07323766e+00 8.38759303e-01
3.28544587e-01 -7.26004168e-02 7.56815434e-01 3.03923368e-01
-8.73713374e-01 6.12657189e-01 -8.03864062e-01 1.09680835e-02
8.08918625e-02 4.50492501e-01 2.32173968e-02 2.22612604e-01
-8.65848184e-01 8.96487653e-01 7.14202583e-01 -3.30714643e-01
-1.99278700e+00 -1.72442824e-01 -6.03704929e-01 1.66244544e-02
1.18429136e+00 -6.10067129e-01 1.72576177e+00 -1.09388840e+00
-2.09326434e+00 4.82851677e-02 7.59959698e-01 -7.03782380e-01
3.25295806e-01 -4.15220171e-01 1.21846303e-01 2.98709393e-01
2.48445913e-01 5.56611061e-01 9.44928825e-01 -1.36070991e+00
-1.03488481e+00 7.70995021e-02 5.25567174e-01 9.60233390e-01
-1.00087054e-01 -2.46147275e-01 2.13120505e-01 -1.44196928e-01
-5.29993951e-01 -8.45680773e-01 -9.62752581e-01 -7.05923140e-01
1.87889010e-01 -6.38317347e-01 3.56541395e-01 -1.68000475e-01
8.48901868e-01 -1.70853555e+00 7.43592501e-01 9.98898596e-02
2.61548817e-01 -5.39919883e-02 -6.67640030e-01 8.93911719e-01
6.03211820e-01 -4.81749594e-01 2.74135977e-01 6.14963546e-02
1.05140164e-01 4.34903413e-01 -1.38470575e-01 3.91379267e-01
4.90694940e-02 7.84824908e-01 -1.57000411e+00 -5.07219017e-01
4.05657701e-02 -3.27392071e-01 -8.31251442e-01 7.66455889e-01
-7.45104134e-01 3.78347963e-01 -9.82727408e-01 6.68558657e-01
2.86437303e-01 1.20255165e-01 6.64134860e-01 5.02428412e-01
-1.95138931e-01 -2.37162411e-01 -1.37783241e+00 1.70930827e+00
-4.24646795e-01 -2.25379169e-01 2.21990764e-01 -8.83244455e-01
9.79058862e-01 4.09846723e-01 9.95728433e-01 -9.15441632e-01
2.02778041e-01 1.62950590e-01 2.09086612e-02 -4.02413309e-01
8.35042655e-01 -8.87905527e-03 -4.35371041e-01 6.41820490e-01
3.76759857e-01 -4.30452615e-01 4.66303885e-01 3.75048399e-01
1.28219700e+00 4.51582998e-01 7.09583938e-01 8.63835141e-02
3.01925361e-01 3.27038616e-01 9.10828471e-01 1.00917494e+00
-5.96002460e-01 -2.92757541e-01 8.37326527e-01 -2.20090687e-01
-8.41183066e-01 -1.29896450e+00 1.00440943e+00 1.22577798e+00
4.83291686e-01 -3.36467743e-01 -6.30146563e-01 -9.25857723e-01
5.47566302e-02 5.83579004e-01 -3.43935579e-01 -4.09576446e-02
-6.23448491e-01 -6.08967841e-01 1.01311527e-01 2.44897798e-01
5.36923289e-01 -1.63088846e+00 -1.43492293e+00 7.42014050e-01
1.08611107e-01 -8.00682783e-01 -3.12136054e-01 2.18054727e-01
-8.00630748e-01 -1.08833408e+00 -6.39685869e-01 -6.03861690e-01
4.25191700e-01 1.24353476e-01 9.53373134e-01 6.51205406e-02
-2.86393482e-02 7.37811565e-01 -7.51087487e-01 -2.03498706e-01
-4.39281374e-01 -8.78498927e-02 2.45167866e-01 -2.82500982e-01
-2.02283308e-01 -3.93575609e-01 -7.79184461e-01 1.71372861e-01
-8.02126050e-01 -6.95763528e-02 8.79506588e-01 9.30369318e-01
5.14258683e-01 2.63024658e-01 1.26971483e+00 -1.19817100e-01
1.06627953e+00 -6.60742342e-01 -1.01138401e+00 1.99121386e-01
-6.08765185e-01 1.42383486e-01 9.45256114e-01 -6.68303907e-01
-1.15234625e+00 3.28141376e-02 3.38637203e-01 -2.15139478e-01
1.63095474e-01 4.97777313e-01 1.20313779e-01 2.16887504e-01
7.14264989e-01 2.70076662e-01 3.86317104e-01 1.47765473e-01
6.02896452e-01 3.47784102e-01 1.81887642e-01 -9.67995346e-01
5.35462797e-01 1.19716302e-02 8.38362575e-02 -3.41622025e-01
-4.85052109e-01 -2.24932119e-01 7.19551221e-02 -6.42483592e-01
5.13974667e-01 -1.02790582e+00 -1.43988431e+00 3.59202355e-01
-8.17446053e-01 -6.80566251e-01 -6.55843556e-01 5.75694203e-01
-1.41537583e+00 1.27859980e-01 -6.16092384e-01 -1.18388140e+00
-3.46545249e-01 -1.15521562e+00 5.11048615e-01 6.05935574e-01
1.76046342e-01 -6.42625272e-01 5.43928683e-01 4.55308743e-02
2.77886808e-01 3.05810750e-01 3.06544751e-01 -4.05714035e-01
-8.12194109e-01 2.47849062e-01 3.38994533e-01 1.97043940e-01
-1.29055902e-01 -4.22363222e-01 -2.40359083e-01 -7.68291593e-01
-2.73787111e-01 -1.16986358e+00 1.84159189e-01 3.78293812e-01
2.71708161e-01 -6.89486444e-01 4.53506038e-02 -5.19178361e-02
1.67636907e+00 5.05713463e-01 4.88651186e-01 6.90828860e-01
5.78309856e-02 4.38699186e-01 1.42153823e+00 1.19276333e+00
4.69780147e-01 4.72650796e-01 1.10609078e+00 4.50061172e-01
3.49967629e-01 -2.33061746e-01 9.50132191e-01 4.56555784e-01
-2.21324608e-01 -1.28123358e-01 -2.88176715e-01 5.70495784e-01
-2.67483711e+00 -1.07103777e+00 4.90945280e-01 2.02506304e+00
8.18613172e-01 5.68734221e-02 6.78606868e-01 -2.65588820e-01
2.77419239e-01 1.65838990e-02 -8.89425397e-01 -4.44262087e-01
4.67544124e-02 -1.17149577e-01 5.09910882e-01 4.27381158e-01
-6.14758968e-01 1.22834063e+00 6.20953798e+00 9.22103167e-01
-4.57802862e-01 1.17677778e-01 2.74035484e-01 -3.30814719e-01
-4.86688539e-02 1.83927625e-01 -6.55304909e-01 1.88663229e-01
8.44182670e-01 -6.93988681e-01 9.12786603e-01 1.17324543e+00
6.39361262e-01 -6.80690408e-01 -7.80871332e-01 7.12250948e-01
-2.72369742e-01 -1.18089128e+00 -3.80778044e-01 -1.01156645e-01
1.08457696e+00 -3.93600166e-02 -1.98410943e-01 7.93656111e-01
1.19872403e+00 -6.90639019e-01 9.00564909e-01 4.10432994e-01
2.10777551e-01 -1.23189569e+00 7.01836467e-01 5.75886846e-01
-1.17794323e+00 -6.94020391e-01 -3.58709246e-01 -5.38775921e-01
1.83889687e-01 3.96320559e-02 -8.75500083e-01 7.13489413e-01
2.27080524e-01 5.61125517e-01 -4.93366532e-02 1.01785946e+00
-3.90782207e-01 3.40584964e-01 9.83444527e-02 -5.07642746e-01
7.39872873e-01 -3.90771955e-01 6.02071583e-01 4.30412620e-01
1.68378487e-01 1.16403580e-01 8.37613881e-01 6.43785179e-01
2.00837269e-01 2.33836338e-01 -4.75284040e-01 -1.29871264e-01
5.86773038e-01 1.52634454e+00 -6.33116603e-01 -4.00251657e-01
-6.49879202e-02 7.78335214e-01 5.80652535e-01 3.66626412e-01
-7.98100770e-01 -2.86567092e-01 7.34962940e-01 -4.49914396e-01
1.71171591e-01 -2.90165216e-01 3.70819092e-01 -8.13343942e-01
-3.14126641e-01 -1.10147953e+00 4.89585817e-01 -6.71975613e-01
-1.08437896e+00 4.28786606e-01 -3.89986336e-02 -1.41935563e+00
-7.77339637e-01 -1.57206506e-01 -6.51307404e-01 1.57746047e-01
-1.52035999e+00 -9.03788447e-01 2.75092758e-02 5.63144207e-01
6.50068402e-01 -7.62821734e-01 5.92150033e-01 -4.19595718e-01
-4.13122505e-01 1.58401921e-01 1.77676886e-01 -4.33924586e-01
3.89352888e-01 -1.40763581e+00 -2.44279988e-02 3.76416713e-01
-4.00658876e-01 -9.45488513e-02 8.44945908e-01 -7.77689457e-01
-1.72787654e+00 -8.67234886e-01 -1.88501284e-01 9.70805511e-02
8.94453406e-01 3.14838737e-01 -2.36461371e-01 3.90158057e-01
5.78555405e-01 -2.62039959e-01 8.60118419e-02 -2.82190144e-01
2.68733650e-01 -2.58834481e-01 -1.25599432e+00 7.82308102e-01
5.50270796e-01 1.69070393e-01 -5.82358062e-01 1.80641294e-01
9.04461622e-01 -4.86975729e-01 -9.28018689e-01 -8.39272663e-02
1.93114787e-01 -9.62563455e-01 6.79405570e-01 -6.87739313e-01
4.01368588e-01 -2.75693983e-01 -5.49548194e-02 -1.95906317e+00
-2.00744182e-01 -1.17484248e+00 -4.10999656e-01 6.58363402e-01
-1.15661576e-01 -4.67977971e-01 9.56934333e-01 2.47506592e-02
-1.76279560e-01 -7.45880246e-01 -1.07729650e+00 -1.07802892e+00
2.41038892e-02 2.15007827e-01 4.98659134e-01 4.32263136e-01
6.16920173e-01 4.08972591e-01 -7.55139768e-01 6.39206311e-03
1.06749618e+00 6.34377077e-02 9.31988299e-01 -6.02634907e-01
-7.94538617e-01 -1.99645221e-01 2.51797549e-02 -7.71246254e-01
2.60850728e-01 -4.39307988e-01 2.71037340e-01 -1.71853256e+00
2.83332050e-01 -3.75239432e-01 -2.18391031e-01 5.67955136e-01
-4.96481583e-02 -2.73395479e-01 6.51400089e-01 5.40807024e-02
-1.61939859e+00 8.94162416e-01 1.70271266e+00 -2.73544937e-02
-7.40849435e-01 -7.99644664e-02 -5.77173889e-01 4.54257607e-01
1.08220232e+00 -3.35105360e-01 -5.64435065e-01 1.94605812e-02
3.89496088e-01 1.07845736e+00 2.00382024e-01 -9.95978057e-01
7.26446584e-02 -1.06826401e+00 -2.01487139e-01 -4.30073380e-01
2.30603904e-01 -5.93334317e-01 -1.97777197e-01 1.01935863e+00
-3.85108262e-01 3.86202574e-01 -1.39586672e-01 7.42887855e-01
-4.59042415e-02 -4.58611906e-01 5.84692121e-01 -3.78453314e-01
-8.22927594e-01 4.27974574e-02 -1.08179748e+00 1.57118171e-01
1.63505840e+00 2.92358637e-01 -4.53420907e-01 -6.85773015e-01
-5.42974651e-01 9.59861755e-01 2.34364390e-01 2.26208419e-01
8.96540523e-01 -1.29408538e+00 -7.65615582e-01 -2.92433381e-01
-1.30693033e-01 -2.73631185e-01 3.43650639e-01 4.13431525e-01
-1.72342271e-01 -6.81191310e-02 -7.98071921e-01 -4.27282266e-02
-1.02304626e+00 5.89738727e-01 4.06418085e-01 -7.27262855e-01
-4.95815903e-01 1.00524507e-01 -2.98255503e-01 -2.74361372e-01
2.05652684e-01 -7.57157207e-02 -2.94592053e-01 -5.56224771e-02
4.35910732e-01 7.03385055e-01 -4.20786858e-01 -9.87982899e-02
-1.07448861e-01 -4.62892614e-02 4.72064540e-02 -5.27291417e-01
1.43066573e+00 -3.54776345e-02 1.62296593e-01 6.45411909e-02
3.25569212e-01 -3.17523420e-01 -1.80440521e+00 -1.35089979e-01
-5.45576140e-02 -2.75131494e-01 -9.55011025e-02 -8.59198689e-01
-9.26120639e-01 7.92457955e-04 4.04395521e-01 1.19108543e-01
9.30125296e-01 -1.36653543e-01 6.87012196e-01 8.80111337e-01
9.81348395e-01 -1.77186942e+00 1.17046046e+00 5.11822641e-01
8.42434466e-01 -1.03152299e+00 4.03526016e-02 3.34681243e-01
-1.34118140e+00 8.12902153e-01 1.13300848e+00 -6.04860127e-01
-4.40066494e-02 2.36565024e-01 8.29325337e-03 -7.62192309e-02
-1.34514403e+00 -4.44972277e-01 -5.60987294e-01 8.56621146e-01
-1.60711303e-01 2.17995465e-01 -5.45938492e-01 5.00166476e-01
1.23401038e-01 3.24476250e-02 1.02320719e+00 1.33220243e+00
-9.90276873e-01 -1.16511071e+00 -5.16999900e-01 3.12061638e-01
-1.10885993e-01 5.32035053e-01 5.04029691e-02 5.71259558e-01
-2.83846736e-01 1.13274205e+00 -2.07641900e-01 -1.64863348e-01
2.09329635e-01 -6.27514184e-01 7.43857861e-01 -5.08676529e-01
-9.82823610e-01 2.37227440e-01 1.01867788e-01 -8.19087982e-01
-3.88070703e-01 -4.84334409e-01 -1.84936082e+00 -1.18414015e-01
2.91428668e-03 5.42588532e-01 3.96613538e-01 8.91375601e-01
1.94487840e-01 7.17293262e-01 1.00894678e+00 -9.28930163e-01
-1.26582146e+00 -7.36560524e-01 -8.75159204e-01 2.76782423e-01
3.57831359e-01 -8.27717543e-01 1.90240052e-02 -4.61681277e-01] | [3.812676191329956, 2.023343086242676] |
fd4b5b16-2db9-4ad5-9528-6b0129f8df9f | malceiver-perceiver-with-hierarchical-and | 2204.05994 | null | https://arxiv.org/abs/2204.05994v1 | https://arxiv.org/pdf/2204.05994v1.pdf | Malceiver: Perceiver with Hierarchical and Multi-modal Features for Android Malware Detection | We propose the Malceiver, a hierarchical Perceiver model for Android malware detection that makes use of multi-modal features. The primary inputs are the opcode sequence and the requested permissions of a given Android APK file. To reach a malware classification decision the model combines hierarchical features extracted from the opcode sequence together with the requested permissions. The model's architecture is based on the Perceiver/PerceiverIO which allows for very long opcode sequences to be processed efficiently. Our proposed model can be easily extended to use multi-modal features. We show experimentally that this model outperforms a conventional CNN architecture for opcode sequence based malware detection. We then show that using additional modalities improves performance. Our proposed architecture opens new avenues for the use of Transformer-style networks in malware research. | ['Niall McLaughlin'] | 2022-04-12 | null | null | null | null | ['android-malware-detection'] | ['miscellaneous'] | [ 5.93239129e-01 -5.05296066e-02 -5.48372328e-01 -2.79783934e-01
-4.27275151e-01 -8.59469175e-01 7.01807737e-01 -4.44363281e-02
-2.48960614e-01 -6.16204105e-02 5.85029647e-02 -7.82617569e-01
2.16971815e-01 -6.40588284e-01 -6.94738388e-01 -3.86137187e-01
-1.68587655e-01 7.92034790e-02 3.88652980e-01 -2.50619203e-01
2.37090513e-01 3.81419837e-01 -1.69557774e+00 8.74376893e-01
2.29189590e-01 1.35989475e+00 1.69424921e-01 1.13633025e+00
1.98636159e-01 8.68174016e-01 -6.25391424e-01 -2.83737391e-01
2.45400339e-01 1.49016501e-02 -5.57247281e-01 -3.88190389e-01
3.94524574e-01 -6.84044957e-01 -2.88650423e-01 1.07906830e+00
1.42039865e-01 -4.71727937e-01 6.20442569e-01 -1.26285470e+00
-6.71866238e-01 4.40438151e-01 -2.65088290e-01 4.77712214e-01
7.26259530e-01 2.43743777e-01 9.78577733e-01 -4.50910538e-01
2.93273598e-01 1.40054500e+00 8.71058583e-01 6.52499616e-01
-8.28181386e-01 -3.12434554e-01 1.36327771e-02 5.05026877e-01
-9.54303384e-01 -4.86782223e-01 6.10289633e-01 -5.54872096e-01
1.28953588e+00 4.50412601e-01 6.85055614e-01 1.77354920e+00
6.30371392e-01 6.40792131e-01 1.18585527e+00 1.72804482e-02
2.89351493e-01 6.14873171e-02 3.52882206e-01 1.13950682e+00
-8.65832120e-02 2.41453335e-01 -3.31731200e-01 -4.43455577e-01
1.09946169e-01 4.71669346e-01 -6.46479279e-02 -1.33485585e-01
-8.83306205e-01 9.07675445e-01 5.59034884e-01 4.18999404e-01
-6.51326925e-02 3.57523799e-01 7.11130381e-01 3.86057466e-01
1.33813500e-01 3.89171600e-01 -5.15361607e-01 -5.94813406e-01
-9.18440819e-01 -3.37449133e-01 9.12651420e-01 5.33725679e-01
5.84194064e-01 -8.60006809e-02 1.28143445e-01 4.81466323e-01
6.72574878e-01 5.76278210e-01 6.03299022e-01 -1.14739430e+00
2.14956254e-01 7.01728463e-01 -3.83044451e-01 -1.09860873e+00
-1.17722206e-01 1.15456268e-01 -2.75671065e-01 2.11089253e-02
1.44667462e-01 1.67803273e-01 -8.71410012e-01 1.37925851e+00
-1.84271634e-02 2.23728687e-01 -4.60773408e-02 2.61497945e-01
7.13390768e-01 4.96701747e-01 -7.72621334e-02 -5.97183257e-02
1.67581677e+00 -5.82457900e-01 -3.63596708e-01 -1.26295671e-01
5.18667698e-01 -3.74966383e-01 1.18490064e+00 3.14496487e-01
-5.57039559e-01 -4.61859196e-01 -1.28008926e+00 -2.51050711e-01
-9.18523490e-01 2.07260668e-01 4.48512107e-01 1.17814052e+00
-1.34065211e+00 3.02511245e-01 -8.81535292e-01 -4.15039539e-01
6.19303346e-01 8.07027578e-01 -8.28986242e-02 2.67176479e-01
-9.05099094e-01 4.24879432e-01 2.11875990e-01 -6.00309297e-02
-1.66377878e+00 -2.71317393e-01 -9.37737584e-01 2.23044813e-01
3.40017617e-01 -3.85966808e-01 1.45103490e+00 -8.69878650e-01
-1.65667439e+00 6.44434869e-01 -3.53833646e-01 -5.69489300e-01
2.55508479e-02 -5.43457158e-02 -2.86828160e-01 4.07324106e-01
-2.18874335e-01 4.98625487e-01 1.59432924e+00 -1.14410412e+00
-4.00754780e-01 -6.23515546e-01 5.86839437e-01 -1.84584275e-01
-7.80197620e-01 1.03565075e-01 -1.56504735e-01 -1.18099116e-01
-8.00430477e-01 -9.99462843e-01 2.25554869e-01 -3.85047734e-01
-5.96198618e-01 -1.54407779e-02 1.50134790e+00 -6.22835159e-01
9.49490964e-01 -2.38572574e+00 4.92087901e-02 7.04572126e-02
8.64396274e-01 6.08293533e-01 -2.00155526e-01 3.69839519e-01
1.97161391e-01 3.98395300e-01 -8.64876360e-02 -4.27144885e-01
2.65352577e-02 1.72295108e-01 -6.99963748e-01 4.16807234e-01
8.04819539e-02 1.04099476e+00 -8.62649858e-01 -2.23415032e-01
1.86780080e-01 7.78579235e-01 -5.82366467e-01 2.53887922e-01
-4.64336932e-01 1.95310310e-01 -4.15233046e-01 7.94886827e-01
4.31754142e-01 -2.00390399e-01 2.31435418e-01 -1.35074839e-01
9.36216265e-02 5.99348545e-01 -7.78006539e-02 1.07603514e+00
-7.36333847e-01 9.18718874e-01 2.72605062e-01 -7.24256873e-01
5.31738937e-01 1.23899989e-01 2.58629203e-01 -3.29519868e-01
5.64989090e-01 7.42280483e-02 3.26652139e-01 -5.98344445e-01
4.75786328e-01 4.86376405e-01 -6.42625615e-02 3.15665036e-01
-1.32044386e-02 1.74620971e-01 -2.86214929e-02 2.41509810e-01
1.60588455e+00 -2.70826429e-01 4.18612868e-01 -1.42951265e-01
1.00779855e+00 -2.98856497e-01 1.96451306e-01 7.57801354e-01
-5.24629295e-01 -1.82390690e-01 9.58314717e-01 -3.68774980e-01
-6.68316782e-01 -1.11454701e+00 1.92792118e-01 1.56279492e+00
5.25128208e-02 -1.00753796e+00 -1.05435550e+00 -1.45359564e+00
-5.00791781e-02 2.47436374e-01 -9.35630202e-01 -2.56390184e-01
-5.95200062e-01 -2.71369398e-01 1.07275355e+00 4.48553532e-01
5.10403156e-01 -9.32177126e-01 -9.64483142e-01 -2.86729544e-01
2.19797924e-01 -1.18105698e+00 -4.16427672e-01 9.80497450e-02
-6.81016445e-01 -1.27370930e+00 -2.94741802e-02 -6.68177128e-01
4.23735172e-01 1.33692876e-01 6.89416587e-01 2.83448100e-01
-9.48621258e-02 9.66725588e-01 -3.68640035e-01 -2.85610080e-01
-7.67247081e-01 3.38904768e-01 1.92536294e-01 6.89342842e-02
6.01676285e-01 -5.61618745e-01 -3.08590382e-01 1.54408619e-01
-8.87955070e-01 -5.61416447e-01 3.03406119e-01 5.76115847e-01
1.38026208e-01 -9.32992324e-02 1.15095668e-01 -7.89881945e-01
8.56329620e-01 -5.95715404e-01 -4.38916475e-01 -3.90291214e-04
-3.03340346e-01 -2.27712214e-01 1.20328355e+00 -6.20646298e-01
-6.45882785e-01 9.08586383e-02 -5.20436227e-01 -4.43906903e-01
-2.77526528e-01 6.14092126e-02 -2.41629779e-01 -3.85627896e-01
6.16596699e-01 4.72160906e-01 1.09483860e-01 -2.71591604e-01
4.76992249e-01 1.17480576e+00 2.45571569e-01 -2.37877324e-01
7.10271358e-01 6.74238861e-01 -4.55847234e-02 -1.20467329e+00
-3.05972487e-01 -3.09967667e-01 -3.99333239e-01 -9.12723988e-02
9.62622344e-01 -6.13352060e-01 -1.65520930e+00 4.46557432e-01
-1.16672182e+00 -3.87938648e-01 5.91573603e-02 -1.61195412e-01
-6.30598426e-01 6.32643402e-01 -1.04965138e+00 -8.09838474e-01
-2.98847824e-01 -1.77859461e+00 1.64238918e+00 -1.78800581e-03
-1.18769042e-01 -1.10093260e+00 -7.88066313e-02 3.56896549e-01
5.88523269e-01 -1.92993805e-01 9.74426925e-01 -1.03587413e+00
-5.36894381e-01 3.46660689e-02 -2.54950255e-01 3.14093143e-01
1.00335643e-01 2.00106576e-01 -1.27412128e+00 -4.37496334e-01
3.86900634e-01 -4.36680675e-01 1.06431508e+00 -9.87979695e-02
1.52308881e+00 -7.55681992e-01 -6.79126084e-01 7.15009570e-01
1.06548142e+00 3.64380568e-01 4.85605448e-01 -5.11031076e-02
1.03437769e+00 3.99286002e-01 1.87312409e-01 1.70356214e-01
5.26988447e-01 6.70689762e-01 9.91731882e-01 3.46199036e-01
1.04280196e-01 -3.03590119e-01 1.14123130e+00 9.21157062e-01
3.30703914e-01 -2.18777284e-01 -8.67278457e-01 2.21230224e-01
-1.40225720e+00 -1.01747382e+00 1.84138313e-01 1.89807498e+00
4.38920528e-01 1.49871081e-01 3.76456439e-01 1.53770849e-01
3.31292212e-01 4.77809757e-01 -5.95801711e-01 -1.06379724e+00
4.56383377e-01 1.57931417e-01 4.80922997e-01 7.61332333e-01
-1.16943741e+00 8.02427530e-01 6.97105598e+00 8.24818492e-01
-1.35574341e+00 5.12312949e-01 4.20939296e-01 2.09865913e-01
-9.53258649e-02 -3.23912174e-01 -9.32553053e-01 8.10692310e-01
1.52520156e+00 5.33403456e-01 8.72868598e-01 9.95534182e-01
-1.30925000e-01 -4.87356111e-02 -1.37377501e+00 1.07515740e+00
3.43300432e-01 -1.14012015e+00 -1.38708472e-01 6.31928623e-01
1.24239691e-01 3.37014973e-01 5.89959681e-01 4.73354876e-01
1.88422263e-01 -1.00620186e+00 3.72357488e-01 9.52263027e-02
8.81405413e-01 -4.16970134e-01 3.37462902e-01 2.88396180e-01
-1.35569906e+00 -8.27673197e-01 -2.01513916e-01 -5.84495403e-02
-3.58644515e-01 2.15146586e-01 -1.39907026e+00 -2.49366206e-03
5.40894091e-01 1.10211623e+00 -9.87434268e-01 4.13210124e-01
-9.88122150e-02 8.07094932e-01 -2.05285609e-01 -2.56434947e-01
9.40624923e-02 2.63222814e-01 5.79860330e-01 1.29611325e+00
2.65127838e-01 -5.36907196e-01 1.24762259e-01 6.51549697e-01
-1.99938729e-01 -1.40630081e-01 -1.13158071e+00 -5.57413638e-01
3.65745604e-01 1.33544493e+00 -6.45991445e-01 -5.86190283e-01
-2.34617546e-01 1.09623408e+00 3.10700148e-01 8.97736400e-02
-8.72381866e-01 -2.36318707e-01 9.21189606e-01 -6.27786890e-02
5.38210809e-01 -3.54400069e-01 -3.00261099e-02 -1.09031045e+00
-2.49322534e-01 -1.32481599e+00 1.70975193e-01 -4.94555891e-01
-9.66257215e-01 7.33288527e-01 -2.03500971e-01 -9.23142612e-01
-5.61414659e-01 -1.33105099e+00 -4.89529133e-01 2.58100510e-01
-1.16149831e+00 -1.20928574e+00 -1.88905895e-01 6.61642075e-01
5.16495943e-01 -3.63989323e-01 8.59699488e-01 1.14406623e-01
-5.02152324e-01 5.65758169e-01 -1.11435913e-01 1.77636489e-01
1.49282873e-01 -1.39351475e+00 5.98437428e-01 5.78878164e-01
7.19854534e-02 1.12463760e+00 4.12070781e-01 -7.88370609e-01
-1.70499849e+00 -9.62346017e-01 5.28973281e-01 -1.01998007e+00
8.16130459e-01 -9.27148163e-01 -4.49764013e-01 8.18651795e-01
4.68172908e-01 -1.43664777e-01 9.33033884e-01 -1.82991728e-01
-1.05033934e+00 -3.09075862e-01 -1.28657532e+00 4.70804155e-01
9.16184843e-01 -1.22331488e+00 -3.49647045e-01 2.76217699e-01
8.72900069e-01 -1.69313580e-01 -8.39260995e-01 3.42589974e-01
7.62765706e-01 -1.16324317e+00 9.38156009e-01 -4.74615157e-01
2.44219437e-01 -3.36002916e-01 -2.95944899e-01 -9.88910317e-01
7.22021013e-02 -7.05279410e-01 -8.13559413e-01 6.59070492e-01
1.59192890e-01 -7.42270291e-01 4.46833849e-01 -2.55713254e-01
7.35569149e-02 -8.60434890e-01 -1.09334457e+00 -6.60649538e-01
-3.51357192e-01 -7.23990917e-01 6.01084292e-01 5.25405765e-01
1.56138241e-01 4.10134315e-01 -2.83992589e-01 1.96713924e-01
3.00121695e-01 -2.26020142e-01 4.78721946e-01 -1.13675940e+00
-7.19358981e-01 -4.30608064e-01 -6.34244025e-01 -1.29022253e+00
4.07715023e-01 -8.63750994e-01 -7.24516017e-03 -7.65023053e-01
1.30703509e-01 6.60103336e-02 -1.54556809e-02 6.66034281e-01
3.16073895e-01 7.12842703e-01 4.18593585e-01 1.16205700e-01
-7.23527610e-01 2.60540307e-01 7.27535784e-01 -4.24392521e-01
-1.19066782e-01 -2.88665388e-02 -5.62321365e-01 7.81004131e-01
9.83029127e-01 -2.60838121e-01 -5.42287827e-01 -1.66031003e-01
3.26943725e-01 -2.69590557e-01 4.43833172e-01 -9.08024669e-01
-1.14479557e-01 3.52483064e-01 2.97680736e-01 -1.98810518e-01
8.67383242e-01 -9.28437412e-01 -4.62831914e-01 7.94055521e-01
-3.52046162e-01 3.78892750e-01 1.04809120e-01 7.38922775e-01
2.56877124e-01 4.78121415e-02 5.16227067e-01 -1.09039694e-01
-4.35828924e-01 6.94505125e-02 -1.03569591e+00 -4.11275536e-01
1.06696248e+00 -1.74663559e-01 -7.67192960e-01 -3.59157741e-01
-5.44472337e-01 -3.15191090e-01 7.45371580e-01 6.71206772e-01
7.57360458e-01 -8.91378939e-01 1.51852667e-01 4.60613310e-01
1.00698590e-01 -9.72963870e-01 -2.97410607e-01 9.29109573e-01
-2.63747334e-01 6.33993804e-01 -2.40805969e-01 -7.67446816e-01
-1.64839852e+00 8.82575333e-01 3.75602156e-01 -2.35831350e-01
7.12300390e-02 5.29857159e-01 -7.07657486e-02 -4.79831845e-01
2.26059780e-01 -5.30035794e-01 -4.30436522e-01 -4.24487628e-02
9.24602866e-01 4.48121190e-01 7.16078142e-03 -1.14328527e+00
-6.88880265e-01 4.02359575e-01 -2.35159338e-01 1.56901538e-01
9.09747183e-01 2.14259103e-02 -5.18459678e-01 4.21248317e-01
1.53441262e+00 4.17875081e-01 -8.80536139e-01 3.42750549e-01
-1.44507051e-01 -3.31262767e-01 -2.34911039e-01 -7.08577216e-01
-9.00703907e-01 1.17400849e+00 8.87473583e-01 7.28746235e-01
1.11395645e+00 3.72357331e-02 8.66091669e-01 5.17511904e-01
3.95890474e-01 -4.59739000e-01 2.27769002e-01 6.99004173e-01
3.79340857e-01 -9.94989157e-01 -6.08598351e-01 -4.55534160e-01
-1.55866817e-01 1.18910491e+00 5.78919768e-01 -1.34064943e-01
6.37471139e-01 4.20968801e-01 -3.64992917e-02 -4.12004054e-01
-6.44335091e-01 -2.23960221e-01 2.80697435e-01 1.03693485e+00
2.34047860e-01 1.98410362e-01 -6.35995865e-02 1.94932967e-01
-2.03153104e-01 -2.25332141e-01 6.09817624e-01 6.71646714e-01
-3.94717723e-01 -1.38563240e+00 -3.22356164e-01 7.40391135e-01
-6.95263624e-01 -2.24139154e-01 -9.98597562e-01 2.35066935e-01
4.51549441e-01 1.23116529e+00 -7.65500963e-02 -1.14540696e+00
-2.96111971e-01 7.89215863e-02 4.03774917e-01 -7.21859694e-01
-8.73664737e-01 -5.02383053e-01 -8.55706930e-02 -1.00154209e+00
-1.12407826e-01 -4.27259624e-01 -8.40754092e-01 -3.12347770e-01
2.25663856e-01 -2.13850766e-01 7.00752914e-01 9.60525036e-01
7.63046384e-01 4.05056655e-01 6.64807260e-01 -9.00962055e-01
-5.30226767e-01 -7.44631886e-01 6.57548681e-02 -3.07545858e-03
1.03906572e+00 -5.08103430e-01 -2.02994600e-01 -7.88233578e-02] | [14.447157859802246, 9.70645523071289] |
4eb8d1bb-6137-409e-b70b-0fe77c9100d0 | proximal-gradient-based-unfolding-for-massive | 2212.01839 | null | https://arxiv.org/abs/2212.01839v1 | https://arxiv.org/pdf/2212.01839v1.pdf | Proximal Gradient-Based Unfolding for Massive Random Access in IoT Networks | Grant-free random access is an effective technology for enabling low-overhead and low-latency massive access, where joint activity detection and channel estimation (JADCE) is a critical issue. Although existing compressive sensing algorithms can be applied for JADCE, they usually fail to simultaneously harvest the following properties: effective sparsity inducing, fast convergence, robust to different pilot sequences, and adaptive to time-varying networks. To this end, we propose an unfolding framework for JADCE based on the proximal gradient method. Specifically, we formulate the JADCE problem as a group-row-sparse matrix recovery problem and leverage a minimax concave penalty rather than the widely-used $\ell_1$-norm to induce sparsity. We then develop a proximal gradient-based unfolding neural network that parameterizes the algorithmic iterations. To improve convergence rate, we incorporate momentum into the unfolding neural network, and prove the accelerated convergence theoretically. Based on the convergence analysis, we further develop an adaptive-tuning algorithm, which adjusts its parameters to different signal-to-noise ratio settings. Simulations show that the proposed unfolding neural network achieves better recovery performance, convergence rate, and adaptivity than current baselines. | ['Yonina C. Eldar', 'Xu Chen', 'Yong Zhou', 'Yinan Zou'] | 2022-12-04 | null | null | null | null | ['compressive-sensing', 'activity-detection'] | ['computer-vision', 'computer-vision'] | [ 2.75900543e-01 -4.99109030e-01 -5.67650378e-01 -1.80185914e-01
-7.80817389e-01 -1.03798181e-01 -3.21381278e-02 -5.74260771e-01
-2.39701286e-01 8.29951346e-01 4.33967113e-01 -7.06739008e-01
-2.20696628e-01 -3.44473302e-01 -6.08011425e-01 -7.16148913e-01
-6.26715899e-01 -2.52802700e-01 -4.08263594e-01 -1.92203000e-02
-5.66659123e-02 2.47267544e-01 -4.49362189e-01 -3.13382059e-01
8.77125919e-01 1.35348976e+00 1.44574046e-01 3.65090579e-01
5.24621487e-01 8.90715599e-01 -3.43406022e-01 1.68377861e-01
5.24478316e-01 -6.76443875e-01 -7.15250820e-02 8.38429779e-02
-9.68782753e-02 -5.80702007e-01 -7.47293234e-01 9.56383526e-01
8.66630614e-01 3.03180981e-03 2.65931934e-01 -9.90957797e-01
-4.14120674e-01 7.55460203e-01 -9.33203101e-01 3.42136681e-01
3.05170417e-01 -9.35183316e-02 9.30167198e-01 -1.24901688e+00
2.10688427e-01 8.66458595e-01 9.86666799e-01 5.03634699e-02
-1.06225491e+00 -1.11527336e+00 2.15968430e-01 1.18178144e-01
-1.88805270e+00 -7.83922136e-01 7.46520758e-01 1.07873879e-01
5.31643271e-01 1.84539676e-01 5.62183201e-01 1.00920677e+00
-1.17974266e-01 9.66045737e-01 9.04504597e-01 -4.60603058e-01
2.82874644e-01 -4.19312477e-01 -2.64665037e-01 7.48229861e-01
4.45767850e-01 -1.04040094e-01 -6.07873976e-01 -2.68441558e-01
1.03939533e+00 1.60142705e-01 -7.89282560e-01 -2.41143450e-01
-1.34206665e+00 6.72679961e-01 3.52335989e-01 2.25473270e-01
-7.58947968e-01 5.26637793e-01 3.96469384e-02 4.76327807e-01
8.35444629e-02 1.42247200e-01 1.03692994e-01 -9.78119150e-02
-1.21844983e+00 2.19199602e-02 9.13166225e-01 8.89333129e-01
4.86575991e-01 5.65387011e-01 -4.17333961e-01 8.90432715e-01
5.08515835e-01 7.62035429e-01 3.40902060e-01 -9.20083940e-01
7.76335359e-01 -1.47519499e-01 3.66023928e-02 -1.37660718e+00
-3.16989511e-01 -1.21661174e+00 -1.37547457e+00 -5.32781720e-01
-5.12511842e-02 -7.06108689e-01 -5.53053200e-01 1.79464579e+00
2.11356226e-02 8.27029645e-01 -2.24759132e-02 1.11224520e+00
6.16795160e-02 6.86143696e-01 -5.92103302e-01 -8.93121898e-01
7.71212041e-01 -9.84287798e-01 -1.11567807e+00 -3.62228781e-01
4.08780038e-01 -4.37449694e-01 6.39259040e-01 3.82546782e-01
-1.07678854e+00 -7.95752332e-02 -1.51809490e+00 4.35907453e-01
6.09337449e-01 4.00824934e-01 6.02839708e-01 7.48576403e-01
-7.34875202e-01 1.27450675e-01 -7.32778609e-01 4.18381728e-02
5.80665290e-01 4.52978939e-01 1.06014021e-01 -1.93878934e-01
-1.17790377e+00 3.01784128e-02 2.65815705e-01 3.79793614e-01
-9.15189326e-01 -4.61484849e-01 -5.44906795e-01 2.66624480e-01
7.15151966e-01 -6.00548625e-01 1.10861290e+00 -6.22813225e-01
-1.72032881e+00 2.33376220e-01 -3.40983242e-01 -8.94753873e-01
2.11594656e-01 -3.30830991e-01 -7.25217998e-01 1.29407629e-01
4.63093482e-02 -2.34714672e-01 1.13007438e+00 -1.07754052e+00
-4.27695990e-01 -5.75841963e-02 -2.12037772e-01 2.00150967e-01
-6.94869041e-01 -3.42790693e-01 -8.44521821e-01 -1.11077809e+00
5.66246867e-01 -8.61521840e-01 -5.77797890e-01 -2.66868740e-01
-5.70525646e-01 5.72446644e-01 6.39386773e-01 -6.63266718e-01
2.00614381e+00 -2.36511946e+00 -2.83038449e-02 9.09408152e-01
4.56492960e-01 1.65967315e-01 -4.26875241e-02 4.10135955e-01
2.50351399e-01 -2.08571717e-01 -2.83253700e-01 -2.56569475e-01
-1.03163265e-01 4.02837470e-02 -4.12802607e-01 5.65979481e-01
-3.96677256e-01 7.31224239e-01 -8.00150096e-01 -2.01587081e-02
-2.02174902e-01 2.33094543e-01 -6.72650158e-01 1.51719660e-01
1.77823022e-01 3.57009917e-01 -6.10201001e-01 8.12131703e-01
6.30566716e-01 -7.42604613e-01 5.78237355e-01 -3.12110752e-01
9.46458578e-02 -2.49873269e-02 -1.49050486e+00 1.87569296e+00
-4.93668795e-01 4.47236240e-01 7.05606163e-01 -1.30586576e+00
7.76476324e-01 3.31218600e-01 4.95461047e-01 -9.69004393e-01
2.16489583e-01 6.86077952e-01 -1.06746502e-01 -2.20123380e-01
3.53871584e-01 -5.34891002e-02 -9.25049931e-02 7.10908115e-01
-3.00661385e-01 4.96735066e-01 -1.61075175e-01 4.09989655e-01
1.16108978e+00 -4.02155250e-01 5.28366089e-01 -1.05478369e-01
3.89912784e-01 -7.47902572e-01 8.55770648e-01 1.11542833e+00
1.52940089e-02 3.56845707e-01 1.79758921e-01 -6.78072944e-02
-6.31446004e-01 -7.39349604e-01 3.17991935e-02 1.06444383e+00
3.90157044e-01 -3.35324526e-01 -5.20693302e-01 -1.81152999e-01
-9.50785875e-02 2.80629486e-01 -1.43278763e-01 -1.61362365e-01
-5.30400455e-01 -1.04927552e+00 6.28429234e-01 2.96369910e-01
9.15185750e-01 -2.82710344e-01 -2.99908817e-01 6.07751667e-01
-3.96809250e-01 -1.17965734e+00 -8.03100705e-01 2.18704596e-01
-5.78199983e-01 -4.80140060e-01 -9.28878367e-01 -6.44053280e-01
5.27043700e-01 8.11856449e-01 7.09325135e-01 1.51927009e-01
1.89244986e-01 4.23863262e-01 -3.22904825e-01 2.60363724e-02
1.55807614e-01 2.33412206e-01 3.68741751e-01 5.44645727e-01
2.37548798e-02 -1.16708541e+00 -8.44345868e-01 3.00480455e-01
-7.22337782e-01 -1.16248548e-01 9.68464851e-01 1.01298809e+00
6.38058424e-01 -2.67652929e-01 8.06541860e-01 -7.26171494e-01
1.02915049e+00 -8.40587914e-01 -3.59640777e-01 1.26891628e-01
-7.40655184e-01 1.55931832e-02 6.62645519e-01 -3.24456453e-01
-6.65746510e-01 1.57005768e-02 -2.14593783e-01 -6.76202238e-01
6.58954024e-01 1.09709346e+00 -1.75163388e-01 -5.08620858e-01
6.24183118e-01 5.06070852e-01 -1.19320862e-01 -1.30847663e-01
3.61162633e-01 9.77883756e-01 6.11032844e-01 -5.69798350e-01
1.04916394e+00 5.87696254e-01 -2.18894258e-01 -7.45029449e-01
-1.08185875e+00 -4.12709981e-01 1.12710536e-01 4.84340303e-02
-7.67163327e-03 -1.53360426e+00 -5.08751214e-01 2.12795496e-01
-6.18126571e-01 -4.66836125e-01 7.85471350e-02 6.80255115e-01
-3.49965453e-01 5.71221709e-01 -5.84015191e-01 -7.90221274e-01
-5.55318236e-01 -7.97673881e-01 6.28971815e-01 -3.51876691e-02
1.36253953e-01 -6.79476023e-01 -9.00534168e-02 4.97175343e-02
8.05562496e-01 1.97884202e-01 4.28663164e-01 -1.65212393e-01
-7.06795335e-01 -2.31902927e-01 -3.38562071e-01 1.47415951e-01
4.75065447e-02 -7.57309914e-01 -5.99699080e-01 -8.22382867e-01
1.38034776e-01 -2.01381385e-01 8.00769806e-01 7.02333629e-01
1.37787747e+00 -7.51697421e-01 -5.29241502e-01 1.28420281e+00
1.35609388e+00 9.93245915e-02 6.95701241e-01 1.93716601e-01
6.43574476e-01 -3.20806026e-01 3.98858964e-01 1.11517811e+00
3.88182342e-01 6.84635162e-01 3.99782687e-01 -1.41443551e-01
3.08476686e-01 -2.84870774e-01 3.75720203e-01 1.33164048e+00
5.57759218e-02 -1.30818337e-01 -4.52618718e-01 4.62015778e-01
-2.00264359e+00 -1.05695391e+00 1.19078562e-01 2.37657475e+00
7.97149360e-01 1.74480766e-01 1.90006435e-01 2.33757257e-01
5.65179050e-01 4.09333408e-01 -6.39384508e-01 2.36232549e-01
-2.39932910e-01 9.38881636e-02 8.26774657e-01 4.18772697e-01
-9.76904094e-01 5.39826453e-01 6.15311384e+00 1.10873902e+00
-1.41753125e+00 2.45768204e-01 4.60145622e-01 -2.59964138e-01
-3.50775301e-01 -5.54868057e-02 -3.98201853e-01 5.58488786e-01
7.64357746e-01 -1.61335826e-01 8.18917692e-01 6.22253239e-01
5.45381606e-01 2.16754362e-01 -7.19459832e-01 1.47899413e+00
6.50968403e-02 -1.53112316e+00 -3.12700927e-01 1.95643187e-01
8.99848640e-01 6.62088916e-02 9.73243192e-02 3.74836028e-01
1.57448918e-01 -9.80376720e-01 6.17928505e-01 4.10500228e-01
1.00592661e+00 -6.17839634e-01 5.56045234e-01 3.29231173e-01
-1.37710905e+00 -5.83778799e-01 -1.37780011e-01 -2.68455803e-01
5.65701604e-01 1.09203660e+00 -4.75965023e-01 5.23409605e-01
3.44627082e-01 8.39869201e-01 5.90524785e-02 1.17807353e+00
-3.10228974e-01 9.86518145e-01 -5.50359845e-01 -3.85698266e-02
2.25300163e-01 -1.61658257e-01 6.29807353e-01 1.16700935e+00
8.43454659e-01 2.73487180e-01 6.20186090e-01 5.05020022e-01
-4.75673884e-01 -3.89325321e-02 -4.04737562e-01 4.11443673e-02
1.26444566e+00 9.25209284e-01 -1.59815162e-01 -9.86162424e-02
-4.31989521e-01 9.27954853e-01 4.45043705e-02 8.64314556e-01
-7.14614928e-01 -5.80997884e-01 3.19754392e-01 1.17995059e-02
5.22528529e-01 -5.71642697e-01 -2.63245046e-01 -1.31803501e+00
2.94874877e-01 -1.17936397e+00 9.75847989e-02 -2.97909707e-01
-9.77798700e-01 2.22613066e-01 -4.94258463e-01 -1.48126709e+00
-2.04908699e-01 -1.20284609e-01 -6.12407982e-01 6.18806183e-01
-1.61760044e+00 -7.62627065e-01 -2.52239019e-01 8.99551868e-01
7.65690804e-02 -3.50587785e-01 5.72777331e-01 7.35362768e-01
-8.22581768e-01 1.05235863e+00 5.25829852e-01 1.62671804e-01
3.98153067e-01 -7.49452472e-01 -1.32934764e-01 1.31885743e+00
1.31425736e-02 8.93762529e-01 4.68341947e-01 -3.17094237e-01
-1.82720673e+00 -1.00371206e+00 2.58846909e-01 6.27598882e-01
7.14154005e-01 -3.11245114e-01 -6.31198585e-01 6.90576911e-01
1.90867502e-02 2.99131215e-01 8.63458037e-01 1.86425790e-01
-2.71069348e-01 -3.58020484e-01 -6.66882157e-01 6.87703729e-01
1.08721113e+00 -7.19492018e-01 2.14769259e-01 2.63978928e-01
4.91632015e-01 -5.87148070e-01 -7.41675794e-01 3.97507101e-01
6.62419975e-01 -8.14250648e-01 9.62474406e-01 -7.26435035e-02
-5.25757708e-02 -4.20288652e-01 -4.70137239e-01 -1.01439333e+00
-5.39099157e-01 -1.41908228e+00 -8.66058230e-01 6.15628839e-01
4.62616712e-01 -5.12588799e-01 8.31201911e-01 -1.67954624e-01
-1.99571311e-01 -9.65806305e-01 -1.10353565e+00 -7.40445614e-01
-5.55569291e-01 -6.84871972e-01 4.04374391e-01 1.08874214e+00
1.22291252e-01 5.72595596e-01 -1.24714100e+00 4.45065260e-01
6.80527687e-01 7.47476518e-02 9.17551219e-01 -8.25564206e-01
-9.19196546e-01 -2.27769911e-01 -2.09167469e-02 -2.01567197e+00
-2.88022637e-01 -7.24843383e-01 5.02392910e-02 -1.24060965e+00
1.27994074e-02 -8.28556180e-01 -6.79879189e-01 4.24056828e-01
-2.55612910e-01 1.27169952e-01 1.45071566e-01 5.70993662e-01
-9.70923543e-01 7.49671876e-01 9.90390182e-01 1.25749297e-02
-4.13642019e-01 2.14282915e-01 -1.02121222e+00 3.80794168e-01
6.88129604e-01 -2.70365745e-01 -3.38577509e-01 -4.56679553e-01
4.73513216e-01 4.39828068e-01 6.17542164e-03 -1.33029020e+00
3.56899083e-01 -2.51835585e-02 -2.72366907e-02 -1.82111278e-01
3.01954985e-01 -7.66765535e-01 6.21080995e-02 5.57666302e-01
-1.32713750e-01 -3.09702516e-01 -3.18587184e-01 9.47700322e-01
-2.22495154e-01 4.53136042e-02 3.67434382e-01 3.30972165e-01
-5.45935154e-01 6.76388085e-01 -4.59449172e-01 1.02411903e-01
6.42402232e-01 -2.88108289e-01 1.85496598e-01 -1.00315809e+00
-4.19682622e-01 4.43479806e-01 9.66435764e-03 -1.65314585e-01
7.04133213e-01 -1.62704861e+00 -7.33443975e-01 1.41625181e-01
-1.50882766e-01 -2.22236037e-01 2.08665937e-01 1.35241640e+00
-2.23011449e-01 3.44701290e-01 3.12341928e-01 -5.04277766e-01
-6.94591939e-01 7.49811009e-02 3.15534353e-01 -3.94691646e-01
-3.44749361e-01 8.48556936e-01 -2.88761288e-01 -1.75612494e-01
5.76446414e-01 7.20480755e-02 1.78243995e-01 -3.34724396e-01
6.17044330e-01 2.61315346e-01 -1.75783500e-01 -1.11039765e-01
-1.09800413e-01 2.39552304e-01 -1.05757095e-01 -1.38377056e-01
1.10404348e+00 -3.98369253e-01 6.10586144e-02 1.02111988e-01
1.16764271e+00 6.10743582e-01 -1.23096156e+00 -7.85272598e-01
-2.04761952e-01 -5.40953457e-01 3.32841039e-01 -3.10402572e-01
-1.17834294e+00 5.75165689e-01 6.69634283e-01 1.78901583e-01
1.40703762e+00 -6.21830702e-01 1.12731123e+00 6.79462373e-01
6.52275503e-01 -1.15608239e+00 2.30636790e-01 5.97927690e-01
5.54221928e-01 -8.18325758e-01 2.98855931e-01 -2.55540669e-01
-2.45819971e-01 9.02807891e-01 9.13636461e-02 2.69333925e-02
6.44302487e-01 2.36306086e-01 -1.38199270e-01 -2.15107854e-02
-3.76841933e-01 5.08458540e-02 -4.62965146e-02 4.36493576e-01
2.97369182e-01 2.27553278e-01 -5.05932271e-01 7.05133498e-01
1.00856483e-01 3.27866435e-01 5.36722839e-01 9.30013180e-01
-5.73012710e-01 -8.80206227e-01 -3.83582324e-01 7.99713612e-01
-5.84391177e-01 -3.50113302e-01 2.63646424e-01 2.94837117e-01
-2.82339275e-01 8.16776574e-01 -4.54699516e-01 -5.62105417e-01
-1.05582196e-02 -6.17670476e-01 1.63474217e-01 -3.16367418e-01
5.70383780e-02 3.99762958e-01 1.21693201e-01 -6.18749201e-01
-4.50991094e-01 -4.57902849e-01 -1.07167041e+00 -3.27306688e-01
-5.06322443e-01 3.14039528e-01 2.39075154e-01 1.11487114e+00
6.77510679e-01 5.61519504e-01 1.21646142e+00 -7.15562403e-01
-7.41233349e-01 -6.93261862e-01 -7.75632501e-01 -1.46182850e-01
4.64004397e-01 -2.99129725e-01 -2.28121012e-01 -3.43359351e-01] | [6.3526387214660645, 1.4119846820831299] |
f8e58509-8f1a-4133-97ad-3cfcc1eb97d4 | fine-grained-off-road-semantic-segmentation | 2103.03651 | null | https://arxiv.org/abs/2103.03651v1 | https://arxiv.org/pdf/2103.03651v1.pdf | Fine-Grained Off-Road Semantic Segmentation and Mapping via Contrastive Learning | Road detection or traversability analysis has been a key technique for a mobile robot to traverse complex off-road scenes. The problem has been mainly formulated in early works as a binary classification one, e.g. associating pixels with road or non-road labels. Whereas understanding scenes with fine-grained labels are needed for off-road robots, as scenes are very diverse, and the various mechanical performance of off-road robots may lead to different definitions of safe regions to traverse. How to define and annotate fine-grained labels to achieve meaningful scene understanding for a robot to traverse off-road is still an open question. This research proposes a contrastive learning based method. With a set of human-annotated anchor patches, a feature representation is learned to discriminate regions with different traversability, a method of fine-grained semantic segmentation and mapping is subsequently developed for off-road scene understanding. Experiments are conducted on a dataset of three driving segments that represent very diverse off-road scenes. An anchor accuracy of 89.8% is achieved by evaluating the matching with human-annotated image patches in cross-scene validation. Examined by associated 3D LiDAR data, the fine-grained segments of visual images are demonstrated to have different levels of toughness and terrain elevation, which represents their semantical meaningfulness. The resultant maps contain both fine-grained labels and confidence values, providing rich information to support a robot traversing complex off-road scenes. | ['Huijing Zhao', 'Xijun Zhao', 'Shaochi Hu', 'Biao Gao'] | 2021-03-05 | null | null | null | null | ['road-scene-understanding'] | ['computer-vision'] | [ 4.11001205e-01 2.25265607e-01 -3.75107974e-01 -8.41839015e-01
-5.78401327e-01 -4.10721570e-01 3.79751593e-01 1.51077732e-01
-2.83797592e-01 6.29412234e-01 -3.32266808e-01 -3.85741115e-01
-4.49953824e-01 -1.21630883e+00 -6.69381678e-01 -3.70769292e-01
2.83699972e-03 7.54243195e-01 5.76348066e-01 -4.55645382e-01
6.44282222e-01 6.43043995e-01 -2.11678362e+00 1.22886039e-01
1.35925651e+00 1.10948849e+00 5.88859618e-01 2.36384556e-01
-3.93804610e-01 1.24898233e-01 -2.37773776e-01 -7.67123699e-02
4.49365765e-01 1.10990308e-01 -7.64487386e-01 1.09118573e-01
5.64698219e-01 1.30077600e-01 1.82135552e-01 1.20724666e+00
-8.71176347e-02 9.61433128e-02 1.19537187e+00 -1.31210518e+00
-2.48948023e-01 7.93834552e-02 -5.59975863e-01 -1.77715808e-01
1.26421481e-01 -1.74245209e-01 7.79742241e-01 -6.84427440e-01
2.10498825e-01 1.37457001e+00 5.76445341e-01 -1.45829186e-01
-9.64691043e-01 -8.29836905e-01 2.67720968e-01 5.32683790e-01
-1.73064077e+00 -6.83985502e-02 6.43197298e-01 -8.47215056e-01
4.45324510e-01 2.01520860e-01 5.11300743e-01 3.41206342e-01
5.11524439e-01 1.64710775e-01 1.45404387e+00 -1.24703750e-01
1.82966352e-01 3.12284321e-01 2.12167844e-01 6.93443000e-01
5.87165117e-01 1.47051454e-01 -2.28948608e-01 5.46666741e-01
5.74534059e-01 -2.15003163e-01 7.42924288e-02 -6.04764104e-01
-8.22879791e-01 8.70483696e-01 8.38889003e-01 -5.73030747e-02
-2.86736220e-01 -2.69333780e-01 4.56366956e-01 8.22302401e-02
1.49712995e-01 2.52032310e-01 -3.48895252e-01 5.81188835e-02
-5.51498175e-01 1.08720414e-01 5.25702238e-01 1.07179797e+00
1.52092767e+00 -1.27215916e-02 2.35483482e-01 8.91317487e-01
2.47356147e-01 7.10773766e-01 1.03650242e-01 -9.06823218e-01
4.99160290e-01 8.64619792e-01 7.94226900e-02 -1.35480475e+00
-5.11187315e-01 -2.26647332e-01 -7.04718709e-01 7.24462211e-01
2.85392284e-01 2.75442153e-01 -1.30466616e+00 1.16155779e+00
2.83940792e-01 -2.61363655e-01 1.06863923e-01 9.13776159e-01
6.13925219e-01 5.35105765e-01 1.73214510e-01 3.83598179e-01
1.44636774e+00 -8.45907748e-01 -4.43146944e-01 -6.58418298e-01
5.21707594e-01 -4.54106629e-01 1.27793610e+00 2.90535212e-01
-2.86045134e-01 -9.49813604e-01 -1.25610459e+00 -2.53361501e-02
-8.21285367e-01 1.48134232e-01 4.98921603e-01 6.67276025e-01
-6.78903401e-01 2.67858773e-01 -2.39833847e-01 -4.57143813e-01
5.51750481e-01 7.94883072e-02 -4.39126998e-01 -2.07121000e-01
-1.31069362e+00 1.28425360e+00 7.46179760e-01 2.10995272e-01
-9.09262061e-01 -2.91561574e-01 -1.09122682e+00 -3.64948452e-01
3.04907590e-01 -2.45769620e-01 7.30567515e-01 -7.70609975e-01
-1.03453660e+00 1.18800521e+00 -6.86972067e-02 -2.16591343e-01
4.36368197e-01 -1.42762154e-01 -6.29419386e-01 1.33941755e-01
8.85263026e-01 6.89917028e-01 7.36193657e-01 -1.61478603e+00
-1.36681366e+00 -4.68566835e-01 2.07965448e-01 5.98625660e-01
2.92573303e-01 -5.31329632e-01 -2.75916338e-01 -1.99940056e-01
5.26974201e-01 -9.72835243e-01 -2.76431441e-01 -9.70997661e-02
-5.20244718e-01 -2.08412305e-01 9.20566142e-01 -4.14090633e-01
8.23210835e-01 -1.91715109e+00 -3.41425538e-01 5.14817357e-01
-1.62818804e-01 -1.61921695e-01 2.28753947e-02 5.40708564e-02
2.41931856e-01 2.70477921e-01 -6.36209905e-01 5.34657896e-01
-3.82499732e-02 3.62435669e-01 -1.61018506e-01 3.00887793e-01
3.01625192e-01 5.72825193e-01 -7.36689985e-01 -7.31670260e-01
4.26648259e-01 1.63615331e-01 4.68634702e-02 1.33327559e-01
6.03534505e-02 5.14583707e-01 -9.96981144e-01 7.87179410e-01
8.91388476e-01 4.07404035e-01 -1.82324514e-01 -4.18266803e-01
-3.33317339e-01 2.45844405e-02 -1.24171185e+00 1.26405883e+00
-6.14660978e-01 7.53580928e-01 -9.60149840e-02 -1.16185343e+00
1.58349323e+00 -4.57668394e-01 -3.52804922e-02 -1.08579779e+00
7.07737878e-02 5.57463646e-01 -2.69223481e-01 -7.26141989e-01
8.98565173e-01 -8.09464008e-02 -5.89654446e-01 -1.29908055e-01
-8.47452879e-01 -8.52156937e-01 -2.42059361e-02 -3.19844544e-01
5.13683259e-01 3.98265213e-01 6.26576170e-02 -5.65757632e-01
5.94021022e-01 7.80038595e-01 4.50766623e-01 6.74043715e-01
-7.82796368e-02 4.87148315e-01 2.30414048e-02 -4.64746296e-01
-1.12988222e+00 -1.22627389e+00 -6.78904414e-01 8.15430164e-01
1.17024922e+00 2.85668522e-01 -7.55144477e-01 -3.43889207e-01
-7.30090290e-02 5.73725700e-01 -5.55453360e-01 -1.01422332e-01
-4.79359001e-01 -4.12762463e-01 4.88711923e-01 4.81539130e-01
1.04148078e+00 -1.06548226e+00 -1.03520966e+00 4.96447012e-02
-4.02193636e-01 -1.08454072e+00 1.24336243e-01 4.00759995e-01
-5.01765430e-01 -1.22860098e+00 -1.49256736e-01 -1.02008247e+00
7.67163455e-01 7.24750161e-01 1.04838610e+00 -1.70220152e-01
-3.52446049e-01 -3.98887061e-02 -4.66945648e-01 -3.88549834e-01
-2.49270782e-01 1.23421542e-01 -4.04826663e-02 5.22628911e-02
2.68396527e-01 -2.69470841e-01 -5.33107162e-01 1.13472986e+00
-5.48060000e-01 2.17395410e-01 8.53952229e-01 5.86359859e-01
1.03347373e+00 6.73250616e-01 4.95815992e-01 -8.50643635e-01
1.82182804e-01 -6.29239678e-01 -6.14339411e-01 1.31297067e-01
-5.34405112e-01 -1.66737381e-02 4.72676724e-01 -1.06585780e-02
-1.19666123e+00 6.83247531e-03 6.56660944e-02 2.07559600e-01
-6.61556005e-01 4.04680222e-01 -5.36266148e-01 -1.11893177e-01
7.62247145e-01 -4.27934565e-02 -2.94198483e-01 -1.64220147e-02
5.34580469e-01 8.89900744e-01 8.07604969e-01 -7.02010930e-01
1.04972982e+00 6.07327640e-01 5.78013472e-02 -9.31188226e-01
-7.93929160e-01 -5.82888067e-01 -8.63221228e-01 -3.57064456e-01
1.13013208e+00 -9.47385430e-01 -2.99080819e-01 5.47193646e-01
-6.04821622e-01 -4.43746060e-01 -1.38711631e-01 4.35418487e-01
-7.36721635e-01 9.02719945e-02 1.13959365e-01 -6.37372971e-01
5.38010746e-02 -1.17155576e+00 1.04828429e+00 4.82004881e-01
-1.94914162e-01 -5.81506908e-01 -3.56412202e-01 7.03563273e-01
2.00335726e-01 3.34085345e-01 1.06827998e+00 -2.13760570e-01
-7.32363760e-01 -1.50808126e-01 -7.46850848e-01 7.41740987e-02
2.45629936e-01 -1.06612720e-01 -1.01140761e+00 2.15568900e-01
-3.49227995e-01 -3.71066451e-01 7.62184501e-01 3.30390602e-01
1.13507593e+00 9.94950756e-02 -5.90253294e-01 5.16088247e-01
1.31323981e+00 5.23222744e-01 8.46198738e-01 9.55034673e-01
8.90377164e-01 1.31232047e+00 1.75712001e+00 3.53313312e-02
6.55992031e-01 6.53695107e-01 7.89380848e-01 -2.22327173e-01
1.50117189e-01 -3.96210998e-01 -3.26635018e-02 2.09074020e-01
1.90553479e-02 -6.89150542e-02 -1.00463510e+00 6.67712510e-01
-1.63298714e+00 -7.32899606e-01 -6.19619071e-01 2.03196454e+00
4.64828014e-01 5.35300910e-01 -8.92940238e-02 2.31327042e-01
1.01153922e+00 8.74983445e-02 -7.99540579e-01 -6.79893255e-01
-1.58853665e-01 -4.09055129e-02 9.78927672e-01 5.06936431e-01
-1.24552608e+00 1.19259465e+00 5.31248856e+00 1.24406588e+00
-1.06366074e+00 -2.27924749e-01 6.63590968e-01 9.61655736e-01
-4.79553431e-01 1.03823513e-01 -1.05641031e+00 1.44880399e-01
7.06281245e-01 1.63718015e-01 -2.53873952e-02 9.52248573e-01
4.10246342e-01 -5.62036574e-01 -6.28644228e-01 9.34834719e-01
-2.90497601e-01 -9.58668828e-01 7.88957924e-02 -4.42601554e-02
7.52342582e-01 -6.12550750e-02 -4.45367768e-02 2.46366858e-01
3.74678791e-01 -1.35409141e+00 1.09210372e+00 4.79795188e-01
1.20590961e+00 -6.79752648e-01 6.99689448e-01 3.48893970e-01
-1.81377494e+00 -1.05130769e-01 -4.99778450e-01 -7.58142620e-02
1.99928463e-01 5.35179973e-01 -9.18308496e-01 8.63765121e-01
1.00936568e+00 7.49869883e-01 -5.48459589e-01 8.70335758e-01
-3.41298461e-01 1.84815213e-01 -2.55666852e-01 9.95644554e-02
3.18006366e-01 -6.57237828e-01 1.07565507e-01 1.10502946e+00
3.92323375e-01 -8.89341533e-02 3.84917229e-01 7.35122323e-01
4.33803916e-01 6.15277439e-02 -7.99605608e-01 3.83419633e-01
7.22867072e-01 1.30541873e+00 -1.02248323e+00 -2.89822608e-01
-6.88045770e-02 6.23204589e-01 1.06806785e-01 2.75268346e-01
-9.17958617e-01 -8.13984752e-01 5.82386076e-01 5.93225121e-01
-3.48443203e-02 -3.20604593e-01 -7.47974873e-01 -3.59498203e-01
1.05932400e-01 -3.56986225e-01 1.56405568e-01 -9.05884624e-01
-9.97434020e-01 6.78808928e-01 2.24850640e-01 -1.56608057e+00
1.98466524e-01 -6.07373714e-01 -4.56284046e-01 8.95547509e-01
-2.02602577e+00 -1.24905014e+00 -1.13417053e+00 3.63845885e-01
7.38543510e-01 1.10924080e-01 5.25948107e-01 2.12612092e-01
-3.26880366e-01 1.59490198e-01 -5.62408045e-02 -9.62831229e-02
4.51504678e-01 -1.08482444e+00 1.87363297e-01 6.05638266e-01
-4.12693620e-01 3.24452557e-02 6.44479573e-01 -8.29672098e-01
-7.75716186e-01 -1.56907368e+00 3.85634959e-01 -8.13140869e-02
3.50001872e-01 -2.21695588e-03 -1.03939664e+00 2.09236979e-01
-4.76873010e-01 -2.47364238e-01 2.23620191e-01 -1.33333370e-01
-2.63753772e-01 -3.87157857e-01 -1.23615873e+00 5.05487561e-01
1.29326367e+00 -4.88699764e-01 -6.64047599e-01 -2.05817912e-02
4.75114524e-01 -1.29061222e-01 -7.66795814e-01 9.48204637e-01
6.62605107e-01 -9.48444128e-01 9.21189964e-01 8.00515413e-02
3.08050990e-01 -8.74863744e-01 -2.84387052e-01 -1.12149739e+00
-2.74868190e-01 1.81269243e-01 9.99152958e-01 9.97580707e-01
3.83044988e-01 -8.37072074e-01 9.01064813e-01 2.73302406e-01
-7.25673139e-01 -4.67588931e-01 -7.68616676e-01 -9.53489780e-01
1.63263548e-02 -6.15847290e-01 4.52766120e-01 7.70196736e-01
-5.15340745e-01 2.30158448e-01 1.35007277e-01 4.47476238e-01
5.76959133e-01 3.65739673e-01 1.02254057e+00 -1.62659562e+00
5.93770146e-01 -4.91642833e-01 -6.96292937e-01 -1.02806115e+00
2.93887675e-01 -8.43578577e-01 5.94062746e-01 -1.93966866e+00
-9.37995911e-02 -1.33746898e+00 1.76548231e-02 4.14870501e-01
5.71864359e-02 5.02631247e-01 -5.15941739e-01 3.69114697e-01
-2.28010088e-01 5.54520488e-01 1.21039414e+00 -4.95190173e-01
-2.18851402e-01 -4.11566086e-02 -4.93529141e-01 9.36956704e-01
1.07181215e+00 -2.57556409e-01 -6.41198754e-01 -3.63249816e-02
7.77120218e-02 -2.26585731e-01 3.13749433e-01 -1.25043964e+00
-3.78778800e-02 -4.96355116e-01 6.10354617e-02 -1.10993564e+00
3.76310915e-01 -8.95208836e-01 3.18347842e-01 4.37480807e-01
2.71910522e-02 -2.90765077e-01 2.51768738e-01 5.59586227e-01
-3.11689138e-01 -2.72913218e-01 1.04394388e+00 5.04197739e-02
-1.73145604e+00 1.53830379e-01 -3.61102879e-01 1.98873043e-01
1.37871361e+00 -1.26140058e+00 -1.27171606e-01 -2.52788025e-03
-4.14542079e-01 4.55121994e-01 6.58924341e-01 5.09348750e-01
8.35486650e-01 -1.12074721e+00 -6.44946158e-01 3.37743789e-01
7.17671752e-01 4.70967233e-01 4.55129534e-01 4.02604610e-01
-8.87148976e-01 3.48229766e-01 -7.47931242e-01 -9.51879859e-01
-1.05073118e+00 1.75510362e-01 2.49357551e-01 2.49980390e-01
-5.74139535e-01 5.83747685e-01 7.12030530e-01 -6.27634346e-01
-2.57757246e-01 -4.69923109e-01 -7.43205309e-01 1.83072284e-01
-9.81449783e-02 5.15700102e-01 -9.56195742e-02 -1.12273228e+00
-2.85188228e-01 1.62390316e+00 2.53958702e-01 1.38554230e-01
7.60262787e-01 -5.75128138e-01 1.81397468e-01 5.06135702e-01
8.99251461e-01 -1.12190686e-01 -1.38438177e+00 4.55637798e-02
1.23086609e-01 -6.52467012e-01 -1.96882784e-01 -6.16230726e-01
-6.72329307e-01 9.37440932e-01 8.96298349e-01 8.82949010e-02
9.77255821e-01 1.24635182e-01 6.67352915e-01 3.22927505e-01
9.51949298e-01 -1.29789841e+00 -9.39867571e-02 5.11333764e-01
7.75591552e-01 -1.61439979e+00 -1.26019493e-01 -8.65752161e-01
-7.22778857e-01 1.04642940e+00 7.17879176e-01 7.60138258e-02
3.82827073e-01 -4.13132571e-02 1.73811525e-01 -4.27914053e-01
3.09733748e-01 -5.34931719e-01 4.12888467e-01 9.19963241e-01
-2.78837174e-01 5.66017330e-01 -1.04929127e-01 2.50960976e-01
-5.79940975e-01 -4.64326978e-01 4.12666798e-01 6.92134738e-01
-1.20862412e+00 -7.24275529e-01 -5.63587964e-01 5.52587926e-01
2.27319777e-01 2.99154371e-01 -1.31446552e-02 9.65039492e-01
5.49870908e-01 1.24258244e+00 2.37956390e-01 -5.94573855e-01
5.65041363e-01 -5.38072288e-01 5.33616617e-02 -7.77408898e-01
1.85084119e-01 -5.21664500e-01 4.27110016e-01 -3.69280070e-01
-3.48672330e-01 -3.26605558e-01 -1.74502802e+00 -6.04100823e-02
-2.17208296e-01 2.82519639e-01 8.15101624e-01 7.79506147e-01
-4.20440659e-02 4.42326427e-01 8.69700909e-01 -1.01762235e+00
1.29899874e-01 -5.71298778e-01 -7.66830742e-01 3.54874372e-01
-4.34983941e-03 -1.29646492e+00 -3.61605436e-01 -1.10003509e-01] | [8.295819282531738, -1.8684076070785522] |
f66a9162-4dfa-4785-b3df-cd43a9d5725e | temporally-coherent-person-matting-trained-on | 2109.04843 | null | https://arxiv.org/abs/2109.04843v1 | https://arxiv.org/pdf/2109.04843v1.pdf | Temporally Coherent Person Matting Trained on Fake-Motion Dataset | We propose a novel neural-network-based method to perform matting of videos depicting people that does not require additional user input such as trimaps. Our architecture achieves temporal stability of the resulting alpha mattes by using motion-estimation-based smoothing of image-segmentation algorithm outputs, combined with convolutional-LSTM modules on U-Net skip connections. We also propose a fake-motion algorithm that generates training clips for the video-matting network given photos with ground-truth alpha mattes and background videos. We apply random motion to photos and their mattes to simulate movement one would find in real videos and composite the result with the background clips. It lets us train a deep neural network operating on videos in an absence of a large annotated video dataset and provides ground-truth training-clip foreground optical flow for use in loss functions. | ['Dmitry Vatolin', 'Andrey Moskalenko', 'Mikhail Erofeev', 'Ivan Molodetskikh'] | 2021-09-10 | null | null | null | null | ['image-matting', 'video-matting'] | ['computer-vision', 'computer-vision'] | [ 3.18336278e-01 2.28312507e-01 1.19419694e-01 -3.57194155e-01
-5.05509377e-01 -4.28617597e-01 3.92196178e-01 -7.49484599e-01
-5.13437033e-01 6.24151111e-01 -5.36506958e-02 -1.87488109e-01
6.18245959e-01 -7.33436346e-01 -1.42885649e+00 -4.56472784e-01
-1.96732014e-01 3.02811384e-01 3.29412669e-01 1.75604627e-01
-1.17862254e-01 1.70974314e-01 -1.22480190e+00 8.64103019e-01
6.30068421e-01 8.27257037e-01 -4.29953746e-02 1.33891296e+00
6.17290102e-02 1.16840076e+00 -8.34105730e-01 -7.79907823e-01
7.79945016e-01 -7.92901874e-01 -7.06926227e-01 5.48046529e-01
1.41599131e+00 -9.95707750e-01 -1.06601763e+00 8.75717342e-01
-2.40268886e-01 3.89429301e-01 4.61588740e-01 -1.39310503e+00
-6.67607486e-01 6.12307966e-01 -4.64517415e-01 2.93768436e-01
4.27656144e-01 9.08615470e-01 5.09921849e-01 -4.49014872e-01
1.00771880e+00 1.60216844e+00 8.43690932e-01 9.11895037e-01
-1.38189781e+00 -3.10448855e-01 5.24579026e-02 9.94071960e-02
-8.65454733e-01 -4.39711422e-01 5.24251521e-01 -5.19552171e-01
8.06958318e-01 3.17315578e-01 1.18160892e+00 1.58270693e+00
2.64581274e-02 1.10995293e+00 6.94674909e-01 -2.85600703e-02
9.26305354e-02 -2.33452857e-01 -4.36400175e-01 1.01890779e+00
1.32665873e-01 -1.07089020e-01 -2.70708293e-01 1.93178847e-01
1.31253731e+00 7.04526231e-02 -2.94360369e-01 -7.21282437e-02
-1.53961790e+00 4.47922945e-01 6.06507659e-01 1.97529778e-01
-4.62100327e-01 9.07549679e-01 2.28218526e-01 3.49771976e-01
4.40502077e-01 2.87791818e-01 -8.66342038e-02 -5.46949841e-02
-1.97448337e+00 4.51627195e-01 6.05373800e-01 7.78134525e-01
6.32181108e-01 8.31403613e-01 -2.12677896e-01 -2.87987962e-02
-1.15795366e-01 7.33052194e-01 3.21357429e-01 -1.75740635e+00
5.65128624e-01 2.63230801e-01 4.64203089e-01 -1.09298325e+00
-4.60588783e-02 6.04303926e-02 -7.77313411e-01 4.16673005e-01
1.10610247e+00 -3.98655921e-01 -1.47796512e+00 1.65153074e+00
-9.57041420e-03 1.08691406e+00 1.01822475e-02 1.19541276e+00
4.18360770e-01 9.32859480e-01 -8.92990977e-02 -9.88139585e-02
7.32311189e-01 -1.19501674e+00 -6.15287602e-01 -3.03260058e-01
2.99240202e-01 -4.36343789e-01 9.53883410e-01 4.18575317e-01
-1.71251523e+00 -6.80153310e-01 -9.66164947e-01 -2.65083283e-01
-7.87530839e-02 7.57156163e-02 3.36242497e-01 4.56811249e-01
-1.49924183e+00 1.08726549e+00 -1.25764465e+00 -1.25578031e-01
6.85430944e-01 3.17025661e-01 -3.19679797e-01 2.47742370e-01
-9.23977613e-01 8.93027306e-01 3.52002114e-01 4.08177227e-01
-1.62885690e+00 -7.52383590e-01 -9.76153553e-01 -1.06807806e-01
1.27609015e-01 -1.19062686e+00 1.16364396e+00 -2.26129389e+00
-1.31176484e+00 8.15259337e-01 -3.37451957e-02 -9.89875972e-01
1.28328860e+00 -4.40855056e-01 -1.32010400e-01 7.98554778e-01
-6.23866804e-02 1.29199433e+00 1.45877790e+00 -1.29156721e+00
-4.23795134e-01 3.03576618e-01 1.67506039e-01 -3.23494151e-02
6.31993562e-02 -2.80701965e-01 -4.84682560e-01 -8.14071834e-01
-3.91696572e-01 -8.12252045e-01 -4.18770730e-01 2.50604719e-01
-5.38804173e-01 5.73937833e-01 1.01420891e+00 -1.28893495e+00
8.05820405e-01 -1.82238948e+00 3.24849427e-01 1.54350773e-01
2.04194844e-01 3.72732133e-01 -4.23843116e-01 -2.03324080e-01
1.03764385e-02 -1.10615551e-01 -4.59925622e-01 -5.45386374e-01
-3.58150214e-01 2.22898841e-01 -4.13358897e-01 6.34685934e-01
5.32274842e-01 1.24362957e+00 -1.06082475e+00 -4.59780723e-01
4.95392025e-01 6.12295449e-01 -6.37165666e-01 4.89275366e-01
-8.00896406e-01 5.10432303e-01 1.68374956e-01 4.32320625e-01
5.97065330e-01 -9.42059681e-02 3.60720418e-02 -1.84745640e-01
7.22460002e-02 -7.32594505e-02 -9.00020540e-01 1.93904924e+00
-2.54720867e-01 1.06947839e+00 1.87368378e-01 -7.61956871e-01
2.67808348e-01 2.71147490e-01 5.70787251e-01 -8.69219378e-02
2.98019528e-01 -1.82682723e-02 -9.93670747e-02 -7.89300263e-01
6.04364514e-01 1.82577908e-01 3.02031726e-01 4.95635897e-01
1.92445993e-01 1.11466311e-02 2.48707011e-01 4.32913840e-01
1.21274197e+00 5.78981459e-01 -6.06282175e-01 5.00061736e-02
1.49371356e-01 7.86552206e-02 1.72957420e-01 6.64775908e-01
-2.26142466e-01 8.87001038e-01 3.96182299e-01 -8.23061585e-01
-1.79925358e+00 -1.14135456e+00 5.01182616e-01 8.05988133e-01
7.26410151e-02 -3.47480737e-02 -1.42385089e+00 -5.78705430e-01
-9.68064144e-02 7.21698761e-01 -1.00576234e+00 6.29696203e-03
-1.08298051e+00 -2.94950426e-01 9.39834952e-01 4.97956336e-01
7.06073046e-01 -1.19882250e+00 -9.44952846e-01 2.11422980e-01
-4.54197496e-01 -1.17597592e+00 -7.73911595e-01 -3.32123280e-01
-9.23140943e-01 -1.08922791e+00 -1.18496263e+00 -8.38952005e-01
8.03362489e-01 1.23507798e-01 1.21324122e+00 2.03870401e-01
-2.07136467e-01 4.40256506e-01 1.49835959e-01 3.55457127e-01
-7.47139871e-01 -3.58845353e-01 -5.99573329e-02 4.07054365e-01
9.52465385e-02 -5.13458967e-01 -9.45618391e-01 -1.31781161e-01
-1.15760672e+00 3.34629714e-01 3.79576474e-01 5.52996159e-01
5.61632104e-02 -3.91638011e-01 4.07381468e-02 -6.60377264e-01
1.06555186e-01 -2.17787370e-01 -5.16805053e-01 1.05638951e-01
2.75477260e-01 -1.30842745e-01 5.26032686e-01 -6.98503375e-01
-1.04250574e+00 2.41593555e-01 1.61711022e-01 -1.42296672e+00
-6.98980838e-02 -1.11837596e-01 7.52291158e-02 5.33930771e-02
8.23291183e-01 1.40104204e-01 5.35213947e-02 -4.79076505e-02
8.48728180e-01 2.15059385e-01 1.29698634e+00 -3.16349655e-01
9.16129708e-01 8.31068337e-01 -1.24397025e-01 -7.11258173e-01
-7.71223366e-01 5.14924042e-02 -7.28366375e-01 -5.55672050e-01
1.42209017e+00 -9.64476466e-01 -7.46457100e-01 5.17717719e-01
-1.41472554e+00 -9.89931941e-01 -1.99541584e-01 2.69862562e-01
-8.93937290e-01 4.79597300e-01 -1.13244164e+00 -7.59778559e-01
-2.56987065e-01 -1.05298126e+00 8.67289364e-01 2.64121056e-01
-2.65285790e-01 -9.69331443e-01 -3.60907912e-01 4.25183058e-01
2.37917587e-01 6.98686481e-01 2.13799849e-01 -1.77979097e-01
-1.28764498e+00 5.71612604e-02 -2.09538609e-01 6.09715462e-01
-2.01059327e-01 4.85324651e-01 -1.12814212e+00 -1.98557034e-01
-1.95416182e-01 -2.12754712e-01 1.21937239e+00 9.39497411e-01
1.20666981e+00 -8.88201058e-01 -6.42847717e-02 9.34646010e-01
1.08121300e+00 -8.74048620e-02 9.89142060e-01 2.64615625e-01
1.17677867e+00 4.24620062e-01 2.66345948e-01 2.69980934e-02
5.17407730e-02 2.95865029e-01 4.52961147e-01 -2.37767458e-01
-3.82560343e-01 -3.72262418e-01 8.03291082e-01 3.21418315e-01
-3.41338187e-01 -1.80356234e-01 -4.40704882e-01 6.99167609e-01
-2.00731325e+00 -1.61515522e+00 -1.64213374e-01 2.04088211e+00
9.09661114e-01 2.57960826e-01 5.23729146e-01 -2.90864915e-01
8.10713828e-01 8.63418654e-02 -4.63394880e-01 -2.55464077e-01
-2.60972887e-01 1.48487806e-01 5.14977455e-01 8.32504988e-01
-1.47396529e+00 1.22873735e+00 6.18438005e+00 2.46395737e-01
-1.19084871e+00 8.19131285e-02 1.07527566e+00 -4.13766474e-01
-1.92025274e-01 -1.33115903e-01 -1.38842225e-01 6.50144815e-01
1.17521787e+00 1.73549861e-01 6.54300094e-01 6.46546483e-01
6.71938777e-01 -6.41817749e-02 -1.28730309e+00 9.18564379e-01
1.88133195e-01 -1.82992435e+00 1.84752226e-01 -2.76717395e-01
1.06975996e+00 -1.73723206e-01 2.33861774e-01 -4.92198728e-02
4.44798797e-01 -1.13113308e+00 1.08158445e+00 7.09304094e-01
8.14440727e-01 -4.34207350e-01 3.40700269e-01 -1.10339388e-01
-7.19077706e-01 9.10653919e-02 -2.95617849e-01 -6.53256848e-02
5.21115065e-01 2.26938471e-01 -8.86408269e-01 9.01397467e-02
4.57044721e-01 8.25318456e-01 -4.66573238e-01 9.90326345e-01
1.67749228e-03 7.19185948e-01 -2.47399047e-01 5.16900241e-01
5.33689618e-01 -3.72692019e-01 6.73393607e-01 1.61729765e+00
2.59620011e-01 -3.17715079e-01 1.90747529e-01 1.24310553e+00
-2.15447694e-01 -5.62869966e-01 -6.19631708e-01 -2.38485232e-01
-7.28139579e-02 1.19012260e+00 -9.21751142e-01 -1.02477419e+00
-1.86969802e-01 1.59375715e+00 -1.74558774e-01 7.91700661e-01
-1.06312954e+00 -2.07498968e-02 5.39257646e-01 4.34251130e-01
3.61325592e-01 -3.66762519e-01 -2.29970440e-01 -1.58673167e+00
-2.03200117e-01 -8.62319231e-01 2.16599479e-01 -1.22072780e+00
-1.01952434e+00 6.51088178e-01 3.71482745e-02 -9.88413632e-01
-5.28861940e-01 -4.59045321e-01 -8.80047977e-01 5.04703045e-01
-6.90874815e-01 -1.38305151e+00 -4.28271770e-01 6.47891223e-01
7.15665400e-01 5.00094220e-02 1.02089748e-01 3.36438388e-01
-3.87152731e-01 3.53239268e-01 -4.27939832e-01 5.98687708e-01
6.26878977e-01 -1.37303877e+00 8.26307058e-01 1.47820926e+00
1.80115029e-01 4.16408539e-01 1.00868464e+00 -9.53632295e-01
-1.01388800e+00 -1.56522918e+00 3.08700651e-01 -6.43994987e-01
7.68395424e-01 -2.96569794e-01 -9.01789367e-01 1.29981887e+00
7.29874611e-01 -1.91791132e-01 -4.65890244e-02 -8.64664733e-01
-2.22559035e-01 1.81261331e-01 -1.07753825e+00 9.03437436e-01
9.90381837e-01 -3.33754659e-01 -5.49638987e-01 3.54118377e-01
8.04491878e-01 -6.84859157e-01 -3.53574634e-01 -1.47539675e-01
4.84755129e-01 -8.67205262e-01 1.13210726e+00 -9.28142726e-01
8.37735713e-01 -5.12401044e-01 2.02912435e-01 -1.13080895e+00
1.80535480e-01 -1.07369792e+00 -2.70854294e-01 6.31783366e-01
2.29522794e-01 1.62779659e-01 1.20434260e+00 9.61375058e-01
-1.89467058e-01 -1.33574843e-01 -7.68973768e-01 -4.37714726e-01
-5.41944653e-02 -3.40193391e-01 1.31846033e-02 9.38908756e-01
-4.52621162e-01 -2.30087876e-01 -9.02600765e-01 5.33099286e-02
8.03925276e-01 -1.52609169e-01 9.90403533e-01 -4.47001576e-01
-4.10280675e-01 -4.02713567e-01 -4.87810403e-01 -9.75574553e-01
2.96014130e-01 -6.09046221e-01 9.57209691e-02 -1.25638556e+00
-1.26557630e-02 2.79451400e-01 2.86347240e-01 4.83155638e-01
-1.07908092e-01 7.43502676e-01 3.99650931e-01 7.98538700e-02
-7.55345523e-01 2.35065043e-01 1.41681480e+00 -4.07255292e-01
-3.14263880e-01 -5.78383952e-02 1.53810233e-01 7.59491503e-01
5.48200190e-01 -2.42494941e-01 -2.02632144e-01 -5.16041219e-01
-9.22144949e-02 1.69177428e-01 1.00441611e+00 -1.23080373e+00
-1.40046433e-01 -3.33048314e-01 9.88672137e-01 -2.76717246e-01
4.78392810e-01 -5.63372076e-01 4.38030690e-01 8.36950719e-01
-5.26525557e-01 2.52653778e-01 1.34732977e-01 5.61814845e-01
2.50841796e-01 -1.44563884e-01 9.31959093e-01 -5.32850146e-01
-5.48075855e-01 2.93949962e-01 -5.57645082e-01 -6.47314116e-02
9.00136650e-01 -3.27652693e-01 -2.37797141e-01 -7.97417045e-01
-1.02652907e+00 -1.03274941e-01 8.61682713e-01 2.01103076e-01
6.48226738e-01 -1.14619493e+00 -6.09223187e-01 1.12733766e-01
-8.72887731e-01 8.75454322e-02 2.48844996e-01 6.95142031e-01
-1.43821168e+00 -8.12496021e-02 -5.89127004e-01 -6.34900570e-01
-9.21308458e-01 6.73293471e-01 6.00126326e-01 2.91629404e-01
-8.71822417e-01 8.74550581e-01 2.24319100e-01 3.32464427e-01
2.25108564e-01 -6.74551070e-01 5.20271301e-01 -2.90435404e-01
7.67463505e-01 2.16223523e-01 -3.73115748e-01 -5.63072145e-01
-4.06533759e-03 -8.94359648e-02 2.15799376e-01 -5.37397265e-01
1.08754075e+00 -1.10852197e-01 3.62798683e-02 3.44414353e-01
1.01033151e+00 -3.01887482e-01 -2.13745332e+00 3.17086399e-01
-2.02299714e-01 -7.11832166e-01 -2.66842723e-01 -5.18461704e-01
-1.46531641e+00 9.80727255e-01 3.28928202e-01 -7.54830465e-02
8.26299787e-01 -3.26289475e-01 1.20372581e+00 3.46232742e-01
-2.08574027e-01 -9.28485453e-01 6.11511886e-01 3.35140489e-02
5.70763052e-01 -9.04726505e-01 -2.80160218e-01 1.17386617e-01
-6.81027412e-01 1.33911431e+00 9.11482155e-01 -7.85673022e-01
1.22896060e-01 2.17478245e-01 4.35637712e-01 1.95615813e-01
-7.01941550e-01 -8.76807570e-02 3.02669555e-01 7.16426671e-01
6.14533089e-02 -9.33590159e-02 2.67963439e-01 3.95875387e-02
-1.60146937e-01 1.58245027e-01 1.01650047e+00 5.22885144e-01
-4.22171742e-01 -5.84430397e-01 -3.70226681e-01 4.12989467e-01
-5.09733200e-01 -7.41172805e-02 -4.21453923e-01 6.61691070e-01
1.44984394e-01 5.46485424e-01 5.63080430e-01 -3.41324747e-01
-3.50424379e-01 8.73833001e-02 8.63586605e-01 -1.09086506e-01
-6.67340934e-01 2.62967497e-01 1.31294146e-01 -7.70664930e-01
-7.20681190e-01 -5.12552738e-01 -1.10983479e+00 -7.07362592e-01
2.93982327e-01 -2.84530759e-01 2.26768002e-01 9.14088428e-01
9.77918953e-02 4.70623106e-01 1.60662800e-01 -1.54909587e+00
3.50970365e-02 -8.27391982e-01 -1.81437194e-01 9.52384174e-01
6.71356201e-01 -2.09961116e-01 -2.75599957e-01 9.23315823e-01] | [10.706969261169434, -0.8086175322532654] |
baac132c-cfad-4dbb-a0af-237f3340be63 | revisiting-gray-pixel-for-statistical | 1803.08326 | null | http://arxiv.org/abs/1803.08326v4 | http://arxiv.org/pdf/1803.08326v4.pdf | Revisiting Gray Pixel for Statistical Illumination Estimation | We present a statistical color constancy method that relies on novel gray
pixel detection and mean shift clustering. The method, called Mean Shifted Grey
Pixel -- MSGP, is based on the observation: true-gray pixels are aligned
towards one single direction. Our solution is compact, easy to compute and
requires no training. Experiments on two real-world benchmarks show that the
proposed approach outperforms state-of-the-art methods in the camera-agnostic
scenario. In the setting where the camera is known, MSGP outperforms all
statistical methods. | ['Joni-Kristian Kämäräinen', 'Jarno Nikkanen', 'Yanlin Qian', 'Jiri Matas', 'Said Pertuz'] | 2018-03-22 | null | null | null | null | ['color-constancy'] | ['computer-vision'] | [ 4.42820787e-01 -3.13654184e-01 -1.77540109e-01 -1.17396265e-01
-7.18507469e-01 -5.26100099e-01 4.74728405e-01 -2.97816545e-01
-3.79245847e-01 5.97208500e-01 -5.65369785e-01 -3.25018317e-01
3.71734679e-01 -2.67877519e-01 -8.56632590e-01 -1.24420238e+00
3.45463939e-02 -2.00972538e-02 6.95927858e-01 -2.28752159e-02
7.43791401e-01 2.08302930e-01 -1.41289198e+00 -1.53384447e-01
7.23058105e-01 1.10812914e+00 2.59091258e-01 8.36918473e-01
-1.56417377e-02 5.42253435e-01 -3.22208345e-01 -3.28391999e-01
8.74807179e-01 -6.62827909e-01 -2.67358840e-01 7.04952598e-01
9.08371270e-01 2.48026587e-02 3.23408730e-02 1.64174449e+00
2.71664470e-01 2.57556260e-01 4.11534756e-01 -1.20999479e+00
-8.65534067e-01 1.28736705e-01 -1.42221928e+00 2.86660910e-01
1.46579042e-01 1.60721794e-01 8.21709573e-01 -8.87099028e-01
5.16197741e-01 1.24808681e+00 5.99418759e-01 4.93085265e-01
-1.47163832e+00 -1.38808951e-01 4.18060988e-01 8.54500830e-02
-1.39680636e+00 -2.61480242e-01 7.11528838e-01 -2.11385235e-01
4.10212576e-01 1.18017793e-01 3.43358546e-01 6.87165022e-01
2.99651235e-01 7.33415484e-01 2.14922714e+00 -8.36746275e-01
5.24277866e-01 6.18666736e-03 -4.19286638e-02 1.04698825e+00
5.67428231e-01 3.96661252e-01 -5.55275559e-01 1.53116882e-01
8.97976100e-01 -6.21717274e-02 -4.23596740e-01 -7.51375079e-01
-1.36506343e+00 5.00773251e-01 2.02604860e-01 1.00496441e-01
-6.90369979e-02 5.25466025e-01 -1.76338017e-01 4.55606990e-02
4.54683900e-01 1.83577418e-01 -5.69003165e-01 -1.77304149e-01
-1.13397312e+00 -1.60353437e-01 7.23239005e-01 1.20292759e+00
9.78675961e-01 2.30906144e-01 -3.85986939e-02 3.92111421e-01
1.54990971e-01 1.08687937e+00 4.87891257e-01 -1.31197751e+00
7.21294284e-02 3.87621909e-01 2.68085837e-01 -9.40780938e-01
-2.61440098e-01 -1.96821868e-01 -8.98977399e-01 8.57905328e-01
5.73596358e-01 -1.73311844e-01 -1.35741663e+00 1.37086344e+00
2.32563108e-01 4.82270390e-01 3.87044959e-02 9.06921804e-01
1.81827560e-01 3.39867443e-01 -2.99767435e-01 -2.58410156e-01
1.12785435e+00 -1.05187571e+00 -5.98843217e-01 -4.27086115e-01
6.63181096e-02 -8.21149468e-01 9.33642447e-01 8.62729073e-01
-1.05220473e+00 -6.30690694e-01 -1.21580029e+00 2.19616383e-01
-6.56719387e-01 1.99931070e-01 6.16453528e-01 1.35251021e+00
-1.36582220e+00 6.35544240e-01 -7.16883421e-01 -3.92156094e-01
1.51461035e-01 -2.62582991e-02 -2.26120427e-01 -1.53824195e-01
-3.98588002e-01 5.49213886e-01 3.93746763e-01 4.58726920e-02
-4.56593841e-01 -3.05449009e-01 -7.32849002e-01 -4.60898995e-01
6.12739921e-01 -4.66105282e-01 1.08170867e+00 -1.35398889e+00
-1.95419908e+00 9.74129617e-01 -4.79068547e-01 -3.70929360e-01
8.35322022e-01 -9.79811549e-02 -3.28394443e-01 2.82928437e-01
-3.94993350e-02 6.05939567e-01 1.34226167e+00 -1.90444183e+00
-9.95523095e-01 -2.61271358e-01 -4.05682176e-01 4.51377444e-02
2.28557453e-01 -2.01926962e-01 -1.09852242e+00 -4.62820083e-01
4.08611268e-01 -1.26133549e+00 -4.70343024e-01 4.38253060e-02
-6.35768950e-01 3.99555147e-01 7.43941009e-01 -3.05688262e-01
8.53476107e-01 -2.18961096e+00 -2.88156956e-01 6.26460135e-01
-1.61404051e-02 -1.55154526e-01 1.43246740e-01 -1.53502658e-01
-1.42570734e-01 -2.92224199e-01 -5.45864820e-01 -4.22464788e-01
2.87916213e-02 2.36439750e-01 -2.64068712e-02 9.10118520e-01
-1.07108310e-01 6.90180957e-01 -1.13052523e+00 -7.44197071e-01
5.84817767e-01 1.99103266e-01 -1.10778764e-01 -3.23371857e-01
-8.54384154e-02 1.99187189e-01 1.45882741e-01 1.02515459e+00
1.43418884e+00 -2.92504460e-01 8.65503624e-02 -1.68675467e-01
-3.36593062e-01 -7.61717618e-01 -1.52851427e+00 1.60868073e+00
-6.07017353e-02 7.78243721e-01 -1.92787256e-02 -6.00833118e-01
8.72422516e-01 -5.04301369e-01 3.78230065e-01 -7.67610848e-01
1.17782868e-01 2.00492665e-01 -2.40136907e-01 -8.48677829e-02
7.93097615e-01 2.26591676e-01 1.12405501e-01 2.96474814e-01
-3.06451827e-01 -3.72267902e-01 4.85552698e-01 1.83871225e-01
7.79970407e-01 3.91114384e-01 3.72372150e-01 -4.63212162e-01
4.77855891e-01 -8.39891434e-02 6.01633787e-01 1.04953170e+00
-5.53575635e-01 9.31216598e-01 3.55894059e-01 6.24041185e-02
-5.61028779e-01 -1.19265687e+00 2.04530358e-03 1.07667315e+00
8.38296652e-01 6.97790906e-02 -1.05150354e+00 -5.99363863e-01
4.41587567e-02 5.80346107e-01 -1.02957380e+00 3.05719972e-01
-2.42656425e-01 -9.51402009e-01 -1.74289599e-01 5.32711089e-01
7.48626709e-01 -3.70766729e-01 -9.01898444e-01 -1.55662298e-01
1.57338738e-01 -1.41424966e+00 -6.26548588e-01 3.86105776e-01
-8.37128162e-01 -1.21322191e+00 -8.37710559e-01 -6.49160504e-01
1.07613039e+00 9.96003568e-01 1.21430027e+00 -1.00434214e-01
-3.77212465e-01 8.16876352e-01 -3.61756891e-01 -1.08361617e-01
-1.56779379e-01 -4.54504371e-01 -3.58838886e-01 4.41741675e-01
4.81589228e-01 4.79213893e-02 -9.67521131e-01 4.33198839e-01
-7.59589672e-01 -8.79918560e-02 1.04760122e+00 7.59942353e-01
1.23101580e+00 1.78983346e-01 -4.10342097e-01 -1.09711170e+00
8.99143741e-02 1.29263058e-01 -1.00155294e+00 5.18324852e-01
-9.98179853e-01 1.64275095e-01 1.83132946e-01 -3.07613283e-01
-1.16516089e+00 6.96774542e-01 8.23921144e-01 -3.93292516e-01
-1.32818192e-01 -2.80665517e-01 -2.07445454e-02 -6.47929788e-01
4.92048234e-01 4.30573493e-01 -2.63612121e-01 -2.22096696e-01
8.85506451e-01 2.28298888e-01 1.17237163e+00 -3.78067166e-01
8.63016605e-01 1.14544177e+00 5.61508536e-01 -7.40069389e-01
-5.03957033e-01 -7.40291238e-01 -9.07812655e-01 -3.15305322e-01
8.86372745e-01 -7.41884410e-01 -5.17901361e-01 7.11876392e-01
-8.01620781e-01 -5.82580149e-01 -1.43675104e-01 2.06458896e-01
-8.15711975e-01 7.41789401e-01 -1.62465852e-02 -1.03398275e+00
1.08464779e-02 -9.41019773e-01 1.26815629e+00 5.46682954e-01
4.14303243e-01 -1.22983289e+00 -4.13399972e-02 -1.11025795e-01
3.30944508e-01 4.03745949e-01 2.21997932e-01 2.63585020e-02
-7.90160656e-01 -1.72551781e-01 -5.91537058e-01 4.67762202e-01
2.03341872e-01 4.80512798e-01 -9.07775879e-01 -6.87344894e-02
-2.49493003e-01 2.98117876e-01 1.14544094e+00 7.06070423e-01
1.17786443e+00 1.73127770e-01 -2.23531336e-01 7.42432058e-01
1.91712844e+00 -2.15131734e-02 5.69226742e-01 5.40123761e-01
6.11952424e-01 1.95105344e-01 8.95786643e-01 4.12741005e-01
3.17000657e-01 5.08433521e-01 6.36203408e-01 -5.17497122e-01
-2.40199849e-01 8.00952166e-02 3.94897312e-01 1.41335964e-01
-1.84614807e-01 7.69151002e-02 -5.89160323e-01 5.52713335e-01
-1.99808931e+00 -9.25543547e-01 -5.51180482e-01 2.26460743e+00
7.04187453e-01 7.38421977e-02 1.18739709e-01 3.42820063e-02
8.71271610e-01 6.58688769e-02 -5.16731203e-01 -2.30429322e-01
-7.17377484e-01 8.06684494e-02 1.58567631e+00 4.87386465e-01
-1.37959957e+00 1.00820422e+00 7.09320307e+00 5.52150786e-01
-1.01567066e+00 2.20909536e-01 9.44482267e-01 3.18568766e-01
3.44801754e-01 6.61117285e-02 -7.76602864e-01 5.75357616e-01
4.21453297e-01 5.97277805e-02 3.72162879e-01 1.01144195e+00
8.07609558e-02 -9.75754738e-01 -8.53384197e-01 1.29207170e+00
6.00051284e-01 -8.70750964e-01 -5.33648074e-01 -1.60410464e-01
1.47707367e+00 1.34831946e-02 6.11111581e-01 -4.66885448e-01
6.89242959e-01 -4.61138338e-01 7.55082726e-01 7.99760282e-01
7.39653826e-01 -4.74239647e-01 5.58225930e-01 -3.39971632e-01
-1.10900140e+00 4.56161499e-02 -4.03761089e-01 4.76811856e-01
1.36170208e-01 5.51843047e-01 -2.80646384e-01 3.29699367e-01
1.02586102e+00 7.88213372e-01 -1.13706374e+00 1.16459179e+00
-5.21293342e-01 2.59497792e-01 -2.38682956e-01 2.76746787e-02
3.53597939e-01 -4.35084730e-01 8.00320357e-02 1.49684799e+00
3.29985946e-01 -1.96516648e-01 1.94917336e-01 6.54956222e-01
2.79772609e-01 -1.24865882e-01 -1.86508700e-01 4.64302361e-01
1.65377200e-01 1.44754291e+00 -1.33043349e+00 -2.91196018e-01
-6.32843852e-01 1.65571952e+00 -3.05053622e-01 6.75723195e-01
-5.15068173e-01 -5.26068270e-01 2.96472341e-01 -2.43565336e-01
9.04591084e-01 -4.04808342e-01 -5.88671744e-01 -8.92117381e-01
-1.10344507e-01 -5.90375602e-01 3.68888378e-01 -1.11133897e+00
-1.22615552e+00 3.90288413e-01 -3.78209017e-02 -1.48949158e+00
-1.45480722e-01 -1.18490911e+00 -5.83317280e-01 5.94953775e-01
-2.12815666e+00 -1.03487933e+00 -7.69749165e-01 9.49037611e-01
3.53553832e-01 1.48505056e-02 2.84620672e-01 -3.08566213e-01
-5.06956518e-01 5.84923089e-01 7.28294015e-01 -2.59673484e-02
9.49136615e-01 -1.85859513e+00 3.72783244e-01 1.59329808e+00
1.78950503e-01 3.09180647e-01 8.87911320e-01 -2.52123505e-01
-1.55611515e+00 -1.06750095e+00 2.62476832e-01 -4.93093103e-01
9.18669522e-01 -1.36301696e-01 -5.28295279e-01 3.25817525e-01
4.75745589e-01 3.78280848e-01 5.19904912e-01 -4.77466434e-01
-3.50189447e-01 -3.36088210e-01 -1.04434776e+00 4.86692250e-01
7.36991107e-01 -1.48728505e-01 -1.97134674e-01 3.50635797e-01
1.53032809e-01 -5.26403487e-01 -3.88515681e-01 -1.70760080e-01
2.26067469e-01 -1.13480079e+00 8.71969879e-01 3.83424945e-02
4.70038503e-02 -7.07179666e-01 -3.66622716e-01 -1.30570340e+00
-1.59683689e-01 -1.08525658e+00 -1.31152600e-01 8.39488089e-01
3.28949392e-01 -7.44254112e-01 7.94754982e-01 6.47908866e-01
5.33421598e-02 -1.84572741e-01 -9.34838295e-01 -1.20386136e+00
-2.17681974e-01 -2.87628680e-01 1.92991242e-01 5.50698042e-01
-3.47534180e-01 -1.86390758e-01 -4.13911521e-01 3.80641669e-01
1.20777476e+00 4.81037766e-01 9.42539036e-01 -9.48523343e-01
-5.28226137e-01 -7.25700200e-01 -6.39688849e-01 -1.13775826e+00
-1.57886241e-02 -3.67139801e-02 4.09266233e-01 -1.24324059e+00
1.65881559e-01 -1.17065847e-01 -4.80100483e-01 2.73741484e-01
-5.59058845e-01 7.20881104e-01 1.02892995e-01 -6.01979531e-02
-1.12733352e+00 1.30338907e-01 1.03424811e+00 -1.68755934e-01
-9.90713313e-02 -6.76137283e-02 -7.00341821e-01 7.02275693e-01
7.65220106e-01 -1.52729884e-01 -1.36936009e-01 -1.66378506e-02
6.12407438e-02 -7.93041170e-01 2.57801443e-01 -1.09079492e+00
3.84240329e-01 -3.76065910e-01 4.90646571e-01 -6.77413404e-01
3.32644939e-01 -1.00861120e+00 -3.58870000e-01 3.96193683e-01
2.50621140e-01 2.49392360e-01 3.26278731e-02 8.88067067e-01
-1.12257905e-01 -1.43125743e-01 1.19361126e+00 -9.20168087e-02
-1.43196166e+00 6.45780787e-02 -1.30600676e-01 8.68640617e-02
1.37321532e+00 -6.37526274e-01 -5.36810219e-01 -3.30663443e-01
-2.95300245e-01 -9.37442258e-02 9.71830487e-01 9.22219977e-02
5.33373058e-01 -1.10454381e+00 -4.50114518e-01 2.46456444e-01
2.89990246e-01 -5.16642332e-01 1.72949716e-01 1.15827191e+00
-7.34955251e-01 2.17316747e-01 4.25523296e-02 -9.42769468e-01
-1.34322929e+00 8.30060363e-01 3.54638278e-01 2.38732263e-01
-5.16497016e-01 8.22022855e-01 1.74464896e-01 2.13796824e-01
9.37447548e-02 -5.78640163e-01 4.79697615e-01 -4.27610755e-01
6.16135538e-01 4.89275157e-01 -1.88029986e-02 -7.36330748e-01
-3.18981171e-01 1.02245235e+00 1.49064422e-01 -4.14195657e-01
9.48580265e-01 -6.20376050e-01 -1.25960052e-01 6.37166500e-01
8.65752757e-01 1.16175815e-01 -1.89991999e+00 -4.88918483e-01
-5.89594766e-02 -1.18496621e+00 4.23245579e-01 -8.25786233e-01
-1.49117815e+00 7.66254604e-01 1.10633123e+00 2.64320016e-01
1.54329169e+00 -1.18039235e-01 1.26455352e-01 4.20575619e-01
2.93067783e-01 -1.66846919e+00 1.84903681e-01 2.58487165e-01
3.38319778e-01 -1.65102017e+00 4.41141278e-01 -4.71246272e-01
-9.75783348e-01 1.23585117e+00 5.33722520e-01 -2.61371791e-01
7.46637642e-01 1.44719899e-01 3.45293611e-01 2.30517477e-01
-3.83255482e-01 -7.30199456e-01 2.27095619e-01 9.82877016e-01
1.65925864e-02 6.11406006e-02 8.43953714e-02 -1.25708967e-01
2.21054465e-01 -3.77327472e-01 7.07784355e-01 1.04916501e+00
-4.53233033e-01 -9.58914757e-01 -6.81630194e-01 3.37377340e-02
-2.61919737e-01 -2.25805253e-01 -6.52557075e-01 9.61540103e-01
-1.06465509e-02 1.06314838e+00 1.57923207e-01 -1.16610035e-01
4.63427417e-02 -2.56002754e-01 7.45423496e-01 -3.32722366e-02
7.81352669e-02 3.09701830e-01 -5.46066344e-01 -1.05494344e+00
-1.08562434e+00 -1.01800823e+00 -1.00206792e+00 -2.67193854e-01
-3.66955996e-01 -3.52291286e-01 8.42080832e-01 4.56993759e-01
2.63830960e-01 1.87905446e-01 9.27775443e-01 -1.03217447e+00
-2.14497566e-01 -4.84979928e-01 -1.05814254e+00 3.53056401e-01
4.64486897e-01 -5.53923607e-01 -6.23298764e-01 4.79370832e-01] | [10.47368049621582, -2.5575246810913086] |
df0e8615-058a-4ef2-9163-f592b6ca60c8 | modeling-user-behavior-with-graph-convolution | 2202.06081 | null | https://arxiv.org/abs/2202.06081v1 | https://arxiv.org/pdf/2202.06081v1.pdf | Modeling User Behavior with Graph Convolution for Personalized Product Search | User preference modeling is a vital yet challenging problem in personalized product search. In recent years, latent space based methods have achieved state-of-the-art performance by jointly learning semantic representations of products, users, and text tokens. However, existing methods are limited in their ability to model user preferences. They typically represent users by the products they visited in a short span of time using attentive models and lack the ability to exploit relational information such as user-product interactions or item co-occurrence relations. In this work, we propose to address the limitations of prior arts by exploring local and global user behavior patterns on a user successive behavior graph, which is constructed by utilizing short-term actions of all users. To capture implicit user preference signals and collaborative patterns, we use an efficient jumping graph convolution to explore high-order relations to enrich product representations for user preference modeling. Our approach can be seamlessly integrated with existing latent space based methods and be potentially applied in any product retrieval method that uses purchase history to model user preferences. Extensive experiments on eight Amazon benchmarks demonstrate the effectiveness and potential of our approach. The source code is available at \url{https://github.com/floatSDSDS/SBG}. | ['Keping Yang', 'Taiwei Jin', 'Sen Li', 'Guli Lin', 'Fuyu Lv', 'Xiaotong Zhang', 'Xiao-Ming Wu', 'Bo Liu', 'Qimai Li', 'Fan Lu'] | 2022-02-12 | null | null | null | null | ['learning-semantic-representations'] | ['methodology'] | [ 6.62576780e-02 -2.77210653e-01 -8.42674613e-01 -7.09534407e-01
-3.59773487e-01 -6.00062013e-01 7.56165385e-01 4.54146117e-01
-1.99243680e-01 8.08821172e-02 6.37760937e-01 -2.15778753e-01
-5.05123496e-01 -8.12912464e-01 -2.93642163e-01 -3.15742135e-01
-2.30503097e-01 5.78149915e-01 -1.09748496e-02 -3.77640545e-01
2.24157989e-01 4.29514050e-02 -1.44437909e+00 4.92203802e-01
7.96193480e-01 1.02984810e+00 2.32310310e-01 2.29600161e-01
-2.35952541e-01 5.30132413e-01 2.78110862e-01 -2.46580511e-01
-1.09853661e-02 -1.15372054e-01 -9.06657636e-01 1.84325885e-03
1.42718688e-01 -3.04555565e-01 -6.28543913e-01 1.02413082e+00
3.50771606e-01 5.98167837e-01 4.72467273e-01 -9.85081911e-01
-1.19281101e+00 9.28179204e-01 -4.29320931e-01 2.66252995e-01
5.04508257e-01 -3.52368057e-01 1.59538329e+00 -9.95858788e-01
3.23322296e-01 1.28796971e+00 3.65038872e-01 4.07696486e-01
-1.46484792e+00 -4.97669339e-01 7.91144431e-01 3.22693050e-01
-1.32807958e+00 -1.55995786e-01 9.70562756e-01 -4.09060419e-01
1.07354021e+00 3.47077072e-01 4.93776411e-01 1.21219850e+00
-3.19644868e-01 1.33370066e+00 6.39920950e-01 -2.38704219e-01
9.16511416e-02 1.66640297e-01 7.57037759e-01 5.91583014e-01
-8.10775161e-02 -1.93536475e-01 -6.71219349e-01 -5.63749373e-01
7.97754288e-01 8.09793532e-01 -1.23473875e-01 -6.44870460e-01
-8.26253414e-01 1.11382711e+00 6.32198393e-01 4.89697784e-01
-5.72837532e-01 7.52542317e-02 2.87261396e-03 2.05331758e-01
9.24267590e-01 3.18901032e-01 -5.91058195e-01 -1.64561775e-02
-8.68119657e-01 2.57910013e-01 7.32377052e-01 1.02180517e+00
7.71547735e-01 -4.63385522e-01 -3.03686857e-01 1.13864017e+00
7.71771789e-01 -5.14993146e-02 8.59811187e-01 -5.56281388e-01
1.85349718e-01 7.09299445e-01 1.93480402e-01 -1.07834291e+00
-2.10165769e-01 -6.57688916e-01 -3.54966819e-01 -5.97786129e-01
-1.25664324e-01 3.16966683e-01 -1.14736402e+00 1.53299558e+00
9.32062939e-02 2.94469863e-01 -3.14918280e-01 9.56368923e-01
6.79540336e-01 6.74916625e-01 3.54197353e-01 -4.46605161e-02
1.33372140e+00 -1.19183636e+00 -6.76028848e-01 -3.37222755e-01
7.67224908e-01 -7.83072889e-01 1.26050711e+00 2.05168188e-01
-1.01351285e+00 -4.33644176e-01 -6.65565431e-01 -1.55717045e-01
-5.64178944e-01 -2.28613243e-01 1.23108828e+00 5.64279079e-01
-8.58094752e-01 7.85298586e-01 -1.03632712e+00 -5.82723737e-01
5.27295411e-01 5.73643267e-01 1.50207490e-01 -2.77618468e-01
-1.53715038e+00 4.21371460e-01 -3.02139129e-02 7.74729177e-02
-6.39547110e-01 -7.30997264e-01 -8.95406187e-01 5.06851852e-01
4.85856563e-01 -7.40656614e-01 1.36577690e+00 -6.55941725e-01
-1.39916575e+00 3.16408932e-01 -3.79302442e-01 -2.02610359e-01
-4.97099012e-02 -4.47563767e-01 -6.76806509e-01 -3.93655419e-01
-2.38208305e-02 2.74334490e-01 3.89733255e-01 -1.01688933e+00
-6.13906622e-01 -6.29619658e-01 2.22045526e-01 3.97784382e-01
-7.84331381e-01 7.26373047e-02 -9.52676535e-01 -6.13355875e-01
3.16946059e-01 -9.92286146e-01 -5.19082248e-01 -4.86549407e-01
-1.15848012e-01 -6.00258887e-01 5.23590744e-01 -2.14569151e-01
1.50561011e+00 -2.16956925e+00 7.47161731e-02 4.11578804e-01
1.28166988e-01 1.39874101e-01 -3.17132086e-01 7.92510569e-01
1.34130582e-01 2.74820268e-01 2.44101316e-01 -7.95167089e-01
4.82934117e-01 1.61142170e-01 -3.81557196e-01 3.38603437e-01
-4.17271495e-01 1.33623803e+00 -1.20166445e+00 -1.88947394e-01
1.65013358e-01 7.54534006e-01 -6.62034810e-01 5.44933230e-02
-2.70369023e-01 2.39825957e-02 -1.00098586e+00 6.41191304e-01
4.05892372e-01 -7.78331876e-01 4.40342098e-01 -1.08083956e-01
1.18934013e-01 7.93487251e-01 -9.96906996e-01 2.09219193e+00
-6.12866163e-01 4.14244682e-02 -2.81567186e-01 -6.77643657e-01
5.52174985e-01 2.08371505e-01 7.71610379e-01 -9.11703229e-01
2.98997425e-02 9.44091305e-02 -4.93994087e-01 -2.37113521e-01
7.20290899e-01 3.13854635e-01 -5.94956540e-02 5.78016698e-01
7.44720921e-02 4.57411438e-01 8.34310427e-02 4.15059566e-01
7.98194587e-01 1.36324555e-01 7.85804391e-02 -1.97991148e-01
2.98692614e-01 -3.88125598e-01 2.55425215e-01 8.46617520e-01
2.21912324e-01 4.10978019e-01 6.74002096e-02 -3.84458989e-01
-4.40611690e-01 -1.00457108e+00 -5.44818118e-03 1.72115338e+00
2.26491511e-01 -8.34494233e-01 -1.53413445e-01 -7.60732293e-01
1.37629911e-01 7.52611995e-01 -7.10555613e-01 -3.22298765e-01
-3.32808405e-01 -8.19290519e-01 -1.97390661e-01 7.26542473e-01
4.90949415e-02 -9.67902005e-01 7.49344006e-02 2.23099455e-01
-2.28628144e-01 -7.11500227e-01 -9.59102809e-01 -8.51422846e-02
-9.15624738e-01 -5.77022851e-01 -7.88454950e-01 -7.38852799e-01
5.81247330e-01 5.51738024e-01 1.17490554e+00 1.24157868e-01
-1.41777501e-01 6.91659570e-01 -5.35493553e-01 -2.15119366e-02
5.08299112e-01 3.13652903e-01 2.30095878e-01 4.32424217e-01
8.92603159e-01 -7.46843040e-01 -1.21983206e+00 4.44385767e-01
-9.99513328e-01 -2.83373058e-01 3.44323844e-01 7.54587889e-01
5.57471931e-01 1.12819061e-01 9.23020467e-02 -1.24825466e+00
9.16866601e-01 -9.28249538e-01 -3.27923834e-01 2.32284218e-01
-1.16476297e+00 5.43874167e-02 4.70305281e-03 -5.69850206e-01
-1.01532888e+00 4.89021763e-02 7.18392283e-02 -2.69519597e-01
-1.06842361e-01 1.05014324e+00 -5.72257414e-02 3.97834420e-01
3.47479492e-01 8.78138766e-02 -4.93607819e-01 -1.03210378e+00
7.35614836e-01 6.78398371e-01 -6.74506202e-02 -3.79502743e-01
1.88961208e-01 4.24074233e-01 -4.79710221e-01 -4.92997319e-01
-1.00282824e+00 -1.25814545e+00 -3.70088458e-01 3.98201466e-01
4.99987900e-01 -9.23504531e-01 -6.47923410e-01 1.91467609e-02
-5.97372651e-01 -2.58566946e-01 -1.57749355e-01 6.64313376e-01
-2.37941727e-01 4.66581166e-01 -7.84087420e-01 -9.08475459e-01
-2.74829626e-01 -7.90277719e-01 1.09809864e+00 1.41577005e-01
-5.20427346e-01 -1.50875163e+00 7.08685666e-02 4.01203245e-01
6.73957705e-01 -4.19661373e-01 8.72823238e-01 -8.20825934e-01
-4.70119089e-01 -4.05774057e-01 -1.68429345e-01 -1.42008722e-01
5.22880137e-01 -5.34333169e-01 -7.76178479e-01 -3.61752570e-01
-1.49474025e-01 -1.52728662e-01 1.25647366e+00 4.74093080e-01
1.22974193e+00 -4.65027630e-01 -8.41715455e-01 3.15421611e-01
1.30114806e+00 1.05598103e-02 3.61407220e-01 -4.32393886e-02
7.46652544e-01 6.55249536e-01 6.13754034e-01 6.13481343e-01
6.30234599e-01 1.04547513e+00 3.08380097e-01 -7.01836869e-02
1.89844206e-01 -6.36313796e-01 1.83200285e-01 5.73231816e-01
-7.07011744e-02 -5.82077026e-01 -5.57748914e-01 6.58316672e-01
-2.33544731e+00 -9.08153713e-01 -6.32228553e-02 2.16037440e+00
6.28890336e-01 -3.03932130e-02 1.18345559e-01 -2.16455653e-01
4.22951192e-01 3.44899535e-01 -5.12251496e-01 -1.09325849e-01
3.03666323e-01 1.19643658e-01 5.52169323e-01 8.34939420e-01
-1.04331636e+00 1.12202477e+00 5.60156059e+00 6.88935637e-01
-6.85948849e-01 2.59327888e-01 4.59810197e-01 -3.39143842e-01
-9.03679490e-01 -5.42128161e-02 -7.91540504e-01 3.92373919e-01
7.03393757e-01 2.89475825e-03 5.54410756e-01 7.48146355e-01
7.34604970e-02 3.33018601e-01 -1.19621706e+00 1.01185763e+00
9.39254761e-02 -1.13780367e+00 8.60951990e-02 3.06488961e-01
6.33370221e-01 1.21788032e-01 6.47919774e-01 3.34848136e-01
6.69537783e-01 -8.83202612e-01 2.19511077e-01 5.59628189e-01
2.94118166e-01 -4.46672052e-01 5.19017816e-01 1.74498141e-01
-1.44740486e+00 -3.24124694e-01 -2.41722316e-01 -1.42894700e-01
4.01759118e-01 5.11446655e-01 -7.26823628e-01 6.33020759e-01
7.31791496e-01 1.22572732e+00 -5.19734561e-01 8.19699824e-01
8.72858614e-02 7.10155725e-01 -3.61848682e-01 -7.74941146e-02
3.64959419e-01 -2.19738290e-01 1.09892040e-01 1.18616951e+00
4.96162117e-01 9.64003876e-02 3.45188707e-01 8.03091228e-01
9.71580520e-02 3.69432837e-01 -3.83925527e-01 -5.98477185e-01
2.01256573e-01 1.19411004e+00 -6.31305516e-01 -1.78475589e-01
-6.78148806e-01 1.27212179e+00 2.99416423e-01 7.34281003e-01
-5.58884859e-01 -2.10939217e-02 8.71497214e-01 3.35532278e-01
4.79070723e-01 -2.61640877e-01 -1.11540779e-01 -1.22739053e+00
-6.83277771e-02 -3.74756545e-01 8.46009016e-01 -4.40124035e-01
-1.61985898e+00 5.24460018e-01 -8.47491175e-02 -9.19135869e-01
-2.43674964e-01 -4.94069934e-01 -2.67710477e-01 1.04080415e+00
-1.41087496e+00 -1.47300410e+00 -9.33484286e-02 7.60886073e-01
6.59885108e-01 -1.16815209e-01 1.01811659e+00 5.50468504e-01
-2.32378170e-01 5.22448599e-01 4.44293730e-02 -2.17322782e-01
4.33683276e-01 -1.14793313e+00 5.58701158e-01 2.99898893e-01
6.26098633e-01 1.14825404e+00 5.89564145e-01 -8.13585699e-01
-1.38503397e+00 -7.64199078e-01 1.33873641e+00 -7.31408477e-01
7.03005373e-01 -5.64179838e-01 -7.35388517e-01 9.43563044e-01
1.25201985e-01 -7.74820745e-02 1.22737503e+00 9.22492802e-01
-4.53508556e-01 9.25883204e-02 -5.16247571e-01 7.15482533e-01
1.29225886e+00 -6.44761503e-01 -3.89796048e-01 4.18946415e-01
8.03531051e-01 3.86537425e-02 -8.24562967e-01 4.17693675e-01
6.21436536e-01 -4.88679141e-01 1.26643789e+00 -8.45212936e-01
5.51569946e-02 6.94246218e-02 -2.81425536e-01 -1.28555930e+00
-9.09096539e-01 -6.80224597e-01 -5.05579412e-01 1.26264405e+00
6.66941106e-01 -7.31166720e-01 9.97552395e-01 1.05497229e+00
4.17443849e-02 -9.03980613e-01 -1.19081952e-01 -3.76381665e-01
-3.20139289e-01 -2.85209477e-01 8.83056760e-01 1.04949236e+00
4.18615431e-01 4.61569458e-01 -5.13138175e-01 2.39555731e-01
4.66123790e-01 5.67670703e-01 1.13273837e-01 -1.34722221e+00
-5.73288023e-01 -6.23860359e-01 9.23531279e-02 -1.65141535e+00
1.71606522e-02 -1.04779363e+00 -4.03911054e-01 -1.87740088e+00
3.27679962e-01 -6.45437062e-01 -1.14046431e+00 4.92604077e-01
-1.82542473e-01 2.52190053e-01 -1.38254240e-01 3.42657655e-01
-9.21295583e-01 5.79309344e-01 9.53011334e-01 -2.15517551e-01
-5.78943312e-01 3.55220705e-01 -9.81175303e-01 3.99758637e-01
8.42269063e-01 -4.14140999e-01 -9.64968681e-01 -5.61880589e-01
6.66820407e-01 -2.82196403e-01 8.94074887e-02 -1.90506354e-01
3.79155397e-01 -1.99772641e-01 2.92731971e-01 -3.96008044e-01
7.16108441e-01 -8.62205207e-01 1.30436137e-01 5.18739745e-02
-6.68711543e-01 -8.29777047e-02 -1.78405270e-01 8.40742230e-01
-2.70176142e-01 3.50086316e-02 -8.49478468e-02 -2.79629499e-01
-7.85330713e-01 8.88347924e-01 -1.91694096e-01 -5.76968372e-01
4.09948826e-01 1.30493566e-01 2.00313292e-02 -5.82781613e-01
-9.76299524e-01 5.14307380e-01 2.87732154e-01 9.74789321e-01
5.71165860e-01 -1.53335750e+00 -2.45773643e-01 2.74442613e-01
4.70127702e-01 -4.80311930e-01 4.65673059e-01 5.81022382e-01
2.77788937e-01 8.23211372e-01 2.16018543e-01 -2.21614510e-01
-1.17191136e+00 8.46359611e-01 5.21297157e-02 -4.15981859e-01
-3.46245348e-01 1.08530855e+00 3.47943604e-01 -3.04551274e-01
3.69958997e-01 -1.31237715e-01 -6.77072465e-01 1.23520538e-01
4.61882412e-01 2.46792763e-01 -1.57501504e-01 -5.48648417e-01
-2.63316840e-01 3.42891306e-01 -4.87365931e-01 -9.22416896e-02
1.43579245e+00 -5.55322051e-01 5.31656407e-02 4.88851666e-01
1.24393654e+00 -4.01087403e-02 -1.03099155e+00 -8.23250592e-01
2.00412780e-01 -6.83132768e-01 3.13172221e-01 -7.40287244e-01
-1.01818371e+00 6.71570301e-01 5.59205413e-01 5.92761576e-01
9.18978870e-01 4.46313411e-01 9.72574353e-01 2.40162332e-02
3.57727826e-01 -9.57561553e-01 2.70721279e-02 2.32813209e-01
5.60698152e-01 -1.19104087e+00 -9.23031345e-02 -6.34246826e-01
-7.47430444e-01 6.18818104e-01 3.06902885e-01 5.72372749e-02
1.23042524e+00 -4.19325352e-01 -5.09705693e-02 -4.79621023e-01
-8.03018212e-01 -4.11843032e-01 8.25839341e-01 5.10272570e-02
9.10233974e-01 2.73189634e-01 -3.62800121e-01 8.44804525e-01
6.31954148e-02 5.32796159e-02 -1.66652396e-01 9.66422677e-01
-1.43771380e-01 -1.68790770e+00 4.54298586e-01 4.65511531e-01
-4.20708418e-01 -4.46643919e-01 -2.57396132e-01 -1.40012882e-03
-2.05309302e-01 1.02710581e+00 3.84911671e-02 -4.01296556e-01
2.56573766e-01 2.43314996e-01 2.85097152e-01 -8.93433213e-01
-2.93546557e-01 2.38327965e-01 -7.09430873e-02 -9.95454073e-01
-3.67474437e-01 -6.93331659e-01 -9.45998371e-01 -9.25136209e-02
-2.71590501e-01 3.28599244e-01 5.66639245e-01 6.88883126e-01
7.98380971e-01 3.49422961e-01 5.00409186e-01 -7.55166292e-01
-4.66965973e-01 -1.03593731e+00 -7.80984104e-01 6.61595285e-01
2.77948737e-01 -7.63239503e-01 -1.19636089e-01 -1.04054064e-01] | [10.151416778564453, 5.6860575675964355] |
e383ca6c-98e5-4863-9d88-c40fd8309044 | autoencoder-based-anomaly-detection-in | 2305.08977 | null | https://arxiv.org/abs/2305.08977v1 | https://arxiv.org/pdf/2305.08977v1.pdf | Autoencoder-based Anomaly Detection in Streaming Data with Incremental Learning and Concept Drift Adaptation | In our digital universe nowadays, enormous amount of data are produced in a streaming manner in a variety of application areas. These data are often unlabelled. In this case, identifying infrequent events, such as anomalies, poses a great challenge. This problem becomes even more difficult in non-stationary environments, which can cause deterioration of the predictive performance of a model. To address the above challenges, the paper proposes an autoencoder-based incremental learning method with drift detection (strAEm++DD). Our proposed method strAEm++DD leverages on the advantages of both incremental learning and drift detection. We conduct an experimental study using real-world and synthetic datasets with severe or extreme class imbalance, and provide an empirical analysis of strAEm++DD. We further conduct a comparative study, showing that the proposed method significantly outperforms existing baseline and advanced methods. | ['Marios M. Polycarpou', 'Kleanthis Malialis', 'Jin Li'] | 2023-05-15 | null | null | null | null | ['incremental-learning'] | ['methodology'] | [ 1.21089078e-01 -5.81383348e-01 -1.42521843e-01 -3.76928955e-01
-3.89182329e-01 -1.22856110e-01 3.42454374e-01 5.53985715e-01
-2.85463810e-01 8.52477849e-01 9.11890343e-03 -5.85793965e-02
-2.89224446e-01 -7.73389459e-01 -4.84119117e-01 -7.28985310e-01
-3.02539259e-01 3.00867021e-01 3.05989355e-01 -5.89740761e-02
2.68340915e-01 1.13665923e-01 -1.71602798e+00 1.91664964e-01
9.85506237e-01 1.22398269e+00 -1.30035535e-01 4.01938617e-01
-1.92753389e-01 8.77455115e-01 -9.46590483e-01 -2.49722674e-01
7.82302991e-02 -2.58356601e-01 -2.73363322e-01 -5.13875112e-02
-1.91828404e-02 -6.14138365e-01 -1.36758655e-01 8.99771869e-01
7.91989565e-01 1.31370440e-01 3.79288942e-01 -1.53745914e+00
-2.03302935e-01 5.68050027e-01 -9.66440499e-01 8.32456291e-01
1.36980012e-01 -9.72704515e-02 7.05356181e-01 -6.77515924e-01
2.74450779e-01 9.47971761e-01 9.27454889e-01 2.95844436e-01
-8.58693540e-01 -8.90412867e-01 5.74949086e-01 5.26958108e-01
-1.19893599e+00 -2.27447525e-01 1.06998837e+00 -4.49210107e-01
6.95110559e-01 -6.19151518e-02 4.51864868e-01 1.14086998e+00
8.66477117e-02 1.01175988e+00 6.90270364e-01 -2.82011658e-01
6.12504065e-01 -1.23272523e-01 2.62021035e-01 -1.35818332e-01
6.15055978e-01 -4.19098027e-02 -6.18362427e-01 -2.00643823e-01
4.51379538e-01 4.85857308e-01 -2.68568873e-01 5.87809496e-02
-9.27692890e-01 5.59684753e-01 5.13682403e-02 2.73240596e-01
-7.02097774e-01 -2.49773905e-01 7.67575324e-01 3.66345853e-01
8.04318845e-01 -2.84762323e-01 -5.26468635e-01 -5.16521573e-01
-1.00743020e+00 3.21695834e-01 5.60849190e-01 6.63712680e-01
2.90339649e-01 3.48173350e-01 1.48405924e-01 8.25196862e-01
6.61884472e-02 1.08028173e-01 9.36458349e-01 -4.59785014e-01
6.11973643e-01 7.10116208e-01 3.76795352e-01 -1.15053916e+00
-4.71137226e-01 -7.43051827e-01 -1.20471680e+00 -1.31385475e-01
6.19547516e-02 -3.62639844e-01 -6.39980078e-01 1.54604352e+00
7.03834295e-01 7.62749791e-01 1.42655894e-01 5.85308909e-01
6.42308414e-01 9.58804071e-01 1.82918310e-01 -7.04665899e-01
8.06042016e-01 -5.95728040e-01 -9.87050176e-01 -1.47572726e-01
4.62532133e-01 -3.81181836e-01 7.08142936e-01 6.01250470e-01
-6.96331322e-01 -5.22204101e-01 -1.02551973e+00 6.32510126e-01
1.16326965e-01 -1.57321468e-01 5.99207640e-01 4.15659845e-01
-5.32854855e-01 6.06489360e-01 -9.60225761e-01 -1.95113122e-01
5.56905568e-01 1.77892908e-01 -3.88634056e-02 9.97879356e-03
-1.29531193e+00 1.98857144e-01 6.12099648e-01 1.39563575e-01
-6.31023705e-01 -7.03864336e-01 -5.93574643e-01 1.43729243e-02
3.49251717e-01 -2.16824755e-01 1.50513303e+00 -1.00943446e+00
-9.86656964e-01 2.73928702e-01 -1.41259909e-01 -8.77693117e-01
6.85970902e-01 -5.84671795e-01 -7.83757687e-01 -1.89583927e-01
-1.45635664e-01 -1.29973248e-01 8.27917159e-01 -9.72221136e-01
-1.01737440e+00 -3.56110930e-01 -2.99313039e-01 1.86250031e-01
-8.61249924e-01 -5.41378893e-02 3.87032004e-03 -7.95793831e-01
-3.42535451e-02 -3.64112377e-01 -6.69924393e-02 -4.31943864e-01
-5.85389249e-02 -4.99846697e-01 1.20979118e+00 -6.57718539e-01
1.82513320e+00 -2.24898267e+00 -3.81581873e-01 -1.68336272e-01
4.84845713e-02 6.65420175e-01 2.82103360e-01 6.13404393e-01
-2.21300811e-01 -6.92884400e-02 -4.18805778e-01 -2.48231649e-01
-3.48211497e-01 2.07874432e-01 -4.28844333e-01 1.18492074e-01
1.63732097e-02 3.01178843e-01 -9.09927547e-01 -3.67909610e-01
-1.17851421e-01 2.51675695e-01 -3.34383875e-01 2.55677402e-01
-2.25534379e-01 2.28190064e-01 -4.81559843e-01 9.11513746e-01
7.43101001e-01 -2.80558050e-01 -7.93550089e-02 2.88977087e-01
5.81272282e-02 -2.10520774e-01 -1.54651964e+00 1.11239898e+00
-3.07441920e-01 6.62738383e-01 -4.06682581e-01 -1.38390386e+00
9.87313032e-01 3.56919765e-01 5.77755392e-01 -7.29948223e-01
6.74954727e-02 1.88462973e-01 -1.52929664e-01 -6.61726058e-01
3.97626668e-01 -9.51039195e-02 1.79364726e-01 3.86344045e-01
-2.47034535e-01 8.26441348e-01 2.55700111e-01 1.22064851e-01
1.33043921e+00 -4.04739939e-03 3.75754207e-01 1.56555071e-01
5.22095025e-01 -1.89817384e-01 1.35227680e+00 6.30305946e-01
-5.99912524e-01 3.40568572e-01 3.70266169e-01 -8.77791286e-01
-8.43616426e-01 -7.68482983e-01 -1.78115457e-01 7.23451197e-01
9.40174758e-02 -2.08554521e-01 -5.60117841e-01 -5.60819268e-01
1.59480855e-01 6.03839457e-01 -5.32088161e-01 -3.02081674e-01
-5.17605722e-01 -1.59065259e+00 3.66549492e-01 7.17646360e-01
7.31996000e-01 -9.79623020e-01 -7.55081356e-01 5.46837926e-01
-2.78293878e-01 -1.00444210e+00 6.34286404e-02 -6.00962602e-02
-1.24322426e+00 -9.27798927e-01 -4.71651882e-01 -5.38264155e-01
3.56031716e-01 2.68216640e-01 1.10851932e+00 9.03060287e-03
-2.49616385e-01 -6.78361133e-02 -5.77129185e-01 -9.64447021e-01
-1.82466462e-01 9.33004171e-02 2.81887501e-01 4.62430507e-01
8.94852281e-01 -8.61805439e-01 -8.75174880e-01 1.68471038e-01
-1.21447897e+00 -5.15042424e-01 3.84684056e-01 8.60976994e-01
4.97465014e-01 7.12458491e-01 1.40968502e+00 -8.46191049e-01
7.22564936e-01 -1.04602087e+00 -6.77524149e-01 -1.46510735e-01
-7.91199446e-01 -3.58622044e-01 6.91535413e-01 -6.38174117e-01
-1.09191620e+00 -2.74185061e-01 -2.15691715e-01 -3.01798791e-01
-2.38307670e-01 7.26377189e-01 -1.72544882e-01 5.24596393e-01
5.73144495e-01 2.31397316e-01 -3.94258648e-01 -6.74517334e-01
-3.76505792e-01 1.19898283e+00 5.51126957e-01 -1.81149542e-01
5.95804811e-01 5.32156587e-01 -3.26795906e-01 -7.85224020e-01
-6.88681364e-01 -4.45568919e-01 -3.99001420e-01 -2.60223210e-01
1.84136748e-01 -1.12445879e+00 -4.06690300e-01 9.45298135e-01
-8.40458989e-01 -9.40901190e-02 -2.19466373e-01 3.75811607e-01
-5.78846671e-02 2.49120444e-01 -4.40494746e-01 -1.18559670e+00
-4.38985974e-01 -5.24106801e-01 7.04922199e-01 5.92910349e-01
-1.26256809e-01 -8.76138270e-01 4.17234302e-01 5.26664741e-02
5.12859285e-01 6.81837261e-01 7.84169018e-01 -1.03220367e+00
-3.33011359e-01 -5.71262360e-01 6.51764795e-02 4.31775540e-01
2.24437609e-01 -1.06945001e-02 -8.76014113e-01 -4.13301498e-01
8.58954787e-02 -3.65747437e-02 7.22699404e-01 3.25514287e-01
1.27930868e+00 -1.45042703e-01 -2.95449883e-01 3.53575140e-01
1.59265387e+00 7.78174579e-01 4.96994644e-01 6.95985913e-01
5.18275082e-01 2.78146386e-01 7.55210102e-01 1.01717710e+00
4.52299267e-01 3.25846434e-01 5.03465593e-01 3.18472773e-01
3.24695706e-01 -7.36766830e-02 2.38832057e-01 1.00586271e+00
7.62689859e-02 -4.34009671e-01 -1.10298276e+00 9.26776052e-01
-2.06075215e+00 -1.01207030e+00 1.09580949e-01 2.33615088e+00
8.69661510e-01 4.68048543e-01 2.58334577e-01 9.26217198e-01
1.00328636e+00 1.48916662e-01 -9.06916380e-01 -1.52939871e-01
1.32244208e-03 -8.89923349e-02 -1.56135038e-01 -1.41166925e-01
-1.19721234e+00 4.20901030e-01 5.86705542e+00 6.20783567e-01
-1.28325832e+00 7.94392601e-02 7.66310096e-01 -1.75898820e-01
2.08775774e-01 -2.48897552e-01 -7.68483222e-01 1.06129420e+00
1.19863772e+00 -3.78645629e-01 -1.42101333e-01 1.05410647e+00
1.94224834e-01 -1.54980600e-01 -8.38042438e-01 1.22844934e+00
-4.22850549e-02 -1.06863987e+00 -1.87070385e-01 -3.96865726e-01
7.88108885e-01 -4.86372039e-02 -1.17672570e-01 4.17754769e-01
1.67314380e-01 -6.43926144e-01 2.76200444e-01 4.11048621e-01
3.39044362e-01 -1.12253582e+00 1.12246418e+00 6.29416347e-01
-8.42447758e-01 -4.71300185e-01 -2.27998346e-01 -3.63220870e-01
2.12625682e-01 1.33520651e+00 -5.06516576e-01 6.19209886e-01
9.56266880e-01 8.69475365e-01 -4.06057954e-01 1.40422571e+00
9.88172144e-02 1.15599883e+00 -3.37354541e-01 6.69650361e-02
-1.68590039e-01 2.10075527e-01 5.94489098e-01 1.00191784e+00
3.65485489e-01 -1.55062536e-02 7.86200389e-02 1.56508878e-01
-2.73683101e-01 1.90205798e-01 -5.14547825e-01 1.10154219e-01
9.15288210e-01 8.94960046e-01 -5.51412404e-01 -5.54959714e-01
-4.80578184e-01 7.68115699e-01 -5.49642667e-02 3.19201112e-01
-6.91208482e-01 -6.74208343e-01 5.38214087e-01 5.93026355e-02
2.77936906e-01 8.24423209e-02 -4.67138514e-02 -1.30090070e+00
5.33557713e-01 -9.51153100e-01 6.11187041e-01 -3.82223368e-01
-1.28576112e+00 5.29447377e-01 -2.97459289e-02 -1.50086725e+00
-4.00242150e-01 -1.33103818e-01 -8.61609161e-01 3.59144837e-01
-1.53743947e+00 -5.63585103e-01 -6.68538690e-01 3.97359908e-01
7.56785750e-01 -2.21178234e-01 5.27207971e-01 8.02931845e-01
-1.01941538e+00 5.94010890e-01 5.75648189e-01 1.40527442e-01
6.81526721e-01 -1.09836900e+00 5.11126399e-01 1.12530804e+00
-1.54222056e-01 3.08257103e-01 7.47832000e-01 -6.57558322e-01
-9.31355298e-01 -1.38618839e+00 6.06963515e-01 -1.27596751e-01
4.26474214e-01 -1.03719361e-01 -1.31261945e+00 7.06340730e-01
-3.44317518e-02 7.51327053e-02 7.58593917e-01 1.04094511e-02
-1.19137876e-01 -6.10245287e-01 -1.11026144e+00 2.07210153e-01
8.16375554e-01 -3.97532992e-02 -5.92695415e-01 2.32179552e-01
6.01825297e-01 -3.60570461e-01 -6.74497843e-01 7.38422692e-01
4.09450978e-01 -1.08512962e+00 6.25472784e-01 -6.01628482e-01
4.45284337e-01 -2.58557647e-01 -1.63422115e-02 -1.49492359e+00
5.60015440e-02 -4.68569309e-01 -7.97519028e-01 1.53313124e+00
-5.09401448e-02 -6.49670362e-01 6.75936759e-01 1.89661905e-01
1.94356322e-01 -7.76199877e-01 -9.61214364e-01 -9.43392277e-01
-1.29184037e-01 -4.24812496e-01 9.01741385e-01 1.22535634e+00
-2.19961122e-01 1.09412640e-01 -5.52257240e-01 2.16495335e-01
5.87323248e-01 1.19566917e-01 6.82002783e-01 -1.66665685e+00
-1.65783659e-01 2.78371363e-03 -7.61124015e-01 -7.11001456e-01
-2.60838032e-01 -2.52838910e-01 -1.25695303e-01 -1.26319587e+00
1.76827312e-02 -2.97789961e-01 -9.38270330e-01 2.22755000e-01
-5.38949311e-01 8.52494501e-03 -2.93066442e-01 4.63500857e-01
-6.59293711e-01 7.20718622e-01 2.99416900e-01 9.18851197e-02
-3.79965365e-01 3.02774966e-01 -6.42032385e-01 8.25966835e-01
1.00331926e+00 -6.05015218e-01 -5.65948546e-01 -4.19979662e-01
3.19971591e-01 -4.48346771e-02 9.91336722e-03 -1.34371018e+00
1.86137542e-01 -4.54068668e-02 5.23747921e-01 -8.70444357e-01
-2.88583755e-01 -6.30549908e-01 -1.49672911e-01 4.50743020e-01
2.55208500e-02 4.57997143e-01 3.21892500e-01 9.02651131e-01
-5.98931432e-01 2.34657079e-02 5.11878133e-01 1.60416573e-01
-8.92090201e-01 3.42270643e-01 -9.19013619e-02 1.10281430e-01
1.12598002e+00 4.22174623e-03 -2.36229420e-01 -5.01097918e-01
-4.58363861e-01 4.98059720e-01 8.54151696e-02 6.63311422e-01
5.53867936e-01 -1.44688523e+00 -7.27953076e-01 2.07300335e-01
1.62432522e-01 4.24996704e-01 5.50085962e-01 7.05200970e-01
-3.78643066e-01 1.36549678e-02 -3.85918990e-02 -6.66145325e-01
-1.11158192e+00 6.27383411e-01 1.00913607e-01 -2.42093951e-01
-7.10772693e-01 6.55285478e-01 -1.86858401e-01 -4.30248305e-03
5.46407163e-01 -2.86541693e-02 -3.35915148e-01 4.25666273e-01
1.05679727e+00 5.77032089e-01 4.04771209e-01 -3.62587363e-01
-3.25288922e-01 1.83334917e-01 -4.02157873e-01 2.07995668e-01
1.48046029e+00 -1.61345080e-01 1.13218889e-01 7.60047138e-01
7.35248923e-01 -3.41186404e-01 -1.24403858e+00 -5.34919024e-01
2.98376381e-01 -5.46834469e-01 1.04433715e-01 -5.99026144e-01
-1.08033669e+00 8.51255059e-01 9.55867529e-01 3.14789265e-01
1.62738776e+00 -5.69563448e-01 1.15448773e+00 3.11737657e-01
2.75473893e-01 -1.25486934e+00 6.38938844e-02 2.43421927e-01
3.55939478e-01 -1.46858156e+00 -1.28845632e-01 1.26104623e-01
-4.60568309e-01 8.20044935e-01 6.49271667e-01 -1.00281850e-01
7.13815510e-01 1.65450767e-01 2.77165234e-01 1.96837679e-01
-1.01422501e+00 1.82640463e-01 -3.98047954e-01 5.01012146e-01
2.77861506e-01 -1.10928625e-01 -4.39433187e-01 7.84798086e-01
1.26711741e-01 4.11708206e-01 5.78924179e-01 1.41681683e+00
-3.80500883e-01 -1.01634884e+00 -4.98349607e-01 5.47732055e-01
-7.83505142e-01 2.12967977e-01 -5.51397633e-03 5.80196261e-01
1.68439060e-01 8.45216274e-01 1.95402667e-01 -5.24308980e-01
3.08822274e-01 2.93667316e-01 -2.26655647e-01 -1.54534891e-01
-2.28584111e-01 -9.50867776e-03 -2.67640620e-01 -3.08612376e-01
-4.42005605e-01 -9.03083563e-01 -1.12776446e+00 -1.65325120e-01
-2.39720747e-01 1.04499623e-01 6.22166812e-01 9.30026114e-01
5.32114029e-01 7.99944401e-01 9.11003053e-01 -3.57366651e-01
-6.59411371e-01 -1.11811054e+00 -5.82502007e-01 5.14455438e-01
6.60983443e-01 -8.25395226e-01 -4.30079937e-01 1.18456602e-01] | [7.375077724456787, 2.85905385017395] |
96610dd2-01a1-49a0-90a2-a52dabc809a1 | handheld-mobile-photography-in-very-low-light | 1910.11336 | null | https://arxiv.org/abs/1910.11336v1 | https://arxiv.org/pdf/1910.11336v1.pdf | Handheld Mobile Photography in Very Low Light | Taking photographs in low light using a mobile phone is challenging and rarely produces pleasing results. Aside from the physical limits imposed by read noise and photon shot noise, these cameras are typically handheld, have small apertures and sensors, use mass-produced analog electronics that cannot easily be cooled, and are commonly used to photograph subjects that move, like children and pets. In this paper we describe a system for capturing clean, sharp, colorful photographs in light as low as 0.3~lux, where human vision becomes monochromatic and indistinct. To permit handheld photography without flash illumination, we capture, align, and combine multiple frames. Our system employs "motion metering", which uses an estimate of motion magnitudes (whether due to handshake or moving objects) to identify the number of frames and the per-frame exposure times that together minimize both noise and motion blur in a captured burst. We combine these frames using robust alignment and merging techniques that are specialized for high-noise imagery. To ensure accurate colors in such low light, we employ a learning-based auto white balancing algorithm. To prevent the photographs from looking like they were shot in daylight, we use tone mapping techniques inspired by illusionistic painting: increasing contrast, crushing shadows to black, and surrounding the scene with darkness. All of these processes are performed using the limited computational resources of a mobile device. Our system can be used by novice photographers to produce shareable pictures in a few seconds based on a single shutter press, even in environments so dim that humans cannot see clearly. | ['Tim Brooks', 'Yun-Ta Tsai', 'Qiurui He', 'Nikhil Karnad', 'Marc Levoy', 'Dillon Sharlet', 'Orly Liba', 'Jonathan T. Barron', 'Yael Pritch', 'Samuel W. Hasinoff', 'Kiran Murthy', 'Tianfan Xue', 'Ryan Geiss'] | 2019-10-24 | null | null | null | null | ['tone-mapping'] | ['computer-vision'] | [ 6.47656620e-01 -4.77474838e-01 3.04266274e-01 -4.57866602e-02
-2.52792627e-01 -9.26203728e-01 2.26948038e-01 -4.44708347e-01
-5.91109633e-01 8.08342218e-01 -2.43778735e-01 -3.08067858e-01
3.29699129e-01 -4.94048119e-01 -4.79465157e-01 -7.16168940e-01
4.23474908e-01 -2.49546513e-01 5.23513913e-01 1.34219155e-01
2.76491195e-01 4.69350815e-01 -1.80167556e+00 -5.51335365e-02
7.38399863e-01 6.28949642e-01 4.88764137e-01 1.11106229e+00
1.96772441e-02 1.09955668e+00 -8.96924794e-01 -3.10230255e-01
4.91649270e-01 -7.61794806e-01 -1.80977881e-01 4.17963207e-01
7.52435982e-01 -8.89487505e-01 -3.76036227e-01 1.02837837e+00
3.38018566e-01 2.37980485e-01 2.46993408e-01 -9.15121317e-01
-4.78403568e-01 -3.22866887e-01 -7.54911244e-01 2.55429089e-01
7.84647048e-01 6.79331839e-01 7.60128424e-02 -5.30688763e-01
4.66979921e-01 8.64369392e-01 6.20936632e-01 5.47786355e-01
-1.26020217e+00 -6.81995690e-01 -1.76208839e-01 9.93145257e-02
-1.45173442e+00 -8.23932707e-01 6.17225468e-01 -2.65072733e-01
8.44727516e-01 5.67910731e-01 8.48886132e-01 7.91545510e-01
4.70399737e-01 -7.77051374e-02 1.44120169e+00 -6.34904683e-01
2.70793736e-01 3.68540585e-01 -5.27431726e-01 6.15574181e-01
2.86628902e-01 -7.51693025e-02 -4.80935007e-01 -9.66833159e-03
9.34857249e-01 8.61804187e-02 -8.55758071e-01 6.15172051e-02
-1.14643872e+00 1.96273714e-01 1.65483922e-01 6.93375245e-02
-1.33443147e-01 2.95212239e-01 -3.34786475e-01 1.43315271e-01
2.76549071e-01 4.34101611e-01 -1.46766365e-01 -3.77678633e-01
-8.98919761e-01 -3.85483295e-01 6.46353960e-01 8.17317486e-01
4.83993143e-01 9.18221567e-03 3.71663898e-01 7.52644598e-01
4.50003855e-02 7.13578999e-01 3.55114043e-01 -1.34697008e+00
2.14092642e-01 2.00540349e-01 5.86671114e-01 -7.81339526e-01
-1.29921764e-01 3.92213196e-01 -5.29289544e-01 1.08111572e+00
6.44558609e-01 -4.15324986e-01 -8.21923375e-01 1.30003273e+00
3.50359917e-01 2.87488192e-01 -2.12618083e-01 1.26455796e+00
2.92627692e-01 7.67401338e-01 -2.84809560e-01 -6.55865014e-01
1.37583303e+00 -5.50206244e-01 -8.74059319e-01 -2.66048759e-01
3.87765728e-02 -1.32107294e+00 1.50117302e+00 7.29703367e-01
-1.39570320e+00 -4.56989080e-01 -1.35493076e+00 -3.59551758e-01
-3.80287804e-02 2.95061897e-02 2.79790968e-01 1.03033268e+00
-9.91998315e-01 5.97197950e-01 -8.29316020e-01 -3.81662726e-01
5.55908196e-02 1.06400602e-01 -1.28932074e-01 -3.00815374e-01
-5.19439995e-01 1.12958455e+00 -1.70182392e-01 3.51563245e-02
-3.69187266e-01 -4.95153159e-01 -5.52106917e-01 -8.45965818e-02
3.29199642e-01 -7.67749012e-01 1.16597998e+00 -1.37330270e+00
-2.10395622e+00 8.58360708e-01 -3.73705477e-01 -1.15215920e-01
5.42549074e-01 -3.43904138e-01 -4.71983373e-01 6.76113307e-01
-3.82008672e-01 3.91309261e-01 1.12095761e+00 -1.20963526e+00
-3.94126713e-01 8.73529539e-03 9.71668437e-02 5.39354622e-01
-1.81838080e-01 3.47326100e-01 -5.23498118e-01 -2.20143497e-01
-9.25708190e-02 -7.74459064e-01 -6.16832357e-03 5.04345357e-01
-6.21466115e-02 5.98941326e-01 1.08680403e+00 -5.48595309e-01
8.77248704e-01 -2.24963593e+00 -5.03139913e-01 -1.76879302e-01
-6.36952892e-02 3.65332603e-01 1.71962351e-01 2.32685477e-01
2.61541456e-01 -1.99717551e-01 -2.22662821e-01 -1.72149003e-01
-5.65416753e-01 1.14600651e-01 -2.16361359e-01 6.02167904e-01
-2.12100312e-01 3.89552385e-01 -8.73001039e-01 -4.41896588e-01
6.22627139e-01 6.22256637e-01 -3.72161381e-02 2.60257572e-01
1.48524404e-01 2.34594494e-01 2.62282759e-01 6.97341383e-01
7.70882487e-01 7.05997795e-02 2.15099588e-01 -2.22502485e-01
-4.35109019e-01 3.92066911e-02 -1.26261556e+00 1.36902094e+00
-6.89603746e-01 1.14238060e+00 2.46410057e-01 1.28547363e-02
7.87921548e-01 8.60575736e-02 1.40873045e-01 -6.92581296e-01
1.81360900e-01 1.44360960e-01 -4.07233089e-01 -8.05180073e-01
6.59496605e-01 -4.28969532e-01 3.80359262e-01 5.96923649e-01
-5.27818680e-01 -7.68636405e-01 1.58309355e-01 3.47019136e-02
9.20794189e-01 1.59168750e-01 3.56064171e-01 1.12382025e-01
8.30561370e-02 -1.74833342e-01 3.04159641e-01 5.26122034e-01
-1.00620598e-01 1.18045259e+00 1.93373431e-02 -3.26826960e-01
-1.05749965e+00 -1.08158600e+00 9.35588628e-02 6.62876844e-01
5.77223957e-01 1.26397675e-02 -8.57948065e-01 6.55183792e-02
-4.32356060e-01 6.47767246e-01 -1.20518319e-02 5.53316213e-02
-5.49220622e-01 -5.06557167e-01 1.36568442e-01 9.38436985e-02
5.17600298e-01 -8.43589246e-01 -1.36528063e+00 1.14748172e-01
-1.19408138e-01 -1.06919134e+00 -6.88125551e-01 -5.71081378e-02
-6.47056997e-01 -1.19992399e+00 -8.86867285e-01 -5.00341177e-01
7.49726772e-01 1.00434422e+00 8.87558341e-01 9.42901522e-02
-7.47420371e-01 5.42587042e-01 -2.93128610e-01 -3.68308514e-01
-8.27571601e-02 -7.69240916e-01 2.75678597e-02 1.86958481e-02
2.27549583e-01 -5.84304154e-01 -9.74112272e-01 3.17842633e-01
-9.90115106e-01 1.57965094e-01 3.52821440e-01 3.99801224e-01
2.86993146e-01 2.08498493e-01 -1.65535539e-01 -3.23365390e-01
3.61450702e-01 1.10078454e-01 -8.64696801e-01 7.74697885e-02
-2.24084824e-01 -5.79340935e-01 7.29130864e-01 -8.29566658e-01
-1.24835527e+00 5.14966175e-02 4.79139209e-01 -5.11781335e-01
-3.10626149e-01 -5.30449212e-01 3.83685902e-02 -4.93405491e-01
8.49544406e-01 5.68788014e-02 -2.65679788e-02 -1.11789539e-01
2.93281049e-01 8.42877686e-01 8.64581406e-01 5.39061725e-02
7.83602595e-01 9.95917022e-01 -1.57123625e-01 -1.24843752e+00
-4.51414406e-01 -2.33226672e-01 -3.79540056e-01 -6.07309580e-01
9.57960129e-01 -7.80213237e-01 -9.60482538e-01 7.15248287e-01
-9.60916936e-01 -5.06938398e-01 -3.79399896e-01 7.34608531e-01
-2.61461645e-01 5.29849589e-01 -6.60401344e-01 -1.09750879e+00
-1.43498048e-01 -8.19632947e-01 8.29940856e-01 9.62098300e-01
-2.31756762e-01 -6.70130610e-01 -1.10224783e-01 3.83109123e-01
4.37910527e-01 1.62737519e-01 3.56052876e-01 7.84025788e-01
-8.67653728e-01 -1.12172281e-02 -1.03897028e-01 4.11189348e-01
4.28121954e-01 3.26117009e-01 -1.22070444e+00 -1.67156652e-01
2.39125371e-01 -1.90482453e-01 5.02844512e-01 5.12700856e-01
9.57078397e-01 -1.74216107e-01 -8.48581195e-02 8.08283806e-01
1.60042596e+00 7.29642153e-01 9.56720412e-01 3.23699057e-01
5.85665584e-01 4.45164144e-01 4.93920147e-01 2.25446612e-01
1.62819959e-02 6.65380597e-01 2.10199296e-01 -5.26204526e-01
-4.11622643e-01 6.10954985e-02 4.14751530e-01 4.57501501e-01
-1.46670058e-01 -3.46460164e-01 -4.12447423e-01 3.23010355e-01
-1.06231356e+00 -1.11908805e+00 -4.03660238e-01 2.55680108e+00
9.00232852e-01 -2.17262059e-02 -9.62093822e-04 2.44655199e-02
7.46194124e-01 -6.37682602e-02 -4.65414703e-01 -4.67606753e-01
-2.16216192e-01 2.82613933e-01 9.59653318e-01 7.26382673e-01
-7.29742050e-01 6.59487069e-01 6.78461695e+00 1.75812677e-01
-1.32826281e+00 -1.26844989e-02 6.14602685e-01 -8.25776756e-01
-2.70754278e-01 7.85022527e-02 -2.31384978e-01 8.50960195e-01
6.34830117e-01 8.81158113e-02 7.70285964e-01 4.67046559e-01
5.88263810e-01 -1.03618133e+00 -7.47015774e-01 1.43326473e+00
2.46357918e-01 -7.74268270e-01 -4.62228775e-01 -8.15829784e-02
6.65967286e-01 -3.16763669e-01 1.50276914e-01 -5.95819235e-01
1.38625935e-01 -9.44170415e-01 7.03731060e-01 7.11182177e-01
1.02046108e+00 -6.38664663e-01 2.70094752e-01 1.59223184e-01
-8.65252316e-01 3.03219631e-02 -6.18694603e-01 -4.24599975e-01
5.86336553e-01 6.16424978e-01 -6.54244006e-01 -9.40212905e-02
6.89136147e-01 4.37673926e-03 -2.55353898e-01 1.21376598e+00
-3.01281035e-01 2.92493999e-01 -6.00250065e-01 -1.00851044e-01
-9.90467742e-02 -6.39329672e-01 3.82891893e-01 1.03030813e+00
6.22573733e-01 7.42788911e-01 -2.94692159e-01 6.45835459e-01
1.05666220e-01 -3.75507355e-01 -6.08282745e-01 4.09475476e-01
4.42713916e-01 1.33759642e+00 -9.61264670e-01 -3.38333398e-01
-5.60703337e-01 1.59195316e+00 -4.65857029e-01 6.27125978e-01
-7.38661051e-01 -8.02185595e-01 5.87978244e-01 3.16512257e-01
1.21728167e-01 -3.89219999e-01 -4.17512625e-01 -1.14444947e+00
3.12385768e-01 -6.98537290e-01 -9.85201076e-02 -1.56982136e+00
-6.70260429e-01 5.55905759e-01 -1.19302295e-01 -1.44083858e+00
6.42746538e-02 -4.26964849e-01 -8.17712009e-01 8.37579906e-01
-1.29640603e+00 -4.86700535e-01 -6.66200817e-01 7.82008648e-01
5.85916340e-01 3.93716633e-01 6.02120936e-01 1.89386278e-01
-2.79918462e-01 1.99267134e-01 1.91000432e-01 -4.00836170e-01
1.00486982e+00 -1.01462948e+00 1.85790688e-01 1.28941321e+00
1.62146747e-01 4.85893697e-01 7.99946845e-01 -4.29170102e-01
-1.35636675e+00 -5.04082978e-01 6.17079616e-01 -4.46032584e-01
2.07919285e-01 -4.45071936e-01 -8.90420794e-01 1.92967087e-01
4.52216268e-01 -6.48828819e-02 5.23857236e-01 -5.48174918e-01
-8.86487961e-02 -4.13303137e-01 -1.18756068e+00 7.76637912e-01
7.51248598e-01 -6.24848366e-01 -4.40398037e-01 3.18895698e-01
4.13432181e-01 -4.49395269e-01 -1.86454892e-01 -3.39766294e-01
8.80233407e-01 -1.24539387e+00 8.63925993e-01 3.04809213e-01
1.78568959e-02 -7.20197499e-01 2.63294548e-01 -1.22954869e+00
-5.71224876e-02 -1.15152550e+00 4.85824436e-01 1.13718987e+00
4.33746204e-02 -6.89444602e-01 5.11504352e-01 8.87751818e-01
1.50346652e-01 -5.47538064e-02 -7.34377325e-01 -6.63147330e-01
-7.00213313e-01 -1.33638144e-01 1.44203380e-01 7.77838826e-01
1.32979274e-01 1.13517635e-01 -7.19354928e-01 1.60516515e-01
7.30406880e-01 -2.25328356e-02 7.86343455e-01 -7.98367620e-01
-3.37475687e-01 4.81742108e-03 -1.89464912e-01 -8.50001216e-01
-4.70345378e-01 -9.37324297e-03 2.03956457e-04 -1.28732693e+00
4.52684946e-02 -7.63073713e-02 2.46570438e-01 3.92720103e-02
-2.64454097e-01 6.96375728e-01 5.39101958e-01 2.45963469e-01
-2.65637189e-01 -1.68669354e-02 9.20414686e-01 2.53010243e-01
-5.10752797e-01 -2.42213085e-01 -5.19050837e-01 1.01963127e+00
6.42745912e-01 -2.91574389e-01 -6.05814874e-01 -6.13053143e-01
1.46545738e-01 -3.24706808e-02 4.14886653e-01 -1.23727632e+00
4.34134275e-01 -3.03758264e-01 7.64910281e-01 -1.83850989e-01
6.01524293e-01 -1.07728601e+00 7.08775282e-01 5.12937963e-01
1.03111692e-01 -5.04723638e-02 7.04535246e-02 3.31887484e-01
8.77005011e-02 -2.90019810e-01 1.27302182e+00 -4.50305164e-01
-5.30949116e-01 -2.95531362e-01 -6.91210508e-01 -2.81670183e-01
1.23349690e+00 -7.12503195e-01 -6.21751308e-01 -8.28895628e-01
-1.58228368e-01 -2.93438405e-01 1.20086563e+00 3.04752737e-02
7.08442032e-01 -1.04329324e+00 -1.50742099e-01 2.42643848e-01
-3.14685941e-01 -2.14420453e-01 5.76052547e-01 6.85779572e-01
-1.28203535e+00 -8.23044851e-02 -2.96532154e-01 -4.10891622e-01
-1.55677402e+00 5.91941595e-01 3.14333588e-01 5.20512760e-01
-8.11209917e-01 7.93951035e-01 7.41562024e-02 6.17988288e-01
5.22882037e-04 -4.33577418e-01 2.78796494e-01 -2.32164770e-01
1.08819652e+00 6.59044206e-01 -7.91554749e-02 -1.94835469e-01
-2.62075394e-01 8.62149000e-01 2.41379023e-01 -4.19264853e-01
9.22093391e-01 -6.88199461e-01 1.51728898e-01 4.72700536e-01
9.41211700e-01 5.35283208e-01 -1.70470548e+00 2.36331925e-01
-8.30809772e-01 -1.10210156e+00 -8.88032094e-02 -5.93460321e-01
-6.89079702e-01 7.03069389e-01 7.11379349e-01 2.99373031e-01
1.63679445e+00 -3.29987228e-01 7.82355607e-01 8.68177190e-02
3.32058936e-01 -1.31722808e+00 1.72043324e-01 -6.29601330e-02
3.23896289e-01 -8.14806938e-01 1.31181553e-01 -2.83079028e-01
-5.81915677e-01 1.22263467e+00 5.22818625e-01 5.75335696e-02
-8.70065298e-03 6.57562375e-01 4.93812770e-01 1.74682751e-01
-6.45126879e-01 -2.90076673e-01 -1.45924762e-01 8.66471410e-01
1.02685243e-01 -1.66425943e-01 -1.08913258e-01 -3.15260679e-01
-2.25663647e-01 1.16774507e-01 1.22008884e+00 9.31111574e-01
-6.87531710e-01 -6.58763349e-01 -9.62331176e-01 1.77469522e-01
-5.52102685e-01 1.93145107e-02 -3.63985211e-01 5.87930143e-01
2.93645024e-01 1.35115278e+00 2.55850911e-01 -4.81363572e-02
3.64217132e-01 -2.48343229e-01 7.09494531e-01 -1.59656078e-01
-1.62591815e-01 2.02485681e-01 1.76410154e-02 -7.28129506e-01
-6.22659564e-01 -5.32411754e-01 -7.34813690e-01 -4.72580045e-01
-1.87812105e-01 -3.70889515e-01 7.05430627e-01 5.75244784e-01
-2.50375662e-02 2.58921087e-01 6.09644234e-01 -9.25992966e-01
6.51473850e-02 -6.74449742e-01 -7.89427698e-01 3.15205872e-01
4.69633400e-01 -3.21092218e-01 -6.60655558e-01 5.18421412e-01] | [10.545787811279297, -2.5930142402648926] |
77629c39-f65a-4a82-991e-5ec8f2d8d3c9 | neuron-coverage-guided-domain-generalization | 2103.00229 | null | https://arxiv.org/abs/2103.00229v2 | https://arxiv.org/pdf/2103.00229v2.pdf | Neuron Coverage-Guided Domain Generalization | This paper focuses on the domain generalization task where domain knowledge is unavailable, and even worse, only samples from a single domain can be utilized during training. Our motivation originates from the recent progresses in deep neural network (DNN) testing, which has shown that maximizing neuron coverage of DNN can help to explore possible defects of DNN (i.e., misclassification). More specifically, by treating the DNN as a program and each neuron as a functional point of the code, during the network training we aim to improve the generalization capability by maximizing the neuron coverage of DNN with the gradient similarity regularization between the original and augmented samples. As such, the decision behavior of the DNN is optimized, avoiding the arbitrary neurons that are deleterious for the unseen samples, and leading to the trained DNN that can be better generalized to out-of-distribution samples. Extensive studies on various domain generalization tasks based on both single and multiple domain(s) setting demonstrate the effectiveness of our proposed approach compared with state-of-the-art baseline methods. We also analyze our method by conducting visualization based on network dissection. The results further provide useful evidence on the rationality and effectiveness of our approach. | ['Shiqi Wang', 'Yang Liu', 'Xiaofei Xie', 'Haoliang Li', 'Chris Xing Tian'] | 2021-02-27 | null | null | null | null | ['dnn-testing'] | ['adversarial'] | [ 2.38469869e-01 1.08160302e-01 -1.67793900e-01 -4.18359101e-01
1.82058647e-01 -7.33673751e-01 1.54516205e-01 -1.36219323e-01
-1.28565609e-01 8.86420250e-01 -3.58442277e-01 -4.68766719e-01
-2.91568786e-01 -1.06747746e+00 -1.00641716e+00 -5.50755084e-01
1.72102869e-01 2.80908942e-01 2.09370598e-01 -1.01358138e-01
2.31318682e-01 5.61015248e-01 -1.74408853e+00 2.14090958e-01
1.37146723e+00 1.08375490e+00 2.82298386e-01 -2.50043441e-02
-5.41907139e-02 5.35143614e-01 -1.18058300e+00 -2.83709407e-01
3.51301134e-01 -2.75361300e-01 -5.62308967e-01 7.76075125e-02
3.64762127e-01 -5.00875115e-01 -3.32590520e-01 1.67488658e+00
2.78665841e-01 3.67156044e-02 5.48870444e-01 -1.54424930e+00
-9.42476749e-01 7.31200278e-01 -3.55813682e-01 1.32495597e-01
-1.25801027e-01 8.43069851e-02 5.86741924e-01 -5.41977286e-01
5.73225260e-01 1.26327538e+00 4.30905968e-01 9.58498955e-01
-1.28454661e+00 -1.04554224e+00 3.05353820e-01 9.16169062e-02
-1.29989624e+00 -1.09484084e-01 8.96872818e-01 -4.88025218e-01
7.63953090e-01 -2.30010614e-01 5.74839950e-01 1.47932804e+00
-2.33069941e-01 8.63872707e-01 6.37086213e-01 -2.89409548e-01
5.34969985e-01 4.15306926e-01 3.50499064e-01 3.59060109e-01
7.86461592e-01 2.85027504e-01 -1.91481441e-01 5.14843762e-02
7.03504324e-01 2.24946961e-01 -3.27047229e-01 -5.61808825e-01
-7.68797815e-01 6.23049676e-01 2.93657631e-01 1.21477813e-01
-1.60248086e-01 -2.84006238e-01 5.43912649e-01 2.90251851e-01
2.93467760e-01 5.02955616e-01 -5.81267059e-01 -6.07747063e-02
-8.52089763e-01 2.37543628e-01 5.62094331e-01 1.31311071e+00
8.80629301e-01 3.51631999e-01 -3.47807892e-02 8.43572199e-01
3.73268984e-02 2.29122341e-01 6.58977032e-01 -6.89909220e-01
7.06101120e-01 1.37326074e+00 -1.85538873e-01 -7.00970590e-01
-1.41914621e-01 -7.39369631e-01 -8.72203469e-01 4.70657796e-01
3.67897153e-01 -4.51175272e-01 -7.67412484e-01 1.89691591e+00
7.87096620e-02 2.89736032e-01 3.44958782e-01 6.95423841e-01
6.44177675e-01 4.53614771e-01 -1.16763011e-01 2.38023564e-01
9.45769489e-01 -8.23277116e-01 -3.84523124e-01 -2.93505639e-01
1.00840735e+00 2.47716717e-02 1.19358099e+00 7.55183578e-01
-7.38102436e-01 -7.80198693e-01 -1.44796801e+00 4.40462530e-01
-4.88105029e-01 2.67137259e-01 3.30728412e-01 9.31162655e-01
-7.26399601e-01 7.98823595e-01 -5.92841029e-01 -2.40127206e-01
9.99887466e-01 4.07115161e-01 -1.98643640e-01 -4.93080169e-01
-1.19420302e+00 5.06773770e-01 8.58791888e-01 -1.87348425e-02
-1.21411467e+00 -9.93191659e-01 -5.61202168e-01 2.30203271e-01
3.30573201e-01 -3.53135049e-01 9.89139140e-01 -1.18940914e+00
-9.55831170e-01 5.21514416e-01 1.42522335e-01 -3.50791752e-01
4.38390166e-01 5.65572781e-03 -6.29199564e-01 -7.06334785e-02
-6.99716136e-02 5.48631132e-01 3.71131718e-01 -1.29736233e+00
-7.16723680e-01 -4.68968093e-01 2.39003778e-01 1.30376546e-04
-9.04054821e-01 -4.05426681e-01 -1.84454408e-03 -6.27994359e-01
8.23164638e-03 -6.15083158e-01 3.10904205e-01 7.03548118e-02
-6.25566959e-01 -4.17393208e-01 1.14434183e+00 -5.78700185e-01
1.45350957e+00 -2.45932126e+00 -1.06783479e-01 3.66979241e-01
2.54017919e-01 5.13909936e-01 -1.82952464e-01 -6.23055436e-02
-3.53986859e-01 4.96316701e-01 -2.67119616e-01 -6.61616493e-03
-4.67377156e-02 2.61679024e-01 -3.56674433e-01 2.58222848e-01
3.07291418e-01 5.52329779e-01 -6.87502503e-01 -2.09805578e-01
-1.49637595e-01 5.07894251e-03 -6.81172252e-01 1.40370905e-01
-3.01345706e-01 2.90101379e-01 -3.20776373e-01 9.13101792e-01
8.95283699e-01 -9.09416750e-02 2.48901889e-01 5.94474152e-02
2.71446079e-01 1.51540548e-01 -1.15148318e+00 1.38107634e+00
-1.74196288e-01 7.54830599e-01 -2.23458633e-01 -1.50767601e+00
1.31154096e+00 7.11097792e-02 -1.44213095e-01 -7.23301291e-01
-1.39581829e-01 3.50846291e-01 3.69995683e-01 -4.36702818e-01
1.75009921e-01 1.84396192e-01 1.61372066e-01 3.38821530e-01
1.30809292e-01 5.67330480e-01 2.56992817e-01 -1.45930976e-01
1.10717976e+00 1.54862087e-02 6.37477487e-02 -5.25996566e-01
2.21651495e-01 4.07934561e-03 7.97267795e-01 8.06485772e-01
-2.23805889e-01 1.95562184e-01 1.04466105e+00 -1.36039108e-01
-1.15379751e+00 -1.01666212e+00 -2.78187364e-01 9.13801849e-01
1.65756792e-01 3.56042653e-01 -8.92378032e-01 -9.35488224e-01
9.22617763e-02 1.06721842e+00 -7.96532393e-01 -6.72885716e-01
-4.07429785e-01 -4.93360758e-01 9.36261654e-01 9.54442322e-01
7.65572011e-01 -9.68302727e-01 -4.27029639e-01 -4.65144031e-02
4.66614626e-02 -8.94545496e-01 2.43240707e-02 2.65860587e-01
-1.37756085e+00 -1.10513926e+00 -6.64196908e-01 -1.03402793e+00
1.03426003e+00 -4.46257070e-02 9.73814428e-01 2.16282547e-01
1.24719985e-01 -2.07189664e-01 -2.78652608e-01 -3.88501465e-01
-5.58611512e-01 7.83329550e-03 1.67143956e-01 -2.70832151e-01
6.56113505e-01 -7.97046006e-01 -3.67897421e-01 7.00514853e-01
-1.08273256e+00 -3.01063716e-01 5.11626720e-01 8.82622242e-01
2.84550816e-01 6.39726400e-01 9.13696647e-01 -1.13479233e+00
8.85399699e-01 -7.48503625e-01 -8.10395479e-01 5.88093340e-01
-7.66966522e-01 4.63424623e-03 1.02068162e+00 -9.93086994e-01
-1.07490146e+00 -3.50390226e-01 3.67690802e-01 -6.22045100e-01
-3.83622378e-01 4.42110538e-01 -8.21279943e-01 1.13376319e-01
1.06102180e+00 4.55554366e-01 -4.55195382e-02 -5.81937194e-01
-1.84913456e-01 6.03220582e-01 2.56190866e-01 -8.17830741e-01
4.43966359e-01 1.33169353e-01 -2.22876370e-01 -6.31256998e-01
-4.26635563e-01 8.37256387e-02 -4.60477233e-01 -8.53304565e-02
3.34259659e-01 -7.53046989e-01 -7.02362955e-01 5.32123268e-01
-1.22951007e+00 -5.13154805e-01 -1.74289510e-01 2.69945323e-01
-5.43763340e-02 3.16611052e-01 -2.75257260e-01 -8.84184003e-01
2.20885664e-01 -1.34374070e+00 4.24028784e-01 3.65339518e-01
-1.78598285e-01 -1.17627549e+00 -4.08509433e-01 -7.58552551e-02
1.95436090e-01 1.17068194e-01 1.59542513e+00 -1.30956829e+00
-4.25706476e-01 -1.76773146e-01 -4.09097075e-01 9.91847157e-01
-8.00726265e-02 6.15880042e-02 -1.07898092e+00 -2.14988440e-01
9.48844478e-02 -2.52189517e-01 6.81296527e-01 2.48737916e-01
1.59398854e+00 -2.74356663e-01 -5.85128546e-01 5.73143840e-01
1.46678436e+00 7.05630362e-01 7.50656009e-01 4.36312735e-01
5.18186688e-01 6.94217861e-01 5.58668196e-01 4.59770888e-01
-2.07027882e-01 2.34393895e-01 7.01760590e-01 7.23137185e-02
1.67747512e-01 -3.71739954e-01 2.24288911e-01 7.21070766e-02
2.23689318e-01 -6.69984818e-01 -1.04168856e+00 7.68620849e-01
-1.63847411e+00 -5.94119251e-01 1.52019083e-01 2.14771295e+00
5.72982550e-01 2.75649607e-01 5.06495386e-02 3.25784534e-01
1.24421203e+00 -4.50347573e-01 -1.17342210e+00 -1.49863064e-01
-2.75851488e-01 5.08484878e-02 2.39097282e-01 -1.24018930e-01
-8.00916553e-01 5.82619667e-01 5.98654366e+00 9.28868473e-01
-9.73125815e-01 -2.84859389e-01 6.57603979e-01 -1.22516295e-02
-4.05742556e-01 -1.80266559e-01 -1.05662620e+00 6.57707632e-01
5.76740861e-01 -1.04604363e-01 6.62403703e-01 1.51414502e+00
-3.96279097e-02 5.25093451e-02 -1.60641325e+00 8.06125224e-01
-2.28688553e-01 -1.15844214e+00 2.40191504e-01 3.17071199e-01
8.81130993e-01 -2.83401102e-01 1.53440714e-01 6.37727082e-01
1.67528197e-01 -1.03537273e+00 7.71031022e-01 2.79809594e-01
7.92742491e-01 -8.88345480e-01 8.84028137e-01 7.08330154e-01
-6.80295050e-01 -4.40400153e-01 -7.02857316e-01 -7.43228495e-02
-5.39011419e-01 7.54180670e-01 -9.72766995e-01 2.73583949e-01
7.18695402e-01 6.84627831e-01 -6.09882593e-01 9.32070374e-01
-3.02055359e-01 5.78824639e-01 -1.48141339e-01 -2.63100445e-01
-1.71192542e-01 -7.30684102e-02 3.63796264e-01 9.28048611e-01
6.57527506e-01 -1.41470626e-01 -1.58272251e-01 1.52567375e+00
-4.30212498e-01 -2.91020602e-01 -9.47192371e-01 -1.51333556e-01
8.94941211e-01 7.09860563e-01 -5.20338714e-01 -3.54859859e-01
-4.04386491e-01 6.08960748e-01 5.19838274e-01 6.98527992e-01
-7.80394852e-01 -6.22676015e-01 7.85475910e-01 1.95303671e-02
3.93109858e-01 2.08786711e-01 -7.57539630e-01 -8.83976817e-01
3.63826662e-01 -8.94499838e-01 2.03680471e-01 -5.85759580e-01
-1.37723684e+00 5.82387686e-01 9.18947682e-02 -1.54398477e+00
1.80733819e-02 -1.00424016e+00 -5.78162730e-01 1.10044169e+00
-1.17054439e+00 -4.78814393e-01 -3.27609062e-01 5.63995361e-01
2.17201814e-01 -6.00652158e-01 5.87860405e-01 4.46478575e-01
-8.76306057e-01 9.84611988e-01 1.88809142e-01 2.87814617e-01
4.10289228e-01 -9.43950057e-01 2.02068210e-01 1.07084119e+00
-4.69385326e-01 9.56998706e-01 3.64660889e-01 -8.33673060e-01
-8.21568549e-01 -1.37134063e+00 3.01714063e-01 -8.86766240e-02
4.93643761e-01 -6.00984573e-01 -1.18297827e+00 6.11707449e-01
-2.67146915e-01 -8.36096033e-02 6.91922426e-01 1.04515783e-01
-2.69237161e-01 -2.58426845e-01 -1.45528674e+00 6.38146698e-01
1.25916207e+00 -3.59193623e-01 -4.46845502e-01 2.19579018e-03
7.78897583e-01 -2.04222932e-01 -6.59818053e-01 4.14413214e-01
2.74858683e-01 -1.11318970e+00 6.38475955e-01 -8.44150245e-01
8.63907993e-01 -2.12340832e-01 -2.66818911e-01 -1.44057119e+00
2.23535043e-03 7.50342160e-02 -2.65625894e-01 1.36542666e+00
6.20144427e-01 -8.10005486e-01 9.76532340e-01 7.33146369e-01
-3.41873676e-01 -7.46624172e-01 -7.60804892e-01 -1.25609052e+00
1.40149251e-01 -5.60603261e-01 9.18500006e-01 9.46119308e-01
-9.13735479e-02 1.93671938e-02 -3.46773490e-02 2.81855196e-01
3.95260513e-01 -2.49371231e-01 5.39598286e-01 -1.58842552e+00
-2.49611244e-01 -5.77186644e-01 -4.67744619e-01 -1.40856731e+00
3.85719150e-01 -8.60710561e-01 -5.55406325e-02 -1.07509780e+00
1.90293625e-01 -4.02661145e-01 -2.22286448e-01 5.76952338e-01
1.60586521e-01 -2.23595172e-01 -1.87550321e-01 6.61863983e-02
-2.40788415e-01 3.32983792e-01 1.40941727e+00 -2.45114207e-01
-2.00117916e-01 1.69236034e-01 -1.10482788e+00 6.36556447e-01
7.93184161e-01 -6.04182899e-01 -7.65865862e-01 -4.77667242e-01
1.28834262e-01 -1.28053010e-01 5.01785457e-01 -1.17875826e+00
1.10134967e-01 -2.38569472e-02 4.71348494e-01 -5.06473601e-01
-1.13623135e-01 -1.06701505e+00 -1.43287778e-01 5.14540315e-01
-3.33041370e-01 -2.00983688e-01 4.89082992e-01 7.27554440e-01
-2.73495734e-01 -8.52875650e-01 7.53264189e-01 2.15225853e-02
-8.48134458e-01 1.18445136e-01 -2.12740839e-01 1.27258629e-01
1.04069245e+00 -7.25493014e-01 -7.87732422e-01 2.80526206e-02
-4.46676016e-01 2.86087334e-01 6.28678381e-01 2.73612440e-01
6.50351167e-01 -1.30424190e+00 -3.39785457e-01 4.67665732e-01
2.54228085e-01 3.70319515e-01 3.90638500e-01 3.82797629e-01
-3.36733341e-01 2.80311495e-01 -3.63006353e-01 -7.14489996e-01
-9.35971320e-01 7.21277893e-01 4.76595849e-01 -1.26630634e-01
-3.31274271e-01 8.34681094e-01 6.67147577e-01 -7.04346597e-01
6.40383899e-01 -5.65130711e-01 -1.43804893e-01 -2.22043380e-01
4.00308579e-01 6.96552813e-01 1.20377876e-01 3.16565901e-01
-3.44996065e-01 -6.46236585e-03 -1.42170832e-01 2.52863407e-01
1.25107431e+00 2.36180469e-01 2.00733561e-02 3.62104565e-01
1.16146755e+00 -4.44873422e-01 -1.41913271e+00 -5.77211604e-02
-7.53381550e-02 -2.19415024e-01 -2.66656816e-01 -9.81950521e-01
-1.04009604e+00 1.20479453e+00 6.90409422e-01 3.55481535e-01
1.26061285e+00 -2.52870709e-01 4.17794853e-01 6.23360097e-01
1.20107964e-01 -1.19442272e+00 1.61574662e-01 4.47382838e-01
6.16447568e-01 -1.14671731e+00 -3.61714929e-01 -2.41601899e-01
-4.83159751e-01 1.37552965e+00 1.22153795e+00 -1.35896191e-01
4.67011511e-01 3.26823071e-02 -4.10019755e-01 1.26166523e-01
-5.37507057e-01 4.87037480e-01 -1.42341806e-02 1.00582731e+00
1.04653567e-01 -2.72890907e-02 2.06836104e-01 1.22774029e+00
-4.98596840e-02 9.46073607e-02 5.70151508e-01 7.34679282e-01
-3.59378725e-01 -8.94324660e-01 -2.26305589e-01 6.34563506e-01
-5.92996590e-02 -1.30546063e-01 -3.02051634e-01 1.10936606e+00
4.05102193e-01 7.18841553e-01 1.10454433e-01 -6.20965719e-01
6.46577060e-01 2.31592596e-01 2.77461618e-01 -7.51896739e-01
-2.77882785e-01 -3.61366034e-01 -7.42474794e-02 -1.45237088e-01
-4.81119864e-02 -4.77274299e-01 -1.09850121e+00 -3.30003470e-01
-3.76754165e-01 -1.85146406e-01 4.71027911e-01 8.97171080e-01
3.97878379e-01 8.34225297e-01 4.71987635e-01 -3.64178181e-01
-8.99480522e-01 -9.67390180e-01 -9.62168157e-01 4.47058022e-01
3.20002317e-01 -8.92081797e-01 -5.19125998e-01 -1.63314849e-01] | [6.5683207511901855, 7.630397796630859] |
17b78cbc-0806-4723-9efa-88b491947a8f | self-constructing-graph-convolutional | 2003.06932 | null | https://arxiv.org/abs/2003.06932v2 | https://arxiv.org/pdf/2003.06932v2.pdf | Self-Constructing Graph Convolutional Networks for Semantic Labeling | Graph Neural Networks (GNNs) have received increasing attention in many fields. However, due to the lack of prior graphs, their use for semantic labeling has been limited. Here, we propose a novel architecture called the Self-Constructing Graph (SCG), which makes use of learnable latent variables to generate embeddings and to self-construct the underlying graphs directly from the input features without relying on manually built prior knowledge graphs. SCG can automatically obtain optimized non-local context graphs from complex-shaped objects in aerial imagery. We optimize SCG via an adaptive diagonal enhancement method and a variational lower bound that consists of a customized graph reconstruction term and a Kullback-Leibler divergence regularization term. We demonstrate the effectiveness and flexibility of the proposed SCG on the publicly available ISPRS Vaihingen dataset and our model SCG-Net achieves competitive results in terms of F1-score with much fewer parameters and at a lower computational cost compared to related pure-CNN based work. Our code will be made public soon. | ['Arnt-Børre Salberg', 'Michael Kampffmeyer', 'Qinghui Liu', 'Robert Jenssen'] | 2020-03-15 | null | null | null | null | ['graph-reconstruction'] | ['graphs'] | [ 4.84275818e-02 3.98421079e-01 6.02085982e-03 -3.35368603e-01
-2.52010852e-01 -6.65935755e-01 3.74306977e-01 1.62537545e-01
-2.03135386e-01 4.30731446e-01 -1.03044743e-03 -3.45474035e-01
-3.09190929e-01 -1.06661296e+00 -8.08013797e-01 -6.31142795e-01
-1.67513028e-01 4.01576132e-01 1.31572232e-01 3.36823091e-02
-1.89127132e-01 5.38313508e-01 -1.17911649e+00 -2.60230303e-01
9.56152439e-01 9.65189934e-01 4.36216414e-01 4.53826636e-01
1.05208278e-01 4.74911958e-01 -1.02528431e-01 -4.82138574e-01
4.32152152e-01 -3.38282764e-01 -7.09459186e-01 4.60022688e-01
5.71428418e-01 -1.43578559e-01 -4.53128636e-01 1.27675760e+00
2.87530243e-01 2.86881238e-01 5.05618870e-01 -1.18714964e+00
-1.19539785e+00 5.97163975e-01 -3.20006818e-01 -2.36358538e-01
-2.43083060e-01 4.85872142e-02 1.43617857e+00 -6.84299290e-01
8.18090439e-01 1.11096537e+00 8.77791762e-01 2.94510186e-01
-1.41616881e+00 -3.74581784e-01 3.11658382e-01 1.00103974e-01
-1.51750374e+00 1.38721034e-01 1.00717580e+00 -5.26485980e-01
8.53560686e-01 -1.30358979e-01 8.19285572e-01 7.82203138e-01
7.34628588e-02 4.12086338e-01 6.06207073e-01 -2.66632140e-01
2.97327548e-01 -1.75478369e-01 2.74245739e-02 1.35528278e+00
3.12410146e-01 -3.40676397e-01 -1.07057422e-01 -7.90265426e-02
9.13431823e-01 1.61927789e-01 -4.14591730e-01 -7.78476357e-01
-8.80002499e-01 1.03426778e+00 1.05323899e+00 1.80077106e-02
-4.07172948e-01 4.50562596e-01 -1.82599965e-02 -4.74397503e-02
7.54421175e-01 3.87276649e-01 -1.46816269e-01 5.15746355e-01
-6.93938196e-01 5.63474000e-03 6.57165468e-01 1.22660398e+00
9.98187304e-01 2.18940407e-01 -1.58004779e-02 8.23638618e-01
3.95662785e-01 3.14678460e-01 4.99892607e-02 -7.95200527e-01
3.40776831e-01 9.84413207e-01 -1.58300415e-01 -1.38296986e+00
-3.81150723e-01 -6.72156990e-01 -9.07264769e-01 -9.60225761e-02
1.06090009e-01 -9.70336422e-02 -1.00891793e+00 1.78296340e+00
3.54632288e-01 3.30862314e-01 -7.79425427e-02 9.11380529e-01
8.96027863e-01 5.72273493e-01 -1.20968763e-02 2.70443588e-01
8.73645544e-01 -1.06598818e+00 -5.35839796e-01 -1.66421711e-01
5.74792325e-01 -2.28340939e-01 1.21076930e+00 1.32830054e-01
-7.13685274e-01 -2.49210060e-01 -1.20995522e+00 -1.55447468e-01
-4.39988226e-01 4.65935796e-01 8.25554729e-01 3.24917048e-01
-1.40119016e+00 8.24023962e-01 -1.08125925e+00 -3.49063903e-01
7.55846024e-01 3.30254048e-01 -6.37452960e-01 -5.41050285e-02
-7.40967989e-01 2.27835611e-01 5.51959038e-01 3.54266763e-01
-1.02210939e+00 -3.59390557e-01 -1.18854129e+00 2.39024863e-01
6.94142818e-01 -8.06853652e-01 4.70866233e-01 -9.21061218e-01
-1.56050527e+00 8.09304416e-01 3.21519613e-01 -5.25043547e-01
3.56888771e-01 -1.39662147e-01 -4.52302918e-02 3.02448124e-01
-2.69655348e-03 6.46608651e-01 9.05521333e-01 -1.05390584e+00
-1.68236271e-02 -4.74785805e-01 2.53371000e-01 9.46289673e-02
-5.47630847e-01 -5.50891161e-01 -6.89096987e-01 -7.67843962e-01
2.69205272e-01 -1.23447943e+00 -3.36923689e-01 2.50489354e-01
-4.70907956e-01 -1.42013937e-01 8.24537277e-01 -5.99790394e-01
9.95868266e-01 -2.12470460e+00 4.94323134e-01 1.26556158e-01
6.35782361e-01 3.55526179e-01 -4.18976277e-01 3.40152383e-01
6.15331717e-03 2.71104336e-01 -6.22355998e-01 -3.65518898e-01
1.85241122e-02 2.98173696e-01 -1.03460632e-01 5.17492115e-01
4.45278734e-01 1.16279137e+00 -1.14081347e+00 -2.53280967e-01
3.45648617e-01 6.21598840e-01 -6.71305299e-01 1.87677145e-01
-4.81938928e-01 2.93674972e-02 -4.20598954e-01 3.93644005e-01
6.85240388e-01 -8.53565991e-01 4.43277717e-01 -2.08960518e-01
1.24734841e-01 -6.10726885e-02 -9.56191123e-01 1.83765483e+00
-3.10871601e-01 5.30435205e-01 1.14694260e-01 -1.09776926e+00
1.01535726e+00 -1.12957940e-01 1.81512475e-01 -3.45562175e-02
2.13593215e-01 -2.35207640e-02 -3.38389426e-01 -2.35132426e-01
3.50500911e-01 1.24513611e-01 2.58295268e-01 1.07376173e-01
4.04315680e-01 -1.37146086e-01 1.76231459e-01 4.10692096e-01
1.18076611e+00 3.50220174e-01 1.54860869e-01 -4.37092274e-01
2.55517751e-01 -7.19551817e-02 4.53595489e-01 3.83732736e-01
5.38660549e-02 7.51175046e-01 5.77776074e-01 -5.50707936e-01
-8.26443255e-01 -9.39146459e-01 1.85934305e-01 7.90225267e-01
1.75162598e-01 -6.42321050e-01 -7.84800351e-01 -9.60467875e-01
1.07388191e-01 5.89266837e-01 -6.28930390e-01 -8.19109231e-02
-3.70098382e-01 -5.80532849e-01 3.42537194e-01 5.17384112e-01
6.65242732e-01 -8.95061851e-01 -2.16878876e-01 1.62424773e-01
-1.30938282e-02 -1.36516643e+00 -4.15564328e-01 4.49145697e-02
-1.04202843e+00 -1.05581129e+00 -6.21137798e-01 -8.76512945e-01
9.47459757e-01 3.74048859e-01 1.00259972e+00 2.30320677e-01
-3.15857291e-01 2.93206513e-01 -3.72429341e-01 -9.23592672e-02
-6.87224418e-03 2.67745852e-01 -4.59998339e-01 2.37431362e-01
3.77149619e-02 -7.62912214e-01 -5.68908989e-01 1.07722946e-01
-1.05572951e+00 2.13021651e-01 4.77111697e-01 8.42068851e-01
7.95103371e-01 5.69933578e-02 2.31776103e-01 -1.07645905e+00
4.33332622e-01 -3.40159744e-01 -1.11133218e+00 1.97439194e-01
-7.68554389e-01 3.01835328e-01 8.20142567e-01 -2.42759481e-01
-5.81619203e-01 2.67121553e-01 6.32001534e-02 -7.95044243e-01
8.81014913e-02 7.34327614e-01 -1.71231717e-01 -3.44169110e-01
4.33013171e-01 9.73936636e-03 3.47330235e-02 -3.35324287e-01
5.98974109e-01 1.77303344e-01 3.31785023e-01 -2.67522097e-01
8.52350652e-01 5.41296661e-01 3.14697534e-01 -9.59849060e-01
-9.86191511e-01 -3.22604209e-01 -6.26183510e-01 -1.89450696e-01
1.08137465e+00 -8.96157742e-01 -3.45737934e-01 3.33414316e-01
-9.51990724e-01 -6.37936354e-01 -2.04754129e-01 4.77017343e-01
-3.86783123e-01 5.81207991e-01 -5.28417587e-01 -5.08727491e-01
-3.35527956e-01 -1.10346174e+00 1.11954391e+00 5.17144985e-02
3.50787342e-01 -1.27760577e+00 -2.62938946e-01 1.49494737e-01
2.71729946e-01 6.39600873e-01 9.98494148e-01 -3.21496993e-01
-1.02071345e+00 -9.35977995e-02 -5.03702044e-01 6.29265189e-01
1.69055924e-01 -9.93013307e-02 -6.92606747e-01 -4.20460105e-01
-4.31770265e-01 -2.40701124e-01 1.05165958e+00 5.70205808e-01
1.33466113e+00 -4.50379461e-01 -2.22574845e-01 1.05644822e+00
1.78320789e+00 -3.07668209e-01 4.74056721e-01 8.62719268e-02
1.23816693e+00 4.58277285e-01 6.11095168e-02 3.69674027e-01
3.92865390e-01 3.84547383e-01 8.70316267e-01 -6.00715168e-03
-3.10080767e-01 -4.09142941e-01 1.18369319e-01 9.09610689e-01
-2.98042387e-01 -6.10950351e-01 -9.40582871e-01 7.15781927e-01
-2.03372526e+00 -5.55941999e-01 -1.37065098e-01 2.02138138e+00
3.41087937e-01 -5.39409518e-02 -3.53474021e-01 -2.84735203e-01
8.80808711e-01 3.92964929e-01 -5.62018812e-01 1.67987227e-01
-9.76651907e-02 2.20100179e-01 8.05078626e-01 5.65227866e-01
-1.22705758e+00 1.29649866e+00 5.47434759e+00 5.86681724e-01
-1.04600751e+00 -9.75363702e-03 3.27134043e-01 2.71506965e-01
-3.98851186e-01 1.92845911e-01 -5.23263216e-01 8.44793022e-02
5.12059212e-01 5.45249209e-02 6.44217789e-01 1.11100376e+00
-1.35837331e-01 3.62151682e-01 -7.80027986e-01 1.16685629e+00
1.58793271e-01 -1.48055553e+00 3.22017968e-01 1.46216303e-01
7.14209259e-01 1.94181412e-01 6.88551040e-03 1.52635248e-02
6.05206370e-01 -9.26211298e-01 6.33415163e-01 4.13191855e-01
8.32319081e-01 -6.19720340e-01 7.00017393e-01 9.99553576e-02
-1.48441458e+00 2.93780677e-02 -5.56475163e-01 4.50959615e-02
9.38384384e-02 7.25748181e-01 -7.84677327e-01 8.58059347e-01
4.56313223e-01 1.01037621e+00 -6.57738864e-01 8.56807828e-01
-6.42678261e-01 8.35345089e-01 -2.50910670e-01 -1.82597637e-02
5.30865014e-01 -6.80712044e-01 5.91976345e-01 8.66179109e-01
4.13024276e-01 -1.78030171e-02 3.98175538e-01 1.18745399e+00
-4.16418135e-01 1.30954832e-01 -7.50441611e-01 -6.21870339e-01
2.00681925e-01 1.46119285e+00 -1.03290761e+00 -1.32372826e-01
-1.68638095e-01 1.08572888e+00 7.46482313e-01 4.88760263e-01
-8.20662439e-01 -4.90171999e-01 2.64063925e-01 1.58669487e-01
5.26946127e-01 -5.54992616e-01 3.80193489e-03 -1.29925704e+00
1.06321461e-01 -3.03715259e-01 3.69696200e-01 -8.90037835e-01
-1.27079785e+00 8.36452603e-01 -1.16058299e-02 -8.79241824e-01
1.04270920e-01 -7.77773082e-01 -2.14864567e-01 7.05082178e-01
-1.56730652e+00 -1.52684402e+00 -8.12403798e-01 3.75265688e-01
1.80219501e-01 -1.66128069e-01 8.01857948e-01 8.55818689e-02
-5.35484076e-01 3.43464315e-01 -7.28011355e-02 3.09511125e-01
3.02095085e-01 -1.40221846e+00 6.15461528e-01 8.86331975e-01
2.80980706e-01 5.06332099e-01 4.38373536e-01 -7.86630511e-01
-1.37465310e+00 -1.75509286e+00 6.75783813e-01 -1.51309088e-01
7.24304140e-01 -5.89560747e-01 -9.44041252e-01 9.98377979e-01
1.63440779e-01 4.23073411e-01 4.28234577e-01 1.06604360e-01
-3.99310887e-01 -7.71870092e-02 -9.05529976e-01 5.43115616e-01
1.40911949e+00 -4.96652842e-01 -2.53873020e-01 5.36743760e-01
1.09370267e+00 -2.33075649e-01 -7.09211230e-01 3.69805872e-01
2.63648421e-01 -6.62180603e-01 8.35285246e-01 -4.68474388e-01
2.91557342e-01 -3.98948520e-01 -1.70983136e-01 -1.42583883e+00
-7.07911491e-01 -6.11604571e-01 1.32830098e-01 9.21018958e-01
3.17184240e-01 -8.85431945e-01 7.36824095e-01 3.24182123e-01
-4.75820720e-01 -7.52121389e-01 -4.91145223e-01 -9.57361460e-01
-3.04423690e-01 -3.23566318e-01 6.16758227e-01 9.82711911e-01
-4.66585875e-01 5.06322205e-01 -2.84237355e-01 5.63153028e-01
8.17226350e-01 1.61691070e-01 7.45853543e-01 -1.33921385e+00
-3.48786056e-01 -2.22328663e-01 -7.07976639e-01 -1.00099039e+00
3.95695686e-01 -1.37022412e+00 -4.86585833e-02 -1.90049005e+00
-5.13797700e-02 -3.30543846e-01 -1.34934559e-01 7.64121294e-01
-1.26954436e-01 2.53086209e-01 1.50782421e-01 -2.38835625e-02
-7.50868320e-01 1.05254138e+00 1.21295214e+00 -2.41404966e-01
-1.77499443e-01 -4.74540412e-01 -3.99045289e-01 7.50977099e-01
7.33184695e-01 -5.35317123e-01 -7.41236687e-01 -6.18885696e-01
4.51658517e-01 -2.27218375e-01 5.87865531e-01 -1.02304757e+00
-5.98522276e-03 -8.31297711e-02 1.13292802e-02 -4.31764990e-01
1.24005713e-01 -7.01322854e-01 3.46807957e-01 1.41877308e-01
-1.62717238e-01 1.19683698e-01 9.95754302e-02 1.03111410e+00
-1.30159780e-01 -2.24332377e-01 8.14626575e-01 -1.52000591e-01
-6.45853519e-01 7.61678874e-01 9.14651155e-02 -1.14612523e-02
1.04433143e+00 7.06855133e-02 -2.10276291e-01 -3.26412380e-01
-7.50736594e-01 1.03337817e-01 6.66490555e-01 2.36590981e-01
7.81611145e-01 -1.42508781e+00 -5.26787817e-01 2.26786017e-01
2.85616547e-01 2.44588792e-01 1.77760065e-01 5.13633966e-01
-8.75915587e-01 2.64231056e-01 -1.06804274e-01 -6.49856627e-01
-1.01000059e+00 5.29647529e-01 1.40884995e-01 -3.78629625e-01
-9.92958784e-01 1.01263046e+00 3.49410415e-01 -6.21795058e-01
-6.63138134e-03 -4.25042748e-01 -9.66563150e-02 -1.43277586e-01
1.05531894e-01 1.69891685e-01 7.17365965e-02 -5.76315999e-01
-2.00303420e-01 4.85996664e-01 2.40172938e-01 1.66558981e-01
1.61903691e+00 1.15075387e-01 -1.73941344e-01 2.26765573e-01
1.24086607e+00 -3.00817966e-01 -1.45864487e+00 -1.86626956e-01
-1.72413275e-01 -4.48371440e-01 3.62738967e-01 -2.25334316e-01
-1.41177738e+00 8.24071348e-01 4.40975815e-01 3.18391472e-01
9.96932924e-01 -9.67680663e-03 7.12497592e-01 4.47097749e-01
4.62850690e-01 -8.86758745e-01 2.46134713e-01 4.95786488e-01
9.43572283e-01 -1.10312772e+00 5.02109788e-02 -6.15252793e-01
-4.97020245e-01 1.14505327e+00 2.54471570e-01 -6.71182752e-01
7.56443381e-01 -1.81393862e-01 -2.74519056e-01 -4.99846160e-01
-4.72797453e-01 -4.65056568e-01 5.16225576e-01 5.65962732e-01
1.61243334e-01 3.00677478e-01 -7.90969208e-02 3.84367555e-01
-1.16270266e-01 -3.75924516e-03 3.97166848e-01 7.01078594e-01
-1.26684487e-01 -9.16447759e-01 2.93214113e-01 5.12847841e-01
-2.99248636e-01 -2.61836410e-01 -3.43903154e-01 7.12951303e-01
-1.08311906e-01 6.16261363e-01 -1.17270708e-01 -4.03380692e-01
1.50223181e-01 -1.02765843e-01 4.00379956e-01 -8.77871215e-01
-9.06219780e-02 -6.69467971e-02 -2.89805187e-03 -7.35507131e-01
-5.23891270e-01 -2.72401929e-01 -1.15223897e+00 1.32854404e-02
-5.51997721e-01 3.22604063e-03 6.32654309e-01 7.17221558e-01
7.11159706e-01 5.56983471e-01 2.13292554e-01 -8.39824319e-01
-3.71510446e-01 -8.32849741e-01 -7.72463143e-01 4.66399640e-01
1.62227780e-01 -8.85153294e-01 -4.25499856e-01 1.78421542e-01] | [9.627215385437012, 0.7686147689819336] |
a1cf99b2-0675-429b-a6db-c191bbe4cd4b | learning-from-bootstrapping-and-stepwise | 2205.09324 | null | https://arxiv.org/abs/2205.09324v1 | https://arxiv.org/pdf/2205.09324v1.pdf | Learning from Bootstrapping and Stepwise Reinforcement Reward: A Semi-Supervised Framework for Text Style Transfer | Text style transfer is an important task in controllable language generation. Supervised approaches have pushed performance improvement on style-oriented rewriting such as formality conversion. However, challenges remain due to the scarcity of large-scale parallel data in many domains. While unsupervised approaches do not rely on annotated sentence pairs for each style, they are often plagued with instability issues such as mode collapse or quality degradation. To take advantage of both supervised and unsupervised paradigms and tackle the challenges, in this work, we propose a semi-supervised framework for text style transfer. First, the learning process is bootstrapped with supervision guided by automatically constructed pseudo-parallel pairs using lexical and semantic-based methods. Then the model learns from unlabeled data via reinforcement rewards. Specifically, we propose to improve the sequence-to-sequence policy gradient via stepwise reward optimization, providing fine-grained learning signals and stabilizing the reinforced learning process. Experimental results show that the proposed approach achieves state-of-the-art performance on multiple datasets, and produces effective generation with as minimal as 10\% of training data. | ['Nancy F. Chen', 'Zhengyuan Liu'] | 2022-05-19 | null | https://aclanthology.org/2022.findings-naacl.201 | https://aclanthology.org/2022.findings-naacl.201.pdf | findings-naacl-2022-7 | ['text-style-transfoer'] | ['natural-language-processing'] | [ 8.04824352e-01 2.04672664e-01 -2.91659921e-01 -4.68801707e-01
-1.10045099e+00 -6.36578381e-01 8.19315314e-01 2.33115461e-02
-3.09573501e-01 1.15459919e+00 2.84902692e-01 -1.48309261e-01
2.37457618e-01 -6.89602435e-01 -7.90151834e-01 -4.70592469e-01
4.07542974e-01 6.62280858e-01 -4.30982700e-03 -5.66487610e-01
3.13740253e-01 7.59548843e-02 -1.25484908e+00 1.81456178e-01
1.48271859e+00 5.97834885e-01 2.94185132e-01 5.36888540e-01
-4.01574135e-01 7.39381433e-01 -5.51039994e-01 -5.03019512e-01
9.45637822e-02 -9.52341080e-01 -8.30823481e-01 1.16144292e-01
5.44109158e-02 -1.50328413e-01 2.65501682e-02 1.09629393e+00
5.45227647e-01 5.86599670e-02 5.67804515e-01 -1.04717422e+00
-7.23527908e-01 8.47885430e-01 -3.76094162e-01 -9.73502770e-02
3.58813018e-01 2.69934148e-01 1.19958472e+00 -8.94349933e-01
8.98888052e-01 1.26570463e+00 6.04749843e-02 1.04222870e+00
-1.50465202e+00 -4.69280690e-01 2.26234525e-01 -2.35933974e-01
-8.75807106e-01 -5.38752198e-01 1.00046790e+00 -2.48345703e-01
7.54717827e-01 3.93841416e-02 4.86384064e-01 1.42870045e+00
-4.69469232e-03 9.33100700e-01 1.29240167e+00 -7.08447814e-01
3.00558656e-01 2.33540341e-01 -2.43536666e-01 6.53070867e-01
-6.96979314e-02 2.34352842e-01 -6.96520269e-01 -5.71702719e-02
5.48019350e-01 -3.67558300e-01 -6.88516051e-02 -4.39338088e-01
-1.24981618e+00 8.10735106e-01 1.78394333e-01 2.53425837e-01
-1.46459520e-01 -2.68298737e-03 6.14147544e-01 4.82125074e-01
4.63553488e-01 8.07974875e-01 -2.98823684e-01 -4.09837663e-01
-8.96156907e-01 4.14387912e-01 7.38200486e-01 1.22860217e+00
7.83425391e-01 2.38963053e-01 -3.37067008e-01 1.03124940e+00
-3.35768098e-03 6.18482888e-01 6.30248666e-01 -7.22439766e-01
9.06766236e-01 5.86647272e-01 8.10638443e-02 -5.49706101e-01
-1.19383149e-02 -2.60077268e-01 -8.18987072e-01 8.53895247e-02
3.63498271e-01 -3.92409950e-01 -7.29272604e-01 2.01484966e+00
1.96136162e-01 -2.23449707e-01 1.77240521e-01 7.54996598e-01
3.34764630e-01 5.69698751e-01 8.00760463e-02 -4.63299006e-01
1.02348912e+00 -1.13944960e+00 -6.79052591e-01 -2.74274170e-01
5.97058594e-01 -8.33190501e-01 1.46630549e+00 2.37318277e-01
-1.27619171e+00 -5.43787241e-01 -9.90660369e-01 1.11789063e-01
-7.48556778e-02 4.40909453e-02 2.36740768e-01 3.95684659e-01
-8.11303377e-01 7.81737685e-01 -6.53936505e-01 -4.22523558e-01
2.80974835e-01 2.49573424e-01 -7.26098777e-04 1.62328556e-01
-1.18594193e+00 7.58060217e-01 4.78316933e-01 -3.91701274e-02
-8.72399449e-01 -4.94132608e-01 -8.07098448e-01 -1.12090886e-01
5.42735398e-01 -8.60927284e-01 1.45697331e+00 -1.32299328e+00
-2.21559906e+00 7.27474213e-01 -1.11555532e-01 -4.37932700e-01
6.84204936e-01 -4.95576560e-01 -3.42615813e-01 -1.20807946e-01
1.12705685e-01 7.78461814e-01 9.97203112e-01 -1.33431089e+00
-6.74984038e-01 -4.63766940e-02 6.73945546e-02 4.34614480e-01
-5.52376270e-01 1.06896102e-01 -8.34504738e-02 -7.75508463e-01
-3.14446181e-01 -1.00476050e+00 -3.47030252e-01 -3.75444442e-01
-4.11859810e-01 -2.68441796e-01 5.70809960e-01 -4.42076921e-01
1.13947344e+00 -1.72314990e+00 5.82648575e-01 -7.84094185e-02
-1.03865743e-01 3.52659285e-01 -2.92272896e-01 5.51828265e-01
3.01581442e-01 7.83792213e-02 -4.27164555e-01 -4.46479619e-01
1.01048453e-03 8.43421593e-02 -5.87707520e-01 -1.27490595e-01
6.08138740e-01 8.61382186e-01 -1.26500869e+00 -4.87834424e-01
-1.55801177e-01 4.68603196e-03 -6.85865700e-01 6.17497921e-01
-7.21722066e-01 8.45372260e-01 -6.23124361e-01 3.95388663e-01
1.63899332e-01 -1.42808616e-01 4.87791061e-01 2.49484822e-01
-3.33565250e-02 6.41784668e-01 -8.97050738e-01 2.18758845e+00
-6.19838059e-01 1.42401800e-01 -2.94809967e-01 -1.01760113e+00
1.23809218e+00 3.15182924e-01 6.71713650e-02 -6.02049947e-01
6.82301745e-02 4.04387146e-01 -4.25080806e-02 -2.77014136e-01
6.12602472e-01 -2.77286053e-01 -2.46901318e-01 6.31929040e-01
2.41971463e-01 -5.49511969e-01 4.00017589e-01 1.35914087e-01
8.06273997e-01 6.36348128e-01 2.23409638e-01 -2.43095413e-01
6.21933579e-01 -6.74464479e-02 7.13384509e-01 5.69693089e-01
3.12005412e-02 4.78031963e-01 5.91653526e-01 -1.64013460e-01
-1.27594471e+00 -1.01742661e+00 3.24742794e-01 1.21114302e+00
-1.51303271e-02 -4.60293353e-01 -1.01057863e+00 -9.75506425e-01
-3.59600842e-01 8.33005488e-01 -3.97195429e-01 -3.26262712e-01
-8.50714564e-01 -5.31402647e-01 5.52842736e-01 4.78845567e-01
2.77008832e-01 -1.55582535e+00 -3.45261604e-01 5.26137352e-01
-2.69783318e-01 -1.05195832e+00 -7.37081349e-01 1.75216496e-02
-1.09296894e+00 -6.35850191e-01 -6.19517207e-01 -8.82708609e-01
7.38405943e-01 -1.84521779e-01 1.26613772e+00 -1.91672102e-01
-2.72881053e-03 2.19054949e-02 -3.34123254e-01 -1.97040915e-01
-9.31754887e-01 6.23412192e-01 1.57153443e-01 8.74495134e-02
-2.53513246e-03 -7.33165920e-01 -4.13102746e-01 1.02024078e-01
-7.78938413e-01 3.14328820e-01 6.70963049e-01 1.35934210e+00
5.47134280e-01 -6.12841249e-01 1.06623399e+00 -1.21605575e+00
1.00087178e+00 -1.27239108e-01 -6.61764681e-01 3.44252616e-01
-9.46187437e-01 5.15048325e-01 1.09424579e+00 -4.02947366e-01
-1.39309442e+00 6.28353059e-02 1.42611951e-01 -2.81379640e-01
-2.91822851e-01 2.93340027e-01 -2.89948553e-01 4.14148420e-01
8.32525969e-01 4.61202145e-01 -2.60212850e-02 -3.16555858e-01
6.45451665e-01 6.50483727e-01 3.32898587e-01 -1.16991639e+00
1.00693858e+00 9.59152877e-02 -2.65679002e-01 -5.69317043e-01
-9.35426235e-01 3.60981710e-02 -5.60600221e-01 -5.39810210e-02
7.21113622e-01 -7.37417102e-01 -2.38971397e-01 2.45702758e-01
-1.11425042e+00 -5.52311718e-01 -4.74372745e-01 2.05522373e-01
-9.25435603e-01 3.27181071e-01 -6.97330117e-01 -7.59628177e-01
-6.69234335e-01 -1.20018995e+00 9.96133924e-01 3.13419640e-01
-4.09220934e-01 -9.32911694e-01 2.80446887e-01 3.20846498e-01
3.28455418e-01 1.51595101e-01 9.72508192e-01 -5.66961586e-01
-5.16981661e-01 1.42797559e-01 -6.19970867e-03 3.71537179e-01
3.82521421e-01 -1.37561515e-01 -7.75598884e-01 -3.16148341e-01
-2.69338280e-01 -9.32809234e-01 4.01583254e-01 -2.07648948e-01
9.74306762e-01 -3.14609319e-01 -1.57684572e-02 4.89904284e-01
1.12767529e+00 1.28542095e-01 2.82990277e-01 2.01470718e-01
7.24728048e-01 7.25822389e-01 7.34267771e-01 3.55188757e-01
1.78382590e-01 6.39671445e-01 3.12151364e-03 1.22865967e-01
-7.49652684e-02 -8.44218433e-01 4.81262773e-01 1.05079782e+00
-3.16153578e-02 -7.64597729e-02 -6.41908169e-01 5.05408227e-01
-1.92434561e+00 -7.33916461e-01 3.06721032e-01 2.18910241e+00
1.40903950e+00 5.73590517e-01 1.71643585e-01 -5.12616187e-02
7.67817378e-01 2.11552873e-01 -6.20180905e-01 -5.35155475e-01
-8.13048184e-02 2.49082804e-01 6.34541586e-02 5.94047129e-01
-7.82424808e-01 1.50892138e+00 5.47904968e+00 7.99954414e-01
-1.21135283e+00 -1.38738528e-01 6.37333930e-01 -4.30595800e-02
-6.00110173e-01 1.28309816e-01 -8.11954319e-01 4.20998663e-01
8.82174015e-01 -3.22207153e-01 8.07646632e-01 7.39249170e-01
3.06328595e-01 3.33429068e-01 -1.28979969e+00 6.27625704e-01
3.51886410e-04 -1.15562272e+00 1.56006172e-01 -1.84614316e-01
1.13664925e+00 -2.84373760e-01 -9.73381400e-02 4.36818182e-01
5.85151434e-01 -8.02465081e-01 7.62359321e-01 3.05175215e-01
1.06426704e+00 -7.96334982e-01 2.60954142e-01 4.30252492e-01
-9.16949630e-01 1.54372990e-01 -1.98562592e-01 -1.16072014e-01
2.01132253e-01 4.75408643e-01 -8.57964933e-01 5.48791587e-01
1.59256294e-01 8.07492971e-01 -3.37565511e-01 3.58239770e-01
-6.97932839e-01 7.06758738e-01 5.91307096e-02 -3.04190069e-01
1.20638072e-01 -3.91384006e-01 5.32660484e-01 1.08679402e+00
9.86464024e-02 -2.19982237e-01 3.37688982e-01 1.04186690e+00
-3.22880000e-01 3.25480551e-01 -8.18752468e-01 -3.57508332e-01
4.01654363e-01 1.24989498e+00 -5.34177840e-01 -4.12684590e-01
-1.90665200e-01 1.24400425e+00 7.98267603e-01 4.27286863e-01
-6.49707437e-01 -4.20870274e-01 3.77596200e-01 -1.11196667e-01
2.12297603e-01 -2.04412594e-01 -5.03799677e-01 -1.33134174e+00
1.73716947e-01 -1.19884157e+00 7.94070959e-02 -3.55131716e-01
-1.25301623e+00 7.12260067e-01 -2.60895908e-01 -1.33347523e+00
-6.61549985e-01 -2.28057429e-01 -7.34428823e-01 9.03927565e-01
-1.55359650e+00 -1.04359972e+00 8.86305049e-02 3.91750455e-01
9.78140950e-01 -4.82845277e-01 9.62864339e-01 1.57154370e-02
-4.81621742e-01 9.33118403e-01 1.93381205e-01 -7.02545568e-02
1.13721681e+00 -1.38716245e+00 6.23574615e-01 7.84712493e-01
-7.49356230e-04 5.35264850e-01 7.03015149e-01 -6.86444402e-01
-1.31191945e+00 -1.05909574e+00 1.05273831e+00 -2.30745748e-01
8.12766612e-01 -6.23220563e-01 -6.76712811e-01 3.75376761e-01
3.79503131e-01 -2.14160353e-01 5.26461124e-01 1.94688722e-01
-3.04335475e-01 -1.69387057e-01 -9.29026186e-01 9.81877506e-01
1.36247420e+00 -4.79769051e-01 -6.07925892e-01 3.03164035e-01
8.20384026e-01 -5.27020276e-01 -7.95648932e-01 2.13918582e-01
1.87456056e-01 -6.25201762e-01 4.46840018e-01 -9.23688173e-01
9.67323601e-01 -1.66859478e-01 7.37911910e-02 -1.63965952e+00
-1.20337769e-01 -1.04927051e+00 1.67571828e-02 1.62804663e+00
6.63233161e-01 -4.43277329e-01 7.16509342e-01 2.70876139e-01
-1.75067529e-01 -7.72753596e-01 -6.59798145e-01 -7.17134297e-01
2.27289677e-01 7.80351162e-02 5.20424306e-01 7.38743186e-01
2.96454400e-01 9.99068201e-01 -6.05306864e-01 -2.58475214e-01
5.65227449e-01 4.09294873e-01 8.74675810e-01 -9.29579198e-01
-5.74474335e-01 -4.06958014e-01 2.46993914e-01 -1.05918920e+00
4.04521435e-01 -9.74575758e-01 2.81774223e-01 -1.14035547e+00
9.21343416e-02 -4.42235053e-01 -2.07199484e-01 2.65974760e-01
-5.75977743e-01 -7.44936913e-02 3.40853423e-01 1.69828475e-01
-7.04944849e-01 1.04010653e+00 1.48537838e+00 -1.91246882e-01
-3.20797622e-01 -4.36857007e-02 -7.26512730e-01 5.48251867e-01
9.87892687e-01 -4.51187491e-01 -6.44572616e-01 -5.62410414e-01
1.21232644e-01 3.67209584e-01 -1.30476817e-01 -8.17784429e-01
-1.73180059e-01 -4.32186455e-01 4.67146225e-02 -7.15335831e-02
2.66073383e-02 -3.81161213e-01 -3.74689192e-01 4.53769058e-01
-8.83336544e-01 3.37551162e-02 -1.57534227e-01 6.17393374e-01
-1.13294490e-01 -1.81748286e-01 7.88438797e-01 -1.16891950e-01
-2.57096022e-01 1.62103266e-01 -1.31531179e-01 5.96635103e-01
6.90856516e-01 2.82467306e-01 -1.06899917e-01 -2.77405828e-01
-4.19896871e-01 2.40219265e-01 5.49324989e-01 7.63292909e-01
3.30518037e-01 -1.45415831e+00 -7.44007528e-01 1.67132840e-01
3.47172439e-01 7.17831999e-02 -1.44419402e-01 2.65149057e-01
-1.00271784e-01 3.03056687e-01 -2.34131068e-01 -5.26804924e-01
-8.29699934e-01 5.18910825e-01 1.16012380e-01 -7.82530248e-01
-4.55590785e-01 5.83993316e-01 1.04992293e-01 -7.51357734e-01
8.57025310e-02 -1.55862272e-01 3.14138271e-02 -1.57218263e-01
3.08003247e-01 1.36251643e-01 -3.66550796e-02 -2.58123934e-01
1.10695116e-01 3.38708073e-01 -3.90839547e-01 -4.98179764e-01
1.07298088e+00 -5.36292382e-02 1.24802396e-01 5.79051793e-01
8.55925381e-01 -7.61227831e-02 -1.50878906e+00 -4.91592348e-01
1.45685092e-01 -2.51924902e-01 -4.56571370e-01 -9.00169194e-01
-5.95114052e-01 9.14211512e-01 4.02574450e-01 2.89711095e-02
9.19049263e-01 -3.07220340e-01 1.04830587e+00 5.00300944e-01
3.69668961e-01 -1.40512323e+00 3.52105230e-01 6.58675253e-01
8.10791850e-01 -1.28028107e+00 -4.54021871e-01 -2.19779149e-01
-8.77707958e-01 9.73578036e-01 8.86871815e-01 -1.77692994e-01
8.66181627e-02 5.38814552e-02 1.21984184e-01 3.56143087e-01
-9.33942795e-01 8.43408983e-03 1.67264372e-01 4.88962293e-01
7.45140553e-01 1.20605938e-01 -6.66665912e-01 5.15014768e-01
-3.76951575e-01 1.30371317e-01 3.46034765e-01 1.08210361e+00
-3.29505384e-01 -1.83144760e+00 -9.79312211e-02 4.20719445e-01
-2.82169104e-01 -1.60737485e-01 -3.68146390e-01 2.76118755e-01
-3.07303131e-01 8.12478006e-01 -2.67288417e-01 -2.38111526e-01
3.98711830e-01 3.34299862e-01 5.65252125e-01 -7.99303353e-01
-6.99612498e-01 1.94473013e-01 2.72202134e-01 -2.36996755e-01
-5.11568412e-03 -7.41847575e-01 -1.18532228e+00 -5.62504306e-02
-2.19161436e-01 3.20027560e-01 4.04235333e-01 1.06123340e+00
3.65024447e-01 6.83417320e-01 1.02427924e+00 -8.07135701e-01
-1.04995418e+00 -1.05416965e+00 -2.29296595e-01 6.91095769e-01
1.85302850e-02 -3.84043276e-01 -3.01967263e-02 4.06247526e-01] | [11.67066764831543, 9.306537628173828] |
c34572a3-5780-4684-834f-96e21e1078e0 | dreamartist-towards-controllable-one-shot | 2211.11337 | null | https://arxiv.org/abs/2211.11337v3 | https://arxiv.org/pdf/2211.11337v3.pdf | DreamArtist: Towards Controllable One-Shot Text-to-Image Generation via Positive-Negative Prompt-Tuning | Large-scale text-to-image generation models have achieved remarkable progress in synthesizing high-quality, feature-rich images with high resolution guided by texts. However, these models often struggle with novel concepts, eg, new styles, object entities, etc. Although recent attempts have employed fine-tuning or prompt-tuning strategies to teach the pre-trained diffusion model novel concepts from a reference image set,they have the drawback of overfitting to the given reference images, particularly in one-shot applications, which is harmful to generate diverse and high-quality images while maintaining generation controllability. To tackle this challenge, we present a simple yet effective method called DreamArtist, which employs a positive-negative prompt-tuning learning strategy. Specifically, DreamArtist incorporates both positive and negative embeddings and jointly trains them. The positive embedding aggressively captures the salient characteristics of the reference image to drive diversified generation and the negative embedding rectifies inadequacies from the positive embedding. It learns not only what is correct, but also what can be avoided or improved. We have conducted extensive experiments and evaluated the proposed method from image similarity and diversity, generation controllability, and style cloning. And our DreamArtist has achieved a superior generation performance over existing methods. Besides, our additional evaluation on extended tasks, including concept compositions and prompt-guided image editing, demonstrates its effectiveness for more applications. | ['Liang Lin', 'Pengxu Wei', 'Ziyi Dong'] | 2022-11-21 | null | null | null | null | ['novel-concepts'] | ['reasoning'] | [ 5.06547272e-01 -1.29767790e-01 7.20391944e-02 -2.56489366e-01
-4.77625191e-01 -4.08481777e-01 8.80984187e-01 -2.94175774e-01
-2.76027679e-01 7.54225552e-01 3.80355239e-01 2.96907067e-01
-3.90769914e-02 -9.38071728e-01 -6.47044778e-01 -8.06547344e-01
5.51319301e-01 3.88721913e-01 1.11440271e-01 -6.05438888e-01
3.95945340e-01 2.93368280e-01 -1.72428727e+00 2.71117598e-01
1.23782897e+00 7.03977883e-01 6.31820679e-01 5.29084444e-01
-3.17370474e-01 6.12369478e-01 -7.57511258e-01 -6.19500101e-01
2.90572643e-01 -6.15494132e-01 -5.16994178e-01 2.63536453e-01
5.26779830e-01 -4.58056331e-01 -1.93740323e-01 9.82460320e-01
7.40552306e-01 2.21875906e-01 7.42805004e-01 -1.29187226e+00
-1.51081645e+00 5.64442933e-01 -6.37607515e-01 8.39760303e-02
3.85374725e-01 6.37286305e-01 8.58992696e-01 -8.27310085e-01
1.02267790e+00 1.40807760e+00 2.15197220e-01 1.06717026e+00
-1.33511186e+00 -9.29432273e-01 2.40454689e-01 -6.64023235e-02
-1.32276487e+00 -4.61697370e-01 8.26403558e-01 -3.56350631e-01
7.28058696e-01 1.40139386e-01 7.25974441e-01 1.48361087e+00
9.77118909e-02 8.94865572e-01 1.04838037e+00 -2.03332722e-01
1.38769299e-03 5.18299460e-01 -5.47963381e-01 5.59201777e-01
9.55866575e-02 2.48466432e-01 -4.80930567e-01 1.13802068e-01
1.02843475e+00 1.53961163e-02 -4.82583404e-01 -3.66872191e-01
-1.54111040e+00 8.43276083e-01 4.52404261e-01 3.33769649e-01
-3.13124329e-01 -7.52135813e-02 1.23214915e-01 3.13651770e-01
2.16985717e-01 1.01753557e+00 -2.35015482e-01 -8.06612521e-02
-9.24589336e-01 4.19089049e-01 4.26748425e-01 1.26692212e+00
8.03500175e-01 3.14519107e-01 -7.19071388e-01 1.08673334e+00
-1.66647390e-01 6.14266574e-01 7.65290499e-01 -8.27746093e-01
4.30645019e-01 8.04022789e-01 -4.04339880e-02 -1.13382435e+00
5.93547970e-02 -4.83965099e-01 -1.15789199e+00 7.81589970e-02
-9.30677801e-02 -2.74099380e-01 -1.02331269e+00 1.95482516e+00
1.40501395e-01 -2.08318591e-01 7.76617900e-02 8.41901243e-01
7.85707295e-01 9.10548449e-01 -2.48832628e-03 -1.50058806e-01
1.01021504e+00 -1.14890862e+00 -6.89541519e-01 -1.20977908e-01
2.99091250e-01 -8.83369803e-01 1.40777874e+00 1.69746399e-01
-1.19124603e+00 -9.12068009e-01 -1.04120791e+00 6.66558323e-03
-3.58302027e-01 2.30370611e-01 4.60005462e-01 3.93500775e-01
-1.03775287e+00 5.93690276e-01 -1.21550925e-01 -3.92347038e-01
5.80912173e-01 1.28893107e-01 -2.82822162e-01 -1.71922773e-01
-1.16703820e+00 6.97595954e-01 4.53338444e-01 -3.66688788e-01
-8.37296903e-01 -8.75267088e-01 -7.07692564e-01 4.22383435e-02
4.30026889e-01 -1.09472442e+00 9.52669322e-01 -1.29672921e+00
-1.63577032e+00 6.46833003e-01 2.36107171e-01 -2.70678997e-02
5.72293818e-01 6.99060708e-02 -6.16564572e-01 -9.02088806e-02
2.58143812e-01 1.32551229e+00 1.27839804e+00 -1.50829995e+00
-8.24489355e-01 -3.00993156e-02 2.13214487e-01 4.50707346e-01
-8.21389139e-01 -3.16779971e-01 -4.16355014e-01 -1.07100058e+00
-1.79065838e-01 -8.33881736e-01 -1.56772509e-01 1.75675392e-01
-3.53701770e-01 -6.27205223e-02 6.79110289e-01 -2.17394888e-01
1.23377776e+00 -2.25241327e+00 3.33950132e-01 -1.16327986e-01
1.79960892e-01 3.63072336e-01 -5.54367721e-01 4.01423663e-01
7.85244703e-02 2.95595527e-01 -1.26253724e-01 -7.22109154e-02
-3.39307375e-02 1.45585418e-01 -3.59952241e-01 -3.35340083e-01
5.61228752e-01 9.49435472e-01 -1.12529051e+00 -5.39451420e-01
7.44013190e-02 4.20384675e-01 -7.52806604e-01 3.38315040e-01
-3.96831274e-01 3.87918651e-01 -3.88080329e-01 4.98229146e-01
5.01594901e-01 -3.75017226e-01 -1.07367113e-01 -4.40783888e-01
1.10184431e-01 -3.01787525e-01 -1.23167479e+00 1.75382590e+00
-5.04681170e-01 4.92862016e-01 -2.94954926e-01 -5.13779461e-01
1.09554636e+00 7.15968832e-02 1.28967032e-01 -1.01509857e+00
-8.76324177e-02 2.02624425e-02 -1.49103343e-01 -5.52671611e-01
9.26495671e-01 -2.35828534e-01 1.14351474e-01 5.23752034e-01
1.52902484e-01 -3.85075748e-01 4.20220226e-01 5.20095646e-01
7.36086369e-01 3.16905230e-01 6.50463924e-02 -7.10737258e-02
4.76442724e-01 1.77669376e-02 4.88040686e-01 7.40810037e-01
2.90810224e-02 9.01376307e-01 2.57902563e-01 -2.46120393e-01
-1.16869545e+00 -1.11921835e+00 2.56696731e-01 1.06195021e+00
4.01701331e-01 -3.99631798e-01 -5.54232180e-01 -6.99873865e-01
-1.64702520e-01 8.45339954e-01 -6.61213279e-01 -4.66381162e-01
-3.74412030e-01 -6.78601146e-01 3.69984925e-01 4.82944310e-01
7.26105630e-01 -1.33043838e+00 -2.54170179e-01 3.13676447e-01
-2.95012504e-01 -8.45664442e-01 -9.44813132e-01 -3.32905591e-01
-6.87485099e-01 -7.60377169e-01 -1.00550783e+00 -8.88807774e-01
9.07201946e-01 4.55076188e-01 1.22188163e+00 1.46677047e-01
-3.54766041e-01 3.04804116e-01 -3.54616940e-01 -2.25795582e-01
-6.00669920e-01 1.81480888e-02 -1.08883612e-01 1.78917140e-01
-5.14616668e-02 -5.00706315e-01 -7.69012332e-01 3.31335813e-01
-1.25063586e+00 4.93599862e-01 8.15880358e-01 1.11875081e+00
5.18818378e-01 1.76261112e-01 7.99895465e-01 -9.28432465e-01
1.04525816e+00 -1.47866666e-01 -2.35224441e-01 5.31566322e-01
-8.76085043e-01 2.44289145e-01 7.51351237e-01 -8.24895620e-01
-1.44620848e+00 -1.45671099e-01 -1.14015080e-02 -3.99761707e-01
5.75246811e-02 3.50576565e-02 -2.63495654e-01 1.07809961e-01
8.11496854e-01 6.26632392e-01 -3.61788878e-03 -1.48404464e-01
7.87567258e-01 4.50360745e-01 5.38388669e-01 -7.41392374e-01
9.99984622e-01 3.70443285e-01 -5.27239382e-01 -4.73883152e-01
-5.66332042e-01 -6.11533178e-03 -4.52763617e-01 -1.17672615e-01
7.70586789e-01 -1.02410972e+00 -1.35310411e-01 4.81782824e-01
-1.01794827e+00 -1.72103524e-01 -6.11030281e-01 -5.36052585e-02
-4.61987287e-01 6.12319075e-02 -3.65418941e-01 -4.37414795e-01
-4.79518414e-01 -1.10612452e+00 9.73999262e-01 6.36073351e-01
-1.96518928e-01 -8.99175584e-01 -9.96798202e-02 2.19865009e-01
7.28159308e-01 2.34340012e-01 1.07487333e+00 -2.34185338e-01
-8.08532894e-01 2.51309127e-02 -3.24497014e-01 3.84246320e-01
4.37801063e-01 2.91729152e-01 -6.22255087e-01 -3.46579671e-01
-3.24535519e-01 -4.65677977e-01 8.36290777e-01 -1.71265498e-01
1.03821921e+00 -5.03817558e-01 -2.21024171e-01 6.50255024e-01
1.40574253e+00 2.22419992e-01 6.89831853e-01 2.89717406e-01
7.76368201e-01 3.63979518e-01 5.85649669e-01 4.97874975e-01
3.47613871e-01 5.81849635e-01 1.41775668e-01 -2.83614248e-01
-5.19484341e-01 -6.41321838e-01 2.66983539e-01 7.18733311e-01
3.78650287e-03 -3.81383389e-01 -3.09737831e-01 7.00863838e-01
-1.65561616e+00 -1.15806961e+00 3.81121188e-01 1.89544725e+00
1.29593003e+00 1.05430476e-01 -9.38955043e-03 -2.65810937e-01
6.50858462e-01 2.42661759e-01 -6.80328488e-01 -2.76466936e-01
-4.38939989e-01 1.24447076e-02 -2.91331653e-02 -1.73103213e-02
-7.04496562e-01 1.16559649e+00 5.74036455e+00 1.19427121e+00
-1.34010506e+00 2.67984699e-02 6.52716756e-01 -3.69255543e-01
-7.80201614e-01 -1.84871927e-01 -7.43273318e-01 5.01985073e-01
2.63603002e-01 -3.90927017e-01 6.49750710e-01 8.22466969e-01
-2.32347697e-02 2.20370531e-01 -9.32531655e-01 1.11138582e+00
3.99781734e-01 -1.69026458e+00 8.22144330e-01 -1.93246529e-01
1.24285591e+00 -5.13299227e-01 3.65073115e-01 5.02239645e-01
4.10553187e-01 -8.61715615e-01 7.26853013e-01 5.20506442e-01
1.09564292e+00 -7.33548939e-01 3.47583592e-01 3.12321484e-01
-9.14800882e-01 -1.72614768e-01 -4.09164310e-01 2.40495473e-01
1.44001991e-01 2.82586038e-01 -6.63432062e-01 4.90908623e-01
5.24807215e-01 7.97177434e-01 -9.06175196e-01 5.94694018e-01
-2.81460106e-01 5.05008548e-02 2.00751796e-01 -1.04536042e-01
2.25763172e-01 -1.40413195e-01 4.93275553e-01 1.00321269e+00
5.08667648e-01 2.38665789e-01 1.77150872e-02 1.29943216e+00
-2.73031861e-01 1.26337424e-01 -5.89579821e-01 -3.65867764e-01
3.26169133e-01 1.45213163e+00 -5.00868976e-01 -5.68324864e-01
-2.55408794e-01 1.16348469e+00 3.52780342e-01 4.07314926e-01
-7.43370354e-01 -4.65579033e-01 6.50526106e-01 1.62917510e-01
2.60828197e-01 9.16511789e-02 -3.06370914e-01 -1.30211854e+00
-9.15908515e-02 -1.16134024e+00 1.41763970e-01 -1.13890588e+00
-1.50515437e+00 9.10293341e-01 -1.49576962e-01 -1.24028230e+00
-1.94772631e-01 -3.17331493e-01 -7.05883563e-01 7.25384951e-01
-1.34314096e+00 -1.23205972e+00 -4.40831184e-01 7.48911262e-01
8.85138512e-01 -5.39462090e-01 6.15978897e-01 2.20337600e-01
-5.96941948e-01 7.86807597e-01 2.95085460e-02 -1.37893319e-01
1.03788877e+00 -1.17411041e+00 3.94268990e-01 7.42868900e-01
1.03644408e-01 6.48104429e-01 5.79477370e-01 -7.11242616e-01
-1.25158095e+00 -1.29110873e+00 6.61473691e-01 -2.67747343e-01
3.71481210e-01 -2.74596810e-01 -8.55135381e-01 2.53470689e-01
4.19162005e-01 -4.16472256e-01 3.84510428e-01 -2.10603103e-01
-3.80013078e-01 -2.48771936e-01 -1.02712309e+00 1.10282409e+00
1.30710459e+00 -2.25454971e-01 -4.65223670e-01 2.69025594e-01
9.43660796e-01 -2.35541448e-01 -7.10831106e-01 4.00132865e-01
5.22669315e-01 -1.01954293e+00 9.31790352e-01 -5.07657528e-01
8.95770788e-01 -2.48217359e-01 7.28725195e-02 -1.64441848e+00
-7.58597255e-01 -6.78410769e-01 1.97729945e-01 1.61650968e+00
4.82538670e-01 -3.75372142e-01 5.25285065e-01 5.09465456e-01
4.34101932e-02 -8.11269999e-01 -2.91156799e-01 -7.29862869e-01
6.82251453e-02 1.98887885e-01 9.77334559e-01 9.78265464e-01
-4.25727427e-01 6.44452691e-01 -6.47448659e-01 -2.80757934e-01
3.88486654e-01 3.74765754e-01 9.86369848e-01 -8.00555110e-01
-2.50917435e-01 -6.39422417e-01 -1.14261523e-01 -1.15301239e+00
-5.59499860e-02 -6.97623372e-01 3.21549848e-02 -1.57999969e+00
4.83654976e-01 -5.27299464e-01 -1.03412360e-01 3.09472680e-01
-5.26101768e-01 2.46880010e-01 4.44890887e-01 2.52673656e-01
-6.32162511e-01 8.94306779e-01 1.99359441e+00 -4.83050436e-01
-2.96358585e-01 -2.64310449e-01 -1.20061409e+00 3.73965502e-01
7.56003857e-01 -2.44834408e-01 -7.89953947e-01 -6.22690499e-01
2.70269871e-01 -1.67277098e-01 1.74051881e-01 -9.13974762e-01
8.96401703e-02 -4.35664654e-01 7.02749431e-01 -2.93332577e-01
1.99745506e-01 -5.56375325e-01 2.92280495e-01 1.89511225e-01
-5.91588199e-01 1.48719221e-01 4.52861153e-02 6.78457797e-01
-1.49575591e-01 -1.93804726e-02 8.62175047e-01 -3.45316440e-01
-9.67563570e-01 5.15895784e-01 2.77671535e-02 2.74535090e-01
9.56392825e-01 -4.64566350e-01 -4.31829542e-01 -4.75751579e-01
-4.31646436e-01 2.74633944e-01 7.86597311e-01 1.06030619e+00
9.23478603e-01 -1.71802735e+00 -9.69012141e-01 4.58062738e-01
4.86203849e-01 -1.34431168e-01 4.38734591e-01 2.79592395e-01
-1.66385278e-01 6.94944290e-03 -5.38840294e-01 -5.03426254e-01
-1.01223457e+00 7.90671706e-01 1.64406598e-01 -2.02489510e-01
-5.43969989e-01 8.26943517e-01 6.51190817e-01 -2.96188533e-01
7.83021674e-02 5.24574518e-02 -2.10196838e-01 -2.59862058e-02
5.73703229e-01 1.32150933e-01 -3.88060898e-01 -4.59296942e-01
1.10222206e-01 5.66732049e-01 -4.23640311e-01 -6.81198984e-02
1.19038880e+00 -3.28077257e-01 8.11514109e-02 1.03130043e-01
9.44567561e-01 -1.44552663e-01 -1.52052248e+00 -3.73608053e-01
-3.92353535e-01 -6.65264845e-01 -2.01085016e-01 -1.17094278e+00
-1.30012012e+00 7.25830495e-01 5.07089734e-01 -5.52780516e-02
1.43754804e+00 -3.04554105e-01 1.02887607e+00 1.96333990e-01
3.05852383e-01 -1.21326363e+00 7.87415564e-01 2.66195863e-01
1.25323284e+00 -1.29823077e+00 -1.52369931e-01 -1.47458777e-01
-1.13013732e+00 9.52170968e-01 1.11200869e+00 1.05325572e-01
2.05586627e-01 -1.23571686e-01 3.30294557e-02 4.32609729e-02
-9.78243589e-01 -7.13944361e-02 2.79941291e-01 8.57460916e-01
2.06147224e-01 -6.76039830e-02 -1.17265165e-01 5.24342597e-01
-2.43623778e-01 8.93719681e-03 4.34555113e-01 6.54262602e-01
-3.33950520e-01 -1.11837852e+00 -1.55053481e-01 5.13413429e-01
-1.31036639e-02 -1.82854682e-01 -4.48845178e-01 7.88612247e-01
4.86887753e-01 8.24362993e-01 -2.28292383e-02 -4.89125103e-01
5.37577927e-01 -2.23979756e-01 4.85341132e-01 -6.37135923e-01
-6.69715047e-01 -4.71254885e-02 -2.47128665e-01 -3.90260696e-01
-2.84572363e-01 -4.13002014e-01 -9.84616280e-01 -3.78326923e-01
-3.23484033e-01 -6.75702617e-02 4.26637828e-01 6.40591264e-01
6.40476942e-01 7.90415347e-01 7.99787462e-01 -6.62273049e-01
-6.12158120e-01 -7.81329215e-01 -5.53820193e-01 9.09309030e-01
4.65844683e-02 -5.88134408e-01 -1.85695350e-01 3.47786665e-01] | [11.472586631774902, -0.37425002455711365] |
6031d38d-e48d-4a79-a422-e8256f4fe9e0 | automatic-detection-and-rectification-of | 2303.05763 | null | https://arxiv.org/abs/2303.05763v1 | https://arxiv.org/pdf/2303.05763v1.pdf | Automatic Detection and Rectification of Paper Receipts on Smartphones | We describe the development of a real-time smartphone app that allows the user to digitize paper receipts in a novel way by "waving" their phone over the receipts and letting the app automatically detect and rectify the receipts for subsequent text recognition. We show that traditional computer vision algorithms for edge and corner detection do not robustly detect the non-linear and discontinuous edges and corners of a typical paper receipt in real-world settings. This is particularly the case when the colors of the receipt and background are similar, or where other interfering rectangular objects are present. Inaccurate detection of a receipt's corner positions then results in distorted images when using an affine projective transformation to rectify the perspective. We propose an innovative solution to receipt corner detection by treating each of the four corners as a unique "object", and training a Single Shot Detection MobileNet object detection model. We use a small amount of real data and a large amount of automatically generated synthetic data that is designed to be similar to real-world imaging scenarios. We show that our proposed method robustly detects the four corners of a receipt, giving a receipt detection accuracy of 85.3% on real-world data, compared to only 36.9% with a traditional edge detection-based approach. Our method works even when the color of the receipt is virtually indistinguishable from the background. Moreover, our method is trained to detect only the corners of the central target receipt and implicitly learns to ignore other receipts, and other rectangular objects. Including synthetic data allows us to train an even better model. These factors are a major advantage over traditional edge detection-based approaches, allowing us to deliver a much better experience to the user. | ['Ikuo Kitagishi', 'Masashi Tanaka', 'Edward Whittaker'] | 2023-03-10 | null | null | null | null | ['edge-detection'] | ['computer-vision'] | [ 4.61817682e-01 -3.90124410e-01 2.95566052e-01 1.28882721e-01
-5.02681375e-01 -7.44168460e-01 5.41705728e-01 -5.22786789e-02
-2.87319213e-01 1.80630505e-01 -3.53548914e-01 -4.09352750e-01
4.61953938e-01 -5.78781724e-01 -9.53610063e-01 -3.47610146e-01
7.37639964e-02 3.39554280e-01 5.96988440e-01 -1.21582737e-02
3.94136310e-01 7.05775619e-01 -1.48733175e+00 3.79317522e-01
5.76931357e-01 8.12481880e-01 3.96985412e-01 1.31120539e+00
1.36771023e-01 3.60619128e-01 -7.04552650e-01 -1.92318156e-01
4.93961573e-01 -1.06172331e-01 2.43602437e-03 2.71515697e-01
9.11614835e-01 -7.57068098e-01 -4.06753898e-01 9.15969908e-01
3.71668756e-01 -9.18857753e-02 6.72731876e-01 -1.26641154e+00
-5.07020712e-01 -1.65821701e-01 -1.15163672e+00 -5.85799739e-02
7.14555025e-01 3.58591639e-02 3.38065416e-01 -1.09001791e+00
6.85114026e-01 1.01398849e+00 1.21823597e+00 4.16498959e-01
-9.72231328e-01 -5.47596693e-01 6.24606498e-02 1.04076222e-01
-1.51939738e+00 -5.28207064e-01 7.25371778e-01 -2.83328354e-01
6.45091891e-01 5.24793565e-01 4.49994266e-01 1.09758902e+00
2.28968635e-01 8.51143301e-01 8.42727244e-01 -7.20010638e-01
1.45958945e-01 3.54845792e-01 7.46447891e-02 8.56883287e-01
3.40076715e-01 -2.19322830e-01 -2.62518674e-01 -2.43821532e-01
8.48169148e-01 3.21586967e-01 -5.29941857e-01 -6.29687011e-01
-1.28813982e+00 1.05760254e-01 1.76879600e-01 8.99019167e-02
-4.38526362e-01 1.37642354e-01 -8.25201198e-02 -1.64887413e-01
8.27071741e-02 2.63914149e-02 -1.77157030e-01 -1.97995588e-01
-1.10705924e+00 3.12867798e-02 9.57490504e-01 1.29751158e+00
4.91768181e-01 -5.79702994e-03 9.50414389e-02 5.25644422e-01
6.30600005e-02 7.51141310e-01 4.07667384e-02 -8.07556510e-01
5.17766118e-01 5.25877416e-01 6.85171902e-01 -1.19542229e+00
-4.18487459e-01 1.52711729e-02 -5.40568411e-01 5.60711086e-01
8.54272485e-01 -4.89422232e-01 -1.03863561e+00 1.11256599e+00
4.09757912e-01 5.09525299e-01 -2.23987296e-01 8.80551875e-01
4.67445850e-01 5.54701030e-01 -5.91117978e-01 1.36582643e-01
1.39425075e+00 -6.59658074e-01 -5.62336683e-01 -3.11909646e-01
4.33005095e-01 -1.09765732e+00 1.23143959e+00 4.65898931e-01
-8.62333179e-01 -2.16574907e-01 -1.30555069e+00 1.05564609e-01
-4.37303811e-01 4.71620470e-01 4.84707117e-01 9.39384937e-01
-1.01097953e+00 1.79860547e-01 -5.15224993e-01 -4.36524421e-01
1.75869718e-01 3.30924898e-01 -1.33103043e-01 -3.31331521e-01
-4.66092557e-01 7.15183616e-01 -1.43540189e-01 1.17836803e-01
-4.60396171e-01 -6.41703308e-01 -6.43808663e-01 8.26201960e-02
5.47840357e-01 -3.65633667e-01 1.30785012e+00 -1.20502222e+00
-1.16117203e+00 7.27824211e-01 -4.14924353e-01 2.85905898e-02
9.26486373e-01 -1.61198780e-01 -4.29619491e-01 2.46612191e-01
-9.10122097e-02 1.66551977e-01 1.20353758e+00 -1.68434989e+00
-8.00468683e-01 -3.66886526e-01 -7.42659196e-02 1.93116933e-01
-1.85578257e-01 -5.80467517e-03 -9.32282031e-01 -4.97953624e-01
-3.20325196e-02 -1.03569829e+00 2.38909096e-01 2.62177855e-01
-4.09666061e-01 4.14465487e-01 1.47122526e+00 -6.96889043e-01
8.25652540e-01 -2.11015296e+00 -7.29310393e-01 3.03933114e-01
1.69626296e-01 4.41401482e-01 -2.18487963e-01 3.19184810e-01
1.39414653e-01 -2.19818890e-01 5.49718216e-02 -5.24360061e-01
-2.61251003e-01 -3.40680271e-01 -2.73848087e-01 6.19319022e-01
-1.13862537e-01 6.27444744e-01 -9.90715981e-01 -7.24190623e-02
3.15460443e-01 7.70959258e-01 -2.29129925e-01 -1.04044318e-01
6.79560006e-02 -1.93721324e-01 -8.58093649e-02 8.84478092e-01
1.16949642e+00 -3.74985747e-02 2.39981353e-01 -1.58151686e-01
-5.00716716e-02 -2.63074696e-01 -1.67856514e+00 1.04163218e+00
-3.54550630e-01 1.13758123e+00 3.14755291e-01 -2.78000861e-01
6.44169331e-01 9.16260779e-02 3.15603614e-01 -7.11158335e-01
-3.39693651e-02 -3.11780870e-02 -2.36887991e-01 -4.34760481e-01
9.18179870e-01 4.00354385e-01 1.89394593e-01 5.69204211e-01
-6.23408377e-01 1.02887936e-02 -1.27099916e-01 4.13094044e-01
1.28524303e+00 4.83903550e-02 -8.97111148e-02 2.63913751e-01
5.44896424e-02 -1.01287119e-01 3.35439086e-01 1.18992805e+00
-1.24165677e-01 1.20574617e+00 2.42824659e-01 -1.33968487e-01
-9.84255433e-01 -1.10366869e+00 1.37439042e-01 8.51237953e-01
7.01125383e-01 -1.59206942e-01 -1.03491163e+00 -6.83426380e-01
7.71738663e-02 7.16448426e-01 -5.03664732e-01 9.58309695e-02
-5.35901606e-01 -4.57717001e-01 4.99652356e-01 5.18842220e-01
5.69553018e-01 -6.08193219e-01 -8.04376423e-01 -6.44492656e-02
-1.51578739e-01 -1.26252711e+00 -7.79491067e-01 -1.65471226e-01
-5.93168497e-01 -1.37527132e+00 -9.30892885e-01 -9.06495452e-01
9.84164238e-01 1.21593451e+00 7.17788875e-01 2.93915808e-01
-5.48401833e-01 9.47363794e-01 -9.76081342e-02 -6.12188935e-01
-2.84209847e-01 -4.52602446e-01 8.89565349e-02 3.40381503e-01
2.78386116e-01 2.47856099e-02 -7.39663363e-01 4.56411839e-01
-7.68469930e-01 5.99733889e-02 3.16235691e-01 4.79409277e-01
1.84938982e-01 1.01779491e-01 1.76646322e-01 -7.78629780e-01
6.75639451e-01 -1.68565288e-01 -5.97065628e-01 3.51767719e-01
-2.35427678e-01 -4.59733039e-01 6.56839550e-01 -7.95730531e-01
-1.21486378e+00 2.47695684e-01 4.74598318e-01 -4.50415552e-01
-2.80044764e-01 -2.03546315e-01 -1.50355007e-02 -3.85717034e-01
6.94741905e-01 1.83026358e-01 -2.18083993e-01 -3.00510436e-01
1.68370381e-01 9.96637702e-01 6.31016552e-01 -6.16294108e-02
8.00282240e-01 8.56500268e-01 -1.57587066e-01 -1.32983947e+00
1.42707769e-02 -6.88076317e-01 -4.71805602e-01 -3.38896185e-01
2.56094426e-01 -6.38427854e-01 -8.13532233e-01 9.28345621e-01
-1.44544101e+00 -4.69026536e-01 1.53312132e-01 3.37492853e-01
-2.29485497e-01 6.49987698e-01 -5.41218758e-01 -1.01109493e+00
-1.02535047e-01 -8.69840741e-01 1.25973499e+00 4.57458943e-01
-2.33653598e-02 -8.16259742e-01 -3.62049937e-01 2.18526348e-02
3.01180959e-01 -1.27458116e-02 5.38389087e-01 -1.96094096e-01
-6.98047578e-01 -8.02342296e-01 -5.29592872e-01 -2.48360131e-02
2.49070242e-01 2.65821904e-01 -1.14332485e+00 -2.86862999e-01
6.79241568e-02 3.99504900e-01 6.00051284e-01 3.93932849e-01
7.25200653e-01 -4.12717946e-02 -6.75417244e-01 4.18980122e-01
1.37088215e+00 4.86893505e-01 6.68679118e-01 3.80614430e-01
9.16264474e-01 2.59937584e-01 6.83192313e-01 3.75030428e-01
4.40076053e-01 6.74714923e-01 3.74143124e-01 -4.59946662e-01
-3.20284307e-01 -1.29675969e-01 4.06285733e-01 8.48584846e-02
-3.56481131e-03 -3.97574812e-01 -8.94677758e-01 3.99213910e-01
-1.71124780e+00 -1.02006304e+00 -5.03535509e-01 2.50694895e+00
2.21211165e-01 1.00935269e-02 1.23490259e-01 -5.80100119e-02
1.10239661e+00 -1.80911005e-01 -4.42155898e-01 -1.90396667e-01
-2.36664843e-02 -2.46418864e-01 1.00640810e+00 5.44018030e-01
-1.08034992e+00 7.98843205e-01 6.94458437e+00 3.62770468e-01
-1.42972791e+00 -2.48655513e-01 2.84427106e-01 5.87717257e-02
-4.33583744e-02 -5.35313226e-02 -8.04335296e-01 5.39885581e-01
3.51188570e-01 3.00250292e-01 6.34061337e-01 7.06729054e-01
3.45630258e-01 -5.70347011e-01 -1.06483507e+00 1.36848319e+00
5.20921946e-01 -1.19931781e+00 -3.00133020e-01 -1.50662735e-01
6.26209497e-01 -3.47718559e-02 1.27514988e-01 -9.92755964e-02
-1.18820034e-01 -6.56360090e-01 7.87617266e-01 5.71750402e-01
7.98443258e-01 -3.16862792e-01 3.58715177e-01 5.57648897e-01
-1.05388570e+00 2.60634776e-02 -1.64359629e-01 -1.20133720e-01
-7.67720938e-02 2.94544011e-01 -1.48414314e+00 -2.88105570e-02
5.66479981e-01 1.94036201e-01 -6.49830520e-01 1.36635208e+00
1.19916134e-01 2.88239390e-01 -6.33747458e-01 -1.05663814e-01
-4.25930060e-02 -1.48391813e-01 6.81158960e-01 1.61651731e+00
4.91198897e-01 -1.90565318e-01 -1.61617070e-01 5.95271349e-01
-1.80945277e-01 -2.12126702e-01 -9.33222532e-01 3.01013917e-01
5.67349553e-01 1.31407785e+00 -1.09764147e+00 -3.59411597e-01
-7.30568945e-01 1.45354366e+00 -9.74043682e-02 8.51812899e-01
-9.41177428e-01 -9.26328421e-01 2.84238040e-01 4.29762065e-01
5.29116273e-01 -3.94852668e-01 -3.67746592e-01 -1.33231556e+00
4.15349424e-01 -8.73304963e-01 -1.62617583e-02 -1.12662923e+00
-7.42437959e-01 1.04110852e-01 -3.42614323e-01 -1.34388554e+00
6.62239194e-02 -7.84901500e-01 -9.73681450e-01 6.15517676e-01
-1.35631001e+00 -1.18950582e+00 -7.50218332e-01 5.36306918e-01
5.56787074e-01 1.21553898e-01 7.31154442e-01 3.43352824e-01
-3.98341924e-01 7.60658145e-01 5.99076033e-01 2.91867286e-01
8.41103256e-01 -1.18862963e+00 6.52174592e-01 1.15152633e+00
2.09565073e-01 6.41473353e-01 6.25388682e-01 -7.24855721e-01
-1.83194482e+00 -7.45183587e-01 4.30731446e-01 -5.47619820e-01
3.74295205e-01 -7.52811372e-01 -8.90819013e-01 8.06290388e-01
-8.22660550e-02 -3.42224315e-02 2.05096647e-01 -2.27220684e-01
-3.00266147e-01 -9.68992561e-02 -1.22854829e+00 1.04213965e+00
8.70144486e-01 -5.24183095e-01 -3.27567339e-01 3.74732614e-01
3.92218456e-02 -6.89898372e-01 -5.70541397e-02 6.51889518e-02
9.47828531e-01 -1.04738235e+00 1.05004704e+00 -2.60252152e-02
-2.11025387e-01 -4.63974625e-01 1.20551102e-01 -1.05738318e+00
-4.31712195e-02 -7.88859785e-01 -2.90765941e-01 1.03792655e+00
1.79863885e-01 -6.45103157e-01 1.01676977e+00 7.19947100e-01
3.80133167e-02 -3.55877459e-01 -6.79505527e-01 -5.62736094e-01
-6.76440716e-01 -6.72744632e-01 2.33368814e-01 8.37395728e-01
-8.50230306e-02 4.43913229e-02 -4.24569100e-01 4.74994063e-01
6.17688358e-01 1.07692748e-01 1.12604654e+00 -8.05757821e-01
-2.26545304e-01 -2.37942293e-01 -4.46472377e-01 -1.29644763e+00
-4.58627969e-01 -3.71096611e-01 1.89976320e-01 -1.46639395e+00
9.01738778e-02 -6.22392714e-01 1.34125829e-01 2.13391840e-01
-1.73006341e-01 4.67821360e-01 2.85559207e-01 4.00325703e-03
-5.12491643e-01 -1.90787822e-01 8.87738705e-01 -1.75946802e-01
-5.00949264e-01 2.64490098e-01 -6.04644418e-01 1.18442106e+00
5.96631706e-01 -2.53140092e-01 -1.76249370e-01 -5.20751715e-01
1.88863594e-02 -2.07685545e-01 6.08530045e-01 -1.01901531e+00
5.24420440e-01 -5.47878295e-02 9.51962829e-01 -6.99220896e-01
5.01220226e-01 -1.10868704e+00 -1.42130956e-01 3.18553001e-01
2.79929996e-01 -4.15445901e-02 4.00449574e-01 6.92214251e-01
5.45928180e-01 -4.00187254e-01 5.03917158e-01 3.97236124e-02
-8.20287526e-01 -2.06460804e-01 -6.18768632e-01 -3.24110597e-01
1.32015014e+00 -8.12300503e-01 -3.53739411e-01 -5.99915743e-01
-2.18021408e-01 -6.71003833e-02 1.00017798e+00 5.47863245e-01
8.95747602e-01 -9.98744130e-01 -2.78823584e-01 5.12152791e-01
-7.47057572e-02 -2.48348221e-01 1.65946737e-01 7.73546278e-01
-9.21658278e-01 1.02142289e-01 1.46821156e-01 -6.69313490e-01
-1.63854122e+00 4.46482331e-01 3.68295193e-01 3.42384905e-01
-9.07452524e-01 4.21036750e-01 1.09217539e-01 2.75002589e-04
5.72288215e-01 -4.27677006e-01 2.60702521e-01 -1.77560166e-01
8.34257543e-01 6.79333508e-01 2.26509809e-01 -4.85394746e-01
-3.78069967e-01 7.78843641e-01 -2.70866036e-01 2.63717934e-03
8.19963217e-01 -1.10923909e-01 5.18798530e-01 2.30515257e-01
9.76567149e-01 6.84716225e-01 -1.45886278e+00 -5.47222011e-02
-2.60621697e-01 -7.38401473e-01 -4.56461199e-02 -6.45920813e-01
-6.79056287e-01 7.32680738e-01 8.85319054e-01 2.27191463e-01
8.94413710e-01 -4.38095719e-01 7.99536467e-01 5.36118150e-01
3.86963516e-01 -1.36924970e+00 8.99938196e-02 4.80527282e-01
5.21418452e-01 -1.19901383e+00 -7.95775205e-02 -5.57906806e-01
-5.85561693e-01 1.31982148e+00 4.86362070e-01 -1.44073993e-01
1.49615347e-01 6.99627161e-01 2.02348694e-01 1.44964773e-02
-4.62549962e-02 1.30238757e-01 1.46610931e-01 9.68413770e-01
6.94328547e-02 -3.87113704e-03 2.87648082e-01 1.92243427e-01
3.24840754e-01 -2.03527585e-02 8.95710349e-01 1.22773933e+00
-3.92582953e-01 -6.18444026e-01 -1.01528943e+00 4.76619512e-01
-3.43499154e-01 1.82621852e-02 -5.59632421e-01 8.55086863e-01
-6.51922002e-02 8.29598963e-01 2.31742680e-01 -3.76886010e-01
5.03081322e-01 -3.86839174e-02 5.53810179e-01 -4.74046439e-01
-2.42156625e-01 1.45520434e-01 -1.93616271e-01 -4.22859997e-01
8.93275961e-02 -5.41301250e-01 -1.43853807e+00 -3.90044004e-01
-5.65785110e-01 -3.76916081e-01 9.28951681e-01 6.27227247e-01
5.25265098e-01 2.93147385e-01 5.66402972e-01 -1.40647328e+00
-2.98259705e-01 -5.23115039e-01 -6.14575028e-01 3.49182278e-01
5.27179897e-01 -4.33371663e-01 -3.42496127e-01 7.59171247e-02] | [8.549877166748047, -1.7416656017303467] |
0ddfdedd-1523-4c17-9180-2744e0987c00 | deep-ahs-a-deep-learning-approach-to-acoustic | 2302.09252 | null | https://arxiv.org/abs/2302.09252v1 | https://arxiv.org/pdf/2302.09252v1.pdf | Deep AHS: A Deep Learning Approach to Acoustic Howling Suppression | In this paper, we formulate acoustic howling suppression (AHS) as a supervised learning problem and propose a deep learning approach, called Deep AHS, to address it. Deep AHS is trained in a teacher forcing way which converts the recurrent howling suppression process into an instantaneous speech separation process to simplify the problem and accelerate the model training. The proposed method utilizes properly designed features and trains an attention based recurrent neural network (RNN) to extract the target signal from the microphone recording, thus attenuating the playback signal that may lead to howling. Different training strategies are investigated and a streaming inference method implemented in a recurrent mode used to evaluate the performance of the proposed method for real-time howling suppression. Deep AHS avoids howling detection and intrinsically prohibits howling from happening, allowing for more flexibility in the design of audio systems. Experimental results show the effectiveness of the proposed method for howling suppression under different scenarios. | ['Dong Yu', 'Meng Yu', 'Hao Zhang'] | 2023-02-18 | null | null | null | null | ['speech-separation'] | ['speech'] | [ 2.95073569e-01 6.47619814e-02 2.33035877e-01 -3.49134728e-02
-8.64088297e-01 -8.80365893e-02 3.29155564e-01 -2.18827501e-01
-2.35348552e-01 1.96257696e-01 5.77262819e-01 -2.87926674e-01
-1.68980613e-01 -4.19119537e-01 -5.62937856e-01 -8.98947001e-01
2.11279705e-01 -3.37701082e-01 7.77899772e-02 -1.51166812e-01
7.23528042e-02 4.36915100e-01 -1.86193883e+00 2.94721305e-01
7.24239767e-01 8.78800213e-01 7.03966260e-01 9.60794926e-01
1.01164751e-01 9.72626030e-01 -1.16836512e+00 3.74003559e-01
2.19254345e-01 -4.97714072e-01 -1.42633528e-01 -1.06167167e-01
2.86364168e-01 -2.07488209e-01 -2.32807487e-01 8.54074657e-01
1.06112111e+00 4.77669418e-01 4.45366383e-01 -7.40217090e-01
-5.25003932e-02 7.46570170e-01 -2.37844065e-01 3.04531753e-01
1.41718730e-01 3.32686938e-02 6.82219565e-01 -9.06660855e-01
-2.64550447e-01 1.23469555e+00 8.05756807e-01 6.46157682e-01
-9.14926708e-01 -8.23247790e-01 1.13336816e-02 1.86408594e-01
-1.11997771e+00 -7.84199119e-01 1.10740566e+00 -1.76693693e-01
1.07229912e+00 4.30317372e-01 6.38300657e-01 8.55359375e-01
-6.20993562e-02 8.95341754e-01 7.36344993e-01 -8.31653237e-01
2.96100080e-01 -9.73679796e-02 2.69910127e-01 1.45335585e-01
-3.01686764e-01 2.67967612e-01 -7.66513705e-01 1.46078959e-01
3.94117206e-01 -2.10597470e-01 -3.91451895e-01 3.68217140e-01
-5.89449763e-01 5.65353096e-01 9.98523682e-02 3.44610274e-01
-3.89966369e-01 -1.54200783e-02 5.59046388e-01 2.68513858e-01
6.27242267e-01 3.35690141e-01 -3.00748110e-01 4.81627323e-03
-1.21692610e+00 1.13379613e-01 5.34474969e-01 5.38738132e-01
1.32966414e-01 8.05058718e-01 -2.83209622e-01 1.32466829e+00
2.32813463e-01 4.75662380e-01 8.50709438e-01 -8.78596485e-01
4.13514316e-01 1.17123872e-01 -3.29670943e-02 -6.61917448e-01
-4.20095831e-01 -6.32862747e-01 -7.80162454e-01 2.59687901e-01
-8.80410820e-02 -3.37365121e-01 -9.33177292e-01 1.54795134e+00
2.09519282e-01 4.50491726e-01 1.26903281e-01 8.27462912e-01
7.59190142e-01 1.30757797e+00 -1.93195850e-01 -5.69333851e-01
8.54705513e-01 -1.39373970e+00 -1.17719710e+00 -1.03013933e-01
3.03948045e-01 -9.98473942e-01 1.16580677e+00 8.49491477e-01
-1.15518403e+00 -8.64348173e-01 -1.30416179e+00 7.61697590e-02
-8.17673951e-02 2.98933119e-01 4.75901216e-02 7.01710105e-01
-9.74958062e-01 5.49024642e-01 -7.83162951e-01 1.14368699e-01
-3.14975977e-01 4.52415913e-01 2.37844765e-01 3.62376511e-01
-1.28953290e+00 5.85673213e-01 6.92392737e-02 5.88249505e-01
-1.09847665e+00 -7.19546676e-01 -8.10047925e-01 3.15034568e-01
3.29596013e-01 -2.52641469e-01 1.39051962e+00 -9.05421853e-01
-2.06772327e+00 2.95912057e-01 -2.56709963e-01 -7.50883102e-01
2.75355399e-01 -9.11613941e-01 -6.97459519e-01 5.82596436e-02
-6.40498251e-02 2.25433156e-01 1.35184407e+00 -9.23929214e-01
-4.06787992e-01 7.15956017e-02 -3.78210515e-01 2.77923554e-01
-4.56628740e-01 3.70316990e-02 -1.20687336e-01 -1.13121057e+00
7.15467483e-02 -6.67889953e-01 -9.78994891e-02 -7.22692251e-01
-5.01474798e-01 -2.80379713e-01 1.08944046e+00 -7.95759022e-01
1.50685871e+00 -2.45420575e+00 2.90307701e-02 -5.81842624e-02
-2.16071755e-01 1.11204898e+00 -2.28130460e-01 6.28182411e-01
-2.39852741e-01 -2.08197907e-01 -2.47919619e-01 -6.92925572e-01
-2.79683411e-01 4.47773375e-02 -6.20684743e-01 2.30246797e-01
6.11411594e-02 1.91399306e-01 -5.76097369e-01 -6.15209974e-02
6.34142637e-01 8.10840011e-01 -5.71861386e-01 8.50669205e-01
-6.28327802e-02 4.16583985e-01 7.45538324e-02 2.61856347e-01
6.62379622e-01 8.07816684e-01 -1.96933866e-01 -1.42718434e-01
-5.74292064e-01 7.34064341e-01 -1.08205664e+00 1.27964008e+00
-9.34296191e-01 7.42829323e-01 4.66479003e-01 -1.11958373e+00
1.17224455e+00 5.22007823e-01 7.20082298e-02 -8.72405350e-01
6.12895451e-02 1.31614119e-01 8.19598604e-03 -7.26362348e-01
2.05476955e-01 -1.12278596e-01 3.04226249e-01 1.61735281e-01
-4.60663252e-02 -1.57722861e-01 -1.60249755e-01 -4.01489474e-02
1.00439787e+00 8.53067171e-03 -6.65817782e-02 -2.36391202e-01
9.13174987e-01 -7.28680432e-01 6.84020221e-01 8.46856475e-01
-1.86067878e-03 5.04333496e-01 5.80632873e-02 -1.99274868e-01
-7.17788160e-01 -9.95293200e-01 1.88691854e-01 1.32531273e+00
-1.37477472e-01 -3.39135170e-01 -9.75277603e-01 -7.69609958e-02
-3.70366782e-01 9.55762982e-01 -7.82315806e-02 -4.87056285e-01
-9.75958645e-01 -3.02648276e-01 6.31194353e-01 2.01687902e-01
4.47540373e-01 -1.53370428e+00 -6.60223544e-01 4.29068863e-01
-2.28900954e-01 -7.42743731e-01 -5.77638388e-01 6.79814458e-01
-6.72534108e-01 -3.79074901e-01 -8.34477365e-01 -1.03276408e+00
1.85224265e-01 3.45785797e-01 5.52153349e-01 -7.13401362e-02
-2.08981812e-01 1.07615843e-01 -2.90443331e-01 -5.75790584e-01
-6.11501575e-01 -8.36957619e-03 3.00769210e-01 1.19907215e-01
1.08704329e-01 -7.77556717e-01 -6.22728050e-01 1.32986367e-01
-7.79527903e-01 -1.88583702e-01 5.56833625e-01 7.84439564e-01
3.25774223e-01 4.40753937e-01 1.18434453e+00 -5.05115867e-01
9.73167360e-01 -1.89217418e-01 -6.67827487e-01 -1.30998597e-01
-3.67433012e-01 -8.64181481e-03 1.05729210e+00 -5.53341210e-01
-1.24992228e+00 -3.88988331e-02 -5.99277556e-01 -4.67213690e-01
-1.14530340e-01 3.47958386e-01 -3.79961669e-01 2.50000268e-01
2.36559331e-01 3.87086421e-01 -1.53435200e-01 -9.80985463e-01
9.73720923e-02 9.66494739e-01 6.97918832e-01 -5.92134185e-02
4.63558167e-01 -2.11610626e-02 -3.33136916e-01 -1.42344165e+00
-1.08717239e+00 -6.63141727e-01 -1.85682282e-01 -3.99540871e-01
5.18237770e-01 -9.42556381e-01 -6.42329276e-01 7.09524512e-01
-1.10430181e+00 -4.36584622e-01 -2.66343683e-01 5.61455011e-01
-4.86792862e-01 1.07566163e-01 -7.37950683e-01 -1.50275040e+00
-6.48520529e-01 -9.89016593e-01 9.20929253e-01 2.04023421e-01
-1.13060318e-01 -5.90856314e-01 2.51398951e-01 3.25966746e-01
4.43153381e-01 -4.50169295e-01 7.43604302e-01 -4.40086484e-01
-2.05388293e-01 -9.59482715e-02 2.74852961e-01 1.01820838e+00
1.97222099e-01 -1.40721947e-01 -1.51027930e+00 -1.48180857e-01
6.96084380e-01 -6.62448332e-02 1.04192948e+00 7.42228150e-01
1.11126840e+00 -4.20754939e-01 1.82777509e-01 4.87529069e-01
1.11430824e+00 5.94887495e-01 7.53027320e-01 1.08151384e-01
5.80344975e-01 5.95357418e-01 6.26219392e-01 3.61717612e-01
-3.32066864e-01 6.12883627e-01 3.59502137e-01 -4.39228892e-01
-5.68053126e-01 -3.39569807e-01 7.60234714e-01 1.59431696e+00
2.77760148e-01 -3.03871959e-01 -3.49707425e-01 5.82798719e-01
-1.64290404e+00 -8.27400982e-01 -4.24453393e-02 2.23684478e+00
8.74030650e-01 2.93921292e-01 3.10402587e-02 8.49416792e-01
8.81716371e-01 3.17702413e-01 -3.35277259e-01 -8.78887713e-01
1.15748920e-01 5.28622270e-01 2.34696761e-01 8.38903487e-01
-9.12681043e-01 8.75635803e-01 6.12513638e+00 1.05381489e+00
-1.44871390e+00 -6.48676082e-02 2.54893303e-01 -2.82553315e-01
2.94970404e-02 -4.97849464e-01 -8.49375665e-01 3.73928666e-01
1.07012475e+00 1.41153023e-01 5.02922773e-01 6.52792811e-01
9.45578754e-01 -1.78424083e-02 -7.59490490e-01 8.61836791e-01
4.59270887e-02 -8.05071771e-01 -8.46623350e-03 -5.48318505e-01
2.98453599e-01 -3.05924296e-01 3.07012707e-01 3.90219837e-01
-1.94318578e-01 -8.83708835e-01 8.58099997e-01 4.69999284e-01
3.43096077e-01 -1.15266645e+00 5.32164454e-01 5.29161453e-01
-1.12943566e+00 -2.59702265e-01 -1.90659225e-01 -4.80413705e-01
1.06994376e-01 8.22974741e-01 -1.02241933e+00 1.20017946e-01
5.80112517e-01 4.89865988e-01 -2.15524971e-01 1.07810760e+00
-4.52930182e-01 1.32154667e+00 -4.47857499e-01 3.14780101e-02
2.61482149e-01 2.77075893e-03 9.95592535e-01 1.44076693e+00
3.06155324e-01 -1.73426881e-01 -1.00813493e-01 4.46654141e-01
1.83069497e-01 -3.14569175e-02 -4.50593799e-01 1.60008445e-01
5.16664267e-01 9.47924137e-01 -3.71727794e-01 -6.01904355e-02
-2.97534224e-02 7.05333412e-01 -1.12482853e-01 5.76953471e-01
-7.23920107e-01 -7.50950456e-01 3.77275497e-01 1.72804624e-01
3.85055453e-01 1.79807749e-02 -2.40103398e-02 -6.60623789e-01
3.29190269e-02 -7.81567812e-01 2.99260188e-02 -9.47208703e-01
-7.93295622e-01 6.83252931e-01 -2.85870284e-01 -1.24564195e+00
-3.43059093e-01 -3.04172158e-01 -9.78049517e-01 7.45730937e-01
-1.46125162e+00 -7.09709406e-01 1.94553569e-01 4.99246299e-01
1.16790819e+00 -1.42896771e-01 6.85915351e-01 6.36732876e-01
-8.54622066e-01 5.79175770e-01 8.28812048e-02 -2.71274298e-01
6.05288267e-01 -1.03178072e+00 1.95549011e-01 1.03943300e+00
1.70652658e-01 5.22565246e-01 1.05936193e+00 -3.43594879e-01
-9.51152921e-01 -1.17298293e+00 8.07303667e-01 3.26884598e-01
5.14180779e-01 -5.64610660e-01 -9.25952315e-01 2.39444926e-01
3.93764615e-01 -4.65812534e-01 4.68865961e-01 -1.73466310e-01
-9.56445783e-02 -5.10698318e-01 -7.31638670e-01 5.64451516e-01
6.14021778e-01 -6.39548779e-01 -8.18429232e-01 1.34622268e-02
1.05267799e+00 1.68547109e-02 -1.92090884e-01 4.18286145e-01
4.12106961e-01 -1.01611340e+00 9.50824440e-01 -2.85622720e-02
2.64522702e-01 -3.53799611e-01 -8.74493942e-02 -1.43773961e+00
-7.46853203e-02 -9.72351611e-01 -2.13278115e-01 1.58867097e+00
2.42046461e-01 -4.46998030e-01 4.44250315e-01 -1.10611983e-01
-7.02577829e-01 -5.54927528e-01 -8.79881918e-01 -7.21747458e-01
-3.23235929e-01 -6.17634892e-01 2.07083791e-01 3.95733148e-01
-2.01278210e-01 4.83906060e-01 -8.24323535e-01 4.89396214e-01
5.25604665e-01 -1.47617385e-01 4.73586917e-01 -6.77397966e-01
-4.15301234e-01 -1.80252880e-01 -7.08824545e-02 -1.25068414e+00
1.06967531e-01 -3.86253387e-01 5.12833118e-01 -1.10059023e+00
-5.20262241e-01 -4.54562344e-02 -5.85338295e-01 1.95333540e-01
-7.52913579e-02 4.70942780e-02 1.83506638e-01 -1.29186705e-01
-3.08130443e-01 7.52238810e-01 8.68685603e-01 -2.47843817e-01
-5.78410387e-01 4.99677151e-01 -2.95296311e-01 9.77366090e-01
8.10671508e-01 -5.82475185e-01 -6.23266101e-01 -3.68695676e-01
-2.55913258e-01 3.64828736e-01 2.52621889e-01 -1.31576574e+00
3.02899539e-01 3.67849976e-01 -7.42865577e-02 -1.05565906e+00
5.22246063e-01 -7.93803513e-01 -2.41707489e-01 5.80022931e-01
-6.60194695e-01 -5.06902575e-01 4.04898405e-01 4.22982723e-01
-4.73055482e-01 -4.00564790e-01 9.89658892e-01 3.39327186e-01
-3.57936025e-01 -3.84217471e-01 -8.83658648e-01 -1.47307605e-01
4.43486720e-01 1.02474049e-01 2.30466574e-01 -5.85999429e-01
-9.19954658e-01 -5.16593165e-04 -4.76717323e-01 3.20158809e-01
6.54208004e-01 -1.04446542e+00 -5.22879422e-01 5.70383787e-01
-3.86319518e-01 -7.01832548e-02 3.82109523e-01 4.46799457e-01
-3.31471652e-01 3.26316327e-01 2.67455220e-01 -4.42461282e-01
-1.39579117e+00 5.35861492e-01 4.00470376e-01 -1.76085502e-01
-8.85156393e-01 9.49273705e-01 4.14882898e-01 -2.02519193e-01
8.87950480e-01 -3.27069849e-01 -4.88619924e-01 -1.76670030e-01
7.64225662e-01 6.00190878e-01 3.27487618e-01 -2.82646418e-01
-9.82262939e-02 4.16977286e-01 -8.69410560e-02 -1.70663327e-01
1.26701498e+00 -2.99015969e-01 1.23229578e-01 7.89026499e-01
1.06828022e+00 4.66087252e-01 -1.09817028e+00 1.33564100e-01
-4.60197963e-02 -1.10134706e-01 3.93746227e-01 -8.26778233e-01
-1.01070809e+00 1.04445481e+00 7.75985539e-01 4.71259922e-01
1.55555665e+00 -3.85444969e-01 1.09401870e+00 2.45327324e-01
-3.00061524e-01 -1.27102005e+00 1.31583408e-01 4.83847737e-01
1.03647649e+00 -7.23135054e-01 -2.72642761e-01 -3.09393317e-01
-2.36570433e-01 1.00366414e+00 4.25317049e-01 -3.77302676e-01
6.90248013e-01 6.30231380e-01 3.33974808e-01 1.82274193e-01
-7.26414263e-01 3.22166160e-02 -3.50294299e-02 5.06928802e-01
3.85819286e-01 -8.26377571e-02 -1.90688848e-01 7.57942259e-01
-3.73915732e-01 -2.27460384e-01 3.42066556e-01 5.86282015e-01
-5.93478382e-01 -8.37042391e-01 -5.54268122e-01 2.47161865e-01
-6.24067545e-01 -2.69137502e-01 -2.68871456e-01 2.77782768e-01
3.05652738e-01 1.31899452e+00 -7.32038915e-02 -3.43468726e-01
5.24029434e-01 3.86517681e-02 5.48394844e-02 -5.54040313e-01
-8.37197959e-01 7.68702149e-01 -8.33425857e-03 -2.92366952e-01
-3.16205770e-01 -1.78984746e-01 -1.14714372e+00 2.86071688e-01
-6.46049738e-01 4.24463183e-01 5.60710251e-01 1.02261615e+00
1.57528907e-01 1.13661659e+00 9.85585570e-01 -8.71284485e-01
-5.66512644e-01 -1.09192920e+00 -6.51628375e-01 1.26008585e-01
8.22758019e-01 -3.54763389e-01 -7.75620461e-01 3.56514826e-02] | [15.015043258666992, 5.902990341186523] |
991ec2ed-6b07-410d-a681-6696a8fec149 | refusion-enabling-large-size-realistic-image | 2304.08291 | null | https://arxiv.org/abs/2304.08291v1 | https://arxiv.org/pdf/2304.08291v1.pdf | Refusion: Enabling Large-Size Realistic Image Restoration with Latent-Space Diffusion Models | This work aims to improve the applicability of diffusion models in realistic image restoration. Specifically, we enhance the diffusion model in several aspects such as network architecture, noise level, denoising steps, training image size, and optimizer/scheduler. We show that tuning these hyperparameters allows us to achieve better performance on both distortion and perceptual scores. We also propose a U-Net based latent diffusion model which performs diffusion in a low-resolution latent space while preserving high-resolution information from the original input for the decoding process. Compared to the previous latent-diffusion model which trains a VAE-GAN to compress the image, our proposed U-Net compression strategy is significantly more stable and can recover highly accurate images without relying on adversarial optimization. Importantly, these modifications allow us to apply diffusion models to various image restoration tasks, including real-world shadow removal, HR non-homogeneous dehazing, stereo super-resolution, and bokeh effect transformation. By simply replacing the datasets and slightly changing the noise network, our model, named Refusion, is able to deal with large-size images (e.g., 6000 x 4000 x 3 in HR dehazing) and produces good results on all the above restoration problems. Our Refusion achieves the best perceptual performance in the NTIRE 2023 Image Shadow Removal Challenge and wins 2nd place overall. | ['Thomas B. Schön', 'Jens Sjölund', 'Zheng Zhao', 'Fredrik K. Gustafsson', 'Ziwei Luo'] | 2023-04-17 | null | null | null | null | ['shadow-removal', 'bokeh-effect-rendering', 'image-dehazing', 'image-shadow-removal', 'stereo-image-super-resolution'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 5.73614001e-01 -1.47553474e-01 -6.54841540e-03 -1.64386034e-01
-9.18657780e-01 -1.94368616e-01 4.30706620e-01 -6.44375861e-01
-2.86914110e-01 6.37089133e-01 6.39412642e-01 -2.80717343e-01
3.48259369e-03 -7.89561450e-01 -7.45205164e-01 -1.06350422e+00
7.79349655e-02 -1.46666663e-02 1.77571699e-01 -3.80608439e-01
4.23305333e-02 5.04492581e-01 -1.27486086e+00 1.97389215e-01
1.04190004e+00 8.71613979e-01 4.17488217e-01 9.79382813e-01
5.02135932e-01 1.16793668e+00 -6.97624385e-01 -2.97999293e-01
5.76587558e-01 -5.54343224e-01 -5.53949118e-01 1.60197482e-01
7.24530220e-01 -9.35079336e-01 -8.24943781e-01 1.05785370e+00
9.97501731e-01 2.47362360e-01 3.95887733e-01 -8.68566632e-01
-1.17329562e+00 4.56021845e-01 -5.66944778e-01 3.80833507e-01
-6.78062662e-02 4.47065800e-01 5.82127154e-01 -6.51792109e-01
7.31211603e-01 1.44933116e+00 7.41609812e-01 6.76862001e-01
-1.32454348e+00 -6.18982434e-01 -2.33762026e-01 3.91844511e-01
-1.23028159e+00 -8.53703201e-01 6.07323706e-01 1.11092322e-01
9.50192750e-01 4.22285527e-01 1.73085436e-01 1.27718627e+00
4.62874532e-01 6.92691445e-01 1.36530507e+00 -2.53238410e-01
2.22459212e-01 -4.12702441e-01 -3.58606875e-01 3.09809923e-01
-2.44031906e-01 2.91690618e-01 -4.42867488e-01 5.21491021e-02
1.13033009e+00 -1.25017017e-01 -6.69637442e-01 6.16552345e-02
-1.10340571e+00 6.82239652e-01 5.55450559e-01 3.15700434e-02
-4.36272949e-01 5.10545850e-01 1.47317812e-01 3.85590464e-01
7.90031612e-01 2.21472427e-01 -1.24996088e-01 7.64918923e-02
-1.24387920e+00 1.07553892e-01 2.88875759e-01 7.50895441e-01
2.89589971e-01 6.02001548e-01 -4.97571498e-01 1.08420575e+00
-5.27103851e-03 7.33284831e-01 4.57022160e-01 -1.60579824e+00
3.46486270e-01 -3.58234018e-01 -7.29474332e-03 -8.67973030e-01
6.79425448e-02 -5.92308581e-01 -1.49208403e+00 4.97902870e-01
-8.40895995e-02 -1.68038338e-01 -1.52237487e+00 1.68089294e+00
-7.05898851e-02 4.64100271e-01 3.14519644e-01 1.22443056e+00
6.80795610e-01 8.90709221e-01 -3.21220845e-01 -3.89506310e-01
8.98338854e-01 -1.29066086e+00 -1.01874018e+00 -2.79929280e-01
7.28126168e-02 -9.59814548e-01 1.01870596e+00 5.14121830e-01
-1.47376335e+00 -6.16197109e-01 -1.22053623e+00 -4.97711122e-01
2.65551597e-01 -1.82423428e-01 5.38935184e-01 6.98941350e-01
-1.67273760e+00 7.90907860e-01 -7.87384808e-01 -1.24412596e-01
5.84900081e-01 2.68051982e-01 -1.31828442e-01 -7.00016558e-01
-1.03676164e+00 7.02753127e-01 -3.59822139e-02 1.29127160e-01
-1.36062193e+00 -5.17090499e-01 -7.34475613e-01 1.18492983e-01
3.70884746e-01 -9.07810092e-01 8.58930230e-01 -8.28489304e-01
-1.64899290e+00 3.89383942e-01 -6.87850639e-02 -8.87148798e-01
6.58505142e-01 -1.73488975e-01 -4.93032336e-01 2.11691380e-01
-1.21853478e-01 7.36403763e-01 1.36096478e+00 -1.34533668e+00
-2.88127065e-01 -1.22296922e-01 -1.87854432e-02 2.57221907e-01
-1.54052854e-01 -7.27170259e-02 -7.72245407e-01 -1.17833984e+00
8.71609673e-02 -8.10476005e-01 -4.31425422e-01 1.84027795e-02
-3.07093710e-01 6.52159750e-01 1.09668517e+00 -1.05126500e+00
8.14082444e-01 -2.34076715e+00 3.38318318e-01 8.76026112e-04
5.89029789e-01 3.01034838e-01 -5.58937430e-01 9.87592898e-03
-7.98999891e-02 8.71006548e-02 -5.48995256e-01 -7.74347603e-01
-2.67866582e-01 5.83442450e-01 -5.62594891e-01 5.68272352e-01
-2.60448843e-01 1.06572342e+00 -4.83524442e-01 -2.89307952e-01
3.22563469e-01 9.88617599e-01 -6.42412364e-01 2.34522328e-01
1.12514056e-01 5.25466263e-01 2.04801172e-01 5.42314053e-01
9.82464314e-01 -1.47244141e-01 -6.17425144e-02 -2.42126912e-01
1.21795960e-01 -1.66701168e-01 -9.50974405e-01 1.69095564e+00
-7.44504452e-01 1.00039852e+00 5.45070589e-01 -5.89267492e-01
7.05275357e-01 2.15573251e-01 3.63212138e-01 -1.11849034e+00
-1.04133733e-01 5.82819544e-02 -2.89255917e-01 -2.13029176e-01
7.48041689e-01 -1.00679107e-01 3.69380116e-01 3.79947722e-01
1.17984682e-03 -1.92639962e-01 -1.16322070e-01 3.01296860e-01
1.20361018e+00 -2.32718334e-01 -2.21542060e-01 -7.08838413e-03
1.61533013e-01 -5.10800600e-01 5.46629488e-01 9.05468225e-01
-5.85669503e-02 1.15936434e+00 2.42034093e-01 -1.76345915e-01
-1.37471700e+00 -1.16927636e+00 3.73775773e-02 8.96352887e-01
2.78696239e-01 -3.35061878e-01 -8.27979147e-01 -3.68535668e-01
-3.36351961e-01 7.98781991e-01 -5.32208264e-01 -2.71284401e-01
-7.49118030e-01 -1.03948998e+00 6.42124474e-01 3.68052632e-01
9.47019041e-01 -9.71068203e-01 -1.47061586e-01 9.51820463e-02
-5.50990522e-01 -1.21101618e+00 -7.28872061e-01 2.38803700e-01
-8.56002688e-01 -5.55802345e-01 -1.00611401e+00 -6.39529169e-01
5.18096149e-01 5.31189919e-01 1.07743585e+00 1.27023190e-01
-1.72307000e-01 8.08405429e-02 -3.39522302e-01 7.88230076e-02
-6.05888486e-01 -2.60649294e-01 -8.21443871e-02 -2.28493642e-02
-4.78333354e-01 -9.99410093e-01 -9.05442894e-01 3.33775252e-01
-1.38825703e+00 9.33123082e-02 7.23306179e-01 9.63530123e-01
6.99614584e-01 6.06832445e-01 2.35484377e-01 -8.35706592e-01
7.38396347e-01 -1.78025663e-01 -4.90308374e-01 4.71643768e-02
-9.68634009e-01 8.23014528e-02 6.71978891e-01 -4.71547514e-01
-1.17904782e+00 -3.23923737e-01 -4.67384577e-01 -7.57698178e-01
1.12081461e-01 9.63435844e-02 -1.97535247e-01 -3.07578743e-01
7.26232052e-01 4.75939035e-01 1.07515650e-02 -5.74127316e-01
5.56110620e-01 4.28382456e-01 9.47442234e-01 -2.05973610e-01
1.07146287e+00 7.00413167e-01 -4.80297729e-02 -6.45932615e-01
-6.20653629e-01 -3.09019797e-02 -3.03909838e-01 9.50918812e-03
8.01199436e-01 -1.06256843e+00 -3.72203052e-01 8.56079578e-01
-9.30384755e-01 -8.60265493e-01 -5.81112802e-01 1.29508138e-01
-6.40552163e-01 6.55440271e-01 -1.12813139e+00 -3.98108125e-01
-5.95721483e-01 -1.40461373e+00 1.00829375e+00 -3.56259830e-02
5.06406546e-01 -7.85089791e-01 -2.34927043e-01 4.85221922e-01
1.08503199e+00 7.14170337e-02 8.76797974e-01 3.04988652e-01
-8.91866624e-01 2.69700259e-01 -4.02517527e-01 9.45330203e-01
8.70476216e-02 -4.79890645e-01 -9.23928440e-01 -5.95939636e-01
4.00761425e-01 -1.87136233e-01 1.45910215e+00 8.96065235e-01
1.34520435e+00 -4.29195881e-01 1.43720224e-01 1.44463527e+00
1.45244253e+00 -1.28664583e-01 1.36535358e+00 2.48701155e-01
7.83661544e-01 9.69877690e-02 3.35871011e-01 2.83381611e-01
1.67236999e-01 7.20171452e-01 6.50301158e-01 -7.67447829e-01
-8.23311090e-01 6.03123829e-02 5.27141035e-01 6.76212668e-01
-3.20332140e-01 -6.25823557e-01 -5.29203176e-01 3.33068430e-01
-1.58434927e+00 -9.89557743e-01 1.81603312e-01 1.96887970e+00
8.26481104e-01 7.16001019e-02 -4.11277562e-01 7.29436129e-02
5.14342964e-01 6.93027616e-01 -7.53691375e-01 -1.25825271e-01
-7.48842776e-01 3.36800903e-01 1.01547313e+00 8.57419133e-01
-8.54782224e-01 1.15372646e+00 6.90314484e+00 1.19284558e+00
-1.22038245e+00 5.57908714e-01 1.01590371e+00 -3.05507839e-01
-2.67519087e-01 -1.84186876e-01 -3.33688051e-01 3.26690078e-01
9.15706098e-01 1.09849758e-01 1.18577063e+00 3.99263144e-01
3.84413481e-01 -3.47372666e-02 -5.98034561e-01 1.11685383e+00
3.71759295e-01 -1.50210142e+00 1.31130725e-01 2.65944809e-01
9.83285546e-01 2.45745540e-01 5.78116894e-01 1.34600163e-01
5.13083398e-01 -1.35054731e+00 5.93352795e-01 3.78020912e-01
1.20858407e+00 -8.11110020e-01 7.99419940e-01 2.10933432e-01
-7.06105292e-01 -1.51130930e-01 -5.21133661e-01 3.96278322e-01
5.19498646e-01 6.89129353e-01 -3.74612510e-01 4.53060329e-01
8.71262014e-01 6.08009756e-01 -4.64058727e-01 7.49807715e-01
-3.11619073e-01 7.21644461e-01 -8.34666118e-02 1.00618887e+00
1.60472319e-02 -1.66370556e-01 7.21708000e-01 1.06236207e+00
3.56317848e-01 2.00832337e-01 -1.49103835e-01 6.95918381e-01
-3.28238696e-01 -3.37032139e-01 -2.48195484e-01 4.67547148e-01
7.50511792e-03 9.55466092e-01 -5.08646011e-01 -3.13779980e-01
6.98089227e-03 1.65427423e+00 -1.12332478e-01 8.41144025e-01
-9.73040760e-01 1.34491129e-02 7.60889709e-01 1.30898446e-01
4.75847930e-01 -3.84643495e-01 -1.33242756e-01 -1.21087909e+00
-9.12445337e-02 -1.22973311e+00 -1.47572234e-02 -1.10202181e+00
-1.07367396e+00 9.74016845e-01 -2.25021690e-01 -9.44106162e-01
-1.30618244e-01 -1.96161240e-01 -3.11001539e-01 1.01278543e+00
-1.94731247e+00 -1.14799845e+00 -5.42657018e-01 9.04026031e-01
6.79699421e-01 -1.64945751e-01 4.71064955e-01 5.89824677e-01
-4.63215649e-01 6.53279662e-01 4.30563331e-01 -9.70326439e-02
1.01077986e+00 -1.16271222e+00 7.20997155e-01 1.32832944e+00
-9.03512314e-02 3.54345649e-01 8.80673468e-01 -7.23988771e-01
-1.32482171e+00 -1.27383196e+00 2.37197340e-01 -4.82029207e-02
2.08258897e-01 -2.90876597e-01 -8.18348825e-01 5.93074560e-01
4.23859745e-01 2.34616488e-01 1.56297147e-01 -3.16348940e-01
-2.75120318e-01 -1.34859368e-01 -1.26033390e+00 7.09504724e-01
1.20877361e+00 -6.35323107e-01 1.60333633e-01 5.12155354e-01
1.00525212e+00 -8.06806564e-01 -8.17157805e-01 3.59493226e-01
1.03006743e-01 -1.08467472e+00 1.42562675e+00 1.15288302e-01
7.70018816e-01 -1.24737322e-01 -4.35839236e-01 -1.34861040e+00
-5.21584988e-01 -9.48513389e-01 -3.01664293e-01 1.08216095e+00
6.47887215e-02 -6.07812107e-01 7.36308873e-01 2.94477552e-01
-2.71988988e-01 -4.94105667e-01 -9.15793121e-01 -7.04453409e-01
-1.42730936e-01 -3.44617724e-01 6.30856633e-01 8.56363654e-01
-1.10675097e+00 1.92377657e-01 -1.12411189e+00 3.78197700e-01
9.36764359e-01 8.70531052e-03 6.71482265e-01 -5.11082947e-01
-6.02703869e-01 -1.88365415e-01 -3.54930498e-02 -1.57870090e+00
-4.67712805e-02 -6.99891388e-01 4.03641611e-02 -1.59677875e+00
9.50635448e-02 -4.06966835e-01 -2.80661523e-01 3.78924876e-01
-6.67278767e-02 7.46181250e-01 3.40900600e-01 5.52570224e-01
-3.16034704e-01 8.02081823e-01 1.64935052e+00 -2.65805364e-01
-2.29358152e-01 -1.50416970e-01 -6.72970593e-01 4.01192337e-01
6.00674510e-01 -4.00194407e-01 -4.46184248e-01 -9.35532749e-01
-1.84238181e-01 3.23709428e-01 5.44506609e-01 -1.03602874e+00
3.14490557e-01 -1.02485098e-01 4.46382046e-01 -2.76088446e-01
7.15551794e-01 -8.10000360e-01 2.42914319e-01 2.22559959e-01
-3.31031680e-01 4.75568660e-02 7.96346068e-02 6.12213433e-01
-2.92772025e-01 1.58783019e-01 1.00583255e+00 -6.45985315e-03
-6.61787391e-01 3.91105115e-01 -2.08289191e-01 -1.76163673e-01
6.28459156e-01 -3.11968565e-01 -7.85756946e-01 -8.82621050e-01
-7.13028967e-01 -1.17269449e-01 7.51268506e-01 3.23019326e-01
8.35907638e-01 -1.14160562e+00 -1.03161919e+00 3.02665651e-01
-4.50952351e-01 -8.64203647e-02 6.64286017e-01 7.77306318e-01
-7.40714788e-01 -4.95613515e-02 -3.84815603e-01 -3.92397940e-01
-1.36107206e+00 4.66496527e-01 4.26934212e-01 -3.41012806e-01
-1.09545529e+00 8.34855437e-01 2.82316297e-01 -8.35592449e-02
2.93836206e-01 -1.33459657e-01 1.26274638e-02 -4.06239539e-01
5.58958769e-01 3.03081959e-01 2.92104166e-02 -6.78634942e-01
1.84258342e-01 3.63010794e-01 -1.24029577e-01 -2.73561746e-01
1.61711109e+00 -4.90425825e-01 -2.17014179e-01 -2.95412749e-01
1.09347868e+00 3.44770581e-01 -1.65433908e+00 -3.71798515e-01
-7.89659321e-01 -7.68485904e-01 7.88704038e-01 -9.08158481e-01
-1.76631439e+00 6.43242121e-01 1.07620823e+00 -1.27651721e-01
1.75142252e+00 -2.27966040e-01 1.30750918e+00 -1.13611437e-01
1.90811485e-01 -8.30703199e-01 1.31362945e-01 2.88006574e-01
1.11314297e+00 -9.99663174e-01 2.75804162e-01 -4.25246477e-01
-6.23756051e-01 7.01088250e-01 3.59936953e-01 -9.73393675e-03
2.79544413e-01 6.35944843e-01 1.33419618e-01 2.79886331e-02
-7.03801692e-01 4.77263778e-02 1.40312344e-01 7.20622897e-01
-8.03910494e-02 -8.94096717e-02 1.43163934e-01 -4.58275579e-04
-1.90349966e-01 -1.23906724e-01 7.77034223e-01 5.89803159e-01
-1.81725889e-01 -1.03206873e+00 -5.17425418e-01 3.24234605e-01
-4.96676326e-01 -5.53956211e-01 4.72701564e-02 3.59266013e-01
9.00308937e-02 9.96323287e-01 -1.55319288e-01 -4.55460757e-01
1.39713943e-01 -7.09495783e-01 3.89771223e-01 -1.88556671e-01
-4.68216449e-01 3.34924191e-01 -1.23164892e-01 -8.69489610e-01
-4.19419080e-01 -3.01154822e-01 -7.65938461e-01 -8.77961338e-01
-1.93374351e-01 -2.51764446e-01 6.00936890e-01 7.09109426e-01
4.59530264e-01 8.16027105e-01 5.27854919e-01 -1.03403509e+00
-4.10466522e-01 -8.41801882e-01 -6.21498466e-01 2.93061852e-01
7.24502385e-01 -1.52774215e-01 -4.83189076e-01 1.84326902e-01] | [11.27173900604248, -2.1838276386260986] |
d9d4d5b2-a315-4e42-af96-c8bfccc9f0c0 | data-driven-summarization-of-scientific | 1804.08875 | null | http://arxiv.org/abs/1804.08875v1 | http://arxiv.org/pdf/1804.08875v1.pdf | Data-driven Summarization of Scientific Articles | Data-driven approaches to sequence-to-sequence modelling have been
successfully applied to short text summarization of news articles. Such models
are typically trained on input-summary pairs consisting of only a single or a
few sentences, partially due to limited availability of multi-sentence training
data. Here, we propose to use scientific articles as a new milestone for text
summarization: large-scale training data come almost for free with two types of
high-quality summaries at different levels - the title and the abstract. We
generate two novel multi-sentence summarization datasets from scientific
articles and test the suitability of a wide range of existing extractive and
abstractive neural network-based summarization approaches. Our analysis
demonstrates that scientific papers are suitable for data-driven text
summarization. Our results could serve as valuable benchmarks for scaling
sequence-to-sequence models to very long sequences. | ['Michael Pfeiffer', 'Richard H. R. Hahnloser', 'Nikola I. Nikolov'] | 2018-04-24 | null | null | null | null | ['abstractive-sentence-summarization'] | ['natural-language-processing'] | [ 8.34260046e-01 3.64646077e-01 -4.73360151e-01 -3.23868871e-01
-1.23147249e+00 -5.47306478e-01 8.04409385e-01 8.65007937e-01
-4.42794412e-01 1.23293066e+00 1.02258503e+00 -3.71064693e-01
3.41433026e-02 -4.77910250e-01 -7.73339212e-01 -1.03670128e-01
2.33013451e-01 6.01079047e-01 3.16707231e-02 -3.60506415e-01
8.29492927e-01 1.41169533e-01 -1.39928603e+00 7.78649330e-01
1.17324018e+00 1.95434988e-01 4.52706128e-01 1.33545852e+00
-4.91910934e-01 6.24350607e-01 -1.18416584e+00 -1.84424102e-01
-2.29395807e-01 -8.49344790e-01 -8.47939730e-01 -8.39264914e-02
7.75577724e-01 -2.47695595e-01 -2.88577080e-01 7.16045320e-01
8.25153530e-01 1.48866951e-01 8.17142248e-01 -6.15043998e-01
-4.91973758e-01 1.02151394e+00 -4.04065311e-01 4.53579128e-01
6.43409967e-01 -1.39058139e-02 1.22961533e+00 -4.48295325e-01
1.05351710e+00 9.76916373e-01 6.16839349e-01 7.80275166e-01
-1.20435894e+00 1.08739704e-01 2.16098526e-03 -1.22485526e-01
-3.71570796e-01 -8.50681245e-01 5.68462849e-01 -2.17645079e-01
1.47687006e+00 6.94556832e-01 6.74824893e-01 1.27514637e+00
5.75590193e-01 1.02729511e+00 5.47802329e-01 -5.37689626e-01
3.24022323e-01 -2.16863289e-01 4.64410514e-01 9.39570963e-02
7.95235574e-01 -6.05585694e-01 -6.55222952e-01 -4.01846975e-01
1.14222556e-01 -3.16950500e-01 -2.35668957e-01 3.08986396e-01
-1.24541283e+00 8.42585325e-01 -3.16675574e-01 2.83988506e-01
-5.62759697e-01 -1.24553941e-01 1.05673850e+00 3.99904937e-01
7.54859030e-01 1.11147273e+00 -5.82175910e-01 -4.71298516e-01
-1.67805099e+00 6.88965976e-01 1.31948328e+00 9.56919670e-01
9.55733955e-02 3.82384568e-01 -6.23865068e-01 9.30203617e-01
-3.08713526e-01 4.19045568e-01 8.78056586e-01 -8.29662561e-01
8.47617209e-01 2.65305012e-01 2.33373210e-01 -8.67554426e-01
-4.36108798e-01 -3.11926097e-01 -8.91580045e-01 -4.65349972e-01
8.11278820e-04 -4.46954072e-01 -7.05124915e-01 1.28430498e+00
-2.57367224e-01 -3.97242814e-01 4.50565696e-01 2.35105768e-01
1.37025356e+00 1.05195224e+00 -5.93124405e-02 -6.38927221e-01
1.13940609e+00 -9.29853082e-01 -7.29019165e-01 -3.17849457e-01
8.37001562e-01 -7.73274660e-01 6.58163965e-01 2.44773567e-01
-1.67072153e+00 -4.18712348e-01 -1.01984394e+00 -2.34299619e-02
-1.79586247e-01 1.81655377e-01 2.20175177e-01 2.69188851e-01
-1.08468938e+00 1.07967961e+00 -6.19946241e-01 -6.23198569e-01
5.18219054e-01 -1.25732152e-02 -2.05913678e-01 2.10524499e-01
-9.99579370e-01 1.04609406e+00 7.78040707e-01 -2.90590614e-01
-4.56049234e-01 -6.46387994e-01 -7.81281531e-01 2.99232632e-01
4.78545055e-02 -1.08334529e+00 1.58313596e+00 -6.64165258e-01
-1.45306826e+00 6.32172227e-01 -3.47917110e-01 -8.08629632e-01
3.48094553e-01 -2.72435874e-01 -1.95869148e-01 3.45459431e-01
1.87689722e-01 4.83683676e-01 5.36309719e-01 -1.06904936e+00
-6.58351541e-01 -1.02385134e-01 -4.43610638e-01 2.37245306e-01
-4.06647503e-01 1.91164538e-01 6.04773127e-02 -7.91635096e-01
-5.13745725e-01 -4.79213446e-01 -3.92858952e-01 -8.63979578e-01
-9.64871347e-01 -1.62540644e-01 3.35692614e-01 -1.05942905e+00
1.52935207e+00 -1.28845477e+00 2.68643588e-01 -4.84680474e-01
-1.94687163e-03 5.62511563e-01 -6.03933811e-01 1.09584689e+00
-7.35789761e-02 3.81139308e-01 -2.93620229e-01 -5.20110965e-01
-1.26806945e-01 -1.84248403e-01 -5.83198965e-01 8.89563747e-03
3.10657442e-01 1.07408881e+00 -1.00179720e+00 -6.16891623e-01
-2.60307133e-01 -2.18048826e-01 -2.25664616e-01 1.93463072e-01
-5.98408818e-01 1.81636184e-01 -6.08970404e-01 1.79225028e-01
2.84737796e-01 5.64090349e-03 -4.63929176e-02 1.95853189e-01
-2.72912323e-01 7.51164019e-01 -4.25830483e-01 1.92778790e+00
-2.03911394e-01 9.77689385e-01 -2.10887045e-01 -1.21614373e+00
7.89129019e-01 3.15719813e-01 3.20819259e-01 -5.06851435e-01
-1.49037186e-02 2.15392441e-01 4.75378521e-02 -6.92948580e-01
1.28345692e+00 -9.65151712e-02 -2.47135535e-01 8.61618280e-01
1.20817401e-01 -4.59452808e-01 1.01276922e+00 5.79744101e-01
1.30638933e+00 7.78615922e-02 5.43106019e-01 -2.44032681e-01
2.21657038e-01 5.08805096e-01 2.89318413e-01 1.30689323e+00
3.24980378e-01 1.00649869e+00 6.59016550e-01 -1.48586929e-01
-1.65714991e+00 -5.04776478e-01 1.75275102e-01 9.57312942e-01
-6.08152211e-01 -6.84572160e-01 -9.57877398e-01 -3.64777863e-01
-2.31740430e-01 1.35123706e+00 -3.59820068e-01 -2.72033811e-01
-7.16841161e-01 -7.85460472e-01 7.98845828e-01 3.34971756e-01
-4.40603085e-02 -1.30524182e+00 -5.34377277e-01 5.21581709e-01
-2.89905459e-01 -8.14597845e-01 -4.77395147e-01 2.36832902e-01
-1.28995812e+00 -5.14090300e-01 -1.03713179e+00 -7.01948822e-01
4.46241498e-01 2.28299513e-01 1.37904382e+00 -1.50881708e-01
-3.92814614e-02 2.07557335e-01 -3.73946369e-01 -8.80882978e-01
-1.32373953e+00 7.72387922e-01 1.89365614e-02 -6.44689381e-01
-2.43482962e-02 -4.95155066e-01 -2.43991897e-01 -5.40656149e-01
-1.06204903e+00 2.44986579e-01 8.96912038e-01 7.38784730e-01
1.25276119e-01 -5.57513237e-01 1.24491167e+00 -1.01459956e+00
1.46145666e+00 -4.81660575e-01 1.62755772e-02 3.39371890e-01
-3.97846252e-01 1.57326236e-01 9.70619559e-01 -4.30295855e-01
-1.10374641e+00 -5.44032395e-01 -2.76262790e-01 3.04604053e-01
-3.17254096e-01 1.13196611e+00 2.15854317e-01 7.24791169e-01
9.08705473e-01 7.62321234e-01 2.40140721e-01 -4.47185516e-01
4.61017489e-01 9.10311282e-01 5.26345015e-01 -3.49791110e-01
4.67889756e-01 5.54478168e-02 -1.82842731e-01 -1.27501059e+00
-9.53313053e-01 -4.75460708e-01 -5.54160058e-01 1.27878532e-01
4.61809516e-01 -6.96078718e-01 -2.26194471e-01 1.48969859e-01
-1.50254285e+00 -1.89121380e-01 -7.05494165e-01 1.69119030e-01
-7.00375617e-01 7.52889037e-01 -7.45884120e-01 -4.98444051e-01
-1.15254629e+00 -4.72580045e-01 1.09802377e+00 4.21070963e-01
-9.32038248e-01 -1.00391281e+00 4.16715324e-01 2.05535159e-01
3.89720559e-01 2.76026517e-01 9.79124606e-01 -1.42494798e+00
2.58610528e-02 -5.20473897e-01 1.62735522e-01 3.73311132e-01
7.73980170e-02 3.02355587e-01 -4.68730837e-01 -2.74739832e-01
-3.42871584e-02 -4.71469581e-01 1.32761121e+00 7.44063199e-01
8.54031563e-01 -7.09749937e-01 -2.51216859e-01 4.62251902e-02
9.37882662e-01 -1.14785498e-02 5.48601508e-01 5.54323018e-01
5.27547896e-01 7.20123351e-01 4.19730544e-01 4.75371182e-01
1.69191897e-01 1.66416228e-01 -2.76897401e-01 1.71404973e-01
-4.60337959e-02 -2.95692861e-01 2.90293396e-01 1.42769039e+00
1.65428206e-01 -7.97005832e-01 -5.89290619e-01 7.28505194e-01
-1.99959385e+00 -1.47037363e+00 -2.76833832e-01 1.96767664e+00
1.22151613e+00 2.62764484e-01 2.68911004e-01 -1.64082274e-01
5.32731712e-01 5.09790480e-01 -4.48974103e-01 -9.68007982e-01
-4.00717348e-01 -5.32185659e-02 3.38169545e-01 2.20694721e-01
-8.29828978e-01 6.28299236e-01 7.01598406e+00 1.02866244e+00
-8.76010180e-01 -3.18246573e-01 5.91494381e-01 -5.03640115e-01
-5.51119030e-01 -1.34641500e-02 -8.71466577e-01 5.45270205e-01
1.59794927e+00 -7.35150754e-01 -1.84908047e-01 6.07549131e-01
6.84318721e-01 -2.99751788e-01 -1.15976179e+00 4.52843577e-01
3.25730354e-01 -2.10055780e+00 6.09706461e-01 -2.31850445e-01
1.08192432e+00 1.74461082e-02 -5.18075824e-01 2.53751457e-01
2.44228542e-01 -1.00021946e+00 4.40196514e-01 7.41603851e-01
6.13602817e-01 -5.56137025e-01 7.26693809e-01 6.44631565e-01
-3.89152348e-01 2.11155325e-01 -6.70416772e-01 -1.31741509e-01
4.28595752e-01 6.44592464e-01 -8.67307842e-01 8.21972907e-01
1.76488180e-02 9.28490281e-01 -5.87441444e-01 1.00083578e+00
1.03256553e-01 7.98168719e-01 -4.51922528e-02 -6.21785760e-01
2.23845020e-01 -9.08896923e-02 1.07237613e+00 1.82392776e+00
3.37201387e-01 -1.10048339e-01 -3.12390048e-02 5.54278314e-01
-3.42075884e-01 2.67482311e-01 -6.57729089e-01 -5.21288097e-01
2.86328137e-01 1.09883916e+00 -5.97480237e-01 -8.29756498e-01
-2.17347369e-01 8.72537613e-01 2.55790323e-01 2.79798031e-01
-1.70187369e-01 -7.51135051e-01 3.62289399e-02 -1.31128371e-01
2.79341191e-01 -5.40191196e-02 -7.00918317e-01 -1.28794587e+00
5.87109011e-03 -1.10097647e+00 1.60867497e-01 -7.80756712e-01
-1.00072706e+00 4.81429309e-01 1.32450834e-01 -1.02921319e+00
-5.65651000e-01 -1.71739087e-02 -1.19959342e+00 8.04153502e-01
-1.07975209e+00 -8.79618585e-01 2.36876622e-01 -4.12704200e-01
1.13003552e+00 -5.41108489e-01 8.50763917e-01 -3.67443264e-01
-6.13633335e-01 2.01592460e-01 9.02454734e-01 -1.28048554e-01
6.96032763e-01 -1.25687969e+00 1.07624745e+00 7.39126742e-01
-1.24672502e-02 7.24798203e-01 1.37739515e+00 -9.60625708e-01
-1.33448970e+00 -1.10020268e+00 1.29105675e+00 -5.85226417e-01
4.94315356e-01 -1.39132321e-01 -9.98880744e-01 4.56795782e-01
7.64762700e-01 -1.08420181e+00 7.32892454e-01 7.46806562e-02
3.49783510e-01 2.18781725e-01 -7.09875882e-01 8.20401430e-01
6.48101091e-01 -1.33817747e-01 -1.13440859e+00 5.86582124e-01
8.31393480e-01 -2.87879646e-01 -8.31492960e-01 4.92239930e-02
4.02737021e-01 -6.69215024e-01 6.84554517e-01 -1.01735198e+00
1.32552803e+00 2.39092916e-01 4.28310692e-01 -1.80472386e+00
-1.41775012e-01 -8.04952264e-01 -2.73707032e-01 1.33108699e+00
6.18323565e-01 -3.12204868e-01 7.97404289e-01 2.06905618e-01
-4.78417665e-01 -6.79951906e-01 -6.16383672e-01 -6.36319518e-01
3.75313550e-01 6.83818236e-02 2.00191751e-01 5.94110370e-01
4.44464207e-01 9.03873622e-01 -3.37643862e-01 -7.80515313e-01
3.19221199e-01 2.40479648e-01 8.72286558e-01 -1.31788099e+00
-5.41869085e-03 -9.18545306e-01 1.29188180e-01 -1.05143499e+00
2.30596632e-01 -9.38326120e-01 2.43854746e-01 -2.33249974e+00
6.80485845e-01 4.00358796e-01 2.64253736e-01 3.67172211e-02
-5.21267116e-01 -3.86793911e-01 4.12155725e-02 2.96200067e-01
-7.65558958e-01 6.24069035e-01 1.12671971e+00 -2.68209815e-01
-3.83791149e-01 1.20448500e-01 -1.15688109e+00 4.73224908e-01
9.64938223e-01 -5.96744895e-01 -2.69583821e-01 -1.65491477e-01
2.16461137e-01 3.51997197e-01 -2.52508551e-01 -7.20729947e-01
3.26559335e-01 -3.20320398e-01 3.14757586e-01 -9.15179312e-01
-1.03607424e-01 1.43352762e-01 -1.86633542e-01 4.33525771e-01
-1.17121792e+00 5.12542389e-02 4.30595040e-01 6.74124002e-01
-1.41115651e-01 -7.13698447e-01 4.04473126e-01 -4.42710966e-01
-1.71705872e-01 -1.35216370e-01 -8.14613342e-01 3.63479793e-01
4.52973276e-01 -3.79920840e-01 -5.93535662e-01 -6.24125361e-01
-2.74684459e-01 2.80606389e-01 4.24187750e-01 3.89010698e-01
5.82674265e-01 -7.42325068e-01 -1.45396101e+00 -3.53792131e-01
-7.29041323e-02 3.83886625e-03 1.45423964e-01 5.15160978e-01
-5.49098969e-01 9.80199873e-01 -3.08814466e-01 -2.79198408e-01
-1.42662239e+00 2.59509057e-01 -3.00641149e-01 -5.97647667e-01
-7.52759576e-01 4.69101995e-01 -3.48898768e-01 -3.50778699e-01
-6.00575767e-02 -2.83645451e-01 -4.08439994e-01 3.95040065e-01
7.49513030e-01 4.43488389e-01 7.44504705e-02 -4.13310498e-01
1.57489061e-01 -2.31477804e-02 -4.88725662e-01 -2.28491291e-01
1.69579589e+00 -5.72853796e-02 -1.88354537e-01 6.75600588e-01
1.09016681e+00 -6.64995536e-02 -7.16141880e-01 -3.95857543e-02
2.76548684e-01 1.34162664e-01 -2.94880480e-01 -6.72017813e-01
-1.98640525e-01 6.46690190e-01 -4.22427148e-01 4.91233826e-01
6.02075160e-01 -1.78500324e-01 9.83088136e-01 8.93035114e-01
-3.05885732e-01 -1.39084470e+00 7.53694475e-02 7.01985896e-01
1.09642136e+00 -1.05368149e+00 6.01680040e-01 3.34310502e-01
-7.06402361e-01 1.38128352e+00 1.02051422e-01 -1.40023157e-01
-2.23356873e-01 -5.81396334e-02 -2.52266020e-01 -5.76152913e-02
-1.17086554e+00 3.69363129e-01 3.88670832e-01 4.86143410e-01
6.69678330e-01 -2.02552021e-01 -6.85517251e-01 4.07010168e-01
-4.32770550e-01 -5.79521209e-02 1.32979572e+00 1.20395088e+00
-9.19181466e-01 -9.19531584e-01 -2.07615197e-01 1.27231550e+00
-6.76685870e-01 -3.76003921e-01 -7.22190678e-01 4.24138367e-01
-8.85885596e-01 1.12227750e+00 -4.59742434e-02 7.10444897e-02
4.04767245e-01 2.64964283e-01 2.66870379e-01 -9.59820807e-01
-8.13501358e-01 -5.40449731e-02 7.04436660e-01 1.68402016e-01
-3.31533045e-01 -9.60159898e-01 -9.32035506e-01 -4.09415364e-01
-2.06303254e-01 4.16629672e-01 7.73315609e-01 9.82345164e-01
7.27269709e-01 8.00252736e-01 3.52407187e-01 -1.16898274e+00
-9.00164008e-01 -1.50463247e+00 -2.38747537e-01 3.13694596e-01
4.98981684e-01 3.76735866e-01 -1.15284257e-01 3.40780467e-01] | [12.501297950744629, 9.522834777832031] |
b437fcbc-fe96-4fc8-b0b6-e192c0afbc17 | decomposed-prototype-learning-for-few-shot | 2303.10863 | null | https://arxiv.org/abs/2303.10863v1 | https://arxiv.org/pdf/2303.10863v1.pdf | Decomposed Prototype Learning for Few-Shot Scene Graph Generation | Today's scene graph generation (SGG) models typically require abundant manual annotations to learn new predicate types. Thus, it is difficult to apply them to real-world applications with a long-tailed distribution of predicates. In this paper, we focus on a new promising task of SGG: few-shot SGG (FSSGG). FSSGG encourages models to be able to quickly transfer previous knowledge and recognize novel predicates well with only a few examples. Although many advanced approaches have achieved great success on few-shot learning (FSL) tasks, straightforwardly extending them into FSSGG is not applicable due to two intrinsic characteristics of predicate concepts: 1) Each predicate category commonly has multiple semantic meanings under different contexts. 2) The visual appearance of relation triplets with the same predicate differs greatly under different subject-object pairs. Both issues make it hard to model conventional latent representations for predicate categories with state-of-the-art FSL methods. To this end, we propose a novel Decomposed Prototype Learning (DPL). Specifically, we first construct a decomposable prototype space to capture intrinsic visual patterns of subjects and objects for predicates, and enhance their feature representations with these decomposed prototypes. Then, we devise an intelligent metric learner to assign adaptive weights to each support sample by considering the relevance of their subject-object pairs. We further re-split the VG dataset and compare DPL with various FSL methods to benchmark this task. Extensive results show that DPL achieves excellent performance in both base and novel categories. | ['Jun Xiao', 'Yi Yang', 'Yinfu Feng', 'Guikun Chen', 'Long Chen', 'Xingchen Li'] | 2023-03-20 | null | null | null | null | ['scene-graph-generation'] | ['computer-vision'] | [ 3.89929622e-01 8.48413482e-02 -3.44502240e-01 -4.88105386e-01
-4.06368852e-01 -4.92480248e-01 7.30657458e-01 2.88933426e-01
1.67714387e-01 5.97240031e-01 1.64875403e-01 -1.26320183e-01
-1.89329997e-01 -9.25669551e-01 -6.24593616e-01 -5.42576313e-01
2.56688390e-02 5.72324097e-01 5.94507694e-01 -1.74015462e-01
-1.21674918e-01 2.69831330e-01 -1.79833531e+00 4.41113144e-01
9.98708606e-01 9.44267094e-01 3.47927332e-01 1.01865262e-01
-4.84915227e-01 5.71946502e-01 -5.76931000e-01 -4.09926653e-01
2.69739360e-01 -4.74823892e-01 -6.91003740e-01 3.13031465e-01
4.92494524e-01 2.64747024e-01 -2.36854702e-01 1.17872894e+00
2.77180493e-01 5.24584174e-01 8.69158328e-01 -1.61527371e+00
-1.10947859e+00 5.29191434e-01 -5.14114261e-01 1.78547859e-01
5.41790307e-01 1.44706890e-01 1.37667489e+00 -1.04405892e+00
7.31120944e-01 1.42585623e+00 3.10068786e-01 5.66475868e-01
-1.23351181e+00 -4.85275805e-01 5.95187545e-01 7.67237246e-01
-1.23619270e+00 -1.39860541e-01 1.04068720e+00 -4.26191479e-01
7.17026353e-01 1.27902105e-01 6.69122994e-01 1.14386988e+00
-2.89001673e-01 1.03202522e+00 1.00315356e+00 -3.96447629e-01
3.50847453e-01 8.82579386e-02 2.49093324e-01 5.10925174e-01
4.82323766e-01 -2.00028449e-01 -5.01854122e-01 4.47552558e-03
2.92978227e-01 1.11384682e-01 -4.43306506e-01 -8.10756087e-01
-1.10813355e+00 7.17835546e-01 5.69287539e-01 2.82050312e-01
-1.64789796e-01 -1.91851243e-01 3.00067455e-01 2.38923579e-01
3.03877264e-01 2.45489687e-01 -4.04368639e-01 1.69446915e-01
-4.46794957e-01 2.50004590e-01 5.47406733e-01 1.36375618e+00
1.07765448e+00 -2.17547029e-01 -5.37175059e-01 9.70513284e-01
2.07398340e-01 2.21596524e-01 4.77686375e-01 -2.59754390e-01
4.41840500e-01 1.01857197e+00 -1.15607455e-01 -1.19530296e+00
-1.91920161e-01 -2.08579972e-01 -7.74111688e-01 -1.44660637e-01
1.93751067e-01 1.89852208e-01 -9.98376906e-01 1.70357025e+00
5.18102229e-01 2.74593145e-01 1.06378309e-01 6.26755059e-01
1.11115181e+00 8.30564618e-01 2.82427758e-01 -4.35033739e-01
1.17745876e+00 -1.00731468e+00 -4.98217255e-01 -4.28827345e-01
4.53417271e-01 -4.07786071e-01 1.61995697e+00 5.04583046e-02
-4.12015289e-01 -7.34475374e-01 -9.46697116e-01 3.25086415e-02
-5.82524538e-01 -7.07206130e-02 9.22587395e-01 4.21274722e-01
-4.93502349e-01 5.02387702e-01 -6.97799802e-01 -4.85352010e-01
7.19525337e-01 -1.37423519e-02 -1.71492726e-01 -4.37313557e-01
-1.11994803e+00 5.25884867e-01 8.24053764e-01 -2.65762568e-01
-7.22860396e-01 -7.69805610e-01 -1.13852322e+00 2.66997546e-01
8.83686364e-01 -5.66692472e-01 9.86830413e-01 -7.82496154e-01
-1.08410430e+00 8.34272921e-01 -2.42061615e-01 -7.24335238e-02
2.42023706e-01 -4.14897092e-02 -7.13348627e-01 -6.82382584e-02
3.86420459e-01 4.51849818e-01 8.76118124e-01 -1.35013914e+00
-7.33374357e-01 -2.57500738e-01 1.33211970e-01 2.85794228e-01
-3.39070082e-01 -1.53108656e-01 -4.43206728e-01 -7.42388844e-01
3.28882009e-01 -5.98357975e-01 2.28752084e-02 6.89531385e-04
-5.62399745e-01 -7.35708833e-01 9.46780086e-01 6.03013821e-02
1.09783602e+00 -2.30319047e+00 -2.86931563e-02 -5.53899929e-02
1.59299582e-01 3.39170963e-01 -1.63223326e-01 3.39056551e-01
-1.27026930e-01 -2.84027696e-01 -1.74257666e-01 -4.48290743e-02
2.03501388e-01 4.54086006e-01 -5.79486728e-01 3.44782285e-02
1.83212385e-01 1.08331513e+00 -1.26215601e+00 -5.44300079e-01
9.82429534e-02 -6.33574137e-03 -4.25567031e-01 3.12995285e-01
-5.37504613e-01 1.19410150e-01 -4.86489862e-01 9.62352931e-01
5.87577343e-01 -6.13393724e-01 2.97658980e-01 -4.87176269e-01
2.87442416e-01 3.15332897e-02 -1.24480033e+00 1.50639665e+00
-8.74087214e-02 2.95548081e-01 -7.26731598e-01 -1.18879640e+00
1.10731542e+00 5.48340678e-02 2.89864630e-01 -3.77304167e-01
-7.96373039e-02 3.93456854e-02 2.07551173e-03 -5.90629578e-01
1.92242727e-01 -3.85365993e-01 -1.22950040e-01 1.35751382e-01
3.06765735e-01 -6.20270260e-02 4.65670645e-01 2.98282474e-01
1.02516747e+00 1.33100405e-01 7.02348173e-01 1.48929758e-02
4.34195101e-01 2.04331591e-03 1.09388554e+00 7.52692163e-01
-3.53102237e-01 5.38417697e-01 5.78365624e-01 -4.11905527e-01
-5.21101296e-01 -1.35905206e+00 9.33945253e-02 1.26142859e+00
6.15037799e-01 -6.44231677e-01 -1.06763721e-01 -1.06206763e+00
1.61377698e-01 9.79197383e-01 -5.41756749e-01 -4.06072468e-01
-2.03499004e-01 -7.42341578e-01 1.74409207e-02 7.09851563e-01
4.72427845e-01 -1.36228240e+00 -4.31371748e-01 2.24117890e-01
2.02846769e-02 -1.04285824e+00 -2.39443168e-01 -3.87416072e-02
-5.24904907e-01 -1.30960333e+00 -4.43459541e-01 -1.03963983e+00
6.72218740e-01 3.71993095e-01 1.15125144e+00 -2.68671781e-01
-3.38066459e-01 2.35616148e-01 -7.36402273e-01 -4.40037459e-01
-6.00766167e-02 -3.16194952e-01 7.19115287e-02 2.41347477e-01
7.38782465e-01 -6.56731129e-01 -2.62827396e-01 1.52525246e-01
-6.71183765e-01 1.21913873e-01 5.42814136e-01 8.70544851e-01
8.95848572e-01 2.42174461e-01 5.66073298e-01 -1.24040425e+00
5.27815700e-01 -5.08279443e-01 -5.21718860e-01 7.66405821e-01
-5.53649783e-01 -1.29637823e-01 7.51940787e-01 -6.52249217e-01
-1.09426582e+00 -1.23950258e-01 4.73000258e-01 -7.64319599e-01
-2.92280197e-01 5.23023307e-01 -7.49171615e-01 2.79686987e-01
5.31931877e-01 2.68851906e-01 -4.73035187e-01 -4.05185074e-01
6.25429273e-01 3.28311712e-01 5.84891856e-01 -7.84736216e-01
9.57574427e-01 3.40034276e-01 -3.42397317e-02 -7.38065898e-01
-1.29686368e+00 -4.73083109e-01 -5.38585365e-01 -6.22354867e-03
6.27506852e-01 -6.52967691e-01 -4.55275863e-01 3.04705322e-01
-8.00534964e-01 -1.48057938e-01 -6.86802685e-01 3.07319343e-01
-5.84218264e-01 4.78495479e-01 -1.37065440e-01 -5.00155509e-01
9.44014522e-04 -7.65017331e-01 8.91099215e-01 3.63408804e-01
-3.05251647e-02 -8.90937388e-01 3.66284586e-02 9.44578573e-02
1.05848901e-01 3.24874282e-01 1.28745210e+00 -8.13202202e-01
-6.09296918e-01 6.79318756e-02 -4.62892085e-01 2.18043789e-01
5.61796129e-01 -1.29571021e-01 -8.43371272e-01 -3.06827724e-01
-9.81769413e-02 -4.72844690e-01 8.88755620e-01 1.38531655e-01
1.36000788e+00 -2.19556540e-01 -6.12174869e-01 7.09847450e-01
1.33440244e+00 3.33448678e-01 3.43536019e-01 3.59439664e-02
9.56413090e-01 4.71461862e-01 1.05520701e+00 4.23179388e-01
4.52596426e-01 5.93454242e-01 1.15680732e-01 2.03679085e-01
-3.03032368e-01 -6.67429030e-01 1.27946466e-01 5.96891165e-01
5.54640368e-02 -3.11212748e-01 -9.29495871e-01 6.81879282e-01
-2.04992509e+00 -9.06363368e-01 2.22449943e-01 1.94942498e+00
7.53771782e-01 2.39602596e-01 -1.70761079e-01 -7.23047974e-03
7.74657249e-01 3.19329739e-01 -6.14391804e-01 2.21844599e-01
-2.20035434e-01 1.64374217e-01 -1.27108470e-01 -3.49630229e-02
-1.29623961e+00 1.27457798e+00 5.14434576e+00 9.48421180e-01
-6.77273631e-01 1.03777526e-02 3.81487519e-01 1.69422075e-01
-4.68579024e-01 3.94300252e-01 -9.28543210e-01 4.58270729e-01
2.31203839e-01 -6.22506797e-01 1.85267553e-01 1.07387650e+00
-3.37658167e-01 1.04188286e-01 -1.31907737e+00 1.29517734e+00
3.84616286e-01 -1.20780182e+00 4.21793044e-01 -3.42120051e-01
8.43996167e-01 -3.56332481e-01 -8.34064260e-02 7.42784083e-01
2.74608672e-01 -7.25857198e-01 3.43399107e-01 3.91177863e-01
8.96876812e-01 -4.92210686e-01 3.36020827e-01 2.51362473e-01
-1.50890303e+00 -2.25582853e-01 -8.25742483e-01 -1.36860702e-02
1.68936923e-01 5.31334043e-01 -9.45287049e-01 6.49298728e-01
6.41821444e-01 1.06981742e+00 -8.08815122e-01 1.11591101e+00
-6.77067220e-01 5.45362413e-01 -1.50859967e-01 -1.53614223e-01
3.97240706e-02 -2.54195724e-02 5.44628680e-01 7.59516358e-01
2.82467335e-01 2.39991918e-01 6.18234456e-01 9.43393230e-01
-3.13749202e-02 1.98464736e-01 -6.67279243e-01 -1.80648565e-01
6.78901672e-01 1.14853692e+00 -8.88562322e-01 -5.93079627e-01
-7.74840713e-01 8.04565310e-01 4.86637324e-01 4.40369785e-01
-6.31750226e-01 -4.15252775e-01 6.49438381e-01 -3.91316488e-02
4.32230681e-01 1.47060707e-01 -2.05441602e-02 -1.46307898e+00
8.68078992e-02 -7.09711909e-01 9.31176007e-01 -7.91850567e-01
-1.81047630e+00 3.94464254e-01 3.07419956e-01 -1.48775363e+00
-8.79005566e-02 -5.53122580e-01 -6.84859157e-01 4.91089076e-01
-1.41422451e+00 -1.34806263e+00 -4.62596506e-01 8.39672565e-01
6.23416603e-01 -3.64610612e-01 7.92338669e-01 -1.65122263e-02
-3.71814638e-01 6.26971245e-01 -2.00361460e-01 9.75261163e-03
6.77727342e-01 -1.25018895e+00 4.35582429e-01 9.50065792e-01
4.95124221e-01 6.18657351e-01 5.90335786e-01 -8.86184216e-01
-1.08916748e+00 -1.29508305e+00 7.99602449e-01 -3.67287546e-01
6.40469313e-01 -4.15125519e-01 -1.10826361e+00 9.21501756e-01
-3.03881019e-01 4.83713925e-01 7.51859725e-01 4.86125469e-01
-5.82243919e-01 -2.14578480e-01 -9.09192383e-01 6.20533645e-01
1.44171822e+00 -3.66463065e-01 -1.12686288e+00 5.65691113e-01
9.13115859e-01 -4.97719720e-02 -4.14920479e-01 7.55123734e-01
-2.74800155e-02 -7.90766060e-01 8.15048099e-01 -8.45595896e-01
1.69740573e-01 -6.19134784e-01 -4.01147246e-01 -1.42599022e+00
-6.13561749e-01 -2.77018726e-01 -4.34711784e-01 1.32201326e+00
1.77338228e-01 -5.71758926e-01 7.83389270e-01 4.11254406e-01
-3.28521341e-01 -7.66290188e-01 -6.43645823e-01 -1.07660031e+00
-4.53494966e-01 -3.32013428e-01 7.86725581e-01 1.13191223e+00
1.33835822e-01 5.23033738e-01 -2.02132314e-01 1.50526762e-01
6.65344834e-01 8.93113732e-01 8.50113809e-01 -1.37674236e+00
-5.04561067e-01 -1.76861450e-01 -8.60365272e-01 -8.33881676e-01
1.59114808e-01 -1.05528414e+00 1.44344633e-02 -1.67154241e+00
4.59397644e-01 -5.08399069e-01 -4.25123841e-01 9.29231524e-01
-6.68535352e-01 -5.57550825e-02 2.85952359e-01 1.29743695e-01
-8.44442248e-01 8.84830058e-01 1.29861283e+00 -2.98890829e-01
-2.35027313e-01 -7.60187358e-02 -7.97234833e-01 7.82281637e-01
4.90664154e-01 -2.26422340e-01 -9.93019283e-01 -3.80448215e-02
1.25475213e-01 -3.57522279e-01 2.49007344e-01 -1.06053138e+00
8.69775936e-02 -5.09701192e-01 3.99434149e-01 -6.41456008e-01
2.52130181e-01 -6.12437010e-01 2.56659091e-02 1.07343653e-02
-7.77533352e-02 -4.11743373e-01 -1.13259926e-01 9.07402515e-01
-3.56799155e-01 -7.06898645e-02 6.10626876e-01 -2.06941754e-01
-1.44185722e+00 5.92586398e-01 2.90527850e-01 4.69101042e-01
1.25641441e+00 -3.47048044e-01 -4.98073488e-01 -6.18462935e-02
-7.61037052e-01 2.79602945e-01 5.22362411e-01 6.68249547e-01
7.95957029e-01 -1.43233407e+00 -4.22602713e-01 4.57744598e-01
7.76561558e-01 1.78650409e-01 3.60899985e-01 3.44882220e-01
-5.33769373e-03 2.12735347e-02 -1.32881030e-01 -5.63531637e-01
-1.20261574e+00 1.06988859e+00 -5.92874512e-02 -2.63971481e-02
-9.22025621e-01 1.11203706e+00 6.68852746e-01 -3.10353607e-01
1.88611940e-01 -2.48457879e-01 -3.64686757e-01 1.73173457e-01
5.24122000e-01 3.73014361e-02 -2.47687370e-01 -5.20239711e-01
-3.75336736e-01 4.73664761e-01 -2.76915759e-01 5.19926906e-01
1.31240785e+00 1.52835652e-01 1.14404835e-01 8.33381176e-01
9.50876653e-01 -2.25798145e-01 -1.21498859e+00 -5.93513370e-01
9.49112028e-02 -7.78067291e-01 -5.92001617e-01 -6.40371144e-01
-7.94219732e-01 7.50631154e-01 3.60365838e-01 1.47045165e-01
1.20425713e+00 4.78585184e-01 5.98786056e-01 4.86049712e-01
6.27742648e-01 -9.40027177e-01 2.86486864e-01 4.87303615e-01
6.54927611e-01 -1.22692740e+00 3.11197676e-02 -8.86430085e-01
-7.81674445e-01 8.04577351e-01 8.34052086e-01 4.42946926e-02
5.73211789e-01 -3.36472183e-01 -1.89042091e-01 -5.12345612e-01
-8.41288030e-01 -4.09137160e-01 5.75806320e-01 8.49086642e-01
1.56951681e-01 3.50667030e-01 -2.68605351e-01 6.82737768e-01
-2.13488609e-01 -1.69563100e-01 3.06248635e-01 9.86398280e-01
-5.14931083e-01 -1.09216928e+00 2.11378001e-02 6.47365212e-01
1.33616775e-01 5.41715696e-02 -2.80308843e-01 7.33659983e-01
3.29525292e-01 7.06738532e-01 -8.38592127e-02 -2.69720972e-01
4.51885462e-01 1.81798369e-01 4.87601638e-01 -1.25329340e+00
1.66877508e-01 -1.35122195e-01 -8.07594731e-02 -5.08067667e-01
-4.42426056e-01 -6.65326118e-01 -1.20419979e+00 3.00479978e-01
-2.49534652e-01 3.44836153e-02 -4.69351523e-02 9.61288154e-01
2.13725850e-01 4.77704674e-01 4.06133384e-01 -4.63994533e-01
-4.67037797e-01 -8.27894568e-01 -8.98146987e-01 9.31565285e-01
-1.98358774e-01 -1.27847564e+00 -2.97461390e-01 3.01747378e-02] | [10.182536125183105, 1.9371331930160522] |
93ff669c-99d8-44ae-8844-30c3682d6cf1 | verifiable-and-provably-secure-machine | 2210.09126 | null | https://arxiv.org/abs/2210.09126v2 | https://arxiv.org/pdf/2210.09126v2.pdf | Verifiable and Provably Secure Machine Unlearning | Machine unlearning aims to remove points from the training dataset of a machine learning model after training; for example when a user requests their data to be deleted. While many machine unlearning methods have been proposed, none of them enable users to audit the procedure. Furthermore, recent work shows a user is unable to verify if their data was unlearnt from an inspection of the model alone. Rather than reasoning about model parameters, we propose to view verifiable unlearning as a security problem. To this end, we present the first cryptographic definition of verifiable unlearning to formally capture the guarantees of a machine unlearning system. In this framework, the server first computes a proof that the model was trained on a dataset $D$. Given a user data point $d$ requested to be deleted, the server updates the model using an unlearning algorithm. It then provides a proof of the correct execution of unlearning and that $d \notin D'$, where $D'$ is the new training dataset. Our framework is generally applicable to different unlearning techniques that we abstract as admissible functions. We instantiate the framework, based on cryptographic assumptions, using SNARKs and hash chains. Finally, we implement the protocol for three different unlearning techniques (retraining-based, amnesiac, and optimization-based) to validate its feasibility for linear regression, logistic regression, and neural networks. | ['Nicolas Papernot', 'Olga Ohrimenko', 'Esha Ghosh', 'Varun Chandrasekaran', 'Doreen Riepel', 'Thorsten Eisenhofer'] | 2022-10-17 | null | null | null | null | ['snarks'] | ['natural-language-processing'] | [ 3.30212563e-01 2.95943528e-01 -4.29021269e-01 -1.90802023e-01
-4.71923411e-01 -1.23194575e+00 3.16631466e-01 3.51690471e-01
-6.07492864e-01 9.05538440e-01 -7.34569907e-01 -1.07425773e+00
9.96838138e-03 -1.16063166e+00 -1.40824485e+00 -7.22114146e-01
-1.40522972e-01 3.97092760e-01 1.62285671e-01 2.99387932e-01
3.18229347e-01 5.35178304e-01 -1.59108436e+00 2.49907643e-01
2.22408175e-01 7.76663840e-01 -5.91747522e-01 1.03878200e+00
3.46254319e-01 2.98608243e-01 -6.45963728e-01 -7.36530185e-01
5.66933751e-01 -1.48786873e-01 -1.00338864e+00 -5.59709609e-01
4.63922918e-01 -6.73378766e-01 6.53174892e-03 1.19208050e+00
1.55387282e-01 -4.87027824e-01 2.48641863e-01 -1.95858610e+00
-5.59716411e-02 1.06429338e+00 1.46305850e-02 -2.24660233e-01
2.84394830e-01 1.62065670e-01 1.07267654e+00 -2.85472453e-01
4.44970965e-01 6.55819774e-01 4.71099705e-01 7.19241500e-01
-1.09667826e+00 -1.01162219e+00 -2.56173551e-01 2.55070299e-01
-1.24312532e+00 -2.56330997e-01 3.41385692e-01 -1.32686332e-01
9.69509840e-01 8.02915871e-01 4.12842125e-01 8.56384575e-01
2.35378221e-01 4.54844058e-01 1.24478841e+00 -7.02181041e-01
3.81932259e-01 6.20659769e-01 5.29124737e-01 7.77610362e-01
1.09847975e+00 5.54312289e-01 -4.79571581e-01 -6.18406892e-01
4.48848248e-01 -1.18227638e-01 -1.61976039e-01 -6.07516646e-01
-8.95218313e-01 8.04362178e-01 -2.01212645e-01 1.66695006e-02
2.94922799e-01 4.81120765e-01 5.22338152e-01 8.09439838e-01
-1.00004628e-01 7.36488327e-02 -1.07033288e+00 -1.92826971e-01
-6.23775303e-01 1.03421405e-01 1.34517765e+00 7.93781698e-01
9.69036639e-01 6.37555914e-03 6.62912726e-01 -4.75144923e-01
3.04064244e-01 5.96122146e-01 4.86423105e-01 -6.40649438e-01
1.18910067e-01 4.42468047e-01 4.79153823e-03 -7.84761012e-01
1.79739371e-02 -1.78277865e-01 -4.83162969e-01 5.03201544e-01
4.12873000e-01 -2.08450660e-01 -4.21412379e-01 1.84343219e+00
3.71570796e-01 5.58184803e-01 3.20720702e-01 4.19789791e-01
2.75031328e-01 3.44527096e-01 -3.69040012e-01 -5.02136052e-01
9.65348959e-01 -5.40645361e-01 -1.71521425e-01 1.37291014e-01
1.01268899e+00 -2.45055646e-01 9.34546351e-01 9.68604147e-01
-1.05829060e+00 -1.78251877e-01 -1.37718892e+00 3.47180575e-01
-6.14179373e-01 -4.36757445e-01 6.03835821e-01 1.25455880e+00
-1.14772308e+00 4.73224401e-01 -7.44231880e-01 4.93912697e-02
1.13398090e-01 8.76905501e-01 -6.93084061e-01 2.13652760e-01
-1.09365833e+00 4.69012767e-01 8.52516174e-01 -3.07506979e-01
-1.20917308e+00 -2.75803298e-01 -8.94224763e-01 9.10699889e-02
3.41405779e-01 -4.87920493e-01 1.39989805e+00 -1.01422477e+00
-1.16391397e+00 9.35971320e-01 1.38189718e-01 -9.85423744e-01
6.36871040e-01 3.71467993e-02 -2.80949175e-01 7.79745653e-02
-6.19446814e-01 -7.39845261e-02 1.24190140e+00 -1.44467556e+00
-7.72344768e-01 -5.33769429e-01 4.01894659e-01 -4.86710191e-01
-5.10283768e-01 -9.08914432e-02 1.79959740e-02 -3.90428126e-01
-6.62134439e-02 -1.08768857e+00 -9.26836301e-03 -2.11687192e-01
-4.14853692e-01 3.34524214e-01 1.05497408e+00 -1.07571661e-01
1.30427265e+00 -2.09869480e+00 -4.09759581e-01 1.01193440e+00
2.09964946e-01 5.45337796e-01 -8.02137256e-02 2.89581954e-01
-2.18713030e-01 6.35602832e-01 -3.14501256e-01 -1.11374810e-01
2.40338594e-01 5.43537855e-01 -7.28227377e-01 7.17010736e-01
-3.80661160e-01 7.16855109e-01 -5.80485046e-01 -2.35447511e-01
-1.22826342e-02 2.95915365e-01 -9.09849346e-01 2.57686645e-01
-4.93870407e-01 -1.13388980e-02 -1.24282345e-01 4.88543481e-01
7.91198790e-01 -2.01750830e-01 7.58718193e-01 1.49284676e-01
2.45454654e-01 2.09912732e-01 -1.55892467e+00 1.09915876e+00
-3.28798831e-01 2.95925587e-01 -4.31481330e-03 -8.25076342e-01
4.86152083e-01 4.28899407e-01 8.78859684e-02 -1.95676014e-01
1.12897158e-01 3.13562125e-01 -1.85684323e-01 -2.98586786e-01
4.58688557e-01 -3.94744538e-02 -1.21581815e-01 1.28097165e+00
-1.18744578e-02 9.69367549e-02 -3.22488189e-01 9.52198803e-02
1.04695201e+00 7.62260333e-02 4.10058320e-01 2.17187151e-01
6.82167649e-01 -5.10305285e-01 2.33010039e-01 1.08385015e+00
4.83575910e-02 1.44554945e-02 5.64067304e-01 -3.67683500e-01
-7.51212358e-01 -9.46774006e-01 1.17274396e-01 1.06403196e+00
-1.42292008e-02 -7.92096555e-01 -9.30561185e-01 -1.52077806e+00
2.03421369e-01 6.58212423e-01 -5.61424792e-01 -3.50265950e-01
-5.32478631e-01 -1.16720259e-01 9.98923540e-01 2.35346958e-01
3.84711385e-01 -7.70486414e-01 -7.71039665e-01 -8.07011351e-02
1.17745906e-01 -7.21301556e-01 -2.30158195e-01 3.64986688e-01
-1.02484262e+00 -1.62440777e+00 4.23170626e-01 -6.09469831e-01
1.04507267e+00 -6.69978335e-02 6.92879379e-01 9.06236589e-01
7.27812573e-02 6.48604453e-01 -9.27604288e-02 -4.80915070e-01
-8.93049836e-01 1.91092461e-01 2.35316381e-01 -1.99397385e-01
4.29574221e-01 -4.96337354e-01 -3.31193209e-01 3.25923562e-01
-1.43484783e+00 -5.11215925e-01 3.31744820e-01 5.60957074e-01
3.54144484e-01 4.32828039e-01 1.30700797e-01 -1.44026268e+00
4.16199833e-01 -2.45315954e-01 -1.06593335e+00 4.17301595e-01
-1.31024146e+00 3.14907312e-01 9.06956911e-01 -6.54503584e-01
-1.96952879e-01 -3.40731554e-02 -1.80914614e-03 -2.17244059e-01
-2.06772581e-01 4.17739540e-01 -3.73337716e-01 -2.64101267e-01
6.20046556e-01 3.09363633e-01 1.77505150e-01 -2.14604780e-01
4.01550651e-01 6.88853085e-01 4.74198937e-01 -8.42517436e-01
1.29601550e+00 4.30717349e-01 7.88067728e-02 -1.28352880e-01
-2.68037498e-01 9.73219424e-02 -4.41515356e-01 -5.29426849e-03
-3.26302927e-03 -1.90364733e-01 -1.66615307e+00 5.83538115e-01
-9.57611382e-01 -6.78440571e-01 -2.94694304e-01 2.57380128e-01
-4.06527936e-01 6.17896676e-01 -2.86855400e-01 -9.94177222e-01
-5.35566568e-01 -1.18535757e+00 3.97753358e-01 -1.15341745e-01
-1.02565274e-01 -1.00718355e+00 -8.84721279e-02 2.82391999e-02
1.58253804e-01 1.06220290e-01 1.04553330e+00 -1.35345721e+00
-6.55175567e-01 -6.85647964e-01 1.35892972e-01 6.85770333e-01
-1.16462924e-01 2.41665751e-01 -1.17969739e+00 -8.03498089e-01
1.46208972e-01 -2.89282352e-01 2.80254215e-01 -3.00948679e-01
1.38412619e+00 -1.32236505e+00 -1.23551942e-01 5.46692550e-01
1.37011206e+00 -3.76083306e-03 6.16017818e-01 4.89847600e-01
1.09492935e-01 7.61154741e-02 2.88498789e-01 5.75413585e-01
1.59296319e-01 1.57945946e-01 7.79386461e-01 3.18724871e-01
7.76597857e-01 -5.15075982e-01 6.77133322e-01 3.28414798e-01
5.79816848e-02 1.31429866e-01 -5.01647413e-01 -6.36718124e-02
-1.52987444e+00 -7.12742925e-01 1.20183453e-01 2.75843024e+00
1.30785108e+00 6.30876958e-01 7.39998147e-02 6.39016807e-01
2.72287965e-01 -2.46371627e-01 -5.11539161e-01 -9.21871781e-01
2.34558702e-01 8.37765694e-01 9.50618625e-01 8.00364077e-01
-6.77519381e-01 8.09936881e-01 6.09709597e+00 3.54559481e-01
-1.38677526e+00 7.48506710e-02 2.02116176e-01 1.54129252e-01
-5.95008194e-01 5.95325708e-01 -8.64345491e-01 3.02250743e-01
1.27017748e+00 -3.85968208e-01 6.54502332e-01 1.11712897e+00
-4.03462231e-01 -6.25929832e-02 -1.58634102e+00 5.42251468e-01
2.66787559e-01 -1.05641365e+00 1.09454514e-02 1.00010559e-01
4.33026697e-04 -4.62073445e-01 2.66515613e-01 5.35746336e-01
5.16907513e-01 -1.00549710e+00 7.97446132e-01 2.35673010e-01
1.00639260e+00 -1.02401793e+00 6.28877521e-01 6.41226053e-01
-6.67781591e-01 -3.46455686e-02 -1.92787703e-02 -9.34875682e-02
-5.93687117e-01 1.12059861e-01 -1.32062387e+00 4.67734247e-01
3.92945260e-01 1.19526312e-01 -6.51665151e-01 6.93951130e-01
-6.31412983e-01 8.44822586e-01 -4.65463847e-01 -2.70957537e-02
-2.23461866e-01 -2.38182694e-02 3.18400681e-01 1.04954004e+00
2.94130355e-01 -1.18899949e-01 -2.27404252e-01 2.40972847e-01
-3.28500569e-01 -2.21534997e-01 -6.75078988e-01 -3.73527855e-02
4.86825019e-01 8.63195956e-01 -2.24474341e-01 -2.53192544e-01
-1.12959668e-01 8.13937247e-01 1.86218470e-02 2.12386340e-01
-8.43554676e-01 -2.41300941e-01 4.79711533e-01 9.41286311e-02
1.70457795e-01 -4.11601603e-01 -9.15306155e-03 -1.19565797e+00
-1.18064228e-02 -1.41400981e+00 9.30762708e-01 -5.71281493e-01
-7.41799593e-01 2.97662467e-01 -2.98119728e-02 -9.20597434e-01
-3.13744962e-01 -5.12229204e-01 -2.36249909e-01 4.00619715e-01
-1.57758772e+00 -1.11871493e+00 1.56656012e-01 8.71313393e-01
-3.68226379e-01 -5.86682111e-02 1.39616132e+00 -5.10642119e-02
-3.00808489e-01 1.23261595e+00 -3.48289728e-01 1.29528761e-01
8.09314013e-01 -1.10859156e+00 9.41538960e-02 1.13424075e+00
3.34805548e-01 1.17688775e+00 7.73003519e-01 -5.92563391e-01
-1.75237226e+00 -9.23421562e-01 8.98862481e-01 -5.13038158e-01
5.59114933e-01 -4.30723608e-01 -8.32056880e-01 1.41520011e+00
-8.94695446e-02 1.19473964e-01 9.37209606e-01 -2.01209545e-01
-9.73089576e-01 -3.56837064e-01 -1.45877469e+00 3.55435193e-01
3.34200770e-01 -6.68024063e-01 -3.68477911e-01 -4.68183495e-02
7.64490426e-01 -4.81207281e-01 -7.23760426e-01 3.90336215e-01
1.00094855e+00 -8.42198014e-01 6.53466284e-01 -1.09251571e+00
-2.76515633e-01 -3.41264307e-01 -2.52118647e-01 -6.55996501e-01
6.35990918e-01 -1.09647071e+00 -8.48191559e-01 1.06862557e+00
5.10230899e-01 -1.18753946e+00 9.94538724e-01 7.76331306e-01
5.71671665e-01 -2.60979325e-01 -1.10186315e+00 -6.09880626e-01
2.42545307e-01 -9.01236117e-01 1.16259575e+00 1.05637789e+00
2.59407073e-01 -1.65530175e-01 -2.96114892e-01 6.34400368e-01
6.67189837e-01 9.92237106e-02 1.25309134e+00 -1.06521547e+00
-6.30358875e-01 -1.03846282e-01 -2.95346498e-01 -6.54869854e-01
4.06772554e-01 -1.12118137e+00 -3.14801723e-01 -8.45418811e-01
1.35830849e-01 -7.01931059e-01 -4.34099227e-01 1.00926459e+00
3.59046966e-01 9.54596400e-02 -2.38661412e-02 8.88043568e-02
-3.16012233e-01 -3.35370362e-01 5.96811414e-01 -5.44724911e-02
-5.50729875e-03 5.26060700e-01 -6.98902249e-01 6.27334535e-01
9.77746904e-01 -7.30340362e-01 -3.12326163e-01 1.25352025e-01
7.88302898e-01 -4.99308519e-02 5.71042895e-01 -6.90621018e-01
4.04101223e-01 -3.12462509e-01 -1.62069321e-01 -2.36887589e-01
-2.25193217e-01 -1.41649723e+00 3.65448087e-01 9.28153276e-01
-3.75855416e-01 1.34853601e-01 4.84902784e-02 3.65219772e-01
2.65679300e-01 -4.17443633e-01 5.96227467e-01 2.50648379e-01
-3.05756569e-01 4.40482020e-01 -1.90479726e-01 -1.62486687e-01
1.19446123e+00 -1.34638548e-01 -5.31369925e-01 -1.79934666e-01
-5.96657336e-01 -8.65839645e-02 8.93180013e-01 -1.03414379e-01
8.42107236e-01 -6.38766944e-01 -2.73193359e-01 9.89248395e-01
5.72812259e-02 -1.32956551e-02 -3.53034616e-01 5.92332184e-01
-5.22901118e-01 8.70749652e-02 1.24154530e-01 -2.61327267e-01
-1.85081685e+00 8.94945443e-01 2.88329870e-01 -3.03082496e-01
-1.35830477e-01 3.37009907e-01 -5.89218497e-01 -5.65064549e-01
7.24845350e-01 -6.48723602e-01 3.77856493e-01 -2.73287147e-01
5.21501899e-01 1.34466067e-01 4.00845379e-01 -1.22838125e-01
-1.84152916e-01 6.84192106e-02 -9.05341506e-02 -2.84627646e-01
1.03720272e+00 1.53985992e-01 -3.88329983e-01 3.04547921e-02
1.20402575e+00 3.65359753e-01 -7.22229779e-01 -2.69248009e-01
-1.09771140e-01 -3.45711529e-01 -4.20594215e-01 -7.35328853e-01
-9.56541479e-01 8.49613667e-01 6.16924167e-01 4.15184200e-01
9.85852063e-01 -3.49635303e-01 8.01936209e-01 8.25980306e-01
8.19151342e-01 -8.03385615e-01 -9.40097719e-02 2.73646802e-01
2.32469738e-01 -8.24446321e-01 1.77658960e-01 3.49037386e-02
-1.42890468e-01 1.26056027e+00 4.60776180e-01 2.51269519e-01
8.99777889e-01 6.83561385e-01 -1.33118346e-01 4.02974486e-01
-7.47534096e-01 4.75799978e-01 -3.11859429e-01 6.21188343e-01
-4.23662364e-01 1.22129628e-02 -2.93306023e-01 9.18098867e-01
-4.19608295e-01 5.40572584e-01 8.29802096e-01 1.39913332e+00
-4.19867814e-01 -1.73427308e+00 -4.16326255e-01 2.14136884e-01
-7.24364042e-01 -1.95668429e-01 -3.11629832e-01 9.93497610e-01
-1.24635189e-04 6.25764787e-01 -2.85658717e-01 -5.31027615e-01
7.54171517e-04 1.89676806e-01 5.32137156e-01 -5.71559370e-01
-8.34466875e-01 -3.14563185e-01 -1.06094191e-02 -2.99287766e-01
-2.31541812e-01 -2.55784690e-01 -1.56765449e+00 -7.63087988e-01
-6.44265532e-01 4.09493208e-01 9.24659252e-01 6.69341087e-01
1.88781191e-02 -3.20985675e-01 1.00738466e+00 5.05400822e-02
-8.34623098e-01 -2.51870155e-01 -6.18545473e-01 5.34237400e-02
5.03039598e-01 2.18003020e-02 -9.11366224e-01 4.29995835e-01] | [5.912057399749756, 7.226564407348633] |
53fe9363-374e-4bcb-bada-08e0ee4331b0 | curriculum-learning-for-compositional-visual | 2303.15006 | null | https://arxiv.org/abs/2303.15006v1 | https://arxiv.org/pdf/2303.15006v1.pdf | Curriculum Learning for Compositional Visual Reasoning | Visual Question Answering (VQA) is a complex task requiring large datasets and expensive training. Neural Module Networks (NMN) first translate the question to a reasoning path, then follow that path to analyze the image and provide an answer. We propose an NMN method that relies on predefined cross-modal embeddings to ``warm start'' learning on the GQA dataset, then focus on Curriculum Learning (CL) as a way to improve training and make a better use of the data. Several difficulty criteria are employed for defining CL methods. We show that by an appropriate selection of the CL method the cost of training and the amount of training data can be greatly reduced, with a limited impact on the final VQA accuracy. Furthermore, we introduce intermediate losses during training and find that this allows to simplify the CL strategy. | ['Michel Crucianu', 'Marin Ferecatu', 'Wafa Aissa'] | 2023-03-27 | null | null | null | null | ['visual-reasoning', 'visual-reasoning'] | ['computer-vision', 'reasoning'] | [ 2.11235046e-01 2.22072318e-01 -5.20102158e-02 -3.54314983e-01
-7.94317067e-01 -7.45402217e-01 5.52808583e-01 2.51411140e-01
-7.65306294e-01 5.02756119e-01 -1.57441109e-01 -5.10191023e-01
-1.58757955e-01 -1.06710792e+00 -8.69950473e-01 -5.16019344e-01
4.40052181e-01 5.23036003e-01 3.57140392e-01 -1.59218401e-01
1.06273800e-01 5.55079818e-01 -1.40622365e+00 2.04007372e-01
8.98073792e-01 1.05529952e+00 3.61794889e-01 5.99844217e-01
-5.16465783e-01 1.09697080e+00 -5.42310894e-01 -7.27132082e-01
1.77228779e-01 -6.69021785e-01 -1.18457150e+00 -6.54004142e-02
3.63431573e-01 -2.09474102e-01 -1.26716062e-01 9.49092090e-01
4.04511869e-01 2.68191755e-01 7.84732342e-01 -1.30859518e+00
-6.02870941e-01 3.53136271e-01 -2.37635970e-01 4.78059016e-02
6.35532513e-02 2.43657768e-01 1.26372302e+00 -8.83198202e-01
8.26375842e-01 1.05638981e+00 4.39702600e-01 8.70038688e-01
-1.31378472e+00 -3.57519388e-01 1.57502174e-01 6.08572006e-01
-1.13633525e+00 -9.09951478e-02 9.70369279e-01 -3.19055438e-01
6.43578649e-01 1.55545413e-01 6.26876116e-01 9.54421103e-01
-1.03954613e-01 9.20900822e-01 8.67987037e-01 -6.24959528e-01
4.08326745e-01 2.82499939e-01 2.62951136e-01 9.23015833e-01
-3.19865681e-02 -2.11173832e-01 -1.45877153e-01 2.35126540e-01
5.05334496e-01 -2.42901474e-01 -3.03288966e-01 -8.34904134e-01
-8.27885807e-01 1.20658839e+00 8.46983552e-01 1.08964644e-01
-1.93782598e-01 3.10681850e-01 3.51915747e-01 5.84994376e-01
3.12924162e-02 7.09479630e-01 -2.82419056e-01 2.19464824e-01
-7.55948961e-01 2.71826595e-01 6.53721809e-01 5.89737773e-01
1.12799954e+00 -1.39374211e-01 -2.72872090e-01 7.37376094e-01
9.11090821e-02 4.05902922e-01 1.27319261e-01 -1.12750006e+00
5.74453413e-01 8.40589762e-01 -1.51627734e-02 -7.28315949e-01
-3.33489895e-01 -3.64557594e-01 -5.59291959e-01 6.16878748e-01
7.86879957e-01 -6.10271767e-02 -1.07221794e+00 1.86646080e+00
3.23422968e-01 -2.79512197e-01 1.32017717e-01 8.96882296e-01
8.73791456e-01 7.46325433e-01 3.03193659e-01 5.48182093e-02
1.35529482e+00 -1.03159261e+00 -5.95152497e-01 -2.38888338e-01
7.44319201e-01 -4.55981165e-01 1.50522482e+00 3.23719591e-01
-1.14903963e+00 -5.98139584e-01 -1.11454046e+00 -6.02441072e-01
-5.52343190e-01 2.50384241e-01 5.39372265e-01 4.90547836e-01
-1.10518467e+00 5.24543285e-01 -5.57392120e-01 -8.05271491e-02
7.76650369e-01 2.91840553e-01 -3.03008378e-01 -3.63135964e-01
-1.10569775e+00 1.10964382e+00 4.01566714e-01 1.03401475e-01
-9.98428345e-01 -8.62665176e-01 -9.11657870e-01 2.52052605e-01
5.13738155e-01 -9.90181983e-01 1.01030028e+00 -1.25030553e+00
-1.45011556e+00 8.60496998e-01 -5.75602129e-02 -4.15810823e-01
6.67026103e-01 -7.44524077e-02 1.39085054e-01 5.60694396e-01
-1.08518086e-01 1.04747915e+00 8.92898917e-01 -1.23295534e+00
-3.88988256e-01 -1.62884533e-01 6.15866423e-01 1.66090146e-01
-1.65899500e-01 -2.61429667e-01 -5.07275999e-01 -2.53130823e-01
-1.06090568e-01 -7.32968867e-01 -2.67914474e-01 2.79970139e-01
-3.00441217e-02 -4.66837138e-01 4.59721893e-01 -7.60179698e-01
8.47624362e-01 -1.94454038e+00 5.04921794e-01 2.88940579e-01
2.98046172e-01 3.43248159e-01 -3.51175249e-01 2.32016161e-01
1.46733448e-02 1.63981423e-03 -3.30805391e-01 -3.63005430e-01
1.23435045e-02 4.40027982e-01 -3.21666479e-01 2.65693396e-01
5.79671025e-01 1.18174326e+00 -7.99893141e-01 -5.35266697e-01
1.50564924e-01 2.97341317e-01 -7.22296119e-01 4.94149208e-01
-6.34332359e-01 3.60705316e-01 -2.26527065e-01 2.68012226e-01
4.42237347e-01 -3.68212193e-01 3.59576233e-02 -3.87047917e-01
1.95838004e-01 1.50323331e-01 -9.83438671e-01 1.61446798e+00
-6.21481955e-01 7.73533344e-01 -3.68539579e-02 -1.22802794e+00
8.18588436e-01 2.90435944e-02 2.04367917e-02 -9.35068727e-01
3.59357297e-01 8.81678537e-02 -5.24750836e-02 -6.89759433e-01
2.14223474e-01 -1.15266867e-01 1.70977473e-01 3.99575025e-01
3.01387221e-01 -4.01718430e-02 2.45797724e-01 2.62136132e-01
7.26825655e-01 1.98257148e-01 9.12578851e-02 -5.73545769e-02
7.33927190e-01 3.11755806e-01 2.90292799e-01 6.03954852e-01
-2.74077430e-02 4.50245321e-01 8.11665177e-01 -3.36690605e-01
-1.16524529e+00 -1.03881729e+00 2.40478352e-01 1.04746103e+00
7.12843537e-02 -7.92113468e-02 -8.14547598e-01 -1.00079322e+00
-2.03646973e-01 8.85373652e-01 -8.80350292e-01 -2.48340234e-01
-7.76578248e-01 -3.51104587e-01 2.20439225e-01 6.96207643e-01
2.90626287e-01 -1.31971681e+00 -6.70720100e-01 -1.87783763e-01
-7.92367682e-02 -1.02224660e+00 -1.72097012e-01 3.34641844e-01
-6.74417615e-01 -1.21598554e+00 -7.60793865e-01 -9.19190109e-01
9.68972981e-01 2.83348449e-02 1.21562397e+00 2.95146942e-01
-9.49289277e-02 3.98667753e-01 -3.43588650e-01 -1.81760013e-01
-4.81793016e-01 2.17782542e-01 -5.16860664e-01 -1.58857405e-02
1.06571242e-01 -4.23859566e-01 -5.20437777e-01 -1.23787917e-01
-8.95824730e-01 1.50102511e-01 5.34664035e-01 8.69337678e-01
6.13231778e-01 -2.65190482e-01 6.09387696e-01 -9.62406039e-01
6.09114707e-01 -3.00963402e-01 -8.23860168e-01 3.95458728e-01
-6.78892851e-01 4.06978428e-01 8.75544786e-01 -4.82990205e-01
-8.93122911e-01 -9.13981870e-02 -3.48544836e-01 -7.76762426e-01
1.30448863e-01 4.29978341e-01 -2.67639309e-01 -2.01050445e-01
6.22110188e-01 7.68747032e-02 8.81235451e-02 -3.17509055e-01
7.67420053e-01 -3.82191949e-02 4.70653117e-01 -4.95256543e-01
8.22991252e-01 4.48386580e-01 1.68498024e-01 -4.59028393e-01
-9.27877009e-01 -1.00216553e-01 -5.40463746e-01 -1.95439711e-01
1.20736313e+00 -5.09317815e-01 -9.32400227e-01 -1.09667487e-01
-1.04584408e+00 -5.54049730e-01 -5.31234443e-01 2.47464880e-01
-5.40359795e-01 2.29501426e-01 -4.68293697e-01 -6.84114516e-01
-3.70514303e-01 -1.11603165e+00 4.81063902e-01 3.20847750e-01
-4.54178415e-02 -1.08860636e+00 -1.39159206e-02 5.26919246e-01
3.41593266e-01 7.36271963e-02 1.40124416e+00 -5.50440609e-01
-8.29501867e-01 1.73208490e-01 -6.32898748e-01 6.02193058e-01
-1.78416103e-01 -2.04833001e-01 -9.49480891e-01 -2.91388392e-01
-9.67051387e-02 -5.99643588e-01 1.05525398e+00 2.13870749e-01
1.31269217e+00 -1.66411340e-01 -9.95886279e-04 6.92106843e-01
1.61813128e+00 2.42762975e-02 6.25299215e-01 3.82450879e-01
8.73556137e-01 7.90446341e-01 4.56478685e-01 -2.35477626e-01
4.94085371e-01 4.31837976e-01 6.41498864e-01 -3.00401896e-01
-3.31886023e-01 -3.18056673e-01 2.51509309e-01 4.96032536e-01
2.21524984e-01 -1.17637210e-01 -9.17487442e-01 8.19226563e-01
-1.71592510e+00 -6.84590459e-01 4.85852268e-03 2.02128720e+00
8.48609090e-01 6.32424504e-02 1.19472757e-01 2.80067027e-01
2.25385532e-01 9.21277329e-02 -4.28986430e-01 -5.79530597e-01
8.45911428e-02 5.14377713e-01 2.81486928e-01 7.57149696e-01
-9.32529330e-01 9.96390402e-01 6.11771011e+00 7.87865400e-01
-1.06948543e+00 1.26975283e-01 3.93964916e-01 -3.24406065e-02
-6.80276513e-01 -6.99324207e-03 -4.40359920e-01 1.09714523e-01
8.89767945e-01 2.05033079e-01 4.30902034e-01 7.52795219e-01
-1.88515380e-01 1.73213542e-03 -1.20228899e+00 9.45491493e-01
1.18164428e-01 -1.54313159e+00 3.77795249e-01 -3.18825215e-01
5.31349301e-01 -3.03335041e-01 -1.03753559e-01 5.26896477e-01
2.34592214e-01 -1.06751275e+00 6.21069491e-01 5.27426243e-01
5.46124637e-01 -1.04711163e+00 6.45398557e-01 1.68001801e-01
-9.18108284e-01 -1.98402405e-01 -4.48959619e-01 1.52080312e-01
1.49434909e-01 2.96138048e-01 -7.28487790e-01 6.25279069e-01
6.15612328e-01 1.29883543e-01 -8.69042337e-01 9.31743205e-01
-6.19992793e-01 5.26871860e-01 -8.44267011e-02 -2.19054028e-01
3.85278761e-01 -3.17387581e-01 2.87858725e-01 8.99563551e-01
-3.93735878e-02 -5.22364303e-02 5.46805421e-03 1.16311371e+00
-2.26075485e-01 2.07480699e-01 -2.90066421e-01 7.66486153e-02
7.55986497e-02 1.23247778e+00 -7.12446749e-01 -3.12019527e-01
-2.90646881e-01 1.08617353e+00 7.92301953e-01 5.02468169e-01
-7.42604017e-01 -4.49372530e-01 3.12832803e-01 -1.83223616e-02
5.60939968e-01 -4.82188985e-02 -2.29845718e-01 -9.79031026e-01
5.56209236e-02 -8.13176513e-01 5.65844595e-01 -9.34462011e-01
-1.09203362e+00 5.69046915e-01 -1.24439642e-01 -9.17447269e-01
-3.75031531e-01 -9.07106578e-01 -5.33066511e-01 9.36360002e-01
-1.93541455e+00 -1.21733987e+00 -2.97980934e-01 6.83605790e-01
2.66192019e-01 6.93278164e-02 6.11389518e-01 3.82658035e-01
-3.96856457e-01 7.24395275e-01 -2.67756820e-01 2.80628890e-01
4.29112345e-01 -1.39364719e+00 1.99534837e-02 7.29726911e-01
2.20388487e-01 4.74942267e-01 5.98784745e-01 -8.45036134e-02
-1.15456772e+00 -1.01046216e+00 9.68376338e-01 -6.33979499e-01
5.30500829e-01 -4.81232703e-01 -1.09319973e+00 6.01194680e-01
3.29267889e-01 -7.79135153e-02 5.87472737e-01 1.35661572e-01
-5.20071924e-01 -2.49902800e-01 -1.08375752e+00 7.11704493e-01
6.32862985e-01 -7.26399362e-01 -7.83364713e-01 7.37653226e-02
8.37472558e-01 -1.83012292e-01 -5.99775732e-01 2.51691639e-01
3.18098634e-01 -7.61650383e-01 9.16359067e-01 -8.66135120e-01
5.28093755e-01 -3.80060196e-01 1.27449915e-01 -1.22283018e+00
-2.49936983e-01 -1.04674339e-01 -5.72263375e-02 1.02948904e+00
5.28503537e-01 -3.86543721e-01 7.70318568e-01 3.72613013e-01
4.45534624e-02 -9.20867205e-01 -8.16720247e-01 -5.07730126e-01
3.97761613e-01 -3.69524896e-01 3.05694520e-01 8.38820398e-01
-3.42546791e-01 5.64767778e-01 -1.20606065e-01 1.42432362e-01
5.11256278e-01 4.20061231e-01 7.50005543e-01 -1.19458652e+00
-2.81852275e-01 -4.62363869e-01 -1.83387250e-01 -1.05266666e+00
2.64271408e-01 -9.39180672e-01 -1.10343866e-01 -1.74465668e+00
7.55852014e-02 -3.95880550e-01 -2.07792714e-01 5.69771290e-01
-3.13414067e-01 1.74856514e-01 3.77177745e-01 -9.75348800e-02
-5.56937039e-01 5.59608161e-01 1.41899288e+00 -2.79265940e-01
-3.27473693e-03 -7.53473863e-02 -5.30466497e-01 4.58503902e-01
7.54123092e-01 -3.36915314e-01 -7.48048067e-01 -5.99173427e-01
5.19961417e-01 -9.87743214e-03 5.81863880e-01 -7.29960263e-01
7.33151063e-02 -5.88865578e-02 4.46194649e-01 -4.42984849e-01
2.92971939e-01 -8.60296488e-01 -2.93703943e-01 4.05827433e-01
-5.12625039e-01 4.48427536e-02 2.98436314e-01 3.14293653e-01
-2.12595209e-01 -7.12446392e-01 9.23550129e-01 -4.83231358e-02
-6.63481295e-01 1.59588501e-01 -4.50110994e-03 1.44304456e-02
8.58260989e-01 -7.00107589e-02 -2.23705471e-01 -4.60740000e-01
-8.21151853e-01 6.10037506e-01 3.84263605e-01 3.41840774e-01
4.54796672e-01 -1.30946362e+00 -4.83332247e-01 -5.06825699e-03
9.05446056e-03 1.54316975e-02 2.97826320e-01 6.51072741e-01
-7.24276245e-01 4.42130476e-01 -2.62727201e-01 -4.17444885e-01
-1.12277877e+00 8.81146193e-01 5.06462991e-01 -4.34939742e-01
-6.13930464e-01 9.29124296e-01 1.79895937e-01 -4.99443740e-01
3.08605313e-01 -3.56055856e-01 -5.97895324e-01 2.55861074e-01
4.25627708e-01 1.14367746e-01 2.38803960e-02 -2.39639491e-01
-2.45213732e-01 5.14532268e-01 -1.28312744e-02 -1.81564376e-01
1.08821785e+00 -3.24433930e-02 9.35896933e-02 2.75045216e-01
1.25916588e+00 -1.52388260e-01 -1.24853218e+00 -9.14438441e-02
-7.39126354e-02 -1.86538175e-01 -3.11267748e-02 -6.97085440e-01
-1.24404585e+00 1.28637528e+00 6.98113859e-01 2.05918878e-01
1.22033107e+00 1.68735281e-01 5.47151923e-01 4.37100708e-01
5.04289977e-02 -1.02569973e+00 3.41560870e-01 3.61968637e-01
8.55095029e-01 -1.17997026e+00 -1.64856181e-01 -3.61911327e-01
-6.28800750e-01 1.03721714e+00 6.48669600e-01 -2.05644727e-01
2.92009145e-01 -3.16313177e-01 2.53093034e-01 -3.03979754e-01
-6.22867167e-01 -5.38628936e-01 3.31185579e-01 6.07660115e-01
-4.00381275e-02 -3.38229179e-01 -3.67454559e-01 4.95673954e-01
-1.39179751e-01 -9.31994244e-02 2.91048110e-01 5.54089904e-01
-3.58805954e-01 -1.22626352e+00 -4.31619808e-02 2.02753872e-01
-2.46856645e-01 -5.73003478e-02 -2.98093557e-01 8.91219079e-01
2.24406093e-01 6.73182130e-01 7.64487833e-02 -5.33235520e-02
4.15910661e-01 3.06052715e-01 8.12158704e-01 -3.85894179e-01
-5.39528906e-01 -4.27698374e-01 1.22537144e-01 -6.09245300e-01
-3.58663499e-01 -2.16298133e-01 -1.31399131e+00 -2.08810091e-01
-1.65649846e-01 1.57868147e-01 4.02128339e-01 1.12288046e+00
1.98419482e-01 6.89270735e-01 3.32661688e-01 -2.91960090e-01
-5.12722135e-01 -6.56171739e-01 -8.90396610e-02 5.85479021e-01
3.71144205e-01 -6.04461968e-01 -4.09270674e-01 4.65040617e-02] | [10.722506523132324, 1.9020401239395142] |
d1d9e289-ed80-4d09-a921-a47623bb08c4 | deep-reinforcement-learning-based-path | 2301.05980 | null | https://arxiv.org/abs/2301.05980v1 | https://arxiv.org/pdf/2301.05980v1.pdf | Deep-Reinforcement-Learning-based Path Planning for Industrial Robots using Distance Sensors as Observation | Industrial robots are widely used in various manufacturing environments due to their efficiency in doing repetitive tasks such as assembly or welding. A common problem for these applications is to reach a destination without colliding with obstacles or other robot arms. Commonly used sampling-based path planning approaches such as RRT require long computation times, especially in complex environments. Furthermore, the environment in which they are employed needs to be known beforehand. When utilizing the approaches in new environments, a tedious engineering effort in setting hyperparameters needs to be conducted, which is time- and cost-intensive. On the other hand, Deep Reinforcement Learning has shown remarkable results in dealing with unknown environments, generalizing new problem instances, and solving motion planning problems efficiently. On that account, this paper proposes a Deep-Reinforcement-Learning-based motion planner for robotic manipulators. We evaluated our model against state-of-the-art sampling-based planners in several experiments. The results show the superiority of our planner in terms of path length and execution time. | ['Jens Lambrecht', 'Benno Kutschank', 'Yifan Hu', 'Linh Kästner', 'Teham Bhuiyan'] | 2023-01-14 | null | null | null | null | ['industrial-robots', 'motion-planning'] | ['robots', 'robots'] | [ 1.23001739e-01 -1.86945394e-01 -4.65211645e-02 -1.42780006e-01
-3.79132032e-01 -3.49232763e-01 1.92640156e-01 2.32454211e-01
-5.67803442e-01 8.45216036e-01 -5.38111687e-01 -2.54299730e-01
-5.99269211e-01 -1.03393865e+00 -5.54246902e-01 -8.05186570e-01
-1.80677086e-01 9.21126902e-01 2.63249248e-01 -4.70274061e-01
5.36491156e-01 8.22756886e-01 -1.25500715e+00 -4.37242389e-01
9.30671930e-01 5.09636819e-01 1.05020511e+00 2.61781305e-01
-9.05681923e-02 2.17735648e-01 -3.92518431e-01 1.52496904e-01
3.59425962e-01 -2.02698201e-01 -8.77271175e-01 3.22359920e-01
-4.19329941e-01 -3.46367240e-01 -5.44031918e-01 8.70085180e-01
3.58970076e-01 6.95474446e-01 6.25497639e-01 -1.22284949e+00
-1.65454075e-02 4.21998262e-01 -6.51783347e-01 -2.77580470e-01
2.61420250e-01 4.57726181e-01 4.87971961e-01 -3.84924263e-01
6.60918593e-01 1.15432942e+00 2.70390242e-01 1.94491908e-01
-8.31662953e-01 -4.66983497e-01 8.02886114e-02 4.28830087e-01
-1.14158952e+00 4.81006764e-02 7.33343601e-01 -9.40707326e-02
9.55945015e-01 -2.75881022e-01 3.76564473e-01 9.60864246e-01
7.24331737e-01 5.74028075e-01 6.66265965e-01 -2.57681698e-01
5.13241947e-01 -4.27890301e-01 -4.00624484e-01 4.90105122e-01
3.27825814e-01 9.17790085e-02 2.92563558e-01 2.48331368e-01
1.17448068e+00 5.60138114e-02 3.72924330e-03 -6.14311099e-01
-1.28975141e+00 7.21372724e-01 5.24755359e-01 2.51847535e-01
-7.66274154e-01 3.07117254e-01 5.07497132e-01 2.04869822e-01
-2.88396835e-01 8.85045588e-01 -4.97836798e-01 -3.57033521e-01
-3.08649957e-01 5.13268650e-01 7.43154943e-01 1.35308647e+00
7.74798632e-01 -4.44299029e-03 3.17629099e-01 8.15983295e-01
3.69817801e-02 1.81357026e-01 1.82666793e-01 -1.17613864e+00
7.48358727e-01 3.47606093e-01 3.90840471e-01 -1.01780474e+00
-8.24745476e-01 -3.29791367e-01 -9.82946813e-01 4.70772445e-01
5.26637375e-01 -2.35194653e-01 -8.02314222e-01 1.34389603e+00
4.42774385e-01 -8.23885053e-02 7.11996704e-02 1.04662955e+00
-3.03633343e-02 8.02781045e-01 -1.17080927e-01 -2.19097286e-01
1.10792708e+00 -1.28304684e+00 -6.61276579e-01 -3.48602414e-01
6.09914482e-01 -7.93143690e-01 1.05313277e+00 7.03670919e-01
-7.93956876e-01 -7.22170651e-01 -1.03777838e+00 2.01822713e-01
-1.12931810e-01 8.86158720e-02 7.47606277e-01 2.47239470e-01
-5.59997618e-01 1.05560565e+00 -9.25917745e-01 -3.49836886e-01
2.34369069e-01 5.22989094e-01 -2.67237604e-01 -4.47647244e-01
-8.49693835e-01 1.07477677e+00 5.44955254e-01 3.96039575e-01
-1.07048297e+00 -6.94883764e-02 -7.53977597e-01 1.51567161e-01
9.44095254e-01 -5.01842380e-01 1.51076257e+00 -7.49825761e-02
-1.93609977e+00 -1.08532682e-01 9.92616266e-02 -1.15177639e-01
6.54631853e-01 -3.32641095e-01 -1.18925653e-01 -2.65657138e-02
1.13754310e-01 3.56528670e-01 5.91944277e-01 -1.08863831e+00
-6.96709096e-01 -1.67178184e-01 3.96970332e-01 3.67361903e-01
6.61829412e-02 -5.26763201e-01 -3.74480397e-01 -2.06916913e-01
3.32119465e-01 -1.31768560e+00 -9.87550735e-01 -1.83100507e-01
-3.89190227e-01 -5.10690987e-01 7.79694200e-01 -2.61792839e-01
6.96762621e-01 -1.87283564e+00 3.26252133e-01 2.16491923e-01
-2.38879487e-01 9.82375592e-02 -2.63590962e-01 8.07761133e-01
3.63258362e-01 -3.24534953e-01 3.16895321e-02 2.31628850e-01
3.27863954e-02 4.52769518e-01 2.62644384e-02 3.28031778e-01
-3.53321359e-02 4.70272928e-01 -1.25777292e+00 -4.05086249e-01
5.28608859e-01 9.28708166e-03 -4.35082823e-01 8.75057355e-02
-4.65597659e-01 7.97579885e-01 -8.88970852e-01 3.70438874e-01
5.05457222e-01 -5.92395058e-03 3.45620841e-01 1.18852876e-01
-3.84231508e-01 1.99970469e-01 -1.36387062e+00 2.24079323e+00
-9.96095836e-01 3.65305990e-01 7.62876123e-02 -1.04681337e+00
1.08914471e+00 1.14286110e-01 5.23486674e-01 -6.82408988e-01
2.48450592e-01 3.57656747e-01 5.97907417e-02 -7.06197321e-01
7.05800831e-01 1.43802539e-01 -9.54188928e-02 3.29543173e-01
-2.85368770e-01 -7.27391422e-01 5.41432559e-01 -3.97389472e-01
1.13487387e+00 4.63476837e-01 3.04455012e-01 -1.15029179e-01
5.04776657e-01 3.47223610e-01 6.78250194e-01 4.99535948e-01
1.09607585e-01 1.51433766e-01 2.69404173e-01 -6.18312120e-01
-1.04090440e+00 -7.28915215e-01 1.96034878e-01 6.05172992e-01
7.28806257e-01 -2.32570656e-02 -4.20444369e-01 -3.86931866e-01
-1.76836029e-01 8.58933091e-01 -4.69305366e-02 -2.76031811e-02
-1.03957438e+00 -2.30350092e-01 1.18516879e-02 5.57405114e-01
5.75909793e-01 -1.30349874e+00 -1.11967599e+00 8.36847246e-01
5.20882197e-03 -1.10838139e+00 -1.98612988e-01 1.97515920e-01
-1.13363409e+00 -1.26182187e+00 -6.88531399e-01 -1.04003882e+00
7.00119019e-01 5.74935198e-01 5.79617798e-01 4.22023349e-02
-4.82922673e-01 -1.54375985e-01 -3.72134447e-01 -3.68975967e-01
-4.01364625e-01 5.18397868e-01 5.17815724e-02 -7.91548491e-01
-1.83698550e-01 -5.30676305e-01 -5.92091560e-01 7.42097497e-01
-7.08418787e-01 -1.21079840e-01 1.01772296e+00 7.77847350e-01
4.52334285e-01 8.70618224e-01 7.82684028e-01 -4.00517583e-01
6.94632769e-01 -3.80280018e-01 -9.65764165e-01 8.88655409e-02
-3.12248647e-01 1.44784823e-01 9.55993533e-01 -4.21044469e-01
-1.18692613e+00 2.50735343e-01 -1.45888433e-01 -3.50616090e-02
-3.44293535e-01 6.32757545e-01 -1.55725420e-01 6.01358563e-02
4.30954427e-01 1.52610140e-02 1.00104757e-01 -3.46699238e-01
1.19334444e-01 4.18581337e-01 3.21659267e-01 -8.32504809e-01
7.58214474e-01 1.03138730e-01 4.92768735e-01 -8.32971096e-01
-1.82462096e-01 -4.51520383e-01 -7.20691323e-01 -2.40164444e-01
5.09805739e-01 -2.70558566e-01 -9.42154884e-01 4.58686084e-01
-1.27059400e+00 -4.00425106e-01 4.66603450e-02 6.38612330e-01
-9.05272722e-01 4.86180365e-01 -4.81616974e-01 -6.24984324e-01
3.48122679e-02 -1.71498358e+00 7.41749167e-01 3.22214663e-01
-2.23024681e-01 -6.81413829e-01 -4.82069962e-02 4.74758893e-02
4.32458997e-01 3.79362166e-01 9.83864665e-01 -1.93163812e-01
-9.12926972e-01 -1.47438288e-01 -6.72292709e-03 -1.31358117e-01
4.50768173e-01 -2.91162252e-01 -2.24497691e-01 -3.44140798e-01
-1.35609999e-01 -2.36965731e-01 1.78248152e-01 3.33580583e-01
1.27728415e+00 2.59019602e-02 -5.66905677e-01 1.74166590e-01
1.33923030e+00 8.27061355e-01 5.31439900e-01 7.60162771e-01
3.26567709e-01 7.24332333e-01 1.42990828e+00 6.18187726e-01
2.89200507e-02 8.72740984e-01 8.13603878e-01 2.04207987e-01
4.56188023e-01 -6.65962324e-02 -1.06012568e-01 5.71819544e-01
-2.35968471e-01 -3.16096544e-01 -1.01823759e+00 6.62092209e-01
-2.13035774e+00 -6.29290104e-01 -1.38767466e-01 1.91714275e+00
5.46813846e-01 1.63551122e-01 -2.02007264e-01 2.84849465e-01
5.00606060e-01 -1.74346849e-01 -7.44651556e-01 -4.13913399e-01
6.47869587e-01 5.99528924e-02 5.63044846e-01 2.79476434e-01
-9.41122711e-01 8.97987068e-01 5.04333591e+00 7.10978687e-01
-9.47407901e-01 -3.13727975e-01 -1.96125340e-02 1.20953642e-01
2.52572626e-01 -8.14288631e-02 -4.80487972e-01 2.45870724e-01
4.93597925e-01 2.70495415e-02 6.56787455e-01 1.17042732e+00
2.38868609e-01 -4.84616071e-01 -1.07377803e+00 9.10784543e-01
-6.74388647e-01 -1.11519790e+00 -4.12459373e-01 5.85811362e-02
5.95051527e-01 -1.77936703e-01 -1.93741679e-01 4.18446213e-01
4.32183027e-01 -9.06577110e-01 5.38753092e-01 7.43620992e-02
3.50116998e-01 -1.14792621e+00 9.95073140e-01 5.75276196e-01
-1.22068954e+00 -2.59754390e-01 -7.66370595e-01 -2.41948023e-01
5.82030535e-01 5.66792905e-01 -1.26363873e+00 9.78957534e-01
4.39281970e-01 2.68329352e-01 3.14539939e-01 1.31948018e+00
-4.47704643e-01 -1.06217943e-01 -3.45020115e-01 -3.21993679e-01
6.61271393e-01 -4.73575324e-01 4.43933129e-01 7.01110899e-01
7.04505324e-01 -1.66801382e-02 6.24864101e-01 6.90255761e-01
2.83404946e-01 -1.01759642e-01 -7.55708873e-01 1.40335709e-01
5.64199746e-01 1.15919352e+00 -9.55061257e-01 1.22211009e-01
-6.69422522e-02 8.34048808e-01 1.11865267e-01 3.04341644e-01
-9.71712589e-01 -8.28918159e-01 6.20307863e-01 -8.36560056e-02
1.93262398e-01 -9.58457291e-01 -9.22603980e-02 -4.64735776e-01
3.60827520e-02 -5.97718894e-01 -1.47388384e-01 -5.57315767e-01
-7.68415034e-01 3.72909129e-01 1.20288387e-01 -1.33834088e+00
-4.37368482e-01 -7.50683546e-01 -3.60787541e-01 5.98541200e-01
-1.58926713e+00 -5.22796273e-01 -5.52061141e-01 3.12406421e-01
1.13292229e+00 -1.52364045e-01 7.11944580e-01 2.07675844e-01
-5.22075534e-01 1.34130679e-02 1.30547762e-01 -1.54846966e-01
5.58317602e-01 -8.90015244e-01 3.80904794e-01 3.66012275e-01
-3.11320990e-01 5.07679343e-01 8.10634434e-01 -6.24721169e-01
-1.79398537e+00 -8.85746360e-01 2.65737265e-01 2.17074662e-01
4.90867674e-01 5.79371490e-02 -7.14514434e-01 4.54219669e-01
1.93512857e-01 -3.36370081e-01 -1.11999497e-01 1.51097193e-01
5.46771526e-01 -8.81116241e-02 -1.00947881e+00 9.13617253e-01
1.22751844e+00 2.37490475e-01 -3.19625497e-01 4.91655380e-01
5.63956797e-01 -8.03266525e-01 -7.66673863e-01 5.33847928e-01
4.74951297e-01 -6.79518402e-01 8.46108675e-01 -3.86377126e-01
4.43322182e-01 -2.73690820e-01 2.76419818e-01 -1.73127699e+00
-4.52965587e-01 -7.97478080e-01 3.03506702e-01 7.64752150e-01
1.79884180e-01 -6.95655882e-01 8.66174877e-01 3.38134021e-01
-4.70373511e-01 -1.00828052e+00 -8.27221036e-01 -1.11342478e+00
-1.90194234e-01 -1.54716641e-01 8.23531806e-01 6.60588264e-01
-8.18093866e-02 1.30239666e-01 -1.46326214e-01 3.65053982e-01
4.00077015e-01 1.85298666e-01 1.07228792e+00 -1.03816521e+00
-9.30759460e-02 -1.13432318e-01 -2.97809780e-01 -1.00536752e+00
1.27251208e-01 -3.63302201e-01 6.32873833e-01 -1.88867354e+00
-4.10638601e-01 -1.08701253e+00 1.59000624e-02 3.55734736e-01
2.26623744e-01 -6.42875671e-01 1.55454576e-01 1.83398664e-01
-4.28478271e-01 6.96710706e-01 1.78413379e+00 -1.39999717e-01
-5.71434855e-01 3.52004141e-01 -1.49901193e-02 7.43773222e-01
1.35725772e+00 -4.55655396e-01 -7.76488960e-01 -7.35042691e-01
2.41403192e-01 4.89097148e-01 -1.59679666e-01 -1.21682382e+00
4.30868894e-01 -6.75853610e-01 2.40434837e-02 -7.00177073e-01
4.02458638e-01 -1.16132545e+00 5.54848351e-02 8.61845315e-01
-2.47915415e-03 3.57844681e-01 2.91892588e-01 6.82950079e-01
-1.41838387e-01 -7.65340567e-01 4.79016393e-01 -3.55703712e-01
-1.04004943e+00 1.86532989e-01 -6.08199656e-01 -3.85049313e-01
1.48465991e+00 -1.55232906e-01 -8.76364037e-02 -2.40773574e-01
-4.92439866e-01 4.59241122e-01 4.52326685e-01 5.15676558e-01
6.79265320e-01 -9.81279194e-01 -4.03330356e-01 3.56489047e-02
-1.24854095e-01 6.46769404e-01 2.79935896e-01 6.66074872e-01
-9.62631226e-01 5.19555807e-01 -5.74502349e-01 -6.29141927e-01
-8.43196988e-01 6.80342078e-01 -1.61208212e-01 -5.39592803e-01
-7.62075245e-01 3.35450470e-01 2.90049575e-02 -5.91319323e-01
2.44574979e-01 -6.56027615e-01 3.59580480e-02 -3.07809174e-01
6.51178434e-02 8.07390332e-01 5.72330058e-02 2.69146979e-01
-2.38703772e-01 5.52506685e-01 -5.93887037e-03 -1.07767982e-02
1.41698730e+00 4.99289855e-02 3.26209404e-02 3.30150239e-02
7.35764742e-01 -3.99987310e-01 -1.37207091e+00 2.66150925e-02
3.19701344e-01 -5.62303364e-01 -1.57649219e-01 -4.53807354e-01
-8.85072112e-01 7.63494074e-01 1.55258030e-01 -1.28746241e-01
8.46959770e-01 -4.60485846e-01 1.02834451e+00 9.85573173e-01
1.32769656e+00 -1.37868249e+00 3.03548157e-01 7.70833552e-01
9.72261906e-01 -1.10316420e+00 2.31540799e-02 -6.21708751e-01
-5.36956489e-01 1.41741407e+00 7.00252175e-01 -2.14237884e-01
2.61191815e-01 1.65811777e-01 -3.58776376e-02 8.15008208e-02
-3.00897211e-01 1.30418509e-01 -2.32358471e-01 6.46319091e-01
-2.13668644e-02 1.01186885e-02 -3.62114280e-01 -2.45982353e-02
-1.21229254e-01 -3.95896882e-02 5.80556095e-01 1.39044869e+00
-5.65137565e-01 -1.36688793e+00 -3.96901697e-01 3.07813194e-02
6.11054376e-02 6.40576720e-01 3.00693750e-01 1.19622302e+00
-1.33245841e-01 1.01620603e+00 -2.41205528e-01 -1.15424514e-01
5.15509307e-01 -3.95400941e-01 7.14867294e-01 -6.66843057e-01
-2.45202333e-01 -1.19295396e-01 3.03927809e-01 -6.97448730e-01
-2.21202821e-01 -5.28694510e-01 -1.74324787e+00 -5.13596535e-02
-3.48219156e-01 9.71256346e-02 1.09886956e+00 8.66973996e-01
2.47935414e-01 8.79988611e-01 6.29359663e-01 -1.24789608e+00
-9.48529124e-01 -8.32049549e-01 -4.38570708e-01 2.54196636e-02
-5.48142679e-02 -1.05920875e+00 3.33038837e-01 -4.81960952e-01] | [4.83468770980835, 1.3807153701782227] |
a9ab733f-8a1c-444f-898c-99dbc73eab5f | xf2t-cross-lingual-fact-to-text-generation | 2209.11252 | null | https://arxiv.org/abs/2209.11252v1 | https://arxiv.org/pdf/2209.11252v1.pdf | XF2T: Cross-lingual Fact-to-Text Generation for Low-Resource Languages | Multiple business scenarios require an automated generation of descriptive human-readable text from structured input data. Hence, fact-to-text generation systems have been developed for various downstream tasks like generating soccer reports, weather and financial reports, medical reports, person biographies, etc. Unfortunately, previous work on fact-to-text (F2T) generation has focused primarily on English mainly due to the high availability of relevant datasets. Only recently, the problem of cross-lingual fact-to-text (XF2T) was proposed for generation across multiple languages alongwith a dataset, XALIGN for eight languages. However, there has been no rigorous work on the actual XF2T generation problem. We extend XALIGN dataset with annotated data for four more languages: Punjabi, Malayalam, Assamese and Oriya. We conduct an extensive study using popular Transformer-based text generation models on our extended multi-lingual dataset, which we call XALIGNV2. Further, we investigate the performance of different text generation strategies: multiple variations of pretraining, fact-aware embeddings and structure-aware input encoding. Our extensive experiments show that a multi-lingual mT5 model which uses fact-aware embeddings with structure-aware input encoding leads to best results on average across the twelve languages. We make our code, dataset and model publicly available, and hope that this will help advance further research in this critical area. | ['Vasudeva Varma', 'Manish Gupta', 'Anubhav Sharma', 'Bhavyajeet Singh', 'Tushar Abhishek', 'Shivprasad Sagare'] | 2022-09-22 | null | null | null | null | ['data-to-text-generation'] | ['natural-language-processing'] | [ 9.69130844e-02 3.26585323e-01 -1.19151406e-01 -4.57180887e-01
-1.16177237e+00 -6.51620686e-01 1.23122549e+00 9.54022557e-02
-1.85933128e-01 1.36679971e+00 8.15992296e-01 -5.52967668e-01
1.47609845e-01 -1.09542418e+00 -6.81228518e-01 -1.68585047e-01
2.17203572e-01 8.42501819e-01 -1.98319077e-01 -5.34913838e-01
1.73163906e-01 -1.23798819e-02 -1.23538494e+00 6.33159637e-01
1.16219246e+00 6.19651318e-01 3.69030447e-03 7.19318211e-01
-5.51447093e-01 7.61274278e-01 -8.77917647e-01 -1.06622303e+00
2.60172039e-01 -6.45612836e-01 -1.03526187e+00 -9.23625156e-02
4.07616168e-01 -2.15145163e-02 -4.17635478e-02 5.79471111e-01
7.18812823e-01 -4.49848175e-03 8.65370572e-01 -1.24074984e+00
-1.26679111e+00 1.34175634e+00 -2.95172095e-01 7.67915025e-02
5.97470105e-01 4.76029590e-02 9.82125640e-01 -9.41434264e-01
1.05145466e+00 1.34144378e+00 5.57557762e-01 6.35939002e-01
-8.62172067e-01 -6.10364318e-01 -2.99698621e-01 -6.20544963e-02
-1.23927343e+00 -2.82996237e-01 4.56247032e-01 -2.60371059e-01
1.19306135e+00 1.35942191e-01 3.91437650e-01 1.50098777e+00
4.42128658e-01 8.70151937e-01 1.27461410e+00 -5.69126070e-01
-8.00305977e-02 4.99316216e-01 -2.83781260e-01 4.65921819e-01
5.32785535e-01 -5.18326089e-02 -6.84365571e-01 -3.98532674e-02
5.14767170e-01 -4.95209217e-01 -1.21864088e-01 4.12823617e-01
-1.49146473e+00 1.15746284e+00 -9.75568891e-02 3.63444418e-01
-2.82280624e-01 -6.06304929e-02 5.44954598e-01 4.23972994e-01
7.89668739e-01 5.62207878e-01 -5.64737737e-01 -3.62967938e-01
-1.01457381e+00 7.78123260e-01 9.24442828e-01 1.39919424e+00
4.42243189e-01 3.43225926e-01 -3.98814440e-01 9.33398783e-01
2.81655807e-02 5.42588174e-01 9.04794514e-01 -3.35853785e-01
1.18139577e+00 6.21944606e-01 9.43350047e-02 -7.55325139e-01
-1.38603941e-01 -1.73747778e-01 -8.68306518e-01 -3.71280372e-01
1.31861255e-01 -7.85258114e-01 -1.05805695e+00 1.45634615e+00
1.09113522e-01 -4.10941929e-01 5.48411191e-01 5.89551210e-01
1.09539759e+00 9.26452518e-01 -3.42022590e-02 5.58474138e-02
1.37788296e+00 -8.44524086e-01 -8.73880804e-01 -2.20796287e-01
8.06454837e-01 -1.07538188e+00 9.46862459e-01 1.05794571e-01
-9.79532123e-01 -6.22506976e-01 -8.31563890e-01 -4.31903958e-01
-1.07858384e+00 4.31684285e-01 7.46001065e-01 8.89730334e-01
-1.00029707e+00 1.34746909e-01 -4.36951876e-01 -6.70814514e-01
1.39126033e-01 -9.58322659e-02 -5.05809009e-01 -2.52759755e-01
-1.64982033e+00 9.60408449e-01 8.66894782e-01 -2.52330214e-01
-4.94473726e-01 -7.23751962e-01 -1.13543534e+00 -3.97492647e-01
1.29581600e-01 -8.47639441e-01 1.05035567e+00 -4.00846481e-01
-1.24873471e+00 6.76603734e-01 -1.33186787e-01 -6.76708817e-01
6.22032523e-01 -2.20537543e-01 -7.75887728e-01 -2.68090904e-01
5.25746584e-01 8.00729334e-01 5.32345235e-01 -1.03675783e+00
-6.82527959e-01 -8.27108789e-03 -8.80081300e-03 1.47882193e-01
-1.93982124e-01 1.03256814e-01 -1.19446762e-01 -1.22376359e+00
-5.29656112e-01 -7.07293153e-01 -8.86199549e-02 -7.21880674e-01
-7.16718256e-01 -4.76715922e-01 6.23093903e-01 -8.92089248e-01
1.41335082e+00 -1.52537966e+00 -2.11005732e-01 -2.95822144e-01
-3.31614584e-01 1.09626994e-01 -1.85949415e-01 1.02823818e+00
-1.11112975e-01 5.79864323e-01 -1.86172828e-01 -2.47554109e-01
2.43867859e-01 1.63107634e-01 -5.25071383e-01 -2.91018546e-01
6.46389663e-01 1.32598174e+00 -8.39335859e-01 -7.63280153e-01
9.21952501e-02 3.03612649e-01 -4.98029441e-01 6.75904453e-02
-2.98157841e-01 1.81471445e-02 -5.15098631e-01 6.13279343e-01
4.42618757e-01 9.72120464e-02 4.23363447e-02 -3.46587747e-02
-9.54461545e-02 5.59180439e-01 -8.64282966e-01 1.66884959e+00
-7.57311046e-01 5.86347878e-01 -6.04307473e-01 -6.10464752e-01
1.05003679e+00 6.62095070e-01 2.48416513e-01 -5.70909917e-01
1.85624305e-02 4.41403240e-01 -1.77307725e-01 -4.67198014e-01
1.33579659e+00 -4.09398109e-01 -6.44272983e-01 7.06922233e-01
1.74673900e-01 -5.68959475e-01 8.03087413e-01 5.60285270e-01
8.60707939e-01 3.53111178e-01 2.26738289e-01 -3.67238000e-03
2.27641508e-01 3.69610608e-01 1.14279665e-01 6.72823310e-01
2.93878704e-01 7.05280006e-01 1.89780936e-01 -3.45084995e-01
-1.14708209e+00 -9.93002236e-01 -9.09385830e-02 7.63131201e-01
-4.96964008e-01 -7.89299607e-01 -6.38064146e-01 -9.75160062e-01
-2.61666924e-02 1.34739053e+00 -6.06089890e-01 8.98927823e-02
-7.23565638e-01 -1.09358716e+00 8.09811234e-01 4.40071791e-01
4.82520431e-01 -1.37076092e+00 -1.91466972e-01 5.38509548e-01
-5.72421074e-01 -1.28960896e+00 -6.01123929e-01 -8.93474668e-02
-6.55624926e-01 -7.33273983e-01 -7.93568611e-01 -7.18408704e-01
4.44265962e-01 -3.75957310e-01 1.26647878e+00 -5.56977868e-01
-1.93053469e-01 1.79675043e-01 -7.04891443e-01 -7.12236166e-01
-8.13283682e-01 3.50308985e-01 -4.08185367e-03 -2.52555400e-01
3.73236030e-01 -3.01377699e-02 6.49006218e-02 -1.67454720e-01
-1.18758309e+00 2.94087023e-01 5.59929252e-01 7.09377170e-01
2.67886758e-01 1.70889139e-01 1.04387558e+00 -1.27589619e+00
1.32603252e+00 -6.47966444e-01 -6.48236573e-02 3.49847466e-01
-5.15661299e-01 2.99756318e-01 8.65114689e-01 -1.63733482e-01
-1.43063414e+00 -4.47140753e-01 -2.20986500e-01 1.15799092e-01
-5.51302321e-02 8.85480523e-01 -2.57795583e-02 7.29401827e-01
6.91418409e-01 5.35316348e-01 -5.83457410e-01 -2.00923890e-01
7.27837861e-01 9.60177958e-01 4.96302396e-01 -6.98485494e-01
8.98560166e-01 4.01982628e-02 -4.04556185e-01 -6.82480395e-01
-7.77485549e-01 -4.19965088e-02 -4.68432486e-01 -4.91592661e-03
8.07969689e-01 -9.12915885e-01 1.04350917e-01 2.99684942e-01
-1.33581102e+00 -1.81289569e-01 -3.79891902e-01 3.90518814e-01
-4.97580498e-01 1.85590073e-01 -6.14821970e-01 -7.36633837e-01
-7.38783300e-01 -7.71109581e-01 1.26813114e+00 1.35707040e-03
-5.03058434e-01 -1.39475894e+00 1.49063259e-01 5.47603786e-01
3.55647147e-01 5.06935179e-01 1.01015532e+00 -7.08144128e-01
-3.68611187e-01 -2.57828981e-01 -1.59405202e-01 2.22536206e-01
4.75095153e-01 -1.82614878e-01 -5.95402777e-01 -5.15591167e-02
-5.13214707e-01 -6.37134552e-01 7.96083391e-01 1.83146596e-01
7.02531695e-01 -5.51620364e-01 -2.25195125e-01 2.88056731e-01
1.34314775e+00 1.50394410e-01 5.58950543e-01 2.02887848e-01
7.00687945e-01 7.15793073e-01 7.42462277e-01 4.21761125e-01
9.04895008e-01 4.70475286e-01 -2.19459042e-01 7.26817399e-02
-3.62580866e-01 -6.31963491e-01 6.30829930e-01 1.20063806e+00
1.05567761e-02 -8.58880520e-01 -7.84429312e-01 1.08723986e+00
-1.54380369e+00 -1.09376812e+00 -1.91073805e-01 1.83762527e+00
1.29366314e+00 6.10682443e-02 -1.34267762e-01 6.29275292e-02
3.85448188e-01 7.78768510e-02 -1.03418127e-01 -8.71352136e-01
-2.60830224e-01 5.21532178e-01 3.26187760e-01 3.41904491e-01
-1.02255440e+00 1.20971286e+00 6.09749269e+00 1.07080448e+00
-8.90770435e-01 1.55141398e-01 6.05440915e-01 1.14789598e-01
-6.73558474e-01 -1.75166070e-01 -1.17565155e+00 5.16264081e-01
1.22223663e+00 -7.26310551e-01 6.41831383e-02 6.09019399e-01
3.01401675e-01 2.07127959e-01 -1.07725430e+00 7.79765546e-01
3.36760551e-01 -1.53015947e+00 6.93411708e-01 -1.00828949e-02
1.10782444e+00 -7.92504027e-02 -1.98430300e-01 6.74689293e-01
8.74773502e-01 -1.12423480e+00 7.67535985e-01 3.28960121e-01
1.10651863e+00 -8.72296154e-01 9.21876252e-01 1.91256195e-01
-1.18520951e+00 3.11643332e-01 -2.83898652e-01 1.95375293e-01
6.45956457e-01 7.74797797e-01 -1.34383118e+00 1.14893162e+00
3.82538527e-01 7.26762652e-01 -5.91060221e-01 5.14709711e-01
-2.43788898e-01 6.29943371e-01 -1.31388813e-01 -2.31530383e-01
3.68885309e-01 1.75710488e-02 3.34367484e-01 1.50994444e+00
7.82295406e-01 -9.90070254e-02 1.75649196e-01 9.33268964e-01
-1.66291967e-01 4.06711131e-01 -8.42441082e-01 -5.43063343e-01
2.90218472e-01 1.02460384e+00 -5.19826353e-01 -5.89501977e-01
-2.93342799e-01 1.01466405e+00 1.49557307e-01 2.22176746e-01
-6.85602188e-01 -7.69652069e-01 3.48844737e-01 1.44942358e-01
2.40403473e-01 -2.19940990e-01 -3.25648576e-01 -1.25884461e+00
9.02955607e-02 -9.80108023e-01 5.17059982e-01 -9.20993090e-01
-1.55595362e+00 1.01643586e+00 2.83885449e-01 -1.06743634e+00
-9.47366953e-01 -4.51093405e-01 -4.10198718e-01 1.08109581e+00
-1.50396049e+00 -1.45502961e+00 1.98334187e-01 4.16098863e-01
9.22324896e-01 -4.60609585e-01 8.34784865e-01 2.34802172e-01
-3.11257780e-01 7.50836790e-01 -8.45111310e-02 3.95581245e-01
8.13320160e-01 -1.35160899e+00 9.63325679e-01 9.08617318e-01
1.88448250e-01 7.17487276e-01 5.28844655e-01 -1.11941600e+00
-1.15159428e+00 -1.68807304e+00 1.73407710e+00 -8.42044652e-01
7.52615809e-01 -5.45048535e-01 -3.83584261e-01 8.62655699e-01
7.54343152e-01 -5.98235130e-01 6.21903837e-01 -9.25360247e-02
-6.97287470e-02 1.89051673e-01 -1.00427997e+00 7.56184161e-01
8.68514001e-01 -3.32032353e-01 -7.31185377e-01 4.81951088e-01
1.00347340e+00 -3.64980280e-01 -1.05879164e+00 9.17237103e-02
3.02796751e-01 -5.31682253e-01 6.30249143e-01 -5.93633890e-01
9.64196026e-01 -2.19270796e-01 -6.09765984e-02 -1.64477515e+00
-2.49209702e-02 -6.40693426e-01 1.42605931e-01 1.68543601e+00
8.78483593e-01 -7.28670239e-01 5.44162571e-01 2.14738414e-01
-3.43685597e-01 -6.04991496e-01 -7.79642522e-01 -7.76762545e-01
4.52937692e-01 -7.25201428e-01 9.20291901e-01 1.10720503e+00
-6.40584305e-02 6.07255876e-01 -5.49776137e-01 -2.76678652e-01
1.71724066e-01 1.70837969e-01 8.13393831e-01 -7.25730419e-01
-5.67146726e-02 -1.18188314e-01 -1.34769291e-01 -6.30764604e-01
7.55053386e-02 -1.27066982e+00 -8.62268358e-02 -2.01118445e+00
1.14819907e-01 -5.28831482e-01 3.98052186e-01 6.78903282e-01
-3.10405076e-01 2.81812519e-01 1.52387857e-01 -2.67123431e-01
-5.43797016e-02 5.99598706e-01 1.37191701e+00 -2.04134390e-01
-2.28830855e-02 -2.85265356e-01 -8.41169834e-01 1.04743257e-01
9.07557666e-01 -5.35852492e-01 -6.01854682e-01 -5.71160853e-01
2.49032289e-01 -4.25863899e-02 -2.08185595e-02 -8.23380113e-01
-1.50544330e-01 -2.66851068e-01 3.62716168e-01 -7.52456844e-01
1.93071231e-01 -2.83966303e-01 1.72808439e-01 7.83759654e-02
-4.86213773e-01 4.89666879e-01 2.77954191e-01 1.47028551e-01
-3.34279418e-01 -2.35534355e-01 1.59604788e-01 -3.37011844e-01
-4.49256420e-01 2.29975104e-01 -4.97772723e-01 4.84289050e-01
7.67518401e-01 -2.14322582e-01 -5.85457444e-01 -6.19258463e-01
-1.84620947e-01 1.90201312e-01 1.48444459e-01 9.05619740e-01
6.08104110e-01 -1.56054592e+00 -1.40077341e+00 2.33342260e-01
3.08332056e-01 -2.73733228e-01 -9.60297510e-02 3.85658324e-01
-4.48469818e-01 9.02440369e-01 -1.03576735e-01 4.64961790e-02
-8.37074518e-01 2.80041486e-01 -1.45754546e-01 -7.65837848e-01
-2.33221769e-01 5.76648653e-01 -1.24953523e-01 -9.52199280e-01
-4.59341764e-01 -5.58134437e-01 -1.55560866e-01 1.14470556e-01
4.48911488e-01 2.31157348e-01 1.65256739e-01 -7.24154294e-01
-3.28278430e-02 1.54206261e-01 -2.85819739e-01 -5.19566953e-01
1.14992058e+00 7.87956491e-02 -4.09490652e-02 3.64809602e-01
1.03227878e+00 3.10948372e-01 -3.42207968e-01 6.55357242e-02
1.75195467e-02 -1.09296255e-01 -2.01496929e-01 -1.10000670e+00
-9.00347650e-01 6.31543100e-01 9.99056734e-03 2.05679208e-01
9.52059805e-01 -1.12798706e-01 1.04184508e+00 1.54164419e-01
5.55936515e-01 -1.11164033e+00 -1.16445556e-01 7.50617743e-01
1.16386211e+00 -1.09772444e+00 -7.39763603e-02 -1.90299392e-01
-9.94358599e-01 9.92173254e-01 5.87364018e-01 3.71126384e-01
3.88270825e-01 3.21382165e-01 2.24596977e-01 6.79052342e-03
-1.08196628e+00 -1.41926542e-01 2.94699490e-01 8.72607470e-01
9.28450167e-01 2.00989991e-01 -5.14204562e-01 5.91896951e-01
-1.11363435e+00 1.89918965e-01 8.64538550e-01 8.07626605e-01
5.74402846e-02 -1.54516828e+00 -3.40868264e-01 8.33986700e-01
-6.39163196e-01 -5.83579004e-01 -5.38808525e-01 9.66685474e-01
3.12130809e-01 1.08220041e+00 -1.35026336e-01 -3.10424536e-01
1.60585999e-01 2.69679159e-01 2.74072409e-01 -9.44665015e-01
-6.68192565e-01 -2.88510084e-01 7.30207443e-01 6.35338426e-02
-3.43070775e-01 -7.54408777e-01 -1.14817548e+00 -4.35274810e-01
-1.22420773e-01 3.08344215e-01 5.79432666e-01 8.38628471e-01
3.65642190e-01 4.39218998e-01 2.65283436e-01 -3.95856857e-01
-1.14029959e-01 -1.33860338e+00 -4.18784052e-01 4.30426031e-01
-1.80037051e-01 -1.95901543e-01 7.43300095e-02 3.94505650e-01] | [11.506692886352539, 9.423125267028809] |
b2141344-616a-42b9-bd83-580425348123 | semeval-2015-task-6-clinical-tempeval | null | null | https://aclanthology.org/S15-2136 | https://aclanthology.org/S15-2136.pdf | SemEval-2015 Task 6: Clinical TempEval | null | ['Marc Verhagen', 'Leon Derczynski', 'Guergana Savova', 'Steven Bethard', 'James Pustejovsky'] | 2015-06-01 | null | null | null | semeval-2015-6 | ['temporal-information-extraction'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.273398399353027, 3.7930564880371094] |
d6fc5618-1277-44bf-a452-85fc5f8486ee | named-entity-recognition-in-the-style-of | 2101.11122 | null | https://arxiv.org/abs/2101.11122v1 | https://arxiv.org/pdf/2101.11122v1.pdf | Named Entity Recognition in the Style of Object Detection | In this work, we propose a two-stage method for named entity recognition (NER), especially for nested NER. We borrowed the idea from the two-stage Object Detection in computer vision and the way how they construct the loss function. First, a region proposal network generates region candidates and then a second-stage model discriminates and classifies the entity and makes the final prediction. We also designed a special loss function for the second-stage training that predicts the entityness and entity type at the same time. The model is built on top of pretrained BERT encoders, and we tried both BERT base and BERT large models. For experiments, we first applied it to flat NER tasks such as CoNLL2003 and OntoNotes 5.0 and got comparable results with traditional NER models using sequence labeling methodology. We then tested the model on the nested named entity recognition task ACE2005 and Genia, and got F1 score of 85.6$\%$ and 76.8$\%$ respectively. In terms of the second-stage training, we found that adding extra randomly selected regions plays an important role in improving the precision. We also did error profiling to better evaluate the performance of the model in different circumstances for potential improvements in the future. | ['Bing Li'] | 2021-01-26 | null | null | null | null | ['nested-named-entity-recognition'] | ['natural-language-processing'] | [-3.37051868e-01 2.49349385e-01 1.48774996e-01 -6.91617370e-01
-9.07419086e-01 -6.86839819e-01 4.20657843e-01 1.52092233e-01
-1.14943361e+00 7.80038595e-01 6.78153336e-02 -3.36971492e-01
3.31352323e-01 -7.83033669e-01 -6.60948634e-01 -3.80243003e-01
-1.99943095e-01 6.22268617e-01 5.20247877e-01 -2.70142611e-02
9.70847383e-02 5.77882349e-01 -1.03251541e+00 3.19422990e-01
7.76967525e-01 9.09694254e-01 2.80603170e-01 7.38220394e-01
-3.44891757e-01 9.27257717e-01 -8.74024153e-01 -8.40443850e-01
1.99838161e-01 -1.58888325e-01 -1.10662580e+00 -2.59578973e-01
1.51066020e-01 9.78628732e-03 -2.78591305e-01 8.45686257e-01
7.83717215e-01 3.81381750e-01 5.70138931e-01 -8.78424287e-01
-5.35090923e-01 9.43696141e-01 -2.74369836e-01 2.89428115e-01
-2.91750208e-02 -9.37375799e-02 1.09568834e+00 -9.61376190e-01
6.99190617e-01 1.07777643e+00 8.83471608e-01 8.26850533e-01
-8.11178446e-01 -6.86983883e-01 1.68169886e-01 9.14158523e-02
-1.47777283e+00 -5.35476327e-01 2.54657984e-01 -1.90039709e-01
9.76222336e-01 1.06483459e-01 -1.55521873e-02 7.65699863e-01
-2.43478585e-02 9.50379491e-01 8.31738293e-01 -5.55328727e-01
2.47787684e-01 2.69405633e-01 3.03357035e-01 8.61243308e-01
-1.11076914e-01 1.27963275e-01 5.37604094e-02 -1.30088836e-01
6.11524940e-01 -3.54636461e-01 -1.94084346e-01 9.49448049e-02
-1.13723779e+00 7.93736994e-01 7.02221990e-01 5.98075032e-01
-3.85453194e-01 -3.80754769e-02 4.86807406e-01 -1.97376981e-02
4.44729656e-01 5.07517934e-01 -8.89067054e-01 8.24915320e-02
-1.13894379e+00 -2.13795349e-01 9.54417467e-01 9.71528471e-01
5.85311055e-01 -2.15622574e-01 -4.93542790e-01 1.00492311e+00
3.87625396e-01 -3.59335728e-03 4.88229424e-01 -6.25464439e-01
6.86812282e-01 4.62887675e-01 1.92558348e-01 -4.00070250e-01
-6.07499659e-01 -4.58950639e-01 -7.82960415e-01 -2.13702656e-02
5.07318914e-01 -5.91812313e-01 -1.41249442e+00 1.75228691e+00
2.80581445e-01 1.95149451e-01 1.99743092e-01 7.33877301e-01
9.25266802e-01 8.42846692e-01 4.42756832e-01 6.99289963e-02
1.67019451e+00 -1.13996685e+00 -6.72683418e-01 -6.37693256e-02
8.49410057e-01 -6.36497915e-01 6.76705003e-01 -3.43833491e-02
-9.60945129e-01 -7.19345510e-01 -7.45161355e-01 -7.83829242e-02
-5.40207207e-01 7.31723368e-01 4.97514069e-01 6.15417957e-01
-1.01728714e+00 6.95170879e-01 -8.69470835e-01 -4.07546759e-01
2.10612908e-01 3.27690512e-01 -2.90606499e-01 1.79640099e-01
-1.52411699e+00 8.96333039e-01 8.64266157e-01 4.05159742e-01
-7.71030009e-01 -4.50349152e-01 -8.82151127e-01 3.55489016e-01
1.46217406e-01 -3.68530929e-01 1.38818789e+00 -5.84064364e-01
-1.31410718e+00 9.42721248e-01 -1.53169885e-01 -5.89709044e-01
4.19161648e-01 -1.06247477e-01 -7.80268610e-01 -1.01576500e-01
2.37080097e-01 9.84772742e-01 2.38725781e-01 -1.02454007e+00
-8.66153061e-01 -2.22236782e-01 2.11949334e-01 -4.89882231e-02
-1.45410758e-03 3.45384330e-01 -6.08444333e-01 -6.23362660e-01
-1.94259420e-01 -7.88269877e-01 -5.10743499e-01 -3.79160374e-01
-5.27990162e-01 -5.37430584e-01 2.82286137e-01 -9.53413844e-01
1.24454713e+00 -2.18173814e+00 -2.49600887e-01 1.40834913e-01
1.08441468e-02 5.65534174e-01 -2.17605740e-01 3.40388685e-01
-2.34832883e-01 5.34025192e-01 -3.14273268e-01 -4.78055328e-01
4.83540930e-02 -1.64597973e-01 -1.28132969e-01 1.97907522e-01
3.07646930e-01 7.07529962e-01 -6.92158341e-01 -5.32620728e-01
-2.44764626e-01 4.35677975e-01 -3.21742117e-01 4.17984515e-01
-1.20735258e-01 -4.53934306e-03 -2.16914088e-01 4.60545391e-01
5.78838944e-01 -1.64624020e-01 1.05575003e-01 -2.08947197e-01
-3.17130059e-01 5.08476555e-01 -1.35941100e+00 1.47989941e+00
-5.07833302e-01 5.63422501e-01 1.01657799e-02 -7.50798762e-01
1.00045085e+00 6.98274553e-01 1.84561059e-01 -6.04487658e-01
7.79885128e-02 1.72231451e-01 -1.39323799e-02 -2.41188318e-01
6.75039172e-01 -1.70359105e-01 -2.37052381e-01 6.32609427e-02
3.69466126e-01 3.88631046e-01 3.06349814e-01 5.51189668e-03
1.15988493e+00 2.22121492e-01 4.42234963e-01 -9.80939791e-02
5.30546010e-01 1.47239938e-01 9.11889017e-01 7.10916877e-01
-5.62386274e-01 6.50136232e-01 3.29819143e-01 -4.40599293e-01
-9.04266357e-01 -8.56263936e-01 -1.70223206e-01 1.25203121e+00
2.64206273e-03 -2.96531886e-01 -8.40404153e-01 -1.41679370e+00
-4.41207260e-01 1.03947890e+00 -4.38017219e-01 1.71491951e-01
-7.52933919e-01 -7.80895412e-01 1.14524972e+00 8.17123175e-01
7.84966528e-01 -1.55580676e+00 -1.16885513e-01 3.47448051e-01
-1.83866054e-01 -1.13558125e+00 -6.98457599e-01 6.76278412e-01
-6.86924160e-01 -9.30172503e-01 -9.96705651e-01 -1.32537818e+00
5.20028114e-01 -3.40804815e-01 1.14915526e+00 -1.29661739e-01
-2.73145251e-02 -1.36038661e-01 -4.30889726e-01 -4.13831145e-01
-3.44292521e-01 4.43569869e-01 -2.68396497e-01 -2.26313487e-01
2.61323601e-01 4.93051074e-02 -4.18626577e-01 4.37654704e-01
-7.18008637e-01 -4.04951096e-01 8.58253658e-01 9.69168961e-01
5.10037005e-01 2.27469727e-01 6.16254449e-01 -1.13215947e+00
2.91383892e-01 -3.49617749e-01 -6.21890724e-01 5.05869389e-01
-4.00980920e-01 3.46599162e-01 7.36711085e-01 -2.18934879e-01
-1.39542115e+00 2.83047676e-01 -7.34463036e-01 -9.12922770e-02
-4.34290290e-01 3.24616760e-01 -5.26430309e-01 3.57883871e-01
6.59313321e-01 3.37798260e-02 -7.86548734e-01 -8.37548018e-01
3.95314038e-01 9.12455082e-01 3.59142512e-01 -3.65292907e-01
5.38912833e-01 1.15676351e-01 -6.99080825e-01 -6.25045955e-01
-8.58828723e-01 -6.83568418e-01 -5.71495354e-01 2.37188786e-01
1.07957006e+00 -1.09572947e+00 -6.38178706e-01 3.25179815e-01
-1.54399014e+00 -3.35446954e-01 -3.88422012e-01 5.79598546e-01
-1.71499133e-01 3.61294121e-01 -1.06195998e+00 -8.31604362e-01
-5.18119633e-01 -8.63086641e-01 1.02683282e+00 3.99092823e-01
1.69032305e-01 -9.95746732e-01 3.29398602e-01 2.23166659e-01
1.89947352e-01 -3.26276392e-01 6.65749967e-01 -1.19392991e+00
-2.05233604e-01 -8.65716487e-02 -3.86591882e-01 3.99253815e-01
-2.96074718e-01 -2.36148909e-01 -1.15188539e+00 -1.41272858e-01
-1.56198710e-01 -1.14600241e-01 1.08152330e+00 1.09031424e-01
1.03741336e+00 -1.96553797e-01 -4.84670103e-01 4.83442217e-01
1.41245902e+00 5.67757726e-01 9.33998108e-01 4.72099632e-01
6.77754819e-01 5.89166641e-01 6.48058236e-01 1.02862753e-01
4.00541455e-01 5.57515621e-01 1.69211447e-01 -4.02428299e-01
-6.99669048e-02 -3.16595256e-01 3.98666263e-01 6.56098306e-01
2.11840812e-02 -6.67728961e-01 -9.38680530e-01 6.82751954e-01
-1.57019162e+00 -1.09363151e+00 2.60741655e-02 1.90047336e+00
8.99371147e-01 2.03018546e-01 1.61536634e-01 -2.79339522e-01
1.24327576e+00 -8.65925383e-03 -3.93596321e-01 -3.70891452e-01
-3.44568975e-02 4.07348275e-01 7.18586504e-01 3.07236999e-01
-1.49218154e+00 1.31771457e+00 5.89716148e+00 9.68433380e-01
-1.02755332e+00 7.87192360e-02 8.59790266e-01 5.26495278e-01
-2.66143261e-03 9.12287310e-02 -1.35976076e+00 3.68347853e-01
1.30302083e+00 3.78508806e-01 -5.31971408e-03 9.80612218e-01
-7.07731321e-02 3.01978201e-01 -8.74800384e-01 5.96453130e-01
-1.92439318e-01 -1.10051036e+00 -2.98678249e-01 -3.51586491e-02
4.73123908e-01 4.23864007e-01 -6.89313412e-01 7.37502873e-01
5.85648894e-01 -7.71040857e-01 7.93777168e-01 4.22127038e-01
6.24403417e-01 -7.73838103e-01 1.16154027e+00 5.50854087e-01
-1.38221788e+00 6.17084354e-02 -5.29262960e-01 4.15734261e-01
3.21097583e-01 6.40760839e-01 -1.25540805e+00 6.57103777e-01
6.02279603e-01 1.52194679e-01 -4.42945838e-01 1.50964332e+00
-4.60184544e-01 8.35272968e-01 -4.05659080e-01 -1.36067808e-01
1.81663841e-01 1.65895179e-01 3.25463474e-01 1.70889044e+00
2.11397126e-01 -6.59464225e-02 1.38265803e-01 7.13080227e-01
-4.78097051e-01 2.81670123e-01 -1.84885293e-01 -2.45966353e-02
4.95309830e-01 1.41687667e+00 -9.14639533e-01 -5.06224513e-01
-2.75039017e-01 1.24439228e+00 6.29275143e-01 3.46338272e-01
-1.14136052e+00 -8.78859699e-01 2.72284210e-01 -2.22501084e-01
8.55922520e-01 -1.43195316e-01 -1.21636786e-01 -1.43680000e+00
-1.62081867e-01 -4.49512094e-01 7.25270033e-01 -5.13980567e-01
-1.37460625e+00 1.10039032e+00 -3.94308835e-01 -8.00885081e-01
2.43236292e-02 -6.55634642e-01 -6.87038124e-01 8.63960385e-01
-1.53498161e+00 -8.67805362e-01 1.28285035e-01 2.44568169e-01
5.68103850e-01 6.34016246e-02 9.33216333e-01 5.36267102e-01
-9.40638959e-01 9.23787117e-01 7.12256432e-02 9.18480158e-01
7.73706853e-01 -1.54285753e+00 8.29106629e-01 9.12237227e-01
5.31585217e-01 5.85022509e-01 1.95227951e-01 -5.15976191e-01
-4.80908304e-01 -1.30239737e+00 1.44801664e+00 -1.94871098e-01
2.83210963e-01 -4.97506618e-01 -7.87314892e-01 7.37424374e-01
2.79742241e-01 -5.87531878e-03 6.35394275e-01 2.07346439e-01
-2.50034004e-01 1.52646959e-01 -1.45692551e+00 4.94257987e-01
9.78823245e-01 -3.35043818e-01 -8.02164316e-01 1.22268453e-01
9.82847631e-01 -1.37321264e-01 -9.78226960e-01 3.19048941e-01
2.78079242e-01 -6.82324708e-01 7.84052074e-01 -7.11294174e-01
-1.04867615e-01 -2.91137069e-01 -1.68036431e-01 -1.28629088e+00
-5.82584321e-01 -2.88528830e-01 1.55458629e-01 1.81236875e+00
9.52650011e-01 -4.37520385e-01 7.75242329e-01 5.51572680e-01
-3.84149671e-01 -5.64991057e-01 -7.40798891e-01 -8.04473162e-01
3.26029509e-02 -3.03764343e-01 4.83420342e-01 8.78093243e-01
-3.03168982e-01 6.60376728e-01 -6.47598207e-02 3.95917386e-01
3.54452729e-01 -2.31045201e-01 2.02548370e-01 -1.01567829e+00
-2.00294614e-01 -1.73044041e-01 -2.60046035e-01 -1.18185973e+00
2.39993632e-01 -8.54543746e-01 5.63124001e-01 -1.53997195e+00
1.98694542e-02 -7.62834072e-01 -5.78812420e-01 7.92737186e-01
-3.32055122e-01 7.15622306e-02 2.80087173e-01 1.50029123e-01
-7.36863673e-01 3.93529892e-01 7.74317443e-01 -4.03287821e-02
-3.36273193e-01 2.11522803e-01 -6.93180680e-01 6.21879160e-01
8.85494530e-01 -6.37457550e-01 8.68871659e-02 -4.08184320e-01
-1.46956429e-01 -4.74440157e-02 6.08275272e-02 -9.46137428e-01
3.11720669e-01 6.70961365e-02 4.92382973e-01 -7.33785629e-01
5.33651300e-02 -7.55690455e-01 -2.48696551e-01 3.18683803e-01
-4.25577402e-01 2.92843059e-02 1.30565763e-01 3.85977626e-01
-3.26067328e-01 -5.37154675e-01 8.17706823e-01 -2.98448771e-01
-1.12635529e+00 2.64650583e-01 -1.42512903e-01 2.23453835e-01
8.61522853e-01 1.33275509e-01 4.96320007e-03 5.72441518e-02
-9.31797981e-01 3.03134292e-01 1.79886475e-01 1.25197396e-01
2.96693981e-01 -1.14306808e+00 -7.56925285e-01 -3.15482840e-02
-3.56006343e-03 -3.55934016e-02 3.89701240e-02 4.01575446e-01
-6.03401661e-01 4.47444737e-01 1.31002665e-01 -2.87222147e-01
-9.28402126e-01 5.27297437e-01 5.06996751e-01 -6.83743834e-01
-3.98506314e-01 1.02668607e+00 2.25665405e-01 -9.14875567e-01
5.18282056e-01 -4.15928811e-01 -4.80528980e-01 8.24545324e-02
5.57136834e-01 3.81266296e-01 1.99339181e-01 -6.02359176e-01
-6.30851924e-01 2.09556401e-01 -2.60166526e-01 -3.32480460e-01
1.38413203e+00 -1.09873097e-02 1.66212678e-01 2.42123708e-01
1.20557368e+00 1.35432303e-01 -9.47801411e-01 -8.75681937e-02
5.53826034e-01 1.55415237e-01 -1.09560646e-01 -1.08406842e+00
-1.11080563e+00 7.72739708e-01 5.93495369e-01 1.96326628e-01
1.02539504e+00 2.13208459e-02 1.01123559e+00 5.14097571e-01
3.25658917e-01 -9.89680350e-01 -4.57347631e-01 8.79975200e-01
3.55016351e-01 -1.17860138e+00 -4.63501245e-01 -3.68141830e-01
-7.14604318e-01 1.01542115e+00 6.01047218e-01 -1.22956678e-01
6.29866064e-01 2.89807826e-01 1.05932288e-01 -1.43225297e-01
-5.85615933e-01 -4.42823946e-01 3.21846336e-01 3.67501915e-01
8.58818054e-01 3.41230035e-02 -5.20128429e-01 8.78639400e-01
-1.70311574e-02 -9.21728089e-03 2.27335691e-01 7.53109872e-01
-4.98179793e-01 -1.11223328e+00 -2.71606207e-01 9.85194147e-02
-8.65146577e-01 -2.68741816e-01 -3.77504975e-01 8.54370654e-01
4.28730488e-01 9.43628967e-01 -1.14453144e-01 -4.06277418e-01
6.04390383e-01 3.90292794e-01 1.02244839e-01 -8.04974258e-01
-1.03230858e+00 -1.15347311e-01 5.22192299e-01 -4.52242106e-01
-1.70199841e-01 -4.18159783e-01 -1.69527721e+00 9.64362547e-02
-8.04756165e-01 8.28653693e-01 8.34665179e-01 8.48757505e-01
4.90562946e-01 5.31754136e-01 6.61638796e-01 -5.10715544e-01
-5.97011268e-01 -1.12847996e+00 -6.31614149e-01 2.98001379e-01
-7.30923489e-02 -3.36080641e-01 -2.87569851e-01 4.00830433e-02] | [9.674333572387695, 9.600103378295898] |
58961f36-2d76-42ab-9021-6fccd3cfa532 | identifying-copied-fragments-in-a-18th | null | null | https://aclanthology.org/2022.lrec-1.631 | https://aclanthology.org/2022.lrec-1.631.pdf | Identifying Copied Fragments in a 18th Century Dutch Chronicle | We apply computational stylometric techniques to an 18th century Dutch chronicle to determine which fragments of the manuscript represent the author’s own original work and which show signs of external source use through either direct copying or paraphrasing. Through stylometric methods the majority of text fragments in the chronicle can be correctly labelled as either the author’s own words, direct copies from sources or paraphrasing. Our results show that clustering text fragments based on stylometric measures is an effective methodology for authorship verification of this document; however, this approach is less effective when personal writing style is masked by author independent styles or when applied to paraphrased text. | ['Erika Kuijpers', 'Alie Lassche', 'Lianne Wilhelmus', 'Eleanor L. T. Smith', 'Roser Morante'] | null | null | null | null | lrec-2022-6 | ['authorship-verification'] | ['natural-language-processing'] | [ 5.56313470e-02 -9.96974576e-03 -4.12900925e-01 2.77211726e-01
-4.36728150e-01 -1.02227736e+00 1.02677846e+00 3.98349017e-01
-4.69738513e-01 8.84379089e-01 5.41648746e-01 -6.78155124e-01
-1.24571860e-01 -5.19397557e-01 -2.48022333e-01 -2.18879879e-01
8.47525001e-01 7.23440647e-01 -1.16386609e-02 -1.67837054e-01
1.02304423e+00 7.32419312e-01 -1.09282112e+00 -5.63079305e-02
8.96020591e-01 1.59352511e-01 -9.64197069e-02 9.02519107e-01
-6.42704904e-01 7.08125114e-01 -1.48137367e+00 -7.82197356e-01
1.20060384e-01 -6.46706283e-01 -7.43459642e-01 1.99562997e-01
6.55654132e-01 -1.22653499e-01 -3.21445763e-01 1.03799510e+00
1.13941148e-01 -6.11870408e-01 1.00390720e+00 -9.46271360e-01
-9.09506083e-01 8.00418198e-01 -5.99763095e-01 3.05275708e-01
9.03692722e-01 -2.07884073e-01 7.26067245e-01 -8.71533453e-01
9.56371665e-01 1.41602969e+00 6.36802912e-01 -9.10687596e-02
-1.43007219e+00 -9.41620469e-01 -4.54259068e-01 -1.94164351e-01
-1.23238516e+00 -5.38403571e-01 8.30418348e-01 -8.00063610e-01
3.31924886e-01 1.98161989e-01 7.02682674e-01 1.16405261e+00
4.05045241e-01 4.25045878e-01 1.46703506e+00 -8.31807911e-01
5.82516566e-03 4.29479420e-01 3.96727234e-01 5.58620691e-01
7.43551552e-01 -2.74009526e-01 -7.81817436e-01 -5.98679721e-01
5.83919466e-01 -3.09553087e-01 -1.66103721e-01 3.05890173e-01
-1.43142915e+00 6.25283301e-01 -2.84306079e-01 8.78767848e-01
-6.38562515e-02 -1.78706780e-01 4.03397590e-01 6.17483854e-01
2.42281839e-01 9.52220082e-01 -3.03910710e-02 -3.26687366e-01
-1.73088861e+00 2.76955694e-01 1.04897237e+00 1.03808427e+00
5.50819278e-01 -9.80176106e-02 8.22314024e-02 7.79780388e-01
2.45686471e-01 7.92166531e-01 8.79991770e-01 -8.27185094e-01
4.03613955e-01 1.04400039e+00 2.10217625e-01 -1.24236834e+00
8.37590173e-02 -2.80948758e-01 -5.45849800e-01 4.63106036e-02
6.04812145e-01 2.78305531e-01 -6.86270058e-01 1.00731146e+00
-3.36590409e-01 -9.47304964e-01 -2.10673571e-01 1.78225249e-01
9.11313951e-01 3.28794539e-01 -2.00676709e-01 -3.68113369e-01
1.35863316e+00 -3.96220535e-01 -1.01050377e+00 9.79735926e-02
3.56147438e-01 -1.47732055e+00 9.64352250e-01 5.10197282e-01
-1.23668945e+00 -4.80656594e-01 -1.18049181e+00 2.98251007e-02
-4.42443639e-01 4.16714638e-01 -2.83569753e-01 1.14033651e+00
-7.54867196e-01 8.14892769e-01 -2.92757094e-01 -4.50267315e-01
4.50900167e-01 -6.01612329e-02 -2.10723534e-01 4.51627284e-01
-8.46564591e-01 8.88292968e-01 1.01857610e-01 -3.26356471e-01
5.39803244e-02 -5.93760848e-01 -4.41357017e-01 -3.13292980e-01
5.11233024e-02 -1.58578724e-01 9.04327095e-01 -1.08059227e+00
-1.30838501e+00 1.62242758e+00 -4.93267745e-01 1.14871703e-01
9.74152565e-01 1.90232992e-01 -7.86633909e-01 3.04448694e-01
3.98797750e-01 -2.24183753e-01 1.33437669e+00 -1.21098375e+00
-5.14075100e-01 -5.78320086e-01 -5.47555566e-01 -2.43461028e-01
-3.28682959e-01 4.89985377e-01 -2.28449032e-01 -1.26393211e+00
3.96368891e-01 -8.53115618e-01 5.92957079e-01 -1.79682523e-01
-8.04409504e-01 -2.35434160e-01 7.18176007e-01 -1.25867069e+00
1.79256535e+00 -1.72207129e+00 -4.23016250e-02 7.46264756e-01
5.93730211e-01 3.40300590e-01 4.09184784e-01 8.90385866e-01
1.56906798e-01 8.07776570e-01 -2.02235833e-01 -2.00089827e-01
1.82088137e-01 -3.71629745e-02 -3.71186346e-01 6.32901728e-01
-3.66919726e-01 9.81761336e-01 -7.75608301e-01 -6.83931291e-01
-2.06722692e-01 -1.31891787e-01 3.90494555e-01 -1.15774274e-01
4.88099933e-01 -7.11243525e-02 -9.58793983e-03 9.43941891e-01
5.59991300e-01 1.25056561e-02 2.79556751e-01 2.51954347e-01
-2.08371595e-01 4.51544702e-01 -9.58055615e-01 1.16123080e+00
-1.94956020e-01 1.12830663e+00 7.85140395e-02 -1.53903320e-01
1.07483447e+00 1.19143605e-01 -3.08937252e-01 -4.71296281e-01
-4.08663005e-02 6.90836132e-01 -4.86058667e-02 -1.70038104e-01
9.90709484e-01 -2.10930780e-01 -1.69857651e-01 1.05811501e+00
-3.88373226e-01 -2.76267439e-01 5.07013083e-01 6.06575966e-01
7.38405645e-01 -1.08305238e-01 7.10394025e-01 -6.12291873e-01
7.46695995e-01 -4.47254702e-02 2.12233976e-01 7.97860801e-01
-8.86843875e-02 4.74263698e-01 9.18645442e-01 2.33786609e-02
-1.34113169e+00 -1.11525154e+00 -3.25418293e-01 6.49744987e-01
-2.75175095e-01 -6.64339721e-01 -6.39695346e-01 -6.31792307e-01
2.96052426e-01 5.37368357e-01 -6.52961135e-01 1.07984759e-01
-5.04990458e-01 -2.39927724e-01 9.42121923e-01 2.58054495e-01
2.21124828e-01 -9.21225548e-01 -5.53482473e-01 1.19757779e-01
-1.91413745e-01 -4.33994532e-01 -5.65012455e-01 -3.30748349e-01
-7.43966103e-01 -1.26037991e+00 -1.40451145e+00 -4.31294292e-01
6.90877736e-01 5.12230992e-02 9.96384382e-01 5.12078881e-01
-1.96426451e-01 1.77495643e-01 -2.77163029e-01 -2.67794937e-01
-1.34792912e+00 2.65125841e-01 1.62576120e-02 -3.36130619e-01
3.31641734e-01 -3.96896362e-01 -1.01105846e-01 3.73851597e-01
-7.48663306e-01 -5.00804901e-01 1.61634192e-01 5.62934875e-01
-1.75502405e-01 -2.11580217e-01 3.57016653e-01 -1.28692532e+00
9.98491228e-01 -2.67713398e-01 -2.17426553e-01 2.20468491e-01
-1.01937509e+00 -1.32468864e-01 5.05292356e-01 -3.71138841e-01
-8.02943587e-01 -5.82417071e-01 3.70653808e-01 -1.72680229e-01
-1.00923821e-01 4.33513552e-01 -3.65096740e-02 5.17385453e-02
7.79662192e-01 2.80870885e-01 1.90244615e-01 -5.84620059e-01
1.17197009e-02 1.23552334e+00 7.58258224e-01 -4.13971573e-01
1.05460274e+00 3.51355284e-01 -1.48028776e-01 -9.76850688e-01
-2.35193893e-01 -6.48272634e-01 -1.10108328e+00 -3.44339728e-01
2.73094863e-01 -5.84629178e-01 -5.46136916e-01 5.86107373e-01
-1.16056168e+00 1.78040564e-01 -3.24310780e-01 1.57000264e-03
-9.95273963e-02 9.86787736e-01 -5.26637018e-01 -7.77106404e-01
-1.56361043e-01 -8.02145958e-01 9.08665299e-01 -7.39293247e-02
-1.30502987e+00 -9.99623358e-01 2.50842124e-02 4.83036101e-01
-2.38526259e-02 1.81897044e-01 1.18420231e+00 -5.56301594e-01
9.47989598e-02 -7.58891940e-01 -1.76074699e-01 -6.80988701e-03
4.26821798e-01 6.60638273e-01 -6.21963024e-01 -1.64554417e-01
-1.47528693e-01 6.91467747e-02 5.14538467e-01 -2.15161636e-01
3.99959534e-01 -5.22280872e-01 -3.59894156e-01 1.90476552e-01
1.12567616e+00 2.13127419e-01 8.52570295e-01 6.39729738e-01
5.69533706e-01 7.73476362e-01 -6.15780009e-03 2.70774096e-01
-6.70390576e-02 4.14716363e-01 -4.56028640e-01 2.81655759e-01
-2.67506689e-01 -5.97013414e-01 3.26630950e-01 9.62196350e-01
-3.75867225e-02 -4.09144789e-01 -1.06270301e+00 6.05341852e-01
-1.38051832e+00 -1.14123666e+00 -7.90387392e-01 2.27800584e+00
6.58418179e-01 3.45338076e-01 5.55375695e-01 8.39179158e-01
9.19275999e-01 1.93686653e-02 -1.19166017e-01 -6.65282607e-01
-5.47296822e-01 1.39893994e-01 5.55752635e-01 4.19984251e-01
-2.85566449e-01 7.40711093e-01 8.02366352e+00 9.42853093e-01
-5.20208538e-01 -1.27993464e-01 3.15455645e-01 7.20428899e-02
-3.86069983e-01 -5.30841686e-02 -7.41174400e-01 8.86004567e-01
7.04230249e-01 -6.19624317e-01 8.02490786e-02 1.68063387e-01
2.22517103e-01 -6.06253266e-01 -7.98722327e-01 1.03198373e+00
4.51597929e-01 -1.23262930e+00 3.22652251e-01 4.36977893e-01
7.54564583e-01 -8.77949595e-01 -1.07124329e-01 -2.29719579e-01
1.62331641e-01 -9.07244921e-01 1.30887222e+00 4.92568970e-01
1.04342628e+00 -5.63326955e-01 6.41411781e-01 7.17496648e-02
-7.68879890e-01 1.02441221e-01 -2.98940212e-01 -2.77177185e-01
6.18679775e-03 6.05202675e-01 -5.74803650e-01 3.25890988e-01
1.89634323e-01 7.16597795e-01 -1.04342818e+00 5.33519924e-01
-4.87781405e-01 9.66083825e-01 5.32667264e-02 -1.85979754e-01
-1.12560764e-02 -3.94093335e-01 8.77840102e-01 1.37317610e+00
3.03862363e-01 -3.81254673e-01 -5.46054840e-01 9.19915676e-01
-3.01323980e-01 3.32848012e-01 -4.86988872e-01 -5.58043420e-01
7.53695965e-01 9.96985912e-01 -1.24970841e+00 -6.60712838e-01
6.07624687e-02 1.23989868e+00 -2.25286484e-02 3.32112759e-01
-1.08220786e-01 -8.00706923e-01 5.28379343e-02 6.56734824e-01
-1.14208527e-01 -2.01924667e-01 -9.37724590e-01 -9.03654039e-01
2.88947523e-01 -9.67205524e-01 8.41928050e-02 -7.48021364e-01
-1.15284383e+00 3.32028717e-01 -3.62000078e-01 -1.00330579e+00
3.55846137e-02 -5.20884871e-01 -7.45391488e-01 1.27146542e+00
-4.93468314e-01 -9.12642956e-01 -7.88274556e-02 1.30150527e-01
2.08282396e-01 -7.32804477e-01 5.36409676e-01 -1.05027527e-01
-2.84458965e-01 7.83472717e-01 6.06416821e-01 2.16763556e-01
9.15800333e-01 -1.44857788e+00 4.96417493e-01 7.23463535e-01
1.71035603e-01 8.79729331e-01 8.54910076e-01 -1.20858395e+00
-8.48404646e-01 -4.09687459e-01 1.62289977e+00 -7.81550646e-01
1.10415030e+00 -2.98751026e-01 -7.14738905e-01 4.20229703e-01
3.38759094e-01 -1.23682070e+00 7.42697179e-01 8.21163580e-02
-3.95162702e-01 4.20167178e-01 -1.04904556e+00 6.89630330e-01
9.99828577e-01 -8.67774010e-01 -1.28364730e+00 3.52675825e-01
-1.46974698e-02 1.88016027e-01 -8.96482408e-01 -4.88087445e-01
9.42727149e-01 -8.22580874e-01 5.55175781e-01 -3.02173108e-01
7.84990847e-01 -1.49345741e-01 2.68343776e-01 -1.03789127e+00
-3.48588794e-01 -1.00941515e+00 2.82134265e-01 1.59365284e+00
3.90948147e-01 -6.71442330e-01 6.85297966e-01 4.44804341e-01
2.27163553e-01 1.75536610e-02 -8.59310508e-01 -1.03137159e+00
2.82513231e-01 9.42210257e-02 3.81441802e-01 1.21453631e+00
4.33586240e-01 2.60812521e-01 -1.42580062e-01 -8.07011783e-01
6.10334694e-01 6.12331666e-02 8.96789432e-01 -1.61978436e+00
1.14336751e-01 -1.13413167e+00 -4.22733992e-01 -4.18625563e-01
3.12795162e-01 -9.29721415e-01 -3.50043744e-01 -1.43849528e+00
1.94946691e-01 -1.47053465e-01 5.32868445e-01 -5.57996556e-02
-3.97078926e-03 3.03398490e-01 4.38277185e-01 1.02746212e+00
1.75250441e-01 -1.69389725e-01 1.04190683e+00 -6.04528859e-02
-3.41034502e-01 4.16723229e-02 -8.11347246e-01 7.93801308e-01
6.08181119e-01 -5.31992197e-01 2.73239523e-01 2.11361825e-01
6.96954608e-01 -1.19581908e-01 2.09300533e-01 -7.61025488e-01
1.49872988e-01 -1.20409973e-01 5.65633774e-01 -7.47362912e-01
-3.54189277e-01 -5.01946628e-01 2.93327183e-01 6.32469594e-01
-3.01876396e-01 5.15942693e-01 1.54875338e-01 5.38983285e-01
2.90361214e-02 -8.99227679e-01 6.77281380e-01 -2.20193848e-01
1.50227591e-01 -4.67190653e-01 -1.22481441e+00 -6.80046678e-02
4.88125652e-01 -9.21667218e-01 -2.65561908e-01 -2.53741413e-01
-5.28841913e-01 -2.99473941e-01 1.27795029e+00 1.01897962e-01
3.02739531e-01 -1.17331719e+00 -7.34249711e-01 -1.85280349e-02
3.60649191e-02 -9.06826735e-01 -5.34117043e-01 6.57575846e-01
-9.99944270e-01 3.25462282e-01 -2.46607035e-01 -3.16220336e-02
-1.48484337e+00 4.42794949e-01 -2.44694158e-01 -1.17190115e-01
-5.89204967e-01 3.51007581e-01 -3.57417822e-01 -3.33543075e-03
8.69689435e-02 -1.41364047e-02 -1.79007813e-01 7.94536471e-01
5.25114417e-01 1.21885252e+00 8.24634805e-02 -1.17497945e+00
-2.51127154e-01 6.64304674e-01 -1.25795111e-01 -6.61759973e-01
5.58997869e-01 -4.48067933e-01 -3.15712631e-01 1.07002878e+00
1.12865472e+00 8.68425488e-01 -6.99114203e-02 -1.90113932e-01
1.69218302e-01 -6.93663418e-01 -2.04472452e-01 -9.08586204e-01
-4.50877041e-01 6.03855968e-01 -3.02381348e-03 3.38622123e-01
5.70044518e-01 -2.27434516e-01 6.73766494e-01 2.64411181e-01
8.95469263e-02 -1.64584482e+00 -1.77930981e-01 4.23440486e-01
1.09752655e+00 -8.13405633e-01 5.62560022e-01 -2.31597528e-01
-3.18832368e-01 1.47484112e+00 -1.92338377e-01 -6.07222915e-02
5.97450078e-01 -6.92402869e-02 9.26011726e-02 -3.15549344e-01
8.43981951e-02 1.78630233e-01 4.56714422e-01 3.66260678e-01
6.29150510e-01 1.01236798e-01 -9.58271980e-01 3.38319868e-01
-1.00647271e+00 -1.81981072e-01 1.24086154e+00 9.35916424e-01
-2.87930876e-01 -9.93168712e-01 -9.35329854e-01 9.18235958e-01
-6.73226058e-01 -8.62547681e-02 -1.40238702e+00 9.30192888e-01
-1.59505129e-01 9.65231538e-01 1.17683284e-01 -2.37323672e-01
3.61734703e-02 4.06452000e-01 5.64096630e-01 -5.02561212e-01
-1.09304583e+00 2.54293442e-01 1.64438322e-01 1.88415721e-01
-5.78811646e-01 -1.02435970e+00 -8.38176847e-01 -8.64270627e-01
-2.22374797e-01 1.70972735e-01 5.45799851e-01 9.72560048e-01
-1.43296823e-01 6.07375503e-02 3.02056789e-01 -3.94698679e-01
-2.28683367e-01 -1.07357907e+00 -9.98826563e-01 4.74919379e-01
1.34541169e-01 -4.25514102e-01 -8.03594053e-01 3.47218394e-01] | [9.588790893554688, 10.574206352233887] |
efa237eb-99ea-4018-b911-914ba6690cb5 | self-knowledge-distillation-based-self | 2206.03009 | null | https://arxiv.org/abs/2206.03009v1 | https://arxiv.org/pdf/2206.03009v1.pdf | Self-Knowledge Distillation based Self-Supervised Learning for Covid-19 Detection from Chest X-Ray Images | The global outbreak of the Coronavirus 2019 (COVID-19) has overloaded worldwide healthcare systems. Computer-aided diagnosis for COVID-19 fast detection and patient triage is becoming critical. This paper proposes a novel self-knowledge distillation based self-supervised learning method for COVID-19 detection from chest X-ray images. Our method can use self-knowledge of images based on similarities of their visual features for self-supervised learning. Experimental results show that our method achieved an HM score of 0.988, an AUC of 0.999, and an accuracy of 0.957 on the largest open COVID-19 chest X-ray dataset. | ['Miki Haseyama', 'Takahiro Ogawa', 'Ren Togo', 'Guang Li'] | 2022-06-07 | null | null | null | null | ['self-knowledge-distillation'] | ['computer-vision'] | [-5.85679673e-02 -2.60035187e-01 -1.92462340e-01 -3.01984936e-01
-5.80779731e-01 -4.89724219e-01 1.66338384e-01 7.96520174e-01
-4.81289148e-01 6.27661645e-01 -1.98541135e-01 -3.15463543e-01
-3.80440295e-01 -3.90284330e-01 -2.28302434e-01 -5.33514678e-01
-6.41126454e-01 8.46779585e-01 1.64147481e-01 5.01478076e-01
-3.07873189e-01 5.78637421e-01 -1.06106520e+00 5.18654168e-01
7.79816508e-01 7.76732504e-01 4.85078365e-01 1.32811809e+00
5.20667136e-01 6.19719684e-01 -4.71616745e-01 -1.05675235e-02
9.35980380e-02 -3.96985769e-01 -6.22529566e-01 -1.44521534e-01
2.72468209e-01 -4.68572944e-01 5.46336845e-02 4.59140956e-01
3.20760697e-01 -3.89399171e-01 1.21069515e+00 -1.21476734e+00
-4.31717187e-02 -2.71436393e-01 -4.88854170e-01 7.29265273e-01
2.54061073e-01 3.08004588e-01 4.97580349e-01 -5.31950355e-01
1.11090612e+00 6.88638031e-01 1.14620519e+00 4.68542993e-01
-9.16361034e-01 -5.68722665e-01 -5.44939339e-01 3.06514174e-01
-1.47193873e+00 4.40755814e-01 1.76113814e-01 -9.00454462e-01
1.07806957e+00 1.63537920e-01 9.67536569e-01 7.85328209e-01
5.66557229e-01 3.79466146e-01 1.14366603e+00 -1.77051678e-01
6.44866601e-02 2.64286935e-01 2.82630175e-01 1.22754896e+00
8.63565862e-01 1.06771670e-01 1.05439492e-01 -9.58568931e-01
8.65592837e-01 8.13014984e-01 -1.08798146e-02 -3.70166630e-01
-1.19622219e+00 1.11948931e+00 4.29547638e-01 2.79128820e-01
-4.93570685e-01 -4.99304801e-01 5.22590101e-01 1.78641036e-01
1.71010718e-01 4.40128088e-01 -4.95650589e-01 2.95942158e-01
-7.79607773e-01 -7.62717202e-02 5.77045619e-01 4.74427134e-01
3.77425224e-01 -3.08663011e-01 -8.23292881e-02 4.90256011e-01
3.40927452e-01 1.33497441e+00 3.64295095e-01 -3.31215054e-01
-1.32051393e-01 6.52074814e-01 1.13064781e-01 -7.74927557e-01
-1.03424466e+00 -2.55209088e-01 -1.13918710e+00 -2.19083533e-01
-6.00363538e-02 -5.89829683e-01 -9.98863280e-01 1.26616430e+00
4.93694127e-01 3.18909824e-01 2.96346128e-01 6.77795947e-01
9.62961853e-01 5.25503874e-01 1.86528906e-01 -6.71357095e-01
1.72581649e+00 -6.49130285e-01 -6.73529208e-01 6.52466342e-02
8.98753583e-01 -6.89361930e-01 3.10400248e-01 2.64033586e-01
-3.38429570e-01 -3.07655334e-01 -1.06469142e+00 8.90563726e-01
-1.14841208e-01 3.60549480e-01 4.14484143e-01 6.65183127e-01
-6.63804531e-01 1.82010949e-01 -1.00750625e+00 -1.00431037e+00
6.47348583e-01 1.75464466e-01 -4.35078770e-01 -2.34418064e-02
-7.74341464e-01 9.71903503e-01 4.05276507e-01 -6.56651795e-01
-8.43770027e-01 -8.46500695e-01 -5.29341459e-01 -5.67642391e-01
8.37498680e-02 -1.28099322e+00 9.64008510e-01 -2.47389987e-01
-5.31593621e-01 1.29132795e+00 -7.37349465e-02 -5.75004399e-01
3.45209599e-01 -1.70213625e-01 -8.17279220e-01 8.67411375e-01
1.58315375e-01 3.62969190e-01 9.27554071e-01 -1.38081539e+00
-8.91477346e-01 -4.64845359e-01 -6.29018068e-01 -1.67643234e-01
-3.44633088e-02 1.97824940e-01 -3.43373395e-03 -6.13064170e-01
-1.94880813e-01 -1.29844785e+00 -3.75664622e-01 8.47020820e-02
-4.72528264e-02 -2.69363344e-01 1.02963567e+00 -5.91813624e-01
1.03886247e+00 -2.06055713e+00 -9.25590158e-01 4.00391638e-01
7.31583059e-01 1.08433783e+00 3.03485412e-02 4.50283617e-01
-7.73282945e-02 -1.93663865e-01 -2.24845037e-01 3.80664259e-01
-7.15487480e-01 1.64250389e-01 5.67851700e-02 6.72432780e-01
2.43414149e-01 9.57777977e-01 -1.14050925e+00 -9.18562233e-01
3.06650698e-01 5.20543098e-01 -4.73037928e-01 8.62452507e-01
1.58819314e-02 4.72785354e-01 -4.75384325e-01 4.47477490e-01
7.02688634e-01 -9.10896778e-01 3.69675189e-01 -1.97305888e-01
3.27852547e-01 -6.15016282e-01 -6.65351689e-01 1.02919495e+00
1.00580983e-01 4.40142065e-01 -2.61883795e-01 -3.86708885e-01
7.22395241e-01 6.18325114e-01 8.63504350e-01 -6.75937384e-02
2.59393513e-01 -1.65374622e-01 -2.08385110e-01 -1.12496340e+00
-2.40320578e-01 -2.35148519e-01 3.18197668e-01 8.12488854e-01
-1.75232664e-01 -1.07219793e-01 1.25138357e-01 3.44076872e-01
1.30813134e+00 -6.52891576e-01 7.94578910e-01 -1.43819898e-01
3.57720017e-01 5.88743508e-01 2.69884169e-01 9.23679888e-01
-3.59597981e-01 5.23220837e-01 -1.91390455e-01 -6.10543430e-01
-8.26724648e-01 -1.53506815e+00 -3.09975207e-01 3.44789565e-01
-2.01806933e-01 -3.50888997e-01 -6.99461281e-01 -9.45849717e-01
-7.32523277e-02 2.80480742e-01 -5.49130797e-01 -2.13023555e-02
-4.22090799e-01 -6.47497058e-01 4.99469519e-01 5.33057034e-01
2.38699615e-01 -8.44177663e-01 -1.27646351e+00 5.72817586e-02
-1.70935720e-01 -1.22207057e+00 -3.03284794e-01 -1.45628616e-01
-9.25016820e-01 -1.68167114e+00 -8.74694884e-01 -1.23035359e+00
8.77637744e-01 4.03208733e-01 8.01574051e-01 3.11382860e-01
-1.11504745e+00 7.81293035e-01 -3.54446739e-01 -7.59440005e-01
-6.05344534e-01 -2.65677243e-01 3.86433601e-01 -2.70876437e-01
5.74854732e-01 4.96568866e-02 -9.60139930e-01 3.06541085e-01
-6.28500938e-01 -1.87994659e-01 5.34485877e-01 7.77879775e-01
7.32857049e-01 -1.86605245e-01 6.51609659e-01 -1.02742922e+00
5.47771931e-01 -6.86198175e-01 -4.20023233e-01 3.48361671e-01
-9.98727024e-01 -4.78633255e-01 2.62905568e-01 -3.42741340e-01
-9.51630056e-01 3.12922984e-01 2.34548047e-01 -6.74529076e-01
-3.44614565e-01 8.69624242e-02 9.32617188e-01 2.83523113e-01
9.07739818e-01 -2.78347824e-02 4.53709930e-01 -2.02653870e-01
9.97599121e-03 9.78737473e-01 6.74324393e-01 3.59537423e-01
6.96294546e-01 7.55295694e-01 -2.58799177e-02 -1.10499358e+00
-8.73896062e-01 -1.19684708e+00 -4.75518852e-01 -2.07848996e-01
1.41738200e+00 -1.19202328e+00 -8.71897578e-01 4.58665878e-01
-1.09687257e+00 1.45967469e-01 -6.04187846e-02 1.07697463e+00
-4.62106973e-01 5.12012482e-01 -4.69158262e-01 -6.72953188e-01
-8.35768342e-01 -6.36505365e-01 7.13030636e-01 7.48603195e-02
-6.01452649e-01 -1.08661628e+00 8.68511975e-01 4.47290361e-01
2.76952952e-01 4.50082392e-01 7.93154538e-01 -1.12200046e+00
-1.81786343e-01 -4.56847399e-01 -6.68474674e-01 2.38043576e-01
5.90388477e-01 -1.96878910e-01 -8.88819754e-01 -6.76469088e-01
2.49138147e-01 -3.18276107e-01 8.00026476e-01 4.48777407e-01
4.82434213e-01 -3.41314971e-01 -9.35978651e-01 4.72747147e-01
1.51727629e+00 4.76576775e-01 -1.05845686e-02 -1.84562325e-01
6.28713310e-01 2.72902369e-01 7.58508325e-01 7.47469723e-01
4.38377798e-01 -3.51387523e-02 1.93096533e-01 -4.26337719e-01
8.55773464e-02 6.05207123e-02 -3.51823926e-01 6.40734971e-01
-1.57243580e-01 -6.66702166e-02 -1.38539219e+00 6.16676271e-01
-1.70308685e+00 -9.08100724e-01 -3.12311053e-01 1.81927192e+00
5.16867340e-01 -2.76947498e-01 2.79043108e-01 -2.31039658e-01
7.99101889e-01 -3.94757539e-01 -5.28922021e-01 -2.66660094e-01
8.63190275e-03 9.20773670e-03 1.92176029e-01 5.24240509e-02
-1.42436707e+00 2.74544209e-01 6.99748468e+00 3.22985560e-01
-8.86994481e-01 3.66155207e-01 2.56385714e-01 2.00780869e-01
3.95558625e-01 -5.95591664e-01 -5.08547127e-01 2.07039386e-01
6.21318638e-01 -9.89692137e-02 -1.69372186e-01 8.65739048e-01
4.80081476e-02 -2.47446284e-01 -7.50158072e-01 1.27052736e+00
5.13077080e-01 -1.56059754e+00 -4.90988977e-02 -2.81087071e-01
1.00257730e+00 3.83993745e-01 -2.28324994e-01 -2.64201939e-01
2.18811601e-01 -6.98274434e-01 -3.77546757e-01 6.36510432e-01
1.14512038e+00 -5.25102437e-01 9.66505110e-01 2.74260670e-01
-1.12315166e+00 1.62677169e-01 -9.12902877e-03 3.59045327e-01
2.54765302e-01 4.53324556e-01 -1.88464224e+00 2.02683523e-01
8.12229097e-01 4.49919432e-01 -3.00553888e-01 1.28941965e+00
1.83247119e-01 7.30747104e-01 -3.16288590e-01 -1.17010862e-01
4.48095128e-02 2.59630203e-01 5.37254453e-01 1.75432611e+00
-2.85219681e-02 6.02942467e-01 3.33969563e-01 2.03820057e-02
4.75775421e-01 3.84666443e-01 -8.73671591e-01 -3.87172885e-02
5.08653484e-02 1.11180007e+00 -9.81335163e-01 -7.89265394e-01
-1.90451920e-01 7.54416943e-01 -1.91393092e-01 1.94208100e-02
-8.41407061e-01 -2.32542649e-01 4.37645167e-01 3.45044285e-01
7.51646161e-01 1.46559387e-01 8.93731117e-02 -7.68190980e-01
-5.10019064e-01 -7.40131974e-01 1.18158150e+00 -7.86244512e-01
-1.70665896e+00 9.36423540e-01 1.62443608e-01 -1.46212363e+00
-3.75169575e-01 -4.94831771e-01 -5.13720334e-01 1.53559342e-01
-1.28144848e+00 -1.20795929e+00 -3.86661530e-01 6.27109528e-01
8.54410380e-02 -4.51693118e-01 1.41929257e+00 -2.57599205e-01
-4.32831086e-02 4.29886162e-01 3.35142732e-01 2.79649772e-04
7.42720962e-01 -8.43920231e-01 -1.36936843e-01 3.27167690e-01
-3.19717616e-01 6.57673240e-01 3.70346129e-01 -1.13254189e+00
-1.02981186e+00 -1.39962912e+00 9.84967709e-01 -5.00216067e-01
4.20434386e-01 1.90411747e-01 -8.60518456e-01 4.60865200e-01
3.47713888e-01 1.27640337e-01 1.17904019e+00 -5.51243424e-01
-4.60233331e-01 1.41836822e-01 -1.63387179e+00 2.26762276e-02
7.46239007e-01 -4.51159656e-01 -9.77237046e-01 7.86113560e-01
4.50406343e-01 -5.05410396e-02 -9.80380118e-01 7.98536837e-01
6.39585555e-01 -7.11871922e-01 1.11168540e+00 -7.85299063e-01
-2.28709042e-01 -2.40991309e-01 1.68248683e-01 -8.54343235e-01
-3.51675332e-01 -5.39184570e-01 8.51886049e-02 2.44937822e-01
2.07498133e-01 -7.89301157e-01 5.54335594e-01 -2.23370597e-01
5.36132216e-01 -7.63305902e-01 -5.77853739e-01 -8.48881364e-01
-6.13036215e-01 -1.09076157e-01 2.08955258e-01 1.11391592e+00
1.16326325e-01 3.04367810e-01 -1.48588911e-01 3.79946798e-01
8.06495667e-01 3.75029057e-01 5.33359408e-01 -1.49512017e+00
-1.27291694e-01 1.48297414e-01 -5.36766350e-01 -2.81428874e-01
-6.51416957e-01 -7.54644632e-01 -3.22204143e-01 -1.79851866e+00
8.35910141e-01 -4.91616040e-01 -4.65205699e-01 3.76147211e-01
-2.75964886e-01 6.38353586e-01 5.32687455e-02 4.47853655e-01
-8.37259591e-01 -3.19119543e-01 1.04833686e+00 1.38606585e-03
-2.19037920e-01 9.55899060e-02 8.43356997e-02 9.60191011e-01
1.27738523e+00 -9.20788705e-01 -4.97562766e-01 1.79733425e-01
1.65513222e-04 1.72217906e-01 2.41024375e-01 -9.35649991e-01
5.76271899e-02 -1.96440548e-01 4.30078179e-01 -1.14436948e+00
1.47246927e-01 -7.76173234e-01 3.40820760e-01 1.56077683e+00
2.09185332e-01 2.65534431e-01 2.49474138e-01 7.70279348e-01
2.03502730e-01 -1.98393911e-02 8.42813611e-01 -8.13287646e-02
-4.22449678e-01 1.51881039e-01 -1.01125336e+00 3.21276248e-01
1.54849458e+00 -1.23820417e-01 -3.04183125e-01 8.10326487e-02
-6.29869163e-01 7.60277808e-02 3.70952696e-01 2.86799133e-01
9.42116618e-01 -8.50448310e-01 -1.01398635e+00 5.97422421e-01
6.90393150e-01 -3.76978785e-01 3.96162361e-01 9.00066614e-01
-9.43779826e-01 5.69485068e-01 -2.32441485e-01 -1.23025930e+00
-2.05281258e+00 8.73217940e-01 -7.49068409e-02 -5.11988819e-01
-6.11719847e-01 4.39559907e-01 2.44585544e-01 -1.48656080e-02
6.41771629e-02 1.38724232e-02 -3.23360533e-01 1.29446298e-01
6.24829113e-01 3.33445400e-01 -1.60863429e-01 -6.66008294e-01
-7.91459501e-01 7.96986759e-01 -2.51986474e-01 5.00692725e-01
1.37425852e+00 1.58935949e-01 4.62364182e-02 2.32167348e-01
1.18744826e+00 -2.56346345e-01 -5.44910729e-01 -3.07104945e-01
-4.16338712e-01 -2.75105804e-01 -3.11227024e-01 -9.54126656e-01
-5.62430620e-01 3.72853309e-01 1.35714972e+00 2.21411791e-02
1.06138086e+00 4.65872914e-01 8.47589493e-01 7.94915140e-01
1.41324058e-01 -7.29545653e-01 3.46163422e-01 2.53667891e-01
3.94406199e-01 -1.55550063e+00 3.00808877e-01 -3.61413538e-01
-9.90615547e-01 6.50105357e-01 2.45705411e-01 -5.61405160e-02
1.06634259e+00 3.49853456e-01 7.74695873e-01 -6.76540136e-01
-7.40332961e-01 -7.82245845e-02 2.39662901e-01 1.17927206e+00
-7.69907013e-02 4.72423851e-01 -3.52078050e-01 3.54443878e-01
3.31412554e-01 1.84540704e-01 2.97553111e-02 1.27724349e+00
-6.71116471e-01 -5.57639778e-01 -4.66293275e-01 9.51022506e-01
-2.93467492e-01 1.00185610e-01 -1.99368954e-01 5.70820570e-01
2.83263832e-01 8.93962979e-01 1.14029095e-01 -2.65021652e-01
3.53445075e-02 -1.76241882e-02 4.49905664e-01 -5.52275717e-01
-4.69839752e-01 -1.43324314e-02 -3.33331674e-02 -1.51788622e-01
-7.44161487e-01 -5.18851578e-01 -1.58733487e+00 2.14931414e-01
-4.26091850e-01 1.08451992e-01 5.85069180e-01 6.17070317e-01
4.31917906e-01 1.29885435e-01 7.90063024e-01 -1.08131031e-02
-2.92210460e-01 -5.80322444e-01 -3.67234051e-01 4.70237792e-01
7.37882257e-01 -3.75977188e-01 -1.61202475e-01 4.27471071e-01] | [15.57408618927002, -1.6978635787963867] |
45f2797c-26d0-4d00-81e9-a4b94ad14b3e | common-pets-in-3d-dynamic-new-view-synthesis | 2211.03889 | null | https://arxiv.org/abs/2211.03889v1 | https://arxiv.org/pdf/2211.03889v1.pdf | Common Pets in 3D: Dynamic New-View Synthesis of Real-Life Deformable Categories | Obtaining photorealistic reconstructions of objects from sparse views is inherently ambiguous and can only be achieved by learning suitable reconstruction priors. Earlier works on sparse rigid object reconstruction successfully learned such priors from large datasets such as CO3D. In this paper, we extend this approach to dynamic objects. We use cats and dogs as a representative example and introduce Common Pets in 3D (CoP3D), a collection of crowd-sourced videos showing around 4,200 distinct pets. CoP3D is one of the first large-scale datasets for benchmarking non-rigid 3D reconstruction "in the wild". We also propose Tracker-NeRF, a method for learning 4D reconstruction from our dataset. At test time, given a small number of video frames of an unseen object, Tracker-NeRF predicts the trajectories of its 3D points and generates new views, interpolating viewpoint and time. Results on CoP3D reveal significantly better non-rigid new-view synthesis performance than existing baselines. | ['David Novotny', 'Andrea Vedaldi', 'Natalia Neverova', 'Ignacio Rocco', 'Jeremy Reizenstein', 'Roman Shapovalov', 'Samarth Sinha'] | 2022-11-07 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Sinha_Common_Pets_in_3D_Dynamic_New-View_Synthesis_of_Real-Life_Deformable_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Sinha_Common_Pets_in_3D_Dynamic_New-View_Synthesis_of_Real-Life_Deformable_CVPR_2023_paper.pdf | cvpr-2023-1 | ['object-reconstruction'] | ['computer-vision'] | [-1.47590131e-01 1.25426218e-01 1.19737998e-01 -2.15230137e-01
-7.95858324e-01 -8.79117012e-01 9.08540308e-01 -6.99736774e-01
2.02199340e-01 4.02045637e-01 6.14357829e-01 2.09125593e-01
4.48761225e-01 -2.47229293e-01 -1.26205564e+00 -4.08061087e-01
1.39258504e-01 9.77813423e-01 6.01003766e-01 6.54625520e-02
1.21677935e-01 8.73761058e-01 -1.62188089e+00 3.93734455e-01
7.84208477e-02 5.57721317e-01 3.00881952e-01 6.99695766e-01
2.97201902e-01 5.31035304e-01 -4.18083727e-01 -4.57683712e-01
8.23931515e-01 -9.68695283e-02 -5.27620852e-01 7.43689120e-01
9.15963411e-01 -7.97455013e-01 -6.83520317e-01 6.34375572e-01
3.64060640e-01 1.42170817e-01 6.51293933e-01 -1.14134562e+00
-5.48649192e-01 1.10283546e-01 -6.75281584e-01 1.44142985e-01
1.00125086e+00 6.11516178e-01 5.50181270e-01 -1.18386793e+00
1.42780948e+00 1.68403399e+00 8.17766249e-01 7.98824251e-01
-1.63394868e+00 -3.26738536e-01 1.82269260e-01 -2.24152580e-01
-1.18838847e+00 -7.94536054e-01 6.24603629e-01 -7.04557955e-01
9.09940660e-01 2.01837122e-01 9.52767909e-01 1.63323629e+00
-2.28687450e-02 7.26856172e-01 9.41642523e-01 -6.33443445e-02
1.22989483e-01 -1.19091183e-01 -5.36685765e-01 5.40970504e-01
1.94589317e-01 4.43561316e-01 -7.39268482e-01 -2.30838105e-01
1.22399354e+00 1.43327996e-01 -5.58586597e-01 -9.63585079e-01
-1.88462591e+00 4.20701563e-01 2.09297523e-01 -2.27017328e-01
-3.44312072e-01 2.88567781e-01 -1.18913345e-01 4.38289065e-03
4.50601012e-01 5.03522679e-02 -3.68883908e-01 -1.76706582e-01
-7.04070270e-01 6.33595169e-01 8.33406210e-01 1.43739665e+00
4.64988261e-01 2.09379569e-01 -6.02471158e-02 3.69139940e-01
3.08418036e-01 7.31396675e-01 1.37645081e-01 -1.65867972e+00
4.01422232e-01 1.56510234e-01 5.41843355e-01 -6.17391944e-01
-6.34923056e-02 7.00529888e-02 -4.72469032e-01 3.25289965e-01
3.88738602e-01 2.20176741e-01 -1.12992513e+00 1.42108488e+00
8.52635980e-01 4.05468971e-01 4.76045795e-02 1.08006561e+00
1.01450527e+00 5.95690727e-01 -4.44940746e-01 -3.91488783e-02
9.22281921e-01 -1.08464205e+00 -2.64879942e-01 -1.82660162e-01
3.72137986e-02 -9.46019173e-01 8.69563818e-01 3.42493176e-01
-1.34075403e+00 -5.53838491e-01 -5.51617920e-01 -2.43192792e-01
2.03161284e-01 1.09971976e-02 4.38517839e-01 1.68929845e-01
-1.03093207e+00 4.93317634e-01 -9.83692825e-01 -5.36910295e-01
4.20898259e-01 1.70533247e-02 -8.94056439e-01 -3.44400138e-01
-1.94448099e-01 9.24542069e-01 4.65478152e-02 -6.34951741e-02
-1.64484692e+00 -9.72135484e-01 -8.00590217e-01 -4.95755225e-01
2.73876011e-01 -1.07271147e+00 1.34037876e+00 -5.00996411e-01
-1.55525243e+00 1.43155265e+00 -1.42614454e-01 -3.83827746e-01
8.92769098e-01 -3.23990583e-01 -1.20130993e-01 2.89617717e-01
2.22725660e-01 1.04755414e+00 1.13336992e+00 -1.71967971e+00
-3.49880069e-01 -2.37740681e-01 9.38154683e-02 2.39193454e-01
6.47009254e-01 -1.39359117e-01 -7.61636317e-01 -6.58254385e-01
3.37414503e-01 -1.30570984e+00 -1.59554720e-01 3.45445007e-01
-4.12366986e-01 1.73162315e-02 1.01746619e+00 -5.71824849e-01
2.31050536e-01 -2.11291695e+00 5.87718606e-01 -2.39971295e-01
2.45924070e-01 2.48679935e-05 -1.60265863e-01 2.09156603e-01
-1.40134856e-01 -3.02422434e-01 1.09586595e-02 -8.39979112e-01
8.41103196e-02 4.52576995e-01 -3.59002501e-01 8.28090012e-01
1.48241088e-01 9.28760529e-01 -7.90182471e-01 -2.86319554e-01
4.52563107e-01 6.60349905e-01 -8.83109272e-01 4.81693655e-01
-5.11539698e-01 8.36001873e-01 -1.64704934e-01 9.06373203e-01
6.85686946e-01 -4.10905868e-01 1.81338364e-05 -4.84241903e-01
1.68674979e-02 -7.80627057e-02 -1.29662800e+00 2.13354754e+00
-1.17506860e-02 5.92562854e-01 -2.28508282e-02 -6.68784499e-01
8.07301819e-01 2.58339316e-01 6.84534013e-01 -2.20674172e-01
-4.97774519e-02 5.47627918e-02 -5.02223313e-01 -7.33849943e-01
5.36237299e-01 -8.75874683e-02 1.15894951e-01 2.90214658e-01
2.72201300e-01 -7.20460773e-01 2.31265314e-02 1.82618603e-01
1.18019640e+00 8.38328481e-01 6.80844337e-02 9.41084102e-02
8.83586928e-02 3.16690840e-02 5.52347004e-01 3.56347471e-01
-8.95878579e-03 1.44344723e+00 6.02067932e-02 -7.78561890e-01
-1.40147460e+00 -1.33880258e+00 -7.24101737e-02 3.95625144e-01
2.46627167e-01 -3.83425653e-01 -4.35225397e-01 -6.11536860e-01
3.47464979e-01 5.91787994e-01 -6.36719227e-01 3.76613021e-01
-7.30789006e-01 -1.62965938e-01 1.65663380e-02 2.40745276e-01
1.74687743e-01 -9.63947356e-01 -7.42494464e-01 9.70745310e-02
-3.86964023e-01 -1.55918062e+00 -6.82147682e-01 -5.88285066e-02
-1.02741361e+00 -1.20936453e+00 -8.73612046e-01 -6.30679250e-01
7.10036337e-01 7.29027092e-01 1.57730377e+00 -2.20610276e-01
-2.09114343e-01 8.10428381e-01 -1.29158169e-01 -1.54094025e-01
-5.95719337e-01 -4.35716540e-01 2.64935911e-01 -1.34886160e-01
5.76702952e-02 -9.17891383e-01 -6.25003040e-01 5.74410558e-01
-5.94277442e-01 3.42500091e-01 2.59544343e-01 4.40450549e-01
1.05894005e+00 -5.36444426e-01 -9.46464464e-02 -6.42847240e-01
-2.71950752e-01 -5.21194696e-01 -8.30668449e-01 3.30788940e-02
1.47929087e-01 -5.07553704e-02 1.80939287e-01 -6.82429433e-01
-1.09945512e+00 5.30962169e-01 1.22175552e-01 -1.39828944e+00
-3.39013726e-01 -3.50719154e-01 -5.04099578e-03 4.90367040e-03
8.33671212e-01 1.55905008e-01 -1.94329917e-02 -7.11615741e-01
3.74450207e-01 -5.17781936e-02 8.29823971e-01 -7.64167011e-01
1.15743971e+00 9.14140821e-01 -4.64545004e-02 -6.72601759e-01
-8.08639824e-01 -2.79404670e-01 -7.54224122e-01 -2.52846301e-01
9.02163982e-01 -1.41063797e+00 -7.62381911e-01 2.25449651e-01
-1.26693428e+00 -5.26793540e-01 -7.90550709e-01 6.74275577e-01
-1.06325102e+00 2.09565997e-01 -3.03482294e-01 -4.55571264e-01
-2.81710345e-02 -1.19670796e+00 1.73708653e+00 -9.87910554e-02
-2.74659127e-01 -6.48180962e-01 3.03169131e-01 6.66195869e-01
-2.62447577e-02 7.00843751e-01 8.33854377e-02 -1.20536149e-01
-1.25410736e+00 4.37804945e-02 1.23608723e-01 2.25283414e-01
-9.56400335e-02 1.69327796e-01 -9.27879095e-01 -3.71072292e-01
8.69365409e-02 -4.03251052e-01 4.93614167e-01 4.44593251e-01
8.68609726e-01 -2.92239308e-01 -3.88990879e-01 9.47154641e-01
1.30407596e+00 -1.29006848e-01 6.27154648e-01 7.32519999e-02
8.74763012e-01 4.26978618e-01 6.00069523e-01 4.81677860e-01
6.23448133e-01 9.29256141e-01 6.30595028e-01 2.91743606e-01
-4.98756886e-01 -6.77079082e-01 5.38436413e-01 5.96654952e-01
-4.17659491e-01 -1.33147046e-01 -5.93021512e-01 6.89816177e-01
-1.61501956e+00 -1.15983427e+00 6.11955998e-03 2.11328387e+00
6.95474088e-01 -9.75535363e-02 3.10111165e-01 -3.03189903e-01
5.65132678e-01 1.59614518e-01 -6.96053624e-01 2.80241430e-01
-1.92034289e-01 -2.23679636e-02 3.63580734e-01 3.81309360e-01
-7.78287053e-01 8.00157607e-01 6.17943001e+00 2.75488496e-01
-8.39697301e-01 8.23636353e-02 2.47978926e-01 -4.41821754e-01
-4.82360244e-01 3.12363386e-01 -9.18500125e-01 3.70543599e-01
5.75524032e-01 5.64047806e-02 4.42357540e-01 9.13754761e-01
2.05998585e-01 -1.72012180e-01 -1.66593480e+00 1.31146371e+00
3.26426178e-01 -1.68281567e+00 6.48930296e-02 2.35820755e-01
1.12364626e+00 4.85555917e-01 -1.23677649e-01 1.41668683e-02
5.88475943e-01 -8.16906035e-01 1.25607479e+00 7.10123241e-01
7.04879582e-01 -2.22679704e-01 2.63334773e-02 4.11229163e-01
-9.24314380e-01 3.00211877e-01 -4.49281722e-01 4.39910561e-01
5.58246136e-01 2.24256024e-01 -7.32119143e-01 4.32145804e-01
1.12274504e+00 1.24171591e+00 -5.14286220e-01 1.02819288e+00
1.63565911e-02 2.01731771e-01 -6.42486274e-01 5.42024255e-01
-8.22672099e-02 -2.43253514e-01 1.00385594e+00 8.38880002e-01
6.74427927e-01 3.56778473e-01 9.19365436e-02 8.91324818e-01
-1.82185441e-01 -5.69317639e-01 -1.00008941e+00 3.04381698e-01
3.21747154e-01 1.05817258e+00 -5.74312925e-01 -3.09250414e-01
-2.36641720e-01 9.38682795e-01 1.22474477e-01 2.60615230e-01
-8.43446314e-01 5.93115926e-01 7.41584182e-01 4.97893274e-01
7.70430684e-01 -4.16250229e-01 4.31300431e-01 -1.74549341e+00
6.38812855e-02 -9.87887681e-01 3.73318419e-02 -1.44744933e+00
-1.20863891e+00 3.79671544e-01 3.17400157e-01 -1.72221768e+00
-3.40958685e-01 -2.62039751e-01 -1.92432970e-01 4.37962174e-01
-1.11338174e+00 -1.27701819e+00 -5.56258798e-01 9.25149083e-01
1.00617123e+00 -1.15721144e-01 6.00101829e-01 5.94637617e-02
2.46893149e-02 8.18786398e-03 -1.78435054e-02 -3.29844594e-01
7.10011661e-01 -1.07661235e+00 7.60372818e-01 7.12815702e-01
5.37026107e-01 2.43070647e-01 8.82399321e-01 -4.89665151e-01
-1.83854961e+00 -1.01114905e+00 3.73119086e-01 -1.36350298e+00
2.80522406e-01 -4.59724277e-01 -7.15431511e-01 1.12668610e+00
5.40653504e-02 6.57075763e-01 1.85174979e-02 -5.34946263e-01
-4.00617033e-01 4.28028293e-02 -1.28039312e+00 6.13290012e-01
1.69226491e+00 -3.14533472e-01 -8.95152092e-01 6.29239261e-01
9.32778895e-01 -1.14540792e+00 -1.07905591e+00 1.75663441e-01
5.80663741e-01 -1.33202696e+00 1.42358887e+00 -3.65129441e-01
5.32712877e-01 -6.73921347e-01 -5.11005163e-01 -1.25103283e+00
-1.18487976e-01 -1.00023246e+00 -4.47038084e-01 9.22443151e-01
-2.42423072e-01 -2.08014727e-01 8.88637185e-01 4.72958058e-01
-2.36404940e-01 -3.54817450e-01 -7.60420799e-01 -8.31290424e-01
-3.12137485e-01 -2.35940859e-01 6.61177993e-01 6.55566633e-01
-8.28935444e-01 2.12200105e-01 -6.49631441e-01 1.69203013e-01
1.06590092e+00 4.40200955e-01 1.64990294e+00 -1.09559739e+00
-4.16641921e-01 2.04078749e-01 -5.33839703e-01 -1.38929439e+00
2.05083191e-01 -7.15529740e-01 2.61413213e-02 -1.27675629e+00
1.14708379e-01 -2.85322130e-01 7.65323877e-01 1.60173714e-01
3.22189927e-01 4.02296305e-01 3.16629410e-01 5.27952254e-01
-5.77624142e-01 6.72994852e-01 1.65931308e+00 1.09881543e-01
-6.41615912e-02 -4.76760156e-02 -2.25736424e-01 9.03872252e-01
1.77353904e-01 -5.08074641e-01 -4.05720890e-01 -6.96715415e-01
-7.27967173e-02 5.37893772e-01 7.90532649e-01 -9.39837039e-01
-4.36456017e-02 -3.10078233e-01 6.90462470e-01 -1.06895792e+00
8.46701205e-01 -9.83776748e-01 1.17314744e+00 9.61225182e-02
-4.85803671e-02 2.35572547e-01 6.40451014e-02 1.03021705e+00
1.04815066e-01 1.65365562e-01 7.59330928e-01 -5.28980911e-01
-5.73637843e-01 6.62874758e-01 7.60855898e-02 3.66913885e-01
1.16634035e+00 -4.76819247e-01 -1.91177174e-01 -1.72395900e-01
-8.22931111e-01 -2.02714037e-02 1.07003474e+00 6.76714957e-01
9.39091265e-01 -1.45742047e+00 -7.79921293e-01 3.57060909e-01
5.36465980e-02 6.28632128e-01 2.77030408e-01 5.89741707e-01
-7.78483570e-01 1.63500726e-01 -2.51096308e-01 -1.23097515e+00
-1.28021407e+00 7.03670561e-01 2.45530412e-01 1.66345522e-01
-1.30114269e+00 6.10041916e-01 4.29460615e-01 -5.38302124e-01
1.96512520e-01 -4.76087391e-01 3.24824691e-01 -3.88350695e-01
2.68201113e-01 1.73108533e-01 -1.44954085e-01 -1.03320408e+00
-2.99845278e-01 8.79308462e-01 2.07326129e-01 -2.43256807e-01
1.81779170e+00 -1.77922755e-01 2.59169251e-01 3.19549233e-01
1.09171987e+00 1.51822686e-01 -2.05456805e+00 -2.77353913e-01
-4.39197302e-01 -1.01651394e+00 -5.16238511e-01 -2.41802886e-01
-1.18685257e+00 5.35862207e-01 3.73573691e-01 -2.94563115e-01
6.14564717e-01 3.67462456e-01 6.18516147e-01 3.92563283e-01
8.03420842e-01 -5.55215120e-01 4.66653913e-01 2.44330123e-01
1.30245101e+00 -1.21681321e+00 2.12340072e-01 -1.79506570e-01
-7.09606528e-01 1.01812112e+00 6.40091717e-01 -4.74943042e-01
5.19568920e-01 1.79631397e-01 -2.68577546e-01 -2.95164347e-01
-9.89678800e-01 1.93349302e-01 2.07763448e-01 9.37925279e-01
-1.45167619e-01 -1.61821961e-01 5.20459175e-01 -4.79586236e-03
-2.88990498e-01 9.71816406e-02 7.18866229e-01 8.97256434e-01
1.33875303e-03 -7.87489653e-01 -6.38394356e-01 3.88883017e-02
-2.01723695e-01 4.07384604e-01 -4.74925786e-02 8.71255040e-01
-6.46457949e-04 5.02039969e-01 4.25787121e-02 -1.40229180e-01
6.29589856e-01 -3.71166915e-01 1.01572311e+00 -6.83619678e-01
-2.75966227e-01 1.65298328e-01 1.24419093e-01 -9.79449391e-01
-1.04014874e+00 -1.04684043e+00 -8.76914740e-01 -5.29861093e-01
4.49719429e-02 -3.16601127e-01 5.10657012e-01 5.45773387e-01
4.55577075e-01 7.57107092e-03 5.74474990e-01 -1.73576987e+00
-4.91762429e-01 -5.18269956e-01 -3.63436073e-01 6.82620049e-01
4.59239572e-01 -8.74997258e-01 -5.89477062e-01 7.30584681e-01] | [8.629915237426758, -2.8107123374938965] |
c31824fa-13e4-4fb5-bfea-47a936451d66 | defense-against-adversarial-cloud-attack-on | 2306.17431 | null | https://arxiv.org/abs/2306.17431v2 | https://arxiv.org/pdf/2306.17431v2.pdf | Defense against Adversarial Cloud Attack on Remote Sensing Salient Object Detection | Detecting the salient objects in a remote sensing image has wide applications for the interdisciplinary research. Many existing deep learning methods have been proposed for Salient Object Detection (SOD) in remote sensing images and get remarkable results. However, the recent adversarial attack examples, generated by changing a few pixel values on the original remote sensing image, could result in a collapse for the well-trained deep learning based SOD model. Different with existing methods adding perturbation to original images, we propose to jointly tune adversarial exposure and additive perturbation for attack and constrain image close to cloudy image as Adversarial Cloud. Cloud is natural and common in remote sensing images, however, camouflaging cloud based adversarial attack and defense for remote sensing images are not well studied before. Furthermore, we design DefenseNet as a learn-able pre-processing to the adversarial cloudy images so as to preserve the performance of the deep learning based remote sensing SOD model, without tuning the already deployed deep SOD model. By considering both regular and generalized adversarial examples, the proposed DefenseNet can defend the proposed Adversarial Cloud in white-box setting and other attack methods in black-box setting. Experimental results on a synthesized benchmark from the public remote sensing SOD dataset (EORSSD) show the promising defense against adversarial cloud attacks. | ['Hongkai Yu', 'Yuewei Lin', 'Tianyun Zhang', 'Zibo Meng', 'Qing Guo', 'Jinlong Li', 'Lan Fu', 'Huiming Sun'] | 2023-06-30 | null | null | null | null | ['adversarial-attack', 'object-detection', 'salient-object-detection-1'] | ['adversarial', 'computer-vision', 'computer-vision'] | [ 4.41904306e-01 -8.20006132e-02 5.10844588e-01 7.85121247e-02
-4.28288639e-01 -9.16436136e-01 5.21614730e-01 -2.56948829e-01
-2.56710261e-01 3.64947826e-01 8.89679790e-02 -4.27134722e-01
1.36598840e-01 -9.91920769e-01 -9.10148680e-01 -1.10763812e+00
-1.42097965e-01 -8.22289288e-02 2.93229014e-01 -6.88505292e-01
1.19384468e-01 8.55829716e-01 -1.27624512e+00 1.14208527e-01
9.12797391e-01 9.08604860e-01 2.51042664e-01 8.14720094e-01
2.85202146e-01 6.97645605e-01 -8.40677321e-01 -2.87883788e-01
1.08969021e+00 -1.36802599e-01 -5.19983411e-01 4.98088747e-02
6.66782260e-01 -2.78155178e-01 -5.44784188e-01 1.66117191e+00
7.93469429e-01 1.13443591e-01 4.40962106e-01 -1.21016729e+00
-1.33996570e+00 5.96338034e-01 -8.44867289e-01 6.49625063e-01
-2.71554083e-01 4.32513416e-01 4.19386357e-01 -8.11155379e-01
2.66745001e-01 1.31530952e+00 3.85911494e-01 6.48056149e-01
-5.04667222e-01 -9.11747098e-01 6.43077493e-01 2.52703547e-01
-1.34536910e+00 6.82328492e-02 9.86074924e-01 -2.72630483e-01
3.79226506e-01 7.13611603e-01 5.44956267e-01 8.69540155e-01
1.92310020e-01 6.06835365e-01 1.19650567e+00 1.55336186e-02
1.80094942e-01 1.22400895e-01 -4.28830057e-01 3.60061646e-01
3.02874088e-01 5.01483500e-01 2.37097532e-01 -2.47091338e-01
4.74763632e-01 6.38704240e-01 -4.24918503e-01 2.32323408e-02
-9.23368573e-01 8.38951826e-01 1.26363003e+00 -1.00890152e-01
-2.35414147e-01 -1.12228524e-02 3.05623889e-01 3.63017470e-01
5.79061151e-01 4.72999603e-01 -2.43556485e-01 9.83602524e-01
-7.21019506e-01 4.32407767e-01 2.61616379e-01 6.93820059e-01
5.69287956e-01 7.82722354e-01 -1.60582155e-01 1.84326902e-01
5.11598624e-02 1.11440599e+00 3.31822634e-01 -3.40065479e-01
3.84276539e-01 4.33332473e-01 2.22793519e-01 -1.48705101e+00
-2.20478296e-01 -7.53916979e-01 -1.22838914e+00 8.85383427e-01
-1.52464444e-02 -5.01307011e-01 -8.90471041e-01 1.33036160e+00
6.18093133e-01 4.59506124e-01 4.34616774e-01 1.50074005e+00
8.00382435e-01 8.17993820e-01 2.15470511e-02 6.24407344e-02
1.10684109e+00 -9.81482327e-01 -3.65160078e-01 -4.70203698e-01
-2.00271636e-01 -7.81518698e-01 1.20580447e+00 2.02942193e-01
-5.98656058e-01 -4.78729367e-01 -1.14513946e+00 3.32839042e-01
-7.94621587e-01 -2.62935936e-01 5.90825677e-01 5.40797830e-01
-7.50858605e-01 2.35934019e-01 -4.06739056e-01 2.00579464e-02
6.44391119e-01 1.13310046e-01 -7.75815770e-02 1.07538328e-01
-1.46351635e+00 9.14694309e-01 2.71317989e-01 5.40399730e-01
-1.68192291e+00 -8.40337396e-01 -5.95075190e-01 2.67472953e-01
4.50075388e-01 -3.02962512e-01 5.46486080e-01 -1.45342898e+00
-9.54225481e-01 8.58413219e-01 8.25177193e-01 -7.29104459e-01
7.19599605e-01 -3.69227618e-01 -5.87813854e-01 2.73290187e-01
-2.51650214e-01 3.86091024e-01 1.54282105e+00 -1.42293870e+00
-3.51645201e-01 -2.56371289e-01 4.95910883e-01 3.62699121e-01
-2.70356894e-01 2.74086922e-01 3.71332288e-01 -1.05270267e+00
-9.28314626e-02 -9.72243309e-01 -4.28141117e-01 2.51557171e-01
-4.78799671e-01 4.88393486e-01 1.46830904e+00 -5.22272825e-01
4.22995806e-01 -2.14965224e+00 -1.88861206e-01 -6.18685931e-02
2.01764375e-01 8.64163041e-01 -3.32295179e-01 2.36834854e-01
-4.49903995e-01 3.27608645e-01 -3.62138152e-01 4.77536559e-01
-1.82884097e-01 -8.62626452e-03 -1.18633902e+00 9.28436279e-01
5.95793962e-01 7.97663569e-01 -9.31026340e-01 -2.69616872e-01
3.66182685e-01 4.10424948e-01 -3.14676613e-01 3.22337478e-01
-1.48554876e-01 5.36337554e-01 -5.29941559e-01 9.31257904e-01
1.44728291e+00 1.67917415e-01 -4.96552974e-01 1.81361109e-01
4.21789251e-02 -6.43936992e-01 -1.07603979e+00 8.97301018e-01
-1.45894721e-01 4.60872382e-01 4.97418642e-01 -9.39529777e-01
9.50071871e-01 -8.30552503e-02 9.24038216e-02 -4.05213684e-01
6.52517155e-02 -1.48581624e-01 -2.04832666e-02 -5.29776812e-01
4.13112968e-01 -1.58020437e-01 -1.53802365e-01 2.27058694e-01
-7.02446043e-01 -3.76211673e-01 -6.80489480e-01 2.88560241e-01
6.87413633e-01 -9.69082490e-02 9.39798169e-03 -3.02019626e-01
6.98329926e-01 1.12356752e-01 6.40532553e-01 8.91072392e-01
-3.83567452e-01 7.30226040e-01 -8.00421089e-02 -8.54922771e-01
-8.92058074e-01 -1.02203989e+00 -5.51614277e-02 1.24928403e+00
6.32390976e-01 5.35451591e-01 -7.00284779e-01 -9.17148709e-01
6.79683238e-02 -2.31042095e-02 -7.21831083e-01 -4.79782104e-01
-3.80410761e-01 -9.42831278e-01 6.60075307e-01 2.32056454e-01
9.80010271e-01 -1.33342612e+00 -6.99994862e-01 -1.07341006e-01
-1.37568796e-02 -9.88767207e-01 -3.75972420e-01 -7.37756789e-02
-4.33301032e-01 -1.24096036e+00 -7.13883281e-01 -6.60050035e-01
6.40355170e-01 9.19730008e-01 9.54340458e-01 5.46103239e-01
-3.60838979e-01 -1.52060106e-01 -6.33876443e-01 -9.79044437e-01
-2.19439149e-01 -3.15601140e-01 1.35541409e-01 1.71722114e-01
5.14839180e-02 -5.71507156e-01 -9.17870879e-01 1.46539852e-01
-1.57621217e+00 -3.75269771e-01 6.61321282e-01 8.01844776e-01
3.50493640e-01 3.24273556e-01 4.25814331e-01 -8.03891480e-01
2.51362234e-01 -6.76235855e-01 -8.71906757e-01 1.66534364e-01
-2.86333442e-01 -3.05929750e-01 1.05823708e+00 -6.98327899e-01
-8.91437531e-01 -4.25021257e-03 -1.87832918e-02 -9.10719097e-01
-1.52855828e-01 2.92385429e-01 -2.25590155e-01 -6.73628926e-01
9.67211604e-01 4.32168871e-01 -3.43562216e-01 -1.38279602e-01
3.86948317e-01 6.76492214e-01 7.91775167e-01 -3.39695752e-01
1.79931366e+00 1.02377832e+00 -4.96036746e-02 -5.81847668e-01
-1.06717348e+00 -1.54731110e-01 -1.20301448e-01 -4.39791195e-02
8.42338264e-01 -1.25948513e+00 -5.34388900e-01 5.81653953e-01
-8.97816002e-01 -1.29851654e-01 -1.71400577e-01 1.25186965e-01
-1.47093311e-02 4.31332916e-01 -8.56011361e-02 -8.52342784e-01
-9.41269696e-01 -7.85495460e-01 1.04662526e+00 5.08356094e-01
7.79750288e-01 -6.72934830e-01 -1.28152356e-01 1.48792177e-01
7.09975839e-01 1.10538709e+00 5.56987405e-01 -5.04196405e-01
-8.28834593e-01 -1.11221157e-01 -2.91690737e-01 5.51321030e-01
3.47517990e-02 1.31300628e-01 -1.05874789e+00 -7.18325973e-01
4.30469662e-01 -3.60113293e-01 1.00529325e+00 9.14987549e-02
1.39889574e+00 -7.49991357e-01 5.28100692e-02 1.12894380e+00
1.85920584e+00 -3.42412181e-02 7.49726474e-01 5.29679358e-01
9.81706917e-01 4.80123878e-01 8.60266685e-01 3.57061893e-01
1.33113181e-02 4.81895432e-02 1.49741995e+00 -7.83320487e-01
2.39547491e-01 9.10778195e-02 4.48237896e-01 -2.30084490e-02
6.69092387e-02 -1.23699188e-01 -7.89695442e-01 6.19021058e-01
-1.61229229e+00 -1.34677732e+00 -6.28162399e-02 1.77135277e+00
4.84936625e-01 -4.50784862e-02 -2.03531191e-01 -9.90376025e-02
8.95506978e-01 8.05088282e-01 -9.87695932e-01 -2.48078495e-01
-6.76480591e-01 1.91558778e-01 9.36224580e-01 2.64590770e-01
-1.53520095e+00 1.17095864e+00 5.66252232e+00 7.14539468e-01
-1.73682678e+00 3.41563880e-01 4.35123980e-01 -2.88391888e-01
-3.40819746e-01 1.40430816e-02 -5.73255062e-01 5.51668584e-01
2.78759331e-01 -2.47469634e-01 4.82787043e-01 1.17181242e+00
4.02759522e-01 2.33838066e-01 -2.38679871e-01 7.48868823e-01
2.82162074e-02 -1.42754793e+00 4.82593834e-01 -9.37739983e-02
1.10963869e+00 4.24560726e-01 6.50417864e-01 3.02595347e-01
5.66506982e-01 -9.82918203e-01 7.76850939e-01 2.95521021e-01
4.61323351e-01 -8.33984971e-01 7.47115195e-01 3.89307171e-01
-1.14283812e+00 -4.32880670e-01 -1.03900135e+00 -2.19879791e-01
-4.86246586e-01 3.76129925e-01 -5.11181056e-01 5.30930758e-01
1.03237963e+00 4.70631748e-01 -7.94330478e-01 6.90563262e-01
-2.82463253e-01 4.30657595e-01 -3.76192331e-02 1.92276955e-01
6.50163233e-01 -1.32734194e-01 9.65617836e-01 8.59462738e-01
-1.26994774e-01 2.42982879e-01 4.51305658e-01 9.50230598e-01
-7.55765513e-02 -1.15758516e-01 -9.36308265e-01 1.37079522e-01
3.46207470e-01 1.55103827e+00 -3.04220587e-01 -1.78420588e-01
-5.57978749e-02 7.51899064e-01 -1.23869248e-01 4.63411301e-01
-1.20853746e+00 -4.62580442e-01 1.01052856e+00 1.91999543e-02
3.80448639e-01 4.00734544e-02 -3.63679416e-02 -1.34162164e+00
6.38954937e-02 -1.19498396e+00 4.01506722e-01 -8.79025877e-01
-1.40407085e+00 8.03480983e-01 -2.36742020e-01 -1.69791830e+00
3.80749553e-01 -3.92328262e-01 -1.34708488e+00 1.09111249e+00
-2.22707319e+00 -1.53390384e+00 -6.97407961e-01 1.00380075e+00
2.88213581e-01 -3.63157153e-01 4.76961732e-01 -1.50909081e-01
-4.37084168e-01 4.77801830e-01 1.35599717e-01 2.85405815e-01
4.59525615e-01 -1.18864620e+00 6.64793551e-01 1.70916247e+00
-9.25780013e-02 4.20761317e-01 8.98037612e-01 -5.25608420e-01
-1.43422377e+00 -1.96344757e+00 -2.09724084e-01 -2.82876641e-01
6.34107351e-01 -2.95819253e-01 -1.13869226e+00 5.00237644e-01
4.01336551e-01 6.24703228e-01 3.29156786e-01 -7.21025288e-01
-4.69491065e-01 -3.26907396e-01 -1.52332354e+00 6.08987451e-01
8.20807159e-01 -5.80809951e-01 -6.58532977e-01 9.08547044e-01
1.06516695e+00 -5.82504094e-01 -4.28733081e-01 4.20556933e-01
-6.85112178e-02 -8.57390821e-01 1.31059110e+00 -7.92291880e-01
4.73823845e-01 -8.36682975e-01 -1.55069336e-01 -1.24955201e+00
-3.69950950e-01 -6.88380778e-01 1.81568325e-01 9.76541400e-01
-9.77921113e-02 -6.41296744e-01 7.26232588e-01 7.87574276e-02
-3.30686688e-01 -3.60076755e-01 -5.14998913e-01 -8.14060748e-01
4.59476739e-01 3.14952433e-01 9.61065650e-01 1.27470291e+00
-8.79125237e-01 -2.90977716e-01 -5.85491002e-01 1.26223314e+00
7.61321425e-01 4.91324157e-01 9.05518234e-01 -9.59275126e-01
-1.75109610e-01 -1.74582928e-01 -4.51890677e-01 -2.62025684e-01
1.94895998e-01 -7.62827039e-01 -1.38119504e-01 -1.01259208e+00
-6.36928007e-02 -3.87921005e-01 -3.56931478e-01 5.46452522e-01
-7.71552384e-01 5.47142386e-01 4.48679864e-01 1.48914799e-01
-2.05734044e-01 6.39737189e-01 1.26826167e+00 -7.37040043e-01
1.05149344e-01 -7.27953762e-02 -1.14599538e+00 7.12335229e-01
8.53487134e-01 -6.95025980e-01 -2.88074404e-01 -6.55211210e-01
2.29370788e-01 -3.38139415e-01 8.04970682e-01 -1.11653805e+00
1.17682479e-01 -6.19707704e-01 5.17078876e-01 -6.47677124e-01
-9.39686596e-02 -1.03594553e+00 7.96074495e-02 7.07564831e-01
-1.15272654e-02 8.38459432e-02 3.66526216e-01 6.49219453e-01
-2.53966630e-01 3.30977850e-02 1.31619287e+00 -5.10351062e-01
-9.18230236e-01 6.49202168e-01 -2.19125822e-02 1.59741238e-01
1.31489444e+00 -1.06319197e-01 -8.52566242e-01 -1.77015617e-01
-6.13748312e-01 3.67988259e-01 5.30096412e-01 5.07529199e-01
8.40598583e-01 -1.15456152e+00 -1.09241664e+00 2.69153595e-01
1.45173892e-01 2.28198886e-01 5.78184605e-01 2.99489737e-01
-8.40953171e-01 -4.43086892e-01 -6.15908861e-01 -3.82402182e-01
-1.17102551e+00 1.16030633e+00 6.79266334e-01 6.75698221e-02
-5.94042957e-01 9.46665466e-01 4.62840915e-01 -4.62847412e-01
-3.62385325e-02 -6.25152439e-02 -2.38050848e-01 -2.81225502e-01
8.34003687e-01 2.19795760e-02 6.13227068e-03 -6.24452174e-01
-5.01629055e-01 5.19593894e-01 -2.07158960e-02 4.14512694e-01
1.42417145e+00 -9.59809572e-02 -1.25419706e-01 -3.30577701e-01
9.19821024e-01 3.31041515e-01 -1.39098608e+00 -3.52181166e-01
-6.37492537e-01 -7.09495008e-01 1.14045866e-01 -5.32775998e-01
-1.58353889e+00 1.01847339e+00 1.09578717e+00 4.26743090e-01
1.36368036e+00 -5.16647041e-01 7.13154316e-01 4.76074457e-01
1.56785354e-01 -8.17993522e-01 6.57740459e-02 4.86853629e-01
1.21437180e+00 -1.62471485e+00 7.41371140e-02 -3.23659539e-01
-8.49101663e-01 7.56447554e-01 8.22379291e-01 -8.60707402e-01
6.78242624e-01 1.69294029e-01 6.16783321e-01 -2.22971365e-01
-3.66035968e-01 -3.23723555e-01 -6.89533539e-03 9.83062029e-01
-6.16291165e-01 1.04205973e-01 1.44199550e-01 1.86768115e-01
-1.51961088e-01 -6.81317210e-01 7.86237299e-01 8.15628409e-01
-5.01811743e-01 -3.48785818e-01 -8.78165901e-01 4.18138504e-02
-7.96238422e-01 -4.53299850e-01 -6.55609250e-01 7.58169711e-01
2.71242440e-01 8.65629673e-01 3.76858525e-02 -5.34996390e-01
1.77695170e-01 -6.67133570e-01 -1.81156769e-01 -6.38443112e-01
-1.27495837e+00 -2.45221898e-01 -7.25473106e-01 -3.35434049e-01
-3.89749944e-01 -2.87751406e-01 -1.00941622e+00 -4.45224732e-01
-3.59741837e-01 2.05247045e-01 4.25360084e-01 7.52054155e-01
2.66904593e-01 4.27831829e-01 1.39079571e+00 -1.05219090e+00
-9.07933056e-01 -8.34545732e-01 -6.71745658e-01 5.53158343e-01
7.12635636e-01 -3.28866959e-01 -6.60862446e-01 -1.36797860e-01] | [5.505424499511719, 7.94519567489624] |
ba26f0e0-59b0-466f-8f04-de56db6a25a7 | development-of-a-thermodynamics-of-human | 2212.12795 | null | https://arxiv.org/abs/2212.12795v2 | https://arxiv.org/pdf/2212.12795v2.pdf | Development of a Thermodynamics of Human Cognition and Human Culture | Inspired by foundational studies in classical and quantum physics, and by information retrieval studies in quantum information theory, we prove that the notions of 'energy' and 'entropy' can be consistently introduced in human language and, more generally, in human culture. More explicitly, if energy is attributed to words according to their frequency of appearance in a text, then the ensuing energy levels are distributed non-classically, namely, they obey Bose-Einstein, rather than Maxwell-Boltzmann, statistics, as a consequence of the genuinely 'quantum indistinguishability' of the words that appear in the text. Secondly, the 'quantum entanglement' due to the way meaning is carried by a text reduces the (von Neumann) entropy of the words that appear in the text, a behaviour which cannot be explained within classical (thermodynamic or information) entropy. We claim here that this 'quantum-type behaviour is valid in general in human language', namely, any text is conceptually more concrete than the words composing it, which entails that the entropy of the overall text decreases. In addition, we provide examples taken from cognition, where quantization of energy appears in categorical perception, and from culture, where entities collaborate, thus 'entangle', to decrease overall entropy. We use these findings to propose the development of a new 'non-classical thermodynamic theory' for human cognition, which also covers broad parts of human culture and its artefacts and bridges concepts with quantum physics entities. | ['Sandro Sozzo', 'Lester Beltran', 'Jonito Aerts Arguëlles', 'Diederik Aerts'] | 2022-12-24 | null | null | null | null | ['culture'] | ['speech'] | [ 1.99135467e-01 1.06608726e-01 2.85190791e-01 -7.08789006e-02
2.23806828e-01 -7.89454103e-01 1.03512096e+00 5.10632277e-01
-6.63035810e-01 8.30808163e-01 3.84190470e-01 -2.67011881e-01
-2.55477011e-01 -1.10338962e+00 -2.97661781e-01 -9.59147274e-01
2.16608614e-01 3.12302202e-01 1.19965509e-01 -5.86769104e-01
4.86371964e-01 1.30011246e-01 -1.76036155e+00 1.05522778e-02
7.59250402e-01 5.90686083e-01 1.01875186e-01 6.68884635e-01
-2.71811932e-01 9.70509410e-01 -6.11240387e-01 -4.31229889e-01
-2.03363687e-01 -1.07322299e+00 -1.24389899e+00 -1.45834908e-01
-2.30113894e-01 1.14332505e-01 -3.68726552e-01 1.31746161e+00
-8.70304927e-02 4.79709029e-01 8.97829771e-01 -7.51800656e-01
-1.34245980e+00 4.44014728e-01 1.41217664e-01 9.21228826e-02
7.95575261e-01 -2.37197310e-01 1.15715528e+00 -2.18670368e-01
6.17940605e-01 1.19311857e+00 3.74216735e-01 4.92705256e-01
-1.21734178e+00 -3.77040878e-02 -5.13717771e-01 2.62245148e-01
-1.54188323e+00 -1.59756586e-01 2.21355557e-01 -5.82366765e-01
9.89318848e-01 5.58077633e-01 1.10940754e+00 7.10594177e-01
9.00023580e-01 7.57706016e-02 1.35251546e+00 -9.08945739e-01
6.39146090e-01 3.16343129e-01 1.91935092e-01 5.46136200e-01
4.84501868e-01 4.00818735e-01 -7.76937962e-01 1.83517084e-01
3.86869162e-01 -2.49059424e-01 -1.93122119e-01 -2.36191861e-02
-1.24943209e+00 6.14868581e-01 1.62785456e-01 1.16459620e+00
-4.37403351e-01 2.91811615e-01 2.21089959e-01 2.56710947e-01
1.95961893e-01 2.98870683e-01 -2.35595703e-01 -2.73350000e-01
-4.02162939e-01 1.70482188e-01 1.20499527e+00 6.73549116e-01
9.58055317e-01 -5.21480858e-01 2.98733801e-01 1.03438377e-01
6.42214000e-01 8.75786245e-01 6.41044617e-01 -7.80001700e-01
-3.61739874e-01 4.08918023e-01 1.40596852e-01 -1.01109481e+00
-2.50163049e-01 -3.59399468e-02 -8.78779829e-01 -1.51653141e-01
1.69261679e-01 3.58978182e-01 -5.46281934e-01 1.96918869e+00
-3.29208747e-02 -5.61973006e-02 4.42436099e-01 8.01993191e-01
5.78181744e-01 6.20205104e-01 1.18553929e-01 -7.18602538e-01
1.55572712e+00 -5.23176417e-02 -1.14604616e+00 3.13758761e-01
9.10927415e-01 -6.67781055e-01 7.35613942e-01 3.06528747e-01
-8.65279138e-01 -1.76805139e-01 -9.29347456e-01 -1.22706883e-01
-9.79676843e-01 -8.02391827e-01 7.09319174e-01 1.06166172e+00
-1.10474479e+00 8.85381699e-01 -5.04751086e-01 -5.16716838e-01
-1.44581841e-02 4.50730883e-03 -2.20585093e-01 4.03418779e-01
-1.53374183e+00 1.30074203e+00 6.86434388e-01 1.99780211e-01
-1.72892287e-01 -7.65236095e-02 -5.20864844e-01 -9.95247737e-02
1.65419802e-01 -9.85106468e-01 1.00331128e+00 -1.03393531e+00
-1.53301418e+00 1.13629127e+00 -5.85385524e-02 -5.93689792e-02
8.92089605e-02 1.96862165e-02 -6.78165019e-01 1.03128813e-01
-3.99489924e-02 3.08527201e-01 4.06581163e-01 -1.19273174e+00
-5.84511757e-02 -4.26279545e-01 2.87464231e-01 2.72959143e-01
2.19391659e-02 -1.33654103e-01 2.27801010e-01 -1.00946603e-02
3.69180471e-01 -9.26247060e-01 5.09914309e-02 -6.49268568e-01
-2.57841140e-01 -5.21795392e-01 1.56787544e-01 -1.48559526e-01
1.38524139e+00 -2.35649300e+00 3.39828551e-01 5.71191788e-01
2.94292837e-01 -1.08647764e-01 2.70529151e-01 7.56411076e-01
-4.56111506e-02 4.88180786e-01 -3.72354358e-01 2.79969394e-01
6.14073217e-01 6.44494534e-01 -6.56037331e-02 7.05687106e-01
-1.75981432e-01 8.28207254e-01 -1.05123103e+00 -5.31848609e-01
3.48354042e-01 5.17563045e-01 -4.51058567e-01 -3.71128589e-01
3.59612964e-02 4.39733028e-01 -4.64200795e-01 7.71960393e-02
3.15458864e-01 -1.49470821e-01 3.64269853e-01 1.43974841e-01
-4.23534244e-01 4.08995450e-01 -1.07715952e+00 1.59934449e+00
-3.55651438e-01 7.65363932e-01 -4.24738973e-01 -8.88118386e-01
4.47490901e-01 3.57575655e-01 1.04805768e-01 -1.00192308e+00
3.19489866e-01 3.92619014e-01 4.80489880e-01 -5.93548059e-01
7.64069378e-01 -6.55274034e-01 -3.73865992e-01 6.54177070e-01
-1.38583392e-01 -5.80753446e-01 4.78616387e-01 3.26236874e-01
8.82002175e-01 1.00879043e-01 6.10731483e-01 -6.52483225e-01
8.76023650e-01 -4.09698546e-01 8.87433365e-02 8.00611019e-01
-3.69849622e-01 2.54310016e-02 3.74738783e-01 -2.11675927e-01
-1.12884212e+00 -1.36578727e+00 -7.46562362e-01 9.67228293e-01
7.20653415e-01 -6.13032758e-01 -8.50086570e-01 2.73717195e-01
-2.42781237e-01 1.01795387e+00 -6.95452392e-01 -2.44601250e-01
-1.04685269e-01 -6.76445067e-01 2.94968724e-01 -9.68846455e-02
3.99697989e-01 -1.24622929e+00 -1.07357550e+00 3.00803065e-01
-1.65893301e-01 -7.47384965e-01 1.54164702e-01 2.44998455e-01
-5.46458662e-01 -6.45978928e-01 3.01528294e-02 -1.67445410e-02
3.19486737e-01 -2.41055742e-01 1.07995462e+00 4.77425694e-01
-3.47614944e-01 6.24462366e-01 -7.89248288e-01 -3.51202041e-01
-9.72398758e-01 -4.29232925e-01 2.68816113e-01 -2.90321738e-01
6.03464544e-01 -6.04828060e-01 -4.60787624e-01 -4.98924166e-01
-1.56383169e+00 -1.83160335e-01 3.56418341e-01 5.81629813e-01
-1.39039561e-01 1.95963189e-01 1.09900855e-01 -5.49988806e-01
7.06875384e-01 -2.57720619e-01 -2.30719689e-02 3.05233002e-01
-4.93892312e-01 4.08859342e-01 5.67417920e-01 9.20643955e-02
-8.98308754e-01 -7.83383191e-01 1.41695693e-01 6.67148948e-01
-2.44775817e-01 5.20661116e-01 -7.78822005e-02 1.90451160e-01
6.32371843e-01 5.21109402e-01 5.01156524e-02 7.94031993e-02
6.45527065e-01 7.62871325e-01 4.24729705e-01 -5.72042406e-01
5.80499232e-01 5.66181660e-01 6.20418727e-01 -1.16391420e+00
-7.04688787e-01 -4.17054057e-01 -8.89181793e-01 -1.89071998e-01
1.37302041e+00 -3.66418272e-01 -1.23951864e+00 2.39015132e-01
-1.25662661e+00 3.24392468e-01 -5.91269612e-01 9.24143314e-01
-7.03069329e-01 6.88263237e-01 -5.68395436e-01 -1.31770885e+00
1.18112661e-01 -7.43348598e-01 7.32065141e-01 3.23751360e-01
-5.34248352e-01 -1.54127264e+00 2.23261192e-01 -5.08461520e-02
3.00094396e-01 -2.20273416e-02 1.06772351e+00 -6.13668680e-01
-4.64890718e-01 -1.03724331e-01 -5.44580482e-02 3.40914220e-01
1.35708153e-01 9.58363563e-02 -7.30131209e-01 7.13302270e-02
2.65825689e-01 8.71519074e-02 5.18439710e-01 7.02236321e-07
1.93539768e-01 -1.84783936e-01 -3.06549110e-02 -2.02153727e-01
1.79576969e+00 3.39404464e-01 8.73024702e-01 1.84479058e-01
3.88387769e-01 6.78821445e-01 2.31387466e-02 3.98847729e-01
1.55568406e-01 3.84765804e-01 -7.78881758e-02 4.71020639e-01
2.93268263e-01 1.30397260e-01 4.35389340e-01 1.60443473e+00
-6.14236176e-01 -2.38738909e-01 -6.85532391e-01 1.89147890e-01
-1.56698155e+00 -1.34400201e+00 -6.72526777e-01 2.22218561e+00
9.65643466e-01 -2.02899668e-02 -3.15677643e-01 9.44464747e-03
5.73316634e-01 -3.94570157e-02 1.64889619e-01 -1.03622675e+00
-3.38856995e-01 6.66928351e-01 3.90455186e-01 6.98508680e-01
-7.09878743e-01 8.51033270e-01 6.66959763e+00 6.18602037e-01
-6.28543556e-01 3.47191125e-01 -1.46272376e-01 2.85249501e-01
-6.77984536e-01 4.47257489e-01 -3.51252109e-02 6.72971427e-01
1.44232678e+00 -5.29197872e-01 6.12762153e-01 7.30438530e-02
1.28053159e-01 -7.32947648e-01 -1.21464443e+00 8.43499303e-01
-2.66190767e-02 -9.19620872e-01 1.73737064e-01 3.72315586e-01
5.32777429e-01 -5.59563518e-01 -2.83014864e-01 -2.76720617e-02
4.08917591e-02 -9.26510334e-01 8.79456162e-01 9.04911578e-01
3.18960130e-01 -6.89081013e-01 9.06272054e-01 4.32856649e-01
-9.92070019e-01 1.77690655e-01 -3.79918873e-01 -6.14967227e-01
2.63341308e-01 6.68696642e-01 -3.17864539e-03 8.79519522e-01
3.67979139e-01 2.49912694e-01 -3.22717100e-01 3.83030146e-01
-1.33202821e-01 8.51684958e-02 -3.00032407e-01 -6.86082244e-01
1.50128752e-01 -9.57408607e-01 4.98739928e-01 1.05714703e+00
1.95224985e-01 5.50878882e-01 -4.45732504e-01 1.19321346e+00
3.83762658e-01 3.42930019e-01 -5.92263997e-01 -3.59878451e-01
3.33433867e-01 8.30038309e-01 -9.99034703e-01 -6.60975456e-01
-3.31839472e-01 1.03975904e+00 -2.91742325e-01 1.04616016e-01
-6.97912514e-01 -3.12245220e-01 2.71179169e-01 -5.07229567e-01
-6.09796420e-02 -2.67371535e-01 -1.16270855e-01 -1.07187653e+00
-1.63733155e-01 -2.15511695e-01 -4.67440963e-01 -6.70250356e-01
-1.28274012e+00 2.93112546e-01 -4.67073470e-02 -7.43069649e-01
-1.55267924e-01 -8.11953664e-01 -4.71166149e-02 1.02382195e+00
-7.74886072e-01 -5.66654384e-01 1.88243747e-01 4.32266116e-01
-5.42590439e-01 3.74477208e-01 1.34267509e+00 -2.69204061e-02
-2.19717711e-01 -3.59041207e-02 5.69950044e-01 -8.05787519e-02
1.94790348e-01 -1.41466916e+00 1.73180364e-02 5.71565688e-01
3.14232081e-01 9.75659549e-01 1.07758522e+00 -5.54319918e-01
-1.54827738e+00 -7.06714094e-02 1.31441617e+00 -6.33401632e-01
1.06944907e+00 -3.90128344e-01 -9.20309544e-01 2.89424956e-01
6.02769554e-01 -5.08524179e-01 1.05744576e+00 1.66623414e-01
-2.71388292e-01 2.02659249e-01 -8.85288358e-01 5.02552807e-01
1.08955479e+00 -1.20559251e+00 -1.18902230e+00 5.42089820e-01
6.21901989e-01 1.00541919e-01 -1.11980247e+00 9.75629762e-02
8.29632998e-01 -1.45062649e+00 4.93477255e-01 -4.64846253e-01
2.58132130e-01 -1.92232653e-01 -2.52959132e-01 -9.99391079e-01
-5.00843883e-01 -4.06846404e-01 3.62516552e-01 7.99559414e-01
1.50861934e-01 -9.98316348e-01 2.15774099e-03 6.63772583e-01
-1.19179487e-02 -1.41463354e-01 -1.25679803e+00 -8.15759957e-01
5.09108901e-01 -5.27124524e-01 3.87099385e-01 1.06086552e+00
6.96995199e-01 5.33667862e-01 2.41917789e-01 -3.24526370e-01
4.58575934e-01 -2.67053396e-01 -3.39853838e-02 -1.50906956e+00
-1.67484641e-01 -6.87378109e-01 -8.00326228e-01 -7.47445524e-01
2.65937984e-01 -1.04638410e+00 3.77115756e-01 -1.24303126e+00
3.63832951e-01 6.99419379e-02 -1.95903018e-01 -2.76322156e-01
2.39677250e-01 1.71223775e-01 2.63259441e-01 5.47037721e-01
-6.39203250e-01 3.25663894e-01 1.27850115e+00 3.90043944e-01
-5.30264378e-02 -8.23212028e-01 -3.77500474e-01 6.51454270e-01
4.69805419e-01 -1.92934975e-01 -9.05767158e-02 2.38320276e-01
1.08946335e+00 -3.81090343e-02 6.45829976e-01 -8.18328142e-01
4.40114401e-02 -1.87402666e-01 -6.69309348e-02 -2.18489170e-01
3.81203741e-02 -7.78576255e-01 4.21867162e-01 6.49247825e-01
-1.68690309e-01 -2.56594807e-01 -1.88495308e-01 5.08340359e-01
1.46228522e-01 -5.87860405e-01 6.16376221e-01 -2.58007258e-01
-5.68010569e-01 -5.41346312e-01 -9.50665891e-01 -1.18838236e-01
1.00267458e+00 -4.61574614e-01 -5.36305666e-01 -1.03525206e-01
-7.36227274e-01 -5.82223356e-01 6.74935520e-01 -1.34905130e-01
2.00421572e-01 -1.13306689e+00 -3.57529253e-01 -1.45107493e-01
-1.78286582e-01 -4.70671833e-01 4.52316642e-01 1.21395671e+00
-8.34498107e-01 8.06696415e-01 -5.72472773e-02 -4.81609404e-01
-5.61055064e-01 6.66006267e-01 1.07026853e-01 2.59823143e-01
-4.39644367e-01 4.13163632e-01 2.72512972e-01 -3.91523540e-02
-6.26663327e-01 -3.99400830e-01 2.46766414e-02 1.42087922e-01
2.82317728e-01 2.66352028e-01 -2.09802672e-01 -1.24670780e+00
-6.18325353e-01 5.91065109e-01 2.29625121e-01 -5.49425662e-01
6.76909983e-01 -5.02330899e-01 -1.01268744e+00 1.35127103e+00
1.25211108e+00 7.31574325e-03 -1.17648169e-01 7.79175907e-02
8.64739716e-02 -2.18884602e-01 -1.78706169e-01 -5.58806658e-01
8.53229687e-02 5.60968578e-01 5.93937039e-01 1.29354060e+00
8.88174653e-01 2.62830198e-01 4.85216796e-01 3.87034595e-01
4.95191187e-01 -1.37967968e+00 -1.67648345e-01 6.68258548e-01
4.71846640e-01 -7.66673088e-01 5.50871454e-02 -2.98662931e-01
-4.89737913e-02 1.20863259e+00 -2.84746289e-01 -4.41414118e-03
7.12521374e-01 -9.19380561e-02 -3.65275472e-01 -2.48869732e-01
-6.59845233e-01 -5.21162033e-01 2.27723360e-01 2.45398283e-01
8.89958501e-01 4.91156578e-01 -1.14041150e+00 8.60772505e-02
-7.80473709e-01 -1.44661412e-01 9.73882377e-01 1.05383372e+00
-7.61639714e-01 -1.08847725e+00 -5.38430095e-01 7.14609325e-02
-3.47590744e-01 -3.75916839e-01 -5.28628469e-01 6.85408831e-01
7.88084924e-01 1.08618724e+00 3.80349338e-01 -1.02788940e-01
-1.89998686e-01 3.87685061e-01 1.00390255e+00 -6.36521101e-01
-2.30953380e-01 -2.02687621e-01 -6.67686537e-02 -3.09531748e-01
-1.24519885e+00 -6.14618242e-01 -1.72679293e+00 -9.12192166e-01
-4.74144429e-01 6.15574241e-01 8.76499653e-01 1.65385985e+00
-8.85997936e-02 6.59036338e-01 3.05415839e-01 -3.55819345e-01
-4.34454590e-01 -8.97445858e-01 -1.15160072e+00 6.33641601e-01
4.11616117e-02 -5.68469405e-01 -7.91752756e-01 1.50528520e-01] | [5.792938709259033, 5.018588066101074] |
aa58301a-bc44-4ff7-b9b5-5070185fc935 | transition-is-a-process-pair-to-video-change | null | null | https://ieeexplore.ieee.org/document/9975266 | https://ieeexplore.ieee.org/document/9975266 | Transition Is a Process: Pair-to-Video Change Detection Networks for Very High Resolution Remote Sensing Images | As an important yet challenging task in Earth observation, change detection (CD) is undergoing a technological revolution, given the broadening application of deep learning. Nevertheless, existing deep learning-based CD methods still suffer from two salient issues: 1) incomplete temporal modeling, and 2) space-time coupling. In view of these issues, we propose a more explicit and sophisticated modeling of time and accordingly establish a pair-to-video change detection (P2V-CD) framework. First, a pseudo transition video that carries rich temporal information is constructed from the input image pair, interpreting CD as a problem of video understanding. Then, two decoupled encoders are utilized to spatially and temporally recognize the type of transition, and the encoders are laterally connected for mutual promotion. Furthermore, the deep supervision technique is applied to accelerate the model training. We illustrate experimentally that the P2V-CD method compares favorably to other state-of-the-art CD approaches in terms of both the visual effect and the evaluation metrics, with a moderate model size and relatively lower computational overhead. Extensive feature map visualization experiments demonstrate how our method works beyond making contrasts between bi-temporal images. Source code is available at https://github.com/Bobholamovic/CDLab. | ['Hongyan zhang', 'Guangyi Yang', 'Manhui Lin'] | 2022-12-07 | null | null | null | ieee-transactions-on-image-processing-2022-12 | ['change-detection', 'video-understanding', 'change-detection-for-remote-sensing-images', 'building-change-detection-for-remote-sensing'] | ['computer-vision', 'computer-vision', 'miscellaneous', 'miscellaneous'] | [ 1.17086032e-02 -3.89441788e-01 -8.64605159e-02 -1.78718299e-01
-3.97184998e-01 -5.74813366e-01 9.34915245e-01 7.61755416e-03
-2.28617758e-01 3.28996480e-01 7.43076876e-02 -2.82621562e-01
-1.15769863e-01 -7.62477338e-01 -7.41925359e-01 -7.54605055e-01
-4.03940350e-01 -5.97623326e-02 5.11533797e-01 -6.24853559e-02
8.50839391e-02 4.09290731e-01 -1.52140820e+00 7.19773546e-02
8.21000695e-01 1.27963376e+00 3.26938957e-01 7.50087738e-01
1.74092487e-01 9.55274940e-01 -4.27287109e-02 3.92070524e-02
1.73776433e-01 -4.90087628e-01 -7.58236349e-01 2.45149303e-02
3.25100392e-01 -4.82891530e-01 -8.25084925e-01 9.86024618e-01
3.27028692e-01 2.78585464e-01 3.59404027e-01 -1.05790317e+00
-5.37870646e-01 6.71754554e-02 -6.24460757e-01 6.59628093e-01
2.69679308e-01 1.76258236e-01 9.32279348e-01 -9.34998870e-01
8.08624148e-01 9.86377537e-01 4.99090314e-01 2.30116889e-01
-1.17575121e+00 -5.10429919e-01 4.30687785e-01 7.50553429e-01
-1.32635915e+00 -3.68654341e-01 9.59441304e-01 -8.82375419e-01
8.25496554e-01 1.97686613e-01 1.08514452e+00 8.93739223e-01
1.56137168e-01 8.28524590e-01 8.96361470e-01 -2.62889743e-01
1.87789097e-01 -2.17347711e-01 -9.33605954e-02 6.84757054e-01
-2.50977367e-01 4.09054190e-01 -5.34273744e-01 2.04317763e-01
1.00627959e+00 1.30516097e-01 -6.39508069e-01 -4.19878930e-01
-1.39061868e+00 6.10976756e-01 6.17602706e-01 4.20807600e-01
-2.51150668e-01 3.11619073e-01 4.18693125e-01 4.22561169e-01
6.85400605e-01 1.29071385e-01 -3.31184506e-01 -2.92770296e-01
-9.90231693e-01 1.73638046e-01 4.67683256e-01 7.34201372e-01
7.73837686e-01 9.55306739e-02 6.76731765e-02 4.69084114e-01
2.99144804e-01 2.29256347e-01 3.90900612e-01 -9.08333480e-01
3.15852612e-01 3.95070344e-01 9.64271873e-02 -1.24072146e+00
-3.11643094e-01 -4.69890833e-01 -9.49491382e-01 1.91734016e-01
1.95378080e-01 1.37578532e-01 -7.17813492e-01 1.74684167e+00
3.62208754e-01 6.93849325e-01 1.61367422e-03 9.92201865e-01
7.30796158e-01 9.39737260e-01 -2.95296758e-01 -2.47395054e-01
1.17404425e+00 -1.00083756e+00 -7.89682329e-01 -2.51460493e-01
4.85819250e-01 -3.14657152e-01 9.00660098e-01 6.40931204e-02
-9.77268279e-01 -6.27359331e-01 -1.05854809e+00 -1.10242806e-01
-1.76480353e-01 1.35354623e-01 4.40297574e-01 -7.56321847e-02
-1.09078813e+00 6.57272577e-01 -1.27494907e+00 -3.74317914e-01
2.86105126e-01 4.16244119e-02 -3.20305109e-01 -6.77227834e-03
-1.30837953e+00 6.66067421e-01 1.80229485e-01 3.52638274e-01
-1.04643655e+00 -7.19255567e-01 -9.72728789e-01 9.08289179e-02
1.43141896e-01 -6.12595022e-01 1.32173991e+00 -1.05284035e+00
-1.45572436e+00 8.38827789e-01 -2.66574413e-01 -5.02954781e-01
8.20620298e-01 -3.37691247e-01 -4.31104660e-01 4.92252380e-01
-7.52195111e-03 6.72820687e-01 7.83097506e-01 -1.18608677e+00
-7.11464703e-01 -8.63290504e-02 8.51962939e-02 3.48002225e-01
-1.65336847e-01 -1.49661094e-01 -7.43129730e-01 -7.46437371e-01
2.78148949e-01 -8.84998322e-01 -5.51301315e-02 4.80198473e-01
1.45728625e-02 6.23871125e-02 1.04441535e+00 -9.02277350e-01
1.42525506e+00 -2.35461807e+00 4.43872303e-01 -1.40505388e-01
2.90028006e-01 1.84530362e-01 7.54988492e-02 5.18289685e-01
-2.43852302e-01 -5.44235185e-02 -5.43986499e-01 -3.16866308e-01
-2.74765730e-01 7.64203295e-02 -3.79289031e-01 6.13713384e-01
1.22634836e-01 7.54323125e-01 -1.09976840e+00 -3.13457698e-01
3.38350862e-01 4.54452723e-01 -5.10601878e-01 3.13028336e-01
-1.16921537e-01 8.70935678e-01 -2.43576095e-01 5.56797087e-01
5.81778467e-01 -4.14309442e-01 1.66703284e-01 -1.90446630e-01
-4.95234281e-01 2.01331973e-01 -1.01149774e+00 1.66537309e+00
-2.03250051e-01 1.19030237e+00 -9.18179527e-02 -1.17383051e+00
5.83264351e-01 3.56715977e-01 6.32550538e-01 -8.73056471e-01
-1.30239934e-01 1.72438145e-01 -9.88129005e-02 -7.70229459e-01
4.19173539e-01 5.26998453e-02 3.38801891e-01 2.09731549e-01
-7.11429119e-02 -6.12069890e-02 4.27351147e-02 1.72832280e-01
9.55836475e-01 4.42485422e-01 2.71408647e-01 -1.47041678e-01
3.72311622e-01 -1.26182854e-01 7.90818036e-01 3.45883280e-01
-3.96023005e-01 6.20425820e-01 3.41801047e-01 -5.56234002e-01
-8.56929183e-01 -8.90227079e-01 -1.34078339e-01 6.38617277e-01
5.44144392e-01 -3.41625184e-01 -4.20818657e-01 -3.40005428e-01
-1.33958504e-01 3.75655472e-01 -6.37462854e-01 -1.25818476e-01
-7.42506266e-01 -5.52472353e-01 2.82966644e-01 5.19627750e-01
6.61546826e-01 -8.09318364e-01 -7.83470988e-01 2.45564461e-01
-4.79647577e-01 -1.11061561e+00 -3.32723320e-01 -1.36743141e-02
-9.78005707e-01 -9.92750287e-01 -5.92571735e-01 -8.51434231e-01
4.97962892e-01 7.02140629e-01 8.92380834e-01 4.68985848e-02
-2.72700012e-01 5.22395074e-01 -4.13619876e-01 3.46856974e-02
-8.25929418e-02 -4.25737023e-01 -5.32625392e-02 1.34575620e-01
3.77716906e-02 -8.29508662e-01 -9.82457280e-01 2.05295309e-01
-1.01516581e+00 4.67902035e-01 3.84066790e-01 9.08694983e-01
4.99818891e-01 1.08143650e-01 1.88399315e-01 -3.03754658e-01
-4.18100208e-02 -6.07958078e-01 -7.88357258e-01 1.32687643e-01
-4.15029317e-01 -2.05230713e-01 2.69283623e-01 -4.10544157e-01
-1.19921017e+00 1.19461328e-01 -4.96604741e-02 -5.78508615e-01
-2.43616924e-02 8.21609735e-01 4.36390527e-02 1.19310012e-03
3.77442837e-01 5.30941069e-01 -7.25434422e-02 -4.70542759e-01
2.90848196e-01 3.14520538e-01 6.34948373e-01 -2.91027933e-01
8.43234658e-01 8.78391266e-01 -3.02038938e-01 -9.37805057e-01
-6.17549419e-01 -5.25337160e-01 -8.57707143e-01 -4.88649040e-01
8.75624895e-01 -1.32470274e+00 -3.21805954e-01 8.08400452e-01
-1.11868465e+00 -6.45618081e-01 -5.61513044e-02 7.42077589e-01
-5.52006304e-01 5.41264594e-01 -7.99918354e-01 -4.92441088e-01
-6.10785559e-02 -9.13656771e-01 9.45867836e-01 1.53872028e-01
1.68949008e-01 -1.15829027e+00 2.74598420e-01 5.76180220e-02
2.51971871e-01 3.43420655e-01 8.42925012e-01 -3.46978791e-02
-9.34474945e-01 -1.39082242e-02 -2.70635337e-01 1.93441823e-01
1.62151307e-01 2.46891618e-01 -1.02417850e+00 -4.89433944e-01
2.18692392e-01 6.15289472e-02 9.82801795e-01 6.15029275e-01
1.14304924e+00 -2.59779513e-01 -3.78984213e-01 9.15031791e-01
1.48246884e+00 4.15912777e-01 6.28183544e-01 5.37274361e-01
7.77480841e-01 5.61644554e-01 6.17383838e-01 5.37629724e-01
6.17192566e-01 8.61419678e-01 6.37497962e-01 -2.20307291e-01
-1.22720495e-01 -2.31113493e-01 4.94811356e-01 1.06823039e+00
-1.85456619e-01 -1.76656961e-01 -1.03156412e+00 7.28166103e-01
-2.11306453e+00 -1.23163033e+00 -1.37217298e-01 2.09446049e+00
7.22698927e-01 -8.16212595e-02 -1.96921006e-01 1.38956651e-01
7.22739160e-01 5.63932180e-01 -5.52598953e-01 3.52820158e-01
-1.35224268e-01 -3.89588565e-01 1.56076476e-01 5.18319011e-01
-1.33659172e+00 8.31511974e-01 5.79205370e+00 4.11170691e-01
-1.53151906e+00 9.19093192e-02 4.81118470e-01 -8.09301063e-02
-2.47596800e-01 1.57749802e-01 -3.98378015e-01 5.88814199e-01
5.39428711e-01 -2.92734414e-01 3.49314541e-01 5.15977621e-01
5.35994470e-01 -6.68595508e-02 -1.15883899e+00 1.12918425e+00
-3.87861803e-02 -1.47665286e+00 -5.84399104e-02 -5.92983738e-02
7.22056329e-01 2.92828083e-01 -4.98946495e-02 9.34242904e-02
-1.52881429e-01 -5.55057287e-01 1.07571089e+00 6.21737003e-01
8.12250555e-01 -3.26850504e-01 4.10423368e-01 2.35169351e-01
-1.66905284e+00 -1.36884153e-01 -1.35777384e-01 -2.40411654e-01
3.69156241e-01 6.24974310e-01 -2.03269914e-01 7.90333688e-01
9.61241305e-01 1.31816137e+00 -4.57310289e-01 1.15379763e+00
-2.97578096e-01 6.24486148e-01 -1.71675682e-01 5.02133489e-01
3.14063668e-01 -2.97543854e-01 6.96397483e-01 1.25015938e+00
6.57181144e-01 1.99075446e-01 3.80953662e-02 7.82902181e-01
8.24983865e-02 -2.68075675e-01 -6.15431786e-01 5.89181781e-02
5.52868903e-01 1.03665042e+00 -6.38493538e-01 -4.90534455e-01
-7.19113648e-01 1.15822053e+00 2.49906942e-01 5.17880499e-01
-1.02114046e+00 -1.80231735e-01 8.22877645e-01 2.67045479e-02
4.88018215e-01 -4.90534961e-01 1.18736021e-01 -1.57964242e+00
3.33950400e-01 -5.84040940e-01 2.84623086e-01 -7.54611552e-01
-8.69635403e-01 5.85117936e-01 -1.92305017e-02 -1.90793443e+00
-3.33692282e-02 -2.73192167e-01 -7.83347547e-01 4.78809416e-01
-1.88295865e+00 -1.05747736e+00 -7.50380993e-01 3.97368342e-01
5.90570927e-01 2.45192766e-01 3.64414066e-01 5.76700628e-01
-6.33600891e-01 7.37845451e-02 3.27413559e-01 3.96742485e-02
6.29507303e-01 -1.16841245e+00 6.52711689e-01 1.27221382e+00
6.59514591e-02 2.12836415e-01 5.89018822e-01 -4.46810633e-01
-1.31597638e+00 -1.18020630e+00 7.46378124e-01 -1.24659635e-01
8.17475438e-01 -2.93319613e-01 -1.23193598e+00 7.72033334e-01
2.03112975e-01 2.59170532e-01 4.07932490e-01 -3.76666754e-01
-2.54259735e-01 -8.58114213e-02 -5.57377160e-01 5.56948960e-01
1.08709049e+00 -8.92173231e-01 -5.40725231e-01 1.94453984e-01
8.43895853e-01 -4.61362094e-01 -8.70436251e-01 3.14898401e-01
5.15807629e-01 -1.04606771e+00 8.38359594e-01 -1.87552512e-01
5.56057096e-01 -6.48585677e-01 -1.20456755e-01 -1.33539701e+00
-5.57869136e-01 -6.12932682e-01 -3.74073923e-01 1.01985896e+00
-1.15979295e-02 -4.93900329e-01 2.96131432e-01 2.22221315e-01
-3.17565709e-01 -7.05470264e-01 -9.14058506e-01 -8.67378414e-01
-8.45386982e-02 -3.68937880e-01 2.22057641e-01 1.29769635e+00
-3.14916205e-03 1.17286339e-01 -5.38766682e-01 4.64174896e-01
3.47159386e-01 3.61103684e-01 5.39337099e-01 -1.16579795e+00
-3.57637435e-01 -4.82994646e-01 -4.41690475e-01 -1.57031965e+00
-1.69734716e-01 -6.08284473e-01 9.59806815e-02 -1.55987620e+00
1.41844332e-01 -2.55708605e-01 -2.59328038e-01 3.01401585e-01
-1.32942006e-01 -5.37256896e-02 1.19912960e-01 5.97126007e-01
-4.23857868e-01 9.11066115e-01 1.25577605e+00 -1.27414584e-01
-1.98557526e-01 -2.27746055e-01 -1.58389539e-01 5.75449169e-01
5.39334655e-01 -3.26610714e-01 -3.68229449e-01 -7.50435412e-01
2.02453822e-01 3.14716816e-01 6.78274512e-01 -1.05323911e+00
4.12138700e-01 -1.59809500e-01 1.61027342e-01 -5.62013447e-01
3.27277601e-01 -6.03907943e-01 3.81495446e-01 7.04533279e-01
-8.03526789e-02 1.64192185e-01 2.14614078e-01 8.34192216e-01
-7.31330037e-01 1.61329523e-01 9.12716866e-01 -5.71947284e-02
-1.19229805e+00 5.33703387e-01 -4.67192590e-01 -1.66589186e-01
1.03852904e+00 -2.46230394e-01 -4.38314199e-01 -4.98406708e-01
-6.91826463e-01 2.70163357e-01 5.19296885e-01 3.83667111e-01
5.26552618e-01 -1.37089622e+00 -5.39592147e-01 2.13121697e-01
2.20749855e-01 1.31170109e-01 6.56896055e-01 1.24098623e+00
-6.88150942e-01 2.67228335e-01 -2.38574892e-01 -9.10034895e-01
-1.09225869e+00 4.60263222e-01 5.44099212e-01 -2.99751256e-02
-1.25436759e+00 7.67744124e-01 6.82614267e-01 -1.09810598e-01
2.00530201e-01 -5.00395477e-01 -1.15361795e-01 1.88302770e-01
5.77752054e-01 3.15601617e-01 -9.82253402e-02 -4.96635497e-01
-4.02659029e-01 7.50855327e-01 1.75490662e-01 -1.50458023e-01
1.51372755e+00 -4.52062339e-01 1.21869976e-02 5.41839004e-01
1.22927463e+00 -4.17284936e-01 -1.85653639e+00 -5.04790246e-01
-7.00998157e-02 -5.57143807e-01 1.17808253e-01 -3.87517869e-01
-1.15110314e+00 1.10008752e+00 7.98037648e-01 2.35275581e-01
1.36440766e+00 -1.69574901e-01 6.31161809e-01 2.50926673e-01
7.00193867e-02 -8.94063771e-01 2.56792784e-01 5.75571537e-01
9.70974565e-01 -1.38489878e+00 -5.21130450e-02 -2.61592001e-01
-5.96945107e-01 1.04986548e+00 5.30170321e-01 6.22152723e-02
8.24707150e-01 5.01524508e-02 1.17656350e-01 -3.28296751e-01
-9.93702710e-01 -1.44032240e-01 2.00701192e-01 2.26081893e-01
3.33044916e-01 -1.30518034e-01 -8.10294151e-02 1.96051359e-01
2.69085854e-01 -4.63922173e-02 4.95510846e-01 1.11742938e+00
-2.92859107e-01 -7.22717285e-01 -2.76214276e-02 5.60069308e-02
-1.51152417e-01 -3.55845876e-02 -3.66116874e-02 8.22293878e-01
-6.30159825e-02 6.21954381e-01 1.99121282e-01 -4.18550640e-01
8.66713673e-02 -2.83370525e-01 2.87378371e-01 -3.55450004e-01
-1.75515994e-01 2.27121875e-01 -1.75933316e-01 -6.96046591e-01
-7.22964883e-01 -7.71602154e-01 -1.13245070e+00 -1.64960474e-01
-2.35706046e-01 2.11809669e-02 5.00728369e-01 8.75058174e-01
4.87750739e-01 6.09590828e-01 8.29523265e-01 -1.21210659e+00
-1.43423155e-01 -8.69324446e-01 -4.34418261e-01 3.94407183e-01
8.47778320e-01 -8.04081976e-01 -5.76228440e-01 5.23621202e-01] | [9.259491920471191, -0.7665409445762634] |
76dcae2d-2abb-4a88-8613-4e676d6cee4d | scgc-self-supervised-contrastive-graph | 2204.12656 | null | https://arxiv.org/abs/2204.12656v1 | https://arxiv.org/pdf/2204.12656v1.pdf | SCGC : Self-Supervised Contrastive Graph Clustering | Graph clustering discovers groups or communities within networks. Deep learning methods such as autoencoders (AE) extract effective clustering and downstream representations but cannot incorporate rich structural information. While Graph Neural Networks (GNN) have shown great success in encoding graph structure, typical GNNs based on convolution or attention variants suffer from over-smoothing, noise, heterophily, are computationally expensive and typically require the complete graph being present. Instead, we propose Self-Supervised Contrastive Graph Clustering (SCGC), which imposes graph-structure via contrastive loss signals to learn discriminative node representations and iteratively refined soft cluster labels. We also propose SCGC*, with a more effective, novel, Influence Augmented Contrastive (IAC) loss to fuse richer structural information, and half the original model parameters. SCGC(*) is faster with simple linear units, completely eliminate convolutions and attention of traditional GNNs, yet efficiently incorporates structure. It is impervious to layer depth and robust to over-smoothing, incorrect edges and heterophily. It is scalable by batching, a limitation in many prior GNN models, and trivially parallelizable. We obtain significant improvements over state-of-the-art on a wide range of benchmark graph datasets, including images, sensor data, text, and citation networks efficiently. Specifically, 20% on ARI and 18% on NMI for DBLP; overall 55% reduction in training time and overall, 81% reduction on inference time. Our code is available at : https://github.com/gayanku/SCGC | ['Shekhar S. Chandra', 'Marius Portmann', 'Gayan K. Kulatilleke'] | 2022-04-27 | null | null | null | null | ['graph-clustering'] | ['graphs'] | [ 1.09033711e-01 3.29955131e-01 -9.58945081e-02 -2.54802972e-01
-1.41620547e-01 -5.62618554e-01 6.16460145e-01 2.00341746e-01
-2.31307223e-01 4.96557266e-01 1.34474292e-01 -4.26797718e-01
-2.86328018e-01 -1.07733023e+00 -9.42793787e-01 -6.72160387e-01
-4.66937512e-01 4.66274053e-01 -1.09586157e-01 1.14659369e-01
-1.37239605e-01 4.42878276e-01 -1.05756533e+00 -3.27468775e-02
7.59585559e-01 8.17771614e-01 1.34904012e-01 7.51991689e-01
-6.37367144e-02 9.42624629e-01 -3.52627724e-01 -3.69198084e-01
1.51391104e-01 -1.96748912e-01 -7.21580863e-01 2.48719174e-02
4.38340038e-01 -1.20271735e-01 -9.57610488e-01 1.13907421e+00
5.52343011e-01 3.72591391e-02 6.07076883e-01 -1.41078627e+00
-1.04214263e+00 1.11633801e+00 -6.91664398e-01 7.45675340e-02
-2.57004827e-01 2.20506221e-01 1.17927909e+00 -5.87070704e-01
6.40768230e-01 1.26158857e+00 1.04877198e+00 4.57408369e-01
-1.64316285e+00 -8.62914860e-01 4.61552650e-01 -6.52473047e-02
-1.41197407e+00 -2.53784120e-01 7.96465933e-01 -4.04145449e-01
1.09273612e+00 -2.58004721e-02 6.89046085e-01 1.23816895e+00
-8.89850967e-03 6.61628485e-01 4.82407898e-01 5.32603376e-02
1.89614058e-01 -4.37653065e-01 4.61894721e-02 1.08792984e+00
5.14907241e-01 4.84727230e-03 -1.93116099e-01 5.04340120e-02
9.56705511e-01 4.38447475e-01 -2.94656843e-01 -1.87347144e-01
-8.65018547e-01 8.33961904e-01 1.05140352e+00 1.24657147e-01
-3.27647567e-01 7.55055964e-01 3.12108159e-01 3.79152447e-01
5.79748571e-01 2.95792013e-01 -3.59138280e-01 1.69391617e-01
-8.37285817e-01 -1.66565463e-01 9.53438044e-01 9.88744676e-01
1.10313356e+00 3.45784813e-01 6.86171204e-02 6.59070313e-01
3.30419600e-01 3.89439613e-01 4.86093462e-02 -1.08296573e+00
3.17841142e-01 8.60436440e-01 -7.20439076e-01 -1.35382116e+00
-5.94467759e-01 -8.77958894e-01 -1.42951202e+00 -3.53211239e-02
2.39123199e-02 -3.40374023e-01 -1.25318742e+00 1.84285939e+00
-7.67595768e-02 4.94092733e-01 -1.72480419e-01 6.78106368e-01
1.04966307e+00 6.68014169e-01 1.06779210e-01 3.12316030e-01
8.74569416e-01 -1.05163503e+00 -4.94485646e-01 -4.69681680e-01
4.84519929e-01 -2.77864754e-01 7.76716352e-01 2.08375081e-01
-1.10766423e+00 -2.26433501e-01 -9.89012122e-01 -1.86413080e-01
-5.35484195e-01 -1.30997658e-01 1.11847675e+00 3.48872513e-01
-1.65339243e+00 1.05341709e+00 -1.05419409e+00 -3.31106782e-01
8.39170575e-01 6.74219131e-01 -3.20262671e-01 -3.48657250e-01
-9.47510600e-01 3.88618827e-01 2.48144209e-01 9.17884558e-02
-8.90435755e-01 -8.73839498e-01 -1.12585378e+00 4.65041101e-01
3.83025557e-01 -8.11213017e-01 5.32391310e-01 -1.06781185e+00
-1.28580606e+00 4.76262391e-01 -1.03331022e-02 -5.54553747e-01
-3.04314401e-03 1.91181794e-01 -4.88452017e-01 2.10169077e-01
-4.62743975e-02 9.64627564e-01 6.13283634e-01 -1.04077685e+00
6.03404976e-02 -2.59365827e-01 -8.29684064e-02 2.47794874e-02
-4.75182861e-01 -2.23289564e-01 -8.19559157e-01 -6.90003455e-01
1.09235458e-01 -8.84117305e-01 -4.21711504e-01 1.21937744e-01
-5.26032567e-01 -1.78360552e-01 9.06290770e-01 -7.31369436e-01
1.27014732e+00 -2.01139235e+00 1.65627047e-01 5.01919627e-01
9.74955142e-01 2.24808127e-01 -4.57929701e-01 4.04100269e-01
-1.58846918e-02 5.19395173e-01 -5.19594848e-01 -4.45792794e-01
7.36469552e-02 3.32139105e-01 2.42903054e-01 4.46814984e-01
4.64927405e-01 1.29987001e+00 -8.24201167e-01 -2.67239332e-01
2.58069754e-01 8.70755076e-01 -8.16288769e-01 -1.45367667e-01
-3.12164068e-01 6.16808683e-02 -1.11939944e-01 7.04416871e-01
6.64775610e-01 -1.01488495e+00 5.69706857e-01 -2.17607245e-01
3.34730387e-01 2.78319120e-01 -9.91859734e-01 1.60796511e+00
-2.32431874e-01 8.17868054e-01 5.45281887e-01 -1.33032739e+00
7.26387739e-01 -9.33685079e-02 4.62338448e-01 -5.65900445e-01
2.25478500e-01 -5.85343763e-02 1.11878015e-01 -6.14455603e-02
7.46773332e-02 4.60527182e-01 1.47271976e-01 4.54469025e-01
5.13680875e-01 3.68954986e-01 2.86318839e-01 6.66990161e-01
1.75333381e+00 -2.64435560e-01 -2.12875456e-01 -4.26327318e-01
-1.25708908e-01 -3.09629232e-01 5.57223260e-01 6.32373631e-01
2.03687381e-02 5.26453495e-01 7.09173024e-01 -3.78918469e-01
-8.98028553e-01 -1.11695755e+00 3.73044878e-01 9.21230018e-01
5.50206341e-02 -5.84916234e-01 -6.32204652e-01 -5.64817786e-01
2.46608242e-01 2.85137355e-01 -5.16839921e-01 -2.66808510e-01
-4.65796500e-01 -8.61805022e-01 7.44441032e-01 7.62928426e-01
4.98382479e-01 -9.64459717e-01 3.77320826e-01 3.60774010e-01
1.23628512e-01 -1.12464750e+00 -3.60845864e-01 3.18753690e-01
-9.54943597e-01 -1.13543606e+00 -2.84492284e-01 -8.94967854e-01
9.75848615e-01 1.98163360e-01 1.49807918e+00 4.31404650e-01
-3.75758797e-01 2.78767884e-01 -2.25849990e-02 -1.32627681e-01
-1.42527178e-01 1.57782510e-01 -1.94059998e-01 -1.96082488e-01
2.67509580e-01 -1.05221868e+00 -7.47244120e-01 -3.89398783e-02
-6.29937410e-01 -6.92770928e-02 6.99331999e-01 8.58669758e-01
5.06996155e-01 2.83982426e-01 3.35366845e-01 -1.17321932e+00
4.96787578e-01 -6.90967619e-01 -5.98269939e-01 -7.59583190e-02
-1.00606167e+00 5.72902299e-02 8.11244488e-01 -2.60096490e-01
-6.27412438e-01 -2.76817411e-01 -3.88155528e-03 -7.73983836e-01
4.01241258e-02 7.94823349e-01 -4.48354259e-02 -2.78936118e-01
5.80585837e-01 -1.65851697e-01 2.89161772e-01 -2.69525081e-01
3.98606926e-01 1.49024457e-01 4.31513488e-01 -3.75577837e-01
7.75309861e-01 6.30382001e-01 5.80054261e-02 -7.90751219e-01
-4.38491672e-01 -3.25172633e-01 -2.98018992e-01 -1.31764829e-01
6.55152142e-01 -1.17399287e+00 -7.73174345e-01 3.36843282e-01
-9.07549918e-01 -7.93649852e-01 -5.75521328e-02 2.53174722e-01
1.32379010e-01 4.63264406e-01 -1.17582583e+00 -3.89908284e-01
-5.93345881e-01 -8.40866983e-01 8.67902517e-01 1.83952436e-01
1.94053221e-02 -1.44686234e+00 -4.30570781e-01 8.64453912e-02
6.41257167e-01 5.18089652e-01 8.86345446e-01 -4.54846323e-01
-8.15711141e-01 -5.06841019e-02 -6.34193420e-01 3.55505675e-01
4.03070077e-02 2.61578888e-01 -8.22852612e-01 -6.81839228e-01
-7.83728123e-01 -1.91372618e-01 1.26374686e+00 6.74743950e-01
1.50373650e+00 -6.27572477e-01 -5.54758132e-01 9.71534848e-01
1.59926271e+00 -2.30454251e-01 5.27190089e-01 -2.28603467e-01
1.35937810e+00 3.31454575e-01 -3.67695093e-01 1.73053712e-01
5.87508559e-01 -2.68764310e-02 6.56694591e-01 -5.27481794e-01
-4.96746421e-01 -2.48984575e-01 3.88336360e-01 1.11106277e+00
-4.15127203e-02 -7.20129430e-01 -9.96686101e-01 6.38934314e-01
-1.77285743e+00 -8.50031674e-01 -1.72166228e-01 1.60396576e+00
5.58995724e-01 2.22527385e-01 -1.50041401e-01 -1.01091549e-01
8.47101748e-01 2.99530923e-01 -8.65843594e-01 -1.64674252e-01
-2.41373122e-01 4.26550746e-01 8.30823004e-01 6.27474904e-01
-1.14073312e+00 1.06356525e+00 5.82436562e+00 7.13910460e-01
-1.11000240e+00 -2.25777589e-02 7.63327301e-01 -8.55488181e-02
-5.64313054e-01 -4.22999408e-04 -3.73673022e-01 3.93472463e-01
1.05130613e+00 1.00935645e-01 9.33818400e-01 7.44382322e-01
-5.10154665e-02 4.50463414e-01 -1.02956474e+00 9.43418384e-01
-1.88842241e-03 -1.75409067e+00 -3.40456665e-02 3.01372319e-01
8.47860277e-01 8.48120034e-01 -2.36448143e-02 3.05265903e-01
1.04321170e+00 -1.31929517e+00 2.75774330e-01 2.34590247e-01
9.26426888e-01 -7.24321187e-01 6.66577160e-01 1.93318457e-03
-1.34152317e+00 -7.31267780e-02 -5.34308434e-01 -2.73628891e-01
-2.24572107e-01 9.29276109e-01 -6.55279577e-01 4.31798995e-01
8.62932146e-01 1.18202543e+00 -5.50966740e-01 6.74901366e-01
-2.07305267e-01 1.05313826e+00 -5.83157897e-01 -1.84449449e-01
4.23719913e-01 -2.80397385e-01 4.30693805e-01 1.23035955e+00
5.82108460e-02 -1.46579385e-01 3.52788091e-01 1.18113768e+00
-6.13101304e-01 -4.14895028e-01 -6.58164859e-01 -4.80162412e-01
6.58582389e-01 1.31163847e+00 -8.48902225e-01 -3.42267215e-01
-4.27201897e-01 9.58844960e-01 6.29938960e-01 7.79571235e-01
-7.35047579e-01 -5.61768651e-01 7.24385738e-01 1.53031290e-01
4.87853706e-01 -2.01686710e-01 -2.82250077e-01 -9.88041341e-01
-1.71089560e-01 -6.46092892e-01 4.59070086e-01 -4.70240325e-01
-1.45475245e+00 5.61926603e-01 -5.14884531e-01 -6.32209539e-01
7.88358152e-02 -7.33382285e-01 -7.31626987e-01 6.77353799e-01
-1.39011371e+00 -1.20076180e+00 -5.36920071e-01 5.59391677e-01
1.01869397e-01 -1.25732841e-02 6.65024638e-01 6.67751849e-01
-7.86280215e-01 6.23089254e-01 1.07544385e-01 5.74495256e-01
3.69708389e-01 -1.42344415e+00 8.01001489e-01 9.88215268e-01
9.40617546e-02 7.46378183e-01 2.61315763e-01 -7.63664961e-01
-1.47637439e+00 -1.62974811e+00 6.15648508e-01 -1.37468204e-01
8.17140579e-01 -6.91396475e-01 -8.92461419e-01 9.32482421e-01
3.07522416e-01 2.77542770e-01 5.00675142e-01 4.20103222e-01
-6.48979425e-01 -1.68873072e-01 -9.67095912e-01 7.22124338e-01
1.52253914e+00 -5.50748467e-01 2.72363007e-01 3.33438277e-01
1.07364547e+00 -1.10001780e-01 -1.11053336e+00 4.41502333e-01
2.24084839e-01 -7.43635297e-01 8.88335705e-01 -3.85023743e-01
3.41156512e-01 -3.03385854e-01 1.83099747e-01 -1.30413628e+00
-8.55525196e-01 -6.61443353e-01 -3.69096100e-01 1.08830905e+00
5.47350705e-01 -7.02774048e-01 1.11514401e+00 4.03185427e-01
-4.00142759e-01 -6.95791245e-01 -4.51475888e-01 -6.43803596e-01
-7.81402662e-02 -2.71645099e-01 5.48664570e-01 1.35325098e+00
-1.80709913e-01 5.23998976e-01 -2.96098529e-03 3.48025709e-01
9.59368587e-01 -1.37714922e-01 5.86097896e-01 -1.49689972e+00
-3.30049962e-01 -7.87487805e-01 -4.11161602e-01 -9.57588792e-01
2.96354383e-01 -1.34912705e+00 -3.07485640e-01 -1.93479264e+00
8.42252672e-02 -3.85491192e-01 -3.90202790e-01 9.15320039e-01
2.44584121e-02 1.99067220e-01 -9.99453589e-02 -9.22080576e-02
-7.83654392e-01 4.10748720e-01 1.00468886e+00 -5.70197105e-01
-1.39234081e-01 -4.34195101e-01 -9.58864093e-01 5.89084625e-01
1.04432559e+00 -3.68291348e-01 -5.95127583e-01 -8.34051430e-01
4.08383697e-01 -2.99024731e-01 5.98900437e-01 -9.32912767e-01
5.10290861e-01 3.31675887e-01 5.30618429e-01 -3.43379378e-01
-1.96239091e-02 -5.08262396e-01 2.89871097e-01 4.88513440e-01
-7.80971348e-02 5.32171689e-02 3.34231734e-01 9.48763669e-01
-5.74303977e-02 4.66312885e-01 6.00485206e-01 -2.15111360e-01
-6.78962171e-01 7.70447850e-01 -2.18755260e-01 9.81499702e-02
5.08856416e-01 -2.44905502e-01 -6.22609317e-01 -4.61284816e-01
-6.98310256e-01 6.00723982e-01 4.71072972e-01 2.00417385e-01
5.64402342e-01 -1.18352640e+00 -6.59925044e-01 1.77421153e-01
-2.44873717e-01 4.24423695e-01 3.92410278e-01 7.23182619e-01
-6.65923715e-01 -1.43945869e-02 8.32687840e-02 -5.43144584e-01
-9.24295068e-01 5.32315314e-01 2.49323189e-01 -2.77665377e-01
-8.78802955e-01 1.17838430e+00 2.02821180e-01 -7.32655227e-01
3.90273511e-01 -2.90370107e-01 1.17946275e-01 -2.70784080e-01
5.54155819e-02 3.82442117e-01 -1.07432681e-03 -1.70190543e-01
-3.80527258e-01 2.95038313e-01 -2.88137118e-03 4.59831446e-01
1.54925013e+00 -5.59073128e-02 -5.09782553e-01 -2.57494394e-02
1.44381976e+00 -3.45448762e-01 -1.48891306e+00 -3.15310836e-01
-6.31495863e-02 2.76853561e-01 3.84179085e-01 -8.03057134e-01
-1.63809288e+00 9.20788825e-01 1.66532442e-01 3.07030261e-01
1.09510827e+00 1.49436325e-01 7.87753761e-01 4.86767411e-01
-3.80374305e-02 -9.23197329e-01 1.06903747e-01 4.75127220e-01
6.39631093e-01 -1.10321987e+00 1.53722167e-01 -5.24485886e-01
-2.28583142e-01 7.75276840e-01 7.10837066e-01 -4.16254401e-01
1.09497893e+00 5.82358241e-01 -2.93559432e-01 -7.04357088e-01
-8.25878203e-01 -1.60778105e-01 2.76179537e-02 6.25019610e-01
2.59096622e-01 2.37694621e-01 2.47218356e-01 4.28577572e-01
2.45159045e-02 -4.93397772e-01 3.56158912e-01 5.26646137e-01
-1.25386089e-01 -6.98303580e-01 2.81967700e-01 8.42786849e-01
-3.56327057e-01 -5.41337311e-01 -4.92548168e-01 7.93129623e-01
-2.12248802e-01 8.99995625e-01 4.38584477e-01 -6.65082872e-01
-7.14552775e-02 -4.01405364e-01 1.75025806e-01 -5.28392971e-01
-4.44052100e-01 1.48781538e-01 2.04463497e-01 -7.77700305e-01
-3.51790011e-01 -2.58387715e-01 -1.41151941e+00 -8.67880344e-01
-3.79148901e-01 -2.80777328e-02 4.65864807e-01 4.45653409e-01
9.83512402e-01 9.20120478e-01 3.62248480e-01 -9.18415129e-01
-7.61398301e-02 -9.23782408e-01 -5.60805857e-01 2.72827476e-01
2.52853721e-01 -4.36636239e-01 -6.68868124e-01 -2.17383474e-01] | [6.9586029052734375, 6.182433605194092] |
25b474e7-774f-4623-ba0a-802d2203881f | video-pose-distillation-for-few-shot-fine | 2109.01305 | null | https://arxiv.org/abs/2109.01305v1 | https://arxiv.org/pdf/2109.01305v1.pdf | Video Pose Distillation for Few-Shot, Fine-Grained Sports Action Recognition | Human pose is a useful feature for fine-grained sports action understanding. However, pose estimators are often unreliable when run on sports video due to domain shift and factors such as motion blur and occlusions. This leads to poor accuracy when downstream tasks, such as action recognition, depend on pose. End-to-end learning circumvents pose, but requires more labels to generalize. We introduce Video Pose Distillation (VPD), a weakly-supervised technique to learn features for new video domains, such as individual sports that challenge pose estimation. Under VPD, a student network learns to extract robust pose features from RGB frames in the sports video, such that, whenever pose is considered reliable, the features match the output of a pretrained teacher pose detector. Our strategy retains the best of both pose and end-to-end worlds, exploiting the rich visual patterns in raw video frames, while learning features that agree with the athletes' pose and motion in the target video domain to avoid over-fitting to patterns unrelated to athletes' motion. VPD features improve performance on few-shot, fine-grained action recognition, retrieval, and detection tasks in four real-world sports video datasets, without requiring additional ground-truth pose annotations. | ['Kayvon Fatahalian', 'Michaël Gharbi', 'Matthew Fisher', 'James Hong'] | 2021-09-03 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Hong_Video_Pose_Distillation_for_Few-Shot_Fine-Grained_Sports_Action_Recognition_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Hong_Video_Pose_Distillation_for_Few-Shot_Fine-Grained_Sports_Action_Recognition_ICCV_2021_paper.pdf | iccv-2021-1 | ['action-understanding', 'fine-grained-action-recognition'] | ['computer-vision', 'computer-vision'] | [ 1.56480342e-01 -1.73760474e-01 -4.52306181e-01 -4.29865241e-01
-1.10162210e+00 -7.60279775e-01 3.13662767e-01 -2.76356433e-02
-6.38156891e-01 5.38395584e-01 5.43904424e-01 5.81557274e-01
4.88539860e-02 -2.66997993e-01 -1.10111880e+00 -4.98277396e-01
-1.52412653e-01 6.15067124e-01 5.35913885e-01 -7.58343562e-02
2.02804208e-01 3.46955746e-01 -1.76153910e+00 4.84496146e-01
2.36737400e-01 1.09330213e+00 -1.05576821e-01 8.53699028e-01
4.47014987e-01 7.55898058e-01 -6.33875847e-01 -2.76460797e-01
3.89168829e-01 -2.04642758e-01 -6.79156184e-01 4.04039800e-01
1.16638601e+00 -7.17462778e-01 -7.72228122e-01 7.40915120e-01
4.15289015e-01 4.13727999e-01 5.38159251e-01 -1.24536514e+00
1.52372569e-02 3.55566777e-02 -5.64614713e-01 5.24813831e-01
7.90124118e-01 5.21128714e-01 8.61920595e-01 -8.34495783e-01
9.32919860e-01 1.21327627e+00 7.59479284e-01 5.58098674e-01
-1.07238495e+00 -5.85542202e-01 3.04118484e-01 2.68491268e-01
-1.32240105e+00 -3.97023529e-01 6.10105455e-01 -5.69761872e-01
8.78461480e-01 1.11834764e-01 9.51091170e-01 1.32728004e+00
1.17825121e-01 1.09955025e+00 6.70232832e-01 -3.35868672e-02
1.93162143e-01 -4.63405520e-01 -3.97093803e-01 6.29535437e-01
6.32670000e-02 2.41202325e-01 -1.35788000e+00 2.10725918e-01
9.26440656e-01 7.44941160e-02 -2.33884320e-01 -1.02321792e+00
-1.26444590e+00 5.08843601e-01 3.68936896e-01 -2.05229983e-01
-2.80723721e-01 3.21726710e-01 3.81133318e-01 2.16006488e-01
4.76412773e-01 4.73003775e-01 -6.22285664e-01 -7.22549617e-01
-8.42282593e-01 4.69687015e-01 5.23044646e-01 8.76660347e-01
6.83438778e-01 -4.58179936e-02 -2.77142018e-01 3.64687860e-01
5.45625426e-02 5.08166075e-01 4.44049001e-01 -1.33423924e+00
6.59030020e-01 5.21817625e-01 1.12787165e-01 -9.40705061e-01
-4.06749517e-01 -4.25323635e-01 -2.15225723e-02 1.53373107e-01
8.71886849e-01 1.17577292e-01 -1.02994514e+00 1.66732585e+00
5.74963689e-01 5.39825559e-01 -2.25132689e-01 1.48992825e+00
7.51832247e-01 1.55570984e-01 1.67802453e-01 1.30730271e-01
1.16328323e+00 -8.72188687e-01 -4.75175604e-02 -6.66480303e-01
6.59116209e-01 -6.27117991e-01 9.32933807e-01 5.63440204e-01
-8.70375752e-01 -6.36226654e-01 -8.31474781e-01 -2.59651452e-01
4.68212813e-02 5.82867414e-02 4.38137263e-01 4.15442795e-01
-5.23783267e-01 7.58008063e-01 -1.02700770e+00 -2.95488715e-01
4.86770302e-01 3.40723842e-01 -8.18494201e-01 -1.89212933e-01
-8.69966030e-01 8.02009046e-01 2.99675494e-01 1.21349096e-01
-1.26554430e+00 -6.93689644e-01 -1.18688381e+00 -3.09711725e-01
7.81556129e-01 -6.10824168e-01 1.37162209e+00 -1.11353111e+00
-1.33449328e+00 1.07893813e+00 1.85487702e-01 -3.94496053e-01
8.35850298e-01 -9.15373623e-01 6.10777549e-02 5.21719337e-01
3.27864885e-01 7.31167972e-01 1.14700758e+00 -7.14328945e-01
-7.44743586e-01 -6.04720533e-01 1.46950439e-01 5.61688542e-01
-8.26274306e-02 -3.02183658e-01 -8.82225633e-01 -6.47441328e-01
2.66922951e-01 -1.21628523e+00 -1.86708435e-01 3.65555733e-01
3.97776328e-02 -1.55505955e-01 7.38950253e-01 -5.49800932e-01
6.26691341e-01 -2.23117971e+00 4.32420135e-01 8.19212645e-02
1.20707445e-01 8.30218270e-02 -2.79536456e-01 5.73914573e-02
-3.61866131e-02 -5.49720645e-01 3.12837750e-01 -3.15532237e-02
-1.33560449e-01 4.36208576e-01 -1.36639684e-01 8.02170098e-01
3.22395056e-01 9.38127935e-01 -1.06260538e+00 -5.61202288e-01
4.49367642e-01 2.76607692e-01 -8.51213038e-01 3.37517738e-01
-4.11816448e-01 7.73355782e-01 -4.77331042e-01 7.92136967e-01
1.17037006e-01 -4.34259418e-03 -1.26787961e-01 -4.00021851e-01
2.21109509e-01 1.63474098e-01 -1.46764481e+00 2.33194566e+00
-7.41657913e-02 5.08640409e-01 -9.44601446e-02 -9.59460258e-01
5.47386944e-01 4.11366522e-02 6.83258295e-01 -6.28039002e-01
-1.73020493e-02 6.56295940e-02 -2.32657984e-01 -7.83815205e-01
3.54375899e-01 -6.52474090e-02 -3.62763315e-01 4.04881313e-03
4.33681995e-01 -1.78167894e-01 3.18242580e-01 1.08680986e-01
1.14694643e+00 8.84025812e-01 1.32414669e-01 2.32690886e-01
-7.01554269e-02 1.61735177e-01 7.46758342e-01 7.34834373e-01
-4.67871785e-01 9.56558645e-01 4.22096878e-01 -6.75138414e-01
-9.31918263e-01 -1.18462265e+00 3.51416349e-01 1.52326465e+00
2.99148262e-01 -5.05837619e-01 -5.76790035e-01 -7.47294784e-01
1.44747853e-01 -1.05084620e-01 -6.21492922e-01 -3.61573368e-01
-8.59590650e-01 -9.35343206e-02 5.02405763e-01 9.68552053e-01
1.71870410e-01 -7.50075340e-01 -7.95045018e-01 2.49410719e-01
-4.03972745e-01 -1.40124726e+00 -7.19266772e-01 1.01316556e-01
-8.14158499e-01 -1.41585577e+00 -6.58774912e-01 -4.82755870e-01
3.23437452e-01 1.71457931e-01 1.09789944e+00 -1.09273523e-01
-6.53050065e-01 7.55806267e-01 -3.65021497e-01 8.71299207e-02
1.12587668e-01 -2.22754031e-01 3.22823435e-01 -3.75476889e-02
5.13875306e-01 -2.63389289e-01 -7.77114391e-01 3.90496880e-01
-5.41067243e-01 -3.51954192e-01 6.85172260e-01 8.24903369e-01
8.53935361e-01 -4.32086051e-01 -8.49204436e-02 -4.68662888e-01
-7.63205290e-02 3.40726562e-02 -3.90072972e-01 6.17687330e-02
-2.97427606e-02 1.48427650e-01 1.23257264e-01 -6.34948850e-01
-6.26500010e-01 6.88026190e-01 8.20799917e-02 -1.06261587e+00
-3.71093959e-01 1.69020638e-01 -8.15329924e-02 -9.10591036e-02
9.67953086e-01 2.90442817e-02 8.77099633e-02 -4.14037228e-01
3.16016644e-01 2.18781710e-01 1.11198103e+00 -7.82960832e-01
8.10227931e-01 4.33788568e-01 6.94311187e-02 -7.35826731e-01
-1.24303806e+00 -7.73106098e-01 -1.00128305e+00 -5.61798990e-01
1.04599380e+00 -1.43551135e+00 -8.92811835e-01 4.11444813e-01
-7.98285365e-01 -1.34164244e-01 -4.49512362e-01 9.53972697e-01
-9.11819875e-01 3.60438913e-01 -5.01919985e-01 -4.68549758e-01
1.95585832e-01 -1.08235359e+00 1.70166337e+00 2.06901714e-01
-5.65888643e-01 -5.03445446e-01 -1.12601314e-02 7.46867537e-01
-3.78734499e-01 4.72640276e-01 1.45753533e-01 -4.48897392e-01
-6.59807801e-01 -4.40948159e-01 1.97980940e-01 3.05069208e-01
-3.05809855e-01 -2.40730524e-01 -9.66198921e-01 -4.24403608e-01
-2.83114612e-01 -8.67088854e-01 9.22225237e-01 4.03066754e-01
1.10010350e+00 -3.41215394e-02 -1.38516948e-01 8.13078523e-01
8.81557047e-01 -4.38485026e-01 2.69257754e-01 4.05911565e-01
9.64178681e-01 7.49928236e-01 1.08073616e+00 5.32402337e-01
2.57371336e-01 9.23796594e-01 4.36759800e-01 1.14172898e-01
-2.45724529e-01 -6.38853550e-01 5.43294489e-01 1.83017835e-01
-3.29842985e-01 2.74580628e-01 -8.03901911e-01 2.39979237e-01
-1.85631716e+00 -1.07432568e+00 1.18124746e-01 2.20739651e+00
8.44088018e-01 4.45313364e-01 5.07410645e-01 -8.97700265e-02
4.30268437e-01 1.78007022e-01 -8.68237495e-01 2.12417468e-01
3.57467271e-02 1.77369326e-01 6.65645242e-01 2.49180198e-01
-1.42046034e+00 9.78483200e-01 5.66992474e+00 6.09784245e-01
-1.05571222e+00 -1.23657256e-01 3.33429068e-01 -6.50103450e-01
1.51070818e-01 -1.46241784e-01 -6.75877035e-01 1.30666018e-01
4.41472411e-01 3.34209681e-01 1.05131306e-01 1.12732005e+00
1.50444120e-01 -1.74424976e-01 -1.58527601e+00 1.27133429e+00
3.53430748e-01 -1.15384316e+00 -2.27213711e-01 -2.53413260e-01
5.14057696e-01 -1.55861536e-02 1.44225154e-02 5.81060588e-01
4.17449549e-02 -9.94811356e-01 8.60308707e-01 4.05097663e-01
7.69043684e-01 -6.60954714e-01 4.02391672e-01 4.20337200e-01
-1.25420487e+00 -1.82349309e-01 -1.95822015e-01 -3.55479360e-01
-4.68548462e-02 5.71994372e-02 -8.21892977e-01 1.31499991e-01
9.12752211e-01 1.06349087e+00 -7.50330448e-01 1.01813984e+00
-3.21535140e-01 2.89000124e-01 -3.98773551e-01 1.62775412e-01
2.42919341e-01 1.93172038e-01 6.08205795e-01 9.49604571e-01
5.41383214e-02 1.33707613e-01 7.13593841e-01 1.80595160e-01
8.03163424e-02 -2.85241693e-01 -4.84095842e-01 3.63816023e-02
2.77006656e-01 8.83426964e-01 -6.13301575e-01 -1.70675963e-01
-2.70237178e-01 1.04538524e+00 2.21625313e-01 3.00371587e-01
-6.91855550e-01 2.03996971e-01 1.05187094e+00 3.69649827e-01
4.82872635e-01 -3.15419048e-01 4.74005677e-02 -1.35897708e+00
2.89557904e-01 -1.21787250e+00 7.06596673e-01 -7.30470836e-01
-1.03913820e+00 -2.14175265e-02 1.04147896e-01 -1.77290046e+00
-7.51449645e-01 -7.34660387e-01 -1.60053447e-01 3.06136847e-01
-9.05510008e-01 -1.17981708e+00 -3.82985443e-01 7.22152293e-01
9.62020040e-01 4.28940468e-02 5.89370549e-01 1.84365347e-01
-2.03595519e-01 4.86724854e-01 -3.93796861e-01 3.71327311e-01
1.08139253e+00 -1.12542093e+00 2.06049368e-01 7.87714541e-01
5.63392520e-01 1.44884601e-01 9.78818476e-01 -6.28128111e-01
-1.68108714e+00 -9.18605149e-01 4.85476136e-01 -9.69742477e-01
5.00002444e-01 -4.58839923e-01 -6.19892120e-01 6.07571840e-01
-4.97955024e-01 4.95579690e-01 4.49940205e-01 3.24324220e-01
-6.21109903e-01 -1.64797500e-01 -7.02129841e-01 4.50790584e-01
1.48068511e+00 -7.03412116e-01 -7.61730671e-01 4.91229415e-01
3.41654181e-01 -1.06665325e+00 -7.54414976e-01 2.92988628e-01
8.32701325e-01 -7.00087130e-01 1.22481310e+00 -1.05984688e+00
4.84328568e-01 -3.59417737e-01 -1.44816801e-01 -1.14336598e+00
-1.91883743e-01 -4.20743138e-01 -1.66942134e-01 6.07196212e-01
-5.59523515e-02 2.69667119e-01 1.24601460e+00 7.12025166e-01
-2.04757694e-02 -5.59042454e-01 -1.19053483e+00 -8.98230791e-01
-2.41570324e-01 -5.40624261e-01 1.07667871e-01 6.11596107e-01
-1.59106050e-02 2.84900367e-01 -5.85143864e-01 1.81155056e-01
6.92941129e-01 1.95574000e-01 1.26954317e+00 -1.12543297e+00
-6.85968637e-01 -1.72653735e-01 -1.18563461e+00 -1.40536511e+00
1.28333136e-01 -4.57637399e-01 3.83593440e-01 -1.10160768e+00
7.68210143e-02 2.08871976e-01 -3.10898572e-02 4.54209030e-01
-8.15879256e-02 7.10912228e-01 3.34326148e-01 2.28037268e-01
-1.08753085e+00 3.87475669e-01 1.33394063e+00 -7.73769245e-02
-3.59533988e-02 1.97504070e-02 -1.98589981e-01 9.93023992e-01
2.45084077e-01 -6.04465425e-01 -3.78130615e-01 -5.38962305e-01
6.95033446e-02 4.24211293e-01 7.51895547e-01 -1.19166815e+00
2.04166234e-01 -2.12597668e-01 9.44918454e-01 -4.69758272e-01
7.74206102e-01 -5.67236245e-01 -4.76448350e-02 5.05224288e-01
-4.59975392e-01 -2.08906084e-01 8.05836171e-02 8.05026233e-01
-2.54379869e-01 4.61403057e-02 5.41848719e-01 -2.16081098e-01
-1.29620433e+00 5.12082577e-01 -5.52594662e-02 6.48928106e-01
1.20198691e+00 -5.98538995e-01 -1.58343330e-01 -4.42684352e-01
-1.11788380e+00 2.86117494e-01 4.08114314e-01 7.90201247e-01
6.75236106e-01 -1.16847885e+00 -6.97270751e-01 2.83367485e-01
4.71804500e-01 2.58437097e-01 4.65109646e-01 8.71233344e-01
-4.78129834e-01 1.01898819e-01 -2.95857847e-01 -1.18127286e+00
-1.36666036e+00 3.35506797e-01 3.06436539e-01 4.05787788e-02
-8.26110959e-01 1.23079967e+00 2.53022432e-01 -1.70025960e-01
5.21981657e-01 -3.72362792e-01 9.58487019e-02 1.78780660e-01
5.03527582e-01 1.84891909e-01 4.37207334e-02 -7.89716601e-01
-5.41853487e-01 8.20604086e-01 -5.29700667e-02 -5.00408038e-02
1.21347606e+00 -3.07486746e-02 6.38006866e-01 3.59505445e-01
1.14919937e+00 -2.13628575e-01 -2.11824512e+00 -2.84682363e-01
-1.45677879e-01 -9.82096612e-01 -1.10903434e-01 -4.95700032e-01
-9.49494660e-01 8.06983471e-01 6.41271412e-01 -5.72757244e-01
7.72494614e-01 2.71650374e-01 7.03201234e-01 4.77419347e-01
4.57932115e-01 -1.51638198e+00 6.60393238e-01 4.13770169e-01
7.61111081e-01 -1.45699024e+00 1.56495750e-01 -2.22557127e-01
-9.11293447e-01 1.10813344e+00 1.04460239e+00 -2.83421397e-01
3.29471618e-01 1.11939564e-01 1.74066767e-01 -1.46219358e-01
-6.65269196e-01 -3.83009404e-01 5.65955877e-01 8.61486137e-01
1.64291412e-01 -7.28409961e-02 1.85816690e-01 3.75892341e-01
-3.24207872e-01 4.25644927e-02 1.99538022e-01 1.05374157e+00
-5.45041740e-01 -6.96523011e-01 -3.86323899e-01 2.83553004e-01
-2.58297324e-01 3.57615560e-01 -4.18911487e-01 8.62308323e-01
3.29150379e-01 6.07046008e-01 3.03153187e-01 -4.72347498e-01
6.69299960e-01 -2.17492227e-02 8.59668672e-01 -8.93429220e-01
-4.79591161e-01 2.29095489e-01 2.70224273e-01 -1.11686552e+00
-5.02710223e-01 -9.38959956e-01 -1.19388723e+00 2.67765801e-02
-1.99611438e-03 -1.70677483e-01 2.41541803e-01 1.23647809e+00
7.27899894e-02 2.48554602e-01 1.80065811e-01 -1.21103013e+00
-7.23552585e-01 -7.04365253e-01 -5.18931866e-01 9.69632745e-01
4.87748355e-01 -1.06529009e+00 -2.03044098e-02 3.20057303e-01] | [7.881523132324219, 0.13941660523414612] |
dab71276-9bbe-41a6-8eec-4c3e5636743b | a-robust-design-of-time-varying-internal | 2304.06451 | null | https://arxiv.org/abs/2304.06451v1 | https://arxiv.org/pdf/2304.06451v1.pdf | A robust design of time-varying internal model principle-based control for ultra-precision tracking in a direct-drive servo stage | This paper proposes a robust design of the time-varying internal model principle-based control (TV-IMPC) for tracking sophisticated references generated by linear time-varying (LTV) autonomous systems. The existing TV-IMPC design usually requires a complete knowledge of the plant I/O (input/output) model, leading to the lack of structural robustness. To tackle this issue, we, in the paper, design a gray-box extended state observer (ESO) to estimate and compensate unknown model uncertainties and external disturbances. By means of the ESO feedback, the plant model is kept as nominal, and hence the structural robustness is achieved for the time-varying internal model. It is shown that the proposed design has bounded ESO estimation errors, which can be further adjusted by modifying the corresponding control gains. To stabilize the ESO-based TV-IMPC, a time-varying stabilizer is developed by employing Linear Matrix Inequalities (LMIs). Extensive simulation and experimental studies are conducted on a direct-drive servo stage to validate the proposed robust TV-IMPC with ultra-precision tracking performance ($\sim 60$nm RMSE out of $\pm80$mm stroke). | ['Zhen Zhang', 'Yue Cao'] | 2023-04-13 | null | null | null | null | ['robust-design'] | ['miscellaneous'] | [ 1.43719181e-01 6.86852038e-01 -3.98487151e-01 5.66711009e-01
-1.93930436e-02 -6.66837871e-01 3.93368274e-01 -2.64961928e-01
-1.44251334e-02 8.13955843e-01 -5.21232188e-01 -7.65657842e-01
-6.47078574e-01 -1.18635632e-01 -6.44181788e-01 -8.98128688e-01
3.38037342e-01 -1.90117314e-01 -2.80293282e-02 -3.80361855e-01
-9.46214702e-03 3.61507297e-01 -9.10511076e-01 -9.49497283e-01
9.49095130e-01 9.83918965e-01 3.65874648e-01 4.42925692e-01
7.96146452e-01 4.02513057e-01 -3.52507353e-01 4.69544262e-01
3.67587507e-01 -4.25668776e-01 -1.30191103e-01 4.27390665e-01
-2.58979440e-01 -7.15050846e-02 -5.14455020e-01 1.36180043e+00
5.57502389e-01 3.80197555e-01 6.50268316e-01 -1.38354743e+00
-3.53880495e-01 6.45458028e-02 -3.60253423e-01 -1.65605813e-01
2.12212168e-02 2.41257936e-01 2.78424472e-01 -9.43509042e-01
4.63256568e-01 1.21351051e+00 5.20543933e-01 4.24739957e-01
-1.25456262e+00 -5.54852486e-01 1.75509587e-01 -9.49674770e-02
-1.62177241e+00 -3.71898770e-01 5.71812868e-01 -5.39648414e-01
2.73654073e-01 5.13260067e-01 3.94495696e-01 3.78503084e-01
5.57731628e-01 2.71895170e-01 9.72925723e-01 -2.71549463e-01
-6.61060400e-03 2.31266484e-01 1.24275908e-01 5.44158101e-01
6.71295643e-01 1.89685971e-01 3.66076678e-01 8.46224800e-02
1.30665386e+00 -1.90929040e-01 -6.74447060e-01 -6.13322020e-01
-1.08830440e+00 4.04579192e-01 4.33849603e-01 2.72588909e-01
-3.03978026e-01 -1.23060696e-01 3.40000302e-01 5.05032897e-01
-4.94342521e-02 4.19027179e-01 -3.11657637e-01 2.12285951e-01
3.41374725e-02 1.48919821e-01 7.17311680e-01 1.61350393e+00
1.30746230e-01 9.84452665e-01 2.75225818e-01 5.17080903e-01
4.30068165e-01 9.42054808e-01 3.26610386e-01 -8.93833458e-01
3.15356284e-01 7.02645481e-01 8.25120687e-01 -1.05050075e+00
-3.75265658e-01 -5.27464747e-01 -1.05651593e+00 4.23540205e-01
2.07461387e-01 -6.44033492e-01 -5.70964634e-01 1.60084212e+00
5.56377053e-01 -4.60334152e-01 2.21523821e-01 1.00686109e+00
-2.24571913e-01 1.01918936e+00 -5.08016527e-01 -9.61774766e-01
1.00735009e+00 -4.53308493e-01 -1.41388977e+00 -2.06969306e-01
1.98024809e-01 -5.42952180e-01 7.42719293e-01 1.73203140e-01
-8.95563424e-01 -6.62922978e-01 -1.36990321e+00 5.52099645e-01
3.58152986e-02 6.66258991e-01 -8.06040704e-01 9.32205915e-02
-5.56737125e-01 4.51299220e-01 -6.74661577e-01 -1.17468014e-01
-7.94670105e-01 2.22305372e-01 -1.41248360e-01 7.65968263e-01
-1.46391869e+00 1.46672571e+00 6.60555482e-01 7.31379032e-01
-4.48629797e-01 -6.25820875e-01 -5.80774128e-01 -5.18378019e-01
1.07957554e+00 -4.61874545e-01 1.00063217e+00 -7.61942863e-01
-2.12820840e+00 -5.67277931e-02 -1.19679108e-01 -1.59527734e-01
3.84942114e-01 -3.53814177e-02 -3.46739441e-01 1.58750728e-01
-2.34501004e-01 -4.42862988e-01 1.11306286e+00 -1.01813543e+00
-2.69365460e-01 -2.42661592e-02 -2.42406771e-01 1.97332427e-01
-1.42709389e-01 -3.17645550e-01 -1.34182442e-02 -4.83395129e-01
3.32219362e-01 -1.27523959e+00 -3.68280262e-01 1.74210340e-01
-3.35822493e-01 9.01528671e-02 1.12951541e+00 -3.97437125e-01
1.57152700e+00 -2.11984587e+00 4.93863910e-01 3.00096929e-01
-2.23087117e-01 8.26304197e-01 4.70050395e-01 6.57808423e-01
4.35475446e-02 -8.77816230e-02 2.16807574e-01 4.27236080e-01
1.51248639e-02 8.75550788e-03 -2.91207850e-01 7.74150312e-01
1.61603063e-01 3.29078227e-01 -5.00121951e-01 -4.68413606e-02
6.66246295e-01 4.03850496e-01 2.08236158e-01 -9.87329930e-02
7.65896589e-02 6.04211748e-01 -9.05187011e-01 2.87659049e-01
4.73787278e-01 -4.17400412e-02 2.50183076e-01 -2.78274029e-01
-8.16157460e-01 -4.75116163e-01 -1.77677190e+00 6.92599177e-01
-4.93730634e-01 4.60118681e-01 1.18275523e+00 -9.53881919e-01
1.26794732e+00 7.52571166e-01 6.12349994e-02 -4.75921810e-01
7.95060873e-01 5.61786830e-01 5.39609455e-02 -2.57380694e-01
-1.03867054e-01 -2.10020795e-01 1.93028405e-01 -2.72127420e-01
-2.45128125e-01 -5.06926417e-01 -1.32224843e-01 -1.30610809e-01
5.16436934e-01 -1.00157946e-01 8.31842899e-01 -8.25021684e-01
1.52997506e+00 5.89273795e-02 1.05153799e+00 -1.49443880e-01
-3.65412205e-01 -2.66166329e-01 2.61573672e-01 2.53653914e-01
-1.32571602e+00 -2.15422869e-01 -2.24061653e-01 1.62723199e-01
4.85831648e-01 2.21285298e-01 -4.90987241e-01 3.54228973e-01
1.54291749e-01 5.88800669e-01 -2.08028331e-01 -7.15304196e-01
-4.38524932e-01 1.76105484e-01 -4.33592536e-02 4.42497253e-01
2.37720370e-01 6.73520789e-02 -6.85740411e-01 6.81167543e-01
3.14543486e-01 -9.12026942e-01 -3.52424800e-01 -7.38923848e-02
-9.15684521e-01 -1.26332426e+00 -7.50028491e-01 -9.10604775e-01
8.97612512e-01 4.81441990e-02 -2.15164453e-01 -1.80749774e-01
-3.99650354e-03 4.54040289e-01 -7.75734521e-03 -6.76762044e-01
-4.71592396e-01 -4.60516810e-01 7.63257027e-01 1.69868678e-01
-6.72500670e-01 2.88756430e-01 -4.15501386e-01 1.09265041e+00
-5.41538537e-01 2.17620179e-01 1.93156153e-01 1.07913387e+00
6.27298057e-01 2.07552224e-01 8.96558225e-01 -1.86783895e-01
6.87802196e-01 1.99140921e-01 -1.62796283e+00 1.31001145e-01
-8.14615726e-01 -2.61314124e-01 1.37860405e+00 -7.22985804e-01
-1.10324395e+00 2.43488044e-01 2.79897898e-01 -8.89252782e-01
5.72859347e-01 5.40458739e-01 -4.30912852e-01 -2.82619178e-01
3.78958642e-01 2.07828209e-01 9.91243064e-01 -2.73743391e-01
3.19981009e-01 9.15907621e-01 6.04753137e-01 -1.31857976e-01
1.11919570e+00 -4.44598757e-02 7.53554940e-01 -8.81068289e-01
-1.67718649e-01 -3.20087433e-01 -4.61889952e-01 -4.04409617e-01
3.52748066e-01 -9.98390257e-01 -9.89095509e-01 6.57310247e-01
-7.82352865e-01 -2.94222772e-01 -1.36556625e-01 9.05134737e-01
-8.55476260e-01 -7.54855350e-02 -6.46204591e-01 -1.54953325e+00
-3.56635392e-01 -1.12928414e+00 3.05256158e-01 2.55332142e-01
-1.55312613e-01 -9.83754754e-01 -2.44385526e-01 -1.76225781e-01
6.49371088e-01 7.06173241e-01 6.04020238e-01 1.64233968e-01
-4.43830341e-01 -5.76699972e-01 3.32604349e-01 3.90790075e-01
-1.84424445e-02 7.89732933e-02 -2.59711474e-01 -9.87218022e-01
6.13530338e-01 7.73628950e-02 -2.75368601e-01 3.99273694e-01
1.82853844e-02 -6.15268588e-01 -5.02923548e-01 1.53524697e-01
2.00162840e+00 7.75560915e-01 3.41196984e-01 5.07678211e-01
6.79422915e-01 5.04271209e-01 1.12391233e+00 2.97357917e-01
-4.02233973e-02 5.96305370e-01 5.77240825e-01 -4.88116108e-02
5.62734365e-01 9.77143925e-03 4.55424070e-01 6.99352443e-01
6.28179535e-02 6.50136918e-03 -5.45872033e-01 6.45723462e-01
-1.74547064e+00 -2.69601762e-01 -7.02989399e-01 2.27394891e+00
7.44525313e-01 -2.27771819e-01 -2.51935869e-01 7.32617974e-01
1.37985659e+00 -3.57998192e-01 -8.37499738e-01 -4.95889813e-01
9.83167589e-02 -4.01996404e-01 9.78336573e-01 6.73362017e-01
-7.49583662e-01 8.41065869e-02 4.54739141e+00 8.12140942e-01
-1.47299099e+00 -4.86754984e-01 -5.30767888e-02 2.49404564e-01
4.98124480e-01 1.68509871e-01 -7.75459766e-01 4.94050950e-01
1.15256751e+00 -1.06453669e+00 2.23885506e-01 1.04653084e+00
9.23267782e-01 -1.18776988e-02 -4.86833602e-01 6.15535915e-01
-2.86960840e-01 -6.32148027e-01 -6.47881806e-01 -2.11925969e-01
7.96871603e-01 -9.98467624e-01 2.83356965e-01 1.08966604e-01
-2.56309897e-01 5.66932112e-02 8.73515129e-01 5.01854420e-01
7.45975554e-01 -6.11056745e-01 6.25547886e-01 6.91819251e-01
-1.34341979e+00 -5.96186101e-01 -3.73061270e-01 -1.03567734e-01
5.25058448e-01 2.51732856e-01 -4.11302537e-01 9.81089592e-01
-2.33495221e-01 6.09670103e-01 -1.75407946e-01 8.19539785e-01
-3.93232703e-01 3.04578751e-01 -3.40627939e-01 -5.80911100e-01
2.00401723e-01 -5.25994420e-01 1.18948078e+00 3.14863771e-01
2.19917044e-01 4.84116107e-01 7.86550567e-02 8.87538075e-01
7.36141980e-01 8.43491182e-02 -5.41581452e-01 1.19733870e-01
4.16557431e-01 1.01198316e+00 3.16049121e-02 -5.86988568e-01
1.38645053e-01 4.43875223e-01 -4.73516226e-01 3.07357311e-01
-9.20626879e-01 -1.12352991e+00 4.47003722e-01 3.36438008e-02
9.93927941e-02 -5.06190658e-01 -5.34950383e-02 -6.71007216e-01
1.84904695e-01 -6.72312856e-01 -2.13764489e-01 -7.07797408e-01
-7.26320088e-01 2.78923541e-01 -4.61198054e-02 -2.18933749e+00
-7.15926886e-01 -4.10106689e-01 -3.67706299e-01 1.17727983e+00
-1.13637245e+00 -4.31670368e-01 3.04675579e-01 4.33591366e-01
3.08303207e-01 1.87512755e-01 3.23073417e-01 2.44896770e-01
-9.88202631e-01 -5.23593947e-02 9.64238524e-01 -3.68001342e-01
7.09508061e-01 -8.88313413e-01 -4.65745032e-01 1.04066241e+00
-1.29683578e+00 1.01711321e+00 1.11279273e+00 -4.68391538e-01
-2.07355380e+00 -1.45858014e+00 5.59004843e-01 3.19074094e-01
1.04701567e+00 4.79867421e-02 -1.09056079e+00 5.59349298e-01
1.36349171e-01 2.11997211e-01 -5.75221658e-01 -1.16480589e+00
4.07079130e-01 -3.49809170e-01 -1.14754045e+00 6.78208172e-01
4.11124378e-01 -3.90579402e-01 -6.00985110e-01 -2.48433679e-01
8.40739667e-01 -7.33814716e-01 -1.22286868e+00 6.82185948e-01
4.38001901e-01 2.17798114e-01 5.83457053e-01 -4.30226102e-02
-4.14106667e-01 -1.17092907e+00 2.89262235e-01 -1.64695525e+00
-2.73845166e-01 -1.08469284e+00 -2.29925349e-01 1.05317199e+00
2.02351838e-01 -1.02224374e+00 7.08754137e-02 5.57146430e-01
-1.11493081e-01 -7.03512371e-01 -8.90067339e-01 -1.42242360e+00
-1.10622413e-01 3.04375917e-01 -8.60381499e-02 7.12470591e-01
5.77857196e-01 5.19803405e-01 -5.09393930e-01 8.96848321e-01
4.33531016e-01 -3.39982420e-01 8.74777436e-01 -8.21770251e-01
7.40206689e-02 1.34547269e-02 -1.03591263e-01 -9.11937892e-01
-5.47193736e-02 -8.54427218e-02 2.54699826e-01 -1.69541991e+00
-7.86221683e-01 -1.57643303e-01 -9.12582651e-02 1.09133206e-01
-2.38701805e-01 -2.79390126e-01 1.83126450e-01 2.90322483e-01
-1.31204203e-01 7.05130696e-01 1.47032070e+00 -1.86704889e-01
-5.22780836e-01 3.58060598e-01 1.48772210e-01 5.96815169e-01
8.85883868e-01 1.68427397e-02 -9.32459116e-01 -1.71783969e-01
-3.65892291e-01 9.15091097e-01 1.78425118e-01 -1.39524412e+00
1.97047174e-01 -3.27672988e-01 -1.72820702e-01 -8.00154269e-01
6.46175295e-02 -1.32895339e+00 8.65530789e-01 1.30156684e+00
-1.80620164e-01 -2.80950814e-01 9.71841067e-02 7.62296140e-01
-3.95301044e-01 -1.31094530e-01 1.14510655e+00 4.89546955e-01
-3.15025836e-01 -1.81171194e-01 -8.19088817e-01 -5.40946364e-01
1.24989247e+00 -3.54677409e-01 -2.59704471e-01 -3.61082941e-01
-6.00602925e-01 7.09499180e-01 2.86402315e-01 2.42782623e-01
1.86615765e-01 -1.16408527e+00 -8.43216181e-02 7.87355378e-02
-1.49375409e-01 -2.67905921e-01 3.94814223e-01 1.20495605e+00
9.33564454e-03 1.16987634e+00 -1.97280318e-01 -5.52003443e-01
-1.33752441e+00 9.17794406e-01 4.89758641e-01 4.06319588e-01
-3.49703997e-01 -4.86967675e-02 -2.09380522e-01 -1.84402689e-01
-5.17923087e-02 -5.73714077e-01 -1.03472807e-01 -1.05747029e-01
2.69196063e-01 6.09798372e-01 -1.03896640e-01 -5.76518297e-01
-5.71240373e-02 9.41856503e-01 5.99033058e-01 -5.08291163e-02
8.40757787e-01 -6.43228948e-01 2.10039079e-01 6.01566732e-01
9.62599695e-01 -4.57510531e-01 -1.56713295e+00 -1.45417601e-01
-3.09109747e-01 -2.44021669e-01 -2.56472170e-01 -2.98723340e-01
-7.97321856e-01 3.70006055e-01 5.54009974e-01 3.04013163e-01
8.26536953e-01 -9.52754855e-01 3.56617272e-01 3.45804065e-01
6.62281752e-01 -1.30639756e+00 -2.75312811e-01 8.32669139e-01
1.18980980e+00 -5.67876399e-01 2.53694296e-01 -7.45409787e-01
-5.93080878e-01 1.12048650e+00 8.78513753e-01 -6.82036817e-01
5.43116570e-01 2.29517877e-01 -8.53856802e-02 5.76183498e-01
-7.43998647e-01 1.79142803e-02 2.92526126e-01 2.55298018e-01
1.78127170e-01 -2.08811373e-01 -1.25600910e+00 2.64381230e-01
7.44685769e-01 2.55729053e-02 8.85379076e-01 1.16432405e+00
-7.05660820e-01 -7.39847302e-01 -8.99240732e-01 -1.85132965e-01
-3.71919274e-01 9.01886523e-01 6.06845841e-02 1.33067918e+00
-3.19230616e-01 9.33639288e-01 -1.37232825e-01 -6.90261349e-02
1.20046687e+00 -1.71135336e-01 2.33703274e-02 -1.85172170e-01
-1.72177866e-01 5.23687959e-01 1.69438068e-02 -3.29706609e-01
-1.17985711e-01 -1.54448241e-01 -1.35859060e+00 1.74035147e-01
-1.17694438e+00 3.42409611e-01 6.96336448e-01 2.68725127e-01
3.27369273e-01 6.11677289e-01 9.41703916e-01 -4.85719621e-01
-1.21494114e+00 -1.01773512e+00 -6.13275051e-01 -2.54062973e-02
6.00581348e-01 -1.18803084e+00 -9.11501527e-01 -1.72125828e-02] | [5.3046488761901855, 2.457193613052368] |
a8d0d04b-980c-4a82-b4d7-fe09cab34f63 | prediction-of-icd-codes-with-clinical-bert | 2008.10492 | null | https://arxiv.org/abs/2008.10492v1 | https://arxiv.org/pdf/2008.10492v1.pdf | Prediction of ICD Codes with Clinical BERT Embeddings and Text Augmentation with Label Balancing using MIMIC-III | This paper achieves state of the art results for the ICD code prediction task using the MIMIC-III dataset. This was achieved through the use of Clinical BERT (Alsentzer et al., 2019). embeddings and text augmentation and label balancing to improve F1 scores for both ICD Chapter as well as ICD disease codes. We attribute the improved performance mainly to the use of novel text augmentation to shuffle the order of sentences during training. In comparison to the Top-32 ICD code prediction (Keyang Xu, et. al.) with an F1 score of 0.76, we achieve a final F1 score of 0.75 but on a total of the top 50 ICD codes. | ['Haifeng Lin', 'Gaurav Desai', 'Brent Biseda', 'Anish Philip'] | 2020-08-24 | null | null | null | null | ['text-augmentation'] | ['natural-language-processing'] | [ 2.03771561e-01 7.12355494e-01 -3.99427980e-01 -4.01131958e-01
-9.91056442e-01 -2.82669187e-01 4.46553200e-01 7.40354061e-01
-4.55983132e-01 6.83050156e-01 7.02796638e-01 -6.11866593e-01
-1.43437654e-01 -4.08486903e-01 -3.61554325e-01 -3.00678253e-01
-4.96254295e-01 8.08346391e-01 -2.28517935e-01 -1.37766972e-01
-2.15004712e-01 3.14979911e-01 -1.14904726e+00 7.26759732e-01
6.62898302e-01 9.04847026e-01 -2.80904055e-01 7.94478178e-01
-5.03521189e-02 1.06575143e+00 -3.17038268e-01 -7.80773878e-01
6.12927675e-02 -3.09490800e-01 -9.41034675e-01 -2.21488953e-01
3.11007053e-01 -4.41177964e-01 -6.41686976e-01 6.37298465e-01
7.74260581e-01 -2.63351053e-01 8.73106420e-01 -1.00518918e+00
-6.78093910e-01 8.86085093e-01 -3.28579813e-01 8.46407115e-02
2.09165141e-01 -3.19265783e-01 1.20394945e+00 -9.62110043e-01
6.66492820e-01 8.89280617e-01 1.03094983e+00 7.95544446e-01
-1.25405157e+00 -6.89299226e-01 -2.91691244e-01 1.41614035e-01
-1.45704675e+00 -3.35521072e-01 4.71443117e-01 -6.08666837e-01
1.19383395e+00 -1.73196755e-02 6.04388058e-01 1.12466192e+00
4.75201994e-01 6.57001197e-01 5.71550608e-01 -5.47466338e-01
-2.46487316e-02 2.55833924e-01 1.58450589e-01 7.19391525e-01
1.58720270e-01 1.43603250e-01 -5.81028759e-02 -6.60224319e-01
2.61828184e-01 2.56421536e-01 3.70861776e-02 -2.34419733e-01
-1.39593744e+00 1.07032669e+00 4.66345340e-01 2.77356267e-01
-3.99443477e-01 6.48883358e-02 9.97900426e-01 1.74660444e-01
5.18473983e-01 4.67705727e-01 -8.88500273e-01 -3.49070549e-01
-9.65865850e-01 2.61386991e-01 6.79446638e-01 7.14238763e-01
1.11881703e-01 -1.24179251e-01 -5.17011106e-01 1.18561029e+00
2.36022353e-01 2.07363844e-01 6.01283014e-01 -6.45266175e-01
6.17831409e-01 5.44087350e-01 -1.10623576e-01 -6.07231438e-01
-7.40315735e-01 -6.78609848e-01 -1.07570052e+00 -2.03994185e-01
-8.00552890e-02 -3.97807658e-01 -1.23284006e+00 1.57134104e+00
-1.67184353e-01 -1.38231784e-01 9.87143293e-02 2.71584839e-01
8.53926003e-01 3.56596559e-01 3.96676600e-01 -2.12717969e-02
1.39586246e+00 -7.82516181e-01 -7.68695772e-01 -7.56769627e-02
1.38569319e+00 -7.26630211e-01 6.02324128e-01 1.79646075e-01
-7.72949517e-01 -6.39202178e-01 -6.86550915e-01 -2.58420277e-02
-1.41895011e-01 4.12967145e-01 5.66746950e-01 6.76929772e-01
-1.11421740e+00 4.95309412e-01 -7.65468776e-01 -3.94926161e-01
7.56676018e-01 5.40156662e-01 -4.52268094e-01 -9.60908681e-02
-1.24930286e+00 9.37398374e-01 5.56451082e-01 -6.44840598e-01
-4.28301692e-01 -1.11043572e+00 -7.05141485e-01 2.54405644e-02
-3.10517818e-01 -6.13261878e-01 9.22062933e-01 -5.96759737e-01
-1.00369251e+00 1.29367709e+00 1.36553109e-01 -8.47199380e-01
5.87341905e-01 -1.53398454e-01 -6.83898926e-01 1.21922947e-01
1.17335573e-01 1.19476807e+00 9.89781134e-03 -7.95697272e-01
-5.19476175e-01 -2.30148017e-01 -3.60166460e-01 -8.29554647e-02
-5.90562582e-01 -4.81789410e-02 9.50393174e-03 -9.44185257e-01
-1.23466529e-01 -1.21840429e+00 -3.24450493e-01 -2.87884157e-02
-2.71609575e-01 -2.28430703e-01 1.70442343e-01 -8.09650004e-01
1.52408206e+00 -2.41300249e+00 -2.74946094e-01 8.36193711e-02
4.56030399e-01 3.70725781e-01 -1.35700360e-01 7.19105899e-01
-7.63080299e-01 4.55645591e-01 -3.21472377e-01 -5.26296139e-01
-1.44078299e-01 2.71679908e-01 -2.80523539e-01 3.49725515e-01
2.96145409e-01 8.20908666e-01 -5.44989347e-01 -6.43968046e-01
1.80664003e-01 4.23738897e-01 -1.08059704e+00 2.89593488e-02
2.77761102e-01 8.77048969e-02 -1.18135802e-01 4.11592782e-01
6.15785241e-01 -3.40938568e-01 2.04089642e-01 9.31020901e-02
2.97530890e-01 4.35888588e-01 -6.71152413e-01 1.65951860e+00
-3.79992396e-01 5.36435187e-01 -4.75360274e-01 -1.04948473e+00
8.99492085e-01 7.76663721e-01 1.12520349e+00 -1.71252936e-01
2.00658023e-01 1.87026292e-01 5.16884625e-01 -3.44510317e-01
1.45714372e-01 -3.38997930e-01 -2.26691365e-01 1.15342140e-01
2.87622474e-02 1.78014025e-01 1.56438723e-01 3.19365531e-01
1.37377369e+00 -3.54080945e-01 3.19679618e-01 -4.45366979e-01
2.15405643e-01 1.89825997e-01 4.22792345e-01 8.59812796e-01
-3.27757239e-01 9.83405292e-01 5.65442801e-01 -8.09020698e-01
-1.32551003e+00 -1.04111195e+00 -8.77845347e-01 7.48882294e-01
-6.79883957e-01 -7.97245204e-01 -4.04293925e-01 -9.33180571e-01
2.84398496e-01 7.62470782e-01 -1.08794284e+00 -4.91823226e-01
-3.51973176e-01 -9.83406305e-01 1.03881562e+00 7.72909224e-01
1.17654428e-01 -9.71954584e-01 -2.66870409e-01 5.41419983e-01
-2.27859825e-01 -1.01406276e+00 -3.64098459e-01 5.89458764e-01
-1.01245630e+00 -9.55071270e-01 -6.90213144e-01 -9.75556493e-01
4.17341053e-01 -9.01531577e-01 1.29356945e+00 2.34083645e-02
-4.19734627e-01 -1.10701732e-02 -6.79594576e-01 -5.59806049e-01
-7.35079348e-01 3.28607112e-01 9.94018763e-02 -3.49893689e-01
4.36191827e-01 -2.76157528e-01 -5.89999318e-01 -3.62777650e-01
-7.70805120e-01 1.59045964e-01 5.78693807e-01 1.16615236e+00
4.24216628e-01 -3.87886018e-01 6.26714349e-01 -1.07417107e+00
4.36791331e-01 -4.53066111e-01 1.18298166e-01 5.44927306e-02
-1.03106213e+00 -1.01326667e-02 5.93256235e-01 -3.97281766e-01
-3.87629539e-01 1.85348719e-01 -8.50793719e-01 -2.35072121e-01
-1.45107642e-01 3.67139250e-01 5.86214066e-01 3.75200421e-01
7.29273736e-01 7.85692930e-02 -3.41512971e-02 -5.06630301e-01
3.47567350e-02 9.86702621e-01 -7.44277388e-02 -7.48239309e-02
2.41835639e-01 2.95864027e-02 -5.89170046e-02 -3.26879740e-01
-7.93398380e-01 -3.50412518e-01 -5.95028937e-01 3.74636322e-01
1.07466829e+00 -1.14373672e+00 -4.16627377e-01 -2.87013594e-02
-1.10887516e+00 3.50738615e-02 -5.45950472e-01 7.14272380e-01
-5.03936887e-01 1.15403175e-01 -9.77135479e-01 -4.72907007e-01
-5.07809937e-01 -9.40918505e-01 1.04474580e+00 -5.79766691e-01
-8.49556029e-01 -9.88889694e-01 3.87109667e-01 1.21114969e-01
1.84719384e-01 3.79262984e-01 1.31616259e+00 -1.10826707e+00
3.93526763e-01 -4.10808474e-01 -2.33942062e-01 4.59943742e-01
3.36074144e-01 -3.45363557e-01 -8.50813746e-01 -3.80064785e-01
-4.62392032e-01 -3.66185099e-01 1.03977144e+00 4.25311446e-01
1.33942235e+00 -1.43698931e-01 -5.65471590e-01 5.18966198e-01
1.34471154e+00 4.27015722e-01 6.10392690e-01 1.67409360e-01
7.55679488e-01 2.26601288e-01 1.62595510e-01 7.28889167e-01
3.35935682e-01 6.50167823e-01 1.55408159e-01 -3.45129788e-01
-2.39561692e-01 -3.08453679e-01 -1.29643515e-01 8.28955114e-01
4.12010521e-01 -1.62085503e-01 -1.47513556e+00 8.61340582e-01
-1.57382238e+00 -6.92936420e-01 -2.94805646e-01 1.82209802e+00
1.16482842e+00 1.98043078e-01 -5.09221926e-02 3.17557812e-01
4.53071207e-01 -1.80345565e-01 -2.58929521e-01 -8.28550696e-01
1.38545752e-01 4.18587118e-01 6.27197683e-01 2.94966787e-01
-1.23542273e+00 5.66178441e-01 7.40766144e+00 6.97191775e-01
-6.71029985e-01 1.17888264e-01 8.51896286e-01 -1.96513534e-01
-6.11133501e-02 -3.61723721e-01 -7.30700135e-01 7.86373138e-01
1.40665662e+00 1.05156645e-01 -1.82491653e-02 7.44093120e-01
-1.17154025e-01 4.55033749e-01 -1.01146102e+00 1.06164360e+00
6.73725605e-02 -1.35101593e+00 1.90142244e-01 2.97557026e-01
9.28524852e-01 3.93709362e-01 -5.01482077e-02 5.64845145e-01
3.73384774e-01 -1.05725861e+00 3.64539951e-01 2.06528410e-01
1.16085708e+00 -9.07597303e-01 1.21874893e+00 7.29420483e-02
-7.28437066e-01 -3.97590637e-01 -1.28340557e-01 1.22775137e-02
5.55663183e-02 9.98819292e-01 -1.06470513e+00 3.37098777e-01
7.48483062e-01 8.55473936e-01 -5.12848198e-01 8.49217176e-01
1.05426989e-01 8.65625739e-01 -1.56178132e-01 9.28772390e-02
5.55139072e-02 5.66686451e-01 8.57163668e-02 1.49461555e+00
2.76746392e-01 4.01483662e-02 2.82616735e-01 1.75290868e-01
-4.36453819e-01 2.78768629e-01 -5.77070713e-01 9.27914865e-03
2.95166552e-01 7.69211113e-01 -3.15302432e-01 -7.12551236e-01
-2.24460229e-01 9.72647429e-01 2.11986691e-01 -7.09502846e-02
-8.48641038e-01 -6.58819497e-01 6.97803974e-01 2.65289456e-01
3.72654706e-01 2.58866102e-01 -6.38639092e-01 -9.75113869e-01
-3.26401979e-01 -7.52329350e-01 5.47457874e-01 -3.25818062e-01
-1.25815821e+00 7.42526174e-01 -2.11820796e-01 -1.45678663e+00
-3.53947043e-01 -6.81009531e-01 -5.60623743e-02 7.55283177e-01
-1.14412761e+00 -1.01550567e+00 2.83539504e-01 2.48308390e-01
2.53784388e-01 -3.97437125e-01 1.36478865e+00 7.63867259e-01
-2.51756012e-01 9.75206733e-01 5.50925493e-01 4.98121977e-01
8.58071804e-01 -1.31734371e+00 5.15814006e-01 1.16159521e-01
-8.30565859e-03 4.66268361e-01 4.85465258e-01 -5.67888737e-01
-5.84258020e-01 -1.25137806e+00 1.56208849e+00 -6.87235832e-01
5.44268191e-01 -3.08165252e-01 -6.12411559e-01 7.83268809e-01
1.55903324e-01 7.67888203e-02 1.08772814e+00 3.39352697e-01
-2.46431679e-01 -8.10624585e-02 -1.26633704e+00 3.21602672e-01
1.06074321e+00 -4.92480427e-01 -5.67426324e-01 5.52526414e-01
6.57797933e-01 -3.43299270e-01 -1.31813395e+00 7.77699709e-01
5.64197600e-01 -6.89063907e-01 9.70900834e-01 -8.44987333e-01
8.77219617e-01 2.00255260e-01 -1.29355639e-01 -1.28316009e+00
-6.48270547e-01 9.48413163e-02 1.29884109e-01 9.93064880e-01
6.63866282e-01 -3.92608494e-01 7.53983855e-01 3.71316284e-01
-1.73488915e-01 -1.13505614e+00 -9.59759474e-01 -5.22677660e-01
4.44316208e-01 -4.15306628e-01 3.21262032e-01 1.02799535e+00
3.17826271e-01 2.33028844e-01 -4.66477722e-01 -2.60632962e-01
2.99512684e-01 -1.73092842e-01 1.52201310e-01 -1.52942979e+00
-1.38276473e-01 -2.38505080e-01 -5.86510956e-01 -5.37634850e-01
1.78340286e-01 -1.41541851e+00 -1.82141438e-01 -1.77569962e+00
3.89358997e-01 -5.09805739e-01 -7.45962203e-01 7.83121109e-01
-1.18728802e-01 4.83858347e-01 1.56501338e-01 1.67240351e-01
-3.49362969e-01 3.21988434e-01 7.80220330e-01 -9.48240608e-02
-1.29009679e-01 -8.65670294e-02 -7.69710720e-01 4.31563795e-01
1.08863604e+00 -9.44461763e-01 -1.72900081e-01 -3.95623922e-01
9.14850757e-02 1.41157657e-01 1.12398583e-02 -9.59004343e-01
-2.85314530e-01 4.62873816e-01 7.00606763e-01 -5.02477229e-01
1.70155972e-01 -9.40508962e-01 3.98657285e-02 1.19403458e+00
-8.06607187e-01 3.69498879e-01 5.53205907e-01 3.76580417e-01
-2.45690361e-01 -1.39171064e-01 8.42534900e-01 1.04571179e-01
-2.38960698e-01 1.96603820e-01 -5.84017813e-01 1.91841081e-01
8.41856599e-01 -3.28956656e-02 8.67117122e-02 -1.30846262e-01
-1.16541600e+00 1.28107518e-01 4.36360128e-02 5.76810300e-01
4.91510212e-01 -1.54179144e+00 -1.06208205e+00 3.89490038e-01
4.90517735e-01 -5.29559851e-01 1.40070558e-01 7.83318818e-01
-6.96132779e-01 5.98126829e-01 -1.18858539e-01 -4.47602153e-01
-1.26116085e+00 2.92997658e-01 1.96311980e-01 -7.05796480e-01
-7.49240041e-01 7.80498803e-01 -3.93949710e-02 -7.15750873e-01
1.84707597e-01 -3.35259467e-01 -1.99714616e-01 1.11281693e-01
4.92277563e-01 2.78296232e-01 3.68002176e-01 -2.63877630e-01
-7.88737953e-01 4.54736620e-01 -4.40208912e-01 7.29087694e-03
1.46316075e+00 2.34111145e-01 1.31686822e-01 1.71505317e-01
1.73160493e+00 -3.64158824e-02 -5.43397307e-01 2.80655492e-02
5.54072410e-02 -4.72443141e-02 7.70497322e-02 -1.26921535e+00
-8.74399424e-01 7.85709083e-01 1.17074585e+00 -2.20191316e-04
8.96194696e-01 1.32669762e-01 8.26671779e-01 2.12507799e-01
-6.11776784e-02 -8.35793674e-01 -2.49400422e-01 6.54766917e-01
6.10912979e-01 -1.34083569e+00 -6.81349784e-02 -5.59441783e-02
-7.98746705e-01 9.64999318e-01 2.23293342e-02 -1.74822912e-01
9.69413638e-01 4.10328567e-01 1.59982979e-01 -2.98893321e-02
-9.17192280e-01 6.92648590e-02 2.42815986e-01 5.06678641e-01
1.00221789e+00 2.67223001e-01 -5.32514572e-01 5.35806060e-01
-2.76220858e-01 2.26219982e-01 3.09236079e-01 6.64972544e-01
-1.65408552e-01 -1.34262228e+00 1.11767147e-02 1.22199523e+00
-1.03113866e+00 -4.83766913e-01 -2.35740930e-01 7.07896233e-01
4.38715696e-01 8.95789683e-01 1.70237884e-01 -6.34623110e-01
4.91260380e-01 4.71247822e-01 1.21392518e-01 -7.68545866e-01
-8.41115415e-01 -4.00116920e-01 2.46139601e-01 -2.23502412e-01
-1.44704372e-01 -6.58433914e-01 -1.32391679e+00 -2.46883720e-01
6.94738477e-02 1.50030732e-01 2.96559572e-01 5.37301719e-01
7.00873733e-01 8.59731555e-01 4.22753900e-01 -1.48789480e-01
-4.79018509e-01 -1.18608737e+00 -6.93103611e-01 5.98610163e-01
6.17950559e-01 -4.02398884e-01 -3.12472165e-01 1.00040287e-01] | [8.031961441040039, 6.852139949798584] |
b402e27e-e30f-4da3-989d-13f538256235 | improved-active-fire-detection-using | 2304.09721 | null | https://arxiv.org/abs/2304.09721v1 | https://arxiv.org/pdf/2304.09721v1.pdf | Improved Active Fire Detection using Operational U-Nets | As a consequence of global warming and climate change, the risk and extent of wildfires have been increasing in many areas worldwide. Warmer temperatures and drier conditions can cause quickly spreading fires and make them harder to control; therefore, early detection and accurate locating of active fires are crucial in environmental monitoring. Using satellite imagery to monitor and detect active fires has been critical for managing forests and public land. Many traditional statistical-based methods and more recent deep-learning techniques have been proposed for active fire detection. In this study, we propose a novel approach called Operational U-Nets for the improved early detection of active fires. The proposed approach utilizes Self-Organized Operational Neural Network (Self-ONN) layers in a compact U-Net architecture. The preliminary experimental results demonstrate that Operational U-Nets not only achieve superior detection performance but can also significantly reduce computational complexity. | ['Moncef Gabbouj', 'Turker Ince', 'Fahad Sohrab', 'Mete Ahishali', 'Ozer Can Devecioglu'] | 2023-04-19 | null | null | null | null | ['fire-detection'] | ['time-series'] | [ 4.43675399e-01 -4.95561123e-01 -3.83106440e-01 3.16083319e-02
8.69292393e-03 -2.71154165e-01 5.75279832e-01 2.98671961e-01
-7.97259510e-01 8.70590627e-01 -4.62173261e-02 -4.33808059e-01
-1.47320583e-01 -1.36027551e+00 -1.72620699e-01 -8.34697783e-01
-6.42109096e-01 -7.59646529e-03 3.92154992e-01 -1.39478505e-01
1.23118773e-01 8.86768818e-01 -1.65028918e+00 -1.27839491e-01
1.09067523e+00 9.06530917e-01 7.13198721e-01 5.90330005e-01
-1.39398007e-02 3.54546815e-01 -4.61318702e-01 5.13027430e-01
4.06151474e-01 -5.02033904e-02 -3.31540912e-01 -1.71695054e-01
-8.63615498e-02 -7.70033419e-01 -2.68282771e-01 1.20040846e+00
3.94807607e-01 4.41612422e-01 6.37774289e-01 -8.52155566e-01
-2.72786051e-01 6.07652962e-01 -8.29024911e-01 4.27431375e-01
-3.70680273e-01 2.81033158e-01 8.55600417e-01 -6.95880413e-01
2.50385731e-01 1.04916704e+00 8.13531995e-01 7.31868087e-04
-1.30020714e+00 -9.36582565e-01 2.99150944e-01 2.44747788e-01
-1.55171108e+00 -3.23425382e-01 4.23733115e-01 -3.44478339e-01
9.51805830e-01 3.13144922e-01 9.96857882e-01 7.71019459e-01
2.42397159e-01 6.25626028e-01 9.90237892e-01 -3.51142794e-01
2.88955808e-01 -5.86801350e-01 -1.95946217e-01 4.64712262e-01
6.46114647e-01 7.90637910e-01 1.46131366e-01 3.47536989e-02
1.15605497e+00 6.44961476e-01 -1.54452249e-01 2.25017488e-01
-9.52651560e-01 9.05098021e-01 1.22041619e+00 5.03805041e-01
-9.64451373e-01 2.43515715e-01 4.13799167e-01 -2.54207432e-01
5.91481388e-01 3.89426976e-01 -1.46062180e-01 1.28811821e-01
-1.21396208e+00 2.32715800e-01 2.84308881e-01 2.18487084e-01
7.06221044e-01 3.30659211e-01 1.32884800e-01 8.55953872e-01
1.96108684e-01 8.33239317e-01 1.01805598e-01 -6.69556499e-01
5.35680503e-02 7.86631405e-01 1.84292838e-01 -1.20836139e+00
-5.40826857e-01 -5.46146750e-01 -1.55679810e+00 7.26529658e-01
-3.10766138e-02 -1.38890445e-01 -1.24505270e+00 1.29704487e+00
-7.68871307e-02 2.01172501e-01 -3.05978626e-01 8.88533950e-01
3.28007668e-01 1.03711283e+00 5.36221921e-01 -3.18058938e-01
1.05362022e+00 -4.59058017e-01 -5.76935768e-01 -4.87190634e-01
1.68909878e-02 -3.32090884e-01 6.65815651e-01 -5.19580022e-03
-3.20547849e-01 -5.15350103e-01 -1.21062505e+00 5.69916010e-01
-6.66226566e-01 2.86702067e-01 7.28862703e-01 3.42804432e-01
-8.36254597e-01 6.23252690e-01 -9.68695223e-01 -6.87322378e-01
7.00875878e-01 5.47337346e-02 -6.68722391e-02 1.25959381e-01
-1.19328392e+00 9.61941183e-01 8.15015018e-01 8.72395873e-01
-1.01707017e+00 -1.38987914e-01 -5.90685010e-01 3.21541041e-01
4.39117968e-01 -8.28665793e-02 1.02905607e+00 -7.40993321e-01
-7.75384724e-01 4.84375000e-01 1.05486335e-02 -6.25284135e-01
2.18098402e-01 -2.13295504e-01 -4.82957363e-01 -9.15835574e-02
1.09301291e-01 6.38052821e-01 5.62980056e-01 -1.19513392e+00
-1.06766212e+00 -4.63137388e-01 -1.42259493e-01 3.64392363e-02
-2.27154300e-01 6.22650348e-02 5.56272566e-01 -7.98567414e-01
2.70762324e-01 -7.15015233e-01 -7.72018850e-01 2.87890553e-01
-6.82321638e-02 -6.79432452e-02 1.09507847e+00 -6.33146763e-01
1.30274785e+00 -1.83587348e+00 -2.69325405e-01 2.02459499e-01
3.13295811e-01 8.89445722e-01 -1.02851335e-02 2.59372503e-01
6.90238476e-02 4.20156866e-01 -6.89636588e-01 5.45269966e-01
-4.43523765e-01 5.93239367e-01 -1.90350428e-01 9.86874849e-02
1.91976771e-01 5.00167608e-01 -1.04670799e+00 -2.43183151e-01
4.98512655e-01 4.31029350e-01 9.98274796e-03 7.23897517e-02
-1.87429413e-01 6.92751482e-02 -4.15690601e-01 8.08190227e-01
7.41152883e-01 3.09190273e-01 5.37666008e-02 1.70528188e-01
-9.05535698e-01 5.34529015e-02 -1.21179283e+00 9.18444395e-01
-3.15358371e-01 7.72103369e-01 4.52536792e-01 -1.04527795e+00
1.13315845e+00 2.63145179e-01 2.94788122e-01 -8.01100969e-01
-6.20310660e-04 2.85070896e-01 1.28363773e-01 -3.78928006e-01
5.11325955e-01 -2.23428339e-01 2.92117566e-01 2.88541377e-01
-3.63502413e-01 4.24348235e-01 2.22740322e-01 -2.75017798e-01
8.82403135e-01 -1.25138164e-01 7.65636683e-01 -2.75652856e-01
2.56958783e-01 1.21172532e-01 8.65161657e-01 7.96674132e-01
-6.60214186e-01 1.32408235e-02 -9.67542678e-02 -8.86105716e-01
-8.18973064e-01 -1.13253772e+00 -1.94485277e-01 1.01056480e+00
-1.35587662e-01 -5.07176630e-02 -3.29593807e-01 -2.87088245e-01
1.21845789e-01 6.40128255e-01 -4.50703651e-01 -4.94496860e-02
-4.28443879e-01 -1.00015891e+00 6.51692390e-01 9.54416871e-01
1.25625694e+00 -1.34718812e+00 -8.67186368e-01 6.32736266e-01
-1.80150256e-01 -5.35045624e-01 1.84741825e-01 5.79669774e-01
-1.22786045e+00 -9.69744444e-01 -7.31862962e-01 -6.20010614e-01
6.01788402e-01 7.51146078e-01 8.19373071e-01 1.54645950e-01
-6.42459214e-01 -4.02305514e-01 -4.78678465e-01 -5.61878264e-01
-1.85349792e-01 3.70709896e-01 8.37426260e-02 -2.71635264e-01
3.89899999e-01 -9.30727303e-01 -7.60795653e-01 2.18765557e-01
-9.01912868e-01 -1.38042361e-01 1.07026231e+00 6.15530193e-01
3.22171539e-01 4.96338099e-01 5.21119058e-01 -4.84370291e-01
4.94915605e-01 -2.11383253e-01 -7.68874347e-01 1.01871088e-01
-6.25952601e-01 -7.80869052e-02 3.35536420e-01 -8.67185891e-02
-9.79115486e-01 2.23410085e-01 -3.88136171e-02 4.11681905e-02
-5.50849974e-01 7.62790859e-01 1.04522727e-01 7.49396607e-02
8.31391752e-01 6.58863634e-02 -3.12169015e-01 -4.45904285e-01
-5.67544326e-02 8.33375096e-01 5.85813820e-01 -1.64485931e-01
1.13966084e+00 2.96474129e-01 -1.14629772e-02 -1.49192834e+00
-5.32180846e-01 -5.90525508e-01 -7.46886075e-01 -6.55451417e-01
7.20368326e-01 -8.27741444e-01 -6.22143090e-01 7.56052256e-01
-1.21098077e+00 -1.36419386e-01 9.14859772e-02 5.02039135e-01
1.54847547e-01 1.42420288e-02 -9.96908098e-02 -1.33979678e+00
-6.12267554e-01 -5.90759456e-01 3.73860121e-01 5.41061163e-01
-1.58271603e-02 -7.48248279e-01 2.86039889e-01 -2.11553469e-01
8.26593101e-01 4.66004282e-01 6.87550366e-01 6.92354292e-02
-6.27539515e-01 -3.20648015e-01 -5.46674311e-01 4.71172094e-01
5.47631681e-01 3.28186810e-01 -8.96675348e-01 -1.80006102e-02
-4.38404113e-01 2.18031436e-01 1.46627128e+00 8.06074500e-01
8.37193012e-01 -2.23576203e-01 -5.42043447e-01 5.10176420e-01
1.66534960e+00 6.74285829e-01 8.51093948e-01 5.94073117e-01
3.44160408e-01 2.78197825e-01 5.36399782e-01 5.52737534e-01
-3.26965362e-01 7.61552230e-02 8.53398442e-01 -4.99557048e-01
3.28761071e-01 4.08502016e-03 3.35465252e-01 3.13932925e-01
-9.81029809e-01 -1.56901870e-02 -1.22346127e+00 7.55291760e-01
-1.89051390e+00 -1.73453701e+00 -1.63893893e-01 2.28145623e+00
5.25358737e-01 1.27448559e-01 9.91888642e-02 1.58351883e-01
1.02614903e+00 6.19455040e-01 -5.49728096e-01 -1.38652816e-01
-1.00344099e-01 5.39899230e-01 9.99952912e-01 3.28022212e-01
-1.58468509e+00 9.46953475e-01 6.70649433e+00 4.60098654e-01
-1.33866346e+00 3.95027921e-02 3.96082401e-01 5.42466231e-02
3.01431924e-01 -7.58167123e-03 -6.49124742e-01 -4.32771677e-03
4.73191530e-01 -6.57567084e-02 4.41289991e-01 8.00261497e-01
8.67172658e-01 -5.05035818e-01 -7.43747652e-02 5.70577025e-01
-6.67481005e-01 -1.19616771e+00 -1.11019962e-01 4.46067899e-02
5.46963573e-01 2.71026969e-01 -4.40617025e-01 2.53251731e-01
5.28554142e-01 -9.34876859e-01 2.80055434e-01 3.62273306e-01
7.24227965e-01 -9.99403358e-01 8.68381202e-01 5.17014205e-01
-1.86154020e+00 -3.06335628e-01 -6.26434803e-01 -5.62330604e-01
5.79165109e-02 8.54063570e-01 -6.31805480e-01 3.39214206e-01
9.13693130e-01 7.49109745e-01 -2.80143261e-01 1.26805294e+00
-5.31946957e-01 7.89400816e-01 -5.92132330e-01 -9.02542546e-02
4.75950390e-01 -4.04787898e-01 6.70190990e-01 9.47862029e-01
3.51855695e-01 1.41179964e-01 5.00963628e-01 9.40019369e-01
2.14392498e-01 -2.22748041e-01 -8.80360365e-01 -3.24015468e-01
5.01612782e-01 1.34650862e+00 -9.14301991e-01 -1.07799456e-01
-7.67486170e-02 6.16580665e-01 2.40622517e-02 2.66810864e-01
-6.36847079e-01 -4.62812424e-01 9.02171254e-01 3.28479439e-01
8.14526454e-02 -7.47127533e-01 -2.44458050e-01 -8.23813498e-01
-2.70683259e-01 -4.64143664e-01 2.34423265e-01 -4.50786442e-01
-7.41172552e-01 3.88595879e-01 -1.62021369e-01 -1.10572267e+00
5.92197515e-02 -6.00445628e-01 -1.04835296e+00 8.35303664e-01
-1.53537059e+00 -9.13036227e-01 -5.40465117e-01 1.63197204e-01
4.61835325e-01 6.43466711e-02 1.15374994e+00 6.42944425e-02
-7.03878939e-01 -3.36652488e-01 3.77961785e-01 3.83942425e-01
3.46778840e-01 -9.45890844e-01 3.86570603e-01 1.32610285e+00
-1.53864324e-01 4.45286393e-01 3.72628421e-01 -8.16399455e-01
-6.99438632e-01 -1.37859571e+00 6.70349300e-01 6.76858425e-01
4.57859248e-01 -2.08499521e-01 -6.73147678e-01 5.27143896e-01
5.25397100e-02 -3.41307431e-01 3.10148805e-01 3.96725625e-01
-1.48988992e-01 -5.10299921e-01 -1.09937918e+00 4.89112198e-01
9.14241433e-01 -2.86831170e-01 -3.14316958e-01 3.15818563e-02
1.27650589e-01 5.10406256e-01 -3.92005652e-01 4.24680650e-01
7.54416049e-01 -8.12208891e-01 7.78670669e-01 -2.45091245e-01
2.15433002e-01 -3.65014642e-01 -1.09737448e-01 -1.47788954e+00
-1.00450790e+00 -6.01018704e-02 2.93106049e-01 1.00437808e+00
2.28891119e-01 -6.90788329e-01 5.82832634e-01 1.47003308e-03
-1.66079849e-02 -1.77373901e-01 -7.43400335e-01 -9.02519166e-01
-3.19120526e-01 -2.03631282e-01 8.47194120e-02 8.11034203e-01
-3.63377929e-01 -3.99410399e-03 -1.77818835e-01 4.28090572e-01
6.74370885e-01 -1.27986714e-01 3.25436175e-01 -1.94038987e+00
3.51199865e-01 -6.57105148e-01 -5.77110946e-01 -5.51497102e-01
-2.69594908e-01 -6.08801425e-01 3.13984364e-01 -1.83803868e+00
2.29790211e-01 -3.81172627e-01 -4.35043246e-01 1.00209630e+00
-2.92956144e-01 1.14639774e-01 -1.11287907e-01 2.27405682e-01
2.35683858e-01 6.66298628e-01 7.64953196e-01 -4.45172817e-01
-3.74094248e-01 9.01335757e-03 -1.66830122e-01 7.17474997e-01
1.08495080e+00 -5.60738981e-01 1.28519803e-03 -3.29918265e-01
2.43229792e-02 -3.10964882e-01 6.50888920e-01 -1.55769897e+00
8.15543607e-02 -5.42537928e-01 5.30126154e-01 -9.30109501e-01
-1.60090700e-01 -9.37145531e-01 -5.40532470e-02 1.01584089e+00
-1.63999405e-02 -1.13211975e-01 4.18626875e-01 6.52935505e-01
-2.32054666e-01 -1.24502085e-01 9.54966843e-01 -3.15890908e-01
-1.22849894e+00 4.81998950e-01 -1.10385096e+00 -5.75586021e-01
1.01247859e+00 -1.59384891e-01 -2.47327402e-01 -2.16521040e-01
-6.34427845e-01 2.65590310e-01 1.05854198e-02 4.06941921e-01
6.90089524e-01 -1.17015672e+00 -8.09617221e-01 2.31173337e-01
-4.80600968e-02 1.74558498e-02 2.08469152e-01 5.65076649e-01
-9.01481748e-01 3.71400893e-01 -6.78133547e-01 -4.30617929e-01
-1.24689960e+00 2.33238935e-01 4.28575933e-01 -3.13468903e-01
-5.34426033e-01 7.44307160e-01 -1.38844594e-01 -3.69476497e-01
1.73340797e-01 -4.23208743e-01 -2.63285339e-01 3.21880132e-01
7.35546291e-01 5.52590311e-01 -1.66008100e-01 -3.45382899e-01
-3.36822420e-01 -4.17275429e-02 -1.15141101e-01 -7.66542321e-03
1.66982102e+00 3.08944643e-01 -2.27426827e-01 2.96155572e-01
4.97316271e-01 -7.05999732e-01 -1.05362535e+00 -1.50079966e-01
1.09766759e-01 -3.88969541e-01 8.70340407e-01 -8.73827279e-01
-1.33980513e+00 8.79719079e-01 9.90621686e-01 3.55940998e-01
1.27279758e+00 -4.91471499e-01 6.63950741e-01 6.74167991e-01
3.11783731e-01 -1.16716671e+00 -5.40536821e-01 5.49052715e-01
8.77555907e-01 -1.12005270e+00 1.74666718e-02 -8.85451362e-02
9.37648043e-02 1.32402277e+00 5.00984132e-01 -8.43399614e-02
6.11654401e-01 4.11781341e-01 1.34036511e-01 1.10599305e-02
-3.20913225e-01 -8.43873739e-01 -3.56905133e-01 8.33533645e-01
3.51657182e-01 6.06932521e-01 -2.79465079e-01 3.87722515e-02
4.55402911e-01 8.65076110e-02 -8.47220328e-03 1.23094237e+00
-1.23628294e+00 -9.00810659e-01 -8.39610875e-01 8.12492669e-01
1.48889408e-01 -3.04464996e-01 -5.72770476e-01 6.96833134e-01
1.45753279e-01 9.70714271e-01 1.95767716e-01 -4.21654999e-01
1.66990966e-01 4.61708568e-02 1.24254361e-01 -3.64828587e-01
-4.37679082e-01 -1.05059996e-01 9.45733115e-03 -2.51940101e-01
-6.08242273e-01 -5.64341187e-01 -1.01816428e+00 -6.73184395e-01
-6.17001534e-01 -1.31436750e-01 8.29396963e-01 8.86646569e-01
1.39141232e-01 5.24463534e-01 6.77831233e-01 -1.14054859e+00
-4.35726523e-01 -1.38816094e+00 -9.70429957e-01 -2.89007604e-01
6.69273362e-02 -8.87269139e-01 -3.68579894e-01 -2.80736566e-01] | [9.308920860290527, -1.369176983833313] |
15cf4019-9aca-4583-8645-719e006eb51e | discriminative-multi-modality-speech | 2005.05592 | null | https://arxiv.org/abs/2005.05592v2 | https://arxiv.org/pdf/2005.05592v2.pdf | Discriminative Multi-modality Speech Recognition | Vision is often used as a complementary modality for audio speech recognition (ASR), especially in the noisy environment where performance of solo audio modality significantly deteriorates. After combining visual modality, ASR is upgraded to the multi-modality speech recognition (MSR). In this paper, we propose a two-stage speech recognition model. In the first stage, the target voice is separated from background noises with help from the corresponding visual information of lip movements, making the model 'listen' clearly. At the second stage, the audio modality combines visual modality again to better understand the speech by a MSR sub-network, further improving the recognition rate. There are some other key contributions: we introduce a pseudo-3D residual convolution (P3D)-based visual front-end to extract more discriminative features; we upgrade the temporal convolution block from 1D ResNet with the temporal convolutional network (TCN), which is more suitable for the temporal tasks; the MSR sub-network is built on the top of Element-wise-Attention Gated Recurrent Unit (EleAtt-GRU), which is more effective than Transformer in long sequences. We conducted extensive experiments on the LRS3-TED and the LRW datasets. Our two-stage model (audio enhanced multi-modality speech recognition, AE-MSR) consistently achieves the state-of-the-art performance by a significant margin, which demonstrates the necessity and effectiveness of AE-MSR. | ['Jacob Wang', 'Yandong Guo', 'Bo Xu', 'Cheng Lu'] | 2020-05-12 | discriminative-multi-modality-speech-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Xu_Discriminative_Multi-Modality_Speech_Recognition_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Xu_Discriminative_Multi-Modality_Speech_Recognition_CVPR_2020_paper.pdf | cvpr-2020-6 | ['lipreading', 'audio-visual-speech-recognition'] | ['computer-vision', 'speech'] | [ 1.70216188e-01 -4.54165518e-01 8.88924394e-03 -2.98258606e-02
-9.69920516e-01 -4.12345305e-02 4.01898414e-01 -6.24232531e-01
-4.75276530e-01 1.73362628e-01 5.58598578e-01 -5.16256392e-01
3.66563052e-01 -1.59531102e-01 -4.90550160e-01 -8.83731902e-01
3.77753943e-01 -1.86949790e-01 3.13373983e-01 -3.27456534e-01
-1.61335841e-02 5.47452629e-01 -1.64128125e+00 5.02213180e-01
4.90733773e-01 1.17438555e+00 6.00282311e-01 9.84818995e-01
-2.28488401e-01 8.94189537e-01 -5.04533231e-01 1.32673383e-01
-1.73319742e-01 -2.62682080e-01 -5.57228446e-01 1.10896513e-01
1.09017696e-02 -2.81330317e-01 -7.82844067e-01 7.83068120e-01
1.08701754e+00 5.40969908e-01 3.11360538e-01 -1.04029667e+00
-8.16287696e-01 3.47036570e-01 -7.53181815e-01 4.38560456e-01
4.97546643e-01 4.61898535e-01 8.08645546e-01 -1.39055598e+00
1.25424713e-01 1.57383502e+00 2.98108250e-01 8.12988102e-01
-7.75026143e-01 -6.97022021e-01 3.71196568e-01 6.00041628e-01
-1.40991032e+00 -9.44453955e-01 7.82614827e-01 -4.73577343e-03
1.19238365e+00 3.55315238e-01 5.20041704e-01 1.34638107e+00
-2.98957914e-01 1.18097842e+00 9.88943040e-01 -2.22659975e-01
-1.26799136e-01 -1.42533749e-01 9.05085504e-02 2.63992310e-01
-5.58980763e-01 3.57911140e-01 -6.66659236e-01 4.11190480e-01
7.70457804e-01 2.03195021e-01 -7.06879973e-01 3.19458991e-01
-1.17836583e+00 3.44370008e-01 4.17086035e-01 6.18920982e-01
-4.14810270e-01 -7.42418021e-02 4.63350534e-01 3.56376171e-01
5.55798590e-01 -2.95993000e-01 -3.35748106e-01 -2.80928999e-01
-9.22214985e-01 -4.77583706e-01 1.27868503e-01 6.20230615e-01
3.14310789e-01 6.26706898e-01 -4.16494936e-01 1.46789300e+00
7.15775192e-01 5.91145515e-01 8.83620024e-01 -2.98773140e-01
4.84177262e-01 2.04378933e-01 -2.51776904e-01 -2.23726630e-01
-2.05474600e-01 -5.81492960e-01 -1.11431205e+00 8.97046030e-02
-2.18957856e-01 8.45255107e-02 -1.39320111e+00 1.44656599e+00
1.98578998e-01 6.30501151e-01 3.85315388e-01 1.20350301e+00
1.57578933e+00 1.02981889e+00 3.86149622e-02 -4.45015907e-01
1.45278943e+00 -1.36046135e+00 -9.65966284e-01 -1.34774253e-01
1.18199363e-01 -8.93650413e-01 1.07514310e+00 1.42113701e-01
-1.01368582e+00 -1.01698446e+00 -7.41256773e-01 -2.08066925e-01
-1.65568203e-01 3.87611657e-01 2.79520694e-02 2.60162741e-01
-1.38505924e+00 1.01073146e-01 -6.47193372e-01 -1.02021486e-01
3.01778764e-01 6.91582412e-02 -5.05004227e-01 -1.18952863e-01
-1.27300572e+00 8.58099818e-01 1.33184254e-01 6.58371031e-01
-1.34072042e+00 -5.09087920e-01 -1.08228350e+00 1.06028527e-01
3.69716585e-01 -3.52501154e-01 1.30171144e+00 -8.59603524e-01
-1.84553659e+00 6.60408735e-01 -7.08025455e-01 -1.75276771e-01
3.49029839e-01 -2.81859725e-03 -8.69360089e-01 2.37372741e-01
-3.64926994e-01 5.91692924e-01 1.28109491e+00 -1.21815968e+00
-5.43525100e-01 -2.86760241e-01 -1.72272921e-01 4.53112751e-01
-9.29095298e-02 3.27616930e-01 -8.45942855e-01 -9.62414384e-01
1.43335149e-01 -6.64202213e-01 9.12178084e-02 -3.02489996e-01
-2.04255089e-01 -4.79455233e-01 1.05296326e+00 -1.13846982e+00
1.27745008e+00 -2.72680712e+00 2.12730259e-01 -2.06107318e-01
1.08198337e-01 8.23630869e-01 -5.99454105e-01 1.00237839e-01
-5.47362566e-01 1.80429798e-02 3.33678387e-02 -8.57151270e-01
-2.55401462e-01 4.05306369e-02 -4.71786559e-01 3.48437250e-01
2.56158948e-01 9.60490346e-01 -5.61971664e-01 -3.54626596e-01
4.96664435e-01 9.16367769e-01 2.64959000e-02 4.26111877e-01
2.33117193e-01 5.30540109e-01 -1.66629836e-01 7.59157836e-01
7.26361930e-01 -7.17843883e-03 -3.80140930e-01 -2.43178457e-01
-1.56656116e-01 4.14427787e-01 -1.05190742e+00 1.72587657e+00
-8.34132612e-01 6.62982881e-01 3.34612072e-01 -7.51157701e-01
1.13175547e+00 8.67006660e-01 1.33403584e-01 -9.83242154e-01
4.39349636e-02 1.54005408e-01 -3.45656052e-02 -6.74683511e-01
1.88557118e-01 -1.39872551e-01 4.75945711e-01 4.88332845e-02
1.14102915e-01 1.05965756e-01 -4.62160707e-01 -2.07014307e-02
8.38400662e-01 -5.62966391e-02 -5.09853428e-03 3.67500097e-01
9.69960093e-01 -7.01654673e-01 5.88781238e-01 3.87500465e-01
-3.73315424e-01 8.52721751e-01 -6.55844137e-02 -2.84807011e-02
-6.31654561e-01 -1.06245041e+00 3.38029340e-02 1.06378639e+00
1.75837815e-01 -1.51213318e-01 -3.07753474e-01 -5.84675968e-01
-2.75793999e-01 5.67159414e-01 -3.47625762e-01 -2.53570914e-01
-4.79594916e-01 -2.69010246e-01 5.86029828e-01 7.82495379e-01
7.66343534e-01 -1.40058303e+00 2.71018059e-03 9.65193361e-02
-3.69352788e-01 -1.13273835e+00 -8.16860735e-01 -3.02069746e-02
-5.21485746e-01 -5.99447250e-01 -1.49045432e+00 -1.02342868e+00
3.61709267e-01 7.59220004e-01 2.99256355e-01 2.96377409e-02
6.16932996e-02 4.78868812e-01 -5.99938154e-01 -1.24048598e-01
-3.06787580e-01 -2.87622303e-01 1.64893165e-01 4.84816939e-01
1.38259828e-01 -6.25910342e-01 -5.41131377e-01 2.99507111e-01
-7.34974146e-01 8.06586072e-02 7.24380791e-01 9.97560859e-01
6.08905017e-01 -9.83841643e-02 7.11663425e-01 -1.84168469e-03
6.11659527e-01 -1.08490460e-01 -1.55007526e-01 2.78470814e-01
-1.54404014e-01 -3.13034147e-01 6.36440158e-01 -8.56159329e-01
-1.26840687e+00 -1.15116730e-01 -7.80181885e-01 -1.20110142e+00
-2.78522581e-01 3.21369946e-01 -4.75760102e-01 4.01707226e-03
2.59556109e-04 8.06301117e-01 1.24119610e-01 -8.92969251e-01
1.87799588e-01 1.33424914e+00 5.69274306e-01 8.09442773e-02
5.54187477e-01 1.43583417e-01 -3.85893077e-01 -1.31566715e+00
-5.27968705e-01 -7.69021511e-01 -1.13985017e-01 -3.47309738e-01
8.28951001e-01 -1.30219030e+00 -9.39355016e-01 9.12828624e-01
-1.22311151e+00 -3.43989789e-01 -8.13526288e-02 7.95168757e-01
-1.65088490e-01 5.46029449e-01 -6.89040661e-01 -1.10388303e+00
-4.92172092e-01 -1.48409808e+00 1.15965068e+00 4.77530479e-01
4.40138429e-01 -6.44461751e-01 -1.60778880e-01 6.58263266e-01
5.39842486e-01 -5.86173475e-01 5.64965367e-01 -8.41517389e-01
-2.79278159e-01 2.26612046e-01 -3.70505661e-01 7.40423322e-01
1.24387860e-01 -2.20288306e-01 -1.57463872e+00 -4.70417231e-01
6.68381155e-02 -1.48345515e-01 1.31191742e+00 4.31563407e-01
1.26095331e+00 1.06627485e-02 -1.06904313e-01 5.45434952e-01
8.83804142e-01 6.27865553e-01 9.43843782e-01 -4.07130718e-02
7.90993452e-01 3.45138192e-01 6.05530858e-01 1.23144756e-03
3.07315022e-01 8.37622106e-01 3.45400453e-01 -5.27631402e-01
-7.15149462e-01 -2.95115173e-01 7.72564709e-01 1.29243553e+00
-1.19347617e-01 -2.21042439e-01 -6.91693544e-01 5.47702730e-01
-1.72825062e+00 -9.18410778e-01 2.25890353e-01 2.01946592e+00
5.90828657e-01 -1.21492958e-02 9.24350619e-02 3.79598260e-01
1.04369509e+00 5.29965520e-01 -5.34428596e-01 -3.02960426e-01
-3.46523255e-01 2.94963717e-01 -7.71927312e-02 6.76184058e-01
-8.73084247e-01 9.33582902e-01 5.36132479e+00 1.39696240e+00
-1.48739755e+00 2.32831597e-01 4.50584292e-01 -2.95065232e-02
-1.80678055e-01 -3.00162554e-01 -7.29136050e-01 3.58618706e-01
8.40544820e-01 2.21970826e-01 6.18982017e-01 6.48552418e-01
5.48586667e-01 2.18175530e-01 -6.99397385e-01 1.54059327e+00
2.23974526e-01 -9.66768324e-01 1.14277408e-01 -4.03039493e-02
1.00552380e-01 5.02292290e-02 2.16033265e-01 6.14122689e-01
-2.03983605e-01 -1.05382836e+00 6.43863380e-01 4.24630672e-01
1.33487463e+00 -7.71535814e-01 5.91788590e-01 1.84462756e-01
-1.78661621e+00 -7.80635923e-02 -1.57036677e-01 3.56848717e-01
2.61433929e-01 2.72851586e-01 -7.20786035e-01 6.40137553e-01
8.11702669e-01 8.98415744e-01 -3.25396955e-01 9.77722883e-01
-3.63969505e-01 6.93508923e-01 -1.19132318e-01 8.92870724e-02
1.28405243e-01 1.11371174e-01 9.49633956e-01 1.34325945e+00
1.43903196e-01 8.21373612e-02 -8.26007798e-02 3.80002797e-01
-8.67653787e-02 -3.51272710e-02 -4.18976992e-01 -9.33149904e-02
4.33593124e-01 1.20536184e+00 -2.28389069e-01 -2.23788276e-01
-5.26470184e-01 1.02012682e+00 -2.74859779e-02 6.97016060e-01
-8.23569655e-01 -5.40963769e-01 7.71080792e-01 -4.71407264e-01
6.88632309e-01 -1.26383707e-01 1.66509673e-01 -1.29826510e+00
1.32293850e-01 -1.07391810e+00 1.74636811e-01 -1.19385445e+00
-1.05381382e+00 8.53604496e-01 -5.54010570e-01 -1.28245330e+00
-4.12339866e-02 -5.50457954e-01 -6.58524930e-01 1.24193037e+00
-2.00359511e+00 -1.29328501e+00 -1.42284021e-01 9.90333259e-01
1.15361810e+00 -2.29447126e-01 5.77925444e-01 5.37597537e-01
-7.93625772e-01 8.12770307e-01 -1.23785712e-01 2.12490439e-01
5.96641183e-01 -7.50248015e-01 4.55414534e-01 1.01567364e+00
9.90326256e-02 3.66417736e-01 2.10356921e-01 -5.35970747e-01
-1.35822594e+00 -1.09611261e+00 7.43986428e-01 2.44168088e-01
4.33442384e-01 -2.30195820e-01 -1.15482819e+00 4.54813927e-01
5.01959287e-02 1.36776388e-01 4.71957564e-01 -1.96146116e-01
-4.79645938e-01 -2.97233611e-01 -7.82727480e-01 5.79246521e-01
1.01048851e+00 -1.17404127e+00 -7.48288333e-01 -9.92237404e-02
1.28667188e+00 -3.69300395e-01 -6.54731750e-01 5.66778660e-01
4.35906917e-01 -5.59239745e-01 1.11037266e+00 -3.70174080e-01
1.19317189e-01 -6.24891222e-01 -3.43081206e-01 -1.35904527e+00
-1.36458948e-01 -7.42308974e-01 -1.46597564e-01 1.53759873e+00
3.21685791e-01 -6.63702488e-01 4.32749391e-02 -1.50329158e-01
-4.52702045e-01 -7.44257808e-01 -1.26997733e+00 -6.55624866e-01
-5.67346156e-01 -7.64258087e-01 4.95480835e-01 7.09200740e-01
-3.24742943e-01 5.09756327e-01 -5.36008835e-01 3.14089209e-01
1.53013274e-01 -1.30209982e-01 3.51495802e-01 -7.63407767e-01
-2.44691193e-01 -3.18047374e-01 -2.34694928e-01 -1.56721854e+00
2.16814741e-01 -7.09409595e-01 1.52642265e-01 -1.58013797e+00
-9.81284454e-02 -5.61341178e-03 -6.00399792e-01 5.60138226e-01
-2.05327064e-01 8.22710693e-02 3.48015845e-01 1.63034752e-01
-4.52146918e-01 9.94387805e-01 1.57105374e+00 -2.66818494e-01
-3.87117356e-01 1.93867609e-01 -4.29789782e-01 5.31908631e-01
3.63410711e-01 4.96712886e-02 -2.34252259e-01 -4.07250196e-01
-5.15512109e-01 4.99189198e-01 4.23647404e-01 -7.44746089e-01
4.15381640e-01 1.01188153e-01 2.61226952e-01 -9.26317990e-01
8.34721804e-01 -7.70908952e-01 -1.65413365e-01 1.78596526e-01
-3.24674785e-01 -1.33805469e-01 4.39677000e-01 4.87142503e-01
-6.00613296e-01 6.28231987e-02 9.24969852e-01 1.46291450e-01
-9.29133773e-01 4.17782873e-01 -4.00319487e-01 -2.11390570e-01
6.31178617e-01 -2.19146341e-01 -3.50237131e-01 -5.30813873e-01
-9.30292845e-01 2.75068432e-01 -1.10179760e-01 7.04817772e-01
1.26175547e+00 -1.40990341e+00 -7.19058216e-01 3.61308217e-01
-4.76067550e-02 -2.45247614e-02 8.96926522e-01 9.67165112e-01
1.89645719e-02 2.54605412e-01 1.50112569e-01 -6.44087017e-01
-1.58640742e+00 6.13703668e-01 4.78818446e-01 1.56857759e-01
-8.02410662e-01 9.42367315e-01 4.21725541e-01 -9.54044312e-02
6.87959492e-01 -1.35342538e-01 -6.41114891e-01 1.22508980e-01
8.74516904e-01 4.29745376e-01 2.11832859e-02 -1.00239134e+00
-5.55472136e-01 5.77614307e-01 -1.73848361e-01 -4.41090196e-01
1.39545465e+00 -4.19076324e-01 9.76574197e-02 5.36669910e-01
1.27814806e+00 1.05474144e-02 -1.13871765e+00 -4.05116200e-01
-7.31117249e-01 -2.93717057e-01 4.89055157e-01 -8.12989533e-01
-1.29661393e+00 1.45718420e+00 7.88097441e-01 1.55179240e-02
1.52716827e+00 -6.23902157e-02 9.00487185e-01 -6.92966878e-02
-2.02590302e-01 -7.26688623e-01 2.93231696e-01 6.87873363e-01
1.23206568e+00 -1.12049365e+00 -6.55318975e-01 -1.88284934e-01
-9.48771477e-01 1.12250364e+00 5.45869231e-01 3.60351980e-01
5.92387438e-01 1.74114570e-01 3.12831312e-01 2.93675780e-01
-7.25835681e-01 -6.43255472e-01 4.86809313e-01 6.59812391e-01
1.80616796e-01 -8.76657590e-02 3.20521742e-01 7.96903908e-01
1.42387167e-01 -2.11629957e-01 5.38804419e-02 5.23774087e-01
-3.34984630e-01 -7.30940342e-01 -4.14089531e-01 -7.04501942e-02
-3.45197052e-01 -3.99643302e-01 -1.57061130e-01 3.99212033e-01
-1.67113543e-01 1.29677701e+00 1.79021768e-02 -7.54562795e-01
5.09997487e-01 9.15623233e-02 1.13590047e-01 -4.02541548e-01
-6.05312765e-01 7.20901668e-01 -7.73842409e-02 -4.85587120e-01
-3.31601322e-01 -3.97438258e-01 -1.13269544e+00 3.40532102e-02
-4.76174951e-01 3.29179205e-02 6.73128903e-01 1.03605366e+00
4.19772863e-01 9.73314762e-01 1.00458908e+00 -9.32230353e-01
-3.21175307e-01 -1.16702449e+00 -5.37730575e-01 1.30313680e-01
9.25315440e-01 -5.58553994e-01 -6.89494252e-01 -2.25778177e-01] | [14.34458065032959, 5.219583511352539] |
2eacdd81-81db-4cfa-b41c-f0875f548d6c | multi-stage-temporal-difference-learning-for | 1606.07374 | null | http://arxiv.org/abs/1606.07374v2 | http://arxiv.org/pdf/1606.07374v2.pdf | Multi-Stage Temporal Difference Learning for 2048-like Games | Szubert and Jaskowski successfully used temporal difference (TD) learning
together with n-tuple networks for playing the game 2048. However, we observed
a phenomenon that the programs based on TD learning still hardly reach large
tiles. In this paper, we propose multi-stage TD (MS-TD) learning, a kind of
hierarchical reinforcement learning method, to effectively improve the
performance for the rates of reaching large tiles, which are good metrics to
analyze the strength of 2048 programs. Our experiments showed significant
improvements over the one without using MS-TD learning. Namely, using 3-ply
expectimax search, the program with MS-TD learning reached 32768-tiles with a
rate of 18.31%, while the one with TD learning did not reach any. After further
tuned, our 2048 program reached 32768-tiles with a rate of 31.75% in 10,000
games, and one among these games even reached a 65536-tile, which is the first
ever reaching a 65536-tile to our knowledge. In addition, MS-TD learning method
can be easily applied to other 2048-like games, such as Threes. Based on MS-TD
learning, our experiments for Threes also demonstrated similar performance
improvement, where the program with MS-TD learning reached 6144-tiles with a
rate of 7.83%, while the one with TD learning only reached 0.45%. | ['Chao-Chin Liang', 'Chia-Chuan Chang', 'Chu-Hsuan Hsueh', 'I-Chen Wu', 'Kun-Hao Yeh', 'Han Chiang'] | 2016-06-23 | null | null | null | null | ['2048'] | ['playing-games'] | [-4.15161461e-01 -1.21990256e-01 -4.22559291e-01 3.20103616e-01
-8.82462859e-01 -4.41599041e-01 3.60193580e-01 -1.14668813e-02
-7.40380526e-01 9.18209195e-01 -2.26746887e-01 -6.77565813e-01
-2.23570779e-01 -1.25119150e+00 -8.85872841e-01 -5.35485029e-01
-4.91164178e-01 3.36217165e-01 9.53572690e-01 -6.15521789e-01
3.85705471e-01 -1.70955613e-01 -1.62625086e+00 6.43139184e-01
1.02439535e+00 1.11969995e+00 7.49358833e-02 7.48137355e-01
-1.21916585e-01 1.02723432e+00 -7.69065320e-01 -1.18455989e-02
3.65197092e-01 -1.88608274e-01 -7.37399995e-01 -6.32465184e-01
-3.83300453e-01 -3.79880816e-01 -3.54062706e-01 8.88380349e-01
3.69210303e-01 3.47838365e-02 2.57265896e-01 -1.32513118e+00
-1.11388214e-01 1.11531746e+00 -1.07842612e+00 4.40287441e-02
1.92516610e-01 2.77129292e-01 9.22414482e-01 -5.11537015e-01
1.63320825e-01 9.98042047e-01 5.03015041e-01 5.15516043e-01
-8.13272119e-01 -1.15386915e+00 -9.45096090e-02 2.63906747e-01
-1.24816871e+00 3.26260589e-02 1.86042145e-01 -8.38961229e-02
1.00736392e+00 2.73193568e-01 8.31868470e-01 4.92565393e-01
3.29987854e-01 8.98357987e-01 1.48889351e+00 -5.96762240e-01
3.65980744e-01 -3.40545714e-01 -2.37088017e-02 7.58384049e-01
1.52796373e-01 5.68933308e-01 -6.50463939e-01 -1.26160875e-01
9.84171569e-01 -5.01151145e-01 4.85349566e-01 7.79805258e-02
-1.22250926e+00 8.02128315e-01 2.45278344e-01 4.64862406e-01
-1.79949209e-01 3.86095285e-01 5.91692448e-01 4.85027730e-01
2.67828703e-01 6.10521019e-01 -5.32722890e-01 -8.18987012e-01
-7.66767383e-01 4.02570605e-01 6.84409380e-01 8.45314264e-01
8.80852580e-01 2.56677508e-01 -2.13374928e-01 4.82753158e-01
-1.16679631e-01 4.80020165e-01 5.42670190e-01 -1.19353437e+00
5.88010013e-01 7.01889157e-01 7.85460919e-02 -7.51170695e-01
-5.45575261e-01 -1.53625607e-01 -9.55157280e-01 6.27741456e-01
7.65795648e-01 -6.65229797e-01 -5.77882349e-01 1.79083145e+00
2.20894068e-02 1.84092551e-01 1.57903418e-01 5.15816271e-01
5.80322146e-01 8.75803709e-01 -2.84837484e-01 -2.42087871e-01
1.02715743e+00 -1.00328410e+00 -7.68153295e-02 -9.51711759e-02
1.07935727e+00 -5.44967234e-01 1.16238832e+00 7.80937791e-01
-1.13961840e+00 -5.31736195e-01 -1.14395428e+00 8.99255157e-01
-1.41901001e-01 -1.83498815e-01 7.74824858e-01 9.18671191e-01
-1.31007147e+00 6.30996883e-01 -6.73701227e-01 -1.50241792e-01
-5.69303893e-03 6.61814809e-01 6.96954131e-02 1.78517520e-01
-1.48143268e+00 6.91669583e-01 7.13566780e-01 -5.21263361e-01
-8.92433822e-01 -5.64903438e-01 -4.80786920e-01 -2.29284028e-03
9.41624403e-01 -8.67343470e-02 1.42839324e+00 -4.29596215e-01
-1.88852882e+00 2.85843074e-01 1.29598707e-01 -5.48775792e-01
3.43018949e-01 7.22040841e-03 -4.59683329e-01 -2.32407406e-01
1.96613729e-01 5.50353646e-01 2.16775343e-01 -7.23874331e-01
-9.05440807e-01 -2.40466878e-01 5.71161568e-01 4.93934266e-02
-4.14810598e-01 -9.04482976e-02 -1.91924319e-01 -1.42791033e-01
-1.73441485e-01 -1.10112000e+00 -4.01474208e-01 -5.53876758e-01
-1.54644176e-01 -6.11460447e-01 4.49098259e-01 -9.53579396e-02
1.41653490e+00 -2.00247908e+00 -3.13481480e-01 2.34529719e-01
2.05770314e-01 3.59196186e-01 -3.45698625e-01 4.79163885e-01
1.82073429e-01 3.33749145e-01 1.88232183e-01 3.37522537e-01
1.73055772e-02 9.93739516e-02 8.10962468e-02 -3.84304635e-02
1.11541331e-01 7.85685420e-01 -9.96608019e-01 -4.27480310e-01
3.03656198e-02 -6.84172332e-01 -7.77806580e-01 1.18107580e-01
-5.46171546e-01 4.09898646e-02 -4.58538562e-01 3.81854147e-01
6.29045963e-01 -6.37862533e-02 2.40502059e-01 3.91628444e-01
-6.62647367e-01 8.63498524e-02 -1.27020621e+00 1.62267971e+00
-5.43494284e-01 4.63965148e-01 -5.11821210e-01 -9.08935308e-01
1.11866951e+00 9.64514315e-02 5.63745439e-01 -1.33759618e+00
-1.07807301e-01 5.41670144e-01 3.10212553e-01 -3.65914822e-01
7.28080511e-01 -3.10287774e-02 -7.72608221e-01 6.90564513e-01
-4.35443282e-01 -2.17137396e-01 4.71766829e-01 9.98306200e-02
1.45245445e+00 -3.01329158e-02 1.24486238e-01 -4.34848934e-01
3.79419714e-01 1.41950518e-01 7.17293203e-01 1.18952799e+00
6.41140193e-02 1.43904284e-01 1.10138142e+00 -5.35610914e-01
-8.38765442e-01 -6.86868608e-01 4.77463961e-01 1.33540940e+00
3.46550077e-01 -5.91739774e-01 -5.05350471e-01 -6.76580906e-01
-2.33566225e-01 6.59612536e-01 -4.70848173e-01 -3.10799837e-01
-6.19413793e-01 -6.18796408e-01 1.04807603e+00 5.12757421e-01
1.16908097e+00 -1.24256623e+00 -7.51996398e-01 2.79763699e-01
-1.68386623e-01 -6.67631388e-01 -3.50506902e-01 4.27363485e-01
-6.60145283e-01 -1.11107516e+00 -6.61778808e-01 -7.27228403e-01
-8.33821297e-02 1.64204672e-01 8.91371369e-01 1.00111827e-01
1.50625616e-01 -2.49070153e-01 -4.11199272e-01 -1.88031539e-01
-4.47583854e-01 3.05608124e-01 1.48258120e-01 -6.93080068e-01
1.08935788e-01 -5.98331630e-01 -3.14828545e-01 5.90164006e-01
-8.30628335e-01 -1.05648406e-01 6.14395678e-01 9.77471530e-01
1.08892985e-01 5.43745935e-01 7.33551025e-01 -5.55533588e-01
6.96433365e-01 -4.15666938e-01 -8.76615644e-01 1.34297162e-01
-7.28569031e-01 3.55369687e-01 8.44187319e-01 -5.00481129e-01
-8.80160511e-01 -3.19226325e-01 -1.22898288e-01 1.35613242e-02
2.21186668e-01 6.86971366e-01 1.44433036e-01 8.54475647e-02
8.37489247e-01 3.07544202e-01 -2.90377140e-02 4.69693914e-02
2.09201425e-01 6.32302940e-01 2.09288269e-01 -1.05906534e+00
5.84417522e-01 -1.86570734e-01 -9.41587985e-02 -3.98942500e-01
-3.85723054e-01 -2.65654981e-01 7.33229220e-02 -1.70061857e-01
6.53363049e-01 -6.97532535e-01 -1.42086780e+00 9.90016043e-01
-9.87508357e-01 -1.05917597e+00 -1.68135047e-01 3.93425375e-01
-6.43768787e-01 2.10844114e-01 -7.82413840e-01 -8.25157642e-01
-5.62892668e-02 -1.18140423e+00 5.73373616e-01 3.58580112e-01
-1.26516521e-01 -5.92027485e-01 1.35644034e-01 1.87373143e-02
4.88418311e-01 1.24917008e-01 1.23054016e+00 -4.10814703e-01
-4.01024669e-01 3.09071869e-01 -2.54243910e-01 1.66198656e-01
2.49288362e-02 -7.27175772e-02 -3.83872122e-01 -1.82508171e-01
-3.54474872e-01 -6.30799055e-01 4.69499677e-01 2.66327947e-01
1.31317258e+00 -1.12228565e-01 4.31001224e-02 1.83251664e-01
1.38657403e+00 9.65742469e-01 9.56896365e-01 9.87157345e-01
2.00291842e-01 1.56674802e-01 1.05388403e+00 8.38461459e-01
2.58245975e-01 6.84826434e-01 6.79827690e-01 9.42170992e-02
2.10815042e-01 -2.92872220e-01 5.94058990e-01 8.49573195e-01
-3.36190075e-01 -2.43534282e-01 -1.20362592e+00 3.98830146e-01
-1.94050157e+00 -7.75295675e-01 -1.18396237e-01 2.32604766e+00
1.10884285e+00 7.78389037e-01 6.69059634e-01 2.73687422e-01
7.41331041e-01 3.23479027e-01 -6.79117858e-01 -7.91687250e-01
5.74903600e-02 5.93663990e-01 6.50806606e-01 2.67814606e-01
-6.76500857e-01 1.17142379e+00 5.99083853e+00 1.50737345e+00
-9.82435405e-01 -2.00975984e-01 5.19151866e-01 4.86994796e-02
-9.61017385e-02 6.40660971e-02 -6.52930081e-01 6.13662541e-01
1.10605156e+00 -7.12011695e-01 5.55537045e-01 8.12848389e-01
8.25591087e-02 -6.00519121e-01 -8.00381601e-01 8.09286654e-01
-6.20486557e-01 -1.38633847e+00 -2.74548709e-01 1.02342248e-01
9.89255846e-01 -3.21352184e-01 2.14920208e-01 9.45823014e-01
8.92263949e-01 -1.14628422e+00 5.47960103e-01 6.62721843e-02
1.04240608e+00 -1.34317338e+00 9.09974635e-01 8.05683672e-01
-1.29753470e+00 -1.25194043e-01 -3.06052506e-01 -6.11637771e-01
-4.47008908e-01 5.95039606e-01 -7.43240297e-01 7.48360455e-01
9.78669345e-01 4.25973386e-01 -2.21106350e-01 9.50600207e-01
7.48321926e-03 6.04594111e-01 -4.66045678e-01 -6.74255610e-01
5.48084795e-01 1.24689184e-01 2.20166415e-01 7.41842866e-01
6.02040529e-01 8.93616527e-02 1.47669986e-01 4.89105850e-01
-6.54696077e-02 -6.09556772e-02 -5.41675806e-01 1.80090621e-01
5.76779842e-01 8.90988708e-01 -5.60244620e-01 -2.62393445e-01
-2.37421483e-01 4.28626508e-01 2.00434625e-01 -1.01214640e-01
-1.22937369e+00 -5.82891405e-01 3.88964623e-01 6.15603849e-02
1.93886057e-01 -3.23318332e-01 -2.58125842e-01 -7.15633869e-01
-2.77838651e-02 -1.33395839e+00 2.62813479e-01 -5.05980134e-01
-8.67529571e-01 5.76213181e-01 1.31271631e-01 -1.39371121e+00
-3.96498799e-01 -3.75339240e-01 -7.31816947e-01 4.17616367e-01
-1.03407705e+00 -4.96566236e-01 -2.81417072e-01 4.95212525e-01
4.72766787e-01 -4.28989202e-01 7.42561638e-01 1.59364119e-01
-4.41338599e-01 9.12597418e-01 3.27581853e-01 -1.33779019e-01
4.95555133e-01 -1.27349520e+00 2.95489252e-01 3.87371093e-01
-1.99770257e-01 1.49552464e-01 3.09859961e-01 -5.18284976e-01
-1.27557325e+00 -9.12572086e-01 3.73576790e-01 3.10864866e-01
7.65486121e-01 -8.00400153e-02 -5.85205019e-01 2.13741049e-01
1.95561811e-01 -3.57040584e-01 1.86399594e-01 2.46361271e-01
-2.08858356e-01 -3.57922912e-01 -9.67880189e-01 8.50292802e-01
1.11217546e+00 -1.18533812e-01 -1.89879671e-01 1.51966855e-01
8.94285023e-01 -7.53653109e-01 -8.96668136e-01 4.78691816e-01
5.11612296e-01 -1.48235905e+00 5.53064644e-01 -2.93211669e-01
7.28069663e-01 -2.19154894e-01 -1.96390320e-02 -1.35131836e+00
-3.35019618e-01 -6.56835377e-01 2.23670691e-01 9.20046687e-01
4.35793161e-01 -8.69254529e-01 1.18249202e+00 1.59122273e-01
-1.02550341e-02 -9.23205793e-01 -1.00278223e+00 -1.27694631e+00
6.13660574e-01 -6.11514509e-01 9.73852754e-01 5.33243716e-01
3.19576859e-01 -8.72561783e-02 -6.81776464e-01 -2.65791595e-01
3.09748083e-01 9.95296463e-02 8.00280035e-01 -8.05660486e-01
-7.47149467e-01 -5.06561160e-01 -1.24072962e-01 -1.31004584e+00
-5.99384820e-03 -5.57309508e-01 1.24769583e-01 -1.08126819e+00
3.60265493e-01 -8.90133977e-01 -3.87110084e-01 6.67800128e-01
8.80522355e-02 4.12383750e-02 6.64805546e-02 6.50177943e-03
-8.74435842e-01 3.76804560e-01 1.46527004e+00 -8.75586495e-02
-4.70266223e-01 3.72814387e-01 -6.93012357e-01 6.27851248e-01
1.24327445e+00 -6.06996000e-01 -3.87037396e-01 -1.98484644e-01
4.67916429e-01 6.90287411e-01 -9.75714102e-02 -1.13794887e+00
5.28081298e-01 -4.25372124e-01 2.35617701e-02 -8.04705977e-01
6.60579428e-02 -2.90926158e-01 8.77822191e-03 1.05163682e+00
-2.73427695e-01 2.80919135e-01 6.73396885e-01 3.32143486e-01
-2.52073139e-01 -3.04485887e-01 3.75633121e-01 -1.90969273e-01
-1.05719411e+00 1.21844970e-01 -9.78206754e-01 3.65932435e-01
1.35389256e+00 -1.70243874e-01 -5.74823678e-01 -5.15365422e-01
-4.23901916e-01 4.34454590e-01 2.24827617e-01 1.08736716e-01
5.85129142e-01 -1.47411287e+00 -5.77278316e-01 -1.55174201e-02
-1.45029649e-01 1.18469909e-01 1.30629897e-01 6.62466407e-01
-5.52881598e-01 3.23058546e-01 -4.72025037e-01 -4.64102715e-01
-1.20021272e+00 1.82402834e-01 2.27374047e-01 -1.05541074e+00
-2.06904203e-01 5.21055043e-01 4.91697192e-02 -5.65785825e-01
-2.15884540e-02 -3.63963962e-01 3.14619280e-02 -1.22304291e-01
1.62255496e-01 7.12853670e-01 4.22918387e-02 3.50147694e-01
-5.64495921e-01 5.98140776e-01 -1.36935087e-02 -4.33578372e-01
1.34018898e+00 5.36106825e-01 -1.66091487e-01 3.59808862e-01
8.96672308e-01 -2.80468408e-02 -9.39014971e-01 -2.05425203e-01
2.76366711e-01 -3.01759601e-01 -2.81397760e-01 -6.67574942e-01
-1.09534383e+00 6.42518282e-01 3.71197402e-01 3.52405995e-01
1.24908864e+00 -1.74242944e-01 9.40386057e-01 5.05410433e-01
1.26875377e+00 -1.05098343e+00 7.02230871e-01 1.11715400e+00
4.20747280e-01 -1.04561388e+00 -2.28276879e-01 -1.13787308e-01
-5.94130456e-01 1.25226486e+00 1.14334977e+00 -3.05129260e-01
2.70087540e-01 6.05444670e-01 -3.31554443e-01 1.50790095e-01
-9.66111243e-01 -1.57683402e-01 -1.94110677e-01 5.01936793e-01
-3.58767365e-03 2.45935664e-01 -2.55874842e-01 6.90559924e-01
-3.52348715e-01 2.33284235e-01 7.79084980e-01 1.04333603e+00
-5.63259244e-01 -1.57039678e+00 -2.96444952e-01 4.53313202e-01
-2.73069650e-01 -2.57846117e-01 2.64324248e-02 1.12633681e+00
3.09338748e-01 1.10800993e+00 1.39713556e-01 -1.10780513e+00
3.89262468e-01 -3.67501110e-01 4.41380918e-01 -5.08842945e-01
-8.24029982e-01 -1.77782252e-01 3.20282042e-01 -6.40136182e-01
-1.79799601e-01 -4.38260913e-01 -1.39495730e+00 -9.44908142e-01
-3.13345283e-01 3.62382531e-01 8.38409141e-02 6.50335789e-01
5.52727468e-02 5.67487538e-01 9.25871968e-01 -2.02029586e-01
-6.09193146e-01 -6.52120948e-01 -7.85883129e-01 -4.08541977e-01
-1.84345782e-01 -4.89888608e-01 -4.08785999e-01 -7.80352831e-01] | [3.6176819801330566, 1.56121826171875] |
c3aaae69-0ced-4604-8298-93c21bd61559 | ian-the-individual-aggregation-network-for | 1705.05552 | null | http://arxiv.org/abs/1705.05552v1 | http://arxiv.org/pdf/1705.05552v1.pdf | IAN: The Individual Aggregation Network for Person Search | Person search in real-world scenarios is a new challenging computer version
task with many meaningful applications. The challenge of this task mainly comes
from: (1) unavailable bounding boxes for pedestrians and the model needs to
search for the person over the whole gallery images; (2) huge variance of
visual appearance of a particular person owing to varying poses, lighting
conditions, and occlusions. To address these two critical issues in modern
person search applications, we propose a novel Individual Aggregation Network
(IAN) that can accurately localize persons by learning to minimize intra-person
feature variations. IAN is built upon the state-of-the-art object detection
framework, i.e., faster R-CNN, so that high-quality region proposals for
pedestrians can be produced in an online manner. In addition, to relieve the
negative effect caused by varying visual appearances of the same individual,
IAN introduces a novel center loss that can increase the intra-class
compactness of feature representations. The engaged center loss encourages
persons with the same identity to have similar feature characteristics.
Extensive experimental results on two benchmarks, i.e., CUHK-SYSU and PRW, well
demonstrate the superiority of the proposed model. In particular, IAN achieves
77.23% mAP and 80.45% top-1 accuracy on CUHK-SYSU, which outperform the
state-of-the-art by 1.7% and 1.85%, respectively. | ['Kai-Zhu Huang', 'Jiashi Feng', 'Jimin Xiao', 'Tammam Tillo', 'Yunchao Wei', 'Yanchun Xie'] | 2017-05-16 | null | null | null | null | ['person-search'] | ['computer-vision'] | [-4.55346495e-01 -5.07115662e-01 7.88233355e-02 -2.91778475e-01
-4.72462624e-01 -2.31175467e-01 3.32591891e-01 -2.89398193e-01
-6.95595264e-01 6.54769719e-01 1.70849398e-01 3.07251215e-01
2.18477368e-01 -6.31093383e-01 -6.11707330e-01 -7.77511239e-01
1.87639281e-01 2.88155526e-01 3.96017611e-01 -1.06023811e-01
-2.42750823e-01 2.48799503e-01 -1.41990781e+00 -1.67427003e-01
8.39173913e-01 1.21775091e+00 2.08888903e-01 1.64362311e-01
2.01138109e-01 1.04029179e-01 -6.75148845e-01 -6.42464519e-01
5.34434736e-01 -7.47809783e-02 -2.47824132e-01 2.05389753e-01
6.03266358e-01 -4.84664768e-01 -8.68651390e-01 1.15120578e+00
8.41744363e-01 4.13822263e-01 4.63134408e-01 -1.19252741e+00
-1.00030637e+00 1.09455615e-01 -1.14309013e+00 2.96121538e-01
2.72621065e-01 4.11949545e-01 6.88966990e-01 -9.65891778e-01
1.61528438e-01 1.72437572e+00 4.88601476e-01 6.97686076e-01
-1.03792536e+00 -1.05987072e+00 5.79575896e-01 4.35596943e-01
-1.89950156e+00 -2.96033204e-01 4.86612529e-01 -1.90478623e-01
3.98705304e-01 3.41228366e-01 6.45586848e-01 8.60754490e-01
-2.32088372e-01 9.86226857e-01 8.58620107e-01 -1.37922376e-01
-1.37209028e-01 4.02839482e-01 1.81513190e-01 8.29594493e-01
5.48150897e-01 1.32166609e-01 -3.14758033e-01 -4.76657227e-02
7.99133241e-01 2.86888361e-01 -4.05125350e-01 -3.42734009e-01
-9.70990062e-01 6.40611887e-01 1.01137793e+00 -2.18308821e-01
-1.31864026e-01 4.39453200e-02 4.59539115e-01 -2.82836974e-01
3.00587445e-01 -1.35973617e-01 -1.76536307e-01 3.06025118e-01
-5.21961153e-01 5.05999506e-01 2.03832820e-01 1.17222261e+00
4.83326465e-01 -1.99558869e-01 -5.35362363e-01 1.02437270e+00
4.59505141e-01 7.80868709e-01 3.97985250e-01 -3.63054544e-01
6.53234065e-01 7.08740294e-01 3.48382592e-01 -1.11520851e+00
-1.59922183e-01 -8.45783770e-01 -1.05614841e+00 -9.66086239e-02
4.86577779e-01 -3.15329172e-02 -9.58340287e-01 1.73151672e+00
5.64521372e-01 1.58286557e-01 -1.95206061e-01 1.36406064e+00
1.17416918e+00 6.40541553e-01 2.09012568e-01 6.62868321e-02
1.69803441e+00 -1.24410236e+00 -4.33517873e-01 -4.70978916e-01
-4.92519215e-02 -5.96228182e-01 8.98403227e-01 -2.47152634e-02
-9.54328120e-01 -9.54309225e-01 -7.89600074e-01 7.03600869e-02
-3.31966460e-01 5.66336632e-01 3.50916654e-01 8.03109467e-01
-7.82410920e-01 -1.06008746e-01 -3.53229165e-01 -4.29279476e-01
6.96191728e-01 3.43507171e-01 -2.77297258e-01 -3.18278462e-01
-1.04844189e+00 5.35061598e-01 1.81374356e-01 3.66181046e-01
-6.25628471e-01 -6.91278338e-01 -7.56625831e-01 1.99762642e-01
4.96850222e-01 -7.95363784e-01 9.50659871e-01 -6.74394548e-01
-9.30342197e-01 8.29432786e-01 -4.20630366e-01 -2.82907844e-01
9.70325589e-01 -4.08937424e-01 -4.72433567e-01 -9.47035328e-02
3.66469860e-01 6.58838034e-01 5.46902180e-01 -1.31311119e+00
-8.62610042e-01 -7.32484877e-01 -2.62815297e-01 2.93255895e-01
-4.67174232e-01 3.07349086e-01 -1.25255013e+00 -6.45625532e-01
-7.06950799e-02 -9.18437243e-01 -2.85705149e-01 3.91063482e-01
-4.78550404e-01 -4.60836142e-01 7.56575048e-01 -8.04488242e-01
1.13469231e+00 -1.90137470e+00 -1.71727296e-02 1.36438996e-01
1.95846632e-01 5.08589208e-01 5.08955754e-02 -3.12972039e-01
1.65335000e-01 -1.06304593e-01 2.20764309e-01 -4.68433678e-01
-8.99172947e-03 -2.29902148e-01 -1.10029608e-01 7.13412046e-01
-1.54228536e-02 9.16670978e-01 -7.52391756e-01 -6.91618204e-01
2.43219003e-01 5.52816987e-01 -1.94449157e-01 2.05085680e-01
3.52983981e-01 2.75158346e-01 -5.26304901e-01 9.11080241e-01
1.01901591e+00 -1.92642957e-01 -3.96350473e-01 -2.57305890e-01
-2.97370236e-02 -3.65536898e-01 -1.42197227e+00 1.33613563e+00
-1.80968925e-01 3.70659590e-01 9.16394219e-02 -6.35107756e-01
9.14386034e-01 -5.01646213e-02 5.18553853e-02 -8.65781724e-01
9.97282416e-02 2.44261641e-02 -1.97118804e-01 -1.93755075e-01
3.65692109e-01 3.82471055e-01 -3.91064212e-02 2.10895389e-02
-3.20818037e-01 8.35540414e-01 2.04333067e-01 5.94896190e-02
5.17150283e-01 -1.11493781e-01 1.94220200e-01 -2.70607919e-01
8.87997031e-01 -5.42382658e-01 9.09214079e-01 8.31666052e-01
-6.95732355e-01 5.99618435e-01 -1.77488208e-01 -8.03761065e-01
-9.08128440e-01 -1.14912188e+00 -2.24061415e-01 9.59598243e-01
8.00484896e-01 -8.55519623e-02 -6.00442648e-01 -5.88332772e-01
1.80744290e-01 1.58437848e-01 -6.35537565e-01 -8.83170664e-02
-6.55288935e-01 -1.00899351e+00 3.37765783e-01 6.72273755e-01
1.07414234e+00 -8.12147975e-01 -3.66211116e-01 -3.79967466e-02
-3.15860122e-01 -1.26239455e+00 -9.32436168e-01 -6.63227260e-01
-1.73055708e-01 -9.31376874e-01 -1.31423354e+00 -9.44416881e-01
9.52183545e-01 8.77958655e-01 8.71868193e-01 2.15206578e-01
-6.69711888e-01 1.52029857e-01 -1.59648880e-01 -2.99674720e-01
4.05583113e-01 -1.83342874e-01 2.67225057e-01 3.09669316e-01
5.29181480e-01 -8.64159167e-02 -1.17320859e+00 8.92232120e-01
-3.51754159e-01 -6.48348704e-02 5.24365187e-01 8.01100552e-01
6.95892632e-01 3.06191355e-01 3.07457060e-01 -9.10043716e-02
4.28828359e-01 -4.22688574e-01 -6.95308328e-01 5.62042356e-01
-3.79145026e-01 -4.68915910e-01 5.25264204e-01 -5.53065479e-01
-1.08050406e+00 1.14004165e-01 2.75247157e-01 -4.21119034e-01
-7.27748275e-02 -3.65791202e-01 -4.56811845e-01 -2.37334058e-01
4.07815725e-01 5.15416980e-01 -1.86065108e-01 -4.36304718e-01
1.99441031e-01 6.35812461e-01 6.58025742e-01 -5.29759645e-01
1.05783451e+00 6.23752296e-01 -1.76550195e-01 -7.10325301e-01
-6.29922628e-01 -8.33216429e-01 -3.00923616e-01 -3.25780571e-01
8.33857417e-01 -1.43970680e+00 -9.39023912e-01 6.60774708e-01
-1.18227303e+00 1.24953575e-01 1.66310549e-01 2.43054628e-01
5.77525310e-02 5.28307199e-01 -3.68400961e-01 -1.14540434e+00
-6.47643626e-01 -1.18195546e+00 1.05188191e+00 1.01580465e+00
3.31457943e-01 -3.67531300e-01 -4.46332902e-01 5.64808607e-01
2.11007297e-01 1.24358103e-01 2.87127823e-01 -2.86282629e-01
-7.87358403e-01 -3.95408690e-01 -9.16366637e-01 1.30480409e-01
8.20162445e-02 -3.95675987e-01 -9.20942366e-01 -5.38435638e-01
-5.35532475e-01 -2.03258708e-01 1.01603937e+00 3.82732779e-01
1.35142815e+00 -2.54821986e-01 -7.23047018e-01 6.48449481e-01
1.23441255e+00 1.68126702e-01 4.30453628e-01 3.04593712e-01
7.22999036e-01 5.51563263e-01 6.58758044e-01 5.38097382e-01
5.72706401e-01 1.15253115e+00 2.27506772e-01 -3.01629364e-01
-3.06975782e-01 -3.54775161e-01 7.70086655e-03 -6.95914254e-02
-1.34075075e-01 -1.71408907e-01 -7.34689891e-01 6.18475616e-01
-2.20732069e+00 -1.04370975e+00 -8.65302905e-02 2.32263637e+00
4.93438244e-01 8.21146145e-02 4.84794796e-01 -3.68992716e-01
1.11587095e+00 4.93412875e-02 -5.98461986e-01 5.08403182e-01
-2.33118623e-01 -2.81519413e-01 5.95768809e-01 5.66074178e-02
-1.45185053e+00 8.45826864e-01 4.96355534e+00 1.02909946e+00
-5.82025409e-01 7.02796206e-02 1.01852250e+00 -3.07840496e-01
3.66494715e-01 -4.71625686e-01 -1.43736041e+00 9.45373654e-01
7.56357983e-02 -1.39875755e-01 3.40894610e-01 1.05864251e+00
2.62458354e-01 -2.67350841e-02 -7.48295069e-01 1.39343572e+00
2.38878578e-01 -1.00950313e+00 -5.00880554e-02 3.67607921e-02
6.97023034e-01 -2.62054414e-01 2.19223380e-01 4.28566486e-01
2.13535354e-01 -9.97283638e-01 7.09579647e-01 4.40661907e-01
6.33645535e-01 -9.59692955e-01 9.05408442e-01 2.22015232e-01
-1.83039534e+00 -2.88265973e-01 -7.70493150e-01 2.23831385e-01
2.62276620e-01 3.42959613e-01 -3.21477205e-01 4.23752815e-01
1.25514841e+00 3.92931521e-01 -8.36941302e-01 1.55962574e+00
-5.41144758e-02 8.42388272e-02 -4.76693332e-01 -1.74718902e-01
1.81509912e-01 -1.42926291e-01 4.63157147e-01 1.25493848e+00
1.12923086e-01 1.23820819e-01 6.00814402e-01 1.03192413e+00
-1.09166630e-01 1.88557461e-01 -1.50554717e-01 6.74375594e-01
5.48444629e-01 1.32891810e+00 -5.12571573e-01 -2.14913249e-01
-5.24976015e-01 1.19615555e+00 3.94291013e-01 5.44768810e-01
-1.06173658e+00 -1.89134851e-01 9.53963637e-01 2.30452821e-01
4.37095404e-01 1.93322115e-02 -3.12493630e-02 -1.09464848e+00
5.73887169e-01 -5.97340167e-01 4.66840506e-01 -5.32125652e-01
-1.52677178e+00 5.62194824e-01 -1.76338866e-01 -1.09461939e+00
3.79610330e-01 -4.17255163e-01 -7.60717928e-01 1.14638126e+00
-1.60085785e+00 -1.41150105e+00 -8.40899289e-01 6.63917124e-01
6.05300248e-01 -4.30523396e-01 4.24123287e-01 6.04112148e-01
-1.10238826e+00 1.13138783e+00 -4.86394167e-02 5.46076775e-01
7.64026344e-01 -9.09276128e-01 5.80871761e-01 9.78441656e-01
-2.81883121e-01 7.83557177e-01 3.82718503e-01 -6.47424042e-01
-1.09972560e+00 -1.38068628e+00 6.72549009e-01 -3.53545547e-01
2.17983052e-01 -4.72768277e-01 -6.35561168e-01 3.64088118e-01
-1.85985386e-01 5.15989363e-01 4.11387146e-01 6.05041310e-02
-4.44904745e-01 -4.27208930e-01 -1.16100788e+00 8.16626966e-01
1.25096512e+00 -1.31193548e-01 -1.11623563e-01 4.95200187e-01
6.15560353e-01 -4.22555029e-01 -3.73014390e-01 1.71784252e-01
6.49931848e-01 -9.17788744e-01 1.54470420e+00 -4.66271251e-01
-3.84818725e-02 -5.71121752e-01 -6.10023998e-02 -9.01610196e-01
-6.24291718e-01 -3.94376099e-01 -1.28630653e-01 1.33725727e+00
8.70809599e-04 -6.58957422e-01 7.66144574e-01 9.31134522e-01
2.63234466e-01 -9.67462420e-01 -1.05461192e+00 -9.71694350e-01
-2.68750906e-01 1.92321002e-01 7.09508061e-01 2.87662357e-01
-5.61260343e-01 1.68358207e-01 -6.54179931e-01 4.02004659e-01
1.08346295e+00 5.55948503e-02 1.05191219e+00 -1.08764589e+00
-1.43881276e-01 -4.88221765e-01 -6.10164523e-01 -1.40416157e+00
-2.17294786e-02 -4.48302180e-01 2.45306604e-02 -1.33768046e+00
7.56938279e-01 -6.40300870e-01 -3.30363184e-01 1.74448818e-01
-7.89454103e-01 2.34885097e-01 3.43163401e-01 3.20907682e-01
-7.87519813e-01 8.84361088e-01 1.10778618e+00 -2.73597091e-01
-8.91872048e-02 3.24889779e-01 -7.40268767e-01 6.61583543e-01
6.31492496e-01 -1.10328950e-01 -1.64198056e-01 -4.48692501e-01
-3.46368968e-01 -4.75182652e-01 9.03563619e-01 -1.15503597e+00
4.21166152e-01 2.71546580e-02 1.01531446e+00 -7.46169329e-01
5.14978528e-01 -7.77149141e-01 -1.72600180e-01 5.31810343e-01
4.64456826e-02 3.58904675e-02 1.56313539e-01 8.79828870e-01
-1.99058913e-02 1.32917404e-01 1.08160114e+00 -2.44603679e-01
-9.77347493e-01 6.88607097e-01 4.32575375e-01 4.56023365e-02
1.26163971e+00 -3.89517039e-01 -3.72868150e-01 -2.96341091e-01
-1.74033076e-01 7.37925351e-01 2.78158486e-01 6.23302758e-01
6.73349559e-01 -1.52903044e+00 -9.10729289e-01 1.10830054e-01
2.55807042e-01 1.04998007e-01 5.55535614e-01 5.18706441e-01
-2.23314390e-01 4.64694798e-01 -3.06388568e-02 -5.49742222e-01
-1.55586159e+00 4.86317605e-01 5.25415719e-01 -1.01455249e-01
-5.84717333e-01 1.21637416e+00 8.06352735e-01 -2.58368850e-01
5.45312285e-01 2.55194485e-01 -9.25485268e-02 -1.67099908e-01
1.00342226e+00 5.06177962e-01 -3.21394265e-01 -9.60896850e-01
-6.00862801e-01 7.31771290e-01 -4.49029595e-01 3.17194611e-01
9.59376514e-01 -3.10232431e-01 2.11487964e-01 -3.85396391e-01
1.01357579e+00 -2.70693123e-01 -1.42919350e+00 -5.70506215e-01
-6.01036251e-01 -9.67544317e-01 -7.37737343e-02 -7.61053085e-01
-1.21793199e+00 5.64298987e-01 1.01950514e+00 -2.85303205e-01
9.20474946e-01 4.89661023e-02 1.08278883e+00 1.72298938e-01
4.36518937e-01 -1.20086467e+00 1.37454525e-01 2.74274527e-04
8.81597757e-01 -1.49249828e+00 2.72177517e-01 -5.67772150e-01
-5.93522131e-01 6.76862299e-01 1.10575354e+00 4.57870029e-02
4.11491126e-01 -4.21329856e-01 -8.00333247e-02 1.19378567e-01
-1.27747431e-01 -3.20783824e-01 5.83183289e-01 6.63903832e-01
-5.37258424e-02 2.39229739e-01 -8.42927322e-02 8.75152051e-01
-2.67691538e-02 -3.15952539e-01 -1.36164933e-01 4.57963169e-01
-4.67108250e-01 -5.90845287e-01 -7.65852809e-01 2.94592649e-01
-1.37123317e-01 6.28787056e-02 -3.53273988e-01 7.69959211e-01
4.51558650e-01 1.00537968e+00 -3.61473151e-02 -2.41551191e-01
5.11772394e-01 -4.41799074e-01 2.30030358e-01 -2.32217580e-01
-4.26461190e-01 8.85027694e-04 -1.91725999e-01 -4.83680278e-01
-4.31128731e-03 -6.38780117e-01 -9.34087574e-01 -4.61112589e-01
-4.21798468e-01 -8.79492909e-02 2.74009943e-01 6.47415042e-01
3.05011541e-01 3.38375956e-01 5.24504960e-01 -6.92971885e-01
-6.21293724e-01 -7.69497454e-01 -4.50125933e-01 4.56658125e-01
1.45753652e-01 -9.00060892e-01 -5.00169583e-02 -2.39494771e-01] | [14.780203819274902, 0.8364280462265015] |
9f967cc5-0f96-4430-9a8f-3687038d5462 | facial-expression-recognition-and-image | 2208.06117 | null | https://arxiv.org/abs/2208.06117v1 | https://arxiv.org/pdf/2208.06117v1.pdf | Facial Expression Recognition and Image Description Generation in Vietnamese | This paper discusses a facial expression recognition model and a description generation model to build descriptive sentences for images and facial expressions of people in images. Our study shows that YOLOv5 achieves better results than a traditional CNN for all emotions on the KDEF dataset. In particular, the accuracies of the CNN and YOLOv5 models for emotion recognition are 0.853 and 0.938, respectively. A model for generating descriptions for images based on a merged architecture is proposed using VGG16 with the descriptions encoded over an LSTM model. YOLOv5 is also used to recognize dominant colors of objects in the images and correct the color words in the descriptions generated if it is necessary. If the description contains words referring to a person, we recognize the emotion of the person in the image. Finally, we combine the results of all models to create sentences that describe the visual content and the human emotions in the images. Experimental results on the Flickr8k dataset in Vietnamese achieve BLEU-1, BLEU-2, BLEU-3, BLEU-4 scores of 0.628; 0.425; 0.280; and 0.174, respectively. | ['Jugal Kalita', 'Loc Huu Nguy', 'Kim-Ngoc Thi Nguyen', 'Khang Nhut Lam'] | 2022-08-12 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [-4.30550069e-01 -8.10266957e-02 2.24602073e-01 -7.52926588e-01
-3.98549080e-01 -3.73037070e-01 4.26777482e-01 -3.55281353e-01
-3.66635740e-01 7.40204513e-01 2.46615306e-01 5.13217151e-01
4.94414270e-01 -6.14961028e-01 -5.54324448e-01 -8.07632089e-01
1.43394217e-01 -2.24562269e-02 -5.55680275e-01 -1.86210006e-01
-7.65805990e-02 6.47677958e-01 -1.50620234e+00 6.78550422e-01
1.81404397e-01 1.50295341e+00 -2.91783392e-01 8.09071004e-01
-1.74383283e-01 1.35334933e+00 -8.23232770e-01 -6.29182220e-01
-1.62379384e-01 -5.72549880e-01 -6.55728936e-01 5.85256279e-01
5.03469408e-01 -4.87892866e-01 -4.32323128e-01 1.15915322e+00
5.11484981e-01 2.53259480e-01 8.84041667e-01 -1.50496185e+00
-1.05665922e+00 3.03858548e-01 -6.15287066e-01 -4.99382198e-01
1.27780259e-01 -5.79495803e-02 5.43952763e-01 -9.46133971e-01
6.55872822e-01 1.44965506e+00 2.21957907e-01 1.18966091e+00
-6.20694757e-01 -1.07222283e+00 -5.59594296e-02 4.23567563e-01
-1.66579998e+00 -4.09068465e-01 6.23996377e-01 -2.12801442e-01
9.15311754e-01 2.04442486e-01 5.95628142e-01 1.03382087e+00
2.47356445e-01 8.35654497e-01 9.42644477e-01 -3.89871299e-01
-7.58790225e-02 4.38965321e-01 7.99610466e-03 9.88693774e-01
-3.21156800e-01 -2.84051180e-01 -4.65868771e-01 2.82229513e-01
6.87234044e-01 -9.23281312e-02 2.80422568e-02 3.68571877e-01
-7.74310172e-01 9.58956063e-01 6.56849623e-01 2.78763145e-01
-6.60913110e-01 5.27523220e-01 6.30722940e-01 1.57412618e-01
4.57369447e-01 9.51022655e-03 -1.08502343e-01 1.66841999e-01
-7.92946339e-01 1.26560524e-01 4.06239003e-01 1.32849288e+00
8.22054207e-01 4.90599394e-01 -2.95450419e-01 9.25628603e-01
3.00372988e-01 7.73289502e-01 3.95885289e-01 -1.06814742e+00
-1.68880016e-01 6.41627610e-01 2.96456784e-01 -1.36992848e+00
-2.85721064e-01 5.05045913e-02 -1.05002999e+00 5.66827431e-02
-9.90542024e-02 -3.65023404e-01 -1.32007039e+00 1.79365599e+00
1.66116643e-05 -8.78941640e-02 5.91741621e-01 1.20590031e+00
1.77593815e+00 1.26279306e+00 3.73869121e-01 -4.39146273e-02
1.48846817e+00 -1.27930140e+00 -1.15565634e+00 -3.86346914e-02
4.38049644e-01 -6.65910125e-01 8.24091673e-01 4.16519612e-01
-9.36539590e-01 -8.64953279e-01 -7.10657060e-01 -1.68281987e-01
-4.54536498e-01 8.78278255e-01 4.88168865e-01 4.71735507e-01
-1.41364884e+00 6.20791577e-02 -4.45216596e-02 -5.11330783e-01
2.38956735e-01 1.79690391e-01 -4.93096262e-01 -2.16526352e-03
-1.25155509e+00 7.35070288e-01 4.98274207e-01 3.87292236e-01
-1.25428438e+00 5.86395040e-02 -9.44858611e-01 -5.58404485e-03
-1.59914091e-01 -1.61228403e-01 1.06589913e+00 -1.83477509e+00
-1.32167089e+00 1.15145814e+00 -9.36370715e-02 -2.64220059e-01
1.68785259e-01 4.15604413e-02 -7.31260419e-01 4.84308720e-01
-1.01836443e-01 1.45563149e+00 6.33700192e-01 -1.35778546e+00
-6.57610476e-01 -3.19220334e-01 4.13482636e-02 2.61930525e-01
-1.67596579e-01 2.75359541e-01 -7.81893194e-01 -2.55281597e-01
-4.65797395e-01 -7.56651521e-01 3.70988771e-02 -2.67384690e-03
-3.68185788e-01 -4.60319191e-01 9.19190824e-01 -8.95821631e-01
9.11466599e-01 -2.44595075e+00 -2.66373962e-01 2.16123208e-01
-5.20579936e-03 2.99499184e-01 -4.66658622e-01 1.88557610e-01
-2.01158989e-02 9.10158083e-02 2.63051748e-01 -1.38897181e-01
6.98615760e-02 2.35086530e-01 -1.01958871e-01 2.44405344e-01
3.22639287e-01 9.19318199e-01 -6.18255377e-01 -6.17948830e-01
1.61724135e-01 8.15300465e-01 -2.28646711e-01 4.61284786e-01
-1.84749469e-01 -3.72594967e-02 -3.66449147e-01 6.63145125e-01
6.99646294e-01 -4.95199151e-02 -1.60874918e-01 -3.98607552e-01
1.96447581e-01 -7.20016360e-01 -8.68030131e-01 1.34464383e+00
-4.57996547e-01 8.16759884e-01 -8.76130164e-02 -5.58507323e-01
1.48634493e+00 5.67364454e-01 1.77200377e-01 -6.29275560e-01
5.69619954e-01 -1.23205557e-02 -4.23292965e-01 -7.93821156e-01
7.32493758e-01 -3.70659202e-01 -3.29941928e-01 2.02631399e-01
3.91548485e-01 1.76862422e-02 3.47727597e-01 1.96773440e-01
4.44662541e-01 1.00098746e-02 1.04338720e-01 1.17972896e-01
7.17937827e-01 1.72844976e-02 3.55592847e-01 5.07891476e-01
-2.87265450e-01 5.46916664e-01 5.13370156e-01 -6.85913205e-01
-9.60256219e-01 -7.76997805e-01 3.15671206e-01 9.51514065e-01
4.28450592e-02 -3.22695404e-01 -9.65486944e-01 -5.83710611e-01
-3.89976650e-01 8.20410192e-01 -8.03023279e-01 -2.62819827e-01
1.02907673e-01 -5.18409550e-01 7.81547248e-01 4.85346794e-01
1.06288147e+00 -1.64165235e+00 -5.66723287e-01 7.52910832e-03
-5.21423638e-01 -1.23486400e+00 -2.20477253e-01 -3.16243470e-01
-1.94651306e-01 -7.21918583e-01 -8.00728023e-01 -1.12787819e+00
8.99487674e-01 -9.61160883e-02 8.33888829e-01 9.62923169e-02
-3.54470253e-01 4.11011487e-01 -7.32908368e-01 -3.81388217e-01
-2.48729333e-01 -6.75274014e-01 -1.66449428e-01 3.80203307e-01
7.73086011e-01 3.12411219e-01 -3.79564673e-01 4.34061065e-02
-8.01425457e-01 1.61883697e-01 3.94667804e-01 7.74863541e-01
4.78154212e-01 3.43229957e-02 2.98240066e-01 -4.50520456e-01
6.65091634e-01 -1.23059876e-01 -1.84020564e-01 4.09397572e-01
-2.64033586e-01 -3.94404203e-01 5.89534581e-01 -4.21420902e-01
-1.23330331e+00 3.11523050e-01 -1.28253385e-01 -7.18861282e-01
-4.63683516e-01 2.43391722e-01 -6.14232533e-02 1.80924892e-01
2.87738025e-01 4.19447303e-01 -7.28407130e-02 1.09180540e-01
3.98684263e-01 1.07309985e+00 5.58128417e-01 -1.25633612e-01
6.29792437e-02 2.79092431e-01 -3.56701672e-01 -1.14322412e+00
-6.78069413e-01 -1.67013094e-01 -2.92565525e-01 -9.17006254e-01
1.40851831e+00 -1.28230512e+00 -8.81574035e-01 6.77367389e-01
-1.40715694e+00 -7.95094483e-03 8.67579803e-02 3.31249595e-01
-5.14722824e-01 1.66590177e-02 -9.07693923e-01 -1.01773524e+00
-7.83338010e-01 -1.13685262e+00 1.08839834e+00 6.44160330e-01
-2.10649878e-01 -6.76728189e-01 -5.83199501e-01 2.78943568e-01
2.57601440e-01 4.10670549e-01 6.86517894e-01 -3.21376681e-01
-4.43439446e-02 -3.00401628e-01 -6.52372360e-01 7.50457883e-01
-5.42070642e-02 4.20942187e-01 -1.08646119e+00 9.94174704e-02
-2.99651921e-01 -8.45193684e-01 8.18063557e-01 4.30457622e-01
1.27162719e+00 -5.39992154e-01 2.44127661e-01 4.79646206e-01
1.51261330e+00 6.31212831e-01 9.92120206e-01 6.21024892e-03
5.97414136e-01 8.18299949e-01 7.25951195e-01 5.66700935e-01
3.70037496e-01 4.04153973e-01 3.71914685e-01 -5.45134723e-01
-3.94028425e-02 -1.63984984e-01 5.77537477e-01 5.52681565e-01
-1.26165152e-01 -4.87979352e-01 -6.08761191e-01 3.83524925e-01
-1.58827114e+00 -1.07519031e+00 -3.64076644e-01 1.36527205e+00
5.99946916e-01 -5.59310555e-01 -6.85994476e-02 -2.70133644e-01
8.52983892e-01 2.06894740e-01 -1.75538823e-01 -1.08950496e+00
-3.53480399e-01 -1.62781496e-02 2.15501025e-01 3.86105835e-01
-1.14738834e+00 1.47905064e+00 5.75777245e+00 7.39005804e-01
-1.39384508e+00 -8.88642594e-02 1.15648162e+00 -6.99397624e-02
1.39928520e-01 -6.89823925e-01 -7.22899079e-01 2.46892363e-01
8.40217412e-01 -7.08101913e-02 3.12892646e-01 9.34146762e-01
3.21155310e-01 -4.12773602e-02 -5.89085996e-01 1.53522837e+00
7.51245558e-01 -9.75950122e-01 6.57532334e-01 -3.33833009e-01
8.16728771e-01 -4.38120604e-01 1.13494478e-01 2.96729535e-01
2.71695733e-01 -1.26192975e+00 9.04574215e-01 8.15281034e-01
9.32294548e-01 -1.14066207e+00 1.21708834e+00 -1.16839901e-01
-8.50180209e-01 1.33938968e-01 -8.02903533e-01 1.29946485e-01
-7.35856518e-02 1.75847352e-01 -8.97557497e-01 2.19179109e-01
9.88049269e-01 7.43936121e-01 -5.35444200e-01 3.82476538e-01
-5.64333498e-01 3.33976567e-01 1.92137942e-01 -6.22242033e-01
7.38779604e-01 -1.21468648e-01 -1.18950382e-01 1.55627728e+00
4.52993304e-01 3.61179441e-01 -2.09223777e-01 8.89635742e-01
-2.43915945e-01 4.86761719e-01 -4.48578537e-01 -3.24440002e-01
6.42562434e-02 1.63897967e+00 -4.81235087e-01 -6.57613337e-01
-1.52824312e-01 1.31068540e+00 1.27494056e-02 6.40845776e-01
-1.09418213e+00 -5.03390372e-01 4.40317839e-01 -4.68410224e-01
2.71854669e-01 2.80291319e-01 2.76515484e-01 -9.21334743e-01
-3.75119001e-01 -9.12913322e-01 3.29307288e-01 -1.70943952e+00
-1.12646103e+00 1.09587669e+00 -2.41738394e-01 -9.98188078e-01
-2.23841563e-01 -7.80624688e-01 -3.32600564e-01 8.30366969e-01
-1.23429167e+00 -1.40972269e+00 -7.11423337e-01 7.31589377e-01
5.30264616e-01 -3.47798705e-01 1.10354853e+00 1.79570884e-01
-5.01355171e-01 4.88543928e-01 -2.75133193e-01 6.67109132e-01
7.35590696e-01 -8.08767855e-01 -4.93600108e-02 6.06519997e-01
-8.77241511e-03 3.86819214e-01 4.92246896e-01 -5.01598477e-01
-1.09504247e+00 -1.38321340e+00 1.16649973e+00 1.61250621e-01
1.81187183e-01 -1.43783495e-01 -4.56706017e-01 6.04840696e-01
6.51831210e-01 -1.10016063e-01 8.02764118e-01 -3.54765385e-01
-2.42084041e-01 -2.95733035e-01 -1.34647560e+00 5.01432002e-01
4.25529361e-01 -3.15692365e-01 -4.09392081e-02 2.07916439e-01
3.42681438e-01 -2.15390727e-01 -7.34431624e-01 8.15223232e-02
7.58119226e-01 -7.90675819e-01 5.03443658e-01 -7.11687684e-01
8.55247557e-01 -3.44919741e-01 -4.24829870e-01 -1.15333104e+00
-3.57763886e-01 5.74312843e-02 6.47003472e-01 1.42024326e+00
4.20397282e-01 5.92084303e-02 5.38658619e-01 4.84224409e-01
1.13973200e-01 -2.80446082e-01 -6.46010041e-01 -2.95342326e-01
-2.95060605e-01 -2.23024487e-01 5.67669034e-01 8.95215273e-01
-2.41849601e-01 4.31015134e-01 -9.52863395e-01 -1.66936696e-01
3.59009445e-01 -8.42213556e-02 8.05627882e-01 -5.52420080e-01
3.61754924e-01 -2.01168075e-01 -5.84527731e-01 -6.02002919e-01
4.96730834e-01 -7.39806890e-01 1.29647240e-01 -1.73897696e+00
5.09533584e-01 2.10472807e-01 -2.74935633e-01 8.28921974e-01
1.80052355e-01 7.67643154e-01 3.83349031e-01 -2.18239069e-01
-7.36745715e-01 5.93656242e-01 1.33980942e+00 -4.06967342e-01
1.28142819e-01 -6.04428351e-01 -6.44546151e-01 7.72338510e-01
9.62018847e-01 -7.35663325e-02 -1.62338689e-01 -5.36618888e-01
5.73613420e-02 2.19405610e-02 3.46644849e-01 -8.21649909e-01
-9.20931026e-02 -4.42346707e-02 1.10376203e+00 -7.11685956e-01
5.78156888e-01 -9.48752761e-01 2.41213143e-01 4.74744111e-01
-5.74805558e-01 1.21158667e-01 2.24824339e-01 1.19267330e-01
-5.98948300e-01 -1.65023580e-01 9.83024776e-01 -3.50413948e-01
-1.38875639e+00 1.92186594e-01 -6.50768757e-01 -4.80003685e-01
1.19156623e+00 -2.30042517e-01 -1.47707701e-01 -1.13104832e+00
-9.34900761e-01 2.71003157e-01 1.72656372e-01 5.51822186e-01
1.21879470e+00 -1.75298405e+00 -9.27018523e-01 2.26529509e-01
4.07970876e-01 -5.37819982e-01 5.72052300e-01 2.70107180e-01
-8.20311368e-01 2.58192420e-01 -5.95869541e-01 -2.97337055e-01
-1.74039388e+00 3.35730672e-01 5.07597327e-01 2.74929762e-01
-4.75692749e-02 1.02989244e+00 3.99627358e-01 -6.37434274e-02
2.86124408e-01 -3.77251916e-02 -5.95086813e-01 2.68485308e-01
7.95994222e-01 1.34023070e-01 -4.34869707e-01 -1.42058289e+00
-3.61256123e-01 6.47570550e-01 -1.54789761e-02 -6.02297671e-02
1.22031128e+00 -6.98518157e-02 -3.09073567e-01 3.27524334e-01
1.37897110e+00 -4.67998087e-01 -7.73548782e-01 9.44347158e-02
-5.85857511e-01 -1.25415027e-01 -4.45568375e-02 -1.13260019e+00
-1.55789804e+00 1.03555501e+00 7.30934083e-01 -1.96233958e-01
1.61415517e+00 -2.41821390e-02 6.66227400e-01 1.57377645e-01
1.48485810e-01 -1.29874408e+00 2.45310664e-01 5.65625012e-01
1.12692130e+00 -1.05400801e+00 -2.98272520e-01 -8.36654976e-02
-1.39830053e+00 1.40111649e+00 8.36023927e-01 -2.00389862e-01
2.27203161e-01 -1.18162213e-02 5.58938861e-01 -3.11272562e-01
-7.92741179e-01 -1.94738954e-01 2.85338461e-01 4.10115272e-01
4.48504716e-01 2.79418826e-01 -3.87491167e-01 6.37392700e-01
-1.98797211e-01 8.53603408e-02 6.05313599e-01 4.32427555e-01
-5.32371342e-01 -5.22602260e-01 -3.56441796e-01 4.55417223e-02
-6.34017944e-01 1.02350235e-01 -1.01880670e+00 8.84217143e-01
2.67692953e-01 1.18604386e+00 2.86320150e-01 -8.04421961e-01
3.39865416e-01 1.52528316e-01 4.40851539e-01 -2.60996342e-01
-4.93790865e-01 3.42501193e-01 3.03917646e-01 -4.97807086e-01
-7.70954788e-01 -8.88226628e-02 -1.49457538e+00 -2.78801858e-01
-7.12032393e-02 2.72780895e-01 6.24951720e-01 4.63539034e-01
1.25637293e-01 2.97572106e-01 5.97757041e-01 -3.79725069e-01
1.26074944e-02 -1.08791399e+00 -1.12018490e+00 6.29418731e-01
6.75140396e-02 -2.77100891e-01 -2.76521713e-01 4.75507826e-01] | [13.282247543334961, 4.98299503326416] |
d4511e11-08ab-4c36-8511-1c291b3a5195 | gfte-graph-based-financial-table-extraction | 2003.07560 | null | https://arxiv.org/abs/2003.07560v1 | https://arxiv.org/pdf/2003.07560v1.pdf | GFTE: Graph-based Financial Table Extraction | Tabular data is a crucial form of information expression, which can organize data in a standard structure for easy information retrieval and comparison. However, in financial industry and many other fields tables are often disclosed in unstructured digital files, e.g. Portable Document Format (PDF) and images, which are difficult to be extracted directly. In this paper, to facilitate deep learning based table extraction from unstructured digital files, we publish a standard Chinese dataset named FinTab, which contains more than 1,600 financial tables of diverse kinds and their corresponding structure representation in JSON. In addition, we propose a novel graph-based convolutional neural network model named GFTE as a baseline for future comparison. GFTE integrates image feature, position feature and textual feature together for precise edge prediction and reaches overall good results. | ['Zheng Huang', 'Yiren Li', 'Xianhui Liu', 'Junchi Yan', 'Yi Zhou', 'Fan Ye'] | 2020-03-17 | null | null | null | null | ['table-extraction'] | ['miscellaneous'] | [-3.23083133e-01 -2.89764494e-01 -4.59634364e-01 -2.19594091e-01
-4.41734701e-01 -6.09749973e-01 2.38392591e-01 6.59448743e-01
-2.13461727e-01 9.60937202e-01 4.44988251e-01 -6.68426812e-01
-2.92197578e-02 -1.31387258e+00 -7.43937492e-01 -1.50685102e-01
1.36182848e-02 3.90487134e-01 2.67967582e-02 -3.18520784e-01
3.57529074e-01 3.89977038e-01 -9.43452060e-01 7.17612565e-01
8.54641259e-01 1.62199497e+00 -7.67410547e-02 1.10477008e-01
-1.04600072e+00 1.41618359e+00 -7.02760160e-01 -9.21419322e-01
1.53334096e-01 -1.29993707e-01 -6.66142762e-01 1.85655840e-02
2.10343748e-01 -4.08398807e-01 -8.31441045e-01 1.08876395e+00
3.09103429e-01 -4.20093864e-01 5.80812871e-01 -1.17027402e+00
-1.30252004e+00 1.08495486e+00 -7.32336640e-01 1.28390286e-02
4.32827085e-01 -3.14482450e-01 1.12376761e+00 -7.09097505e-01
1.00359249e+00 1.07071173e+00 5.84009767e-01 -1.40053183e-01
-5.37699640e-01 -6.97864056e-01 1.18912935e-01 4.26949680e-01
-1.17880034e+00 -1.53842226e-01 1.00032258e+00 -3.39766026e-01
8.92594337e-01 2.14649111e-01 7.51249611e-01 7.98306942e-01
6.94729030e-01 8.52297604e-01 5.33044457e-01 -6.03649803e-02
-2.58681566e-01 1.27100796e-01 -9.83907357e-02 6.04290128e-01
8.21602464e-01 -6.19694591e-01 -4.71603692e-01 4.32623774e-01
4.60984409e-01 1.21175483e-01 -1.59154490e-01 -3.82919252e-01
-1.35717165e+00 6.71624124e-01 8.09311867e-01 2.67920136e-01
-7.67922029e-02 -8.42819065e-02 1.04616237e+00 4.86290157e-01
3.51354539e-01 2.18476623e-01 -4.75110441e-01 -9.05402761e-04
-4.53883171e-01 1.29115760e-01 8.21613669e-01 1.42050195e+00
4.75546539e-01 -2.36829068e-03 -3.60607386e-01 6.21489704e-01
1.01346813e-01 3.12038511e-01 4.54933077e-01 -3.80763292e-01
1.46564639e+00 1.22215879e+00 -1.12516642e-01 -1.69128013e+00
-4.65737164e-01 -3.93965632e-01 -1.37364459e+00 -3.23996454e-01
1.31287023e-01 -6.65870681e-02 -5.68896651e-01 7.04594433e-01
-2.93386215e-03 -6.98838830e-01 -1.63342237e-01 6.18932068e-01
1.43414879e+00 9.55299735e-01 -3.53582919e-01 -2.29131654e-01
1.43826973e+00 -8.23735595e-01 -1.16521025e+00 7.85383657e-02
8.02323937e-01 -7.33325899e-01 9.63597298e-01 5.97676575e-01
-9.83770549e-01 -4.28765863e-01 -1.20747113e+00 -7.20501781e-01
-1.08952928e+00 1.00141928e-01 9.31483269e-01 4.99870777e-01
-8.37965012e-01 5.52305818e-01 -2.86844283e-01 1.40935168e-01
9.41960156e-01 1.43116206e-01 -6.07025445e-01 8.45870376e-03
-1.24650347e+00 3.26678008e-01 7.28513598e-01 1.75331235e-01
1.41441181e-01 -4.03570831e-01 -1.02560651e+00 4.09964651e-01
4.71510738e-01 -4.95966613e-01 9.07236874e-01 -2.86059827e-01
-8.95098805e-01 7.55551159e-01 2.24886209e-01 -5.21390855e-01
5.63885987e-01 9.51500311e-02 -7.27000058e-01 -3.39286588e-02
-2.34269053e-02 2.30954796e-01 5.08116901e-01 -9.63221192e-01
-5.28913438e-01 -3.93697828e-01 -1.25936002e-01 6.72849566e-02
-8.59875083e-01 1.91213433e-02 -9.38568473e-01 -9.75152612e-01
9.12625343e-02 -2.08994657e-01 -7.32171396e-03 8.66562873e-02
-9.54751790e-01 -2.40289852e-01 7.40332127e-01 -9.11498249e-01
1.96289599e+00 -1.86212850e+00 -2.47172356e-01 1.64003998e-01
5.99071622e-01 -3.58131714e-02 2.99711168e-01 4.74983037e-01
3.90681811e-02 4.90046978e-01 -1.28262550e-01 -1.59374140e-02
1.99678391e-01 -1.53474525e-01 -3.24911386e-01 -5.67086041e-02
1.29891157e-01 1.36591399e+00 -5.33273220e-01 -1.02062953e+00
-2.09926799e-01 3.27164859e-01 -2.82851845e-01 -1.06331207e-01
-1.69572398e-01 -1.85643081e-02 -8.22433770e-01 9.70969498e-01
7.65141189e-01 -6.83989227e-01 2.03670353e-01 -4.83279169e-01
-8.58065933e-02 5.53880095e-01 -1.01999032e+00 1.32120311e+00
-2.38555282e-01 7.65384912e-01 -2.34359831e-01 -1.03372359e+00
1.13307500e+00 -4.92364541e-03 6.76928759e-01 -1.13885379e+00
4.72314358e-01 2.04501942e-01 -2.76254058e-01 -2.07474083e-01
9.38218474e-01 5.31567693e-01 -3.73119414e-01 1.84369355e-01
-2.77643144e-01 8.86177719e-02 6.41204000e-01 5.63890576e-01
6.95748568e-01 -1.10793293e-01 3.02052826e-01 -1.71617307e-02
4.71246064e-01 1.35232374e-01 5.15574098e-01 3.31969738e-01
2.52757013e-01 6.79071069e-01 1.03023887e+00 -7.61327088e-01
-1.06731999e+00 -5.98658323e-01 -3.54813814e-01 2.93871820e-01
8.40619951e-02 -6.59555078e-01 -5.59909046e-01 -7.51663566e-01
2.71902144e-01 1.77077293e-01 -5.32032430e-01 8.65033269e-03
-6.11435413e-01 -5.73483586e-01 2.22754776e-02 6.43102109e-01
8.83296967e-01 -1.26718211e+00 7.38349780e-02 4.30556923e-01
-1.36673257e-01 -1.06720614e+00 -8.30728531e-01 2.24648863e-01
-7.32033610e-01 -1.05037498e+00 -5.69582224e-01 -9.22517598e-01
4.96758699e-01 8.93800929e-02 1.42624056e+00 1.99646637e-01
-7.83659071e-02 -3.69031966e-01 -3.82835150e-01 -4.04477030e-01
-4.29702476e-02 4.10041422e-01 -4.22943115e-01 -2.04338163e-01
3.93569589e-01 -1.39579579e-01 -4.48246866e-01 -1.33705020e-01
-1.09251535e+00 1.35618165e-01 4.89422321e-01 7.94323623e-01
7.02091038e-01 3.15547407e-01 3.46011400e-01 -1.24670410e+00
9.27292466e-01 -4.52158928e-01 -9.32009339e-01 4.31830317e-01
-8.67053151e-01 2.24882126e-01 1.02390957e+00 1.27621308e-01
-7.47335136e-01 -3.64747077e-01 -8.12314004e-02 -1.69655979e-01
3.30240577e-01 1.36542726e+00 -5.57039738e-01 7.41237551e-02
2.12119535e-01 3.19650739e-01 -2.48492867e-01 -7.25394726e-01
-4.46020365e-02 9.10220921e-01 4.82187361e-01 -1.21570088e-01
8.42758834e-01 -2.10259852e-04 3.76541689e-02 -1.28739521e-01
-8.32668662e-01 -4.04328406e-02 -7.22067654e-01 -8.18275437e-02
7.46567369e-01 -8.88370335e-01 -1.02319467e+00 5.07480264e-01
-1.00371325e+00 7.13338703e-03 1.98810145e-01 1.82234555e-01
-1.59885168e-01 4.73946631e-01 -9.60611641e-01 -2.84758598e-01
-5.19785881e-01 -8.92053604e-01 8.24797809e-01 2.55036503e-01
3.04116249e-01 -9.78589833e-01 -4.42985922e-01 3.20922911e-01
3.07126313e-01 4.22715127e-01 1.12912583e+00 -5.92530429e-01
-9.69841182e-01 -4.12677675e-01 -7.76160061e-01 1.14601548e-03
3.50915015e-01 7.02062100e-02 -3.57677430e-01 -6.22903407e-02
-4.10740793e-01 -2.96782196e-01 1.17404723e+00 5.59410527e-02
1.98805082e+00 -6.61123395e-01 -4.29572403e-01 9.45332170e-01
1.50498235e+00 8.12386632e-01 1.06718755e+00 9.74922359e-01
1.20279789e+00 3.80987406e-01 4.23369229e-01 6.52622342e-01
6.82835221e-01 4.05806929e-01 2.64760435e-01 4.79543880e-02
2.29390115e-01 -4.80621517e-01 -1.79010332e-02 1.27939248e+00
2.13802531e-01 -8.53573799e-01 -9.40554738e-01 2.89965183e-01
-1.57568812e+00 -8.85896683e-01 -2.07239583e-01 1.96417212e+00
9.01369870e-01 6.65635288e-01 -1.70841187e-01 3.95467728e-01
6.67050302e-01 2.51470983e-01 -4.59644616e-01 -2.71349818e-01
-4.86480892e-01 -1.59174681e-01 7.91831136e-01 -1.60958037e-01
-1.31474912e+00 4.89345938e-01 5.44955349e+00 8.34302545e-01
-1.07161021e+00 -5.10541856e-01 1.33753622e+00 2.81409472e-01
-5.73559642e-01 -4.75329250e-01 -8.89517009e-01 9.45298791e-01
6.31821275e-01 -6.79394066e-01 1.91296384e-01 8.42523873e-01
-2.43245512e-01 4.05108273e-01 -6.41188025e-01 1.31808352e+00
-2.16628522e-01 -2.21696734e+00 6.45911276e-01 2.12356612e-01
5.63306808e-01 -4.26127672e-01 2.77170599e-01 3.06194603e-01
5.41787073e-02 -1.16515183e+00 6.64357305e-01 4.26729113e-01
1.14030874e+00 -1.02688372e+00 1.09715879e+00 -3.02681744e-01
-1.53320444e+00 -1.51403815e-01 -5.20370364e-01 2.33386278e-01
-5.29493503e-02 5.94529033e-01 -3.12960923e-01 1.08144641e+00
9.82195377e-01 1.28975666e+00 -7.80739486e-01 1.25438559e+00
2.85222113e-01 2.78922379e-01 3.95809039e-02 -3.93352449e-01
2.95142442e-01 -3.83019596e-01 -9.79511887e-02 1.14688241e+00
6.57358468e-01 -1.02937341e-01 5.44344783e-02 5.02208829e-01
-9.44973350e-01 4.92121935e-01 -6.50351167e-01 -5.39434731e-01
3.66658121e-01 1.33441222e+00 -1.09675598e+00 -3.79384309e-01
-8.17475021e-01 5.42045772e-01 4.53229576e-01 1.06412768e-01
-7.30233550e-01 -1.12937653e+00 2.16070890e-01 1.84438750e-01
5.72238326e-01 -1.39498472e-01 -5.73489189e-01 -1.54866433e+00
4.53354537e-01 -7.98595905e-01 7.63276041e-01 -9.69541252e-01
-1.32569969e+00 8.49513352e-01 -2.13749766e-01 -1.83273661e+00
-1.81172803e-01 -1.09492397e+00 -3.62424970e-01 6.49850070e-01
-1.45554066e+00 -7.40129769e-01 -4.31985229e-01 6.33882344e-01
6.65781200e-02 -5.51115751e-01 3.06217492e-01 7.24243701e-01
-8.31032395e-01 7.88916767e-01 6.98574722e-01 1.08176351e+00
5.82078457e-01 -1.40785098e+00 8.87170434e-01 5.22334635e-01
3.21741670e-01 5.08317828e-01 2.46618122e-01 -8.62167358e-01
-1.67599225e+00 -1.20559430e+00 1.06682515e+00 -1.21279299e-01
9.42900836e-01 -5.53052187e-01 -1.11103058e+00 6.06224716e-01
4.40445751e-01 3.39100882e-02 3.06459963e-01 -2.15757519e-01
-3.52778852e-01 -5.46073616e-01 -8.11992049e-01 5.76982796e-01
9.83612061e-01 -4.91424501e-01 -1.84565410e-01 4.66432869e-01
1.00741529e+00 -5.63213468e-01 -1.06127346e+00 7.72460103e-02
3.59678209e-01 -8.12767804e-01 7.63493955e-01 -5.25573075e-01
9.32024479e-01 -2.00085551e-01 1.36469409e-01 -1.05854297e+00
-3.70629340e-01 -4.97578472e-01 -5.38295984e-01 1.34089303e+00
6.45655990e-01 -5.99239171e-01 1.02476513e+00 1.29875809e-01
-8.61725062e-02 -8.47666264e-01 -7.37582803e-01 -7.47833610e-01
6.28601164e-02 -2.44761392e-01 1.16085970e+00 9.98159587e-01
2.08010837e-01 4.74409431e-01 -3.34187686e-01 -5.30612588e-01
2.55591393e-01 4.99092817e-01 4.91010845e-01 -1.50441766e+00
3.80551279e-01 -6.68226123e-01 -5.01537204e-01 -9.87778485e-01
-2.21865654e-01 -9.52798545e-01 -5.15836239e-01 -2.22583270e+00
1.97277233e-01 -3.39491904e-01 -5.41533351e-01 3.23155344e-01
8.47782344e-02 1.54011771e-01 -1.05254509e-01 1.88847929e-01
-6.70961380e-01 6.13611341e-01 1.67991865e+00 -6.64988160e-01
9.82513577e-02 -4.03545678e-01 -9.81683373e-01 1.47559002e-01
6.48360312e-01 -3.37123811e-01 -1.59992158e-01 -5.22098660e-01
8.93363118e-01 3.64191204e-01 -2.22965524e-01 -8.87664497e-01
3.65545571e-01 -2.03159839e-01 9.54033077e-01 -1.14635444e+00
-2.33457088e-01 -9.05832112e-01 -1.24879768e-02 2.10680485e-01
-2.71688700e-01 4.73654419e-01 1.98817238e-01 3.97049516e-01
-7.91314065e-01 -1.24555575e-02 2.09077522e-02 -1.46101400e-01
-6.88400388e-01 9.80075657e-01 -3.45729068e-02 1.04635119e-01
5.97180247e-01 -3.00619930e-01 -6.97976530e-01 -5.41905940e-01
-1.68143600e-01 3.62470835e-01 2.59993702e-01 5.81837058e-01
7.89472342e-01 -1.71281660e+00 -6.08815491e-01 1.81077883e-01
3.70515287e-01 9.33848172e-02 1.25038892e-01 5.95801890e-01
-1.01379704e+00 8.38467479e-01 -4.26589340e-01 -2.06382424e-02
-8.13094318e-01 1.00400341e+00 -3.41199264e-02 -6.12107158e-01
-7.80371904e-01 5.36813438e-01 -6.53572455e-02 -2.65739381e-01
3.07060540e-01 -6.54139400e-01 -6.78899586e-01 4.22701538e-01
7.26532638e-01 -1.27086103e-01 4.02436018e-01 -3.47077936e-01
-1.32061794e-01 2.56004840e-01 -4.84210163e-01 4.37368333e-01
1.36644495e+00 -1.26028270e-01 -4.36550051e-01 3.23461592e-01
1.46384168e+00 2.89129555e-01 -9.09576714e-01 -5.66432253e-02
3.03470582e-01 -5.06647587e-01 -1.43576320e-02 -4.44214404e-01
-1.69277120e+00 7.75778949e-01 -1.00180851e-02 6.51679397e-01
1.19387019e+00 -3.03462565e-01 1.12609303e+00 9.30739999e-01
4.20151651e-01 -1.15430546e+00 -6.36733174e-02 5.60676396e-01
8.78435731e-01 -1.20656884e+00 3.02633464e-01 -4.13002104e-01
-5.52166760e-01 1.48003006e+00 5.30876458e-01 3.27667773e-01
7.49361396e-01 4.84278262e-01 1.40476540e-01 -2.56127447e-01
-7.54249930e-01 1.80339590e-01 5.94676852e-01 4.79336649e-01
7.57579327e-01 -2.92923152e-01 -2.24697307e-01 8.95032287e-01
-4.82726574e-01 -4.90980186e-02 5.30745089e-01 8.50929618e-01
-2.80066073e-01 -9.82179761e-01 -1.31149694e-01 1.14136243e+00
-7.46482730e-01 -3.83726507e-01 -4.60514694e-01 8.54208469e-01
-2.49648333e-01 5.67998230e-01 4.11529779e-01 -5.01037955e-01
2.42864370e-01 -1.98919013e-01 -6.50165305e-02 -1.55783266e-01
-6.73235774e-01 -3.25430572e-01 1.18026927e-01 -2.64661252e-01
-7.16572162e-03 -4.56732363e-01 -1.12481523e+00 -5.55161059e-01
3.62498090e-02 1.71037033e-01 2.74737835e-01 5.82105279e-01
2.14581639e-01 7.95298278e-01 6.40144110e-01 -1.73010901e-01
-6.73029572e-03 -6.66833043e-01 -7.55836487e-01 2.99173415e-01
2.09674314e-01 -4.41964865e-01 6.77777752e-02 1.35715529e-01] | [11.629671096801758, 3.130505084991455] |
2ff4af69-2df1-42f0-8987-22b538a4dd97 | semantic-rule-web-based-diagnosis-and | 2301.03013 | null | https://arxiv.org/abs/2301.03013v2 | https://arxiv.org/pdf/2301.03013v2.pdf | Semantic rule Web-based Diagnosis and Treatment of Vector-Borne Diseases using SWRL rules | Vector-borne diseases (VBDs) are a kind of infection caused through the transmission of vectors generated by the bites of infected parasites, bacteria, and viruses, such as ticks, mosquitoes, triatomine bugs, blackflies, and sandflies. If these diseases are not properly treated within a reasonable time frame, the mortality rate may rise. In this work, we propose a set of ontologies that will help in the diagnosis and treatment of vector-borne diseases. For developing VBD's ontology, electronic health records taken from the Indian Health Records website, text data generated from Indian government medical mobile applications, and doctors' prescribed handwritten notes of patients are used as input. This data is then converted into correct text using Optical Character Recognition (OCR) and a spelling checker after pre-processing. Natural Language Processing (NLP) is applied for entity extraction from text data for making Resource Description Framework (RDF) medical data with the help of the Patient Clinical Data (PCD) ontology. Afterwards, Basic Formal Ontology (BFO), National Vector Borne Disease Control Program (NVBDCP) guidelines, and RDF medical data are used to develop ontologies for VBDs, and Semantic Web Rule Language (SWRL) rules are applied for diagnosis and treatment. The developed ontology helps in the construction of decision support systems (DSS) for the NVBDCP to control these diseases. | ['Navjot Singh', 'Sonali Agarwal', 'Sadhana Tiwari', 'Ritesh Chandra'] | 2023-01-08 | null | null | null | null | ['optical-character-recognition'] | ['computer-vision'] | [ 3.43475342e-01 -7.46519342e-02 -4.18870628e-01 -1.80000946e-01
1.67817548e-01 -3.44002128e-01 4.81271863e-01 8.63963187e-01
-3.93370658e-01 7.43632197e-01 5.18393040e-01 -4.55776215e-01
-5.08087456e-01 -1.16061842e+00 -9.73515119e-03 -3.97884905e-01
-1.65045917e-01 7.38961279e-01 1.84424311e-01 -2.46416941e-01
9.77774709e-02 6.69011116e-01 -1.45702803e+00 5.13136744e-01
1.05169201e+00 6.46041214e-01 4.03781474e-01 4.29634124e-01
-6.59573257e-01 8.57846975e-01 -7.42084324e-01 -2.29056045e-01
-7.75690526e-02 4.55170088e-02 -5.88143826e-01 -8.99164826e-02
-6.03826880e-01 -5.02409995e-01 9.77604315e-02 9.65116799e-01
2.41111949e-01 -3.29009354e-01 7.70002782e-01 -1.19999576e+00
-6.93562746e-01 -4.01070248e-03 7.23935515e-02 -7.05446526e-02
7.27089286e-01 -2.72586904e-02 2.82205522e-01 -3.72169644e-01
1.23216844e+00 1.41426063e+00 3.45127970e-01 8.22238207e-01
-3.78374934e-01 -1.62840873e-01 -6.21338427e-01 6.97505772e-02
-1.08061838e+00 -1.23985611e-01 -2.34410912e-02 -7.75678933e-01
1.25819337e+00 2.66715854e-01 7.81004369e-01 5.31224072e-01
7.90230930e-01 3.94147217e-01 4.66233969e-01 -5.74584544e-01
3.23261529e-01 1.34384260e-01 2.23576754e-01 6.59709156e-01
9.49173868e-01 -1.22202978e-01 -2.60253936e-01 -5.65568984e-01
4.33230937e-01 5.87444186e-01 -2.44040385e-01 -1.35031827e-02
-7.98943102e-01 8.64277422e-01 2.27892384e-01 4.50892329e-01
-7.52675653e-01 -7.05287933e-01 4.80079800e-01 -1.03007868e-01
1.69954166e-01 3.73313352e-02 -4.35561568e-01 3.03713799e-01
-1.06220275e-01 2.12214351e-01 8.53257596e-01 6.02187753e-01
3.00064683e-01 -3.08671653e-01 1.13434672e-01 6.58875227e-01
8.54804814e-01 9.07867134e-01 5.46014965e-01 -4.11595792e-01
1.75906315e-01 1.17674494e+00 2.10304871e-01 -1.13691461e+00
-2.65395403e-01 3.09712350e-01 -6.17335439e-01 -3.26051444e-01
1.27460407e-02 -7.05513582e-02 -1.12709117e+00 9.98977184e-01
7.31756449e-01 -2.71175802e-01 7.08056629e-01 6.16529405e-01
8.28209639e-01 7.48013198e-01 5.41324258e-01 -4.28904474e-01
1.85374904e+00 -1.72577828e-01 -1.31671441e+00 1.68129399e-01
9.05288100e-01 -5.08685648e-01 3.65364760e-01 2.82004237e-01
-4.55769897e-01 9.64846388e-02 -7.91470945e-01 1.76399007e-01
-1.02165389e+00 -3.02408338e-01 6.61109209e-01 5.26585579e-01
-5.32238126e-01 2.03888863e-01 -8.91885996e-01 -1.09920382e+00
3.70712131e-01 3.67961109e-01 -4.92247134e-01 -4.35198247e-01
-1.32338321e+00 1.11140859e+00 3.92320007e-01 -3.27219032e-02
-4.94578749e-01 -5.11807561e-01 -1.06330919e+00 -4.04342830e-01
-2.83640195e-02 -9.48557734e-01 5.06403625e-01 -2.87654012e-01
-8.70365798e-01 1.03486979e+00 -1.20916575e-01 -4.98058349e-01
-9.35032591e-02 -1.51149139e-01 -9.62526679e-01 4.08284217e-01
3.53547007e-01 3.13963681e-01 9.29404721e-02 -7.17125773e-01
-9.59328175e-01 -9.51906681e-01 -1.92740247e-01 -9.77208540e-02
-2.43370727e-01 4.15700585e-01 -3.87855142e-01 -5.32227039e-01
-3.43370527e-01 -5.05096436e-01 -8.91077444e-02 -9.22244713e-02
-1.31260246e-01 -4.24049437e-01 8.26648891e-01 -1.05050540e+00
1.43808365e+00 -2.00041437e+00 4.10577990e-02 3.26095462e-01
-1.15942813e-01 7.14554727e-01 1.44617232e-02 5.81545651e-01
5.24226487e-01 4.00667191e-01 -2.40621850e-01 6.86316192e-01
-2.82603949e-01 7.67557323e-01 3.19332555e-02 4.50881524e-03
2.07468122e-01 2.31873900e-01 -8.78615379e-01 -6.79440856e-01
4.18117404e-01 1.13091731e+00 -2.80963957e-01 9.13998112e-02
-3.35898250e-01 1.54491246e-01 -1.03568304e+00 8.97489727e-01
4.51304615e-01 -4.36193161e-02 4.21281576e-01 -1.73763588e-01
-1.10921441e-02 2.92699277e-01 -8.30895782e-01 9.28822100e-01
-1.65481135e-01 -1.24679036e-01 1.02834627e-01 -5.08764923e-01
7.84600079e-01 7.16543078e-01 7.33492136e-01 -6.86181366e-01
8.06022584e-02 3.08068514e-01 -3.34499627e-01 -1.32736218e+00
-8.21574107e-02 2.73913387e-02 2.94639140e-01 2.42774606e-01
-2.87009567e-01 1.98158219e-01 6.24334574e-01 6.48299977e-02
9.82906938e-01 2.93932054e-02 6.55624986e-01 -1.43771365e-01
7.51267195e-01 6.25978947e-01 8.80863786e-01 -1.58317998e-01
6.69301227e-02 -3.51285368e-01 1.75499126e-01 -8.16173851e-01
-7.55268991e-01 -8.40328217e-01 -6.22905850e-01 3.80951166e-01
-9.43409652e-02 -4.54930276e-01 -7.32330084e-01 -1.27493277e-01
1.58369914e-01 5.76740146e-01 -3.41570318e-01 1.97785914e-01
-2.52648622e-01 -1.03411698e+00 5.19798279e-01 -2.10410706e-03
4.77014393e-01 -1.18685615e+00 -9.38434958e-01 5.36064029e-01
-1.03212602e-01 -9.40619171e-01 2.77482331e-01 -3.49650621e-01
-9.31777775e-01 -1.68887568e+00 -3.57326210e-01 -6.99619472e-01
7.54386365e-01 -1.13754973e-01 4.01705563e-01 2.81235665e-01
-7.39876628e-01 7.77883381e-02 -3.81930828e-01 -7.80216634e-01
-6.92547619e-01 -6.49263442e-01 3.22654903e-01 -3.63084018e-01
1.00824940e+00 8.77224579e-02 -3.60749692e-01 -2.84872930e-02
-1.49620676e+00 -7.71429837e-02 1.19335055e-01 2.99073517e-01
5.92998028e-01 5.03852546e-01 3.55951041e-01 -8.44624639e-01
7.77673602e-01 -5.09381890e-01 -5.11442065e-01 6.89307451e-01
-4.74931806e-01 -5.06949462e-02 2.20432624e-01 -2.49241590e-02
-8.91628981e-01 -2.47863665e-01 -3.58481258e-02 1.08969696e-01
-3.35413754e-01 6.13389075e-01 -2.04363406e-01 4.51274306e-01
5.74156106e-01 7.18532270e-03 2.47289926e-01 -5.64117730e-01
-4.24859673e-01 1.46187508e+00 3.41704994e-01 -5.96969537e-02
9.19976085e-02 5.61234772e-01 -9.78330746e-02 -1.05058479e+00
-3.49293470e-01 -3.17066967e-01 -2.50208564e-02 -9.01174247e-02
1.51887059e+00 -6.55078888e-01 -6.76931262e-01 3.75472277e-01
-1.17671275e+00 1.99292764e-01 2.74410129e-01 7.07130373e-01
9.95437801e-02 5.49667142e-02 -7.39905059e-01 -8.71754527e-01
-6.44163311e-01 -9.20319378e-01 6.99563920e-01 1.66108802e-01
-2.13517860e-01 -1.03663361e+00 2.59075969e-01 6.53100252e-01
3.08138967e-01 8.17423403e-01 1.51023626e+00 -6.34359837e-01
-1.11673482e-01 -3.05968761e-01 5.79829738e-02 2.50932992e-01
8.65844965e-01 4.50043917e-01 -1.88854337e-01 2.59961337e-01
8.73776153e-02 2.34089583e-01 1.28101066e-01 3.20320606e-01
2.99269110e-01 -6.93141520e-01 -7.45661139e-01 3.30717057e-01
1.66391265e+00 9.61456418e-01 7.28773057e-01 3.74986351e-01
3.75716060e-01 1.02743971e+00 8.04938614e-01 5.15116990e-01
4.80754286e-01 3.71502668e-01 2.11244375e-01 4.57930446e-01
2.42537722e-01 3.99406403e-02 1.96475238e-01 6.37669086e-01
-4.82308805e-01 -5.21643341e-01 -1.32750309e+00 5.80500722e-01
-1.49719071e+00 -7.13189125e-01 -2.76626647e-01 2.11144948e+00
7.82252014e-01 -3.94583434e-01 -8.69242400e-02 4.07286659e-02
7.64172673e-01 -5.72078168e-01 -2.23801658e-01 -5.10851562e-01
2.21885219e-02 3.46729517e-01 5.16179681e-01 5.33028781e-01
-7.09286451e-01 6.49663270e-01 5.28464508e+00 1.98477387e-01
-1.11370897e+00 -9.87433568e-02 -8.66461545e-02 3.32602084e-01
-3.78827035e-01 -3.80782872e-01 -7.46839523e-01 4.68362659e-01
8.25121462e-01 -2.10717753e-01 4.81750369e-01 4.73735154e-01
4.80068624e-01 -1.85219213e-01 -4.90538150e-01 7.49854445e-01
-5.05780205e-02 -1.48644280e+00 3.77955943e-01 2.74046838e-01
2.29224324e-01 -6.01678453e-02 -7.09544241e-01 -3.75210345e-01
3.72233421e-01 -7.43420601e-01 -5.29952645e-02 6.84212208e-01
5.57334006e-01 -5.76678336e-01 1.06165302e+00 6.86792582e-02
-1.00118256e+00 3.60029400e-03 -4.92194831e-01 4.46849644e-01
3.64421815e-01 6.64366603e-01 -8.71989548e-01 5.94005108e-01
7.63636053e-01 4.94315207e-01 -1.30485579e-01 7.46117234e-01
-9.20005292e-02 5.53341433e-02 -4.23274159e-01 5.08107431e-02
-7.75108412e-02 -2.45046124e-01 4.17067409e-01 8.50770593e-01
4.21286047e-01 5.40587306e-01 -3.46910357e-01 2.71556467e-01
3.82945091e-01 3.54053408e-01 -7.53484607e-01 -6.70354128e-01
4.14164186e-01 7.45829344e-01 -4.82037544e-01 -5.76625109e-01
-1.07840553e-01 4.96462613e-01 -4.07455593e-01 2.44196996e-01
-3.49055111e-01 -6.10748827e-01 8.51139128e-01 3.73999119e-01
-2.41093407e-03 3.21550637e-01 3.44874859e-01 -1.03530121e+00
-2.86615282e-01 -8.62224638e-01 6.66583478e-01 -6.90319657e-01
-1.18032026e+00 6.23460770e-01 8.30683783e-02 -9.04324293e-01
-3.31869811e-01 -7.18096912e-01 -1.86223000e-01 5.04493296e-01
-1.19667876e+00 -8.59816492e-01 -7.47107640e-02 7.38503337e-01
9.11252722e-02 -3.00361037e-01 1.26735055e+00 4.58618492e-01
-4.37274128e-01 -3.12928438e-01 5.13429791e-02 1.62910089e-01
4.09734070e-01 -4.83601928e-01 -4.37934756e-01 2.41266608e-01
-7.55424380e-01 8.20014179e-01 5.53782761e-01 -1.32874703e+00
-1.46635687e+00 -1.18546665e+00 1.76972437e+00 1.94320213e-02
4.19738561e-01 1.29168689e-01 -7.18253076e-01 3.64024937e-01
1.21308036e-01 -2.84427732e-01 1.23183167e+00 -8.12451780e-01
-4.05530035e-02 1.42698120e-02 -1.94887674e+00 3.34263831e-01
5.09181857e-01 -2.30234429e-01 -1.02168298e+00 7.65558004e-01
5.86111069e-01 3.60122062e-02 -1.22856593e+00 5.24251044e-01
6.08678997e-01 -3.50677043e-01 1.00590062e+00 -9.33394015e-01
1.61663473e-01 -6.39987886e-01 -3.74036282e-01 -7.40963101e-01
3.58983219e-01 -1.26832768e-01 2.69235998e-01 9.77302492e-01
-9.80280563e-02 -7.76664913e-01 9.45994258e-03 6.49049103e-01
1.39235422e-01 -4.27905947e-01 -5.45597315e-01 -4.56494331e-01
-7.65513957e-01 -1.37367681e-01 1.00042748e+00 1.16722524e+00
1.24423802e-01 -9.41957999e-03 -1.03506530e-02 3.26093912e-01
7.37464488e-01 -5.32263517e-01 2.29546696e-01 -1.30171418e+00
2.41518915e-01 4.04859245e-01 -6.18433595e-01 1.82273835e-01
-5.16353011e-01 -6.41217470e-01 -4.42441255e-01 -2.34686685e+00
-2.06562251e-01 -3.95935208e-01 -8.35034177e-02 7.88886130e-01
3.62045854e-01 -1.71496972e-01 7.94258416e-02 3.08400840e-01
9.27694812e-02 -8.35107341e-02 8.93794537e-01 -1.98163427e-02
-4.40770298e-01 -3.24001878e-01 -2.29986474e-01 7.14593887e-01
7.61708677e-01 -8.00348401e-01 -2.50974715e-01 -5.83869338e-01
3.88457596e-01 3.64280343e-01 1.25230372e-01 -5.94269276e-01
1.04636243e-02 -5.98878801e-01 -1.01478323e-02 -5.84439039e-01
-6.36890009e-02 -1.20649493e+00 8.19420218e-01 1.34426522e+00
1.25860915e-01 -7.20827505e-02 1.60156831e-01 2.38354445e-01
-1.95832729e-01 -2.80461758e-01 3.08347493e-01 8.80443305e-02
-4.82804149e-01 1.21562354e-01 -8.60035479e-01 -3.83223891e-01
1.26042235e+00 -3.05331886e-01 -6.92291021e-01 2.71080017e-01
-6.83811009e-01 4.37204361e-01 3.71384323e-01 4.43028629e-01
6.10549629e-01 -1.21058118e+00 -6.29472613e-01 1.95984185e-01
1.63901553e-01 3.35994102e-02 3.01090151e-01 6.10554814e-01
-1.57617140e+00 8.15702736e-01 -5.26220858e-01 -2.42548406e-01
-1.21603787e+00 6.92985535e-01 5.39789163e-02 -2.74227023e-01
-3.13897997e-01 1.19074114e-01 -5.56155778e-02 -4.16937053e-01
1.34302750e-01 -3.57357651e-01 -6.46239161e-01 6.35256246e-02
1.00230491e+00 3.57593566e-01 -9.35880020e-02 -8.29728782e-01
-9.97555554e-01 6.65022016e-01 1.14343986e-01 1.56477600e-01
1.75271273e+00 2.16347978e-01 -4.93643016e-01 -1.35748670e-01
8.11427891e-01 -1.05160892e-01 -1.82801023e-01 1.48354217e-01
8.09946060e-02 -2.78760135e-01 -1.98987037e-01 -8.22652221e-01
-8.51310551e-01 4.48907256e-01 6.08584046e-01 6.27273023e-02
1.19740295e+00 -7.24576563e-02 7.17847884e-01 4.11573142e-01
4.19369400e-01 -9.85017478e-01 -7.35747695e-01 -6.40237927e-02
6.09112442e-01 -6.16655767e-01 6.63768798e-02 -6.18066907e-01
-6.26690865e-01 1.22160113e+00 -1.90574042e-02 2.39183888e-01
7.85321116e-01 5.41601419e-01 1.89854711e-01 -4.09897447e-01
-8.31953526e-01 2.10056584e-02 -2.01956138e-01 1.07041490e+00
2.62050450e-01 1.51996955e-01 -7.52351701e-01 4.47178751e-01
2.46941850e-01 7.22214103e-01 1.46779954e-01 1.35295188e+00
-7.56811440e-01 -1.34148085e+00 -6.92055404e-01 4.90731835e-01
-8.09250474e-01 -2.83367615e-02 -3.64055157e-01 7.59182334e-01
7.04016268e-01 1.03552783e+00 4.14279640e-01 1.61133781e-02
2.15204358e-01 -2.13769898e-02 2.51935750e-01 -5.00873089e-01
-4.95853066e-01 -5.87383518e-03 1.81913689e-01 -2.01829195e-01
-9.16877031e-01 -2.74907947e-01 -1.77653670e+00 -2.63558745e-01
1.38551788e-02 2.06875190e-01 1.05555880e+00 1.00932539e+00
3.27133030e-01 2.00436980e-01 3.20431478e-02 2.84539551e-01
2.31911466e-01 -6.63209379e-01 -4.60581005e-01 4.35202837e-01
8.52144882e-02 -3.59889597e-01 -1.16772167e-01 3.94174308e-01] | [8.51614761352539, 8.465448379516602] |
b2d554eb-1cc0-48a8-b113-23fab57079ee | recurrent-neural-network-transducer-for-audio | 1911.04890 | null | https://arxiv.org/abs/1911.04890v1 | https://arxiv.org/pdf/1911.04890v1.pdf | Recurrent Neural Network Transducer for Audio-Visual Speech Recognition | This work presents a large-scale audio-visual speech recognition system based on a recurrent neural network transducer (RNN-T) architecture. To support the development of such a system, we built a large audio-visual (A/V) dataset of segmented utterances extracted from YouTube public videos, leading to 31k hours of audio-visual training content. The performance of an audio-only, visual-only, and audio-visual system are compared on two large-vocabulary test sets: a set of utterance segments from public YouTube videos called YTDEV18 and the publicly available LRS3-TED set. To highlight the contribution of the visual modality, we also evaluated the performance of our system on the YTDEV18 set artificially corrupted with background noise and overlapping speech. To the best of our knowledge, our system significantly improves the state-of-the-art on the LRS3-TED set. | ['Hank Liao', 'Otavio Braga', 'Basilio Garcia', 'Yannis Assael', 'Takaki Makino', 'Olivier Siohan', 'Brendan Shillingford'] | 2019-11-08 | null | null | null | null | ['lipreading', 'audio-visual-speech-recognition'] | ['computer-vision', 'speech'] | [-1.40320053e-02 -2.59474684e-02 8.03736672e-02 -3.03732485e-01
-1.26990211e+00 -3.81693363e-01 4.33528364e-01 -3.77449691e-01
-2.77010560e-01 1.24798603e-01 3.74111354e-01 -3.98981154e-01
7.05191672e-01 1.13969659e-02 -8.55216980e-01 -5.37680864e-01
-1.50752477e-02 1.69363528e-01 3.80613357e-01 5.63603677e-02
-9.81697813e-02 2.28890613e-01 -1.91246426e+00 7.68351257e-01
1.24388210e-01 1.48883581e+00 2.35335931e-01 1.12190998e+00
1.72170103e-01 9.92058814e-01 -6.74481630e-01 1.15662567e-01
1.57434002e-01 -2.18269303e-01 -5.45622647e-01 2.09126934e-01
6.98942602e-01 -5.46712875e-01 -8.26957524e-01 5.98159373e-01
7.29752183e-01 2.25040227e-01 5.53435564e-01 -1.35273349e+00
-7.10539758e-01 3.67778420e-01 -6.17876723e-02 3.22496712e-01
5.41733265e-01 3.30397069e-01 9.22674537e-01 -1.41133761e+00
8.63470852e-01 1.26103723e+00 3.08598727e-01 4.28391248e-01
-8.66119862e-01 -7.17776775e-01 5.15767187e-02 4.32394713e-01
-1.58817518e+00 -1.11207902e+00 7.53225684e-01 -4.75283384e-01
1.60897219e+00 2.06098780e-01 7.55589247e-01 1.81377912e+00
-2.35353857e-01 1.29919982e+00 7.88705289e-01 -3.96331012e-01
2.01706365e-01 2.11979657e-01 4.46517505e-02 5.83087206e-01
-6.91770971e-01 1.60321027e-01 -9.79202211e-01 -1.35859728e-01
5.69653273e-01 -2.98581451e-01 -4.73422587e-01 -1.48552939e-01
-9.31900024e-01 7.42718101e-01 2.06718128e-02 2.10712224e-01
-1.46249726e-01 -2.53296411e-03 9.15403724e-01 8.01172554e-01
5.76938152e-01 -3.42038959e-01 -4.44540650e-01 -3.56663615e-01
-1.09746361e+00 -2.02366412e-01 7.69333601e-01 1.07239389e+00
2.38899276e-01 8.83681536e-01 -2.33164340e-01 1.04574013e+00
5.86510539e-01 6.96890414e-01 7.65698016e-01 -8.33740950e-01
6.83146894e-01 5.74016571e-02 -3.21551919e-01 -4.08891320e-01
-4.67401594e-02 -1.61796827e-02 -4.59955782e-01 1.15653642e-01
8.03795457e-02 9.10638198e-02 -1.44049692e+00 1.27085912e+00
1.03384122e-01 2.12941527e-01 2.64511138e-01 9.94823277e-01
1.89580214e+00 1.17699730e+00 -2.46229157e-01 -3.65107983e-01
1.05931604e+00 -1.23002839e+00 -7.02858567e-01 2.16358125e-01
4.56906885e-01 -6.71585977e-01 1.19945800e+00 4.41379100e-01
-9.61817026e-01 -6.89077556e-01 -9.25844491e-01 -1.13525003e-01
-2.12435037e-01 4.39776927e-01 -1.83335498e-01 4.69511122e-01
-1.48438120e+00 -1.86771706e-01 -6.20813012e-01 -3.14226747e-01
1.39325559e-01 -4.68685739e-02 -5.93240559e-01 6.33546486e-02
-1.13030660e+00 3.74072343e-01 -1.14051826e-01 -9.32443440e-02
-1.79874611e+00 -2.94115156e-01 -1.13505375e+00 6.06166981e-02
7.45908856e-01 5.44405431e-02 1.67222416e+00 -9.94371712e-01
-1.70257342e+00 6.77289546e-01 -4.12062526e-01 -5.67407966e-01
2.41394475e-01 3.43737146e-03 -7.05242217e-01 5.89293540e-01
-1.71147719e-01 7.45642126e-01 1.27039671e+00 -1.00125825e+00
-1.90202564e-01 -1.15556717e-01 -3.95749986e-01 6.84225559e-03
-2.97650605e-01 3.21575403e-01 -8.70250404e-01 -8.14569533e-01
-2.40145937e-01 -8.90105784e-01 4.45932239e-01 -2.18975380e-01
-5.15847266e-01 -4.56671447e-01 1.20111263e+00 -8.97363901e-01
9.55638289e-01 -2.72512674e+00 7.73091451e-05 -2.39937031e-03
1.15517452e-02 4.88223463e-01 -6.24054015e-01 4.50146824e-01
-1.31445706e-01 1.19733505e-01 3.45093280e-01 -6.88372850e-01
-2.09527001e-01 1.35728225e-01 -6.04900002e-01 3.08993131e-01
-3.37299071e-02 7.57674813e-01 -4.38171089e-01 -5.13827026e-01
2.12168023e-01 7.79853582e-01 -3.42564166e-01 4.38290268e-01
-8.37943703e-02 1.41468257e-01 7.41426945e-02 1.19068813e+00
-3.76290977e-02 -6.44625872e-02 -1.83734983e-01 -3.74526940e-02
2.87916075e-04 3.60392988e-01 -6.35860920e-01 1.41237009e+00
-2.50211746e-01 1.41460001e+00 2.45378822e-01 -7.81404257e-01
8.93051803e-01 1.13370085e+00 2.48689964e-01 -9.82705057e-01
1.37365773e-01 1.80430830e-01 -3.43008220e-01 -6.99986100e-01
5.97810924e-01 3.15879434e-01 1.97703436e-01 2.03443348e-01
6.50760949e-01 5.05205281e-02 -2.28236150e-02 4.08451855e-01
1.02999210e+00 -7.42867887e-02 -8.72656778e-02 4.65106279e-01
2.66845018e-01 -2.17398718e-01 2.77293205e-01 6.79205656e-01
-4.39713836e-01 1.01121938e+00 5.65397441e-01 -1.91647410e-01
-1.17583048e+00 -1.08123922e+00 -6.67586131e-03 1.16969264e+00
-5.23626566e-01 -7.38210440e-01 -5.69256186e-01 -6.07274592e-01
-1.24037161e-01 4.07936871e-01 -6.10435426e-01 3.08592081e-01
-1.30683675e-01 1.79551601e-01 9.27682519e-01 6.08945489e-01
1.05867118e-01 -1.45403349e+00 -1.69850692e-01 -3.98135968e-02
-2.91246384e-01 -1.61071873e+00 -5.44197500e-01 -8.56861025e-02
-3.14339429e-01 -9.80567217e-01 -1.04366553e+00 -9.79748785e-01
1.40932843e-01 4.50214058e-01 8.69485676e-01 -4.08142745e-01
-1.93859056e-01 7.67959952e-01 -6.58752918e-01 -3.72047126e-01
-6.71114385e-01 -3.49684268e-01 2.32408315e-01 8.77131000e-02
2.96296757e-02 -2.56181210e-01 -1.67294011e-01 5.18866897e-01
-6.55263364e-01 -6.43647239e-02 2.38381639e-01 8.18625033e-01
7.72600830e-01 -5.41494191e-01 4.53281701e-01 -1.07288361e-01
5.29364944e-01 -2.98273653e-01 -6.04103088e-01 2.59174347e-01
-2.05916613e-01 -6.09321117e-01 3.85434210e-01 -6.90973103e-01
-6.77058578e-01 8.99034142e-02 -2.61099935e-01 -1.35075426e+00
-1.61503837e-01 6.13633454e-01 1.01377301e-01 9.37482864e-02
3.94170016e-01 4.81215864e-01 6.05766848e-02 -6.60459757e-01
3.20199519e-01 1.37354600e+00 8.17550421e-01 1.35378048e-01
1.27258867e-01 1.62906259e-01 -6.29527450e-01 -1.64764071e+00
3.98723023e-05 -7.09179938e-01 -1.37847066e-01 -5.23848832e-01
7.32520044e-01 -1.48074567e+00 -6.17644608e-01 7.84756839e-01
-1.11528659e+00 -6.42967403e-01 -1.88557491e-01 4.66122955e-01
-5.20980000e-01 4.15934861e-01 -8.62310529e-01 -1.07566452e+00
-4.80399400e-01 -1.56810439e+00 1.22167718e+00 -3.30204457e-01
-3.52951623e-02 -4.28126007e-01 1.37757048e-01 5.06683469e-01
1.25194013e-01 -2.04850778e-01 1.67475715e-01 -8.46765995e-01
-2.99409121e-01 -1.78216755e-01 -1.89926863e-01 6.28578126e-01
-2.52105266e-01 4.13071990e-01 -1.41782176e+00 -4.57884014e-01
-1.83892816e-01 -8.82903218e-01 1.20285022e+00 4.00453717e-01
1.00045860e+00 -2.13762835e-01 6.83034807e-02 3.04848254e-01
7.65022755e-01 5.05570710e-01 6.12526417e-01 -8.08368996e-02
7.83169329e-01 3.79055232e-01 5.25910914e-01 3.21147949e-01
4.12161469e-01 9.67544913e-01 4.21707779e-01 -6.77535031e-03
-4.06729043e-01 -2.44047597e-01 9.23293352e-01 1.41184592e+00
9.04717892e-02 -6.14473224e-01 -9.27767456e-01 8.74311447e-01
-1.58551490e+00 -8.73454690e-01 2.59326786e-01 1.93717623e+00
5.25916755e-01 -3.41679677e-02 4.82217729e-01 2.98049390e-01
4.91770595e-01 3.83111924e-01 -3.08485210e-01 -4.85117793e-01
-1.87949166e-01 -7.83793535e-03 -1.62947357e-01 3.60092580e-01
-1.14418936e+00 8.34018648e-01 6.90659380e+00 1.02950919e+00
-1.55398679e+00 3.14443320e-01 5.05943179e-01 -6.67204976e-01
-6.05791360e-02 -5.80965161e-01 -6.22134030e-01 3.50916535e-01
1.64475417e+00 2.62744308e-01 3.02848041e-01 9.90173697e-01
3.60972166e-01 1.61223799e-01 -8.15019011e-01 1.21023786e+00
5.00709295e-01 -1.28363132e+00 3.01886611e-02 2.87742019e-02
2.90512592e-01 8.45053613e-01 1.82135135e-01 5.00954747e-01
-1.42201930e-01 -1.02709055e+00 1.02145433e+00 2.01208759e-02
1.48020887e+00 -4.41645801e-01 5.22902012e-01 -4.01740931e-02
-1.35719860e+00 1.18784897e-01 5.64586464e-03 3.90216917e-01
2.98150182e-01 5.89767359e-02 -1.15445793e+00 1.73470452e-01
1.16033399e+00 8.85416687e-01 -4.15872723e-01 8.11183631e-01
-2.22712159e-01 1.04353166e+00 -3.14562529e-01 -1.01083845e-01
3.60396177e-01 4.58459258e-01 8.07541907e-01 1.36757803e+00
2.12798133e-01 -1.02584951e-01 4.13390528e-03 1.55519247e-01
-2.42209300e-01 1.71270907e-01 -9.30358708e-01 -5.11087537e-01
4.44042236e-01 9.24984515e-01 -3.72922540e-01 -6.15534067e-01
-7.66559124e-01 5.78046083e-01 1.35220245e-01 6.92498088e-01
-6.95039868e-01 -2.44042307e-01 4.63881433e-01 -1.27083138e-01
7.88530350e-01 -2.06613123e-01 4.85999554e-01 -1.23180294e+00
1.83954552e-01 -1.11047077e+00 5.59564307e-02 -1.27289200e+00
-7.67588139e-01 1.04485774e+00 -6.17393777e-02 -1.42679715e+00
-6.66103125e-01 -5.46609640e-01 -3.18406016e-01 4.59168553e-01
-1.08546853e+00 -1.05249882e+00 -1.29409790e-01 7.61947155e-01
1.11097026e+00 -8.77684891e-01 6.99988604e-01 2.97157735e-01
-4.96961534e-01 7.51833022e-01 1.34027392e-01 3.85721058e-01
6.76466346e-01 -5.68523526e-01 7.21395016e-01 6.50217295e-01
5.84353030e-01 1.21150367e-01 5.13704836e-01 -4.47482377e-01
-1.42562532e+00 -1.11407912e+00 6.55238271e-01 -1.95724770e-01
6.35891497e-01 -6.73863292e-01 -9.62459207e-01 7.82811463e-01
4.10980642e-01 1.77108020e-01 6.59994662e-01 -3.59792411e-01
-6.76015913e-01 2.00492099e-01 -7.43752778e-01 3.61127466e-01
7.05503643e-01 -1.31255543e+00 -6.12158775e-01 2.17860844e-02
1.15106893e+00 -5.42766392e-01 -7.10637569e-01 3.43860328e-01
8.16265941e-01 -7.55104899e-01 9.10003185e-01 -4.86114591e-01
3.70330989e-01 -1.33216560e-01 -5.35227299e-01 -1.25555253e+00
3.59845966e-01 -6.20025277e-01 -2.34438822e-01 1.11052167e+00
4.24424767e-01 -4.22620177e-01 4.44351375e-01 -6.24969304e-02
-4.23554003e-01 -7.44399369e-01 -1.56241763e+00 -7.97191560e-01
-5.18566072e-01 -9.89183009e-01 9.54836234e-02 6.30730152e-01
3.62582430e-02 3.36547762e-01 -7.42666245e-01 -1.06791787e-01
1.02036349e-01 -2.31391490e-01 8.79651964e-01 -8.28138292e-01
-2.92635947e-01 -3.32502499e-02 -6.81959152e-01 -1.09588110e+00
1.82864875e-01 -6.34302199e-01 7.45306835e-02 -1.25858343e+00
-3.59534509e-02 3.66436452e-01 -1.85532883e-01 7.66286612e-01
5.46068311e-01 6.34428382e-01 3.62028092e-01 2.79284537e-01
-8.36329818e-01 7.63644636e-01 8.57370198e-01 -4.80072558e-01
-3.37598532e-01 -2.93731093e-01 -4.32754643e-02 3.50386381e-01
4.30207610e-01 -2.70438731e-01 -4.36097145e-01 -3.13747227e-01
-2.60924101e-01 5.85657179e-01 3.88006568e-01 -6.77571356e-01
2.27822661e-01 3.64476442e-01 9.33564305e-02 -9.86384094e-01
9.34790969e-01 -3.96730572e-01 -7.60234743e-02 1.27212793e-01
-2.89510578e-01 -9.31295976e-02 3.81544918e-01 5.07565916e-01
-6.44933581e-01 2.98344523e-01 3.22964191e-01 1.30955234e-01
-8.82820785e-01 8.53037313e-02 -8.93978953e-01 -2.28694286e-02
7.10796714e-01 -2.66398013e-01 -5.75796127e-01 -9.66841638e-01
-9.52294767e-01 1.70351580e-01 1.72657460e-01 8.73922050e-01
1.19199753e+00 -1.37059247e+00 -7.23094165e-01 5.45927167e-01
4.68063861e-01 -4.09991592e-01 3.11845213e-01 8.46571982e-01
-1.77546144e-01 8.03849578e-01 2.70152651e-02 -9.08125103e-01
-1.96849942e+00 5.35320401e-01 1.18922803e-03 4.71759766e-01
-7.14800954e-01 9.07870412e-01 3.00121665e-01 -1.28532022e-01
8.73985052e-01 -5.55271924e-01 -4.51084673e-01 2.08063409e-01
6.63658977e-01 3.24917614e-01 1.41533181e-01 -1.24535334e+00
-3.89174849e-01 1.64047658e-01 -8.41423869e-03 -5.44161081e-01
1.12057900e+00 -6.61892742e-02 4.12895977e-01 9.00210798e-01
1.49706149e+00 -6.45622313e-02 -1.01801121e+00 -3.04134697e-01
-7.13908553e-01 -2.16610238e-01 1.69769123e-01 -4.86629069e-01
-1.25046635e+00 9.86724138e-01 8.05733800e-01 4.62173462e-01
8.91514122e-01 2.28047073e-01 7.85281956e-01 4.45657909e-01
1.08362585e-01 -1.12546885e+00 3.31401795e-01 7.27699459e-01
1.31665754e+00 -1.32543063e+00 -4.42214876e-01 -2.04560265e-01
-1.00965798e+00 9.69620407e-01 3.18412155e-01 2.83348799e-01
6.76630676e-01 1.55025557e-01 5.78126371e-01 -8.70878771e-02
-1.11151707e+00 -1.62638322e-01 5.11847734e-01 5.26166141e-01
2.73273170e-01 -5.95378056e-02 5.04101932e-01 3.62224281e-01
-1.38387650e-01 2.48888377e-02 4.29117501e-01 8.05650592e-01
-2.71624625e-01 -4.93409514e-01 -3.85762274e-01 1.73650652e-01
-4.09637660e-01 -2.42568329e-01 -6.59481227e-01 7.10807562e-01
-6.05737031e-01 1.27235186e+00 2.20999151e-01 -7.28784204e-01
2.43415505e-01 1.60149947e-01 8.64228159e-02 -3.81178141e-01
-3.73946965e-01 8.07478726e-01 3.02941978e-01 -8.16260695e-01
-3.10499251e-01 -6.15468562e-01 -9.39372420e-01 1.07348002e-01
-7.13148490e-02 7.73098618e-02 8.63501251e-01 7.98319161e-01
4.56620812e-01 6.52842462e-01 5.73016286e-01 -1.16692150e+00
-2.19120890e-01 -1.18189800e+00 -4.84167516e-01 1.43676758e-01
8.31723273e-01 -6.04211509e-01 -6.41002357e-01 -4.16149497e-02] | [14.34278678894043, 5.134781360626221] |
9941a522-1527-4ed9-b6c1-4f7e07801f66 | information-retrieval-recent-advances-and | 2301.08801 | null | https://arxiv.org/abs/2301.08801v1 | https://arxiv.org/pdf/2301.08801v1.pdf | Information Retrieval: Recent Advances and Beyond | In this paper, we provide a detailed overview of the models used for information retrieval in the first and second stages of the typical processing chain. We discuss the current state-of-the-art models, including methods based on terms, semantic retrieval, and neural. Additionally, we delve into the key topics related to the learning process of these models. This way, this survey offers a comprehensive understanding of the field and is of interest for for researchers and practitioners entering/working in the information retrieval domain. | ['Hugo Proenca', 'Kailash A. Hambarde'] | 2023-01-20 | null | null | null | null | ['semantic-retrieval'] | ['natural-language-processing'] | [ 1.57934517e-01 -7.13171512e-02 -7.12070525e-01 -1.26950085e-01
-7.57521093e-01 -4.38213646e-01 7.56656468e-01 7.25839257e-01
-6.25493705e-01 1.88265458e-01 9.68317166e-02 -6.86741918e-02
-7.12173343e-01 -7.71263659e-01 -6.76578283e-02 -4.57428008e-01
-2.63059556e-01 6.79975986e-01 7.78536201e-02 -4.40934449e-01
1.15474916e+00 4.74430650e-01 -2.08113050e+00 6.62808836e-01
6.40190482e-01 1.31417537e+00 3.78200263e-01 4.12829697e-01
-8.00397575e-01 6.59636557e-01 -7.11523235e-01 -2.80149281e-01
-1.39247000e-01 -1.88465975e-02 -1.22452557e+00 -2.87527353e-01
3.54604334e-01 -2.15383902e-01 -5.76275349e-01 1.03375316e+00
3.70095938e-01 4.06481773e-01 1.02234674e+00 -7.48651624e-01
-9.08009708e-01 5.59740961e-01 9.86673776e-03 6.03327453e-01
7.90586174e-01 -5.93929052e-01 1.23909044e+00 -1.15698707e+00
7.37789929e-01 1.34528100e+00 3.05594832e-01 4.74661291e-01
-4.31068629e-01 -2.26598635e-01 4.02863294e-01 6.82911098e-01
-1.36201704e+00 -2.64238715e-01 8.03833246e-01 -4.81746823e-01
1.38060749e+00 2.72776663e-01 7.60445356e-01 9.11608517e-01
3.76032233e-01 1.29322708e+00 6.36733592e-01 -1.07262516e+00
-6.20784201e-02 1.60438612e-01 1.15097106e+00 6.37201965e-01
2.96754003e-01 7.59254256e-03 -7.11253643e-01 -1.25262246e-01
6.45654798e-01 1.55626044e-01 -1.86744541e-01 -1.58648297e-01
-7.03134835e-01 9.09941018e-01 5.34860194e-01 8.57571959e-01
-4.67178643e-01 -1.76703319e-01 4.12337035e-01 5.68417907e-01
5.37554264e-01 7.82094955e-01 -4.00514632e-01 1.09339520e-01
-1.14314520e+00 4.40939516e-01 9.23474193e-01 8.69609416e-01
5.44450641e-01 -3.82773608e-01 -4.38005865e-01 1.04562187e+00
8.90648007e-01 1.16590425e-01 8.23868811e-01 -6.96744740e-01
2.44388789e-01 5.03028870e-01 -3.27700153e-02 -8.68913233e-01
-3.39151561e-01 -2.03344166e-01 -4.58210588e-01 -3.40980023e-01
-2.52645433e-01 4.62576896e-01 -1.16163528e+00 8.69621158e-01
-3.08243722e-01 -5.15886247e-01 8.65544453e-02 6.67559087e-01
1.56283760e+00 9.08335686e-01 3.69220942e-01 -3.64904672e-01
1.65145397e+00 -1.54485965e+00 -1.27802050e+00 -2.92771816e-01
3.71168494e-01 -1.03819501e+00 5.67013502e-01 2.86672384e-01
-1.58422077e+00 -6.63564503e-01 -8.10032070e-01 -5.01869678e-01
-1.16201246e+00 -1.93371810e-02 9.65224028e-01 3.54487926e-01
-1.22279036e+00 7.08079576e-01 -5.92111766e-01 -8.78412247e-01
2.21266244e-02 1.44441664e-01 1.54447377e-01 -3.83728683e-01
-1.73453021e+00 1.43382096e+00 6.80118442e-01 -5.40374890e-02
-8.04889202e-01 -3.90043586e-01 -8.46824765e-01 2.02936634e-01
1.51381135e-01 -8.37331653e-01 1.52682877e+00 -5.19905448e-01
-1.19388795e+00 1.06045914e+00 -4.06751812e-01 -1.38823181e-01
-1.94786400e-01 -5.30753732e-01 -3.93273950e-01 6.76746845e-01
-2.59475708e-01 5.66461205e-01 6.36902094e-01 -1.31465161e+00
-6.23906851e-01 -4.05294985e-01 2.68386871e-01 7.71625102e-01
-4.77059007e-01 7.38592923e-01 -8.83452475e-01 -6.49850070e-01
5.38049266e-02 -4.44050610e-01 7.40608945e-02 1.78238228e-02
1.59883901e-01 -8.94747078e-01 3.31517488e-01 -6.48498833e-01
1.80520082e+00 -1.85729563e+00 1.59355626e-01 1.24156266e-01
-3.38838808e-02 4.93018448e-01 -1.41015530e-01 1.40656817e+00
7.14770406e-02 3.62430543e-01 2.28173137e-01 -4.51593816e-01
4.53081951e-02 -4.07640338e-02 -4.09639269e-01 -8.87696519e-02
-3.14376771e-01 1.05937326e+00 -9.93609071e-01 -6.25166595e-01
1.21632427e-01 5.71999013e-01 5.11654243e-02 4.72051799e-01
1.07445911e-01 -1.78713933e-01 -9.24471796e-01 8.46247852e-01
1.13790937e-01 -1.79986119e-01 -1.37084976e-01 4.54724953e-02
-2.61243105e-01 7.95002162e-01 -6.37511551e-01 1.92406130e+00
-4.87282097e-01 7.13813126e-01 2.35359129e-02 -1.04123151e+00
5.75526655e-01 6.17944241e-01 3.41996163e-01 -7.58786261e-01
3.06867242e-01 2.98656970e-01 -3.42870742e-01 -5.38785934e-01
8.94021392e-01 1.80293217e-01 2.12904230e-01 6.01321340e-01
3.67495596e-01 -5.50242588e-02 6.94251359e-01 2.34954149e-01
6.75381243e-01 -1.38686240e-01 6.04997814e-01 -2.42654964e-01
6.39213204e-01 -2.14995649e-02 -4.84726101e-01 1.13100457e+00
-3.50444354e-02 3.34964544e-01 -3.06196272e-01 -6.13552213e-01
-6.02310240e-01 -8.79842281e-01 -2.96077579e-01 1.37292719e+00
2.08364159e-01 -6.28578067e-01 -6.67498112e-01 -3.19197059e-01
-1.37834489e-01 4.48032826e-01 -7.14379907e-01 -2.73673773e-01
-2.97484696e-01 -6.63512468e-01 3.17836255e-01 4.34360951e-01
4.89629626e-01 -1.60363019e+00 -3.15677673e-01 -6.59971535e-02
-3.39451462e-01 -4.44237083e-01 3.42195742e-02 4.20430750e-02
-1.23199105e+00 -9.95515943e-01 -9.41743135e-01 -1.13401318e+00
6.35975420e-01 6.83077335e-01 1.36780214e+00 7.58894086e-01
-3.67035985e-01 8.28515649e-01 -7.56653845e-01 -3.71041238e-01
7.27259070e-02 3.65641356e-01 -2.38864601e-01 -5.14770269e-01
9.60597813e-01 -1.17382722e-03 -7.51780689e-01 -1.73767850e-01
-1.26973689e+00 -5.75199783e-01 5.74371576e-01 5.76518893e-01
1.69713587e-01 1.49016321e-01 2.69377917e-01 -8.08958769e-01
1.34814513e+00 -5.04706740e-01 -2.10592166e-01 6.69504106e-01
-1.19036722e+00 -2.36128066e-02 3.66507657e-02 -6.91131726e-02
-1.14496207e+00 -6.45280540e-01 -1.70489222e-01 2.18824923e-01
-2.07203463e-01 1.06322229e+00 3.69561344e-01 -4.53350484e-01
6.95150971e-01 3.58059049e-01 -3.52885723e-01 -8.77289355e-01
4.41438407e-01 9.68236268e-01 -4.12427872e-01 -3.53406250e-01
3.47891182e-01 2.48393983e-01 -4.34994757e-01 -8.63046348e-01
-1.11475992e+00 -1.25738907e+00 -5.88953555e-01 -3.48761618e-01
5.48968375e-01 -5.68046212e-01 -2.87854552e-01 6.98656082e-01
-1.30531001e+00 2.08174959e-01 -1.16741516e-01 2.23253921e-01
-4.52127278e-01 5.62938035e-01 -1.25455320e+00 -7.67405391e-01
-9.03578281e-01 -1.08943522e+00 1.00900865e+00 4.71917897e-01
-2.36206830e-01 -1.47606528e+00 1.92071170e-01 5.09491205e-01
8.24386716e-01 -7.71759093e-01 1.03730786e+00 -1.05374885e+00
-2.64313996e-01 -5.87834537e-01 -1.66681066e-01 1.63361117e-01
-7.22751766e-02 -1.19930103e-01 -1.00060987e+00 -1.48824319e-01
1.76890135e-01 -4.32162285e-01 1.38955069e+00 3.84531677e-01
1.21427774e+00 -1.68751061e-01 -7.13488996e-01 3.93783897e-02
1.38118219e+00 4.41872090e-01 5.99484026e-01 7.34577656e-01
6.68121949e-02 1.10383427e+00 7.64426172e-01 -3.21394764e-02
3.41651052e-01 3.96919817e-01 5.87582439e-02 1.99437305e-01
-1.67779908e-01 -1.60436735e-01 1.76835656e-02 1.23697329e+00
-1.90950081e-01 -4.80739295e-01 -8.29478085e-01 6.18585587e-01
-1.78681135e+00 -7.66725361e-01 2.96231419e-01 2.01594520e+00
9.28371847e-01 -1.13519438e-01 -4.21916932e-01 2.03019038e-01
6.82905495e-01 3.32463890e-01 -8.83204266e-02 -4.78196293e-01
2.13580191e-01 3.76120329e-01 -3.74801755e-02 7.35459387e-01
-1.18109310e+00 1.06462538e+00 8.82676697e+00 1.06132936e+00
-6.81332409e-01 -3.09416980e-01 2.67190456e-01 2.55828589e-01
-7.24102259e-02 -9.93016735e-02 -1.05610490e+00 3.15451287e-02
1.16667104e+00 -2.57015735e-01 4.86072481e-01 7.56401420e-01
-1.31578967e-01 -3.61754179e-01 -1.00255966e+00 8.30299914e-01
5.77032983e-01 -1.07460892e+00 5.75102687e-01 -3.33648354e-01
4.55819964e-01 7.68810557e-03 8.16144571e-02 4.45144802e-01
-4.24348526e-02 -8.23062956e-01 3.40641081e-01 6.26980543e-01
2.65017152e-01 -3.86280775e-01 9.35608804e-01 3.84837955e-01
-9.67772126e-01 -1.02238648e-01 -4.39521909e-01 -8.15932453e-02
1.48599625e-01 3.00697207e-01 -2.00251073e-01 9.26670551e-01
7.77835608e-01 8.04903030e-01 -5.56885421e-01 1.50200760e+00
-4.72386956e-01 1.61563039e-01 -1.21861979e-01 -2.86899149e-01
3.33144873e-01 -1.85794942e-02 3.59655648e-01 1.60047376e+00
1.82956189e-01 1.19729698e-01 1.68495849e-01 2.89381266e-01
8.15863907e-02 3.37450057e-01 -5.44468224e-01 -3.66189748e-01
5.28518736e-01 1.22805560e+00 -5.90650022e-01 -6.65773869e-01
-3.48050684e-01 7.31076598e-01 3.27642709e-01 5.71752429e-01
-1.01474561e-02 -8.60531032e-01 3.09912235e-01 -1.20939746e-01
-1.35785505e-01 1.20734125e-02 -1.67488649e-01 -1.07839787e+00
-1.46658242e-01 -6.62634075e-01 7.65867114e-01 -7.46855736e-01
-1.51118588e+00 7.06223547e-01 6.82826340e-01 -8.52602780e-01
-4.09336954e-01 -8.35652351e-01 -1.56617433e-01 9.00305867e-01
-2.13041520e+00 -8.53944719e-01 -1.49905935e-01 3.08634430e-01
7.89830387e-01 -2.04791948e-01 1.19511712e+00 3.55762929e-01
-1.19742356e-01 1.04593873e-01 2.71433800e-01 3.62273082e-02
6.28678918e-01 -9.07394528e-01 3.13720524e-01 2.24723354e-01
3.80016953e-01 1.35332513e+00 3.34638268e-01 -4.01292771e-01
-1.18937206e+00 -4.15826410e-01 1.56634235e+00 -3.52151960e-01
5.86826682e-01 2.40060270e-01 -8.45014870e-01 5.44002235e-01
7.63370275e-01 -6.86872780e-01 9.36829627e-01 1.40602916e-01
-2.71577209e-01 1.69422925e-01 -1.00761676e+00 4.29393291e-01
6.12800717e-01 -6.25206828e-01 -1.18952155e+00 5.92926383e-01
3.72129023e-01 -2.28971794e-01 -8.44624043e-01 3.12935531e-01
7.59135425e-01 -4.58265364e-01 1.27794003e+00 -7.09587812e-01
2.04834685e-01 2.05953389e-01 1.74794436e-01 -1.16605639e+00
-5.72795630e-01 -1.95493728e-01 -4.42268074e-01 6.58065915e-01
3.16775113e-01 -4.43728745e-01 1.69046268e-01 5.94401062e-01
-2.10086197e-01 -8.32463384e-01 -5.40246725e-01 -4.98869300e-01
2.58773774e-01 -3.70917767e-01 -2.85323896e-02 6.85448527e-01
5.13053179e-01 3.08254421e-01 2.32711568e-01 -4.75730419e-01
4.18736577e-01 -8.22261628e-03 -5.80512024e-02 -1.53307068e+00
3.13784122e-01 -9.17248726e-01 -1.02010265e-01 -1.65645087e+00
4.67726961e-02 -9.28189814e-01 1.08005129e-01 -2.25376654e+00
4.81595933e-01 1.74440332e-02 -7.44828463e-01 3.22800010e-01
-1.95442706e-01 2.90864736e-01 1.20825447e-01 6.82462275e-01
-9.47802722e-01 2.93915361e-01 1.03853154e+00 -1.65667892e-01
1.12892449e-01 5.30042015e-02 -9.03297365e-01 7.43950844e-01
7.08885550e-01 -5.12851894e-01 -4.17790741e-01 -8.00550342e-01
5.32280326e-01 -3.44369054e-01 -2.59992063e-01 -5.13194799e-01
6.29627287e-01 1.01102367e-01 1.38205186e-01 -8.55578363e-01
5.80845892e-01 -8.46235275e-01 -4.79549080e-01 3.33920270e-01
-8.23547900e-01 2.34119654e-01 6.14371896e-02 4.15911704e-01
-8.36609602e-01 -1.18232167e+00 4.17231351e-01 -6.95770979e-01
-8.28894615e-01 1.19033851e-01 -7.41939127e-01 -2.39890754e-01
6.28588021e-01 3.61629464e-02 -2.28095308e-01 -4.91229147e-01
-7.78558552e-01 4.09127831e-01 1.47469416e-01 7.80372381e-01
8.26960027e-01 -1.10178113e+00 -3.83993119e-01 -2.89022960e-02
4.39507544e-01 -3.84887040e-01 7.82896057e-02 3.47323895e-01
-6.44004166e-01 1.42295730e+00 5.19410707e-02 -2.35828012e-02
-1.25446916e+00 7.89670825e-01 7.12082675e-03 -7.42574453e-01
-1.44005150e-01 7.47809708e-01 -9.37502980e-02 -4.43018340e-02
7.83327043e-01 4.02804539e-02 -1.22416031e+00 4.60305154e-01
9.33313906e-01 3.44000816e-01 3.93319905e-01 -3.37952912e-01
-2.95563489e-01 6.97228491e-01 -6.24644578e-01 -2.51267433e-01
1.01890075e+00 -3.47490549e-01 -6.85358226e-01 7.68590987e-01
1.27614212e+00 -8.05926263e-01 2.37584990e-02 -6.27124906e-01
3.52042109e-01 -3.14841926e-01 4.59066451e-01 -8.84694517e-01
-8.43529284e-01 1.13041747e+00 5.22128046e-01 7.04700053e-01
1.10312617e+00 1.62094891e-01 6.04075134e-01 8.64367902e-01
3.67620528e-01 -1.41650796e+00 7.48086795e-02 9.32498872e-01
9.16045845e-01 -1.04157889e+00 2.38144919e-01 -3.78006876e-01
-1.57662660e-01 1.41262877e+00 5.28575368e-02 -2.93907493e-01
1.22391009e+00 -2.56731540e-01 2.14466423e-01 -5.19910216e-01
-8.44295442e-01 -3.30688626e-01 9.03219819e-01 4.28826779e-01
9.48467493e-01 -6.22646213e-01 -9.72217739e-01 1.75202042e-01
4.04950976e-02 -1.23331480e-01 -1.07853241e-01 1.38123071e+00
-9.58454311e-01 -1.39531898e+00 -3.29563588e-01 2.72941172e-01
-8.24024260e-01 -6.62265718e-01 -6.97912037e-01 6.06608868e-01
-5.86299837e-01 1.12736452e+00 6.30134298e-03 1.83951184e-01
1.75617367e-01 4.61646825e-01 5.71217716e-01 -7.99975038e-01
-9.27911401e-01 5.38440309e-02 2.39845309e-02 -5.22737145e-01
-8.53228509e-01 -4.94834006e-01 -7.94105470e-01 1.25589594e-01
-4.02161449e-01 5.00730395e-01 8.18226576e-01 9.88715470e-01
2.56092936e-01 3.81913722e-01 2.41284698e-01 -9.14879501e-01
-4.44261193e-01 -1.59053159e+00 -5.41802526e-01 7.67939910e-02
2.26426855e-01 -7.09531665e-01 -5.18824935e-01 -7.34212920e-02] | [11.41093635559082, 7.690741539001465] |
6eba21d9-7c4f-405c-968b-0b793283247d | qualities-challenges-and-future-of-genetic | 2011.05277 | null | https://arxiv.org/abs/2011.05277v3 | https://arxiv.org/pdf/2011.05277v3.pdf | Qualities, challenges and future of genetic algorithms: a literature review | Genetic algorithms, computer programs that simulate natural evolution, are increasingly applied across many disciplines. They have been used to solve various optimisation problems from neural network architecture search to strategic games, and to model phenomena of adaptation and learning. Expertise on the qualities and drawbacks of this technique is largely scattered across the literature or former, motivating an compilation of this knowledge at the light of the most recent developments of the field. In this review, we present genetic algorithms, their qualities, limitations and challenges, as well as some future development perspectives. Genetic algorithms are capable of exploring large and complex spaces of possible solutions, to quickly locate promising elements, and provide an adequate modelling tool to describe evolutionary systems, from games to economies. They however suffer from high computation costs, difficult parameter configuration, and crucial representation of the solutions. Recent developments such as GPU, parallel and quantum computing, conception of powerful parameter control methods, and novel approaches in representation strategies, may be keys to overcome those limitations. This compiling review aims at informing practitioners and newcomers in the field alike in their genetic algorithm research, and at outlining promising avenues for future research. It highlights the potential for interdisciplinary research associating genetic algorithms to pulse original discoveries in social sciences, open ended evolution, artificial life and AI. | ['Doyne J. Farmer', 'Alissa M. Kleinnijenhuis', 'Aymeric Vie'] | 2020-11-05 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [ 2.24382475e-01 -1.88376367e-01 -3.02551743e-02 3.51711750e-01
3.00230086e-01 -5.12069166e-01 5.26365280e-01 9.50967055e-03
-6.28576756e-01 1.10216129e+00 -3.90288591e-01 -2.30596259e-01
-7.82941401e-01 -8.60106885e-01 -1.20705113e-01 -1.17922044e+00
-2.69019276e-01 5.14877498e-01 -1.40170408e-02 -8.65867496e-01
8.47895026e-01 4.35144126e-01 -2.26677752e+00 -2.87862331e-01
1.22488225e+00 6.56452537e-01 3.16077799e-01 5.47646165e-01
-1.60897315e-01 2.47564271e-01 -5.94169617e-01 -4.86291885e-01
8.22807923e-02 -9.04039264e-01 -3.57753456e-01 -3.22338551e-01
-7.74678409e-01 7.76201963e-01 4.03889902e-02 1.03740335e+00
9.66371238e-01 3.37833405e-01 6.14363670e-01 -1.06868064e+00
-8.79059613e-01 4.16251510e-01 -2.47755110e-01 4.94626284e-01
4.02789891e-01 3.21696132e-01 6.94654465e-01 -3.98053080e-01
5.43654084e-01 9.81298089e-01 6.08245552e-01 9.68677104e-01
-7.31394112e-01 -3.92737627e-01 -1.72118545e-01 6.16007984e-01
-1.39917886e+00 -2.59958595e-01 4.73336190e-01 -7.71285146e-02
1.28873944e+00 7.07380354e-01 1.42511368e+00 6.99103832e-01
3.44005555e-01 4.83886003e-01 1.06163454e+00 -8.88606966e-01
6.72581136e-01 2.28146359e-01 -5.33255339e-01 5.95795035e-01
3.21682185e-01 7.30635583e-01 -6.25408828e-01 -2.65969545e-01
5.47695339e-01 -5.07417679e-01 -2.02225626e-01 -4.39520627e-01
-8.57926667e-01 1.12874639e+00 1.16801724e-01 5.79570830e-01
-5.39458871e-01 9.72346067e-02 1.67694747e-01 3.15439731e-01
2.85629947e-02 1.19174850e+00 -3.27687234e-01 -5.99565804e-01
-5.64566135e-01 5.83844423e-01 8.42248440e-01 3.79140258e-01
2.95011401e-01 3.06243896e-01 3.75668049e-01 8.14207375e-01
3.00556332e-01 2.37697050e-01 8.35036993e-01 -9.70608234e-01
-2.71835387e-01 7.28599787e-01 -1.54843643e-01 -9.74850595e-01
-6.78483903e-01 -6.93660259e-01 -7.54422545e-01 2.97528207e-01
7.69963786e-02 -2.92113274e-01 -5.02388597e-01 1.53440201e+00
4.89444643e-01 -3.00033297e-02 1.73700273e-01 7.11460948e-01
4.72358197e-01 7.49025941e-01 -3.93324606e-02 -5.75630784e-01
1.31617033e+00 -7.69424736e-01 -5.77739656e-01 -1.34109989e-01
4.20231760e-01 -6.88356936e-01 5.08468449e-01 5.28151095e-01
-1.43631327e+00 6.57364260e-03 -8.73718262e-01 7.14162409e-01
-7.56345749e-01 -3.47445399e-01 1.11084533e+00 1.44585741e+00
-1.40479517e+00 6.87003434e-01 -7.51781940e-01 -7.24368453e-01
1.92906246e-01 7.60732949e-01 4.03941065e-01 2.61701703e-01
-1.47756600e+00 1.25137806e+00 6.95887089e-01 1.88702837e-01
2.88442075e-02 -4.20522481e-01 -4.91932958e-01 3.66143547e-02
4.93279755e-01 -1.15234792e+00 6.71163082e-01 -1.11715519e+00
-2.02418447e+00 8.45208466e-01 1.41870320e-01 -4.81490940e-01
2.29749784e-01 6.05466127e-01 -6.24666750e-01 -2.50295848e-01
-5.04267097e-01 3.89553457e-01 3.48229110e-01 -8.00340712e-01
-5.77918768e-01 -4.29634303e-01 -3.32172602e-01 6.23679221e-01
-2.33494475e-01 3.96047503e-01 -1.16716675e-01 -6.55722082e-01
1.44618321e-02 -1.09772348e+00 -8.89399886e-01 -4.33184743e-01
3.16749066e-01 -2.29156718e-01 1.67352527e-01 -9.74506289e-02
1.49757564e+00 -1.73270237e+00 7.07342088e-01 4.56592381e-01
-1.50152475e-01 6.64693296e-01 -3.10820844e-02 8.27536643e-01
2.58308351e-01 1.94493845e-01 -3.79830718e-01 4.75433320e-01
-3.83261591e-02 2.71423638e-01 1.68870136e-01 2.81915814e-01
3.95416170e-02 1.15464282e+00 -8.69442940e-01 5.75382076e-02
2.64375240e-01 4.37376738e-01 -5.47993541e-01 -2.79849350e-01
-4.76679578e-02 2.86989927e-01 -5.76967418e-01 6.89657748e-01
2.24169329e-01 4.34389897e-02 2.56834507e-01 5.52839220e-01
-5.47611415e-01 -1.09438367e-01 -1.24030674e+00 1.31553352e+00
-1.36967704e-01 4.71777022e-01 -2.99525801e-02 -1.33120489e+00
1.04680943e+00 2.34476760e-01 2.98003614e-01 -1.02468276e+00
4.50158894e-01 6.21334553e-01 7.30735242e-01 -4.92783308e-01
4.99415725e-01 7.17482576e-03 1.06813826e-01 5.25341749e-01
-7.56210089e-02 -4.56009269e-01 5.28770030e-01 -4.64603513e-01
6.28832877e-01 1.11086190e-01 6.16098940e-01 -3.87103140e-01
8.93510640e-01 2.43276786e-02 3.83533299e-01 8.58246207e-01
-2.77078569e-01 5.47984727e-02 2.31493771e-01 -7.67731488e-01
-1.06196427e+00 -4.57478434e-01 -1.92460492e-01 1.06835961e+00
1.70706958e-01 -3.44974607e-01 -8.39319348e-01 3.58891398e-01
-3.13564807e-01 6.50986731e-01 -5.89181364e-01 -5.89978755e-01
-6.45523489e-01 -1.70944381e+00 3.87653828e-01 -9.74220410e-02
2.63235122e-01 -1.42976904e+00 -1.14231575e+00 5.08214116e-01
2.78171420e-01 -3.47647816e-01 4.20779586e-01 8.00569355e-02
-8.74874711e-01 -8.08289170e-01 -7.67935336e-01 -7.07621455e-01
2.23984510e-01 -3.53954509e-02 1.02791321e+00 5.11175215e-01
-8.01090002e-01 2.00188115e-01 -2.92571247e-01 -5.61820209e-01
-6.12242937e-01 1.11434355e-01 1.77060604e-01 -4.37074184e-01
3.88344437e-01 -6.98911667e-01 -5.17103434e-01 3.52599293e-01
-7.39601135e-01 -1.81290239e-01 4.36964035e-01 1.17151022e+00
2.47056484e-01 3.08934391e-01 7.99875319e-01 -4.67924237e-01
1.03736758e+00 -6.44321680e-01 -6.99140906e-01 4.61704195e-01
-1.07136619e+00 9.36711580e-02 4.05340374e-01 -2.64842272e-01
-9.08161640e-01 -4.82589930e-01 -7.92449489e-02 3.53165388e-01
-1.48089016e-02 6.17856979e-01 1.89173922e-01 -6.63071811e-01
8.86800766e-01 4.86290097e-01 2.91156948e-01 -6.76584095e-02
2.51479715e-01 4.59643215e-01 1.56530395e-01 -4.61416990e-01
4.44421381e-01 4.06646505e-02 2.91508734e-01 -8.88988435e-01
3.11763193e-02 -5.41282296e-02 -2.05391675e-01 -3.57550293e-01
5.98320365e-01 -1.34323493e-01 -7.54490852e-01 6.28519237e-01
-6.93393290e-01 7.55984262e-02 -3.98579627e-01 3.07164669e-01
-6.81726098e-01 9.56122130e-02 -5.41359261e-02 -1.22130513e+00
-4.27511960e-01 -1.18195963e+00 2.79783845e-01 1.03953779e+00
-1.78824425e-01 -1.25664639e+00 3.47391725e-01 2.07188830e-01
7.51719654e-01 1.92902327e-01 9.24122870e-01 -4.53350306e-01
-4.00301814e-01 -7.12623298e-02 5.64613998e-01 -2.23571569e-01
-1.98152885e-01 4.63178843e-01 -4.53233480e-01 -1.91909984e-01
6.29309863e-02 -6.27249628e-02 5.37614167e-01 5.46095192e-01
8.32170248e-01 -3.10775898e-02 -5.48689425e-01 7.00232387e-01
1.29453075e+00 1.11614656e+00 7.34379053e-01 9.47337329e-01
-3.36713530e-02 8.57803762e-01 5.06897628e-01 7.20408738e-01
1.44201785e-01 7.28507519e-01 3.67756277e-01 2.05942973e-01
3.05784345e-01 4.20931518e-01 1.32213905e-01 9.95989919e-01
-7.93114007e-01 -4.99399692e-01 -1.00924361e+00 2.06022039e-01
-1.85024035e+00 -1.10243273e+00 8.82856473e-02 2.20116663e+00
5.07164299e-01 -9.75583866e-02 3.65212411e-01 2.02912033e-01
9.63484704e-01 -3.16583127e-01 -6.63038671e-01 -1.07474077e+00
-6.42999172e-01 5.38137257e-01 2.67217845e-01 1.54084459e-01
-5.24315000e-01 1.00857699e+00 7.24280787e+00 1.05272460e+00
-1.11796999e+00 -2.56658643e-01 5.00823498e-01 -2.11941376e-01
-3.13017964e-01 2.20074039e-02 -3.50637794e-01 6.08183742e-01
1.22119474e+00 -6.03218615e-01 1.11533296e+00 3.46167862e-01
1.30255088e-01 -2.59556502e-01 -1.76731780e-01 8.69267702e-01
1.43616542e-01 -1.51383185e+00 -2.10872814e-01 1.68048024e-01
1.07516265e+00 -5.18592969e-02 3.82470697e-01 5.42260259e-02
1.54597357e-01 -1.28768265e+00 5.29003263e-01 4.23049569e-01
1.78653687e-01 -1.24237418e+00 7.30570674e-01 2.78116941e-01
-7.02665329e-01 -4.51549888e-01 -4.94706929e-01 -5.25961518e-01
8.44737515e-02 2.52620317e-02 -2.30969101e-01 5.99489689e-01
7.23759770e-01 2.53117949e-01 -4.54159021e-01 1.65449703e+00
3.46419550e-02 1.19138561e-01 -4.14240509e-01 -1.05776775e+00
3.88770610e-01 -8.25687349e-01 7.29100108e-01 5.11538923e-01
5.88973105e-01 3.21621716e-01 -5.91607213e-01 9.97632563e-01
6.41572416e-01 3.85837376e-01 -3.62032831e-01 -4.57989156e-01
6.47078216e-01 9.42960978e-01 -1.13239408e+00 1.15237109e-01
-1.05741300e-01 6.19113684e-01 5.74127119e-03 6.83281794e-02
-8.07216763e-01 -4.96129721e-01 6.07244730e-01 -3.72367322e-01
4.20078933e-02 -9.37735289e-02 -4.02047873e-01 -8.95742118e-01
-5.04688859e-01 -1.28061056e+00 5.08411109e-01 -3.90824199e-01
-6.37007177e-01 6.99429750e-01 -1.60564065e-01 -6.32961750e-01
-4.49822962e-01 -6.84184730e-01 -7.35184610e-01 7.88404644e-01
-1.15880764e+00 -4.96141136e-01 1.73263684e-01 2.19498888e-01
3.11888516e-01 -7.74551094e-01 9.38198507e-01 -2.66618095e-02
-8.59701931e-01 5.04144430e-01 8.58681142e-01 -9.57014382e-01
-2.70543527e-03 -9.15509045e-01 5.31957567e-01 4.35062498e-01
6.65928721e-02 5.62999189e-01 8.68817985e-01 -2.60457486e-01
-1.58028591e+00 -1.58811599e-01 6.21966064e-01 -1.88495904e-01
4.89688039e-01 -2.05592550e-02 -5.92307150e-01 -2.26196051e-01
2.73375034e-01 -8.15182328e-01 9.06378150e-01 3.74924652e-02
6.20981097e-01 4.29858804e-01 -1.16847742e+00 1.07200789e+00
1.02354944e+00 1.14554174e-01 -2.71219343e-01 1.16323300e-01
6.68981820e-02 -3.65051001e-01 -5.06648540e-01 2.43183747e-01
7.68826783e-01 -1.34385109e+00 1.18742371e+00 -5.82463205e-01
3.54389139e-02 -5.13706058e-02 3.45162779e-01 -1.50448096e+00
-5.91037035e-01 -1.23393321e+00 6.79469034e-02 9.14443433e-01
4.05724615e-01 -1.11253655e+00 7.74932623e-01 5.82421422e-01
-1.42302010e-02 -1.22927332e+00 -1.13961029e+00 -5.69810450e-01
2.42871866e-01 -1.16414502e-01 9.78157759e-01 8.47280443e-01
2.41893187e-01 -1.29752830e-01 -1.99095964e-01 -4.29758400e-01
3.39598954e-01 -1.19740613e-01 2.73820579e-01 -1.28029239e+00
-3.33550006e-01 -1.31214857e+00 -6.88555300e-01 -1.37470290e-01
-1.37071431e-01 -6.44935906e-01 -4.54421639e-01 -1.20264804e+00
-1.46943003e-01 -5.38199544e-01 -3.05687398e-01 -2.25488275e-01
-1.94156155e-01 3.53359520e-01 -1.01860762e-02 8.85108635e-02
-2.46426702e-01 5.76047659e-01 1.18416703e+00 3.73854458e-01
-5.70368767e-01 1.77095219e-01 -9.38925385e-01 7.96731651e-01
1.06351161e+00 -3.75368118e-01 -3.83426964e-01 -1.00223850e-02
9.85244811e-01 -1.95821211e-01 -7.11782500e-02 -1.06659222e+00
3.66115659e-01 -5.24444282e-01 8.34831223e-02 1.13051109e-01
2.85343617e-01 -4.65857476e-01 6.21583760e-01 1.03962350e+00
-3.94677036e-02 5.09091854e-01 2.25872532e-01 3.30358595e-01
-2.99510118e-02 -7.32385993e-01 7.85753667e-01 -5.07275641e-01
-8.31454158e-01 -7.63716772e-02 -9.00670886e-01 1.40311429e-02
1.48578477e+00 -8.15349102e-01 -5.63331656e-02 -1.25595480e-01
-7.07258403e-01 2.55399406e-01 6.41923547e-01 4.33608145e-01
3.65591735e-01 -9.68233824e-01 -4.76752669e-01 1.87541589e-01
-2.31872737e-01 -6.27255559e-01 3.10600966e-01 7.16506422e-01
-1.04285288e+00 5.53685844e-01 -6.58827484e-01 -2.21765041e-01
-1.12631309e+00 6.41254365e-01 7.22611547e-01 1.01828173e-01
-6.68974519e-02 1.15447545e+00 -2.22652450e-01 -4.60201323e-01
-3.79006588e-03 4.48421180e-01 -5.40522039e-01 7.11886911e-03
4.15246934e-01 7.25699306e-01 1.01567032e-02 -4.68176186e-01
-4.80266929e-01 9.08543587e-01 3.13974530e-01 -9.32545140e-02
1.42206550e+00 -2.38957793e-01 -5.89623690e-01 1.89030126e-01
3.79978448e-01 -4.11776602e-01 -3.81859273e-01 2.33843997e-01
7.30222091e-02 -3.22288483e-01 4.38598283e-02 -9.63373899e-01
-8.58566165e-01 5.94774783e-01 6.79955363e-01 5.32657385e-01
1.46988857e+00 -2.92806655e-01 3.39723349e-01 3.10236275e-01
4.70963806e-01 -1.31274056e+00 -3.16951483e-01 5.13766170e-01
6.83023632e-01 -8.24595273e-01 2.36421190e-02 -2.12219685e-01
-4.88515198e-01 1.29225624e+00 4.56767440e-01 -8.86086971e-02
3.78508478e-01 1.85353369e-01 -1.51943907e-01 -2.61010617e-01
-1.08477747e+00 -2.99793929e-01 1.33134037e-01 8.39447498e-01
4.64690387e-01 -8.71125385e-02 -1.11753464e+00 2.13604480e-01
-5.48779786e-01 -2.01636702e-01 5.55605352e-01 9.68471110e-01
-4.89351392e-01 -1.68485224e+00 -6.26645207e-01 2.66454160e-01
-4.65367109e-01 -1.51870936e-01 -6.14709914e-01 6.26059592e-01
3.26667339e-01 7.86176980e-01 -1.76342830e-01 -1.49355188e-01
3.12379245e-02 -8.80384371e-02 8.10092509e-01 -1.04360037e-01
-1.09290457e+00 -2.46332243e-01 2.72986237e-02 -2.40971908e-01
-5.01173139e-01 -9.77229953e-01 -8.81677806e-01 -5.88937819e-01
-6.83851600e-01 7.15496540e-01 9.45383132e-01 8.02270412e-01
4.95270461e-01 5.00935972e-01 3.80567372e-01 -8.31533372e-01
-3.28981996e-01 -3.26271951e-01 -5.55584490e-01 -3.99520159e-01
-3.87204975e-01 -8.97860348e-01 -1.91448256e-01 -5.99447370e-01] | [5.731607913970947, 3.8675084114074707] |
79f0d305-d1a5-48e6-aa2c-6be15654b5da | cutmib-boosting-light-field-super-resolution | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Xiao_CutMIB_Boosting_Light_Field_Super-Resolution_via_Multi-View_Image_Blending_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Xiao_CutMIB_Boosting_Light_Field_Super-Resolution_via_Multi-View_Image_Blending_CVPR_2023_paper.pdf | CutMIB: Boosting Light Field Super-Resolution via Multi-View Image Blending | Data augmentation (DA) is an efficient strategy for improving the performance of deep neural networks. Recent DA strategies have demonstrated utility in single image super-resolution (SR). Little research has, however, focused on the DA strategy for light field SR, in which multi-view information utilization is required. For the first time in light field SR, we propose a potent DA strategy called CutMIB to improve the performance of existing light field SR networks while keeping their structures unchanged. Specifically, CutMIB first cuts low-resolution (LR) patches from each view at the same location. Then CutMIB blends all LR patches to generate the blended patch and finally pastes the blended patch to the corresponding regions of high-resolution light field views, and vice versa. By doing so, CutMIB enables light field SR networks to learn from implicit geometric information during the training stage. Experimental results demonstrate that CutMIB can improve the reconstruction performance and the angular consistency of existing light field SR networks. We further verify the effectiveness of CutMIB on real-world light field SR and light field denoising. The implementation code is available at https://github.com/zeyuxiao1997/CutMIB. | ['Zhiwei Xiong', 'Ruisheng Gao', 'Yutong Liu', 'Zeyu Xiao'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['image-super-resolution'] | ['computer-vision'] | [ 4.24767286e-01 -3.26284081e-01 2.28888076e-02 -3.57710183e-01
-5.24048686e-01 -1.55418217e-01 1.27015844e-01 -6.63209736e-01
-6.51422739e-02 7.20093608e-01 1.47349268e-01 1.96849093e-01
-1.38756558e-01 -9.58608329e-01 -7.25001812e-01 -9.97844040e-01
5.63874483e-01 -1.12828836e-01 3.55176240e-01 -1.65660590e-01
1.12269625e-01 6.65018916e-01 -1.36423564e+00 3.36709380e-01
8.47892225e-01 8.53230596e-01 6.74325407e-01 3.20274889e-01
-8.64517242e-02 8.86730850e-01 -8.75450075e-02 -4.34798673e-02
5.04944384e-01 -2.64343381e-01 -6.33680642e-01 -3.28586809e-02
9.57559586e-01 -8.17964375e-01 -6.62031233e-01 1.29631770e+00
5.30037105e-01 1.47921413e-01 -2.99324393e-02 -7.95622110e-01
-7.88310051e-01 3.70297968e-01 -9.64442372e-01 5.25582910e-01
9.88508016e-02 6.20521791e-02 6.61578357e-01 -1.13066387e+00
7.59471416e-01 1.09193873e+00 4.65664238e-01 6.47946119e-01
-1.21263528e+00 -1.09868419e+00 -8.01357266e-04 3.92610170e-02
-1.08159304e+00 -4.75303113e-01 1.17457974e+00 -9.71883535e-02
5.49748182e-01 2.23007426e-02 6.55968726e-01 7.04049468e-01
1.51627094e-01 7.13214695e-01 1.45000482e+00 -1.73260719e-01
-1.58872083e-01 -2.22551718e-01 2.41127878e-01 6.44028068e-01
1.56889796e-01 4.25722480e-01 -7.52326787e-01 2.36734822e-01
1.51035380e+00 2.47969270e-01 -7.11421132e-01 -1.21696755e-01
-1.14911211e+00 4.25810307e-01 7.82267869e-01 4.80666876e-01
-3.76129836e-01 1.26258910e-01 -9.28984955e-02 1.38811857e-01
8.03644001e-01 1.87737569e-01 -2.08510801e-01 4.05381292e-01
-7.55099833e-01 8.64486694e-02 -1.32814813e-02 6.66742444e-01
9.96994257e-01 2.31278881e-01 7.46921226e-02 1.12664354e+00
-4.86411562e-04 6.20210290e-01 4.04126495e-02 -1.29712570e+00
3.12893122e-01 4.02481854e-01 1.41868154e-02 -1.01654923e+00
-3.13548833e-01 -5.16711771e-01 -1.20619762e+00 3.68874937e-01
1.26167700e-01 1.38272882e-01 -8.69077921e-01 1.50520420e+00
4.22044605e-01 4.08373058e-01 -2.46714056e-01 1.13694358e+00
1.30909717e+00 7.98169315e-01 -4.36672777e-01 -3.42182934e-01
8.65711153e-01 -7.44543254e-01 -7.13440537e-01 -1.07651442e-01
-1.04771256e-01 -8.58041525e-01 9.54517841e-01 6.60133600e-01
-1.55042791e+00 -6.75611496e-01 -7.67800570e-01 -3.86765867e-01
2.90494561e-01 2.33255640e-01 7.31066465e-01 3.17749232e-02
-1.02922833e+00 3.52937669e-01 -7.69000232e-01 1.99564323e-01
7.66163409e-01 1.12459399e-01 -3.41554701e-01 -6.49412394e-01
-8.51310730e-01 4.84963417e-01 -4.72856760e-02 2.65755117e-01
-8.30346942e-01 -1.11674762e+00 -6.90228224e-01 -1.84250101e-01
2.39383489e-01 -7.56434023e-01 1.04545176e+00 -7.07621396e-01
-1.43409228e+00 7.14946687e-01 -2.65488416e-01 9.20789130e-03
8.66748020e-02 -1.08812854e-01 -4.05133605e-01 3.80666941e-01
1.28501445e-01 6.27493024e-01 1.00546277e+00 -1.61113298e+00
-7.18882203e-01 -4.66216326e-01 3.62980925e-02 1.27694756e-01
-6.08330891e-02 -2.53531262e-02 -3.91678780e-01 -7.38620222e-01
5.57645380e-01 -5.74786007e-01 -6.03916459e-02 1.54614344e-01
-1.42010927e-01 -1.86124891e-01 8.78662527e-01 -6.47210002e-01
9.49488759e-01 -2.21692920e+00 1.74480394e-01 -6.26977161e-02
5.61032355e-01 3.51365358e-01 -2.45484620e-01 -1.67740837e-01
-2.56773621e-01 -3.92317504e-01 -3.02958101e-01 -2.89793402e-01
-7.16961265e-01 2.20258355e-01 -4.59787041e-01 5.01888990e-01
-3.62752527e-02 7.43163884e-01 -8.07313204e-01 -2.11009994e-01
4.42176312e-01 7.94441283e-01 -5.91583133e-01 9.75313038e-02
-1.22105531e-01 8.77731442e-01 -3.19701403e-01 6.88579202e-01
1.18838286e+00 -3.38897586e-01 -2.45736554e-01 -7.99624741e-01
-3.70816171e-01 1.14511676e-01 -1.34274805e+00 1.77963829e+00
-7.73335814e-01 6.71784282e-01 3.36679280e-01 -6.26885891e-01
9.33467507e-01 -6.41076937e-02 7.24449992e-01 -9.14440930e-01
-3.20706554e-02 1.81718186e-01 -3.15397292e-01 -3.51689070e-01
2.74841487e-01 -3.57875109e-01 7.38432646e-01 4.49905753e-01
-1.96420372e-01 -2.46679589e-01 -3.48359197e-02 9.83629748e-02
5.46356797e-01 2.02216223e-01 -7.13838339e-02 1.14628430e-02
5.53139806e-01 -4.43723381e-01 8.43108058e-01 5.16602755e-01
1.57866795e-02 9.59477544e-01 -3.36728424e-01 -5.95880985e-01
-1.02504528e+00 -1.23046577e+00 -3.59473139e-01 8.16294074e-01
4.83795285e-01 9.87079646e-03 -5.28419614e-01 -4.48349774e-01
-2.32077643e-01 4.00434256e-01 -3.38833064e-01 1.85189545e-01
-9.46934581e-01 -6.81589246e-01 -3.38330641e-02 4.85671848e-01
1.10444629e+00 -1.02331173e+00 -3.51225257e-01 1.03726581e-01
-5.12156844e-01 -1.15072727e+00 -5.22463024e-01 -2.67020136e-01
-8.20670247e-01 -9.89690721e-01 -7.06793725e-01 -7.39790678e-01
7.62781024e-01 1.02847838e+00 9.80052471e-01 2.44663075e-01
-2.74380207e-01 1.32990882e-01 -7.10359216e-02 1.34369554e-02
-1.11347020e-01 -3.58939201e-01 -1.24129586e-01 1.81124389e-01
4.64775003e-02 -1.01842427e+00 -9.81938183e-01 3.75884533e-01
-1.14577174e+00 5.10715067e-01 6.46630347e-01 8.05312812e-01
8.73812377e-01 2.84252763e-01 2.94079840e-01 -6.77230179e-01
1.50647938e-01 -7.78980330e-02 -8.19397449e-01 4.89654765e-02
-5.03647387e-01 -3.27894479e-01 4.74436343e-01 -4.02150393e-01
-1.56711686e+00 -2.22815052e-02 -3.08991998e-01 -6.74293280e-01
-6.27162978e-02 8.23134482e-02 -8.57938081e-02 -5.98899245e-01
4.27475274e-01 6.22127593e-01 2.14326219e-03 -7.25624740e-01
2.01608062e-01 4.10912216e-01 6.97897732e-01 -2.57550299e-01
8.68983030e-01 9.97088253e-01 1.64574131e-01 -9.07025516e-01
-1.19771314e+00 -4.33122993e-01 -4.45448488e-01 -3.20121080e-01
8.28589737e-01 -9.65092421e-01 -6.20329201e-01 7.61720657e-01
-9.99068856e-01 -3.75863761e-01 -1.34522229e-01 4.73737836e-01
-2.84941882e-01 4.78205085e-01 -9.05157566e-01 -3.10289145e-01
-4.04513448e-01 -1.00947893e+00 1.01727104e+00 5.05227983e-01
6.26962721e-01 -7.21929789e-01 -7.94365853e-02 6.54125392e-01
4.66662914e-01 -2.15690836e-01 6.34820819e-01 5.63989401e-01
-1.09034288e+00 3.26500118e-01 -6.46250725e-01 7.01885641e-01
2.92431325e-01 -2.19042853e-01 -9.73320603e-01 -3.55292976e-01
3.62468004e-01 -7.33357817e-02 9.78375316e-01 7.66915321e-01
1.25043118e+00 -7.07330480e-02 -1.26895428e-01 1.18819594e+00
1.73871076e+00 1.47162572e-01 8.15159559e-01 3.86669725e-01
1.03366983e+00 3.35331142e-01 5.01564980e-01 1.48403019e-01
2.27425545e-01 8.14166963e-01 4.36179698e-01 -4.20688540e-01
-7.11311400e-01 -1.50346160e-01 2.00744212e-01 6.54279411e-01
-2.50083447e-01 2.51405358e-01 -5.74698448e-01 3.31151605e-01
-1.29795384e+00 -1.10915017e+00 -3.74238223e-01 1.96864200e+00
9.15270329e-01 -1.81806982e-01 -2.42541283e-01 4.23455052e-02
5.48032880e-01 3.66575599e-01 -5.64706385e-01 2.46400669e-01
-3.31588507e-01 2.92063326e-01 3.52122009e-01 7.77510166e-01
-7.97636569e-01 8.51837933e-01 5.28845072e+00 9.03066695e-01
-1.37127125e+00 2.73065865e-01 5.32955050e-01 -2.02336937e-01
-4.23114598e-01 -1.17688902e-01 -9.65505362e-01 4.01478231e-01
-9.85550359e-02 1.42172739e-01 7.20032096e-01 5.55789232e-01
3.76899183e-01 -2.58321971e-01 -6.58741713e-01 1.16865301e+00
4.91102748e-02 -1.50741208e+00 8.57224837e-02 -6.76918700e-02
9.77181137e-01 2.66645372e-01 1.51222497e-01 -2.40851864e-01
2.05092609e-01 -8.16597521e-01 1.79039180e-01 6.12242281e-01
8.56630147e-01 -6.88172460e-01 4.31944668e-01 3.32238942e-01
-1.08565235e+00 -4.71820347e-02 -5.21437407e-01 2.47975647e-01
3.73921871e-01 1.02564955e+00 -3.82898062e-01 7.17197955e-01
9.96258557e-01 1.14539838e+00 -4.27992761e-01 8.31801832e-01
-2.98602313e-01 4.25933838e-01 -2.71508247e-01 7.57100761e-01
-9.40043777e-02 -6.17781043e-01 8.26382518e-01 7.18925238e-01
1.41379148e-01 3.45085979e-01 3.41608264e-02 1.08167481e+00
1.22270873e-03 -3.55202794e-01 -4.94496644e-01 6.33914530e-01
3.04208279e-01 1.41306937e+00 -4.22118485e-01 -2.00258136e-01
-6.01652443e-01 8.33495021e-01 1.17029913e-01 5.74943542e-01
-5.51100075e-01 6.39399961e-02 5.63453197e-01 5.08587837e-01
2.25658938e-01 -2.07982287e-01 -4.29408014e-01 -1.29744613e+00
2.08171755e-01 -8.10053110e-01 2.75276393e-01 -1.32411933e+00
-1.37863398e+00 6.80814564e-01 -2.74149701e-02 -1.35552192e+00
3.61320496e-01 -2.78831035e-01 -4.31786895e-01 9.49405253e-01
-1.94477391e+00 -1.35502684e+00 -6.88178360e-01 7.80570984e-01
5.44586599e-01 5.87458275e-02 1.23650409e-01 4.74490672e-01
-6.02121234e-01 1.46156296e-01 1.62128687e-01 5.13753034e-02
5.16414046e-01 -8.85788441e-01 2.16805004e-02 1.04756618e+00
-2.36846842e-02 6.44013286e-01 3.91303420e-01 -7.60562241e-01
-1.17205977e+00 -1.08088899e+00 3.71510506e-01 -1.12344131e-01
2.22558901e-01 -6.37400001e-02 -1.11775267e+00 5.68123162e-01
6.09104373e-02 3.89022648e-01 1.03019290e-01 -2.82392204e-01
-3.01413447e-01 -5.85382819e-01 -1.18219316e+00 3.14575434e-01
1.16752326e+00 -4.24859911e-01 -3.83247495e-01 2.59371161e-01
5.15370429e-01 -6.54491186e-01 -8.39769661e-01 6.77397847e-01
2.92492747e-01 -1.32362902e+00 1.32261205e+00 -7.50852376e-02
6.80039465e-01 -6.58194542e-01 -7.06234127e-02 -1.25823009e+00
-4.03201491e-01 -3.17554981e-01 -8.13321248e-02 1.08647859e+00
-1.56118929e-01 -6.90579116e-01 7.51428604e-01 1.95191145e-01
-1.84244856e-01 -7.25358546e-01 -7.31642008e-01 -5.62423408e-01
-7.46340007e-02 -1.33032292e-01 3.88598025e-01 1.05447876e+00
-8.68089974e-01 2.31482565e-01 -4.14735854e-01 5.23095071e-01
1.12624419e+00 5.86541235e-01 6.55172467e-01 -1.05289841e+00
-3.16787899e-01 -2.95544088e-01 1.70095026e-01 -1.36706734e+00
1.20489271e-02 -8.61820400e-01 -1.10162251e-01 -1.66852307e+00
4.34343606e-01 -6.37440324e-01 4.87570744e-03 2.68490285e-01
-1.93674922e-01 6.67279661e-01 2.12228879e-01 4.60101247e-01
-1.64185435e-01 5.31960845e-01 2.03794193e+00 1.46817446e-01
-1.65582940e-01 4.46590520e-02 -5.71011066e-01 8.35209608e-01
6.40202880e-01 -2.16763660e-01 -4.80045885e-01 -8.57847452e-01
2.34546199e-01 8.15228224e-02 6.68196082e-01 -8.99453998e-01
2.66368687e-01 -3.41090471e-01 5.00255048e-01 -7.36896753e-01
3.36811632e-01 -7.99065948e-01 3.57795537e-01 2.11125389e-01
-1.32003903e-01 -3.17721665e-01 -4.30992953e-02 4.25965309e-01
-2.60619611e-01 -7.79600302e-03 1.29656863e+00 -3.24069321e-01
-6.74102783e-01 6.88212633e-01 1.44341797e-01 -9.87938717e-02
6.18258595e-01 -1.53974608e-01 -5.04469931e-01 -2.57133722e-01
-5.24873674e-01 -6.00705715e-03 4.86576289e-01 9.68529955e-02
9.64879990e-01 -1.19079041e+00 -7.83576787e-01 6.03535831e-01
-3.04499090e-01 6.04052603e-01 7.06268430e-01 8.96739125e-01
-5.10236263e-01 9.76116210e-02 -3.86350334e-01 -6.39888227e-01
-1.35087633e+00 4.41803724e-01 6.57516956e-01 3.82554419e-02
-1.16805732e+00 9.85355318e-01 7.05365837e-01 -1.63211718e-01
1.26063004e-01 -2.46722594e-01 -1.68647215e-01 -4.17301834e-01
8.46529007e-01 4.75552976e-01 1.70334009e-03 -4.03803349e-01
-3.88620980e-02 1.05231321e+00 -3.28767419e-01 8.92315805e-02
1.78781986e+00 -3.58963400e-01 -4.87518430e-01 4.22587432e-02
1.05384564e+00 2.84651071e-01 -1.42722964e+00 -5.98037899e-01
-6.56808138e-01 -1.00024629e+00 7.21781135e-01 -6.93612695e-01
-1.81613493e+00 7.72919178e-01 5.96263826e-01 -1.61567822e-01
1.70601666e+00 1.00415945e-02 1.17606521e+00 5.11276256e-03
4.31099236e-01 -6.25315785e-01 2.19416320e-01 2.35437289e-01
1.02014661e+00 -1.23205948e+00 1.47359312e-01 -8.32525492e-01
-3.45029980e-01 1.01007700e+00 8.32946599e-01 -2.05875367e-01
7.13309705e-01 3.23152095e-01 3.45698856e-02 -3.73198420e-01
-4.94690418e-01 -6.80512860e-02 2.33610436e-01 5.41837752e-01
2.62497395e-01 -2.92548209e-01 -1.93441898e-01 1.75722063e-01
1.31357580e-01 3.90674472e-01 6.13708973e-01 6.12907112e-01
-3.24466050e-01 -8.86918962e-01 -2.19734684e-01 2.27503270e-01
-3.44900876e-01 -2.97917783e-01 1.13800868e-01 3.48608345e-01
2.32665464e-01 7.15796113e-01 1.22493887e-02 -1.29411399e-01
2.53111660e-01 -6.09911263e-01 6.68732285e-01 -5.24921954e-01
-2.34130889e-01 2.49762982e-01 -3.12471837e-01 -7.19673932e-01
-6.84163451e-01 -4.58224922e-01 -1.13795125e+00 -2.94304758e-01
-1.13768183e-01 -3.55070382e-01 4.24370676e-01 5.70643902e-01
3.08998406e-01 4.41377252e-01 9.89769280e-01 -8.64901304e-01
-1.93278387e-01 -7.66903877e-01 -5.61743855e-01 3.22225660e-01
7.29843795e-01 -6.68360710e-01 -4.16312128e-01 -1.13651464e-02] | [10.690568923950195, -2.3334834575653076] |
0e1489e6-96a3-492f-aab6-95acf567488a | unsupervised-landmark-detection-based | 2109.14805 | null | https://arxiv.org/abs/2109.14805v3 | https://arxiv.org/pdf/2109.14805v3.pdf | Unsupervised Landmark Detection Based Spatiotemporal Motion Estimation for 4D Dynamic Medical Images | Motion estimation is a fundamental step in dynamic medical image processing for the assessment of target organ anatomy and function. However, existing image-based motion estimation methods, which optimize the motion field by evaluating the local image similarity, are prone to produce implausible estimation, especially in the presence of large motion. In this study, we provide a novel motion estimation framework of Dense-Sparse-Dense (DSD), which comprises two stages. In the first stage, we process the raw dense image to extract sparse landmarks to represent the target organ anatomical topology and discard the redundant information that is unnecessary for motion estimation. For this purpose, we introduce an unsupervised 3D landmark detection network to extract spatially sparse but representative landmarks for the target organ motion estimation. In the second stage, we derive the sparse motion displacement from the extracted sparse landmarks of two images of different time points. Then, we present a motion reconstruction network to construct the motion field by projecting the sparse landmarks displacement back into the dense image domain. Furthermore, we employ the estimated motion field from our two-stage DSD framework as initialization and boost the motion estimation quality in light-weight yet effective iterative optimization. We evaluate our method on two dynamic medical imaging tasks to model cardiac motion and lung respiratory motion, respectively. Our method has produced superior motion estimation accuracy compared to existing comparative methods. Besides, the extensive experimental results demonstrate that our solution can extract well representative anatomical landmarks without any requirement of manual annotation. Our code is publicly available online. | ['Jinman Kim', 'Qian Wang', 'Ruiyan Zhang', 'Dagan Feng', 'Zhengbin Zhu', 'Liyun Chen', 'Dongming Wei', 'Lei Bi', 'Yuyu Guo'] | 2021-09-30 | null | null | null | null | ['unsupervised-landmark-detection'] | ['computer-vision'] | [-5.10842875e-02 -3.35175335e-01 -3.72257292e-01 -5.76941222e-02
-7.39920199e-01 -4.37925875e-01 1.61872149e-01 -1.80212721e-01
-3.58960450e-01 3.87882233e-01 6.08048677e-01 4.33725603e-02
-2.21268743e-01 -4.65863228e-01 -2.38536909e-01 -1.00958502e+00
-5.49251065e-02 1.68733999e-01 3.37144732e-01 2.69243777e-01
-2.49453429e-02 4.41208303e-01 -5.76599300e-01 7.91056838e-04
7.03967929e-01 6.84002638e-01 5.13825476e-01 5.70909262e-01
-5.63127548e-02 9.66430783e-01 -1.84625715e-01 3.59121650e-01
2.34282956e-01 -5.73206782e-01 -1.01272333e+00 2.40382046e-01
1.92297414e-01 -4.99523908e-01 -4.57773596e-01 1.05681586e+00
8.06004703e-01 2.18856454e-01 4.14425343e-01 -6.29486799e-01
-2.25672364e-01 3.17991614e-01 -7.68512905e-01 7.25733399e-01
1.91109389e-01 2.13090405e-01 3.74501079e-01 -1.05789471e+00
8.86133194e-01 8.37955117e-01 7.42468596e-01 5.31413972e-01
-1.00572526e+00 -2.62738168e-01 1.28513854e-02 -2.34157946e-02
-1.55079329e+00 -3.58971328e-01 1.07919717e+00 -5.44000208e-01
4.76010174e-01 3.48239601e-01 6.82453156e-01 6.29272759e-01
3.86532307e-01 7.43634701e-01 5.26271641e-01 -3.76129039e-02
-4.32414189e-02 -2.59534389e-01 -5.52829131e-02 1.02944207e+00
2.55470186e-01 -1.48662012e-02 -2.31326655e-01 -2.58490086e-01
1.08344340e+00 2.35596731e-01 -7.39644051e-01 -4.91866708e-01
-1.87036669e+00 5.64151406e-01 6.40205026e-01 7.14643359e-01
-6.86327100e-01 1.01103023e-01 2.06947014e-01 -3.22099268e-01
2.52273917e-01 -1.05561860e-01 9.94250178e-02 1.03841228e-02
-1.09097672e+00 -2.06637204e-01 5.72092950e-01 5.89329481e-01
6.61472857e-01 1.45517334e-01 -3.16699594e-01 6.97625816e-01
7.19016492e-01 3.02706420e-01 9.87650394e-01 -1.05350614e+00
3.58777344e-01 5.30335367e-01 -6.75622597e-02 -1.40957963e+00
-5.85267901e-01 -4.37551826e-01 -1.47809577e+00 -3.61446232e-01
2.95335561e-01 -1.57315165e-01 -8.11012089e-01 1.37838399e+00
8.63236606e-01 7.35853970e-01 -2.91606467e-02 1.30586576e+00
1.07466042e+00 7.31512606e-01 1.33403003e-01 -6.16267979e-01
1.11009908e+00 -1.09045029e+00 -8.44523191e-01 -6.71498924e-02
8.92335296e-01 -7.83839881e-01 7.86525786e-01 -6.53616115e-02
-1.19097459e+00 -4.48147148e-01 -6.88362658e-01 1.43535778e-01
5.40210724e-01 2.43803650e-01 6.24022722e-01 3.33186090e-01
-8.56247127e-01 5.76382816e-01 -1.40304387e+00 -1.02469949e-02
3.37439358e-01 3.89362276e-01 -3.84763181e-01 -3.02151173e-01
-8.76613975e-01 6.88569248e-01 3.18786114e-01 6.45621836e-01
-1.01601720e+00 -8.13890696e-01 -1.06362939e+00 -2.18726650e-01
6.92687854e-02 -1.10201478e+00 1.00197351e+00 -5.44154108e-01
-1.33014250e+00 6.27886891e-01 -3.90056521e-01 1.74151715e-02
3.37018609e-01 1.77616626e-01 -9.37631056e-02 4.61627215e-01
2.03384399e-01 4.97184306e-01 5.82521558e-01 -9.91629899e-01
-1.96837887e-01 -8.59899297e-02 -4.90033656e-01 4.82277423e-01
-2.08827928e-01 -1.84183270e-01 -8.20426404e-01 -9.84857559e-01
6.41249001e-01 -1.01834726e+00 -8.25286329e-01 1.70354798e-01
-2.76890963e-01 2.87391216e-01 6.77604318e-01 -8.08107853e-01
1.61847174e+00 -2.25367999e+00 4.01164412e-01 3.59732121e-01
4.54743236e-01 1.12635620e-01 7.55912736e-02 -2.94906288e-01
9.92949158e-02 -6.68773130e-02 -6.72863364e-01 -1.46728382e-01
-6.54800951e-01 1.97047859e-01 3.21740270e-01 8.32566261e-01
-6.15380704e-02 1.00557196e+00 -1.12958527e+00 -1.12503004e+00
4.91076976e-01 5.35600245e-01 -5.92058301e-01 2.81903923e-01
4.67529804e-01 1.08262277e+00 -9.15735960e-01 6.93731129e-01
6.70513213e-01 -6.05782390e-01 2.24530295e-01 -6.73815012e-01
-9.86527577e-02 -1.46370446e-02 -1.30979335e+00 2.39031935e+00
-3.18465322e-01 1.35769397e-01 1.83225721e-01 -9.08484340e-01
4.49665755e-01 5.68599582e-01 1.34728527e+00 -2.50256002e-01
2.46652871e-01 2.51637727e-01 3.69354300e-02 -6.59625590e-01
1.27127558e-01 -2.69034714e-01 9.56922546e-02 2.73766577e-01
-1.25220835e-01 -7.45544359e-02 -5.04868217e-02 1.09137967e-01
1.07870734e+00 2.70114616e-02 3.43187869e-01 -2.82385856e-01
9.92016494e-01 1.95337236e-01 9.01951551e-01 2.62321025e-01
-5.23083925e-01 9.43780243e-01 -1.01860568e-01 -5.40974736e-01
-6.61562026e-01 -1.00445664e+00 -1.67326927e-01 1.81233421e-01
4.62726057e-01 -2.12302446e-01 -4.68523055e-01 -6.86461091e-01
-2.75429219e-01 -4.83188517e-02 -3.00636351e-01 -6.19851314e-02
-1.10546839e+00 -1.13032126e+00 2.71808565e-01 4.05729920e-01
4.64061260e-01 -8.24622929e-01 -5.45504391e-01 4.62572485e-01
-6.47556901e-01 -1.02051950e+00 -1.02806342e+00 -3.35107177e-01
-1.24563766e+00 -8.89204323e-01 -1.16893697e+00 -1.16940951e+00
1.07271075e+00 6.50563419e-01 9.01927948e-01 3.79099071e-01
-4.64956492e-01 3.19174707e-01 -4.62742746e-02 3.32905173e-01
-2.02179775e-01 -6.98810667e-02 -1.02775775e-01 3.80585864e-02
-2.10439444e-01 -5.21554351e-01 -1.20174205e+00 3.94793987e-01
-8.83604527e-01 2.02514961e-01 6.84833288e-01 7.59947002e-01
9.31364834e-01 -1.50531083e-01 4.78836857e-02 -8.00396323e-01
1.78662196e-01 -8.07201743e-01 -3.83707076e-01 7.84972310e-02
-2.54571319e-01 9.55542400e-02 3.51945519e-01 -4.78180379e-01
-1.02261305e+00 6.78484082e-01 -2.58256733e-01 -7.19734371e-01
1.36680976e-01 6.68889642e-01 1.98233530e-01 -4.60864693e-01
4.15647268e-01 5.31392157e-01 4.98182289e-02 -3.07655394e-01
2.12257162e-01 1.14342712e-01 7.23937094e-01 -3.47635716e-01
1.02116406e+00 8.53457987e-01 2.74619550e-01 -6.71566963e-01
-6.68501735e-01 -9.82722044e-01 -8.24453115e-01 -2.36793831e-01
1.15074563e+00 -8.91127646e-01 -5.79356134e-01 2.79851973e-01
-1.07025659e+00 -8.24900493e-02 -2.12023541e-01 1.10236359e+00
-3.39202374e-01 1.03566933e+00 -8.42290103e-01 -2.78361082e-01
-6.30160868e-01 -1.34341943e+00 9.61314559e-01 1.45137638e-01
-1.00402728e-01 -1.38379776e+00 4.06648815e-01 1.44140020e-01
4.03413951e-01 5.64964056e-01 5.50823331e-01 -3.19050625e-02
-9.06212509e-01 -5.96294850e-02 4.57202345e-02 -1.91094261e-02
4.13721085e-01 -3.24839145e-01 -3.50697577e-01 -3.27930778e-01
3.50228459e-01 1.68844059e-01 5.58209002e-01 9.75780010e-01
1.07883537e+00 -2.33097255e-01 -3.66309583e-01 1.16377330e+00
1.37592471e+00 9.31186005e-02 3.66561651e-01 1.14034906e-01
1.13345230e+00 3.44017327e-01 7.14974165e-01 4.85734224e-01
2.70292044e-01 5.44357181e-01 2.07793504e-01 -3.85908872e-01
-2.44919002e-01 -1.31939754e-01 1.51137009e-01 1.62990046e+00
-2.51573265e-01 1.14321969e-01 -1.02263343e+00 6.85886264e-01
-1.85735345e+00 -8.01394701e-01 -2.10608408e-01 2.18276167e+00
7.80980945e-01 -2.85723954e-01 -9.58100632e-02 -2.93857902e-01
6.32266998e-01 3.54113519e-01 -3.16715688e-01 4.02963251e-01
2.46026576e-01 -4.75925282e-02 3.57649744e-01 6.48412228e-01
-1.05008101e+00 6.68950140e-01 6.23757601e+00 5.27439833e-01
-1.19570744e+00 3.54466498e-01 5.03625989e-01 -1.41890615e-01
-3.94963771e-01 3.06431968e-02 -5.55529833e-01 5.46246827e-01
6.04789674e-01 -2.09341079e-01 4.48386669e-02 6.04461014e-01
5.81681609e-01 -1.15489937e-01 -7.02943385e-01 1.09487927e+00
-6.67679012e-02 -1.58369648e+00 5.96588962e-02 9.25189909e-03
7.47238755e-01 -8.30038562e-02 -1.43533066e-01 -1.57211080e-01
-1.09450333e-01 -9.64577258e-01 2.23731071e-01 5.74934363e-01
5.14776051e-01 -4.77785826e-01 7.35389590e-01 5.92698991e-01
-1.58195388e+00 3.83256495e-01 -4.37722057e-01 3.99968386e-01
5.45262396e-01 6.84298813e-01 -7.98771560e-01 8.17533672e-01
4.04822141e-01 1.02878571e+00 -1.99720174e-01 1.16331124e+00
8.51199925e-02 5.02813995e-01 -3.08001310e-01 3.19575667e-01
3.44872296e-01 -3.89691472e-01 7.53366172e-01 1.10267854e+00
5.16401231e-01 3.65317494e-01 4.50329363e-01 6.63836420e-01
1.27634391e-01 3.69052470e-01 -5.96990943e-01 3.68814349e-01
2.33842120e-01 1.49928904e+00 -8.73497665e-01 -4.25925910e-01
-5.36332309e-01 1.05735528e+00 -2.29357015e-02 1.49837092e-01
-8.32540691e-01 1.16989978e-01 1.54944792e-01 9.19813961e-02
-7.77546838e-02 -4.48862731e-01 1.07569829e-01 -1.47646499e+00
-1.02133118e-01 -4.69402581e-01 5.81817627e-01 -3.49637806e-01
-9.25997555e-01 3.77136618e-01 -3.53395715e-02 -1.60931230e+00
-3.24191839e-01 -1.81904659e-01 -8.22185636e-01 8.69144022e-01
-1.47902322e+00 -8.92419159e-01 -7.82438338e-01 7.94025838e-01
5.90749145e-01 2.73488045e-01 4.80092108e-01 7.16635942e-01
-6.90430939e-01 1.83759570e-01 -5.77533208e-02 3.56656581e-01
5.10643780e-01 -9.70771253e-01 1.31484261e-02 9.16678846e-01
-3.26374173e-02 6.69448733e-01 1.53437868e-01 -8.10486794e-01
-1.45697284e+00 -1.22578442e+00 5.83908916e-01 -3.42568457e-01
3.55501860e-01 3.40194702e-01 -9.78745222e-01 5.69338202e-01
-1.75678656e-01 7.78576910e-01 7.26297140e-01 -7.16123700e-01
6.17674530e-01 1.34996817e-01 -1.08430445e+00 4.08924550e-01
1.05481613e+00 -1.42815471e-01 -4.98661160e-01 4.21622217e-01
5.57259798e-01 -8.37604225e-01 -1.33770895e+00 5.62977552e-01
3.72287631e-01 -4.93659198e-01 1.37100899e+00 -1.83974966e-01
2.49768555e-01 -7.74852157e-01 1.64711729e-01 -8.68316114e-01
-6.75745189e-01 -6.95298851e-01 -3.18172008e-01 8.41054857e-01
1.00329712e-01 -2.64958978e-01 9.93995905e-01 3.54410261e-01
-3.54843289e-01 -7.57867992e-01 -9.45277154e-01 -2.84530103e-01
-1.39580682e-01 -9.45829079e-02 -1.10499086e-02 1.25720119e+00
-2.30777413e-01 5.02026007e-02 -4.71050441e-01 3.44273061e-01
7.42924929e-01 1.42365903e-01 4.88903791e-01 -6.87441647e-01
-2.86466390e-01 -1.25287324e-01 -3.83215517e-01 -1.29968441e+00
-1.53689412e-02 -1.15895605e+00 2.10793182e-01 -1.64109302e+00
4.12780136e-01 -2.79456735e-01 -5.06012857e-01 1.06225543e-01
-4.72564906e-01 4.06753838e-01 -9.60103644e-04 8.13513935e-01
-5.10979235e-01 4.62063670e-01 1.89835167e+00 -1.07767172e-01
-3.22069108e-01 7.98641518e-02 -2.99022615e-01 9.27427590e-01
4.62168276e-01 -6.05798006e-01 -5.17009377e-01 -3.57919216e-01
-3.60267192e-01 4.38821673e-01 3.07684124e-01 -1.03233349e+00
3.53121817e-01 -2.49025255e-01 5.96306145e-01 -6.40644133e-01
7.55819157e-02 -9.10557687e-01 3.27234030e-01 9.98892367e-01
-1.38093848e-02 1.02231048e-01 1.05656376e-02 5.60386539e-01
-4.83092338e-01 -2.99689204e-01 8.63459826e-01 -4.54402834e-01
-7.19205320e-01 8.34851205e-01 -2.78078437e-01 9.17231664e-02
1.02774334e+00 -7.80734792e-02 2.43496835e-01 -1.53371572e-01
-9.59419608e-01 3.04114878e-01 2.07788780e-01 -2.55380739e-02
8.84193361e-01 -1.65016162e+00 -6.35607243e-01 2.61853337e-02
-3.38219076e-01 6.10113025e-01 6.04844093e-01 1.58126390e+00
-1.06801772e+00 2.49306113e-01 5.50900511e-02 -1.07707143e+00
-1.05971599e+00 6.93894446e-01 4.35842335e-01 -5.01590014e-01
-9.82123792e-01 5.98625004e-01 5.41356921e-01 -1.38381764e-01
-5.10738045e-02 -5.63138902e-01 -2.17386201e-01 -3.73981714e-01
3.72697204e-01 2.21604615e-01 -2.24363282e-01 -9.41903114e-01
-6.81248069e-01 1.08391130e+00 2.16939300e-01 1.28811747e-01
1.10799515e+00 -3.93886179e-01 -1.21370114e-01 2.03311533e-01
1.47288561e+00 1.51331276e-01 -1.19292676e+00 -1.72336757e-01
-5.52269444e-03 -7.18530774e-01 2.45050415e-01 -1.08791981e-02
-1.33881950e+00 8.52764130e-01 6.13224626e-01 -3.79705220e-01
1.25611687e+00 -2.04087079e-01 1.10710418e+00 -1.13051124e-01
1.30514503e-01 -5.43233693e-01 1.48074746e-01 1.04063906e-01
5.38028359e-01 -1.26882529e+00 3.09965909e-01 -7.10330963e-01
-6.39198482e-01 1.01707959e+00 5.23363292e-01 -1.37165919e-01
7.09472299e-01 1.34857461e-01 1.11629188e-01 -2.22841576e-01
-3.73803884e-01 3.59621532e-02 6.46518052e-01 3.41142088e-01
5.78907013e-01 -3.76304716e-01 -4.08949256e-01 1.73734188e-01
3.56198460e-01 7.57224709e-02 1.72178894e-01 1.02355671e+00
-3.34744185e-01 -7.86260486e-01 -3.85100067e-01 3.08887102e-03
-6.47302747e-01 -7.47345015e-02 1.50030896e-01 6.34535253e-01
-6.91537187e-02 3.39268148e-01 -1.85725734e-01 -7.55873397e-02
1.83270708e-01 -2.81985372e-01 4.49663520e-01 -6.86338782e-01
-5.22847950e-01 4.32704031e-01 -2.79691696e-01 -7.12727308e-01
-6.84001803e-01 -5.51017821e-01 -1.54597628e+00 -5.36395237e-02
-1.14864349e-01 1.63616538e-01 3.95733893e-01 6.85642004e-01
1.31017283e-01 5.46395421e-01 8.49736571e-01 -9.73227501e-01
-3.00627768e-01 -5.99024653e-01 -2.36968771e-01 5.46957314e-01
4.42538887e-01 -4.39575851e-01 -3.28452259e-01 1.72793552e-01] | [14.1820707321167, -2.498248338699341] |
42849620-079a-414b-b0ea-84bac7fade33 | cascadexml-rethinking-transformers-for-end-to | 2211.00640 | null | https://arxiv.org/abs/2211.00640v1 | https://arxiv.org/pdf/2211.00640v1.pdf | CascadeXML: Rethinking Transformers for End-to-end Multi-resolution Training in Extreme Multi-label Classification | Extreme Multi-label Text Classification (XMC) involves learning a classifier that can assign an input with a subset of most relevant labels from millions of label choices. Recent approaches, such as XR-Transformer and LightXML, leverage a transformer instance to achieve state-of-the-art performance. However, in this process, these approaches need to make various trade-offs between performance and computational requirements. A major shortcoming, as compared to the Bi-LSTM based AttentionXML, is that they fail to keep separate feature representations for each resolution in a label tree. We thus propose CascadeXML, an end-to-end multi-resolution learning pipeline, which can harness the multi-layered architecture of a transformer model for attending to different label resolutions with separate feature representations. CascadeXML significantly outperforms all existing approaches with non-trivial gains obtained on benchmark datasets consisting of up to three million labels. Code for CascadeXML will be made publicly available at \url{https://github.com/xmc-aalto/cascadexml}. | ['Rohit Babbar', 'Erik Schultheis', 'Atmadeep Banerjee', 'Siddhant Kharbanda'] | 2022-10-29 | null | null | null | null | ['multi-label-text-classification', 'extreme-multi-label-classification', 'multi-label-text-classification'] | ['methodology', 'methodology', 'natural-language-processing'] | [ 2.72201419e-01 1.30008971e-02 -1.45043790e-01 -7.20285594e-01
-1.44331515e+00 -6.39740944e-01 5.99044383e-01 2.30234832e-01
-3.81740123e-01 5.68613112e-01 1.85910270e-01 -2.84346610e-01
-3.21374722e-02 -5.25940955e-01 -5.39204240e-01 -3.43311697e-01
4.41601008e-01 7.29707301e-01 -1.38326779e-01 1.13899551e-01
1.91966146e-01 4.07450572e-02 -1.59359765e+00 1.05787981e+00
4.14453179e-01 1.30368018e+00 1.40860700e-03 7.10797846e-01
-3.28400582e-01 1.09239531e+00 -1.62562296e-01 -5.68415523e-01
-1.37829736e-01 -2.49198362e-01 -1.27364993e+00 -2.78989613e-01
9.43802953e-01 -1.64793149e-01 9.42547619e-02 8.92851114e-01
5.84728122e-01 -1.31928086e-01 8.71794283e-01 -1.12930739e+00
-7.77116537e-01 1.04024446e+00 -7.86991537e-01 5.77162392e-02
5.70555404e-02 -1.05362795e-01 1.48818505e+00 -1.23710263e+00
3.65729600e-01 1.38840401e+00 8.93671274e-01 8.07877064e-01
-1.50008678e+00 -1.03934324e+00 3.79411578e-01 1.64783448e-01
-1.15412033e+00 -5.03629923e-01 3.93807709e-01 -3.05972010e-01
1.03026235e+00 2.60418385e-01 -1.04133010e-01 1.37452579e+00
1.02574006e-01 9.19706583e-01 1.42676055e+00 -5.38300276e-01
-7.58556556e-03 6.60300814e-03 4.93381053e-01 9.00089979e-01
-2.60265321e-01 -2.40853444e-01 -6.63365543e-01 -2.29078114e-01
3.41502964e-01 1.90629080e-01 -8.11465308e-02 6.27090782e-02
-1.13093507e+00 8.96109581e-01 6.71159923e-01 3.05140734e-01
-1.58572227e-01 5.99891543e-01 5.82451761e-01 3.16941440e-01
9.49862480e-01 5.30533493e-01 -8.50037694e-01 4.70907353e-02
-8.19130480e-01 -6.77910983e-04 3.70765895e-01 1.02260923e+00
7.75966406e-01 -4.62522388e-01 -4.29041386e-01 1.03976905e+00
4.24214393e-01 7.23448545e-02 6.31155849e-01 -1.03754032e+00
5.28074682e-01 8.33074570e-01 -1.91818714e-01 -2.94473618e-01
-7.00872064e-01 -6.16118371e-01 -7.74425447e-01 2.45204657e-01
3.25945050e-01 -1.16518900e-01 -9.13861513e-01 1.79398215e+00
3.33961576e-01 1.84941947e-01 -1.66492939e-01 6.48421168e-01
1.07596469e+00 5.88802278e-01 5.09161830e-01 1.47450984e-01
1.71625185e+00 -1.28738475e+00 -5.55128813e-01 -2.99861729e-01
1.07520604e+00 -7.48259664e-01 1.35184813e+00 1.60829738e-01
-1.09992909e+00 -4.61880893e-01 -8.32978070e-01 -6.17854476e-01
-5.45364320e-01 2.09926262e-01 4.75476444e-01 1.90333411e-01
-1.13379502e+00 6.71536446e-01 -5.14901042e-01 -3.27223003e-01
5.76129496e-01 3.16488385e-01 -1.83975741e-01 -1.36613637e-01
-1.20244884e+00 8.58738244e-01 2.22500324e-01 -1.34712234e-01
-7.40009964e-01 -9.40988064e-01 -5.22482872e-01 2.02358767e-01
2.52405494e-01 -6.57418907e-01 1.83062530e+00 -8.39148164e-01
-1.16102636e+00 1.13409257e+00 -9.87748429e-02 -2.05743715e-01
4.31941390e-01 -4.89203095e-01 -1.22517534e-01 -1.02952503e-01
3.87520880e-01 1.01156878e+00 5.25179744e-01 -1.22933221e+00
-1.07482159e+00 -3.54227245e-01 9.07234251e-02 3.12574029e-01
-3.58092576e-01 2.55432963e-01 -2.10519671e-01 -4.62313980e-01
-3.92649248e-02 -9.28111434e-01 -1.28278732e-01 -1.52286768e-01
-4.88082796e-01 -8.14530611e-01 6.68760836e-01 -2.63599336e-01
1.04619718e+00 -1.97970164e+00 1.41548023e-01 -5.08245051e-01
4.45056856e-01 5.87057648e-03 -3.13626319e-01 4.63013291e-01
-1.25821769e-01 3.40583384e-01 -2.99270041e-02 -1.03891075e+00
2.76730746e-01 -3.32337350e-01 -4.26236331e-01 4.08017099e-01
3.77895743e-01 1.12335193e+00 -8.64699304e-01 -5.51523209e-01
4.58390005e-02 6.07620120e-01 -2.52186179e-01 1.28194407e-01
-5.74974597e-01 3.09630096e-01 -3.31957161e-01 7.45714486e-01
3.33465517e-01 -8.98048043e-01 4.22804430e-02 -8.80646557e-02
-6.08714856e-02 4.90014255e-01 -6.63128436e-01 1.82184923e+00
-8.13144565e-01 3.93395364e-01 -1.54712006e-01 -5.06800056e-01
4.87087846e-01 5.82190990e-01 3.80510181e-01 -5.44345975e-01
4.57844645e-01 3.31313729e-01 -5.53970039e-01 -8.40386972e-02
2.68727392e-01 -1.70613572e-01 -4.50283229e-01 9.98952091e-01
2.30896369e-01 4.73559201e-01 2.08268277e-02 1.80539280e-01
1.12311900e+00 1.76658824e-01 2.40053475e-01 -2.38616318e-01
2.40457743e-01 -1.62481010e-01 4.59287077e-01 6.76321745e-01
-1.82068888e-02 6.15208864e-01 4.09436464e-01 -6.40975058e-01
-8.70970309e-01 -6.92966342e-01 -3.04481894e-01 1.92294586e+00
-4.00940984e-01 -3.86033893e-01 -6.62592173e-01 -9.57817912e-01
-1.17471209e-02 8.61877263e-01 -8.46901536e-01 -1.07668042e-01
-4.68218148e-01 -6.58670306e-01 6.31036222e-01 5.55517495e-01
3.33812952e-01 -1.22179294e+00 -6.45208001e-01 2.55028576e-01
-2.50963986e-01 -1.04035878e+00 -4.91377532e-01 6.15476191e-01
-6.86881065e-01 -8.41934681e-01 -5.51838100e-01 -8.97139788e-01
4.21536565e-01 7.35778660e-02 1.46820438e+00 5.23355454e-02
-1.74747378e-01 9.89455804e-02 -4.75770414e-01 -2.52228707e-01
-4.55850124e-01 7.53476799e-01 -2.58862734e-01 1.02418913e-02
5.70946276e-01 -5.13945401e-01 -5.54086387e-01 1.55559167e-01
-5.49998462e-01 4.49574381e-01 5.29110909e-01 7.75570869e-01
7.29790509e-01 -3.32806766e-01 9.17998850e-01 -1.44842672e+00
4.68702018e-01 -6.49094701e-01 -3.58547926e-01 4.40677822e-01
-8.46600354e-01 6.76089004e-02 8.00677419e-01 -3.90108049e-01
-7.66150475e-01 1.75833285e-01 -2.10396141e-01 -3.87519240e-01
-2.85161406e-01 3.41377735e-01 1.59862861e-02 2.59843528e-01
5.57587802e-01 -1.67611837e-01 -3.91712248e-01 -7.67551184e-01
8.66930008e-01 7.87627339e-01 2.73597658e-01 -5.06158888e-01
2.10448086e-01 1.45258635e-01 -2.62915701e-01 8.98453966e-02
-1.78774059e+00 -4.44561541e-01 -7.93828189e-01 -2.43624777e-01
1.00027812e+00 -1.06053936e+00 -7.20908940e-01 4.56058055e-01
-1.05846918e+00 -4.22752917e-01 -4.08872180e-02 5.42213507e-02
-5.39727747e-01 -2.44583577e-01 -1.18680871e+00 -6.17549837e-01
-6.35378063e-01 -1.06351364e+00 1.42933762e+00 1.55918911e-01
-3.01744848e-01 -9.86288249e-01 -7.15139583e-02 4.44686145e-01
4.81041163e-01 1.76223129e-01 1.22933114e+00 -7.18621671e-01
-4.01163042e-01 -1.46030635e-01 -4.47884113e-01 -9.83941033e-02
-5.51086627e-02 -3.58953387e-01 -1.45786583e+00 -6.10120595e-01
-2.32933119e-01 -1.07332778e+00 1.28113031e+00 1.78247020e-01
1.24941027e+00 -2.87299424e-01 -4.98908460e-01 5.90404451e-01
1.50678265e+00 -3.96861523e-01 2.21730303e-02 3.21933240e-01
9.17993903e-01 5.15533984e-01 4.23002154e-01 2.09602162e-01
5.91862440e-01 6.68869317e-01 5.81709921e-01 -2.17050120e-01
-3.26292694e-01 -1.18254721e-01 1.24605507e-01 7.64165819e-01
4.04978454e-01 -2.65546381e-01 -9.88428414e-01 2.16952220e-01
-1.88342357e+00 -7.34243333e-01 -1.67253897e-01 1.90799165e+00
1.07614934e+00 4.14002389e-02 -1.65824722e-02 2.36755572e-02
7.90520489e-01 1.79244176e-01 -8.52101088e-01 -4.05065507e-01
1.25978142e-01 5.03883325e-02 3.25105339e-01 5.86410105e-01
-1.48489821e+00 1.12968075e+00 5.86524868e+00 7.42769957e-01
-1.09688926e+00 5.89855015e-01 9.68443215e-01 -5.36605120e-01
-9.99142826e-02 -1.58426329e-01 -1.24263871e+00 3.19897681e-01
1.39678812e+00 2.11783707e-01 2.63625026e-01 6.54755592e-01
-4.02372837e-01 4.16606247e-01 -1.30679977e+00 6.57932580e-01
-1.05619580e-02 -1.12945855e+00 2.51931194e-02 -1.41492197e-02
6.65735006e-01 4.18821037e-01 4.59610611e-01 6.96222126e-01
7.67125309e-01 -1.11972857e+00 7.75339484e-01 4.07304794e-01
1.34828424e+00 -6.58050120e-01 4.94194210e-01 3.42738122e-01
-1.24450660e+00 -3.71681690e-01 -3.95921648e-01 6.19498976e-02
-3.77670699e-03 5.87927461e-01 -6.83072805e-01 2.79458135e-01
7.44734466e-01 6.68035805e-01 -8.29022944e-01 4.83216494e-01
-9.48427022e-02 6.56082869e-01 -7.07032084e-02 4.58690897e-02
4.31075573e-01 3.62427235e-01 -8.89620557e-02 1.39792609e+00
1.03036277e-01 -1.03389479e-01 3.36582482e-01 7.29274988e-01
-5.81131577e-01 1.69992879e-01 -2.62704074e-01 1.63350657e-01
7.06776857e-01 1.57251573e+00 -7.00925946e-01 -3.77197057e-01
-6.85374200e-01 8.91055524e-01 1.14400196e+00 2.34376028e-01
-9.32918370e-01 -4.91849408e-02 4.24987346e-01 -4.55980040e-02
1.59336463e-01 3.83294940e-01 -4.70306247e-01 -9.06168461e-01
-2.50035465e-01 -8.22305202e-01 6.42900407e-01 -5.83697498e-01
-1.59646177e+00 8.58360231e-01 -2.98035502e-01 -1.05285311e+00
-1.93726212e-01 -5.96800268e-01 -1.76257864e-01 9.74253058e-01
-1.55911911e+00 -1.71042061e+00 -1.32105738e-01 2.33372092e-01
9.38332379e-01 -5.62640093e-02 1.27733767e+00 3.55343342e-01
-5.91860771e-01 8.68500710e-01 3.44152033e-01 2.32993998e-02
9.13982511e-01 -1.52534199e+00 6.56625211e-01 2.86960632e-01
2.96326578e-01 2.18865603e-01 2.40469307e-01 -2.86965638e-01
-7.95581639e-01 -1.57548368e+00 1.32272899e+00 -9.27461922e-01
5.45523643e-01 -6.67451322e-01 -8.77509713e-01 1.05734515e+00
3.65698814e-01 2.13804156e-01 1.09139562e+00 6.17825210e-01
-8.34326625e-01 9.81290415e-02 -8.14710140e-01 3.91891539e-01
1.02334654e+00 -6.68618441e-01 -2.12125972e-01 5.22152662e-01
9.38550234e-01 -2.06399649e-01 -1.02708530e+00 3.93226981e-01
4.84416872e-01 -8.42631042e-01 7.52380908e-01 -7.53029764e-01
7.83033192e-01 -4.67783585e-02 -1.90232888e-01 -1.21710837e+00
-9.42248344e-01 -8.13637301e-03 -2.10004792e-01 1.45153201e+00
8.84594738e-01 -6.07706547e-01 4.37041074e-01 3.74864876e-01
-6.38143793e-02 -1.32582355e+00 -8.99629891e-01 -2.39035428e-01
6.27643287e-01 -2.52249688e-01 6.78980112e-01 1.19383979e+00
-1.08917907e-01 1.00263619e+00 -3.79878819e-01 6.22407645e-02
7.14795113e-01 4.14811015e-01 1.73496902e-01 -1.49105453e+00
-2.23628700e-01 -6.62705779e-01 2.56004542e-01 -8.02424729e-01
4.68016684e-01 -1.49845779e+00 -7.08330497e-02 -1.77497029e+00
5.03592432e-01 -6.86252296e-01 -8.69073391e-01 9.85270917e-01
-4.68956620e-01 5.55573881e-01 2.17393324e-01 5.05784571e-01
-1.08898771e+00 3.85050476e-01 9.58785176e-01 -1.11435853e-01
3.13845307e-01 -1.50687307e-01 -1.04565072e+00 6.64270580e-01
9.13255453e-01 -9.17020380e-01 -2.36355990e-01 -8.41795027e-01
3.19109023e-01 1.48226805e-02 8.23196694e-02 -9.41655874e-01
2.68349975e-01 2.23206148e-01 5.39507747e-01 -6.28744245e-01
4.06264603e-01 -5.03419280e-01 -9.51643214e-02 2.36616671e-01
-1.12669790e+00 3.15271676e-01 -3.67201725e-03 3.96348983e-01
1.46635145e-01 -2.30982646e-01 8.76764536e-01 -3.40795159e-01
-4.09648299e-01 3.46567333e-01 -2.85251290e-01 1.69096082e-01
8.02918792e-01 3.61849606e-01 -6.95044577e-01 5.06431470e-03
-4.72783208e-01 4.02300686e-01 4.84944671e-01 5.07337689e-01
1.89859048e-01 -1.52322304e+00 -9.73835707e-01 -1.32111922e-01
2.81572640e-01 1.44258365e-01 2.07148075e-01 5.15123367e-01
-2.35584565e-02 4.12452847e-01 8.90897438e-02 -5.30717909e-01
-1.19195628e+00 7.32153535e-01 3.89202088e-01 -6.36125028e-01
-6.68002784e-01 1.21646833e+00 3.17116141e-01 -7.62045443e-01
2.87894785e-01 -8.15351754e-02 -2.96357632e-01 1.98558360e-01
6.71051562e-01 6.96756467e-02 2.01130640e-02 -5.97209096e-01
-3.00073653e-01 7.06959069e-01 -6.33886099e-01 9.56785772e-03
1.14004362e+00 -1.70675084e-01 -5.84550910e-02 9.06587958e-01
1.43389058e+00 -4.75316852e-01 -1.38125801e+00 -5.51973522e-01
2.94989884e-01 -5.21721505e-02 2.01069176e-01 -1.23178387e+00
-1.00236726e+00 8.89036715e-01 6.09225452e-01 1.35935590e-01
1.05108154e+00 1.98481545e-01 7.89384186e-01 2.78043538e-01
2.56074429e-01 -8.60286951e-01 1.20604530e-01 6.80333614e-01
7.50382304e-01 -1.31218708e+00 -2.54049182e-01 -1.02420121e-01
-5.61710536e-01 8.94512236e-01 8.12197089e-01 1.25923738e-01
6.07683480e-01 3.41117650e-01 4.74379212e-01 -3.00788254e-01
-1.51735604e+00 -1.93297818e-01 9.98085365e-02 -3.04282382e-02
1.04545450e+00 1.42065629e-01 -1.66201349e-02 4.63151664e-01
-5.20342924e-02 -2.34080911e-01 1.87423438e-01 7.94155657e-01
-5.21972597e-01 -1.04164577e+00 -4.81032580e-02 6.68698907e-01
-7.39212036e-01 -3.26284975e-01 -3.92048329e-01 3.24512780e-01
8.28978419e-02 9.13497031e-01 -1.08024143e-02 -3.42576712e-01
1.71397001e-01 5.33273518e-01 2.01332092e-01 -8.71161819e-01
-8.92689347e-01 -6.73997551e-02 2.19262857e-02 -4.97632861e-01
-1.70625284e-01 -5.53524196e-01 -1.33361804e+00 -2.46581852e-01
-3.61797571e-01 1.60913896e-02 4.45668459e-01 7.27507710e-01
5.03235936e-01 6.60350382e-01 5.06418347e-01 -7.52227664e-01
-7.61770248e-01 -1.33947682e+00 -4.20013428e-01 3.62183660e-01
2.97976106e-01 -5.60054719e-01 -1.40394971e-01 -1.01451077e-01] | [9.606673240661621, 4.422499179840088] |
b63fb2c6-f278-4840-b8e2-c432d9b6d446 | multi-view-data-classification-with-a-label | 2201.00714 | null | https://arxiv.org/abs/2201.00714v1 | https://arxiv.org/pdf/2201.00714v1.pdf | Multi-view Data Classification with a Label-driven Auto-weighted Strategy | Distinguishing the importance of views has proven to be quite helpful for semi-supervised multi-view learning models. However, existing strategies cannot take advantage of semi-supervised information, only distinguishing the importance of views from a data feature perspective, which is often influenced by low-quality views then leading to poor performance. In this paper, by establishing a link between labeled data and the importance of different views, we propose an auto-weighted strategy to evaluate the importance of views from a label perspective to avoid the negative impact of unimportant or low-quality views. Based on this strategy, we propose a transductive semi-supervised auto-weighted multi-view classification model. The initialization of the proposed model can be effectively determined by labeled data, which is practical. The model is decoupled into three small-scale sub-problems that can efficiently be optimized with a local convergence guarantee. The experimental results on classification tasks show that the proposed method achieves optimal or sub-optimal classification accuracy at the lowest computational cost compared to other related methods, and the weight change experiments show that our proposed strategy can distinguish view importance more accurately than other related strategies on multi-view datasets with low-quality views. | ['Qibin Zhao', 'Shengli Xie', 'Haonan Huang', 'Guoxu Zhou', 'Yuyuan Yu'] | 2022-01-03 | null | null | null | null | ['multi-view-learning'] | ['computer-vision'] | [-5.65535165e-02 -3.87836620e-02 -4.70472038e-01 -5.49305737e-01
-8.47359002e-01 -4.76353586e-01 4.19653654e-01 -3.45798433e-02
-1.64848894e-01 4.58434373e-01 4.78378683e-01 1.73761696e-01
-3.80746990e-01 -6.75580919e-01 -2.95573473e-01 -9.30014908e-01
4.32855010e-01 4.21227396e-01 1.03727669e-01 8.02674517e-02
2.71854699e-01 -9.55698043e-02 -1.63902104e+00 4.44615126e-01
8.70155454e-01 9.67801452e-01 8.85924175e-02 1.99920878e-01
-7.54277855e-02 8.59826565e-01 -1.68169156e-01 -3.51416707e-01
3.74104083e-01 -5.30433714e-01 -6.18289292e-01 5.49953997e-01
3.72546047e-01 -2.28459537e-01 4.11406815e-01 1.22368181e+00
4.54060465e-01 -1.66647136e-01 5.75934350e-01 -1.38537514e+00
-5.28169215e-01 3.38338077e-01 -8.81432772e-01 9.57773551e-02
4.03497182e-02 -4.34852481e-01 1.31207824e+00 -1.08258390e+00
4.48796600e-01 1.31390202e+00 3.51669371e-01 1.44210055e-01
-1.22230816e+00 -6.37505412e-01 6.98266089e-01 3.66602570e-01
-1.05460525e+00 -3.14051837e-01 1.12961638e+00 -4.13501292e-01
1.39097437e-01 1.48055911e-01 5.50317645e-01 7.75498629e-01
-1.83985326e-02 6.57496750e-01 1.77783453e+00 -1.96238831e-01
2.42599130e-01 5.74562728e-01 2.69486457e-01 8.47141206e-01
3.89716119e-01 -6.48323298e-02 -6.87784731e-01 -3.71632338e-01
3.24454755e-01 5.29094756e-01 -4.17502165e-01 -9.49592054e-01
-1.23073637e+00 8.67647290e-01 1.92093417e-01 1.70949578e-01
-2.45183751e-01 -4.26993281e-01 3.94047022e-01 4.94974256e-01
8.09647977e-01 -1.52754575e-01 -5.50404429e-01 5.84506392e-02
-5.62518060e-01 -2.94548750e-01 5.27890563e-01 7.53896415e-01
8.02480996e-01 -1.44083679e-01 1.55036703e-01 1.01752532e+00
5.33392072e-01 5.45531213e-01 3.62633884e-01 -8.05580020e-01
5.42584121e-01 9.67833161e-01 1.29510373e-01 -1.11825502e+00
-7.45266378e-02 -4.57790583e-01 -9.90874469e-01 2.98077673e-01
2.54810780e-01 2.79187150e-02 -4.33798760e-01 1.64229393e+00
6.69186056e-01 -1.35398090e-01 1.11356638e-01 9.24474835e-01
7.28677571e-01 4.50789183e-01 -3.56076300e-01 -7.01819718e-01
1.37246203e+00 -1.02750289e+00 -7.38053143e-01 -5.31187616e-02
4.96622562e-01 -6.62758112e-01 1.03042448e+00 5.71060479e-01
-8.46641839e-01 -5.65299332e-01 -1.19882119e+00 3.09238225e-01
-1.92405343e-01 2.49782696e-01 4.67384458e-01 6.14687800e-01
-6.81768656e-01 3.95532250e-01 -5.31661808e-01 -1.59077570e-01
3.76107365e-01 2.33406812e-01 -5.82057893e-01 -3.32736105e-01
-8.90546262e-01 7.63584912e-01 4.16141838e-01 6.59501031e-02
-6.59212768e-01 -5.29732049e-01 -6.65438056e-01 -1.23673715e-01
7.84417868e-01 -4.83626217e-01 7.19980955e-01 -1.32021224e+00
-1.16367066e+00 8.20891559e-01 -3.07727426e-01 1.00112922e-01
4.60843623e-01 -2.12102160e-01 -3.97339344e-01 2.07772523e-01
2.10157171e-01 1.24840461e-01 1.00102782e+00 -1.65413523e+00
-9.24129307e-01 -6.90270782e-01 3.41794610e-01 6.60900712e-01
-7.66014695e-01 -1.99943900e-01 -5.21665454e-01 -5.03467560e-01
3.44690084e-01 -8.62691343e-01 -4.20424938e-02 7.28937835e-02
-4.48297635e-02 -7.73919821e-02 8.19160640e-01 -3.90867323e-01
1.09044731e+00 -2.01162887e+00 3.83430600e-01 8.83779954e-03
5.13535261e-01 2.72193681e-02 9.97519419e-02 3.11782628e-01
-1.06943019e-01 -1.27384961e-01 -1.24090001e-01 -1.01398237e-01
-3.35481316e-01 2.98166126e-01 -6.93152994e-02 5.26426613e-01
-1.94048062e-01 4.34561640e-01 -1.02589262e+00 -7.26610839e-01
1.30066141e-01 2.59725988e-01 -3.15960586e-01 3.79508197e-01
4.03976440e-01 1.48241103e-01 -6.22661114e-01 5.65610170e-01
7.45173037e-01 -6.01362467e-01 3.93434674e-01 -4.99373108e-01
2.31156334e-01 -1.34990558e-01 -1.46391058e+00 1.33310449e+00
-6.38496161e-01 -5.70310280e-02 2.15090156e-01 -1.10525942e+00
8.05415392e-01 3.01320285e-01 5.53852618e-01 -4.24860358e-01
-5.37707880e-02 -4.36081318e-03 -1.82085395e-01 -2.63526857e-01
-2.53189020e-02 -4.41043258e-01 1.00586228e-01 8.54896963e-01
-3.72410030e-03 3.45226169e-01 -2.37046857e-03 2.75143534e-01
5.57222903e-01 8.95881355e-02 8.70099843e-01 -3.44714403e-01
8.29962373e-01 -1.89835981e-01 8.89827132e-01 2.59018272e-01
-1.95023909e-01 5.68102360e-01 4.73409057e-01 -4.89347130e-01
-6.13737583e-01 -8.57793212e-01 1.67621642e-01 1.10529089e+00
4.52467978e-01 -5.36201179e-01 -5.77187121e-01 -1.30668926e+00
-1.95001811e-01 3.36956888e-01 -7.42939591e-01 -1.80573776e-01
-1.23668641e-01 -9.00839508e-01 -2.71477997e-01 5.08327961e-01
4.17995125e-01 -5.46141565e-01 -3.89771312e-01 -8.77303258e-02
-3.38732302e-01 -9.79442537e-01 -5.03775597e-01 9.79897827e-02
-1.06365108e+00 -1.36531460e+00 -7.38525152e-01 -5.38909197e-01
1.19804585e+00 1.00830698e+00 1.10291731e+00 6.98043182e-02
4.70413536e-01 5.35232008e-01 -5.15758097e-01 -2.82352477e-01
-1.96402282e-01 -1.20981447e-01 1.41614392e-01 6.30459666e-01
2.85876900e-01 -5.88626206e-01 -6.15936935e-01 8.00114632e-01
-8.64684880e-01 3.59182447e-01 6.46283507e-01 1.09179366e+00
9.35715318e-01 1.69705078e-01 6.53067946e-01 -1.36111248e+00
2.05209166e-01 -3.97687107e-01 -4.00495201e-01 5.47920346e-01
-1.19810820e+00 2.39952996e-01 8.65036130e-01 -2.24062502e-01
-1.33776593e+00 -4.67663854e-02 3.62004489e-01 -5.31717539e-01
7.95000941e-02 3.73516709e-01 -6.09400511e-01 -4.24634404e-02
9.13081095e-02 3.65550399e-01 3.84191871e-02 -3.78769964e-01
4.03970033e-01 5.61199725e-01 -1.56561583e-01 -4.63729531e-01
8.45805407e-01 9.26108837e-01 -1.58011715e-03 -3.10395926e-01
-1.44187021e+00 -6.38411164e-01 -8.10866117e-01 -4.22060251e-01
5.22513866e-01 -1.12266147e+00 -3.66967022e-01 1.93425387e-01
-6.70613050e-01 4.84297872e-01 -7.05943927e-02 4.76386756e-01
-2.89595157e-01 7.14102268e-01 -1.82127133e-01 -7.02948213e-01
-3.03689092e-01 -1.08629882e+00 8.38829696e-01 8.81543010e-03
1.49314493e-01 -1.09528470e+00 9.08751227e-03 8.21850657e-01
5.58101609e-02 9.48956758e-02 8.21018040e-01 -6.69641376e-01
-3.81310284e-01 -8.08468014e-02 -3.34217668e-01 5.40587842e-01
5.25883257e-01 -2.97517866e-01 -1.06916142e+00 -5.00506043e-01
5.00935495e-01 -5.25134683e-01 8.34987700e-01 3.29644382e-02
9.91318285e-01 -3.42676908e-01 -9.77664888e-02 4.27729666e-01
1.63934171e+00 1.11857876e-02 1.17430806e-01 2.56679151e-02
9.67455506e-01 7.61722863e-01 1.05997586e+00 6.03211105e-01
6.61696732e-01 4.23271418e-01 6.11569643e-01 -2.04634041e-01
2.63046741e-01 -1.63844317e-01 4.04092312e-01 1.43937778e+00
-2.72752851e-01 4.03587101e-03 -4.78847712e-01 4.60899711e-01
-2.08899164e+00 -9.56747115e-01 -8.39686245e-02 2.46309948e+00
6.36202395e-01 2.49024332e-01 2.12391585e-01 2.14472115e-01
7.56090820e-01 5.81790268e-01 -5.72810948e-01 -7.65129030e-02
6.20657727e-02 -4.63930607e-01 2.05634490e-01 2.84434944e-01
-1.10643518e+00 3.53314817e-01 5.68756962e+00 8.88618946e-01
-7.74481714e-01 4.58730161e-01 7.72339642e-01 -3.25036258e-01
-5.99016368e-01 1.08058766e-01 -7.12646484e-01 5.01276493e-01
4.04528350e-01 -2.44460121e-01 2.81299949e-01 1.05547547e+00
1.86114103e-01 -2.90488247e-02 -1.14670551e+00 1.15954709e+00
5.81060350e-01 -8.05388510e-01 2.55125463e-01 2.10582480e-01
7.18471944e-01 -3.76482159e-01 -1.06720775e-01 5.72700016e-02
2.87769526e-01 -3.64362270e-01 4.93610531e-01 3.80777895e-01
4.70434874e-01 -9.82321501e-01 7.97434688e-01 5.80538750e-01
-1.46979034e+00 -1.44459650e-01 -4.59948808e-01 -2.27557719e-02
-1.24595605e-01 8.61076593e-01 -1.46288559e-01 9.88599300e-01
6.07993007e-01 9.75508451e-01 -5.72318017e-01 2.57978290e-01
-8.06409791e-02 4.09092993e-01 9.12478194e-02 1.67510837e-01
-1.89045053e-02 -4.99030679e-01 2.40236506e-01 6.40006483e-01
1.65024027e-01 1.79902673e-01 3.64602268e-01 4.23764050e-01
4.08362597e-02 2.51864225e-01 -6.40160501e-01 2.57904053e-01
1.94952041e-01 1.54841220e+00 -7.77250648e-01 -4.95463610e-01
-8.72290671e-01 8.54192674e-01 5.00201344e-01 4.59113345e-03
-8.04668367e-01 1.49796173e-01 3.08501303e-01 -3.13150580e-03
3.74358058e-01 1.60080835e-01 -2.02227220e-01 -1.47623193e+00
2.97656596e-01 -1.04545641e+00 7.87403166e-01 -6.12572789e-01
-1.49816716e+00 4.98943716e-01 4.08026576e-02 -1.90770376e+00
-8.02497729e-04 -4.03924167e-01 -3.62793893e-01 6.10290766e-01
-1.68077946e+00 -1.21851313e+00 -4.25872445e-01 5.41747510e-01
5.72857380e-01 -1.55491114e-01 6.65646970e-01 1.17454156e-01
-4.81713623e-01 4.10922647e-01 1.52497500e-01 -1.06975675e-01
9.90412474e-01 -1.41848409e+00 -2.91395575e-01 8.34430218e-01
1.48671553e-01 4.26517248e-01 2.10299537e-01 -3.54029506e-01
-1.38883686e+00 -9.80951965e-01 6.61931336e-01 -3.32888544e-01
3.97426277e-01 -1.17829435e-01 -8.84551704e-01 2.73570508e-01
2.45795831e-01 2.90303051e-01 1.17042673e+00 3.09970230e-01
-5.15495420e-01 -5.79743087e-01 -1.02868664e+00 2.61556834e-01
1.07505715e+00 -4.17811573e-01 -6.28701627e-01 2.30235294e-01
4.26680595e-01 1.82817075e-02 -9.14557636e-01 4.26206291e-01
6.75479412e-01 -1.37960625e+00 8.88523996e-01 -5.74599445e-01
6.46240950e-01 -3.81626844e-01 -1.85078636e-01 -1.48045146e+00
-5.26606679e-01 6.47212937e-02 -2.39586309e-01 1.42117262e+00
1.98901191e-01 -7.98133969e-01 6.18392169e-01 3.36686730e-01
3.87088746e-01 -9.46912885e-01 -7.94152319e-01 -6.43300951e-01
-4.69679236e-01 -3.79842110e-02 3.08989495e-01 1.15959191e+00
-1.08408853e-01 5.78897834e-01 -6.79728746e-01 3.08337003e-01
1.00196481e+00 8.66421044e-01 6.69144452e-01 -1.36218774e+00
-5.01325905e-01 -9.64391530e-02 -2.28957951e-01 -7.66572773e-01
8.70836899e-03 -7.29259789e-01 -2.08379030e-01 -1.53082848e+00
8.45516264e-01 -2.23582253e-01 -8.96517456e-01 2.88469940e-01
-6.62883997e-01 -1.58587098e-02 1.37761340e-01 4.26409572e-01
-9.59375024e-01 6.32937789e-01 1.31012261e+00 -9.62971970e-02
1.90846592e-01 1.69429835e-02 -1.02347159e+00 1.20100534e+00
4.98176336e-01 -6.07767761e-01 -8.91867399e-01 -2.60360926e-01
3.50895762e-01 2.75857896e-01 6.46448508e-02 -7.28099167e-01
-4.77410853e-02 -3.77179801e-01 2.88027644e-01 -7.06468463e-01
2.33632520e-01 -1.25077486e+00 6.03089407e-02 3.04932415e-01
-2.73731083e-01 2.15662643e-01 -5.19519269e-01 1.01144195e+00
-4.57325369e-01 -9.14661586e-02 8.12695503e-01 -4.28833723e-01
-5.95892310e-01 3.21214259e-01 9.58639234e-02 2.55614161e-01
1.07954133e+00 -2.13401541e-01 -1.04238346e-01 -4.42582011e-01
-5.51185429e-01 3.41463834e-01 5.11917830e-01 4.82909381e-01
6.39761090e-01 -1.72259927e+00 -5.36555529e-01 6.17308654e-02
5.63615859e-01 -2.39381447e-01 4.23953176e-01 8.83050203e-01
1.36805296e-01 9.99644920e-02 -2.03789279e-01 -6.10864699e-01
-1.66498280e+00 8.68470967e-01 9.68441926e-03 -5.83040774e-01
-2.33026549e-01 3.63993108e-01 5.63778758e-01 -4.23920840e-01
3.83527577e-02 2.53348029e-04 -6.21965826e-01 5.94689071e-01
5.56120038e-01 5.66060007e-01 -3.98053825e-02 -7.25337923e-01
-3.36725444e-01 8.90197635e-01 -2.24953681e-01 5.74591048e-02
1.34413040e+00 -4.97754306e-01 -7.06335828e-02 9.27313268e-01
1.18524837e+00 5.22831939e-02 -1.15555871e+00 -6.06025577e-01
-3.34547400e-01 -7.17319429e-01 1.94667846e-01 -6.29130840e-01
-1.45667529e+00 8.91738355e-01 6.63085461e-01 2.36902490e-01
1.38565195e+00 -1.81051299e-01 5.00038683e-01 3.21580768e-01
6.06922925e-01 -1.12900245e+00 1.94689006e-01 -8.65143761e-02
7.41036236e-01 -1.68343353e+00 4.07672226e-01 -6.13010347e-01
-1.06722069e+00 8.97075355e-01 8.04025292e-01 1.32366791e-01
8.01910818e-01 -1.96808521e-02 2.03171983e-01 -2.87285507e-01
-9.66996908e-01 -1.82581797e-01 3.85850102e-01 2.97273725e-01
2.43077457e-01 1.13852195e-01 -3.09424758e-01 5.32814443e-01
4.66751993e-01 -1.28243178e-01 3.37398529e-01 8.44918311e-01
-4.10011947e-01 -1.10879815e+00 -3.98672789e-01 6.22159004e-01
-3.69927108e-01 1.14103973e-01 -4.12463516e-01 5.11975229e-01
1.79738045e-01 9.59478676e-01 -3.31380099e-01 -5.18510103e-01
3.28343928e-01 6.34128973e-03 2.01377973e-01 -5.76645553e-01
-3.32304537e-01 2.43831098e-01 8.21735263e-02 -5.51270723e-01
-1.05404711e+00 -5.92852175e-01 -7.34943390e-01 -6.36089221e-02
-5.99548340e-01 1.80333823e-01 4.76065651e-02 1.02484667e+00
4.72281933e-01 3.64027768e-01 1.32270992e+00 -4.51034278e-01
-7.90528536e-01 -6.34133756e-01 -7.48322725e-01 6.69138014e-01
4.03779047e-03 -7.81722903e-01 -5.48794985e-01 1.02177605e-01] | [8.548012733459473, 4.498826503753662] |
a55cb40c-7e9b-4431-9ef4-2bddca1b29c0 | few-shot-domain-adaptation-for-in-situ | 2007.15422 | null | https://arxiv.org/abs/2007.15422v1 | https://arxiv.org/pdf/2007.15422v1.pdf | Few shot domain adaptation for in situ macromolecule structural classification in cryo-electron tomograms | Motivation: Cryo-Electron Tomography (cryo-ET) visualizes structure and spatial organization of macromolecules and their interactions with other subcellular components inside single cells in the close-to-native state at sub-molecular resolution. Such information is critical for the accurate understanding of cellular processes. However, subtomogram classification remains one of the major challenges for the systematic recognition and recovery of the macromolecule structures in cryo-ET because of imaging limits and data quantity. Recently, deep learning has significantly improved the throughput and accuracy of large-scale subtomogram classification. However often it is difficult to get enough high-quality annotated subtomogram data for supervised training due to the enormous expense of labeling. To tackle this problem, it is beneficial to utilize another already annotated dataset to assist the training process. However, due to the discrepancy of image intensity distribution between source domain and target domain, the model trained on subtomograms in source domainmay perform poorly in predicting subtomogram classes in the target domain. Results: In this paper, we adapt a few shot domain adaptation method for deep learning based cross-domain subtomogram classification. The essential idea of our method consists of two parts: 1) take full advantage of the distribution of plentiful unlabeled target domain data, and 2) exploit the correlation between the whole source domain dataset and few labeled target domain data. Experiments conducted on simulated and real datasets show that our method achieves significant improvement on cross domain subtomogram classification compared with baseline methods. | ['Xiangrui Zeng', 'Ge Yang', 'Rui Jiang', 'Ran Li', 'Jie Jin', 'Hongyi Wang', 'Min Xu', 'Liangyong Yu'] | 2020-07-30 | null | null | null | null | ['electron-tomography'] | ['medical'] | [ 1.24460034e-01 -4.63609874e-01 -9.28890407e-02 -2.81102449e-01
-1.03052270e+00 -6.21068656e-01 4.36370194e-01 1.26652867e-01
-3.11897099e-01 1.35069299e+00 -2.03143746e-01 -4.74329442e-02
1.26791820e-01 -6.35897100e-01 -6.35435700e-01 -1.30860686e+00
2.79885769e-01 7.48008132e-01 2.93907911e-01 5.46615571e-02
3.04061413e-01 7.13127196e-01 -1.04657912e+00 2.86958784e-01
9.38876629e-01 9.14424479e-01 7.42651224e-01 6.50540471e-01
-4.70260680e-01 6.42638206e-01 -4.47363675e-01 1.46946669e-01
5.43791689e-02 -6.23860478e-01 -9.18172359e-01 1.49499744e-01
1.20660976e-01 6.92443848e-02 -6.38250560e-02 1.13105667e+00
6.89080298e-01 5.47153316e-02 1.00922811e+00 -5.93904436e-01
-4.27989066e-01 -3.93020719e-01 -5.79392314e-01 6.20385766e-01
7.10237548e-02 4.02346887e-02 5.97088158e-01 -9.64208484e-01
1.05224979e+00 6.46683693e-01 3.12823415e-01 5.09790480e-01
-1.58169711e+00 -7.11089075e-01 -1.59844443e-01 3.29113662e-01
-9.61994588e-01 -4.97781634e-01 7.58116841e-01 -5.38927853e-01
8.25978279e-01 -1.03291027e-01 5.65555990e-01 1.12273538e+00
7.64488935e-01 3.75616848e-01 1.49831164e+00 -2.98645616e-01
5.12568891e-01 1.45137042e-01 1.84335858e-01 5.69946945e-01
-1.79324642e-01 6.34088181e-03 -1.72281981e-01 -1.57316163e-01
8.35196435e-01 1.21833317e-01 -4.59291935e-01 -7.82708585e-01
-1.00909257e+00 4.82080400e-01 3.00451577e-01 6.09929442e-01
-2.05305725e-01 -4.71957952e-01 6.71346784e-01 4.07104939e-01
4.95561361e-01 6.30062461e-01 -5.36909044e-01 -7.08500445e-02
-8.16375732e-01 -3.40071283e-02 5.86975992e-01 5.88328719e-01
7.77146041e-01 -1.72434077e-01 5.26052237e-01 1.06123269e+00
-2.40072668e-01 2.96036392e-01 8.45638037e-01 -6.77461326e-01
1.95304319e-01 6.73216879e-01 2.33944103e-01 -5.29718339e-01
-5.49102664e-01 -4.50379476e-02 -1.02659345e+00 2.96864569e-01
8.19434524e-01 3.15246843e-02 -1.20437813e+00 1.45306373e+00
4.15987611e-01 2.62014382e-02 1.17601618e-01 1.20508707e+00
7.04979062e-01 8.19248915e-01 7.54570812e-02 -3.83266330e-01
1.27165675e+00 -4.99465376e-01 -5.38844526e-01 -5.71894087e-02
6.38608634e-01 -7.37753808e-01 1.06737030e+00 1.14696160e-01
-5.58531284e-01 -5.83279014e-01 -1.10278451e+00 5.13605699e-02
-5.91525555e-01 3.91136147e-02 4.54943776e-01 3.20678860e-01
-4.27073896e-01 5.99720299e-01 -8.90378654e-01 -6.47068560e-01
5.43361843e-01 3.15184712e-01 -8.08280647e-01 -2.47085437e-01
-9.60559249e-01 7.29171813e-01 3.96302044e-01 -4.76976424e-01
-1.04075527e+00 -7.25151598e-01 -5.79442859e-01 1.11544408e-01
1.20943189e-01 -3.30125928e-01 9.25327957e-01 -8.38503778e-01
-1.33950698e+00 1.09151244e+00 -2.96805918e-01 -1.36045069e-01
1.59252018e-01 3.16771746e-01 -2.75617331e-01 4.20562804e-01
1.80001765e-01 2.79607475e-01 4.67165202e-01 -1.20705402e+00
-3.46803486e-01 -7.34352708e-01 -3.09988946e-01 6.92810640e-02
-5.52736819e-02 -1.88803256e-01 -4.43853699e-02 -2.03216866e-01
2.15598658e-01 -6.28270328e-01 -1.17601983e-01 -1.14332221e-01
-1.42200097e-01 -3.83765660e-02 1.04637742e+00 -5.41573048e-01
3.88963103e-01 -2.04188585e+00 1.75027609e-01 -2.61655539e-01
6.10065460e-01 3.80333364e-01 2.49820873e-02 4.81473893e-01
-4.45947081e-01 -3.13575625e-01 -1.09842069e-01 9.31078643e-02
-5.01014113e-01 4.63154987e-02 -1.75773010e-01 7.89625704e-01
5.23140803e-02 8.31100225e-01 -8.77179444e-01 -6.58478498e-01
4.43740427e-01 2.19922364e-01 2.63573825e-01 4.46261793e-01
-4.29490618e-02 8.57840478e-01 -4.27899987e-01 9.40301538e-01
8.83215010e-01 -6.69011652e-01 6.12446725e-01 -3.12161982e-01
6.20842688e-02 2.04788983e-01 -6.73281848e-01 1.55205727e+00
-2.11579666e-01 5.89095592e-01 2.63058126e-01 -1.58007836e+00
9.50672209e-01 3.60922307e-01 7.30009615e-01 -8.89793038e-01
9.78122726e-02 3.74064237e-01 2.15307511e-02 -4.37638432e-01
-7.74799958e-02 -7.22268879e-01 6.40677363e-02 4.37645882e-01
2.66216785e-01 9.23800319e-02 1.06894575e-01 8.11226815e-02
9.80313420e-01 8.56384039e-02 6.08516991e-01 -4.32052255e-01
5.64331114e-01 1.88161165e-01 7.04995811e-01 3.75956595e-01
-7.16186225e-01 6.19511724e-01 3.84351283e-01 -8.98497522e-01
-1.60759413e+00 -1.07374084e+00 -4.26266909e-01 8.91445041e-01
2.56584406e-01 2.19774470e-01 -6.67643487e-01 -7.90147245e-01
-2.63912916e-01 -2.34337285e-01 -6.79513693e-01 1.91928893e-02
-5.00786543e-01 -1.21559298e+00 3.37653607e-01 6.78328037e-01
4.70434040e-01 -8.84530008e-01 -5.57154655e-01 2.99341530e-01
-1.84629828e-01 -1.11165202e+00 -1.86432630e-01 7.35323787e-01
-9.04420376e-01 -1.41336513e+00 -9.84679580e-01 -9.21099603e-01
7.23426461e-01 4.19468224e-01 8.74845266e-01 -2.14494422e-01
-5.23356616e-01 -2.07411408e-01 -1.77270070e-01 -4.32273559e-02
-2.55634785e-01 5.37232421e-02 1.39588460e-01 -2.77029812e-01
6.44653022e-01 -6.72189593e-01 -4.63775545e-01 4.69914943e-01
-5.48317909e-01 -5.17202169e-02 5.38775623e-01 1.24619675e+00
9.11547363e-01 5.22103943e-02 8.26225460e-01 -1.04382098e+00
3.85126114e-01 -3.05097699e-01 -6.68037415e-01 2.11247936e-01
-4.71294016e-01 -8.21558833e-02 1.19693029e+00 -4.09030318e-01
-9.96639609e-01 1.51140839e-01 -4.28270735e-02 -3.68303984e-01
-4.80055690e-01 2.05263734e-01 -4.04782325e-01 -1.85773417e-01
6.97035372e-01 7.04248607e-01 2.15191916e-01 -3.43798816e-01
-2.29668885e-01 6.41842604e-01 4.11462307e-01 -7.31799483e-01
1.99605972e-01 6.83074117e-01 2.53430665e-01 -8.69175494e-01
-6.61586404e-01 -6.99089408e-01 -8.05376232e-01 -1.45792693e-01
7.94161737e-01 -6.88873470e-01 -6.31587565e-01 3.84368837e-01
-8.88112009e-01 -2.96934515e-01 3.84791680e-02 4.27955210e-01
-6.31772041e-01 7.34253466e-01 -7.79945731e-01 -5.06314218e-01
-2.21810222e-01 -1.15688920e+00 9.97578859e-01 1.88705266e-01
1.59555182e-01 -1.10903859e+00 3.28414232e-01 2.97031999e-01
1.17592420e-02 1.59216359e-01 1.32567656e+00 -6.97910070e-01
-5.39593279e-01 -1.82730392e-01 -5.45158386e-01 1.68592669e-02
3.93680543e-01 -3.88835251e-01 -9.76268172e-01 -4.11754787e-01
2.85197377e-01 -8.02218616e-01 7.18197227e-01 5.66871285e-01
1.01898837e+00 3.85790259e-01 -6.27449691e-01 5.68228781e-01
1.50204420e+00 5.07639766e-01 6.52583539e-01 3.11627030e-01
4.54203010e-01 6.70179069e-01 7.44507670e-01 1.89751327e-01
-1.11263342e-01 6.11446023e-01 1.83872506e-01 -5.74629158e-02
-8.77307579e-02 1.04276143e-01 -2.39193551e-02 6.36990130e-01
3.24520655e-02 -9.65346172e-02 -1.00287616e+00 5.85455835e-01
-1.57818031e+00 -9.94097829e-01 1.10892251e-01 2.07632184e+00
7.60733664e-01 -5.27776219e-02 1.46725908e-01 -1.11066792e-02
8.04473519e-01 -6.18031770e-02 -9.91831958e-01 -1.28723532e-01
-9.94137377e-02 -1.25856951e-01 2.76889116e-01 1.54928967e-01
-1.16038048e+00 7.20200241e-01 6.20573664e+00 8.38377595e-01
-1.32258236e+00 1.14619397e-01 8.23927402e-01 2.07006961e-01
3.79604965e-01 -2.18803227e-01 -8.73453617e-01 6.91316187e-01
9.46867168e-01 -1.00544170e-01 2.78547257e-01 8.07657003e-01
1.23358034e-01 -2.24735424e-01 -1.12799501e+00 1.00630188e+00
-1.92393526e-01 -1.63141191e+00 -3.89719337e-01 2.32420310e-01
6.46389365e-01 1.32184789e-01 -3.23688537e-01 2.54974484e-01
1.23980120e-01 -8.19560647e-01 -9.83151942e-02 2.67014652e-01
9.73466933e-01 -7.13485837e-01 9.27692473e-01 6.37367308e-01
-1.18843663e+00 1.82285011e-01 -9.09133852e-01 1.37037262e-01
6.42533898e-02 8.00390780e-01 -1.02981937e+00 5.49419522e-01
5.97888291e-01 7.20732629e-01 -9.35944468e-02 8.09748650e-01
3.72693688e-01 3.22058529e-01 -1.88766234e-02 4.27507795e-02
1.93564612e-02 -4.57103640e-01 2.65469968e-01 1.09127569e+00
7.79625252e-02 1.29031509e-01 3.80185902e-01 7.12563455e-01
-2.92619884e-01 7.77151436e-02 -8.08982134e-01 -3.76527905e-01
4.16551590e-01 1.30499470e+00 -9.45998609e-01 -4.48236495e-01
-3.82273495e-01 8.92758369e-01 6.54817879e-01 2.96790093e-01
-6.05921149e-01 -5.56378961e-01 7.06941485e-01 2.22265303e-01
2.20407173e-01 -1.78668126e-01 -4.48863447e-01 -1.23552108e+00
-3.73964697e-01 -6.56914651e-01 2.72641629e-01 -6.26813114e-01
-1.61912060e+00 5.02071857e-01 -3.83022994e-01 -1.10061574e+00
-3.35818864e-02 -8.54735434e-01 -3.68382663e-01 9.32088673e-01
-1.51693320e+00 -9.33250248e-01 -3.90713006e-01 3.58464122e-01
7.76496530e-01 -2.06853300e-01 9.47345912e-01 3.39509070e-01
-4.54248786e-01 1.63863793e-01 7.50919402e-01 9.55205932e-02
9.32705164e-01 -1.44990087e+00 5.39733581e-02 5.07347465e-01
-2.49847621e-01 6.51498437e-01 6.25305176e-01 -6.75586462e-01
-1.03360283e+00 -1.02092874e+00 5.76469064e-01 -5.09338677e-01
4.64473277e-01 -4.83702958e-01 -1.25892031e+00 4.58507150e-01
-2.48593781e-02 3.64617616e-01 9.26675081e-01 -1.17068719e-02
-1.59569904e-01 -4.92730141e-02 -1.35132611e+00 2.33010784e-01
5.01957059e-01 -7.88561821e-01 -6.37737036e-01 4.68779951e-01
2.08116412e-01 -3.40136141e-01 -1.06870079e+00 2.03079298e-01
6.90007627e-01 -1.08201742e+00 8.59059095e-01 -6.74890637e-01
4.84368622e-01 -2.40952045e-01 -1.15719095e-01 -1.32869709e+00
-3.24372977e-01 8.58472940e-03 -3.94996069e-03 7.89892733e-01
9.78716998e-04 -5.23748040e-01 1.26283884e+00 4.27503824e-01
-2.17409462e-01 -8.37840855e-01 -9.89039004e-01 -8.77548993e-01
3.44252527e-01 3.43370140e-01 3.59546244e-01 1.15038323e+00
2.18865782e-01 5.57906270e-01 -2.45511636e-01 -1.04707442e-01
6.79470658e-01 7.19214261e-01 7.51571655e-01 -1.37443399e+00
-1.81255609e-01 1.30252853e-01 -5.49876153e-01 -1.00043976e+00
2.85880774e-01 -7.15099335e-01 1.26281127e-01 -1.29889262e+00
7.99663126e-01 -2.45141625e-01 -4.55745757e-01 1.12689845e-01
5.15838340e-02 2.96447635e-01 -1.93594038e-01 4.39074129e-01
-6.34978235e-01 5.28698921e-01 1.54558301e+00 -7.02645704e-02
6.38420805e-02 -2.75577754e-01 -3.42860311e-01 4.52189505e-01
7.31212258e-01 -4.59061384e-01 -3.66380244e-01 1.58898477e-02
-4.54945624e-01 2.61786163e-01 3.02486241e-01 -1.15661061e+00
9.48851183e-02 -4.09467399e-01 8.46086144e-01 -5.89729071e-01
5.36963761e-01 -8.03089321e-01 2.05117706e-02 4.40843344e-01
-1.41138047e-01 -3.36391568e-01 -9.18624252e-02 8.57429862e-01
-1.93794787e-01 -1.74831480e-01 1.44385409e+00 -5.33895910e-01
-6.13682926e-01 2.34699801e-01 -6.37687027e-01 -4.34233993e-02
1.12832332e+00 -4.92195547e-01 -5.58527052e-01 -7.82514736e-02
-8.80536437e-01 -1.22988440e-01 1.03514469e+00 -1.77191004e-01
6.09170914e-01 -1.02280807e+00 -6.89703673e-02 3.17369401e-01
2.32512131e-01 -3.03362280e-01 4.04943764e-01 8.95361423e-01
-4.18892145e-01 7.74435639e-01 -7.28861094e-01 -6.92821741e-01
-1.23565495e+00 7.33850241e-01 5.74270606e-01 -4.85594183e-01
-6.21863365e-01 5.33582509e-01 7.20000029e-01 -6.20647073e-01
-2.97899365e-01 -1.03300810e-01 -1.87267601e-01 -2.45491818e-01
6.34787321e-01 1.71041086e-01 1.16005979e-01 -5.00103474e-01
-3.49717557e-01 5.95361233e-01 -6.50002122e-01 4.58303243e-01
1.33472729e+00 -3.41445923e-01 -1.12924747e-01 4.20538425e-01
1.25303853e+00 -4.84004885e-01 -1.40945387e+00 -7.23652095e-02
-1.14196852e-01 -5.12183785e-01 -6.88771904e-02 -7.13240802e-01
-7.16143727e-01 1.20730639e+00 7.46103227e-01 1.59260314e-02
1.08679712e+00 1.02741174e-01 9.42198515e-01 5.18710136e-01
6.91270709e-01 -1.14506578e+00 2.68561542e-01 4.27393824e-01
2.13751197e-01 -1.53058887e+00 -2.82790419e-03 -3.62329185e-01
-4.63252336e-01 1.18991911e+00 8.58573020e-01 -2.38475353e-02
3.76650900e-01 2.67840475e-01 1.68684393e-01 -4.22511309e-01
-8.25347543e-01 1.76296145e-01 -3.72443020e-01 9.26042199e-01
5.35714209e-01 -1.74639046e-01 -2.48990595e-01 4.11507189e-01
4.80498165e-01 1.80702642e-01 4.78973806e-01 9.90611970e-01
-8.22718799e-01 -1.21876764e+00 -2.14892119e-01 6.63776577e-01
-6.28605783e-01 2.84193277e-01 -3.53353322e-01 5.69065034e-01
-1.78871021e-01 5.56687474e-01 -1.57282725e-01 -2.03335851e-01
-7.76485950e-02 2.43907228e-01 5.83682835e-01 -6.57569170e-01
-1.99897904e-02 2.65634328e-01 -1.78364292e-01 -3.78753573e-01
-4.22805130e-01 -3.81202161e-01 -1.57557571e+00 -2.71816909e-01
-3.26504320e-01 3.69448781e-01 4.55691576e-01 1.10469353e+00
4.51902270e-01 3.09374094e-01 6.59422636e-01 -9.57791448e-01
-3.74321759e-01 -9.00043249e-01 -1.09242463e+00 5.18888950e-01
3.75519723e-01 -8.64700496e-01 -7.02361986e-02 3.26390475e-01] | [13.474211692810059, -3.1197354793548584] |
2fd59b15-7567-417b-a4ac-e204aa0bfe86 | novel-class-discovery-without-forgetting | 2207.10659 | null | https://arxiv.org/abs/2207.10659v1 | https://arxiv.org/pdf/2207.10659v1.pdf | Novel Class Discovery without Forgetting | Humans possess an innate ability to identify and differentiate instances that they are not familiar with, by leveraging and adapting the knowledge that they have acquired so far. Importantly, they achieve this without deteriorating the performance on their earlier learning. Inspired by this, we identify and formulate a new, pragmatic problem setting of NCDwF: Novel Class Discovery without Forgetting, which tasks a machine learning model to incrementally discover novel categories of instances from unlabeled data, while maintaining its performance on the previously seen categories. We propose 1) a method to generate pseudo-latent representations which act as a proxy for (no longer available) labeled data, thereby alleviating forgetting, 2) a mutual-information based regularizer which enhances unsupervised discovery of novel classes, and 3) a simple Known Class Identifier which aids generalized inference when the testing data contains instances form both seen and unseen categories. We introduce experimental protocols based on CIFAR-10, CIFAR-100 and ImageNet-1000 to measure the trade-off between knowledge retention and novel class discovery. Our extensive evaluations reveal that existing models catastrophically forget previously seen categories while identifying novel categories, while our method is able to effectively balance between the competing objectives. We hope our work will attract further research into this newly identified pragmatic problem setting. | ['Vineeth N Balasubramanian', 'Kai Han', 'Piyush Rai', 'Soma Biswas', 'Gaurav Aggarwal', 'Sujoy Paul', 'K J Joseph'] | 2022-07-21 | null | null | null | null | ['novel-class-discovery', 'novel-class-discovery'] | ['computer-vision', 'methodology'] | [ 3.84617597e-01 1.50769427e-01 -7.98424110e-02 -5.09558022e-01
-3.59280050e-01 -6.02480173e-01 5.64247370e-01 2.70313352e-01
-6.95130229e-01 1.08667624e+00 1.21923663e-01 -6.11553118e-02
-2.58642852e-01 -7.99544513e-01 -6.47763491e-01 -6.72715068e-01
-2.29606867e-01 7.03578949e-01 2.87375569e-01 2.60209978e-01
1.48181975e-01 4.93348479e-01 -1.95748472e+00 3.32265049e-01
1.06774700e+00 6.32106781e-01 3.09798598e-01 3.78808200e-01
4.19818237e-02 8.73798966e-01 -3.82434934e-01 -3.33084643e-01
9.76952389e-02 -2.91495055e-01 -9.94215488e-01 9.65495110e-02
5.88509619e-01 -2.48465255e-01 -4.53822583e-01 8.79952788e-01
3.23508590e-01 4.51563150e-01 7.01040030e-01 -1.11140311e+00
-1.22675598e+00 8.68617237e-01 -2.19490081e-01 6.76171780e-01
-9.63575765e-02 3.54881883e-01 7.72499621e-01 -1.16148651e+00
5.84651768e-01 1.12582016e+00 7.34608591e-01 9.67782378e-01
-1.35633230e+00 -8.49846840e-01 5.64728498e-01 3.91134232e-01
-1.58612823e+00 -6.57169342e-01 5.05633473e-01 -3.94312710e-01
9.08922017e-01 9.75911096e-02 4.91275549e-01 1.31958723e+00
-2.30459850e-02 9.20840025e-01 1.11080253e+00 -3.35593343e-01
5.20895302e-01 4.75478530e-01 6.85520530e-01 4.75500941e-01
5.29275656e-01 2.85888314e-01 -8.73010635e-01 -1.52749106e-01
3.58383119e-01 4.10076946e-01 -3.77442181e-01 -5.36886275e-01
-1.15601182e+00 5.81633210e-01 4.68790799e-01 5.10420024e-01
-2.77158350e-01 -1.66008562e-01 -9.72175971e-02 6.06840968e-01
3.27532321e-01 6.17686212e-01 -8.77388299e-01 9.32138264e-02
-9.39275861e-01 -1.27112940e-01 5.35526812e-01 9.18930471e-01
1.09652996e+00 -1.39025599e-02 -2.94614583e-01 7.59459972e-01
1.09146029e-01 5.35258472e-01 1.11577582e+00 -5.11633396e-01
-5.28976880e-03 6.96748018e-01 7.24316388e-03 -5.85929871e-01
-3.01005870e-01 -9.36747134e-01 -6.34488702e-01 -1.71174884e-01
1.67216852e-01 2.05904528e-01 -1.23389304e+00 2.23656535e+00
1.12736531e-01 5.17502785e-01 3.67239356e-01 5.55355310e-01
4.05893207e-01 2.33949319e-01 2.11190805e-01 -4.02855277e-01
1.06123352e+00 -9.10216093e-01 -4.29132193e-01 -4.98307705e-01
4.30325836e-01 -2.01980799e-01 1.23693621e+00 2.76065499e-01
-8.26210082e-01 -6.69981241e-01 -1.05474925e+00 1.61594808e-01
-6.13662004e-01 -1.47904873e-01 8.14216197e-01 4.99027461e-01
-1.17244017e+00 7.86438823e-01 -6.57489181e-01 -3.85231256e-01
7.23710835e-01 2.52380311e-01 -1.90824524e-01 -2.71809012e-01
-1.14962697e+00 7.88071930e-01 7.56883025e-01 -7.45191947e-02
-1.41022289e+00 -1.04024816e+00 -4.55139041e-01 2.28411153e-01
3.64697337e-01 -8.65745187e-01 1.14940131e+00 -1.12935805e+00
-9.98216748e-01 9.80379403e-01 -2.28153020e-01 -7.13260174e-01
3.06516767e-01 -4.09705877e-01 -7.90639937e-01 -1.28105074e-01
3.81301463e-01 8.39071512e-01 1.04211807e+00 -1.35291100e+00
-9.53548908e-01 -4.71163124e-01 -2.35937592e-02 2.30678231e-01
-6.30021095e-01 -8.27071786e-01 -1.26421135e-02 -7.80099452e-01
3.89013529e-01 -8.27489078e-01 5.47335558e-02 -1.53780922e-01
-2.19205782e-01 -2.68953025e-01 7.78777421e-01 -2.36774668e-01
1.14382136e+00 -2.19424510e+00 2.34970078e-02 1.32472843e-01
5.57590723e-01 2.86405563e-01 -3.31930161e-01 1.33226141e-02
-3.48607004e-02 5.60414754e-02 -3.99960130e-01 -1.98263153e-01
-2.99778968e-01 4.25744832e-01 -7.92251587e-01 1.33171648e-01
1.82006508e-01 9.29888546e-01 -1.28695345e+00 6.72646612e-02
-2.49114275e-01 2.67871082e-01 -5.05944014e-01 1.26760975e-01
-1.22229360e-01 3.59685540e-01 -7.00383559e-02 7.07832873e-01
6.79402232e-01 -5.43131173e-01 1.11185506e-01 1.16976937e-02
2.65225083e-01 4.68508936e-02 -1.15011549e+00 1.54483795e+00
-3.14261019e-01 2.38680854e-01 -6.02877319e-01 -8.60957444e-01
7.49369085e-01 3.28887887e-02 -9.16442573e-02 -7.25818276e-01
-2.70585090e-01 2.31034473e-01 7.08750216e-03 -1.87216312e-01
3.68740469e-01 -3.42245311e-01 2.05709472e-01 5.40065467e-01
4.88764912e-01 5.05287468e-01 9.81919616e-02 4.78433549e-01
1.29814184e+00 -3.68804127e-01 1.62183180e-01 -3.38865876e-01
2.71578252e-01 -1.57385394e-01 6.53031349e-01 1.45020342e+00
-3.23145479e-01 2.99487948e-01 -2.17066616e-01 -6.63356006e-01
-6.04009986e-01 -1.92602062e+00 -2.23916158e-01 1.28446472e+00
1.67660460e-01 1.95031129e-02 -1.12525277e-01 -1.10707366e+00
3.28023702e-01 1.07437706e+00 -8.62051547e-01 -7.96146810e-01
-2.14844033e-01 -6.96340978e-01 2.93126732e-01 5.08157969e-01
4.82340306e-01 -1.15623915e+00 -5.07844687e-01 2.51950115e-01
-1.12183161e-01 -5.85653126e-01 -4.33588684e-01 6.68888986e-01
-1.16358221e+00 -1.37291765e+00 -4.38144386e-01 -8.74150872e-01
1.06036973e+00 5.16625345e-01 1.30384338e+00 1.38758495e-01
-2.27443606e-01 8.01727891e-01 -4.06388462e-01 -2.39982605e-01
-8.49485993e-02 8.41990188e-02 6.44192696e-01 8.10553730e-02
6.89530730e-01 -9.18058693e-01 -5.53380907e-01 1.80927366e-01
-1.04950202e+00 1.41110998e-02 8.43886554e-01 1.05661774e+00
5.96859574e-01 3.44114155e-01 1.11953378e+00 -1.22278285e+00
3.20365608e-01 -8.63055706e-01 -1.41909555e-01 4.82663602e-01
-1.02955329e+00 3.91461939e-01 5.01387358e-01 -9.43036795e-01
-1.25775266e+00 -4.62212302e-02 3.48449945e-01 -5.01404405e-01
-1.12771444e-01 4.87688661e-01 -7.69642591e-02 1.00296475e-01
8.59225929e-01 5.60645640e-01 -3.96425873e-01 -6.40504360e-01
5.74365318e-01 3.72152686e-01 8.53753269e-01 -6.97089732e-01
9.45972323e-01 5.19947112e-01 -7.57945359e-01 -4.93707120e-01
-1.39151156e+00 -4.95489120e-01 -9.85153317e-01 1.03010535e-01
1.65497869e-01 -1.02944517e+00 -2.91024804e-01 4.11001593e-01
-7.62720287e-01 -3.95925194e-01 -8.97555232e-01 4.54125404e-01
-2.90305972e-01 1.56268537e-01 -3.94492626e-01 -6.70112789e-01
-1.93883613e-01 -4.61940706e-01 5.86122811e-01 4.12857115e-01
-1.26856029e-01 -9.44761157e-01 6.82567880e-02 5.39420955e-02
6.17848277e-01 -1.64505973e-01 1.16205251e+00 -8.90989542e-01
-7.00112820e-01 1.23940825e-01 -1.18073151e-01 3.22292238e-01
2.90855348e-01 -6.85457230e-01 -1.26338422e+00 -8.30684960e-01
-6.21571653e-02 -4.81352270e-01 1.50291240e+00 -6.86598867e-02
1.13095105e+00 -4.93432701e-01 -5.36005437e-01 3.88849288e-01
1.29152679e+00 2.90608823e-01 3.83545458e-01 2.05877885e-01
4.58074361e-01 1.87045097e-01 2.25891620e-01 4.11055446e-01
3.38305026e-01 2.23721698e-01 4.37406860e-02 3.96187246e-01
-4.28741217e-01 -4.61891353e-01 1.30074367e-01 9.66653943e-01
2.29913637e-01 -2.06905007e-02 -8.88911188e-01 1.00106955e+00
-1.78701925e+00 -9.30387735e-01 5.91170728e-01 2.49818325e+00
1.34686983e+00 2.70266265e-01 -3.66094530e-01 1.37585521e-01
6.15857005e-01 -3.66582930e-01 -1.15699494e+00 2.25390851e-01
-1.83575541e-01 5.38446188e-01 1.18088491e-01 3.09484452e-01
-1.09610760e+00 9.83075380e-01 6.33107090e+00 7.03669488e-01
-8.97049129e-01 2.32495859e-01 5.98920226e-01 -1.39965668e-01
-4.69324946e-01 9.20425057e-02 -9.91391182e-01 4.10240531e-01
9.85553503e-01 -5.14587045e-01 5.16555965e-01 9.70257401e-01
-5.51132679e-01 -4.56103422e-02 -1.55849504e+00 9.29185271e-01
1.35331020e-01 -1.16799939e+00 5.56627214e-01 -2.96850175e-01
8.26250494e-01 -2.86861267e-02 4.80937749e-01 8.20314288e-01
5.04866421e-01 -7.28429973e-01 7.13896632e-01 9.62966800e-01
6.05044365e-01 -5.10546327e-01 4.03856546e-01 4.01355505e-01
-8.83746266e-01 -5.36525846e-01 -5.25162399e-01 -1.04700722e-01
-4.32070166e-01 9.12413001e-01 -1.04111123e+00 1.34012431e-01
8.54254961e-01 8.00082743e-01 -1.12288272e+00 1.18386173e+00
-3.21060956e-01 7.46204615e-01 -7.99515322e-02 4.00903881e-01
-7.88198486e-02 5.05772889e-01 3.86414379e-01 8.17994714e-01
3.35144758e-01 2.03584284e-01 1.66248083e-01 9.06530380e-01
-3.92849207e-01 -3.23583961e-01 -4.99904603e-01 -5.96545488e-02
8.88832211e-01 8.91643524e-01 -9.59846795e-01 -6.33570015e-01
-1.10086553e-01 1.06358182e+00 7.05071151e-01 5.19654512e-01
-5.50253093e-01 -2.54186124e-01 5.04088640e-01 3.52640748e-02
3.47670376e-01 -3.66028473e-02 3.94763146e-03 -1.54043984e+00
-7.46182427e-02 -6.32652938e-01 8.02404463e-01 -6.01066291e-01
-1.87789989e+00 6.67892933e-01 -5.90050891e-02 -8.06065500e-01
-2.02680044e-02 -1.98754981e-01 -1.95016056e-01 5.04132509e-01
-1.84737492e+00 -8.72643471e-01 -2.50124723e-01 6.63336575e-01
5.76758087e-01 -4.05134380e-01 8.88372958e-01 3.44909072e-01
-3.37032706e-01 8.89574170e-01 3.65233123e-01 -1.72016069e-01
7.77601838e-01 -1.13264024e+00 1.57273173e-01 9.74380851e-01
5.97336054e-01 9.40257251e-01 3.87624115e-01 -9.34197605e-01
-1.03926361e+00 -1.43326628e+00 9.76605892e-01 -7.69170225e-01
3.67758691e-01 -2.51343846e-01 -1.26178145e+00 9.08866346e-01
-3.84122580e-01 -2.79365163e-02 7.75099874e-01 3.96328121e-01
-9.30054307e-01 -1.82086036e-01 -1.26716876e+00 4.47309285e-01
1.47590601e+00 -5.81502378e-01 -1.16705823e+00 2.66003519e-01
1.00020027e+00 1.62595697e-02 -3.55806977e-01 5.05348623e-01
3.37883323e-01 -7.38831103e-01 1.02971911e+00 -8.19023967e-01
-6.62454963e-02 -3.57363254e-01 -1.27173439e-01 -1.36743200e+00
-6.87124908e-01 -2.58051127e-01 -5.26530087e-01 1.25384259e+00
4.01216000e-01 -7.37013102e-01 8.02965045e-01 5.62827647e-01
-3.97331156e-02 -4.01804626e-01 -9.21058953e-01 -1.10011661e+00
-1.80031974e-02 -3.51921879e-02 5.62728047e-01 1.28893006e+00
-3.59367788e-01 3.70262206e-01 -2.71633595e-01 3.29870999e-01
7.90235400e-01 1.75133646e-01 2.79719740e-01 -1.59071481e+00
-1.17412068e-01 -9.63940248e-02 -3.14308554e-01 -8.78218114e-01
7.28034750e-02 -1.17675662e+00 -1.32245698e-03 -9.76913989e-01
6.46534860e-01 -7.56528795e-01 -1.10894847e+00 1.10453856e+00
-3.90907228e-01 1.87293768e-01 1.03990398e-01 4.92769092e-01
-1.07076442e+00 5.87408006e-01 8.99410605e-01 -3.70925665e-01
-3.83910209e-01 -7.79707804e-02 -1.28601646e+00 5.79418242e-01
6.03860080e-01 -7.28127420e-01 -7.24259734e-01 -4.45241988e-01
7.36498758e-02 -5.15166759e-01 4.14311558e-01 -1.42767072e+00
5.03605962e-01 2.81476360e-02 7.32712388e-01 -4.75590259e-01
-4.20910539e-03 -7.91384339e-01 1.21124767e-01 5.66106021e-01
-6.58649087e-01 -2.21397847e-01 2.18724653e-01 1.05053711e+00
6.49071783e-02 -1.89327061e-01 9.18434322e-01 -2.79732615e-01
-1.27445149e+00 4.55790877e-01 -1.48258299e-01 4.58182096e-01
9.69566405e-01 -1.49094671e-01 -5.70238948e-01 3.32263671e-03
-1.19695282e+00 1.78648040e-01 3.96829635e-01 7.54808187e-01
9.25703764e-01 -1.41116083e+00 -5.07599890e-01 4.77183670e-01
3.68282706e-01 -2.90656984e-01 3.79874229e-01 3.68896455e-01
1.81190267e-01 2.80725300e-01 -2.63598293e-01 -4.72399265e-01
-7.58925915e-01 9.57464516e-01 6.36944175e-02 -2.21896812e-01
-4.22113657e-01 1.10280287e+00 4.03260529e-01 -5.04377544e-01
4.08487499e-01 -2.03914642e-01 -1.53661385e-01 3.97296585e-02
7.34046280e-01 1.75967515e-01 1.25521824e-01 -1.80258229e-01
-5.01519382e-01 -2.31344765e-03 -7.58500338e-01 3.80284011e-01
1.34441733e+00 -4.30566639e-01 7.82164931e-02 6.99445128e-01
9.39846873e-01 -2.31532082e-01 -1.24136138e+00 -7.87156045e-01
3.43292892e-01 -5.40664673e-01 -2.51549423e-01 -1.33313012e+00
-8.61075938e-01 5.45185983e-01 1.09674847e+00 -1.52170792e-01
1.23410237e+00 9.44558531e-02 6.33048177e-01 7.25588858e-01
7.23486722e-01 -1.01356328e+00 5.91499627e-01 7.35834301e-01
6.65063858e-01 -1.11664343e+00 -1.85356587e-01 1.00028310e-02
-4.14705038e-01 8.56560886e-01 8.67113531e-01 1.10856332e-01
7.43119955e-01 -2.24205792e-01 -1.11468382e-01 -9.96232554e-02
-9.94032979e-01 -4.16872352e-01 2.11429372e-01 8.19904566e-01
-4.44667861e-02 2.61058309e-03 1.53720438e-01 9.52375233e-01
-3.32058594e-02 1.24526680e-01 3.07592034e-01 1.01395476e+00
-6.74512744e-01 -8.37918043e-01 -2.69013997e-02 8.18160295e-01
-7.17434809e-02 -4.17344749e-01 -9.32164565e-02 5.52098334e-01
4.10034835e-01 7.98999548e-01 1.26594126e-01 -3.79584283e-01
2.65369415e-01 4.43966210e-01 3.58217150e-01 -1.01865125e+00
-1.73559278e-01 -6.47165120e-01 -4.72254902e-01 -3.33522767e-01
-3.70113701e-01 -4.63010788e-01 -1.00583708e+00 -2.33342983e-02
-2.74251789e-01 4.76985335e-01 -7.58156627e-02 9.15629923e-01
6.68407917e-01 4.04791027e-01 5.97379923e-01 -3.71565640e-01
-6.64786398e-01 -6.40297949e-01 -7.66394675e-01 6.34964585e-01
4.29023594e-01 -1.09239256e+00 -6.33926451e-01 2.67191738e-01] | [9.837520599365234, 3.3080875873565674] |
93887b33-6794-4468-be2a-142ed1b01ea0 | intrinsic-scene-decomposition-from-rgb-d | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Hachama_Intrinsic_Scene_Decomposition_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Hachama_Intrinsic_Scene_Decomposition_ICCV_2015_paper.pdf | Intrinsic Scene Decomposition From RGB-D images | In this paper, we address the problem of computing an intrinsic decomposition of the colors of a surface into an albedo and a shading term. The surface is reconstructed from a single or multiple RGB-D images of a static scene obtained from different views. We thereby extend and improve existing works in the area of intrinsic image decomposition. In a variational framework, we formulate the problem as a minimization of an energy composed of two terms: a data term and a regularity term. The first term is related to the image formation process and expresses the relation between the albedo, the surface normals, and the incident illumination. We use an affine shading model, a combination of a Lambertian model, and an ambient lighting term. This model is relevant for Lambertian surfaces. When available, multiple views can be used to handle view-dependent non-Lambertian reflections. The second term contains an efficient combination of l2 and l1-regularizers on the illumination vector field and albedo respectively. Unlike most previous approaches, especially Retinex-like techniques, these terms do not depend on the image gradient or texture, thus reducing the mixing shading/reflectance artifacts and leading to better results. The obtained non-linear optimization problem is efficiently solved using a cyclic block coordinate descent algorithm. Our method outperforms a range of state-of-the-art algorithms on a popular benchmark dataset. | ['Peter Wonka', 'Mohammed Hachama', 'Bernard Ghanem'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['intrinsic-image-decomposition'] | ['computer-vision'] | [ 5.92000484e-01 -4.20763820e-01 4.23743188e-01 -3.45957637e-01
-3.45587999e-01 -3.98807138e-01 5.94231248e-01 -6.42018840e-02
-4.30193037e-01 4.68791157e-01 -2.46649864e-03 1.15078717e-01
1.45363554e-01 -8.77752125e-01 -7.36915112e-01 -1.22520018e+00
6.53110921e-01 1.44748732e-01 1.34329215e-01 -3.02064568e-01
2.90724993e-01 5.23304760e-01 -1.87622118e+00 6.89197034e-02
8.13727021e-01 1.07105207e+00 2.08135888e-01 6.04454637e-01
-2.56180286e-01 5.54227114e-01 -2.45954514e-01 -5.06006368e-03
4.81940329e-01 -5.21434605e-01 -4.54274625e-01 4.77751166e-01
6.44732177e-01 -1.42408982e-01 7.83575624e-02 1.10343182e+00
3.34132522e-01 3.88782978e-01 5.93729019e-01 -8.40860307e-01
-2.70533383e-01 -5.93049347e-01 -8.32645953e-01 -2.77025282e-01
4.88040686e-01 -2.01557547e-01 7.95442104e-01 -9.79120433e-01
5.34297347e-01 1.07832336e+00 3.41106117e-01 3.23328316e-01
-1.27858269e+00 2.59548835e-02 9.01481211e-02 1.30594954e-01
-1.16228211e+00 -4.17260289e-01 1.22502172e+00 -5.50707817e-01
5.30890822e-01 5.20076811e-01 6.91668510e-01 5.46976149e-01
5.60266487e-02 3.71768802e-01 1.53926432e+00 -8.27442348e-01
3.95186007e-01 2.52272308e-01 1.50339872e-01 6.91213846e-01
1.79202154e-01 -1.27862301e-02 -5.82146168e-01 -3.45648408e-01
7.13410497e-01 8.98701921e-02 -6.92517221e-01 -6.94833159e-01
-7.90297210e-01 5.98729014e-01 1.47516251e-01 1.93586528e-01
-5.28576374e-01 -1.38299033e-01 -2.53757596e-01 9.26271677e-02
6.48922443e-01 -4.92891893e-02 -1.80516243e-01 3.34406763e-01
-7.19988704e-01 2.71728076e-02 1.00922656e+00 3.86333257e-01
1.24767315e+00 -4.62417193e-02 3.21966588e-01 9.39116657e-01
7.30321348e-01 8.95379841e-01 2.55587865e-02 -9.72258449e-01
2.92080820e-01 5.77361405e-01 4.93549973e-01 -1.03301871e+00
-1.52895093e-01 -2.00782031e-01 -7.12324977e-01 7.01638043e-01
5.09922981e-01 6.45797029e-02 -6.56481862e-01 1.45885408e+00
7.95153201e-01 1.76802233e-01 4.33653444e-02 1.05848861e+00
5.94484568e-01 6.51004314e-01 -7.05831170e-01 -5.00302374e-01
1.27996933e+00 -8.98846805e-01 -8.24362516e-01 -1.70291960e-01
7.46662915e-02 -1.22927523e+00 8.31427097e-01 6.96658671e-01
-1.29547060e+00 -2.87558138e-01 -1.04063535e+00 -3.19351077e-01
-1.29629783e-02 2.85052329e-01 1.84547529e-01 7.16409028e-01
-1.11338806e+00 3.75264913e-01 -7.46533573e-01 -1.34806752e-01
-1.94434702e-01 1.12599261e-01 -1.98489457e-01 -3.53943795e-01
-5.24449825e-01 7.69083321e-01 -3.96160960e-01 4.03732151e-01
-5.26976585e-01 -4.56895709e-01 -7.11372375e-01 -2.19887406e-01
3.65932375e-01 -6.53139055e-01 4.50441629e-01 -1.18197465e+00
-1.85733473e+00 1.03020680e+00 -5.53605258e-01 2.52457798e-01
5.08166671e-01 -2.31218174e-01 -4.86580059e-02 1.82684883e-01
-2.71231443e-01 -2.98184693e-01 1.08759296e+00 -1.86901069e+00
-2.32771829e-01 -6.46072686e-01 6.82408884e-02 6.61008894e-01
1.43587068e-02 -3.31895292e-01 -7.40816712e-01 -2.67901748e-01
3.20501715e-01 -9.41514611e-01 -1.15095275e-02 2.13788509e-01
-3.08400989e-01 3.97528440e-01 6.20840132e-01 -7.37491548e-01
6.66307509e-01 -2.19984412e+00 5.23497999e-01 2.08193794e-01
2.31420379e-02 8.82788748e-02 6.35837018e-02 4.17268544e-01
-7.41246715e-02 -5.28821766e-01 -4.60967481e-01 -6.78901374e-01
-2.68851846e-01 1.56823337e-01 -1.26921922e-01 9.23146725e-01
-2.29963690e-01 2.27265388e-01 -6.91602051e-01 -1.55157179e-01
4.55436260e-01 1.11067176e+00 -3.19676071e-01 3.89081866e-01
-1.65771499e-01 7.06634820e-01 -2.66087502e-01 1.98430166e-01
1.14511573e+00 9.39387754e-02 1.74044222e-01 -2.38774642e-01
-5.26450217e-01 1.34513080e-01 -1.61919296e+00 1.73378968e+00
-7.76658654e-01 3.16811770e-01 7.78926671e-01 -8.44609141e-01
1.02063990e+00 1.72514841e-01 6.01744056e-01 -6.44410670e-01
5.06323949e-02 1.95531011e-01 -5.10917127e-01 -3.55498552e-01
2.30014890e-01 -2.20565826e-01 8.48153889e-01 3.74340236e-01
-3.76273870e-01 -2.49906227e-01 9.80953723e-02 -1.33278251e-01
7.94083238e-01 4.60585684e-01 4.53627110e-02 -4.29779619e-01
9.79050398e-01 -4.31443632e-01 5.11789441e-01 2.29275420e-01
4.75005895e-01 8.26426506e-01 4.41066623e-01 -4.69134241e-01
-7.85541415e-01 -7.81756938e-01 -2.39825130e-01 8.17363858e-01
3.84241611e-01 -6.58412352e-02 -7.88380265e-01 -1.79586317e-02
-2.12575734e-01 6.83206081e-01 -5.12244523e-01 2.59436309e-01
-5.24715424e-01 -8.60166550e-01 -3.20909053e-01 -2.02135488e-01
5.34116268e-01 -5.36166191e-01 -8.51199329e-01 -1.02595866e-01
-2.73204982e-01 -1.11059153e+00 -3.84015918e-01 -3.07599027e-02
-9.09789860e-01 -1.22206402e+00 -8.37879360e-01 -3.86623204e-01
8.86586547e-01 6.83484197e-01 1.17790902e+00 1.51672304e-01
-2.43205383e-01 8.55543137e-01 -1.63930774e-01 -2.26372018e-01
-1.50017381e-01 -6.18068874e-01 -3.25083673e-01 9.44960177e-01
-3.04275732e-02 -6.21115923e-01 -7.85953045e-01 4.17345226e-01
-1.11311913e+00 3.55260760e-01 1.30570024e-01 6.08355761e-01
8.47579479e-01 -2.28852138e-01 -5.14920533e-01 -8.54006827e-01
-2.41005681e-02 -1.14316225e-01 -1.05630565e+00 3.24633330e-01
-4.82064068e-01 1.21824302e-01 5.11543810e-01 -5.48102856e-02
-1.50980246e+00 3.31142694e-01 4.97111157e-02 -3.21593374e-01
-8.60500559e-02 -5.83146662e-02 -2.79232621e-01 -4.36574429e-01
4.14318115e-01 3.92889351e-01 -1.27739266e-01 -7.04885542e-01
2.19228446e-01 2.71069437e-01 1.65638655e-01 -5.46579063e-01
8.02997768e-01 1.16772199e+00 4.93575245e-01 -1.43091559e+00
-6.55037463e-01 -7.52954185e-01 -6.11558616e-01 -3.29148024e-01
8.47193480e-01 -5.91545820e-01 -6.73873425e-01 7.15221286e-01
-1.23245895e+00 -3.59920830e-01 -3.14925760e-01 3.17988336e-01
-4.66769993e-01 6.92075253e-01 -2.36267582e-01 -1.03042698e+00
-8.35627392e-02 -1.10393918e+00 1.20614171e+00 1.26274526e-01
6.13382161e-01 -1.21160400e+00 5.44252038e-01 4.82762128e-01
3.28120768e-01 4.32841510e-01 7.30721056e-01 4.28502351e-01
-7.35723972e-01 1.62799522e-01 -8.50403160e-02 6.53034985e-01
1.45409256e-01 5.55890659e-03 -1.15732884e+00 -2.48896256e-01
7.93467522e-01 -1.80404764e-02 9.11583722e-01 4.11451578e-01
6.39127076e-01 -2.53858089e-01 9.25646052e-02 1.02049983e+00
2.12033725e+00 1.88547466e-02 6.64588153e-01 2.33819172e-01
9.45012867e-01 7.68734932e-01 2.88114190e-01 6.06371045e-01
4.06842202e-01 9.84554946e-01 7.66630769e-01 -4.35982257e-01
-2.16377735e-01 3.68062586e-01 1.95302531e-01 7.26367056e-01
-6.35439157e-01 -2.26836562e-01 -8.05422187e-01 2.50326991e-01
-1.72158504e+00 -6.41215324e-01 -7.72067368e-01 2.78028870e+00
5.07489085e-01 -4.67188239e-01 -3.08487326e-01 1.82100132e-01
3.78710330e-01 3.66049558e-01 -4.23793584e-01 -3.01744133e-01
-3.71700376e-01 1.91475853e-01 4.98602211e-01 1.08622015e+00
-6.82431221e-01 3.88548493e-01 5.23549843e+00 2.94645160e-01
-1.25804460e+00 1.27183586e-01 3.17036748e-01 8.97973776e-03
-8.22202265e-01 1.23265751e-01 -4.79524761e-01 1.96134731e-01
3.25280398e-01 4.34938550e-01 8.25374186e-01 4.75821823e-01
1.91362128e-01 -5.10637403e-01 -8.35271120e-01 1.02811539e+00
4.24034983e-01 -7.56119668e-01 -1.74766302e-01 1.92576483e-01
8.51854265e-01 8.46959949e-02 5.64454682e-03 -6.70247436e-01
-1.32826015e-01 -6.56075120e-01 5.10893941e-01 8.06093454e-01
5.92513323e-01 -3.23531955e-01 3.75559211e-01 2.25257888e-01
-1.05543959e+00 3.40593427e-01 -2.46379972e-01 -2.65192930e-02
1.47616029e-01 9.37764347e-01 9.21897218e-02 7.77899504e-01
5.89713871e-01 6.18022382e-01 -1.63821518e-01 6.73976719e-01
-4.37528163e-01 1.59624010e-01 -6.07581913e-01 2.92873025e-01
5.93082719e-02 -1.05073893e+00 6.98085666e-01 7.53993869e-01
9.20435265e-02 3.35294187e-01 1.43922660e-02 7.39084363e-01
2.35124096e-01 3.12977433e-01 -4.10953134e-01 7.10357785e-01
-3.52273703e-01 1.47666764e+00 -5.72902560e-01 5.97803108e-03
-7.50587642e-01 1.21542895e+00 1.49436831e-01 8.67406487e-01
-4.05419469e-01 5.35269210e-04 4.14478302e-01 1.92102671e-01
2.35183820e-01 -3.45776081e-01 -2.39187166e-01 -1.45623481e+00
4.02472615e-01 -7.67589748e-01 -8.52952972e-02 -7.73855388e-01
-8.47216368e-01 5.22036076e-01 -1.29588544e-01 -9.97293055e-01
2.06528991e-01 -8.12397659e-01 -6.07815742e-01 1.18739152e+00
-2.07524347e+00 -1.08009756e+00 -6.37493491e-01 9.22215343e-01
2.42260650e-01 3.36720198e-01 7.95251906e-01 2.68173426e-01
-4.12428409e-01 -2.36205861e-01 5.64650953e-01 -3.94935191e-01
6.10664666e-01 -1.27806401e+00 -2.27303222e-01 9.69800830e-01
2.17190403e-02 5.00410438e-01 7.16012955e-01 -3.02825153e-01
-1.80197573e+00 -5.12002170e-01 5.87758660e-01 -8.48261192e-02
1.35034949e-01 -3.27006936e-01 -9.17838216e-01 4.52679753e-01
3.21463019e-01 2.33040839e-01 6.44078374e-01 -1.57514811e-01
-2.94913620e-01 -3.82852346e-01 -9.54774916e-01 1.89459771e-01
6.47132337e-01 -3.34985048e-01 -1.12579174e-01 3.97638172e-01
2.58887522e-02 -5.18666148e-01 -5.79795659e-01 1.48651704e-01
6.20637059e-01 -1.51878572e+00 9.98540819e-01 8.27430040e-02
2.50469327e-01 -5.01992702e-01 -3.89597923e-01 -1.18959844e+00
-1.32964626e-01 -6.70041442e-01 2.14855298e-02 1.06304133e+00
-7.32002920e-03 -7.35638797e-01 5.41774869e-01 6.63348317e-01
-1.60776880e-02 -6.79092467e-01 -7.80940711e-01 -3.32027346e-01
-3.38836879e-01 -1.23901710e-01 1.25265121e-01 7.72882938e-01
-6.51419342e-01 2.97954798e-01 -4.84719157e-01 3.48216653e-01
1.15074003e+00 3.86646509e-01 8.47300351e-01 -1.22858441e+00
-4.38094944e-01 -2.49462038e-01 1.38396472e-02 -1.11924434e+00
-7.57789239e-02 -4.71542418e-01 2.11252809e-01 -1.44728887e+00
-4.95899096e-03 -4.06858057e-01 -5.08975163e-02 5.71937999e-03
-1.03995308e-01 2.57107526e-01 7.40414038e-02 3.39438885e-01
-1.07705250e-01 6.06686115e-01 1.11779630e+00 1.25030667e-01
-3.91503602e-01 1.44005582e-01 -1.69614494e-01 1.02882075e+00
4.16437298e-01 -3.29352826e-01 -3.96723807e-01 -8.27593923e-01
5.25273263e-01 -8.10047495e-04 3.50949496e-01 -6.64711595e-01
2.23863602e-01 -2.50101537e-01 4.73569557e-02 -4.10118461e-01
7.16216803e-01 -1.01322329e+00 3.33243549e-01 1.36691555e-01
1.20336175e-01 -3.03947210e-01 -1.58464208e-01 6.46723628e-01
-1.02631897e-01 -3.94425809e-01 9.56927776e-01 -2.57201761e-01
-1.98286578e-01 1.28463998e-01 -8.96667540e-02 -1.47112831e-01
6.82978511e-01 -1.69771537e-01 -1.08490296e-01 -3.03242385e-01
-3.13371927e-01 -1.52463868e-01 9.56732571e-01 -4.39557396e-02
5.84102809e-01 -1.10513961e+00 -7.01282680e-01 3.59515041e-01
-1.98390499e-01 1.59444645e-01 3.47016841e-01 1.02700448e+00
-8.77890885e-01 -8.42676014e-02 -1.07864827e-01 -6.34218574e-01
-1.35802734e+00 2.69875228e-01 4.51644570e-01 -2.82297194e-01
-5.25101602e-01 6.71574891e-01 6.43964291e-01 -3.62285912e-01
4.86440472e-02 -1.39394298e-01 -2.35519022e-01 -1.35915712e-01
4.57901716e-01 8.23078334e-01 1.90157160e-01 -1.04276419e+00
-4.66127872e-01 1.34409487e+00 5.39411306e-01 -3.61461282e-01
1.41255057e+00 -3.55826497e-01 -6.00558877e-01 7.07304895e-01
1.22858071e+00 4.11088824e-01 -1.31059682e+00 -2.70096272e-01
-5.55606484e-01 -6.24867618e-01 4.73074436e-01 -4.67433840e-01
-1.22720301e+00 1.01595914e+00 5.35996854e-01 3.23655725e-01
1.48110938e+00 -3.90740454e-01 5.07766664e-01 2.58679669e-02
2.83483148e-01 -9.18626130e-01 -1.06628858e-01 4.12281036e-01
1.05285954e+00 -1.17205548e+00 4.32398140e-01 -6.75307631e-01
-3.04653555e-01 1.24613905e+00 8.40708539e-02 -2.55749404e-01
7.03116179e-01 5.84967732e-02 2.92811126e-01 -1.31926566e-01
-3.66062462e-01 -3.05519789e-01 6.63487256e-01 2.92953432e-01
5.18725276e-01 -2.23282441e-01 -5.02483845e-01 -2.80389845e-01
3.53214264e-01 -2.50574976e-01 4.12428200e-01 6.77890539e-01
-2.23119587e-01 -1.24455333e+00 -7.26262152e-01 -1.20867275e-01
-3.30939740e-01 2.92398855e-02 -1.60853162e-01 3.67334574e-01
2.54363418e-01 9.54021215e-01 -4.90711704e-02 3.04704875e-01
4.46889997e-01 -4.61488143e-02 6.24153435e-01 -4.86631513e-01
-2.44933262e-01 5.42543113e-01 -2.42509946e-01 -7.71475255e-01
-9.53209698e-01 -8.32612574e-01 -9.05557752e-01 2.47794501e-02
-5.43228745e-01 -1.40291736e-01 1.15629709e+00 8.18264008e-01
1.75878350e-02 2.07540840e-01 8.87143552e-01 -1.16583037e+00
-1.55047461e-01 -4.72437799e-01 -8.21941137e-01 5.53453684e-01
5.95530093e-01 -4.88427877e-01 -6.14755571e-01 1.59360081e-01] | [9.925837516784668, -2.9478325843811035] |
4def3508-df31-4b92-84af-3fae05f0f77a | adversarial-synthesis-based-data-augmentation | 2205.15747 | null | https://arxiv.org/abs/2205.15747v2 | https://arxiv.org/pdf/2205.15747v2.pdf | Adversarial synthesis based data-augmentation for code-switched spoken language identification | Spoken Language Identification (LID) is an important sub-task of Automatic Speech Recognition(ASR) that is used to classify the language(s) in an audio segment. Automatic LID plays an useful role in multilingual countries. In various countries, identifying a language becomes hard, due to the multilingual scenario where two or more than two languages are mixed together during conversation. Such phenomenon of speech is called as code-mixing or code-switching. This nature is followed not only in India but also in many Asian countries. Such code-mixed data is hard to find, which further reduces the capabilities of the spoken LID. Hence, this work primarily addresses this problem using data augmentation as a solution on the on the data scarcity of the code-switched class. This study focuses on Indic language code-mixed with English. Spoken LID is performed on Hindi, code-mixed with English. This research proposes Generative Adversarial Network (GAN) based data augmentation technique performed using Mel spectrograms for audio data. GANs have already been proven to be accurate in representing the real data distribution in the image domain. Proposed research exploits these capabilities of GANs in speech domains such as speech classification, automatic speech recognition, etc. GANs are trained to generate Mel spectrograms of the minority code-mixed class which are then used to augment data for the classifier. Utilizing GANs give an overall improvement on Unweighted Average Recall by an amount of 3.5% as compared to a Convolutional Recurrent Neural Network (CRNN) classifier used as the baseline reference. | ['Dr. Hardik Sailor', 'Dr. Abhishek Bhatt', 'Dr. Shrinivas Mahajan', 'Poorval Wanere', 'Chirag Patil', 'Parth Shastri'] | 2022-05-30 | null | null | null | null | ['spoken-language-identification'] | ['speech'] | [ 1.80242643e-01 -7.57848620e-02 1.71072870e-01 -3.14157568e-02
-1.01673532e+00 -7.36677051e-01 6.15776181e-01 -1.63449183e-01
-2.36510739e-01 8.57741416e-01 4.61106509e-01 -4.40099090e-01
5.42531013e-01 -4.75523919e-01 -3.40529710e-01 -8.33725154e-01
3.25976193e-01 4.74678338e-01 -2.95187116e-01 -3.61211389e-01
1.26821011e-01 4.10367638e-01 -1.65702784e+00 4.38816845e-01
6.53896689e-01 6.02837682e-01 2.91756958e-01 1.00689089e+00
-5.03397882e-01 8.92018139e-01 -8.17022800e-01 -1.58512801e-01
2.81827748e-01 -8.20199549e-01 -5.85152209e-01 1.47662535e-01
9.99222249e-02 2.16822661e-02 -2.34401941e-01 1.18576682e+00
6.75578475e-01 1.20859277e-02 7.53466845e-01 -1.19719577e+00
-5.40892363e-01 7.01684713e-01 -5.57568431e-01 1.89048961e-01
4.79629219e-01 -4.96864691e-02 5.50491333e-01 -8.73139918e-01
2.34110564e-01 1.18409598e+00 5.66212654e-01 7.24235654e-01
-9.79802430e-01 -1.01350176e+00 -4.16287005e-01 -1.76368654e-02
-1.46118128e+00 -5.78082383e-01 8.43446016e-01 -5.34192502e-01
8.62059772e-01 3.03554684e-01 3.32671225e-01 8.24199796e-01
1.24603391e-01 5.86219132e-01 1.46246660e+00 -7.84076393e-01
-3.73496599e-02 5.60348213e-01 -1.17893167e-01 3.21038216e-01
-4.43969637e-01 9.09647718e-02 -4.23252642e-01 2.84302711e-01
2.06109852e-01 -3.10212672e-01 -1.17095627e-01 4.39311206e-01
-8.00503075e-01 1.13634443e+00 1.62331954e-01 5.72288513e-01
-3.15641403e-01 -3.50410074e-01 5.62312484e-01 6.17796600e-01
5.05819857e-01 1.16022378e-01 -1.96333036e-01 -4.58478093e-01
-1.06911552e+00 -2.13483736e-01 6.23741627e-01 6.17505074e-01
4.65847492e-01 6.29997969e-01 1.64664477e-01 1.26525700e+00
3.01995337e-01 5.13161659e-01 1.15347052e+00 -1.59142524e-01
5.00731885e-01 6.23979211e-01 -3.34350735e-01 -6.93975985e-01
-1.32402316e-01 -5.40029287e-01 -9.82131481e-01 3.91682416e-01
6.05958067e-02 -3.15658271e-01 -1.34540951e+00 1.64340246e+00
1.36743858e-01 -3.02055310e-02 5.04978120e-01 6.34385467e-01
7.62385666e-01 1.16738379e+00 -8.83271545e-02 -1.01762131e-01
1.33478439e+00 -9.91157949e-01 -8.28406334e-01 -1.47955254e-01
3.68155301e-01 -1.35843074e+00 1.10409212e+00 5.09833038e-01
-6.39603794e-01 -6.88845456e-01 -1.07259977e+00 2.68998504e-01
-6.51405215e-01 3.04847091e-01 2.01247960e-01 1.13396478e+00
-1.16855407e+00 -3.63262653e-01 -3.35723579e-01 -4.67458665e-01
7.12691322e-02 4.82345790e-01 -6.13995373e-01 2.25656591e-02
-1.14383245e+00 8.21458399e-01 3.28767210e-01 -1.88635796e-01
-9.13576245e-01 -2.08188817e-01 -8.35764468e-01 -3.23066056e-01
-1.72506496e-01 2.07080498e-01 1.02883506e+00 -1.58129966e+00
-1.57572365e+00 9.96551394e-01 -4.81487066e-02 -6.74289346e-01
5.20156205e-01 2.28082553e-01 -8.95772576e-01 -2.31884956e-01
-1.27798237e-03 6.44289970e-01 9.22892630e-01 -1.04624844e+00
-6.61740780e-01 -1.54062778e-01 -4.03002113e-01 3.65078598e-01
-2.44608685e-01 1.85574979e-01 -2.55136695e-02 -1.06452751e+00
1.81180909e-01 -1.14576387e+00 4.57506239e-01 -8.00830245e-01
-4.73208636e-01 -6.69352189e-02 1.25469077e+00 -1.19638395e+00
1.32867908e+00 -2.32131362e+00 -1.65759206e-01 9.26477313e-02
-3.77764881e-01 4.90542203e-01 2.18273196e-02 5.60721755e-01
-2.89432973e-01 5.47471456e-02 -4.29544657e-01 -3.98494899e-01
-4.71921712e-01 1.94258928e-01 -2.14959174e-01 2.86509335e-01
1.83855861e-01 2.29662240e-01 -2.38875955e-01 -2.48619273e-01
2.01621369e-01 6.45356834e-01 -2.31690243e-01 3.31052095e-01
1.10789165e-01 7.59304464e-01 1.39713511e-01 6.98363006e-01
6.33973658e-01 6.51170909e-01 -1.66268185e-01 -6.27991464e-03
-3.07905406e-01 1.93773076e-01 -1.22552621e+00 1.38489354e+00
-8.45914900e-01 8.49257469e-01 1.79435760e-01 -1.05984581e+00
1.34216201e+00 7.10286438e-01 6.64928257e-02 -4.09196824e-01
7.06836656e-02 4.24743950e-01 3.76199096e-01 -3.85269314e-01
3.87813270e-01 -2.91722417e-01 -1.33158669e-01 4.17562097e-01
1.50673717e-01 -2.34213322e-01 -4.75626253e-02 -7.59512186e-02
7.61138260e-01 -3.38243693e-01 2.85464615e-01 -9.47951451e-02
1.01363838e+00 -1.11546740e-01 1.96293056e-01 3.78866524e-01
-6.79740533e-02 7.23206758e-01 4.83717322e-02 -8.91361758e-02
-1.24365115e+00 -7.59590626e-01 -1.14168927e-01 8.82417560e-01
-5.97020090e-01 2.50475764e-01 -8.37244987e-01 -3.93349081e-01
-3.67037326e-01 5.57129562e-01 -3.25071037e-01 -1.82060391e-01
-3.26411486e-01 -5.66541910e-01 1.01855731e+00 1.47043258e-01
8.75978112e-01 -1.36349511e+00 -5.23285270e-02 4.38207954e-01
-6.32177219e-02 -9.10647869e-01 -5.30865490e-01 4.48615879e-01
-3.39077741e-01 -6.73219621e-01 -9.53328669e-01 -1.17937326e+00
5.30894279e-01 -1.20922849e-01 7.37604082e-01 -3.35682362e-01
-2.99353659e-01 1.82010576e-01 -6.11449540e-01 -6.79992974e-01
-1.20970452e+00 1.53640196e-01 2.51441926e-01 3.63183677e-01
4.37423617e-01 -3.41602653e-01 -2.00120687e-01 1.31904945e-01
-8.51035416e-01 -9.91203412e-02 6.30272150e-01 9.71684575e-01
1.10000074e-01 3.88827801e-01 1.00173831e+00 -7.18197763e-01
9.07546520e-01 -5.88709295e-01 -4.22684610e-01 1.68822870e-01
-3.51437956e-01 -4.95840050e-02 7.56781697e-01 -5.76077402e-01
-9.96968687e-01 3.26091468e-01 -5.17481565e-01 -1.34073496e-01
-3.59627843e-01 5.78534961e-01 -2.06440777e-01 -6.33617789e-02
3.82614374e-01 6.02696359e-01 2.96102494e-01 -4.36553210e-01
-9.17941630e-02 1.67109001e+00 5.97398639e-01 -1.46293432e-01
6.07449830e-01 -7.37262368e-02 -4.87597257e-01 -1.13040566e+00
-1.93972364e-01 -6.22021556e-01 -3.08314651e-01 -2.92006910e-01
1.02996898e+00 -1.10721469e+00 -2.08533138e-01 8.82207215e-01
-9.82483685e-01 -1.11234151e-01 1.44180238e-01 6.31604731e-01
-1.67035326e-01 -4.09445167e-02 -4.49918151e-01 -1.24834442e+00
-5.23006618e-01 -1.34780657e+00 7.36708283e-01 2.20846757e-01
-2.44683191e-01 -8.01623881e-01 8.55651721e-02 5.17215610e-01
4.85507131e-01 3.82765271e-02 7.98248172e-01 -9.81927335e-01
-5.05570555e-03 -4.06849116e-01 1.81156740e-01 9.29806411e-01
5.78975976e-01 -7.32032210e-02 -1.21776509e+00 -3.04353505e-01
1.22671470e-01 -4.41753060e-01 5.56090295e-01 1.68069690e-01
6.59405231e-01 -3.52841675e-01 3.25306743e-01 2.11389586e-01
1.30719900e+00 8.08352530e-01 5.96973538e-01 6.92624077e-02
6.39239669e-01 4.77824479e-01 4.40578848e-01 2.19387248e-01
7.82341789e-03 6.18442297e-01 6.63917959e-02 -5.00565693e-02
-4.43216860e-01 -3.06818843e-01 7.29915142e-01 1.39429379e+00
3.97798330e-01 -2.94235289e-01 -1.22185373e+00 6.00620747e-01
-1.08307135e+00 -7.75784791e-01 6.56149536e-02 2.21794438e+00
1.04667842e+00 3.04158460e-02 2.02971756e-01 8.58600914e-01
1.05458665e+00 -1.54976413e-01 -2.51967013e-01 -9.11746502e-01
-2.34243095e-01 6.27500117e-01 4.96574849e-01 6.96870148e-01
-1.10645425e+00 9.37118709e-01 5.01741314e+00 1.08393943e+00
-1.73458564e+00 2.96692193e-01 7.25916862e-01 4.50216234e-01
4.41100933e-02 -2.72919953e-01 -6.71653092e-01 8.06124866e-01
1.15575039e+00 1.35542378e-01 6.84090197e-01 6.83053017e-01
2.06055671e-01 -1.28040195e-01 -6.74529195e-01 1.24731565e+00
3.34800124e-01 -7.66992271e-01 2.19835922e-01 1.00091778e-01
8.60669672e-01 -1.88232064e-02 3.69593799e-01 3.88283551e-01
3.45533133e-01 -1.31109560e+00 6.01227939e-01 -2.96323118e-03
1.21460748e+00 -1.14682412e+00 1.06845975e+00 3.78358960e-01
-1.20218503e+00 -9.87206995e-02 2.36683227e-02 1.03716180e-01
-1.51955247e-01 2.38037258e-01 -1.36545897e+00 1.22823067e-01
4.25973147e-01 1.27372995e-01 -3.68304074e-01 7.25381136e-01
9.16252881e-02 9.20728624e-01 -3.26707155e-01 9.70043615e-02
4.53911901e-01 -1.45666167e-01 5.61327100e-01 1.25737643e+00
6.77506030e-01 -2.81333447e-01 -5.18373176e-02 3.22961539e-01
-2.13526011e-01 3.66687149e-01 -8.93367946e-01 -3.00532401e-01
4.98966992e-01 9.01514471e-01 -6.69674695e-01 -1.82177387e-02
-2.90208846e-01 1.05623996e+00 -3.00664544e-01 1.04973510e-01
-5.38195908e-01 -4.29283887e-01 3.55246246e-01 1.89682823e-02
-1.47046655e-01 -5.21208830e-02 -3.21919098e-03 -6.16060495e-01
-1.60173103e-01 -1.42428362e+00 2.17104062e-01 -6.23006165e-01
-1.01214600e+00 1.01203883e+00 -4.57829982e-01 -1.40144134e+00
-4.93256092e-01 -3.44601244e-01 -5.92913091e-01 1.17475593e+00
-1.12803745e+00 -1.45812774e+00 -2.84560144e-01 6.52259707e-01
1.31483746e+00 -1.08338559e+00 1.09809911e+00 6.10012949e-01
-3.60489368e-01 8.45154107e-01 2.05528021e-01 4.45386052e-01
6.12264097e-01 -1.14799583e+00 9.41235945e-02 1.02048588e+00
4.11446303e-01 2.43697405e-01 6.39912188e-01 -5.79116762e-01
-1.00604665e+00 -1.04335082e+00 7.09151447e-01 5.81223378e-03
3.69714051e-01 -4.53344017e-01 -6.21460319e-01 3.34853828e-01
5.46787679e-01 -2.82890111e-01 7.53078282e-01 -4.65695173e-01
-1.18078083e-01 -3.27249646e-01 -1.42637885e+00 1.32891387e-01
1.65734738e-01 -9.51338530e-01 -3.60791266e-01 1.01801530e-01
3.80989373e-01 -1.72530636e-01 -4.26319867e-01 3.60712521e-02
3.32467407e-01 -9.23877597e-01 4.97864366e-01 -2.22309306e-01
2.75977314e-01 -3.40564400e-01 -3.80801469e-01 -1.43425620e+00
4.11621690e-01 -5.85164070e-01 4.52752054e-01 1.92116296e+00
4.21064675e-01 -7.14112520e-01 6.41184807e-01 -8.15611109e-02
-8.69998112e-02 -1.78179100e-01 -1.11909509e+00 -5.68293571e-01
5.57969101e-02 -2.22305715e-01 3.76569599e-01 1.09647560e+00
-4.37024057e-01 1.78999528e-01 -4.35633630e-01 7.91371800e-03
2.29415655e-01 -6.73419476e-01 6.90136909e-01 -1.01129484e+00
-7.55569264e-02 -1.64095059e-01 -6.68162644e-01 -4.46643919e-01
2.60311216e-01 -9.29453731e-01 2.45170179e-03 -1.09823537e+00
-1.31663218e-01 -6.19883597e-01 -2.01257408e-01 3.99812400e-01
3.13352972e-01 4.60739613e-01 9.64561328e-02 1.69205472e-01
3.63783419e-01 1.26101613e-01 6.81514561e-01 -3.75008315e-01
-3.31917971e-01 2.94526041e-01 -3.49483967e-01 4.51273978e-01
1.21987236e+00 -5.25544047e-01 -5.63667476e-01 8.17186758e-02
-9.85805616e-02 1.43569157e-01 -4.14511748e-02 -1.42108238e+00
1.39261588e-01 3.51296812e-01 6.59552142e-02 -6.33522868e-01
3.21985781e-01 -1.02847099e+00 3.89929444e-01 5.46801031e-01
-2.96945184e-01 2.58945853e-01 3.63787889e-01 1.76969334e-01
-8.26708496e-01 -3.64649922e-01 1.12703764e+00 -2.03911975e-01
-6.01487875e-01 -2.29699329e-01 -6.94662213e-01 -1.10441431e-01
1.14104903e+00 -5.10338783e-01 1.33065076e-03 -8.26569498e-01
-6.15002632e-01 -2.36246899e-01 1.29999429e-01 5.60884595e-01
4.71966952e-01 -1.28613675e+00 -8.92729342e-01 6.24441862e-01
5.01648970e-02 -3.21811080e-01 2.37429783e-01 5.73601723e-01
-9.59423959e-01 4.82597798e-01 -3.96177888e-01 -3.88596237e-01
-1.69883418e+00 1.13597699e-01 2.93449074e-01 -8.71117339e-02
-1.08471513e-01 8.96198928e-01 -1.85649797e-01 -4.48901117e-01
3.93189073e-01 -5.52752316e-02 -5.65569282e-01 2.70472884e-01
2.98704714e-01 1.20267816e-01 2.89097130e-01 -1.05163538e+00
-2.37858474e-01 4.66903746e-01 -1.87245365e-02 -4.64382559e-01
1.02772486e+00 -1.11868612e-01 -1.96177676e-01 7.92494178e-01
1.42769289e+00 3.90189439e-01 -6.75185144e-01 2.20725030e-01
-1.34770632e-01 -1.19644832e-02 9.51135084e-02 -9.95571792e-01
-1.03482568e+00 9.88277912e-01 1.15767217e+00 2.90962905e-01
1.15220559e+00 -2.34844103e-01 6.53477311e-01 -5.62498197e-02
1.99610874e-01 -1.27878904e+00 -1.68009683e-01 4.69975144e-01
9.89388287e-01 -1.36475801e+00 -6.06898665e-01 9.75084901e-02
-9.35360730e-01 1.05213022e+00 4.30663079e-01 4.84003834e-02
7.85155773e-01 5.60297370e-01 5.05811572e-01 3.87984157e-01
-3.31110060e-01 -1.01740904e-01 4.67724577e-02 7.95262575e-01
7.07343936e-01 1.80934146e-01 -1.63942307e-01 2.28330031e-01
-6.34508252e-01 -3.63884985e-01 5.93771219e-01 8.21814597e-01
-2.46111840e-01 -1.28548026e+00 -8.17324579e-01 5.00200629e-01
-8.30008268e-01 -2.90113360e-01 -4.20190662e-01 5.36594033e-01
3.85116398e-01 1.30994713e+00 1.85403094e-01 -7.02197373e-01
-1.43148288e-01 4.70225692e-01 -7.88440332e-02 -6.31107271e-01
-8.51572871e-01 -2.85940594e-03 -3.54816429e-02 3.05436492e-01
-4.91457134e-01 -4.82495129e-01 -1.13690150e+00 -1.01965830e-01
-1.70080498e-01 3.88621032e-01 1.14957106e+00 6.57982707e-01
-5.94632551e-02 5.15147567e-01 8.71633649e-01 -3.61100525e-01
-3.27858597e-01 -1.37893713e+00 -6.01993978e-01 2.85261631e-01
6.21482134e-01 -5.08687496e-01 -6.57294869e-01 2.32030526e-01] | [14.248126029968262, 6.529244899749756] |
4ffc4e95-237c-43cd-8007-e55154c470ba | modelling-uncertainty-in-deep-learning-for | 1509.05909 | null | http://arxiv.org/abs/1509.05909v2 | http://arxiv.org/pdf/1509.05909v2.pdf | Modelling Uncertainty in Deep Learning for Camera Relocalization | We present a robust and real-time monocular six degree of freedom visual
relocalization system. We use a Bayesian convolutional neural network to
regress the 6-DOF camera pose from a single RGB image. It is trained in an
end-to-end manner with no need of additional engineering or graph optimisation.
The algorithm can operate indoors and outdoors in real time, taking under 6ms
to compute. It obtains approximately 2m and 6 degrees accuracy for very large
scale outdoor scenes and 0.5m and 10 degrees accuracy indoors. Using a Bayesian
convolutional neural network implementation we obtain an estimate of the
model's relocalization uncertainty and improve state of the art localization
accuracy on a large scale outdoor dataset. We leverage the uncertainty measure
to estimate metric relocalization error and to detect the presence or absence
of the scene in the input image. We show that the model's uncertainty is caused
by images being dissimilar to the training dataset in either pose or
appearance. | ['Roberto Cipolla', 'Alex Kendall'] | 2015-09-19 | null | null | null | null | ['camera-relocalization'] | ['computer-vision'] | [ 1.65182769e-01 1.60790503e-01 1.06560625e-01 -5.24751067e-01
-7.57850766e-01 -8.26866984e-01 4.97693628e-01 -1.11699410e-01
-6.15424812e-01 6.31233037e-01 -1.25440195e-01 -3.24639171e-01
-1.20026484e-01 -4.44119215e-01 -1.30876505e+00 -3.93666834e-01
8.16666260e-02 6.35674536e-01 1.49173275e-01 2.46008962e-01
1.32942408e-01 6.42664313e-01 -1.12248325e+00 -6.78646982e-01
2.87670374e-01 1.18778229e+00 1.34135410e-01 1.20345652e+00
8.49295199e-01 6.51833832e-01 -4.86979336e-01 2.15223998e-01
6.28286242e-01 2.54531115e-01 -4.36749458e-01 2.46987984e-01
1.05154169e+00 -5.35640597e-01 -5.98729908e-01 9.32690203e-01
5.11397541e-01 1.21757381e-01 4.64571446e-01 -9.54387128e-01
-2.22099766e-01 -3.31969708e-01 -4.02105302e-01 -3.38632837e-02
7.43336022e-01 2.20860094e-01 5.41620195e-01 -7.62062013e-01
4.69585121e-01 9.24619734e-01 9.91314411e-01 -2.02568583e-02
-1.31459260e+00 -4.11478162e-01 -6.74459757e-03 -2.15756088e-01
-1.97590935e+00 -5.54599226e-01 4.35952038e-01 -6.62624061e-01
1.04658186e+00 -4.07729521e-02 5.95227838e-01 8.70559156e-01
3.97142142e-01 1.59838185e-01 9.62308168e-01 -3.15097898e-01
3.44116300e-01 -2.29492128e-01 -2.85923362e-01 8.98324192e-01
3.40047687e-01 3.03150415e-01 -5.32558084e-01 9.14964527e-02
8.97524536e-01 -1.70771331e-02 -2.22426862e-01 -9.91213739e-01
-1.25061297e+00 4.38354194e-01 1.04219186e+00 -4.61044401e-01
4.51141894e-02 8.84988546e-01 -4.26584631e-01 -4.56727296e-02
2.59277463e-01 4.01307464e-01 -5.04024148e-01 -1.44709453e-01
-8.34973395e-01 -1.30596042e-01 5.40059924e-01 1.29514480e+00
1.08832276e+00 -9.98562388e-03 7.05947042e-01 2.12192625e-01
6.74571574e-01 8.92067790e-01 9.35443044e-02 -1.32553303e+00
2.08727151e-01 2.91162431e-01 4.62803721e-01 -1.10294199e+00
-5.93185246e-01 -3.41315687e-01 -5.55468202e-01 3.74217957e-01
3.43058467e-01 -1.68186620e-01 -1.14417672e+00 1.33828437e+00
4.01742339e-01 1.39855489e-01 -2.09518895e-01 1.10215735e+00
4.12070096e-01 2.89572001e-01 -7.09623933e-01 4.05343026e-01
7.89910614e-01 -6.08952343e-01 -6.57859966e-02 -7.59064674e-01
3.71143490e-01 -8.59600544e-01 5.14868557e-01 4.28715825e-01
-5.97189426e-01 -4.69533086e-01 -1.48316264e+00 -1.63186267e-01
-1.32537663e-01 3.24532896e-01 4.41328228e-01 6.04196072e-01
-1.29778278e+00 4.67293292e-01 -1.12433112e+00 -4.48830724e-01
5.28690889e-02 6.29585862e-01 -6.38609231e-01 -2.58129060e-01
-5.35360992e-01 9.25255477e-01 2.69224048e-01 1.77149951e-01
-9.78468239e-01 -5.96966922e-01 -1.26740229e+00 -3.93302351e-01
3.78241211e-01 -7.98747003e-01 1.18987429e+00 -6.22539282e-01
-1.47148728e+00 6.37134433e-01 -1.90263718e-01 -4.48270738e-01
5.44883251e-01 -5.96095920e-01 -1.24001466e-02 4.60551679e-02
2.70080626e-01 7.91034222e-01 8.69941652e-01 -1.26631927e+00
-5.07437348e-01 -6.21656001e-01 3.54210697e-02 3.91005069e-01
4.95366216e-01 -4.92949367e-01 -6.62344992e-01 -1.98992431e-01
8.97996902e-01 -1.33330405e+00 -4.41952944e-01 2.95303702e-01
-4.51759994e-01 6.37355328e-01 6.55481815e-01 -4.58372831e-01
1.56775609e-01 -1.94725633e+00 -3.50999534e-02 3.71493667e-01
2.06442863e-01 -4.61591989e-01 2.46024519e-01 3.89204100e-02
1.14624679e-01 -1.71997115e-01 3.91815268e-02 -6.98163807e-01
-6.51763007e-02 3.00559878e-01 -7.68507412e-03 1.25948501e+00
-1.46900296e-01 5.31065762e-01 -9.59452748e-01 2.33399887e-02
7.49618351e-01 7.01467276e-01 -4.84092742e-01 3.02540153e-01
-3.86666879e-02 7.59836912e-01 1.19160920e-01 7.93552995e-01
7.31463134e-01 -1.65816978e-01 3.53590609e-03 -2.47229293e-01
9.04406160e-02 3.46401066e-01 -1.41888368e+00 2.20939064e+00
-6.58525586e-01 1.10898256e+00 9.19235572e-02 -1.79030776e-01
8.91975522e-01 -1.61535308e-01 2.77428746e-01 -4.64436024e-01
1.87461525e-01 5.39622903e-02 -5.15046299e-01 1.75199628e-01
8.21213484e-01 9.63479429e-02 -1.80256546e-01 1.68525219e-01
9.67196077e-02 -6.13344312e-01 -4.38526899e-01 1.88700974e-01
1.21081018e+00 5.74539125e-01 2.82347828e-01 -1.87233254e-01
-1.03370041e-01 -9.34879482e-02 4.70488042e-01 6.93744302e-01
-2.45814249e-02 9.08172905e-01 -8.58043507e-02 -6.85181975e-01
-9.52353835e-01 -1.38911343e+00 5.13409302e-02 4.49828029e-01
6.91213012e-01 -3.24523181e-01 -3.55243474e-01 -3.31280619e-01
9.15650278e-02 2.69066662e-01 -3.87360811e-01 -1.28540665e-01
-1.15592606e-01 -4.00532305e-01 3.14225554e-01 7.10383713e-01
4.68613207e-01 -3.23990807e-02 -9.50842738e-01 -5.66636696e-02
-6.52915612e-02 -1.41759324e+00 -2.62187570e-01 4.20323640e-01
-4.84548807e-01 -1.12820172e+00 -2.29767740e-01 -2.73590982e-01
9.61471319e-01 4.18489844e-01 1.16597939e+00 -4.01271462e-01
-4.38961089e-01 7.08750904e-01 -6.84801582e-03 -1.59729004e-01
1.37509882e-01 -1.95071548e-01 6.09384954e-01 -2.68017173e-01
1.48217514e-01 -6.06657565e-01 -6.44513190e-01 6.03682578e-01
-3.02053064e-01 -1.33282661e-01 3.48566741e-01 4.93390858e-01
7.42384553e-01 -5.05362265e-02 -6.60942912e-01 -1.82010725e-01
-1.05883040e-01 -2.38333955e-01 -1.57631838e+00 -2.13964105e-01
-6.89677298e-01 1.93448231e-01 2.34041438e-01 -1.55939221e-01
-4.87032831e-01 8.75933707e-01 3.01008880e-01 -6.74702764e-01
-3.75624925e-01 2.63226628e-01 6.63787425e-02 -7.49690950e-01
8.72177124e-01 -2.09975615e-01 -4.20928329e-01 -3.34040999e-01
4.20688152e-01 4.88018900e-01 8.44132006e-01 -5.91947317e-01
9.67215121e-01 7.55370140e-01 4.69937503e-01 -8.07638407e-01
-6.49414659e-01 -6.58026040e-01 -1.06114411e+00 -3.27487320e-01
8.54913831e-01 -1.47681761e+00 -9.16254163e-01 2.86615163e-01
-1.15640676e+00 -5.56889653e-01 6.11484461e-02 6.94967449e-01
-5.63628376e-01 1.89400285e-01 2.92741135e-02 -7.89831996e-01
1.52267069e-01 -1.16614592e+00 1.52582514e+00 1.99340671e-01
-7.31549188e-02 -6.94522500e-01 6.83892369e-02 2.48128548e-01
8.73180404e-02 4.12754089e-01 -4.98911962e-02 2.35356599e-01
-1.16894126e+00 -4.65872318e-01 -3.51182818e-01 1.46889746e-01
-8.85625035e-02 -5.71006797e-02 -1.06749201e+00 -6.03043020e-01
-2.48511165e-01 -3.37875545e-01 6.30073607e-01 4.26675618e-01
5.47592461e-01 -6.80141076e-02 -2.73907840e-01 1.17316186e+00
1.80827105e+00 -1.13022655e-01 3.54944497e-01 3.31685185e-01
1.02633202e+00 2.01673537e-01 5.78954756e-01 3.39635700e-01
5.62677085e-01 8.30385447e-01 1.03766716e+00 -5.08648437e-03
6.20329715e-02 -4.26596314e-01 2.06008092e-01 2.59906977e-01
1.94096759e-01 -7.25634694e-02 -1.22057855e+00 5.10436177e-01
-1.78788304e+00 -1.66366771e-01 -7.19913170e-02 2.56462002e+00
2.38466099e-01 2.10942760e-01 -3.39102119e-01 -2.18935698e-01
4.20578778e-01 7.53731653e-02 -5.89706779e-01 -1.07259393e-01
2.08000317e-01 -1.55896500e-01 1.56629515e+00 1.14048779e+00
-1.09583533e+00 9.38203752e-01 6.37850857e+00 7.71047398e-02
-1.07453620e+00 -2.49687672e-01 3.45641702e-01 -1.60126776e-01
1.31219327e-01 4.35769469e-01 -8.12753975e-01 1.08264223e-01
9.08437729e-01 5.12099147e-01 8.11229050e-01 1.05094385e+00
-8.85589495e-02 -7.12827444e-01 -1.13295877e+00 1.38908982e+00
2.54898459e-01 -1.16565359e+00 -6.46799088e-01 3.27805042e-01
7.21042752e-01 7.97578931e-01 -1.24496482e-01 -2.64887661e-01
7.90559411e-01 -1.10917294e+00 1.00270951e+00 6.62885010e-01
9.74729896e-01 -8.02684963e-01 5.68499207e-01 5.12898445e-01
-1.12265122e+00 2.03664362e-01 -5.81387699e-01 -3.41804683e-01
2.70400196e-02 6.21046484e-01 -1.34638846e+00 4.54674661e-01
9.18392897e-01 6.38762593e-01 -8.12992334e-01 1.06078172e+00
-5.53959548e-01 1.57640055e-01 -9.40440238e-01 1.27370343e-01
4.09043469e-02 -1.02415659e-01 5.20537078e-01 6.46542728e-01
3.94327939e-01 -1.70501754e-01 2.82196164e-01 6.55537546e-01
-1.22678690e-01 -6.99300528e-01 -7.61255741e-01 4.37799066e-01
3.96726191e-01 9.82663155e-01 -5.38191020e-01 4.57634553e-02
-1.10061057e-02 1.21066475e+00 1.61678120e-01 3.77554536e-01
-6.40410602e-01 -4.30415928e-01 6.62276745e-01 1.59189224e-01
4.27136719e-01 -8.30724597e-01 -2.94027865e-01 -1.19539464e+00
1.79944769e-01 -2.66465098e-01 -2.44986147e-01 -1.44222856e+00
-7.09997475e-01 4.14675146e-01 -8.43563080e-02 -1.05696332e+00
-5.15902936e-01 -8.76668215e-01 7.57187903e-02 9.15587783e-01
-1.25234509e+00 -1.00445712e+00 -7.62104273e-01 5.10114193e-01
-2.00335942e-02 9.72418413e-02 8.20350945e-01 -3.83295603e-02
-4.78558317e-02 2.64039785e-01 2.05714315e-01 2.22139508e-01
8.10608208e-01 -1.48904514e+00 5.72594643e-01 1.24702990e+00
2.37612471e-01 5.87102294e-01 8.56143534e-01 -5.47152042e-01
-1.66572642e+00 -1.12231743e+00 5.89786172e-01 -1.08771074e+00
5.71215332e-01 -6.10603511e-01 5.10439910e-02 1.03185356e+00
-1.04609676e-01 6.09132230e-01 1.13201723e-01 1.35111645e-01
-5.71449757e-01 -2.87804365e-01 -1.18097389e+00 4.71718341e-01
9.62023914e-01 -8.05130839e-01 -1.72020540e-01 4.57314342e-01
7.63956428e-01 -1.05940723e+00 -7.35596001e-01 4.72370982e-01
6.56567752e-01 -1.00887072e+00 1.17304397e+00 1.12005048e-01
-1.36320874e-01 -7.67730713e-01 -8.05045366e-01 -1.08291626e+00
-2.06531465e-01 -7.11182177e-01 -9.45743397e-02 5.97243905e-01
1.04072064e-01 -4.61899728e-01 9.75433409e-01 5.69028795e-01
-1.23994406e-02 -2.74208456e-01 -1.32030725e+00 -8.79205048e-01
-6.45454288e-01 -7.59523034e-01 2.44713306e-01 4.40687388e-01
-6.75091922e-01 4.17559892e-01 -4.47319448e-01 9.74022746e-01
9.75234210e-01 -7.37358928e-02 1.33221638e+00 -1.01396525e+00
-4.23398316e-01 1.83552191e-01 -1.04115951e+00 -1.41203344e+00
8.59561935e-03 -3.34893942e-01 4.62894291e-01 -1.48787105e+00
-3.33673626e-01 -2.42949262e-01 -4.84384503e-03 3.02201629e-01
3.25006247e-01 6.39040470e-01 -1.11735545e-01 4.70147319e-02
-8.43785942e-01 3.32021266e-01 3.60145211e-01 -7.92224929e-02
1.42456055e-01 6.58213049e-02 -2.79303551e-01 9.63478804e-01
5.79973042e-01 -5.55331528e-01 -3.49053741e-01 -8.28042924e-01
6.97879374e-01 -1.53234363e-01 7.20752358e-01 -1.56917250e+00
3.01211536e-01 4.92667928e-02 9.25123096e-01 -6.19233370e-01
7.82832563e-01 -1.15578079e+00 1.75855950e-01 2.73769468e-01
1.06084600e-01 1.65679529e-01 2.38009930e-01 8.42985272e-01
2.70899147e-01 1.19555138e-01 8.34037483e-01 -1.23295277e-01
-8.06968212e-01 2.58665800e-01 -1.81147084e-01 -2.64943749e-01
8.36005032e-01 -3.34033817e-01 -4.80763502e-02 -7.95693874e-01
-5.14774680e-01 -7.55508021e-02 1.15996695e+00 2.25951612e-01
6.53416336e-01 -1.26794696e+00 -1.85777634e-01 4.23177034e-01
4.40521538e-01 4.85297590e-01 -2.04225808e-01 5.48233211e-01
-1.13277280e+00 3.80716830e-01 1.36537462e-01 -1.24470484e+00
-1.06766367e+00 4.37269777e-01 6.43689334e-01 2.76793033e-01
-2.31338874e-01 9.27401781e-01 -2.51764864e-01 -7.18422592e-01
3.62515688e-01 -5.15599012e-01 7.18076289e-01 -6.83425963e-01
1.97708830e-01 2.18719289e-01 1.33083919e-02 -8.66501570e-01
-6.93530679e-01 8.02864969e-01 4.58258390e-01 -5.38240016e-01
1.04809034e+00 -5.43603122e-01 4.08912599e-02 3.35503042e-01
1.34782159e+00 2.56474882e-01 -1.88927639e+00 -1.51499793e-01
-2.67392933e-01 -8.32084954e-01 5.38984954e-01 -7.16435015e-01
-5.54583013e-01 8.06226850e-01 9.77249384e-01 -4.37238425e-01
6.97478414e-01 -4.05271202e-02 3.05379599e-01 8.63177836e-01
8.27156544e-01 -1.07708466e+00 -4.46995087e-02 7.77437449e-01
5.08420348e-01 -1.49194551e+00 5.18227994e-01 -2.24627722e-02
-1.21596999e-01 1.03927553e+00 4.58473980e-01 -4.42601174e-01
5.82639813e-01 3.10473859e-01 1.81695655e-01 -1.99272837e-02
-1.04445577e-01 1.03709750e-01 4.01798576e-01 8.32469165e-01
-7.26601109e-02 1.27312526e-01 8.27287912e-01 -2.58796662e-01
-3.49790394e-01 -3.30567867e-01 4.36018348e-01 9.42711353e-01
-3.96946281e-01 -5.41574955e-01 -6.13164127e-01 -1.26023874e-01
1.39035001e-01 -1.03126794e-01 -5.05226851e-01 6.27651930e-01
1.42010972e-01 1.01295078e+00 5.40895797e-02 -8.12816262e-01
7.28507265e-02 -3.84073436e-01 8.48453164e-01 -4.96986568e-01
1.49587259e-01 7.51393884e-02 3.19544226e-02 -1.17340899e+00
-2.97420979e-01 -4.48522419e-01 -1.02258098e+00 -2.71953464e-01
-3.65580052e-01 -4.24463063e-01 1.35183394e+00 9.62712228e-01
4.77397680e-01 2.23906711e-01 5.13031960e-01 -1.30144954e+00
-1.23567328e-01 -7.18013108e-01 -3.30480665e-01 -3.31709117e-01
7.59026706e-01 -5.20809114e-01 -3.62221897e-01 -1.18938349e-01] | [7.657641887664795, -2.270331859588623] |
2ac93ef3-5045-4764-966c-4be99fa39668 | cqe-a-comprehensive-quantity-extractor | 2305.08853 | null | https://arxiv.org/abs/2305.08853v1 | https://arxiv.org/pdf/2305.08853v1.pdf | CQE: A Comprehensive Quantity Extractor | Quantities are essential in documents to describe factual information. They are ubiquitous in application domains such as finance, business, medicine, and science in general. Compared to other information extraction approaches, interestingly only a few works exist that describe methods for a proper extraction and representation of quantities in text. In this paper, we present such a comprehensive quantity extraction framework from text data. It efficiently detects combinations of values and units, the behavior of a quantity (e.g., rising or falling), and the concept a quantity is associated with. Our framework makes use of dependency parsing and a dictionary of units, and it provides for a proper normalization and standardization of detected quantities. Using a novel dataset for evaluation, we show that our open source framework outperforms other systems and -- to the best of our knowledge -- is the first to detect concepts associated with identified quantities. The code and data underlying our framework are available at https://github.com/vivkaz/CQE. | ['Michael Gertz', 'Philip Göldner', 'Vivian Kazakova', 'Satya Almasian'] | 2023-05-15 | null | null | null | null | ['dependency-parsing'] | ['natural-language-processing'] | [-1.68892398e-01 -2.47404784e-01 -3.57814342e-01 -2.21707702e-01
-5.89331150e-01 -7.61674702e-01 9.52469468e-01 1.06621552e+00
-4.69533890e-01 7.51319706e-01 1.56617373e-01 -4.03201610e-01
-2.11709023e-01 -1.03583705e+00 -3.53832334e-01 -4.30745721e-01
-8.16749223e-03 4.77428973e-01 2.28010237e-01 -2.89505988e-01
6.35674357e-01 6.21671259e-01 -1.68184352e+00 9.10545290e-02
6.08029842e-01 1.27125084e+00 -1.44400254e-01 5.37784815e-01
-6.12680197e-01 1.03817022e+00 -8.02455604e-01 -4.99275327e-01
2.28401244e-01 -3.36688906e-01 -7.54606307e-01 -7.99754113e-02
1.37327313e-01 -4.13890809e-01 -8.97981375e-02 8.49066913e-01
4.17540222e-02 3.51691693e-02 8.72426808e-01 -1.32862115e+00
-5.09736180e-01 8.71851981e-01 -4.29340959e-01 6.35963202e-01
7.28956997e-01 -1.38185546e-01 1.38828230e+00 -6.53262973e-01
7.57869959e-01 9.36526477e-01 3.13082248e-01 2.77345389e-01
-8.77857864e-01 -5.14864028e-01 -1.03583097e-01 -1.86627619e-02
-1.22493112e+00 -4.35472935e-01 6.00050330e-01 -4.95510817e-01
1.25147867e+00 4.57915157e-01 6.34931207e-01 7.65597343e-01
1.85401931e-01 6.80094779e-01 1.07483339e+00 -7.31988013e-01
1.89062506e-01 2.30072528e-01 4.89891559e-01 3.36067468e-01
8.19448531e-01 -4.25148100e-01 -5.73956013e-01 -1.14003308e-01
5.68051159e-01 -5.43685965e-02 -1.81912221e-02 -2.19778605e-02
-1.28397775e+00 5.71701288e-01 -2.75362730e-01 8.07466567e-01
-5.25909364e-01 2.86347985e-01 4.55146372e-01 2.57355660e-01
6.39210880e-01 4.36497062e-01 -6.76949322e-01 -5.09105980e-01
-9.33042765e-01 4.64874715e-01 1.36943173e+00 1.19562376e+00
6.38696611e-01 -5.29132068e-01 -1.14492349e-01 3.97173464e-01
-3.90876923e-03 4.29950386e-01 4.86399651e-01 -6.53690577e-01
7.63144553e-01 8.88319612e-01 4.12140340e-01 -1.10384023e+00
-6.15289211e-01 9.95546207e-02 -3.79230827e-01 -3.83099258e-01
5.96131682e-01 -6.42923499e-03 -4.78580356e-01 1.24431276e+00
4.43797678e-01 -1.42073363e-01 -7.18318895e-02 5.63166142e-01
1.09692812e+00 4.26573694e-01 1.85502991e-01 -2.80402184e-01
1.69139218e+00 -1.07882716e-01 -1.14530313e+00 1.10480130e-01
7.93603480e-01 -8.12092066e-01 8.52735102e-01 2.99577445e-01
-9.76097882e-01 3.86390463e-02 -7.97893286e-01 -4.03370231e-01
-1.07785439e+00 -1.63606942e-01 9.38534617e-01 7.52596438e-01
-6.45931482e-01 8.95370483e-01 -7.56221354e-01 -3.87010723e-01
2.70882636e-01 2.01728940e-01 -2.37919554e-01 4.38187718e-01
-1.24844146e+00 8.89552355e-01 4.65191066e-01 -2.66262144e-01
1.24380663e-01 -6.29909158e-01 -1.02987921e+00 4.72270921e-02
6.03076220e-01 -3.51738393e-01 1.54111946e+00 -3.66875947e-01
-1.12473464e+00 8.99258792e-01 -6.68347403e-02 -4.98033792e-01
2.88089812e-01 -2.19563991e-01 -5.11995912e-01 2.22148150e-01
3.57631147e-01 1.10339113e-01 4.18502003e-01 -7.75221467e-01
-8.68789077e-01 -2.95813233e-01 2.47210607e-01 -1.30853951e-01
-4.01723146e-01 5.67753911e-01 -3.37489009e-01 -6.13496900e-01
3.96197699e-02 -3.38978171e-01 3.49275954e-02 -4.27365988e-01
-3.88994932e-01 -5.37760496e-01 2.67175108e-01 -6.52747869e-01
1.69370282e+00 -1.77899539e+00 -1.18503816e-01 3.03943306e-01
4.26370740e-01 -2.68296272e-01 4.27914560e-01 8.31374109e-01
3.00817583e-02 5.28161049e-01 -3.66351902e-01 -1.89560086e-01
3.78666461e-01 1.43966451e-01 -1.68788224e-01 5.79474390e-01
3.52579296e-01 9.92892444e-01 -1.00279200e+00 -7.22282588e-01
3.77480209e-01 3.41108859e-01 -2.02035233e-02 4.01805937e-02
-3.55966508e-01 -1.10302843e-01 -6.02061927e-01 7.74686575e-01
5.70247769e-01 -2.54874110e-01 2.99844533e-01 -1.73842944e-02
-3.93622935e-01 7.54991829e-01 -1.45466304e+00 1.31162333e+00
-3.60875666e-01 5.16935647e-01 -2.44413272e-01 -8.81331563e-01
8.01936746e-01 3.24422270e-01 8.92301202e-01 -8.35828483e-01
4.99336898e-01 4.92660969e-01 -1.97937846e-01 -4.64134544e-01
9.66396630e-01 -1.99305024e-02 -5.07536352e-01 5.00055432e-01
1.71697363e-01 -6.41361833e-01 1.09557056e+00 3.55768621e-01
1.18849564e+00 1.50114194e-01 1.08198714e+00 -2.90842593e-01
7.80673504e-01 8.51015300e-02 2.74720788e-01 4.42457408e-01
-5.09625226e-02 1.80444747e-01 8.94689918e-01 -1.12967417e-01
-1.10048723e+00 -6.08156741e-01 -4.26234752e-01 8.14755499e-01
-8.67509395e-02 -8.92574430e-01 -5.72159648e-01 -2.78444499e-01
3.57767880e-01 7.84228444e-01 -6.26609564e-01 5.48726320e-01
-5.50314724e-01 -5.92515469e-01 1.73638642e-01 4.18978870e-01
-2.03022547e-02 -1.23692632e+00 -7.67030001e-01 3.77610534e-01
-9.09766033e-02 -1.41210163e+00 -1.40532061e-01 2.59779662e-01
-7.07462609e-01 -1.23397207e+00 -1.94998294e-01 -2.33275652e-01
3.02607477e-01 -1.58591136e-01 1.45014095e+00 2.56910384e-01
-4.59914625e-01 4.76350695e-01 -6.15348339e-01 -6.94452643e-01
-4.69363153e-01 3.02244369e-02 -1.36583388e-01 -4.21117336e-01
7.92991221e-01 -3.00061762e-01 -2.96371013e-01 -1.68943167e-01
-1.04613829e+00 -3.06702495e-01 3.50404888e-01 3.19861412e-01
2.75519907e-01 1.78642124e-01 3.16815555e-01 -1.17247534e+00
1.10068786e+00 -7.05358684e-01 -7.76124895e-01 8.29615965e-02
-7.33146429e-01 1.24308825e-01 6.48785710e-01 -1.30119458e-01
-6.50292635e-01 -2.75848001e-01 -7.56501704e-02 3.16592366e-01
-4.53749239e-01 5.68080664e-01 -1.46338388e-01 5.01347005e-01
4.82067764e-01 -1.43670395e-01 -4.97885525e-01 -6.52528226e-01
6.11767828e-01 8.23022783e-01 4.15847361e-01 -7.53989697e-01
5.60921669e-01 4.29949403e-01 1.42365947e-01 -5.70999444e-01
-7.89880931e-01 -1.25998771e+00 -1.09807611e+00 -8.17283690e-02
5.21808326e-01 -5.73472261e-01 -7.02179313e-01 3.27033401e-01
-1.14557612e+00 2.44334303e-02 -4.41966563e-01 3.14908147e-01
-4.37459737e-01 3.83637607e-01 -7.28488743e-01 -1.08356655e+00
-5.07123113e-01 -6.61147833e-01 1.06909215e+00 2.10981458e-01
-3.83833140e-01 -1.10068488e+00 -1.34274559e-02 2.89762411e-02
2.50385731e-01 4.54178780e-01 4.30135041e-01 -6.68410599e-01
-2.69243091e-01 -4.34589446e-01 -2.56497443e-01 1.50918946e-01
4.59138870e-01 3.59116167e-01 -5.70734859e-01 2.91103661e-01
-9.57571417e-02 3.75383608e-02 6.60286367e-01 1.35798514e-01
8.44837725e-01 -5.98320246e-01 -3.25738549e-01 7.59464875e-02
1.43632638e+00 7.66757876e-02 5.14641881e-01 7.59996116e-01
4.81233895e-01 7.62760878e-01 8.10500085e-01 9.51288640e-01
4.11002010e-01 3.90430063e-01 7.20921233e-02 2.46642232e-01
2.12576151e-01 1.77123606e-01 9.93870795e-02 8.05944920e-01
-2.01241076e-01 -3.19561452e-01 -1.08087027e+00 7.88238049e-01
-1.63030577e+00 -1.08310318e+00 -4.17251825e-01 1.89956653e+00
1.25670815e+00 3.85535121e-01 2.93609589e-01 4.47473645e-01
3.85128945e-01 1.31764263e-01 -4.82178815e-02 -5.67027390e-01
1.12039611e-01 6.34756923e-01 7.51982868e-01 2.65481979e-01
-1.25454748e+00 9.04218853e-01 6.01311207e+00 6.69258893e-01
-7.20208645e-01 -7.76043385e-02 2.08452955e-01 -8.08250843e-05
-3.57757956e-01 -3.17309680e-03 -9.73984599e-01 4.66745406e-01
1.32959163e+00 -5.32072961e-01 1.20049566e-01 6.97572827e-01
3.07915390e-01 -3.09579372e-01 -1.11770892e+00 7.95073926e-01
-7.76445493e-02 -1.19258153e+00 -1.04004852e-01 3.35065648e-02
2.78110474e-01 -2.65831947e-01 -2.82816291e-01 2.83344649e-03
1.85209721e-01 -9.12808418e-01 9.76519465e-01 5.02307653e-01
5.86970985e-01 -5.77488899e-01 9.79657829e-01 2.01327190e-01
-1.25918818e+00 1.86951220e-01 -2.10545138e-01 -2.84159243e-01
4.25770618e-02 1.04204452e+00 -6.56242192e-01 7.84425020e-01
6.48495495e-01 8.26451898e-01 -6.54624343e-01 7.79639781e-01
-4.43154782e-01 5.58935523e-01 -3.88745934e-01 -5.27851105e-01
4.36540060e-02 -4.83724713e-01 2.32292637e-01 1.64466453e+00
2.73056865e-01 2.25727335e-01 -1.72892362e-01 7.24597156e-01
-1.84971437e-01 7.30597377e-01 -4.00177240e-01 -4.70971406e-01
5.15829384e-01 1.42183065e+00 -1.03371561e+00 -5.99963725e-01
-6.70357466e-01 4.78149801e-01 3.09590757e-01 -3.90506014e-02
-6.05909348e-01 -5.63486040e-01 5.72948217e-01 3.57435197e-01
2.23554388e-01 -4.59601343e-01 -4.91888255e-01 -1.27003944e+00
4.27309930e-01 -7.00808525e-01 5.17556131e-01 -4.42188710e-01
-1.31886947e+00 3.33580732e-01 3.41305226e-01 -1.13937557e+00
-5.26962817e-01 -9.66834664e-01 -1.38627231e-01 6.86839938e-01
-1.44722891e+00 -6.47343934e-01 -3.15548003e-01 2.64182180e-01
2.20014304e-02 1.64639980e-01 7.85213947e-01 2.32484758e-01
-4.37739819e-01 9.68505517e-02 -7.73226796e-03 3.78185570e-01
5.04880965e-01 -1.58032286e+00 7.57609665e-01 8.37540150e-01
4.09243524e-01 7.70744205e-01 9.15059209e-01 -6.80544496e-01
-1.25334096e+00 -5.33706129e-01 1.32544672e+00 -7.01624036e-01
1.47177184e+00 -5.41648209e-01 -7.57860959e-01 5.74815035e-01
1.19416714e-01 -2.38362402e-01 6.77801311e-01 1.56281337e-01
-5.43900728e-01 6.44720197e-02 -1.17423332e+00 3.93856227e-01
8.10861170e-01 -4.96081829e-01 -6.81283653e-01 5.28102994e-01
6.43543303e-01 -5.60349166e-01 -1.30248809e+00 8.39703232e-02
5.17544329e-01 -9.38865066e-01 7.60376334e-01 -4.43072677e-01
5.32456815e-01 -1.13785945e-01 -6.54417798e-02 -8.52566600e-01
8.79459828e-02 -5.87756395e-01 -5.15679181e-01 1.40687656e+00
5.19801140e-01 -7.72634685e-01 4.28260088e-01 8.35410178e-01
1.70126900e-01 -5.98574102e-01 -7.63285756e-01 -8.77683580e-01
1.80720583e-01 -7.31671214e-01 8.50150943e-01 9.85161424e-01
3.50508392e-01 9.74409655e-02 -1.09612625e-02 -1.18683845e-01
4.93842125e-01 2.73339033e-01 6.56970561e-01 -1.21339262e+00
4.31486852e-02 -7.92947412e-01 -5.75071812e-01 -5.01827121e-01
-4.14914265e-03 -6.98882282e-01 -2.58331418e-01 -1.95321000e+00
6.73963428e-02 -4.01258439e-01 -1.81044191e-01 5.29236436e-01
-5.96233346e-02 3.78452688e-02 5.92040829e-02 3.64241630e-01
-5.61956167e-01 -1.78308059e-02 9.59090829e-01 4.95796390e-02
-2.50405461e-01 -1.33109838e-01 -7.96878815e-01 4.37119514e-01
1.48715651e+00 -6.38486266e-01 1.43425062e-01 1.52060300e-01
5.83616555e-01 -1.69658244e-01 1.48423195e-01 -6.51458979e-01
9.21067372e-02 -4.38050419e-01 6.38457686e-02 -7.51711011e-01
-5.40920198e-02 -8.16396534e-01 -1.23266511e-01 5.80123514e-02
7.93701336e-02 2.03782290e-01 2.58123666e-01 7.43520483e-02
-2.23576441e-01 -6.42715514e-01 2.54827589e-01 -2.01061413e-01
-8.58577967e-01 1.77062843e-02 -1.70862228e-01 1.02797605e-01
9.15152490e-01 1.77524686e-01 -6.38984799e-01 -1.83174506e-01
-1.92239031e-01 8.39590356e-02 4.86656189e-01 3.52745473e-01
2.82704026e-01 -9.55417633e-01 -6.49333298e-01 -2.97706217e-01
2.36924589e-01 -1.36765927e-01 -3.99250418e-01 6.92157626e-01
-8.23061049e-01 6.08639240e-01 5.31445369e-02 -8.44156146e-02
-9.13118124e-01 7.37253428e-01 1.35540098e-01 -4.78650898e-01
-6.93156064e-01 3.56513232e-01 -4.55094099e-01 -2.25557312e-01
-1.84614986e-01 -1.03204191e+00 -4.21689421e-01 4.79195744e-01
6.10860169e-01 3.33203852e-01 3.23057234e-01 -7.22596943e-01
-6.92757428e-01 4.21794236e-01 9.50147510e-02 -2.57012069e-01
1.41806078e+00 -4.16062921e-02 -5.81744969e-01 8.89312267e-01
1.04022944e+00 3.91461313e-01 -5.17287850e-01 -1.49589181e-01
7.52795398e-01 -3.77208054e-01 -2.22768486e-01 -5.68777323e-01
-6.35445952e-01 3.51458758e-01 -1.97083950e-01 7.73462415e-01
1.04669452e+00 1.87877327e-01 5.78924239e-01 3.67477119e-01
3.43223393e-01 -1.33809662e+00 -3.05514544e-01 4.83785331e-01
8.13115895e-01 -1.09328735e+00 5.59104443e-01 -6.62316859e-01
-3.33626598e-01 1.19649673e+00 5.18531092e-02 -3.69906984e-02
9.05358672e-01 7.76246846e-01 9.59518403e-02 -4.68716294e-01
-6.11639261e-01 -6.62029505e-01 4.06699121e-01 2.20669076e-01
9.02430415e-01 2.22968414e-01 -8.85869741e-01 5.56335628e-01
-6.86375082e-01 -5.32276854e-02 8.09719861e-01 1.21098673e+00
-4.35332417e-01 -1.31351519e+00 -4.65220958e-01 7.59726822e-01
-8.23693752e-01 -4.16301817e-01 -5.05776942e-01 1.18540239e+00
4.44053262e-02 1.16708708e+00 2.89287865e-01 1.11618079e-01
5.15555084e-01 1.89181030e-01 6.35861516e-01 -7.99374700e-01
-7.75751948e-01 4.48943377e-02 2.67026722e-01 -5.81240654e-01
-8.70275080e-01 -1.01665831e+00 -1.60435128e+00 -4.99138862e-01
-1.93346962e-01 1.51403606e-01 7.63033271e-01 8.48050654e-01
-1.15993842e-01 4.48258430e-01 2.16819808e-01 -3.39028060e-01
-2.19859451e-01 -1.24520040e+00 -1.07334971e+00 6.63559556e-01
1.51352778e-01 -7.66384900e-01 -6.30280197e-01 1.92166358e-01] | [9.428722381591797, 8.730173110961914] |
9f399398-0445-4ea4-9d56-5a8a53b5dfad | a-variational-perspective-on-solving-inverse | 2305.04391 | null | https://arxiv.org/abs/2305.04391v1 | https://arxiv.org/pdf/2305.04391v1.pdf | A Variational Perspective on Solving Inverse Problems with Diffusion Models | Diffusion models have emerged as a key pillar of foundation models in visual domains. One of their critical applications is to universally solve different downstream inverse tasks via a single diffusion prior without re-training for each task. Most inverse tasks can be formulated as inferring a posterior distribution over data (e.g., a full image) given a measurement (e.g., a masked image). This is however challenging in diffusion models since the nonlinear and iterative nature of the diffusion process renders the posterior intractable. To cope with this challenge, we propose a variational approach that by design seeks to approximate the true posterior distribution. We show that our approach naturally leads to regularization by denoising diffusion process (RED-Diff) where denoisers at different timesteps concurrently impose different structural constraints over the image. To gauge the contribution of denoisers from different timesteps, we propose a weighting mechanism based on signal-to-noise-ratio (SNR). Our approach provides a new variational perspective for solving inverse problems with diffusion models, allowing us to formulate sampling as stochastic optimization, where one can simply apply off-the-shelf solvers with lightweight iterates. Our experiments for image restoration tasks such as inpainting and superresolution demonstrate the strengths of our method compared with state-of-the-art sampling-based diffusion models. | ['Arash Vahdat', 'Jan Kautz', 'Jiaming Song', 'Morteza Mardani'] | 2023-05-07 | null | null | null | null | ['stochastic-optimization'] | ['methodology'] | [ 3.89774710e-01 4.16663624e-02 2.64113754e-01 1.63343437e-02
-8.92332852e-01 -4.32142317e-01 7.55395949e-01 -5.22991419e-01
-3.07074368e-01 6.19930804e-01 3.80200148e-01 -5.14687486e-02
-2.72385001e-01 -5.61087310e-01 -7.01056719e-01 -1.05147576e+00
3.17642808e-01 3.63003165e-01 1.12977549e-01 -2.74176389e-01
2.02078640e-01 3.81515682e-01 -1.15827048e+00 3.44642878e-01
9.22728658e-01 7.55830586e-01 6.48720920e-01 7.31853843e-01
2.29005273e-02 8.03553283e-01 -4.12335962e-01 -6.71646968e-02
3.92580122e-01 -7.80070961e-01 -5.80479324e-01 4.03378874e-01
2.18999475e-01 -6.17292047e-01 -4.88453060e-01 1.17382431e+00
4.72970575e-01 1.45553902e-01 7.13537872e-01 -7.02983201e-01
-9.39413548e-01 1.11889958e-01 -9.64291215e-01 3.69282693e-01
1.55045360e-01 3.42403322e-01 7.88685024e-01 -9.86593723e-01
8.69549394e-01 1.13286757e+00 6.98370218e-01 5.20145953e-01
-2.03081131e+00 -5.66713847e-02 6.27869368e-02 1.15489632e-01
-1.12290347e+00 -7.20430017e-01 7.65468776e-01 -5.22544622e-01
5.33200622e-01 2.29530647e-01 5.81971943e-01 1.06713295e+00
2.53239155e-01 6.73173010e-01 1.45160985e+00 -3.47626865e-01
3.98106247e-01 -2.56693453e-01 -2.82770544e-01 4.64169085e-01
-1.71594501e-01 9.72206295e-02 -7.25193202e-01 -8.64314064e-02
1.07760358e+00 -9.38241258e-02 -6.43544316e-01 -3.50967407e-01
-1.24082363e+00 7.81746089e-01 2.03865334e-01 2.05995627e-02
-6.88048303e-01 2.35046700e-01 -2.72963811e-02 3.72231394e-01
9.05092955e-01 9.74291265e-02 -8.76868740e-02 1.47549540e-01
-1.20504463e+00 4.97641146e-01 6.50644779e-01 5.41936398e-01
8.06054831e-01 1.15027614e-01 -3.21408540e-01 9.06978190e-01
4.91341501e-01 5.10996103e-01 1.47644982e-01 -1.68512690e+00
5.74169308e-02 -2.87753642e-01 3.25852305e-01 -6.47165060e-01
1.70862034e-01 -4.45001811e-01 -1.03945005e+00 7.64833510e-01
5.63840270e-01 -1.27313852e-01 -1.02831364e+00 1.91346681e+00
4.97442544e-01 4.37546343e-01 -1.44816637e-01 1.20938706e+00
3.24214041e-01 7.96012700e-01 -3.65293294e-01 -5.32512248e-01
1.07834268e+00 -8.44972432e-01 -8.25489163e-01 -3.24369282e-01
-2.15937141e-02 -8.81598473e-01 8.45631421e-01 6.26409173e-01
-1.50841248e+00 -2.79153585e-01 -9.22843933e-01 -3.83207500e-01
3.56509864e-01 -3.01765203e-01 2.11513147e-01 3.81479502e-01
-1.26302588e+00 1.07527745e+00 -1.16059446e+00 -1.06406346e-01
3.21031839e-01 3.91849205e-02 -7.26812258e-02 -2.58906573e-01
-7.14918315e-01 9.73247111e-01 -4.20716733e-01 1.69650793e-01
-1.30824995e+00 -9.83987570e-01 -5.03695488e-01 -1.05179161e-01
3.02464396e-01 -9.76042211e-01 1.15315378e+00 -9.00730431e-01
-1.82737255e+00 7.81216621e-01 -3.24153960e-01 -3.42559010e-01
8.84247243e-01 -1.90193266e-01 1.01185612e-01 2.85866231e-01
2.77864665e-01 2.45135427e-01 1.47668338e+00 -1.45446348e+00
-5.06236138e-05 -3.59531939e-01 -5.32911234e-02 1.90275177e-01
-1.36832213e-02 -3.28256935e-02 -4.93149102e-01 -7.79767632e-01
4.05359775e-01 -8.13580990e-01 -4.66033131e-01 5.50296307e-01
-1.78565487e-01 5.03495157e-01 7.95966148e-01 -9.97991800e-01
8.14594924e-01 -2.01327443e+00 9.34621572e-01 4.88520451e-02
5.79407215e-01 -2.21150685e-02 -1.85377568e-01 3.79789412e-01
1.31931538e-02 -2.29413524e-01 -7.32525766e-01 -8.29811513e-01
-7.90531710e-02 4.52872485e-01 -4.70031768e-01 8.42565835e-01
1.38116911e-01 7.12107062e-01 -8.95848215e-01 -2.48447791e-01
3.32212895e-01 9.98173654e-01 -7.47621715e-01 2.72298932e-01
-2.19446138e-01 1.15784311e+00 -3.68046105e-01 1.41043589e-01
8.04331720e-01 -2.88469642e-01 9.48707536e-02 -2.91070163e-01
-2.76912153e-01 3.51512991e-02 -1.36629915e+00 2.12012935e+00
-6.91850126e-01 5.09314537e-01 9.65493798e-01 -1.20734560e+00
3.29271913e-01 3.65147799e-01 4.46558446e-01 -5.87642789e-01
-1.76655710e-01 2.73512453e-01 -1.43210664e-01 -4.40160602e-01
2.48831153e-01 -5.82496881e-01 5.34854531e-01 6.20084584e-01
5.01829460e-02 -5.46241581e-01 2.12380454e-01 2.66399801e-01
1.20217633e+00 5.38708091e-01 -9.00620148e-02 -4.50646013e-01
2.53791809e-01 -1.86203837e-01 5.19422054e-01 8.65921021e-01
5.18648624e-02 1.13626814e+00 3.86758506e-01 -1.55301601e-01
-1.30449164e+00 -1.19937563e+00 -1.78858206e-01 6.44553721e-01
-8.99028629e-02 -1.23046100e-01 -9.58236516e-01 -1.27684399e-01
-1.90025643e-01 4.34192330e-01 -6.53397202e-01 2.09026515e-01
-6.26020968e-01 -9.96265829e-01 9.47190598e-02 4.83733490e-02
4.05657947e-01 -6.35634363e-01 -5.27205646e-01 3.99936140e-01
-3.31567615e-01 -1.10161531e+00 -4.96957541e-01 -1.14423211e-03
-9.71631527e-01 -6.97618544e-01 -1.30723989e+00 -2.98492014e-01
5.78618884e-01 4.01589572e-01 1.13587499e+00 -4.34518121e-02
-2.29487404e-01 5.69875717e-01 7.04179481e-02 1.81825608e-01
-5.68289280e-01 -4.60835487e-01 -4.97003570e-02 4.76327926e-01
-5.62929094e-01 -1.11770856e+00 -1.02430117e+00 1.64555371e-01
-1.16098332e+00 7.85270408e-02 5.16379178e-01 9.84405875e-01
8.02947640e-01 -1.27597183e-01 2.28692710e-01 -7.40619540e-01
8.78429413e-01 -5.24851084e-01 -5.35713792e-01 2.18692068e-02
-4.80610102e-01 2.45425254e-01 4.73588318e-01 -6.68408990e-01
-1.29561079e+00 -7.93763101e-02 -1.85936019e-01 -7.38442421e-01
2.46845871e-01 5.20287573e-01 8.26214328e-02 -2.30349660e-01
6.90523982e-01 4.28510576e-01 3.64041090e-01 -7.00537443e-01
6.82139933e-01 2.33726382e-01 6.29614353e-01 -8.15342963e-01
7.06227005e-01 1.08938670e+00 2.36468866e-01 -1.12926793e+00
-8.83585989e-01 -1.40899673e-01 -3.99556845e-01 -2.83509791e-01
9.54429984e-01 -8.99765313e-01 -5.86324513e-01 5.98092198e-01
-1.24251997e+00 -6.50360227e-01 -5.67234635e-01 1.83533639e-01
-8.21867645e-01 6.43648326e-01 -8.48440588e-01 -8.16815019e-01
-2.13984791e-02 -1.28090322e+00 1.13869798e+00 -1.63150564e-01
-1.41035795e-01 -1.14423776e+00 8.50855187e-02 3.04035336e-01
7.20335901e-01 2.09101915e-01 4.60061640e-01 4.33863014e-01
-8.87434483e-01 3.45118999e-01 -1.59988821e-01 5.64198673e-01
-1.35685727e-01 -2.13097647e-01 -1.02042663e+00 -2.76725322e-01
8.65485847e-01 -1.63005710e-01 1.16503549e+00 8.27686667e-01
8.73152018e-01 -2.38794491e-01 -8.20060968e-02 9.09221113e-01
1.35626233e+00 -4.58617806e-01 8.01990747e-01 5.56143709e-02
5.20853281e-01 5.91729045e-01 1.22426197e-01 4.56914395e-01
1.26163855e-01 7.33985066e-01 3.42868507e-01 -1.08681001e-01
-6.25569046e-01 -7.53268087e-03 3.23535085e-01 6.98143482e-01
-2.24438474e-01 -9.14265215e-02 -6.70008838e-01 6.32885098e-01
-1.73954642e+00 -9.76142585e-01 -3.86181474e-01 2.15580416e+00
8.86798203e-01 -1.15405448e-01 -2.39942595e-01 6.11845106e-02
4.30965573e-01 4.14099813e-01 -7.37311244e-01 1.26562163e-01
-1.28495976e-01 2.57552445e-01 2.41997108e-01 1.04202330e+00
-6.79323256e-01 5.29162586e-01 6.39461946e+00 8.90393376e-01
-9.84195113e-01 5.78837693e-01 6.71470761e-01 -2.78670579e-01
-4.86632496e-01 2.25810751e-01 -2.33524203e-01 4.83038247e-01
5.88432014e-01 -2.88799405e-02 1.01657093e+00 1.32974029e-01
6.31614745e-01 -3.24652880e-01 -1.01748610e+00 8.52335334e-01
-4.42285463e-02 -1.40797663e+00 -1.66241780e-01 2.93456823e-01
9.90550816e-01 2.37818807e-01 3.06070060e-01 -3.09332877e-01
5.15956700e-01 -8.87549937e-01 9.13499177e-01 7.68552959e-01
8.01871836e-01 -7.90934488e-02 -1.84666310e-02 5.47849178e-01
-7.03509510e-01 1.86246812e-01 -2.42006749e-01 -1.75136864e-01
7.84023106e-01 1.13375735e+00 1.20709926e-01 3.56190383e-01
7.17239857e-01 8.31839740e-01 2.54656136e-01 6.95908487e-01
-4.05860394e-01 4.74320501e-01 -4.95643675e-01 7.55472839e-01
-9.02368873e-02 -8.41760278e-01 9.31851029e-01 8.49351346e-01
5.25079370e-01 4.08263922e-01 -4.10661772e-02 1.41980410e+00
5.48771545e-02 -3.83431375e-01 -3.53599936e-01 2.38479942e-01
-4.17634025e-02 1.12386310e+00 -6.26668215e-01 -2.70473897e-01
-4.00688618e-01 1.40475094e+00 3.46383423e-01 8.84061396e-01
-5.82771659e-01 3.31574291e-01 8.39950860e-01 2.67188758e-01
4.30310309e-01 -4.96000022e-01 -3.80407810e-01 -1.39179945e+00
1.57108471e-01 -8.87070477e-01 -1.69565007e-01 -9.35282767e-01
-1.35284376e+00 2.32457176e-01 -1.49730235e-01 -9.63732839e-01
-1.12753205e-01 -3.56926471e-01 -6.13052249e-01 1.30550551e+00
-1.63459444e+00 -9.11669493e-01 -1.19360402e-01 5.53563118e-01
5.81774712e-01 4.19969201e-01 4.44773614e-01 3.08888823e-01
-2.86356241e-01 -2.23858401e-01 4.94305134e-01 -3.97158504e-01
5.41980922e-01 -1.22676837e+00 3.29373479e-01 1.04450655e+00
-4.17832620e-02 5.72582185e-01 1.18619025e+00 -6.20249808e-01
-1.63065195e+00 -5.49726903e-01 4.90818799e-01 -4.78276134e-01
8.45043480e-01 -1.33827016e-01 -1.10331202e+00 6.19546831e-01
3.13859284e-01 2.73189038e-01 1.67868078e-01 -2.64305472e-01
-8.78041983e-02 7.18827620e-02 -1.08006096e+00 5.81405342e-01
1.09755695e+00 -6.40126467e-01 -2.81267941e-01 3.40814918e-01
2.57460773e-01 -6.52400553e-01 -7.64363110e-01 4.89787646e-02
4.09175277e-01 -1.28676641e+00 1.28326130e+00 -1.39077023e-01
8.18497181e-01 -4.25524861e-01 -3.75756882e-02 -1.56967092e+00
-1.80467755e-01 -1.03740740e+00 -4.14535671e-01 9.43709731e-01
5.74409515e-02 -5.86631536e-01 5.46106756e-01 6.65912867e-01
-1.24875605e-01 -5.24010539e-01 -1.04863346e+00 -6.58137560e-01
1.88631400e-01 -3.88593912e-01 -1.18588572e-02 8.33772659e-01
-4.84236240e-01 2.11878031e-01 -6.71684861e-01 2.25412577e-01
1.14763343e+00 -1.14678718e-01 5.72618902e-01 -9.24624622e-01
-8.25260222e-01 -3.61922085e-01 1.08725257e-01 -1.67244887e+00
-6.62910566e-02 -7.35489488e-01 2.08610728e-01 -1.64489377e+00
1.54621854e-01 -2.31521800e-01 1.09877624e-01 2.47611422e-02
-9.34318975e-02 3.59627903e-01 1.27663344e-01 5.84311545e-01
-3.09616849e-02 7.90450931e-01 1.64215159e+00 -1.73343122e-01
-3.30905765e-02 -1.44270554e-01 -5.91637850e-01 7.81073749e-01
2.76386678e-01 -6.33445799e-01 -4.83312100e-01 -9.83441114e-01
2.84723639e-01 4.75489467e-01 6.50089502e-01 -7.23586261e-01
4.30268198e-01 -1.12749130e-01 1.63915798e-01 5.10540716e-02
6.62402987e-01 -5.71863949e-01 3.20657939e-01 2.74922431e-01
-1.48378178e-01 -2.84654677e-01 -7.89977908e-02 7.95129299e-01
-1.36911556e-01 -3.00006568e-01 9.20439243e-01 -2.96068013e-01
-2.24707991e-01 1.81366935e-01 -4.20125693e-01 2.70482332e-01
6.25446141e-01 -6.43154085e-02 -4.37932387e-02 -6.23074532e-01
-1.11209297e+00 -4.38632034e-02 6.93416834e-01 -1.44727141e-01
5.41204929e-01 -9.74624038e-01 -1.07518017e+00 1.78370640e-01
-5.23799360e-01 9.71749872e-02 4.93717164e-01 1.24665332e+00
-4.09033507e-01 -3.85421187e-01 9.18631405e-02 -6.98019743e-01
-9.01444376e-01 3.56369823e-01 6.03372693e-01 -2.60643899e-01
-8.91928077e-01 9.21546817e-01 4.69438255e-01 -9.94234458e-02
-1.63447663e-01 -1.84508756e-01 3.99538696e-01 -1.52423829e-01
6.23779058e-01 4.99142826e-01 -1.01310797e-01 -5.69568515e-01
5.80949076e-02 7.72881687e-01 5.93599044e-02 -8.18900466e-01
1.52146173e+00 -4.17409509e-01 -3.62764448e-01 2.44557291e-01
9.68325734e-01 9.29617882e-02 -1.96428668e+00 -2.81137913e-01
-4.22629029e-01 -5.60582757e-01 6.09091938e-01 -5.30803800e-01
-1.24647856e+00 9.04007316e-01 4.06262577e-01 2.30557531e-01
1.11792707e+00 -6.86529800e-02 6.84472561e-01 -7.22095668e-02
1.08829036e-01 -1.01929855e+00 1.81794558e-02 3.03389311e-01
1.24081171e+00 -1.00847626e+00 3.88655603e-01 -3.92070144e-01
-3.02033335e-01 7.75943100e-01 -1.64166465e-01 -3.63734901e-01
8.09332252e-01 3.81851256e-01 8.76988173e-02 -1.62157416e-01
-5.56280255e-01 -1.38699904e-01 1.92296982e-01 5.67362070e-01
2.28816241e-01 -3.88428867e-01 -3.84237200e-01 -1.56870782e-02
2.63809323e-01 1.87723383e-01 5.73545158e-01 8.87417793e-01
-2.04416409e-01 -1.26430929e+00 -5.14904141e-01 1.93621024e-01
-4.63340789e-01 -2.62452006e-01 7.61885047e-02 2.65157819e-01
-5.91813438e-02 9.74759877e-01 -2.59384900e-01 2.50691593e-01
1.51598513e-01 -1.95942387e-01 7.33502746e-01 -5.03784895e-01
-1.32127896e-01 4.66959983e-01 -2.28016168e-01 -6.95158780e-01
-7.28159368e-01 -9.71938729e-01 -9.30686772e-01 -3.08575481e-01
-1.88811533e-02 -1.41558409e-01 6.54823899e-01 9.56426084e-01
3.89615834e-01 5.90141594e-01 2.52648592e-01 -1.44744122e+00
-6.71637893e-01 -8.26835990e-01 -6.91467524e-01 4.88452077e-01
8.15601289e-01 -5.89305162e-01 -5.86064935e-01 2.90221781e-01] | [11.698702812194824, -2.395348072052002] |
ef7759fb-58de-441e-a8fb-1fa060e9444d | robot-navigation-in-constrained-pedestrian | 2010.08600 | null | https://arxiv.org/abs/2010.08600v2 | https://arxiv.org/pdf/2010.08600v2.pdf | Robot Navigation in Constrained Pedestrian Environments using Reinforcement Learning | Navigating fluently around pedestrians is a necessary capability for mobile robots deployed in human environments, such as buildings and homes. While research on social navigation has focused mainly on the scalability with the number of pedestrians in open spaces, typical indoor environments present the additional challenge of constrained spaces such as corridors and doorways that limit maneuverability and influence patterns of pedestrian interaction. We present an approach based on reinforcement learning (RL) to learn policies capable of dynamic adaptation to the presence of moving pedestrians while navigating between desired locations in constrained environments. The policy network receives guidance from a motion planner that provides waypoints to follow a globally planned trajectory, whereas RL handles the local interactions. We explore a compositional principle for multi-layout training and find that policies trained in a small set of geometrically simple layouts successfully generalize to more complex unseen layouts that exhibit composition of the structural elements available during training. Going beyond walls-world like domains, we show transfer of the learned policy to unseen 3D reconstructions of two real environments. These results support the applicability of the compositional principle to navigation in real-world buildings and indicate promising usage of multi-agent simulation within reconstructed environments for tasks that involve interaction. | ['Silvio Savarese', 'Roberto Martín-Martín', 'Patrick Goebel', 'Can Liu', "Claudia Pérez-D'Arpino"] | 2020-10-16 | null | null | null | null | ['social-navigation'] | ['robots'] | [-1.14877731e-01 2.49350056e-01 2.62026668e-01 -3.36733669e-01
-1.08329266e-01 -7.01647162e-01 6.67746663e-01 9.71521139e-02
-8.36379766e-01 1.15357327e+00 3.86180371e-01 -6.24709606e-01
-3.39393526e-01 -1.11000574e+00 -8.51426721e-01 -6.61859930e-01
-7.68301487e-01 9.08779323e-01 5.50148189e-01 -8.83987665e-01
1.55762453e-02 6.89486086e-01 -1.78717566e+00 -1.06910892e-01
6.67546451e-01 7.23926499e-02 5.58401525e-01 9.50512350e-01
1.21124543e-01 2.79791355e-01 -1.75447911e-01 3.07897508e-01
3.78818125e-01 1.02402493e-01 -7.39463389e-01 1.96777076e-01
1.26838401e-01 -2.15355545e-01 -4.83262002e-01 4.59920704e-01
6.94903135e-01 9.76812840e-01 6.39759958e-01 -1.43312383e+00
-4.54511374e-01 2.53791600e-01 1.83608338e-01 2.05811754e-01
7.21087396e-01 5.68933308e-01 6.54923260e-01 -2.51943976e-01
6.90026760e-01 1.52069581e+00 5.25180280e-01 6.91922665e-01
-1.08868146e+00 1.28395990e-01 7.76555479e-01 3.04807305e-01
-9.16711390e-01 -3.67148340e-01 2.97105342e-01 -1.30633458e-01
1.34015667e+00 2.78467476e-01 8.11462760e-01 1.45715415e+00
2.86232114e-01 6.66902423e-01 7.50638962e-01 -2.61820465e-01
8.66107881e-01 5.40881380e-02 -3.03383440e-01 7.57795870e-01
3.58138382e-01 3.42191339e-01 -5.72185554e-02 4.66937125e-02
6.69727266e-01 -2.56491154e-01 -1.12105040e-02 -1.09320903e+00
-1.38480902e+00 6.89135373e-01 7.85688818e-01 1.38189271e-01
-3.24270010e-01 1.63284078e-01 2.72624999e-01 2.08261132e-01
-2.69125819e-01 5.41220784e-01 -6.73844159e-01 -1.95582703e-01
-1.61193699e-01 8.23471367e-01 1.16837466e+00 1.33350289e+00
6.57280326e-01 2.71443259e-02 2.60879338e-01 6.00167274e-01
4.28009480e-01 6.29435003e-01 3.04591298e-01 -1.31173062e+00
6.10997379e-01 2.89866447e-01 6.35951102e-01 -9.02162552e-01
-1.06462348e+00 -4.12598908e-01 -8.44414413e-01 5.70829809e-01
7.52934039e-01 -3.36939991e-01 -7.72182167e-01 1.69484818e+00
8.28781605e-01 -2.68842727e-01 8.99957865e-02 8.55955660e-01
3.72041613e-01 3.84023607e-01 2.27251127e-01 3.33816648e-01
9.43707943e-01 -1.25583673e+00 -6.77293614e-02 -6.16791785e-01
7.57669866e-01 -1.28040612e-01 1.08695543e+00 2.60314457e-02
-7.33404577e-01 -6.24680400e-01 -9.22726333e-01 1.50267512e-01
-8.53996158e-01 -4.48107630e-01 5.98031938e-01 9.39514279e-01
-1.51013756e+00 5.61170101e-01 -8.61903965e-01 -1.09252524e+00
2.74568081e-01 5.17617762e-01 -5.05269468e-01 -1.53596342e-01
-9.39831018e-01 1.12300622e+00 2.20596969e-01 1.20075874e-01
-1.08284378e+00 -2.77446717e-01 -9.59002137e-01 -1.24492899e-01
2.96015769e-01 -1.17217338e+00 1.17534363e+00 -2.35878646e-01
-1.50072813e+00 3.73796821e-01 -1.03421710e-01 -3.59076113e-01
8.94737899e-01 -1.75306514e-01 -2.67999560e-01 -3.11155498e-01
5.21155000e-01 8.52633834e-01 2.54712462e-01 -1.63405979e+00
-1.05733597e+00 -1.66737765e-01 4.80297059e-01 6.21085823e-01
2.38801762e-01 -8.32519233e-01 7.06012100e-02 -1.13110058e-02
1.85542077e-01 -1.32566631e+00 -1.13524258e+00 -8.71989802e-02
-3.81975979e-01 4.59885858e-02 7.65169740e-01 -1.58987060e-01
5.66865802e-01 -1.69920588e+00 2.38673106e-01 4.35377061e-01
-3.62012625e-01 -2.10421532e-01 -2.18668655e-01 6.97342038e-01
3.55619907e-01 -1.45442747e-02 -2.55542789e-02 -3.63215238e-01
4.73367870e-01 7.72276580e-01 1.03910014e-01 3.89838219e-01
-3.75756711e-01 8.54985952e-01 -1.36441636e+00 -1.17861725e-01
5.91521442e-01 2.88202882e-01 -8.23223650e-01 -3.31832394e-02
-2.43643716e-01 1.07062781e+00 -7.04930067e-01 4.47298318e-01
5.10537744e-01 1.43979102e-01 1.91352844e-01 6.39281094e-01
-4.13680732e-01 1.88488811e-01 -1.54761779e+00 1.73364842e+00
-6.35709345e-01 2.78388351e-01 3.22986126e-01 -7.73338020e-01
5.85956037e-01 -1.78901106e-01 3.71707708e-01 -9.38964367e-01
-1.91245392e-01 1.07506283e-01 -1.16819382e-01 -7.31908619e-01
7.53462970e-01 2.43652821e-01 -2.92348683e-01 2.58681118e-01
-3.23999465e-01 -1.70136228e-01 2.51698941e-01 -1.50069147e-01
1.48579180e+00 5.37946045e-01 3.12425584e-01 -4.53154057e-01
5.38017929e-01 2.22163215e-01 1.67809293e-01 1.35283244e+00
-5.61751068e-01 2.18388021e-01 -1.15010828e-01 -8.16629648e-01
-1.36351407e+00 -1.48413765e+00 1.77590907e-01 1.47176397e+00
3.54004323e-01 -1.53032154e-01 -5.68872750e-01 -5.26873946e-01
3.13313566e-02 9.53588903e-01 -4.89774376e-01 1.43298090e-01
-1.18736660e+00 -6.49907708e-01 2.18525961e-01 2.57268399e-01
7.41141081e-01 -1.32088804e+00 -1.26417661e+00 4.99577075e-01
-1.24568067e-01 -1.07620025e+00 1.48252537e-02 2.63221860e-01
-5.61781466e-01 -1.06779182e+00 -5.00719368e-01 -1.04147208e+00
7.39955306e-01 3.78513128e-01 8.92757952e-01 5.51981293e-03
-4.14951816e-02 1.06729281e+00 -3.81854862e-01 -2.21194159e-02
-3.28761369e-01 3.97618681e-01 3.08049530e-01 -4.61096078e-01
-4.52263542e-02 -1.01773489e+00 -7.89320111e-01 7.19696164e-01
-1.92594647e-01 6.69567958e-02 3.71053785e-01 6.85366929e-01
1.94654703e-01 3.81790906e-01 3.26336503e-01 -4.70140427e-01
5.65161526e-01 -8.24548900e-01 -2.71052808e-01 -3.52979475e-03
-1.26611590e-01 1.29494578e-01 7.75447607e-01 -3.22206438e-01
-1.04792130e+00 2.45330021e-01 -3.05980831e-01 7.89183676e-01
-9.83035266e-01 -1.55436754e-01 -5.57231188e-01 -1.82267297e-02
7.45888829e-01 4.68334742e-02 -4.78786498e-01 -6.59000427e-02
7.29109526e-01 6.42179400e-02 3.45089257e-01 -1.04773259e+00
8.40258658e-01 6.90800428e-01 3.98806840e-01 -1.16214573e+00
4.36141118e-02 -2.88275093e-01 -9.47154641e-01 -3.53373200e-01
7.75414705e-01 -5.37567139e-01 -1.04489923e+00 7.31667653e-02
-7.64675558e-01 -1.03957736e+00 -5.02058089e-01 2.49933362e-01
-1.10942280e+00 2.06530526e-01 -1.59558684e-01 -9.23698843e-01
4.91649866e-01 -1.08989489e+00 9.27214682e-01 3.14675331e-01
-3.34544927e-01 -1.33752847e+00 2.28302479e-01 -1.11758426e-01
6.19652092e-01 4.16436136e-01 9.92991149e-01 -3.53908628e-01
-9.66519713e-01 1.34287938e-01 3.45081806e-01 -5.57792664e-01
1.45985305e-01 -6.03767157e-01 -6.36234701e-01 -3.88141304e-01
-5.09727955e-01 3.67775857e-02 4.14750755e-01 5.93899965e-01
5.02094865e-01 -2.57406324e-01 -8.35798323e-01 2.70245641e-01
1.18978345e+00 4.35676962e-01 5.21174192e-01 1.20502424e+00
4.51047540e-01 9.54400778e-01 5.06568849e-01 2.92796135e-01
9.90284741e-01 5.24888813e-01 7.02689230e-01 7.66341612e-02
1.76253498e-01 -5.54521322e-01 1.56542763e-01 9.81850773e-02
-1.62585706e-01 -4.75283355e-01 -9.29561257e-01 5.81607878e-01
-2.28667188e+00 -9.84795511e-01 1.26171466e-02 2.07698369e+00
-1.33721735e-02 3.49712819e-01 1.95135072e-01 -8.93017426e-02
4.12429065e-01 -1.59225974e-03 -6.18796706e-01 -3.19649279e-01
-1.51974007e-01 -1.97033986e-01 7.07010806e-01 9.97704983e-01
-1.09840000e+00 9.67335701e-01 6.62302971e+00 1.44222230e-01
-2.96883762e-01 3.78023041e-03 3.77677232e-01 1.32632583e-01
-2.73444295e-01 3.05730328e-02 -8.41096759e-01 1.95142984e-01
1.02426994e+00 4.79520917e-01 6.43538773e-01 1.04722166e+00
7.58015394e-01 -4.86090481e-01 -1.14570439e+00 4.78531331e-01
-5.70220828e-01 -1.05015981e+00 -2.55447000e-01 2.91411161e-01
7.16118813e-01 -2.87008528e-02 1.59671471e-01 7.57404506e-01
8.90816092e-01 -9.39643562e-01 8.53487492e-01 4.45424706e-01
-3.31069417e-02 -7.38984704e-01 2.72444725e-01 8.78122330e-01
-1.34704280e+00 -5.33900738e-01 -3.76744419e-01 -3.88357431e-01
6.69616580e-01 -1.72400758e-01 -1.35976386e+00 3.38115543e-01
8.78378391e-01 4.10413921e-01 -5.81659853e-01 1.38147879e+00
-1.81190684e-01 -1.82395130e-01 -4.69560713e-01 -5.88060141e-01
5.71545839e-01 -3.34795743e-01 8.48047972e-01 1.17206585e+00
3.77524018e-01 -8.18992183e-02 4.50007677e-01 3.20983648e-01
8.06065023e-01 -1.49465099e-01 -1.16676068e+00 8.23538959e-01
3.07003886e-01 8.96306992e-01 -1.01538181e+00 1.65965259e-02
-4.27935831e-02 9.95911896e-01 4.55750525e-01 7.67335176e-01
-6.67445600e-01 1.26687214e-01 8.77194047e-01 4.15389478e-01
4.23145354e-01 -9.21950161e-01 -1.54294848e-01 -6.42826021e-01
-1.68154482e-02 -5.30872107e-01 5.60759902e-02 -5.88302255e-01
-9.10476804e-01 3.91244292e-01 1.27774581e-01 -1.09442067e+00
-3.27359229e-01 -6.76398218e-01 -5.07973015e-01 4.80879098e-01
-1.34363222e+00 -1.27303469e+00 -3.54272962e-01 5.28720140e-01
8.14808369e-01 -2.80880988e-01 1.09413922e+00 1.32328033e-01
-1.05184458e-01 1.21941149e-01 2.69330859e-01 -3.08790863e-01
1.83753774e-01 -1.52965665e+00 8.19418311e-01 6.85766280e-01
-2.68695325e-01 6.42067730e-01 1.17084241e+00 -7.23474145e-01
-1.16849685e+00 -1.06954563e+00 5.77884793e-01 -5.84877729e-01
2.95641392e-01 -6.01029336e-01 -3.71782064e-01 7.78430521e-01
-1.53914280e-02 -2.41341427e-01 4.04880494e-01 2.91864365e-01
1.81518048e-01 2.44278073e-01 -1.27488720e+00 1.27134979e+00
1.84675014e+00 7.39390552e-02 -3.03422362e-01 3.25485796e-01
5.36798894e-01 -5.46830893e-01 -1.91904709e-01 7.39324316e-02
6.22491419e-01 -1.10342348e+00 1.31350672e+00 -8.12853456e-01
-2.73500681e-01 -4.90892202e-01 -5.76286972e-01 -1.69012094e+00
-5.19575179e-01 -6.03013337e-01 1.60242543e-01 6.19702578e-01
4.72566992e-01 -7.87613809e-01 1.01062214e+00 7.36466467e-01
-3.93113583e-01 -3.97278219e-01 -1.22981703e+00 -5.52554786e-01
1.60946310e-01 -4.83883291e-01 6.34030640e-01 3.40222538e-01
-1.42764822e-01 1.34988800e-01 -1.67711154e-01 6.29903734e-01
6.94177449e-01 -5.71311772e-01 1.08275330e+00 -9.61084545e-01
-2.23847881e-01 -2.93621927e-01 -3.78227323e-01 -1.47758293e+00
3.04731786e-01 -6.51387095e-01 3.24549019e-01 -2.00589442e+00
-5.46979189e-01 -9.90802705e-01 3.84643555e-01 5.61150908e-02
3.43375951e-01 -6.43212438e-01 5.00915479e-03 -2.15451971e-01
-8.76662552e-01 6.69867516e-01 1.39246774e+00 -1.77298427e-01
-6.28495276e-01 5.33523321e-01 -3.75053525e-01 7.73691237e-01
8.60086739e-01 -1.26466155e-01 -6.83088541e-01 -3.98656040e-01
3.01443547e-01 -1.99375734e-01 5.52577615e-01 -1.43673170e+00
5.36106169e-01 -3.45256507e-01 4.64408815e-01 -4.93811548e-01
4.93249863e-01 -1.09961092e+00 7.09956661e-02 6.35251105e-01
-1.97291702e-01 4.08112586e-01 2.51702040e-01 1.02582324e+00
5.91680229e-01 6.14488460e-02 4.13632512e-01 -6.72675431e-01
-1.25308084e+00 3.35873663e-02 -1.03216422e+00 -3.44931483e-02
1.19376493e+00 -5.85147679e-01 -2.55187273e-01 -4.64798063e-01
-1.37747371e+00 7.10906744e-01 5.22423148e-01 5.93723118e-01
7.15864956e-01 -1.22414041e+00 -3.56703699e-01 3.67290378e-01
-2.95623839e-01 2.64603585e-01 3.82092714e-01 2.78505802e-01
-7.22469747e-01 5.29599428e-01 -6.16021693e-01 -6.18198752e-01
-8.40092659e-01 4.75636303e-01 6.07738554e-01 -1.63576856e-01
-9.51437354e-01 5.22820830e-01 5.36982715e-02 -1.14671922e+00
2.33978972e-01 -5.25971413e-01 -2.55572259e-01 -4.13527668e-01
2.58278370e-01 6.70058966e-01 -2.36093044e-01 -7.05930233e-01
-2.39077583e-01 3.66254568e-01 2.72701830e-01 -4.83710825e-01
1.40744174e+00 -7.43588507e-01 4.33962911e-01 3.27949375e-01
6.08084917e-01 -3.29682261e-01 -1.65635216e+00 1.66777521e-01
2.15898931e-01 -4.77450967e-01 -7.17721999e-01 -6.17031932e-01
-1.20051436e-01 3.77056837e-01 7.32972264e-01 1.70950904e-01
4.67329800e-01 -7.95006230e-02 4.82711852e-01 1.00659108e+00
1.10174203e+00 -1.22272968e+00 2.26292789e-01 7.95791328e-01
7.63221562e-01 -1.19908762e+00 -3.90884548e-01 -8.05799812e-02
-5.43746710e-01 1.05642450e+00 7.35893965e-01 -2.01411545e-01
7.70930290e-01 8.04035813e-02 -2.59618521e-01 1.49611980e-01
-4.73427862e-01 -4.93076324e-01 -3.66479039e-01 1.60101235e+00
-3.32509875e-01 2.37559423e-01 3.40449095e-01 8.06550607e-02
-6.88380837e-01 -6.63861215e-01 6.58520401e-01 1.31204879e+00
-9.10784483e-01 -1.09759927e+00 -4.47437912e-01 -5.54435030e-02
6.79952264e-01 4.95817214e-01 1.37406603e-01 1.27019620e+00
4.90977824e-01 1.08723330e+00 -1.13762112e-03 -8.46866146e-02
8.26698363e-01 -1.03852905e-01 5.08018494e-01 -5.72108507e-01
-3.08029264e-01 -3.20538551e-01 4.55087274e-01 -6.69854164e-01
-3.14303428e-01 -1.09174740e+00 -1.46742320e+00 -3.83211225e-01
4.26405281e-01 -2.06224509e-02 5.73794484e-01 9.86759245e-01
9.04125273e-02 6.09028697e-01 2.12160736e-01 -1.42635429e+00
-2.03452483e-01 -5.40430009e-01 -4.49590564e-01 2.00465456e-01
7.26011157e-01 -8.47850859e-01 -1.55772150e-01 -2.42886126e-01] | [4.76202917098999, 0.8568902015686035] |
5b1b4363-3916-4473-b212-62b278293a38 | meta-batch-instance-normalization-for | 2011.14670 | null | https://arxiv.org/abs/2011.14670v2 | https://arxiv.org/pdf/2011.14670v2.pdf | Meta Batch-Instance Normalization for Generalizable Person Re-Identification | Although supervised person re-identification (Re-ID) methods have shown impressive performance, they suffer from a poor generalization capability on unseen domains. Therefore, generalizable Re-ID has recently attracted growing attention. Many existing methods have employed an instance normalization technique to reduce style variations, but the loss of discriminative information could not be avoided. In this paper, we propose a novel generalizable Re-ID framework, named Meta Batch-Instance Normalization (MetaBIN). Our main idea is to generalize normalization layers by simulating unsuccessful generalization scenarios beforehand in the meta-learning pipeline. To this end, we combine learnable batch-instance normalization layers with meta-learning and investigate the challenging cases caused by both batch and instance normalization layers. Moreover, we diversify the virtual simulations via our meta-train loss accompanied by a cyclic inner-updating manner to boost generalization capability. After all, the MetaBIN framework prevents our model from overfitting to the given source styles and improves the generalization capability to unseen domains without additional data augmentation or complicated network design. Extensive experimental results show that our model outperforms the state-of-the-art methods on the large-scale domain generalization Re-ID benchmark and the cross-domain Re-ID problem. The source code is available at: https://github.com/bismex/MetaBIN. | ['Changick Kim', 'Hyoungseob Park', 'Minki Jeong', 'Taekyung Kim', 'Seokeon Choi'] | 2020-11-30 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Choi_Meta_Batch-Instance_Normalization_for_Generalizable_Person_Re-Identification_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Choi_Meta_Batch-Instance_Normalization_for_Generalizable_Person_Re-Identification_CVPR_2021_paper.pdf | cvpr-2021-1 | ['generalizable-person-re-identification'] | ['computer-vision'] | [-3.59722041e-02 -3.62304360e-01 -4.11805660e-02 -6.56526327e-01
-3.21954846e-01 -4.20757145e-01 6.47526026e-01 -2.19182104e-01
-6.01750493e-01 6.81793094e-01 4.57170084e-02 1.99246287e-01
-1.42560318e-01 -6.74319386e-01 -5.23539901e-01 -5.23211241e-01
4.44557160e-01 5.13524175e-01 -1.80143975e-02 -4.18803483e-01
-1.10958405e-01 2.98321337e-01 -1.41497731e+00 7.13300705e-02
1.25903499e+00 6.71475351e-01 -1.18585322e-02 1.34323344e-01
-3.81972943e-03 2.74619728e-01 -6.63467824e-01 -9.26490307e-01
3.51561934e-01 -4.95962858e-01 -5.75229943e-01 6.53479546e-02
5.26559055e-01 -4.30188984e-01 -5.03982365e-01 1.25143635e+00
9.08491254e-01 4.43173379e-01 4.50558931e-01 -1.32587993e+00
-1.04078650e+00 3.93061936e-01 -6.29424930e-01 2.04237580e-01
6.31920919e-02 5.52903078e-02 5.26413381e-01 -8.55587065e-01
1.71033338e-01 1.36589468e+00 7.36233175e-01 1.19923353e+00
-1.26883817e+00 -1.10067558e+00 5.50809622e-01 2.88910359e-01
-1.59019613e+00 -3.83389711e-01 8.94190073e-01 -2.91252136e-01
6.16015077e-01 4.00740057e-02 1.24027520e-01 1.53047132e+00
-5.45141697e-01 6.04597032e-01 9.09897327e-01 -2.46884331e-01
-7.66683510e-03 3.14728677e-01 2.70835280e-01 2.41148725e-01
3.14338326e-01 7.14017730e-03 -1.42008051e-01 -1.19948424e-02
8.76179874e-01 3.48964304e-01 -1.63471043e-01 -1.14346288e-01
-1.04324710e+00 4.27140117e-01 4.34145898e-01 1.94931716e-01
3.82733089e-03 -2.75982529e-01 6.32384539e-01 3.41721714e-01
6.16434872e-01 2.47837499e-01 -4.64176118e-01 4.66514565e-02
-7.11743593e-01 3.56537759e-01 4.02153105e-01 1.08458960e+00
5.24352252e-01 -9.01180059e-02 -3.44402313e-01 1.57658041e+00
-2.27060005e-01 3.95534545e-01 9.00761962e-01 -4.97526944e-01
8.19865227e-01 8.27606499e-01 5.84760755e-02 -7.57175028e-01
-4.80710208e-01 -8.70194852e-01 -1.40440834e+00 -9.75869596e-02
4.91224021e-01 -2.88915277e-01 -8.58704925e-01 2.06643128e+00
2.30911568e-01 4.50327337e-01 -3.99112981e-03 9.26851809e-01
8.76835942e-01 3.79664153e-01 2.89054006e-01 1.35677099e-01
1.29267883e+00 -1.06758547e+00 -4.67551708e-01 -2.97003061e-01
5.48539579e-01 -4.01256740e-01 1.22383082e+00 3.54322046e-01
-8.40042710e-01 -1.01425827e+00 -1.11158001e+00 -8.59856233e-02
-4.13033843e-01 4.29140925e-01 2.84709454e-01 7.32853949e-01
-6.38610899e-01 5.96711516e-01 -6.00984633e-01 -4.02849197e-01
4.59708244e-01 3.73822957e-01 -4.75434035e-01 -2.60175496e-01
-1.19731748e+00 5.72724760e-01 6.68392658e-01 1.89581349e-01
-6.44685984e-01 -7.58465111e-01 -6.00696623e-01 3.28348875e-02
3.08532536e-01 -8.26886117e-01 1.12473452e+00 -1.18419778e+00
-1.51192999e+00 9.77282643e-01 -1.72546253e-01 -1.88139245e-01
7.63049006e-01 -4.25352871e-01 -8.67484152e-01 -3.70798796e-01
-8.63422360e-03 4.40507293e-01 7.75610328e-01 -1.13389277e+00
-5.27577400e-01 -6.84593678e-01 8.99208859e-02 2.06368998e-01
-8.97777200e-01 -8.66019875e-02 -6.31336689e-01 -1.10198545e+00
-8.94014835e-02 -9.72342372e-01 -8.08010623e-02 -4.33148503e-01
-3.80175143e-01 -3.42167854e-01 3.73134971e-01 -7.41237879e-01
1.40165329e+00 -2.19604158e+00 1.80053428e-01 4.90580276e-02
1.60056502e-01 6.44591510e-01 -3.36269230e-01 1.12979360e-01
-3.96309584e-01 7.76048154e-02 -2.41070032e-01 -7.30095327e-01
-3.45350280e-02 1.05825707e-01 -1.73069209e-01 2.18919948e-01
2.12603826e-02 7.84551263e-01 -7.52660573e-01 -1.38232574e-01
8.14468712e-02 5.22201538e-01 -6.65438592e-01 3.90200227e-01
-4.14923718e-03 7.61070013e-01 -2.94971257e-01 4.72904354e-01
1.00993240e+00 -2.38043979e-01 9.69682783e-02 -3.04414868e-01
1.58574656e-01 1.04117520e-01 -1.24241805e+00 1.79428494e+00
-4.17392641e-01 1.24552846e-02 -2.43084028e-01 -1.15497386e+00
1.10384691e+00 1.16472892e-01 2.22353041e-01 -9.95380580e-01
3.46350744e-02 1.75176397e-01 -1.94474503e-01 -2.74585098e-01
3.75267565e-01 -1.28555179e-01 -2.04816405e-02 1.47200897e-01
1.21799998e-01 8.17525983e-01 1.74652576e-01 4.14949507e-02
5.34312129e-01 1.74088702e-01 6.40985370e-02 -5.35062030e-02
9.69980299e-01 -2.73799181e-01 9.52884436e-01 7.68889189e-01
-2.37127140e-01 7.42948234e-01 5.33540621e-02 -3.39214563e-01
-1.17153561e+00 -9.53582287e-01 -2.10790470e-01 1.36536646e+00
3.43418717e-01 -4.07584935e-01 -8.67857099e-01 -8.71495426e-01
3.12561430e-02 5.47239244e-01 -6.00652635e-01 -2.87929416e-01
-6.89729571e-01 -1.29615963e+00 7.44164765e-01 7.88843155e-01
1.04781687e+00 -6.63467586e-01 4.73370939e-01 2.08867535e-01
-3.26400429e-01 -1.09178722e+00 -6.30569398e-01 -2.78129071e-01
-9.02419925e-01 -7.73885965e-01 -1.26535535e+00 -8.72755408e-01
7.67301023e-01 3.51673275e-01 9.81492281e-01 6.61276653e-02
-1.00004233e-01 2.90712595e-01 -4.18411493e-01 -1.26632690e-01
-1.92046449e-01 3.69269639e-01 3.71108890e-01 1.61959872e-01
8.13013375e-01 -7.83026874e-01 -7.94734180e-01 7.41926730e-01
-7.39037275e-01 1.94266681e-02 3.72692555e-01 1.01327682e+00
5.45329332e-01 1.19410031e-01 9.45937932e-01 -8.30276608e-01
5.67825973e-01 -5.25445402e-01 -5.14750183e-01 4.13576931e-01
-6.55913293e-01 -4.48013358e-02 8.93114090e-01 -7.63395607e-01
-1.43138182e+00 -1.96152583e-01 -3.32651228e-01 -3.13601255e-01
-4.50951546e-01 4.90248911e-02 -6.85312271e-01 1.53218731e-01
7.08687842e-01 2.68501371e-01 -2.48263761e-01 -1.01763964e+00
2.84335464e-01 8.19947660e-01 7.10394919e-01 -8.11914146e-01
9.76630032e-01 4.51121807e-01 -3.76232892e-01 -5.48385382e-01
-9.40119684e-01 -4.98492241e-01 -7.38077462e-01 1.88871175e-01
6.56489551e-01 -1.18158090e+00 -6.49665177e-01 8.18510115e-01
-1.03816915e+00 -2.45613292e-01 -2.73391642e-02 3.93936962e-01
-1.52004421e-01 5.65660000e-01 -4.60622579e-01 -5.03335595e-01
-4.23421293e-01 -9.19008076e-01 6.59727097e-01 4.83657151e-01
7.29424506e-02 -9.29940522e-01 3.77641581e-02 4.19957787e-01
3.92131239e-01 -2.50169300e-02 4.92803007e-01 -9.64550972e-01
-9.32235420e-02 -7.91051537e-02 -4.52723175e-01 7.05330849e-01
3.27763736e-01 -5.91933072e-01 -1.13881814e+00 -5.90213656e-01
-2.75107294e-01 -1.35727659e-01 8.62267673e-01 -3.18970531e-02
1.52559197e+00 -3.16071689e-01 -3.07141989e-01 1.12966108e+00
1.21512926e+00 -7.19003603e-02 6.17368162e-01 7.92255998e-01
9.46393013e-01 5.06242156e-01 3.40705961e-01 5.39373934e-01
5.40905654e-01 8.83532703e-01 5.95029891e-02 -1.97133154e-01
-1.71396583e-01 -3.78717929e-01 1.78806245e-01 5.43505430e-01
-3.77584040e-01 -2.31595054e-01 -6.83536410e-01 4.90953773e-01
-1.89846480e+00 -1.00866723e+00 1.19935259e-01 2.39246750e+00
7.83488274e-01 3.29076163e-02 5.62929511e-01 -1.98355317e-02
1.10662246e+00 -1.99542135e-01 -8.40401173e-01 -4.11445573e-02
-2.17907533e-01 -9.69866812e-02 4.05714780e-01 1.27818018e-01
-1.19555354e+00 8.96024764e-01 4.38812828e+00 9.94529426e-01
-8.61139894e-01 3.60469729e-01 6.44371748e-01 -1.65357590e-01
6.74106032e-02 -3.85600835e-01 -1.33820188e+00 7.32057273e-01
5.98271966e-01 -2.05332354e-01 5.73077261e-01 9.45154190e-01
-2.81981658e-02 5.84421039e-01 -1.06223261e+00 1.37092745e+00
1.56058967e-01 -8.60219836e-01 2.48795003e-01 -6.75016940e-02
7.24265814e-01 -1.46668702e-01 1.06602721e-01 6.36064470e-01
1.68367967e-01 -8.33181143e-01 3.64187002e-01 4.56165373e-01
9.16429877e-01 -9.88617957e-01 6.90187216e-01 3.53659660e-01
-1.12042022e+00 -3.52636546e-01 -7.09548295e-01 1.30460069e-01
1.77151319e-02 5.10481179e-01 -3.40057522e-01 7.30932117e-01
8.94427538e-01 8.46278727e-01 -8.87959898e-01 1.05067885e+00
-1.06791884e-01 4.15900946e-01 -1.61851615e-01 4.19033289e-01
-1.86041072e-01 -1.70397863e-01 4.95205373e-01 1.28641355e+00
2.26953134e-01 1.53997943e-01 8.63209888e-02 8.73684943e-01
-2.92439848e-01 2.90609151e-02 -1.18442692e-01 2.89643645e-01
4.87886101e-01 1.11085546e+00 -1.29157439e-01 -3.55315447e-01
-5.90560615e-01 1.47127366e+00 5.93526423e-01 6.19023085e-01
-8.33192050e-01 -4.64543104e-01 8.73743296e-01 2.40751773e-01
8.72201771e-02 -8.65671039e-02 -2.23036259e-01 -1.65130925e+00
1.95128828e-01 -9.33026493e-01 7.00771689e-01 -2.81294554e-01
-1.93353117e+00 7.21320868e-01 1.03248440e-01 -1.42469680e+00
-4.11649533e-02 -6.32177770e-01 -5.35321891e-01 1.09222770e+00
-1.61188555e+00 -1.31955361e+00 -7.25882649e-01 9.19022739e-01
4.11719829e-01 -4.69600677e-01 5.17932236e-01 7.83008933e-01
-1.17607498e+00 1.45907319e+00 3.28726381e-01 4.59680080e-01
1.14490640e+00 -1.05873370e+00 6.18078113e-01 8.71650934e-01
-3.56908113e-01 1.00658596e+00 3.03762645e-01 -6.59663856e-01
-8.69574785e-01 -1.40466690e+00 5.75971663e-01 -4.18340713e-01
4.25809473e-01 -5.22496045e-01 -1.25471640e+00 8.14617753e-01
-2.24316776e-01 -1.35644814e-02 7.04671144e-01 3.32468778e-01
-6.44402742e-01 -5.07659554e-01 -1.14757514e+00 6.14261091e-01
1.47122967e+00 -3.34833950e-01 -5.57573259e-01 2.62181342e-01
5.22560418e-01 -2.86781609e-01 -9.61889267e-01 5.35261929e-01
5.42423487e-01 -7.53698647e-01 1.31310940e+00 -7.54728138e-01
1.97481826e-01 -3.06489378e-01 2.10676417e-02 -1.37523532e+00
-7.87070453e-01 -3.88229012e-01 8.98767114e-02 1.94492817e+00
1.46421060e-01 -9.19720411e-01 7.17875779e-01 8.22496235e-01
7.63229206e-02 -2.71448612e-01 -7.19531953e-01 -1.17485416e+00
3.23889017e-01 -1.00661345e-01 1.03457451e+00 1.12019289e+00
-1.99484050e-01 3.03922445e-01 -6.63009644e-01 2.69163579e-01
9.23763990e-01 -1.87361047e-01 1.06842494e+00 -1.32749617e+00
-4.06391114e-01 -4.44170773e-01 -2.64418423e-01 -1.29234004e+00
2.38566056e-01 -1.06440544e+00 -4.49551970e-01 -1.14764452e+00
4.36838597e-01 -6.73815846e-01 -5.48625469e-01 5.40086448e-01
-6.05047584e-01 2.33122647e-01 2.41000563e-01 3.97984207e-01
-6.16599143e-01 7.97082543e-01 1.14216614e+00 -1.85787827e-01
-4.03210640e-01 8.45708176e-02 -9.63490903e-01 5.56239188e-01
1.17281938e+00 -3.18465412e-01 -4.65317547e-01 -6.11329257e-01
-3.31188999e-02 -5.69796622e-01 6.80720747e-01 -1.12927294e+00
2.07584947e-01 2.86821648e-02 7.63073623e-01 -2.68695146e-01
1.47571683e-01 -6.92480564e-01 7.09616169e-02 -6.31047636e-02
-3.25267285e-01 6.38060272e-03 3.78030449e-01 5.21095991e-01
-7.94880465e-02 -1.82135805e-01 9.41524744e-01 -1.16198204e-01
-8.30708802e-01 5.78676999e-01 3.87821347e-01 2.52054393e-01
6.89026058e-01 -2.30599672e-01 -5.06600261e-01 4.71564345e-02
-7.43642151e-01 3.36907089e-01 6.01152897e-01 7.22060084e-01
3.24205309e-01 -1.52536237e+00 -9.03480053e-01 3.52289557e-01
3.72933567e-01 -1.77279394e-02 7.03530848e-01 5.61478734e-01
1.16908938e-01 1.27617627e-01 -2.81225264e-01 -4.11657423e-01
-1.00686407e+00 8.01400304e-01 6.25061810e-01 -2.08131179e-01
-6.72253847e-01 8.22353959e-01 6.04700983e-01 -7.92668998e-01
3.55809927e-01 2.28565156e-01 -2.97330260e-01 -1.98173761e-01
9.27255332e-01 6.42450213e-01 6.81581572e-02 -6.27133489e-01
-4.03560042e-01 5.77340484e-01 -5.32843471e-01 2.53529280e-01
1.20010102e+00 -2.94076413e-01 2.10668564e-01 7.92249441e-02
1.17623866e+00 -2.21658051e-01 -1.27437770e+00 -5.10525465e-01
-1.77087247e-01 -3.66752267e-01 -2.97847539e-01 -9.06487644e-01
-1.01957810e+00 7.83767819e-01 9.05642390e-01 -1.88209713e-01
1.32587814e+00 -1.93568826e-01 1.03725147e+00 3.52180630e-01
1.43190742e-01 -1.23815000e+00 -4.89911512e-02 4.63646978e-01
7.91646838e-01 -1.43654823e+00 -9.98028591e-02 -2.19775602e-01
-6.63289070e-01 8.86238635e-01 1.11115146e+00 -5.45133427e-02
4.99415904e-01 -1.86001807e-01 -7.86206201e-02 3.18455070e-01
-1.91841707e-01 -1.22761346e-01 3.96682203e-01 7.88258135e-01
3.90208304e-01 4.49405164e-02 -2.41363868e-01 1.30781484e+00
-7.51718581e-02 1.84474468e-01 6.14761300e-02 3.82528335e-01
-1.11213443e-03 -1.35142589e+00 -4.25376087e-01 2.29228988e-01
-3.83967072e-01 -6.34144396e-02 -2.42678434e-01 6.95002019e-01
3.73146594e-01 8.49973321e-01 -6.79610148e-02 -6.44110501e-01
7.49413252e-01 -1.66637748e-02 4.08950090e-01 -4.61620867e-01
-6.29601836e-01 -3.51634532e-01 -1.20836556e-01 -3.34694266e-01
-3.02760690e-01 -4.93784577e-01 -8.77918720e-01 -4.27232504e-01
-8.80220458e-02 -1.70682203e-02 3.31346810e-01 7.62614071e-01
5.51992536e-01 5.12884915e-01 5.56336820e-01 -7.63964951e-01
-7.55106747e-01 -1.05740416e+00 -4.64933842e-01 8.72744143e-01
2.38516927e-01 -8.56672823e-01 -3.41866672e-01 1.79125778e-02] | [14.748296737670898, 1.0680232048034668] |
5d270299-e4db-44e3-a13c-7d4d2cc57b1e | flowlens-seeing-beyond-the-fov-via-flow | 2211.11293 | null | https://arxiv.org/abs/2211.11293v1 | https://arxiv.org/pdf/2211.11293v1.pdf | FlowLens: Seeing Beyond the FoV via Flow-guided Clip-Recurrent Transformer | Limited by hardware cost and system size, camera's Field-of-View (FoV) is not always satisfactory. However, from a spatio-temporal perspective, information beyond the camera's physical FoV is off-the-shelf and can actually be obtained "for free" from the past. In this paper, we propose a novel task termed Beyond-FoV Estimation, aiming to exploit past visual cues and bidirectional break through the physical FoV of a camera. We put forward a FlowLens architecture to expand the FoV by achieving feature propagation explicitly by optical flow and implicitly by a novel clip-recurrent transformer, which has two appealing features: 1) FlowLens comprises a newly proposed Clip-Recurrent Hub with 3D-Decoupled Cross Attention (DDCA) to progressively process global information accumulated in the temporal dimension. 2) A multi-branch Mix Fusion Feed Forward Network (MixF3N) is integrated to enhance the spatially-precise flow of local features. To foster training and evaluation, we establish KITTI360-EX, a dataset for outer- and inner FoV expansion. Extensive experiments on both video inpainting and beyond-FoV estimation tasks show that FlowLens achieves state-of-the-art performance. Code will be made publicly available at https://github.com/MasterHow/FlowLens. | ['Kaiwei Wang', 'Xiaoting Yin', 'Kailun Yang', 'Qi Jiang', 'Hao Shi'] | 2022-11-21 | null | null | null | null | ['seeing-beyond-the-visible', 'video-inpainting'] | ['computer-vision', 'computer-vision'] | [-2.82524943e-01 -3.53246242e-01 -1.58146083e-01 -2.10308149e-01
-4.67954427e-01 -4.04855847e-01 3.78735900e-01 -5.94103098e-01
-1.12819493e-01 4.86775666e-01 4.27836686e-01 -2.68553555e-01
8.03562924e-02 -4.56723422e-01 -8.01390946e-01 -5.00046194e-01
1.53788313e-01 -5.23853362e-01 2.63601750e-01 3.55129950e-02
4.69657965e-02 5.07954478e-01 -1.24721956e+00 3.76471996e-01
7.26008296e-01 1.19404471e+00 5.68969846e-01 7.16239572e-01
1.43436730e-01 1.34915495e+00 -1.06421687e-01 -2.29730427e-01
4.48379993e-01 -2.80789763e-01 -7.33012021e-01 3.16964835e-01
6.23621762e-01 -9.94822621e-01 -1.10907161e+00 7.30338752e-01
2.39797354e-01 2.06238627e-01 -3.89913619e-02 -1.05867028e+00
-7.91720212e-01 2.55501896e-01 -6.17243171e-01 4.24808830e-01
4.38175291e-01 8.08646858e-01 6.57957733e-01 -1.02292645e+00
8.63652289e-01 1.10420763e+00 4.76001114e-01 3.23924512e-01
-8.77847612e-01 -7.15932667e-01 4.59263861e-01 3.08750749e-01
-1.32987177e+00 -5.78738391e-01 1.05026567e+00 -5.11862397e-01
8.23909521e-01 -1.06268311e-02 7.85292208e-01 1.12897289e+00
1.57592207e-01 9.21244740e-01 7.81925440e-01 2.60331249e-03
-1.79570884e-01 -1.28404841e-01 -2.74743825e-01 8.34206581e-01
-3.17128509e-01 4.78524119e-01 -9.23983574e-01 4.32268918e-01
1.17648828e+00 2.97442108e-01 -7.45857894e-01 -1.39085919e-01
-1.41091061e+00 5.65980136e-01 7.59880960e-01 1.78765073e-01
-2.80595094e-01 5.08464456e-01 2.86035269e-01 1.74086571e-01
5.12695134e-01 3.39820027e-01 -3.62316817e-01 -3.40647191e-01
-1.10865140e+00 6.38711080e-02 2.24781722e-01 1.26323771e+00
8.57760370e-01 1.48721203e-01 -1.09335646e-01 4.73642588e-01
2.92856067e-01 5.49500048e-01 1.86968759e-01 -1.61698902e+00
7.32629180e-01 2.89703816e-01 3.17030460e-01 -8.37079942e-01
-2.17018664e-01 -5.04533291e-01 -9.85696554e-01 1.34283274e-01
3.71305913e-01 -2.13551253e-01 -5.24113238e-01 1.58377624e+00
4.22034860e-01 5.36057889e-01 -2.97494948e-01 1.34299076e+00
7.00509369e-01 9.80017900e-01 -3.04828882e-01 -3.09696466e-01
1.07278252e+00 -1.60055089e+00 -8.00855875e-01 -2.63324171e-01
5.14692247e-01 -8.44394207e-01 9.08029616e-01 4.63101774e-01
-1.25493038e+00 -9.48897839e-01 -9.65584219e-01 -5.52890658e-01
-3.25978510e-02 2.32149258e-01 7.00323939e-01 -3.01071703e-02
-1.07341146e+00 5.97307742e-01 -9.81421411e-01 8.15760866e-02
5.88772953e-01 6.43233582e-02 -4.93010759e-01 -4.45145875e-01
-1.06378984e+00 4.49946076e-01 6.68920055e-02 4.08302546e-01
-9.40251291e-01 -1.05604446e+00 -9.87245440e-01 6.22964650e-02
4.29751396e-01 -1.08273637e+00 1.09347761e+00 -8.17137003e-01
-1.51976454e+00 2.39208564e-01 -4.13434952e-01 -2.82722175e-01
6.09238148e-01 -5.11092782e-01 -2.28953972e-01 5.42876601e-01
1.05570726e-01 8.53866994e-01 1.04544246e+00 -9.59676206e-01
-5.90814888e-01 -1.70170262e-01 2.49500737e-01 2.32647151e-01
-2.93465793e-01 -2.36537252e-02 -7.44076014e-01 -8.32984805e-01
-1.39986157e-01 -7.82725334e-01 -1.09273292e-01 4.99397308e-01
-2.66016304e-01 1.34056568e-01 1.02231634e+00 -8.38478208e-01
1.49134815e+00 -2.28478241e+00 6.16634078e-02 -4.32162941e-01
5.20711482e-01 4.65606451e-01 -4.95742410e-02 3.56145144e-01
-5.14712781e-02 -2.39777952e-01 -1.18520884e-02 -6.40952885e-01
-4.73610610e-01 -9.12148729e-02 -4.85672623e-01 5.92120349e-01
3.04839581e-01 9.40041780e-01 -1.08997536e+00 -4.47566956e-01
7.98162103e-01 7.12536991e-01 -7.47505665e-01 4.58245575e-01
3.03964466e-02 8.13124597e-01 -5.50389349e-01 5.87442279e-01
1.00320697e+00 -5.03220975e-01 -2.79835105e-01 -4.78037626e-01
-4.80270326e-01 1.13465168e-01 -1.00369608e+00 2.38584709e+00
-6.88653171e-01 9.02682781e-01 1.02893986e-01 -4.07177418e-01
6.00460172e-01 6.70743510e-02 6.07778788e-01 -6.86147273e-01
5.01760170e-02 -1.29962424e-02 -4.43405122e-01 -6.43409610e-01
6.06667578e-01 3.19737256e-01 2.79954493e-01 1.55618310e-01
2.32955128e-01 2.02687666e-01 2.28340596e-01 3.09335291e-01
1.14723718e+00 6.11755848e-01 5.86701483e-02 8.89583454e-02
7.11698115e-01 -3.11081648e-01 5.76138198e-01 3.18624049e-01
-3.33151132e-01 8.52272093e-01 3.28373611e-01 -4.78304803e-01
-1.05063939e+00 -1.09792924e+00 -4.34262853e-04 6.79175198e-01
4.50443506e-01 -5.39012671e-01 -4.18626934e-01 -6.12927318e-01
-1.09261297e-01 4.17659760e-01 -5.53099215e-01 1.30153764e-02
-6.93161845e-01 1.58176288e-01 9.29315612e-02 6.80657446e-01
7.98938990e-01 -7.70981550e-01 -7.71570742e-01 2.35282063e-01
-4.50606853e-01 -1.52042079e+00 -1.13759577e+00 -3.05213839e-01
-7.92269289e-01 -1.01637781e+00 -8.33311200e-01 -3.52066278e-01
4.24900264e-01 8.52561176e-01 8.97914648e-01 -7.68638477e-02
-1.86759531e-01 2.95191437e-01 -4.27520722e-01 2.43638739e-01
-2.00753603e-02 -9.21700448e-02 -2.84348369e-01 1.24412730e-01
-1.04281753e-01 -7.47857332e-01 -1.24065351e+00 3.37108135e-01
-9.19440627e-01 4.73649591e-01 5.49616933e-01 9.89761293e-01
3.26170027e-01 -3.36586803e-01 2.47305512e-01 -4.04402852e-01
-1.68242790e-02 -5.11838794e-01 -5.99692106e-01 -8.94727632e-02
-2.66203523e-01 -7.46254399e-02 8.91092777e-01 -3.23747188e-01
-1.42355490e+00 1.31176934e-01 -5.63303083e-02 -1.30021644e+00
4.17283401e-02 1.05268925e-01 -1.56665057e-01 -4.03253399e-02
2.47915000e-01 2.92700291e-01 2.33469158e-02 -4.47477341e-01
5.79978108e-01 4.69688892e-01 7.45940149e-01 -1.87657118e-01
6.70243800e-01 7.98804820e-01 -9.14206915e-03 -6.76583767e-01
-1.02859795e+00 -6.82144940e-01 -5.76713204e-01 -4.84475374e-01
7.77406037e-01 -1.34702492e+00 -9.21026409e-01 6.74171686e-01
-1.29425120e+00 -5.38089097e-01 -2.95898169e-01 6.89631879e-01
-5.61501920e-01 5.48374712e-01 -7.53953338e-01 -4.76255864e-01
-4.73861456e-01 -1.28976536e+00 1.21829844e+00 3.87810439e-01
2.86314845e-01 -9.31195915e-01 -1.38559029e-01 5.04299879e-01
3.99772793e-01 3.92053276e-02 -2.99706645e-02 3.94309908e-01
-1.16455710e+00 2.28260189e-01 -6.37473762e-01 6.19191766e-01
-6.93489760e-02 -2.22433414e-02 -1.13070846e+00 -3.33378524e-01
8.36395696e-02 -1.61630899e-01 1.09769130e+00 5.67200720e-01
1.29398239e+00 -2.71249682e-01 -1.89999983e-01 1.36669910e+00
1.47733891e+00 6.06594011e-02 6.90248787e-01 1.06915876e-01
1.03662479e+00 3.82648230e-01 7.67180204e-01 7.92073667e-01
5.60201585e-01 7.09955096e-01 6.12087727e-01 -4.40498963e-02
-6.97109163e-01 -4.88945931e-01 4.74735022e-01 7.86028385e-01
-1.15571059e-02 -3.67120475e-01 -4.38004464e-01 5.12080491e-01
-1.80727732e+00 -1.05896986e+00 -1.58008561e-01 1.95150590e+00
5.51174819e-01 -1.09985501e-01 -2.03640103e-01 -1.86010033e-01
4.93109852e-01 6.26377106e-01 -6.46299362e-01 4.42332067e-02
-1.66894674e-01 -2.64915347e-01 4.45821673e-01 7.40732610e-01
-1.10646749e+00 8.46356690e-01 4.60877466e+00 9.16252851e-01
-1.26826155e+00 3.07440639e-01 7.14301705e-01 -5.21040618e-01
-1.48819476e-01 1.50238231e-01 -7.91976094e-01 8.09651136e-01
5.46335638e-01 1.57518402e-01 5.87579966e-01 5.15862525e-01
4.74771172e-01 -2.05092937e-01 -1.02342606e+00 1.21837068e+00
-1.08269356e-01 -1.69876957e+00 -2.14707255e-01 1.41664952e-01
7.54616261e-01 2.24443778e-01 8.96337256e-02 1.00187324e-01
-8.49955007e-02 -6.62494004e-01 8.25400591e-01 7.67396152e-01
1.20038664e+00 -5.92835188e-01 4.40947145e-01 1.54215887e-01
-1.48872972e+00 -2.99529523e-01 -2.02734217e-01 -3.16209067e-03
5.83283365e-01 7.08391905e-01 -1.90471768e-01 8.02927852e-01
8.50592196e-01 1.34663761e+00 -5.35361469e-01 9.61524606e-01
-1.02045342e-01 3.73866796e-01 -3.22552621e-01 6.20323896e-01
5.40022433e-01 -2.11294800e-01 6.18591785e-01 1.06406140e+00
5.36767304e-01 1.31739154e-01 -3.34575586e-02 9.46662009e-01
-1.02284789e-01 -3.66993815e-01 -6.07976198e-01 3.52871299e-01
3.85388225e-01 1.35323763e+00 -1.87853888e-01 -2.48774096e-01
-7.25237906e-01 1.03740525e+00 2.25559250e-01 4.63486671e-01
-1.12896228e+00 -2.77053922e-01 7.28018463e-01 3.37231755e-01
5.89619517e-01 -2.20193595e-01 1.27984717e-01 -1.59664774e+00
3.11265767e-01 -3.24101537e-01 8.87339339e-02 -1.27848554e+00
-1.06487513e+00 6.46405935e-01 -1.29934266e-01 -1.64801538e+00
-1.18040077e-01 -4.24562514e-01 -5.18054485e-01 7.74207056e-01
-1.80911732e+00 -1.40466440e+00 -7.16209114e-01 8.47024500e-01
7.17357159e-01 2.23042786e-01 1.93875819e-01 5.87522984e-01
-5.37616253e-01 4.88898486e-01 5.64901046e-02 6.64118007e-02
9.43412125e-01 -7.75404572e-01 3.34564537e-01 1.19793713e+00
-1.35817468e-01 3.64451259e-01 3.93845052e-01 -3.81939501e-01
-1.63397062e+00 -1.43434501e+00 6.20705724e-01 -2.11479425e-01
6.89447641e-01 -2.22591519e-01 -7.93807268e-01 7.92463422e-01
4.44153070e-01 6.68371379e-01 1.23604365e-01 -4.74280506e-01
-4.48180467e-01 -3.97991627e-01 -6.77339256e-01 4.12390739e-01
1.33381832e+00 -5.71507275e-01 -3.86643521e-02 2.86896914e-01
1.08991730e+00 -5.45578599e-01 -1.00066757e+00 1.91909000e-01
4.09412146e-01 -1.32309878e+00 1.05905342e+00 -3.57324481e-02
8.13738644e-01 -4.58614260e-01 2.16764342e-02 -1.07682371e+00
-2.37433299e-01 -1.31910217e+00 -7.05478907e-01 1.20829916e+00
-1.98011875e-01 -5.61305940e-01 6.31915450e-01 2.84114003e-01
-5.42757094e-01 -1.03949976e+00 -8.26716065e-01 -6.33011818e-01
-2.07117304e-01 -6.31767511e-01 4.32251036e-01 7.39701331e-01
-8.23065713e-02 3.96552205e-01 -7.91278899e-01 6.85347915e-02
5.60828865e-01 1.78493053e-01 7.49899268e-01 -5.59856951e-01
-4.81980890e-01 -3.34762454e-01 -2.13682637e-01 -1.93722057e+00
-6.84711859e-02 -5.74834943e-01 -3.08229655e-01 -1.23716044e+00
-8.18571728e-03 -2.19334736e-01 4.75485027e-02 1.31922215e-01
-8.87764320e-02 7.58777633e-02 6.56973004e-01 2.93872982e-01
-8.19528103e-01 8.48336518e-01 1.88201344e+00 5.67994043e-02
-1.29767135e-01 -2.50769943e-01 -4.53570038e-01 6.33039117e-01
4.12233710e-01 -1.20896362e-01 -6.02548659e-01 -7.76673138e-01
1.40557453e-01 7.65699029e-01 5.44369519e-01 -1.05882645e+00
4.26376224e-01 -1.43071385e-02 4.61437225e-01 -7.97467947e-01
7.12467074e-01 -7.53378570e-01 8.47853646e-02 1.50074571e-01
-1.19218946e-01 1.04604572e-01 -4.19589207e-02 8.09996784e-01
-4.32173342e-01 1.10781282e-01 7.43157089e-01 -9.14265588e-03
-7.79416800e-01 7.72074401e-01 -2.99243852e-02 1.39743626e-01
1.05152822e+00 -8.86136815e-02 -5.37344933e-01 -4.52925116e-01
-5.44299364e-01 2.90793896e-01 5.42104423e-01 3.86284411e-01
8.72763276e-01 -1.29336822e+00 -5.76845288e-01 3.49896431e-01
5.19399606e-02 2.10517824e-01 1.01025033e+00 1.15578902e+00
-7.32467413e-01 5.06820977e-01 -1.20904557e-02 -7.59199083e-01
-8.35689664e-01 9.27627325e-01 2.21500725e-01 -3.39770377e-01
-9.59996939e-01 9.36567366e-01 6.58177614e-01 1.02618784e-02
5.23252226e-02 -4.32074457e-01 1.35934964e-01 -2.33485430e-01
7.85397112e-01 4.92302537e-01 -3.99820060e-01 -5.70001304e-01
-1.10029116e-01 6.63734734e-01 4.92571034e-02 1.11234069e-01
1.13114345e+00 -7.72571743e-01 1.25601768e-01 7.17882439e-02
1.59473586e+00 -1.83545515e-01 -2.17593884e+00 -4.06244695e-01
-7.47264564e-01 -1.03057456e+00 2.81905949e-01 -3.91312033e-01
-1.60020387e+00 1.10565114e+00 3.07332963e-01 -2.89776087e-01
1.44816136e+00 -1.69997245e-01 1.22892070e+00 -2.18941327e-02
1.76774189e-01 -7.28328943e-01 1.37056917e-01 3.55520487e-01
9.37977195e-01 -1.30200469e+00 6.29875138e-02 -5.29929817e-01
-6.59327686e-01 1.29918671e+00 6.58597052e-01 -2.13692844e-01
8.43400657e-01 2.44475588e-01 -2.48271391e-01 -2.92643439e-03
-1.15448380e+00 2.26931591e-02 1.70924231e-01 4.68869656e-01
2.44724378e-01 -3.21798265e-01 3.00260365e-01 3.66718501e-01
1.96630031e-01 3.50426137e-01 6.10988736e-01 6.19529426e-01
3.82013358e-02 -5.36556125e-01 -1.83461249e-01 1.34399638e-01
-2.74100721e-01 -1.60954818e-01 4.28569973e-01 7.66794086e-01
2.80565917e-01 9.30099726e-01 9.68900397e-02 -3.92557949e-01
9.45814624e-02 -4.83905733e-01 4.59141374e-01 -4.27822396e-02
-3.78859639e-01 2.91708320e-01 -2.49816664e-02 -1.25305820e+00
-4.56795156e-01 -6.61612332e-01 -8.99758816e-01 -5.74607432e-01
-1.95081919e-01 -2.45402768e-01 4.09725815e-01 6.32300198e-01
7.51485884e-01 7.60468960e-01 8.98726225e-01 -1.06111276e+00
-2.72973031e-01 -8.96555245e-01 -3.75164032e-01 1.00455023e-01
7.26119339e-01 -5.63543081e-01 -4.32378203e-01 3.45054358e-01] | [10.75388240814209, -1.4666528701782227] |
2177af99-69bc-43eb-ad38-c718dd6fefe4 | collision-avoidance-robotics-via-meta | 2007.08616 | null | https://arxiv.org/abs/2007.08616v1 | https://arxiv.org/pdf/2007.08616v1.pdf | Collision Avoidance Robotics Via Meta-Learning (CARML) | This paper presents an approach to exploring a multi-objective reinforcement learning problem with Model-Agnostic Meta-Learning. The environment we used consists of a 2D vehicle equipped with a LIDAR sensor. The goal of the environment is to reach some pre-determined target location but also effectively avoid any obstacles it may find along its path. We also compare this approach against a baseline TD3 solution that attempts to solve the same problem. | ['Abhiram Iyer', 'Aravind Mahadevan'] | 2020-07-16 | null | null | null | null | ['multi-objective-reinforcement-learning'] | ['methodology'] | [-1.80568919e-02 2.67581671e-01 -4.69875038e-01 -7.27166161e-02
-9.31844234e-01 -4.01156634e-01 5.17999828e-01 1.10833198e-01
-1.03509855e+00 1.25102198e+00 -5.52085936e-01 -4.87693667e-01
-6.10963643e-01 -9.63504672e-01 -7.57247210e-01 -7.67202079e-01
-4.69841748e-01 1.02229154e+00 3.54208022e-01 -5.18865108e-01
5.32985926e-01 6.23858333e-01 -1.76391518e+00 -3.60200554e-01
5.68924546e-01 8.75783563e-01 7.40997672e-01 7.90825129e-01
9.51777678e-03 5.90632975e-01 -7.59790421e-01 4.26218480e-01
5.37637711e-01 2.88086027e-01 -9.26400721e-01 1.22025631e-01
9.53381732e-02 6.94946349e-02 2.51432974e-02 1.02165401e+00
3.82494956e-01 4.07095104e-01 5.49881220e-01 -1.74498737e+00
2.51304269e-01 4.34343740e-02 -5.08987844e-01 4.15120274e-01
2.38428622e-01 4.82656032e-01 4.47554320e-01 -1.58033267e-01
6.11193597e-01 1.38840330e+00 2.96292126e-01 7.12103546e-01
-1.12515819e+00 -2.51691967e-01 1.57993838e-01 4.94283736e-01
-1.23752809e+00 -2.15738282e-01 6.58190310e-01 3.74447927e-02
1.46082294e+00 -5.54569960e-02 6.64900482e-01 8.75807583e-01
8.93626750e-01 3.69694620e-01 1.51619112e+00 -3.63954484e-01
8.60714614e-01 2.37546619e-02 -3.97910297e-01 8.15907300e-01
1.42625436e-01 8.70491803e-01 -1.68535456e-01 -6.53514415e-02
3.32592130e-01 -3.40357274e-01 5.35401225e-01 -7.94844687e-01
-7.52346516e-01 1.04078877e+00 3.29105109e-01 6.28060102e-02
-4.83273268e-01 4.64479387e-01 1.64832771e-01 5.50655425e-01
-2.19880372e-01 7.17866600e-01 -6.62437558e-01 -2.18683109e-01
-6.78074300e-01 6.48822188e-01 6.17075443e-01 8.55188310e-01
1.32782900e+00 4.04451996e-01 4.92797077e-01 1.88422590e-01
5.58838665e-01 4.92148340e-01 2.61763662e-01 -1.56518042e+00
5.06784976e-01 3.30660790e-01 4.22913164e-01 -3.35230500e-01
-6.36093855e-01 -1.24221191e-01 -1.02809861e-01 1.31952310e+00
1.71562005e-02 -6.48274899e-01 -1.02877116e+00 1.42490792e+00
5.33686996e-01 2.32365042e-01 6.02903187e-01 6.06346905e-01
3.97385269e-01 5.39768696e-01 1.13129981e-01 -2.28875875e-01
7.21326530e-01 -9.60775793e-01 -3.60041946e-01 -5.92628300e-01
4.63412672e-01 -2.41975635e-01 4.96267915e-01 5.05112827e-01
-1.03931999e+00 -4.52476293e-01 -1.29199672e+00 5.30459106e-01
-6.90155685e-01 -6.39800489e-01 4.54449803e-01 7.40589738e-01
-1.34331203e+00 6.50459886e-01 -7.47835696e-01 -1.95563659e-01
1.53153226e-01 6.30270422e-01 -1.39063701e-01 1.15689084e-01
-9.67539132e-01 1.84659123e+00 8.97934377e-01 -3.35223407e-01
-1.44669390e+00 -2.40152419e-01 -9.92210388e-01 -4.80993658e-01
8.66991162e-01 -5.30313849e-01 1.52946806e+00 -4.47648615e-01
-1.81507993e+00 6.69137359e-01 -1.42445534e-01 -6.60521507e-01
3.76885504e-01 9.98081360e-03 -4.31953967e-01 -6.82655945e-02
1.33748949e-01 8.15623820e-01 7.31973648e-01 -1.55446756e+00
-1.20261371e+00 -3.96296173e-01 4.51176554e-01 5.00425816e-01
3.08130383e-01 -3.51316929e-01 1.16986327e-01 -1.80954356e-02
-3.68533432e-01 -1.08247662e+00 -1.04551375e+00 -4.96269315e-01
-2.34044462e-01 -2.27770701e-01 1.32083201e+00 2.17326850e-01
6.49790466e-01 -1.44348228e+00 3.06226850e-01 3.40555996e-01
-1.03375874e-01 2.01364476e-02 -5.75831771e-01 6.77514970e-01
4.40111101e-01 1.21989965e-01 -1.84580743e-01 -2.82059848e-01
-1.51045993e-01 8.46961439e-01 2.18171310e-02 2.64642954e-01
2.14556724e-01 6.98714316e-01 -1.08295166e+00 -6.52081192e-01
4.35352802e-01 -2.87175700e-02 -2.74107039e-01 -3.07253306e-03
-8.62068772e-01 3.16751242e-01 -7.32488513e-01 8.71320903e-01
6.32124364e-01 4.31892365e-01 2.83060014e-01 8.23402941e-01
-7.17210948e-01 4.34854999e-02 -1.49597609e+00 1.71837151e+00
-5.86664379e-01 3.92980814e-01 1.55928329e-01 -1.16446555e+00
8.50599170e-01 -6.85399994e-02 7.63586819e-01 -1.06409967e+00
5.76877743e-02 2.50460654e-01 -1.39267161e-01 -4.39570665e-01
7.41158247e-01 -1.77684426e-01 -1.81174740e-01 2.64084220e-01
-1.72400489e-01 -7.11075127e-01 1.47735357e-01 -3.68676513e-01
1.37199962e+00 4.68890846e-01 3.51845920e-01 -3.75915885e-01
5.40188491e-01 8.65094066e-01 5.32435715e-01 1.03658640e+00
-4.46624845e-01 -1.76863000e-01 1.15764648e-01 -5.16996264e-01
-6.98775828e-01 -1.31704199e+00 5.02860919e-02 6.92654252e-01
5.45892835e-01 -6.95409328e-02 -3.44790310e-01 -7.68198490e-01
2.66341329e-01 1.03096521e+00 -5.30745685e-01 7.78890923e-02
-6.20693266e-01 -2.80703455e-01 2.72472024e-01 1.13676250e-01
6.76980495e-01 -9.98768747e-01 -1.45231044e+00 4.27687556e-01
1.99316010e-01 -6.53260171e-01 3.06074530e-01 8.18893015e-01
-1.08945680e+00 -1.25947177e+00 3.32413614e-02 -7.21687376e-01
2.94763595e-01 2.73151308e-01 9.61371243e-01 -1.06180102e-01
-1.69216573e-01 7.59783626e-01 -5.56667037e-02 -9.61957216e-01
-5.24433911e-01 -1.41582131e-01 1.03978895e-01 -6.66139901e-01
3.62789929e-01 -2.91608781e-01 -3.08404297e-01 2.91974604e-01
-6.94756091e-01 -5.04544914e-01 5.34054041e-01 5.69323421e-01
7.78012097e-01 8.82313550e-01 3.79454195e-01 -1.60261869e-01
7.40342081e-01 -5.56721747e-01 -9.97986078e-01 1.41584307e-01
-7.84049392e-01 1.92258894e-01 3.36996138e-01 -3.46935719e-01
-6.88066661e-01 3.90891761e-01 1.24150641e-01 -4.10301328e-01
-4.61246192e-01 5.13423204e-01 -1.82806447e-01 -4.90556598e-01
3.93782526e-01 4.59468104e-02 2.12159753e-01 -1.20387986e-01
2.17130661e-01 1.78794071e-01 2.04629049e-01 -5.71116626e-01
1.03493857e+00 1.89564630e-01 6.39286339e-01 -8.30942631e-01
-2.97781378e-01 -2.23199323e-01 -5.27038038e-01 -4.78262842e-01
5.38231492e-01 -6.80610895e-01 -1.00333285e+00 1.14790536e-01
-1.00792313e+00 -8.12453508e-01 -4.20107335e-01 2.51058936e-01
-1.21075058e+00 -2.55226612e-01 1.62694290e-01 -9.76849437e-01
2.57932097e-01 -1.12810862e+00 8.00848246e-01 1.96512237e-01
1.91192493e-01 -1.24515843e+00 4.99725521e-01 2.59587634e-02
6.01531088e-01 5.43852508e-01 9.53784287e-01 -3.56310606e-01
-6.47553384e-01 3.26518357e-01 4.55092520e-01 -1.77074671e-01
3.24013047e-02 -1.55584574e-01 -6.97955847e-01 -4.36292410e-01
6.36564568e-02 -6.32524073e-01 6.67162418e-01 4.40349370e-01
9.85322595e-01 -9.13231224e-02 -5.08923650e-01 -1.28887277e-02
1.83597994e+00 6.01914227e-01 3.18705022e-01 1.39166749e+00
-4.34150621e-02 5.87199748e-01 1.30142498e+00 3.97342980e-01
7.58920789e-01 7.88986385e-01 1.24320555e+00 1.40986919e-01
4.79092032e-01 -3.36440131e-02 2.50336736e-01 -2.35494152e-01
1.06709935e-02 -2.70100027e-01 -1.22392917e+00 5.00801444e-01
-1.98407924e+00 -1.06097615e+00 3.46821189e-01 1.95769823e+00
2.84078509e-01 4.26428527e-01 3.13956678e-01 1.62667751e-01
4.06392276e-01 1.24674983e-01 -9.11455989e-01 -1.00451112e+00
9.49820131e-02 2.68191338e-01 9.44695473e-01 8.61306906e-01
-1.10910642e+00 1.02840507e+00 7.41262484e+00 5.20793021e-01
-1.14027905e+00 -1.08041875e-01 2.21300516e-02 -2.80009031e-01
-1.33666664e-01 6.30022511e-02 -8.47063243e-01 2.84431428e-02
1.35097051e+00 -4.30827737e-01 6.24472499e-01 8.17180693e-01
4.54250813e-01 -7.63104677e-01 -8.51907849e-01 4.01751727e-01
-3.31582338e-01 -1.21899724e+00 -3.06357682e-01 4.53471929e-01
6.95926845e-01 3.60289127e-01 2.61396199e-01 6.80303931e-01
8.12001705e-01 -1.07820582e+00 7.65774071e-01 3.32363605e-01
4.58702713e-01 -1.32692766e+00 2.92474061e-01 8.86531174e-01
-1.00575221e+00 -6.21686935e-01 -2.62448817e-01 -3.65035385e-01
1.41758353e-01 -4.61883098e-02 -1.17552233e+00 7.14809716e-01
7.08886385e-01 2.27737561e-01 -3.30241591e-01 1.38959742e+00
-2.95974255e-01 2.01446423e-03 -1.63815245e-01 -3.70520353e-01
1.08054566e+00 -2.13234335e-01 9.73585546e-01 6.32418156e-01
3.71556908e-01 -1.38138562e-01 7.94392526e-01 3.97705942e-01
5.31893611e-01 -2.78492093e-01 -1.24136496e+00 3.72213960e-01
4.23741400e-01 9.31461453e-01 -4.23724920e-01 -1.23629168e-01
-1.66139558e-01 3.15267593e-01 4.40018803e-01 3.19426775e-01
-7.64752626e-01 -3.81731838e-01 7.64504969e-01 -2.16854796e-01
2.96828777e-01 -5.43058991e-01 2.49088556e-02 -3.73608470e-01
-3.76060843e-01 -7.63349593e-01 2.90090054e-01 -5.67501485e-01
-6.25859141e-01 4.15991068e-01 3.11244547e-01 -1.14668763e+00
-5.22975206e-01 -5.17598391e-01 -6.30715191e-01 7.18476117e-01
-2.20539784e+00 -9.79704857e-01 -7.92311057e-02 6.05624437e-01
7.71058440e-01 -5.67075193e-01 8.74896169e-01 -2.81772733e-01
-2.54707217e-01 -2.12591320e-01 1.87741444e-01 -6.61999822e-01
2.14492097e-01 -1.46840322e+00 9.62829888e-02 4.00973260e-01
-3.11850637e-01 1.61165878e-01 1.02349937e+00 -9.75601554e-01
-1.88525844e+00 -1.25715053e+00 4.40167844e-01 -3.26795489e-01
6.82503879e-01 2.81782091e-01 -3.75392169e-01 7.59752035e-01
5.11870027e-01 -2.25354031e-01 1.24951474e-01 -1.98390651e-02
3.14479172e-01 -7.17701837e-02 -1.61320376e+00 5.31195164e-01
7.13353693e-01 2.27053706e-02 -8.00071776e-01 1.09593138e-01
4.76549536e-01 -7.48445094e-01 -3.47424060e-01 5.96317947e-01
1.37827158e-01 -7.53795564e-01 7.55560875e-01 -8.71221423e-01
1.69841096e-01 -2.44950444e-01 -2.69715637e-01 -1.95787728e+00
-1.48586959e-01 -6.16638601e-01 -1.60592183e-01 4.57936347e-01
3.14987719e-01 -7.34318256e-01 1.21349216e+00 2.31737778e-01
-2.49558285e-01 -6.73416138e-01 -1.74919736e+00 -1.24881291e+00
4.59447414e-01 -5.40994048e-01 5.20267904e-01 2.62739688e-01
-3.19624633e-01 5.51372170e-02 -2.13067085e-01 4.85201418e-01
1.13514030e+00 -4.85767275e-02 7.36928999e-01 -1.20801795e+00
7.70047978e-02 -4.69836652e-01 -3.42249632e-01 -5.33457220e-01
4.74090695e-01 -5.98815560e-01 3.10395122e-01 -1.49234593e+00
-4.39983338e-01 -7.83741057e-01 -3.99590939e-01 3.81942272e-01
4.29364085e-01 -3.52243870e-01 1.06873676e-01 -3.33493620e-01
-9.46009815e-01 4.98496741e-01 1.03204894e+00 -4.15762007e-01
-5.16515553e-01 3.52206826e-01 -3.35308790e-01 6.89228117e-01
1.24980855e+00 -7.71204174e-01 -6.22506738e-01 -4.38765436e-01
-4.10873406e-02 6.55631125e-01 3.14093113e-01 -1.17642128e+00
4.87492561e-01 -9.59365070e-01 1.04580536e-01 -8.22088301e-01
6.64311230e-01 -1.20990622e+00 -2.00532824e-02 1.08691013e+00
-1.91354021e-01 6.34763718e-01 7.95255542e-01 6.96589291e-01
-1.69659689e-01 -6.74128473e-01 8.26998830e-01 -5.15339792e-01
-1.40765977e+00 9.49804261e-02 -1.02215397e+00 -6.13162033e-02
1.76807618e+00 -3.47564578e-01 -3.64182174e-01 -1.70860931e-01
-5.41218102e-01 1.03743923e+00 6.37593865e-01 6.80771768e-01
1.05266917e+00 -1.05111408e+00 -3.87138635e-01 -4.63527255e-02
-2.55444068e-02 -2.84965903e-01 -1.71921715e-01 1.37527794e-01
-1.88021138e-01 5.06917655e-01 -7.60068357e-01 -4.53004450e-01
-1.27083373e+00 6.58644557e-01 8.32991600e-01 -3.39667976e-01
-3.32578391e-01 2.43448243e-01 -7.53185511e-01 -7.41948783e-01
4.16933179e-01 1.73538044e-01 -4.43512529e-01 -1.67024687e-01
4.39918071e-01 7.39828706e-01 -6.71068504e-02 -3.05328190e-01
-4.61310118e-01 5.39525330e-01 3.98906499e-01 -6.13484681e-01
1.45403540e+00 -2.33182177e-01 3.35835457e-01 5.25995970e-01
5.33287108e-01 -3.60317409e-01 -1.22361875e+00 -5.74382469e-02
4.87354666e-01 -4.50422525e-01 3.38935256e-01 -9.56366718e-01
-6.79626286e-01 5.36846459e-01 9.33630943e-01 1.29979342e-01
7.53549933e-01 -4.19101834e-01 3.47786605e-01 9.59726274e-01
1.16487575e+00 -1.47141051e+00 3.15126568e-01 9.34369326e-01
6.97013378e-01 -1.63601375e+00 9.73074436e-02 2.45119423e-01
-7.61638045e-01 1.17431951e+00 9.34159875e-01 -1.10105954e-01
4.06141877e-01 4.38916951e-01 1.99104130e-01 -1.92457333e-01
-1.19944358e+00 -7.30559766e-01 -2.51248360e-01 1.31233430e+00
-6.60674155e-01 -1.38100892e-01 1.51542723e-01 -5.64852834e-01
-1.75784118e-02 -9.06177312e-02 7.27660060e-01 1.41813779e+00
-1.12884283e+00 -1.29878414e+00 -5.13727844e-01 3.06032393e-02
-1.29518077e-01 5.22197604e-01 -1.71679169e-01 1.14771521e+00
1.76195741e-01 1.47164321e+00 3.44315395e-02 -3.13520521e-01
5.03230095e-01 -2.26129815e-01 5.78514457e-01 -6.19470894e-01
-4.87254262e-01 -1.37138456e-01 2.57533312e-01 -7.62248397e-01
-5.26468039e-01 -7.34567583e-01 -1.70463824e+00 -1.43757299e-01
2.70763934e-01 4.16562319e-01 1.08017385e+00 8.61921668e-01
8.31967741e-02 3.63087445e-01 1.01619101e+00 -9.92857337e-01
-6.85853601e-01 -3.86785060e-01 -5.02586067e-01 -6.00394309e-01
7.77599454e-01 -9.91266370e-01 8.85867551e-02 -7.39917338e-01] | [5.003396511077881, 1.3238885402679443] |
185f2540-1b9d-4a7d-a4c6-dffcf767221f | slic-hf-sequence-likelihood-calibration-with | 2305.10425 | null | https://arxiv.org/abs/2305.10425v1 | https://arxiv.org/pdf/2305.10425v1.pdf | SLiC-HF: Sequence Likelihood Calibration with Human Feedback | Learning from human feedback has been shown to be effective at aligning language models with human preferences. Past work has often relied on Reinforcement Learning from Human Feedback (RLHF), which optimizes the language model using reward scores assigned from a reward model trained on human preference data. In this work we show how the recently introduced Sequence Likelihood Calibration (SLiC), can also be used to effectively learn from human preferences (SLiC-HF). Furthermore, we demonstrate this can be done with human feedback data collected for a different model, similar to off-policy, offline RL data. Automatic and human evaluation experiments on the TL;DR summarization task show that SLiC-HF significantly improves supervised fine-tuning baselines. Furthermore, SLiC-HF presents a competitive alternative to the PPO RLHF implementation used in past work while being much simpler to implement, easier to tune and more computationally efficient in practice. | ['Peter J. Liu', 'Mohammad Saleh', 'Misha Khalman', 'Tianqi Liu', 'Rishabh Joshi', 'Yao Zhao'] | 2023-05-17 | null | null | null | null | ['offline-rl'] | ['playing-games'] | [ 3.37003142e-01 1.89915821e-01 -5.68223476e-01 -5.65546215e-01
-1.32049060e+00 -8.31514239e-01 6.11213624e-01 2.77725101e-01
-8.98258090e-01 1.03049791e+00 8.27351630e-01 -4.39464092e-01
1.89834833e-01 -1.67558640e-01 -5.91349900e-01 -4.42726344e-01
1.54689014e-01 9.04223204e-01 -2.72073820e-02 -2.82532036e-01
4.54628170e-01 1.03818439e-01 -1.11025500e+00 3.89378667e-01
7.79680610e-01 4.08635795e-01 3.99641126e-01 9.08839464e-01
2.29100868e-01 9.64379013e-01 -5.01392245e-01 -3.12567115e-01
1.99929878e-01 -3.74290884e-01 -1.08613944e+00 -1.28411412e-01
5.08199096e-01 -5.25091410e-01 -1.61854178e-02 6.88265324e-01
9.28598225e-01 3.76655281e-01 6.35174394e-01 -6.73569381e-01
-6.41142368e-01 8.21080923e-01 -4.14958566e-01 9.47539583e-02
6.81661904e-01 3.16949040e-01 1.36611259e+00 -5.19613087e-01
7.40348876e-01 1.66304445e+00 3.61106664e-01 8.33172560e-01
-1.71776581e+00 -5.78481197e-01 1.83941558e-01 2.51547117e-02
-6.11376524e-01 -3.81234765e-01 3.63109410e-01 -2.68726289e-01
1.33660471e+00 2.18677759e-01 5.42332351e-01 1.39673662e+00
-7.70755112e-02 1.38781130e+00 1.18487632e+00 -6.95259154e-01
2.18184680e-01 3.21565151e-01 1.91278517e-01 4.11906451e-01
-2.02914953e-01 3.91157806e-01 -5.93742788e-01 -4.39629912e-01
4.29845244e-01 -5.39864659e-01 -4.79978442e-01 -6.57967806e-01
-1.17088389e+00 1.15677071e+00 2.82960832e-01 3.13247070e-02
-3.77288729e-01 2.04537660e-01 6.05692089e-01 5.47525167e-01
4.04791296e-01 9.13501322e-01 -8.04507911e-01 -3.43856126e-01
-1.01480246e+00 4.50923264e-01 8.46060395e-01 8.57606411e-01
5.85980296e-01 -1.58584900e-02 -6.84342325e-01 1.06616306e+00
2.89983898e-01 5.45271039e-01 9.08527672e-01 -1.33069575e+00
5.32306671e-01 1.38238713e-01 4.91541952e-01 -4.26782250e-01
-4.20486480e-01 -2.64010310e-01 -3.57703269e-01 -3.21892835e-02
4.14875418e-01 -1.58769980e-01 -7.10046232e-01 1.92468834e+00
-1.26241311e-01 -2.97564536e-01 1.53684869e-01 8.71426940e-01
2.57095098e-01 7.00744152e-01 3.41884941e-01 -3.16234142e-01
8.03832829e-01 -1.01321828e+00 -3.77381623e-01 -3.73307288e-01
1.07733607e+00 -6.76997900e-01 1.56748128e+00 7.71459103e-01
-9.51790154e-01 -3.54362786e-01 -8.89247000e-01 -3.02681364e-02
9.41543058e-02 1.77281454e-01 6.46205187e-01 7.38512993e-01
-1.35281730e+00 7.95182645e-01 -6.05873823e-01 -4.76064622e-01
7.43413046e-02 6.47647858e-01 -2.30793655e-01 -8.53635278e-03
-1.28263295e+00 1.24362636e+00 5.99196017e-01 -5.03191054e-01
-7.68390298e-01 -6.30060494e-01 -8.77944171e-01 -1.23766631e-01
4.54509676e-01 -7.11643696e-01 2.07073665e+00 -1.42915928e+00
-2.00304461e+00 9.32446122e-01 -1.88811138e-01 -8.22665751e-01
6.07716203e-01 -7.16048539e-01 -1.15779571e-01 -2.35601533e-02
2.18302105e-03 9.67722833e-01 6.62536681e-01 -1.05035818e+00
-4.21941310e-01 -9.66369063e-02 -1.05740860e-01 3.53391856e-01
-1.86823130e-01 3.15129399e-01 -1.11542188e-01 -6.42513633e-01
-7.21678495e-01 -1.07787025e+00 -4.36020583e-01 -6.63074613e-01
-5.41696921e-02 -4.46965337e-01 2.76600212e-01 -8.68156672e-01
1.49549150e+00 -1.67955601e+00 2.10943669e-01 3.23459297e-01
-2.77508050e-01 6.48707449e-01 -4.48227763e-01 6.07090533e-01
3.88563350e-02 1.19482450e-01 -3.72318506e-01 -1.03215694e-01
2.71673769e-01 2.95021355e-01 -4.76783216e-01 1.83727100e-01
6.22168519e-02 9.94737148e-01 -1.39990973e+00 -2.60094762e-01
1.66379243e-01 1.59959882e-01 -9.53733742e-01 3.82046789e-01
-4.65372592e-01 4.72264081e-01 -2.22946212e-01 4.31273431e-02
1.60282701e-01 -1.60918683e-01 4.99976814e-01 2.75644153e-01
8.55680034e-02 5.90968966e-01 -9.03931081e-01 1.81613696e+00
-5.62936842e-01 3.67781758e-01 -2.10524037e-01 -7.68941462e-01
9.04726982e-01 4.06261563e-01 9.51052681e-02 -8.70986462e-01
8.83007348e-02 3.18720311e-01 -7.51662115e-03 -3.67077827e-01
6.45365834e-01 -5.94209395e-02 -2.11405039e-01 7.08149850e-01
3.56964350e-01 -3.11258465e-01 2.97005415e-01 3.41616094e-01
9.73266602e-01 4.19673681e-01 7.85540819e-01 -2.23799631e-01
6.96166635e-01 -2.00244248e-01 5.26867151e-01 1.03987968e+00
-1.98243201e-01 4.15937901e-01 2.27584839e-01 -7.24779963e-02
-1.06576383e+00 -8.42159808e-01 3.15360308e-01 1.45907021e+00
-3.31496358e-01 -5.53632140e-01 -8.73441815e-01 -8.61518621e-01
-9.41770598e-02 1.27500379e+00 -1.73429608e-01 -1.50656626e-01
-8.53337765e-01 -5.40874004e-01 3.89149874e-01 5.59307098e-01
8.51529688e-02 -1.44541347e+00 -8.01409125e-01 5.15464008e-01
-2.29086369e-01 -8.16549242e-01 -9.92870688e-01 3.68072003e-01
-8.45847964e-01 -6.93411529e-01 -9.10015106e-01 -5.29345870e-01
2.61737168e-01 1.98690477e-03 1.61585212e+00 1.26045952e-02
4.13458496e-02 6.60199046e-01 -3.45126837e-01 -4.21170563e-01
-6.68994904e-01 2.50482798e-01 1.45455971e-01 -3.52370560e-01
4.46300864e-01 -2.67436296e-01 -4.57058817e-01 1.44910544e-01
-7.49541581e-01 -5.92740290e-02 7.35899270e-01 1.29893935e+00
2.40153402e-01 -7.77222097e-01 7.41208911e-01 -1.30273342e+00
1.10717189e+00 -1.74310565e-01 -4.99692678e-01 4.41039354e-01
-1.01407301e+00 7.79410183e-01 5.96077621e-01 -3.39874476e-01
-1.09585392e+00 2.07530916e-01 -1.48408666e-01 -1.19541660e-01
-1.79580063e-01 3.37868422e-01 -1.08048625e-01 2.65781164e-01
8.30661416e-01 -1.28135592e-01 -1.55080378e-01 -4.48063463e-01
7.77507961e-01 8.00720215e-01 5.62226534e-01 -8.94030809e-01
3.55292141e-01 -1.23290971e-01 -5.85590303e-01 -4.86229688e-01
-8.42753053e-01 -7.87694454e-01 -5.95592499e-01 1.35559961e-01
6.63426816e-01 -7.31202006e-01 -5.38695514e-01 -7.93863609e-02
-1.02738452e+00 -8.48537385e-01 -1.39556140e-01 5.38533807e-01
-9.38844383e-01 4.77911055e-01 -5.59532642e-01 -1.02688324e+00
-5.48773646e-01 -9.79205310e-01 1.13254201e+00 1.30111203e-01
-9.46328700e-01 -1.14572060e+00 4.10309434e-01 2.88617879e-01
2.35447690e-01 -3.25359583e-01 8.82581174e-01 -8.06072354e-01
-5.81161752e-02 3.64807211e-02 1.49681255e-01 4.44245428e-01
-1.89601690e-01 -1.35529444e-01 -9.57577527e-01 -5.57725847e-01
-2.80601710e-01 -9.91920829e-01 9.46896613e-01 3.42106491e-01
1.00543749e+00 -5.18033266e-01 3.09750549e-02 7.64313936e-02
1.11671436e+00 -7.22268745e-02 3.41072112e-01 5.65425813e-01
2.83333212e-01 6.56385660e-01 9.21760023e-01 3.31213176e-01
2.33429372e-01 8.73463392e-01 -1.90204516e-01 -1.72253530e-02
1.87980741e-01 -6.26779020e-01 8.77362132e-01 7.13962674e-01
1.40605897e-01 -3.25978607e-01 -7.32455671e-01 3.15732360e-01
-2.32546186e+00 -9.80112970e-01 2.90263474e-01 2.46710300e+00
1.17656255e+00 1.45396814e-01 5.76784253e-01 -6.77220076e-02
3.96117538e-01 -9.06673446e-02 -4.65514541e-01 -9.27330256e-01
-9.08254981e-02 4.16966766e-01 8.31167638e-01 9.41220462e-01
-1.00625277e+00 1.15913582e+00 6.78489876e+00 7.31899858e-01
-9.80020106e-01 -1.64148165e-03 5.49367309e-01 -1.90171093e-01
-5.69606781e-01 8.95994753e-02 -7.07065463e-01 1.25350505e-01
1.29505539e+00 -2.10918590e-01 7.45579481e-01 6.81267560e-01
4.83815283e-01 -2.52308488e-01 -1.43109059e+00 9.60634828e-01
-1.44728902e-03 -9.09104466e-01 6.83496594e-02 -1.20731503e-01
6.32978261e-01 1.02825463e-01 -1.61588397e-02 7.21204877e-01
1.04333174e+00 -9.84381378e-01 6.54279947e-01 2.46661469e-01
4.97980326e-01 -9.42779362e-01 5.56625426e-01 8.25522840e-01
-4.89288837e-01 -2.33943701e-01 -5.09479821e-01 -2.71650814e-02
1.97374374e-01 1.23440988e-01 -1.36825609e+00 5.49140692e-01
3.54716867e-01 7.41548896e-01 -5.72506666e-01 7.68243968e-01
-5.26002765e-01 8.84134293e-01 -1.67329982e-02 -2.29452863e-01
4.71966416e-01 -1.27670288e-01 4.35714632e-01 1.73966110e+00
-4.69791032e-02 -2.81828791e-01 5.01027882e-01 6.16422057e-01
1.54033685e-02 4.23249036e-01 -5.73517621e-01 -9.71550494e-02
2.85250932e-01 1.09989762e+00 -2.16731578e-01 -4.33843821e-01
-1.03624947e-01 1.10522985e+00 6.32964492e-01 2.77129889e-01
-4.75777060e-01 1.15072452e-01 1.94008335e-01 -1.82486370e-01
2.26838693e-01 -1.26672406e-02 7.03209862e-02 -1.01732481e+00
-4.11176234e-01 -1.60874403e+00 6.15836143e-01 -6.78251505e-01
-1.24612415e+00 3.99262071e-01 2.25591242e-01 -1.10041118e+00
-9.01601732e-01 -6.33484066e-01 -1.76944494e-01 9.53626871e-01
-1.32835948e+00 -6.53573692e-01 5.15565574e-01 3.61555696e-01
8.12244058e-01 -1.16041899e-01 1.10647833e+00 -1.67172790e-01
-2.92596519e-01 7.12172687e-01 1.56525686e-01 -2.38020584e-01
1.11517119e+00 -1.81462479e+00 4.03981805e-01 4.36562717e-01
3.60254437e-01 7.35724688e-01 9.38490033e-01 -5.19012392e-01
-1.03437710e+00 -7.82040238e-01 9.66675580e-01 -5.65413833e-01
4.89637017e-01 -1.31210700e-01 -8.31111431e-01 5.81613004e-01
7.24080563e-01 -6.89346373e-01 7.87075222e-01 1.36433572e-01
-3.75898659e-01 1.55592903e-01 -9.80838835e-01 6.83504164e-01
7.76917875e-01 -5.22236466e-01 -9.16410983e-01 2.48667359e-01
5.89400232e-01 -2.65101641e-01 -5.27528942e-01 2.50315458e-01
4.82649326e-01 -7.58908093e-01 9.46511447e-01 -8.92138720e-01
9.13328007e-02 -1.36119291e-01 -1.39394656e-01 -1.78975797e+00
-4.51343596e-01 -1.06750309e+00 -3.62631083e-02 1.07590759e+00
5.53640246e-01 -3.82749021e-01 3.55705291e-01 6.73662663e-01
1.21354222e-01 -4.92448241e-01 -5.13611019e-01 -7.57623076e-01
2.80879170e-01 -2.01575860e-01 2.69504279e-01 6.54586434e-01
1.79479837e-01 7.77473748e-01 -7.63898790e-01 -3.59242558e-01
4.32745248e-01 -1.90254822e-01 9.10583854e-01 -1.10536861e+00
-8.95425618e-01 -4.91121918e-01 2.21561283e-01 -1.23566759e+00
4.90569562e-01 -1.18932641e+00 3.05452794e-01 -1.33448291e+00
7.63323009e-02 5.34839649e-03 -4.97261167e-01 4.58128363e-01
-3.84011447e-01 -1.89190015e-01 3.70890915e-01 1.78237051e-01
-6.70173883e-01 4.94528592e-01 1.06731761e+00 -1.54553771e-01
-3.79176944e-01 -4.70568202e-02 -8.15530598e-01 6.18628860e-01
9.05367374e-01 -5.11944652e-01 -5.00301659e-01 -1.79164976e-01
2.96360046e-01 6.49408028e-02 -1.00629866e-01 -4.56494927e-01
1.45916026e-02 -4.22947481e-02 2.58462429e-01 -3.96413982e-01
-2.13464200e-02 -3.78148705e-01 -3.81436437e-01 4.51764524e-01
-1.03352702e+00 3.11281830e-01 2.30537757e-01 6.61504984e-01
-8.35601427e-03 -3.72991025e-01 7.01136649e-01 -4.85201597e-01
-9.00790870e-01 -1.63791671e-01 -4.14634913e-01 2.84712195e-01
4.62318182e-01 1.90908089e-01 3.24015990e-02 -6.77713275e-01
-6.00067437e-01 5.77404499e-01 3.98023665e-01 4.72591788e-01
3.22527736e-01 -1.13767529e+00 -7.72514582e-01 -4.97904085e-02
1.90333724e-01 -4.37191963e-01 -3.27301949e-01 7.74090350e-01
-1.84170261e-01 7.48718083e-01 2.24374309e-02 -5.38638234e-01
-1.35982561e+00 6.47868633e-01 1.85497120e-01 -9.43131089e-01
-3.99550557e-01 6.55453920e-01 4.00780588e-02 -8.57811511e-01
5.34287930e-01 -2.53620356e-01 -2.62089401e-01 -2.12286711e-02
4.87130761e-01 3.14989924e-01 4.07443196e-02 -3.27582479e-01
-1.41324580e-01 2.62375265e-01 -3.36961716e-01 -7.25289345e-01
1.19493723e+00 -5.37630171e-02 1.70762166e-01 6.19432747e-01
9.75869775e-01 1.10672839e-01 -1.24416983e+00 -4.10313606e-01
7.95539737e-01 -2.58469164e-01 -8.44492316e-02 -1.18357134e+00
-2.00745866e-01 8.64465356e-01 4.80546087e-01 -2.48175502e-01
8.16312671e-01 -4.07760590e-01 6.76023364e-01 6.56676471e-01
4.12130922e-01 -1.52557766e+00 1.37351036e-01 5.26088059e-01
1.10939419e+00 -1.17113042e+00 -1.69275045e-01 2.31143683e-01
-1.24349117e+00 1.15433919e+00 4.52306926e-01 -1.74909174e-01
6.17798865e-02 -1.36582464e-01 3.49320501e-01 4.02476698e-01
-9.88299787e-01 -8.36099312e-02 4.40656960e-01 5.58140516e-01
9.82357979e-01 2.00022623e-01 -4.18947577e-01 3.36436272e-01
-4.10327792e-01 2.17529297e-01 4.25620407e-01 7.52398908e-01
-5.08460522e-01 -1.86139250e+00 -2.10245937e-01 3.84177774e-01
-3.52274597e-01 -3.03568304e-01 -7.63711512e-01 4.76147234e-01
-4.93739516e-01 1.06539345e+00 -4.49905604e-01 -1.26815394e-01
3.18994313e-01 3.85315746e-01 6.86056972e-01 -9.04363632e-01
-1.04541337e+00 2.77355790e-01 3.52977067e-01 -7.12980092e-01
-4.52703655e-01 -7.58397579e-01 -1.17906809e+00 1.80107117e-01
-1.38054267e-01 3.60209554e-01 1.29469842e-01 8.73557270e-01
1.42908156e-01 2.24733427e-01 6.08474910e-01 -9.28033233e-01
-1.33799994e+00 -1.09084404e+00 -2.60240614e-01 4.50451583e-01
4.28757191e-01 -3.45390528e-01 -2.67459989e-01 -1.72725886e-01] | [11.80644416809082, 8.904767036437988] |
17234e82-ef8f-45f1-a305-6884295c3aac | lstfe-net-long-short-term-feature-enhancement | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Xiao_LSTFE-NetLong_Short-Term_Feature_Enhancement_Network_for_Video_Small_Object_Detection_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Xiao_LSTFE-NetLong_Short-Term_Feature_Enhancement_Network_for_Video_Small_Object_Detection_CVPR_2023_paper.pdf | LSTFE-Net:Long Short-Term Feature Enhancement Network for Video Small Object Detection | Video small object detection is a difficult task due to the lack of object information. Recent methods focus on adding more temporal information to obtain more potent high-level features, which often fail to specify the most vital information for small objects, resulting in insufficient or inappropriate features. Since information from frames at different positions contributes differently to small objects, it is not ideal to assume that using one universal method will extract proper features. We find that context information from the long-term frame and temporal information from the short-term frame are two useful cues for video small object detection. To fully utilize these two cues, we propose a long short-term feature enhancement network (LSTFE-Net) for video small object detection. First, we develop a plug-and-play spatio-temporal feature alignment module to create temporal correspondences between the short-term and current frames. Then, we propose a frame selection module to select the long-term frame that can provide the most additional context information. Finally, we propose a long short-term feature aggregation module to fuse long short-term features. Compared to other state-of-the-art methods, our LSTFE-Net achieves 4.4% absolute boosts in AP on the FL-Drones dataset. More details can be found at https://github.com/xiaojs18/LSTFE-Net. | ['Jiayi Ma', 'Zhongyuan Wang', 'Shurui Wang', 'Yunhua Chen', 'Yuanxu Wu', 'Jinsheng Xiao'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['small-object-detection'] | ['computer-vision'] | [-1.15647294e-01 -6.16313756e-01 -2.70407051e-01 -4.82207805e-01
-6.16047144e-01 -4.31082785e-01 4.14897352e-01 -3.72986123e-02
-5.27024269e-01 5.07900953e-01 2.21340805e-01 2.55956560e-01
6.59086779e-02 -8.26423407e-01 -7.08579659e-01 -6.74343705e-01
-3.01364571e-01 -3.46981376e-01 9.75408316e-01 -2.71060914e-01
1.96824968e-03 4.19398755e-01 -1.61457109e+00 4.74274278e-01
6.27186537e-01 1.36437047e+00 4.84712124e-01 5.30470252e-01
1.75299779e-01 6.64031088e-01 -4.41696763e-01 -3.30590755e-02
5.31081676e-01 -3.35107476e-01 -2.75235623e-01 -3.87052707e-02
7.66333163e-01 -7.82774210e-01 -5.44567287e-01 1.00505304e+00
5.86752892e-01 2.04590037e-01 2.05438778e-01 -1.26868927e+00
-1.23874620e-02 2.89118677e-01 -7.30338037e-01 7.49399185e-01
5.61776817e-01 4.13925856e-01 8.70195806e-01 -1.11216199e+00
5.24667382e-01 1.12913203e+00 5.83062053e-01 3.19294244e-01
-7.54206300e-01 -9.86454129e-01 3.43102068e-01 4.83219862e-01
-1.45043862e+00 -6.79709613e-01 6.25095665e-01 -3.14552307e-01
6.10199094e-01 2.65222311e-01 7.58136749e-01 8.83475363e-01
8.44289958e-02 7.98597634e-01 5.57091355e-01 -1.19900294e-02
-1.76734596e-01 -2.35066295e-01 5.28354384e-02 6.52761817e-01
2.06976444e-01 3.56944591e-01 -7.26929903e-01 -2.70692352e-02
6.72045112e-01 3.90517652e-01 -4.77295101e-01 -3.55796516e-02
-1.35438585e+00 5.71035326e-01 5.92324853e-01 4.18586224e-01
-4.37987715e-01 3.16441417e-01 1.53780967e-01 1.66636392e-01
3.56206328e-01 1.27182871e-01 -4.78721768e-01 -1.88534647e-01
-9.72043514e-01 2.80212820e-01 2.81106323e-01 9.44861650e-01
9.50559020e-01 -4.67984825e-02 -5.23688674e-01 6.37187123e-01
2.58180082e-01 4.94828701e-01 1.97028279e-01 -1.03740263e+00
5.84373593e-01 4.61906284e-01 1.95206642e-01 -1.13629258e+00
-3.65847856e-01 -4.78659123e-01 -3.71932149e-01 4.04035486e-02
2.82980293e-01 -2.74414986e-01 -7.13153720e-01 1.66820657e+00
4.44900632e-01 6.04506910e-01 -2.38763675e-01 1.18364477e+00
9.68596339e-01 7.55891323e-01 -9.53387171e-02 -2.82975048e-01
1.34812987e+00 -9.68487620e-01 -5.29626012e-01 -3.15422237e-01
3.52977902e-01 -7.05000579e-01 7.10406601e-01 1.11401372e-01
-8.81636083e-01 -7.51946032e-01 -1.02074850e+00 9.32496265e-02
-2.16927886e-01 2.84629017e-01 4.34992254e-01 2.13471517e-01
-7.35435009e-01 3.40229213e-01 -7.27856934e-01 -3.38320464e-01
4.33491141e-01 3.82062972e-01 -3.41330737e-01 -1.93913370e-01
-1.09481537e+00 6.10889912e-01 4.00696039e-01 2.52306908e-01
-1.02163112e+00 -5.14630914e-01 -8.30514312e-01 6.00470863e-02
9.00582910e-01 -5.87494254e-01 1.14920223e+00 -9.38364029e-01
-9.90072727e-01 1.73329890e-01 -4.44413036e-01 -3.71227652e-01
3.65719110e-01 -2.94131428e-01 -3.52664948e-01 4.77404207e-01
1.96921274e-01 8.10244560e-01 9.70487177e-01 -1.02351511e+00
-1.28298569e+00 -1.65306970e-01 3.59076113e-01 2.17926398e-01
-5.44017434e-01 2.93601751e-01 -9.38781500e-01 -9.22506690e-01
8.65663886e-02 -7.63186038e-01 -3.60816121e-02 2.42547750e-01
2.10793819e-02 -1.66051209e-01 1.00432754e+00 -4.60065603e-01
1.51649523e+00 -2.29533839e+00 -4.48460020e-02 -1.15570940e-01
3.49387497e-01 4.92619008e-01 -2.04694986e-01 3.82161848e-02
1.15693592e-01 -3.29456069e-02 1.13061450e-01 -2.63688087e-01
-2.86920786e-01 7.02729588e-03 -1.11459292e-01 3.40658963e-01
2.24199012e-01 7.55937338e-01 -1.07604516e+00 -5.29794455e-01
3.69934022e-01 4.98067498e-01 -4.89851207e-01 1.67180374e-01
2.72773970e-02 2.64516652e-01 -6.68219090e-01 8.06168556e-01
7.39667237e-01 -1.17142655e-01 -4.25728470e-01 -5.24597943e-01
-3.81922901e-01 1.39096528e-01 -1.21360230e+00 1.63412488e+00
-3.72869261e-02 7.30861843e-01 -2.60857735e-02 -5.82766473e-01
5.39289713e-01 2.47269168e-01 7.26343632e-01 -6.55436397e-01
3.32790285e-01 2.64165848e-01 -5.63880019e-02 -6.24978364e-01
5.63928246e-01 1.92802459e-01 1.28822356e-01 -6.73260018e-02
1.72405884e-01 3.21995258e-01 6.70912862e-01 2.02394024e-01
1.15824628e+00 1.27363533e-01 2.48330906e-01 4.78264131e-02
5.91963589e-01 -4.14772183e-01 1.08350325e+00 5.21881998e-01
-4.59243566e-01 8.94694865e-01 2.40772039e-01 -4.46607411e-01
-4.97514129e-01 -7.21951842e-01 4.66786139e-02 1.19507253e+00
5.25599957e-01 -8.70555460e-01 -5.46329737e-01 -8.38269293e-01
-6.04081936e-02 1.45004690e-01 -6.06945574e-01 -2.24584505e-01
-7.12627769e-01 -5.95607400e-01 3.55532855e-01 4.77399707e-01
6.55471325e-01 -7.96659172e-01 -9.59144294e-01 2.67882794e-01
-4.90827441e-01 -1.31496513e+00 -8.40432525e-01 -9.87540558e-02
-7.12016046e-01 -1.00343323e+00 -7.53093600e-01 -3.76723319e-01
4.61179167e-01 8.29298258e-01 7.74854898e-01 2.71026105e-01
-1.85639858e-01 5.84062636e-02 -7.23864198e-01 -2.23069116e-01
2.01853156e-01 -1.50777802e-01 2.98639871e-02 2.75220454e-01
3.20364594e-01 -4.68435854e-01 -8.89778495e-01 5.95884681e-01
-7.85273612e-01 1.10882102e-02 4.77352232e-01 6.37786627e-01
5.02971351e-01 -5.51159047e-02 3.35103840e-01 -1.92928687e-01
1.93796605e-02 -4.48062003e-01 -5.67333162e-01 4.85699065e-02
-9.64951068e-02 -2.84460634e-01 5.24656534e-01 -5.93574345e-01
-7.83137798e-01 5.11471517e-02 -1.66718870e-01 -7.05865860e-01
-2.92976815e-02 2.62932330e-01 -2.65818894e-01 -2.05468670e-01
3.98486733e-01 1.86109066e-01 -3.35315108e-01 -3.89553457e-01
-1.21308491e-01 4.57687885e-01 3.07792902e-01 -2.51596600e-01
7.79778838e-01 5.60890555e-01 -1.44452721e-01 -7.38959968e-01
-9.80488658e-01 -6.66686773e-01 -5.79220653e-01 -4.22362179e-01
7.75755823e-01 -1.14933097e+00 -5.67512095e-01 4.34755296e-01
-9.83499527e-01 -9.24871266e-02 -1.87173024e-01 6.31881833e-01
-2.41246760e-01 3.43562096e-01 -3.65755826e-01 -4.70232993e-01
-2.20248416e-01 -1.36541998e+00 1.18233764e+00 4.83160108e-01
1.78079277e-01 -4.10858661e-01 -3.39517921e-01 1.29012138e-01
4.28527951e-01 1.46850899e-01 -6.13171756e-02 -2.58482307e-01
-9.07166719e-01 -1.60141870e-01 -3.23824853e-01 2.86289722e-01
2.16755986e-01 9.35192406e-02 -7.97814369e-01 -3.93263012e-01
-1.01347782e-01 2.22128001e-03 1.27078843e+00 6.14369154e-01
9.51728761e-01 -1.49849191e-01 -4.84889537e-01 1.03195047e+00
1.25789261e+00 3.28301609e-01 5.45529068e-01 3.20930988e-01
8.44660163e-01 2.23827347e-01 1.22793186e+00 6.38125420e-01
5.09378016e-01 9.38156307e-01 4.13797110e-01 -4.29191738e-02
-3.91508788e-01 -1.49940997e-02 7.02315927e-01 2.70701438e-01
-2.68086493e-01 -2.57019877e-01 -6.20406270e-01 4.32336956e-01
-1.99637175e+00 -1.20834887e+00 -5.16242906e-03 2.22126222e+00
5.38030684e-01 2.70483524e-01 2.92450219e-01 -1.16587393e-01
7.07149029e-01 3.81394088e-01 -4.08205867e-01 5.37388742e-01
-2.53695190e-01 -2.51292557e-01 5.74400008e-01 3.12436014e-01
-1.36558628e+00 9.06136036e-01 5.14364719e+00 1.04536641e+00
-1.29663348e+00 1.69324011e-01 3.34332913e-01 -5.38169205e-01
8.38576630e-02 4.54445146e-02 -1.12346923e+00 8.08007598e-01
6.55108929e-01 -9.99905989e-02 1.08895659e-01 6.96187556e-01
5.01230299e-01 -4.22251761e-01 -1.00241566e+00 1.08511627e+00
1.35095313e-01 -1.23966670e+00 5.24349324e-02 -7.96618871e-03
4.96305794e-01 2.49588236e-01 -1.28730685e-01 8.40233117e-02
-3.62479419e-01 -5.57452321e-01 1.13335562e+00 4.63049680e-01
6.33670926e-01 -7.57190228e-01 7.49354064e-01 1.99242517e-01
-1.88632715e+00 -4.00935173e-01 -3.88869405e-01 7.58099332e-02
2.65472919e-01 5.87390065e-01 -4.05568630e-01 5.42365491e-01
9.96009767e-01 1.18797398e+00 -8.34528387e-01 1.51295364e+00
-1.13831848e-01 5.76485336e-01 -5.38320780e-01 2.32271150e-01
1.12918809e-01 6.95443433e-03 8.53887618e-01 1.29424036e+00
5.53597927e-01 4.13875371e-01 4.75943178e-01 5.57641447e-01
4.86496128e-02 -6.24533221e-02 -4.39912707e-01 1.85271546e-01
4.28274244e-01 1.36362779e+00 -7.70183086e-01 -5.46064138e-01
-7.46953607e-01 9.05608118e-01 -1.48859948e-01 3.69315743e-01
-1.13690078e+00 -3.52082402e-01 7.89868951e-01 1.78785101e-01
6.34341478e-01 -2.56204247e-01 2.25700349e-01 -1.53129876e+00
3.02563369e-01 -6.77634835e-01 5.06344020e-01 -7.11071849e-01
-8.18544030e-01 7.20166743e-01 7.36623928e-02 -1.73034680e+00
-8.61137137e-02 -2.85083085e-01 -4.76248145e-01 4.91179764e-01
-1.63986361e+00 -1.18770707e+00 -6.90268159e-01 6.77546978e-01
8.37407291e-01 -3.35749499e-02 2.31473312e-01 7.99024045e-01
-5.41789114e-01 5.82945585e-01 -2.50069171e-01 2.92687565e-01
8.97529900e-01 -6.93403900e-01 1.49495587e-01 1.30138445e+00
2.23516095e-02 4.57481533e-01 6.47426844e-01 -6.76312923e-01
-1.32238817e+00 -1.16254139e+00 6.07440233e-01 -3.20295632e-01
2.88967520e-01 -2.99671441e-01 -7.08002150e-01 6.06762111e-01
-1.00902684e-01 6.24551117e-01 2.86806434e-01 -3.30092847e-01
-1.26814321e-01 -5.07663548e-01 -7.92879939e-01 3.96763891e-01
1.13060343e+00 -2.10966542e-01 -4.42333937e-01 -2.39580516e-02
8.37779403e-01 -3.21938932e-01 -5.61499476e-01 6.99371457e-01
7.42395878e-01 -1.20637751e+00 1.12906730e+00 -2.17048749e-02
2.46883109e-01 -8.28203976e-01 -2.45785072e-01 -9.40921366e-01
-3.98401678e-01 -6.66238189e-01 -2.69214809e-01 1.18829715e+00
1.66262478e-01 -4.59650785e-01 5.88328242e-01 2.28531212e-01
-2.52685010e-01 -6.93336606e-01 -8.91490698e-01 -8.53516042e-01
-5.68585575e-01 -6.02494895e-01 5.39075017e-01 6.47366643e-01
-2.09155962e-01 1.29649475e-01 -5.14132977e-01 2.24763870e-01
4.19401586e-01 2.58769214e-01 7.65767515e-01 -1.10404861e+00
-6.63869828e-02 -3.30313295e-01 -5.70463777e-01 -1.29352129e+00
-1.95381597e-01 -3.62777114e-01 2.33691305e-01 -1.45157397e+00
4.12554681e-01 -3.77814502e-01 -4.42247123e-01 5.17550230e-01
-4.71157014e-01 5.10214508e-01 4.77087468e-01 8.84852782e-02
-9.87354755e-01 6.11682296e-01 1.13333976e+00 8.38006660e-02
-2.04900682e-01 3.27349976e-02 -4.28182095e-01 8.20584834e-01
6.55885875e-01 -5.65730333e-01 -1.83150515e-01 -3.81857425e-01
6.60773925e-03 5.81378788e-02 4.94085401e-01 -1.33118939e+00
4.15525585e-01 -3.70047033e-01 5.90243399e-01 -8.05343807e-01
6.61446273e-01 -8.81915867e-01 -7.24549815e-02 2.67476887e-01
8.03818405e-02 -2.21101530e-02 2.42726177e-01 4.72416699e-01
-4.78192180e-01 -1.55106217e-01 7.19543934e-01 -1.11210123e-01
-1.08857536e+00 7.54919171e-01 -3.33075434e-01 -1.23737402e-01
1.22555888e+00 -3.71161252e-01 -2.33705640e-01 -3.12396467e-01
-5.05028844e-01 4.43750471e-01 4.75915790e-01 6.08667433e-01
8.44570339e-01 -1.35062802e+00 -7.13182986e-01 2.49659061e-01
1.97573006e-01 -2.33683698e-02 3.86025965e-01 1.08917916e+00
-2.16148704e-01 3.49531710e-01 -3.28056961e-01 -7.26657093e-01
-1.64326513e+00 4.88702416e-01 2.67801523e-01 1.36848286e-01
-5.26258886e-01 1.07082403e+00 4.17088211e-01 3.49908739e-01
1.41599461e-01 -5.22290289e-01 -3.54830295e-01 4.02435750e-01
8.73138011e-01 3.26076269e-01 -1.81195959e-01 -1.06809628e+00
-5.29339910e-01 9.07451212e-01 7.82913417e-02 7.25503489e-02
1.42630076e+00 -4.68063563e-01 1.74377132e-02 1.82507217e-01
1.24699199e+00 1.19930476e-01 -1.56810367e+00 -2.75539845e-01
-3.63633901e-01 -9.59731221e-01 2.09524393e-01 -3.86339098e-01
-1.32703769e+00 8.14669788e-01 7.38845706e-01 -1.02239504e-01
1.49583197e+00 -5.82808256e-03 9.04910922e-01 2.39524379e-01
4.80054289e-01 -9.00357127e-01 1.96618333e-01 5.53084314e-01
8.46630931e-01 -1.46177137e+00 1.79787099e-01 -6.39908850e-01
-4.27483946e-01 1.27541971e+00 7.98360825e-01 5.01928367e-02
6.21787131e-01 2.55251706e-01 -1.24353960e-01 -1.54639497e-01
-8.28292847e-01 -6.48350239e-01 6.20971441e-01 2.96307862e-01
2.10473225e-01 -2.47453943e-01 -1.38433889e-01 5.95317900e-01
1.15588769e-01 1.45546943e-01 4.67054695e-01 9.51882184e-01
-6.76085234e-01 -9.63348627e-01 -3.94586265e-01 4.62722510e-01
-6.01542354e-01 -1.09204039e-01 2.12360509e-02 5.71881175e-01
5.70534885e-01 1.00404513e+00 2.33051311e-02 -7.11148798e-01
2.99421012e-01 -4.28208798e-01 3.05803031e-01 -4.83996093e-01
-4.62169319e-01 5.18084168e-01 6.87567294e-02 -1.05681384e+00
-6.68679237e-01 -7.34201312e-01 -1.29745531e+00 -1.53760687e-01
-5.27137220e-01 -9.94703248e-02 2.58808762e-01 6.94777966e-01
4.67837274e-01 5.81010938e-01 5.40150881e-01 -1.26925230e+00
1.09736519e-02 -9.52709377e-01 -3.62120390e-01 3.27250093e-01
7.75503635e-01 -9.43937778e-01 -4.23178107e-01 2.37332299e-01] | [9.071020126342773, -0.3598281741142273] |
8ae81d02-6af0-4760-a99b-f0d33ea998f9 | exploiting-spatial-information-with-the | 2210.15512 | null | https://arxiv.org/abs/2210.15512v2 | https://arxiv.org/pdf/2210.15512v2.pdf | Exploiting spatial information with the informed complex-valued spatial autoencoder for target speaker extraction | In conventional multichannel audio signal enhancement, spatial and spectral filtering are often performed sequentially. In contrast, it has been shown that for neural spatial filtering a joint approach of spectro-spatial filtering is more beneficial. In this contribution, we investigate the spatial filtering performed by such a time-varying spectro-spatial filter. We extend the recently proposed complex-valued spatial autoencoder (COSPA) for the task of target speaker extraction by leveraging its interpretable structure and purposefully informing the network of the target speaker's position. We show that the resulting informed COSPA (iCOSPA) effectively and flexibly extracts a target speaker from a mixture of speakers. We also find that the proposed architecture is well capable of learning pronounced spatial selectivity patterns and show that the results depend significantly on the training target and the reference signal when computing various evaluation metrics. | ['Walter Kellermann', 'Mhd Modar Halimeh', 'Annika Briegleb'] | 2022-10-27 | null | null | null | null | ['target-speaker-extraction'] | ['audio'] | [ 4.62955952e-01 -1.55715466e-01 3.85278374e-01 -2.68403232e-01
-8.74294341e-01 -6.17708921e-01 6.72184408e-01 -7.10505545e-02
-6.73388362e-01 6.45074010e-01 3.20117176e-01 -5.76026700e-02
-4.95102763e-01 -3.50470513e-01 -6.82421625e-01 -1.00122511e+00
-2.15759888e-01 -1.00182585e-01 2.81720161e-01 -1.54374778e-01
2.21503034e-01 7.68479109e-01 -1.70429146e+00 3.91755342e-01
5.41077733e-01 1.01865256e+00 4.37968373e-01 1.02059555e+00
4.21796858e-01 5.83122671e-01 -1.00892568e+00 1.44517198e-01
8.30316097e-02 -6.92154825e-01 -6.21784270e-01 -3.74932960e-02
5.74406087e-01 -3.24965194e-02 -2.93931514e-01 9.50040162e-01
6.76967859e-01 4.55157042e-01 6.77504718e-01 -7.31559396e-01
-1.67900100e-01 8.46016526e-01 -1.18671969e-01 5.66926003e-01
1.09501205e-01 -7.22238198e-02 9.85937893e-01 -1.06551838e+00
3.39326262e-01 1.00724173e+00 7.94376075e-01 2.74674624e-01
-1.20292866e+00 -5.04053295e-01 3.02058250e-01 3.13113630e-01
-1.36281455e+00 -6.98804975e-01 1.08258080e+00 -3.72595787e-01
8.32878292e-01 5.16561568e-01 5.38516581e-01 1.01328945e+00
-9.35784951e-02 7.56244361e-01 1.12717891e+00 -6.87171102e-01
3.55408937e-01 6.81443587e-02 8.37505311e-02 1.62255719e-01
-3.99377584e-01 4.81795907e-01 -7.49029338e-01 6.87008873e-02
7.16448188e-01 -4.63790625e-01 -6.79215372e-01 -5.16062006e-02
-1.23183489e+00 4.12318915e-01 4.35426742e-01 9.70279098e-01
-5.41210592e-01 1.95859537e-01 2.05267608e-01 2.78919339e-01
2.53580153e-01 6.84339166e-01 -4.58924800e-01 -8.70340392e-02
-1.46223140e+00 1.79443419e-01 4.74260360e-01 2.95622736e-01
3.40735197e-01 4.77946699e-01 -2.73220688e-01 7.27432251e-01
6.28793091e-02 2.64244199e-01 5.57615101e-01 -1.02122271e+00
5.42966351e-02 -1.31895021e-01 1.81907609e-01 -7.63856530e-01
-4.11236107e-01 -9.11149383e-01 -6.22513413e-01 4.95185643e-01
7.54315257e-01 -3.12452704e-01 -6.23197258e-01 1.97413826e+00
2.38817573e-01 4.92690176e-01 8.11706409e-02 1.01109195e+00
4.35229152e-01 5.84732354e-01 4.52745035e-02 -2.28156716e-01
1.16191339e+00 -4.33286726e-01 -9.35748756e-01 -1.30420968e-01
-2.58047017e-04 -7.22978592e-01 8.79845023e-01 7.64376402e-01
-1.20998347e+00 -7.44834960e-01 -1.20969009e+00 1.48857862e-01
-4.76050556e-01 5.61146975e-01 2.31735319e-01 7.72358358e-01
-1.11630499e+00 8.64878237e-01 -6.56150401e-01 2.53665168e-02
1.16797127e-01 5.56056559e-01 -3.08733016e-01 5.05953491e-01
-1.39815354e+00 8.50250661e-01 2.80427545e-01 1.69126943e-01
-7.97285259e-01 -7.22386897e-01 -5.58134437e-01 3.05001765e-01
4.19582985e-03 -2.51745909e-01 1.21315897e+00 -1.16318989e+00
-1.64276981e+00 3.95543367e-01 -1.59310341e-01 -6.79569721e-01
3.40884179e-01 -1.94589466e-01 -6.07428908e-01 4.62414056e-01
-3.15929681e-01 5.84925234e-01 1.36690795e+00 -1.08883119e+00
-6.26326203e-01 -4.58909750e-01 -1.41179651e-01 1.70263007e-01
-4.85849142e-01 1.84944980e-02 -2.78326180e-02 -1.16541207e+00
-1.68253370e-02 -5.90162277e-01 7.87451640e-02 -2.05952615e-01
-3.26774389e-01 3.70134786e-02 7.85650671e-01 -8.79178822e-01
1.27699435e+00 -2.53518963e+00 3.74428511e-01 2.29547739e-01
-4.85606901e-02 3.82057905e-01 -3.15464199e-01 2.65922427e-01
-3.60254884e-01 -3.25299591e-01 -2.82190770e-01 -2.71120936e-01
6.50566667e-02 -3.04865181e-01 -4.04022932e-01 6.30378187e-01
1.91908255e-01 4.12971854e-01 -6.62800312e-01 -1.83358610e-01
3.02731991e-01 1.17605710e+00 -5.97133815e-01 2.42801625e-02
6.70350045e-02 5.51006198e-01 -1.95560642e-02 8.70829374e-02
5.70678294e-01 1.78508341e-01 1.72999829e-01 -4.09555554e-01
-2.23663017e-01 3.92408222e-01 -1.41496885e+00 1.50784719e+00
-6.03719056e-01 1.02883613e+00 6.63533628e-01 -7.93305159e-01
7.18496382e-01 7.01379240e-01 4.18566912e-01 -6.09858572e-01
1.36533797e-01 1.51539594e-01 2.05290020e-01 -1.33335605e-01
3.62771183e-01 -2.33326346e-01 2.30577275e-01 2.91429013e-01
4.93539184e-01 2.66533732e-01 -3.89238112e-02 -1.30818844e-01
8.77740264e-01 3.57771553e-02 2.11744770e-01 -4.90931988e-01
8.08816135e-01 -4.74992156e-01 1.51468217e-01 8.33173394e-01
-2.93611109e-01 6.57440245e-01 1.09886117e-01 3.12083475e-02
-7.32356250e-01 -1.13793492e+00 -3.25661421e-01 1.48190117e+00
-2.97008157e-01 -1.53189609e-02 -1.04990470e+00 -3.03917646e-01
-2.83695310e-01 9.00293291e-01 -5.60217083e-01 -5.16415350e-02
-7.80232072e-01 -3.95754397e-01 7.49419451e-01 5.37301600e-01
1.53771088e-01 -1.04051328e+00 -6.86690152e-01 4.37038302e-01
-5.08002192e-02 -8.71226192e-01 -5.87816596e-01 5.45715928e-01
-6.04124427e-01 -6.86027288e-01 -9.18460429e-01 -6.95678115e-01
1.69365838e-01 1.67770106e-02 7.09359944e-01 -3.73023987e-01
-2.61652172e-02 4.19445068e-01 -5.92482127e-02 -3.47081363e-01
-1.87303767e-01 6.75318018e-02 9.13862884e-02 3.68099779e-01
1.71302259e-01 -9.64622438e-01 -6.00394785e-01 2.00932726e-01
-9.38427210e-01 -3.00411582e-01 4.38842386e-01 8.06168675e-01
3.30834568e-01 3.60110462e-01 8.27646554e-01 -5.17430425e-01
7.92043149e-01 1.87401678e-02 -4.42584425e-01 -7.24122971e-02
-5.03597856e-02 1.07034609e-01 6.72369540e-01 -6.60566390e-01
-1.17799020e+00 1.93408459e-01 -4.16442692e-01 -3.71514320e-01
-5.89176118e-01 2.61962533e-01 -1.19285472e-01 -2.81095445e-01
9.06497538e-01 4.76086497e-01 -6.69203475e-02 -5.68231463e-01
2.53342807e-01 6.03327930e-01 6.35551929e-01 -1.65747002e-01
5.52199900e-01 3.64048511e-01 -6.63318038e-02 -1.22586644e+00
-4.81056660e-01 -4.35179472e-01 -6.29516602e-01 -3.66302401e-01
8.48120332e-01 -7.21000075e-01 -8.17928016e-01 3.77471834e-01
-1.23355222e+00 -2.75460869e-01 -3.94714981e-01 7.25873768e-01
-5.90805590e-01 4.13088128e-02 -5.02317429e-01 -1.03389609e+00
7.89498836e-02 -1.08119583e+00 1.01877666e+00 -6.72757328e-02
-2.77935475e-01 -9.82976615e-01 -9.86404344e-03 -2.30938837e-01
5.59281230e-01 -1.18651271e-01 5.56097686e-01 -7.22851574e-01
-4.29972857e-01 -2.34170318e-01 2.24381253e-01 3.51482004e-01
8.27446207e-03 -2.19283402e-01 -1.59876800e+00 -1.91361696e-01
3.15743536e-01 1.15771450e-01 1.04423821e+00 8.10195029e-01
1.03369439e+00 -3.14419210e-01 2.05702204e-02 4.35149699e-01
1.16302049e+00 2.50189155e-01 4.70871061e-01 6.88835308e-02
2.79284120e-01 8.33219349e-01 2.01229274e-01 3.29847187e-01
-3.57287049e-01 7.31076300e-01 3.01561296e-01 -5.58550172e-02
-3.00832689e-01 -5.48610045e-03 3.49918246e-01 5.24740100e-01
-1.97201937e-01 -1.48916602e-01 -4.61673528e-01 5.98940969e-01
-1.40252757e+00 -1.23692989e+00 2.66638875e-01 2.14474034e+00
8.09073508e-01 1.34673929e-02 2.63481915e-01 8.64967585e-01
7.23440886e-01 2.78563142e-01 -2.03843966e-01 -1.93248048e-01
-2.86705226e-01 5.22962809e-01 3.19405943e-01 9.91915643e-01
-1.28232419e+00 5.50794005e-01 6.84934568e+00 1.04435647e+00
-1.38678932e+00 3.17063592e-02 2.21154362e-01 -2.70237565e-01
-1.25679653e-02 -6.16339207e-01 -5.17833292e-01 2.54452586e-01
1.17644811e+00 1.81937575e-01 6.61070228e-01 4.99468893e-01
3.41257960e-01 9.94904265e-02 -1.03558600e+00 8.71058524e-01
-8.81134123e-02 -1.10461056e+00 -3.31654936e-01 -2.97455788e-02
1.80194318e-01 -3.33479971e-01 4.82822686e-01 8.59606490e-02
-2.16550976e-01 -1.11686969e+00 1.04215586e+00 4.96119887e-01
4.34651822e-01 -7.19611526e-01 2.43426904e-01 3.44506055e-01
-1.17030489e+00 -4.31059301e-01 1.21010110e-01 5.23559861e-02
-8.79168212e-02 5.14937043e-01 -7.79168725e-01 1.70138553e-01
5.15912175e-01 3.81343484e-01 -3.90148193e-01 1.10447621e+00
-4.55266647e-02 8.59645426e-01 -4.19020653e-01 8.71744528e-02
2.40741879e-01 9.64521989e-02 8.68617237e-01 1.50359321e+00
4.41142648e-01 3.46399955e-02 -4.47469592e-01 7.24139750e-01
3.48971218e-01 -1.73755232e-02 -3.59304905e-01 4.83839698e-02
4.13294435e-01 1.02732468e+00 -4.60107207e-01 -9.10638571e-02
-1.34298325e-01 8.65762651e-01 3.11917216e-02 7.62382686e-01
-5.61992168e-01 -5.43440700e-01 5.99206150e-01 4.34201770e-02
8.67790163e-01 -1.52229354e-01 -2.86234409e-01 -7.67849624e-01
-6.53181076e-02 -7.96411455e-01 2.15553552e-01 -7.62995899e-01
-8.92392814e-01 9.34867620e-01 -1.71114817e-01 -1.08719587e+00
-3.90427381e-01 -8.86430562e-01 -4.07284260e-01 1.18951404e+00
-1.36932361e+00 -9.32533205e-01 2.70629108e-01 7.61618435e-01
3.70427102e-01 -9.68500003e-02 7.64335632e-01 4.07013327e-01
-2.19603226e-01 6.32142842e-01 -2.69109290e-02 7.91406706e-02
4.37314391e-01 -1.42412579e+00 -1.27539277e-01 9.04822409e-01
5.01457036e-01 6.78885520e-01 1.16704917e+00 -9.68714207e-02
-1.08602178e+00 -8.47655177e-01 9.71064091e-01 1.04621118e-02
5.37935436e-01 -3.81246120e-01 -9.92846966e-01 4.44091082e-01
4.72399563e-01 -8.99224356e-02 5.70864856e-01 4.96362858e-02
-2.63026416e-01 -2.97410190e-01 -9.55736339e-01 4.39327776e-01
6.85654461e-01 -8.69401515e-01 -7.57604420e-01 7.61711597e-02
4.37106162e-01 -1.91050410e-01 -5.44750333e-01 2.51097232e-01
4.42235351e-01 -1.31196558e+00 1.26436937e+00 -3.89969349e-01
-8.85274448e-03 -4.70400363e-01 -2.53131747e-01 -1.38772511e+00
-4.71486360e-01 -4.71784741e-01 -2.66835038e-02 8.80879641e-01
3.28914225e-01 -5.49922466e-01 6.15983903e-01 -5.29623479e-02
-2.11206019e-01 -3.94733220e-01 -1.19742596e+00 -5.58912933e-01
-2.37951502e-02 -5.56765556e-01 3.10397536e-01 7.34661162e-01
-5.05707636e-02 1.64271176e-01 -3.01099837e-01 4.25593317e-01
5.02468228e-01 -1.42117158e-01 1.12754747e-01 -1.22377884e+00
-6.64601266e-01 -6.22807264e-01 -2.56355673e-01 -1.15909302e+00
1.83921456e-01 -5.82374930e-01 2.37655208e-01 -7.73455501e-01
-6.53776884e-01 7.11380765e-02 -6.82102621e-01 -1.36768678e-03
4.79632020e-02 2.60702342e-01 3.41301113e-02 -2.28662774e-01
-1.40214533e-01 3.47349077e-01 1.19941640e+00 -1.94850881e-02
-4.07237619e-01 3.82446736e-01 -4.52005714e-01 6.50735259e-01
6.33586287e-01 -1.42424777e-01 -4.23706591e-01 -1.71254188e-01
-4.52561527e-02 2.63994664e-01 6.88210428e-01 -1.30461824e+00
3.52607936e-01 2.36492425e-01 6.38685167e-01 -5.39309978e-01
8.19726169e-01 -9.46182609e-01 2.74853129e-02 2.01023161e-01
-6.96780384e-01 -3.49653035e-01 4.02012318e-01 6.15310609e-01
-5.58123708e-01 -2.37444609e-01 9.03954804e-01 -5.37817851e-02
-6.22349203e-01 -2.08102152e-01 -7.41851985e-01 -3.01180810e-01
5.36832511e-01 -2.45966762e-01 1.22698292e-01 -4.64058876e-01
-1.29743946e+00 -5.96686482e-01 -1.56874180e-01 -1.08436324e-01
4.95598108e-01 -1.16988766e+00 -5.28444290e-01 3.26344401e-01
-4.11451757e-01 -6.23122633e-01 5.27866602e-01 9.25426781e-01
-1.42682984e-01 6.35859787e-01 -1.64935887e-01 -5.50116479e-01
-1.32873452e+00 4.02277023e-01 8.16352606e-01 1.30329520e-01
-4.22739804e-01 1.08269882e+00 2.36787990e-01 -8.57772082e-02
5.94412088e-01 -3.59594345e-01 -4.80341256e-01 2.46111661e-01
7.37891734e-01 2.66262352e-01 2.89193124e-01 -8.13791275e-01
-2.97281027e-01 4.43861693e-01 2.74214238e-01 -6.29462063e-01
1.42899454e+00 -9.04379562e-02 1.14929333e-01 6.11572504e-01
1.03288317e+00 3.46993029e-01 -1.48231328e+00 -2.41729498e-01
-1.85944978e-02 -3.05951893e-01 4.48669791e-01 -8.81886601e-01
-9.17509556e-01 1.24226689e+00 7.76585400e-01 4.49016452e-01
1.60741031e+00 -3.04055095e-01 1.55790597e-01 1.11567244e-01
6.20498545e-02 -1.05271828e+00 3.41033936e-02 3.73246908e-01
1.11083424e+00 -6.07390225e-01 -3.05382639e-01 -3.14165890e-01
-4.14241403e-01 1.18544018e+00 7.82863274e-02 -2.10620746e-01
8.49910498e-01 3.89334142e-01 1.99468344e-01 1.39501095e-01
-5.34544647e-01 -2.86515534e-01 5.29116571e-01 8.01869690e-01
5.14592350e-01 -2.98496112e-02 2.17837304e-01 6.54743493e-01
-3.49251956e-01 -3.81114095e-01 1.71723008e-01 6.63912833e-01
-6.20291412e-01 -8.90697241e-01 -7.70287871e-01 -4.63517494e-02
-6.92674458e-01 -2.20852822e-01 -3.52661818e-01 6.75360143e-01
8.34735930e-02 8.19382787e-01 8.35880712e-02 -2.16253221e-01
2.41321713e-01 4.12521362e-01 5.85320950e-01 -2.15614706e-01
-9.11931276e-01 5.84911287e-01 -1.57282680e-01 -3.09684634e-01
-5.03042877e-01 -7.53257573e-01 -1.02252579e+00 2.48799518e-01
-3.09265822e-01 2.02198029e-01 7.28512645e-01 8.92049491e-01
1.50948405e-01 9.02296066e-01 3.56828481e-01 -1.12239361e+00
-4.60351974e-01 -1.10869837e+00 -8.75634730e-01 2.21076533e-01
1.00578189e+00 -5.36555171e-01 -6.09182894e-01 1.51407227e-01] | [15.279093742370605, 5.68610954284668] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.