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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ba628895-75f8-4d99-9778-0d7e9f7a0d35 | combining-reinforcement-learning-and-barrier | 2306.07013 | null | https://arxiv.org/abs/2306.07013v1 | https://arxiv.org/pdf/2306.07013v1.pdf | Combining Reinforcement Learning and Barrier Functions for Adaptive Risk Management in Portfolio Optimization | Reinforcement learning (RL) based investment strategies have been widely adopted in portfolio management (PM) in recent years. Nevertheless, most RL-based approaches may often emphasize on pursuing returns while ignoring the risks of the underlying trading strategies that may potentially lead to great losses especially under high market volatility. Therefore, a risk-manageable PM investment framework integrating both RL and barrier functions (BF) is proposed to carefully balance the needs for high returns and acceptable risk exposure in PM applications. Up to our understanding, this work represents the first attempt to combine BF and RL for financial applications. While the involved RL approach may aggressively search for more profitable trading strategies, the BF-based risk controller will continuously monitor the market states to dynamically adjust the investment portfolio as a controllable measure for avoiding potential losses particularly in downtrend markets. Additionally, two adaptive mechanisms are provided to dynamically adjust the impact of risk controllers such that the proposed framework can be flexibly adapted to uptrend and downtrend markets. The empirical results of our proposed framework clearly reveal such advantages against most well-known RL-based approaches on real-world data sets. More importantly, our proposed framework shed lights on many possible directions for future investigation. | ['Vincent Tam', 'Hejun Huang', 'Zhenglong Li'] | 2023-06-12 | null | null | null | null | ['portfolio-optimization'] | ['time-series'] | [-2.03830078e-01 2.02514693e-01 -2.89915651e-01 -8.37491080e-03
-5.44823945e-01 -5.24751306e-01 5.75595260e-01 1.95114613e-01
-3.77852768e-01 7.76407301e-01 -1.88954756e-01 -4.63613272e-01
-5.87011337e-01 -1.23401034e+00 -2.94437408e-01 -7.10834861e-01
-2.77969599e-01 2.37798363e-01 2.41479158e-01 -2.99671561e-01
7.70473361e-01 5.99297345e-01 -1.25459588e+00 -2.83852190e-01
8.97004426e-01 1.37785673e+00 -1.79544076e-01 1.83107078e-01
-3.08907218e-02 7.64506698e-01 -6.38691902e-01 -5.19336820e-01
7.53664494e-01 -1.78984404e-01 -6.77970201e-02 -8.57236981e-02
-4.27443236e-01 -3.69011998e-01 2.31786117e-01 1.03478932e+00
3.88248295e-01 2.20311850e-01 5.63349843e-01 -1.10761428e+00
-2.20417604e-01 7.74712026e-01 -7.91079044e-01 5.09957016e-01
-1.02151468e-01 4.48664337e-01 1.13360798e+00 -4.34711754e-01
-6.73805550e-02 9.60772455e-01 1.92954987e-01 2.80665874e-01
-9.72538948e-01 -7.61338532e-01 5.20690978e-01 -1.50525376e-01
-7.19924808e-01 -4.15109880e-02 1.04038918e+00 -4.62327003e-01
8.16421509e-01 1.91599220e-01 8.51722479e-01 6.17464006e-01
6.54892981e-01 3.85234654e-01 1.33503127e+00 -2.82035798e-01
4.33647335e-01 3.39566946e-01 -2.07939729e-01 -1.54027957e-02
5.90138376e-01 6.19390190e-01 -1.85770586e-01 -2.82455146e-01
9.01731849e-01 1.87089406e-02 7.36032948e-02 -2.49678358e-01
-7.67576635e-01 9.36957538e-01 1.62337884e-01 8.89739096e-02
-8.17326963e-01 5.38881160e-02 3.44046235e-01 6.21785939e-01
4.30551797e-01 6.94114864e-01 -4.15785104e-01 -3.07428986e-01
-7.73724020e-01 4.15432543e-01 6.71924174e-01 5.01042545e-01
3.03135335e-01 4.86269176e-01 -1.96504205e-01 4.01546896e-01
3.59929830e-01 1.24413468e-01 4.80466664e-01 -8.76830995e-01
8.27191949e-01 6.01913810e-01 4.82469797e-01 -9.72825527e-01
-1.71772733e-01 -6.95243895e-01 -4.17263031e-01 7.19254673e-01
2.46173203e-01 -4.31669354e-01 -3.52652147e-02 1.48838520e+00
2.70053774e-01 8.91109332e-02 1.33806944e-01 3.22974026e-01
-3.18146855e-01 4.86488461e-01 -8.18957016e-02 -6.11992896e-01
1.12869596e+00 -6.10087931e-01 -7.36448824e-01 5.98374270e-02
8.53593871e-02 -5.19348383e-01 9.65681851e-01 6.18452847e-01
-1.32782388e+00 -2.14707866e-01 -1.13431716e+00 1.06659853e+00
-1.54259056e-01 -2.82699645e-01 4.36530113e-01 7.71129668e-01
-6.88901186e-01 9.19843197e-01 -8.38700831e-01 1.54244214e-01
2.29573101e-01 3.56915414e-01 6.04601026e-01 8.16999972e-01
-1.45286298e+00 9.34349358e-01 4.29033458e-01 4.16381896e-01
-8.47090900e-01 -5.38605630e-01 -1.71183228e-01 1.74643084e-01
9.53517437e-01 -2.90982217e-01 1.29078114e+00 -8.57422054e-01
-1.95300341e+00 2.11535156e-01 7.78714418e-01 -1.04076302e+00
1.15246105e+00 -4.22921807e-01 -3.33369493e-01 1.24225788e-01
-4.04487103e-01 -1.03644945e-01 9.21550691e-01 -7.68749118e-01
-6.97885156e-01 -1.62352130e-01 2.19403327e-01 4.19022858e-01
-5.15236259e-01 4.47096191e-02 4.89883810e-01 -1.08107114e+00
-2.10868895e-01 -7.30491459e-01 -4.18405443e-01 -2.87626535e-01
-1.62875623e-01 9.79435369e-02 3.18265468e-01 -2.79450983e-01
1.50671744e+00 -1.63151920e+00 -1.21623300e-01 3.22137892e-01
-2.34616712e-01 2.47808456e-01 2.61310518e-01 7.63170421e-01
8.85047298e-03 1.17050670e-01 -5.48867956e-02 2.58246928e-01
1.21516883e-01 -1.02718540e-01 -6.59989119e-01 2.26541594e-01
3.22262794e-01 6.46187603e-01 -4.82529074e-01 6.26068264e-02
2.49864429e-01 -6.87908530e-02 -4.68761742e-01 5.11553586e-01
-3.72314304e-01 2.71171153e-01 -8.34203422e-01 6.94264114e-01
4.89825606e-01 1.84045210e-01 8.20163414e-02 3.11612576e-01
-4.16796237e-01 1.44310042e-01 -1.35334527e+00 3.49648446e-01
-5.63245893e-01 -2.04564873e-02 3.37383784e-02 -1.14348853e+00
1.28596532e+00 2.74073422e-01 5.09262979e-01 -6.72058225e-01
1.07433692e-01 2.86884785e-01 5.84367737e-02 -9.33454558e-02
4.40881342e-01 -5.36922872e-01 6.14986382e-02 6.67694569e-01
-6.98108017e-01 -2.11399198e-02 -8.39583725e-02 -4.63024199e-01
7.06189156e-01 1.37584597e-01 5.49795687e-01 -3.34854633e-01
8.17787349e-01 -3.65700275e-01 8.10926497e-01 5.15388370e-01
-5.22713244e-01 -2.07195655e-01 1.00077438e+00 -2.61066079e-01
-5.26352346e-01 -9.25492346e-01 7.22212642e-02 6.72910213e-01
-1.24664456e-01 2.26059154e-01 -5.20810068e-01 -5.84469497e-01
2.82707334e-01 8.22326183e-01 -4.24244165e-01 -1.83370322e-01
-5.99721253e-01 -9.79874969e-01 2.32069299e-01 3.99395376e-01
5.97518563e-01 -1.37106681e+00 -1.24686396e+00 6.32721603e-01
6.13252521e-01 -4.93185878e-01 -1.42616719e-01 3.32110316e-01
-1.16687036e+00 -1.15880394e+00 -7.49366283e-01 -1.63840503e-02
3.65449250e-01 -8.10661316e-02 8.66372228e-01 -7.86738768e-02
1.18154228e-01 3.63983333e-01 -2.64141679e-01 -6.60997450e-01
-3.96822155e-01 -1.98480651e-01 -6.55715987e-02 1.01609804e-01
1.07695386e-01 -4.99065280e-01 -8.02974224e-01 3.68962497e-01
-8.16479146e-01 -4.29851830e-01 4.65123266e-01 8.69096875e-01
6.06652379e-01 6.46418989e-01 1.31664038e+00 -9.20692503e-01
1.05460835e+00 -6.16667032e-01 -1.46318388e+00 4.23283309e-01
-1.34413970e+00 1.26130581e-01 7.04693377e-01 -5.11500001e-01
-1.51990283e+00 -6.68403029e-01 7.56099746e-02 -1.73957542e-01
3.05368155e-01 4.92722780e-01 -1.34973824e-01 -4.50830534e-02
3.96928713e-02 1.04137294e-01 -4.20979336e-02 -3.63396674e-01
1.09139495e-01 3.54661256e-01 4.89984341e-02 -7.32467413e-01
8.70613813e-01 1.96363971e-01 1.15135014e-01 -2.30595052e-01
-5.09478390e-01 9.66704786e-02 -3.80518109e-01 -2.91789949e-01
3.64806801e-01 -6.45973921e-01 -8.92950773e-01 5.12649536e-01
-3.48231405e-01 -1.05657466e-01 -3.39411646e-01 3.57748806e-01
-9.15358305e-01 2.26858318e-01 -6.18204534e-01 -1.52245367e+00
-5.80291986e-01 -1.02433419e+00 1.72174126e-01 4.64040518e-01
-9.92930084e-02 -1.13960254e+00 1.68233261e-01 2.11993128e-01
6.79127157e-01 6.57560587e-01 9.84069943e-01 -8.27337980e-01
-8.36179733e-01 -2.72234648e-01 2.17058331e-01 5.08401930e-01
3.65688235e-01 1.73930213e-01 -5.14963865e-01 -4.58156705e-01
6.14089966e-01 -2.13326320e-01 5.27190983e-01 3.23359847e-01
7.29991078e-01 -5.96363604e-01 2.26522297e-01 2.67816156e-01
1.53659785e+00 9.18965936e-01 3.75325173e-01 1.16587591e+00
4.46145311e-02 9.06412363e-01 1.37130225e+00 1.06569505e+00
7.49377608e-02 5.29228568e-01 6.10203683e-01 3.42916578e-01
9.02977586e-01 -2.90868074e-01 5.76725125e-01 3.58706713e-01
-3.92449200e-02 5.06695919e-02 -5.97312868e-01 1.02784418e-01
-1.73355293e+00 -1.02430677e+00 6.15753412e-01 2.66453600e+00
7.29298651e-01 9.92904186e-01 5.20247936e-01 2.00976580e-01
5.85120916e-01 3.20810735e-01 -9.18386042e-01 -7.73454130e-01
5.56852929e-02 1.26556292e-01 6.14300609e-01 1.56469017e-01
-9.08199906e-01 5.11215031e-01 6.01055050e+00 6.23583317e-01
-1.15158331e+00 -4.67618376e-01 9.62212265e-01 -2.12575138e-01
-5.10015428e-01 2.13043362e-01 -8.57486248e-01 5.60367167e-01
1.02322686e+00 -8.72385085e-01 2.58632779e-01 8.90233815e-01
6.65207922e-01 -4.73279953e-02 -7.75584459e-01 3.20510030e-01
-6.61624551e-01 -1.14093268e+00 -1.55264959e-01 2.51595974e-01
3.79388541e-01 -6.57541811e-01 3.92284036e-01 3.83573055e-01
1.37983009e-01 -7.55068481e-01 8.62573028e-01 8.39188218e-01
1.95676442e-02 -1.41655397e+00 7.69332826e-01 4.70544785e-01
-1.12641704e+00 -5.41191459e-01 -2.80392170e-01 -2.57344276e-01
1.73065126e-01 4.44806695e-01 -4.47793871e-01 5.00701249e-01
3.64533216e-01 4.41493243e-01 -3.47847849e-01 7.77742028e-01
-1.50552586e-01 5.57477057e-01 -9.38517228e-02 -2.28758961e-01
3.04307431e-01 -6.98660493e-01 5.20355046e-01 5.55223346e-01
4.00481850e-01 -3.12190428e-02 6.81316340e-03 9.88784075e-01
4.27639872e-01 1.46679193e-01 -6.13738537e-01 -4.01748836e-01
5.59207857e-01 1.00963247e+00 -8.59734535e-01 -1.24930166e-01
-3.39784592e-01 2.59726107e-01 9.27203670e-02 1.95566431e-01
-8.48565936e-01 -2.93817014e-01 5.31839728e-01 3.91436577e-01
3.53165954e-01 -4.60336953e-02 -4.41857070e-01 -9.32940125e-01
1.36429980e-01 -8.84294629e-01 6.75974309e-01 8.19968730e-02
-1.24900830e+00 4.14971143e-01 1.09478869e-01 -1.55612814e+00
-4.17750835e-01 -3.45194042e-01 -1.03833032e+00 6.56277359e-01
-1.96296525e+00 -4.05038387e-01 5.16152799e-01 2.03978762e-01
5.33359349e-01 -4.67033446e-01 3.24943364e-01 -2.06471622e-01
-8.80456209e-01 4.85038161e-01 3.17525119e-01 -3.87962729e-01
3.41992527e-01 -1.19473386e+00 1.70054421e-01 8.45588088e-01
-2.32598364e-01 6.18883431e-01 6.80992901e-01 -8.37659538e-01
-1.06676459e+00 -7.13475406e-01 2.41594255e-01 2.17685327e-02
1.13302827e+00 4.81002107e-02 -9.87897515e-01 3.49455297e-01
2.75024354e-01 -4.93678331e-01 4.69101071e-01 -3.64309162e-01
9.99768004e-02 -4.78489906e-01 -1.22786975e+00 6.95588291e-01
2.42167115e-01 -6.53123856e-02 -6.14018440e-01 -1.42114550e-01
5.97255111e-01 -7.97885209e-02 -9.87432063e-01 4.71804529e-01
4.21897113e-01 -1.32311893e+00 8.26433778e-01 -2.83162743e-01
5.16082272e-02 9.41581950e-02 1.17247351e-01 -1.14872479e+00
7.48917460e-02 -1.12812638e+00 -2.56511271e-01 1.44622302e+00
2.67340362e-01 -1.12128592e+00 7.65125513e-01 8.65407884e-01
4.82606947e-01 -1.16510284e+00 -1.04890370e+00 -9.84351933e-01
4.26121086e-01 -3.15299243e-01 7.36905038e-01 5.38829207e-01
-8.88246596e-02 -5.40744662e-01 -4.46010232e-01 -4.59482484e-02
8.35644960e-01 3.92704606e-01 4.14507240e-01 -8.70816827e-01
-6.48113847e-01 -8.44825625e-01 7.78301880e-02 -3.56883764e-01
-2.78049782e-02 -2.74981707e-01 -4.25778598e-01 -6.84355915e-01
-2.02301338e-01 -4.53996032e-01 -1.06278861e+00 1.06708318e-01
-1.67432487e-01 -5.59818625e-01 3.71289462e-01 1.47476494e-01
-5.87390959e-02 7.86894023e-01 1.13509750e+00 1.24446221e-01
-5.16054988e-01 8.01767170e-01 -8.62974346e-01 7.37937987e-01
1.01142263e+00 -3.75892878e-01 -6.87887192e-01 3.87724012e-01
3.35133702e-01 5.92882752e-01 2.04091538e-02 -6.40961170e-01
-6.48721866e-03 -8.92165005e-01 -6.11439906e-02 -6.51330531e-01
-1.82818234e-01 -8.61028612e-01 2.12578103e-01 7.98965514e-01
-5.60912609e-01 4.34019744e-01 1.37801081e-01 7.96848893e-01
-3.06774676e-01 -1.80335358e-01 8.38447273e-01 -2.49160960e-01
-2.03600451e-01 1.70406178e-01 -5.59309781e-01 6.62934780e-02
1.54614985e+00 -3.50754201e-01 -1.03887208e-01 -4.22044665e-01
-4.47039396e-01 5.77234864e-01 1.52468488e-01 3.96385133e-01
6.34460330e-01 -1.07153904e+00 -3.47320378e-01 2.81223375e-02
-2.92334676e-01 -3.41926366e-01 1.79740787e-01 5.82613587e-01
-4.36969548e-01 5.05298495e-01 -3.79712462e-01 1.93195939e-01
-7.02277720e-01 5.59801579e-01 6.43508494e-01 -8.59075129e-01
-6.69967949e-01 5.06574810e-01 1.47321969e-01 1.00546405e-01
4.23268914e-01 -3.96052718e-01 -4.31518048e-01 4.78788793e-01
4.53843504e-01 7.30930030e-01 -8.57667997e-02 -3.34310792e-02
-1.73958138e-01 4.66292024e-01 -8.57713744e-02 -3.21766198e-01
1.50080192e+00 -2.43311927e-01 1.82360217e-01 5.40229082e-01
3.48360479e-01 2.73114396e-03 -1.76727796e+00 -3.17689180e-02
8.63090098e-01 -5.15823781e-01 -8.71771723e-02 -6.34405673e-01
-1.27529275e+00 7.61222959e-01 3.79568487e-01 4.39685315e-01
1.40799105e+00 -7.90971935e-01 6.42735422e-01 2.25066513e-01
5.16409338e-01 -1.46975076e+00 2.82179952e-01 -5.41173331e-02
9.08505619e-01 -8.27454805e-01 9.56542715e-02 3.25138867e-02
-1.05715430e+00 1.22546184e+00 7.39343703e-01 -4.83388752e-01
8.55525255e-01 3.98973256e-01 5.19003198e-02 3.81232947e-02
-1.10537052e+00 2.44513705e-01 6.60893247e-02 2.08562493e-01
1.81503564e-01 1.20817982e-01 -5.35768390e-01 8.36578071e-01
-4.57856478e-03 -1.02886453e-01 7.57298827e-01 1.23758721e+00
-5.59905708e-01 -1.36207938e+00 -5.39520741e-01 5.14992595e-01
-7.26609051e-01 1.23625509e-01 -1.92721128e-01 1.14901078e+00
-3.27252358e-01 6.14061534e-01 -6.78290501e-02 5.56148104e-02
6.37274325e-01 -4.74514961e-02 4.18075696e-02 -4.38054413e-01
-1.03777707e+00 6.69976652e-01 -7.41609633e-02 -4.66327071e-01
-2.03588471e-01 -8.96937609e-01 -1.23264015e+00 2.94705220e-02
-4.26679730e-01 8.57573748e-02 6.54665753e-02 8.00076723e-01
6.71424670e-03 4.32562321e-01 1.21326792e+00 -3.53108943e-01
-1.48590243e+00 -5.20181119e-01 -1.03196442e+00 -1.92395702e-01
1.71535268e-01 -1.02491307e+00 -5.99949062e-01 -7.16235161e-01] | [4.568074703216553, 3.9259541034698486] |
a0acce55-75bc-419b-a253-0cb7b01ba7a8 | language-agnostic-representation-learning-of-1 | 2103.11318 | null | https://arxiv.org/abs/2103.11318v1 | https://arxiv.org/pdf/2103.11318v1.pdf | Language-Agnostic Representation Learning of Source Code from Structure and Context | Source code (Context) and its parsed abstract syntax tree (AST; Structure) are two complementary representations of the same computer program. Traditionally, designers of machine learning models have relied predominantly either on Structure or Context. We propose a new model, which jointly learns on Context and Structure of source code. In contrast to previous approaches, our model uses only language-agnostic features, i.e., source code and features that can be computed directly from the AST. Besides obtaining state-of-the-art on monolingual code summarization on all five programming languages considered in this work, we propose the first multilingual code summarization model. We show that jointly training on non-parallel data from multiple programming languages improves results on all individual languages, where the strongest gains are on low-resource languages. Remarkably, multilingual training only from Context does not lead to the same improvements, highlighting the benefits of combining Structure and Context for representation learning on code. | ['Stephan Günnemann', 'Jure Leskovec', 'Michele Catasta', 'Tobias Kirschstein', 'Daniel Zügner'] | 2021-03-21 | language-agnostic-representation-learning-of | https://openreview.net/forum?id=Xh5eMZVONGF | https://openreview.net/pdf?id=Xh5eMZVONGF | iclr-2021-1 | ['code-summarization'] | ['computer-code'] | [-4.89464146e-04 -4.14945707e-02 -6.25157773e-01 -3.28160822e-01
-1.12374115e+00 -8.37553918e-01 4.59320694e-01 7.82087207e-01
-1.42562643e-01 1.89778358e-01 6.37549698e-01 -7.04718649e-01
4.14439887e-01 -3.90598267e-01 -7.46129692e-01 2.50960924e-02
-1.74212977e-01 -1.47968084e-01 1.53006017e-01 -3.77445191e-01
4.39293921e-01 -6.13146573e-02 -1.46325839e+00 7.36317277e-01
1.12171555e+00 2.47189820e-01 3.34389478e-01 7.60688543e-01
-8.40932608e-01 1.27584708e+00 -4.23804283e-01 -5.28028488e-01
-2.24649161e-02 -3.66554022e-01 -9.70816374e-01 -3.37254435e-01
7.88244128e-01 7.12080970e-02 -1.23478189e-01 1.03947341e+00
4.48839404e-02 -4.76465046e-01 4.00978595e-01 -9.01594758e-01
-6.43146157e-01 1.13466382e+00 -8.76217008e-01 1.32227659e-01
6.62660897e-01 -1.27798647e-01 1.41856837e+00 -8.35066199e-01
7.14627445e-01 9.75506246e-01 9.40197349e-01 2.82318205e-01
-1.55676556e+00 -2.36084610e-01 3.56728941e-01 -1.53190494e-01
-9.99112070e-01 -3.69877219e-01 6.75825119e-01 -7.89499044e-01
1.56326008e+00 -4.57339473e-02 2.56828725e-01 6.03921652e-01
4.14509952e-01 9.61477041e-01 7.27366507e-01 -7.19135165e-01
-2.53863186e-01 1.00776359e-01 6.08768940e-01 1.15568757e+00
4.42120254e-01 -3.79626989e-01 -2.20653802e-01 -5.33905506e-01
7.89646432e-02 -1.59839556e-01 -5.49360737e-02 -6.77002430e-01
-1.28835213e+00 8.70300591e-01 2.68242210e-01 4.28373516e-01
2.51333844e-02 4.14087027e-01 9.61441278e-01 5.65726101e-01
1.72855273e-01 4.47709560e-01 -7.39120781e-01 -3.31761628e-01
-1.01561546e+00 1.70010164e-01 9.10823703e-01 1.35427415e+00
1.12664497e+00 1.53341711e-01 1.29623875e-01 9.21023369e-01
1.88714504e-01 6.09963536e-01 5.84046066e-01 -7.99483538e-01
1.02369726e+00 1.07042086e+00 -3.14633816e-01 -7.55822122e-01
-3.72537285e-01 -4.15835112e-01 -2.19253987e-01 -5.78438677e-02
3.14319640e-01 -1.70734599e-01 -4.67038721e-01 1.77829289e+00
-1.81900933e-01 -4.17979956e-01 2.42363051e-01 1.91964984e-01
9.08082128e-01 5.99027574e-01 -6.25033677e-02 1.48937270e-01
1.18279338e+00 -1.15989625e+00 -1.36104316e-01 -6.19806051e-01
1.37426579e+00 -9.88486826e-01 1.09027374e+00 2.66952187e-01
-9.81883228e-01 -4.38261867e-01 -1.08871496e+00 -5.18137753e-01
-2.01422185e-01 4.24580365e-01 8.92821193e-01 6.70469463e-01
-1.22587800e+00 3.93227726e-01 -9.34297204e-01 -4.39882249e-01
3.84450816e-02 8.37817639e-02 -5.07057309e-01 -3.24304253e-01
-3.19298327e-01 6.64995313e-01 2.45956346e-01 -7.07811952e-01
-6.42683029e-01 -8.32194328e-01 -1.15371752e+00 8.55127499e-02
3.16254348e-01 -5.57467759e-01 1.56450188e+00 -1.27904224e+00
-1.10509896e+00 9.26322758e-01 -4.28636134e-01 -3.94230962e-01
-5.26526535e-04 -4.23300117e-01 -2.01757923e-01 -2.16532290e-01
2.69348264e-01 2.13402539e-01 5.64800560e-01 -1.28112006e+00
-7.99567640e-01 -1.83625624e-01 4.48348105e-01 -1.96101546e-01
-3.32980365e-01 5.04342973e-01 -4.58521307e-01 -6.23101771e-01
-1.37217447e-01 -9.53741193e-01 -1.67569309e-01 -3.86882395e-01
-2.37622276e-01 -2.72998542e-01 4.22973394e-01 -9.72997427e-01
1.56242323e+00 -2.24545765e+00 5.27828336e-01 -9.52185318e-02
1.82768628e-01 5.06153591e-02 -6.75989628e-01 6.60713613e-01
-2.57704109e-01 2.66568363e-01 -5.69230616e-01 -2.97979593e-01
6.51873499e-02 1.67222843e-01 -5.30151784e-01 3.60292822e-01
2.89018482e-01 7.96160579e-01 -1.06514013e+00 -4.42697048e-01
-3.25660825e-01 1.38378441e-01 -1.15960169e+00 1.80705905e-01
-3.88604790e-01 1.65928110e-01 -3.98811251e-01 5.35650432e-01
4.20889944e-01 -1.67346358e-01 5.77840090e-01 2.93887198e-01
-3.05021495e-01 6.90431178e-01 -7.04525948e-01 2.41514277e+00
-1.18166578e+00 6.35667443e-01 1.11013666e-01 -8.45572114e-01
7.89993584e-01 2.77005881e-01 2.55668163e-01 -6.10068977e-01
-3.57432932e-01 5.02337754e-01 8.68145227e-02 -6.08300030e-01
6.21922970e-01 3.33384335e-01 -7.39757121e-01 7.85561085e-01
2.48074666e-01 -2.11488336e-01 4.83866066e-01 5.94016850e-01
1.22812283e+00 4.62283850e-01 7.50141740e-01 -5.34051657e-01
6.76559150e-01 2.10546270e-01 5.71435213e-01 6.99414253e-01
3.31773937e-01 2.59309471e-01 1.02295232e+00 -3.52062136e-01
-1.07968867e+00 -8.70151639e-01 7.34447241e-02 1.60755837e+00
-2.85532534e-01 -1.31462359e+00 -6.96658134e-01 -1.13172233e+00
1.06299870e-01 6.63170755e-01 -4.34625298e-01 1.20618753e-01
-1.07731628e+00 -4.81424928e-01 7.09612727e-01 6.03500783e-01
-3.18707824e-02 -5.17044485e-01 -5.62674522e-01 2.12609023e-01
-1.02865517e-01 -8.37698340e-01 -5.52951157e-01 2.30312422e-01
-1.12523246e+00 -1.06755817e+00 -3.34566802e-01 -7.69381285e-01
5.17323732e-01 2.74971604e-01 1.59504807e+00 5.29600859e-01
-1.23777628e-01 4.97553527e-01 -3.78172964e-01 -7.47438297e-02
-1.04685581e+00 6.14173591e-01 -4.79931414e-01 -6.39146209e-01
1.50439903e-01 -7.30717301e-01 2.01498177e-02 -3.22845072e-01
-7.41702676e-01 3.57552841e-02 8.61506045e-01 7.15402484e-01
4.21185941e-02 -4.11372036e-01 4.11595911e-01 -1.33055639e+00
3.86265457e-01 -7.96163082e-01 -5.90541244e-01 3.58766228e-01
-4.29072440e-01 5.40554821e-01 1.08424127e+00 -4.02951725e-02
-1.10037231e+00 3.93901207e-02 -6.19449914e-02 1.89324245e-01
-1.40650392e-01 1.07855988e+00 1.64079189e-01 -4.90571745e-02
8.30192983e-01 2.06623316e-01 -3.65582585e-01 -7.55442500e-01
6.00614786e-01 5.14103234e-01 5.34984410e-01 -1.21297634e+00
6.41799510e-01 1.74632803e-01 -2.34942794e-01 -6.57195508e-01
-5.33655524e-01 -5.81926942e-01 -8.99047852e-01 5.33322334e-01
4.46477830e-01 -1.11619210e+00 3.49919684e-02 1.41993299e-01
-1.33815706e+00 -2.94608116e-01 -8.42466056e-02 8.30710307e-02
-4.31380749e-01 8.18699062e-01 -6.64539933e-01 -2.56376982e-01
-1.65769503e-01 -1.38002598e+00 1.10711646e+00 -2.40271762e-01
-3.02556455e-01 -1.03210521e+00 4.39604998e-01 1.38618737e-01
5.44285476e-01 1.73756450e-01 1.67969191e+00 -6.01805627e-01
-6.35607958e-01 -6.05028011e-02 -2.81940609e-01 2.14931563e-01
1.67154208e-01 1.92644656e-01 -7.57594466e-01 -5.12648165e-01
-2.03629687e-01 -4.80634660e-01 1.00355053e+00 -2.35152140e-01
8.48901749e-01 -4.34637636e-01 -1.97655290e-01 7.09255159e-01
1.78508818e+00 -2.70369619e-01 2.47248068e-01 2.10161701e-01
8.46846282e-01 5.59476733e-01 8.57834592e-02 4.69810724e-01
6.98202252e-01 4.74192113e-01 2.06903458e-01 1.87110677e-01
-3.40502709e-01 -2.71465182e-01 8.56344819e-01 1.30534899e+00
1.54162601e-01 3.06456447e-01 -1.42594039e+00 9.35314357e-01
-1.73396289e+00 -7.16020584e-01 -4.33409691e-01 2.11644316e+00
1.17204690e+00 -1.02794215e-01 8.70409384e-02 -3.45184684e-01
2.61907756e-01 2.12833405e-01 -2.02354610e-01 -7.22649515e-01
8.33319053e-02 2.69329101e-01 3.89825523e-01 4.81734842e-01
-9.27509904e-01 8.96155894e-01 6.41708803e+00 5.63191235e-01
-1.17356765e+00 3.46231371e-01 -8.59372839e-02 6.52951524e-02
-7.78565228e-01 5.49599767e-01 -6.99945688e-01 2.00330466e-01
1.13372517e+00 -5.18746078e-01 5.71848452e-01 1.21227479e+00
-4.33396965e-01 -1.85797304e-01 -1.49646294e+00 6.78657830e-01
2.01394066e-01 -1.17938471e+00 9.46648195e-02 -1.75288931e-01
1.10923707e+00 6.24960423e-01 -1.90346867e-01 8.31814945e-01
7.45046258e-01 -7.79522538e-01 1.03250575e+00 1.46510422e-01
7.92129517e-01 -7.21787810e-01 6.38298631e-01 2.52773374e-01
-1.42226422e+00 -3.29381227e-01 -2.08624423e-01 6.58798814e-02
-3.49887967e-01 4.23684269e-01 -4.70541596e-01 8.68328273e-01
2.90158838e-01 1.11257362e+00 -1.24805272e+00 7.61901319e-01
-3.36965024e-01 6.05503023e-01 -8.13787244e-03 2.27593571e-01
2.08394259e-01 1.40040293e-01 5.32233298e-01 1.98382521e+00
2.98611909e-01 -4.59944725e-01 6.04856610e-01 8.99035990e-01
-2.67445207e-01 3.96732956e-01 -8.89588833e-01 -2.26495370e-01
1.92994744e-01 1.11981058e+00 -3.44832987e-01 -4.98874217e-01
-1.25937665e+00 5.12149394e-01 6.68543875e-01 1.47884756e-01
-5.12063265e-01 -6.73331499e-01 4.87588942e-01 -9.59040821e-02
4.47999626e-01 -4.52000856e-01 -3.18783879e-01 -1.71329272e+00
2.28859186e-01 -1.29467499e+00 5.01123846e-01 -3.65022391e-01
-9.44111228e-01 4.82508153e-01 5.99427968e-02 -9.26398933e-01
-4.27953422e-01 -5.24690509e-01 -7.03804672e-01 8.02807987e-01
-1.71432698e+00 -1.22606337e+00 8.02428052e-02 2.97880560e-01
5.51727831e-01 -3.24624062e-01 8.68923604e-01 2.22802028e-01
-3.66752476e-01 7.64859915e-01 3.40229303e-01 3.95381659e-01
6.87099159e-01 -1.53286362e+00 7.77573287e-01 1.19262314e+00
2.96880126e-01 1.25567222e+00 4.80168998e-01 -5.87954640e-01
-1.85354459e+00 -1.03647506e+00 1.07578957e+00 -7.39681542e-01
9.90634561e-01 -3.94412160e-01 -9.78339255e-01 9.83976007e-01
5.08567870e-01 -1.02123283e-01 7.01177716e-01 5.45446992e-01
-1.11872840e+00 1.20139502e-01 -5.74035347e-01 4.85038429e-01
9.12588775e-01 -9.42514062e-01 -8.97900820e-01 6.13176599e-02
6.72423422e-01 -3.54678124e-01 -1.04236126e+00 1.33732706e-01
3.83000910e-01 -9.41985905e-01 6.77733481e-01 -7.44827688e-01
8.88650298e-01 -2.29354858e-01 -5.69775403e-01 -1.25504887e+00
-2.25423157e-01 -4.53682899e-01 -7.24100322e-02 1.38202405e+00
5.59903979e-01 -5.00042379e-01 3.17028016e-01 9.55493599e-02
-3.91313761e-01 -4.32586819e-01 -5.37159085e-01 -7.43769646e-01
6.45212054e-01 -4.27637279e-01 6.39027119e-01 1.08099031e+00
5.52409768e-01 5.14556348e-01 1.13490544e-01 5.20517165e-03
4.69106078e-01 8.29113841e-01 1.09387684e+00 -1.15340447e+00
-5.21133840e-01 -6.02179289e-01 -1.67755201e-01 -9.47955251e-01
6.26334012e-01 -1.61364925e+00 -2.63707042e-02 -1.41047490e+00
6.51764095e-01 -4.30784315e-01 -1.05146565e-01 8.17519486e-01
-2.46300489e-01 -3.68960708e-01 2.96926349e-01 3.37312490e-01
-5.99848866e-01 2.63140589e-01 4.38260704e-01 -2.79314697e-01
1.07196376e-01 -3.45470816e-01 -9.33902085e-01 7.91693687e-01
6.80584729e-01 -8.37672591e-01 -3.66743058e-01 -1.03524208e+00
5.89882016e-01 3.31265628e-01 -6.99215336e-04 -9.18718398e-01
6.04520738e-02 -4.28175926e-02 -3.18061799e-01 -1.12480931e-01
-4.03500110e-01 -5.33989131e-01 -2.32802808e-01 4.36917126e-01
-4.76051569e-01 4.23864841e-01 5.42009294e-01 5.26361823e-01
-2.84085751e-01 -7.32408643e-01 6.15520656e-01 -3.33400697e-01
-8.14056098e-01 -1.58683047e-01 -5.16011178e-01 4.57399756e-01
4.37986702e-01 3.38848859e-01 -6.09707415e-01 8.12659189e-02
-8.37556124e-02 -2.21729763e-02 1.09374261e+00 6.51494741e-01
4.15507033e-02 -9.89756465e-01 -8.44042063e-01 1.84640259e-01
5.90998709e-01 -2.83097327e-01 -8.59119296e-02 7.12548077e-01
-5.60520709e-01 6.11385942e-01 -1.22692481e-01 -6.05810761e-01
-1.30762005e+00 7.54588723e-01 7.81300664e-02 -7.02536047e-01
-4.61094677e-01 4.06893343e-01 2.86136955e-01 -9.27280545e-01
-1.28313139e-01 -7.20544338e-01 5.37496060e-02 -7.94791803e-02
4.09124464e-01 6.10608011e-02 3.53704870e-01 -6.79642677e-01
-3.82046968e-01 6.26260757e-01 -4.20253634e-01 1.18153974e-01
1.52106011e+00 1.76377818e-01 -7.02248573e-01 5.46513319e-01
1.41268480e+00 7.38580763e-01 -9.25715029e-01 -5.27085781e-01
6.10865593e-01 -4.44342703e-01 -1.73576623e-01 -6.82902515e-01
-1.03342128e+00 1.00736427e+00 -6.32165326e-03 -5.99016994e-02
7.93765366e-01 9.65577289e-02 5.45878291e-01 7.69192934e-01
7.72237420e-01 -6.27969623e-01 5.45669720e-02 8.73816788e-01
7.28204131e-01 -1.04265809e+00 9.90651697e-02 -1.63223773e-01
-4.34170365e-01 1.26233482e+00 4.27372247e-01 -1.67797208e-01
3.56246233e-01 5.58365166e-01 -1.14830665e-01 -3.63757089e-02
-9.67614710e-01 2.00248230e-02 2.25747496e-01 4.15005237e-01
1.18041515e+00 -1.57145351e-01 -1.51712194e-01 6.50403738e-01
-7.83003494e-02 -1.17550448e-01 9.59000289e-01 1.53267086e+00
-4.03302759e-01 -1.56670392e+00 -2.45030850e-01 5.29399812e-01
-7.32118428e-01 -6.17383778e-01 -2.32424900e-01 7.78024435e-01
-6.85527027e-02 7.79379010e-01 -1.28878728e-01 -1.47986531e-01
3.00170183e-01 1.83850363e-01 6.16996348e-01 -1.30552626e+00
-1.05516183e+00 -3.47830892e-01 1.34100094e-01 -6.10512316e-01
-2.13968813e-01 -7.77989805e-01 -1.25970221e+00 -3.83361846e-01
-6.72778711e-02 2.52718151e-01 6.53404176e-01 6.49040282e-01
6.26496553e-01 4.33444709e-01 4.45889413e-01 -5.89255750e-01
-5.46502590e-01 -6.66307986e-01 -2.12903261e-01 2.96247214e-01
6.54026151e-01 -1.94199368e-01 -1.42535686e-01 4.11561787e-01] | [7.665132522583008, 7.92950439453125] |
073ca3fa-b863-4499-981d-2c2a7b6c9bfe | migration-reframed-a-multilingual-analysis-on | 2302.02813 | null | https://arxiv.org/abs/2302.02813v2 | https://arxiv.org/pdf/2302.02813v2.pdf | Migration Reframed? A multilingual analysis on the stance shift in Europe during the Ukrainian crisis | The war in Ukraine seems to have positively changed the attitude toward the critical societal topic of migration in Europe -- at least towards refugees from Ukraine. We investigate whether this impression is substantiated by how the topic is reflected in online news and social media, thus linking the representation of the issue on the Web to its perception in society. For this purpose, we combine and adapt leading-edge automatic text processing for a novel multilingual stance detection approach. Starting from 5.5M Twitter posts published by 565 European news outlets in one year, beginning September 2021, plus replies, we perform a multilingual analysis of migration-related media coverage and associated social media interaction for Europe and selected European countries. The results of our analysis show that there is actually a reframing of the discussion illustrated by the terminology change, e.g., from "migrant" to "refugee", often even accentuated with phrases such as "real refugees". However, concerning a stance shift in public perception, the picture is more diverse than expected. All analyzed cases show a noticeable temporal stance shift around the start of the war in Ukraine. Still, there are apparent national differences in the size and stability of this shift. | ['Erick Elejalde', 'Claudia Niederée', 'Sergej Wildemann'] | 2023-02-06 | null | null | null | null | ['stance-detection'] | ['natural-language-processing'] | [-3.38207006e-01 4.31880116e-01 -1.85523286e-01 8.66095647e-02
-3.33105713e-01 -9.57304180e-01 1.24940729e+00 8.94067943e-01
-9.45924640e-01 1.00352609e+00 1.19085538e+00 -5.40061116e-01
-9.39433128e-02 -9.15166318e-01 -5.00705779e-01 -6.07013822e-01
1.30876675e-01 3.27506393e-01 -2.21622596e-03 -1.05431473e+00
1.47088990e-01 1.48480549e-01 -1.21645153e+00 2.25746304e-01
8.06452453e-01 9.97890383e-02 -2.00885445e-01 1.08607717e-01
-4.44546729e-01 3.42847139e-01 -8.30781221e-01 -7.58635640e-01
-1.18119851e-01 -1.52153388e-01 -8.73544991e-01 -1.08832136e-01
1.53982371e-01 2.50506073e-01 -1.19289778e-01 1.13906550e+00
3.68614405e-01 -2.77823418e-01 5.09928226e-01 -5.28858364e-01
1.50086386e-02 1.08176339e+00 -4.43613857e-01 5.29924870e-01
8.06014061e-01 -1.91115275e-01 5.97577274e-01 -8.80910456e-01
1.59454930e+00 1.25783288e+00 6.52390003e-01 -1.21310085e-01
-1.05070984e+00 -3.89260232e-01 3.94564539e-01 -1.94944710e-01
-1.25280547e+00 -1.76013425e-01 5.00632226e-01 -1.07020736e+00
5.28731763e-01 5.79161763e-01 1.09152365e+00 1.27362692e+00
3.31655353e-01 -1.53364331e-01 1.46749008e+00 -3.39753926e-01
-1.67078927e-01 5.62221110e-01 1.43701166e-01 1.16338871e-01
7.32807875e-01 -5.13214827e-01 -4.61845934e-01 -3.56805652e-01
-3.68496507e-01 -2.79832274e-01 -3.82613063e-01 3.52155119e-01
-1.40374053e+00 7.14817762e-01 1.87125251e-01 1.01445174e+00
-4.86245751e-01 -7.98223913e-01 8.01561654e-01 7.08908558e-01
1.25918996e+00 4.82649431e-02 -4.08625454e-01 -3.65281224e-01
-7.18070149e-01 3.57782930e-01 9.89685476e-01 1.38331905e-01
5.89472711e-01 -4.11503017e-01 3.81611466e-01 6.82994545e-01
2.19786212e-01 6.22239828e-01 2.67248154e-01 -1.33211136e-01
7.27757275e-01 7.27869928e-01 1.42676547e-01 -1.57230198e+00
-7.54989028e-01 -7.52949774e-01 -4.95301366e-01 1.10278151e-03
6.39122665e-01 -6.73326612e-01 -1.20440736e-01 1.76840007e+00
5.63079357e-01 -7.45194077e-01 1.07733145e-01 8.21047723e-01
6.49806261e-01 8.03323388e-01 1.83314472e-01 -6.04288280e-01
1.54490733e+00 -1.27537623e-01 -8.71940374e-01 -1.16934739e-02
6.36235535e-01 -1.17568803e+00 5.59396207e-01 8.06764066e-02
-8.95199835e-01 -2.89513707e-01 -8.90334725e-01 1.95632741e-01
-9.00147796e-01 -2.59770155e-01 -3.81929167e-02 7.57956743e-01
-8.31801295e-01 4.48437065e-01 -3.36407512e-01 -1.07744336e+00
-1.99156567e-01 -1.45644769e-01 -6.33650780e-01 4.70226288e-01
-1.41925216e+00 1.01774621e+00 4.39199567e-01 -3.10818460e-02
4.64767516e-01 -4.23678368e-01 -7.36834228e-01 -6.27953768e-01
2.20764667e-01 -3.34097475e-01 5.79529166e-01 -1.20939589e+00
-1.08870924e+00 1.41058350e+00 1.42470405e-01 -4.80571002e-01
9.96902943e-01 -4.81096432e-02 -1.21359646e+00 -6.63025603e-02
3.40666205e-01 1.47891566e-01 3.67571205e-01 -1.14071429e+00
-8.27780128e-01 -4.48422074e-01 1.51586235e-01 1.27522245e-01
-4.12869513e-01 7.93314695e-01 6.47210777e-02 -9.19792533e-01
1.13582395e-01 -9.67212260e-01 -3.67427468e-02 -7.71243274e-01
-1.22020662e-01 -5.35052828e-02 4.97621804e-01 -1.06359303e+00
1.82659304e+00 -2.18355322e+00 -8.72889161e-02 2.37651736e-01
2.80676454e-01 -6.64920360e-02 6.26480460e-01 1.28441072e+00
1.27060339e-02 3.58743012e-01 -1.65619045e-01 9.66736749e-02
-2.86487546e-02 8.78992677e-02 -4.97273564e-01 9.58416581e-01
-3.15407991e-01 4.58869517e-01 -1.02367508e+00 -3.77450585e-01
-2.37120494e-01 3.60663593e-01 -3.83199573e-01 -7.44772315e-01
2.10384101e-01 6.57169163e-01 -4.15738635e-02 3.83873969e-01
7.06209481e-01 2.33747274e-01 5.63829184e-01 -5.03069833e-02
-1.26112878e+00 5.70365310e-01 -7.38087118e-01 1.23599029e+00
-4.83719520e-02 1.17718625e+00 3.44592184e-01 -4.26487744e-01
8.82899463e-01 3.24545711e-01 4.09333378e-01 -7.18083322e-01
3.41212153e-01 4.34279561e-01 3.26696396e-01 -2.60422856e-01
1.04064190e+00 -1.62206769e-01 -6.77233338e-01 3.47139508e-01
-3.31607759e-01 2.77745813e-01 4.13705289e-01 9.00723562e-02
4.24941927e-01 -1.25362068e-01 5.53716063e-01 -8.79496694e-01
6.04733706e-01 2.80274540e-01 5.36124468e-01 2.20077828e-01
1.70509238e-02 2.02031568e-01 6.08974040e-01 -5.00536740e-01
-6.54795587e-01 -7.14728892e-01 -4.17110920e-01 7.85656571e-01
3.13970773e-03 -7.54079401e-01 -7.83087194e-01 -3.81229222e-01
-1.30681261e-01 5.85873663e-01 -6.44282997e-01 3.23035300e-01
-8.39563727e-01 -1.12369382e+00 9.79752466e-02 -6.39185846e-01
5.86627185e-01 -7.63460040e-01 -6.11479342e-01 2.34269127e-01
-6.87887788e-01 -1.00329995e+00 1.32159293e-01 -2.31512517e-01
-5.32410264e-01 -6.24778032e-01 -8.22106123e-01 -3.98773074e-01
5.74912786e-01 -7.02935830e-02 9.04492974e-01 -3.89222354e-01
3.38536292e-01 1.66882351e-01 -4.12247092e-01 -5.56979001e-01
-8.92216980e-01 3.95810753e-01 1.77851811e-01 1.36419935e-02
5.47226109e-02 -2.81563997e-01 -3.47366333e-01 1.46450587e-02
-9.11200762e-01 -2.90469110e-01 9.99675319e-02 -1.62212402e-02
-3.10468048e-01 -8.65586028e-02 5.57043493e-01 -1.16953027e+00
7.83718646e-01 -1.04895544e+00 -1.07390739e-01 -2.90178627e-01
-4.89786863e-01 -5.83213091e-01 2.51530021e-01 -1.60386115e-01
-1.02127314e+00 -9.16918337e-01 -3.59233737e-01 8.54472816e-01
-2.23092616e-01 1.13479567e+00 1.53268889e-01 2.62945563e-01
6.82335496e-01 -1.65063575e-01 -4.88009118e-02 -4.73731786e-01
8.56577903e-02 8.20091784e-01 2.93161720e-01 -2.94252247e-01
9.49386120e-01 6.83070838e-01 -6.05772078e-01 -1.18087709e+00
-2.95610696e-01 -4.32755917e-01 -5.54021776e-01 -7.71796584e-01
8.45362127e-01 -9.46841300e-01 -5.39983213e-01 6.75400496e-01
-1.17993259e+00 -1.05552599e-01 -9.85070020e-02 4.98927325e-01
7.46868998e-02 3.13172966e-01 -5.75227976e-01 -6.68297529e-01
-1.22116737e-01 -5.64733863e-01 2.22722992e-01 -1.74775049e-01
-9.31032836e-01 -1.42765057e+00 6.66058362e-01 3.14950258e-01
3.88145238e-01 8.11957061e-01 9.00955141e-01 -7.00934172e-01
4.99464452e-01 -5.79304472e-02 5.10307103e-02 -2.82001823e-01
4.60257351e-01 5.60615025e-02 -5.51702201e-01 -3.95809799e-01
-4.40999074e-03 4.53338623e-01 5.85771024e-01 -1.65173218e-01
-3.43183726e-01 -7.64965594e-01 -3.41186166e-01 -3.11905563e-01
1.42224836e+00 7.59542584e-02 3.62111628e-01 9.85384643e-01
-3.55941913e-04 1.09403861e+00 6.58676863e-01 5.25287449e-01
4.36022967e-01 9.07036126e-01 1.78176761e-01 -1.35843366e-01
6.46647885e-02 -1.49346948e-01 1.03059804e+00 1.25203681e+00
-2.35248998e-01 -1.10302299e-01 -1.31602430e+00 6.44066930e-01
-1.54907250e+00 -8.84342074e-01 -6.30405903e-01 2.25842714e+00
6.74575329e-01 5.33998132e-01 5.14393508e-01 6.32394254e-02
7.27938116e-01 6.07946634e-01 6.00643337e-01 -4.50061619e-01
-4.30159479e-01 -2.85909861e-01 7.56266773e-01 9.49655652e-01
-7.40750372e-01 6.95024252e-01 5.84080172e+00 3.44244331e-01
-1.34576988e+00 3.11288327e-01 3.80740881e-01 1.26098037e-01
-5.13668954e-01 -3.62725765e-03 -6.09589398e-01 6.11283302e-01
9.59601521e-01 -5.64433038e-01 -2.18962103e-01 8.77121612e-02
6.40036047e-01 -4.21624005e-01 -1.31736591e-01 4.99893695e-01
1.37396097e-01 -1.29923546e+00 -2.29859967e-02 4.58964527e-01
7.09395170e-01 1.75188377e-01 -8.42683297e-03 1.22910395e-01
-2.03430012e-01 -1.45394102e-01 1.32578576e+00 4.30218279e-01
5.92651367e-01 -6.96514368e-01 8.88708472e-01 3.66803497e-01
-8.16561520e-01 2.46202886e-01 2.17662789e-02 -5.04458308e-01
4.49103892e-01 8.59321833e-01 -6.85567141e-01 8.47111762e-01
4.03572679e-01 4.58999962e-01 -5.17945170e-01 4.86969739e-01
2.23039929e-02 7.16047823e-01 -1.44475490e-01 6.47640899e-02
4.77657735e-01 -6.61298335e-01 1.30566490e+00 1.65075517e+00
5.37046671e-01 -2.51803339e-01 -1.43253222e-01 1.72320768e-01
6.29463345e-02 6.57023311e-01 -8.29489768e-01 -1.05651103e-01
8.89526457e-02 1.16356409e+00 -1.13757539e+00 -2.21731186e-01
-5.78276277e-01 6.06080770e-01 9.55967326e-03 3.04246694e-01
-7.36388862e-01 -1.38969213e-01 6.32297695e-01 8.49067450e-01
-2.00317204e-01 -3.80786419e-01 1.23165131e-01 -9.23008978e-01
4.05302569e-02 -8.14488888e-01 2.92924672e-01 -1.65534884e-01
-7.28052199e-01 7.62772322e-01 1.42751843e-01 -1.11862135e+00
-1.68578759e-01 -2.95832247e-01 -5.00138700e-01 5.82896650e-01
-1.00525987e+00 -7.59392619e-01 3.35843831e-01 1.41608700e-01
2.28216991e-01 1.49454609e-01 6.42973065e-01 5.18788934e-01
-3.93922955e-01 4.01021577e-02 -4.08117361e-02 3.44602354e-02
9.32342827e-01 -9.01975393e-01 4.35808033e-01 6.54826224e-01
-2.12680221e-01 6.22324646e-01 1.33077776e+00 -1.10599792e+00
-5.41296601e-01 -6.40535593e-01 1.95434427e+00 -2.99436182e-01
1.39907014e+00 -3.73014539e-01 -4.92454797e-01 5.40938020e-01
1.00236690e+00 -9.40692961e-01 8.18748832e-01 4.54783261e-01
-1.82129126e-02 2.17664391e-01 -1.04451418e+00 1.05153203e+00
1.00104570e+00 -3.10786337e-01 -8.19722116e-01 4.06866938e-01
5.77495873e-01 -1.73184082e-01 -1.04621911e+00 2.38905922e-01
5.03362477e-01 -8.66401076e-01 5.68743110e-01 -4.31401879e-01
1.77582234e-01 -4.35938127e-03 8.24351981e-02 -1.50045824e+00
-2.97203630e-01 -9.26708400e-01 8.23063850e-01 1.50778615e+00
7.61966646e-01 -1.26551425e+00 2.60855973e-01 9.77231711e-02
-9.74982232e-02 -1.83132976e-01 -1.13178360e+00 -3.41403425e-01
3.09667736e-01 -3.35169941e-01 1.49268314e-01 1.39823902e+00
6.38961911e-01 1.83345035e-01 -5.94825968e-02 -5.69482483e-02
1.91581789e-02 -6.44836649e-02 7.76200354e-01 -1.43557763e+00
3.31028253e-01 -6.23152554e-01 -3.88357401e-01 -2.92136639e-01
-1.09019004e-01 -9.72406864e-01 -8.06697190e-01 -1.49505925e+00
-1.55535594e-01 5.84810562e-02 1.97621211e-01 -4.49884444e-01
2.77576149e-01 1.01602398e-01 3.53639454e-01 3.35059643e-01
-1.67767480e-01 6.22833781e-02 9.54441547e-01 1.36065230e-01
-3.68361354e-01 -7.25893378e-02 -7.93464422e-01 1.05650806e+00
8.31962705e-01 -5.25294125e-01 2.49730706e-01 -9.77163855e-03
1.28803384e+00 -1.12457626e-01 1.76374286e-01 -1.07400239e+00
1.21541798e-01 6.66772127e-02 -1.99146062e-01 -6.51066363e-01
6.10328317e-02 -7.90241420e-01 5.55514932e-01 9.56782997e-01
1.71072390e-02 4.81902063e-01 3.76689881e-01 1.64773002e-01
-5.03075302e-01 -7.30840042e-02 2.70894140e-01 3.88126783e-02
-1.76511854e-01 -2.51141369e-01 -1.26576972e+00 7.24092796e-02
7.91100502e-01 -3.06534976e-01 -5.33921897e-01 -3.65485102e-01
-1.21201646e+00 -7.59261623e-02 7.45188713e-01 4.76509869e-01
-8.55731666e-02 -1.16426980e+00 -1.17542470e+00 -3.24218810e-01
2.06483349e-01 -8.81530643e-01 2.57417172e-01 1.32043123e+00
-4.64402229e-01 4.07018751e-01 -3.49065393e-01 9.75009054e-02
-1.28866935e+00 3.30331177e-01 -3.88127603e-02 -3.97155881e-01
-5.12958825e-01 1.01929747e-01 -9.05777141e-02 -5.88082075e-01
-3.07286084e-01 -3.97218585e-01 -6.25633895e-01 1.42895234e+00
2.95209318e-01 6.27540886e-01 -1.02789607e-02 -1.29463172e+00
-3.78890365e-01 4.92075533e-01 -1.15345661e-02 -6.34759068e-01
1.13066351e+00 -6.38162434e-01 -5.64180613e-01 1.13954008e+00
1.03770053e+00 1.16806614e+00 -3.29157501e-01 -5.52567542e-02
3.37974966e-01 -2.82390773e-01 -5.18051684e-01 -7.52889216e-01
-4.89645064e-01 1.38573393e-01 2.24668041e-01 7.38069415e-01
5.11021793e-01 8.67090076e-02 5.16377449e-01 -8.05134550e-02
4.77319092e-01 -1.30238128e+00 -8.09465349e-01 6.87875807e-01
1.18546581e+00 -6.69304132e-01 1.86282676e-02 -6.19497955e-01
-3.99678171e-01 1.05149090e+00 -1.12176649e-01 2.10289896e-01
9.24991786e-01 6.25911504e-02 3.24190080e-01 -4.03481543e-01
-3.58630598e-01 -1.67677462e-01 -1.17872609e-02 1.12211615e-01
6.74496472e-01 3.22681069e-01 -1.45296466e+00 2.85105675e-01
-6.22525871e-01 -4.41107988e-01 8.10737491e-01 5.79296529e-01
-3.61448377e-01 -1.13275242e+00 -5.97983420e-01 -1.18920676e-01
-8.50681961e-01 5.28261177e-02 -8.29960227e-01 1.30171955e+00
3.58143300e-01 7.58112729e-01 2.82958090e-01 -3.37696314e-01
2.44188070e-01 9.78343561e-02 3.79637852e-02 -3.35986972e-01
-1.24902272e+00 1.47968277e-01 9.41259503e-01 6.16489351e-02
-7.30558574e-01 -1.20246565e+00 -1.06873119e+00 -6.70185030e-01
1.69688255e-01 3.98182988e-01 7.54083276e-01 1.03863668e+00
5.66695444e-02 3.11965942e-01 4.25477415e-01 -4.15189713e-01
2.21591190e-01 -9.32599068e-01 -3.38686556e-01 2.16220871e-01
3.24046969e-01 -2.87813842e-01 -4.12116289e-01 -2.94742703e-01] | [8.643515586853027, 9.884262084960938] |
90afc015-416a-4ae7-a3ac-eddea7f9e75a | aser-towards-large-scale-commonsense | 2104.02137 | null | https://arxiv.org/abs/2104.02137v2 | https://arxiv.org/pdf/2104.02137v2.pdf | ASER: Towards Large-scale Commonsense Knowledge Acquisition via Higher-order Selectional Preference over Eventualities | Commonsense knowledge acquisition and reasoning have long been a core artificial intelligence problem. However, in the past, there has been a lack of scalable methods to collect commonsense knowledge. In this paper, we propose to develop principles for collecting commonsense knowledge based on selectional preference. We generalize the definition of selectional preference from one-hop linguistic syntactic relations to higher-order relations over linguistic graphs. Unlike previous commonsense knowledge definition (e.g., ConceptNet), selectional preference (SP) knowledge only relies on statistical distribution over linguistic graphs, which can be efficiently and accurately acquired from the unlabeled corpus with modern tools. Following this principle, we develop a large-scale eventuality (a linguistic term covering activity, state, and event)-based knowledge graph ASER, where each eventuality is represented as a dependency graph, and the relation between them is a discourse relation defined in shallow discourse parsing. The higher-order selectional preference over collected linguistic graphs reflects various kinds of commonsense knowledge. Moreover, motivated by the observation that humans understand events by abstracting the observed events to a higher level and can thus transfer their knowledge to new events, we propose a conceptualization module to significantly boost the coverage of ASER. In total, ASER contains 648 million edges between 438 million eventualities. After conceptualization with Probase, a selectional preference based concept-instance relational knowledge base, our concept graph contains 15 million conceptualized eventualities and 224 million edges between them. Detailed analysis is provided to demonstrate its quality. All the collected data, APIs, and tools are available at https://github.com/HKUST-KnowComp/ASER. | ['Yangqiu Song', 'Tianqing Fang', 'Jiefu Ou', 'Haowen Ke', 'Haojie Pan', 'Xin Liu', 'Hongming Zhang'] | 2021-04-05 | null | null | null | null | ['discourse-parsing'] | ['natural-language-processing'] | [ 8.82518142e-02 5.44840217e-01 -4.54120308e-01 -4.98423696e-01
-4.52773064e-01 -7.75441945e-01 7.19684064e-01 6.12324417e-01
-2.77104646e-01 9.91917133e-01 5.34455895e-01 -3.28903019e-01
-3.04319918e-01 -1.31175411e+00 -7.86738873e-01 -1.43764257e-01
9.91000049e-03 5.72758377e-01 4.55534905e-01 -3.93815249e-01
2.50841558e-01 1.63263381e-01 -1.39853942e+00 4.28481162e-01
1.17057514e+00 7.03816354e-01 1.13498129e-01 9.71907154e-02
-4.01062667e-01 1.38082492e+00 -6.63867533e-01 -6.42554998e-01
-4.19534683e-01 -5.50882697e-01 -1.37154138e+00 -1.69309080e-01
-1.90432742e-01 6.26406074e-02 -2.36029148e-01 1.31288314e+00
-8.10512714e-03 4.16846484e-01 5.92506886e-01 -1.32299376e+00
-1.12875521e+00 1.48881769e+00 -8.49158838e-02 2.94078231e-01
7.26552188e-01 -1.59098506e-01 1.47457409e+00 -4.64554936e-01
1.03026986e+00 1.42388117e+00 2.94156313e-01 6.47433162e-01
-9.08277988e-01 -5.30240178e-01 1.74423024e-01 7.15881526e-01
-1.19736576e+00 -9.60046202e-02 9.30880547e-01 -2.63857633e-01
1.15744066e+00 3.23275387e-01 9.32629228e-01 1.14419663e+00
-2.00110406e-01 7.95393050e-01 1.01786613e+00 -6.49573684e-01
5.24621189e-01 4.96345991e-03 8.13850343e-01 7.28424728e-01
6.57014787e-01 -2.98571140e-01 -6.86932921e-01 -1.89373970e-01
5.26644707e-01 -8.10212269e-02 -2.71063060e-01 4.62476648e-02
-1.07387698e+00 7.59587348e-01 4.50700283e-01 6.61596298e-01
-2.86730975e-01 2.27059349e-01 5.42824209e-01 7.43859708e-02
2.50765920e-01 5.25422037e-01 -5.92476368e-01 -2.23625049e-01
-4.08680797e-01 7.26884678e-02 9.08660173e-01 1.09761858e+00
8.61735940e-01 -2.28420451e-01 4.40278649e-02 6.76657617e-01
2.92594105e-01 3.85498315e-01 5.76300740e-01 -1.13409638e+00
3.90942574e-01 1.06618607e+00 -1.46776885e-01 -1.11014235e+00
-3.44463557e-01 5.20186275e-02 -4.98117000e-01 -5.79079986e-01
9.09346938e-02 -2.76754778e-02 -5.76478720e-01 1.99312377e+00
3.79383475e-01 7.27059543e-02 3.13638031e-01 6.88550591e-01
1.03979409e+00 5.81536531e-01 1.80421203e-01 -3.88572484e-01
1.67373538e+00 -2.77212679e-01 -8.44652057e-01 -2.82869726e-01
8.10415328e-01 3.72214243e-02 1.37349880e+00 1.75836131e-01
-7.34576702e-01 8.88762847e-02 -8.98913801e-01 -4.07219499e-01
-5.62435567e-01 -3.89491916e-01 1.10523903e+00 3.62396508e-01
-6.30157411e-01 3.36722583e-01 -7.19231486e-01 -5.09617448e-01
4.64386076e-01 -2.34244853e-01 -2.70111054e-01 -9.86683965e-02
-1.86307955e+00 1.02134681e+00 1.27355886e+00 -1.16633832e-01
-5.46426535e-01 -6.20656312e-01 -1.17893040e+00 1.00071087e-01
1.04559124e+00 -6.29861355e-01 8.24552774e-01 -5.17604530e-01
-1.32441020e+00 1.08638370e+00 -1.35060981e-01 -6.21664464e-01
-8.44347849e-02 -1.75865546e-01 -6.74537599e-01 1.93670481e-01
3.10075551e-01 2.01846659e-01 7.77171403e-02 -1.32855511e+00
-6.30546629e-01 -4.12381142e-01 7.77742803e-01 1.60532802e-01
-3.16831023e-01 1.16574839e-01 -1.75439447e-01 -3.68128091e-01
1.28166199e-01 -4.92265403e-01 1.86809599e-01 -5.26381850e-01
-7.39861012e-01 -5.99555671e-01 3.42539966e-01 -3.87569904e-01
1.58823824e+00 -2.04257512e+00 2.52777219e-01 1.11567356e-01
4.69862193e-01 -2.87051409e-01 3.90350431e-01 3.23004246e-01
-8.16852078e-02 4.56668615e-01 -5.56556642e-01 2.41639972e-01
4.52230185e-01 6.58338308e-01 -4.75374103e-01 4.26578484e-02
6.99480549e-02 9.55946088e-01 -1.46224916e+00 -8.31438482e-01
1.62303373e-01 6.39928654e-02 -4.67290163e-01 -5.86390197e-02
-5.36256015e-01 -1.46697894e-01 -7.32683241e-01 7.03497469e-01
1.34010077e-01 -4.44435507e-01 5.50809920e-01 -3.33696634e-01
1.54481664e-01 6.95167840e-01 -8.79967034e-01 1.73257673e+00
-3.75123620e-01 4.81430799e-01 -4.92126554e-01 -1.09408379e+00
7.68932521e-01 2.12915435e-01 1.57616377e-01 -2.52197713e-01
2.65098542e-01 1.14382535e-01 6.43371120e-02 -6.44001365e-01
4.82967079e-01 -5.88950098e-01 -6.83172405e-01 4.16114986e-01
2.35045627e-01 -4.60664630e-01 7.75700986e-01 6.42507970e-01
1.19413972e+00 2.04933621e-02 8.24167728e-01 -3.48163098e-01
5.38737714e-01 5.07525921e-01 9.22951519e-01 3.21297318e-01
-6.87070116e-02 -5.84215857e-02 7.93244123e-01 -2.67980158e-01
-5.30954182e-01 -1.17361498e+00 -2.39422932e-01 1.00478888e+00
3.31071526e-01 -6.72236323e-01 -5.51286936e-01 -6.70339823e-01
-1.87645450e-01 1.64308822e+00 -6.63220704e-01 -2.36373141e-01
-6.43594325e-01 -6.47337556e-01 7.68712103e-01 4.85506028e-01
6.75586104e-01 -1.46358311e+00 -7.76966095e-01 1.19373500e-01
-6.26969457e-01 -1.23788226e+00 -7.33727887e-02 2.42613226e-01
-5.29257476e-01 -1.34658027e+00 3.66602033e-01 -5.56134701e-01
4.46154624e-01 -2.81638443e-01 1.42552102e+00 -2.75942590e-03
-2.03788653e-01 3.88345361e-01 -8.16069603e-01 -4.77608383e-01
-4.30431157e-01 -3.18444073e-01 4.55333367e-02 -4.86628741e-01
8.94199491e-01 -8.08918715e-01 -2.03945418e-03 -3.04724097e-01
-1.14482594e+00 -1.26911700e-01 1.10635646e-01 6.00264251e-01
6.16570711e-01 5.10381579e-01 5.04997075e-01 -1.15949571e+00
8.78003120e-01 -7.26886213e-01 -3.29563111e-01 4.93496597e-01
-2.75823802e-01 2.15923905e-01 5.57435393e-01 -2.32404903e-01
-1.28703344e+00 -5.88965058e-01 2.95643628e-01 4.37361076e-02
-2.74001807e-01 1.18918061e+00 -4.35334980e-01 8.73861730e-01
9.12173986e-01 1.60019428e-01 -5.81005991e-01 1.76917519e-02
8.32321227e-01 4.31381226e-01 4.83311445e-01 -1.35828197e+00
4.11676675e-01 4.44056720e-01 -1.22686379e-01 -7.44128823e-01
-1.42782688e+00 -2.95695662e-01 -5.95968246e-01 1.43014546e-02
9.84268129e-01 -5.60654938e-01 -7.60349095e-01 -1.39961258e-01
-1.30173516e+00 -2.26477936e-01 -7.97390401e-01 4.52186286e-01
-6.57834888e-01 3.49232912e-01 -6.92610860e-01 -7.84606993e-01
-2.63598919e-01 -4.67332453e-01 6.06967568e-01 1.04077727e-01
-5.40941715e-01 -1.20443642e+00 1.66984333e-03 2.64265209e-01
-1.76648125e-01 4.98089164e-01 1.31589580e+00 -9.08206761e-01
-2.73221284e-01 1.30785611e-02 -1.48569316e-01 2.74436921e-01
2.64400601e-01 2.52319183e-02 -4.80753273e-01 4.30715740e-01
1.88906237e-01 -5.59371769e-01 6.17714882e-01 2.23461702e-01
1.02656686e+00 -3.69006962e-01 -3.33675772e-01 -3.63668129e-02
1.36382401e+00 2.12659419e-01 7.64496624e-01 3.56550962e-01
5.95862329e-01 6.59748614e-01 5.22270977e-01 4.35263664e-01
7.99958944e-01 1.56225443e-01 1.34251103e-01 5.65113783e-01
-1.30174592e-01 -2.47383609e-01 4.43393737e-01 9.59799707e-01
-5.89544833e-01 -1.83047533e-01 -1.11419749e+00 7.84754097e-01
-1.85264289e+00 -1.45114112e+00 -7.16256574e-02 1.75532436e+00
1.44522452e+00 2.26767689e-01 -1.62963852e-01 1.22282520e-01
8.33799064e-01 5.63602559e-02 -5.21746516e-01 -2.21218154e-01
-4.10445452e-01 1.14213653e-01 -4.15908396e-02 7.73244917e-01
-5.73157489e-01 1.35960925e+00 5.05069494e+00 9.24224794e-01
-4.07181412e-01 1.87970519e-01 7.92002454e-02 2.14250505e-01
-7.16375709e-01 3.12359959e-01 -5.47835410e-01 2.32831761e-01
6.71805739e-01 -7.28747725e-01 4.87326384e-01 7.40079820e-01
-2.23471820e-01 -1.39022633e-01 -1.23498571e+00 8.23819995e-01
6.37678206e-02 -1.39241695e+00 3.60155910e-01 -1.37474179e-01
4.36498463e-01 -2.14924946e-01 -5.35800576e-01 6.03195727e-01
9.01482105e-01 -7.89148510e-01 7.96289444e-01 4.00553524e-01
7.32634723e-01 -6.66432798e-01 6.43117249e-01 3.77596319e-01
-1.35787380e+00 3.74673456e-02 -3.69524211e-01 -1.65287048e-01
3.43027204e-01 1.08987510e+00 -4.42745864e-01 8.95484626e-01
5.59400797e-01 8.20392251e-01 -1.42241701e-01 2.27640301e-01
-9.74611402e-01 6.86801553e-01 -3.42917174e-01 -3.00266594e-01
-1.21201478e-01 -2.13271469e-01 5.21458209e-01 1.23046982e+00
1.13339879e-01 8.13545644e-01 9.20999348e-02 1.34530377e+00
-3.71854335e-01 -1.23969376e-01 -4.97198105e-01 -2.23881543e-01
9.67729807e-01 9.68118250e-01 -7.08978236e-01 -7.65221655e-01
-2.47937456e-01 6.74644291e-01 6.17449999e-01 2.52535194e-01
-9.81946826e-01 -5.02726257e-01 3.99818450e-01 -1.35821238e-01
-2.03813493e-01 -2.32196432e-02 -1.14871867e-01 -1.50111485e+00
1.47700712e-01 -5.61163425e-01 7.10013866e-01 -8.59783173e-01
-1.62991190e+00 6.12351954e-01 5.57315111e-01 -5.72251320e-01
-2.31892124e-01 -6.89888418e-01 -4.47667032e-01 4.87290055e-01
-1.11166680e+00 -1.02554393e+00 -1.78629801e-01 7.12544203e-01
3.72322708e-01 2.48088509e-01 1.09711921e+00 -1.23580076e-01
-4.13593680e-01 8.72180536e-02 -6.76299512e-01 3.02255183e-01
3.67613286e-01 -1.39818645e+00 -7.73508251e-02 8.18020821e-01
1.72340676e-01 1.06995678e+00 7.86321163e-01 -9.95606184e-01
-1.09201336e+00 -8.82970154e-01 1.09718800e+00 -7.67168999e-01
1.33698821e+00 -1.38421595e-01 -1.08706272e+00 1.16875720e+00
1.51105821e-01 3.68464999e-02 8.79131138e-01 6.07416868e-01
-7.41613269e-01 2.09111556e-01 -1.13668394e+00 6.10172927e-01
1.60000527e+00 -7.35044539e-01 -1.60923076e+00 3.26206416e-01
1.10072017e+00 -2.18935296e-01 -9.49090779e-01 2.79051811e-01
-4.51708883e-02 -6.25415921e-01 6.00421846e-01 -7.54222035e-01
6.91614270e-01 -4.68257964e-01 -6.09915197e-01 -1.20893264e+00
-1.85109839e-01 -3.80507261e-01 -5.28206110e-01 1.22051919e+00
6.85414195e-01 -6.96930707e-01 3.84318978e-01 7.94579864e-01
-2.54374802e-01 -3.75476271e-01 -9.40098286e-01 -9.17172432e-01
1.33520976e-01 -7.30619490e-01 6.75121844e-01 1.46404445e+00
9.75337684e-01 7.35303760e-01 1.33610845e-01 1.02336407e-01
6.95583940e-01 4.80222434e-01 9.86041278e-02 -1.41273022e+00
-2.28558496e-01 -3.55145991e-01 -3.48585039e-01 -6.12825096e-01
5.27238011e-01 -1.19273293e+00 4.31411937e-02 -1.89909554e+00
5.91129124e-01 -3.52401942e-01 -2.83120632e-01 7.82707453e-01
-2.13768825e-01 -4.42346901e-01 -3.97440828e-02 6.98182592e-03
-8.43516052e-01 5.88011682e-01 1.23139870e+00 -9.51738209e-02
-2.59529471e-01 -8.19199026e-01 -8.41441035e-01 1.13123250e+00
9.67954397e-01 -3.82627428e-01 -5.42068243e-01 -3.60091060e-01
6.90393925e-01 -1.62323952e-01 4.40300643e-01 -5.89213967e-01
2.99158007e-01 -5.70395768e-01 -3.47108006e-01 -3.16050380e-01
2.32405096e-01 -6.78139806e-01 1.05415933e-01 2.19847828e-01
-3.85053307e-01 -3.04903269e-01 1.05553031e-01 4.25135732e-01
-4.20066357e-01 -2.78975874e-01 2.57954150e-01 -3.24536771e-01
-1.16575217e+00 -1.81675419e-01 1.71844065e-02 6.28239393e-01
9.56905603e-01 7.67218769e-02 -7.86727190e-01 1.81444734e-02
-9.49133337e-01 -1.04187243e-02 2.96512097e-01 1.63545236e-01
7.41682887e-01 -1.29290020e+00 -7.64136255e-01 -6.02064967e-01
3.95038456e-01 2.12204024e-01 1.49724141e-01 4.43354458e-01
-2.81250566e-01 2.55601436e-01 -7.26883858e-02 -1.70969460e-02
-7.89475262e-01 7.64606833e-01 6.23501884e-03 -1.89627528e-01
-6.23723388e-01 8.16125631e-01 -3.09471171e-02 -3.97459686e-01
-2.51202673e-01 -6.67190373e-01 -3.27278107e-01 3.51440944e-02
4.79591638e-01 3.29011828e-01 -3.86574447e-01 -5.91885746e-01
-7.34617651e-01 2.95475811e-01 9.61295590e-02 -1.09542459e-01
1.40788352e+00 -9.52258408e-02 -8.30522120e-01 8.50033760e-01
5.73752344e-01 2.13738874e-01 -3.69328737e-01 -2.85620749e-01
2.27071732e-01 -8.37197676e-02 -1.32935554e-01 -8.53550792e-01
-5.68685532e-01 5.17031312e-01 -4.76374358e-01 5.84088206e-01
9.08577025e-01 6.96557760e-01 4.66340750e-01 6.20712876e-01
8.88993442e-01 -1.15590310e+00 -1.36320814e-01 9.07919288e-01
1.07545733e+00 -9.38152909e-01 2.14280188e-02 -1.00155067e+00
-7.82494426e-01 9.22712922e-01 5.77230990e-01 -4.87101823e-02
5.52501321e-01 1.09697312e-01 -4.29909140e-01 -7.17609286e-01
-6.46582782e-01 -3.90833348e-01 9.63190123e-02 6.27761424e-01
5.41386604e-01 5.47120392e-01 -4.31676865e-01 1.16307199e+00
-5.79805195e-01 -3.02344561e-02 6.31562710e-01 8.73355150e-01
-5.73503554e-01 -8.78761768e-01 -1.50746077e-01 3.01649392e-01
-1.73625812e-01 -4.87417161e-01 -5.27497709e-01 8.87247443e-01
3.19154918e-01 1.11858654e+00 -2.05806971e-01 -2.30485037e-01
4.14381325e-01 1.22209005e-01 5.82295477e-01 -1.06369519e+00
-1.83536753e-01 -6.18309617e-01 6.05942905e-01 -5.00450373e-01
-7.88512766e-01 -5.93099117e-01 -2.18390203e+00 -4.91773725e-01
-3.53766471e-01 3.59402895e-01 5.36754578e-02 1.24177313e+00
1.55367896e-01 7.40149140e-01 -1.68139905e-01 2.59572584e-02
-1.37152910e-01 -8.89240801e-01 -7.59637833e-01 6.28510237e-01
-2.62183160e-01 -8.20817351e-01 -4.77655590e-01 4.13834602e-01] | [9.901439666748047, 8.175593376159668] |
8d05cfc9-10a9-440d-b18c-c11a12e16703 | graph-representation-learning-via-graphical | 2002.01169 | null | https://arxiv.org/abs/2002.01169v1 | https://arxiv.org/pdf/2002.01169v1.pdf | Graph Representation Learning via Graphical Mutual Information Maximization | The richness in the content of various information networks such as social networks and communication networks provides the unprecedented potential for learning high-quality expressive representations without external supervision. This paper investigates how to preserve and extract the abundant information from graph-structured data into embedding space in an unsupervised manner. To this end, we propose a novel concept, Graphical Mutual Information (GMI), to measure the correlation between input graphs and high-level hidden representations. GMI generalizes the idea of conventional mutual information computations from vector space to the graph domain where measuring mutual information from two aspects of node features and topological structure is indispensable. GMI exhibits several benefits: First, it is invariant to the isomorphic transformation of input graphs---an inevitable constraint in many existing graph representation learning algorithms; Besides, it can be efficiently estimated and maximized by current mutual information estimation methods such as MINE; Finally, our theoretical analysis confirms its correctness and rationality. With the aid of GMI, we develop an unsupervised learning model trained by maximizing GMI between the input and output of a graph neural encoder. Considerable experiments on transductive as well as inductive node classification and link prediction demonstrate that our method outperforms state-of-the-art unsupervised counterparts, and even sometimes exceeds the performance of supervised ones. | ['Yu Rong', 'Qinghua Zheng', 'Zhen Peng', 'Tingyang Xu', 'Junzhou Huang', 'Wenbing Huang', 'Minnan Luo'] | 2020-02-04 | null | null | null | null | ['mutual-information-estimation'] | ['methodology'] | [ 2.99877495e-01 6.36320114e-01 -4.99862611e-01 -3.22726607e-01
-1.03634093e-02 -3.91788334e-01 6.46804035e-01 3.50843698e-01
-2.66910382e-02 6.66523695e-01 2.05488041e-01 -3.85560274e-01
-6.09970391e-01 -1.24170983e+00 -5.39985240e-01 -6.00087523e-01
-5.42567015e-01 3.12303215e-01 -1.06773362e-01 -3.19662035e-01
-8.34519342e-02 3.78537953e-01 -1.16767085e+00 -2.54713029e-01
9.03273821e-01 8.61649692e-01 -1.99398939e-02 4.95333552e-01
-3.53927076e-01 1.16724980e+00 -8.33980516e-02 -4.71780300e-01
7.19574913e-02 -5.74798107e-01 -9.23280239e-01 3.25811245e-02
-1.80304095e-01 9.65335071e-02 -8.86090040e-01 1.16830051e+00
7.33734816e-02 -5.06265759e-02 9.40980256e-01 -1.51891339e+00
-9.25561070e-01 9.56159174e-01 -3.02238435e-01 3.85013781e-02
3.83884609e-01 -3.01708817e-01 1.62770391e+00 -5.16595602e-01
7.10961401e-01 9.21603620e-01 5.88107407e-01 1.97885633e-01
-1.46200204e+00 -4.17399883e-01 -1.01182796e-01 -2.04262454e-02
-1.29096949e+00 -3.61747444e-02 1.36906052e+00 -5.54724395e-01
6.99547827e-01 1.76979944e-01 6.43020093e-01 6.74976707e-01
1.27078548e-01 6.99995875e-01 7.86515772e-01 -5.24144590e-01
-1.00500546e-01 2.78022081e-01 1.63823217e-01 1.30847442e+00
3.84401113e-01 1.38357639e-01 -3.91237170e-01 -8.30412582e-02
8.63746762e-01 7.95472711e-02 -2.83140451e-01 -8.25392842e-01
-1.21768951e+00 9.25892949e-01 9.87668753e-01 8.40340972e-01
-6.42294139e-02 2.15723827e-01 2.28180647e-01 8.18219602e-01
4.38054144e-01 4.24578607e-01 2.60545686e-02 3.54149014e-01
-4.23422635e-01 -5.48570216e-01 1.06732225e+00 9.93798435e-01
1.15879464e+00 1.46659613e-01 3.55733633e-01 5.46041191e-01
4.99297500e-01 2.40709201e-01 4.23802316e-01 -3.65786999e-01
4.40433174e-01 1.14549887e+00 -6.14453852e-01 -1.71325183e+00
-4.66307521e-01 -6.28385305e-01 -1.41826165e+00 -2.49441534e-01
6.56531379e-02 6.05147593e-02 -3.47091496e-01 1.83645475e+00
4.86435853e-02 -1.67016819e-01 -4.31356244e-02 5.16558826e-01
9.31534767e-01 5.13970137e-01 -4.52095494e-02 -3.77421618e-01
7.26620376e-01 -5.21857083e-01 -7.42139280e-01 -5.31242415e-02
1.06030071e+00 -1.90282688e-01 6.25981867e-01 -2.35627398e-01
-8.13588262e-01 -3.27426642e-01 -1.17753768e+00 1.50402352e-01
-5.92231274e-01 -7.00341314e-02 1.10663855e+00 3.79275858e-01
-1.24381495e+00 8.59548807e-01 -4.84375089e-01 -4.05327141e-01
1.86688676e-01 4.30384368e-01 -9.35323238e-01 1.07211217e-01
-1.30700994e+00 6.72957718e-01 6.59987330e-01 -8.23679715e-02
-1.71262726e-01 -3.33584964e-01 -1.17225277e+00 1.79031953e-01
3.05717200e-01 -6.97769403e-01 4.08483297e-01 -1.01560652e+00
-1.36013007e+00 7.97029734e-01 1.14184298e-01 -3.14525783e-01
3.56242508e-02 3.72160971e-01 -4.69540626e-01 1.67629004e-01
-5.37574589e-02 3.56702536e-01 7.36012697e-01 -1.16650724e+00
1.20417021e-01 -4.01137471e-01 1.08888708e-01 3.81905809e-02
-8.73807430e-01 -4.74344373e-01 -6.44987524e-02 -5.73073804e-01
5.94666302e-01 -7.10378766e-01 -2.60308236e-01 1.33468479e-01
-6.40250564e-01 -2.74627924e-01 6.47043407e-01 -3.02562296e-01
1.40165424e+00 -2.00003171e+00 6.30456150e-01 7.07593501e-01
8.52619410e-01 3.72573777e-05 -1.52745917e-01 8.25706780e-01
-4.34695929e-01 1.61710545e-01 -4.00288492e-01 1.66211799e-01
-8.54518414e-02 3.64756525e-01 -2.17188522e-02 4.64930266e-01
3.16600680e-01 1.20446444e+00 -1.16537106e+00 -7.13793814e-01
3.11001688e-01 3.88190657e-01 -5.03290415e-01 3.87877345e-01
1.40559375e-01 3.65608603e-01 -5.35294712e-01 3.57202530e-01
2.12516189e-01 -8.89649332e-01 6.94436550e-01 -3.38724434e-01
2.76094347e-01 1.95829004e-01 -8.87993157e-01 1.54110157e+00
-3.77847135e-01 7.52497196e-01 -1.90128073e-01 -1.56702459e+00
1.22119451e+00 1.99126422e-01 6.73995912e-01 -5.61217308e-01
2.09765851e-01 4.66484986e-02 6.44759536e-02 -1.64023519e-01
1.95237800e-01 -1.49819767e-02 -1.54896855e-01 5.99597573e-01
5.21879554e-01 1.38657853e-01 1.50278002e-01 8.15708756e-01
1.21397889e+00 -4.11002547e-01 7.19569623e-01 -3.66006345e-01
4.99521643e-01 -6.17308974e-01 2.17483699e-01 4.58567351e-01
1.15719534e-01 -7.54833296e-02 8.65846395e-01 -2.18722910e-01
-8.17737937e-01 -1.18330336e+00 5.41474037e-02 8.60760868e-01
-6.48360094e-03 -7.10447669e-01 -4.42044765e-01 -6.72637761e-01
5.73712699e-02 1.94174111e-01 -7.56258309e-01 -4.60403025e-01
-2.28573740e-01 -4.89811152e-01 3.53716403e-01 4.34624493e-01
3.63660008e-01 -8.45025063e-01 2.30291381e-01 1.04352765e-01
-3.72354984e-02 -9.76724386e-01 -2.29841396e-01 2.82123119e-01
-9.90797222e-01 -1.11968148e+00 -3.34665775e-01 -1.04140389e+00
9.71042037e-01 2.94963121e-01 1.17481339e+00 3.38667572e-01
-2.97390789e-01 4.46059763e-01 -3.06614220e-01 2.08107695e-01
-4.87277001e-01 2.41732672e-01 -4.68164831e-02 1.76616475e-01
6.22350909e-02 -1.09443629e+00 -1.80340737e-01 2.02733368e-01
-9.34979379e-01 2.14748815e-01 8.33662927e-01 9.26203668e-01
2.66089857e-01 1.38022676e-01 6.70182467e-01 -1.05354774e+00
7.41155326e-01 -6.54038668e-01 -3.50464165e-01 3.38965267e-01
-8.51531744e-01 6.23713911e-01 6.15280032e-01 -2.66960058e-02
-5.61686635e-01 -1.09780416e-01 1.88482478e-01 -6.51468560e-02
3.43343318e-01 8.43644857e-01 -3.80234241e-01 -3.03975433e-01
4.42515999e-01 3.30225468e-01 2.33278438e-01 -2.28698850e-01
6.98511302e-01 4.24134552e-01 4.28406864e-01 -2.56397575e-01
1.06448662e+00 1.17730983e-01 4.65459287e-01 -9.09241080e-01
-6.75788939e-01 -5.28537035e-01 -1.01933372e+00 -2.54282773e-01
5.05185008e-01 -6.28130972e-01 -8.11900795e-01 9.76738557e-02
-9.86346960e-01 6.12465777e-02 -2.50017285e-01 3.79815191e-01
-5.01389086e-01 5.91517925e-01 -8.12284231e-01 -6.42593145e-01
-2.45850369e-01 -6.57475948e-01 4.89496678e-01 -1.71618372e-01
2.82474719e-02 -1.69443560e+00 2.14766309e-01 -3.85056138e-02
3.51571590e-01 4.36048180e-01 1.26264405e+00 -5.92290044e-01
-4.25173908e-01 -2.77955234e-01 -4.83856261e-01 2.32166871e-01
4.36768770e-01 -2.06600562e-01 -4.80167449e-01 -2.80347973e-01
-4.40088511e-01 -1.82057321e-01 9.74987149e-01 -6.50890321e-02
9.45707321e-01 -6.38947666e-01 -5.16039312e-01 4.93874997e-01
1.48953044e+00 -3.30817997e-01 5.02499580e-01 -2.05659524e-01
1.09085214e+00 5.84517002e-01 -4.71020266e-02 2.41387159e-01
3.89918566e-01 4.01085973e-01 5.67804337e-01 -5.81220090e-02
1.04775853e-01 -7.68782318e-01 2.24347338e-01 1.83081698e+00
-1.29343614e-01 -2.39650726e-01 -6.39609337e-01 3.57676089e-01
-1.80114532e+00 -9.63641644e-01 -2.96997875e-02 2.13013077e+00
8.19324136e-01 1.00264147e-01 5.93834184e-02 3.13086927e-01
7.47869670e-01 4.47171897e-01 -2.41938069e-01 -2.36114681e-01
-1.56923920e-01 6.34215400e-02 4.39264148e-01 5.12826800e-01
-9.27628100e-01 8.51528287e-01 6.28005648e+00 5.59702754e-01
-7.75510788e-01 -9.16429982e-02 3.86682749e-01 5.31238616e-01
-6.77580237e-01 1.92787290e-01 4.53714617e-02 1.70639560e-01
9.22256947e-01 -3.75634432e-01 4.67344195e-01 7.82704830e-01
-4.76485312e-01 4.91382033e-01 -1.25935411e+00 9.21835780e-01
5.96414804e-02 -1.40259862e+00 1.84798837e-01 3.37017357e-01
6.15899682e-01 -1.44348964e-01 -5.41727338e-03 1.44567296e-01
5.19523144e-01 -1.05875778e+00 1.93990022e-02 7.53017485e-01
7.29537666e-01 -6.99335814e-01 7.07198143e-01 2.10993871e-01
-1.51928353e+00 7.48321190e-02 -4.65679258e-01 -2.46337563e-01
-1.46535998e-02 9.12124336e-01 -9.36282218e-01 8.93670201e-01
1.41299115e-02 1.20588863e+00 -5.86648405e-01 6.18460715e-01
-3.88445497e-01 5.20458579e-01 -1.79201469e-01 -3.04225862e-01
2.18660519e-01 -5.38794398e-01 5.35297394e-01 1.06876373e+00
7.21098408e-02 -3.46231386e-02 1.14587560e-01 8.93591583e-01
-4.17883068e-01 2.94961751e-01 -1.37302804e+00 -5.40012777e-01
4.24900234e-01 1.27682471e+00 -5.03371477e-01 -1.64024875e-01
-3.56256366e-01 8.83284867e-01 8.76469195e-01 1.92529336e-01
-5.30969799e-01 -5.94863594e-01 1.92257479e-01 2.15731300e-02
1.15607120e-01 -2.96297938e-01 8.38215351e-02 -1.25245881e+00
1.98237419e-01 -3.76478255e-01 4.14472997e-01 -3.80955160e-01
-1.42302299e+00 7.32899904e-01 -2.39603415e-01 -1.30973196e+00
-4.85181868e-01 -7.49756813e-01 -6.24413073e-01 3.37078303e-01
-1.28407466e+00 -1.13410032e+00 -1.31175414e-01 7.01235950e-01
-3.26668680e-01 -2.80725896e-01 1.04627407e+00 2.23970070e-01
-4.66998667e-01 7.61366606e-01 2.53708035e-01 5.66282392e-01
6.57251999e-02 -1.35654342e+00 1.42790243e-01 4.50565875e-01
6.71550989e-01 6.35219693e-01 3.03447008e-01 -4.20992434e-01
-1.74910831e+00 -1.03020740e+00 1.05163717e+00 -1.83311954e-01
1.22162294e+00 -2.84880519e-01 -8.61782432e-01 8.57462406e-01
-6.26614736e-03 2.87310272e-01 8.56807888e-01 3.51412505e-01
-6.95869386e-01 -7.46015608e-02 -7.76768267e-01 4.98669714e-01
1.40835249e+00 -8.99039447e-01 -3.63948524e-01 4.78472501e-01
8.89517784e-01 2.31104285e-01 -1.39585006e+00 3.06643516e-01
4.37239528e-01 -8.91043723e-01 9.07313406e-01 -8.37266266e-01
3.42033386e-01 4.84349988e-02 -2.34044075e-01 -1.43013525e+00
-5.63736260e-01 -7.21226871e-01 -3.76802057e-01 1.10782111e+00
6.04540825e-01 -8.74039769e-01 6.08894467e-01 3.05659890e-01
2.73561597e-01 -7.51223862e-01 -8.05263221e-01 -6.81185126e-01
-1.61031201e-01 -1.80736676e-01 4.46600378e-01 1.31535423e+00
7.96094298e-01 7.09087551e-01 -5.05841374e-01 -7.85717592e-02
8.68900716e-01 1.61293268e-01 5.88913202e-01 -1.63043165e+00
-3.78600746e-01 -4.99773234e-01 -1.20904374e+00 -9.47325230e-01
4.71169591e-01 -1.55656290e+00 -3.96555245e-01 -1.44893527e+00
2.77259231e-01 -4.15290147e-01 -6.06931448e-01 3.64228487e-01
1.96606547e-01 1.38578545e-02 1.58428952e-01 2.65281349e-01
-7.66622305e-01 8.12472403e-01 1.21969187e+00 -3.09348911e-01
-1.29324302e-01 -9.84690636e-02 -6.66697085e-01 6.59635007e-01
6.61655843e-01 -3.46710145e-01 -4.72990811e-01 -6.58188611e-02
5.57677269e-01 2.75937438e-01 4.15951818e-01 -8.09985280e-01
3.50593358e-01 8.26193839e-02 6.65301308e-02 -1.54072300e-01
1.60881095e-02 -9.05531943e-01 6.36458546e-02 4.46500838e-01
-5.89499354e-01 -3.94765623e-02 -4.82983232e-01 8.87962162e-01
-3.46372932e-01 -1.42384231e-01 4.34150934e-01 -5.14603890e-02
-6.28600240e-01 4.42392141e-01 -6.13678843e-02 -3.47475857e-02
7.46455550e-01 3.07016000e-02 -2.23734856e-01 -6.38673663e-01
-6.93906963e-01 -1.22475382e-02 2.51524895e-01 3.74623120e-01
7.84631312e-01 -1.70312822e+00 -4.72318739e-01 3.07807624e-01
3.22487563e-01 -5.84453762e-01 -1.89682424e-01 8.89664710e-01
-1.40637577e-01 5.32691777e-01 -1.13644883e-01 -4.43604112e-01
-9.02347982e-01 6.52024209e-01 3.48427449e-03 -6.37450814e-01
-6.45939350e-01 3.96218479e-01 1.78080767e-01 -6.63800478e-01
-7.04592317e-02 -1.42571265e-02 -1.51157290e-01 -1.15944780e-01
1.78445220e-01 1.67979255e-01 -2.29026213e-01 -8.50505292e-01
-2.07383499e-01 4.10969883e-01 5.94493411e-02 2.60292023e-01
1.27225089e+00 2.20684391e-02 -5.96719384e-01 5.26564658e-01
1.88034987e+00 -2.49155134e-01 -6.39312148e-01 -6.96008146e-01
1.88282162e-01 -2.36916989e-01 7.26964548e-02 -6.90142885e-02
-1.27211523e+00 8.34972739e-01 5.49867861e-02 6.67623401e-01
8.52756619e-01 3.82751882e-01 4.62287694e-01 7.65270114e-01
4.91616994e-01 -7.58982062e-01 2.80977279e-01 3.00730884e-01
7.31691360e-01 -1.23040926e+00 3.33845578e-02 -6.22698247e-01
-3.77983898e-01 1.06060255e+00 2.58753777e-01 -1.73956692e-01
1.12368500e+00 -7.68535137e-02 -6.70525372e-01 -4.74730760e-01
-7.39260197e-01 -3.36129218e-01 6.99412942e-01 5.48324168e-01
6.37802899e-01 4.23395932e-01 -2.33737111e-01 2.06090182e-01
-2.10977763e-01 -4.11730796e-01 2.27089286e-01 6.71842635e-01
-5.38286626e-01 -1.00883043e+00 4.36802119e-01 6.88191295e-01
5.73457256e-02 -1.16135860e-02 -7.27673113e-01 8.75508189e-01
-4.71017778e-01 7.79545248e-01 -8.18889886e-02 -7.58073807e-01
-1.58715799e-01 -2.01159894e-01 4.00238156e-01 -4.30848062e-01
8.78296513e-03 -5.60460865e-01 1.58520844e-02 -5.20921886e-01
-5.67588747e-01 -8.85976255e-02 -1.09563577e+00 -3.58727485e-01
-6.91744685e-01 3.70975286e-01 4.74850327e-01 9.13307548e-01
2.92217076e-01 5.23796856e-01 9.91314292e-01 -6.24358773e-01
-3.88915330e-01 -8.64665508e-01 -8.84243548e-01 4.69726086e-01
1.27221644e-01 -6.29739761e-01 -5.52032113e-01 -2.58261889e-01] | [7.1574225425720215, 6.20635461807251] |
6bc1058c-d6a8-4351-a0db-844271293b1e | probabilistic-linguistic-knowledge-and-token | 2306.16644 | null | https://arxiv.org/abs/2306.16644v2 | https://arxiv.org/pdf/2306.16644v2.pdf | Probabilistic Linguistic Knowledge and Token-level Text Augmentation | This paper investigates the effectiveness of token-level text augmentation and the role of probabilistic linguistic knowledge within a linguistically-motivated evaluation context. Two text augmentation programs, REDA and REDA$_{NG}$, were developed, both implementing five token-level text editing operations: Synonym Replacement (SR), Random Swap (RS), Random Insertion (RI), Random Deletion (RD), and Random Mix (RM). REDA$_{NG}$ leverages pretrained $n$-gram language models to select the most likely augmented texts from REDA's output. Comprehensive and fine-grained experiments were conducted on a binary question matching classification task in both Chinese and English. The results strongly refute the general effectiveness of the five token-level text augmentation techniques under investigation, whether applied together or separately, and irrespective of various common classification model types used, including transformers. Furthermore, the role of probabilistic linguistic knowledge is found to be minimal. | ['Zhengxiang Wang'] | 2023-06-29 | null | null | null | null | ['text-augmentation'] | ['natural-language-processing'] | [ 5.43697119e-01 1.38826549e-01 -2.04447374e-01 -6.20304227e-01
-1.06086516e+00 -4.77351010e-01 8.56272161e-01 5.77452242e-01
-1.06866491e+00 6.95450962e-01 4.64413285e-01 -8.89238298e-01
-1.70807555e-01 -7.66020834e-01 -5.12127757e-01 -1.98462024e-01
3.03454787e-01 5.52073359e-01 -1.45433605e-01 -3.99340063e-01
3.24878782e-01 2.21745253e-01 -1.50901008e+00 4.47911799e-01
9.59457159e-01 5.79869688e-01 3.30939703e-02 5.28524816e-01
-6.44348741e-01 8.87154281e-01 -7.69500017e-01 -9.56135333e-01
7.71670640e-02 -8.99957046e-02 -1.08360195e+00 -2.34776542e-01
4.12975818e-01 -1.10257708e-01 -1.47940189e-01 8.42822671e-01
5.22155523e-01 3.32385957e-01 5.63140690e-01 -9.32398558e-01
-6.88208044e-01 1.10840786e+00 -4.20234293e-01 2.74812609e-01
5.19574821e-01 2.24483937e-01 1.52411902e+00 -1.00886273e+00
4.74841177e-01 1.24090934e+00 9.02021945e-01 3.36940587e-01
-1.23719430e+00 -7.16196120e-01 1.53410733e-01 -2.81797677e-01
-1.33971250e+00 -4.67277378e-01 6.42136514e-01 -2.99120426e-01
1.55137718e+00 4.02884632e-01 5.16477190e-02 7.80343592e-01
-2.04628468e-01 8.33552241e-01 1.30075443e+00 -1.19632077e+00
-1.19842805e-01 3.12572300e-01 4.45649445e-01 5.37499368e-01
3.22824538e-01 -3.99201997e-02 -4.70247269e-01 -5.33035755e-01
2.33800992e-01 -5.67779779e-01 6.49684817e-02 2.83558041e-01
-1.11709702e+00 9.57234681e-01 -2.28449836e-01 5.26112020e-01
-3.11808169e-01 1.11819424e-01 7.23645687e-01 2.88393825e-01
4.69998121e-01 7.79854298e-01 -9.00027633e-01 -2.27241665e-01
-8.17023337e-01 4.07679260e-01 5.05981684e-01 1.01861799e+00
7.21127927e-01 1.71578273e-01 -5.64024866e-01 1.26268291e+00
2.83164799e-01 4.60915148e-01 7.97269106e-01 -8.39078069e-01
7.42465198e-01 6.90395772e-01 7.13914931e-02 -6.84697151e-01
-3.27550679e-01 -2.07596779e-01 -2.78475672e-01 -1.53763711e-01
5.42484939e-01 -2.77403951e-01 -8.49220335e-01 2.03459430e+00
1.40160426e-01 -4.16383266e-01 -3.59810889e-02 1.40820667e-01
7.34892488e-01 3.36973339e-01 6.85270786e-01 1.36587307e-01
1.84318042e+00 -5.69430470e-01 -6.17652059e-01 -5.57721794e-01
1.31167936e+00 -1.06751454e+00 1.64932668e+00 1.35234231e-02
-1.13696706e+00 -5.87907672e-01 -6.91049993e-01 -1.79955602e-01
-6.89475119e-01 3.29029053e-01 8.03496420e-01 1.32146931e+00
-9.55304801e-01 2.45328799e-01 -3.24067235e-01 -3.03001255e-01
3.70429419e-02 2.71384984e-01 -2.53417581e-01 -2.58508921e-01
-1.53789282e+00 1.21460545e+00 4.38370824e-01 5.45254909e-03
-8.12404603e-02 -5.61034262e-01 -9.73961532e-01 -2.00148314e-01
8.70325938e-02 -5.56781471e-01 1.38451457e+00 -7.69168258e-01
-1.13475943e+00 1.19582522e+00 -4.68465447e-01 -3.25449735e-01
2.91822433e-01 -3.75837088e-01 -4.51178342e-01 -2.52128601e-01
1.85448840e-01 6.43674672e-01 4.34305787e-01 -8.87064695e-01
-6.98992908e-01 -2.27724507e-01 5.85559867e-02 4.06233430e-01
-3.30429226e-01 7.75258601e-01 -2.74494924e-02 -1.15143502e+00
-1.98482990e-01 -6.62954986e-01 -1.58139735e-01 -6.10444009e-01
-3.06211263e-01 -5.82647562e-01 5.26968658e-01 -8.51427794e-01
1.52493489e+00 -1.85792089e+00 -4.37724322e-01 2.65878797e-01
-2.66406804e-01 4.13357824e-01 -2.42395878e-01 4.87892002e-01
-3.69366646e-01 5.53437293e-01 -3.68448675e-01 -4.17010039e-01
1.36068270e-01 1.87855586e-01 -2.63823807e-01 -2.48801261e-02
3.68945390e-01 1.08855021e+00 -7.99637675e-01 -4.74156380e-01
8.16454962e-02 1.30848616e-01 -5.96337616e-01 -1.78571016e-01
-4.03616935e-01 -2.22107008e-01 -2.40944669e-01 5.27211249e-01
2.67912388e-01 1.41117647e-01 2.25162789e-01 -3.74012589e-02
-2.16998369e-01 1.04350603e+00 -1.05404770e+00 1.26509368e+00
-6.00576997e-01 4.30031389e-01 3.57580483e-02 -6.98404074e-01
9.18227613e-01 3.39625359e-01 1.25197977e-01 -6.46164179e-01
1.63908347e-01 2.98946112e-01 1.50760293e-01 -3.16793382e-01
1.11890721e+00 -3.00209731e-01 -3.54693085e-01 8.89307022e-01
-1.07295513e-01 -5.29487319e-02 2.61603624e-01 3.21079880e-01
1.19707024e+00 2.35200033e-01 2.70837903e-01 -3.15955102e-01
5.32050788e-01 2.64811307e-01 3.38508725e-01 1.24552500e+00
-1.34708583e-01 2.76539147e-01 3.20356250e-01 1.00484975e-01
-1.10478675e+00 -6.58914924e-01 -2.10726395e-01 1.77704251e+00
-4.85434860e-01 -3.99254501e-01 -5.57049036e-01 -8.07620585e-01
9.36944112e-02 1.36550534e+00 -5.87551355e-01 -1.93385288e-01
-8.81346941e-01 -1.09604371e+00 1.15655828e+00 5.09322345e-01
2.17905939e-01 -1.33383965e+00 -3.28771502e-01 4.15540189e-01
-4.82466966e-01 -9.94907379e-01 -4.69378144e-01 3.81828666e-01
-6.94546044e-01 -8.34546745e-01 -4.23297584e-01 -8.55416596e-01
4.38174993e-01 -1.87851060e-02 1.34400856e+00 2.11837545e-01
-1.13783598e-01 4.96814817e-01 -6.27180278e-01 -3.86893511e-01
-6.02404118e-01 3.00861061e-01 -1.06885649e-01 -3.49530011e-01
8.54951799e-01 -2.44978234e-01 -1.51668027e-01 1.37473673e-01
-9.79403853e-01 -2.56259948e-01 7.88962364e-01 9.20826435e-01
2.34869495e-01 -8.91770571e-02 8.15716624e-01 -1.37001538e+00
1.04198837e+00 -2.31787756e-01 -1.67116180e-01 4.53687996e-01
-8.71882617e-01 1.42886847e-01 4.20044869e-01 -4.58782524e-01
-1.23601496e+00 -3.22510600e-01 -5.02694547e-01 1.98774785e-01
-3.65791172e-01 8.02219272e-01 -2.57115364e-01 1.86618403e-01
7.22551286e-01 5.40898181e-02 -1.67715773e-01 -4.73145902e-01
5.53971350e-01 7.63996601e-01 4.20266777e-01 -9.02083218e-01
7.29915440e-01 7.28576183e-02 -7.26659358e-01 -8.54534984e-01
-5.84922671e-01 -5.22082150e-01 -5.36266267e-01 7.31059760e-02
6.46910191e-01 -7.35858262e-01 -3.10867190e-01 6.23610914e-01
-1.01167190e+00 -4.33826864e-01 -5.69483399e-01 4.69524682e-01
-9.57995504e-02 4.86138076e-01 -7.68782258e-01 -8.15283358e-01
-4.47788686e-01 -8.35682869e-01 6.76128209e-01 -1.50748253e-01
-7.26987839e-01 -8.63410592e-01 -2.43071765e-01 7.50305414e-01
4.13850099e-01 -1.96821898e-01 1.57725465e+00 -1.13234973e+00
-3.95531878e-02 -4.55925792e-01 -2.05784470e-01 4.31917042e-01
1.63636386e-01 7.67051056e-02 -8.83269489e-01 -5.61295897e-02
-1.76583067e-01 -4.48446453e-01 5.69982946e-01 1.48740798e-01
8.31616044e-01 -2.94315994e-01 8.08896497e-02 -2.08445992e-02
9.77003157e-01 1.69278011e-01 5.42690992e-01 4.83634233e-01
4.47044224e-01 6.78642154e-01 6.10399127e-01 2.16657981e-01
5.38191855e-01 4.46915656e-01 -3.38286795e-02 -1.51717551e-02
-1.10047847e-01 -2.30957955e-01 3.90550226e-01 9.67071533e-01
4.39356744e-01 -2.77414322e-01 -1.11585391e+00 6.60141349e-01
-1.26947570e+00 -7.88461864e-01 -1.47199079e-01 2.28909540e+00
1.19497466e+00 3.82402897e-01 -5.89918792e-02 2.61099964e-01
7.42947936e-01 2.41258875e-01 -9.40876231e-02 -7.13666677e-01
-4.59803760e-01 7.60302544e-01 5.26453495e-01 5.86470425e-01
-1.00437629e+00 1.09352577e+00 6.80586386e+00 8.85318458e-01
-6.36915743e-01 7.09986985e-02 6.57850862e-01 1.95114627e-01
-7.59464264e-01 1.30785629e-01 -8.98723185e-01 2.75449544e-01
9.34698045e-01 -1.10575326e-01 1.71318725e-01 6.12180114e-01
-4.68580723e-02 -3.08017880e-01 -8.37510288e-01 4.25889283e-01
8.06544572e-02 -1.01950645e+00 3.45522672e-01 -3.31314802e-01
3.93290430e-01 8.83571133e-02 5.62852621e-02 8.41829062e-01
8.75018418e-01 -9.32821214e-01 7.79506385e-01 1.17323555e-01
8.50689054e-01 -7.05939651e-01 7.83006668e-01 2.90757209e-01
-1.07746255e+00 9.24923085e-03 -6.20492995e-02 -7.75796100e-02
1.97486222e-01 3.57844442e-01 -7.94509411e-01 4.41755980e-01
5.32765746e-01 -1.03378452e-01 -8.53734910e-01 3.82678926e-01
-2.49696687e-01 1.01373088e+00 -3.52594525e-01 -2.30273947e-01
3.12445670e-01 -1.24990873e-01 4.59717095e-01 1.65420496e+00
-1.07413188e-01 1.51110917e-01 -1.79422811e-01 4.89659667e-01
-3.36189002e-01 4.76503164e-01 -8.94727260e-02 -1.04472481e-01
8.14144433e-01 1.11543238e+00 -6.13813937e-01 -5.02614558e-01
-5.56988418e-01 6.13752961e-01 3.86990458e-01 2.40912065e-01
-4.92593110e-01 -8.51879179e-01 4.85124648e-01 1.18260927e-01
8.74852091e-02 1.10514024e-02 -6.00646377e-01 -8.86976361e-01
1.56096891e-01 -1.21779466e+00 6.72191739e-01 -7.70122409e-01
-1.24366724e+00 3.86728704e-01 1.04912445e-01 -4.77749765e-01
-3.81076604e-01 -5.45165539e-01 -5.51891446e-01 1.36553693e+00
-1.37357318e+00 -1.10192811e+00 1.38174310e-01 3.35663408e-01
4.05657858e-01 -1.04265012e-01 9.14135754e-01 4.84926850e-01
-5.36896706e-01 1.09728301e+00 -1.51415747e-02 5.79080105e-01
6.69324994e-01 -1.33002996e+00 7.54555225e-01 9.01663780e-01
5.57738245e-02 9.75628436e-01 5.76521575e-01 -8.12926769e-01
-1.09024429e+00 -1.07418120e+00 1.62822747e+00 -6.15508258e-01
6.67157531e-01 -4.28857952e-01 -1.19576323e+00 7.27283895e-01
1.41033188e-01 -5.68157971e-01 7.83360183e-01 5.21276057e-01
-6.04660094e-01 1.71294853e-01 -1.27671158e+00 7.19988644e-01
7.09850967e-01 -9.49545920e-01 -9.11719143e-01 2.50531912e-01
8.23988855e-01 -5.46842581e-03 -8.49042594e-01 4.79773492e-01
3.76319379e-01 -5.15176058e-01 8.36073935e-01 -8.73741448e-01
2.28227928e-01 3.20593789e-02 -3.04181784e-01 -8.98233235e-01
-2.27610409e-01 -4.32623655e-01 4.10941511e-01 1.66754520e+00
8.89260590e-01 -8.26081216e-01 6.93706274e-01 9.38957810e-01
-2.93440521e-01 -5.42205632e-01 -1.03219867e+00 -6.72875464e-01
5.54409564e-01 -8.81242812e-01 7.03053057e-01 1.35301971e+00
4.72255610e-02 3.61984342e-01 -4.19001374e-03 -2.25998119e-01
2.07890630e-01 -3.70319039e-01 5.54545879e-01 -8.88957441e-01
-2.98664629e-01 -7.86909580e-01 6.96933419e-02 -8.35655630e-01
2.23398298e-01 -1.05724657e+00 5.66577446e-03 -1.41335618e+00
-6.16340004e-02 -1.04511106e+00 -2.61643052e-01 7.52240181e-01
-6.78768039e-01 -8.58595967e-03 -8.42715707e-03 -1.06169984e-01
-8.29812437e-02 4.16761279e-01 4.91340756e-01 6.28062412e-02
-1.41966522e-01 -5.71122095e-02 -9.11583126e-01 7.09430814e-01
8.89458001e-01 -5.12418211e-01 -9.13660694e-03 -6.93122149e-01
3.32346380e-01 -2.72284210e-01 -9.73282382e-03 -4.04591620e-01
1.19964831e-01 7.25551182e-03 7.10610151e-02 -4.98672605e-01
2.86632776e-01 -4.16999012e-01 -3.76286417e-01 3.20409536e-01
-8.23904812e-01 5.89731276e-01 5.04536688e-01 1.13390341e-01
-6.31929114e-02 -6.85965598e-01 6.27130628e-01 -1.49785295e-01
-5.10559559e-01 -1.81141585e-01 -6.86331987e-01 6.04322255e-01
4.62294906e-01 -3.51820379e-01 -2.25091770e-01 -1.44010723e-01
-3.73131961e-01 2.03484327e-01 1.20177209e-01 5.46107471e-01
1.82410195e-01 -1.13237953e+00 -8.71139288e-01 1.66097924e-01
4.03155714e-01 -2.65212119e-01 1.41686397e-02 4.23952967e-01
-3.28413218e-01 5.97873151e-01 4.24915403e-01 1.21620046e-02
-1.04569912e+00 2.09734380e-01 2.64941484e-01 -6.07290268e-01
-1.99727193e-01 1.04361463e+00 -4.13290024e-01 -1.01125228e+00
2.96718627e-01 -3.06744307e-01 -6.78956434e-02 -3.39333666e-03
2.17972189e-01 4.53875452e-01 3.33871633e-01 -7.09931254e-01
-1.98269576e-01 -9.20838416e-02 -3.76676857e-01 -5.31832337e-01
7.85674453e-01 -1.07188798e-01 -2.06441775e-01 2.79628694e-01
8.08247924e-01 2.46942684e-01 -2.62515426e-01 -7.04160452e-01
5.83495736e-01 -1.90020248e-01 -1.09684013e-01 -1.10869873e+00
-4.64277565e-01 6.63791597e-01 1.28956437e-01 -2.18400024e-02
8.73852730e-01 -1.94855198e-01 5.64971626e-01 4.37570810e-01
1.83218047e-02 -1.39844728e+00 -3.62107903e-01 7.62536585e-01
5.46611547e-01 -9.32262659e-01 -8.03725049e-02 -2.03459546e-01
-6.45988405e-01 4.73983616e-01 6.73404276e-01 3.91203761e-01
5.72042942e-01 1.56152278e-01 1.79613814e-01 -2.71121447e-04
-7.01374471e-01 -2.47499585e-01 1.02897801e-01 4.06149864e-01
1.02271962e+00 -9.27151814e-02 -4.96743977e-01 5.47134399e-01
-5.84427953e-01 -4.41638261e-01 2.89557576e-01 1.19256926e+00
-3.26371431e-01 -1.43993676e+00 -3.94521445e-01 1.01687574e+00
-7.14382768e-01 -7.55517423e-01 -4.11212474e-01 1.14966130e+00
1.26860619e-01 1.12257504e+00 5.44631332e-02 -2.24986374e-01
3.67400438e-01 7.94406056e-01 1.10252321e-01 -8.46416771e-01
-1.32159495e+00 -1.98279873e-01 4.81047660e-01 9.48789567e-02
-2.62850940e-01 -1.16050065e+00 -1.07990611e+00 -2.71926820e-01
-6.11218989e-01 4.40919876e-01 6.51821434e-01 1.12508667e+00
1.31527305e-01 2.54970372e-01 2.80945599e-01 -4.19090129e-02
-6.83510005e-01 -1.29270589e+00 -3.01619440e-01 5.58966458e-01
-1.91828668e-01 -2.77599305e-01 -3.16930652e-01 -1.48990974e-01] | [10.78712272644043, 9.611377716064453] |
60c50b80-1a40-475f-a4cf-49923340bc6c | learning-to-segment-dominant-object-motion | 2111.14160 | null | https://arxiv.org/abs/2111.14160v1 | https://arxiv.org/pdf/2111.14160v1.pdf | Learning To Segment Dominant Object Motion From Watching Videos | Existing deep learning based unsupervised video object segmentation methods still rely on ground-truth segmentation masks to train. Unsupervised in this context only means that no annotated frames are used during inference. As obtaining ground-truth segmentation masks for real image scenes is a laborious task, we envision a simple framework for dominant moving object segmentation that neither requires annotated data to train nor relies on saliency priors or pre-trained optical flow maps. Inspired by a layered image representation, we introduce a technique to group pixel regions according to their affine parametric motion. This enables our network to learn segmentation of the dominant foreground object using only RGB image pairs as input for both training and inference. We establish a baseline for this novel task using a new MovingCars dataset and show competitive performance against recent methods that require annotated masks to train. | ['Nick Barnes', 'Hongdong Li', 'Mohammad Ali Armin', 'Sahir Shrestha'] | 2021-11-28 | null | null | null | null | ['unsupervised-video-object-segmentation'] | ['computer-vision'] | [ 7.44162261e-01 3.38463187e-01 -2.85962164e-01 -5.13655543e-01
-7.00331330e-01 -7.58620560e-01 5.99728942e-01 -8.85322317e-02
-8.14488769e-01 6.05082095e-01 -3.04192960e-01 -2.65337139e-01
3.72646093e-01 -6.89370155e-01 -1.13671708e+00 -6.15397155e-01
1.13856360e-01 4.59518850e-01 9.80495751e-01 -2.02372484e-02
1.79024726e-01 3.69920075e-01 -1.38324225e+00 2.34678224e-01
8.04641604e-01 8.11857998e-01 5.24720490e-01 8.08581710e-01
-2.10582703e-01 8.84349048e-01 -5.09770811e-01 -2.12433025e-01
5.85734189e-01 -6.61118329e-01 -1.34291518e+00 5.60339630e-01
5.80423236e-01 -6.17460907e-01 -1.18238568e-01 1.04367578e+00
-2.14576721e-01 3.85726213e-01 6.30741417e-01 -1.21658337e+00
-1.14446424e-01 5.78371525e-01 -4.16973919e-01 3.08217347e-01
4.21547927e-02 3.73116106e-01 9.54356015e-01 -5.23312926e-01
8.02187324e-01 9.22775269e-01 5.12445152e-01 5.80157816e-01
-1.48302007e+00 -3.33462469e-02 4.35985088e-01 2.01922923e-01
-1.01756120e+00 -4.88518864e-01 8.34193528e-01 -5.17566681e-01
5.51464677e-01 2.18488574e-01 8.68545175e-01 9.14034545e-01
-3.25333059e-01 1.12909126e+00 9.01540875e-01 -3.60568196e-01
5.25108397e-01 1.72998756e-02 8.29048753e-02 8.06734562e-01
2.20518798e-01 -1.74031466e-01 -3.00341129e-01 4.68917757e-01
1.19669175e+00 -1.43154353e-01 -3.35153133e-01 -7.38373935e-01
-1.22369611e+00 8.46916378e-01 6.03243589e-01 1.71116441e-01
-3.48047405e-01 5.08995473e-01 4.17063534e-02 -1.84058636e-01
2.70100296e-01 4.01988029e-01 -5.95982969e-01 9.53402519e-02
-1.72957921e+00 1.19505428e-01 7.00307250e-01 8.41606915e-01
1.33692956e+00 3.80597860e-02 -1.32146269e-01 2.70067990e-01
2.46504351e-01 2.32385442e-01 1.64559692e-01 -1.47964454e+00
3.79138999e-02 5.48609138e-01 3.74667078e-01 -7.87774265e-01
-3.09651107e-01 -2.75141329e-01 -4.13751513e-01 2.82003582e-01
9.92391586e-01 -1.13027036e-01 -1.47397840e+00 1.37331688e+00
4.34472382e-01 4.85645831e-01 3.59349027e-02 1.17066801e+00
5.68264306e-01 5.14061630e-01 -7.98120052e-02 -1.40302017e-01
9.61616218e-01 -1.27628720e+00 -4.43542719e-01 -4.35797065e-01
4.83338177e-01 -4.16242272e-01 1.00904715e+00 2.49181122e-01
-1.24033403e+00 -5.15125334e-01 -9.80903625e-01 -2.90331066e-01
-2.98294723e-01 1.30573064e-02 8.33070576e-01 8.02772701e-01
-1.20506048e+00 6.50785625e-01 -1.26047480e+00 -2.64710426e-01
7.88075030e-01 4.71529573e-01 -2.03159377e-01 1.30766243e-01
-7.69133627e-01 6.76530361e-01 7.30216622e-01 2.02498958e-01
-1.29082036e+00 -5.21253467e-01 -9.54570591e-01 -2.04423934e-01
5.88173747e-01 -7.91375458e-01 1.18764567e+00 -1.58204782e+00
-1.66357493e+00 8.26947808e-01 -2.69783616e-01 -8.54809761e-01
8.10248971e-01 -3.26878846e-01 5.21820366e-01 7.93474317e-01
1.37384813e-02 1.30712664e+00 1.09423614e+00 -1.60089469e+00
-6.37660146e-01 1.56693906e-01 4.99315351e-01 -8.13816637e-02
9.76482406e-02 -1.94510534e-01 -7.98865736e-01 -5.43049455e-01
1.60455257e-01 -8.07622313e-01 -5.59032917e-01 -4.76580374e-02
-6.11341953e-01 -9.14255902e-03 1.15346289e+00 -6.83730543e-01
6.73028529e-01 -1.50656521e+00 2.76891708e-01 1.84123427e-01
7.94122592e-02 2.04883173e-01 -1.13347754e-01 -2.67615736e-01
2.86414951e-01 1.16555668e-01 -9.38670754e-01 -4.99664873e-01
-1.01245627e-01 5.06658316e-01 -2.97298253e-01 5.80167234e-01
6.42277539e-01 1.06908488e+00 -1.04130733e+00 -7.77544856e-01
6.05442882e-01 4.69698519e-01 -9.53810811e-01 3.02888721e-01
-7.71770120e-01 9.02974248e-01 -1.97264344e-01 5.05846024e-01
5.55183530e-01 -2.41688728e-01 -6.18056133e-02 -1.26792744e-01
-1.30712584e-01 9.70209017e-02 -1.03359008e+00 2.07479620e+00
-1.40455276e-01 7.01962948e-01 1.00845233e-01 -1.45547140e+00
4.11357552e-01 -2.28253007e-02 6.63726628e-01 -2.04356864e-01
3.52736056e-01 7.37250447e-02 -1.24883033e-01 -4.77571398e-01
4.07679468e-01 -4.82397387e-03 1.48229599e-01 3.15203607e-01
3.71841431e-01 -5.75495362e-01 5.29793561e-01 1.80006564e-01
9.88077939e-01 8.00380826e-01 -3.37159008e-01 -3.48296851e-01
5.70980430e-01 3.79602998e-01 7.52430677e-01 8.01538646e-01
-1.42967030e-01 1.08770621e+00 4.82580006e-01 -4.85102713e-01
-9.17761981e-01 -1.05613446e+00 -2.86461855e-03 1.02244067e+00
5.32695591e-01 7.82572627e-02 -1.35519850e+00 -9.23726261e-01
-5.50280452e-01 5.87847114e-01 -6.54699147e-01 3.85960102e-01
-6.99414372e-01 -6.13799572e-01 2.59831816e-01 4.95964706e-01
6.51346445e-01 -1.08596742e+00 -1.17942965e+00 1.42831773e-01
-2.65701383e-01 -1.63466334e+00 -3.80918115e-01 2.92678326e-01
-9.53691483e-01 -1.26225162e+00 -8.79872143e-01 -8.46797466e-01
9.06785190e-01 1.53942078e-01 1.26160669e+00 2.49997765e-01
-2.89750725e-01 4.37627971e-01 -2.19952583e-01 -2.20711142e-01
-2.12564066e-01 1.66673332e-01 -4.13325012e-01 1.76237032e-01
-6.40824884e-02 -3.79025877e-01 -9.25214171e-01 1.56115562e-01
-1.03942096e+00 4.50149834e-01 5.84505677e-01 4.33783233e-01
6.26419425e-01 -2.39038855e-01 1.79440051e-01 -8.90417576e-01
-3.42382580e-01 -9.65277031e-02 -9.61079001e-01 4.60621230e-02
6.31813779e-02 8.41029137e-02 3.55237991e-01 -9.87448692e-02
-9.05029953e-01 7.47643113e-01 7.70765683e-03 -5.59863746e-01
-5.01445770e-01 1.01419255e-01 -1.66807041e-01 -9.89789516e-02
3.46509874e-01 1.93172976e-01 -2.13714391e-01 -2.80373007e-01
6.13954425e-01 8.56551304e-02 9.23343539e-01 -4.43700880e-01
9.83091950e-01 8.69657993e-01 -1.09752551e-01 -8.96992862e-01
-8.85867894e-01 -5.29141128e-01 -1.33176267e+00 -3.17564398e-01
1.53570724e+00 -7.42553890e-01 -5.91955960e-01 2.92300791e-01
-1.38029170e+00 -9.31762516e-01 -2.62847841e-01 4.24350351e-01
-8.54338646e-01 3.51767719e-01 -3.95597011e-01 -7.43758023e-01
9.56410356e-03 -1.37428594e+00 1.29467225e+00 2.97032773e-01
-3.34517434e-02 -1.06786561e+00 -4.01158512e-01 5.67095160e-01
1.67862803e-01 5.65762520e-01 4.27832931e-01 -3.15196574e-01
-1.41563082e+00 1.29580438e-01 -3.00886780e-01 4.36616600e-01
1.80762604e-01 1.52900666e-01 -9.56538320e-01 -1.92751060e-03
-1.49873048e-01 -2.04201773e-01 1.18611550e+00 5.29132366e-01
1.27319562e+00 -2.44121268e-01 -3.31481755e-01 8.67655933e-01
1.32573986e+00 -2.78939866e-02 5.53373456e-01 2.22764775e-01
1.27613521e+00 7.28716195e-01 3.43982369e-01 -5.90934679e-02
4.11647558e-01 4.80539322e-01 4.79153216e-01 -3.83157670e-01
-9.49205905e-02 -5.59428446e-02 3.26137900e-01 3.41711909e-01
-2.26034969e-01 -1.05853043e-01 -9.09631073e-01 8.67361248e-01
-1.73996544e+00 -8.15510929e-01 -1.29994243e-01 2.11430931e+00
1.08738613e+00 2.86734402e-01 3.33676964e-01 1.27641693e-01
5.09736657e-01 7.95012116e-02 -5.64403713e-01 -1.42017584e-02
-3.03935912e-02 4.41265970e-01 6.66763544e-01 8.09751630e-01
-1.53594613e+00 1.50146389e+00 5.86016560e+00 3.60874325e-01
-1.25657105e+00 1.79430768e-01 9.33748603e-01 -1.69878304e-02
-2.79801726e-01 3.22090149e-01 -5.23150742e-01 2.83841074e-01
6.51080906e-01 3.92840058e-01 3.62136453e-01 4.61078733e-01
4.03580099e-01 -6.16044044e-01 -1.15770054e+00 7.52700329e-01
-2.18826294e-01 -1.48179305e+00 3.67209390e-02 -2.93567091e-01
1.03374827e+00 4.84415032e-02 -1.71552822e-01 -5.75989485e-02
4.73557860e-01 -9.41061974e-01 9.00808156e-01 5.72139382e-01
1.67249769e-01 -5.10255635e-01 4.55523193e-01 2.95402139e-01
-9.50347722e-01 3.68476570e-01 -1.82143569e-01 -4.16123867e-03
3.78558576e-01 4.05089080e-01 -8.04062843e-01 3.19321245e-01
5.69415092e-01 6.97195292e-01 -6.35204911e-01 1.18396354e+00
-3.64937156e-01 8.26858521e-01 -5.10734916e-01 3.87703419e-01
5.71149230e-01 -3.84894043e-01 3.81435484e-01 1.14400601e+00
-1.35999411e-01 -9.34157595e-02 6.03298008e-01 1.04611897e+00
-1.85627751e-02 -2.15808630e-01 -2.04240128e-01 1.19192405e-02
-9.66841802e-02 1.22976315e+00 -1.69649541e+00 -4.79880244e-01
-3.28076184e-01 1.43568063e+00 5.71187846e-02 6.87665164e-01
-9.45885479e-01 -1.20621182e-01 3.82604182e-01 1.73993826e-01
7.11478233e-01 -5.74930489e-01 -2.82147706e-01 -1.11692810e+00
-3.27353179e-01 -3.25594991e-01 4.96571176e-02 -5.84688306e-01
-6.24511182e-01 3.80298018e-01 2.21810728e-01 -8.68177712e-01
-3.47191572e-01 -6.12142801e-01 -6.13020778e-01 5.38827896e-01
-1.67760634e+00 -1.25109816e+00 -3.45770776e-01 5.62038243e-01
6.96478903e-01 4.88366216e-01 3.38573545e-01 -3.08490992e-02
-5.38631678e-01 1.23050511e-01 -4.45953697e-01 5.43680072e-01
1.95249185e-01 -1.55107701e+00 3.55866194e-01 1.26184356e+00
4.71142292e-01 4.58075941e-01 8.11375618e-01 -4.37306583e-01
-1.20222819e+00 -1.25640452e+00 1.73358008e-01 -7.23214149e-01
3.71681154e-01 -5.33880770e-01 -8.49091172e-01 6.97023571e-01
4.01561350e-01 4.73536193e-01 1.58338100e-01 -4.82202917e-01
2.26321176e-01 8.42217058e-02 -8.80156875e-01 4.84338343e-01
1.01505303e+00 -3.02974284e-01 -4.07917798e-01 4.43067461e-01
8.50630581e-01 -4.89284217e-01 -4.42637771e-01 4.16849852e-01
9.83181223e-02 -9.49480236e-01 1.00457597e+00 -5.20484328e-01
3.15318108e-01 -7.66553283e-01 7.02723712e-02 -7.78772235e-01
4.46142524e-01 -7.94408977e-01 1.87839530e-02 9.72936749e-01
3.37267935e-01 -1.31840721e-01 1.16282368e+00 6.28950775e-01
-2.05457643e-01 -4.95207369e-01 -6.38336778e-01 -4.61426169e-01
-8.56394395e-02 -6.93272114e-01 8.77589136e-02 9.08446014e-01
-6.64265633e-01 4.64827754e-02 6.27538562e-02 2.99433589e-01
8.20788562e-01 2.38632992e-01 8.75353634e-01 -1.01706207e+00
-2.62371004e-01 -4.71872509e-01 -4.52777624e-01 -1.24762797e+00
3.42271924e-01 -7.27635026e-01 6.21122360e-01 -1.73749280e+00
-1.08360365e-01 -3.50417376e-01 -1.84507817e-01 5.52540839e-01
-2.79865623e-01 8.10397923e-01 2.22632855e-01 6.42345622e-02
-9.49551225e-01 3.32761586e-01 1.18055129e+00 -2.44947433e-01
-4.23084557e-01 -6.01569712e-02 -2.32746035e-01 9.67672050e-01
6.92573667e-01 -2.69545764e-01 -5.25050640e-01 -4.40994442e-01
-1.57930151e-01 -2.25161031e-01 7.74989247e-01 -1.04828429e+00
1.73893750e-01 -2.92095542e-01 4.29578871e-01 -4.97655302e-01
2.14996621e-01 -8.07242215e-01 -2.93491900e-01 3.49202722e-01
-1.98422879e-01 -4.01115984e-01 1.89416990e-01 4.95504469e-01
1.24745478e-03 -4.10073847e-01 7.64145017e-01 -3.41469556e-01
-1.11047113e+00 2.73114562e-01 -4.58265692e-01 1.31799862e-01
1.05114019e+00 -4.20249939e-01 1.01706192e-01 -2.61815697e-01
-8.90960693e-01 1.39315724e-01 6.82076931e-01 2.97183663e-01
5.47341704e-01 -7.84652412e-01 -3.16730112e-01 8.25147852e-02
-3.77471447e-01 6.77883625e-01 -1.11719333e-01 9.34492469e-01
-1.03292537e+00 2.54043669e-01 -1.59229323e-01 -1.08574045e+00
-8.58002067e-01 6.01690233e-01 4.53252405e-01 1.89637959e-01
-5.41801631e-01 1.09623015e+00 4.71439838e-01 -1.11818351e-01
1.39741823e-01 -8.96327496e-01 -4.04757746e-02 4.52180542e-02
2.69115150e-01 -5.62275760e-02 -1.36584550e-01 -7.84011781e-01
-2.85774976e-01 4.99190241e-01 9.80681628e-02 -3.56009305e-01
1.16721308e+00 -2.00427011e-01 -1.06219485e-01 5.61798215e-01
1.15638185e+00 -2.95279026e-01 -2.00162268e+00 4.17139009e-02
3.59669060e-01 -3.78042281e-01 2.77451456e-01 -4.07077610e-01
-1.37847888e+00 1.03742957e+00 3.91245306e-01 1.09137572e-01
1.01832199e+00 1.34130819e-02 7.59405434e-01 3.80730569e-01
3.32674891e-01 -1.10943627e+00 2.68884629e-01 4.20718968e-01
4.48170662e-01 -1.40630496e+00 -1.31201282e-01 -5.85807085e-01
-4.50331837e-01 1.04826939e+00 5.82173109e-01 -4.26586151e-01
3.69212866e-01 2.09422156e-01 3.46656024e-01 1.11822844e-01
-2.06623882e-01 -7.79814363e-01 4.13659722e-01 5.25397778e-01
3.16579759e-01 -3.55584443e-01 -5.61785586e-02 1.52846396e-01
-1.15419462e-01 -8.92562568e-02 5.66220760e-01 1.00683105e+00
-4.58294958e-01 -1.02017999e+00 -3.10776323e-01 1.81430787e-01
-5.57826698e-01 -1.22201862e-02 -4.45914418e-01 6.83957160e-01
2.79496998e-01 8.40614796e-01 3.36259544e-01 3.72655019e-02
-1.52980804e-01 2.70195398e-02 7.48350561e-01 -7.57317245e-01
-2.99125701e-01 3.34089845e-01 -3.43537062e-01 -6.32067859e-01
-1.02138650e+00 -6.68995678e-01 -1.70356679e+00 2.75922686e-01
-2.20550269e-01 -1.87297743e-02 5.72847247e-01 1.31684899e+00
-1.01965606e-01 7.11768389e-01 2.27314994e-01 -1.59351277e+00
3.47023338e-01 -7.09279180e-01 -1.10302016e-01 4.48601902e-01
6.21207118e-01 -5.53926766e-01 -1.07171677e-01 8.92806232e-01] | [9.114644050598145, -0.20419543981552124] |
0475017a-6adf-4f8c-bf07-afe3ce4e3e82 | liplearner-customizable-silent-speech | 2302.05907 | null | https://arxiv.org/abs/2302.05907v3 | https://arxiv.org/pdf/2302.05907v3.pdf | LipLearner: Customizable Silent Speech Interactions on Mobile Devices | Silent speech interface is a promising technology that enables private communications in natural language. However, previous approaches only support a small and inflexible vocabulary, which leads to limited expressiveness. We leverage contrastive learning to learn efficient lipreading representations, enabling few-shot command customization with minimal user effort. Our model exhibits high robustness to different lighting, posture, and gesture conditions on an in-the-wild dataset. For 25-command classification, an F1-score of 0.8947 is achievable only using one shot, and its performance can be further boosted by adaptively learning from more data. This generalizability allowed us to develop a mobile silent speech interface empowered with on-device fine-tuning and visual keyword spotting. A user study demonstrated that with LipLearner, users could define their own commands with high reliability guaranteed by an online incremental learning scheme. Subjective feedback indicated that our system provides essential functionalities for customizable silent speech interactions with high usability and learnability. | ['Jun Rekimoto', 'Shitao Fang', 'Zixiong Su'] | 2023-02-12 | null | null | null | null | ['lipreading', 'visual-keyword-spotting', 'keyword-spotting'] | ['computer-vision', 'computer-vision', 'speech'] | [-9.70448852e-02 5.26634157e-02 -5.64999580e-01 -2.51559794e-01
-1.08188331e+00 -7.01077104e-01 2.40776613e-01 -4.25830156e-01
-3.70150238e-01 4.98483330e-01 2.93063402e-01 -5.08669436e-01
1.00451246e-01 -1.27408013e-01 -5.39253891e-01 -3.92997772e-01
1.71871036e-01 -6.41515553e-02 1.81346565e-01 -1.35636538e-01
4.86042909e-02 2.61721164e-01 -1.93436790e+00 3.84420812e-01
8.57653737e-01 9.85643983e-01 8.62019062e-01 1.02357769e+00
-3.66809428e-01 2.53536373e-01 -7.71018267e-01 -1.05335124e-01
-6.45893216e-02 1.43795088e-01 -3.70023668e-01 -2.42103428e-01
2.51159519e-01 -7.26121366e-01 -2.90317237e-01 5.01951933e-01
1.20342922e+00 5.84725803e-03 1.16311856e-01 -1.18019533e+00
-4.58180159e-01 5.61359107e-01 1.24004260e-01 -2.67533332e-01
9.51314747e-01 7.14197814e-01 8.96222472e-01 -9.92577493e-01
1.84231535e-01 7.57592916e-01 6.21322155e-01 9.67683256e-01
-1.22018981e+00 -9.17698324e-01 2.93324053e-01 2.18916945e-02
-1.53380501e+00 -1.17880571e+00 5.19562364e-01 -1.65040612e-01
1.23191345e+00 5.69996715e-01 6.54226959e-01 1.51731038e+00
-3.66199642e-01 8.88580501e-01 7.94604421e-01 -4.07085866e-01
3.56123179e-01 6.37127578e-01 -2.30396658e-01 6.37884259e-01
-3.02470177e-01 -1.26296654e-01 -1.13644469e+00 2.52578296e-02
5.47111809e-01 9.29314494e-02 -5.62081099e-01 -2.69257158e-01
-1.06151223e+00 3.28354686e-01 -1.06158674e-01 1.22475229e-01
-7.09962100e-02 -1.32542446e-01 4.10508543e-01 4.36687618e-01
1.13136228e-02 3.38429630e-01 -7.25829363e-01 -1.00579357e+00
-7.78204978e-01 -2.13067383e-01 1.07385182e+00 1.30672634e+00
3.51189077e-01 7.33261704e-02 -1.78236455e-01 1.13215947e+00
1.19490385e-01 9.88649130e-01 7.16077089e-01 -8.46517444e-01
5.11683881e-01 1.88624129e-01 9.09759030e-02 -1.94737539e-01
-3.35448503e-01 1.05364239e-02 -3.33995074e-01 1.82392240e-01
2.14130655e-01 -3.63844484e-01 -7.46584713e-01 1.80779588e+00
1.00728143e-02 -8.07149857e-02 -1.20848976e-01 5.64291239e-01
8.56239796e-01 3.84768158e-01 1.21971004e-01 -4.79110301e-01
1.22984517e+00 -7.02907145e-01 -9.75779831e-01 1.13355801e-01
4.86909658e-01 -6.34477794e-01 2.32742023e+00 5.72019279e-01
-9.33784604e-01 -5.35262167e-01 -9.78916466e-01 1.12588242e-01
-2.97335207e-01 2.14344963e-01 7.35947549e-01 1.35575044e+00
-1.03627193e+00 1.45219177e-01 -6.65013015e-01 -4.32343602e-01
3.23560238e-01 7.97918260e-01 -3.23873699e-01 3.79960179e-01
-8.06273341e-01 4.09952164e-01 -1.43510118e-01 -5.20079792e-01
-6.30017817e-01 -8.74668658e-01 -5.56030393e-01 2.52323121e-01
6.35106742e-01 -5.41486502e-01 1.60146415e+00 -5.35250545e-01
-2.60824132e+00 5.56777060e-01 -3.40559304e-01 -1.11595288e-01
3.71136159e-01 -3.68327677e-01 -5.28688550e-01 1.16122037e-01
-2.96966732e-01 5.73568046e-01 1.10186028e+00 -1.18339431e+00
-5.75850546e-01 -1.81860551e-01 2.31718630e-01 4.72156614e-01
-1.01464272e+00 6.22816086e-02 -7.20595539e-01 -4.47922140e-01
-3.98332208e-01 -8.06861043e-01 5.68175077e-01 2.13543952e-01
-4.14948761e-02 -1.37014970e-01 8.60120833e-01 -5.10182321e-01
1.45240617e+00 -2.25786376e+00 -1.94941267e-01 -5.64481653e-02
6.79684803e-02 5.04183650e-01 7.12375564e-04 1.98372647e-01
3.87449175e-01 2.96448112e-01 4.05534416e-01 -5.27881384e-01
2.72081643e-01 -2.84938589e-02 -1.97495803e-01 6.60339221e-02
-4.47389543e-01 9.16486442e-01 -6.90315127e-01 -3.54636192e-01
4.45355296e-01 5.44026732e-01 -8.11904192e-01 6.77561164e-01
-5.46657033e-02 4.51109745e-02 -1.49349705e-03 9.15035665e-01
2.72100657e-01 -6.94020465e-02 1.17399395e-01 -4.85484488e-02
-2.11399019e-01 5.07136345e-01 -1.03340578e+00 1.96754396e+00
-1.33607864e+00 6.66772604e-01 4.79985148e-01 -3.21642160e-01
7.41829872e-01 3.98924798e-01 4.41593319e-01 -7.69832909e-01
-4.24488820e-02 3.03622354e-02 -3.49683374e-01 -1.08660424e+00
3.30418736e-01 4.00367767e-01 2.87414342e-02 5.39777756e-01
5.02013192e-02 -2.09953129e-01 -5.18578291e-01 -7.10826591e-02
7.49807715e-01 9.43786874e-02 1.17137164e-01 6.56994507e-02
1.92334235e-01 -8.47631454e-01 -1.44654870e-01 7.81598568e-01
-2.23652288e-01 4.35196459e-01 -5.06149456e-02 4.02947441e-02
-5.47595322e-01 -1.20972586e+00 1.38403580e-01 2.01464415e+00
9.61458385e-02 -6.71949506e-01 -8.91363084e-01 -5.47858596e-01
-1.99903205e-01 6.92655802e-01 1.86466482e-02 -7.63892755e-02
-2.48030931e-01 -2.64290906e-02 7.23270416e-01 3.81747693e-01
3.83198202e-01 -9.52660322e-01 -6.57283008e-01 -1.23525724e-01
-2.42306039e-01 -1.11784089e+00 -9.09281611e-01 5.01050726e-02
-4.67289120e-01 -5.13940573e-01 -9.01723027e-01 -1.01363838e+00
2.39869773e-01 4.13839072e-01 5.27362406e-01 -2.34260038e-01
-8.15306976e-02 9.21373248e-01 -3.14026028e-01 -3.20111811e-01
-3.90508324e-01 4.49544162e-01 6.98188066e-01 -1.21863879e-01
1.35124594e-01 -7.08215296e-01 -6.57714248e-01 4.26385731e-01
-3.29551876e-01 3.67057393e-04 5.54112315e-01 9.17765319e-01
1.19619191e-01 -4.80009139e-01 6.64136171e-01 -2.31896594e-01
1.02553928e+00 -5.69779463e-02 -1.35202572e-01 4.14606422e-01
-8.14992011e-01 -6.67269975e-02 5.99776208e-01 -1.04939926e+00
-1.16609609e+00 4.92536128e-02 -5.28605402e-01 -1.76473275e-01
-2.93939263e-01 -7.64920786e-02 -6.42905772e-01 -1.60823494e-01
6.05534971e-01 4.87303793e-01 2.75944799e-01 -4.61486548e-01
5.36193371e-01 1.75099254e+00 4.52620655e-01 -5.83021760e-01
3.13390672e-01 2.75661387e-02 -8.44107866e-01 -1.50781369e+00
3.02243372e-03 -5.12794614e-01 -3.13547134e-01 -2.83466041e-01
3.72417033e-01 -1.01055372e+00 -1.58401561e+00 5.60348809e-01
-7.09452212e-01 -7.84754515e-01 -1.25893027e-01 4.51926887e-01
-7.12896526e-01 1.77046537e-01 -1.59789726e-01 -1.20800877e+00
-5.15156627e-01 -1.24020374e+00 1.25103509e+00 3.63451868e-01
-5.60235858e-01 -4.63741928e-01 -3.86120737e-01 6.77492678e-01
7.35284567e-01 -7.19333053e-01 5.69163561e-01 -5.46596289e-01
-3.89757156e-01 -1.98029429e-01 1.23646244e-01 9.91835222e-02
5.61977267e-01 -2.36194596e-01 -1.43536174e+00 -4.15225089e-01
-3.08523089e-01 -6.91376746e-01 2.22490683e-01 6.17819950e-02
1.60475123e+00 -6.75754130e-01 -2.46430904e-01 8.84052992e-01
7.29500294e-01 1.57412529e-01 3.98396552e-01 -5.34610786e-02
5.72286606e-01 2.38931879e-01 3.35121244e-01 6.10172153e-01
3.66398364e-01 1.13144839e+00 4.05411646e-02 7.27908611e-02
-2.89690852e-01 -5.87144494e-01 6.00016236e-01 7.20293820e-01
-6.99083731e-02 -2.35273659e-01 -6.31780207e-01 2.32876763e-01
-1.33849192e+00 -8.84553075e-01 8.57443810e-01 2.43016315e+00
1.11265266e+00 2.63512194e-01 4.67531890e-01 2.49730097e-03
3.83968711e-01 -5.52229509e-02 -6.85992360e-01 -3.67466480e-01
2.15026438e-01 4.30893302e-01 3.50290895e-01 6.62599027e-01
-7.44977355e-01 1.10923707e+00 6.44489956e+00 8.47060621e-01
-1.56615961e+00 9.87304598e-02 2.04988360e-01 -5.36531627e-01
-3.66786808e-01 -6.11794710e-01 -8.66126895e-01 7.09814966e-01
1.07838702e+00 -1.04766183e-01 7.17405558e-01 8.83084476e-01
5.03494501e-01 2.10300237e-01 -8.35614204e-01 1.48446798e+00
1.48598537e-01 -1.27351213e+00 -1.86008707e-01 -1.44618480e-02
2.70910144e-01 5.25194108e-02 3.78903866e-01 5.88012815e-01
-2.62013614e-01 -9.92446899e-01 4.54469740e-01 3.66332620e-01
1.82607889e+00 -5.94079316e-01 1.16390832e-01 5.67636490e-01
-1.08305299e+00 -3.39622319e-01 2.55085200e-01 4.42937352e-02
-2.63962243e-02 -4.33293819e-01 -1.40075183e+00 -2.47175530e-01
8.19308102e-01 -7.13257119e-02 -2.10457236e-01 6.39137864e-01
7.12227449e-03 6.76379621e-01 -4.96446460e-01 -7.82806098e-01
-5.25727689e-01 4.46679801e-01 2.88846284e-01 1.33002687e+00
2.88379043e-01 3.26703303e-02 8.85304809e-02 3.31043661e-01
-1.03195027e-01 3.68422002e-01 -6.99761331e-01 -9.05845016e-02
9.40107048e-01 9.98844028e-01 -1.65036395e-01 4.59325798e-02
-4.10830319e-01 1.26995838e+00 7.16050193e-02 5.50897181e-01
-6.34186625e-01 -6.18362069e-01 9.85481381e-01 3.60401571e-01
2.32963175e-01 -4.81096417e-01 -3.98969769e-01 -1.08267498e+00
1.01858072e-01 -1.17235482e+00 -1.71278447e-01 -6.40652955e-01
-5.74615121e-01 4.19241339e-01 -1.05472073e-01 -1.11524189e+00
-5.40834308e-01 -6.45684004e-01 -3.60318899e-01 7.37227559e-01
-1.01737916e+00 -1.38071930e+00 -4.43736464e-01 6.57863736e-01
1.09990907e+00 -6.20189428e-01 1.37501907e+00 1.38176650e-01
-3.16275716e-01 1.55229843e+00 -6.09917566e-02 -2.51567692e-01
9.06244397e-01 -9.18136537e-01 1.70173556e-01 2.67321438e-01
1.19653761e-01 1.01032174e+00 5.27837396e-01 -2.24151537e-01
-1.72479427e+00 -3.69586200e-01 6.22900188e-01 -3.54572624e-01
4.15944129e-01 -9.51499581e-01 -6.08179331e-01 1.52274281e-01
1.30370423e-01 -1.72589496e-01 1.10473382e+00 2.79055715e-01
-3.77912104e-01 -3.94794136e-01 -1.10060859e+00 1.10417640e+00
1.40464532e+00 -1.07726204e+00 -2.11348191e-01 7.85821527e-02
1.04723835e+00 -3.65757316e-01 -6.93289995e-01 1.71079129e-01
1.24509370e+00 -7.97560096e-01 9.18653607e-01 -3.89599144e-01
-5.26026726e-01 3.70090939e-02 -2.24564806e-01 -9.80127215e-01
1.58528581e-01 -1.49958968e+00 -5.12934506e-01 1.27937412e+00
5.49746573e-01 -6.14379048e-01 8.28180313e-01 8.00398171e-01
-4.52651940e-02 -8.08102012e-01 -8.17936599e-01 -7.73533046e-01
-3.87457699e-01 -7.80375600e-01 8.94209027e-01 2.97773540e-01
6.69776022e-01 2.96378940e-01 -4.73339140e-01 8.51408914e-02
1.86419502e-01 -3.21401894e-01 1.28788507e+00 -8.63338172e-01
-5.53365290e-01 -4.84323978e-01 -1.27157509e-01 -1.77152896e+00
2.92697728e-01 -5.80482304e-01 1.70605257e-01 -8.98741543e-01
-6.98043853e-02 -4.51473296e-01 4.46067415e-02 5.64909637e-01
-1.14491787e-02 -8.99739489e-02 1.47336185e-01 7.09558427e-02
-5.36612332e-01 5.43633044e-01 7.78139651e-01 -3.30510676e-01
-7.78784692e-01 5.52074909e-01 -6.92803323e-01 5.24968147e-01
9.31699812e-01 2.81095922e-01 -9.09419060e-01 -1.46110713e-01
-1.02809407e-01 -2.17141919e-02 -4.89605702e-02 -8.91926825e-01
4.45608020e-01 -2.09333763e-01 -1.01936283e-02 -3.01555935e-02
6.63104236e-01 -7.50017166e-01 -3.31489414e-01 7.69318640e-02
-4.85047579e-01 -3.87735367e-01 3.90965551e-01 3.83796036e-01
3.44481200e-01 6.77430779e-02 3.28375399e-01 2.64189631e-01
-8.30766082e-01 1.67554200e-01 -5.26933670e-01 -7.96219334e-02
8.15422475e-01 -5.12926161e-01 -5.99621609e-02 -8.78983796e-01
-7.26207793e-01 -1.07204914e-01 4.25931394e-01 6.82179034e-01
8.70550811e-01 -1.01559091e+00 6.20292090e-02 7.65923738e-01
3.62043828e-01 -2.97403842e-01 5.89907616e-02 4.44483757e-01
-1.91003814e-01 3.61540854e-01 -2.98539493e-02 -7.75810540e-01
-1.78941548e+00 3.29597980e-01 3.33249569e-01 4.85211283e-01
-6.20250642e-01 7.89382219e-01 -4.12715584e-01 -3.97753626e-01
1.10304534e+00 -3.13163817e-01 -9.91160143e-03 -1.22077614e-01
7.28780627e-01 1.74837396e-01 1.26802266e-01 -1.53227597e-01
-3.53553593e-01 4.42898154e-01 1.56367719e-01 -3.85196537e-01
9.85759497e-01 -4.80284214e-01 7.94814527e-01 6.17862046e-01
1.14099336e+00 4.97192949e-01 -1.44799066e+00 9.15802047e-02
-4.36058462e-01 -3.97020876e-01 -7.33614117e-02 -9.63273108e-01
-4.15847749e-01 7.11695015e-01 9.49050903e-01 1.04558049e-02
1.08633995e+00 1.99930947e-02 8.64357829e-01 8.53049040e-01
6.71831906e-01 -1.36109722e+00 2.72520214e-01 4.10732090e-01
8.26976120e-01 -1.41411483e+00 -5.06249189e-01 -1.92159995e-01
-8.90032530e-01 9.95419204e-01 5.45732617e-01 8.17859411e-01
6.72664165e-01 7.51703918e-01 2.98949420e-01 3.18366170e-01
-6.85803890e-01 -3.37991536e-01 1.94513977e-01 1.12837291e+00
4.42601651e-01 2.91938335e-01 2.17925832e-01 8.43958795e-01
-3.66308093e-01 1.20380841e-01 -2.05329433e-02 6.91050231e-01
-4.83861953e-01 -9.49201584e-01 -3.58069316e-02 2.90699899e-01
-2.99369127e-01 -9.64954719e-02 -1.85324892e-01 5.56032062e-01
-6.55698031e-02 1.21051383e+00 -1.00204393e-01 -8.23252797e-01
3.56033087e-01 3.83742124e-01 4.81194615e-01 -6.71071112e-01
-3.81729960e-01 2.81814095e-02 5.80395274e-02 -7.68489540e-01
1.32842094e-01 -4.53794867e-01 -1.14529598e+00 -1.44984767e-01
-3.54137331e-01 -1.77240223e-01 9.14396644e-01 7.01366127e-01
6.92906976e-01 3.42703134e-01 7.85672009e-01 -9.03396964e-01
-8.03364456e-01 -1.07048798e+00 -2.62563467e-01 1.34929180e-01
5.21268129e-01 -4.59479034e-01 -3.64363104e-01 1.09831847e-01] | [14.312551498413086, 6.182345867156982] |
fdafb9ac-fb9e-43ba-a2c1-10f6279b9868 | training-sound-event-detection-with-soft | 2302.14572 | null | https://arxiv.org/abs/2302.14572v1 | https://arxiv.org/pdf/2302.14572v1.pdf | Training sound event detection with soft labels from crowdsourced annotations | In this paper, we study the use of soft labels to train a system for sound event detection (SED). Soft labels can result from annotations which account for human uncertainty about categories, or emerge as a natural representation of multiple opinions in annotation. Converting annotations to hard labels results in unambiguous categories for training, at the cost of losing the details about the labels distribution. This work investigates how soft labels can be used, and what benefits they bring in training a SED system. The results show that the system is capable of learning information about the activity of the sounds which is reflected in the soft labels and is able to detect sounds that are missed in the typical binary target training setup. We also release a new dataset produced through crowdsourcing, containing temporally strong labels for sound events in real-life recordings, with both soft and hard labels. | ['Annamaria Mesaros', 'Paul Ahokas', 'Manu Harju', 'Irene Martín-Morató'] | 2023-02-28 | null | null | null | null | ['sound-event-detection'] | ['audio'] | [ 4.27496016e-01 4.03232902e-01 4.32737440e-01 -6.80057228e-01
-9.03101087e-01 -1.09784758e+00 6.96558297e-01 3.29546452e-01
-3.08710128e-01 6.59482002e-01 2.33151004e-01 1.79666162e-01
1.49961367e-01 -4.85608160e-01 -6.15298867e-01 -6.57427311e-01
5.41966856e-02 4.18945163e-01 8.10570598e-01 -1.03523821e-01
-1.64481491e-01 3.82759720e-02 -1.70904481e+00 8.74985754e-01
1.13821693e-01 1.15075493e+00 7.98045993e-02 7.79048741e-01
-2.40321934e-01 1.16883898e+00 -1.25010955e+00 -3.49122345e-01
4.69838157e-02 -3.11873764e-01 -9.04718161e-01 -9.19485241e-02
5.20764709e-01 1.97553694e-01 3.64108562e-01 9.51427639e-01
6.30861878e-01 -5.14568910e-02 6.45137966e-01 -1.31369340e+00
-1.60797134e-01 7.90086389e-01 1.77521095e-01 1.88944474e-01
7.89229631e-01 -1.68367937e-01 9.86158133e-01 -8.96265328e-01
4.03109699e-01 1.30160141e+00 9.78889167e-01 5.24562657e-01
-1.05657673e+00 -5.12825131e-01 -1.82910077e-02 8.53064284e-02
-1.26796591e+00 -6.69929206e-01 4.11516309e-01 -8.81183147e-01
6.64675713e-01 7.00708628e-01 1.72638103e-01 1.28540611e+00
-3.99945855e-01 5.42110026e-01 1.29025054e+00 -8.42230797e-01
7.48375237e-01 5.87650418e-01 2.03498095e-01 2.91345239e-01
-1.73754200e-01 1.38291731e-01 -8.33116889e-01 -5.22339404e-01
8.01578164e-02 -4.86521155e-01 -2.28263378e-01 3.25336069e-01
-9.77688551e-01 4.34679240e-01 9.58987027e-02 5.42946100e-01
-1.49333373e-01 6.12421706e-02 4.20062989e-01 3.16871792e-01
7.65116036e-01 2.48701200e-01 -5.91303647e-01 -1.57095462e-01
-8.52148652e-01 5.36362231e-02 1.07627463e+00 8.44263375e-01
5.02805352e-01 4.29080240e-02 -5.11656821e-01 9.97763395e-01
3.86960804e-01 4.91849542e-01 3.58517081e-01 -1.01779008e+00
-7.24422862e-04 2.56280154e-01 3.85620207e-01 -5.64572692e-01
-4.87663865e-01 2.44122762e-02 -8.34471881e-02 2.64439702e-01
5.55145442e-01 -2.46864885e-01 -8.31340611e-01 1.77363789e+00
2.76785553e-01 2.59726316e-01 -2.58647025e-01 6.79727316e-01
1.01594734e+00 4.94815171e-01 2.86047131e-01 -1.89110324e-01
1.42812335e+00 -4.90734667e-01 -9.56070423e-01 -4.14862275e-01
4.69974041e-01 -8.81731749e-01 1.11651504e+00 5.38735449e-01
-8.17257226e-01 -7.13680029e-01 -5.80942154e-01 2.40535438e-01
-6.08073831e-01 2.49637723e-01 2.23984942e-01 7.40660191e-01
-8.89992177e-01 4.73699808e-01 -6.03901684e-01 -2.84558028e-01
5.18848933e-02 1.44636884e-01 -2.13140771e-02 3.56859326e-01
-1.31506026e+00 6.96764588e-01 2.89599776e-01 4.17984910e-02
-9.85299528e-01 -2.44875357e-01 -7.08101511e-01 -1.24090806e-01
5.62830627e-01 2.36465678e-01 1.86596429e+00 -9.86870468e-01
-1.29852653e+00 1.00561333e+00 -8.53846446e-02 -1.79153308e-01
5.97743928e-01 -5.09430245e-02 -6.47080183e-01 2.17194870e-01
3.03237110e-01 5.41165233e-01 8.78228009e-01 -1.40050316e+00
-7.63891816e-01 1.13259807e-01 3.90443578e-02 -3.91538799e-01
-7.79732838e-02 4.60514754e-01 9.71217304e-02 -7.55678773e-01
-1.09673604e-01 -1.09584785e+00 1.72456205e-01 -1.29703224e-01
-5.43143809e-01 -3.96040201e-01 6.07066631e-01 -5.79688370e-01
9.76718366e-01 -2.25707889e+00 -4.45813596e-01 9.92797390e-02
-8.97115022e-02 3.28771919e-01 1.91063926e-01 3.68322790e-01
2.57897004e-02 2.24072590e-01 -3.38232219e-01 -5.41871548e-01
3.94246578e-01 6.35959268e-01 -8.15395355e-01 1.86510589e-02
2.22952127e-01 5.09973228e-01 -1.05791759e+00 -4.77517039e-01
-1.83321536e-02 2.14240029e-01 1.15821883e-02 4.22582597e-01
-4.42214072e-01 5.00157118e-01 -5.20658679e-02 4.49927419e-01
3.61648500e-01 1.27877697e-01 -4.92711477e-02 1.24267919e-03
-1.28603742e-01 4.20504928e-01 -1.42477787e+00 9.23274934e-01
-3.38368386e-01 8.26882005e-01 1.61434665e-01 -5.81955254e-01
7.36035824e-01 7.49171376e-01 7.73996040e-02 -4.09586251e-01
1.55931056e-01 2.96411872e-01 -5.40143490e-01 -7.34595597e-01
3.81709039e-01 -2.73570985e-01 -2.98050523e-01 5.81420839e-01
2.54423290e-01 -4.01800305e-01 1.59697533e-01 -2.02293009e-01
1.03620005e+00 9.23151523e-02 2.56483126e-02 -6.78608045e-02
2.53444731e-01 -8.12407583e-02 6.35167301e-01 1.05265367e+00
-8.26563090e-02 7.22691715e-01 4.04601336e-01 -4.15489703e-01
-5.60889006e-01 -1.05006135e+00 -2.56530225e-01 1.54918599e+00
-1.16534352e-01 -3.64779532e-01 -6.73292637e-01 -8.46895099e-01
-4.09709036e-01 7.87081718e-01 -4.79764372e-01 -1.67935550e-01
-4.15068984e-01 -5.38028598e-01 9.22313094e-01 5.06196320e-01
1.28638139e-02 -1.33287311e+00 -6.91986203e-01 2.10108995e-01
-5.23428321e-01 -1.38988233e+00 -2.63765603e-01 7.32712984e-01
-2.67109066e-01 -9.18635547e-01 -3.26657236e-01 -7.48772621e-01
3.11447561e-01 -1.31226912e-01 1.29731727e+00 -1.41189501e-01
-2.41760835e-01 5.72597146e-01 -7.13650107e-01 -1.07561243e+00
-8.46127272e-01 -4.52393115e-01 7.88993537e-02 1.37473643e-01
4.88600641e-01 -3.51177305e-01 -6.06646687e-02 6.71732426e-01
-9.72508907e-01 -5.54453552e-01 -3.62818241e-02 5.69508016e-01
3.58185261e-01 1.23340316e-01 8.52892458e-01 -8.83364260e-01
6.76899314e-01 -4.48481083e-01 -3.66942316e-01 3.20476562e-01
-2.41137326e-01 -1.61700398e-02 4.61605102e-01 -7.76510179e-01
-1.07343221e+00 2.63914913e-01 -2.09747627e-01 -8.77349079e-02
-6.77309036e-01 7.50792818e-03 1.92349941e-01 1.25533866e-03
1.03812337e+00 -3.14043164e-01 -5.69378078e-01 -5.84830523e-01
1.14259236e-01 1.08611536e+00 5.96439183e-01 -5.84829092e-01
3.22692275e-01 5.44664204e-01 -3.84675026e-01 -7.76311100e-01
-1.43183637e+00 -6.69575512e-01 -6.03265643e-01 -4.33611631e-01
9.32041943e-01 -9.03710723e-01 -3.72119218e-01 3.85959655e-01
-1.08904397e+00 -4.81821269e-01 -8.51297617e-01 4.80311960e-01
-4.67924118e-01 2.03644320e-01 -4.52875584e-01 -1.18638289e+00
2.46271715e-01 -6.86837435e-01 1.49875414e+00 4.22158428e-02
-6.34630799e-01 -9.38574135e-01 3.68836403e-01 1.11091837e-01
2.19683364e-01 1.76967755e-01 4.65295553e-01 -1.08313239e+00
-5.38879726e-03 -3.71429354e-01 1.77031606e-01 6.61567986e-01
2.48766653e-02 -4.61494587e-02 -2.01868010e+00 1.07678227e-01
1.98676720e-01 -5.24479926e-01 7.77698219e-01 1.63783401e-01
7.83373415e-01 -3.93426031e-01 -1.38841569e-01 -3.27033609e-01
8.03792715e-01 1.25188664e-01 2.39805520e-01 -6.41346276e-02
3.67036730e-01 7.16868997e-01 5.67729712e-01 5.43595791e-01
1.95865661e-01 8.39204490e-01 4.10857260e-01 -2.14632973e-01
-1.83635876e-01 -1.43619627e-01 5.34121573e-01 6.07352257e-01
1.42963409e-01 -4.91804779e-01 -8.50802243e-01 5.19850492e-01
-1.73674285e+00 -1.01169837e+00 -3.46813887e-01 2.13327360e+00
1.12235129e+00 2.83359647e-01 2.71338254e-01 5.20227015e-01
9.80269551e-01 -3.20752859e-02 2.44352758e-01 -5.06104589e-01
-9.08426568e-02 2.74044961e-01 1.73412740e-01 6.26878619e-01
-1.35423255e+00 5.55261850e-01 7.41039467e+00 6.08146369e-01
-8.90600085e-01 4.75847274e-01 9.96604860e-02 1.56233436e-03
-5.85238002e-02 -4.93833609e-02 -6.84729874e-01 6.31562948e-01
1.04929185e+00 4.39726830e-01 8.05174485e-02 5.79574883e-01
1.74415618e-01 -4.05101359e-01 -1.33993745e+00 7.17404544e-01
7.94679746e-02 -7.70900905e-01 -5.29903233e-01 -1.90954894e-01
6.47222042e-01 -3.10738292e-02 -3.36788535e-01 3.59717220e-01
4.69291925e-01 -7.48436809e-01 1.57062387e+00 6.39334917e-01
5.64805925e-01 -1.60091251e-01 1.07633221e+00 6.39103711e-01
-9.55390155e-01 -1.56482473e-01 -2.52550304e-01 -3.69077265e-01
2.73359507e-01 8.58380318e-01 -1.27211475e+00 6.65831044e-02
8.62600923e-01 1.76850855e-01 -7.72496581e-01 1.14164805e+00
-5.69752455e-01 1.18217778e+00 -7.36706614e-01 -5.32143656e-03
-2.81856325e-03 4.56871361e-01 5.71871340e-01 1.67002785e+00
3.16217989e-01 -9.18763652e-02 5.62690258e-01 7.27475584e-01
8.63401890e-02 -1.61620855e-01 -5.79740405e-01 1.74056709e-01
4.81861681e-01 1.14479005e+00 -7.93125033e-01 -3.54771197e-01
-3.01987696e-02 6.61707044e-01 -1.29334778e-01 1.28748745e-01
-7.56323755e-01 -1.92026019e-01 2.15693880e-02 6.90872408e-03
4.80417795e-02 2.27362499e-01 -3.74201462e-02 -7.34607697e-01
2.79623687e-01 -5.22332847e-01 4.78951663e-01 -1.15372908e+00
-1.37641120e+00 8.39592040e-01 1.61135629e-01 -1.34178853e+00
-4.20050591e-01 -5.41545868e-01 -7.36003697e-01 6.07859969e-01
-1.15108168e+00 -8.18196654e-01 -3.88542682e-01 2.29938030e-01
4.85875070e-01 1.94568843e-01 9.75208819e-01 3.87864918e-01
-1.42450735e-01 2.68803060e-01 -3.47603142e-01 3.55406217e-02
1.12627590e+00 -1.64134336e+00 7.77872950e-02 5.48966587e-01
7.33696163e-01 -1.66068859e-02 9.36429024e-01 -4.72164035e-01
-1.15177564e-01 -1.03838086e+00 1.13292658e+00 -9.19415712e-01
5.20483613e-01 -5.44363797e-01 -8.81182969e-01 4.54139292e-01
-5.52304694e-03 2.21495450e-01 8.89016390e-01 -1.07566677e-01
-4.69236374e-01 2.15257660e-01 -1.27788150e+00 -9.29116830e-02
7.64674425e-01 -9.99419093e-01 -9.04381931e-01 7.42924333e-01
6.28675222e-01 -2.83672571e-01 -5.45543671e-01 2.84383714e-01
2.97335774e-01 -8.53611350e-01 7.82482624e-01 -3.05469155e-01
-1.04816258e-01 -5.07719219e-01 -2.71969497e-01 -1.34837699e+00
-8.25225934e-02 -4.91990626e-01 1.07993402e-01 1.60435152e+00
4.67089146e-01 -3.09436679e-01 1.05612375e-01 5.12515843e-01
-1.89203098e-01 4.54012081e-02 -1.08150613e+00 -9.00177836e-01
-3.50978076e-01 -8.55180204e-01 4.03723001e-01 1.02462566e+00
-8.45399033e-03 2.17585996e-01 -3.01326811e-01 4.77926224e-01
2.75205523e-01 -2.67763138e-01 2.19823956e-01 -1.64748263e+00
-3.36589485e-01 2.34624855e-02 -4.35791552e-01 -4.20009196e-01
2.03219399e-01 -7.79327512e-01 7.96394944e-01 -1.27855277e+00
-3.08899254e-01 -5.42029500e-01 -1.30907640e-01 1.00070822e+00
5.61306067e-02 7.64635146e-01 1.02573484e-01 1.84493795e-01
-7.58718371e-01 -1.10631268e-02 6.17050052e-01 -9.74229202e-02
-1.63123891e-01 3.58358234e-01 -3.79720867e-01 1.27809799e+00
6.83806181e-01 -1.02658033e+00 -6.21311255e-02 -2.38031417e-01
6.28046155e-01 -1.59441069e-01 6.52226031e-01 -1.18965852e+00
8.77000391e-02 3.12470142e-02 8.05401355e-02 -3.92569274e-01
3.17662477e-01 -8.48294497e-01 9.03403461e-02 6.04661694e-03
-7.30558753e-01 -4.68684167e-01 2.00289145e-01 5.43070078e-01
-4.61214781e-01 -8.50676060e-01 7.19024360e-01 -1.84680447e-01
-6.13198280e-01 -4.56153274e-01 -5.76702833e-01 3.82279426e-01
7.92806387e-01 1.64673179e-02 -2.73211747e-01 -3.64381164e-01
-1.32663333e+00 1.69077574e-03 7.29954755e-03 4.79389399e-01
-5.02376780e-02 -1.25012970e+00 -6.61884665e-01 2.55446341e-02
2.30007380e-01 -1.59042791e-01 -1.58741787e-01 5.34352839e-01
-1.04252964e-01 -4.29531001e-02 1.98099062e-01 -8.44728649e-01
-1.35473728e+00 2.79306144e-01 4.16810244e-01 1.63411602e-01
-2.81036109e-01 1.00099194e+00 1.23158405e-02 -6.34207487e-01
6.73566580e-01 -5.66846192e-01 -3.72970641e-01 5.13419390e-01
6.98134303e-01 4.53187019e-01 3.19781840e-01 -6.26302838e-01
-5.25363028e-01 5.35054147e-01 5.86684883e-01 -3.63307953e-01
1.09231424e+00 -6.57493547e-02 -6.36085197e-02 1.07530153e+00
6.90901339e-01 5.28389037e-01 -9.89615381e-01 -2.82286018e-01
3.16166013e-01 -1.70017391e-01 -1.07813567e-01 -1.26631820e+00
-5.32289743e-01 9.70738173e-01 9.02610660e-01 1.05412376e+00
8.67501915e-01 4.93539721e-01 3.69844645e-01 4.60908532e-01
4.06265974e-01 -1.29655337e+00 9.69627425e-02 4.97516245e-01
8.48040164e-01 -1.24305665e+00 -4.13621813e-01 -5.41230500e-01
-7.86989093e-01 1.13195670e+00 3.06519151e-01 1.82703391e-01
6.01983905e-01 6.32591486e-01 3.86230409e-01 -2.54947066e-01
-5.50093949e-01 -5.18485367e-01 3.45347881e-01 7.65248656e-01
1.78775176e-01 2.72881985e-01 1.88938543e-01 8.27053010e-01
-1.64921895e-01 1.98001415e-03 6.93596721e-01 1.04440451e+00
-7.44861960e-01 -1.02264977e+00 -6.30241811e-01 1.44833341e-01
-6.59830868e-01 2.05732629e-01 -8.93813074e-01 -6.06570720e-05
8.31276536e-01 1.53966665e+00 -1.39838740e-01 -3.31045270e-01
6.10700011e-01 5.83766043e-01 1.65647447e-01 -9.43105280e-01
-8.21805239e-01 3.42220813e-02 5.75522363e-01 -3.31181675e-01
-6.12793922e-01 -4.77661580e-01 -1.24015260e+00 6.30661130e-01
-5.97292721e-01 3.47479790e-01 5.49354136e-01 1.12062907e+00
-5.98021187e-02 6.39373004e-01 6.47590339e-01 -9.29621398e-01
-5.54396391e-01 -1.08806980e+00 -4.69843566e-01 4.59967315e-01
4.70257640e-01 -6.82667494e-01 -8.97775710e-01 5.98532796e-01] | [15.157950401306152, 5.149257659912109] |
6c586694-1dcd-4101-b5e3-ce06cd07827a | is-ggt-iterative-scene-graph-generation-with | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Kundu_IS-GGT_Iterative_Scene_Graph_Generation_With_Generative_Transformers_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Kundu_IS-GGT_Iterative_Scene_Graph_Generation_With_Generative_Transformers_CVPR_2023_paper.pdf | IS-GGT: Iterative Scene Graph Generation With Generative Transformers | Scene graphs provide a rich, structured representation of a scene by encoding the entities (objects) and their spatial relationships in a graphical format. This representation has proven useful in several tasks, such as question answering, captioning, and even object detection, to name a few. Current approaches take a generation-by-classification approach where the scene graph is generated through labeling of all possible edges between objects in a scene, which adds computational overhead to the approach. This work introduces a generative transformer-based approach to generating scene graphs beyond link prediction. Using two transformer-based components, we first sample a possible scene graph structure from detected objects and their visual features. We then perform predicate classification on the sampled edges to generate the final scene graph. This approach allows us to efficiently generate scene graphs from images with minimal inference overhead. Extensive experiments on the Visual Genome dataset demonstrate the efficiency of the proposed approach. Without bells and whistles, we obtain, on average, 20.7% mean recall (mR@100) across different settings for scene graph generation (SGG), outperforming state-of-the-art SGG approaches while offering competitive performance to unbiased SGG approaches. | ['Sathyanarayanan N. Aakur', 'Sanjoy Kundu'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['scene-graph-generation'] | ['computer-vision'] | [ 6.17430389e-01 4.16257262e-01 7.58473203e-02 -4.63864118e-01
-6.91477180e-01 -6.50392532e-01 9.05740499e-01 5.88488340e-01
1.41470125e-02 4.84353632e-01 1.40320882e-01 -3.71919513e-01
2.68511958e-02 -1.11797690e+00 -9.93949175e-01 -4.05639142e-01
-1.80345789e-01 6.01151466e-01 6.31337702e-01 1.77129999e-01
2.18546465e-01 3.95723879e-01 -1.77161407e+00 5.11145711e-01
8.57330084e-01 9.05148387e-01 5.27016699e-01 8.08246732e-01
-5.14388859e-01 1.13155651e+00 -6.15016997e-01 -6.70139194e-01
-5.01622148e-02 -4.29809242e-01 -8.55429828e-01 5.89471281e-01
7.36403525e-01 -1.59079835e-01 -3.12219650e-01 8.79874468e-01
7.19045997e-02 1.60518467e-01 5.53015411e-01 -1.34292436e+00
-5.62620759e-01 6.28457427e-01 -5.04302740e-01 -7.86460787e-02
7.22133338e-01 -1.09162465e-01 1.37659359e+00 -1.05787361e+00
9.23012614e-01 1.54154229e+00 1.36997491e-01 2.16843411e-01
-1.33994544e+00 -2.24936739e-01 4.19434756e-01 2.63309240e-01
-1.35140479e+00 -1.80422395e-01 8.92973185e-01 -4.50768173e-01
7.64622748e-01 3.82796794e-01 7.60293067e-01 5.54418862e-01
-2.85203636e-01 9.92245197e-01 9.17820752e-01 -6.00480914e-01
3.22859317e-01 2.25324780e-01 3.68780196e-02 9.74765182e-01
3.80670518e-01 -5.47580004e-01 -5.00095308e-01 -2.50895292e-01
6.43969893e-01 -1.14843421e-01 -1.96988598e-01 -7.00873077e-01
-1.25968993e+00 7.39659250e-01 8.80016863e-01 -1.83686957e-01
-3.92245173e-01 2.30931684e-01 1.72310770e-02 -2.19135508e-01
4.03345197e-01 2.69728839e-01 1.87505528e-01 3.66924465e-01
-6.84368491e-01 4.95775878e-01 7.34692216e-01 1.31231475e+00
8.52965951e-01 -8.32829252e-02 -5.48136830e-01 7.52717972e-01
3.07998151e-01 3.84200454e-01 -1.60345614e-01 -6.51685238e-01
6.91784918e-01 9.09436464e-01 2.18317285e-02 -1.33124006e+00
-1.34407207e-01 -1.59607857e-01 -5.64440429e-01 -2.34580234e-01
3.04799736e-01 1.65551513e-01 -1.20094728e+00 1.48539710e+00
7.57863164e-01 2.06218511e-01 -1.44453451e-01 5.95454276e-01
1.01832807e+00 8.23778391e-01 2.11469665e-01 2.13659942e-01
1.56284606e+00 -1.03368974e+00 -4.25500333e-01 -5.31921566e-01
4.90175635e-01 -6.60901785e-01 1.10515058e+00 1.42320141e-01
-7.76587427e-01 -3.65638882e-01 -8.93716931e-01 -5.91321215e-02
-4.61858839e-01 1.77831352e-01 8.51776659e-01 4.79990184e-01
-1.00830925e+00 1.80676937e-01 -5.11749923e-01 -5.87494254e-01
6.44790947e-01 7.15037249e-03 -2.71699846e-01 -3.06521565e-01
-5.88948727e-01 4.35127109e-01 7.80501604e-01 -1.07468583e-01
-8.85710835e-01 -4.22387242e-01 -1.19688404e+00 1.99160259e-02
6.53268874e-01 -7.71807551e-01 1.13446629e+00 -4.26680684e-01
-8.33041072e-01 7.86056936e-01 -4.91576821e-01 -5.42255461e-01
2.23091900e-01 -1.29323244e-01 -2.10266978e-01 4.15099889e-01
2.74468064e-01 8.84647369e-01 6.36033714e-01 -1.58529401e+00
-6.67440832e-01 -9.66485366e-02 3.66806746e-01 4.60144162e-01
1.73639506e-01 -6.17771558e-02 -8.15175712e-01 -4.01529729e-01
3.97276282e-01 -8.94137621e-01 -3.37115377e-01 1.34945154e-01
-9.62731421e-01 -2.48251051e-01 9.17206466e-01 -4.71542478e-01
1.07725465e+00 -1.92762494e+00 -2.76570529e-01 3.50568146e-01
2.85077840e-01 3.67199369e-02 -1.47688702e-01 6.58333182e-01
3.38655114e-02 4.28144559e-02 -2.73580968e-01 -4.69081610e-01
-4.13738303e-02 6.97674677e-02 -4.76222903e-01 1.90858215e-01
3.54788482e-01 1.06948590e+00 -1.13030565e+00 -7.56860912e-01
5.51630914e-01 2.83594310e-01 -6.67090535e-01 3.24367374e-01
-7.86859810e-01 1.37405381e-01 -5.70224166e-01 6.19604528e-01
5.31883240e-01 -6.50314629e-01 3.61252993e-01 -2.44117931e-01
3.42710435e-01 4.67026979e-01 -1.26542914e+00 1.45062923e+00
-3.05636674e-01 7.50873148e-01 -4.94005561e-01 -8.89738321e-01
1.00670159e+00 -1.21851675e-01 2.43658628e-02 -5.22016525e-01
-1.59701675e-01 -1.73271284e-01 -2.04035670e-01 -2.15066031e-01
6.86147749e-01 2.25165442e-01 -2.29757890e-01 2.91613221e-01
2.31605582e-02 -2.27833852e-01 4.89706248e-01 8.68398309e-01
9.98266041e-01 2.20812231e-01 4.89355505e-01 1.37486150e-02
2.54133940e-01 2.80686885e-01 2.43132794e-03 8.87963474e-01
4.54503149e-01 6.96865797e-01 6.70150757e-01 -2.21522734e-01
-9.00060177e-01 -1.19791615e+00 3.20389748e-01 7.69413650e-01
3.66580725e-01 -7.90663540e-01 -7.53973305e-01 -6.17632508e-01
-2.15009734e-01 9.60461736e-01 -5.68730235e-01 1.27255425e-01
-4.17732447e-01 -4.88388658e-01 1.82480440e-01 4.14934665e-01
4.62460399e-01 -1.18324161e+00 -7.02254653e-01 4.69945818e-02
-3.69243860e-01 -1.55539668e+00 -2.36800671e-01 -3.11038345e-01
-6.36049688e-01 -1.10850489e+00 -3.00928056e-01 -8.56665552e-01
1.07581544e+00 5.08602977e-01 1.35333514e+00 1.06681503e-01
-5.78568220e-01 4.60705072e-01 -3.55487257e-01 -2.85269678e-01
-3.01190674e-01 -2.21364498e-01 -5.16490221e-01 2.66327888e-01
-5.75977229e-02 -2.41833776e-01 -5.40724218e-01 6.48507103e-02
-8.12811613e-01 5.81135452e-01 4.84007984e-01 5.54248571e-01
9.54021275e-01 -7.22350436e-04 2.06841722e-01 -1.36399686e+00
4.12848949e-01 -4.15771365e-01 -7.51595199e-01 5.21191657e-01
-2.44990349e-01 1.57905638e-01 5.68874955e-01 -2.47539774e-01
-1.15934563e+00 2.10576966e-01 2.88796037e-01 -3.32880080e-01
-2.59861380e-01 4.37776744e-01 -3.20940465e-01 1.45630881e-01
6.00962877e-01 4.37080681e-01 -4.56936896e-01 -2.80841142e-01
8.46547246e-01 2.02815458e-01 6.96476340e-01 -5.51914096e-01
9.14754212e-01 5.54341078e-01 1.97624147e-01 -9.27895546e-01
-1.07093406e+00 -5.44649065e-01 -4.72787291e-01 -3.41472805e-01
8.38395238e-01 -8.77629757e-01 -4.77270573e-01 1.00224800e-01
-1.24000883e+00 -2.63645351e-01 -2.59744972e-01 1.12470537e-01
-4.21879917e-01 2.76590824e-01 -1.57786995e-01 -9.06947672e-01
-7.04099536e-02 -7.44817376e-01 1.53300416e+00 2.40949675e-01
-5.11809550e-02 -8.30952227e-01 -3.25477391e-01 3.37730438e-01
1.07439449e-02 6.55807912e-01 1.07988656e+00 -5.27598023e-01
-1.24521661e+00 -2.15007886e-01 -5.94194233e-01 -1.96757749e-01
1.68374881e-01 -1.15040056e-01 -8.94711435e-01 2.94190552e-02
-6.98503435e-01 -6.33006096e-02 6.45680964e-01 2.36013532e-01
1.42011404e+00 -5.56480825e-01 -6.54457390e-01 3.80524457e-01
1.61766613e+00 2.17892632e-01 6.83794498e-01 7.46334791e-02
1.10296559e+00 7.41301894e-01 7.62068093e-01 4.50108767e-01
6.70125663e-01 7.71761417e-01 5.93883157e-01 -2.24357009e-01
-4.13721323e-01 -8.60748649e-01 -1.06318884e-01 2.32384935e-01
3.63525212e-01 -6.75638020e-01 -8.95053387e-01 8.19005668e-01
-1.81967604e+00 -9.60319817e-01 -5.29626966e-01 2.10867763e+00
3.82804841e-01 8.20869505e-02 7.83623233e-02 3.07181291e-02
8.61449182e-01 2.87896454e-01 -3.26577216e-01 -2.48574749e-01
1.52723957e-02 9.00907442e-02 3.73106241e-01 4.76590782e-01
-1.10027277e+00 1.18469214e+00 5.89401484e+00 7.00137496e-01
-6.67326450e-01 -3.04284632e-01 7.10822284e-01 2.87001729e-01
-4.69773233e-01 3.32213402e-01 -8.56554508e-01 2.06962749e-01
4.58789080e-01 -3.70510072e-01 3.79011959e-01 1.05522561e+00
2.15676408e-02 -3.36774319e-01 -9.95116413e-01 1.05070353e+00
3.29678208e-01 -1.46207464e+00 5.69296062e-01 -1.03930116e-01
6.99409366e-01 -2.94289500e-01 -2.63255656e-01 -2.38703359e-02
5.29989660e-01 -8.49894285e-01 1.00188196e+00 2.72252530e-01
6.78733110e-01 -5.14373422e-01 2.76111454e-01 3.39057326e-01
-1.56404126e+00 2.25964785e-01 -3.26139122e-01 7.91320428e-02
4.15560305e-01 6.83422446e-01 -1.62443268e+00 7.73102045e-01
4.85485494e-01 3.97014946e-01 -9.71415520e-01 1.34751904e+00
-4.81235713e-01 6.60628021e-01 -3.28566462e-01 -2.97194242e-01
1.19767517e-01 -1.50391743e-01 4.89091933e-01 1.32559121e+00
3.39493960e-01 -4.77761403e-02 3.62347126e-01 1.00541794e+00
-1.50085539e-01 8.81911293e-02 -9.13677394e-01 -1.09451478e-02
7.09691048e-01 1.37918997e+00 -1.31176865e+00 -5.82181692e-01
-2.31294498e-01 1.04479396e+00 3.55481148e-01 3.28680307e-01
-8.81602705e-01 -4.74181294e-01 1.35150477e-01 3.75066489e-01
4.80301499e-01 -1.68575943e-01 -5.84799424e-02 -8.80256414e-01
2.29154333e-01 -6.24489367e-01 4.93611485e-01 -1.15656495e+00
-8.99802327e-01 6.73643231e-01 4.67685491e-01 -1.19498944e+00
-4.17394131e-01 -3.68069381e-01 -4.37269330e-01 6.11957490e-01
-1.17850792e+00 -1.31952488e+00 -7.72965133e-01 4.15426075e-01
5.89259148e-01 3.90609741e-01 5.91683805e-01 -2.21642219e-02
-2.27670491e-01 1.24862045e-01 -3.67887884e-01 2.24911049e-01
1.57522514e-01 -1.47818363e+00 8.80211532e-01 1.17120433e+00
7.87468016e-01 4.71338660e-01 6.61877990e-01 -7.14967251e-01
-1.23867559e+00 -1.54180980e+00 1.03822029e+00 -4.18558210e-01
4.26976353e-01 -8.61293495e-01 -7.18657732e-01 7.41968691e-01
9.37967971e-02 7.80659169e-02 2.60952264e-01 -7.96141848e-02
-4.64774549e-01 4.96956147e-02 -9.65610981e-01 9.81997848e-01
1.29672682e+00 -3.96924078e-01 -3.16965014e-01 5.53984284e-01
8.10507953e-01 -5.54431975e-01 -3.15019965e-01 2.07538515e-01
1.92470342e-01 -9.30262864e-01 1.04377615e+00 -4.15201217e-01
3.81518483e-01 -7.05422044e-01 -1.73916459e-01 -1.10772860e+00
-1.84609741e-01 -4.94614989e-01 -1.67987451e-01 1.28365970e+00
4.85180974e-01 -4.26039785e-01 7.92920530e-01 4.85293776e-01
-4.69391160e-02 -6.29765987e-01 -3.56674463e-01 -6.29013598e-01
-8.50210130e-01 -5.77143013e-01 6.66066229e-01 5.83684385e-01
-3.76973301e-01 6.41346455e-01 -2.37580210e-01 2.88020253e-01
8.56403232e-01 5.73248684e-01 1.05025017e+00 -1.14504755e+00
-3.80422056e-01 -1.09665297e-01 -6.13524854e-01 -1.28920484e+00
-1.45437330e-01 -9.52034056e-01 1.35993421e-01 -2.28410387e+00
4.04736370e-01 -4.53830749e-01 3.04963768e-01 5.12801945e-01
-4.16524231e-01 2.62128681e-01 3.71446162e-01 -1.41210347e-01
-1.01995480e+00 2.97154158e-01 1.21181417e+00 -1.02245584e-01
-1.01187654e-01 -3.01621705e-01 -7.98945427e-01 7.28553236e-01
5.86775124e-01 -3.43201518e-01 -7.97452390e-01 -3.62502217e-01
9.15474519e-02 3.81842740e-02 7.08276570e-01 -9.08951998e-01
1.55191764e-01 -6.70386702e-02 2.21019760e-01 -8.34347844e-01
5.47790408e-01 -5.51069617e-01 4.11151320e-01 1.90653399e-01
-1.89076900e-01 -1.40316063e-03 1.48810059e-01 7.99487591e-01
-2.28034362e-01 -5.00960015e-02 4.66216415e-01 -1.65282518e-01
-9.09813285e-01 2.34265238e-01 1.49373813e-02 3.83486077e-02
1.14996517e+00 -3.14122468e-01 -5.33209741e-01 -6.16828442e-01
-5.53963900e-01 8.55175480e-02 4.16235358e-01 3.92776936e-01
8.66033852e-01 -1.29773808e+00 -5.59496760e-01 -2.17855182e-02
4.65585887e-01 3.38077813e-01 1.10769086e-01 4.05059308e-01
-7.21129477e-01 4.64827299e-01 3.02478492e-01 -8.61176193e-01
-1.52010596e+00 5.37972689e-01 -1.01663671e-01 -2.45040804e-01
-8.16973448e-01 8.71121585e-01 7.39007950e-01 -1.22608826e-01
5.70255704e-02 -3.64457190e-01 -1.92659795e-01 -1.12937607e-01
3.61379892e-01 1.85927451e-01 -7.19114323e-04 -6.22747421e-01
-3.41306031e-01 3.98277879e-01 -1.18313869e-02 -1.09866343e-01
9.13947880e-01 4.29880545e-02 2.30547450e-02 3.36224645e-01
8.99765849e-01 7.69920722e-02 -9.91806388e-01 -1.77067772e-01
8.69890973e-02 -6.82031333e-01 -3.07160616e-01 -6.50775075e-01
-7.15596199e-01 7.07961082e-01 1.89564973e-01 5.75537980e-01
1.10598278e+00 5.86121619e-01 3.90456170e-01 3.47073078e-01
5.75522661e-01 -3.76907170e-01 2.61337787e-01 9.53661874e-02
8.29513967e-01 -1.00379229e+00 5.00929123e-03 -1.23006356e+00
-8.20381641e-01 6.29886627e-01 5.28686345e-01 -1.65525556e-01
2.88126737e-01 -4.73353751e-02 -3.29349488e-01 -4.44538951e-01
-8.77864838e-01 -5.39894640e-01 6.10927045e-01 6.47083938e-01
2.44138360e-01 3.02600235e-01 -4.73039038e-02 -2.81287637e-02
-3.00783157e-01 -4.12680387e-01 4.16246176e-01 7.76374340e-01
-4.81310338e-01 -1.00614786e+00 -2.93619633e-01 7.35092342e-01
-1.90861598e-01 -3.15041631e-01 -6.09359145e-01 8.19792867e-01
-1.29455939e-01 1.06116557e+00 1.98468223e-01 -1.62516221e-01
3.77005488e-01 -1.58918023e-01 4.87162530e-01 -9.28686917e-01
-3.09130494e-02 -2.92746536e-02 4.72498119e-01 -6.59851670e-01
-4.15210873e-01 -6.23737752e-01 -1.19222879e+00 1.88887656e-01
-3.65787685e-01 3.50161344e-02 5.64011574e-01 6.54000401e-01
5.38863480e-01 4.98163998e-01 5.36468387e-01 -6.84134305e-01
2.23640710e-01 -6.85688972e-01 -3.40916067e-01 6.59392774e-01
4.48749103e-02 -7.46154428e-01 -3.90565619e-02 3.79690140e-01] | [10.368812561035156, 1.5904008150100708] |
39db779f-10e1-4dbf-9186-f05295ed0865 | genetic-algorithm-based-floor-planning-system | 1704.06016 | null | http://arxiv.org/abs/1704.06016v1 | http://arxiv.org/pdf/1704.06016v1.pdf | Genetic Algorithm Based Floor Planning System | Genetic Algorithms are widely used in many different optimization problems
including layout design. The layout of the shelves play an important role in
the total sales metrics for superstores since this affects the customers'
shopping behaviour. This paper employed a genetic algorithm based approach to
design shelf layout of superstores. The layout design problem was tackled by
using a novel chromosome representation which takes many different parameters
to prevent dead-ends and improve shelf visibility into consideration. Results
show that the approach can produce reasonably good layout designs in very short
amounts of time. | ['Mehmet Serdar Guzel', 'Hamide Ozlem Dalgic', 'Erkan Bostanci'] | 2017-04-20 | null | null | null | null | ['layout-design'] | ['computer-vision'] | [-3.57764363e-01 -1.11925535e-01 -1.22793391e-01 -5.34066200e-01
-2.79268250e-02 -8.83062720e-01 -6.72936887e-02 3.86070728e-01
-3.59381378e-01 9.82659280e-01 -6.02098070e-02 -4.11946505e-01
-7.66403675e-01 -1.27824998e+00 -5.56268752e-01 -7.99458146e-01
-9.96736661e-02 7.54053116e-01 1.73877850e-01 -8.77648115e-01
8.89377534e-01 5.56016803e-01 -1.78185844e+00 -5.52265085e-02
7.69708335e-01 5.42055011e-01 4.47874516e-01 5.15922964e-01
-2.08897963e-01 1.38749465e-01 -7.56580532e-01 -3.73698235e-01
3.63938302e-01 -4.61109936e-01 -3.53784621e-01 4.25926656e-01
-5.90292692e-01 1.74663782e-01 2.90419161e-01 6.51545465e-01
5.18601120e-01 4.38250363e-01 6.64542615e-01 -1.18479371e+00
-4.19966280e-01 8.76022100e-01 -5.88699698e-01 2.37128586e-01
3.10674220e-01 -2.88288772e-01 8.98775995e-01 -2.27907717e-01
5.52620888e-01 1.17094910e+00 3.33523393e-01 -3.23754013e-01
-1.27949977e+00 -1.45986900e-01 -3.08733284e-02 1.17602505e-01
-1.49844432e+00 4.69366424e-02 5.04013002e-01 4.01068740e-02
1.02834451e+00 6.11994028e-01 9.84206319e-01 -3.03685874e-01
5.98668575e-01 5.81017315e-01 7.93954134e-01 -4.51161265e-01
6.16188109e-01 4.11053330e-01 3.64983156e-02 2.04267427e-01
9.18895364e-01 9.76092517e-02 1.63215935e-01 1.48240298e-01
2.87333101e-01 -2.65854448e-01 7.14669004e-02 -5.37405074e-01
-3.24363679e-01 1.47514451e+00 4.24886763e-01 2.99061984e-01
-1.10359959e-01 -4.65535372e-02 4.63968128e-01 3.62898290e-01
-2.47311652e-01 9.14469302e-01 -4.09177095e-01 -1.34662524e-01
-4.82650548e-01 6.25202060e-01 1.03081346e+00 1.18664062e+00
-1.06852263e-01 -4.55303863e-02 4.46597457e-01 8.20530534e-01
3.00596267e-01 4.01997238e-01 -4.76548709e-02 -4.01334643e-01
4.96657163e-01 5.44673800e-01 3.47285092e-01 -1.07448411e+00
-7.82400191e-01 -6.45575702e-01 -3.91947299e-01 3.13242853e-01
2.29776725e-01 -3.53597820e-01 -6.50894582e-01 8.71571422e-01
3.34471017e-01 -7.75416732e-01 -2.23137736e-02 7.43460953e-01
5.57698548e-01 8.96827161e-01 -3.76277864e-01 -3.20564330e-01
1.27976096e+00 -9.19319868e-01 -1.06588590e+00 4.27113026e-02
7.02785373e-01 -1.12656391e+00 4.47006583e-01 8.82804871e-01
-1.44038415e+00 -3.11507940e-01 -1.40903354e+00 3.83006632e-01
-7.45336950e-01 1.51343256e-01 7.03795433e-01 1.17685640e+00
-9.98870790e-01 2.67967463e-01 7.56100118e-02 -3.76798958e-01
-1.68741554e-01 8.72172356e-01 8.85367095e-02 -1.12407424e-01
-7.31338382e-01 1.03793478e+00 5.98997474e-01 3.55455577e-01
1.16681764e-02 8.55840684e-04 -7.59047329e-01 2.37202018e-01
6.87999845e-01 -4.20516849e-01 9.18241143e-01 -5.70019484e-01
-1.34395862e+00 5.48969388e-01 2.27815509e-01 -3.70238721e-01
2.87293553e-01 4.15889561e-01 -3.60506415e-01 -3.09779733e-01
-3.20691437e-01 2.33515516e-01 2.32192412e-01 -1.42702842e+00
-8.31010401e-01 -3.23386490e-01 1.57391176e-01 4.70433682e-01
3.47726047e-01 -1.11403510e-01 -2.44653925e-01 -5.34654558e-01
2.50695318e-01 -1.06791723e+00 -6.04437053e-01 -7.67943263e-01
-1.74685657e-01 -7.34228827e-03 3.96631747e-01 -4.02666628e-01
1.46047306e+00 -1.79288816e+00 2.93029010e-01 8.07737291e-01
-5.16185582e-01 1.23960257e-01 2.18629599e-01 8.74191284e-01
2.64398336e-01 8.46244842e-02 1.17642798e-01 2.05935612e-01
3.04765970e-01 7.40165949e-01 5.69380343e-01 2.80454427e-01
-3.84989411e-01 6.69798672e-01 -4.04973716e-01 -2.10774511e-01
2.82126188e-01 1.05844393e-01 -6.52311087e-01 4.94739898e-02
3.80547494e-02 -3.26590449e-01 -3.90324861e-01 5.79631448e-01
9.94259417e-01 5.88902056e-01 3.34220052e-01 2.29923233e-01
-4.97464597e-01 -9.52865705e-02 -1.47509539e+00 1.18874478e+00
-4.55223441e-01 1.47312745e-01 3.32273573e-01 -1.08537626e+00
1.39709496e+00 -4.38192077e-02 2.72115648e-01 -1.06715488e+00
7.58785903e-01 1.36895850e-01 4.19067651e-01 -5.98044455e-01
8.63071799e-01 -2.96815969e-02 -3.91692638e-01 3.00845146e-01
-5.23904622e-01 -5.14712036e-01 6.20223582e-01 -8.11705440e-02
4.83593315e-01 -1.84402183e-01 3.82734567e-01 -7.51740873e-01
5.28679132e-01 4.61909711e-01 4.35430080e-01 3.53785306e-01
1.54815599e-01 2.64390111e-01 6.75043523e-01 -2.76839554e-01
-1.39275706e+00 -6.94841862e-01 -1.11364275e-01 5.39309084e-01
4.85976249e-01 -1.56358052e-02 -8.30626607e-01 -3.75007093e-02
2.98126698e-01 1.09040725e+00 -6.90541387e-01 -9.32233632e-02
-6.78738534e-01 -7.81025231e-01 -1.90256104e-01 4.43120807e-01
1.49237856e-01 -6.85276568e-01 -7.31016695e-01 6.37855411e-01
4.21237707e-01 -4.12726432e-01 -3.81919175e-01 6.07703805e-01
-1.13194203e+00 -8.49802434e-01 -5.30004442e-01 -1.10375512e+00
9.94283676e-01 5.40265977e-01 8.97398293e-01 3.03001255e-01
-5.14260948e-01 -3.41707498e-01 -6.77420795e-01 -4.29197371e-01
-1.58664837e-01 3.42993200e-01 -5.15335441e-01 -6.18776798e-01
3.20132256e-01 -2.37054050e-01 -4.97266501e-01 9.21895623e-01
-9.10772681e-01 -4.91399735e-01 3.47859204e-01 8.01659107e-01
3.79542023e-01 1.20384610e+00 8.31796646e-01 -7.40567923e-01
1.00515020e+00 -4.89393324e-01 -1.09903979e+00 1.78846791e-01
-5.11144519e-01 2.91348636e-01 7.19683707e-01 1.55004397e-01
-8.97586763e-01 -7.57718310e-02 -3.64511728e-01 6.51370943e-01
4.54486795e-02 3.13004225e-01 -5.01928627e-01 -3.31545204e-01
-5.73249161e-02 -1.14785813e-01 1.72113746e-01 -4.93337214e-01
2.66171675e-02 3.77527952e-01 4.44534458e-02 -3.35184038e-01
4.98975575e-01 -4.85845655e-03 2.63128012e-01 -7.67902434e-01
2.33135521e-01 -6.67723358e-01 -5.18307507e-01 -1.90228000e-01
4.39048350e-01 -7.49713555e-02 -1.01661348e+00 8.64026789e-03
-4.45439041e-01 1.05481498e-01 -2.27965880e-02 2.91761756e-01
-4.28639412e-01 -2.27812946e-01 5.15790172e-02 -1.09017551e+00
-1.17075227e-01 -1.31455326e+00 3.67115676e-01 4.17656749e-01
-2.30619654e-01 -9.45017755e-01 -1.57959506e-01 3.63710046e-01
1.83276266e-01 5.02088130e-01 1.10017264e+00 -3.03469837e-01
-5.91056943e-01 -2.37308264e-01 1.24820761e-01 -1.22603148e-01
-9.15249437e-02 7.42404461e-02 4.37676720e-02 -2.49690518e-01
-2.15431586e-01 4.46843773e-01 2.94369608e-01 9.18863773e-01
2.84377396e-01 -1.12126105e-01 -4.65990037e-01 3.96253884e-01
2.15486050e+00 1.50881410e+00 9.62496698e-01 9.24120486e-01
-8.87761042e-02 8.81193817e-01 1.33410275e+00 6.79621994e-01
2.02524483e-01 7.36869931e-01 6.14052296e-01 -7.27546364e-02
4.45291996e-01 -8.03413913e-02 -2.34637678e-01 4.31560397e-01
2.76508987e-01 -9.86123741e-01 -4.77289826e-01 7.90963948e-01
-1.75961614e+00 -6.62055612e-01 -5.32160640e-01 2.20061636e+00
5.29237017e-02 3.55352491e-01 4.85616237e-01 7.70011246e-01
6.40606642e-01 -4.39980596e-01 -9.24157649e-02 -1.65496731e+00
-1.70583203e-01 -2.85624489e-02 1.21132350e+00 3.77640575e-01
-5.35346389e-01 4.38379765e-01 7.07387304e+00 6.61004364e-01
-4.42812592e-01 -2.71382809e-01 4.44009274e-01 -3.81759912e-01
-5.51869154e-01 2.04705857e-02 -9.55923676e-01 6.55773997e-01
5.81188977e-01 -1.80905372e-01 3.96426618e-01 5.12141764e-01
6.91208720e-01 -1.00963211e+00 -3.64876777e-01 6.30880713e-01
-6.35617301e-02 -1.21616757e+00 -5.52903712e-01 3.63985777e-01
1.09175169e+00 -9.14933324e-01 7.63886720e-02 -2.87580431e-01
1.90576017e-01 -9.06129599e-01 7.34490633e-01 6.23834841e-02
-7.92077556e-02 -1.79134464e+00 1.27492130e+00 -2.52593006e-03
-1.14786279e+00 -3.25633407e-01 -3.53436291e-01 -9.51129496e-02
9.64878261e-01 2.43333876e-01 -8.40637624e-01 5.07053733e-01
5.63108802e-01 -2.89464772e-01 -1.51786879e-01 1.98058331e+00
1.98275328e-01 -1.84954070e-02 -6.23500347e-01 -9.68676865e-01
8.53492379e-01 -8.75338078e-01 2.90345192e-01 6.11656964e-01
5.30744016e-01 2.83656448e-01 -2.40731001e-01 4.55741912e-01
4.84271348e-01 5.78185618e-01 -5.55800200e-01 1.46836251e-01
2.68816948e-01 7.60339439e-01 -1.42510092e+00 2.51349628e-01
-5.31547004e-03 4.13983136e-01 -6.29465938e-01 -1.83759048e-03
-9.67503726e-01 -8.76333833e-01 4.49527621e-01 5.12899160e-01
8.21418107e-01 -3.15860748e-01 -6.64761424e-01 -7.39930421e-02
-3.42413783e-01 -5.66195726e-01 3.33135575e-01 -5.82696974e-01
-3.77215415e-01 3.30490112e-01 1.97775960e-01 -7.48907328e-01
7.52858669e-02 -6.16804481e-01 -6.59708083e-01 5.35306811e-01
-1.09649289e+00 -6.76275074e-01 2.32152462e-01 6.32921699e-03
7.78090477e-01 -2.16348439e-01 2.84170210e-01 4.09698009e-01
-6.30928814e-01 5.53359210e-01 9.79561090e-01 -6.59770370e-01
6.21644929e-02 -1.24617660e+00 1.53704926e-01 6.09039009e-01
-6.25970423e-01 5.54389298e-01 1.29189897e+00 -7.21642196e-01
-1.63906956e+00 -2.44220406e-01 9.05267298e-01 5.36337733e-01
2.91219175e-01 -2.52858371e-01 -2.14248270e-01 -1.55123218e-03
5.16712487e-01 -1.11610341e+00 7.47706294e-01 1.67991698e-01
7.52024531e-01 -1.89918578e-01 -1.49199855e+00 4.69868869e-01
8.34606290e-01 6.80313468e-01 -3.25224787e-01 7.81461373e-02
2.26577163e-01 -3.08131009e-01 -7.45605171e-01 -1.81020200e-01
5.57761610e-01 -1.06333053e+00 8.31559062e-01 -4.25697826e-02
-4.36579809e-02 -1.46340653e-01 1.64735973e-01 -1.55434728e+00
-3.84400278e-01 -6.49615943e-01 7.64447808e-01 1.22332764e+00
7.30740607e-01 -8.27162027e-01 8.96287262e-01 6.25774145e-01
-2.90453225e-01 -9.99618649e-01 -5.16203225e-01 -7.43626297e-01
-2.99545765e-01 3.58542472e-01 1.13187385e+00 1.77354038e-01
1.45658225e-01 1.84397817e-01 -2.02715263e-01 -1.29087657e-01
5.26525080e-01 2.74032146e-01 4.85445112e-01 -1.00384521e+00
-1.99745163e-01 -5.48350811e-01 -4.61043715e-01 -5.56695998e-01
-5.28606713e-01 -3.34488690e-01 1.29366620e-02 -1.84474194e+00
-4.82590258e-01 -8.12789798e-01 2.09579855e-01 -3.34554851e-01
4.14675325e-01 6.73506036e-03 1.68322742e-01 -5.30338466e-01
-1.33590773e-01 1.65161803e-01 1.34361720e+00 1.72644541e-01
-4.74410176e-01 3.45876545e-01 -6.95038676e-01 2.62594938e-01
9.36818004e-01 -4.20317918e-01 -9.52877104e-01 -2.62219429e-01
7.44036913e-01 1.37667298e-01 -5.98597884e-01 -5.38742483e-01
7.15623647e-02 -2.54452765e-01 1.78411976e-02 -1.21104813e+00
2.33005628e-01 -1.25417304e+00 9.29163575e-01 6.81703925e-01
-7.07608983e-02 6.59319878e-01 3.85972619e-01 3.14382493e-01
-2.16311604e-01 -1.12551570e+00 6.89804077e-01 -2.14169044e-02
-6.79804385e-01 -4.19914097e-01 -6.60385072e-01 -6.16989970e-01
1.77489817e+00 -9.60483432e-01 3.03894430e-02 -3.13527793e-01
-6.23732150e-01 7.04111278e-01 5.94100296e-01 4.21109825e-01
3.27740788e-01 -9.69651937e-01 -1.99280754e-01 2.39619136e-01
-2.37521440e-01 -3.68842244e-01 2.85738409e-01 4.75202858e-01
-1.50037670e+00 9.26065803e-01 -6.52190745e-01 -1.06663913e-01
-1.50915599e+00 8.95092487e-01 5.57426848e-02 -1.42312512e-01
-1.64463669e-01 1.25535202e+00 -4.03351247e-01 -2.67774649e-02
8.46339762e-02 -2.78038949e-01 -5.83746612e-01 4.28966641e-01
1.50083318e-01 7.52496004e-01 2.10984349e-01 -7.33636618e-01
-1.92547485e-01 8.92415643e-01 4.22764383e-02 -8.75304043e-02
1.51432312e+00 -5.60736299e-01 1.05647586e-01 -1.68488279e-01
1.23769605e+00 -1.71992242e-01 -7.41450429e-01 4.94260609e-01
-1.11020487e-02 -9.75229740e-01 3.36116344e-01 -8.15794051e-01
-1.19973052e+00 2.44581386e-01 4.47267652e-01 6.63231134e-01
1.42851269e+00 -5.44336021e-01 1.00061357e+00 4.35102805e-02
6.55387044e-01 -1.64167321e+00 -5.06785810e-01 3.61274816e-02
6.33181214e-01 -9.46133137e-01 2.61477143e-01 -6.85739756e-01
-7.82111883e-01 1.15578115e+00 1.95652917e-01 -2.69618243e-01
4.14503813e-01 7.57874370e-01 -3.04203480e-01 -3.01180989e-01
-4.73958701e-01 -3.98838788e-01 -1.87545866e-01 5.02794206e-01
1.82085559e-01 1.63961083e-01 -1.38997233e+00 3.45709622e-01
-3.56624603e-01 -2.96968371e-01 5.81243575e-01 1.29279757e+00
-7.50996709e-01 -1.73511171e+00 -8.53747010e-01 3.79047573e-01
-3.13249767e-01 2.10178584e-01 -2.79312253e-01 1.10982382e+00
4.84332681e-01 1.19891870e+00 -1.24373566e-02 -1.81631893e-01
6.32180035e-01 -4.74807858e-01 4.26343173e-01 -2.34986469e-01
-1.11554217e+00 3.67055058e-01 6.54811978e-01 -6.05624681e-03
-7.12585747e-02 -8.03730547e-01 -1.33863449e+00 -9.38572884e-01
-8.69426370e-01 9.20239866e-01 1.06785512e+00 4.59555626e-01
-4.47350461e-03 7.01954603e-01 8.00193548e-01 -3.37680995e-01
-1.06196463e-01 -1.88089177e-01 -1.23175693e+00 -1.83369383e-01
-1.98173836e-01 -9.57210183e-01 1.99656356e-02 -2.59695679e-01] | [5.729421138763428, 3.566906690597534] |
a42bc2e6-77a9-4b35-8595-3f8ec2459f70 | a-generalised-seizure-prediction-with | 1707.01976 | null | http://arxiv.org/abs/1707.01976v2 | http://arxiv.org/pdf/1707.01976v2.pdf | A Generalised Seizure Prediction with Convolutional Neural Networks for Intracranial and Scalp Electroencephalogram Data Analysis | Seizure prediction has attracted a growing attention as one of the most
challenging predictive data analysis efforts in order to improve the life of
patients living with drug-resistant epilepsy and tonic seizures. Many
outstanding works have been reporting great results in providing a sensible
indirect (warning systems) or direct (interactive neural-stimulation) control
over refractory seizures, some of which achieved high performance. However,
many works put heavily handcraft feature extraction and/or carefully tailored
feature engineering to each patient to achieve very high sensitivity and low
false prediction rate for a particular dataset. This limits the benefit of
their approaches if a different dataset is used. In this paper we apply
Convolutional Neural Networks (CNNs) on different intracranial and scalp
electroencephalogram (EEG) datasets and proposed a generalized retrospective
and patient-specific seizure prediction method. We use Short-Time Fourier
Transform (STFT) on 30-second EEG windows with 50% overlapping to extract
information in both frequency and time domains. A standardization step is then
applied on STFT components across the whole frequency range to prevent high
frequencies features being influenced by those at lower frequencies. A
convolutional neural network model is used for both feature extraction and
classification to separate preictal segments from interictal ones. The proposed
approach achieves sensitivity of 81.4%, 81.2%, 82.3% and false prediction rate
(FPR) of 0.06/h, 0.16/h, 0.22/h on Freiburg Hospital intracranial EEG (iEEG)
dataset, Children's Hospital of Boston-MIT scalp EEG (sEEG) dataset, and Kaggle
American Epilepsy Society Seizure Prediction Challenge's dataset, respectively.
Our prediction method is also statistically better than an unspecific random
predictor for most of patients in all three datasets. | ['Mohammad Reza Bonyadi', 'Jiawei Yang', 'Anh Duy Nguyen', 'Nhan Duy Truong', 'Levin Kuhlmann', 'Omid Kavehei'] | 2017-07-06 | null | null | null | null | ['seizure-prediction'] | ['medical'] | [ 1.30895019e-01 -1.88314825e-01 2.48212993e-01 -3.29247534e-01
-6.80862784e-01 -2.53988922e-01 3.21934611e-01 1.06669575e-01
-4.85000312e-01 1.00738132e+00 8.08542371e-02 -1.60287321e-01
-6.30712867e-01 -4.87687200e-01 -2.65030414e-01 -7.60455430e-01
-7.29977727e-01 5.48431799e-02 -1.75090525e-02 -2.43968949e-01
3.08615685e-01 5.46863437e-01 -1.34876668e+00 3.33527267e-01
8.96946251e-01 1.12790966e+00 1.88479498e-01 4.38539386e-01
3.95401478e-01 4.08069789e-01 -8.06867361e-01 2.83398837e-01
1.54313713e-01 -3.50104034e-01 -4.64450270e-01 -3.33517939e-01
-4.27155912e-01 -7.43490644e-03 -2.53278852e-01 9.12106276e-01
1.01828349e+00 -3.40580404e-01 6.47214413e-01 -1.11838186e+00
-2.34468699e-01 6.40201211e-01 -5.13038516e-01 5.82895458e-01
2.64006287e-01 8.76315236e-02 1.93608701e-01 -6.31582379e-01
1.24803282e-01 6.69366196e-02 6.81115627e-01 5.10581613e-01
-1.03453839e+00 -1.25875497e+00 -2.27362752e-01 4.11586791e-01
-1.71175754e+00 -2.23925114e-01 7.79371262e-01 -7.01022685e-01
1.31525898e+00 3.10742527e-01 1.11461627e+00 1.19107485e+00
8.26774776e-01 1.85422271e-01 1.14240253e+00 -1.07037298e-01
1.18845508e-01 2.76104789e-02 1.66614354e-01 -1.17337272e-01
-7.42154419e-02 3.69835466e-01 -5.75635254e-01 -2.48329848e-01
5.59637785e-01 6.75183758e-02 -9.29192960e-01 2.62570113e-01
-1.11583555e+00 6.43748939e-01 1.83910549e-01 8.23885202e-01
-7.19491780e-01 -2.65119046e-01 3.49592268e-01 4.22242492e-01
3.89945388e-01 6.64303839e-01 -9.08722520e-01 -2.66014934e-01
-1.16402960e+00 1.46274567e-01 5.94301164e-01 4.90229666e-01
2.73069531e-01 3.10074538e-01 -1.94961414e-01 7.68281579e-01
-2.15171561e-01 2.59505868e-01 1.30810440e+00 9.93760154e-02
2.89144218e-02 4.44514632e-01 -8.59807525e-03 -8.46571505e-01
-1.01490498e+00 -8.83701265e-01 -1.15507889e+00 -6.35639280e-02
-6.61617965e-02 -5.11662245e-01 -9.93710995e-01 1.58822834e+00
-3.22900653e-01 3.26789469e-01 1.19975142e-01 7.26906061e-01
7.56313384e-01 3.90200466e-01 4.88485098e-02 -5.58798671e-01
1.33830488e+00 -1.79120809e-01 -7.10777462e-01 -8.82273018e-02
6.96998835e-01 -6.36849761e-01 5.59065700e-01 9.90767419e-01
-7.49449492e-01 -1.81107938e-01 -1.08423090e+00 6.75613999e-01
-3.05968851e-01 2.66102195e-01 4.79966402e-01 5.08370399e-01
-1.00173807e+00 7.06393600e-01 -9.31154549e-01 -1.72056466e-01
4.32885081e-01 1.09060812e+00 -4.58126873e-01 5.80016077e-01
-1.41767490e+00 1.02837503e+00 3.93598735e-01 -1.19278513e-01
-8.23404551e-01 -8.50535512e-01 -2.80203015e-01 1.00825481e-01
-2.47249365e-01 -1.78728655e-01 7.69861102e-01 -9.36469793e-01
-1.27019143e+00 5.08298695e-01 2.18788877e-01 -7.80222535e-01
2.93426633e-01 -2.61918664e-01 -9.76181448e-01 -7.52188489e-02
-1.58942655e-01 3.34446222e-01 7.45174468e-01 -3.72054487e-01
-5.92326462e-01 -5.77584445e-01 -6.44156814e-01 -1.35601133e-01
-3.64346892e-01 1.34706721e-01 3.31506848e-01 -8.98448288e-01
2.01262251e-01 -6.40898645e-01 2.01540999e-02 -9.97464240e-01
-4.47843134e-01 -1.99590892e-01 6.98054254e-01 -8.97744358e-01
1.43042815e+00 -1.98397076e+00 -9.13461223e-02 1.98671162e-01
1.79354414e-01 3.66890550e-01 -1.57793108e-02 3.90303791e-01
-8.03715646e-01 -1.00209653e-01 -2.23020494e-01 3.64420891e-01
-4.15250242e-01 -4.24123645e-01 -2.76548505e-01 5.60292542e-01
2.43760392e-01 6.21261060e-01 -4.60417032e-01 1.95159227e-01
3.39862734e-01 9.85476911e-01 -2.40864575e-01 1.74357504e-01
4.63371187e-01 7.01776326e-01 -3.92180592e-01 4.70000029e-01
5.07874250e-01 7.90677965e-02 -1.66017964e-01 -2.96623617e-01
-1.54639766e-01 1.44867480e-01 -8.51018310e-01 1.36395490e+00
-3.09615910e-01 7.73233652e-01 -6.18453681e-01 -1.21635854e+00
1.26500416e+00 9.41871107e-01 7.92343140e-01 -7.72641480e-01
4.93847877e-01 4.66330886e-01 4.89088416e-01 -7.07155287e-01
-2.68000484e-01 -1.82644576e-01 2.79203206e-01 1.11017846e-01
9.06483829e-02 -5.64468242e-02 -1.99973539e-01 -2.84142256e-01
1.26195347e+00 -3.18811625e-01 5.86161673e-01 -6.36628211e-01
5.63449800e-01 -2.68917441e-01 5.86482823e-01 4.30028170e-01
-1.28988430e-01 6.55521750e-01 2.68881261e-01 -7.29022443e-01
-6.12874746e-01 -5.20595908e-01 -7.39266396e-01 3.10961366e-01
-2.22171828e-01 -3.06839406e-01 -7.51632750e-01 -3.29571038e-01
-3.89498085e-01 6.26784563e-01 -6.08049333e-01 -4.47294652e-01
-4.98984814e-01 -1.25084126e+00 6.08442485e-01 4.12804693e-01
4.73001987e-01 -1.31505632e+00 -1.00998878e+00 5.54842472e-01
1.11405522e-01 -6.75658047e-01 6.75481111e-02 7.81993210e-01
-6.48684740e-01 -1.14481819e+00 -8.74582112e-01 -7.56330311e-01
3.11271667e-01 -5.80991387e-01 6.93225026e-01 -1.69826061e-01
-4.43307757e-01 -1.56409740e-01 -4.34825927e-01 -5.58150232e-01
1.72511131e-01 1.64766222e-01 3.38288039e-01 2.65589133e-02
8.64103496e-01 -1.01455557e+00 -7.46733904e-01 2.01368690e-01
-5.67364991e-01 -8.19089338e-02 5.71618855e-01 8.82420838e-01
4.10225689e-01 1.15103237e-01 1.08733475e+00 -2.80931145e-01
9.27234352e-01 -5.95872819e-01 -6.08033538e-01 5.74116260e-02
-7.22401738e-01 -3.23627591e-01 8.82781029e-01 -7.96996713e-01
-3.89136702e-01 1.75359741e-01 -2.06534654e-01 -2.19937235e-01
-3.91293973e-01 3.45311075e-01 1.95403904e-01 -1.63142160e-01
7.76810229e-01 6.66212261e-01 -4.88536537e-01 -3.13981831e-01
-7.39325166e-01 9.29248929e-01 4.21435356e-01 9.47687924e-02
3.34743679e-01 -9.66401771e-02 -1.02996588e-01 -6.67850971e-01
-1.00813061e-01 -1.05889328e-01 -3.52186084e-01 1.03682339e-01
8.08418334e-01 -9.19087112e-01 -6.42721474e-01 6.05577290e-01
-9.29976702e-01 -1.07787915e-01 8.60629007e-02 1.06850028e+00
-4.72756892e-01 -1.65331572e-01 -3.20934534e-01 -8.38615477e-01
-8.87158334e-01 -1.45996058e+00 5.69073260e-01 1.42571419e-01
-5.17784178e-01 -5.03528953e-01 4.67329249e-02 -4.22788113e-01
7.06819057e-01 4.96689886e-01 7.23015189e-01 -1.22316229e+00
-8.19803029e-02 -5.18373191e-01 5.74525334e-02 3.30602676e-01
3.55818778e-01 -3.88595849e-01 -1.12589669e+00 -3.84575754e-01
4.05778646e-01 4.03439775e-02 6.29708111e-01 8.05562377e-01
1.27195549e+00 -8.93037766e-02 -4.34127986e-01 8.70884180e-01
1.47277761e+00 1.11260569e+00 6.91063941e-01 3.18355292e-01
9.85564142e-02 1.57135755e-01 5.28978072e-02 6.65187120e-01
-1.82835788e-01 4.53820050e-01 1.60180658e-01 2.28843391e-01
2.51840025e-01 4.06309068e-01 1.21423215e-01 6.76647604e-01
-2.30482697e-01 -8.39786455e-02 -1.19504619e+00 7.50427783e-01
-1.42905486e+00 -6.87425017e-01 -1.32918879e-01 2.28918242e+00
7.63561964e-01 7.26386905e-02 -9.55139473e-02 6.91305757e-01
6.38747156e-01 -4.52945679e-01 -4.75121230e-01 -1.93202421e-01
-1.84871688e-01 8.27631176e-01 4.19556260e-01 7.14827031e-02
-1.02200389e+00 4.75616097e-01 5.31236553e+00 7.38517404e-01
-1.80652130e+00 3.44296768e-02 6.46013200e-01 -2.93963790e-01
3.50672275e-01 -4.12646830e-01 -4.11559105e-01 6.97904587e-01
1.19161677e+00 -4.14247245e-01 5.33843994e-01 5.48695683e-01
2.86656827e-01 -1.10149942e-02 -9.60446537e-01 1.44279981e+00
-1.91980720e-01 -1.16083670e+00 -3.17644477e-01 6.22085072e-02
4.76510555e-01 2.92230874e-01 1.54260062e-02 1.09858774e-01
-4.43074912e-01 -1.48035181e+00 2.82611638e-01 6.93707168e-01
9.22363281e-01 -1.09408009e+00 1.02929020e+00 3.71856868e-01
-1.02191448e+00 -1.78870797e-01 -3.82271484e-02 -1.70868784e-01
-1.29484043e-01 6.95691288e-01 -8.20705116e-01 3.17213714e-01
1.03315496e+00 7.42414653e-01 -3.63803953e-01 1.25455213e+00
-1.44320577e-01 7.19532669e-01 -4.56111699e-01 -6.83300123e-02
4.64891642e-02 1.82409614e-01 5.82299590e-01 1.09488976e+00
7.45724559e-01 4.55112815e-01 -4.14050281e-01 5.64430773e-01
2.59217173e-01 1.97871909e-01 -5.22633314e-01 5.99623173e-02
3.61458004e-01 1.14446890e+00 -5.87622166e-01 -1.46534279e-01
-1.52710363e-01 6.05136573e-01 -1.38329118e-01 3.40628624e-01
-7.59966552e-01 -9.96932566e-01 3.09075981e-01 1.05276383e-01
-1.31759360e-01 2.91891187e-01 -4.23503160e-01 -1.17241693e+00
7.57049304e-03 -7.22076774e-01 1.18242674e-01 -6.73063099e-01
-9.56965327e-01 1.25550640e+00 -1.35068476e-01 -1.43936777e+00
-4.65364784e-01 -3.64638299e-01 -7.74111271e-01 1.16291618e+00
-1.34864020e+00 -6.05770171e-01 -1.91897109e-01 1.02068579e+00
4.22623962e-01 -4.53419626e-01 1.12478149e+00 3.68726492e-01
-4.65093940e-01 5.78493595e-01 -1.81345999e-01 8.10797215e-02
6.17630959e-01 -8.97259057e-01 -1.63258061e-01 5.41733384e-01
-2.97208786e-01 5.14340937e-01 6.34294212e-01 -5.21337986e-01
-9.91502762e-01 -1.05269575e+00 1.00045276e+00 1.04953079e-02
5.16123116e-01 -1.57867625e-01 -9.68228400e-01 5.63863456e-01
3.29218179e-01 -1.75833218e-02 6.74820483e-01 -2.34912440e-01
3.35392237e-01 -1.74848035e-01 -1.31098974e+00 2.61652052e-01
5.31943500e-01 -1.63642138e-01 -8.09440017e-01 3.37160140e-01
1.41507775e-01 -1.80264115e-01 -1.12062979e+00 8.93153489e-01
6.13284826e-01 -1.12728560e+00 5.80589294e-01 -2.33760685e-01
2.14518979e-01 -2.39346772e-02 6.86204284e-02 -1.60489881e+00
-3.05351764e-01 -8.03910017e-01 1.43295467e-01 6.32090867e-01
4.25942153e-01 -1.03740597e+00 4.83965248e-01 3.46255600e-01
-3.26369852e-01 -1.22990477e+00 -8.71797562e-01 -6.11568213e-01
9.21870247e-02 -6.02319002e-01 8.65317523e-01 9.15342867e-01
3.80874336e-01 3.70274335e-02 -3.61145794e-01 2.28080899e-01
1.98529720e-01 -2.05043703e-01 1.50812566e-01 -1.32609093e+00
6.65732697e-02 -4.37854201e-01 -7.79871821e-01 3.47222574e-02
-2.57742107e-01 -8.10324490e-01 -2.71734983e-01 -1.20188510e+00
2.44661830e-02 -1.07764125e-01 -9.36479270e-01 7.58965850e-01
2.73422509e-01 2.92797953e-01 -3.35540771e-01 -8.76507908e-03
3.49115491e-01 2.49897435e-01 7.86387503e-01 4.48237779e-03
-6.60833299e-01 9.43622589e-02 -3.36360902e-01 8.00052106e-01
1.19982195e+00 -6.05015397e-01 -4.84619230e-01 -1.09232292e-01
-7.84247275e-03 3.49617392e-01 1.94734335e-01 -1.48671544e+00
2.88194418e-01 1.24348290e-01 1.05310059e+00 -6.15971446e-01
2.20410630e-01 -7.05545425e-01 5.66266775e-01 6.90042794e-01
-8.32544342e-02 1.20207504e-03 4.84983504e-01 -3.50426361e-02
-4.06574935e-01 2.61729926e-01 8.67195904e-01 1.20303154e-01
-4.25164223e-01 4.97030377e-01 -4.44281816e-01 -2.44274616e-01
1.26068735e+00 -4.00509685e-01 6.72835633e-02 -1.67739734e-01
-1.15763938e+00 -1.04751490e-01 -2.40302309e-01 4.41828579e-01
7.57986844e-01 -1.06462204e+00 -7.57105708e-01 7.82760799e-01
1.50038317e-01 -5.10664880e-01 3.64476144e-01 1.32370341e+00
-1.46797895e-01 6.76691949e-01 -6.23117208e-01 -5.46170712e-01
-1.09340286e+00 2.08942115e-01 5.79792857e-01 6.86461851e-02
-8.82029653e-01 9.16269004e-01 1.82944208e-01 2.09590137e-01
2.35401019e-01 -3.81396890e-01 -6.08332932e-01 5.32539599e-02
7.70245850e-01 5.21555319e-02 6.25521183e-01 -4.54683810e-01
-6.88160956e-01 6.60261452e-01 -9.63671803e-02 2.49081016e-01
1.78686106e+00 4.93508250e-01 4.70077358e-02 8.30677673e-02
1.11426926e+00 -3.67586076e-01 -8.06725562e-01 4.07765418e-01
-2.84771305e-02 -5.06353639e-02 3.93890023e-01 -1.37331045e+00
-1.32345653e+00 9.32050407e-01 1.24447775e+00 4.08465087e-01
1.72823989e+00 -3.32013369e-01 6.71673000e-01 1.33682534e-01
5.62336087e-01 -8.49620163e-01 -6.36025846e-01 2.33775958e-01
1.11800146e+00 -8.35412025e-01 -2.41412222e-01 1.89237878e-01
-4.17661607e-01 1.42492366e+00 3.80496472e-01 -5.37857294e-01
1.01388812e+00 5.10466278e-01 -1.38538331e-01 -2.53777474e-01
-8.00037920e-01 1.62181258e-01 4.03379560e-01 6.38109386e-01
6.17786765e-01 2.11463198e-01 -7.30425358e-01 1.33657777e+00
-5.42613029e-01 4.88355577e-01 4.01287973e-01 8.43484581e-01
-3.21223527e-01 -8.36722016e-01 -1.67443722e-01 1.08925164e+00
-8.20318162e-01 -3.83568227e-01 -1.31995827e-01 9.57690120e-01
3.69978189e-01 8.95530462e-01 -1.09944474e-02 -7.36197591e-01
2.89983720e-01 3.03725183e-01 3.47334862e-01 -3.72818500e-01
-1.00643647e+00 4.48518813e-01 -2.85277814e-01 -5.75738668e-01
-1.50597855e-01 -3.62491488e-01 -1.32859242e+00 2.98012674e-01
-4.27533597e-01 3.45551908e-01 7.33207941e-01 8.51195574e-01
5.31872511e-01 7.32759476e-01 5.99602282e-01 -6.53986990e-01
-1.62962019e-01 -1.46660280e+00 -9.78038669e-01 7.26340711e-02
6.14537537e-01 -6.49608493e-01 -6.51470721e-01 -1.15151800e-01] | [13.22880744934082, 3.5133612155914307] |
1419e9a5-a150-4bed-8f22-2e08148cacb5 | inspro-propagating-instance-query-and | 2301.01882 | null | https://arxiv.org/abs/2301.01882v1 | https://arxiv.org/pdf/2301.01882v1.pdf | InsPro: Propagating Instance Query and Proposal for Online Video Instance Segmentation | Video instance segmentation (VIS) aims at segmenting and tracking objects in videos. Prior methods typically generate frame-level or clip-level object instances first and then associate them by either additional tracking heads or complex instance matching algorithms. This explicit instance association approach increases system complexity and fails to fully exploit temporal cues in videos. In this paper, we design a simple, fast and yet effective query-based framework for online VIS. Relying on an instance query and proposal propagation mechanism with several specially developed components, this framework can perform accurate instance association implicitly. Specifically, we generate frame-level object instances based on a set of instance query-proposal pairs propagated from previous frames. This instance query-proposal pair is learned to bind with one specific object across frames through conscientiously developed strategies. When using such a pair to predict an object instance on the current frame, not only the generated instance is automatically associated with its precursors on previous frames, but the model gets a good prior for predicting the same object. In this way, we naturally achieve implicit instance association in parallel with segmentation and elegantly take advantage of temporal clues in videos. To show the effectiveness of our method InsPro, we evaluate it on two popular VIS benchmarks, i.e., YouTube-VIS 2019 and YouTube-VIS 2021. Without bells-and-whistles, our InsPro with ResNet-50 backbone achieves 43.2 AP and 37.6 AP on these two benchmarks respectively, outperforming all other online VIS methods. | ['Kaiqi Huang', 'Xin Zhao', 'Yanhu Shan', 'Jian Jia', 'Naiyu Gao', 'Haoyang Zhang', 'Fei He'] | 2023-01-05 | null | null | null | null | ['video-instance-segmentation'] | ['computer-vision'] | [ 1.60174239e-02 -3.85817960e-02 -4.93681014e-01 -2.13200957e-01
-9.19499636e-01 -5.87115705e-01 6.29724860e-01 1.09424189e-01
-4.15779650e-01 7.04255760e-01 -2.53930122e-01 1.64318278e-01
1.04314752e-01 -6.54638350e-01 -9.87398446e-01 -4.56104547e-01
-3.14708173e-01 5.70854187e-01 1.00876009e+00 3.35089974e-02
1.47900105e-01 3.27174127e-01 -1.56492710e+00 4.41164941e-01
4.02148873e-01 1.22735155e+00 2.27160588e-01 8.04323077e-01
-1.78644061e-01 1.10029399e+00 -5.70177853e-01 -4.92708981e-01
5.00658333e-01 -3.00491989e-01 -9.40936327e-01 4.44813043e-01
8.57654870e-01 -5.55295348e-01 -4.44649130e-01 7.86812723e-01
2.49951929e-01 2.50886798e-01 4.09517229e-01 -1.31605899e+00
5.21666696e-03 7.85675943e-01 -8.74534249e-01 5.67719519e-01
4.45635527e-01 4.50365663e-01 1.06790721e+00 -9.49673533e-01
1.08272004e+00 9.92212951e-01 5.09540558e-01 4.78585631e-01
-1.07396269e+00 -5.85537553e-01 5.44824839e-01 3.50153595e-01
-1.35862732e+00 -4.82447565e-01 5.20467103e-01 -4.51975644e-01
6.29179239e-01 2.57279456e-01 9.30767000e-01 7.56324291e-01
-2.60397881e-01 1.14138937e+00 6.16874516e-01 2.02487648e-01
1.09928109e-01 -1.14660361e-03 3.20135564e-01 7.09157109e-01
-1.54957637e-01 -1.51900025e-02 -7.48411179e-01 9.00079310e-02
8.80042970e-01 1.99838057e-01 -3.11678588e-01 -1.14634335e-01
-1.41727817e+00 2.97687173e-01 4.40766275e-01 1.06648862e-01
-6.08271837e-01 5.02249420e-01 4.44380254e-01 -6.38837973e-03
1.09482557e-01 1.02181010e-01 -5.90853512e-01 -1.67938009e-01
-1.51010072e+00 2.43150294e-01 6.98027492e-01 1.31867492e+00
9.53802168e-01 -1.99020401e-01 -7.10669696e-01 4.78100836e-01
8.47859234e-02 1.87464595e-01 1.19184919e-01 -1.31547654e+00
4.17578459e-01 4.85126734e-01 2.38367394e-01 -9.84969616e-01
-2.33049709e-02 -4.37020004e-01 -4.30575192e-01 -9.69201624e-02
5.67393303e-01 -8.33281949e-02 -1.02466404e+00 1.65881968e+00
7.18634427e-01 1.07486939e+00 -1.26397878e-01 9.68387365e-01
8.66694748e-01 7.30409384e-01 2.61181414e-01 -4.46476847e-01
1.51028228e+00 -1.27676094e+00 -3.45022023e-01 -2.92637702e-02
4.34104472e-01 -6.60203576e-01 6.75717175e-01 5.03097177e-01
-1.41349065e+00 -8.01760495e-01 -6.07524335e-01 1.64696485e-01
-1.20461583e-02 2.42300630e-02 4.65600312e-01 6.82258904e-02
-7.74200916e-01 6.35169744e-01 -9.18499887e-01 -1.18689939e-01
6.50457859e-01 4.34819937e-01 -3.12248737e-01 4.16718386e-02
-8.63435626e-01 1.12579100e-01 5.18627465e-01 -1.90802682e-02
-1.20686865e+00 -8.71135414e-01 -5.84894657e-01 -2.48231664e-02
9.72839236e-01 -7.25520194e-01 1.23953617e+00 -1.20135713e+00
-1.18017280e+00 7.37998307e-01 -3.27438742e-01 -8.17374051e-01
7.28090942e-01 -4.30089563e-01 -2.34903321e-01 6.10334635e-01
2.11226851e-01 1.13632870e+00 1.07339275e+00 -1.28584540e+00
-1.19762099e+00 -6.80187419e-02 5.12547612e-01 2.18703430e-02
9.11220256e-03 4.48153242e-02 -1.38005435e+00 -5.42293549e-01
4.45003901e-03 -8.91957581e-01 -1.69029057e-01 1.72339246e-01
-4.40979213e-01 -4.18252885e-01 8.71631086e-01 -4.07759666e-01
1.42682409e+00 -2.20799828e+00 3.66251655e-02 1.43055305e-01
4.48960185e-01 4.28082645e-01 -6.77244365e-02 1.49161398e-01
1.37189493e-01 -3.42231654e-02 7.84475356e-03 -3.65754455e-01
-5.79999834e-02 2.00666770e-01 -2.38275796e-01 2.16299012e-01
2.32744947e-01 9.60936785e-01 -1.26324713e+00 -8.30447972e-01
1.57862023e-01 4.21697170e-01 -8.68878305e-01 2.56182611e-01
-6.35446608e-01 2.86529183e-01 -4.05017287e-01 6.51199341e-01
4.30412054e-01 -6.07066929e-01 6.24654293e-02 -5.22773325e-01
2.66954843e-02 3.11738928e-04 -1.33276641e+00 1.80088580e+00
-3.80023867e-02 5.57715714e-01 -9.26109031e-02 -8.53009164e-01
4.91975516e-01 3.79797697e-01 7.49689341e-01 -4.26778764e-01
-1.69437230e-01 1.00183472e-01 -3.29428434e-01 -5.72602212e-01
5.78002453e-01 3.69910806e-01 1.92825809e-01 2.38315955e-01
1.83042631e-01 3.08912486e-01 7.50437617e-01 7.07985342e-01
1.14747488e+00 5.20214975e-01 4.18632664e-02 1.32650241e-01
6.45989835e-01 6.26158938e-02 6.92815065e-01 9.32726681e-01
-3.10978353e-01 6.17128074e-01 5.55035233e-01 -4.76701170e-01
-7.55825281e-01 -9.00692225e-01 1.36014931e-02 1.18754888e+00
5.97389400e-01 -7.63673723e-01 -6.57959819e-01 -1.08317447e+00
-1.48881942e-01 3.24325174e-01 -5.32550752e-01 2.99672931e-01
-8.10280323e-01 -2.96157330e-01 2.76530951e-01 5.42208910e-01
5.49327791e-01 -1.09463084e+00 -6.30944669e-01 5.93252838e-01
-3.59159172e-01 -1.36505103e+00 -7.92792916e-01 -1.55167326e-01
-6.87324345e-01 -1.36340725e+00 -7.23704219e-01 -4.71215248e-01
4.83825177e-01 3.41398895e-01 1.36933362e+00 4.52962995e-01
-2.74954051e-01 5.43663919e-01 -4.04747814e-01 5.58601655e-02
-1.53660715e-01 2.44037099e-02 -1.13410912e-01 3.74263436e-01
2.67698437e-01 -3.41822058e-01 -9.74365950e-01 6.25424325e-01
-8.37295353e-01 1.46244287e-01 4.09281790e-01 5.36357284e-01
9.49977279e-01 -2.08725050e-01 4.50443953e-01 -8.54293108e-01
-2.94689447e-01 -6.55382335e-01 -7.14037657e-01 1.55385002e-01
-1.63281038e-01 -2.78384030e-01 3.00235063e-01 -6.00980639e-01
-7.18672633e-01 4.28370208e-01 1.56579353e-02 -7.86660433e-01
-2.10989311e-01 3.01715285e-01 2.04921097e-01 2.65882343e-01
5.73183537e-01 3.77009332e-01 -2.44209588e-01 -2.10875213e-01
2.81426668e-01 2.07021102e-01 8.97730589e-01 -7.17583835e-01
7.66071141e-01 4.88846183e-01 -2.22697586e-01 -5.72529018e-01
-1.04824829e+00 -7.06236720e-01 -4.93987501e-01 -5.81569791e-01
9.56759095e-01 -1.09317720e+00 -8.27193737e-01 2.55934209e-01
-1.18564177e+00 -6.14144683e-01 -3.35863203e-01 2.94331908e-01
-6.17946148e-01 3.74526531e-01 -7.06750035e-01 -4.86581802e-01
-1.65835708e-01 -1.23483241e+00 1.27288914e+00 1.89599976e-01
-1.34563476e-01 -6.48605645e-01 -2.41519436e-01 2.73412168e-01
-5.26215741e-03 2.05264241e-01 1.16203062e-01 -5.17416835e-01
-1.31750047e+00 -7.34244436e-02 -3.46080959e-01 2.84948871e-02
-8.87116715e-02 3.88190717e-01 -7.75944650e-01 -2.02406168e-01
-3.96296680e-01 1.89163722e-02 8.50160897e-01 3.56317371e-01
1.41570306e+00 -5.03319740e-01 -5.04104137e-01 5.47674954e-01
1.49447906e+00 8.74419734e-02 7.09229469e-01 1.50509030e-01
7.05115139e-01 3.75006825e-01 9.22968268e-01 5.69095612e-01
3.45926166e-01 9.95429039e-01 5.53255856e-01 1.40962988e-01
-3.20579141e-01 -6.76745251e-02 4.60583419e-01 3.71422023e-01
-4.41461802e-01 -2.58021295e-01 -6.31330729e-01 5.89579046e-01
-2.09177065e+00 -1.36463225e+00 -3.37339759e-01 2.06123972e+00
9.75083292e-01 3.26193273e-01 6.55233681e-01 -6.87405169e-02
8.41989636e-01 5.44594675e-02 -3.78014147e-01 4.01657313e-01
2.06093386e-01 5.59671270e-03 3.81585509e-01 3.54627758e-01
-1.12374938e+00 1.14261651e+00 5.26679134e+00 9.82004106e-01
-9.26985621e-01 7.67872036e-02 7.36421943e-01 -3.35880458e-01
1.79695278e-01 -1.46489115e-02 -1.11663175e+00 7.15108812e-01
6.57295704e-01 -8.19397941e-02 2.78194040e-01 7.61512637e-01
2.25788847e-01 -1.99310601e-01 -1.44538367e+00 9.22638118e-01
-3.23799253e-01 -1.77698731e+00 9.19637904e-02 -9.48933885e-02
6.69733763e-01 -6.20173588e-02 -1.51841536e-01 3.30103219e-01
-1.31613286e-02 -5.71160555e-01 1.05427885e+00 5.96541643e-01
6.01427019e-01 -5.21263301e-01 4.88436222e-01 1.86941311e-01
-1.52483296e+00 5.11200652e-02 -1.59435451e-01 2.46626139e-01
4.57288563e-01 4.35669452e-01 -5.61425090e-01 4.96511012e-01
7.42089033e-01 9.66383219e-01 -3.66391361e-01 1.38084602e+00
-1.17167532e-01 7.85959661e-01 -5.23776174e-01 4.26250994e-01
4.88887012e-01 1.65343896e-01 5.80920279e-01 1.40039921e+00
1.04831330e-01 3.91192228e-01 5.23402810e-01 7.35882223e-01
-1.40181646e-01 -7.31558800e-02 -1.53121367e-01 2.49702603e-01
5.64113557e-01 1.46451652e+00 -9.09468293e-01 -1.00605047e+00
-4.50973511e-01 9.29968357e-01 1.35804459e-01 4.54261929e-01
-1.38894105e+00 1.04220755e-01 5.67407966e-01 3.80610377e-01
7.44301736e-01 -1.36661783e-01 3.00857008e-01 -1.07980156e+00
4.89498228e-02 -8.24029386e-01 4.46080416e-01 -8.29279482e-01
-1.07162333e+00 6.09376729e-01 -1.45765697e-03 -1.36923802e+00
-2.28380337e-01 -1.94053590e-01 -5.81220031e-01 2.09736869e-01
-1.49839199e+00 -9.20491159e-01 -6.38979137e-01 9.62599337e-01
9.12603021e-01 2.15236008e-01 1.93489969e-01 6.02160990e-01
-7.36044228e-01 6.08587503e-01 -6.63618088e-01 3.42515677e-01
6.82500720e-01 -1.19990993e+00 2.21596435e-01 8.44659030e-01
5.98742545e-01 4.53827411e-01 6.50449574e-01 -7.10038841e-01
-1.26013803e+00 -1.38195980e+00 4.48832244e-01 -6.18708372e-01
7.17520297e-01 -2.00470723e-03 -9.68518913e-01 7.96249866e-01
1.50195211e-01 4.68642682e-01 1.89335510e-01 -3.80564958e-01
-2.92945176e-01 -2.32969895e-01 -7.99247682e-01 4.84987736e-01
1.30206561e+00 -2.73006380e-01 -3.74734730e-01 6.02967381e-01
7.91459501e-01 -5.85612476e-01 -9.42571700e-01 2.61036992e-01
4.41443443e-01 -1.21698070e+00 1.16718328e+00 -6.02118492e-01
6.77307844e-01 -6.04709983e-01 2.33207569e-02 -5.10245144e-01
-1.57641262e-01 -1.11378145e+00 -7.14163065e-01 1.40892553e+00
3.09231490e-01 -1.32593811e-01 9.93121207e-01 5.52079976e-01
-1.79585949e-01 -9.19246376e-01 -6.32568836e-01 -8.43777120e-01
-5.86755276e-01 -5.71593881e-01 5.90659618e-01 7.77696192e-01
-2.95530289e-01 5.55464551e-02 -3.38912964e-01 3.13931972e-01
8.10426414e-01 1.29399404e-01 1.14316189e+00 -8.79082203e-01
-6.13045096e-01 -3.85311842e-01 -5.10902107e-01 -1.53555501e+00
-3.11184861e-02 -6.39294803e-01 -3.82080153e-02 -1.23765779e+00
2.92552531e-01 -6.36609316e-01 -3.40612799e-01 5.65369189e-01
-4.17871952e-01 7.29772925e-01 4.90700305e-01 5.00174701e-01
-1.29766762e+00 2.36326642e-02 1.13978744e+00 -8.50908756e-02
-3.29367459e-01 5.93901239e-02 -2.03062460e-01 7.95873523e-01
4.59845155e-01 -5.59490502e-01 -2.99305588e-01 -2.14546427e-01
1.36438519e-01 5.13977110e-01 7.72523344e-01 -1.15749681e+00
4.53276277e-01 -1.17875941e-01 2.61678994e-01 -8.13076437e-01
4.08450395e-01 -8.52728367e-01 4.42780077e-01 3.39986801e-01
-1.56484455e-01 -1.36162043e-01 3.27460468e-02 8.41282189e-01
-2.39305690e-01 -7.90227056e-02 7.24640489e-01 -2.71489561e-01
-1.21426785e+00 8.93920422e-01 -8.09902474e-02 1.91983759e-01
1.32720983e+00 -5.28437078e-01 -1.96193591e-01 -1.48499012e-01
-1.09365630e+00 4.97459352e-01 3.29946637e-01 2.76969999e-01
4.22523499e-01 -1.26411474e+00 -6.54317021e-01 -8.99523348e-02
1.09556161e-01 1.19483285e-01 2.78983384e-01 1.32124889e+00
-3.65029693e-01 -3.56265754e-02 8.05891231e-02 -1.05929208e+00
-1.21273005e+00 7.31058478e-01 2.69584268e-01 -1.52037635e-01
-8.16888213e-01 9.88512397e-01 4.31800485e-01 2.33773530e-01
3.46312016e-01 -1.82500958e-01 -1.58335298e-01 1.66037992e-01
6.49027884e-01 3.65363479e-01 -3.69755834e-01 -6.65483057e-01
-2.53846556e-01 5.53557873e-01 -1.96587101e-01 2.61105150e-02
1.13736618e+00 -1.54823601e-01 1.26884952e-01 1.53577849e-01
9.42886055e-01 -3.23360339e-02 -1.84171116e+00 -4.02558476e-01
-3.12220715e-02 -6.52987719e-01 -3.34669441e-01 -5.27851820e-01
-1.47975981e+00 6.01012111e-01 1.70496941e-01 3.58457297e-01
1.02709198e+00 2.44294629e-01 1.02727771e+00 8.57578069e-02
5.38529634e-01 -8.37851822e-01 4.15554345e-01 1.54836386e-01
5.48624218e-01 -1.07126594e+00 -5.16103022e-02 -6.15405381e-01
-4.52717245e-01 9.80248034e-01 8.50117207e-01 -1.97677284e-01
4.20189261e-01 1.04580075e-01 -2.72803277e-01 -2.68860132e-01
-9.21212971e-01 -2.89277881e-01 4.18663681e-01 2.63223410e-01
1.84765473e-01 -2.32922867e-01 -1.81794554e-01 3.74170929e-01
2.27274209e-01 2.86268085e-01 4.51721638e-01 5.76913297e-01
-5.61575532e-01 -9.11750972e-01 -2.70903349e-01 4.56215054e-01
-7.15291619e-01 -7.30263516e-02 1.11773424e-01 8.59532773e-01
1.48682892e-01 6.56186819e-01 1.49042845e-01 -1.88289717e-01
1.20506190e-01 -2.26098105e-01 4.24200326e-01 -6.66976571e-01
-7.63489783e-01 3.04316252e-01 1.41722700e-02 -1.11025167e+00
-7.67071784e-01 -6.92838311e-01 -1.50660670e+00 -2.46560246e-01
-4.17969137e-01 3.72041464e-02 1.76217049e-01 9.08534110e-01
4.90836054e-01 4.93010789e-01 4.46183354e-01 -1.13245940e+00
-1.32293016e-01 -4.45762843e-01 -2.89312840e-01 4.62494791e-01
2.99712241e-01 -5.83636522e-01 -1.75955012e-01 4.83346969e-01] | [9.163708686828613, -0.07813439518213272] |
32e2d53f-3412-4006-91bb-f782e78cf0c6 | few-shot-action-recognition-via-intra-and | 2305.06114 | null | https://arxiv.org/abs/2305.06114v1 | https://arxiv.org/pdf/2305.06114v1.pdf | Few-shot Action Recognition via Intra- and Inter-Video Information Maximization | Current few-shot action recognition involves two primary sources of information for classification:(1) intra-video information, determined by frame content within a single video clip, and (2) inter-video information, measured by relationships (e.g., feature similarity) among videos. However, existing methods inadequately exploit these two information sources. In terms of intra-video information, current sampling operations for input videos may omit critical action information, reducing the utilization efficiency of video data. For the inter-video information, the action misalignment among videos makes it challenging to calculate precise relationships. Moreover, how to jointly consider both inter- and intra-video information remains under-explored for few-shot action recognition. To this end, we propose a novel framework, Video Information Maximization (VIM), for few-shot video action recognition. VIM is equipped with an adaptive spatial-temporal video sampler and a spatiotemporal action alignment model to maximize intra- and inter-video information, respectively. The video sampler adaptively selects important frames and amplifies critical spatial regions for each input video based on the task at hand. This preserves and emphasizes informative parts of video clips while eliminating interference at the data level. The alignment model performs temporal and spatial action alignment sequentially at the feature level, leading to more precise measurements of inter-video similarity. Finally, These goals are facilitated by incorporating additional loss terms based on mutual information measurement. Consequently, VIM acts to maximize the distinctiveness of video information from limited video data. Extensive experimental results on public datasets for few-shot action recognition demonstrate the effectiveness and benefits of our framework. | ['John See', 'Shuyuan Li', 'Yuxi Li', 'Tieyuan Chen', 'Weiyao Lin', 'Huabin Liu'] | 2023-05-10 | null | null | null | null | ['few-shot-action-recognition', 'action-recognition-in-videos', 'video-similarity', 'action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 4.28690165e-01 -4.47184622e-01 -6.20161951e-01 -2.88280606e-01
-7.22010314e-01 -3.40968430e-01 3.55051607e-01 -9.61063206e-02
-2.68747807e-01 3.95430386e-01 4.16422278e-01 2.96320647e-01
-4.43828821e-01 -4.48508650e-01 -5.22125661e-01 -8.96863222e-01
-4.15703561e-03 -2.69633681e-01 3.17922980e-01 2.75726169e-01
3.42866868e-01 3.03435951e-01 -1.72477782e+00 4.64641809e-01
6.87219918e-01 1.21327746e+00 4.27541763e-01 5.34192204e-01
-1.64573759e-01 1.21147025e+00 -3.55497301e-01 -3.30114812e-02
2.67281413e-01 -6.83792770e-01 -3.96078974e-01 5.82332194e-01
3.28121096e-01 -5.63359022e-01 -5.79262078e-01 1.10631728e+00
2.49002427e-01 4.30377543e-01 2.69148916e-01 -1.44993985e+00
-1.95415184e-01 4.87469345e-01 -6.46964848e-01 6.47454560e-01
4.96018648e-01 3.04687172e-01 1.08838403e+00 -8.84533703e-01
6.50179684e-01 1.08348370e+00 3.92880321e-01 1.60155326e-01
-7.74060965e-01 -5.56099951e-01 3.50054562e-01 5.82404494e-01
-1.43151486e+00 -5.58012545e-01 9.82192814e-01 -5.07299602e-01
5.82860112e-01 2.68758267e-01 8.02806318e-01 8.02025676e-01
1.03417076e-01 1.19122267e+00 4.62298125e-01 -5.76001033e-02
1.47122309e-01 -1.91673622e-01 4.31123488e-02 5.35188198e-01
-1.05081618e-01 -7.52672181e-02 -9.03623402e-01 -3.31237586e-03
7.10581362e-01 5.28761804e-01 -5.98533988e-01 -4.30369705e-01
-1.20669770e+00 5.33252478e-01 -9.57526639e-02 5.38367629e-01
-5.70198417e-01 -2.00581357e-01 5.35945714e-01 2.31056049e-01
2.83193648e-01 8.53334088e-03 -1.51322857e-01 -5.59511840e-01
-1.14179349e+00 -9.96112153e-02 3.92831355e-01 1.05361176e+00
9.14226174e-01 6.36997372e-02 -3.89247656e-01 8.32380056e-01
1.07454397e-01 3.59296203e-01 6.15450382e-01 -1.05158710e+00
7.92967975e-01 7.11558700e-01 -7.96499699e-02 -1.45726788e+00
-4.40866733e-03 -6.89735115e-02 -7.76612937e-01 -2.59165466e-01
4.05875891e-01 8.50983411e-02 -4.66019899e-01 1.74106097e+00
3.16732943e-01 6.12183571e-01 -1.61690831e-01 1.12034202e+00
5.79118848e-01 6.17504478e-01 -8.50547478e-02 -8.07906926e-01
1.21979618e+00 -9.01233733e-01 -8.33752692e-01 1.75181776e-02
5.65168202e-01 -6.05892956e-01 8.35943878e-01 1.37018129e-01
-1.11491585e+00 -6.96446598e-01 -9.17488217e-01 2.97715604e-01
4.74975035e-02 1.29340455e-01 1.47755355e-01 2.94668078e-01
-6.04849517e-01 3.33701521e-01 -7.37408102e-01 -1.43314242e-01
4.46657062e-01 2.23851845e-01 -4.16611433e-01 -3.13900143e-01
-1.13666940e+00 1.82979673e-01 4.30044532e-01 -5.59361540e-02
-7.51523912e-01 -8.13312888e-01 -9.57746923e-01 7.49686584e-02
9.46775079e-01 -3.20349127e-01 9.23689544e-01 -1.28852975e+00
-1.32370472e+00 3.54849309e-01 -2.75169760e-01 -4.36771542e-01
4.57091480e-01 -1.89622149e-01 -4.37262833e-01 7.40245461e-01
1.88892230e-01 3.37869674e-01 1.06982100e+00 -8.43216121e-01
-1.01372325e+00 -3.68529767e-01 1.32276133e-01 4.54723030e-01
-5.91510773e-01 1.29198954e-02 -9.24676776e-01 -9.53749955e-01
1.07182346e-01 -6.27137184e-01 1.25085756e-01 6.11474775e-02
-2.55049229e-01 3.39203812e-02 1.02245998e+00 -6.79468751e-01
1.51129663e+00 -2.44677925e+00 2.10735515e-01 1.33729041e-01
1.32944956e-01 2.60932267e-01 -2.43760809e-01 2.62243152e-01
9.36260354e-03 -2.45076850e-01 -1.20231383e-01 -2.12788749e-02
-3.39439750e-01 1.04633048e-01 -1.76912159e-01 5.06364882e-01
9.90815833e-02 6.67801917e-01 -1.00627100e+00 -8.00606310e-01
4.84560400e-01 3.69379550e-01 -5.07619381e-01 3.28541994e-01
1.70455664e-01 4.71912891e-01 -7.30736315e-01 8.09377670e-01
5.09530425e-01 -1.93305433e-01 1.25792921e-01 -4.92652982e-01
-2.31133208e-01 -1.06766604e-01 -1.32812428e+00 1.74065423e+00
-2.05785230e-01 4.83242691e-01 2.28076428e-02 -1.25695074e+00
5.82293153e-01 3.05905133e-01 1.21513665e+00 -7.63489187e-01
-4.55139019e-02 -9.66391414e-02 -2.62462765e-01 -7.34308958e-01
3.99899095e-01 1.98364168e-01 9.40675139e-02 4.36427742e-01
3.25184800e-02 5.40745914e-01 4.57004815e-01 2.72402525e-01
1.16391838e+00 3.06749437e-02 6.37535095e-01 1.89436506e-02
7.18136847e-01 -3.37101042e-01 9.19325829e-01 5.62602162e-01
-6.12609088e-01 6.65559947e-01 3.19816589e-01 -2.78667659e-01
-7.55946875e-01 -8.59387696e-01 1.46813422e-01 9.82120097e-01
6.89406812e-01 -6.38666213e-01 -6.04630351e-01 -7.08842158e-01
-7.90497214e-02 2.57387131e-01 -4.45742995e-01 -3.09273273e-01
-7.05184579e-01 -2.88698554e-01 3.52448225e-01 4.33174610e-01
6.17890418e-01 -9.15698409e-01 -6.70987368e-01 2.64156252e-01
-3.87827367e-01 -1.24537003e+00 -1.00725448e+00 -4.08374220e-01
-8.42617154e-01 -1.36232018e+00 -7.08694756e-01 -4.77756679e-01
5.42044520e-01 9.31856513e-01 7.03932285e-01 -6.67880252e-02
-2.99424946e-01 6.89725041e-01 -6.32700443e-01 2.88430452e-01
-2.38848981e-02 -3.51346344e-01 3.02913017e-03 6.10423326e-01
4.35028493e-01 -5.63593268e-01 -7.52617836e-01 7.29397416e-01
-1.03952360e+00 8.28478932e-02 6.92994714e-01 7.94811845e-01
7.72931159e-01 3.06880951e-01 3.00406039e-01 -3.63232404e-01
7.62537718e-02 -6.49203897e-01 -1.93138257e-01 3.73008966e-01
-1.63311884e-01 -2.41826937e-01 6.61357760e-01 -5.99330664e-01
-1.05450416e+00 8.34491178e-02 3.84571493e-01 -9.77704823e-01
-4.18266021e-02 4.33603734e-01 -5.19770026e-01 1.27771020e-01
2.85459822e-03 7.41191030e-01 8.36661831e-02 -2.80564636e-01
6.38503581e-02 6.60810709e-01 4.23891246e-01 -2.45250285e-01
4.10121202e-01 5.82007110e-01 -1.92503035e-01 -1.19521511e+00
-5.96576273e-01 -9.07647133e-01 -6.81923926e-01 -5.77898502e-01
8.27684939e-01 -9.80230570e-01 -6.45944774e-01 5.38120627e-01
-6.42386615e-01 1.27358451e-01 -2.72318900e-01 7.97501922e-01
-5.95841944e-01 7.53817797e-01 -4.45170313e-01 -8.79782319e-01
-7.64207691e-02 -1.24987495e+00 9.95352745e-01 1.86327755e-01
-9.36506763e-02 -7.72019267e-01 -2.53201276e-01 5.42273521e-01
-1.60237458e-02 -1.22044690e-01 4.06269729e-01 -3.49048138e-01
-7.89269984e-01 -1.60508379e-01 -2.34188035e-01 3.86364251e-01
5.41264951e-01 1.58563666e-02 -5.43144643e-01 -3.01625669e-01
1.81615591e-01 -2.39292100e-01 9.48695421e-01 5.57196140e-01
1.31543088e+00 -3.88889462e-01 -2.51157194e-01 5.43205917e-01
1.16493702e+00 4.93013650e-01 6.69570386e-01 2.36145444e-02
8.88889432e-01 4.11675155e-01 1.16172814e+00 9.15845692e-01
2.09335327e-01 9.40023720e-01 2.41508394e-01 3.48395735e-01
-9.10450239e-03 -2.65750945e-01 6.25420153e-01 8.76396060e-01
-2.11755186e-01 -1.75330803e-01 -4.18000817e-01 4.20265943e-01
-2.10434175e+00 -1.55900955e+00 3.53081614e-01 2.31407237e+00
5.76944590e-01 -3.08277663e-02 3.69026393e-01 1.69034868e-01
9.87883031e-01 6.15167797e-01 -7.93546915e-01 3.39449197e-01
-1.85024783e-01 -5.36809802e-01 2.73285925e-01 7.20391497e-02
-1.22356200e+00 5.36274672e-01 4.88579416e+00 1.20469081e+00
-8.52165043e-01 -2.61290818e-02 5.55632293e-01 -4.69741642e-01
-6.46689236e-02 -7.79186636e-02 -7.39337802e-01 9.69568968e-01
4.88148063e-01 -2.52956212e-01 3.47485244e-01 8.85555625e-01
5.13167858e-01 -2.69147813e-01 -1.01823401e+00 1.26984477e+00
3.49626571e-01 -1.20350122e+00 2.68743992e-01 -6.44711182e-02
5.93222320e-01 -3.64660323e-01 -1.02446556e-01 5.28726280e-02
-3.36559981e-01 -3.80131602e-01 7.40553558e-01 6.75889313e-01
6.56497777e-01 -8.13198686e-01 4.14816111e-01 3.59801918e-01
-1.69842589e+00 -2.99119562e-01 -2.62321651e-01 6.37485161e-02
2.65726864e-01 6.39655948e-01 -1.26255408e-01 6.68806911e-01
5.23239672e-01 1.43291783e+00 -2.46302351e-01 9.49847817e-01
8.09897557e-02 4.03534353e-01 -9.20030773e-02 1.69645622e-01
3.17302078e-01 -4.10090208e-01 7.99458265e-01 1.08480346e+00
3.63461494e-01 3.93978506e-01 5.39780915e-01 4.83680964e-01
1.69502735e-01 2.13771299e-01 -5.16978681e-01 -1.91634670e-01
5.81967413e-01 1.00135803e+00 -6.50280237e-01 -3.96669090e-01
-7.51327157e-01 1.00567198e+00 7.46692866e-02 3.66797507e-01
-9.83627021e-01 -1.66160166e-01 9.66701746e-01 -1.28330886e-01
5.01598477e-01 -6.55967891e-02 1.22731656e-01 -1.48501480e+00
3.96080405e-01 -9.63274717e-01 6.86586916e-01 -4.37362164e-01
-1.05806887e+00 1.94550365e-01 3.58393155e-02 -1.90805542e+00
-2.44244590e-01 -1.50047109e-01 -4.19824421e-01 2.78157651e-01
-1.19900954e+00 -8.95033836e-01 -3.75558406e-01 8.37321997e-01
1.12300396e+00 -2.15159848e-01 1.80924565e-01 4.88731474e-01
-8.02460074e-01 6.52350247e-01 8.70926231e-02 1.13327131e-01
5.92605650e-01 -6.86488152e-01 -2.25900009e-01 1.01683211e+00
3.28813374e-01 3.98359895e-01 3.02659035e-01 -6.89652920e-01
-1.76560557e+00 -1.03758466e+00 5.56369960e-01 -3.16772866e-03
5.99471390e-01 1.01385519e-01 -9.41692948e-01 3.30320895e-01
-3.11243355e-01 6.85434416e-02 8.47360015e-01 -2.67690986e-01
-2.18333051e-01 -2.75736809e-01 -7.98825860e-01 5.80348670e-01
1.23922884e+00 -7.77950048e-01 -4.27869737e-01 1.33634299e-01
5.80194831e-01 -4.03480940e-02 -8.39713573e-01 4.01842654e-01
7.77290225e-01 -1.17482316e+00 1.01305854e+00 -4.78831589e-01
5.68770051e-01 -4.65087622e-01 -4.30954456e-01 -9.13868487e-01
-3.02151591e-01 -5.74391425e-01 -3.59516233e-01 1.33924508e+00
-1.31481454e-01 -1.62836894e-01 7.02606738e-01 6.58986568e-01
-9.37587842e-02 -7.87105978e-01 -9.37642574e-01 -8.34821999e-01
-8.11887205e-01 -6.68433607e-01 4.02947098e-01 9.16112006e-01
2.60645837e-01 -7.09341168e-02 -7.62294888e-01 1.41479865e-01
5.55395722e-01 4.10334826e-01 7.40701556e-01 -6.93287969e-01
-3.81538659e-01 -4.26936477e-01 -7.31224597e-01 -1.32567620e+00
1.23607427e-01 -4.60196793e-01 -8.16377578e-04 -1.00948703e+00
5.82637846e-01 -3.53725851e-02 -6.18379712e-01 2.52029806e-01
-2.92411536e-01 2.35940009e-01 3.92512828e-01 6.76012814e-01
-1.08657384e+00 7.12242246e-01 1.14737749e+00 -1.75764397e-01
-2.93050468e-01 -4.30865213e-02 -2.82585382e-01 8.02556753e-01
5.00577331e-01 -1.54504135e-01 -6.98049009e-01 -1.59303740e-01
-2.88475335e-01 4.48166072e-01 2.25067392e-01 -1.11739099e+00
3.47061515e-01 -5.23530006e-01 2.56071806e-01 -6.18551731e-01
3.53965253e-01 -8.65879655e-01 1.16514206e-01 3.24188232e-01
-4.31227237e-01 -2.91650474e-01 -2.44925603e-01 9.84961033e-01
-6.54109180e-01 -4.04789224e-02 6.94128454e-01 3.59847322e-02
-1.18633246e+00 6.24123394e-01 -3.17376494e-01 9.79964286e-02
1.29212821e+00 -5.88274419e-01 1.68430042e-02 -5.12552202e-01
-5.17624438e-01 2.88493574e-01 4.41101819e-01 5.17335474e-01
8.57621193e-01 -1.54446924e+00 -4.45323437e-01 3.40838760e-01
3.41218680e-01 -4.54737961e-01 8.47745538e-01 1.10157895e+00
4.67164256e-02 2.57902652e-01 -2.49326900e-01 -6.67215168e-01
-1.71106184e+00 7.15209663e-01 -3.93225923e-02 -1.34830505e-01
-7.00301290e-01 5.69049716e-01 5.48113525e-01 4.22930956e-01
2.22368568e-01 -2.53514796e-01 -4.60057765e-01 4.01203305e-01
1.04334831e+00 5.38836896e-01 -3.82598311e-01 -1.08822894e+00
-3.30935150e-01 6.65154934e-01 -9.09258202e-02 1.85210407e-01
1.10692620e+00 -5.59975386e-01 1.98120758e-01 5.42721152e-01
1.42406309e+00 -2.31178284e-01 -1.68273401e+00 -4.44301933e-01
-2.18347952e-01 -1.04077911e+00 3.49928737e-02 -1.41385272e-01
-1.21943986e+00 6.35711432e-01 2.57860243e-01 -6.96476316e-03
1.52307117e+00 -2.44934946e-01 9.96918738e-01 1.91888765e-01
4.07459736e-01 -1.29557192e+00 4.55652952e-01 1.92604214e-01
5.53603947e-01 -1.19263744e+00 1.85108911e-02 -5.61285317e-01
-7.16271341e-01 1.05126727e+00 4.90321308e-01 2.26963401e-01
6.57897770e-01 7.24834949e-02 -3.29071909e-01 -7.01392591e-02
-6.29768312e-01 -1.88026994e-01 4.82418299e-01 2.75342226e-01
-6.41555246e-03 -8.30821022e-02 -2.70912677e-01 6.01082504e-01
4.51133251e-01 1.64264277e-01 1.91642314e-01 9.95724916e-01
-5.46484232e-01 -6.93183720e-01 -2.00631410e-01 4.77348089e-01
-3.91825080e-01 6.92546964e-02 -2.31913015e-01 5.61672091e-01
8.20424557e-02 1.03478217e+00 2.04854593e-01 -6.50215328e-01
8.35505500e-02 -2.14579567e-01 1.86484694e-01 -2.46460974e-01
-6.85196891e-02 2.62536347e-01 -5.32752089e-02 -1.02893651e+00
-7.42824912e-01 -9.02026892e-01 -1.12139571e+00 -1.72150522e-01
-2.69761443e-01 1.15926363e-01 1.39242068e-01 1.16483903e+00
5.24051607e-01 3.07748348e-01 1.04155803e+00 -8.25605989e-01
-4.64426100e-01 -6.58022821e-01 -8.73953521e-01 7.84826636e-01
1.53583393e-01 -7.19748795e-01 -4.08909321e-01 3.43666583e-01] | [8.618916511535645, 0.5240207314491272] |
81ec73eb-d795-4b6c-ab96-b59a434d91a7 | iqpp-a-benchmark-for-image-query-performance | 2302.10126 | null | https://arxiv.org/abs/2302.10126v3 | https://arxiv.org/pdf/2302.10126v3.pdf | iQPP: A Benchmark for Image Query Performance Prediction | To date, query performance prediction (QPP) in the context of content-based image retrieval remains a largely unexplored task, especially in the query-by-example scenario, where the query is an image. To boost the exploration of the QPP task in image retrieval, we propose the first benchmark for image query performance prediction (iQPP). First, we establish a set of four data sets (PASCAL VOC 2012, Caltech-101, ROxford5k and RParis6k) and estimate the ground-truth difficulty of each query as the average precision or the precision@k, using two state-of-the-art image retrieval models. Next, we propose and evaluate novel pre-retrieval and post-retrieval query performance predictors, comparing them with existing or adapted (from text to image) predictors. The empirical results show that most predictors do not generalize across evaluation scenarios. Our comprehensive experiments indicate that iQPP is a challenging benchmark, revealing an important research gap that needs to be addressed in future work. We release our code and data as open source at https://github.com/Eduard6421/iQPP, to foster future research. | ['Josiane Mothe', 'Radu Tudor Ionescu', 'Eduard Poesina'] | 2023-02-20 | null | null | null | null | ['content-based-image-retrieval'] | ['computer-vision'] | [ 8.31933692e-02 -7.35757291e-01 -4.79236692e-01 -3.60541940e-01
-1.43150997e+00 -6.86371207e-01 6.85819566e-01 6.16557859e-02
-3.96257132e-01 1.86031833e-01 1.24018915e-01 -2.44786181e-02
-3.72322768e-01 -4.48408604e-01 -6.48279905e-01 -5.57161033e-01
-1.54832443e-02 3.02361161e-01 3.71636659e-01 -5.78835718e-02
5.67096472e-01 4.97679293e-01 -1.92897594e+00 3.88468742e-01
6.48932815e-01 1.30044639e+00 3.35552841e-01 8.13493252e-01
3.50889683e-01 7.21274555e-01 -6.57993913e-01 -4.01873797e-01
4.50489223e-01 -3.23224999e-02 -1.05155933e+00 -2.22729638e-01
7.06805408e-01 -3.87507498e-01 -6.18015349e-01 8.91060591e-01
7.48096824e-01 -1.82005018e-02 7.03587472e-01 -1.65435064e+00
-8.87917161e-01 4.14208658e-02 -5.70191562e-01 4.30659682e-01
3.05598825e-01 2.80461431e-01 1.21070087e+00 -1.00779212e+00
6.36866212e-01 1.18656719e+00 1.25585347e-01 1.06523387e-01
-9.68183577e-01 -7.30686188e-01 -1.23813920e-01 6.61443949e-01
-1.81710303e+00 -3.69153857e-01 4.74539936e-01 -3.91522318e-01
7.05119848e-01 5.09020627e-01 2.31134266e-01 8.41435313e-01
8.31543282e-02 1.22730899e+00 1.04881680e+00 -3.86857718e-01
-9.76930484e-02 2.20146533e-02 2.67661065e-01 4.22566086e-01
-1.26319379e-01 1.74876839e-01 -5.73268950e-01 -2.74184644e-01
4.57669139e-01 5.63431997e-03 -4.93487686e-01 -4.02422458e-01
-1.29568970e+00 7.15758324e-01 7.36343682e-01 6.82453439e-02
-3.44557911e-01 3.28561813e-01 4.24215615e-01 5.08261502e-01
4.20163572e-01 5.78374386e-01 -5.61294377e-01 -6.06777593e-02
-9.28402364e-01 4.73929584e-01 6.90084696e-01 1.19688559e+00
9.27431524e-01 -6.10956669e-01 -6.42724693e-01 1.05450392e+00
1.10983960e-01 7.68538356e-01 3.94438833e-01 -1.00879014e+00
2.55326182e-01 4.25673723e-01 2.76167899e-01 -1.20991647e+00
3.60493474e-02 -1.57265052e-01 -4.42082673e-01 -4.16972876e-01
8.07816014e-02 4.42673236e-01 -9.16941941e-01 1.25594103e+00
5.61465174e-02 1.48491472e-01 -9.59573016e-02 1.14587927e+00
1.05759263e+00 9.77787197e-01 2.40610197e-01 -1.20687656e-01
1.37571573e+00 -1.46053660e+00 -3.17198694e-01 -7.39006847e-02
6.93215549e-01 -1.15251613e+00 1.44993734e+00 2.64580995e-01
-7.91502178e-01 -6.22674644e-01 -6.71111941e-01 -2.30559632e-01
-5.30839324e-01 5.17619014e-01 3.30350459e-01 1.39317378e-01
-1.10132718e+00 2.36781448e-01 -4.34587836e-01 -4.22454804e-01
1.51754692e-01 7.60331973e-02 -1.92506611e-01 -4.84100610e-01
-1.18122339e+00 7.95306325e-01 4.28047389e-01 -2.04155028e-01
-1.16085625e+00 -9.64581549e-01 -3.95939410e-01 5.54115847e-02
5.41168094e-01 -4.76778924e-01 1.36968338e+00 -7.64059007e-01
-1.14472449e+00 1.19916153e+00 -2.76669674e-02 -2.22696275e-01
2.55292118e-01 -4.89131808e-01 -3.47353697e-01 4.92251605e-01
1.63162723e-01 1.09446704e+00 7.97297001e-01 -1.50877023e+00
-7.05626428e-01 -3.22156042e-01 2.80006409e-01 3.14687967e-01
-3.16374660e-01 2.85062701e-01 -1.39768016e+00 -5.58921635e-01
-2.20807940e-01 -1.11310542e+00 6.87741628e-03 8.81863087e-02
-3.69521737e-01 -4.80069697e-01 5.86808205e-01 -4.05524790e-01
1.38780379e+00 -2.37015200e+00 -7.42454007e-02 4.68324423e-02
-5.76969385e-02 3.60423625e-01 -6.88706577e-01 6.58490062e-01
9.34518278e-02 1.68326780e-01 1.01794071e-01 -1.16049595e-01
-4.96570580e-02 1.42807141e-02 -3.79882574e-01 3.43532294e-01
6.81264624e-02 1.19969201e+00 -8.47601891e-01 -8.48126292e-01
1.80383667e-01 4.01666492e-01 -2.57280320e-01 3.23032528e-01
-2.43080482e-01 1.83209926e-01 -6.61638200e-01 1.07380283e+00
6.30598724e-01 -6.17489159e-01 -4.15034443e-01 -4.63173836e-01
5.45839127e-03 -1.54303968e-01 -8.53768706e-01 1.72084880e+00
-3.06399912e-01 6.57236040e-01 -2.06351161e-01 -7.29660869e-01
7.42118657e-01 -2.44827624e-02 6.08844221e-01 -1.14878333e+00
-9.04623568e-02 1.91392303e-01 -3.36182266e-01 -5.87378025e-01
8.25612366e-01 7.36460865e-01 -9.02272612e-02 2.15139613e-01
6.27789721e-02 -2.88753867e-01 4.35799122e-01 3.02793980e-01
1.01050115e+00 -2.16596782e-01 7.98320621e-02 -3.33860785e-01
5.17704070e-01 3.33552212e-01 2.03809321e-01 9.88719821e-01
-4.38845664e-01 9.74187076e-01 2.75440097e-01 -1.44541144e-01
-9.93787766e-01 -9.43319857e-01 -3.26165289e-01 1.56771207e+00
5.70835054e-01 -5.75774312e-01 -5.36732495e-01 -5.49455404e-01
1.37134761e-01 3.37213367e-01 -5.40547967e-01 -7.76246935e-02
-2.44033456e-01 -6.15511358e-01 4.85862255e-01 2.59954095e-01
5.53558588e-01 -1.11269593e+00 -4.50759262e-01 -2.84417152e-01
-3.48951936e-01 -1.25366497e+00 -6.75145030e-01 -4.06819403e-01
-5.61348200e-01 -1.29598284e+00 -1.13576901e+00 -8.07912827e-01
3.68363589e-01 8.31024051e-01 1.46888113e+00 4.00823325e-01
-4.51880813e-01 9.87912476e-01 -7.32976913e-01 -3.78177673e-01
-5.02603166e-02 2.24034131e-01 -3.00466835e-01 -2.52684474e-01
4.39046800e-01 4.69163507e-02 -1.27718496e+00 7.77892590e-01
-1.06024373e+00 -8.06168616e-02 8.30780327e-01 7.18938649e-01
1.14685678e+00 -6.43522143e-02 2.34086245e-01 -5.28548777e-01
7.49186933e-01 -5.17125964e-01 -7.31330633e-01 7.07820773e-01
-8.46978068e-01 -1.78198427e-01 6.61501288e-02 -4.14636791e-01
-6.44170165e-01 -2.82313339e-02 2.12426722e-01 -7.57409096e-01
-1.84781943e-02 5.71913719e-01 7.23336935e-02 -1.67613402e-01
6.67028606e-01 4.77488458e-01 -2.17950836e-01 -3.93814206e-01
4.36042041e-01 7.34514713e-01 5.44754505e-01 -6.37639046e-01
5.60938597e-01 2.58846462e-01 -2.45367959e-01 -7.93793380e-01
-9.02001500e-01 -1.15550101e+00 -4.06636506e-01 -1.92762092e-01
4.56542432e-01 -1.14824128e+00 -6.55968070e-01 3.25647295e-01
-9.92015421e-01 -2.90797293e-01 4.63253586e-03 2.01762065e-01
-4.90677595e-01 4.24425960e-01 -6.46939814e-01 -4.21504587e-01
-7.24180758e-01 -1.50600445e+00 1.60753942e+00 2.50950813e-01
1.71466112e-01 -4.78676826e-01 1.05385870e-01 6.86306715e-01
4.58842993e-01 -2.25351542e-01 6.47383094e-01 -5.64810574e-01
-7.84812450e-01 -3.88291538e-01 -8.34390581e-01 3.28337610e-01
-2.81923562e-01 1.50906980e-01 -8.34124804e-01 -4.78520006e-01
-3.23092997e-01 -6.70896173e-01 9.70676601e-01 2.72680581e-01
1.65277147e+00 -1.04891829e-01 -3.23830366e-01 3.71774048e-01
1.65777433e+00 -1.96607448e-02 9.09919918e-01 5.18075287e-01
3.91022474e-01 4.21315014e-01 1.20140266e+00 2.59588152e-01
5.47647238e-01 7.77773321e-01 4.40990180e-01 -1.97508596e-02
-1.15819499e-01 -1.39413476e-01 -3.39702121e-03 6.21926188e-01
1.09941453e-01 -4.73902375e-01 -1.21551406e+00 6.38845325e-01
-1.83445883e+00 -5.15341461e-01 -7.48689771e-02 2.25590324e+00
7.05702603e-01 -3.67786527e-01 -1.22110723e-02 -1.50780037e-01
4.31946546e-01 1.35228023e-01 -4.95555758e-01 1.82214469e-01
-1.05316555e-02 4.22700308e-02 5.76032460e-01 1.28927991e-01
-1.17348480e+00 1.10226154e+00 6.48883009e+00 1.19882882e+00
-1.15527308e+00 1.39866084e-01 8.80614281e-01 2.38969754e-02
1.02334686e-01 -4.29934449e-02 -8.11024725e-01 4.49260056e-01
8.77622068e-01 -3.34944040e-01 2.38311678e-01 1.02257061e+00
-8.50440413e-02 -3.05943549e-01 -1.18992412e+00 1.35608792e+00
1.63428426e-01 -9.06534195e-01 1.71834156e-01 -5.39318472e-02
6.42908931e-01 4.76803571e-01 3.24077219e-01 5.66802323e-01
-5.25510050e-02 -8.90448093e-01 3.20936143e-01 6.59154713e-01
6.60387635e-01 -4.49889004e-01 6.57359362e-01 2.09226653e-01
-1.04027975e+00 -8.40993524e-02 -5.99785328e-01 5.61851978e-01
-2.51935720e-01 3.12880039e-01 -4.54353690e-01 5.03611088e-01
1.08184850e+00 6.42299175e-01 -1.07925487e+00 1.35064065e+00
4.89733294e-02 4.74061638e-01 -2.10423991e-01 2.86313798e-02
2.20238045e-01 1.86532661e-01 2.16759980e-01 1.35662663e+00
2.59928793e-01 2.17500076e-01 2.67407298e-01 5.40992141e-01
-4.08791691e-01 5.86977839e-01 -3.43008339e-01 -4.89127077e-02
6.48175955e-01 1.33618343e+00 -5.53218782e-01 -4.07532990e-01
-3.97842795e-01 9.64175344e-01 5.06418012e-02 6.13744020e-01
-8.03325653e-01 -1.84580043e-01 5.26667178e-01 7.85722286e-02
1.98663741e-01 7.60123506e-02 2.03520685e-01 -1.14056802e+00
8.32582191e-02 -9.73741591e-01 4.37548161e-01 -1.12337255e+00
-1.40692449e+00 3.81719470e-01 3.59575868e-01 -1.44518292e+00
3.79399285e-02 -5.88512778e-01 -1.81633592e-01 5.61113358e-01
-1.94864202e+00 -1.13094306e+00 -6.72736049e-01 5.21721005e-01
7.48243392e-01 -5.08070327e-02 6.50553465e-01 4.87414062e-01
-3.77929181e-01 6.92727268e-01 3.07918042e-01 1.64300084e-01
1.12924564e+00 -8.21446061e-01 -4.63589048e-03 3.77798676e-01
3.28348756e-01 3.44157428e-01 6.30804896e-01 -1.23248599e-01
-1.76116645e+00 -1.06947827e+00 5.59620619e-01 -7.25106776e-01
7.04083443e-01 3.05784568e-02 -9.25004661e-01 3.02408725e-01
2.32680902e-01 3.91228497e-01 5.96235514e-01 -3.44850644e-02
-4.70507652e-01 -2.70104021e-01 -8.31082582e-01 4.35285687e-01
7.52385259e-01 -5.97737372e-01 -1.27572581e-01 7.05870211e-01
8.76782715e-01 -3.43452573e-01 -1.23346162e+00 7.45317459e-01
6.53256774e-01 -7.52414167e-01 1.34968066e+00 -2.04554081e-01
5.43685973e-01 -1.22946732e-01 -4.86608505e-01 -9.22362506e-01
-2.42528215e-01 2.81250826e-03 -6.75630942e-02 1.10936618e+00
2.81535715e-01 -2.07205653e-01 7.90530741e-01 5.08555353e-01
1.64054960e-01 -9.74434316e-01 -5.42833507e-01 -9.67924654e-01
1.99069589e-01 -3.34417850e-01 4.95772272e-01 7.28237033e-01
-5.70646167e-01 2.05113024e-01 -2.73250222e-01 1.14869401e-01
4.31764215e-01 3.47958416e-01 1.06005836e+00 -8.78500998e-01
-1.31290361e-01 -5.39829314e-01 -5.11619866e-01 -1.35060680e+00
5.27541116e-02 -7.44574010e-01 7.73713887e-02 -1.31671596e+00
6.66009843e-01 -4.51165855e-01 -5.95781028e-01 3.59944165e-01
-2.46953398e-01 4.36942697e-01 4.32199210e-01 9.06538069e-01
-1.19440150e+00 6.18329585e-01 1.35455716e+00 -4.63393688e-01
4.13351804e-02 -8.19862261e-02 -4.90422130e-01 2.63590753e-01
7.67384052e-01 -4.17748481e-01 -5.88879108e-01 -5.25697947e-01
7.06954598e-02 1.19069636e-01 4.93516415e-01 -8.75274420e-01
3.08878064e-01 -2.01550931e-01 8.00544024e-02 -8.39021921e-01
3.73636842e-01 -7.55044401e-01 -1.40302747e-01 2.31888101e-01
-5.08546054e-01 3.10692102e-01 1.40207544e-01 5.31720996e-01
-6.74405217e-01 -1.86545670e-01 5.67068517e-01 2.47310344e-02
-1.16216385e+00 6.40827179e-01 2.05197081e-01 1.71940774e-01
9.51304853e-01 1.52724028e-01 -6.35489643e-01 -4.12486523e-01
-2.87164897e-01 7.32564807e-01 4.53118354e-01 7.02486157e-01
7.69847631e-01 -1.29768229e+00 -8.41866195e-01 -3.30159515e-01
9.46501672e-01 -3.08341026e-01 2.84596413e-01 6.73426747e-01
-6.79190874e-01 7.16947973e-01 8.30491632e-02 -1.00172722e+00
-1.53363836e+00 8.54259610e-01 1.21933945e-01 -3.07584733e-01
-2.53476024e-01 7.73548365e-01 3.32815796e-01 -2.29817316e-01
4.51044917e-01 -1.55608393e-02 -1.36193722e-01 -5.92768192e-02
6.32163942e-01 2.56421715e-01 6.80311173e-02 -7.54003644e-01
-2.15078384e-01 5.29853404e-01 -3.06316733e-01 1.46600977e-01
1.09776652e+00 -2.62789577e-01 -1.51665822e-01 2.50434220e-01
1.66069949e+00 -6.27610326e-01 -7.75615513e-01 -4.56031322e-01
1.04640916e-01 -8.62361073e-01 1.24039277e-01 -8.29927027e-01
-1.20043719e+00 7.25540400e-01 1.06648934e+00 -1.14685027e-02
1.48105323e+00 2.85255700e-01 5.87274492e-01 6.66424274e-01
3.98340285e-01 -1.24394429e+00 4.55262244e-01 3.99595499e-01
1.33096468e+00 -1.71153104e+00 1.84551105e-01 -3.98620933e-01
-5.67052364e-01 6.90433919e-01 7.48651087e-01 -1.00704931e-01
8.63195300e-01 -3.29194278e-01 8.36685747e-02 -3.78890038e-01
-9.13946331e-01 -2.97875226e-01 6.61835194e-01 1.60105765e-01
4.80036408e-01 -2.22945656e-03 -3.24161083e-01 -1.21261354e-03
-6.55595288e-02 1.21091299e-01 -2.11476665e-02 1.02737379e+00
-2.47680634e-01 -1.00222707e+00 -3.08803976e-01 5.83863854e-01
-4.85621065e-01 -2.79367387e-01 -4.28666830e-01 8.58775556e-01
-3.62975389e-01 8.68645430e-01 -9.90034342e-02 -3.66735250e-01
4.80953068e-01 -2.16953591e-01 3.97486299e-01 -3.64271522e-01
-3.07975411e-01 5.78489900e-02 -2.41286978e-01 -7.83090472e-01
-5.65844119e-01 -3.34962130e-01 -6.82179689e-01 -2.17080817e-01
-5.41348636e-01 9.95427445e-02 7.46906519e-01 4.59384739e-01
6.71267033e-01 1.18840523e-01 7.15358675e-01 -7.31718242e-01
-4.82341826e-01 -9.88993406e-01 -3.43372613e-01 6.03629708e-01
2.68651396e-02 -5.91066182e-01 -2.24966675e-01 3.45865823e-02] | [10.801763534545898, 0.8486325740814209] |
2c08d1b1-8707-4985-b01a-cfd61b1ffb2d | graph-classification-gaussian-processes-via | 2306.03770 | null | https://arxiv.org/abs/2306.03770v1 | https://arxiv.org/pdf/2306.03770v1.pdf | Graph Classification Gaussian Processes via Spectral Features | Graph classification aims to categorise graphs based on their structure and node attributes. In this work, we propose to tackle this task using tools from graph signal processing by deriving spectral features, which we then use to design two variants of Gaussian process models for graph classification. The first variant uses spectral features based on the distribution of energy of a node feature signal over the spectrum of the graph. We show that even such a simple approach, having no learned parameters, can yield competitive performance compared to strong neural network and graph kernel baselines. A second, more sophisticated variant is designed to capture multi-scale and localised patterns in the graph by learning spectral graph wavelet filters, obtaining improved performance on synthetic and real-world data sets. Finally, we show that both models produce well calibrated uncertainty estimates, enabling reliable decision making based on the model predictions. | ['Xiaowen Dong', 'Pietro Liò', 'Yin-Cong Zhi', 'Felix L. Opolka'] | 2023-06-06 | null | null | null | null | ['graph-classification', 'gaussian-processes'] | ['graphs', 'methodology'] | [ 3.31179678e-01 3.77577364e-01 1.52137280e-01 -4.92943600e-02
-5.20177186e-01 -5.32833636e-01 9.00466800e-01 6.65316701e-01
-2.81768609e-02 3.47575575e-01 1.84238821e-01 -2.09631383e-01
-4.45612520e-01 -1.06399071e+00 -5.04465938e-01 -9.12135005e-01
-5.70864618e-01 2.78823256e-01 2.09890857e-01 -1.89820807e-02
4.46305722e-02 5.78776598e-01 -1.09188867e+00 -1.62764102e-01
6.82815254e-01 1.03412926e+00 -8.70986432e-02 7.73714244e-01
-5.90374358e-02 5.31942844e-01 -4.84749973e-01 -3.67543280e-01
4.41672504e-02 -3.33553612e-01 -5.75250506e-01 1.39181495e-01
3.68941389e-03 5.87000370e-01 -3.41173142e-01 1.22313857e+00
3.87771845e-01 7.73087516e-02 1.00257993e+00 -1.03300142e+00
-5.25649011e-01 6.85055792e-01 -3.44500691e-01 1.82836816e-01
4.24459189e-01 -1.34406433e-01 1.18620181e+00 -4.98978466e-01
4.62441355e-01 1.08906233e+00 1.12510288e+00 5.62715670e-03
-2.05793548e+00 -2.06360698e-01 -9.59889404e-03 2.10339129e-01
-1.36002755e+00 -2.78562814e-01 1.19422686e+00 -5.73428571e-01
6.09561563e-01 6.08078651e-02 5.91388583e-01 1.27496541e+00
2.94447780e-01 2.80297667e-01 1.16978860e+00 -5.87924004e-01
4.25188512e-01 -1.42653421e-01 1.31602421e-01 9.57507789e-01
1.35913923e-01 8.09509605e-02 -2.83964366e-01 -5.41227221e-01
5.36941409e-01 -4.02481973e-01 -6.25480950e-01 -6.82247400e-01
-1.10223114e+00 1.04090750e+00 4.74634618e-01 6.02339685e-01
-6.49350762e-01 4.33661550e-01 2.22482041e-01 3.41293961e-01
7.76350558e-01 2.30947509e-01 -1.08458318e-01 1.43978149e-01
-8.93641114e-01 -1.91547915e-01 1.10312641e+00 5.72984874e-01
8.89101267e-01 1.85978144e-01 -1.38176933e-01 6.47525966e-01
6.35427415e-01 4.40429181e-01 1.44668669e-01 -6.58591330e-01
1.39277559e-02 3.25990945e-01 -3.99813056e-01 -1.21324599e+00
-8.11672151e-01 -6.63830996e-01 -1.02054405e+00 1.28798261e-01
4.46835548e-01 -2.75464475e-01 -9.08439517e-01 1.65020823e+00
6.68579191e-02 5.57625651e-01 -6.03445955e-02 3.64293694e-01
5.05020857e-01 6.41911685e-01 1.94628481e-02 -3.40795636e-01
1.03123379e+00 -3.13429832e-01 -4.07932520e-01 1.11205809e-01
2.37484097e-01 -4.74285722e-01 7.05581784e-01 4.42295998e-01
-7.56765425e-01 -3.94846678e-01 -1.07561803e+00 7.45371103e-01
-2.91966319e-01 -1.13791972e-02 6.90394223e-01 9.71153319e-01
-1.38761890e+00 1.11602080e+00 -8.17043245e-01 -4.96006906e-01
-8.12358316e-03 1.01278588e-01 -3.16032559e-01 2.17820659e-01
-1.17530978e+00 6.67808652e-01 3.66426528e-01 3.54194343e-02
-4.89642531e-01 -6.07172489e-01 -9.66564298e-01 3.57968271e-01
3.37997347e-01 -5.70877731e-01 8.62890780e-01 -8.80333364e-01
-1.66867721e+00 5.36045432e-01 1.96342468e-01 -6.99231327e-01
3.00848365e-01 5.33515632e-01 -5.32381237e-01 3.52789879e-01
-2.05809265e-01 -4.32137549e-02 1.43864572e+00 -1.21281219e+00
-1.01612940e-01 -2.06140816e-01 -3.88730288e-01 -3.99159372e-01
-1.36041448e-01 -1.72143951e-01 -1.09673552e-01 -7.63819754e-01
9.92258713e-02 -8.76658440e-01 -3.39701056e-01 -4.23038810e-01
-3.39369237e-01 -3.05513203e-01 4.94248509e-01 -7.46816993e-01
9.03158128e-01 -2.17951131e+00 2.86689997e-01 8.31655800e-01
4.73497212e-01 -5.09607717e-02 -1.80358887e-01 9.09913003e-01
-2.23695025e-01 1.74253806e-01 -1.99598521e-01 -2.77692258e-01
2.07257271e-01 1.31892458e-01 -7.84319118e-02 7.97186434e-01
5.00532329e-01 8.19902241e-01 -7.37844765e-01 -1.24283336e-01
2.38354787e-01 5.64932764e-01 -2.03087434e-01 -1.43640950e-01
-4.24872674e-02 1.84286281e-01 -2.64727861e-01 2.93266833e-01
3.99616748e-01 -2.71331728e-01 1.45261258e-01 -2.08921492e-01
3.54675919e-01 -1.93355963e-01 -1.27966702e+00 1.43237960e+00
-4.27852035e-01 5.98133385e-01 4.20107782e-01 -1.52527428e+00
1.03813148e+00 3.55821133e-01 6.58419251e-01 -2.79002398e-01
1.29508421e-01 3.05784377e-03 4.73213494e-02 8.33724663e-02
-4.60166410e-02 -2.17835173e-01 -1.69373572e-01 1.94860399e-01
6.58319354e-01 -2.16218606e-01 1.73614472e-01 1.28836319e-01
1.49776495e+00 -5.88260703e-02 3.45160544e-01 -5.77086866e-01
6.76723182e-01 -4.77555484e-01 1.89675197e-01 7.22604990e-01
-1.23208098e-01 4.05515939e-01 9.64916885e-01 -1.64581448e-01
-7.73880899e-01 -1.38305128e+00 -7.50400592e-03 8.40901554e-01
-1.16967425e-01 -5.70427835e-01 -7.11297035e-01 -6.71078742e-01
1.31285056e-01 6.42065346e-01 -5.76777160e-01 -4.64122802e-01
-2.35740885e-01 -9.00229394e-01 3.69017035e-01 2.56342798e-01
6.33316189e-02 -8.49209130e-01 -5.38534746e-02 2.46415317e-01
2.45537952e-01 -1.10251844e+00 -1.92200661e-01 5.20381749e-01
-5.10452747e-01 -9.98609543e-01 -5.25265098e-01 -4.74886388e-01
2.88659722e-01 -2.58350283e-01 1.09530985e+00 -2.22360864e-01
-1.83642015e-01 9.59188342e-01 -3.72118354e-01 -3.40409964e-01
-6.47689879e-01 -2.81703994e-02 4.13543452e-03 4.39921737e-01
-5.63814975e-02 -1.01003540e+00 -4.51593734e-02 -7.32742473e-02
-6.87541008e-01 -3.97115231e-01 7.68446684e-01 7.42256820e-01
3.66532713e-01 5.16999543e-01 4.36138541e-01 -6.63186908e-01
1.19264197e+00 -5.18893778e-01 -6.14758074e-01 1.81647018e-01
-8.39798808e-01 5.21362603e-01 8.74470532e-01 -3.96635234e-01
-5.97068787e-01 6.80796802e-02 6.54827580e-02 -3.29128206e-01
-2.74414152e-01 8.30181956e-01 -5.55785932e-02 -4.04973328e-01
7.64797926e-01 3.29352498e-01 9.77364704e-02 -2.36790046e-01
5.76740801e-01 1.64999083e-01 5.33335686e-01 -6.08786523e-01
1.19115901e+00 2.30571970e-01 5.33676744e-01 -1.22425115e+00
-1.31424293e-01 -5.03976882e-01 -5.69204628e-01 -3.68475229e-01
8.14388752e-01 -5.87699294e-01 -5.54541409e-01 3.01947236e-01
-8.86928320e-01 -2.66949594e-01 -2.90852398e-01 4.58571851e-01
-7.60340512e-01 7.18694448e-01 -6.82295501e-01 -9.55534995e-01
-1.67894900e-01 -6.99437857e-01 1.12007678e+00 3.22104618e-02
-9.78505146e-03 -1.43984675e+00 2.06306905e-01 -3.21783304e-01
4.63248938e-01 6.36562169e-01 1.16703784e+00 -9.35331285e-01
-2.81456381e-01 -3.94080341e-01 -1.32089123e-01 2.83809602e-01
-3.22351977e-02 1.63954198e-01 -7.62770534e-01 -2.95785636e-01
6.17151558e-02 1.31368011e-01 8.75530422e-01 5.27486563e-01
7.50222027e-01 5.19519160e-03 -2.89893985e-01 4.98883486e-01
1.46702218e+00 -1.69160634e-01 3.14052075e-01 -2.60162592e-01
6.11350417e-01 5.59971809e-01 -7.78101832e-02 1.34226963e-01
2.89003216e-02 6.85162604e-01 2.74420202e-01 4.01570424e-02
-6.54710084e-02 -2.33022749e-01 3.59065920e-01 7.38926053e-01
-2.73129046e-01 -1.34258673e-01 -1.14245903e+00 2.74536729e-01
-1.90486765e+00 -1.03178668e+00 -3.85780394e-01 2.11713934e+00
3.43390524e-01 3.13495487e-01 3.27047497e-01 2.15906262e-01
8.60897779e-01 3.06978434e-01 1.21421544e-02 -2.59128690e-01
6.92106038e-02 5.10801673e-01 7.54011512e-01 5.26450694e-01
-1.20333159e+00 4.57451642e-01 6.63937807e+00 9.59239244e-01
-1.04227114e+00 -2.21362263e-02 2.54239917e-01 5.96447647e-01
-2.43643150e-01 2.87815351e-02 -1.15695931e-01 4.42876548e-01
1.43095529e+00 -4.28667337e-01 5.58112442e-01 6.04474008e-01
1.56332955e-01 1.16424084e-01 -9.77817535e-01 9.26057458e-01
1.12146974e-01 -1.08727634e+00 -3.44354182e-01 2.48890758e-01
4.00082022e-01 -1.72551461e-02 -2.10578695e-01 3.58723819e-01
5.12742460e-01 -1.10331011e+00 5.36584079e-01 7.93332458e-01
2.68480569e-01 -7.42827833e-01 6.96335495e-01 3.89064282e-01
-1.48169100e+00 -3.25971134e-02 -2.03885049e-01 1.38204277e-01
1.42108575e-01 9.04514313e-01 -8.99335623e-01 9.91361618e-01
2.97652304e-01 6.68181121e-01 -7.71597445e-01 1.13939333e+00
-2.31810078e-01 8.85578930e-01 -3.86699706e-01 -3.94131094e-02
1.13916971e-01 -5.08808017e-01 7.55937099e-01 1.24671316e+00
3.69887948e-01 -4.33539033e-01 4.09513563e-01 9.63907301e-01
1.52216688e-01 1.26125231e-01 -7.03829587e-01 -2.78884023e-01
8.62472951e-02 1.54236472e+00 -1.22577810e+00 6.87525868e-02
-4.13961202e-01 8.50786030e-01 3.06480974e-01 5.13194263e-01
-6.93775117e-01 -3.58088017e-01 2.08714858e-01 6.35470003e-02
3.77116531e-01 -4.64592040e-01 1.79244503e-01 -9.97856557e-01
-5.67005016e-02 -4.20509189e-01 4.19652849e-01 -4.88864154e-01
-1.66013205e+00 4.27336097e-01 1.40095994e-01 -8.24242055e-01
-4.90451217e-01 -8.12247396e-01 -9.06757951e-01 9.39119935e-01
-1.13665605e+00 -1.33494341e+00 -2.51735181e-01 5.18424153e-01
8.68029136e-04 -1.25241190e-01 8.52700293e-01 -7.93301389e-02
-3.32171738e-01 3.47609222e-02 1.28888577e-01 1.54916167e-01
2.76407868e-01 -1.98360252e+00 5.46491027e-01 9.13782895e-01
7.58990288e-01 1.52056932e-01 9.79771733e-01 -5.41831911e-01
-1.40003419e+00 -8.98313105e-01 4.03203219e-01 -2.83573985e-01
1.22660244e+00 -6.20342791e-01 -8.25492382e-01 3.84784579e-01
9.78332832e-02 3.92772168e-01 5.99133492e-01 3.51139903e-01
-2.34039471e-01 6.03838302e-02 -9.61043060e-01 2.31214687e-01
8.74274552e-01 -7.11682677e-01 -5.32270670e-01 3.13744813e-01
4.70428199e-01 1.68354452e-01 -1.21004903e+00 3.90220225e-01
7.82007575e-02 -9.24854159e-01 9.61572409e-01 -4.58272129e-01
-1.20247021e-01 -1.42702922e-01 -3.91829200e-02 -1.91127002e+00
-7.00885117e-01 -9.45094287e-01 -2.19055802e-01 1.31010306e+00
4.64625627e-01 -8.70579302e-01 8.09718728e-01 7.69374445e-02
1.07993141e-01 -3.87833685e-01 -1.04663229e+00 -8.18364799e-01
-2.25039888e-02 -4.28809434e-01 2.21685439e-01 6.74600363e-01
7.51007944e-02 6.33815765e-01 -9.67009291e-02 4.04148996e-01
8.32424521e-01 1.04731441e-01 5.48102558e-01 -1.76835537e+00
-6.02553844e-01 -7.58604586e-01 -1.13047290e+00 -3.01435560e-01
5.37418664e-01 -1.03699708e+00 -4.95345779e-02 -1.40246105e+00
-2.70998836e-01 -1.63931519e-01 -1.65219188e-01 -5.26587926e-02
-1.29113451e-01 1.14145011e-01 5.23436069e-02 -1.87621891e-01
-3.10127914e-01 5.09858251e-01 4.13608402e-01 -1.48704454e-01
-7.59316087e-02 3.24275076e-01 -3.42245579e-01 7.05879986e-01
6.36976182e-01 -1.58926412e-01 -4.46976125e-01 3.82437974e-01
5.44982664e-02 1.52374238e-01 5.12426615e-01 -1.34148979e+00
9.85733569e-02 2.09554002e-01 3.92235875e-01 -9.38827097e-02
1.51022911e-01 -9.33437884e-01 3.98790181e-01 5.06831467e-01
-1.86088100e-01 -8.30160528e-02 -4.48728800e-02 1.24195540e+00
-1.62140325e-01 -3.03673655e-01 7.04540610e-01 1.41439885e-01
-5.74872315e-01 2.01442048e-01 -4.63570446e-01 -2.26035848e-01
1.15162432e+00 8.32887143e-02 -9.20723155e-02 -7.69329250e-01
-1.17850041e+00 -3.57012711e-02 3.39096874e-01 9.09232497e-02
2.38486528e-01 -1.35766673e+00 -7.15199649e-01 1.56043500e-01
1.03148006e-01 -5.81700861e-01 -1.22972272e-01 9.80979800e-01
-3.90189230e-01 1.99749768e-01 1.98624849e-01 -7.86183178e-01
-1.23380470e+00 8.78802955e-01 4.43364352e-01 -3.48964512e-01
-6.53717160e-01 4.53189015e-01 -2.94944078e-01 -3.13768953e-01
-1.65486056e-02 -5.10891080e-01 -2.26729959e-01 1.79657429e-01
1.03554972e-01 1.44657835e-01 2.58987188e-01 -7.85869181e-01
-2.04980835e-01 5.97669244e-01 7.38002181e-01 -1.34449331e-02
1.29457664e+00 -4.64810897e-03 -1.38988674e-01 5.57939589e-01
1.16761649e+00 4.50382560e-01 -1.09341645e+00 -2.91824430e-01
5.65690815e-01 -4.21704613e-02 1.53391168e-01 -2.74758279e-01
-8.06708872e-01 6.23527050e-01 4.25735116e-01 1.20307136e+00
1.14308918e+00 2.31058791e-01 8.54813531e-02 2.78434932e-01
4.08104748e-01 -8.54394317e-01 -1.49122402e-01 1.84705153e-01
6.49635196e-01 -8.30945432e-01 -1.01959541e-01 -8.15434098e-01
-2.50342667e-01 1.43852913e+00 -2.32338727e-01 -5.15374243e-01
1.02791965e+00 7.69035444e-02 -3.09694499e-01 -3.93337905e-01
-5.56823969e-01 -4.88415867e-01 5.20119488e-01 1.05736005e+00
9.27475095e-02 1.88960701e-01 -1.86278433e-01 4.87739056e-01
-1.99260309e-01 -5.24366915e-01 4.96423930e-01 4.77228642e-01
-3.88909638e-01 -1.11786151e+00 -4.43016917e-01 5.87482870e-01
-3.31786782e-01 6.80024996e-02 -4.85574841e-01 6.97043955e-01
-3.52273852e-01 9.71645892e-01 -2.24888593e-01 -3.31946522e-01
2.97797620e-01 4.44928944e-01 5.21233797e-01 -5.68025053e-01
-2.55607337e-01 7.94465691e-02 3.44216287e-01 -5.87942064e-01
-3.66392851e-01 -6.20893419e-01 -7.67909825e-01 -1.28948078e-01
-4.47896838e-01 3.58593374e-01 7.14429438e-01 7.85655022e-01
1.75601766e-01 6.92995965e-01 6.15561843e-01 -1.09227049e+00
-7.86046386e-01 -9.74821925e-01 -1.03320777e+00 3.33543867e-01
2.44656280e-01 -6.68334901e-01 -5.00758171e-01 -1.19845890e-01] | [6.992751121520996, 5.3981242179870605] |
d7600ce1-5b67-4e7b-a87e-51f193f45c2e | from-argumentation-mining-to-stance | null | null | https://aclanthology.org/W15-0509 | https://aclanthology.org/W15-0509.pdf | From Argumentation Mining to Stance Classification | null | ['Stan Matwin', 'Parinaz Sobhani', 'Diana Inkpen'] | 2015-06-01 | null | null | null | ws-2015-6 | ['subjectivity-analysis'] | ['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] | [-7.234838485717773, 3.840644359588623] |
dbff3537-cbce-415f-b1ba-97be52a31f1c | zscrgan-a-gan-based-expectation-maximization | 2007.12212 | null | https://arxiv.org/abs/2007.12212v3 | https://arxiv.org/pdf/2007.12212v3.pdf | ZSCRGAN: A GAN-based Expectation Maximization Model for Zero-Shot Retrieval of Images from Textual Descriptions | Most existing algorithms for cross-modal Information Retrieval are based on a supervised train-test setup, where a model learns to align the mode of the query (e.g., text) to the mode of the documents (e.g., images) from a given training set. Such a setup assumes that the training set contains an exhaustive representation of all possible classes of queries. In reality, a retrieval model may need to be deployed on previously unseen classes, which implies a zero-shot IR setup. In this paper, we propose a novel GAN-based model for zero-shot text to image retrieval. When given a textual description as the query, our model can retrieve relevant images in a zero-shot setup. The proposed model is trained using an Expectation-Maximization framework. Experiments on multiple benchmark datasets show that our proposed model comfortably outperforms several state-of-the-art zero-shot text to image retrieval models, as well as zero-shot classification and hashing models suitably used for retrieval. | ['Vinay Kumar Verma', 'Saptarshi Ghosh', 'Kripabandhu Ghosh', 'Anurag Roy'] | 2020-07-23 | null | null | null | null | ['cross-modal-information-retrieval'] | ['miscellaneous'] | [ 3.28201801e-01 -3.28572571e-01 -4.14719135e-01 -3.45828176e-01
-1.60260427e+00 -5.59498608e-01 9.81444955e-01 1.00058727e-01
-2.45536655e-01 2.17501760e-01 -5.11201248e-02 1.44702584e-01
-2.06372350e-01 -8.32154930e-01 -6.49953842e-01 -7.43997693e-01
4.69201028e-01 1.06209981e+00 3.10149372e-01 -3.45851928e-01
5.11712074e-01 -1.70120418e-01 -1.94061387e+00 2.98023105e-01
2.62180090e-01 1.13105667e+00 3.24012309e-01 5.94172418e-01
-6.11514039e-02 5.48499823e-01 -4.89348441e-01 -3.80496651e-01
2.61825085e-01 -5.39807856e-01 -9.00199950e-01 1.65410172e-02
6.42636478e-01 -5.97424030e-01 -4.50940132e-01 1.02736235e+00
5.95332742e-01 2.84839094e-01 1.04941154e+00 -1.25806546e+00
-1.00021529e+00 2.60820210e-01 -3.63415837e-01 -1.87304914e-01
5.96918643e-01 -1.63102403e-01 1.33645916e+00 -1.07042480e+00
9.16475892e-01 1.12353742e+00 -1.03347301e-01 6.23810828e-01
-1.05447602e+00 -4.71258283e-01 -3.16081643e-01 3.90136600e-01
-1.77838326e+00 -4.69058901e-01 6.59357429e-01 -3.34805369e-01
6.73496366e-01 1.56717986e-01 3.45844537e-01 9.89234090e-01
1.45356879e-01 9.23515737e-01 6.60731792e-01 -8.27156007e-01
4.34688836e-01 3.94158602e-01 3.06273311e-01 2.64555365e-01
-1.70372352e-02 -2.16484040e-01 -5.18737495e-01 -4.45451230e-01
2.43722588e-01 2.58740038e-01 -1.26720101e-01 -9.19788361e-01
-9.49931443e-01 1.03632820e+00 3.40261519e-01 3.12827855e-01
-1.48499086e-01 1.14619575e-01 4.60228980e-01 3.12378436e-01
2.79990643e-01 1.52197361e-01 1.56673938e-01 3.20297658e-01
-1.16185522e+00 7.69154802e-02 7.91458726e-01 1.36286759e+00
9.49249864e-01 -4.45059866e-01 -5.04108489e-01 9.50015008e-01
4.41030502e-01 8.00313950e-01 7.57614374e-01 -4.37385350e-01
2.36570179e-01 3.08359802e-01 1.60947442e-01 -7.79578447e-01
4.75041389e-01 3.58326845e-02 -5.50086737e-01 -4.61504608e-01
-2.74057835e-01 6.93202198e-01 -1.23801959e+00 1.52042782e+00
2.82864988e-01 8.62900689e-02 3.77064317e-01 9.55590725e-01
8.75604749e-01 8.25089037e-01 -1.18193276e-01 -1.61223561e-01
1.34530747e+00 -1.04226112e+00 -6.57875478e-01 -1.77401468e-01
6.65952802e-01 -9.27607298e-01 1.13021231e+00 3.84778753e-02
-9.56900775e-01 -5.62758267e-01 -1.05250061e+00 -1.71358213e-01
-6.56574130e-01 -3.38201001e-02 1.11135185e-01 5.52049458e-01
-1.05147028e+00 7.58944452e-02 -4.99383479e-01 -9.98361945e-01
-1.66350350e-01 6.94911405e-02 -2.84909159e-01 -6.99727893e-01
-1.27390075e+00 6.79239392e-01 7.02704668e-01 -3.15781504e-01
-1.42945230e+00 -7.59422630e-02 -8.26965749e-01 3.68593693e-01
4.68540698e-01 -6.91384375e-01 1.23860836e+00 -6.75543487e-01
-9.14488554e-01 1.21205330e+00 -1.56164676e-01 -2.16841310e-01
3.01266741e-02 -1.96407408e-01 -2.08375096e-01 7.30487823e-01
3.01825374e-01 8.90217900e-01 1.36311305e+00 -1.41800034e+00
-2.36397341e-01 -4.71161932e-01 1.64884835e-01 2.10929319e-01
-6.02362990e-01 1.91648126e-01 -9.82772827e-01 -3.77663583e-01
-1.03834374e-02 -1.02006447e+00 3.01565498e-01 -5.03681563e-02
-3.41405511e-01 -2.60222495e-01 1.05790663e+00 -1.05806790e-01
1.22390342e+00 -2.14732575e+00 5.61666712e-02 1.65902138e-01
-2.11064145e-01 5.35529777e-02 -2.18904018e-01 9.90508616e-01
3.72156948e-02 -5.32748140e-02 -3.93313989e-02 -4.97673810e-01
2.21849620e-01 3.07324708e-01 -7.06547260e-01 5.44284701e-01
-2.45817006e-01 9.54674423e-01 -7.91578770e-01 -1.07384515e+00
3.47232789e-01 5.19405305e-01 -3.67913157e-01 5.24045527e-01
-2.40750536e-01 -8.23451504e-02 -5.54523349e-01 7.60870337e-01
5.27026534e-01 -5.03708601e-01 -1.55627262e-03 -2.28441283e-02
4.20374215e-01 -1.16076089e-01 -8.40691328e-01 2.06228113e+00
-3.22263926e-01 4.47594434e-01 -4.16212022e-01 -1.03484321e+00
7.52187610e-01 5.83151042e-01 3.01329046e-01 -8.25024843e-01
2.07275167e-01 2.34318748e-01 -5.41423500e-01 -3.31862003e-01
8.29364955e-01 -4.13474031e-02 -4.46601212e-01 8.76157761e-01
4.09168273e-01 -1.39010653e-01 2.18157217e-01 6.24632776e-01
8.49775076e-01 -1.01977577e-02 2.97387809e-01 -7.45353997e-02
3.09937865e-01 3.30924205e-02 -1.30799398e-01 1.32571161e+00
1.97176844e-01 9.83271778e-01 -1.05429158e-01 -1.02702245e-01
-1.12275684e+00 -8.53933394e-01 -3.18926334e-01 1.20159352e+00
5.34960628e-01 -4.39582020e-01 -7.86605775e-01 -3.32271457e-01
-4.17498767e-01 6.16530836e-01 -7.26778507e-01 -5.20382702e-01
1.31605342e-01 -2.90283203e-01 3.09815049e-01 5.06283455e-02
3.47584873e-01 -1.01466322e+00 -7.39964485e-01 -8.08758959e-02
-3.01066279e-01 -1.06095541e+00 -6.87822938e-01 7.30955079e-02
-4.71976131e-01 -8.76418829e-01 -9.84091103e-01 -1.08589280e+00
7.01767921e-01 7.95834005e-01 1.11948967e+00 3.14471871e-01
-5.42150259e-01 1.16861033e+00 -9.12297845e-01 -8.02398473e-02
-2.40463123e-01 1.10286877e-01 -1.51294768e-01 1.95947215e-01
6.56838894e-01 -2.15475224e-02 -8.94699693e-01 4.85990733e-01
-1.63656032e+00 -3.64325672e-01 4.80736226e-01 1.13115346e+00
6.20591879e-01 -8.56045485e-02 1.95822403e-01 -6.11098409e-01
5.16710103e-01 -6.34348869e-01 -3.40323627e-01 1.04319620e+00
-7.43599415e-01 2.15526491e-01 2.04101533e-01 -6.58388317e-01
-7.56717443e-01 -9.57701355e-03 4.37183708e-01 -9.53140020e-01
-2.02637255e-01 5.68870485e-01 -6.45485371e-02 5.24213957e-03
6.51929617e-01 6.24272466e-01 -3.38063836e-01 -3.53921592e-01
4.54363048e-01 1.01270723e+00 3.39302391e-01 -6.67598844e-01
9.12103295e-01 4.68071282e-01 -2.05861926e-01 -7.16520011e-01
-8.50336730e-01 -1.17829978e+00 -7.64205277e-01 -1.67258561e-01
8.86372030e-01 -9.48195219e-01 -1.52136698e-01 3.06556761e-01
-1.06331444e+00 1.37680337e-01 -1.71325251e-01 2.66586393e-01
-6.82735860e-01 3.56650382e-01 -2.74938732e-01 -8.66227448e-01
-6.08907700e-01 -1.14389646e+00 1.79216647e+00 5.20655178e-02
-3.93084548e-02 -9.83322144e-01 3.54960442e-01 4.03082788e-01
4.32937771e-01 -3.40116590e-01 1.01155245e+00 -1.03928328e+00
-6.94475472e-01 -7.24003792e-01 -9.85953286e-02 -8.62925779e-03
-5.22437021e-02 -1.42759055e-01 -1.22572672e+00 -6.57105029e-01
-7.32612144e-03 -1.13558936e+00 8.21933031e-01 1.27956375e-01
8.81323218e-01 -2.98986018e-01 -4.60793763e-01 1.69057518e-01
1.68744540e+00 3.12113643e-01 8.66568208e-01 2.92886138e-01
2.51287818e-01 4.24463570e-01 1.06836367e+00 5.01772702e-01
1.28164455e-01 8.42989862e-01 4.20931667e-01 4.32426557e-02
2.08460554e-01 -5.17292857e-01 -9.82693359e-02 6.00191295e-01
6.68343067e-01 -8.13122869e-01 -6.57496333e-01 7.82920718e-01
-1.87415171e+00 -1.07085180e+00 5.41054189e-01 2.57060719e+00
6.58742189e-01 -2.62700438e-01 -3.49191725e-01 -9.46889818e-02
7.60068119e-01 2.24495634e-01 -5.61301053e-01 -4.24963981e-02
5.96768372e-02 2.34712034e-01 5.03836013e-02 2.47687474e-01
-1.08417094e+00 9.29466307e-01 5.99576855e+00 1.18270648e+00
-1.03430426e+00 1.93442702e-01 3.32377285e-01 -4.90538627e-02
-3.49353224e-01 3.71159673e-01 -8.99455369e-01 1.86261326e-01
8.00680041e-01 -4.65771198e-01 2.22300231e-01 1.03527069e+00
-5.47979653e-01 -1.61498576e-01 -1.42454314e+00 1.18536675e+00
8.87364686e-01 -8.78623426e-01 5.57848692e-01 3.56650501e-02
6.98954225e-01 -2.27773935e-01 2.55674899e-01 6.40721858e-01
6.05881959e-02 -8.03293049e-01 5.50067425e-01 5.75585246e-01
9.72092986e-01 -4.10355657e-01 7.79120922e-01 4.54771131e-01
-9.82748866e-01 3.05311292e-01 -6.08718991e-01 6.47264302e-01
2.40313485e-02 -6.74513951e-02 -8.25105011e-01 6.14499927e-01
6.87253356e-01 5.38022280e-01 -7.22164333e-01 1.02646542e+00
2.54049718e-01 1.55881360e-01 -2.01260120e-01 -7.27644712e-02
2.60089099e-01 8.22850764e-02 3.26176196e-01 9.18903708e-01
6.14144862e-01 -3.51649150e-02 3.11540097e-01 5.61899543e-01
-1.78214654e-01 4.20731217e-01 -9.80425715e-01 -1.75794140e-01
4.85505462e-01 1.25668323e+00 -6.32286429e-01 -6.96742713e-01
-3.83311152e-01 1.24389911e+00 3.76349594e-03 4.34269935e-01
-6.01668000e-01 -5.78781128e-01 3.96007113e-02 -4.22745347e-02
4.87126768e-01 2.96145022e-01 6.81380391e-01 -1.33604205e+00
6.57597035e-02 -7.66091704e-01 7.59229720e-01 -1.39854920e+00
-1.37158310e+00 6.31252527e-01 4.56482828e-01 -1.81280780e+00
-6.26977921e-01 -1.28565371e-01 -3.87511551e-01 5.76209843e-01
-1.46274269e+00 -1.48813701e+00 -3.38851631e-01 9.56276655e-01
6.55600429e-01 -2.94464260e-01 1.14091289e+00 2.34133244e-01
8.16930179e-03 6.18775129e-01 3.71804804e-01 1.46151170e-01
1.14834011e+00 -8.15672338e-01 -1.68661222e-01 5.78925908e-01
4.71040785e-01 9.70574081e-01 5.91327548e-01 -3.78560126e-01
-1.76121950e+00 -9.60952163e-01 6.94429874e-01 -2.48899370e-01
6.97533488e-01 -3.30335885e-01 -9.32929873e-01 7.32194602e-01
5.75197756e-01 9.98125784e-03 9.21217918e-01 -2.82935798e-01
-5.89654863e-01 -1.52525932e-01 -9.62567151e-01 3.68373901e-01
3.06865752e-01 -1.10872138e+00 -7.83543229e-01 7.57876039e-01
7.32450902e-01 -1.99564815e-01 -6.35255158e-01 1.39423147e-01
6.98055089e-01 -5.22946000e-01 9.76482093e-01 -5.94272137e-01
3.33558470e-01 -1.34912878e-01 -6.97117567e-01 -8.52489948e-01
7.02276975e-02 -2.01680258e-01 -7.42848963e-02 1.13264930e+00
1.25936896e-01 -2.00164363e-01 3.58610511e-01 7.15565503e-01
1.56515136e-01 -2.30254516e-01 -9.90466297e-01 -8.24440360e-01
-1.44382194e-01 -2.40129054e-01 3.20335418e-01 8.49174380e-01
-4.68135588e-02 6.75612569e-01 -6.64130688e-01 2.62667090e-02
8.64575803e-01 5.17079592e-01 9.12685275e-01 -1.15039945e+00
-4.19495255e-01 5.84204979e-02 -6.08568668e-01 -1.16570485e+00
2.42855012e-01 -9.72959220e-01 2.56276995e-01 -1.43469226e+00
1.02928531e+00 -9.93386209e-02 -4.12123561e-01 3.56572300e-01
1.54596698e-02 4.29617375e-01 4.05731611e-02 7.03638315e-01
-1.21291316e+00 7.25095570e-01 8.07398736e-01 -5.50131023e-01
8.23620036e-02 -2.87620813e-01 -3.79759431e-01 7.65789067e-03
5.44172943e-01 -7.31225967e-01 -7.37041056e-01 -3.67338002e-01
1.20168172e-01 3.71624023e-01 4.34446603e-01 -7.62294173e-01
7.64468431e-01 7.27885365e-02 -4.72555012e-02 -8.67179811e-01
7.73076653e-01 -9.75477934e-01 3.19171771e-02 1.47588074e-01
-6.62048459e-01 -2.11961046e-01 -2.02588022e-01 7.54543185e-01
-6.29685700e-01 -7.37472236e-01 5.13868511e-01 -1.32820755e-01
-6.17835939e-01 4.04944420e-01 -8.33853781e-02 -3.66303064e-02
1.00458384e+00 -1.76692203e-01 -4.96848643e-01 -8.29562724e-01
-4.66005266e-01 3.64954144e-01 6.73555613e-01 5.05266428e-01
9.19815183e-01 -1.47615612e+00 -3.74716491e-01 5.22081666e-02
1.13113165e+00 -3.65464360e-01 2.91779190e-01 3.21336031e-01
-6.80987462e-02 9.22432363e-01 5.87952696e-02 -8.90834808e-01
-1.40243328e+00 1.11229086e+00 -1.30641758e-01 -2.98119515e-01
-3.85671586e-01 3.86444241e-01 3.69018346e-01 -1.50313303e-01
3.39229941e-01 3.89507204e-01 -4.31552716e-02 6.62473291e-02
6.60361886e-01 -1.88799933e-01 4.36226837e-02 -7.78691113e-01
-1.23974301e-01 8.53029847e-01 -4.43305939e-01 -4.30948853e-01
9.37360525e-01 -2.09340990e-01 -1.22034103e-01 6.85349941e-01
1.69126129e+00 -5.69790483e-01 -4.03338075e-01 -6.69402480e-01
-1.17892921e-01 -6.27188504e-01 5.88213019e-02 -2.90056288e-01
-8.47283542e-01 1.03884780e+00 8.59100461e-01 1.28612489e-01
1.12548983e+00 4.16814208e-01 7.98274040e-01 9.45087314e-01
7.52027869e-01 -1.16238320e+00 4.29092884e-01 2.49268845e-01
6.40179157e-01 -1.58394015e+00 -8.54477733e-02 1.87397480e-01
-4.73232955e-01 9.75630999e-01 3.76902282e-01 4.46594618e-02
6.98913336e-01 -4.00432765e-01 -3.60570811e-02 -4.86980319e-01
-1.01936495e+00 -3.84678751e-01 4.96056765e-01 3.24315220e-01
7.43732080e-02 -3.44955951e-01 5.82330301e-02 5.92255360e-03
1.24027334e-01 -7.21642077e-02 1.73429057e-01 1.24964559e+00
-5.39590240e-01 -1.09482908e+00 -4.36465979e-01 1.86401337e-01
-2.83519566e-01 -3.24215293e-01 -4.55216229e-01 7.07968712e-01
-6.04214132e-01 1.05934548e+00 1.46065667e-01 -2.39305183e-01
-6.92941155e-03 3.96826506e-01 3.16403568e-01 -8.37358892e-01
-1.52113289e-01 3.47675085e-01 -4.83832151e-01 -2.29084253e-01
-5.71393490e-01 -3.33929569e-01 -7.32207239e-01 1.23867027e-01
-6.73007131e-01 4.59723055e-01 6.70834959e-01 8.54828298e-01
2.29992643e-01 -1.14498824e-01 7.73644924e-01 -6.82829618e-01
-5.63928783e-01 -1.11193013e+00 -6.75581932e-01 6.36609197e-01
1.89481243e-01 -6.74691260e-01 -5.06063163e-01 2.32795253e-01] | [11.299522399902344, 0.9635635614395142] |
c941071c-108c-4b09-8be0-e8c2ed7a4e8f | good-robot-now-watch-this-repurposing | null | null | https://proceedings.mlr.press/v164/hundt22a.html | https://proceedings.mlr.press/v164/hundt22a/hundt22a.pdf | "Good Robot! Now Watch This!": Repurposing Reinforcement Learning for Task-to-Task Transfer | Modern Reinforcement Learning (RL) algorithms are not sample efficient to train on multi-step tasks in complex domains, impeding their wider deployment in the real world. We address this problem by leveraging the insight that RL models trained to complete one set of tasks can be repurposed to complete related tasks when given just a handful of demonstrations. Based upon this insight, we propose See-SPOT-Run (SSR), a new computational approach to robot learning that enables a robot to complete a variety of real robot tasks in novel problem domains without task-specific training. SSR uses pretrained RL models to create vectors that represent model, task, and action relevance in demonstration and test scenes. SSR then compares these vectors via our Cycle Consistency Distance (CCD) metric to determine the next action to take. SSR completes 58% more task steps and 20% more trials than a baseline few-shot learning method that requires task-specific training. SSR also achieves a four order of magnitude improvement in compute efficiency and a 20% to three order of magnitude improvement in sample efficiency compared to the baseline and to training RL models from scratch. To our knowledge, we are the first to address multi-step tasks from demonstration on a real robot without task-specific training, where both the visual input and action space output are high dimensional. Code is available in the supplement. | ['Gregory D. Hager', 'Matthew Gombolay', 'Nakul Gopalan', 'Ran Liu', 'Priyanka Hubli', 'Aditya Murali', 'Andrew Hundt'] | 2021-11-08 | null | null | null | conference-on-robot-learning-corl-2021-11 | ['robotic-grasping'] | ['robots'] | [ 4.74321038e-01 -2.83261556e-02 -1.46374553e-01 -1.92619532e-01
-1.02697492e+00 -5.86820722e-01 7.24955916e-01 -8.65189582e-02
-7.92415619e-01 8.98542583e-01 -4.69305068e-02 -2.65194237e-01
-1.36904404e-01 -1.85892537e-01 -7.89932311e-01 -5.19048393e-01
-2.22145945e-01 5.90003669e-01 3.33128065e-01 -2.11682647e-01
5.38854659e-01 5.64910412e-01 -1.70660996e+00 3.70554365e-02
4.15011853e-01 6.12371445e-01 8.85692775e-01 9.54892874e-01
4.71099913e-01 9.20174122e-01 -5.09442031e-01 6.20471776e-01
4.87066805e-01 -4.46378887e-01 -8.32451046e-01 8.52890611e-02
2.33611584e-01 -5.69676995e-01 -5.06663144e-01 6.20096445e-01
5.45190394e-01 7.54872739e-01 7.91145742e-01 -1.52709639e+00
-3.05915494e-02 2.31022313e-02 -3.47373992e-01 1.62573367e-01
4.07199264e-01 8.60014915e-01 7.59375453e-01 -8.95737171e-01
6.38208389e-01 1.26826000e+00 3.55841130e-01 7.02666342e-01
-1.31859183e+00 -5.41510522e-01 3.29103991e-02 4.46155071e-01
-8.21263909e-01 -5.65940082e-01 3.45118999e-01 -4.47472513e-01
1.72836459e+00 -4.64221030e-01 6.49296045e-01 1.12354732e+00
2.27676198e-01 8.86080146e-01 1.03243220e+00 -3.43181908e-01
5.02385259e-01 -3.34985912e-01 -6.31366447e-02 8.34755778e-01
4.41764146e-02 6.11454487e-01 -4.48522985e-01 1.43761262e-01
8.72867167e-01 4.66539003e-02 5.31552639e-03 -9.55129743e-01
-1.52119637e+00 8.97892833e-01 3.83096874e-01 5.63026033e-02
-4.01385218e-01 6.38314188e-01 7.03905165e-01 6.30186200e-01
-3.72390866e-01 1.01630461e+00 -5.55729747e-01 -6.20857418e-01
-5.09163499e-01 4.56000537e-01 7.70444214e-01 1.18900108e+00
8.80488336e-01 2.82765269e-01 4.06117104e-02 5.58449984e-01
-9.73312333e-02 3.81515980e-01 6.72612667e-01 -1.61896408e+00
5.57002485e-01 2.37750888e-01 3.24730843e-01 -2.23232239e-01
-7.06924915e-01 4.33520265e-02 -2.80316591e-01 1.01413083e+00
3.20950627e-01 -2.58547097e-01 -1.01882768e+00 1.49648297e+00
1.97379634e-01 -9.28500853e-03 3.91176671e-01 7.46486425e-01
3.36229146e-01 6.16912186e-01 -8.04204419e-02 -1.44525066e-01
9.24090207e-01 -1.45351851e+00 -5.20001531e-01 -6.12665176e-01
9.24517572e-01 -4.86297339e-01 1.26999283e+00 5.40268779e-01
-9.18769419e-01 -9.30774271e-01 -1.41433620e+00 -7.46664703e-02
-2.91535318e-01 3.00498381e-02 7.23230779e-01 7.83159286e-02
-9.13788140e-01 9.14473474e-01 -1.13039339e+00 -4.92301017e-01
4.00164485e-01 4.97707814e-01 -4.06379789e-01 -3.09769869e-01
-5.75962007e-01 1.48129213e+00 5.57960153e-01 -3.40594888e-01
-1.76108789e+00 -5.69508672e-01 -1.12026858e+00 -1.00893445e-01
7.61481583e-01 -4.68582720e-01 1.92138648e+00 -5.09125650e-01
-1.72923481e+00 3.08736682e-01 -6.66153207e-02 -4.41890776e-01
2.96908587e-01 -4.08196896e-01 1.17400616e-01 1.43806726e-01
2.61373401e-01 9.44779277e-01 9.32658076e-01 -1.11725032e+00
-9.05667365e-01 -1.61601260e-01 1.73862353e-01 6.25700891e-01
1.75287887e-01 -3.39037538e-01 -2.55075991e-01 -1.27535999e-01
-2.31419995e-01 -1.21175802e+00 -4.57488775e-01 1.74194947e-01
1.70040011e-01 -4.44633722e-01 8.72783780e-01 -3.82559240e-01
4.90044773e-01 -2.06433296e+00 6.03921294e-01 -2.96882838e-01
3.32685150e-02 1.99169606e-01 -5.50950468e-01 6.68554008e-01
6.18750751e-02 -4.97094125e-01 -2.44283676e-01 -2.50612199e-01
8.58909488e-02 4.94643986e-01 -1.90856591e-01 4.65016007e-01
1.25130951e-01 9.13337946e-01 -1.36570680e+00 -3.41636002e-01
5.39430976e-01 -2.03260705e-02 -3.63545060e-01 3.98957163e-01
-3.29288632e-01 4.64352071e-01 -2.94512004e-01 4.54104811e-01
2.19288990e-02 -1.06801532e-01 2.14029968e-01 5.10072447e-02
7.06420615e-02 1.53022423e-01 -8.66871238e-01 2.14670229e+00
-7.38114059e-01 8.72126579e-01 -7.98187554e-02 -1.23307669e+00
8.25503707e-01 2.69683272e-01 4.92263168e-01 -8.39517534e-01
-2.58318074e-02 1.75601646e-01 1.84265494e-01 -6.95014954e-01
3.92654002e-01 -1.11313827e-01 -2.72261351e-01 6.01092994e-01
2.78094828e-01 -8.21816564e-01 5.47285318e-01 -2.66527385e-02
1.62675738e+00 7.84172595e-01 5.81315398e-01 2.72176355e-01
-2.66023763e-02 5.07867992e-01 2.03595176e-01 8.48252952e-01
-5.30705631e-01 1.39111891e-01 2.38274291e-01 -3.62702191e-01
-1.35212624e+00 -1.13528788e+00 4.30755854e-01 1.35705924e+00
2.14086473e-02 -5.72196469e-02 -2.47368887e-01 -6.90646470e-01
1.09378725e-01 9.71744180e-01 -4.76431817e-01 -2.72168726e-01
-8.66359591e-01 -7.69871399e-02 2.59908646e-01 6.77532911e-01
3.57491523e-01 -1.65014946e+00 -1.40336704e+00 2.81695634e-01
1.48079455e-01 -1.06724179e+00 -3.14850390e-01 7.15413868e-01
-1.02069509e+00 -1.42043436e+00 -7.29907811e-01 -1.14893651e+00
6.47413135e-01 4.75221306e-01 8.80944312e-01 -3.01738292e-01
-5.84894001e-01 7.00169742e-01 -3.49701196e-01 -2.96488076e-01
-4.85880077e-01 -1.73248023e-01 2.92560011e-01 -9.99283671e-01
1.41489297e-01 -5.70013404e-01 -3.73288482e-01 1.93687350e-01
-4.79287505e-01 1.37622982e-01 1.00677919e+00 1.17418575e+00
3.60978633e-01 -8.12839270e-02 6.47536278e-01 -3.59561116e-01
7.42179811e-01 -3.48889142e-01 -7.39449382e-01 8.73060226e-02
-6.93491578e-01 4.62670922e-01 6.08056009e-01 -7.66830325e-01
-8.53700161e-01 6.74460173e-01 2.81759411e-01 -5.23914576e-01
-2.40956873e-01 2.88584083e-01 4.41635013e-01 1.55401863e-02
8.68108630e-01 3.32802445e-01 1.84817836e-01 -1.38016894e-01
5.03893018e-01 2.60591179e-01 5.25494039e-01 -4.16827828e-01
7.82291532e-01 1.45304754e-01 2.22206503e-01 -6.35851085e-01
-4.26939785e-01 -7.18248188e-01 -8.49304974e-01 -1.47447243e-01
7.59393156e-01 -8.72278392e-01 -9.45471108e-01 1.39179260e-01
-9.56158638e-01 -1.28161919e+00 -4.36336070e-01 7.37003565e-01
-1.57603264e+00 2.28823543e-01 -3.55276465e-01 -7.80866146e-01
1.47719579e-02 -1.16490495e+00 9.36640501e-01 4.28144932e-02
-2.44871199e-01 -5.94614506e-01 2.24029347e-01 -7.23697767e-02
5.88165671e-02 8.77545699e-02 8.39853942e-01 -4.86641467e-01
-4.45821792e-01 1.03992909e-01 -1.80500552e-01 2.31480479e-01
1.51889354e-01 -6.82923257e-01 -5.14191389e-01 -7.42175877e-01
7.52582774e-02 -1.10521209e+00 7.02099264e-01 3.46284389e-01
8.12604189e-01 9.53786969e-02 -5.01861751e-01 2.01126859e-01
1.40044129e+00 6.44023657e-01 4.64223564e-01 5.12174964e-01
3.98474693e-01 4.82033193e-01 1.30058765e+00 4.96265739e-01
9.94890630e-02 7.34310269e-01 5.13315678e-01 1.43037513e-01
-1.76361054e-01 -3.82955402e-01 6.34949505e-01 3.00614923e-01
1.19822718e-01 1.93086073e-01 -7.38592625e-01 7.06020534e-01
-2.03343534e+00 -1.10179853e+00 4.85265702e-01 2.06028986e+00
6.55385435e-01 2.50147045e-01 2.14497074e-01 -2.94104293e-02
2.80216485e-01 -3.67221758e-02 -1.27584314e+00 -3.55857193e-01
6.39930010e-01 2.92231679e-01 4.88194495e-01 4.25594956e-01
-8.76282394e-01 1.04791331e+00 6.69599819e+00 6.84547603e-01
-6.73515022e-01 -2.55636349e-02 5.95104918e-02 -3.75630140e-01
5.84378839e-01 -6.26526494e-03 -5.62179744e-01 1.91897023e-02
9.37081337e-01 -2.06385404e-01 8.80265892e-01 1.30848646e+00
3.31112415e-01 -5.80630958e-01 -1.72796857e+00 1.17541957e+00
8.56473297e-02 -9.92108107e-01 -5.03666282e-01 -6.83703274e-02
6.58879340e-01 2.24916011e-01 -3.83859575e-02 9.83162820e-01
8.96727145e-01 -1.21052349e+00 3.88415694e-01 2.36628324e-01
8.02926481e-01 -6.03779435e-01 2.61874080e-01 6.65761411e-01
-1.09965074e+00 -5.02673566e-01 -5.33716917e-01 -3.58377188e-01
2.01354608e-01 -3.63853544e-01 -1.52172029e+00 1.29181996e-01
5.18341005e-01 8.02703500e-01 -2.88831055e-01 1.01988959e+00
-2.87656039e-01 1.36919022e-01 -7.82536268e-02 -4.24288124e-01
4.92845863e-01 1.50381982e-01 3.27056408e-01 8.67296159e-01
4.73583847e-01 4.59815264e-02 4.15903687e-01 4.45768446e-01
4.33373749e-01 -4.71439391e-01 -1.03816950e+00 -8.65603164e-02
5.11402309e-01 1.05477428e+00 -5.14064550e-01 -5.10717571e-01
-2.40822449e-01 1.29357433e+00 6.47161007e-01 5.06658554e-01
-7.28831112e-01 -6.66895747e-01 6.15827739e-01 -4.69161212e-01
5.60609996e-01 -8.21959019e-01 -5.86692467e-02 -5.88084400e-01
-7.02053681e-02 -1.06268430e+00 6.55703247e-02 -1.01584518e+00
-1.00710917e+00 2.90915191e-01 1.26985297e-01 -1.51611137e+00
-8.34616721e-01 -8.82636011e-01 -3.88049513e-01 5.83311617e-01
-1.46942949e+00 -7.07872152e-01 -3.27372134e-01 5.55655777e-01
1.35388637e+00 -4.37858790e-01 9.79160607e-01 -4.27420974e-01
-1.84266493e-01 1.76855013e-01 2.59788543e-01 -2.81388283e-01
7.37927675e-01 -1.18656719e+00 4.76513088e-01 3.59310806e-01
-7.93607980e-02 2.59102106e-01 9.03638899e-01 -5.59981227e-01
-1.72166467e+00 -8.18874359e-01 3.15057665e-01 -5.05521238e-01
6.04003012e-01 -2.32709214e-01 -4.40001279e-01 8.21741343e-01
1.17071211e-01 -1.79475367e-01 2.41515726e-01 -2.38223337e-02
-1.67678475e-01 1.48584530e-01 -1.02012634e+00 8.28616142e-01
1.28467381e+00 -4.14372623e-01 -1.01275408e+00 5.99683404e-01
7.57925630e-01 -3.43630821e-01 -6.51096642e-01 2.69602954e-01
5.83419561e-01 -6.34384334e-01 9.95382547e-01 -5.96298933e-01
4.82913613e-01 -2.24363729e-01 -1.21382043e-01 -1.74781537e+00
-3.91658306e-01 -5.53006887e-01 -2.26786360e-01 2.09964976e-01
3.07783723e-01 -3.51281255e-01 5.73941648e-01 1.58619493e-01
-4.84310061e-01 -6.19844377e-01 -7.66636014e-01 -1.18841851e+00
-1.00546688e-01 -3.47623110e-01 3.27119492e-02 4.11087722e-01
4.74538386e-01 5.95739305e-01 -4.01939124e-01 -2.03712642e-01
5.85837424e-01 8.47242475e-02 1.18629515e+00 -9.24149036e-01
-5.06685913e-01 -3.31049889e-01 -7.27396086e-02 -1.34261572e+00
3.26653630e-01 -6.70424104e-01 7.52170563e-01 -1.76434708e+00
5.70931435e-02 -4.10834163e-01 -1.29102245e-01 6.59020603e-01
1.59372687e-01 -1.82293728e-01 4.06765431e-01 2.78658122e-01
-8.95747900e-01 7.26824462e-01 1.52164376e+00 -1.01731956e-01
-4.73186553e-01 6.76336756e-04 -2.60282397e-01 4.89637882e-01
8.91174853e-01 -3.61930639e-01 -7.82713890e-01 -3.44252616e-01
-2.23568752e-01 3.59929800e-01 2.80334473e-01 -1.52588117e+00
3.20105076e-01 -4.58152205e-01 4.74666506e-01 -4.80070114e-01
7.94579685e-01 -7.44451463e-01 -2.69686937e-01 9.72478628e-01
-6.02775514e-01 2.13571966e-01 4.70213622e-01 8.53414476e-01
2.17742637e-01 -5.61946809e-01 6.00435615e-01 -4.33443040e-01
-1.46380579e+00 1.24308378e-01 -8.19969237e-01 -1.97246432e-01
1.38606906e+00 -3.66532832e-01 -1.26651734e-01 -3.91488642e-01
-7.33276188e-01 4.60481763e-01 4.00045365e-01 5.13185501e-01
8.98609400e-01 -1.16522801e+00 -3.90996963e-01 4.50862013e-02
2.29612365e-01 -6.08837977e-02 1.03112303e-01 5.92290759e-01
-2.02983797e-01 6.80204153e-01 -7.04961061e-01 -4.30739522e-01
-1.24619877e+00 8.56108367e-01 3.37425172e-02 -3.85759920e-01
-7.81339765e-01 5.43464124e-01 9.96119007e-02 -6.42600298e-01
5.01761734e-01 -3.07682425e-01 -5.89273199e-02 -2.99484372e-01
3.52357984e-01 4.96077776e-01 -4.51536000e-01 -1.35394081e-01
-1.26728877e-01 3.84279132e-01 -4.41121683e-02 -5.47482312e-01
1.35584855e+00 7.26312101e-02 5.87888837e-01 6.46781266e-01
1.14917982e+00 -8.70560527e-01 -2.07069397e+00 -1.41691580e-01
8.87359083e-02 -3.66063505e-01 -2.01581270e-01 -9.41889584e-01
-3.08004349e-01 1.04408526e+00 7.12348700e-01 -2.80886054e-01
7.83924699e-01 -9.06022042e-02 5.14417231e-01 1.26882732e+00
6.96606636e-01 -1.49127150e+00 1.11918890e+00 8.48823845e-01
1.12181532e+00 -1.42548490e+00 1.09168425e-01 1.55832440e-01
-9.30605829e-01 1.12411582e+00 9.02081013e-01 -4.29255038e-01
1.22535065e-01 1.68641493e-01 -2.73408592e-01 -1.36648551e-01
-9.58722770e-01 -3.74116361e-01 -1.91446275e-01 1.17201734e+00
-1.23820297e-01 -1.48935795e-01 5.11679426e-02 -6.96707889e-02
2.41182130e-02 2.01003268e-01 4.65862453e-01 1.43175304e+00
-9.06949580e-01 -8.32597494e-01 6.65439963e-02 4.74706292e-01
4.44960862e-01 1.79637864e-01 1.27582163e-01 9.41572785e-01
-3.57597411e-01 9.86054003e-01 1.03116468e-01 -3.94779325e-01
3.66023839e-01 2.16873527e-01 9.50491011e-01 -9.80480373e-01
-1.79362699e-01 -1.63612336e-01 1.55224890e-01 -8.60383868e-01
-2.11994395e-01 -7.24037588e-01 -1.69420600e+00 -2.17058249e-02
4.04173024e-02 -2.48003721e-01 8.20804715e-01 1.00736737e+00
2.84128934e-01 7.52783000e-01 5.00111222e-01 -1.62658918e+00
-1.11635363e+00 -1.14552045e+00 -3.05406898e-01 2.14631200e-01
4.60663885e-01 -1.11636710e+00 -2.46402621e-01 3.23223695e-02] | [4.49249792098999, 0.9343538284301758] |
f8bc9661-bcdc-4005-83d6-318d5dbf6fed | mgpfusion-predicting-protein-stability | 1802.02852 | null | http://arxiv.org/abs/1802.02852v2 | http://arxiv.org/pdf/1802.02852v2.pdf | mGPfusion: Predicting protein stability changes with Gaussian process kernel learning and data fusion | Proteins are commonly used by biochemical industry for numerous processes.
Refining these proteins' properties via mutations causes stability effects as
well. Accurate computational method to predict how mutations affect protein
stability are necessary to facilitate efficient protein design. However,
accuracy of predictive models is ultimately constrained by the limited
availability of experimental data. We have developed mGPfusion, a novel
Gaussian process (GP) method for predicting protein's stability changes upon
single and multiple mutations. This method complements the limited experimental
data with large amounts of molecular simulation data. We introduce a Bayesian
data fusion model that re-calibrates the experimental and in silico data
sources and then learns a predictive GP model from the combined data. Our
protein-specific model requires experimental data only regarding the protein of
interest and performs well even with few experimental measurements. The
mGPfusion models proteins by contact maps and infers the stability effects
caused by mutations with a mixture of graph kernels. Our results show that
mGPfusion outperforms state-of-the-art methods in predicting protein stability
on a dataset of 15 different proteins and that incorporating molecular
simulation data improves the model learning and prediction accuracy. | ['Harri Lähdesmäki', 'Markus Heinonen', 'Emmi Jokinen'] | 2018-02-08 | null | null | null | null | ['protein-design'] | ['medical'] | [ 1.85824677e-01 -2.12110072e-01 -1.59323186e-01 -3.08185309e-01
-5.63915312e-01 -5.86858690e-01 2.41501436e-01 7.38687038e-01
-1.46823153e-01 1.22336781e+00 -1.46258786e-01 -4.51953143e-01
7.54859205e-03 -4.98924643e-01 -1.04209387e+00 -1.23152268e+00
7.45752305e-02 8.96298289e-01 8.72861445e-01 -2.83727944e-01
1.87766895e-01 5.86581767e-01 -1.27519667e+00 3.21402326e-02
1.40985215e+00 5.33353448e-01 6.58479989e-01 6.20985985e-01
-5.82578999e-04 2.33654842e-01 -1.51993304e-01 -3.66872579e-01
4.43106405e-02 -1.47637740e-01 -3.69955868e-01 -4.63373512e-01
-1.00594319e-01 4.85550433e-01 -5.40358499e-02 8.39438558e-01
7.20984161e-01 5.95269129e-02 8.19606006e-01 -9.04928267e-01
-7.78531730e-01 1.35067835e-01 -5.78584492e-01 2.74207890e-02
3.97039086e-01 6.87781274e-01 8.16261232e-01 -6.81438804e-01
8.39031100e-01 1.23997939e+00 8.40317249e-01 2.68890291e-01
-2.05226350e+00 -2.17627555e-01 -1.25091359e-01 3.41322213e-01
-1.20411360e+00 -1.32863194e-01 6.37061656e-01 -7.00554371e-01
1.42045987e+00 -1.08812265e-01 5.32926559e-01 8.41838717e-01
6.83119476e-01 3.17410856e-01 1.06150162e+00 -1.89834550e-01
5.06303608e-01 -3.78898561e-01 3.93409997e-01 6.76207125e-01
3.23142737e-01 2.29018733e-01 -6.60367072e-01 -9.08967435e-01
1.79702654e-01 1.00952975e-01 -4.54148918e-01 -1.00811648e+00
-8.14619601e-01 6.85634077e-01 -4.80526276e-02 -2.41461977e-01
-6.08940065e-01 -1.12080626e-01 1.71798304e-01 -3.16503942e-02
2.92549610e-01 4.56199706e-01 -1.24267519e+00 -2.58128583e-01
-7.22449183e-01 5.27302027e-01 8.69093060e-01 5.41297138e-01
8.04000258e-01 -5.10099947e-01 1.15722433e-01 6.89324737e-01
4.78986919e-01 5.26564181e-01 2.32344851e-01 -4.28446293e-01
-2.29231808e-02 6.89715266e-01 5.10628998e-01 -6.30685389e-01
-5.44345379e-01 1.19860053e-01 -3.66687119e-01 2.23110273e-01
6.06483936e-01 3.34239453e-02 -9.89131510e-01 1.86035621e+00
4.43655759e-01 4.19097617e-02 1.37355447e-01 5.86910605e-01
4.34774458e-01 7.78896034e-01 5.64728320e-01 -6.48116410e-01
1.16438293e+00 -5.11865914e-01 -5.32007039e-01 2.56874889e-01
5.61539769e-01 -7.14322507e-01 8.98222387e-01 5.60234189e-01
-8.73106003e-01 -4.11523819e-01 -1.15446389e+00 1.96001232e-02
-4.83464807e-01 4.12735306e-02 5.75357258e-01 4.67134416e-01
-7.12002158e-01 1.26304495e+00 -1.22057557e+00 -5.58928847e-01
2.10383400e-01 7.26017177e-01 -4.89465564e-01 1.66182354e-01
-1.17516267e+00 1.45031953e+00 5.23329198e-01 -4.15886134e-01
-5.46524882e-01 -9.49728310e-01 -5.20133495e-01 6.18868321e-03
2.24416852e-01 -6.48440063e-01 9.27099228e-01 -2.75999993e-01
-1.64927876e+00 4.53815997e-01 -4.91463929e-01 -6.63377762e-01
1.27436474e-01 -1.80742577e-01 -1.72290117e-01 -8.10021162e-03
-3.82420272e-01 4.34604913e-01 4.78459120e-01 -9.66463566e-01
-2.27085352e-02 -4.70872551e-01 -5.69337249e-01 -1.12748452e-01
5.61154425e-01 -1.32701173e-01 -9.73776449e-03 -1.84873968e-01
2.13770181e-01 -1.07197285e+00 -4.81287897e-01 -1.19020328e-01
-1.86703920e-01 -2.91749179e-01 7.60108531e-01 -7.19411373e-01
9.05467868e-01 -1.51855409e+00 6.87035978e-01 3.14239115e-01
4.11469996e-01 5.69175541e-01 9.95415449e-02 8.13679576e-01
-3.96660268e-01 -1.03254512e-01 -4.47382450e-01 3.29121113e-01
-1.56828389e-01 3.33295166e-02 2.58250926e-02 5.23930848e-01
3.02066058e-01 8.63754451e-01 -7.97397256e-01 -2.37319753e-01
3.33505481e-01 7.87452281e-01 -4.03202772e-01 2.54417479e-01
-6.74831808e-01 4.33300346e-01 -4.37773496e-01 4.95627165e-01
8.76095235e-01 -4.62848485e-01 5.39179683e-01 -4.61194724e-01
-2.57782843e-02 -3.21851880e-03 -6.39403641e-01 1.43792033e+00
4.23022330e-01 -2.86083966e-01 -2.40834787e-01 -8.10659111e-01
1.18484366e+00 1.36777848e-01 6.81174397e-01 1.15915373e-01
-1.52211040e-02 4.95707132e-02 3.23491126e-01 -3.72492731e-01
-1.08063422e-01 -3.46224695e-01 2.87343949e-01 2.71526240e-02
2.33620360e-01 -5.11412248e-02 1.70143202e-01 2.12077454e-01
1.26485717e+00 7.99116910e-01 4.82700884e-01 -4.94145602e-01
5.91290295e-01 1.23284206e-01 1.09903741e+00 2.33110979e-01
-5.13487339e-01 2.79140502e-01 8.66496921e-01 -3.97143215e-01
-1.34059238e+00 -1.05972111e+00 -1.53637424e-01 1.09328103e+00
1.24248937e-01 -7.30171859e-01 -6.91953003e-01 -4.22379583e-01
3.68793666e-01 5.13482749e-01 -4.57651198e-01 -4.35088843e-01
-1.39484510e-01 -1.57003188e+00 1.77981958e-01 1.67565078e-01
-2.37645581e-01 -7.15690732e-01 -6.62044808e-02 5.79100609e-01
1.08641073e-01 -5.95903277e-01 -3.28217030e-01 7.03566432e-01
-7.12192059e-01 -1.54457176e+00 -2.89156497e-01 -3.69955480e-01
4.10156459e-01 -3.02523021e-02 8.78038108e-01 -1.58181593e-01
-5.12746155e-01 -3.13077539e-01 -2.07788467e-01 -3.37421566e-01
-7.88485169e-01 -2.16565013e-01 5.53273261e-01 -3.20447862e-01
8.48600984e-01 -8.68847430e-01 -6.05583727e-01 3.58488053e-01
-5.40128469e-01 1.04709983e-01 6.55392230e-01 9.57584202e-01
9.88135993e-01 -3.14088427e-02 5.47826171e-01 -8.22783411e-01
7.22799838e-01 -3.87266845e-01 -7.98448443e-01 5.24908841e-01
-9.94488418e-01 7.17755973e-01 4.04006004e-01 -6.03494763e-01
-1.16472304e+00 6.18042409e-01 -2.94052828e-02 -1.93385139e-01
-1.26340151e-01 4.97808188e-01 -4.87949073e-01 -1.77834108e-01
7.78307259e-01 2.59871393e-01 5.15112579e-01 -6.12559974e-01
1.76378444e-01 1.73447639e-01 2.70869762e-01 -8.76981199e-01
3.43696535e-01 -4.08500731e-02 4.13996696e-01 -6.84424579e-01
-2.54370987e-01 -4.44369733e-01 -9.15801108e-01 1.61230639e-01
7.12720811e-01 -5.66901088e-01 -1.18102777e+00 6.43851578e-01
-8.73107374e-01 -1.56706974e-01 2.99468189e-01 2.85671324e-01
-8.42426419e-01 1.03252566e+00 -8.17992747e-01 -6.49202347e-01
-4.55678999e-01 -1.58778787e+00 8.38123679e-01 9.60242227e-02
-2.84610003e-01 -6.80808723e-01 5.65124691e-01 3.34268957e-01
4.40709889e-02 3.39549690e-01 1.29478514e+00 -7.76042759e-01
-3.19045037e-01 -1.07319340e-01 1.26882538e-01 5.82558177e-02
2.28681058e-01 5.34943998e-01 -4.31820899e-01 -2.58728921e-01
-2.68412143e-01 -3.29167038e-01 9.49932635e-01 7.12192535e-01
7.59839714e-01 2.01651737e-01 -7.71869838e-01 2.11433202e-01
1.43339705e+00 1.71893984e-01 7.43861735e-01 -8.85205343e-02
6.14424288e-01 5.14741838e-01 6.81762218e-01 1.35978714e-01
2.76650824e-02 7.83344924e-01 3.26505035e-01 2.52987713e-01
3.25005472e-01 -4.08056408e-01 4.59143788e-01 4.19299036e-01
-3.93454224e-01 -9.24501941e-02 -9.11801219e-01 -1.13771506e-01
-2.29342341e+00 -9.86339152e-01 -4.59479094e-01 2.29477644e+00
1.34698462e+00 1.13983110e-01 2.26957962e-01 -2.11727545e-01
7.89117336e-01 -4.72244352e-01 -1.12614596e+00 -1.45524740e-01
-4.68412608e-01 2.59056985e-01 7.78971374e-01 5.97130358e-01
-8.46187174e-01 1.05582368e+00 7.32703543e+00 8.72926354e-01
-8.19563568e-01 -1.47495180e-01 3.94069403e-01 1.65022373e-01
9.78354439e-02 2.35631704e-01 -9.62669790e-01 6.25218451e-01
1.31488693e+00 -2.72919178e-01 1.28142163e-01 7.74789155e-01
6.34575009e-01 -2.88933575e-01 -8.45826924e-01 6.13441825e-01
-3.47309738e-01 -1.49863839e+00 -2.64636129e-01 1.48827478e-01
5.61995327e-01 1.37978718e-01 -3.28986496e-01 -1.34900376e-01
7.63859749e-01 -8.29151094e-01 9.74779204e-02 1.06801605e+00
1.88271120e-01 -7.65723526e-01 7.49669433e-01 4.56630498e-01
-7.92257607e-01 3.54305416e-01 -5.80940783e-01 1.04100890e-01
2.18724489e-01 8.00837398e-01 -1.07381737e+00 2.13742331e-01
3.99074942e-01 5.86775362e-01 -5.75532913e-01 8.68725240e-01
-1.09280311e-01 7.36108363e-01 -4.26891863e-01 -5.61560728e-02
-4.85720694e-01 -7.82021105e-01 2.11843476e-01 8.44984353e-01
4.12975065e-02 1.94178686e-01 2.44577721e-01 9.96042132e-01
3.34300607e-01 2.44791612e-01 -2.14355946e-01 -3.09142649e-01
4.61474001e-01 9.59322155e-01 -6.13299072e-01 -1.53183177e-01
-3.01337540e-01 8.15456748e-01 6.26869500e-01 4.07617390e-01
-6.96759522e-01 -7.16225952e-02 1.05364680e+00 2.68756002e-01
3.91895503e-01 -3.64353985e-01 1.80482894e-01 -9.93793070e-01
-3.49102765e-01 -7.04992414e-01 5.31789586e-02 -8.57911944e-01
-1.52964377e+00 -2.61702668e-02 -2.13001326e-01 -4.79011834e-01
1.48612589e-01 -1.00662887e+00 -2.81417698e-01 1.24535930e+00
-1.14310586e+00 -1.02158964e+00 3.90071243e-01 -4.41282941e-03
2.19006941e-01 -6.55557066e-02 8.78917158e-01 -4.20179754e-01
-7.20347106e-01 9.26244631e-02 6.19545400e-01 -7.81656146e-01
1.10409749e+00 -1.42591691e+00 5.40617347e-01 6.15512252e-01
-6.70972407e-01 9.79609489e-01 1.55463684e+00 -1.25878584e+00
-1.58438003e+00 -8.05193126e-01 6.25919461e-01 -7.63303638e-01
8.44328046e-01 -1.30132169e-01 -1.47274613e+00 2.29699060e-01
-2.80185372e-01 -1.37724638e-01 1.11571801e+00 8.02917331e-02
-2.74528682e-01 3.85968417e-01 -1.41920161e+00 3.65131587e-01
8.20381224e-01 -2.50436813e-01 -7.55265415e-01 4.71543938e-01
8.39684188e-01 -1.70434818e-01 -1.39740491e+00 5.27597785e-01
2.98395813e-01 -7.89382279e-01 1.04432392e+00 -1.05307662e+00
6.44123852e-02 -7.82921374e-01 -1.45180076e-01 -1.28740489e+00
-6.88046515e-01 -6.19725883e-01 -6.21123672e-01 7.35566676e-01
6.88752770e-01 -6.54188991e-01 8.67317379e-01 9.92479265e-01
1.42551912e-02 -9.34113681e-01 -5.18351018e-01 -5.75742602e-01
1.90751553e-01 -4.92581651e-02 5.09915948e-01 4.91142064e-01
4.06174451e-01 4.14239615e-01 -3.57611388e-01 1.43926933e-01
6.44555986e-01 -6.96863905e-02 6.41885400e-01 -1.60421932e+00
-4.89685297e-01 -4.43429761e-02 -5.95134079e-01 -6.86021984e-01
2.15912491e-01 -5.32296002e-01 -7.94568844e-03 -1.09199059e+00
5.80769956e-01 -9.15111303e-02 -2.71650106e-01 3.48085731e-01
-5.37863612e-01 -3.72978151e-01 -2.33090803e-01 2.46418580e-01
-4.17601913e-01 6.82082295e-01 9.25560832e-01 5.71830943e-02
-2.67885447e-01 -6.92606568e-02 -5.16852736e-01 4.26631063e-01
9.13672864e-01 -1.43776625e-01 -4.51926649e-01 7.02507317e-01
3.15166622e-01 -8.36477131e-02 -8.83643422e-03 -7.37771451e-01
-8.82644951e-03 -4.16636139e-01 4.62407261e-01 -8.38252187e-01
3.58289063e-01 -4.71115351e-01 6.76840246e-01 6.37309849e-01
-1.21897772e-01 -6.74256608e-02 1.18951306e-01 1.18521762e+00
4.09343481e-01 9.71955713e-03 9.99140322e-01 8.65609124e-02
-2.84290731e-01 4.80370998e-01 -5.02147555e-01 -4.45210546e-01
1.05009532e+00 -1.39030024e-01 -2.18886167e-01 6.29799888e-02
-1.26616919e+00 1.30123690e-01 1.18199980e+00 1.24801055e-01
3.81701410e-01 -8.08827400e-01 -3.45475942e-01 1.58163071e-01
2.26229027e-01 -3.70525211e-01 3.75004597e-02 7.78539598e-01
-8.00619125e-01 5.43391764e-01 -1.99943259e-01 -7.14615226e-01
-1.69026017e+00 1.13979256e+00 2.89266825e-01 -1.97470263e-01
-2.24107742e-01 5.89232445e-01 -1.11272082e-01 -5.20452440e-01
-3.82437795e-01 -2.63885200e-01 1.93253513e-02 -3.76270473e-01
3.28261793e-01 3.46314847e-01 1.68133572e-01 -7.55549908e-01
-2.30074167e-01 4.77546006e-01 -3.83617073e-01 4.13614959e-01
1.46003234e+00 -3.58198211e-02 -2.58341134e-01 3.69666517e-01
6.60870969e-01 -3.71840268e-01 -1.79154778e+00 -2.19843030e-01
9.61843058e-02 -4.05651666e-02 -2.47957304e-01 -9.34627473e-01
-2.25427076e-01 4.89103734e-01 7.42355466e-01 -3.95989448e-01
6.91617370e-01 7.35630020e-02 5.61010897e-01 5.62390208e-01
5.99076390e-01 -1.10759628e+00 -2.88052499e-01 3.32272947e-01
4.95426029e-01 -1.26557052e+00 2.68825561e-01 -6.82344854e-01
-5.61891258e-01 8.47778141e-01 4.42928880e-01 7.17425868e-02
7.43004739e-01 1.35963917e-01 -2.48227879e-01 -2.41257101e-01
-1.14616871e+00 -4.75832894e-02 9.95063931e-02 9.95807111e-01
6.27479970e-01 3.49363655e-01 -7.14671969e-01 8.10090899e-01
8.02333429e-02 7.03249052e-02 2.29021996e-01 1.09769225e+00
-9.30208564e-01 -1.77453804e+00 -3.78811836e-01 4.73094791e-01
-3.26908022e-01 -9.41634923e-02 -7.10734487e-01 1.82386443e-01
4.23185043e-02 7.50216782e-01 -7.49832988e-01 -2.52045661e-01
3.06829903e-02 6.68556631e-01 6.78091347e-01 -3.81234378e-01
-9.78564695e-02 -4.71076965e-02 -1.29099265e-02 -4.90439743e-01
-2.98919678e-01 -9.31872904e-01 -1.49754906e+00 -5.54755151e-01
-7.20791698e-01 5.37877977e-01 6.05162323e-01 7.66963363e-01
9.41794813e-01 1.69952944e-01 1.50105640e-01 -8.24984372e-01
-5.27726889e-01 -1.01312435e+00 -6.51017845e-01 5.52138567e-01
-1.29814312e-01 -1.05615902e+00 -6.28742352e-02 3.71282041e-01] | [4.776124954223633, 5.545172214508057] |
85f2c7f4-db55-4762-bcab-285d4ddb88a5 | navigation-in-urban-environments-amongst | 2110.05205 | null | https://arxiv.org/abs/2110.05205v1 | https://arxiv.org/pdf/2110.05205v1.pdf | Navigation In Urban Environments Amongst Pedestrians Using Multi-Objective Deep Reinforcement Learning | Urban autonomous driving in the presence of pedestrians as vulnerable road users is still a challenging and less examined research problem. This work formulates navigation in urban environments as a multi objective reinforcement learning problem. A deep learning variant of thresholded lexicographic Q-learning is presented for autonomous navigation amongst pedestrians. The multi objective DQN agent is trained on a custom urban environment developed in CARLA simulator. The proposed method is evaluated by comparing it with a single objective DQN variant on known and unknown environments. Evaluation results show that the proposed method outperforms the single objective DQN variant with respect to all aspects. | ['Anne Spalanzani', 'Dominique Vaufreydaz', 'Niranjan Deshpande'] | 2021-10-11 | null | null | null | null | ['multi-objective-reinforcement-learning'] | ['methodology'] | [-2.60732502e-01 -7.82204419e-03 -1.01345904e-01 -3.25182855e-01
-6.42338514e-01 -1.98689222e-01 7.36049354e-01 -4.22337949e-02
-1.28543460e+00 1.39798760e+00 -4.46511656e-02 -6.34450793e-01
-4.20473278e-01 -1.10681927e+00 -6.11994088e-01 -6.45716488e-01
-1.90723523e-01 6.99519813e-01 6.65150583e-01 -1.00839806e+00
4.71711904e-01 2.42041096e-01 -1.97624409e+00 -6.82720304e-01
1.09044564e+00 4.18641806e-01 2.41104051e-01 9.51586962e-01
3.41750383e-01 6.60061955e-01 -7.21314609e-01 -3.58146966e-01
6.77816927e-01 -5.79593778e-02 -6.40332103e-01 -2.12404042e-01
5.38280427e-01 -3.20895404e-01 -3.18391770e-01 1.04326653e+00
1.05347967e+00 7.62818336e-01 6.14605904e-01 -1.71099079e+00
-2.18942210e-01 -1.07659623e-01 -2.99953390e-02 6.71273291e-01
1.30929142e-01 5.50076604e-01 8.35706592e-01 -2.25911602e-01
3.63338530e-01 1.71971416e+00 3.32199335e-01 4.33956116e-01
-9.06778932e-01 -5.12703180e-01 2.72729784e-01 1.00505936e+00
-1.01198530e+00 -9.42554027e-02 2.14948371e-01 -3.30114275e-01
1.26454997e+00 -1.42029360e-01 7.16782510e-01 9.53570247e-01
6.73421979e-01 6.60432220e-01 1.25520730e+00 8.50231424e-02
8.69481087e-01 -3.46379209e-04 1.06244966e-01 6.65694058e-01
4.04743910e-01 1.06382728e+00 -2.29473114e-01 2.17306599e-01
-6.49476796e-02 -7.44372249e-01 7.83825219e-01 -3.30641687e-01
-6.02249384e-01 1.44549549e+00 2.45237425e-01 -4.63201940e-01
-6.19189143e-01 3.16539258e-01 5.01890361e-01 4.44288850e-01
1.26114577e-01 1.95049327e-02 -3.22361618e-01 -4.89412665e-01
-3.94977272e-01 1.02723718e+00 7.75625765e-01 6.64785564e-01
8.55697930e-01 7.01472938e-01 1.59120545e-01 4.47517842e-01
5.90614796e-01 9.64607418e-01 2.75777161e-01 -1.14639807e+00
2.64258116e-01 1.34647593e-01 2.99477696e-01 -8.20333183e-01
-6.07223630e-01 -5.07601559e-01 -2.79323757e-01 1.22764742e+00
3.88208121e-01 -7.91671813e-01 -7.70315230e-01 1.51469266e+00
5.04051268e-01 1.80953324e-01 5.92454731e-01 7.81885028e-01
6.81304812e-01 6.49175704e-01 5.78511834e-01 3.15097779e-01
1.00597310e+00 -1.04816735e+00 -6.93348110e-01 -4.42540675e-01
3.56189072e-01 -4.41463083e-01 4.60203260e-01 4.76552099e-01
-7.64763892e-01 -7.63747811e-01 -1.27575350e+00 2.63016820e-01
-9.01702762e-01 -7.21469462e-01 2.04498380e-01 1.37823093e+00
-1.49035096e+00 1.48925126e-01 -2.32187033e-01 -6.58542812e-01
3.91487241e-01 7.08666146e-01 1.00994101e-02 -4.14455235e-02
-2.00717187e+00 1.45342875e+00 6.33382976e-01 -1.72243431e-01
-1.35610461e+00 -2.40445688e-01 -9.34223592e-01 -5.39360046e-01
6.54324770e-01 -5.44996560e-01 1.30680776e+00 -5.28645098e-01
-1.81998134e+00 4.87492532e-01 8.86244774e-02 -1.03960192e+00
9.02260959e-01 -5.72509281e-02 -6.55699432e-01 -2.50401765e-01
5.90300202e-01 9.59987104e-01 4.62442189e-01 -1.10850406e+00
-1.63786268e+00 -2.64183193e-01 4.83365625e-01 7.91962028e-01
4.69969273e-01 -3.40395510e-01 3.22982401e-01 -1.54617935e-01
-8.69724870e-01 -7.79107988e-01 -8.28157485e-01 -2.75805563e-01
3.73027474e-02 -4.89623606e-01 9.16253507e-01 -3.64207089e-01
7.19920933e-01 -1.42508769e+00 -3.08375686e-01 3.62975389e-01
-9.00964960e-02 4.55435246e-01 -2.13403240e-01 3.05804074e-01
5.11628509e-01 -2.39606440e-01 -5.90267964e-02 2.40830202e-02
3.83902401e-01 7.55406380e-01 5.41519523e-01 4.55461770e-01
1.33256137e-01 7.40311444e-01 -1.16032517e+00 -7.37746060e-01
5.50283611e-01 1.72064856e-01 -3.23116004e-01 7.53936470e-02
-5.26666865e-02 3.14630151e-01 -4.46417540e-01 5.34702361e-01
8.48994017e-01 8.59497488e-01 -3.22896719e-01 5.79899549e-01
-7.11819887e-01 -1.87163249e-01 -1.50725091e+00 9.41664040e-01
-4.95782375e-01 8.73443902e-01 -1.24765351e-01 -1.35089087e+00
9.44798291e-01 6.10127710e-02 2.68535793e-01 -1.74986339e+00
5.95831759e-02 2.98314929e-01 3.95424038e-01 -8.01426113e-01
7.84434140e-01 -2.85314500e-01 -1.67749971e-01 -2.87283305e-02
-3.90907116e-02 -6.95021404e-03 6.09419763e-01 -1.51321694e-01
9.79216158e-01 1.36124462e-01 3.92794460e-01 -7.02409863e-01
9.67043757e-01 3.50904077e-01 5.76572180e-01 1.24550307e+00
-1.26818860e+00 -4.28572804e-01 3.62636328e-01 -5.43877602e-01
-1.02149177e+00 -1.32397091e+00 -1.74550235e-01 1.32364511e+00
5.95246494e-01 3.51267785e-01 -7.18528867e-01 -6.50154173e-01
2.84640193e-01 9.14163053e-01 -6.34437442e-01 -8.04941729e-02
-8.38410854e-01 -8.49541605e-01 4.38228101e-01 1.93585500e-01
7.92356849e-01 -1.18749213e+00 -1.11173987e+00 5.94237685e-01
1.70025721e-01 -8.67451847e-01 2.08351538e-01 1.78629071e-01
-2.72067219e-01 -1.04361558e+00 -6.24850094e-01 -8.68895173e-01
-1.69217765e-01 1.27500534e-01 1.02376115e+00 -2.10492626e-01
-7.94573054e-02 2.19188035e-01 -2.22868875e-01 -8.30922246e-01
-5.32124758e-01 1.33831263e-01 6.67585284e-02 -7.75377527e-02
6.20194793e-01 -2.95687169e-01 -9.39506352e-01 5.45434177e-01
-3.94489974e-01 -5.89150488e-01 6.05078638e-01 6.82472646e-01
1.56994238e-01 4.47537482e-01 1.21718860e+00 -4.30852830e-01
8.87476683e-01 -7.63995945e-01 -1.15625525e+00 -2.94363052e-01
-7.65182614e-01 2.00313717e-01 5.11194766e-01 -2.97602899e-02
-9.45511281e-01 3.03128082e-02 -5.31551063e-01 8.60983312e-01
-4.59937304e-01 7.20055401e-02 -4.85713065e-01 -3.10803771e-01
6.82076871e-01 4.50679995e-02 8.95069316e-02 4.79713194e-02
4.31659132e-01 4.69666272e-01 4.04892594e-01 -3.46762449e-01
8.59413028e-01 4.24975246e-01 3.80717546e-01 -9.90271628e-01
-5.78182079e-02 -4.99730647e-01 -4.93221164e-01 -9.83905375e-01
1.22948515e+00 -7.73290157e-01 -1.17519319e+00 6.36921108e-01
-6.62710786e-01 -5.71025014e-01 -1.06334813e-01 4.08990711e-01
-9.91645873e-01 9.66920108e-02 3.26889902e-01 -9.80565667e-01
1.08541153e-01 -1.52850711e+00 6.04075134e-01 5.36548018e-01
2.99266160e-01 -1.22216916e+00 3.43744248e-01 4.37658995e-01
5.75346291e-01 3.88233453e-01 5.92202067e-01 -5.12448609e-01
-5.17847657e-01 1.02458626e-01 -1.21816166e-01 1.36373028e-01
-2.44623929e-01 -3.43445063e-01 -6.36000752e-01 -1.57139033e-01
-5.12898862e-01 -3.39456648e-01 7.43145764e-01 5.37462294e-01
2.29324371e-01 -1.27435029e-01 -1.71673913e-02 1.12253718e-01
1.76565278e+00 9.42113698e-01 5.47205031e-01 1.32090962e+00
2.93245465e-01 8.45226228e-01 1.04384494e+00 3.10414672e-01
8.07298779e-01 4.69131082e-01 1.15504360e+00 -1.18864574e-01
2.08545834e-01 1.95808142e-01 3.66351157e-01 -1.61795393e-01
9.37520266e-02 -4.61780727e-01 -1.15448916e+00 8.85774434e-01
-2.10265970e+00 -1.20607960e+00 -3.37100178e-01 1.74005926e+00
8.02263170e-02 6.59646571e-01 7.03823984e-01 3.06013048e-01
6.70643270e-01 3.53710018e-02 -8.92205358e-01 -1.20234179e+00
-3.17809552e-01 3.25874120e-01 1.14892673e+00 1.24648452e+00
-1.23803186e+00 1.02262819e+00 6.09029913e+00 7.97928751e-01
-7.20312953e-01 2.55548179e-01 3.72418553e-01 2.54467487e-01
1.25032529e-01 -1.49707973e-01 -9.16125655e-01 4.12976325e-01
1.49959326e+00 -5.23035169e-01 1.20251939e-01 6.65430903e-01
8.43852580e-01 -7.42712677e-01 -3.81942451e-01 4.82874960e-01
-3.04749608e-01 -8.10139239e-01 -2.39383802e-01 3.65074724e-01
7.49669850e-01 2.51545042e-01 3.58267307e-01 8.22662830e-01
1.11145949e+00 -1.18509591e+00 8.68533850e-01 4.65174139e-01
-1.58822141e-03 -1.49021542e+00 1.02071619e+00 3.54037106e-01
-9.44354177e-01 -4.00897473e-01 -2.36266762e-01 -2.22963840e-01
5.41922927e-01 -7.85550624e-02 -8.28094840e-01 5.74081540e-01
5.84688783e-01 5.04458487e-01 -6.88417017e-01 1.50173426e+00
-8.73241946e-03 4.58011866e-01 -3.78739685e-02 -5.75561941e-01
1.25156343e+00 -3.82549733e-01 8.53843212e-01 1.18511796e+00
8.94802064e-02 -2.22112775e-01 3.11254352e-01 1.44877821e-01
5.83422422e-01 2.40152568e-01 -8.17067862e-01 8.39584827e-01
-2.07685772e-02 7.94757366e-01 -3.75038922e-01 -2.67806441e-01
-3.45396072e-01 4.03416216e-01 8.89816042e-03 2.89119035e-01
-8.48304391e-01 -3.96157175e-01 1.00148022e+00 3.74935493e-02
4.47114795e-01 -2.31487080e-01 -2.20108759e-02 -1.15999676e-01
-3.77177089e-01 -6.07677519e-01 1.87677413e-01 -2.89645553e-01
-9.46998119e-01 6.13981426e-01 3.40807885e-01 -1.13360476e+00
-3.94056439e-01 -8.63449454e-01 -4.24417078e-01 9.17596161e-01
-2.06842637e+00 -6.67366922e-01 -1.46356985e-01 5.06074607e-01
8.22074294e-01 -7.30338931e-01 4.90449369e-01 3.68493259e-01
-5.62198639e-01 3.87994796e-01 4.33687180e-01 -3.00894260e-01
4.52082545e-01 -1.57134128e+00 4.54345614e-01 8.56909931e-01
-9.40105498e-01 -3.01398695e-01 1.23308015e+00 -7.13568032e-01
-7.77671933e-01 -1.43261647e+00 6.51255488e-01 -8.99955407e-02
5.20519555e-01 -1.55541599e-01 -3.03163916e-01 5.29399216e-02
8.12736928e-01 -3.16154391e-01 4.59486455e-01 -3.71427685e-01
3.82564664e-01 -2.77178049e-01 -1.45638537e+00 8.02706361e-01
7.80621409e-01 6.99842498e-02 -4.16872114e-01 1.71946794e-01
3.86923254e-01 -6.63516298e-03 -4.59703535e-01 1.06632426e-01
4.79152799e-01 -1.02349448e+00 9.42859709e-01 -6.49598181e-01
-4.12057266e-02 -2.88867295e-01 -3.83294612e-01 -1.50061786e+00
-7.05807805e-02 -5.99049330e-01 3.65351319e-01 5.45733929e-01
4.60466564e-01 -7.50329196e-01 1.08579969e+00 1.83211103e-01
-1.16881318e-01 -6.01996005e-01 -1.59277987e+00 -1.02307999e+00
5.95799804e-01 -3.53868932e-01 4.75585878e-01 1.69903770e-01
-8.22199225e-01 4.21923339e-01 -4.29695308e-01 2.62779504e-01
1.13663518e+00 -9.11028445e-01 7.45137572e-01 -1.33620942e+00
3.71598750e-01 -4.79912311e-01 -1.07462323e+00 -4.35170352e-01
4.95785266e-01 -5.15518844e-01 2.59255499e-01 -1.86565685e+00
-5.32090783e-01 -1.77294299e-01 -2.29340658e-01 -7.88958520e-02
-2.03912884e-01 6.92796782e-02 -1.38314351e-01 -6.31709695e-01
-8.28017354e-01 8.13331246e-01 1.04777646e+00 -3.92140001e-01
-2.37476043e-02 5.55736303e-01 -4.57282007e-01 5.04915893e-01
1.31977320e+00 -5.29885471e-01 -6.01607800e-01 1.03444187e-02
1.30504876e-01 1.16683759e-01 5.30406952e-01 -1.60168719e+00
3.00646245e-01 -5.25272131e-01 -1.78366870e-01 -1.02847672e+00
2.54473746e-01 -6.92809880e-01 -3.68576616e-01 9.01435375e-01
-3.38296071e-02 6.66250527e-01 3.51911068e-01 7.79774547e-01
-5.77996075e-02 -2.91568190e-01 1.18458652e+00 -1.62610888e-01
-1.73993039e+00 1.83978915e-01 -1.31912637e+00 2.45227069e-01
1.43107116e+00 -7.63714552e-01 -9.41097140e-02 -5.11236310e-01
-6.46855295e-01 8.67232502e-01 -1.97479948e-01 5.56002021e-01
7.03146994e-01 -1.22410655e+00 -1.06189716e+00 -1.77103177e-01
-4.03321767e-03 -5.76845288e-01 2.65977621e-01 2.57110775e-01
-7.68671811e-01 6.90069854e-01 -1.04979372e+00 -2.82643855e-01
-1.15155768e+00 3.39643449e-01 9.25221920e-01 -1.82991937e-01
-1.49578005e-01 3.73088896e-01 -2.34711826e-01 -8.06723654e-01
2.29151085e-01 2.97063559e-01 -7.66783655e-01 7.98551068e-02
1.16359368e-01 1.06141067e+00 6.10530935e-02 -1.05368626e+00
-3.50186199e-01 2.65899360e-01 6.83749244e-02 -8.46850276e-01
9.81846154e-01 -5.02380133e-01 5.62111139e-01 5.89960478e-02
1.09213662e+00 -6.34673715e-01 -1.40075910e+00 -1.68530941e-02
1.97261870e-01 -2.96345204e-01 5.52914776e-02 -1.01110399e+00
-5.86965859e-01 6.73298717e-01 1.53147709e+00 1.98063012e-02
6.51126504e-01 -7.09897220e-01 7.91480005e-01 6.12837255e-01
1.71191081e-01 -1.69784653e+00 -3.61100282e-03 9.16389823e-01
5.11902452e-01 -1.80558026e+00 -2.57995307e-01 4.84550774e-01
-6.78972960e-01 8.47560167e-01 1.02093530e+00 -4.02597964e-01
9.38690960e-01 -1.26520246e-01 4.69889104e-01 -2.04003649e-03
-6.59148932e-01 -1.07376218e+00 -1.89207911e-01 1.43652833e+00
-3.18218976e-01 2.23604918e-01 -5.94170213e-01 -2.62990892e-01
-5.26313007e-01 -3.38001758e-01 6.73807263e-01 1.00510144e+00
-9.43773627e-01 -1.28190935e+00 -3.46030116e-01 2.02416047e-01
-1.67154044e-01 2.42644668e-01 3.01425764e-03 1.26044524e+00
6.02661490e-01 1.54239535e+00 -1.15743198e-01 -2.39305109e-01
7.35116482e-01 -4.86475199e-01 -1.12390481e-01 -8.34557936e-02
-7.73133814e-01 -6.09747410e-01 4.05970335e-01 -4.13040966e-01
-5.22514760e-01 -1.00219727e+00 -1.11998415e+00 -3.42779577e-01
2.36643538e-01 1.44574583e-01 9.95566726e-01 9.93484020e-01
-1.33198842e-01 5.40078342e-01 7.05158651e-01 -4.10499930e-01
-5.43152452e-01 -3.68384123e-01 -2.80190021e-01 -9.60125029e-02
6.55845523e-01 -1.02931929e+00 -6.29990324e-02 -5.83012521e-01] | [5.237006187438965, 1.1922353506088257] |
ed8fe2b8-23b6-4171-a9be-e24ed18698a7 | latent-video-diffusion-models-for-high | 2211.13221 | null | https://arxiv.org/abs/2211.13221v2 | https://arxiv.org/pdf/2211.13221v2.pdf | Latent Video Diffusion Models for High-Fidelity Long Video Generation | AI-generated content has attracted lots of attention recently, but photo-realistic video synthesis is still challenging. Although many attempts using GANs and autoregressive models have been made in this area, the visual quality and length of generated videos are far from satisfactory. Diffusion models have shown remarkable results recently but require significant computational resources. To address this, we introduce lightweight video diffusion models by leveraging a low-dimensional 3D latent space, significantly outperforming previous pixel-space video diffusion models under a limited computational budget. In addition, we propose hierarchical diffusion in the latent space such that longer videos with more than one thousand frames can be produced. To further overcome the performance degradation issue for long video generation, we propose conditional latent perturbation and unconditional guidance that effectively mitigate the accumulated errors during the extension of video length. Extensive experiments on small domain datasets of different categories suggest that our framework generates more realistic and longer videos than previous strong baselines. We additionally provide an extension to large-scale text-to-video generation to demonstrate the superiority of our work. Our code and models will be made publicly available. | ['Qifeng Chen', 'Ying Shan', 'Yong Zhang', 'Tianyu Yang', 'Yingqing He'] | 2022-11-23 | null | null | null | null | ['video-generation', 'text-to-video-generation'] | ['computer-vision', 'natural-language-processing'] | [ 3.12035799e-01 7.08342195e-02 -2.13163897e-01 -8.15405250e-02
-9.74376559e-01 -4.54428166e-01 8.13108683e-01 -7.35984743e-01
1.85969621e-02 8.79153430e-01 7.56748736e-01 -1.29528027e-02
3.74586970e-01 -6.92503512e-01 -8.18058610e-01 -7.32345462e-01
1.99850783e-01 2.15909649e-02 2.60082424e-01 6.76856712e-02
8.58972222e-02 -7.28607783e-03 -1.30814707e+00 3.33234727e-01
1.01624453e+00 7.90295243e-01 2.94453025e-01 6.84047997e-01
1.08528465e-01 9.42639709e-01 -6.74135923e-01 -4.11710650e-01
3.89510125e-01 -8.11924398e-01 -4.57481414e-01 4.66984630e-01
5.45671284e-01 -9.14321482e-01 -5.44553876e-01 8.24740171e-01
7.31133938e-01 8.62135291e-02 7.70390332e-01 -1.34024358e+00
-1.18153059e+00 6.27938271e-01 -6.01150453e-01 -6.88974634e-02
5.45749247e-01 5.36724210e-01 7.78497458e-01 -8.28678310e-01
8.80501509e-01 1.41680992e+00 4.51564521e-01 8.90095294e-01
-1.13887513e+00 -8.29877496e-01 2.90753335e-01 9.66733973e-03
-1.25982916e+00 -6.54434621e-01 8.24808180e-01 -4.49833810e-01
7.14417696e-01 -6.35157898e-02 6.17705345e-01 1.83563662e+00
1.44088417e-02 1.16725993e+00 7.32827902e-01 -2.78960973e-01
1.43618450e-01 -1.73670247e-01 -8.52043331e-01 5.44761837e-01
9.93743688e-02 -9.29219127e-02 -5.33539772e-01 -3.65287922e-02
1.19033134e+00 -5.46604991e-02 -3.71170759e-01 -2.95496225e-01
-1.39729428e+00 9.79785204e-01 4.16455008e-02 8.84229466e-02
-6.20870948e-01 4.30740714e-01 2.12484166e-01 1.23925526e-02
7.82384753e-01 3.02186579e-01 5.79859614e-02 -4.79611099e-01
-1.32182705e+00 4.72701728e-01 4.18049216e-01 1.36304843e+00
3.10981184e-01 5.73807299e-01 -5.29889643e-01 8.08067203e-01
1.79173753e-01 5.56444049e-01 4.05854523e-01 -1.49451363e+00
6.29027605e-01 1.41026005e-01 2.43204743e-01 -1.10316110e+00
3.43388021e-01 -1.01300135e-01 -9.61869240e-01 -1.35421278e-02
2.86274999e-01 -4.88817662e-01 -8.31628859e-01 1.73720896e+00
1.45349175e-01 4.44097608e-01 3.50403301e-02 9.49706495e-01
5.69346428e-01 1.07304692e+00 -8.79293587e-03 -3.87431830e-01
7.92041183e-01 -1.46032274e+00 -9.41659689e-01 -7.51065537e-02
3.33354026e-01 -8.47439647e-01 1.13423395e+00 3.50869149e-01
-1.49500763e+00 -5.29470623e-01 -8.63422990e-01 -4.70020548e-02
3.24163824e-01 3.66569459e-02 5.49513221e-01 6.31551564e-01
-1.27316678e+00 3.32305253e-01 -8.11554313e-01 -2.88993955e-01
5.80476344e-01 -1.19245484e-01 -1.48457482e-01 -2.78715521e-01
-1.03550553e+00 3.45270187e-01 1.51366800e-01 -3.71007398e-02
-1.25325775e+00 -6.32236302e-01 -8.70455086e-01 -1.39478490e-01
4.74501133e-01 -9.21715379e-01 1.15918493e+00 -1.00529480e+00
-1.74030757e+00 2.29628891e-01 -1.96634471e-01 -5.56163251e-01
1.00238073e+00 -4.79836553e-01 -1.72501594e-01 3.22449356e-01
1.24453545e-01 1.32406509e+00 1.23430133e+00 -1.23043025e+00
-4.08673853e-01 2.47472525e-01 9.14802179e-02 2.70340115e-01
-7.41374969e-01 -1.36566237e-01 -8.51603150e-01 -1.29933822e+00
-3.53266478e-01 -1.15032291e+00 -3.84253800e-01 5.45870885e-02
-2.82846510e-01 -4.88705114e-02 9.40083742e-01 -6.51057303e-01
1.29455388e+00 -2.10418177e+00 3.36775541e-01 -4.32890177e-01
1.46512538e-01 9.73044038e-02 -4.45624918e-01 4.15544331e-01
2.21849725e-01 2.77024090e-01 -1.52833402e-01 -4.57462937e-01
1.11345598e-03 6.87692240e-02 -6.10315621e-01 9.73977670e-02
1.25549987e-01 1.06180918e+00 -9.02464628e-01 -5.71020484e-01
9.69222337e-02 6.40828967e-01 -9.37899649e-01 2.54668534e-01
-5.25646687e-01 5.09083033e-01 -4.15291041e-01 6.40735984e-01
5.03639042e-01 -4.41422433e-01 -5.56117222e-02 -9.11831632e-02
2.13623136e-01 -4.33065183e-02 -9.27524626e-01 1.95357406e+00
-3.43927622e-01 8.20800602e-01 -2.25153193e-01 -5.45864284e-01
6.40511692e-01 4.54677194e-01 5.23816228e-01 -5.41640460e-01
-4.67149355e-02 -6.41586483e-02 -3.38500142e-01 -4.96929526e-01
8.54591489e-01 -2.65113870e-03 8.20648968e-02 4.98893380e-01
-2.33752549e-01 -3.61168087e-01 4.23412591e-01 5.17506361e-01
9.13296759e-01 5.96421301e-01 -2.12755591e-01 2.57672489e-01
8.13866705e-02 -2.00083494e-01 6.46567106e-01 6.57538652e-01
-1.61601484e-01 1.02980006e+00 4.75941122e-01 -7.58748278e-02
-1.41174555e+00 -8.93628955e-01 3.74335408e-01 6.59105897e-01
3.83671075e-02 -6.52138352e-01 -1.08163178e+00 -5.97265363e-01
-2.72410035e-01 7.72579014e-01 -4.10786241e-01 8.88739154e-03
-5.81466079e-01 -5.63475609e-01 6.88525617e-01 7.23196387e-01
5.39168119e-01 -9.84992921e-01 -2.50303179e-01 3.20482969e-01
-5.30914962e-01 -1.41783535e+00 -7.84613609e-01 -8.41893077e-01
-8.73298228e-01 -4.83069748e-01 -1.55146635e+00 -5.79183578e-01
5.87334156e-01 4.26699489e-01 9.93341625e-01 -1.36690855e-01
-1.29389912e-01 4.99004573e-01 -5.26363909e-01 -1.53992459e-01
-5.89165986e-01 -1.07802927e-01 1.71510771e-01 -5.26524894e-02
-1.00839652e-01 -4.94241655e-01 -9.49275076e-01 3.75351220e-01
-1.14566040e+00 4.55748022e-01 5.09033561e-01 7.65461564e-01
4.32727545e-01 -5.63607030e-02 5.88612437e-01 -6.03543043e-01
7.33210087e-01 -5.09713590e-01 -4.24523711e-01 1.58946421e-02
-4.83741909e-01 -3.20407599e-02 5.49003422e-01 -6.70836806e-01
-1.30408049e+00 -2.14334488e-01 -6.62004054e-02 -7.70082653e-01
3.01502794e-02 2.02295229e-01 -5.32680862e-02 2.48475313e-01
4.83997941e-01 3.03602874e-01 -1.02883264e-01 -2.53598899e-01
6.18677020e-01 4.40225393e-01 2.46326730e-01 -5.62644541e-01
8.03023875e-01 5.06927490e-01 -3.75428587e-01 -7.38978326e-01
-5.26127756e-01 1.48147494e-01 -1.72318920e-01 -3.42469454e-01
9.19286430e-01 -1.33033502e+00 -2.30949819e-01 7.09991336e-01
-1.11168909e+00 -7.09316194e-01 -2.52803028e-01 5.21478474e-01
-7.52003670e-01 7.92657793e-01 -9.64548588e-01 -7.06104636e-01
-2.20423296e-01 -1.40155864e+00 1.20844305e+00 -2.15560906e-02
-2.83686697e-01 -8.91504169e-01 -9.74753648e-02 5.07497489e-01
5.22822499e-01 2.15919346e-01 5.01815617e-01 1.80036321e-01
-1.04191172e+00 3.18126976e-02 -2.20384404e-01 4.19754237e-01
1.44081026e-01 2.22268388e-01 -5.61894715e-01 -3.95358384e-01
-1.90505385e-01 -4.80162710e-01 9.72747087e-01 6.54454172e-01
1.26320004e+00 -4.18987423e-01 -2.95400769e-01 6.12844527e-01
1.01082945e+00 8.21882039e-02 7.80499518e-01 2.14730471e-01
8.78934979e-01 4.63289857e-01 6.65890038e-01 6.42154515e-01
4.49313670e-01 7.06321299e-01 2.41586357e-01 -1.10177584e-01
-5.54320335e-01 -5.43697178e-01 7.17293739e-01 1.05971837e+00
-2.38426968e-01 -8.21551919e-01 -4.81678367e-01 7.18462765e-01
-1.79103005e+00 -1.29701972e+00 5.50565086e-02 1.82794130e+00
7.72419989e-01 1.79687470e-01 2.00811043e-01 -1.90273732e-01
5.74477553e-01 5.11779487e-01 -5.73297024e-01 -3.62430490e-03
-3.47877651e-01 -4.13526893e-01 3.87364417e-01 3.68563414e-01
-8.61406505e-01 1.22671258e+00 7.16451883e+00 1.09906101e+00
-9.04356360e-01 2.65614558e-02 8.60089779e-01 -4.61195081e-01
-6.63632870e-01 -2.05312558e-02 -7.36157358e-01 7.19089329e-01
7.49905467e-01 -3.43634896e-02 3.25693637e-01 8.74997616e-01
5.05477488e-01 -1.60545649e-04 -8.59605491e-01 1.15090144e+00
3.16779405e-01 -1.64818668e+00 4.99132812e-01 2.99120247e-01
1.33235025e+00 -2.13981181e-01 3.82179320e-01 2.22541109e-01
4.55239683e-01 -8.72691751e-01 8.80623996e-01 5.48523843e-01
1.07175314e+00 -5.39261341e-01 2.68865287e-01 2.64185309e-01
-1.06574011e+00 -4.12924662e-02 -3.71772736e-01 8.64355266e-02
6.56734943e-01 5.52525997e-01 -5.12710810e-01 2.82494366e-01
5.22234857e-01 9.36903119e-01 -4.67737347e-01 7.33572066e-01
-3.46608639e-01 8.22608411e-01 -1.38518035e-01 2.06466049e-01
3.77912611e-01 -2.42556617e-01 6.32257938e-01 1.11231744e+00
8.21271837e-01 8.06752443e-02 9.04284045e-02 9.85617280e-01
-2.22095251e-01 -5.76040633e-02 -8.10442567e-01 -3.86347204e-01
3.21411371e-01 9.16238010e-01 -5.74299335e-01 -5.26306987e-01
-5.07820547e-01 1.47387552e+00 1.99307203e-02 7.61338949e-01
-1.15831101e+00 -3.08678299e-02 6.11623287e-01 9.64540765e-02
4.73488003e-01 -4.66452271e-01 1.01699755e-01 -1.53463411e+00
1.27553359e-01 -9.96412933e-01 -9.70082078e-03 -1.05283844e+00
-1.15254879e+00 5.21905005e-01 -2.89999563e-02 -1.34971654e+00
-7.24931180e-01 -1.11142755e-01 -2.91654080e-01 5.62725306e-01
-1.28698480e+00 -1.13822973e+00 -2.93172598e-01 5.56965053e-01
1.19032562e+00 -2.86846250e-01 4.74922359e-01 2.89902270e-01
-4.84983385e-01 6.86855793e-01 2.05902696e-01 -5.03417104e-02
1.01362693e+00 -8.42018485e-01 7.73822427e-01 1.10789871e+00
2.57036593e-02 4.06523198e-01 7.49372303e-01 -9.11185503e-01
-1.41637444e+00 -1.07529533e+00 4.98755366e-01 -4.79173899e-01
4.71667945e-01 -4.31831390e-01 -6.44110024e-01 6.70119286e-01
5.75494170e-01 -3.63611490e-01 5.23078918e-01 -4.24569517e-01
-1.65470779e-01 1.85753524e-01 -7.64325202e-01 1.09299409e+00
1.38885307e+00 -2.53554255e-01 -9.59404483e-02 2.77924091e-01
1.08911932e+00 -3.87824714e-01 -8.03184927e-01 3.35468918e-01
4.52382833e-01 -1.03109515e+00 9.08221602e-01 -1.41952008e-01
9.96995807e-01 -2.55536944e-01 -5.06659150e-02 -1.16380870e+00
-3.75783533e-01 -1.02989614e+00 -4.23652798e-01 1.42023587e+00
1.80619150e-01 -2.26957295e-02 1.01771879e+00 5.99839866e-01
5.01601472e-02 -5.31570792e-01 -4.26146954e-01 -7.73727536e-01
1.78210847e-02 -5.79857171e-01 4.85748291e-01 7.78465748e-01
-3.50638986e-01 3.10840040e-01 -1.16197944e+00 -2.22114950e-01
5.91969013e-01 1.08623197e-02 1.10311806e+00 -5.32898724e-01
-4.22216624e-01 -4.04208422e-01 -1.89371228e-01 -1.81981230e+00
1.78666756e-01 -4.73133355e-01 7.62973959e-03 -1.68730760e+00
3.01701605e-01 -2.20618710e-01 2.07988769e-01 1.44078583e-01
-2.92475164e-01 5.58159709e-01 3.34342957e-01 3.76890957e-01
-6.34119093e-01 1.05490673e+00 1.62712657e+00 -1.37537062e-01
-2.13946477e-02 -3.71867448e-01 -6.49396539e-01 6.72728240e-01
4.68652666e-01 -1.81375057e-01 -7.26740181e-01 -8.34527433e-01
1.16650835e-01 2.16107979e-01 1.65777162e-01 -1.00006497e+00
7.23132268e-02 -2.73895621e-01 5.37224531e-01 -4.90260005e-01
5.99406838e-01 -4.98795599e-01 2.98230886e-01 1.07840493e-01
-4.87404913e-01 4.89956439e-02 1.31914597e-02 9.04628217e-01
-2.25030094e-01 1.37093291e-01 4.94791001e-01 -1.78680010e-02
-5.55039585e-01 6.22904420e-01 -5.62485754e-01 2.30205338e-02
1.13594651e+00 -3.55553240e-01 -7.83072934e-02 -1.08967566e+00
-4.03584719e-01 1.58940747e-01 9.41118598e-01 6.94145143e-01
7.89253235e-01 -1.51845908e+00 -9.38884377e-01 -4.82064113e-02
-5.62141165e-02 -6.16125092e-02 5.19522369e-01 5.91880202e-01
-5.32305360e-01 2.87129104e-01 -1.11074641e-01 -5.63531339e-01
-1.06050622e+00 6.93542778e-01 -2.87244529e-01 -1.05692849e-01
-6.97009265e-01 8.54784191e-01 5.30974865e-01 1.35038972e-01
2.04475433e-01 -8.57987553e-02 5.16902320e-02 -1.22295067e-01
5.94067812e-01 3.80645543e-01 -5.79147577e-01 -7.07499743e-01
1.68540969e-01 4.44955915e-01 -1.68485671e-01 -3.74013662e-01
1.26795459e+00 -4.60373819e-01 3.46527934e-01 1.29115313e-01
9.22138035e-01 1.62146002e-01 -2.02235961e+00 -1.01364724e-01
-4.81478125e-01 -7.54253983e-01 -1.37533143e-01 -4.29605186e-01
-1.21890152e+00 9.46454525e-01 2.51843542e-01 -6.35146117e-03
1.16275585e+00 -2.20415115e-01 1.35717928e+00 -8.80613998e-02
2.82622516e-01 -1.05429029e+00 5.58609366e-01 2.75772810e-01
9.45694327e-01 -1.16256225e+00 5.09273037e-02 -3.27257186e-01
-1.00646138e+00 9.16552067e-01 6.18387043e-01 -5.43391183e-02
7.10309669e-02 2.46976256e-01 -2.15973947e-02 3.32025588e-01
-9.57311153e-01 1.45408809e-01 1.82676733e-01 6.58594489e-01
5.36675334e-01 -2.26624027e-01 -3.02193582e-01 2.29258299e-01
1.13744866e-02 2.20985234e-01 7.26569593e-01 7.12801039e-01
-7.88668394e-02 -1.19387209e+00 -3.37514192e-01 1.42097801e-01
-5.69371045e-01 -2.65399665e-01 -9.63560268e-02 4.02838051e-01
-2.19003037e-01 1.00314093e+00 -4.11123894e-02 -1.92742988e-01
-9.01435688e-02 -1.93793803e-01 4.79769349e-01 -2.77842700e-01
7.00807646e-02 7.09982038e-01 9.93020926e-03 -6.43093050e-01
-6.06192589e-01 -6.58604622e-01 -7.75132179e-01 -5.31154811e-01
-2.17510238e-01 -1.00047424e-01 3.54412854e-01 6.24864280e-01
6.00102425e-01 5.24556935e-01 4.30319428e-01 -1.15892172e+00
-3.59230131e-01 -9.46928978e-01 -2.92150050e-01 4.97820318e-01
1.01726897e-01 -5.83090365e-01 -2.33169973e-01 6.06377184e-01] | [10.90688419342041, -0.5223299860954285] |
b82eeacf-e035-4d0c-a1c3-32574a8cc900 | unsupervised-augmentation-optimization-for | 2306.05107 | null | https://arxiv.org/abs/2306.05107v1 | https://arxiv.org/pdf/2306.05107v1.pdf | Unsupervised augmentation optimization for few-shot medical image segmentation | The augmentation parameters matter to few-shot semantic segmentation since they directly affect the training outcome by feeding the networks with varying perturbated samples. However, searching optimal augmentation parameters for few-shot segmentation models without annotations is a challenge that current methods fail to address. In this paper, we first propose a framework to determine the ``optimal'' parameters without human annotations by solving a distribution-matching problem between the intra-instance and intra-class similarity distribution, with the intra-instance similarity describing the similarity between the original sample of a particular anatomy and its augmented ones and the intra-class similarity representing the similarity between the selected sample and the others in the same class. Extensive experiments demonstrate the superiority of our optimized augmentation in boosting few-shot segmentation models. We greatly improve the top competing method by 1.27\% and 1.11\% on Abd-MRI and Abd-CT datasets, respectively, and even achieve a significant improvement for SSL-ALP on the left kidney by 3.39\% on the Abd-CT dataset. | ['S. Kevin Zhou', 'Heqin Zhu', 'Qingsong Yao', 'Shang Zhao', 'Quan Quan'] | 2023-06-08 | null | null | null | null | ['few-shot-image-segmentation', 'anatomy'] | ['computer-vision', 'miscellaneous'] | [ 3.64073545e-01 4.53504741e-01 -4.05344754e-01 -5.86678684e-01
-1.17869651e+00 -1.41056627e-01 2.59945512e-01 4.30667520e-01
-6.19674385e-01 4.85189170e-01 5.57988742e-03 2.25718752e-01
6.63439091e-03 -6.17919028e-01 -7.84573257e-01 -9.63928819e-01
2.64176968e-02 7.65232205e-01 4.38671529e-01 -2.38334119e-01
-1.11487336e-01 -1.76296458e-02 -1.47653067e+00 2.39644140e-01
9.61468637e-01 1.25729382e+00 1.83344036e-01 5.46870947e-01
-2.22583577e-01 6.95245802e-01 -5.45353115e-01 -2.65950412e-01
3.09517860e-01 -7.30686545e-01 -7.02425420e-01 -2.35396102e-02
6.38662040e-01 -1.45600259e-01 -1.09769285e-01 1.23795319e+00
7.52536595e-01 4.40664470e-01 6.14660442e-01 -1.23071170e+00
-4.51417238e-01 8.35830748e-01 -6.58740819e-01 3.20636988e-01
-1.45901814e-01 1.59132481e-01 7.89243877e-01 -6.09976411e-01
7.52930045e-01 8.64298165e-01 6.75375402e-01 9.76408124e-01
-1.11680353e+00 -5.59212625e-01 9.63333845e-02 1.25808895e-01
-1.05694270e+00 -2.83894300e-01 5.87384760e-01 -4.38557267e-01
4.42852378e-01 6.28891587e-02 7.62021840e-01 9.76104736e-01
-1.44246981e-01 9.77521837e-01 8.65453064e-01 -2.74130642e-01
5.55380642e-01 1.44329563e-01 4.93557334e-01 6.70017600e-01
-1.10771675e-02 -1.24972142e-01 -1.83908507e-01 -5.84735908e-02
5.08180201e-01 1.30624965e-01 -1.92070007e-01 -6.75310969e-01
-9.57140565e-01 9.32996690e-01 7.32439280e-01 5.01440525e-01
-3.40286821e-01 2.52243161e-01 5.11455417e-01 -1.20011590e-01
4.40289110e-01 4.83874142e-01 -3.11225206e-01 -6.83719218e-02
-1.01749444e+00 2.73699373e-01 6.18889809e-01 1.04128551e+00
6.14585280e-01 -5.64802401e-02 -5.73610663e-01 1.20043969e+00
-1.71419784e-01 1.34590730e-01 7.73026586e-01 -8.61061335e-01
2.38756061e-01 4.37687308e-01 -1.45781428e-01 -3.59804779e-01
-4.68930423e-01 -6.44675970e-01 -7.59136319e-01 -1.22286350e-01
5.52189827e-01 -2.52291173e-01 -1.63279808e+00 1.90069485e+00
6.29962265e-01 5.05646884e-01 -1.47079155e-01 1.03649068e+00
1.35327613e+00 3.29072773e-01 4.97333735e-01 -2.07837656e-01
1.52042305e+00 -1.20902967e+00 -7.08018124e-01 -5.38768589e-01
9.08867836e-01 -3.54653031e-01 1.23009300e+00 -1.33831903e-01
-1.14557755e+00 -4.49632108e-01 -1.13240552e+00 1.17422402e-01
3.15513881e-03 -2.45472878e-01 6.18364334e-01 5.09250462e-01
-6.27925932e-01 7.51823366e-01 -8.65148187e-01 -1.06020644e-01
8.05120170e-01 1.09155796e-01 -4.60263109e-04 -1.98843718e-01
-1.13235784e+00 6.50289237e-01 3.63752097e-01 -3.97965193e-01
-1.01166713e+00 -1.47192776e+00 -8.94233465e-01 2.52557814e-01
6.38141870e-01 -5.48470974e-01 1.46118164e+00 -7.63118207e-01
-1.15894794e+00 1.08066678e+00 8.64028782e-02 -6.37932539e-01
6.53686941e-01 1.40821725e-01 2.26914398e-02 1.57458514e-01
3.08366060e-01 9.47461843e-01 5.74401975e-01 -1.10370564e+00
-5.42946994e-01 -5.15892446e-01 -1.19669713e-01 3.08300436e-01
-3.51934917e-02 -4.80964839e-01 -5.51577806e-01 -7.49738932e-01
3.79290104e-01 -9.48422134e-01 -6.64231837e-01 2.33486846e-01
-2.56407857e-01 -6.72960058e-02 4.18551624e-01 -4.44745243e-01
7.67950118e-01 -2.21972156e+00 -6.57260127e-04 -1.03142798e-01
4.99909669e-02 2.63131320e-01 -3.19666602e-02 -2.75440067e-01
-7.74787739e-02 -1.11314960e-01 -6.81030691e-01 -3.26241016e-01
-1.42994359e-01 4.58807856e-01 -8.90232064e-03 4.54950273e-01
6.54414482e-03 8.99474859e-01 -1.12798131e+00 -7.15713561e-01
1.35798126e-01 2.69806445e-01 -6.21002793e-01 1.68451160e-01
-1.73158944e-01 4.68836993e-01 -2.29578897e-01 5.48630416e-01
5.01780152e-01 -3.64143580e-01 6.34648977e-03 -2.55717151e-02
4.31941718e-01 -1.97551668e-01 -1.03748488e+00 2.11611962e+00
-2.39010170e-01 1.21053182e-01 3.27191837e-02 -1.14797914e+00
7.84125745e-01 1.97987556e-01 7.66392648e-01 -8.06634903e-01
2.58020759e-01 1.86352447e-01 2.36829311e-01 -3.51111650e-01
3.29428732e-01 -8.20277035e-01 -2.44864494e-01 3.67694110e-01
4.42090690e-01 -3.18085670e-01 2.60440081e-01 2.86420316e-01
8.50766659e-01 -1.24213822e-01 3.72920960e-01 -2.67082214e-01
1.51064023e-01 -7.24501610e-02 7.11843908e-01 1.06929207e+00
-6.88293755e-01 9.42032218e-01 6.58876836e-01 -3.69271010e-01
-8.92648935e-01 -1.02432239e+00 -3.84935349e-01 1.27406430e+00
4.17221457e-01 8.06956515e-02 -9.51518118e-01 -9.19747174e-01
-8.92274305e-02 1.01750863e+00 -9.96294498e-01 -3.84892166e-01
-5.09459198e-01 -9.29636776e-01 4.06077385e-01 8.39965224e-01
3.31930608e-01 -8.65431666e-01 -7.17314422e-01 2.18808666e-01
-3.86078537e-01 -1.05925810e+00 -5.15729666e-01 4.44426239e-01
-1.04244912e+00 -8.73858869e-01 -1.20319974e+00 -8.28456640e-01
6.78727806e-01 -2.72986609e-02 9.83685374e-01 1.48741379e-01
-6.55487299e-01 -6.58785105e-02 -2.57983267e-01 -4.61802483e-01
-2.02737242e-01 1.21519655e-01 -4.61163819e-01 -3.35242897e-01
2.45732181e-02 -3.88205975e-01 -7.05646396e-01 3.20207953e-01
-7.09977210e-01 -2.34559309e-02 1.95663109e-01 9.71092761e-01
8.34749818e-01 -5.84456146e-01 5.94137967e-01 -1.23482251e+00
9.89062861e-02 -5.83946645e-01 -7.30186626e-02 1.93090290e-01
-6.12054467e-01 -6.06299378e-02 2.92205095e-01 -6.80904925e-01
-9.18415010e-01 9.73003432e-02 -3.54129635e-02 -7.19589114e-01
-9.41211805e-02 3.40929955e-01 2.00255558e-01 1.76654339e-01
8.84806752e-01 6.75373748e-02 -2.45569050e-02 -3.54022771e-01
5.64470351e-01 1.90494180e-01 7.46381998e-01 -4.84796613e-01
2.25164294e-01 7.25951016e-01 -1.95154518e-01 -6.92331851e-01
-1.27207184e+00 -7.46328175e-01 -4.68092650e-01 -1.71197563e-01
9.70123172e-01 -7.50061750e-01 -1.37821510e-01 2.80629992e-01
-6.09706640e-01 -5.20890951e-01 -9.63733256e-01 4.33770359e-01
-7.24894345e-01 2.34996915e-01 -6.62921607e-01 -5.00351369e-01
-5.33019543e-01 -1.39640772e+00 1.02591991e+00 4.39833730e-01
-2.58768022e-01 -7.61732638e-01 9.37474668e-02 5.10267794e-01
3.23004991e-01 2.47024208e-01 1.08957493e+00 -9.08143699e-01
-2.05350503e-01 -3.12264502e-01 -1.77877530e-01 1.50987402e-01
-2.81087942e-02 -4.00850743e-01 -9.38454449e-01 -2.07813606e-01
8.08918849e-02 -4.49238688e-01 9.53712523e-01 9.87062156e-01
1.30491662e+00 9.46265459e-02 -3.84892970e-01 8.79974723e-01
1.33995485e+00 2.31803909e-01 4.98571247e-01 1.13278227e-02
6.45146966e-01 3.51327240e-01 7.67373860e-01 4.59167331e-01
7.00521618e-02 5.41399956e-01 4.45254862e-01 -6.09129854e-03
-3.77232403e-01 -7.02504367e-02 -4.12135988e-01 5.54830551e-01
7.28393644e-02 6.06098101e-02 -1.06097937e+00 8.32538545e-01
-1.81063306e+00 -8.18843544e-01 1.41779333e-01 2.12827396e+00
8.55002165e-01 2.68839866e-01 1.43380135e-01 -1.59308881e-01
8.84472191e-01 3.15485507e-01 -8.22034836e-01 -6.08594231e-02
3.11321318e-01 3.89544129e-01 3.51088822e-01 2.59629011e-01
-1.20589542e+00 8.51055682e-01 5.80288982e+00 9.60934341e-01
-1.01371717e+00 3.12503725e-01 8.67142379e-01 -4.65302974e-01
-4.92663980e-02 -1.84556752e-01 -7.14554071e-01 5.85254073e-01
8.40298414e-01 -2.68260151e-01 4.68216315e-02 1.16835690e+00
-2.21291468e-01 -2.02491954e-01 -1.17662311e+00 7.33067393e-01
2.04535961e-01 -1.28439748e+00 -8.80659744e-02 -2.37998143e-01
1.05951893e+00 5.27764484e-02 -3.15508135e-02 7.21588433e-01
2.91734897e-02 -8.87138188e-01 5.74863255e-01 2.83432811e-01
6.86810255e-01 -6.66821063e-01 8.19923699e-01 3.70706111e-01
-9.62113082e-01 3.71048003e-02 -4.00543213e-01 2.90706754e-01
2.17034638e-01 4.55205679e-01 -1.03719997e+00 2.57093519e-01
7.03828692e-01 4.21351492e-01 -2.52109706e-01 1.23037231e+00
-1.22884713e-01 4.74222541e-01 -1.26283795e-01 3.81468013e-02
4.93761241e-01 -8.02831724e-02 4.65713799e-01 9.85413492e-01
1.11657701e-01 2.71893263e-01 3.05908024e-01 6.45425975e-01
-2.67508179e-01 1.66041002e-01 -1.11765891e-01 1.57204717e-01
2.67947257e-01 1.17468524e+00 -1.09461749e+00 -8.49877954e-01
-5.58507293e-02 6.22317493e-01 3.14079612e-01 1.65171444e-01
-1.12447226e+00 -2.74877727e-01 5.23309827e-01 3.11700422e-02
4.30535078e-01 5.08414447e-01 -6.90188706e-01 -9.46397185e-01
-1.44329801e-01 -6.42495632e-01 7.95562267e-01 -6.33331180e-01
-1.21012282e+00 3.36019576e-01 -1.90900918e-02 -1.08155489e+00
-1.64319754e-01 -9.34170708e-02 -6.64817631e-01 5.75781286e-01
-1.06192410e+00 -9.17235315e-01 -5.56252062e-01 2.16853425e-01
7.84875095e-01 8.77764523e-02 7.31528163e-01 3.59455645e-01
-2.90483385e-01 6.62238181e-01 -3.91856320e-02 1.55459732e-01
5.84820986e-01 -1.35711408e+00 2.66252279e-01 5.24219871e-01
-1.29838064e-01 1.75412297e-01 7.88600624e-01 -6.25955164e-01
-6.89982474e-01 -1.00525367e+00 4.77584630e-01 -9.58791375e-02
4.19289500e-01 -4.90538329e-02 -1.20053911e+00 5.98705471e-01
-3.01345497e-01 4.67400074e-01 7.33461082e-01 5.57073355e-02
-1.88397229e-01 1.57151952e-01 -1.50623190e+00 5.86241364e-01
1.14476430e+00 -1.96418107e-01 -5.89062929e-01 4.11883682e-01
8.73899698e-01 -8.49782228e-01 -1.01767755e+00 5.79791546e-01
4.43453640e-01 -6.89204335e-01 1.05795932e+00 -9.59680557e-01
8.07270348e-01 1.50519654e-01 -1.89138979e-01 -1.38236332e+00
-1.53228015e-01 -1.01850815e-02 -1.49752453e-01 7.67526209e-01
2.81903863e-01 -2.72287250e-01 1.17323518e+00 6.85695410e-01
-3.90782505e-01 -1.15377104e+00 -1.05676126e+00 -5.88270426e-01
1.89306140e-01 -4.23121095e-01 5.63250959e-01 1.17770088e+00
-3.08458861e-02 1.57940239e-01 -2.01071963e-01 -4.66254838e-02
8.95454705e-01 2.05130845e-01 4.55647767e-01 -1.08394563e+00
-3.50929141e-01 -4.28023338e-01 -4.65177864e-01 -5.67693353e-01
-2.57883817e-02 -1.08313000e+00 3.91068518e-01 -1.59617114e+00
5.74696898e-01 -5.37854433e-01 -4.43560481e-01 5.82132697e-01
-5.83110034e-01 2.25033283e-01 2.63133079e-01 1.01935323e-02
-5.67262173e-01 6.08621597e-01 1.51974475e+00 -2.29160368e-01
-2.12874249e-01 1.15637586e-01 -6.43842280e-01 6.64954960e-01
6.49127305e-01 -7.38295436e-01 -5.39411783e-01 -7.62082934e-02
-3.68718803e-01 3.14092696e-01 3.44199300e-01 -1.02877760e+00
4.04169783e-02 -9.71573777e-03 3.09489608e-01 -5.05139530e-01
3.80431771e-01 -4.72824484e-01 -1.92265660e-01 7.68075109e-01
-6.84312046e-01 -5.17859459e-01 1.34235635e-01 7.73308635e-01
8.63132402e-02 -6.73098743e-01 1.39045072e+00 -4.92762357e-01
-6.95739508e-01 5.74055970e-01 8.65559131e-02 7.48068571e-01
1.24484587e+00 -1.18815750e-01 -2.22598433e-01 -3.32771420e-01
-1.14008462e+00 3.75600457e-01 2.61788309e-01 2.32558727e-01
3.46472204e-01 -1.22657001e+00 -5.34524977e-01 1.53803334e-01
2.54968077e-01 3.20529163e-01 8.11073601e-01 9.15405154e-01
-3.05311859e-01 -3.61323841e-02 -3.28269988e-01 -8.45921874e-01
-1.18629873e+00 3.60539436e-01 5.03626168e-01 -3.55282634e-01
-8.34853649e-01 1.28115654e+00 3.87225717e-01 -4.91357893e-01
4.34428960e-01 -3.48295271e-01 -1.90433264e-02 1.66869178e-01
5.58264077e-01 4.73871529e-01 1.30843922e-01 -2.95251399e-01
-2.31449440e-01 2.11809084e-01 -3.29687953e-01 -5.71110810e-04
1.45856047e+00 2.48122096e-01 2.91746289e-01 5.67171335e-01
1.27175999e+00 -7.26770937e-01 -1.32741630e+00 -4.44085479e-01
-1.78432018e-01 -3.41104180e-01 1.11525893e-01 -9.30052757e-01
-1.29906809e+00 8.17379475e-01 8.19991291e-01 -1.41385719e-01
6.20806217e-01 3.89747441e-01 8.74946356e-01 8.62616152e-02
1.78517208e-01 -1.22797632e+00 1.48868829e-01 2.71846175e-01
5.45207262e-01 -1.30471849e+00 -2.39775162e-02 -3.50615203e-01
-1.00537205e+00 6.06458485e-01 7.65045047e-01 -1.65223971e-01
5.63647509e-01 1.03529200e-01 1.79262623e-01 -3.33271682e-01
-5.16834855e-01 -1.22882225e-01 3.71038258e-01 4.13791388e-01
2.97875255e-01 2.61039943e-01 -2.35627413e-01 5.81740975e-01
-1.49921447e-01 3.02813072e-02 4.09153283e-01 9.00263250e-01
-5.20788372e-01 -5.11998773e-01 -1.92189775e-02 8.59837532e-01
-3.93992037e-01 2.19521560e-02 7.79997781e-02 8.26470137e-01
1.30069591e-02 3.12602937e-01 2.86218226e-01 3.12044248e-02
4.51693952e-01 3.74188691e-01 5.75771034e-01 -8.78197968e-01
-5.88187337e-01 1.82305723e-01 -5.95067628e-02 -4.75684941e-01
-8.02022368e-02 -6.60916448e-01 -1.64892876e+00 1.64479807e-01
-2.50242949e-01 1.40280202e-01 6.23579979e-01 1.00281107e+00
-6.75598234e-02 8.01166832e-01 2.12092653e-01 -8.38405967e-01
-9.71416473e-01 -8.83505046e-01 -6.16981268e-01 6.14851117e-01
1.63848996e-01 -7.38164604e-01 -3.98516387e-01 -2.32083946e-01] | [14.563264846801758, -2.1341946125030518] |
2dd08c65-fba7-47a8-9316-ebbe730340cd | a-simple-method-for-predicting-covariance | 2305.19484 | null | https://arxiv.org/abs/2305.19484v1 | https://arxiv.org/pdf/2305.19484v1.pdf | A Simple Method for Predicting Covariance Matrices of Financial Returns | We consider the well-studied problem of predicting the time-varying covariance matrix of a vector of financial returns. Popular methods range from simple predictors like rolling window or exponentially weighted moving average (EWMA) to more sophisticated predictors such as generalized autoregressive conditional heteroscedastic (GARCH) type methods. Building on a specific covariance estimator suggested by Engle in 2002, we propose a relatively simple extension that requires little or no tuning or fitting, is interpretable, and produces results at least as good as MGARCH, a popular extension of GARCH that handles multiple assets. To evaluate predictors we introduce a novel approach, evaluating the regret of the log-likelihood over a time period such as a quarter. This metric allows us to see not only how well a covariance predictor does over all, but also how quickly it reacts to changes in market conditions. Our simple predictor outperforms MGARCH in terms of regret. We also test covariance predictors on downstream applications such as portfolio optimization methods that depend on the covariance matrix. For these applications our simple covariance predictor and MGARCH perform similarly. | ['Stephen Boyd', 'Thomas Schmelzer', 'Markus Pelger', 'Mehmet Giray Ogut', 'Kasper Johansson'] | 2023-05-31 | null | null | null | null | ['portfolio-optimization'] | ['time-series'] | [-6.06588960e-01 -1.41639531e-01 2.81863123e-01 -5.27329326e-01
-8.83779824e-01 -8.97207975e-01 7.20467210e-01 4.95361909e-02
-2.68916875e-01 7.73500443e-01 1.34909719e-01 -7.04166472e-01
-4.08861160e-01 -8.86353970e-01 -5.73575616e-01 -7.07272410e-01
-4.13247645e-01 5.25342703e-01 9.68187377e-02 1.81868137e-03
4.86631095e-01 4.82832432e-01 -9.96003330e-01 -3.69755596e-01
4.08189386e-01 1.36719286e+00 -2.76412457e-01 6.91900790e-01
5.47615401e-02 7.46459842e-01 -3.76704723e-01 -7.27760196e-01
7.00708985e-01 -3.02228481e-01 -2.50499040e-01 -1.80790529e-01
-2.31342703e-01 -3.60719502e-01 8.59556496e-02 7.70658016e-01
2.30315864e-01 2.91631162e-01 8.45636785e-01 -7.81512141e-01
-3.90149683e-01 8.47071350e-01 -5.00212669e-01 4.27165687e-01
-9.46288183e-02 8.75181034e-02 1.14673555e+00 -8.47146511e-01
2.38667652e-01 1.28324652e+00 9.26056147e-01 -2.48789098e-02
-1.43891716e+00 -5.58422983e-01 1.55306295e-01 -1.89912543e-01
-1.01124597e+00 -3.44662964e-01 2.80374587e-01 -7.67631948e-01
9.26751554e-01 2.57955372e-01 3.64413261e-01 6.77926838e-01
6.35441720e-01 3.86495978e-01 1.22327006e+00 -2.92060345e-01
5.99594414e-01 2.10792348e-01 2.02819228e-01 6.31308230e-03
2.36523360e-01 3.86956930e-01 -2.38478288e-01 -5.16950607e-01
5.51596701e-01 1.81682035e-01 9.32607502e-02 -2.36412764e-01
-9.37338471e-01 1.10966313e+00 -2.05965757e-01 -1.91667452e-02
-6.81651413e-01 1.85217068e-01 5.70760369e-02 7.79223263e-01
7.68349648e-01 5.11991143e-01 -7.39564955e-01 -4.93424058e-01
-1.00311339e+00 4.06420529e-01 1.26208639e+00 8.22711706e-01
5.57642460e-01 3.76601487e-01 -2.48383492e-01 4.55604345e-01
3.14822286e-01 6.36419475e-01 2.82885313e-01 -1.05661297e+00
3.80971998e-01 8.37677717e-02 4.08387154e-01 -5.41066468e-01
-4.14312243e-01 -5.38586140e-01 -5.17356098e-01 4.37985599e-01
6.36360765e-01 -8.38569939e-01 -4.11768913e-01 1.51113117e+00
3.29856819e-04 5.03236204e-02 5.64662740e-02 5.55635333e-01
-2.61747748e-01 5.31364202e-01 -3.82289499e-01 -5.15292704e-01
8.94857585e-01 -6.32757843e-01 -3.90781492e-01 -1.04703955e-01
4.45094824e-01 -1.01735389e+00 3.82204682e-01 5.47420681e-01
-1.22550094e+00 -3.61063294e-02 -8.11730087e-01 4.99480635e-01
-1.25741646e-01 -3.34875852e-01 7.37730682e-01 1.02951884e+00
-9.69984472e-01 1.13906348e+00 -1.03118217e+00 2.08998084e-01
-2.00164065e-01 2.47464195e-01 2.01025724e-01 5.50506830e-01
-8.76617670e-01 9.44007754e-01 4.74390090e-02 -1.11521240e-02
-6.40141547e-01 -8.75836015e-01 -2.89806902e-01 4.98893440e-01
2.83500433e-01 -5.12902975e-01 1.72761595e+00 -8.61956775e-01
-1.83367300e+00 2.08233103e-01 1.06379330e-01 -9.25429285e-01
9.72258806e-01 -3.15196455e-01 -3.66515309e-01 -4.35467243e-01
-2.22570017e-01 -2.60201454e-01 7.56447673e-01 -2.72373319e-01
-6.58824563e-01 -4.32021379e-01 -4.13099289e-01 -2.82553494e-01
-2.35156622e-02 3.93217266e-01 1.70688882e-01 -1.01591241e+00
-2.46761907e-02 -1.06011462e+00 -5.18443465e-01 -5.98063409e-01
-2.60523885e-01 1.36085570e-01 2.20167115e-01 -6.99941516e-01
1.40458369e+00 -1.92435122e+00 -2.58377463e-01 6.24463439e-01
-3.60715926e-01 -4.31474775e-01 2.06702754e-01 7.19560325e-01
-2.94284433e-01 1.44817159e-01 -1.83387265e-01 -2.63704270e-01
2.53563464e-01 -1.75767258e-01 -7.19160080e-01 5.25390446e-01
1.44536287e-01 8.18544447e-01 -3.62484962e-01 2.38756120e-01
-1.18436344e-01 1.33108646e-01 -5.26969016e-01 1.57186866e-01
-1.24451950e-01 3.58620465e-01 -3.90571266e-01 4.07005191e-01
4.20907408e-01 -3.93313110e-01 -1.60081424e-02 5.91481149e-01
-5.02965569e-01 2.53324240e-01 -1.52971017e+00 7.80548334e-01
-2.96830058e-01 5.37602186e-01 -2.84387499e-01 -8.53489995e-01
1.35152316e+00 2.85853118e-01 3.63899529e-01 -2.78263092e-01
1.26872778e-01 4.20894563e-01 -5.18913455e-02 1.08864211e-01
4.46185648e-01 -4.64700043e-01 -1.34598434e-01 8.85734499e-01
-2.13459358e-01 1.55228600e-01 6.35409681e-03 -1.99783891e-01
1.19732380e+00 -8.74883682e-02 4.72973734e-01 -5.67856371e-01
9.56931338e-02 -1.97342560e-01 7.17441738e-01 9.26804245e-01
3.31818968e-01 5.66110313e-01 9.56596076e-01 -3.41194272e-01
-1.11755443e+00 -1.03050208e+00 -2.87558168e-01 9.63419437e-01
-5.78148901e-01 -1.10596836e-01 -2.70643800e-01 -1.37807414e-01
4.46691424e-01 1.03302789e+00 -7.02696502e-01 2.78329015e-01
-2.75834829e-01 -9.55536187e-01 -2.85113528e-02 5.98391652e-01
1.13508604e-01 -8.79411876e-01 -5.23068368e-01 6.64010525e-01
7.32387483e-01 -5.53854287e-01 -3.32132071e-01 4.01655465e-01
-1.01999795e+00 -7.99525142e-01 -1.05138338e+00 1.65811285e-01
1.13529757e-01 -1.84737295e-01 1.17478585e+00 -6.87993109e-01
5.15479408e-02 5.23759842e-01 -7.59887099e-02 -9.04552698e-01
-2.67603219e-01 -3.77893537e-01 -4.53466848e-02 3.75297338e-01
3.63232732e-01 -6.46540582e-01 -7.23620594e-01 3.65710914e-01
-4.81247723e-01 -7.61067390e-01 4.78669345e-01 6.50185704e-01
5.06385028e-01 -1.05006993e-01 5.87698162e-01 -1.05001545e+00
8.92334104e-01 -6.93955600e-01 -1.40655518e+00 4.12862837e-01
-1.21342301e+00 3.09177071e-01 4.29101497e-01 -5.82890689e-01
-1.12351370e+00 -1.28904879e-01 4.59856182e-01 -3.64881039e-01
4.02353913e-01 9.04376090e-01 5.00637054e-01 1.42777681e-01
3.90294820e-01 -1.79443434e-02 -8.29053596e-02 -7.64550626e-01
1.49739504e-01 3.24851185e-01 4.98070270e-01 -5.27071059e-01
7.32148945e-01 2.41657808e-01 3.82591128e-01 -3.73742878e-01
-5.15701294e-01 -3.99237126e-01 -3.52701664e-01 7.16631189e-02
6.32753849e-01 -8.84265006e-01 -8.54634702e-01 3.78759772e-01
-8.15702617e-01 -4.14571494e-01 -3.59018683e-01 9.51624155e-01
-8.27726662e-01 -4.98801209e-02 -7.56914020e-01 -1.27530181e+00
-3.83700997e-01 -6.76886320e-01 6.83051407e-01 9.26037282e-02
-3.25933635e-01 -1.37544978e+00 5.25912642e-01 -4.10015553e-01
6.92471683e-01 3.50746661e-01 8.74085605e-01 -1.24184632e+00
-5.99817038e-01 -5.65978587e-01 3.38536911e-02 3.91783297e-01
-2.33492881e-01 3.19782674e-01 -5.40684462e-01 -3.88720930e-01
3.79919410e-01 2.24684253e-01 7.17652619e-01 6.35270953e-01
6.63551927e-01 -4.63326782e-01 1.04656257e-01 5.57211161e-01
1.30604136e+00 4.00818676e-01 4.17889982e-01 7.27470815e-01
-1.08412996e-01 7.64462948e-01 6.95815325e-01 1.02802980e+00
1.91801846e-01 3.82551044e-01 4.60958034e-02 3.92999560e-01
1.01490808e+00 6.77077286e-03 5.37348092e-01 7.82983541e-01
-1.74844936e-01 1.06577262e-01 -9.54295576e-01 6.60292134e-02
-2.01895738e+00 -1.06184375e+00 -2.63949186e-01 2.67780900e+00
5.70883214e-01 2.73553938e-01 3.33075851e-01 -4.30035323e-01
5.48007369e-01 -2.85020098e-02 -5.19716203e-01 -6.40091717e-01
-2.14058816e-01 3.19942236e-01 9.90460932e-01 3.86881322e-01
-1.07161820e+00 3.51433247e-01 7.36789131e+00 3.85444194e-01
-9.11023319e-01 -2.23052204e-01 8.82183790e-01 -1.89418063e-01
-3.92426461e-01 5.14340758e-01 -1.05544329e+00 6.14925981e-01
1.43383551e+00 -9.72748399e-01 3.63587886e-01 1.15150464e+00
2.80834466e-01 -6.92147836e-02 -1.22129333e+00 6.32093370e-01
-4.62100863e-01 -1.22351003e+00 -5.94112337e-01 3.67080182e-01
8.72358084e-01 1.55968145e-01 3.01584065e-01 5.22019625e-01
7.89836228e-01 -8.97957742e-01 7.02775240e-01 1.10918498e+00
6.46669716e-02 -8.40015054e-01 7.53667057e-01 3.74166727e-01
-9.17189717e-01 -2.44431764e-01 -6.91770315e-01 -1.87873811e-01
3.26951772e-01 7.80335605e-01 -5.63557148e-01 2.50664949e-01
6.50313079e-01 5.65896869e-01 -5.51812768e-01 1.37438166e+00
2.82128006e-02 7.98366249e-01 -5.20609200e-01 -2.44485382e-02
3.96543294e-01 -8.24018419e-01 5.12248933e-01 1.09143150e+00
1.02475703e+00 2.17336640e-01 -1.83647230e-01 7.46018350e-01
4.60223883e-01 3.65452945e-01 -4.78094876e-01 3.85690592e-02
3.40506524e-01 9.29119587e-01 -4.60270435e-01 -3.53283972e-01
-7.36586094e-01 2.36955941e-01 -7.49069005e-02 5.03189683e-01
-4.62621480e-01 -6.88162372e-02 4.57040548e-01 1.74111016e-02
8.16691339e-01 -2.09994122e-01 -3.14454257e-01 -1.07677269e+00
1.65212512e-01 -6.56842589e-01 7.30146110e-01 -5.83324909e-01
-1.50998175e+00 3.74935627e-01 2.25061807e-03 -1.46734834e+00
-9.75739658e-01 -5.08498728e-01 -9.67082202e-01 1.07556450e+00
-1.35124314e+00 -1.53267264e-01 3.61238062e-01 3.99908721e-01
2.46529356e-01 -5.98286033e-01 6.79636419e-01 -1.64763764e-01
-6.92070425e-01 2.68823445e-01 9.40815747e-01 -1.47378013e-01
7.37217844e-01 -1.53684759e+00 6.40291452e-01 7.66202927e-01
-3.14252265e-02 7.05496430e-01 9.80919421e-01 -7.10955083e-01
-1.22512865e+00 -8.36581349e-01 7.31676757e-01 -3.77774477e-01
1.43680406e+00 1.37872592e-01 -1.00230742e+00 1.05390143e+00
-2.03342345e-02 -2.68849403e-01 6.05350852e-01 3.93748969e-01
-2.98624963e-01 -2.51877874e-01 -7.46334970e-01 4.08435971e-01
1.30656555e-01 -9.86887813e-02 -4.92103368e-01 2.57050842e-01
3.82663429e-01 -6.79078558e-03 -1.22486699e+00 4.45115477e-01
6.24856949e-01 -1.43518043e+00 5.60508609e-01 -8.70470345e-01
1.57025129e-01 2.69982100e-01 -2.36984447e-01 -1.29181719e+00
-4.26250428e-01 -1.40542698e+00 -2.07402676e-01 1.30328345e+00
7.38544285e-01 -1.23766565e+00 7.55716503e-01 1.24761569e+00
2.52141237e-01 -5.16463041e-01 -8.45284581e-01 -1.31362772e+00
5.04469812e-01 -3.84423763e-01 7.77357399e-01 6.62752509e-01
-2.92174131e-01 -1.54329628e-01 -4.80607897e-01 3.59918885e-02
7.84314156e-01 4.94693607e-01 8.26592803e-01 -1.38575172e+00
-8.35279763e-01 -7.38432586e-01 -2.29225010e-01 -8.66301239e-01
-4.82587144e-03 -5.70537746e-01 -2.48320118e-01 -5.60569167e-01
1.06017694e-01 -2.23723233e-01 -5.25042593e-01 2.14071237e-02
-5.80141991e-02 -4.12526906e-01 1.04328707e-01 3.13426107e-01
-2.96418101e-01 3.16050678e-01 5.81869185e-01 2.53772378e-01
-2.44449064e-01 7.64198005e-01 -4.53689754e-01 6.84792817e-01
8.22973490e-01 -6.26547456e-01 -8.19732249e-02 1.38214648e-01
4.33493763e-01 7.47245014e-01 1.96643278e-01 -4.64525193e-01
2.11681455e-01 -3.54455084e-01 3.75068307e-01 -6.23226166e-01
1.11634552e-01 -4.64506149e-01 7.23555267e-01 3.34652722e-01
-3.15426707e-01 6.42659605e-01 -6.85809478e-02 6.79608524e-01
-4.52279031e-01 -5.54923236e-01 6.23674810e-01 -8.76483619e-02
-2.87654757e-01 2.52823621e-01 -4.05067980e-01 9.90137309e-02
1.06816876e+00 4.65622470e-02 -2.13478088e-01 -9.69323158e-01
-8.59788835e-01 4.12528604e-01 3.49686384e-01 1.30854994e-01
2.50896692e-01 -1.10364497e+00 -9.20791268e-01 5.85427415e-03
-2.35718504e-01 -6.29387319e-01 -6.82617947e-02 1.12222755e+00
-3.38389307e-01 7.34446764e-01 1.44849628e-01 -2.58905113e-01
-9.39799190e-01 6.07889235e-01 2.84139842e-01 -3.57895583e-01
-5.03228128e-01 5.86098313e-01 2.05992624e-01 5.26071936e-02
2.02049524e-01 -1.73431233e-01 7.85895530e-03 3.80053699e-01
6.83435678e-01 6.81299984e-01 -3.13867033e-02 -1.35713875e-01
-5.66386916e-02 5.45341253e-01 5.99977002e-02 -4.01841193e-01
1.65925539e+00 -1.71512291e-01 -1.27814159e-01 9.16153729e-01
1.04460227e+00 -8.73848498e-02 -1.60983634e+00 -2.34345108e-01
7.91954458e-01 -3.84956598e-01 -1.68682173e-01 -3.65819156e-01
-9.26616609e-01 8.23533177e-01 2.79406458e-01 5.84114969e-01
8.01112831e-01 -2.91756660e-01 3.14603180e-01 6.33190513e-01
3.02479506e-01 -1.01287329e+00 -2.99313068e-01 4.87763196e-01
8.44709516e-01 -1.03702712e+00 2.35104084e-01 2.15505347e-01
-8.76050234e-01 1.38422596e+00 -2.95478135e-01 -5.26465297e-01
1.18367124e+00 3.33015829e-01 1.57857463e-01 6.06449917e-02
-1.32967675e+00 1.21283293e-01 3.49435925e-01 9.32174921e-02
3.33064795e-01 1.47459999e-01 -2.58409590e-01 6.84016049e-01
-6.31654382e-01 -3.89953673e-01 7.16090441e-01 7.45615840e-01
-5.06929219e-01 -1.14849651e+00 -3.84308845e-01 7.47517288e-01
-9.42677319e-01 -4.64305356e-02 -2.58303583e-01 8.63272429e-01
-1.01004529e+00 8.32482576e-01 3.25175256e-01 1.27970710e-01
3.85375738e-01 2.21293718e-01 -9.08078104e-02 -4.60309774e-01
-7.18511820e-01 6.08830094e-01 1.06608123e-01 -4.61991131e-01
1.88297369e-02 -1.35887992e+00 -5.43952763e-01 -5.99105537e-01
-3.10921043e-01 2.63808578e-01 6.32836938e-01 7.83728242e-01
6.98531270e-02 1.00484714e-01 1.14601493e+00 -6.12431228e-01
-1.58874881e+00 -9.54270542e-01 -1.12508297e+00 1.20203204e-01
2.03488812e-01 -3.15439999e-01 -6.82565391e-01 -2.21874043e-01] | [4.913432598114014, 4.040506839752197] |
989ee75d-5028-428c-95f5-f9cba67d4a4e | conceptual-study-and-performance-analysis-of | 2306.10246 | null | https://arxiv.org/abs/2306.10246v1 | https://arxiv.org/pdf/2306.10246v1.pdf | Conceptual Study and Performance Analysis of Tandem Dual-Antenna Spaceborne SAR Interferometry | Multi-baseline synthetic aperture radar interferometry (MB-InSAR), capable of mapping 3D surface model with high precision, is able to overcome the ill-posed problem in the single-baseline InSAR by use of the baseline diversity. Single pass MB acquisition with the advantages of high coherence and simple phase components has a more practical capability in 3D reconstruction than conventional repeat-pass MB acquisition. Using an asymptotic 3D phase unwrapping (PU), it is possible to get a reliable 3D reconstruction using very sparse acquisitions but the interferograms should follow the optimal baseline design. However, current spaceborne SAR system doesn't satisfy this principle, inducing more difficulties in practical application. In this article, a new concept of Tandem Dual-Antenna SAR Interferometry (TDA-InSAR) system for single-pass reliable 3D surface mapping using the asymptotic 3D PU is proposed. Its optimal MB acquisition is analyzed to achieve both good relative height precision and flexible baseline design. Two indicators, i.e., expected relative height precision and successful phase unwrapping rate, are selected to optimize the system parameters and evaluate the performance of various baseline configurations. Additionally, simulation-based demonstrations are conducted to evaluate the performance in typical scenarios and investigate the impact of various error sources. The results indicate that the proposed TDA-InSAR is able to get the specified MB acquisition for the asymptotic 3D PU, which offers a feasible solution for single-pass 3D SAR imaging. | ['YaQiu Jin', 'Chibiao Ding', 'Xiaolan Qiu', 'Feng Xu', 'Fengming Hu'] | 2023-06-17 | null | null | null | null | ['3d-reconstruction'] | ['computer-vision'] | [ 3.32712024e-01 -3.28775048e-01 4.54780459e-01 -2.88084328e-01
-8.66568446e-01 -3.49671632e-01 3.88528347e-01 -4.08616185e-01
-1.84835538e-01 7.83397257e-01 -2.18402594e-01 -1.79260045e-01
-8.80369186e-01 -9.19494629e-01 -2.91399896e-01 -1.00825250e+00
-4.48101670e-01 6.18496537e-01 -2.99434885e-02 -5.56705356e-01
1.63963616e-01 8.30752254e-01 -1.21172118e+00 -5.46832740e-01
1.36073303e+00 7.68907905e-01 4.09462571e-01 8.24702680e-02
4.22857583e-01 -8.50817114e-02 -4.37865138e-01 4.22284693e-01
5.97783983e-01 -5.49917519e-01 -1.79697439e-01 -1.49350971e-01
3.69215786e-01 -3.58512759e-01 1.82670876e-01 1.16385865e+00
4.43949133e-01 3.90124886e-04 5.67167103e-01 -5.81156492e-01
5.29039577e-02 2.35259101e-01 -9.30101275e-01 9.68135670e-02
2.47202218e-01 -1.03281319e-01 5.77586710e-01 -7.92133927e-01
1.91580191e-01 6.72425747e-01 6.23028815e-01 -3.92861098e-01
-1.20486879e+00 -5.05697131e-01 -6.63196266e-01 2.18657330e-02
-1.61518228e+00 -2.49678701e-01 8.49411190e-01 -5.21552563e-01
6.95647657e-01 3.31797659e-01 8.61027002e-01 1.79959297e-01
4.81984288e-01 -1.33266151e-01 1.53621292e+00 -5.19936144e-01
6.12058528e-02 -2.65890539e-01 4.00873423e-01 2.89496869e-01
1.14600706e+00 4.48717326e-01 -2.01170638e-01 2.45536049e-03
7.08415508e-01 -3.12572151e-01 -8.72346520e-01 -5.02719939e-01
-1.27644444e+00 6.38007581e-01 4.76269633e-01 5.53487659e-01
-8.33251059e-01 -4.86809909e-01 -2.08980009e-01 3.39118332e-01
2.58082151e-01 9.32166636e-01 -2.06288412e-01 2.85482466e-01
-1.23931742e+00 4.15535897e-01 4.14218277e-01 5.59770763e-01
8.51260185e-01 8.08269441e-01 1.23524435e-01 2.96357334e-01
3.77260387e-01 1.64282191e+00 2.05578104e-01 -4.33982253e-01
2.16478318e-01 3.60944301e-01 3.98317456e-01 -8.20512831e-01
-5.17833292e-01 -1.16071951e+00 -8.97875130e-01 3.59610468e-01
2.96379346e-02 -2.13383794e-01 -7.70306766e-01 1.33592105e+00
2.56636977e-01 -1.25560939e-01 6.78406179e-01 9.87161338e-01
2.57788002e-01 8.32016647e-01 -5.78933001e-01 -8.23460996e-01
1.40373075e+00 -1.32907793e-01 -9.43425357e-01 -7.58286715e-01
4.29697424e-01 -9.71577585e-01 5.00393808e-01 4.37941730e-01
-6.94338500e-01 -4.33289230e-01 -1.60117018e+00 8.89494419e-01
3.15245569e-01 3.64402443e-01 3.69212568e-01 6.06732726e-01
-6.76674604e-01 2.75427282e-01 -7.77270675e-01 -9.92694572e-02
-2.70202488e-01 4.91091125e-02 -2.59387255e-01 -7.35755041e-02
-1.10685730e+00 1.03006279e+00 3.65416110e-01 5.38658142e-01
-3.61055791e-01 -8.32476437e-01 -8.60077441e-01 -1.47918954e-01
-7.78361112e-02 -5.95811784e-01 6.34904504e-01 -4.54270482e-01
-1.38269401e+00 5.88767469e-01 -6.75503463e-02 -8.45174491e-01
-1.60444528e-02 -6.53236210e-01 -5.93074739e-01 3.85369629e-01
2.08804682e-01 -1.46050051e-01 6.94230378e-01 -1.05518758e+00
-7.11191595e-02 -6.10931993e-01 -5.21066785e-01 3.98697674e-01
3.85232389e-01 -4.41935807e-01 6.93667293e-01 -2.43403688e-01
1.05658269e+00 -9.19035733e-01 -2.55772591e-01 -7.44945347e-01
-9.60161835e-02 7.08395720e-01 7.95877755e-01 -5.26309550e-01
8.25056314e-01 -2.16777444e+00 -3.50293815e-02 3.46210808e-01
-4.66005355e-01 2.61448056e-01 1.75425872e-01 5.96267879e-01
7.41857709e-03 -5.72131991e-01 -6.47449017e-01 3.11436027e-01
-7.46079385e-01 -1.21093027e-01 -3.58565062e-01 8.68982673e-01
-8.62977060e-04 6.03110790e-01 -3.54266465e-01 4.50482927e-02
2.21519515e-01 4.24932778e-01 -2.26458862e-01 1.90147132e-01
8.24822485e-02 6.93601787e-01 -6.42286658e-01 6.11107767e-01
1.37549090e+00 1.66513026e-01 2.53005207e-01 -7.93150663e-01
-9.03638899e-01 6.09961487e-02 -1.33514667e+00 1.23569083e+00
-2.97720343e-01 5.10845065e-01 5.48658013e-01 -9.62950647e-01
1.65889597e+00 8.09043571e-02 3.41518551e-01 -1.03916025e+00
-4.39726487e-02 7.32312858e-01 2.89539516e-01 -3.58355850e-01
4.84190583e-01 -5.77212632e-01 -1.27416193e-01 4.04967368e-02
-6.16206169e-01 -7.48660386e-01 -3.24295998e-01 -2.73018569e-01
5.87999284e-01 1.91197261e-01 7.48713911e-01 -9.13532853e-01
8.06476235e-01 4.45959806e-01 6.27722442e-01 3.26126277e-01
3.59005183e-01 3.67826045e-01 -2.79615432e-01 -3.45053017e-01
-7.34756708e-01 -1.00666916e+00 -4.36309099e-01 -5.70552826e-01
5.96699059e-01 4.13860887e-01 -1.21316247e-01 3.13899100e-01
-1.20974749e-01 7.92395353e-01 2.80644018e-02 8.72565508e-02
-8.02013814e-01 -1.23954391e+00 1.53688610e-01 -2.12927878e-01
1.06336200e+00 -3.53132248e-01 -1.31412339e+00 2.11019084e-01
-3.24674815e-01 -9.35422242e-01 3.52711320e-01 -8.95130783e-02
-1.31376374e+00 -1.03534889e+00 -4.63593453e-01 -3.84862006e-01
4.32514191e-01 9.92342651e-01 7.71250904e-01 -4.12232041e-01
-7.43550807e-02 1.39977813e-01 -6.35589123e-01 -3.52074429e-02
-1.60507560e-01 -3.50119114e-01 2.51354337e-01 6.22259900e-02
1.81633569e-02 -9.12404001e-01 -6.88550472e-01 5.67822635e-01
-4.99218851e-01 9.31366980e-02 9.98326898e-01 7.51526892e-01
6.66663885e-01 1.15655564e-01 6.55149579e-01 -3.98522705e-01
1.36049315e-01 -1.22247562e-02 -1.15345228e+00 5.08839600e-02
-6.53299630e-01 8.51770211e-03 1.70119852e-01 8.81842226e-02
-1.27799308e+00 -1.89293087e-01 -1.48235634e-01 1.54896125e-01
4.54853848e-02 9.49346781e-01 -1.72898754e-01 -5.36581457e-01
8.60815942e-01 5.89418232e-01 3.71715039e-01 -4.61316794e-01
-1.73193291e-01 5.06805539e-01 3.45481277e-01 -1.62901297e-01
1.40572023e+00 6.18261278e-01 5.71689188e-01 -1.54095721e+00
-9.73254859e-01 -5.16815364e-01 -4.56179947e-01 -5.14076538e-02
7.28094339e-01 -1.30225146e+00 -2.15434223e-01 6.06850863e-01
-6.52787447e-01 -1.23201095e-01 -1.19830348e-01 1.31449938e+00
-4.02488917e-01 7.00317383e-01 5.62373325e-02 -1.02436578e+00
-8.93052995e-01 -8.97729814e-01 8.12294364e-01 3.18111688e-01
-1.04830571e-01 -4.97149497e-01 2.12510809e-01 2.34436139e-01
6.77455127e-01 5.83655357e-01 5.80289006e-01 -1.08245455e-01
-1.02675474e+00 -1.07782558e-01 -3.15162987e-02 9.21569392e-02
2.94433326e-01 -4.94422793e-01 -4.07189459e-01 -7.09563255e-01
6.60464168e-01 7.91289732e-02 5.42089522e-01 9.20619726e-01
-2.99250960e-01 -5.41799925e-02 -3.37431133e-01 9.92909372e-01
2.14005566e+00 5.61302900e-01 7.98030496e-01 6.24407470e-01
-1.46883177e-02 3.30432028e-01 1.59158957e+00 3.96874607e-01
-2.76533157e-01 8.02388668e-01 5.45607686e-01 3.50853503e-02
1.61880955e-01 1.98572785e-01 2.50938654e-01 5.58196902e-01
-2.02571705e-01 4.36909758e-02 -6.13681138e-01 5.36996305e-01
-1.24007916e+00 -9.29687321e-01 -6.15798831e-01 2.71740079e+00
1.31708518e-01 6.35259971e-02 -6.34137928e-01 2.56211370e-01
2.63851613e-01 4.64487165e-01 1.15256302e-01 6.43159449e-02
-4.99264330e-01 4.38705802e-01 8.44347775e-01 8.05157721e-01
-7.39911377e-01 4.13950503e-01 4.95754480e+00 3.59799802e-01
-1.27636313e+00 -6.59666732e-02 -4.32830215e-01 4.66661364e-01
-6.06121480e-01 3.73721093e-01 -1.00123465e+00 -5.41382749e-03
7.04253554e-01 -2.44411275e-01 -1.79748893e-01 3.16518962e-01
4.61455286e-01 -6.81534290e-01 -1.74977854e-01 8.12424064e-01
-1.12796977e-01 -1.33550382e+00 1.09702408e-01 1.67229623e-01
5.87551296e-01 -6.24648072e-02 -2.17199281e-01 -1.16170473e-01
-5.10859668e-01 -3.58081132e-01 4.13849533e-01 5.52508891e-01
9.37345684e-01 -6.81113601e-01 9.58642781e-01 4.84934211e-01
-1.14794374e+00 6.94143325e-02 -1.46759242e-01 -2.52938211e-01
8.59833300e-01 1.17028940e+00 -7.89894700e-01 1.51813519e+00
3.45866293e-01 5.02636552e-01 -2.76700463e-02 1.14475167e+00
-2.38393024e-01 4.19478893e-01 -6.04501486e-01 1.76682174e-01
3.15514863e-01 -9.52453077e-01 1.09086919e+00 4.28201020e-01
1.37460160e+00 5.67561686e-01 4.07901965e-02 6.53916597e-01
1.00584459e+00 3.71705107e-02 -8.64934802e-01 1.71609432e-01
5.81988692e-01 9.13316071e-01 -2.14285061e-01 1.20383389e-01
-7.42150620e-02 2.70042926e-01 -7.79512644e-01 3.32827687e-01
-5.79393864e-01 -2.95016676e-01 7.40839124e-01 7.00212836e-01
3.04124206e-01 -8.95145476e-01 -1.17599972e-01 -7.66451299e-01
-6.78747054e-03 -6.73862875e-01 2.04646692e-01 -9.79450285e-01
-4.34250325e-01 7.92789876e-01 3.72944087e-01 -1.82867193e+00
-2.39921033e-01 -2.24169880e-01 -4.71792638e-01 1.09832215e+00
-1.73283756e+00 -1.21448183e+00 -5.43927550e-01 1.87376812e-01
2.02464417e-01 -1.81753904e-01 8.95032525e-01 5.89314401e-02
-3.44731733e-02 -2.08834067e-01 1.51017860e-01 -5.03153205e-01
5.74089468e-01 -5.55716991e-01 7.69696292e-03 1.15345228e+00
-2.03109488e-01 4.63016748e-01 1.43100822e+00 -1.08073330e+00
-1.71825397e+00 -6.42637312e-01 6.09862685e-01 4.81214195e-01
3.65596533e-01 1.75722376e-01 -8.41097713e-01 6.43846273e-01
1.00922160e-01 -4.34176058e-01 2.57488966e-01 -1.87694892e-01
1.43999279e-01 -6.88827872e-01 -1.20491755e+00 1.61973387e-01
5.95617175e-01 1.03799246e-01 -9.18097734e-01 1.54063866e-01
5.10103047e-01 -4.21431422e-01 -8.17406714e-01 1.37422991e+00
5.72711170e-01 -1.26658237e+00 1.03940535e+00 4.71172035e-01
-2.37494156e-01 -7.71787703e-01 -4.12897199e-01 -1.36030209e+00
-3.10775191e-01 -6.80330276e-01 4.37161118e-01 8.21113586e-01
2.14120552e-01 -1.22336006e+00 5.97338200e-01 -5.85793197e-01
-3.75016272e-01 -2.20577329e-01 -1.21808457e+00 -1.11839771e+00
-4.20169771e-01 1.65402383e-01 2.97823846e-01 6.57484591e-01
-5.12561858e-01 3.93166333e-01 -4.05403316e-01 1.15542901e+00
1.28064740e+00 9.14460659e-01 7.65872478e-01 -1.55580032e+00
-2.38852650e-01 2.61560678e-01 -2.26911426e-01 -1.06588697e+00
-3.27629864e-01 -3.50930303e-01 3.87763157e-02 -1.47229314e+00
-5.47642648e-01 -7.92333245e-01 3.47785830e-01 -3.51075590e-01
3.28358620e-01 7.21924454e-02 -2.70251572e-01 5.07767916e-01
5.16679049e-01 7.48904526e-01 1.23562062e+00 4.91138622e-02
-5.04191518e-01 4.56715792e-01 -5.00041097e-02 4.41816270e-01
6.47621512e-01 -2.16631383e-01 -5.09737313e-01 -5.33268809e-01
3.48279476e-01 7.09042966e-01 3.32401723e-01 -1.57754076e+00
1.08399138e-01 -1.33389190e-01 2.65020281e-02 -1.24436045e+00
6.71699822e-01 -9.05545473e-01 9.43033695e-01 7.74784386e-01
7.18679965e-01 -1.73418313e-01 1.59153029e-01 3.37572813e-01
-7.20353425e-01 -3.70014280e-01 1.10220671e+00 -5.86799113e-03
-8.14569950e-01 1.91057322e-03 -1.91534758e-01 -3.75424922e-01
9.79639411e-01 -5.14319003e-01 -4.07767951e-01 -1.62449628e-01
-3.75567526e-01 3.86850946e-02 4.80006009e-01 -3.92193764e-01
7.90873826e-01 -1.01123261e+00 -9.17001843e-01 7.18716681e-01
1.29819006e-01 -8.23362544e-02 8.94449949e-01 1.19880068e+00
-9.45280254e-01 5.36557555e-01 -4.63118136e-01 -8.16599965e-01
-1.10425806e+00 4.71962895e-03 5.24349272e-01 -3.34065855e-01
-8.06805611e-01 2.52533853e-01 -3.68355922e-02 -1.26434147e-01
-7.93083429e-01 7.80670792e-02 -2.63491422e-01 1.72478169e-01
6.01159573e-01 -1.00428434e-02 1.97478101e-01 -5.40021122e-01
-4.41331983e-01 1.30750299e+00 4.60841298e-01 -1.71302825e-01
1.38961399e+00 -2.53885776e-01 -1.56222403e-01 9.16277841e-02
5.67283511e-01 5.76523781e-01 -8.21454346e-01 -7.99525753e-02
-2.60274172e-01 -5.89878380e-01 2.57289290e-01 -5.23694873e-01
-5.46153009e-01 5.87622046e-01 5.71345747e-01 1.50508985e-01
1.32666624e+00 -3.88362855e-01 5.75703681e-01 5.03116667e-01
8.55742514e-01 -3.40466112e-01 -3.60446453e-01 4.94134337e-01
9.88245249e-01 -8.46479654e-01 6.95361853e-01 -6.99215055e-01
-3.20461988e-01 1.11227834e+00 1.75048754e-01 -3.14649850e-01
5.32739520e-01 2.46514156e-01 3.00101973e-02 -4.44729835e-01
-3.13557461e-02 -2.42236421e-01 7.76330829e-02 5.26098192e-01
1.41688049e-01 3.00315738e-01 -8.49496603e-01 1.79341331e-01
-1.71639547e-01 -3.77559483e-01 8.01667809e-01 8.05100739e-01
-1.01611221e+00 -9.03431118e-01 -8.33738804e-01 1.36259034e-01
2.05638446e-02 7.30262548e-02 4.83107209e-01 1.22717059e+00
-3.66407782e-01 5.22867560e-01 1.49221346e-01 -2.92791128e-02
6.24286950e-01 -4.60537300e-02 6.28418624e-01 -4.20665681e-01
-7.03232065e-02 2.50781178e-01 4.70755190e-01 -3.06476384e-01
-7.81291187e-01 -7.00122893e-01 -1.03029835e+00 1.98243693e-01
-3.87620032e-01 7.39340365e-01 6.61656857e-01 8.80157053e-01
2.51604140e-01 6.50906637e-02 7.32791245e-01 -5.13339758e-01
-7.29568303e-01 -1.01944435e+00 -1.05256402e+00 -5.10710478e-01
3.98701161e-01 -7.91133106e-01 -6.21198952e-01 -6.00871980e-01] | [6.790736198425293, 0.9825596809387207] |
33a97f19-7046-435d-8f90-04bb6843b61d | face-alignment-by-explicit-shape-regression | null | null | https://ieeexplore.ieee.org/document/6248015 | https://www.microsoft.com/en-us/research/wp-content/uploads/2013/01/Face-Alignment-by-Explicit-Shape-Regression.pdf | Face alignment by explicit shape regression | We present a very efficient, highly accurate, “Explicit Shape Regression” approach for face alignment. Unlike previous regression-based approaches, we directly learn a vectorial regression function to infer the whole facial shape (a set of facial landmarks) from the image and explicitly minimize the alignment errors over the training data. The inherent shape constraint is naturally encoded into the regressor in a cascaded learning framework and applied from coarse to fine during the test, without using a fixed parametric shape model as in most previous methods. To make the regression more effective and efficient, we design a two-level boosted regression, shape-indexed features and a correlation-based feature selection method. This combination enables us to learn accurate models from large training data in a short time (20 minutes for 2,000 training images), and run regression extremely fast in test (15 ms for a 87 landmarks shape). Experiments on challenging data show that our approach significantly outperforms the state-of-the-art in terms of both accuracy and efficiency. | ['Jian Sun', 'Fang Wen', 'Yichen Wei', 'Xudong Cao'] | 2013-12-13 | null | null | null | int-j-comput-vis-2013-12 | ['face-alignment'] | ['computer-vision'] | [-9.96401161e-02 -1.13019466e-01 -1.63424566e-01 -8.04936469e-01
-9.78343964e-01 -3.62249732e-01 4.01143730e-01 -7.39525557e-02
-3.65307927e-01 4.36469227e-01 -8.65191668e-02 -4.58966494e-02
1.07092731e-01 -5.34688115e-01 -8.48883212e-01 -4.75669086e-01
-5.49307503e-02 8.21783841e-01 4.92272619e-03 -1.63182870e-01
3.00590843e-01 1.02716863e+00 -1.36655390e+00 -1.65930152e-01
3.11405122e-01 9.56421793e-01 -3.03504676e-01 5.94475508e-01
7.71031231e-02 4.33430642e-01 -2.25523829e-01 -6.25699878e-01
4.37110037e-01 -7.39724413e-02 -6.58580303e-01 3.54357123e-01
9.31650162e-01 -2.89777577e-01 -2.43249862e-03 6.43117487e-01
6.31534755e-01 6.28761128e-02 5.49831390e-01 -1.06633663e+00
-1.95139706e-01 -8.71965736e-02 -1.08504605e+00 -1.86410204e-01
2.88944721e-01 -1.10581756e-01 8.62641871e-01 -1.53094923e+00
5.44103026e-01 1.30548155e+00 9.25469279e-01 6.43330216e-01
-1.51279008e+00 -8.04450035e-01 1.49061471e-01 -9.32400972e-02
-1.63465023e+00 -9.40865040e-01 7.82247722e-01 -4.18087989e-01
8.51654112e-01 2.93308586e-01 6.38227940e-01 7.19634712e-01
-6.18551672e-02 4.00968045e-01 9.22489345e-01 -4.45335954e-01
1.01874005e-02 4.78425734e-02 -1.60803303e-01 1.21286035e+00
2.22939886e-02 5.47882579e-02 -5.15055060e-01 -2.67965764e-01
1.11563587e+00 -5.10940924e-02 1.13797642e-01 -6.25980794e-01
-6.65033460e-01 8.64120722e-01 3.96828771e-01 2.71495860e-02
-2.53640443e-01 3.09966981e-01 2.04278380e-01 3.56358737e-01
5.81784964e-01 2.01067179e-01 -6.47418678e-01 2.45617349e-02
-1.30237079e+00 3.06940168e-01 7.55515337e-01 6.80907011e-01
1.02717745e+00 8.15574825e-02 1.47933692e-01 1.00274277e+00
3.92766029e-01 5.17707944e-01 1.38944611e-01 -9.90137160e-01
2.26688623e-01 5.24908483e-01 -1.62510872e-01 -1.18442059e+00
-5.74967504e-01 -2.05029726e-01 -7.02274680e-01 5.26867211e-01
5.51516354e-01 -2.22459629e-01 -9.24511731e-01 1.58605540e+00
7.65235126e-01 3.80471438e-01 -5.50381124e-01 7.40954459e-01
5.95356286e-01 2.24545151e-01 -2.97941566e-01 -4.34757501e-01
1.13058472e+00 -9.28182304e-01 -3.20469290e-01 -1.33990258e-01
5.20986557e-01 -8.70698810e-01 8.03482711e-01 4.52220887e-01
-1.26958108e+00 -5.12620389e-01 -8.33015680e-01 -7.77111873e-02
8.62864126e-03 2.33604565e-01 7.33866334e-01 5.62122226e-01
-1.26087260e+00 6.99415267e-01 -9.05335665e-01 -1.44121185e-01
7.48628318e-01 9.94553745e-01 -8.00699055e-01 -5.51262349e-02
-3.54737282e-01 9.48055208e-01 -3.39483023e-01 1.56085640e-01
-7.35385120e-01 -8.27607691e-01 -8.82216156e-01 -9.84889045e-02
4.76669371e-01 -5.11622906e-01 1.10867035e+00 -1.17654729e+00
-1.69847095e+00 1.07229996e+00 -6.42307103e-01 2.01452393e-02
4.62178558e-01 -2.44060278e-01 5.61119094e-02 -2.01800689e-02
-2.12805927e-01 7.29245663e-01 1.35133624e+00 -1.19485676e+00
-1.47917479e-01 -7.18316853e-01 -3.07681441e-01 7.97047615e-02
-1.18935153e-01 4.20744658e-01 -8.59543264e-01 -3.70469928e-01
1.52788550e-01 -9.90190029e-01 -4.70511079e-01 2.99425870e-01
1.33332182e-02 -1.97352767e-01 6.11244977e-01 -7.47759581e-01
9.84216750e-01 -1.92009103e+00 1.79721966e-01 6.99118018e-01
3.54738265e-01 1.92532048e-01 -4.07570064e-01 -8.42599645e-02
-2.68430293e-01 -1.09522618e-01 -2.06235409e-01 -6.34060025e-01
-3.84217948e-01 9.02924687e-02 -5.48403114e-02 7.50585020e-01
3.42233807e-01 1.03350437e+00 -5.81740499e-01 -5.94760954e-01
-4.37587127e-02 7.02906609e-01 -8.10653627e-01 2.86787570e-01
8.72657746e-02 6.16067410e-01 -3.28058451e-01 7.85631418e-01
8.07547808e-01 -7.27025196e-02 1.62429884e-01 -2.18343839e-01
-6.26817942e-02 -8.47499296e-02 -1.11019874e+00 1.66433132e+00
-6.71860278e-01 4.05555397e-01 1.08510010e-01 -1.01251769e+00
1.16655970e+00 2.36721024e-01 8.60573769e-01 -6.05950236e-01
5.67121729e-02 1.58475637e-01 -2.32480198e-01 -1.54152036e-01
1.87262297e-02 -3.61643732e-02 2.46078596e-01 4.46542472e-01
3.39547426e-01 -2.73264170e-01 -1.04045630e-01 -2.93566793e-01
8.08257937e-01 5.07372916e-01 4.00130630e-01 -1.58632517e-01
7.39836156e-01 -2.88135558e-01 5.87481260e-01 1.57379061e-01
1.63447186e-01 7.16985285e-01 4.77884680e-01 -8.35488737e-01
-1.07230592e+00 -6.04378343e-01 -1.97081000e-01 1.29356635e+00
-4.16221797e-01 -4.72327173e-01 -9.28084373e-01 -8.12959671e-01
2.94660956e-01 6.53389245e-02 -6.81004167e-01 1.87540308e-01
-1.04422414e+00 -6.86035275e-01 2.47092918e-01 6.63579106e-01
1.68913797e-01 -7.93964565e-01 -1.57638267e-01 8.31553489e-02
2.89890766e-01 -1.08331037e+00 -6.58476770e-01 -6.39717504e-02
-1.03591859e+00 -1.06928301e+00 -4.55028445e-01 -7.69519508e-01
1.18754625e+00 2.42752023e-02 1.12446928e+00 4.85972732e-01
-6.13227725e-01 2.81228423e-01 1.80650681e-01 -2.21628249e-01
1.25208005e-01 -2.06654415e-01 7.80258626e-02 1.69662029e-01
3.66376966e-01 -6.28633797e-01 -3.77427995e-01 4.27673608e-01
-2.02088535e-01 -2.78674394e-01 6.19854391e-01 1.01533377e+00
7.95739591e-01 -4.66524452e-01 1.99615732e-01 -8.40399921e-01
1.54604807e-01 -1.40948936e-01 -1.06605101e+00 1.78754777e-01
-6.86804414e-01 1.54174775e-01 5.22598982e-01 -4.61507499e-01
-6.15171611e-01 5.17880440e-01 -3.25062603e-01 -7.08162904e-01
1.47406548e-01 2.06577748e-01 8.89445618e-02 -7.45672226e-01
6.16704404e-01 1.13034196e-01 4.17165846e-01 -5.51701486e-01
3.83124143e-01 2.72760779e-01 4.13362950e-01 -7.27733195e-01
1.10164237e+00 4.76650417e-01 4.94480073e-01 -8.06710839e-01
-7.83965945e-01 -2.99759775e-01 -1.31323719e+00 -1.19391039e-01
4.28838581e-01 -8.26005280e-01 -9.89467502e-01 1.67712927e-01
-9.66237783e-01 -3.06795448e-01 -9.79500487e-02 3.60735387e-01
-6.59129202e-01 1.12930782e-01 -2.51596838e-01 -6.74167395e-01
-3.67630750e-01 -1.11038435e+00 1.19621837e+00 4.98236902e-02
-3.48830104e-01 -9.17500556e-01 2.24075064e-01 2.45702982e-01
3.78902495e-01 3.34223390e-01 4.91616100e-01 -4.94907886e-01
-5.30921519e-01 -3.31314981e-01 -2.94190556e-01 1.92240313e-01
1.00367531e-01 3.29229742e-01 -9.19888616e-01 -4.38035160e-01
-1.53522640e-01 -4.26573783e-01 6.52798176e-01 3.91221732e-01
1.36032891e+00 -4.21105504e-01 -2.76349217e-01 9.76245105e-01
1.32615435e+00 -3.09605509e-01 4.38575566e-01 -5.68043180e-02
8.40092123e-01 5.40450215e-01 6.42513096e-01 3.41440856e-01
2.91049331e-01 9.81886864e-01 2.70547509e-01 -4.89203185e-01
-6.65165707e-02 -1.48363754e-01 2.05744147e-01 6.19826674e-01
-5.72375059e-01 6.65416718e-01 -7.76275635e-01 3.01995486e-01
-1.72631466e+00 -9.50769365e-01 1.45169884e-01 2.55810547e+00
9.69155073e-01 -3.02157789e-01 3.17673236e-01 -2.11989023e-02
3.41346532e-01 -1.84413940e-02 -4.10388321e-01 -6.24386549e-01
3.24538589e-01 7.06346512e-01 3.72176915e-01 8.49400640e-01
-1.04641140e+00 1.27003682e+00 7.64702082e+00 6.34081423e-01
-1.40585470e+00 -3.95636447e-02 8.20831180e-01 -2.51741171e-01
-1.16482310e-01 -1.69669181e-01 -1.02624226e+00 8.99311900e-03
8.05781662e-01 8.03664103e-02 7.58137047e-01 1.05702531e+00
1.19095266e-01 1.23809449e-01 -1.20304894e+00 1.23097134e+00
2.92786330e-01 -1.23491299e+00 -2.89001524e-01 1.92544252e-01
6.36457264e-01 -3.48892175e-02 2.62160972e-02 1.17350303e-01
2.13266239e-01 -1.56492329e+00 3.78640592e-01 5.58638930e-01
1.21759140e+00 -9.53642190e-01 3.86853695e-01 7.72741735e-02
-1.25251555e+00 5.42819798e-02 -5.81400812e-01 2.06939220e-01
-3.50879937e-01 4.86995727e-01 -8.06365967e-01 1.96176395e-01
5.14623642e-01 5.58007538e-01 -5.67546546e-01 7.20797956e-01
-1.67031437e-01 4.64152306e-01 -5.22494853e-01 1.78212419e-01
-2.16627866e-02 -3.39987695e-01 1.81265041e-01 1.07045853e+00
1.91631496e-01 9.17628407e-02 8.98974314e-02 5.54352105e-01
-2.31054842e-01 3.98303896e-01 -4.41241145e-01 3.86290997e-01
3.14642489e-01 1.73003805e+00 -5.06739080e-01 -2.07082648e-02
-5.63233078e-01 8.73025894e-01 7.43984759e-01 1.15195341e-01
-7.16359854e-01 6.36056215e-02 6.80487216e-01 2.63060272e-01
5.50181031e-01 -3.46077293e-01 -3.48058790e-01 -1.01720059e+00
1.94720536e-01 -9.94218707e-01 4.98796478e-02 -2.71055073e-01
-1.04295027e+00 6.18449152e-01 -2.47695670e-01 -9.19987321e-01
-5.05971253e-01 -6.24448955e-01 -6.06035233e-01 8.90964508e-01
-1.64985108e+00 -1.36931098e+00 -3.23942095e-01 8.58617842e-01
3.34911734e-01 -2.61725634e-01 1.07228470e+00 2.72927254e-01
-5.31434059e-01 1.07284224e+00 -2.91141450e-01 3.28868449e-01
8.84333193e-01 -1.01797616e+00 4.83434618e-01 4.10998464e-01
4.61285412e-01 7.17344999e-01 3.88036847e-01 -3.31593424e-01
-1.61715806e+00 -9.70683396e-01 9.92084265e-01 -6.44499540e-01
4.34535742e-01 -4.46163118e-01 -8.94811690e-01 7.73883879e-01
-2.42883846e-01 7.32786596e-01 7.87780344e-01 4.54755396e-01
-6.57102108e-01 -5.28995931e-01 -1.21986878e+00 5.32963872e-01
1.09062290e+00 -3.23938072e-01 -2.73783118e-01 2.84635007e-01
1.64025173e-01 -5.62635481e-01 -9.09785628e-01 4.43886131e-01
9.52435076e-01 -7.46691585e-01 1.07058239e+00 -7.38871634e-01
2.11736187e-01 -1.17952995e-01 4.83995071e-03 -1.04307985e+00
-4.18075055e-01 -8.98409069e-01 -2.85633534e-01 9.35193777e-01
5.31320810e-01 -4.73291337e-01 9.02781546e-01 6.09834373e-01
3.00872743e-01 -1.33697593e+00 -9.71027792e-01 -4.33362633e-01
-1.32524725e-02 -2.16186821e-01 7.52151251e-01 8.19536448e-01
-2.66380847e-01 2.95561880e-01 -4.23597544e-01 1.25369847e-01
6.47824466e-01 1.31577164e-01 1.16742671e+00 -1.39454389e+00
-1.38067037e-01 -4.14500624e-01 -6.41446888e-01 -9.54993665e-01
3.42135102e-01 -6.87578321e-01 -1.58296898e-01 -9.06707823e-01
2.06980601e-01 -5.92440486e-01 -1.53697878e-02 9.69491482e-01
-1.89070612e-01 5.99299133e-01 -3.11525576e-02 1.24087751e-01
-2.53718764e-01 2.80295074e-01 1.08931315e+00 1.53672040e-01
-7.94669017e-02 1.90052688e-01 -7.01474965e-01 8.08326662e-01
4.03570741e-01 -5.44376194e-01 -1.23802684e-01 -4.04725939e-01
9.33970138e-03 1.09133855e-01 9.98362079e-02 -5.90069771e-01
2.26997197e-01 -2.61341423e-01 9.26949561e-01 -4.19453084e-01
5.93651295e-01 -9.49792922e-01 -1.35105729e-01 2.23809764e-01
-9.00352001e-02 1.07661344e-01 2.10769832e-01 -1.91612449e-02
-1.10357240e-01 8.59598517e-02 1.08109653e+00 1.29353479e-01
-2.75510639e-01 8.57329011e-01 4.38936770e-01 -1.72134876e-01
1.08128929e+00 -2.98170477e-01 4.28011090e-01 -2.69971550e-01
-7.72813141e-01 6.54104501e-02 4.55790490e-01 2.76109427e-01
7.84759164e-01 -1.47527051e+00 -9.65633273e-01 7.68949330e-01
-1.72605097e-01 1.80913024e-02 -2.10319653e-01 9.27775741e-01
-4.34712112e-01 1.36566550e-01 -2.53309429e-01 -7.50608742e-01
-1.68864393e+00 3.36308181e-01 4.96518314e-01 -2.95644969e-01
-4.65140373e-01 1.09498036e+00 -1.32354334e-01 -5.07011771e-01
8.37155804e-02 2.77774751e-01 -2.23838568e-01 9.47875977e-02
7.19833493e-01 1.54199213e-01 3.05037558e-01 -8.05470526e-01
-5.56011200e-01 1.30447316e+00 -1.04763642e-01 -1.53646231e-01
1.58278453e+00 1.83988407e-01 -2.38827184e-01 2.01843232e-02
1.44857240e+00 4.00991738e-01 -1.46990478e+00 -2.80706614e-01
-2.83247083e-02 -7.15897441e-01 8.37562457e-02 -4.96101707e-01
-1.32087255e+00 8.13074350e-01 3.42729300e-01 -4.28993821e-01
1.07498193e+00 -1.71551153e-01 3.77875328e-01 3.21857989e-01
2.78494030e-01 -8.99534643e-01 1.21978112e-01 3.72565061e-01
1.15591836e+00 -1.32862759e+00 4.24643099e-01 -5.19528270e-01
-5.64086854e-01 1.39950359e+00 5.71668923e-01 -5.71900368e-01
7.94442654e-01 6.37055635e-01 1.95531666e-01 -1.53114974e-01
-8.11719656e-01 -2.41956171e-02 6.71594977e-01 4.54822630e-01
5.81979573e-01 -3.63062710e-01 -6.16679341e-02 1.47756398e-01
-3.23612720e-01 -9.11013708e-02 -7.12739676e-02 6.26934052e-01
-3.40714693e-01 -1.42938375e+00 -3.51845413e-01 3.33385408e-01
-4.54839557e-01 5.31271379e-03 -2.75741071e-01 7.82861769e-01
-1.62819654e-01 5.82150459e-01 2.90190190e-01 -2.79201955e-01
2.62410074e-01 8.06412399e-02 9.68115389e-01 -5.02677321e-01
-6.34121656e-01 3.04374516e-01 -1.51095524e-01 -1.11234343e+00
-2.45069072e-01 -7.78300643e-01 -1.12818480e+00 -4.66226578e-01
-3.62312913e-01 2.03275271e-02 8.92897427e-01 9.54948485e-01
3.27444464e-01 -1.06479099e-03 1.35287166e+00 -1.12666261e+00
-6.23777926e-01 -7.89093435e-01 -1.92096025e-01 9.36302543e-02
4.49739993e-01 -6.96840763e-01 -9.71873179e-02 3.11825919e-04] | [13.430732727050781, 0.2757883071899414] |
b2b0e995-3609-43ee-b187-5845459ac784 | efficient-encoder-decoder-and-dual-path | 2306.05861 | null | https://arxiv.org/abs/2306.05861v1 | https://arxiv.org/pdf/2306.05861v1.pdf | Efficient Encoder-Decoder and Dual-Path Conformer for Comprehensive Feature Learning in Speech Enhancement | Current speech enhancement (SE) research has largely neglected channel attention and spatial attention, and encoder-decoder architecture-based networks have not adequately considered how to provide efficient inputs to the intermediate enhancement layer. To address these issues, this paper proposes a time-frequency (T-F) domain SE network (DPCFCS-Net) that incorporates improved densely connected blocks, dual-path modules, convolution-augmented transformers (conformers), channel attention, and spatial attention. Compared with previous models, our proposed model has a more efficient encoder-decoder and can learn comprehensive features. Experimental results on the VCTK+DEMAND dataset demonstrate that our method outperforms existing techniques in SE performance. Furthermore, the improved densely connected block and two dimensions attention module developed in this work are highly adaptable and easily integrated into existing networks. | ['Junyu Wang'] | 2023-06-09 | null | null | null | null | ['speech-enhancement'] | ['speech'] | [-4.77214456e-02 5.32299355e-02 -1.03791758e-01 -2.86607683e-01
-8.02510440e-01 1.65203884e-01 2.83059061e-01 -3.26932758e-01
-5.15717149e-01 6.40686333e-01 7.54142821e-01 -3.69689673e-01
1.84088841e-01 -6.22487426e-01 -6.04039609e-01 -3.97865117e-01
-5.03258780e-02 -3.85348588e-01 3.53677660e-01 -4.24934566e-01
-6.01373240e-02 1.03749692e-01 -1.07623136e+00 5.80281973e-01
1.06050682e+00 1.08085179e+00 6.55806541e-01 8.93467784e-01
3.78109850e-02 9.31304932e-01 -4.16719198e-01 -5.65108061e-01
-2.32550994e-01 -4.08560216e-01 -5.92203677e-01 -2.29151752e-02
-7.01805875e-02 -6.69667900e-01 -9.18621540e-01 8.92698050e-01
9.41923201e-01 1.61982790e-01 3.33240390e-01 -1.05964983e+00
-1.25928104e+00 7.74030387e-01 -4.77107078e-01 6.87659740e-01
-5.74508533e-02 -5.88720590e-02 9.56409991e-01 -1.25325453e+00
2.69256085e-01 1.13748693e+00 9.45098698e-01 5.72049379e-01
-7.48534203e-01 -5.65570951e-01 3.16062242e-01 7.82897770e-01
-1.17636347e+00 -6.91764355e-01 7.82743514e-01 2.83085525e-01
1.73188519e+00 -1.68857351e-02 6.71415389e-01 1.14387774e+00
1.48046732e-01 1.23776114e+00 8.60723734e-01 -3.57302397e-01
-1.37653589e-01 2.12042198e-01 -9.18819755e-02 4.47580844e-01
-3.97691369e-01 5.32966256e-01 -6.21522605e-01 4.21062142e-01
8.67654741e-01 -1.23814672e-01 -4.39510852e-01 2.03565463e-01
-9.83550906e-01 6.12650394e-01 8.35770607e-01 3.68363649e-01
-5.15719354e-01 2.34940395e-01 6.15840614e-01 4.54428613e-01
6.47324264e-01 4.96425554e-02 -7.28722155e-01 -4.07351226e-01
-1.11882472e+00 -2.05437049e-01 3.81941944e-01 1.23885024e+00
4.49151099e-01 6.47703111e-01 -5.06088197e-01 1.06139719e+00
2.56839573e-01 3.55344176e-01 5.18261075e-01 -7.09684372e-01
6.62703812e-01 4.33907062e-02 -4.83273864e-01 -6.79090619e-01
-1.22362651e-01 -9.05647159e-01 -1.14770639e+00 -2.21597850e-01
-4.25156623e-01 -3.06313396e-01 -9.02911842e-01 1.63730359e+00
-1.38044819e-01 4.57905829e-01 1.79368913e-01 9.77482855e-01
1.05804503e+00 8.23174715e-01 3.72218072e-01 -2.24207006e-02
1.18732309e+00 -1.59329927e+00 -1.25209486e+00 -1.12474002e-01
3.75987351e-01 -7.32975841e-01 8.25932562e-01 1.79407626e-01
-1.43768549e+00 -8.77574384e-01 -1.16018033e+00 -4.09578919e-01
-3.82852703e-01 3.60378116e-01 6.56311989e-01 7.03617573e-01
-1.61181271e+00 6.23304069e-01 -5.23477435e-01 -4.19181325e-02
6.29964709e-01 4.56505239e-01 -2.32831880e-01 8.97994637e-02
-1.67857766e+00 8.49715829e-01 3.22451949e-01 1.41102925e-01
-1.03525627e+00 -6.77753985e-01 -9.84425783e-01 5.87821245e-01
1.63465664e-01 -6.53137028e-01 1.60825884e+00 -1.05091965e+00
-1.74385273e+00 -2.82202754e-02 -2.94204354e-01 -7.00392544e-01
-1.00293472e-01 -1.46932438e-01 -8.72365534e-01 2.95685649e-01
-3.55060667e-01 1.01400399e+00 9.98376250e-01 -8.39893043e-01
-7.88474560e-01 -3.35973129e-02 -2.71989200e-02 3.93648475e-01
-8.29831660e-01 2.17882767e-01 -5.29558361e-01 -1.11971760e+00
-7.25202635e-02 -2.72546947e-01 -1.66583374e-01 -3.55948061e-02
-1.99951053e-01 1.12960093e-01 1.21097684e+00 -1.28093195e+00
1.77162659e+00 -2.28151226e+00 4.82889749e-02 -1.92701951e-01
2.76825309e-01 6.58772230e-01 -4.26097691e-01 2.93327153e-01
-3.20175320e-01 1.77926809e-01 -1.43485114e-01 -5.72691679e-01
1.14267722e-01 6.98138773e-02 -6.81571066e-02 2.09507644e-01
4.47590232e-01 1.13418090e+00 -6.96403980e-01 -3.96622181e-01
3.18653792e-01 1.15825415e+00 -8.15504670e-01 1.94432750e-01
1.91672355e-01 2.33329028e-01 -2.35915512e-01 5.78201175e-01
7.32872725e-01 -8.90611559e-02 -1.76659495e-01 -3.38137656e-01
-2.25648463e-01 6.96297228e-01 -8.30933571e-01 1.59561920e+00
-7.58719265e-01 9.89536405e-01 1.57334253e-01 -8.38237286e-01
7.13293076e-01 8.20392489e-01 2.69531280e-01 -1.27122080e+00
1.62289441e-01 1.13603778e-01 4.05569263e-02 -4.45867211e-01
7.81840444e-01 1.46583728e-02 4.39753652e-01 1.99527606e-01
4.81284827e-01 4.09300089e-01 -1.96255386e-01 2.75496751e-01
1.00705528e+00 -2.64061000e-02 1.32866621e-01 4.11119349e-02
6.62211120e-01 -5.44808149e-01 5.63956678e-01 3.07029456e-01
-4.63377595e-01 6.43624365e-01 1.37116596e-01 6.99535757e-02
-1.44019294e+00 -9.89320159e-01 -1.47931203e-01 1.29454160e+00
4.54707704e-02 -6.31785870e-01 -8.61185849e-01 -5.43855727e-01
-4.79905427e-01 3.84399563e-01 -4.29181904e-01 -2.56576210e-01
-5.53328931e-01 -3.63698125e-01 6.95066392e-01 1.24741220e+00
1.10315466e+00 -1.15492010e+00 -1.59926176e-01 6.89661920e-01
-5.76707125e-01 -1.15609050e+00 -9.11004424e-01 2.09002241e-01
-8.70482087e-01 -4.97026443e-01 -1.11156523e+00 -1.06553984e+00
4.21753824e-01 4.82114106e-01 9.31495845e-01 -3.09146140e-02
1.41462320e-02 1.76012307e-01 -7.25271881e-01 -3.22893083e-01
-9.24668014e-02 2.29389325e-01 -4.67405647e-01 -9.54814851e-02
3.08655769e-01 -5.90462863e-01 -8.11763287e-01 1.47732645e-01
-6.12284780e-01 1.07560277e-01 8.72274995e-01 1.24408555e+00
4.49381053e-01 7.62090310e-02 9.17094707e-01 -4.64624584e-01
7.81408250e-01 -4.71017033e-01 -3.40369679e-02 1.00174464e-01
-6.83303833e-01 4.74720116e-04 5.23941636e-01 -2.94557422e-01
-1.55479038e+00 -3.26147228e-01 -8.50588918e-01 -3.91710043e-01
-8.37962981e-03 4.37937021e-01 -4.18194741e-01 -2.07833365e-01
2.52467513e-01 5.50006747e-01 -1.64906949e-01 -6.39539599e-01
3.38861287e-01 1.08658719e+00 5.47412515e-01 -1.95156142e-01
4.03389215e-01 1.29132763e-01 -6.42107546e-01 -8.24160874e-01
-2.05721557e-01 -4.36516643e-01 -4.83386904e-01 -1.52674094e-01
7.98963964e-01 -1.41524947e+00 -3.66359681e-01 4.33815747e-01
-1.23529446e+00 -2.39738837e-01 -2.86352426e-01 6.62392914e-01
-4.63432401e-01 4.92123306e-01 -1.08716464e+00 -7.23493576e-01
-6.98312044e-01 -1.22782838e+00 9.85706508e-01 1.55433208e-01
2.50635862e-01 -1.02863836e+00 -3.21259171e-01 2.13017225e-01
1.08433461e+00 -5.56257665e-01 5.63149273e-01 -1.09618209e-01
-5.62055647e-01 2.99396038e-01 -5.09121418e-01 6.47393584e-01
-2.00458393e-01 -2.90585369e-01 -1.30344152e+00 -3.47417891e-01
-1.06618673e-01 -6.95804879e-02 1.22559118e+00 6.93700016e-01
1.48743761e+00 -3.44193101e-01 -2.33311772e-01 9.85946476e-01
1.17849970e+00 4.62134212e-01 1.12302160e+00 3.29934478e-01
3.55843425e-01 9.81709585e-02 3.18027318e-01 5.35600364e-01
7.06734419e-01 5.09686172e-01 3.43343437e-01 -6.96564853e-01
-6.53815687e-01 -2.95235693e-01 6.39094591e-01 1.40896034e+00
-1.17961302e-01 -5.32479465e-01 -2.82697678e-01 8.41008902e-01
-1.54136312e+00 -1.13680995e+00 3.21579799e-02 1.39313316e+00
6.89354241e-01 -1.26223534e-01 -6.24697320e-02 4.89368528e-01
7.93961883e-01 2.32324436e-01 -2.93415129e-01 -6.96279287e-01
-3.51703674e-01 5.84722757e-01 3.16335648e-01 3.23704422e-01
-1.23136365e+00 8.54763150e-01 6.52217054e+00 1.14796257e+00
-1.02792037e+00 6.30988598e-01 7.65608549e-01 1.09239116e-01
-3.25140893e-01 -3.09958607e-01 -6.07814491e-01 3.27116489e-01
1.13106775e+00 1.31551186e-02 3.84970278e-01 7.96592891e-01
3.01249176e-01 5.37424147e-01 -5.75782120e-01 8.05234969e-01
-1.55235231e-01 -1.21660304e+00 -3.32389325e-01 -2.51379684e-02
6.45170033e-01 1.28373593e-01 4.87496078e-01 7.58320570e-01
-2.81216390e-03 -1.02277827e+00 7.35598564e-01 2.03601837e-01
1.22763455e+00 -8.49960864e-01 7.96563983e-01 6.15696050e-02
-1.77931476e+00 -2.47074842e-01 -4.08638805e-01 8.59442130e-02
3.28672141e-01 2.89949507e-01 -7.02132463e-01 5.66201091e-01
8.92619073e-01 9.60293174e-01 -2.18544424e-01 1.17125964e+00
-3.59718204e-01 7.38849342e-01 1.67029956e-03 9.56457108e-02
4.67307955e-01 3.87238830e-01 3.72191131e-01 1.60381186e+00
5.55444300e-01 2.14800864e-01 -3.96130443e-01 5.86969912e-01
-2.11549670e-01 1.61208082e-02 -2.48718768e-01 4.47716378e-02
3.63555133e-01 1.07531977e+00 -1.24601752e-01 -2.84232080e-01
-9.01173770e-01 1.12656629e+00 3.20281148e-01 6.44532084e-01
-1.04605937e+00 -5.88587701e-01 7.20204651e-01 -8.17587301e-02
8.78361404e-01 -8.64603743e-02 -2.16262996e-01 -1.07699156e+00
-1.34823307e-01 -8.83516192e-01 1.16808973e-01 -9.89487708e-01
-9.94806409e-01 1.07927346e+00 -5.70460975e-01 -1.09995818e+00
1.58324212e-01 -5.52285135e-01 -6.22216463e-01 1.07372904e+00
-2.31880069e+00 -1.38046026e+00 -1.40787289e-01 1.05648768e+00
8.61705303e-01 -3.20264071e-01 6.07140958e-01 9.74710047e-01
-6.01818383e-01 1.05739546e+00 3.53644751e-02 2.51228720e-01
5.12729764e-01 -1.12645578e+00 5.04427493e-01 1.00194061e+00
-6.32097125e-02 3.46140653e-01 9.52090472e-02 -5.08178174e-01
-9.99054253e-01 -1.36823857e+00 1.09188378e+00 3.86791050e-01
3.47214729e-01 -1.76205277e-01 -1.02232897e+00 5.56696832e-01
8.19442749e-01 3.66838127e-02 5.27743280e-01 -8.08774233e-02
-3.37637484e-01 -1.86916590e-01 -9.46645796e-01 4.33509886e-01
1.14554060e+00 -7.88478851e-01 -3.58146518e-01 -3.66037488e-02
9.05131817e-01 -5.18745542e-01 -9.73081470e-01 3.31840247e-01
2.90155143e-01 -8.63888144e-01 1.15281212e+00 -4.22231674e-01
5.33848822e-01 1.69335923e-03 -1.33858398e-01 -1.58039153e+00
-6.70462847e-01 -5.99091709e-01 -4.29708809e-01 1.18292797e+00
6.02087975e-01 -3.79823744e-01 4.66134220e-01 -5.30193783e-02
-1.01918793e+00 -9.67764020e-01 -9.02183354e-01 -6.16547585e-01
-7.79002532e-02 -6.70626521e-01 1.00033319e+00 6.11399591e-01
1.17036730e-01 2.18816683e-01 -6.56327903e-01 2.98697919e-01
2.70985991e-01 -4.71409023e-01 6.64163455e-02 -6.17354631e-01
-4.42059487e-01 -5.67619205e-01 -1.57974273e-01 -1.48482108e+00
5.74312657e-02 -7.48350024e-01 1.07062690e-01 -1.53406549e+00
2.89303511e-01 -3.50969315e-01 -5.30575335e-01 4.97648180e-01
-3.34855616e-01 1.23759046e-01 1.50769249e-01 -2.45042354e-01
-6.01038158e-01 1.11975121e+00 1.52988398e+00 -2.18949944e-01
8.50204006e-02 -2.62390941e-01 -7.83490062e-01 2.81368941e-01
8.02761555e-01 -1.62936047e-01 -4.74017918e-01 -8.10017824e-01
-1.89432964e-01 1.71352997e-01 3.61141294e-01 -1.03752315e+00
6.92436397e-01 4.34913874e-01 4.95021999e-01 -6.69373095e-01
4.97558683e-01 -7.98612475e-01 -1.98012814e-01 2.46468678e-01
-3.62023711e-01 2.20171854e-01 4.07195628e-01 4.96338904e-01
-7.24049926e-01 1.23654388e-01 8.58143151e-01 6.25045449e-02
-1.17858016e+00 7.15396166e-01 -4.71324831e-01 -1.84942067e-01
7.49182820e-01 -2.84945250e-01 -4.08280432e-01 -5.80256879e-01
-8.70711267e-01 1.40586510e-01 -3.31709802e-01 3.35832447e-01
1.21364439e+00 -1.47633028e+00 -8.20848346e-01 4.17924613e-01
-8.12267959e-02 -5.80217957e-01 7.92226732e-01 8.92440438e-01
-3.10820967e-01 7.00255156e-01 -2.69650847e-01 -1.82659373e-01
-1.08151388e+00 4.73103642e-01 4.03618604e-01 -3.09089899e-01
-7.18582451e-01 1.07821333e+00 1.95201412e-01 -1.95445389e-01
4.63300020e-01 -3.02617669e-01 -3.11716139e-01 -3.46642345e-01
6.49725258e-01 4.82750088e-01 9.29273218e-02 -6.26063704e-01
-8.48345906e-02 2.17937469e-01 -1.07086740e-01 -1.83393508e-01
1.53913438e+00 -4.93487746e-01 3.42264354e-01 -3.40946376e-01
1.30623996e+00 -5.15820205e-01 -1.47857368e+00 -5.25140464e-01
-4.53723341e-01 -2.33629405e-01 7.34066725e-01 -9.65569913e-01
-1.63491714e+00 1.44579983e+00 7.42995501e-01 -8.63432512e-02
1.84167981e+00 -2.47895539e-01 1.30508626e+00 -9.02672485e-02
-6.93738833e-02 -1.29475212e+00 2.78155714e-01 8.21131945e-01
9.80293274e-01 -1.16329598e+00 -4.82222766e-01 -5.01794696e-01
-8.08974087e-01 1.07817805e+00 8.60787928e-01 2.98429340e-01
8.13732803e-01 7.37184942e-01 -1.96220577e-01 1.69043094e-01
-1.04515541e+00 -4.49463785e-01 2.95608759e-01 9.77708399e-01
6.74822390e-01 -1.27179980e-01 -1.93519503e-01 9.03578401e-01
3.99371013e-02 -5.06381281e-02 9.12174135e-02 6.91164494e-01
-4.66282725e-01 -1.03227532e+00 -6.89958483e-02 2.40520433e-01
-5.98692238e-01 -7.49118090e-01 1.90836772e-01 6.34946585e-01
2.41170377e-01 9.89140689e-01 1.05773531e-01 -7.35144436e-01
3.72286528e-01 -2.69640595e-01 1.25775397e-01 -2.51694441e-01
-9.90517139e-01 4.68349040e-01 1.17666036e-01 -4.42888677e-01
-3.44924659e-01 -3.15295428e-01 -1.13321447e+00 -4.17754143e-01
-6.08219087e-01 1.29985735e-01 6.94429696e-01 6.93731010e-01
7.76844144e-01 1.27785242e+00 7.30401039e-01 -6.49537921e-01
-2.23429471e-01 -1.14883614e+00 -7.08865821e-01 -6.68430608e-03
5.57834923e-01 -4.59670752e-01 -8.80350247e-02 2.67800707e-02] | [14.740074157714844, 5.923710823059082] |
31f8e96e-7208-4adf-b0e0-76e68f91e995 | fingerprint-recognition-under-missing-image | 1902.05389 | null | http://arxiv.org/abs/1902.05389v1 | http://arxiv.org/pdf/1902.05389v1.pdf | Fingerprint Recognition under Missing Image Pixels Scenario | This work observed the problem of fingerprint image recognition in the case
of missing pixels from the original image. The possibility of missing pixels
recovery is tested by applying the Compressive Sensing approach. Namely,
different percentage of missing pixels is observed and the image reconstruction
is done by applying commonly used approach for sparse image reconstruction. The
theory is verified by experiments, showing successful image reconstruction and
later person identification even if less then 90% of the image pixels is
missing. | ['Kristina Tomovic', 'Dejan Brajovic', 'Jovan Radonjic'] | 2019-02-06 | null | null | null | null | ['person-identification'] | ['computer-vision'] | [ 1.45655692e+00 5.15870936e-02 -1.91233993e-01 4.57186066e-02
-2.00169325e-01 -2.62855381e-01 2.34970510e-01 -5.94036460e-01
-1.70230180e-01 1.02198136e+00 1.49033636e-01 -1.10676207e-01
-2.64808893e-01 -8.39101851e-01 -8.89596581e-01 -7.64704704e-01
1.77607372e-01 3.11760902e-02 -3.93658578e-01 8.45667496e-02
5.30263305e-01 5.42070031e-01 -1.70906031e+00 4.07626510e-01
6.10354006e-01 7.80612350e-01 3.00497055e-01 4.15459752e-01
-5.58475591e-02 8.72304440e-01 -4.84282345e-01 -1.66961908e-01
6.14962339e-01 -5.96850574e-01 -3.13348383e-01 4.74344164e-01
4.08801198e-01 -5.04836679e-01 -4.83717442e-01 1.44900107e+00
1.46617278e-01 -2.11534917e-01 4.09161538e-01 -8.60175252e-01
-6.55051827e-01 5.00043988e-01 -6.22490048e-01 2.43939217e-02
7.11712301e-01 -4.00842458e-01 1.44646376e-01 -8.88672113e-01
6.77589178e-01 9.27968860e-01 5.40959358e-01 5.35819903e-02
-1.19177806e+00 -6.45775914e-01 -7.60731697e-01 2.78996646e-01
-1.58799875e+00 -5.06381929e-01 1.10185635e+00 -7.21959621e-02
2.35460132e-01 4.17814761e-01 1.44070581e-01 6.83445334e-01
1.19253159e-01 3.72899890e-01 1.56669164e+00 -7.56069422e-01
-1.15957089e-01 1.83579490e-01 6.53686002e-02 5.46780109e-01
7.98433781e-01 4.74513024e-01 -5.07838190e-01 -1.95832163e-01
7.38898277e-01 5.59148848e-01 -1.78953633e-01 -2.59314962e-02
-1.22072411e+00 4.95117396e-01 8.77836198e-02 6.55267656e-01
-7.27747321e-01 -3.33895922e-01 -2.60809630e-01 4.61257935e-01
-3.34868759e-01 -9.68363881e-02 6.06485665e-01 4.15409863e-01
-1.56848156e+00 -1.85147524e-02 6.17009223e-01 7.49893010e-01
6.54043496e-01 3.75656962e-01 -2.50100233e-02 5.85826278e-01
3.41584533e-01 1.02648818e+00 3.39832336e-01 -1.25910509e+00
4.33489472e-01 3.77990395e-01 4.37359363e-01 -1.28278160e+00
3.25215191e-01 -7.11185098e-01 -1.07224667e+00 2.90565491e-01
2.01179713e-01 -7.01754913e-02 -8.44648600e-01 1.09100902e+00
-1.43551469e-01 8.70176673e-01 3.52542490e-01 7.16310620e-01
3.84910315e-01 6.70114279e-01 -5.12161493e-01 -4.87096012e-01
1.01102817e+00 -2.60760844e-01 -9.38119531e-01 -1.11164600e-01
-5.05510449e-01 -1.02745116e+00 1.02130845e-01 8.05489242e-01
-7.80947864e-01 -8.41088414e-01 -1.35356092e+00 5.29942334e-01
3.42014700e-01 2.78829485e-01 2.89893627e-01 1.07509613e+00
-4.84460503e-01 4.63194519e-01 -1.19505353e-01 -3.32824811e-02
1.98912024e-01 3.69539380e-01 -5.77959716e-01 -9.13782597e-01
-8.58981848e-01 4.12088990e-01 2.78016061e-01 5.03426611e-01
-4.47685510e-01 -1.29693970e-01 -6.26484871e-01 -4.45646569e-02
-2.65728682e-01 -3.38088870e-01 3.43202323e-01 -1.20377183e+00
-9.46757853e-01 7.35846043e-01 -4.43877339e-01 -5.71965933e-01
6.26010001e-01 1.03843115e-01 -8.22070122e-01 4.54009682e-01
5.35785668e-02 -2.42415324e-01 1.38606048e+00 -1.25840390e+00
-6.65282160e-02 -6.28545761e-01 -7.41197348e-01 -1.04271844e-01
2.52554625e-01 -3.40320766e-01 1.15114234e-01 -4.80983436e-01
7.63017893e-01 -9.19950366e-01 -1.17063574e-01 -3.08567077e-01
-5.15958071e-01 7.48164177e-01 1.03779924e+00 -1.00526750e+00
6.72074914e-01 -2.43129826e+00 -3.44160289e-01 9.27414596e-01
-2.84969389e-01 1.91781834e-01 -9.02416110e-02 7.71938264e-01
-3.42420377e-02 -1.91817790e-01 -5.07103503e-01 -2.19979812e-03
-4.19293880e-01 1.40367508e-01 -4.31049168e-01 7.15592802e-01
-3.23093563e-01 1.89993039e-01 -3.57300282e-01 -3.23894233e-01
4.13067132e-01 7.60990739e-01 -9.43896100e-02 -1.90505683e-01
5.69607854e-01 7.31533647e-01 -3.29172432e-01 9.72831488e-01
1.34003651e+00 -6.67975768e-02 4.11165327e-01 -2.26653799e-01
-1.26294300e-01 -4.38995838e-01 -1.84216475e+00 1.10855532e+00
1.54344499e-01 5.37204742e-01 2.39278734e-01 -1.33447540e+00
1.27799225e+00 3.79844993e-01 8.06725562e-01 -9.48655188e-01
-5.29802740e-02 3.41555417e-01 -9.83807594e-02 -7.12813139e-01
4.59730476e-01 -1.56709358e-01 3.97050053e-01 4.81427163e-01
-3.69089425e-01 4.36318636e-01 -1.62951350e-01 -4.91232350e-02
9.63366628e-01 -1.52156487e-01 2.20920309e-01 -1.27572134e-01
4.26442087e-01 -1.75902341e-02 2.63751686e-01 1.04019690e+00
3.19487080e-02 7.55562484e-01 -2.79678315e-01 -2.02639941e-02
-1.26863575e+00 -9.48290825e-01 -5.45876563e-01 1.01523876e-01
6.12320006e-01 5.69930196e-01 -4.13364738e-01 1.07505478e-01
2.66995966e-01 2.96584368e-01 -5.16565681e-01 1.93297476e-01
-4.68807012e-01 -6.52856529e-01 5.24644136e-01 -1.25935078e-01
8.59530807e-01 -1.12319076e+00 -4.92449105e-01 2.16398194e-01
-2.66240448e-01 -1.10956538e+00 1.73420906e-01 -4.60606337e-01
-1.15536010e+00 -1.18094325e+00 -8.70515108e-01 -8.08334529e-01
1.10938740e+00 7.70798266e-01 4.02978688e-01 4.84476298e-01
-4.32620704e-01 2.53740728e-01 -2.42312923e-01 -5.99314161e-02
-2.01853409e-01 -7.92542040e-01 -2.23941669e-01 6.23489380e-01
2.41481185e-01 -4.76080060e-01 -8.04243326e-01 1.12144589e-01
-9.04741406e-01 -3.23123366e-01 8.36558700e-01 8.29109013e-01
5.98609567e-01 5.15792608e-01 2.88334012e-01 -1.11393166e+00
4.71542865e-01 -5.50711572e-01 -2.98553437e-01 1.32180586e-01
-6.02542341e-01 1.38305783e-01 2.21010506e-01 -1.93355396e-01
-8.60828817e-01 2.71889150e-01 6.28819168e-02 -2.57085830e-01
-8.67447853e-02 4.46744204e-01 5.58192059e-02 -5.83264649e-01
4.18913096e-01 9.68256950e-01 4.04900104e-01 -3.57036322e-01
-2.25123972e-01 7.94715285e-01 8.55505526e-01 -2.54545748e-01
7.32051790e-01 7.29384184e-01 2.29491681e-01 -1.09535265e+00
-4.37573716e-02 -4.05006021e-01 -1.87981069e-01 -2.81701207e-01
2.23787755e-01 -7.37757444e-01 -5.43340027e-01 6.02511287e-01
-8.88773024e-01 5.80892861e-01 3.56022716e-02 7.40805745e-01
-1.28404602e-01 8.30813587e-01 -1.47093967e-01 -1.11642969e+00
-1.09750971e-01 -9.51338172e-01 5.19692004e-01 1.24808013e-01
3.69285256e-01 -3.31767529e-01 -1.52563915e-01 8.05259168e-01
4.17484879e-01 4.28633034e-01 4.88135278e-01 -9.77362320e-02
-7.12275386e-01 -7.73976147e-01 -2.16251835e-01 2.23275885e-01
1.80079058e-01 -3.11134815e-01 -8.65119576e-01 -3.98634076e-01
3.69127065e-01 2.98679411e-01 8.34737659e-01 4.70965147e-01
8.09970021e-01 -3.59647721e-01 -2.19042614e-01 4.13931906e-01
2.01356006e+00 3.89735579e-01 1.45239151e+00 6.36838153e-02
8.42531398e-02 3.69498432e-01 4.13204610e-01 4.96436298e-01
-3.32784176e-01 3.76095146e-01 4.06727731e-01 9.91745517e-02
-8.86644050e-02 -2.89796561e-01 7.59501532e-02 2.84991920e-01
-2.58003354e-01 -2.90859908e-01 -3.52376193e-01 4.04111981e-01
-1.44182599e+00 -1.70402050e+00 -4.93360728e-01 2.41450930e+00
2.13603154e-01 -2.24339887e-01 -2.45666444e-01 8.14384103e-01
1.15095675e+00 1.91860706e-01 -3.49651814e-01 -9.78093818e-02
-3.79423976e-01 3.29227984e-01 8.66953790e-01 6.68869376e-01
-5.84267497e-01 1.39533579e-01 7.21443844e+00 2.64029920e-01
-1.23011911e+00 -1.22921243e-01 1.71287522e-01 3.38129550e-01
-5.27721345e-01 1.59709156e-01 -4.49831486e-01 9.01128411e-01
3.12896430e-01 3.33912015e-01 5.32264769e-01 2.63178319e-01
-2.66334265e-02 -4.39513326e-01 -2.93349862e-01 1.33789277e+00
3.93400162e-01 -1.24573076e+00 -5.56467380e-03 3.53839844e-01
8.28911066e-01 -5.77440798e-01 1.54734552e-01 -3.76759589e-01
-1.00546181e-01 -1.01808977e+00 4.14092749e-01 1.00548148e+00
7.38784790e-01 -4.99537408e-01 8.09616446e-01 5.52139044e-01
-7.53991365e-01 -2.29004458e-01 -5.31930685e-01 -1.60866797e-01
6.82096258e-02 9.13654983e-01 -7.41418600e-01 5.54307401e-01
1.92874044e-01 4.85294253e-01 -1.50418440e-02 1.08952987e+00
1.94751903e-01 7.88845599e-01 -4.65859294e-01 3.29291224e-01
-1.21636920e-01 -5.54695666e-01 7.05594718e-01 7.48473346e-01
1.02581275e+00 1.80693775e-01 -1.21041976e-01 9.22192812e-01
-8.82076025e-02 1.13712959e-02 -1.13064051e+00 1.30661130e-01
1.00846338e+00 4.13122207e-01 -2.91316539e-01 -1.80089399e-01
-2.80106395e-01 1.25055623e+00 -6.71142161e-01 5.77838123e-01
-2.81911433e-01 -1.88252345e-01 -3.04429621e-01 3.89064610e-01
5.98559260e-01 -2.36782283e-01 -6.24593616e-01 -8.76232207e-01
1.92610338e-01 -7.32408464e-01 1.54855149e-02 -6.11158490e-01
-8.14455152e-01 1.83208227e-01 -3.45395327e-01 -1.38708484e+00
-4.32370827e-02 -1.18787833e-01 -6.35765016e-01 9.13493037e-01
-1.40560043e+00 -9.16208923e-01 -2.79131621e-01 7.68321216e-01
1.85738578e-01 -6.57300889e-01 6.07857406e-01 4.25947219e-01
-2.26261452e-01 4.43709344e-01 5.65986991e-01 -1.27078801e-01
2.19040379e-01 -4.72663403e-01 -5.28321981e-01 1.25382674e+00
1.49532095e-01 9.14856613e-01 9.74109232e-01 -9.09150720e-01
-1.58931744e+00 -6.11354411e-01 1.00861037e+00 4.61412311e-01
-1.38735309e-01 3.31010133e-01 -5.60839772e-01 4.34605479e-01
2.73685038e-01 8.88858214e-02 7.99051046e-01 -4.48467672e-01
-2.54913509e-01 -1.70747668e-01 -1.89663017e+00 -9.30301473e-02
5.59687912e-01 -5.95722198e-01 -4.78085756e-01 4.60714810e-02
-3.62767190e-01 1.11002766e-01 -8.07608724e-01 4.82578099e-01
8.81391287e-01 -9.75142896e-01 1.35431743e+00 1.69083193e-01
2.00481206e-01 -4.96739924e-01 -8.11999500e-01 -3.18299919e-01
-2.08952770e-01 -1.69616491e-01 3.40220928e-02 9.61982965e-01
1.98382586e-01 -3.22455376e-01 1.04590654e+00 7.77783170e-02
5.40535033e-01 2.04503015e-01 -8.86896253e-01 -5.91945112e-01
-8.19076598e-01 -7.34920576e-02 3.22664291e-01 1.10808933e+00
-4.19135213e-01 -3.23574096e-01 -1.15486634e+00 3.64178210e-01
1.41666806e+00 4.38314348e-01 4.04873163e-01 -1.13566792e+00
-4.02529776e-01 3.40427220e-01 -5.46712399e-01 -6.02834046e-01
-2.19922423e-01 -7.26328731e-01 -3.56089622e-01 -1.26075423e+00
3.29767197e-01 -2.18325630e-01 -5.21979332e-01 3.64536680e-02
1.83636397e-01 8.71393561e-01 2.36161292e-01 7.19366550e-01
7.81426355e-02 -1.71577096e-01 1.27541482e+00 -3.67919147e-01
2.65780658e-01 3.40616435e-01 -7.00612068e-01 -3.65760848e-02
5.20986676e-01 -5.50498962e-01 -3.48648995e-01 -7.96748176e-02
3.94529523e-03 6.11517310e-01 7.63098657e-01 -1.45842588e+00
1.63487375e-01 -1.32276520e-01 6.11015081e-01 -6.64864302e-01
3.25876057e-01 -1.25740433e+00 1.02138758e+00 7.90376723e-01
-1.07889622e-01 -3.60009640e-01 -4.02065694e-01 6.21070921e-01
-4.67966378e-01 -6.46965623e-01 6.87129200e-01 -2.86321223e-01
-9.43983018e-01 -3.71981263e-01 -2.15782404e-01 -9.27634180e-01
9.22323585e-01 -1.09312987e+00 -1.12105198e-01 -2.76284754e-01
-9.23394620e-01 -6.65323615e-01 4.07835394e-01 -1.55027434e-01
1.01177835e+00 -1.29958463e+00 -9.82675016e-01 7.81116724e-01
-3.31186771e-01 -8.33656907e-01 6.74972951e-01 7.43855774e-01
-7.30707467e-01 2.37949163e-01 -5.98058760e-01 -4.59586203e-01
-1.31611145e+00 5.00626981e-01 1.92480072e-01 1.67331234e-01
-7.63122499e-01 1.76294193e-01 -7.38494694e-01 1.62132978e-01
-8.01659599e-02 4.65508014e-01 -1.66186601e-01 -3.84668976e-01
6.65533721e-01 5.57657599e-01 -1.36538371e-01 -7.28777766e-01
-1.83123946e-01 8.05781066e-01 3.36810291e-01 -1.17227808e-01
1.18438590e+00 -2.60150343e-01 -2.55797416e-01 4.39618640e-02
9.63115752e-01 5.94236732e-01 -8.77164364e-01 -4.61333126e-01
-2.43196353e-01 -8.83308232e-01 2.94672176e-02 -6.72985375e-01
-8.39193821e-01 4.55451638e-01 1.19318044e+00 1.14708580e-01
1.01076138e+00 -5.59149683e-01 6.38033032e-01 3.21802497e-01
6.64075494e-01 -7.91626930e-01 -3.50596368e-01 -5.40604964e-02
7.54501343e-01 -1.22017503e+00 4.55021501e-01 -3.49751323e-01
-3.23840529e-01 9.82956290e-01 -4.37443614e-01 -6.32813156e-01
6.91665888e-01 3.07219177e-01 -2.13694677e-01 -1.13438204e-01
9.54099968e-02 2.44309440e-01 -8.08322728e-02 6.15466237e-01
9.20472667e-02 2.59753793e-01 -6.84686899e-01 1.22924253e-01
-5.77515289e-02 4.96545911e-01 5.15026510e-01 1.14574325e+00
-6.58074439e-01 -1.13738465e+00 -1.13663304e+00 3.70110482e-01
-6.77611470e-01 2.76902169e-02 -1.98771462e-01 3.84970844e-01
3.10188591e-01 1.13917136e+00 -2.55027086e-01 -3.08043540e-01
-8.41501355e-02 -1.00335166e-01 7.60072947e-01 -1.40512019e-01
-3.21652107e-02 -1.16076335e-01 -2.00941533e-01 -2.43892863e-01
-7.46650875e-01 -9.47574556e-01 -1.05879939e+00 -2.59566128e-01
-2.96273176e-02 2.73330659e-01 9.70921755e-01 8.38094771e-01
9.62025970e-02 9.50106010e-02 9.71407831e-01 -3.51695329e-01
-2.24737465e-01 -8.12897563e-01 -1.03755069e+00 4.85308409e-01
6.31236136e-01 -3.61560851e-01 -2.76585728e-01 2.90029615e-01] | [12.430561065673828, 0.31963491439819336] |
9812ce72-b56a-4f40-90b2-95730ca26480 | depgraph-towards-any-structural-pruning | 2301.12900 | null | https://arxiv.org/abs/2301.12900v2 | https://arxiv.org/pdf/2301.12900v2.pdf | DepGraph: Towards Any Structural Pruning | Structural pruning enables model acceleration by removing structurally-grouped parameters from neural networks. However, the parameter-grouping patterns vary widely across different models, making architecture-specific pruners, which rely on manually-designed grouping schemes, non-generalizable to new architectures. In this work, we study a highly-challenging yet barely-explored task, any structural pruning, to tackle general structural pruning of arbitrary architecture like CNNs, RNNs, GNNs and Transformers. The most prominent obstacle towards this goal lies in the structural coupling, which not only forces different layers to be pruned simultaneously, but also expects all removed parameters to be consistently unimportant, thereby avoiding structural issues and significant performance degradation after pruning. To address this problem, we propose a general and {fully automatic} method, \emph{Dependency Graph} (DepGraph), to explicitly model the dependency between layers and comprehensively group coupled parameters for pruning. In this work, we extensively evaluate our method on several architectures and tasks, including ResNe(X)t, DenseNet, MobileNet and Vision transformer for images, GAT for graph, DGCNN for 3D point cloud, alongside LSTM for language, and demonstrate that, even with a simple norm-based criterion, the proposed method consistently yields gratifying performances. | ['Xinchao Wang', 'Michael Bi Mi', 'Mingli Song', 'Xinyin Ma', 'Gongfan Fang'] | 2023-01-30 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Fang_DepGraph_Towards_Any_Structural_Pruning_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Fang_DepGraph_Towards_Any_Structural_Pruning_CVPR_2023_paper.pdf | cvpr-2023-1 | ['neural-network-compression', 'neural-network-compression'] | ['methodology', 'miscellaneous'] | [ 6.60695955e-02 1.78069234e-01 2.91750759e-01 -2.00339973e-01
5.04371151e-03 -5.33653259e-01 4.36834604e-01 3.82122979e-03
-6.13271415e-01 3.74469995e-01 -1.47713006e-01 -6.32353902e-01
-2.68467844e-01 -5.66368759e-01 -7.19370723e-01 -6.00947678e-01
-8.35659448e-03 3.68884325e-01 3.01319808e-01 -3.35876763e-01
7.71581084e-02 8.54485333e-01 -1.35832047e+00 3.76468115e-02
8.53065491e-01 1.00055623e+00 4.67101693e-01 2.48701781e-01
-1.63751170e-01 7.86136985e-01 -6.23858809e-01 -7.24212885e-01
4.55087990e-01 1.94972120e-02 -7.37902999e-01 -2.72148371e-01
6.46450400e-01 -8.13016742e-02 -2.13990808e-01 1.24600720e+00
2.83121675e-01 1.12148218e-01 4.83451188e-01 -8.35961521e-01
-5.06627083e-01 9.26730096e-01 -5.42478621e-01 2.06097484e-01
-4.65038508e-01 3.32600236e-01 1.19616997e+00 -6.69125974e-01
3.35209996e-01 1.32843673e+00 9.37597752e-01 5.26466370e-01
-1.29259348e+00 -6.49540424e-01 7.06158280e-01 1.70780867e-01
-1.56032503e+00 -3.32516134e-01 7.19545066e-01 -2.48555318e-01
1.38219607e+00 3.36524308e-01 7.83080757e-01 1.13717222e+00
9.57939103e-02 4.26881790e-01 6.77763820e-01 -2.06288040e-01
9.37647521e-02 -7.72034079e-02 2.58641303e-01 9.36919689e-01
4.08913910e-01 5.96712995e-03 -1.18508510e-01 1.50467262e-01
5.09111881e-01 -3.58265489e-02 -1.87768936e-01 -3.49225193e-01
-9.47487473e-01 6.42915785e-01 8.22913766e-01 2.68944830e-01
-2.44239226e-01 2.92104751e-01 4.40300375e-01 3.90072316e-01
4.14363921e-01 5.12783945e-01 -7.34750032e-01 8.06284994e-02
-9.90595222e-01 5.19326143e-02 7.34875739e-01 1.06836987e+00
7.89912820e-01 4.44762081e-01 -7.14986622e-02 1.06666017e+00
1.69056877e-01 2.14574456e-01 3.18823963e-01 -6.60263240e-01
6.23441398e-01 9.49930310e-01 -6.33585989e-01 -1.19669700e+00
-6.85840905e-01 -1.09504819e+00 -1.38778007e+00 8.13725740e-02
9.36091617e-02 -2.11311355e-02 -1.26871872e+00 1.86528730e+00
6.08395338e-02 3.13292034e-02 -1.77432388e-01 8.58335793e-01
9.88005757e-01 3.14649403e-01 1.66967154e-01 1.32501781e-01
1.22892666e+00 -1.23893368e+00 -7.24956095e-02 -4.61013794e-01
6.99261785e-01 -4.21586245e-01 1.21116948e+00 6.48544848e-01
-1.12649846e+00 -5.09327114e-01 -1.09883618e+00 -3.02359164e-01
-3.05196881e-01 3.37560087e-01 6.25824571e-01 4.00170237e-01
-1.48912621e+00 8.75044107e-01 -9.08864558e-01 -3.25075686e-01
5.57786524e-01 7.37406969e-01 -2.34585106e-01 1.14726182e-02
-8.29172134e-01 1.02835858e+00 7.25376725e-01 5.12215495e-01
-7.97525644e-01 -6.65651321e-01 -6.00475669e-01 5.46596587e-01
5.97209692e-01 -9.67309117e-01 9.60161805e-01 -6.94856226e-01
-1.44419158e+00 6.61302865e-01 1.26384810e-01 -8.86761427e-01
4.78038192e-01 -1.87544435e-01 -5.89964725e-02 -1.63037717e-01
-4.22474861e-01 1.07039750e+00 9.57636595e-01 -1.21671236e+00
-2.50992894e-01 -2.68843740e-01 2.69042492e-01 1.99214667e-01
-4.50621963e-01 -9.12601352e-02 -7.96432018e-01 -7.57055461e-01
1.56536460e-01 -9.29486513e-01 -6.04720891e-01 -3.58831853e-01
-7.70429134e-01 -1.73627645e-01 7.88110554e-01 -6.08145535e-01
1.14281142e+00 -2.21432996e+00 4.57775384e-01 2.07570180e-01
6.40440881e-01 4.85635728e-01 -4.40202475e-01 5.33103533e-02
-3.03742196e-02 3.23209107e-01 -2.95355767e-01 -7.79980421e-01
-1.10008918e-01 5.74487627e-01 -1.75194964e-01 2.02097788e-01
2.89382041e-01 1.02493513e+00 -5.85503936e-01 -3.04094344e-01
2.18368575e-01 5.57902634e-01 -8.37749302e-01 -2.94996232e-01
-2.89892048e-01 2.32797250e-01 -2.36203760e-01 6.04376912e-01
6.60159588e-01 -2.57749647e-01 -3.72734945e-03 -4.43307936e-01
-2.80702002e-02 3.21778387e-01 -9.17787015e-01 1.57408738e+00
-4.23888594e-01 3.66895556e-01 2.39919946e-01 -1.19248939e+00
1.02163243e+00 -2.00250760e-01 1.79297328e-01 -6.79212511e-01
3.90365124e-01 2.04936564e-01 4.03808802e-01 -1.39443604e-02
5.41084707e-01 2.33128533e-01 2.49949694e-01 3.25789154e-02
2.93965161e-01 -5.29479235e-03 3.56971890e-01 7.30583072e-02
1.26362133e+00 -1.50678709e-01 -3.48218381e-02 -3.99390966e-01
4.55562681e-01 -1.20292813e-01 6.05112195e-01 8.72751415e-01
1.39768600e-01 6.99401438e-01 6.89023674e-01 -4.37677473e-01
-9.57717478e-01 -7.41318822e-01 1.78688884e-01 1.03847277e+00
-1.39762713e-02 -5.93429029e-01 -8.65529120e-01 -7.36014128e-01
-1.17032982e-01 5.67959785e-01 -4.93173242e-01 -3.76779824e-01
-8.62765789e-01 -8.94668818e-01 7.85581768e-01 4.42933589e-01
4.92529750e-01 -1.11533833e+00 -4.33220565e-01 1.94945276e-01
2.03763291e-01 -1.30713892e+00 -1.71133414e-01 6.31386697e-01
-1.29020870e+00 -8.96251678e-01 -5.08227110e-01 -6.77685499e-01
7.25734711e-01 2.80208290e-01 1.24448216e+00 3.57378602e-01
-6.32053167e-02 -1.54920489e-01 -1.16926812e-01 -2.61269063e-01
-1.60947919e-01 6.69600368e-01 -2.02440530e-01 -1.83586434e-01
1.37162618e-02 -9.29375708e-01 -6.36082649e-01 2.80734718e-01
-7.01712012e-01 2.49121025e-01 9.83423293e-01 6.50874734e-01
7.13061571e-01 -3.59750949e-02 1.87619522e-01 -9.57164109e-01
7.22216010e-01 -1.28681272e-01 -5.83904803e-01 2.90578812e-01
-5.99534750e-01 8.54736865e-02 1.02535534e+00 -4.83595759e-01
-5.06300151e-01 -1.31593227e-01 -4.03011709e-01 -8.88882101e-01
-6.28859922e-02 6.11688435e-01 -2.48649284e-01 -2.93311864e-01
5.92751920e-01 2.71306038e-01 -2.23517731e-01 -7.96490014e-01
4.41648662e-01 -1.34518161e-01 5.65623224e-01 -3.95030200e-01
8.32341671e-01 3.28471839e-01 1.78781807e-01 -8.94859552e-01
-7.54839897e-01 -6.39982745e-02 -5.71742475e-01 2.33112589e-01
5.93946099e-01 -7.68126070e-01 -5.20054221e-01 3.14005941e-01
-1.33187211e+00 -5.35765529e-01 -3.59157890e-01 2.96261966e-01
-6.13200776e-02 3.46533418e-01 -6.87131464e-01 -2.87117600e-01
-7.09661663e-01 -1.34400415e+00 7.77258575e-01 -4.23319936e-02
-1.15584023e-02 -8.63288105e-01 -3.20031673e-01 1.02608919e-01
5.29722154e-01 1.77533269e-01 1.28766358e+00 -7.10646629e-01
-7.58041918e-01 1.39442340e-01 -6.23117089e-01 8.85624647e-01
-2.41668716e-01 3.51123400e-02 -7.84249544e-01 -4.21378016e-01
1.86744601e-01 -7.71939978e-02 1.39824629e+00 3.57245326e-01
1.54474199e+00 -4.75171208e-01 -2.63246536e-01 1.30936444e+00
1.22455370e+00 -1.32263169e-01 5.25777161e-01 2.54234970e-01
1.20830274e+00 4.41524595e-01 -1.27303094e-01 3.79220136e-02
2.63981611e-01 4.12600905e-01 9.85243738e-01 -2.65540451e-01
-2.73531884e-01 1.39037175e-02 4.24039364e-02 1.22461045e+00
-4.38770473e-01 -3.81179690e-01 -8.11429679e-01 3.36640745e-01
-1.63459444e+00 -4.81478393e-01 -1.59821108e-01 1.73132133e+00
5.44630468e-01 4.33517933e-01 -2.84808010e-01 -1.88869946e-02
4.60552335e-01 3.29612970e-01 -6.21837199e-01 -4.11331534e-01
-1.99862003e-01 4.99933869e-01 7.24776626e-01 2.50241339e-01
-9.17793989e-01 1.18678415e+00 5.35202503e+00 9.09656584e-01
-1.25894296e+00 1.74411267e-01 6.41891658e-01 -2.96172291e-01
-2.38563508e-01 4.72517498e-02 -1.01519716e+00 1.72993585e-01
6.97757363e-01 2.65249193e-01 6.80234313e-01 8.80277693e-01
1.29079863e-01 3.95578057e-01 -1.13704276e+00 1.00518310e+00
-1.12855203e-01 -1.29419506e+00 4.89755154e-01 8.95847678e-02
5.08746922e-01 6.08835816e-01 1.53589785e-01 6.25929773e-01
3.24167401e-01 -1.16148353e+00 8.91940415e-01 1.96740031e-01
5.51199198e-01 -7.90479898e-01 5.02520442e-01 3.46470863e-01
-9.79890168e-01 -2.00588852e-01 -7.78937042e-01 2.82363687e-02
-1.20137259e-02 7.85032570e-01 -6.20998383e-01 6.26830578e-01
9.40044701e-01 7.11053908e-01 -9.51448262e-01 1.10856926e+00
-2.75070995e-01 5.94029427e-01 -5.48625886e-01 1.00669160e-01
6.32000387e-01 -2.93405980e-01 6.36325479e-01 1.29965985e+00
4.33841079e-01 -2.46442378e-01 -9.96844471e-02 1.02125001e+00
-2.82307863e-01 3.73097770e-02 -4.87701535e-01 1.09317809e-01
2.49630168e-01 1.45046222e+00 -9.99291956e-01 -7.61999413e-02
-2.00159401e-01 6.69268906e-01 7.97282994e-01 4.84996796e-01
-9.60898697e-01 -2.78027151e-02 5.85090220e-01 1.35514559e-02
5.93154013e-01 -4.14620310e-01 -4.37829882e-01 -1.12123573e+00
2.13178158e-01 -9.60803151e-01 1.69079825e-01 -4.33716983e-01
-9.51672316e-01 1.06213176e+00 -7.94770718e-02 -9.17417228e-01
1.19725130e-01 -7.27209330e-01 -6.17480457e-01 5.91628373e-01
-1.60933876e+00 -1.18954182e+00 -2.74770051e-01 6.81092322e-01
3.59437406e-01 -2.65530765e-01 5.26623666e-01 4.80547130e-01
-8.96982610e-01 7.92821229e-01 -1.83965594e-01 5.01645217e-03
2.41565242e-01 -1.12400508e+00 7.05734015e-01 8.91311526e-01
3.36243302e-01 8.24234784e-01 4.31983918e-01 -4.50605839e-01
-1.19246423e+00 -1.41704643e+00 5.42781115e-01 -1.41713262e-01
6.12415493e-01 -7.66900241e-01 -1.13237369e+00 6.48295701e-01
8.58967677e-02 -3.47593650e-02 6.26803935e-02 3.56907606e-01
-4.62484658e-01 -1.90542728e-01 -8.51095319e-01 9.05961633e-01
1.59021568e+00 -9.55352932e-02 -4.32484895e-01 2.37908944e-01
1.13683414e+00 -4.53327954e-01 -4.54960108e-01 6.50851965e-01
1.47512496e-01 -1.12372041e+00 1.13430607e+00 -4.46689457e-01
2.02745438e-01 -8.92183855e-02 -2.11991388e-02 -1.23232567e+00
-3.77180338e-01 -6.32760584e-01 -4.02862936e-01 1.09037888e+00
6.24374568e-01 -7.46316731e-01 9.27999496e-01 1.49973974e-01
-6.51966751e-01 -9.70740914e-01 -1.00228727e+00 -8.31563056e-01
8.51604715e-02 -5.25281787e-01 6.90245807e-01 8.23303580e-01
-9.18261826e-01 5.57206154e-01 -2.76974410e-01 2.33318359e-01
4.06440467e-01 -3.31483036e-01 7.40913868e-01 -1.40048325e+00
-4.95210826e-01 -1.04070723e+00 -3.15757245e-01 -1.28121257e+00
2.33542994e-02 -9.17135000e-01 -2.16119364e-01 -1.48920083e+00
-1.16379999e-01 -7.63152242e-01 -4.17723894e-01 7.79393196e-01
9.12685916e-02 2.91919857e-01 2.99160928e-01 1.93121746e-01
-5.75437129e-01 5.94325423e-01 1.27566099e+00 -3.53453964e-01
-3.42504591e-01 3.42355520e-02 -8.49529147e-01 1.00364995e+00
7.20959783e-01 -5.42357922e-01 -6.66459680e-01 -1.05912578e+00
4.93523419e-01 -5.62085867e-01 6.85441554e-01 -1.18357515e+00
3.00874621e-01 1.43559501e-01 5.68617694e-02 -7.16425240e-01
3.13671619e-01 -7.88375378e-01 1.62182629e-01 4.58834231e-01
6.28278404e-02 3.66612583e-01 4.42109942e-01 4.34058547e-01
-1.70771405e-01 -3.65196019e-01 6.44823730e-01 -3.04222107e-01
-5.59607387e-01 4.91395563e-01 7.25510269e-02 7.84154981e-02
5.18954217e-01 -3.61434489e-01 -4.09916401e-01 1.60950884e-01
-9.17432725e-01 2.11105198e-01 2.59336203e-01 3.16343635e-01
4.93876934e-01 -8.63855898e-01 -5.69827437e-01 1.37396723e-01
-2.37103939e-01 7.28205740e-01 4.15495187e-01 1.09461701e+00
-6.34559274e-01 3.81683081e-01 1.79566871e-02 -7.40836263e-01
-1.14320290e+00 5.32256007e-01 5.11430144e-01 -7.33573318e-01
-1.01757705e+00 1.08929241e+00 5.11554420e-01 -5.18238604e-01
4.56964403e-01 -9.98514295e-01 -1.96880668e-01 -7.02610984e-02
6.88138977e-02 4.85633500e-02 6.18634760e-01 -3.36808890e-01
-1.73505098e-01 4.62517530e-01 -4.28398877e-01 5.41866601e-01
1.59674287e+00 3.49702090e-02 -2.35411733e-01 -5.60206994e-02
1.09279501e+00 -2.95113146e-01 -1.24863803e+00 -1.46921843e-01
-2.50786301e-02 1.84332520e-01 8.78426433e-02 -6.69057190e-01
-1.70801055e+00 1.05251813e+00 2.60810524e-01 3.54574919e-01
1.15811872e+00 -1.50616318e-01 7.50025749e-01 7.69505978e-01
2.58391827e-01 -9.14337456e-01 -1.60408661e-01 9.75389004e-01
9.24792051e-01 -7.36974537e-01 1.63802709e-02 -3.35324258e-01
-2.32490331e-01 8.17831039e-01 7.63714075e-01 -2.82959402e-01
5.25043070e-01 2.12658554e-01 -2.79663771e-01 -3.37995499e-01
-8.37317824e-01 -2.18171507e-01 3.20868760e-01 4.53039199e-01
-2.16743480e-02 -2.38981605e-01 -7.21645951e-02 4.47214991e-01
-5.61701655e-01 -3.55417699e-01 7.87033066e-02 5.48803568e-01
-1.56180575e-01 -1.01956975e+00 5.33111878e-02 6.23329818e-01
-4.11779672e-01 -5.09428561e-01 -4.90524828e-01 1.11294198e+00
3.58408004e-01 3.73448193e-01 2.85520293e-02 -5.08049428e-01
5.77079296e-01 -2.65463144e-01 4.66073960e-01 -6.87624991e-01
-1.03521705e+00 -5.37133999e-02 8.01809281e-02 -3.88630331e-01
-1.07135899e-01 -2.42608950e-01 -9.53530908e-01 -4.35372949e-01
-4.29714501e-01 -1.76270649e-01 5.65979838e-01 9.10238683e-01
5.34550548e-01 8.71909380e-01 8.93085822e-02 -8.86314631e-01
-5.73893785e-01 -9.41060066e-01 -2.36032486e-01 1.42899379e-01
1.04399055e-01 -6.72216594e-01 -3.69238853e-01 -3.90689820e-01] | [8.602266311645508, 3.2327702045440674] |
ac29775c-3e1b-469d-9c2c-4501ac890b84 | learning-content-enhanced-mask-transformer | 2307.00371 | null | https://arxiv.org/abs/2307.00371v1 | https://arxiv.org/pdf/2307.00371v1.pdf | Learning Content-enhanced Mask Transformer for Domain Generalized Urban-Scene Segmentation | Domain-generalized urban-scene semantic segmentation (USSS) aims to learn generalized semantic predictions across diverse urban-scene styles. Unlike domain gap challenges, USSS is unique in that the semantic categories are often similar in different urban scenes, while the styles can vary significantly due to changes in urban landscapes, weather conditions, lighting, and other factors. Existing approaches typically rely on convolutional neural networks (CNNs) to learn the content of urban scenes. In this paper, we propose a Content-enhanced Mask TransFormer (CMFormer) for domain-generalized USSS. The main idea is to enhance the focus of the fundamental component, the mask attention mechanism, in Transformer segmentation models on content information. To achieve this, we introduce a novel content-enhanced mask attention mechanism. It learns mask queries from both the image feature and its down-sampled counterpart, as lower-resolution image features usually contain more robust content information and are less sensitive to style variations. These features are fused into a Transformer decoder and integrated into a multi-resolution content-enhanced mask attention learning scheme. Extensive experiments conducted on various domain-generalized urban-scene segmentation datasets demonstrate that the proposed CMFormer significantly outperforms existing CNN-based methods for domain-generalized semantic segmentation, achieving improvements of up to 14.00\% in terms of mIoU (mean intersection over union). The source code for CMFormer will be made available at this \href{https://github.com/BiQiWHU/domain-generalized-urban-scene-segmentation}{repository}. | ['Theo Gevers', 'ShaoDi You', 'Qi Bi'] | 2023-07-01 | null | null | null | null | ['scene-segmentation'] | ['computer-vision'] | [ 3.42086166e-01 -3.96401882e-02 -1.25549659e-01 -5.53060949e-01
-8.18733335e-01 -4.61212367e-01 3.93542916e-01 -1.02123111e-01
-1.77044600e-01 4.83138204e-01 2.08384424e-01 -5.43456115e-02
8.72336999e-02 -9.15347278e-01 -9.03015912e-01 -4.58161563e-01
4.99687731e-01 2.50928879e-01 4.40251678e-01 -4.33913797e-01
9.95913371e-02 2.96489388e-01 -1.44536352e+00 2.93837845e-01
1.34048796e+00 1.00792480e+00 9.54643130e-01 3.87022257e-01
-3.22039038e-01 4.61776406e-01 -4.71532822e-01 -7.27441087e-02
4.20335889e-01 -4.04073060e-01 -6.71611309e-01 3.71828198e-01
4.81736779e-01 -1.96375445e-01 -3.43524516e-01 1.24128866e+00
1.97337091e-01 2.80295730e-01 4.68407393e-01 -1.10127866e+00
-7.43556798e-01 5.09224951e-01 -4.82227534e-01 9.02582556e-02
-1.08102814e-03 1.75124854e-01 8.44801009e-01 -8.86220336e-01
4.97593671e-01 1.35430241e+00 4.71883148e-01 3.97724956e-01
-1.21178424e+00 -8.43194067e-01 6.04860187e-01 2.09308892e-01
-1.55383682e+00 -2.41823018e-01 9.66156781e-01 -4.28934038e-01
6.83999181e-01 3.30528855e-01 5.31423271e-01 8.48253608e-01
-1.08121105e-01 1.06207895e+00 1.06383038e+00 -9.71251205e-02
1.16889663e-01 1.77636862e-01 -1.48457691e-01 5.38772225e-01
1.41993806e-01 -1.23257421e-01 -1.92677975e-01 5.33217728e-01
9.33030427e-01 1.15194112e-01 -3.32038581e-01 -2.48658702e-01
-1.10949528e+00 7.82183528e-01 8.75062406e-01 3.29964072e-01
-1.32391065e-01 1.04031824e-01 1.18266352e-01 1.62806705e-01
7.39826500e-01 3.61846834e-01 -6.80274844e-01 7.55443722e-02
-9.05191243e-01 3.17432463e-01 2.69136369e-01 1.17959988e+00
1.14214623e+00 2.88336068e-01 -2.69094080e-01 1.25578511e+00
2.71184117e-01 6.21964693e-01 4.14994836e-01 -1.05635059e+00
4.87710118e-01 7.23591208e-01 -4.17772420e-02 -8.44015121e-01
-2.21835330e-01 -7.44372904e-01 -7.49659956e-01 -7.58045092e-02
1.80147097e-01 -1.72001533e-02 -1.25514913e+00 1.82476926e+00
3.07631761e-01 3.68659467e-01 -2.23028079e-01 9.27893162e-01
1.08059251e+00 6.09116495e-01 3.35341036e-01 3.67863268e-01
1.19242406e+00 -1.07208371e+00 -4.35588568e-01 -7.90619373e-01
3.87642771e-01 -7.39149153e-01 1.21551454e+00 -2.16922238e-02
-8.48114192e-01 -8.85545492e-01 -1.04067016e+00 -1.63687408e-01
-6.55378222e-01 1.61072657e-01 2.22685024e-01 5.43390036e-01
-9.74930108e-01 3.56868476e-01 -5.06809890e-01 -5.56753874e-01
7.61337876e-01 1.25045851e-01 -1.12251259e-01 -2.16470882e-01
-1.17677438e+00 5.57198763e-01 6.12650275e-01 -1.86976165e-01
-7.03709126e-01 -8.02628696e-01 -1.13710833e+00 1.19961752e-03
5.33247650e-01 -5.93890071e-01 1.21300268e+00 -1.23296249e+00
-1.25384665e+00 9.60154176e-01 -3.55549037e-01 -2.42176622e-01
4.00305063e-01 -2.36147687e-01 -3.35323513e-01 1.29445279e-02
6.40179336e-01 1.06552470e+00 8.61885011e-01 -1.59209585e+00
-7.73908615e-01 -1.70866743e-01 8.29839259e-02 3.91328692e-01
2.87702363e-02 -3.25585902e-01 -7.91299224e-01 -1.00840735e+00
2.52879560e-01 -7.54800618e-01 -3.91353339e-01 -1.79046109e-01
-4.94954020e-01 5.49063608e-02 1.04107535e+00 -6.46097839e-01
1.02717757e+00 -2.31135035e+00 4.79339436e-02 -7.30057582e-02
-3.13084126e-02 2.91240513e-01 -3.36459905e-01 1.10768981e-01
-9.62750465e-02 2.59141713e-01 -5.52875876e-01 -4.10453081e-01
6.58898056e-02 2.95480043e-01 -1.34332657e-01 1.84067041e-01
4.28353369e-01 1.07973254e+00 -8.72884095e-01 -4.17587966e-01
6.31914914e-01 2.56278843e-01 -5.29815793e-01 1.29619762e-01
-4.89556283e-01 8.24170530e-01 -5.31584561e-01 7.78622270e-01
8.56093168e-01 -1.00691520e-01 -1.65473908e-01 -2.22717911e-01
-9.50966999e-02 1.24458581e-01 -1.19761503e+00 1.88908207e+00
-5.50161183e-01 7.36530840e-01 1.67932570e-01 -1.14080310e+00
9.19885933e-01 -7.56139234e-02 4.35087085e-01 -9.85855997e-01
3.19197357e-01 2.10956156e-01 -4.34558004e-01 -1.52101383e-01
6.54863477e-01 3.52625698e-02 -3.57371658e-01 -3.41387279e-02
-9.56954435e-03 -5.62665045e-01 2.18904674e-01 5.47293089e-02
5.33013046e-01 7.02498257e-02 2.36480609e-02 -4.04684365e-01
5.47430158e-01 1.02380209e-01 7.79116690e-01 7.15145886e-01
-2.87948161e-01 7.63255954e-01 1.70224771e-01 -1.12907387e-01
-9.07664776e-01 -1.31245708e+00 -2.59803265e-01 1.11860275e+00
6.95178986e-01 -1.50978819e-01 -8.91100168e-01 -4.91179943e-01
-1.06471509e-01 9.00857627e-01 -4.96983021e-01 -1.29396737e-01
-4.74610686e-01 -4.68348026e-01 2.18131587e-01 6.50274456e-01
1.21334660e+00 -8.88255119e-01 -3.76919389e-01 2.66900420e-01
-4.35831636e-01 -1.39550912e+00 -5.46170831e-01 -6.27016323e-03
-6.75685346e-01 -9.46010411e-01 -6.70954823e-01 -8.94667923e-01
5.12167394e-01 5.76410115e-01 1.06421292e+00 -5.96560836e-02
-1.77651778e-01 3.67566019e-01 -4.29002076e-01 -5.09986222e-01
-1.09464727e-01 2.81482756e-01 -4.03559536e-01 1.88179255e-01
2.63263732e-01 -5.38690448e-01 -7.19674110e-01 3.87052357e-01
-1.00038290e+00 4.77137715e-01 6.14118278e-01 7.51566529e-01
7.24655032e-01 1.74393818e-01 4.16490793e-01 -9.92728651e-01
1.83574319e-01 -6.12107456e-01 -4.69729096e-01 -7.89315253e-02
-2.07425162e-01 -3.58714759e-01 4.30893421e-01 -2.62393653e-01
-1.25209773e+00 4.64220010e-02 -3.06526184e-01 -4.52678651e-01
-5.85980713e-01 2.32384846e-01 -6.41672909e-01 1.71154603e-01
3.85319412e-01 3.05033535e-01 -4.62223321e-01 -5.25852263e-01
4.92428154e-01 5.26282132e-01 5.40808439e-01 -6.06691182e-01
9.27630782e-01 5.62448800e-01 -3.81715328e-01 -9.90723848e-01
-9.74989176e-01 -4.88596827e-01 -6.81073129e-01 -1.87002584e-01
1.06742465e+00 -1.25299573e+00 9.76263732e-02 7.14579344e-01
-8.29438508e-01 -9.07753408e-01 -3.68407100e-01 1.45165801e-01
-6.13498688e-01 9.59231108e-02 -4.08532202e-01 -4.16153193e-01
-3.60509828e-02 -1.19880962e+00 1.34306967e+00 4.93065029e-01
-1.20095491e-01 -8.98880303e-01 -4.18691605e-01 5.84550023e-01
4.60501224e-01 3.65229607e-01 9.07402098e-01 -3.17686081e-01
-8.79829526e-01 1.59959823e-01 -6.08636856e-01 4.49029148e-01
3.36196989e-01 -2.71656305e-01 -9.71519351e-01 -6.47776201e-02
-2.35878617e-01 4.58932668e-03 1.05808973e+00 6.71821177e-01
1.50169146e+00 3.31826173e-02 -3.11634094e-01 8.33588421e-01
1.71272826e+00 2.65866935e-01 5.86333632e-01 3.20515007e-01
1.01154208e+00 4.13623482e-01 8.20421755e-01 2.38626465e-01
4.51731980e-01 6.24542177e-01 4.43975806e-01 -4.49809462e-01
-4.02621418e-01 -2.99742788e-01 9.26473662e-02 3.76007229e-01
2.70264924e-01 -3.25015515e-01 -7.33321488e-01 8.97633433e-01
-1.78337812e+00 -7.22215474e-01 -2.26472363e-01 1.79151464e+00
4.97974753e-01 2.04092354e-01 -1.04834840e-01 -2.32957881e-02
7.30338633e-01 4.57493395e-01 -7.97275424e-01 -3.55937302e-01
-3.20743829e-01 3.40951711e-01 7.16883779e-01 5.15401542e-01
-1.32084167e+00 1.45537400e+00 4.93913746e+00 1.14042974e+00
-1.14870024e+00 2.60051221e-01 8.48786354e-01 9.35613513e-02
-3.67979318e-01 -2.15447590e-01 -4.98045325e-01 5.32632053e-01
4.24399078e-01 3.55403163e-02 3.28083247e-01 9.15000677e-01
3.66981179e-01 -1.99173406e-01 -5.62024474e-01 9.71011519e-01
-1.81111604e-01 -1.19342494e+00 -3.84199247e-02 -1.87474117e-01
1.02169919e+00 2.37175912e-01 3.59027922e-01 4.04145181e-01
4.34185624e-01 -9.49090540e-01 1.02792859e+00 2.86200106e-01
9.91000414e-01 -6.71914816e-01 5.03803313e-01 1.01359241e-01
-1.72285473e+00 -1.45643681e-01 -2.35293806e-01 9.94217023e-03
4.85165752e-02 5.23546278e-01 -4.87748712e-01 6.86547577e-01
9.23579812e-01 1.02410114e+00 -6.08581126e-01 8.05243492e-01
-2.22021505e-01 6.08365953e-01 -7.83227682e-02 3.27932090e-01
5.12059867e-01 -2.30958343e-01 6.09393299e-01 1.26592612e+00
3.28270167e-01 6.80818558e-02 4.21773791e-01 1.11756730e+00
1.07596759e-02 7.86947012e-02 -4.09840226e-01 1.28181815e-01
4.74133343e-01 1.04552829e+00 -9.53149855e-01 -5.04333973e-01
-5.05042613e-01 1.13435531e+00 7.06725046e-02 6.76843941e-01
-9.76490676e-01 -1.51543811e-01 1.11706018e+00 2.65484929e-01
6.67759180e-01 -2.47307792e-01 -6.74994886e-01 -1.02263224e+00
-2.40353987e-01 -6.97615206e-01 2.02285901e-01 -8.72958422e-01
-1.09364760e+00 4.71986800e-01 1.33528650e-01 -1.10675716e+00
3.07626456e-01 -5.22135079e-01 -5.82808554e-01 9.07713771e-01
-1.55539954e+00 -1.48278868e+00 -5.72185636e-01 6.75827920e-01
1.31165528e+00 5.98729625e-02 3.76497805e-01 4.07422900e-01
-6.96580589e-01 3.82752419e-01 1.80821806e-01 1.59735441e-01
5.05408585e-01 -1.26969182e+00 3.86673898e-01 1.05797398e+00
-1.17429756e-01 9.91620496e-02 5.66620052e-01 -6.75161123e-01
-7.50409961e-01 -1.83842552e+00 5.49025476e-01 -1.87989116e-01
3.61327589e-01 -3.70029747e-01 -7.84746289e-01 7.11678267e-01
6.43519685e-02 -2.98680305e-01 3.44217151e-01 -2.24302426e-01
-2.91682661e-01 -2.47564033e-01 -1.12501311e+00 6.97405756e-01
1.35028040e+00 -4.91890728e-01 -4.35072601e-01 8.34666118e-02
9.14851665e-01 -5.17509341e-01 -5.88424563e-01 5.83671987e-01
9.80703607e-02 -1.06585038e+00 1.10118055e+00 3.06094550e-02
4.01115656e-01 -4.57757026e-01 -5.77675283e-01 -1.33018470e+00
-6.80498719e-01 -5.94960824e-02 3.67755324e-01 1.18819296e+00
1.83019593e-01 -6.60334826e-01 7.21668780e-01 2.97579199e-01
-6.35724068e-01 -6.92756355e-01 -7.65671074e-01 -5.92639327e-01
2.23313153e-01 -7.34614730e-01 9.88389730e-01 8.81628513e-01
-6.61932230e-01 2.17358708e-01 -1.97920412e-01 3.42043638e-01
4.63625342e-01 2.65607864e-01 7.45035231e-01 -1.02412760e+00
-8.51183012e-02 -7.76986063e-01 -1.96401924e-01 -1.31020391e+00
1.32275164e-01 -7.94207990e-01 1.55327082e-01 -1.85031843e+00
4.80202027e-03 -4.93344009e-01 -2.85067081e-01 3.45377684e-01
-4.32311714e-01 5.09932101e-01 3.22983325e-01 -3.09883840e-02
-6.09698832e-01 6.83277905e-01 1.37387431e+00 -4.23870504e-01
-2.54598916e-01 -5.11463061e-02 -8.96741927e-01 7.09786117e-01
1.16058457e+00 -1.82803869e-01 -5.96183181e-01 -6.41434312e-01
-3.98344398e-01 -3.88706923e-01 4.69010949e-01 -1.22883809e+00
-2.16058716e-01 -3.92272621e-01 4.45044547e-01 -7.57425487e-01
4.25352752e-01 -7.04679072e-01 2.48248696e-01 4.30449456e-01
3.97316590e-02 -3.73470753e-01 4.21193689e-01 5.47226548e-01
-2.35966384e-01 1.65759921e-01 1.00768173e+00 -4.73684371e-01
-1.33607256e+00 4.25681382e-01 -5.37040472e-01 1.88254490e-01
8.96035135e-01 -5.34786582e-01 -7.33242854e-02 -2.81844318e-01
-6.65018559e-01 3.83325696e-01 6.83881223e-01 6.25365138e-01
5.76875269e-01 -1.23322296e+00 -6.59272134e-01 2.89300799e-01
2.19777495e-01 4.61909801e-01 7.23338068e-01 5.13771415e-01
-4.43237990e-01 3.95666301e-01 -2.05403194e-01 -7.39640057e-01
-9.64523435e-01 2.54650563e-01 5.23748338e-01 -2.88067348e-02
-3.95174950e-01 1.03019953e+00 7.82946229e-01 -5.68601310e-01
-1.54063120e-01 -5.52716732e-01 -1.72090530e-03 -1.51689857e-01
7.18389452e-02 1.74980491e-01 -1.16385795e-01 -1.00003231e+00
-1.11919329e-01 7.86069691e-01 1.86137274e-01 1.69261903e-01
1.15642703e+00 -4.71809626e-01 1.40508100e-01 3.08058798e-01
1.08869803e+00 -2.18486086e-01 -1.50578976e+00 -5.53128660e-01
-2.29005516e-01 -6.96860909e-01 5.90398982e-02 -7.71182001e-01
-1.23981786e+00 7.13290572e-01 7.04314232e-01 -8.30204189e-02
1.51728773e+00 2.78195798e-01 1.13931978e+00 -2.01517463e-01
2.40486145e-01 -1.21302438e+00 1.61027506e-01 6.07249737e-01
8.55075300e-01 -1.43370402e+00 -2.81178385e-01 -7.61883974e-01
-6.41435921e-01 5.70597649e-01 8.59418273e-01 -2.25572214e-01
7.07333505e-01 -1.59033120e-01 2.54942983e-01 -1.80568278e-01
-1.67188525e-01 -7.34651983e-01 3.73969644e-01 7.38705754e-01
1.12534605e-01 3.93797368e-01 2.47720405e-02 6.55505061e-01
-3.45598608e-01 -3.83954138e-01 1.23101838e-01 7.11268306e-01
-6.03545070e-01 -1.00641072e+00 -4.01898026e-01 5.10088682e-01
-2.99857222e-02 -1.11446455e-01 -2.63703167e-01 6.69436336e-01
7.66085267e-01 1.13295579e+00 2.15687945e-01 -4.50823069e-01
4.30867434e-01 -1.25908434e-01 2.35205024e-01 -8.17267537e-01
-2.05667824e-01 1.65291965e-01 4.04327400e-02 -7.18155086e-01
-4.00337011e-01 -6.02287114e-01 -1.11627138e+00 -2.52299845e-01
4.97191846e-02 -2.14155793e-01 4.22806084e-01 9.21921968e-01
3.21959496e-01 7.71882474e-01 5.50864935e-01 -1.03098273e+00
3.66613060e-01 -8.13585877e-01 -5.19150853e-01 6.39494777e-01
1.84997380e-01 -8.68793964e-01 5.73121756e-03 9.63169560e-02] | [9.675786018371582, 1.136521816253662] |
bc92107a-f356-47a2-99a0-8adaea07f349 | spatio-temporal-latent-graph-structure | 2202.12586 | null | https://arxiv.org/abs/2202.12586v2 | https://arxiv.org/pdf/2202.12586v2.pdf | Spatio-Temporal Latent Graph Structure Learning for Traffic Forecasting | Accurate traffic forecasting, the foundation of intelligent transportation systems (ITS), has never been more significant than nowadays due to the prosperity of smart cities and urban computing. Recently, Graph Neural Network truly outperforms the traditional methods. Nevertheless, the most conventional GNN-based model works well while given a pre-defined graph structure. And the existing methods of defining the graph structures focus purely on spatial dependencies and ignore the temporal correlation. Besides, the semantics of the static pre-defined graph adjacency applied during the whole training progress is always incomplete, thus overlooking the latent topologies that may fine-tune the model. To tackle these challenges, we propose a new traffic forecasting framework -- Spatio-Temporal Latent Graph Structure Learning networks (ST-LGSL). More specifically, the model employs a graph generator based on Multilayer perceptron and K-Nearest Neighbor, which learns the latent graph topological information from the entire data considering both spatial and temporal dynamics. Furthermore, with the initialization of MLP-kNN based on ground-truth adjacency matrix and similarity metric in kNN, ST-LGSL aggregates the topologies focusing on geography and node similarity. Additionally, the generated graphs act as the input of the Spatio-temporal prediction module combined with the Diffusion Graph Convolutions and Gated Temporal Convolutions Networks. Experimental results on two benchmarking datasets in real world demonstrate that ST-LGSL outperforms various types of state-of-art baselines. | ['Tianrui Li', 'Jie Hu', 'Shengdong Du', 'Shijing Liu', 'Tang Qian', 'Jiabin Tang'] | 2022-02-25 | null | null | null | null | ['graph-structure-learning'] | ['graphs'] | [-1.89082280e-01 -9.53653306e-02 -2.67621607e-01 -1.77534357e-01
1.64629161e-01 -2.43369713e-01 7.74369061e-01 1.19015299e-01
-1.05935037e-01 6.10016406e-01 9.47793573e-02 -7.73672342e-01
-4.79930967e-01 -1.66683900e+00 -5.76073050e-01 -7.63391554e-01
-3.83828342e-01 6.32881522e-01 5.29883564e-01 -4.43956167e-01
1.33995309e-01 5.83920777e-01 -1.56045997e+00 1.05425954e-01
1.21849608e+00 1.06823015e+00 2.02146918e-01 4.75328773e-01
-4.25766230e-01 8.33429694e-01 -2.87471451e-02 -3.43615621e-01
5.19217066e-02 -6.99665397e-02 -4.34575409e-01 -1.03179716e-01
3.48705798e-01 1.69411171e-02 -1.05821133e+00 8.76248896e-01
1.58524647e-01 5.08912504e-01 4.32851344e-01 -1.47475278e+00
-8.29271555e-01 6.64927065e-01 -1.32068902e-01 4.61968154e-01
-5.35646006e-02 5.02591670e-01 9.51663435e-01 -3.79754663e-01
5.31731427e-01 1.15713048e+00 8.01247120e-01 2.07306296e-02
-1.08541036e+00 -5.78970492e-01 7.07149923e-01 6.23655856e-01
-1.31721997e+00 -4.42224592e-02 1.18750286e+00 -5.43063939e-01
1.03261316e+00 1.54820204e-01 8.75609577e-01 1.03373969e+00
4.02211010e-01 3.57164115e-01 7.71180511e-01 8.92671291e-03
1.41203254e-01 -3.51529896e-01 8.09078887e-02 9.74668801e-01
1.28556609e-01 2.77220964e-01 -1.73942670e-01 2.56953120e-01
6.36634707e-01 1.21272333e-01 8.59071687e-02 -2.47826964e-01
-1.10617185e+00 6.33943260e-01 1.02310669e+00 2.98452437e-01
-4.87232864e-01 3.40671569e-01 3.48597646e-01 9.83761027e-02
7.27115333e-01 -1.67561784e-01 -4.68996018e-02 7.10516572e-02
-8.92968774e-01 2.60399804e-02 5.32413423e-01 8.23293746e-01
1.04270041e+00 2.43159845e-01 -1.32892177e-01 3.90235513e-01
2.84727812e-01 3.52921605e-01 2.40122706e-01 -4.29856509e-01
8.85335803e-01 1.20483804e+00 -3.28644931e-01 -1.76184893e+00
-5.45027614e-01 -6.83566034e-01 -1.18873549e+00 -1.55752495e-01
4.63978440e-01 -3.07825976e-03 -1.06994283e+00 1.53029525e+00
2.13208064e-01 9.72619891e-01 -1.53557777e-01 6.03570640e-01
9.43709135e-01 8.37244391e-01 1.55038998e-01 7.08507597e-02
6.52930975e-01 -1.18910027e+00 -4.40769434e-01 -1.79147840e-01
7.12237298e-01 -1.38858140e-01 9.47633028e-01 -2.77582079e-01
-5.66085875e-01 -6.61370218e-01 -7.23639846e-01 1.81262568e-01
-1.01582050e+00 -6.66714013e-02 9.41408515e-01 4.99343872e-01
-1.37174726e+00 6.40768230e-01 -7.80896187e-01 -5.64221382e-01
2.92022914e-01 2.86555827e-01 -1.82803661e-01 -1.18910700e-01
-1.37877655e+00 6.69506729e-01 5.80670178e-01 6.94003880e-01
-5.03604591e-01 -6.03745639e-01 -8.04679632e-01 1.89170033e-01
3.67837697e-01 -7.16432869e-01 3.57031375e-01 -6.09627008e-01
-1.38834345e+00 2.84627140e-01 3.48628163e-02 -5.84840894e-01
5.20985603e-01 5.39186478e-01 -9.80476975e-01 -1.39396429e-01
9.72043127e-02 4.40956444e-01 4.57498789e-01 -9.20311332e-01
-6.72852933e-01 2.04871036e-02 2.19125122e-01 1.09049886e-01
-1.73358485e-01 -5.93146265e-01 -6.03172898e-01 -5.01885593e-01
1.92080528e-01 -7.86240399e-01 -5.01229346e-01 -4.83184487e-01
-4.88843918e-01 -3.75899762e-01 9.80960011e-01 -5.07747471e-01
1.65322638e+00 -1.81099546e+00 -1.69909149e-01 6.10031188e-01
4.70238954e-01 2.61430502e-01 -3.02878439e-01 6.20272636e-01
-1.37687594e-01 -1.49954498e-01 -6.90690875e-02 -5.84465303e-02
1.61247700e-02 4.21862781e-01 -3.55280578e-01 3.14270288e-01
-5.97907640e-02 1.39985204e+00 -1.25638461e+00 -3.95577610e-01
6.06405854e-01 5.15045166e-01 -2.54921436e-01 -1.53582379e-01
-4.54398990e-01 5.75727761e-01 -5.15776634e-01 3.65073562e-01
6.27015233e-01 -5.37657976e-01 3.04011911e-01 -1.96827888e-01
-2.40954638e-01 3.59933168e-01 -9.77447271e-01 1.25546372e+00
-3.95560712e-01 5.60359478e-01 -5.83428741e-01 -1.39325273e+00
1.11318862e+00 -9.45412815e-02 6.44634008e-01 -1.05864489e+00
-1.95841696e-02 3.89163755e-02 -8.13909918e-02 -4.95300502e-01
2.91565090e-01 3.79140943e-01 1.00044169e-01 2.63888687e-01
-8.43523368e-02 6.27123952e-01 4.72549230e-01 3.20630014e-01
1.23626924e+00 -7.66668469e-02 -2.99844891e-01 -4.96245138e-02
7.44396627e-01 6.84253946e-02 3.80813181e-01 5.18349469e-01
-2.64024083e-02 1.62880301e-01 6.31289303e-01 -9.99379396e-01
-6.70275211e-01 -1.13277328e+00 2.89668202e-01 8.78126323e-01
2.73253292e-01 -3.35388333e-01 -4.87783581e-01 -8.68084133e-01
4.56274599e-02 6.68386698e-01 -6.87383354e-01 -2.36912385e-01
-7.97374904e-01 -7.83204913e-01 4.12686527e-01 3.67231935e-01
6.94400787e-01 -1.15719521e+00 2.10770473e-01 4.54245508e-01
-1.34362429e-01 -1.35257351e+00 -4.52095509e-01 -3.69538426e-01
-8.73680353e-01 -1.02572763e+00 -1.93061128e-01 -7.85727561e-01
9.19085860e-01 4.08304214e-01 1.06654286e+00 3.73745739e-01
1.72590151e-01 1.03442542e-01 -1.28167957e-01 3.68428409e-01
-2.97020495e-01 4.58842993e-01 -2.56850868e-01 5.43600202e-01
4.52356666e-01 -1.25954247e+00 -6.34024322e-01 2.77468115e-01
-4.54776794e-01 4.69591677e-01 6.91053808e-01 4.52395350e-01
4.28734452e-01 6.41389012e-01 5.21237135e-01 -9.38375413e-01
6.31804585e-01 -8.27630103e-01 -7.69659698e-01 3.53823721e-01
-8.67803752e-01 2.17482336e-02 8.55586469e-01 -3.47150415e-01
-8.20477068e-01 -1.70009151e-01 7.27391317e-02 -3.16303790e-01
-2.25307405e-01 1.02198482e+00 -2.61385739e-01 -1.85892195e-01
4.02990222e-01 4.74195838e-01 -2.11588204e-01 -1.85480312e-01
6.63312852e-01 1.34859875e-01 5.27957916e-01 -3.33428890e-01
9.71855938e-01 5.31472623e-01 3.22747886e-01 -7.47544289e-01
-5.27958453e-01 -3.15498680e-01 -8.50860596e-01 -6.91713214e-01
6.24605417e-01 -5.59426665e-01 -8.48137677e-01 5.73092997e-01
-1.03922677e+00 -6.91010177e-01 -1.09369546e-01 4.09215093e-01
-2.44459778e-01 4.45325643e-01 -4.45556164e-01 -6.32937372e-01
-5.37903607e-03 -9.68310177e-01 7.25913227e-01 1.16620682e-01
4.22182679e-01 -1.61567700e+00 1.20981922e-03 1.02682166e-01
6.54892564e-01 6.04479134e-01 1.16122019e+00 -2.13127702e-01
-1.17507887e+00 -2.47934520e-01 -6.17109597e-01 3.12374290e-02
2.44288966e-01 -6.63199974e-03 -4.01854128e-01 7.63349757e-02
-7.58852363e-01 4.63157862e-01 1.03787255e+00 5.69406331e-01
1.04387188e+00 -5.76271713e-01 -6.20577574e-01 7.74517775e-01
1.39212739e+00 3.76428925e-02 6.89470410e-01 2.05267861e-01
1.19780219e+00 6.78367257e-01 6.97948635e-02 -7.35841831e-03
1.20355582e+00 5.11136770e-01 5.36181986e-01 -1.17054813e-01
-3.60593677e-01 -5.72642922e-01 2.14225248e-01 1.22266388e+00
-3.92743051e-01 -4.96840358e-01 -1.16446829e+00 5.16492963e-01
-2.26225924e+00 -1.15374160e+00 -5.92591643e-01 2.01716948e+00
1.55520409e-01 5.68447888e-01 1.55820787e-01 -8.51729736e-02
8.82738888e-01 4.85322922e-01 -5.50849020e-01 -7.28128776e-02
-2.81124353e-01 -8.60077292e-02 8.40629816e-01 6.46592975e-01
-9.92662132e-01 1.18546391e+00 5.48038960e+00 8.48771334e-01
-1.45152080e+00 -6.22097962e-02 5.49361110e-01 3.23472232e-01
-4.85097349e-01 1.03066981e-01 -6.19659364e-01 7.56840289e-01
1.16333711e+00 -1.68170989e-01 8.53989482e-01 6.49901509e-01
4.47398186e-01 1.88299254e-01 -6.99111819e-01 8.14059377e-01
-2.07389966e-01 -1.82972491e+00 3.28804731e-01 2.61375636e-01
7.26621926e-01 5.20987451e-01 1.43848687e-01 4.73257005e-01
5.35862207e-01 -1.02119696e+00 4.37549829e-01 9.73203123e-01
6.46186233e-01 -6.40743315e-01 5.76447725e-01 4.58312452e-01
-1.86751485e+00 -1.18886597e-01 -1.49786279e-01 -9.49852243e-02
3.01058859e-01 6.52188599e-01 -7.29167163e-01 9.47956026e-01
3.98637265e-01 1.10176969e+00 -8.43664348e-01 1.03629887e+00
-3.36710244e-01 8.89141083e-01 -4.17314917e-01 1.97237544e-02
6.54441714e-01 -6.82357192e-01 4.57381785e-01 1.18641222e+00
2.44707435e-01 -1.98033467e-01 3.43704104e-01 8.67995799e-01
8.37290809e-02 -8.00065100e-02 -7.96757758e-01 -1.10784046e-01
3.97236377e-01 1.22104096e+00 -1.00921524e+00 -2.22595885e-01
-4.41976964e-01 4.44021672e-01 5.06261051e-01 8.53491604e-01
-9.90310133e-01 -2.38091454e-01 2.92191535e-01 4.58257496e-01
2.48057872e-01 -6.47062480e-01 -2.08738938e-01 -9.84527826e-01
-1.37029113e-02 -2.48167768e-01 4.69250649e-01 -5.12884796e-01
-1.38950801e+00 8.51085663e-01 -1.76676502e-03 -1.08778036e+00
-2.87429299e-02 -4.51920420e-01 -1.01688075e+00 8.71079743e-01
-1.75690532e+00 -1.73034632e+00 -5.73330104e-01 6.28625512e-01
5.58833405e-02 -2.53651023e-01 4.21671987e-01 6.31032109e-01
-7.29868531e-01 4.22982723e-01 2.53434051e-02 3.04364711e-01
-5.14115952e-02 -1.07516205e+00 1.01593792e+00 1.02842152e+00
6.38317987e-02 3.59157532e-01 3.30492169e-01 -9.83477414e-01
-1.25229204e+00 -1.69064379e+00 1.18142569e+00 -3.26622039e-01
1.13693225e+00 -4.40101683e-01 -8.51324856e-01 7.92420924e-01
-1.07386552e-01 2.87650675e-01 9.32891518e-02 8.44287053e-02
-1.72929212e-01 -5.40914059e-01 -5.74562132e-01 7.21792817e-01
1.50362432e+00 -5.03817320e-01 -1.87238138e-02 5.11898220e-01
7.13213503e-01 -3.28716397e-01 -5.43865144e-01 3.65455180e-01
3.29800338e-01 -8.84775221e-01 9.58923578e-01 -6.25351608e-01
1.19882193e-03 -5.30281425e-01 1.15613751e-01 -1.18420279e+00
-6.79733396e-01 -5.55171490e-01 -2.95518547e-01 1.04372573e+00
3.65208060e-01 -1.13827419e+00 1.08336627e+00 3.56960833e-01
-3.11579406e-01 -7.40724564e-01 -1.02331388e+00 -8.19358468e-01
-2.71949887e-01 -7.59631932e-01 1.09883118e+00 1.02651656e+00
-4.16565329e-01 4.15607095e-01 -1.89726397e-01 5.07881284e-01
6.77100360e-01 2.30013207e-01 8.01731408e-01 -1.32370400e+00
5.15882336e-02 -8.21637452e-01 -8.46144259e-01 -1.16709864e+00
3.47639829e-01 -1.24874747e+00 -4.51024622e-01 -1.93902719e+00
-4.81085449e-01 -8.38479936e-01 -4.88419116e-01 3.98964465e-01
-7.98616782e-02 -1.42215751e-02 -2.40543634e-01 -7.12796953e-03
-7.49760687e-01 6.26014471e-01 1.23051417e+00 -4.06961441e-01
-3.32583815e-01 1.47740826e-01 -3.45001698e-01 3.79422128e-01
8.46676648e-01 -1.88073233e-01 -8.18698108e-01 -4.15878415e-01
4.69586343e-01 -1.09963179e-01 5.99086225e-01 -1.15134370e+00
7.71472633e-01 -4.40243185e-01 -1.16813473e-01 -9.71047580e-01
-4.87584211e-02 -7.75851190e-01 4.38976288e-01 1.80308551e-01
5.58852367e-02 1.63131610e-01 -2.79977042e-02 8.59383106e-01
-1.70564860e-01 5.85809112e-01 2.36867994e-01 6.34656399e-02
-9.78952348e-01 9.73507881e-01 -2.90349185e-01 -8.12014565e-02
9.49046016e-01 -6.25207305e-01 -5.46379864e-01 -3.44212055e-01
-7.04908490e-01 5.37905097e-01 7.30158836e-02 5.50551116e-01
5.21333575e-01 -1.59814358e+00 -4.73838955e-01 3.14463675e-01
5.44459075e-02 -7.87471831e-02 5.40470779e-01 1.00381517e+00
-5.45663357e-01 6.11738443e-01 -1.33044481e-01 -5.80285549e-01
-5.54925442e-01 7.27892935e-01 3.97967905e-01 -5.55808008e-01
-9.64037478e-01 3.90956908e-01 5.29294051e-02 -7.17144072e-01
1.47317931e-01 -5.89679718e-01 -4.45654631e-01 -7.25249574e-02
-2.65671965e-02 4.86463755e-01 9.71769318e-02 -9.92767990e-01
-2.31545404e-01 4.15059507e-01 4.22490537e-01 2.99563229e-01
1.43075764e+00 -2.86648929e-01 -2.40023315e-01 4.19971108e-01
9.41780984e-01 -2.58077234e-01 -1.13607216e+00 -3.76042664e-01
1.61539599e-01 -2.13095680e-01 7.22968057e-02 -5.24490118e-01
-1.47689295e+00 8.10892224e-01 2.73178220e-01 5.60582995e-01
9.22808230e-01 -3.45665842e-01 1.09198034e+00 3.46910477e-01
2.67932802e-01 -8.75121891e-01 -1.26926988e-01 7.05534637e-01
5.93299627e-01 -9.02840793e-01 -2.09926829e-01 -5.24519980e-01
-1.72292486e-01 1.25123799e+00 7.07911074e-01 -2.82240689e-01
1.11480570e+00 -2.82510817e-01 -2.59444267e-01 -4.40069348e-01
-6.77472115e-01 -4.78719980e-01 5.73361754e-01 6.51407003e-01
-8.58174413e-02 3.10920775e-01 1.24402121e-01 2.12255448e-01
-2.86404878e-01 -6.71663061e-02 -2.94512627e-03 3.42344522e-01
-3.15173328e-01 -1.07705581e+00 1.74595147e-01 6.39462829e-01
3.96350503e-01 -7.85437301e-02 1.70882177e-02 6.89456046e-01
3.59969348e-01 8.31313312e-01 2.22277552e-01 -8.32656503e-01
3.11181605e-01 -2.44132429e-01 8.68848413e-02 -2.67061323e-01
-2.39693701e-01 -4.41711426e-01 1.86464027e-01 -6.68449521e-01
-2.24656641e-01 -3.80733877e-01 -1.20397186e+00 -7.79007971e-01
-1.11932606e-01 1.66892841e-01 4.49978262e-01 1.04101467e+00
6.12857699e-01 7.11940944e-01 6.45164847e-01 -8.66803229e-01
2.84709960e-01 -7.64723957e-01 -3.94722879e-01 3.63083512e-01
1.64427415e-01 -8.04104269e-01 -1.90089196e-01 -1.59479767e-01] | [6.47738790512085, 2.0761375427246094] |
125eb805-55b9-4c48-929e-82d86f149771 | cross-modal-and-hierarchical-modeling-of | 1810.07212 | null | http://arxiv.org/abs/1810.07212v1 | http://arxiv.org/pdf/1810.07212v1.pdf | Cross-Modal and Hierarchical Modeling of Video and Text | Visual data and text data are composed of information at multiple
granularities. A video can describe a complex scene that is composed of
multiple clips or shots, where each depicts a semantically coherent event or
action. Similarly, a paragraph may contain sentences with different topics,
which collectively conveys a coherent message or story. In this paper, we
investigate the modeling techniques for such hierarchical sequential data where
there are correspondences across multiple modalities. Specifically, we
introduce hierarchical sequence embedding (HSE), a generic model for embedding
sequential data of different modalities into hierarchically semantic spaces,
with either explicit or implicit correspondence information. We perform
empirical studies on large-scale video and paragraph retrieval datasets and
demonstrated superior performance by the proposed methods. Furthermore, we
examine the effectiveness of our learned embeddings when applied to downstream
tasks. We show its utility in zero-shot action recognition and video
captioning. | ['Bowen Zhang', 'Hexiang Hu', 'Fei Sha'] | 2018-10-16 | cross-modal-and-hierarchical-modeling-of-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Bowen_Zhang_Cross-Modal_and_Hierarchical_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Bowen_Zhang_Cross-Modal_and_Hierarchical_ECCV_2018_paper.pdf | eccv-2018-9 | ['zero-shot-action-recognition'] | ['computer-vision'] | [ 3.31555814e-01 -2.68098474e-01 -4.46013063e-01 -2.92802304e-01
-8.62650454e-01 -4.52541620e-01 1.01345789e+00 3.47971141e-01
-6.25979751e-02 5.28770626e-01 1.18575680e+00 2.70291805e-01
1.30609378e-01 -4.04517084e-01 -7.34267831e-01 -5.56086779e-01
-4.64883111e-02 -3.20869498e-02 3.06355506e-01 1.45230547e-01
2.84757227e-01 1.10760778e-01 -1.65675044e+00 8.30882132e-01
1.72507782e-02 1.07552838e+00 3.64588350e-01 7.95058250e-01
-3.70382190e-01 1.23577642e+00 -5.73616087e-01 -3.72364551e-01
-1.82913542e-01 -5.73123991e-01 -7.23714650e-01 4.76686865e-01
6.10417902e-01 -5.61690867e-01 -7.87496805e-01 7.78413177e-01
2.30047420e-01 1.27784774e-01 7.28807449e-01 -1.54850590e+00
-9.26301062e-01 4.03606683e-01 -6.30570590e-01 2.33808771e-01
7.95414329e-01 1.19970748e-02 1.30624533e+00 -8.80128324e-01
7.94581473e-01 1.36965561e+00 3.24962318e-01 4.72972095e-01
-1.08933604e+00 -2.68297702e-01 3.15332294e-01 5.88452160e-01
-1.20535564e+00 -6.00914121e-01 5.97416222e-01 -7.75357306e-01
8.54909718e-01 1.94963843e-01 6.70123339e-01 1.42996216e+00
5.24361245e-02 1.17740762e+00 4.26422626e-01 -1.31568491e-01
1.01344474e-01 -1.31107355e-02 2.43431315e-01 5.31397760e-01
1.55901611e-01 -4.08804387e-01 -1.03696930e+00 -1.87933668e-01
7.58773744e-01 5.60322821e-01 -3.74426007e-01 -4.37171847e-01
-1.61817694e+00 8.38085532e-01 1.34758651e-01 4.23368245e-01
-3.13724607e-01 3.67509097e-01 8.17838609e-01 1.77891001e-01
3.00017297e-01 1.94109008e-01 5.94127215e-02 -2.88665473e-01
-9.03503001e-01 9.11312625e-02 6.06229842e-01 1.22240520e+00
4.87369806e-01 -1.30904824e-01 -6.06030822e-01 8.28346968e-01
2.89397120e-01 2.72801280e-01 5.56135893e-01 -1.14906883e+00
7.61449873e-01 5.16326845e-01 3.34505558e-01 -1.33211434e+00
2.67863944e-02 3.92590731e-01 -8.11168671e-01 -4.86616969e-01
-1.40074249e-02 3.08632195e-01 -8.58688951e-01 1.77921748e+00
7.25193992e-02 5.57318628e-01 1.94229707e-01 1.10179794e+00
1.07554662e+00 1.11983728e+00 2.76120752e-01 -2.21105114e-01
1.62812197e+00 -1.08883715e+00 -1.03716588e+00 4.91637625e-02
3.42696309e-01 -4.50191468e-01 9.13970053e-01 -1.19351648e-01
-1.22278965e+00 -5.84704340e-01 -9.15515304e-01 -4.69620019e-01
-3.23691964e-01 -4.63117845e-02 2.05269992e-01 -6.66200817e-02
-9.55792367e-01 2.73665637e-01 -7.35621452e-01 -6.74139559e-01
2.72492051e-01 -3.13538074e-01 -6.73891902e-01 -2.21841350e-01
-1.26478839e+00 5.33765614e-01 3.57960671e-01 -2.23085627e-01
-9.90083516e-01 -5.84006250e-01 -1.17629540e+00 2.65762448e-01
3.49055320e-01 -8.60408783e-01 1.16990507e+00 -7.83041894e-01
-1.11843228e+00 8.02830637e-01 -6.00748301e-01 -4.44544971e-01
1.56469122e-01 -2.28822410e-01 -4.60887998e-01 7.95095086e-01
2.48389900e-01 8.03022742e-01 1.03785992e+00 -1.06466532e+00
-6.39587164e-01 -3.50186706e-01 4.69855130e-01 3.70974451e-01
-5.83468139e-01 1.95206508e-01 -7.51726210e-01 -8.02893639e-01
-2.75671463e-02 -5.34037650e-01 1.88513711e-01 1.98244944e-01
-3.07180762e-01 -2.10866034e-01 8.03386152e-01 -7.19132602e-01
1.48474765e+00 -2.46702433e+00 6.27957284e-01 -4.03092861e-01
1.91582963e-01 -2.46275172e-01 -2.09005535e-01 1.01608479e+00
1.98988304e-01 3.28542650e-01 -1.53100505e-01 -3.90958101e-01
3.06619287e-01 2.44928822e-01 -6.28525913e-01 2.44052842e-01
1.72609314e-01 9.82836485e-01 -8.66333365e-01 -7.71420240e-01
8.28375518e-02 4.70186621e-01 -2.65267968e-01 4.18658167e-01
-2.58385658e-01 1.34876698e-01 -6.69955015e-01 6.46795869e-01
5.54341674e-02 -6.84192240e-01 1.70315847e-01 -5.58855116e-01
-7.31072761e-03 -6.26283437e-02 -8.88199747e-01 2.05592799e+00
-1.98641554e-01 1.03311944e+00 -1.96210623e-01 -1.04609787e+00
3.76140475e-01 7.09541440e-01 6.01609826e-01 -5.33009470e-01
-1.31232455e-01 -3.91431898e-01 -7.23203897e-01 -8.61140788e-01
8.60904336e-01 -6.90045133e-02 -4.33139622e-01 5.01988053e-01
7.09310174e-02 2.85953641e-01 1.65922374e-01 5.71452320e-01
1.20374537e+00 -4.13279906e-02 4.87314999e-01 2.27941886e-01
3.12949181e-01 8.38114973e-03 3.61789972e-01 8.85752916e-01
-3.63584608e-01 8.10664773e-01 6.86397791e-01 -1.45008892e-01
-1.28428507e+00 -1.13038790e+00 3.50087993e-02 1.21901405e+00
5.30603647e-01 -8.11506510e-01 -2.26933718e-01 -3.44855011e-01
5.75363189e-02 3.53362203e-01 -7.06012666e-01 -2.20298823e-02
-2.46898755e-01 -9.80964229e-02 3.61759156e-01 7.57083476e-01
2.80644625e-01 -9.94203985e-01 -6.17697060e-01 1.62093580e-01
-5.44433534e-01 -1.37304795e+00 -8.02237451e-01 -4.98053551e-01
-6.47847533e-01 -9.82503474e-01 -8.78565371e-01 -8.16500604e-01
3.22462022e-01 7.45135367e-01 1.05640495e+00 -9.66164991e-02
-2.93414026e-01 1.04257929e+00 -7.87824571e-01 1.57478675e-01
-9.25869942e-02 -5.65232337e-01 5.61166778e-02 3.11583519e-01
3.63666445e-01 -2.92696655e-01 -6.10926211e-01 1.87140599e-01
-1.31535459e+00 3.74964923e-01 2.69225538e-01 7.69330561e-01
5.50480604e-01 -4.64098662e-01 3.63720030e-01 -4.75329131e-01
5.28983355e-01 -8.57035279e-01 6.94879442e-02 5.62211275e-01
2.15624154e-01 -1.92429703e-02 3.65473926e-01 -4.17683333e-01
-9.22916234e-01 -2.91226149e-01 3.67449552e-01 -8.79245937e-01
-3.28417480e-01 6.15605175e-01 -1.51772186e-01 5.56541681e-01
3.18570584e-02 5.17945111e-01 -1.02051400e-01 -4.03353333e-01
7.12875128e-01 7.21520007e-01 4.01648015e-01 -4.79948699e-01
4.26145196e-01 7.29034960e-01 -2.72335589e-01 -1.09803808e+00
-8.16330314e-01 -7.98814774e-01 -5.90120852e-01 -3.73378366e-01
1.27938068e+00 -1.13799965e+00 -5.08345068e-01 2.17774734e-01
-1.30392349e+00 1.16897061e-01 -9.91472453e-02 4.61321235e-01
-7.26461649e-01 6.42360747e-01 -9.21824634e-01 -4.54578400e-01
7.49132559e-02 -9.33062613e-01 1.45464981e+00 8.42692852e-02
-3.75187367e-01 -9.57598686e-01 2.25216635e-02 4.49303716e-01
-4.66622598e-02 3.34219337e-01 6.84477866e-01 -5.33230364e-01
-6.42811894e-01 -2.26703356e-03 -3.24196428e-01 1.05188772e-01
1.70225278e-01 -9.33048967e-03 -6.25146329e-01 -3.60372573e-01
-2.39315376e-01 -6.97272897e-01 1.02417338e+00 2.19805494e-01
1.29950655e+00 -5.44555724e-01 -3.41281503e-01 1.94741383e-01
1.46574819e+00 1.59040987e-01 5.85854590e-01 3.19243312e-01
8.22054267e-01 6.53623223e-01 6.06885910e-01 8.12424421e-01
3.51101846e-01 7.12556422e-01 1.08760670e-01 1.56930581e-01
-4.00389694e-02 -5.21563590e-01 5.45666218e-01 1.02763700e+00
1.94346473e-01 -4.60238725e-01 -6.00348890e-01 7.93661296e-01
-2.22386146e+00 -1.52705801e+00 1.72433197e-01 1.75374365e+00
6.63606107e-01 -2.93677181e-01 1.29933849e-01 -2.27236971e-01
9.02451694e-01 7.89429903e-01 -5.33541858e-01 -1.59124509e-01
-1.57317981e-01 -4.33335572e-01 1.18848987e-01 1.70114428e-01
-1.07872820e+00 7.13018775e-01 6.39440107e+00 7.07966626e-01
-7.14532256e-01 2.13205159e-01 2.21474648e-01 -2.40630671e-01
-4.12188172e-01 -1.81970134e-01 -5.30041993e-01 6.57533467e-01
8.48018825e-01 -5.42377412e-01 6.15776032e-02 5.32045960e-01
2.96394795e-01 -7.83544034e-02 -1.39934039e+00 1.21528900e+00
5.43100238e-01 -1.76976657e+00 6.19584799e-01 -2.83893943e-01
7.64826536e-01 -3.64303559e-01 1.70891024e-02 1.98436841e-01
-2.80477721e-02 -8.43565881e-01 7.71200180e-01 7.25176156e-01
6.94833517e-01 -2.31833071e-01 3.18051487e-01 1.41878799e-01
-1.47646070e+00 -8.36606920e-02 -3.53625476e-01 6.57975748e-02
5.27739584e-01 -5.44950217e-02 -2.54029781e-01 6.52720392e-01
6.45713091e-01 1.45423484e+00 -2.82201409e-01 9.24902320e-01
6.78661391e-02 3.18988621e-01 2.09231466e-01 -8.90710279e-02
4.42376226e-01 -1.41494691e-01 4.98706132e-01 1.42856264e+00
4.06756043e-01 4.69475746e-01 2.17992172e-01 5.33651769e-01
-1.93455398e-01 -1.77985001e-02 -9.19556379e-01 -4.93493259e-01
5.93135595e-01 8.71042967e-01 -3.92184347e-01 -7.56558478e-01
-9.20560062e-01 1.26110399e+00 2.31974214e-01 6.58323824e-01
-9.86808181e-01 -1.04655832e-01 1.00074577e+00 -1.14865929e-01
5.11864483e-01 -2.67765462e-01 2.35604331e-01 -1.60774696e+00
1.50247365e-01 -5.88232458e-01 6.52173936e-01 -1.17968166e+00
-1.46132326e+00 2.49903381e-01 2.09585384e-01 -1.57162023e+00
-1.53955996e-01 -4.16593105e-01 -4.40307260e-01 3.65511715e-01
-1.22149837e+00 -9.90199149e-01 -2.53612399e-01 7.30506897e-01
9.98346508e-01 -1.96214676e-01 6.88947916e-01 1.73946872e-01
-3.95813286e-01 1.19604357e-01 4.04533744e-01 1.75945342e-01
6.20769858e-01 -9.81322527e-01 -5.33763655e-02 6.26445174e-01
3.60671490e-01 4.48369145e-01 6.63280070e-01 -4.94602114e-01
-1.66631746e+00 -1.07256889e+00 9.34472978e-01 -3.92106742e-01
1.08199418e+00 -2.91407317e-01 -1.00639331e+00 8.82725298e-01
4.35889333e-01 -2.90600806e-01 9.94268239e-01 -1.24296553e-01
-5.48013031e-01 1.60578504e-01 -7.35205889e-01 6.85811937e-01
1.09050858e+00 -1.03571296e+00 -1.11457407e+00 2.79846787e-01
1.14212787e+00 -4.59635518e-02 -9.76188660e-01 1.85623825e-01
6.66601002e-01 -8.82166445e-01 1.00976443e+00 -1.16432953e+00
9.99190271e-01 -1.97673783e-01 -6.12467110e-01 -1.01594770e+00
-4.27607566e-01 -1.52811408e-01 -4.33214188e-01 1.11228704e+00
2.14305874e-02 -1.77324824e-02 4.79304314e-01 5.37550509e-01
-9.18880999e-02 -3.26928735e-01 -8.78076017e-01 -6.99353278e-01
-4.25861448e-01 -4.04524982e-01 4.07281697e-01 1.03988826e+00
3.98487598e-01 2.73617625e-01 -7.73416162e-01 1.40571073e-01
4.99148846e-01 5.10406733e-01 5.92907667e-01 -7.45408833e-01
-1.76083684e-01 -2.80652702e-01 -8.16352606e-01 -1.36105454e+00
2.85833150e-01 -7.61149347e-01 6.65221438e-02 -1.76989019e+00
8.31153870e-01 4.64875400e-01 -3.97374511e-01 3.09927106e-01
-6.68333247e-02 1.56125128e-01 4.05134380e-01 5.93531966e-01
-1.18019986e+00 7.30334520e-01 9.90255296e-01 -4.42009807e-01
2.53845364e-01 -6.45515800e-01 -3.74542058e-01 4.69960570e-01
4.94294673e-01 -1.41068935e-01 -5.76388597e-01 -6.84505165e-01
3.89886014e-02 6.01888120e-01 5.15043974e-01 -6.95554376e-01
2.21112162e-01 -2.73638278e-01 1.99508414e-01 -5.69328487e-01
6.39137328e-01 -8.53409171e-01 1.86041832e-01 6.01187497e-02
-8.59192610e-01 -5.05291075e-02 8.37997273e-02 1.11465633e+00
-7.83143997e-01 7.33339861e-02 2.74485797e-01 -2.08944958e-02
-1.20989501e+00 5.73389053e-01 -4.52121466e-01 1.65229198e-02
1.19649768e+00 -3.20903301e-01 -3.16921175e-01 -6.45995855e-01
-1.05395317e+00 3.04339141e-01 5.36787927e-01 8.85333478e-01
9.99917269e-01 -1.85655367e+00 -5.52013814e-01 -1.17136642e-01
6.11054063e-01 -6.33289218e-01 4.58460301e-01 6.39398992e-01
-2.02156886e-01 6.58755839e-01 -2.95005828e-01 -6.37269258e-01
-1.47169673e+00 6.43886983e-01 -2.78076082e-01 1.79225460e-01
-7.99073637e-01 5.85595489e-01 5.74961424e-01 2.74299055e-01
5.18288553e-01 -1.94107994e-01 -4.96735543e-01 3.87133718e-01
8.60688627e-01 2.61632323e-01 -7.72107363e-01 -9.88915503e-01
-3.18849474e-01 5.46264529e-01 8.76089260e-02 -2.09353969e-01
1.20329404e+00 -5.40399611e-01 -5.46414964e-02 1.09925151e+00
1.65393794e+00 -6.13028467e-01 -1.29425740e+00 -3.78790587e-01
1.18269034e-01 -6.92835093e-01 -1.86159134e-01 -2.23335132e-01
-6.65504158e-01 8.96210432e-01 1.50471568e-01 3.43741238e-01
9.00693655e-01 5.09292603e-01 8.44139338e-01 2.25704715e-01
2.94547826e-01 -9.27322149e-01 5.42622507e-01 3.05617332e-01
1.02209926e+00 -1.28923881e+00 -2.06649527e-01 -4.67808992e-02
-1.09009981e+00 1.19379282e+00 3.81185651e-01 1.93324432e-01
4.57695365e-01 -2.36762047e-01 -3.90196443e-01 -3.29023570e-01
-1.15427792e+00 -1.97274670e-01 3.34234625e-01 1.72944725e-01
3.51394206e-01 2.39947978e-02 -2.60258973e-01 5.94143152e-01
4.89923477e-01 -1.51072033e-02 4.70529199e-01 1.08738554e+00
-6.39504194e-01 -7.24249244e-01 -2.07153603e-01 3.10755134e-01
-2.93532193e-01 -3.19445692e-02 -4.57138717e-01 6.10330045e-01
-2.95367509e-01 8.91890764e-01 4.21291173e-01 -3.72899145e-01
1.26542911e-01 2.30747789e-01 3.78015399e-01 -6.76676214e-01
4.65212800e-02 -3.97496251e-03 2.25513950e-02 -8.54398489e-01
-8.41648638e-01 -8.41579437e-01 -1.09771872e+00 -1.13853216e-01
4.61124331e-01 1.63191900e-01 3.76030654e-01 9.08491075e-01
4.51884776e-01 3.51831824e-01 6.25821173e-01 -7.78803408e-01
-1.30019650e-01 -8.06804836e-01 -7.84979224e-01 8.67327273e-01
4.70219254e-01 -5.78746080e-01 -3.74233276e-01 7.28600740e-01] | [10.230262756347656, 0.8733316659927368] |
fedff40f-c1c8-4dcd-94fd-89963d65fa56 | morphological-classification-of-extragalactic | 2304.12729 | null | https://arxiv.org/abs/2304.12729v1 | https://arxiv.org/pdf/2304.12729v1.pdf | Morphological Classification of Extragalactic Radio Sources Using Gradient Boosting Methods | The field of radio astronomy is witnessing a boom in the amount of data produced per day due to newly commissioned radio telescopes. One of the most crucial problems in this field is the automatic classification of extragalactic radio sources based on their morphologies. Most recent contributions in the field of morphological classification of extragalactic radio sources have proposed classifiers based on convolutional neural networks. Alternatively, this work proposes gradient boosting machine learning methods accompanied by principal component analysis as data-efficient alternatives to convolutional neural networks. Recent findings have shown the efficacy of gradient boosting methods in outperforming deep learning methods for classification problems with tabular data. The gradient boosting methods considered in this work are based on the XGBoost, LightGBM, and CatBoost implementations. This work also studies the effect of dataset size on classifier performance. A three-class classification problem is considered in this work based on the three main Fanaroff-Riley classes: class 0, class I, and class II, using radio sources from the Best-Heckman sample. All three proposed gradient boosting methods outperformed a state-of-the-art convolutional neural networks-based classifier using less than a quarter of the number of images, with CatBoost having the highest accuracy. This was mainly due to the superior accuracy of gradient boosting methods in classifying Fanaroff-Riley class II sources, with 3--4\% higher recall. | ['Abir Hussain', 'Marley Vellasco', 'Ilias Fernini', 'Abdollah Masoud Darya'] | 2023-04-25 | null | null | null | null | ['astronomy'] | ['miscellaneous'] | [-3.56441170e-01 -4.00338620e-01 -1.83076069e-01 -3.28585744e-01
-3.34965080e-01 -4.81469333e-01 1.03768229e+00 -2.40117833e-02
-6.20292544e-01 7.20433772e-01 -4.16729972e-02 -7.58869886e-01
-5.24669051e-01 -9.04533565e-01 -3.86493653e-01 -8.43397081e-01
-2.52117720e-02 5.26616275e-01 3.56423318e-01 -4.15314823e-01
5.14665604e-01 9.36081827e-01 -1.52557921e+00 2.86467552e-01
2.21886799e-01 1.50137722e+00 2.47686028e-01 7.75029480e-01
-1.93251386e-01 9.03376877e-01 -5.76903164e-01 -4.12278503e-01
5.15400887e-01 -4.67666566e-01 -5.41015387e-01 -1.25323489e-01
3.69487256e-01 8.71782973e-02 -1.57770246e-01 5.40467739e-01
6.10586524e-01 -6.59184111e-03 8.90679419e-01 -1.04561496e+00
-3.85851592e-01 4.63219248e-02 -4.30995435e-01 7.79728174e-01
-1.16856880e-01 -3.37879241e-01 8.78155410e-01 -8.90572965e-01
3.53991777e-01 1.03626204e+00 8.96385610e-01 -5.40213697e-02
-7.45439589e-01 -7.14689612e-01 -4.02465075e-01 5.53395212e-01
-9.20219183e-01 -1.30599692e-01 9.22314882e-01 -7.07159281e-01
1.14503872e+00 3.34432513e-01 6.98524237e-01 4.68812943e-01
3.66830051e-01 2.00983629e-01 1.65440810e+00 -7.70048797e-01
6.01470292e-01 3.14450920e-01 2.21526727e-01 5.89237332e-01
5.10842144e-01 3.04597169e-01 -5.05846083e-01 -4.12661731e-01
5.70758522e-01 -2.16184407e-01 1.20921463e-01 -1.97695687e-01
-7.14984953e-01 1.36273062e+00 7.18044996e-01 7.11687565e-01
-3.44998628e-01 -3.64751108e-02 2.81715304e-01 3.51757020e-01
8.76925349e-01 6.16950750e-01 -7.81336904e-01 2.76040137e-01
-6.79112017e-01 3.11058104e-01 4.79860008e-01 4.09276098e-01
6.25503778e-01 2.59418994e-01 4.74695623e-01 1.00090766e+00
3.90561998e-01 1.86457992e-01 8.75288069e-01 -2.84894288e-01
1.96408153e-01 6.41118169e-01 -4.66342457e-02 -9.28887904e-01
-7.66779661e-01 -1.03474426e+00 -4.92272407e-01 7.30019927e-01
5.00375628e-01 5.09373620e-02 -9.88255918e-01 9.27796662e-01
3.93184870e-01 -4.34358597e-01 1.18810542e-01 6.82135403e-01
9.94626582e-01 3.98120195e-01 -1.75713256e-01 -1.39408074e-02
1.55024219e+00 -8.11863780e-01 -3.81193645e-02 -3.35624337e-01
5.19016922e-01 -9.36974704e-01 3.88855934e-01 5.77727795e-01
-6.82353437e-01 -7.22808599e-01 -1.39986515e+00 2.40203008e-01
-7.18568623e-01 3.85384172e-01 1.25485730e+00 1.26141334e+00
-8.49515915e-01 4.23539221e-01 -4.89082694e-01 -3.78774732e-01
5.06253779e-01 2.88454533e-01 -3.53993446e-01 2.95821011e-01
-5.92048764e-01 8.61571431e-01 5.07407129e-01 -1.34344280e-01
-4.37039226e-01 -7.07184076e-01 -2.29980305e-01 -3.43056321e-02
-1.44501686e-01 -3.34787488e-01 1.26884329e+00 -1.12055612e+00
-1.30485082e+00 1.00687909e+00 3.37670714e-01 -5.29753149e-01
3.00657213e-01 3.52395564e-01 -4.67899770e-01 1.53264135e-01
-1.57568639e-03 3.36400926e-01 6.39477968e-01 -8.90838861e-01
-1.14432466e+00 -5.53949952e-01 5.34841828e-02 -1.91952184e-01
-1.59728140e-01 4.86733586e-01 1.49441287e-01 -6.72995865e-01
5.05060136e-01 -9.90392148e-01 1.20466702e-01 -4.40712482e-01
3.85696143e-01 -5.76760352e-01 8.34069431e-01 -7.05618620e-01
3.96903157e-01 -1.85642600e+00 -3.58502209e-01 -3.97583507e-02
2.29426231e-02 3.15304935e-01 4.72251117e-01 2.50821337e-02
-3.59968156e-01 -7.70452246e-02 1.33332208e-01 6.65118396e-02
-2.92533964e-01 -4.57627699e-02 -2.12732926e-01 5.13284683e-01
-8.86478722e-02 5.09669185e-01 -5.41786969e-01 -4.03147377e-02
2.70104289e-01 2.92902976e-01 -1.13591477e-01 9.74550918e-02
1.28993064e-01 2.71822453e-01 -1.23010002e-01 7.14090407e-01
7.36655653e-01 4.21632230e-02 -5.41097522e-02 -4.30590590e-05
-3.85944605e-01 2.49154493e-01 -7.83815086e-01 1.15360665e+00
-6.64587677e-01 8.69577587e-01 -2.43307456e-01 -1.28854406e+00
1.46392238e+00 3.17293108e-01 2.74434328e-01 -8.96714509e-01
2.50695795e-01 6.06767178e-01 3.00332785e-01 -1.01863377e-01
1.94867119e-01 -4.42682564e-01 3.44436198e-01 1.20997787e-01
4.72803086e-01 -1.19249828e-01 2.77599752e-01 -2.08716065e-01
8.00033808e-01 1.09474264e-01 5.04735887e-01 -4.73487228e-01
3.82576406e-01 2.17756137e-01 4.90427107e-01 8.67946088e-01
2.66893357e-02 5.92283785e-01 4.68683064e-01 -1.09884226e+00
-9.79211807e-01 -5.53107798e-01 -5.79560459e-01 1.23865807e+00
-5.73724687e-01 1.42920524e-01 -3.19765508e-01 -6.93864584e-01
-1.45855501e-01 4.86170858e-01 -7.36068785e-01 2.43370503e-01
-3.48018676e-01 -1.69800055e+00 4.45227116e-01 4.02691931e-01
8.47402453e-01 -1.15626574e+00 -9.63809133e-01 6.34456202e-02
1.99985266e-01 -7.52421856e-01 8.64296377e-01 7.04527318e-01
-1.12056363e+00 -1.12021518e+00 -6.59359574e-01 -6.47935629e-01
2.71686584e-01 3.89144540e-01 1.19847524e+00 1.10293105e-01
-4.83819932e-01 -1.09793842e-01 -7.00142384e-01 -1.05767691e+00
-2.99461871e-01 1.92034632e-01 -3.21974933e-01 -1.35660199e-02
4.34865445e-01 -3.37267131e-01 -4.59308684e-01 2.55265653e-01
-5.81789613e-01 -2.46610418e-01 7.79619217e-01 7.47642756e-01
4.16795835e-02 2.57099956e-01 7.67839611e-01 -6.21357679e-01
5.46040274e-02 -7.57026255e-01 -8.99187446e-01 -8.56391862e-02
-7.69529045e-01 -1.12754926e-01 4.60336715e-01 -5.19504212e-02
-1.36444867e+00 5.43719083e-02 -2.56406248e-01 2.40444571e-01
-3.45337778e-01 4.20673639e-01 1.95826456e-01 -6.00083172e-01
1.12458408e+00 -2.20201492e-01 -3.11094433e-01 -5.57986259e-01
-1.23351172e-01 8.31382632e-01 3.19037199e-01 -2.32351407e-01
5.43849826e-01 7.29316115e-01 3.29164952e-01 -1.01266479e+00
-9.45350468e-01 -8.65388691e-01 -5.57559073e-01 -4.51186240e-01
9.89645481e-01 -7.16876268e-01 -2.00731099e-01 6.99744344e-01
-8.53086054e-01 4.44562919e-02 2.29874298e-01 8.56387377e-01
-3.80111039e-01 -6.93394169e-02 -2.45406151e-01 -8.96289349e-01
-2.49862403e-01 -8.89043510e-01 6.58822358e-01 2.74948031e-01
3.61257672e-01 -9.81004059e-01 1.86723635e-01 5.50430655e-01
5.58140576e-01 2.51847416e-01 9.38635647e-01 -8.35271418e-01
-1.08265586e-01 -4.85622704e-01 -2.59119123e-01 2.19915643e-01
-1.05315469e-01 -3.27884525e-01 -1.24386382e+00 -2.12676540e-01
4.19351727e-01 -3.11598122e-01 1.03789937e+00 4.72767621e-01
7.85532236e-01 2.52371013e-01 -7.07899854e-02 6.33761287e-01
1.59950256e+00 7.84140229e-01 5.41467726e-01 1.02846050e+00
2.27116391e-01 5.75732946e-01 5.03248930e-01 1.67891577e-01
-2.67840087e-01 6.75767958e-01 5.69717348e-01 -1.31691501e-01
-1.43962726e-01 3.06720257e-01 -1.87078953e-01 4.08137351e-01
-4.32723045e-01 1.55183047e-01 -1.07231557e+00 3.58564645e-01
-1.58506835e+00 -8.89252245e-01 -5.89588702e-01 2.17136025e+00
1.15910836e-01 3.87536556e-01 2.50446111e-01 5.34631670e-01
4.31307167e-01 -6.87677488e-02 8.62809792e-02 -4.87252116e-01
-6.68217391e-02 5.03381968e-01 7.87618041e-01 1.42992102e-02
-1.31024742e+00 3.09269965e-01 6.13113737e+00 6.56636655e-01
-1.41925323e+00 3.25185537e-01 9.59254444e-01 8.11877847e-02
5.78052223e-01 1.13453843e-01 -9.41078305e-01 4.42768842e-01
9.98139501e-01 4.63795930e-01 1.41813934e-01 1.19382274e+00
-2.41955876e-01 -4.52889532e-01 -3.53326201e-01 8.68680298e-01
2.52019614e-01 -1.51371884e+00 -6.10540807e-01 1.76927730e-01
6.98018670e-01 4.89040792e-01 -2.83295512e-01 4.83355731e-01
1.13379046e-01 -6.69351161e-01 7.50307620e-01 1.73522770e-01
2.81011134e-01 -9.04778481e-01 1.02846897e+00 2.67332494e-01
-8.91743541e-01 -4.94114459e-01 -9.14248049e-01 -5.05232692e-01
-4.77861345e-01 6.26066744e-01 -1.07564807e+00 7.52413273e-01
1.26859558e+00 -4.35274886e-03 -9.98742819e-01 1.34768450e+00
-6.26654401e-02 9.31166589e-01 -3.13584596e-01 -3.15217704e-01
3.63663882e-01 -1.69340983e-01 1.16981946e-01 1.23584199e+00
4.82298732e-01 -1.40298560e-01 -1.22361742e-01 2.01271489e-01
2.51797825e-01 2.37134233e-01 -4.13713843e-01 3.55987132e-01
1.06460169e-01 1.69323301e+00 -1.44945419e+00 -3.70036542e-01
-7.83349276e-01 1.58615306e-01 1.45555973e-01 -1.43416554e-01
-4.03618366e-01 -6.29498184e-01 3.32578458e-02 1.54991180e-01
5.64713597e-01 -5.54224327e-02 -4.66522634e-01 -5.59807122e-01
-1.05836570e-01 -5.97471476e-01 6.74825728e-01 -9.22324061e-01
-1.11791253e+00 6.72368407e-01 1.48485959e-01 -9.01032031e-01
-2.14093670e-01 -1.33396053e+00 -7.02529967e-01 8.69017065e-01
-1.33487439e+00 -1.16377771e+00 -5.16392171e-01 9.56031680e-02
4.44356143e-01 -1.04914033e+00 5.93016922e-01 1.36308298e-01
-1.13359362e-01 -1.22165643e-02 4.86533672e-01 1.07088946e-01
4.39726710e-01 -1.39879453e+00 1.74891248e-01 6.47113204e-01
2.58426279e-01 6.37531579e-02 5.09445786e-01 -4.43978608e-01
-9.62961733e-01 -8.76020610e-01 8.67145598e-01 -1.12115867e-01
6.74101174e-01 -2.73724079e-01 -5.96421063e-01 3.32502395e-01
1.75156862e-01 2.55553216e-01 5.74702084e-01 2.23103225e-01
-4.63068932e-01 -3.45769703e-01 -1.23609650e+00 -1.35790959e-01
3.92350197e-01 -2.04862028e-01 -6.63050115e-01 5.52705467e-01
-7.82801360e-02 3.79040353e-02 -5.71667671e-01 4.56006587e-01
4.77963001e-01 -1.34373534e+00 8.77178371e-01 -5.37111223e-01
4.25359219e-01 -1.77652791e-01 -3.91114969e-03 -1.27987194e+00
-5.71059167e-01 -4.95536160e-03 3.93467069e-01 8.78227890e-01
2.41669074e-01 -6.85861707e-01 1.07016897e+00 -2.18104377e-01
-1.74557790e-01 -5.38180828e-01 -1.20384920e+00 -8.56652856e-01
2.96699703e-01 -3.91334832e-01 2.99089283e-01 9.32329237e-01
-5.49976170e-01 4.55792576e-01 -7.79467225e-02 -8.15840513e-02
4.34104800e-01 1.16689034e-01 7.55481362e-01 -1.57894647e+00
-6.94292784e-01 -6.35958135e-01 -8.32283616e-01 -6.02660961e-02
-8.78963768e-02 -1.13978839e+00 -1.92426294e-01 -1.48859417e+00
8.60671327e-02 -6.51248813e-01 -3.88539851e-01 1.92442358e-01
1.07451022e-01 6.93316519e-01 1.26350783e-02 3.39415342e-01
1.03745170e-01 -2.09004998e-01 3.69978726e-01 5.39737456e-02
6.91112056e-02 2.73394018e-01 -3.89859021e-01 8.49907637e-01
1.25186634e+00 -6.62661552e-01 -2.12179303e-01 -2.56804645e-01
5.05224168e-01 3.14274952e-02 4.17557865e-01 -1.48687005e+00
-1.28314853e-01 1.85678676e-01 9.76910830e-01 -6.11157477e-01
2.95050532e-01 -4.75007653e-01 1.45879298e-01 5.14592111e-01
9.03970003e-02 1.61516234e-01 1.43376902e-01 2.40735367e-01
-1.13470174e-01 -9.05168474e-01 9.06888425e-01 -3.94284278e-01
-6.46364868e-01 -1.52589753e-01 -6.12316966e-01 -4.65108424e-01
8.68311405e-01 -4.46545631e-02 -4.19903755e-01 -6.21643066e-02
-3.94422740e-01 -7.37862170e-01 -3.59935760e-02 4.01195943e-01
1.64416470e-02 -1.15022898e+00 -8.57202053e-01 1.02620952e-01
1.86976448e-01 -5.08088410e-01 -1.27550736e-01 5.64823925e-01
-7.98885524e-01 7.83847272e-01 -7.55124271e-01 -4.78080839e-01
-1.40579665e+00 5.21455646e-01 5.87540030e-01 -3.45524371e-01
-9.94755253e-02 9.74670529e-01 1.52066246e-01 -4.95600998e-01
-8.46020430e-02 1.44734159e-01 -4.40995425e-01 3.45048159e-01
3.95130545e-01 7.84540534e-01 9.79573131e-01 -4.33533370e-01
-2.46253937e-01 1.92714199e-01 -8.67992342e-02 -2.66156822e-01
1.48126662e+00 5.75644612e-01 -3.17591757e-01 3.46556336e-01
1.03407502e+00 -6.90285936e-02 -6.18058145e-01 -5.76877184e-02
3.92221063e-01 -4.98739839e-01 5.68704128e-01 -1.05684531e+00
-8.95469666e-01 8.05886984e-01 9.91383195e-01 4.88774210e-01
1.10039091e+00 1.20718472e-01 -6.15124069e-02 5.40975749e-01
4.27984893e-01 -1.08617973e+00 -2.37710699e-02 2.97051519e-01
1.01278532e+00 -1.30741847e+00 4.29484814e-01 -1.26159504e-01
-1.64974645e-01 1.38294792e+00 2.29742706e-01 -1.98261678e-01
7.11257935e-01 -1.17802493e-01 1.61127523e-01 -3.09711635e-01
-5.06097674e-01 -2.50606626e-01 2.76555866e-01 6.11598551e-01
6.75185680e-01 1.47106558e-01 -6.83020115e-01 2.43755326e-01
-4.66329336e-01 -2.30041683e-01 3.33409220e-01 1.09787750e+00
-6.71762705e-01 -9.86316621e-01 -1.05645704e+00 8.18885207e-01
-8.78347576e-01 1.44941851e-01 -5.61322927e-01 9.88638759e-01
3.51254284e-01 1.13182950e+00 2.05855206e-01 -4.58502285e-02
-1.02171838e-01 1.27576977e-01 4.28173572e-01 -1.90942526e-01
-9.43975687e-01 1.19527601e-01 2.48867452e-01 2.04809576e-01
-5.39428055e-01 -7.45876729e-01 -5.97446680e-01 -2.76025217e-02
-5.31282246e-01 3.89141500e-01 1.60238647e+00 9.58409548e-01
-2.12031379e-01 4.41246688e-01 8.25828969e-01 -9.61118460e-01
-4.25647259e-01 -1.30818045e+00 -4.33928847e-01 9.98791009e-02
1.83139704e-02 -9.26938772e-01 -5.27502716e-01 -2.71670669e-01] | [7.729050159454346, 3.0335752964019775] |
99324c05-c00f-48d7-8caa-85d7b4e8abf0 | on-learning-to-summarize-with-large-language | 2305.14239 | null | https://arxiv.org/abs/2305.14239v1 | https://arxiv.org/pdf/2305.14239v1.pdf | On Learning to Summarize with Large Language Models as References | Recent studies have found that summaries generated by large language models (LLMs) are favored by human annotators over the original reference summaries in commonly used summarization datasets. Therefore, we investigate a new learning paradigm of text summarization models that considers the LLMs as the reference or the gold-standard oracle on commonly used summarization datasets such as the CNN/DailyMail dataset. To examine the standard practices that are aligned with the new learning setting, we propose a novel training method that is based on contrastive learning with LLMs as a summarization quality evaluator. For this reward-based training method, we investigate two different methods of utilizing LLMs for summary quality evaluation, namely GPTScore and GPTRank. Our experiments on the CNN/DailyMail dataset demonstrate that smaller summarization models trained by our proposed method can achieve performance equal to or surpass that of the reference LLMs, as evaluated by the LLMs themselves. This underscores the efficacy of our proposed paradigm in enhancing model performance over the standard maximum likelihood estimation (MLE) training method, and its efficiency since it only requires a small budget to access the LLMs. We release the training scripts, model outputs, and LLM-based evaluation results to facilitate future studies. | ['Arman Cohan', 'Dragomir Radev', 'PengFei Liu', 'Alexander R. Fabbri', 'Yixin Liu'] | 2023-05-23 | null | null | null | null | ['text-summarization'] | ['natural-language-processing'] | [ 2.55849928e-01 3.05855483e-01 -4.52930659e-01 -3.92581165e-01
-1.30927575e+00 -5.95194101e-01 8.35377693e-01 3.83917630e-01
-5.72413504e-01 9.33411419e-01 6.84242368e-01 -1.23344041e-01
8.66063759e-02 -6.42597973e-01 -8.72798979e-01 -2.82812297e-01
2.91786492e-01 3.44429404e-01 2.94507947e-02 5.15268184e-02
6.66502118e-01 -9.29188207e-02 -1.23304892e+00 5.31953573e-01
1.66535056e+00 8.46392751e-01 1.43357337e-01 9.95172083e-01
-7.04785110e-04 8.15981150e-01 -1.32593775e+00 -8.31319809e-01
5.26709966e-02 -5.49396396e-01 -7.22233415e-01 -1.54632360e-01
9.49107528e-01 -5.73294640e-01 -3.47883254e-01 9.51666653e-01
8.06428909e-01 2.31249124e-01 8.77679884e-01 -1.17470706e+00
-8.82913947e-01 1.14415526e+00 -3.36025000e-01 2.36809283e-01
3.57571602e-01 2.25830168e-01 1.38015151e+00 -4.88281548e-01
4.71882224e-01 1.19486368e+00 6.74402893e-01 4.60940093e-01
-9.12356555e-01 -4.28026825e-01 8.82510766e-02 1.69123083e-01
-1.05531514e+00 -6.58498883e-01 5.41521013e-01 -8.73756856e-02
1.23415518e+00 5.60014904e-01 3.47904652e-01 1.34196544e+00
3.59906644e-01 1.34818733e+00 5.75986922e-01 -3.10366511e-01
2.68675953e-01 6.48385510e-02 4.48880076e-01 4.33338583e-01
7.28359759e-01 -3.55650604e-01 -6.25439763e-01 -3.48684013e-01
1.67813897e-01 -3.84820342e-01 -5.23997068e-01 2.45650485e-01
-1.27891445e+00 7.84923911e-01 1.47870094e-01 6.29998371e-02
-4.94412512e-01 2.61406213e-01 6.50375962e-01 1.40828744e-01
6.70153618e-01 8.01090300e-01 -3.77548039e-01 -3.19196761e-01
-1.51630723e+00 3.67565125e-01 1.09321976e+00 1.08361745e+00
3.40625256e-01 8.55964273e-02 -9.20102298e-01 8.86216581e-01
1.69323221e-01 5.64146340e-01 8.17127109e-01 -1.08830798e+00
8.54374468e-01 6.48537636e-01 1.84770748e-01 -9.31971192e-01
-1.02859743e-01 -8.06158781e-01 -1.15337563e+00 -4.67208505e-01
-6.03492185e-02 -2.12457851e-01 -3.75219673e-01 1.56201053e+00
-2.52101928e-01 1.14076711e-01 3.96384448e-01 3.56245339e-01
1.18074811e+00 9.20075953e-01 -1.18109360e-01 -4.85907823e-01
9.20913279e-01 -1.47162151e+00 -8.67366016e-01 -2.34217614e-01
7.97796726e-01 -5.62933505e-01 1.29651904e+00 4.09157783e-01
-1.14634395e+00 -4.70625401e-01 -1.37907326e+00 -2.41973647e-03
9.24660116e-02 5.82519054e-01 1.71855584e-01 5.04583240e-01
-1.16589618e+00 1.06989002e+00 -7.55429149e-01 -4.07091647e-01
4.20857161e-01 -1.75563805e-02 -7.77051225e-02 2.98745453e-01
-1.13718235e+00 9.00595009e-01 6.52323365e-01 -1.15050606e-01
-8.73591781e-01 -4.87558126e-01 -8.44858885e-01 4.35396224e-01
1.94088697e-01 -9.08204556e-01 1.74759936e+00 -6.88022733e-01
-1.50075877e+00 5.04700005e-01 -1.42713130e-01 -8.41845870e-01
6.96388364e-01 -6.84762299e-01 -1.67972848e-01 1.21631615e-01
2.54976302e-01 4.89125431e-01 6.63646042e-01 -1.06877863e+00
-5.57845652e-01 1.61219165e-01 -5.52532673e-02 2.99645364e-01
-4.64611381e-01 -1.96933538e-01 -2.16817513e-01 -6.93583071e-01
-4.74935025e-01 -5.14421642e-01 -5.68982586e-02 -8.15825164e-01
-9.77070391e-01 -5.07130265e-01 3.27785134e-01 -8.66291761e-01
1.89695084e+00 -1.55356252e+00 -1.18315816e-01 -2.81123489e-01
1.71338737e-01 6.77015007e-01 -4.51643139e-01 7.41381764e-01
3.87405604e-01 4.01941001e-01 -3.83815289e-01 -8.51098418e-01
1.81150824e-01 -9.53688323e-02 -4.15321171e-01 9.79564786e-02
2.29918674e-01 1.09964824e+00 -9.74624574e-01 -7.15935707e-01
-1.68219864e-01 -2.14776434e-02 -3.96512091e-01 5.37707329e-01
-2.84697592e-01 4.31290977e-02 -3.50620151e-01 3.46843272e-01
4.92938340e-01 -4.43790317e-01 -4.50539365e-02 -2.38248080e-01
-6.22161962e-02 5.25265157e-01 -9.22402859e-01 1.68078971e+00
-2.05608204e-01 6.59492314e-01 -5.54914176e-01 -7.93368638e-01
9.91709590e-01 3.47662777e-01 -8.49931911e-02 -5.55722952e-01
6.99735805e-02 3.33234280e-01 -2.76446551e-01 -4.74229902e-01
1.09029365e+00 4.52090323e-01 -3.31744179e-02 7.52992988e-01
3.54002982e-01 -1.88596338e-01 5.00550151e-01 6.22398436e-01
1.21746826e+00 -1.29543409e-01 6.30181432e-01 -1.74294189e-01
4.48431313e-01 -1.24877430e-01 6.23996437e-01 1.39853156e+00
-1.00983948e-01 7.14152753e-01 7.22702086e-01 1.39042690e-01
-1.13457048e+00 -6.56806707e-01 7.66552985e-02 1.02516675e+00
-2.24893034e-01 -7.04837263e-01 -1.01978672e+00 -9.90657449e-01
-2.75778323e-01 1.22766960e+00 -3.14294994e-01 -3.09013247e-01
-4.49256092e-01 -7.88966835e-01 9.35484767e-01 4.90059078e-01
7.49837816e-01 -1.18925381e+00 -6.78725362e-01 8.44086781e-02
-4.04803425e-01 -9.94783401e-01 -6.56395137e-01 -1.98622540e-01
-1.07119083e+00 -8.39995623e-01 -8.26173365e-01 -3.70853812e-01
2.06866786e-01 7.60294357e-03 1.33596766e+00 -4.80337627e-02
4.78698611e-01 4.20508981e-01 -4.27009493e-01 -4.13199753e-01
-8.79542112e-01 7.58313954e-01 -4.59293136e-03 -1.73680380e-01
-1.17807679e-01 -4.20955747e-01 -7.23130763e-01 -1.74052700e-01
-1.06031859e+00 1.89232469e-01 9.28529501e-01 6.64575756e-01
3.26353282e-01 -3.19057107e-01 1.27497303e+00 -1.03718603e+00
1.42839897e+00 -3.91122162e-01 -2.16977507e-01 6.31128132e-01
-9.35100555e-01 2.07448050e-01 6.85042739e-01 -3.16110820e-01
-1.04194272e+00 -5.06036699e-01 -1.14727490e-01 -3.65871638e-02
1.07466385e-01 8.89904737e-01 -1.62113383e-01 5.07914662e-01
8.50417972e-01 3.35235387e-01 -2.54446328e-01 -4.27939802e-01
3.91585052e-01 1.00108624e+00 7.51704991e-01 -3.20159316e-01
4.05902982e-01 -5.44370636e-02 -4.95084733e-01 -5.87938249e-01
-1.10871613e+00 -3.50831777e-01 -3.94757330e-01 -6.82860315e-02
5.16679585e-01 -8.33911419e-01 -5.42593658e-01 4.47859585e-01
-1.53583074e+00 -1.34658739e-01 -2.76416421e-01 4.16408360e-01
-5.73303223e-01 8.75465035e-01 -5.64026594e-01 -7.31415749e-01
-1.38784671e+00 -9.06273246e-01 1.15772343e+00 4.15163368e-01
-5.99452198e-01 -1.05700886e+00 3.45623702e-01 3.18777949e-01
5.69060147e-01 1.55963182e-01 9.65818167e-01 -1.35858226e+00
-2.15737954e-01 -4.67927784e-01 -7.42408410e-02 9.20620143e-01
-8.54591094e-03 2.40965173e-01 -9.15640712e-01 -3.58729392e-01
-6.18045405e-03 -3.60878378e-01 1.17249382e+00 5.48735023e-01
9.58811879e-01 -7.52508461e-01 -8.58872309e-02 1.29262760e-01
1.13092351e+00 -2.29925796e-01 5.60315192e-01 4.78713125e-01
4.97492760e-01 2.49146581e-01 6.61657691e-01 5.22351444e-01
4.54211861e-01 2.17854559e-01 1.94644883e-01 2.35020638e-01
-5.13601862e-02 -5.53498507e-01 6.75660253e-01 1.36266434e+00
1.83699220e-01 -8.54220629e-01 -6.86037064e-01 5.19871891e-01
-2.27717710e+00 -9.65070307e-01 -5.99906445e-02 2.30937433e+00
1.10194075e+00 2.66814023e-01 -1.95637062e-01 -1.11906603e-01
6.31985128e-01 5.55676460e-01 -6.43162429e-01 -5.20375252e-01
-2.72547185e-01 -3.50597911e-02 2.72857487e-01 3.16674501e-01
-1.03767407e+00 7.77912259e-01 6.61981344e+00 1.12305188e+00
-7.89161623e-01 3.33682299e-02 6.59647822e-01 -8.81884247e-02
-3.87970865e-01 -1.95870683e-01 -8.12974811e-01 5.24921715e-01
1.25872171e+00 -7.01757610e-01 -4.96245213e-02 9.04254079e-01
5.15392184e-01 -1.89325705e-01 -1.43608105e+00 9.19224560e-01
4.02186275e-01 -1.46060669e+00 5.20160675e-01 -3.26729000e-01
1.10568762e+00 1.00901484e-01 -1.69105947e-01 6.97569132e-01
1.59505293e-01 -9.77853537e-01 7.01822042e-01 7.25362003e-01
6.14329696e-01 -6.08051121e-01 1.12196946e+00 7.83102989e-01
-5.77243209e-01 1.21194631e-01 -5.54221034e-01 1.37581542e-01
2.15395153e-01 6.97257042e-01 -1.04591775e+00 8.31069350e-01
2.03772977e-01 9.00019228e-01 -1.00765276e+00 1.21939945e+00
-2.87913352e-01 9.29315865e-01 7.53839388e-02 -1.92693040e-01
1.93462789e-01 1.14878930e-01 7.18443155e-01 1.51352620e+00
3.98793221e-01 -4.07223135e-01 -1.23375088e-01 8.81055355e-01
-6.76367879e-01 3.19505304e-01 -3.97570223e-01 -2.63017297e-01
6.51646912e-01 1.24467289e+00 -2.01427266e-01 -8.21416914e-01
-7.70487189e-02 7.26122975e-01 3.63046438e-01 2.91291147e-01
-8.81385267e-01 -5.55630922e-01 -3.54084074e-02 -2.92564660e-01
2.06693634e-03 8.76264796e-02 -3.78056407e-01 -1.39840412e+00
1.25696942e-01 -1.14150965e+00 3.41444224e-01 -6.82397068e-01
-1.23990703e+00 7.82648146e-01 2.02306613e-01 -1.25619996e+00
-4.24335778e-01 -4.10308316e-02 -1.13475513e+00 5.74189723e-01
-1.67848241e+00 -7.96521842e-01 -3.68376434e-01 -4.40921308e-03
9.60697949e-01 -3.31670940e-01 5.76174915e-01 -3.83763458e-04
-7.75445819e-01 8.25793207e-01 4.26099509e-01 -1.22195184e-02
9.66476858e-01 -1.28877807e+00 5.25169015e-01 9.09655690e-01
4.68388386e-02 5.34987807e-01 8.39536726e-01 -7.88807034e-01
-9.20284867e-01 -1.26897764e+00 1.13090277e+00 -5.11968434e-01
3.16557080e-01 1.02946594e-01 -1.06229126e+00 7.67182469e-01
8.05441022e-01 -7.92396724e-01 7.61211812e-01 -4.06439640e-02
-8.89265910e-03 -3.94639634e-02 -8.86816621e-01 5.37382841e-01
7.18436062e-01 -2.06024528e-01 -9.98371720e-01 4.03738797e-01
1.09482312e+00 -2.21349135e-01 -7.55251944e-01 7.05327511e-01
4.17878270e-01 -7.49304295e-01 5.08485317e-01 -7.03246236e-01
9.45632041e-01 1.24459881e-02 -1.39985502e-01 -1.55457246e+00
7.14304000e-02 -6.05093002e-01 -6.64475918e-01 1.64031029e+00
5.39015591e-01 -5.11532187e-01 4.19162303e-01 4.88895595e-01
-4.16007131e-01 -8.72818172e-01 -8.04311275e-01 -6.95934832e-01
1.18977062e-01 -2.54760653e-01 5.89961052e-01 6.70578182e-01
-1.54095531e-01 6.89519167e-01 -3.41817200e-01 5.58939716e-03
5.48104286e-01 9.80917588e-02 9.55642760e-01 -9.48709548e-01
-2.82982379e-01 -6.77259326e-01 6.39868155e-02 -1.24127638e+00
2.67067701e-01 -1.02610791e+00 -3.77071425e-02 -2.05147552e+00
6.65510118e-01 8.32020938e-02 -2.47827500e-01 1.90212354e-01
-5.33716500e-01 -2.53943175e-01 7.23337904e-02 3.97700310e-01
-1.33503771e+00 9.80133414e-01 1.08632672e+00 -3.45320821e-01
-1.82867616e-01 2.20158488e-01 -1.02041459e+00 5.83273411e-01
8.15584183e-01 -5.61834991e-01 -3.76566291e-01 -5.26039481e-01
3.28721017e-01 3.83867025e-02 1.07777953e-01 -1.05633020e+00
3.86423916e-01 1.69344366e-01 1.06059890e-02 -9.05195415e-01
-2.21934438e-01 -2.98800375e-02 -2.50446618e-01 2.45085418e-01
-1.05998743e+00 1.03165828e-01 -9.87290312e-03 5.72979689e-01
-2.81803608e-01 -8.14115703e-01 3.97359520e-01 -1.05576389e-01
-1.60343140e-01 -1.19538419e-02 -2.55019397e-01 2.37664580e-01
4.63113904e-01 1.05156638e-01 -8.35120380e-01 -7.51880527e-01
-2.33434319e-01 5.23927093e-01 2.60687828e-01 2.99498558e-01
5.65775156e-01 -1.15760827e+00 -1.25835633e+00 -4.07843173e-01
1.00513533e-01 9.92514938e-02 1.56679675e-01 9.23376322e-01
-4.80915189e-01 7.36666203e-01 1.51508600e-01 -5.07391393e-01
-1.11340046e+00 1.99655443e-01 1.68855414e-01 -8.41866255e-01
-3.82497936e-01 5.12162268e-01 -1.36509808e-02 -5.57484865e-01
4.03743863e-01 -6.00047946e-01 -4.02162015e-01 6.27511516e-02
5.56604922e-01 8.26605260e-01 2.37250820e-01 -1.75722316e-01
1.40619159e-01 -1.09624602e-01 -2.28162870e-01 -5.22497520e-02
1.15336442e+00 -1.12049356e-01 -1.08482212e-01 5.75418115e-01
1.06124341e+00 1.01172961e-01 -9.98143733e-01 -3.39997560e-01
2.55874276e-01 -1.70884654e-01 -1.50412008e-01 -8.52543533e-01
-5.99234521e-01 7.00432658e-01 1.56352922e-01 3.06249559e-01
9.31212723e-01 -1.32456511e-01 8.99408281e-01 7.56072521e-01
-6.08339757e-02 -1.17374814e+00 3.27169061e-01 6.02189600e-01
1.16830695e+00 -1.37271273e+00 1.19217485e-01 2.33425245e-01
-1.02133358e+00 1.03363347e+00 6.55789137e-01 -9.98468250e-02
-7.79994503e-02 -3.25382650e-01 -1.30174264e-01 3.56534719e-02
-1.13798892e+00 3.29628497e-01 4.95387495e-01 2.07620338e-01
5.44315279e-01 8.73516649e-02 -5.82432091e-01 9.99682546e-01
-5.42530179e-01 -1.42526552e-01 8.15124869e-01 6.43941760e-01
-6.02994561e-01 -7.96726286e-01 -2.40810569e-02 9.42791820e-01
-5.04639745e-01 -3.12713444e-01 -4.27852601e-01 4.39025193e-01
-4.84276682e-01 1.21746802e+00 -1.36075750e-01 -3.02811086e-01
5.66974163e-01 1.04723722e-01 1.16500221e-01 -7.79454172e-01
-7.53939271e-01 -1.83337495e-01 2.83839583e-01 -3.20977807e-01
-3.38562369e-01 -4.83189285e-01 -9.47418690e-01 -3.53994817e-01
-5.85605323e-01 4.15412039e-01 4.18083638e-01 8.58386874e-01
5.24771333e-01 7.27431476e-01 6.36957705e-01 -7.18749464e-01
-1.18656659e+00 -1.55405140e+00 -2.49610603e-01 2.73899525e-01
2.86686867e-01 -4.14940007e-02 -3.86235237e-01 -9.25658941e-02] | [12.320684432983398, 9.346410751342773] |
4bc8159c-71a6-4d17-b6c1-80ab0bbbb7a1 | bert-post-training-for-review-reading | 1904.02232 | null | https://arxiv.org/abs/1904.02232v2 | https://arxiv.org/pdf/1904.02232v2.pdf | BERT Post-Training for Review Reading Comprehension and Aspect-based Sentiment Analysis | Question-answering plays an important role in e-commerce as it allows potential customers to actively seek crucial information about products or services to help their purchase decision making. Inspired by the recent success of machine reading comprehension (MRC) on formal documents, this paper explores the potential of turning customer reviews into a large source of knowledge that can be exploited to answer user questions.~We call this problem Review Reading Comprehension (RRC). To the best of our knowledge, no existing work has been done on RRC. In this work, we first build an RRC dataset called ReviewRC based on a popular benchmark for aspect-based sentiment analysis. Since ReviewRC has limited training examples for RRC (and also for aspect-based sentiment analysis), we then explore a novel post-training approach on the popular language model BERT to enhance the performance of fine-tuning of BERT for RRC. To show the generality of the approach, the proposed post-training is also applied to some other review-based tasks such as aspect extraction and aspect sentiment classification in aspect-based sentiment analysis. Experimental results demonstrate that the proposed post-training is highly effective. The datasets and code are available at https://www.cs.uic.edu/~hxu/. | ['Bing Liu', 'Philip S. Yu', 'Hu Xu', 'Lei Shu'] | 2019-04-03 | bert-post-training-for-review-reading-1 | https://aclanthology.org/N19-1242 | https://aclanthology.org/N19-1242.pdf | naacl-2019-6 | ['aspect-extraction'] | ['natural-language-processing'] | [ 3.91699970e-01 2.43902683e-01 -3.84172410e-01 -7.07412124e-01
-9.66980934e-01 -5.74959755e-01 5.89591563e-01 6.03262067e-01
-3.98929715e-01 2.36502409e-01 2.86919236e-01 -9.25100327e-01
7.64626563e-02 -9.22955871e-01 -5.32028675e-01 -1.58961922e-01
3.41431081e-01 3.33153367e-01 1.69950753e-01 -8.01496863e-01
7.45891392e-01 -1.11472502e-01 -1.42574656e+00 5.18020391e-01
9.49234366e-01 9.84547377e-01 2.44934469e-01 6.74796045e-01
-5.94825268e-01 7.52758741e-01 -5.79021335e-01 -9.03884053e-01
-5.54131418e-02 -4.60057586e-01 -1.04485166e+00 7.92169422e-02
-2.61059970e-01 2.45436393e-02 4.64893937e-01 9.20512438e-01
2.10521370e-01 2.66570985e-01 4.13986087e-01 -1.09882927e+00
-8.62426519e-01 9.05192971e-01 -6.26195550e-01 5.41489050e-02
6.19998932e-01 -2.85449237e-01 1.52322018e+00 -9.00348604e-01
3.61106217e-01 1.13108444e+00 2.41075993e-01 4.04089034e-01
-3.87594938e-01 -2.38312259e-01 6.15482688e-01 4.11427736e-01
-7.11115599e-01 -1.69526666e-01 1.04622960e+00 6.61713779e-02
1.24068308e+00 3.62803042e-01 6.91500187e-01 5.85355699e-01
1.50534302e-01 1.29392242e+00 1.15403748e+00 -7.70170271e-01
3.99823606e-01 5.64118385e-01 8.23753595e-01 4.39205170e-01
2.58137614e-01 -5.64182281e-01 -4.15336847e-01 6.09365851e-02
1.86471850e-01 -6.57950640e-02 -1.92627713e-01 1.23182088e-01
-8.71490240e-01 1.21277881e+00 4.42193374e-02 3.27578127e-01
-3.52372527e-01 -1.42667353e-01 4.75549161e-01 6.08014464e-01
5.76989174e-01 7.40997076e-01 -9.91813779e-01 -3.33648473e-01
-3.09199274e-01 2.04055831e-01 1.41481686e+00 1.15326333e+00
5.64335942e-01 -3.92541170e-01 1.53642952e-01 9.98121381e-01
4.89782572e-01 6.43089652e-01 6.93558931e-01 -5.97704172e-01
6.05365634e-01 9.81795073e-01 -1.94800138e-01 -9.97087896e-01
-1.88158005e-01 -6.27764463e-01 -4.98122811e-01 -4.56705093e-01
-1.07368544e-01 -7.83791840e-02 -6.06531978e-01 1.15728402e+00
2.58317888e-01 -3.21457386e-01 2.77959377e-01 7.35757411e-01
1.15077519e+00 6.89360619e-01 -1.85105707e-02 -9.37430710e-02
1.84504795e+00 -1.29579723e+00 -8.32742095e-01 -3.53971481e-01
6.46004379e-01 -1.03305292e+00 1.20527911e+00 5.62317491e-01
-7.80763984e-01 -3.05207491e-01 -1.03365874e+00 -2.96220273e-01
-8.72220516e-01 -1.62848085e-02 1.14003062e+00 7.20364451e-01
-7.41327286e-01 -5.99017516e-02 -6.22582555e-01 -5.23875177e-01
2.66902000e-01 4.50495370e-02 9.58599746e-02 -3.58576834e-01
-1.38650727e+00 6.77171350e-01 -2.39050388e-02 2.20210105e-01
-2.61555612e-01 -4.68506753e-01 -9.52628851e-01 -3.22318636e-02
9.19696569e-01 -5.50264120e-01 1.75581372e+00 -9.67439711e-01
-1.56970787e+00 5.91167212e-01 -5.13318241e-01 -2.95942605e-01
-2.32548401e-01 -4.58144963e-01 -5.19631445e-01 1.73928306e-01
3.04063618e-01 3.92805725e-01 7.42647409e-01 -1.16421247e+00
-7.89769888e-01 -4.48395252e-01 7.11098433e-01 3.07286561e-01
-1.85879752e-01 2.48496756e-01 -7.65088081e-01 -5.75772285e-01
-4.43145484e-02 -7.32654989e-01 -3.78764600e-01 -6.30869150e-01
-2.33891010e-01 -5.28927386e-01 5.30674577e-01 -5.83351135e-01
1.27646637e+00 -1.58581388e+00 -2.43943155e-01 3.26836526e-01
1.17402533e-02 3.18837017e-01 -4.80095565e-01 7.09826410e-01
-1.84108950e-02 4.01700824e-01 -2.36567259e-01 -2.95839816e-01
1.44696847e-01 1.33633748e-01 -4.00093973e-01 -1.92915887e-01
2.21290797e-01 1.41321182e+00 -7.83753693e-01 -1.96205765e-01
-5.97371385e-02 3.11446935e-01 -4.63903725e-01 2.83015706e-03
-5.93734980e-01 3.13942358e-02 -8.91463041e-01 9.64524567e-01
5.50209939e-01 -3.35981131e-01 -1.06762677e-01 1.17256120e-01
1.82161257e-01 6.02577209e-01 -7.15356529e-01 1.43857551e+00
-7.83794284e-01 4.19446170e-01 -1.48394048e-01 -1.19495583e+00
9.40389156e-01 7.32442141e-02 1.85279906e-01 -9.59943652e-01
2.85579205e-01 2.08376646e-01 -8.52392763e-02 -4.32853878e-01
9.60225165e-01 -2.19862834e-01 -1.56113014e-01 8.29437017e-01
-3.14893156e-01 -4.85408485e-01 5.36946237e-01 4.89223629e-01
8.22761655e-01 -1.52293146e-01 5.96688211e-01 -9.34496820e-02
9.84716892e-01 2.77827650e-01 1.96757659e-01 6.30955815e-01
-4.65373248e-02 3.42940569e-01 6.92571044e-01 4.52540815e-02
-4.65200454e-01 -4.33767319e-01 -7.83447549e-03 1.26160610e+00
1.34843916e-01 -9.07281756e-01 -6.82258487e-01 -8.46995056e-01
-1.56136647e-01 1.13836646e+00 -4.72013950e-01 1.27095524e-02
-3.63187432e-01 -7.38049746e-01 -1.04235530e-01 5.11996508e-01
5.90274036e-01 -1.20447040e+00 -3.58847737e-01 2.07240805e-01
-3.25231910e-01 -1.25565767e+00 -3.36166024e-01 1.29135430e-01
-8.92492235e-01 -1.20567811e+00 -4.32562411e-01 -9.36127305e-01
5.73745608e-01 6.91855907e-01 1.51649761e+00 2.14408830e-01
1.25791460e-01 9.69451845e-01 -1.17277217e+00 -9.38698888e-01
-3.57888818e-01 6.46089613e-01 -4.70691264e-01 -2.69252598e-01
1.09668136e+00 -1.73836261e-01 -6.31175041e-01 3.60970646e-01
-1.10970581e+00 -8.56466405e-03 6.99189305e-01 3.61278713e-01
5.49481034e-01 2.34726205e-01 1.09950221e+00 -1.52370870e+00
1.21402943e+00 -4.54653442e-01 -3.84047210e-01 3.44160080e-01
-9.47367013e-01 -2.39028081e-01 4.60627198e-01 -4.89453301e-02
-1.06988072e+00 -5.54722071e-01 -5.14071822e-01 6.00010931e-01
1.07415281e-02 1.18104506e+00 -1.76734701e-01 3.49830091e-01
8.20360631e-02 1.78662196e-01 -9.81613770e-02 -4.69860047e-01
5.15195131e-01 8.49596679e-01 -1.57742780e-02 -3.62064421e-01
5.77444553e-01 5.68522215e-02 -4.15020496e-01 -9.07031953e-01
-1.35819483e+00 -1.02291131e+00 -3.42728913e-01 -2.14545019e-02
7.15826511e-01 -8.22114110e-01 -6.76645994e-01 2.96494365e-01
-9.54450667e-01 4.43480425e-02 -1.68122113e-01 2.25598320e-01
-2.94086456e-01 3.85833085e-01 -4.58666176e-01 -7.27577448e-01
-6.23346329e-01 -1.05583930e+00 1.12323284e+00 3.45279336e-01
-2.11264834e-01 -1.28237152e+00 -5.75248264e-02 9.55414474e-01
5.72047412e-01 -3.59376848e-01 1.13285744e+00 -8.93737972e-01
-6.58189654e-01 -4.92844850e-01 4.29784618e-02 6.71551108e-01
1.26108110e-01 -4.04031992e-01 -7.33714581e-01 4.55077365e-02
4.11272138e-01 -1.87491149e-01 9.17926431e-01 1.67138934e-01
9.23056781e-01 -4.71816398e-02 1.27720818e-01 -8.50242078e-02
1.37429380e+00 1.64885446e-01 7.56768286e-01 7.26943851e-01
4.84646976e-01 8.37902129e-01 1.13799345e+00 2.81257272e-01
9.12767649e-01 3.32981944e-01 1.79108441e-01 1.12391174e-01
2.94987619e-01 -2.02632368e-01 5.33783019e-01 1.47207975e+00
6.01104349e-02 -2.24874645e-01 -6.25939488e-01 5.51208496e-01
-1.76899922e+00 -5.62776268e-01 -2.95685202e-01 1.65224946e+00
7.55906582e-01 2.07398504e-01 -2.48954564e-01 1.94741488e-01
2.66873479e-01 2.12953851e-01 -4.77740407e-01 -8.64247978e-01
-1.91498324e-02 5.11231780e-01 1.88235402e-01 5.54993749e-01
-9.85111296e-01 1.09095311e+00 4.75251675e+00 7.59959936e-01
-6.13853335e-01 5.82522601e-02 7.05147982e-01 3.57350081e-01
-8.25489402e-01 1.49312660e-01 -9.85878229e-01 -9.18798968e-02
8.01067173e-01 -2.44425774e-01 2.67629266e-01 1.00072777e+00
1.71641812e-01 -5.62262833e-01 -9.91109371e-01 6.70772493e-01
4.70276177e-01 -9.27651763e-01 2.00504750e-01 -1.97572485e-01
6.90220058e-01 -1.87867299e-01 2.75334418e-02 8.74637365e-01
8.85111392e-02 -7.87639439e-01 1.98327184e-01 2.80377984e-01
1.98866874e-01 -1.00584102e+00 1.29330707e+00 4.31303412e-01
-1.13791966e+00 1.65909201e-01 -5.38338304e-01 -1.46190971e-01
2.82111287e-01 6.68352783e-01 -7.16652513e-01 7.24460721e-01
4.85866696e-01 9.60679650e-01 -9.47345853e-01 5.05436540e-01
-6.19625688e-01 7.67666698e-01 5.05160093e-02 -6.27167702e-01
2.63440132e-01 -4.24769938e-01 3.29291046e-01 9.50098276e-01
1.61640570e-02 3.03361326e-01 -2.32915580e-02 4.04780805e-01
-3.70538905e-02 5.60346365e-01 -3.28382730e-01 -3.28516722e-01
-1.15111217e-01 1.47973728e+00 -8.27898443e-01 -3.09887767e-01
-7.93542504e-01 7.57374704e-01 1.48567166e-02 4.73876089e-01
-4.80979443e-01 -5.84095120e-01 3.16966236e-01 -6.78966120e-02
6.36975050e-01 -5.04677817e-02 -3.94300103e-01 -1.35816455e+00
3.20524305e-01 -1.33168936e+00 1.54255927e-01 -5.69706440e-01
-1.27829576e+00 5.71388841e-01 -1.59015119e-01 -8.76272440e-01
-4.00347799e-01 -7.73598850e-01 -7.03876972e-01 8.04143429e-01
-2.06128001e+00 -1.05677068e+00 -8.33651200e-02 2.69385070e-01
9.08772111e-01 -2.13305935e-01 7.82266200e-01 -4.38781716e-02
-3.79820228e-01 3.97489905e-01 -1.89394131e-01 -9.40407589e-02
5.50923347e-01 -1.43302178e+00 5.78348339e-01 6.49246573e-01
1.10095575e-01 1.14716208e+00 5.48098862e-01 -6.37799859e-01
-1.88872242e+00 -9.26091313e-01 1.28370619e+00 -6.72232747e-01
8.61364961e-01 -3.58431667e-01 -8.83399487e-01 5.93427420e-01
3.98443162e-01 -8.99007022e-01 1.23208380e+00 5.68042099e-01
-2.21705705e-01 -2.76739478e-01 -7.60306954e-01 6.26060843e-01
5.02908111e-01 -5.64438343e-01 -9.68774736e-01 2.13676631e-01
1.09039474e+00 7.79685602e-02 -8.18208694e-01 3.69527757e-01
4.14445460e-01 -7.45613694e-01 6.54572487e-01 -4.52108264e-01
7.02569842e-01 -1.63401589e-01 -3.61083865e-01 -1.23595142e+00
2.53662080e-01 -3.52330148e-01 3.98261733e-02 1.58269072e+00
8.10802698e-01 -7.90977240e-01 5.84695816e-01 7.50674546e-01
-6.01528883e-02 -1.19168377e+00 -2.94071436e-01 -1.13961235e-01
1.36202604e-01 -1.05322564e+00 7.06769586e-01 4.81356829e-01
1.49264216e-01 9.36345637e-01 2.13543802e-01 -1.06828094e-01
1.59283072e-01 5.54435313e-01 8.44276428e-01 -9.49536562e-01
-4.40221339e-01 -3.70867521e-01 1.32451266e-01 -1.44717205e+00
1.09498702e-01 -8.88368964e-01 -8.09100941e-02 -1.87910140e+00
2.48008043e-01 -2.66703218e-01 -2.29245782e-01 2.58762538e-01
-5.10042191e-01 -6.02720939e-02 1.71757773e-01 -4.36741039e-02
-1.04474390e+00 6.65777564e-01 1.62571990e+00 -8.03515017e-02
-1.57595217e-01 5.79724967e-01 -1.66445804e+00 6.22636139e-01
9.65213895e-01 -2.13091671e-01 -7.71425903e-01 -1.67537957e-01
8.47287297e-01 -7.07357824e-02 -1.28499508e-01 -1.17744744e-01
9.03338194e-02 1.81347966e-01 -1.28516436e-01 -9.69860017e-01
1.43298674e-02 -8.42800319e-01 -8.06983888e-01 5.89223765e-02
-5.37477970e-01 3.30211341e-01 1.32673802e-02 5.40500402e-01
-6.17108762e-01 -6.94097757e-01 3.11782658e-02 -2.63745070e-01
-7.19579101e-01 -7.44024664e-02 -5.56691229e-01 9.96424109e-02
6.29596949e-01 1.50609180e-01 -3.88792127e-01 -9.76256013e-01
-1.54172838e-01 7.25286603e-01 7.33811408e-02 7.92733550e-01
7.45858431e-01 -8.95226717e-01 -5.31108975e-01 1.40956447e-01
4.49628770e-01 4.41632904e-02 -3.54340747e-02 6.56131446e-01
-3.50901425e-01 8.89832497e-01 5.10350466e-01 -4.19292271e-01
-1.18011308e+00 4.58422720e-01 -2.28144467e-01 -6.30162120e-01
-2.37766147e-01 5.77223778e-01 -2.24434175e-02 -8.03362012e-01
8.48996192e-02 -5.23963988e-01 -7.88101852e-01 1.66110992e-01
7.43080854e-01 3.03256642e-02 3.16302329e-01 -2.84082532e-01
-1.62523925e-01 6.50513887e-01 -5.68240762e-01 -2.41274923e-01
1.39804089e+00 -4.98222262e-01 -4.96877611e-01 5.00522017e-01
1.28893423e+00 1.17633611e-01 -3.49947095e-01 -3.05108309e-01
3.09921324e-01 -1.89741492e-01 9.14477333e-02 -1.02347445e+00
-8.88744473e-01 7.76427209e-01 4.87164361e-03 4.20890450e-01
1.32608318e+00 1.33279964e-01 1.01853907e+00 8.05706620e-01
3.69101465e-01 -1.19495916e+00 1.78148448e-02 8.63443255e-01
9.23684299e-01 -1.62465191e+00 1.00574888e-01 -6.75807714e-01
-1.01455307e+00 9.57149029e-01 4.53735769e-01 1.42246321e-01
9.15042102e-01 -6.13186732e-02 3.71899873e-01 -4.63531494e-01
-9.67044711e-01 -4.70547497e-01 3.23980272e-01 3.96696776e-01
6.60691023e-01 2.03607529e-02 -6.63717985e-01 1.07288599e+00
-5.90585351e-01 -9.68739092e-02 6.83960617e-01 1.19134521e+00
-5.55322051e-01 -1.47132838e+00 5.50051853e-02 6.38419688e-01
-6.98710918e-01 -4.89179194e-01 -4.50649738e-01 7.20873654e-01
-3.86083633e-01 1.58369160e+00 -3.09987545e-01 -9.47271809e-02
4.56933707e-01 2.64243521e-02 1.85912430e-01 -8.56580853e-01
-7.88429141e-01 2.30967268e-01 4.29168314e-01 -3.00121129e-01
-8.22852850e-01 -5.98502398e-01 -1.17240024e+00 -1.23036414e-01
-5.04914880e-01 5.60974598e-01 9.98923421e-01 9.86075401e-01
1.65953398e-01 5.42307436e-01 7.57789493e-01 -1.38631733e-02
-3.75210822e-01 -1.09644544e+00 -4.24228936e-01 2.44002268e-01
6.90473691e-02 -1.65134251e-01 -3.59867483e-01 -1.35571763e-01] | [11.467771530151367, 6.650811195373535] |
75f9b016-08bf-43f3-b1ed-617ca5b6012f | improving-audio-caption-fluency-with | 2306.10090 | null | https://arxiv.org/abs/2306.10090v1 | https://arxiv.org/pdf/2306.10090v1.pdf | Improving Audio Caption Fluency with Automatic Error Correction | Automated audio captioning (AAC) is an important cross-modality translation task, aiming at generating descriptions for audio clips. However, captions generated by previous AAC models have faced ``false-repetition'' errors due to the training objective. In such scenarios, we propose a new task of AAC error correction and hope to reduce such errors by post-processing AAC outputs. To tackle this problem, we use observation-based rules to corrupt captions without errors, for pseudo grammatically-erroneous sentence generation. One pair of corrupted and clean sentences can thus be used for training. We train a neural network-based model on the synthetic error dataset and apply the model to correct real errors in AAC outputs. Results on two benchmark datasets indicate that our approach significantly improves fluency while maintaining semantic information. | ['Kai Yu', 'Mengyue Wu', 'Xuenan Xu', 'Zeyu Xie', 'Hanxue Zhang'] | 2023-06-16 | null | null | null | null | ['audio-captioning'] | ['audio'] | [ 8.92074227e-01 4.58627015e-01 4.28365409e-01 -2.09675357e-01
-1.38297904e+00 -3.24056119e-01 3.93812627e-01 6.04100302e-02
-1.29020840e-01 1.12717342e+00 6.00818992e-01 -1.66901276e-01
4.71029162e-01 -4.63990390e-01 -1.12048960e+00 -2.74069104e-02
3.03591460e-01 5.14094055e-01 -2.42383912e-01 -2.48414382e-01
6.86412901e-02 -1.28143460e-01 -1.55004501e+00 1.07612312e+00
1.24306023e+00 6.46528244e-01 1.34462923e-01 9.31595027e-01
-1.20609447e-01 9.98257041e-01 -1.08901596e+00 -6.20639026e-01
-1.86725482e-01 -1.03735960e+00 -1.02137506e+00 -5.39714545e-02
3.02345306e-01 -1.86213627e-01 -3.24213356e-02 1.11677921e+00
6.52442813e-01 1.62856057e-02 5.76905370e-01 -1.26658571e+00
-6.65405691e-01 1.02843261e+00 2.68380523e-01 -2.19241396e-01
1.12446427e+00 1.32014409e-01 7.77532399e-01 -9.50177789e-01
5.54048181e-01 1.17569172e+00 7.84435987e-01 1.11863840e+00
-1.35458457e+00 -3.90925884e-01 -1.69609368e-01 4.52357918e-01
-1.21237087e+00 -9.02991295e-01 7.60099411e-01 -2.27465332e-01
8.79271507e-01 8.43282640e-01 5.51076472e-01 1.93789279e+00
-7.40920380e-02 8.37397814e-01 9.17114615e-01 -7.70116568e-01
2.44654104e-01 2.25159675e-01 -4.14448202e-01 1.91615894e-01
-2.16543034e-01 -1.02416448e-01 -7.51473010e-01 -2.48045828e-02
1.00278027e-01 -7.00043023e-01 -6.71655297e-01 3.69657725e-01
-1.37160647e+00 5.36153734e-01 1.63035452e-01 2.49865785e-01
-3.19942027e-01 1.09244160e-01 5.42001486e-01 6.27535522e-01
4.35830116e-01 9.33727384e-01 -3.29750210e-01 -5.88037848e-01
-8.83545935e-01 2.47201025e-01 7.58326113e-01 9.19140279e-01
1.16091490e-01 2.09476516e-01 -6.59434795e-01 1.11459684e+00
3.74687905e-03 4.74811226e-01 7.79284358e-01 -8.42631757e-01
8.37414384e-01 4.90299128e-02 2.62052208e-01 -7.73713946e-01
-1.22537769e-01 -3.59575570e-01 -8.30934048e-01 -7.85461739e-02
4.46076334e-01 -2.05740258e-01 -9.20992792e-01 1.83083689e+00
-1.93772867e-01 2.48229325e-01 3.47199202e-01 9.93776023e-01
6.85501575e-01 8.37708175e-01 1.01754509e-01 -3.82139623e-01
9.61283386e-01 -1.20471716e+00 -1.09899294e+00 -1.62444040e-01
7.45826840e-01 -1.05067825e+00 1.43167889e+00 5.09308219e-01
-1.27263308e+00 -5.16633332e-01 -8.28261852e-01 1.03321865e-01
-1.40757859e-01 2.91572183e-01 -1.06351912e-01 4.85845774e-01
-9.62513328e-01 6.70741439e-01 -3.10366750e-01 -1.68289244e-01
2.58180976e-01 -1.32485610e-02 -3.08473080e-01 -2.86979564e-02
-1.47167861e+00 8.35971057e-01 3.32236260e-01 1.05296232e-01
-8.07972193e-01 -8.39004397e-01 -8.79474878e-01 3.77680063e-02
1.05387591e-01 -8.46463323e-01 1.70036674e+00 -1.71453893e+00
-1.86397481e+00 4.02292430e-01 -1.73836485e-01 -5.97281158e-01
8.26191366e-01 -4.56754506e-01 -7.19052076e-01 1.33943141e-01
-2.47977301e-02 9.44969833e-01 1.13073444e+00 -1.38818848e+00
-3.59718055e-01 3.15260231e-01 -3.54794592e-01 2.65557736e-01
-1.73932627e-01 -2.58322842e-02 1.47498846e-01 -9.99917746e-01
-2.26310089e-01 -8.85790050e-01 1.17838249e-01 -3.58603984e-01
-7.58698702e-01 9.03581306e-02 4.73772168e-01 -1.21313095e+00
1.13046134e+00 -2.05916691e+00 3.33696276e-01 -8.73589963e-02
-3.89933497e-01 4.47325677e-01 -5.65922201e-01 3.95019799e-01
-1.57284439e-01 2.24263132e-01 -4.92084444e-01 -7.50597000e-01
-6.70939758e-02 8.56955051e-02 -7.53325880e-01 -2.61861145e-01
5.45361876e-01 6.11302972e-01 -1.11948824e+00 -2.73141891e-01
-9.79693327e-03 5.08979142e-01 -7.28952587e-01 6.26787066e-01
-5.93186200e-01 8.27828467e-01 1.62734538e-02 2.89955705e-01
4.11792129e-01 2.86939830e-01 -2.32485239e-03 -1.23960145e-01
1.74200073e-01 7.09314883e-01 -7.23108053e-01 1.97712839e+00
-8.83921862e-01 7.55813062e-01 -4.41308618e-01 -6.95675611e-01
7.58572400e-01 7.58480310e-01 7.86640644e-02 -7.37329662e-01
1.48271441e-01 5.32325625e-01 -1.00091137e-01 -7.12241709e-01
7.13610709e-01 -1.09748699e-01 -6.46938905e-02 2.33720422e-01
4.24299166e-02 -4.23456907e-01 3.23115177e-02 4.15978245e-02
1.21727371e+00 2.28076577e-01 -2.06527516e-01 5.00016212e-01
6.83928013e-01 1.03349440e-01 3.16655189e-01 7.90837586e-01
-7.50910118e-02 1.31736207e+00 3.20160121e-01 -2.29614899e-01
-1.30384600e+00 -9.01612759e-01 3.93018365e-01 7.65500546e-01
-3.28643978e-01 -4.37164634e-01 -1.13472962e+00 -7.65676022e-01
-4.47120577e-01 1.38662791e+00 -3.80580962e-01 -4.44742233e-01
-5.51230013e-01 -1.30839035e-01 9.88261759e-01 3.34505618e-01
2.30238736e-01 -1.35544276e+00 -2.75672019e-01 5.70830345e-01
-1.00496244e+00 -1.25317192e+00 -4.74321634e-01 -1.57245144e-01
-4.82675374e-01 -6.99209511e-01 -8.33593845e-01 -7.55289733e-01
5.55120885e-01 -1.90012634e-01 1.15099895e+00 8.38431064e-03
8.16468298e-02 1.61437064e-01 -7.43932903e-01 -3.80905211e-01
-1.39927411e+00 1.99158490e-02 1.99960396e-01 1.62677810e-01
-5.15821017e-02 -5.49322844e-01 -8.38137940e-02 -6.25139102e-03
-8.07479203e-01 4.68665898e-01 5.01642346e-01 9.85911250e-01
3.48838478e-01 -5.12950361e-01 9.33258593e-01 -5.79594731e-01
1.07843411e+00 -2.55584776e-01 -1.48236072e-02 2.54028738e-01
-1.97353795e-01 1.64372295e-01 1.19144320e+00 -5.78204215e-01
-1.05152869e+00 -1.06465900e-02 -4.61971223e-01 -5.86629808e-01
-3.74345809e-01 3.87599230e-01 -2.10432589e-01 4.85768080e-01
7.46999979e-01 3.93766582e-01 -1.30724818e-01 -5.43938637e-01
2.68214554e-01 1.23795784e+00 7.97040582e-01 -5.16424417e-01
4.32056487e-01 -4.53605764e-02 -3.29732805e-01 -4.92839575e-01
-9.22004044e-01 2.38093957e-01 -1.35462672e-01 -4.62766021e-01
6.84294403e-01 -9.24977779e-01 -3.08045328e-01 3.33918124e-01
-1.77730155e+00 -2.71494269e-01 -4.88090605e-01 5.27816594e-01
-1.02303588e+00 1.85394242e-01 -4.02771145e-01 -8.05026889e-01
-4.56381053e-01 -1.09460592e+00 1.14206910e+00 -1.15932077e-01
-7.11989224e-01 -5.43219268e-01 2.74803787e-01 4.77944940e-01
3.52886975e-01 4.47497852e-02 8.30099404e-01 -7.14147866e-01
-4.97018099e-02 -8.55229720e-02 5.74172437e-02 6.66143179e-01
1.00627109e-01 -2.22284064e-01 -1.21682405e+00 -3.68553698e-02
-2.87764311e-01 -5.06400824e-01 5.28903008e-01 -2.24330500e-01
1.26709354e+00 -8.24046373e-01 9.49917138e-02 1.83023542e-01
8.66511583e-01 1.61048491e-02 8.96121562e-01 1.62210733e-01
4.80955064e-01 4.57253337e-01 5.59970975e-01 4.09615040e-01
8.64469260e-02 9.45147872e-01 3.32779258e-01 2.57631093e-01
-6.53381228e-01 -8.67816985e-01 7.02661991e-01 1.14802802e+00
2.04549432e-01 -6.98716342e-01 -8.09026480e-01 6.42945290e-01
-1.78471053e+00 -9.42066908e-01 -3.59207809e-01 2.06097293e+00
1.24041772e+00 2.89113000e-02 -1.23630285e-01 4.53748524e-01
8.93214285e-01 -2.89098859e-01 -8.32411423e-02 -8.51745427e-01
-5.89944646e-02 2.92098761e-01 -1.34558395e-01 5.33690870e-01
-8.19676459e-01 8.75890493e-01 6.14832163e+00 7.31407225e-01
-1.03198576e+00 2.38368705e-01 5.08001268e-01 1.17280602e-03
-5.25830865e-01 -2.41404340e-01 -1.60385489e-01 8.37106407e-01
1.33901787e+00 6.66198134e-02 6.57670736e-01 4.78857070e-01
7.01220751e-01 2.00668007e-01 -1.29809666e+00 1.10607874e+00
3.80675375e-01 -1.24091947e+00 5.80473125e-01 -5.81978858e-01
6.64369762e-01 -4.29138273e-01 2.19346881e-02 3.28414530e-01
-7.69705838e-03 -1.01234174e+00 1.16394544e+00 7.28580356e-01
9.81489420e-01 -6.10089481e-01 7.83147991e-01 2.47219846e-01
-5.52315533e-01 2.74152812e-02 -1.32432520e-01 -2.61768788e-01
3.55563879e-01 6.19026840e-01 -1.20571315e+00 5.79035521e-01
4.32490081e-01 4.15727496e-01 -5.27939677e-01 1.14127290e+00
-4.75615501e-01 7.66833186e-01 -5.19793481e-02 7.30930120e-02
-7.19061866e-02 1.33817315e-01 8.16791952e-01 1.32946777e+00
8.48028779e-01 -3.19891810e-01 -3.16932231e-01 9.75633979e-01
-3.67187738e-01 1.53216481e-01 -6.23984873e-01 -1.67330459e-01
4.94006157e-01 6.66172981e-01 -8.31264351e-03 -3.96462619e-01
2.95811221e-02 1.62096512e+00 2.67085284e-01 2.67413110e-01
-9.92301762e-01 -4.58465755e-01 5.36094427e-01 -1.22138590e-01
-3.45327482e-02 1.57876208e-01 -3.39913815e-01 -1.23835921e+00
2.39898711e-01 -1.23695683e+00 3.15251239e-02 -1.39732707e+00
-1.00891984e+00 8.29052866e-01 -5.12898564e-01 -1.63585031e+00
-5.42832851e-01 -2.09323511e-01 -5.25080621e-01 6.47209704e-01
-1.43892944e+00 -9.66520131e-01 -2.37678811e-01 2.79934645e-01
9.84985173e-01 -1.57459795e-01 1.08826947e+00 4.96009558e-01
-3.25317293e-01 8.43644321e-01 -1.20595582e-01 2.62614880e-02
9.88259196e-01 -1.06457686e+00 3.33578169e-01 8.62979829e-01
1.29282355e-01 2.07977532e-03 1.21813798e+00 -5.46677589e-01
-1.01401281e+00 -1.56117678e+00 1.38078904e+00 -4.08479422e-01
5.00075758e-01 -2.79975235e-01 -9.41379488e-01 4.41705197e-01
5.23672104e-01 -3.14662576e-01 6.14092410e-01 -4.70607489e-01
-3.29596490e-01 1.65189669e-01 -1.09246445e+00 7.32810616e-01
1.12657261e+00 -5.97663879e-01 -8.26571703e-01 6.34583831e-01
1.14962959e+00 -5.73001504e-01 -5.57528436e-01 2.91469842e-01
4.17388260e-01 -6.63630188e-01 6.03378654e-01 -9.03549373e-01
9.14096236e-01 -2.33181447e-01 -3.83504406e-02 -2.00047088e+00
2.25525722e-01 -9.46034074e-01 -2.14600693e-02 1.39847517e+00
7.00210094e-01 -1.54905319e-01 3.20151210e-01 2.31109709e-01
-7.09313035e-01 -1.05713055e-01 -1.21592379e+00 -9.32151616e-01
-6.72446713e-02 -6.54546201e-01 7.55682707e-01 8.65352094e-01
3.03052425e-01 3.40368062e-01 -7.42784560e-01 9.87801626e-02
2.13412538e-01 -4.36802983e-01 7.65637457e-01 -8.00518811e-01
-2.61806786e-01 -2.55250663e-01 -1.76600724e-01 -6.67829633e-01
4.03184116e-01 -8.39110017e-01 6.24331236e-01 -1.22068048e+00
-2.05830827e-01 -3.38138402e-01 1.60589680e-01 4.74281222e-01
-1.90618113e-01 5.19503951e-01 3.53518575e-01 3.06547638e-02
-4.79658306e-01 8.49138975e-01 1.11474609e+00 -4.62787211e-01
-3.10764492e-01 -4.48246934e-02 -3.87656569e-01 4.35874283e-01
9.73752797e-01 -7.21433103e-01 -1.00059777e-01 -7.09646046e-01
4.28093463e-01 1.76140606e-01 3.97868633e-01 -1.33813512e+00
-1.47330225e-01 9.43948850e-02 -1.79834627e-02 -1.47554010e-01
4.37701523e-01 -8.16557527e-01 2.93368816e-01 4.47261870e-01
-9.20870006e-01 -4.09583412e-02 2.06318825e-01 3.73160541e-01
-4.79165167e-01 -4.79842871e-01 4.95005190e-01 1.03500180e-01
-1.15846798e-01 -4.14935261e-01 -7.27916598e-01 2.72923876e-02
6.10772610e-01 5.44890463e-02 -2.97813654e-01 -7.47205853e-01
-1.01559126e+00 -1.57677025e-01 2.39001125e-01 6.97363257e-01
7.76228189e-01 -1.63012171e+00 -1.12805164e+00 1.47666320e-01
3.23286504e-01 -2.80013531e-01 6.38384968e-02 6.95195079e-01
-4.16482329e-01 3.56176496e-01 -5.66832349e-02 -4.55055088e-01
-1.21742225e+00 3.60604465e-01 4.63206053e-01 -4.19811085e-02
-4.18384790e-01 6.96585894e-01 -5.50180674e-01 -3.96093875e-01
1.90903321e-01 -4.58339214e-01 -1.43579736e-01 -1.51786476e-01
6.32601678e-01 2.59856701e-01 4.93335485e-01 -5.40362537e-01
-1.83006134e-02 -3.77054722e-03 1.97170779e-01 -5.18199682e-01
1.06022489e+00 -1.26335263e-01 2.35991344e-01 4.04097527e-01
1.10090673e+00 4.47475277e-02 -8.81073117e-01 1.16391055e-01
-2.19294414e-01 -4.78159875e-01 -2.68175036e-01 -1.20419419e+00
-4.84732449e-01 8.69419217e-01 6.20767176e-01 2.49904722e-01
1.00421536e+00 -3.02838564e-01 9.83577609e-01 4.17693585e-01
1.14311844e-01 -1.18655014e+00 2.25533843e-01 6.47812843e-01
1.39289689e+00 -1.04045439e+00 -7.39212990e-01 -4.20932680e-01
-7.06601799e-01 1.10527694e+00 5.49834669e-01 3.52664858e-01
-1.99813485e-01 -1.42604142e-01 2.26611286e-01 5.73589444e-01
-7.77764440e-01 5.38808107e-02 9.59249735e-02 8.90724182e-01
3.53694707e-01 1.69798061e-01 -3.58185679e-01 5.70382953e-01
-5.71454525e-01 3.22909027e-01 9.03660715e-01 7.08704233e-01
-2.11209938e-01 -1.19084167e+00 -6.49452567e-01 1.38427973e-01
-1.61388919e-01 -2.90903568e-01 -6.82763219e-01 1.16590276e-01
1.12271436e-01 1.27007389e+00 9.51562002e-02 -5.77698588e-01
4.81498659e-01 4.80991423e-01 3.74537736e-01 -5.18850803e-01
-6.61140442e-01 -3.46979737e-01 5.27017653e-01 -4.19291973e-01
-2.65740842e-01 -5.80225110e-01 -1.02563274e+00 -3.81367141e-03
-2.87399977e-01 3.84208113e-01 7.31131494e-01 1.02710164e+00
7.04701900e-01 9.08194959e-01 7.59098887e-01 -6.01553619e-01
-6.03880167e-01 -1.27621126e+00 6.27179444e-02 8.30907643e-01
3.44008327e-01 -1.74084395e-01 -4.12883282e-01 4.27769691e-01] | [15.2745943069458, 4.896003246307373] |
0d2cb716-5b1e-4ef4-838d-1d7c10f42c9e | on-strengthening-and-defending-graph | 2306.09104 | null | https://arxiv.org/abs/2306.09104v1 | https://arxiv.org/pdf/2306.09104v1.pdf | On Strengthening and Defending Graph Reconstruction Attack with Markov Chain Approximation | Although powerful graph neural networks (GNNs) have boosted numerous real-world applications, the potential privacy risk is still underexplored. To close this gap, we perform the first comprehensive study of graph reconstruction attack that aims to reconstruct the adjacency of nodes. We show that a range of factors in GNNs can lead to the surprising leakage of private links. Especially by taking GNNs as a Markov chain and attacking GNNs via a flexible chain approximation, we systematically explore the underneath principles of graph reconstruction attack, and propose two information theory-guided mechanisms: (1) the chain-based attack method with adaptive designs for extracting more private information; (2) the chain-based defense method that sharply reduces the attack fidelity with moderate accuracy loss. Such two objectives disclose a critical belief that to recover better in attack, you must extract more multi-aspect knowledge from the trained GNN; while to learn safer for defense, you must forget more link-sensitive information in training GNNs. Empirically, we achieve state-of-the-art results on six datasets and three common GNNs. The code is publicly available at: https://github.com/tmlr-group/MC-GRA. | ['Bo Han', 'Quanming Yao', 'Jiangchao Yao', 'Xuan Li', 'Chenyu Zhou', 'Zhanke Zhou'] | 2023-06-15 | null | null | null | null | ['graph-reconstruction'] | ['graphs'] | [ 3.86933297e-01 5.49390674e-01 -4.49465066e-01 1.15122475e-01
-5.57967603e-01 -9.25346017e-01 2.67844260e-01 -1.42774852e-02
-1.87418722e-02 5.35362601e-01 2.31357008e-01 -9.79444623e-01
-2.71541864e-01 -1.06740880e+00 -9.89507139e-01 -7.35889256e-01
-4.74585801e-01 7.59593770e-02 -4.87746000e-02 -3.24583679e-01
1.28382877e-01 7.53533661e-01 -4.84912306e-01 3.08453366e-02
6.51157737e-01 7.44127870e-01 -6.51170969e-01 5.15202224e-01
3.80076081e-01 6.86504543e-01 -2.88287789e-01 -1.22198927e+00
8.53010714e-01 -4.27425027e-01 -6.08209133e-01 -3.88616592e-01
3.38063359e-01 -6.13858759e-01 -1.13884997e+00 1.63054872e+00
4.53355432e-01 -2.75778830e-01 3.52361113e-01 -1.42946601e+00
-7.43274271e-01 1.27752340e+00 -6.83329821e-01 8.62454027e-02
1.01565093e-01 2.66991377e-01 1.27811658e+00 -2.64816850e-01
4.65250939e-01 1.10248101e+00 7.75878131e-01 6.67499244e-01
-1.36624002e+00 -1.11478639e+00 3.75197344e-02 1.36529431e-01
-1.46833277e+00 -4.95466739e-01 1.08558297e+00 1.63253203e-01
3.75003099e-01 4.56177592e-01 5.81047297e-01 1.60401118e+00
3.96118969e-01 5.81540883e-01 9.05216217e-01 -7.94013143e-02
6.96273297e-02 8.51865113e-02 9.16476399e-02 7.71512449e-01
9.87151802e-01 6.03516459e-01 -4.39998031e-01 -6.38349533e-01
6.68336630e-01 2.44055778e-01 -7.04888999e-01 -6.52939618e-01
-4.59066361e-01 1.12178946e+00 6.89997971e-01 -2.73284651e-02
-1.25747696e-01 4.80043709e-01 3.75192344e-01 5.38537800e-01
2.73165464e-01 3.42989028e-01 -3.57724398e-01 4.78309423e-01
-5.21646738e-01 1.91075064e-03 1.32646739e+00 7.56219208e-01
5.27779102e-01 1.50083706e-01 2.55225211e-01 -3.32940333e-02
3.30761254e-01 4.19809818e-01 -3.21352512e-01 -6.39226735e-01
6.36353135e-01 3.02718818e-01 -5.05988896e-01 -1.46342742e+00
-2.18382284e-01 -8.59456122e-01 -1.21282339e+00 -2.50316318e-02
4.77345496e-01 -4.88285035e-01 -5.85270882e-01 1.99269319e+00
2.14486763e-01 2.51359046e-01 1.90272897e-01 6.65748954e-01
5.07810950e-01 4.05781090e-01 -1.80452764e-01 -4.32200469e-02
1.19505143e+00 -6.59999251e-01 -5.38595557e-01 -2.07443789e-01
7.44361341e-01 -1.94100216e-01 5.99952221e-01 4.84997749e-01
-9.50827777e-01 2.37776130e-01 -1.31859112e+00 1.21585295e-01
-1.98400408e-01 -5.37990868e-01 9.80905116e-01 1.34037924e+00
-1.11936128e+00 8.32166314e-01 -5.97908914e-01 -3.97457406e-02
8.34366858e-01 3.70394915e-01 -4.78016675e-01 -3.06854308e-01
-1.62927043e+00 4.59251463e-01 2.49338135e-01 1.30636245e-01
-1.19513381e+00 -7.94279456e-01 -6.59540236e-01 2.18905851e-01
7.99600422e-01 -6.50922060e-01 6.74497426e-01 -5.44251204e-01
-1.27282488e+00 4.88928467e-01 6.25422239e-01 -9.38549757e-01
5.02291739e-01 -1.39062908e-02 -5.66315949e-01 4.92803216e-01
-4.40521568e-01 5.48907258e-02 1.00617707e+00 -1.37038732e+00
9.51489657e-02 -6.63135529e-01 1.93036348e-01 -1.42521635e-01
-5.82187653e-01 -2.28089288e-01 -1.60829410e-01 -6.63969696e-01
2.10131183e-02 -1.07812881e+00 -3.87786299e-01 2.54771143e-01
-1.02170074e+00 3.53147417e-01 8.88971448e-01 -6.58223152e-01
1.21156847e+00 -2.01643682e+00 -3.20828892e-02 7.48261750e-01
9.55690980e-01 4.94582862e-01 -2.69942135e-01 7.80516326e-01
-1.15685932e-01 4.44468677e-01 -1.92299604e-01 -1.13861285e-01
2.34888680e-02 -1.08710974e-01 -7.69212723e-01 8.11084628e-01
-3.16715896e-01 1.08964074e+00 -6.29836977e-01 4.12687808e-02
-2.54175544e-01 6.38445139e-01 -7.59455442e-01 -3.67615446e-02
-6.65854812e-02 2.45900229e-01 -5.62471330e-01 6.81919575e-01
1.05619299e+00 -4.03875500e-01 6.06217206e-01 -3.98335516e-01
6.99282706e-01 2.11796254e-01 -8.92167509e-01 1.30745399e+00
1.97304204e-01 4.47175145e-01 3.59984368e-01 -8.91994476e-01
7.39601612e-01 1.21441133e-01 2.16346607e-01 -4.00052249e-01
3.70537490e-01 -8.88731778e-02 1.45737588e-01 -7.97937810e-02
1.07578322e-01 -5.18134236e-03 -1.73753083e-01 7.15352774e-01
-1.43688485e-01 3.39741170e-01 -4.95727867e-01 7.71151841e-01
1.63039052e+00 -5.72083950e-01 1.91265494e-01 -1.96844384e-01
4.64008236e-03 -5.24252772e-01 5.78371346e-01 9.76890147e-01
-2.99609959e-01 1.52277112e-01 1.04621255e+00 -3.90604049e-01
-8.10487688e-01 -1.30808473e+00 3.02614689e-01 6.97565198e-01
2.56797612e-01 -8.00965846e-01 -7.66924977e-01 -1.09203196e+00
1.48703992e-01 7.71322966e-01 -6.20358884e-01 -6.91464603e-01
-2.63857424e-01 -8.50737691e-01 1.33346105e+00 1.25861332e-01
6.72372818e-01 -3.77676219e-01 5.29531091e-02 -2.64441669e-01
-2.12384257e-02 -9.37005103e-01 -6.35970235e-01 1.18146703e-01
-8.22562635e-01 -1.12846875e+00 -5.96268065e-02 -1.88530371e-01
8.21108758e-01 4.39560533e-01 8.55936050e-01 4.79399443e-01
-6.62488192e-02 3.40751559e-01 -1.85934797e-01 -2.64073372e-01
-4.61086601e-01 1.44270629e-01 -2.51336191e-02 -1.34249078e-02
1.72213897e-01 -1.18451548e+00 -6.12067759e-01 8.55142996e-02
-9.22572017e-01 -2.42745832e-01 8.42347920e-01 4.81385350e-01
8.48152414e-02 2.34888867e-01 1.93579897e-01 -1.34505701e+00
7.84665048e-01 -6.15394771e-01 -5.66544116e-01 2.20076904e-01
-8.44945967e-01 2.06839353e-01 7.07384706e-01 -5.17979145e-01
-4.26254570e-01 -2.36970484e-01 -2.51302660e-01 -5.15891790e-01
3.62664968e-01 3.16301316e-01 -4.99330699e-01 -7.15438545e-01
7.64847338e-01 1.29855305e-01 1.36714876e-01 -1.80173457e-01
6.27814353e-01 -5.14490791e-02 3.27785581e-01 -6.73238635e-01
1.43881595e+00 4.87345099e-01 5.01023054e-01 -4.86903071e-01
-6.57959223e-01 3.27172488e-01 -2.95423754e-02 1.00172367e-02
4.45063353e-01 -7.28102863e-01 -1.07544947e+00 3.61615747e-01
-9.74996805e-01 -2.87519693e-01 -1.29067361e-01 9.20458809e-02
-2.25993674e-02 8.15756440e-01 -9.52357650e-01 -6.87064588e-01
-7.35551238e-01 -8.98299336e-01 2.89720714e-01 1.24450952e-01
2.76839584e-01 -1.11544728e+00 -1.33133262e-01 3.13001573e-01
4.24645513e-01 4.79875892e-01 8.84744525e-01 -1.08924901e+00
-9.48649466e-01 -4.77863222e-01 -4.37105477e-01 3.32203716e-01
-2.43551478e-01 -3.11035395e-01 -9.66866612e-01 -7.47971117e-01
1.62511483e-01 -2.66067743e-01 9.38880444e-01 6.86912611e-02
1.33441925e+00 -1.07138860e+00 -3.68946165e-01 1.27371156e+00
1.49581206e+00 -1.24776557e-01 9.31771338e-01 -1.86997831e-01
1.04426384e+00 4.29594189e-01 -2.42396727e-01 3.25560898e-01
2.84606993e-01 9.72573683e-02 8.75165641e-01 1.43838366e-02
9.84479785e-02 -9.34031963e-01 3.11134994e-01 5.04324257e-01
1.44529998e-01 -7.14705408e-01 -6.63188636e-01 -1.36300027e-02
-1.56008935e+00 -9.40257847e-01 7.60943256e-03 2.08264804e+00
6.71210408e-01 4.05076534e-01 1.18862791e-02 1.07422676e-02
7.11493671e-01 6.45496488e-01 -6.80563450e-01 -1.30519092e-01
-2.49553006e-02 3.12424973e-02 1.17614591e+00 5.84389508e-01
-8.01192522e-01 8.14959645e-01 6.02321005e+00 9.64739621e-01
-7.90913761e-01 1.30873611e-02 8.19582641e-01 1.07607014e-01
-7.82985806e-01 4.91799623e-01 -7.98764646e-01 2.80350834e-01
9.83822227e-01 -3.29666585e-01 8.83728743e-01 7.00924635e-01
-5.16502738e-01 5.14894426e-01 -8.13479483e-01 6.85628235e-01
5.90193495e-02 -1.55270481e+00 3.17204356e-01 7.41355240e-01
3.87684166e-01 -1.98796526e-01 2.95657128e-01 4.03316244e-02
7.85221398e-01 -1.03870213e+00 1.33675411e-01 3.67713094e-01
7.58305669e-01 -1.12982023e+00 3.31336796e-01 2.14649975e-01
-9.47025716e-01 -1.79418728e-01 -4.14381415e-01 1.51135206e-01
-5.03448583e-02 8.02565753e-01 -5.87741196e-01 8.21269453e-01
3.70934695e-01 4.73277330e-01 -5.19166410e-01 6.14048123e-01
-5.14395475e-01 8.99585307e-01 -3.08484435e-01 -2.25023855e-03
9.33826864e-02 -3.45275283e-01 9.62184966e-01 7.33665705e-01
7.67520294e-02 2.16940641e-01 -9.42726284e-02 1.01017749e+00
-6.50587201e-01 -3.89687300e-01 -1.00105906e+00 -3.54654342e-01
8.12063217e-01 1.25597727e+00 -4.95875031e-01 2.72300541e-01
-7.90151488e-03 8.98215294e-01 4.49613184e-01 4.70362276e-01
-7.58872509e-01 -5.46718776e-01 7.01299667e-01 1.89435691e-01
1.84046850e-01 -1.71404734e-01 -3.29204381e-01 -1.16750479e+00
-1.62585288e-01 -1.19843996e+00 5.76353252e-01 -2.00344890e-01
-1.48351097e+00 5.03178000e-01 -1.90567166e-01 -8.57159078e-01
1.38649613e-01 -3.84270579e-01 -8.44507575e-01 5.50988197e-01
-1.23514366e+00 -1.20011163e+00 1.65827483e-01 7.42061138e-01
-3.54939789e-01 -2.53804624e-01 7.03347862e-01 2.67082006e-01
-7.43882358e-01 1.29724503e+00 -4.64578345e-02 5.74102759e-01
4.26568091e-01 -7.04311371e-01 7.35492647e-01 1.29752254e+00
2.59302765e-01 8.94365907e-01 7.11826921e-01 -8.62974584e-01
-2.03299594e+00 -9.50069427e-01 3.65015179e-01 -3.95063102e-01
8.80094945e-01 -6.95709586e-01 -8.62781644e-01 9.67088938e-01
-5.37834829e-03 2.07382932e-01 8.27344298e-01 4.37969305e-02
-1.03790021e+00 -2.60066658e-01 -1.58205056e+00 8.32807779e-01
1.40813470e+00 -7.11121976e-01 1.11097999e-01 2.98203260e-01
1.20054126e+00 -3.52737844e-01 -6.99953020e-01 2.28757069e-01
5.30666769e-01 -1.08609831e+00 1.24728680e+00 -6.48072124e-01
1.56251833e-01 1.01521514e-01 -2.82929957e-01 -1.14333630e+00
-2.36220300e-01 -1.32113290e+00 -5.88395953e-01 1.00988472e+00
4.65767503e-01 -1.16441619e+00 1.16422737e+00 6.17342591e-01
3.63206744e-01 -8.17959726e-01 -7.99811065e-01 -6.95600986e-01
1.11501358e-01 -2.53641814e-01 7.39186347e-01 1.00333095e+00
-2.19223380e-01 1.99001566e-01 -8.18115473e-01 5.84264576e-01
1.26804531e+00 -3.05850238e-01 7.91992903e-01 -1.12186778e+00
-4.65911150e-01 -2.36010656e-01 -3.89301956e-01 -1.00428045e+00
1.87012479e-01 -1.04541433e+00 -5.58039069e-01 -9.47086155e-01
1.93643495e-01 -2.05204919e-01 -2.08121210e-01 5.06869316e-01
1.54526949e-01 1.35230228e-01 2.50059217e-01 4.29023802e-02
-3.86845052e-01 4.57366556e-01 1.06244075e+00 -6.16066568e-02
1.02763034e-01 3.57710898e-01 -1.71515965e+00 3.90847862e-01
9.85098481e-01 -7.40769982e-01 -8.00529599e-01 -1.00180820e-01
6.71387315e-01 3.12985450e-01 6.56902254e-01 -7.08690822e-01
3.74139726e-01 4.23485786e-02 1.98602170e-01 -3.85594040e-01
1.89916357e-01 -9.29902732e-01 3.81017238e-01 8.17566514e-01
-3.96980613e-01 -4.74405400e-02 -1.83068737e-02 1.10624588e+00
5.28932989e-01 7.23662451e-02 7.73027182e-01 5.06523922e-02
-6.07133284e-02 8.54268014e-01 3.73923182e-02 1.21339440e-01
7.42565751e-01 5.94259277e-02 -8.93458903e-01 -8.78238618e-01
-3.40747684e-01 2.33139526e-02 4.48529541e-01 3.19324695e-02
7.53450155e-01 -1.25318444e+00 -6.45781100e-01 3.39569449e-01
-2.95613855e-01 -4.28853065e-01 2.72180647e-01 6.79314315e-01
-3.09387356e-01 9.90075525e-03 -1.01518542e-01 8.42330419e-03
-9.89571273e-01 9.37806845e-01 3.26397240e-01 -5.16699970e-01
-8.35008025e-01 9.08319950e-01 1.20590046e-01 -3.73324871e-01
5.09241104e-01 3.64431679e-01 2.85666853e-01 -3.87037307e-01
5.73013246e-01 3.69115442e-01 -2.63046503e-01 -1.58217400e-01
-1.92888185e-01 4.38420549e-02 -4.19205755e-01 6.86578229e-02
1.22066140e+00 -1.15771830e-01 -1.98232695e-01 -3.63602668e-01
1.39701569e+00 3.48702848e-01 -1.24597466e+00 -4.02905881e-01
-3.09226602e-01 -7.11202979e-01 1.24175720e-01 -5.93457341e-01
-1.56819713e+00 7.15311587e-01 2.11021513e-01 5.29932618e-01
1.17856538e+00 -1.15529224e-01 1.03195369e+00 5.29379070e-01
5.68046391e-01 -4.77819055e-01 1.67433068e-01 2.07569748e-01
5.79731643e-01 -6.22274280e-01 3.43154460e-01 -6.58863187e-01
-5.03785193e-01 8.60449672e-01 4.02007997e-01 -2.55968988e-01
9.78201210e-01 2.78046995e-01 -4.98321384e-01 -3.63069445e-01
-7.52455354e-01 4.10589993e-01 1.66885760e-02 7.87727714e-01
-2.87312984e-01 8.61560628e-02 8.45777169e-02 9.27948833e-01
-1.83620989e-01 -5.29128850e-01 7.05962658e-01 6.24279618e-01
-2.08901152e-01 -1.17976451e+00 -4.23988700e-02 5.19677579e-01
-7.22320020e-01 -3.37494403e-01 -7.20961869e-01 5.57730436e-01
-5.03306925e-01 7.77341783e-01 -6.12549186e-01 -1.00116980e+00
6.88777305e-03 -3.83357376e-01 3.24241966e-01 -1.80210307e-01
-4.90132898e-01 -3.86420041e-01 2.86774755e-01 -8.65208626e-01
4.01096195e-01 -3.99718791e-01 -7.81420350e-01 -1.16824746e+00
-3.63285780e-01 9.51780453e-02 4.41253483e-01 5.19538760e-01
6.64289355e-01 8.15531388e-02 9.86144900e-01 -4.32375908e-01
-9.21470046e-01 -3.10410887e-01 -7.79447615e-01 8.85199197e-03
5.00863075e-01 -2.68803626e-01 -9.33809578e-01 -7.89486825e-01] | [6.034623622894287, 7.1940598487854] |
6cb8cd8b-dfec-4fb0-9dde-22c2b8c67a62 | geometric-graph-filters-and-neural-networks | 2305.18467 | null | https://arxiv.org/abs/2305.18467v2 | https://arxiv.org/pdf/2305.18467v2.pdf | Geometric Graph Filters and Neural Networks: Limit Properties and Discriminability Trade-offs | This paper studies the relationship between a graph neural network (GNN) and a manifold neural network (MNN) when the graph is constructed from a set of points sampled from the manifold, thus encoding geometric information. We consider convolutional MNNs and GNNs where the manifold and the graph convolutions are respectively defined in terms of the Laplace-Beltrami operator and the graph Laplacian. Using the appropriate kernels, we analyze both dense and moderately sparse graphs. We prove non-asymptotic error bounds showing that convolutional filters and neural networks on these graphs converge to convolutional filters and neural networks on the continuous manifold. As a byproduct of this analysis, we observe an important trade-off between the discriminability of graph filters and their ability to approximate the desired behavior of manifold filters. We then discuss how this trade-off is ameliorated in neural networks due to the frequency mixing property of nonlinearities. We further derive a transferability corollary for geometric graphs sampled from the same manifold. We validate our results numerically on a navigation control problem and a point cloud classification task. | ['Alejandro Ribeiro', 'Luana Ruiz', 'Zhiyang Wang'] | 2023-05-29 | null | null | null | null | ['point-cloud-classification'] | ['computer-vision'] | [-1.53659225e-01 6.13921821e-01 2.41501927e-01 -2.78935977e-03
7.70320520e-02 -6.07477486e-01 4.33541387e-01 2.36092940e-01
-2.06464782e-01 2.31923386e-01 -3.85091780e-03 -2.18319580e-01
-4.19070095e-01 -9.99825656e-01 -1.32544315e+00 -6.26466393e-01
-6.22111440e-01 2.83365026e-02 -1.01229131e-01 -2.81215638e-01
5.15549863e-03 7.31438279e-01 -1.43000531e+00 -3.90464783e-01
1.06174970e+00 1.01532567e+00 -3.82772856e-03 9.67983246e-01
2.71236487e-02 5.89640379e-01 -2.49813423e-01 -7.17555806e-02
6.23027265e-01 -3.82863969e-01 -6.08662784e-01 1.73359618e-01
1.09542775e+00 1.67125240e-01 -8.96323740e-01 1.67173517e+00
5.75632267e-02 5.19776583e-01 8.46255600e-01 -1.36453092e+00
-1.23353159e+00 3.51243436e-01 -8.65672827e-02 1.56530142e-01
-2.61955857e-01 -1.21092550e-01 8.95290017e-01 -7.40942061e-01
5.60537398e-01 1.31229162e+00 9.61174905e-01 3.99700314e-01
-1.31715810e+00 -4.12413627e-01 1.17812343e-01 -1.93913832e-01
-1.14607370e+00 -3.17483433e-02 6.41028941e-01 -7.33448505e-01
4.84347075e-01 1.54770818e-02 8.31758857e-01 5.38072169e-01
3.59521508e-01 1.47860169e-01 4.92224604e-01 -3.92177671e-01
2.13715822e-01 -8.43041576e-03 3.83558869e-01 1.28817534e+00
5.05928516e-01 1.91382378e-01 -1.12792306e-01 -1.12372331e-01
1.41060781e+00 1.30189732e-01 -8.56414139e-01 -6.76046431e-01
-8.78176749e-01 1.12191272e+00 1.08474016e+00 3.38637531e-01
-3.62178832e-01 6.23208702e-01 -1.05012171e-01 5.76408565e-01
3.49577636e-01 3.33781391e-01 1.57739386e-01 5.11286199e-01
-2.96556801e-01 -1.02375776e-01 1.02637768e+00 1.20396233e+00
1.02288640e+00 2.37493128e-01 1.72787309e-01 3.09146583e-01
2.71609038e-01 5.78585207e-01 5.86817190e-02 -8.20509195e-01
2.35983476e-01 6.01754725e-01 -1.93608373e-01 -1.24571538e+00
-5.03585219e-01 -7.06150830e-01 -1.09102809e+00 3.47192585e-01
7.42299974e-01 -2.19332740e-01 -8.97148907e-01 2.05533934e+00
-8.17151964e-02 3.93709093e-01 -8.47875327e-02 9.00080144e-01
4.21539247e-01 5.85453212e-01 -8.69109854e-02 2.70064294e-01
9.22666728e-01 -5.14128745e-01 -5.10819972e-01 1.33081198e-01
5.88069379e-01 -3.53373885e-02 9.16667819e-01 -1.20038226e-01
-1.12460196e+00 -6.42531693e-01 -1.26898253e+00 -1.81643814e-01
-5.36194265e-01 1.12531051e-01 2.87006676e-01 3.96843642e-01
-1.53256738e+00 1.16072595e+00 -9.30267751e-01 -6.30420804e-01
2.66162395e-01 3.51268649e-01 -3.86762798e-01 2.61803061e-01
-9.47782874e-01 6.62939489e-01 1.20610654e-01 3.77572685e-01
-5.27238250e-01 -8.88002336e-01 -9.67727244e-01 3.80924612e-01
-2.00586021e-01 -7.85809994e-01 7.91072190e-01 -1.22084701e+00
-1.17822182e+00 5.99311888e-01 4.09722120e-01 -6.41340971e-01
3.89938742e-01 -6.63434789e-02 -3.72875452e-01 4.05568302e-01
-1.30759776e-01 4.14892972e-01 9.05564427e-01 -8.64682496e-01
-4.28528875e-01 -4.48814154e-01 3.70109230e-01 5.98630160e-02
-3.58238697e-01 -8.18245590e-01 1.21829592e-01 -4.34950858e-01
7.38277063e-02 -1.07112610e+00 -2.98281372e-01 3.30253571e-01
-3.91758710e-01 5.71493246e-02 8.51655006e-01 -4.87683326e-01
7.46055067e-01 -2.42158556e+00 4.31694776e-01 5.58722913e-01
6.43527448e-01 -8.93444195e-03 -3.15304607e-01 5.02909958e-01
-3.31802577e-01 -2.14980394e-02 -1.32745594e-01 6.33941814e-02
-1.82345957e-02 2.06537768e-02 -2.73907363e-01 1.06516242e+00
4.11552280e-01 9.29527164e-01 -9.61156726e-01 2.40593463e-01
2.59859502e-01 7.28674650e-01 -6.49680555e-01 9.92062911e-02
-1.58634260e-01 4.58862603e-01 -4.12108630e-01 -1.53148502e-01
5.88917017e-01 -6.41996086e-01 -1.66027874e-01 -2.90238857e-01
1.38824552e-01 -1.96188554e-01 -1.03896546e+00 1.58202112e+00
-4.42979157e-01 8.55985701e-01 4.24740434e-01 -1.40148282e+00
8.16014171e-01 6.91571087e-02 1.73398241e-01 -2.61438102e-01
2.41865024e-01 2.38751769e-01 3.63827944e-02 -2.95260906e-01
1.77083656e-01 -1.59332350e-01 2.50685275e-01 -7.22669214e-02
4.72171128e-01 2.49668315e-01 -5.15784919e-02 2.25477234e-01
1.17353904e+00 -4.57579702e-01 -7.69577548e-02 -1.12099910e+00
4.79979634e-01 -1.90529391e-01 -1.11303002e-01 8.36921930e-01
1.25578806e-01 3.87432694e-01 6.58800125e-01 -3.50341588e-01
-9.20885742e-01 -1.39322031e+00 -7.96467997e-03 5.92006505e-01
1.20487653e-01 -6.85865283e-02 -8.38908017e-01 -4.28976864e-01
2.31581599e-01 4.04381663e-01 -9.35629547e-01 -6.36999786e-01
-4.79473919e-01 -8.79554749e-02 3.08120519e-01 4.42214906e-01
5.74025810e-01 -4.74216223e-01 -5.12334645e-01 3.18326242e-02
3.37689459e-01 -1.04279566e+00 -7.01290667e-01 1.28371045e-01
-9.51977074e-01 -1.35522580e+00 -7.36405790e-01 -1.22952437e+00
9.52860236e-01 2.50208765e-01 1.03888345e+00 1.43486038e-01
2.37699877e-02 9.52499807e-01 1.44593909e-01 -1.67755410e-01
-5.23511589e-01 9.46532935e-02 -9.46738794e-02 4.07896519e-01
-3.92863937e-02 -8.89922976e-01 -5.86066186e-01 5.80807216e-02
-9.99806643e-01 -9.81286243e-02 2.35650539e-01 7.54299283e-01
2.07143664e-01 -2.00515445e-02 2.98888683e-01 -7.06878424e-01
6.51629448e-01 -5.38664639e-01 -1.15886402e+00 1.15690172e-01
-2.89003551e-01 6.25547051e-01 9.54792678e-01 -4.96020526e-01
-2.79965192e-01 1.49744347e-01 3.37598383e-01 -7.76908696e-01
1.23270884e-01 4.50803190e-01 -5.37525490e-02 -7.77155578e-01
7.82714844e-01 -1.17210656e-01 4.68698263e-01 -1.75307021e-01
7.46100843e-01 1.18824653e-01 7.21127748e-01 -3.07443023e-01
1.03708541e+00 6.48764312e-01 4.49591279e-01 -1.39691257e+00
-5.16358912e-01 -3.01282555e-01 -4.57980305e-01 -4.45696473e-01
9.66993868e-01 -6.73645258e-01 -9.13830817e-01 2.29555979e-01
-9.59087014e-01 -5.02013564e-01 -6.05266571e-01 6.16825581e-01
-7.79147863e-01 1.52684376e-01 -8.50917459e-01 -7.25374520e-01
8.13188702e-02 -7.22873688e-01 8.42290580e-01 3.24570000e-01
3.32872897e-01 -1.63319349e+00 -1.83379557e-02 -7.47466028e-01
5.02799094e-01 6.52134001e-01 1.12750816e+00 -4.44522381e-01
-8.11220288e-01 -1.95596263e-01 -3.79707187e-01 4.17908072e-01
-6.47302391e-03 -2.59226680e-01 -6.41088843e-01 -4.20748115e-01
1.42759532e-01 1.72438160e-01 9.65679884e-01 7.76582539e-01
7.52498865e-01 -4.20963198e-01 -1.51713282e-01 9.24148142e-01
1.83503377e+00 -1.40684336e-01 4.09055799e-01 -1.22854978e-01
9.40893233e-01 4.82592523e-01 -3.71545285e-01 -1.69533461e-01
-6.69498518e-02 1.68924674e-01 6.59137368e-01 -7.29168728e-02
-6.80491403e-02 -3.50933045e-01 2.56459951e-01 8.38883936e-01
-2.15161532e-01 -6.38144687e-02 -7.79536724e-01 5.31007051e-01
-1.82990324e+00 -5.63211799e-01 -2.54063934e-01 2.33415294e+00
-3.23112705e-03 -1.06767975e-01 1.03059247e-01 -1.40153155e-01
1.14947605e+00 -1.00072604e-02 -5.17767727e-01 -1.23950340e-01
-2.23371044e-01 3.77753109e-01 9.88883853e-01 8.59324753e-01
-1.10512972e+00 4.18660313e-01 6.22731733e+00 2.97732830e-01
-1.15431762e+00 -9.93533283e-02 2.46035576e-01 3.91981483e-01
-1.63618952e-01 -2.27696300e-01 -2.71231174e-01 1.70859337e-01
1.08673251e+00 -3.85601133e-01 6.86991036e-01 7.48669267e-01
-2.23500088e-01 4.74078745e-01 -1.24873841e+00 9.12563860e-01
-1.56353593e-01 -1.63450670e+00 2.06502780e-01 4.63450909e-01
5.38641393e-01 1.60471067e-01 3.18483442e-01 6.90984279e-02
4.56404239e-01 -1.15859532e+00 7.73849845e-01 6.45335257e-01
6.16129577e-01 -6.41257107e-01 1.67794749e-01 5.50335087e-02
-1.38952148e+00 -1.39161572e-01 -5.12529135e-01 -2.86207505e-04
-7.77186379e-02 4.07869518e-01 -3.29956025e-01 6.68345273e-01
2.67803162e-01 9.97472644e-01 -4.51895624e-01 1.15309405e+00
1.65232837e-01 4.04679835e-01 -3.46685857e-01 -1.49963886e-01
5.42538643e-01 -8.25453401e-01 6.95803046e-01 9.47459877e-01
4.87069190e-01 1.21723071e-01 7.12108165e-02 1.33634746e+00
-1.15816422e-01 -1.22376457e-01 -9.30886865e-01 -2.18677685e-01
-3.98396365e-02 1.14530957e+00 -8.41648281e-01 -7.13056475e-02
-6.12480104e-01 7.67803907e-01 6.38719559e-01 9.16760087e-01
-6.10777736e-01 -6.50831461e-01 8.28590989e-01 2.86485225e-01
3.46661001e-01 -4.83458042e-01 6.53636605e-02 -1.15205240e+00
5.20873219e-02 -1.60960063e-01 1.57756001e-01 -5.91736138e-01
-1.30819917e+00 5.25552928e-01 -1.40615940e-01 -1.07683814e+00
6.93467185e-02 -1.03915322e+00 -5.59367836e-01 1.00995946e+00
-1.15405726e+00 -7.41174102e-01 -4.23811078e-01 7.07479298e-01
-3.46877903e-01 1.28314257e-01 5.10482371e-01 2.12277547e-01
-1.25279024e-01 2.66632289e-01 1.52991384e-01 6.48908019e-01
-2.70694494e-01 -1.53648186e+00 7.19071627e-01 8.61325383e-01
1.42359018e-01 6.26929700e-01 5.44980884e-01 -4.90510315e-01
-1.68914652e+00 -1.50227153e+00 2.40610555e-01 -3.40977937e-01
9.54693139e-01 -4.60122555e-01 -1.10032177e+00 1.01317668e+00
1.04275912e-01 3.63693058e-01 1.48582265e-01 -1.44315034e-01
-3.86372656e-01 2.63972819e-01 -9.08927500e-01 5.57865977e-01
1.29050410e+00 -6.55963421e-01 -2.64850140e-01 4.42706823e-01
7.97129929e-01 -4.01614845e-01 -8.11875403e-01 9.02952775e-02
1.97724283e-01 -7.73279905e-01 8.08527052e-01 -9.57999349e-01
1.96288347e-01 -1.42591715e-01 -2.61754483e-01 -1.65521836e+00
-5.51488400e-01 -6.56862617e-01 9.07658425e-04 4.35443223e-01
2.37772450e-01 -1.11838293e+00 6.84893250e-01 2.42152825e-01
-1.79786608e-01 -3.73052686e-01 -8.36606860e-01 -1.08279622e+00
3.92948985e-01 6.31482974e-02 2.04115286e-01 8.74324143e-01
-1.76899042e-02 4.41209465e-01 1.23653196e-01 6.36453688e-01
8.64273727e-01 -2.06910759e-01 3.53453726e-01 -1.42658043e+00
-2.14426562e-01 -6.38856828e-01 -1.08355379e+00 -1.18311632e+00
2.20673680e-01 -1.23382831e+00 -1.49291739e-01 -1.21409774e+00
-4.81517851e-01 -2.25618213e-01 -7.76809901e-02 -2.67917067e-01
1.24577895e-01 -7.12554902e-02 1.38202846e-01 -1.20364740e-01
-2.78853569e-02 4.36152786e-01 1.26952338e+00 -1.60263162e-02
-1.46108329e-01 -3.54645438e-02 -3.63498449e-01 5.36007643e-01
5.93355060e-01 3.40276286e-02 -6.12341166e-01 -4.46496159e-01
3.36995155e-01 3.45804021e-02 8.51420820e-01 -1.30451334e+00
4.83762503e-01 3.99073422e-01 1.22122772e-01 -5.77602312e-02
7.04753175e-02 -7.98058450e-01 3.10168803e-01 6.65651977e-01
-4.18313533e-01 1.56683579e-01 1.10467508e-01 1.15026677e+00
-2.25235038e-02 1.34526461e-01 9.77023840e-01 -1.01233898e-02
-2.87803918e-01 6.59455657e-01 -2.73981482e-01 2.43509576e-01
6.99057996e-01 -7.98041448e-02 -2.92433381e-01 -7.46174276e-01
-1.11292505e+00 1.16966195e-01 4.65023637e-01 2.35059887e-01
3.63684207e-01 -1.69329286e+00 -3.82662445e-01 5.66256464e-01
-9.94294360e-02 -5.98585382e-02 7.65501410e-02 9.10407543e-01
-7.58717954e-01 2.64572263e-01 -3.33610773e-01 -7.77544320e-01
-5.02366483e-01 6.81079686e-01 8.97914410e-01 3.26181114e-01
-9.62100744e-01 6.73113704e-01 5.33576012e-01 -3.24933231e-01
3.95093590e-01 -1.02880394e+00 1.46019414e-01 -4.10090268e-01
2.52869874e-01 3.27181607e-01 5.71437702e-02 -5.28037071e-01
4.49805846e-03 6.63123488e-01 5.84144711e-01 3.74845378e-02
1.07061386e+00 1.01522483e-01 -1.10350505e-01 5.76185763e-01
1.62045050e+00 -2.10773557e-01 -1.38583207e+00 -1.57987356e-01
7.27537423e-02 -8.05168673e-02 -9.07062441e-02 2.01733783e-01
-1.39650917e+00 8.62093627e-01 5.96837461e-01 9.87493217e-01
8.83681476e-01 1.25656456e-01 4.38924789e-01 2.86891878e-01
1.57457337e-01 -6.91522598e-01 -1.42342955e-01 5.61822295e-01
8.54749501e-01 -6.88904941e-01 -6.88032150e-01 -6.97605491e-01
2.69779891e-01 1.43991661e+00 1.80063739e-01 -1.20643747e+00
1.39935732e+00 -7.68658742e-02 -3.30946565e-01 -5.27230740e-01
-3.38708073e-01 -3.41425210e-01 4.13182825e-01 6.75902188e-01
6.84136823e-02 1.72256649e-01 2.19756573e-01 9.30253491e-02
-3.67482573e-01 -2.80411363e-01 6.79799557e-01 5.00060856e-01
-5.97120702e-01 -3.18577677e-01 -1.17586823e-02 4.91503626e-01
-1.13111198e-01 1.40097626e-02 -2.49585152e-01 1.04973257e+00
-3.50144833e-01 5.35488009e-01 4.05446202e-01 -2.86882967e-01
4.28122580e-01 -6.78054541e-02 6.74050331e-01 -5.54116905e-01
-1.82828516e-01 -2.87172794e-01 -3.04267138e-01 -6.59359336e-01
-2.79057831e-01 -2.76005387e-01 -1.14509845e+00 -3.57976288e-01
-2.67198473e-01 3.71054709e-01 4.93872851e-01 5.89231610e-01
5.02289414e-01 5.89035809e-01 3.04375291e-01 -7.86826968e-01
-7.08228588e-01 -7.60689080e-01 -1.04925668e+00 4.16167647e-01
8.88327718e-01 -6.23779595e-01 -7.80072570e-01 -8.67888257e-02] | [6.835962295532227, 6.059092998504639] |
569c20d5-ce9e-4c33-afe6-45d98bbb9826 | length-of-stay-prediction-for-hospital | 2306.16823 | null | https://arxiv.org/abs/2306.16823v1 | https://arxiv.org/pdf/2306.16823v1.pdf | Length of Stay prediction for Hospital Management using Domain Adaptation | Inpatient length of stay (LoS) is an important managerial metric which if known in advance can be used to efficiently plan admissions, allocate resources and improve care. Using historical patient data and machine learning techniques, LoS prediction models can be developed. Ethically, these models can not be used for patient discharge in lieu of unit heads but are of utmost necessity for hospital management systems in charge of effective hospital planning. Therefore, the design of the prediction system should be adapted to work in a true hospital setting. In this study, we predict early hospital LoS at the granular level of admission units by applying domain adaptation to leverage information learned from a potential source domain. Time-varying data from 110,079 and 60,492 patient stays to 8 and 9 intensive care units were respectively extracted from eICU-CRD and MIMIC-IV. These were fed into a Long-Short Term Memory and a Fully connected network to train a source domain model, the weights of which were transferred either partially or fully to initiate training in target domains. Shapley Additive exPlanations (SHAP) algorithms were used to study the effect of weight transfer on model explanability. Compared to the benchmark, the proposed weight transfer model showed statistically significant gains in prediction accuracy (between 1% and 5%) as well as computation time (up to 2hrs) for some target domains. The proposed method thus provides an adapted clinical decision support system for hospital management that can ease processes of data access via ethical committee, computation infrastructures and time. | ['Bart De Moor', 'Frank Rademakers', 'Elaine O. Nsoesie', 'Nyalleng Moorosi', 'Lyse Naomi Wamba Momo'] | 2023-06-29 | null | null | null | null | ['length-of-stay-prediction', 'management'] | ['medical', 'miscellaneous'] | [ 4.76797074e-02 3.18291694e-01 -4.38647389e-01 -4.43606436e-01
-4.92836982e-01 -2.10276544e-01 -2.49558672e-01 7.45523512e-01
-6.02909923e-01 1.10389042e+00 3.91468197e-01 -7.78799713e-01
-7.30423450e-01 -8.53957653e-01 -3.26344222e-01 -4.58443582e-01
-2.47935429e-01 8.64451110e-01 -4.16368127e-01 2.34264862e-02
1.45146772e-01 6.97297275e-01 -1.05199635e+00 7.67643869e-01
9.06359076e-01 9.58697855e-01 1.20952882e-01 7.14839339e-01
7.66375288e-02 1.00045991e+00 -2.06416413e-01 -2.90477555e-02
4.21183616e-01 -3.69701207e-01 -6.52430832e-01 -3.76447976e-01
-5.70205927e-01 -6.63411140e-01 -3.14682156e-01 1.49637505e-01
8.16001177e-01 5.48891872e-02 7.60739446e-01 -9.03018177e-01
-7.24524409e-02 7.59987295e-01 1.13349058e-01 3.78533840e-01
1.75609246e-01 1.16332859e-01 5.23038685e-01 -6.91565156e-01
2.19110653e-01 6.16206229e-01 7.08682895e-01 4.34946299e-01
-1.00202298e+00 -5.22483408e-01 -1.77011520e-01 8.77282843e-02
-1.10706592e+00 -8.03387687e-02 4.56781924e-01 -5.57849705e-01
1.14693785e+00 3.57329667e-01 6.66937768e-01 1.04816175e+00
5.74205935e-01 1.37531057e-01 7.82255054e-01 -5.16265154e-01
3.30528229e-01 3.85789275e-01 3.23834121e-01 5.06641984e-01
6.56686008e-01 1.71307981e-01 -1.55958980e-01 -4.19917554e-01
8.22245479e-01 7.76539803e-01 -3.69485736e-01 -1.65752783e-01
-1.28718126e+00 9.03007507e-01 4.91911829e-01 2.44221807e-01
-8.54387760e-01 -2.71226853e-01 4.80374277e-01 3.63772422e-01
1.73852399e-01 7.04744995e-01 -9.06466365e-01 -1.57307774e-01
-7.53759921e-01 -1.81122079e-01 8.35313678e-01 8.96487772e-01
1.30606398e-01 -1.33911341e-01 -3.00427735e-01 5.37584066e-01
-6.05367012e-02 4.28939462e-01 8.73124003e-01 -6.44068062e-01
6.73416317e-01 9.54650283e-01 2.76498169e-01 -6.79369211e-01
-9.84877467e-01 -4.34297055e-01 -1.07493126e+00 -1.95224807e-01
2.60123581e-01 -4.70088810e-01 -7.16501176e-01 1.32956004e+00
-1.18549913e-01 2.44864166e-01 4.22774911e-01 6.48777187e-01
6.13621414e-01 4.07151014e-01 5.12636863e-02 -4.72761154e-01
1.25992656e+00 -6.05707705e-01 -4.78523821e-01 -1.20038554e-01
1.25550306e+00 -4.92930382e-01 5.87726176e-01 3.37230951e-01
-1.16862762e+00 -2.67580032e-01 -7.50200748e-01 5.40283799e-01
-2.01615319e-01 1.97071917e-02 3.01282644e-01 2.09316507e-01
-6.44636214e-01 9.16071415e-01 -8.89610469e-01 -4.70709175e-01
4.38688487e-01 7.38269746e-01 -2.24183887e-01 -3.36257875e-01
-1.22054422e+00 9.96305585e-01 6.35028780e-01 -1.36344954e-01
-5.71856618e-01 -9.40436661e-01 -5.91174662e-01 4.66252178e-01
-1.14074491e-01 -1.15074074e+00 9.98451769e-01 -6.81002259e-01
-8.99747968e-01 5.00319600e-01 7.43525699e-02 -9.10294414e-01
4.99281824e-01 -3.14818978e-01 -4.14824367e-01 1.80892259e-01
-1.84082329e-01 1.72481298e-01 2.37914428e-01 -9.45342541e-01
-5.61221361e-01 -3.05679381e-01 -5.61834909e-02 2.33468518e-01
-3.99002641e-01 -3.65506321e-01 3.90337080e-01 -5.25166512e-01
9.50844213e-02 -9.88515317e-01 -4.83178556e-01 -4.19046164e-01
-1.10930115e-01 5.03144205e-01 2.05798358e-01 -7.65116990e-01
1.65295315e+00 -1.91204774e+00 -2.15499148e-01 4.53359485e-01
1.23674259e-01 1.95357680e-01 2.32295603e-01 7.51596987e-01
-5.92891097e-01 -9.58841890e-02 -2.39157930e-01 4.82759923e-02
-4.55816299e-01 4.24600184e-01 -1.63059562e-01 1.42445222e-01
-1.42648278e-04 7.74427772e-01 -7.08467305e-01 -4.20540839e-01
5.16072929e-01 3.20288926e-01 -7.81987429e-01 7.40564883e-01
3.84649932e-01 5.42226791e-01 -4.47832227e-01 5.05214572e-01
2.54423946e-01 -4.22421247e-01 2.64136702e-01 2.64512599e-01
1.29422083e-01 1.06982291e-01 -8.89780760e-01 1.37269402e+00
-6.75138175e-01 1.12100735e-01 -5.80707610e-01 -1.13844359e+00
9.87586915e-01 7.12521672e-01 8.75753582e-01 -5.42554975e-01
4.75362599e-01 2.86301643e-01 2.65693575e-01 -7.99553514e-01
-9.24526453e-02 -3.24148715e-01 -1.83927901e-02 4.44166541e-01
-4.76101220e-01 3.83705884e-01 -3.45061988e-01 -4.14473750e-02
1.37133014e+00 -4.84829605e-01 7.30819106e-01 -3.42693031e-01
4.76642370e-01 2.16454685e-01 5.43180346e-01 5.55194914e-01
-2.82010250e-02 5.74402750e-01 3.29802454e-01 -1.03153551e+00
-1.10258234e+00 -8.72880697e-01 -2.90261805e-01 7.88167894e-01
-2.72948980e-01 2.11442798e-01 -5.04059732e-01 -4.84799564e-01
3.73303257e-02 1.12985528e+00 -6.75598085e-01 -6.11038446e-01
-6.32351875e-01 -6.82995975e-01 3.19614083e-01 9.08218026e-01
1.09058030e-01 -1.13241613e+00 -1.23066556e+00 6.91685319e-01
-8.88389871e-02 -7.73931682e-01 -1.63513035e-01 5.97611189e-01
-1.44659281e+00 -1.18578815e+00 -7.32109249e-01 -5.21495581e-01
8.12919974e-01 4.01755348e-02 9.86524940e-01 1.92447692e-01
-3.36872220e-01 3.05553257e-01 -3.03428262e-01 -7.07215548e-01
-4.34080422e-01 1.96287081e-01 2.98036546e-01 -2.47161508e-01
5.86308897e-01 -5.58284044e-01 -1.03654170e+00 1.46445736e-01
-8.50761175e-01 1.04175642e-01 7.41631627e-01 1.25852418e+00
2.82001555e-01 -4.04937893e-01 1.00104570e+00 -9.86701965e-01
6.88275635e-01 -9.23429906e-01 -2.65467733e-01 1.62668914e-01
-1.22185731e+00 1.88742191e-01 9.09002841e-01 -2.18152359e-01
-9.25101280e-01 -8.45234618e-02 2.43768066e-01 -4.65360641e-01
-3.16653281e-01 6.26184285e-01 3.93566787e-01 5.84759176e-01
6.82356894e-01 -6.41365945e-02 4.72056819e-03 -3.80890340e-01
-3.81855369e-01 9.33282018e-01 1.70714766e-01 -3.22376311e-01
2.91864514e-01 2.33258441e-01 2.12519124e-01 2.23023966e-02
-7.61779010e-01 -7.20066667e-01 -8.78929019e-01 1.63782477e-01
8.17896366e-01 -9.57956135e-01 -4.99923348e-01 -3.26989561e-01
-1.05307913e+00 -2.28088081e-01 -3.20079327e-01 9.24801469e-01
-6.47529185e-01 -2.36068070e-01 -5.20052493e-01 -4.94309127e-01
-5.41545630e-01 -9.89780962e-01 3.85467827e-01 -1.43094510e-01
-5.26402831e-01 -1.27723300e+00 6.18900433e-02 1.86275765e-01
4.60695863e-01 6.19269431e-01 1.37577748e+00 -1.15446866e+00
-2.56482989e-01 -5.78763127e-01 -1.92761526e-01 3.47705007e-01
2.09412158e-01 -3.81520778e-01 -7.80209839e-01 -3.76790076e-01
5.27687445e-02 4.91439402e-02 3.26649547e-01 6.83287978e-01
1.25004530e+00 -5.76767802e-01 -5.13194144e-01 6.00523889e-01
1.61959541e+00 8.09667230e-01 4.58677500e-01 5.04609287e-01
2.93761015e-01 4.70117360e-01 6.96626544e-01 1.07737696e+00
2.78134704e-01 3.81115340e-02 4.57942545e-01 -2.25924149e-01
3.79786134e-01 9.09278989e-02 -9.84973982e-02 9.33132231e-01
-4.58621889e-01 -1.06894612e-01 -1.38856184e+00 4.62407827e-01
-1.99647832e+00 -9.01086390e-01 -7.53723532e-02 2.49917746e+00
4.08320397e-01 1.83987275e-01 -2.12417305e-01 3.35282326e-01
4.02953833e-01 -6.54403865e-01 -5.89668930e-01 -8.94469798e-01
4.53022152e-01 2.39893749e-01 8.00212979e-01 1.48478672e-01
-4.98271883e-01 1.08255923e-01 5.83387518e+00 -6.43298924e-02
-9.70301032e-01 -1.00231156e-01 9.29166973e-01 -3.81900489e-01
9.39231366e-02 -1.70193136e-01 -3.68159205e-01 4.53756601e-01
1.54462743e+00 -2.64515221e-01 3.09512496e-01 1.16202521e+00
7.02038109e-01 4.90910634e-02 -1.48643088e+00 9.41397071e-01
-2.56418675e-01 -1.36020291e+00 -1.32887825e-01 -1.15231238e-03
7.73109615e-01 -1.96765617e-01 -1.23397447e-01 2.74165392e-01
8.65534320e-02 -1.26608574e+00 2.18798611e-02 9.21320856e-01
8.84768128e-01 -1.06391454e+00 1.52117944e+00 5.28794408e-01
-8.15965533e-01 -7.01683402e-01 -3.25799495e-01 -3.53374809e-01
8.87541473e-02 3.27064097e-01 -1.59881592e+00 6.27741337e-01
5.85414588e-01 3.52421314e-01 -1.41889513e-01 9.57396448e-01
3.86418313e-01 5.97277224e-01 -2.19596215e-02 2.99939275e-01
3.22330415e-01 6.53376281e-02 -9.56374046e-04 1.06548309e+00
7.91744709e-01 6.22419178e-01 7.15400511e-03 3.21352422e-01
8.87210295e-02 2.30639964e-01 -9.22024667e-01 2.57753491e-01
4.43182915e-01 8.04678917e-01 -4.26885307e-01 -6.11861348e-01
-4.41068053e-01 6.32568181e-01 1.38283715e-01 3.00661057e-01
-1.01717341e+00 -3.14578056e-01 4.65644747e-01 6.76509321e-01
-1.30440533e-01 1.90123647e-01 -9.09897923e-01 -8.49536955e-01
-4.04259205e-01 -6.64703667e-01 6.05266392e-01 -6.19057655e-01
-9.78132665e-01 7.54863560e-01 2.06666142e-02 -1.58518302e+00
-6.06146336e-01 -5.39212346e-01 -5.59003651e-01 9.35180485e-01
-1.50538635e+00 -5.09505093e-01 -6.92792237e-01 6.62389278e-01
4.07415569e-01 -3.52790266e-01 1.25862980e+00 3.90503019e-01
-5.74234962e-01 4.15501058e-01 4.11214888e-01 4.65092026e-02
6.97585285e-01 -8.56214345e-01 -4.73449707e-01 3.81976157e-01
-7.00155020e-01 6.12431169e-01 4.58610713e-01 -4.51992422e-01
-9.00382996e-01 -1.38267052e+00 1.06321526e+00 -3.84658754e-01
3.03699344e-01 3.26242328e-01 -9.12799716e-01 7.79287815e-01
1.53480945e-02 -1.42480448e-01 1.07420075e+00 -2.36856952e-01
3.07447672e-01 -1.44796088e-01 -1.23892093e+00 3.74597043e-01
9.68586624e-01 2.60986462e-02 -8.43308628e-01 3.60100627e-01
7.00674653e-01 -3.16865563e-01 -1.29682875e+00 6.39016688e-01
5.05788863e-01 -1.09233320e+00 8.77626121e-01 -1.07680523e+00
6.76336288e-01 4.06813204e-01 -6.74892291e-02 -1.22802389e+00
-4.01637018e-01 -1.15894541e-01 4.20596302e-02 5.55313587e-01
5.35015643e-01 -9.37523246e-01 5.68491995e-01 1.23261225e+00
-3.80774796e-01 -1.32805157e+00 -7.54114747e-01 -4.80524361e-01
3.93208750e-02 -3.46143156e-01 9.71395254e-01 8.90098870e-01
2.12015480e-01 5.10487780e-02 -1.83346152e-01 2.47466162e-01
2.17047960e-01 6.37933537e-02 3.00081372e-01 -1.43099868e+00
-2.22248688e-01 -2.12831095e-01 -2.27353081e-01 -1.48325071e-01
-2.36441225e-01 -9.68516946e-01 -3.52939397e-01 -1.98506975e+00
2.44828373e-01 -7.29975343e-01 -8.83250415e-01 5.39944649e-01
5.38044982e-02 -4.86002207e-01 -1.27345249e-01 4.45771009e-01
7.57556036e-02 1.00891970e-01 9.76309299e-01 2.90950090e-01
-5.48251510e-01 2.14260951e-01 -5.88182390e-01 8.62615407e-01
1.07302868e+00 -7.92886496e-01 -5.96587658e-01 -3.48819703e-01
2.15022396e-02 7.82115459e-01 1.40337020e-01 -1.15330315e+00
1.34968475e-01 -4.13942188e-01 6.76395297e-01 -3.36875588e-01
1.30562454e-01 -1.43220806e+00 2.68315792e-01 1.10650635e+00
-4.76734132e-01 4.82451797e-01 2.28745610e-01 3.61385822e-01
-1.38363078e-01 -2.60881752e-01 6.82990730e-01 -2.64064819e-01
-2.01501727e-01 3.60480875e-01 -3.49483043e-01 -1.47457376e-01
1.56804502e+00 -3.56968850e-01 -2.78106406e-02 -1.53693751e-01
-8.90697122e-01 6.82581142e-02 2.03840658e-01 6.34472296e-02
8.67544055e-01 -1.16206098e+00 -7.00876951e-01 2.80691564e-01
2.81537503e-01 1.80383831e-01 4.83416378e-01 1.02211654e+00
-8.44089746e-01 9.65065539e-01 -3.62687171e-01 -3.26105982e-01
-1.04760563e+00 8.58044922e-01 1.36091977e-01 -5.51866174e-01
-7.08882332e-01 4.35030997e-01 6.25237077e-02 -1.40643358e-01
2.61436999e-01 -6.76414549e-01 -2.01495990e-01 -9.49515700e-02
4.11791801e-01 4.98148710e-01 8.70951563e-02 -6.79963902e-02
-3.57958853e-01 1.22236393e-01 5.98012134e-02 3.53947669e-01
1.60499299e+00 -9.70703736e-02 2.34271690e-01 4.87107217e-01
1.08306968e+00 -6.19981468e-01 -8.50053966e-01 -5.61445877e-02
1.32236227e-01 -2.60709167e-01 -1.63236916e-01 -9.06677842e-01
-6.38854921e-01 9.72898304e-01 7.52614617e-01 -6.56482056e-02
1.33520639e+00 -2.95094669e-01 8.07140350e-01 4.60586548e-01
2.67079175e-01 -9.93386030e-01 6.40297635e-03 9.03735235e-02
7.25371540e-01 -1.34307468e+00 -2.66081721e-01 1.98485956e-01
-9.34972405e-01 1.34726644e+00 3.74495268e-01 -2.18980670e-01
6.93004966e-01 2.10295528e-01 9.84745696e-02 -6.99317381e-02
-9.99585450e-01 4.48890865e-01 9.90395024e-02 3.12895834e-01
5.14054239e-01 3.82284790e-01 -2.60795772e-01 1.03412616e+00
5.07274494e-02 4.56958920e-01 8.00516486e-01 9.35232460e-01
-4.51433152e-01 -9.11419630e-01 -4.18756455e-01 1.12158585e+00
-4.43465143e-01 -4.14551020e-01 2.73688793e-01 7.22936630e-01
1.65247321e-01 6.83249772e-01 2.86188573e-02 -2.73229301e-01
6.14786685e-01 5.47208250e-01 -1.91022277e-01 -6.04948342e-01
-6.64085388e-01 -2.21413687e-01 1.11899927e-01 -3.41815054e-01
-1.02858186e-01 -4.14190710e-01 -1.64180040e+00 -3.59285772e-01
4.82013114e-02 2.30785966e-01 4.57775414e-01 8.34994912e-01
7.49556065e-01 1.07392430e+00 7.54175901e-01 -2.75897741e-01
-8.28101158e-01 -1.04472291e+00 -4.93623942e-01 2.78358907e-01
2.19179079e-01 -5.02224863e-01 -3.36751968e-01 1.21697791e-01] | [7.974412441253662, 6.209477424621582] |
dcb8a591-038a-4cc8-aaff-12bdbc449a58 | lsd-c-linearly-separable-deep-clusters | 2006.10039 | null | https://arxiv.org/abs/2006.10039v1 | https://arxiv.org/pdf/2006.10039v1.pdf | LSD-C: Linearly Separable Deep Clusters | We present LSD-C, a novel method to identify clusters in an unlabeled dataset. Our algorithm first establishes pairwise connections in the feature space between the samples of the minibatch based on a similarity metric. Then it regroups in clusters the connected samples and enforces a linear separation between clusters. This is achieved by using the pairwise connections as targets together with a binary cross-entropy loss on the predictions that the associated pairs of samples belong to the same cluster. This way, the feature representation of the network will evolve such that similar samples in this feature space will belong to the same linearly separated cluster. Our method draws inspiration from recent semi-supervised learning practice and proposes to combine our clustering algorithm with self-supervised pretraining and strong data augmentation. We show that our approach significantly outperforms competitors on popular public image benchmarks including CIFAR 10/100, STL 10 and MNIST, as well as the document classification dataset Reuters 10K. | ['Andrew Zisserman', 'Sylvestre-Alvise Rebuffi', 'Kai Han', 'Sebastien Ehrhardt', 'Andrea Vedaldi'] | 2020-06-17 | null | null | null | null | ['image-clustering'] | ['computer-vision'] | [ 9.57367383e-03 2.09937170e-01 -4.01033491e-01 -6.52401090e-01
-5.59208572e-01 -9.04289424e-01 7.65835762e-01 3.11157823e-01
-4.64846969e-01 4.43945169e-01 -2.86738779e-02 -1.04626648e-01
-3.68198305e-01 -3.68301243e-01 -6.98486626e-01 -9.63660538e-01
-3.94309282e-01 8.30586433e-01 4.26475406e-02 3.97792578e-01
2.49890387e-01 4.73900080e-01 -1.10171044e+00 5.00259221e-01
6.10563695e-01 9.34737980e-01 -2.59077758e-01 4.76431400e-01
1.10019512e-01 6.30370677e-01 -2.52899766e-01 -1.36922047e-01
5.76531947e-01 -5.67845941e-01 -1.17501020e+00 3.66366029e-01
4.86735672e-01 5.31778276e-01 -4.05271769e-01 9.85715091e-01
-2.76975539e-02 3.26208144e-01 1.03613758e+00 -1.33263028e+00
-4.83397722e-01 9.54482317e-01 -7.90464401e-01 1.67682812e-01
-3.36051136e-01 -1.92820117e-01 1.11420417e+00 -7.73388684e-01
6.85812294e-01 1.00821292e+00 5.26872456e-01 5.17274678e-01
-1.75033987e+00 -6.64534271e-01 3.36943388e-01 1.65179685e-01
-1.52622092e+00 -3.76084685e-01 6.55871093e-01 -5.16464949e-01
5.69606304e-01 1.73438311e-01 6.20338678e-01 7.82965183e-01
-3.09470683e-01 8.87162447e-01 8.59230578e-01 -5.24459720e-01
5.72536886e-01 4.27369326e-01 6.40689671e-01 3.89684081e-01
-8.54724571e-02 -5.29269986e-02 -4.38722372e-01 -3.00976038e-02
3.47648859e-01 5.88388257e-02 -1.97508559e-01 -1.12017620e+00
-1.10866845e+00 9.13895607e-01 9.28085327e-01 4.65300888e-01
8.33553150e-02 6.46444783e-02 2.32274607e-01 2.92596340e-01
4.01106119e-01 4.86806482e-01 -3.40612024e-01 4.16224808e-01
-1.22008264e+00 -3.12227666e-01 7.07107365e-01 6.52999222e-01
9.12970543e-01 -3.70653480e-01 7.97103196e-02 6.22641981e-01
4.37813371e-01 -9.50584039e-02 7.06815541e-01 -9.91811931e-01
3.74228239e-01 8.20831478e-01 -3.63489300e-01 -7.05414593e-01
-3.12268406e-01 -4.25400645e-01 -8.63493383e-01 1.95090577e-01
3.45861942e-01 -2.41072178e-01 -9.42333460e-01 1.70353854e+00
2.97740817e-01 5.47612965e-01 2.18056202e-01 5.29253185e-01
3.31515044e-01 6.22134745e-01 -1.65146306e-01 -1.94841906e-01
6.98613942e-01 -1.13227701e+00 -1.78383544e-01 -1.32381365e-01
6.81845367e-01 -2.96910852e-01 7.91435421e-01 3.34973812e-01
-8.26644182e-01 -6.61242485e-01 -1.09938085e+00 5.53804815e-01
-4.48732764e-01 2.25447357e-01 4.84627336e-01 5.40017068e-01
-1.17340124e+00 9.28829372e-01 -1.14198506e+00 -4.14336562e-01
7.90255368e-01 6.77676499e-01 -4.68265742e-01 6.14963733e-02
-4.30985093e-01 3.63245100e-01 6.48950279e-01 -1.63773298e-01
-7.05526650e-01 -4.96330112e-01 -6.48445010e-01 1.40169606e-01
6.16337545e-02 -1.63626559e-02 6.89488471e-01 -1.42595673e+00
-1.26999366e+00 1.04157484e+00 1.13616981e-01 -8.33475113e-01
3.90571386e-01 7.11871758e-02 -2.51701981e-01 1.74279734e-01
-6.23917393e-02 1.12282050e+00 7.63403237e-01 -1.38407838e+00
-6.55064762e-01 -5.40994585e-01 -6.29786193e-01 3.44163030e-01
-7.96953619e-01 -1.63688049e-01 -5.35440981e-01 -3.39595467e-01
2.95617253e-01 -1.21230996e+00 -4.06032473e-01 -2.90798634e-01
-8.75281334e-01 -3.88025522e-01 1.04720831e+00 1.03473403e-01
8.42806041e-01 -2.48686433e+00 2.34685823e-01 9.51142967e-01
3.63566041e-01 1.80268452e-01 -1.97935775e-01 4.33295257e-02
-5.24618685e-01 8.77978727e-02 -2.30882004e-01 -5.38575470e-01
-2.84011871e-01 -2.00334769e-02 -2.44390294e-01 7.46676981e-01
3.44458669e-02 7.41370380e-01 -7.78029025e-01 -3.02639574e-01
1.33901998e-01 2.13664249e-01 -6.11319184e-01 7.91173428e-04
2.78162230e-02 3.65312576e-01 -2.77372915e-02 -1.79256629e-02
4.09326315e-01 -5.93097866e-01 5.13682544e-01 4.90835197e-02
2.38753065e-01 1.75059721e-01 -1.10708869e+00 1.56856263e+00
1.65895119e-01 7.73221791e-01 -3.37333709e-01 -1.48488271e+00
9.37157393e-01 1.00925982e-01 6.19193017e-01 -4.79155593e-02
1.49715781e-01 -1.71625897e-01 2.57753134e-01 9.55569968e-02
1.01092733e-01 2.16265872e-01 8.96437764e-02 4.48288083e-01
3.30855280e-01 1.52279899e-01 2.73225844e-01 6.08605325e-01
8.73691261e-01 -1.60024777e-01 -9.78130028e-02 -6.04239523e-01
3.92663091e-01 -1.47965834e-01 3.69732767e-01 7.91336358e-01
-1.16923429e-01 7.01742649e-01 5.56480348e-01 -2.80154705e-01
-8.96688759e-01 -1.37203848e+00 -1.80515409e-01 7.06081212e-01
4.91608568e-02 -3.60522389e-01 -8.11238945e-01 -1.35396254e+00
-4.17878516e-02 5.12062132e-01 -9.65594053e-01 -4.36925888e-01
-2.57396996e-01 -7.28034317e-01 2.69663841e-01 4.59815830e-01
2.88800240e-01 -9.77919996e-01 -9.18405876e-02 -1.05876260e-01
2.49953672e-01 -7.48776376e-01 -4.87859726e-01 9.29922163e-01
-9.40764725e-01 -1.28297186e+00 -4.19479698e-01 -1.28869534e+00
1.17805374e+00 1.45310253e-01 8.48649561e-01 -1.69216692e-01
-2.83225864e-01 3.22580524e-02 -2.96432048e-01 1.56108871e-01
-3.10153037e-01 3.78949732e-01 1.29838228e-01 3.27034920e-01
6.26236200e-01 -5.21481872e-01 -5.51011503e-01 3.00022215e-01
-5.40928304e-01 -1.52589768e-01 2.32348964e-01 6.88578188e-01
6.63069308e-01 2.41362497e-01 4.14685339e-01 -9.14514720e-01
1.84445277e-01 -6.00410104e-01 -6.07234895e-01 3.68507653e-01
-6.67410672e-01 1.08396567e-01 6.76649272e-01 -7.04483747e-01
-5.57415664e-01 7.08601654e-01 5.96293628e-01 -6.87668741e-01
-2.99076319e-01 4.11655754e-01 6.42446429e-03 8.25510100e-02
9.38890278e-01 1.35151967e-01 -7.91406259e-03 -3.84045690e-01
5.51092207e-01 6.95664704e-01 7.52814293e-01 -2.85306692e-01
9.98100519e-01 6.76521003e-01 -2.07859084e-01 -6.02131367e-01
-7.19284594e-01 -7.18978226e-01 -1.14173985e+00 -1.85408909e-02
7.77332962e-01 -8.38431537e-01 -6.07471347e-01 1.55767098e-01
-5.62340796e-01 -5.30811429e-01 -7.09894359e-01 7.13307381e-01
-4.00716931e-01 3.55814099e-01 -6.67251408e-01 -5.56778848e-01
-4.89558093e-02 -9.59981978e-01 4.21515495e-01 3.06940794e-01
-3.78353715e-01 -9.20463383e-01 2.78129071e-01 9.44285393e-02
-1.33682847e-01 3.48971598e-02 7.86797345e-01 -1.20087683e+00
-2.24992886e-01 -3.67025465e-01 1.64162240e-03 6.88448131e-01
3.25790972e-01 3.47171336e-01 -7.99174488e-01 -5.23640513e-01
-3.24080378e-01 -5.59747159e-01 1.27111673e+00 5.21712601e-01
1.27287519e+00 -3.45460176e-01 -7.43768692e-01 5.61566770e-01
1.24788857e+00 2.86356598e-01 4.30742890e-01 1.15372613e-03
7.34388053e-01 8.28841686e-01 7.48125613e-02 -3.89275104e-02
1.67021882e-02 3.33413124e-01 1.12620354e-01 -2.65087873e-01
1.53731912e-01 -1.35510564e-01 3.38738054e-01 8.20341289e-01
5.15716851e-01 -2.74849772e-01 -9.48834956e-01 6.17828608e-01
-1.90263736e+00 -9.12396729e-01 -1.67940501e-02 2.33637190e+00
8.56919467e-01 2.75912762e-01 2.41148576e-01 2.23526046e-01
8.95810544e-01 -1.46121040e-01 -6.34660661e-01 -8.22263435e-02
-2.11687818e-01 2.59875983e-01 3.72243315e-01 3.56969744e-01
-1.52948058e+00 9.86811578e-01 6.24946022e+00 7.10235894e-01
-1.06306088e+00 -2.98880130e-01 1.37611246e+00 -3.77281040e-01
1.73687920e-01 1.19990326e-01 -7.68528938e-01 4.33400303e-01
9.84587133e-01 6.71399832e-02 3.27583820e-01 8.27474594e-01
-2.46302322e-01 1.42141864e-01 -1.46958792e+00 6.54973984e-01
8.32455531e-02 -1.53273201e+00 -1.60523996e-01 2.54047513e-01
1.26497483e+00 5.69254577e-01 4.29661632e-01 1.38905337e-02
9.34732735e-01 -1.06173623e+00 4.50612038e-01 1.48277596e-01
4.55199331e-01 -9.73382473e-01 3.14721912e-01 3.14308554e-01
-9.38416660e-01 -1.34711966e-01 -3.69232982e-01 2.82145172e-01
-4.24951553e-01 5.23830175e-01 -9.52014148e-01 1.79119214e-01
7.19851971e-01 1.02759504e+00 -7.21342027e-01 1.22642529e+00
2.43979812e-04 8.69209945e-01 -4.15812463e-01 2.32928708e-01
4.59684521e-01 -4.97583002e-01 6.02736212e-02 1.08455038e+00
-6.54419884e-02 -2.52833545e-01 3.87658805e-01 6.84263885e-01
-4.64535147e-01 1.60840943e-01 -3.95937562e-01 -5.74350357e-02
4.12139207e-01 1.34671080e+00 -1.36276186e+00 -5.61235189e-01
-2.10649922e-01 9.25872982e-01 6.72082126e-01 3.40545088e-01
-4.76153284e-01 -3.21727842e-01 4.06132132e-01 -1.09351486e-01
5.33592403e-01 3.87682877e-02 -1.66691825e-01 -1.03005075e+00
-3.19107682e-01 -6.43599272e-01 6.66825116e-01 -3.50783437e-01
-1.40012562e+00 7.41106689e-01 -3.72171074e-01 -1.16853654e+00
-1.49361938e-01 -3.62622261e-01 -7.94522345e-01 4.79334235e-01
-1.08790624e+00 -7.42926419e-01 -8.03072751e-03 5.88741720e-01
1.49139762e-01 -3.40819120e-01 7.37064600e-01 -1.05933389e-02
-7.68781364e-01 6.89902604e-01 8.42816532e-01 5.29560089e-01
6.69830024e-01 -1.35063517e+00 4.52010274e-01 6.64697886e-01
8.62702191e-01 5.34436047e-01 5.11192381e-02 -4.83218402e-01
-3.98436546e-01 -1.33682954e+00 6.40099943e-01 -4.63572651e-01
5.41935921e-01 -7.02008784e-01 -9.25508201e-01 9.32632685e-01
3.05239677e-01 3.97759676e-01 9.38793361e-01 7.90450275e-02
-6.54148996e-01 -7.44619444e-02 -1.11408794e+00 4.21910524e-01
7.43212759e-01 -4.45543438e-01 -4.10366088e-01 6.10898077e-01
4.71452326e-01 9.43191051e-02 -6.15574419e-01 1.40951887e-01
2.27721572e-01 -8.35837185e-01 8.16084802e-01 -1.01210368e+00
3.83240461e-01 -2.80389458e-01 -3.70264202e-02 -1.37588990e+00
-4.65265065e-01 -5.69794953e-01 9.08048823e-02 1.19888794e+00
6.61967695e-01 -1.87672421e-01 1.30462205e+00 3.55125695e-01
1.12804294e-01 -5.92798769e-01 -7.58176506e-01 -9.20088470e-01
4.48344558e-01 1.25552520e-01 1.82000563e-01 1.29696155e+00
4.40882206e-01 4.63101119e-01 6.57427236e-02 8.44019055e-02
8.24330807e-01 1.80263713e-01 5.00833571e-01 -1.65652215e+00
-3.27100515e-01 -6.15014851e-01 -4.93791342e-01 -9.02177274e-01
5.21612406e-01 -1.39478648e+00 1.35953352e-01 -9.96825397e-01
5.17994761e-01 -8.04096043e-01 -5.84385574e-01 5.61292171e-01
6.02858625e-02 4.49356079e-01 1.69628803e-02 4.82017756e-01
-9.85039234e-01 3.82808059e-01 4.67359215e-01 -1.16977662e-01
-5.21498561e-01 4.19684500e-02 -5.61110079e-01 6.88772857e-01
9.47657347e-01 -6.83071673e-01 -3.35778326e-01 -1.63251795e-02
-1.60465941e-01 -4.04915899e-01 2.89188270e-02 -1.15583420e+00
3.95419508e-01 2.14609087e-01 8.62027168e-01 -5.59459150e-01
5.89081831e-02 -8.57915640e-01 -1.79515481e-02 5.95403433e-01
-1.02896631e+00 -3.43942553e-01 -1.37609104e-02 6.51986897e-01
-1.05009310e-01 -1.09971918e-01 1.22504103e+00 1.45093009e-01
-1.91596821e-01 3.47409725e-01 -3.09843719e-01 1.51644722e-01
1.37300777e+00 -1.83240816e-01 -5.39547242e-02 -6.15606494e-02
-1.01487923e+00 6.27779782e-01 4.51075882e-01 2.70060748e-01
2.92381018e-01 -1.28985691e+00 -5.53569376e-01 3.83151770e-01
9.50914845e-02 1.02901496e-01 -6.22189008e-02 5.87929070e-01
-7.09061548e-02 3.32147300e-01 -7.44127631e-02 -8.74070823e-01
-1.40177822e+00 5.83536625e-01 4.00262207e-01 -2.42857456e-01
-3.48819822e-01 1.14687860e+00 2.80884594e-01 -6.16617382e-01
5.77686727e-01 -3.93982157e-02 -2.47485831e-01 1.40734419e-01
3.62727046e-01 1.63583420e-02 4.92128357e-03 -5.62412798e-01
-3.01679939e-01 2.60064870e-01 -5.63421309e-01 -1.00915223e-01
1.43549764e+00 -2.81289611e-02 7.69679621e-03 5.41955709e-01
1.81791162e+00 -3.13990027e-01 -1.45649159e+00 -6.17058873e-01
4.08125013e-01 -7.52664283e-02 -2.44778290e-01 -4.69572306e-01
-1.44955003e+00 6.50147736e-01 6.23914182e-01 2.24550724e-01
7.39301145e-01 4.29927021e-01 1.30392298e-01 5.96331060e-01
-2.32510746e-01 -1.28802931e+00 5.19353688e-01 3.77292246e-01
2.45635614e-01 -1.10610414e+00 -1.09829433e-01 -1.17117502e-01
-6.56522930e-01 9.06452656e-01 5.94123065e-01 -5.75978220e-01
9.01712418e-01 1.32140577e-01 -2.27543619e-02 -2.63704687e-01
-7.36697197e-01 1.00033648e-01 3.50902557e-01 6.41135752e-01
3.88533622e-01 -1.09996730e-02 1.88868240e-01 2.47799054e-01
-2.04805374e-01 -4.01732445e-01 3.16118002e-01 5.76506376e-01
-3.35628510e-01 -9.28604603e-01 -1.02621861e-01 7.63364017e-01
-2.13996097e-01 -1.37806311e-01 -7.54661739e-01 4.92151767e-01
-1.09700203e-01 7.29235232e-01 6.50863707e-01 -4.72733587e-01
-3.12884122e-01 2.86116034e-01 3.32696259e-01 -6.84386492e-01
-6.47340477e-01 1.75660014e-01 -4.53551859e-01 -3.52894038e-01
-4.59490925e-01 -9.23347533e-01 -1.38615084e+00 1.28615675e-02
-4.95846838e-01 6.28646612e-01 4.56862360e-01 7.81246662e-01
3.88067722e-01 2.00333640e-01 1.24160612e+00 -6.88193858e-01
-5.27411163e-01 -7.39292622e-01 -6.64088070e-01 5.30108273e-01
1.38607278e-01 -1.16601102e-01 -6.36023104e-01 2.90555865e-01] | [9.30469799041748, 3.0021655559539795] |
6af4b8ea-7cd3-499e-bfad-c1fc8c9df9cf | sparse-over-complete-patch-matching | 1806.03556 | null | http://arxiv.org/abs/1806.03556v2 | http://arxiv.org/pdf/1806.03556v2.pdf | Sparse Over-complete Patch Matching | Image patch matching, which is the process of identifying corresponding
patches across images, has been used as a subroutine for many computer vision
and image processing tasks. State -of-the-art patch matching techniques take
image patches as input to a convolutional neural network to extract the patch
features and evaluate their similarity. Our aim in this paper is to improve on
the state of the art patch matching techniques by observing the fact that a
sparse-overcomplete representation of an image posses statistical properties of
natural visual scenes which can be exploited for patch matching. We propose a
new paradigm which encodes image patch details by encoding the patch and
subsequently using this sparse representation as input to a neural network to
compare the patches. As sparse coding is based on a generative model of natural
image patches, it can represent the patch in terms of the fundamental visual
components from which it has been composed of, leading to similar sparse codes
for patches which are built from similar components. Once the sparse coded
features are extracted, we employ a fully-connected neural network, which
captures the non-linear relationships between features, for comparison. We have
evaluated our approach using the Liberty and Notredame subsets of the popular
UBC patch dataset and set a new benchmark outperforming all state-of-the-art
patch matching techniques for these datasets. | ['Clinton Fookes', 'Sridha Sridharan', 'Kien Nguyen', 'Akila Pemasiri'] | 2018-06-09 | null | null | null | null | ['patch-matching'] | ['computer-vision'] | [ 4.79074746e-01 -2.59662598e-01 -2.52325207e-01 -2.73631811e-01
-5.69277346e-01 -3.36999685e-01 8.72820735e-01 4.57536966e-01
-1.62686542e-01 2.38169611e-01 1.72228038e-01 1.83144093e-01
-2.92759418e-01 -1.11850178e+00 -1.01011527e+00 -6.39694691e-01
-2.79222727e-01 1.21139936e-01 2.46427059e-01 -2.31894076e-01
5.14356136e-01 6.12599373e-01 -2.24700308e+00 9.39336777e-01
3.59926969e-01 1.24723458e+00 1.50357649e-01 4.40567315e-01
-9.14066210e-02 8.29577088e-01 -2.13763654e-01 -3.75947617e-02
5.56663275e-01 -5.96758842e-01 -9.18487489e-01 1.36541083e-01
1.10668290e+00 1.87216029e-01 -6.75195754e-01 1.14417529e+00
1.68665856e-01 1.74640894e-01 7.33217537e-01 -1.04168034e+00
-5.18306911e-01 1.78195715e-01 -4.18013543e-01 2.47576579e-01
6.48749590e-01 -2.64751107e-01 1.07580960e+00 -8.00610483e-01
7.47451365e-01 1.03242075e+00 8.45155358e-01 -3.43404487e-02
-1.36551309e+00 -3.16284567e-01 -2.10923463e-01 5.11738062e-01
-1.61718535e+00 -4.16490912e-01 9.87010777e-01 -4.61094886e-01
1.03786194e+00 3.71110708e-01 6.98970556e-01 4.97877449e-01
2.24361405e-01 6.75337553e-01 1.02264583e+00 -6.66624129e-01
2.63674080e-01 -1.63834076e-02 -4.35300032e-03 7.41530895e-01
-1.91747859e-01 3.94117862e-01 -4.83847648e-01 -3.81851614e-01
7.07806945e-01 5.72734296e-01 -4.11874086e-01 -5.76998055e-01
-1.01463425e+00 9.36695874e-01 9.79083478e-01 7.87854075e-01
-8.47133994e-01 -2.96039730e-02 1.55431941e-01 4.23860103e-01
2.34802961e-01 1.92177773e-01 1.42983347e-01 3.81247848e-01
-1.29919767e+00 4.47567552e-01 8.49930584e-01 6.25064492e-01
1.51781142e+00 -3.39399636e-01 -2.66389281e-01 1.03996921e+00
1.93455368e-01 3.54615957e-01 5.91886580e-01 -8.03285003e-01
1.62550405e-01 1.02887416e+00 -4.90461797e-01 -1.64539087e+00
6.35871589e-02 -3.80095780e-01 -1.14467680e+00 3.39411527e-01
1.73941646e-02 6.83466017e-01 -8.63074601e-01 1.31106913e+00
1.77651241e-01 5.02607405e-01 7.65361562e-02 8.37003946e-01
1.02709913e+00 6.21931911e-01 -3.17367375e-01 2.41827250e-01
1.40303481e+00 -8.22617650e-01 6.39222413e-02 -1.97974384e-01
1.37306556e-01 -1.03157330e+00 5.64743996e-01 1.61309451e-01
-8.76202583e-01 -9.02420282e-01 -1.04567266e+00 1.85420632e-01
-4.85939085e-01 -1.93524379e-02 2.83643991e-01 1.46217257e-01
-1.09585512e+00 1.01194966e+00 -3.96356225e-01 -5.38274348e-01
6.36453390e-01 1.91420078e-01 -1.00480461e+00 -4.59984362e-01
-7.74541497e-01 5.94004929e-01 3.23685348e-01 -1.22686930e-01
-7.70628631e-01 -7.93372929e-01 -1.22791100e+00 2.56163269e-01
-1.67539075e-01 -4.54335362e-01 6.78029537e-01 -1.38499963e+00
-9.62874830e-01 1.12989259e+00 -1.69108927e-01 -4.65858161e-01
-1.37448937e-01 3.28571171e-01 -3.29091281e-01 5.43303013e-01
3.06809247e-01 7.38075256e-01 1.28209198e+00 -1.19273889e+00
-6.88734889e-01 -2.66040236e-01 -1.09023497e-01 -3.58774178e-02
8.99494439e-02 -1.66574985e-01 -3.50940645e-01 -7.27244854e-01
3.56256276e-01 -8.52107286e-01 -1.78216591e-01 2.21708700e-01
5.03903022e-03 -3.68863158e-02 9.73079443e-01 -4.25821304e-01
9.02275264e-01 -2.27779722e+00 2.91308492e-01 6.47285938e-01
2.40165949e-01 2.35282928e-01 -5.65035939e-01 6.94774091e-01
-4.35135990e-01 -4.51412827e-01 -6.38856053e-01 -3.41407329e-01
-1.76233903e-01 2.44064704e-01 -2.03646794e-01 8.60093236e-01
3.97619963e-01 7.36145675e-01 -6.64789379e-01 -3.80794495e-01
3.65580976e-01 6.88526511e-01 -5.38767636e-01 4.38885003e-01
1.41922817e-01 7.93651789e-02 -1.28843235e-02 6.20710194e-01
7.77082205e-01 -1.08651601e-01 -1.52274802e-01 -5.27757645e-01
-1.19310968e-01 -1.71630376e-03 -1.23413396e+00 1.96814895e+00
-3.76249760e-01 8.08260322e-01 -5.39991856e-02 -1.68486238e+00
1.16116059e+00 7.55631253e-02 5.76416731e-01 -9.89500821e-01
4.16179150e-02 3.12460244e-01 -1.43510327e-01 -3.96314800e-01
3.78504604e-01 -5.85722849e-02 1.74138322e-01 4.63301063e-01
3.02269310e-01 9.79956985e-03 1.86512128e-01 7.35056922e-02
1.21579504e+00 -1.34420663e-01 4.49690580e-01 -4.46653008e-01
8.23966980e-01 1.08497038e-01 1.84587285e-01 6.45289302e-01
1.52714521e-01 1.00590360e+00 4.06886563e-02 -1.00314748e+00
-1.13964868e+00 -7.57405043e-01 -2.84940541e-01 5.93573987e-01
1.48811072e-01 -4.19917524e-01 -6.04823112e-01 -4.97071356e-01
2.93706387e-01 -6.23874366e-02 -1.07044578e+00 -1.48022637e-01
-6.11278355e-01 -1.96208730e-01 2.54686207e-01 2.72590101e-01
5.63480794e-01 -1.36464250e+00 -8.58398199e-01 2.32362062e-01
2.10266188e-02 -7.73317397e-01 -2.75073767e-01 2.16295883e-01
-5.29888570e-01 -1.45683169e+00 -8.56379628e-01 -1.21752679e+00
8.22735310e-01 6.46946490e-01 1.43884861e+00 5.46263576e-01
-8.23045135e-01 4.82927799e-01 -5.10089636e-01 4.81183529e-02
-3.82182449e-01 -1.66157812e-01 -3.70278955e-01 5.42845845e-01
3.27510417e-01 -8.77553284e-01 -7.92684317e-01 2.47990593e-01
-1.13847089e+00 -2.74271518e-01 7.67756820e-01 9.46070075e-01
1.10687292e+00 1.32555127e-01 -7.00603276e-02 -8.16028714e-01
2.34877482e-01 -6.39653981e-01 -5.21977067e-01 2.06430778e-01
-9.41504017e-02 2.13499695e-01 6.49407208e-01 -2.90248334e-01
-3.73009175e-01 3.40339214e-01 -1.58524141e-01 -6.81946158e-01
-4.24750566e-01 5.57509184e-01 1.88897222e-01 -6.90868914e-01
7.58795440e-01 5.96117020e-01 2.23270833e-01 -5.12893796e-01
2.79697150e-01 6.11170769e-01 6.46234274e-01 -4.73703593e-01
7.56220698e-01 5.98171175e-01 2.49502033e-01 -8.86285484e-01
-4.42362905e-01 -9.38385427e-01 -7.05092967e-01 3.13747376e-02
6.49240136e-01 -8.11445832e-01 -3.79719138e-01 3.73902470e-01
-1.08075738e+00 -1.52031137e-02 -4.52488869e-01 2.25943014e-01
-6.23905480e-01 5.84182382e-01 -1.97370112e-01 -2.66777009e-01
-4.10349429e-01 -1.04344690e+00 1.23884237e+00 2.58381635e-01
-3.27421576e-02 -8.22043419e-01 5.98058045e-01 -7.97441080e-02
5.88192999e-01 3.08345556e-01 9.02297258e-01 -4.48939323e-01
-5.28539360e-01 -4.69284922e-01 -2.54701465e-01 3.46589446e-01
2.69173801e-01 -3.13046247e-01 -9.68675911e-01 -5.05011499e-01
-7.05285668e-02 -7.49167055e-02 1.04163206e+00 2.49689132e-01
1.07354259e+00 -4.50010657e-01 -3.41118783e-01 8.05623531e-01
2.04951334e+00 -1.11505747e-01 1.00934982e+00 4.08852398e-01
5.31503975e-01 6.67650878e-01 1.65877759e-01 2.14401767e-01
6.15909770e-02 8.10446024e-01 4.65897441e-01 -2.96919465e-01
-2.39745691e-01 -2.92067319e-01 -9.75208208e-02 4.93867934e-01
6.45854920e-02 3.58368963e-01 -6.57484710e-01 6.82875335e-01
-1.94301271e+00 -1.21119261e+00 1.27621889e-01 2.34598160e+00
4.90433097e-01 -3.01671922e-01 9.39046219e-02 3.02422881e-01
6.18617356e-01 2.80098706e-01 -1.01066530e-01 -2.63684183e-01
-1.46204934e-01 8.77208710e-01 3.03034037e-01 2.77896583e-01
-1.30350161e+00 5.42692840e-01 6.10161734e+00 8.77820015e-01
-1.18318689e+00 -1.64577812e-01 4.39884722e-01 3.87444079e-01
-1.53010353e-01 1.66916594e-01 -2.00626016e-01 4.21157300e-01
5.78796744e-01 9.45003256e-02 7.38936722e-01 8.16369832e-01
-3.92259836e-01 -1.63470536e-01 -1.30492902e+00 1.29470253e+00
5.20277619e-01 -1.71315789e+00 1.66564956e-01 7.41051091e-03
7.96258509e-01 1.21841997e-01 -4.13381457e-02 8.91714990e-02
-3.04956138e-01 -1.28745079e+00 5.99567115e-01 5.29349923e-01
5.09295881e-01 -5.58593988e-01 8.00452530e-01 9.63494629e-02
-1.60657454e+00 -1.31925615e-02 -9.49464500e-01 -1.49501354e-01
-2.87026346e-01 3.93262625e-01 -3.53226125e-01 6.70369208e-01
8.63248169e-01 1.14387333e+00 -7.13559091e-01 1.44694948e+00
1.01666883e-01 2.51505196e-01 -1.88689843e-01 4.56873685e-01
3.97467703e-01 -3.01375002e-01 3.58914226e-01 1.21336722e+00
3.79446000e-01 -1.60970643e-01 2.87757546e-01 1.15473127e+00
-1.08749000e-02 3.58997881e-01 -1.00535953e+00 2.45253071e-01
3.13977778e-01 1.36768138e+00 -6.49491549e-01 -4.20484126e-01
-6.51037991e-01 1.06378686e+00 4.14302617e-01 4.29082662e-02
-3.34242195e-01 -5.02341330e-01 6.81321979e-01 5.52456677e-02
7.95603335e-01 1.46966159e-01 2.50098795e-01 -9.08205926e-01
3.73604782e-02 -1.20747066e+00 2.82718539e-01 -5.12541354e-01
-1.45370889e+00 9.71691370e-01 -2.13942647e-01 -1.61218834e+00
-2.46930838e-01 -4.32347000e-01 -8.79680812e-01 1.02523625e+00
-1.80552208e+00 -1.18733120e+00 -5.48446178e-01 1.12322927e+00
3.39464068e-01 -4.13925320e-01 1.12371624e+00 3.90871853e-01
-3.83309089e-02 4.65550065e-01 5.23782037e-02 3.29876870e-01
3.17463934e-01 -1.04462254e+00 3.34458709e-01 7.32061625e-01
7.34521687e-01 5.13037503e-01 3.02284241e-01 -2.88749635e-01
-1.48557460e+00 -1.09510040e+00 8.02783132e-01 5.04338592e-02
4.01973993e-01 -1.42306983e-01 -1.15959537e+00 1.69717655e-01
4.37932551e-01 6.53775156e-01 6.56697333e-01 -1.86820924e-01
-7.82643318e-01 -2.06531659e-01 -1.10203850e+00 2.93498160e-03
8.00421119e-01 -9.58904147e-01 -6.79214060e-01 3.82490084e-02
3.73807520e-01 -4.71676663e-02 -9.54455018e-01 2.78070390e-01
6.37531698e-01 -1.20472538e+00 1.20386124e+00 -4.89394575e-01
6.04114234e-01 -3.59415829e-01 -5.48907042e-01 -1.30586684e+00
-6.23563826e-01 -2.20177338e-01 3.12016159e-01 9.09817755e-01
9.65081826e-02 -3.42169464e-01 8.85117650e-01 -1.04784094e-01
1.16376346e-03 -6.74963415e-01 -8.87411475e-01 -6.64330065e-01
-1.51127025e-01 -1.18466012e-01 6.82174504e-01 8.63394499e-01
-1.88876510e-01 1.22956425e-01 -2.63364017e-01 1.34991199e-01
6.41818762e-01 8.19912016e-01 6.60666943e-01 -1.28281343e+00
-4.11258310e-01 -5.62284648e-01 -1.08605611e+00 -6.53879166e-01
3.35504502e-01 -1.13439441e+00 1.02282852e-01 -1.33566082e+00
3.54572147e-01 -5.09849727e-01 -4.00128275e-01 5.26133478e-01
-3.08557302e-02 7.90723026e-01 3.05460840e-01 4.63196486e-01
-5.13275445e-01 4.79354590e-01 8.05816412e-01 -7.05089033e-01
2.31992006e-02 -1.53265521e-01 -4.68260199e-01 3.38880301e-01
4.42625046e-01 -5.98629773e-01 -5.44112101e-02 -2.45928779e-01
-5.56871807e-03 -1.81320503e-01 7.67671466e-01 -1.38595963e+00
3.88981640e-01 2.70959884e-01 3.17224532e-01 -4.10791904e-01
1.28492817e-01 -1.18526077e+00 5.13294280e-01 6.15603030e-01
-1.69859037e-01 1.01438470e-01 1.61862373e-01 4.48007464e-01
-8.69578063e-01 -4.18840706e-01 7.78643668e-01 -3.78370523e-01
-8.65812004e-01 5.13614357e-01 -1.99389815e-01 -1.83963344e-01
8.45542789e-01 -4.04980808e-01 -2.22626105e-02 -1.62362173e-01
-3.58407885e-01 -3.57117116e-01 6.76606297e-01 3.60427886e-01
8.81015241e-01 -1.69699407e+00 -8.89687181e-01 8.10884655e-01
4.23815817e-01 -2.57627308e-01 3.31477404e-01 5.85390449e-01
-7.12503850e-01 2.21278146e-01 -6.94628417e-01 -8.67241383e-01
-1.35739088e+00 8.71672571e-01 4.13519621e-01 1.07134869e-02
-8.71938765e-01 5.69185317e-01 8.65815282e-02 -2.63009369e-01
1.95704669e-01 -1.32821187e-01 -4.61358815e-01 -9.34635773e-02
7.43435562e-01 8.08923040e-03 2.52263278e-01 -1.09812164e+00
-4.24274445e-01 1.13403702e+00 1.35466889e-01 4.06208873e-01
1.52901697e+00 2.87615627e-01 -5.88294446e-01 -1.04717605e-01
1.91356015e+00 -1.60156712e-01 -8.01519871e-01 -6.80582762e-01
-1.22188151e-01 -8.25424731e-01 6.91377968e-02 -1.95308119e-01
-1.34689808e+00 8.02423596e-01 8.67322564e-01 2.89412171e-01
1.25051177e+00 1.67086899e-01 6.31999552e-01 3.09542418e-01
4.84404385e-01 -4.95648116e-01 -2.88329609e-02 2.00258866e-01
1.03591681e+00 -1.35533345e+00 -8.56524613e-03 -2.91975200e-01
-1.29449040e-01 1.18649888e+00 6.53337629e-04 -8.62603486e-01
9.35993075e-01 2.02684645e-02 -1.78782955e-01 -6.01528823e-01
-3.30645204e-01 -4.51396823e-01 7.82045782e-01 6.63787365e-01
2.83609957e-01 1.70964859e-02 -1.41145423e-01 9.22087803e-02
-1.04792111e-01 -2.36013085e-01 -1.46322042e-01 9.22782421e-01
-4.93024707e-01 -1.40406036e+00 -4.46345121e-01 4.09606993e-01
-2.26954505e-01 -2.44971648e-01 -4.04269069e-01 4.72403765e-01
3.75444233e-01 6.39463842e-01 2.90860116e-01 -4.24125105e-01
3.98988605e-01 -3.46533567e-01 5.58607101e-01 -6.57800436e-01
-8.61305594e-01 -2.42892712e-01 -4.46587324e-01 -1.03555751e+00
-8.22059512e-01 -5.31825721e-01 -6.68326795e-01 -8.70004296e-02
1.18799679e-01 1.96854442e-01 4.81284022e-01 8.87975991e-01
4.53082711e-01 1.67923287e-01 9.23054099e-01 -1.24007523e+00
-1.51201993e-01 -7.39077270e-01 -6.65597379e-01 9.80670333e-01
3.39457631e-01 -6.34891987e-01 -1.87108546e-01 7.63204917e-02] | [10.258938789367676, 0.08440513908863068] |
ffa045df-02d4-46bb-8d64-2efb3b736f14 | deeplidarflow-a-deep-learning-architecture | 2008.08136 | null | https://arxiv.org/abs/2008.08136v1 | https://arxiv.org/pdf/2008.08136v1.pdf | DeepLiDARFlow: A Deep Learning Architecture For Scene Flow Estimation Using Monocular Camera and Sparse LiDAR | Scene flow is the dense 3D reconstruction of motion and geometry of a scene. Most state-of-the-art methods use a pair of stereo images as input for full scene reconstruction. These methods depend a lot on the quality of the RGB images and perform poorly in regions with reflective objects, shadows, ill-conditioned light environment and so on. LiDAR measurements are much less sensitive to the aforementioned conditions but LiDAR features are in general unsuitable for matching tasks due to their sparse nature. Hence, using both LiDAR and RGB can potentially overcome the individual disadvantages of each sensor by mutual improvement and yield robust features which can improve the matching process. In this paper, we present DeepLiDARFlow, a novel deep learning architecture which fuses high level RGB and LiDAR features at multiple scales in a monocular setup to predict dense scene flow. Its performance is much better in the critical regions where image-only and LiDAR-only methods are inaccurate. We verify our DeepLiDARFlow using the established data sets KITTI and FlyingThings3D and we show strong robustness compared to several state-of-the-art methods which used other input modalities. The code of our paper is available at https://github.com/dfki-av/DeepLiDARFlow. | ['Didier Stricker', 'Oliver Wasenmüller', 'René Schuster', 'Ramy Battrawy', 'Rishav'] | 2020-08-18 | null | null | null | null | ['scene-flow-estimation'] | ['computer-vision'] | [-1.86651275e-01 -6.54818177e-01 2.57436335e-01 -4.19094503e-01
-5.00782490e-01 -5.51735163e-01 5.40286720e-01 -7.26500675e-02
-5.58073163e-01 6.36407197e-01 -6.80701956e-02 -2.03662571e-02
-5.79666463e-04 -9.87029672e-01 -6.46370411e-01 -6.16115153e-01
2.25633547e-01 5.06053746e-01 5.43483555e-01 -3.00581425e-01
1.33049071e-01 7.96401560e-01 -1.99825191e+00 1.72517762e-01
6.04759634e-01 1.07452524e+00 3.55904073e-01 8.37094486e-01
-2.58540839e-01 6.76270783e-01 -2.89717503e-02 -7.39045739e-02
7.25201488e-01 -1.94076657e-01 -5.18732011e-01 -7.87168667e-02
9.42308128e-01 -6.25440359e-01 -8.00081849e-01 8.64190161e-01
6.21888399e-01 5.10969348e-02 3.97952914e-01 -1.13714576e+00
-8.84421095e-02 -2.53398716e-01 -6.12948477e-01 1.81390211e-01
6.24665082e-01 3.95912260e-01 6.87379479e-01 -9.90797222e-01
5.14184773e-01 1.20088911e+00 7.27466643e-01 7.63540789e-02
-9.56563890e-01 -5.56765556e-01 -6.33729026e-02 3.11614692e-01
-1.30174434e+00 -5.06672025e-01 9.08768654e-01 -4.89744514e-01
8.15647602e-01 8.61987546e-02 6.40852392e-01 8.29089224e-01
-1.22014591e-02 4.81641918e-01 1.30684149e+00 -1.83420137e-01
7.85739124e-02 1.09427072e-01 -1.85393617e-01 8.41511011e-01
2.09247649e-01 5.94977856e-01 -6.76806569e-01 2.48717964e-01
7.95666039e-01 3.46046537e-01 -4.43112284e-01 -5.87374032e-01
-1.24242651e+00 6.80752337e-01 8.74615967e-01 1.56625077e-01
-1.74111143e-01 1.64149076e-01 9.67848375e-02 3.26139778e-01
2.46547982e-01 7.29924515e-02 -4.16152120e-01 -4.40602675e-02
-9.69547153e-01 2.69489348e-01 5.42747438e-01 8.33294392e-01
1.23925006e+00 2.00221967e-02 4.75518793e-01 4.75320190e-01
4.95741218e-01 9.21115816e-01 2.26525262e-01 -1.20135522e+00
5.46326041e-01 6.18414581e-01 1.05954312e-01 -1.00280404e+00
-5.26603937e-01 -1.88263506e-01 -7.46795952e-01 6.23356342e-01
5.17080903e-01 8.68987963e-02 -8.29410493e-01 1.17571104e+00
5.10307491e-01 3.04198146e-01 5.51513396e-02 1.26979494e+00
1.26206028e+00 5.37151754e-01 -5.53896785e-01 2.22090095e-01
9.19527709e-01 -6.69490993e-01 -3.11958283e-01 -3.93598586e-01
3.51477593e-01 -1.02440929e+00 1.04832518e+00 2.84911841e-01
-8.58863294e-01 -7.62720764e-01 -9.67866182e-01 -2.95310706e-01
-6.47409320e-01 1.22673449e-03 7.00545311e-01 3.57953072e-01
-1.06047058e+00 6.36566520e-01 -7.28514910e-01 -5.49587190e-01
3.87397915e-01 2.82901615e-01 -6.85932219e-01 -5.22408903e-01
-8.74550104e-01 9.63704944e-01 8.26520771e-02 2.77634770e-01
-7.22401977e-01 -7.50653803e-01 -1.06817865e+00 -3.19382250e-01
9.30901617e-02 -7.46297002e-01 8.53098869e-01 -4.02948499e-01
-1.35523427e+00 1.00691020e+00 -1.95541531e-01 -2.00775549e-01
8.57752025e-01 -4.07903910e-01 -6.62960531e-03 2.27237716e-01
-4.09370847e-02 7.53095210e-01 6.97928429e-01 -1.25419974e+00
-6.22595787e-01 -5.05394399e-01 1.89378396e-01 2.36055464e-01
1.98412776e-01 -4.03827161e-01 -3.30437243e-01 1.50578562e-02
2.02959627e-01 -8.32753897e-01 -2.13561401e-01 4.46317792e-01
-1.86321273e-01 3.11394662e-01 1.09581196e+00 -2.94611692e-01
4.91851687e-01 -1.95663095e+00 -2.00343374e-02 -2.39414290e-01
-3.97163369e-02 4.53627527e-01 -3.48488167e-02 5.38178682e-01
2.52814919e-01 -3.19851488e-01 -3.09621900e-01 -5.57343781e-01
-2.52296507e-01 3.54183584e-01 -2.82262325e-01 9.69164968e-01
9.23425779e-02 6.47488654e-01 -8.56092155e-01 -4.69555944e-01
1.10161662e+00 8.04487586e-01 -3.95967484e-01 4.35960919e-01
5.37430607e-02 8.25991929e-01 -2.01137558e-01 8.09455514e-01
1.10290360e+00 9.28613842e-02 -4.33056027e-01 -4.50933367e-01
-5.21388829e-01 2.42540479e-01 -1.46467769e+00 2.04949737e+00
-6.58145726e-01 8.42332363e-01 1.24562122e-01 -6.96986139e-01
9.33485091e-01 -7.47872889e-02 5.04644930e-01 -8.66072834e-01
2.01371059e-01 4.10565913e-01 -2.69651681e-01 -5.55636108e-01
4.62026358e-01 -2.42104530e-01 1.55570194e-01 1.10403806e-01
3.35914642e-02 -7.83567429e-01 -1.18040042e-02 1.67076752e-01
1.00421453e+00 4.91511106e-01 1.30549863e-01 -6.29782900e-02
6.92337275e-01 2.38486007e-03 5.07619858e-01 6.32770598e-01
-3.02317798e-01 1.06172097e+00 -6.90200403e-02 -5.69188237e-01
-9.69692171e-01 -1.14270747e+00 -3.37550789e-01 3.10509920e-01
4.81182605e-01 -1.64846122e-01 -1.45354167e-01 -3.88365507e-01
3.62241805e-01 2.94623643e-01 -4.03339893e-01 1.66082963e-01
-4.62528259e-01 -4.88146394e-01 2.81053156e-01 4.17406440e-01
8.01977396e-01 -7.71390557e-01 -1.08570981e+00 -2.47434061e-02
-5.57631105e-02 -1.44173419e+00 5.12402542e-02 1.57929938e-02
-8.85235965e-01 -1.28515804e+00 -4.06480193e-01 -2.35325068e-01
3.71361703e-01 7.01646745e-01 1.08009887e+00 7.62401521e-02
-5.16675651e-01 3.91220927e-01 -2.89497107e-01 -2.38156199e-01
-4.87938821e-02 -2.52787083e-01 1.06524549e-01 -9.62492153e-02
9.21581686e-02 -7.73150265e-01 -7.41862655e-01 3.59572023e-01
-7.57016361e-01 6.99389130e-02 3.20042461e-01 5.58977902e-01
4.34568167e-01 -1.44098058e-01 -1.12419970e-01 -5.29057086e-01
-2.15742111e-01 -2.33423293e-01 -1.02803481e+00 -2.44268268e-01
-1.93110526e-01 -1.29498718e-02 4.45349813e-01 1.39303401e-01
-1.04863095e+00 4.97548819e-01 -1.89058214e-01 -8.26372087e-01
-5.21843433e-01 1.09685652e-01 -7.03203231e-02 -3.53378743e-01
5.63782811e-01 1.97473099e-03 -7.33194277e-02 -6.16703093e-01
2.45516330e-01 5.05237222e-01 6.70686424e-01 -1.76582903e-01
1.09983099e+00 1.21586430e+00 4.84181613e-01 -1.06939197e+00
-8.49832356e-01 -8.87759626e-01 -1.18325281e+00 -4.45883214e-01
6.77293479e-01 -1.12356293e+00 -6.27713382e-01 7.50861824e-01
-1.11105478e+00 -3.36404979e-01 -2.30006948e-01 6.69501543e-01
-5.71978569e-01 3.96879822e-01 -3.74007702e-01 -8.62176776e-01
-1.02820612e-01 -1.18670869e+00 1.35069442e+00 3.81174326e-01
2.79108196e-01 -1.04699922e+00 1.54486552e-01 3.87355208e-01
3.79743397e-01 5.68267107e-01 1.33287698e-01 2.32674912e-01
-9.36185181e-01 -1.33724928e-01 -4.06642705e-01 2.83746898e-01
2.58253485e-01 2.09580541e-01 -1.48342454e+00 -1.89325482e-01
-1.86124623e-01 -4.82595950e-01 1.12015629e+00 2.63889462e-01
5.76808333e-01 2.66960412e-01 -8.91222730e-02 1.18516386e+00
1.91368103e+00 -9.95955914e-02 6.20716453e-01 4.61650938e-01
1.03718054e+00 7.50428200e-01 8.05388391e-01 4.58637774e-01
4.96734619e-01 7.49038875e-01 9.20090079e-01 -2.81360179e-01
-3.40159208e-01 -1.81930363e-01 2.87811458e-01 5.52391052e-01
-6.78631291e-02 -6.27804548e-02 -1.04240286e+00 5.78117669e-01
-1.80066919e+00 -8.18253577e-01 -6.39435053e-01 2.32567453e+00
3.51259202e-01 -2.16608425e-03 -1.52827293e-01 3.24991584e-01
1.80296466e-01 2.41028130e-01 -4.97867584e-01 -9.85160917e-02
-4.50350285e-01 1.72902063e-01 7.97667682e-01 6.84436560e-01
-1.15525818e+00 9.53802884e-01 5.33127880e+00 1.58786118e-01
-1.30693328e+00 7.54092708e-02 3.75410691e-02 -2.19090417e-01
-1.10464469e-01 1.24582268e-01 -8.11629832e-01 2.70737231e-01
6.59711182e-01 2.70911545e-01 2.46397048e-01 5.06734073e-01
3.10979873e-01 -6.09289408e-01 -9.92072523e-01 1.30040324e+00
-4.95289154e-02 -1.26016676e+00 -4.04028893e-01 1.09755300e-01
6.44772351e-01 7.75919616e-01 -1.30291179e-01 -1.13881938e-01
2.12007016e-01 -8.29138815e-01 6.51417494e-01 6.08392954e-01
6.78371787e-01 -4.46514517e-01 8.28336537e-01 3.41631889e-01
-1.24643457e+00 2.45806705e-02 -5.64621747e-01 -2.92031705e-01
3.16225201e-01 7.86808252e-01 -5.83524227e-01 7.63830245e-01
1.11097932e+00 1.14511621e+00 -6.87651932e-01 1.32993913e+00
-1.37973875e-01 1.61462769e-01 -7.56575525e-01 5.15094280e-01
2.95478880e-01 -3.28421682e-01 5.26604116e-01 1.19516337e+00
4.75781888e-01 -4.80996780e-02 1.58869699e-01 7.99113989e-01
2.16740578e-01 -2.53217995e-01 -1.08843780e+00 4.29759979e-01
1.62613451e-01 1.32971430e+00 -4.64576960e-01 -9.33044031e-02
-5.75581551e-01 8.92268121e-01 1.73927009e-01 1.81617305e-01
-5.87629020e-01 -1.46947771e-01 9.79695201e-01 3.10171753e-01
3.79498810e-01 -5.22913873e-01 -2.07741037e-01 -1.43546581e+00
1.88320572e-03 -2.21119210e-01 3.21215659e-01 -1.05704153e+00
-1.16843915e+00 5.43396890e-01 -2.41292696e-02 -1.60110128e+00
-9.73292738e-02 -7.85052419e-01 -5.16430974e-01 8.39543879e-01
-2.14297438e+00 -1.14420438e+00 -9.15264130e-01 8.67036164e-01
3.91541302e-01 1.53837562e-01 5.73059976e-01 4.81785536e-01
-2.70958155e-01 -8.22702497e-02 6.71078637e-02 1.08622881e-02
7.55965173e-01 -1.12245917e+00 2.35768512e-01 1.05572367e+00
2.67241746e-01 1.06861919e-01 7.16047466e-01 -4.06422734e-01
-1.69172871e+00 -9.69245434e-01 6.67354882e-01 -4.95786220e-01
4.86945897e-01 -4.35891449e-01 -7.69805253e-01 3.79905730e-01
3.86263579e-02 5.41718125e-01 1.43163562e-01 -3.32285523e-01
-2.50061512e-01 -5.23688138e-01 -1.14900362e+00 7.72420093e-02
1.15631199e+00 -6.40131056e-01 -3.45001370e-01 2.41324037e-01
4.62648034e-01 -6.65717542e-01 -7.06487954e-01 5.39592564e-01
4.57229972e-01 -1.73262596e+00 1.07693398e+00 -1.02843486e-01
2.84742802e-01 -6.35102212e-01 -4.60860699e-01 -1.19547915e+00
5.00462465e-02 -2.85477221e-01 5.20810783e-02 1.03686833e+00
-1.99228656e-02 -6.31985724e-01 6.96117342e-01 2.57631510e-01
-1.73790291e-01 -2.75229424e-01 -1.07286012e+00 -7.17130661e-01
-2.21402973e-01 -6.38302147e-01 4.65516418e-01 7.30847776e-01
-5.38280666e-01 1.90000162e-01 -4.24968719e-01 3.17865968e-01
8.07965517e-01 5.54701149e-01 1.11919808e+00 -1.30675864e+00
-6.71030283e-02 -1.82195157e-01 -7.58437335e-01 -8.06475639e-01
-1.50090479e-03 -5.85637629e-01 7.57028759e-02 -1.67698705e+00
-3.02348018e-01 -6.73825264e-01 3.09113823e-02 2.42816344e-01
1.45029038e-01 5.75719595e-01 4.32144701e-01 3.21950078e-01
-3.70235831e-01 6.26594901e-01 1.18743229e+00 8.46397653e-02
-4.95518558e-02 -6.09652922e-02 -2.20797248e-02 8.06478024e-01
7.30891168e-01 -2.99890876e-01 -2.19114438e-01 -7.58910120e-01
9.03678164e-02 6.49862085e-03 6.66053295e-01 -1.42245042e+00
2.56667614e-01 -6.59765750e-02 4.85298246e-01 -8.85858953e-01
8.27959418e-01 -1.07717621e+00 2.71719784e-01 4.62539315e-01
1.76750094e-01 1.21902533e-01 2.68953651e-01 4.16024536e-01
-2.64817148e-01 7.09013492e-02 9.52498496e-01 -4.89479452e-01
-9.50448871e-01 5.63349009e-01 -6.69977665e-02 -6.17081486e-02
7.75093973e-01 -3.69098783e-01 -2.45076045e-01 -4.55929071e-01
-2.15865657e-01 6.87407851e-02 8.33381772e-01 4.16193634e-01
8.80563378e-01 -1.16989970e+00 -7.08135962e-01 5.25828123e-01
2.16447428e-01 4.39411819e-01 1.35800868e-01 1.03300166e+00
-8.74908864e-01 4.57521051e-01 -4.33310390e-01 -9.93997931e-01
-1.17498755e+00 3.61080498e-01 5.76933086e-01 1.30353704e-01
-7.97661781e-01 7.17795491e-01 1.72909245e-01 -7.11973727e-01
-1.39397243e-02 -3.69062781e-01 -1.06361751e-02 -3.84085849e-02
3.76942039e-01 4.07423437e-01 3.24544132e-01 -1.14572716e+00
-6.12780929e-01 1.09089172e+00 4.45117831e-01 1.38252024e-02
1.32930589e+00 -3.28157514e-01 3.10205072e-02 5.24552882e-01
1.44201922e+00 -2.17666440e-02 -1.54734826e+00 -3.16827595e-01
-5.07528067e-01 -1.08061123e+00 4.84614670e-01 -4.22294468e-01
-1.45781791e+00 1.39101815e+00 7.61574745e-01 -1.19563378e-01
1.19647419e+00 -8.08069408e-02 7.11142540e-01 1.92834601e-01
6.03184521e-01 -6.30346835e-01 -1.84683904e-01 6.27444029e-01
6.64400876e-01 -1.63603365e+00 2.38158137e-01 -3.41897905e-01
-4.06404525e-01 1.27480030e+00 5.66439867e-01 -2.28313357e-01
7.06006587e-01 3.55745882e-01 3.36825401e-01 -2.44525194e-01
-4.12523955e-01 -7.31564105e-01 1.36615977e-01 6.65969610e-01
1.85353607e-01 -1.83925733e-01 4.44788277e-01 -4.28627640e-01
-2.50204712e-01 3.59690227e-02 4.99252617e-01 9.66717958e-01
-3.13217163e-01 -9.71469998e-01 -5.09421706e-01 1.40159756e-01
-4.70407195e-02 8.82169828e-02 -3.21892202e-01 9.07713294e-01
2.82989323e-01 9.55537200e-01 1.63548231e-01 -4.26492751e-01
5.50068617e-01 -4.41487193e-01 6.85958445e-01 -3.25568438e-01
-4.12254363e-01 -1.06967211e-01 -8.61768797e-02 -1.13233590e+00
-8.05534542e-01 -9.73372042e-01 -1.14906669e+00 -5.02340496e-01
-1.01631165e-01 -3.58686388e-01 7.64880836e-01 7.95026124e-01
2.21926719e-01 4.94519547e-02 6.38870835e-01 -1.36803365e+00
-5.43323047e-02 -6.70605063e-01 -5.14431775e-01 4.03225690e-01
8.24114382e-01 -9.63387251e-01 -4.49447513e-01 -1.17928134e-02] | [8.526579856872559, -2.3272433280944824] |
2a45de15-9cec-4fae-8ee1-530c404cb6fe | hard-exudate-segmentation-supplemented-by | 2211.09404 | null | https://arxiv.org/abs/2211.09404v1 | https://arxiv.org/pdf/2211.09404v1.pdf | Hard Exudate Segmentation Supplemented by Super-Resolution with Multi-scale Attention Fusion Module | Hard exudates (HE) is the most specific biomarker for retina edema. Precise HE segmentation is vital for disease diagnosis and treatment, but automatic segmentation is challenged by its large variation of characteristics including size, shape and position, which makes it difficult to detect tiny lesions and lesion boundaries. Considering the complementary features between segmentation and super-resolution tasks, this paper proposes a novel hard exudates segmentation method named SS-MAF with an auxiliary super-resolution task, which brings in helpful detailed features for tiny lesion and boundaries detection. Specifically, we propose a fusion module named Multi-scale Attention Fusion (MAF) module for our dual-stream framework to effectively integrate features of the two tasks. MAF first adopts split spatial convolutional (SSC) layer for multi-scale features extraction and then utilize attention mechanism for features fusion of the two tasks. Considering pixel dependency, we introduce region mutual information (RMI) loss to optimize MAF module for tiny lesions and boundary detection. We evaluate our method on two public lesion datasets, IDRiD and E-Ophtha. Our method shows competitive performance with low-resolution inputs, both quantitatively and qualitatively. On E-Ophtha dataset, the method can achieve $\geq3\%$ higher dice and recall compared with the state-of-the-art methods. | ['Jiang Liu', 'Yan Hu', 'Mingming Yang', 'Zhongxi Qiu', 'Xiaoshan Chen', 'Jiayi Zhang'] | 2022-11-17 | null | null | null | null | ['boundary-detection'] | ['computer-vision'] | [ 1.89555854e-01 -2.27841213e-01 -7.66958296e-02 -2.30897307e-01
-9.61865246e-01 -1.43245563e-01 1.05569243e-01 -6.87525654e-03
-4.90287006e-01 5.60311139e-01 2.31079906e-01 -1.64365396e-01
-1.21597148e-01 -6.28090799e-01 -3.31726134e-01 -7.13036895e-01
1.54123276e-01 -1.23393899e-02 7.48702526e-01 8.61888230e-02
2.28198469e-01 6.36404276e-01 -1.54183698e+00 4.69143808e-01
1.44174182e+00 1.33225954e+00 5.01872420e-01 6.86217427e-01
-2.19780982e-01 6.17183268e-01 -1.22111022e-01 -2.21634656e-01
3.25434804e-01 -4.03430343e-01 -8.23205113e-01 3.38097848e-02
5.48956037e-01 -4.63782638e-01 -2.53685206e-01 1.16635776e+00
8.21469963e-01 -3.05582404e-01 8.15112829e-01 -6.04686379e-01
-6.08664095e-01 6.49283752e-02 -1.08894193e+00 7.83865392e-01
-2.33577341e-01 5.05834639e-01 6.84166431e-01 -7.60945559e-01
4.70391005e-01 9.47176337e-01 5.41696727e-01 2.57190645e-01
-8.59230936e-01 -5.30879855e-01 -1.71140134e-02 3.95426601e-01
-1.27983415e+00 -2.55377203e-01 2.88472801e-01 -5.97916842e-01
5.92481434e-01 2.91073412e-01 6.68871641e-01 5.20584285e-01
3.07035387e-01 8.57511818e-01 1.66475463e+00 -5.90264462e-02
-1.04867868e-01 -7.73409083e-02 2.07747206e-01 8.05135131e-01
2.80604392e-01 -2.12419584e-01 6.45776317e-02 8.68764296e-02
1.14496171e+00 8.29411000e-02 -7.61062205e-01 1.55555055e-01
-1.08784890e+00 4.90873128e-01 8.55872989e-01 1.29294172e-01
-4.50849593e-01 -4.85381372e-02 1.41618997e-01 -2.39499778e-01
3.25935692e-01 3.01538467e-01 -1.89408377e-01 1.44443527e-01
-7.63297737e-01 1.10777341e-01 7.94188827e-02 3.89077663e-01
4.62622702e-01 -5.58646023e-01 -8.25967550e-01 9.57376063e-01
9.96241570e-02 3.97024065e-01 5.86457491e-01 -7.81221807e-01
2.90372342e-01 9.16091859e-01 9.59255323e-02 -6.17759407e-01
-4.63857263e-01 -5.95506608e-01 -9.31408286e-01 4.07665223e-01
3.67945999e-01 -1.72855690e-01 -1.35197294e+00 1.26257324e+00
4.38701302e-01 2.55018592e-01 -3.55987519e-01 1.31693161e+00
1.12658787e+00 2.97144145e-01 1.02861166e-01 -2.88645476e-01
1.53718781e+00 -1.03183806e+00 -6.36689305e-01 -2.11163625e-01
4.25857365e-01 -7.92570531e-01 1.19552445e+00 1.29783198e-01
-1.27163804e+00 -3.28493685e-01 -9.01474655e-01 -4.93712634e-01
-1.52703449e-01 4.54824209e-01 6.01503730e-01 1.37522697e-01
-1.08683884e+00 2.95718968e-01 -7.42347896e-01 -2.95826226e-01
1.11455703e+00 3.92779708e-01 -2.03001231e-01 -1.35668471e-01
-7.90710092e-01 7.87692070e-01 -7.32388496e-02 1.89634323e-01
-3.04920793e-01 -8.43221426e-01 -4.38454449e-01 -7.65153468e-02
1.20526008e-01 -9.36956108e-01 9.85679209e-01 -4.23670888e-01
-1.31647408e+00 9.47152197e-01 -5.24330616e-01 -3.44755799e-01
5.85284948e-01 -2.45547459e-01 -3.28165710e-01 5.72714388e-01
2.04032108e-01 6.81336284e-01 7.49561727e-01 -9.85888422e-01
-9.31048393e-01 -5.87680697e-01 -9.75673497e-02 3.10023010e-01
-3.13507877e-02 2.01814532e-01 -5.65828145e-01 -4.06538785e-01
2.07956716e-01 -4.31449145e-01 -3.74098182e-01 2.93244809e-01
-4.61585104e-01 -1.93134129e-01 7.47042179e-01 -9.11861777e-01
1.47140598e+00 -1.77615273e+00 1.07061155e-02 8.62383172e-02
8.30148160e-01 6.18248582e-01 -1.09787159e-01 -5.63717425e-01
1.58066619e-02 2.19059095e-01 -3.43869805e-01 -1.94872320e-01
-5.44515610e-01 -1.59366921e-01 3.11581586e-02 4.41383660e-01
4.65214103e-01 1.30027640e+00 -6.06355608e-01 -7.37383544e-01
1.76634133e-01 5.99198341e-01 -2.33972281e-01 7.78540522e-02
5.08705527e-02 4.03430402e-01 -6.34028852e-01 8.77186239e-01
8.02503884e-01 -7.13322580e-01 -5.34411490e-01 -5.30640483e-01
-2.56461471e-01 -1.60178125e-01 -1.02897763e+00 1.33650064e+00
-4.33206260e-01 4.97893244e-01 1.35731772e-01 -5.17689586e-01
6.52524412e-01 -6.45226762e-02 4.58722562e-01 -6.27757728e-01
4.93686438e-01 3.63378167e-01 1.53978318e-01 -8.25192332e-01
-1.50022991e-02 5.69017082e-02 4.24961716e-01 1.23446479e-01
-3.71149540e-01 1.27599895e-01 6.96621612e-02 6.55501187e-02
1.07837903e+00 -1.25469550e-01 3.33446294e-01 4.49733697e-02
6.42829359e-01 -3.18272263e-01 5.37036479e-01 4.26208079e-01
-5.72070718e-01 1.04927719e+00 6.29127443e-01 -1.92490265e-01
-8.94176662e-01 -1.13485694e+00 -5.25588155e-01 3.86550695e-01
4.88720119e-01 -8.38030651e-02 -8.00616801e-01 -7.54421890e-01
1.03849076e-01 -2.33308807e-01 -9.29194748e-01 1.34700298e-01
-3.68217260e-01 -1.03901756e+00 2.69064665e-01 6.01396203e-01
9.76390898e-01 -8.11147869e-01 -6.61437333e-01 1.06764391e-01
-1.84891745e-01 -1.13416553e+00 -6.84214950e-01 -3.10982257e-01
-6.23623371e-01 -1.22872615e+00 -1.08284450e+00 -7.48235047e-01
5.96221805e-01 4.45427030e-01 5.81601441e-01 8.58442783e-02
-9.50740635e-01 -1.61713883e-01 -2.29624391e-01 -2.98898637e-01
1.60981908e-01 -7.41729289e-02 -4.53294009e-01 3.58492225e-01
2.94817984e-01 -5.82491696e-01 -1.27185500e+00 2.30850592e-01
-6.96490407e-01 1.19151674e-01 1.37021542e+00 7.37331629e-01
1.05068040e+00 -1.90299347e-01 5.28240144e-01 -8.37199032e-01
5.97460032e-01 -3.82998019e-01 -3.75381231e-01 3.82152766e-01
-5.21986127e-01 -2.88884729e-01 2.09412336e-01 -2.05014497e-01
-8.44566226e-01 -4.35672924e-02 1.29156768e-01 -5.64316392e-01
-2.04381004e-01 2.13581935e-01 -1.83301404e-01 -3.71817201e-01
6.57050133e-01 1.99353054e-01 2.80111164e-01 -5.07836163e-01
6.98847771e-02 1.08946419e+00 8.02935302e-01 -3.18404869e-03
3.39521796e-01 6.86923265e-01 1.76740602e-01 -5.73093474e-01
-1.17663872e+00 -7.40062535e-01 -5.11587679e-01 -1.08555801e-01
1.08595157e+00 -7.91360438e-01 -7.19970465e-01 8.26374710e-01
-1.06712174e+00 -1.74734220e-01 -1.18084110e-01 4.01641428e-01
-3.09078574e-01 3.85026872e-01 -6.43453717e-01 -4.78474587e-01
-6.64837182e-01 -1.41644108e+00 1.16896904e+00 8.72831345e-01
3.79327774e-01 -5.74676037e-01 -2.93562114e-01 7.52036035e-01
6.10363662e-01 4.58453983e-01 8.96500945e-01 -7.47512206e-02
-7.74296761e-01 -1.14728130e-01 -1.23575675e+00 5.71604609e-01
2.15712562e-01 -9.22244936e-02 -9.77656901e-01 -7.39372149e-02
-2.92782307e-01 -7.87003562e-02 1.36610055e+00 9.64875102e-01
1.53006387e+00 -5.77974468e-02 -4.08937097e-01 9.66530263e-01
1.48329389e+00 -1.50477096e-01 9.48836207e-01 1.47789672e-01
7.80656695e-01 4.40601796e-01 5.91141284e-01 3.36614102e-01
5.47943473e-01 4.28614885e-01 3.72070462e-01 -6.07226193e-01
-7.24427164e-01 2.76536822e-01 -1.69780441e-02 3.05016160e-01
-3.17630321e-01 6.41415045e-02 -8.67761910e-01 6.76355839e-01
-1.78958631e+00 -6.70582294e-01 -4.39220339e-01 2.05636406e+00
9.49389160e-01 -7.95937032e-02 1.29867494e-01 -2.02991068e-01
9.31920588e-01 -4.38150368e-04 -8.64546359e-01 7.02423677e-02
-3.48040134e-01 5.63225627e-01 6.45690739e-01 3.67731899e-01
-1.32657278e+00 8.23677242e-01 5.14227057e+00 1.08613729e+00
-1.29699779e+00 2.76096970e-01 8.51809621e-01 -3.42270464e-01
-3.27616483e-02 -2.91734308e-01 -7.84455180e-01 6.67330980e-01
3.77203017e-01 4.72114868e-02 1.83045045e-01 1.40742034e-01
3.63690257e-01 -3.80970180e-01 -5.68463624e-01 1.07884717e+00
-9.32505429e-02 -1.42673135e+00 1.03185803e-01 1.82463735e-01
7.43505597e-01 3.25000495e-01 3.00541490e-01 -2.02324525e-01
-8.63966495e-02 -1.34914315e+00 -4.21463251e-02 8.97789538e-01
1.14548385e+00 -6.48491681e-01 1.04367280e+00 -1.58874050e-01
-1.22296321e+00 -2.72898227e-01 -3.54930729e-01 3.56306881e-01
3.76716070e-02 8.12150240e-01 -5.53845704e-01 4.23011363e-01
8.03256929e-01 9.32755291e-01 -8.78776252e-01 1.80741751e+00
-1.87756300e-01 4.01926279e-01 -2.57552415e-01 2.17394844e-01
1.26009062e-01 -2.28393823e-01 6.53705895e-01 1.01657987e+00
3.36436778e-01 4.13691729e-01 1.04414515e-01 1.06029356e+00
6.67081848e-02 4.40344632e-01 4.03042696e-02 2.33248994e-01
3.87482673e-01 1.48105609e+00 -6.11873984e-01 -2.30032980e-01
-3.61374080e-01 9.28433061e-01 2.33011872e-01 4.40305322e-01
-6.99649096e-01 -6.39525652e-01 7.57924259e-01 5.99666357e-01
1.78198785e-01 2.34914571e-01 -5.78029096e-01 -9.88782883e-01
2.49330595e-01 -4.59604412e-01 3.48894209e-01 -8.49901617e-01
-1.27615988e+00 5.38132787e-01 -5.43673515e-01 -1.19797325e+00
5.01283467e-01 -4.47111517e-01 -7.83990979e-01 1.31505859e+00
-2.06694627e+00 -1.40154779e+00 -8.90378773e-01 5.42239785e-01
2.19456166e-01 6.34245798e-02 2.85925418e-01 2.94749647e-01
-8.46283495e-01 5.44657230e-01 -8.49781409e-02 8.57074633e-02
8.97591531e-01 -1.25839186e+00 6.72099143e-02 7.99260676e-01
-6.27864838e-01 3.28793257e-01 1.62890665e-02 -7.16322422e-01
-7.01033711e-01 -1.45191491e+00 4.40238059e-01 -3.35322589e-01
5.06897867e-01 2.06015840e-01 -1.08644116e+00 1.93057448e-01
-2.65853852e-01 6.11172199e-01 3.95826429e-01 -5.03411174e-01
-1.23588033e-01 -9.65041220e-02 -1.26819563e+00 5.29498518e-01
9.14117336e-01 -3.26249063e-01 -2.94820130e-01 3.72036099e-01
8.22443306e-01 -6.49410367e-01 -1.09906209e+00 8.32867861e-01
4.26910013e-01 -1.06175911e+00 8.56572926e-01 -3.93909127e-01
6.16660058e-01 -4.13987935e-01 1.41722322e-01 -8.12426329e-01
-4.33993667e-01 -5.16367316e-01 -1.10061606e-02 9.25525129e-01
2.41634607e-01 -8.11257064e-01 6.58814013e-01 2.92459846e-01
-3.37160408e-01 -1.47022581e+00 -7.42562413e-01 -2.67473251e-01
7.49864206e-02 -6.23858310e-02 5.82221448e-01 5.62896252e-01
-4.50435877e-01 1.37765720e-01 1.17610373e-01 3.61144662e-01
6.57627225e-01 2.30108514e-01 4.52522695e-01 -1.30726957e+00
-5.08847423e-02 -8.40673566e-01 -7.00227678e-01 -8.59092414e-01
-2.62773335e-01 -8.85819793e-01 -3.09003353e-01 -1.87356484e+00
5.19035637e-01 -3.87842745e-01 -4.74518448e-01 4.50444102e-01
-5.12038350e-01 4.93611574e-01 -1.41781047e-01 3.64349544e-01
-4.98166978e-01 3.37507993e-01 1.96309841e+00 -3.10290419e-02
-4.58022714e-01 -7.66686490e-03 -9.01829898e-01 7.61414647e-01
6.61209822e-01 1.54654771e-01 -2.34096810e-01 -2.41419822e-01
-2.53990054e-01 -1.24392472e-01 6.57789528e-01 -1.21596539e+00
3.05139005e-01 -7.07774609e-02 5.86367190e-01 -5.68234026e-01
1.57298714e-01 -1.90356359e-01 -5.06146610e-01 2.77581871e-01
-1.41372725e-01 -6.02315128e-01 1.51976764e-01 5.76520562e-01
-1.64589360e-01 2.20443875e-01 1.33730638e+00 -9.28939506e-03
-6.02787912e-01 7.43584275e-01 -2.46123374e-02 2.44938567e-01
1.13323820e+00 -3.98805320e-01 -7.13235557e-01 2.61983834e-02
-8.05694222e-01 4.84681785e-01 5.57063162e-01 2.09601000e-01
9.33160901e-01 -9.03939188e-01 -1.11781681e+00 1.63842276e-01
2.10078254e-01 4.48522538e-01 6.62308514e-01 1.60534227e+00
-5.80480278e-01 2.73702651e-01 -2.72709250e-01 -7.13372231e-01
-1.39047742e+00 2.71923281e-02 7.02163637e-01 -1.99539095e-01
-7.69829690e-01 1.04993927e+00 4.07729536e-01 2.41102204e-01
8.46488327e-02 -6.78332210e-01 -4.21941549e-01 -6.38439273e-03
8.57049286e-01 4.12584215e-01 8.57255384e-02 -5.56460917e-01
-1.08231597e-01 9.61826921e-01 -4.94969726e-01 4.32575792e-01
1.15481198e+00 -3.82633150e-01 -3.92175525e-01 -1.70410499e-02
1.07283032e+00 -1.96299896e-01 -1.20792699e+00 -4.41113144e-01
-4.03245926e-01 -6.33523285e-01 5.87378383e-01 -1.15977907e+00
-1.26990831e+00 1.09180260e+00 1.10612035e+00 -3.28905195e-01
1.27921367e+00 1.02879405e-01 1.31677389e+00 -2.86052674e-01
6.10529780e-02 -7.39023805e-01 -1.27606988e-01 5.26125208e-02
8.08285534e-01 -1.38783073e+00 2.81379092e-03 -7.79485404e-01
-6.64223731e-01 8.13351274e-01 8.55687916e-01 -8.68658498e-02
6.14650488e-01 9.01079401e-02 3.40244733e-02 -2.05393672e-01
-5.05117655e-01 -6.98702037e-01 8.03045988e-01 6.03277326e-01
2.35767469e-01 6.57309964e-02 -4.98609334e-01 9.60348129e-01
2.01205283e-01 3.06651324e-01 6.88822269e-01 4.30659115e-01
-8.10899138e-01 -7.18748510e-01 -2.43238769e-02 1.18028033e+00
-6.20617449e-01 -1.71973199e-01 -3.39080900e-01 6.48693323e-01
4.69340712e-01 5.36623180e-01 1.62138790e-01 -2.59136826e-01
1.20663971e-01 -2.97552347e-01 4.28741276e-01 -5.73489487e-01
-5.03552616e-01 1.63067132e-01 -2.41791919e-01 -8.09958875e-01
-1.74356118e-01 -5.09626567e-01 -1.50911129e+00 1.04453139e-01
-1.71057120e-01 -2.99191236e-01 3.34033608e-01 7.98239768e-01
5.65392971e-01 5.69403589e-01 6.10797346e-01 -5.81731200e-01
-2.21752182e-01 -8.86598170e-01 -7.89493561e-01 2.61040002e-01
5.67959130e-01 -7.58386016e-01 -2.70337939e-01 -9.86019894e-02] | [15.774075508117676, -3.947934150695801] |
2b7bd203-7196-40fe-bf48-bc3e9900c1b6 | cased-curriculum-adaptive-sampling-for | 1807.10819 | null | http://arxiv.org/abs/1807.10819v1 | http://arxiv.org/pdf/1807.10819v1.pdf | CASED: Curriculum Adaptive Sampling for Extreme Data Imbalance | We introduce CASED, a novel curriculum sampling algorithm that facilitates
the optimization of deep learning segmentation or detection models on data sets
with extreme class imbalance. We evaluate the CASED learning framework on the
task of lung nodule detection in chest CT. In contrast to two-stage solutions,
wherein nodule candidates are first proposed by a segmentation model and
refined by a second detection stage, CASED improves the training of deep nodule
segmentation models (e.g. UNet) to the point where state of the art results are
achieved using only a trivial detection stage. CASED improves the optimization
of deep segmentation models by allowing them to first learn how to distinguish
nodules from their immediate surroundings, while continuously adding a greater
proportion of difficult-to-classify global context, until uniformly sampling
from the empirical data distribution. Using CASED during training yields a
minimalist proposal to the lung nodule detection problem that tops the LUNA16
nodule detection benchmark with an average sensitivity score of 88.35%.
Furthermore, we find that models trained using CASED are robust to nodule
annotation quality by showing that comparable results can be achieved when only
a point and radius for each ground truth nodule are provided during training.
Finally, the CASED learning framework makes no assumptions with regard to
imaging modality or segmentation target and should generalize to other medical
imaging problems where class imbalance is a persistent problem. | ['Nicolas Chapados', 'Florian Soudan', 'Andrew Jesson', 'Nicolas Guizard', 'Sina Hamidi Ghalehjegh', 'Damien Goblot'] | 2018-07-27 | null | null | null | null | ['lung-nodule-detection'] | ['medical'] | [ 3.79733950e-01 7.01691091e-01 -4.72699016e-01 -2.42643535e-01
-1.17753363e+00 -5.17174602e-01 4.34499830e-01 2.12822437e-01
-4.14049685e-01 3.94552290e-01 -1.77809149e-01 -6.23276651e-01
-1.93063572e-01 -6.07907116e-01 -6.22028589e-01 -8.92070174e-01
8.04602727e-02 1.17298126e+00 7.69818544e-01 1.88313931e-01
-2.47875467e-01 4.86696035e-01 -1.10862374e+00 4.12511766e-01
6.46725655e-01 9.79496717e-01 3.11406791e-01 8.15666020e-01
1.21017739e-01 6.12527549e-01 -4.56679374e-01 -1.32535219e-01
5.33950210e-01 -4.94355500e-01 -9.75942910e-01 4.47345674e-01
6.57786846e-01 -4.53348517e-01 -1.34664522e-02 7.62068272e-01
7.40992785e-01 -4.03150380e-01 9.75328684e-01 -7.74118781e-01
5.67171301e-05 5.68164527e-01 -6.51059985e-01 3.28382999e-01
-2.66103059e-01 3.41445684e-01 9.59670901e-01 -6.72002137e-01
5.48039436e-01 6.36109889e-01 8.57025504e-01 7.43980050e-01
-1.19027150e+00 -2.34972388e-01 -1.11118928e-01 -2.34575793e-01
-1.02820253e+00 1.43696159e-01 1.14881977e-01 -6.41691208e-01
6.33855045e-01 5.63514292e-01 1.01872861e+00 7.26609945e-01
1.44979715e-01 9.43671465e-01 7.65305042e-01 -3.32512885e-01
1.56155869e-01 2.91066796e-01 1.40552998e-01 9.50911701e-01
7.09079742e-01 4.11749743e-02 3.25352699e-01 -1.30543396e-01
9.78015602e-01 -6.65731952e-02 -1.48327962e-01 -9.54028904e-01
-1.24364889e+00 8.68178844e-01 8.08140039e-01 4.02223945e-01
-3.72867286e-01 2.44680509e-01 4.51248050e-01 1.35747381e-02
2.40807235e-01 6.82766140e-01 -2.71671772e-01 5.06358266e-01
-1.31752360e+00 2.54478753e-01 8.67844343e-01 7.15206861e-01
4.36329633e-01 -3.14604104e-01 -6.52162910e-01 4.53423709e-01
1.90944165e-01 2.19623283e-01 5.87478995e-01 -8.70766640e-01
-4.05602045e-02 6.82951093e-01 -4.93444912e-02 -3.25879008e-01
-7.59146690e-01 -1.06426275e+00 -6.96489930e-01 4.31222528e-01
8.14436853e-01 -1.41228914e-01 -1.30652237e+00 1.24481535e+00
5.32710791e-01 6.58457428e-02 -1.45239681e-01 8.08290184e-01
8.49368751e-01 6.15240587e-03 -3.72814643e-03 -1.78003348e-02
1.56189060e+00 -1.07463837e+00 -1.55602723e-01 -2.36855179e-01
1.08709431e+00 -6.40853465e-01 1.02493310e+00 2.60318726e-01
-1.19258940e+00 -5.13756335e-01 -1.02639008e+00 2.69680731e-02
2.29206905e-02 2.72678614e-01 4.42348272e-01 8.44074726e-01
-1.27445996e+00 3.91290575e-01 -1.18389153e+00 -3.95309776e-01
8.87302339e-01 6.41166866e-01 1.33156657e-01 -3.48181911e-02
-6.71659887e-01 1.11206031e+00 5.28380573e-01 -2.60131210e-01
-1.22467327e+00 -1.21789646e+00 -5.27881861e-01 -1.57558359e-02
6.47772133e-01 -9.65751052e-01 1.75063980e+00 -9.94388938e-01
-1.03181148e+00 1.16624963e+00 7.44557083e-02 -7.68178105e-01
8.70005608e-01 3.51882309e-01 2.61893779e-01 2.30342001e-01
1.54272020e-01 9.13165748e-01 6.79208219e-01 -1.06540775e+00
-7.31523454e-01 8.89237225e-02 -1.63816214e-02 2.89174974e-01
2.16227368e-01 -4.70262319e-01 -5.16682267e-01 -2.57322311e-01
2.00901762e-01 -1.08396924e+00 -7.76132226e-01 3.45769316e-01
-5.19293904e-01 -1.41038433e-01 8.42382133e-01 -3.34912717e-01
8.67758870e-01 -1.82746255e+00 -1.72131851e-01 4.84954506e-01
7.19815850e-01 1.70306504e-01 1.95298180e-01 -5.31251848e-01
-1.77162200e-01 3.20796520e-01 -3.83353859e-01 8.57221633e-02
-1.78470667e-02 3.05067867e-01 4.21862900e-01 4.45729703e-01
4.19117719e-01 1.13547635e+00 -8.61985505e-01 -9.44432378e-01
2.68390864e-01 1.49382085e-01 -9.18721378e-01 3.97804528e-02
-3.88831198e-01 2.96437681e-01 -4.19607639e-01 5.09769380e-01
4.79307055e-01 -9.00636315e-01 1.73023880e-01 -1.36909917e-01
1.34785369e-01 2.41946802e-01 -1.00070846e+00 1.33858097e+00
-3.29572916e-01 4.97539937e-01 9.79555994e-02 -8.70473921e-01
4.48752850e-01 5.25167227e-01 8.31593752e-01 -3.42441380e-01
2.08140686e-01 5.47989905e-01 7.84558117e-01 -5.00311315e-01
-9.50313956e-02 -2.54713237e-01 2.23796919e-01 5.57035685e-01
9.71841738e-02 -6.01266801e-01 3.22324723e-01 9.24802497e-02
1.40119612e+00 -2.15471610e-01 6.11288249e-01 -5.13327599e-01
2.96155632e-01 4.88864601e-01 2.02721879e-01 1.05329299e+00
-3.43811780e-01 8.98171127e-01 8.74980867e-01 -2.37850383e-01
-1.13035703e+00 -1.06139588e+00 -6.74631119e-01 8.36411655e-01
-5.48186675e-02 9.13170353e-02 -9.42781448e-01 -1.26921713e+00
-1.50226265e-01 4.53027725e-01 -1.06986248e+00 1.13485612e-01
-6.76019490e-01 -1.15900218e+00 4.16582346e-01 5.02970576e-01
2.54578769e-01 -8.30918968e-01 -1.06169724e+00 1.53412476e-01
5.03838696e-02 -7.45028079e-01 -3.28868598e-01 7.10854113e-01
-1.03399253e+00 -1.40532184e+00 -9.70299602e-01 -9.62058723e-01
8.45231652e-01 2.31836010e-02 1.63398051e+00 2.86512762e-01
-7.93618977e-01 3.90225381e-01 2.59412732e-02 -5.82726896e-01
-7.19847620e-01 5.09574592e-01 -6.64636910e-01 -5.49386621e-01
1.60776898e-01 3.64392661e-02 -8.59439790e-01 4.52711642e-01
-8.41939986e-01 7.30074197e-02 1.02661180e+00 1.00102341e+00
9.02554989e-01 -1.13951735e-01 2.44692117e-01 -1.17585790e+00
6.35598078e-02 -5.22087276e-01 -3.60886425e-01 1.37659594e-01
-3.77469569e-01 -9.01719779e-02 8.55396688e-02 -4.16285574e-01
-8.43447566e-01 3.81110966e-01 -1.90369740e-01 -2.47556224e-01
-3.22323501e-01 2.40768552e-01 2.55218118e-01 -2.06866607e-01
1.14638698e+00 -1.76544249e-01 2.65141696e-01 -2.86414800e-03
6.65690890e-03 2.35464666e-02 4.15160865e-01 -2.28789598e-01
6.98880136e-01 5.47513247e-01 1.12835117e-01 -5.63388348e-01
-1.18677163e+00 -7.01454341e-01 -8.80645156e-01 -1.62275538e-01
1.04419720e+00 -6.67407930e-01 -1.77766442e-01 -6.31429330e-02
-6.24420881e-01 -6.80653989e-01 -1.02641582e+00 5.01246333e-01
-6.49824798e-01 -9.56044495e-02 -5.84915936e-01 -2.45235130e-01
-1.12051569e-01 -1.39750028e+00 1.17594421e+00 1.17813334e-01
-3.01772803e-01 -1.18398237e+00 -1.23224735e-01 7.25945532e-02
4.57094461e-01 3.83109093e-01 1.08448589e+00 -1.21469796e+00
-7.51463294e-01 -2.99350590e-01 -2.47161746e-01 1.02432974e-01
1.74909934e-01 -9.30620655e-02 -8.81726682e-01 -3.59744608e-01
3.23955297e-01 -3.86421531e-01 9.36009407e-01 8.81873906e-01
1.42040098e+00 7.72038996e-02 -7.54629970e-01 6.03049397e-01
1.25586653e+00 3.32117430e-03 1.64956719e-01 2.31315911e-01
5.70186734e-01 4.15263027e-01 3.97432774e-01 2.35100985e-02
-1.80276126e-01 2.97895640e-01 9.38107491e-01 -7.40443051e-01
-5.98855972e-01 2.68704802e-01 -3.98288727e-01 2.56429821e-01
1.87706947e-01 -2.26074442e-01 -1.30176413e+00 6.33674562e-01
-1.33472371e+00 -6.60290658e-01 -3.13872844e-01 1.95228457e+00
8.58832598e-01 5.15816033e-01 1.64415807e-01 1.06307432e-01
4.45347160e-01 -4.97013219e-02 -7.35854983e-01 1.28978223e-01
3.06269675e-01 4.35029089e-01 7.50943005e-01 4.96954679e-01
-1.42102242e+00 4.33081806e-01 6.74830627e+00 8.81393254e-01
-1.20020068e+00 2.57759094e-01 1.00739932e+00 -2.51730114e-01
-2.19369709e-01 -3.39621127e-01 -7.58693039e-01 -8.49362835e-03
5.88591933e-01 8.77270177e-02 -3.59570265e-01 8.85941863e-01
-3.95630673e-02 -2.26730123e-01 -1.52224350e+00 4.45539951e-01
-1.34934247e-01 -1.41478121e+00 -2.68284172e-01 1.86147839e-01
9.43866611e-01 2.81490058e-01 2.22547710e-01 3.42582405e-01
3.84727210e-01 -1.23029768e+00 1.68728516e-01 1.67236015e-01
7.16814578e-01 -3.07313472e-01 9.12144005e-01 5.47971606e-01
-7.73011267e-01 1.09423734e-01 -3.60833824e-01 5.75115860e-01
-2.55109608e-01 6.44174218e-01 -1.88317454e+00 3.37903947e-01
4.55311745e-01 4.12096530e-01 -9.02789533e-01 1.54467297e+00
1.39361858e-01 9.18240786e-01 -5.12495697e-01 6.10574149e-02
5.55193603e-01 2.55779117e-01 4.26573217e-01 1.31375849e+00
2.46934056e-01 -1.31078005e-01 2.98842698e-01 7.85210013e-01
1.02971338e-01 9.08477381e-02 -2.72006780e-01 6.11064255e-01
4.73644808e-02 1.32847857e+00 -1.17152750e+00 -4.37664866e-01
-2.42654115e-01 5.21036685e-01 -1.19235501e-01 5.39080165e-02
-9.43365037e-01 1.50818661e-01 -7.46848574e-03 6.37738824e-01
6.13615215e-01 3.48643273e-01 -4.29155082e-01 -6.37869656e-01
-2.43083134e-01 -8.40808630e-01 4.49151069e-01 -3.21669728e-01
-1.03704166e+00 4.71575528e-01 1.09445699e-03 -1.29837310e+00
-4.28639412e-01 -9.31683660e-01 -6.83855653e-01 6.88168466e-01
-1.43461728e+00 -8.85917723e-01 -4.76941258e-01 3.08171898e-01
5.38644373e-01 1.59340352e-01 5.89989245e-01 8.89649317e-02
-1.95800662e-01 5.07160544e-01 -5.86204864e-02 2.25841612e-01
6.58911526e-01 -1.80376136e+00 1.88671485e-01 3.76328975e-01
-1.01915091e-01 -9.08134785e-03 4.55530047e-01 -4.86441672e-01
-6.65124953e-01 -1.40771163e+00 3.77444804e-01 -6.78338945e-01
5.49177051e-01 -1.38843134e-01 -8.94504547e-01 6.74581707e-01
2.85404138e-02 5.76041460e-01 5.72326541e-01 -1.93177968e-01
2.24682361e-01 9.03861150e-02 -1.36422873e+00 4.48162287e-01
7.07412720e-01 8.07585078e-04 -3.92359346e-01 7.28931129e-01
5.63362479e-01 -9.44768488e-01 -7.36946821e-01 6.70472443e-01
2.89114445e-01 -9.71556723e-01 9.40055907e-01 -4.88348275e-01
2.78298736e-01 -1.30819589e-01 2.49884203e-01 -9.59090829e-01
-3.99407148e-01 -1.08725935e-01 5.77218011e-02 4.07030016e-01
8.30416620e-01 -2.48735443e-01 1.40165997e+00 2.70426422e-01
-3.55118424e-01 -1.15039062e+00 -8.34342420e-01 -4.26418036e-01
3.60101521e-01 -2.25101978e-01 1.45147592e-01 7.44040191e-01
-7.38645971e-01 -6.43122494e-02 3.77438098e-01 9.44598764e-02
6.57155335e-01 6.30362928e-02 4.38868195e-01 -1.31504774e+00
-6.60205305e-01 -8.16842854e-01 -2.19914645e-01 -1.09853232e+00
-3.84148955e-01 -1.10459054e+00 1.63118571e-01 -1.63184643e+00
4.63409126e-01 -5.78620851e-01 -2.25488186e-01 3.46890152e-01
-2.57095158e-01 3.87689799e-01 -1.46490991e-01 2.46052831e-01
-5.12176275e-01 -1.79051325e-01 1.80021012e+00 -1.05878264e-01
-2.89023407e-02 7.68952072e-01 -7.19356239e-01 9.72506821e-01
6.94479644e-01 -6.01918340e-01 -4.47777182e-01 -9.92609262e-02
-4.56430130e-02 1.23282738e-01 5.99896789e-01 -1.06209087e+00
1.29769325e-01 7.49109611e-02 7.39903688e-01 -8.29462886e-01
5.84642850e-02 -9.12713110e-01 -1.52286664e-01 1.12663484e+00
-5.07589042e-01 -3.83721292e-01 1.91812962e-01 4.62278098e-01
1.91075150e-02 -6.40217483e-01 1.19408023e+00 -5.81678629e-01
-2.45962009e-01 3.87057811e-01 -5.11609733e-01 2.60339200e-01
1.21282053e+00 -4.65166152e-01 -1.33402854e-01 -8.79729167e-03
-1.27865124e+00 4.17851120e-01 3.33498448e-01 -2.50295997e-01
2.84926668e-02 -1.09823668e+00 -8.98039460e-01 7.45738223e-02
-1.85997620e-01 7.55840361e-01 1.29818637e-02 1.30023777e+00
-7.81925023e-01 4.73482072e-01 1.87950984e-01 -1.30224431e+00
-1.12563407e+00 3.71467590e-01 1.22568297e+00 -8.78545642e-01
-6.67969882e-01 1.03369510e+00 6.18315399e-01 -5.49121439e-01
4.99432474e-01 -7.19335914e-01 2.74842922e-02 2.15280987e-02
8.32084417e-02 4.04366031e-02 5.15297472e-01 7.78392106e-02
-1.99898168e-01 1.98448911e-01 -3.78531903e-01 1.22150339e-01
8.60333741e-01 3.53909165e-01 2.42956012e-01 2.99561560e-01
8.71160507e-01 -1.71512127e-01 -1.33143198e+00 -2.10624814e-01
2.30892941e-01 1.50598707e-02 1.03298262e-01 -1.06056941e+00
-1.20730865e+00 7.93216109e-01 9.04091179e-01 4.07497972e-01
1.01277292e+00 3.88671756e-01 3.00708175e-01 3.94620359e-01
-3.70413274e-01 -5.52010179e-01 4.12898362e-01 4.48372185e-01
6.22306406e-01 -1.50155509e+00 1.65516302e-01 -4.33261395e-01
-5.00731051e-01 1.11014152e+00 9.18567121e-01 -3.51334542e-01
7.57863164e-01 6.42381787e-01 1.67966887e-01 -3.35476458e-01
-8.09641838e-01 -2.45428920e-01 5.76996386e-01 6.05678618e-01
5.39707124e-01 2.81470325e-02 -1.22391567e-01 3.72966647e-01
-1.82179183e-01 -3.62271786e-01 4.72172558e-01 8.67312491e-01
-7.68722415e-01 -9.32979107e-01 -4.11671609e-01 1.05010653e+00
-6.29570544e-01 -2.89216894e-03 -4.16093618e-01 1.48068309e+00
3.45381349e-01 6.98094070e-02 4.03589100e-01 3.60636443e-01
2.23875031e-01 8.69298878e-04 4.97768611e-01 -1.11263454e+00
-9.32928681e-01 5.97917438e-01 -1.78769425e-01 -3.28221977e-01
-2.36624733e-01 -8.90898287e-01 -1.20095372e+00 3.14388394e-01
-5.58323026e-01 1.02633186e-01 3.00917327e-01 8.76204669e-01
-2.35977873e-01 1.04347360e+00 2.97531635e-01 -8.24206293e-01
-7.97327995e-01 -7.07984388e-01 -3.97847027e-01 3.12181935e-02
6.03977203e-01 -3.34225982e-01 -2.79683709e-01 -7.85881877e-02] | [15.331160545349121, -2.1710824966430664] |
3a2df2c9-13f6-4ee2-806b-4da531b05405 | taqyim-evaluating-arabic-nlp-tasks-using | 2306.16322 | null | https://arxiv.org/abs/2306.16322v1 | https://arxiv.org/pdf/2306.16322v1.pdf | Taqyim: Evaluating Arabic NLP Tasks Using ChatGPT Models | Large language models (LLMs) have demonstrated impressive performance on various downstream tasks without requiring fine-tuning, including ChatGPT, a chat-based model built on top of LLMs such as GPT-3.5 and GPT-4. Despite having a lower training proportion compared to English, these models also exhibit remarkable capabilities in other languages. In this study, we assess the performance of GPT-3.5 and GPT-4 models on seven distinct Arabic NLP tasks: sentiment analysis, translation, transliteration, paraphrasing, part of speech tagging, summarization, and diacritization. Our findings reveal that GPT-4 outperforms GPT-3.5 on five out of the seven tasks. Furthermore, we conduct an extensive analysis of the sentiment analysis task, providing insights into how LLMs achieve exceptional results on a challenging dialectal dataset. Additionally, we introduce a new Python interface https://github.com/ARBML/Taqyim that facilitates the evaluation of these tasks effortlessly. | ['Ali Fadel', 'Ebrahim Alareqi', 'Hamzah Luqman', 'Badr AlKhamissi', 'Maged S. Alshaibani', 'Zaid Alyafeai'] | 2023-06-28 | null | null | null | null | ['sentiment-analysis', 'part-of-speech-tagging', 'transliteration'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [-1.63286045e-01 6.89790100e-02 -6.70606494e-02 -3.98473501e-01
-1.22303343e+00 -8.07384431e-01 6.42792046e-01 1.72437429e-01
-4.50359851e-01 6.99908078e-01 3.37801337e-01 -6.86797559e-01
4.78651732e-01 -3.65252107e-01 -4.89888936e-01 -2.05666721e-01
1.22771353e-01 6.57495499e-01 -2.57453680e-01 -6.55520737e-01
3.23791921e-01 -6.05817996e-02 -5.82885981e-01 7.28912055e-01
1.31373060e+00 4.75760549e-01 1.05949402e-01 6.18973672e-01
-2.41905913e-01 8.70180190e-01 -6.52496874e-01 -9.90210176e-01
-1.37530789e-01 -4.25319761e-01 -1.07798207e+00 -1.41989678e-01
2.70608515e-01 -8.36950466e-02 8.09173584e-02 6.02331936e-01
5.87004542e-01 -8.26765820e-02 6.03483856e-01 -1.07528639e+00
-1.06523740e+00 1.00447917e+00 -3.11232805e-01 -6.78220689e-02
6.17002964e-01 1.85698748e-01 1.25627351e+00 -1.08478737e+00
5.47154784e-01 1.31557083e+00 9.03224230e-01 4.56645906e-01
-9.90505457e-01 -4.08216625e-01 7.20387474e-02 -1.16825342e-01
-1.13474679e+00 -6.48469329e-01 2.74534702e-01 -2.80428290e-01
1.54279375e+00 2.07922056e-01 2.59298116e-01 1.17509294e+00
2.81406522e-01 1.20854449e+00 1.25362551e+00 -5.89467227e-01
-8.68584365e-02 2.41855025e-01 1.12102397e-01 7.44958699e-01
-1.59923568e-01 -9.23004091e-01 -7.80843735e-01 -1.20857425e-01
1.06438205e-01 -3.65670353e-01 -7.70403370e-02 6.03092790e-01
-1.27984619e+00 8.91676724e-01 9.91915539e-02 2.54691184e-01
-2.03972742e-01 -1.11975506e-01 5.69592297e-01 6.41245067e-01
8.30417871e-01 7.29738712e-01 -9.94405985e-01 -4.91237104e-01
-8.33903551e-01 -1.47362694e-01 1.30905139e+00 1.11394739e+00
4.34403867e-01 -5.91805205e-02 -1.51310712e-01 1.26570022e+00
2.86494881e-01 6.35286033e-01 7.65032530e-01 -9.04644608e-01
1.05353940e+00 4.68385965e-01 -1.10332832e-01 -5.17390072e-01
-4.00063396e-01 -1.38296843e-01 -4.87816393e-01 -5.19969225e-01
6.12826884e-01 -5.23681343e-01 -7.69223809e-01 1.45358932e+00
-1.69095919e-01 -5.83167434e-01 3.61015826e-01 4.18621510e-01
9.74206626e-01 1.05338466e+00 1.83901861e-02 1.07205316e-01
1.37630773e+00 -1.52340686e+00 -5.49095988e-01 -7.44189858e-01
1.14689839e+00 -1.40874195e+00 1.51294112e+00 3.75363111e-01
-1.29175794e+00 -2.15720937e-01 -6.58624887e-01 -3.10044765e-01
-3.65277857e-01 6.05710089e-01 6.50886893e-01 5.71435273e-01
-1.31987882e+00 4.86248851e-01 -8.81873488e-01 -8.13962460e-01
-1.92275867e-02 4.97964658e-02 -4.03618723e-01 -5.46418056e-02
-1.06550026e+00 1.33065927e+00 8.36476609e-02 3.21492791e-01
-4.72009182e-01 -3.77488226e-01 -9.23753142e-01 -1.62808627e-01
-2.10089032e-02 -3.17534238e-01 1.73273325e+00 -8.95783007e-01
-2.03015995e+00 9.67127621e-01 -5.03732085e-01 -3.51666063e-01
4.16928411e-01 -4.33474272e-01 -2.91635990e-01 -5.89278601e-02
3.20944935e-01 5.69086075e-01 4.13268209e-01 -6.84068084e-01
-4.13680434e-01 -3.41250226e-02 -4.13480438e-02 3.60818684e-01
-2.18741208e-01 5.95722914e-01 -4.06417698e-01 -5.93058407e-01
-8.26795176e-02 -1.13904798e+00 -9.40116644e-02 -7.07217097e-01
-6.28422618e-01 -3.12663078e-01 3.22554886e-01 -1.20142353e+00
1.24782288e+00 -1.98531508e+00 1.68548658e-01 -1.96523726e-01
-2.56207854e-01 2.70626873e-01 -5.32072365e-01 1.08398211e+00
2.67517716e-01 3.98053139e-01 -3.67801517e-01 -7.07620800e-01
2.07617253e-01 1.85666561e-01 -1.29306242e-01 9.43409875e-02
5.46593606e-01 1.20580244e+00 -7.43591964e-01 -3.20628703e-01
-1.50391385e-01 1.25914901e-01 -5.25401115e-01 -2.44926456e-02
-3.19105446e-01 4.62875426e-01 -1.97393268e-01 9.33065593e-01
3.40452701e-01 -3.22840035e-01 3.67068380e-01 4.50500250e-01
-2.80454606e-01 1.00804186e+00 -3.93759817e-01 1.72545826e+00
-9.08323586e-01 6.58741176e-01 9.25896466e-02 -5.24793923e-01
7.38061309e-01 4.45157796e-01 -4.12481911e-02 -6.66600943e-01
1.42239645e-01 6.17933571e-01 1.62526950e-01 -5.01224399e-01
7.77700961e-01 1.03227161e-01 -5.40986359e-01 8.64071906e-01
2.83053070e-01 -3.70179057e-01 4.09271747e-01 3.74916017e-01
8.28580678e-01 6.11345544e-02 2.76862353e-01 -1.94724381e-01
4.84650850e-01 1.11438222e-01 1.55320972e-01 6.65892720e-01
-1.29476875e-01 6.25554264e-01 7.36460030e-01 3.61100249e-02
-8.73490393e-01 -7.52140999e-01 9.41716060e-02 1.55453336e+00
-5.41600943e-01 -8.51028204e-01 -9.75403547e-01 -7.76431024e-01
-2.58297652e-01 9.69399154e-01 -2.99705446e-01 1.62975132e-01
-7.55334377e-01 -1.13377452e+00 1.02799070e+00 2.67907500e-01
5.32088101e-01 -1.14512312e+00 9.49329585e-02 1.88235745e-01
-8.27500045e-01 -1.17232871e+00 -5.43677270e-01 7.46060908e-02
-7.64551520e-01 -5.45369625e-01 -6.69517636e-01 -1.20968723e+00
3.81320447e-01 -5.94283603e-02 1.26721501e+00 -7.73908496e-02
3.82054836e-01 1.87511012e-01 -7.41829038e-01 -4.78461415e-01
-8.76091301e-01 7.68038273e-01 -2.04672366e-01 -3.55415672e-01
4.90583986e-01 -2.48777047e-01 -1.81563020e-01 1.45474166e-01
-4.02739108e-01 2.61853039e-02 6.91346645e-01 6.96095288e-01
-2.92322543e-02 -7.32107282e-01 7.29051173e-01 -1.01657045e+00
1.00068283e+00 -5.57675362e-01 -8.42456669e-02 4.79969114e-01
-4.07197624e-01 -6.49936944e-02 7.18590081e-01 -2.73139775e-01
-1.11323810e+00 -3.57388109e-01 -6.62940025e-01 6.42220616e-01
1.10485502e-01 9.95878577e-01 9.48551204e-03 1.94198534e-01
5.32636046e-01 3.51763278e-01 1.24380782e-01 -6.32265210e-01
5.33229113e-01 1.14046669e+00 1.77491799e-01 -5.66836655e-01
3.76522779e-01 3.99905927e-02 -7.32374907e-01 -9.72663641e-01
-1.02763689e+00 -2.38628373e-01 -6.80076003e-01 1.77320108e-01
7.43087828e-01 -1.18160021e+00 -4.79034573e-01 7.93566823e-01
-1.23658013e+00 -8.30178797e-01 2.80525923e-01 3.30472827e-01
-2.82434285e-01 5.52963555e-01 -1.36706877e+00 -4.72499371e-01
-7.77522266e-01 -1.14992750e+00 1.06108654e+00 -3.46690603e-03
-8.03192437e-01 -1.20369565e+00 1.10351600e-01 7.83750057e-01
4.76251185e-01 -2.00903416e-01 9.71544683e-01 -1.02760100e+00
-1.23638339e-01 -1.96690887e-01 -3.25377807e-02 5.51935554e-01
2.37265024e-02 2.67396122e-01 -7.44676590e-01 -2.57948637e-01
-2.09595382e-01 -5.74076831e-01 6.10021234e-01 1.56772003e-01
4.51988399e-01 -6.02054298e-01 1.93379655e-01 4.44803655e-01
8.42184126e-01 2.37256512e-02 4.23605293e-01 7.37936258e-01
5.28870225e-01 6.30781054e-01 5.70742905e-01 1.02896906e-01
9.57239687e-01 3.04495394e-01 4.91623580e-03 1.29153505e-01
-3.25288251e-02 -1.94966644e-01 1.03813088e+00 1.66809499e+00
1.20442770e-01 -4.14635807e-01 -1.11834264e+00 5.39421916e-01
-1.80509841e+00 -5.08668959e-01 -4.27918702e-01 1.72443736e+00
1.16596103e+00 -5.77451289e-02 -3.08352504e-02 -3.47321928e-01
3.05869818e-01 1.79737061e-01 -6.12246171e-02 -1.01382601e+00
-2.86825508e-01 2.19726250e-01 1.01099253e-01 6.90280914e-01
-9.27706838e-01 1.63041425e+00 6.29805279e+00 7.55895495e-01
-1.14565182e+00 3.68781120e-01 5.67439377e-01 1.49590805e-01
-8.03073496e-02 1.04568209e-02 -8.31009805e-01 5.22166669e-01
1.23759472e+00 -3.90609428e-02 4.95975822e-01 6.55435264e-01
2.69840479e-01 -1.67702809e-01 -1.04882431e+00 5.14331460e-01
2.45241880e-01 -1.02405286e+00 1.24651842e-01 -2.80543506e-01
8.53054702e-01 6.41507387e-01 1.31911382e-01 6.49654925e-01
4.65343207e-01 -1.05838728e+00 8.37485194e-01 2.86353789e-02
6.45033896e-01 -4.98032272e-01 9.05764699e-01 3.36575270e-01
-6.07863963e-01 1.41459092e-01 -2.52819955e-01 -2.98917919e-01
3.81706446e-01 2.84139663e-01 -1.12594604e+00 5.34009576e-01
5.49162865e-01 8.98833692e-01 -8.09858739e-01 6.39622748e-01
-8.25421154e-01 1.15387607e+00 -1.79922074e-01 -1.85790047e-01
6.16863251e-01 -4.85884756e-01 3.90415430e-01 1.83017373e+00
3.59313905e-01 -1.91640571e-01 -6.53163269e-02 3.64250153e-01
-2.43085384e-01 4.40123737e-01 -2.49675855e-01 -4.15052772e-01
3.87389630e-01 1.41612589e+00 -4.91909683e-01 -3.89453471e-01
-4.77094531e-01 1.21791124e+00 5.37099183e-01 3.25376809e-01
-6.96784675e-01 -6.45190001e-01 5.87584853e-01 -2.35215321e-01
8.81707221e-02 -5.29471874e-01 -3.34957570e-01 -1.46135795e+00
2.89669279e-02 -1.30576897e+00 3.75502288e-01 -9.77763236e-01
-1.42449272e+00 9.42861974e-01 -4.50002432e-01 -9.25719798e-01
-3.90142053e-01 -6.75104499e-01 -7.29050875e-01 1.08175063e+00
-1.51105464e+00 -1.40211654e+00 1.82471037e-01 3.81996930e-01
7.99043953e-01 -2.25071251e-01 9.31985855e-01 2.58083463e-01
-9.11696315e-01 7.56126881e-01 3.65653813e-01 5.70402086e-01
1.15025413e+00 -1.24149525e+00 9.08316851e-01 9.23029602e-01
4.27000038e-02 7.39685416e-01 3.90060484e-01 -5.33820391e-01
-1.13427174e+00 -1.11950803e+00 1.72792697e+00 -9.08087969e-01
1.19309807e+00 -6.07983589e-01 -8.39607537e-01 1.20970273e+00
7.71822810e-01 -8.81195068e-01 7.95017421e-01 3.58188152e-01
-2.75077313e-01 2.26692185e-01 -7.55922616e-01 8.12203884e-01
6.40974879e-01 -6.99124515e-01 -6.97308958e-01 6.82089806e-01
7.94216156e-01 -3.86796653e-01 -9.90749121e-01 -1.06550358e-01
5.42523503e-01 -6.51045263e-01 5.58752000e-01 -6.27459526e-01
8.35051477e-01 1.70604736e-01 -1.22626044e-01 -1.81227386e+00
-1.13069504e-01 -9.18633401e-01 2.39631340e-01 1.49270070e+00
1.09766829e+00 -1.06823874e+00 2.84642935e-01 3.88804883e-01
-3.98651779e-01 -4.95217234e-01 -6.55355453e-01 -7.56908238e-01
5.41745067e-01 -3.72119755e-01 2.79065579e-01 1.09704757e+00
5.48147082e-01 8.18555415e-01 -1.91438496e-01 -6.20048270e-02
1.14369027e-01 8.61301124e-02 6.70030951e-01 -6.95322394e-01
-2.85935760e-01 -5.52051306e-01 2.60819435e-01 -1.05022085e+00
3.86412144e-01 -1.23095417e+00 7.03322887e-02 -1.82666254e+00
-7.20927194e-02 -3.59663308e-01 3.41977447e-01 8.92953813e-01
-2.14331344e-01 3.95710945e-01 3.11881363e-01 3.58381212e-01
-5.97263515e-01 4.90744799e-01 1.01858068e+00 1.33174425e-02
-3.57107431e-01 1.22291252e-01 -9.18183625e-01 7.47095585e-01
1.27665579e+00 -4.58538234e-01 1.18103549e-01 -1.04010141e+00
5.29572070e-01 -5.31312525e-02 -2.49027714e-01 -4.45915788e-01
1.82670671e-02 3.25880982e-02 -2.76882201e-03 -2.86943674e-01
3.70579123e-01 -3.60617861e-02 -4.51160491e-01 2.90426344e-01
-2.67305136e-01 5.09806752e-01 3.24361622e-01 -1.63670048e-01
-3.80679071e-01 -3.46191853e-01 4.11108404e-01 -2.67107844e-01
-4.56435531e-01 -1.38158992e-01 -1.01230168e+00 3.15256923e-01
4.63545501e-01 1.25527695e-01 -6.97901070e-01 -6.16265416e-01
-7.01682448e-01 4.70683634e-01 2.97850728e-01 4.56840634e-01
1.92486867e-01 -8.13581347e-01 -1.11081374e+00 2.14351099e-02
1.54875398e-01 -2.11275771e-01 -1.94099888e-01 1.37891638e+00
-8.94863844e-01 6.75478280e-01 -6.14768341e-02 -3.55096877e-01
-1.20092165e+00 -7.05913454e-02 1.56196818e-01 -5.12485504e-01
-1.46793887e-01 9.01296616e-01 -1.67630121e-01 -1.20802748e+00
-1.83108881e-01 -4.40110087e-01 -7.91499615e-02 1.38673112e-01
2.19954848e-01 3.93998981e-01 3.12680364e-01 -7.30127573e-01
-3.71552110e-01 2.91927189e-01 -3.36051196e-01 -3.90535772e-01
1.35647964e+00 -4.16170537e-01 -7.34495342e-01 5.30149102e-01
9.83209312e-01 4.04978395e-01 -6.58777058e-01 -1.32772952e-01
3.75807852e-01 1.58553824e-01 -5.28163254e-01 -1.24123728e+00
-3.84916842e-01 8.83023441e-01 -4.68079478e-01 1.47397593e-01
8.45089197e-01 -3.31067517e-02 1.01548743e+00 6.44102931e-01
2.51987636e-01 -1.17834854e+00 5.09541854e-02 1.41413283e+00
9.17353868e-01 -1.27726233e+00 -3.33218724e-01 -2.21854776e-01
-1.18508434e+00 9.53242004e-01 5.71316481e-01 1.50406241e-01
3.63873035e-01 1.63958609e-01 5.00000119e-01 3.09574697e-02
-9.31205928e-01 -3.63532733e-03 9.43495035e-02 2.32992396e-01
1.01720202e+00 1.57791033e-01 -4.52837259e-01 6.63812339e-01
-8.35727811e-01 -3.45059782e-01 8.29255760e-01 9.98241961e-01
-1.77099243e-01 -1.16619182e+00 -9.56501067e-02 2.41113529e-01
-8.17084134e-01 -8.16232085e-01 -7.24268615e-01 8.34772289e-01
-5.36123097e-01 1.24074304e+00 -1.07541740e-01 -2.09173262e-01
2.73374528e-01 5.09759128e-01 3.04749101e-01 -8.81853819e-01
-1.06033397e+00 7.60817230e-02 5.96592367e-01 -3.02868336e-01
-1.38574839e-01 -7.50592649e-01 -1.05591726e+00 -5.55930972e-01
-1.15937844e-01 5.02046525e-01 6.79228008e-01 9.83250499e-01
4.05796200e-01 7.22615570e-02 2.94947058e-01 -6.89486921e-01
-5.21919608e-01 -1.50604963e+00 -3.48161608e-01 -1.15116015e-01
1.41680673e-01 3.00950289e-01 -3.15324485e-01 1.01281684e-02] | [11.151154518127441, 9.681744575500488] |
abd67cdf-0590-491f-ba8f-d35e81f2cde2 | speaker-diaphragm-excursion-prediction-deep | 2305.06640 | null | https://arxiv.org/abs/2305.06640v1 | https://arxiv.org/pdf/2305.06640v1.pdf | Speaker Diaphragm Excursion Prediction: deep attention and online adaptation | Speaker protection algorithm is to leverage the playback signal properties to prevent over excursion while maintaining maximum loudness, especially for the mobile phone with tiny loudspeakers. This paper proposes efficient DL solutions to accurately model and predict the nonlinear excursion, which is challenging for conventional solutions. Firstly, we build the experiment and pre-processing pipeline, where the feedback current and voltage are sampled as input, and laser is employed to measure the excursion as ground truth. Secondly, one FFTNet model is proposed to explore the dominant low-frequency and other unknown harmonics, and compares to a baseline ConvNet model. In addition, BN re-estimation is designed to explore the online adaptation; and INT8 quantization based on AI Model efficiency toolkit (AIMET\footnote{AIMET is a product of Qualcomm Innovation Center, Inc.}) is applied to further reduce the complexity. The proposed algorithm is verified in two speakers and 3 typical deployment scenarios, and $>$99\% residual DC is less than 0.1 mm, much better than traditional solutions. | ['Hao Xu', 'Chirag Patel', 'Eddie Choy', 'Yin Huang', 'Matt Zivney', 'Yuwei Ren'] | 2023-05-11 | null | null | null | null | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [-3.38144638e-02 -1.58792824e-01 7.71261603e-02 -2.30117232e-01
-6.20379090e-01 -4.50389326e-01 -1.10792555e-01 -2.15282366e-01
-2.55660787e-02 6.81367040e-01 3.76166463e-01 -3.35746288e-01
2.83246208e-02 -1.92577496e-01 -3.67402971e-01 -7.04385579e-01
-2.36172795e-01 -4.91162717e-01 -1.72539473e-01 -2.38409474e-01
1.27973884e-01 2.26008758e-01 -1.52341115e+00 2.28454098e-02
1.21182334e+00 1.47178924e+00 -1.27301132e-02 7.40101933e-01
1.85595468e-01 7.24506855e-01 -1.13973010e+00 -2.63822734e-01
1.55098245e-01 -3.90610754e-01 -2.21795395e-01 -4.39900279e-01
4.95288283e-01 -5.83822668e-01 -4.09923077e-01 1.15879822e+00
1.35171592e+00 2.37001646e-02 3.46988231e-01 -1.24814427e+00
-3.67715135e-02 8.33616912e-01 -3.93935114e-01 5.64909935e-01
1.87940374e-01 4.28841412e-01 6.85197771e-01 -6.52923286e-01
1.28428251e-01 9.86653626e-01 1.03628433e+00 3.45057696e-01
-1.02643955e+00 -1.38918245e+00 6.55256957e-02 4.89761293e-01
-1.59272814e+00 -8.38358700e-01 1.11104131e+00 5.35496399e-02
1.18475044e+00 4.78094578e-01 7.92354405e-01 8.90055120e-01
2.09536135e-01 5.82312405e-01 8.34053338e-01 -3.07615817e-01
4.41625476e-01 3.64643782e-01 1.90314148e-02 1.77115902e-01
-3.47790837e-01 -8.89153555e-02 -1.09196806e+00 -2.04036370e-01
5.51978588e-01 -3.39644819e-01 -8.21971953e-01 4.01447684e-01
-4.83142823e-01 3.89203876e-01 4.26476359e-01 2.43164599e-02
-2.26032630e-01 2.67572999e-01 2.85143942e-01 3.40335310e-01
2.94082105e-01 3.20780516e-01 -2.23126158e-01 -5.12126684e-01
-1.33302462e+00 6.46010861e-02 7.37791717e-01 9.33008432e-01
4.05545563e-01 9.33021903e-01 -6.94019347e-02 9.09708917e-01
5.36015689e-01 6.83947802e-01 5.79273880e-01 -1.05112147e+00
4.58507091e-01 9.16286111e-02 1.51407514e-02 -8.80616009e-01
-4.61386025e-01 -9.11027551e-01 -9.97044563e-01 -2.33671799e-01
6.26755357e-02 -4.17440057e-01 -7.10475743e-01 1.66659760e+00
1.92787081e-01 6.20081961e-01 -8.60872343e-02 8.97870839e-01
7.87993670e-01 8.43315184e-01 -3.77210468e-01 -6.68165863e-01
9.76709843e-01 -8.06046009e-01 -1.24223781e+00 -2.29449451e-01
5.49211502e-02 -7.10036695e-01 1.17512524e+00 9.05834496e-01
-1.19466543e+00 -6.96026027e-01 -1.42415845e+00 -3.24730277e-02
2.29797229e-01 7.18705654e-02 1.93740144e-01 1.28590858e+00
-1.13301098e+00 4.49791074e-01 -6.92383230e-01 3.31576109e-01
2.96919793e-01 3.98529291e-01 6.12568557e-01 3.92891347e-01
-1.64011908e+00 4.83486503e-01 -2.84231246e-01 5.15650570e-01
-8.23560715e-01 -1.20634329e+00 -5.89854836e-01 1.86957672e-01
-1.62456512e-01 -1.87352657e-01 1.41365600e+00 -8.28076124e-01
-2.11758637e+00 -1.60037503e-01 -4.15716290e-01 -7.52471864e-01
3.84034187e-01 -2.94606477e-01 -1.04923069e+00 7.52733424e-02
-4.04636204e-01 1.24842510e-01 1.12214601e+00 -9.47110772e-01
-5.49447715e-01 -1.15558431e-01 -6.46153092e-02 2.46635497e-01
-8.55869234e-01 -2.47006550e-01 -3.52741241e-01 -7.19266534e-01
1.72825322e-01 -4.74029899e-01 -1.14309572e-01 -3.74164760e-01
-5.11850834e-01 2.63636768e-01 4.70106423e-01 -9.28225994e-01
1.89318788e+00 -2.30195689e+00 -4.84558582e-01 6.81293190e-01
5.86402453e-02 3.50802243e-01 4.27220255e-01 2.46890500e-01
1.11981109e-01 1.98833011e-02 1.29310144e-02 -2.31063247e-01
9.62506160e-02 -4.35648382e-01 -6.52499259e-01 5.64255893e-01
-6.41684383e-02 3.80242139e-01 -1.76317632e-01 -4.57620695e-02
4.21177559e-02 6.66577518e-01 -5.95514715e-01 1.14136003e-02
3.78736943e-01 2.81097054e-01 -1.12108039e-02 4.11390126e-01
1.04808044e+00 1.03533231e-01 2.48578824e-02 -5.68713546e-01
-2.44821772e-01 9.72603500e-01 -1.79269636e+00 1.55843043e+00
-7.55085826e-01 9.90456998e-01 6.30307555e-01 -4.71182168e-01
9.69794869e-01 4.76720691e-01 8.31110179e-02 -8.89965236e-01
4.07106906e-01 1.50595680e-01 2.18102992e-01 -3.50731730e-01
2.09504575e-01 1.23709880e-01 3.47437382e-01 2.40879685e-01
7.17355385e-02 -4.49794531e-01 -3.36267591e-01 1.43392727e-01
7.48782516e-01 -1.75542444e-01 -3.98796163e-02 -5.69477439e-01
6.57822609e-01 -4.95265186e-01 9.59716260e-01 4.67015386e-01
-3.36323649e-01 4.18579936e-01 1.74828872e-01 -5.19583523e-02
-4.92707759e-01 -8.95319521e-01 -4.83750850e-01 9.50935721e-01
3.92451227e-01 -5.69659173e-01 -1.11152112e+00 2.03886554e-01
-2.53662467e-01 9.98810649e-01 -3.03485971e-02 -2.04313338e-01
-5.42172551e-01 -5.62941551e-01 9.56236839e-01 5.77500880e-01
7.40446746e-01 -5.88114738e-01 -5.66262782e-01 3.40393215e-01
-1.64778292e-01 -7.34642923e-01 -6.85701132e-01 3.90943885e-01
-5.06879389e-01 -2.42867455e-01 -4.73403186e-01 -6.70307040e-01
1.79252207e-01 -2.40619451e-01 6.43420041e-01 -6.24370836e-02
-3.00200075e-01 1.20705143e-01 1.67551443e-01 -8.12367320e-01
7.27601000e-04 -1.26422375e-01 5.57341456e-01 2.05769558e-02
2.54990965e-01 -8.99611175e-01 -1.02957618e+00 2.71106541e-01
-3.78510356e-01 -3.65399688e-01 1.64094806e-01 4.91064578e-01
4.00354326e-01 4.44763839e-01 9.04347539e-01 -2.40907535e-01
8.80954266e-01 -2.39193588e-02 -5.15148520e-01 -1.92951024e-01
-6.19708419e-01 -4.35831904e-01 1.06269896e+00 -7.55854607e-01
-1.29748380e+00 -2.62713552e-01 -7.16900945e-01 -1.93061829e-01
2.60842770e-01 1.62222326e-01 -6.32346809e-01 -5.31337298e-02
6.89835548e-01 2.24895060e-01 -1.78927109e-01 -3.11318070e-01
7.82627016e-02 1.34764278e+00 6.69013023e-01 -5.93326539e-02
5.72259963e-01 3.89248550e-01 -5.27369797e-01 -1.22423160e+00
-2.36547872e-01 -2.58740723e-01 -3.49198729e-02 -1.72108084e-01
3.33850175e-01 -1.32952642e+00 -1.02900541e+00 6.46954715e-01
-9.61483896e-01 -3.45279038e-01 -5.97267449e-02 8.34448278e-01
-2.70046890e-01 -1.17443256e-01 -9.36719239e-01 -1.13791227e+00
-1.05723333e+00 -1.13308299e+00 4.50868875e-01 6.61557078e-01
-2.39915237e-01 -6.74385965e-01 -3.28509957e-01 1.46906808e-01
7.77002454e-01 -2.97527879e-01 4.56094593e-01 -2.24445537e-01
-2.00365156e-01 7.90950283e-02 4.11672175e-01 5.62538326e-01
-1.12373419e-01 -1.34327337e-01 -1.69193649e+00 -2.79848993e-01
8.90174925e-01 2.54740827e-02 4.80233997e-01 7.56541669e-01
1.23307812e+00 -3.64868343e-01 -2.93049067e-01 9.41546917e-01
1.01370549e+00 4.92121279e-01 9.07583952e-01 -8.64902213e-02
5.22413015e-01 2.80646622e-01 2.21446902e-01 4.50840652e-01
3.78751516e-01 3.15416753e-01 1.90777719e-01 -5.31671122e-02
-5.80643117e-02 -8.97000507e-02 4.88378406e-01 1.31284916e+00
3.67786586e-01 -3.07098627e-01 -5.50528407e-01 3.08664382e-01
-1.08932447e+00 -7.73168445e-01 -2.91225433e-01 2.14879704e+00
9.84651923e-01 4.81354028e-01 -1.28750816e-01 5.83735287e-01
5.70689321e-01 3.44534144e-02 -1.05884802e+00 -6.57500505e-01
-8.97907987e-02 4.92440611e-01 6.30250394e-01 6.63798928e-01
-7.05596209e-01 3.02377760e-01 6.07246161e+00 9.85805154e-01
-1.86062932e+00 1.38477311e-01 5.61617136e-01 -4.29112077e-01
-4.44018573e-01 -3.77544552e-01 -9.98338640e-01 6.33845925e-01
1.42906451e+00 -3.50876480e-01 5.04303217e-01 6.84776306e-01
4.96193320e-01 1.91812813e-01 -8.68280590e-01 1.25669992e+00
2.89687425e-01 -1.03284061e+00 -5.96750617e-01 -2.52030611e-01
4.10635054e-01 -2.46878356e-01 3.84140283e-01 3.55838835e-01
-2.66076833e-01 -1.00876081e+00 9.92134333e-01 5.03142416e-01
1.01715302e+00 -1.06191516e+00 4.64029074e-01 5.01880169e-01
-1.30814135e+00 -3.37989748e-01 -1.51183799e-01 -2.09644213e-01
2.99812376e-01 7.05728769e-01 -8.94840837e-01 2.88490206e-01
1.13666224e+00 3.05686384e-01 -2.35185802e-01 1.13732803e+00
-4.19688165e-01 1.39656818e+00 -7.06947803e-01 -2.11054668e-01
-3.34169209e-01 2.12037921e-01 5.63358068e-01 1.11664104e+00
4.12852556e-01 1.61812708e-01 -3.50450993e-01 7.25255549e-01
-1.42390907e-01 -1.42154232e-01 2.12527692e-01 6.52415633e-01
9.42329109e-01 1.27861071e+00 -1.90741107e-01 -1.77771196e-01
-2.08285511e-01 7.42199421e-01 -4.14382905e-01 6.46066785e-01
-1.16490519e+00 -1.01502764e+00 8.02511394e-01 2.66127855e-01
9.74044502e-02 -6.30304366e-02 -4.66139495e-01 -7.21986949e-01
2.35899299e-01 -1.02531719e+00 -1.95839912e-01 -7.23325312e-01
-8.34062934e-01 7.91058660e-01 -4.96142477e-01 -1.47162652e+00
-4.12954062e-01 -1.96996897e-01 -9.73440468e-01 1.19722331e+00
-1.39360762e+00 -5.02013326e-01 -2.23157153e-01 5.61575472e-01
6.64091349e-01 7.79943392e-02 5.49200058e-01 6.61478043e-01
-5.37688851e-01 1.31054616e+00 2.69385159e-01 -2.73064226e-01
6.73980474e-01 -1.00409448e+00 4.62846309e-01 9.33666945e-01
-2.34280735e-01 9.90173221e-01 9.33731973e-01 -3.34032804e-01
-1.61826754e+00 -9.32675123e-01 4.54324901e-01 7.44320601e-02
4.58149701e-01 -9.78953421e-01 -1.00195777e+00 1.26121402e-01
5.72377145e-01 -1.77557454e-01 7.64781296e-01 -3.33374441e-01
7.23374933e-02 -8.03209066e-01 -1.20288157e+00 6.55152500e-01
8.75650167e-01 -5.67856193e-01 -8.21553171e-02 -9.11310688e-02
8.14360321e-01 -7.45404303e-01 -8.40711415e-01 1.44313484e-01
6.07081532e-01 -8.11312735e-01 8.16885173e-01 1.64503008e-01
-1.35014191e-01 -6.36162877e-01 -1.44318521e-01 -1.45364654e+00
-5.92465550e-02 -1.43380630e+00 -3.93360198e-01 1.62945735e+00
5.67287087e-01 -7.20815420e-01 5.61060786e-01 5.24542332e-01
-4.34746981e-01 -6.74798608e-01 -1.20076883e+00 -7.01069474e-01
-1.38783887e-01 -7.62762606e-01 7.45802164e-01 4.79949385e-01
3.08666348e-01 4.12043750e-01 -4.46510166e-01 5.26310921e-01
5.21767139e-01 -3.89362872e-01 4.61516351e-01 -6.71507835e-01
-3.22399795e-01 -2.81172454e-01 -1.86524704e-01 -1.56004727e+00
-1.88879758e-01 -3.59348953e-01 1.72861189e-01 -1.20338607e+00
-6.77092791e-01 -2.82089621e-01 -3.26350182e-01 -2.72354316e-02
-1.20724641e-01 3.28660041e-01 -7.43537098e-02 -1.88896716e-01
-4.17905338e-02 5.95851719e-01 6.56418383e-01 -4.31300968e-01
-6.41892850e-01 2.96108693e-01 -7.35997498e-01 8.51038158e-01
6.90672517e-01 -1.17190748e-01 -7.22050071e-01 -4.42868859e-01
3.15714814e-02 5.80350757e-02 1.32271275e-01 -1.41646457e+00
7.34380186e-01 3.48348469e-01 4.27309573e-01 -6.68510914e-01
5.52134991e-01 -7.23148644e-01 9.69220325e-02 3.69447678e-01
-3.68945628e-01 -2.04214677e-01 4.57938462e-01 2.07889900e-01
-2.34420910e-01 -1.77986119e-02 7.24203765e-01 5.15278637e-01
-1.99165255e-01 4.75985482e-02 -5.41867316e-01 8.75579845e-03
4.06323254e-01 -2.93841898e-01 -2.56802976e-01 -7.71079063e-01
-4.55762416e-01 4.09139305e-01 -1.56994849e-01 1.18785076e-01
5.04804611e-01 -1.24958169e+00 -3.38982075e-01 4.70138073e-01
-5.14739215e-01 -6.14700541e-02 8.18265557e-01 9.33809042e-01
-3.84252131e-01 2.10995138e-01 2.69297898e-01 -6.15574479e-01
-1.07518756e+00 -2.41992082e-02 6.88390553e-01 3.19976240e-01
-3.95744801e-01 1.25614464e+00 -1.32966787e-01 1.32234290e-01
7.95281649e-01 -5.33820629e-01 -2.01520026e-01 -9.90354456e-03
9.53413785e-01 6.51566207e-01 4.41118360e-01 -3.57901871e-01
-4.17895317e-01 4.90288556e-01 3.24325971e-02 -2.78664887e-01
1.02208054e+00 -6.07506573e-01 2.02576369e-01 4.78469223e-01
1.21623957e+00 4.82120514e-01 -1.41216958e+00 7.28107840e-02
-4.95091975e-01 -2.33293727e-01 3.82000744e-01 -7.89240360e-01
-9.90439296e-01 1.26106536e+00 1.02674949e+00 1.34939238e-01
1.59774697e+00 -7.34817028e-01 1.08391070e+00 4.89437133e-02
1.94563091e-01 -1.38189781e+00 -1.16508275e-01 2.28455424e-01
8.43946993e-01 -4.40873921e-01 -3.15342098e-02 -2.08793163e-01
-7.29327857e-01 9.16064739e-01 6.38749003e-01 1.00936659e-01
8.49747419e-01 7.16385365e-01 4.75832313e-01 4.14945066e-01
-5.20504117e-01 3.61197948e-01 1.62515581e-01 6.83207929e-01
4.58626360e-01 -2.85883009e-01 -2.10841253e-01 1.18669724e+00
-8.25439572e-01 -3.66581455e-02 5.19925237e-01 5.56608558e-01
-3.73811156e-01 -4.31419790e-01 -4.89343226e-01 6.06620014e-01
-8.25876057e-01 -2.96809465e-01 7.47551546e-02 7.75279105e-02
3.47132623e-01 1.40630543e+00 3.14305276e-01 -6.81121707e-01
5.84717512e-01 -2.01577276e-01 -1.35134906e-02 1.30690346e-02
-9.23167288e-01 6.14926457e-01 -1.25683919e-01 -6.40263557e-01
2.56842822e-01 -4.59805965e-01 -1.59219551e+00 -1.99922651e-01
-6.99806571e-01 1.52539566e-01 8.25755715e-01 5.35648465e-01
3.80588561e-01 9.38731551e-01 9.14645195e-01 -6.95260465e-01
-4.99430001e-01 -1.08954263e+00 -7.57475257e-01 -2.15129331e-01
6.98255897e-01 -3.09364945e-01 -7.09181130e-01 3.67156826e-02] | [14.81933307647705, 5.887237071990967] |
f455f514-f579-4160-9a91-8e8019b1f0fa | adaptive-rate-sparse-signal-reconstruction | 1503.03231 | null | http://arxiv.org/abs/1503.03231v1 | http://arxiv.org/pdf/1503.03231v1.pdf | Adaptive-Rate Sparse Signal Reconstruction With Application in Compressive Background Subtraction | We propose and analyze an online algorithm for reconstructing a sequence of
signals from a limited number of linear measurements. The signals are assumed
sparse, with unknown support, and evolve over time according to a generic
nonlinear dynamical model. Our algorithm, based on recent theoretical results
for $\ell_1$-$\ell_1$ minimization, is recursive and computes the number of
measurements to be taken at each time on-the-fly. As an example, we apply the
algorithm to compressive video background subtraction, a problem that can be
stated as follows: given a set of measurements of a sequence of images with a
static background, simultaneously reconstruct each image while separating its
foreground from the background. The performance of our method is illustrated on
sequences of real images: we observe that it allows a dramatic reduction in the
number of measurements with respect to state-of-the-art compressive background
subtraction schemes. | ['Aswin C. Sankaranarayanan', 'Nikos Deligiannis', 'Volkan Cevher', 'Miguel R. D. Rodrigues', 'Joao F. C. Mota'] | 2015-03-11 | null | null | null | null | ['video-background-subtraction'] | ['computer-vision'] | [ 1.03275108e+00 -3.64158452e-01 4.52089161e-01 -1.04758538e-01
-5.22932589e-01 -4.96820033e-01 5.25637448e-01 -4.06402498e-01
-3.46439779e-01 5.39675117e-01 -3.54094028e-01 -3.13824683e-01
-5.68821467e-02 -4.43315148e-01 -6.48605227e-01 -1.05117214e+00
-3.22707117e-01 2.38648325e-01 1.03628740e-01 -7.92534947e-02
1.37553364e-01 2.84191817e-01 -1.35392952e+00 -2.80756742e-01
5.37012875e-01 1.16391599e+00 4.72158581e-01 1.20250154e+00
2.20528319e-01 1.18559682e+00 -6.12289071e-01 -1.61588401e-01
7.02420294e-01 -9.21703875e-01 -2.39781544e-01 8.46733749e-01
5.94936907e-01 -4.10781682e-01 -8.71244967e-01 1.54962528e+00
3.70397747e-01 1.17283665e-01 1.32469997e-01 -8.55207741e-01
-7.33726099e-02 2.82917231e-01 -6.86406732e-01 7.13181198e-01
2.07572773e-01 -1.38522405e-02 3.39136988e-01 -6.83284163e-01
6.69987023e-01 9.06292498e-01 5.60005665e-01 4.22451764e-01
-1.39999187e+00 -3.45908403e-01 2.44466111e-01 3.84211093e-01
-1.36279440e+00 -9.93879974e-01 8.71361673e-01 -3.45670193e-01
3.30954283e-01 5.78087151e-01 7.24379063e-01 5.79459667e-01
-3.56534831e-02 6.75333500e-01 9.73676622e-01 -6.35395586e-01
5.09758234e-01 -3.99757177e-01 -1.29289376e-02 7.01644480e-01
6.74411118e-01 6.94443583e-02 -6.18770540e-01 -2.86601782e-01
7.68389940e-01 4.32953686e-02 -8.05048466e-01 -4.51773971e-01
-1.17527318e+00 4.22785133e-01 -4.19420034e-01 2.80876040e-01
-5.27861476e-01 2.66641527e-01 -1.32842407e-01 5.29352248e-01
4.22459900e-01 -1.94833785e-01 -5.14947362e-02 3.35847959e-02
-1.47941768e+00 1.98217943e-01 1.04838574e+00 9.16497946e-01
6.17634714e-01 5.11154056e-01 3.80534142e-01 3.84281486e-01
-5.06838076e-02 1.18773043e+00 1.24806114e-01 -1.44946015e+00
1.78703159e-01 -2.40044385e-01 4.75060344e-01 -1.09569085e+00
1.03730172e-01 -3.48127633e-01 -1.14321053e+00 2.09140688e-01
2.90381879e-01 -3.57349545e-01 -6.43589079e-01 1.83140302e+00
3.72601658e-01 1.20028210e+00 3.90864350e-02 7.46015370e-01
1.30571738e-01 6.31524801e-01 -9.26760316e-01 -1.08525026e+00
8.52750838e-01 -6.05712950e-01 -9.30235147e-01 -5.84457278e-01
-1.45228878e-01 -7.97361434e-01 6.59171939e-02 7.50132501e-01
-1.36123240e+00 -4.84290689e-01 -1.17444551e+00 5.75829744e-01
3.48168969e-01 1.28910065e-01 8.46576169e-02 8.73589516e-01
-1.24262726e+00 6.23712718e-01 -1.19028127e+00 -2.18331918e-01
-5.09705134e-02 2.87216604e-01 -1.07255042e-01 -4.23262805e-01
-5.13082445e-01 6.92440748e-01 6.22792952e-02 2.74126977e-01
-1.21879852e+00 -3.86165768e-01 -7.03503668e-01 -6.85196668e-02
7.27258325e-01 -6.34560764e-01 1.12925112e+00 -1.22719860e+00
-1.48193562e+00 7.56481171e-01 -5.87161303e-01 -6.53477967e-01
5.79877198e-01 -2.91280299e-01 -5.61239362e-01 5.37425935e-01
-3.21957916e-02 -1.94695294e-01 1.51919222e+00 -1.37362504e+00
-5.34757853e-01 -3.55480134e-01 -2.10292071e-01 7.93897267e-03
3.70257981e-02 -6.92204833e-02 -6.21768296e-01 -5.74904919e-01
5.84432125e-01 -1.19521606e+00 -6.14811599e-01 -1.34462953e-01
-1.89879164e-01 1.08300173e+00 1.08569920e+00 -1.00743759e+00
1.03686821e+00 -2.18282151e+00 5.59894443e-01 6.36300966e-02
2.62923390e-01 2.68108606e-01 -1.37263104e-01 2.26572871e-01
-5.27608916e-02 -4.94023442e-01 -7.63604462e-01 -4.88836586e-01
-3.81857276e-01 1.97844952e-01 -3.14799130e-01 1.18111122e+00
-2.83084691e-01 3.39932889e-01 -8.48666430e-01 -1.60431623e-01
6.09252453e-01 3.58900666e-01 -2.17410296e-01 2.09440112e-01
-2.29585972e-02 6.87893927e-01 -2.09888875e-01 3.72224450e-01
1.15023482e+00 -2.67503738e-01 7.08373547e-01 1.11389779e-01
-1.22405827e-01 -3.99577200e-01 -1.82567918e+00 1.44690871e+00
9.88820419e-02 1.03572226e+00 1.12736642e+00 -1.46802926e+00
3.27650338e-01 3.62572193e-01 8.47996652e-01 -3.96683723e-01
3.51373643e-01 1.56389058e-01 -9.44503322e-02 -4.00510848e-01
4.08412963e-01 -2.53702074e-01 1.62836090e-01 3.99148554e-01
-1.37286186e-02 -1.36713475e-01 4.13549513e-01 4.81793225e-01
1.52016020e+00 -1.58985198e-01 4.38425809e-01 -3.11451524e-01
8.27198923e-01 -2.37162516e-01 5.79225063e-01 1.06395864e+00
-1.61473781e-01 4.67594475e-01 -8.53380561e-02 -5.22925965e-02
-7.62323260e-01 -9.34441745e-01 2.95979202e-01 4.97339904e-01
4.32917595e-01 -1.60268426e-01 -9.53042448e-01 1.31006435e-01
-2.42926791e-01 5.45873702e-01 -4.15531635e-01 2.06706256e-01
-8.15421939e-01 -9.88264918e-01 2.83227935e-02 -1.32761344e-01
5.42090535e-01 -5.94225526e-01 -1.21930218e+00 3.53858143e-01
-5.72183967e-01 -1.66391778e+00 -3.17963034e-01 9.88364220e-02
-9.62483287e-01 -1.25303316e+00 -6.95301175e-01 -4.12336230e-01
8.77989531e-01 9.89971161e-01 1.15680182e+00 1.24207027e-01
-3.52907807e-01 1.14740908e+00 -1.05801776e-01 -3.79343033e-01
-3.83619696e-01 -8.03535163e-01 2.39463180e-01 6.41198814e-01
-1.20160095e-01 -6.95241570e-01 -5.14276147e-01 1.08053513e-01
-1.12613642e+00 -1.98748171e-01 1.42702386e-01 7.38928616e-01
7.06715941e-01 5.32165587e-01 -2.50744224e-01 -7.22753644e-01
1.79626971e-01 -2.53505766e-01 -1.16134727e+00 1.57710329e-01
-1.33857965e-01 -3.89987767e-01 5.33013463e-01 -5.51593542e-01
-1.07478118e+00 2.61293977e-01 5.63221872e-01 -8.09415221e-01
2.04441741e-01 2.63090104e-01 -6.76664412e-02 -6.66832149e-01
2.24940553e-01 8.34625900e-01 1.23774208e-01 -2.02560276e-01
4.92188096e-01 2.19176948e-01 1.13274670e+00 -4.64002997e-01
1.08342338e+00 1.17303371e+00 3.54877025e-01 -1.52902675e+00
-5.08927464e-01 -6.67486012e-01 -6.60177350e-01 -5.10104477e-01
3.84807765e-01 -1.05062485e+00 -6.75731778e-01 7.73709893e-01
-8.07244003e-01 -3.83058161e-01 -3.54223758e-01 5.25341392e-01
-7.42260695e-01 8.62344980e-01 -5.13536632e-01 -1.23820686e+00
2.33886857e-02 -8.46859932e-01 7.86064029e-01 1.28114358e-01
3.83520007e-01 -9.57889140e-01 1.92152768e-01 1.28735676e-01
4.50218946e-01 2.95789570e-01 1.90078720e-01 -7.57428408e-02
-1.03695869e+00 -2.55098522e-01 1.87889814e-01 6.07584476e-01
3.12847681e-02 -2.30217606e-01 -7.87776649e-01 -7.67959118e-01
1.03153431e+00 3.07645977e-01 8.63069236e-01 8.49265099e-01
7.27031410e-01 -2.79465407e-01 -3.16133082e-01 6.07929409e-01
1.75321531e+00 3.70282769e-01 5.78559101e-01 3.11558973e-02
3.68591785e-01 1.75590217e-01 4.25639749e-01 7.83452690e-01
-1.39375195e-01 3.92209291e-01 5.72248101e-01 1.40910044e-01
3.58688310e-02 4.84612912e-01 5.34705400e-01 1.15003920e+00
-5.80892190e-02 -4.21528250e-01 -4.22238082e-01 6.35693610e-01
-1.83417308e+00 -1.28232670e+00 -1.19323455e-01 2.54278755e+00
2.09600627e-01 -1.82381272e-01 -2.72033572e-01 2.34587461e-01
8.80511284e-01 5.98393083e-01 -7.30898678e-01 5.51122785e-01
-4.66386110e-01 4.07252014e-01 8.71244669e-01 6.50022805e-01
-1.08636916e+00 5.24419069e-01 6.84389019e+00 2.71496058e-01
-1.27490842e+00 1.57162964e-01 2.75515109e-01 -5.24118066e-01
1.73264697e-01 1.16509773e-01 -2.27057695e-01 4.83766407e-01
1.12353098e+00 -5.03754437e-01 9.07585382e-01 3.41734588e-01
1.53295964e-01 -3.96239221e-01 -9.02948797e-01 1.41865492e+00
5.83287597e-01 -1.39497590e+00 -2.73974746e-01 1.24557234e-01
8.48099351e-01 1.46918267e-01 -1.05258353e-01 -2.52614737e-01
3.60657066e-01 -4.52323437e-01 9.29529965e-01 4.48201805e-01
4.30507213e-01 -3.29206824e-01 2.75214761e-01 6.58760011e-01
-1.14134431e+00 -1.69242248e-01 -3.54989648e-01 -2.22167045e-01
7.00800240e-01 9.09644008e-01 -1.58197209e-01 7.76064694e-01
5.13093531e-01 6.38908088e-01 -1.21819511e-01 9.76206720e-01
1.82191297e-01 7.51645267e-01 -5.42343080e-01 4.67979312e-01
-1.29478902e-01 -8.98693204e-01 9.99860406e-01 1.09317863e+00
8.24572384e-01 8.67281258e-01 3.84076208e-01 2.38727674e-01
-1.75756291e-02 -3.12530756e-01 -5.32079995e-01 4.53504473e-02
3.23711962e-01 9.37470496e-01 -8.00727785e-01 -8.06083322e-01
-5.18781483e-01 1.26515150e+00 -3.77171397e-01 4.62276071e-01
-6.84289455e-01 1.16587162e-01 5.20291805e-01 1.03989756e-03
7.64213860e-01 -5.31949699e-01 1.93357527e-01 -1.69237053e+00
1.49624720e-01 -1.04595959e+00 2.13115081e-01 -6.56740367e-01
-8.51717293e-01 2.47325346e-01 -1.42058313e-01 -1.24829388e+00
-7.94499144e-02 -3.18937957e-01 -4.23072010e-01 5.33555806e-01
-1.24496150e+00 -3.98232639e-01 -5.13750613e-01 7.70546019e-01
6.01203799e-01 -5.37701659e-02 5.25861442e-01 3.26110989e-01
-4.69015628e-01 -3.04222137e-01 6.16798043e-01 -2.30325423e-02
2.14234501e-01 -8.54625046e-01 3.54912311e-01 1.76068962e+00
3.98688138e-01 3.38688999e-01 1.13843620e+00 -5.52996278e-01
-1.86991119e+00 -7.78924704e-01 2.70827323e-01 -1.30779028e-01
6.92257702e-01 -1.67414993e-01 -6.43680990e-01 1.05845594e+00
3.79185617e-01 2.72644579e-01 5.89444041e-01 -6.79934144e-01
8.79442915e-02 3.34325954e-02 -1.14991283e+00 4.34935272e-01
1.14144325e+00 -3.61764610e-01 -3.73538107e-01 3.35721463e-01
9.79116261e-02 -5.60087800e-01 -3.43229145e-01 1.47977993e-01
3.77902865e-01 -1.22072971e+00 1.13530612e+00 -6.80972189e-02
-1.84189826e-01 -5.49865782e-01 -7.99818993e-01 -1.02603388e+00
-3.56046438e-01 -1.25171721e+00 -4.74077344e-01 4.89824504e-01
-3.75498950e-01 -6.54109240e-01 8.54179442e-01 1.61692888e-01
3.71434540e-01 8.27128142e-02 -1.18920541e+00 -7.86183417e-01
-5.83221674e-01 -5.02404034e-01 -3.60583072e-03 8.78148377e-01
-4.95608956e-01 6.90517053e-02 -8.80910039e-01 7.08823681e-01
1.48584342e+00 2.31804952e-01 9.77172017e-01 -1.26743686e+00
-7.17927516e-01 -8.31308737e-02 -4.99658376e-01 -1.38991320e+00
2.81106308e-02 -2.13758051e-01 1.73188522e-01 -9.84019995e-01
3.27626944e-01 4.31676097e-02 -2.47480735e-01 -3.74415368e-01
-8.81317183e-02 1.46249652e-01 7.01460660e-01 1.56127125e-01
-6.60273850e-01 2.06108645e-01 6.67926192e-01 -3.79904211e-01
2.17108876e-01 1.98744666e-02 -4.27986860e-01 8.11551452e-01
1.92956910e-01 -4.82764065e-01 -4.37329203e-01 -4.82468247e-01
-2.96877563e-01 8.63709450e-01 2.30168849e-01 -1.18873870e+00
3.24032187e-01 -2.19487831e-01 2.36224368e-01 -5.47000051e-01
7.49039650e-01 -9.11496639e-01 6.63713634e-01 7.28711903e-01
1.39164165e-01 -1.09057158e-01 1.09057419e-01 9.45775568e-01
-1.53590634e-01 -1.89879268e-01 8.63143146e-01 -3.60734671e-01
-6.51550829e-01 1.97578937e-01 -5.93371987e-01 -6.54723421e-02
9.57756341e-01 -1.28796816e-01 -9.05018151e-02 -8.28096688e-01
-1.02119708e+00 -1.68708026e-01 5.91279149e-01 -1.78523406e-01
4.45748478e-01 -8.39694381e-01 -8.20816815e-01 3.43362123e-01
-6.36136293e-01 -2.42478177e-01 5.16663492e-01 9.49586630e-01
-6.23770893e-01 2.22099811e-01 9.25120786e-02 -6.99816823e-01
-1.66020715e+00 7.19349325e-01 3.97787988e-01 -2.06255451e-01
-6.13954067e-01 7.04615712e-01 1.26319334e-01 2.32342824e-01
-9.65149477e-02 1.47876978e-01 3.97719920e-01 -2.84553945e-01
9.76548195e-01 8.08277726e-01 -1.25742838e-01 -8.09244335e-01
-3.30000281e-01 7.17353046e-01 4.29030180e-01 -5.91625154e-01
1.27774787e+00 -5.75075626e-01 -3.83384794e-01 4.26897496e-01
9.13599312e-01 4.52091277e-01 -1.52383268e+00 -4.42864299e-01
-2.33713523e-01 -8.88736129e-01 2.98738480e-02 -3.09106976e-01
-1.36617434e+00 5.78540266e-01 7.05344677e-01 3.57134223e-01
1.52977097e+00 -3.74902725e-01 6.68299615e-01 5.01577318e-01
6.19072199e-01 -7.56482601e-01 -7.73990899e-02 3.01361889e-01
4.59565818e-01 -8.54926825e-01 3.19287956e-01 -6.83966637e-01
1.68878597e-03 9.53411400e-01 -2.71420926e-01 -4.40766633e-01
7.84698486e-01 7.16416419e-01 -2.64900923e-03 7.99950883e-02
-6.04871452e-01 -1.17540963e-01 -2.89092541e-01 4.94849563e-01
-2.17837796e-01 1.45145014e-01 8.67625028e-02 -2.03034118e-01
1.82129771e-01 2.00037748e-01 9.47086930e-01 1.36888766e+00
-5.48983216e-01 -8.64627063e-01 -9.41938221e-01 2.00221851e-01
-5.27831316e-01 3.69232856e-02 -1.64035946e-01 4.50677931e-01
-1.28996342e-01 1.20243788e+00 -1.64052740e-01 8.96752905e-03
1.10491551e-02 -1.33205116e-01 9.47181821e-01 -2.03624457e-01
5.16714752e-02 4.75280464e-01 -1.65514156e-01 -6.52629733e-01
-9.07704890e-01 -1.17972255e+00 -8.08203936e-01 -3.83083194e-01
-3.14862430e-01 5.56797944e-02 5.29742181e-01 9.56275642e-01
1.38105005e-01 5.52327037e-01 6.11355960e-01 -1.18878448e+00
-5.54387748e-01 -6.27483964e-01 -8.32883060e-01 2.70201802e-01
6.44585729e-01 -3.38424265e-01 -6.42689109e-01 7.08918452e-01] | [9.057456016540527, -0.8683642148971558] |
52b07dbb-dfcc-42d7-8157-e44425f4652d | knowda-all-in-one-knowledge-mixture-model-for | 2206.10265 | null | https://arxiv.org/abs/2206.10265v2 | https://arxiv.org/pdf/2206.10265v2.pdf | KnowDA: All-in-One Knowledge Mixture Model for Data Augmentation in Low-Resource NLP | This paper focuses on the data augmentation for low-resource NLP tasks where the training set is limited. The existing solutions either leverage task-independent heuristic rules (e.g., Synonym Replacement) or fine-tune general-purpose pre-trained language models (e.g., GPT2) using the limited training instances to produce new synthetic data. Consequently, they have trivial task-specific knowledge and are limited to yielding low-quality synthetic data. To combat this issue, we propose Knowledge Mixture Data Augmentation Model (KnowDA) which is an Seq2Seq language model pre-trained on a mixture of diverse NLP tasks under a novel framework of Knowledge Mixture Training (KoMT). The goal of KoMT is to condense diverse NLP task-specific knowledge into the single KnowDA model (i.e., all-in-one) such that KnowDA could utilize these knowledge to quickly grasp the inherent synthesis law of the target task through limited training instances. Specifically, KoMT reformulates input examples from various heterogeneous NLP tasks into a unified text-to-text format, and employs denoising training objectives in different granularity to learn to reconstruct partial or complete samples. To the best of our knowledge, we are the first attempt to apply 100+ NLP multi-task training for data augmentation. Extensive experiments show that i) the synthetic data produced by KnowDA successfully improves performance of the strong pre-trained language models (i.e., Bert, ALBert and Deberta) by a large margin on the low-resource NLP benchmark FewGLUE, CoNLL'03 and WikiAnn; ii) KnowDA successfully transfers the task knowledge to NLP tasks whose types are seen and unseen in KoMT. | ['Daxin Jiang', 'Chongyang Tao', 'Tao Shen', 'Xiubo Geng', 'Can Xu', 'Jiayi Zheng', 'YuFei Wang'] | 2022-06-21 | null | null | null | null | ['few-shot-ner'] | ['natural-language-processing'] | [ 2.10413516e-01 2.24960193e-01 -5.11780441e-01 -1.99464783e-01
-1.08516979e+00 -6.15770340e-01 6.30930066e-01 -4.65911388e-01
-4.97516125e-01 1.16519892e+00 1.79991782e-01 -3.27697515e-01
1.35107905e-01 -6.61411226e-01 -9.79450166e-01 -6.49831593e-01
5.97275078e-01 1.04972279e+00 -2.53362864e-01 -4.00780410e-01
-3.31818521e-01 6.44138306e-02 -1.13908315e+00 5.99881947e-01
1.37664652e+00 6.13636196e-01 4.76087898e-01 3.66711318e-01
-6.94028318e-01 5.83604813e-01 -7.23322153e-01 -5.73678970e-01
3.19989532e-01 -1.90144360e-01 -8.35550904e-01 -1.48392692e-01
1.42472237e-01 -4.68365438e-02 -1.94000110e-01 8.53006065e-01
5.18853605e-01 7.69664198e-02 7.87503898e-01 -1.35046005e+00
-9.14847314e-01 1.19146407e+00 -6.48331583e-01 -5.11680506e-02
-4.29029576e-02 2.51715899e-01 7.33038247e-01 -1.29152727e+00
5.79178095e-01 1.31673229e+00 6.27210975e-01 7.21419573e-01
-1.25928080e+00 -9.26948786e-01 1.85309157e-01 -1.43810302e-01
-1.49324536e+00 -4.84159589e-01 6.00157559e-01 -8.09185579e-02
1.19160354e+00 -2.29552239e-02 2.55811572e-01 1.85355282e+00
-3.73005390e-01 1.23962712e+00 1.17679572e+00 -4.61705297e-01
1.01186719e-03 2.51173943e-01 1.60443500e-01 3.23573202e-01
2.73673475e-01 -1.32968828e-01 -6.12077177e-01 -2.33896300e-01
6.70006633e-01 -2.18242168e-01 -2.62455642e-01 -9.17962403e-04
-1.66641712e+00 6.16385102e-01 -2.54119225e-02 2.64832199e-01
-4.30722088e-01 1.58647739e-03 4.89516705e-01 3.90578806e-01
6.73295379e-01 6.95205271e-01 -1.29819465e+00 -1.46459281e-01
-7.80562520e-01 3.55529726e-01 8.68858874e-01 1.46050978e+00
7.10045576e-01 4.54956055e-01 -6.16869450e-01 1.15862453e+00
9.56830978e-02 7.52413809e-01 9.34651911e-01 -7.95788825e-01
1.04809391e+00 2.84581542e-01 1.35175377e-01 -1.46733165e-01
-2.07756430e-01 -5.97832561e-01 -1.03623235e+00 -3.37350130e-01
3.64282936e-01 -5.62803745e-01 -1.30299079e+00 2.01005936e+00
1.63444161e-01 2.76732475e-01 8.33766460e-01 3.54245782e-01
1.12053776e+00 1.04007995e+00 2.13827074e-01 -2.68869966e-01
1.30753982e+00 -1.29141462e+00 -8.60600650e-01 -6.37093723e-01
8.27278197e-01 -5.62134564e-01 1.66275954e+00 4.95899498e-01
-9.15072143e-01 -8.23540688e-01 -7.65624404e-01 4.74864021e-02
-6.02649629e-01 4.36248124e-01 6.22396350e-01 5.75431883e-01
-5.59421301e-01 1.43258020e-01 -5.36698401e-01 -1.46308094e-01
5.65622330e-01 9.93345678e-02 -3.94400179e-01 -5.79811931e-01
-1.45921445e+00 8.59510005e-01 1.12474084e+00 -2.08150484e-02
-1.04072309e+00 -1.13244081e+00 -8.57594550e-01 -8.37898329e-02
7.16929615e-01 -1.03572702e+00 1.28642011e+00 -1.02658522e+00
-1.65852642e+00 6.61062241e-01 -1.24250457e-01 -5.12238860e-01
4.45453227e-01 -4.53753561e-01 -4.28250939e-01 -3.93852353e-01
1.69188097e-01 1.01779366e+00 8.64857256e-01 -1.46745801e+00
-4.83903736e-01 9.78646874e-02 -3.05948913e-01 4.64989215e-01
-4.79074299e-01 1.58983953e-02 -5.65437853e-01 -1.06308281e+00
-2.55613983e-01 -7.49100447e-01 -1.77880406e-01 -7.59985924e-01
-6.55589998e-01 -2.53729582e-01 6.51435316e-01 -6.13917351e-01
9.14918363e-01 -1.88725948e+00 1.61033511e-01 -1.64495617e-01
6.09572753e-02 7.93443203e-01 -6.60849094e-01 2.57771254e-01
-1.27767816e-01 4.82527435e-01 -3.56862009e-01 -5.88887334e-01
-9.36262831e-02 7.84546852e-01 -5.79071224e-01 -2.54462540e-01
4.75354880e-01 1.31275463e+00 -1.00930595e+00 -4.91282701e-01
-5.59598729e-02 2.44503275e-01 -3.86354029e-01 2.62328982e-01
-8.41269493e-01 6.77334368e-01 -5.37062585e-01 6.10872686e-01
6.87120378e-01 -9.70159620e-02 -1.34255113e-02 -1.18828014e-01
2.23825470e-01 3.63365948e-01 -9.98545587e-01 1.94096732e+00
-8.19426477e-01 1.56682387e-01 -2.25590184e-01 -9.87376869e-01
8.99612188e-01 5.47046423e-01 1.10540055e-01 -3.62151444e-01
-5.00052199e-02 2.98133582e-01 1.64348874e-02 -5.42914987e-01
5.31520307e-01 -2.90894032e-01 -2.30932429e-01 2.24814385e-01
4.80232209e-01 -3.95995140e-01 3.94706637e-01 6.30493090e-02
1.03405368e+00 4.23367739e-01 2.62345701e-01 1.21785924e-02
2.11199433e-01 2.86790967e-01 7.78896809e-01 1.05479109e+00
2.54353225e-01 6.13986909e-01 1.94060206e-01 -6.13843501e-02
-1.12233686e+00 -1.15698552e+00 -6.02905601e-02 1.16565275e+00
-5.32778561e-01 -2.70697802e-01 -4.23269033e-01 -8.72583807e-01
-5.98648097e-03 1.28824866e+00 -3.97001386e-01 -1.72040790e-01
-5.47131419e-01 -1.20893967e+00 1.00955713e+00 4.74907994e-01
8.39466333e-01 -1.38933706e+00 4.40235943e-01 2.42572352e-01
-4.49414760e-01 -1.47874713e+00 -2.98261106e-01 4.25443918e-01
-7.88636863e-01 -6.17714286e-01 -8.44301462e-01 -8.99146974e-01
6.20828152e-01 1.10376462e-01 1.26171684e+00 -6.71600103e-01
2.54332393e-01 2.20950693e-01 -7.09600210e-01 -5.31880677e-01
-7.09105670e-01 3.41249883e-01 2.77633101e-01 -1.23429760e-01
4.00568634e-01 -6.26852810e-01 2.00908273e-01 1.36819690e-01
-9.76435304e-01 4.62969959e-01 1.10312474e+00 1.09260428e+00
9.81530547e-01 2.06659317e-01 1.14617026e+00 -1.23723435e+00
8.65135193e-01 -6.82531834e-01 -8.96064043e-02 5.10437667e-01
-2.83809543e-01 1.78714097e-01 1.00461960e+00 -1.04897630e+00
-1.49795973e+00 5.53903058e-02 -9.16103646e-02 -7.55855918e-01
-2.17656732e-01 8.20636988e-01 -6.79961622e-01 2.15493456e-01
8.62196565e-01 6.68558300e-01 -4.13009048e-01 -5.87415040e-01
8.99764299e-01 6.43838584e-01 6.81046784e-01 -1.08033061e+00
9.15971398e-01 6.11955449e-02 -4.99756217e-01 -6.72936916e-01
-1.27065384e+00 -3.06648284e-01 -5.82060754e-01 5.09307146e-01
5.21797299e-01 -1.25318325e+00 -7.69132301e-02 5.72344065e-01
-1.35507250e+00 -7.38248765e-01 -6.72174275e-01 5.18589497e-01
-6.04926884e-01 1.55777499e-01 -5.34945726e-01 -5.77035785e-01
-5.12532711e-01 -8.33016574e-01 9.62792218e-01 -8.07880238e-03
-1.60745755e-01 -8.66355896e-01 1.06609240e-01 6.81865633e-01
3.44592154e-01 -1.49158537e-01 1.08154273e+00 -1.20277750e+00
-2.96285570e-01 -5.48606403e-02 -1.79012343e-01 8.17962468e-01
1.64593205e-01 -2.17557475e-01 -1.08951557e+00 -1.17326947e-02
7.88458735e-02 -7.95606315e-01 9.41014767e-01 1.82549119e-01
1.32702637e+00 -5.21928072e-01 -2.95695662e-01 6.53132558e-01
9.48928237e-01 -8.66148844e-02 6.06744945e-01 9.39582884e-02
8.65287781e-01 3.78417611e-01 8.10777545e-01 1.41505167e-01
3.55698496e-01 5.26744604e-01 -7.18951300e-02 4.00911197e-02
-3.15786928e-01 -4.49158788e-01 5.18542409e-01 1.17384350e+00
1.76500306e-01 -6.37996137e-01 -1.19241452e+00 6.59782648e-01
-1.82736742e+00 -5.51678777e-01 -4.80096489e-02 1.90239882e+00
1.64624882e+00 1.36018649e-01 -3.49595875e-01 -2.57813305e-01
5.82234442e-01 -4.23754752e-02 -6.45904958e-01 4.09391001e-02
-5.80813646e-01 2.02039391e-01 3.34795266e-01 2.33734563e-01
-1.00534964e+00 1.55572164e+00 5.12350035e+00 1.51709843e+00
-6.11884117e-01 4.32494223e-01 5.32422900e-01 -5.27026989e-02
-3.88853341e-01 -1.60003871e-01 -1.20510674e+00 4.82917279e-01
9.52030718e-01 -2.67754316e-01 4.43580717e-01 7.61728704e-01
-6.95528165e-02 2.74193019e-01 -9.76596773e-01 9.69170034e-01
4.64933813e-02 -1.05584240e+00 5.14660001e-01 -2.29479715e-01
1.10787761e+00 3.40318531e-01 2.14503512e-01 1.19626606e+00
8.02314043e-01 -1.17434335e+00 3.57536525e-01 4.00673300e-01
8.85177255e-01 -5.28400302e-01 8.02340806e-01 9.00528371e-01
-7.50131249e-01 1.45666689e-01 -4.75987285e-01 1.99619785e-01
2.70146638e-01 8.72932017e-01 -1.16009629e+00 7.96040118e-01
3.33484471e-01 4.83562171e-01 -3.22289020e-01 6.80136204e-01
-5.11402607e-01 9.49255943e-01 -5.31556785e-01 2.45860189e-01
1.62991732e-01 6.95538521e-02 5.71813524e-01 1.23859882e+00
3.04840297e-01 -2.94945370e-02 2.57542074e-01 1.05782855e+00
-5.81175804e-01 2.10564062e-01 -5.87334096e-01 -4.04030949e-01
5.79936922e-01 1.25957739e+00 -1.30591184e-01 -5.73924839e-01
-5.16081989e-01 9.55493212e-01 5.02512276e-01 7.36469388e-01
-7.19341993e-01 -1.49912030e-01 4.28952754e-01 -3.34897041e-01
1.58167839e-01 -6.61473125e-02 -3.71223629e-01 -1.42854750e+00
4.53447811e-02 -1.11895859e+00 2.23890692e-01 -9.39763725e-01
-1.72748828e+00 6.94171846e-01 2.31966212e-01 -1.03735149e+00
-3.19096029e-01 -5.49253285e-01 -4.31298107e-01 1.25095379e+00
-1.64365375e+00 -1.53155625e+00 -1.48023471e-01 5.44374287e-01
8.02513719e-01 -6.35579407e-01 9.60084796e-01 1.14843071e-01
-7.54528582e-01 6.60151064e-01 2.30412766e-01 2.15823442e-01
9.83230770e-01 -1.16690934e+00 6.83215797e-01 6.21064126e-01
2.21582398e-01 5.37783206e-01 4.86550570e-01 -1.04501808e+00
-1.16554821e+00 -1.53115726e+00 7.73654580e-01 -7.42294908e-01
7.05337286e-01 -4.91294444e-01 -1.09943473e+00 9.42973733e-01
-1.64435450e-02 1.93923071e-01 6.78067744e-01 2.08431944e-01
-5.18626630e-01 9.81870517e-02 -1.07685328e+00 7.55644202e-01
1.00836599e+00 -2.69504577e-01 -9.57410395e-01 6.21567011e-01
1.29761636e+00 -5.73982418e-01 -8.28710496e-01 7.11541414e-01
4.98068780e-02 -4.02650125e-02 1.06836116e+00 -1.21540892e+00
4.92193937e-01 -2.00979292e-01 -2.07084745e-01 -1.63755417e+00
9.95589271e-02 -5.40587008e-01 -2.41909772e-01 1.60104358e+00
8.50062668e-01 -6.65902257e-01 8.09059203e-01 3.25277925e-01
-4.83068585e-01 -6.82023823e-01 -9.31428373e-01 -1.02880359e+00
3.90790522e-01 -6.75234556e-01 4.51377064e-01 1.35616601e+00
-3.35807234e-01 6.43516958e-01 -5.38773954e-01 4.70750667e-02
3.05911660e-01 -2.32581571e-01 9.56652462e-01 -9.37297344e-01
-4.39743251e-01 -7.13853389e-02 4.55607653e-01 -1.13054276e+00
6.01952314e-01 -1.25550807e+00 6.32524490e-02 -1.41054797e+00
2.66145349e-01 -7.89248168e-01 -1.73577577e-01 1.04699481e+00
-4.78550166e-01 -2.51027465e-01 -7.31037408e-02 2.47630984e-01
-4.75016892e-01 1.03128445e+00 1.43028915e+00 -1.63833484e-01
-4.81959611e-01 1.32017121e-01 -8.73395860e-01 6.77897692e-01
1.03965008e+00 -6.59066081e-01 -6.15144432e-01 -6.58926904e-01
3.60972792e-01 -1.10285737e-01 2.06925511e-01 -6.12617910e-01
9.01521072e-02 -2.18330041e-01 2.49326453e-01 -5.13874650e-01
4.87318307e-01 -6.85936570e-01 -7.99904764e-02 1.10094277e-02
-3.08887511e-01 -3.18343610e-01 5.84360719e-01 5.59529960e-01
-2.70620078e-01 -4.28661644e-01 4.20181006e-01 -3.13598186e-01
-6.15902841e-01 3.51088583e-01 -1.54357374e-01 6.01310372e-01
7.38259315e-01 1.74037695e-01 -6.93137765e-01 -1.07738204e-01
-5.96245646e-01 5.25837779e-01 6.66276217e-02 4.83856916e-01
4.16830420e-01 -1.40791440e+00 -1.12219083e+00 7.05047920e-02
2.00598195e-01 6.24338150e-01 3.71334970e-01 6.24750018e-01
-4.20366786e-03 4.68455881e-01 4.01313454e-02 -2.83414721e-01
-7.39256263e-01 7.51988769e-01 2.33905807e-01 -7.55520225e-01
-4.58549619e-01 8.95363808e-01 3.48768204e-01 -1.15130138e+00
1.00784965e-01 -4.08548862e-01 -7.29786232e-02 2.00915784e-02
3.59453201e-01 -5.12734149e-03 9.20088869e-03 -2.19751462e-01
-5.76286111e-03 8.45296010e-02 -4.00486648e-01 -1.68927267e-01
1.19338942e+00 1.91181526e-01 6.33949190e-02 4.70960081e-01
9.49086249e-01 7.53011182e-02 -1.00263262e+00 -7.49701440e-01
-1.00723140e-01 -8.38698596e-02 -1.60663307e-01 -1.28021991e+00
-8.70505929e-01 7.98261642e-01 -8.70190710e-02 -4.27279860e-01
9.64598596e-01 2.59541012e-02 8.10809374e-01 9.59855676e-01
2.21118987e-01 -9.79645789e-01 2.31292471e-01 9.03023541e-01
1.04887617e+00 -1.26863539e+00 -2.81309932e-01 -5.64244807e-01
-1.09862590e+00 6.89046562e-01 8.77974629e-01 3.42313707e-01
3.03126425e-01 3.73588860e-01 -6.39321804e-02 2.00625256e-01
-1.06095314e+00 -3.15614134e-01 1.45482272e-01 9.04621959e-01
1.14031225e-01 2.98787430e-02 1.70327410e-01 1.31008196e+00
-2.61098206e-01 1.27479747e-01 2.04895064e-01 5.06345272e-01
-2.44143501e-01 -1.26450014e+00 -2.54789233e-01 6.11536622e-01
-2.76008308e-01 -6.90146804e-01 -1.90032095e-01 9.05700684e-01
2.48073637e-01 6.00304961e-01 -3.75494421e-01 -2.27749810e-01
3.77701074e-01 5.61864376e-01 3.02662522e-01 -1.06293118e+00
-5.73227644e-01 -1.30945370e-01 4.13056105e-01 3.94376963e-02
-2.18787283e-01 -4.15765226e-01 -1.04906118e+00 3.59756351e-02
-5.39659224e-02 1.14104539e-01 4.91361082e-01 1.07239056e+00
3.93434495e-01 7.62630522e-01 6.71450943e-02 -6.21286094e-01
-7.18869567e-01 -1.47299302e+00 -3.73536885e-01 2.54392266e-01
-1.20177105e-01 -4.81816590e-01 -2.19193920e-01 -4.45081964e-02] | [11.004700660705566, 8.420083999633789] |
3dc2a1fc-d726-4f2a-8a81-00df555bda94 | automatic-right-ventricle-segmentation-using | 2004.02317 | null | https://arxiv.org/abs/2004.02317v1 | https://arxiv.org/pdf/2004.02317v1.pdf | Automatic Right Ventricle Segmentation using Multi-Label Fusion in Cardiac MRI | Accurate segmentation of the right ventricle (RV) is a crucial step in the assessment of the ventricular structure and function. Yet, due to its complex anatomy and motion segmentation of the RV has not been as largely studied as the left ventricle. This paper presents a fully automatic method for the segmentation of the RV in cardiac magnetic resonance images (MRI). The method uses a coarse-to-fine segmentation strategy in combination with a multi-atlas propagation segmentation framework. Based on a cross correlation metric, our method selects the best atlases for propagation allowing the refinement of the segmentation at each iteration of the propagation. The proposed method was evaluated on 32 cardiac MRI datasets provided by the RV Segmentation Challenge in Cardiac MRI. | ['Sébastien Ourselin', 'Maria A. Zuluaga', 'M. Jorge Cardoso'] | 2020-04-05 | null | null | null | null | ['motion-segmentation'] | ['computer-vision'] | [-7.09149707e-03 5.61155826e-02 2.96467572e-01 -2.32269078e-01
-4.04692948e-01 -7.07335234e-01 1.04061142e-01 2.07305685e-01
-4.56639677e-01 6.40028119e-01 -3.61577012e-02 -2.55954325e-01
3.95706072e-02 -4.15950924e-01 8.70693401e-02 -7.47308671e-01
4.70247632e-03 1.11903799e+00 7.18412399e-01 7.66018480e-02
1.86340809e-01 9.69304562e-01 -6.15229309e-01 -1.08889900e-01
6.97922647e-01 4.92551982e-01 3.31579745e-01 9.72289801e-01
-1.12218425e-01 5.31338215e-01 -3.36925924e-01 -6.73000962e-02
2.35575765e-01 -7.45614171e-01 -1.05790186e+00 2.15797961e-01
1.49164766e-01 -2.28176892e-01 3.23271364e-01 7.47101784e-01
6.79972053e-01 -7.37412646e-02 6.03217542e-01 -5.27421594e-01
2.26968169e-01 6.08599901e-01 -4.35908943e-01 8.20145726e-01
-4.04597998e-01 -9.34224799e-02 6.42905533e-01 -1.05187678e+00
9.74802554e-01 3.12690556e-01 7.61411190e-01 3.51884544e-01
-1.33960009e+00 -3.50187480e-01 -3.89559090e-01 -2.48374507e-01
-1.46758771e+00 -9.30878799e-03 7.33843803e-01 -1.10081649e+00
5.55022657e-01 -4.54630665e-02 1.05044925e+00 -3.07516884e-02
3.39650422e-01 3.05139393e-01 1.24470484e+00 -2.09313467e-01
2.32043684e-01 -1.57112062e-01 2.62152374e-01 6.84529960e-01
1.45866439e-01 -2.00506449e-01 7.06961453e-02 -2.69535124e-01
1.10758626e+00 -3.25513810e-01 -1.60024047e-01 -5.29128194e-01
-1.18556142e+00 7.38875508e-01 2.34517813e-01 8.94087970e-01
-7.06278086e-01 2.96197869e-02 4.61501002e-01 -1.34834439e-01
4.56972033e-01 2.04616979e-01 -2.25248531e-01 3.11146956e-02
-1.68784463e+00 2.64218420e-01 5.05916297e-01 2.46900365e-01
2.89692104e-01 -2.52313679e-03 -2.90478170e-01 7.48398900e-01
6.24357462e-01 2.30794609e-01 1.25451639e-01 -1.03871906e+00
7.98177421e-02 3.78054142e-01 -1.64154157e-01 -7.59515643e-01
-8.41863394e-01 -7.92672276e-01 -1.00955820e+00 4.95615631e-01
6.55989170e-01 -2.35644922e-01 -9.16900218e-01 1.09452045e+00
7.67776608e-01 -1.03957005e-01 -4.07173246e-01 1.02736509e+00
7.49596179e-01 1.58267319e-01 1.64372340e-01 -5.98063052e-01
1.24306166e+00 -7.33268082e-01 -6.18078291e-01 1.41093984e-01
5.72181821e-01 -6.98088050e-01 2.85957456e-01 1.32358611e-01
-1.23991632e+00 -4.03929412e-01 -8.99218798e-01 3.23852956e-01
3.15411866e-01 6.47374019e-02 6.84225708e-02 7.17193604e-01
-1.19003236e+00 5.57701409e-01 -1.17573977e+00 1.09965168e-01
5.94575882e-01 3.99103373e-01 -3.78099680e-01 4.15116549e-01
-7.86058068e-01 1.05608630e+00 4.43664908e-01 3.21110964e-01
-4.95657384e-01 -6.93264961e-01 -4.64038789e-01 -3.30179513e-01
2.26993486e-02 -9.27184522e-01 9.69642639e-01 -7.08120763e-01
-1.36016285e+00 1.24342382e+00 -5.60554750e-02 -4.89942968e-01
9.09544945e-01 2.01464370e-01 1.56844482e-01 6.38308227e-01
9.82281640e-02 3.72133881e-01 8.95894170e-01 -1.31846750e+00
-4.34887439e-01 -6.29057884e-01 -4.75029409e-01 1.84380904e-01
7.91661859e-01 3.44897360e-01 -3.31300586e-01 -8.35829318e-01
5.99337459e-01 -1.09290707e+00 -5.69754243e-01 -3.82074237e-01
-1.21340334e-01 2.63036042e-01 5.38677454e-01 -1.27818310e+00
1.36272311e+00 -1.77108979e+00 3.08973461e-01 6.48912132e-01
8.84329021e-01 3.06867778e-01 6.42704964e-01 -1.50400504e-01
6.76298290e-02 1.36882916e-01 -6.80069327e-01 -1.71758726e-01
-7.17818379e-01 1.13586374e-01 2.40632772e-01 8.24214756e-01
-4.46359664e-01 9.12130535e-01 -7.22787201e-01 -9.57526743e-01
2.48951524e-01 4.76776987e-01 -2.61135012e-01 1.41262129e-01
2.81022877e-01 1.36004114e+00 -5.10921001e-01 2.32537761e-01
6.22551858e-01 -2.69441247e-01 3.44701588e-01 -4.44581285e-02
-1.49243400e-01 -3.19887340e-01 -9.11128819e-01 1.50022233e+00
-9.30372030e-02 2.38284588e-01 3.73943269e-01 -7.47452557e-01
9.08862889e-01 7.34360814e-01 1.07728696e+00 -2.58668870e-01
3.09692770e-01 4.56886351e-01 4.22809750e-01 -1.82142824e-01
-9.12706368e-03 -6.17055416e-01 4.24714088e-01 4.78182405e-01
4.34091724e-02 -4.63967443e-01 2.78762192e-01 1.03476988e-02
8.27434063e-01 3.55282694e-01 4.09560144e-01 -5.58091938e-01
7.68943846e-01 1.58109128e-01 6.59273624e-01 4.13554490e-01
-3.60887468e-01 1.02910268e+00 2.55897284e-01 -5.48166573e-01
-1.11014915e+00 -8.91725540e-01 -5.17160296e-01 3.33751202e-01
-2.50821769e-01 -6.05713464e-02 -1.12045300e+00 -7.12596059e-01
-3.22370052e-01 2.62612879e-01 -4.25455540e-01 5.75820029e-01
-1.04369640e+00 -8.56668532e-01 5.82911670e-01 3.26804012e-01
2.71477103e-01 -1.13697124e+00 -1.20046890e+00 5.31913757e-01
-3.06199610e-01 -1.16887522e+00 -4.25408602e-01 -1.82518885e-01
-1.49719143e+00 -1.14073920e+00 -1.11432767e+00 -7.44872093e-01
7.59826422e-01 -4.24078107e-01 1.27847970e+00 4.09407914e-01
-4.65177506e-01 1.62161201e-01 -1.50293112e-01 -6.39541820e-03
-9.27215338e-01 -1.13093061e-02 -4.16124195e-01 -8.24704021e-02
-4.26744223e-01 -5.65013051e-01 -6.79621816e-01 4.22173232e-01
-6.54691637e-01 1.96073472e-01 2.18092039e-01 6.61194563e-01
1.05598187e+00 -2.07589239e-01 4.35197026e-01 -1.17720401e+00
4.70357746e-01 -3.65047187e-01 -8.59380841e-01 1.47147149e-01
-6.86996877e-01 -3.32909435e-01 2.88820356e-01 1.15174457e-01
-7.44063377e-01 2.73918569e-01 -4.12674159e-01 -2.84383684e-01
-2.72418767e-01 3.84056509e-01 5.27545631e-01 -4.84198242e-01
3.68977904e-01 1.54813109e-02 2.14970391e-02 -3.78649652e-01
2.68807352e-01 3.71298529e-02 4.51500744e-01 -2.57061899e-01
5.87782204e-01 4.31567341e-01 5.51100612e-01 -8.12431335e-01
-2.63725609e-01 -6.01699531e-01 -1.52017951e+00 -6.99060917e-01
1.38371754e+00 -1.51997909e-01 -3.09876859e-01 3.67827445e-01
-1.12476182e+00 -3.62749726e-01 -4.33830976e-01 6.14006162e-01
-6.06043458e-01 5.62481999e-01 -7.20154643e-01 -5.49013793e-01
-7.91001201e-01 -1.41622329e+00 6.25252485e-01 -3.05292234e-02
-3.05436254e-01 -1.29742467e+00 4.92264926e-01 5.04699588e-01
5.61874628e-01 6.81251824e-01 9.55193341e-01 -5.79775631e-01
-4.09193248e-01 7.02256570e-03 2.29334623e-01 4.57159132e-01
3.98125798e-02 2.69484557e-02 -5.06789565e-01 -5.90798147e-02
2.14411721e-01 2.38002777e-01 6.36821628e-01 1.02546275e+00
7.44551241e-01 3.22031438e-01 -1.43144664e-03 6.19874895e-01
1.41121066e+00 2.55693465e-01 4.08746928e-01 8.86484236e-02
8.43278050e-01 5.32486260e-01 5.95812738e-01 5.04507065e-01
3.61429125e-01 6.09886289e-01 2.52519965e-01 -2.93207794e-01
-3.75715166e-01 3.69534940e-01 -4.35277939e-01 1.01632297e+00
-6.97199404e-01 3.75075430e-01 -1.44699919e+00 5.21006703e-01
-1.41291571e+00 -6.55982673e-01 -6.05901480e-01 2.33290672e+00
6.00351334e-01 7.47850239e-02 2.98382849e-01 1.00500442e-01
7.39821196e-01 -1.08157426e-01 -4.86691706e-02 -3.83903265e-01
2.98000518e-02 4.08865422e-01 4.67258513e-01 8.12029302e-01
-1.02909315e+00 6.09120548e-01 7.00560188e+00 2.60810703e-01
-1.01690567e+00 5.15439451e-01 7.33020484e-01 5.22103488e-01
-5.44166490e-02 6.03160188e-02 -7.19951153e-01 2.32873976e-01
4.99756664e-01 1.69508010e-01 2.37627968e-01 2.96086311e-01
3.51733953e-01 -3.30336392e-01 -5.87052703e-01 6.66562617e-01
-1.95043489e-01 -1.41548216e+00 -4.15048361e-01 5.72830923e-02
6.78888619e-01 2.38373905e-01 -4.76907313e-01 -2.79154807e-01
-2.90700555e-01 -9.49420094e-01 6.60780668e-01 6.44912481e-01
6.29100621e-01 -6.39623284e-01 9.52661633e-01 4.35361773e-01
-1.17865944e+00 4.03817624e-01 1.63547620e-01 5.31314433e-01
5.37561357e-01 6.50875330e-01 -1.16700923e+00 3.98490757e-01
3.76255184e-01 1.37524456e-01 -4.52140391e-01 1.24456227e+00
-9.94434506e-02 9.03607845e-01 -1.44837767e-01 6.26665115e-01
-5.05306274e-02 -6.76233649e-01 9.83965456e-01 1.25744438e+00
3.94104347e-02 2.18249068e-01 4.67474699e-01 9.14879680e-01
6.11517467e-02 6.94356382e-01 -3.45921248e-01 2.66020507e-01
7.89952651e-03 1.43368387e+00 -1.54169083e+00 -2.68618792e-01
-1.05307236e-01 5.97665846e-01 1.13659091e-01 3.08311880e-01
-4.93713915e-01 3.10410410e-01 -2.82138675e-01 6.27424955e-01
2.21296713e-01 -3.43863785e-01 -6.68138385e-01 -6.69843197e-01
-8.17155465e-02 -4.80923563e-01 4.78227526e-01 -4.15654719e-01
-7.09900200e-01 8.98710847e-01 1.58372983e-01 -9.11584139e-01
-3.42715412e-01 -5.38698435e-02 -4.15317208e-01 1.32407606e+00
-1.10598481e+00 -1.08591819e+00 -6.16506375e-02 2.54514873e-01
3.22773814e-01 3.24390754e-02 8.21393073e-01 3.44240099e-01
-9.66213122e-02 -1.32217973e-01 -2.79143572e-01 2.70687699e-01
1.20815516e-01 -1.55036533e+00 2.04100564e-01 9.31840479e-01
-2.58258432e-01 4.91325706e-01 5.77901244e-01 -1.03597236e+00
-5.40792346e-01 -8.82949591e-01 8.32061529e-01 -2.51812875e-01
3.39182764e-01 3.19288224e-01 -9.03750479e-01 5.84590495e-01
4.50912379e-02 5.89883387e-01 7.60856926e-01 -5.10639191e-01
5.18951416e-01 2.05080464e-01 -1.33405852e+00 4.83301133e-02
3.61110359e-01 -8.68868083e-02 -5.60904503e-01 1.05750775e-02
-5.75271901e-03 -7.85777688e-01 -1.45334244e+00 5.72700500e-01
4.50122684e-01 -9.62867200e-01 9.64567602e-01 4.85571520e-03
6.73651993e-02 -5.64630866e-01 3.79450351e-01 -1.02048790e+00
-3.42979580e-01 -5.46544909e-01 -1.09197050e-01 9.60055709e-01
2.80420929e-01 -4.52833891e-01 7.69157052e-01 3.93795013e-01
-1.02111563e-01 -9.30461049e-01 -1.09897316e+00 -2.02571154e-01
2.39375040e-01 -2.57480353e-01 9.25740413e-03 6.91721678e-01
-7.57747531e-01 1.14699215e-01 4.21261089e-03 4.84587587e-02
9.26161110e-01 1.42817810e-01 3.58521074e-01 -1.53334177e+00
-3.36105615e-01 -7.20472336e-01 -5.08620083e-01 -4.48515415e-01
-1.14365429e-01 -1.14901745e+00 -4.11478989e-02 -1.81350541e+00
2.39572391e-01 -7.06191659e-01 -2.54651040e-01 -4.32282016e-02
-1.89433545e-01 4.27470922e-01 2.54159898e-01 4.66219723e-01
-1.14298150e-01 -8.78062695e-02 1.70971811e+00 4.43233818e-01
-3.53693038e-01 4.50862557e-01 5.75748011e-02 8.89227748e-01
8.79162133e-01 -6.89411640e-01 -2.29292661e-01 6.01105839e-02
-3.94755974e-02 6.80460751e-01 3.14739108e-01 -7.39476621e-01
4.29170206e-03 2.97373623e-01 3.18368375e-01 -9.67455804e-01
-2.81837466e-03 -7.66177952e-01 3.24488431e-01 7.75433481e-01
-8.23531225e-02 4.08680260e-01 -7.31023699e-02 5.57465777e-02
-1.85095698e-01 -4.59360838e-01 1.41756713e+00 -4.60963815e-01
-2.07425445e-01 3.99507970e-01 -5.12414634e-01 4.83828038e-01
1.07550240e+00 -3.13465118e-01 5.45745492e-01 6.78426176e-02
-1.44874310e+00 8.43049139e-02 2.52322823e-01 -2.02222720e-01
7.09594667e-01 -8.77413809e-01 -1.07093000e+00 -7.46860504e-02
-4.41262186e-01 1.69034228e-01 3.06650639e-01 1.69986892e+00
-1.34792554e+00 2.53764927e-01 -2.73681581e-01 -9.69780922e-01
-1.60517240e+00 1.60791203e-01 1.03532052e+00 -8.61400962e-01
-1.13865745e+00 5.80365002e-01 1.35506719e-01 -3.07182372e-01
-1.76841587e-01 -3.28176647e-01 -6.93319321e-01 -2.34237183e-02
2.01506600e-01 5.68864167e-01 8.01368207e-02 -1.32986450e+00
-4.49108720e-01 8.93197060e-01 4.16484177e-01 -2.51028210e-01
1.07673049e+00 -4.35389668e-01 -5.51427543e-01 5.47471941e-01
6.83747113e-01 1.61534622e-01 -8.40338886e-01 -5.73004819e-02
2.77162105e-01 -2.45916665e-01 4.37648714e-01 -7.70620704e-01
-1.29037249e+00 9.23258185e-01 7.02585280e-01 8.69204327e-02
9.58569407e-01 -2.44045362e-01 5.83246469e-01 -5.77815652e-01
3.85879457e-01 -1.01892626e+00 -5.15482903e-01 2.54435480e-01
7.96917021e-01 -8.82268906e-01 1.22092396e-01 -5.04312873e-01
-9.31155145e-01 1.22790849e+00 1.14867792e-01 -1.23820670e-01
9.00601745e-01 3.62891555e-01 3.32113951e-01 -2.94935852e-01
-4.98929471e-02 -1.62381127e-01 5.22022426e-01 3.61639440e-01
6.89543366e-01 1.38040453e-01 -7.71316051e-01 1.92051098e-01
2.89313048e-01 1.56856239e-01 3.34172577e-01 7.29030132e-01
-4.43510771e-01 -1.09772801e+00 -4.45633590e-01 2.83566207e-01
-1.08521867e+00 -8.80021527e-02 -9.70473886e-02 5.17468989e-01
3.02775115e-01 5.60759366e-01 -4.10360575e-01 3.08659226e-01
1.86484918e-01 1.29929602e-01 6.56507969e-01 -6.58367991e-01
-1.13349175e+00 4.53811914e-01 -1.18612036e-01 -2.58017570e-01
-4.55022782e-01 -8.79767179e-01 -1.93709385e+00 2.31109381e-01
1.51659234e-03 1.04776487e-01 8.71417105e-01 1.15171123e+00
-1.03072502e-01 7.23681092e-01 6.02878749e-01 -7.19239295e-01
-1.88373059e-01 -7.67044842e-01 -7.24635482e-01 3.59403312e-01
1.82789445e-01 -4.15139705e-01 1.06272958e-02 1.23577967e-01] | [14.143900871276855, -2.521662473678589] |
c761733a-2b35-44fd-997f-534ea64e1ef2 | recurrent-models-for-situation-recognition | 1703.06233 | null | http://arxiv.org/abs/1703.06233v2 | http://arxiv.org/pdf/1703.06233v2.pdf | Recurrent Models for Situation Recognition | This work proposes Recurrent Neural Network (RNN) models to predict
structured 'image situations' -- actions and noun entities fulfilling semantic
roles related to the action. In contrast to prior work relying on Conditional
Random Fields (CRFs), we use a specialized action prediction network followed
by an RNN for noun prediction. Our system obtains state-of-the-art accuracy on
the challenging recent imSitu dataset, beating CRF-based models, including ones
trained with additional data. Further, we show that specialized features
learned from situation prediction can be transferred to the task of image
captioning to more accurately describe human-object interactions. | ['Svetlana Lazebnik', 'Arun Mallya'] | 2017-03-18 | recurrent-models-for-situation-recognition-1 | http://openaccess.thecvf.com/content_iccv_2017/html/Mallya_Recurrent_Models_for_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Mallya_Recurrent_Models_for_ICCV_2017_paper.pdf | iccv-2017-10 | ['grounded-situation-recognition', 'situation-recognition'] | ['computer-vision', 'computer-vision'] | [ 7.36545503e-01 6.53632581e-01 -3.12158376e-01 -8.20427358e-01
-6.10763431e-01 -2.90845424e-01 9.11207557e-01 -1.17355533e-01
-6.71345890e-01 7.04933226e-01 1.01848614e+00 -1.26922235e-01
2.16608420e-01 -4.69905287e-01 -8.80195975e-01 -3.54748517e-01
2.37320900e-01 7.28224754e-01 9.12116002e-03 5.62352547e-03
1.55988887e-01 2.79323161e-01 -1.43515933e+00 8.46963763e-01
3.04698676e-01 8.71865332e-01 3.86233956e-01 4.54596311e-01
-6.80340379e-02 1.65699780e+00 -3.33592713e-01 -5.13184488e-01
-6.27833083e-02 -1.87741056e-01 -1.45284081e+00 3.37575376e-01
2.93810785e-01 -4.53852534e-01 -7.54478872e-01 7.11130500e-01
2.80301929e-01 3.41782033e-01 6.86891139e-01 -1.06958258e+00
-1.05636716e+00 8.01925480e-01 3.48175466e-02 6.05774391e-03
5.92623413e-01 -2.17286963e-03 1.24417841e+00 -7.33910143e-01
1.32586265e+00 1.33990121e+00 3.03014249e-01 1.08612990e+00
-1.20358884e+00 -3.54874313e-01 3.03094715e-01 5.08792043e-01
-9.36942160e-01 -4.02995050e-01 5.01406968e-01 -2.49355748e-01
1.47091186e+00 -1.37149170e-01 2.98209846e-01 1.64120209e+00
-1.02413207e-01 1.42708373e+00 7.24184394e-01 -3.32579225e-01
-1.11918382e-01 -1.97170824e-02 1.37479156e-01 3.74983877e-01
-3.60781372e-01 -2.00816736e-01 -3.73912543e-01 3.59743655e-01
7.90605307e-01 -8.29012766e-02 -1.58623680e-01 -2.99938232e-01
-1.71964967e+00 5.60947418e-01 4.32628334e-01 5.47512829e-01
-8.31308246e-01 4.86048520e-01 5.24665296e-01 -5.27363755e-02
5.45636594e-01 5.11479497e-01 -9.32742238e-01 -3.86101961e-01
-6.73443615e-01 2.85835803e-01 7.32684195e-01 1.07362962e+00
4.96237129e-01 -1.04550235e-01 -6.59550667e-01 7.53075838e-01
2.71641821e-01 2.73107469e-01 4.22849953e-01 -1.18251753e+00
5.86716115e-01 4.44217652e-01 3.05707335e-01 -5.07434666e-01
-2.67229080e-01 -1.54269248e-01 -6.36570692e-01 -4.81648266e-01
3.80286604e-01 9.54246372e-02 -1.09523320e+00 1.76255798e+00
-1.34700999e-01 4.72589552e-01 4.78236467e-01 7.73520529e-01
1.06574094e+00 7.80267715e-01 8.15037608e-01 -2.26984873e-01
1.39970839e+00 -1.06774390e+00 -8.22070241e-01 -1.97138757e-01
7.10055172e-01 -4.16161984e-01 8.80049706e-01 1.64391056e-01
-1.16136813e+00 -4.07336175e-01 -1.81066081e-01 -4.31982130e-01
-4.78893250e-01 2.18302906e-01 8.15508544e-01 7.44216740e-02
-1.40441167e+00 5.44232726e-01 -5.59409797e-01 -5.00171542e-01
7.42594719e-01 5.01155138e-01 -5.65249264e-01 -9.20598954e-02
-1.12320900e+00 1.11973488e+00 5.86587012e-01 1.64863706e-01
-1.10435486e+00 -5.06264508e-01 -1.09162068e+00 1.07579879e-01
5.24434447e-01 -6.17866993e-01 1.62660992e+00 -1.15834224e+00
-1.42093503e+00 1.23142433e+00 -5.01730680e-01 -9.71731007e-01
-1.97505746e-02 -3.39676112e-01 -2.78367788e-01 4.11553353e-01
2.52711743e-01 1.47464013e+00 4.23476249e-01 -1.17786193e+00
-4.00721431e-01 -1.04572237e-01 3.90570879e-01 2.61109859e-01
-1.79331079e-01 7.32811809e-01 -4.99517202e-01 -5.93656957e-01
-5.46223484e-02 -9.19014335e-01 -6.41572237e-01 -4.47909147e-01
-5.33029556e-01 -8.30716431e-01 8.27673078e-01 -6.84640467e-01
8.03073227e-01 -1.98261285e+00 3.85026723e-01 -2.64881343e-01
6.96480274e-02 2.45312050e-01 -4.29434687e-01 9.89058912e-02
-4.99382436e-01 8.76788571e-02 1.01335887e-02 -5.44001579e-01
2.62951672e-01 5.17937839e-01 -4.98691589e-01 2.08546594e-01
4.02766734e-01 1.42898464e+00 -8.47129405e-01 -5.69107175e-01
2.48807311e-01 4.66161042e-01 -3.93975824e-01 4.06974316e-01
-8.26249540e-01 6.03539050e-01 -4.65584546e-01 6.17969751e-01
2.30155930e-01 -6.63797617e-01 4.52950329e-01 -2.83627033e-01
1.68278232e-01 5.72405338e-01 -6.47429466e-01 1.67756987e+00
-5.13589859e-01 4.68470812e-01 -2.21456856e-01 -1.24907184e+00
4.91193473e-01 8.06638539e-01 5.32204509e-01 -6.50180399e-01
1.15175754e-01 -2.17747718e-01 -4.03387159e-01 -6.79777682e-01
4.46812272e-01 5.38522527e-02 -2.81661183e-01 4.76879537e-01
5.67405403e-01 2.14195281e-01 1.94487944e-01 3.98795158e-01
9.62096274e-01 7.95748889e-01 5.05943775e-01 1.12972505e-01
5.78309357e-01 -2.56659746e-01 6.39231861e-01 6.65556133e-01
-3.75277966e-01 4.39310789e-01 6.15511954e-01 -6.15084350e-01
-9.70576882e-01 -8.16062748e-01 7.56673887e-02 1.53892648e+00
8.00532997e-02 -5.03809869e-01 -6.56303585e-01 -1.07943070e+00
-4.88713741e-01 1.06472385e+00 -6.67515695e-01 3.30450982e-01
-8.46197188e-01 -4.60943758e-01 4.16700304e-01 9.87604499e-01
5.05968392e-01 -1.95056474e+00 -5.40037096e-01 2.85117775e-01
-4.23952013e-01 -1.83142197e+00 -1.55451775e-01 1.55165672e-01
-5.48364401e-01 -8.67563963e-01 -6.26223087e-01 -9.94690895e-01
7.27840602e-01 -1.93565607e-01 1.47347212e+00 -2.09061995e-01
1.52028263e-01 7.71576524e-01 -4.85949129e-01 -3.92272532e-01
-5.21185398e-01 -1.42534807e-01 -1.42204955e-01 -1.26436595e-02
4.64265555e-01 -5.40447772e-01 -3.84917855e-01 2.53506064e-01
-9.35318112e-01 4.87016320e-01 6.91760480e-01 5.38812876e-01
5.18103659e-01 -4.48323905e-01 3.82857293e-01 -9.82229590e-01
1.24219067e-01 -4.86293077e-01 -3.10625285e-01 5.25018334e-01
-1.86200008e-01 1.67269081e-01 4.46291000e-01 -4.36246127e-01
-1.59884477e+00 6.24392807e-01 -1.28299713e-01 -4.34132725e-01
-8.38966966e-01 1.79408044e-01 -3.10448885e-01 4.85492259e-01
2.86068290e-01 3.35373074e-01 -4.39105779e-01 -3.83079708e-01
6.02098644e-01 4.91733700e-01 8.19061399e-01 -5.15872657e-01
3.96830976e-01 4.41659987e-01 9.75254737e-03 -3.58938813e-01
-1.32478666e+00 -5.90687275e-01 -8.85697365e-01 -3.04381073e-01
1.60114110e+00 -1.06705630e+00 -8.06603074e-01 2.81982452e-01
-1.74697959e+00 -3.17236781e-01 -3.57882082e-01 5.06648719e-01
-9.47209597e-01 2.50785947e-01 -1.03234327e+00 -7.90902078e-01
-1.57659113e-01 -1.14568269e+00 1.47835934e+00 1.29481375e-01
-2.01236472e-01 -8.47100019e-01 -3.29673290e-01 7.62081146e-01
1.57793090e-01 -2.53371187e-02 7.27819145e-01 -9.45107639e-01
-8.71552229e-01 8.02480336e-03 -4.42609370e-01 2.58614689e-01
-2.97223985e-01 -5.82158208e-01 -7.88995385e-01 3.36011946e-01
-6.89305440e-02 -6.90382421e-01 9.15403783e-01 4.99570310e-01
1.45566452e+00 -3.91613275e-01 -4.49980587e-01 1.14314072e-01
1.06771970e+00 4.93706129e-02 1.24782658e+00 1.84711039e-01
6.71361923e-01 7.55916476e-01 6.51368558e-01 2.10956439e-01
6.19509220e-01 7.03973830e-01 5.52875698e-01 6.61702082e-02
-1.08641796e-01 -2.31289282e-01 4.62994128e-01 8.88398364e-02
-3.56655419e-01 -4.79683518e-01 -9.74580228e-01 6.80207551e-01
-2.14858842e+00 -1.27315533e+00 -1.13409549e-01 1.41925776e+00
7.67706335e-01 -6.30276799e-02 -6.05871305e-02 -4.39403117e-01
8.29375744e-01 3.09858650e-01 -2.41118401e-01 -2.06227764e-01
-1.76770195e-01 1.29314065e-01 4.16583002e-01 1.52837738e-01
-1.48467219e+00 1.70430410e+00 6.99083996e+00 6.24004543e-01
-6.41761482e-01 2.93288410e-01 7.23527610e-01 1.67433292e-01
-2.43633151e-01 1.32072166e-01 -8.66638780e-01 9.93597806e-02
1.28515625e+00 4.96372372e-01 2.76346207e-01 8.92242193e-01
3.47748578e-01 -1.17709167e-01 -1.16519630e+00 7.26877451e-01
3.55587959e-01 -1.46724439e+00 1.42869234e-01 1.02010228e-01
4.63904142e-01 2.16688871e-01 -1.30383328e-01 4.99681324e-01
6.37642801e-01 -1.10876548e+00 5.52561104e-01 7.85443842e-01
5.82438529e-01 -2.93975770e-01 7.16075480e-01 2.97158986e-01
-8.71505439e-01 -1.55294314e-01 -1.19901292e-01 7.22567514e-02
5.59672952e-01 2.73498118e-01 -1.24296749e+00 4.06484246e-01
5.57639301e-01 1.14711940e+00 -5.15922904e-01 4.06527877e-01
-5.76799572e-01 7.25755394e-01 2.76228897e-02 -4.91387621e-02
3.60477537e-01 6.91707358e-02 4.80205297e-01 1.32358587e+00
-2.80434698e-01 2.53200591e-01 1.39384300e-01 8.96177828e-01
-5.81875741e-01 1.58366039e-01 -8.11844826e-01 -2.17973098e-01
-2.10749477e-01 1.10269701e+00 -8.27203631e-01 -6.75947845e-01
-5.75880051e-01 1.27034712e+00 4.06147867e-01 4.23122555e-01
-9.43482399e-01 1.05110526e-01 1.93778515e-01 -6.29637614e-02
6.64268196e-01 -1.46920040e-01 1.52372748e-01 -1.47445548e+00
-8.85952264e-02 -6.66991949e-01 3.43825042e-01 -1.32773888e+00
-1.23143148e+00 6.13008797e-01 2.25451082e-01 -9.87877548e-01
-6.25558972e-01 -6.59349680e-01 -3.30746859e-01 2.88044810e-01
-1.26370132e+00 -1.85350478e+00 3.31116378e-01 5.18899560e-01
7.66487181e-01 -1.81847990e-01 1.10183692e+00 1.75716281e-01
-3.82818192e-01 -3.32244113e-02 -5.85174799e-01 4.67980713e-01
5.33302248e-01 -1.24509180e+00 3.44022125e-01 8.01362514e-01
6.03127539e-01 4.86913085e-01 7.33841956e-01 -8.49309444e-01
-8.83383751e-01 -1.20046949e+00 1.67751586e+00 -7.86302388e-01
4.78419244e-01 -4.34723794e-01 -5.19690692e-01 1.38634276e+00
5.52849829e-01 2.59435028e-01 6.13584936e-01 2.90829390e-01
-4.24342722e-01 4.68622625e-01 -9.99103129e-01 6.41021848e-01
1.38226485e+00 -7.42469728e-01 -9.35446441e-01 7.58453250e-01
1.01125550e+00 -2.95682311e-01 -7.69069314e-01 5.41322291e-01
2.98908412e-01 -5.05550146e-01 1.05526614e+00 -1.24532139e+00
8.12262714e-01 -4.07943428e-02 -3.69109631e-01 -7.47828722e-01
-1.37196288e-01 -4.67672348e-01 -2.54098684e-01 1.02492118e+00
5.92630923e-01 1.00846887e-01 9.19055939e-01 8.98558557e-01
-1.86156288e-01 -6.86139166e-01 -7.94827402e-01 -3.14927250e-01
-2.30345055e-01 -7.27264822e-01 3.37624013e-01 8.32036257e-01
-5.62760681e-02 6.73477709e-01 -6.64062440e-01 2.24008085e-03
1.50207311e-01 -5.79758547e-02 3.90396535e-01 -1.05638945e+00
-2.24307001e-01 -3.17293853e-02 -4.30370539e-01 -1.31589472e+00
1.22698534e+00 -9.81525004e-01 2.17090264e-01 -1.94196784e+00
4.33959663e-01 2.70111710e-01 -3.13046992e-01 1.00753844e+00
7.43364245e-02 3.84290397e-01 4.03261364e-01 1.52049348e-01
-1.63818169e+00 5.52631736e-01 1.11606610e+00 -6.18453138e-02
-7.22038299e-02 5.73576940e-03 -5.76047301e-01 9.11277115e-01
4.51869696e-01 -4.01335210e-01 -1.03730805e-01 -4.91227359e-01
3.81256789e-01 3.64251167e-01 5.91799617e-01 -7.70674050e-01
2.17365384e-01 -1.70061305e-01 4.37662691e-01 -7.30551541e-01
4.06399280e-01 -8.03517461e-01 -3.00106406e-01 -2.95474147e-03
-9.61886823e-01 -1.85741112e-01 -1.53377876e-02 7.40593255e-01
-3.06487888e-01 -3.39302689e-01 2.20051214e-01 -4.49367464e-01
-1.02774930e+00 2.16047451e-01 -7.67990172e-01 -3.10137212e-01
1.07798755e+00 3.41503359e-02 -7.07747741e-03 -8.29278111e-01
-1.26503837e+00 1.42893791e-01 -6.28540060e-03 3.85553628e-01
6.61518097e-01 -1.10174036e+00 -5.95959365e-01 -3.80729407e-01
2.82720067e-02 -1.03508286e-01 2.49980226e-01 6.72228634e-01
4.25566733e-02 7.34348774e-01 -1.63777098e-01 -4.50674057e-01
-1.22272348e+00 4.94544744e-01 1.88398957e-01 -6.70664072e-01
-5.77163577e-01 7.40676761e-01 3.55574459e-01 -6.57696724e-01
2.19868213e-01 -1.61962107e-01 -6.06673598e-01 -1.34804726e-01
3.99123788e-01 -2.31170848e-01 -2.89867610e-01 -1.04429555e+00
-2.52978563e-01 6.31784499e-02 -3.31511110e-01 -2.47058496e-01
1.72332847e+00 -1.80897355e-01 -2.61270791e-01 1.23109572e-01
1.20154107e+00 -4.70244765e-01 -1.12022555e+00 -3.51658881e-01
3.44245046e-01 1.38285458e-01 -1.17035016e-01 -1.02424073e+00
-8.68250966e-01 6.49039090e-01 2.37645924e-01 -1.66833445e-01
1.09209859e+00 6.15120828e-01 8.36151004e-01 8.11065257e-01
1.84170380e-01 -1.15531337e+00 2.58877516e-01 8.96513283e-01
8.69687557e-01 -1.25745595e+00 -2.85975814e-01 -5.54142356e-01
-1.17579806e+00 1.02363157e+00 7.23844290e-01 -9.51278061e-02
5.34765720e-01 -5.41566163e-02 -2.62344420e-01 -2.86586136e-01
-1.03088629e+00 -5.03539324e-01 2.87718087e-01 5.97377777e-01
5.88262975e-01 -1.35716304e-01 -4.01512533e-02 7.00132787e-01
1.88972890e-01 3.61699760e-01 5.08875132e-01 8.29035759e-01
-6.44966867e-03 -1.39225507e+00 7.34305605e-02 5.84100962e-01
-7.72091925e-01 -3.47467661e-01 -2.66106755e-01 5.07490218e-01
2.82894105e-01 7.48668730e-01 3.06555778e-01 3.84496041e-02
2.60822028e-01 4.25179660e-01 4.98191029e-01 -9.90968525e-01
-6.75203264e-01 9.96830757e-04 5.23304462e-01 -1.06735480e+00
-1.19954324e+00 -9.22964811e-01 -1.33028936e+00 4.82172132e-01
5.86484186e-02 -1.58151940e-01 5.51495850e-01 1.38483441e+00
3.90932173e-01 3.93751919e-01 1.68016687e-01 -8.02419007e-01
-2.20312446e-01 -1.08216393e+00 -1.90554708e-01 7.73507416e-01
-1.37824073e-01 -3.17562789e-01 7.79258367e-03 4.71854210e-01] | [10.358128547668457, 1.3294955492019653] |
7ea3240d-8c8f-4fc3-9360-10d5f1ee0e92 | a-case-study-on-profiling-of-an-eeg-based | 2110.02785 | null | https://arxiv.org/abs/2110.02785v1 | https://arxiv.org/pdf/2110.02785v1.pdf | A case study on profiling of an EEG-based brain decoding interface on Cloud and Edge servers | Brain-Computer Interfaces (BCIs) enable converting the brain electrical activity of an interface user to the user commands. BCI research studies demonstrated encouraging results in different areas such as neurorehabilitation, control of artificial limbs, control of computer environments, communication and detection of diseases. Most of BCIs use scalp-electroencephalography (EEG), which is a non-invasive method to capture the brain activity. Although EEG monitoring devices are available in the market, these devices are generally lab-oriented and expensive. Day-to-day use of BCIs is impractical at this time due to the complex techniques required for data preprocessing and signal analysis. This implies that BCI technologies should be improved to facilitate its widespread adoption in Cloud and Edge datacenters. This paper presents a case study on profiling the accuracy and performance of a brain-computer interface which runs on typical Cloud and Edge servers. In particular, we investigate how the accuracy and execution time of the preprocessing phase, i.e. the brain signal filtering phase, of a brain-computer interface varies when processing static and live streaming data obtained in real time BCI devices. We identify the optimal size of the packets for sampling brain signals which provides the best trade-off between the accuracy and performance. Finally, we discuss the pros and cons of using typical Cloud and Edge servers to perform the BCI filtering phase. | ['Lev Mukhanov', 'Georgios Karakonstantis', 'Barry J. Devereux', 'Alexandra Samsonova'] | 2021-10-04 | null | null | null | null | ['brain-decoding', 'brain-decoding'] | ['medical', 'miscellaneous'] | [ 1.81437805e-01 -5.92155933e-01 3.79290164e-01 -1.82759628e-01
3.53062041e-02 -4.92487758e-01 1.14345081e-01 -2.35287733e-02
-4.92165625e-01 7.65640855e-01 -2.17886195e-01 -4.46868241e-01
-2.62516886e-01 -5.97609520e-01 -3.27765614e-01 -6.69894040e-01
-1.95525214e-01 2.44751886e-01 1.39640421e-01 1.63865566e-01
2.45694429e-01 7.75895953e-01 -1.72062206e+00 2.31792599e-01
7.87762463e-01 1.40051675e+00 1.86358064e-01 7.38408148e-01
-3.35052311e-02 6.81836680e-02 -1.13185668e+00 1.06150441e-01
6.55857027e-02 -4.44048703e-01 -1.81554362e-01 -3.32908481e-01
-3.21393162e-01 -3.41029674e-01 -1.58585533e-01 1.08733368e+00
8.68105292e-01 -4.22931015e-01 3.83256733e-01 -1.62309992e+00
2.11797580e-01 1.03388697e-01 -5.09306073e-01 9.10917222e-01
4.65516299e-01 -4.87003196e-03 -9.14910249e-03 -4.00081903e-01
2.42012680e-01 4.02766377e-01 4.50195819e-01 3.19553465e-01
-1.04462099e+00 -1.25958121e+00 -2.35093206e-01 8.38311613e-01
-1.57722712e+00 -5.69263279e-01 6.31400764e-01 -2.90090740e-01
1.14637232e+00 4.31458056e-01 1.21834242e+00 9.33457255e-01
8.10148954e-01 2.51143008e-01 1.23180711e+00 -1.80554509e-01
5.63636661e-01 2.63995767e-01 3.73647153e-01 -2.31496319e-01
5.34438491e-01 -1.42612413e-01 -9.27256346e-01 -2.49418676e-01
8.03631723e-01 5.14135044e-03 -7.41054296e-01 2.86519468e-01
-1.03305292e+00 1.79601654e-01 -2.13580817e-01 6.19611382e-01
-9.37724948e-01 -1.16121292e-03 5.49919486e-01 6.40117705e-01
2.70967245e-01 1.29167244e-01 -3.58070761e-01 -7.62791812e-01
-1.12600148e+00 -5.89967053e-03 9.22783613e-01 1.06554925e+00
8.97799581e-02 -1.14505373e-01 -9.73727629e-02 4.02085990e-01
-5.40038012e-02 5.11624038e-01 7.05326259e-01 -3.50848049e-01
1.09870046e-01 2.44656891e-01 4.33709435e-02 -5.76289296e-01
-5.40409327e-01 -2.36642972e-01 -6.75822198e-01 1.33298144e-01
3.03863347e-01 -6.12948298e-01 -4.35346246e-01 1.11768353e+00
1.89572662e-01 5.84519267e-01 -4.54176039e-01 9.23271120e-01
4.15120542e-01 3.61761123e-01 -1.08524486e-01 -5.16675413e-01
1.49019873e+00 -1.67529136e-02 -9.72580254e-01 -7.22951517e-02
8.27583596e-02 -4.96261001e-01 8.82171214e-01 8.37130368e-01
-1.16739964e+00 -2.02743024e-01 -1.04602134e+00 6.27312422e-01
-3.12464207e-01 -2.08414003e-01 4.89957124e-01 1.07291591e+00
-1.20513511e+00 2.81087816e-01 -1.06628764e+00 -3.94433916e-01
2.21644297e-01 7.35281467e-01 -3.16628397e-01 3.58679116e-01
-8.18993628e-01 8.95499110e-01 -1.10635169e-01 3.19155976e-02
-4.86812651e-01 -8.50065470e-01 -5.81016280e-02 3.10051858e-01
-8.40205029e-02 -2.06415311e-01 9.35367942e-01 -1.13891053e+00
-1.76017106e+00 5.79956174e-01 -1.37785807e-01 -4.12684679e-01
3.34440261e-01 6.10877201e-02 -6.58794045e-01 2.16280714e-01
-3.57718617e-01 2.45062802e-02 9.07546282e-01 -4.16867018e-01
-5.87712228e-01 -6.89269364e-01 -2.96579570e-01 -1.46603659e-01
-3.15685898e-01 5.34303546e-01 -1.01096429e-01 -1.37009904e-01
-2.25654934e-02 -7.88957953e-01 5.14282882e-01 -4.73050773e-01
6.92722052e-02 8.02196737e-05 8.43867123e-01 -6.39147401e-01
1.09354568e+00 -2.12735248e+00 -2.14302793e-01 4.62509990e-01
1.28840968e-01 2.01060787e-01 2.19596967e-01 2.90715665e-01
-3.05118412e-01 -2.61163443e-01 3.95014137e-01 3.28581333e-01
-3.22489560e-01 -1.41740993e-01 3.72735225e-02 7.49773264e-01
-2.02062994e-01 4.83884901e-01 -3.30467701e-01 -8.42570886e-02
1.74589053e-01 6.18115008e-01 -3.66415590e-01 3.81265432e-01
6.18095636e-01 8.25697362e-01 -1.40511006e-01 5.08117199e-01
5.81901729e-01 8.40933844e-02 1.22885639e-02 -8.80733803e-02
-2.18283266e-01 5.45339763e-01 -1.10079944e+00 1.42011774e+00
-4.85594273e-01 9.54719424e-01 5.64323723e-01 -1.17581820e+00
7.23685026e-01 9.03129458e-01 7.35828280e-01 -7.96026111e-01
5.90947688e-01 1.87786013e-01 5.09518206e-01 -7.17569709e-01
-2.11293802e-01 1.31854475e-01 6.26267970e-01 6.07010365e-01
-9.96985063e-02 -8.94588232e-02 1.17601939e-01 -5.61098792e-02
1.33237612e+00 -2.10286215e-01 1.84870854e-01 -6.74284220e-01
3.53052616e-01 -2.59957314e-01 2.95101076e-01 2.43557289e-01
-3.65169317e-01 1.17503621e-01 3.78143966e-01 -7.80777261e-02
-5.55196285e-01 -7.09727526e-01 -3.97238225e-01 8.03280711e-01
3.67557481e-02 -2.63948202e-01 -1.23824859e+00 7.93428123e-02
-1.95975080e-01 6.92909420e-01 -8.21038112e-02 -3.00462902e-01
-2.24331945e-01 -7.55596101e-01 1.31136179e-01 4.38007027e-01
4.46365267e-01 -9.91498053e-01 -1.54595065e+00 4.96179014e-01
-9.53966528e-02 -9.37291563e-01 -1.36562020e-01 3.51803601e-01
-9.36995387e-01 -9.30094898e-01 -4.51993436e-01 -2.51668841e-01
6.11755967e-01 3.96020055e-01 6.49582922e-01 1.12527221e-01
-5.11902273e-01 5.92320621e-01 -2.96333283e-01 -1.10887814e+00
1.23462446e-01 -1.70481384e-01 3.00147533e-01 1.99442565e-01
8.66073966e-01 -1.18365407e+00 -7.82290220e-01 2.82090932e-01
-7.10917115e-01 -2.87384298e-02 2.84901083e-01 2.90542156e-01
2.97297895e-01 2.65134305e-01 7.13691354e-01 -2.72320569e-01
1.23950171e+00 -6.09389305e-01 -6.75287008e-01 1.19624458e-01
-5.85838258e-01 -4.63858664e-01 5.73216975e-01 -7.66187787e-01
-6.86250687e-01 -2.45340422e-01 1.51974380e-01 -2.11165637e-01
-3.55123520e-01 3.65548968e-01 -2.72431582e-01 -1.66499227e-01
4.89673644e-01 1.61178499e-01 1.91765314e-03 -2.63526022e-01
-5.13246715e-01 1.40147793e+00 5.66430390e-01 -2.06743255e-01
8.40780605e-03 3.25067133e-01 -4.19514358e-01 -1.13715374e+00
2.15538114e-01 -4.07163173e-01 -4.11544293e-01 -5.46511114e-01
5.98038495e-01 -6.44269943e-01 -1.10983777e+00 5.80257356e-01
-1.23498130e+00 -3.02635819e-01 2.43406281e-01 7.87923396e-01
-4.52743441e-01 -1.96489885e-01 -4.69457895e-01 -1.02256906e+00
-8.48016977e-01 -1.11582613e+00 6.88801467e-01 4.46664929e-01
-5.10561585e-01 -3.50612909e-01 -2.45732933e-01 7.13253692e-02
8.69445682e-01 -3.28988805e-02 6.01159155e-01 -4.15039659e-01
-2.59046167e-01 -4.41292346e-01 -1.12782858e-01 6.56508952e-02
1.18331112e-01 -2.44835421e-01 -1.03018022e+00 -4.60157096e-01
6.11980319e-01 3.21255118e-01 -2.47599140e-01 7.79668510e-01
1.26070654e+00 1.31274015e-01 -3.83821309e-01 5.80178797e-01
1.22150791e+00 7.17714608e-01 8.54127765e-01 1.40889376e-01
-6.43008277e-02 4.57760900e-01 2.12681457e-01 6.43174052e-01
-9.45750922e-02 4.85616088e-01 1.56617999e-01 3.43115628e-01
1.81199521e-01 4.60925251e-01 3.65593642e-01 8.34875584e-01
-4.01697427e-01 5.59640825e-02 -9.66419876e-01 3.52631360e-01
-1.30870092e+00 -7.24455357e-01 -8.96863714e-02 2.64056849e+00
7.31746316e-01 1.83069170e-01 2.34483138e-01 5.48857987e-01
6.18800640e-01 -8.04537654e-01 -7.12953627e-01 -4.29739624e-01
4.64556307e-01 7.15464413e-01 6.95225120e-01 -1.73092648e-01
-3.29311728e-01 3.16192508e-01 6.26084089e+00 2.63369888e-01
-1.74125600e+00 6.16514146e-01 1.98831916e-01 -7.69321322e-01
2.99978435e-01 -5.41129768e-01 -4.80514914e-01 1.07863033e+00
1.73126364e+00 -5.23797870e-01 8.78637075e-01 5.88389337e-01
7.16636777e-01 -5.90919793e-01 -1.22563970e+00 1.76350427e+00
-2.23978639e-01 -9.52262223e-01 -6.68066144e-01 2.01474056e-01
-6.23303503e-02 1.74745023e-01 -4.81172651e-01 -1.75985590e-01
-5.87605000e-01 -9.58485305e-01 8.11042309e-01 5.62403440e-01
1.06859601e+00 -8.87902737e-01 7.99730659e-01 4.16093022e-01
-9.73303199e-01 -7.70294443e-02 -1.84007034e-01 -4.05329108e-01
9.21523571e-02 6.32310987e-01 -3.62546295e-01 -1.18516229e-01
1.12737441e+00 9.17524099e-02 -1.16353944e-01 1.33240104e+00
3.75945903e-02 7.79595494e-01 -5.80074131e-01 -3.41299117e-01
-3.99996489e-01 -4.68275160e-01 3.15815389e-01 1.01780069e+00
5.11398375e-01 5.71166992e-01 -4.76761788e-01 7.58756936e-01
2.28380352e-01 -1.30870854e-02 -5.65878034e-01 7.97878802e-02
6.83499634e-01 1.21407497e+00 -9.07539070e-01 -2.27897510e-01
-4.89942640e-01 9.64834034e-01 -1.71698451e-01 4.31327522e-01
-9.55056429e-01 -8.12319636e-01 9.59763288e-01 4.14901882e-01
-2.25333348e-01 -4.50808406e-01 -6.65021181e-01 -1.00279021e+00
2.91115612e-01 -8.79235685e-01 -1.32855713e-01 -8.05609882e-01
-6.93687975e-01 7.14222729e-01 1.60967156e-01 -1.00004649e+00
-1.43272042e-01 -5.29692888e-01 -8.90617728e-01 1.00629628e+00
-1.15098286e+00 -1.37149021e-01 -7.45219111e-01 1.09682584e+00
3.72056276e-01 4.00656350e-02 8.49135816e-01 6.00149751e-01
-6.07335687e-01 3.73174220e-01 -7.52383918e-02 -2.21169889e-01
5.53608716e-01 -6.14747584e-01 2.20490135e-02 8.35240901e-01
-2.16210976e-01 5.18168569e-01 7.63047934e-01 -5.19098580e-01
-1.96862578e+00 -2.62424797e-01 3.91306847e-01 9.90795940e-02
5.61742365e-01 -6.05649412e-01 -7.80731082e-01 2.58646190e-01
4.20837224e-01 -2.49338940e-01 7.41901815e-01 -3.54995131e-01
4.30618167e-01 -4.78263050e-01 -1.29876447e+00 4.96041745e-01
8.46247256e-01 -5.15029073e-01 -3.08119506e-01 3.09673995e-01
-9.09575224e-02 -9.31593701e-02 -8.93300593e-01 -2.20503762e-01
6.60539806e-01 -9.18168128e-01 4.62239176e-01 -2.34270066e-01
-2.68430918e-01 6.64732829e-02 3.50412458e-01 -1.51047552e+00
-1.30694047e-01 -9.36728001e-01 9.46767777e-02 9.20789361e-01
1.17460929e-01 -1.18796790e+00 7.02355206e-01 1.20180023e+00
2.43279219e-01 -4.42986339e-01 -1.15039682e+00 -7.20946729e-01
-5.10120869e-01 -7.30845511e-01 7.78409421e-01 4.39787865e-01
8.01433206e-01 1.81140572e-01 1.25551865e-01 4.50193807e-02
3.29781026e-01 -2.23285452e-01 4.78830338e-01 -1.42063606e+00
-6.69315457e-02 -3.95126015e-01 -8.31303895e-01 -4.82629389e-01
-1.69209167e-01 -5.81083536e-01 -2.01541945e-01 -1.24146879e+00
-1.04078710e-01 -1.21120565e-01 -3.39778066e-01 2.03061700e-01
1.76583573e-01 -7.03015402e-02 8.30405112e-03 6.79116100e-02
2.05258176e-01 -6.17416613e-02 5.37059247e-01 7.72042647e-02
-4.83001500e-01 3.20341028e-02 -3.06271851e-01 4.10748571e-01
1.00480902e+00 -5.06868780e-01 -6.11638844e-01 -4.53936696e-01
-2.43633520e-02 2.03420669e-01 6.19828962e-02 -1.37890041e+00
7.29243755e-01 8.48740563e-02 4.09854650e-01 -2.81075656e-01
1.81830958e-01 -1.26873732e+00 4.52633649e-01 5.08865714e-01
2.68906742e-01 3.07250053e-01 2.38770172e-01 3.19484353e-01
-2.89314277e-02 -9.33854952e-02 7.78684318e-01 2.11709276e-01
-1.47619337e-01 1.96106672e-01 -1.03221083e+00 -1.81981325e-01
1.33143306e+00 -6.17136776e-01 -1.41244143e-01 -3.88052255e-01
-4.78978455e-01 -1.55479431e-01 1.51505515e-01 2.31513128e-01
5.36916018e-01 -8.29209089e-01 -5.76681077e-01 8.69131267e-01
-2.78749228e-01 -4.91879791e-01 -4.37762327e-02 1.58559239e+00
-5.86465478e-01 3.43114525e-01 -8.39611590e-01 -6.07095659e-01
-1.46468484e+00 2.82458484e-01 4.08980161e-01 3.62417459e-01
-8.57124448e-01 6.61128342e-01 -2.57356077e-01 7.20918775e-01
5.99958301e-01 -4.57082272e-01 -9.98721868e-02 8.31154883e-02
1.05548751e+00 5.65272272e-01 7.27001190e-01 -3.45163122e-02
-6.80684865e-01 4.13228385e-02 2.91955829e-01 -3.91423911e-01
1.36686647e+00 -6.30537495e-02 -4.08318698e-01 3.30862999e-01
8.65723848e-01 -3.80332410e-01 -8.27741683e-01 3.64007682e-01
1.64709669e-02 -4.53056842e-01 6.07810915e-01 -6.61190331e-01
-1.35515296e+00 9.30600286e-01 9.54581022e-01 3.66453141e-01
1.67982697e+00 -5.84451556e-01 8.51033270e-01 1.55513033e-01
1.13593495e+00 -1.38504016e+00 -7.05429196e-01 -5.47986478e-02
8.49631608e-01 -4.92472410e-01 -2.36757189e-01 -6.55577779e-02
-3.16321522e-01 1.19845021e+00 2.33206272e-01 -1.66960046e-01
1.18256032e+00 1.05722260e+00 -1.75419882e-01 -2.16597557e-01
-7.72754073e-01 2.37144664e-01 -1.07765555e-01 6.91361189e-01
5.20995617e-01 3.23781580e-01 -7.28839815e-01 8.48140776e-01
-3.10458630e-01 5.28767645e-01 6.01692915e-01 1.04875040e+00
-2.38583058e-01 -6.40685022e-01 -7.73753047e-01 1.02805138e+00
-6.93579376e-01 8.43659230e-03 -2.36643791e-01 3.15452278e-01
-2.23078262e-02 1.43239748e+00 2.24470645e-01 -2.73009330e-01
5.28968751e-01 4.28189546e-01 5.69146335e-01 -3.79531324e-01
-8.39224279e-01 8.04244056e-02 -1.02706775e-01 -9.55650151e-01
-1.66118518e-01 -5.41162312e-01 -1.20799625e+00 -4.72699612e-01
-3.45267624e-01 6.27231076e-02 1.50165212e+00 7.64459848e-01
7.67692685e-01 7.01557755e-01 2.47530445e-01 -9.92279589e-01
-2.12435365e-01 -1.22389913e+00 -1.07840073e+00 9.09993500e-02
-1.19758196e-01 -6.87097073e-01 -2.52995223e-01 2.77375220e-03] | [13.200429916381836, 3.3742973804473877] |
1921e259-d3fb-4309-baea-61fe97ea70a4 | sinc-spatial-composition-of-3d-human-motions | 2304.10417 | null | https://arxiv.org/abs/2304.10417v1 | https://arxiv.org/pdf/2304.10417v1.pdf | SINC: Spatial Composition of 3D Human Motions for Simultaneous Action Generation | Our goal is to synthesize 3D human motions given textual inputs describing simultaneous actions, for example 'waving hand' while 'walking' at the same time. We refer to generating such simultaneous movements as performing 'spatial compositions'. In contrast to temporal compositions that seek to transition from one action to another, spatial compositing requires understanding which body parts are involved in which action, to be able to move them simultaneously. Motivated by the observation that the correspondence between actions and body parts is encoded in powerful language models, we extract this knowledge by prompting GPT-3 with text such as "what are the body parts involved in the action <action name>?", while also providing the parts list and few-shot examples. Given this action-part mapping, we combine body parts from two motions together and establish the first automated method to spatially compose two actions. However, training data with compositional actions is always limited by the combinatorics. Hence, we further create synthetic data with this approach, and use it to train a new state-of-the-art text-to-motion generation model, called SINC ("SImultaneous actioN Compositions for 3D human motions"). In our experiments, we find training on additional synthetic GPT-guided compositional motions improves text-to-motion generation. | ['Gül Varol', 'Michael J. Black', 'Mathis Petrovich', 'Nikos Athanasiou'] | 2023-04-20 | null | null | null | null | ['action-generation'] | ['computer-vision'] | [ 4.30994362e-01 3.45376551e-01 -1.74875900e-01 1.47918269e-01
-4.73327100e-01 -7.10109830e-01 1.13084674e+00 -3.48337084e-01
6.48832843e-02 8.13786864e-01 8.08444083e-01 -2.05424994e-01
1.34428423e-02 -9.95718479e-01 -7.70858705e-01 -6.02087796e-01
8.35178792e-02 7.57456958e-01 3.31217557e-01 -5.71515381e-01
5.63375689e-02 4.76202846e-01 -1.58081520e+00 6.26177311e-01
5.42474389e-01 3.38243544e-01 2.59464711e-01 8.21601450e-01
-5.64294040e-01 1.11931324e+00 -5.55829287e-01 -3.22079301e-01
2.37277135e-01 -1.27166915e+00 -1.13187301e+00 4.92367417e-01
1.74497858e-01 -2.68668652e-01 -2.60520965e-01 5.25152743e-01
2.20683366e-01 4.45964038e-01 8.35361421e-01 -1.35545564e+00
-7.21194923e-01 8.74508798e-01 -2.50635892e-01 -3.84397894e-01
9.55386519e-01 6.40416086e-01 1.07682407e+00 -4.98278588e-01
1.27597964e+00 1.35717690e+00 3.60266119e-01 1.15837646e+00
-1.38728642e+00 -5.40084355e-02 1.39662445e-01 -9.80055556e-02
-1.04626513e+00 -4.51659679e-01 6.11664236e-01 -5.44339478e-01
1.19571602e+00 4.21421111e-01 1.07757664e+00 1.33784556e+00
-3.13827060e-02 1.11311269e+00 5.65693736e-01 -5.47516882e-01
1.91138178e-01 -5.22719920e-01 -5.11570215e-01 6.72025323e-01
-1.73340023e-01 7.76838288e-02 -7.72710502e-01 1.19347140e-01
1.09078825e+00 -2.24391222e-01 -9.37124789e-02 -3.93245101e-01
-1.78950751e+00 4.99207109e-01 1.57008320e-01 4.57642406e-01
-4.26293701e-01 7.41581798e-01 2.78372377e-01 1.82240501e-01
1.96709767e-01 6.97049081e-01 -3.47580373e-01 -2.85919249e-01
-7.38733947e-01 1.03142083e+00 8.59271467e-01 1.27930355e+00
5.89251161e-01 1.69510506e-02 -5.59421182e-01 3.69605184e-01
-1.33914784e-01 4.01866823e-01 3.88900608e-01 -1.38510537e+00
5.34547508e-01 6.25858188e-01 4.23546046e-01 -5.01407087e-01
-1.84912935e-01 3.07409335e-02 -7.31296957e-01 1.40552193e-01
6.02735639e-01 -2.03867763e-01 -9.26310778e-01 1.87902856e+00
3.88349503e-01 -6.86648712e-02 2.19078466e-01 8.76221180e-01
5.18248856e-01 7.39185333e-01 2.07164377e-01 -1.01288758e-01
1.37223482e+00 -9.58151400e-01 -7.24177957e-01 -9.22492817e-02
1.00074553e+00 -4.97531086e-01 1.11769176e+00 9.72464085e-02
-1.31315172e+00 -5.87047100e-01 -5.99193871e-01 -2.10056990e-01
-1.76732048e-01 6.31293207e-02 6.95400119e-01 1.99819908e-01
-8.09543073e-01 9.68662143e-01 -8.47552955e-01 -4.54929352e-01
1.42664284e-01 -6.73864260e-02 -4.44481969e-01 1.11985259e-01
-1.09444344e+00 9.57749426e-01 4.22482401e-01 -3.31692934e-01
-9.54739809e-01 -5.18925905e-01 -9.64350879e-01 -2.16278493e-01
6.41606212e-01 -1.32860577e+00 1.58124292e+00 -9.41056550e-01
-1.65003967e+00 7.73178518e-01 -2.38421917e-01 -4.56699789e-01
7.29931712e-01 -2.64755577e-01 -1.74329534e-01 1.36598185e-01
4.77248490e-01 1.29947984e+00 6.09696507e-01 -1.09390664e+00
-6.98688328e-01 -2.06423309e-02 3.46170276e-01 4.32048708e-01
4.20083344e-01 2.30619609e-02 -5.54881632e-01 -8.19459975e-01
8.17658678e-02 -1.04612982e+00 -2.79993623e-01 2.19119608e-01
-7.54930913e-01 -1.87302291e-01 6.28914237e-01 -4.94394034e-01
1.09248662e+00 -1.91025889e+00 8.81594121e-01 -3.80807184e-02
-1.50630325e-01 -2.49859422e-01 -2.86880791e-01 9.09041524e-01
-7.77193010e-02 2.46432871e-01 -5.24083138e-01 -4.84488547e-01
3.15244377e-01 5.47497630e-01 -2.49782398e-01 -4.00259085e-02
3.62016737e-01 1.24412739e+00 -1.13497078e+00 -5.58403552e-01
1.54721677e-01 1.90507367e-01 -6.75391614e-01 2.49456227e-01
-9.00680363e-01 9.40976560e-01 -4.71277326e-01 4.04061586e-01
-4.26805355e-02 -7.72427022e-02 1.80158064e-01 -1.97908152e-02
-8.44534487e-02 4.45064187e-01 -1.19804955e+00 2.08202481e+00
-2.55724162e-01 2.12625340e-01 -4.19278949e-01 -6.16061211e-01
5.27056277e-01 5.84576726e-01 5.81932247e-01 -4.95859534e-01
-9.38070863e-02 2.47490212e-01 -1.24106938e-02 -7.12538779e-01
3.99762541e-01 -4.97915179e-01 -4.25671816e-01 6.61129534e-01
-5.09245507e-02 -7.34267294e-01 6.53057694e-01 2.87763089e-01
1.16802585e+00 1.12915444e+00 5.24155200e-01 3.33312899e-03
3.03290159e-01 5.08059800e-01 2.54201651e-01 4.76855725e-01
2.77399778e-01 6.95045471e-01 6.14415169e-01 -4.67437595e-01
-1.23741591e+00 -1.08355260e+00 7.61480510e-01 9.20528889e-01
-1.86087061e-02 -6.83200061e-01 -8.73582959e-01 -5.60917497e-01
-2.56605476e-01 1.03224468e+00 -7.69716144e-01 1.15991428e-01
-1.11637545e+00 2.16924455e-02 6.68458283e-01 7.29666948e-01
3.98044914e-01 -1.44641185e+00 -1.19304955e+00 4.17603731e-01
-5.18559933e-01 -9.27162111e-01 -8.51882279e-01 -1.11075593e-02
-8.03826272e-01 -1.00002003e+00 -1.03071868e+00 -6.23083413e-01
5.09275913e-01 -5.27604334e-02 1.25942826e+00 1.58849165e-01
-2.71429151e-01 4.02023435e-01 -5.64623356e-01 -1.74626976e-01
-7.82505393e-01 -3.06333713e-02 -1.18576378e-01 -2.19436750e-01
-2.74326175e-01 -7.74438560e-01 -4.37602103e-01 1.06982067e-01
-1.13672948e+00 9.32611763e-01 5.49314916e-01 5.46035647e-01
4.93657321e-01 -1.57837406e-01 7.36386608e-03 -5.50071716e-01
5.23956776e-01 -1.82216287e-01 -6.44116998e-02 2.84470469e-01
4.62175757e-01 5.79503179e-01 6.62978411e-01 -7.90087938e-01
-1.03172767e+00 3.51856530e-01 -6.64686933e-02 -3.85308325e-01
-4.11617607e-01 3.81704003e-01 -3.45546246e-01 6.75421536e-01
8.10869813e-01 4.66060907e-01 -1.54821081e-02 -2.83867717e-01
9.69464183e-01 -2.44785607e-01 8.54456067e-01 -1.08210409e+00
8.59321952e-01 4.96124268e-01 3.55580777e-01 -6.93193913e-01
-4.07799214e-01 -1.45499095e-01 -7.86972106e-01 -1.88737184e-01
1.24965787e+00 -6.14442170e-01 -7.16178894e-01 3.23900580e-01
-1.44438052e+00 -8.10514987e-01 -7.68522143e-01 9.43240598e-02
-1.18569446e+00 2.73798376e-01 -5.66319108e-01 -7.69152701e-01
-5.59315607e-02 -1.04056549e+00 1.39281189e+00 -2.68291235e-01
-1.00810027e+00 -7.49398232e-01 1.40136987e-01 -1.11903027e-01
3.82489040e-02 5.17211080e-01 1.14532197e+00 -2.11041033e-01
-7.05574870e-01 1.28079906e-01 3.59527349e-01 -1.36686280e-01
3.78043264e-01 7.63895130e-03 -1.97443336e-01 3.46368819e-01
-3.82280111e-01 -1.05309486e-01 3.13125819e-01 2.10281283e-01
8.37420821e-01 -6.26416683e-01 -3.71581048e-01 2.23549113e-01
1.09370792e+00 1.77208051e-01 8.46736670e-01 1.99037343e-02
8.22988272e-01 1.03126347e+00 4.95755732e-01 2.88269192e-01
2.16383934e-01 1.05146313e+00 1.09200567e-01 2.65760183e-01
-5.42641282e-01 -8.47426951e-01 4.55184758e-01 3.17676544e-01
-4.50345635e-01 -3.14047277e-01 -7.85586178e-01 6.33969009e-01
-1.97061360e+00 -1.37624550e+00 -1.51766613e-01 1.70623577e+00
1.05717480e+00 -6.11973889e-02 4.34750587e-01 1.93372548e-01
3.28754544e-01 1.57646090e-01 -1.47984028e-01 -5.96377067e-02
-6.19893223e-02 3.03417444e-01 1.70112208e-01 6.35860324e-01
-8.51238251e-01 1.29037046e+00 5.76489449e+00 8.76008570e-01
-7.71063387e-01 -1.93547443e-01 1.85847431e-01 -9.58215818e-02
-7.22343028e-01 2.41890699e-01 -5.43338537e-01 2.87587434e-01
4.70452487e-01 -6.73532858e-02 6.81070030e-01 3.73105019e-01
4.06397611e-01 -2.01887935e-01 -1.55548048e+00 7.33467400e-01
-7.15979636e-02 -1.61745405e+00 6.58898592e-01 -2.30345845e-01
6.96232557e-01 -8.62653494e-01 -3.64149243e-01 1.26835972e-01
6.15653872e-01 -1.08765459e+00 1.36463165e+00 5.70268750e-01
8.84123683e-01 -4.10886168e-01 -7.79443681e-02 8.02580833e-01
-1.50749218e+00 2.16157675e-01 2.57119864e-01 -1.98738247e-01
7.81027913e-01 3.89780253e-02 -5.76039970e-01 7.94068456e-01
1.47404104e-01 5.14178813e-01 -7.75424540e-02 5.94035447e-01
-7.56897509e-01 6.32923990e-02 -1.16293907e-01 -1.30805910e-01
2.74098992e-01 -3.10149968e-01 6.91583931e-01 9.73556697e-01
7.13167131e-01 4.52754974e-01 9.42932218e-02 1.24164236e+00
2.85817593e-01 -1.61011577e-01 -1.02132928e+00 -2.64202565e-01
1.58142075e-01 6.57952845e-01 -7.92186081e-01 -7.27071464e-01
-4.65949662e-02 1.43311131e+00 -3.95716690e-02 3.77752990e-01
-9.13660347e-01 -6.93513826e-02 7.15497255e-01 3.02297980e-01
2.31086090e-01 -5.76962411e-01 -2.41720140e-01 -9.71209288e-01
-1.00863121e-01 -1.00021470e+00 -3.30250487e-02 -1.21478713e+00
-7.86813259e-01 3.55537355e-01 5.00493824e-01 -1.34955692e+00
-6.79220438e-01 -3.68814051e-01 -6.57437384e-01 8.20240200e-01
-5.61076224e-01 -1.41365695e+00 2.61375704e-03 5.71370244e-01
9.48734641e-01 2.70448744e-01 9.07995999e-01 -8.92449319e-02
1.63936000e-02 4.99831289e-02 -8.27837825e-01 1.38689518e-01
2.81147242e-01 -1.14006758e+00 9.45804298e-01 8.62359107e-01
4.23222572e-01 6.26775742e-01 9.58591402e-01 -9.78225291e-01
-1.42625868e+00 -9.12505746e-01 1.21606851e+00 -6.70430064e-01
5.50145447e-01 -3.09890926e-01 -5.11737585e-01 8.43826115e-01
1.47316471e-01 -5.10101318e-01 3.17691267e-01 -4.81534570e-01
-1.20031305e-01 5.87046206e-01 -7.03535676e-01 1.44731748e+00
2.00527883e+00 -4.21839893e-01 -9.75165665e-01 2.44221091e-01
9.27401841e-01 -5.86046219e-01 -4.44506764e-01 3.28947544e-01
4.68179256e-01 -8.45908701e-01 1.07889950e+00 -1.07797134e+00
1.00942540e+00 -6.11519337e-01 -1.35996431e-01 -1.42301869e+00
-8.94954205e-02 -8.52475822e-01 -1.88110441e-01 7.95469761e-01
4.45005000e-01 -3.50612663e-02 8.61429751e-01 4.52585876e-01
-4.37681019e-01 -4.79078859e-01 -7.10843146e-01 -9.46017504e-01
7.60103390e-02 -6.31469011e-01 7.83521771e-01 7.43074536e-01
1.23641476e-01 2.65009880e-01 -6.48587644e-01 -2.66463190e-01
1.25837296e-01 1.64652020e-01 1.14779794e+00 -7.10804522e-01
-7.03020036e-01 -5.84837615e-01 3.14060897e-02 -1.46345317e+00
8.20910558e-02 -8.26442063e-01 3.29327196e-01 -1.81583118e+00
-1.39197269e-02 -1.43477783e-01 6.75968766e-01 7.30899692e-01
-1.06013112e-01 -1.92279033e-02 4.38090473e-01 2.43313193e-01
-2.98845768e-01 5.56060195e-01 1.84101665e+00 -2.12088779e-01
-4.05305594e-01 -1.97388783e-01 -4.40601438e-01 5.47136605e-01
4.80719566e-01 -2.39728808e-01 -5.19639790e-01 -5.06635308e-01
3.59737039e-01 6.15618765e-01 5.09531856e-01 -1.08980680e+00
1.29537344e-01 -6.64898813e-01 7.40042925e-02 -3.31316769e-01
4.66123194e-01 -5.74133575e-01 8.52364123e-01 7.07761765e-01
-5.19012868e-01 1.27371728e-01 1.20230049e-01 1.65699825e-01
-7.28121921e-02 -3.74792367e-02 2.83320159e-01 -6.98210537e-01
-8.30610275e-01 1.36819795e-01 -5.61426401e-01 -1.33232951e-01
1.00413156e+00 -4.66610938e-01 -1.02581441e-01 -6.05206013e-01
-1.05837023e+00 -4.72286120e-02 6.15729034e-01 4.11939263e-01
5.21747291e-01 -1.59807360e+00 -6.52099848e-01 -5.48026375e-02
1.25180259e-01 1.71239406e-01 2.72816151e-01 7.82566965e-01
-8.67211103e-01 2.74063021e-01 -3.51681560e-01 -3.37656140e-01
-1.11707580e+00 7.10088313e-01 3.13039273e-01 -2.28620633e-01
-7.86928833e-01 6.90572202e-01 2.73734927e-01 -3.63320261e-01
-3.29535194e-02 -5.58230698e-01 1.65265456e-01 -2.09912777e-01
2.99192488e-01 2.43040085e-01 -5.68222284e-01 -6.01678312e-01
-1.12114593e-01 7.28832781e-01 6.41639113e-01 -8.01795185e-01
8.74909818e-01 1.83863506e-01 -8.46095756e-02 5.15447676e-01
8.03830564e-01 1.44567406e-02 -1.20032525e+00 9.67666730e-02
-1.03498800e-02 -3.61138910e-01 -1.14123964e+00 -6.45702064e-01
-5.99362135e-01 8.05918992e-01 -2.36580819e-01 7.44664446e-02
8.93765509e-01 2.85308152e-01 7.35685766e-01 4.13291335e-01
5.98755419e-01 -8.09993744e-01 5.35935640e-01 7.23168075e-01
1.28310585e+00 -6.12803817e-01 -1.76820830e-01 -5.79102457e-01
-9.06355023e-01 9.49293196e-01 5.17428219e-01 -2.63037882e-03
-1.82736628e-02 2.00000599e-01 -2.33990237e-01 -1.97316825e-01
-8.36954892e-01 -4.89152044e-01 3.09715271e-01 7.26116717e-01
3.55878770e-01 2.54070848e-01 -4.45032001e-01 9.31253210e-02
-5.04251659e-01 1.26412854e-01 1.49787143e-01 1.13819456e+00
-2.38929525e-01 -1.62292671e+00 -2.71022469e-01 1.54423518e-02
-3.37330997e-02 1.23932380e-02 -7.40888000e-01 1.00887895e+00
4.18725610e-01 6.46052539e-01 4.54801247e-02 -3.59801888e-01
4.19073373e-01 3.55418682e-01 8.64597082e-01 -1.01025093e+00
-4.57785428e-01 1.17353506e-01 4.47368830e-01 -6.54478133e-01
-6.80113614e-01 -7.06259251e-01 -1.55660677e+00 -3.27396393e-01
4.63390678e-01 -5.94119020e-02 2.18755111e-01 1.15938747e+00
6.15688227e-02 7.36950636e-01 -1.83331762e-02 -1.13163674e+00
-3.48727077e-01 -8.62445414e-01 -2.55947143e-01 9.18302000e-01
5.20878546e-02 -4.59551722e-01 3.80648039e-02 6.05229378e-01] | [7.344239234924316, -0.15215519070625305] |
418cc1f0-a7c1-4d88-a716-972a43772d87 | macroeconomic-forecasting-and-sovereign-risk | 2301.09856 | null | https://arxiv.org/abs/2301.09856v1 | https://arxiv.org/pdf/2301.09856v1.pdf | Macroeconomic forecasting and sovereign risk assessment using deep learning techniques | In this study, we propose a novel approach of nowcasting and forecasting the macroeconomic status of a country using deep learning techniques. We focus particularly on the US economy but the methodology can be applied also to other economies. Specifically US economy has suffered a severe recession from 2008 to 2010 which practically breaks out conventional econometrics model attempts. Deep learning has the advantage that it models all macro variables simultaneously taking into account all interdependencies among them and detecting non-linear patterns which cannot be easily addressed under a univariate modelling framework. Our empirical results indicate that the deep learning methods have a superior out-of-sample performance when compared to traditional econometric techniques such as Bayesian Model Averaging (BMA). Therefore our results provide a concise view of a more robust method for assessing sovereign risk which is a crucial component in investment and monetary decisions. | ['Sotirios Chatzis', 'Loukas Papadoulas', 'Konstantinos P. Panousis', 'Vassilis Siakoulis', 'Anastasios Petropoulos'] | 2023-01-24 | null | null | null | null | ['econometrics'] | ['miscellaneous'] | [-7.25516677e-01 -2.02542379e-01 -1.30707979e-01 -2.73038924e-01
-6.73037231e-01 -3.09034556e-01 1.10740125e+00 -1.02750331e-01
-2.93844312e-01 9.20694411e-01 4.51632529e-01 -8.87971044e-01
-2.83507228e-01 -9.71405923e-01 -3.81533086e-01 -1.04614627e+00
-1.02973208e-01 3.98761451e-01 -4.90003943e-01 -5.14204055e-02
3.21507722e-01 3.98066550e-01 -7.86417902e-01 -2.45799720e-01
7.34531820e-01 7.70393610e-01 3.40961688e-03 4.85043466e-01
-7.17212260e-02 1.18078554e+00 -6.29387915e-01 -5.89354396e-01
1.53263018e-01 -8.09461847e-02 -3.97154242e-01 -2.39033550e-01
1.71608031e-01 -4.80958730e-01 -3.02913129e-01 9.88434374e-01
4.97725844e-01 1.61769360e-01 1.05395603e+00 -5.54112673e-01
-4.82031912e-01 6.78377271e-01 -6.74313545e-01 7.15209544e-01
8.37333649e-02 1.95408210e-01 9.77773607e-01 -7.68249810e-01
-1.98556166e-02 1.43677235e+00 8.54443789e-01 -2.51145750e-01
-1.15520966e+00 -7.10044742e-01 1.04235023e-01 1.68086827e-01
-6.62326932e-01 -4.25258785e-01 6.53572500e-01 -7.82893181e-01
1.10815454e+00 -1.90617606e-01 3.51431817e-01 9.90068555e-01
8.77195716e-01 2.58344740e-01 1.40387285e+00 -5.57363629e-01
2.13275790e-01 -2.68828094e-01 2.40935326e-01 2.10761338e-01
4.68178302e-01 4.25158113e-01 -5.65123484e-02 -8.46722201e-02
8.84783745e-01 4.04068492e-02 3.45453829e-01 2.39780396e-01
-1.08129203e+00 1.27986610e+00 9.54568163e-02 5.82857072e-01
-1.02345645e+00 2.58930415e-01 3.47942889e-01 4.66044247e-01
8.35895717e-01 1.67688921e-01 -8.44196260e-01 -2.54846960e-01
-1.07185519e+00 2.78907955e-01 8.27583253e-01 -1.37779787e-01
4.55520213e-01 7.61527896e-01 2.07230255e-01 6.25085592e-01
2.49690190e-01 8.03983688e-01 4.40477192e-01 -1.13784218e+00
4.64935154e-01 6.19015619e-02 2.46820509e-01 -1.29525352e+00
-5.81871331e-01 -5.26092887e-01 -1.22222900e+00 5.18304586e-01
7.80813992e-01 -7.80153930e-01 -5.99475324e-01 1.54084754e+00
-1.16371140e-02 3.13647151e-01 2.88390338e-01 2.75737464e-01
5.33088267e-01 8.42547655e-01 2.58625656e-01 -3.83391410e-01
1.38545871e+00 -5.84362805e-01 -8.54578793e-01 -6.49083927e-02
5.09488165e-01 -7.15605497e-01 5.01460850e-01 6.81300223e-01
-9.65130806e-01 -3.92730445e-01 -5.70082605e-01 3.17286700e-01
-6.11559629e-01 2.84919683e-02 1.02221787e+00 7.00754523e-01
-1.04790914e+00 6.49364531e-01 -1.03977847e+00 3.48749384e-02
1.39404267e-01 1.48807228e-01 -7.30785951e-02 4.30047870e-01
-1.25107396e+00 1.34330451e+00 2.76694685e-01 1.95889130e-01
-8.46030951e-01 -3.94227892e-01 -7.45595336e-01 4.32069987e-01
3.63085158e-02 -5.18186569e-01 1.48561704e+00 -9.42113638e-01
-1.63533998e+00 2.32577518e-01 -9.61322188e-02 -7.82727242e-01
2.77191222e-01 -1.81897715e-01 -4.62829858e-01 -2.83178926e-01
-2.55520195e-02 -3.21856707e-01 4.09710228e-01 -6.74778938e-01
-6.47904634e-01 -3.76754999e-01 1.79036379e-01 -3.00507098e-01
-7.68562481e-02 5.09270906e-01 4.94528681e-01 -1.20991051e+00
-1.38822585e-01 -4.99840111e-01 -2.58166909e-01 -1.10280943e+00
-2.68073343e-02 -4.89008844e-01 3.32502395e-01 -1.29547632e+00
1.45561862e+00 -1.40573847e+00 4.08428395e-03 1.67307220e-02
-2.06160977e-01 3.21578979e-01 3.67142171e-01 7.24578977e-01
-4.15394157e-01 1.18216695e-02 -2.93668002e-01 -4.91494536e-01
1.87903821e-01 8.09684768e-02 -4.29404944e-01 5.61975420e-01
3.84294242e-02 8.09467256e-01 -6.84372246e-01 -1.39907926e-01
4.48827684e-01 4.29860175e-01 -1.04783386e-01 -3.99060063e-02
1.60100058e-01 2.69697100e-01 -3.05490196e-01 6.92799032e-01
8.61098170e-01 1.23711661e-01 1.59907445e-01 3.79650563e-01
-4.87010479e-01 6.35434628e-01 -1.04049718e+00 1.10800433e+00
-7.78054535e-01 8.92127693e-01 -1.32686242e-01 -1.77394843e+00
8.96675706e-01 7.54020095e-01 1.29234239e-01 -7.84458578e-01
9.74006876e-02 1.62298650e-01 1.55391738e-01 -4.74904865e-01
1.26337633e-01 -6.04017973e-01 -4.55872901e-02 6.96031094e-01
5.11217304e-02 1.71890184e-01 1.26541406e-01 -2.11232170e-01
6.16639435e-01 -5.20912781e-02 5.33724964e-01 -6.90253794e-01
3.71145785e-01 -5.77726185e-01 7.99343467e-01 8.01396906e-01
2.04696581e-01 5.64891919e-02 9.84385252e-01 -7.58697629e-01
-8.07610810e-01 -8.56719792e-01 -3.65049481e-01 8.77434015e-01
-1.11822081e+00 2.16514334e-01 -5.49027622e-01 -5.28348923e-01
1.96368396e-01 8.81252885e-01 -9.44544435e-01 3.94805759e-01
-5.21660566e-01 -1.46378303e+00 3.17268431e-01 5.58481693e-01
4.82250184e-01 -1.15770900e+00 -7.07177699e-01 3.82239282e-01
8.87536481e-02 -5.53082764e-01 4.29702163e-01 3.30175161e-01
-1.09688962e+00 -9.96853292e-01 -9.13367271e-01 -4.73167360e-01
-7.40082711e-02 -2.55500197e-01 1.39603961e+00 -4.81869310e-01
4.74708796e-01 -1.16159871e-01 9.72893760e-02 -8.82560134e-01
-5.22820890e-01 2.83337012e-02 4.17714827e-02 -6.45189360e-02
7.44471312e-01 -6.96813881e-01 -4.57175702e-01 -3.72824222e-01
-4.79549021e-01 -3.67739648e-01 5.28388321e-01 8.79169464e-01
-1.53484643e-01 5.06952047e-01 1.27025032e+00 -6.14831984e-01
5.76682925e-01 -9.15245652e-01 -9.63495135e-01 2.09878698e-01
-8.17034960e-01 -2.12165520e-01 3.35460037e-01 -1.18477710e-01
-1.56485462e+00 -5.71991861e-01 -1.86664760e-02 1.84666395e-01
-2.20032454e-01 1.22814286e+00 1.72828183e-01 4.97215688e-01
-2.86869742e-02 -7.46989027e-02 -5.76910824e-02 -1.00102234e+00
-1.19989626e-01 5.28666973e-01 5.56580961e-01 -4.86037701e-01
6.16110563e-01 3.80611539e-01 7.54449442e-02 -5.56641698e-01
-4.15050209e-01 -3.44682902e-01 -7.25574136e-01 -9.69289318e-02
7.97438502e-01 -1.29485703e+00 -5.83680868e-01 1.05873835e+00
-1.23433578e+00 -3.00794125e-01 1.81522414e-01 9.72177625e-01
-2.50536770e-01 1.76409379e-01 -1.08440089e+00 -1.34129727e+00
-2.28761941e-01 -1.06555033e+00 5.90178609e-01 3.03643078e-01
-7.58088231e-02 -1.86548913e+00 6.68330908e-01 1.06430419e-01
5.98214090e-01 5.37325382e-01 9.51016843e-01 -6.25633061e-01
8.72421041e-02 -1.82614639e-01 -1.65916130e-01 4.25261140e-01
2.70468175e-01 3.27032775e-01 -1.20398939e+00 -1.50718570e-01
2.61399120e-01 -4.83917892e-02 1.16031969e+00 1.13378966e+00
5.22704422e-01 -4.41708416e-01 2.86203206e-01 4.77769226e-01
1.59776533e+00 5.40799558e-01 5.75376391e-01 8.88174832e-01
3.61505300e-01 7.60854602e-01 1.86032891e-01 7.52776742e-01
4.81681585e-01 3.65970075e-01 3.36274266e-01 -2.60151744e-01
4.05449420e-01 2.62250006e-01 4.07510370e-01 7.67431378e-01
-5.70519984e-01 -8.09284747e-02 -1.28406310e+00 6.57341063e-01
-1.81805348e+00 -1.30355811e+00 -3.56248558e-01 2.05820417e+00
6.72840118e-01 1.63055480e-01 3.32846910e-01 1.21734492e-01
3.76283169e-01 4.05771136e-01 -4.59356233e-04 -9.27660704e-01
-2.96176076e-01 6.45441830e-01 6.78763568e-01 6.28349662e-01
-1.24271154e+00 6.49360895e-01 7.25803995e+00 5.22715569e-01
-9.97130513e-01 7.85196275e-02 1.12748134e+00 2.89804548e-01
-2.57000178e-01 1.91413894e-01 -5.23627758e-01 5.73103845e-01
1.43156278e+00 -1.11012772e-01 5.03506102e-02 5.69530785e-01
8.01199198e-01 -5.13602972e-01 -5.84623575e-01 4.80496079e-01
-4.12152141e-01 -1.25637829e+00 -2.59985179e-01 6.46440744e-01
1.04834282e+00 1.84188530e-01 3.12052697e-01 4.24994826e-01
8.84207010e-01 -1.20738137e+00 5.17029524e-01 9.56893206e-01
2.13070393e-01 -1.24177909e+00 1.39536643e+00 2.12840095e-01
-6.31085813e-01 -4.61949050e-01 -5.47560155e-01 -9.06992555e-01
2.56481200e-01 1.04969037e+00 -3.13111395e-01 7.57691860e-01
8.42662990e-01 7.97226131e-01 -2.71612167e-01 7.05675840e-01
-7.15873316e-02 9.59071279e-01 -8.10038969e-02 5.43356776e-01
6.49566472e-01 -6.11625195e-01 1.19318433e-01 1.33911586e+00
6.05947554e-01 3.53152975e-02 -4.70846385e-01 5.10318160e-01
3.10294300e-01 -5.05113043e-02 -8.08028460e-01 1.78708211e-01
1.82350025e-01 6.66011393e-01 -4.63882804e-01 -6.37777209e-01
-1.01246619e+00 5.22790886e-02 1.28106996e-01 5.77369571e-01
-5.33085465e-01 -1.53398201e-01 6.33562386e-01 -2.95380831e-01
5.78017712e-01 -3.41652602e-01 -6.23212278e-01 -1.35726714e+00
-3.02447349e-01 -1.01079667e+00 4.93260562e-01 -3.51984918e-01
-1.17276227e+00 1.46363378e-02 1.69349253e-01 -6.16591454e-01
-7.85576105e-01 -6.56012654e-01 -1.26184905e+00 1.08805835e+00
-1.75747240e+00 -8.62923324e-01 4.49233174e-01 3.46470624e-01
6.26710951e-01 -6.04232967e-01 8.44571710e-01 4.63558249e-02
-9.33353722e-01 -5.95495701e-02 9.34988439e-01 2.32678175e-01
4.65996146e-01 -1.52937913e+00 5.58617294e-01 1.18448615e+00
1.18147442e-02 4.90209788e-01 7.10301459e-01 -5.78938484e-01
-6.87174618e-01 -3.97575021e-01 1.43198395e+00 -3.01271796e-01
1.08476889e+00 4.50234413e-02 -9.16759014e-01 9.69728947e-01
7.09796607e-01 -5.33645988e-01 6.01832807e-01 4.62391615e-01
-2.09811181e-01 -2.10157380e-01 -8.25356483e-01 2.45264731e-02
-1.60508707e-01 -4.98821735e-01 -9.69640195e-01 1.15754418e-01
7.86726996e-02 3.45445156e-01 -1.08636546e+00 3.60618234e-01
7.99484730e-01 -1.35079348e+00 8.96984696e-01 -7.32299685e-01
4.64032650e-01 4.06287879e-01 -1.20875925e-01 -1.51689672e+00
-5.69892764e-01 -5.76173067e-01 -3.96425128e-01 1.48543036e+00
6.42491877e-02 -1.05644739e+00 3.71984959e-01 5.51898360e-01
1.83205277e-01 -4.87076849e-01 -1.17719388e+00 -7.34253168e-01
1.11222422e+00 -4.40180153e-01 9.88419771e-01 1.31262231e+00
-1.69455633e-01 -2.88922071e-01 -4.73120332e-01 2.12585866e-01
7.54855692e-01 2.00943455e-01 6.31800294e-01 -1.60342193e+00
-2.30819449e-01 -8.61305475e-01 1.84197854e-02 -4.83965665e-01
5.48575699e-01 -2.20575064e-01 -5.16503513e-01 -1.33563018e+00
2.19673023e-01 -2.83100665e-01 -7.27323353e-01 5.17333150e-02
4.47405828e-03 -8.46894681e-02 -1.54904053e-01 -2.74744853e-02
2.85327256e-01 3.24026912e-01 6.63888216e-01 -4.18090634e-02
1.84780687e-01 2.71410614e-01 -5.57543457e-01 1.02931714e+00
1.03645110e+00 -3.25792760e-01 -2.24245414e-02 -6.37369931e-01
2.85187095e-01 4.84183520e-01 4.35070962e-01 -6.38559639e-01
-8.77055228e-02 -5.11416197e-01 4.22123253e-01 -6.88734114e-01
-1.28573969e-01 -4.81424302e-01 1.02037966e-01 5.24333298e-01
1.09698802e-01 6.39361858e-01 2.48532444e-01 3.27561349e-01
-4.85372126e-01 -2.33446404e-01 5.74090123e-01 -5.47732234e-01
-2.59803832e-01 -1.43511081e-02 -9.37329233e-01 -1.32579505e-01
6.40208721e-01 1.54195517e-01 -8.63842741e-02 -7.40868330e-01
-7.03763604e-01 9.82589722e-02 2.55557299e-01 -2.38739308e-02
5.96266873e-02 -1.24474728e+00 -1.11532211e+00 -1.18641071e-02
-6.38340652e-01 -3.29812616e-01 2.03678295e-01 9.88906741e-01
-6.39775872e-01 1.14895022e+00 -1.27437204e-01 8.00162926e-02
-6.63541436e-01 4.31417853e-01 4.60344225e-01 -7.43309438e-01
-4.55943406e-01 5.75870275e-01 4.71650720e-01 -2.14148462e-02
1.67980880e-01 -3.86975765e-01 -7.03971624e-01 6.58862352e-01
5.13113499e-01 7.38432050e-01 5.36331199e-02 -8.48938882e-01
-1.93761453e-01 4.92157340e-01 2.42324308e-01 -2.54282206e-01
1.87923467e+00 -3.61491859e-01 -3.19384843e-01 8.80579412e-01
9.23466146e-01 -1.98709086e-01 -1.34371030e+00 -1.74399272e-01
5.38046360e-01 -2.37041697e-01 6.45499706e-01 -7.39318490e-01
-1.38336313e+00 1.14132452e+00 3.12485665e-01 4.85199511e-01
8.85760844e-01 -3.93730938e-01 6.44476056e-01 2.49069631e-01
2.91032791e-02 -1.26281226e+00 -3.69462848e-01 5.27741551e-01
8.20615828e-01 -1.37618864e+00 2.13388801e-01 6.42322123e-01
-2.61848897e-01 1.36770594e+00 -1.32016972e-01 -3.43079746e-01
1.09501553e+00 1.08288907e-01 1.33537412e-01 -2.23495275e-01
-9.16766942e-01 1.55721698e-02 4.57069501e-02 5.57595789e-01
6.55432701e-01 3.83359581e-01 -4.24782306e-01 5.27021468e-01
-1.75098151e-01 -2.05563053e-01 6.01603508e-01 5.80089390e-01
-2.69205511e-01 -9.17873979e-01 -7.36652076e-01 3.34345281e-01
-1.38645613e+00 -2.98788130e-01 1.52418420e-01 1.03751230e+00
-1.93909720e-01 1.03081906e+00 3.62193137e-01 1.51767164e-01
-2.72320718e-01 4.73480999e-01 -6.47324696e-02 -2.94738531e-01
-5.10863423e-01 4.10322279e-01 2.35417664e-01 -1.72864944e-01
-7.07364321e-01 -1.32253432e+00 -4.47426468e-01 -7.64555812e-01
-3.61707091e-01 3.07911690e-02 5.75704217e-01 1.17944860e+00
-1.53173089e-01 3.45227718e-01 9.58130240e-01 -1.05881953e+00
-7.80195415e-01 -1.57028949e+00 -8.67796838e-01 -1.63272843e-01
7.21974790e-01 -8.88793766e-01 -7.63975441e-01 1.26669905e-03] | [5.8377485275268555, 3.798964500427246] |
12fb8ab7-4ccb-4789-b183-6f2f36e5dd94 | icdar-2023-competition-on-reading-the-seal | 2304.11966 | null | https://arxiv.org/abs/2304.11966v2 | https://arxiv.org/pdf/2304.11966v2.pdf | ICDAR 2023 Competition on Reading the Seal Title | Reading seal title text is a challenging task due to the variable shapes of seals, curved text, background noise, and overlapped text. However, this important element is commonly found in official and financial scenarios, and has not received the attention it deserves in the field of OCR technology. To promote research in this area, we organized ICDAR 2023 competition on reading the seal title (ReST), which included two tasks: seal title text detection (Task 1) and end-to-end seal title recognition (Task 2). We constructed a dataset of 10,000 real seal data, covering the most common classes of seals, and labeled all seal title texts with text polygons and text contents. The competition opened on 30th December, 2022 and closed on 20th March, 2023. The competition attracted 53 participants from academia and industry including 28 submissions for Task 1 and 25 submissions for Task 2, which demonstrated significant interest in this challenging task. In this report, we present an overview of the competition, including the organization, challenges, and results. We describe the dataset and tasks, and summarize the submissions and evaluation results. The results show that significant progress has been made in the field of seal title text reading, and we hope that this competition will inspire further research and development in this important area of OCR technology. | ['Xiang Bai', 'Dimosthenis Karatzas', 'Yuliang Liu', 'Yinlong Wen', 'Ning Lu', 'Mingrui Chen', 'MingYu Liu', 'Wenwen Yu'] | 2023-04-24 | null | null | null | null | ['optical-character-recognition'] | ['computer-vision'] | [ 5.25399208e-01 1.67754088e-02 1.06551521e-01 -1.43753991e-01
-1.11522770e+00 -7.39536464e-01 5.30009568e-01 4.36363667e-01
-4.81597066e-01 5.52283704e-01 3.20925832e-01 -2.66084522e-01
3.87409568e-01 -2.89219648e-01 -8.59972537e-01 -2.74886757e-01
-5.76567426e-02 1.65637657e-01 2.75140196e-01 -6.03675991e-02
7.10735679e-01 2.07556516e-01 -1.45978820e+00 5.38321316e-01
8.33713472e-01 9.93271232e-01 3.94342065e-01 9.65565026e-01
-2.49259666e-01 8.51911664e-01 -1.24390697e+00 -6.61498249e-01
-6.89679459e-02 -2.00182274e-01 -9.95620608e-01 -2.69532144e-01
8.27931285e-01 4.43796404e-02 -4.63418603e-01 7.82802045e-01
6.50712430e-01 -8.30679312e-02 7.22646892e-01 -8.52227211e-01
-7.47494817e-01 5.96857131e-01 -7.81316698e-01 3.85644376e-01
6.73250914e-01 -3.30819845e-01 1.19180191e+00 -1.10253882e+00
5.36415815e-01 8.83227468e-01 9.35288727e-01 5.20093560e-01
-6.72078848e-01 -5.10920584e-01 1.28889769e-01 -3.12599204e-02
-1.54630470e+00 -3.64117116e-01 3.63306761e-01 -7.32579350e-01
1.16926861e+00 5.91623783e-01 3.16884398e-01 1.04138422e+00
3.43662202e-01 1.41508496e+00 1.08290708e+00 -6.26422048e-01
4.91887257e-02 2.51459450e-01 4.37140912e-01 1.18871316e-01
2.60979980e-01 -2.27393433e-01 -6.37091815e-01 -1.74686406e-02
3.07509501e-04 -4.32211578e-01 -3.81899774e-01 1.59032390e-01
-1.04913592e+00 5.08446574e-01 2.39399504e-02 3.12558413e-01
2.46762991e-01 -1.16058812e-01 5.55939734e-01 1.66150063e-01
6.27574086e-01 6.67142630e-01 -3.96203578e-01 -3.88286710e-01
-1.33078074e+00 6.67247176e-01 7.14316726e-01 1.17253745e+00
2.79096663e-01 -4.50111598e-01 -2.85606593e-01 1.23330724e+00
1.72630340e-01 8.26425135e-01 3.42146218e-01 -3.44728053e-01
9.53378022e-01 3.80782366e-01 1.12928398e-01 -8.68175924e-01
-5.21121025e-01 -1.93923727e-01 -5.40800750e-01 -1.04805179e-01
2.35735878e-01 -8.44189450e-02 -1.12151515e+00 9.11769867e-01
-7.91040808e-02 -5.06465137e-01 -5.72619699e-02 6.71049893e-01
1.32821476e+00 7.87302554e-01 -2.10851714e-01 1.05082810e-01
1.72937298e+00 -1.00745809e+00 -9.45398688e-01 -3.15250546e-01
6.52810514e-01 -1.53042269e+00 1.11448205e+00 4.48445112e-01
-9.85194087e-01 -3.61521810e-01 -1.39154255e+00 -3.47006977e-01
-7.28655100e-01 3.72543126e-01 2.62942072e-02 7.90735960e-01
-7.99415827e-01 2.88594425e-01 -3.53110105e-01 -3.47655356e-01
3.85307103e-01 7.98405185e-02 -1.23755813e-01 7.98064545e-02
-1.29702950e+00 7.13951707e-01 2.10483260e-02 1.78752705e-01
-3.02805990e-01 -7.08161652e-01 -8.65306973e-01 -3.24515641e-01
2.53543615e-01 -1.15487248e-01 1.72847247e+00 -8.59013423e-02
-1.08342195e+00 1.26612937e+00 -8.28527063e-02 -4.82261389e-01
8.32578242e-01 -7.28678465e-01 -5.87518513e-01 7.03212246e-02
3.63645926e-02 4.34283465e-01 7.59445548e-01 -1.08642244e+00
-8.12519670e-01 -2.65590698e-01 -3.70220184e-01 2.39957184e-01
-3.13710690e-01 4.73753333e-01 -7.51831591e-01 -1.14379776e+00
-1.76246107e-01 -7.24561691e-01 3.65097433e-01 -3.71765077e-01
-6.30781591e-01 -3.70392919e-01 9.75875318e-01 -9.86007154e-01
1.50102401e+00 -2.38611031e+00 -5.03311038e-01 -1.60771504e-01
1.24640413e-01 2.75172651e-01 2.29908496e-01 8.36692989e-01
1.06776215e-01 5.21939814e-01 -4.68498021e-01 -8.82533193e-01
7.48490542e-02 -2.52419740e-01 -9.25424814e-01 5.71219802e-01
-9.99731347e-02 7.97750235e-01 -5.61733186e-01 -6.47766113e-01
-3.74216773e-02 2.80312836e-01 3.11674774e-01 3.04742813e-01
-1.69066891e-01 -1.35674879e-01 -3.39487106e-01 8.07553887e-01
8.93763661e-01 -6.31458033e-03 -5.82227468e-01 1.18018605e-01
-3.98144394e-01 4.61948186e-01 -1.19176245e+00 1.21278036e+00
-2.19786003e-01 1.22586441e+00 -6.16177619e-02 -4.71839398e-01
1.08075786e+00 2.62407124e-01 2.42166445e-01 -1.00637484e+00
-2.20481306e-02 4.02396142e-01 -6.92401230e-01 -5.11898041e-01
1.42661011e+00 3.92639101e-01 -3.15534562e-01 3.46400529e-01
-5.24467766e-01 -7.35505223e-01 6.04700327e-01 4.14911777e-01
8.70484531e-01 3.32373939e-02 -4.80612973e-03 -1.88405529e-01
4.42082345e-01 2.72302912e-03 8.89648348e-02 8.78052413e-01
-6.57871692e-03 1.28956950e+00 4.47886616e-01 -3.92765582e-01
-1.08042991e+00 -8.75258088e-01 -6.66545570e-01 8.28604162e-01
3.68400753e-01 -6.57711148e-01 -1.00014281e+00 -5.27541935e-01
-2.05440000e-02 2.64635861e-01 -8.89321089e-01 2.26261988e-01
-8.56800675e-01 -6.75514758e-01 1.08564687e+00 6.64589226e-01
6.94843650e-01 -1.31815350e+00 -7.17107892e-01 -2.74908673e-02
-2.94014901e-01 -1.33855128e+00 -8.82033765e-01 2.62717128e-01
-6.14365578e-01 -1.14360952e+00 -1.36440957e+00 -1.03913856e+00
3.74918073e-01 1.99850962e-01 9.36519325e-01 8.70219991e-02
-7.61943460e-01 3.76785994e-01 -5.77267051e-01 -8.18679392e-01
-3.73861551e-01 4.53489929e-01 -2.58903116e-01 -3.36984068e-01
2.71246284e-01 4.12793964e-01 -3.46764863e-01 4.30274189e-01
-1.03573048e+00 -2.18479797e-01 3.61664087e-01 5.86580753e-01
4.69996870e-01 -2.30721712e-01 2.52400935e-01 -9.33241308e-01
8.61338913e-01 1.07906304e-01 -7.98547626e-01 4.74892825e-01
-6.56255245e-01 -2.03486651e-01 3.24194878e-01 -1.04333222e-01
-9.68847811e-01 -1.34297043e-01 -2.65084743e-01 2.93889731e-01
1.67866781e-01 3.25152010e-01 2.56596748e-02 1.18095763e-01
7.86698639e-01 2.20882326e-01 -4.07723367e-01 -9.69842494e-01
1.92890689e-02 1.36867797e+00 8.19853127e-01 -5.48075259e-01
6.89423501e-01 2.73760676e-01 -5.67990959e-01 -1.16060245e+00
-7.93480337e-01 -8.77015531e-01 -2.97727644e-01 7.91129423e-04
8.89943838e-01 -8.80237460e-01 -2.86898702e-01 1.23090005e+00
-1.21652043e+00 -3.13043177e-01 -2.76789963e-01 4.35270406e-02
-2.98368871e-01 6.71506464e-01 -7.52455413e-01 -8.59645069e-01
-7.28167772e-01 -1.07888055e+00 1.48932505e+00 5.24753213e-01
-4.61587757e-01 -8.06060433e-01 2.82819659e-01 5.78384876e-01
1.90341145e-01 1.53349070e-02 5.02627611e-01 -5.69413006e-01
-2.79068500e-01 -5.09336710e-01 -2.65543401e-01 4.32470888e-01
-7.07989410e-02 2.77927041e-01 -1.23609841e+00 -2.42689654e-01
-3.57501119e-01 -4.99946147e-01 1.16676009e+00 2.57638365e-01
1.18311143e+00 1.70368329e-01 -6.13681614e-01 3.35550964e-01
1.14002001e+00 2.30304554e-01 8.96690547e-01 6.13700688e-01
4.83367264e-01 6.36368155e-01 8.36773157e-01 3.54272932e-01
3.02107811e-01 8.40335011e-01 2.01913461e-01 -2.95188963e-01
-2.38991752e-01 -4.86514747e-01 2.92586833e-01 6.17960572e-01
2.85344541e-01 -5.92156649e-01 -1.18209171e+00 6.85149610e-01
-1.40412450e+00 -6.93813801e-01 -5.53843915e-01 2.18576574e+00
9.47385788e-01 2.48133078e-01 1.04659617e-01 5.12190104e-01
9.10527468e-01 3.44550222e-01 -2.12169334e-01 -6.80449903e-01
-5.35771370e-01 4.77589667e-02 5.87200999e-01 3.68953615e-01
-1.56028450e+00 8.38012159e-01 7.25612259e+00 9.43106353e-01
-9.25155222e-01 -4.02237266e-01 7.57704914e-01 2.80979335e-01
-5.70859052e-02 -3.01961720e-01 -1.42241514e+00 6.05147779e-01
6.78427935e-01 -7.30081946e-02 -1.55368268e-01 7.71393597e-01
-5.99467233e-02 -4.05783385e-01 -8.25909019e-01 9.74551916e-01
3.79229456e-01 -1.31518948e+00 -3.97373706e-01 7.83502758e-02
7.61985600e-01 4.23943475e-02 3.99040967e-01 2.91335404e-01
-7.39039257e-02 -1.22764373e+00 1.16957235e+00 1.73948273e-01
9.85519767e-01 -3.70595753e-01 9.31945801e-01 -1.23838028e-02
-1.27184618e+00 9.34663489e-02 -1.52050927e-01 1.50441170e-01
1.06469810e-01 5.56216896e-01 -6.67025626e-01 3.40654075e-01
1.19742322e+00 5.90856314e-01 -7.67192304e-01 1.48427570e+00
-2.74463177e-01 5.07406771e-01 -3.48719478e-01 -3.51054072e-01
9.37454924e-02 2.51351118e-01 3.70513290e-01 1.89288282e+00
-5.91571964e-02 -2.56132185e-01 -1.13600791e-01 4.61201251e-01
-4.40627784e-01 3.93982351e-01 -2.79270917e-01 -1.70958191e-01
4.40831810e-01 9.39242423e-01 -9.37726736e-01 -3.56511086e-01
-2.75521815e-01 9.05449986e-01 -1.00842118e-01 2.33734295e-01
-6.54724002e-01 -9.26803946e-01 3.87278080e-01 9.77606773e-02
4.65846151e-01 -2.69209631e-02 -8.58641446e-01 -8.67132843e-01
6.23227596e-01 -1.01285791e+00 5.22082686e-01 -7.56019592e-01
-9.65216517e-01 7.01430976e-01 -7.02716261e-02 -1.16582632e+00
1.61073893e-01 -5.50030351e-01 -3.18053365e-01 8.46599460e-01
-1.58471620e+00 -7.66865194e-01 -2.37488523e-01 1.10247351e-01
9.05820787e-01 1.31466180e-01 6.04876995e-01 3.99543196e-01
-3.54444504e-01 1.24084592e+00 5.65436602e-01 4.55477923e-01
1.10416412e+00 -1.41008139e+00 9.38935637e-01 5.38537204e-01
3.73565167e-01 5.30947208e-01 8.48209679e-01 -6.48113489e-01
-1.07558179e+00 -7.55043328e-01 1.38291430e+00 -6.87154949e-01
5.92628300e-01 -7.45561004e-01 -7.43328273e-01 3.21685910e-01
2.75000691e-01 -2.44180605e-01 4.96393979e-01 -1.83980837e-01
-5.38190305e-01 1.63188457e-01 -8.78716230e-01 4.49451655e-01
6.27814591e-01 -5.94351768e-01 -8.17991257e-01 4.16141838e-01
8.34410310e-01 -1.12021661e+00 -7.50017285e-01 3.47041517e-01
4.95451540e-01 -5.57225525e-01 5.49899161e-01 8.98292847e-03
5.66317856e-01 -2.58549660e-01 -3.29876058e-02 -7.10227191e-01
4.64057833e-01 -7.61103511e-01 6.78639710e-02 1.35818768e+00
8.24582756e-01 -2.49708891e-01 7.42289484e-01 3.69551450e-01
-5.65621674e-01 -9.96542275e-01 -9.81301188e-01 -9.42468822e-01
4.74826783e-01 -4.65532362e-01 4.23417419e-01 3.73028278e-01
1.59297895e-03 1.81438535e-01 -4.41464603e-01 -2.02303559e-01
2.37097636e-01 8.35766420e-02 7.77395248e-01 -9.76437092e-01
-1.93261146e-03 -5.74548781e-01 -2.64883667e-01 -1.53258669e+00
-3.75184149e-01 -4.42688823e-01 3.24772656e-01 -1.69804263e+00
1.39246672e-01 -2.91689962e-01 1.33168325e-01 4.01359618e-01
-3.69725198e-01 6.04475737e-01 2.99952120e-01 3.61933917e-01
-6.97723448e-01 1.57976255e-01 1.01965761e+00 -5.53950369e-01
-5.87215982e-02 2.08157256e-01 -5.12135267e-01 3.28230649e-01
5.44163346e-01 -3.93930316e-01 -3.25356759e-02 -3.10671508e-01
3.90439898e-01 -4.62467909e-01 6.17029369e-02 -9.23511922e-01
1.66519135e-01 3.13835561e-01 4.44924146e-01 -1.13995135e+00
3.17034990e-01 -4.24257308e-01 -4.74256814e-01 7.30233043e-02
-4.62104082e-01 7.66382217e-02 4.16834503e-01 2.23157987e-01
-3.81217182e-01 -5.09129763e-01 6.18489742e-01 5.02308570e-02
-5.13601959e-01 -8.01714435e-02 -5.10530472e-01 6.61745310e-01
8.85930061e-01 -5.52996874e-01 -4.16958541e-01 -1.26020595e-01
-4.39147204e-01 3.53302628e-01 6.34065866e-01 8.73913646e-01
6.31650388e-01 -6.82737827e-01 -7.42624044e-01 6.86559752e-02
4.05548543e-01 1.14871144e-01 2.01850861e-01 5.49548090e-01
-7.80627966e-01 6.38886154e-01 5.12307703e-01 -6.50370896e-01
-1.62803340e+00 3.83104801e-01 3.03389877e-01 -4.55069155e-01
-9.38841760e-01 1.04557455e+00 2.81159412e-02 -5.24168946e-02
7.23749399e-01 -2.98452914e-01 -4.67556030e-01 2.63820231e-01
1.06339598e+00 2.76658207e-01 5.82045734e-01 -5.29970825e-01
-3.51389915e-01 9.10388291e-01 -5.78313410e-01 -3.31211761e-02
9.37600017e-01 -4.32610922e-02 7.92161599e-02 5.13070822e-01
1.32790482e+00 2.27355033e-01 -9.80906129e-01 -9.06163007e-02
4.50417280e-01 -2.77424484e-01 -3.34609479e-01 -9.65694249e-01
-6.90679908e-01 1.02290487e+00 4.10367578e-01 3.38190466e-01
9.31297839e-01 6.84359372e-02 1.09611166e+00 1.94665283e-01
3.86903435e-02 -1.68035662e+00 -3.17591242e-02 7.48018682e-01
9.57094193e-01 -9.80321288e-01 2.53649801e-01 -3.97567689e-01
-5.53262055e-01 1.12147415e+00 4.09866393e-01 2.62182534e-01
4.01020586e-01 7.22642422e-01 2.93738872e-01 -1.66962206e-01
-3.03132176e-01 2.48827189e-01 4.71534550e-01 5.17261267e-01
6.63591146e-01 -9.13791209e-02 -4.27137434e-01 4.78623539e-01
-5.59020996e-01 -2.77315319e-01 6.70486212e-01 1.14248741e+00
-4.08463627e-01 -1.10157859e+00 -6.63506508e-01 5.05511343e-01
-8.76184404e-01 -4.18282062e-01 -8.58014345e-01 8.56177270e-01
-2.84205467e-01 9.57283437e-01 -9.16315764e-02 -2.70637870e-01
5.86856663e-01 -2.18115300e-01 2.23616913e-01 -5.48150837e-01
-8.63324106e-01 -1.16867065e-01 1.96531832e-01 -1.37932464e-01
1.71010092e-01 -8.19730222e-01 -1.27643454e+00 -2.11827338e-01
-4.63316351e-01 3.52922857e-01 9.45438385e-01 7.41059005e-01
1.65628716e-01 1.90761954e-01 4.97825831e-01 -3.18154007e-01
-6.19811475e-01 -1.01174307e+00 -6.03412628e-01 9.83496904e-02
5.23435056e-01 -3.46242845e-01 -5.92797935e-01 3.95837098e-01] | [11.838737487792969, 2.53881573677063] |
05bc2ba3-4c77-4644-9778-136fd0ce78d3 | effective-hierarchical-information-threading | null | null | https://link.springer.com/chapter/10.1007/978-3-031-28244-7_44 | https://doi.org/10.1007/978-3-031-28244-7_44 | Effective Hierarchical Information Threading Using Network Community Detection | With the tremendous growth in the volume of information produced online every day (e.g. news articles), there is a need for automatic methods to identify related information about events as the events evolve over time (i.e., information threads). In this work, we propose a novel unsupervised approach, called HINT, which identifies coherent Hierarchical Information Threads. These threads can enable users to easily interpret a hierarchical association of diverse evolving information about an event or discussion. In particular, HINT deploys a scalable architecture based on network community detection to effectively identify hierarchical links between documents based on their chronological relatedness and answers to the 5W1H questions (i.e., who, what, where, when, why & how). On the NewSHead collection, we show that HINT markedly outperforms existing state-of-the-art approaches in terms of the quality of the identified threads. We also conducted a user study that shows that our proposed network-based hierarchical threads are significantly (p<0.05) preferred by users compared to cluster-based sequential threads. | ['Iadh Ounis', 'Graham McDonald', 'Hitarth Narvala'] | 2023-03-17 | null | null | null | european-conference-on-information-retrieval-3 | ['community-detection'] | ['graphs'] | [-3.73575121e-01 -1.13585591e-01 -1.85716152e-01 1.98605144e-03
-1.77293837e-01 -8.29174817e-01 8.12720120e-01 1.08207345e+00
-1.98868886e-01 2.57897288e-01 9.60260332e-01 -3.51220280e-01
-3.87587249e-01 -7.46685207e-01 -1.07046038e-01 -2.50278115e-01
-4.41188127e-01 4.20357704e-01 8.05331171e-01 -3.31724465e-01
6.87120259e-01 -1.73341837e-02 -1.73033512e+00 5.92866123e-01
7.97878563e-01 6.00322783e-01 1.61894515e-01 5.29788136e-01
-5.64360082e-01 9.63128984e-01 -5.48035204e-01 -3.49947780e-01
-2.51908749e-01 -5.68565786e-01 -1.03314197e+00 -5.61434170e-03
9.82188135e-02 4.74220477e-02 -1.99074805e-01 7.90344775e-01
1.83773860e-01 1.97403222e-01 4.56848204e-01 -1.25934386e+00
-4.27946635e-02 9.29375172e-01 -7.27859437e-01 7.27300942e-01
7.39420950e-01 -5.19652903e-01 1.56927359e+00 -5.88432431e-01
1.19572151e+00 1.20736194e+00 5.46411872e-01 -2.35183194e-01
-1.11440504e+00 -4.95869547e-01 2.20321715e-01 5.13467729e-01
-1.19601381e+00 -1.21374540e-01 6.79894984e-01 -7.79989898e-01
5.84473670e-01 5.14118791e-01 7.72163510e-01 9.60997641e-01
1.93989500e-01 6.01737618e-01 8.48597527e-01 -3.71102154e-01
3.36479843e-01 -6.09984174e-02 7.43070126e-01 5.46552479e-01
1.56727344e-01 -6.75665915e-01 -1.05398452e+00 -6.92986012e-01
3.84613812e-01 6.39290884e-02 -1.72985092e-01 -3.20703466e-03
-1.39747238e+00 9.78467524e-01 3.50019723e-01 5.06235421e-01
-6.05204225e-01 -4.06446546e-01 6.85358226e-01 1.95404619e-01
6.35966361e-01 3.32729608e-01 -1.71037361e-01 -3.25342566e-01
-8.75396729e-01 2.81341672e-01 1.19484937e+00 6.25512779e-01
5.70242882e-01 -8.70463729e-01 -1.93676278e-01 5.97053587e-01
4.90433723e-01 -2.81634092e-01 3.62690985e-01 -7.09340274e-01
3.79914820e-01 1.06321537e+00 2.30985731e-01 -1.58417654e+00
-7.30232179e-01 -5.12755632e-01 -9.09103751e-01 -4.96428102e-01
3.91314507e-01 -8.54211375e-02 -1.10853940e-01 1.42447400e+00
6.70118630e-01 1.24304317e-01 -1.25540435e-01 5.80054641e-01
9.77388024e-01 8.76042962e-01 -9.48632956e-02 -5.41245699e-01
1.84474003e+00 -7.34944761e-01 -9.77771759e-01 1.78988710e-01
4.03941631e-01 -1.12530065e+00 9.20233786e-01 2.75887519e-01
-8.83413315e-01 -3.28834355e-01 -6.97993994e-01 2.30944306e-01
-1.88856140e-01 -3.59477252e-01 5.27049541e-01 1.28496021e-01
-9.75171089e-01 3.56536865e-01 -6.79018795e-01 -1.15110993e+00
-1.16066702e-01 -2.64813095e-01 1.33275047e-01 2.45619103e-01
-1.25547183e+00 4.00723368e-01 5.89214921e-01 -3.44200164e-01
-4.83129084e-01 -4.38336432e-01 -2.39449218e-01 1.32829651e-01
4.58300114e-01 -5.25880814e-01 1.24676037e+00 -4.71028268e-01
-9.90644038e-01 5.69655597e-01 -4.39386070e-01 -4.55848157e-01
1.76067948e-01 -2.49113306e-01 -4.10507053e-01 3.82565975e-01
2.90599495e-01 1.92233756e-01 4.15432453e-01 -9.77281868e-01
-1.18057418e+00 -1.00554608e-01 1.21392749e-01 2.39624873e-01
-8.80601883e-01 6.35621607e-01 -6.25936329e-01 -5.59450328e-01
3.44978184e-01 -8.55146110e-01 -1.86493486e-01 -4.07582551e-01
-5.34104288e-01 -9.60981667e-01 8.23388755e-01 -5.46530843e-01
1.83805108e+00 -1.94447351e+00 2.65755534e-01 5.07279932e-01
7.66382813e-01 -4.71369356e-01 3.36323053e-01 1.17992151e+00
3.43816459e-01 3.58325452e-01 4.74327475e-01 -5.11164702e-02
-4.65073474e-02 1.18602075e-01 -3.60676676e-01 1.89741567e-01
-6.31803572e-01 2.27972165e-01 -1.17439318e+00 -6.76282704e-01
-5.22211790e-01 1.11848190e-01 -4.43518430e-01 3.43166441e-01
-2.26377770e-01 4.48355675e-01 -5.64801097e-01 1.65764645e-01
9.73247215e-02 -1.02608693e+00 3.61212641e-01 -2.21609071e-01
-5.94738185e-01 5.37280798e-01 -1.02946413e+00 1.24588335e+00
1.90614183e-02 8.56830001e-01 -7.29771033e-02 -6.33468390e-01
8.10908020e-01 5.80096960e-01 4.35646713e-01 -3.95656168e-01
-6.01315089e-02 -7.13798776e-02 1.94964215e-01 -4.58481461e-01
6.60223484e-01 5.45939028e-01 -1.81925923e-01 1.00518799e+00
-2.42366627e-01 6.59528196e-01 7.41732776e-01 1.06858265e+00
1.45612049e+00 -6.90281630e-01 4.68940973e-01 -4.90556628e-01
2.72978544e-01 1.83314621e-01 4.68336701e-01 8.53299677e-01
-7.26416381e-03 2.16506675e-01 8.54155362e-01 -4.98304188e-01
-9.31381941e-01 -8.13526571e-01 1.67097703e-01 1.42903829e+00
2.57156044e-01 -1.22561491e+00 -4.69786763e-01 -3.70164961e-01
-1.83833882e-01 4.14843708e-01 -6.57613158e-01 4.85396206e-01
-3.42765689e-01 -6.55279398e-01 -4.81801527e-03 4.95793708e-02
5.07828832e-01 -9.66144562e-01 -5.04192293e-01 4.52292532e-01
-7.81222701e-01 -9.24593449e-01 -4.58532035e-01 -1.86193928e-01
-7.61687934e-01 -9.72696066e-01 -3.28648955e-01 -5.07247388e-01
5.14242947e-01 6.13993168e-01 1.33152020e+00 2.93017656e-01
9.50995274e-03 3.42969835e-01 -1.11458313e+00 -1.88580632e-01
-3.29123497e-01 4.65230972e-01 -5.67627288e-02 2.14051872e-01
1.69140622e-01 -6.92225993e-01 -9.59258914e-01 7.08637297e-01
-7.90560126e-01 2.22296000e-01 1.56758159e-01 2.95419902e-01
4.90070544e-02 5.54362714e-01 5.25926888e-01 -1.05399728e+00
1.01842511e+00 -1.03811276e+00 -3.51215392e-01 1.55184224e-01
-7.28543341e-01 -4.54544388e-02 3.55060488e-01 -2.98768282e-01
-9.71120596e-01 -4.38156694e-01 2.76965976e-01 4.46329355e-01
3.87536995e-02 1.13058126e+00 5.35717010e-01 4.21918780e-01
7.08854437e-01 -1.55993372e-01 -4.83464122e-01 -6.43375218e-01
2.27599680e-01 7.47652411e-01 2.97163159e-01 -4.13854748e-01
7.84848571e-01 5.88999152e-01 -4.97184366e-01 -9.17052090e-01
-1.17784142e+00 -1.24582458e+00 -3.48428249e-01 -7.10425436e-01
8.49798679e-01 -7.02339292e-01 -8.09562862e-01 2.14878976e-01
-1.30810773e+00 1.37173906e-01 4.33177799e-01 3.29062968e-01
1.52628973e-01 4.87192154e-01 -8.52356672e-01 -6.16116285e-01
-4.13749874e-01 -4.81276542e-01 3.04737568e-01 3.98515135e-01
-1.08475780e+00 -1.07073879e+00 3.15419346e-01 4.52630281e-01
5.07110953e-01 2.92889535e-01 9.55721319e-01 -9.04187262e-01
-5.52898347e-01 -7.81768933e-02 -2.81123370e-01 -7.32961655e-01
1.65918633e-01 2.31249556e-01 -3.49401951e-01 -2.88964331e-01
-2.57568687e-01 1.82875648e-01 5.58707952e-01 1.96864530e-02
6.56217635e-01 -7.28096068e-01 -6.08704090e-01 -2.61142820e-01
8.63877594e-01 2.11511254e-01 4.04115260e-01 7.00987756e-01
4.97626692e-01 8.72485161e-01 3.54828477e-01 1.00700200e+00
6.99603021e-01 5.67981005e-01 2.51333237e-01 6.56728968e-02
2.73440689e-01 -2.11691067e-01 1.85759038e-01 1.35296917e+00
-1.31757379e-01 -5.02418280e-01 -1.15673137e+00 6.44567549e-01
-2.32743549e+00 -1.15933871e+00 -6.88690841e-01 1.85621047e+00
8.33519995e-01 5.13222575e-01 3.78841370e-01 9.37946886e-02
9.40395176e-01 2.19694108e-01 -1.95699677e-01 -1.59256503e-01
1.80940151e-01 -3.10914725e-01 -1.36694359e-02 1.60124823e-01
-8.62493694e-01 5.08265734e-01 5.82447100e+00 5.39080858e-01
-5.49144685e-01 1.57301590e-01 4.43685293e-01 2.78545260e-01
-2.89104581e-01 1.95968941e-01 -8.32969844e-01 5.18334925e-01
1.04061711e+00 -6.37197614e-01 1.13018285e-02 5.48158348e-01
4.98641133e-01 -3.32952201e-01 -7.56553710e-01 6.06714010e-01
-3.58537436e-02 -1.66435599e+00 -8.46810639e-02 7.61260539e-02
7.24542856e-01 -8.88321623e-02 -3.32018256e-01 -5.53092174e-02
7.23430157e-01 -1.58616289e-01 5.93956172e-01 3.78741741e-01
-1.57375544e-01 -6.92695022e-01 6.58799887e-01 6.89641953e-01
-1.52604651e+00 5.18182106e-03 2.75177546e-02 -2.49802917e-01
3.11498910e-01 8.61663938e-01 -7.76188076e-01 5.67544758e-01
1.18170929e+00 8.45058441e-01 -6.12441123e-01 1.27611029e+00
-2.88668573e-01 1.21412039e+00 -2.39180356e-01 -3.96594077e-01
2.13924080e-01 -8.21605250e-02 5.98495185e-01 1.26013720e+00
4.38865483e-01 2.32447699e-01 4.58028555e-01 3.16910565e-01
-2.39033904e-02 3.50278586e-01 -1.54792771e-01 -1.15559839e-01
6.40420079e-01 1.58637905e+00 -1.48438931e+00 -6.14258230e-01
-3.63824159e-01 6.49069488e-01 2.94820994e-01 1.86999068e-01
-3.88606191e-01 -3.84527385e-01 2.61479288e-01 4.22689140e-01
1.97427884e-01 -2.80644476e-01 7.29403049e-02 -1.03002858e+00
-2.74612218e-01 -6.83860838e-01 9.84858990e-01 -7.25850463e-01
-1.56932509e+00 7.37192035e-01 3.92479300e-02 -1.12094808e+00
-1.60807654e-01 2.27246985e-01 -9.77368593e-01 4.20811892e-01
-8.00312042e-01 -4.36928451e-01 -5.92553675e-01 4.33533907e-01
4.89601165e-01 7.16744810e-02 5.46685755e-01 2.20862851e-01
-5.99392354e-01 -6.39154240e-02 1.68589607e-01 1.88988611e-01
7.03401089e-01 -1.11745393e+00 4.46147501e-01 7.38745749e-01
2.93079823e-01 8.62328053e-01 1.09140027e+00 -8.70188057e-01
-9.67476785e-01 -6.87089503e-01 1.27960968e+00 -2.20268145e-01
1.35409462e+00 -4.95135188e-01 -9.14976358e-01 3.64289910e-01
8.51652205e-01 -5.94410002e-01 1.07393050e+00 6.60453916e-01
-3.79508644e-01 7.56873637e-02 -6.50612533e-01 8.41256678e-01
1.03295946e+00 -2.19665438e-01 -7.87866950e-01 6.11226022e-01
7.70332515e-01 -4.02896255e-02 -9.19705272e-01 -1.45458207e-01
3.98811340e-01 -1.05729961e+00 6.11421943e-01 -4.34323281e-01
4.26900566e-01 -4.35463667e-01 2.97416121e-01 -1.09232295e+00
-5.68707645e-01 -1.07740128e+00 -2.70133168e-01 1.45012105e+00
4.43528354e-01 -5.47825575e-01 7.07900524e-01 1.72052339e-01
1.91687927e-01 -2.08211720e-01 -6.63565218e-01 -4.18075204e-01
-8.11280131e-01 1.62732438e-03 9.39597785e-02 1.24108255e+00
5.88978827e-01 8.88328612e-01 -1.81721911e-01 1.80597261e-01
6.15845382e-01 5.62358797e-01 6.34637654e-01 -1.92545962e+00
-2.13562965e-01 -4.42561001e-01 -2.31819913e-01 -9.30540979e-01
-4.86635208e-01 -6.92388177e-01 -2.54398465e-01 -1.88720775e+00
6.34560645e-01 -1.76922590e-01 -3.20076376e-01 1.49854407e-01
-2.76371032e-01 -5.07401414e-02 -3.47191654e-02 9.34818983e-01
-1.29224002e+00 3.34315598e-02 7.71057785e-01 2.87136048e-01
-4.43716317e-01 5.91854714e-02 -6.15169585e-01 8.06862056e-01
8.00911844e-01 -7.15935707e-01 -1.92159876e-01 1.73491761e-01
1.01699781e+00 7.51816183e-02 -1.12326503e-01 -8.57692242e-01
8.16168308e-01 -2.49466393e-02 -5.36960438e-02 -9.68935370e-01
-2.82171011e-01 -3.78081411e-01 2.61427462e-01 5.60820282e-01
-6.43956542e-01 5.63754499e-01 -1.71666667e-01 7.85103023e-01
-3.90032828e-01 -3.04573085e-02 3.19954753e-02 -8.16993415e-02
-5.73454559e-01 4.82589342e-02 -9.45927620e-01 2.11962789e-01
7.17135012e-01 2.42093429e-01 -5.56002796e-01 -7.85702586e-01
-6.69911683e-01 4.24926758e-01 1.27461910e-01 5.70577085e-01
3.88011873e-01 -1.17753661e+00 -8.54924023e-01 -5.21932662e-01
3.53206128e-01 -2.63216138e-01 1.01844490e-01 9.00023997e-01
-4.25901502e-01 1.70314685e-01 7.99392164e-02 -6.22636795e-01
-1.76658773e+00 3.17717522e-01 -5.40304959e-01 -6.23088479e-01
-7.57614613e-01 6.21118486e-01 1.52812181e-02 1.20145939e-01
3.95559043e-01 -7.71207586e-02 -7.86644876e-01 8.61499250e-01
8.03065896e-01 4.37610567e-01 -6.35205582e-02 -3.69266927e-01
-1.96500316e-01 2.93084502e-01 -4.37437594e-01 -2.10590139e-01
1.33042860e+00 -3.32506567e-01 -5.60656369e-01 7.55990803e-01
9.57024753e-01 5.86492009e-02 -6.87650740e-01 -5.19706011e-01
8.54475439e-01 -2.43864179e-01 -1.29668415e-01 -4.47805196e-01
-6.60539448e-01 3.32821995e-01 2.04556704e-01 1.32665324e+00
7.27172911e-01 4.77249295e-01 8.16540003e-01 3.85998428e-01
1.19914092e-01 -9.24760640e-01 3.98941964e-01 7.11102605e-01
7.44069040e-01 -9.21485245e-01 1.60131603e-01 -4.92757350e-01
-4.27386373e-01 1.23775554e+00 2.26632714e-01 2.13462129e-01
7.71953225e-01 -1.47603340e-02 8.03583711e-02 -6.60899103e-01
-1.18276680e+00 -1.20118491e-01 3.63683730e-01 -1.44703582e-01
7.16476023e-01 -1.70309782e-01 -6.04867876e-01 9.15028304e-02
-2.32583582e-01 -4.12701786e-01 5.75273812e-01 1.01721239e+00
-8.10479164e-01 -8.84384632e-01 -4.78621453e-01 4.76522058e-01
-4.66728091e-01 3.91894691e-02 -7.08636403e-01 5.52236438e-01
-2.87064552e-01 1.31132567e+00 3.49134356e-01 -3.30980182e-01
1.40792772e-01 -6.59672618e-02 -3.24182123e-01 -5.69159746e-01
-9.04392242e-01 2.78907627e-01 3.44357193e-01 -4.27269846e-01
-6.43199503e-01 -9.17922199e-01 -1.16330397e+00 -4.55539554e-01
-1.91113576e-01 7.66098022e-01 4.32790548e-01 7.56736577e-01
5.70511162e-01 5.34823954e-01 6.77843809e-01 -4.31430936e-01
6.12508878e-02 -9.61918831e-01 -1.77825853e-01 5.91060519e-01
-1.07171424e-01 -4.31527227e-01 -5.79133213e-01 1.79800496e-01] | [10.361712455749512, 7.42232608795166] |
f6310b2e-a90e-4f04-81f9-164c1f0a9e40 | efficient-gender-debiasing-of-pre-trained | 2209.03661 | null | https://arxiv.org/abs/2209.03661v1 | https://arxiv.org/pdf/2209.03661v1.pdf | Efficient Gender Debiasing of Pre-trained Indic Language Models | The gender bias present in the data on which language models are pre-trained gets reflected in the systems that use these models. The model's intrinsic gender bias shows an outdated and unequal view of women in our culture and encourages discrimination. Therefore, in order to establish more equitable systems and increase fairness, it is crucial to identify and mitigate the bias existing in these models. While there is a significant amount of work in this area in English, there is a dearth of research being done in other gendered and low resources languages, particularly the Indian languages. English is a non-gendered language, where it has genderless nouns. The methodologies for bias detection in English cannot be directly deployed in other gendered languages, where the syntax and semantics vary. In our paper, we measure gender bias associated with occupations in Hindi language models. Our major contributions in this paper are the construction of a novel corpus to evaluate occupational gender bias in Hindi, quantify this existing bias in these systems using a well-defined metric, and mitigate it by efficiently fine-tuning our model. Our results reflect that the bias is reduced post-introduction of our proposed mitigation techniques. Our codebase is available publicly. | ['Aditya Kane', 'V Manushree', 'Neeraja Kirtane'] | 2022-09-08 | null | null | null | null | ['culture'] | ['speech'] | [-2.14717150e-01 2.74317056e-01 -4.39283758e-01 -8.42825711e-01
-2.95626409e-02 -4.29521650e-01 8.62456441e-01 1.13816358e-01
-9.64186072e-01 9.40543354e-01 6.36402667e-01 -4.73906994e-01
1.15070399e-02 -8.33994508e-01 -2.51965165e-01 -3.98173660e-01
5.11389732e-01 7.76031017e-01 -5.16748354e-02 -8.44824195e-01
6.82666659e-01 1.73434153e-01 -1.47396684e+00 -7.08837509e-02
8.06262553e-01 1.17132485e-01 -8.66598040e-02 2.97402322e-01
6.96077347e-02 8.65491271e-01 -7.28264689e-01 -8.83014500e-01
1.44967139e-01 -3.32541972e-01 -1.01638377e+00 -5.27177155e-01
6.64500654e-01 -2.29319811e-01 -5.68677299e-02 1.00232327e+00
8.18238080e-01 -2.49268979e-01 8.04774463e-01 -1.16470993e+00
-6.99327946e-01 9.56962943e-01 -6.39757574e-01 2.49423608e-01
6.72261193e-02 -3.36290926e-01 8.50783706e-01 -8.32982779e-01
7.94077039e-01 1.88225043e+00 6.67816222e-01 7.69237459e-01
-1.19330263e+00 -1.14059722e+00 3.97977643e-02 -6.38990030e-02
-1.29867208e+00 -4.06190723e-01 6.74975455e-01 -6.20212853e-01
6.29424512e-01 5.22296727e-01 5.43407738e-01 1.33854854e+00
4.53307509e-01 2.70193934e-01 1.65000570e+00 -6.67197406e-01
-2.05615401e-01 6.91034913e-01 3.74329507e-01 4.26487535e-01
4.86850977e-01 1.85639247e-01 -7.11115718e-01 -1.15267180e-01
3.11935812e-01 -4.34030779e-02 5.64187825e-01 -9.19554904e-02
-1.14239538e+00 9.71427917e-01 4.97967470e-03 6.09205723e-01
4.21289466e-02 5.39469495e-02 5.39380193e-01 2.19403759e-01
6.81278169e-01 4.27573085e-01 -3.82735014e-01 -4.52013105e-01
-1.18115389e+00 8.20011139e-01 7.67974436e-01 7.46500313e-01
4.99305457e-01 -1.11913636e-01 -1.47550851e-01 1.17704380e+00
4.97371584e-01 7.57919431e-01 4.04782712e-01 -1.06698573e+00
4.79296833e-01 7.51984060e-01 4.52384241e-02 -1.41407430e+00
-4.71875101e-01 -3.46222818e-01 -6.45510674e-01 2.40890428e-01
7.46546984e-01 7.65886828e-02 -6.88894451e-01 1.94160604e+00
1.86381061e-02 -9.32549000e-01 -1.76858306e-01 8.31823766e-01
7.18207598e-01 4.02390845e-02 5.85508764e-01 1.72305003e-01
1.59834802e+00 -4.44869280e-01 -7.23252118e-01 -5.71290076e-01
5.29078364e-01 -1.16714883e+00 1.12568688e+00 7.67743811e-02
-9.39485490e-01 -4.49558228e-01 -8.92293692e-01 -3.16865891e-01
-6.88763201e-01 -5.36839999e-02 7.45476544e-01 1.47014380e+00
-6.90064728e-01 2.39426300e-01 -4.10627753e-01 -7.12107301e-01
3.69018853e-01 5.17919719e-01 -1.88754991e-01 1.07352316e-01
-1.46612179e+00 1.21987593e+00 2.93623935e-02 -1.08754233e-01
-3.55861694e-01 -6.62891626e-01 -8.24161649e-01 -5.86146355e-01
2.33484298e-01 -3.41392934e-01 1.07512832e+00 -1.06271112e+00
-5.25716126e-01 1.54907346e+00 -2.84301519e-01 -1.48571759e-01
7.40246773e-01 7.95245469e-02 -5.24577498e-01 -6.52435660e-01
6.86024547e-01 7.40512908e-01 4.16431844e-01 -1.53235805e+00
-9.64504838e-01 -8.10995579e-01 2.76852548e-01 3.34067270e-02
-4.34098542e-01 8.07987154e-01 1.67179599e-01 -7.99493194e-01
-1.23710353e-02 -1.09170496e+00 -1.55063137e-01 -6.87385082e-01
-6.84512407e-02 -2.19751105e-01 4.41721439e-01 -9.63769615e-01
1.78738356e+00 -1.89934504e+00 -2.61512071e-01 3.02052677e-01
3.26134026e-01 -2.03307215e-02 6.07790828e-01 5.53585112e-01
1.82610586e-01 5.04450738e-01 -1.52293876e-01 -4.12442796e-02
3.79542947e-01 5.74004412e-01 -2.16107011e-01 3.95099193e-01
-1.51120171e-01 4.42952037e-01 -8.03779602e-01 -9.89232004e-01
-1.87779188e-01 2.81718075e-01 -9.27374125e-01 -2.65843093e-01
6.03697717e-01 1.82617530e-01 1.17884830e-01 9.01582658e-01
8.72769594e-01 6.83398724e-01 3.94970804e-01 -1.80597734e-02
-5.92858374e-01 4.57257509e-01 -1.13391471e+00 1.10323834e+00
-3.04836452e-01 4.52724874e-01 5.03673591e-02 -8.99009228e-01
1.04180217e+00 -1.70028627e-01 2.44209632e-01 -8.72397780e-01
2.08334729e-01 6.99433506e-01 5.07559836e-01 -3.09221327e-01
1.18899286e+00 -4.06215012e-01 -7.14685142e-01 3.99848074e-01
-1.27136871e-01 -4.09179479e-02 6.04154885e-01 1.03690587e-01
3.55528682e-01 2.00997248e-01 2.56762981e-01 -8.52287114e-01
4.43041921e-01 -2.65726186e-02 7.28604198e-01 7.02966332e-01
-5.66037953e-01 5.19443035e-01 6.01595104e-01 -5.54472566e-01
-1.07683372e+00 -6.96431875e-01 -5.07936656e-01 1.79608810e+00
-2.09940702e-01 -3.35768968e-01 -7.13770509e-01 -7.03224361e-01
2.68677592e-01 8.20952952e-01 -7.04504430e-01 -1.38067067e-01
-7.80276775e-01 -1.12678874e+00 9.89995480e-01 4.43927884e-01
4.69902575e-01 -8.07152212e-01 -8.17828834e-01 7.41731450e-02
-3.47025573e-01 -6.53115571e-01 -2.11944729e-01 1.71692580e-01
-6.55620456e-01 -8.37282300e-01 -5.10122776e-01 -8.54026735e-01
4.39174265e-01 -1.38280675e-01 1.60045707e+00 1.30187437e-01
-1.49166035e-02 2.46088137e-03 1.90295037e-02 -1.28946245e+00
-7.04465270e-01 4.61156756e-01 4.00604829e-02 -6.01454318e-01
1.19094551e+00 -1.78734764e-01 -4.32816982e-01 5.21758437e-01
-5.90391397e-01 -1.77084908e-01 2.88477570e-01 6.90967083e-01
-7.30139241e-02 -1.81241244e-01 5.83002448e-01 -1.45700002e+00
5.79499543e-01 -3.58135611e-01 -3.16254012e-02 -1.47226706e-01
-1.08502519e+00 -2.46856540e-01 -2.38744710e-02 5.10823987e-02
-1.21469522e+00 -5.44972003e-01 -3.34587216e-01 8.37321222e-01
-1.17868274e-01 3.30787152e-01 -6.36922419e-02 2.83440977e-01
7.83710301e-01 -4.55550522e-01 1.33765368e-02 -3.78603876e-01
-8.22617114e-02 1.19047701e+00 2.35316634e-01 -9.34413373e-01
5.13470829e-01 4.47583556e-01 -1.08324893e-01 -7.20903218e-01
-6.39638007e-01 -3.85451376e-01 -8.77743363e-01 -3.07039499e-01
7.48342097e-01 -8.86987627e-01 -3.91234875e-01 3.75705093e-01
-7.18490660e-01 -5.77920936e-02 9.23334807e-02 5.00866771e-01
-1.80296198e-01 1.23450167e-01 -2.90985703e-01 -1.05493760e+00
-6.28619418e-02 -8.73326063e-01 8.86335790e-01 2.87627634e-02
-1.12373567e+00 -1.03133416e+00 -9.10809077e-03 6.56980157e-01
3.25192362e-01 3.39046470e-03 1.18480897e+00 -5.89484334e-01
2.51259029e-01 1.01381592e-01 -2.59483218e-01 3.19627851e-01
2.07734630e-02 1.88463837e-01 -9.35092032e-01 -5.29334508e-02
-1.88536406e-01 -2.42625535e-01 7.35565603e-01 1.62534341e-01
7.30376601e-01 -1.81762666e-01 -8.57969820e-02 9.99058560e-02
1.21217847e+00 2.45644152e-01 5.28729498e-01 8.05827320e-01
7.15466499e-01 1.19198382e+00 9.96335626e-01 2.42230162e-01
1.05805194e+00 5.35693705e-01 8.95350501e-02 -2.62416601e-01
-5.72706200e-02 -1.22569412e-01 1.56986445e-01 8.68586063e-01
-8.79903197e-01 4.05437768e-01 -1.46046376e+00 7.89278984e-01
-1.77526057e+00 -1.16087413e+00 -4.01236802e-01 2.06646562e+00
1.20480645e+00 3.08259815e-01 5.01299202e-01 4.44334179e-01
4.78865564e-01 9.91495028e-02 2.22807392e-01 -1.13837039e+00
-2.23367646e-01 2.12816805e-01 7.06226289e-01 5.78846395e-01
-1.17173183e+00 9.25643146e-01 6.75211811e+00 4.48911905e-01
-9.27871108e-01 2.54889410e-02 7.03069448e-01 8.41905858e-05
-3.73507053e-01 5.90614881e-03 -9.58643079e-01 3.92256349e-01
8.54341328e-01 -2.59623557e-01 1.36166304e-01 8.28466296e-01
2.65559912e-01 -1.05898917e-01 -8.72537971e-01 9.17025566e-01
9.26198810e-02 -6.37536883e-01 -2.81012177e-01 4.40294117e-01
4.86261755e-01 -3.86800677e-01 2.03716248e-01 7.24618018e-01
2.19974950e-01 -1.16367781e+00 1.37102914e+00 1.56071723e-01
5.92243016e-01 -1.15420091e+00 1.04033327e+00 7.55745023e-02
-3.55961084e-01 -2.60183930e-01 -6.44774139e-01 -7.43286729e-01
-3.99172723e-01 3.97563875e-01 -6.76798820e-01 1.68992430e-01
1.00939405e+00 4.32277083e-01 -1.01768672e+00 1.83759928e-01
1.07235067e-01 8.83049130e-01 1.54751733e-01 -7.44088143e-02
6.81703985e-02 -2.24040449e-01 2.08461732e-01 1.71471417e+00
8.52390379e-02 -4.06605780e-01 2.18247576e-03 5.63707948e-01
4.62200940e-02 4.19198960e-01 -7.09315062e-01 -7.88871273e-02
4.46645945e-01 1.16492605e+00 -7.62892485e-01 -1.69956610e-01
-6.66907132e-01 3.62481922e-01 -1.27442628e-01 1.23892881e-01
-7.97078907e-01 -3.62661988e-01 8.12235594e-01 5.62165916e-01
-4.53137845e-01 -1.83135763e-01 -9.02713597e-01 -7.55207837e-01
-3.08900237e-01 -1.39669621e+00 4.75154698e-01 -1.12087280e-01
-1.15150714e+00 -1.05543949e-01 3.07752609e-01 -5.45651078e-01
-4.80565041e-01 -6.90300882e-01 2.04049632e-01 1.10170043e+00
-1.06243026e+00 -1.56814277e+00 -6.31708652e-02 1.90290108e-01
8.85240436e-02 -4.46964949e-01 9.24638391e-01 7.83176839e-01
-3.23452622e-01 8.27463269e-01 -2.93040991e-01 1.32462054e-01
1.43098998e+00 -1.49825346e+00 1.34103715e-01 6.07252717e-01
-3.05145264e-01 1.12845922e+00 1.13217378e+00 -8.11875522e-01
-7.16158032e-01 -5.71483374e-01 1.88010097e+00 -9.02203321e-01
5.60467243e-01 -7.23200142e-01 -4.64247525e-01 7.39503324e-01
3.49508494e-01 -6.76698565e-01 1.03579688e+00 6.55338645e-01
-4.05218720e-01 -2.73924381e-01 -1.23906898e+00 7.53809690e-01
1.29682505e+00 -5.03121257e-01 -8.81575644e-01 -1.60557047e-01
1.00315593e-01 -2.24527389e-01 -7.37131476e-01 3.20739955e-01
1.08239090e+00 -1.14432812e+00 8.94981027e-01 -6.64912999e-01
9.45813656e-01 6.33461699e-02 -4.45286393e-01 -1.18221390e+00
-4.40743655e-01 -9.61691365e-02 6.90028012e-01 1.82207680e+00
4.90264714e-01 -8.54452908e-01 6.68170393e-01 8.97195339e-01
4.72065434e-02 -5.08499682e-01 -7.50181258e-01 -5.21963179e-01
7.24951386e-01 -3.06351006e-01 8.36874306e-01 1.22696102e+00
-2.82096356e-01 3.04065913e-01 -3.88131440e-01 -3.96386623e-01
5.13801813e-01 2.26573236e-02 1.00564849e+00 -1.38322222e+00
1.87598854e-01 -4.89154249e-01 -3.98382783e-01 8.16632584e-02
2.44277045e-01 -9.22456443e-01 -2.01445594e-01 -1.28902650e+00
5.87762117e-01 -5.84313214e-01 -1.07336812e-01 4.33340341e-01
-8.43543410e-02 5.93925714e-01 1.70635596e-01 1.97063208e-01
-3.33542228e-02 -4.27916236e-02 1.00117862e+00 -1.36882856e-01
1.53946012e-01 -1.84519202e-01 -1.50243998e+00 1.20029807e+00
8.09943199e-01 -8.74337792e-01 -1.19954936e-01 -3.30142736e-01
6.88068688e-01 -8.58606935e-01 3.86017919e-01 -7.56783783e-01
-3.77722263e-01 -4.82905090e-01 5.89021266e-01 -4.75955784e-01
-2.62763165e-02 -7.58260071e-01 -1.32136837e-01 5.30892670e-01
-7.64031932e-02 2.50419855e-01 -8.25739652e-02 -2.33178318e-01
-2.91308761e-01 -3.21460664e-01 6.18264079e-01 -3.39066565e-01
-5.12939870e-01 -3.06178480e-01 -5.15606046e-01 3.74420643e-01
5.95815957e-01 -3.20764542e-01 -3.22178602e-01 -1.17009587e-03
-1.68389529e-01 -2.91829109e-02 8.62946570e-01 6.47865117e-01
-1.15176871e-01 -1.40698969e+00 -8.52909267e-01 1.54339418e-01
3.30229372e-01 -6.99843824e-01 -2.27058902e-01 6.74852848e-01
-5.43602884e-01 4.30707216e-01 -6.60371363e-01 -1.67290807e-01
-1.63145876e+00 1.12518109e-01 2.17287969e-02 -1.85297400e-01
1.83114409e-01 5.64061105e-01 5.46784699e-02 -1.13413525e+00
-1.35824131e-02 -8.38099942e-02 -6.26299262e-01 6.44527137e-01
3.98400873e-02 8.64244580e-01 4.94637340e-02 -1.16187954e+00
-7.92618275e-01 2.52335936e-01 -3.94055573e-03 -4.70121086e-01
1.19127810e+00 -3.02007258e-01 -7.16356754e-01 7.75755107e-01
6.77055478e-01 8.40518236e-01 -2.76003152e-01 4.60049301e-01
2.93761402e-01 -8.33576202e-01 -2.76894659e-01 -8.64046037e-01
-5.06652892e-01 6.09811306e-01 7.19581902e-01 1.90404564e-01
7.31619775e-01 -2.42594853e-01 3.99638504e-01 7.31046870e-02
4.13422704e-01 -1.99205363e+00 -4.96042609e-01 8.49266291e-01
8.31243932e-01 -1.20668960e+00 1.78194195e-01 -2.90532529e-01
-6.08665228e-01 6.44060850e-01 7.04211116e-01 1.01162411e-01
6.52494490e-01 2.38833502e-01 7.22489774e-01 -2.14533567e-01
-2.07973734e-01 -7.08487853e-02 6.22038171e-02 6.98288262e-01
1.38817632e+00 3.96327168e-01 -1.45410836e+00 9.87020075e-01
-1.07890368e+00 -1.07186519e-01 5.84784925e-01 1.07407689e+00
-1.78579077e-01 -1.56405306e+00 -6.97289765e-01 4.52979237e-01
-1.19060338e+00 7.53323361e-02 -7.85765886e-01 1.26162481e+00
7.95454800e-01 9.94536996e-01 -7.38013238e-02 -3.90085459e-01
5.84836841e-01 3.08539540e-01 5.92674732e-01 -5.23479164e-01
-9.56392169e-01 -2.34923586e-01 6.33764207e-01 -2.05172569e-01
-7.27673411e-01 -7.93468177e-01 -1.04166079e+00 -8.05484056e-01
2.95634300e-01 -1.07102402e-01 7.95921624e-01 6.56139612e-01
-1.93512037e-01 3.82937759e-01 1.66189954e-01 -5.40351510e-01
-5.85115373e-01 -1.06745112e+00 -8.11299801e-01 6.44431293e-01
-2.32703894e-01 -8.66724491e-01 -1.74403414e-01 2.21061899e-04] | [9.390625, 10.220355987548828] |
d1f3bae6-2f89-40fb-b5ad-fdbbb9e8d0e2 | title-guided-encoding-for-keyphrase | 1808.08575 | null | http://arxiv.org/abs/1808.08575v5 | http://arxiv.org/pdf/1808.08575v5.pdf | Title-Guided Encoding for Keyphrase Generation | Keyphrase generation (KG) aims to generate a set of keyphrases given a
document, which is a fundamental task in natural language processing (NLP).
Most previous methods solve this problem in an extractive manner, while
recently, several attempts are made under the generative setting using deep
neural networks. However, the state-of-the-art generative methods simply treat
the document title and the document main body equally, ignoring the leading
role of the title to the overall document. To solve this problem, we introduce
a new model called Title-Guided Network (TG-Net) for automatic keyphrase
generation task based on the encoder-decoder architecture with two new
features: (i) the title is additionally employed as a query-like input, and
(ii) a title-guided encoder gathers the relevant information from the title to
each word in the document. Experiments on a range of KG datasets demonstrate
that our model outperforms the state-of-the-art models with a large margin,
especially for documents with either very low or very high title length ratios. | ['Yifan Gao', 'Michael R. Lyu', 'Irwin King', 'Wang Chen', 'Jiani Zhang'] | 2018-08-26 | null | null | null | null | ['keyphrase-generation'] | ['natural-language-processing'] | [ 3.18242818e-01 1.03335008e-01 -4.13372576e-01 3.97991985e-02
-9.50038850e-01 -6.43163741e-01 1.15964234e+00 2.51355678e-01
-3.40321660e-01 8.17534029e-01 8.24148059e-01 -3.25407475e-01
1.24507695e-01 -1.09775746e+00 -8.70889723e-01 -6.70473278e-01
2.93774188e-01 5.45325935e-01 8.57175067e-02 -2.99156904e-01
4.71883923e-01 8.83292481e-02 -1.16113818e+00 3.94222111e-01
9.11232948e-01 8.69222581e-01 2.70873517e-01 5.71274459e-01
-5.15408576e-01 8.49246919e-01 -7.59325445e-01 -6.37938321e-01
2.44009435e-01 -6.94025576e-01 -9.23726380e-01 6.34377152e-02
3.60511184e-01 -5.77742338e-01 -5.91245770e-01 9.66872156e-01
4.94985789e-01 1.70818530e-02 7.39616573e-01 -1.25957978e+00
-9.90701318e-01 1.24362171e+00 -3.99875104e-01 -7.86996037e-02
3.51796210e-01 -9.62310210e-02 1.42128015e+00 -9.71635342e-01
7.57133663e-01 1.05756307e+00 1.01137944e-01 3.38687658e-01
-1.03898621e+00 -5.17350495e-01 1.76787257e-01 -5.64119080e-03
-1.31434560e+00 -1.86123952e-01 9.54457164e-01 -1.34279922e-01
8.08263421e-01 -5.20569645e-02 5.55765092e-01 1.13265741e+00
4.64365721e-01 9.64848578e-01 7.20798552e-01 -4.13767546e-01
1.33655947e-02 3.04861758e-02 -4.34223488e-02 5.60434222e-01
2.98505962e-01 -2.51886785e-01 -6.98360443e-01 -1.75578192e-01
5.18560946e-01 -5.32866977e-02 -2.71833658e-01 -1.31374389e-01
-1.52690899e+00 9.72487211e-01 2.59048402e-01 2.38243192e-01
-5.92694640e-01 1.67562693e-01 3.74094039e-01 1.98475927e-01
4.16931123e-01 6.13575459e-01 -2.73727566e-01 2.86138104e-03
-1.25704861e+00 8.88826549e-01 1.14272177e+00 1.08841360e+00
5.56879520e-01 -4.41871025e-02 -7.65795529e-01 5.28093755e-01
2.04844996e-01 3.75587523e-01 4.53390628e-01 -2.25174159e-01
8.82763803e-01 6.83947146e-01 9.20807645e-02 -8.63247693e-01
5.12142777e-02 -6.53333306e-01 -1.05197489e+00 -4.55520481e-01
1.01582631e-01 -3.24818254e-01 -1.12352931e+00 1.58950913e+00
9.58686322e-02 -3.60925168e-01 1.48870930e-01 5.41944385e-01
1.06745708e+00 1.19464874e+00 -1.58151343e-01 -2.90390581e-01
1.44094014e+00 -1.02346909e+00 -9.10758197e-01 -2.02933192e-01
3.53404403e-01 -9.31014299e-01 7.99532771e-01 2.31934220e-01
-1.02565360e+00 -3.79657716e-01 -9.34385359e-01 -3.80433828e-01
-5.66224337e-01 4.11349326e-01 4.91862237e-01 1.83013499e-01
-8.99379492e-01 3.54742795e-01 -3.10428530e-01 -8.42478648e-02
2.09625393e-01 9.59003866e-02 -1.38833463e-01 -5.56334928e-02
-1.53381717e+00 4.95144814e-01 8.05208087e-01 -8.85042101e-02
-9.05102491e-01 -7.66025722e-01 -8.21759343e-01 3.82995963e-01
8.11866820e-01 -1.05611932e+00 1.39936960e+00 -3.30159903e-01
-1.36122644e+00 4.94046956e-01 -2.67275035e-01 -4.93478775e-01
4.33460563e-01 -5.52318335e-01 -1.88258976e-01 2.62319744e-01
3.07336122e-01 7.71243989e-01 9.97695386e-01 -1.26868582e+00
-6.84002519e-01 -6.31160438e-02 2.15259150e-01 3.25962961e-01
-3.10029626e-01 4.08494622e-02 -7.14001298e-01 -1.16748548e+00
-1.02388807e-01 -7.59229541e-01 -5.69450222e-02 -4.01056796e-01
-1.05337012e+00 -4.13611919e-01 6.88786864e-01 -7.69717813e-01
1.63216913e+00 -1.84768784e+00 2.73632795e-01 5.91258518e-02
2.17619121e-01 1.95117921e-01 1.07688047e-02 1.13231409e+00
3.29483449e-02 2.45167956e-01 -9.94379744e-02 -2.58036435e-01
2.37324730e-01 -4.84295674e-02 -1.05200291e+00 -1.86952725e-01
1.15288854e-01 1.16802466e+00 -9.82119918e-01 -5.48640311e-01
-3.09773386e-01 2.65212625e-01 -4.89699483e-01 5.11614621e-01
-6.31790876e-01 -3.61092724e-02 -5.76512814e-01 3.85017842e-01
3.90598357e-01 -2.91607410e-01 -1.01958662e-01 -2.31315508e-01
-3.66056450e-02 8.50796223e-01 -1.17648220e+00 1.52692866e+00
-2.05347985e-01 3.55543941e-01 -1.80799782e-01 -5.61034262e-01
8.33683848e-01 5.44201732e-01 3.04834217e-01 -4.09664392e-01
-6.72661960e-02 1.72651455e-01 -1.79819167e-01 -1.01271309e-01
1.16036451e+00 2.78048888e-02 -1.92270562e-01 9.26554561e-01
1.34271324e-01 -2.79659659e-01 7.25110292e-01 8.81656349e-01
8.59219074e-01 -7.87822306e-02 3.35406691e-01 2.66200788e-02
3.67322654e-01 -1.93334237e-01 2.00966954e-01 1.01645172e+00
5.14143467e-01 6.65054739e-01 8.74338686e-01 -2.45400086e-01
-1.08882475e+00 -7.70429790e-01 3.69432330e-01 9.21888351e-01
-9.55771804e-02 -9.53947604e-01 -8.10881853e-01 -8.07112277e-01
-9.51127857e-02 8.78500760e-01 -6.57287300e-01 -2.82638669e-01
-5.07254303e-01 -5.61113477e-01 5.36510468e-01 5.67382097e-01
6.77956820e-01 -1.21166360e+00 -2.55553573e-01 3.73934418e-01
-4.03055400e-01 -1.11350822e+00 -9.08379197e-01 2.25783572e-01
-4.86509502e-01 -7.03100562e-01 -1.12045550e+00 -6.58675551e-01
6.92075729e-01 2.30217159e-01 1.06505048e+00 -1.86715588e-01
9.53120589e-02 7.98601210e-02 -5.30774355e-01 -4.43729579e-01
-5.94087899e-01 5.99931002e-01 -2.73762137e-01 1.16810367e-01
1.34377912e-01 -2.48315498e-01 -3.10121119e-01 -2.56301612e-01
-1.35000622e+00 4.67472017e-01 9.14782226e-01 9.16937113e-01
4.54676121e-01 4.15752053e-01 4.45386380e-01 -1.04326177e+00
1.29004455e+00 -3.66537690e-01 -4.77466017e-01 3.96030962e-01
-7.35009730e-01 5.85834086e-01 8.27333450e-01 -3.71163487e-01
-1.07347167e+00 -2.85788357e-01 -9.06239823e-02 -1.93657711e-01
1.00071870e-01 8.52970481e-01 -4.72357363e-01 5.98808050e-01
2.67498493e-01 9.78303254e-01 -4.02640790e-01 -4.87610340e-01
5.53612351e-01 5.23434401e-01 6.14325345e-01 -6.05425000e-01
1.16479027e+00 1.13913901e-01 -8.63259360e-02 -5.70613146e-01
-1.02143073e+00 -3.55468184e-01 -6.07042432e-01 1.11467488e-01
5.61282516e-01 -8.79898667e-01 -1.81405962e-01 6.06799781e-01
-1.51505220e+00 1.48531552e-02 -2.89492339e-01 1.94057971e-01
-2.74518967e-01 3.47851157e-01 -5.11513352e-01 -4.94172961e-01
-9.79910731e-01 -1.15069520e+00 1.34024847e+00 2.85556018e-01
-2.58940101e-01 -8.56446326e-01 -3.58551852e-02 6.05040230e-02
2.46978000e-01 2.79760920e-02 1.37208259e+00 -8.68185461e-01
-7.42379427e-01 -3.09904188e-01 -3.26570004e-01 1.91284969e-01
3.18605989e-01 7.79873803e-02 -5.47658801e-01 -6.98003843e-02
-2.94027328e-01 -2.54354328e-01 1.18163049e+00 2.67615300e-02
9.92458701e-01 -9.11853790e-01 -3.04893464e-01 4.43859518e-01
1.15880883e+00 1.55089542e-01 5.70651591e-01 2.18487889e-01
8.33103716e-01 4.02081937e-01 2.85275966e-01 3.62939537e-01
6.43633604e-01 4.80880558e-01 3.03305537e-01 5.12568988e-02
-1.52164027e-01 -8.28626931e-01 3.16698372e-01 7.90445983e-01
1.90748170e-01 -6.76998913e-01 -6.41273677e-01 6.29282832e-01
-1.66232014e+00 -1.07751167e+00 1.82213336e-01 2.05104733e+00
1.35303748e+00 4.15544629e-01 -5.12726754e-02 2.04439729e-01
4.53991354e-01 6.81751728e-01 -2.35389024e-01 -1.77378446e-01
-3.82615253e-02 1.12463146e-01 2.38460556e-01 1.85507596e-01
-1.09333813e+00 1.13259089e+00 6.05325031e+00 8.16411078e-01
-1.09110081e+00 -4.28279549e-01 3.31107080e-01 8.77540186e-02
-6.03704453e-01 1.78048149e-01 -1.40124238e+00 5.17718375e-01
6.09640479e-01 -5.06212771e-01 2.07039639e-01 9.06070471e-01
-4.05178331e-02 -5.40999812e-04 -1.16016662e+00 7.03607142e-01
2.40487143e-01 -1.48726571e+00 8.46921384e-01 1.42877936e-01
7.42580652e-01 -3.37945491e-01 3.23666027e-03 3.21995944e-01
2.94920981e-01 -7.86335409e-01 8.50689113e-01 4.63439882e-01
6.78079307e-01 -8.28691006e-01 6.27271295e-01 4.02994603e-01
-1.09937799e+00 1.83777750e-01 -2.38655329e-01 2.62776166e-01
2.54328936e-01 8.67582321e-01 -1.00474274e+00 7.01978743e-01
2.88478434e-01 5.53621590e-01 -5.19302011e-01 8.45113516e-01
-7.53068209e-01 6.52019441e-01 -4.27243412e-02 -2.46264234e-01
6.24321520e-01 1.57893617e-02 6.40562296e-01 1.30920613e+00
3.04140747e-01 -3.10095642e-02 1.89080596e-01 1.16293931e+00
-6.38055682e-01 1.88271955e-01 -5.26631236e-01 -7.29387879e-01
3.76820952e-01 1.28838062e+00 -8.10395539e-01 -5.67176998e-01
-1.40165702e-01 9.43297148e-01 1.16915204e-01 5.00878155e-01
-4.47938710e-01 -8.66317332e-01 2.93698132e-01 8.06130767e-02
4.63818938e-01 -3.40873450e-01 -4.82600555e-02 -1.17302334e+00
2.35149026e-01 -1.11971223e+00 2.14520052e-01 -8.43585968e-01
-9.05033767e-01 5.00964999e-01 1.54217049e-01 -1.13172936e+00
-6.29458070e-01 -2.08624795e-01 -6.06366754e-01 9.93708253e-01
-1.48862481e+00 -1.32628977e+00 -4.13346104e-02 3.05947244e-01
6.74008727e-01 -1.61185563e-01 7.39616275e-01 -5.22339940e-02
-3.05543602e-01 4.51374114e-01 5.60226180e-02 5.83243966e-01
7.17471838e-01 -1.37488043e+00 7.50586212e-01 1.12482619e+00
3.97544146e-01 1.06912649e+00 6.12946928e-01 -8.80177855e-01
-1.51891398e+00 -1.11588502e+00 1.47575009e+00 -1.79282859e-01
6.31345630e-01 -6.68124259e-01 -8.24600518e-01 5.21297395e-01
6.91586435e-01 -4.81373519e-01 5.36465704e-01 -1.33646101e-01
-4.56643283e-01 -3.60413222e-03 -4.13681328e-01 9.06314075e-01
6.50143683e-01 -5.50610244e-01 -8.15425158e-01 3.76814544e-01
1.14788520e+00 -6.11425877e-01 -4.64683771e-01 1.67000622e-01
5.41237414e-01 -4.99471456e-01 7.08272040e-01 -5.09998798e-01
1.06246400e+00 -2.42620826e-01 1.28234655e-01 -1.58957171e+00
-3.10115278e-01 -9.30467546e-01 -6.23012602e-01 1.40936077e+00
4.12017405e-01 -2.06351027e-01 4.19634074e-01 2.95815587e-01
-1.00817323e-01 -8.76759648e-01 -4.68807995e-01 -5.88812947e-01
-1.07731700e-01 -2.04599231e-01 8.34719419e-01 5.32910645e-01
-1.15409926e-01 9.13628697e-01 -6.13019049e-01 -1.74181357e-01
2.97452629e-01 3.89406979e-01 9.99643028e-01 -9.77865458e-01
-2.20215157e-01 -5.18660367e-01 2.90797263e-01 -1.43123925e+00
7.51551660e-03 -8.39037478e-01 7.11767077e-02 -2.06573439e+00
3.50696921e-01 3.31053659e-02 1.70408398e-01 5.79279363e-01
-5.25099993e-01 -3.36058497e-01 2.43811771e-01 3.46769363e-01
-3.55902553e-01 7.60878921e-01 1.44774795e+00 -3.71251523e-01
-1.59753010e-01 1.02809049e-01 -1.12647545e+00 3.51694733e-01
3.75679821e-01 -5.28580904e-01 -5.56853473e-01 -3.51570874e-01
5.50633550e-01 7.91956484e-03 1.88579708e-01 -6.68959796e-01
5.52237332e-01 -2.60751218e-01 3.86513412e-01 -9.92345691e-01
9.72236991e-02 -4.66515243e-01 -1.85139716e-01 3.01811934e-01
-7.33542860e-01 9.98716950e-02 -2.67422404e-02 4.12273079e-01
-3.31487447e-01 -3.27614933e-01 2.80742496e-01 -2.95119405e-01
-3.92627299e-01 4.67091084e-01 -4.25437719e-01 2.48125792e-01
4.94719654e-01 1.83470950e-01 -3.95008832e-01 -7.29417741e-01
3.99274230e-02 1.02109492e-01 -1.14947965e-04 5.87028563e-01
6.30472064e-01 -1.17919850e+00 -8.97713065e-01 2.59838939e-01
2.62083143e-01 2.11038649e-01 -2.77750343e-01 4.44320053e-01
-3.30721319e-01 9.39885020e-01 2.01528892e-01 1.95465107e-02
-9.06211913e-01 5.54217637e-01 -1.35764554e-01 -8.48750055e-01
-5.90168953e-01 7.77489483e-01 3.59018683e-01 -8.46803412e-02
3.03259075e-01 -5.30105472e-01 -2.52225697e-01 3.67427349e-01
7.43064106e-01 -1.03649847e-01 1.55612960e-01 -4.68766302e-01
1.47906199e-01 1.29398704e-01 -7.88994133e-01 -2.88259864e-01
1.13810897e+00 2.14617312e-01 -2.31411859e-01 2.70257860e-01
1.05909693e+00 4.18487042e-02 -8.94318998e-01 -6.31789386e-01
-1.40764937e-01 -1.07448108e-01 1.40103444e-01 -7.53552496e-01
-8.20456982e-01 8.22049677e-01 -3.04771662e-01 1.84032768e-01
9.79655206e-01 -3.01895719e-02 1.14003289e+00 6.14698946e-01
1.76237509e-01 -1.12625682e+00 3.68962973e-01 7.77105808e-01
1.10157204e+00 -9.75129962e-01 2.41118163e-01 -2.55118430e-01
-6.46057844e-01 1.12095308e+00 3.55838716e-01 1.98122218e-01
4.99798149e-01 1.84592754e-01 -1.80743307e-01 -1.58985123e-01
-8.50389600e-01 -1.04882315e-01 6.16020143e-01 1.61157757e-01
4.61161166e-01 -3.05954665e-01 -4.47756797e-01 7.00710356e-01
-5.48738003e-01 -9.20698512e-03 5.06631196e-01 1.05633950e+00
-3.93191785e-01 -1.17086709e+00 -2.65970677e-01 6.26302361e-01
-6.85223997e-01 -6.12192988e-01 -8.06707382e-01 6.70501709e-01
-1.48733839e-01 6.91888750e-01 -1.89072385e-01 -2.96975940e-01
1.18217915e-01 3.03136110e-01 1.50963396e-01 -8.85594666e-01
-7.12175071e-01 2.77002186e-01 -7.64124319e-02 -1.85834467e-01
-9.39822644e-02 -3.44232768e-01 -1.11410952e+00 -2.09998667e-01
-3.17725420e-01 5.16632617e-01 5.86652756e-01 1.01251602e+00
4.60283786e-01 6.82364762e-01 5.74322641e-01 -5.36112726e-01
-7.93817878e-01 -1.20222187e+00 -6.73281133e-01 2.76875645e-01
3.23122591e-01 -2.67358005e-01 -1.36357501e-01 4.53918815e-01] | [12.312376976013184, 8.927029609680176] |
1738cff4-7976-4eb1-9da8-18f2f7c6f98f | bayesian-knowledge-driven-critiquing-with | 2306.05636 | null | https://arxiv.org/abs/2306.05636v1 | https://arxiv.org/pdf/2306.05636v1.pdf | Bayesian Knowledge-driven Critiquing with Indirect Evidence | Conversational recommender systems (CRS) enhance the expressivity and personalization of recommendations through multiple turns of user-system interaction. Critiquing is a well-known paradigm for CRS that allows users to iteratively refine recommendations by providing feedback about attributes of recommended items. While existing critiquing methodologies utilize direct attributes of items to address user requests such as 'I prefer Western movies', the opportunity of incorporating richer contextual and side information about items stored in Knowledge Graphs (KG) into the critiquing paradigm has been overlooked. Employing this substantial knowledge together with a well-established reasoning methodology paves the way for critique-based recommenders to allow for complex knowledge-based feedback (e.g., 'I like movies featuring war side effects on veterans') which may arise in natural user-system conversations. In this work, we aim to increase the flexibility of critique-based recommendation by integrating KGs and propose a novel Bayesian inference framework that enables reasoning with relational knowledge-based feedback. We study and formulate the framework considering a Gaussian likelihood and evaluate it on two well-known recommendation datasets with KGs. Our evaluations demonstrate the effectiveness of our framework in leveraging indirect KG-based feedback (i.e., preferred relational properties of items rather than preferred items themselves), often improving personalized recommendations over a one-shot recommender by more than 15%. This work enables a new paradigm for using rich knowledge content and reasoning over indirect evidence as a mechanism for critiquing interactions with CRS. | ['Scott Sanner', 'Zhenwei Tang', 'Griffin Floto', 'Armin Toroghi'] | 2023-06-09 | null | null | null | null | ['knowledge-graphs', 'bayesian-inference'] | ['knowledge-base', 'methodology'] | [-1.76301911e-01 4.45992768e-01 -3.07422876e-01 -4.24540043e-01
-3.00877512e-01 -7.73151457e-01 6.96416497e-01 3.88037384e-01
-3.66066754e-01 7.73440301e-01 9.05971587e-01 -4.59432751e-01
-9.51134384e-01 -8.35806787e-01 -5.56816876e-01 -3.73435915e-01
3.29735912e-02 6.15752518e-01 1.83791265e-01 -5.69890320e-01
5.51893711e-01 3.28422427e-01 -1.73767722e+00 6.02700710e-01
7.99830377e-01 7.13011324e-01 1.97236225e-01 6.27153575e-01
-8.18720236e-02 7.34981358e-01 -4.40726161e-01 -8.31684113e-01
-4.21336219e-02 3.37359495e-02 -6.92163467e-01 -1.59971431e-01
2.30514094e-01 -4.63267446e-01 4.01639268e-02 6.09799683e-01
1.94827452e-01 8.47772181e-01 8.48730445e-01 -8.78299117e-01
-8.89520526e-01 1.16029656e+00 2.03886271e-01 1.59185603e-01
8.14868271e-01 -1.47377431e-01 1.40129483e+00 -9.02511895e-01
5.07372141e-01 1.17106104e+00 6.27878547e-01 4.80099916e-01
-1.07865477e+00 -2.09294021e-01 6.51252925e-01 4.33345258e-01
-1.17627168e+00 2.08267290e-02 4.69967365e-01 -6.33565426e-01
7.92669952e-01 8.74688268e-01 6.20039225e-01 1.08832014e+00
-3.28987688e-02 4.82849628e-01 8.30438673e-01 -9.54501927e-02
5.33881903e-01 9.09096241e-01 5.50458908e-01 3.60763311e-01
2.45915040e-01 9.09676105e-02 -6.53931975e-01 -4.93974149e-01
3.38093638e-01 3.46876532e-01 -3.13768208e-01 -4.31830436e-02
-8.45050335e-01 8.90603483e-01 2.89078414e-01 9.49527770e-02
-6.42158985e-01 -2.88336247e-01 2.08299682e-02 5.37358597e-02
6.43766284e-01 8.46640050e-01 -5.09980142e-01 -2.50595152e-01
-6.28080487e-01 4.25154537e-01 1.16176999e+00 9.73942220e-01
6.82015181e-01 -3.98517787e-01 -6.19358838e-01 8.58793974e-01
7.01375246e-01 4.23320413e-01 6.93680644e-02 -8.08569312e-01
1.97843593e-02 6.29857898e-01 4.58469808e-01 -1.25736988e+00
-4.74963963e-01 -6.76547348e-01 -2.44701236e-01 -2.36597016e-01
2.42746890e-01 -2.29721278e-01 -2.56533861e-01 1.55590641e+00
5.75405896e-01 1.47308379e-01 1.91631373e-02 9.53509510e-01
8.72766972e-01 3.71800750e-01 3.02020349e-02 -4.46541786e-01
1.51956940e+00 -7.64124572e-01 -7.05694854e-01 4.15226817e-01
5.76417685e-01 -5.72312236e-01 1.23369396e+00 9.00356650e-01
-6.90220237e-01 -2.91575700e-01 -6.03938997e-01 3.55401188e-01
-4.52776730e-01 -1.19477242e-01 7.03137457e-01 9.14416373e-01
-9.19712901e-01 6.67562425e-01 -1.75082505e-01 -4.08015698e-01
-5.55773228e-02 2.74736106e-01 -2.86847558e-02 2.58077830e-02
-1.51244521e+00 9.69566882e-01 -1.00767231e-02 3.98409888e-02
-5.43039560e-01 -1.04068077e+00 -3.82695466e-01 2.69170374e-01
1.02177382e+00 -9.13778722e-01 1.24751747e+00 -4.07645166e-01
-1.78217137e+00 -2.90613592e-01 1.70398787e-01 -2.05060557e-01
1.67617545e-01 -6.66104376e-01 -6.97607040e-01 1.02921806e-01
-2.68128842e-01 -1.41155720e-02 7.13546574e-01 -1.27332759e+00
-3.90652239e-01 -4.31438237e-02 7.29340434e-01 3.30136567e-01
-4.63580698e-01 -6.48872703e-02 -1.54898807e-01 -4.94998455e-01
-5.16694248e-01 -9.12789762e-01 -3.06394309e-01 -4.70217735e-01
-4.45017427e-01 -5.35969734e-01 3.92265111e-01 -3.14884156e-01
1.69452345e+00 -1.77088094e+00 1.79375455e-01 6.23731792e-01
2.46947005e-01 2.69503266e-01 1.17736854e-01 8.95654142e-01
3.83171767e-01 2.56046414e-01 3.67402583e-01 -1.53990939e-01
2.40992263e-01 3.39667857e-01 -5.22630274e-01 -9.37958527e-03
-2.21148401e-01 5.87253511e-01 -1.08348274e+00 -1.25484273e-01
1.48151129e-01 6.75393283e-01 -1.18758059e+00 3.14555496e-01
-5.93016982e-01 3.03865135e-01 -4.78129148e-01 1.77477926e-01
3.37932974e-01 -4.06832397e-01 4.01438475e-01 -5.29592276e-01
-1.00343429e-01 5.37007928e-01 -1.28232241e+00 1.13858485e+00
-5.40062129e-01 -1.26770988e-01 -3.59340996e-01 -5.03087580e-01
7.67184973e-01 2.64428645e-01 2.20923215e-01 -1.66250974e-01
8.32460448e-02 -4.13987249e-01 -2.79925078e-01 -6.16740167e-01
9.98174191e-01 -1.74839392e-01 8.92962366e-02 8.93887877e-01
-2.70187836e-02 1.13727972e-01 -1.07427258e-02 8.96441102e-01
1.04969430e+00 2.33579114e-01 2.71185666e-01 -3.70418847e-01
4.55067307e-01 -2.57533550e-01 1.57416761e-01 9.82811332e-01
5.10250449e-01 1.99164912e-01 1.56281367e-01 -7.19425380e-02
-3.63539189e-01 -9.50715959e-01 -6.91246102e-03 1.49676657e+00
5.33119217e-02 -1.09462142e+00 -4.66021150e-01 -9.15216625e-01
7.17994422e-02 1.25944030e+00 -8.63248706e-01 -1.47898436e-01
-1.49672050e-02 -4.44261521e-01 1.43893257e-01 3.78233612e-01
-1.14629328e-01 -6.45728528e-01 -2.99592137e-01 2.95842052e-01
-2.93918937e-01 -7.72590160e-01 -5.41628599e-01 -1.25214934e-01
-7.59620905e-01 -9.19740379e-01 -3.13933015e-01 3.60845238e-01
5.31834245e-01 3.99467975e-01 1.05636156e+00 2.42543980e-01
2.81382114e-01 1.22446418e+00 -8.55012000e-01 -1.01770110e-01
-1.89669028e-01 -2.13466927e-01 6.46052897e-01 2.76913494e-01
2.09078878e-01 -7.34086812e-01 -8.18615019e-01 7.09519506e-01
-8.58359814e-01 -1.18633121e-01 3.64394039e-01 6.42263353e-01
1.56039089e-01 -1.43696025e-01 8.70095134e-01 -1.52602506e+00
1.06418192e+00 -1.21281600e+00 -1.69806629e-01 4.37256366e-01
-1.25112474e+00 1.93727046e-01 5.94583869e-01 -6.51831627e-01
-1.57697368e+00 -6.82564914e-01 -1.44883320e-01 -6.17231766e-04
-9.92172211e-02 1.02172971e+00 2.05348656e-01 2.24778026e-01
1.10846543e+00 -1.86163470e-01 -3.42769027e-01 -6.97388530e-01
9.68180776e-01 6.94224179e-01 3.36951792e-01 -1.03554428e+00
4.84205633e-01 2.65002638e-01 -2.93954998e-01 -5.02614498e-01
-1.15157676e+00 -7.71184266e-01 -1.33668214e-01 -6.36318982e-01
4.86060739e-01 -5.29526889e-01 -1.38120055e+00 -5.08665860e-01
-8.10835361e-01 -2.05340213e-03 -3.66414517e-01 7.94903457e-01
-4.69361424e-01 5.23809493e-01 -4.26112205e-01 -1.23128927e+00
-2.81620979e-01 -7.51991749e-01 4.60510969e-01 2.77267516e-01
-5.61744153e-01 -1.20925009e+00 3.01062614e-01 5.40446997e-01
7.60776937e-01 -4.43462908e-01 7.79282272e-01 -1.13928461e+00
-4.65103388e-01 6.21250309e-02 2.74355933e-02 1.20582812e-01
1.59248322e-01 -8.82231519e-02 -9.03357685e-01 2.36586016e-02
-1.95025355e-01 2.45909654e-02 3.96740198e-01 -2.94022821e-02
7.64467955e-01 -7.66771495e-01 -1.98116019e-01 -4.15235087e-02
9.34973300e-01 -3.60821486e-02 5.12302637e-01 -1.32953703e-01
4.87023920e-01 9.37094569e-01 9.18330729e-01 9.46170807e-01
6.68412864e-01 8.70930076e-01 4.13501769e-01 5.69274604e-01
1.10374980e-01 -4.24223334e-01 3.58154029e-01 7.64371753e-01
-5.99551260e-01 -3.28411758e-01 -3.53722930e-01 1.15129776e-01
-2.05930305e+00 -1.12711656e+00 -1.74854234e-01 2.14780045e+00
9.13370490e-01 -8.73223767e-02 1.82969615e-01 -1.36575907e-01
4.71538067e-01 -3.76026660e-01 -3.19575787e-01 -4.77944076e-01
3.55443776e-01 -4.34536971e-02 9.31130126e-02 7.58491039e-01
-4.06232506e-01 6.05596423e-01 5.84560251e+00 6.00227594e-01
-5.24170101e-01 5.12726279e-03 4.49234545e-02 -1.20544285e-01
-1.06766307e+00 8.15090388e-02 -1.09125984e+00 2.84191549e-01
1.16190100e+00 -1.13789164e-01 7.52081752e-01 9.10411417e-01
3.44223350e-01 -3.75377119e-01 -1.12021112e+00 6.84244335e-01
1.59315020e-01 -1.49773061e+00 1.38188779e-01 1.07936203e-01
6.00167930e-01 -3.45530629e-01 4.23680134e-02 4.95775044e-01
8.23297441e-01 -6.03855968e-01 4.28570777e-01 1.21396267e+00
2.69396961e-01 -5.23203492e-01 6.84438825e-01 4.81616050e-01
-8.12608600e-01 -2.01587796e-01 -2.18860582e-01 -3.24142963e-01
3.21822196e-01 8.01956654e-01 -1.24663746e+00 9.02815759e-01
6.50348067e-01 8.45342398e-01 -3.46992910e-01 8.69560063e-01
-2.19049662e-01 8.94736409e-01 -3.35129708e-01 -4.45971102e-01
-7.22120628e-02 -3.32725525e-01 6.77561402e-01 1.31549466e+00
3.64896268e-01 5.89939356e-01 -4.71157879e-02 7.21043587e-01
2.77839601e-01 4.39860970e-01 -4.76602256e-01 1.59082133e-02
5.82046092e-01 1.47565222e+00 -4.30463135e-01 -4.54867929e-01
-2.65543312e-01 3.94317716e-01 3.15583020e-01 3.47652406e-01
-5.84418952e-01 2.72234708e-01 5.98269403e-01 3.26283544e-01
3.83373827e-01 1.93385687e-02 5.01084179e-02 -1.10960686e+00
-4.78547961e-01 -8.40755641e-01 5.65003872e-01 -8.24639201e-01
-1.70606923e+00 5.45665383e-01 4.77858543e-01 -1.21873486e+00
-2.52610832e-01 -3.30417365e-01 -4.56795901e-01 7.22551346e-01
-1.08509672e+00 -8.91046584e-01 -1.14661641e-01 7.39675045e-01
2.40601569e-01 1.88495088e-02 9.19942200e-01 1.83420226e-01
-3.86597663e-02 5.03184736e-01 -1.69396907e-01 -8.26333165e-01
9.06812429e-01 -1.31669509e+00 -2.88601488e-01 3.93596381e-01
2.42396876e-01 1.43825197e+00 9.64705706e-01 -9.12675083e-01
-1.55584633e+00 -6.13744020e-01 6.26190007e-01 -8.79842699e-01
6.87547624e-01 -1.24624871e-01 -9.13778424e-01 5.51008999e-01
2.18351167e-02 -5.71895599e-01 1.40826750e+00 1.04249668e+00
-6.12952828e-01 -6.87220097e-02 -1.09514117e+00 7.68485069e-01
1.00685370e+00 -5.18287301e-01 -7.46787190e-01 3.64908993e-01
8.06722581e-01 3.40795191e-03 -1.16241693e+00 2.36287445e-01
6.82340026e-01 -7.99642444e-01 9.54996347e-01 -8.64960730e-01
5.29714115e-03 -5.19516170e-01 -2.82346070e-01 -1.52557170e+00
-4.94768202e-01 -7.54408181e-01 -4.57186520e-01 1.25088263e+00
5.57472706e-01 -5.80947161e-01 4.05057907e-01 1.00353503e+00
-4.07539636e-01 -8.02742839e-01 -2.59892136e-01 -3.29414248e-01
-4.84292805e-01 -8.29513550e-01 5.82417965e-01 8.90081167e-01
7.72680223e-01 4.60367173e-01 -8.30482662e-01 4.11954373e-01
3.03324252e-01 -1.38261750e-01 7.98618257e-01 -1.58727801e+00
-8.97249937e-01 -1.61331668e-01 1.78567797e-01 -1.18550324e+00
-2.68014669e-01 -8.96153808e-01 -9.16790217e-02 -1.49736273e+00
8.17981362e-02 -7.38747120e-01 -4.05982971e-01 2.53434956e-01
-2.43155152e-01 6.09806776e-02 1.07000656e-01 1.32537661e-02
-8.71201456e-01 3.46234411e-01 1.22341883e+00 1.99575350e-01
-2.34742910e-01 4.08895880e-01 -1.30312133e+00 5.51782489e-01
5.32001495e-01 -4.33452696e-01 -9.64803100e-01 2.53517359e-01
1.29123938e+00 1.12244889e-01 1.67887688e-01 -4.16703999e-01
5.95363259e-01 -4.52468276e-01 -2.64244229e-01 -4.76956964e-01
3.96962464e-01 -8.97834718e-01 5.99514186e-01 -2.35155880e-01
-6.59216881e-01 -2.33154014e-01 5.88246919e-02 1.13748097e+00
2.78275281e-01 -3.14786851e-01 -9.54044834e-02 9.16173458e-02
-3.20922434e-01 -2.61532683e-02 -6.75476909e-01 -2.76909947e-01
5.70555449e-01 7.04698935e-02 -6.27696335e-01 -7.95704842e-01
-1.17360151e+00 1.01619340e-01 -8.84616971e-02 5.21768332e-01
6.85922861e-01 -1.07815611e+00 -4.08418000e-01 -2.05920473e-01
2.59317487e-01 -6.98238492e-01 5.21427751e-01 1.03808296e+00
3.39586824e-01 2.81510144e-01 1.34480566e-01 -3.20255309e-01
-1.15543878e+00 7.62449443e-01 -2.68746875e-02 -1.58750221e-01
-5.29786587e-01 8.69667113e-01 8.81565735e-02 -4.64092642e-01
2.98767030e-01 -3.09349924e-01 -9.17392552e-01 2.67815739e-01
8.29392374e-01 5.46108902e-01 4.26304154e-02 -1.21615328e-01
-2.36314133e-01 1.96825519e-01 -3.76006842e-01 -2.24665254e-01
1.14090109e+00 -2.88717091e-01 1.20588318e-01 5.37638366e-01
3.20335150e-01 7.42621601e-01 -9.74898398e-01 -5.12986958e-01
5.24960048e-02 -5.03910840e-01 -3.95377129e-02 -1.41678965e+00
-4.87535089e-01 3.99104118e-01 2.20989138e-02 5.49567044e-01
6.81487024e-01 9.46059972e-02 2.72259891e-01 8.05143952e-01
5.22365332e-01 -9.79625762e-01 1.88190103e-01 3.64824623e-01
9.31861699e-01 -8.93121779e-01 1.98589474e-01 -4.59532619e-01
-1.00531924e+00 1.10470259e+00 2.55731732e-01 2.38461047e-01
1.10006070e+00 9.91619751e-02 -5.87022416e-02 -3.04630131e-01
-1.16243947e+00 -2.40713760e-01 7.76208460e-01 4.63908970e-01
4.65849072e-01 3.38633358e-01 -6.74510598e-01 1.35617077e+00
-1.49187297e-01 2.48862673e-02 6.39385164e-01 4.27417785e-01
-5.23293793e-01 -1.19524908e+00 -1.40342504e-01 6.78060949e-01
-1.73348576e-01 -2.99405843e-01 -1.60967827e-01 1.09271191e-01
-5.43526150e-02 1.44281232e+00 -4.24227744e-01 -9.50912595e-01
3.19374800e-01 -2.64241174e-02 1.56849444e-01 -9.66258883e-01
-1.16619849e+00 -1.50837198e-01 7.04250991e-01 -5.37719965e-01
-4.75143731e-01 -5.89426100e-01 -9.82390165e-01 -4.22553986e-01
-7.23508596e-01 7.67233968e-01 6.21099532e-01 1.24930596e+00
6.10521913e-01 5.53339362e-01 4.95296299e-01 -5.72893560e-01
-6.67635560e-01 -9.85161364e-01 -5.02946496e-01 2.99214602e-01
-2.63415687e-02 -1.18420589e+00 -3.86026084e-01 -3.81901525e-02] | [10.020988464355469, 5.846883296966553] |
7a686ff3-8975-45fe-a325-12a23879ba50 | efficient-stereo-depth-estimation-for-pseudo | 2205.08089 | null | https://arxiv.org/abs/2205.08089v1 | https://arxiv.org/pdf/2205.08089v1.pdf | Efficient Stereo Depth Estimation for Pseudo LiDAR: A Self-Supervised Approach Based on Multi-Input ResNet Encoder | Perception and localization are essential for autonomous delivery vehicles, mostly estimated from 3D LiDAR sensors due to their precise distance measurement capability. This paper presents a strategy to obtain the real-time pseudo point cloud instead of the laser sensor from the image sensor. We propose an approach to use different depth estimators to obtain pseudo point clouds like LiDAR to obtain better performance. Moreover, the training and validating strategy of the depth estimator has adopted stereo imagery data to estimate more accurate depth estimation as well as point cloud results. Our approach to generating depth maps outperforms on KITTI benchmark while yielding point clouds significantly faster than other approaches. | ['Xianke Lin', 'Sabir Hossain'] | 2022-05-17 | null | null | null | null | ['stereo-depth-estimation'] | ['computer-vision'] | [-9.46470946e-02 -1.30166292e-01 8.91230032e-02 -7.11861134e-01
-8.15562367e-01 -5.00668883e-01 6.30562186e-01 3.88509519e-02
-5.35197914e-01 9.14055347e-01 -5.93926907e-01 -3.06633234e-01
2.13379636e-01 -1.26556742e+00 -9.46573734e-01 -4.96244162e-01
8.56501758e-02 1.17724681e+00 5.59391260e-01 -1.16098173e-01
7.01092064e-01 1.10503209e+00 -1.95890307e+00 -2.26988181e-01
1.11533475e+00 1.11364734e+00 7.30335057e-01 6.82795644e-01
-5.61993778e-01 -6.16437867e-02 -2.34246552e-01 -3.16020139e-02
5.34432232e-01 3.05382222e-01 -1.63045511e-01 6.33821785e-02
6.37347221e-01 -8.38735878e-01 1.76415429e-01 1.05522525e+00
1.80456102e-01 -3.80226582e-01 7.81730294e-01 -1.56330144e+00
1.56564098e-02 2.67836042e-02 -7.46094704e-01 -3.48758966e-01
4.87192035e-01 -4.95547615e-02 2.80328512e-01 -1.04903722e+00
6.43345594e-01 1.09592724e+00 6.38925970e-01 3.02428901e-01
-8.51884186e-01 -8.28470469e-01 -4.85051721e-01 5.30702472e-01
-1.84360993e+00 -4.10233468e-01 7.02622831e-01 -2.61260659e-01
7.83620358e-01 -1.56784117e-01 5.39298296e-01 6.23557806e-01
3.51284951e-01 1.17777601e-01 1.20635188e+00 -1.93135783e-01
3.91293436e-01 4.90303606e-01 -2.34570920e-01 6.10107541e-01
6.03216052e-01 3.87822419e-01 -1.98504716e-01 -2.65041646e-02
8.82654786e-01 6.46846667e-02 -8.65733698e-02 -8.22539032e-01
-8.25821817e-01 8.00364196e-01 5.36696076e-01 -6.93160146e-02
-5.71373880e-01 2.60545343e-01 5.61766773e-02 1.52923614e-01
1.94008678e-01 -2.87253201e-01 -4.36798066e-01 -1.40087053e-01
-8.98928106e-01 2.52381295e-01 6.39498889e-01 1.60108376e+00
1.49376643e+00 -1.77271944e-02 4.23725724e-01 2.09374085e-01
5.30133724e-01 1.06895745e+00 1.47944987e-01 -1.27379715e+00
7.12872684e-01 7.23776042e-01 3.80455047e-01 -6.76144183e-01
-1.19023353e-01 1.65160908e-03 -5.89081049e-01 8.14693630e-01
1.31136790e-01 2.50617355e-01 -1.03107154e+00 7.64877081e-01
4.30668563e-01 5.76579452e-01 3.36241126e-01 6.66054606e-01
6.55603707e-01 5.97025871e-01 -3.38610172e-01 -1.19561762e-01
9.24107909e-01 -3.56746495e-01 -3.89517933e-01 -2.86125243e-01
7.26252079e-01 -5.91266990e-01 6.18208170e-01 6.20675981e-01
-8.42503130e-01 -6.68162227e-01 -1.22358513e+00 -1.26099646e-01
-3.79093438e-01 1.34165017e-02 2.65097171e-01 5.27853608e-01
-1.39051831e+00 6.67706490e-01 -7.66260862e-01 -3.70089024e-01
3.42296511e-01 2.90401012e-01 -5.01543283e-01 -3.34668636e-01
-7.56955445e-01 1.22191310e+00 5.04759610e-01 7.12012351e-02
-7.68489897e-01 -4.33325291e-01 -8.41905832e-01 -3.12271357e-01
-6.23775609e-02 -5.67120373e-01 1.10187101e+00 -1.92157283e-01
-1.44188607e+00 7.30798423e-01 -5.86645842e-01 -7.66421378e-01
5.65697491e-01 -6.03203438e-02 1.31878152e-01 3.37788880e-01
2.53757596e-01 1.01041377e+00 5.71131885e-01 -1.67679954e+00
-8.46843600e-01 -6.51445210e-01 -3.41352105e-01 2.04675108e-01
1.97539404e-01 -7.18065619e-01 -2.93159932e-01 6.13911510e-01
8.02999437e-01 -6.14703059e-01 -3.60493332e-01 2.90092438e-01
-4.57753055e-02 -9.26232263e-02 1.11727381e+00 -1.69866532e-01
2.97066897e-01 -1.72470236e+00 -4.14064646e-01 2.00857788e-01
8.85495469e-02 8.10963381e-03 1.66414022e-01 3.71294111e-01
3.77813190e-01 -7.30442628e-02 -1.24844471e-02 -6.27830386e-01
-2.92345524e-01 7.05840528e-01 -3.34857017e-01 5.55070579e-01
-4.09883521e-02 6.13818169e-01 -7.79124737e-01 -9.43438292e-01
8.23857963e-01 6.52793646e-01 -2.48623192e-01 -4.17706855e-02
-7.31038749e-02 4.72899973e-01 -5.80583811e-01 7.65856862e-01
1.55543995e+00 3.96368682e-01 -3.18204641e-01 -1.64137170e-01
-6.73525870e-01 3.57541107e-02 -1.07527161e+00 1.76572955e+00
-6.55165076e-01 5.52228987e-01 7.74918869e-02 -5.91267228e-01
1.41241443e+00 1.50109112e-01 5.38167179e-01 -7.74149001e-01
2.46836215e-01 4.65728790e-01 -4.92767096e-01 -2.25863278e-01
8.72739315e-01 -1.49974421e-01 2.62598932e-01 -4.62172240e-01
-3.37252140e-01 -1.03506052e+00 -1.79995179e-01 9.07179862e-02
7.03414440e-01 4.03323531e-01 1.48388937e-01 -3.14586721e-02
6.33012176e-01 5.13019025e-01 4.35919046e-01 4.16381836e-01
-7.13645592e-02 6.54339552e-01 -7.36367851e-02 -2.44748250e-01
-1.34964752e+00 -1.26963580e+00 -6.48120642e-01 -1.59068704e-01
5.75591445e-01 -3.02681830e-02 -3.56502563e-01 -2.10186824e-01
4.70417649e-01 7.56657839e-01 -2.47408554e-01 4.45729405e-01
-5.47928452e-01 1.06910758e-01 1.81538180e-01 6.06156409e-01
6.19409740e-01 -4.63424563e-01 -7.12573647e-01 2.59266824e-01
1.90177724e-01 -1.22266459e+00 3.01044673e-01 6.50401637e-02
-1.32669747e+00 -9.90656495e-01 -3.03742826e-01 -3.54422778e-01
5.15374720e-01 6.21846557e-01 7.86355555e-01 1.13989525e-01
1.12884715e-01 1.35881066e-01 -3.86858016e-01 -8.30155671e-01
-4.17138219e-01 -2.96552002e-01 1.15401223e-01 -6.24863803e-01
5.07698238e-01 -5.98994672e-01 -5.48498511e-01 3.40420395e-01
-5.19877851e-01 -9.70176160e-02 5.65793753e-01 6.47039190e-02
8.54672909e-01 7.85590634e-02 6.14460781e-02 -4.42437857e-01
2.32151747e-02 -4.11246121e-01 -1.32069051e+00 -4.01561677e-01
-7.93175280e-01 -1.31334648e-01 3.31094742e-01 6.61839843e-02
-6.12754703e-01 4.21703964e-01 -3.06982517e-01 -8.13863397e-01
-2.85859972e-01 -9.04659256e-02 -9.09659266e-02 -3.39175493e-01
6.55713558e-01 3.89399111e-01 2.78656900e-01 -4.61913496e-01
1.02839880e-01 7.72688806e-01 5.16662300e-01 -2.34196380e-01
1.29949749e+00 9.50118601e-01 4.16656047e-01 -9.88632619e-01
-2.49469534e-01 -6.49960279e-01 -9.94828761e-01 -2.62058020e-01
6.22932017e-01 -1.13172638e+00 -8.16112816e-01 2.78851002e-01
-1.69322813e+00 1.40082747e-01 5.74721210e-02 5.55241823e-01
-6.74780071e-01 4.46030408e-01 -2.14200303e-01 -1.02609205e+00
-2.17330113e-01 -1.26991022e+00 1.50863397e+00 1.34032607e-01
3.77900273e-01 -7.74149537e-01 2.99817231e-02 5.28633073e-02
2.98800170e-01 3.59107584e-01 2.29842708e-01 1.59762368e-01
-1.40736973e+00 -4.06891614e-01 -5.30727804e-01 1.70759797e-01
3.97777259e-02 3.32046896e-01 -1.01166058e+00 -1.45276204e-01
1.07860073e-01 1.71501815e-01 5.29783547e-01 3.54979545e-01
8.19617450e-01 2.75166214e-01 -8.03676724e-01 6.32303178e-01
2.06433105e+00 2.18290269e-01 9.43998635e-01 4.76119548e-01
5.00954747e-01 3.79516065e-01 1.31170070e+00 3.19440693e-01
8.13160956e-01 7.18554735e-01 1.01850486e+00 2.96182275e-01
1.97523236e-01 -3.86157990e-01 1.10005304e-01 5.21748662e-01
1.39037333e-02 -2.67063472e-02 -1.03905034e+00 3.88210356e-01
-1.52510190e+00 -7.52689302e-01 -8.34410250e-01 2.34683657e+00
2.92255580e-01 2.81893462e-01 -2.57561028e-01 3.16571295e-01
4.21013057e-01 -2.98837870e-01 -2.95517206e-01 -5.16438007e-01
2.32934013e-01 2.44560197e-01 1.20458496e+00 8.61162066e-01
-4.53740597e-01 1.07678056e+00 6.31272507e+00 4.69620764e-01
-1.17447102e+00 1.60041466e-01 -2.23585561e-01 3.36605161e-01
-4.26720679e-01 1.90799981e-01 -1.29079270e+00 5.29650331e-01
9.84512329e-01 -1.56601995e-01 -1.04525864e-01 1.17127645e+00
5.08384526e-01 -7.08292484e-01 -9.40009475e-01 1.19527543e+00
-2.51225144e-01 -1.29370546e+00 1.30930888e-02 4.66184616e-01
4.73973453e-01 4.50669527e-01 -3.80395234e-01 -5.08958846e-03
1.54845998e-01 -7.42095053e-01 7.87476182e-01 6.35931790e-01
8.53923559e-01 -8.06114435e-01 9.25029695e-01 9.36654031e-01
-1.30506146e+00 2.50482500e-01 -9.87462699e-01 -2.66152620e-01
3.07952881e-01 6.71019733e-01 -1.62429655e+00 7.69466579e-01
7.03076482e-01 5.09419978e-01 -4.79255766e-01 1.28424585e+00
-1.55772537e-01 -1.61528904e-02 -6.78620517e-01 -6.18048012e-02
1.07569166e-01 -5.41950583e-01 3.79867136e-01 6.48813605e-01
8.80794108e-01 -9.14970636e-02 4.12809057e-03 7.96042383e-01
4.07700419e-01 -4.09809016e-02 -1.25999677e+00 6.84332073e-01
9.99236822e-01 1.09167910e+00 -4.94437307e-01 -3.84875029e-01
-1.61160171e-01 6.74065650e-01 1.64946802e-02 -3.34971137e-02
-9.35527563e-01 -2.34240890e-01 6.36742353e-01 6.04969919e-01
1.38879001e-01 -8.58073235e-01 -5.37760317e-01 -6.72737360e-01
-4.85236123e-02 1.10650368e-01 -3.96975279e-01 -1.26787901e+00
-6.38732135e-01 7.59413183e-01 2.33990535e-01 -1.70942211e+00
-5.59039295e-01 -7.25748301e-01 -5.06045461e-01 1.14411402e+00
-2.09243822e+00 -9.40724134e-01 -9.53938723e-01 4.44919020e-01
3.98565501e-01 2.33109936e-01 4.66100276e-01 2.90724635e-01
1.56732857e-01 -1.52184486e-01 -2.60989983e-02 -3.97862196e-01
4.38729525e-01 -1.01559794e+00 3.75148624e-01 5.05834520e-01
-2.51304120e-01 1.32120982e-01 9.52909529e-01 -9.07794416e-01
-1.62484419e+00 -9.90738153e-01 5.85070491e-01 -7.88563907e-01
2.61654377e-01 -2.01666728e-01 -8.53877008e-01 6.73172355e-01
-5.19143604e-02 -1.34688333e-01 -2.66309708e-01 -5.67579150e-01
2.05021203e-01 -4.77347940e-01 -1.58348048e+00 1.50522143e-01
9.69370246e-01 -2.19372019e-01 -4.19748098e-01 5.39132208e-03
6.48728192e-01 -7.97801673e-01 -4.24275815e-01 7.45851755e-01
4.75699246e-01 -1.34458125e+00 9.66946602e-01 4.49946225e-01
1.55745447e-01 -5.89137495e-01 -3.90422702e-01 -1.01182127e+00
2.32707694e-01 1.74473867e-01 1.87704533e-01 1.13613617e+00
1.75318480e-01 -9.18027759e-01 1.40473986e+00 2.21497327e-01
-2.11515084e-01 -3.97067547e-01 -1.24023962e+00 -8.19747925e-01
7.45471660e-03 -7.14457989e-01 8.51791501e-01 5.20692408e-01
-8.00687551e-01 -2.37947367e-02 1.09415554e-01 6.96640491e-01
1.10177863e+00 5.15155159e-02 1.19160700e+00 -1.60203505e+00
5.34434557e-01 -5.06537557e-02 -9.27359164e-01 -1.13344979e+00
9.99559164e-02 -6.07834458e-01 3.04305404e-01 -1.94006395e+00
-3.93287569e-01 -8.58545005e-01 5.32280207e-01 -1.62426189e-01
5.13469279e-01 4.07599777e-01 -1.17630735e-01 2.13343889e-01
-8.04921985e-02 4.36112046e-01 1.08774507e+00 2.58155435e-01
-6.55111820e-02 2.26204038e-01 -1.10191286e-01 7.60911405e-01
7.84256876e-01 -5.08459866e-01 -5.56904912e-01 -7.45510876e-01
2.90709380e-02 3.68496060e-01 5.30285120e-01 -1.39716828e+00
3.97353441e-01 -2.78663337e-01 4.70622897e-01 -1.75257003e+00
1.00829792e+00 -1.36659014e+00 2.24332631e-01 4.83627021e-01
7.93111563e-01 2.04340890e-01 1.47019655e-01 3.41356575e-01
-2.01718748e-01 -5.17478883e-01 7.81334043e-01 -2.23488703e-01
-8.19988132e-01 5.09216785e-01 -7.48542100e-02 -8.41056526e-01
1.35346115e+00 -1.01018274e+00 -2.11510152e-01 -3.17940772e-01
-1.23025142e-01 4.25595820e-01 1.11022723e+00 1.62429839e-01
1.14523840e+00 -1.32016838e+00 -7.22762883e-01 3.34137887e-01
1.35659501e-01 5.11668801e-01 -4.02753130e-02 5.06020665e-01
-1.22001421e+00 6.31633699e-01 -2.28830218e-01 -1.22383714e+00
-1.12208903e+00 4.80004549e-01 2.59907186e-01 3.99837375e-01
-4.94533449e-01 5.03993154e-01 -2.42743164e-01 -5.54057658e-01
-5.03799170e-02 -6.07127845e-01 -5.70512600e-02 -1.83190331e-01
3.37664545e-01 5.04165113e-01 2.32539788e-01 -6.66017234e-01
-5.78930616e-01 1.20551860e+00 3.51830423e-01 -2.46347472e-01
1.02043927e+00 -4.54312474e-01 -1.73878595e-02 3.34683299e-01
1.32154584e+00 2.16467418e-02 -1.24921668e+00 4.59949449e-02
7.12950304e-02 -8.52315843e-01 1.44213751e-01 -1.44414008e-01
-6.85620546e-01 1.09873343e+00 7.79840946e-01 -1.47881880e-02
7.24770486e-01 -7.33256713e-03 5.92666745e-01 3.45576674e-01
1.38404608e+00 -7.66502142e-01 -4.05675113e-01 4.02746916e-01
8.32864225e-01 -1.51104736e+00 2.24461854e-01 -6.98018849e-01
-4.02808726e-01 1.23270380e+00 6.15330577e-01 -2.67521501e-01
5.45683324e-01 5.92595696e-01 1.87721044e-01 2.71861944e-02
-5.69134116e-01 -3.38586330e-01 -2.61215180e-01 9.56649542e-01
-7.83283189e-02 -1.03192978e-01 -4.82410267e-02 -2.80107081e-01
-3.92805815e-01 2.48008102e-01 6.95685387e-01 8.74177754e-01
-1.12849379e+00 -1.26819646e+00 -6.52432263e-01 2.64339745e-01
3.58421564e-01 1.59423426e-01 2.58882064e-02 9.65008378e-01
1.69263542e-01 8.06805849e-01 4.50614154e-01 -3.12784880e-01
4.58869100e-01 -1.75749719e-01 6.27336442e-01 -5.79170763e-01
2.44994208e-01 -2.55435884e-01 -1.27707422e-01 -7.10352600e-01
-1.66306630e-01 -6.19942725e-01 -1.69155014e+00 -5.55399954e-01
-3.54622185e-01 3.27280879e-01 1.58186615e+00 6.21457219e-01
2.29808301e-01 -2.28483796e-01 1.02079213e+00 -1.29909992e+00
-3.28729898e-01 -9.37054217e-01 -5.99172354e-01 -8.39979574e-02
2.69191772e-01 -7.97067106e-01 -3.39483768e-01 -4.09407079e-01] | [7.76619815826416, -2.546610116958618] |
14c89f3c-81da-4534-bf96-bbc0b7d93cf2 | learning-3d-photography-videos-via-self | 2302.10781 | null | https://arxiv.org/abs/2302.10781v1 | https://arxiv.org/pdf/2302.10781v1.pdf | Learning 3D Photography Videos via Self-supervised Diffusion on Single Images | 3D photography renders a static image into a video with appealing 3D visual effects. Existing approaches typically first conduct monocular depth estimation, then render the input frame to subsequent frames with various viewpoints, and finally use an inpainting model to fill those missing/occluded regions. The inpainting model plays a crucial role in rendering quality, but it is normally trained on out-of-domain data. To reduce the training and inference gap, we propose a novel self-supervised diffusion model as the inpainting module. Given a single input image, we automatically construct a training pair of the masked occluded image and the ground-truth image with random cycle-rendering. The constructed training samples are closely aligned to the testing instances, without the need of data annotation. To make full use of the masked images, we design a Masked Enhanced Block (MEB), which can be easily plugged into the UNet and enhance the semantic conditions. Towards real-world animation, we present a novel task: out-animation, which extends the space and time of input objects. Extensive experiments on real datasets show that our method achieves competitive results with existing SOTA methods. | ['Nan Duan', 'Yuejian Fang', 'Zicheng Liu', 'Lijuan Wang', 'Fan Yang', 'Zhengyuan Yang', 'Linjie Li', 'JianFeng Wang', 'Minheng Ni', 'Shengming Yin', 'Chenfei Wu', 'Xiaodong Wang'] | 2023-02-21 | null | null | null | null | ['image-outpainting'] | ['computer-vision'] | [ 3.48725259e-01 7.66770169e-02 -1.96752876e-01 -3.44271839e-01
-5.83303571e-01 -3.06110173e-01 4.06328440e-01 -6.92906380e-01
-6.54701665e-02 5.69583833e-01 1.08520880e-01 -1.31474286e-01
6.05536401e-01 -8.24603975e-01 -7.62494981e-01 -5.78903258e-01
2.68126398e-01 2.57831573e-01 4.67978179e-01 1.64302826e-01
1.47600606e-01 3.50526720e-01 -1.50705123e+00 3.53296250e-01
9.65428889e-01 9.83869314e-01 5.28117120e-01 6.35950625e-01
-2.72150576e-01 1.10366809e+00 -6.07904494e-01 -2.34457791e-01
5.01903474e-01 -7.45976865e-01 -5.79490781e-01 8.14688146e-01
3.84839982e-01 -9.74125981e-01 -4.07422334e-01 1.08592820e+00
3.75899583e-01 1.54206499e-01 3.44254196e-01 -1.24311996e+00
-4.58284855e-01 5.75299598e-02 -9.52106059e-01 -1.02149844e-02
5.73849320e-01 3.15241963e-01 4.48458999e-01 -9.68155324e-01
9.62050974e-01 1.42854702e+00 1.45680934e-01 6.57812476e-01
-1.19114923e+00 -6.89868510e-01 4.15675223e-01 1.03677526e-01
-1.21699369e+00 -4.75581884e-01 1.30374038e+00 -2.93364316e-01
4.43268508e-01 3.02695900e-01 9.59390938e-01 8.61423552e-01
-1.10301651e-01 9.75516379e-01 1.25900781e+00 -4.37430471e-01
2.84540534e-01 2.87932437e-02 -5.22489011e-01 7.89963186e-01
-3.13237280e-01 3.13013375e-01 -4.81807321e-01 -3.78358550e-02
1.31879830e+00 8.84977207e-02 -6.67574048e-01 -4.19568270e-01
-1.05895424e+00 5.61006367e-01 2.48855025e-01 -1.58632219e-01
-2.82931179e-01 7.55078122e-02 1.35608777e-01 2.99780965e-01
9.05651152e-01 -1.09024554e-01 -1.66883439e-01 1.04443856e-01
-1.05901170e+00 1.31078467e-01 4.22788382e-01 1.07179856e+00
9.68424141e-01 2.06422970e-01 8.03563520e-02 8.57115388e-01
1.31039888e-01 3.41309696e-01 3.68769020e-01 -1.43787181e+00
5.49784780e-01 6.35162055e-01 1.76391259e-01 -9.51883197e-01
1.66710511e-01 8.64526350e-03 -8.57105434e-01 6.64703608e-01
1.82350129e-01 1.24071464e-01 -8.90412271e-01 1.52172434e+00
8.96871209e-01 5.16208589e-01 -1.69442445e-01 1.21494317e+00
6.65200055e-01 8.93637180e-01 -2.69157439e-01 -5.13612628e-01
1.00304008e+00 -1.26772738e+00 -7.95968115e-01 -4.16408151e-01
2.82080323e-01 -8.46363366e-01 1.20588791e+00 6.22630894e-01
-1.44410443e+00 -7.46660531e-01 -1.00794315e+00 -5.07387996e-01
2.28798330e-01 5.68471290e-02 3.29397291e-01 3.62576544e-01
-8.55264485e-01 5.61134636e-01 -7.85541713e-01 2.89310008e-01
3.66166860e-01 -1.14910431e-01 -3.75464231e-01 -4.15764391e-01
-9.40465391e-01 6.14748836e-01 6.27267733e-02 8.39034766e-02
-1.09600699e+00 -5.51126897e-01 -1.01924145e+00 -3.67647707e-01
4.22159433e-01 -7.54747927e-01 9.32622790e-01 -1.35870421e+00
-1.82402539e+00 1.09870231e+00 -3.16322386e-01 5.70483133e-03
9.27270770e-01 -1.65297553e-01 -1.19675800e-01 3.25170040e-01
1.59343421e-01 9.42996800e-01 1.34331346e+00 -1.66908216e+00
-5.40889740e-01 -2.38761410e-01 1.93031907e-01 4.15253520e-01
-5.74236400e-02 -7.19201118e-02 -9.66495216e-01 -8.78810883e-01
2.49256790e-01 -7.23486245e-01 -3.10698986e-01 5.46168208e-01
-3.12043905e-01 1.56924009e-01 1.02390337e+00 -9.56402421e-01
1.05590224e+00 -2.22972393e+00 3.85547370e-01 -1.45941630e-01
2.83075243e-01 1.07302338e-01 -6.80014417e-02 1.40745386e-01
-1.18987888e-01 -3.89266461e-01 -3.58461350e-01 -8.33321393e-01
-4.33321476e-01 4.03118461e-01 -3.85683894e-01 4.56556618e-01
1.16852678e-01 6.58595502e-01 -9.89977479e-01 -7.88254321e-01
5.76379299e-01 4.87429917e-01 -7.61661589e-01 6.86585069e-01
-5.73112905e-01 8.40359628e-01 -3.26958805e-01 5.94862103e-01
1.04954445e+00 -2.37595350e-01 -2.54798587e-02 -2.03225836e-01
-5.60751334e-02 2.27486461e-01 -1.33114004e+00 2.20760632e+00
-4.66270715e-01 6.60333812e-01 4.72462811e-02 -6.52795076e-01
9.78193820e-01 2.06730496e-02 2.81260163e-01 -6.39135420e-01
1.81741789e-01 6.81429170e-03 -3.91061664e-01 -7.97867477e-01
4.61536825e-01 -5.69262961e-03 3.58602047e-01 5.75004339e-01
-3.21449161e-01 -5.04290581e-01 -1.22130197e-02 2.40745842e-01
6.37140155e-01 7.18184531e-01 1.69702500e-01 2.37988397e-01
5.06021976e-01 -1.00429796e-01 7.78887391e-01 6.71030805e-02
-4.11330238e-02 1.17238533e+00 4.32575345e-01 -5.96954465e-01
-1.27290285e+00 -8.06974709e-01 1.67457312e-01 6.79495513e-01
5.62799335e-01 -2.76927739e-01 -8.72418046e-01 -5.18856883e-01
-3.06101322e-01 3.80603999e-01 -6.81161582e-01 5.28319739e-02
-6.42269313e-01 -2.04824761e-01 1.16142202e-02 3.76758784e-01
8.06509554e-01 -1.06874394e+00 -6.41693652e-01 1.41985819e-01
-4.67825025e-01 -1.22638381e+00 -7.86485910e-01 -2.43673384e-01
-9.53308105e-01 -9.17496741e-01 -8.91120136e-01 -8.43886733e-01
7.71147907e-01 6.24237597e-01 1.05334985e+00 3.40345144e-01
-1.22978739e-01 -1.11389749e-01 -2.22948238e-01 1.46263644e-01
-6.18470013e-01 -4.22224373e-01 -2.80037045e-01 2.54109085e-01
-1.58661574e-01 -8.24625075e-01 -1.02162361e+00 4.21521664e-01
-1.16520739e+00 8.31416368e-01 3.34117055e-01 8.20031166e-01
7.30047584e-01 -1.01576403e-01 9.10957754e-02 -8.88007700e-01
8.51574913e-03 -1.46028593e-01 -6.24579012e-01 -5.77727258e-02
-3.19560260e-01 -2.19955191e-01 4.94933546e-01 -7.02076197e-01
-1.20378625e+00 2.55016953e-01 -6.88497871e-02 -1.19636118e+00
9.42073390e-03 6.79992288e-02 -5.23058057e-01 -4.88130860e-02
3.64003897e-01 3.51344168e-01 1.27540946e-01 -4.56944227e-01
5.23707390e-01 5.98441303e-01 6.31026447e-01 -4.63081241e-01
7.31413007e-01 8.49039674e-01 -2.46326804e-01 -6.24457300e-01
-7.86562085e-01 -1.04680672e-01 -5.49039543e-01 -3.82645518e-01
8.56939077e-01 -1.15614831e+00 -2.77679026e-01 4.53667313e-01
-1.34761190e+00 -6.67075336e-01 -3.38498592e-01 2.39264950e-01
-6.71970248e-01 7.13399053e-01 -7.74280846e-01 -6.58520997e-01
-8.13351721e-02 -1.21252251e+00 1.37849450e+00 1.27541414e-02
-5.91181102e-04 -8.58394802e-01 1.41520659e-02 4.67739493e-01
-3.38629372e-02 2.65983701e-01 7.23236978e-01 3.08849603e-01
-9.90757585e-01 -5.24416799e-03 -2.75871009e-01 6.48764789e-01
2.07039759e-01 -6.40766602e-03 -1.07458413e+00 -1.55924618e-01
3.66358489e-01 -2.90582210e-01 5.96219063e-01 3.22262794e-01
1.39046335e+00 -1.53828204e-01 -1.90331027e-01 8.99033189e-01
1.29626930e+00 2.19355091e-01 8.85968208e-01 2.09260389e-01
9.12588239e-01 6.98905289e-01 8.05205047e-01 3.75875741e-01
3.78262848e-01 6.67185485e-01 4.75008756e-01 -2.96284556e-01
-4.86741006e-01 -6.38188243e-01 3.12623501e-01 9.59425092e-01
-1.04797021e-01 -2.04559132e-01 -3.14960003e-01 3.85760963e-01
-1.65400493e+00 -9.21011984e-01 1.57768540e-02 2.04477763e+00
1.07158899e+00 4.76776659e-02 -2.53067501e-02 2.86155224e-01
7.34974623e-01 4.32939619e-01 -6.38347805e-01 3.25080901e-02
-1.04347616e-01 -9.77912992e-02 1.43616050e-01 8.53700697e-01
-7.38681018e-01 9.93393600e-01 5.64607334e+00 1.03905272e+00
-1.15491676e+00 1.57506183e-01 9.77282643e-01 -1.34509310e-01
-5.64286590e-01 2.64464915e-01 -3.91990036e-01 6.59123242e-01
2.51268446e-01 1.25054196e-01 5.13987362e-01 7.71880269e-01
5.53666055e-01 -3.95028412e-01 -1.20393109e+00 1.26619244e+00
2.62738079e-01 -1.28292811e+00 1.49950087e-01 -1.54304178e-02
1.02497983e+00 -4.98130143e-01 -1.23540521e-01 -1.74997658e-01
1.34452626e-01 -6.64237201e-01 9.02117848e-01 3.71074617e-01
1.13942778e+00 -5.37373722e-01 9.32838619e-02 4.81945187e-01
-1.05748117e+00 2.57735401e-01 -5.10917366e-01 -1.96847752e-01
5.43012500e-01 6.93655133e-01 -2.36741439e-01 5.30359030e-01
5.81598401e-01 9.90264833e-01 -2.83406198e-01 8.00853431e-01
-4.32182580e-01 2.07689628e-01 -1.36739567e-01 4.92442220e-01
-1.08852096e-01 -5.64492524e-01 4.02217597e-01 6.55161440e-01
2.39255384e-01 3.84694040e-01 2.34184206e-01 8.65576029e-01
-6.65061995e-02 -2.39711087e-02 -6.27642751e-01 4.23918068e-01
3.60176027e-01 1.13270938e+00 -6.92242205e-01 -6.37047589e-01
-5.00442982e-01 1.58757281e+00 2.70138353e-01 4.58472937e-01
-9.72103179e-01 1.38228498e-02 4.42024320e-01 2.20826402e-01
2.41720721e-01 -4.49815467e-02 -2.23504439e-01 -1.47839677e+00
2.28770316e-01 -1.00632346e+00 9.49330255e-02 -1.32553780e+00
-1.09738135e+00 6.77373469e-01 -9.10496786e-02 -1.88197303e+00
-2.47780725e-01 -8.52515250e-02 -4.83957261e-01 7.68851578e-01
-1.59646165e+00 -9.27655518e-01 -7.73624361e-01 6.95697546e-01
1.01514101e+00 2.14533240e-01 3.89173239e-01 4.98477936e-01
-3.64090621e-01 1.23487383e-01 -1.96986943e-01 -8.08729827e-02
7.93398917e-01 -7.44146287e-01 4.41036433e-01 8.97042334e-01
-7.14253262e-02 1.45788684e-01 5.96698463e-01 -7.20842898e-01
-1.21934152e+00 -1.15210450e+00 4.49678898e-01 -3.09957176e-01
1.50359854e-01 -3.13322067e-01 -1.00415862e+00 6.42859876e-01
3.20418149e-01 2.08673239e-01 1.71781808e-01 -7.73478210e-01
-2.69212753e-01 1.55150779e-02 -1.06476200e+00 7.42955506e-01
1.37395024e+00 -5.77050686e-01 -3.38464707e-01 2.86336213e-01
9.58698869e-01 -8.90602231e-01 -4.68077779e-01 1.29344150e-01
3.43713462e-01 -1.32329524e+00 1.04628646e+00 -1.78010985e-01
1.04512691e+00 -5.57491481e-01 -6.22553343e-04 -1.20207059e+00
4.83986400e-02 -8.13249230e-01 -7.49771148e-02 1.17442346e+00
-2.35292241e-01 -2.49585927e-01 1.05953884e+00 7.38259315e-01
-9.95069146e-02 -6.44620836e-01 -7.68538952e-01 -5.39388001e-01
-1.87799841e-01 -4.80052888e-01 5.51573992e-01 1.06741893e+00
-3.37448627e-01 2.00056478e-01 -7.47460306e-01 1.36894524e-01
7.16196358e-01 5.35471141e-01 1.19268644e+00 -7.63499916e-01
-4.93430614e-01 -5.45911230e-02 -2.53063172e-01 -1.70139813e+00
1.00415714e-01 -4.19663221e-01 -4.09162305e-02 -1.24411750e+00
1.20469548e-01 -4.70311314e-01 4.60549951e-01 8.43957514e-02
-3.71960610e-01 4.99127656e-01 -4.34199639e-04 4.79277968e-01
-3.07686120e-01 8.84412706e-01 1.92980444e+00 3.60557251e-02
-2.44058296e-01 -1.81671962e-01 -2.47352660e-01 9.38483536e-01
3.38937849e-01 -4.45757776e-01 -7.58889139e-01 -7.14019120e-01
-1.22495547e-01 6.02381051e-01 4.14636970e-01 -6.62154198e-01
2.67334264e-02 -3.87046695e-01 5.31345665e-01 -7.48231173e-01
6.82543993e-01 -8.65655541e-01 4.54076141e-01 2.79128045e-01
-2.24810943e-01 9.16785672e-02 -1.30217224e-01 6.08896971e-01
-2.63906777e-01 -2.08420306e-01 8.41735899e-01 -2.07989514e-01
-6.18719220e-01 5.80620646e-01 1.08834587e-01 8.48046504e-03
1.08697462e+00 -4.08993781e-01 -1.08141057e-01 -5.23798943e-01
-5.90732872e-01 6.73022196e-02 1.01776373e+00 2.30889618e-01
9.06653106e-01 -1.41180098e+00 -5.08117557e-01 4.99076068e-01
-1.79877043e-01 5.82525611e-01 4.93680120e-01 4.34973121e-01
-8.55548322e-01 -3.45062643e-01 -8.62212926e-02 -7.17565775e-01
-1.15748858e+00 8.73196483e-01 2.52345294e-01 4.43453714e-02
-1.05755723e+00 6.59004092e-01 7.32990026e-01 -3.22707780e-02
2.27271050e-01 -3.61671954e-01 1.05299108e-01 -2.37841770e-01
6.76613033e-01 2.08256111e-01 -2.44980544e-01 -3.75528932e-01
8.07590634e-02 6.92797720e-01 3.22738998e-02 -3.93694580e-01
1.17335033e+00 -4.77315903e-01 -1.34187922e-01 4.20512497e-01
1.32250488e+00 1.43308207e-01 -1.94916105e+00 -3.20793778e-01
-7.15758860e-01 -1.24973643e+00 4.78434144e-03 -2.05332011e-01
-1.40854013e+00 9.68364120e-01 3.88468117e-01 -4.53911647e-02
1.35611296e+00 -1.67115450e-01 1.03628170e+00 -3.10090154e-01
3.53109092e-01 -6.94452584e-01 3.01221102e-01 -1.13091499e-01
8.97833943e-01 -1.25610256e+00 1.52652875e-01 -8.76335680e-01
-6.60769641e-01 9.44086134e-01 7.21871316e-01 -3.93541873e-01
4.20699954e-01 3.52205515e-01 2.49501154e-01 4.32658829e-02
-8.34869504e-01 2.26186682e-02 -3.93583775e-02 7.12331355e-01
1.41248824e-02 -3.49433631e-01 -2.04440475e-01 1.85139149e-01
-3.12820040e-02 2.35365018e-01 6.04957163e-01 7.88202167e-01
-2.08318219e-01 -1.09364259e+00 -3.61142963e-01 -9.78457108e-02
-1.05846673e-02 -3.71924415e-02 -1.43926978e-01 6.16336524e-01
1.00944825e-01 7.44129241e-01 1.39851332e-01 -3.26942056e-01
1.42888725e-01 -2.06648782e-01 6.43474460e-01 -4.57276762e-01
-1.38920411e-01 4.95309860e-01 -3.90313417e-02 -7.08246529e-01
-7.11238563e-01 -4.95247155e-01 -1.01719010e+00 -4.48990643e-01
-3.46044004e-01 -9.87924486e-02 2.81368196e-01 7.39884436e-01
2.89467543e-01 5.10589838e-01 1.07566285e+00 -1.36334074e+00
-8.20565447e-02 -6.57809019e-01 -4.30383533e-01 5.73715210e-01
3.84696960e-01 -5.87078691e-01 -4.75120097e-01 5.43536544e-01] | [10.556764602661133, -1.5521032810211182] |
e9bcc47a-5350-487e-a230-d6e6494465c9 | interpreting-wide-band-neural-activity-using | null | null | https://elifesciences.org/articles/66551 | https://elifesciences.org/articles/66551#downloads | Interpreting wide-band neural activity using convolutional neural networks | Rapid progress in technologies such as calcium imaging and electrophysiology has seen a dramatic increase in the size and extent of neural recordings. Even so, interpretation of this data often depends on manual operations and requires considerable knowledge about the nature of the representation. Decoding provides a means to infer the information content of such recordings but typically requires highly processed data and prior knowledge of the encoding scheme. Here, we
developed a deep-learning-framework able to decode sensory and behavioural variables directly
from wide-band neural data. The network requires little user input and generalizes across stimuli,
behaviours, brain regions, and recording techniques. Once trained, it can be analysed to determine
elements of the neural code that are informative about a given variable. We validated this approach
using data from rodent auditory cortex and hippocampus, identifying a novel representation of
head direction encoded by putative CA1 interneurons. | ['Caswell Barry', 'Christian F Doeller', 'Julie Lefort', 'Daniel Bendor', 'Andrea Banino', 'Jack Kelly', 'Matthias Nau', "Alice O'Leary", 'Catherine Perrodin', 'Sander Tanni', 'Markus Frey'] | 2021-08-02 | null | null | null | elife-2021-8 | ['brain-decoding', 'brain-decoding'] | ['medical', 'miscellaneous'] | [ 6.36509836e-01 -2.23471373e-01 2.32333943e-01 -5.28653979e-01
-6.85852289e-01 -8.21422398e-01 2.52175629e-01 6.58059657e-01
-7.27240860e-01 8.37225556e-01 5.41606322e-02 -2.05192119e-01
7.36336783e-02 -4.71589863e-01 -9.71184254e-01 -9.68793035e-01
-3.09232390e-03 3.42445672e-01 2.67197430e-01 1.48829045e-02
5.69813132e-01 4.53771502e-01 -1.58227897e+00 5.55089831e-01
1.44215629e-01 1.19041741e+00 5.85709512e-01 7.80452609e-01
6.90538716e-03 2.29489446e-01 -6.26479566e-01 1.16894484e-01
-5.27996048e-02 -6.08925819e-01 -6.41455531e-01 -3.35886717e-01
2.21152544e-01 -8.15100893e-02 2.63663232e-01 8.19090545e-01
7.85623491e-01 -3.96029025e-01 5.62687576e-01 -4.61691886e-01
-2.08331555e-01 5.09925365e-01 -2.25130230e-01 8.32090497e-01
1.03861555e-01 1.66508213e-01 8.48259449e-01 -3.16297233e-01
6.78043365e-01 7.13634849e-01 6.41071260e-01 4.67965454e-01
-1.89076316e+00 -7.10964322e-01 -3.05928122e-02 -2.23117962e-01
-1.17350674e+00 -5.86711287e-01 4.24139559e-01 -6.94491923e-01
1.02587867e+00 -7.51374336e-03 1.05122852e+00 1.11275411e+00
3.30412894e-01 2.07495525e-01 1.39574826e+00 -2.96278536e-01
5.87927818e-01 -8.49675611e-02 -4.85113822e-02 1.87287182e-01
9.60052013e-02 -6.39972761e-02 -6.53092921e-01 -9.03808847e-02
8.90485704e-01 -3.19527358e-01 -3.02638978e-01 -8.63733143e-02
-1.20509005e+00 5.85168481e-01 5.02407551e-02 4.12382424e-01
-2.78996825e-01 3.05137098e-01 5.06634653e-01 2.85216331e-01
4.30650869e-03 7.55510807e-01 -7.74769366e-01 -4.08869028e-01
-1.01956153e+00 1.33453026e-01 4.93625402e-01 2.73032248e-01
8.09221566e-01 3.66474055e-02 3.56368124e-01 7.52858043e-01
2.60352552e-01 2.08109766e-01 7.77508318e-01 -1.05586469e+00
-3.32956798e-02 4.52320039e-01 -2.85972446e-01 -7.40050912e-01
-6.68742597e-01 -3.79892975e-01 -3.86613190e-01 5.45971215e-01
6.16047621e-01 -2.00185120e-01 -7.58449078e-01 1.69899571e+00
6.47252076e-04 -1.70588002e-01 -2.30288014e-01 8.43095303e-01
3.47810119e-01 3.58112037e-01 2.11004466e-01 -3.01480979e-01
1.14360809e+00 3.36827546e-01 -3.73307467e-01 -6.99198961e-01
5.58725595e-01 -2.73788065e-01 7.08933771e-01 4.97142136e-01
-9.11339104e-01 -3.21725011e-01 -1.03739917e+00 -1.33780316e-01
-3.83411795e-01 -1.76638201e-01 6.99486792e-01 3.87702197e-01
-1.04256523e+00 5.69723785e-01 -7.76309729e-01 -3.83730620e-01
7.32326865e-01 8.64623129e-01 -5.69501936e-01 4.19339985e-01
-6.36606157e-01 7.90701151e-01 6.40767574e-01 7.39221415e-03
-9.44062769e-01 -6.42534137e-01 -4.04292375e-01 2.01485544e-01
-3.28380436e-01 -5.41456819e-01 1.16929150e+00 -1.07325947e+00
-1.26260519e+00 1.09936666e+00 -8.81092548e-02 -3.43701899e-01
-3.57512444e-01 2.63584107e-01 5.23557812e-02 3.61791700e-01
-1.87480375e-01 8.27121496e-01 7.37661421e-01 -9.34957743e-01
-4.78329211e-01 -9.05477762e-01 -2.63583630e-01 -1.95732161e-01
5.59828132e-02 2.26310447e-01 2.07390070e-01 -5.33327043e-01
1.89616069e-01 -7.11209893e-01 -1.10314496e-01 2.75207072e-01
2.46614560e-01 3.51442099e-01 3.47473055e-01 -5.58614075e-01
8.51297677e-01 -2.57363033e+00 2.02906698e-01 2.09616229e-01
4.32667881e-02 6.45394325e-02 9.18292627e-02 4.04944211e-01
-1.98248342e-01 -2.06328612e-02 -5.07860661e-01 2.72938341e-01
-3.05942565e-01 1.03334151e-01 -3.82228017e-01 6.31402016e-01
3.07910234e-01 6.19585991e-01 -5.60479403e-01 -2.09817469e-01
-2.52626371e-02 6.36590540e-01 -6.24662101e-01 7.07781687e-02
-3.58880639e-01 8.51758480e-01 -1.52363792e-01 3.62219125e-01
3.09905350e-01 -9.22891349e-02 2.91554660e-01 4.21516225e-02
-4.31807488e-01 6.55263960e-01 -7.26167381e-01 1.90148425e+00
-4.22142953e-01 9.62286174e-01 2.23971635e-01 -1.42145896e+00
1.01023519e+00 3.32562298e-01 1.47922799e-01 -9.67169166e-01
6.02005243e-01 3.27720672e-01 4.11646754e-01 -4.41419363e-01
-2.96593368e-01 -4.20917302e-01 7.46791391e-03 4.48996127e-01
6.33465171e-01 -1.54609218e-01 1.41204149e-01 -2.25970268e-01
1.10282695e+00 1.88329279e-01 3.71909320e-01 -3.67451340e-01
-9.14387703e-02 -1.08276993e-01 4.55871761e-01 4.95346159e-01
8.41214284e-02 7.02443123e-01 6.61585331e-01 -3.20981383e-01
-9.27354038e-01 -9.13140357e-01 -7.08707631e-01 1.26850736e+00
-2.89004385e-01 -1.36761203e-01 -6.10552132e-01 1.02079421e-01
-3.41208637e-01 4.69854295e-01 -7.66873479e-01 -3.00399125e-01
-3.32564920e-01 -9.06845450e-01 6.25280321e-01 5.35377383e-01
-4.30452377e-02 -1.17210269e+00 -1.26126611e+00 4.78013873e-01
3.12464207e-01 -8.77881944e-01 3.47143918e-01 1.14012408e+00
-1.20696449e+00 -8.36181641e-01 -3.08360815e-01 -8.31001222e-01
6.42756343e-01 -3.73276830e-01 6.44502342e-01 -2.86981851e-01
-8.31414461e-01 -3.51906940e-02 -2.49246527e-02 -6.18359923e-01
5.46433069e-02 -6.03581499e-03 -4.00331408e-01 -2.31402680e-01
4.37455386e-01 -1.07720006e+00 -4.73749518e-01 -1.66405857e-01
-8.62501979e-01 -2.81412482e-01 7.54868329e-01 6.79991245e-01
8.17640543e-01 -1.76365405e-01 7.48578072e-01 -7.94893146e-01
5.11139870e-01 -4.44697380e-01 -5.84254146e-01 -1.73339307e-01
-2.28170201e-01 3.67642879e-01 6.80633485e-01 -3.06022853e-01
-7.23679543e-01 2.54509717e-01 -4.33418065e-01 4.17164624e-01
-7.73193598e-01 6.38869166e-01 -3.04946955e-02 -1.66685298e-01
8.58458757e-01 2.70305574e-01 -1.93208173e-01 -3.62358302e-01
-1.43746197e-01 5.74554682e-01 7.51066506e-01 -5.16084969e-01
-8.92187506e-02 5.70560575e-01 -1.42892152e-01 -7.70139456e-01
-5.10539830e-01 -3.22165728e-01 -8.17886114e-01 -5.65459281e-02
7.13562429e-01 -6.95832908e-01 -8.43249559e-01 4.94354427e-01
-8.44984353e-01 -4.40419257e-01 -8.49065185e-03 7.25055099e-01
-6.54809177e-01 -2.23182857e-01 -5.07285118e-01 -6.13735437e-01
-1.81409642e-01 -8.80898893e-01 1.02612150e+00 9.49870646e-02
-5.75946450e-01 -8.78785312e-01 5.55836141e-01 -3.32087316e-02
2.74780154e-01 1.04051583e-01 1.17980230e+00 -5.78671396e-01
-3.05497795e-01 -4.45900708e-01 -9.04085115e-02 -3.07493098e-02
-1.12207867e-01 7.41612166e-02 -1.27765298e+00 -5.19631570e-03
-6.48024306e-02 -6.69372141e-01 8.21504831e-01 5.76031923e-01
1.47371387e+00 -6.32570237e-02 -2.20348686e-01 7.45058119e-01
1.36575973e+00 1.70900658e-01 5.30116498e-01 3.50882053e-01
1.35677278e-01 6.57952011e-01 -1.06658421e-01 3.12000573e-01
4.66954708e-02 4.75893438e-01 3.79073739e-01 1.98776022e-01
2.30182320e-01 -2.13528797e-02 1.73620075e-01 6.09588146e-01
-2.76123509e-02 2.11307630e-01 -8.51366520e-01 6.87529683e-01
-1.37751365e+00 -9.04798925e-01 2.07923114e-01 2.20248961e+00
1.20800638e+00 1.91450104e-01 1.98730469e-01 4.48093891e-01
3.57593894e-01 -2.29572713e-01 -7.21790254e-01 -7.81721532e-01
-7.35908449e-02 6.95979774e-01 3.19551528e-01 2.13288710e-01
-5.09671688e-01 5.33606291e-01 7.87909174e+00 3.47363591e-01
-1.56937516e+00 -2.45363876e-01 3.29605758e-01 -2.94900626e-01
-2.28643894e-01 -6.64218068e-02 -6.76773667e-01 5.87054610e-01
1.31277215e+00 8.77139047e-02 7.48450279e-01 3.08350801e-01
8.30713436e-02 -5.13482988e-01 -1.18325400e+00 8.43165517e-01
-1.50656223e-01 -1.32162619e+00 -2.89986432e-01 1.83176681e-01
2.06977427e-01 3.08479488e-01 5.20923138e-02 -3.07007059e-02
4.30723429e-02 -1.21664536e+00 5.98946393e-01 3.60935062e-01
6.74437523e-01 -5.69933832e-01 5.74669182e-01 4.61839288e-01
-6.11782551e-01 -2.22716957e-01 -4.72607791e-01 -4.77665812e-01
-1.79431930e-01 4.20304805e-01 -7.21334398e-01 -4.22630399e-01
7.88417637e-01 4.77671564e-01 -6.69832408e-01 1.18519604e+00
-6.11450113e-02 8.12420249e-01 -5.05684555e-01 6.40556514e-02
-1.39105216e-01 1.25519633e-01 3.18757910e-03 1.28041553e+00
2.90116698e-01 1.80070415e-01 -3.61578166e-01 9.16167319e-01
-8.32598831e-04 -3.32182907e-02 -5.84151089e-01 -5.38058519e-01
3.57498884e-01 1.27806222e+00 -9.99519229e-01 1.68614924e-01
-2.67624438e-01 5.72843373e-01 5.28992295e-01 3.43939364e-01
-2.87644356e-01 -2.90967882e-01 5.01918972e-01 8.10253527e-03
6.62322879e-01 -4.01018560e-01 -5.52448452e-01 -6.17677033e-01
-1.34654418e-01 -6.51069582e-01 1.55758560e-01 -9.14568722e-01
-7.76407599e-01 3.70290220e-01 -2.49118522e-01 -6.04518354e-01
-4.20762986e-01 -7.90830374e-01 -4.33050722e-01 8.72057796e-01
-1.13951921e+00 -4.89081413e-01 7.77995214e-02 3.82903963e-01
1.22677363e-01 1.87377259e-01 1.16887760e+00 1.75104707e-01
-2.45670587e-01 1.68112352e-01 9.82115865e-02 4.33952585e-02
4.85125571e-01 -1.11464667e+00 7.55125433e-02 2.26159483e-01
3.91349584e-01 6.96844399e-01 7.32920468e-01 -2.28313178e-01
-1.17050183e+00 -7.07752168e-01 5.35358787e-01 -2.75248975e-01
7.49468982e-01 -7.35286295e-01 -1.05176616e+00 7.27565229e-01
-1.05848880e-02 1.07872218e-01 1.24182439e+00 2.14311793e-01
-4.06908005e-01 -1.60662070e-01 -9.12950277e-01 1.82794631e-01
6.08695328e-01 -9.05516446e-01 -7.13040829e-01 -3.17448258e-01
1.14668295e-01 -2.87620649e-02 -7.57513046e-01 7.20154494e-02
9.03379202e-01 -1.01622856e+00 6.34899139e-01 -4.48405027e-01
1.18684076e-01 -2.52922803e-01 -2.03433797e-01 -1.39765966e+00
-2.56709933e-01 -1.06560670e-01 3.53056699e-01 1.05663311e+00
6.40753508e-01 -4.30172473e-01 4.02997017e-01 4.51709360e-01
-2.07274482e-01 -5.41364908e-01 -9.77184832e-01 -2.41322950e-01
2.10274324e-01 -4.08251405e-01 7.51057342e-02 4.13838178e-01
4.14566398e-01 4.87510949e-01 2.97679186e-01 7.04597235e-02
2.56655276e-01 1.96633860e-01 3.25296938e-01 -1.24237120e+00
-3.83327186e-01 -4.13773060e-01 -9.57114041e-01 -7.88836539e-01
2.05323130e-01 -9.51346934e-01 2.43743345e-01 -1.37137032e+00
1.77074507e-01 1.93447992e-02 -3.46497059e-01 5.19303143e-01
1.64655492e-01 5.81400752e-01 -3.42488080e-01 1.54829705e-02
-3.76621097e-01 6.22231215e-02 7.26279497e-01 1.63869187e-01
-1.49944369e-02 -2.72122443e-01 -8.14078689e-01 7.14849770e-01
1.05411708e+00 -7.30197549e-01 -1.10924371e-01 -6.57840014e-01
6.87145412e-01 -1.30536795e-01 6.65597677e-01 -1.22493351e+00
1.29852340e-01 1.92191720e-01 7.58176863e-01 -2.18508095e-01
2.52751350e-01 -7.99754441e-01 8.17072690e-02 1.79521844e-01
-8.71841967e-01 -2.43651003e-01 5.10545611e-01 5.17074525e-01
-2.44829774e-01 -2.87984133e-01 8.05693686e-01 -3.38630676e-01
-5.94856203e-01 -1.03669263e-01 -9.68161404e-01 2.30658486e-01
7.04628289e-01 -4.72773761e-01 -2.69158900e-01 7.99972787e-02
-9.43099737e-01 -2.67206311e-01 5.45466125e-01 -1.48760572e-01
3.76836240e-01 -8.32677662e-01 -3.47644776e-01 6.17213309e-01
1.55922696e-01 -1.32133067e-01 7.99759116e-04 8.91779840e-01
-3.08529019e-01 3.16547483e-01 -7.83164144e-01 -7.03041852e-01
-9.65085387e-01 4.17903662e-01 1.80681258e-01 4.32410628e-01
-5.17889857e-01 9.43626881e-01 3.23278069e-01 -1.32650837e-01
-8.72264802e-02 -3.63836020e-01 -3.02595109e-01 1.73180655e-01
6.96008623e-01 -1.86251149e-01 2.10148588e-01 -3.71717304e-01
-3.26510608e-01 5.33845484e-01 3.15860510e-02 -3.13783377e-01
1.62725186e+00 -1.01689801e-01 -2.76259035e-01 1.20972919e+00
1.26562750e+00 -2.37515584e-01 -1.48576140e+00 3.59410942e-01
2.11833790e-02 -3.26222867e-01 2.29465380e-01 -9.70904589e-01
-9.28041875e-01 1.47522891e+00 5.35905182e-01 1.09416582e-01
1.15783083e+00 1.79435357e-01 4.13669467e-01 2.92755127e-01
5.89366615e-01 -1.04947627e+00 -2.60600984e-01 2.55484045e-01
6.04396582e-01 -8.73617053e-01 -1.60909086e-01 2.12225929e-01
-3.15590024e-01 1.25964487e+00 2.90711999e-01 -2.44259879e-01
6.03463948e-01 6.90307200e-01 1.76477537e-01 -3.88867378e-01
-1.02686477e+00 -9.30678025e-02 -8.78211707e-02 9.63083744e-01
7.56104767e-01 -2.52013057e-01 -1.53929129e-01 5.29032528e-01
-2.55347282e-01 2.23721042e-01 4.00982052e-01 9.55985248e-01
-7.07158566e-01 -6.86717570e-01 -1.89049169e-01 5.99528968e-01
-8.91422033e-01 -1.91751063e-01 -6.28887951e-01 3.38382334e-01
1.35658085e-01 4.58766907e-01 2.39394605e-01 -7.84299448e-02
-1.45462662e-01 3.42426836e-01 8.48336041e-01 -8.61196756e-01
-4.99047577e-01 4.34913933e-01 -9.94223654e-02 -4.46855873e-01
-5.47413170e-01 -8.13304245e-01 -1.70370090e+00 1.26950026e-01
-2.60195374e-01 5.65075502e-02 1.03277993e+00 1.00652850e+00
3.85500908e-01 5.56632161e-01 6.59684837e-02 -8.88755023e-01
-3.45980562e-02 -7.38080800e-01 -8.25806797e-01 8.01054612e-02
4.94543105e-01 -3.95161778e-01 -4.63959843e-01 5.14643431e-01] | [9.666563034057617, 2.5298197269439697] |
ee935c73-545c-41c3-b337-eb88713a078a | polygames-improved-zero-learning | 2001.09832 | null | https://arxiv.org/abs/2001.09832v1 | https://arxiv.org/pdf/2001.09832v1.pdf | Polygames: Improved Zero Learning | Since DeepMind's AlphaZero, Zero learning quickly became the state-of-the-art method for many board games. It can be improved using a fully convolutional structure (no fully connected layer). Using such an architecture plus global pooling, we can create bots independent of the board size. The training can be made more robust by keeping track of the best checkpoints during the training and by training against them. Using these features, we release Polygames, our framework for Zero learning, with its library of games and its checkpoints. We won against strong humans at the game of Hex in 19x19, which was often said to be untractable for zero learning; and in Havannah. We also won several first places at the TAAI competitions. | ['Shi-Jim Yen', 'Yi-Jun Ye', 'Shi-Cheng Ye', 'Yu-Jin Lin', 'Hsin-I Lin', 'Cheng-Ling Li', 'Vasil Khalidov', 'Qucheng Gong', 'Maria Elsa', 'Xian-Dong Chiu', 'Shi-Yu Chen', 'Guan-Wei Chen', 'Yen-Chi Chen', 'Xavier Martinet', 'Sergey Zagoruyko', 'Olivier Teytaud', 'Jeremy Rapin', 'Hengyuan Hu', 'Gabriel Synnaeve', 'Fabien Teytaud', 'Julien Dehos', 'Vegard Mella', 'Tristan Cazenave', 'Baptiste Roziere'] | 2020-01-27 | null | null | null | null | ['board-games'] | ['playing-games'] | [-5.81454873e-01 1.78110525e-01 -3.14893693e-01 2.32836440e-01
-5.27665019e-01 -7.35003948e-01 5.10618806e-01 -2.94080317e-01
-8.87020111e-01 9.62156892e-01 -1.65006638e-01 -5.68843126e-01
-2.29517981e-01 -1.14287722e+00 -8.12017262e-01 -4.47265774e-01
-4.41202819e-01 5.31636536e-01 9.53672230e-01 -1.03681207e+00
3.01388234e-01 6.25612587e-02 -1.17340636e+00 5.79479933e-01
3.44878256e-01 7.71469235e-01 1.59695856e-02 1.20455384e+00
6.33250356e-01 1.63692451e+00 -1.19536912e+00 -7.48982131e-01
6.61918342e-01 3.09391078e-02 -1.04361796e+00 -6.89816952e-01
3.90131772e-01 -6.82578385e-01 -5.92880368e-01 9.19072926e-01
6.16567135e-01 -4.58933450e-02 9.03732330e-02 -1.40679765e+00
-2.02201411e-01 9.93901253e-01 -4.14195448e-01 3.63068610e-01
-8.74287263e-02 6.50184751e-01 1.01894927e+00 -4.53485958e-02
6.29553497e-01 1.06433368e+00 9.42272782e-01 5.64527035e-01
-6.53933823e-01 -7.93792725e-01 -1.18156470e-01 1.71349004e-01
-8.65569293e-01 -2.01572210e-01 2.62787551e-01 -3.80705863e-01
1.37425721e+00 1.03378676e-01 8.28492165e-01 1.41369355e+00
4.40425217e-01 5.59057593e-01 9.18165207e-01 -3.94077957e-01
3.21592718e-01 -5.74400187e-01 7.12043494e-02 8.58990550e-01
3.75040680e-01 3.83673817e-01 -5.51646829e-01 -1.35358915e-01
1.18673253e+00 -3.82226884e-01 2.84398168e-01 -1.74512882e-02
-1.15509832e+00 9.02860165e-01 6.87493443e-01 2.55402982e-01
-1.41629159e-01 8.45345497e-01 5.15202999e-01 6.08531117e-01
2.12880135e-01 9.55956578e-01 -7.05737948e-01 -7.30438709e-01
-7.08261013e-01 6.96690679e-01 8.07623267e-01 5.96898973e-01
6.44985497e-01 1.13437854e-01 -7.65913427e-02 3.87606859e-01
-2.46156245e-01 -2.73710996e-01 4.34354782e-01 -1.48261952e+00
5.83216071e-01 5.14871418e-01 2.19101116e-01 -8.04719448e-01
-6.79978848e-01 -4.34046924e-01 -6.58375740e-01 1.01414680e+00
8.00951004e-01 -8.42249691e-01 -8.99793923e-01 1.55389631e+00
-1.29748374e-01 3.84093463e-01 -1.35390861e-02 7.24757373e-01
7.24812865e-01 1.46753117e-01 -3.20977241e-01 6.89879000e-01
1.02304864e+00 -9.79828477e-01 -1.40223771e-01 -4.30031955e-01
1.01976550e+00 -1.61497757e-01 1.03907990e+00 8.26552272e-01
-1.03849411e+00 -2.60839701e-01 -1.41291499e+00 2.17932224e-01
-6.02401555e-01 -1.22057915e-01 1.13282299e+00 9.52528358e-01
-1.23923612e+00 1.06734753e+00 -9.83176529e-01 -1.67013615e-01
4.20073032e-01 8.31620872e-01 -5.87783456e-01 3.64476293e-01
-1.50656104e+00 1.01569211e+00 5.76321542e-01 -1.02494396e-01
-1.30748832e+00 -3.94743055e-01 -6.34012997e-01 -7.17667490e-02
7.47264683e-01 -3.37486744e-01 1.43888974e+00 -8.04996490e-01
-1.69310308e+00 9.41842377e-01 9.42448735e-01 -8.96513343e-01
7.53030658e-01 -3.01151574e-01 -8.43927488e-02 -1.70545548e-01
4.22428936e-01 6.13802314e-01 6.70695901e-01 -5.11125684e-01
-5.33972263e-01 4.94645648e-02 1.25925851e+00 -1.59662236e-02
-2.02246577e-01 1.17862582e-01 -1.88747883e-01 -4.23894107e-01
-3.74365926e-01 -1.01828337e+00 -6.64500833e-01 -4.23434913e-01
-3.93125713e-01 -1.80556804e-01 4.68645960e-01 -3.90705943e-01
9.14723754e-01 -1.89271474e+00 -7.42005697e-03 1.10577606e-01
5.17641008e-01 2.62524694e-01 -1.39769122e-01 1.85117289e-01
-4.27131839e-02 3.90385687e-01 3.52954596e-01 -2.04376429e-02
1.00068107e-01 8.46350417e-02 1.30363300e-01 4.50412571e-01
7.69688338e-02 1.03548181e+00 -1.07682192e+00 -9.94613990e-02
2.10203245e-01 -8.90146643e-02 -9.25161362e-01 -1.20843768e-01
-1.14104867e-01 7.31040612e-02 -1.15440249e-01 5.24721026e-01
2.12036759e-01 -9.23606530e-02 -2.14297771e-02 5.69345951e-01
-2.41011292e-01 2.37273127e-01 -1.06438613e+00 1.74988484e+00
8.60414505e-02 7.33682990e-01 9.51130837e-02 -6.07246339e-01
6.04894876e-01 1.56795233e-01 3.86480302e-01 -5.69479823e-01
5.42119026e-01 9.93063897e-02 3.10186356e-01 -2.75827087e-02
6.99875653e-01 1.65026903e-01 -6.07675195e-01 3.02315384e-01
4.40938801e-01 -2.45082885e-01 3.68102372e-01 3.51493299e-01
1.76260376e+00 4.52259928e-02 -4.17589024e-02 -3.24440897e-01
-1.74913645e-01 1.97391063e-01 4.72383201e-01 1.21881163e+00
-3.84462982e-01 4.82713461e-01 1.24912131e+00 -1.01572740e+00
-9.16963756e-01 -7.95147896e-01 4.55885977e-01 1.35480070e+00
4.08699587e-02 -8.46740842e-01 -1.08882594e+00 -5.95786095e-01
-2.24147871e-01 1.95761964e-01 -1.03124833e+00 -3.07504714e-01
-6.44031823e-01 -4.53870058e-01 1.14470124e+00 3.98879111e-01
8.28252017e-01 -1.34216046e+00 -7.60700524e-01 1.90513641e-01
1.63901493e-01 -7.95036197e-01 1.31492525e-01 8.21640372e-01
-3.45885336e-01 -1.40128648e+00 -4.72228676e-01 -5.81713498e-01
-6.81198686e-02 5.37858903e-02 1.33263910e+00 3.55520934e-01
-2.62795597e-01 -2.25168139e-01 -4.19711828e-01 -7.13675380e-01
-1.77430391e-01 8.30429077e-01 1.97083086e-01 -6.40444756e-01
-3.44492570e-02 -4.47044551e-01 -2.72851884e-01 1.81520626e-01
-3.79523367e-01 -1.36709601e-01 2.62133986e-01 7.60141611e-01
-3.33506405e-01 1.36656433e-01 3.85761559e-01 -9.90686417e-01
8.42660904e-01 -4.10399020e-01 -7.86843181e-01 -2.15317205e-01
3.56825918e-01 -2.41962075e-01 4.10748094e-01 -2.80409306e-01
-3.67921680e-01 -2.89431393e-01 -1.15993164e-01 -2.21007824e-01
3.26994732e-02 1.33602768e-01 2.41397358e-02 -5.82307577e-01
1.29535961e+00 -7.46216059e-01 -3.89459699e-01 -1.26086533e-01
4.61254507e-01 4.87795264e-01 7.45039582e-01 -7.32268929e-01
6.95452750e-01 1.74810037e-01 -7.33485818e-02 -7.07111239e-01
-6.77185833e-01 -2.78054625e-01 -8.34577918e-01 -4.95351136e-01
1.24531579e+00 -9.86042023e-01 -1.24448097e+00 1.09586692e+00
-1.18398595e+00 -1.30067968e+00 -3.85975420e-01 1.33170441e-01
-6.60870850e-01 -1.27362967e-01 -9.87552285e-01 -3.36587012e-01
4.89539206e-02 -9.54648614e-01 4.83719289e-01 2.66586006e-01
-2.94758976e-01 -7.27466285e-01 4.81758744e-01 6.68981224e-02
4.46230084e-01 5.24248123e-01 4.89922523e-01 -4.03550863e-01
-5.47139585e-01 -4.08956826e-01 5.18792979e-02 4.09390539e-01
-2.11102411e-01 -1.67436525e-01 -8.49201202e-01 -4.75719392e-01
-5.80669880e-01 -9.17865396e-01 9.20653820e-01 3.95043969e-01
9.35276985e-01 -1.69444010e-01 -5.32661676e-02 6.55197740e-01
1.19191682e+00 -1.55086040e-01 1.09518850e+00 1.19807410e+00
5.29840767e-01 1.48444578e-01 2.58427471e-01 4.44241673e-01
2.43006229e-01 4.83725458e-01 1.21667504e+00 -2.51625657e-01
1.30501390e-01 -8.12591314e-02 5.05583763e-01 2.79839158e-01
-6.76845253e-01 -6.01505265e-02 -8.31946969e-01 5.18677115e-01
-1.92681003e+00 -1.30570769e+00 -1.57502964e-01 2.02375984e+00
6.27537251e-01 9.11347866e-01 6.17403030e-01 5.05112521e-02
4.58861649e-01 3.44906271e-01 -3.53708446e-01 -6.19176984e-01
-2.87081391e-01 8.28743935e-01 1.04996300e+00 3.80517781e-01
-1.41854858e+00 1.64006352e+00 7.77188635e+00 9.92870331e-01
-8.53572488e-01 2.75307417e-01 5.43119431e-01 -4.80270714e-01
5.96470356e-01 1.48940220e-01 -6.62434697e-01 3.47839177e-01
9.17879224e-01 2.36329362e-01 6.48455322e-01 1.06952262e+00
-2.38959193e-01 -1.50368497e-01 -4.97348964e-01 7.08614051e-01
-2.99661815e-01 -1.78643227e+00 -6.20538890e-01 3.17367226e-01
1.03204870e+00 5.67482829e-01 -6.34969324e-02 8.98485005e-01
1.57889891e+00 -1.38301158e+00 9.61983025e-01 2.13797420e-01
7.88496852e-01 -1.07593095e+00 8.70536983e-01 5.57041764e-01
-9.31333065e-01 -4.72970575e-01 -7.96759009e-01 -8.87127101e-01
-4.30479914e-01 -1.14090011e-01 -3.94220889e-01 2.20860571e-01
9.74085450e-01 4.66168225e-01 -8.23966026e-01 1.09380889e+00
-6.45948768e-01 6.20198250e-01 -3.93044174e-01 -5.73111773e-02
7.96527982e-01 2.96965212e-01 2.69823134e-01 8.23023140e-01
-2.14671018e-03 -3.41248512e-02 2.13724300e-01 4.97541249e-01
-2.28955716e-01 -4.00970459e-01 -8.24262977e-01 2.77648002e-01
1.07892141e-01 1.31426787e+00 -7.03978419e-01 -1.61317974e-01
-1.05576992e-01 9.16024983e-01 7.90749609e-01 1.20158438e-02
-9.75964069e-01 -6.84567451e-01 7.82936454e-01 1.96352348e-01
1.88268855e-01 -4.43688691e-01 -2.77328104e-01 -1.03071845e+00
-2.79906034e-01 -1.02913737e+00 2.35083193e-01 -7.19914675e-01
-9.43636715e-01 6.43735051e-01 -1.79839939e-01 -1.05485213e+00
-2.19487190e-01 -9.91286278e-01 -9.98657107e-01 5.10911107e-01
-8.50010514e-01 -1.26284158e+00 -3.37185636e-02 7.41454124e-01
1.33998945e-01 -6.09653175e-01 7.38022089e-01 5.34292348e-02
-4.40319151e-01 7.19946265e-01 -8.62739459e-02 5.37426770e-01
6.26704693e-01 -1.47663510e+00 9.55276847e-01 6.40945315e-01
-5.04423156e-02 3.62941325e-01 4.19663906e-01 -6.37246966e-01
-7.66537547e-01 -7.36377537e-01 2.08003521e-01 -7.69417882e-01
1.12220216e+00 -5.77901483e-01 -2.99981117e-01 1.02090108e+00
3.13058555e-01 -2.03005835e-01 1.23420976e-01 4.92646933e-01
-2.96225727e-01 1.03985950e-01 -7.56403506e-01 7.11214006e-01
9.77806926e-01 -4.38266098e-01 -2.86276996e-01 3.87521416e-01
8.00786793e-01 -7.33602345e-01 -3.22689295e-01 1.69993024e-02
7.38792717e-01 -1.14868712e+00 7.08799005e-01 -1.03074098e+00
3.97782743e-01 5.19610271e-02 6.75297156e-02 -1.42087507e+00
-6.64522588e-01 -1.06540704e+00 2.97802836e-01 6.91282213e-01
4.25565898e-01 -5.18923759e-01 1.35765982e+00 2.93138832e-01
-2.32635692e-01 -4.04258192e-01 -1.13190436e+00 -1.05196476e+00
5.50915718e-01 -6.44243598e-01 4.66118336e-01 9.45264339e-01
4.51648086e-01 1.66546211e-01 -9.86470222e-01 -1.24089219e-01
3.42800409e-01 -8.74085724e-01 1.43109822e+00 -1.37725186e+00
-7.92819440e-01 -5.09683669e-01 -8.67076278e-01 -7.30064273e-01
4.92602400e-02 -7.21944451e-01 -1.10706948e-01 -1.23235869e+00
-8.48581642e-02 -3.09579700e-01 -3.79095495e-01 9.61802840e-01
1.21823549e-01 5.46511352e-01 5.01254857e-01 2.94241551e-02
-1.00218308e+00 -1.79822639e-01 9.95400727e-01 -1.69582456e-01
-5.71820028e-02 2.30585281e-02 -8.51817489e-01 1.02287889e+00
1.32116127e+00 -6.79578960e-01 1.52247702e-03 -6.15139186e-01
7.96016872e-01 -2.06007913e-01 5.48273087e-01 -1.77558541e+00
4.88276809e-01 -1.24267921e-01 2.54917115e-01 -1.45821288e-01
4.57640469e-01 -3.10724586e-01 -2.46376753e-01 6.70953631e-01
-9.76997167e-02 1.77628487e-01 4.30770069e-01 1.49759769e-01
1.77840993e-01 -2.87954986e-01 6.82820976e-01 -6.90546930e-01
-8.06035995e-01 1.83050618e-01 -7.19528675e-01 9.50061977e-02
1.13077796e+00 -2.52891362e-01 -8.71908307e-01 -4.41271961e-01
-6.99610353e-01 3.19465309e-01 5.91200650e-01 2.75820762e-01
7.07580298e-02 -1.17743099e+00 -7.14231372e-01 2.27452561e-01
-2.86141872e-01 -1.84797585e-01 3.04527313e-01 4.91180897e-01
-1.10452938e+00 -2.72847675e-02 -9.03082848e-01 -1.54789403e-01
-1.14787185e+00 5.12423404e-02 7.97940612e-01 -8.62489343e-01
-4.65100199e-01 1.18516135e+00 -1.23964570e-01 -8.24817538e-01
1.64943218e-01 -1.21343210e-01 -8.25854391e-02 -3.91104102e-01
6.76937938e-01 4.04255211e-01 3.51950735e-01 -1.47723168e-01
-3.08382541e-01 1.18694440e-01 7.61603191e-03 -3.89634788e-01
1.53547347e+00 8.62681210e-01 -1.16233729e-01 2.65096247e-01
4.56870973e-01 -2.58448422e-02 -1.46000981e+00 3.77753764e-01
-5.24515025e-02 -2.92446911e-01 -5.46980202e-02 -8.87186050e-01
-8.97790730e-01 8.91918302e-01 1.88581288e-01 3.72134626e-01
4.20865953e-01 -2.76548147e-01 6.00890040e-01 6.75146699e-01
9.73893046e-01 -1.21623576e+00 4.64113176e-01 1.40268683e+00
5.45520663e-01 -9.54840362e-01 -6.63313717e-02 1.98605612e-01
-3.91770244e-01 1.06286526e+00 9.45255637e-01 -8.60313237e-01
4.32108313e-01 7.18506455e-01 1.14552066e-01 -4.43005234e-01
-9.81541276e-01 -3.05809110e-01 -3.95464927e-01 1.05687332e+00
4.49898988e-02 3.10088098e-01 1.94578633e-01 7.75606394e-01
-8.22391212e-01 -2.57218424e-02 8.27107906e-01 1.12086272e+00
-8.24007690e-01 -1.16893983e+00 -3.07601929e-01 5.10671616e-01
-6.40272975e-01 -8.83553252e-02 -8.39777350e-01 1.07680452e+00
4.43335682e-01 9.26733017e-01 -1.52497500e-01 -8.66550922e-01
4.24502909e-01 -4.45594490e-01 5.91447830e-01 -7.90364146e-01
-1.16231596e+00 -5.70759892e-01 3.86751950e-01 -7.93295383e-01
2.97800958e-01 -7.52300769e-03 -8.62980485e-01 -1.11659944e+00
-5.35369337e-01 -1.12626767e-02 2.59415329e-01 6.51433945e-01
-5.66144474e-02 5.79515100e-01 2.14822456e-01 -1.24727082e+00
-3.74369472e-01 -9.73958373e-01 -8.63227844e-01 -1.91063806e-01
-3.11232619e-02 -6.73042536e-01 -2.08888382e-01 -5.05294502e-01] | [3.4343791007995605, 1.3905863761901855] |
319d5b97-e71f-40c0-9c48-2ec456541677 | towards-single-camera-human-3d-kinematics | 2301.05435 | null | https://arxiv.org/abs/2301.05435v1 | https://arxiv.org/pdf/2301.05435v1.pdf | Towards Single Camera Human 3D-Kinematics | Markerless estimation of 3D Kinematics has the great potential to clinically diagnose and monitor movement disorders without referrals to expensive motion capture labs; however, current approaches are limited by performing multiple de-coupled steps to estimate the kinematics of a person from videos. Most current techniques work in a multi-step approach by first detecting the pose of the body and then fitting a musculoskeletal model to the data for accurate kinematic estimation. Errors in training data of the pose detection algorithms, model scaling, as well the requirement of multiple cameras limit the use of these techniques in a clinical setting. Our goal is to pave the way toward fast, easily applicable and accurate 3D kinematic estimation \xdeleted{in a clinical setting}. To this end, we propose a novel approach for direct 3D human kinematic estimation D3KE from videos using deep neural networks. Our experiments demonstrate that the proposed end-to-end training is robust and outperforms 2D and 3D markerless motion capture based kinematic estimation pipelines in terms of joint angles error by a large margin (35\% from 5.44 to 3.54 degrees). We show that D3KE is superior to the multi-step approach and can run at video framerate speeds. This technology shows the potential for clinical analysis from mobile devices in the future. | ['Frans C. T. van der Helm', 'Jan van Gemert', 'Ajay Seth', 'Xucong Zhang', 'Wei-Tse Yang', 'Marian Bittner'] | 2023-01-13 | null | null | null | null | ['markerless-motion-capture'] | ['computer-vision'] | [-1.44604012e-01 -9.03183743e-02 -2.04415441e-01 -5.03636673e-02
-1.00496471e+00 -3.92570704e-01 -1.83545314e-02 -1.01993144e-01
-8.28851461e-01 3.87105644e-01 1.18729010e-01 -1.74197927e-01
-7.00789737e-03 -5.44716455e-02 -7.03181148e-01 -2.10999310e-01
-3.13275129e-01 7.62870610e-01 2.20557243e-01 -8.29404071e-02
-2.32626081e-01 6.45737410e-01 -8.11709344e-01 -1.67694375e-01
3.15515459e-01 6.10722899e-01 2.02503172e-03 1.20735466e+00
6.29451871e-01 4.24396068e-01 -4.43472564e-01 -8.64003822e-02
2.28316963e-01 -2.68992037e-01 -6.64384544e-01 1.55174971e-01
6.56083882e-01 -9.53968167e-01 -3.14279526e-01 4.37187254e-01
1.11659682e+00 2.84202304e-02 1.20641388e-01 -1.15363121e+00
2.91392416e-01 -2.33807489e-02 -4.57795739e-01 3.42263162e-01
7.74789810e-01 3.34738255e-01 6.44958839e-02 -6.60138667e-01
9.47676241e-01 8.47791731e-01 1.38197470e+00 5.72765648e-01
-8.89682710e-01 -2.86357999e-01 -3.37410778e-01 8.49471390e-02
-1.35612559e+00 -3.72974187e-01 6.07221723e-01 -5.80882967e-01
1.18360734e+00 1.30910754e-01 1.09994280e+00 1.10993767e+00
4.49298292e-01 6.01168334e-01 6.61173940e-01 -2.08374113e-01
4.80628014e-03 -4.70654756e-01 -8.12597200e-02 8.17671597e-01
3.80928963e-01 -1.52873412e-01 -4.76670325e-01 -3.07560507e-02
1.19441211e+00 1.15807252e-02 -2.19042599e-01 -4.76182401e-01
-1.37274945e+00 2.32057855e-01 1.97585881e-01 -1.38607696e-02
-5.08064628e-01 8.14849019e-01 4.88151789e-01 -1.99699804e-01
2.41545081e-01 2.93604374e-01 -4.97367501e-01 -9.53425825e-01
-1.09120905e+00 6.37715876e-01 4.01906222e-01 7.63634562e-01
-1.35168970e-01 -9.62341353e-02 2.07013115e-01 3.20116103e-01
3.55438471e-01 5.76744378e-01 4.29008096e-01 -1.33043981e+00
5.90759337e-01 5.01958847e-01 3.75076085e-01 -8.92509460e-01
-1.19630015e+00 -3.26486647e-01 -3.47799420e-01 2.61114985e-01
7.51787424e-01 -5.64627349e-01 -7.88510084e-01 1.45775652e+00
8.78616929e-01 2.50444233e-01 -4.20215100e-01 1.34671342e+00
4.22487944e-01 8.25447217e-03 -7.29696974e-02 1.34492382e-01
1.36081278e+00 -5.72450340e-01 -5.71068466e-01 -4.55826283e-01
1.10251617e+00 -7.33812034e-01 7.04137146e-01 5.17433882e-01
-1.41725361e+00 -4.01345402e-01 -1.00911987e+00 -1.11260861e-01
4.01892066e-01 3.66621882e-01 5.06752968e-01 6.88422143e-01
-8.79702568e-01 7.85594225e-01 -1.76790631e+00 -4.11438763e-01
2.15422109e-01 9.02327359e-01 -6.31326377e-01 8.52726251e-02
-9.09843683e-01 1.09994626e+00 1.20756753e-01 4.90498990e-01
-3.45847338e-01 -7.12898970e-01 -8.83470595e-01 -4.98422623e-01
3.28656971e-01 -1.03482831e+00 1.43146980e+00 -3.77375633e-01
-1.59325159e+00 7.99587548e-01 3.56560908e-02 -2.67845780e-01
1.03724992e+00 -1.04334068e+00 -3.41480635e-02 6.32989705e-01
-1.42149687e-01 3.80209118e-01 5.08693695e-01 -4.07148600e-01
-2.33925879e-01 -4.82737631e-01 -2.13297859e-01 4.28547263e-01
1.91561114e-02 6.85287490e-02 -8.70396793e-01 -4.53889370e-01
2.48292312e-01 -1.62376547e+00 -3.08526188e-01 5.82226634e-01
-2.34517202e-01 3.36055070e-01 4.93561476e-01 -1.05260122e+00
8.73708069e-01 -1.74036431e+00 1.87911272e-01 6.93105459e-02
3.96438599e-01 5.81762731e-01 5.18310249e-01 1.77601337e-01
-8.44578817e-03 -3.19559157e-01 2.78041780e-01 -2.25611910e-01
-3.90585303e-01 -6.47462532e-02 5.40749013e-01 8.96052778e-01
-1.49615124e-01 1.01911628e+00 -8.60007107e-01 -5.21837533e-01
4.86045420e-01 8.30165803e-01 -5.41016459e-01 1.05668619e-01
3.01912993e-01 7.42289901e-01 -2.89695263e-01 5.45081198e-01
2.30480075e-01 -3.63425046e-01 3.76086622e-01 -2.72125155e-01
1.88625172e-01 1.74019322e-01 -1.51133060e+00 2.14291835e+00
-2.93362379e-01 5.91219902e-01 2.24155962e-01 -6.23855174e-01
2.82689929e-01 5.62619448e-01 1.05185759e+00 -4.93447818e-02
4.39329505e-01 3.60852689e-01 1.23414762e-01 -8.09622467e-01
2.20173001e-01 -2.89410464e-02 -9.82479677e-02 3.29798788e-01
-1.48599997e-01 1.09872073e-01 -1.87949210e-01 -1.06375359e-01
1.26788771e+00 7.19946325e-01 3.25414866e-01 2.29999110e-01
-3.33789550e-02 3.32741052e-01 3.84212792e-01 3.62498373e-01
-5.55212080e-01 8.09231758e-01 1.04124032e-01 -4.11158085e-01
-1.18282342e+00 -1.14237380e+00 3.06544632e-01 4.80875522e-01
-1.89103186e-02 -4.13649887e-01 -8.63415420e-01 -2.14166030e-01
1.62082419e-01 -1.23563938e-01 -2.49513909e-01 -2.49155685e-01
-1.16407776e+00 -4.60479498e-01 8.57498050e-01 1.13044298e+00
2.30983093e-01 -3.14130634e-01 -1.11733413e+00 3.75290394e-01
-2.42499173e-01 -1.44380581e+00 -5.03405511e-01 -2.68935919e-01
-1.13119805e+00 -1.19326568e+00 -1.08074164e+00 -5.49207807e-01
5.12220919e-01 6.05074763e-02 6.24263048e-01 -2.30682209e-01
-4.48090464e-01 6.92533791e-01 -1.47958383e-01 -2.05795944e-01
-1.19355910e-01 -6.56850561e-02 5.99462688e-01 -6.33426905e-01
9.15347636e-02 -4.53939855e-01 -1.04524481e+00 2.69738913e-01
-2.47670665e-01 6.74550906e-02 5.55170536e-01 3.97947252e-01
5.80080807e-01 -6.02664351e-01 4.93400730e-02 -2.79102206e-01
4.99674410e-01 -1.81168035e-01 -3.39625686e-01 -1.82457209e-01
-4.64041591e-01 -2.37678766e-01 3.13075155e-01 -7.31710017e-01
-6.27290785e-01 5.86957633e-01 -3.48111778e-01 -7.20047355e-01
-1.55359283e-01 4.94768769e-01 3.24066222e-01 -1.58588812e-01
6.94725394e-01 -1.89736456e-01 5.77546239e-01 -4.12960529e-01
1.89841613e-01 4.53545392e-01 1.06874466e+00 -2.26924926e-01
4.02576894e-01 7.12188125e-01 3.60266745e-01 -6.86981082e-01
-2.91199684e-01 -6.96059883e-01 -9.68506575e-01 -5.98738253e-01
8.56198728e-01 -1.05875444e+00 -9.69919860e-01 6.83439195e-01
-1.01467347e+00 -3.69537920e-01 7.22990781e-02 1.24744165e+00
-8.67110252e-01 6.13894165e-01 -8.05480599e-01 -6.24819338e-01
-6.83621287e-01 -1.12979031e+00 1.30979562e+00 -8.86430964e-02
-8.30528140e-01 -7.17124343e-01 1.79495990e-01 5.20151913e-01
1.46745622e-01 7.50029564e-01 6.46160841e-02 -2.14167714e-01
-1.11541912e-01 -8.81488323e-01 3.60454500e-01 -1.27222776e-01
1.23971989e-02 -2.35818192e-01 -4.29235846e-01 -4.59713280e-01
-1.70263037e-01 -2.26535350e-01 1.62720829e-02 8.46510470e-01
4.78534371e-01 1.37983754e-01 -4.72927570e-01 7.55774617e-01
1.11586857e+00 -1.41873240e-01 4.02647048e-01 5.97667933e-01
1.11949658e+00 3.37092608e-01 6.87862873e-01 3.66728127e-01
4.11590308e-01 1.15179586e+00 2.80735761e-01 -2.92748380e-02
-1.35668173e-01 3.37247141e-02 3.98993313e-01 7.60928988e-01
-4.35609072e-01 1.92863405e-01 -1.26458621e+00 4.32259977e-01
-1.84086812e+00 -6.68606639e-01 -4.48774427e-01 2.28107643e+00
5.02820253e-01 3.14994156e-01 6.06582105e-01 1.96149319e-01
6.16749644e-01 -4.75288004e-01 -5.40729642e-01 -2.59180039e-01
6.66496873e-01 5.48817404e-02 7.30098724e-01 3.47431362e-01
-9.99707282e-01 5.15185356e-01 6.45375490e+00 1.11919068e-01
-1.36038077e+00 5.89356907e-02 -6.10607602e-02 -8.03936243e-01
5.88236749e-01 -2.54814416e-01 -7.68374026e-01 3.76279324e-01
1.22741258e+00 3.18378955e-01 -1.63666561e-01 8.81127477e-01
7.37915397e-01 -3.04164141e-01 -1.09130299e+00 1.20903409e+00
-1.29624354e-02 -1.22484863e+00 -7.28423715e-01 2.18345553e-01
2.92975962e-01 2.52565384e-01 -2.43095458e-01 -2.20360622e-01
-1.33538172e-01 -7.63638496e-01 5.46549439e-01 6.95587873e-01
8.36037040e-01 -6.44357383e-01 5.84720194e-01 5.76913536e-01
-1.10214508e+00 3.46856773e-01 1.45630136e-01 -2.33745977e-01
6.38531089e-01 3.12184215e-01 -1.22663438e+00 3.86930943e-01
5.70509255e-01 4.48389351e-01 -3.38942975e-01 1.17098761e+00
-7.32621253e-02 4.63242441e-01 -6.57590926e-01 7.85837844e-02
6.57903105e-02 1.74386337e-01 6.94571972e-01 1.08055973e+00
4.15074885e-01 -6.44687042e-02 2.55603433e-01 1.71151042e-01
3.15360695e-01 -2.10489020e-01 -4.55293804e-01 1.59301654e-01
7.69899860e-02 1.12576699e+00 -7.75089860e-01 -1.22547321e-01
-2.02468023e-01 1.17647803e+00 1.37889341e-01 -1.91461682e-01
-1.02237451e+00 -2.14279905e-01 5.75509846e-01 3.88892025e-01
2.47543156e-02 -8.79740059e-01 -8.93197879e-02 -1.02245331e+00
5.23208857e-01 -8.35403919e-01 2.98808157e-01 -9.15687442e-01
-6.14588082e-01 5.61496764e-02 2.07640618e-01 -1.58113575e+00
-1.00499654e+00 -7.84875572e-01 -2.83900976e-01 7.51398027e-01
-7.15873957e-01 -1.12123907e+00 -4.32266802e-01 4.90400612e-01
2.65604019e-01 4.00161445e-01 6.59719408e-01 4.06254441e-01
-5.49376667e-01 4.40642893e-01 -8.13524574e-02 3.12662542e-01
8.63072455e-01 -1.09604442e+00 6.85710907e-01 9.16912675e-01
-9.16616023e-02 8.36611748e-01 7.79708982e-01 -8.64354134e-01
-1.90183055e+00 -6.67459607e-01 6.31380379e-01 -1.00764632e+00
4.89142209e-01 -1.61957860e-01 -4.03833747e-01 7.84576774e-01
-5.54439187e-01 3.49284768e-01 5.66004753e-01 -1.60176262e-01
3.34872395e-01 6.74911514e-02 -9.63576794e-01 5.91109097e-01
1.03145838e+00 -3.84427786e-01 -5.05626202e-01 3.23623508e-01
2.87989467e-01 -1.30679989e+00 -1.22228909e+00 2.24597991e-01
1.24613035e+00 -4.98639435e-01 1.07300580e+00 -5.63821197e-01
1.21241815e-01 -2.17447028e-01 4.02167052e-01 -9.44067955e-01
-8.08754563e-02 -8.64059687e-01 -5.06502211e-01 4.84040171e-01
-6.73444569e-02 -6.11527823e-02 1.32005167e+00 8.18215787e-01
-1.56273544e-01 -9.50937331e-01 -1.17726481e+00 -8.23178351e-01
-1.94263577e-01 -9.97833014e-01 -1.24036647e-01 5.13252735e-01
1.66355669e-01 -4.57255356e-02 -6.10886216e-01 3.83317202e-01
4.91925091e-01 -3.02228928e-01 1.10629153e+00 -8.33843708e-01
-4.52965796e-01 -4.87906858e-02 -9.02113914e-01 -9.93010998e-01
-4.51848119e-01 -4.34965998e-01 -5.89073226e-02 -1.63297892e+00
-2.23454759e-01 5.01265749e-02 3.03948432e-01 2.12821782e-01
-1.49305165e-01 4.34467286e-01 1.36995882e-01 2.93918073e-01
-4.66356963e-01 -1.93066835e-01 1.08681214e+00 3.79426688e-01
-3.58791590e-01 1.11111283e-01 -9.96350050e-02 1.04776275e+00
5.41502774e-01 -4.55376357e-01 -4.07184273e-01 -5.27828455e-01
2.01846927e-01 2.20873743e-01 7.70483434e-01 -1.33094597e+00
2.72542685e-01 2.02635095e-01 8.00213218e-01 -5.98416030e-01
5.75613141e-01 -6.71523631e-01 4.28593576e-01 8.60505998e-01
1.23739809e-01 3.89538854e-01 2.57763684e-01 2.90651083e-01
2.71756023e-01 1.47753507e-01 4.87585574e-01 -3.11433464e-01
-6.02610826e-01 2.47118399e-01 -5.19032896e-01 1.53322265e-01
9.20664012e-01 -8.77535760e-01 1.13121428e-01 -4.71057981e-01
-1.17604458e+00 7.23401979e-02 4.09155756e-01 2.89794415e-01
5.69158256e-01 -1.17060947e+00 -3.67391765e-01 -1.69197857e-01
-1.01258263e-01 8.26331750e-02 2.57331163e-01 1.49113369e+00
-1.13023436e+00 4.68927026e-01 -3.29596162e-01 -1.00465906e+00
-1.51180971e+00 7.48899728e-02 6.07144356e-01 -2.71046102e-01
-1.01591325e+00 5.50524712e-01 -8.25231254e-01 -2.82525152e-01
5.82265183e-02 -4.15567398e-01 1.88032761e-01 -3.21820498e-01
3.29920441e-01 8.52845669e-01 3.81404936e-01 -7.86363900e-01
-6.81494355e-01 7.33425081e-01 1.48405939e-01 -3.67413789e-01
1.23426628e+00 -3.25887382e-01 4.87086564e-01 2.27940887e-01
1.20248652e+00 -1.03975520e-01 -1.44189537e+00 2.59448647e-01
-9.35234278e-02 -2.75356650e-01 8.92948955e-02 -5.70168912e-01
-7.37234652e-01 6.91179335e-01 1.02599907e+00 -5.34679651e-01
7.97710896e-01 -8.63147527e-02 1.17655754e+00 3.45102936e-01
3.34446162e-01 -1.27029407e+00 2.58465316e-02 8.05977732e-02
4.83294427e-01 -9.52893913e-01 3.76586229e-01 -4.06991482e-01
-4.64542001e-01 1.20014858e+00 5.51968873e-01 -3.11695933e-01
3.80376577e-01 4.48577106e-01 2.59649724e-01 -2.68361866e-01
-2.53266454e-01 1.17444219e-02 5.30112743e-01 6.45304024e-01
6.14201665e-01 -1.10757560e-01 -2.85078406e-01 9.64271352e-02
-1.49284035e-01 5.39726257e-01 5.88911891e-01 1.43030667e+00
-6.77609146e-02 -8.91671598e-01 -6.38596714e-01 2.52404571e-01
-8.03715110e-01 4.73205417e-01 -1.59144193e-01 1.26940799e+00
1.34051085e-01 6.76327109e-01 -1.30596802e-01 -3.25607210e-01
5.40555298e-01 1.96354091e-02 8.68775666e-01 -5.03666103e-01
-6.24183893e-01 4.90155816e-01 3.98570240e-01 -1.12540758e+00
-3.83208364e-01 -9.02472258e-01 -1.50990021e+00 -1.54424876e-01
-1.79011568e-01 -4.77994740e-01 7.95212448e-01 1.07360470e+00
5.94009638e-01 4.34250057e-01 -1.37028143e-01 -1.23381269e+00
-4.77334946e-01 -7.56699204e-01 -2.44096264e-01 3.60589623e-01
3.30963463e-01 -7.80965865e-01 2.47297034e-01 1.25553846e-01] | [7.071767807006836, -0.5534766316413879] |
7883e298-0cb2-4123-b383-6e8c5a129638 | sar-based-landslide-classification | 2211.09927 | null | https://arxiv.org/abs/2211.09927v1 | https://arxiv.org/pdf/2211.09927v1.pdf | SAR-based landslide classification pretraining leads to better segmentation | Rapid assessment after a natural disaster is key for prioritizing emergency resources. In the case of landslides, rapid assessment involves determining the extent of the area affected and measuring the size and location of individual landslides. Synthetic Aperture Radar (SAR) is an active remote sensing technique that is unaffected by weather conditions. Deep Learning algorithms can be applied to SAR data, but training them requires large labeled datasets. In the case of landslides, these datasets are laborious to produce for segmentation, and often they are not available for the specific region in which the event occurred. Here, we study how deep learning algorithms for landslide segmentation on SAR products can benefit from pretraining on a simpler task and from data from different regions. The method we explore consists of two training stages. First, we learn the task of identifying whether a SAR image contains any landslides or not. Then, we learn to segment in a sparsely labeled scenario where half of the data do not contain landslides. We test whether the inclusion of feature embeddings derived from stage-1 helps with landslide detection in stage-2. We find that it leads to minor improvements in the Area Under the Precision-Recall Curve, but also to a significantly lower false positive rate in areas without landslides and an improved estimate of the average number of landslide pixels in a chip. A more accurate pixel count allows to identify the most affected areas with higher confidence. This could be valuable in rapid response scenarios where prioritization of resources at a global scale is important. We make our code publicly available at https://github.com/VMBoehm/SAR-landslide-detection-pretraining. | ['Raul Ramos-Pollan', 'Siddha Ganju', 'Freddie Kalaitzis', 'Edoardo Nemni', 'Ioannis Prapas', 'Ragini Bal Mahesh', 'Wei Ji Leong', 'Vanessa Böhm'] | 2022-11-17 | null | null | null | null | ['landslide-segmentation'] | ['computer-vision'] | [ 4.81610060e-01 -2.71919761e-02 -7.76940119e-03 -3.56149167e-01
-9.46696579e-01 -6.93348289e-01 3.59973490e-01 3.95646334e-01
-6.06417358e-01 5.95270395e-01 2.39182308e-01 -7.88254380e-01
-2.12289274e-01 -1.31283903e+00 -5.55042744e-01 -1.00906336e+00
-2.37417370e-01 6.16407514e-01 5.62359355e-02 -2.00243905e-01
-1.26784012e-01 1.10015845e+00 -1.21497822e+00 2.56865472e-01
5.80321848e-01 9.95881021e-01 3.64960790e-01 5.23928761e-01
3.32185656e-01 3.61881256e-01 -3.90650541e-01 4.93813127e-01
6.49887264e-01 -1.53767467e-01 -7.58661509e-01 1.41906455e-01
2.79711276e-01 -7.86416531e-01 -1.22570410e-01 7.89308250e-01
4.98228848e-01 -1.74474224e-01 7.72036493e-01 -4.59215283e-01
1.81253999e-01 4.83484089e-01 -7.85025716e-01 4.37805980e-01
-3.41269225e-01 2.65004545e-01 8.19756746e-01 -8.29925299e-01
3.39880675e-01 8.88429761e-01 8.10444415e-01 -2.29297310e-01
-1.25829327e+00 -4.65987772e-01 -2.24106431e-01 -6.10765517e-02
-1.41969728e+00 -4.76197958e-01 3.54104400e-01 -6.58160627e-01
8.11677635e-01 1.19560845e-01 3.53489429e-01 1.16645969e-01
-3.49654704e-02 5.72122991e-01 9.16978896e-01 -3.26361775e-01
3.04187149e-01 -2.94996172e-01 4.52010632e-02 3.61328930e-01
5.29859483e-01 3.16815674e-01 1.03707165e-01 4.52136695e-02
5.55536151e-01 2.12809548e-01 -2.40424544e-01 -5.23122288e-02
-9.99417186e-01 1.07431579e+00 8.17742169e-01 4.47160274e-01
-8.23042870e-01 -1.71082281e-02 1.94554195e-01 2.18207061e-01
6.63202524e-01 5.17350376e-01 -4.02535111e-01 4.61126566e-01
-1.51161289e+00 1.92783073e-01 2.17757955e-01 -8.62694755e-02
1.19460404e+00 1.11633852e-01 9.10912678e-02 6.16136670e-01
-4.88400944e-02 9.54425454e-01 -1.26826555e-01 -6.90395892e-01
4.15741324e-01 6.10024750e-01 3.51023465e-01 -9.51032698e-01
-5.76835990e-01 -4.92528409e-01 -7.02835441e-01 4.33464527e-01
2.88914442e-01 -5.04482508e-01 -1.24915385e+00 1.20856571e+00
2.12777838e-01 -2.94140518e-01 7.40074888e-02 1.06034660e+00
1.30720794e-01 1.00315225e+00 1.53412402e-01 -7.35392747e-03
1.31392395e+00 -2.24301308e-01 -1.81002647e-01 -8.66923869e-01
9.30346191e-01 -5.16454637e-01 6.32616103e-01 1.01150118e-01
-2.78942496e-01 -3.44247043e-01 -1.04907703e+00 4.31927681e-01
-5.44365942e-01 5.22685945e-01 6.39652729e-01 3.58252019e-01
-7.24187076e-01 4.56538618e-01 -9.09071445e-01 -3.42897326e-01
7.75917292e-01 1.59310505e-01 -3.77010405e-01 -2.32026562e-01
-1.17341399e+00 9.65625584e-01 3.23817074e-01 5.36227703e-01
-9.01979387e-01 -6.56390309e-01 -9.08185482e-01 2.58991390e-01
2.61228263e-01 1.89881116e-01 7.29372919e-01 -8.24997187e-01
-1.76484019e-01 9.88506496e-01 1.78299889e-01 -7.21163690e-01
2.70513594e-01 -3.71366501e-01 -1.57147661e-01 2.57240325e-01
5.07801890e-01 7.35504448e-01 4.89873618e-01 -1.06344247e+00
-9.09421086e-01 -7.06593633e-01 -1.39126837e-01 1.92231789e-01
-1.08177848e-02 7.84528926e-02 3.11524123e-01 -4.61939543e-01
2.65325785e-01 -7.77755499e-01 -4.82719213e-01 -5.31849824e-02
5.40519990e-02 2.53994584e-01 8.69235873e-01 -9.10047650e-01
9.63233531e-01 -2.22737956e+00 -4.24569190e-01 2.19221607e-01
-1.90023765e-01 6.39859915e-01 -4.16018367e-02 5.42244434e-01
6.23168424e-03 3.18631023e-01 -9.02714193e-01 4.75173593e-01
-2.95222014e-01 1.28398240e-01 -6.30352855e-01 5.36057770e-01
6.72366321e-01 5.20220220e-01 -6.62434757e-01 -1.86680168e-01
1.67544916e-01 2.44181335e-01 -2.30605397e-02 1.94775879e-01
-1.30743936e-01 1.99414596e-01 -3.72229636e-01 8.36184084e-01
9.96133626e-01 8.83726850e-02 1.13280468e-01 -2.22599179e-01
-5.01282990e-01 1.96028292e-01 -1.08338368e+00 8.85408878e-01
-4.23990071e-01 7.91908562e-01 2.18575060e-01 -1.39133847e+00
1.05503964e+00 1.98784709e-01 4.50475514e-01 -8.51889908e-01
-1.67596713e-01 2.29210392e-01 -6.21162355e-02 -5.71898341e-01
4.70272958e-01 -4.16837215e-01 -1.88746110e-01 6.98846877e-01
-4.65617895e-01 -1.55144900e-01 9.96767804e-02 -1.55268744e-01
1.23675418e+00 -3.15861911e-01 1.84498861e-01 -3.46851945e-01
-8.57112557e-02 6.84791982e-01 5.91787696e-01 6.77639902e-01
5.76048829e-02 5.54363251e-01 3.08993995e-01 -8.51662099e-01
-9.34104979e-01 -1.16644168e+00 -4.35534805e-01 8.96188736e-01
-2.18394324e-01 4.70641106e-01 -3.39186341e-01 -5.11938632e-01
9.45838317e-02 7.91205525e-01 -4.52297240e-01 7.33245015e-02
-5.56379914e-01 -1.39145148e+00 6.02066755e-01 6.08676910e-01
8.27548862e-01 -1.24732673e+00 -1.18847024e+00 2.06436098e-01
-1.35537699e-01 -8.20128441e-01 2.74848759e-01 6.07105792e-01
-1.08489871e+00 -1.06050408e+00 -6.96219623e-01 -6.44956112e-01
7.60165691e-01 4.79832679e-01 8.58018756e-01 -8.75815749e-02
-4.42831486e-01 8.65255073e-02 -4.61668253e-01 -3.46097082e-01
-2.33681917e-01 4.51116404e-03 -1.85933873e-01 4.40951157e-03
5.99679589e-01 -2.54392415e-01 -7.28072464e-01 3.62594724e-01
-1.08437014e+00 -1.69101104e-01 9.21835780e-01 5.55387139e-01
6.42268062e-01 4.79738623e-01 4.75511491e-01 -8.29938173e-01
5.25967591e-02 -6.39530420e-01 -6.39560699e-01 1.68294922e-01
-3.80748332e-01 -6.37911782e-02 5.87561168e-02 8.01553056e-02
-8.74690354e-01 4.78394002e-01 -1.51724800e-01 1.26597151e-01
-6.72785163e-01 1.13299632e+00 1.41043499e-01 2.71758467e-01
9.21275258e-01 -4.87796031e-02 -3.72435339e-02 -4.35704529e-01
-1.08122945e-01 9.72139597e-01 4.25157279e-01 -6.93749264e-03
8.74190152e-01 7.72251725e-01 -2.08550826e-01 -1.19330728e+00
-8.23296189e-01 -8.28167260e-01 -6.74290836e-01 -1.69865191e-01
7.52996087e-01 -1.11536646e+00 2.84681767e-01 3.82998258e-01
-7.69605160e-01 -8.01406860e-01 -2.77395248e-01 7.39238918e-01
-1.88086316e-01 3.64930891e-02 1.32179093e-02 -6.98547304e-01
-4.28603649e-01 -7.45008945e-01 8.63071680e-01 1.09370016e-01
-1.31012991e-01 -7.14166880e-01 2.03301087e-01 3.15022320e-01
4.99509692e-01 3.70231241e-01 8.35291445e-01 -5.23317575e-01
-3.10163170e-01 -4.79028165e-01 -4.96421784e-01 4.21473950e-01
4.36170369e-01 -3.37441802e-01 -8.41346622e-01 -2.77977288e-01
-5.78310713e-02 -3.08039039e-01 1.45259535e+00 6.39063001e-01
6.17470741e-01 -2.70316780e-01 -3.60004514e-01 3.54830801e-01
1.63848138e+00 2.62603045e-01 8.92447412e-01 3.53631824e-01
2.86057025e-01 8.94856513e-01 1.06758916e+00 3.87345254e-01
-7.94257075e-02 3.35757315e-01 5.23733020e-01 -5.38735211e-01
7.51817971e-02 2.24267244e-01 4.12828207e-01 -2.13111922e-01
5.32820821e-02 -1.40324533e-01 -1.46995556e+00 1.19835949e+00
-1.44719625e+00 -1.16409683e+00 -4.36578579e-02 2.32712340e+00
3.64189893e-01 -8.13135207e-02 -7.71608949e-02 3.96381527e-01
7.02066302e-01 4.37741756e-01 -4.92240161e-01 -9.89098996e-02
-2.01655343e-01 3.72450709e-01 1.10368574e+00 5.81054866e-01
-1.37735355e+00 9.90346730e-01 5.23736238e+00 5.03910244e-01
-1.41540992e+00 -9.18984860e-02 7.68262565e-01 2.17529014e-01
-1.16202205e-01 1.81824729e-01 -7.31245518e-01 5.24200276e-02
8.82702410e-01 4.64583904e-01 -3.19114029e-02 5.26464820e-01
7.51737356e-01 -6.75661981e-01 -4.75465894e-01 3.96370232e-01
-3.06966931e-01 -1.32879353e+00 -1.00172773e-01 1.33382857e-01
4.59130466e-01 4.87334579e-01 -2.03674763e-01 -3.83076407e-02
3.03306282e-01 -9.30642188e-01 3.31176221e-01 3.18037122e-01
9.33189273e-01 -5.86498916e-01 9.86218870e-01 2.60822922e-01
-1.05676877e+00 -2.28636354e-01 -3.84138882e-01 -1.82080790e-01
3.27974916e-01 8.55448663e-01 -1.09487557e+00 1.18060470e-01
7.03014910e-01 4.87520635e-01 -4.27082390e-01 1.03620839e+00
-3.64514709e-01 1.09098351e+00 -5.89891195e-01 4.53173637e-01
4.93654847e-01 -1.32687345e-01 3.26599002e-01 1.05871153e+00
5.74540794e-01 4.67262238e-01 9.46283638e-02 7.78764665e-01
3.50475252e-01 -2.50227213e-01 -7.80707538e-01 -2.08294198e-01
4.47941869e-01 1.14496219e+00 -9.35130715e-01 -6.56691641e-02
-1.43749982e-01 5.49703777e-01 -1.46906391e-01 3.17247570e-01
-4.24975276e-01 -5.18211424e-01 4.50405329e-01 7.04053819e-01
5.59638023e-01 -3.79243702e-01 -3.14146549e-01 -5.87982237e-01
-2.10944667e-01 -6.00644052e-01 5.34455538e-01 -7.29327500e-01
-7.02737033e-01 3.51799816e-01 2.99089029e-02 -1.10722160e+00
-1.47193536e-01 -5.37253499e-01 -8.95771682e-01 7.67389119e-01
-1.50702965e+00 -1.03040576e+00 -4.14585620e-01 -3.01687922e-02
3.45160633e-01 6.69175573e-03 7.20904589e-01 1.68152466e-01
-2.55943328e-01 -1.25194773e-01 1.86585814e-01 4.93917376e-01
3.79640579e-01 -9.09395576e-01 2.89598465e-01 1.13652480e+00
1.13599852e-01 -7.00816438e-02 5.98307371e-01 -7.88867235e-01
-7.76197314e-01 -1.38700044e+00 9.66695011e-01 1.62217423e-01
5.40461957e-01 6.57916293e-02 -8.41135919e-01 6.60331786e-01
-3.29963863e-01 -2.16022596e-01 5.38489223e-01 -5.16613871e-02
6.59015626e-02 -5.03716111e-01 -1.25547981e+00 5.94066121e-02
3.54413718e-01 -3.95049036e-01 -4.33379501e-01 4.03150141e-01
1.35589227e-01 2.28428796e-01 -5.78355849e-01 6.75768137e-01
3.69464785e-01 -8.52058291e-01 8.13837290e-01 -2.28975371e-01
4.70868856e-01 -3.19044650e-01 -4.95719790e-01 -1.28298056e+00
-4.00571644e-01 3.42878938e-01 7.77610123e-01 8.90862107e-01
6.38375163e-01 -4.77833509e-01 8.35217953e-01 1.41918093e-01
-1.72786653e-01 -3.59229147e-01 -6.99610829e-01 -8.01412046e-01
2.33316988e-01 -2.94656754e-01 4.03862566e-01 6.63370848e-01
-7.35527873e-01 -1.87649950e-02 3.73224579e-02 9.66592073e-01
4.96446162e-01 3.46708387e-01 5.02030313e-01 -1.30293405e+00
1.91976056e-01 -7.51528516e-02 -2.28981987e-01 -5.41522980e-01
-2.91668415e-01 -5.69713295e-01 3.35760117e-01 -1.88231492e+00
-4.01542895e-02 -7.93934882e-01 -6.80076256e-02 1.00351608e+00
6.53592497e-02 4.01663631e-01 -2.29973301e-01 3.12114328e-01
7.97926113e-02 2.09571436e-01 3.78887773e-01 -5.25046527e-01
-4.09434766e-01 2.04337671e-01 -4.08489734e-01 6.25552952e-01
1.17031240e+00 -8.78693521e-01 3.46011743e-02 -8.08119833e-01
3.02012056e-01 7.22545013e-02 6.78685904e-01 -1.40328836e+00
1.23069054e-02 -3.46630245e-01 3.25228930e-01 -1.00284708e+00
5.14848018e-03 -7.68262029e-01 2.47946307e-01 7.50250220e-01
2.88676880e-02 -3.05791169e-01 4.09806967e-01 2.62371272e-01
-2.48812467e-01 -4.80506599e-01 9.34252083e-01 -1.20961092e-01
-9.67224360e-01 2.70421147e-01 -9.62487221e-01 -2.37474695e-01
7.47161627e-01 -1.90652177e-01 -9.65184569e-02 -1.78541809e-01
-6.12048328e-01 2.48222440e-01 3.78785759e-01 -2.22663116e-02
5.25390208e-01 -8.35958838e-01 -1.12335289e+00 2.54806310e-01
7.52811879e-02 9.23663825e-02 2.71813154e-01 7.35294044e-01
-1.11496782e+00 3.18167478e-01 -3.71466875e-01 -4.60688323e-01
-1.12576473e+00 7.78880343e-02 4.28252548e-01 -2.94944376e-01
-2.72542685e-01 6.39154494e-01 1.53609008e-01 -3.47650707e-01
-2.05724061e-01 -1.19370565e-01 -2.10335478e-01 7.83164203e-01
5.97851336e-01 2.63931572e-01 2.18750566e-01 -8.31893682e-01
-4.90544081e-01 4.76681739e-01 2.76049376e-02 -2.51667738e-01
1.77284551e+00 6.17459081e-02 -3.33480127e-02 1.45699039e-01
7.96330452e-01 -1.72660053e-01 -1.33599293e+00 -1.62808001e-01
2.69820429e-02 -3.27065408e-01 7.52643943e-01 -9.44146633e-01
-1.33301103e+00 1.14482164e+00 1.04701316e+00 -2.42997482e-02
1.20943487e+00 2.48460080e-02 5.33043921e-01 6.02649033e-01
2.10683346e-01 -1.05288076e+00 -3.74367505e-01 3.49864691e-01
8.70907545e-01 -1.41942596e+00 3.30328256e-01 1.58014670e-01
-5.47490776e-01 9.67366815e-01 1.07589282e-01 -1.05471551e-01
7.98003435e-01 2.19889820e-01 2.93969274e-01 -5.27721047e-01
-1.19901538e-01 -7.33893454e-01 -2.04479724e-01 6.53144240e-01
5.25273383e-02 4.46445167e-01 -1.47967353e-01 2.02511981e-01
1.64529353e-01 -2.02875748e-01 4.23016489e-01 9.76558268e-01
-1.11446488e+00 -6.16223454e-01 -6.23029590e-01 8.54563475e-01
-2.18144014e-01 -1.46380454e-01 -4.13468808e-01 6.90624475e-01
1.66592211e-01 9.01926219e-01 5.87474048e-01 -1.94171295e-01
9.05012190e-02 -2.05351070e-01 -7.08239228e-02 -7.40074933e-01
-2.14343831e-01 4.24199365e-02 1.28083006e-01 -1.96502402e-01
-4.56065595e-01 -8.99280727e-01 -1.23709798e+00 -1.66438311e-01
-2.48230398e-01 2.06771910e-01 6.02438688e-01 8.80403221e-01
1.82570994e-01 -7.21365288e-02 6.79831922e-01 -9.95414436e-01
-4.03029501e-01 -1.00194073e+00 -8.39364290e-01 -1.05046600e-01
4.13447827e-01 -4.76730973e-01 -5.04494786e-01 -4.86636348e-02] | [9.442658424377441, -1.4403406381607056] |
c190a901-37d9-4fdd-944b-853915c24b7f | multi-modal-multi-label-facial-action-unit | 2203.13301 | null | https://arxiv.org/abs/2203.13301v2 | https://arxiv.org/pdf/2203.13301v2.pdf | Multi-modal Multi-label Facial Action Unit Detection with Transformer | Facial Action Coding System is an important approach of facial expression analysis.This paper describes our submission to the third Affective Behavior Analysis (ABAW) 2022 competition. We proposed a transfomer based model to detect facial action unit (FAU) in video. To be specific, we firstly trained a multi-modal model to extract both audio and visual feature. After that, we proposed a action units correlation module to learn relationships between each action unit labels and refine action unit detection result. Experimental results on validation dataset shows that our method achieves better performance than baseline model, which verifies that the effectiveness of proposed network. | ['Jin Qi', 'Shisen Wang', 'Lingfeng Wang'] | 2022-03-24 | null | null | null | null | ['action-unit-detection', 'facial-action-unit-detection'] | ['computer-vision', 'computer-vision'] | [ 2.38685250e-01 1.04561299e-01 -1.73000887e-01 -6.40110552e-01
-5.55648446e-01 -1.92230001e-01 4.66591656e-01 -5.03748953e-01
-2.97250450e-01 5.73201239e-01 6.62808120e-01 6.39874637e-01
4.04553175e-01 -1.88461974e-01 -2.44881794e-01 -6.36978745e-01
-4.11637664e-01 -3.70821297e-01 -4.84636910e-02 -2.57456690e-01
1.65933788e-01 5.37538230e-01 -1.59516644e+00 1.03655112e+00
1.20123418e-03 1.34340203e+00 -6.46444678e-01 7.29146302e-01
1.70946911e-01 1.50614750e+00 -4.03116107e-01 -2.34692544e-01
9.02612135e-02 -7.45579660e-01 -9.70901012e-01 3.49381357e-01
2.68435061e-01 -5.13075948e-01 -1.76515415e-01 1.14471292e+00
5.20805001e-01 3.30987960e-01 8.66647482e-01 -1.68144238e+00
-4.66270626e-01 3.65256757e-01 -8.56262565e-01 3.33170086e-01
6.83679342e-01 -3.55170250e-01 1.01422203e+00 -1.01619065e+00
7.54089594e-01 1.46117067e+00 4.27116960e-01 9.72639263e-01
-5.78957796e-01 -9.66168642e-01 1.35343313e-01 5.68897069e-01
-1.30409849e+00 -7.09399521e-01 8.34937751e-01 -2.84902692e-01
1.17843270e+00 1.25688627e-01 7.74419725e-01 1.02375627e+00
1.94133744e-01 1.26956916e+00 1.06908786e+00 -3.17896336e-01
-9.59654525e-02 -1.40447512e-01 6.19158074e-02 8.84959280e-01
-8.57225955e-01 -2.78768271e-01 -7.94499338e-01 -9.68996361e-02
5.27769685e-01 -2.29032427e-01 -7.59433163e-03 2.87190199e-01
-5.81983984e-01 7.94046283e-01 1.78775012e-01 5.35244584e-01
-6.00200415e-01 4.72794026e-01 9.20116842e-01 4.07298058e-01
3.87487203e-01 -1.32950693e-01 -2.74963249e-02 -5.75175345e-01
-6.40551090e-01 -1.16166539e-01 2.54375726e-01 5.12477696e-01
5.50901473e-01 1.30783200e-01 -3.51550281e-01 1.11996949e+00
3.72687995e-01 6.15020841e-02 7.03797519e-01 -1.34919608e+00
-2.53315061e-01 6.63992643e-01 -2.80042320e-01 -1.21689689e+00
-3.57121021e-01 6.96389139e-01 -7.04754949e-01 2.86603481e-01
-3.32038373e-01 -2.44989917e-01 -6.70810938e-01 1.63797021e+00
1.45861402e-01 5.48812747e-01 -5.34498431e-02 8.91473055e-01
1.11440325e+00 7.61058748e-01 4.59617704e-01 -5.84504545e-01
1.35110974e+00 -9.90505874e-01 -1.22919559e+00 2.43540898e-01
8.25918555e-01 -7.48223424e-01 3.90364051e-01 5.54752111e-01
-9.73748207e-01 -6.00132465e-01 -9.17876720e-01 2.17824697e-01
-1.80946350e-01 4.02148426e-01 1.01970851e+00 6.48108125e-01
-1.14880645e+00 2.00761288e-01 -7.26138115e-01 -4.85054016e-01
7.35517919e-01 5.95530689e-01 -9.14723217e-01 4.69362289e-01
-1.23277330e+00 6.66649759e-01 3.66663009e-01 1.29613519e-01
-8.30652118e-01 -6.12517223e-02 -8.64175081e-01 -1.70418680e-01
-8.55187550e-02 8.56195763e-02 1.50052357e+00 -1.99407291e+00
-1.86342490e+00 9.57876325e-01 -3.82851452e-01 -3.47114265e-01
-2.03744188e-01 -1.78424284e-01 -8.25611711e-01 6.64538622e-01
-3.12682599e-01 1.02107120e+00 9.99184489e-01 -8.18459749e-01
-7.69761741e-01 -3.37913871e-01 7.31945038e-02 2.46800587e-01
-5.28841615e-01 6.58343673e-01 -2.79158264e-01 -3.97966146e-01
-3.16216260e-01 -6.90039039e-01 1.49042815e-01 -1.04042776e-01
1.89046010e-01 -7.78473437e-01 9.95659947e-01 -5.85918128e-01
1.53733420e+00 -2.47852874e+00 -8.53101090e-02 3.82720411e-01
1.35813147e-01 2.64804751e-01 -2.82266647e-01 2.87976533e-01
-5.95148861e-01 -9.93002579e-02 2.89254218e-01 -3.50277483e-01
-2.71609098e-01 5.74751012e-02 -3.78576480e-02 5.06430686e-01
3.31777126e-01 8.61050010e-01 -5.37397206e-01 -8.75948966e-01
2.15047106e-01 5.26470542e-01 -6.85114145e-01 1.47535011e-01
2.57910311e-01 1.46913916e-01 -5.27350962e-01 1.08622479e+00
5.46876609e-01 2.34839633e-01 1.18802719e-01 -4.59107876e-01
2.03492731e-01 -4.39589590e-01 -1.06451070e+00 1.59213173e+00
-2.84824520e-01 7.81245172e-01 4.70046960e-02 -1.07865596e+00
9.76088583e-01 7.39247024e-01 1.06607485e+00 -6.95518911e-01
7.62715459e-01 -1.77828580e-01 -1.44960349e-02 -8.37729096e-01
1.10714361e-01 -4.31125581e-01 6.59824982e-02 3.07932526e-01
7.28197277e-01 4.32750762e-01 6.52137697e-02 1.87372372e-01
1.14024401e+00 2.59059489e-01 6.52299106e-01 1.39478901e-02
1.03361869e+00 -5.21061301e-01 6.01004243e-01 -4.95237522e-02
-1.01042628e+00 3.51331681e-01 6.53993309e-01 -5.83241522e-01
-4.05371636e-01 -5.11196196e-01 1.34932294e-01 1.50787580e+00
-1.88477859e-01 -8.09129059e-01 -8.91166508e-01 -8.10372412e-01
-2.94234127e-01 1.14038676e-01 -1.17220974e+00 -2.97265470e-01
-3.76510710e-01 -5.08214653e-01 7.71477699e-01 6.49341047e-01
7.81397760e-01 -1.48809457e+00 -4.45455343e-01 -9.76304784e-02
-3.06972831e-01 -1.06288898e+00 -4.75342035e-01 -2.42272422e-01
-3.01997244e-01 -1.16039538e+00 -3.80711496e-01 -9.90063667e-01
5.30166626e-01 -6.55015633e-02 5.68787992e-01 -1.25311175e-02
-3.89475226e-01 6.41654253e-01 -6.64870203e-01 -5.50769329e-01
-1.42060876e-01 -5.79895616e-01 6.59595504e-02 8.12112987e-01
9.78314579e-01 -4.28778589e-01 -4.74302918e-01 1.37346774e-01
-6.64169073e-01 -2.88274437e-01 4.67269927e-01 4.73713338e-01
3.02449435e-01 -2.64286876e-01 5.86114883e-01 -4.47969943e-01
8.21729183e-01 -2.60120064e-01 -3.92762199e-02 1.99957132e-01
-3.70182246e-02 -4.23154324e-01 1.80858493e-01 -3.06574136e-01
-1.22272944e+00 6.33465827e-01 -3.13334852e-01 -6.24872625e-01
-1.58571824e-01 3.18458378e-01 -2.08549574e-03 -1.86606616e-01
3.31880391e-01 6.09639883e-02 2.09222585e-01 -6.04307745e-04
1.52248248e-01 1.06439269e+00 3.36604476e-01 -1.28662884e-01
-8.69201347e-02 5.94217062e-01 4.47648615e-02 -9.20070469e-01
-5.48581004e-01 -6.41167045e-01 -7.22678781e-01 -1.06089544e+00
1.28796458e+00 -1.08391392e+00 -1.35054052e+00 5.30560791e-01
-1.15413988e+00 1.06991716e-01 2.48402476e-01 5.45835197e-01
-7.19656229e-01 3.46094728e-01 -8.03072155e-01 -1.12428808e+00
-3.99617732e-01 -7.82892346e-01 1.20566714e+00 1.07102044e-01
-5.81680775e-01 -7.47080505e-01 3.75288725e-01 3.60358864e-01
8.62206742e-02 4.08772230e-01 2.75918007e-01 -4.27936345e-01
2.54688948e-01 -3.87062430e-01 -1.52684361e-01 6.55575514e-01
2.07183614e-01 4.65287089e-01 -1.22470808e+00 1.38333261e-01
-3.19189541e-02 -9.63227630e-01 9.22924221e-01 3.13647568e-01
1.39145708e+00 -7.54067674e-02 -1.79190904e-01 3.63080502e-01
1.02719676e+00 5.24452627e-01 9.17649746e-01 -9.94326547e-02
2.70792991e-01 5.04380405e-01 7.77278125e-01 8.06263387e-01
-7.68911615e-02 6.04696870e-01 3.64378452e-01 -5.18606557e-03
3.95568870e-02 -6.55124057e-03 8.36124718e-01 7.14721084e-01
-6.88139021e-01 4.18603644e-02 -4.23736572e-01 3.52042884e-01
-1.99999130e+00 -1.42733908e+00 7.75857791e-02 1.25702679e+00
6.85162425e-01 -2.29552969e-01 5.16682088e-01 5.83140589e-02
4.75578666e-01 2.42322534e-01 2.51486823e-02 -1.00393152e+00
-9.97964665e-02 4.38722432e-01 -1.38728991e-01 4.94669765e-01
-1.39015460e+00 1.19100738e+00 7.03078365e+00 1.01213455e+00
-1.20437503e+00 2.75803268e-01 5.65089881e-01 -2.28472427e-01
2.99855083e-01 -5.38424611e-01 -5.03214180e-01 2.30659366e-01
9.40654874e-01 -1.24738559e-01 7.20577016e-02 9.41856325e-01
3.12589020e-01 -1.85661525e-01 -7.70065427e-01 1.46060920e+00
5.81151426e-01 -9.99474525e-01 8.06613043e-02 -1.62827626e-01
4.54588234e-01 -3.36601794e-01 -2.36782536e-01 4.35727090e-01
7.44414106e-02 -1.28204465e+00 2.71521598e-01 5.59732974e-01
5.49172401e-01 -1.05239046e+00 8.28215957e-01 -1.80888459e-01
-1.44275653e+00 -1.43150732e-01 -3.64625752e-01 -2.54518270e-01
-1.35967927e-02 -2.95848936e-01 -5.76864123e-01 1.28128454e-01
7.08898127e-01 1.14731467e+00 -4.30580914e-01 4.08294618e-01
-2.34434471e-01 6.07596040e-01 2.26103850e-02 -2.35650659e-01
3.79703045e-01 -7.24647641e-02 -3.40526900e-03 1.47015822e+00
7.58014172e-02 6.20904088e-01 -2.08074033e-01 1.36921480e-01
-4.34046499e-02 5.28710067e-01 -7.52386570e-01 -2.90220231e-01
-1.33593649e-01 1.70782328e+00 -6.25184059e-01 -2.57655054e-01
-7.38563716e-01 1.29399467e+00 2.94595957e-01 1.87334597e-01
-1.09947634e+00 -2.64376074e-01 8.18816841e-01 -3.80412906e-01
1.14059187e-01 1.77812144e-01 3.20569515e-01 -1.01941490e+00
-3.44193578e-01 -8.20394397e-01 6.81445777e-01 -8.91727388e-01
-9.30990219e-01 6.95492685e-01 -1.30023807e-01 -1.35123980e+00
-1.93536520e-01 -8.10541272e-01 -6.25114620e-01 3.13968629e-01
-9.48855877e-01 -1.46388197e+00 -3.01387519e-01 1.04151034e+00
6.36361003e-01 -4.21443373e-01 1.14892995e+00 4.52354580e-01
-4.79684532e-01 6.64256513e-01 -5.29273987e-01 5.23752153e-01
8.38643372e-01 -6.47181451e-01 -5.90344191e-01 5.58491886e-01
2.77607173e-01 4.23667401e-01 4.14074212e-01 -3.49615067e-01
-1.14502335e+00 -9.00611579e-01 8.19871962e-01 -3.02865505e-01
6.44898474e-01 -1.36894017e-01 -3.64786863e-01 8.38470459e-01
6.56078696e-01 -1.99785009e-02 1.30108118e+00 -1.16266027e-01
-1.65533423e-01 -1.72047541e-01 -1.02912068e+00 3.15396011e-01
8.33964169e-01 -6.84975326e-01 -4.28250998e-01 8.25219229e-02
9.46960300e-02 1.05416618e-01 -8.72664273e-01 5.18351495e-01
9.89493608e-01 -1.09889281e+00 5.59598267e-01 -9.97062147e-01
6.95910037e-01 -1.16476603e-01 -3.79795909e-01 -9.93627429e-01
-4.34659004e-01 -4.23193634e-01 -3.00869495e-02 1.28721845e+00
1.05896071e-01 -7.46622905e-02 6.81949258e-01 4.62175727e-01
3.33440118e-02 -7.70955086e-01 -9.95654821e-01 -3.06575596e-01
-4.53725636e-01 -4.13240314e-01 1.16370447e-01 9.72727597e-01
7.07525432e-01 5.34496784e-01 -7.79851258e-01 -2.90520579e-01
-9.39736590e-02 -1.83343023e-01 4.85559314e-01 -7.14913130e-01
-4.01121238e-03 -5.43383479e-01 -1.11186826e+00 -5.05038500e-01
7.03639388e-01 -7.22456634e-01 -1.57435045e-01 -1.09008181e+00
4.06054467e-01 7.08946049e-01 -7.89671481e-01 8.30364704e-01
1.39481027e-03 8.58774602e-01 1.31434709e-01 -2.51119137e-01
-1.03738475e+00 7.71343946e-01 1.00746727e+00 -7.31899142e-02
-2.14508688e-03 -1.58210382e-01 -3.62769604e-01 9.24929380e-01
7.60773301e-01 -3.33446413e-02 -3.78910720e-01 1.04536705e-01
1.49495220e-02 -7.25204796e-02 1.22016065e-01 -9.36677873e-01
-6.42853081e-02 -2.62250453e-01 5.09320199e-01 -3.98856938e-01
6.50481582e-01 -7.99960256e-01 -2.11540282e-01 3.56971800e-01
-4.17727262e-01 -1.86332718e-01 3.07049096e-01 1.33001074e-01
-7.51510143e-01 7.86498096e-03 9.38527524e-01 3.88631858e-02
-1.33040059e+00 2.52775460e-01 -9.21178341e-01 -4.78884876e-01
1.55342650e+00 -2.55623430e-01 2.12603420e-01 -7.31481016e-01
-1.06822038e+00 7.58909881e-02 -2.29347512e-01 6.27038896e-01
9.42613065e-01 -1.81050825e+00 -4.94567156e-01 1.91124827e-01
2.82860160e-01 -1.12050307e+00 3.21649373e-01 1.08349609e+00
-2.72136092e-01 2.34703928e-01 -8.47960532e-01 -1.64884731e-01
-2.09725022e+00 2.88107544e-01 4.63921994e-01 2.58396953e-01
7.20444471e-02 1.28821969e+00 1.29485214e-02 6.24279790e-02
2.58160830e-01 5.75006157e-02 -8.70934069e-01 5.64253032e-01
9.21653986e-01 3.60324800e-01 -2.60989875e-01 -1.24646842e+00
-6.43573225e-01 5.67680299e-01 1.17607638e-01 -1.60528690e-01
1.17397249e+00 -5.69404885e-02 -4.01525289e-01 4.68820393e-01
1.53872931e+00 -3.66676927e-01 -8.19163263e-01 1.26459509e-01
-2.29544833e-01 -3.72953653e-01 2.85723694e-02 -3.68325531e-01
-1.23795557e+00 1.01628196e+00 8.45630288e-01 -1.46917969e-01
1.59261906e+00 -1.50382251e-01 6.12276912e-01 2.58373082e-01
1.27209485e-01 -1.60151029e+00 5.81935644e-01 4.53929752e-01
1.08174193e+00 -1.32422698e+00 -1.00042522e-01 -3.68235886e-01
-9.88077104e-01 1.43892610e+00 1.03560519e+00 -3.61938715e-01
1.12957859e+00 1.60773873e-01 2.93077290e-01 -3.71848822e-01
-9.15791154e-01 -3.35888237e-01 3.59930634e-01 3.85095924e-01
7.84557462e-01 -1.24051929e-01 -6.22259974e-01 7.04032183e-01
1.61497653e-01 4.37360674e-01 1.59732133e-01 9.53781724e-01
-5.59464872e-01 -9.92431760e-01 8.05443898e-02 2.84588695e-01
-6.72484994e-01 2.02069730e-01 -9.80862260e-01 6.36461854e-01
3.24755907e-01 8.61826479e-01 1.91132817e-02 -8.70078087e-01
8.78803432e-02 5.67464232e-01 6.69212282e-01 -4.15539801e-01
-5.22967577e-01 3.04726213e-01 1.73197180e-01 -1.17687464e+00
-1.24388480e+00 -6.86645150e-01 -1.45066440e+00 -3.90543081e-02
1.70431063e-01 1.18181377e-03 2.73124397e-01 8.63496184e-01
2.21303150e-01 3.84040236e-01 8.60128284e-01 -6.26751065e-01
7.85672441e-02 -1.33111572e+00 -9.09699619e-01 6.99723840e-01
1.25134885e-01 -7.25569487e-01 -1.67092547e-01 5.52852869e-01] | [13.579011917114258, 1.9531787633895874] |
ef382686-b047-4534-b862-cd99f93da81e | a-provably-improved-algorithm-for | 2302.07393 | null | https://arxiv.org/abs/2302.07393v1 | https://arxiv.org/pdf/2302.07393v1.pdf | A Provably Improved Algorithm for Crowdsourcing with Hard and Easy Tasks | Crowdsourcing is a popular method used to estimate ground-truth labels by collecting noisy labels from workers. In this work, we are motivated by crowdsourcing applications where each worker can exhibit two levels of accuracy depending on a task's type. Applying algorithms designed for the traditional Dawid-Skene model to such a scenario results in performance which is limited by the hard tasks. Therefore, we first extend the model to allow worker accuracy to vary depending on a task's unknown type. Then we propose a spectral method to partition tasks by type. After separating tasks by type, any Dawid-Skene algorithm (i.e., any algorithm designed for the Dawid-Skene model) can be applied independently to each type to infer the truth values. We theoretically prove that when crowdsourced data contain tasks with varying levels of difficulty, our algorithm infers the true labels with higher accuracy than any Dawid-Skene algorithm. Experiments show that our method is effective in practical applications. | ['R. Srikant', 'Dimitrios Katselis', 'Saptarshi Mandal', 'Seo Taek Kong'] | 2023-02-14 | null | null | null | null | ['type'] | ['speech'] | [-6.08626753e-02 1.20577000e-01 1.43195353e-02 -1.94234625e-01
-1.11766708e+00 -8.80358398e-01 2.78467208e-01 1.67730063e-01
-6.11854732e-01 1.00921142e+00 -2.76698440e-01 1.33254944e-04
9.67531204e-02 -5.52763760e-01 -7.96115041e-01 -6.82874322e-01
4.78840560e-01 9.49506581e-01 5.02222121e-01 -1.62408561e-01
1.01073511e-01 9.30076279e-03 -1.47882736e+00 2.03781426e-01
7.59463668e-01 6.79723084e-01 2.76112288e-01 9.42857385e-01
-9.20160115e-02 8.82759571e-01 -1.02870762e+00 -4.65392828e-01
6.03215754e-01 -3.06002825e-01 -8.98317099e-01 4.01518285e-01
2.20950395e-01 -4.44192141e-02 9.90217030e-02 1.10931361e+00
5.70284009e-01 2.52384752e-01 6.57985985e-01 -1.73903012e+00
-5.86311638e-01 6.20852947e-01 -7.58362889e-01 -1.77808747e-01
5.77972054e-01 -1.38310745e-01 8.01261663e-01 -7.91682065e-01
3.66628945e-01 1.11556244e+00 8.70062292e-01 7.21449852e-01
-1.36618996e+00 -6.38614714e-01 2.71155387e-01 -1.88578695e-01
-1.50520384e+00 -2.17468485e-01 4.15814579e-01 -8.25229943e-01
2.01208055e-01 2.92347014e-01 3.28381397e-02 9.32345212e-01
-3.84294927e-01 8.31146002e-01 1.54559863e+00 -5.14516473e-01
7.24617839e-01 3.30134749e-01 1.89314455e-01 4.93069828e-01
4.49633718e-01 -5.40066719e-01 -7.73174345e-01 -6.15071416e-01
3.20399940e-01 1.01433089e-02 -2.59624302e-01 -1.37227118e-01
-1.18089485e+00 6.77972376e-01 6.50786906e-02 5.88074997e-02
-2.28314564e-01 2.17560679e-01 3.61663222e-01 1.77257583e-01
8.22129190e-01 4.71070021e-01 -3.31692725e-01 3.59381363e-02
-9.57349002e-01 7.41268456e-01 9.90580380e-01 1.22181296e+00
9.58135486e-01 -4.88867283e-01 -6.24017835e-01 8.33088815e-01
-1.08043998e-01 6.20691538e-01 3.68360907e-01 -1.20377481e+00
3.69990379e-01 4.92884457e-01 1.17559719e+00 -8.16001117e-01
-4.73675758e-01 9.24853161e-02 -4.63667035e-01 2.07640484e-01
9.85416651e-01 -3.67402703e-01 -6.34500861e-01 1.70283353e+00
6.85237348e-01 4.92588580e-02 -2.56686091e-01 1.23662722e+00
3.45623642e-01 1.95562124e-01 -5.05867600e-02 -6.14657924e-02
1.45047402e+00 -9.32352424e-01 -6.76034391e-01 -2.11660907e-01
8.48679245e-01 -6.30386949e-01 1.28730679e+00 6.01118326e-01
-8.22304428e-01 -5.56015551e-01 -6.48189366e-01 -1.77502275e-01
-2.79536873e-01 3.23972285e-01 1.39188841e-01 9.85174417e-01
-9.75285530e-01 3.65096062e-01 -5.78237414e-01 -2.27561802e-01
9.25730541e-02 2.24124819e-01 -1.34166032e-01 -1.15306035e-01
-1.05646574e+00 7.16154337e-01 1.95038587e-01 -9.24630463e-02
-1.10508120e+00 -2.03246847e-01 -5.63734949e-01 -2.87127912e-01
9.30121124e-01 -4.62460130e-01 1.82764125e+00 -8.65825951e-01
-1.06098747e+00 1.07360959e+00 -7.04706430e-01 -3.13760787e-01
9.63175178e-01 1.68292865e-01 1.37293205e-01 4.49243858e-02
7.96957076e-01 1.89468041e-01 7.27931142e-01 -1.83650959e+00
-9.74876463e-01 -5.09565473e-01 5.72301805e-01 1.63652226e-01
-2.17499644e-01 -9.65180174e-02 -1.74328089e-01 -2.81250119e-01
-3.57798189e-02 -1.37850535e+00 -1.31954610e-01 -1.07942000e-01
-6.21411324e-01 -5.68393111e-01 2.79587030e-01 -2.07414255e-01
7.79291928e-01 -1.93745911e+00 -3.06042135e-01 5.45230322e-02
7.48927534e-01 8.54999349e-02 1.24503955e-01 3.60974401e-01
5.57152867e-01 3.38290423e-01 -3.33644271e-01 -8.25859666e-01
3.35550576e-01 4.08703029e-01 -5.36599346e-02 7.14320004e-01
-1.24244630e-01 7.21528888e-01 -1.31705189e+00 -5.07182479e-01
-3.45010161e-01 -1.66198418e-01 -9.73630231e-03 3.16176057e-01
-2.58914918e-01 3.68511111e-01 -4.54950809e-01 3.58885229e-01
8.71282995e-01 -4.27830905e-01 1.39826730e-01 5.70546567e-01
6.86554238e-02 -1.04078136e-01 -1.29912114e+00 1.19370496e+00
-5.99971712e-01 3.77818376e-01 5.10586917e-01 -6.25017583e-01
7.39829481e-01 4.34877664e-01 1.76781267e-01 -6.79909214e-02
-1.14014782e-01 4.18324202e-01 -3.99528086e-01 -3.93928140e-01
7.30682671e-01 -4.16664571e-01 -3.77840459e-01 7.62263417e-01
-2.87381828e-01 -5.58470972e-02 2.27030993e-01 2.41549179e-01
1.13321090e+00 -1.19888134e-01 1.96564086e-02 -2.96367973e-01
1.93384707e-01 4.18341547e-01 5.79124510e-01 1.22263515e+00
-4.31202710e-01 5.98464131e-01 5.94719887e-01 -4.59423065e-01
-9.59610581e-01 -8.21889997e-01 1.13929041e-01 1.73810291e+00
4.94525373e-01 1.92171559e-02 -1.16370857e+00 -8.43680263e-01
1.01667009e-01 3.20415378e-01 -8.67600083e-01 3.79691929e-01
1.61028523e-02 -4.29284036e-01 7.48577058e-01 3.46078545e-01
4.85251784e-01 -5.87795436e-01 -4.61632997e-01 5.12900352e-02
-6.57856047e-01 -1.43016589e+00 -7.40872383e-01 -5.55531383e-02
-1.21382773e-01 -1.26268756e+00 -1.13731778e+00 -6.04161561e-01
8.38485718e-01 8.68556142e-01 1.14244843e+00 1.60983235e-01
3.35787743e-01 4.63935852e-01 -5.83942175e-01 -7.31732845e-01
-4.15991604e-01 2.66528368e-01 2.89931804e-01 9.85062048e-02
6.61658347e-01 -3.74261364e-02 -4.16020364e-01 8.07284474e-01
-9.55051184e-01 -2.21732646e-01 -1.23145327e-01 5.18720984e-01
4.12784159e-01 2.28019521e-01 7.23742366e-01 -1.35389709e+00
1.02782857e+00 -6.53810740e-01 -5.80109775e-01 3.38271707e-01
-4.19068545e-01 -1.36435091e-01 8.03430080e-01 -6.88109756e-01
-8.30593467e-01 3.63159418e-01 5.48042953e-01 -3.60000342e-01
-3.09304923e-01 -8.04443732e-02 8.40923116e-02 -3.48382108e-02
1.21642995e+00 2.38036551e-02 -2.26475880e-01 -2.35222444e-01
9.08262581e-02 1.05456352e+00 4.55361694e-01 -1.11469197e+00
7.78249860e-01 7.21346676e-01 -1.26247853e-01 -4.65820909e-01
-1.44257307e+00 -8.53535354e-01 -5.26538312e-01 -2.61458784e-01
9.33790267e-01 -1.12578106e+00 -1.18182158e+00 4.88839030e-01
-1.30843186e+00 -4.82559770e-01 -7.19610751e-02 1.59391016e-01
-4.61006910e-01 4.23503429e-01 -4.17346507e-01 -1.57020986e+00
2.91842192e-01 -1.22212780e+00 1.54930019e+00 3.63993384e-02
-2.06067890e-01 -9.34778988e-01 -5.71096353e-02 6.42944634e-01
-7.99985195e-04 1.13745235e-01 2.75678169e-02 -1.03700769e+00
-1.83969304e-01 -4.44960117e-01 -1.53992206e-01 1.98340729e-01
-8.70114863e-02 -6.38526678e-01 -1.28865170e+00 -1.12157598e-01
1.25977501e-01 -7.11178660e-01 5.54925263e-01 5.02267182e-02
1.22195947e+00 -4.77805793e-01 -1.79989800e-01 -1.44292906e-01
1.13674510e+00 -2.39799231e-01 6.82401136e-02 3.31751406e-01
7.10661471e-01 6.96085155e-01 6.32504582e-01 5.50535917e-01
7.48966157e-01 5.88814735e-01 2.27258533e-01 -2.52554473e-03
4.97691393e-01 -3.59913290e-01 2.29227200e-01 8.81712288e-02
-7.06374496e-02 -4.07762617e-01 -1.02734089e+00 8.36169124e-01
-2.23031640e+00 -6.59835994e-01 -5.36996126e-01 2.32587814e+00
9.93317842e-01 -1.20232068e-01 7.19216883e-01 2.67560720e-01
1.31657040e+00 -2.28058353e-01 -3.64356250e-01 -2.32891157e-01
1.95922032e-01 -1.50642887e-01 8.45061362e-01 4.81136233e-01
-8.49618733e-01 7.47350752e-01 5.85050869e+00 9.81285393e-01
-3.67450833e-01 5.44698656e-01 5.31869113e-01 6.38299808e-02
-8.88830423e-02 -1.22584686e-01 -9.24208820e-01 8.28671634e-01
5.77334166e-01 -2.80789614e-01 6.32016718e-01 8.46412182e-01
3.37473392e-01 -6.08547747e-01 -1.24864447e+00 8.16159844e-01
-1.66191645e-02 -6.18932486e-01 -6.03901505e-01 1.95782363e-01
1.13636816e+00 -5.16675472e-01 -4.37151074e-01 3.01296085e-01
1.02289927e+00 -8.26609790e-01 8.80980790e-01 3.05747479e-01
6.68822110e-01 -6.05335057e-01 8.54298592e-01 1.09533620e+00
-8.74034047e-01 -7.09713697e-02 -5.50361991e-01 -4.94339138e-01
7.09038973e-02 1.03841972e+00 -1.06171322e+00 2.59295046e-01
4.43374902e-01 -2.93584287e-01 -2.07792625e-01 8.45201433e-01
-3.76660675e-01 6.35848939e-01 -1.89762190e-01 -2.49053359e-01
8.14389437e-02 3.26797292e-02 2.30706826e-01 1.15297306e+00
3.12844008e-01 1.05771326e-01 8.41744542e-01 8.24555337e-01
-3.53523821e-01 -1.52846634e-01 -7.80142248e-01 3.55447739e-01
8.16206694e-01 1.26912034e+00 -7.56877244e-01 -6.64887309e-01
-1.37349948e-01 1.22279727e+00 4.04832393e-01 4.22474980e-01
-6.93161070e-01 -2.13313088e-01 3.35390925e-01 4.86256868e-01
-2.02288836e-01 -2.59015709e-01 -4.77995038e-01 -9.02628303e-01
4.33866978e-01 -7.57838905e-01 1.64446071e-01 -9.56925929e-01
-1.73841345e+00 3.14701170e-01 -5.06899916e-02 -9.89273369e-01
-1.35403529e-01 -6.63605988e-01 -2.58927763e-01 1.15844214e+00
-1.29027963e+00 -9.58967745e-01 -6.00846231e-01 6.15355074e-01
4.57043320e-01 2.42411107e-01 5.11397183e-01 -2.44213372e-01
-1.94810957e-01 4.17512625e-01 -6.74848678e-03 2.92176843e-01
9.08827543e-01 -1.62805045e+00 2.68518925e-01 6.18467450e-01
-1.07669093e-01 4.01016921e-01 7.81654358e-01 -6.92929208e-01
-1.01478732e+00 -1.26910281e+00 1.00245035e+00 -8.61971855e-01
6.75331175e-01 -8.01725507e-01 -7.32264042e-01 5.87723970e-01
-2.52178252e-01 1.65791973e-01 6.16380274e-01 2.16994166e-01
-4.44700956e-01 1.10833146e-01 -1.39849579e+00 2.86601216e-01
1.02979362e+00 -8.71371329e-01 -3.83112729e-01 9.35920119e-01
6.31346405e-01 -6.05927169e-01 -5.08426607e-01 -2.66013622e-01
1.73514724e-01 -9.34015989e-01 2.92822421e-01 -4.71518725e-01
3.04556310e-01 -6.09948158e-01 -1.22067928e-01 -1.41895044e+00
-9.54967961e-02 -5.18104970e-01 3.26065987e-01 1.14137936e+00
3.86458516e-01 -7.34541357e-01 6.10929430e-01 1.15563905e+00
2.05031708e-01 -2.01443613e-01 -8.35993290e-01 -1.15629280e+00
-1.54207125e-02 -3.96076649e-01 5.71451128e-01 9.27275717e-01
3.64930257e-02 1.35348901e-01 -4.80250299e-01 5.35091519e-01
5.69282174e-01 -7.89716542e-02 1.18891275e+00 -1.40975654e+00
-4.18297529e-01 1.59519911e-01 -9.74365324e-02 -1.06129491e+00
4.78646100e-01 -7.12212265e-01 7.66082168e-01 -1.42092848e+00
3.95541161e-01 -6.48730755e-01 2.48808697e-01 4.81625348e-01
-5.46152174e-01 5.13622761e-01 2.43184328e-01 5.17635643e-01
-1.07794559e+00 -1.08679868e-01 1.10473967e+00 -3.36227939e-03
-9.51460227e-02 3.37745428e-01 -9.34769869e-01 6.46517873e-01
8.04487705e-01 -7.56250560e-01 -3.16581488e-01 -5.23434460e-01
6.64633214e-01 7.61142327e-03 5.17582417e-01 -8.78685236e-01
5.25151253e-01 -1.91837668e-01 7.41317198e-02 9.25102234e-02
2.03128323e-01 -5.97222507e-01 -1.27237067e-01 2.27534529e-02
-7.13675797e-01 -1.22410074e-01 -3.11079144e-01 9.59703565e-01
1.55549213e-01 -6.33442819e-01 6.07501805e-01 -4.81391609e-01
-7.50694796e-02 1.36240020e-01 -4.72104698e-01 3.60609382e-01
1.10501444e+00 -2.14551583e-01 -4.20586169e-01 -4.89283919e-01
-6.84275508e-01 5.41956902e-01 8.44509661e-01 -1.42972901e-01
-6.13464378e-02 -1.27904224e+00 -7.88032234e-01 -4.00505215e-01
3.22049111e-01 4.93582338e-01 -7.75460005e-02 7.34050691e-01
-4.02847454e-02 -1.78351492e-01 4.48111475e-01 -5.65925777e-01
-1.03911996e+00 7.92693198e-01 3.14897269e-01 -3.09884965e-01
7.69596025e-02 7.34765947e-01 3.03035915e-01 -5.88488996e-01
7.66344815e-02 1.51869711e-02 2.11712629e-01 8.01414624e-02
6.07834995e-01 8.67765784e-01 2.12571636e-01 -6.01789296e-01
-2.91924536e-01 3.12375426e-01 3.79556179e-01 -4.74400580e-01
6.66324377e-01 -3.18112314e-01 -6.69023618e-02 8.73093843e-01
7.35238910e-01 3.35032642e-01 -1.11438239e+00 -4.10384595e-01
9.75575000e-02 -6.34572208e-01 -4.24499661e-01 -6.61422729e-01
-4.94591504e-01 7.12030590e-01 1.32032782e-01 8.73828828e-01
9.26595032e-01 8.39141235e-02 5.71077466e-01 4.82455194e-01
9.50923741e-01 -1.50957584e+00 -1.13295458e-01 3.22673112e-01
4.09218967e-01 -1.50976110e+00 -2.82932192e-01 -4.80316907e-01
-9.46251512e-01 6.69048071e-01 5.33401608e-01 -1.04862727e-01
3.00293446e-01 3.57949585e-01 2.08331645e-01 -1.22015320e-01
-4.73133206e-01 -5.60727596e-01 -1.40894623e-03 9.10538197e-01
2.80555040e-01 5.01083612e-01 -3.44881982e-01 8.05978060e-01
-1.48043573e-01 2.21557319e-01 9.56151783e-01 8.06856811e-01
-7.45541751e-01 -8.46491814e-01 -1.07285440e+00 3.16251427e-01
-5.72710395e-01 3.12951386e-01 -7.64518976e-01 4.50822085e-01
4.62538004e-01 1.62583959e+00 -1.45338267e-01 -2.42560551e-01
3.77601475e-01 1.35352567e-01 2.59290755e-01 -8.83532524e-01
-5.40685833e-01 -2.28484079e-01 7.80021325e-02 -1.89048693e-01
-5.69831789e-01 -2.92856634e-01 -1.23334694e+00 -4.33774769e-01
-4.54689920e-01 5.04038095e-01 6.06159449e-01 1.15482569e+00
6.88666254e-02 -7.61075243e-02 9.09062862e-01 -8.94667208e-01
-9.13279831e-01 -9.25053179e-01 -1.07401335e+00 5.75249016e-01
2.80400306e-01 -7.56712794e-01 -5.74647129e-01 1.92440525e-01] | [9.619168281555176, 4.635506629943848] |
6dd58e5c-877d-4aae-9064-1f1815bbade1 | 3d-feature-pyramid-attention-module-for | 1810.06178 | null | http://arxiv.org/abs/1810.06178v4 | http://arxiv.org/pdf/1810.06178v4.pdf | 3D Feature Pyramid Attention Module for Robust Visual Speech Recognition | Visual speech recognition is the task to decode the speech content from a
video based on visual information, especially the movements of lips. It is also
referenced as lipreading. Motivated by two problems existing in lipreading,
words with similar pronunciation and the variation of word duration, we propose
a novel 3D Feature Pyramid Attention (3D-FPA) module to jointly improve the
representation power of features in both the spatial and temporal domains.
Specifically, the input features are downsampled for 3 times in both the
spatial and temporal dimensions to construct spatiotemporal feature pyramids.
Then high-level features are upsampled and combined with low-level features,
finally generating a pixel-level soft attention mask to be multiplied with the
input features.It enhances the discriminative power of features and exploits
the temporal multi-scale information while decoding the visual speeches. Also,
this module provides a new method to construct and utilize temporal pyramid
structures in video analysis tasks. The field of temporal featrue pyramids are
still under exploring compared to the plentiful works on spatial feature
pyramids for image analysis tasks. To validate the effectiveness and
adaptability of our proposed module, we embed the module in a sentence-level
lipreading model, LipNet, with the result of 3.6% absolute decrease in word
error rate, and a word-level model, with the result of 1.4% absolute
improvement in accuracy. | ['Jing-Yun Xiao'] | 2018-10-15 | null | null | null | null | ['lipreading'] | ['computer-vision'] | [ 6.00243732e-02 -4.80865359e-01 -3.10915172e-01 -2.35674288e-02
-6.36808574e-01 -4.16311435e-02 4.60572511e-01 -2.36844093e-01
-5.22311449e-01 4.98663306e-01 6.77678406e-01 -1.28822736e-02
2.40977034e-01 -2.97913432e-01 -5.81020057e-01 -7.20739841e-01
2.23202109e-01 -7.37609982e-01 5.92209101e-01 1.39648631e-01
5.17356753e-01 3.26647341e-01 -1.97747993e+00 4.59489048e-01
9.26893413e-01 1.22353935e+00 6.43654644e-01 6.68360651e-01
-3.20452124e-01 3.44797045e-01 -3.66689950e-01 -4.00958098e-02
-5.99608272e-02 -3.51841092e-01 -2.96539158e-01 3.18663120e-01
1.56903014e-01 -3.03717673e-01 -6.94935262e-01 1.11482251e+00
7.35252738e-01 1.17255874e-01 6.28668606e-01 -1.15302622e+00
-9.03522611e-01 -5.71639873e-02 -7.80800462e-01 4.30537820e-01
3.71000320e-01 4.93343323e-01 6.17519915e-01 -1.16925287e+00
1.70748815e-01 1.46532261e+00 3.72232258e-01 4.79216635e-01
-7.52806008e-01 -4.37790543e-01 1.30008563e-01 9.26133811e-01
-1.38436103e+00 -7.98843145e-01 8.49807858e-01 -3.13138694e-01
1.05998158e+00 8.07725266e-02 7.05453455e-01 8.16268086e-01
3.97377402e-01 9.73719239e-01 1.12607300e+00 -3.93667877e-01
-1.75579011e-01 1.37350596e-02 -8.29917043e-02 5.63189328e-01
-2.48295903e-01 6.64171055e-02 -9.72283006e-01 4.44219559e-01
6.28064692e-01 3.19719911e-02 -5.23723483e-01 -8.33699182e-02
-1.25433505e+00 5.05189776e-01 2.44722486e-01 3.76900554e-01
-5.70508242e-01 -4.76945266e-02 4.41738904e-01 -1.51120722e-01
4.55482334e-01 -4.09931213e-01 -3.31379324e-01 -2.25029260e-01
-1.01967502e+00 -1.97409466e-01 -7.99989700e-03 6.52186871e-01
6.22786403e-01 1.87134996e-01 -4.75944042e-01 1.15621781e+00
5.91209590e-01 5.25329292e-01 1.09341717e+00 -5.88911235e-01
4.47640419e-01 4.35337096e-01 -8.94545913e-02 -9.22612667e-01
-2.10885555e-01 -1.97144561e-02 -6.48895502e-01 2.76554734e-01
6.10388517e-02 1.25121191e-01 -1.16211069e+00 1.81787109e+00
2.14272484e-01 2.25712910e-01 8.01563412e-02 1.00126016e+00
8.94603729e-01 8.81520629e-01 2.01537773e-01 -5.17130017e-01
1.71680880e+00 -1.03645766e+00 -1.14240503e+00 -5.07571362e-02
-2.59197447e-02 -9.13559258e-01 1.29375148e+00 1.09175779e-01
-1.33767271e+00 -9.62143242e-01 -1.03224134e+00 -2.97669679e-01
-4.91143912e-01 3.18058908e-01 2.70106882e-01 5.08689940e-01
-1.25636744e+00 1.64952159e-01 -4.40051258e-01 -2.96064377e-01
3.73119920e-01 2.82990187e-01 -3.65408301e-01 2.82307006e-02
-1.27338290e+00 7.38596678e-01 2.80489415e-01 2.72901449e-02
-6.36862338e-01 -5.32530546e-01 -1.06050909e+00 1.76319376e-01
-2.78103240e-02 -4.10953313e-01 9.31225777e-01 -7.39981711e-01
-1.65614820e+00 5.24855256e-01 -9.05366182e-01 -1.07116766e-01
2.04112232e-01 1.21411890e-01 -6.23776078e-01 3.40853274e-01
8.56571347e-02 9.46565449e-01 1.33036053e+00 -7.01854467e-01
-6.46514356e-01 -4.11852241e-01 -3.50919396e-01 6.38671219e-01
-4.80089843e-01 1.27024010e-01 -7.91536868e-01 -7.94875205e-01
1.71219945e-01 -3.34174663e-01 3.23811889e-01 1.55698642e-01
4.78715114e-02 -3.48285705e-01 1.11802065e+00 -1.12586641e+00
1.26071894e+00 -2.60131121e+00 2.34388635e-02 -3.26402336e-01
9.49177667e-02 3.98421377e-01 -2.27277339e-01 -5.59038669e-02
4.23722155e-02 1.30044535e-01 -1.42299503e-01 -2.88477004e-01
-1.50453553e-01 -1.52328193e-01 -9.23826266e-03 4.44090992e-01
3.37409914e-01 1.01906669e+00 -6.84348404e-01 -8.01088572e-01
4.75526065e-01 6.51834190e-01 -4.65160549e-01 6.31423146e-02
1.23174198e-01 1.18825309e-01 -2.36119777e-01 7.06034303e-01
8.78428459e-01 9.05077904e-02 -3.83318186e-01 -4.78869259e-01
-3.30474734e-01 2.74702638e-01 -8.71641457e-01 1.64797163e+00
-4.68159974e-01 9.24067140e-01 1.62986681e-01 -6.84805691e-01
8.37530017e-01 5.06590366e-01 3.99560750e-01 -9.92541730e-01
2.01236755e-02 5.89065067e-02 -1.11544304e-01 -9.79964018e-01
4.50593323e-01 -1.46119922e-01 3.29852611e-01 -3.89833599e-02
1.09790921e-01 -2.79534385e-02 -6.05918504e-02 -1.87739089e-01
5.62473536e-01 5.17502427e-02 1.26289174e-01 -8.31522942e-02
1.03983414e+00 -5.98962724e-01 3.97250354e-01 4.35356200e-02
-6.52824223e-01 5.83819091e-01 2.95059502e-01 1.07963365e-02
-1.01731312e+00 -1.07318079e+00 -1.73455879e-01 1.05346739e+00
3.29502195e-01 -3.32061112e-01 -7.06197917e-01 -4.96429771e-01
-3.46869528e-02 3.16530198e-01 -5.08508444e-01 -2.10696220e-01
-2.26979926e-01 -2.34625190e-01 3.67862314e-01 4.47993994e-01
9.46120024e-01 -1.29229379e+00 -4.61438894e-01 2.03032568e-02
-4.10977364e-01 -1.20293474e+00 -1.10092223e+00 -2.50296533e-01
-3.97780836e-01 -7.70839155e-01 -1.15056109e+00 -1.04579604e+00
4.66869444e-01 4.45863545e-01 2.97287375e-01 -2.44455531e-01
-2.85549790e-01 2.70353824e-01 -3.24883997e-01 -3.22820157e-01
-1.35759845e-01 -3.52158815e-01 2.13303059e-01 4.33455676e-01
3.81365776e-01 -4.98187512e-01 -7.40842044e-01 2.38911763e-01
-8.03968430e-01 1.61343738e-01 7.32045293e-01 9.31359768e-01
3.81982297e-01 -5.72410114e-02 7.62760818e-01 4.95011866e-01
9.12417352e-01 -2.43639588e-01 -3.40402246e-01 1.16343133e-01
-2.40350470e-01 1.16862496e-02 4.49023157e-01 -6.72025800e-01
-9.96841311e-01 -1.80464089e-01 -3.19855899e-01 -7.03667998e-01
-8.10223594e-02 2.02981383e-01 -3.66973549e-01 -4.61778156e-02
1.04528882e-01 8.63533497e-01 5.24098337e-01 -4.72775638e-01
2.36463428e-01 1.18116903e+00 5.55600047e-01 2.15100907e-02
2.77461499e-01 2.10733831e-01 -2.09111467e-01 -1.17181313e+00
-6.77990541e-02 -5.35154521e-01 -3.58335763e-01 -3.76920730e-01
1.22029698e+00 -8.17114830e-01 -1.00979745e+00 8.69339466e-01
-1.25191593e+00 2.45329253e-02 -6.30833134e-02 8.14236999e-01
-6.57242596e-01 8.29152226e-01 -5.68150759e-01 -7.94532239e-01
-2.11220294e-01 -1.58333623e+00 1.04643846e+00 4.93632019e-01
2.80972302e-01 -5.98434567e-01 -3.75867158e-01 2.75039405e-01
4.36726749e-01 -4.50610995e-01 9.61005807e-01 -6.93187043e-02
-4.20977145e-01 2.30372712e-01 -5.40976703e-01 6.01534605e-01
3.71561468e-01 -1.76629707e-01 -1.09507060e+00 -3.90586667e-02
8.09368193e-02 -1.68194145e-01 1.02429879e+00 7.13015258e-01
1.23157823e+00 -3.80958885e-01 -2.58453041e-01 4.98452455e-01
1.05802870e+00 4.19744313e-01 8.13047707e-01 -8.89328495e-02
3.87055725e-01 4.97079760e-01 5.27509809e-01 3.19801509e-01
2.80515820e-01 9.48739111e-01 2.82773465e-01 -1.28922388e-01
-7.57920563e-01 -5.02452075e-01 5.74518502e-01 9.36317086e-01
8.61046538e-02 -2.07231175e-02 -5.23796499e-01 8.61518502e-01
-1.58192539e+00 -9.90683794e-01 3.97807866e-01 2.00298595e+00
8.57375860e-01 -7.01706931e-02 1.49994195e-01 4.11106229e-01
1.08049977e+00 4.38986182e-01 -6.03848100e-01 -5.18510640e-01
-1.94167167e-01 2.83617880e-02 2.99586147e-01 7.11213171e-01
-8.47838163e-01 1.06926334e+00 5.98883295e+00 1.24207950e+00
-1.49023402e+00 1.68763265e-01 4.37663317e-01 3.84780318e-02
-2.34761372e-01 -4.00320411e-01 -6.71890676e-01 8.52375567e-01
6.91787839e-01 -1.97581083e-01 4.12032545e-01 4.49660867e-01
6.38642609e-01 -1.99004874e-01 -5.51555514e-01 1.41206312e+00
2.88674384e-01 -1.12911046e+00 1.84016854e-01 4.56225639e-03
2.22845644e-01 -1.33001715e-01 4.22021240e-01 2.16629937e-01
-5.02606273e-01 -1.07029414e+00 7.33112693e-01 5.39379478e-01
1.26791549e+00 -6.24514163e-01 3.89736295e-01 2.72523105e-01
-1.59241426e+00 -1.44150361e-01 -2.81908095e-01 1.16710879e-01
2.52523452e-01 1.40973493e-01 -7.38983452e-01 3.08932781e-01
7.67615378e-01 8.06961894e-01 -2.95181602e-01 1.17558408e+00
1.72910150e-02 3.28834414e-01 -2.00475946e-01 -2.07202107e-01
2.42574334e-01 -5.48952085e-04 5.96635520e-01 1.25930834e+00
3.76136094e-01 4.74029221e-02 -3.20696592e-01 6.49453521e-01
1.26488850e-01 2.79788136e-01 -5.40008247e-01 7.23092631e-02
2.92631507e-01 9.34306443e-01 -1.41973853e-01 -2.47799188e-01
-6.26058638e-01 1.06657505e+00 -6.40048683e-02 6.21734917e-01
-8.82667720e-01 -6.46319807e-01 8.24531794e-01 1.95368901e-01
5.46581805e-01 -2.92551130e-01 -1.55858561e-01 -9.48865473e-01
3.92976195e-01 -8.06196153e-01 -3.16414498e-02 -1.13077426e+00
-8.27282667e-01 4.69337612e-01 -2.73438931e-01 -1.25097036e+00
-7.90854767e-02 -6.32433891e-01 -5.12892127e-01 1.24627388e+00
-1.90275204e+00 -1.03740108e+00 -3.87942463e-01 1.14192712e+00
1.04731596e+00 -1.82174578e-01 4.40543473e-01 2.87955493e-01
-3.19752961e-01 7.58334935e-01 -6.83002919e-02 -6.83853850e-02
6.49920046e-01 -6.10236406e-01 1.43734738e-01 8.50072503e-01
-1.72884315e-01 2.77722687e-01 2.31934980e-01 -5.18935382e-01
-1.25797164e+00 -8.64623964e-01 1.03822982e+00 1.63290024e-01
4.29924339e-01 -3.18963259e-01 -8.44657004e-01 5.60620725e-02
2.58181214e-01 1.09646112e-01 2.52503574e-01 -5.36103308e-01
-2.30315149e-01 -2.12379694e-01 -1.29319763e+00 7.27962554e-01
9.23115253e-01 -8.43066394e-01 -6.89562619e-01 -7.42560476e-02
9.49936211e-01 -1.56995952e-01 -5.71247458e-01 5.08124948e-01
6.73402250e-01 -8.98217916e-01 8.38592649e-01 -2.19791442e-01
1.21503256e-01 -4.25440222e-01 -3.13673645e-01 -1.00712335e+00
-3.09034109e-01 -5.58780372e-01 1.74785517e-02 1.13377500e+00
1.80424690e-01 -6.66825533e-01 4.51739877e-01 -2.95573589e-03
-2.20761791e-01 -8.03222239e-01 -1.22244585e+00 -6.23854220e-01
-2.02936709e-01 -4.41157401e-01 4.39296931e-01 3.35711926e-01
3.53439450e-01 1.30448610e-01 -3.63097727e-01 1.03072681e-01
3.23384166e-01 -2.68576026e-01 2.44538054e-01 -6.70902371e-01
8.38212073e-02 -5.83077312e-01 -7.10214555e-01 -1.51857650e+00
2.23544508e-01 -7.32478440e-01 1.14541084e-01 -1.59069788e+00
1.51848137e-01 9.68303531e-02 -3.73565793e-01 3.35395128e-01
-3.60648900e-01 2.27673188e-01 3.26279700e-01 1.57780394e-01
-2.09209263e-01 9.00137544e-01 1.49819791e+00 -3.38129371e-01
-3.02615345e-01 -4.77608889e-02 -2.94299126e-01 4.64898318e-01
6.78370237e-01 1.71185151e-01 -4.79539245e-01 -2.95302719e-01
-6.77526176e-01 2.27042139e-01 3.59405667e-01 -1.00150919e+00
4.24258769e-01 -1.53290868e-01 5.27223289e-01 -8.60059619e-01
8.62097979e-01 -5.68554759e-01 -3.99880797e-01 4.21934009e-01
-1.48466125e-01 6.70724586e-02 5.13276577e-01 4.76059079e-01
-5.63492179e-01 5.11118248e-02 1.00464666e+00 6.93476647e-02
-1.05739629e+00 4.13753390e-01 -4.42361981e-01 -1.80607930e-01
1.04547977e+00 -6.88918948e-01 -6.81543797e-02 -4.87638563e-01
-7.29911566e-01 -1.33416325e-01 2.01574266e-01 6.22895122e-01
1.06912601e+00 -1.34357953e+00 -6.79395318e-01 7.64842093e-01
-5.78631684e-02 -5.17307520e-01 6.73374116e-01 8.50701213e-01
-1.25851259e-01 6.19278491e-01 -5.73650062e-01 -7.87183166e-01
-1.36695337e+00 8.93971145e-01 3.45723450e-01 2.65711844e-01
-6.19274139e-01 6.57516420e-01 4.10279542e-01 3.46837699e-01
4.13523048e-01 -3.19151133e-01 -4.83550608e-01 6.91325068e-02
6.43543780e-01 2.44620666e-01 -9.34860259e-02 -8.26669991e-01
-4.41410542e-01 9.98519659e-01 7.23981634e-02 -3.53655189e-01
9.88519967e-01 -5.03645360e-01 8.00030455e-02 2.57627636e-01
1.39004970e+00 2.65729904e-01 -1.48768663e+00 -1.64197609e-01
-4.29376692e-01 -5.64869225e-01 6.70332909e-02 -6.70253396e-01
-9.58519757e-01 1.27327442e+00 8.06815147e-01 -1.01267165e-02
1.42371631e+00 8.95319134e-02 6.55427396e-01 -4.32709277e-01
2.45334823e-02 -9.84631121e-01 3.03390235e-01 3.84758234e-01
1.12459922e+00 -1.09437561e+00 -4.31553364e-01 -3.14225882e-01
-6.47158027e-01 9.56098735e-01 4.70065147e-01 3.63296181e-01
5.75066328e-01 2.72027016e-01 -6.81766197e-02 3.68576288e-01
-5.73689163e-01 -5.95633149e-01 4.52636302e-01 8.67642045e-01
1.73001453e-01 -7.80638084e-02 -4.09385473e-01 5.29330015e-01
2.95837745e-02 1.02473371e-01 6.57704100e-02 4.00778502e-01
-6.40048325e-01 -7.12059557e-01 -3.51974279e-01 1.14135824e-01
-3.98948252e-01 -4.25748050e-01 9.95403305e-02 4.70452160e-01
3.28431129e-01 1.11795199e+00 1.36507630e-01 -5.32188177e-01
2.52295017e-01 3.15668821e-01 3.12086344e-01 -1.27513140e-01
-2.96771508e-02 2.56209314e-01 -3.19778472e-01 -4.69403356e-01
-2.28627980e-01 -5.71552694e-01 -1.15933836e+00 -1.22193620e-01
-1.33277208e-01 -4.00646068e-02 8.87634158e-01 7.10763037e-01
5.48514068e-01 5.05320847e-01 7.10609257e-01 -9.10755694e-01
-3.10669780e-01 -1.18190765e+00 -5.38986742e-01 3.08684379e-01
6.78910673e-01 -7.89879680e-01 -5.16580462e-01 7.50913173e-02] | [14.35925006866455, 5.009498596191406] |
74c27754-3077-4ee2-a232-1ca3a0a19c93 | measuring-sentiment-bias-in-machine | 2306.07152 | null | https://arxiv.org/abs/2306.07152v1 | https://arxiv.org/pdf/2306.07152v1.pdf | Measuring Sentiment Bias in Machine Translation | Biases induced to text by generative models have become an increasingly large topic in recent years. In this paper we explore how machine translation might introduce a bias in sentiments as classified by sentiment analysis models. For this, we compare three open access machine translation models for five different languages on two parallel corpora to test if the translation process causes a shift in sentiment classes recognized in the texts. Though our statistic test indicate shifts in the label probability distributions, we find none that appears consistent enough to assume a bias induced by the translation process. | ['Munir Georges', 'Sören Gröttrup', 'Taylan Volkan', 'Khabbab Zakaria', 'Shubham Kurlekar', 'Aaricia Herygers', 'Kai Hartung'] | 2023-06-12 | null | null | null | null | ['sentiment-analysis'] | ['natural-language-processing'] | [ 1.69749022e-01 2.51948535e-01 -4.97103065e-01 -8.23486745e-01
-6.52266502e-01 -8.82861435e-01 1.28370202e+00 1.27828484e-02
-3.90396595e-01 1.07035875e+00 7.37687588e-01 -6.53681934e-01
4.26163793e-01 -7.69627869e-01 -9.23478901e-01 -7.15665400e-01
7.02483416e-01 8.36326599e-01 -1.47056744e-01 -4.35157567e-01
5.10521770e-01 -2.57549901e-02 -9.85865951e-01 5.08915246e-01
6.53675318e-01 -3.50084454e-02 -2.82821417e-01 3.69145721e-01
-2.31539816e-01 6.32380366e-01 -6.51103556e-01 -8.32508802e-01
3.08179229e-01 -6.61895812e-01 -8.44420612e-01 -1.37220221e-02
4.53477174e-01 1.49262011e-01 1.01414114e-01 1.46756375e+00
3.78254741e-01 -3.13246399e-01 1.08505225e+00 -1.01537108e+00
-1.06206524e+00 1.07407475e+00 -5.72258949e-01 4.43477005e-01
2.27812096e-01 1.06059596e-01 1.14832115e+00 -1.01548195e+00
1.05100441e+00 1.53615725e+00 4.83777523e-01 4.58967894e-01
-1.24186504e+00 -6.82770729e-01 5.09253368e-02 -2.54070610e-01
-8.49513948e-01 -2.84343511e-01 8.10223401e-01 -5.43361604e-01
6.67451680e-01 4.01258320e-01 5.46729743e-01 1.41742992e+00
9.48342383e-01 4.30253506e-01 1.86976218e+00 -6.79207861e-01
1.51879251e-01 9.23651636e-01 2.11612865e-01 3.00055593e-01
8.36827695e-01 -4.42511141e-02 -6.06273353e-01 -3.71098965e-01
1.35734171e-01 -2.19254464e-01 -6.86152056e-02 1.21400715e-03
-1.02708328e+00 1.37623107e+00 1.29874572e-01 4.78924513e-01
-3.01967829e-01 -1.59290954e-02 3.00895780e-01 5.40345192e-01
1.21366382e+00 7.33124375e-01 -7.67082214e-01 -1.53188966e-02
-7.83642292e-01 2.40953803e-01 9.12703633e-01 6.99787199e-01
8.26030314e-01 -3.28504220e-02 1.36214703e-01 7.68544734e-01
5.20202160e-01 9.98320222e-01 7.90901601e-01 -4.84708369e-01
3.55350554e-01 4.14105684e-01 1.29988179e-01 -1.01151848e+00
-1.39258057e-01 -3.60932410e-01 -5.63357286e-02 -1.90048799e-01
4.11177069e-01 -5.21653652e-01 -7.03298330e-01 1.74423444e+00
2.88420897e-02 -9.76646841e-01 2.65649080e-01 6.91443682e-01
2.93089479e-01 6.37257934e-01 2.08843559e-01 -7.03064278e-02
1.27981973e+00 -6.59217954e-01 -8.16270769e-01 -7.45937109e-01
9.13738906e-01 -1.35383630e+00 1.00275207e+00 4.27930802e-02
-7.21609771e-01 -1.71417043e-01 -8.79408181e-01 1.00913011e-02
-5.73295951e-01 -2.57986367e-01 7.76270032e-01 8.18747401e-01
-7.99688041e-01 4.05557305e-01 -4.45846230e-01 -7.75789678e-01
1.62564546e-01 1.14235193e-01 6.07045703e-02 2.11886853e-01
-1.39978325e+00 1.30928099e+00 -2.29129463e-01 -2.42120311e-01
-4.10696685e-01 -1.51196077e-01 -6.14689887e-01 -2.75640100e-01
-2.60511458e-01 -5.56397200e-01 1.38905704e+00 -1.68541086e+00
-1.28832746e+00 1.27648795e+00 -5.12074828e-01 -8.26901495e-02
5.25247991e-01 -1.63169429e-01 -5.43561876e-01 -5.50792038e-01
5.39088070e-01 3.72678518e-01 6.37073755e-01 -1.22677362e+00
-6.20556116e-01 -6.02911651e-01 -2.14229077e-01 2.80239850e-01
-1.75964892e-01 7.61549056e-01 2.86850870e-01 -6.46213591e-01
-1.39240623e-01 -1.32476091e+00 -1.74045399e-01 -7.49704897e-01
-9.06165764e-02 -3.14539969e-01 2.77963996e-01 -4.81665224e-01
8.16500843e-01 -1.71957266e+00 -6.73606396e-02 2.50730723e-01
-2.82007545e-01 -3.70639473e-01 1.41588897e-01 4.44718271e-01
-7.33300820e-02 5.98326683e-01 -2.55893753e-03 -9.11571011e-02
2.03037009e-01 3.38755667e-01 -7.97621846e-01 5.39579272e-01
1.96526408e-01 7.74023354e-01 -7.82686055e-01 -1.66128770e-01
-5.51116884e-01 3.05221170e-01 -5.38199663e-01 -1.95094511e-01
-3.04258130e-02 3.64510715e-02 -4.15094048e-01 5.32674253e-01
5.16054451e-01 -1.05085969e-01 2.87753016e-01 2.68593252e-01
-2.82118469e-01 9.55561459e-01 -2.09590986e-01 1.06209791e+00
-2.55724579e-01 9.65690553e-01 -2.92781025e-01 -4.89501953e-01
1.03093004e+00 2.51954824e-01 -2.89403111e-01 -5.12180328e-01
3.14043224e-01 6.87252522e-01 6.65180981e-01 -1.82074383e-01
7.64061809e-01 -8.71542633e-01 -2.74508685e-01 7.43511677e-01
6.13589995e-02 -4.85672325e-01 1.27845351e-02 1.90338902e-02
4.75502521e-01 -6.68062344e-02 -6.51064962e-02 -8.79540861e-01
-2.12719515e-02 5.17607689e-01 4.06698704e-01 5.88621497e-01
-1.43165877e-02 4.38053280e-01 5.38958251e-01 -3.69311541e-01
-1.09530640e+00 -4.92320299e-01 -3.82695049e-01 1.24216759e+00
-3.40108633e-01 -1.35820016e-01 -6.36066437e-01 -8.80270302e-01
-2.12474063e-01 1.00333667e+00 -1.00616848e+00 -1.43592522e-01
-1.47381097e-01 -1.46573687e+00 2.54266858e-01 2.38246456e-01
-1.37367636e-01 -9.93151963e-01 -3.53141785e-01 -7.28809834e-02
-2.08778560e-01 -7.23383069e-01 -3.99744868e-01 6.23890758e-01
-1.06043887e+00 -4.04088855e-01 -7.15110540e-01 -7.74091184e-01
9.34649348e-01 -7.22785965e-02 1.30571151e+00 -3.33017170e-01
3.70713860e-01 8.95240009e-02 -4.37926561e-01 -1.25939679e+00
-8.50678623e-01 4.77652848e-01 2.08671361e-01 -3.41513902e-01
1.22532821e+00 -1.73043367e-02 -2.74133056e-01 2.54942685e-01
-8.11786413e-01 -3.75183076e-01 5.48194170e-01 4.74981785e-01
7.48938248e-02 -3.94459784e-01 5.36961198e-01 -1.43894815e+00
8.90585244e-01 -8.11038852e-01 -3.39746773e-01 -1.01844966e-01
-1.10053968e+00 3.86012085e-02 1.56207770e-01 -4.09483373e-01
-1.02981210e+00 -5.85477769e-01 2.87541807e-05 3.50670516e-01
-2.04119667e-01 7.93683708e-01 1.92966193e-01 3.70648682e-01
8.50676835e-01 -1.40776947e-01 -2.01666936e-01 -1.49477333e-01
1.23398952e-01 9.66608465e-01 -2.74155319e-01 -5.25829911e-01
8.15467000e-01 4.90084529e-01 -4.86524761e-01 -4.57127690e-01
-1.28870118e+00 -1.13817319e-01 -4.38255727e-01 -7.78021812e-02
8.46270561e-01 -1.06560063e+00 3.86513233e-01 1.67088747e-01
-1.37849522e+00 -3.24718356e-01 -2.26780653e-01 6.81255579e-01
-1.94916844e-01 -3.13685626e-01 -6.05679512e-01 -6.83957875e-01
-3.60048503e-01 -1.12490308e+00 8.44916701e-01 1.68522209e-01
-9.93537247e-01 -1.36807966e+00 7.53047705e-01 2.36722067e-01
5.22541344e-01 -2.10416958e-01 7.69952476e-01 -1.08343446e+00
8.50465149e-02 -2.59690136e-01 1.22012176e-01 1.64536849e-01
4.49954182e-01 3.11820984e-01 -1.01752949e+00 -1.87416241e-01
4.49829966e-01 -2.27823973e-01 5.58990896e-01 3.29979509e-01
1.18977606e-01 -3.72606456e-01 -2.45364502e-01 7.25619048e-02
1.30639815e+00 2.92331964e-01 4.45855081e-01 6.54937983e-01
4.68075156e-01 9.55527604e-01 5.55856764e-01 -9.30434912e-02
1.69986978e-01 2.69571126e-01 -3.30315024e-01 -2.77109891e-01
2.05425978e-01 -3.68618548e-01 9.31408167e-01 1.16183007e+00
9.84382033e-02 -2.94943333e-01 -1.06547797e+00 6.89565241e-01
-1.49791741e+00 -8.61769915e-01 -5.01594245e-01 1.70634723e+00
1.07879341e+00 3.38945955e-01 -1.31738290e-01 -5.62629759e-01
6.81260467e-01 2.42615968e-01 -1.02343038e-01 -1.12053490e+00
-3.74344379e-01 2.23648176e-01 7.44101882e-01 7.23130345e-01
-4.17583764e-01 1.02394593e+00 7.91943598e+00 2.33164042e-01
-1.31862903e+00 2.15861917e-01 9.66473579e-01 -4.68671471e-02
-8.40730906e-01 2.65457094e-01 -8.78472805e-01 5.10838509e-01
1.05604327e+00 -3.57289791e-01 -1.76146269e-01 7.58609056e-01
2.03751802e-01 -4.38958257e-02 -1.04262340e+00 2.50083625e-01
4.70628917e-01 -7.55138874e-01 2.98161983e-01 4.46799845e-01
1.30094790e+00 4.74191129e-01 2.82533407e-01 2.87437797e-01
7.67251730e-01 -8.16323459e-01 9.35232401e-01 2.55197823e-01
3.42187375e-01 -6.30447268e-01 1.13504446e+00 1.59478649e-01
1.69885397e-01 4.24666375e-01 -6.71928704e-01 -4.77661937e-01
-8.84566605e-02 7.54719734e-01 -9.72828388e-01 3.62074859e-02
7.06744790e-01 3.53942662e-01 -7.50207603e-01 2.80771285e-01
-6.21445954e-01 1.22923231e+00 8.15265849e-02 -5.37452877e-01
3.15685630e-01 -3.90824795e-01 5.48393726e-01 1.27854931e+00
3.19320351e-01 -3.92951816e-01 -3.28129828e-01 7.61593461e-01
-2.81366765e-01 4.48821783e-01 -1.04696810e+00 -1.86637014e-01
-9.02986377e-02 1.08850884e+00 -9.20156538e-01 -5.49056649e-01
-5.88711917e-01 1.04005027e+00 -8.87603238e-02 3.19577277e-01
-7.06116140e-01 -6.35143071e-02 5.37368357e-01 1.80535257e-01
9.67831239e-02 1.96682662e-01 -5.69835186e-01 -1.48657918e+00
-2.33411610e-01 -1.19661200e+00 -7.36987814e-02 -9.54960346e-01
-1.55085957e+00 6.85338199e-01 -2.40028456e-01 -7.37819731e-01
-2.62608051e-01 -7.56630778e-01 -5.97253740e-01 1.15344191e+00
-1.24742484e+00 -8.21180642e-01 4.57525641e-01 1.29798278e-01
6.37510002e-01 -2.30054170e-01 7.11525798e-01 -9.59186032e-02
-3.72766137e-01 3.12799215e-01 1.63731650e-01 2.00064778e-01
1.51260471e+00 -1.27323771e+00 5.27705669e-01 8.46810400e-01
3.02586526e-01 1.11525917e+00 1.38005233e+00 -8.98687422e-01
-7.29298294e-01 -8.41540992e-01 1.61726332e+00 -1.24164534e+00
8.91367078e-01 -3.74561310e-01 -6.77312672e-01 1.08762145e+00
9.44463670e-01 -6.69952631e-01 1.12825525e+00 4.25970465e-01
-4.48437274e-01 3.55810642e-01 -8.74862194e-01 6.38534248e-01
7.15333343e-01 -5.33321321e-01 -1.04357708e+00 5.17437398e-01
3.98732573e-01 -3.39604430e-02 -5.33109069e-01 2.28208527e-01
4.27220494e-01 -4.63422656e-01 1.32720962e-01 -9.60465968e-01
9.87898946e-01 -9.11241099e-02 -3.82029384e-01 -1.84939981e+00
-4.45927888e-01 -3.03940982e-01 7.87516475e-01 1.34372687e+00
1.18187642e+00 -1.12061131e+00 5.28938591e-01 7.12005615e-01
1.86469749e-01 -5.04605770e-01 -4.68546599e-01 -4.42205608e-01
8.38418663e-01 -1.77275136e-01 4.17904019e-01 1.43330097e+00
2.94859082e-01 1.03146625e+00 -9.35879648e-02 -1.37609258e-01
2.32829273e-01 2.76184648e-01 8.01617026e-01 -1.19484210e+00
-1.65381506e-01 -5.29078364e-01 3.88918743e-02 -5.11596143e-01
2.83800095e-01 -1.09454513e+00 2.07200989e-01 -1.51617575e+00
7.22092509e-01 -2.12086439e-01 -1.35023564e-01 3.63059074e-01
-3.31689864e-01 2.47563094e-01 -1.07663296e-01 4.71844584e-01
-1.42131612e-01 3.05134505e-01 9.85614002e-01 -1.05869688e-01
2.18928680e-01 -1.70645207e-01 -1.37466598e+00 1.04787183e+00
1.09402180e+00 -1.02445853e+00 -1.18683249e-01 -7.04345405e-01
1.09879184e+00 -7.79679656e-01 -3.52505185e-02 -2.68829376e-01
-3.44596356e-01 -1.83230132e-01 5.20286977e-01 -1.17872888e-02
-3.17423761e-01 -5.95948517e-01 -1.05825819e-01 4.57854509e-01
-5.02323985e-01 5.24435401e-01 3.49400192e-01 2.71924227e-01
-3.12750280e-01 -3.57255220e-01 5.34797549e-01 -2.14502648e-01
8.82070139e-02 -3.41914117e-01 -9.43895996e-01 8.77481028e-02
4.55594122e-01 2.73115724e-01 -4.52332169e-01 -3.91016841e-01
-3.58065337e-01 -3.05608958e-01 8.62474263e-01 6.53852224e-01
-2.16947451e-01 -1.28915775e+00 -9.69342053e-01 -2.26829126e-01
1.29289776e-01 -7.96146631e-01 -5.14361918e-01 8.28575253e-01
-1.73644006e-01 6.25929654e-01 -6.06548823e-02 -2.70691156e-01
-1.21093428e+00 2.93684304e-01 1.95886150e-01 -3.26509297e-01
2.32521087e-01 6.86684251e-01 1.64642587e-01 -7.10086167e-01
-5.91368198e-01 -2.44043276e-01 2.17421999e-04 3.89677197e-01
1.43637493e-01 9.54911932e-02 3.43320891e-02 -8.04044604e-01
-3.51931453e-01 4.37933266e-01 -2.75767237e-01 -6.29682779e-01
1.29206371e+00 -1.23721227e-01 -4.14120942e-01 1.19725358e+00
1.01859105e+00 7.21032798e-01 -5.22312105e-01 -2.08286587e-02
-3.91767034e-03 -3.37815732e-01 -5.41725419e-02 -1.14094567e+00
-6.36737466e-01 5.97267091e-01 3.90688837e-01 2.94292867e-01
3.93390447e-01 1.89245194e-01 3.08445185e-01 2.00538039e-01
4.16664064e-01 -1.34051836e+00 -5.04962921e-01 6.96276963e-01
9.58884835e-01 -1.40877807e+00 1.01967864e-01 -2.06752315e-01
-8.01991284e-01 7.37610579e-01 4.20352936e-01 -1.97342068e-01
5.38620114e-01 8.18212852e-02 7.79813945e-01 -4.00423110e-01
-9.79181111e-01 3.27912830e-02 1.15143113e-01 1.07002579e-01
1.21786559e+00 1.95975944e-01 -1.07953322e+00 8.79150182e-02
-6.68160498e-01 -3.49367321e-01 8.30567002e-01 6.52852893e-01
-2.68546999e-01 -1.30558634e+00 -5.49038351e-01 4.45516199e-01
-1.03860736e+00 -3.29609454e-01 -1.20129561e+00 7.58299649e-01
1.14687361e-01 1.01753902e+00 2.93537349e-01 -1.91045254e-01
-1.89596079e-02 7.17159629e-01 3.37320358e-01 -9.82655227e-01
-7.80081451e-01 2.57120281e-01 3.11568201e-01 2.19196141e-01
-6.80088520e-01 -1.17074943e+00 -8.47102821e-01 -1.16367087e-01
-6.53992116e-01 5.05610406e-01 8.77059877e-01 8.62399459e-01
1.65964916e-01 2.66122043e-01 5.21061361e-01 -1.52945787e-01
-4.37714994e-01 -1.53268409e+00 -2.65624702e-01 5.54277718e-01
-8.69803131e-02 -2.50301540e-01 -9.32101250e-01 1.63287997e-01] | [9.85326099395752, 10.140080451965332] |
bfdacfd1-e077-47c2-8c6b-091a8db56506 | end-to-end-lung-nodule-detection-in-computed | 1711.02074 | null | http://arxiv.org/abs/1711.02074v2 | http://arxiv.org/pdf/1711.02074v2.pdf | End-to-end Lung Nodule Detection in Computed Tomography | Computer aided diagnostic (CAD) system is crucial for modern med-ical
imaging. But almost all CAD systems operate on reconstructed images, which were
optimized for radiologists. Computer vision can capture features that is subtle
to human observers, so it is desirable to design a CAD system op-erating on the
raw data. In this paper, we proposed a deep-neural-network-based detection
system for lung nodule detection in computed tomography (CT). A
primal-dual-type deep reconstruction network was applied first to convert the
raw data to the image space, followed by a 3-dimensional convolutional neural
network (3D-CNN) for the nodule detection. For efficient network training, the
deep reconstruction network and the CNN detector was trained sequentially
first, then followed by one epoch of end-to-end fine tuning. The method was
evaluated on the Lung Image Database Consortium image collection (LIDC-IDRI)
with simulated forward projections. With 144 multi-slice fanbeam pro-jections,
the proposed end-to-end detector could achieve comparable sensitivity with the
reference detector, which was trained and applied on the fully-sampled image
data. It also demonstrated superior detection performance compared to detectors
trained on the reconstructed images. The proposed method is general and could
be expanded to most detection tasks in medical imaging. | ['Kyungsang Kim', 'Dufan Wu', 'Bin Dong', 'Quanzheng Li', 'Georges El Fakhri'] | 2017-11-06 | null | null | null | null | ['lung-nodule-detection'] | ['medical'] | [ 3.54998767e-01 2.00028181e-01 -5.67644415e-03 -2.61894494e-01
-9.85345006e-01 -9.70269293e-02 2.02478305e-01 -1.78264990e-01
-7.12019980e-01 -3.47493179e-02 -1.71148881e-01 -7.61652052e-01
6.61789253e-02 -6.10343397e-01 -3.17482799e-01 -7.42039382e-01
-1.28736362e-01 7.75523543e-01 6.41748369e-01 3.62531483e-01
-3.51615012e-01 7.67048836e-01 -8.87045443e-01 4.90584165e-01
1.54657766e-01 1.02946258e+00 8.41961086e-01 1.13088012e+00
3.22494447e-01 1.06572258e+00 3.58971581e-02 -1.90243982e-02
3.38917047e-01 -5.31218410e-01 -8.26524794e-01 1.71293050e-01
-2.65859924e-02 -7.89962888e-01 -6.52603984e-01 7.60668814e-01
8.43451977e-01 -4.73849833e-01 7.90367603e-01 -4.70291495e-01
-3.32302362e-01 3.21439832e-01 -4.53128904e-01 6.39020026e-01
-1.81797385e-01 2.25957349e-01 4.86738890e-01 -1.11849165e+00
3.32131475e-01 8.19769382e-01 8.23904872e-01 6.69270098e-01
-7.84435809e-01 -3.83652866e-01 -7.53420353e-01 -1.84980687e-02
-1.12276983e+00 3.50035541e-02 3.75255585e-01 -6.61625147e-01
7.29398668e-01 2.56320715e-01 7.33002663e-01 7.18344927e-01
5.58002710e-01 6.72873914e-01 6.98760629e-01 -5.27016699e-01
-1.36037201e-01 1.04213335e-01 -1.36280358e-01 1.13800621e+00
4.36361521e-01 6.47435129e-01 3.46881002e-01 -2.38916025e-01
1.36635947e+00 2.13059157e-01 -3.42109025e-01 -5.43954194e-01
-1.47303915e+00 9.61670756e-01 1.01551068e+00 4.42543298e-01
-7.14249432e-01 8.99790674e-02 4.56053525e-01 -2.13440597e-01
3.46386805e-02 4.47464287e-01 -4.80966531e-02 3.73297930e-01
-9.33511436e-01 -2.08404481e-01 5.91538489e-01 3.96145105e-01
-6.80758478e-03 -1.03541650e-01 -6.50843084e-01 6.28734052e-01
4.61415291e-01 4.34775412e-01 9.70193624e-01 -6.23715699e-01
-8.98964610e-03 4.61637974e-01 -2.90373772e-01 -5.30117989e-01
-1.05918479e+00 -7.81959534e-01 -1.26388168e+00 2.65685350e-01
4.14584428e-01 -1.46158874e-01 -1.36489809e+00 1.01907802e+00
2.90591776e-01 -1.73810646e-01 -1.52553797e-01 1.41839385e+00
1.08212924e+00 2.86663383e-01 -9.12769809e-02 -3.86574864e-02
1.69876492e+00 -8.66651297e-01 -2.93543071e-01 -1.18471444e-01
7.59793162e-01 -8.89736176e-01 6.90107584e-01 1.37531608e-01
-1.12268615e+00 -6.15338385e-01 -1.11184537e+00 -5.14832176e-02
2.65103102e-01 6.86503708e-01 3.36536735e-01 6.19414747e-01
-1.01127779e+00 2.36834049e-01 -1.31197882e+00 -4.69975531e-01
6.08862281e-01 5.19118071e-01 -1.67735413e-01 -5.21588810e-02
-8.29583764e-01 9.64805186e-01 5.07547259e-01 1.84832960e-01
-1.15456223e+00 -6.56496644e-01 -4.31697339e-01 2.80676819e-02
2.64863968e-01 -1.12700570e+00 1.90088272e+00 -8.13805282e-01
-1.41507757e+00 1.11101258e+00 2.46920511e-01 -6.10289276e-01
7.44361639e-01 3.32882881e-01 -1.54253036e-01 4.41323608e-01
2.04248309e-01 5.78478634e-01 7.81535268e-01 -7.49636114e-01
-6.04466677e-01 -1.98633686e-01 -3.31702799e-01 1.26206756e-01
1.29771188e-01 -8.62734616e-02 -6.77638412e-01 -3.85679483e-01
5.22048891e-01 -9.43705559e-01 -5.71388543e-01 4.29604977e-01
-4.91042733e-01 2.73042440e-01 7.71284938e-01 -5.96253693e-01
9.22606528e-01 -1.98571086e+00 -3.14405560e-01 1.98391140e-01
5.53459525e-01 4.69583660e-01 4.10280153e-02 -1.82945535e-01
-4.70113665e-01 -3.06265354e-01 -1.14935979e-01 -9.65181179e-03
-6.29725397e-01 4.40945178e-02 3.96082580e-01 7.70054102e-01
8.53632241e-02 1.06970906e+00 -7.55851507e-01 -7.91394114e-01
4.97817367e-01 3.94930452e-01 -5.68396807e-01 5.30341387e-01
3.18422139e-01 5.76331377e-01 -5.15773535e-01 5.80895841e-01
5.29282212e-01 -9.45996046e-01 9.66984257e-02 -4.16827261e-01
-7.13971537e-03 -9.53318030e-02 -7.65234053e-01 1.33680761e+00
-4.78404850e-01 6.62036419e-01 1.71247810e-01 -8.35001945e-01
5.41418254e-01 6.51509285e-01 7.72419810e-01 -7.48735070e-01
4.07736510e-01 3.38278770e-01 5.77466309e-01 -8.76804590e-01
-1.83115080e-01 -5.37597418e-01 2.70614922e-01 3.78097087e-01
-1.42748296e-01 -3.39229435e-01 -1.68909267e-01 -5.41249551e-02
1.36451828e+00 -6.00294232e-01 6.89811885e-01 -1.05638787e-01
7.06920683e-01 4.06229675e-01 1.40000105e-01 9.18555856e-01
-2.71446168e-01 9.79845703e-01 2.91139603e-01 -7.11485982e-01
-1.12665641e+00 -1.15187573e+00 -6.77627981e-01 6.63170755e-01
-1.33660480e-01 2.21731156e-01 -6.89473212e-01 -7.98450828e-01
-2.65788645e-01 2.42215306e-01 -5.12464523e-01 -1.60748288e-01
-6.83431268e-01 -8.41539979e-01 6.18304670e-01 6.63655758e-01
6.42503619e-01 -1.03173912e+00 -1.24634337e+00 3.89055610e-01
6.92573264e-02 -8.85895371e-01 -4.70617145e-01 6.80593371e-01
-1.01927614e+00 -1.45806718e+00 -9.30529833e-01 -1.04713619e+00
8.50380421e-01 4.57447797e-01 1.00551248e+00 1.70088232e-01
-8.15500021e-01 9.39233005e-02 -1.76861063e-02 -2.32784882e-01
-6.93615377e-01 1.78158320e-02 -2.45999604e-01 -3.81506413e-01
2.70174533e-01 9.13650393e-02 -8.83571029e-01 2.89567977e-01
-8.15050364e-01 3.39983195e-01 1.34199512e+00 1.36308813e+00
6.91136122e-01 -1.33853406e-01 1.16139650e-01 -1.05259478e+00
3.19112808e-01 -1.31578177e-01 -6.17674232e-01 4.27057669e-02
-2.43723571e-01 4.13666852e-03 5.40225983e-01 -1.21365547e-01
-9.81940150e-01 6.08623683e-01 -4.44747806e-01 -5.08535147e-01
8.10494572e-02 4.55910563e-01 4.33955491e-01 -2.23802492e-01
1.09334433e+00 1.20648295e-01 1.17564529e-01 -1.49631187e-01
-3.91971320e-02 7.45816588e-01 6.22074366e-01 1.69944108e-01
6.07877314e-01 5.13038397e-01 2.25899667e-01 -6.86516881e-01
-8.68160903e-01 -7.65719950e-01 -8.34922254e-01 -8.59205276e-02
1.19550860e+00 -1.01924992e+00 -3.17346066e-01 3.41372579e-01
-1.10061693e+00 -1.08984016e-01 -3.13783914e-01 1.06743252e+00
-3.77482563e-01 2.33033717e-01 -8.63840997e-01 -2.55964577e-01
-6.47914767e-01 -1.36929381e+00 1.07572114e+00 7.43230134e-02
9.15981233e-02 -8.61491799e-01 1.49918959e-01 8.82275775e-02
5.58713317e-01 -5.28922044e-02 8.48858297e-01 -6.62832081e-01
-4.83512223e-01 -6.42453730e-01 -6.43813789e-01 4.24479723e-01
1.53962448e-01 -2.04479918e-01 -9.63934124e-01 -3.66215169e-01
3.69528443e-01 -2.06335708e-01 9.69276428e-01 9.80190873e-01
1.49378800e+00 1.18686236e-01 -5.54469824e-01 8.98052096e-01
1.31546974e+00 2.86680400e-01 3.47692519e-01 1.39089361e-01
6.02445424e-01 -1.40152529e-01 2.11288169e-01 2.45804518e-01
-2.02591866e-01 4.32988763e-01 5.76343954e-01 -8.13688934e-01
-4.15940434e-01 -1.35766193e-02 -2.98380911e-01 4.95564252e-01
-9.08939764e-02 1.63042508e-02 -1.20749021e+00 2.87038296e-01
-1.35405576e+00 -7.10500062e-01 -3.55983168e-01 1.89745259e+00
4.42545950e-01 2.66232580e-01 -7.50520080e-02 -1.31598532e-01
6.45888150e-01 -2.65276223e-01 -5.55913687e-01 3.23731899e-02
4.15481150e-01 4.50793236e-01 8.86380434e-01 1.22001417e-01
-1.34293032e+00 2.11898655e-01 6.71628761e+00 5.96917689e-01
-1.56709051e+00 4.44944352e-01 7.02053308e-01 2.84477305e-02
4.21881348e-01 -4.58240122e-01 -3.33724856e-01 3.34894508e-02
6.76247537e-01 4.32167649e-02 -1.57537177e-01 1.16092396e+00
2.60483176e-01 -1.84893653e-01 -1.29817307e+00 9.65302527e-01
-2.50325948e-01 -1.32363677e+00 -2.06502080e-01 9.51426849e-02
4.13553238e-01 4.19241846e-01 8.41291770e-02 3.28723788e-01
-9.64347087e-03 -1.13022316e+00 9.26145762e-02 2.18749881e-01
1.07509911e+00 -4.92710561e-01 1.14407945e+00 6.68018758e-01
-9.86907125e-01 3.95851135e-02 -4.97481614e-01 3.62177998e-01
-2.63980255e-02 5.90337455e-01 -1.80571079e+00 3.23003948e-01
4.84568477e-01 3.70889038e-01 -6.65952623e-01 1.26054931e+00
5.61962761e-02 5.31883240e-01 -2.24405482e-01 -1.82471331e-02
2.77239531e-01 2.46556923e-01 2.60014951e-01 1.43886817e+00
6.03321910e-01 2.77080238e-01 1.25404239e-01 7.52852440e-01
-1.08320013e-01 -2.97372967e-01 -4.43608791e-01 2.31223971e-01
1.47727743e-01 1.51787674e+00 -8.49767268e-01 -4.48449254e-01
-4.41284925e-01 8.23266804e-01 3.22014652e-02 1.67847574e-02
-1.05894077e+00 5.12997769e-02 -1.78874612e-01 6.01690829e-01
3.54570389e-01 1.75135061e-01 -1.69078976e-01 -7.73429513e-01
-4.15983707e-01 -5.31099617e-01 4.58441913e-01 -6.95973277e-01
-1.20479441e+00 8.91443670e-01 -2.33530283e-01 -1.45980704e+00
-5.89753330e-01 -1.04740417e+00 -6.76943541e-01 8.00938725e-01
-1.21401501e+00 -9.41894293e-01 -5.16345382e-01 5.59617519e-01
4.08899277e-01 -1.61230475e-01 8.22973013e-01 3.79277080e-01
-4.65809584e-01 4.41420346e-01 5.99671602e-02 5.20231485e-01
3.96401763e-01 -1.21530592e+00 5.63775711e-02 7.08378971e-01
-2.10173607e-01 6.32588044e-02 1.15495861e-01 -2.75093496e-01
-1.32770026e+00 -1.46200430e+00 3.82716537e-01 -2.02106148e-01
3.91770422e-01 2.28057489e-01 -7.60697365e-01 6.43072426e-01
4.03169990e-02 6.64521635e-01 4.34857160e-01 -7.02688515e-01
3.32722425e-01 1.85530186e-01 -1.27811706e+00 1.42507106e-01
4.81009543e-01 -3.38188380e-01 -5.22663951e-01 6.28990173e-01
4.01576072e-01 -9.93062973e-01 -6.05526209e-01 5.46016157e-01
4.38015580e-01 -1.02132320e+00 1.18057191e+00 -2.38200709e-01
3.65183413e-01 -1.79586962e-01 1.29566029e-01 -9.74067211e-01
-8.31562877e-01 1.27111301e-01 1.66537419e-01 -4.04097214e-02
4.31641906e-01 -3.76146942e-01 9.81004715e-01 2.03252897e-01
-3.85367274e-01 -8.95869017e-01 -1.09326029e+00 -4.33487862e-01
-1.87214632e-02 -3.40163916e-01 -4.91547585e-02 4.59940195e-01
-4.18186128e-01 2.33011588e-01 1.27817154e-01 3.18018764e-01
6.08713269e-01 -1.31805331e-01 3.55550289e-01 -9.17663336e-01
-6.66380703e-01 -5.05938053e-01 -3.87226552e-01 -1.12009275e+00
-6.92353010e-01 -1.10408843e+00 1.07011549e-01 -1.60563290e+00
5.11375785e-01 -1.56052321e-01 -2.63617575e-01 7.14048445e-02
-1.77530736e-01 2.56917119e-01 -8.69207010e-02 4.62555498e-01
-1.83273762e-01 2.65944023e-02 1.86288261e+00 -2.35172734e-02
4.69545312e-02 5.28320551e-01 -2.11446449e-01 8.49781096e-01
6.11612022e-01 -6.96623921e-01 -4.29855660e-02 -2.66286612e-01
-3.37332219e-01 6.50988221e-01 7.92570889e-01 -1.41144025e+00
5.26631415e-01 2.67787546e-01 9.89811063e-01 -1.03095412e+00
3.01545590e-01 -1.03141308e+00 1.25404179e-01 1.35948527e+00
-2.04386592e-01 -2.78800756e-01 -4.66229772e-04 3.65935445e-01
-3.06419373e-01 -3.19946170e-01 1.35049689e+00 -5.12715578e-01
-5.00851512e-01 4.42090213e-01 -5.36638141e-01 -2.55711436e-01
1.25744188e+00 -2.99787909e-01 1.13292381e-01 1.05192594e-01
-8.27181518e-01 -1.43944286e-02 3.51415873e-02 -2.06144512e-01
9.33791101e-01 -1.42670643e+00 -9.70498145e-01 5.51610887e-01
7.94617832e-02 3.72583598e-01 2.57086128e-01 1.20936596e+00
-1.20459235e+00 8.66161287e-01 -5.10704592e-02 -1.13393509e+00
-1.12495124e+00 6.42502248e-01 1.15272689e+00 -6.29785120e-01
-7.49732852e-01 1.05789888e+00 5.22481978e-01 -4.45824206e-01
1.32492781e-01 -8.01170170e-01 3.87254953e-02 -3.69789213e-01
3.62831146e-01 -8.44604224e-02 3.74648094e-01 -2.66673654e-01
-4.49778706e-01 4.49262649e-01 -2.58689672e-01 1.39859676e-01
1.04360366e+00 3.12358886e-01 3.62088591e-01 -3.49994153e-02
1.30505860e+00 -6.36151373e-01 -9.21431959e-01 -3.74394715e-01
-1.81874514e-01 -2.97747284e-01 6.12680793e-01 -6.77403033e-01
-1.29840231e+00 1.01862192e+00 1.18899345e+00 1.03101552e-01
1.21612513e+00 2.06768021e-01 5.44044673e-01 4.48746592e-01
7.26155238e-03 -5.41053712e-01 3.56934220e-01 2.88622051e-01
6.92499936e-01 -1.68180394e+00 6.90806657e-02 -2.92905688e-01
-6.29928410e-01 1.55237675e+00 7.06639409e-01 -2.35685810e-01
8.53959143e-01 3.79596055e-01 1.73636794e-01 -6.83864057e-01
-6.12240613e-01 -2.66611814e-01 5.28734803e-01 4.58155513e-01
7.46787667e-01 2.80246317e-01 3.28042693e-02 4.15595412e-01
2.50313431e-01 3.12804818e-01 4.85128105e-01 9.32490885e-01
-7.41541922e-01 -5.73570788e-01 -6.88585639e-01 8.35713267e-01
-4.52873379e-01 3.70426551e-02 -2.99284346e-02 1.19689953e+00
1.64903119e-01 4.95415568e-01 1.58152729e-01 -7.02033043e-02
4.87435162e-01 -4.70823973e-01 3.20179671e-01 -8.65432560e-01
-7.38909423e-01 3.18385273e-01 -2.21094549e-01 -4.56622511e-01
-1.09134465e-01 -4.88639057e-01 -1.37691331e+00 -5.62492087e-02
-4.36484516e-01 -1.79118469e-01 4.63873535e-01 5.89916527e-01
-1.14155382e-01 1.07794464e+00 7.90558696e-01 -7.59540915e-01
-7.96539366e-01 -1.06436741e+00 -5.02075732e-01 -3.94017361e-02
4.66808170e-01 -2.92212039e-01 -2.30499789e-01 -1.40971774e-02] | [15.365561485290527, -2.1189513206481934] |
dd00f8cb-488a-4c32-9138-27e4de44a196 | benchmarking-robot-manipulation-with-the | 2202.07074 | null | https://arxiv.org/abs/2202.07074v1 | https://arxiv.org/pdf/2202.07074v1.pdf | Benchmarking Robot Manipulation with the Rubik's Cube | Benchmarks for robot manipulation are crucial to measuring progress in the field, yet there are few benchmarks that demonstrate critical manipulation skills, possess standardized metrics, and can be attempted by a wide array of robot platforms. To address a lack of such benchmarks, we propose Rubik's cube manipulation as a benchmark to measure simultaneous performance of precise manipulation and sequential manipulation. The sub-structure of the Rubik's cube demands precise positioning of the robot's end effectors, while its highly reconfigurable nature enables tasks that require the robot to manage pose uncertainty throughout long sequences of actions. We present a protocol for quantitatively measuring both the accuracy and speed of Rubik's cube manipulation. This protocol can be attempted by any general-purpose manipulator, and only requires a standard 3x3 Rubik's cube and a flat surface upon which the Rubik's cube initially rests (e.g. a table). We demonstrate this protocol for two distinct baseline approaches on a PR2 robot. The first baseline provides a fundamental approach for pose-based Rubik's cube manipulation. The second baseline demonstrates the benchmark's ability to quantify improved performance by the system, particularly that resulting from the integration of pre-touch sensing. To demonstrate the benchmark's applicability to other robot platforms and algorithmic approaches, we present the functional blocks required to enable the HERB robot to manipulate the Rubik's cube via push-grasping. | ['Joshua R. Smith', 'Siddhartha S. Srinivasa', 'Patrick E. Lancaster', 'Boling Yang'] | 2022-02-14 | null | null | null | null | ['rubik-s-cube', 'robot-manipulation'] | ['graphs', 'robots'] | [ 6.24257326e-02 1.82704687e-01 -5.61682656e-02 1.14832625e-01
-6.08299911e-01 -1.02746904e+00 2.10194990e-01 -3.64268601e-01
-2.08666220e-01 6.01166964e-01 -4.04970706e-01 -2.19669685e-01
-7.13114738e-01 -6.41596019e-01 -1.01713836e+00 -4.66161847e-01
-4.41994011e-01 8.73079777e-01 4.71206248e-01 -8.59915197e-01
5.22683203e-01 9.43461120e-01 -1.51914513e+00 -9.44332853e-02
3.26469719e-01 9.86207962e-01 7.60911226e-01 7.49751389e-01
5.38473606e-01 4.60310549e-01 -5.23955524e-01 4.20156270e-01
6.89583957e-01 2.19368026e-01 -7.60768294e-01 -2.40695462e-01
1.18217640e-01 -5.37170410e-01 -2.85408169e-01 6.82902455e-01
2.88320333e-01 2.64683247e-01 9.07567918e-01 -1.53657031e+00
-1.99308112e-01 7.57036150e-01 -3.42649609e-01 -3.79373580e-01
6.80927575e-01 5.21241069e-01 8.27638745e-01 -6.55153275e-01
8.67141366e-01 1.45578825e+00 6.43741727e-01 2.85589665e-01
-1.07324374e+00 -4.71921623e-01 -1.64712280e-01 -3.15474659e-01
-1.22026193e+00 -2.47137398e-01 1.67669892e-01 -6.30044520e-01
9.94840384e-01 5.01104519e-02 2.94207871e-01 9.89409208e-01
8.74841213e-01 2.07292974e-01 6.59628332e-01 -3.18926364e-01
1.58524841e-01 -4.17100847e-01 -6.04861081e-02 4.90763992e-01
6.66810751e-01 1.66328952e-01 -1.41708180e-01 2.17538074e-01
1.03823388e+00 -6.07096814e-02 -2.29841433e-02 -1.00453115e+00
-1.54290497e+00 1.70698389e-01 6.22975588e-01 2.25049667e-02
-4.98713344e-01 7.59549618e-01 4.03289109e-01 3.67413610e-01
-7.03158677e-01 1.21509719e+00 -6.15534246e-01 -4.34905797e-01
-6.74433336e-02 8.89984012e-01 1.02366626e+00 1.78543079e+00
3.49125594e-01 -3.59324187e-01 3.45313810e-02 4.33422804e-01
3.77116874e-02 5.96255839e-01 -2.10221633e-01 -1.55916059e+00
8.66025686e-01 5.02883613e-01 7.82178402e-01 -9.08814132e-01
-8.52693677e-01 3.36786151e-01 -3.78885061e-01 8.89295518e-01
5.47972560e-01 -1.60099357e-01 -6.32347226e-01 1.31806660e+00
1.68631867e-01 -1.01747715e+00 1.37092531e-01 9.92595673e-01
3.43019933e-01 2.05026001e-01 -5.54501116e-01 1.98689774e-01
1.38848102e+00 -8.38321149e-01 -3.82280856e-01 6.32555559e-02
5.14462888e-01 -5.97125173e-01 1.26792693e+00 7.36461520e-01
-1.09509635e+00 -3.27738494e-01 -1.22959352e+00 -9.37243178e-02
-3.20338190e-01 -5.86515591e-02 5.86670041e-01 1.20383509e-01
-6.79168940e-01 9.34252799e-01 -9.69404876e-01 -4.13900465e-01
1.74773578e-02 7.11201847e-01 -6.77654505e-01 -1.47886276e-01
-6.10123813e-01 1.75576591e+00 5.15238762e-01 2.16699526e-01
-1.10663295e+00 -3.66969526e-01 -4.34673548e-01 -5.54167144e-02
8.12389076e-01 -4.40454364e-01 1.64193249e+00 2.38173068e-01
-2.09187818e+00 3.76201093e-01 6.76223576e-01 8.14574137e-02
6.69230580e-01 -5.39437234e-01 4.30705726e-01 2.49420166e-01
2.19945088e-01 6.28608584e-01 7.65234411e-01 -1.40409935e+00
-3.38711679e-01 -2.63591528e-01 3.30498368e-01 1.87461317e-01
3.91643912e-01 -3.13518256e-01 -1.17320597e-01 -2.37585530e-01
2.80914187e-01 -1.46922171e+00 -2.73366243e-01 5.53109590e-03
-5.07179141e-01 -8.92687291e-02 7.55979240e-01 -5.81099130e-02
3.85115713e-01 -1.91443086e+00 6.04027271e-01 2.32511252e-01
-1.15062296e-01 -2.05683634e-01 -2.53534466e-01 9.92950737e-01
3.61031771e-01 4.69588488e-02 7.01594055e-02 3.75355661e-01
2.58253396e-01 2.87565321e-01 -1.72750264e-01 4.09310490e-01
8.02020058e-02 9.49728668e-01 -9.16057825e-01 3.18879634e-02
7.66045973e-02 -1.06568441e-01 -6.25825346e-01 2.60709137e-01
-4.54670221e-01 4.10401881e-01 -5.24943769e-01 9.10474300e-01
2.59394169e-01 2.86579967e-01 1.54123440e-01 -1.80953562e-01
-3.42495561e-01 1.36696264e-01 -1.11236870e+00 1.69008493e+00
-3.44742209e-01 1.86532319e-01 6.79634988e-01 -5.48868120e-01
1.08473670e+00 1.00439370e-01 5.06328940e-01 -2.78326958e-01
4.25140679e-01 6.94123685e-01 1.67678207e-01 -6.38762951e-01
8.84179592e-01 1.47918895e-01 -4.09697920e-01 3.21277142e-01
-2.99128979e-01 -1.31561363e+00 2.27986172e-01 -2.81528562e-01
1.61144805e+00 7.02639937e-01 2.18998641e-01 -4.17990983e-01
-5.13878204e-02 7.36255407e-01 -7.11411089e-02 8.86642337e-01
1.02769978e-01 4.90860045e-01 3.75032187e-01 -5.36605604e-02
-1.24542642e+00 -1.31256795e+00 1.00732893e-01 1.15216327e+00
6.56265318e-01 -1.27141550e-01 -3.38827938e-01 -7.15490133e-02
8.39586437e-01 6.90726578e-01 -4.79261220e-01 -1.83512434e-01
-7.83791721e-01 5.08419797e-02 5.95211446e-01 7.98661292e-01
7.44301751e-02 -9.74258780e-01 -1.38712478e+00 2.96155572e-01
1.82162449e-01 -9.86427307e-01 1.94194820e-02 5.41758120e-01
-8.85613382e-01 -1.44892991e+00 -2.00268537e-01 -7.10417926e-01
5.24697840e-01 5.34333050e-01 9.61833119e-01 -9.85549688e-02
-3.93345028e-01 7.07030654e-01 -6.83867991e-01 -4.19787169e-01
-4.57530290e-01 2.73071527e-01 4.53553468e-01 -1.41968942e+00
-4.94477659e-01 -7.95976579e-01 -4.24692810e-01 8.35906982e-01
-5.60761988e-01 -3.63745719e-01 1.10220516e+00 5.58245540e-01
3.35444093e-01 1.77291870e-01 3.17985833e-01 -1.97652355e-02
7.39147604e-01 -3.22532356e-01 -7.80855775e-01 -1.08136483e-01
-1.41077310e-01 -3.54741439e-02 4.22182262e-01 -5.24002969e-01
-5.96362591e-01 2.58911490e-01 3.67573172e-01 -2.67625570e-01
1.33004576e-01 3.96645278e-01 1.52401458e-02 -3.06034774e-01
7.28799880e-01 -3.48963141e-01 6.24963641e-01 -4.78197068e-01
5.50015867e-01 6.08912826e-01 8.33715677e-01 -1.22628331e+00
8.14783812e-01 2.11182311e-01 5.49034357e-01 -5.25105774e-01
-5.70701808e-02 -1.31380558e-01 -9.57611859e-01 -9.35213044e-02
6.32323384e-01 -4.23260182e-01 -1.73965013e+00 1.78525284e-01
-1.26313365e+00 -7.84960151e-01 -2.29837060e-01 3.79746497e-01
-1.19753659e+00 -1.40801698e-01 -6.22998774e-01 -7.23154247e-01
-1.55221015e-01 -1.49896824e+00 1.44714046e+00 -2.26916716e-01
-5.12531042e-01 -2.38502864e-02 -2.80345857e-01 -1.64160803e-02
5.01736283e-01 6.18778586e-01 8.68593693e-01 -2.45753586e-01
-7.33815968e-01 -3.83312166e-01 -9.72141773e-02 8.28612503e-03
3.11713934e-01 1.11531295e-01 -1.27207279e-01 -5.38219690e-01
-2.57923156e-01 -6.59079432e-01 6.35520518e-02 -1.15391918e-01
6.19623125e-01 -1.56014964e-01 -5.38162053e-01 1.21953599e-02
1.22579241e+00 2.92327285e-01 6.78080022e-01 8.75338674e-01
5.19338608e-01 8.05579007e-01 1.27383220e+00 7.68247396e-02
9.55896229e-02 1.07834852e+00 1.04410911e+00 8.61824036e-01
3.09694529e-01 1.20957278e-01 6.23669744e-01 4.93776977e-01
-4.74237591e-01 6.62561506e-02 -9.42871332e-01 5.42839393e-02
-1.53012979e+00 -6.29241407e-01 -1.32143319e-01 2.09786868e+00
6.55924261e-01 1.76980361e-01 -3.73352952e-02 8.05915818e-02
2.31966674e-01 -4.86765444e-01 -6.30576909e-01 -3.34929287e-01
7.15408087e-01 -9.94348377e-02 9.11460280e-01 3.63257498e-01
-7.05279410e-01 8.84973228e-01 6.91203403e+00 2.90286422e-01
-1.00842547e+00 -4.53803122e-01 -6.45149291e-01 -2.60581344e-01
3.96648467e-01 -1.95171192e-01 -8.17444384e-01 1.84863731e-01
3.79679382e-01 -9.35104564e-02 6.63814902e-01 1.22769845e+00
-1.12533560e-02 -5.47876596e-01 -1.78586817e+00 4.87782151e-01
-3.94041091e-01 -8.57739687e-01 -2.34801129e-01 -8.04004893e-02
3.16261470e-01 -2.32499205e-02 -2.25839745e-02 5.96799612e-01
5.40870130e-01 -1.20501554e+00 1.17541182e+00 4.60716903e-01
8.81575823e-01 -2.83267200e-01 4.16369021e-01 6.33740544e-01
-1.07959378e+00 -4.38550651e-01 -3.63418579e-01 -4.63775307e-01
2.47307941e-01 1.28001720e-01 -1.22924948e+00 6.77703321e-01
7.31818855e-01 2.55988501e-02 7.02874437e-02 5.35697937e-01
-1.15561768e-01 -3.03260535e-01 -5.90609074e-01 -3.32090825e-01
1.68780670e-01 -3.09169948e-01 8.16751778e-01 7.01477349e-01
3.47209573e-01 1.38806164e-01 3.37421387e-01 7.27877319e-01
1.17536388e-01 -1.60369679e-01 -7.66069829e-01 -4.81670834e-02
7.51360118e-01 1.06365347e+00 -6.54872656e-01 6.76251873e-02
2.84908473e-01 6.66533351e-01 2.09861606e-01 -1.08587526e-01
-7.16211259e-01 -7.44705737e-01 8.06244135e-01 1.09768927e-01
2.85469413e-01 -1.03440773e+00 -3.86322767e-01 -5.35329640e-01
4.87913638e-01 -1.01778889e+00 -3.69540542e-01 -1.23149800e+00
-8.83472979e-01 9.33152586e-02 4.13623601e-01 -1.19708145e+00
-4.65091109e-01 -1.14058077e+00 -1.31873026e-01 5.19358933e-01
-9.92879033e-01 -8.35369110e-01 -7.27697730e-01 1.40371561e-01
3.96878392e-01 2.13947659e-03 8.51802051e-01 -3.43597084e-01
-1.59087889e-02 9.00142044e-02 2.07303390e-02 -3.35668147e-01
7.09633827e-01 -1.03542268e+00 2.49299809e-01 1.00859217e-01
-8.90055597e-01 8.13765764e-01 8.28070164e-01 -7.47761726e-01
-2.60987353e+00 -7.61667669e-01 -2.82877833e-01 -1.03559041e+00
8.13163221e-01 -4.05740082e-01 -5.52774549e-01 9.65674460e-01
-4.61280376e-01 -2.68644571e-01 -3.25994819e-01 2.62833070e-02
-1.55980289e-01 -1.47896677e-01 -1.26183784e+00 7.81458378e-01
1.34830391e+00 5.08076437e-02 -1.06844032e+00 5.37333727e-01
8.01458836e-01 -1.10057390e+00 -1.29411137e+00 6.65015697e-01
1.19028270e+00 -4.91468847e-01 9.90314901e-01 -4.47070777e-01
7.10014522e-01 -5.00334680e-01 -4.99932230e-01 -1.22255349e+00
-4.63536084e-01 -7.52573788e-01 1.81446582e-01 7.32476711e-01
1.74958691e-01 -7.38321960e-01 5.29183805e-01 3.80031228e-01
-4.29953337e-01 -8.07205975e-01 -7.92140067e-01 -1.39974487e+00
2.96317130e-01 -2.11974189e-01 4.88086194e-01 4.14490670e-01
4.93191838e-01 1.41955703e-01 7.41654262e-02 3.58186960e-01
1.24165319e-01 2.90150456e-02 1.47361720e+00 -1.10029769e+00
-1.11480184e-01 -5.88122308e-01 -5.19929051e-01 -9.01321769e-01
-1.22875988e-01 -7.33336031e-01 8.92320514e-01 -1.52836645e+00
-1.82014361e-01 -9.63163972e-01 3.91191065e-01 5.24377167e-01
3.38966697e-01 -1.24723390e-01 4.67198104e-01 4.91848171e-01
-3.80054742e-01 3.46144021e-01 1.48658597e+00 1.39675900e-01
-4.85453069e-01 -2.13061005e-01 -4.01999235e-01 5.35622358e-01
8.91124249e-01 -2.71371633e-01 -2.55978972e-01 -4.50120896e-01
3.27969044e-01 2.18756795e-01 3.37161034e-01 -1.29695177e+00
3.00027609e-01 -3.63796353e-01 4.85491902e-02 -2.85848290e-01
4.34086055e-01 -1.18457460e+00 3.52531552e-01 8.09767008e-01
-1.11542098e-01 3.49948347e-01 1.78208172e-01 3.76493752e-01
3.14923137e-01 -1.21483602e-01 6.46813989e-01 -2.74775207e-01
-3.45338613e-01 -1.85289115e-01 -4.61850137e-01 -2.31687799e-01
1.44792247e+00 -2.44177803e-01 -4.85584199e-01 9.58890393e-02
-5.03547788e-01 4.65137571e-01 9.68315244e-01 6.21979058e-01
3.39417398e-01 -1.09061301e+00 -3.29858124e-01 -2.41057888e-01
-6.61980733e-02 5.02678931e-01 -3.95393550e-01 9.15115535e-01
-1.02927399e+00 3.96988720e-01 -6.67545915e-01 -9.43835437e-01
-9.74645793e-01 6.44633651e-01 2.02838540e-01 3.83853093e-02
-8.25822532e-01 4.13547128e-01 -1.81930974e-01 -7.36408770e-01
1.10332489e-01 -7.62378454e-01 3.10423106e-01 -4.30571854e-01
2.96686124e-03 7.23506093e-01 2.10276619e-02 -1.39071062e-01
-4.38861996e-01 6.23448372e-01 1.14274919e-01 -1.83951482e-01
1.41697383e+00 1.45997837e-01 -2.76882440e-01 4.22450304e-01
7.10745096e-01 1.50216203e-02 -1.09645367e+00 5.15592337e-01
7.92373791e-02 -3.81488532e-01 -7.22041786e-01 -7.83453166e-01
-3.38758767e-01 4.23820883e-01 1.27295060e-02 -1.26608815e-02
3.41575325e-01 -5.15535027e-02 4.54581916e-01 1.33309281e+00
1.30968976e+00 -1.01402390e+00 3.19798201e-01 9.23267245e-01
1.86575711e+00 -9.15899754e-01 3.90606999e-01 -8.07700157e-01
-3.73288006e-01 1.38809621e+00 6.38117194e-01 -4.88238633e-01
1.21447459e-01 7.17217326e-01 -1.45284101e-01 -3.85528564e-01
-5.04528403e-01 2.19057173e-01 -9.48088542e-02 6.88326418e-01
-9.86972451e-02 2.33057931e-01 3.05485372e-02 4.10584897e-01
-6.56972051e-01 3.21762972e-02 6.26522660e-01 1.77959991e+00
-8.54492068e-01 -7.51253843e-01 -7.11190760e-01 4.77482527e-01
6.91922382e-02 5.48715472e-01 -3.14139158e-01 1.33276594e+00
-3.64902437e-01 8.45932126e-01 -1.71140149e-01 -5.25637090e-01
1.06037271e+00 -1.68481275e-01 1.10666049e+00 -7.80845881e-01
-4.02127922e-01 -5.59378862e-01 1.73637018e-01 -1.23127794e+00
4.90662232e-02 -6.29275501e-01 -1.52578449e+00 -1.77255303e-01
-3.57590377e-01 -2.51495898e-01 9.21123505e-01 6.56249464e-01
4.31445003e-01 5.56082904e-01 3.20684344e-01 -1.90643680e+00
-1.37395382e+00 -1.07663870e+00 -6.37662530e-01 1.29753858e-01
1.20818049e-01 -1.55286860e+00 -2.00857297e-01 -2.58421928e-01] | [4.7920002937316895, 0.632317841053009] |
e296c4a6-a615-481c-8815-787ba63a8fbc | bioblp-a-modular-framework-for-learning-on | 2306.03606 | null | https://arxiv.org/abs/2306.03606v1 | https://arxiv.org/pdf/2306.03606v1.pdf | BioBLP: A Modular Framework for Learning on Multimodal Biomedical Knowledge Graphs | Knowledge graphs (KGs) are an important tool for representing complex relationships between entities in the biomedical domain. Several methods have been proposed for learning embeddings that can be used to predict new links in such graphs. Some methods ignore valuable attribute data associated with entities in biomedical KGs, such as protein sequences, or molecular graphs. Other works incorporate such data, but assume that entities can be represented with the same data modality. This is not always the case for biomedical KGs, where entities exhibit heterogeneous modalities that are central to their representation in the subject domain. We propose a modular framework for learning embeddings in KGs with entity attributes, that allows encoding attribute data of different modalities while also supporting entities with missing attributes. We additionally propose an efficient pretraining strategy for reducing the required training runtime. We train models using a biomedical KG containing approximately 2 million triples, and evaluate the performance of the resulting entity embeddings on the tasks of link prediction, and drug-protein interaction prediction, comparing against methods that do not take attribute data into account. In the standard link prediction evaluation, the proposed method results in competitive, yet lower performance than baselines that do not use attribute data. When evaluated in the task of drug-protein interaction prediction, the method compares favorably with the baselines. We find settings involving low degree entities, which make up for a substantial amount of the set of entities in the KG, where our method outperforms the baselines. Our proposed pretraining strategy yields significantly higher performance while reducing the required training runtime. Our implementation is available at https://github.com/elsevier-AI-Lab/BioBLP . | ['Paul Groth', 'Michael Cochez', 'Thom Pijnenburg', 'Payal Mitra', 'Dimitrios Alivanistos', 'Daniel Daza'] | 2023-06-06 | null | null | null | null | ['link-prediction', 'knowledge-graphs', 'entity-embeddings'] | ['graphs', 'knowledge-base', 'methodology'] | [-1.47783440e-02 5.54877460e-01 -5.26652098e-01 -3.56747985e-01
-3.54223311e-01 -3.82306576e-01 3.47482324e-01 1.07294381e+00
-5.15863538e-01 9.84529376e-01 2.95601130e-01 -4.01264936e-01
-1.78936958e-01 -1.10529363e+00 -9.46854830e-01 -5.10947287e-01
-2.55776614e-01 8.21450412e-01 2.23857969e-01 -1.44937396e-01
-2.66560823e-01 3.49701732e-01 -1.19238472e+00 1.99586958e-01
1.03503740e+00 7.40815103e-01 -1.22373044e-01 3.42020363e-01
-2.53828257e-01 5.87296247e-01 -3.22023630e-01 -8.63921762e-01
-1.03077747e-01 1.89970508e-02 -9.61184502e-01 -4.26378667e-01
4.46622252e-01 1.75223321e-01 -4.87739652e-01 9.13875282e-01
7.17796385e-01 -2.08756268e-01 8.86829078e-01 -1.34859705e+00
-9.05985951e-01 6.94522858e-01 -2.47938856e-01 -1.51157351e-02
5.23402452e-01 -4.09424186e-01 1.32361400e+00 -9.22205746e-01
1.02684736e+00 1.03035402e+00 8.41750264e-01 5.42333007e-01
-1.33563650e+00 -4.46312964e-01 -5.38095050e-02 4.99988735e-01
-1.44071853e+00 -2.33841062e-01 4.15112972e-01 -5.25495410e-01
1.09027290e+00 8.37067980e-03 5.28156519e-01 9.87337351e-01
1.63594797e-01 6.80968523e-01 5.65594316e-01 -3.88796419e-01
1.07915878e-01 3.93108219e-01 4.28676963e-01 9.30584311e-01
8.22024286e-01 -2.96275496e-01 -5.60645938e-01 -5.97284377e-01
1.00679964e-01 8.03837180e-02 -4.73748654e-01 -7.28111923e-01
-1.32777417e+00 8.93230140e-01 5.50312996e-01 3.00665885e-01
-3.27202380e-01 -1.45580713e-03 6.06042087e-01 3.10081154e-01
6.03184044e-01 5.85947871e-01 -7.84178972e-01 3.16706181e-01
-4.79223907e-01 1.27600044e-01 1.01616967e+00 1.12937534e+00
7.11011827e-01 -4.03879315e-01 2.23582238e-02 7.51278579e-01
2.37618566e-01 4.45233621e-02 3.69374871e-01 -4.08413172e-01
4.93841976e-01 9.50577497e-01 5.76817989e-03 -1.00154293e+00
-5.42314768e-01 -3.22729051e-01 -7.89202273e-01 -3.07618529e-01
3.37640166e-01 -1.01562642e-01 -8.71029377e-01 1.84363389e+00
7.97295332e-01 3.81624430e-01 4.48146820e-01 3.90469998e-01
1.31677175e+00 3.53060186e-01 4.01808292e-01 -8.65060091e-02
1.56858385e+00 -7.05295742e-01 -9.42257047e-01 2.57875800e-01
1.25424218e+00 -4.45185721e-01 5.63740551e-01 -8.49398449e-02
-7.42719114e-01 -1.87820747e-01 -1.00682342e+00 -3.00095588e-01
-1.11077821e+00 7.50699192e-02 1.02867329e+00 5.35546303e-01
-9.35389996e-01 8.66965711e-01 -7.24742413e-01 -5.69833457e-01
6.32022202e-01 7.00792730e-01 -8.68702054e-01 -1.59069836e-01
-1.71390033e+00 1.13929188e+00 7.30901539e-01 -2.16706142e-01
-2.62740076e-01 -1.10901237e+00 -1.17174780e+00 1.47986457e-01
1.93935096e-01 -1.02351511e+00 5.15898943e-01 -3.07931453e-01
-7.28879511e-01 8.08640838e-01 -1.08521849e-01 -5.01156807e-01
2.09483922e-01 -2.62836535e-02 -6.14714563e-01 8.30048881e-03
-3.32940370e-02 7.07841814e-01 6.48303479e-02 -9.30917323e-01
-3.71689945e-01 -3.76014143e-01 2.79322475e-01 7.49382749e-02
-6.88532650e-01 -3.21309298e-01 -3.90895188e-01 -3.35997909e-01
-2.13806286e-01 -9.02200460e-01 -1.84179351e-01 2.70646587e-02
-6.27610981e-01 -5.21953046e-01 6.15477502e-01 -4.57928210e-01
1.19528031e+00 -1.92806876e+00 2.78369099e-01 2.11545452e-01
3.99026364e-01 3.81726861e-01 -8.59516859e-02 7.55061567e-01
-3.99685413e-01 3.13132048e-01 -1.73282012e-01 -2.60804474e-01
6.03094921e-02 4.13842589e-01 6.84395656e-02 3.07885438e-01
2.63782233e-01 1.00632370e+00 -9.45010185e-01 -5.60823798e-01
-2.01294899e-01 6.59025311e-01 -5.63049495e-01 2.10226271e-02
-3.00287426e-01 1.65988505e-01 -5.19433558e-01 8.05705190e-01
5.49556017e-01 -6.32209063e-01 6.63177848e-01 -6.29328728e-01
4.30033416e-01 2.55571455e-01 -1.13777256e+00 1.50812697e+00
-8.50558430e-02 3.22743088e-01 -4.21889782e-01 -1.17730987e+00
6.34569585e-01 5.73723674e-01 7.56245255e-01 -1.08457476e-01
-1.01985954e-01 3.41269612e-01 1.42517313e-01 -5.23531258e-01
3.15525740e-01 -3.11823413e-02 1.05807513e-01 -4.15403815e-03
4.13494766e-01 4.26047266e-01 3.14801157e-01 5.26279569e-01
1.38229692e+00 -7.70688057e-02 6.83493614e-01 -7.64404386e-02
5.22562802e-01 5.52708283e-02 6.11988783e-01 3.02474320e-01
1.08353868e-01 -8.16455297e-03 7.60362089e-01 -5.25777578e-01
-9.51669455e-01 -8.27959836e-01 -5.26017189e-01 7.63869524e-01
2.20736384e-01 -6.70954347e-01 -3.24033052e-01 -9.89271939e-01
5.44914663e-01 3.11936229e-01 -8.62834632e-01 -1.71345443e-01
-2.59835333e-01 -1.02323854e+00 4.97635990e-01 5.56678236e-01
-1.40779525e-01 -8.04147124e-01 1.41570866e-01 2.23608673e-01
-9.37593263e-03 -1.24149525e+00 -1.36152178e-01 3.21739286e-01
-9.35657620e-01 -1.50711846e+00 -4.72766966e-01 -8.81806374e-01
9.54032898e-01 -4.24245149e-01 1.13567376e+00 1.92058697e-01
-4.20937151e-01 4.31067824e-01 -3.37665498e-01 -4.69434410e-01
-3.14791739e-01 2.58393794e-01 1.95669383e-01 -2.07740963e-01
6.95169628e-01 -5.44533074e-01 -6.00040615e-01 7.84016028e-02
-1.00147915e+00 -8.11536014e-02 5.45217752e-01 1.19660521e+00
6.76188648e-01 -2.98243135e-01 8.53116870e-01 -1.71112287e+00
4.93447334e-01 -8.78278434e-01 -1.84781536e-01 5.35789132e-01
-9.26517785e-01 3.26488227e-01 5.66670954e-01 -3.71620506e-01
-4.50148970e-01 1.45497367e-01 -1.77019835e-01 -2.03268960e-01
-7.59447590e-02 9.75691080e-01 -2.66961306e-01 -2.10938290e-01
6.09173656e-01 -2.37682179e-01 -1.78113788e-01 -4.85666335e-01
3.55336368e-01 5.51335454e-01 1.39790541e-02 -5.46774268e-01
5.96453667e-01 1.23738289e-01 3.05478394e-01 -4.12059635e-01
-5.27354002e-01 -5.55809855e-01 -6.73732638e-01 3.64913583e-01
6.83334053e-01 -1.00360811e+00 -6.10400021e-01 -1.13940366e-01
-1.10532451e+00 1.77412406e-01 -2.27322131e-01 7.67945051e-01
-3.01763117e-01 4.45560038e-01 -7.53979921e-01 -1.90015137e-01
-2.80689865e-01 -9.59122837e-01 8.06579411e-01 -1.19122297e-01
-1.99728623e-01 -1.35968292e+00 1.80494055e-01 2.30600700e-01
4.84459102e-02 4.92046773e-01 1.55890632e+00 -1.35218227e+00
-4.09805149e-01 -3.28757912e-01 -4.27702479e-02 9.93773434e-03
3.89822423e-01 -1.65234968e-01 -7.55043209e-01 -2.45328575e-01
-8.81124735e-01 -2.36711800e-01 9.01169360e-01 9.65889394e-02
9.53159988e-01 -2.34116629e-01 -9.93881345e-01 4.86119986e-01
1.54165244e+00 4.28041629e-02 6.13344073e-01 2.41076931e-01
8.90143156e-01 6.26675010e-01 5.83967924e-01 2.48260126e-01
5.83433867e-01 7.69837618e-01 3.75213534e-01 -3.43147635e-01
2.91886888e-02 -2.76229471e-01 1.54604256e-01 8.69997323e-01
-1.06485240e-01 -4.54329133e-01 -9.05224323e-01 8.99141073e-01
-1.90822577e+00 -7.19388843e-01 -2.92598039e-01 2.04260564e+00
1.32365525e+00 -2.40491539e-01 2.82114446e-02 4.94929068e-02
6.49388552e-01 -2.86392480e-01 -5.81026316e-01 -2.79652238e-01
-1.97799178e-03 4.19266999e-01 5.97845316e-01 5.45269489e-01
-1.14624453e+00 7.95537829e-01 5.75909710e+00 5.17834008e-01
-7.14552999e-01 1.19298577e-01 2.32468769e-01 1.73997968e-01
-5.15531123e-01 -1.11153170e-01 -9.43861783e-01 4.37926739e-01
1.13680732e+00 -3.75937730e-01 -1.77681252e-01 8.08751822e-01
-2.83313453e-01 3.88056397e-01 -1.52234828e+00 6.82140231e-01
-4.50129993e-02 -1.55856276e+00 3.07046473e-02 1.98728606e-01
5.36303580e-01 5.41168731e-04 -1.04079790e-01 3.15070540e-01
4.20199513e-01 -1.02749538e+00 -1.38384849e-01 4.81932759e-01
7.75845528e-01 -5.27074277e-01 1.12435055e+00 -1.05975583e-01
-1.08523750e+00 2.64400452e-01 -5.93041658e-01 5.11579692e-01
-5.36654852e-02 9.02357876e-01 -1.27918875e+00 9.73935544e-01
4.02052730e-01 1.02361929e+00 -6.65595889e-01 1.05989563e+00
-1.99940473e-01 3.67858946e-01 -1.99465781e-01 5.77483624e-02
-1.84107855e-01 7.49574602e-02 2.32016444e-01 1.32226801e+00
3.59622777e-01 5.70240095e-02 1.46631077e-01 4.38533932e-01
-5.72322965e-01 5.36039650e-01 -9.53318179e-01 -2.93804348e-01
6.06296837e-01 1.08340764e+00 -3.50180984e-01 -4.98827577e-01
-6.87544584e-01 7.55477190e-01 5.24028063e-01 2.81187326e-01
-8.02758455e-01 -5.06181598e-01 7.38353610e-01 2.68163323e-01
3.37331623e-01 1.68443978e-01 1.51856080e-01 -1.25266433e+00
-3.85627300e-02 -6.72086179e-01 7.75539339e-01 -4.21547592e-01
-1.57765234e+00 5.60858190e-01 -3.42875645e-02 -1.21098292e+00
-1.68191612e-01 -9.62655485e-01 7.55599514e-02 7.57869542e-01
-1.81313622e+00 -1.27981627e+00 2.12250780e-02 5.38724661e-01
-2.37245470e-01 -1.40583217e-01 1.30000663e+00 7.78038561e-01
-4.34588701e-01 8.16388309e-01 2.92256355e-01 2.21193686e-01
1.01479280e+00 -1.46731091e+00 1.80656388e-01 1.61355123e-01
2.14110345e-01 9.71205533e-01 5.50908744e-01 -7.28617430e-01
-1.34058070e+00 -1.25757110e+00 1.31205392e+00 -6.26049340e-01
7.43161917e-01 -3.71108502e-01 -1.15293026e+00 8.54213357e-01
1.02501303e-01 4.21417713e-01 1.42183578e+00 6.31740570e-01
-5.31101882e-01 9.18629318e-02 -1.25840986e+00 4.50877130e-01
1.14188612e+00 -4.70929027e-01 -5.57174206e-01 7.18959630e-01
6.97792530e-01 -3.50770384e-01 -1.62794399e+00 5.17560840e-01
3.93873811e-01 -1.90102413e-01 1.12784243e+00 -1.19110167e+00
4.12730098e-01 -4.07670736e-01 -4.69687171e-02 -1.44158268e+00
-3.08057219e-01 -7.41658881e-02 -4.54208523e-01 1.05327499e+00
9.01793778e-01 -8.68794143e-01 7.75700092e-01 5.97354054e-01
-1.37373835e-01 -1.11524463e+00 -8.61063123e-01 -6.49350524e-01
4.83772941e-02 1.67440221e-01 6.00844145e-01 1.54534507e+00
4.27472919e-01 3.99139971e-01 -2.23913893e-01 3.80738497e-01
4.50639129e-01 4.54995297e-02 5.25765836e-01 -1.63519943e+00
-2.26724595e-01 5.61356768e-02 -1.15044427e+00 -3.93318355e-01
4.10694540e-01 -1.39985561e+00 -4.85252708e-01 -1.87230873e+00
2.91242748e-01 -5.81013441e-01 -5.97583592e-01 9.82680321e-01
-3.42561454e-01 1.14860512e-01 -2.50917763e-01 5.36178276e-02
-5.18365264e-01 3.92026722e-01 1.01749516e+00 -4.21032310e-01
7.75381848e-02 -3.82639170e-01 -6.77659392e-01 5.23405612e-01
6.53680086e-01 -6.67570591e-01 -4.83177274e-01 -2.17246592e-01
4.05762821e-01 -1.11234918e-01 8.67790058e-02 -7.16132581e-01
4.17950273e-01 4.81747761e-02 3.99206102e-01 -2.87494510e-01
4.05156106e-01 -1.01380420e+00 5.79308748e-01 4.25082594e-01
-3.81371707e-01 -1.82659347e-02 2.34649986e-01 8.41384888e-01
-2.86244541e-01 -2.90471435e-01 4.96477216e-01 1.20447099e-01
-5.46025693e-01 6.00383818e-01 1.23497277e-01 2.13282481e-02
1.15326750e+00 -1.04520403e-01 -5.65017700e-01 4.87462543e-02
-1.27186704e+00 3.01548809e-01 4.69306380e-01 1.84961557e-01
6.15765333e-01 -1.53117740e+00 -6.02401197e-01 -1.53054416e-01
5.07880867e-01 -2.62449294e-01 3.03309895e-02 1.02770638e+00
-3.93732011e-01 4.13034379e-01 -6.28783554e-02 -2.73799777e-01
-1.56646502e+00 7.44293392e-01 2.01097250e-01 -4.91542727e-01
-4.90413487e-01 7.41874158e-01 1.22827381e-01 -6.60193086e-01
1.16265573e-01 -1.75917745e-01 -4.79181886e-01 4.04170096e-01
1.86714947e-01 2.37135794e-02 2.75703371e-01 -4.78870213e-01
-6.56572223e-01 3.97831291e-01 -4.26688671e-01 5.45041919e-01
1.59594798e+00 4.09483284e-01 -2.37696409e-01 3.20724696e-01
1.31636357e+00 1.66988775e-01 -2.72058666e-01 -3.41017604e-01
2.41315052e-01 -2.73075342e-01 -2.68144250e-01 -8.25796723e-01
-1.02513158e+00 7.06620872e-01 5.23030341e-01 1.36777014e-02
6.51544511e-01 1.70083225e-01 5.94594240e-01 5.17058134e-01
4.53270882e-01 -7.54548490e-01 -2.96733052e-01 2.76898444e-01
5.26999414e-01 -1.27237833e+00 3.39503020e-01 -8.25377464e-01
-5.71594834e-01 9.99444544e-01 4.72905517e-01 1.91587761e-01
7.84819067e-01 1.35183990e-01 -2.10923404e-01 -5.41642845e-01
-8.89658093e-01 -2.64012933e-01 4.17193621e-01 7.27916300e-01
8.43384147e-01 1.51895896e-01 -6.58383012e-01 5.72917461e-01
2.27794886e-01 8.85002315e-02 3.67537111e-01 8.34221065e-01
-6.93428218e-02 -1.71642244e+00 1.54875547e-01 7.80604780e-01
-6.57929957e-01 -4.44470674e-01 -4.63321626e-01 8.16988945e-01
2.89645731e-01 5.14117002e-01 -2.63814211e-01 -3.06207508e-01
4.04931545e-01 4.56613630e-01 4.57254082e-01 -8.77229452e-01
-4.09214765e-01 -4.80768353e-01 4.15490299e-01 -3.21452409e-01
-6.39979124e-01 -4.64969814e-01 -1.33024824e+00 -4.68185931e-01
-5.38292646e-01 5.61167479e-01 3.43687862e-01 6.18391812e-01
7.89869428e-01 5.73775470e-01 1.63073957e-01 -8.87843668e-02
-1.07731447e-01 -8.06175530e-01 -6.71228826e-01 5.76742768e-01
1.60246775e-01 -9.49245930e-01 -1.52881697e-01 7.55461007e-02] | [8.493001937866211, 7.8457770347595215] |
f1f160b6-d37f-413d-a233-e6ea1bb2ac55 | effectiveness-of-data-augmentation-and | null | null | https://aclanthology.org/2022.lrec-1.381 | https://aclanthology.org/2022.lrec-1.381.pdf | Effectiveness of Data Augmentation and Pretraining for Improving Neural Headline Generation in Low-Resource Settings | We tackle the problem of neural headline generation in a low-resource setting, where only limited amount of data is available to train a model. We compare the ideal high-resource scenario on English with results obtained on a smaller subset of the same data and also run experiments on two small news corpora covering low-resource languages, Croatian and Estonian. Two options for headline generation in a multilingual low-resource scenario are investigated: a pretrained multilingual encoder-decoder model and a combination of two pretrained language models, one used as an encoder and the other as a decoder, connected with a cross-attention layer that needs to be trained from scratch. The results show that the first approach outperforms the second one by a large margin. We explore several data augmentation and pretraining strategies in order to improve the performance of both models and show that while we can drastically improve the second approach using these strategies, they have little to no effect on the performance of the pretrained encoder-decoder model. Finally, we propose two new measures for evaluating the performance of the models besides the classic ROUGE scores. | ['Elaine Zosa', 'Lidia Pivovarova', 'Syrielle Montariol', 'Matej Martinc'] | null | null | null | null | lrec-2022-6 | ['headline-generation'] | ['natural-language-processing'] | [ 6.52783811e-02 4.79559273e-01 1.09771766e-01 -2.90577263e-01
-1.15128803e+00 -4.02836561e-01 1.03494394e+00 3.95809531e-01
-1.06272268e+00 1.16692114e+00 5.88893712e-01 -3.90409857e-01
3.81328523e-01 -6.88705921e-01 -9.11624730e-01 -5.74138701e-01
3.69911015e-01 8.59953225e-01 3.88709843e-01 -5.56487918e-01
-5.62233739e-02 2.11026222e-01 -1.37640476e+00 2.79117525e-01
9.35836792e-01 6.31064475e-01 5.05153775e-01 5.55093348e-01
-1.80591121e-01 6.64314568e-01 -6.66993678e-01 -6.58586621e-01
3.84229600e-01 -5.40634811e-01 -8.74975383e-01 4.60234806e-02
6.41124398e-02 -4.29394022e-02 -8.98295417e-02 7.78116882e-01
7.32143939e-01 -6.88561872e-02 4.89439368e-01 -5.77836096e-01
-2.13265672e-01 1.09929526e+00 -2.72960573e-01 4.05466966e-02
3.69649917e-01 1.43442361e-03 9.62843776e-01 -6.63832426e-01
8.57961535e-01 1.06091964e+00 5.08472681e-01 2.69870400e-01
-1.12780201e+00 -2.57989764e-01 4.98536006e-02 -9.79693085e-02
-1.21271503e+00 -5.97840846e-01 3.87244254e-01 -3.84414256e-01
1.07641041e+00 -1.54637858e-01 3.45742047e-01 1.02314544e+00
-1.12972938e-01 5.80288231e-01 1.08415210e+00 -8.20085883e-01
3.84685583e-02 7.51341343e-01 -1.95834652e-01 5.07926106e-01
1.44283339e-01 -2.47860160e-02 -2.00560912e-01 -1.06786214e-01
3.76352817e-01 -6.36106789e-01 -3.49147826e-01 -2.43895918e-01
-1.30769980e+00 9.75269437e-01 2.21152961e-01 7.12925375e-01
-5.84987342e-01 -2.13894963e-01 5.88016987e-01 5.69205046e-01
6.75198019e-01 6.77442789e-01 -6.02270126e-01 -1.90357361e-02
-1.16944468e+00 1.75967768e-01 9.96475220e-01 8.06486845e-01
6.99769616e-01 5.04235029e-02 -2.10445732e-01 1.08065975e+00
-1.21492324e-02 3.91084880e-01 8.18546295e-01 -1.12366438e-01
8.33377898e-01 2.99047232e-01 9.29345712e-02 -3.45107555e-01
-3.49958390e-01 -5.28038740e-01 -5.05769432e-01 -9.63353813e-02
7.33280301e-01 -6.91752076e-01 -9.08778489e-01 1.75103474e+00
1.25095204e-01 -1.58052772e-01 3.09832335e-01 6.37890220e-01
4.79259372e-01 1.00789320e+00 -9.62305218e-02 -2.12276787e-01
1.12539077e+00 -1.27382112e+00 -4.27976727e-01 -2.70067662e-01
9.80361640e-01 -1.00740063e+00 1.03534734e+00 3.54810625e-01
-1.35383487e+00 -5.54899573e-01 -1.07652783e+00 -1.13580711e-01
-5.67440391e-01 5.01725435e-01 1.84206530e-01 5.50482512e-01
-1.12699139e+00 5.84519744e-01 -5.22878051e-01 -4.82104778e-01
-1.51768759e-01 2.85183042e-01 -5.07104576e-01 -1.55652955e-01
-1.41978168e+00 1.21086311e+00 7.53071845e-01 -6.07546419e-02
-6.08917773e-01 -3.47024590e-01 -8.53653133e-01 7.13466927e-02
3.21524024e-01 -5.66133559e-01 1.21448338e+00 -1.20483208e+00
-1.73778105e+00 7.53040493e-01 1.89240769e-01 -7.19164789e-01
9.31715786e-01 -2.27183685e-01 -2.25751340e-01 -1.45292521e-01
5.78632839e-02 6.62918746e-01 4.30236071e-01 -1.07664120e+00
-7.26626694e-01 3.47336233e-02 1.23028770e-01 2.59383947e-01
-3.12442452e-01 7.41903856e-02 -5.97841382e-01 -7.18589664e-01
-5.83346605e-01 -9.97076869e-01 -3.61788779e-01 -8.20130944e-01
-4.37093526e-01 -2.11845525e-03 1.67983130e-01 -1.01135123e+00
1.15069151e+00 -1.71193504e+00 2.79999077e-01 2.20144451e-01
-5.90390801e-01 4.95421469e-01 -4.60081249e-01 6.53070211e-01
-9.77698714e-02 -9.25284345e-03 -2.21060872e-01 -4.17411506e-01
-2.04504147e-01 1.20441213e-01 3.61873507e-02 1.67902291e-01
3.25995147e-01 5.77312410e-01 -8.22086453e-01 -2.01176062e-01
-8.39114413e-02 5.54029465e-01 -7.04158604e-01 2.47991130e-01
-3.60196441e-01 4.76377577e-01 -1.01265855e-01 -1.18674703e-01
3.06938678e-01 9.07482207e-02 4.24616009e-01 1.24180011e-01
-2.96070933e-01 7.77244449e-01 -1.15477300e+00 1.60168922e+00
-1.08194494e+00 5.31563282e-01 -1.91638619e-01 -8.82245839e-01
1.00632989e+00 5.78806102e-01 7.86321014e-02 -8.47552896e-01
1.94388270e-01 5.12690902e-01 2.76631415e-01 -3.06970745e-01
7.14509010e-01 -3.05118680e-01 -1.66893750e-01 5.99649787e-01
4.07298177e-01 1.36149511e-01 6.62043333e-01 9.87755656e-02
9.06447828e-01 1.54242992e-01 3.73348445e-01 -4.94673789e-01
7.56733477e-01 -1.02865279e-01 2.75214732e-01 6.51975930e-01
5.89822412e-01 5.91222346e-01 5.85000157e-01 -1.25358388e-01
-1.37950242e+00 -4.47041363e-01 1.13309085e-01 1.17447865e+00
-4.07109380e-01 -5.04140496e-01 -8.94626319e-01 -8.32369745e-01
-3.29502642e-01 9.05464590e-01 -4.46137428e-01 -2.17074025e-02
-8.02192509e-01 -9.34044063e-01 5.46187580e-01 2.94959784e-01
2.65606582e-01 -1.04614329e+00 -4.06967163e-01 3.88860703e-01
-1.62990719e-01 -1.26894736e+00 -3.27137709e-01 3.98241907e-01
-5.82455873e-01 -5.89021802e-01 -1.04717040e+00 -6.43970966e-01
4.76709038e-01 -2.93088913e-01 1.27605569e+00 -5.04545495e-02
2.73517191e-01 -1.22208640e-01 -5.02350926e-01 -3.98352116e-01
-9.47826326e-01 6.91462398e-01 -9.70073789e-02 -9.68265310e-02
9.62875485e-02 -2.34770089e-01 -1.12703927e-01 2.09804112e-03
-9.23382819e-01 1.50968328e-01 9.27184224e-01 9.71043706e-01
1.95151076e-01 -1.47857428e-01 7.70179033e-01 -1.30479372e+00
6.70228243e-01 -5.00702024e-01 -7.07340062e-01 2.60496885e-01
-5.35827219e-01 5.54272294e-01 1.07677650e+00 -3.35382760e-01
-1.06607103e+00 1.57032803e-01 -6.65107310e-01 1.08564094e-01
-1.05791830e-01 7.31880486e-01 -3.85410458e-01 3.89540970e-01
5.11068344e-01 7.60556608e-02 -2.40166083e-01 -9.11689341e-01
3.99627239e-01 6.55561268e-01 3.14534515e-01 -3.68807137e-01
6.21448994e-01 -9.64269713e-02 -4.72261667e-01 -8.33705425e-01
-6.34726048e-01 -3.19467664e-01 -7.87681222e-01 1.81763694e-01
6.74233139e-01 -1.05818892e+00 1.92036361e-01 2.53959626e-01
-1.25048983e+00 -5.57504058e-01 -4.57619250e-01 6.04675710e-01
-4.66425508e-01 1.20550074e-01 -6.12926483e-01 -3.64652127e-01
-2.69189358e-01 -1.32698214e+00 8.88957500e-01 -9.72815305e-02
-7.19606206e-02 -1.24397588e+00 4.04468387e-01 1.27587393e-01
4.03616339e-01 -5.91089241e-02 1.06384289e+00 -1.32331908e+00
-8.35434794e-02 -2.68517792e-01 3.70629737e-03 5.65186739e-01
-9.58831981e-02 -2.88733453e-01 -8.00882995e-01 -3.77190769e-01
-2.72471428e-01 -4.09242153e-01 1.01420200e+00 2.30254680e-02
5.07163227e-01 -3.97154391e-01 -3.26104090e-02 1.65235564e-01
1.57406998e+00 6.97751045e-02 6.34807527e-01 4.39857632e-01
5.71982384e-01 6.70976043e-01 4.02055562e-01 2.51910865e-01
4.46908027e-01 9.16551530e-01 -5.13768606e-02 -3.10400784e-01
-1.66762859e-01 -3.89821231e-01 6.04826391e-01 1.20337617e+00
-1.25568256e-01 -5.68200469e-01 -1.02982271e+00 7.46376455e-01
-1.75817633e+00 -7.35124230e-01 1.17787644e-01 2.50633669e+00
1.02846169e+00 3.58267844e-01 3.90190452e-01 1.17454655e-01
6.28165066e-01 3.25655863e-02 4.68631163e-02 -7.43386507e-01
-1.37727082e-01 3.23491067e-01 5.75669348e-01 7.44582295e-01
-1.01700461e+00 1.02278495e+00 6.37485886e+00 8.27611089e-01
-1.41962075e+00 2.62798786e-01 4.72129852e-01 -2.75737084e-02
-3.81159455e-01 1.00937039e-01 -8.73639047e-01 5.74001551e-01
1.48395896e+00 -1.48172200e-01 2.08082840e-01 7.96069443e-01
1.91009551e-01 -7.72492811e-02 -1.14171278e+00 4.21190232e-01
9.20746699e-02 -1.16313040e+00 1.26418531e-01 1.94146156e-01
7.79983342e-01 3.01901370e-01 -3.97157490e-01 5.24322212e-01
3.42956394e-01 -6.72912657e-01 7.59677112e-01 3.89904320e-01
5.61735988e-01 -8.75604630e-01 1.14620817e+00 4.80949968e-01
-7.48623848e-01 1.87398314e-01 -3.93044710e-01 1.36914000e-01
4.25829679e-01 6.19093835e-01 -9.20916617e-01 7.53295064e-01
4.82968315e-02 3.20298016e-01 -7.33047843e-01 1.09344006e+00
-2.29539812e-01 5.15800357e-01 -2.60558039e-01 -1.59278815e-03
6.49296701e-01 -4.25991490e-02 4.88191128e-01 1.62557614e+00
3.85721684e-01 -5.40851414e-01 1.71717659e-01 3.75918746e-01
-1.72611564e-01 5.91341555e-01 -6.21465862e-01 7.00136200e-02
7.98346251e-02 1.21505451e+00 -4.89036560e-01 -4.84059393e-01
-6.52319312e-01 9.81491864e-01 4.88503337e-01 3.91327739e-02
-6.93537712e-01 -4.74026501e-01 -6.70201844e-03 4.53173131e-01
4.23196346e-01 -1.22756571e-01 2.12470680e-01 -1.26066971e+00
-1.64306343e-01 -1.03303778e+00 2.53741831e-01 -4.56042111e-01
-9.21721578e-01 1.09936738e+00 7.68599939e-03 -9.25889730e-01
-1.05757296e+00 -4.79295313e-01 -4.48036849e-01 1.17175698e+00
-1.68205082e+00 -1.00099313e+00 2.54311353e-01 3.61448914e-01
6.12643182e-01 -3.91003817e-01 7.59408772e-01 5.48685193e-01
-4.96240705e-01 6.31859481e-01 3.33440661e-01 6.17551506e-02
8.29611003e-01 -1.32224095e+00 3.99231911e-01 9.74446595e-01
2.31120601e-01 2.98351377e-01 8.06893468e-01 -4.84727889e-01
-8.14205170e-01 -1.07041109e+00 1.41877437e+00 -4.01495658e-02
6.47109926e-01 -5.10542810e-01 -8.36771905e-01 6.04714751e-01
6.18578196e-01 -4.75346804e-01 6.01220310e-01 3.06889415e-01
-8.21520686e-02 1.44254178e-01 -7.91835010e-01 3.75403970e-01
4.63801295e-01 -1.91979304e-01 -5.99652767e-01 3.04220974e-01
5.56137681e-01 -1.41389728e-01 -8.63595247e-01 2.10809290e-01
4.42911208e-01 -8.16951931e-01 5.50850570e-01 -5.53363144e-01
6.27819717e-01 6.23430125e-02 3.11275776e-02 -1.79410803e+00
-6.31323531e-02 -4.64503199e-01 4.21204597e-01 1.39868343e+00
1.03696883e+00 -5.99409997e-01 3.89381379e-01 1.20828435e-01
-1.06658570e-01 -7.51370370e-01 -7.17703164e-01 -8.48351657e-01
3.99087936e-01 -3.13418098e-02 4.28687155e-01 7.14613855e-01
-7.13302614e-03 9.78820145e-01 -5.57486355e-01 -3.01196426e-01
-7.37107843e-02 -1.39362440e-01 8.84975195e-01 -1.02706420e+00
-5.18751323e-01 -2.04404846e-01 -4.62309867e-02 -9.15195048e-01
3.10835809e-01 -9.94516790e-01 1.65341794e-01 -1.46402824e+00
-3.53934355e-02 -5.58767438e-01 -1.39319152e-01 4.23136532e-01
-1.32181942e-01 1.07206367e-01 3.41020167e-01 1.50686558e-02
-2.50066400e-01 4.61980075e-01 7.79188156e-01 1.52514070e-01
-2.63309866e-01 2.69885100e-02 -6.44625664e-01 6.33062065e-01
6.61900997e-01 -5.44926405e-01 -3.99693936e-01 -6.36086285e-01
2.33110234e-01 1.38820574e-01 -1.13472164e-01 -9.43684638e-01
9.32152495e-02 2.82626957e-01 1.35335475e-01 -1.65160179e-01
9.54541042e-02 -5.21446109e-01 -5.86934835e-02 3.79481345e-01
-5.32289267e-01 3.42755765e-01 2.34437630e-01 2.86562175e-01
-3.67863566e-01 -6.70527816e-01 9.36815977e-01 -2.91459858e-01
-4.48766083e-01 -5.92566468e-02 -4.86057669e-01 9.96125266e-02
7.45371759e-01 1.75174877e-01 -1.31819993e-01 -5.73954105e-01
-7.44443059e-01 4.17532325e-02 4.81622726e-01 6.08338475e-01
-4.24494669e-02 -1.16880631e+00 -1.06882417e+00 2.01321140e-01
4.86765802e-02 -3.82460713e-01 -1.72012106e-01 7.65748203e-01
-5.49193203e-01 7.71094561e-01 -3.14757526e-01 -1.45283133e-01
-1.00909352e+00 5.28958380e-01 2.27800161e-01 -1.06969643e+00
-4.83327836e-01 4.34691697e-01 -3.34422290e-02 -6.44646347e-01
1.31928757e-01 -9.27686766e-02 -4.23444927e-01 3.37144434e-01
4.25768137e-01 9.22821388e-02 5.29635608e-01 -9.10888076e-01
-9.12615880e-02 1.73028156e-01 -3.66047770e-01 -5.54827392e-01
1.40229738e+00 -6.11849986e-02 1.80754110e-01 5.34900069e-01
1.12789690e+00 3.13934863e-01 -8.03408206e-01 -3.06734651e-01
2.45664313e-01 -5.48933633e-02 -1.38368551e-02 -7.91173637e-01
-9.13199842e-01 9.11294460e-01 1.93310916e-01 2.04180360e-01
1.03366125e+00 -1.86685368e-01 6.60128593e-01 3.55110735e-01
2.75212169e-01 -1.23959959e+00 -2.00640738e-01 6.99378908e-01
8.62641573e-01 -1.04321790e+00 -1.75064474e-01 -1.08877838e-01
-9.53519821e-01 9.70418453e-01 3.08456391e-01 -1.79040760e-01
4.46116209e-01 3.87347400e-01 7.67242238e-02 2.98473179e-01
-1.08490658e+00 -5.56415617e-01 3.07953835e-01 2.79401422e-01
8.97656977e-01 -2.78791159e-01 -7.37280965e-01 3.80263954e-01
-5.08995354e-01 -3.16826394e-03 6.87939703e-01 6.23432100e-01
-2.49181971e-01 -1.55355632e+00 -6.38253167e-02 4.24131393e-01
-8.34533989e-01 -3.26906949e-01 -2.40096360e-01 1.11116755e+00
1.98229685e-01 6.55066073e-01 9.20866430e-02 -2.50040561e-01
3.82970363e-01 3.60766888e-01 3.24833870e-01 -7.94864893e-01
-9.24415767e-01 3.70327652e-01 6.20585382e-01 -1.59821197e-01
-3.49297017e-01 -6.94810033e-01 -6.77517951e-01 -9.34856310e-02
-4.55604583e-01 4.19947326e-01 8.67453277e-01 1.04233110e+00
1.25250801e-01 4.77873474e-01 4.39408332e-01 -8.29919994e-01
-5.57452917e-01 -1.21072948e+00 -3.40007514e-01 2.88340062e-01
1.67137444e-01 -1.96856022e-01 -5.42801747e-04 -1.02299929e-03] | [11.487679481506348, 9.947848320007324] |
0d336622-6929-49d2-bb42-0a6fa5fca761 | span-based-joint-entity-and-relation | 1909.07755 | null | https://arxiv.org/abs/1909.07755v4 | https://arxiv.org/pdf/1909.07755v4.pdf | Span-based Joint Entity and Relation Extraction with Transformer Pre-training | We introduce SpERT, an attention model for span-based joint entity and relation extraction. Our key contribution is a light-weight reasoning on BERT embeddings, which features entity recognition and filtering, as well as relation classification with a localized, marker-free context representation. The model is trained using strong within-sentence negative samples, which are efficiently extracted in a single BERT pass. These aspects facilitate a search over all spans in the sentence. In ablation studies, we demonstrate the benefits of pre-training, strong negative sampling and localized context. Our model outperforms prior work by up to 2.6% F1 score on several datasets for joint entity and relation extraction. | ['Markus Eberts', 'Adrian Ulges'] | 2019-09-17 | null | null | null | null | ['joint-entity-and-relation-extraction'] | ['natural-language-processing'] | [-1.39350444e-01 4.30464268e-01 -5.78965127e-01 -4.27954942e-01
-1.18443549e+00 -5.24608254e-01 5.07141590e-01 9.37249362e-01
-8.69031012e-01 8.73877227e-01 4.96417373e-01 -1.60538256e-01
7.15804175e-02 -8.53467107e-01 -5.82412601e-01 1.61419008e-02
-4.37635630e-01 5.01775622e-01 2.55877435e-01 -3.69591117e-01
-2.30752409e-01 4.50487971e-01 -9.30980980e-01 3.29030037e-01
8.19931567e-01 9.54087436e-01 -1.40078962e-01 9.17248666e-01
-5.65294921e-02 7.61255145e-01 -8.01115096e-01 -9.86826658e-01
-1.84341460e-01 2.25816756e-01 -1.24342024e+00 -5.05051434e-01
2.17198864e-01 -2.27915749e-01 -4.29313332e-01 5.88863194e-01
5.47583282e-01 3.14696342e-01 6.57745063e-01 -1.03656733e+00
-8.43716323e-01 7.81846821e-01 -4.20269072e-01 5.41521192e-01
4.85989124e-01 -1.02229320e-01 1.92497373e+00 -1.43090594e+00
8.77780497e-01 9.39951658e-01 8.24903965e-01 3.32555950e-01
-1.44075966e+00 -4.40075427e-01 3.46083969e-01 3.20347637e-01
-1.53414643e+00 -4.19889897e-01 4.40688938e-01 -1.46263197e-01
1.99741781e+00 3.77687752e-01 7.12298274e-01 1.22169864e+00
2.35940903e-01 1.21807897e+00 3.17919344e-01 -6.13608718e-01
7.30702579e-02 -8.65011290e-02 5.71710050e-01 3.17714632e-01
4.33780611e-01 -2.44770478e-02 -7.81516790e-01 -3.43770921e-01
1.42590716e-01 -2.73187786e-01 -1.25203967e-01 2.83538327e-02
-8.21302593e-01 7.85247922e-01 4.28992778e-01 2.73718089e-01
-5.02280891e-01 1.87246546e-01 7.11550176e-01 1.00784555e-01
8.34573925e-01 6.64283395e-01 -1.00839829e+00 -3.78324300e-01
-9.06844437e-01 3.87193263e-01 9.09670949e-01 1.29645097e+00
5.29215097e-01 -4.84957457e-01 -8.66161704e-01 7.30004549e-01
1.79601893e-01 3.43684822e-01 4.61019985e-02 -2.79270649e-01
9.13741946e-01 6.53968155e-01 2.68987805e-01 -7.40524888e-01
-5.38534224e-01 -5.07101059e-01 -3.38596225e-01 -4.88057762e-01
-1.15848809e-01 -5.58303475e-01 -7.40754545e-01 1.72216964e+00
4.68194872e-01 1.22275911e-01 1.24310151e-01 4.12072659e-01
9.95984435e-01 2.32409239e-01 6.16493940e-01 -1.38170704e-01
1.72756326e+00 -1.06499028e+00 -1.08075273e+00 -5.72221637e-01
1.08000529e+00 -5.95638633e-01 8.88091028e-01 2.11758558e-02
-1.10893548e+00 -1.70684025e-01 -1.08347654e+00 -7.07052529e-01
-5.95907032e-01 4.01382387e-01 8.73895764e-01 4.48345274e-01
-6.85096920e-01 6.48315370e-01 -8.31699312e-01 -2.27352783e-01
6.71826482e-01 4.53324378e-01 -5.67956924e-01 1.79539755e-01
-1.62396288e+00 1.16364157e+00 3.78024668e-01 9.76932719e-02
-4.12777364e-01 -8.69382501e-01 -1.19823098e+00 2.70914525e-01
4.86559004e-01 -5.94986916e-01 1.12061405e+00 1.04455791e-01
-8.29320133e-01 8.46071184e-01 -6.62698150e-01 -8.44734311e-01
1.43683925e-01 -1.00759661e+00 -6.66956425e-01 1.34578198e-01
2.71505713e-01 2.96670139e-01 3.50379616e-01 -6.62220836e-01
-4.97012228e-01 -1.05504505e-01 1.05274595e-01 1.04874723e-01
-3.74454528e-01 3.99571270e-01 -6.43140435e-01 -6.53261662e-01
-3.27064753e-01 -3.81669849e-01 -2.78230339e-01 -4.41702545e-01
-6.09313488e-01 -7.53574193e-01 4.60762352e-01 -9.20811296e-01
1.80475152e+00 -1.98557472e+00 -8.66421163e-02 -3.57732996e-02
3.75584006e-01 3.81751955e-01 -2.17046440e-01 8.35860968e-01
-2.25939587e-01 5.96015275e-01 8.65326673e-02 -7.83719540e-01
1.89942703e-01 -2.73864847e-02 -3.82052183e-01 1.61452100e-01
1.18406105e+00 1.52049971e+00 -1.07281113e+00 -7.11738646e-01
-3.09001923e-01 3.00287575e-01 -3.80458593e-01 3.06577444e-01
-1.47980422e-01 -2.77239561e-01 -2.98738152e-01 6.86382949e-01
6.02357090e-01 -2.09364846e-01 3.01208168e-01 -3.71837854e-01
2.78336495e-01 9.95351076e-01 -9.51058626e-01 1.53928113e+00
-3.94221634e-01 7.00029969e-01 -1.73009515e-01 -6.46188855e-01
6.83401108e-01 4.90670323e-01 8.88153836e-02 -4.84229028e-01
7.17104152e-02 7.80123007e-03 -2.63839275e-01 -5.06499588e-01
9.67585087e-01 1.33719668e-01 -4.87341762e-01 2.81658202e-01
4.89781618e-01 5.20369150e-02 7.36013174e-01 7.05403209e-01
1.49810302e+00 -4.25144583e-02 5.54872632e-01 -2.44713724e-02
2.21795261e-01 -2.39913478e-01 7.82225251e-01 6.54541135e-01
-3.71393353e-01 4.55782384e-01 8.48033249e-01 -2.15783194e-01
-7.31399834e-01 -9.95491624e-01 -2.19238654e-01 1.13934577e+00
-1.31126493e-01 -1.14084864e+00 -2.79910028e-01 -1.35391891e+00
1.58640489e-01 7.14483440e-01 -9.69927192e-01 -2.30143115e-01
-7.34017074e-01 -7.22456574e-01 7.28483260e-01 1.03025615e+00
7.23914849e-03 -1.00614130e+00 -2.50927031e-01 3.91729236e-01
-2.01081112e-01 -1.39352918e+00 -3.93502325e-01 7.01771259e-01
-5.13761282e-01 -1.04514623e+00 -1.62864655e-01 -7.57770956e-01
3.78993779e-01 -3.16352844e-01 1.72861230e+00 6.36683106e-02
-4.75480139e-01 2.72492588e-01 -4.99134809e-01 -4.42062050e-01
3.00780684e-01 4.38390076e-01 -3.02571356e-02 -3.88901442e-01
1.01288056e+00 -3.38336587e-01 -5.02506554e-01 -2.66441286e-01
-3.01809162e-01 -4.71571743e-01 5.63526213e-01 1.32692277e+00
5.77431977e-01 -4.90858048e-01 8.33478987e-01 -1.00963509e+00
1.00132442e+00 -5.68423808e-01 -6.23323172e-02 5.37720382e-01
-6.35745466e-01 1.65591557e-02 2.20476970e-01 -2.56934375e-01
-8.37110221e-01 -3.79928350e-01 -3.64770263e-01 2.37279855e-05
2.40249839e-02 6.91548586e-01 -1.59208775e-01 4.59795594e-01
7.95159757e-01 -3.57157171e-01 -6.52679920e-01 -3.53762656e-01
7.62491405e-01 6.02419078e-01 2.62510687e-01 -6.80198550e-01
6.52324736e-01 1.46191493e-01 -2.51566976e-01 -6.71272278e-01
-1.18080747e+00 -6.39531851e-01 -8.35765541e-01 4.94139433e-01
7.76930273e-01 -1.09156251e+00 -5.87698638e-01 -2.93755352e-01
-1.31198108e+00 -2.16201738e-01 -8.34022522e-01 4.41924125e-01
2.90084910e-02 3.48483741e-01 -9.94628549e-01 -1.12350535e+00
-9.11619425e-01 -3.99317622e-01 1.23981464e+00 1.86808869e-01
-9.10847545e-01 -9.09940898e-01 4.85022701e-02 -5.58881797e-02
1.14941172e-01 -5.99732809e-02 5.89197338e-01 -1.19438207e+00
-3.63993287e-01 -6.49630785e-01 -3.80916268e-01 1.35269284e-01
-1.47778124e-01 -4.10185121e-02 -1.09615874e+00 -9.48290750e-02
-4.72546756e-01 -6.49740100e-01 1.06380761e+00 -4.09135334e-02
6.77067518e-01 -1.83173567e-01 -8.38204980e-01 4.41323012e-01
1.13845706e+00 -1.75549269e-01 4.61805046e-01 1.52406350e-01
5.52808404e-01 4.02223796e-01 1.08720827e+00 4.83549774e-01
3.59589368e-01 3.85906845e-01 3.35304551e-02 4.31053117e-02
-6.92508593e-02 -3.14404905e-01 3.44241187e-02 5.09493947e-01
2.44366542e-01 -2.62277335e-01 -8.58495474e-01 1.19913590e+00
-1.76631689e+00 -8.63395214e-01 -3.84911406e-03 1.72628486e+00
1.29838979e+00 3.86142224e-01 -7.91978762e-02 1.80932090e-01
3.02263021e-01 2.50987232e-01 -2.11160317e-01 -5.42943954e-01
-1.69012964e-01 8.90758336e-01 3.31223488e-01 5.92883289e-01
-1.58239663e+00 1.18205369e+00 6.75761557e+00 9.97802734e-01
-4.89205539e-01 1.16935171e-01 3.51543218e-01 -4.19451714e-01
-4.69148099e-01 1.67143151e-01 -1.27378595e+00 1.56270087e-01
1.03952706e+00 -3.00671995e-01 -1.13107815e-01 8.51635158e-01
-2.40951598e-01 -7.62908533e-02 -1.30961215e+00 6.35198116e-01
-1.00145854e-01 -1.42217839e+00 -5.09879708e-01 -2.03699052e-01
4.61352825e-01 1.72848150e-01 -2.74247050e-01 8.75511169e-01
4.82065111e-01 -1.06350076e+00 4.38332319e-01 3.08694959e-01
9.26677048e-01 -8.82034063e-01 9.94594038e-01 4.75050546e-02
-1.51383054e+00 -1.04664043e-02 -1.99708596e-01 5.98675478e-03
4.06675011e-01 9.04423416e-01 -1.02463222e+00 8.88694286e-01
5.22143900e-01 6.90214992e-01 -5.62775612e-01 7.82229722e-01
-7.55544186e-01 8.30724478e-01 -5.19742966e-01 -4.05448765e-01
-1.69894114e-01 2.89106041e-01 4.59544718e-01 1.95240903e+00
-1.87476262e-01 9.03244838e-02 4.27292585e-02 8.31011891e-01
-3.45022887e-01 2.72669047e-01 -4.97135103e-01 -3.23810339e-01
8.40134263e-01 1.39287674e+00 -3.93414676e-01 -5.01148641e-01
-3.82478803e-01 9.38791215e-01 1.12868452e+00 4.99150276e-01
-6.57944739e-01 -9.05906022e-01 6.83645308e-01 -3.82309467e-01
8.54896545e-01 -1.94948047e-01 -5.95939159e-01 -1.29686189e+00
3.12214732e-01 -2.98836678e-01 5.37177682e-01 -1.67509869e-01
-1.49023056e+00 4.80725586e-01 -9.69595984e-02 -8.21224868e-01
-2.89497316e-01 -5.02581060e-01 -7.46364415e-01 1.29579139e+00
-1.52091730e+00 -1.38184118e+00 2.18815476e-01 4.32640910e-02
9.50243324e-02 1.70812130e-01 1.22574651e+00 4.51555490e-01
-8.40057850e-01 1.09524870e+00 -3.55371386e-01 7.23165393e-01
6.29320443e-01 -1.66917324e+00 9.52291250e-01 9.85818803e-01
3.78984720e-01 1.24364591e+00 4.04921174e-01 -8.91746044e-01
-1.05103588e+00 -1.04060137e+00 1.85325670e+00 -7.35185444e-01
8.69434953e-01 -7.34784365e-01 -7.77090251e-01 9.90117311e-01
3.43010128e-01 5.31051040e-01 1.00918961e+00 1.20594573e+00
-4.57846761e-01 8.49553943e-03 -1.09583914e+00 7.11951315e-01
1.18849063e+00 -7.90049851e-01 -9.39875662e-01 1.91670001e-01
1.07672799e+00 -4.33401972e-01 -1.20129609e+00 4.44519699e-01
2.63994634e-01 -3.92517149e-01 9.94565964e-01 -1.05588865e+00
3.81285608e-01 7.99162537e-02 -3.09255365e-02 -1.14018643e+00
-3.31350684e-01 -7.56120801e-01 -1.03294408e+00 1.49934506e+00
9.23994243e-01 -3.34597856e-01 7.20764160e-01 8.99550736e-01
7.13606179e-02 -1.29840684e+00 -1.03055441e+00 -6.97244346e-01
7.26356804e-02 -7.68386364e-01 6.53598726e-01 6.97841227e-01
4.21993703e-01 9.89494801e-01 -2.49734297e-02 1.91119775e-01
2.26023465e-01 1.19239278e-01 5.44466197e-01 -1.07347274e+00
-2.07039192e-01 -1.06878795e-01 -3.21842611e-01 -1.12482524e+00
2.11489499e-01 -8.65561604e-01 -5.86226098e-02 -1.82665241e+00
4.58362363e-02 -2.63670862e-01 -3.22888047e-01 4.87228572e-01
-6.83036447e-01 1.66385453e-02 -1.04612730e-01 -3.51319104e-01
-9.91567373e-01 8.51918876e-01 6.99287832e-01 -1.36227384e-01
-8.48078355e-02 -2.61029273e-01 -7.95463264e-01 2.83395559e-01
6.93346381e-01 -5.56246996e-01 -1.00099809e-01 -2.41435125e-01
5.58458328e-01 -3.52280408e-01 3.44371907e-02 -5.18158972e-01
1.25427932e-01 1.00539602e-01 5.27470291e-01 -9.20755923e-01
5.48147082e-01 -4.70853925e-01 -7.77671099e-01 -4.46729511e-02
-4.34101880e-01 -9.44108814e-02 4.23148364e-01 7.20048010e-01
-2.05420479e-01 -4.20098633e-01 9.51134972e-03 1.94457725e-01
-6.02295160e-01 2.32403457e-01 6.62868395e-02 7.01830506e-01
9.07272100e-01 3.02238584e-01 -2.80536830e-01 3.43459398e-02
-1.03644395e+00 4.62918967e-01 -1.49803698e-01 2.83210784e-01
5.65627158e-01 -1.37997806e+00 -6.40552998e-01 1.13273762e-01
3.67908865e-01 2.76375175e-01 1.80123374e-01 7.99469411e-01
-8.82896557e-02 2.76035547e-01 5.88629484e-01 -1.56037629e-01
-1.32835174e+00 5.78364491e-01 7.69101316e-03 -9.65538502e-01
-5.71921289e-01 1.36185944e+00 -4.33797419e-01 -3.31108391e-01
2.33163342e-01 -5.67727506e-01 -3.53843004e-01 4.25153226e-01
6.85699821e-01 1.07427560e-01 2.92089731e-01 -2.34907642e-01
-7.47282147e-01 4.74724770e-02 -4.68769580e-01 -2.70895004e-01
1.28859329e+00 -6.13989122e-02 -1.32998854e-01 3.80952716e-01
1.26913583e+00 5.01953542e-01 -8.46885860e-01 -2.85844088e-01
6.43158615e-01 -3.11411083e-01 -2.13595435e-01 -8.43072832e-01
-3.22595954e-01 7.39397705e-01 -7.06720538e-03 3.52013379e-01
6.22516155e-01 2.53376245e-01 9.67348695e-01 4.32113886e-01
6.51198551e-02 -1.27729166e+00 -9.79911685e-02 9.14549887e-01
9.11402345e-01 -1.23533547e+00 2.99870610e-01 -5.82079768e-01
-6.50414407e-01 8.55437040e-01 6.44463062e-01 -2.10762307e-01
7.98129082e-01 7.42901027e-01 -3.08976233e-01 -2.75482118e-01
-1.23957920e+00 -5.56143105e-01 5.53669810e-01 7.52496064e-01
9.13631976e-01 1.38896942e-01 -5.33143342e-01 1.31700110e+00
-9.44536701e-02 -1.42237604e-01 -9.26389098e-02 1.12466753e+00
-6.87646121e-02 -1.26103628e+00 3.21926922e-01 6.09360695e-01
-8.65349293e-01 -5.89083195e-01 -4.40490425e-01 7.14271605e-01
-1.77097078e-02 1.00922537e+00 8.13671350e-02 -3.77942204e-01
5.91364801e-01 4.13723260e-01 3.21267754e-01 -1.03243995e+00
-9.03911889e-01 -2.59418756e-01 9.02504563e-01 -5.29809773e-01
6.20734282e-02 -5.68955719e-01 -1.32311761e+00 6.18891083e-02
-7.83808589e-01 3.03104967e-01 5.41678183e-02 9.39032495e-01
8.25989008e-01 6.96145356e-01 1.89255729e-01 -3.90238106e-01
-6.00021064e-01 -1.34445035e+00 -4.62420464e-01 4.31296617e-01
3.57922614e-01 -5.43616235e-01 -1.32883415e-01 -3.67470801e-01] | [9.34901237487793, 8.800207138061523] |
839de106-1e71-4324-abfc-e7ab8934e11f | approximate-information-for-efficient | 2307.01563 | null | https://arxiv.org/abs/2307.01563v1 | https://arxiv.org/pdf/2307.01563v1.pdf | Approximate information for efficient exploration-exploitation strategies | This paper addresses the exploration-exploitation dilemma inherent in decision-making, focusing on multi-armed bandit problems. The problems involve an agent deciding whether to exploit current knowledge for immediate gains or explore new avenues for potential long-term rewards. We here introduce a novel algorithm, approximate information maximization (AIM), which employs an analytical approximation of the entropy gradient to choose which arm to pull at each point in time. AIM matches the performance of Infomax and Thompson sampling while also offering enhanced computational speed, determinism, and tractability. Empirical evaluation of AIM indicates its compliance with the Lai-Robbins asymptotic bound and demonstrates its robustness for a range of priors. Its expression is tunable, which allows for specific optimization in various settings. | ['Jean-Baptiste Masson', 'Christian L. Vestergaard', 'Alex Barbier-Chebbah'] | 2023-07-04 | null | null | null | null | ['thompson-sampling', 'efficient-exploration', 'decision-making'] | ['methodology', 'methodology', 'reasoning'] | [ 8.15532580e-02 2.08126724e-01 -1.28362942e+00 -2.59360671e-01
-1.00864995e+00 -8.06792438e-01 7.36465096e-01 5.54990396e-02
-8.78259301e-01 1.31243610e+00 1.68102235e-01 -8.30156446e-01
-8.87850165e-01 -5.03011048e-01 -3.52858782e-01 -6.82440102e-01
-1.05631940e-01 6.48605943e-01 -3.25182289e-01 2.30076492e-01
5.05028903e-01 5.94482720e-01 -9.30457234e-01 -1.43668130e-01
6.54545784e-01 1.37039661e+00 1.86791435e-01 6.45856619e-01
9.60727632e-02 7.06418335e-01 -3.51514995e-01 -5.82683802e-01
5.60781360e-01 -2.59767085e-01 -8.63462448e-01 1.78000052e-03
-5.08652627e-01 -6.57307923e-01 -1.69189692e-01 9.70624208e-01
2.75206357e-01 5.90906978e-01 6.82828724e-01 -1.02873373e+00
-3.20157617e-01 8.45143676e-01 -8.26482952e-01 6.64105654e-01
2.50889838e-01 2.14227572e-01 1.38313353e+00 -1.64329603e-01
3.43769521e-01 1.24327779e+00 2.50424355e-01 2.25248232e-01
-1.32911170e+00 -3.25542659e-01 3.09550524e-01 -5.91375353e-03
-9.09705102e-01 -4.43307787e-01 6.18772328e-01 -1.60926774e-01
6.49564087e-01 5.54595053e-01 7.97563314e-01 7.80043006e-01
-3.33135203e-03 1.26060772e+00 1.44640446e+00 -5.73019147e-01
7.19435513e-01 1.10835306e-01 2.12238487e-02 3.22603464e-01
5.57685554e-01 6.51380301e-01 -5.77711284e-01 -5.80540776e-01
7.15763569e-01 -8.38704705e-02 -9.75975916e-02 -3.69455159e-01
-8.64861548e-01 1.12234294e+00 1.74497545e-01 -6.94470555e-02
-9.04112935e-01 3.94876748e-01 -1.06268339e-01 3.98550451e-01
4.48800296e-01 7.24977851e-01 -2.89503366e-01 -3.84439558e-01
-9.53380108e-01 5.07181406e-01 8.32617462e-01 5.58830559e-01
5.80725312e-01 -8.81352201e-02 -5.56637585e-01 5.17947495e-01
4.51513380e-01 3.30477536e-01 4.87723947e-01 -1.13829064e+00
7.27816999e-01 8.55723629e-04 8.08527708e-01 -3.35891992e-01
-4.02552605e-01 -6.80946529e-01 -1.95537746e-01 4.56380069e-01
5.06415188e-01 -6.15629673e-01 -6.96377456e-01 1.48131943e+00
4.98089969e-01 -5.90173125e-01 -3.13764028e-02 7.65407383e-01
-3.00180197e-01 5.17422676e-01 5.55315912e-02 -7.44627059e-01
1.13535833e+00 -6.94518030e-01 -7.72709668e-01 -7.34942913e-01
1.85851887e-01 -3.95329833e-01 4.39043075e-01 5.37703872e-01
-1.44311357e+00 2.49279380e-01 -9.50659156e-01 4.57548618e-01
6.79002479e-02 -2.01023608e-01 1.11278987e+00 1.15540767e+00
-6.53700054e-01 6.69817805e-01 -8.71344984e-01 2.08249912e-01
6.78261220e-01 3.70647520e-01 2.91578174e-01 1.74590290e-01
-7.78900206e-01 9.31905389e-01 6.16172254e-01 1.48182556e-01
-8.71799707e-01 -2.89725125e-01 -3.33561152e-01 2.27281958e-01
9.01584089e-01 -5.92553079e-01 1.67468655e+00 -1.05279517e+00
-1.93011904e+00 3.10063899e-01 -1.90511104e-02 -8.82636726e-01
8.31410646e-01 -2.73110837e-01 4.69153896e-02 7.86168799e-02
1.44073218e-01 2.64817744e-01 8.30137253e-01 -7.89292216e-01
-7.34686255e-01 -4.57550198e-01 3.74266207e-01 5.42940795e-01
-8.75590667e-02 5.36301285e-02 5.43901436e-02 -7.77839124e-01
2.04758998e-02 -1.00310016e+00 -5.97049177e-01 -4.30098683e-01
-5.87996364e-01 -7.01948628e-02 -3.17641616e-01 -3.87416244e-01
1.34899831e+00 -1.53463089e+00 -2.00860187e-01 6.17820501e-01
-3.00827295e-01 -1.81379259e-01 8.79561901e-02 5.36774278e-01
2.76957750e-01 2.22395629e-01 1.74888428e-02 2.18227878e-02
3.15107882e-01 -1.15334033e-03 -3.28186631e-01 5.92289150e-01
-2.31974736e-01 9.52555835e-01 -6.63292885e-01 -2.86901355e-01
-5.68902828e-02 -4.58373100e-01 -4.14520532e-01 1.34624308e-02
-3.13415021e-01 5.52729927e-02 -1.07891417e+00 7.99764276e-01
3.41125041e-01 -4.04670984e-01 3.30358535e-01 5.40684760e-01
-2.90027231e-01 3.61954957e-01 -1.21928167e+00 1.21912825e+00
-3.40338379e-01 5.14925718e-01 1.54142871e-01 -1.18110883e+00
4.54695344e-01 3.60619985e-02 2.16002986e-01 -5.01984179e-01
3.01042229e-01 2.44747162e-01 -1.04794033e-01 -2.60292828e-01
3.85999978e-01 -4.33847696e-01 1.03856221e-01 9.55360234e-01
-3.67359430e-01 -1.15661062e-02 2.52704591e-01 -8.92673656e-02
8.08223069e-01 6.06961176e-02 7.55393147e-01 -4.86606687e-01
-2.74269164e-01 -7.60888495e-03 3.10966432e-01 1.45734680e+00
-1.17480345e-01 4.09822166e-02 5.24135232e-01 -3.80806029e-01
-5.87813556e-01 -1.02147007e+00 -2.11136654e-01 1.37767255e+00
9.10608098e-02 2.78754741e-01 -3.02865714e-01 -5.54739594e-01
3.70253563e-01 1.23417139e+00 -8.03665638e-01 4.24010962e-01
-6.39564916e-02 -1.01780593e+00 -9.53070894e-02 2.40711719e-01
4.45672095e-01 -8.37252975e-01 -1.30499589e+00 3.24576527e-01
-4.90761064e-02 -3.65849257e-01 -4.45309460e-01 4.14061934e-01
-9.93075967e-01 -8.17670226e-01 -9.21724975e-01 2.34896034e-01
2.65795499e-01 1.05025545e-01 8.20637226e-01 -5.53888321e-01
1.68024898e-01 4.72346395e-01 -2.07046285e-01 -6.39781415e-01
-1.69528872e-01 1.07633136e-01 -4.78471965e-02 3.32368612e-02
2.65475482e-01 -3.32256764e-01 -8.41672182e-01 1.82027936e-01
-5.87625325e-01 -2.11333141e-01 8.55489910e-01 9.37413156e-01
4.32460219e-01 -1.91849824e-02 6.50089145e-01 -5.22758603e-01
1.05202150e+00 -6.34818196e-01 -1.04291892e+00 3.80984694e-01
-9.61772978e-01 4.77211416e-01 -1.17592737e-01 -4.60856736e-01
-1.49131536e+00 -1.81612983e-01 4.05706376e-01 3.07572484e-02
2.67880648e-01 7.48331189e-01 2.97972828e-01 -7.73834949e-03
6.25623405e-01 2.64205877e-02 7.96979368e-02 -5.11145473e-01
3.85297865e-01 6.43550515e-01 1.33240402e-01 -9.16217446e-01
1.09523542e-01 3.71181190e-01 7.55456015e-02 -4.61674035e-01
-1.15861249e+00 -1.43425614e-01 1.96410358e-01 -8.04921165e-02
2.15820521e-01 -5.59500635e-01 -1.19860947e+00 -2.39154056e-01
-7.27600217e-01 -2.84241438e-01 -6.49016440e-01 9.01631474e-01
-1.02232516e+00 7.67766833e-02 -2.12232590e-01 -1.73632753e+00
-1.81837976e-01 -9.69415307e-01 4.08357352e-01 4.51281071e-01
-2.39845246e-01 -8.06963921e-01 2.87787002e-02 4.61043596e-01
3.14753324e-01 2.07293078e-01 3.98426950e-01 -8.68024468e-01
-8.49296629e-01 -3.49099576e-01 4.04193960e-02 -1.76914185e-01
-6.88456520e-02 -3.84287804e-01 -6.59759402e-01 -5.73397398e-01
1.04967870e-01 -4.49908435e-01 9.34955060e-01 1.07792366e+00
1.01015520e+00 -1.00986898e+00 -4.59501714e-01 4.25466061e-01
1.21535838e+00 5.97258329e-01 1.90431848e-01 1.02057016e+00
-3.80142361e-01 5.19158423e-01 7.60709047e-01 1.09367955e+00
-2.69981455e-02 6.10918701e-01 5.88871598e-01 5.82150996e-01
7.25403368e-01 -2.07269698e-01 -6.41929591e-03 -3.47021997e-01
-1.75914079e-01 -2.74280012e-01 -5.35285354e-01 7.36166179e-01
-1.97418237e+00 -9.62964296e-01 7.53042996e-01 2.49972320e+00
8.45753253e-01 4.91676748e-01 6.12344861e-01 -1.93098798e-01
5.37814915e-01 2.41429821e-01 -1.16246283e+00 -6.59125030e-01
2.97279395e-02 -2.28507034e-02 1.15678859e+00 6.08085930e-01
-8.86525750e-01 4.91056710e-01 7.58945084e+00 1.17846346e+00
-3.36158305e-01 3.88445668e-02 1.27409530e+00 -8.08435321e-01
-3.84962857e-01 1.87939867e-01 -7.94510722e-01 3.44468206e-01
9.44319248e-01 -6.09917581e-01 8.88695061e-01 8.98455799e-01
1.74260631e-01 -6.88120842e-01 -8.42764914e-01 6.36943400e-01
-5.86120427e-01 -1.55229819e+00 -2.91695237e-01 2.43872732e-01
9.32541728e-01 1.31175339e-01 1.52394608e-01 -9.51499268e-02
1.00644755e+00 -8.92288029e-01 8.86063814e-01 5.85284233e-01
5.29453099e-01 -1.05148947e+00 4.04646218e-01 6.09497786e-01
-4.44972217e-01 -8.29285800e-01 -8.78887847e-02 -3.73389274e-01
2.95695543e-01 4.94370729e-01 -7.88871646e-01 3.58221948e-01
3.30141217e-01 -8.50305036e-02 2.04530150e-01 1.18313944e+00
-3.30985427e-01 4.45872486e-01 -8.35164428e-01 -6.83633626e-01
8.58746111e-01 -4.33755845e-01 6.68830097e-01 9.82885838e-01
4.47971284e-01 2.61701316e-01 1.11011378e-02 7.65578747e-01
1.97117761e-01 7.36536011e-02 -2.78490245e-01 -2.84679055e-01
7.41718352e-01 6.97516859e-01 -6.96929157e-01 -1.68093354e-01
-7.81244272e-03 3.31986785e-01 2.52677858e-01 5.57576478e-01
-2.64101416e-01 -2.44108886e-01 5.22558510e-01 -2.20701784e-01
5.26376009e-01 4.11524773e-02 -6.74333811e-01 -7.01988935e-01
1.65826955e-03 -6.66813850e-01 8.74529600e-01 -4.00183350e-01
-8.54311407e-01 9.51324999e-02 5.19857347e-01 -6.95597708e-01
-6.33293986e-01 -3.85061860e-01 -4.68901277e-01 7.60118186e-01
-1.36415672e+00 -5.56934536e-01 6.54049695e-01 1.75211668e-01
4.89982516e-01 -1.47892103e-01 2.56356806e-01 -6.52025878e-01
-4.55552369e-01 4.62777406e-01 7.58483768e-01 -4.88950402e-01
-1.77628219e-01 -1.22296727e+00 6.30673990e-02 5.01318991e-01
4.53197248e-02 4.48887587e-01 1.05289495e+00 -3.97036642e-01
-1.37615860e+00 -2.26442292e-01 3.31442207e-01 -8.17633569e-02
8.70332837e-01 3.95907015e-01 -1.17083460e-01 7.68490732e-01
1.54739663e-01 -5.43809533e-01 5.40721714e-01 3.62430513e-01
-3.04366816e-02 -7.29585290e-02 -1.26616991e+00 7.23592341e-01
6.57512903e-01 -3.64648625e-02 -5.44551909e-01 3.81030113e-01
3.45838189e-01 -1.81725442e-01 -5.45093596e-01 9.84370038e-02
9.51594234e-01 -1.11761534e+00 8.61556411e-01 -7.53112972e-01
1.10718407e-01 3.91192228e-01 -2.46988624e-01 -1.35094500e+00
-4.48327541e-01 -1.59593797e+00 -3.06629777e-01 6.95447385e-01
6.44737780e-01 -9.15049136e-01 1.03739107e+00 1.07928574e+00
5.20288467e-01 -1.04766023e+00 -1.43123555e+00 -9.08986390e-01
7.67188519e-02 -3.73409867e-01 7.70240247e-01 5.85939586e-01
4.02864784e-01 -1.19308040e-01 -5.36620378e-01 -2.15840504e-01
9.73040164e-01 5.64770222e-01 3.65866691e-01 -7.89807320e-01
-7.12056756e-01 -9.08636093e-01 2.60058939e-01 -1.58521116e+00
-1.75703615e-01 -4.00379032e-01 -1.67475894e-01 -1.18343329e+00
2.59478390e-01 -4.55176890e-01 -5.55365741e-01 3.08709204e-01
-7.78579339e-02 -3.51205200e-01 3.22716177e-01 1.89070001e-01
-5.39945483e-01 5.02732873e-01 1.09801722e+00 -2.45119885e-01
-5.87544024e-01 7.88058460e-01 -1.20134711e+00 7.73292542e-01
9.03065205e-01 -4.82905477e-01 -2.53726900e-01 -6.14371933e-02
5.60714602e-01 7.97319233e-01 2.40730960e-02 -3.54966342e-01
1.56452641e-01 -9.19167101e-01 2.70931304e-01 -5.79159021e-01
4.08764035e-01 -5.79691291e-01 2.22968474e-01 5.44095993e-01
-8.01884353e-01 -6.23971857e-02 8.10657740e-02 9.99959707e-01
2.67245382e-01 -7.94014096e-01 6.84963226e-01 -2.96465695e-01
-1.91511586e-01 1.80850193e-01 -5.33028781e-01 6.43116161e-02
1.00662494e+00 -2.17381030e-01 -9.94239748e-02 -8.32304835e-01
-6.88072503e-01 3.96657079e-01 3.11950520e-02 -1.17803194e-01
2.68144071e-01 -1.14938390e+00 -7.23719478e-01 -1.87888384e-01
-2.24763438e-01 -4.08306032e-01 1.69591308e-02 6.02843106e-01
-1.56905770e-01 7.00925708e-01 9.69230011e-02 -5.19231260e-02
-6.67375386e-01 5.69313884e-01 3.24640393e-01 -7.60100007e-01
-1.40124917e-01 8.38725865e-01 -1.16419055e-01 4.49508578e-01
1.41768619e-01 -3.14037129e-02 2.13736221e-02 1.78469464e-01
6.44363582e-01 8.95843327e-01 -3.95543277e-01 1.45664304e-01
-4.37088832e-02 2.67519313e-03 -1.73518047e-01 -7.99142003e-01
1.20621610e+00 -4.51784968e-01 2.00588092e-01 3.51180583e-01
6.08651042e-01 -2.76941925e-01 -1.62208188e+00 -4.82193798e-01
3.89345080e-01 -9.61968780e-01 5.41214406e-01 -1.06516945e+00
-6.81483686e-01 2.38510698e-01 2.97276378e-01 6.99304461e-01
8.31244946e-01 -2.00165078e-01 1.65692508e-01 7.52934694e-01
4.61224169e-01 -1.39233851e+00 -2.73137778e-01 -3.64716575e-02
7.85400569e-01 -1.11267650e+00 5.32551587e-01 4.96345729e-01
-9.03775930e-01 9.05587673e-01 -2.33220920e-01 1.72770694e-01
6.81800663e-01 -1.51375100e-01 -3.86366457e-01 2.39472892e-02
-8.80751431e-01 -2.75667310e-01 6.52230233e-02 2.87276566e-01
-2.56815493e-01 4.69096869e-01 -5.69616079e-01 2.98962593e-01
-2.49621883e-01 3.53777185e-02 3.77032220e-01 9.96916294e-01
-7.12400615e-01 -7.17968166e-01 -5.36132395e-01 9.49675620e-01
-1.00083756e+00 3.79504971e-02 -1.01330347e-01 5.66786349e-01
-7.37439692e-01 1.22992110e+00 -5.95778264e-02 4.80244607e-01
-1.59569368e-01 -1.26093060e-01 5.21375418e-01 2.15136334e-02
-2.17632756e-01 3.45702469e-01 4.68907773e-01 -5.18208027e-01
-5.10474026e-01 -1.03530645e+00 -3.54115039e-01 -2.53965378e-01
-6.61820650e-01 4.51150626e-01 8.36528242e-01 1.14468670e+00
2.81091124e-01 -4.32467600e-03 8.50788772e-01 -7.81840920e-01
-1.50845623e+00 -8.35222542e-01 -9.36305940e-01 -2.47931868e-01
6.37422621e-01 -8.18940639e-01 -4.33267564e-01 -6.41618431e-01] | [4.516361713409424, 3.2281558513641357] |
79744be6-e2c3-40db-9291-834230b23703 | vision-transformer-based-feature-extraction | 2302.00875 | null | https://arxiv.org/abs/2302.00875v1 | https://arxiv.org/pdf/2302.00875v1.pdf | Vision Transformer-based Feature Extraction for Generalized Zero-Shot Learning | Generalized zero-shot learning (GZSL) is a technique to train a deep learning model to identify unseen classes using the image attribute. In this paper, we put forth a new GZSL approach exploiting Vision Transformer (ViT) to maximize the attribute-related information contained in the image feature. In ViT, the entire image region is processed without the degradation of the image resolution and the local image information is preserved in patch features. To fully enjoy these benefits of ViT, we exploit patch features as well as the CLS feature in extracting the attribute-related image feature. In particular, we propose a novel attention-based module, called attribute attention module (AAM), to aggregate the attribute-related information in patch features. In AAM, the correlation between each patch feature and the synthetic image attribute is used as the importance weight for each patch. From extensive experiments on benchmark datasets, we demonstrate that the proposed technique outperforms the state-of-the-art GZSL approaches by a large margin. | ['Byonghyo Shim', 'Junhan Kim', 'Kyuhong Shim', 'Jiseob Kim'] | 2023-02-02 | null | null | null | null | ['generalized-zero-shot-learning', 'generalized-zero-shot-learning'] | ['computer-vision', 'methodology'] | [ 2.64916420e-01 4.60659526e-02 -2.13095769e-02 -5.21404624e-01
-8.42548251e-01 3.09892539e-02 5.49224913e-01 2.00411171e-01
-2.42070511e-01 4.93599206e-01 2.49961153e-01 2.75296539e-01
-1.65479660e-01 -9.15985703e-01 -9.42201793e-01 -9.34480250e-01
5.00652432e-01 -7.52053708e-02 2.62212187e-01 -9.78916660e-02
3.42780322e-01 1.87886029e-01 -1.80882215e+00 2.80645579e-01
1.03110433e+00 1.54091048e+00 4.24580872e-01 5.46382330e-02
-2.70359933e-01 8.69306862e-01 -3.42954308e-01 -1.66656643e-01
2.69786030e-01 -4.05755460e-01 -5.37563205e-01 4.18349624e-01
5.06858587e-01 -4.00650918e-01 -2.91324288e-01 1.18296814e+00
5.75328231e-01 2.69639283e-01 7.00407147e-01 -1.33589363e+00
-8.79639626e-01 1.52816266e-01 -6.40762031e-01 1.00784861e-01
2.27109924e-01 3.43917847e-01 1.08040452e+00 -1.26202476e+00
4.47367966e-01 1.25043535e+00 1.48671106e-01 1.98622510e-01
-1.09580433e+00 -4.70426470e-01 2.92059690e-01 6.66860342e-01
-1.45918584e+00 -3.02881658e-01 1.17791474e+00 -3.26864958e-01
6.74563944e-01 7.72201493e-02 6.12900436e-01 7.64656901e-01
2.83504188e-01 1.21858144e+00 1.00532234e+00 -5.21034896e-01
1.68766275e-01 2.43833631e-01 -3.46512049e-02 6.70536816e-01
-7.63289183e-02 -7.30589405e-02 -6.13196194e-01 6.47861930e-03
6.45363092e-01 3.52806211e-01 -2.86594242e-01 -5.78264713e-01
-1.23644447e+00 8.47544014e-01 8.54573548e-01 -8.98717418e-02
-6.47599161e-01 3.00352015e-02 2.61164904e-01 2.39989161e-01
5.41897893e-01 1.13890544e-01 -2.32266948e-01 4.19595450e-01
-4.58052635e-01 -3.60501409e-02 8.91672745e-02 1.05258334e+00
1.25512767e+00 7.26871639e-02 -7.33782291e-01 8.35936368e-01
2.45638832e-01 3.76034528e-01 5.88513374e-01 -8.88163030e-01
2.30498001e-01 1.00946367e+00 -4.06195410e-02 -1.01671195e+00
1.79710135e-01 -5.15881717e-01 -9.26604629e-01 2.28202999e-01
-1.41103432e-01 1.39457032e-01 -1.00165641e+00 1.70566583e+00
3.92773420e-01 3.33264768e-01 1.16174012e-01 9.17171597e-01
1.00494206e+00 8.55104148e-01 1.85400546e-01 -1.13075718e-01
1.29684889e+00 -1.01331687e+00 -5.63419998e-01 -4.43837464e-01
1.40716419e-01 -7.05412507e-01 1.31767106e+00 1.65756252e-02
-8.28822911e-01 -7.86247134e-01 -1.20524716e+00 -6.04073741e-02
-3.52047920e-01 1.53800532e-01 3.66906703e-01 6.91847652e-02
-6.09092653e-01 4.34431940e-01 -3.24969590e-01 -3.35052639e-01
6.84810460e-01 2.03662157e-01 -3.65325153e-01 -3.34161043e-01
-1.22624576e+00 5.44796646e-01 3.09028178e-01 -6.16301559e-02
-1.07285273e+00 -5.55196881e-01 -1.06808794e+00 3.70081276e-01
4.25942123e-01 -5.32489240e-01 8.83738339e-01 -1.18634546e+00
-1.10945952e+00 7.98562229e-01 -2.88132876e-01 -1.78174049e-01
1.06777713e-01 -1.40560701e-01 -1.39493108e-01 2.70969152e-01
3.52259159e-01 7.29978681e-01 1.16250205e+00 -1.15470326e+00
-6.45733237e-01 -4.52439845e-01 -4.71443310e-02 4.44082677e-01
-5.09192944e-01 -2.74391055e-01 -4.22709942e-01 -7.33695090e-01
-1.39108514e-02 -4.50465262e-01 -1.36982068e-01 2.55663276e-01
-3.67633611e-01 -3.74567121e-01 8.79603386e-01 -2.65895903e-01
8.68849158e-01 -2.48098159e+00 3.45995873e-02 -6.54199347e-02
1.94274232e-01 1.83332577e-01 -2.59413153e-01 3.13066840e-01
5.32819778e-02 -3.19121033e-01 -4.00813341e-01 -1.11100227e-01
-9.80137214e-02 2.41641134e-01 -1.85437113e-01 2.81820625e-01
6.27459347e-01 1.04064965e+00 -8.75140369e-01 -7.37733066e-01
5.21217108e-01 3.89577389e-01 -2.60641664e-01 4.74477530e-01
-1.96990266e-01 3.26834112e-01 -7.24757671e-01 7.07690775e-01
8.22269320e-01 -2.06629321e-01 -6.01930916e-01 -3.58791530e-01
9.19207707e-02 -3.52452993e-01 -8.98080170e-01 1.71235740e+00
-3.65838349e-01 3.51803929e-01 -2.56173283e-01 -9.91297305e-01
1.24756002e+00 6.22170791e-02 3.93244773e-01 -8.17970574e-01
1.75704926e-01 8.25336352e-02 -2.42476046e-01 -6.55627370e-01
5.43979630e-02 6.84683770e-02 -5.92268370e-02 3.29059035e-01
4.27589417e-01 1.60266951e-01 -2.48658627e-01 2.37897113e-01
6.80156648e-01 -2.50797477e-02 4.93675023e-01 -3.71640205e-01
9.09500837e-01 -2.62998641e-01 7.35211194e-01 6.15965068e-01
-3.06827545e-01 6.85527265e-01 3.14722538e-01 -6.74375236e-01
-1.12498927e+00 -1.03791296e+00 -2.68340427e-02 1.02772868e+00
6.02593184e-01 -3.15830633e-02 -8.21848333e-01 -7.08759665e-01
-3.62956412e-02 6.97505414e-01 -8.56613040e-01 -6.68947518e-01
-1.41498387e-01 -4.93700087e-01 -1.32090628e-01 7.04151332e-01
9.84622300e-01 -1.38907719e+00 -4.60729182e-01 2.44629160e-01
-2.33629167e-01 -8.58415604e-01 -6.87049925e-01 -1.92683991e-02
-4.07253146e-01 -8.19414139e-01 -8.25873673e-01 -9.68236804e-01
7.77490556e-01 4.31736916e-01 8.81195486e-01 -9.00401548e-03
-3.52463394e-01 1.70209348e-01 -5.15916765e-01 -5.12736022e-01
1.74419194e-01 -4.96829040e-02 -2.71295518e-01 8.39701235e-01
6.78808987e-01 -4.36525524e-01 -9.52363253e-01 2.30793178e-01
-7.21891224e-01 1.04162181e-02 1.09508359e+00 1.15313256e+00
8.94102097e-01 -1.06139101e-01 7.18008518e-01 -6.32991016e-01
4.44737494e-01 -3.88430595e-01 -3.24458361e-01 3.12750518e-01
-5.74766815e-01 2.90808976e-01 7.21848130e-01 -3.99008006e-01
-1.09466875e+00 2.59108245e-01 4.59551476e-02 -7.36252725e-01
-8.37064311e-02 2.51244098e-01 -7.18281746e-01 -1.07549548e-01
2.32725963e-01 7.80744314e-01 9.32859033e-02 -4.24062759e-01
2.85145760e-01 7.85755098e-01 5.86971521e-01 -4.58007783e-01
5.18745124e-01 4.05443877e-01 3.48906107e-02 -6.17502034e-01
-1.30235589e+00 -4.85774189e-01 -4.35827017e-01 -1.86487064e-01
9.59083676e-01 -8.94947827e-01 -5.33559501e-01 6.36628687e-01
-9.98472631e-01 1.71335429e-01 -5.23026764e-01 4.18135047e-01
-7.10373223e-01 2.06057444e-01 -2.66557753e-01 -7.91959643e-01
-5.62178373e-01 -1.41466570e+00 1.30296433e+00 4.85911220e-01
3.56460571e-01 -4.82392162e-01 -3.58482510e-01 9.74657908e-02
3.79342407e-01 3.29903543e-01 1.03554428e+00 -6.50128841e-01
-8.72325718e-01 -2.31912926e-01 -5.97295403e-01 5.43293238e-01
1.47548571e-01 -2.92994201e-01 -1.19344330e+00 -2.37141848e-01
2.46633589e-01 -4.59376395e-01 1.05843401e+00 3.94505262e-01
1.49609637e+00 -3.49482626e-01 -5.13542406e-02 6.64456427e-01
1.66574967e+00 1.53215647e-01 7.72305071e-01 4.05640990e-01
8.16269040e-01 4.39689666e-01 1.05445004e+00 5.20613790e-01
2.01214164e-01 5.01646996e-01 6.58235848e-01 -1.92105532e-01
-4.43677157e-02 -4.14943248e-01 1.49013400e-01 5.28615892e-01
1.44516885e-01 -1.49711549e-01 -3.90610039e-01 6.77883923e-01
-1.98281622e+00 -8.63157272e-01 1.36996806e-01 2.38953447e+00
4.99448389e-01 7.80511871e-02 -3.32155615e-01 -4.25764509e-02
1.00809979e+00 3.51901829e-01 -7.79339969e-01 -1.97658986e-01
-2.06165761e-01 -3.83621305e-02 1.99616030e-01 2.95584381e-01
-1.23328042e+00 9.20281887e-01 5.31946373e+00 1.24645507e+00
-8.86479676e-01 -1.07146829e-01 9.04219568e-01 1.34008601e-01
-3.73124510e-01 4.71507683e-02 -6.30967200e-01 6.69391096e-01
2.55645722e-01 -3.49755555e-01 2.41986722e-01 1.07858968e+00
-6.60918504e-02 3.59121896e-02 -9.70031321e-01 9.62023973e-01
3.33447516e-01 -1.08475411e+00 4.99286979e-01 4.95068952e-02
5.73784292e-01 -2.76181757e-01 2.74388373e-01 3.66326779e-01
-1.86474964e-01 -8.02872002e-01 4.99736845e-01 7.33824074e-01
8.32571387e-01 -1.02834964e+00 9.30886865e-01 1.38244644e-01
-1.34979188e+00 -2.15771347e-01 -8.75468433e-01 1.08195968e-01
-1.56424031e-01 4.97945696e-01 -3.27123135e-01 5.83445609e-01
7.53775299e-01 9.34879184e-01 -7.14278817e-01 1.02293611e+00
-3.23428005e-01 3.89198303e-01 -2.27697976e-02 8.86689126e-02
4.99423712e-01 -2.17241585e-01 5.33520639e-01 6.36637151e-01
3.70938122e-01 1.64425045e-01 2.08562061e-01 9.55994666e-01
-5.06984368e-02 2.38302156e-01 -6.57712460e-01 2.83347636e-01
6.06698871e-01 1.29463375e+00 -2.02167436e-01 -4.41516459e-01
-5.65754592e-01 1.18010581e+00 2.94463247e-01 3.15305620e-01
-5.43821990e-01 -6.75895035e-01 5.84287226e-01 -1.11642266e-02
5.57114303e-01 4.49783951e-01 2.03388114e-03 -9.38932538e-01
1.73939034e-01 -6.23634219e-01 4.18204010e-01 -1.13769543e+00
-1.43737996e+00 6.68410361e-01 -2.31317773e-01 -1.70692015e+00
-7.15115014e-03 -2.73818284e-01 -9.45091903e-01 9.60535824e-01
-1.65679908e+00 -1.50293446e+00 -7.52165735e-01 8.41012061e-01
7.61806130e-01 -4.02337283e-01 6.40545070e-01 3.09095830e-02
-5.21419704e-01 6.89586401e-01 1.70628071e-01 3.31704095e-02
8.20268333e-01 -1.22910953e+00 3.23332101e-01 7.02366292e-01
-1.50808331e-03 4.19097900e-01 5.19077659e-01 -3.75365496e-01
-1.09154320e+00 -1.36752760e+00 6.25090659e-01 -1.20416179e-01
3.47333521e-01 -2.55767196e-01 -1.10966945e+00 4.92850512e-01
3.70941520e-01 4.70658422e-01 4.21116829e-01 -4.75216925e-01
-3.98001730e-01 -5.37439167e-01 -1.16258657e+00 2.51986891e-01
6.39917910e-01 -4.64884996e-01 -6.64678752e-01 3.16948481e-02
1.11895025e+00 6.36858717e-02 -7.72798061e-01 4.81908560e-01
3.94313037e-01 -9.55699861e-01 9.79049802e-01 -2.96253234e-01
5.86736023e-01 -4.36901689e-01 -3.56014252e-01 -1.46272790e+00
-6.24395192e-01 -1.03522904e-01 1.39917493e-01 1.49106312e+00
1.95188507e-01 -4.72371727e-01 5.38813174e-01 4.16229576e-01
-2.39275292e-01 -1.03356910e+00 -8.01273823e-01 -6.75895691e-01
-2.05208823e-01 1.49350435e-01 9.16873991e-01 7.65240073e-01
-4.42737520e-01 5.28118968e-01 -5.40979564e-01 -7.55025679e-03
9.52497363e-01 3.61516118e-01 4.84070480e-01 -1.28688276e+00
-1.24153122e-01 -1.72388181e-01 -7.33458221e-01 -7.67198682e-01
7.84050599e-02 -6.60455823e-01 1.45698208e-02 -1.59873843e+00
5.73568523e-01 -1.86326116e-01 -7.78981209e-01 4.38322574e-01
-6.55791879e-01 1.18853889e-01 1.56546474e-01 1.89214721e-01
-7.03217983e-01 1.09950876e+00 1.43873727e+00 -5.22376180e-01
1.87981188e-01 -2.09654033e-01 -7.64836669e-01 5.10140121e-01
7.58085549e-01 -3.87349337e-01 -5.69567680e-01 -3.15524399e-01
-4.02594745e-01 -1.63646400e-01 5.31078100e-01 -1.06368399e+00
2.52667874e-01 -2.56825835e-01 6.53209507e-01 -6.42539918e-01
4.21521395e-01 -8.41409206e-01 -3.74839127e-01 2.53455400e-01
-4.32123423e-01 -3.13343406e-01 -7.94366449e-02 7.72031188e-01
-4.58440989e-01 -7.19646737e-02 9.59789693e-01 -2.20668003e-01
-1.20337713e+00 6.73242748e-01 -5.14512956e-02 -5.72073348e-02
1.22966051e+00 -2.33343259e-01 -3.53834212e-01 -3.02587688e-01
-3.84371996e-01 4.42288518e-01 4.93949294e-01 3.65099192e-01
1.02506149e+00 -1.70975327e+00 -7.55744100e-01 5.97421169e-01
8.55573654e-01 1.01935707e-01 5.88130236e-01 5.54779947e-01
-8.35803896e-02 6.32723197e-02 -5.77023923e-01 -5.64791381e-01
-9.58858073e-01 1.06554306e+00 1.38248980e-01 7.36570507e-02
-7.44665742e-01 8.47654998e-01 6.57692492e-01 -6.92284852e-02
7.37983584e-02 1.89159229e-01 -3.56719255e-01 -2.17855245e-01
7.01275885e-01 -2.36299019e-02 -1.23048119e-01 -7.48087823e-01
-3.42624187e-01 8.30196798e-01 -2.11716026e-01 1.43255875e-01
1.19084108e+00 -2.63569146e-01 -1.04945250e-01 3.74326617e-01
1.35826480e+00 -3.37225229e-01 -1.51137996e+00 -6.01199448e-01
-4.13710296e-01 -6.78245783e-01 2.54048645e-01 -5.48007011e-01
-1.23236060e+00 1.13373363e+00 7.48277962e-01 -1.65426120e-01
1.45997190e+00 4.57544476e-02 8.19217145e-01 2.08154887e-01
3.24722797e-01 -9.60019767e-01 3.57860208e-01 9.69997421e-02
8.57698262e-01 -1.55312932e+00 -1.92516848e-01 -3.68331969e-01
-9.20957863e-01 7.37234294e-01 1.00668204e+00 -3.70632172e-01
4.90827531e-01 -2.66830713e-01 -5.77869043e-02 -2.34270170e-01
-6.65348291e-01 -5.02562761e-01 3.72447193e-01 6.74490035e-01
4.12018895e-02 -1.58766001e-01 -1.01107106e-01 6.58144653e-01
3.41295868e-01 -1.76273406e-01 2.39228368e-01 6.87399566e-01
-8.30963850e-01 -8.65987539e-01 -1.89066693e-01 6.20685041e-01
-1.78399637e-01 -1.70220613e-01 -3.99171442e-01 3.70116293e-01
2.70422488e-01 8.64421964e-01 1.52889252e-01 -6.00939691e-01
1.84139997e-01 -1.77715018e-01 1.68137461e-01 -4.26074982e-01
-1.01954162e-01 8.13860670e-02 -4.52846885e-01 -6.07434094e-01
-2.29933068e-01 -2.35590518e-01 -9.35460210e-01 9.77649763e-02
-2.35305384e-01 9.64216962e-02 2.27381900e-01 8.98464680e-01
4.98635083e-01 5.22872210e-01 1.03990459e+00 -6.82837605e-01
-4.29079533e-01 -9.41751778e-01 -7.14004457e-01 5.68649292e-01
3.52330416e-01 -8.21094275e-01 -3.81739944e-01 -9.64328796e-02] | [9.764315605163574, 2.1391022205352783] |
ce759618-a1e0-4478-8ce8-f564c0a1a5e1 | learning-geometric-aware-properties-in-2d | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Arsomngern_Learning_Geometric-Aware_Properties_in_2D_Representation_Using_Lightweight_CAD_Models_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Arsomngern_Learning_Geometric-Aware_Properties_in_2D_Representation_Using_Lightweight_CAD_Models_CVPR_2023_paper.pdf | Learning Geometric-Aware Properties in 2D Representation Using Lightweight CAD Models, or Zero Real 3D Pairs | Cross-modal training using 2D-3D paired datasets, such as those containing multi-view images and 3D scene scans, presents an effective way to enhance 2D scene understanding by introducing geometric and view-invariance priors into 2D features. However, the need for large-scale scene datasets can impede scalability and further improvements. This paper explores an alternative learning method by leveraging a lightweight and publicly available type of 3D data in the form of CAD models. We construct a 3D space with geometric-aware alignment where the similarity in this space reflects the geometric similarity of CAD models based on the Chamfer distance. The acquired geometric-aware properties are then induced into 2D features, which boost performance on downstream tasks more effectively than existing RGB-CAD approaches. Our technique is not limited to paired RGB-CAD datasets. By training exclusively on pseudo pairs generated from CAD-based reconstruction methods, we enhance the performance of SOTA 2D pre-trained models that use ResNet-50 or ViT-B backbones on various 2D understanding tasks. We also achieve comparable results to SOTA methods trained on scene scans on four tasks in NYUv2, SUNRGB-D, indoor ADE20k, and indoor/outdoor COCO, despite using lightweight CAD models or pseudo data. | ['Supasorn Suwajanakorn', 'Sarana Nutanong', 'Pattaramanee Arsomngern'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['scene-understanding'] | ['computer-vision'] | [ 1.83388308e-01 -6.55969977e-02 4.29231673e-02 -8.14988852e-01
-8.61056149e-01 -4.93908972e-01 7.84737229e-01 -2.38141894e-01
-3.45557302e-01 1.42424569e-01 3.16660523e-01 -1.75717488e-01
-1.98836252e-01 -1.07744622e+00 -9.90877807e-01 -1.95274070e-01
2.70147234e-01 7.52540290e-01 2.97041237e-01 -5.22675693e-01
1.67335957e-01 7.92551577e-01 -1.70697832e+00 1.99498713e-01
5.39027929e-01 1.10772264e+00 3.73551458e-01 5.40951431e-01
-2.90630162e-01 6.17070675e-01 5.25677726e-02 -6.86289594e-02
7.55991161e-01 -9.71841365e-02 -6.62535608e-01 1.20567150e-01
1.12565958e+00 -4.86529320e-01 -4.33016241e-01 4.60316658e-01
5.36027193e-01 -2.17851549e-02 3.14151973e-01 -9.37802851e-01
-3.64826053e-01 -2.00163331e-02 -6.91895485e-01 -8.07441697e-02
5.45562565e-01 9.35992077e-02 8.61853600e-01 -1.02003896e+00
7.27194786e-01 1.41059971e+00 8.84317696e-01 2.53331095e-01
-1.19991910e+00 -5.76114714e-01 9.01873931e-02 2.19339937e-01
-1.20110810e+00 -2.86334008e-01 1.15318048e+00 -2.75155008e-01
1.26915169e+00 1.01833142e-01 8.14565480e-01 1.18916249e+00
-6.70732483e-02 7.05235600e-01 1.44027781e+00 -3.76998425e-01
2.42973096e-03 -2.65482455e-01 -5.28484657e-02 5.78639925e-01
-8.58332217e-02 3.56301218e-01 -8.24060977e-01 1.92317247e-01
9.46132243e-01 1.47823364e-01 -2.19445467e-01 -1.01366627e+00
-1.26588917e+00 8.37659955e-01 9.28295493e-01 -2.67286189e-02
-2.72740155e-01 2.19811201e-01 1.16683193e-01 1.93787917e-01
8.00944626e-01 4.81340796e-01 -7.83055007e-01 -7.56030083e-02
-7.71939456e-01 2.78355852e-02 3.94931525e-01 9.82666254e-01
1.13626826e+00 -1.42592698e-01 5.50589383e-01 7.57305801e-01
5.52022159e-01 8.62213135e-01 1.35150656e-01 -1.13807023e+00
6.61179185e-01 7.44818270e-01 -2.83753842e-01 -9.56610262e-01
-5.83679438e-01 -3.92636836e-01 -6.54982805e-01 1.79606423e-01
1.40248194e-01 4.78262097e-01 -1.24644482e+00 1.38313174e+00
4.61995959e-01 2.24519253e-01 -5.32716848e-02 7.93688238e-01
9.77303386e-01 3.06722403e-01 -3.67726415e-01 5.13795674e-01
1.00508666e+00 -9.67232466e-01 -2.14379638e-01 -4.93150532e-01
6.48728788e-01 -7.87317514e-01 1.13975155e+00 2.73369253e-01
-6.65867209e-01 -8.49730730e-01 -1.09399509e+00 -4.91227239e-01
-4.77157116e-01 -2.87205596e-02 7.56626844e-01 5.22507250e-01
-1.17041862e+00 5.54087758e-01 -8.58931541e-01 -5.89782238e-01
6.66685402e-01 1.21265151e-01 -7.22350955e-01 -6.98536754e-01
-7.84252524e-01 1.05452609e+00 2.71695852e-01 4.08104062e-02
-7.38864064e-01 -1.04652369e+00 -1.33292747e+00 -5.17137706e-01
1.57672644e-01 -7.84041941e-01 8.09823930e-01 -3.68712664e-01
-1.46394444e+00 1.22528303e+00 1.83337852e-01 -2.39235222e-01
6.52256250e-01 -5.73684692e-01 -4.96952012e-02 1.87330797e-01
2.57193834e-01 1.05295086e+00 7.01268733e-01 -1.53294718e+00
-2.93721378e-01 -6.99570715e-01 2.84985095e-01 5.46580255e-01
7.13730529e-02 -5.85265756e-01 -4.98026460e-01 -1.66267142e-01
7.94721365e-01 -1.00851643e+00 -3.37850451e-01 2.09439263e-01
-3.46686691e-01 1.75482601e-01 1.08287859e+00 -5.52968740e-01
2.31388450e-01 -2.15894628e+00 3.13600115e-02 1.81663886e-01
5.35215400e-02 -1.84629317e-02 -4.59534228e-01 2.23804623e-01
-1.04752012e-01 -6.19187653e-02 -1.73030943e-01 -5.65332830e-01
-2.48403892e-01 5.22824466e-01 -2.48090774e-01 6.57337666e-01
2.00139821e-01 9.22687531e-01 -8.37194979e-01 -3.16511631e-01
9.49558735e-01 6.12352908e-01 -6.62627399e-01 3.79082948e-01
-2.07248375e-01 7.55887687e-01 -4.72307563e-01 5.51432729e-01
1.07212734e+00 -3.31542701e-01 -2.40024850e-01 -5.92572331e-01
-5.46827428e-02 5.13237298e-01 -9.80496466e-01 2.58172846e+00
-9.99909580e-01 6.31437063e-01 -2.63425201e-01 -8.80325794e-01
9.51708853e-01 -2.52918988e-01 5.21792352e-01 -1.05283237e+00
-1.13201849e-02 1.70234501e-01 -4.23106909e-01 -2.52084464e-01
4.36835974e-01 1.85402893e-02 4.31187935e-02 2.52342433e-01
1.99011520e-01 -1.13787532e+00 -3.08932364e-01 1.31005958e-01
9.97662842e-01 8.25347185e-01 1.52448103e-01 -5.92206754e-02
3.01775038e-01 2.27549002e-01 4.84153032e-02 5.46330333e-01
2.80704975e-01 1.07757354e+00 -2.19316691e-01 -5.97006500e-01
-1.27867746e+00 -1.25947034e+00 -4.05215472e-01 6.21873260e-01
2.86470205e-01 -4.51954931e-01 -3.26422974e-02 -6.36085629e-01
7.35959411e-02 6.76791787e-01 -6.38938785e-01 -2.80832350e-02
-6.29447460e-01 -4.41766471e-01 2.73299366e-01 7.56256759e-01
1.04678571e+00 -4.32287574e-01 -8.00567389e-01 -5.13620265e-02
-1.20320305e-01 -1.46878815e+00 1.49700250e-02 3.78667116e-01
-1.15956342e+00 -1.17375016e+00 -2.76378870e-01 -3.12099755e-01
4.67319071e-01 7.14303434e-01 1.42739820e+00 -2.00425327e-01
-1.99864611e-01 8.51774633e-01 -5.29292226e-01 -2.81288773e-01
-2.52378970e-01 2.71526091e-02 -1.87202811e-01 -2.79022098e-01
1.79801419e-01 -7.99346209e-01 -4.66530919e-01 4.77776259e-01
-6.25497103e-01 5.97476780e-01 6.09156311e-01 7.59775341e-01
8.25271487e-01 -4.99423116e-01 -1.31279975e-01 -6.00359023e-01
-4.00209367e-01 -1.56582817e-01 -3.77561271e-01 8.09806678e-03
-3.49689901e-01 1.05764925e-01 3.18000287e-01 -1.51342509e-04
-1.18376899e+00 4.21067089e-01 -2.94392496e-01 -8.18415701e-01
-4.26416069e-01 2.76950985e-01 -2.56942064e-01 -2.83097118e-01
7.28755355e-01 1.88755020e-01 -8.11789092e-03 -6.83442652e-01
7.16265917e-01 4.59320307e-01 3.95910263e-01 -4.61888134e-01
1.13868129e+00 8.52520227e-01 1.94737434e-01 -9.15753424e-01
-1.17563355e+00 -6.99218333e-01 -1.09711981e+00 -2.78470740e-02
9.81553257e-01 -1.37178826e+00 -1.05607264e-01 5.68490922e-01
-1.10297251e+00 -4.26385432e-01 -4.12557423e-01 6.95204675e-01
-8.29207540e-01 3.40483934e-01 -2.83373624e-01 -2.87555397e-01
-1.48157924e-01 -1.05113828e+00 1.70604599e+00 -9.92346480e-02
7.74389133e-02 -8.51483583e-01 1.52758449e-01 7.41454899e-01
3.86584431e-01 3.98946524e-01 7.84508705e-01 -3.43130350e-01
-8.97749841e-01 -1.11215174e-01 -4.06526268e-01 4.54165906e-01
2.76186585e-01 -2.73158759e-01 -1.33566487e+00 -8.16926360e-02
-5.89833781e-02 -7.49781907e-01 7.83928871e-01 2.75608629e-01
1.18004119e+00 5.03476977e-01 -2.47624427e-01 1.11547720e+00
1.46315420e+00 -2.84323514e-01 5.44346690e-01 4.76023763e-01
1.13519692e+00 4.79214817e-01 6.17475212e-01 2.40750566e-01
7.16019511e-01 8.13871145e-01 8.74160290e-01 -4.35675681e-01
-4.73594546e-01 -6.27696157e-01 3.94893587e-02 9.66180384e-01
-6.81075901e-02 -3.33446823e-02 -1.05524516e+00 3.38205814e-01
-1.42691743e+00 -5.19796550e-01 -3.40100765e-01 2.02249742e+00
5.74806452e-01 7.43286461e-02 -5.41942775e-01 1.04595833e-02
2.12984234e-01 4.38705385e-01 -6.76300287e-01 1.27375051e-01
-3.96998823e-01 5.35863519e-01 6.85968518e-01 2.97974467e-01
-1.03865767e+00 9.07380164e-01 5.36366701e+00 7.08661079e-01
-1.14861500e+00 7.33685195e-02 5.31228602e-01 -5.51786125e-02
-4.59825575e-01 1.67970255e-01 -5.50911784e-01 -1.06417857e-01
7.56974399e-01 3.44265461e-01 1.39909595e-01 1.04408312e+00
-5.23207057e-03 -2.93413728e-01 -1.35957026e+00 1.21909785e+00
2.80853182e-01 -1.46314359e+00 -5.68766072e-02 2.11072311e-01
7.75555015e-01 7.85073459e-01 -4.51934114e-02 1.67042881e-01
4.63190138e-01 -9.61590230e-01 7.55023003e-01 2.88588554e-01
9.53452766e-01 -2.72835821e-01 5.56808352e-01 2.15990931e-01
-1.25281334e+00 2.35974729e-01 -2.56898850e-01 5.59624098e-02
2.81112373e-01 5.60629964e-01 -9.30733085e-01 1.16941953e+00
1.05971050e+00 1.42736149e+00 -9.40387487e-01 7.14112461e-01
-1.96862936e-01 2.36006737e-01 -8.21296394e-01 5.26810586e-01
2.50721335e-01 -1.48527846e-01 1.88243046e-01 6.19062364e-01
3.89139920e-01 -1.66551217e-01 1.25575095e-01 7.75683522e-01
7.14782923e-02 -1.83346987e-01 -9.76284742e-01 4.99849141e-01
2.44142741e-01 1.19181192e+00 -6.68690860e-01 -2.01329947e-01
-5.40041804e-01 8.49016726e-01 1.84828341e-01 -1.07049793e-02
-7.56930590e-01 1.71060279e-01 5.39247453e-01 2.59308726e-01
3.36547285e-01 -6.48066998e-01 -2.84053147e-01 -1.20172024e+00
-1.93793047e-02 -6.95713043e-01 5.67469113e-02 -1.22651887e+00
-1.42080379e+00 7.06943035e-01 3.25901896e-01 -1.44789910e+00
-1.69279218e-01 -7.70718932e-01 -2.64610708e-01 6.86287701e-01
-1.77418494e+00 -1.65182769e+00 -9.94788289e-01 7.35449016e-01
6.43918574e-01 1.60648614e-01 7.25603878e-01 1.63802490e-01
-2.20696274e-02 -3.97186130e-02 -1.10996321e-01 -1.38339922e-02
7.29809403e-01 -1.17641950e+00 7.61899829e-01 5.86692214e-01
4.83141124e-01 1.68358430e-01 2.89535135e-01 -3.87812138e-01
-1.52685678e+00 -1.11372316e+00 1.98868543e-01 -9.09839034e-01
3.18867147e-01 -5.50884604e-01 -7.08050370e-01 5.56125760e-01
-5.97831607e-02 3.91086996e-01 5.14052749e-01 2.62099896e-02
-6.23010397e-01 -1.63747102e-01 -1.07746816e+00 2.73649037e-01
1.73726559e+00 -7.35293984e-01 -5.33953071e-01 4.29563195e-01
8.96801531e-01 -8.22125614e-01 -9.94494915e-01 7.97665000e-01
4.24513549e-01 -1.20062792e+00 1.43568063e+00 -3.93804431e-01
5.58815181e-01 -2.27769047e-01 -8.22195411e-01 -1.35556161e+00
9.62166786e-02 -4.48489711e-02 2.70395219e-01 7.94450223e-01
4.69412953e-02 -4.89893436e-01 9.08429801e-01 2.97335178e-01
-4.30728525e-01 -6.25396371e-01 -1.03631818e+00 -7.06764698e-01
3.87143679e-02 -9.99959171e-01 6.96785688e-01 9.04940963e-01
-9.45124388e-01 3.07781279e-01 -1.60609484e-02 1.39977992e-01
6.16444409e-01 3.80735278e-01 1.29321194e+00 -1.31006694e+00
-1.20135427e-01 -6.08299375e-02 -6.64280653e-01 -1.41828001e+00
2.39957750e-01 -8.80088091e-01 -1.03904076e-01 -1.51986778e+00
-1.38979435e-01 -9.16464567e-01 4.47260700e-02 5.19379973e-01
2.65413672e-01 4.58152145e-01 1.81956574e-01 7.03359544e-02
-3.95703793e-01 8.81462216e-01 1.35269368e+00 -7.46481419e-02
-5.39230928e-02 -3.29904646e-01 -1.69121772e-01 8.30555558e-01
5.32655060e-01 -2.36257806e-01 -5.46425998e-01 -8.36510420e-01
2.02386335e-01 -7.23291934e-02 6.53151095e-01 -1.18061364e+00
1.23389196e-02 2.08454300e-02 5.90074837e-01 -1.12935412e+00
8.71712685e-01 -1.00954926e+00 2.84610212e-01 1.30904466e-01
-4.75055389e-02 9.18629244e-02 2.66588122e-01 6.89981282e-01
-5.13773039e-02 3.33005160e-01 6.34555161e-01 -3.70747149e-01
-1.02808011e+00 3.99307370e-01 4.02390748e-01 -7.28304461e-02
8.02463651e-01 -6.07909679e-01 -2.21668333e-01 -4.22902763e-01
-3.60149086e-01 1.61486074e-01 7.79768944e-01 6.38330102e-01
9.02267635e-01 -1.29044902e+00 -5.29712200e-01 6.54541254e-01
4.64070976e-01 7.98136294e-01 4.60973710e-01 8.35983336e-01
-6.93965554e-01 3.63311887e-01 -4.77042973e-01 -1.32401764e+00
-1.09136772e+00 1.19018830e-01 4.44075733e-01 -1.67687908e-01
-7.95425653e-01 9.58104908e-01 2.65466094e-01 -1.07550025e+00
-1.62008777e-01 -3.89230162e-01 1.26201048e-01 -2.33767509e-01
-5.42302169e-02 7.79856667e-02 4.54147786e-01 -8.06695044e-01
-5.35889626e-01 1.16723132e+00 1.07326351e-01 -3.81479040e-02
1.71919882e+00 -2.11017057e-01 3.63645405e-02 2.93093592e-01
1.55338252e+00 -9.36613530e-02 -1.53626478e+00 -3.86473507e-01
-2.49483109e-01 -9.15357232e-01 4.15047497e-01 -5.85882425e-01
-1.15897334e+00 1.05891907e+00 8.55351508e-01 -3.09032202e-01
1.05932486e+00 2.99136221e-01 6.00904405e-01 5.14527202e-01
9.00457442e-01 -6.37696326e-01 4.04348731e-01 5.85590124e-01
9.96828139e-01 -1.60667813e+00 2.54996032e-01 -5.02907634e-01
-3.86894941e-01 1.19328737e+00 7.73867309e-01 -1.63721576e-01
7.28597581e-01 -8.59626681e-02 1.80252805e-01 -4.74973619e-01
-3.39776367e-01 -3.59551042e-01 3.31970602e-01 8.10142279e-01
1.44352600e-01 -2.33761191e-01 5.25026262e-01 -9.08813253e-02
-5.10751784e-01 -2.51002133e-01 3.92814428e-01 9.98280585e-01
1.54385492e-02 -1.13744283e+00 -2.51045197e-01 4.10693765e-01
2.13027745e-01 -2.89249122e-02 -2.05003321e-01 1.11369288e+00
1.52075052e-01 6.91698074e-01 3.68471563e-01 -6.63185358e-01
4.51036692e-01 -2.98853844e-01 8.67079973e-01 -6.99010849e-01
-1.57323703e-02 -1.77290708e-01 1.47610590e-01 -1.15783179e+00
-9.13305342e-01 -5.87752342e-01 -8.64730358e-01 -1.90176949e-01
-3.79888713e-01 -4.10512745e-01 9.44676042e-01 9.98283386e-01
5.79475522e-01 3.09253186e-01 8.07852268e-01 -1.38074827e+00
-2.91643888e-01 -8.60953808e-01 -2.26386815e-01 4.79554951e-01
2.65008003e-01 -1.02572191e+00 -2.08583653e-01 -1.97843865e-01] | [8.290914535522461, -3.043325901031494] |
de44c27d-0e6a-45c2-bd6c-bdbc1aeab044 | tempered-sigmoid-activations-for-deep | 2007.14191 | null | https://arxiv.org/abs/2007.14191v1 | https://arxiv.org/pdf/2007.14191v1.pdf | Tempered Sigmoid Activations for Deep Learning with Differential Privacy | Because learning sometimes involves sensitive data, machine learning algorithms have been extended to offer privacy for training data. In practice, this has been mostly an afterthought, with privacy-preserving models obtained by re-running training with a different optimizer, but using the model architectures that already performed well in a non-privacy-preserving setting. This approach leads to less than ideal privacy/utility tradeoffs, as we show here. Instead, we propose that model architectures are chosen ab initio explicitly for privacy-preserving training. To provide guarantees under the gold standard of differential privacy, one must bound as strictly as possible how individual training points can possibly affect model updates. In this paper, we are the first to observe that the choice of activation function is central to bounding the sensitivity of privacy-preserving deep learning. We demonstrate analytically and experimentally how a general family of bounded activation functions, the tempered sigmoids, consistently outperform unbounded activation functions like ReLU. Using this paradigm, we achieve new state-of-the-art accuracy on MNIST, FashionMNIST, and CIFAR10 without any modification of the learning procedure fundamentals or differential privacy analysis. | ['Úlfar Erlingsson', 'Shuang Song', 'Steve Chien', 'Abhradeep Thakurta', 'Nicolas Papernot'] | 2020-07-28 | null | null | null | null | ['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning'] | ['methodology', 'natural-language-processing'] | [ 1.89665630e-01 2.96671212e-01 -8.79511237e-03 -7.69639969e-01
-9.29512382e-01 -8.02388310e-01 4.97060001e-01 2.04821959e-01
-9.68295395e-01 9.96252596e-01 -1.45563632e-01 -4.60582763e-01
-1.74995750e-01 -5.80935717e-01 -9.41734731e-01 -1.06872535e+00
-1.57134905e-01 2.47527987e-01 -2.65097082e-01 7.95413405e-02
8.07253793e-02 6.80019081e-01 -1.17219961e+00 2.62052238e-01
5.53558111e-01 9.40034211e-01 -7.72639811e-01 6.02522314e-01
3.56078684e-01 5.34395814e-01 -4.80713576e-01 -1.01078331e+00
7.96135426e-01 -2.27814496e-01 -9.31438744e-01 -3.83779496e-01
5.14626503e-01 -5.25829732e-01 -4.66988266e-01 1.24211884e+00
3.92334789e-01 2.20332623e-01 4.53560770e-01 -1.18812358e+00
-6.07342124e-01 6.80178106e-01 -2.03062117e-01 1.16662849e-02
-3.93072933e-01 9.07731801e-02 1.00579691e+00 -4.04275924e-01
3.63543838e-01 8.36999536e-01 6.74015403e-01 9.46309686e-01
-1.58702183e+00 -7.74009168e-01 7.28816092e-02 -2.29839697e-01
-1.52586317e+00 -7.09019661e-01 3.16996038e-01 -2.53236979e-01
6.92773640e-01 7.54202724e-01 4.87298101e-01 1.13975060e+00
2.13619024e-01 7.98287153e-01 1.01367450e+00 -2.71051615e-01
4.04430509e-01 6.67310536e-01 3.96961778e-01 4.54645723e-01
5.73851764e-01 2.03449801e-01 -3.76061767e-01 -7.05924809e-01
3.55331898e-01 1.04754180e-01 -6.05637968e-01 -9.35227513e-01
-6.93169296e-01 9.77999151e-01 3.13310206e-01 5.88360094e-02
-5.80197712e-03 4.27134633e-01 6.15697145e-01 7.13530898e-01
6.37867272e-01 6.35124445e-01 -6.74419701e-01 2.28486344e-01
-8.82416427e-01 4.82820213e-01 9.67680037e-01 6.97752059e-01
6.33591056e-01 -1.63304716e-01 -7.09496289e-02 4.16339546e-01
3.44352089e-02 -1.26315340e-01 4.06994373e-01 -1.03289700e+00
3.27307791e-01 3.48135307e-02 2.17844114e-01 -6.01262987e-01
-1.06428809e-01 -5.94141841e-01 -7.65841901e-01 5.07811785e-01
7.84474075e-01 -4.52714890e-01 -5.90572715e-01 2.04863334e+00
2.10101575e-01 -1.94093049e-01 3.86157244e-01 7.54577398e-01
-2.98690703e-03 2.60430217e-01 5.92410229e-02 -1.83777586e-01
1.13771594e+00 -5.89715600e-01 -4.79761004e-01 -9.43916291e-03
1.04162931e+00 -2.03515068e-01 1.11168933e+00 4.37383473e-01
-9.50814903e-01 3.03844571e-01 -1.17188287e+00 -3.42384636e-01
-3.24931145e-01 -2.79496759e-01 7.74628103e-01 1.09016466e+00
-1.04898155e+00 9.92793143e-01 -1.02221918e+00 -2.93984473e-01
9.12750781e-01 6.91243529e-01 -7.79313922e-01 2.04645500e-01
-1.27366364e+00 8.64865780e-01 4.02182221e-01 -2.55636871e-02
-7.43721068e-01 -1.01267993e+00 -5.84868193e-01 2.68915653e-01
2.47821540e-01 -7.11620510e-01 1.26135468e+00 -1.14761555e+00
-1.35958016e+00 1.18056238e+00 1.76837608e-01 -1.18062449e+00
1.04098380e+00 -1.91699728e-01 1.34798065e-02 -3.46227348e-01
-6.11769259e-01 5.16119182e-01 7.03965604e-01 -1.13740516e+00
-4.43765640e-01 -7.82318771e-01 8.70094076e-02 1.88787222e-01
-5.96926332e-01 -6.93698460e-03 -1.49177201e-02 -4.84721154e-01
-1.19379297e-01 -1.04656923e+00 -3.59316736e-01 4.97074962e-01
-4.46087599e-01 2.78082520e-01 6.62805796e-01 -5.24106264e-01
9.17598367e-01 -2.41486526e+00 -3.58160228e-01 4.90803093e-01
2.65223414e-01 2.53361404e-01 2.64696419e-01 4.16778624e-02
-7.94683695e-02 3.47072631e-01 -5.95424414e-01 -8.87715399e-01
3.62294108e-01 3.00388902e-01 -5.65488994e-01 1.08619273e+00
-2.00206473e-01 7.83412278e-01 -4.16513324e-01 -2.13006422e-01
-1.57418519e-01 6.37582541e-01 -8.74424934e-01 -1.26633570e-01
3.20902131e-02 2.57983744e-01 -4.55327958e-01 1.54732063e-01
8.23386192e-01 -6.98819980e-02 1.75978482e-01 3.13195705e-01
1.27979949e-01 2.75054932e-01 -8.83238792e-01 1.57145369e+00
-5.81306182e-02 6.54654384e-01 3.87589961e-01 -9.43064153e-01
5.77723920e-01 3.10196519e-01 3.41985732e-01 -2.03217059e-01
2.42753163e-01 1.84321702e-01 2.69839596e-02 -5.08628450e-02
2.71461010e-01 -1.93004653e-01 6.96874484e-02 6.01386726e-01
-1.01941884e-01 2.80485123e-01 -6.00151181e-01 -1.28930047e-01
1.04721594e+00 -2.51943737e-01 2.75847852e-01 -6.53671205e-01
2.06842631e-01 -2.34919190e-01 5.70168495e-01 8.42664301e-01
-4.04651672e-01 7.14011669e-01 6.89159572e-01 -4.90094632e-01
-9.81908202e-01 -7.69771099e-01 -5.25260389e-01 1.05966902e+00
-3.65559101e-01 -2.09333330e-01 -1.11507273e+00 -9.19857979e-01
2.79640496e-01 9.45661128e-01 -9.20904934e-01 -4.17732924e-01
-4.65002090e-01 -1.08046031e+00 9.88511503e-01 7.84368441e-02
4.18016493e-01 -5.50682604e-01 -7.10077345e-01 -2.13577673e-01
4.30887580e-01 -5.05645156e-01 -5.56395292e-01 7.81397045e-01
-9.40661430e-01 -6.97393179e-01 -5.33771396e-01 -2.62313038e-01
7.99866080e-01 -2.08427161e-01 8.11474979e-01 4.28073481e-02
-6.73753619e-02 1.62453547e-01 1.40357688e-01 -4.88823533e-01
-3.23217601e-01 2.76406229e-01 -7.28258444e-03 2.89746106e-01
3.97901982e-01 -7.07265198e-01 -6.10458970e-01 -1.02459662e-01
-1.00616992e+00 -3.33799034e-01 2.97166258e-01 8.71903837e-01
6.10469341e-01 -6.65969327e-02 3.97573523e-02 -1.54302871e+00
6.15569413e-01 -3.59769583e-01 -7.44941473e-01 4.29786712e-01
-1.05844367e+00 3.86710078e-01 6.95337176e-01 -3.45927209e-01
-8.76491129e-01 3.27050209e-01 -1.53624073e-01 -6.31434321e-01
-1.64198026e-01 4.59824614e-02 -3.88646305e-01 -5.36753297e-01
7.82088518e-01 2.02285871e-01 2.57149518e-01 -5.70954442e-01
3.48420441e-01 4.70842332e-01 3.67020637e-01 -6.76483810e-01
5.64964235e-01 6.35294139e-01 8.73221010e-02 -5.00023007e-01
-5.25311232e-01 1.36227131e-01 -4.34508413e-01 4.97448355e-01
6.86919332e-01 -7.34216094e-01 -8.85066032e-01 2.79091150e-01
-9.83082712e-01 -3.00658703e-01 -7.27905214e-01 3.63092214e-01
-5.82826018e-01 2.97639906e-01 -5.80860734e-01 -9.21701252e-01
-4.70728695e-01 -9.83158410e-01 4.32608843e-01 -1.47353932e-01
-2.49237239e-01 -1.06016195e+00 -4.87369448e-02 1.30206034e-01
4.53725517e-01 3.67320925e-01 8.40605378e-01 -1.33598971e+00
-4.64249641e-01 -4.67482716e-01 1.95840970e-01 5.64146578e-01
-1.66808099e-01 -2.24751487e-01 -1.54739153e+00 -5.70639193e-01
5.29788911e-01 -3.37153643e-01 9.31831002e-01 2.49851108e-01
1.62755370e+00 -1.01153398e+00 -1.53164059e-01 1.35484385e+00
1.36173558e+00 5.84911779e-02 6.14257157e-01 3.32545757e-01
5.29677689e-01 8.43002677e-01 5.94684333e-02 4.85758156e-01
5.79185262e-02 4.15926903e-01 4.26032543e-01 3.83820869e-02
7.41744280e-01 -2.48045951e-01 2.32258692e-01 -4.99646701e-02
3.48181814e-01 -1.71709329e-01 -6.48429453e-01 1.82153940e-01
-1.81462920e+00 -8.71963739e-01 1.39466077e-01 2.65371275e+00
1.05225682e+00 -3.26090716e-02 4.91001159e-02 -1.36387572e-01
3.45942825e-01 1.57325342e-01 -7.82519519e-01 -6.02737904e-01
-2.21224576e-01 1.02613203e-01 1.31085789e+00 5.78409374e-01
-1.22327900e+00 6.86754882e-01 6.58745146e+00 7.61758387e-01
-1.11584270e+00 3.06705713e-01 1.20838654e+00 -6.79294527e-01
-4.90221173e-01 9.10156220e-02 -6.08995795e-01 3.03823054e-01
1.04376245e+00 -3.17861259e-01 7.46342003e-01 1.06697059e+00
-2.10952818e-01 3.39325219e-01 -1.51617587e+00 9.20548141e-01
-3.42839748e-01 -1.12847435e+00 -3.51665556e-01 4.59001482e-01
5.36963105e-01 5.45794144e-02 5.13975322e-01 1.30391359e-01
5.39123833e-01 -1.23951089e+00 6.22139454e-01 2.61667520e-01
5.44380963e-01 -8.61367583e-01 4.15438175e-01 4.00534123e-01
-3.50907177e-01 -6.32105023e-02 -4.74660426e-01 1.88072205e-01
-2.86396831e-01 4.17583793e-01 -5.29729128e-01 2.05084994e-01
6.45897925e-01 1.50211751e-01 -3.86798888e-01 8.24307919e-01
6.92997053e-02 5.92325211e-01 -6.48797810e-01 8.47955793e-02
1.71664536e-01 -1.82587028e-01 3.89950454e-01 1.09580195e+00
1.84392691e-01 5.92741407e-02 -5.75568497e-01 8.57796133e-01
-4.31715012e-01 1.05269320e-01 -8.08137953e-01 4.63918149e-02
4.61643100e-01 1.03748536e+00 -1.93748474e-01 1.11262910e-01
-1.65185809e-01 1.10568917e+00 5.56966126e-01 4.68653619e-01
-6.27258301e-01 -1.68367624e-01 1.30888426e+00 1.66390121e-01
1.21557131e-01 1.37910759e-02 -6.32799447e-01 -1.23682702e+00
2.15617359e-01 -9.53019440e-01 7.70194173e-01 9.85596981e-03
-1.33264875e+00 3.32657069e-01 -5.22018038e-02 -8.04990053e-01
-2.75751501e-01 -4.75756168e-01 -3.74455959e-01 1.00114119e+00
-1.28508091e+00 -7.12405145e-01 5.55326104e-01 6.49638653e-01
-1.87903911e-01 -6.14100546e-02 1.05044985e+00 2.50307113e-01
-5.41611016e-01 1.33455884e+00 6.76086426e-01 1.61557809e-01
6.51977658e-01 -1.20071816e+00 4.05676663e-01 1.00171709e+00
3.01994205e-01 1.05457628e+00 8.12278807e-01 -1.41661182e-01
-1.37545371e+00 -1.07162321e+00 7.93905139e-01 -5.06372631e-01
3.89674395e-01 -7.59069264e-01 -1.27055407e+00 1.16296279e+00
-1.21418284e-02 4.16882813e-01 8.38831544e-01 1.27519965e-01
-4.45950001e-01 -3.50670397e-01 -1.72774529e+00 6.03922307e-01
9.19728398e-01 -6.99060082e-01 -8.43951553e-02 4.00483519e-01
8.38216722e-01 -4.81882542e-01 -7.47988939e-01 1.07523091e-01
6.02144957e-01 -9.80349839e-01 8.78101826e-01 -1.05287623e+00
-1.73630834e-01 1.00708559e-01 -3.65417480e-01 -9.94012594e-01
5.00639342e-02 -9.68082190e-01 -2.12290540e-01 1.20293415e+00
5.68774819e-01 -1.11748528e+00 1.30318236e+00 1.73633575e+00
2.45642513e-01 -8.83617342e-01 -1.30234766e+00 -6.78189099e-01
6.12450898e-01 -3.09470266e-01 8.28233600e-01 1.19498587e+00
-2.39798144e-01 -3.66951108e-01 -5.70368588e-01 2.48008758e-01
7.73857355e-01 -3.61847103e-01 7.67284930e-01 -9.32432652e-01
-5.60264528e-01 -5.00721693e-01 -3.17240477e-01 -7.74893701e-01
4.41289693e-01 -9.92413580e-01 -1.75326690e-01 -6.42576933e-01
-1.62926689e-03 -5.35037339e-01 -6.88247800e-01 6.69867754e-01
6.57688007e-02 -4.22056355e-02 5.25609665e-02 1.82098791e-01
-1.45902917e-01 4.88726914e-01 8.18654418e-01 9.57259461e-02
-7.31582120e-02 3.38719100e-01 -1.03555286e+00 5.08059144e-01
8.93953443e-01 -8.59254837e-01 -5.89101613e-01 -5.46133280e-01
6.50646538e-02 -4.06395644e-01 5.82925022e-01 -7.19601214e-01
2.87459075e-01 -9.78136435e-02 3.07227343e-01 1.12192027e-01
3.82938743e-01 -1.21557605e+00 4.61557060e-01 5.75825632e-01
-9.24763620e-01 -1.55688599e-01 1.00894406e-01 4.88461226e-01
1.13477200e-01 -3.27970296e-01 1.07141089e+00 6.57770410e-02
-1.81478918e-01 5.87268591e-01 5.08428290e-02 9.31703448e-02
1.09522247e+00 -5.90673240e-04 -2.83176899e-01 -3.37472230e-01
-6.13305569e-01 1.40217319e-01 8.03119481e-01 -9.71614271e-02
2.12659866e-01 -1.03069246e+00 -5.90992332e-01 4.20001090e-01
-1.69627443e-02 -3.57050970e-02 1.34128332e-01 6.87131166e-01
-4.20565546e-01 2.97278047e-01 3.52316163e-02 -1.44668102e-01
-1.28051186e+00 6.36611223e-01 7.27717817e-01 -2.00352162e-01
-6.78413570e-01 1.14362216e+00 4.00221974e-01 -4.81564283e-01
4.94861215e-01 -2.97415942e-01 4.38111871e-01 -2.11168215e-01
4.03330594e-01 1.12064160e-01 2.55528063e-01 -2.51842111e-01
-3.78633410e-01 -1.27823964e-01 -4.03998017e-01 -3.39167237e-01
1.18484890e+00 2.52008829e-02 2.21797545e-02 3.07431757e-01
1.69402337e+00 -1.89423591e-01 -1.58487558e+00 -1.60001814e-01
-9.02678743e-02 -7.42523015e-01 5.81458136e-02 -6.07470155e-01
-1.15241325e+00 9.63199377e-01 7.44288743e-01 1.64951876e-01
1.00227177e+00 -3.73446018e-01 6.14161313e-01 6.63749337e-01
4.72792655e-01 -8.07136953e-01 -9.63625431e-01 2.00585648e-01
5.83445370e-01 -1.15185630e+00 9.67124179e-02 1.10924877e-01
-5.96113205e-01 7.57471561e-01 3.06384414e-01 -1.44084305e-01
7.70136714e-01 4.23093140e-01 -5.28303124e-02 2.35773399e-01
-8.75271142e-01 4.79740769e-01 1.11073546e-01 4.52127069e-01
1.93565637e-01 3.19286920e-02 -2.74762511e-01 7.79907167e-01
-3.75176907e-01 -1.90294951e-01 1.58041656e-01 1.00515580e+00
-1.78836942e-01 -1.20887256e+00 -1.73339859e-01 4.54902619e-01
-1.11027789e+00 -1.37350336e-01 -5.32182455e-01 9.89641905e-01
-1.82154134e-01 3.33021730e-01 1.52171729e-02 -3.31290692e-01
1.52054161e-01 3.83160055e-01 3.56425047e-01 -2.25995675e-01
-9.63660896e-01 -4.92856264e-01 -7.11220875e-02 -5.55337727e-01
2.50130057e-01 -8.25126171e-01 -1.09144378e+00 -7.98619926e-01
-7.10172802e-02 2.46793091e-01 7.47013211e-01 7.30719745e-01
4.71242875e-01 -8.49204957e-02 4.83196974e-01 -2.96560556e-01
-1.34523857e+00 -4.23397183e-01 -8.41096222e-01 3.93594623e-01
6.32511675e-01 -4.44729552e-02 -6.70753181e-01 -2.87614226e-01] | [5.945583343505859, 6.865639686584473] |
6ba6dc14-8c6a-4722-a9ae-a4d9c49dadfe | unsupervised-machine-translation-on-dravidian | 2103.15877 | null | https://arxiv.org/abs/2103.15877v1 | https://arxiv.org/pdf/2103.15877v1.pdf | Unsupervised Machine Translation On Dravidian Languages | Unsupervised neural machine translation (UNMT) is beneficial especially for low resource languages such as those from the Dravidian family. However, UNMT systems tend to fail in realistic scenarios involving actual low resource languages. Recent works propose to utilize auxiliary parallel data and have achieved state-of-the-art results. In this work, we focus on unsupervised translation between English and Kannada, a low resource Dravidian language. We additionally utilize a limited amount of auxiliary data between English and other related Dravidian languages. We show that unifying the writing systems is essential in unsupervised translation between the Dravidian languages. We explore several model architectures that use the auxiliary data in order to maximize knowledge sharing and enable UNMT for distant language pairs. Our experiments demonstrate that it is crucial to include auxiliary languages that are similar to our focal language, Kannada. Furthermore, we propose a metric to measure language similarity and show that it serves as a good indicator for selecting the auxiliary languages. | ['Jan Niehues', 'Danni Liu', 'Sai Koneru'] | 2021-03-29 | null | https://aclanthology.org/2021.dravidianlangtech-1.7 | https://aclanthology.org/2021.dravidianlangtech-1.7.pdf | eacl-dravidianlangtech-2021-4 | ['unsupervised-machine-translation'] | ['natural-language-processing'] | [-2.16175675e-01 -2.43237570e-01 -4.93387014e-01 -2.15593159e-01
-8.47953916e-01 -7.25957036e-01 7.82316864e-01 -1.10497460e-01
-6.89193249e-01 8.49981546e-01 4.88731027e-01 -5.70856333e-01
1.92911565e-01 -6.77323103e-01 -5.39107740e-01 -4.20131773e-01
3.88021231e-01 8.63420010e-01 -2.82752037e-01 -6.60576344e-01
6.13174811e-02 4.02698308e-01 -8.69516373e-01 1.89615533e-01
1.37090325e+00 -3.83244902e-02 5.25212467e-01 2.42849663e-01
-3.10518593e-01 3.28537107e-01 -5.38367808e-01 -6.09634995e-01
5.37467897e-01 -7.56794810e-01 -1.06234300e+00 -3.12922537e-01
1.41069025e-01 -1.97772354e-01 3.21914209e-03 8.47567916e-01
5.86951911e-01 -4.53395955e-02 7.77246118e-01 -9.05078530e-01
-9.18851614e-01 1.13906693e+00 -4.70517635e-01 2.57657617e-01
5.87090477e-02 -1.29714012e-01 1.09413886e+00 -1.21559501e+00
7.87388802e-01 9.97120619e-01 5.20667851e-01 4.48090047e-01
-1.04247415e+00 -5.81498325e-01 1.38081261e-03 7.58142546e-02
-1.69868994e+00 -5.06937385e-01 6.86382353e-01 -2.26673037e-01
1.02131021e+00 5.62797524e-02 3.05745214e-01 1.13785231e+00
4.51824144e-02 7.49211729e-01 1.15232861e+00 -8.15595031e-01
-1.91268206e-01 3.13931078e-01 3.11277770e-02 5.74177325e-01
1.91818282e-01 -2.61541784e-01 -7.69488633e-01 1.48573034e-02
6.70132279e-01 -2.95990884e-01 -1.99479342e-01 -9.62197930e-02
-1.74482000e+00 9.25250947e-01 1.54766351e-01 7.30297685e-01
-3.20538074e-01 -2.71263510e-01 3.29023361e-01 7.72319674e-01
4.90674049e-01 7.73448229e-01 -4.56839710e-01 -1.42209962e-01
-8.52195919e-01 -4.05143023e-01 8.49546790e-01 1.20635581e+00
9.47861552e-01 6.27464205e-02 7.41827413e-02 1.26260650e+00
1.81186453e-01 6.02642477e-01 6.19521618e-01 -5.93463838e-01
8.81765723e-01 3.75432909e-01 -6.86049536e-02 -4.81347382e-01
-1.93099782e-01 -4.80542600e-01 -9.11725521e-01 -5.10946274e-01
3.22870374e-01 -1.70145854e-01 -5.93989491e-01 1.87346482e+00
3.85658331e-02 -4.64783341e-01 5.07703006e-01 9.26924884e-01
5.80249906e-01 6.92675173e-01 -3.15417826e-01 -2.39837602e-01
1.10823476e+00 -1.25210989e+00 -5.04444122e-01 -2.13226318e-01
1.11039031e+00 -1.19021380e+00 1.53412974e+00 3.46005298e-02
-1.01097536e+00 -5.54023027e-01 -8.46199512e-01 -2.24092335e-01
-3.85544538e-01 6.08474016e-01 5.47505200e-01 5.91296732e-01
-1.13518608e+00 4.33680385e-01 -8.89229298e-01 -9.68271375e-01
-1.75202355e-01 4.42738831e-01 -5.14247179e-01 1.04234442e-02
-1.34509993e+00 1.26956713e+00 3.66496950e-01 3.17808017e-02
-5.66444814e-01 -2.54655331e-01 -4.96666074e-01 -2.87515104e-01
-1.12324588e-01 -7.42929459e-01 9.92716730e-01 -1.23683989e+00
-1.51317167e+00 1.05654156e+00 -1.78376049e-01 -2.07601309e-01
4.64824349e-01 -3.02763790e-01 -3.69141936e-01 3.83823775e-02
1.73213959e-01 5.69667637e-01 3.70813429e-01 -9.06831741e-01
-4.14713711e-01 -3.18751574e-01 -1.16009392e-01 5.21684527e-01
-8.73494208e-01 4.13797349e-01 -6.35761261e-01 -9.00481284e-01
6.73809797e-02 -1.21209002e+00 -5.48694655e-02 -3.19197893e-01
-4.02546942e-01 -1.24647513e-01 3.85475129e-01 -8.01791191e-01
1.13250732e+00 -1.70212233e+00 4.22481298e-01 2.18970329e-01
-9.49870199e-02 7.08358213e-02 -4.19124007e-01 8.80594850e-01
3.82418752e-01 1.99823484e-01 -2.40896717e-01 -3.28787833e-01
-2.15151682e-01 4.83853817e-01 -2.01163828e-01 2.89061397e-01
2.20919937e-01 9.39196110e-01 -7.26047158e-01 -3.60803068e-01
-3.36251348e-01 3.31378698e-01 -3.13636750e-01 1.65099561e-01
7.74440989e-02 6.77466094e-01 -3.17741990e-01 5.57268441e-01
3.83296341e-01 -1.23829365e-01 4.60661799e-01 2.52659339e-02
-2.55611092e-01 7.04002023e-01 -5.04306078e-01 1.91727531e+00
-6.91235900e-01 5.85988700e-01 -1.12982206e-01 -8.31492543e-01
1.16434991e+00 1.80480078e-01 2.53177464e-01 -5.83845377e-01
1.04168870e-01 7.81980574e-01 5.08872092e-01 -1.78373769e-01
8.33510101e-01 -5.89742512e-02 -5.78805059e-02 9.57006276e-01
-1.07901776e-02 -2.17035577e-01 2.75713921e-01 9.04159844e-02
7.22142220e-01 1.86709404e-01 4.36740786e-01 -6.67384386e-01
4.30583090e-01 1.61371663e-01 4.62071240e-01 5.76588809e-01
5.58629334e-02 5.67032576e-01 6.57197610e-02 -2.44014382e-01
-1.29529774e+00 -1.16622329e+00 1.09438501e-01 1.10724735e+00
9.80706736e-02 -4.90036130e-01 -5.94551086e-01 -6.37956917e-01
-4.40287441e-01 7.50781655e-01 -4.18725789e-01 -1.40655741e-01
-9.13782239e-01 -8.36828947e-01 9.59176064e-01 3.96593332e-01
4.11331624e-01 -6.73335075e-01 -1.48196861e-01 6.98866472e-02
-5.70011854e-01 -9.64504242e-01 -7.27238953e-01 4.20410007e-01
-8.68365288e-01 -4.73757654e-01 -1.15914381e+00 -1.06138623e+00
6.69359505e-01 4.13439810e-01 1.36664557e+00 -1.03682376e-01
5.22599936e-01 3.28685232e-02 -5.13932467e-01 -1.89831868e-01
-8.47906351e-01 9.01333749e-01 4.21074092e-01 -1.37736350e-01
6.53402507e-01 -6.78285480e-01 -1.00915700e-01 5.29330313e-01
-6.23480856e-01 3.01258385e-01 7.17187405e-01 8.77469599e-01
4.77682531e-01 -7.07780361e-01 6.15769386e-01 -8.44642580e-01
6.80115223e-01 -6.46418452e-01 -3.24512690e-01 5.78827322e-01
-5.32254219e-01 3.84209484e-01 1.00728285e+00 -6.47369385e-01
-8.89594316e-01 -2.01665029e-01 -2.23704681e-01 -1.45096257e-01
3.26000839e-01 7.47951031e-01 -9.04135555e-02 9.45251137e-02
7.65578747e-01 3.42441410e-01 -5.88291697e-02 -7.17983663e-01
3.88760448e-01 1.07276702e+00 1.74543798e-01 -8.85754704e-01
7.31384218e-01 1.58500895e-01 -3.50193441e-01 -8.92793894e-01
-3.63312155e-01 -3.59441668e-01 -1.07476556e+00 1.79377988e-01
5.60259700e-01 -1.01357079e+00 4.57813144e-02 1.65940478e-01
-1.24119163e+00 -3.89101356e-01 2.33272221e-02 8.70003223e-01
-4.35422629e-01 3.73367578e-01 -8.35726082e-01 -3.73485327e-01
-5.72417259e-01 -1.24704981e+00 8.81054342e-01 -7.78933391e-02
-4.40459549e-01 -1.07787061e+00 3.09589297e-01 1.06687844e-01
4.48129535e-01 -4.96821404e-01 1.18611825e+00 -9.11532938e-01
-3.68660748e-01 4.28101808e-01 -1.23191200e-01 3.10589731e-01
5.72052598e-01 -8.24597925e-02 -3.52660477e-01 -3.15181673e-01
-1.32153049e-01 -3.79093885e-01 6.00296199e-01 -1.36138529e-01
2.82529622e-01 -2.06806868e-01 -1.28866643e-01 6.78172648e-01
1.10424495e+00 7.64684333e-03 2.70378083e-01 3.64098221e-01
7.90401459e-01 7.90687442e-01 5.16163886e-01 -7.74610043e-02
6.93824947e-01 7.77419508e-01 -3.28500539e-01 -3.32383811e-01
-2.01555952e-01 -3.31861138e-01 8.47598255e-01 2.02268887e+00
-7.12075606e-02 -1.97497070e-01 -1.30407166e+00 8.60386848e-01
-1.71114433e+00 -3.95530581e-01 -1.41783878e-01 2.18678570e+00
1.30826402e+00 -3.21188778e-01 3.15458551e-02 -6.00545466e-01
5.65630257e-01 -2.07140341e-01 -2.47746989e-01 -5.82558036e-01
-6.19908869e-01 1.65312722e-01 4.32297051e-01 5.46171665e-01
-6.87159717e-01 1.49893415e+00 6.39389563e+00 6.59603119e-01
-1.30616367e+00 3.39251488e-01 4.59584773e-01 3.28092426e-02
-6.24560237e-01 1.71102032e-01 -8.06063712e-01 2.54459620e-01
1.06364298e+00 -3.02681923e-01 4.47614819e-01 4.68769640e-01
2.47205019e-01 2.55196869e-01 -1.25731707e+00 7.95130610e-01
1.96418568e-01 -1.08138657e+00 3.18977058e-01 -1.62166927e-03
9.96574759e-01 6.92180514e-01 -1.73840970e-01 1.40852928e-01
6.34415925e-01 -8.86670828e-01 5.14855862e-01 1.96860790e-01
7.96655536e-01 -8.18548620e-01 6.23117208e-01 4.47935402e-01
-9.43603218e-01 4.48029190e-01 -5.93161643e-01 1.20835744e-01
7.21692294e-02 2.81595379e-01 -8.14098060e-01 7.79696047e-01
4.98885572e-01 7.14552939e-01 -6.21124208e-01 4.43969429e-01
-4.80883718e-01 6.81478620e-01 -5.53139031e-01 -1.29356399e-01
3.57390404e-01 -6.40533328e-01 4.55861151e-01 1.32361972e+00
6.98469758e-01 -3.33836585e-01 3.31404418e-01 5.58566272e-01
-2.76435494e-01 8.17297101e-01 -9.89822149e-01 -3.42468351e-01
4.70126122e-01 8.80080044e-01 -6.69218123e-01 -3.15639555e-01
-6.48721337e-01 1.46648252e+00 6.75610900e-01 3.59897435e-01
-5.30230463e-01 -2.88187593e-01 6.52374446e-01 -2.00949207e-01
-6.17961101e-02 -5.52441299e-01 -2.76999950e-01 -1.66362083e+00
5.34813218e-02 -1.18275964e+00 2.25708723e-01 -6.21048152e-01
-1.45905149e+00 1.10121763e+00 -1.71340764e-01 -1.44268119e+00
-5.47798634e-01 -3.49894196e-01 -2.88611770e-01 1.27593601e+00
-1.41377878e+00 -1.67722952e+00 2.65219599e-01 7.77369440e-01
5.30697703e-01 -6.15850806e-01 1.01803660e+00 4.33272779e-01
-4.23740476e-01 8.70503962e-01 5.55055380e-01 2.42962748e-01
1.10332251e+00 -8.68690848e-01 5.78091323e-01 1.07572401e+00
6.46897078e-01 1.16070056e+00 4.52455759e-01 -7.22026944e-01
-1.58009100e+00 -1.05823874e+00 1.40479767e+00 -3.66190314e-01
7.76831448e-01 -5.51891565e-01 -7.99115837e-01 8.45341384e-01
5.67769051e-01 -5.44678092e-01 9.52867687e-01 4.64276254e-01
-5.01782060e-01 3.10838744e-02 -6.39019072e-01 1.27237010e+00
1.11413574e+00 -8.72896194e-01 -5.82686901e-01 4.50302392e-01
7.08551288e-01 1.59321994e-01 -8.97387922e-01 1.64539590e-01
4.55429882e-01 -3.98062855e-01 6.45061076e-01 -7.15277731e-01
5.67165852e-01 -3.61552328e-01 -3.16963315e-01 -1.67081285e+00
-2.30817318e-01 -7.76670694e-01 2.08252341e-01 1.31605113e+00
8.04108202e-01 -7.19108105e-01 5.66540182e-01 5.81637546e-02
-1.11409016e-01 -3.21304202e-01 -8.03988516e-01 -1.15599298e+00
5.55405557e-01 -1.08089857e-01 3.80685687e-01 1.41376066e+00
6.72130287e-02 9.33690608e-01 -6.02769136e-01 -2.77014356e-02
1.62560627e-01 2.44798869e-01 9.28498685e-01 -8.91058385e-01
-3.73967230e-01 -4.17956322e-01 8.13458115e-02 -1.22366560e+00
3.11209559e-01 -1.38375711e+00 1.10865692e-02 -1.38903821e+00
1.87219039e-01 -7.06192732e-01 -2.83155173e-01 5.57162404e-01
-2.42371947e-01 5.83140671e-01 1.49370670e-01 7.24034727e-01
-9.50272530e-02 6.97454393e-01 1.02769828e+00 -1.46055847e-01
-4.53760803e-01 -2.30025366e-01 -8.62514377e-01 4.58560467e-01
1.09011662e+00 -5.50407112e-01 -3.80158633e-01 -1.14967358e+00
1.11222751e-01 -2.41737783e-01 -3.42165619e-01 -4.15438920e-01
2.03262255e-01 -3.31580609e-01 2.86902785e-02 -4.47744548e-01
1.96520165e-01 -6.29965186e-01 1.10560738e-01 3.93504769e-01
-3.74928027e-01 7.45029688e-01 3.20647918e-02 -7.76459277e-02
-3.63663882e-01 -2.33772486e-01 6.80391729e-01 -1.51495606e-01
-4.09160376e-01 1.16230607e-01 -5.78363955e-01 -1.36061525e-03
5.29908240e-01 3.18186134e-02 -1.11337125e-01 -4.48979825e-01
-1.29714429e-01 7.45342597e-02 7.83086956e-01 6.39337599e-01
3.27774048e-01 -1.34753704e+00 -1.21079838e+00 2.45623112e-01
4.70549762e-01 -6.60095215e-01 -5.05685270e-01 1.02401102e+00
-6.86612248e-01 3.51430237e-01 -4.01444554e-01 -3.76973599e-01
-1.19788003e+00 3.43174160e-01 8.54335651e-02 -4.21398610e-01
-3.06495756e-01 5.81945956e-01 8.93198550e-02 -1.07100296e+00
-1.51957348e-01 -8.08231905e-02 -4.54531685e-02 -2.24494543e-02
8.54399502e-02 -4.58210111e-02 1.72963738e-01 -9.48188186e-01
-3.50407511e-01 3.91463399e-01 -3.21659952e-01 -4.29737955e-01
1.26204729e+00 -4.10567313e-01 -5.38678586e-01 5.97890317e-01
1.15481901e+00 5.41340232e-01 -4.24677640e-01 -5.76692104e-01
2.90552020e-01 -1.83550000e-01 -3.09074044e-01 -6.81176424e-01
-7.78785288e-01 1.06215787e+00 2.35950828e-01 -3.54014486e-01
1.18527329e+00 -6.59946874e-02 7.63547719e-01 8.12510073e-01
5.51701367e-01 -1.29517138e+00 -3.12038481e-01 1.12025738e+00
6.30791306e-01 -1.22142553e+00 -1.72281355e-01 -2.63207108e-01
-7.33895600e-01 9.94844675e-01 4.89054143e-01 1.86177701e-01
3.15734953e-01 1.00947686e-01 4.79509294e-01 2.54519463e-01
-6.40671134e-01 -3.43260139e-01 3.50878298e-01 5.70577502e-01
9.23896432e-01 4.22274917e-02 -8.91973197e-01 4.49014395e-01
-5.70629537e-01 -3.84982705e-01 4.77481157e-01 7.92147100e-01
-8.86998605e-03 -1.77169335e+00 -2.68153638e-01 2.14528069e-01
-6.11647189e-01 -7.34032273e-01 -7.96876848e-01 8.78287613e-01
-2.12767020e-01 7.87631989e-01 1.86375342e-02 -2.88471937e-01
-8.82013887e-02 1.03468567e-01 4.60439593e-01 -7.10743487e-01
-1.01880968e+00 2.72712737e-01 2.90551215e-01 -2.67579919e-03
-3.89990181e-01 -3.12863767e-01 -9.16656673e-01 -3.73939872e-01
-1.86238959e-01 4.94220495e-01 6.36478186e-01 8.44700396e-01
4.40828919e-01 -3.76017243e-02 6.15849495e-01 -5.16619742e-01
-2.92359084e-01 -1.06151211e+00 -3.19922447e-01 2.14717150e-01
-8.34015310e-02 -2.24595785e-01 1.30644277e-01 7.89266825e-02] | [11.509965896606445, 10.267807006835938] |
d19b624c-e608-4217-9136-b929d65e4b28 | progressive-class-semantic-matching-for-semi | null | null | https://openreview.net/forum?id=FitLLp-Jwa | https://openreview.net/pdf?id=FitLLp-Jwa | Progressive Class Semantic Matching for Semi-supervised Text Classification | Semi-supervised learning is a promising way to reduce the annotation cost for text-classification. Combining with pre-trained language models (PLMs), e.g., BERT, recent semi-supervised learning methods achieved impressive performance. In this work, we further investigate the marriage between semi-supervised learning and a pre-trained language model. Unlike existing approaches that utilize PLMs only for model parameter initialization, we explore the inherent topic matching capability inside PLMs for building a more powerful semi-supervised learning approach. Specifically, we propose a joint semi-supervised learning process that can progressively build a standard $K$-way classifier and a matching network for the input text and the Class Semantic Representation (CSR). The CSR will be initialized from the given labeled sentences and progressively updated through the training process. By means of extensive experiments, we show that our method can not only bring remarkable improvement to baselines, but also overall be more stable, and achieves state-of-the-art performance in semi-supervised text classification. | ['Anonymous'] | 2022-01-16 | null | null | null | acl-arr-january-2022-1 | ['semi-supervised-text-classification-1'] | ['natural-language-processing'] | [ 2.98654288e-01 3.80927682e-01 -6.51776373e-01 -1.02027690e+00
-1.09193301e+00 -3.60690624e-01 8.68672132e-01 3.33938122e-01
-5.30183494e-01 5.15273929e-01 1.32643014e-01 -3.06847245e-01
3.49046856e-01 -5.96221030e-01 -4.97339517e-01 -3.80370915e-01
3.73021662e-01 8.83751631e-01 2.92238295e-01 4.56012562e-02
9.24834535e-02 -4.75833490e-02 -1.54083645e+00 4.90865648e-01
9.32663023e-01 1.03072894e+00 1.63247660e-01 3.59985948e-01
-6.86718881e-01 9.82007027e-01 -3.92507285e-01 -4.13720101e-01
-1.38415605e-01 -3.44580293e-01 -1.38383114e+00 4.81763780e-01
2.94008255e-01 1.15906835e-01 -2.05459118e-01 8.80555451e-01
1.88516781e-01 1.66272640e-01 7.97433138e-01 -1.19223678e+00
-1.08396783e-01 1.16162205e+00 -4.35304433e-01 -3.62618655e-01
1.66981190e-01 -4.64793950e-01 1.16293132e+00 -1.05703807e+00
5.23520350e-01 1.29941881e+00 6.41645491e-01 5.13795912e-01
-1.18687487e+00 -7.63993382e-01 4.01338279e-01 -8.82483795e-02
-1.38205314e+00 -5.35758436e-01 9.99595225e-01 -2.54432172e-01
8.01088452e-01 1.98435970e-02 1.15786314e-01 8.07586253e-01
-2.65983135e-01 1.24193633e+00 1.05802727e+00 -1.03906941e+00
2.19408184e-01 8.25218797e-01 7.37063706e-01 8.39836001e-01
-3.28573734e-02 -4.78037179e-01 -5.78880191e-01 -5.60928404e-01
1.65369883e-01 -6.22490421e-02 3.05810161e-02 -6.16926789e-01
-9.27408576e-01 9.42481160e-01 1.53282791e-01 4.17319238e-01
7.05577731e-02 -7.15583265e-02 5.18352628e-01 1.88834533e-01
9.00669813e-01 2.80820400e-01 -7.29671597e-01 -1.61404479e-02
-1.37520838e+00 -8.04432333e-02 9.38220859e-01 1.03646421e+00
9.05375183e-01 -2.49400914e-01 3.80684473e-02 1.26473379e+00
6.17752314e-01 2.42597982e-01 6.60898149e-01 -5.30351996e-01
6.37241125e-01 9.81369317e-01 -8.80235061e-02 -4.07432109e-01
-3.68953913e-01 -3.75944793e-01 -7.19187677e-01 -3.16699982e-01
2.16391191e-01 -7.81268626e-02 -7.85056055e-01 1.48530328e+00
2.43464604e-01 -1.44116014e-01 1.72310740e-01 2.25553662e-01
7.34936357e-01 6.19667470e-01 3.26815099e-01 -3.12954217e-01
1.30964518e+00 -1.47480834e+00 -6.67946219e-01 -4.70006734e-01
1.28297305e+00 -7.10930705e-01 1.11667502e+00 2.71127403e-01
-9.82481122e-01 -5.96667290e-01 -8.16863060e-01 5.81574365e-02
-5.36572099e-01 5.87750673e-01 6.88786268e-01 7.19309747e-01
-1.00588143e+00 4.72597569e-01 -8.41907084e-01 -5.29837370e-01
4.95016307e-01 3.13568264e-01 -3.00805122e-01 -1.52590409e-01
-1.18851709e+00 7.42712021e-01 7.05556750e-01 -2.74111658e-01
-6.59839511e-01 -4.19099987e-01 -9.71838832e-01 -2.11380441e-02
4.26717728e-01 -3.19958478e-01 1.51771748e+00 -1.08666623e+00
-1.51623046e+00 1.24897552e+00 -6.02932215e-01 -4.90022719e-01
2.87005156e-01 -1.96215123e-01 -2.27096409e-01 7.10357577e-02
-4.57023643e-02 8.03243876e-01 8.75123739e-01 -1.42715013e+00
-7.05237150e-01 -3.03340822e-01 -2.67152846e-01 2.70880342e-01
-8.44362855e-01 5.63665450e-01 -6.55069649e-01 -7.35046089e-01
2.52358288e-01 -9.26620364e-01 -4.57932532e-01 -1.10145934e-01
-5.24296522e-01 -6.81328058e-01 8.37946177e-01 -3.62658352e-01
1.42866981e+00 -2.07625914e+00 -1.15460962e-01 2.20098421e-01
1.25007797e-02 4.99810994e-01 5.88851757e-02 3.75931740e-01
-2.25027382e-01 5.74811921e-02 -4.51396972e-01 -1.10889935e+00
-1.22378310e-02 8.99080113e-02 -4.43060100e-01 2.31165648e-01
3.98620740e-02 8.86014640e-01 -7.64658988e-01 -9.09071684e-01
-1.22138076e-02 2.12709412e-01 -2.75378525e-01 3.48950863e-01
-4.75722253e-01 -1.39744943e-02 -3.36526215e-01 2.90479064e-01
4.11793172e-01 -6.33440614e-01 3.90885502e-01 1.38776809e-01
2.03254312e-01 6.67108119e-01 -1.23795915e+00 1.80075324e+00
-5.28446257e-01 5.82770884e-01 -5.86350895e-02 -1.42665040e+00
1.16751206e+00 3.83456081e-01 2.94632256e-01 -2.11183295e-01
1.20121367e-01 2.28642285e-01 -5.44353485e-01 -4.31009978e-02
5.51814139e-01 -2.98415106e-02 -1.22485176e-01 9.66591001e-01
4.37479705e-01 -1.54854268e-01 1.02789909e-01 4.65067208e-01
5.38780808e-01 1.48916259e-01 3.25334579e-01 -4.90927666e-01
7.74684012e-01 1.69079855e-01 1.91626385e-01 7.49384820e-01
7.51499832e-02 3.61319989e-01 2.00349331e-01 -3.23158413e-01
-6.86862528e-01 -6.01274312e-01 -3.57229978e-01 1.55597496e+00
-1.23116985e-01 -6.99411988e-01 -9.45059061e-01 -1.21119630e+00
-1.40976503e-01 7.64657676e-01 -4.16779995e-01 -1.90047532e-01
-3.59849125e-01 -8.82133543e-01 6.39400184e-01 6.67566597e-01
6.23927057e-01 -7.39982367e-01 1.07468918e-01 1.79423362e-01
-2.19062433e-01 -1.13670051e+00 -3.71879280e-01 5.61722219e-01
-1.09130335e+00 -7.24336505e-01 -6.61321402e-01 -1.30714726e+00
1.04258871e+00 3.68931115e-01 1.11715138e+00 8.96943361e-02
1.23133019e-01 1.58290833e-01 -5.48341155e-01 -3.54073048e-01
-6.44995749e-01 5.16669393e-01 5.58417700e-02 5.16236052e-02
4.74479795e-01 -1.83926478e-01 -1.18242413e-01 4.82256144e-01
-7.97695398e-01 2.10175738e-01 3.27273488e-01 8.27679873e-01
4.79834408e-01 3.23741704e-01 5.86897910e-01 -1.54021525e+00
3.17086220e-01 -2.03137323e-01 -3.24849188e-01 3.52390438e-01
-1.06324470e+00 2.73571819e-01 7.43126035e-01 -3.93239170e-01
-1.43454945e+00 3.51786941e-01 -2.13897288e-01 3.35291028e-02
-3.67191970e-01 5.72351813e-01 -2.25727186e-01 8.74705426e-03
7.35857010e-01 3.03912580e-01 -1.59603819e-01 -6.80939376e-01
4.46073472e-01 1.41349661e+00 9.01444480e-02 -4.95254010e-01
8.89363825e-01 6.12098396e-01 -5.77159941e-01 -6.56957209e-01
-1.47012842e+00 -9.41181898e-01 -1.08920527e+00 1.96551293e-01
4.44958836e-01 -1.20764399e+00 1.09632090e-01 5.36698222e-01
-1.00052822e+00 -6.05092347e-01 -1.33389771e-01 4.02799934e-01
-4.58201379e-01 5.60366213e-01 -6.82069659e-01 -7.35601127e-01
-4.47660565e-01 -6.95359170e-01 1.19010055e+00 -5.99286295e-02
-3.03092033e-01 -1.43755722e+00 1.25560416e-02 6.80000424e-01
1.94761738e-01 -6.34797454e-01 8.15161705e-01 -1.35914242e+00
5.27876429e-02 -4.77791190e-01 -2.40136847e-01 4.70535576e-01
1.00893311e-01 -2.58734554e-01 -1.19749105e+00 -4.31105584e-01
-2.28904095e-02 -7.60076225e-01 1.02542162e+00 1.63100082e-02
1.18589604e+00 -9.90753919e-02 -7.03618467e-01 3.37448865e-01
1.07494462e+00 -7.20482767e-02 4.42060024e-01 3.39217544e-01
6.82510316e-01 8.07410657e-01 8.44505310e-01 2.25373849e-01
7.09294915e-01 6.10885918e-01 -2.73658663e-01 -3.23893487e-01
-1.32344421e-02 -3.96447301e-01 3.27110231e-01 1.05774140e+00
4.44679528e-01 -1.50248945e-01 -1.05304623e+00 3.47217768e-01
-1.91667855e+00 -5.03511310e-01 -1.37403244e-02 2.12013912e+00
1.30719864e+00 3.97314668e-01 1.01662345e-01 3.37728351e-01
6.37356997e-01 2.31144696e-01 -1.90503448e-01 -1.06018588e-01
-5.09185232e-02 2.81104356e-01 3.49512935e-01 5.31930804e-01
-1.44428623e+00 1.41476357e+00 6.66843224e+00 1.04000044e+00
-8.72101128e-01 2.19130024e-01 6.01966262e-01 2.93878824e-01
-1.31825089e-01 9.85932201e-02 -1.30448806e+00 1.07029930e-01
1.16354585e+00 -2.77084317e-02 -2.52658725e-02 1.13148010e+00
-6.58156676e-03 -2.27505658e-02 -1.13204682e+00 8.23700368e-01
4.77387607e-01 -1.15946758e+00 3.24787080e-01 -2.70674318e-01
6.98223710e-01 4.91618142e-02 -3.62620294e-01 5.35805166e-01
5.42674303e-01 -6.84204936e-01 5.28066635e-01 6.33568838e-02
6.77916646e-01 -5.53753555e-01 8.27770114e-01 8.36328089e-01
-1.17556274e+00 1.23015553e-01 -3.66171956e-01 9.96093750e-02
-7.30569437e-02 6.20185971e-01 -8.27385545e-01 6.20596528e-01
4.90129322e-01 8.18614662e-01 -7.90440798e-01 7.00073898e-01
-3.32915574e-01 1.08841670e+00 -2.16977030e-01 -1.39648363e-01
2.07884237e-01 1.67312250e-01 -1.02956347e-01 1.64646101e+00
-3.33171695e-01 -1.55671909e-01 5.52997708e-01 4.58228171e-01
-2.60832816e-01 4.76872891e-01 -3.28492790e-01 -1.46186888e-01
2.77764350e-01 1.15288091e+00 -9.15527403e-01 -9.04949486e-01
-5.52050471e-01 9.09991205e-01 6.03165686e-01 2.17447758e-01
-3.93822968e-01 -5.95014036e-01 -7.19355606e-03 -3.71437632e-02
2.41905317e-01 -1.66822404e-01 -3.56204242e-01 -1.46611965e+00
-7.26389959e-02 -6.68705344e-01 5.58232784e-01 -5.35353661e-01
-1.13085854e+00 6.95963621e-01 4.20078821e-02 -1.08947015e+00
-4.50175077e-01 -5.12166262e-01 -5.15759230e-01 7.24785805e-01
-1.79195917e+00 -1.27351689e+00 -3.82877551e-02 5.79490304e-01
9.85632241e-01 -1.75272644e-01 1.13609290e+00 9.38484147e-02
-5.49760580e-01 8.52730513e-01 2.24827215e-01 4.62508053e-01
9.39232171e-01 -1.26009512e+00 4.42930102e-01 6.10241354e-01
4.90990043e-01 5.69464624e-01 3.98814827e-01 -5.59244514e-01
-9.61257815e-01 -1.29776466e+00 1.34302378e+00 -5.57033181e-01
7.36582696e-01 -7.43328750e-01 -1.09743619e+00 8.53427768e-01
7.20922649e-02 6.39403388e-02 1.03023994e+00 3.33047450e-01
-6.67921364e-01 4.78456803e-02 -8.29891205e-01 3.15851361e-01
7.78111696e-01 -8.32224309e-01 -8.89165640e-01 6.73876822e-01
7.75779963e-01 -2.07703352e-01 -7.08700240e-01 1.18046165e-01
3.31577033e-01 -3.53640109e-01 7.13472664e-01 -7.69562125e-01
1.68162242e-01 -4.75419126e-02 -4.10442706e-03 -1.19249821e+00
1.79629177e-02 -5.83553374e-01 -1.97048653e-02 1.58546364e+00
8.10615957e-01 -5.12014389e-01 1.00951838e+00 6.24342084e-01
3.43235359e-02 -5.35864711e-01 -5.27878284e-01 -8.18398535e-01
2.26704523e-01 -6.47227943e-01 1.11891337e-01 1.17515552e+00
7.04897046e-01 6.61150694e-01 -2.26402283e-01 -7.54876211e-02
8.24859619e-01 2.07115978e-01 6.88533247e-01 -1.55281186e+00
-1.74898282e-01 -1.84099585e-01 3.21405903e-02 -1.39895773e+00
7.65582323e-01 -1.17347777e+00 4.01004642e-01 -1.42143273e+00
5.84447324e-01 -7.51934350e-01 -2.03813091e-01 7.38169134e-01
-4.13960606e-01 1.45986415e-02 -2.20188841e-01 4.49811637e-01
-1.09248638e+00 5.18261611e-01 7.10486174e-01 -3.19350027e-02
-3.95972103e-01 2.59974182e-01 -6.49002731e-01 9.04531538e-01
8.62630725e-01 -6.91161096e-01 -3.79291683e-01 -3.20760369e-01
-1.74330935e-01 -1.50408596e-01 -1.51670605e-01 -6.58670008e-01
5.70669055e-01 6.39397576e-02 1.70543715e-02 -5.33696115e-01
2.60182191e-03 -7.27510750e-01 -5.35287917e-01 2.06486091e-01
-9.30006027e-01 -5.34430861e-01 8.74238685e-02 5.54455936e-01
-4.86165255e-01 -5.46284378e-01 8.39607239e-01 -1.28332332e-01
-4.78850812e-01 1.83070034e-01 -4.36541617e-01 6.72893450e-02
6.50811434e-01 1.08235307e-01 -1.88735098e-01 -2.32140943e-01
-7.10460663e-01 4.83209401e-01 3.36672783e-01 2.67341077e-01
1.80932745e-01 -1.03865635e+00 -4.36542481e-01 1.29937187e-01
4.06949401e-01 1.92435235e-01 -2.18259782e-01 4.30428922e-01
4.49666120e-02 7.73038447e-01 5.15427828e-01 -5.28954685e-01
-1.29811502e+00 4.70479369e-01 1.26384556e-01 -6.13698363e-01
-3.43848199e-01 7.54543543e-01 -9.03995801e-03 -9.86378551e-01
7.83110380e-01 -1.89814270e-01 -2.91867077e-01 7.89904594e-02
5.59035540e-01 -1.03305250e-01 1.28002778e-01 -6.31644785e-01
-3.08105201e-01 3.47413063e-01 -4.61731315e-01 -1.67886972e-01
1.23036122e+00 -2.38892853e-01 -1.06172428e-01 7.12296844e-01
1.31998718e+00 -2.40458965e-01 -8.95156324e-01 -1.08973145e+00
5.19979060e-01 3.87232937e-02 2.97654003e-01 -6.47161901e-01
-6.87856555e-01 9.19015169e-01 1.63065776e-01 2.18562752e-01
8.06184173e-01 4.34245914e-01 6.32458329e-01 9.22369003e-01
3.67004097e-01 -1.35066545e+00 1.90030798e-01 5.93384683e-01
2.67051965e-01 -1.53611112e+00 5.70011996e-02 -8.19972336e-01
-6.88934922e-01 1.06202054e+00 4.77763832e-01 1.38586938e-01
1.00455201e+00 2.21053526e-01 2.72683501e-01 -9.93478149e-02
-9.23024118e-01 -2.60233551e-01 3.48218918e-01 2.90833682e-01
7.94619262e-01 -1.69504419e-01 -1.05858212e-02 7.33709097e-01
-9.04486328e-02 2.77588405e-02 1.06246941e-01 1.21084297e+00
-7.84004331e-01 -1.46018720e+00 -6.95619509e-02 5.25545120e-01
-2.75499851e-01 -2.73095936e-01 -5.11666059e-01 5.32978356e-01
-3.39674503e-01 1.03493059e+00 9.80265141e-02 -2.47199237e-01
2.10125417e-01 7.15073943e-01 7.12775858e-04 -1.22042739e+00
-5.40637016e-01 1.41288742e-01 3.53001922e-01 -2.13643923e-01
-7.68199325e-01 -7.88352907e-01 -1.30324769e+00 2.51198739e-01
-8.94501328e-01 5.87036669e-01 6.68839991e-01 1.25054145e+00
1.14084259e-01 3.25748399e-02 9.56460536e-01 -5.78564703e-01
-6.87771559e-01 -1.19117415e+00 -5.59922099e-01 2.83410668e-01
-6.14965968e-02 -4.23703015e-01 -5.84763765e-01 4.90748256e-01] | [10.597535133361816, 7.3415303230285645] |
caf2b564-cc87-498f-b868-ab3345069d9c | augmenting-data-driven-models-for-energy | 2301.01720 | null | https://arxiv.org/abs/2301.01720v1 | https://arxiv.org/pdf/2301.01720v1.pdf | Augmenting data-driven models for energy systems through feature engineering: A Python framework for feature engineering | Data-driven modeling is an approach in energy systems modeling that has been gaining popularity. In data-driven modeling, machine learning methods such as linear regression, neural networks or decision-tree based methods are being applied. While these methods do not require domain knowledge, they are sensitive to data quality. Therefore, improving data quality in a dataset is beneficial for creating machine learning-based models. The improvement of data quality can be implemented through preprocessing methods. A selected type of preprocessing is feature engineering, which focuses on evaluating and improving the quality of certain features inside the dataset. Feature engineering methods include methods such as feature creation, feature expansion, or feature selection. In this work, a Python framework containing different feature engineering methods is presented. This framework contains different methods for feature creation, expansion and selection; in addition, methods for transforming or filtering data are implemented. The implementation of the framework is based on the Python library scikit-learn. The framework is demonstrated on a case study of a use case from energy demand prediction. A data-driven model is created including selected feature engineering methods. The results show an improvement in prediction accuracy through the engineered features. | ['Sandra Wilfling'] | 2023-01-04 | null | null | null | null | ['feature-engineering'] | ['methodology'] | [-1.10598609e-01 -1.90527484e-01 -3.00439090e-01 -7.88346052e-01
-1.71185583e-01 -2.13665888e-01 6.17775500e-01 6.59326613e-01
-8.69814213e-03 4.03563470e-01 1.71161070e-01 -1.25819102e-01
-4.65665817e-01 -1.45153892e+00 -3.53041172e-01 -4.77288991e-01
2.43019447e-01 1.91283360e-01 5.47307059e-02 -2.46589422e-01
5.76316416e-01 5.92065215e-01 -2.25414014e+00 3.93513501e-01
1.00835299e+00 1.00239217e+00 4.16016616e-02 2.65670091e-01
-7.30966270e-01 8.34833741e-01 -2.44361445e-01 4.15743053e-01
2.15874985e-01 -2.72816569e-01 -4.71357614e-01 -1.96103364e-01
-3.79246265e-01 -1.55150756e-01 4.23046768e-01 6.03300273e-01
4.45179313e-01 1.61629587e-01 5.38030267e-01 -1.74608290e+00
-2.70878345e-01 5.12274563e-01 1.81455165e-01 -2.09597081e-01
4.91091430e-01 -1.07202448e-01 5.03164232e-01 -1.20602810e+00
3.58213633e-01 8.88487935e-01 5.69916010e-01 2.93644905e-01
-1.12543857e+00 -6.17657602e-01 -2.51227111e-01 7.01216638e-01
-1.22117651e+00 -1.94112837e-01 1.09067047e+00 -9.52895522e-01
1.37602711e+00 6.05231345e-01 1.10191619e+00 3.44179392e-01
1.90423682e-01 4.94882435e-01 1.21958947e+00 -8.30436945e-01
8.66952240e-01 6.87687993e-01 6.45201862e-01 7.98011795e-02
3.29812676e-01 4.67404872e-01 -6.44530058e-01 -3.30218136e-01
1.61909405e-02 2.48069510e-01 2.50924796e-01 -5.21560490e-01
-6.03481293e-01 1.02701855e+00 1.97091982e-01 6.32219732e-01
-5.97788751e-01 -2.18366146e-01 4.43655491e-01 2.60872304e-01
5.17797112e-01 3.32153738e-01 -8.94104064e-01 -2.54544497e-01
-1.06167591e+00 2.33331427e-01 8.91792178e-01 9.19252992e-01
1.17510962e+00 1.22757234e-01 -1.53845981e-01 4.39042032e-01
6.59843683e-01 2.66371012e-01 6.58015907e-01 -4.25673693e-01
-6.52176216e-02 1.69848156e+00 -3.74115398e-03 -8.90802562e-01
-7.09618092e-01 1.83253735e-02 -4.94449377e-01 5.12249112e-01
-2.94973999e-01 -1.62295718e-02 -1.04936135e+00 1.20976889e+00
4.94855434e-01 -3.46733421e-01 1.54196218e-01 4.83589500e-01
1.27584136e+00 8.09855103e-01 2.94666916e-01 -4.28519636e-01
1.15347803e+00 -5.21438360e-01 -9.11454976e-01 4.24571306e-01
7.40252852e-01 -5.79266727e-01 8.06276381e-01 6.69313252e-01
-7.20051944e-01 -6.89678431e-01 -9.08787489e-01 1.28626421e-01
-1.25852323e+00 -1.35215029e-01 4.84366506e-01 9.50946093e-01
-6.03466332e-01 6.71935678e-01 -8.77145231e-01 -7.16341674e-01
7.79291391e-02 4.97289956e-01 -1.38367966e-01 4.30636913e-01
-1.09523356e+00 1.13607275e+00 8.50598335e-01 -2.64604181e-01
-6.28515482e-01 -1.17734253e+00 -7.58011937e-01 2.80089766e-01
6.61944896e-02 -5.84987879e-01 8.94072056e-01 -9.94909048e-01
-1.26855099e+00 1.98947012e-01 1.15990965e-02 -4.13855821e-01
1.38145238e-01 -1.72333360e-01 -8.94202471e-01 -5.65647125e-01
-2.84857601e-01 5.24044372e-02 5.70314586e-01 -1.07460213e+00
-7.85595715e-01 -4.37682986e-01 -4.22174156e-01 -1.93317652e-01
-6.50411844e-01 1.54936075e-01 3.34066689e-01 -3.85864854e-01
-1.21642193e-02 -4.20666188e-01 -1.69676825e-01 -2.35075355e-01
-1.89201683e-01 -3.58128369e-01 1.33161163e+00 -6.21120811e-01
1.60215902e+00 -1.86769331e+00 -4.50748056e-01 7.39307523e-01
-2.36897111e-01 2.42278934e-01 5.27693927e-01 9.43193018e-01
-1.45246670e-01 3.41448069e-01 -1.86472535e-01 4.10831302e-01
1.06300965e-01 1.34072050e-01 3.06989640e-01 -1.42548725e-01
2.32950747e-01 4.97480124e-01 -4.30784792e-01 -2.21832335e-01
9.50684786e-01 7.85684943e-01 -4.74215776e-01 2.92094499e-01
-3.90473634e-01 3.31247568e-01 -5.62147796e-01 5.05644381e-01
7.19721913e-01 1.00575924e-01 -2.88750846e-02 -5.67084908e-01
-6.03871703e-01 2.70797342e-01 -1.50754380e+00 1.40132165e+00
-5.42835891e-01 1.29943117e-01 -4.02925372e-01 -9.08069849e-01
1.48593175e+00 3.38958144e-01 9.44045186e-01 -6.82591617e-01
1.87798351e-01 3.18167418e-01 -5.17256781e-02 -8.75535548e-01
3.81267607e-01 1.97100326e-01 3.15080494e-01 4.50488985e-01
-1.20270491e-01 -2.71310925e-01 4.05731916e-01 -2.05049306e-01
8.17113996e-01 4.92120594e-01 7.07745433e-01 -4.06759351e-01
7.62839556e-01 4.99171734e-01 7.11849570e-01 -5.75892888e-02
2.74521589e-01 1.89116091e-01 -4.92100492e-02 -7.85219133e-01
-1.15601146e+00 -3.81841421e-01 -2.76249260e-01 9.55563843e-01
-3.32040071e-01 -9.36579585e-01 -6.79882765e-01 -5.47481477e-01
1.45560056e-01 1.31289887e+00 -5.58446467e-01 -4.65897560e-01
-1.92658126e-01 -8.98185551e-01 -2.89160639e-01 4.64744627e-01
5.68515360e-01 -1.19687307e+00 -1.03972685e+00 4.27385807e-01
3.70206326e-01 -2.66608983e-01 3.74170154e-01 4.75086153e-01
-1.15916467e+00 -1.14732313e+00 3.25337112e-01 -4.60424632e-01
7.79428542e-01 -1.41441941e-01 1.27046108e+00 1.23084508e-01
-4.29808259e-01 2.47783571e-01 -8.92243862e-01 -1.02189875e+00
-7.66026378e-01 1.63925067e-01 -2.49725252e-01 -1.60288870e-01
1.06190419e+00 -4.45185483e-01 -5.62605023e-01 2.32933998e-01
-1.03941631e+00 2.00807855e-01 4.94976759e-01 5.32292902e-01
7.69844174e-01 5.47259033e-01 6.12413228e-01 -7.65563190e-01
5.17216623e-01 -5.28904676e-01 -7.42262065e-01 3.24616790e-01
-1.37652123e+00 1.68212444e-01 5.75627983e-01 -1.27074048e-02
-1.01351428e+00 6.61921918e-01 -2.31647074e-01 3.57702188e-02
-4.77874577e-01 8.23984146e-01 -4.79588181e-01 1.28489405e-01
6.23696446e-01 2.09146142e-01 2.02269316e-01 -9.31006968e-01
3.27940881e-02 8.17878783e-01 -4.79984462e-01 -2.30302185e-01
6.29401505e-01 -5.79443462e-02 -1.09953340e-02 -6.26218081e-01
3.07317134e-02 -3.08072120e-01 -7.20285237e-01 -3.02376151e-01
4.97242808e-01 -4.92361516e-01 -4.15570140e-01 2.03010425e-01
-6.24693334e-01 -1.50011219e-02 -7.48319268e-01 5.05477428e-01
-2.97800511e-01 -5.17131865e-01 3.58978510e-01 -8.18028033e-01
-8.32488775e-01 -1.10912120e+00 4.05347168e-01 5.05652070e-01
-2.99846292e-01 -8.61285627e-01 1.55401289e-01 -1.18635073e-01
8.55469823e-01 5.02919853e-01 1.26484025e+00 -1.16473460e+00
-1.79699153e-01 -3.88438553e-01 3.61327529e-01 4.32976753e-01
6.03375256e-01 7.05748320e-01 -7.87942588e-01 -9.14500505e-02
1.16948206e-02 2.11175829e-01 1.57427475e-01 3.05450201e-01
1.22747195e+00 -2.76925594e-01 -4.40511227e-01 9.03465450e-02
1.58604956e+00 7.15164781e-01 5.94044626e-01 6.30606353e-01
2.96499014e-01 6.67397380e-01 9.78186131e-01 7.33185232e-01
4.25017983e-01 6.40121579e-01 6.32560313e-01 -1.26725331e-01
1.20487131e-01 -7.13583529e-02 1.06059633e-01 5.01564324e-01
3.46947424e-02 4.15094525e-01 -9.86670315e-01 2.94073671e-01
-1.88411295e+00 -9.75765526e-01 -7.31919706e-01 2.23770761e+00
7.14283228e-01 -1.76093176e-01 4.87187594e-01 7.25609064e-01
3.54824126e-01 -6.56518638e-01 -6.26355588e-01 -9.44487870e-01
2.70895362e-01 1.98800594e-01 1.91866755e-01 1.62422597e-01
-9.37096953e-01 2.58291066e-01 5.54445648e+00 2.12926090e-01
-1.22084749e+00 5.80071006e-03 2.42021009e-01 2.51501519e-02
-4.51484323e-01 1.14105664e-01 -7.69866705e-01 6.42071962e-01
9.98007059e-01 -7.06357181e-01 2.10781872e-01 1.36476433e+00
7.47416854e-01 -2.89417148e-01 -1.13595009e+00 6.78881049e-01
-3.15136671e-01 -1.35729587e+00 -2.64642877e-03 2.20826101e-02
5.77266574e-01 -3.13117415e-01 -5.85863531e-01 1.55487657e-01
4.50798459e-02 -8.01116347e-01 3.73646885e-01 9.32996988e-01
1.18304335e-01 -1.01379430e+00 8.60155642e-01 3.61480325e-01
-1.08832002e+00 -2.70610988e-01 -5.74566908e-02 -4.42749076e-03
-2.06138939e-01 1.09705329e+00 -8.35469246e-01 7.99453557e-01
1.09674120e+00 6.63904369e-01 -8.30286324e-01 1.40182447e+00
2.09609717e-01 5.95723212e-01 -3.42424184e-01 -5.59992678e-02
-6.52328551e-01 -1.82296872e-01 1.63287252e-01 1.16587687e+00
6.13056123e-01 -1.66651502e-01 2.33025979e-02 9.16489780e-01
6.71363473e-01 7.11720765e-01 -5.54328680e-01 1.23534024e-01
4.77902085e-01 1.54849958e+00 -3.47122580e-01 -2.52239913e-01
-5.05482554e-01 1.45085558e-01 -3.79262954e-01 -7.09397048e-02
-5.95394015e-01 -6.05952799e-01 4.97571528e-01 5.04910171e-01
1.86754495e-01 7.26929456e-02 -4.72070903e-01 -5.97787559e-01
-8.56304392e-02 -8.36005628e-01 2.31395826e-01 -5.67930043e-01
-1.25131667e+00 4.55798894e-01 1.64059579e-01 -1.42477977e+00
-3.74868870e-01 -4.15671140e-01 -6.69987857e-01 8.35715711e-01
-1.05030584e+00 -1.10958111e+00 -8.58931661e-01 6.23881936e-01
4.27939534e-01 -3.69802833e-01 1.18148398e+00 5.45950651e-01
-5.43821573e-01 -1.32570211e-02 1.88051730e-01 -2.61108339e-01
3.69816273e-01 -1.23787701e+00 2.46258304e-02 6.45373166e-01
-3.49512458e-01 5.25162160e-01 6.77643299e-01 -7.79755712e-01
-1.70590067e+00 -1.17682123e+00 8.65214944e-01 7.18399510e-02
1.45995930e-01 -3.55409175e-01 -1.01150131e+00 3.46962541e-01
9.97973680e-02 1.38125643e-01 1.01588082e+00 -2.27527425e-01
1.12147704e-01 -6.58337474e-01 -1.83399022e+00 -1.93259776e-01
2.34255999e-01 8.90419334e-02 -3.67769897e-01 1.65646836e-01
1.13451026e-01 7.90554956e-02 -1.39717948e+00 7.27006257e-01
4.76440758e-01 -1.01305163e+00 6.49210572e-01 -4.27489191e-01
7.19450563e-02 -7.31455445e-01 -9.46328118e-02 -1.46746302e+00
-2.10048988e-01 -2.13018045e-01 -5.80019414e-01 1.82488823e+00
5.87445676e-01 -5.44091821e-01 7.28644431e-01 1.21309006e+00
8.13484192e-02 -6.09594584e-01 -5.07764459e-01 -7.13159323e-01
-3.68796661e-02 -5.54453969e-01 1.29260516e+00 8.38916600e-01
1.79202020e-01 -1.78220626e-02 1.32335082e-01 -2.25212455e-01
2.25800946e-01 -6.14572763e-02 8.84942830e-01 -1.72559083e+00
1.62502393e-01 -2.83433288e-01 -5.93703270e-01 3.73229712e-01
-4.97977257e-01 -8.62365603e-01 -3.51786882e-01 -2.11030412e+00
1.52250081e-01 -6.33992970e-01 -3.34708989e-01 7.30844200e-01
1.41290009e-01 -2.62884885e-01 8.27071518e-02 1.02946214e-01
2.73657352e-01 4.25566822e-01 5.82012117e-01 -9.62341949e-02
-7.12214589e-01 9.68610123e-02 -4.58291978e-01 5.31522334e-01
1.25089335e+00 -6.44820571e-01 -4.49099213e-01 1.78297967e-01
3.79227370e-01 -6.34309590e-01 2.30867982e-01 -1.09722912e+00
1.73048511e-01 -3.63410771e-01 5.95721424e-01 -8.77254009e-01
3.83100472e-04 -1.65879285e+00 1.00085628e+00 8.00560236e-01
-7.70810097e-02 -2.96139140e-02 2.44583532e-01 -6.75742179e-02
-1.56103194e-01 -3.37465942e-01 5.30826032e-01 3.65499966e-02
-9.88688827e-01 2.93446034e-02 -4.46122944e-01 -6.91264093e-01
1.65793431e+00 -5.37920296e-01 -1.31860450e-01 1.33416533e-01
-9.79713380e-01 1.47797793e-01 3.10421467e-01 6.20156050e-01
4.95203525e-01 -1.44802213e+00 -5.71795166e-01 5.59877992e-01
4.34808254e-01 -2.33410358e-01 -2.40310371e-01 6.44604445e-01
-3.99377018e-01 3.16543311e-01 -3.72916490e-01 -4.49335396e-01
-1.45201468e+00 7.40508199e-01 5.47386110e-01 -9.62387118e-03
-4.05067861e-01 -2.34050583e-03 -6.24290168e-01 -7.06321418e-01
-7.36682266e-02 -4.41442847e-01 -7.09016502e-01 4.57817495e-01
5.93717277e-01 9.52507377e-01 7.57757843e-01 -4.75614280e-01
-6.46800041e-01 6.76075578e-01 3.19977283e-01 3.41877550e-01
1.73460519e+00 6.53380752e-02 -2.22428918e-01 5.14346421e-01
8.50987554e-01 -2.47899696e-01 -6.46742523e-01 1.04674019e-01
4.47416812e-01 -4.60024744e-01 3.62586409e-01 -1.35127664e+00
-1.21940947e+00 5.27800858e-01 1.21735096e+00 7.36917734e-01
1.55143106e+00 -3.94395441e-01 1.31015673e-01 1.16393663e-01
5.23201644e-01 -1.35294533e+00 -5.90706646e-01 8.64137411e-02
1.03856206e+00 -1.33842862e+00 2.05882102e-01 -4.00414914e-01
-2.49172583e-01 1.48075294e+00 6.99322641e-01 1.09826490e-01
1.36396801e+00 5.93893051e-01 -1.70529261e-01 -2.48435766e-01
-6.51736557e-01 -1.67183116e-01 3.04926753e-01 6.32521212e-01
7.40993738e-01 2.09317766e-02 -9.53607738e-01 8.51436198e-01
-2.51448721e-01 7.63558507e-01 1.94082946e-01 1.33357382e+00
-5.33113539e-01 -1.47307324e+00 -5.88970304e-01 8.59458506e-01
-1.73888683e-01 1.03184238e-01 -5.42997062e-01 7.69206643e-01
5.01167715e-01 1.45119095e+00 1.18119247e-01 -5.63187838e-01
8.27890396e-01 3.81438106e-01 1.62510097e-01 -6.16745591e-01
-1.21400344e+00 -3.49230617e-01 -1.72819458e-02 -4.76563096e-01
-4.77422684e-01 -7.62838185e-01 -1.42469716e+00 -3.65431547e-01
-4.80768651e-01 3.43625814e-01 1.50172520e+00 7.98295200e-01
7.08914280e-01 7.19629169e-01 9.38046694e-01 -4.97574419e-01
-1.68293603e-02 -9.78372455e-01 -2.74292022e-01 2.72760808e-01
-4.58399281e-02 -6.35499895e-01 -2.09161431e-01 2.11581439e-01] | [8.394818305969238, 4.756916522979736] |
288f834b-7b2b-4c54-9b01-c7c4a49bdbe2 | todynet-temporal-dynamic-graph-neural-network | 2304.05078 | null | https://arxiv.org/abs/2304.05078v1 | https://arxiv.org/pdf/2304.05078v1.pdf | TodyNet: Temporal Dynamic Graph Neural Network for Multivariate Time Series Classification | Multivariate time series classification (MTSC) is an important data mining task, which can be effectively solved by popular deep learning technology. Unfortunately, the existing deep learning-based methods neglect the hidden dependencies in different dimensions and also rarely consider the unique dynamic features of time series, which lack sufficient feature extraction capability to obtain satisfactory classification accuracy. To address this problem, we propose a novel temporal dynamic graph neural network (TodyNet) that can extract hidden spatio-temporal dependencies without undefined graph structure. It enables information flow among isolated but implicit interdependent variables and captures the associations between different time slots by dynamic graph mechanism, which further improves the classification performance of the model. Meanwhile, the hierarchical representations of graphs cannot be learned due to the limitation of GNNs. Thus, we also design a temporal graph pooling layer to obtain a global graph-level representation for graph learning with learnable temporal parameters. The dynamic graph, graph information propagation, and temporal convolution are jointly learned in an end-to-end framework. The experiments on 26 UEA benchmark datasets illustrate that the proposed TodyNet outperforms existing deep learning-based methods in the MTSC tasks. | ['Jun Gu', 'Yong Cui', 'Hongzhi Wang', 'Zhiyu Liang', 'Donghua Yang', 'Xianzhang Liu', 'Huaiyuan Liu'] | 2023-04-11 | null | null | null | null | ['time-series-classification'] | ['time-series'] | [-3.90004605e-01 -4.58206773e-01 -1.17495313e-01 -1.80451468e-01
3.83060202e-02 -2.50869364e-01 2.51831859e-01 1.78268492e-01
-1.40545890e-01 3.73573899e-01 -5.97910769e-02 -6.10521913e-01
-6.29049420e-01 -8.82916987e-01 -3.84591877e-01 -8.95085096e-01
-7.10730672e-01 7.99579695e-02 3.46178591e-01 -2.74226040e-01
-3.05470020e-01 3.81859660e-01 -8.17410409e-01 1.32669210e-01
7.79735208e-01 1.12657022e+00 1.69454873e-01 4.56080556e-01
-3.31971616e-01 9.42570031e-01 -5.38720429e-01 1.30033595e-02
-1.50427178e-01 -4.21848804e-01 -3.54078233e-01 5.37585355e-02
-2.66140312e-01 -7.57567659e-02 -9.25079107e-01 8.39403749e-01
3.84729654e-01 2.41733521e-01 3.14284474e-01 -1.35094023e+00
-8.99725139e-01 6.26508474e-01 -6.13129914e-01 7.01690018e-01
-1.29445195e-01 1.09411262e-01 1.00614500e+00 -2.29879990e-01
3.13212186e-01 1.27166700e+00 4.53518152e-01 2.03791946e-01
-8.47287953e-01 -9.15372849e-01 1.04057467e+00 4.62631762e-01
-1.27739322e+00 8.06391835e-02 1.18915808e+00 -2.88822234e-01
1.07949102e+00 6.84642941e-02 8.41150761e-01 1.01959658e+00
3.62633824e-01 6.50654912e-01 6.50033414e-01 1.67803884e-01
1.17214993e-02 -6.66924238e-01 3.02651644e-01 1.06319141e+00
7.18818307e-02 -8.39284214e-04 -3.18612427e-01 1.75711721e-01
8.20827961e-01 6.15731835e-01 -3.50170434e-01 -7.72810057e-02
-1.28504634e+00 6.24017894e-01 9.57370400e-01 4.50576246e-01
-2.95435041e-01 4.67597842e-01 7.68832207e-01 6.02659166e-01
5.97826898e-01 -7.68405348e-02 -5.04426777e-01 4.40705419e-02
-2.94596463e-01 -2.97353953e-01 4.04642642e-01 1.10003507e+00
6.12513185e-01 5.37619770e-01 -1.26569882e-01 4.03145194e-01
2.23419711e-01 3.72295707e-01 5.54706216e-01 -1.68545306e-01
9.27717507e-01 1.01798236e+00 -4.51399028e-01 -1.63113642e+00
-8.92317533e-01 -8.17316413e-01 -1.21198440e+00 -4.61924285e-01
-5.26506081e-02 -3.32507253e-01 -9.27970111e-01 1.59775317e+00
1.57358631e-01 6.84563994e-01 7.98626244e-03 9.61390793e-01
8.33217263e-01 7.83714712e-01 8.17023739e-02 -4.44196820e-01
1.07218659e+00 -9.52182114e-01 -1.10820723e+00 -1.43583477e-01
8.01626205e-01 -2.44870365e-01 7.53489912e-01 7.97004402e-02
-2.80813664e-01 -4.83469099e-01 -1.20277154e+00 -3.27680372e-02
-4.96579051e-01 -1.27093047e-01 1.32978082e+00 2.05094025e-01
-8.27329695e-01 5.26792586e-01 -1.19505823e+00 -1.39589742e-01
5.17516673e-01 4.72331882e-01 -2.63672829e-01 -7.02347457e-02
-1.39938211e+00 2.37963170e-01 4.97123331e-01 7.73724437e-01
-6.69355869e-01 -1.65229529e-01 -8.61364365e-01 2.20328882e-01
6.09752357e-01 -4.15173233e-01 6.48086011e-01 -7.92961121e-01
-1.22321641e+00 8.78552347e-02 -5.91459386e-02 -4.68760937e-01
1.38457730e-01 2.28641033e-01 -9.44672883e-01 1.25064582e-01
-2.47271266e-02 -1.96807757e-01 8.24392438e-01 -6.49125218e-01
-5.56579649e-01 -6.21861577e-01 2.51618307e-02 9.94261056e-02
-5.51014006e-01 -2.78313190e-01 -7.60090172e-01 -8.13235104e-01
3.72153699e-01 -8.18012655e-01 -3.39758962e-01 -4.93737847e-01
-2.48851150e-01 -4.09387708e-01 1.24052310e+00 -5.73535204e-01
1.79788864e+00 -2.29576945e+00 2.65687913e-01 2.16361731e-01
7.01013029e-01 4.70385933e-03 -1.68932542e-01 3.86799157e-01
-1.56618789e-01 6.27657995e-02 -2.42019575e-02 -2.36449763e-01
-2.71472991e-01 3.64926010e-01 -2.19183981e-01 4.80782688e-01
5.65306805e-02 1.14228582e+00 -1.07363379e+00 -4.13122654e-01
1.44182220e-01 3.45460534e-01 -1.41034052e-01 1.48599669e-02
-3.09854388e-01 4.94748861e-01 -1.09032595e+00 3.98509055e-01
5.52564800e-01 -7.01672554e-01 3.94384176e-01 -1.90663680e-01
1.16579431e-04 8.21078047e-02 -7.24035800e-01 1.78702438e+00
-3.67870241e-01 5.36038697e-01 -3.25747937e-01 -1.49906564e+00
1.03851151e+00 4.75464284e-01 6.97395265e-01 -1.00722611e+00
2.72889107e-01 9.02389660e-02 2.32029393e-01 -5.46268404e-01
-1.07206330e-01 9.16331857e-02 -2.47126177e-01 2.48346314e-01
1.98373809e-01 6.69176042e-01 -3.06263641e-02 2.05869824e-01
1.24922192e+00 -2.64420360e-01 -9.50798094e-02 -7.04476908e-02
4.45780814e-01 -2.74790645e-01 9.03877199e-01 2.18317777e-01
-2.46022597e-01 1.67727306e-01 8.14600587e-01 -6.84659421e-01
-3.35338116e-01 -7.61696994e-01 4.45315719e-01 8.48554552e-01
3.07191283e-01 -6.01090550e-01 -1.45787641e-01 -9.60540235e-01
-1.38245057e-02 1.01675630e-01 -7.47129381e-01 -5.46177685e-01
-6.76506400e-01 -8.28881145e-01 2.00808629e-01 5.90072870e-01
5.96631944e-01 -1.00145113e+00 -5.40453866e-02 6.57237828e-01
-2.17778683e-01 -1.43134451e+00 -6.92056596e-01 1.44778445e-01
-1.04641235e+00 -1.13042474e+00 -4.11422521e-01 -9.67782974e-01
4.41417515e-01 5.98104477e-01 6.74273431e-01 3.55900645e-01
-2.97145639e-02 1.72396451e-01 -5.49474478e-01 -2.28705958e-01
3.60792339e-01 3.88462692e-01 -2.95393348e-01 3.76921326e-01
4.36617941e-01 -1.04577279e+00 -7.19056547e-01 1.63724855e-01
-7.05865026e-01 1.88500240e-01 3.32365990e-01 7.53752589e-01
6.49816155e-01 6.06700838e-01 7.89202511e-01 -8.74828398e-01
6.90348685e-01 -6.79862142e-01 -6.81670368e-01 3.30495179e-01
-4.81013566e-01 5.53439697e-03 1.11599445e+00 -6.35113180e-01
-6.85003519e-01 -2.88535863e-01 3.15466702e-01 -9.18696940e-01
4.22160923e-01 1.17614961e+00 -2.97494650e-01 6.12435192e-02
1.24290816e-01 4.17614520e-01 -1.51480272e-01 -4.00849789e-01
8.81471783e-02 1.87339842e-01 2.45459512e-01 -3.35036635e-01
6.28030121e-01 4.87900168e-01 2.52789736e-01 -5.79051256e-01
-8.24503839e-01 -4.16586429e-01 -4.99885857e-01 -2.94821501e-01
9.84518707e-01 -9.97675776e-01 -9.36275423e-01 5.72497666e-01
-1.12965465e+00 -4.90321219e-01 2.72538304e-01 7.04045415e-01
-1.36353716e-01 3.13890874e-01 -6.51116014e-01 -5.82456350e-01
-4.04391974e-01 -8.35354924e-01 9.56990123e-01 2.86989748e-01
4.75878477e-01 -1.38933754e+00 -4.20854986e-01 -7.96149075e-02
2.78344333e-01 7.32767045e-01 1.16821563e+00 -2.96772033e-01
-8.90372515e-01 -2.83933848e-01 -3.12650353e-01 -3.41780782e-02
4.14585888e-01 -1.86056226e-01 -4.92195070e-01 -4.42665040e-01
-4.44497094e-02 2.23549768e-01 9.70023274e-01 2.91784942e-01
1.59492803e+00 -3.61479253e-01 -4.45419967e-01 9.16255414e-01
1.32014477e+00 4.80262548e-01 3.20864558e-01 1.06170036e-01
1.32470012e+00 4.31003809e-01 3.03793341e-01 4.81878072e-01
7.04082906e-01 3.37698698e-01 5.25441051e-01 -1.47100031e-01
6.95998445e-02 -2.45582417e-01 1.40562400e-01 1.39847064e+00
-2.57698596e-01 -4.62017626e-01 -9.12994504e-01 4.07045782e-01
-2.34014654e+00 -7.88849533e-01 -3.52923632e-01 1.69642293e+00
2.29278266e-01 4.60508913e-01 -7.46342093e-02 5.14478236e-02
6.41298056e-01 6.26996517e-01 -7.22477436e-01 5.73805422e-02
-6.88198805e-02 2.79778112e-02 5.23229122e-01 8.50573853e-02
-1.25680375e+00 9.38550413e-01 4.57943916e+00 6.79847896e-01
-1.54296529e+00 1.21653900e-01 4.57709461e-01 3.49127427e-02
-3.66863221e-01 -5.07975630e-02 -2.90840268e-01 6.46720350e-01
9.01497245e-01 -2.71389097e-01 6.39927268e-01 6.05173409e-01
4.37022030e-01 6.40221953e-01 -8.48037302e-01 1.23626077e+00
-2.86792189e-01 -1.30764449e+00 1.05823889e-01 9.43640918e-02
5.47974110e-01 6.73444793e-02 1.04331873e-01 6.61365926e-01
1.50208652e-01 -1.05643797e+00 3.03928077e-01 6.99045360e-01
6.11893773e-01 -8.37936640e-01 5.89564979e-01 1.67271078e-01
-1.94270086e+00 -3.54995996e-01 -3.64152730e-01 -2.26734892e-01
3.56098339e-02 5.13980389e-01 -3.37211668e-01 1.14072371e+00
6.31482720e-01 1.51653302e+00 -7.14339077e-01 7.74531245e-01
-9.08778086e-02 7.14123905e-01 -1.84390634e-01 -2.14505509e-01
6.06770396e-01 -3.56818765e-01 3.31583232e-01 8.96753609e-01
2.44277969e-01 2.20563442e-01 5.05302548e-01 5.76282084e-01
-9.57277864e-02 -2.19953451e-02 -5.56113839e-01 -5.52846432e-01
4.00130659e-01 1.09643126e+00 -9.12666261e-01 -2.43642018e-04
-6.90620244e-01 1.03785479e+00 6.08069181e-01 8.64664912e-01
-9.33887243e-01 -5.72832525e-01 3.73216361e-01 -1.86670199e-01
3.51993382e-01 -7.59675503e-01 4.44437489e-02 -1.42194045e+00
2.54618108e-01 -6.27946794e-01 7.87476540e-01 -3.93106937e-01
-1.29875541e+00 7.41417170e-01 -2.98492134e-01 -1.40977991e+00
5.33586517e-02 -4.77409899e-01 -9.37507868e-01 8.52096796e-01
-1.57234979e+00 -1.35154521e+00 -6.10978961e-01 1.12815881e+00
1.91868454e-01 -2.24046692e-01 6.14889681e-01 4.32692945e-01
-1.05901480e+00 5.00406444e-01 1.50264250e-02 4.59697753e-01
2.37915799e-01 -1.11504841e+00 5.06517768e-01 9.36587632e-01
-9.49226096e-02 5.95591128e-01 1.04951583e-01 -6.53419495e-01
-1.74284911e+00 -1.56526101e+00 5.41743219e-01 1.18199356e-01
1.05491102e+00 -5.64665139e-01 -1.12109458e+00 7.84438074e-01
-1.34422734e-01 5.77587247e-01 4.61335629e-01 3.46221298e-01
-3.58841926e-01 -5.08304536e-01 -2.53347933e-01 5.63205540e-01
1.30921185e+00 -7.88278937e-01 1.14897020e-05 4.87717539e-01
1.35609603e+00 -2.19702125e-01 -8.86304498e-01 4.21083331e-01
1.70878515e-01 -5.81088483e-01 6.85202301e-01 -8.59428048e-01
2.75248084e-02 -3.52090508e-01 1.45280197e-01 -1.17239773e+00
-4.35273796e-01 -6.99382126e-01 -4.33543444e-01 9.78068173e-01
3.06182504e-01 -1.08637226e+00 5.92369854e-01 2.27181703e-01
-3.42454433e-01 -8.54683518e-01 -8.24477136e-01 -9.48789656e-01
-2.66938031e-01 -4.31895316e-01 9.93419588e-01 1.26463985e+00
-8.56845006e-02 7.61065602e-01 -3.66684765e-01 3.91541213e-01
4.14928317e-01 5.70085824e-01 5.13799906e-01 -1.26195312e+00
-2.10870132e-01 -4.67679828e-01 -6.72621727e-01 -1.00026059e+00
1.88926205e-01 -1.08352220e+00 -3.80835891e-01 -1.69070148e+00
-3.31006587e-01 -5.78458607e-01 -8.51007998e-01 4.56215531e-01
-2.75975853e-01 -6.40087068e-01 -1.03001803e-01 1.95591599e-01
-8.57411861e-01 9.20021653e-01 1.75474191e+00 -2.22058699e-01
-2.37451702e-01 -1.37172297e-01 -4.09626544e-01 2.52150118e-01
8.86680007e-01 -2.99314559e-01 -9.07275915e-01 -7.73018837e-01
4.84437123e-02 5.23151815e-01 3.05007398e-01 -8.76088679e-01
4.49913353e-01 -2.06982478e-01 3.06760252e-01 -5.69801807e-01
1.65315252e-02 -8.91693532e-01 4.58368938e-03 3.76665652e-01
6.09641932e-02 3.01096648e-01 1.57412708e-01 1.16808379e+00
-3.89346391e-01 6.71674907e-01 2.24050194e-01 6.56793714e-02
-9.63276386e-01 1.24859416e+00 -1.58747379e-02 -2.27763839e-02
8.56248021e-01 1.39882520e-01 -3.95131379e-01 -2.98900515e-01
-7.76165366e-01 7.63146818e-01 -1.60235614e-01 7.56214797e-01
7.38583922e-01 -1.59347618e+00 -3.95526469e-01 2.64194727e-01
1.75216213e-01 1.74697980e-01 7.09078729e-01 1.02531672e+00
-3.69621933e-01 4.06807005e-01 8.29926599e-03 -5.16058028e-01
-8.09174418e-01 9.97414649e-01 4.99540985e-01 -4.45516944e-01
-1.01692307e+00 5.39869487e-01 3.38848144e-01 -1.58868611e-01
2.08012313e-01 -4.28190947e-01 -4.43279207e-01 -2.72371378e-02
1.18930355e-01 -2.39968896e-02 5.86389340e-02 -3.55573386e-01
-4.79733318e-01 3.78728032e-01 -8.17872286e-02 4.29058194e-01
1.28431284e+00 -2.12668449e-01 -1.51868701e-01 6.22951627e-01
1.63456070e+00 -5.15698731e-01 -1.14755094e+00 -4.35486495e-01
2.93303225e-02 -1.96557984e-01 3.05595189e-01 -1.74028218e-01
-1.67914248e+00 1.11736977e+00 3.46601605e-01 5.19021511e-01
1.35378337e+00 -2.47032925e-01 1.09426641e+00 3.66327047e-01
3.89319509e-01 -8.00120354e-01 2.22666398e-01 6.03700399e-01
7.93972909e-01 -1.04822540e+00 -2.39799067e-01 -4.32665229e-01
-4.30395514e-01 1.33638740e+00 7.17515767e-01 -1.88200340e-01
1.18578076e+00 -1.26373127e-01 -2.49665469e-01 -6.10640287e-01
-8.77690136e-01 -3.84842306e-01 3.43683720e-01 3.97441387e-01
1.91198871e-01 1.93318546e-01 -2.44791746e-01 9.44683552e-01
6.04338124e-02 -3.40683043e-01 1.01898059e-01 6.58304453e-01
2.06670444e-02 -6.50967956e-01 3.42621654e-01 5.53554177e-01
-2.28366360e-01 -3.58289629e-02 -4.79258895e-02 9.90853310e-01
-7.62059093e-02 9.26575243e-01 1.12411104e-01 -7.96669960e-01
3.78258020e-01 -3.96356940e-01 1.68808118e-01 -4.94912714e-01
-2.73322284e-01 3.41092855e-01 -1.17065847e-01 -6.20481789e-01
-3.25090855e-01 -3.84487182e-01 -1.58459496e+00 -2.26668686e-01
-2.74803281e-01 2.06803039e-01 8.94132629e-02 9.24419641e-01
4.89886701e-01 1.25153852e+00 9.75303590e-01 -3.33271593e-01
6.04565330e-02 -7.72689044e-01 -7.83056438e-01 4.25177991e-01
5.29894948e-01 -7.20347643e-01 -2.48309225e-01 -2.40194276e-01] | [6.805309295654297, 2.8297643661499023] |
e2cdc78e-e1dc-4e45-9d3b-f3f55f0d9572 | openbox-a-python-toolkit-for-generalized | 2304.13339 | null | https://arxiv.org/abs/2304.13339v1 | https://arxiv.org/pdf/2304.13339v1.pdf | OpenBox: A Python Toolkit for Generalized Black-box Optimization | Black-box optimization (BBO) has a broad range of applications, including automatic machine learning, experimental design, and database knob tuning. However, users still face challenges when applying BBO methods to their problems at hand with existing software packages in terms of applicability, performance, and efficiency. This paper presents OpenBox, an open-source BBO toolkit with improved usability. It implements user-friendly inferfaces and visualization for users to define and manage their tasks. The modular design behind OpenBox facilitates its flexible deployment in existing systems. Experimental results demonstrate the effectiveness and efficiency of OpenBox over existing systems. The source code of OpenBox is available at https://github.com/PKU-DAIR/open-box. | ['Bin Cui', 'Ce Zhang', 'Wentao Zhang', 'Yang Li', 'Yu Shen', 'Huaijun Jiang'] | 2023-04-26 | null | null | null | null | ['experimental-design'] | ['methodology'] | [-8.40581179e-01 -5.19346714e-01 -3.50671262e-01 -3.60084176e-01
-4.70748514e-01 -5.75838327e-01 -6.72748461e-02 2.47598030e-02
-8.68758932e-02 4.72410709e-01 6.36487603e-02 -5.95967829e-01
-9.13840681e-02 -3.15150380e-01 -2.61711031e-01 -5.11749864e-01
-2.77819131e-02 2.69863963e-01 5.81155494e-02 -3.84063125e-01
2.71037579e-01 4.34791803e-01 -1.70383132e+00 1.66270137e-01
7.61789382e-01 9.34132516e-01 3.48523259e-01 7.59162128e-01
2.13162377e-02 3.55256528e-01 -6.28096402e-01 -1.81771100e-01
4.30446744e-01 -2.00099293e-02 -7.03209519e-01 -4.99409795e-01
1.89618096e-01 -1.62638202e-01 6.86760768e-02 7.18491256e-01
9.81864691e-01 2.46285051e-01 -7.98497200e-02 -1.47826672e+00
-5.44569314e-01 4.95406426e-02 -2.01368988e-01 4.41309303e-01
5.05128920e-01 5.82900226e-01 1.27111936e+00 -1.11859584e+00
3.64430815e-01 9.68754768e-01 6.58237338e-01 5.89890242e-01
-1.45715034e+00 -7.89479077e-01 -7.70310238e-02 2.91614652e-01
-1.44325185e+00 -8.28652680e-01 4.02459025e-01 -4.91951078e-01
1.24207973e+00 7.09129512e-01 6.04037523e-01 6.28979445e-01
1.49878338e-02 7.48085380e-01 1.06514227e+00 -4.44949389e-01
2.27347061e-01 3.45146030e-01 4.37809378e-01 7.20714569e-01
1.95421040e-01 9.19722691e-02 -8.83543134e-01 -2.12260470e-01
3.78508508e-01 -2.61279106e-01 -2.20606983e-01 -6.47235751e-01
-1.16890228e+00 5.47986150e-01 3.92067164e-01 9.13305953e-02
-5.77294342e-02 2.62578666e-01 4.22514528e-01 2.90235579e-01
3.33248764e-01 8.57722163e-01 -7.79718220e-01 -4.93973374e-01
-2.90667176e-01 5.55345535e-01 1.17138112e+00 1.37055922e+00
6.42863035e-01 -1.56251922e-01 -9.18565877e-03 1.03563893e+00
2.40179181e-01 6.33014143e-02 5.72788894e-01 -8.26847374e-01
1.47012547e-01 5.16596973e-01 2.02686653e-01 -8.01580906e-01
-5.62517643e-01 -2.90071696e-01 -2.97062039e-01 2.93010175e-01
3.78549278e-01 -1.71507254e-01 -4.20048147e-01 1.23930740e+00
8.56611848e-01 -4.72095311e-01 -1.71796262e-01 1.19247949e+00
1.14695728e+00 5.01803517e-01 -1.36615336e-01 2.08285406e-01
1.58753777e+00 -1.43546665e+00 -8.37941885e-01 -3.34128231e-01
7.98582673e-01 -9.73159313e-01 1.46609116e+00 3.06761742e-01
-8.04049730e-01 -2.94239759e-01 -1.03976941e+00 -4.72637743e-01
-5.65764606e-01 9.29073617e-02 1.13349092e+00 7.49938965e-01
-8.12979162e-01 4.59567606e-01 -8.18642735e-01 -4.29147571e-01
5.37405670e-01 5.58952808e-01 -1.94887117e-01 1.23267956e-01
-8.34924102e-01 7.49665976e-01 4.21934098e-01 4.74345125e-02
-5.16736627e-01 -1.01902544e+00 -7.34790444e-01 2.00499594e-01
9.76446033e-01 -8.93529236e-01 1.62278080e+00 -5.70817649e-01
-1.61514008e+00 4.75019604e-01 -1.83071554e-01 8.75820816e-02
5.06453216e-01 -5.93973875e-01 -8.40825215e-02 -3.46067280e-01
9.09609944e-02 3.14224303e-01 3.53332460e-01 -6.01580143e-01
-4.04209793e-01 -3.29257578e-01 9.55696106e-02 4.91321027e-01
-1.85246885e-01 3.20467502e-01 -8.84103596e-01 -7.22678065e-01
-2.33896941e-01 -8.94689798e-01 -2.23393559e-01 2.25044295e-01
-2.68922448e-01 -2.69756109e-01 5.70193827e-01 -5.36865056e-01
1.57231808e+00 -2.29251289e+00 -1.34502292e-01 1.38587058e-01
2.40758955e-01 1.50179744e-01 1.78179801e-01 7.34492958e-01
-7.68442526e-02 1.73725933e-01 2.85727620e-01 6.44249767e-02
2.89967418e-01 -1.13082834e-01 2.19478503e-01 3.62848252e-01
-2.61377931e-01 7.69945204e-01 -7.30526149e-01 -3.46183181e-01
2.25964591e-01 -3.73641625e-02 -7.37861812e-01 2.67222673e-01
-2.75771797e-01 4.01660085e-01 -3.56597513e-01 9.50531185e-01
3.52339864e-01 -3.65290582e-01 3.17116767e-01 5.17159626e-02
-3.71719748e-01 5.74908435e-01 -1.53387296e+00 1.58218288e+00
-4.89377379e-01 7.09017932e-01 2.48998284e-01 -6.16881728e-01
6.21900678e-01 -4.54366356e-02 3.54391307e-01 -4.23161924e-01
1.54159695e-01 1.56636387e-01 1.73070669e-01 -6.74123228e-01
4.08712983e-01 4.31441903e-01 3.76529209e-02 5.26195526e-01
-3.51905748e-02 -6.72268495e-02 6.35950446e-01 -3.73819694e-02
9.52190936e-01 3.13239485e-01 7.46545255e-01 -4.65518028e-01
2.02431530e-01 2.11533189e-01 4.81772482e-01 6.05254769e-01
-2.34098241e-01 2.49215841e-01 5.06882608e-01 -4.46147442e-01
-9.51105952e-01 -5.78026414e-01 -4.70138043e-01 1.71311319e+00
-8.97923298e-03 -1.00911403e+00 -6.18469656e-01 -2.15951756e-01
3.92349660e-01 5.23217320e-01 -3.87461334e-01 3.22004855e-01
-3.20670664e-01 -6.50576711e-01 1.86677799e-01 3.99407923e-01
2.43921325e-01 -1.06379580e+00 -6.17150545e-01 -3.19967717e-02
-1.49042115e-01 -9.25839663e-01 -8.49452376e-01 1.26200676e-01
-7.43226171e-01 -1.03317845e+00 -2.69429564e-01 -5.41554689e-01
5.66551864e-01 4.31420714e-01 9.57188129e-01 1.27953574e-01
-8.49605083e-01 2.52870053e-01 -2.58658230e-01 -7.43161440e-01
-7.47683123e-02 4.27344620e-01 6.19064718e-02 -2.92692751e-01
4.38861012e-01 -1.80700228e-01 -6.61497951e-01 8.00002456e-01
-5.94533861e-01 2.96807677e-01 2.80871361e-01 9.24095988e-01
4.53356713e-01 -3.91090393e-01 3.18546057e-01 -7.96713829e-01
9.10475016e-01 -3.29409331e-01 -1.21687710e+00 2.21433207e-01
-1.04126561e+00 1.26878649e-01 1.62841618e-01 -1.22407787e-01
-7.28536248e-01 1.21784089e-02 -2.39918873e-01 -9.90483537e-02
-9.23910141e-02 5.55414736e-01 -3.36066097e-01 -1.15877740e-01
7.47964799e-01 -3.12072992e-01 3.47837508e-01 -7.80558825e-01
3.85519624e-01 8.72635782e-01 1.75765142e-01 -7.53885627e-01
3.62486005e-01 1.14612602e-01 -4.58752066e-01 -7.37376869e-01
-7.00387537e-01 -8.93868446e-01 -3.79855424e-01 -1.75128981e-01
4.57061410e-01 -7.00487554e-01 -1.31165195e+00 2.49430016e-01
-6.70680404e-01 -5.63030362e-01 1.14613764e-01 1.87483028e-01
-3.78652066e-01 -2.87130699e-02 -1.18308470e-01 -4.46489632e-01
-6.33579195e-01 -1.25637460e+00 6.55564964e-01 3.57843429e-01
-5.79577625e-01 -8.70636463e-01 9.44332965e-03 8.39470148e-01
3.54003876e-01 6.09882772e-02 5.38894951e-01 -5.98966300e-01
-6.51261449e-01 -3.82554024e-01 5.73733589e-03 1.37201354e-01
-3.69292349e-02 1.08955584e-01 -9.53000665e-01 -3.68603826e-01
-5.32662034e-01 -2.47724727e-01 2.36855209e-01 2.22716883e-01
1.34635961e+00 -3.65731895e-01 -4.80811983e-01 8.52940381e-01
1.29582739e+00 6.33945242e-02 2.10375771e-01 8.17447364e-01
4.99101520e-01 6.62515163e-01 8.61214161e-01 6.76134109e-01
2.87016392e-01 1.00844061e+00 3.14492464e-01 -3.35162841e-02
3.06846619e-01 9.94373634e-02 2.10354954e-01 5.60466409e-01
6.16097301e-02 1.05064109e-01 -1.27629411e+00 2.88685769e-01
-2.08334756e+00 -7.44093239e-01 -1.57166645e-01 2.20319271e+00
9.33374047e-01 -3.35153267e-02 3.78578037e-01 -2.05299690e-01
5.32727242e-01 -2.77133863e-02 -7.20566511e-01 -7.37402201e-01
3.22689325e-01 1.22204229e-01 4.43801790e-01 4.14132416e-01
-9.86115873e-01 8.33830237e-01 6.16675615e+00 7.70590782e-01
-1.16508794e+00 3.64270568e-01 4.96356398e-01 -7.35810637e-01
3.09116036e-01 4.60823327e-02 -1.05115116e+00 3.48237574e-01
7.25530326e-01 -5.08121490e-01 9.51784432e-01 1.35198259e+00
3.46437246e-01 -3.24924916e-01 -1.13405097e+00 9.34078038e-01
-2.91045934e-01 -1.61301303e+00 -7.87237942e-01 2.30619740e-02
3.90735328e-01 2.22355410e-01 -6.65688217e-02 2.74864733e-01
4.02211607e-01 -9.09921110e-01 7.82131493e-01 5.96748069e-02
7.11806655e-01 -4.81249332e-01 6.37856007e-01 1.67582437e-01
-9.18778777e-01 -1.84591785e-01 -1.48053810e-01 -4.08827901e-01
-2.54007071e-01 5.04360139e-01 -1.02241611e+00 4.48120624e-01
1.27811170e+00 2.87786037e-01 -6.39908016e-01 1.23277390e+00
-1.45831287e-01 3.57073933e-01 -4.17050034e-01 -4.93292809e-01
-2.00202093e-01 -2.38260627e-01 3.64120543e-01 1.17658567e+00
-1.29700050e-01 1.12455720e-02 4.38131779e-01 6.62741184e-01
1.24632247e-01 4.92458880e-01 -3.90590638e-01 -4.87927534e-02
6.94581747e-01 1.38295090e+00 -5.09951115e-01 -1.31429747e-01
-4.80806202e-01 3.99597526e-01 4.68739331e-01 2.56458402e-01
-7.63754547e-01 -8.68276954e-01 1.25625277e+00 3.60070840e-02
1.97952583e-01 -3.15752953e-01 -8.01274300e-01 -1.06409669e+00
3.03527504e-01 -1.25840485e+00 5.99134982e-01 -8.01001489e-01
-9.71018851e-01 3.02653849e-01 1.79194793e-01 -9.24623191e-01
1.30345762e-01 -8.44427466e-01 -4.47992176e-01 8.79709661e-01
-8.63366842e-01 -7.17940331e-01 -6.04029715e-01 2.37771869e-01
5.01564384e-01 -4.37986106e-02 7.85087347e-01 4.16186750e-01
-1.19280589e+00 5.05542696e-01 4.98864859e-01 1.49364680e-01
9.22193348e-01 -1.35025108e+00 6.39164567e-01 5.66540956e-01
-3.63369994e-02 1.01889515e+00 8.14966023e-01 -4.08389449e-01
-1.69699121e+00 -6.18277252e-01 5.23087919e-01 -5.96172333e-01
7.98417628e-01 -7.57969856e-01 -6.74256563e-01 5.70649266e-01
1.64342031e-01 1.66972846e-01 1.07876062e+00 6.09629989e-01
-1.67213470e-01 -5.15301228e-01 -8.17185223e-01 8.23539495e-01
8.90376449e-01 -1.90791056e-01 -1.34529144e-01 4.95009571e-01
3.06623518e-01 -9.56149518e-01 -9.17144001e-01 1.35051936e-01
9.97064710e-01 -1.01897812e+00 8.25460672e-01 -5.91215074e-01
5.97306527e-02 -3.25697005e-01 2.82163233e-01 -1.17423487e+00
-2.99636304e-01 -1.14444184e+00 -6.32626424e-03 9.86168385e-01
4.24684078e-01 -9.98455346e-01 5.92137575e-01 8.72720778e-01
-1.95428893e-01 -1.23520505e+00 -6.20517612e-01 -7.05886662e-01
-4.35497820e-01 -6.22343302e-01 9.80393767e-01 8.26417565e-01
4.58977401e-01 4.74418879e-01 -2.07616001e-01 2.59553999e-01
1.41118959e-01 3.72686446e-01 1.39528716e+00 -9.03306961e-01
-6.97079718e-01 -5.22074223e-01 -1.63269043e-01 -9.23282504e-01
-3.46725643e-01 -9.03809905e-01 -1.28066182e-01 -1.40626240e+00
-9.08075795e-02 -7.41509855e-01 6.85370062e-03 8.29685152e-01
-2.26283923e-01 2.84955740e-01 4.20275718e-01 4.66316968e-01
-5.22646844e-01 2.92993397e-01 1.04554152e+00 1.60397470e-01
-6.15600765e-01 1.18078604e-01 -8.73051882e-01 5.74303985e-01
1.19329321e+00 -6.49769008e-01 -1.79884825e-02 -3.21659327e-01
9.44803357e-02 -2.74349511e-01 1.21963747e-01 -7.59613156e-01
8.52893218e-02 -3.31104517e-01 1.35117754e-01 -1.53733402e-01
2.20232442e-01 -5.50242007e-01 2.63188779e-01 6.09548450e-01
-1.16207995e-01 4.41166848e-01 4.42971975e-01 6.82997778e-02
1.51134595e-01 -3.04260015e-01 6.66637957e-01 -1.06467478e-01
-7.52605438e-01 2.07312778e-02 -2.54724532e-01 3.66170593e-02
1.28450263e+00 -1.19203679e-01 -5.59207678e-01 -2.25036770e-01
-8.58313680e-01 6.70554340e-01 4.55569744e-01 5.38676620e-01
2.73319483e-01 -1.04354501e+00 -1.73060969e-01 3.20533991e-01
5.03658533e-01 -1.03106737e-01 -1.17078803e-01 1.04075384e+00
-1.18576777e+00 5.01582444e-01 -3.67312253e-01 -3.41200799e-01
-1.73398304e+00 6.02163732e-01 3.94336283e-01 -2.20968612e-02
-5.34823000e-01 6.57334924e-01 -6.74463660e-02 -5.76881528e-01
3.76990736e-01 -1.93126112e-01 2.11340785e-01 -2.11290717e-01
4.25084531e-01 6.05007231e-01 3.68530840e-01 -1.03918405e-03
-4.89815414e-01 1.29550129e-01 -4.70410436e-02 1.23226449e-01
1.25877333e+00 1.31477043e-01 -2.94049501e-01 4.80647951e-01
7.15689659e-01 -1.77410934e-02 -8.77783298e-01 -8.08870606e-03
7.05055743e-02 -7.69130588e-01 1.86486319e-02 -8.51213217e-01
-6.17019713e-01 7.41685569e-01 6.24724567e-01 8.18766207e-02
9.05255377e-01 -2.40468774e-02 3.20151508e-01 6.36607647e-01
2.08157614e-01 -1.11057723e+00 -4.49240766e-02 3.39616716e-01
1.07334518e+00 -1.34819496e+00 4.56058681e-01 -5.73212445e-01
-6.68773472e-01 1.15281653e+00 1.00639987e+00 2.48400763e-01
8.00863981e-01 3.52335662e-01 3.20625395e-01 -2.18786404e-01
-9.74109054e-01 -2.53059655e-01 2.81466454e-01 3.33329380e-01
6.49660468e-01 3.01645566e-02 -4.92443115e-01 5.69363296e-01
-4.72051591e-01 4.57789153e-02 3.30808640e-01 1.15165937e+00
-1.82019085e-01 -1.20687389e+00 -7.42097557e-01 4.50770319e-01
-4.50921059e-01 -1.43078417e-01 -2.85376549e-01 7.51309454e-01
-1.32258177e-01 7.25914240e-01 -1.54759869e-01 -1.96981788e-01
2.43658379e-01 1.13208398e-01 1.18133262e-01 -8.67224216e-01
-8.66245747e-01 1.84494436e-01 4.32578892e-01 -8.19657445e-01
2.13684127e-01 -5.55725574e-01 -8.71742427e-01 -5.27292669e-01
-4.72265542e-01 2.85116345e-01 1.01201391e+00 4.79165763e-01
9.77948129e-01 4.74230409e-01 2.89748162e-01 -7.29880512e-01
-6.28870726e-01 -9.07912731e-01 -1.52085498e-01 9.72972065e-02
1.86891854e-01 -7.31930614e-01 5.26598319e-02 2.91893911e-02] | [7.336239814758301, 4.401630878448486] |
592f7280-0229-44f8-962c-7b679b35536a | learning-contextually-fused-audio-visual | 2202.07428 | null | https://arxiv.org/abs/2202.07428v2 | https://arxiv.org/pdf/2202.07428v2.pdf | Learning Contextually Fused Audio-visual Representations for Audio-visual Speech Recognition | With the advance in self-supervised learning for audio and visual modalities, it has become possible to learn a robust audio-visual speech representation. This would be beneficial for improving the audio-visual speech recognition (AVSR) performance, as the multi-modal inputs contain more fruitful information in principle. In this paper, based on existing self-supervised representation learning methods for audio modality, we therefore propose an audio-visual representation learning approach. The proposed approach explores both the complementarity of audio-visual modalities and long-term context dependency using a transformer-based fusion module and a flexible masking strategy. After pre-training, the model is able to extract fused representations required by AVSR. Without loss of generality, it can be applied to single-modal tasks, e.g. audio/visual speech recognition by simply masking out one modality in the fusion module. The proposed pre-trained model is evaluated on speech recognition and lipreading tasks using one or two modalities, where the superiority is revealed. | ['Li-Rong Dai', 'Xin Fang', 'Ming-Hui Wu', 'Jian-Shu Zhang', 'Jie Zhang', 'Zi-Qiang Zhang'] | 2022-02-15 | null | null | null | null | ['lipreading', 'audio-visual-speech-recognition'] | ['computer-vision', 'speech'] | [ 6.51561916e-01 2.17536047e-01 -2.63428658e-01 -1.33452371e-01
-1.07730985e+00 -3.19841206e-01 9.61461306e-01 2.21670583e-01
-3.25163335e-01 6.84595823e-01 2.63978988e-01 -3.50876004e-01
-2.38375440e-01 -3.89026552e-01 -3.97656411e-01 -1.05366743e+00
1.98985711e-01 6.74766526e-02 1.06056154e-01 -1.37685224e-01
2.84960568e-01 5.47028065e-01 -2.57240295e+00 6.87167346e-01
7.51220405e-01 1.15297580e+00 5.63400209e-01 5.94451666e-01
-4.18963015e-01 7.90459692e-01 -3.23400050e-01 1.79773476e-02
-1.20054968e-01 -4.07568902e-01 -7.94280529e-01 2.78085142e-01
2.28024200e-01 1.63221583e-01 -9.44101776e-05 7.09286034e-01
5.87184846e-01 3.06672245e-01 7.99884796e-01 -1.08818054e+00
-1.63275227e-01 5.29993951e-01 -3.16656113e-01 5.90277947e-02
6.96945190e-01 -4.86551522e-04 1.00309801e+00 -9.80033696e-01
4.06814188e-01 1.19765759e+00 1.15500189e-01 6.35978341e-01
-1.27305007e+00 -3.05948883e-01 1.34648040e-01 6.28531933e-01
-1.30050778e+00 -9.94999707e-01 1.12967134e+00 -6.20002151e-01
9.16275799e-01 3.31002742e-01 5.62548459e-01 1.11560941e+00
-6.38541207e-02 9.63981032e-01 1.54299939e+00 -9.79750037e-01
2.88212806e-01 5.89842319e-01 -1.16360866e-01 3.12273949e-01
-3.06581467e-01 4.93598849e-01 -7.78264105e-01 -1.63420066e-02
2.15236545e-01 -8.79882351e-02 -2.34688237e-01 -5.95407128e-01
-9.00738657e-01 6.73880935e-01 1.95275858e-01 8.77604544e-01
-4.31552500e-01 -3.24879676e-01 4.47486699e-01 4.10175711e-01
3.82051378e-01 -1.03813134e-01 -1.52134866e-01 7.20960796e-02
-1.39836645e+00 -2.86514729e-01 3.61341059e-01 3.49271923e-01
7.58039832e-01 4.77422446e-01 -2.18968585e-01 1.08944082e+00
7.57258177e-01 4.39871907e-01 7.80212879e-01 -4.37357455e-01
2.70182103e-01 4.44627613e-01 -3.22567910e-01 -5.29684067e-01
-2.61433721e-01 -2.49372676e-01 -7.54436493e-01 7.10351348e-01
1.35236651e-01 3.18656266e-01 -1.00011945e+00 1.66680861e+00
2.75057703e-01 6.82068840e-02 5.28789282e-01 6.83028281e-01
1.14402878e+00 7.01902926e-01 2.19060212e-01 -6.60109997e-01
1.25418925e+00 -6.82617486e-01 -9.31631327e-01 -4.41609472e-02
3.29308361e-01 -9.74870741e-01 7.98559725e-01 3.95050496e-01
-9.86973882e-01 -8.29783261e-01 -1.03751528e+00 1.10504188e-01
-5.77963352e-01 4.51304942e-01 2.76934028e-01 7.41612375e-01
-1.09725845e+00 2.04050392e-01 -5.38140774e-01 -3.86089921e-01
2.17568055e-01 3.73466045e-01 -5.40527701e-01 -1.30842170e-02
-1.07017744e+00 9.26671207e-01 6.31378233e-01 -1.83321182e-02
-9.81353402e-01 -3.19383323e-01 -1.05008113e+00 1.44430026e-01
2.58550465e-01 -3.90635520e-01 9.06653464e-01 -1.32525694e+00
-1.71579671e+00 8.62773836e-01 -4.11487579e-01 -4.11024094e-01
2.58869827e-01 2.79522151e-01 -4.87415761e-01 4.95362550e-01
-3.22431833e-01 6.14790738e-01 1.60722554e+00 -1.61595500e+00
-4.45189744e-01 -4.28272277e-01 -1.42434791e-01 3.07080984e-01
-4.49953914e-01 1.55628845e-02 7.24043921e-02 -7.15370953e-01
5.81550077e-02 -5.99227309e-01 3.07749003e-01 -2.09604397e-01
3.69245536e-03 -2.24965513e-01 8.58906090e-01 -7.34960496e-01
9.93880749e-01 -2.34763622e+00 3.20037335e-01 2.06831634e-01
-1.91061974e-01 6.80614293e-01 -2.61400133e-01 6.03098154e-01
-4.01935577e-01 -3.66994619e-01 -2.09161416e-01 -6.68890834e-01
-2.67195940e-01 2.61437654e-01 -2.02988029e-01 2.79230028e-01
2.99825490e-01 5.33000469e-01 -5.63761592e-01 -8.45436454e-01
7.35159278e-01 9.61752117e-01 -1.06973074e-01 2.58806139e-01
-8.92006084e-02 5.58331728e-01 -1.25528008e-01 7.74715245e-01
6.65118635e-01 1.36051774e-01 1.35860160e-01 -2.35326722e-01
-2.31921747e-01 1.92856520e-01 -1.27005088e+00 1.80325305e+00
-7.85915971e-01 6.73296511e-01 3.14161450e-01 -1.27660954e+00
1.18624949e+00 8.54499817e-01 5.12687325e-01 -8.15250695e-01
1.38686746e-01 1.98501393e-01 -2.20400989e-01 -4.94536847e-01
3.93529713e-01 -4.59675550e-01 4.20194715e-01 1.96956247e-01
4.95041400e-01 1.59914643e-01 -2.59580128e-02 -4.63958271e-02
4.41611260e-01 1.07480220e-01 4.80075449e-01 4.80124839e-02
1.32469726e+00 -4.23304439e-01 6.08233027e-02 2.34105542e-01
2.19368655e-02 4.47041273e-01 -1.19700998e-01 1.57698900e-01
-5.73563695e-01 -1.06362009e+00 -1.18629344e-01 1.20465851e+00
-2.78479993e-01 -2.43729711e-01 -3.60552937e-01 -3.31007868e-01
-2.41252244e-01 6.02773309e-01 -2.92185485e-01 -2.01417223e-01
-9.66252536e-02 -8.82551968e-02 3.13027084e-01 3.21063995e-01
2.06389830e-01 -1.22239649e+00 -3.83259803e-01 6.85041696e-02
-1.52489617e-01 -9.04246807e-01 1.67108953e-01 2.69705743e-01
-9.01139081e-01 -8.38411570e-01 -1.08682752e+00 -9.39524710e-01
2.19419613e-01 4.26755488e-01 5.24069607e-01 -1.07646659e-01
-1.86229900e-01 8.61537337e-01 -5.33296824e-01 -2.25944251e-01
-8.30774307e-01 -6.80072606e-02 -1.36111593e-02 5.77316999e-01
4.80544195e-02 -7.22889543e-01 -1.01610258e-01 1.42673865e-01
-9.75205481e-01 -1.17397927e-01 7.54569232e-01 1.01804650e+00
5.04777610e-01 -1.26938134e-01 8.82282019e-01 -1.56798303e-01
4.23042595e-01 -2.25098193e-01 -5.34956157e-01 5.46711326e-01
-5.28411925e-01 4.89718579e-02 4.04639274e-01 -5.52119017e-01
-1.44782794e+00 4.46286887e-01 -3.04502785e-01 -6.47376537e-01
-5.62100351e-01 4.78651077e-01 -4.53338116e-01 -2.86865592e-01
5.20069718e-01 5.84569871e-01 3.43689114e-01 -6.41639531e-01
4.52360123e-01 1.00531840e+00 3.59229088e-01 -2.57992506e-01
4.97223139e-01 1.90128043e-01 1.98646542e-03 -1.35290861e+00
-6.01406321e-02 -7.46737957e-01 -6.33562982e-01 -4.70618963e-01
8.29411328e-01 -1.03956199e+00 -6.69437945e-01 4.66891408e-01
-1.07973814e+00 1.16343878e-01 -3.56984556e-01 7.24802256e-01
-8.13628852e-01 6.69255674e-01 -7.81657621e-02 -1.39726985e+00
-1.12501733e-01 -1.32353520e+00 1.04539263e+00 1.37965217e-01
2.12918878e-01 -9.82784271e-01 1.35904104e-01 5.86837590e-01
2.37909853e-01 -3.24300408e-01 6.59266531e-01 -8.26142490e-01
-3.41260374e-01 1.09045498e-01 -3.26125100e-02 6.83732212e-01
2.84931302e-01 -1.60738945e-01 -1.58710027e+00 -3.91784400e-01
-9.81207266e-02 -5.24781823e-01 1.15184414e+00 2.56604224e-01
7.65932918e-01 -1.69611230e-01 -1.16016112e-01 1.43387586e-01
1.28699279e+00 2.87611306e-01 6.17984116e-01 4.85028401e-02
3.02486867e-01 7.95607746e-01 6.32906199e-01 4.17826116e-01
1.34991920e-02 9.23652470e-01 4.32768285e-01 -1.06384128e-01
-4.65316951e-01 -1.45277560e-01 5.39924562e-01 8.15958560e-01
-1.03072844e-01 -1.47473618e-01 -7.09496021e-01 5.95388114e-01
-1.67289627e+00 -1.19517660e+00 4.31815833e-01 2.37370396e+00
4.98771280e-01 -1.48006782e-01 2.60851502e-01 8.48452449e-01
8.04736197e-01 3.84979844e-01 -1.56785607e-01 -5.18760920e-01
-3.13141286e-01 3.90389830e-01 -9.74357799e-02 6.57550991e-01
-1.07654643e+00 6.62594557e-01 5.74430275e+00 1.11645794e+00
-1.37312222e+00 1.60261080e-01 2.12557483e-02 1.49629191e-01
-3.80827397e-01 -9.47236717e-02 -5.14414310e-01 1.81483835e-01
8.76572669e-01 1.65276602e-01 4.67128456e-01 5.55215478e-01
2.14540541e-01 -1.48958862e-01 -9.04948115e-01 1.28386593e+00
4.17561114e-01 -1.03051603e+00 2.51912475e-01 2.32146382e-02
2.17556730e-01 -3.23317498e-01 1.85086027e-01 1.21017531e-01
-5.23384452e-01 -9.51139688e-01 5.35814285e-01 5.27950525e-01
7.73099661e-01 -5.75080812e-01 5.92132628e-01 3.72146070e-01
-1.35325360e+00 -2.48448521e-01 7.62105361e-02 2.85662234e-01
2.11030260e-01 1.86785102e-01 -8.92060697e-01 9.93743420e-01
4.67188567e-01 5.98130703e-01 -5.20172536e-01 1.05732942e+00
-1.18347839e-01 5.22786915e-01 -2.13593498e-01 1.43241256e-01
-8.19291696e-02 4.65163626e-02 7.37833917e-01 1.14587784e+00
4.06855464e-01 -2.18769908e-01 -7.79431686e-02 4.81968611e-01
3.98905396e-01 5.49620807e-01 -9.12439823e-01 -2.57555306e-01
2.40401596e-01 1.19063032e+00 -4.68686223e-01 -2.40818709e-01
-4.24949080e-01 5.85725307e-01 -4.24775518e-02 2.69078791e-01
-3.94613355e-01 -1.21140443e-01 3.69446188e-01 -1.51111782e-02
6.43278122e-01 1.47963595e-02 -5.68024479e-02 -1.03423858e+00
-3.17351334e-02 -9.17018950e-01 4.24208492e-01 -8.93252254e-01
-1.03051722e+00 8.51651609e-01 1.03216298e-01 -1.44731939e+00
-7.86112607e-01 -5.15548229e-01 -3.07400018e-01 8.45123470e-01
-1.82758725e+00 -1.52708066e+00 4.95190807e-02 1.02165318e+00
5.80014050e-01 -6.62606955e-01 8.94770026e-01 3.25196952e-01
-5.94700202e-02 6.64332867e-01 3.21965255e-02 -3.77693772e-01
6.45634711e-01 -8.08143258e-01 -7.19049990e-01 6.21456444e-01
4.36002344e-01 3.25349808e-01 6.21141970e-01 -3.85677427e-01
-1.20209420e+00 -6.20812476e-01 1.02269030e+00 2.86656529e-01
4.47750777e-01 -1.30648553e-01 -1.00429475e+00 2.04036713e-01
5.84647000e-01 -1.42686948e-01 9.09478307e-01 -2.07727887e-02
-4.57842380e-01 -3.33981693e-01 -1.11534071e+00 1.68695226e-01
5.35587668e-01 -1.01911926e+00 -9.76018310e-01 6.57758489e-02
4.38496768e-01 1.73890948e-01 -7.31886625e-01 3.46987128e-01
4.49628294e-01 -8.77273738e-01 1.13281286e+00 -3.28300923e-01
2.23023724e-02 -3.41736525e-01 -3.80457371e-01 -1.23629212e+00
1.75020069e-01 -4.65271443e-01 -8.39003772e-02 1.51491237e+00
2.90110677e-01 -5.90841353e-01 2.97302246e-01 -2.05240518e-01
-1.29565254e-01 -2.30987847e-01 -1.55964518e+00 -7.18892157e-01
-2.91648746e-01 -3.71407688e-01 2.22272396e-01 7.28390276e-01
2.57804453e-01 4.54569101e-01 -6.04653001e-01 -8.29925528e-04
4.50776607e-01 1.08259432e-01 4.73494351e-01 -1.43455172e+00
-4.64729875e-01 -4.06471014e-01 -6.07331157e-01 -7.54564047e-01
4.81848955e-01 -9.65500355e-01 -1.00802314e-02 -1.37605560e+00
-1.66642949e-01 -1.22697011e-01 -6.34275198e-01 6.14813864e-01
2.82468081e-01 1.06275119e-01 4.90934223e-01 1.19984552e-01
-2.85307050e-01 8.50428224e-01 9.49561834e-01 -3.64531010e-01
-4.56369996e-01 3.22368532e-01 -3.13919455e-01 3.99142683e-01
7.21096814e-01 -1.52532220e-01 -7.01194227e-01 9.78487879e-02
-2.64651537e-01 4.83194739e-01 5.37358999e-01 -9.93854225e-01
2.29209214e-01 1.40782535e-01 8.19430575e-02 -7.45516837e-01
9.48867083e-01 -9.56661820e-01 5.52138723e-02 1.92705423e-01
-4.94667321e-01 -4.13836449e-01 3.62305492e-01 5.53303361e-01
-6.84475541e-01 -3.75063777e-01 8.47738981e-01 4.75045107e-02
-7.55090654e-01 -3.79108280e-01 -7.80781150e-01 -6.15796685e-01
9.56737876e-01 -4.23558593e-01 -3.32400086e-03 -3.60698253e-01
-1.24635053e+00 -2.87187546e-01 9.96907204e-02 5.45434177e-01
9.42604005e-01 -1.30423498e+00 -4.50206369e-01 3.17782611e-01
4.43205029e-01 -7.14101493e-01 6.44975722e-01 9.47898567e-01
2.56679922e-01 5.53440154e-01 -5.09112835e-01 -8.16422880e-01
-1.81436396e+00 9.15782928e-01 1.48362264e-01 -2.12213118e-02
-2.09061623e-01 2.76809812e-01 -2.61031073e-02 -1.62468791e-01
5.14328420e-01 -3.49616185e-02 -9.01037574e-01 4.91504252e-01
6.18671358e-01 3.04921269e-01 2.77118355e-01 -1.14320850e+00
-4.33755040e-01 6.16281331e-01 1.75508589e-01 -4.84764189e-01
1.16255832e+00 -3.20107430e-01 1.46041527e-01 7.67253578e-01
1.16207087e+00 -3.28476392e-02 -7.34089732e-01 -3.93734097e-01
-1.73443273e-01 -2.61141658e-01 3.35168719e-01 -7.49432206e-01
-9.56875563e-01 1.36650991e+00 9.78299260e-01 2.04043478e-01
1.51623380e+00 1.36555120e-01 1.57354787e-01 1.54923037e-01
2.86194086e-01 -8.97227883e-01 5.09582609e-02 3.79470885e-02
1.01789129e+00 -1.27506149e+00 -1.03579715e-01 -3.87012392e-01
-7.13020802e-01 1.20380890e+00 1.42320931e-01 4.24969524e-01
7.01769650e-01 1.26115261e-02 1.15643121e-01 2.19690204e-01
-7.65390038e-01 -7.79049695e-01 7.05614686e-01 9.30051982e-01
3.85162771e-01 -6.04791082e-02 -4.29027826e-01 3.61195683e-01
3.23746592e-01 3.02940328e-02 1.59619704e-01 9.64636862e-01
-5.76250255e-01 -1.41743314e+00 -7.05090940e-01 -8.75709858e-03
-2.26930995e-02 -6.03570789e-02 -4.29437339e-01 5.94192326e-01
7.21924976e-02 1.36443675e+00 -2.47611046e-01 -4.49938893e-01
1.12734906e-01 5.58742106e-01 5.82883239e-01 -3.28578025e-01
-5.63140392e-01 4.61940616e-01 -1.13195209e-02 -2.88035572e-01
-1.10653150e+00 -8.40632617e-01 -9.14379179e-01 3.27174693e-01
-4.36490417e-01 1.08206898e-01 8.75192344e-01 1.03408968e+00
2.80952871e-01 4.23727453e-01 7.75844395e-01 -1.01357234e+00
-3.82047266e-01 -1.01051378e+00 -4.71359789e-01 1.00559339e-01
6.04959011e-01 -1.04319549e+00 -4.83907402e-01 1.14707105e-01] | [14.346601486206055, 5.110989570617676] |
c4890d8c-62cb-4982-bceb-b6ab4b345f97 | simple-and-effective-semi-supervised-question | 1804.00720 | null | http://arxiv.org/abs/1804.00720v1 | http://arxiv.org/pdf/1804.00720v1.pdf | Simple and Effective Semi-Supervised Question Answering | Recent success of deep learning models for the task of extractive Question
Answering (QA) is hinged on the availability of large annotated corpora.
However, large domain specific annotated corpora are limited and expensive to
construct. In this work, we envision a system where the end user specifies a
set of base documents and only a few labelled examples. Our system exploits the
document structure to create cloze-style questions from these base documents;
pre-trains a powerful neural network on the cloze style questions; and further
fine-tunes the model on the labeled examples. We evaluate our proposed system
across three diverse datasets from different domains, and find it to be highly
effective with very little labeled data. We attain more than 50% F1 score on
SQuAD and TriviaQA with less than a thousand labelled examples. We are also
releasing a set of 3.2M cloze-style questions for practitioners to use while
building QA systems. | ['Dheeraj Rajagopal', 'Danish Pruthi', 'Bhuwan Dhingra'] | 2018-04-02 | simple-and-effective-semi-supervised-question-1 | https://aclanthology.org/N18-2092 | https://aclanthology.org/N18-2092.pdf | naacl-2018-6 | ['triviaqa'] | ['miscellaneous'] | [ 4.19462845e-02 2.60312915e-01 1.67315885e-01 -7.39862025e-01
-1.54378307e+00 -1.03987980e+00 5.10149896e-01 -2.28250697e-02
-4.90146339e-01 7.69654453e-01 1.33465677e-01 -5.41198254e-01
-4.44718935e-02 -6.45209849e-01 -6.69416130e-01 -1.87581256e-01
1.98966458e-01 1.06786680e+00 3.09306711e-01 -5.89812100e-01
1.63013920e-01 1.25162005e-01 -1.05991042e+00 7.07202911e-01
1.20877314e+00 9.53827977e-01 2.36823693e-01 8.88925850e-01
-4.01201010e-01 1.02732885e+00 -1.01351821e+00 -8.09322059e-01
1.18022740e-01 -3.44499052e-01 -1.43675852e+00 3.03457323e-02
7.95645893e-01 -5.61001182e-01 -1.34405777e-01 7.39972949e-01
5.15108764e-01 3.17207910e-02 4.80927169e-01 -8.38185310e-01
-9.80736792e-01 6.71882093e-01 -7.06130415e-02 4.14441794e-01
3.48223090e-01 4.04340625e-01 1.40431976e+00 -8.12936306e-01
5.34735203e-01 1.00454473e+00 3.25416893e-01 9.49014485e-01
-8.59796941e-01 -4.38398242e-01 -1.48044720e-01 1.60310104e-01
-8.28741491e-01 -5.72010517e-01 6.58733130e-01 -1.83598936e-01
9.64113891e-01 2.77713984e-01 3.75470589e-03 9.99641001e-01
-2.86783218e-01 8.96670878e-01 8.85778785e-01 -5.30379653e-01
3.00318688e-01 7.08747134e-02 4.90285486e-01 5.85426509e-01
-1.65898234e-01 -4.33176875e-01 -4.51097339e-02 -2.87670106e-01
3.89311135e-01 -3.17334324e-01 -2.81259120e-01 -6.20587319e-02
-1.02129245e+00 1.28677261e+00 5.02394319e-01 2.85337865e-01
-3.05558443e-01 -2.73740310e-02 4.35493201e-01 6.56905353e-01
2.94773817e-01 1.18459415e+00 -8.35331678e-01 -1.60165578e-01
-7.78691292e-01 6.52957857e-01 1.04565942e+00 1.01121831e+00
5.34278691e-01 -3.37329477e-01 -3.26145560e-01 1.08735514e+00
-2.56585088e-02 5.29261589e-01 4.87183273e-01 -1.23588157e+00
1.02480674e+00 6.74668074e-01 3.94843012e-01 -6.60381079e-01
-2.77158529e-01 -3.29245657e-01 -5.34269452e-01 -4.28240687e-01
7.94917524e-01 -3.97669196e-01 -1.02147603e+00 1.63997817e+00
2.27326125e-01 -4.51755583e-01 2.72900522e-01 8.47502470e-01
1.14449966e+00 9.02447224e-01 1.21666573e-01 1.95048094e-01
1.46893752e+00 -1.30222654e+00 -6.03313446e-01 -6.78703129e-01
8.63718510e-01 -7.24226832e-01 1.73391449e+00 3.96042585e-01
-1.32353461e+00 -4.98407513e-01 -7.84622610e-01 -4.53930199e-01
-1.50634527e-01 1.59523338e-01 4.01873380e-01 5.11041939e-01
-9.76817131e-01 1.28825650e-01 -3.81823748e-01 -1.86516136e-01
3.51880670e-01 3.13382417e-01 -1.20116644e-01 -5.24199724e-01
-1.31083798e+00 9.71681476e-01 3.75667423e-01 -2.96529345e-02
-1.05778313e+00 -5.41948736e-01 -7.58140504e-01 1.04930870e-01
5.22951603e-01 -7.37921178e-01 1.76214695e+00 -7.80991316e-01
-1.41493309e+00 9.04896498e-01 -1.74307451e-02 -4.24783856e-01
2.92749882e-01 -5.35943568e-01 -4.14908141e-01 4.47714776e-01
2.18276709e-01 6.83276117e-01 6.18268669e-01 -9.25336719e-01
-7.12761760e-01 -1.66298151e-01 7.84318388e-01 2.72999853e-01
-3.25962424e-01 2.24028781e-01 -3.71159226e-01 -1.85432449e-01
-1.13544911e-01 -6.35312974e-01 -1.53390467e-01 -5.51254511e-01
-2.66867608e-01 -7.36313939e-01 4.50434357e-01 -8.21116865e-01
1.04333389e+00 -1.81513357e+00 -5.40432148e-02 -1.75281227e-01
3.44064295e-01 5.05584538e-01 -6.42650366e-01 3.10878664e-01
3.57048631e-01 8.00735950e-02 -2.27068216e-01 1.03781998e-01
1.80314019e-01 1.24410801e-01 -4.42502141e-01 -9.10070091e-02
5.59849501e-01 1.04028308e+00 -1.05362439e+00 -4.40330625e-01
-2.41463050e-01 -1.22960158e-01 -7.60052800e-01 7.78576076e-01
-8.75796199e-01 2.10417405e-01 -6.54362977e-01 5.14963329e-01
4.94908243e-01 -5.72761416e-01 5.59524260e-03 6.40374944e-02
4.42228526e-01 9.88266528e-01 -6.64180696e-01 1.72801721e+00
-5.86725652e-01 4.29743737e-01 -4.64097187e-02 -7.73615897e-01
9.40049112e-01 3.88160020e-01 -1.53391540e-01 -7.20698416e-01
1.35988668e-01 3.58150452e-01 1.08599536e-01 -9.22889173e-01
5.29028118e-01 -1.23168975e-01 -3.45454752e-01 6.64708197e-01
4.24150258e-01 -3.93974334e-01 5.80400348e-01 4.21944082e-01
1.43215811e+00 -2.95236081e-01 -3.41374800e-02 -1.21018343e-01
5.57811022e-01 3.25716883e-01 9.94513184e-02 7.95410573e-01
-1.84231356e-01 5.28347492e-01 4.65837389e-01 -5.28510511e-01
-1.23580825e+00 -9.44885910e-01 4.14768867e-02 1.65332031e+00
-2.71793753e-01 -2.48130232e-01 -9.43937004e-01 -1.13358426e+00
-3.66135120e-01 8.37722838e-01 -3.20323586e-01 6.00835606e-02
-8.64526808e-01 -4.29134786e-01 7.85762727e-01 4.39687014e-01
5.74783385e-01 -1.44796395e+00 -2.25819379e-01 2.11971298e-01
-5.19628763e-01 -1.13259029e+00 -4.32949334e-01 2.66739838e-02
-7.52560079e-01 -1.04344571e+00 -7.16788650e-01 -1.14508736e+00
5.13036907e-01 -3.88415083e-02 1.75303805e+00 2.45421723e-01
2.68046677e-01 8.70679691e-02 -6.11214221e-01 -1.99658796e-01
-6.32324815e-01 6.02635741e-01 -4.05522048e-01 -3.84072572e-01
8.73725533e-01 -1.58694535e-01 -4.61147010e-01 1.49901256e-01
-1.02676833e+00 -1.86257258e-01 6.27195835e-01 9.32851374e-01
1.32155463e-01 -3.57396692e-01 9.89805996e-01 -1.22493815e+00
1.07259548e+00 -6.08551264e-01 -5.87199211e-01 4.84703600e-01
-1.71212092e-01 1.39595628e-01 7.75844574e-01 -4.76132184e-01
-1.09362400e+00 -1.87236160e-01 -5.12413979e-01 1.08338490e-01
-3.56633872e-01 5.74023724e-01 -2.22122520e-01 3.04944068e-01
1.22618020e+00 -9.17854905e-02 -4.64703828e-01 -5.68367064e-01
7.38330007e-01 1.00894201e+00 7.02199638e-01 -9.27252054e-01
7.84627438e-01 -5.10009415e-02 -8.23482275e-01 -5.57656825e-01
-1.49371707e+00 -4.59931195e-01 -4.64369118e-01 2.44318992e-01
7.13385522e-01 -7.38276064e-01 -4.76092428e-01 9.70489234e-02
-1.29313779e+00 -4.97943461e-01 -9.98065993e-02 3.43320295e-02
-3.90839219e-01 1.82099715e-01 -9.26145852e-01 -4.93811220e-01
-6.91431224e-01 -9.22813714e-01 8.71579945e-01 2.43037105e-01
-3.25769067e-01 -8.99270594e-01 1.92311242e-01 1.15810478e+00
4.27749366e-01 -9.18133631e-02 1.11260116e+00 -1.16701770e+00
-5.86441278e-01 -2.74362236e-01 -2.03574404e-01 5.45247138e-01
-8.81784596e-03 -4.45551395e-01 -1.12471128e+00 -3.34379822e-01
1.40366063e-01 -1.17893887e+00 5.59196770e-01 -4.23523374e-02
1.08229959e+00 -4.32562739e-01 2.07292005e-01 2.43805781e-01
1.04363298e+00 1.15370028e-01 5.02149761e-01 4.40844655e-01
5.32592714e-01 7.47638345e-01 6.81325614e-01 -1.30296186e-01
4.76196378e-01 2.32171878e-01 2.02701464e-01 2.32566446e-01
7.17468783e-02 -2.19864130e-01 1.75614014e-01 1.00267267e+00
3.22450697e-01 -4.67334270e-01 -1.15184367e+00 9.51916397e-01
-1.34445739e+00 -6.71174228e-01 6.54321983e-02 1.69520426e+00
1.35465074e+00 2.92156011e-01 1.50176054e-02 -2.01443896e-01
5.36420941e-01 -3.28107402e-02 -5.65490782e-01 -4.15163308e-01
1.66931450e-01 6.03010952e-01 -8.27114806e-02 4.94162589e-01
-1.08711708e+00 1.19628716e+00 6.36575890e+00 5.94085932e-01
-8.39668155e-01 1.42085522e-01 7.03518271e-01 -5.06243259e-02
-4.02767360e-01 -1.23848535e-01 -8.44182312e-01 2.54053801e-01
1.32847130e+00 -2.39750370e-02 2.38884985e-01 9.99710381e-01
-2.81728506e-01 2.27471128e-01 -1.24308109e+00 6.32560194e-01
1.08993225e-01 -1.24024272e+00 1.12202592e-01 -2.49534160e-01
8.05577099e-01 3.09920698e-01 -5.94180031e-03 7.82350242e-01
7.04140782e-01 -1.19978762e+00 2.80227810e-01 -2.97937263e-02
8.03245902e-01 -8.13849986e-01 9.48338926e-01 6.64805114e-01
-3.43503475e-01 -1.37138754e-01 -7.71671295e-01 1.91306621e-02
7.65826479e-02 3.63219887e-01 -1.23030651e+00 1.36033624e-01
6.67202115e-01 -3.56744193e-02 -7.84662306e-01 8.54918301e-01
-6.30772412e-01 9.62490499e-01 -3.13876837e-01 -4.27014709e-01
6.80624485e-01 1.89278945e-01 -4.16724086e-02 9.98417914e-01
9.74713862e-02 3.26543570e-01 1.57190375e-02 8.40788662e-01
-6.17101610e-01 2.06742629e-01 -2.07023829e-01 -2.45432451e-01
5.86014450e-01 1.36513400e+00 -7.82128572e-02 -5.18796563e-01
-4.70514357e-01 9.54668641e-01 8.13121676e-01 2.88188398e-01
-5.48757911e-01 -7.15878844e-01 1.73968747e-01 -5.23473658e-02
2.22331211e-01 -1.27564665e-04 -1.26244664e-01 -1.32348454e+00
2.50110835e-01 -1.59190989e+00 6.47784710e-01 -9.52338457e-01
-1.69780171e+00 1.06432343e+00 -2.95112044e-01 -6.86642289e-01
-7.20923781e-01 -6.44777656e-01 -4.85468209e-01 9.87887740e-01
-1.52526343e+00 -1.02987635e+00 -2.75104612e-01 6.00538492e-01
7.47979164e-01 -2.63033986e-01 9.20795083e-01 3.59086454e-01
-2.07092866e-01 7.08785594e-01 1.12645455e-01 6.12703919e-01
8.48543227e-01 -1.52013123e+00 9.49485421e-01 5.06207168e-01
3.55571955e-01 8.34668338e-01 7.43766189e-01 -3.82977575e-01
-1.09716940e+00 -9.98304248e-01 1.28881323e+00 -1.18772042e+00
7.59885430e-01 -4.74164575e-01 -1.21870446e+00 9.24293637e-01
4.75647628e-01 -1.68910637e-01 6.86671317e-01 3.15863937e-01
-4.31591064e-01 6.43093809e-02 -1.25733423e+00 4.18518633e-01
5.99214911e-01 -5.79866052e-01 -1.32779098e+00 6.54221535e-01
1.08840346e+00 -5.29866397e-01 -7.80721247e-01 2.14169174e-01
3.01574506e-02 -5.39671779e-01 7.58719087e-01 -1.16169822e+00
7.11512506e-01 -1.25480667e-01 -2.03707889e-02 -1.30512881e+00
-2.24677593e-01 -4.47538704e-01 -3.46147157e-02 1.23570561e+00
8.59341741e-01 -2.60597318e-01 8.72780621e-01 8.19655836e-01
-1.41981184e-01 -6.87577486e-01 -5.48753798e-01 -4.92355764e-01
4.92031515e-01 -2.83204168e-01 6.88588381e-01 9.72068012e-01
-6.42089546e-02 1.04861522e+00 -1.26142606e-01 2.09058583e-01
2.48126477e-01 1.59870580e-01 9.36696947e-01 -1.02596498e+00
-4.93342072e-01 9.02432427e-02 2.22146854e-01 -1.47720242e+00
1.58280749e-02 -7.61983871e-01 1.27394989e-01 -1.51288211e+00
1.53184757e-01 -5.48902333e-01 -6.18203878e-02 5.34359753e-01
-4.56536442e-01 2.29886353e-01 -9.56359729e-02 5.64621203e-02
-1.01304471e+00 2.87710518e-01 1.24211335e+00 -2.07018033e-01
-5.10432422e-02 4.01704870e-02 -9.79414821e-01 6.28150284e-01
9.82341945e-01 -4.34795320e-01 -5.66931903e-01 -1.04809082e+00
3.45295668e-01 3.30042168e-02 -1.04167603e-01 -7.85799384e-01
4.05400842e-02 -6.58957362e-02 2.93158293e-01 -6.22566760e-01
2.11340874e-01 -3.92231733e-01 -7.69630373e-01 7.55947456e-02
-7.75862277e-01 8.22259486e-02 9.61413309e-02 7.90326744e-02
-3.30387503e-01 -7.05553710e-01 7.20406592e-01 -4.31894124e-01
-5.60804188e-01 2.28264749e-01 1.54628549e-02 8.47564995e-01
4.33854401e-01 3.94991279e-01 -6.75678253e-01 -7.03059614e-01
-4.82315868e-01 5.79043984e-01 1.83128014e-01 4.02761459e-01
4.20192748e-01 -1.20885193e+00 -9.52235878e-01 -9.95433256e-02
2.99566567e-01 3.73565257e-01 1.77817121e-01 3.80692855e-02
-9.09983099e-01 7.61139870e-01 1.59401596e-02 -3.65456402e-01
-9.62856472e-01 5.91817677e-01 3.16605538e-01 -6.00418270e-01
-3.01110536e-01 1.09210145e+00 1.18342705e-01 -9.17005837e-01
1.80972263e-01 -2.75505513e-01 -3.34600121e-01 -2.50072598e-01
8.87370110e-01 8.94435719e-02 1.52242988e-01 -2.19631881e-01
-2.81928740e-02 1.63544849e-01 -5.07563651e-01 -2.18368858e-01
1.40022528e+00 3.35977450e-02 -2.47468855e-02 5.85170053e-02
1.19195104e+00 -2.32734401e-02 -1.04226696e+00 -5.60955405e-01
3.53398561e-01 -1.58943817e-01 -3.34191501e-01 -1.24536908e+00
-6.98505700e-01 1.20900953e+00 1.81918427e-01 3.84498954e-01
1.08351052e+00 3.41850728e-01 1.30312276e+00 1.07679045e+00
2.18977287e-01 -9.38599169e-01 4.04102802e-01 8.85561347e-01
9.33364928e-01 -1.35621881e+00 -3.95829380e-01 -6.34853095e-02
-6.03064418e-01 8.96162748e-01 8.64995420e-01 -2.43491799e-01
-5.93650639e-02 -1.75087616e-01 5.85714340e-01 -5.23051202e-01
-9.15529191e-01 -7.46174380e-02 4.36836272e-01 5.11618018e-01
6.38012052e-01 -1.26353383e-01 -1.64320022e-01 8.35037231e-01
-6.62096024e-01 -1.06312715e-01 4.18086618e-01 7.39882469e-01
-8.62040699e-01 -1.08063042e+00 -3.16769302e-01 4.74462837e-01
-7.40323484e-01 -3.73639256e-01 -5.80622733e-01 7.64095545e-01
-2.46353656e-01 1.24048245e+00 -8.26966614e-02 -9.69017446e-02
3.38832587e-01 3.90543818e-01 3.76986116e-01 -9.55211520e-01
-7.10271180e-01 -2.77301371e-01 4.21110839e-01 -2.23752499e-01
-3.36853601e-02 -1.38317868e-01 -1.04474568e+00 -1.72959566e-01
-2.78840303e-01 6.39441431e-01 4.05731291e-01 1.09959602e+00
2.60998666e-01 1.61375090e-01 5.19222379e-01 -4.43391949e-02
-1.02709723e+00 -1.43446004e+00 -1.61858752e-01 5.90403497e-01
3.96682709e-01 -9.71209407e-02 -4.41705763e-01 7.68512413e-02] | [11.226177215576172, 8.205024719238281] |
2e6fc977-d116-4d91-8faa-6e20149f33b6 | fast-reinforcement-learning-with-generalized | null | null | https://www.pnas.org/doi/full/10.1073/pnas.1907370117 | https://www.pnas.org/doi/epdf/10.1073/pnas.1907370117 | Fast reinforcement learning with generalized policy updates | The combination of reinforcement learning with deep learning is a promising approach to tackle important sequential decision-making problems that are currently intractable. One obstacle to overcome is the amount of data needed by learning systems of this type. In this article, we propose to address this issue through a divide-and-conquer approach. We argue that complex decision problems can be naturally decomposed into multiple tasks that unfold in sequence or in parallel. By associating each task with a reward function, this problem decomposition can be seamlessly accommodated within the standard reinforcement-learning formalism. The specific way we do so is through a generalization of two fundamental operations in reinforcement learning: policy improvement and policy evaluation. The generalized version of these operations allow one to leverage the solution of some tasks to speed up the solution of others. If the reward function of a task can be well approximated as a linear combination of the reward functions of tasks previously solved, we can reduce a reinforcement-learning problem to a simpler linear regression. When this is not the case, the agent can still exploit the task solutions by using them to interact with and learn about the environment. Both strategies considerably reduce the amount of data needed to solve a reinforcement-learning problem. | ['and Doina Precup.', 'David Silver', 'Diana Borsa', 'Shaobo Hou', 'André Barreto'] | 2020-07-09 | null | null | null | proceedings-of-the-national-academy-of-2 | ['problem-decomposition'] | ['miscellaneous'] | [ 2.91406363e-01 2.18768239e-01 -1.44659355e-01 -7.83195123e-02
-4.26825464e-01 -7.80480623e-01 5.42289019e-01 4.02425736e-01
-8.21968198e-01 1.08666658e+00 -7.26704746e-02 -4.30852890e-01
-2.66082317e-01 -7.82150090e-01 -5.90154648e-01 -9.38617349e-01
4.44817264e-03 5.09138763e-01 2.26575240e-01 -4.82627690e-01
3.43122393e-01 5.28560340e-01 -1.53089619e+00 1.38693824e-01
8.49081695e-01 8.78247499e-01 3.74224156e-01 7.12635517e-01
-2.20584705e-01 9.80469763e-01 -4.73193735e-01 1.57884464e-01
4.47931319e-01 -2.87232429e-01 -1.10506332e+00 1.95167229e-01
-1.44596025e-01 -4.60624933e-01 5.54606616e-02 8.48089457e-01
2.43463978e-01 3.80237937e-01 4.87278402e-01 -1.21909404e+00
1.69969097e-01 3.20512682e-01 -4.12067175e-01 4.04970795e-02
2.28819594e-01 1.08994007e-01 9.57191169e-01 -2.25530207e-01
4.43324268e-01 1.21931314e+00 2.01424733e-01 5.48781633e-01
-1.49877596e+00 -2.88583398e-01 5.28212368e-01 2.77644306e-01
-5.48139215e-01 -1.98674068e-01 5.77939093e-01 -4.90036964e-01
1.06799960e+00 3.36655155e-02 7.74308562e-01 7.11028636e-01
1.34477675e-01 9.02580738e-01 1.26243162e+00 -4.96553421e-01
5.69017351e-01 1.77572332e-02 -2.04175548e-03 5.79009593e-01
-5.04439604e-03 5.09410381e-01 -2.10432604e-01 -4.00015488e-02
7.05170453e-01 1.95545241e-01 3.94525528e-02 -7.09225833e-01
-1.05215108e+00 9.71052647e-01 4.02162224e-01 3.85734767e-01
-6.47590220e-01 1.48517236e-01 4.78245735e-01 8.21223974e-01
1.83898389e-01 9.92310882e-01 -6.51503563e-01 -5.15283570e-02
-6.68760300e-01 7.28031874e-01 8.76600981e-01 1.22304894e-01
1.07788706e+00 1.68603361e-01 -3.48495692e-02 4.88695949e-01
1.10492371e-01 1.49060011e-01 4.62791502e-01 -1.29612195e+00
2.72813857e-01 4.79613513e-01 4.98062193e-01 -4.56893265e-01
-4.79355872e-01 -2.36962155e-01 -3.32230538e-01 9.30350423e-01
7.12670028e-01 -4.78867739e-01 -5.85926354e-01 1.78544593e+00
5.05141020e-01 -6.01826981e-02 2.82505065e-01 6.94443822e-01
-1.56932220e-01 6.26573503e-01 1.13312073e-01 -4.86317247e-01
1.02224147e+00 -1.04955304e+00 -4.78986055e-01 -4.23131227e-01
7.31191933e-01 -3.60816985e-01 6.74044609e-01 6.11513197e-01
-1.13108730e+00 -5.49481153e-01 -1.01527488e+00 1.70611203e-01
-2.06262574e-01 -2.85458297e-01 6.95307434e-01 1.76141456e-01
-1.00672889e+00 1.01596749e+00 -7.52413273e-01 -2.60469858e-02
2.20757157e-01 5.81439018e-01 -1.35814831e-01 1.09825909e-01
-9.65185106e-01 1.15948081e+00 5.44784367e-01 -1.38551265e-01
-1.01163495e+00 -2.79619962e-01 -7.07400203e-01 1.98550999e-01
1.04559577e+00 -6.19152009e-01 1.83392835e+00 -1.19500136e+00
-1.64916253e+00 4.31561083e-01 4.44190949e-02 -4.96530175e-01
5.38980722e-01 -1.97415739e-01 2.05348730e-01 -2.84001604e-02
7.96177611e-02 4.40794438e-01 1.10087645e+00 -1.05558062e+00
-1.01396668e+00 -5.38102031e-01 5.58525801e-01 4.01090324e-01
-2.09992751e-02 -1.42448932e-01 2.35529825e-01 -4.48707253e-01
-2.47216016e-01 -1.03359771e+00 -6.83581591e-01 -1.88626528e-01
1.52256593e-01 -5.35986304e-01 5.01565754e-01 -3.09985399e-01
9.20990944e-01 -1.92534661e+00 6.77308261e-01 6.97401240e-02
2.23653734e-01 2.97200888e-01 -3.28148305e-01 5.57487905e-01
-1.77397549e-01 -9.15488228e-02 -1.43272460e-01 -2.64534146e-01
5.37871718e-02 5.74891806e-01 -3.44430894e-01 1.82516143e-01
2.73514092e-01 8.35912406e-01 -1.11509275e+00 -2.41188947e-02
7.64122307e-02 -1.48217633e-01 -6.96225286e-01 4.61241871e-01
-6.23772621e-01 5.98844409e-01 -7.43770063e-01 7.83689618e-02
2.06505656e-01 1.53261647e-01 5.31668663e-01 4.18552965e-01
-1.80997029e-01 5.29457808e-01 -1.47667491e+00 1.55132973e+00
-5.26563048e-01 2.55319625e-01 3.01864594e-01 -1.74522960e+00
6.39790773e-01 2.18851015e-01 7.42705464e-01 -7.13015079e-01
-8.26432481e-02 2.20133871e-01 1.25678435e-01 -6.32336080e-01
2.13702083e-01 -4.86130714e-01 -8.81616101e-02 8.34713519e-01
6.00749366e-02 -2.04136074e-01 5.54643571e-01 -2.67776310e-01
1.19783545e+00 4.60422993e-01 5.12147903e-01 -9.31680351e-02
4.64930058e-01 1.99561536e-01 5.98225534e-01 8.98528814e-01
-1.05874836e-02 -2.39732251e-01 9.27480638e-01 -7.64198244e-01
-9.70229805e-01 -9.07254279e-01 3.02381724e-01 1.37197387e+00
-3.67805719e-01 -1.99875161e-01 -6.16024554e-01 -9.93393719e-01
1.62924960e-01 4.90068287e-01 -6.84002995e-01 -9.93735939e-02
-6.89376831e-01 -5.71349084e-01 3.43497656e-02 4.02205497e-01
3.63732636e-01 -1.33865726e+00 -1.21298206e+00 5.26857018e-01
2.62655020e-01 -7.15592265e-01 6.18838221e-02 7.90931582e-01
-9.73340571e-01 -1.20164061e+00 -5.47438502e-01 -5.92903852e-01
4.18491542e-01 2.08898932e-01 9.04333651e-01 1.45882415e-02
-3.73831503e-02 4.39811677e-01 -2.84091860e-01 -3.76006782e-01
-5.71280301e-01 1.14146084e-01 1.84168890e-01 -6.97516054e-02
3.68631296e-02 -6.67152107e-01 -3.28664184e-01 1.16516622e-02
-1.00835478e+00 6.61860704e-02 5.15270412e-01 9.86639619e-01
2.61917859e-01 2.94430882e-01 7.32441783e-01 -7.74138927e-01
1.00372386e+00 -4.46382791e-01 -9.49877083e-01 2.65216768e-01
-6.74588561e-01 7.58995831e-01 7.07201719e-01 -6.15779161e-01
-8.39824021e-01 3.18857402e-01 1.07091367e-01 -1.75800934e-01
-3.18611354e-01 6.79026902e-01 1.40809596e-01 9.97477546e-02
4.44139779e-01 1.99543536e-01 3.35962117e-01 -4.28608626e-01
2.94558287e-01 3.15921068e-01 1.26943022e-01 -8.09974015e-01
5.69311559e-01 7.98917338e-02 1.57805696e-01 -4.74032074e-01
-6.76598072e-01 -1.86440438e-01 -4.93478686e-01 -2.46545121e-01
7.96286285e-01 -4.70125943e-01 -1.09255838e+00 2.82855332e-01
-9.30855393e-01 -7.96742916e-01 -7.69386888e-01 3.07631373e-01
-8.44250619e-01 1.91343725e-01 -2.86266357e-01 -9.05106544e-01
2.21840695e-01 -1.42903662e+00 7.00076640e-01 1.96604460e-01
-1.48405716e-01 -9.46824610e-01 3.73696983e-01 1.47685260e-01
3.88958335e-01 5.20966016e-02 1.11888540e+00 -7.69612193e-01
-3.23817790e-01 -4.91205007e-02 4.30453457e-02 3.33291352e-01
4.72259521e-02 -4.36560094e-01 -6.77709818e-01 -5.27715087e-01
4.69401985e-01 -5.40538609e-01 8.43883455e-01 2.58674175e-01
9.33217525e-01 -3.99098307e-01 -2.89666117e-03 1.86443746e-01
1.26246250e+00 3.45632821e-01 1.97926179e-01 7.01795638e-01
2.85107493e-01 9.18560863e-01 6.57255948e-01 5.27579486e-01
3.02091002e-01 7.46289313e-01 5.01584888e-01 2.55981833e-01
4.39308703e-01 -1.25756532e-01 4.66754705e-01 1.98532805e-01
-1.78010985e-01 3.33618134e-01 -9.74521220e-01 2.43713975e-01
-2.33071637e+00 -1.09165883e+00 2.33948022e-01 2.06720281e+00
7.81191230e-01 7.78148845e-02 5.46598911e-01 2.83511013e-01
2.30096966e-01 6.81007653e-02 -8.64637315e-01 -8.44037712e-01
4.90539044e-01 2.64930874e-01 2.43996635e-01 6.64264977e-01
-1.01338148e+00 8.29738140e-01 6.59594917e+00 5.44725180e-01
-1.17220223e+00 -1.63266733e-01 2.89419353e-01 -1.38920039e-01
-1.07610844e-01 2.13167638e-01 -5.42371154e-01 1.69650033e-01
8.94288778e-01 -1.99012876e-01 1.08690667e+00 7.97271252e-01
2.25953415e-01 -5.96217394e-01 -1.42894530e+00 5.65835297e-01
-4.28172678e-01 -1.16290629e+00 -2.89342523e-01 2.70962477e-01
5.55347383e-01 -1.95082083e-01 -5.07257879e-02 7.58283556e-01
7.58982062e-01 -9.64786112e-01 4.77634102e-01 2.68474996e-01
2.16460556e-01 -6.17380381e-01 4.41615969e-01 9.01255965e-01
-7.90854812e-01 -7.01513290e-01 -1.33385658e-01 -7.55554020e-01
-1.77960902e-01 2.24833474e-01 -8.52370799e-01 4.69303370e-01
3.90139431e-01 5.01692832e-01 -2.24921867e-01 8.87321711e-01
-4.83292073e-01 8.51672366e-02 -2.32454419e-01 -1.23867668e-01
5.47112763e-01 -4.15027261e-01 4.20787960e-01 7.56536603e-01
1.29958764e-01 -6.63122088e-02 5.56559563e-01 7.03465879e-01
3.46654058e-01 2.29701772e-02 -6.96436584e-01 2.39218660e-02
1.38995321e-02 1.05552995e+00 -5.27743220e-01 -3.24589521e-01
-4.17970806e-01 7.72259474e-01 9.23480868e-01 5.15003681e-01
-3.70588750e-01 3.18181999e-02 6.14184380e-01 -1.34697586e-01
4.31214124e-01 -4.30883944e-01 -1.43768877e-01 -1.24548364e+00
1.45189194e-02 -1.32407832e+00 3.59985948e-01 -4.37075377e-01
-9.37582135e-01 2.79386282e-01 8.20266008e-02 -9.35586631e-01
-7.42827356e-01 -6.09577477e-01 -4.70765173e-01 7.83606589e-01
-1.62493896e+00 -5.86010158e-01 1.70603901e-01 7.16312110e-01
6.27850235e-01 -2.06106663e-01 9.54852581e-01 -2.66392916e-01
-5.23167074e-01 -1.72136903e-01 2.33020037e-01 -2.59177744e-01
3.88328075e-01 -1.54425538e+00 -9.16926563e-02 6.11835539e-01
8.73477533e-02 1.48368299e-01 6.29823506e-01 -4.06848371e-01
-1.32966423e+00 -6.22532189e-01 5.90330064e-01 3.94913368e-03
7.85930932e-01 -1.09604135e-01 -8.49831462e-01 4.56435502e-01
1.74589023e-01 -1.92796633e-01 3.49560499e-01 2.60921597e-01
-2.38158211e-01 -2.04071641e-01 -9.36815560e-01 5.50708175e-01
3.93742889e-01 -4.27450985e-01 -8.87765586e-01 1.45415023e-01
4.13823545e-01 -1.40922740e-01 -6.09514177e-01 1.70869604e-01
3.92508656e-01 -8.40593278e-01 7.20817268e-01 -1.17191935e+00
5.07321835e-01 -1.81977168e-01 2.33746935e-02 -1.75216126e+00
-2.94723004e-01 -8.11018944e-01 -2.85744011e-01 6.05032206e-01
2.39876971e-01 -7.08633721e-01 6.38789773e-01 6.75751865e-01
1.51439860e-01 -8.11534822e-01 -9.07619238e-01 -7.10502267e-01
3.85169357e-01 -2.25686342e-01 5.03630400e-01 7.98963487e-01
3.31869662e-01 4.36576992e-01 -2.11425468e-01 -2.25957051e-01
4.44738448e-01 3.43726635e-01 7.01656163e-01 -1.23160386e+00
-7.15845346e-01 -5.62502921e-01 1.42281204e-01 -1.05712593e+00
3.56933981e-01 -8.15311670e-01 1.68694645e-01 -1.48887277e+00
-2.75365915e-02 -6.59395993e-01 -3.86321127e-01 8.04209232e-01
-1.57289177e-01 -6.08016133e-01 5.36469162e-01 1.56720400e-01
-5.50964355e-01 2.90810317e-01 1.32240963e+00 -4.59457235e-03
-4.48979050e-01 3.85417521e-01 -5.90816319e-01 7.62153089e-01
1.09566128e+00 -5.82790434e-01 -4.01599616e-01 -3.69306445e-01
3.68111759e-01 5.93363643e-01 2.31462777e-01 -7.66682863e-01
3.24640304e-01 -6.14485860e-01 1.88821241e-01 1.64418817e-02
1.87360138e-01 -9.54271257e-01 -2.34184638e-01 7.74452925e-01
-6.50783479e-01 3.17524761e-01 1.95961595e-01 5.73242307e-01
-1.18631259e-01 -6.67452335e-01 6.35561705e-01 -3.83694619e-01
-6.91262662e-01 2.42662542e-02 -8.87818754e-01 -9.97973084e-02
1.29550910e+00 4.33697216e-02 2.33514123e-02 -3.49575937e-01
-1.05339396e+00 5.32946885e-01 1.85033187e-01 3.36584926e-01
3.82041454e-01 -1.00887203e+00 -4.83746141e-01 2.41066441e-01
-3.54167521e-01 4.26619723e-02 -1.10671483e-02 7.86769032e-01
-4.46492061e-02 3.50691646e-01 -6.96551681e-01 -1.77721515e-01
-1.12582958e+00 8.29000652e-01 5.97791672e-01 -1.02453673e+00
-2.91174233e-01 3.11871052e-01 1.78058431e-01 -3.08881313e-01
3.62434983e-01 -3.20732623e-01 -4.09961253e-01 3.70923430e-01
5.08354783e-01 2.61260509e-01 -8.88247341e-02 9.21599492e-02
-7.45736668e-03 2.53505766e-01 -3.19579244e-01 -4.03027117e-01
1.70666707e+00 5.98854646e-02 -1.04088761e-01 3.69485170e-01
8.59486520e-01 -3.81542593e-01 -1.62751150e+00 -3.86667073e-01
2.54960567e-01 -1.13815770e-01 1.35197908e-01 -8.29523087e-01
-7.94282496e-01 9.54260111e-01 3.35703641e-01 6.98035777e-01
1.13697731e+00 -3.21586102e-01 3.59241098e-01 7.62675166e-01
3.62497866e-01 -1.34876227e+00 3.09021235e-01 8.09346914e-01
8.20108771e-01 -1.23420060e+00 -3.19471047e-03 2.49941438e-01
-6.21127367e-01 1.40815067e+00 4.98399675e-01 -4.29026425e-01
3.23105663e-01 2.70445198e-01 -2.80680239e-01 2.00241208e-02
-9.77540731e-01 -5.30199587e-01 -1.78567600e-04 5.44846535e-01
1.61334649e-02 1.53966025e-02 -2.50712663e-01 3.43577594e-01
8.64190981e-02 1.16108321e-01 4.21688139e-01 1.19440722e+00
-7.74837375e-01 -1.68656600e+00 -2.78745443e-01 3.17976296e-01
-3.72026801e-01 2.53575087e-01 -1.86359927e-01 5.33290863e-01
-6.26256270e-03 8.52985442e-01 -6.62947297e-02 -6.07255772e-02
2.88231939e-01 1.85253471e-01 7.89183080e-01 -8.60347927e-01
-7.42808640e-01 1.19230911e-01 1.69676803e-02 -6.75419807e-01
-3.85979116e-01 -7.35363603e-01 -1.15463161e+00 -9.04565156e-02
2.55329192e-01 9.64635313e-02 6.42962456e-01 1.28491843e+00
2.79523507e-02 7.83666611e-01 7.77011096e-01 -1.03575671e+00
-1.21665776e+00 -4.99379069e-01 -4.79784757e-01 2.06921726e-01
7.30544806e-01 -7.66627133e-01 -6.82661161e-02 -2.52400190e-01] | [4.1761088371276855, 1.782096266746521] |
3a6dce11-1b81-4da6-a0b3-4e3665d19eb5 | incorporating-multi-target-in-multi-stage | 2107.04232 | null | https://arxiv.org/abs/2107.04232v1 | https://arxiv.org/pdf/2107.04232v1.pdf | Incorporating Multi-Target in Multi-Stage Speech Enhancement Model for Better Generalization | Recent single-channel speech enhancement methods based on deep neural networks (DNNs) have achieved remarkable results, but there are still generalization problems in real scenes. Like other data-driven methods, DNN-based speech enhancement models produce significant performance degradation on untrained data. In this study, we make full use of the contribution of multi-target joint learning to the model generalization capability, and propose a lightweight and low-computing dilated convolutional network (DCN) model for a more robust speech denoising task. Our goal is to integrate the masking target, the mapping target, and the parameters of the traditional speech enhancement estimator into a DCN model to maximize their complementary advantages. To do this, we build a multi-stage learning framework to deal with multiple targets in stages to achieve their joint learning, namely `MT-in-MS'. Our experimental results show that compared with the state-of-the-art time domain and time-frequency domain models, this proposed low-cost DCN model can achieve better generalization performance in speaker, noise, and channel mismatch cases. | ['Xuyi Zhuang', 'Zehua Zhang', 'Andong Li', 'Mingjiang Wang', 'Lu Zhang'] | 2021-07-09 | null | null | null | null | ['speech-denoising'] | ['speech'] | [ 2.22514898e-01 -4.84020293e-01 2.99488306e-01 -4.94345129e-01
-1.01515782e+00 1.43567011e-01 2.80082822e-01 -4.24272090e-01
-6.25364244e-01 3.96178454e-01 5.39538026e-01 -2.41716281e-01
3.13769560e-03 -3.22263628e-01 -3.66197050e-01 -9.22341049e-01
1.38094246e-01 -5.70797861e-01 2.45603904e-01 -5.36624134e-01
-2.34969202e-02 2.32443795e-01 -1.32503712e+00 3.48161221e-01
1.20444119e+00 1.14480758e+00 6.67526901e-01 6.65035546e-01
1.54832602e-01 6.85461760e-01 -6.89927042e-01 -2.66511470e-01
7.51863644e-02 -3.46895337e-01 4.05786894e-02 1.40478406e-02
9.76985767e-02 -6.50960743e-01 -7.35385060e-01 1.37894213e+00
1.40847468e+00 3.31510127e-01 3.45732152e-01 -8.83045018e-01
-7.18048930e-01 5.57389915e-01 -5.00010848e-01 4.15658087e-01
-2.46118456e-01 6.47467598e-02 5.17991543e-01 -1.15615475e+00
7.51457587e-02 1.37340963e+00 1.09412646e+00 4.65330601e-01
-7.13197470e-01 -1.01037490e+00 3.30692321e-01 5.41832268e-01
-1.08663440e+00 -9.69074547e-01 9.52081442e-01 2.62434125e-01
9.21390712e-01 -3.72134037e-02 1.89498708e-01 1.29508150e+00
-5.02672195e-02 9.91896808e-01 1.17145431e+00 -3.94160420e-01
1.37698397e-01 -6.70463219e-02 -1.64051473e-01 1.54285133e-01
-2.46390820e-01 5.52739441e-01 -7.09649324e-01 2.69674361e-01
8.64940882e-01 -1.39529005e-01 -4.43193406e-01 3.03818792e-01
-7.60269582e-01 4.53612477e-01 4.64138836e-01 5.72209120e-01
-4.04282391e-01 1.92403141e-02 4.09172893e-01 4.87874299e-01
7.82436311e-01 -1.82183251e-01 -6.96093023e-01 -3.26093942e-01
-1.07177353e+00 1.73387229e-02 3.51917386e-01 8.04945052e-01
4.24772739e-01 6.82208061e-01 -2.76630253e-01 1.21513033e+00
3.36619914e-01 4.72430825e-01 7.20881343e-01 -6.41959786e-01
6.26282692e-01 -1.52273059e-01 -1.81857884e-01 -7.08645105e-01
-3.88963014e-01 -1.14589238e+00 -1.36241305e+00 2.22345516e-01
-1.21781848e-01 -4.76238012e-01 -1.01052034e+00 1.85285842e+00
2.22224101e-01 4.36877400e-01 3.77565056e-01 1.08591342e+00
8.14866960e-01 8.62505972e-01 1.72806174e-01 -4.98835951e-01
9.19705987e-01 -1.14980662e+00 -1.43512332e+00 -2.72200972e-01
1.70368269e-01 -1.08910286e+00 5.77150166e-01 6.84551060e-01
-1.18872583e+00 -1.20274353e+00 -1.12307203e+00 -1.32907733e-01
-1.02280483e-01 4.96921092e-01 4.59621102e-01 8.35566819e-01
-1.14834642e+00 7.33916819e-01 -8.14796090e-01 2.82226782e-02
3.39500844e-01 2.66743273e-01 -1.52291509e-03 -1.34693589e-02
-1.45907891e+00 9.57202852e-01 3.23136866e-01 6.43174946e-01
-1.11085427e+00 -4.92079645e-01 -6.77807152e-01 1.06396861e-01
2.68225759e-01 -2.83528626e-01 1.51256061e+00 -9.81898189e-01
-1.74446964e+00 1.73673518e-02 -2.52817929e-01 -4.38404411e-01
2.49176323e-01 -3.49217147e-01 -1.19594944e+00 -1.23986736e-01
-4.06100631e-01 3.82705748e-01 1.26514745e+00 -1.08022928e+00
-7.41374731e-01 -3.30743283e-01 -3.83125454e-01 4.23649102e-01
-7.21352875e-01 2.66023278e-01 -4.56440985e-01 -1.25372791e+00
4.61870462e-01 -2.10103512e-01 -3.50731373e-01 5.04613481e-02
-6.13182746e-02 5.22433035e-02 1.07230723e+00 -1.42546964e+00
1.51294875e+00 -2.55783439e+00 3.43036726e-02 -1.75319672e-01
-6.71424568e-02 8.54674459e-01 -3.63536239e-01 2.46153370e-01
-1.59240946e-01 -3.32745731e-01 -2.50604063e-01 -7.28385985e-01
9.11058262e-02 4.97498922e-02 -1.34995356e-02 3.74637842e-01
2.06723124e-01 5.35022616e-01 -4.81360465e-01 -3.51812422e-01
2.97554672e-01 9.86355126e-01 -4.78610516e-01 3.30053627e-01
2.75614530e-01 2.80020148e-01 7.33739755e-04 6.69592202e-01
1.14111161e+00 3.26135069e-01 4.48754430e-02 -5.73916137e-01
-1.44767433e-01 2.18395650e-01 -1.55221975e+00 1.74919820e+00
-5.91918588e-01 6.77914262e-01 7.86230922e-01 -1.08572137e+00
1.14401317e+00 6.58519089e-01 1.51686355e-01 -9.69384432e-01
1.66666225e-01 5.00470638e-01 2.13004932e-01 -5.64629793e-01
3.68641436e-01 -3.42812628e-01 4.74263161e-01 7.49897503e-04
4.26220000e-01 1.61858499e-01 -3.88578147e-01 -1.50848150e-01
7.96571910e-01 -3.01597826e-02 7.78410509e-02 2.67940369e-02
6.54336274e-01 -8.44206214e-01 9.83272016e-01 7.41611302e-01
-6.01837993e-01 6.33813560e-01 -9.43426862e-02 2.11728364e-02
-1.04137182e+00 -1.08089197e+00 -7.82847181e-02 1.28064215e+00
4.02502865e-02 4.83754762e-02 -8.69771123e-01 -1.09917827e-01
-4.41457719e-01 4.05063868e-01 -1.92782860e-02 -2.99951881e-01
-6.46039128e-01 -8.76715899e-01 5.78969717e-01 7.42625535e-01
1.13952041e+00 -8.12451005e-01 2.52629489e-01 6.13081276e-01
-3.82431477e-01 -1.23712873e+00 -6.59704149e-01 3.95089388e-01
-8.74016225e-01 -2.19589069e-01 -1.00344706e+00 -1.19356227e+00
1.96335346e-01 3.92081797e-01 4.27504271e-01 -2.47618362e-01
2.14180976e-01 -1.82660297e-03 -5.05033553e-01 -5.32642901e-01
-4.36658531e-01 -1.44050658e-01 2.33610392e-01 3.00735801e-01
2.47606888e-01 -9.34659421e-01 -7.06812143e-01 2.78843760e-01
-9.71705437e-01 -2.23186076e-01 9.25314963e-01 1.12227643e+00
2.87550241e-01 5.39848804e-01 1.01556540e+00 -1.69676408e-01
1.01152313e+00 -1.44297928e-01 -4.53381538e-01 1.64445207e-01
-6.44709051e-01 -1.85736135e-01 6.37348473e-01 -8.28064680e-01
-1.80390584e+00 -7.49593824e-02 -8.09149206e-01 -3.90293092e-01
-7.23400265e-02 5.57546616e-01 -6.01896524e-01 -1.72675222e-01
6.22298479e-01 5.43278396e-01 1.18801728e-01 -9.42064226e-01
1.72980562e-01 1.18580580e+00 8.30875933e-01 -1.82525471e-01
7.71949053e-01 1.62965894e-01 -3.68418843e-01 -7.85708904e-01
-6.12774730e-01 -6.27309024e-01 -2.61572510e-01 -1.33434638e-01
6.06901407e-01 -1.49049199e+00 -3.40646625e-01 1.07740688e+00
-1.43635178e+00 -2.03926191e-01 4.76115271e-02 8.53798687e-01
-1.72716811e-01 6.20257080e-01 -8.55668783e-01 -1.02249277e+00
-4.79409605e-01 -1.14413798e+00 8.11150074e-01 3.71557206e-01
5.81470728e-01 -8.77171814e-01 -2.46947557e-01 1.11113504e-01
1.14093590e+00 -6.40230775e-01 4.68875259e-01 -5.39440155e-01
-2.91317642e-01 4.71001826e-02 -2.63926178e-01 1.15414441e+00
1.15110494e-01 -4.86312449e-01 -1.32959712e+00 -4.98792499e-01
7.27676094e-01 7.30818138e-02 9.06655550e-01 7.24274337e-01
1.26462173e+00 -2.15821803e-01 7.91171193e-03 9.47195053e-01
1.35136068e+00 4.67263401e-01 9.52649891e-01 2.12491706e-01
3.00141931e-01 3.42493474e-01 4.82053459e-01 5.21727860e-01
5.45131750e-02 5.65876901e-01 3.07324708e-01 -5.54787397e-01
-5.01840770e-01 -1.00516025e-02 5.84054410e-01 1.32702124e+00
-1.68422367e-02 -2.56736994e-01 -3.66196632e-01 4.97230530e-01
-1.71796751e+00 -9.34378803e-01 1.12641603e-01 1.75570822e+00
8.81131589e-01 1.30389601e-01 -1.82293743e-01 3.55594307e-01
1.02980876e+00 2.99353600e-01 -5.00699818e-01 -1.93813905e-01
-5.17775655e-01 4.32824314e-01 2.59183019e-01 4.44738328e-01
-1.28535533e+00 6.22972250e-01 5.97836399e+00 1.44106257e+00
-1.16066122e+00 4.57620770e-01 5.61480999e-01 6.76638037e-02
1.22115336e-01 -3.31395119e-01 -7.17721879e-01 3.72788161e-01
9.65481877e-01 1.02474280e-01 6.74624443e-01 6.59139812e-01
5.98105252e-01 4.07645971e-01 -4.98224318e-01 1.15597236e+00
-2.34914999e-02 -8.50400150e-01 -3.16462517e-01 -3.27610880e-01
6.71231866e-01 -2.10126340e-02 2.50771254e-01 6.18464768e-01
-1.78754091e-01 -8.07390869e-01 6.04281723e-01 4.17249143e-01
6.69737518e-01 -7.84425139e-01 8.38845909e-01 4.45924520e-01
-1.26796842e+00 -4.94069487e-01 -4.99274522e-01 -1.01660639e-01
3.70595455e-01 8.20417941e-01 -4.24161255e-01 7.05595553e-01
7.68981218e-01 5.97842693e-01 -9.96590331e-02 1.16784489e+00
-3.19864243e-01 6.91056907e-01 -1.40954405e-01 8.05782601e-02
6.81183487e-02 1.62333527e-04 6.07156217e-01 1.35356629e+00
5.03088832e-01 1.68053374e-01 -2.18024686e-01 5.26182413e-01
-2.45027579e-02 -9.48310345e-02 5.37225837e-03 1.51235551e-01
5.20996034e-01 1.21280968e+00 -3.25701647e-02 -1.89080656e-01
-5.02405107e-01 1.02953410e+00 -1.40051013e-02 6.98575079e-01
-9.50611651e-01 -6.97791576e-01 6.49174750e-01 -3.08665991e-01
5.98753035e-01 -3.16264093e-01 -2.58987606e-01 -9.83216405e-01
2.81223834e-01 -1.28926957e+00 -1.15735240e-01 -8.64023328e-01
-1.25422490e+00 7.33327329e-01 -5.59241176e-01 -1.33990955e+00
1.08430935e-02 -7.63479948e-01 -6.29093289e-01 1.20955133e+00
-1.84775794e+00 -1.29360616e+00 -2.21922889e-01 8.34798396e-01
6.72577024e-01 -3.62842113e-01 4.89758193e-01 8.82722676e-01
-5.42275548e-01 7.53225982e-01 4.36811328e-01 1.25911251e-01
8.61905694e-01 -7.99742937e-01 3.54568899e-01 1.40081453e+00
-2.65675515e-01 4.15878922e-01 6.59327149e-01 -4.95571792e-01
-1.08298457e+00 -1.07794511e+00 6.92702115e-01 4.60908234e-01
3.07079881e-01 -3.62877637e-01 -1.06211710e+00 3.33861947e-01
3.60517085e-01 -4.26197276e-02 2.46348664e-01 1.56112209e-01
-1.74131662e-01 -5.85157871e-01 -9.97052550e-01 5.77302873e-01
1.00116265e+00 -6.95570946e-01 -4.48225200e-01 1.03520930e-01
9.48481500e-01 -5.51084161e-01 -6.53324664e-01 6.72247529e-01
3.48888874e-01 -1.01849544e+00 1.13307130e+00 -2.79941499e-01
2.07758680e-01 -2.74668992e-01 -4.36774641e-01 -1.56855142e+00
-3.35989177e-01 -9.27713215e-01 -1.48479328e-01 1.42275035e+00
3.54277581e-01 -5.16377687e-01 4.05806959e-01 9.45387706e-02
-8.45627546e-01 -5.13705909e-01 -1.23378718e+00 -8.76062214e-01
-1.35735616e-01 -6.30448937e-01 6.02194011e-01 6.43507540e-01
-2.46082217e-01 2.22864881e-01 -7.84049511e-01 5.71044326e-01
6.19011402e-01 -5.77931345e-01 3.54569316e-01 -7.99262345e-01
-4.87835467e-01 -2.81340092e-01 -1.94334924e-01 -1.46375263e+00
-1.50473043e-01 -3.04500729e-01 4.12612975e-01 -1.42375612e+00
-1.57316580e-01 -1.88301966e-01 -5.33148229e-01 1.10727258e-01
-4.12413806e-01 -1.72292620e-01 -1.81302670e-02 -1.30978167e-01
-4.04112220e-01 9.84088302e-01 1.28672445e+00 -1.35314226e-01
-1.84543177e-01 1.18305773e-01 -7.42899895e-01 5.62764049e-01
5.63195646e-01 -1.88546017e-01 -3.12133938e-01 -8.27087343e-01
-5.04733145e-01 2.55641460e-01 2.96571016e-01 -1.35157788e+00
7.77515173e-01 2.08926216e-01 4.30227429e-01 -6.79889083e-01
7.47004926e-01 -8.85262430e-01 -7.03664124e-02 3.36041689e-01
-1.84396267e-01 -3.04363787e-01 3.68886858e-01 5.08095205e-01
-7.61141300e-01 -8.59450325e-02 9.36390817e-01 1.39104083e-01
-8.74178767e-01 2.81336397e-01 -3.42818588e-01 -3.41818511e-01
3.96805018e-01 -1.11608021e-01 -3.24153960e-01 -7.58336246e-01
-7.50901818e-01 -1.23214200e-01 -3.73523295e-01 2.89983392e-01
8.77585113e-01 -1.31407738e+00 -9.60743606e-01 3.47013384e-01
-4.38695848e-01 -2.54480809e-01 9.34903920e-01 8.43907237e-01
4.24860939e-02 1.32307231e-01 -1.03416406e-01 -4.67192918e-01
-1.05948842e+00 4.50955510e-01 7.39240408e-01 -2.03872442e-01
-2.30001718e-01 1.05304992e+00 1.32903725e-01 -5.59995830e-01
6.53956532e-01 -1.95225433e-01 -6.11376986e-02 -2.93027639e-01
6.44836962e-01 5.57600856e-01 4.29598719e-01 -3.51406634e-01
-1.12919733e-02 1.57332435e-01 -1.30417585e-01 -3.37539256e-01
1.51328206e+00 -3.43620718e-01 1.05716303e-01 -2.09281445e-01
1.25618303e+00 -1.04922041e-01 -1.48278224e+00 -5.49174607e-01
-5.63317895e-01 -2.12627411e-01 6.06232524e-01 -1.05375028e+00
-1.15882254e+00 1.20172501e+00 1.36423934e+00 -9.86918435e-02
1.94024551e+00 -5.83053291e-01 1.00049555e+00 9.29901153e-02
-5.08839153e-02 -1.49790466e+00 3.09087843e-01 5.95124960e-01
9.41140592e-01 -1.31203651e+00 -1.85445040e-01 -3.79792720e-01
-3.86631459e-01 1.14443064e+00 7.18221247e-01 3.91052246e-01
7.49276102e-01 4.61367339e-01 2.65925378e-01 3.27583104e-01
-5.44386506e-01 -2.85880506e-01 1.30227521e-01 1.04804015e+00
2.30474979e-01 -1.69111684e-01 -1.40642345e-01 1.02247262e+00
1.54940948e-01 -1.00980327e-03 9.08723325e-02 7.28070736e-01
-7.08384216e-01 -1.19375837e+00 -5.53447127e-01 7.85552636e-02
-4.97240603e-01 -5.25194168e-01 3.49240363e-01 4.73632544e-01
1.73022017e-01 1.34535050e+00 -3.09935659e-01 -6.64708853e-01
4.82174665e-01 -2.72331923e-01 8.95627588e-02 1.91260614e-02
-6.62227809e-01 6.64895415e-01 -7.60828704e-03 -1.68401018e-01
-4.58926350e-01 -4.69611913e-01 -8.62490237e-01 -1.27838641e-01
-6.52120173e-01 -1.66125059e-01 9.42577660e-01 9.56455052e-01
3.10550362e-01 9.32664454e-01 8.07582796e-01 -9.15090561e-01
-1.05119896e+00 -1.36888778e+00 -7.43390143e-01 2.09851027e-01
6.77846313e-01 -3.54621410e-01 -5.47844708e-01 -4.81637977e-02] | [14.948225021362305, 5.942246437072754] |
3b7ddef0-d418-4ebc-809f-6fc98853c00a | peer-to-peer-energy-trading-in-smart-grid | null | null | https://ieeexplore.ieee.org/abstract/document/9386079 | https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9386079 | Peer-to-peer energy trading in smart grid through blockchain: A double auction-based game theoretic approach | In a smart grid, each residential unit with renewable energy sources can trade energy with others for profit. Buyers with insufficient energy meet their demand by buying the required energy from other houses with surplus energy. However, they will not be willing to engage in the trade if it is not beneficial. With the aim of improving participants’ profits and reducing the impacts on the grid, we study a peer-to-peer (P2P) energy trading system among prosumers using a double auction-based game theoretic approach, where the buyer adjusts the amount of energy to buy according to varying electricity price in order to maximize benefit, the auctioneer controls the game, and the seller does not participate in the game but finally achieves the maximum social welfare. The proposed method not only benefits the participants but also hides their information, such as their bids and asks, for privacy. We further study individual rationality and incentive compatibility properties in the proposed method’s auction process at the game’s unique Stackelberg equilibrium. For practical applicability, we implement our proposed energy trading system using blockchain technology to show the feasibility of real-time P2P trading. Finally, simulation results under different scenarios demonstrate the effectiveness of the proposed method. | ['D Kim', 'J Cho', 'HT Doan'] | 2021-04-05 | null | null | null | journal-2021-4 | ['energy-management'] | ['time-series'] | [-5.70179462e-01 4.48493332e-01 -8.81648362e-02 5.65843247e-02
-1.87442288e-01 -1.18419373e+00 -4.90061156e-02 1.46670252e-01
-3.61826897e-01 1.10117710e+00 -6.99985325e-02 -2.45588750e-01
-5.49144447e-02 -1.33220744e+00 -2.16969639e-01 -1.36715400e+00
-2.06048280e-01 3.92999262e-01 3.21086147e-03 -4.39863466e-03
4.12035845e-02 -1.95339352e-01 -6.27003968e-01 -4.73263323e-01
1.09485686e+00 1.16477859e+00 -1.63867120e-02 -1.27788365e-01
2.68390715e-01 6.26208723e-01 -5.18466532e-01 -3.73588920e-01
9.50515270e-01 -6.73866689e-01 -2.60006577e-01 -3.93947542e-01
-1.18455911e+00 -1.03383482e+00 5.25420383e-02 1.56282949e+00
4.23939973e-01 -3.46260816e-01 1.42024100e-01 -1.98709047e+00
-7.26800442e-01 1.21237457e+00 -9.10614908e-01 -3.12337607e-01
-5.45672216e-02 1.12199508e-01 1.26676250e+00 1.92892551e-03
3.57524782e-01 3.42297733e-01 -3.61477165e-03 5.88774502e-01
-9.04853523e-01 -1.16532207e+00 -1.72090590e-01 2.28216842e-01
-1.21810472e+00 -2.09686905e-02 9.60264146e-01 2.68880546e-01
7.69838810e-01 6.12813532e-01 1.43547976e+00 -2.62308419e-01
4.43744838e-01 6.61413848e-01 1.38978183e+00 -9.60531682e-02
8.40024292e-01 1.62412688e-01 -2.55126625e-01 -4.10070330e-01
7.49947608e-01 9.62520093e-02 -7.22455382e-02 -5.86795449e-01
4.06725883e-01 3.58230285e-02 -6.16936445e-01 -7.08492875e-01
-1.10971999e+00 7.68280506e-01 4.54376519e-01 2.32956678e-01
-8.15968335e-01 2.04741076e-01 3.90802383e-01 6.51606023e-01
2.34241486e-01 -1.68252200e-01 -3.89436394e-01 -1.08311949e-02
-6.22500300e-01 7.20016137e-02 1.24036860e+00 1.09467959e+00
2.45070234e-01 -6.57421574e-02 1.47009388e-01 1.15276435e-02
6.42068148e-01 5.14368653e-01 1.88411221e-01 -9.46493804e-01
2.45766446e-01 4.32421952e-01 7.98215687e-01 -5.51203668e-01
3.94880492e-03 2.72414107e-02 -1.13766062e+00 3.45425546e-01
3.12938571e-01 -6.06452227e-01 1.32478848e-01 1.54907501e+00
3.49611044e-01 -4.84022826e-01 2.62404531e-01 9.91493106e-01
2.64930218e-01 9.36468780e-01 -1.09811477e-01 -1.12419724e+00
1.49047816e+00 -6.09674096e-01 -8.48986328e-01 3.90116751e-01
1.65240750e-01 -2.75832087e-01 -7.43443295e-02 1.39494643e-01
-1.43265259e+00 8.50261807e-01 -1.00465381e+00 5.98808527e-01
9.11275819e-02 -4.47188288e-01 1.53317168e-01 7.19326258e-01
-1.09744787e+00 1.71410382e-01 -6.90258384e-01 -1.59707233e-01
4.77509350e-01 7.32489407e-01 7.64244348e-02 7.56756425e-01
-1.36627269e+00 4.79091495e-01 6.55380428e-01 1.55070111e-01
-3.66756499e-01 -5.90633631e-01 -3.47982585e-01 7.72295296e-01
5.41056454e-01 -7.79760480e-01 1.27621412e+00 -7.54316688e-01
-1.51961839e+00 3.98356169e-01 5.11852622e-01 -7.83934534e-01
1.08801341e+00 6.76162302e-01 1.58622712e-01 3.29996832e-02
1.34852871e-01 2.41068706e-01 8.66855904e-02 -1.02646399e+00
-9.61154699e-01 -4.19569612e-01 1.31166488e-01 5.55142283e-01
-2.02400565e-01 -9.79288742e-02 5.55450082e-01 -2.66763866e-01
-9.55739021e-02 -8.62245858e-01 -1.10050045e-01 -1.21490605e-01
-3.09546828e-01 -4.33603168e-01 8.46082389e-01 -5.23305893e-01
6.15158677e-01 -2.09713721e+00 -4.78513718e-01 5.09409308e-01
2.44400769e-01 -2.28219166e-01 4.62453812e-01 8.46078336e-01
3.55918646e-01 3.05493563e-01 -6.99942186e-02 3.69749129e-01
5.85735202e-01 9.75898355e-02 -2.12108996e-02 4.78905380e-01
-7.34492421e-01 1.08813739e+00 -1.03097022e+00 -4.33263518e-02
1.57724455e-01 -1.49707049e-01 -2.69414663e-01 -1.60329081e-02
1.07255420e-02 2.80158311e-01 -7.88789213e-01 2.51415402e-01
1.04586327e+00 -2.09799945e-01 7.16185868e-01 1.55823335e-01
-3.35444987e-01 8.17292556e-02 -1.38624358e+00 7.52163529e-01
-3.05049364e-02 1.88941926e-01 8.51205468e-01 -7.54576623e-01
4.48867559e-01 6.40068829e-01 8.55660796e-01 -6.49373353e-01
2.89972663e-01 4.70646828e-01 1.20538302e-01 2.61484981e-01
2.69979179e-01 -1.79118440e-01 -3.69393192e-02 1.24730265e+00
-5.94494581e-01 1.57411788e-02 -4.74206395e-02 2.45893419e-01
8.12641084e-01 -2.96341956e-01 9.61830735e-01 -6.97226107e-01
3.75512064e-01 -2.25499704e-01 9.23810482e-01 3.90270591e-01
-5.04622519e-01 -4.12189782e-01 4.79242623e-01 -1.22176230e-01
-8.48088503e-01 -8.13442588e-01 5.15991628e-01 3.24936897e-01
6.40053630e-01 -6.54685274e-02 -6.56275868e-01 -5.39779425e-01
1.60980403e-01 9.94310558e-01 -9.07243192e-02 2.74672061e-01
-7.79183730e-02 -6.40652716e-01 -2.32465819e-01 2.13050544e-01
1.17257392e+00 -1.15886962e+00 -1.22837734e+00 3.70544523e-01
-3.18914592e-01 -4.79173601e-01 -9.51725721e-01 1.03673041e-01
-2.74320543e-01 -1.01607382e+00 -5.89120626e-01 -5.79008818e-01
8.69216084e-01 5.46195388e-01 6.26327038e-01 1.40788361e-01
4.39950764e-01 1.91900462e-01 -2.82012373e-01 -5.43889523e-01
-3.84820759e-01 -1.67978570e-01 -1.27250820e-01 -2.54738629e-01
2.53981382e-01 -8.88772905e-01 -1.40580845e+00 4.40465569e-01
-7.96236157e-01 6.35243505e-02 1.43925130e-01 6.49518788e-01
3.50971282e-01 8.23385656e-01 1.20023143e+00 -5.32915771e-01
1.00177395e+00 -5.09901822e-01 -1.36676586e+00 2.92586923e-01
-1.02928615e+00 -3.81008089e-01 7.98756480e-01 1.98601693e-01
-8.90214920e-01 -1.74689174e-01 5.53207278e-01 2.07679570e-01
6.42186463e-01 1.58376485e-01 -7.90982008e-01 -2.14451537e-01
-5.97861171e-01 6.05503201e-01 5.14725037e-02 1.29207805e-01
5.16700208e-01 7.09713340e-01 6.60553016e-03 -9.57687274e-02
9.30714250e-01 4.23118860e-01 -7.94637948e-02 -7.72184208e-02
5.40018857e-01 1.73870437e-02 4.26488630e-02 -4.26049195e-02
5.43740749e-01 -1.04333723e+00 -1.96223211e+00 8.19348395e-01
-1.05472434e+00 1.71679884e-01 -5.65449536e-01 2.92083740e-01
-4.45806175e-01 4.24336255e-01 -5.64575016e-01 -1.17694092e+00
-9.57257330e-01 -7.42900729e-01 -5.69419563e-02 6.90297842e-01
9.77842361e-02 -5.25111556e-01 -1.36135176e-01 1.66570097e-01
6.86782479e-01 1.11404680e-01 7.71007955e-01 -8.96969438e-01
-1.37295210e+00 -1.57103743e-02 -9.65241864e-02 1.76594153e-01
6.84778988e-01 -4.67196077e-01 -2.24925995e-01 -9.34904158e-01
4.42387551e-01 7.63921142e-02 -2.08224311e-01 4.09942061e-01
3.93376887e-01 -1.22737539e+00 -1.76299796e-01 -1.89126544e-02
1.76195836e+00 9.24461901e-01 6.36021137e-01 3.75197709e-01
6.90158340e-04 3.76295835e-01 4.33973700e-01 8.09010863e-01
9.65894401e-01 2.49705181e-01 7.74003685e-01 4.72370505e-01
8.60355616e-01 -2.41205141e-01 3.36247832e-01 7.42093980e-01
-5.63812591e-02 -2.87103146e-01 -9.44836512e-02 6.25984251e-01
-2.08967447e+00 -1.02455533e+00 1.36110768e-01 2.35875726e+00
7.37540901e-01 -2.13107750e-01 3.57454062e-01 1.44163081e-02
8.42331588e-01 -7.04273731e-02 -9.84769404e-01 -2.90781438e-01
3.94585617e-02 -3.10403109e-01 1.13334584e+00 4.53129113e-02
-1.27970383e-01 3.01756889e-01 4.48138809e+00 7.43521154e-01
-7.50349760e-01 3.45219076e-01 8.32376957e-01 -3.56679559e-01
-7.57215202e-01 5.43426156e-01 -7.75901452e-02 9.72581804e-01
5.63890517e-01 -1.39544702e+00 1.07748044e+00 5.71867287e-01
6.71571314e-01 -5.24417818e-01 -8.30680668e-01 6.85891271e-01
-7.82216489e-01 -1.31452274e+00 -6.56363308e-01 5.63500881e-01
7.50154674e-01 -1.09280653e-01 -5.42344928e-01 -2.69748926e-01
9.42356825e-01 -3.37075174e-01 1.07636118e+00 -4.08134200e-02
-5.00302240e-02 -1.34431565e+00 7.88537860e-01 6.38796031e-01
-1.47419572e+00 -3.01505476e-01 -1.21586077e-01 -6.52455464e-02
5.09997547e-01 4.96135950e-01 -3.59167218e-01 9.31386292e-01
8.14699709e-01 -1.79426782e-02 3.06596965e-01 8.96691799e-01
-5.65451562e-01 2.92086035e-01 -6.37088239e-01 -4.03611839e-01
-5.13996407e-02 -1.08653629e+00 5.04280329e-01 -7.76079074e-02
6.37835205e-01 9.02122974e-01 2.45361403e-02 1.13211155e+00
-6.04727030e-01 3.12699378e-01 -3.33623797e-01 5.50304763e-02
1.25100851e+00 1.33258915e+00 -1.06478214e+00 -1.80252418e-01
-3.34462345e-01 8.80559683e-01 -4.17607576e-01 1.85420245e-01
-4.47791487e-01 -4.58976060e-01 2.33280450e-01 -9.51181501e-02
4.18740541e-01 2.31457755e-01 -4.87810135e-01 -1.00499988e+00
4.00140405e-01 -4.30107862e-01 3.64771932e-01 -8.36845756e-01
-1.26593065e+00 -1.30528092e-01 -2.79773593e-01 -1.10177505e+00
-3.94726962e-01 5.22577047e-01 -1.26539898e+00 8.70839179e-01
-1.42080939e+00 -8.31129014e-01 2.75936484e-01 6.11607730e-01
-1.62818968e-01 2.93481022e-01 4.31077957e-01 -1.56967744e-01
5.22548966e-02 2.58756399e-01 7.73319185e-01 2.43734151e-01
-1.74000897e-02 -1.26716244e+00 8.85470025e-03 6.76373541e-01
-4.20268774e-01 2.39019051e-01 6.21354282e-01 -7.29247749e-01
-1.60239553e+00 -6.00813925e-01 6.59356833e-01 4.79384750e-01
8.55476677e-01 -2.94145942e-01 -5.14458597e-01 5.52802265e-01
1.13019931e+00 -1.89306036e-01 5.59312105e-01 -8.78635108e-01
3.63094419e-01 -5.24124086e-01 -2.03679347e+00 6.07544243e-01
6.45597339e-01 -1.92992464e-01 -3.53179097e-01 -1.13647096e-02
3.99595201e-01 6.84145018e-02 -7.74532557e-01 -2.36608312e-02
5.94731629e-01 -6.98699415e-01 3.34530801e-01 2.54137069e-01
-3.59217674e-01 -5.51932693e-01 1.93061773e-02 -1.90306091e+00
-1.41275257e-01 -1.26947415e+00 1.35921583e-01 1.48269403e+00
-2.39347368e-02 -1.57390106e+00 7.90205896e-01 1.09447765e+00
8.01108956e-01 -2.49400154e-01 -1.62574613e+00 -7.62472272e-01
4.30984914e-01 5.65204382e-01 1.40037489e+00 8.29894543e-01
9.37745273e-01 -1.54902175e-01 -3.58442992e-01 2.37138718e-01
1.32329834e+00 5.70907950e-01 3.03848863e-01 -7.97119737e-01
9.71514061e-02 -2.71670431e-01 1.24040097e-01 -7.14889050e-01
-1.50012970e-01 -7.48050034e-01 -1.63100176e-02 -1.67174208e+00
5.07532001e-01 -2.75986552e-01 -5.80454051e-01 5.90345562e-01
2.76248038e-01 -1.23983651e-01 9.46125031e-01 5.10310113e-01
-3.86252731e-01 8.13448727e-01 1.05684066e+00 -3.30459893e-01
-1.99009836e-01 2.07011476e-01 -1.10049975e+00 1.26701832e-01
9.11921918e-01 -4.10931379e-01 -6.21280670e-01 1.91815749e-01
2.97114104e-01 7.05724478e-01 1.73075855e-01 -1.62800744e-01
6.87169969e-01 -3.63339424e-01 -1.42123878e-01 -8.26279819e-01
-6.11702323e-01 -1.46321988e+00 9.45346892e-01 1.02681971e+00
3.32231000e-02 5.74651919e-03 -6.17546797e-01 6.48167908e-01
1.11401208e-01 -1.03598215e-01 5.14641702e-01 -2.92749345e-01
6.51093200e-02 2.82677442e-01 -6.92892432e-01 -3.83801192e-01
1.84404433e+00 -7.93246925e-02 -4.85918730e-01 -8.71909022e-01
-5.18053949e-01 1.38941336e+00 8.75432312e-01 2.21607700e-01
1.09622017e-01 -1.51290846e+00 -6.73052728e-01 -2.65547782e-01
-4.58990455e-01 -1.32891074e-01 8.58420059e-02 7.70734012e-01
-1.39657170e-01 4.12547082e-01 -3.29306781e-01 -1.76186875e-01
-1.02962422e+00 5.13638675e-01 4.38137114e-01 -4.31155860e-01
-3.80552709e-01 -1.77125171e-01 1.84050485e-01 -7.80247478e-03
-3.58948439e-01 -2.63009727e-01 1.57359138e-01 3.33039284e-01
2.04173401e-01 6.12897217e-01 -2.10068986e-01 -2.24625617e-01
-4.20053840e-01 -7.99196437e-02 1.92052066e-01 -1.93056121e-01
1.44619870e+00 -7.98585117e-01 -7.11731315e-01 -1.60198197e-01
6.37049377e-01 1.54087931e-01 -9.93452787e-01 6.67397305e-02
-1.84632450e-01 -6.36334538e-01 -7.00756162e-02 -8.83379161e-01
-1.65912127e+00 -1.23796649e-01 3.51729542e-01 1.12249148e+00
1.32307434e+00 -4.50788468e-01 8.59101951e-01 -4.39319275e-02
1.36032307e+00 -1.19863570e+00 -9.36951697e-01 -4.28768158e-01
2.67642647e-01 -8.11009943e-01 -2.19743289e-02 -2.89831012e-01
-5.12105823e-01 7.81259000e-01 6.91747218e-02 -3.52412574e-02
9.11909759e-01 2.50917405e-01 -3.54792684e-01 9.99057963e-02
-6.51359677e-01 4.65533048e-01 -9.03789937e-01 4.44935650e-01
-5.05016804e-01 8.59991431e-01 -1.15793967e+00 9.61299539e-01
-1.61508664e-01 5.57225108e-01 1.25407338e+00 1.21185648e+00
-2.26938084e-01 -1.24241889e+00 -1.37291387e-01 3.65263939e-01
-5.52858293e-01 3.11698485e-02 -2.12968662e-01 5.70152402e-01
-8.05296004e-02 1.15162575e+00 3.32112284e-03 3.75328630e-01
2.00370520e-01 -1.76125959e-01 1.45224676e-01 1.34255616e-02
-7.18161345e-01 1.99827805e-01 -2.73912191e-01 -1.25459477e-01
-3.99999857e-01 -7.72483230e-01 -1.54161346e+00 -7.72200823e-01
-5.45195282e-01 1.08141696e+00 4.19799864e-01 7.49868512e-01
2.05195546e-01 -1.93298683e-01 1.51274776e+00 -8.48021284e-02
-8.64723563e-01 -4.06160891e-01 -1.36902440e+00 -1.40955508e-01
5.24405912e-02 2.42179260e-01 -6.48372173e-01 -4.03974861e-01] | [5.596586227416992, 2.556356906890869] |
4927e23d-48c1-4c8a-afd0-b2e2086e94f2 | reversible-vision-transformers-1 | 2302.04869 | null | https://arxiv.org/abs/2302.04869v1 | https://arxiv.org/pdf/2302.04869v1.pdf | Reversible Vision Transformers | We present Reversible Vision Transformers, a memory efficient architecture design for visual recognition. By decoupling the GPU memory requirement from the depth of the model, Reversible Vision Transformers enable scaling up architectures with efficient memory usage. We adapt two popular models, namely Vision Transformer and Multiscale Vision Transformers, to reversible variants and benchmark extensively across both model sizes and tasks of image classification, object detection and video classification. Reversible Vision Transformers achieve a reduced memory footprint of up to 15.5x at roughly identical model complexity, parameters and accuracy, demonstrating the promise of reversible vision transformers as an efficient backbone for hardware resource limited training regimes. Finally, we find that the additional computational burden of recomputing activations is more than overcome for deeper models, where throughput can increase up to 2.3x over their non-reversible counterparts. Full code and trained models are available at https://github.com/facebookresearch/slowfast. A simpler, easy to understand and modify version is also available at https://github.com/karttikeya/minREV | ['Jitendra Malik', 'Christoph Feichtenhofer', 'Bo Xiong', 'Chao-yuan Wu', 'Yanghao Li', 'Haoqi Fan', 'Karttikeya Mangalam'] | 2023-02-09 | reversible-vision-transformers | http://openaccess.thecvf.com//content/CVPR2022/html/Mangalam_Reversible_Vision_Transformers_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Mangalam_Reversible_Vision_Transformers_CVPR_2022_paper.pdf | cvpr-2022-1 | ['video-classification'] | ['computer-vision'] | [ 2.14321554e-01 -2.11148053e-01 -1.66398168e-01 -1.28270537e-01
-7.34657645e-01 -7.21702099e-01 6.10010862e-01 -1.84307665e-01
-4.31580871e-01 1.08148195e-01 1.67069554e-01 -6.28402591e-01
3.76937538e-01 -6.10431731e-01 -7.43052542e-01 -6.78754389e-01
3.77967298e-01 -3.02333366e-02 2.35803410e-01 4.31903452e-02
4.96911705e-01 5.22071064e-01 -1.71442509e+00 7.23161101e-01
3.36207747e-01 9.93477821e-01 3.19175065e-01 1.17913580e+00
1.29420564e-01 1.19260454e+00 -1.52303159e-01 -2.57676214e-01
2.50611067e-01 5.16526215e-02 -9.03397024e-01 -3.30023728e-02
8.01014125e-01 -6.77754760e-01 -7.62379229e-01 7.91897595e-01
4.69320387e-01 -1.45616412e-01 3.31317008e-01 -8.91335011e-01
-1.05826724e+00 3.47602308e-01 -5.84005415e-01 6.49351478e-01
4.15993966e-02 5.37899613e-01 1.14335275e+00 -1.38730359e+00
2.53226638e-01 1.04761660e+00 7.68893361e-01 6.22770071e-01
-1.18454242e+00 -5.46137810e-01 8.40112194e-02 5.00405371e-01
-1.21979785e+00 -1.01703537e+00 1.95445925e-01 -3.83125991e-01
1.71188009e+00 4.62548077e-01 7.82743275e-01 1.10618401e+00
1.78307727e-01 7.54979372e-01 1.17165577e+00 -2.13173389e-01
-1.32032543e-01 -2.84909159e-01 5.52002370e-01 1.13593018e+00
1.69242978e-01 -9.98477638e-03 -6.19938016e-01 -1.53891407e-02
8.35869074e-01 2.74053425e-01 -2.41330191e-01 2.50941385e-02
-9.29048598e-01 5.68746507e-01 5.97243130e-01 -6.81983531e-02
-1.03928737e-01 7.03070104e-01 4.21821237e-01 2.97938466e-01
1.84069514e-01 2.27762945e-02 -2.61651158e-01 -2.96603739e-01
-7.54932702e-01 -1.62661031e-01 7.13148236e-01 9.95264351e-01
6.41218960e-01 3.48759294e-01 -1.09237544e-01 7.73471177e-01
1.90250561e-01 5.73989868e-01 4.98446107e-01 -1.22138786e+00
1.39574170e-01 5.76562285e-01 -1.85251027e-01 -5.15467644e-01
-3.42823893e-01 -4.73988235e-01 -9.04417634e-01 4.94329743e-02
4.12403136e-01 3.23396772e-01 -1.19564581e+00 1.35929430e+00
1.69047058e-01 1.63301095e-01 -2.60349214e-01 7.00135946e-01
1.20726395e+00 7.59899974e-01 2.28696279e-02 2.59553820e-01
1.74917376e+00 -1.43958545e+00 1.29746139e-01 -4.96540785e-01
6.22812629e-01 -9.60043907e-01 1.48445714e+00 2.10473210e-01
-1.35716546e+00 -3.87515634e-01 -1.14548326e+00 -8.82557809e-01
-2.53400832e-01 3.55663627e-01 8.26418221e-01 5.08471966e-01
-1.56728959e+00 5.15758038e-01 -1.30310869e+00 -5.21595418e-01
6.51521504e-01 6.37405455e-01 -1.56018227e-01 -1.84689283e-01
-4.08121586e-01 6.60527170e-01 -7.58092478e-02 1.33044347e-01
-1.21352994e+00 -7.95741916e-01 -4.27706331e-01 2.05300122e-01
-1.04479752e-01 -9.62926686e-01 1.59791231e+00 -8.57189000e-01
-1.39429021e+00 1.20110214e+00 -3.81975055e-01 -6.43298268e-01
4.02025610e-01 -1.21639974e-01 1.30100176e-01 3.98714133e-02
-3.83849025e-01 7.62144804e-01 9.48400080e-01 -6.55450642e-01
-5.26240528e-01 -2.17742309e-01 2.04152510e-01 1.88712746e-01
-5.90026915e-01 3.12749520e-02 -8.15135539e-01 -2.53128111e-01
-7.99994543e-02 -1.17072701e+00 2.71459832e-03 3.72044772e-01
-2.88677603e-01 3.43428589e-02 7.29261756e-01 -5.51487088e-01
8.52888882e-01 -2.10141683e+00 -1.20196797e-01 -1.33590922e-01
6.76549554e-01 3.87069345e-01 -3.16570818e-01 2.01164320e-01
3.74931209e-02 1.06922397e-02 -1.56094387e-01 -5.82625031e-01
-1.63134560e-01 1.87854424e-01 -3.06463718e-01 3.76274526e-01
3.45806144e-02 1.25480211e+00 -4.24341261e-01 -3.58454771e-02
2.13839337e-01 8.55870783e-01 -5.94740152e-01 -6.28624260e-02
1.73069328e-01 1.16173305e-01 -1.49048194e-01 8.09042215e-01
4.89243686e-01 -8.71890366e-01 2.26050794e-01 -4.51413810e-01
-2.45693535e-01 6.28000438e-01 -5.51711142e-01 1.38971579e+00
-5.34887671e-01 1.05602145e+00 1.15863375e-01 -8.10351014e-01
5.83590627e-01 -3.28832977e-02 2.34716758e-02 -9.65424895e-01
1.73873156e-01 6.09939806e-02 -1.65637955e-01 -1.97953627e-01
6.71715558e-01 2.87633777e-01 3.68184894e-01 4.61031199e-01
-5.74529134e-02 1.80550784e-01 1.24842495e-01 5.01309596e-02
1.46387064e+00 4.35457041e-04 5.13729192e-02 -3.26858521e-01
1.99845001e-01 1.73664585e-01 2.63107598e-01 9.74504530e-01
-1.43780187e-01 5.04422069e-01 2.16839299e-01 -6.68909967e-01
-1.37118661e+00 -1.14566493e+00 -8.47972110e-02 1.53685665e+00
-2.10205793e-01 -5.72266281e-01 -7.44628072e-01 5.50436378e-02
-1.24334432e-01 1.67771652e-01 -5.06250441e-01 -1.44014016e-01
-7.40678370e-01 -9.80220973e-01 6.93617761e-01 8.44575465e-01
7.62136877e-01 -7.05121696e-01 -8.03517640e-01 -1.21332884e-01
-6.66463599e-02 -1.14278436e+00 -4.98193622e-01 1.35853574e-01
-1.22718954e+00 -7.57106960e-01 -6.28516853e-01 -9.34841633e-01
6.56607509e-01 6.27765536e-01 1.26959765e+00 6.04241490e-01
-5.98514438e-01 5.52900791e-01 1.29874095e-01 -2.36337334e-02
-2.88531154e-01 3.54425162e-01 -1.45458445e-01 -4.33264613e-01
1.26815870e-01 -5.45901299e-01 -1.12916756e+00 1.44322857e-01
-5.69514692e-01 5.24078071e-01 6.17119551e-01 8.96256864e-01
6.12848580e-01 -6.57510340e-01 -1.26436755e-01 -7.74613500e-01
2.81700402e-01 -1.51402965e-01 -6.52283013e-01 1.69854820e-01
-7.07625210e-01 4.03110348e-02 4.22201753e-01 -5.52828431e-01
-7.55770087e-01 1.63358152e-01 -3.08741212e-01 -5.34223557e-01
1.92904070e-01 9.33680460e-02 5.13390541e-01 -3.85155886e-01
7.19984651e-01 4.97508496e-01 -7.43920729e-02 -4.52061534e-01
3.55000466e-01 3.91297162e-01 5.94384432e-01 -3.59464049e-01
4.80889499e-01 6.81989610e-01 -9.51808169e-02 -1.00059223e+00
-6.13932252e-01 -2.81123132e-01 -2.85545737e-01 1.76205650e-01
7.09766924e-01 -1.22718620e+00 -9.30351973e-01 6.43014729e-01
-1.03684986e+00 -8.10001791e-01 -1.87422097e-01 9.55341980e-02
-3.74899179e-01 2.81512022e-01 -1.27544630e+00 -4.28496659e-01
-1.01159596e+00 -1.20166981e+00 9.53163445e-01 2.21381322e-01
1.40607711e-02 -9.45364892e-01 -3.18192869e-01 6.33197725e-01
7.00245380e-01 -3.33950251e-01 9.78363335e-01 -8.65997002e-02
-1.01425731e+00 -5.00361286e-02 -6.32481873e-01 2.67697692e-01
-2.47120008e-01 1.84786558e-01 -1.31523335e+00 -6.49459541e-01
-6.96703270e-02 -4.10213977e-01 1.26660097e+00 2.06182510e-01
1.25139523e+00 -5.10897756e-01 -2.13275626e-01 9.19411421e-01
1.42513156e+00 -1.23515040e-01 8.61334085e-01 1.68879211e-01
9.47756529e-01 3.97883850e-04 -8.46453011e-02 3.70611340e-01
6.03523731e-01 4.39810783e-01 2.64685601e-01 -2.29880050e-01
-5.21534979e-01 -1.16557613e-01 6.26480043e-01 1.18397856e+00
-2.95832068e-01 -1.00895531e-01 -1.15086234e+00 4.32011187e-01
-1.62404299e+00 -8.55437458e-01 -1.60695374e-01 2.23992753e+00
6.87582850e-01 1.74418509e-01 1.23357482e-01 -3.33995894e-02
3.74716073e-01 1.52613834e-01 -7.64203668e-01 -6.38119996e-01
-2.46137427e-03 3.91208619e-01 8.04888427e-01 5.98700523e-01
-9.45840001e-01 1.11928272e+00 6.52152491e+00 8.55594218e-01
-1.46121335e+00 4.93866920e-01 8.68192375e-01 -4.36250567e-01
-1.55655757e-01 2.77346432e-01 -9.29549754e-01 2.01712236e-01
1.26087856e+00 6.46588802e-02 6.87915146e-01 7.93251872e-01
6.27613161e-03 4.55980971e-02 -9.77913678e-01 1.24078345e+00
-2.31950864e-01 -1.71116292e+00 1.44931629e-01 2.38703936e-01
4.63221014e-01 8.39846313e-01 5.43894589e-01 8.46781209e-02
1.93860069e-01 -1.01133931e+00 8.22954535e-01 3.03037524e-01
1.04148304e+00 -3.24737132e-01 1.37352154e-01 -3.54038253e-02
-1.39284182e+00 -3.47474575e-01 -6.80231273e-01 -2.68751889e-01
-3.98168176e-01 2.21205041e-01 -6.30770147e-01 -2.82554328e-01
1.03089201e+00 8.61610532e-01 -9.79354620e-01 8.64340782e-01
2.39261594e-02 7.71443903e-01 -4.19339269e-01 -1.86930951e-02
1.92214787e-01 9.55887809e-02 1.71587184e-01 1.44328296e+00
3.70927185e-01 1.24699086e-01 -1.08048894e-01 6.92149460e-01
-3.07343692e-01 -2.34465763e-01 -4.17456418e-01 -1.52673885e-01
5.85578144e-01 1.51211762e+00 -8.62266660e-01 -5.74484229e-01
-5.01833141e-01 1.26135647e+00 5.10905206e-01 2.63346046e-01
-9.22944129e-01 -2.16866896e-01 6.91410244e-01 1.79862335e-01
4.03304815e-01 -5.04897237e-01 -4.78101522e-01 -1.33972585e+00
-6.87722713e-02 -8.00175846e-01 1.82977200e-01 -8.07023406e-01
-7.93299437e-01 5.50523400e-01 -5.61451614e-01 -9.20699656e-01
1.41112566e-01 -9.62461531e-01 -6.72809184e-01 4.11743402e-01
-1.38373446e+00 -1.36282492e+00 -7.48995423e-01 6.20363176e-01
6.17172360e-01 9.05193239e-02 9.29324448e-01 3.91396850e-01
-8.87844741e-01 8.66628647e-01 2.50439912e-01 5.64803183e-02
3.07580620e-01 -8.95681620e-01 1.03882396e+00 8.35243702e-01
1.78816006e-01 7.30327308e-01 4.78987992e-01 -9.19102430e-02
-2.08326077e+00 -9.23916340e-01 6.00495696e-01 -5.81194997e-01
8.85614455e-01 -5.18631220e-01 -8.52281809e-01 9.26178396e-01
2.95857131e-01 8.02983269e-02 4.68654096e-01 5.05305361e-04
-8.79519045e-01 -1.97576463e-01 -8.27446103e-01 8.81187081e-01
1.36148870e+00 -9.57669973e-01 -1.31931379e-02 5.39150119e-01
4.67303276e-01 -5.22865415e-01 -8.50975156e-01 2.02322267e-02
8.34947228e-01 -1.05795228e+00 1.11218619e+00 -2.07870573e-01
3.40850949e-01 -1.31055236e-01 -2.04371735e-01 -5.26557624e-01
-6.51324570e-01 -6.47008181e-01 -4.04634416e-01 7.88651824e-01
4.70499337e-01 -9.19819713e-01 9.30663764e-01 5.86017072e-01
-9.58156288e-02 -8.76749933e-01 -1.00273120e+00 -5.12488961e-01
8.38810205e-02 -2.83418059e-01 1.35796130e-01 5.42194247e-01
-2.89570987e-01 5.88384032e-01 -1.65779650e-01 8.88844952e-02
5.97768843e-01 1.07177868e-01 5.15505373e-01 -6.17514789e-01
-6.80234611e-01 -6.58852816e-01 -3.70619863e-01 -1.50633395e+00
-2.61318952e-01 -1.03776598e+00 -3.84809256e-01 -1.41350210e+00
6.01509869e-01 -3.87985379e-01 -1.97634757e-01 9.55994308e-01
2.20022932e-01 9.80177045e-01 3.66757035e-01 6.52853727e-01
-5.68798184e-01 2.41776884e-01 1.05803251e+00 -2.83516079e-01
8.38553980e-02 -2.50420094e-01 -7.73242056e-01 6.26193047e-01
8.74975741e-01 -2.45275050e-01 -4.33066189e-01 -1.13813305e+00
-7.48543441e-03 -1.70276567e-01 6.75987661e-01 -1.18646479e+00
4.22692508e-01 2.26453289e-01 4.87161070e-01 -8.73589143e-02
7.16488421e-01 -3.07979971e-01 1.43916845e-01 6.94216192e-01
-2.07445890e-01 6.13441944e-01 5.15398264e-01 2.09497362e-01
2.97993273e-01 1.70136951e-02 1.01868963e+00 -2.79814243e-01
-8.46014917e-01 3.76785249e-01 -6.10702217e-01 -9.81308594e-02
6.38356924e-01 -3.83851230e-01 -1.02687812e+00 -2.65368849e-01
-7.01903284e-01 -1.76355690e-01 6.43368721e-01 2.53723472e-01
7.37247705e-01 -8.56083751e-01 -4.00955141e-01 1.64375231e-01
-4.37119193e-02 -3.46279740e-01 4.44359601e-01 1.15467608e+00
-8.89988840e-01 4.18740064e-01 -2.33432144e-01 -8.45020235e-01
-1.60793865e+00 4.08224583e-01 6.15673900e-01 -3.82678525e-04
-7.93760478e-01 1.08704245e+00 3.93131763e-01 4.24256809e-02
8.79477486e-02 -5.65020680e-01 3.23030710e-01 -2.91718662e-01
7.58036554e-01 5.62432885e-01 2.52461225e-01 -4.50042009e-01
-3.51933062e-01 8.06306481e-01 -4.27394897e-01 2.42019713e-01
1.17782748e+00 -4.78959754e-02 -2.11136222e-01 1.45649254e-01
1.20707369e+00 -3.87427211e-01 -1.40976441e+00 -1.67703718e-01
-2.06476629e-01 -1.08653009e-01 3.04562092e-01 -4.81317312e-01
-1.17742860e+00 1.16230035e+00 1.07776701e+00 -4.40450683e-02
1.22273302e+00 -9.60537940e-02 7.44400322e-01 5.68354964e-01
2.08034143e-01 -6.26136005e-01 3.91989574e-02 6.74756348e-01
7.10726023e-01 -1.00327539e+00 1.10199526e-01 -2.53050983e-01
-2.79100716e-01 1.02171564e+00 5.15068173e-01 -3.02631617e-01
5.87638795e-01 6.70194089e-01 6.47732057e-04 -1.25517830e-01
-1.19913912e+00 -1.14148371e-01 1.78695656e-02 3.70560020e-01
5.43653965e-01 1.75717864e-02 2.76364326e-01 -5.50069511e-02
-1.92850381e-01 -4.54151817e-02 4.63773996e-01 1.02701914e+00
-4.20598835e-01 -9.18986380e-01 -4.78658155e-02 6.07179999e-01
-3.60924989e-01 -6.02920413e-01 -2.24861518e-01 3.11933845e-01
-4.13758725e-01 7.81582475e-01 3.97941172e-01 -5.06569505e-01
2.88745649e-02 -1.41044587e-01 7.09343433e-01 -4.51979727e-01
-7.52244771e-01 -9.67668518e-02 -2.16161422e-02 -6.94309175e-01
-8.08795467e-02 -1.94779903e-01 -1.22542202e+00 -1.07368934e+00
2.79658269e-02 -4.70587105e-01 6.08340383e-01 5.99691212e-01
8.69185030e-01 2.39647269e-01 2.72591561e-01 -1.13312364e+00
-6.05139613e-01 -6.08279228e-01 2.97249053e-02 -1.07037574e-01
4.32646066e-01 -2.11390167e-01 -2.58273989e-01 1.61981687e-01] | [9.378416061401367, 1.6451135873794556] |
1dae593d-20c6-40fc-bf86-b5760cb1ad4a | yimmon-at-semeval-2019-task-9-suggestion | null | null | https://aclanthology.org/S19-2222 | https://aclanthology.org/S19-2222.pdf | Yimmon at SemEval-2019 Task 9: Suggestion Mining with Hybrid Augmented Approaches | Suggestion mining task aims to extract tips, advice, and recommendations from unstructured text. The task includes many challenges, such as class imbalance, figurative expressions, context dependency, and long and complex sentences. This paper gives a detailed system description of our submission in SemEval 2019 Task 9 Subtask A. We transfer Self-Attention Network (SAN), a successful model in machine reading comprehension field, into this task. Our model concentrates on modeling long-term dependency which is indispensable to parse long and complex sentences. Besides, we also adopt techniques, such as contextualized embedding, back-translation, and auxiliary loss, to augment the system. Our model achieves a performance of F1=76.3, and rank 4th among 34 participating systems. Further ablation study shows that the techniques used in our system are beneficial to the performance. | ['Yimeng Zhuang'] | 2019-06-01 | null | null | null | semeval-2019-6 | ['suggestion-mining'] | ['natural-language-processing'] | [ 3.19227368e-01 3.55418503e-01 -2.67230779e-01 -6.20810509e-01
-8.13170791e-01 -4.32743609e-01 4.11791921e-01 2.85533369e-01
-5.39141178e-01 7.21780837e-01 7.55279899e-01 -7.20560849e-01
4.40390594e-02 -4.48069453e-01 -5.27823210e-01 -1.90076515e-01
5.88384159e-02 3.19426417e-01 -4.77470234e-02 -6.24041736e-01
7.09091663e-01 -7.73970336e-02 -1.26627910e+00 6.65341079e-01
1.21406734e+00 6.66402757e-01 6.12661131e-02 9.54898655e-01
-4.46603030e-01 9.38151777e-01 -6.77359223e-01 -7.07789421e-01
-2.05818281e-01 -5.58178902e-01 -1.14647555e+00 -6.42176270e-01
6.49159789e-01 -2.91316181e-01 3.84934321e-02 8.88490736e-01
7.54567802e-01 2.00999498e-01 9.36683059e-01 -8.59679282e-01
-9.24830258e-01 1.13658941e+00 -4.66383368e-01 5.47601104e-01
4.08948004e-01 -5.75939044e-02 1.42445290e+00 -1.05365002e+00
2.21598953e-01 1.15940058e+00 5.15287101e-01 6.79310083e-01
-8.88114154e-01 -5.22134542e-01 4.72662896e-01 5.97907424e-01
-7.73492754e-01 -5.75212240e-01 8.98104668e-01 -1.70193478e-01
1.45532894e+00 5.81064999e-01 2.00061217e-01 1.46677721e+00
3.59105289e-01 1.08958030e+00 9.95657265e-01 -7.66637385e-01
-1.98609114e-01 2.38099471e-01 8.50866616e-01 4.77147967e-01
9.76833180e-02 -2.90697455e-01 -5.23403764e-01 5.59785403e-02
-1.09504722e-01 -2.44443849e-01 -3.29938054e-01 5.21929085e-01
-9.61079419e-01 7.63356507e-01 2.70395249e-01 1.85118273e-01
-2.37633064e-01 -2.20610976e-01 4.91533399e-01 5.61615407e-01
6.41888976e-01 9.02804732e-01 -8.74684691e-01 -1.84521303e-01
-5.71162701e-01 1.59453303e-01 1.01885557e+00 8.87138903e-01
3.03347528e-01 -1.93139404e-01 -8.72109532e-01 1.10931456e+00
-3.18823233e-02 5.05360246e-01 5.98065138e-01 -7.25611567e-01
9.47976589e-01 5.41736424e-01 -5.02608065e-03 -1.05753505e+00
-5.73487043e-01 -8.19159925e-01 -1.11763716e+00 -3.29937488e-01
-1.55085728e-01 -4.53185737e-01 -6.28871977e-01 1.68272221e+00
-7.57945403e-02 -1.49929255e-01 2.92762760e-02 6.30401850e-01
1.29208231e+00 5.72345316e-01 7.82600865e-02 -2.37980828e-01
1.29478633e+00 -1.76336420e+00 -8.22166800e-01 -5.00915587e-01
7.19966888e-01 -8.56843293e-01 1.56810844e+00 4.49330986e-01
-1.13588572e+00 -6.24471903e-01 -1.06207812e+00 -5.41461706e-01
-2.78216898e-01 3.43799978e-01 2.87495643e-01 2.95496821e-01
-9.14531946e-01 5.37676632e-01 -3.86117637e-01 -2.04224169e-01
3.34642351e-01 2.14373574e-01 -2.73257997e-02 1.80908665e-02
-1.35144365e+00 1.04388058e+00 5.25159687e-02 1.02157108e-01
-1.67063430e-01 -6.67410195e-01 -7.22750664e-01 4.28708643e-01
4.45270717e-01 -9.33339775e-01 1.44054043e+00 -6.74710929e-01
-1.70015669e+00 5.91704786e-01 -4.19916719e-01 -4.37360048e-01
1.42529592e-01 -8.82439554e-01 -3.74550164e-01 -1.50681764e-01
4.41630036e-02 3.99123073e-01 7.17410684e-01 -5.40824294e-01
-4.97150391e-01 -2.23087311e-01 6.52782097e-02 4.95267034e-01
-5.58771908e-01 2.43642986e-01 -2.71638662e-01 -6.07426643e-01
-3.81093442e-01 -7.02257812e-01 -2.83154905e-01 -8.94753397e-01
-8.39369297e-01 -8.43353093e-01 8.77787843e-02 -1.10823143e+00
1.56696367e+00 -2.00442910e+00 1.78611264e-01 -4.32924703e-02
1.71869144e-01 2.37571448e-01 -4.80362684e-01 3.98804277e-01
1.12473713e-02 4.57382023e-01 -1.36624739e-01 -3.91723812e-01
2.50347108e-01 -1.78811744e-01 -4.11410630e-01 -1.40486270e-01
3.70027035e-01 1.17947793e+00 -7.65129507e-01 -4.18458194e-01
-2.42984638e-01 3.59251574e-02 -6.03654385e-01 4.83462542e-01
-3.56591165e-01 2.26928979e-01 -5.47699511e-01 3.91438693e-01
4.03967291e-01 -2.02188209e-01 -1.86307326e-01 -3.12243775e-02
5.52060045e-02 9.80141938e-01 -4.92041737e-01 1.60610211e+00
-6.96995378e-01 3.15919340e-01 -1.57459855e-01 -9.12380457e-01
7.68062413e-01 1.07665509e-02 -7.71163926e-02 -9.25829530e-01
4.15702239e-02 -1.34447560e-01 2.71718532e-01 -9.46813345e-01
5.41865230e-01 4.46098059e-01 -4.29232232e-02 7.29964674e-01
-1.74928635e-01 5.75792193e-02 4.25584495e-01 4.76692945e-01
1.29114902e+00 -2.02872232e-02 2.66112864e-01 -3.26544642e-01
7.43327498e-01 -3.08023334e-01 3.73166740e-01 8.78772140e-01
3.12435608e-02 4.84480560e-01 8.27024043e-01 -1.95762083e-01
-8.21724355e-01 -4.54159588e-01 1.43219352e-01 1.57834053e+00
-2.39753053e-01 -7.71582305e-01 -5.47948599e-01 -1.15924335e+00
-1.78805336e-01 1.17573023e+00 -7.05841064e-01 -2.28799880e-01
-8.35183978e-01 -8.05502713e-01 3.42605948e-01 7.34513998e-01
3.66215497e-01 -1.39541876e+00 -2.65836775e-01 2.82340348e-01
-4.82313454e-01 -9.39359605e-01 -7.05322266e-01 1.59479871e-01
-8.49388540e-01 -9.35696363e-01 -3.33958477e-01 -8.78969073e-01
5.63314855e-01 -5.25157936e-02 1.65370357e+00 3.12604100e-01
-1.46029620e-02 2.47316360e-02 -7.12400138e-01 -6.92906439e-01
-8.60752314e-02 6.33116782e-01 -1.77285597e-01 -5.03623188e-01
6.12241626e-01 -5.11781931e-01 -6.10024750e-01 -2.29956638e-02
-3.64919215e-01 1.02425054e-01 9.11753535e-01 1.21729290e+00
2.65031427e-01 -3.82103503e-01 9.36464727e-01 -1.49550200e+00
1.33957148e+00 -4.81236339e-01 -1.88151207e-02 5.08583844e-01
-1.15353024e+00 -4.64613885e-02 9.87834692e-01 2.62838136e-02
-1.13756275e+00 -5.00546396e-01 -7.20862389e-01 3.88422012e-01
-2.73543566e-01 7.00173497e-01 -1.18077077e-01 3.89226377e-01
9.72645938e-01 1.32261127e-01 -2.34837070e-01 -8.02085519e-01
4.51780528e-01 9.10404980e-01 4.07455623e-01 -4.14814889e-01
4.42321062e-01 -3.70175689e-01 -4.05437171e-01 -6.78809762e-01
-1.50385499e+00 -4.64867085e-01 -4.69654620e-01 5.17259896e-01
4.40118790e-01 -5.92086017e-01 -5.32474756e-01 5.36552928e-02
-1.32360947e+00 -2.35376939e-01 -1.91669360e-01 2.14770913e-01
-1.02096848e-01 4.34546709e-01 -6.82286084e-01 -5.73989809e-01
-1.33598900e+00 -7.69012630e-01 6.48757577e-01 3.58052820e-01
-4.84309405e-01 -8.30204785e-01 1.55691564e-01 7.87113965e-01
7.69981027e-01 -2.74422675e-01 1.26532567e+00 -9.95439053e-01
-4.81217243e-02 6.32451549e-02 -1.86466590e-01 6.16265893e-01
-4.59259711e-02 -1.37357086e-01 -8.89721513e-01 -8.06169137e-02
-9.95745361e-02 -6.13638580e-01 1.24871457e+00 1.48484930e-01
1.71496701e+00 -6.28394783e-01 -1.00442529e-01 4.60358679e-01
7.58865535e-01 5.75163811e-02 5.22846758e-01 2.13243723e-01
7.38352835e-01 8.18222404e-01 6.39180005e-01 3.12100779e-02
6.87865138e-01 4.01348948e-01 1.08273126e-01 1.03703104e-02
-1.34154662e-01 -2.64540285e-01 5.62251985e-01 1.45957792e+00
1.29938886e-01 -5.39910674e-01 -7.90295422e-01 4.13228244e-01
-1.78098917e+00 -7.22056210e-01 -2.37022400e-01 1.66104424e+00
1.02576602e+00 3.86334002e-01 -1.10121861e-01 -3.00280992e-02
2.16341019e-01 1.83841258e-01 -6.89766467e-01 -1.09171903e+00
-1.13519222e-01 5.17069101e-01 -7.89716393e-02 6.62120879e-01
-1.12077105e+00 1.06032944e+00 6.05591106e+00 5.91952622e-01
-8.81500661e-01 1.18640848e-01 8.89277458e-01 1.23164497e-01
-5.22232354e-01 -2.79903591e-01 -8.75980616e-01 4.56487775e-01
1.10779905e+00 -1.61000147e-01 2.11765796e-01 6.64381802e-01
5.28030284e-02 1.44083828e-01 -1.12236392e+00 5.11767030e-01
5.75105846e-01 -1.09996104e+00 -2.15581302e-02 -5.78033507e-01
7.03569293e-01 1.34948209e-01 1.62333444e-01 9.56461549e-01
1.12972461e-01 -1.30587995e+00 1.59340519e-02 3.37833822e-01
6.58882916e-01 -7.04917729e-01 1.29130185e+00 5.50315082e-01
-3.93191755e-01 -6.85438216e-02 -3.45041901e-01 -2.85172611e-01
3.17831174e-03 6.30594313e-01 -7.59564161e-01 5.53007782e-01
6.74103379e-01 9.40137267e-01 -8.29479277e-01 8.13314199e-01
-8.80326390e-01 1.14244986e+00 -5.56601584e-02 -5.84386945e-01
-1.87753290e-02 -1.86060220e-01 4.16387856e-01 1.49689043e+00
-1.22720405e-01 9.62401405e-02 1.05803367e-02 4.36034411e-01
-4.73223209e-01 6.93017662e-01 -2.85046786e-01 -7.95244500e-02
2.11603537e-01 1.17886686e+00 -2.22054020e-01 -4.09537375e-01
-3.00375044e-01 9.96818364e-01 7.99605548e-01 4.20444310e-01
-4.53776568e-01 -7.34724641e-01 2.06931889e-01 -2.99452126e-01
2.32740715e-01 1.39595613e-01 -6.93872571e-01 -1.32311499e+00
3.56695622e-01 -1.19724917e+00 3.85670334e-01 -4.32827264e-01
-1.48880684e+00 8.42523277e-01 -3.77009511e-01 -7.40182638e-01
-2.77841896e-01 -4.01092559e-01 -1.05912173e+00 9.02815342e-01
-1.78488338e+00 -9.31936562e-01 -1.01035275e-01 1.78476870e-01
9.09972608e-01 -2.80029953e-01 1.02592909e+00 3.20002139e-01
-8.94882798e-01 1.06143939e+00 5.82077242e-02 9.85279586e-03
1.04356134e+00 -1.67677128e+00 6.69328451e-01 8.24654698e-01
1.76241934e-01 8.84278595e-01 6.00389481e-01 -3.91424268e-01
-1.08103549e+00 -1.07509100e+00 1.63910162e+00 -7.48610735e-01
4.00672287e-01 -3.42210799e-01 -8.95371735e-01 5.97427607e-01
7.48216391e-01 -6.26031935e-01 9.20234501e-01 8.93240452e-01
-2.34974280e-01 -2.89025873e-01 -6.90782487e-01 6.92948937e-01
1.05243421e+00 -3.10785383e-01 -9.71633911e-01 5.10258675e-01
1.07427824e+00 -4.24768806e-01 -5.51441848e-01 5.88723481e-01
3.85239989e-01 -8.32538545e-01 7.69118369e-01 -1.12394893e+00
1.11210823e+00 2.84787208e-01 3.15619230e-01 -1.67570245e+00
-6.98199809e-01 -6.51231647e-01 -3.90703291e-01 1.29684615e+00
9.84981775e-01 -3.81538391e-01 6.93630457e-01 4.45049107e-01
-4.37294811e-01 -1.24642992e+00 -4.60805982e-01 -2.70350277e-01
4.33712572e-01 -1.80676505e-01 5.73151171e-01 8.14260602e-01
2.04858184e-01 1.27364016e+00 -4.48917925e-01 -3.03432822e-01
2.44148850e-01 4.40459937e-01 7.16075122e-01 -9.75885868e-01
-3.77486885e-01 -5.59643805e-01 5.03897846e-01 -1.37623739e+00
4.47861165e-01 -9.94964242e-01 -6.19508501e-04 -1.74657559e+00
3.36568862e-01 -3.19899350e-01 -7.49741852e-01 5.04747868e-01
-8.05982530e-01 -1.02880962e-01 -6.01245128e-02 -1.45424679e-01
-8.85303736e-01 7.54318476e-01 1.32239723e+00 -3.74070197e-01
-1.02353096e-01 3.79737496e-01 -1.39904642e+00 6.22511566e-01
1.20574260e+00 -4.44285959e-01 -4.67159390e-01 -7.82479763e-01
4.58352298e-01 -2.66868800e-01 -2.42961973e-01 -2.51703620e-01
2.36951873e-01 -6.60659075e-02 2.20359430e-01 -6.34464860e-01
1.32893264e-01 -3.79791260e-01 -8.34372044e-01 3.01367074e-01
-8.46362889e-01 3.87977600e-01 1.02432795e-01 4.10236269e-01
-1.75816730e-01 -3.90422642e-01 2.34746799e-01 4.45098430e-03
-2.79628992e-01 3.99531536e-02 -1.97678626e-01 4.66573060e-01
2.91117400e-01 5.21431208e-01 -6.52010202e-01 -4.66096312e-01
-3.85146379e-01 6.59826219e-01 -1.86676517e-01 6.63005590e-01
6.29839182e-01 -7.78067112e-01 -1.22256434e+00 1.45505309e-01
9.11308527e-02 4.16441076e-02 2.95106679e-01 9.45367396e-01
-2.64892131e-01 6.73817754e-01 -4.57489975e-02 -1.03164874e-01
-1.29845583e+00 2.30676427e-01 -5.79348616e-02 -6.28651023e-01
-6.00869298e-01 1.27836382e+00 -1.11387096e-01 -7.02890575e-01
4.71807390e-01 -4.46587175e-01 -9.03167009e-01 -4.27024178e-02
6.77049220e-01 3.58899176e-01 3.04225713e-01 7.08553642e-02
-2.78027236e-01 5.07705510e-01 -6.35791898e-01 1.88778386e-01
1.33567274e+00 -1.45404443e-01 -4.17663366e-01 2.82341719e-01
9.72924173e-01 3.26606184e-01 -5.04691720e-01 -2.69605279e-01
1.70148343e-01 3.12236734e-02 -5.40024973e-02 -1.48569810e+00
-4.21805888e-01 1.09753966e+00 8.83150399e-02 5.03459759e-02
1.04935169e+00 -2.05052927e-01 1.09205294e+00 8.79013419e-01
-3.38904768e-01 -1.07626677e+00 -3.00257728e-02 1.05776751e+00
1.21390522e+00 -1.62667668e+00 2.45381612e-02 -2.00561479e-01
-6.46730661e-01 8.65346849e-01 1.08100319e+00 1.55711891e-02
5.35245121e-01 7.77953267e-02 1.47243172e-01 2.71186046e-02
-1.36745465e+00 2.10305125e-01 6.67325377e-01 1.30785406e-01
8.35606933e-01 5.17649315e-02 -8.72476459e-01 1.13675773e+00
-6.54685259e-01 -2.48817265e-01 4.41858470e-01 6.51496828e-01
-3.10194999e-01 -1.17534268e+00 2.07237184e-01 9.39418197e-01
-7.49934018e-01 -5.78857958e-01 -6.90242052e-01 1.82993948e-01
-6.29376620e-02 1.10301125e+00 -1.82564378e-01 -5.06393254e-01
5.10611176e-01 1.54089883e-01 2.28952676e-01 -8.29408586e-01
-9.92297173e-01 -2.81000882e-01 5.71569324e-01 -3.25029284e-01
-9.05011594e-03 -3.39984417e-01 -9.02950346e-01 -1.14340149e-01
-3.33541989e-01 4.72613484e-01 6.94755375e-01 9.18314576e-01
7.51305938e-01 7.36965537e-01 6.54693663e-01 -1.22156208e-02
-8.96279454e-01 -1.66132343e+00 -1.30650308e-02 4.23506945e-01
4.30804402e-01 -1.33906662e-01 -4.37403381e-01 -2.01500624e-01] | [11.128344535827637, 8.372991561889648] |
4d00265a-2dd4-4727-ab3b-63dcdfed5917 | cist-cl-scisumm-2020-longsumm-2020-automatic | null | null | https://aclanthology.org/2020.sdp-1.25 | https://aclanthology.org/2020.sdp-1.25.pdf | CIST@CL-SciSumm 2020, LongSumm 2020: Automatic Scientific Document Summarization | Our system participates in two shared tasks, CL-SciSumm 2020 and LongSumm 2020. In the CL-SciSumm shared task, based on our previous work, we apply more machine learning methods on position features and content features for facet classification in Task1B. And GCN is introduced in Task2 to perform extractive summarization. In the LongSumm shared task, we integrate both the extractive and abstractive summarization ways. Three methods were tested which are T5 Fine-tuning, DPPs Sampling, and GRU-GCN/GAT. | ['Xingyuan Li', 'Siya Qi', 'Yafei Jiang', 'Yinan Liu', 'Wei Liu', 'Yang Xie', 'Lei LI'] | null | null | null | null | emnlp-sdp-2020-11 | ['scientific-article-summarization'] | ['natural-language-processing'] | [ 3.71229947e-01 5.49021661e-02 -4.74084646e-01 -4.20124270e-03
-1.54507983e+00 -6.98250115e-01 9.68909562e-01 3.71747792e-01
-6.60747409e-01 1.54817080e+00 1.12922537e+00 -1.41151488e-01
-4.35842425e-01 -4.60453510e-01 -4.11190897e-01 -4.48813260e-01
7.11646751e-02 4.38272834e-01 3.52326453e-01 -1.59141183e-01
8.49103212e-01 -8.32339078e-02 -1.37829530e+00 1.04558945e+00
1.47003949e+00 1.00678182e+00 3.32648665e-01 9.23391700e-01
-6.47424936e-01 7.59267986e-01 -9.19259310e-01 -4.43212837e-01
-8.62863138e-02 -7.09960699e-01 -1.12931776e+00 -3.82585198e-01
3.89214963e-01 1.80362724e-02 -8.21051896e-02 1.05625498e+00
8.25169444e-01 8.45581889e-02 1.21018684e+00 -9.50619161e-01
-3.69430929e-01 8.65776360e-01 -7.54656374e-01 4.39351350e-01
7.69553244e-01 -2.56928980e-01 1.30086005e+00 -8.56166303e-01
9.26932454e-01 1.08017671e+00 4.68809038e-01 4.68479805e-02
-7.50270665e-01 -5.35751045e-01 8.65039006e-02 5.05850017e-01
-9.39179361e-01 -3.92565966e-01 4.48105484e-01 -2.35925555e-01
1.12453651e+00 9.96164143e-01 6.76256895e-01 1.04423785e+00
5.38074553e-01 1.15681767e+00 1.14184797e+00 -3.53563160e-01
2.35561848e-01 -3.74364525e-01 3.97066087e-01 2.10845783e-01
5.64415991e-01 -5.34204006e-01 -7.95723259e-01 -3.54334235e-01
3.52134630e-02 -1.89921007e-01 -3.36743057e-01 4.41307396e-01
-1.34422791e+00 8.36508393e-01 6.91662654e-02 2.36794904e-01
-5.17080665e-01 3.20257526e-03 7.95264244e-01 3.46104711e-01
1.15150332e+00 8.85842681e-01 -5.80411792e-01 -2.85827428e-01
-1.65131724e+00 6.81548774e-01 8.25607061e-01 8.42826426e-01
5.46160877e-01 -5.13193130e-01 -1.19788778e+00 1.04186058e+00
-1.55302957e-01 1.77985132e-01 8.55401218e-01 -1.21875668e+00
7.85195410e-01 4.98321772e-01 -6.51121289e-02 -2.67198414e-01
-3.46872687e-01 -6.10032260e-01 -9.07621026e-01 -3.34904253e-01
-3.25082451e-01 -3.98608923e-01 -1.00723183e+00 9.75053370e-01
-6.72506019e-02 1.24319755e-01 -6.00968301e-02 6.01100087e-01
1.37770426e+00 6.27162695e-01 -3.09715092e-01 -8.00155878e-01
1.36616516e+00 -1.48042524e+00 -1.24474323e+00 6.09331615e-02
9.01608527e-01 -1.15093184e+00 6.29563749e-01 5.71023226e-01
-1.34418249e+00 -9.87389758e-02 -1.04289341e+00 -2.61400282e-01
-6.41883731e-01 1.41409636e-01 4.31802243e-01 7.20813572e-02
-8.43227029e-01 9.80151117e-01 -1.35057718e-01 -2.63115615e-01
5.55259883e-01 -1.05100445e-01 -2.44193330e-01 -1.19332351e-01
-1.42609906e+00 9.04388309e-01 8.79014730e-01 -2.76967734e-01
-4.63294357e-01 -7.38763928e-01 -4.56621528e-01 -7.88878370e-03
5.82003951e-01 -1.00531209e+00 1.21285200e+00 -3.52296203e-01
-1.32079780e+00 9.01217401e-01 -1.62591219e-01 -4.37889546e-01
5.24983048e-01 -4.43719357e-01 -8.09469074e-02 3.83143164e-02
1.60468265e-01 3.51501971e-01 3.29500765e-01 -6.78018510e-01
-8.75118136e-01 -2.74581730e-01 -1.90650016e-01 6.25448406e-01
-8.37496594e-02 2.65869588e-01 -4.10766572e-01 -9.09609020e-01
7.35963061e-02 -6.02489352e-01 -1.36178449e-01 -7.60060370e-01
-8.11054826e-01 -6.29418254e-01 4.16188747e-01 -9.83785987e-01
1.62527716e+00 -1.55989289e+00 3.25469851e-01 -2.05170020e-01
1.34744421e-01 3.99300128e-01 -3.08394969e-01 9.18847680e-01
1.69511527e-01 3.27598631e-01 -6.15652911e-02 -3.50807309e-01
6.42660409e-02 -7.94683397e-02 -2.65099525e-01 -3.12977917e-02
-1.39694661e-01 1.05516541e+00 -8.63941252e-01 -8.25118959e-01
-3.15903783e-01 -4.04799104e-01 -2.58018315e-01 2.19332173e-01
-2.93647259e-01 -3.94253507e-02 -4.79413509e-01 5.58266938e-01
6.44189775e-01 -6.77513182e-02 -2.42629778e-02 -2.00073957e-01
-3.53939652e-01 9.12583888e-01 -8.63391101e-01 1.90962005e+00
-5.84225468e-02 2.77174830e-01 -9.67067853e-02 -8.85510504e-01
5.23961484e-01 2.95456916e-01 3.44846874e-01 -4.72490489e-01
1.29186526e-01 5.42313755e-01 -2.38852754e-01 -5.64077973e-01
1.05779541e+00 4.63045925e-01 -3.16616774e-01 2.59204894e-01
3.26499194e-01 -4.11824882e-01 8.57370377e-01 6.19975805e-01
1.25493264e+00 1.20068379e-01 7.43501961e-01 -5.69385052e-01
4.19517547e-01 -4.88036275e-02 4.93929207e-01 9.67105508e-01
7.45624080e-02 9.71321881e-01 8.13191831e-01 3.08546927e-02
-8.89541507e-01 -8.25431108e-01 -1.94267686e-02 1.37467670e+00
-1.54412135e-01 -8.99586320e-01 -6.47783518e-01 -7.98297286e-01
-8.54949132e-02 9.35958922e-01 -6.03407323e-01 -2.46210471e-01
-2.98225194e-01 -1.02038693e+00 6.70961559e-01 9.72397551e-02
6.02356195e-01 -1.15128839e+00 -1.09374151e-02 2.08720744e-01
-7.35992551e-01 -6.64097011e-01 -8.28744709e-01 3.61392498e-01
-7.67468512e-01 -1.04939008e+00 -1.37284160e+00 -5.39576352e-01
-1.11317202e-01 2.40579143e-01 9.56816852e-01 -4.39532340e-01
-1.66490361e-01 -1.90456867e-01 -7.79660940e-01 -5.91211140e-01
-1.76530145e-03 8.07134986e-01 -4.02811468e-01 -4.71845269e-01
2.80007850e-02 -5.41784167e-01 -5.85358024e-01 -4.30300832e-01
-5.12135684e-01 2.83374637e-01 7.77881444e-01 5.81872463e-01
6.69528902e-01 -5.47272384e-01 8.61561179e-01 -1.14160991e+00
1.37755132e+00 -5.88998854e-01 1.79284945e-01 4.51933414e-01
-6.87127352e-01 1.20917335e-01 1.65761098e-01 -7.90795758e-02
-8.32964003e-01 -6.67254627e-01 -3.61248046e-01 9.87767801e-02
3.62899214e-01 1.10794139e+00 -7.07471520e-02 4.34840620e-01
4.08111066e-01 6.01778269e-01 -1.36570260e-01 -8.35226238e-01
3.25220287e-01 1.00723529e+00 4.58858818e-01 -2.21228391e-01
2.20628545e-01 -3.11680794e-01 -2.98605472e-01 -8.27942252e-01
-1.37322295e+00 -5.71278393e-01 -1.70903102e-01 7.85342902e-02
8.51468325e-01 -8.81871879e-01 -2.81025678e-01 5.00367403e-01
-1.30030942e+00 4.00463715e-02 -4.59688097e-01 4.75277126e-01
-5.57641149e-01 6.36860490e-01 -5.58048725e-01 -6.07596755e-01
-1.10949576e+00 -6.65531516e-01 1.11656535e+00 1.04385458e-01
-3.40361506e-01 -4.61410195e-01 3.88014078e-01 5.57992399e-01
4.02170092e-01 1.44503474e-01 8.46495926e-01 -1.25935805e+00
1.04795014e-02 -1.18361905e-01 -2.26793274e-01 3.67733061e-01
5.84454089e-02 -2.28808179e-01 -5.61254144e-01 -7.02324212e-02
-3.31044465e-01 -5.59498668e-01 1.78695858e+00 6.73617363e-01
1.39049339e+00 -7.24824905e-01 -3.19645852e-01 4.59044635e-01
9.53072608e-01 -1.17146134e-01 9.31416035e-01 3.74236315e-01
4.19243902e-01 4.86911744e-01 7.88367450e-01 4.14367139e-01
3.26023042e-01 4.45190042e-01 1.56832039e-01 4.37412679e-01
-3.58435661e-01 -1.29318461e-02 3.13656271e-01 1.37494528e+00
-3.10581505e-01 -5.06432474e-01 -4.69650984e-01 4.53712523e-01
-2.10541224e+00 -9.66476381e-01 -3.92574102e-01 1.78295565e+00
1.11048818e+00 2.04701856e-01 1.56419203e-01 1.78803593e-01
4.35290515e-01 6.38369799e-01 -2.33002588e-01 -5.80949247e-01
-3.28032762e-01 1.79337397e-01 5.35746574e-01 3.19341600e-01
-1.06460321e+00 8.56131256e-01 6.40316916e+00 1.54116893e+00
-4.98810291e-01 1.74816102e-01 3.43816847e-01 -5.12850106e-01
-4.26705748e-01 -4.68357094e-02 -7.88311660e-01 9.04904664e-01
8.21545660e-01 -6.16795719e-01 1.47783831e-01 3.36419702e-01
9.16916430e-02 -4.71535653e-01 -5.36304593e-01 6.28440261e-01
2.77093053e-01 -1.59967566e+00 3.45924318e-01 1.00684345e-01
9.48374271e-01 3.36062610e-01 -2.06061199e-01 7.47249246e-01
2.84651995e-01 -7.91333020e-01 4.13904101e-01 8.10438037e-01
8.55434895e-01 -8.62152100e-01 1.24171245e+00 6.49377584e-01
-9.93279278e-01 2.67120153e-01 -3.34867120e-01 -8.97165481e-03
1.67466149e-01 9.12084222e-01 -2.78038383e-01 1.38645983e+00
3.81118327e-01 6.35567009e-01 -5.59858322e-01 1.35471630e+00
-1.75371021e-01 7.42287695e-01 1.71224698e-02 -3.19272935e-01
4.33884680e-01 -2.33621791e-01 7.95389354e-01 1.49795926e+00
5.74299991e-01 -7.56197423e-02 3.76073360e-01 3.35994750e-01
-3.72070730e-01 3.65696371e-01 -2.29839697e-01 -6.50277212e-02
4.47053939e-01 1.34336650e+00 -4.01094854e-01 -5.39797306e-01
9.11797434e-02 1.06983709e+00 1.04767241e-01 2.78782427e-01
-6.37473822e-01 -6.19215369e-01 2.95144349e-01 -1.81580886e-01
3.89390200e-01 2.59021699e-01 -2.92345971e-01 -1.20155036e+00
-2.96688080e-01 -8.53066206e-01 5.62708378e-01 -6.54346824e-01
-1.36917138e+00 5.06563306e-01 1.65158078e-01 -9.32101786e-01
-3.30603778e-01 -2.91782200e-01 -9.35853720e-01 8.06536853e-01
-1.60998070e+00 -1.14057410e+00 1.40220404e-01 1.32084548e-01
8.58793139e-01 -3.47730786e-01 7.19467819e-01 1.36012211e-01
-5.00605345e-01 5.82113743e-01 5.99364221e-01 -3.49393755e-01
1.06932855e+00 -1.30391192e+00 2.43875653e-01 2.85518050e-01
-1.72497392e-01 2.19210446e-01 7.51306057e-01 -8.78608465e-01
-8.49135220e-01 -1.37119555e+00 1.23571706e+00 1.90759599e-01
4.96981382e-01 -1.35860741e-01 -7.08015263e-01 3.90946954e-01
6.39331162e-01 -4.87913519e-01 6.62914693e-01 3.58888060e-01
1.86990444e-02 1.37493417e-01 -7.95718133e-01 3.28675836e-01
1.09482753e+00 -1.38840765e-01 -1.05813241e+00 6.94233656e-01
9.27323818e-01 -3.74027342e-01 -7.97970712e-01 5.07036686e-01
4.72662210e-01 -4.96900797e-01 7.13783920e-01 -6.77355051e-01
9.64655697e-01 1.57539651e-01 -2.91217744e-01 -1.80611968e+00
-4.94276553e-01 -5.69124818e-01 -5.22698522e-01 1.40339243e+00
6.22309268e-01 -5.18446982e-01 5.31952024e-01 -4.17694658e-01
-5.07534742e-01 -1.08609712e+00 -9.80731666e-01 -9.24032450e-01
5.09517789e-01 2.55071428e-02 4.22476470e-01 5.06957710e-01
1.55207142e-01 7.73217082e-01 -4.24322575e-01 -8.13236594e-01
4.23140973e-01 2.95104414e-01 5.25791466e-01 -1.23455274e+00
3.51059325e-02 -7.74292886e-01 2.18216851e-01 -9.53914046e-01
3.56873195e-03 -1.34969568e+00 -4.49014530e-02 -2.25894475e+00
9.40019608e-01 5.07175148e-01 -6.21949613e-01 1.49376646e-01
-6.82144403e-01 -4.08403613e-02 2.57317573e-01 3.74357313e-01
-1.27409065e+00 8.10504854e-01 1.49375117e+00 -1.83485925e-01
3.04712076e-02 9.82800350e-02 -1.00609684e+00 6.52235627e-01
7.31523156e-01 -5.63580573e-01 -6.39613792e-02 1.53398355e-02
2.85094023e-01 -3.84812802e-02 -2.70158440e-01 -8.86515200e-01
2.59623498e-01 -3.42705846e-01 1.53638989e-01 -1.50278556e+00
2.24992990e-01 8.87532160e-02 -3.12542409e-01 3.51871163e-01
-6.89941168e-01 -2.31452286e-01 -1.76631689e-01 3.97867888e-01
-2.00297177e-01 -6.39442682e-01 1.56877309e-01 -3.33174109e-01
-4.27951187e-01 9.07443464e-02 -5.54391623e-01 4.78129297e-01
5.75138569e-01 1.50953345e-02 -7.99890697e-01 -4.01750475e-01
-4.67810839e-01 6.58354580e-01 -8.41536839e-03 1.31275356e-01
4.92707700e-01 -1.28260291e+00 -1.36458516e+00 -5.00327647e-01
2.51466744e-02 -2.85983115e-01 3.20251316e-01 1.16737401e+00
-3.33276063e-01 6.46926701e-01 -2.03566581e-01 -4.69811587e-03
-1.35517514e+00 2.22675920e-01 -2.59325475e-01 -1.13037169e+00
-2.95969129e-01 7.60618448e-01 -2.63345130e-02 -5.81994474e-01
1.94106996e-01 -7.15072900e-02 -6.39043272e-01 6.63628340e-01
6.24764919e-01 9.51434672e-01 2.32261896e-01 -4.08893935e-02
-3.15850317e-01 -2.16444861e-03 -2.99685091e-01 -3.43104899e-01
1.37357521e+00 -1.83547512e-02 -4.92088348e-01 5.63022792e-01
1.32163513e+00 3.78951468e-02 -6.33034706e-01 -4.13304009e-02
3.76024216e-01 1.96152553e-01 3.21603417e-01 -1.17395020e+00
-4.88916010e-01 5.25032341e-01 -1.30769908e-01 2.69423604e-01
8.91232133e-01 -1.66810438e-01 9.90708888e-01 3.80258739e-01
1.66378438e-01 -1.39458144e+00 3.71544212e-02 1.00795019e+00
1.37850142e+00 -9.43675816e-01 6.71864331e-01 -4.32813406e-01
-7.04182267e-01 8.39601636e-01 1.84287965e-01 -2.49744192e-01
3.14180613e-01 1.52765036e-01 -5.96786559e-01 -1.44519970e-01
-1.06185699e+00 -3.96371871e-01 8.02409887e-01 -1.60766561e-02
6.65687978e-01 1.51515663e-01 -1.48953271e+00 1.09421563e+00
-2.42139906e-01 1.08212069e-01 5.50406218e-01 8.38707864e-01
-9.34742451e-01 -8.31116080e-01 -8.72256681e-02 1.36327994e+00
-8.16067457e-01 -3.73456627e-01 -9.21339691e-01 5.52773237e-01
8.49286094e-02 8.50372791e-01 -1.45904616e-01 -4.34146583e-01
1.64762825e-01 4.00894552e-01 5.35539269e-01 -7.89743245e-01
-8.70910525e-01 2.26140589e-01 6.86716855e-01 -3.64423394e-01
-2.70029128e-01 -8.58682871e-01 -7.92137802e-01 -2.05598086e-01
-4.77083951e-01 4.61437523e-01 5.15921891e-01 1.07088387e+00
3.54922771e-01 1.04990888e+00 4.84599769e-01 -7.68509150e-01
-7.81835735e-01 -1.67747343e+00 -7.09578276e-01 2.75061801e-02
-5.80990426e-02 -2.82758474e-01 -3.85461122e-01 -4.24769372e-01] | [12.464003562927246, 9.533185005187988] |
79ff4e77-90c1-4a3b-8a70-594223a88dcf | leveraging-auxiliary-text-for-deep-1 | 1910.12324 | null | https://arxiv.org/abs/1910.12324v1 | https://arxiv.org/pdf/1910.12324v1.pdf | Leveraging Auxiliary Text for Deep Recognition of Unseen Visual Relationships | One of the most difficult tasks in scene understanding is recognizing interactions between objects in an image. This task is often called visual relationship detection (VRD). We consider the question of whether, given auxiliary textual data in addition to the standard visual data used for training VRD models, VRD performance can be improved. We present a new deep model that can leverage additional textual data. Our model relies on a shared text--image representation of subject-verb-object relationships appearing in the text, and object interactions in images. Our method is the first to enable recognition of visual relationships missing in the visual training data and appearing only in the auxiliary text. We test our approach on two different text sources: text originating in images and text originating in books. We test and validate our approach using two large-scale recognition tasks: VRD and Scene Graph Generation. We show a surprising result: Our approach works better with text originating in books, and outperforms the text originating in images on the task of unseen relationship recognition. It is comparable to the model which utilizes text originating in images on the task of seen relationship recognition. | ['Ran El-Yaniv', 'Gal Sadeh Kenigsfield'] | 2019-10-27 | null | null | null | null | ['visual-relationship-detection'] | ['computer-vision'] | [ 5.37596822e-01 1.21626124e-01 1.21116765e-01 -4.06966239e-01
-3.05356205e-01 -5.96329153e-01 1.26150882e+00 2.22489968e-01
-1.81583375e-01 1.64783522e-01 2.70986468e-01 -3.46136719e-01
-8.37242156e-02 -5.91785610e-01 -1.04650426e+00 -1.87123924e-01
2.21884102e-01 8.87660325e-01 4.50279981e-01 -4.46975499e-01
3.25035691e-01 5.81911922e-01 -1.79895580e+00 9.31106091e-01
3.23834062e-01 8.17985296e-01 4.91273493e-01 9.03954983e-01
-1.67083472e-01 1.38208878e+00 -7.68339872e-01 -2.77621597e-01
1.74661770e-01 -4.07166928e-01 -9.74823892e-01 4.84460831e-01
1.16276407e+00 -4.83364046e-01 -7.78945148e-01 4.89529818e-01
2.44912565e-01 2.26842478e-01 8.12635839e-01 -1.49880183e+00
-9.87976491e-01 5.11769414e-01 -6.85580134e-01 3.82610768e-01
9.00504827e-01 -1.72836240e-02 1.18261802e+00 -1.15509057e+00
1.35952044e+00 1.55091965e+00 1.26637876e-01 3.05920452e-01
-1.37197030e+00 -1.74570635e-01 2.91331649e-01 2.96380550e-01
-1.38429475e+00 -5.89383245e-01 5.76045752e-01 -6.83541477e-01
1.47838783e+00 3.64984185e-01 4.28038597e-01 1.32701135e+00
1.04364842e-01 9.27351832e-01 7.78498054e-01 -5.60469687e-01
-1.36320785e-01 4.47663754e-01 1.98825747e-01 6.01647377e-01
1.24600463e-01 -7.01164156e-02 -9.98835862e-01 1.92847013e-01
9.57006037e-01 -2.23524421e-01 -5.16503334e-01 -7.99209237e-01
-1.49331057e+00 5.04828572e-01 3.85608047e-01 3.60230953e-01
6.80866912e-02 1.86196432e-01 1.75770625e-01 5.10293126e-01
4.23273295e-01 3.25338602e-01 -3.05535942e-01 5.33490740e-02
-6.62242174e-01 1.13999456e-01 8.54835093e-01 1.25804281e+00
3.89562100e-01 -2.30795965e-01 -2.45381743e-01 8.29421043e-01
1.04590133e-01 4.78074610e-01 -2.84801796e-03 -6.40637755e-01
9.02365923e-01 7.34008491e-01 -4.74195555e-02 -1.29313743e+00
-2.76596367e-01 -9.47001129e-02 -6.04039013e-01 1.88741490e-01
3.28424931e-01 5.85887015e-01 -9.95739758e-01 1.35322285e+00
1.23785652e-01 -1.43415153e-01 1.15515597e-01 8.22960079e-01
1.63672149e+00 5.90975046e-01 -2.97360092e-01 7.23008960e-02
1.29568267e+00 -9.35652316e-01 -8.05943012e-01 -4.55853403e-01
7.67784417e-01 -7.65504956e-01 1.10415709e+00 1.72993407e-01
-1.00690472e+00 -7.43840575e-01 -9.47563350e-01 -6.47444308e-01
-7.61373043e-01 1.97474286e-01 5.71790338e-01 -4.11345325e-02
-1.37498069e+00 3.49895149e-01 -2.08816886e-01 -8.94649208e-01
3.62840027e-01 3.38925570e-01 -7.94735432e-01 -2.49950796e-01
-7.81040311e-01 1.03414345e+00 1.44717112e-01 1.24916816e-02
-9.29259658e-01 -4.03584719e-01 -1.08185673e+00 -1.37335569e-01
6.26080394e-01 -8.49324703e-01 1.08204520e+00 -8.25978100e-01
-7.32651532e-01 1.59926426e+00 -3.39493126e-01 -1.73741564e-01
6.05325997e-01 -2.05849081e-01 -2.70801485e-01 2.85167724e-01
2.28813916e-01 7.17323542e-01 8.98428977e-01 -1.94078672e+00
-3.57062578e-01 -3.80463213e-01 2.76482701e-01 4.58827108e-01
4.43124548e-02 5.04406393e-02 -8.44352305e-01 -4.79776144e-01
4.59613204e-02 -7.67172217e-01 1.88122794e-01 1.18222341e-01
-7.29535222e-01 -2.07670525e-01 1.13246465e+00 -6.24920666e-01
6.33721948e-01 -2.21611905e+00 3.76557291e-01 1.07816435e-01
5.07230163e-01 -9.41499472e-02 -3.02034885e-01 5.83374977e-01
-4.25893754e-01 9.85972732e-02 2.10895821e-01 -4.57019955e-01
-1.96433682e-02 3.30625206e-01 -5.86481929e-01 2.44729817e-01
2.48879701e-01 1.20004749e+00 -7.12412179e-01 -6.08054221e-01
4.90860283e-01 4.36144590e-01 -6.26830384e-02 3.57774526e-01
-5.14035404e-01 4.60699677e-01 -1.36228740e-01 5.30204237e-01
5.35313845e-01 -5.97244859e-01 4.80109863e-02 -4.50247765e-01
6.12176508e-02 1.69688970e-01 -1.10003865e+00 1.59453237e+00
-2.79900521e-01 1.30185270e+00 -2.75359184e-01 -8.39650750e-01
8.90248477e-01 1.09006815e-01 2.24479318e-01 -1.09265816e+00
4.76260222e-02 -2.07024902e-01 -1.54546678e-01 -6.99720860e-01
7.49918461e-01 4.97459114e-01 1.31317854e-01 5.29970527e-01
7.96241090e-02 -6.19329154e-01 2.14041531e-01 9.58255768e-01
1.14393175e+00 2.64536172e-01 1.53725058e-01 -8.77179578e-03
3.35595489e-01 1.23664066e-01 -4.97597456e-01 1.06539059e+00
2.44631499e-01 7.55560994e-01 8.27054739e-01 -4.00869220e-01
-1.03825307e+00 -1.00265002e+00 -7.36060068e-02 1.02943921e+00
6.47335291e-01 -6.91063464e-01 1.90306455e-03 -7.24587142e-01
-5.61176352e-02 8.29489350e-01 -9.87303317e-01 6.93497434e-02
-4.94973749e-01 -5.72107099e-02 5.97845800e-02 6.27462447e-01
2.76180565e-01 -1.12962174e+00 -2.80619800e-01 -3.91100556e-01
5.22896983e-02 -1.90173423e+00 -2.57775426e-01 2.35719875e-01
-5.46059191e-01 -1.03683722e+00 -3.41542065e-01 -9.89301622e-01
8.34931374e-01 6.31789982e-01 1.54878199e+00 2.26014569e-01
-4.84008998e-01 1.11559463e+00 -3.05702388e-01 -2.96419889e-01
-4.34736460e-01 -3.19441795e-01 -4.64750946e-01 -7.30883852e-02
9.69647802e-03 -2.32181534e-01 -2.03055009e-01 3.13027680e-01
-9.63322759e-01 5.50448537e-01 4.53141481e-01 6.30185366e-01
3.94470304e-01 -3.42549756e-02 -2.97275603e-01 -8.73537421e-01
2.61247873e-01 -3.10285985e-01 -4.04579580e-01 4.42473054e-01
6.09960146e-02 1.63723394e-01 3.53399217e-01 -3.71839911e-01
-9.19798434e-01 2.35276058e-01 2.88819849e-01 -6.20980620e-01
-2.97729075e-01 8.79699886e-02 -1.22681119e-01 7.97640756e-02
5.44457018e-01 9.48969349e-02 -2.41856232e-01 -2.64554888e-01
6.24659300e-01 2.27271795e-01 5.51050663e-01 -2.10959107e-01
9.81755555e-01 7.09447265e-01 1.89632401e-01 -1.11658883e+00
-7.85218656e-01 -6.72870159e-01 -9.75843370e-01 -3.50886017e-01
1.03600705e+00 -8.95460606e-01 -4.43079889e-01 1.10423468e-01
-1.47535920e+00 -5.93951702e-01 -3.98020566e-01 7.26477876e-02
-5.65819681e-01 2.04359338e-01 -3.17011595e-01 -6.06056094e-01
2.55023241e-01 -8.50025117e-01 1.69730246e+00 -1.59224987e-01
-7.74189383e-02 -1.12991869e+00 -2.61426866e-01 4.25611913e-01
-8.87964889e-02 2.70624161e-01 1.11453176e+00 -7.89545774e-01
-1.09087121e+00 4.17342372e-02 -6.60529852e-01 -7.88288563e-02
1.16003351e-02 4.28372957e-02 -9.34505641e-01 8.78356770e-02
-2.63644725e-01 -4.81365860e-01 9.84773099e-01 1.58548012e-01
1.03545582e+00 1.28281251e-01 -6.82143509e-01 4.14095134e-01
1.42012095e+00 1.61791027e-01 6.99953437e-01 3.02123189e-01
1.36181331e+00 1.04413462e+00 7.06280231e-01 7.94153437e-02
4.69813854e-01 9.81557250e-01 6.63678288e-01 -5.19552588e-01
-4.92884994e-01 -2.43975967e-01 2.11947560e-01 3.58812213e-01
-1.26103591e-02 -8.87407482e-01 -1.20250463e+00 5.50245821e-01
-2.06945968e+00 -9.69696641e-01 -7.41830945e-01 1.95772791e+00
4.04638201e-01 1.31976202e-01 -1.57692954e-01 -2.57216431e-02
5.56426227e-01 1.73928991e-01 -2.76218951e-01 -4.30529177e-01
-4.38219696e-01 8.25073272e-02 2.22841054e-01 5.19885302e-01
-1.02369595e+00 1.10503662e+00 6.44111347e+00 3.85845035e-01
-7.11040318e-01 -1.96891144e-01 5.79840720e-01 3.06695681e-02
-4.42744136e-01 3.02363932e-02 -7.43962944e-01 -2.69276708e-01
4.64329153e-01 1.98017403e-01 2.84330606e-01 6.47879422e-01
5.17672263e-02 -4.58686918e-01 -1.85387468e+00 1.38609195e+00
7.54837632e-01 -1.43719387e+00 5.96765399e-01 2.56321952e-02
6.28639817e-01 -3.65527213e-01 8.95567313e-02 9.99976918e-02
3.02177459e-01 -1.31259263e+00 8.01152349e-01 5.41589379e-01
8.29294324e-01 -3.27154875e-01 5.48530579e-01 1.18019968e-01
-1.29018414e+00 3.24383140e-01 -1.53078720e-01 2.13808864e-02
8.48153234e-03 2.15929866e-01 -1.21255100e+00 6.26714408e-01
7.98579395e-01 1.23633599e+00 -1.14943218e+00 5.95193863e-01
-2.03098252e-01 -1.61569044e-01 -4.60501611e-02 8.20172206e-02
-2.55725812e-02 -3.44498046e-02 7.02756166e-01 1.04902720e+00
-1.73758239e-01 -1.02382869e-01 3.37458178e-02 9.87797737e-01
-1.84992477e-01 -4.69579436e-02 -1.38146651e+00 -3.88808101e-02
7.90734664e-02 1.05614007e+00 -8.74130428e-01 -4.45945919e-01
-5.83423257e-01 1.29373503e+00 3.08083594e-01 5.44136286e-01
-8.35740745e-01 -2.73557287e-02 -2.12740786e-02 2.52291501e-01
4.79824841e-01 -3.38421673e-01 -2.55624522e-02 -1.01474059e+00
1.77302539e-01 -7.90465355e-01 4.08590257e-01 -1.48767519e+00
-1.22576737e+00 7.31669664e-01 3.50119501e-01 -1.19628441e+00
-1.26576304e-01 -8.61417651e-01 -2.99521804e-01 5.85400581e-01
-1.25586033e+00 -1.48175621e+00 -8.50334406e-01 9.33199584e-01
8.50932062e-01 -2.48425975e-02 4.96420354e-01 -1.51597578e-02
-2.49996424e-01 2.38416404e-01 -2.25331500e-01 1.25730917e-01
8.17580044e-01 -1.40218210e+00 5.74332952e-01 7.89625764e-01
7.18364537e-01 3.55222881e-01 7.45299160e-01 -6.78122938e-01
-1.62129915e+00 -8.56657863e-01 9.06092823e-01 -1.08403814e+00
6.97532594e-01 -1.00066710e+00 -7.52693355e-01 1.16763043e+00
4.34782445e-01 -1.60887511e-03 2.64282823e-01 -2.32826397e-02
-5.17022729e-01 2.51708388e-01 -5.26125193e-01 8.56223166e-01
1.63288546e+00 -7.85745323e-01 -7.49310851e-01 8.26252639e-01
8.45334113e-01 -6.84348524e-01 -5.77125490e-01 2.43407190e-01
2.87550062e-01 -9.95979249e-01 1.29013431e+00 -5.79653442e-01
8.33905756e-01 -2.11990222e-01 -2.88767070e-01 -9.70321596e-01
7.83764496e-02 -2.60174572e-01 -9.53266323e-02 1.27223277e+00
4.14481938e-01 -3.05271238e-01 5.89011729e-01 5.12015700e-01
3.80290970e-02 -2.89307058e-01 -5.30907452e-01 -7.49100089e-01
-4.49052572e-01 -3.47934693e-01 -6.36369437e-02 1.11949039e+00
-3.58374327e-01 9.11845207e-01 -3.65226358e-01 2.67315079e-02
4.44527388e-01 2.45131016e-01 1.36374974e+00 -1.17086363e+00
-3.62899184e-01 -2.01700047e-01 -6.22993052e-01 -1.20008874e+00
1.87842980e-01 -9.01428640e-01 -5.06860316e-02 -2.17087960e+00
4.48773533e-01 -1.28682077e-01 3.19505960e-01 5.15145421e-01
-1.02459826e-02 2.28817642e-01 3.35191220e-01 3.00403237e-01
-9.89828229e-01 4.22947526e-01 1.49842858e+00 -4.71752346e-01
-1.99852705e-01 -5.24567127e-01 -3.93485934e-01 5.34930825e-01
3.57461095e-01 -2.73686081e-01 -6.79782689e-01 -6.29777014e-01
4.46474642e-01 2.07529664e-01 8.21362019e-01 -6.47175968e-01
3.00233483e-01 4.83556502e-02 6.28165126e-01 -1.14343882e+00
6.26575768e-01 -9.88155246e-01 2.03925326e-01 -3.69834267e-02
-4.76678193e-01 1.49840653e-01 3.70627105e-01 6.37163877e-01
-9.26476195e-02 1.85727086e-02 2.16632843e-01 -2.03654826e-01
-1.03984928e+00 -1.51163666e-02 -2.65312403e-01 1.61167178e-02
1.08447981e+00 -6.37071371e-01 -8.56595993e-01 -7.26522923e-01
-8.67766201e-01 1.40460253e-01 5.67658246e-01 7.98864603e-01
1.07578337e+00 -1.09661555e+00 -7.13839114e-01 1.07622035e-01
6.11161172e-01 9.06672478e-02 1.62288651e-01 6.81079924e-01
-4.23780113e-01 4.58931476e-01 -1.80675721e-04 -1.08008814e+00
-1.73729134e+00 8.55444193e-01 2.33379662e-01 -1.16815746e-01
-9.13947105e-01 8.47779989e-01 8.39161634e-01 -1.32525325e-01
3.59472901e-01 -3.19052398e-01 -3.27318847e-01 -5.50182760e-02
2.97017753e-01 -1.46346092e-01 1.01781242e-01 -8.74256134e-01
-5.06652355e-01 7.96022952e-01 -2.30422065e-01 -2.13791579e-01
1.33358693e+00 -1.71374157e-01 -8.67655352e-02 8.64299715e-01
1.18397105e+00 9.44157317e-02 -8.21710765e-01 -2.35415787e-01
-1.62376165e-02 -7.24146008e-01 -7.44130462e-02 -8.37383747e-01
-7.97149956e-01 7.40852058e-01 3.01690280e-01 3.90210152e-01
8.56533051e-01 6.40190065e-01 9.16162506e-02 5.17811239e-01
1.94162637e-01 -5.98902464e-01 8.13001692e-01 4.45340484e-01
1.37872291e+00 -1.60739613e+00 2.68460959e-01 -8.36081743e-01
-9.13388312e-01 1.05679262e+00 8.86892080e-01 4.34178002e-02
2.60933429e-01 2.00396046e-01 -1.19338788e-01 -8.05088460e-01
-1.05318117e+00 -4.51068491e-01 6.68675065e-01 8.74160349e-01
3.09592724e-01 -5.35431266e-01 3.67437214e-01 -1.39106706e-01
-2.57376060e-02 -6.40328825e-01 7.35310018e-01 1.06314898e+00
1.04162991e-01 -9.65711415e-01 -3.55458587e-01 5.74300528e-01
7.28965849e-02 -3.45406055e-01 -1.27849770e+00 1.26607418e+00
-1.08770467e-01 1.09758568e+00 3.75457019e-01 -3.65041018e-01
5.06128550e-01 -3.46513122e-01 7.42348433e-01 -8.32953751e-01
-4.10842419e-01 -9.29388627e-02 4.77894217e-01 -8.08444440e-01
-6.85646951e-01 -5.48916399e-01 -1.07691884e+00 -1.01586968e-01
-2.94622093e-01 -3.32008183e-01 5.47404468e-01 8.45103502e-01
2.24354312e-01 7.95136929e-01 3.26014519e-01 -8.59488547e-01
3.66616577e-01 -5.37337124e-01 -5.25690973e-01 9.07959938e-01
5.65447569e-01 -7.43410647e-01 -4.92359489e-01 3.98896009e-01] | [10.430617332458496, 1.659075379371643] |
e8e8c3a1-3a4d-44cf-8e77-fac1c1c43fb6 | deep-image-harmonization-by-bridging-the | 2103.17104 | null | https://arxiv.org/abs/2103.17104v3 | https://arxiv.org/pdf/2103.17104v3.pdf | Deep Image Harmonization by Bridging the Reality Gap | Image harmonization has been significantly advanced with large-scale harmonization dataset. However, the current way to build dataset is still labor-intensive, which adversely affects the extendability of dataset. To address this problem, we propose to construct rendered harmonization dataset with fewer human efforts to augment the existing real-world dataset. To leverage both real-world images and rendered images, we propose a cross-domain harmonization network to bridge the domain gap between two domains. Moreover, we also employ well-designed style classifiers and losses to facilitate cross-domain knowledge transfer. Extensive experiments demonstrate the potential of using rendered images for image harmonization and the effectiveness of our proposed network. | ['Junyan Cao', 'Liqing Zhang', 'Jianfu Zhang', 'Li Niu', 'Wenyan Cong'] | 2021-03-31 | null | null | null | null | ['image-harmonization'] | ['computer-vision'] | [ 2.61614025e-02 -2.74154305e-01 -9.06386748e-02 -2.30614617e-01
-7.68887997e-01 -4.18674409e-01 2.19176427e-01 -2.94270009e-01
-3.24269116e-01 8.02662075e-01 6.79129288e-02 2.68456519e-01
-3.54610048e-02 -9.47617948e-01 -7.08273232e-01 -4.48361844e-01
5.97762108e-01 1.58958033e-01 1.13147192e-01 -5.92236161e-01
5.33115827e-02 2.67877340e-01 -1.37789810e+00 1.71201482e-01
1.39260888e+00 1.03044963e+00 1.49365306e-01 5.25147244e-02
-4.35733311e-02 2.96073228e-01 -7.81052411e-01 -7.55540192e-01
6.15713120e-01 -5.29406965e-01 -6.06559753e-01 2.60385007e-01
3.23724955e-01 -1.62190825e-01 -2.88145304e-01 1.21113610e+00
9.75716770e-01 1.23544581e-01 1.13180131e-01 -1.55479908e+00
-1.09930062e+00 4.78245288e-01 -8.00782919e-01 -1.89063311e-01
1.96373269e-01 1.93757311e-01 7.35811651e-01 -6.78003490e-01
7.48605251e-01 1.21144009e+00 9.23714638e-01 2.79831648e-01
-1.05270946e+00 -1.35223591e+00 -1.83928952e-01 2.41019115e-01
-1.48385310e+00 -2.07447425e-01 1.18270016e+00 -1.49208307e-01
4.99470651e-01 6.38191998e-02 8.01004946e-01 1.06650257e+00
-9.21144560e-02 6.66227341e-01 1.19414520e+00 -2.77600735e-01
-1.81074739e-01 2.52109468e-01 -5.17968535e-01 4.54538673e-01
9.97167733e-03 5.24069089e-03 -4.58264917e-01 1.18381724e-01
8.79945517e-01 -7.04603046e-02 -3.52909088e-01 -6.00915432e-01
-1.11188138e+00 8.50997746e-01 6.18459940e-01 1.91502407e-01
-4.93356921e-02 -2.31892809e-01 8.03722024e-01 4.15628880e-01
4.69922572e-01 7.34477878e-01 -2.01525658e-01 5.11473119e-02
-7.05636919e-01 2.17845500e-01 3.73805940e-01 9.84746039e-01
9.92505491e-01 -6.13876246e-02 -3.98450036e-04 1.40976465e+00
-2.87008554e-01 3.32132071e-01 7.15910017e-01 -9.01535690e-01
7.92063117e-01 8.49905372e-01 -9.00817886e-02 -1.25590777e+00
-1.11805893e-01 -4.47727680e-01 -1.14910936e+00 -1.87339988e-02
-1.21635549e-01 3.26240174e-02 -4.59698141e-01 1.89940774e+00
3.56531322e-01 2.71906823e-01 1.29292995e-01 8.57336521e-01
8.30579877e-01 5.29495478e-01 9.08719897e-02 -7.70795345e-02
1.00573254e+00 -1.29879487e+00 -6.42301798e-01 1.03555962e-01
1.87906608e-01 -9.74408448e-01 1.39297569e+00 1.82466760e-01
-1.01649153e+00 -1.05731881e+00 -1.21069860e+00 -2.02932894e-01
-2.46136501e-01 -5.92872873e-02 4.60169256e-01 4.51375961e-01
-8.65627050e-01 3.88353080e-01 -1.83868021e-01 -2.71259308e-01
6.61675930e-01 2.01882109e-01 -5.93516707e-01 1.04606897e-01
-1.50978947e+00 6.44496262e-01 6.87152863e-01 -5.03909886e-02
-1.99207619e-01 -9.95474219e-01 -8.11019957e-01 -7.74377957e-02
1.63271055e-01 -9.12730038e-01 9.24832582e-01 -1.12532961e+00
-1.45520186e+00 1.12288332e+00 3.21520001e-01 -2.78651237e-01
7.71392405e-01 -2.15716407e-01 -7.66838610e-01 8.36817473e-02
3.52414966e-01 9.31512594e-01 7.40238667e-01 -1.48099327e+00
-7.89201319e-01 -3.34164500e-01 -7.70903677e-02 3.62638205e-01
-8.40296924e-01 -2.27417871e-01 -8.49624872e-01 -1.05347157e+00
-3.21072161e-01 -9.81541574e-01 -9.95122790e-02 8.33541453e-02
-2.92970628e-01 -2.01133452e-03 9.15453970e-01 -7.88283348e-01
1.12642562e+00 -2.29724026e+00 -8.52534641e-03 1.53184414e-01
1.09060228e-01 4.41694587e-01 -4.26251143e-01 1.80778787e-01
-1.94261238e-01 -2.34413389e-02 -2.12118983e-01 -3.26072961e-01
-5.79674821e-03 2.35321268e-01 -6.29803419e-01 1.15577295e-01
8.66582692e-02 1.02656054e+00 -6.49736643e-01 -8.79530966e-01
2.70020723e-01 3.22452039e-01 -6.00823283e-01 4.15463805e-01
1.40607998e-01 6.58476710e-01 -5.09172797e-01 4.62158203e-01
1.02147770e+00 -6.37563840e-02 -5.34377582e-02 -7.05018342e-01
1.26240954e-01 -3.48769367e-01 -1.10547173e+00 2.16631460e+00
-7.22119033e-01 4.30182576e-01 -1.81214944e-01 -1.20281804e+00
1.13203609e+00 1.20163243e-02 6.70510411e-01 -1.27959073e+00
1.28217384e-01 1.32458389e-01 -4.16830629e-01 -4.22392339e-01
6.56986654e-01 -1.32232577e-01 -2.64163256e-01 2.69872218e-01
-1.24640644e-01 -4.18042541e-01 2.09363420e-02 -1.44320071e-01
4.88110363e-01 2.23570958e-01 2.97935963e-01 -4.88843285e-02
5.03816366e-01 1.11636825e-01 8.43964577e-01 3.10524940e-01
-4.13419276e-01 8.11216235e-01 1.50617629e-01 -4.94492352e-01
-1.21588159e+00 -9.78043973e-01 -2.18408197e-01 8.60021591e-01
7.08488166e-01 -1.95894524e-01 -8.77129495e-01 -4.57949162e-01
1.00071967e-01 1.93009600e-01 -5.25086761e-01 -4.37268376e-01
-7.38370717e-01 -5.98633051e-01 7.84964859e-01 5.13468266e-01
1.32974529e+00 -1.03618741e+00 -3.37125212e-01 9.27016884e-02
-5.83269119e-01 -1.26692975e+00 -7.39986539e-01 -5.33936262e-01
-6.56655729e-01 -9.93894041e-01 -6.79297864e-01 -1.12337971e+00
3.80393058e-01 3.34738284e-01 1.36863101e+00 -7.24376217e-02
-3.64236027e-01 4.89845313e-02 -2.99425662e-01 -2.06047982e-01
-4.25044507e-01 1.11200392e-01 -7.93336332e-02 -1.27768889e-01
-1.57168955e-02 -8.48649740e-01 -8.77433002e-01 7.07425833e-01
-9.65406060e-01 3.08423579e-01 6.63528442e-01 9.23702776e-01
7.24317670e-01 3.61350089e-01 8.61143291e-01 -7.36298382e-01
9.26661491e-01 -2.24249676e-01 -5.37939429e-01 5.75663447e-01
-6.11460090e-01 -2.26758227e-01 8.18321884e-01 -4.36461568e-01
-1.26749575e+00 -1.09916329e-01 -2.90121529e-02 -6.95069849e-01
1.18568659e-01 4.15186554e-01 -5.01057923e-01 -3.10914248e-01
5.51671207e-01 1.17780015e-01 8.71570557e-02 -2.61362702e-01
6.09328032e-01 5.72528243e-01 1.13506734e+00 -7.01001048e-01
9.40781355e-01 3.34335059e-01 -2.31321901e-01 -2.98142046e-01
-6.24164462e-01 -6.47777915e-02 -4.02536929e-01 -3.18055600e-02
7.09080279e-01 -1.18305564e+00 -4.71724778e-01 5.90345502e-01
-9.55168784e-01 -1.80588797e-01 -3.63506436e-01 1.51572585e-01
-5.24879098e-01 4.68255281e-01 -4.03672308e-01 3.74892317e-02
-6.69795632e-01 -1.11751580e+00 9.01425481e-01 4.01156485e-01
3.73629220e-02 -8.64310741e-01 1.96623489e-01 7.23465681e-01
4.30120081e-01 5.22666395e-01 7.16513932e-01 -8.29359815e-02
-3.27197403e-01 -7.18151703e-02 -5.72021604e-01 4.01104152e-01
3.66926342e-01 -5.85289486e-02 -5.85594296e-01 -3.18651885e-01
-4.44883741e-02 -6.63976729e-01 6.01194143e-01 5.23965992e-02
1.55322719e+00 7.53779560e-02 -1.80581197e-01 9.10872519e-01
1.28583932e+00 8.13641548e-02 7.35080719e-01 6.52264655e-01
7.46442735e-01 4.87094432e-01 8.86801004e-01 6.20857179e-01
5.29558063e-01 7.88899660e-01 2.07620324e-03 -6.61418796e-01
-3.68259162e-01 -5.67486584e-01 -2.37945437e-01 8.82135093e-01
6.73124865e-02 -9.09785181e-02 -7.41850734e-01 5.34530282e-01
-1.92269492e+00 -7.11708307e-01 4.62962240e-01 1.77587175e+00
1.24716389e+00 -1.16169393e-01 1.44284759e-02 -1.89503700e-01
1.06584656e+00 -4.51943576e-02 -8.06515634e-01 1.15688190e-01
-4.09770727e-01 2.93283403e-01 1.45637065e-01 -8.90554115e-02
-1.12247479e+00 9.77049410e-01 6.08017063e+00 1.20027971e+00
-1.32361233e+00 1.20930739e-01 6.53961718e-01 3.57610993e-02
-2.71333277e-01 -2.38716066e-01 -3.31252754e-01 6.60602033e-01
1.73014268e-01 -5.11376321e-01 3.67539614e-01 9.38771665e-01
1.48493433e-02 2.66891927e-01 -8.61027420e-01 1.48504996e+00
8.96044001e-02 -1.37259281e+00 3.22554588e-01 -2.64073253e-01
1.17486882e+00 -4.08352405e-01 3.81777525e-01 4.39023584e-01
2.51304150e-01 -7.87884772e-01 4.60313082e-01 1.27112314e-01
1.07876861e+00 -9.60841775e-01 6.31140113e-01 -4.86142226e-02
-1.44610143e+00 2.95703616e-02 -5.22342801e-01 3.10187548e-01
7.41651878e-02 4.25043792e-01 -4.63750154e-01 9.14559484e-01
8.72193396e-01 9.09694314e-01 -7.56228626e-01 8.75275016e-01
2.94937454e-02 1.11017436e-01 -4.37474586e-02 6.62310243e-01
4.94230678e-03 -4.57385331e-01 9.34369266e-02 8.72952938e-01
4.19558614e-01 -1.37508333e-01 3.79542619e-01 7.89297938e-01
-4.47693199e-01 2.33087435e-01 -5.73334754e-01 2.21780613e-01
7.68643975e-01 1.19152498e+00 -3.68536979e-01 -2.31199652e-01
-3.63020241e-01 1.18954587e+00 5.32111347e-01 2.54602283e-01
-1.17709374e+00 -5.10549486e-01 6.36865079e-01 -2.61145383e-01
1.37254938e-01 1.46192864e-01 -4.63109344e-01 -1.29895139e+00
1.61606923e-01 -1.19346869e+00 5.46194196e-01 -8.50725293e-01
-1.54438341e+00 6.12191081e-01 -1.05019636e-01 -1.52578986e+00
1.90922603e-01 -8.75608325e-02 -6.48951054e-01 7.72943854e-01
-1.57980216e+00 -1.65065217e+00 -8.20295155e-01 1.05270195e+00
3.81629735e-01 -3.55978161e-01 6.05575562e-01 7.03001082e-01
-7.21000552e-01 1.06375599e+00 9.22000632e-02 2.04424173e-01
1.38897693e+00 -7.85547793e-01 3.87168944e-01 6.52981043e-01
-1.04522645e-01 4.94318694e-01 4.96096790e-01 -5.09396613e-01
-1.04578412e+00 -1.33966577e+00 1.89661279e-01 -6.71957852e-03
4.56488788e-01 -8.63426551e-02 -1.00643992e+00 3.70969445e-01
4.86363292e-01 -1.74643274e-03 7.83604503e-01 -1.07051820e-01
-5.70518851e-01 -5.76898098e-01 -1.17766428e+00 6.83051586e-01
1.27063620e+00 -5.71070075e-01 -6.65566325e-01 1.39044672e-01
8.84594738e-01 -4.81699109e-01 -1.13259125e+00 7.15518355e-01
5.81215084e-01 -1.01359928e+00 1.22062480e+00 -3.36724997e-01
7.11162686e-01 -3.05667818e-01 -7.83144012e-02 -1.50957716e+00
-2.92219937e-01 -5.28500140e-01 5.11347830e-01 1.65112710e+00
1.22021131e-01 -7.06266284e-01 6.88929617e-01 5.33790588e-01
-6.86311126e-02 -5.97242713e-01 -6.85465634e-01 -7.79118419e-01
2.07660317e-01 -4.89362665e-02 1.08788812e+00 1.26230025e+00
-2.78840780e-01 3.05595517e-01 -5.78646541e-01 -3.18366438e-02
5.80138624e-01 6.94942832e-01 1.20377421e+00 -8.60927463e-01
-1.94076791e-01 -4.10508841e-01 -3.73502761e-01 -7.76778400e-01
5.37565708e-01 -6.14843130e-01 -1.23768426e-01 -1.11789346e+00
4.84994113e-01 -7.92930543e-01 -5.17161965e-01 3.42035741e-01
-4.62491691e-01 7.36984491e-01 2.23044366e-01 3.80624950e-01
-7.41181672e-01 1.02313030e+00 1.65921915e+00 -1.57762930e-01
-1.93639278e-01 -5.06622851e-01 -9.44494605e-01 4.92186725e-01
9.10003304e-01 -5.44747338e-02 -7.05238700e-01 -5.61958432e-01
4.17255200e-02 1.15713989e-02 2.24662647e-01 -1.10045528e+00
2.36354411e-01 -2.12501943e-01 4.08969790e-01 -5.03369510e-01
9.31524336e-02 -7.43034363e-01 4.66343343e-01 3.30242872e-01
-2.35114917e-01 1.95263416e-01 3.00271213e-01 5.03437579e-01
-7.26726770e-01 3.52353156e-01 1.08873272e+00 7.32512027e-02
-9.17740524e-01 4.54880834e-01 6.52764499e-01 3.19549233e-01
1.26308167e+00 -1.81042552e-01 -2.50667214e-01 -2.56482720e-01
-2.39716187e-01 3.97949547e-01 8.14376652e-01 6.86418295e-01
5.07092357e-01 -1.64106798e+00 -5.53132534e-01 1.79833889e-01
3.50935847e-01 2.59002801e-02 6.41280532e-01 3.36905569e-01
-5.71980953e-01 -1.64762050e-01 -9.22001123e-01 -4.00451988e-01
-1.22818148e+00 7.53663540e-01 2.35961124e-01 -3.13670099e-01
-6.54607236e-01 6.47421956e-01 4.39817071e-01 -7.51798093e-01
5.42453825e-02 1.95659287e-02 -1.52100578e-01 3.39047685e-02
2.32889146e-01 3.25957209e-01 -1.39070928e-01 -4.21309173e-01
-8.47551227e-02 9.36424375e-01 -3.61217465e-03 1.31208286e-01
1.23048604e+00 -3.66485775e-01 -2.66018718e-01 -7.60530159e-02
1.34115624e+00 -1.35479540e-01 -1.22973228e+00 -4.09264892e-01
-4.97834712e-01 -6.69181526e-01 -2.72626638e-01 -5.96220136e-01
-1.33078003e+00 6.04352832e-01 7.41481781e-01 -2.02963158e-01
1.58505881e+00 -3.22578609e-01 1.45327306e+00 3.21152449e-01
2.75229752e-01 -1.37023389e+00 5.41906595e-01 -4.98205125e-02
1.09354115e+00 -1.47873545e+00 -1.37053346e-02 -5.95630288e-01
-8.84978890e-01 8.46981406e-01 1.02870119e+00 -1.73673689e-01
3.78477842e-01 -2.66313385e-02 3.02051872e-01 1.07209183e-01
-1.25192612e-01 1.82105899e-01 5.31678610e-02 6.45113945e-01
1.18584502e-02 5.90890134e-03 -2.97433078e-01 6.28813326e-01
-6.10054672e-01 1.40235141e-01 8.68092850e-02 6.75613284e-01
7.60605140e-03 -1.39420390e+00 -3.66830826e-01 -3.58137600e-02
-9.36932713e-02 1.13428518e-01 -2.59378254e-01 8.86070848e-01
4.50250179e-01 7.96393692e-01 -6.79321587e-02 -4.31081057e-01
6.12581611e-01 -4.68651980e-01 4.05999184e-01 -8.85711983e-02
-5.12374163e-01 -2.09052905e-01 -2.87240714e-01 -4.94389355e-01
-6.09429002e-01 -3.85869332e-02 -9.30042744e-01 -5.62769890e-01
-1.33707030e-02 4.26535457e-02 2.48480007e-01 5.49587965e-01
6.52241468e-01 4.32192177e-01 9.06590343e-01 -5.79107344e-01
-5.59915185e-01 -6.31769717e-01 -6.02509558e-01 1.06445551e+00
-1.48840740e-01 -7.02885926e-01 1.19463347e-01 2.82619238e-01] | [11.241673469543457, -1.1667207479476929] |
ea5947b0-ff19-40ea-82c6-38813717188a | test-time-adaptation-with-calibration-of | 2207.00769 | null | https://arxiv.org/abs/2207.00769v2 | https://arxiv.org/pdf/2207.00769v2.pdf | Test-time Adaptation with Calibration of Medical Image Classification Nets for Label Distribution Shift | Class distribution plays an important role in learning deep classifiers. When the proportion of each class in the test set differs from the training set, the performance of classification nets usually degrades. Such a label distribution shift problem is common in medical diagnosis since the prevalence of disease vary over location and time. In this paper, we propose the first method to tackle label shift for medical image classification, which effectively adapt the model learned from a single training label distribution to arbitrary unknown test label distribution. Our approach innovates distribution calibration to learn multiple representative classifiers, which are capable of handling different one-dominating-class distributions. When given a test image, the diverse classifiers are dynamically aggregated via the consistency-driven test-time adaptation, to deal with the unknown test label distribution. We validate our method on two important medical image classification tasks including liver fibrosis staging and COVID-19 severity prediction. Our experiments clearly show the decreased model performance under label shift. With our method, model performance significantly improves on all the test datasets with different label shifts for both medical image diagnosis tasks. | ['Qi Dou', 'Huimao Zhang', 'Jing Qin', 'Shuang Zheng', 'Cheng Chen', 'Wenao Ma'] | 2022-07-02 | null | null | null | null | ['severity-prediction'] | ['computer-vision'] | [ 3.26095730e-01 -3.70901465e-01 -4.02893424e-01 -8.23742449e-01
-6.48445725e-01 -6.51189268e-01 1.30222946e-01 3.02602321e-01
-3.55695963e-01 9.01038587e-01 -2.77128100e-01 -1.25063866e-01
-3.52549344e-01 -6.31331146e-01 -3.89257580e-01 -1.12968254e+00
1.66838601e-01 1.13995254e+00 1.43976286e-01 3.76373589e-01
-1.52205437e-01 2.34308556e-01 -1.20736539e+00 6.34567022e-01
9.73574400e-01 1.06021643e+00 1.91491053e-01 5.33745170e-01
-1.33852080e-01 6.64169669e-01 -8.62113774e-01 -7.44024739e-02
8.15141723e-02 -5.77341735e-01 -6.89966857e-01 4.16977882e-01
3.95345569e-01 -5.80030680e-02 1.34131089e-01 1.28772259e+00
7.47734666e-01 -2.83835560e-01 1.07795453e+00 -1.28230238e+00
-4.23769236e-01 6.75394952e-01 -7.77023077e-01 3.19079399e-01
-4.39893365e-01 4.52924781e-02 5.27037799e-01 -3.89452755e-01
5.91323793e-01 9.30551112e-01 8.81619573e-01 7.12988675e-01
-1.25597799e+00 -8.54649186e-01 1.82837695e-01 3.19729775e-01
-1.40562057e+00 1.41345039e-01 7.16474950e-01 -7.11630344e-01
-2.39149909e-02 3.32137614e-01 5.69028795e-01 1.00733507e+00
4.28980321e-01 8.02210152e-01 1.32860017e+00 -1.66379407e-01
1.86039641e-01 4.10548270e-01 2.91549027e-01 4.33830082e-01
2.45865300e-01 -1.77262366e-01 7.92306438e-02 -2.21506938e-01
2.70236552e-01 4.34719563e-01 -4.64811742e-01 -5.83315670e-01
-1.13330984e+00 6.43979311e-01 3.22442621e-01 4.86012608e-01
-1.09724924e-01 -1.69733614e-01 7.90703654e-01 4.68526721e-01
6.65092528e-01 1.66213989e-01 -1.01114261e+00 3.82846981e-01
-7.69861996e-01 -1.67575162e-02 7.33546376e-01 6.06916249e-01
4.45544928e-01 -2.57312894e-01 -6.34812891e-01 1.23202264e+00
8.75232592e-02 6.04085863e-01 1.01049578e+00 -4.55248505e-01
8.80163461e-02 7.01965034e-01 -1.82366282e-01 -6.37957811e-01
-7.45054007e-01 -1.17749453e+00 -1.34810507e+00 4.16088402e-02
6.46032751e-01 -2.49491543e-01 -1.20176053e+00 1.81008518e+00
6.05099380e-01 1.19380988e-01 1.10854670e-01 6.42000616e-01
8.45820010e-01 2.87456006e-01 2.12691605e-01 -4.42600340e-01
1.25675380e+00 -7.54924297e-01 -6.89203203e-01 1.01539597e-01
9.93712783e-01 -4.75170493e-01 8.94581378e-01 5.65574408e-01
-5.21502912e-01 -4.62889582e-01 -8.54577661e-01 6.42380536e-01
-1.71875194e-01 3.31792861e-01 3.37349981e-01 3.91122401e-01
-7.24973500e-01 3.85679245e-01 -5.56535065e-01 -3.08264703e-01
6.32748008e-01 3.84664476e-01 -2.96107799e-01 -3.44170153e-01
-1.10122550e+00 5.87108493e-01 6.22333109e-01 -1.02629043e-01
-9.11862195e-01 -8.98408055e-01 -4.03178722e-01 2.47289054e-02
1.88304320e-01 -7.35566616e-01 1.45333874e+00 -1.48282325e+00
-1.08131456e+00 1.00474393e+00 2.87601084e-01 -2.25278720e-01
6.82916224e-01 5.31845570e-01 -4.56118613e-01 -1.31127447e-01
1.08423410e-02 5.04068732e-01 8.53928983e-01 -1.31933522e+00
-8.75521123e-01 -4.81237322e-01 -4.38877851e-01 1.68324843e-01
-3.19432795e-01 -4.83910769e-01 3.45859490e-02 -6.01777613e-01
1.03749499e-01 -1.00468016e+00 -5.00840582e-02 -3.88364121e-02
-2.79496789e-01 -2.99261272e-01 8.58514667e-01 -3.65784198e-01
1.23889315e+00 -2.08406663e+00 1.12070650e-01 1.59324974e-01
2.95949638e-01 1.45095944e-01 -1.68616846e-02 -2.55203307e-01
-2.97626525e-01 4.42359000e-02 -4.28537816e-01 -1.04202451e-02
-1.99830815e-01 4.25578624e-01 9.70341116e-02 5.47188282e-01
-3.11636999e-02 6.24293447e-01 -9.06649232e-01 -5.81714272e-01
-3.31978090e-02 2.12180331e-01 -4.74840134e-01 2.47153476e-01
-2.89107829e-01 9.20105755e-01 -3.43476862e-01 7.81495810e-01
8.52583468e-01 -6.49433613e-01 4.78833526e-01 -2.92925209e-01
5.94281256e-01 -4.66014802e-01 -9.54165578e-01 1.49134576e+00
-2.74124205e-01 2.46445209e-01 -1.72987729e-01 -1.30206788e+00
8.02761078e-01 3.01324457e-01 7.24203527e-01 -4.95857716e-01
1.48604825e-01 3.30277860e-01 6.69079304e-01 -7.15013087e-01
-4.02745396e-01 -4.17289227e-01 -5.86517490e-02 3.45812470e-01
9.66363549e-02 9.89423469e-02 1.21974617e-01 -3.57157201e-01
9.56371188e-01 -2.77205765e-01 4.36174184e-01 -4.96896058e-01
6.50958598e-01 -3.99530828e-02 8.59819412e-01 7.72812307e-01
-5.25811970e-01 5.23826897e-01 6.51145160e-01 -9.26346064e-01
-8.35647821e-01 -9.32435095e-01 -8.47128332e-01 1.00838459e+00
-2.92660315e-02 2.39278227e-01 -6.78068042e-01 -1.50148988e+00
1.00810297e-01 2.31854945e-01 -7.45689571e-01 -3.48954618e-01
-4.21696275e-01 -1.54427135e+00 3.49814832e-01 2.85413742e-01
4.36028421e-01 -9.47865248e-01 -2.58402050e-01 2.23186817e-02
-4.23447229e-02 -3.69495988e-01 -4.21558917e-01 4.26940113e-01
-8.63123298e-01 -1.43158305e+00 -9.88270402e-01 -1.02285039e+00
9.54944491e-01 -8.77259001e-02 1.23204231e+00 1.68235019e-01
-4.24878091e-01 2.11372375e-01 -2.95958281e-01 -4.00595814e-01
-7.35065043e-01 3.27949971e-01 -1.00403905e-01 2.36582354e-01
3.48242700e-01 -3.85879487e-01 -9.05154765e-01 4.49174941e-01
-9.97130930e-01 -5.64221144e-02 6.34614468e-01 1.25601530e+00
8.34295750e-01 3.68314773e-01 8.86530995e-01 -1.39184999e+00
4.78485316e-01 -9.05041575e-01 -1.52243778e-01 5.56900442e-01
-9.25074995e-01 -8.25032592e-02 5.81041276e-01 -8.97835314e-01
-7.57259011e-01 4.67710011e-02 1.04044750e-01 -4.67474580e-01
-2.20138267e-01 6.03320539e-01 -8.52666609e-03 2.33691469e-01
8.56934965e-01 1.36490628e-01 1.03414737e-01 -4.14357126e-01
-1.83463693e-01 7.80200601e-01 2.73417056e-01 -4.61716026e-01
3.74570280e-01 2.74962276e-01 6.03982545e-02 -7.95126855e-02
-9.93889809e-01 -4.85937715e-01 -6.20335937e-01 -1.69034079e-01
6.78627551e-01 -8.08726788e-01 -3.06039423e-01 9.52397645e-01
-8.11490297e-01 -4.96462464e-01 -3.49121213e-01 4.85073149e-01
-1.89998358e-01 2.69553624e-02 -4.62947220e-01 -2.89899230e-01
-5.17509580e-01 -1.46196699e+00 9.69562113e-01 1.58060998e-01
-2.34796088e-02 -1.30002713e+00 2.41660997e-01 -3.52938510e-02
3.55313331e-01 4.09805864e-01 1.22318769e+00 -1.14234567e+00
-5.13943061e-02 -2.37375021e-01 -2.34632701e-01 5.19163668e-01
5.56081057e-01 -1.93722904e-01 -7.52550900e-01 -7.19882607e-01
3.02106254e-02 -3.26958656e-01 8.52392793e-01 5.43738544e-01
1.65762043e+00 -1.87002704e-01 -4.40137088e-01 7.92595506e-01
1.33290660e+00 3.78503591e-01 4.43713591e-02 1.58081844e-01
6.50748909e-01 4.71987039e-01 5.27301908e-01 3.73773158e-01
1.81213006e-01 5.15498579e-01 3.93630415e-01 -2.69313365e-01
-3.22643131e-01 2.21673146e-01 -1.33717760e-01 7.04410374e-01
4.46678728e-01 -3.99392128e-01 -1.16100979e+00 5.12918174e-01
-1.81493342e+00 -5.65532446e-01 -2.84348354e-02 2.10273528e+00
1.20300889e+00 -1.28572151e-01 -3.25120240e-02 4.05868776e-02
9.84468162e-01 -3.05092365e-01 -9.82791722e-01 1.40474647e-01
5.39379045e-02 -7.69993365e-02 4.20748085e-01 1.98208019e-01
-1.16965282e+00 2.78172493e-01 6.31002426e+00 8.94960880e-01
-1.42276299e+00 3.91908944e-01 1.21711779e+00 2.29883846e-02
-4.08704095e-02 -5.16091347e-01 -4.92111742e-01 8.12840879e-01
5.71123660e-01 1.71931628e-02 -1.07512437e-01 8.29583645e-01
-1.57363057e-01 -1.69561096e-02 -1.27580833e+00 1.14511704e+00
1.75438762e-01 -9.57439423e-01 -2.89287604e-03 -3.59731391e-02
9.67649639e-01 -3.69402505e-02 2.74459571e-01 4.51414555e-01
1.51290640e-01 -8.62167418e-01 3.44592422e-01 5.19868731e-01
1.14113414e+00 -6.85192764e-01 1.13911211e+00 4.36630040e-01
-7.70900369e-01 -2.90082961e-01 -1.83282614e-01 4.37147468e-01
-3.00288439e-01 8.18276644e-01 -1.12905872e+00 3.04159313e-01
7.26040661e-01 8.11796188e-01 -6.05698943e-01 1.08672690e+00
6.94877235e-03 6.35029018e-01 -1.16854142e-02 2.12982908e-01
-2.00900257e-01 4.84169908e-02 2.19380111e-01 1.08544767e+00
3.38201970e-01 -2.84120709e-01 5.80167532e-01 4.66385514e-01
-1.62197635e-01 1.00258902e-01 -3.31181109e-01 4.45265353e-01
4.02643651e-01 1.29665685e+00 -9.27113473e-01 -4.72617209e-01
-5.28508425e-02 7.47406483e-01 7.47625530e-02 2.32809991e-01
-9.77484822e-01 8.83507356e-02 2.40839317e-01 1.30807698e-01
-3.89284417e-02 5.06661296e-01 -3.03000867e-01 -1.07036424e+00
-1.50830001e-01 -9.36958373e-01 9.27130222e-01 -4.06342268e-01
-1.87517905e+00 5.90159714e-01 6.11044392e-02 -1.53178751e+00
-1.48995370e-01 -5.88498473e-01 -3.42001051e-01 5.07452965e-01
-1.49912786e+00 -1.10105777e+00 -4.53864574e-01 5.20614922e-01
5.17501473e-01 -4.36241984e-01 6.81634545e-01 6.15151048e-01
-6.07102096e-01 8.92998159e-01 4.27713335e-01 1.31982788e-01
1.25597537e+00 -1.39885783e+00 -5.51372945e-01 2.93368667e-01
-2.59438366e-01 1.32143557e-01 3.44272345e-01 -4.97347802e-01
-6.87258542e-01 -1.50820971e+00 3.81646782e-01 -3.39618474e-01
3.51742566e-01 8.02077632e-03 -1.06385207e+00 5.82620144e-01
-1.89909376e-02 3.94857943e-01 1.07128108e+00 -1.75634455e-02
-3.07634771e-01 -6.81060851e-01 -1.61054504e+00 6.76974133e-02
6.82859063e-01 -7.16121420e-02 -2.55660325e-01 7.09738135e-01
5.07717133e-01 -6.29135609e-01 -1.08274603e+00 8.86301756e-01
4.42685306e-01 -6.18120074e-01 6.01320922e-01 -6.67442858e-01
2.06629723e-01 -5.28395116e-01 -7.52231255e-02 -1.68956137e+00
-4.71303850e-01 1.14808775e-01 3.47942524e-02 1.00539744e+00
4.23820347e-01 -7.00464189e-01 6.26740813e-01 3.07335705e-01
-7.87158832e-02 -8.88011336e-01 -9.91738081e-01 -4.02828962e-01
2.26469144e-01 6.18003458e-02 6.31376386e-01 1.37910950e+00
-3.80855858e-01 2.00654447e-01 -8.77237022e-02 6.52901530e-02
5.00762701e-01 1.39889583e-01 4.38794076e-01 -1.53249788e+00
-4.25579309e-01 -4.24834698e-01 -4.82018769e-01 -3.60457152e-01
1.43004075e-01 -1.28348458e+00 8.50387290e-02 -1.29419374e+00
8.60301495e-01 -8.63660574e-01 -7.87071407e-01 6.37409806e-01
-3.21661443e-01 3.68968934e-01 -2.19913110e-01 3.65403295e-01
-6.98767066e-01 1.11730456e-01 1.54303396e+00 -5.75239003e-01
-6.70727789e-02 2.48314425e-01 -5.83158791e-01 5.95428467e-01
8.78600478e-01 -7.16690361e-01 -8.01275611e-01 -4.15042818e-01
8.06632563e-02 -1.35334104e-01 1.73188329e-01 -1.09132421e+00
-2.39018910e-03 -2.59466916e-01 7.50300109e-01 -4.20830786e-01
-4.82170492e-01 -1.05771482e+00 4.41673517e-01 8.70810866e-01
-3.70193124e-01 2.41479632e-02 5.45668565e-02 5.54677248e-01
-1.71616793e-01 -1.60722703e-01 1.32446110e+00 -2.13523120e-01
-5.00479698e-01 6.93796575e-01 2.99242623e-02 1.85685411e-01
1.27196598e+00 5.72697967e-02 -4.74497199e-01 1.11268707e-01
-1.00267053e+00 3.44501734e-01 2.37993523e-01 2.99206942e-01
2.45083049e-01 -1.49918509e+00 -9.82555389e-01 2.19084486e-01
2.85430282e-01 4.84691024e-01 6.25801861e-01 1.03424227e+00
-3.76171082e-01 -7.36996159e-02 -1.06566690e-01 -1.30493999e+00
-1.21092439e+00 6.32316828e-01 9.36491013e-01 -6.67633951e-01
-3.87925148e-01 7.24232733e-01 6.98631823e-01 -6.85590267e-01
2.11811513e-01 -4.44031835e-01 -2.83977807e-01 2.17174888e-01
5.42013645e-01 1.10670924e-01 2.49389127e-01 -1.98995575e-01
-1.65697798e-01 5.44578791e-01 -3.91855687e-01 5.00393629e-01
1.21837199e+00 1.60371680e-02 -2.36629769e-01 7.06325531e-01
1.16453946e+00 -5.03906369e-01 -1.12238514e+00 -3.07398945e-01
-9.70386118e-02 -1.69467673e-01 -1.39765382e-01 -1.46668839e+00
-1.42247605e+00 6.26703858e-01 1.25374508e+00 1.42795101e-01
1.30128562e+00 3.15013831e-03 4.34438050e-01 4.17688452e-02
1.32585719e-01 -9.04307127e-01 9.66463387e-02 3.23254406e-01
7.32668817e-01 -1.53328061e+00 -1.42470255e-01 -4.29123081e-02
-5.76921225e-01 9.47254658e-01 8.56507242e-01 2.52746969e-01
7.42059231e-01 3.09481531e-01 2.99101323e-01 -5.81887960e-02
-6.44329488e-01 1.86207220e-01 2.31070116e-01 6.77035093e-01
3.53011936e-01 3.31903964e-01 -3.87320012e-01 6.94090724e-01
3.02976280e-01 3.53016034e-02 2.41659954e-01 6.08209610e-01
-3.10193419e-01 -1.33970690e+00 -3.54203850e-01 7.75551677e-01
-4.49238896e-01 1.10171042e-01 -6.57200962e-02 7.53010511e-01
7.20056832e-01 4.40089554e-01 2.39360899e-01 -2.96791524e-01
2.20458835e-01 3.21628809e-01 5.39448619e-01 -7.18110740e-01
-5.26865005e-01 -3.63325700e-02 -4.16987479e-01 1.23760505e-02
-4.71596658e-01 -5.43685496e-01 -1.24245358e+00 1.09520622e-01
-1.75843760e-01 1.30193740e-01 4.06625599e-01 6.89829826e-01
2.54345238e-01 9.02366936e-01 8.19510102e-01 -2.58780628e-01
-8.60462070e-01 -1.12759936e+00 -7.51717389e-01 7.10792720e-01
5.32490790e-01 -7.07813740e-01 -4.64169264e-01 9.09820721e-02] | [14.889549255371094, -2.29597544670105] |
822d0b0d-ca48-404b-9616-71e62d6ba03c | brain-diffusion-for-visual-exploration | 2306.03089 | null | https://arxiv.org/abs/2306.03089v1 | https://arxiv.org/pdf/2306.03089v1.pdf | Brain Diffusion for Visual Exploration: Cortical Discovery using Large Scale Generative Models | A long standing goal in neuroscience has been to elucidate the functional organization of the brain. Within higher visual cortex, functional accounts have remained relatively coarse, focusing on regions of interest (ROIs) and taking the form of selectivity for broad categories such as faces, places, bodies, food, or words. Because the identification of such ROIs has typically relied on manually assembled stimulus sets consisting of isolated objects in non-ecological contexts, exploring functional organization without robust a priori hypotheses has been challenging. To overcome these limitations, we introduce a data-driven approach in which we synthesize images predicted to activate a given brain region using paired natural images and fMRI recordings, bypassing the need for category-specific stimuli. Our approach -- Brain Diffusion for Visual Exploration ("BrainDiVE") -- builds on recent generative methods by combining large-scale diffusion models with brain-guided image synthesis. Validating our method, we demonstrate the ability to synthesize preferred images with appropriate semantic specificity for well-characterized category-selective ROIs. We then show that BrainDiVE can characterize differences between ROIs selective for the same high-level category. Finally we identify novel functional subdivisions within these ROIs, validated with behavioral data. These results advance our understanding of the fine-grained functional organization of human visual cortex, and provide well-specified constraints for further examination of cortical organization using hypothesis-driven methods. | ['Michael J. Tarr', 'Leila Wehbe', 'Margaret M. Henderson', 'Andrew F. Luo'] | 2023-06-05 | null | null | null | null | ['specificity'] | ['natural-language-processing'] | [ 4.84944463e-01 8.27424005e-02 5.58868311e-02 -4.85777795e-01
-2.56831884e-01 -7.78518438e-01 1.02924967e+00 -6.21522591e-03
-5.49864352e-01 5.60875177e-01 5.30501366e-01 -1.43329367e-01
-3.46683502e-01 -5.60564578e-01 -6.01729929e-01 -6.35755897e-01
1.73645020e-02 3.76259625e-01 1.98165476e-01 6.13999739e-02
4.54863995e-01 6.84317648e-01 -1.55255866e+00 4.51916933e-01
7.48930156e-01 6.53714180e-01 6.23607397e-01 1.69784427e-01
1.24475591e-01 6.33305460e-02 -1.04030125e-01 2.25955099e-02
1.78683624e-01 -8.78784955e-01 -7.51938283e-01 1.55452624e-01
5.12241781e-01 -6.29445398e-03 -7.83439204e-02 1.08291852e+00
1.94386691e-01 2.28984535e-01 1.03426671e+00 -7.72093713e-01
-9.36380088e-01 4.15358126e-01 -4.25715476e-01 5.26717722e-01
-1.28924660e-02 4.84490454e-01 1.15637016e+00 -1.05265832e+00
1.07230484e+00 1.49190736e+00 1.34157225e-01 6.64470613e-01
-2.04037046e+00 -4.54077423e-01 3.62317860e-01 -2.95022633e-02
-1.38933945e+00 -6.97920382e-01 6.35627031e-01 -9.74285960e-01
9.33976114e-01 -2.58957432e-03 1.09411883e+00 1.01074839e+00
3.66931915e-01 2.50713348e-01 1.75083184e+00 -2.21594259e-01
4.07652259e-01 6.63284808e-02 -2.42754016e-02 4.27098989e-01
2.66772956e-01 2.35272005e-01 -4.49860036e-01 7.96725526e-02
1.17155159e+00 -1.99693948e-01 -3.30689996e-01 -5.55721045e-01
-1.77464819e+00 8.95184696e-01 7.41725802e-01 6.98239505e-01
-5.06044805e-01 1.64032821e-02 2.09853753e-01 -1.79230407e-01
4.19371903e-01 6.72378004e-01 -2.24725559e-01 4.97261554e-01
-1.34708059e+00 2.51935035e-01 2.07237318e-01 4.95782673e-01
9.12214994e-01 1.11898094e-01 -2.69159317e-01 9.51172292e-01
4.89429682e-01 2.23330230e-01 4.87990946e-01 -1.06641972e+00
-5.06836623e-02 3.07194799e-01 -1.05295643e-01 -8.86576831e-01
-5.34870327e-01 -3.86999309e-01 -5.78727901e-01 4.54938024e-01
4.37849283e-01 2.13751182e-01 -1.01044416e+00 2.26262283e+00
-9.92469024e-03 -5.03944874e-01 -2.90564954e-01 1.10837209e+00
3.83831024e-01 1.75458193e-01 5.86593509e-01 -3.19989324e-01
1.47058260e+00 -4.56210911e-01 -2.36096397e-01 -5.09144783e-01
2.95427203e-01 -1.04799815e-01 1.27097428e+00 1.37459308e-01
-9.61876512e-01 -5.77496588e-01 -7.59511292e-01 -2.01476499e-01
-4.28730696e-01 -1.04751930e-01 5.80195963e-01 5.00282526e-01
-1.45923674e+00 3.69309127e-01 -5.81192732e-01 -7.75667489e-01
9.18253005e-01 2.39366129e-01 -5.33509195e-01 1.60045013e-01
-8.36667478e-01 1.13655460e+00 3.76109272e-01 -2.20049787e-02
-1.26540101e+00 -8.07696819e-01 -7.22355962e-01 1.01333968e-01
-3.62016447e-02 -7.79939055e-01 7.70316660e-01 -1.27528501e+00
-9.99331534e-01 1.43080235e+00 -3.67373168e-01 -2.19262511e-01
-8.05590525e-02 2.64680117e-01 -1.97652116e-01 4.18914288e-01
3.43110949e-01 1.40004289e+00 8.18561792e-01 -1.52954078e+00
3.61949839e-02 -6.64977610e-01 -1.54103816e-01 2.22864419e-01
5.46213761e-02 2.00738862e-01 2.08909169e-01 -7.11363673e-01
3.09426099e-01 -8.43639195e-01 -2.81833261e-01 2.78361231e-01
-1.62374452e-01 -2.99567264e-02 3.74777585e-01 -5.33771515e-01
7.88797796e-01 -2.14995694e+00 4.26442385e-01 2.06895933e-01
5.21804214e-01 -2.36970380e-01 -1.40423432e-01 1.47268474e-01
-3.61257106e-01 3.99497509e-01 -7.66226232e-01 1.35347888e-01
1.04632109e-01 1.48302028e-02 -2.80070454e-01 7.64322758e-01
5.05965531e-01 1.10462511e+00 -8.50849152e-01 -5.10461986e-01
2.03830436e-01 3.91245872e-01 -8.82904530e-01 -1.88780636e-01
-2.51862288e-01 8.63014877e-01 -1.40426576e-01 4.71155643e-01
4.39069360e-01 -2.21705765e-01 4.39872295e-01 -3.53571266e-01
-4.40025955e-01 3.15572441e-01 -6.82976067e-01 1.76572466e+00
-2.38481179e-01 8.03699553e-01 2.32445762e-01 -1.19274306e+00
6.83781981e-01 2.63637789e-02 2.11068034e-01 -6.78987861e-01
1.66448832e-01 1.86128944e-01 7.52914846e-01 -2.66422629e-01
-1.17040090e-01 -6.04164183e-01 1.08214997e-01 5.60604930e-01
4.68515873e-01 -3.65826786e-01 1.16751783e-01 2.25297645e-01
8.72249186e-01 3.37433189e-01 3.62957388e-01 -9.99344885e-01
2.14657992e-01 8.19490384e-03 3.44577610e-01 5.93956411e-01
-3.81775349e-01 6.92303002e-01 5.80795586e-01 1.08587101e-01
-1.13108206e+00 -1.39068687e+00 -5.94608366e-01 1.04795432e+00
-8.58255029e-02 1.53704714e-02 -8.14659536e-01 -3.23226988e-01
-1.44986585e-01 7.97139943e-01 -1.10911191e+00 -1.93831310e-01
-3.46880049e-01 -7.89841950e-01 3.08643728e-01 3.39661598e-01
3.00467581e-01 -1.46371436e+00 -8.48165393e-01 5.42821549e-02
1.43848494e-01 -7.93215990e-01 -2.38354310e-01 2.08199397e-01
-8.51578176e-01 -1.07160318e+00 -8.37895334e-01 -7.83529103e-01
1.03454006e+00 2.31523067e-01 1.10271943e+00 -2.14966491e-01
-6.63105667e-01 5.02245843e-01 8.60959962e-02 2.20780578e-02
-9.27844271e-02 -1.89691141e-01 1.45650789e-01 3.37859653e-02
1.65780842e-01 -8.58185410e-01 -9.35241044e-01 3.69089484e-01
-1.02134836e+00 4.32343073e-02 6.59811735e-01 7.35320508e-01
6.30589604e-01 -3.97783667e-01 8.00332725e-01 -7.82800615e-01
7.37846434e-01 -8.14399958e-01 -3.80726129e-01 6.19676001e-02
-3.57755989e-01 1.57155037e-01 3.37019622e-01 -4.88744467e-01
-1.25672948e+00 -1.21897317e-01 1.00763872e-01 -9.81471092e-02
-4.53971595e-01 4.17751819e-01 -2.27111161e-01 -1.53877959e-01
9.70205069e-01 3.52923483e-01 -2.38753501e-02 -2.11602703e-01
5.91868401e-01 1.71881691e-01 4.28173900e-01 -7.76214838e-01
4.65112090e-01 7.00486898e-01 -1.21695638e-01 -9.60067213e-01
-5.51451862e-01 -4.26455736e-02 -1.07281101e+00 -2.93654650e-01
1.21652484e+00 -7.49241948e-01 -5.50047576e-01 -6.01798780e-02
-9.25106883e-01 -5.18241525e-01 -2.98792422e-01 5.78536391e-01
-8.68962824e-01 1.98162824e-01 -2.12496728e-01 -5.39059877e-01
2.23070353e-01 -1.25733399e+00 9.17154670e-01 1.36109395e-02
-5.88666916e-01 -1.07570159e+00 1.55156344e-01 1.08402863e-01
3.89437377e-01 9.76253003e-02 1.26245010e+00 -4.76574093e-01
-6.25317872e-01 5.14142871e-01 -5.66016614e-01 3.08061577e-02
5.23603670e-02 -1.13117687e-01 -9.22562897e-01 -2.89374944e-02
2.75915973e-02 -5.33141434e-01 1.17637551e+00 5.27534068e-01
1.23212743e+00 -9.99168456e-02 -4.91521358e-01 5.22299230e-01
1.35777617e+00 1.43678516e-01 5.49353480e-01 -4.27263565e-02
3.43741655e-01 1.24391258e+00 9.72498804e-02 1.59213636e-02
1.15085132e-01 4.12550032e-01 1.37202680e-01 9.85766798e-02
-3.69162947e-01 -7.11723343e-02 8.14760402e-02 2.07660019e-01
-1.35340810e-01 7.63849989e-02 -8.97506595e-01 9.81866002e-01
-1.40316963e+00 -1.15716684e+00 1.02965221e-01 2.10619330e+00
8.71735454e-01 2.40663160e-02 1.49044544e-01 -3.08981210e-01
6.58326626e-01 3.00426185e-01 -8.25564623e-01 -1.77915230e-01
-2.61590540e-01 3.20708662e-01 7.99934715e-02 4.96796668e-01
-7.17097461e-01 1.08571494e+00 7.60228777e+00 4.27248418e-01
-1.02295840e+00 2.25683212e-01 7.07014978e-01 -1.69879273e-01
-6.77918792e-01 1.43458873e-01 -5.96881270e-01 2.31676653e-01
8.38776708e-01 -4.78283837e-02 8.48117828e-01 4.78387326e-01
3.71613830e-01 -3.85078490e-01 -1.19708169e+00 6.18125796e-01
9.18514058e-02 -1.28927481e+00 1.92537811e-02 3.75293314e-01
6.21940017e-01 5.93055375e-02 3.04781437e-01 -4.19322178e-02
3.37500066e-01 -1.28259540e+00 8.38709831e-01 6.37806237e-01
1.06120813e+00 -9.90417898e-02 -2.57540673e-01 1.77269161e-01
-8.67495537e-01 -4.66359258e-02 -5.71128845e-01 6.64810240e-02
6.02399595e-02 4.57293183e-01 -5.69179356e-01 -3.57470781e-01
5.60247481e-01 4.68797058e-01 -5.94275832e-01 8.21568191e-01
-5.04302718e-02 4.29652303e-01 -1.43729419e-01 1.76509723e-01
1.31199181e-01 -1.59590960e-01 4.54320490e-01 1.23426056e+00
2.06342340e-01 3.06171000e-01 -1.08234487e-01 1.84508216e+00
1.59550264e-01 1.43199667e-01 -7.16712236e-01 -1.59000561e-01
3.16383541e-01 1.51613200e+00 -1.23483753e+00 -2.07106933e-01
-2.12644339e-01 8.80470574e-01 6.46381080e-01 5.22352695e-01
-5.86509764e-01 7.60698020e-02 6.36265457e-01 3.49232256e-01
2.08504915e-01 -5.97235024e-01 -4.77416158e-01 -1.09092963e+00
-3.19744825e-01 -5.70947409e-01 -6.72345161e-02 -9.38294590e-01
-1.13175023e+00 6.12619460e-01 3.62309128e-01 -4.80378747e-01
5.31872688e-03 -6.62944794e-01 -4.37839150e-01 1.17308056e+00
-1.03425336e+00 -1.12804663e+00 -1.34077609e-01 6.54120445e-01
5.44152558e-01 1.14853837e-01 7.43354082e-01 -3.30728702e-02
-3.42358470e-01 -2.62567960e-02 -6.48712665e-02 -2.47113645e-01
5.15072703e-01 -1.01756883e+00 2.16465205e-01 6.10230923e-01
3.43525767e-01 1.00185382e+00 5.35651147e-01 -8.28726828e-01
-8.55062068e-01 -7.18771875e-01 5.83710551e-01 -5.75770259e-01
6.55604482e-01 -7.97493279e-01 -9.92474079e-01 6.21681273e-01
2.20285550e-01 1.66586235e-01 6.68877959e-01 1.02296785e-01
-4.44647014e-01 2.09691480e-01 -1.18277180e+00 6.51594937e-01
1.50774574e+00 -8.10043097e-01 -7.91128397e-01 3.18199813e-01
3.08623612e-01 4.96573567e-01 -7.69266129e-01 4.79624271e-02
4.63556707e-01 -1.04670799e+00 1.04057312e+00 -6.75662518e-01
3.76203448e-01 -2.60858476e-01 -1.92230955e-01 -1.49312663e+00
-9.72801030e-01 1.34890869e-01 4.93209273e-01 9.93105471e-01
3.87107044e-01 -6.71538353e-01 4.29626495e-01 6.08146369e-01
-1.99340075e-01 -3.30349028e-01 -8.63241374e-01 -5.04572153e-01
3.21001351e-01 -1.75884828e-01 1.67306662e-01 9.13232565e-01
2.33805552e-02 2.31453478e-01 2.54254878e-01 -2.95699626e-01
6.07494414e-01 7.86088686e-03 1.33026913e-01 -1.34080017e+00
-1.26771048e-01 -9.00077343e-01 -1.84588522e-01 -5.73921502e-01
5.01248240e-01 -1.14206195e+00 1.71823487e-01 -1.56057513e+00
4.34424967e-01 -3.29425335e-01 -2.89727837e-01 5.54558218e-01
1.62777454e-01 5.71365595e-01 9.69189405e-02 2.85016954e-01
-1.91946134e-01 4.83519673e-01 1.31002319e+00 2.14262609e-03
3.19947712e-02 -6.64711833e-01 -1.11325943e+00 7.14300692e-01
7.99311101e-01 -2.95282006e-01 -5.33603847e-01 -1.98464185e-01
1.18269548e-01 -3.06135684e-01 8.13531160e-01 -9.24776554e-01
-2.62923568e-01 -4.52921838e-01 7.94356585e-01 4.05654088e-02
2.51623541e-01 -6.17592096e-01 1.20505452e-01 3.85449886e-01
-5.14355302e-01 -2.52781004e-01 2.08441168e-01 4.91469681e-01
6.86501861e-02 1.51519896e-02 1.13544333e+00 -6.21316195e-01
-8.60914290e-01 1.43320605e-01 -7.35420823e-01 2.08138809e-01
1.03971767e+00 -4.11509871e-01 -4.87074345e-01 -1.73241235e-02
-1.07872820e+00 -3.14253151e-01 5.90622604e-01 2.74718672e-01
6.60169601e-01 -1.19194114e+00 -6.19989216e-01 2.75750309e-01
1.93576012e-02 -5.06345868e-01 3.36096436e-01 9.22506988e-01
-1.34520873e-01 5.75896978e-01 -7.29725301e-01 -5.23898840e-01
-7.17723608e-01 7.95380890e-01 4.42089409e-01 3.00837636e-01
-4.14553314e-01 7.86351800e-01 1.20042312e+00 -2.81417072e-01
-5.51725030e-01 -3.03002179e-01 -2.50024289e-01 3.56689900e-01
2.04882205e-01 -1.09012902e-01 -3.31467956e-01 -9.92747843e-01
-4.60857302e-01 3.43428075e-01 1.74464762e-01 -4.85762089e-01
1.28674138e+00 -2.66626894e-01 -3.84723961e-01 5.66342771e-01
9.47057247e-01 -1.10120967e-01 -1.51434243e+00 3.48185152e-02
-1.18882559e-01 -5.29838502e-01 2.14467087e-04 -9.52358603e-01
-8.90961826e-01 1.03431535e+00 4.70692873e-01 2.98099797e-02
9.90951180e-01 4.47581321e-01 -6.77807778e-02 -7.24706054e-02
3.34943682e-01 -9.01019752e-01 1.05384700e-01 1.92037687e-01
1.38104963e+00 -1.01754332e+00 -1.28799379e-01 -1.12513028e-01
-6.62598431e-01 6.26436472e-01 7.44966805e-01 -3.83789271e-01
6.90585673e-01 -3.55476588e-01 -3.30782145e-01 -5.22218823e-01
-7.13139534e-01 -4.45926279e-01 6.23421431e-01 8.19850922e-01
5.85875809e-01 1.15651764e-01 -4.98878688e-01 5.12366235e-01
-1.19864106e-01 -4.54830557e-01 2.08475187e-01 5.73775530e-01
-7.43492424e-01 -8.22503507e-01 -4.38304305e-01 7.43397593e-01
-2.47149050e-01 -3.83155614e-01 -4.48034286e-01 8.32911015e-01
2.76746094e-01 4.94358420e-01 3.19646001e-01 5.38668893e-02
-1.02767043e-01 3.45290542e-01 9.87344027e-01 -9.81438637e-01
-4.49543744e-01 3.03081721e-01 -8.77119005e-02 -3.74578536e-01
-6.26123905e-01 -1.11451876e+00 -1.10512865e+00 1.38595521e-01
8.62299055e-02 -2.54426867e-01 5.50545931e-01 8.81620169e-01
3.40287268e-01 4.60649401e-01 5.25023118e-02 -1.19682515e+00
5.92657998e-02 -8.72946799e-01 -7.85674691e-01 4.09071594e-01
1.23782746e-01 -1.08976829e+00 -1.70420378e-01 2.71775126e-01] | [10.503198623657227, 2.4492762088775635] |
9f476856-5fd6-47eb-8f2e-273b5d560467 | scaling-cross-domain-content-based-image | 2204.11593 | null | https://arxiv.org/abs/2204.11593v1 | https://arxiv.org/pdf/2204.11593v1.pdf | Scaling Cross-Domain Content-Based Image Retrieval for E-commerce Snap and Search Application | In this industry talk at ECIR 2022, we illustrate how we approach the main challenges from large scale cross-domain content-based image retrieval using a cascade method and a combination of our visual search and classification capabilities. Specifically, we present a system that is able to handle the scale of the data for e-commerce usage and the cross-domain nature of the query and gallery image pools. We showcase the approach applied in real-world e-commerce snap and search use case and its impact on ranking and latency performance. | ['Eran Nussinovitch', 'Minh Tran', 'Isaac Kwan Yin Chung'] | 2022-04-13 | null | null | null | null | ['content-based-image-retrieval'] | ['computer-vision'] | [-3.99872780e-01 -6.58450782e-01 -2.03225147e-02 -1.98015898e-01
-1.30842721e+00 -1.40439749e+00 9.18641090e-01 2.05507293e-01
-5.04640877e-01 1.27675667e-01 1.67203158e-01 -2.62764394e-01
-4.74547565e-01 -3.89167219e-01 -2.12745726e-01 -4.97277714e-02
-8.67363811e-02 7.70758390e-01 6.53668821e-01 -7.26831377e-01
4.43964660e-01 7.61334419e-01 -1.64637911e+00 1.31832933e+00
4.64280928e-03 1.46544468e+00 2.11654276e-01 8.15545321e-01
-3.11522186e-01 5.78248322e-01 -8.82111430e-01 -5.82065880e-01
7.08154023e-01 1.10240951e-01 -1.04579914e+00 2.03736484e-01
7.69194424e-01 -3.06001425e-01 -3.73857409e-01 7.71719813e-01
8.16885293e-01 -8.94828960e-02 6.05624020e-01 -1.30273545e+00
-6.94420099e-01 -4.83118743e-03 -5.68613052e-01 5.67837119e-01
4.70451534e-01 -4.87656407e-02 7.60506749e-01 -1.15412760e+00
1.33527732e+00 1.40752673e+00 2.58711159e-01 2.38927171e-01
-9.80187654e-01 -4.66764122e-01 1.05767868e-01 5.06189108e-01
-1.44369197e+00 -2.84928918e-01 5.36492109e-01 -4.42403138e-01
9.46418285e-01 4.72109109e-01 5.63241780e-01 8.69780183e-01
-5.38824499e-02 1.00278175e+00 1.21680951e+00 -5.09393692e-01
1.68098271e-01 5.84696472e-01 1.23307437e-01 4.74584937e-01
-2.11434394e-01 8.67531542e-03 -6.93948448e-01 -4.65673685e-01
8.79607022e-01 1.95571259e-02 3.17916214e-01 -8.64054799e-01
-9.89902437e-01 7.83021152e-01 6.10955358e-01 5.04414439e-01
-3.22448164e-01 5.45389540e-02 8.74951661e-01 7.73982286e-01
4.85862881e-01 2.47504801e-01 -4.68510330e-01 3.55267167e-01
-1.16797829e+00 2.65855342e-01 7.92938948e-01 1.24001288e+00
3.35726202e-01 -6.32345200e-01 -3.55523199e-01 9.44166064e-01
2.97960460e-01 3.37406576e-01 4.68047738e-01 -7.63826430e-01
3.26579303e-01 5.82578480e-01 2.30120212e-01 -8.62082958e-01
3.54019813e-02 -2.63271984e-02 -1.19885549e-01 4.40962940e-01
3.71235311e-01 4.43157017e-01 -1.20113075e+00 9.13341463e-01
2.49690875e-01 -6.81592882e-01 -9.25220549e-02 1.17077100e+00
1.00219142e+00 3.88498157e-01 4.03526664e-01 7.32069910e-02
1.76959729e+00 -1.03211379e+00 -3.82418931e-01 -4.32775728e-02
3.37060243e-01 -1.30507934e+00 9.93513286e-01 3.83176953e-01
-1.32122016e+00 -6.15206957e-01 -9.60508883e-01 -2.04460666e-01
-1.12230945e+00 1.92220986e-01 4.87069041e-01 4.04990494e-01
-1.49723089e+00 3.16947475e-02 -3.01019046e-02 -8.39320958e-01
1.44220933e-01 3.60302776e-01 -4.54502136e-01 -4.00248230e-01
-1.12812531e+00 8.30983460e-01 2.35898539e-01 -2.52372921e-01
-8.19631159e-01 -5.56085408e-01 -2.55692080e-02 -2.39051040e-02
3.58261377e-01 -4.37172264e-01 1.33377719e+00 -1.07650363e+00
-7.95463204e-01 1.35960460e+00 4.27076399e-01 -1.04171649e-01
8.52556169e-01 -2.38864094e-01 -7.92002141e-01 5.47567248e-01
-1.58786196e-02 7.69951642e-01 7.65683949e-01 -1.44816494e+00
-7.99503863e-01 -5.91305435e-01 1.34882376e-01 2.45760351e-01
-3.27782035e-01 6.51279151e-01 -1.15831161e+00 -4.92900968e-01
-2.73246109e-01 -7.05889225e-01 -7.89316222e-02 2.36934051e-01
1.71599299e-01 -1.63219661e-01 1.22989678e+00 -6.15177691e-01
9.68676150e-01 -2.41654134e+00 -2.81882197e-01 5.90780675e-01
6.34282315e-03 4.31970447e-01 -4.80675012e-01 1.10684109e+00
5.55478074e-02 3.61674666e-01 8.37509632e-01 3.26057553e-01
6.78716898e-02 -1.36010125e-01 -2.70445287e-01 3.98170277e-02
-4.27317582e-02 1.06930566e+00 -6.56138062e-01 -9.29900885e-01
1.01667941e-01 3.29875559e-01 -1.01260357e-01 -3.99044640e-02
-9.98353660e-02 -2.75969625e-01 -6.23790503e-01 9.70509529e-01
6.29934609e-01 -4.93959695e-01 5.19813836e-01 -3.94064933e-01
3.99007946e-02 -1.78127125e-01 -1.07639623e+00 1.71090961e+00
-4.95123833e-01 5.72070122e-01 7.58709073e-01 -5.79176068e-01
7.25037456e-01 2.68245399e-01 4.61462975e-01 -1.31295717e+00
-2.82792062e-01 3.28527331e-01 -6.21368825e-01 -2.15596676e-01
6.59743130e-01 2.51407385e-01 -1.03588298e-01 4.17743117e-01
1.40202150e-01 -7.97786415e-02 3.91231000e-01 7.19670475e-01
1.03988993e+00 -1.17261283e-01 5.12777902e-02 -4.15701568e-01
5.05484462e-01 6.88738644e-01 -4.51624364e-01 9.72592294e-01
-1.76419750e-01 7.54033446e-01 1.83060810e-01 -8.44530761e-01
-1.17905247e+00 -1.05644155e+00 7.76042417e-02 1.56124234e+00
3.70505244e-01 -4.25762534e-01 -2.33396053e-01 -1.03218830e+00
2.47913048e-01 -6.63286969e-02 -4.52725202e-01 1.24319114e-01
-3.50722879e-01 -2.21137702e-01 1.40282094e-01 1.86830580e-01
5.20513415e-01 -1.16355681e+00 -6.12283707e-01 7.15393573e-02
-5.96560203e-02 -9.18592513e-01 -6.67885482e-01 -1.14379287e-01
-6.02104306e-01 -1.08776665e+00 -9.89361823e-01 -9.71532583e-01
2.36108169e-01 5.45365751e-01 1.73792195e+00 6.68144524e-02
-1.11193275e+00 1.07150686e+00 -5.50056756e-01 -2.24881381e-01
-3.41766924e-01 -1.62976444e-01 -5.67146063e-01 -3.88901412e-01
5.59273005e-01 3.19827162e-02 -1.21877384e+00 6.09459400e-01
-1.09739375e+00 -4.43178862e-01 7.19449401e-01 5.88487685e-01
5.28944433e-01 -7.27719516e-02 1.95201457e-01 -8.57203543e-01
1.15837395e+00 -4.73406374e-01 -7.47544706e-01 9.41110611e-01
-9.45702255e-01 -3.36637944e-01 -2.73550510e-01 -5.87412000e-01
-7.98808038e-01 2.98919052e-01 4.24463987e-01 -6.87824428e-01
9.34915543e-02 3.67155045e-01 4.01701897e-01 -1.14543043e-01
7.85267055e-01 1.70683920e-01 -7.83331096e-02 -8.97696257e-01
7.40672529e-01 9.03969944e-01 3.36860925e-01 -3.05168271e-01
4.34909195e-01 5.44165075e-01 -3.09633970e-01 -2.65771747e-01
4.13429458e-03 -1.28565764e+00 -5.90892971e-01 -4.82451469e-01
6.52079463e-01 -1.32575047e+00 -5.60424864e-01 8.67296457e-02
-1.23728597e+00 1.26637474e-01 -2.70566165e-01 -1.27614230e-01
-3.11192125e-01 2.13982671e-01 -7.53881633e-01 -7.16360748e-01
-5.93517840e-01 -8.06092083e-01 1.42229176e+00 -3.25008571e-01
2.24190220e-01 -6.19700134e-01 1.31376386e-01 5.10531187e-01
7.02387810e-01 -2.18482479e-01 8.64214540e-01 -1.05376065e+00
-6.49801910e-01 -6.12506211e-01 -8.30068350e-01 2.81209424e-02
-4.21729773e-01 -2.66961098e-01 -6.76720858e-01 -4.95904028e-01
-6.72128856e-01 -5.74607968e-01 6.86827481e-01 -2.02820674e-02
8.25895667e-01 -1.54324144e-01 -4.95562643e-01 -2.42963672e-01
1.98726594e+00 1.63778767e-01 5.69737375e-01 4.73717958e-01
2.32919782e-01 9.79285300e-01 9.74414766e-01 4.77614999e-01
-1.11911342e-01 1.24592519e+00 2.91139543e-01 -5.02487659e-01
-3.68678659e-01 -2.03715369e-01 4.42443453e-02 3.46382260e-01
2.12812111e-01 -2.06314430e-01 -9.29289103e-01 6.97405100e-01
-2.02339363e+00 -9.98829722e-01 3.37229744e-02 1.96692169e+00
4.66060847e-01 -3.01226765e-01 6.07346833e-01 -3.70634615e-01
8.22061777e-01 -1.53342158e-01 -3.08560997e-01 -5.85915148e-01
-2.57644523e-03 8.76465961e-02 6.48888528e-01 1.07295357e-01
-1.09233904e+00 8.40015948e-01 8.27170467e+00 1.41421735e+00
-8.95388067e-01 5.55076182e-01 6.55562103e-01 -2.87177831e-01
1.65887326e-02 -3.93142343e-01 -4.87923503e-01 1.38615578e-01
8.82033765e-01 -1.14775151e-01 3.97056520e-01 1.09822667e+00
-2.88266033e-01 1.64538741e-01 -9.41186607e-01 1.21587324e+00
1.42096788e-01 -1.41618836e+00 1.80865869e-01 2.71424472e-01
5.29827118e-01 3.42046589e-01 2.03002304e-01 2.76531368e-01
5.13765156e-01 -6.43502474e-01 8.87066245e-01 5.24434187e-02
1.10895228e+00 -3.48425567e-01 5.72881877e-01 -8.77780989e-02
-1.08841407e+00 -3.74745697e-01 -4.22840081e-02 6.43943429e-01
-2.02397481e-01 6.55953661e-02 -7.16203272e-01 4.12232280e-01
1.07746696e+00 6.58912808e-02 -8.20944667e-01 1.09099185e+00
6.71365857e-01 -2.85219878e-01 -2.28380382e-01 -1.17398463e-01
2.22242758e-01 2.00593144e-01 1.65861025e-01 1.70874190e+00
2.14631915e-01 -1.44346803e-01 1.84256166e-01 3.28786910e-01
-1.72452405e-02 4.37351376e-01 -6.97461188e-01 1.51689118e-03
2.46604979e-01 1.55934215e+00 -9.81105864e-01 -6.13379896e-01
-5.46613097e-01 1.27707863e+00 6.83990866e-02 5.16346455e-01
-3.49736482e-01 -1.98294431e-01 3.42456371e-01 4.35812324e-01
8.35091710e-01 -8.24193750e-03 4.40521866e-01 -9.65751171e-01
8.76270086e-02 -1.21767700e+00 9.37669456e-01 -1.05453885e+00
-1.53047097e+00 8.14779818e-01 3.00665617e-01 -1.39544201e+00
-3.88944685e-01 -8.36272120e-01 3.06783896e-02 8.30439389e-01
-1.40027404e+00 -1.44073296e+00 -6.07601479e-02 1.09099317e+00
7.49615192e-01 -3.30072582e-01 6.96908951e-01 7.10444868e-01
2.95059681e-01 3.10347974e-01 5.37828267e-01 -1.37428597e-01
9.28960621e-01 -1.04021096e+00 2.39597097e-01 3.72206450e-01
4.31329221e-01 5.96848249e-01 4.74834234e-01 -3.60948861e-01
-1.42829335e+00 -5.66748202e-01 6.13314927e-01 -6.74090326e-01
7.76760697e-01 -6.97484851e-01 -2.47473598e-01 2.23625049e-01
4.90126789e-01 3.75874788e-01 3.76663059e-01 2.60732640e-02
-8.37332308e-01 -4.50953692e-01 -1.23413563e+00 1.61114156e-01
8.55024695e-01 -8.75201285e-01 -2.47127235e-01 8.04606318e-01
1.45592228e-01 -1.04971550e-01 -6.43598497e-01 2.09286198e-01
9.45872724e-01 -9.13145840e-01 1.54856658e+00 -6.71923757e-01
2.56168872e-01 -1.39877856e-01 -2.77138442e-01 -7.42579579e-01
-3.69686931e-01 -2.54976213e-01 3.02810192e-01 1.17501783e+00
4.29468870e-01 3.18972720e-03 7.15105414e-01 7.02667773e-01
6.90314114e-01 -1.60906047e-01 -9.48493004e-01 -8.57290804e-01
-1.54314682e-01 1.26541823e-01 1.86081231e-01 4.90742683e-01
1.41482905e-01 1.76684693e-01 -2.82533377e-01 -3.77856404e-01
3.71008217e-01 4.30096000e-01 5.32393217e-01 -1.07831061e+00
-2.71729529e-01 -2.92401224e-01 -5.04456043e-01 -8.06058288e-01
-7.21325457e-01 -5.96309781e-01 -2.65031368e-01 -1.40055418e+00
6.17268443e-01 -3.43471080e-01 -8.19732249e-01 7.37595484e-02
4.43295151e-01 7.64593542e-01 6.98383689e-01 8.38518202e-01
-1.37910187e+00 -5.20406485e-01 9.16355073e-01 -4.06915992e-01
4.35411632e-02 -3.52246612e-01 -5.37522137e-01 -1.20864533e-01
1.91400394e-01 -2.91922301e-01 -5.61537147e-01 -3.64477634e-01
3.79330963e-01 1.27566248e-01 5.31661570e-01 -6.09149873e-01
3.99785101e-01 1.84064552e-01 5.27569771e-01 -7.03939199e-01
4.09423769e-01 -1.24443126e+00 3.00532043e-01 4.03161258e-01
-6.40048683e-01 5.10420084e-01 2.39350840e-01 8.25013399e-01
-6.02014124e-01 2.21269988e-02 2.75986284e-01 -6.40444815e-01
-1.20932651e+00 -1.91698223e-02 -3.48660588e-01 -1.64216772e-01
1.08207846e+00 3.83362025e-02 -6.19257033e-01 -6.48409963e-01
-8.14449489e-01 2.16484085e-01 3.28630149e-01 7.96427310e-01
4.17237967e-01 -1.33773780e+00 -6.43267155e-01 -6.85819760e-02
7.13460684e-01 -1.17734098e+00 1.12571985e-01 3.50254506e-01
-7.27579355e-01 6.01118028e-01 -5.05538225e-01 -3.93330276e-01
-1.80060947e+00 1.20689046e+00 -5.41710518e-02 -6.79920554e-01
-9.86198261e-02 5.49140692e-01 2.08044752e-01 -1.55310482e-02
1.77327722e-01 4.53117788e-01 -2.48656496e-01 2.07982302e-01
6.82265401e-01 2.07572848e-01 3.82311434e-01 -4.09734905e-01
-6.98531032e-01 6.23823404e-01 -4.05104518e-01 -5.19964874e-01
1.07532024e+00 -5.17537236e-01 1.07199252e-01 1.01776838e-01
1.39634359e+00 -1.59479037e-01 -5.78793824e-01 -3.00173461e-01
1.67962253e-01 -5.57358921e-01 3.12976420e-01 -1.54514217e+00
-1.03904629e+00 8.49494994e-01 1.42449284e+00 7.57734537e-01
1.29000688e+00 4.76498216e-01 3.14897656e-01 2.69730836e-01
4.29375827e-01 -1.51140726e+00 2.21636407e-02 -2.51388133e-01
1.22301745e+00 -1.26672053e+00 2.09427476e-01 -4.49299008e-01
-9.13469970e-01 1.08877480e+00 1.64134204e-01 -1.33098632e-01
8.04922819e-01 1.54168308e-01 7.09900737e-01 -8.44890535e-01
-1.00421512e+00 -2.54861414e-01 3.48901868e-01 6.46004617e-01
2.94360727e-01 -3.76966983e-01 -7.00575471e-01 -6.20123222e-02
7.20592022e-01 5.30721508e-02 -3.89165908e-01 1.04508138e+00
-5.45061789e-02 -1.46984029e+00 -3.08129132e-01 9.75317657e-02
-7.87859321e-01 -3.39170992e-01 -9.69758213e-01 1.05801404e+00
-2.11956739e-01 9.62492049e-01 7.75285903e-03 -3.74600112e-01
5.91473579e-01 1.48207501e-01 2.05931410e-01 -2.76712298e-01
-1.12830698e+00 4.90131408e-01 2.47087821e-01 -9.11027193e-01
-4.36438501e-01 -4.24940825e-01 -4.66771603e-01 3.74411233e-04
-2.95255154e-01 2.32998699e-01 1.23539996e+00 1.96415856e-01
9.60767984e-01 -1.75687417e-01 5.20445883e-01 -6.39630497e-01
-6.03493094e-01 -7.41919994e-01 -9.06624496e-01 8.82127404e-01
6.20728508e-02 -2.59934694e-01 -2.16621310e-01 1.84134662e-01] | [10.866503715515137, 1.0210316181182861] |
119d1546-8b8c-4857-87f1-2b63f43a9664 | a-survey-of-contextual-optimization-methods | 2306.10374 | null | https://arxiv.org/abs/2306.10374v1 | https://arxiv.org/pdf/2306.10374v1.pdf | A Survey of Contextual Optimization Methods for Decision Making under Uncertainty | Recently there has been a surge of interest in operations research (OR) and the machine learning (ML) community in combining prediction algorithms and optimization techniques to solve decision-making problems in the face of uncertainty. This gave rise to the field of contextual optimization, under which data-driven procedures are developed to prescribe actions to the decision-maker that make the best use of the most recently updated information. A large variety of models and methods have been presented in both OR and ML literature under a variety of names, including data-driven optimization, prescriptive optimization, predictive stochastic programming, policy optimization, (smart) predict/estimate-then-optimize, decision-focused learning, (task-based) end-to-end learning/forecasting/optimization, etc. Focusing on single and two-stage stochastic programming problems, this review article identifies three main frameworks for learning policies from data and discusses their strengths and limitations. We present the existing models and methods under a uniform notation and terminology and classify them according to the three main frameworks identified. Our objective with this survey is to both strengthen the general understanding of this active field of research and stimulate further theoretical and algorithmic advancements in integrating ML and stochastic programming. | ['Thibaut Vidal', 'Emma Frejinger', 'Alexandre Forel', 'Erick Delage', 'Abhilash Chenreddy', 'Utsav Sadana'] | 2023-06-17 | null | null | null | null | ['decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['medical', 'reasoning'] | [ 2.12808624e-01 -2.18257122e-02 -8.56841385e-01 -6.71629429e-01
-7.13580728e-01 -2.35482872e-01 3.80797207e-01 3.68917108e-01
-4.21237618e-01 9.73340392e-01 4.47508305e-01 -6.92613065e-01
-9.86970723e-01 -4.93437320e-01 -3.14944476e-01 -7.44890451e-01
7.72449225e-02 8.14335108e-01 -5.25815606e-01 1.79224506e-01
6.07112825e-01 4.07544881e-01 -1.71492827e+00 2.71884412e-01
1.22028744e+00 1.51801491e+00 2.08042502e-01 6.28590584e-01
-3.03289235e-01 8.03855956e-01 -2.25331888e-01 -4.21742886e-01
3.62910360e-01 -9.14739743e-02 -5.06679833e-01 6.53072372e-02
-2.63436198e-01 2.30327225e-03 1.10614002e-01 7.66322136e-01
6.21622145e-01 4.73841339e-01 7.90824115e-01 -1.29915416e+00
-3.40212166e-01 6.48052335e-01 -1.71809107e-01 3.12894166e-01
3.41765165e-01 3.13648373e-01 8.46314251e-01 -6.98540151e-01
1.19698122e-01 1.32977319e+00 3.92259151e-01 2.71204114e-01
-1.05130696e+00 -1.36500373e-01 6.30232751e-01 3.07546258e-01
-9.45102096e-01 -3.55490506e-01 6.40576899e-01 -7.27332652e-01
1.08283663e+00 3.64697993e-01 6.33075476e-01 8.18520367e-01
5.07694840e-01 9.81262147e-01 1.30666053e+00 -7.77620256e-01
4.86600757e-01 5.28434396e-01 2.95341510e-04 4.08498406e-01
1.69697050e-02 6.74299777e-01 -6.20351136e-01 -2.45290741e-01
2.81804055e-01 3.02130193e-01 8.03691372e-02 -3.42942148e-01
-9.46764529e-01 1.07636964e+00 -1.60715759e-01 -1.32205099e-01
-9.20527577e-01 -1.04774393e-01 3.24548304e-01 4.15140033e-01
6.89684689e-01 6.25280321e-01 -8.20549011e-01 -2.91033417e-01
-8.55719209e-01 4.87086862e-01 1.14518464e+00 6.32630467e-01
3.28328907e-01 2.48332933e-01 -4.58120048e-01 8.30031574e-01
7.63213873e-01 4.76893306e-01 5.24552345e-01 -1.04173565e+00
6.45435393e-01 2.76112735e-01 4.95829821e-01 -9.27458704e-01
-5.10574639e-01 -4.67264533e-01 -6.54708922e-01 1.79565534e-01
3.13471667e-02 -5.94607651e-01 -7.08530784e-01 1.22999454e+00
2.05889523e-01 -1.45569116e-01 1.80636346e-01 6.73931539e-01
2.27296889e-01 7.39125013e-01 2.05182269e-01 -8.57236803e-01
9.52821970e-01 -8.62766087e-01 -8.98162782e-01 -3.91760111e-01
2.88895726e-01 -6.90973341e-01 7.39517629e-01 6.72777176e-01
-9.01608825e-01 -2.26724699e-01 -6.21748865e-01 6.01057708e-01
-3.97637248e-01 -1.88225299e-01 6.89344943e-01 8.40199351e-01
-6.13252699e-01 5.69970906e-01 -8.37285578e-01 3.63513976e-02
3.34230214e-01 3.26304674e-01 4.61257547e-01 7.20951008e-03
-1.17944288e+00 1.38762045e+00 5.14852583e-01 3.20975244e-01
-8.24101090e-01 -8.53770912e-01 -7.13119984e-01 -1.75620124e-01
8.58572364e-01 -8.46840918e-01 1.39458191e+00 -9.85982358e-01
-1.94105113e+00 4.92674828e-01 -3.06816459e-01 -6.91689372e-01
7.52617717e-01 -5.22172630e-01 -4.18615997e-01 -5.74700832e-01
-1.57889590e-01 -1.39073506e-01 5.47198713e-01 -1.12416899e+00
-1.05178428e+00 -4.58162874e-01 -1.92147091e-01 4.70564127e-01
1.62117794e-01 3.86839002e-01 -4.85790931e-02 -6.68293893e-01
-2.55523115e-01 -7.40344465e-01 -8.30550492e-01 -6.54401541e-01
-4.09446716e-01 -2.97146618e-01 2.93073267e-01 -5.62858880e-01
1.95530665e+00 -1.52403140e+00 1.84618175e-01 3.96047801e-01
-3.29172850e-01 -3.85836065e-02 4.54366654e-01 7.56331325e-01
4.81314175e-02 1.69994399e-01 -5.03893316e-01 -4.04806525e-01
2.46857964e-02 4.45985824e-01 -5.29170632e-01 2.89324224e-01
-2.66033579e-02 8.31035376e-01 -8.22176576e-01 -3.54223758e-01
7.27189362e-01 -1.89402327e-01 -4.14027750e-01 3.61372411e-01
-6.38132036e-01 5.45550525e-01 -9.43026602e-01 1.02453375e+00
1.10789143e-01 5.27937710e-02 1.40627906e-01 4.11899716e-01
-5.15445113e-01 1.27452716e-01 -1.59328413e+00 1.09325814e+00
-8.16106915e-01 3.47016841e-01 1.21648042e-02 -1.50752568e+00
7.63682723e-01 4.37748045e-01 8.56157839e-01 -3.14309388e-01
8.59803855e-02 3.30422968e-01 -1.72699824e-01 -9.11370218e-01
1.22512490e-01 -3.55381936e-01 -1.50053166e-02 3.61795217e-01
-2.98839867e-01 -2.11300895e-01 6.61092103e-02 -6.20233059e-01
4.84427959e-01 9.55912843e-02 6.79428220e-01 -3.84250015e-01
6.73485935e-01 2.36373261e-01 6.73692167e-01 1.05176628e+00
-2.18157470e-01 3.88018116e-02 4.23196912e-01 -7.75047243e-01
-5.63798249e-01 -7.17956781e-01 -2.93206692e-01 1.36127007e+00
-3.02215159e-01 1.79872196e-02 -1.28203347e-01 -3.84641469e-01
4.21063393e-01 1.23780406e+00 -4.96818215e-01 1.54207900e-01
-3.19060266e-01 -1.26188350e+00 -3.34828049e-01 3.45490068e-01
1.97706223e-02 -1.09840000e+00 -4.97690976e-01 6.99651420e-01
9.99467149e-02 -7.90295541e-01 -1.05138861e-01 2.65553057e-01
-1.19486201e+00 -1.02611220e+00 -4.49520528e-01 -1.99409798e-01
3.95179898e-01 -4.24822092e-01 1.16120470e+00 -5.89770675e-01
1.14336342e-01 6.19640827e-01 -6.24900721e-02 -1.17646635e+00
-2.85355836e-01 -2.28319183e-01 5.30353010e-01 3.43554974e-01
2.85868973e-01 -2.59191841e-01 -4.05495763e-01 3.67638171e-01
-4.95663971e-01 -1.87086433e-01 5.65163076e-01 7.42017090e-01
7.78079152e-01 4.05839831e-02 5.59571683e-01 -1.03441787e+00
1.09540725e+00 -7.07211733e-01 -1.12413919e+00 5.66557467e-01
-1.53084803e+00 4.34250027e-01 4.59644347e-01 -1.45498693e-01
-1.23931026e+00 -1.54815853e-01 1.08571552e-01 -4.06318665e-01
-4.15895320e-02 1.08950508e+00 5.27983233e-02 2.30460688e-01
4.44745600e-01 2.90314913e-01 7.56930411e-02 -5.95506847e-01
3.19459021e-01 8.23284209e-01 7.50009269e-02 -8.40096354e-01
1.74562529e-01 8.57563987e-02 1.18309557e-01 -3.62658471e-01
-1.03316355e+00 -4.34791118e-01 -3.92938882e-01 -3.98780942e-01
6.46824777e-01 -5.23929417e-01 -7.04110205e-01 3.50173086e-01
-7.07679510e-01 -3.37643147e-01 -5.50517261e-01 7.20023274e-01
-1.00061119e+00 -1.50420353e-01 7.87409097e-02 -1.65010870e+00
-2.82931536e-01 -1.59121883e+00 5.29248834e-01 3.11140984e-01
-8.58142134e-03 -1.27720451e+00 2.04075202e-01 4.09719616e-01
5.24487495e-01 5.23025870e-01 8.65438282e-01 -8.12667549e-01
-4.02201861e-01 -4.02440965e-01 4.42510933e-01 6.75051868e-01
-1.15263402e-01 2.14667171e-01 -5.79141021e-01 1.09722674e-01
3.10881048e-01 -6.40147850e-02 5.06576419e-01 1.08468997e+00
1.33988488e+00 -7.59080648e-01 -3.28819096e-01 5.94917178e-01
1.65852213e+00 7.15714812e-01 -3.36899310e-02 5.64300239e-01
1.69917777e-01 9.07830417e-01 9.10646558e-01 9.34959114e-01
4.06268537e-01 5.08181334e-01 4.30243462e-01 5.58758080e-01
8.36286664e-01 -1.33463785e-01 3.83579433e-01 4.38885987e-01
-3.73553604e-01 -3.18186224e-01 -1.21527433e+00 2.72523731e-01
-2.33912492e+00 -8.96210015e-01 1.69710591e-01 2.43654776e+00
6.40428662e-01 8.90534744e-02 1.44260243e-01 -2.54187763e-01
5.25389373e-01 2.39391074e-01 -5.94746888e-01 -9.31299388e-01
1.41782790e-01 -1.58158764e-01 8.73006999e-01 8.33631158e-01
-1.17579234e+00 6.20245755e-01 7.48853970e+00 8.15347075e-01
-9.29934919e-01 -1.94126934e-01 1.12147510e+00 -2.82441616e-01
-2.21476868e-01 -1.03853017e-01 -9.07154202e-01 5.73676348e-01
1.30291259e+00 -4.76426810e-01 6.19248450e-01 1.13519335e+00
8.96460652e-01 -3.65388066e-01 -1.06776965e+00 8.50839913e-01
-2.58953482e-01 -1.53931761e+00 -7.44772553e-02 -4.98912893e-02
1.13681865e+00 1.37947530e-01 2.67979443e-01 4.21308339e-01
6.33848906e-01 -1.20122206e+00 5.94409764e-01 1.15372097e+00
3.10755551e-01 -8.26512516e-01 1.02365077e+00 7.17223823e-01
-6.57979786e-01 -7.60541677e-01 -8.31760466e-03 -3.51915568e-01
3.47259402e-01 8.58944356e-01 -4.89748597e-01 7.07526505e-01
5.55719614e-01 6.13553822e-01 1.80950582e-01 1.14808846e+00
-1.64312407e-01 5.19014657e-01 -3.94775152e-01 -3.61732483e-01
4.69426513e-01 -6.56870127e-01 6.82176948e-01 1.03898847e+00
5.12850434e-02 1.13604963e-01 5.85794687e-01 5.37094235e-01
4.93644714e-01 2.28535444e-01 -2.74200052e-01 -1.03550352e-01
4.67522919e-01 5.75917423e-01 -2.81732026e-02 -2.09859937e-01
-5.81095278e-01 1.52976196e-02 -1.09706089e-01 4.67296511e-01
-3.93400013e-01 3.28159742e-02 7.66741693e-01 -2.35347319e-02
-2.85222769e-01 -2.48943865e-01 -8.94937575e-01 -8.57065499e-01
-8.94861668e-02 -1.01501071e+00 7.93735802e-01 -1.73199713e-01
-1.24049377e+00 1.42746881e-01 5.12652814e-01 -8.06739509e-01
-6.16815865e-01 -7.85537839e-01 -5.02415180e-01 1.07963169e+00
-1.49997914e+00 -4.25097197e-01 2.36821041e-01 2.86708802e-01
7.90516257e-01 -4.83202875e-01 6.60852253e-01 -1.56265348e-01
-8.69514763e-01 3.97017784e-02 9.23574150e-01 -4.07570422e-01
3.13522875e-01 -1.23775363e+00 -3.83132994e-01 5.98070145e-01
-3.49209666e-01 2.97323436e-01 9.14769888e-01 -4.64478463e-01
-1.52848852e+00 -8.69409382e-01 9.43009317e-01 -6.85769022e-01
6.20390177e-01 1.29989192e-01 -3.91709328e-01 5.13471842e-01
-1.69801757e-01 -4.02293116e-01 7.35053003e-01 2.16986403e-01
4.37971622e-01 -4.33204174e-01 -1.20652986e+00 4.67490405e-01
4.60247904e-01 -3.72264609e-02 -6.36382818e-01 6.61898375e-01
4.83235568e-01 -3.43919963e-01 -1.00848913e+00 5.45837760e-01
5.25376320e-01 -8.74900162e-01 8.87921274e-01 -1.05253768e+00
1.94069117e-01 2.67849505e-01 -2.72336096e-01 -1.31480658e+00
-5.48267774e-02 -1.12524772e+00 -4.96466339e-01 6.09825432e-01
5.88162184e-01 -1.00063372e+00 6.12655878e-01 1.23497248e+00
-1.98689207e-01 -1.46127570e+00 -1.09586728e+00 -5.88259220e-01
-4.03292589e-02 -9.88934159e-01 5.36630750e-01 5.46087980e-01
-1.72667116e-01 -2.41713852e-01 -5.47559738e-01 3.76308002e-02
5.73196173e-01 2.75444150e-01 5.14880657e-01 -1.15669811e+00
-1.85408950e-01 -9.54902589e-01 1.01779206e-02 -9.44390059e-01
-6.15927903e-03 -6.41573668e-01 1.83396954e-02 -1.61233997e+00
-2.31058747e-01 -5.89907110e-01 -7.92262316e-01 9.87666547e-02
-2.17061758e-01 -9.16135073e-01 1.43887373e-02 3.17145020e-01
-3.45842242e-01 5.79205096e-01 1.01494479e+00 1.86625034e-01
-7.11491346e-01 9.59454477e-01 -8.55450988e-01 7.80700684e-01
7.47406363e-01 -5.90236366e-01 -3.27314705e-01 -1.68315947e-01
4.03900087e-01 3.96629274e-01 -6.30050078e-02 -4.64996219e-01
1.89071938e-01 -1.26579070e+00 4.44086194e-02 -5.15240848e-01
-6.35225996e-02 -8.79062057e-01 1.36867002e-01 6.24437690e-01
-6.59130692e-01 2.56404907e-01 7.28565902e-02 8.98304641e-01
-3.46393496e-01 -4.45872843e-01 6.68108463e-01 -2.24529102e-01
-7.69168258e-01 5.73939621e-01 -4.33534861e-01 2.52677590e-01
1.30278027e+00 -8.30999203e-03 3.66210073e-01 -4.89618152e-01
-9.96949136e-01 8.32024634e-01 -3.82680237e-01 4.85342056e-01
4.08566236e-01 -8.42509568e-01 -9.33767021e-01 -9.28940102e-02
-1.13837987e-01 -4.17516679e-02 1.62345782e-01 1.04995298e+00
-3.69965672e-01 1.07550251e+00 2.71317840e-01 -3.73710811e-01
-5.80976605e-01 7.47995257e-01 6.11532986e-01 -6.78661227e-01
-9.20266903e-04 8.28926027e-01 -2.25073442e-01 -4.45836514e-01
5.38719356e-01 -2.74239600e-01 -2.22852156e-01 2.14070693e-01
2.63076097e-01 8.78320217e-01 -1.32368654e-01 -1.51460066e-01
-3.33525717e-01 2.95968354e-01 3.10962230e-01 -2.84966588e-01
1.46930361e+00 -1.56968758e-01 1.67060420e-02 8.96865427e-01
6.40089273e-01 -4.09637898e-01 -1.30122066e+00 -5.24494588e-01
5.34500480e-01 -5.26313126e-01 3.88111323e-01 -1.09836698e+00
-8.00560951e-01 6.22593224e-01 4.90478307e-01 2.50146955e-01
1.18771338e+00 -3.18792552e-01 9.03197937e-03 3.83772343e-01
3.16536546e-01 -1.62844670e+00 -3.94331992e-01 5.27388215e-01
1.19985473e+00 -1.58362281e+00 1.98128626e-01 6.43534586e-02
-1.11922407e+00 1.13835084e+00 1.96534172e-01 -7.00183213e-02
1.38126695e+00 8.68065879e-02 -4.04238254e-02 1.12576574e-01
-1.03620994e+00 -4.44323942e-02 6.49406374e-01 6.34132385e-01
2.09421858e-01 3.49913031e-01 -4.18677092e-01 7.24450350e-01
-7.62036517e-02 1.53943092e-01 -6.42913878e-02 1.02021837e+00
-5.73167205e-01 -1.06639588e+00 -5.71851969e-01 9.05263126e-01
-6.40460312e-01 -1.52457997e-01 1.78183615e-02 5.80443084e-01
-1.47638470e-01 1.15183938e+00 -1.85132116e-01 -4.49035652e-02
4.83281285e-01 3.17793220e-01 -1.14099514e-02 -5.32464623e-01
-5.26231170e-01 -3.93593907e-02 1.88552842e-01 -6.63522363e-01
-3.54065895e-01 -9.50788915e-01 -6.35762095e-01 -5.93824759e-02
-2.16293558e-01 1.66604385e-01 9.80065346e-01 1.13805223e+00
3.31438810e-01 5.14459908e-01 7.50994682e-01 -6.95573568e-01
-1.19652474e+00 -8.48420084e-01 -4.12663788e-01 -3.76005948e-01
3.15261453e-01 -7.76448429e-01 -3.28955323e-01 -2.66899496e-01] | [4.521487712860107, 3.1688876152038574] |
d8fd2ac8-ed0f-4ddb-8f4f-69de6754ab05 | a-multi-label-multi-hop-relation-detection | null | null | https://aclanthology.org/2021.findings-emnlp.404 | https://aclanthology.org/2021.findings-emnlp.404.pdf | A Multi-label Multi-hop Relation Detection Model based on Relation-aware Sequence Generation | Multi-hop relation detection in Knowledge Base Question Answering (KBQA) aims at retrieving the relation path starting from the topic entity to the answer node based on a given question, where the relation path may comprise multiple relations. Most of the existing methods treat it as a single-label learning problem while ignoring the fact that for some complex questions, there exist multiple correct relation paths in knowledge bases. Therefore, in this paper, multi-hop relation detection is considered as a multi-label learning problem. However, performing multi-label multi-hop relation detection is challenging since the numbers of both the labels and the hops are unknown. To tackle this challenge, multi-label multi-hop relation detection is formulated as a sequence generation task. A relation-aware sequence relation generation model is proposed to solve the problem in an end-to-end manner. Experimental results show the effectiveness of the proposed method for relation detection and KBQA. | ['Yulan He', 'Chao Lin', 'Deyu Zhou', 'Linhai Zhang'] | null | null | null | null | findings-emnlp-2021-11 | ['knowledge-base-question-answering'] | ['natural-language-processing'] | [ 3.60082954e-01 4.14615393e-01 -2.68972278e-01 -1.64899021e-01
-1.12907982e+00 -7.43657947e-01 3.51431549e-01 5.93387306e-01
-1.11551419e-01 1.01084530e+00 -7.00948909e-02 -5.06878316e-01
-4.51883197e-01 -1.02378893e+00 -6.11794829e-01 -3.00464988e-01
1.11201331e-01 9.36769783e-01 7.72543013e-01 -3.76940519e-01
6.16919361e-02 1.55950695e-01 -1.16632581e+00 5.67694902e-01
1.03134072e+00 6.77891970e-01 5.34761027e-02 7.02066839e-01
-4.01338071e-01 1.15568352e+00 -5.65991342e-01 -7.20032811e-01
-2.59058177e-01 -7.30904818e-01 -1.69277942e+00 -2.59501487e-01
2.27625713e-01 9.41471756e-02 -1.86112970e-02 1.01571774e+00
2.91093618e-01 2.20278352e-01 7.48734832e-01 -1.29115438e+00
-2.52572507e-01 4.79404807e-01 -4.90169108e-01 5.23405254e-01
7.52625227e-01 -6.21736884e-01 1.57337570e+00 -4.16508555e-01
7.19060719e-01 1.43048441e+00 3.03527594e-01 1.33863911e-01
-7.59889960e-01 -4.69085693e-01 1.26970112e-01 8.96145344e-01
-1.63702393e+00 -3.88007760e-02 5.32034457e-01 -3.74035627e-01
1.05550897e+00 3.03420156e-01 1.59131348e-01 4.13230598e-01
-2.49217506e-02 5.38934827e-01 7.20287800e-01 -7.77499139e-01
-6.82717115e-02 3.23640555e-02 4.37853068e-01 8.42841804e-01
1.16525866e-01 -5.13991177e-01 -4.13637817e-01 -4.05841291e-01
2.40664646e-01 -4.79120970e-01 -2.68408984e-01 3.41438614e-02
-9.78485763e-01 7.92672813e-01 4.65114743e-01 3.34274232e-01
-3.62658173e-01 -3.07306945e-01 5.25094271e-01 2.99176902e-01
3.81505996e-01 3.25760871e-01 -7.81821191e-01 1.52988777e-01
-4.11290944e-01 3.50209564e-01 1.33704948e+00 1.12862003e+00
1.06788337e+00 -9.72511470e-01 -1.76980287e-01 7.09328532e-01
3.25257272e-01 -8.70116130e-02 1.78867742e-01 -6.19969487e-01
7.37781048e-01 9.81610060e-01 1.90784082e-01 -1.21047401e+00
-5.47485650e-01 -1.32763505e-01 -5.14214814e-01 -5.74934840e-01
4.46398944e-01 -1.97157338e-01 -4.59027261e-01 1.57859814e+00
1.05974603e+00 5.17205775e-01 3.44035357e-01 7.86562979e-01
1.04130006e+00 9.73894656e-01 2.20588133e-01 -6.34723425e-01
2.01834536e+00 -1.30808067e+00 -9.56819236e-01 -1.93030089e-01
1.02136528e+00 -8.26893508e-01 4.69169825e-01 -6.75284117e-02
-5.21530628e-01 -2.78132349e-01 -9.08096552e-01 -2.61407912e-01
-2.13381559e-01 -6.64997846e-02 4.84313637e-01 3.17963451e-01
-5.48221827e-01 -1.60226643e-01 -2.45748281e-01 -5.38904846e-01
-9.56770107e-02 1.33792907e-01 -1.99924365e-01 -5.35918593e-01
-1.69067371e+00 1.08869123e+00 1.11745918e+00 8.94262921e-03
-5.41980863e-01 -3.53229046e-01 -6.67187154e-01 2.95566535e-03
1.23818231e+00 -7.58319557e-01 1.31271970e+00 -3.04441392e-01
-1.05999720e+00 7.16182709e-01 -5.75508595e-01 -2.66196281e-01
1.49134751e-02 -2.05266029e-01 -5.83575130e-01 3.97819936e-01
2.99908549e-01 7.40639567e-02 3.35327119e-01 -1.18962061e+00
-1.15949643e+00 -2.03746811e-01 4.79945123e-01 6.70674860e-01
2.59022772e-01 4.35432717e-02 -5.49064517e-01 -1.50996700e-01
2.77886987e-01 -9.17905092e-01 7.36205280e-02 -6.75336182e-01
-7.45837212e-01 -8.06764960e-01 7.98302293e-01 -9.73052025e-01
1.28578997e+00 -1.65752852e+00 -1.35030121e-01 1.21300787e-01
1.25435844e-01 2.63907760e-01 2.89375801e-02 6.95091844e-01
4.37401328e-03 4.17776667e-02 -1.47065893e-01 3.58800381e-01
-2.90674537e-01 2.41879717e-01 -1.52896896e-01 6.01293184e-02
2.44018763e-01 9.28419173e-01 -1.18346286e+00 -1.08333063e+00
-3.08567464e-01 1.19922310e-01 -5.18904887e-02 4.27493036e-01
-6.45121813e-01 4.43789721e-01 -6.84880853e-01 5.23964405e-01
4.82523620e-01 -5.14320731e-01 5.89539349e-01 -3.94703716e-01
3.31173092e-01 4.00092602e-01 -1.16920710e+00 1.37561488e+00
-4.69197720e-01 6.95776716e-02 -2.88593352e-01 -9.64992642e-01
8.77101719e-01 7.01096654e-01 3.79006416e-01 -5.73170125e-01
-1.45301029e-01 3.57378751e-01 2.61563361e-01 -9.45164144e-01
4.37879622e-01 -5.34220278e-01 1.06944881e-01 4.38618660e-01
-1.35731036e-02 2.48401403e-01 5.92221320e-01 3.60732287e-01
1.34614134e+00 6.11761250e-02 7.66155720e-01 2.61078954e-01
1.03038800e+00 3.24330240e-01 7.29175091e-01 4.13423806e-01
1.42944381e-01 6.85346723e-02 6.50846660e-01 -3.31097730e-02
-5.74720442e-01 -6.04903281e-01 1.48372576e-01 1.08889830e+00
4.00618732e-01 -5.95863283e-01 -4.34241772e-01 -1.01281714e+00
-2.58654088e-01 5.60170770e-01 -2.13267073e-01 -1.64348617e-01
-7.48035848e-01 -6.84675276e-01 5.03202677e-01 -8.25762600e-02
4.88049269e-01 -1.11840296e+00 -2.87478179e-01 5.30275583e-01
-9.81369436e-01 -1.54735982e+00 -2.70464927e-01 -1.44743435e-02
-5.24740636e-01 -1.52228403e+00 -3.61160070e-01 -1.32263327e+00
3.49990845e-01 2.21418306e-01 1.17546546e+00 2.45510936e-01
-1.91834480e-01 -2.52341647e-02 -8.00976574e-01 -6.31821156e-02
-4.76799309e-01 5.17503083e-01 -5.43189108e-01 -2.96518300e-02
3.89650762e-01 -3.00212234e-01 -3.11069578e-01 3.11907470e-01
-8.31295073e-01 -1.08726189e-01 5.73873460e-01 6.17927551e-01
6.91396356e-01 6.98387444e-01 1.08147871e+00 -1.26587832e+00
9.00109589e-01 -7.24433124e-01 -2.83716589e-01 1.02116024e+00
-4.60225642e-01 1.39612138e-01 4.01246756e-01 -2.43434638e-01
-1.34554446e+00 -1.21677123e-01 -1.99579060e-01 4.10162181e-01
-2.08743274e-01 1.04936469e+00 -6.09643281e-01 1.24814793e-01
3.51762056e-01 1.01612359e-01 -5.06322145e-01 -3.74095082e-01
7.45733917e-01 4.75921631e-01 3.17712069e-01 -6.17604733e-01
4.85749364e-01 -3.63827273e-02 2.28949770e-01 -5.52591264e-01
-1.32883358e+00 -1.03379440e+00 -7.06806362e-01 -7.31981173e-02
8.88069212e-01 -8.02032769e-01 -1.05037427e+00 -3.84528860e-02
-1.34696341e+00 1.30649835e-01 1.35952219e-01 2.54415065e-01
-2.46788010e-01 5.40452421e-01 -6.52790904e-01 -8.86737525e-01
-4.54694629e-01 -7.94224322e-01 8.39589357e-01 2.15221286e-01
-3.03107291e-01 -9.44608450e-01 3.60981464e-01 7.31520116e-01
-4.04930413e-01 6.09101877e-02 1.73957682e+00 -1.06213522e+00
-7.26953447e-01 -3.18438381e-01 -4.58388299e-01 -1.64458558e-01
2.88173527e-01 -5.36888778e-01 -4.62114424e-01 1.43690437e-01
-2.50472426e-01 -4.60624278e-01 5.18451810e-01 -1.55022994e-01
2.68357337e-01 -5.48448145e-01 -5.67474127e-01 -1.65446177e-01
1.38917375e+00 3.70214343e-01 2.93093950e-01 3.09149295e-01
1.02030122e+00 1.08148563e+00 1.18020236e+00 1.33676767e-01
8.66609693e-01 8.48769426e-01 1.95118576e-01 3.33300740e-01
5.21948077e-02 -2.51031071e-01 -2.79335588e-01 8.79751265e-01
3.52566689e-01 -6.13723576e-01 -1.15473461e+00 4.99860376e-01
-2.02285218e+00 -7.49841869e-01 -6.05973423e-01 1.86380637e+00
1.14928639e+00 1.46463281e-04 3.90224345e-02 2.98481286e-01
8.63297880e-01 -8.41231868e-02 -5.68213344e-01 -6.62716478e-02
-5.40087223e-02 -1.34705871e-01 -1.46464230e-02 8.35173011e-01
-9.85169351e-01 1.15024388e+00 4.86937952e+00 1.05636489e+00
-4.65858877e-01 2.51846284e-01 3.25361133e-01 5.90086937e-01
-1.00098848e-01 3.92825663e-01 -9.61261094e-01 2.17629652e-02
7.43126988e-01 -2.41579026e-01 8.27333108e-02 4.61823434e-01
-2.56834745e-01 -4.42834944e-01 -8.29896331e-01 7.11817384e-01
5.15111573e-02 -8.83619905e-01 1.16594531e-01 -3.05649132e-01
5.34326255e-01 -5.46375811e-01 -6.73598349e-01 3.87880057e-01
2.53206044e-01 -7.12719500e-01 2.83453584e-01 3.30262244e-01
5.34223855e-01 -9.43536460e-01 9.46900547e-01 8.54901493e-01
-1.61694646e+00 7.94629380e-02 -1.72007039e-01 1.09229147e-01
6.06872678e-01 8.14545512e-01 -1.22066689e+00 1.19719577e+00
2.89939076e-01 4.02633578e-01 -3.85581344e-01 1.07445860e+00
-6.15955591e-01 5.49716651e-01 -1.90513775e-01 -1.43999949e-01
5.79226762e-02 -2.00885847e-01 2.65646160e-01 1.10677493e+00
1.31284118e-01 5.96758485e-01 3.49829704e-01 1.74204946e-01
-3.25053543e-01 6.11688137e-01 -2.89948612e-01 4.13134359e-02
6.53098762e-01 1.29880953e+00 -8.52703869e-01 -2.81722724e-01
-3.64490271e-01 9.72695589e-01 6.15831852e-01 3.49087059e-01
-5.54321051e-01 -4.80883271e-01 1.31534159e-01 -3.05620044e-01
1.16378888e-01 9.19881910e-02 1.94407538e-01 -7.76789546e-01
1.43806130e-01 -8.21829379e-01 9.15792048e-01 -5.87965071e-01
-1.17797875e+00 5.60130000e-01 -2.45759524e-02 -8.20228040e-01
-3.21448803e-01 1.99082959e-02 -1.39603883e-01 9.13903713e-01
-1.73251486e+00 -1.26396322e+00 -7.60737211e-02 3.54327559e-01
3.20400417e-01 2.52184212e-01 9.74265575e-01 4.65858608e-01
-5.85108936e-01 4.24497068e-01 -2.34165862e-01 -2.51843911e-02
5.26470184e-01 -1.15986001e+00 7.00359046e-03 6.69846058e-01
2.22752959e-01 3.32635581e-01 5.93741000e-01 -8.61251414e-01
-9.67660606e-01 -1.04545784e+00 1.48037088e+00 -9.42010283e-02
6.25738025e-01 2.81263106e-02 -1.19200778e+00 6.09052062e-01
1.46710828e-01 -1.64056808e-01 8.82083356e-01 3.45985621e-01
-3.72058898e-01 5.38689010e-02 -9.94455040e-01 2.96548158e-01
1.00175261e+00 -6.34569287e-01 -6.10836983e-01 7.15323031e-01
1.04613459e+00 -3.80571753e-01 -1.02441370e+00 5.08315146e-01
1.20332055e-01 -3.68387312e-01 9.83406365e-01 -6.00013196e-01
3.35999668e-01 -5.05505085e-01 3.39319557e-02 -9.80152667e-01
-8.63259658e-02 -3.76068532e-01 -7.32339025e-01 1.48579574e+00
7.51751006e-01 -2.55749226e-01 6.83148324e-01 8.52674618e-02
2.06508309e-01 -8.54585350e-01 -1.09658277e+00 -5.01414895e-01
-3.26915652e-01 -1.19369440e-02 5.99707186e-01 9.83767450e-01
2.58816630e-01 1.36075878e+00 -2.95952380e-01 6.05827034e-01
3.80666882e-01 5.37541866e-01 3.49619895e-01 -1.34410870e+00
-2.37669840e-01 1.21123888e-01 -1.64304003e-01 -1.20444822e+00
3.04898947e-01 -1.00220644e+00 3.82829964e-01 -1.97685719e+00
2.23537296e-01 -6.50643229e-01 -3.20375450e-02 2.76586890e-01
-7.77499259e-01 -2.65035719e-01 -2.31458202e-01 2.96508491e-01
-9.55985963e-01 3.94719779e-01 1.33665526e+00 -9.99517813e-02
-3.08570955e-02 1.02448896e-01 -5.83234966e-01 4.34844106e-01
6.23033583e-01 -6.77162409e-01 -8.04713547e-01 -1.26053458e-02
6.42730415e-01 6.99085772e-01 -8.37651920e-03 -7.38550186e-01
4.93895143e-01 -2.18652815e-01 -5.20326197e-01 -7.53804803e-01
3.82957667e-01 -7.74953008e-01 -4.28694002e-02 3.35939527e-01
-4.41605687e-01 -1.58209413e-01 -2.37028360e-01 8.26301455e-01
-4.53274071e-01 -5.59288502e-01 2.83990085e-01 -1.32736385e-01
-9.27813053e-01 3.17549348e-01 -1.53662130e-01 5.02000630e-01
1.20154154e+00 3.88052970e-01 -5.31938672e-01 -3.09473634e-01
-7.11930692e-01 4.53631997e-01 -3.08637887e-01 4.01740342e-01
5.11335790e-01 -1.05557287e+00 -7.60484755e-01 -6.32926464e-01
5.44971824e-01 4.25862521e-01 3.09672058e-01 5.44885874e-01
-5.46388149e-01 5.71540236e-01 4.05024469e-01 -3.66892703e-02
-1.50353467e+00 7.29778588e-01 2.13989869e-01 -9.00859714e-01
-3.22638005e-01 1.16525400e+00 -6.75651357e-02 -5.07015944e-01
2.34291792e-01 1.38850316e-01 -8.75409722e-01 4.06011254e-01
4.66685981e-01 2.74896860e-01 2.59319335e-01 -8.04212451e-01
-4.07365024e-01 4.64847714e-01 -3.17087024e-01 4.99200001e-02
7.41611600e-01 -4.01729107e-01 -6.13232791e-01 4.71743733e-01
1.16225851e+00 -3.87183249e-01 -4.07309949e-01 -7.73023129e-01
6.76782548e-01 -3.64335664e-02 -4.31567013e-01 -8.33785117e-01
-5.09975433e-01 5.49257755e-01 1.34190202e-01 2.70635873e-01
8.81479442e-01 4.07149971e-01 1.13152587e+00 7.07487226e-01
4.17124748e-01 -7.07841933e-01 1.38074934e-01 7.84323215e-01
5.99312007e-01 -1.12034905e+00 -7.72877410e-03 -1.28770518e+00
-5.06574810e-01 8.39190304e-01 8.74097228e-01 5.62623203e-01
6.89440012e-01 -1.43124819e-01 2.08571076e-01 -4.23070937e-01
-7.82628357e-01 -5.58011651e-01 2.91157037e-01 4.79946941e-01
4.25629407e-01 9.61514097e-03 -7.46619225e-01 2.11982787e-01
-1.72413915e-01 -1.39220819e-01 1.46877378e-01 1.17713737e+00
-6.15618825e-01 -1.38474250e+00 -2.22107008e-01 4.38196898e-01
-4.62335199e-01 -2.09051982e-01 -3.27429980e-01 3.10062110e-01
3.88213456e-01 1.51326990e+00 -5.09452581e-01 -4.13275898e-01
4.25750643e-01 3.96508396e-01 3.31639320e-01 -9.75050569e-01
-4.57409322e-01 -2.66315699e-01 6.08742476e-01 1.18535668e-01
-5.11753023e-01 -3.67372155e-01 -1.61042106e+00 -2.48144120e-02
-8.35318685e-01 7.14274049e-01 1.71405807e-01 1.54249322e+00
9.50646400e-02 7.30109215e-01 3.70022357e-01 2.03610912e-01
-2.59663910e-01 -1.00380075e+00 -5.34245849e-01 3.83707821e-01
1.81033209e-01 -5.21007180e-01 2.38731667e-01 1.03770070e-01] | [9.503011703491211, 8.462082862854004] |
d143364e-e400-4dc2-b588-1653caa32672 | adversarial-clean-label-backdoor-attacks-and | 2305.19607 | null | https://arxiv.org/abs/2305.19607v1 | https://arxiv.org/pdf/2305.19607v1.pdf | Adversarial Clean Label Backdoor Attacks and Defenses on Text Classification Systems | Clean-label (CL) attack is a form of data poisoning attack where an adversary modifies only the textual input of the training data, without requiring access to the labeling function. CL attacks are relatively unexplored in NLP, as compared to label flipping (LF) attacks, where the latter additionally requires access to the labeling function as well. While CL attacks are more resilient to data sanitization and manual relabeling methods than LF attacks, they often demand as high as ten times the poisoning budget than LF attacks. In this work, we first introduce an Adversarial Clean Label attack which can adversarially perturb in-class training examples for poisoning the training set. We then show that an adversary can significantly bring down the data requirements for a CL attack, using the aforementioned approach, to as low as 20% of the data otherwise required. We then systematically benchmark and analyze a number of defense methods, for both LF and CL attacks, some previously employed solely for LF attacks in the textual domain and others adapted from computer vision. We find that text-specific defenses greatly vary in their effectiveness depending on their properties. | ['Amrith Krishna', 'Ashim Gupta'] | 2023-05-31 | null | null | null | null | ['data-poisoning'] | ['adversarial'] | [ 6.94007337e-01 2.08297864e-01 -8.79059487e-04 -1.95747688e-01
-1.11786973e+00 -1.74504328e+00 8.32069516e-01 6.84687614e-01
-6.29659772e-01 7.27231264e-01 -1.97598279e-01 -6.56269014e-01
1.55119047e-01 -7.80148268e-01 -7.72683203e-01 -9.76586282e-01
4.47605252e-02 4.78086710e-01 1.09506406e-01 -1.33442700e-01
2.48510882e-01 8.60087156e-01 -1.05807745e+00 1.25067219e-01
4.20476049e-01 5.37921131e-01 -8.02625716e-01 7.51077831e-01
8.40982199e-02 7.80569792e-01 -1.11784697e+00 -6.27679527e-01
6.79086745e-01 -1.27185792e-01 -1.05865860e+00 -1.79750875e-01
8.13722670e-01 -5.68628795e-02 -4.32578117e-01 1.51376748e+00
5.95559359e-01 -1.11046739e-01 7.44815171e-01 -1.56461751e+00
-4.45863903e-01 7.02191412e-01 -4.27438706e-01 1.42822832e-01
4.34746385e-01 3.99575025e-01 7.41839230e-01 -4.23135310e-01
6.09693050e-01 1.20571220e+00 6.01988256e-01 9.86899793e-01
-1.38785434e+00 -9.14641023e-01 -1.00218621e-03 -6.95742443e-02
-1.11894119e+00 -3.91332060e-01 6.55580044e-01 -3.83750916e-01
3.91786575e-01 6.36291981e-01 -1.62066475e-01 1.47965169e+00
-1.95106827e-02 5.21504343e-01 1.44156575e+00 -5.29003024e-01
4.76893723e-01 9.80930105e-02 4.25567955e-01 6.99383914e-01
5.16416728e-01 2.68543363e-01 -2.84506649e-01 -9.54494536e-01
1.40595231e-02 -2.39887863e-01 -3.44560027e-01 -3.15966457e-01
-1.04558372e+00 1.07778621e+00 3.48957092e-01 1.07564427e-01
1.06197976e-01 3.44751000e-01 8.29694390e-01 5.71505845e-01
2.30212301e-01 9.08151448e-01 -5.38642585e-01 4.65673208e-01
-6.49645329e-01 3.64601940e-01 8.79751086e-01 8.93187404e-01
6.89840734e-01 -2.82496270e-02 -3.27221274e-01 5.02686538e-02
-1.23151481e-01 8.10711324e-01 3.19410563e-01 -7.24607408e-01
4.59262252e-01 1.24249995e-01 1.41862601e-01 -9.42698181e-01
-4.25050914e-01 -1.41289026e-01 -6.06005728e-01 3.43456745e-01
7.96053290e-01 -3.89998317e-01 -1.13028634e+00 2.15792871e+00
7.20438182e-01 -1.32842660e-01 1.18264146e-01 3.55386823e-01
4.09428358e-01 2.99820691e-01 4.92425740e-01 -2.44529620e-01
1.53886712e+00 -5.88930368e-01 -7.16336429e-01 -1.33022562e-01
8.93875182e-01 -5.17598212e-01 1.01999664e+00 5.25334895e-01
-8.69546235e-01 -3.38692367e-02 -1.00498593e+00 -3.21667604e-02
-8.61741245e-01 -5.97844362e-01 4.82190758e-01 1.38349760e+00
-6.21628463e-01 4.48890865e-01 -6.62767768e-01 -2.13213563e-01
5.89656651e-01 4.96760935e-01 -6.04573250e-01 -8.02103356e-02
-1.52278674e+00 8.95548820e-01 3.46989125e-01 -4.01678115e-01
-1.29541326e+00 -8.48299384e-01 -8.67451608e-01 -1.94453090e-01
5.38009465e-01 -4.49513614e-01 1.22440886e+00 -4.28325385e-01
-1.02976513e+00 9.66885328e-01 1.70800760e-01 -6.93372130e-01
7.05568135e-01 3.09268516e-02 -3.19072783e-01 3.14939231e-01
-3.73984221e-03 4.78589207e-01 1.15268350e+00 -1.48615110e+00
-5.52745938e-01 -2.67416716e-01 4.00164902e-01 -2.17813730e-01
-4.96763021e-01 1.88195959e-01 1.86376676e-01 -8.91422510e-01
-3.81837934e-01 -1.07861638e+00 -3.56866241e-01 7.86426943e-03
-9.05002177e-01 -2.51073837e-02 1.36222363e+00 -3.00525457e-01
1.01917136e+00 -2.00714636e+00 -9.47369039e-02 3.87805164e-01
3.34626049e-01 6.53355539e-01 -2.57527828e-01 6.00112736e-01
-2.55770117e-01 6.63708329e-01 -6.74799919e-01 -3.87945682e-01
2.03258023e-01 2.08051533e-01 -8.89551878e-01 1.02786064e+00
1.08525893e-02 7.64746428e-01 -9.86668229e-01 -2.17029244e-01
-5.43134175e-02 1.03185020e-01 -4.43118781e-01 1.00143351e-01
-4.54499096e-01 4.74425286e-01 -3.25711906e-01 6.83871508e-01
7.79304445e-01 2.12350383e-01 6.30729869e-02 -1.50852501e-01
4.21579897e-01 1.08319536e-01 -7.96650290e-01 1.22220862e+00
-1.45925537e-01 3.11301261e-01 8.38079005e-02 -7.79573619e-01
5.38612604e-01 4.35215473e-01 2.27154538e-01 -1.12982228e-01
2.93992877e-01 1.16454937e-01 -1.52863622e-01 -3.52380246e-01
7.51170889e-02 -4.44887817e-01 -6.85983896e-01 8.95014048e-01
-1.53662950e-01 -7.78893083e-02 1.52610183e-01 4.08212423e-01
1.64957440e+00 -4.44542795e-01 3.28050345e-01 -2.77088791e-01
6.71898961e-01 1.35985821e-01 3.86587232e-01 1.13054752e+00
-2.32871905e-01 2.20046952e-01 6.32755995e-01 -4.19630140e-01
-9.67783868e-01 -9.13532615e-01 -1.08644158e-01 9.77841377e-01
8.05541500e-02 -4.48660940e-01 -1.16714740e+00 -1.67578924e+00
1.98408931e-01 8.32365990e-01 -8.54207218e-01 -5.30284286e-01
-4.96290505e-01 -7.08869219e-01 1.54835725e+00 2.39087239e-01
3.94420952e-01 -1.12045157e+00 -4.04213607e-01 -1.84570611e-01
-4.05293442e-02 -1.03791666e+00 -5.51700830e-01 6.10490620e-01
-3.59075576e-01 -1.35922801e+00 -6.48926124e-02 -4.91285205e-01
9.78623629e-01 -9.35431719e-02 7.76870310e-01 2.39702687e-01
-2.81969845e-01 3.47551137e-01 -4.43015575e-01 -6.38437629e-01
-9.97963607e-01 2.58935362e-01 2.60375679e-01 -4.49614488e-02
4.80638206e-01 -4.18195456e-01 -7.76116271e-03 1.44118220e-01
-1.49221945e+00 -6.91712201e-01 3.62354033e-02 7.14319706e-01
3.55782837e-01 3.05311561e-01 5.34041107e-01 -1.67891538e+00
6.71423197e-01 -5.68289220e-01 -5.33027172e-01 2.87053078e-01
-4.03495044e-01 1.37241244e-01 1.05345798e+00 -8.94199014e-01
-4.14586574e-01 2.88379222e-01 -1.74878478e-01 -5.93361437e-01
-4.61378485e-01 8.54872689e-02 -5.03038883e-01 -4.62278396e-01
1.07477903e+00 -7.88306594e-02 -3.02616894e-01 -4.42728758e-01
5.41794956e-01 4.90167528e-01 7.38958240e-01 -7.45681345e-01
1.61525655e+00 7.02176392e-01 3.06035966e-01 -3.68195593e-01
-1.18103886e+00 -7.18057677e-02 -4.61509109e-01 2.89742976e-01
7.81133831e-01 -3.55948806e-01 -8.11415732e-01 6.60542011e-01
-1.05600309e+00 -3.05535764e-01 -5.22047102e-01 -2.02933475e-01
-4.20546353e-01 6.65557206e-01 -6.88146234e-01 -4.28098708e-01
-4.07151461e-01 -1.03081465e+00 7.48435855e-01 -2.60236204e-01
-1.97772935e-01 -1.01078415e+00 -2.80433428e-02 2.92933553e-01
1.45435467e-01 9.25851941e-01 1.25115383e+00 -1.46784770e+00
-2.77620628e-02 -6.97664738e-01 1.48047730e-01 2.67458260e-01
3.55382144e-01 -2.88793236e-01 -1.19258797e+00 -7.87218750e-01
5.19502163e-01 -7.40504980e-01 5.65710962e-01 -3.84465545e-01
1.19423389e+00 -7.60136366e-01 -3.35516751e-01 5.35545051e-01
1.30624712e+00 5.22482730e-02 5.57681382e-01 3.61375093e-01
9.41269457e-01 6.73635125e-01 5.94847322e-01 2.64864206e-01
-2.76608497e-01 4.06198204e-01 5.85359991e-01 -1.61141753e-01
1.35713354e-01 -2.00488210e-01 3.02682966e-01 -3.37980166e-02
6.32806182e-01 -5.43471813e-01 -8.01716387e-01 2.23309129e-01
-1.38437688e+00 -9.69401002e-01 -2.28703737e-01 2.24894142e+00
1.25655007e+00 1.44399375e-01 8.42241496e-02 5.61722517e-01
7.94409215e-01 1.40184209e-01 -6.98354602e-01 -6.41027868e-01
-1.87113658e-02 4.83437777e-01 1.00639594e+00 7.01570988e-01
-1.55778086e+00 1.09029436e+00 6.93626165e+00 1.10658360e+00
-9.46244240e-01 3.08473498e-01 4.47197199e-01 -4.80654612e-02
-3.53911608e-01 1.81838509e-03 -7.66728938e-01 4.73133951e-01
1.01746416e+00 -1.09114110e-01 4.61604148e-01 7.21018195e-01
-4.04572546e-01 2.98848659e-01 -1.41680348e+00 5.29108405e-01
1.18725345e-01 -9.45884943e-01 1.46003187e-01 2.56113917e-01
6.42594039e-01 -2.69113034e-01 2.08677828e-01 1.15588307e-01
7.88379014e-01 -1.20451748e+00 7.64062822e-01 -1.34038597e-01
9.97643292e-01 -1.10477674e+00 5.60785770e-01 5.72717369e-01
-6.73854291e-01 -2.12893084e-01 -2.21685857e-01 2.37629265e-01
9.45332181e-03 4.67642993e-01 -5.06589532e-01 3.91460359e-01
4.05574501e-01 -1.05708279e-02 -8.22324097e-01 3.89143646e-01
-5.70591629e-01 7.85093009e-01 -5.06378651e-01 4.07109469e-01
3.57894778e-01 4.99328494e-01 6.88131988e-01 1.31851566e+00
-2.62144685e-01 2.03527771e-02 4.67258543e-01 4.87821937e-01
-4.21735793e-01 -2.22841442e-01 -8.38507712e-01 -2.44982541e-02
7.97164857e-01 1.25476146e+00 -6.12706006e-01 -3.38467240e-01
-1.26840482e-02 9.10925150e-01 2.26349041e-01 1.79275066e-01
-9.72219527e-01 -5.44060469e-01 6.33574367e-01 -7.97282904e-02
1.60523668e-01 -5.32397255e-02 -1.98970124e-01 -9.36683476e-01
-2.59989142e-01 -1.49331200e+00 6.53067112e-01 -1.64311394e-01
-1.56258857e+00 5.53222060e-01 1.82817783e-02 -9.48724151e-01
-1.18390717e-01 -5.42637348e-01 -3.69424731e-01 8.92946422e-01
-1.26497352e+00 -1.12586617e+00 9.16082337e-02 9.85673726e-01
1.78332582e-01 -8.76707435e-02 1.12806296e+00 1.27997458e-01
-5.79874277e-01 1.01300120e+00 -9.05131996e-02 2.57240266e-01
8.17112982e-01 -1.30959320e+00 7.13940680e-01 1.19279563e+00
2.57081389e-01 7.51161695e-01 9.78939652e-01 -7.77340531e-01
-1.22436154e+00 -1.29483461e+00 7.45097101e-01 -9.07130539e-01
7.84448206e-01 -8.41304958e-01 -9.07465458e-01 8.16775680e-01
-7.06663029e-03 3.34751666e-01 8.01734149e-01 -3.69189769e-01
-1.13257074e+00 3.54042500e-01 -1.91305172e+00 7.12568343e-01
8.98171544e-01 -7.71791577e-01 -4.08720851e-01 6.85207844e-01
8.27230215e-01 -3.78941506e-01 -6.64105594e-01 3.61992925e-01
-9.48984548e-03 -5.14569223e-01 9.87557948e-01 -9.05623734e-01
-8.49398375e-02 -5.70177734e-01 -2.46574506e-01 -9.43250299e-01
-2.71141917e-01 -9.20670271e-01 -2.34890580e-01 1.41214502e+00
3.50435197e-01 -7.89340198e-01 7.03106403e-01 5.19273698e-01
3.13369632e-01 -3.05401593e-01 -9.35172379e-01 -9.07413840e-01
5.18227339e-01 -1.99391559e-01 6.44087195e-01 1.37623751e+00
-2.72113979e-01 4.69013229e-02 -3.37898791e-01 6.26067996e-01
8.82147193e-01 -4.33216155e-01 8.05033207e-01 -1.08366251e+00
-2.21509665e-01 -1.10320188e-02 -1.97539821e-01 -3.43117476e-01
3.56716424e-01 -1.11637962e+00 6.53844848e-02 -8.11449170e-01
6.29203336e-04 -5.24106801e-01 -2.19418362e-01 1.07724762e+00
-2.38962606e-01 6.96299791e-01 2.82084137e-01 3.21486562e-01
-1.49021983e-01 -3.28665762e-03 6.78667963e-01 -4.32254970e-01
2.03783438e-01 -9.40833688e-02 -8.96111846e-01 7.98493922e-01
9.06168461e-01 -1.32736778e+00 -4.55933541e-01 -1.36756167e-01
1.31805524e-01 -4.76567626e-01 3.45692247e-01 -8.48989129e-01
2.51454860e-01 -1.63406968e-01 5.16185425e-02 -1.59855574e-01
-7.68502355e-02 -1.03236020e+00 5.36388345e-02 7.25160897e-01
-7.27354348e-01 -7.61386529e-02 2.40101293e-01 7.49238670e-01
2.21145526e-01 -6.81050777e-01 1.09451687e+00 -7.13764429e-02
-6.86862618e-02 3.89944792e-01 -5.20105004e-01 1.52163997e-01
1.33222449e+00 1.13823004e-01 -8.06681633e-01 -1.16322920e-01
-6.59194469e-01 3.53086777e-02 6.86002076e-01 8.90862495e-02
3.57340835e-02 -1.00534797e+00 -5.33704102e-01 8.60611200e-02
-3.48189250e-02 -7.87919015e-02 -1.44064739e-01 4.04668927e-01
-4.28649426e-01 4.26554047e-02 -3.94443870e-02 -8.70354772e-02
-1.50107503e+00 1.53133035e+00 2.10009351e-01 -5.27885377e-01
-5.30562460e-01 7.03078091e-01 3.11884522e-01 -6.10849679e-01
4.54665601e-01 1.83532357e-01 1.53026551e-01 5.54450266e-02
6.52851701e-01 3.24888855e-01 8.29988718e-02 -4.81997788e-01
-4.19298679e-01 3.95993054e-01 -2.89765596e-01 -9.90401506e-02
6.72384560e-01 1.63615867e-01 -1.37014851e-01 -1.86070427e-02
1.14080667e+00 5.65000117e-01 -9.30097222e-01 -2.45117173e-01
2.12503731e-01 -3.52482855e-01 -2.36690849e-01 -9.10678685e-01
-9.28736210e-01 7.83735693e-01 3.72728795e-01 6.92195952e-01
1.16556025e+00 -1.71251312e-01 8.60076904e-01 6.33627653e-01
4.73826915e-01 -6.20843291e-01 1.84895359e-02 2.34823331e-01
5.75379372e-01 -8.19705188e-01 1.01223789e-01 -6.93406045e-01
-5.44465840e-01 7.10891068e-01 2.67552972e-01 -2.36178160e-01
4.57106888e-01 6.57368481e-01 1.65109769e-01 -1.59226924e-01
-4.51168031e-01 1.17052600e-01 -2.27918565e-01 9.26321268e-01
-3.73307586e-01 -2.37932473e-01 -1.75166413e-01 2.56943494e-01
-1.63090020e-01 -6.70473933e-01 7.01339483e-01 1.22692180e+00
-2.67446786e-01 -1.57344103e+00 -7.72707641e-01 3.50892134e-02
-1.02310169e+00 -1.75554708e-01 -7.50156581e-01 7.25661814e-01
3.05707902e-01 1.16974437e+00 -3.29816252e-01 -3.52552056e-01
2.20003992e-01 3.01448107e-01 2.63731867e-01 -7.69158721e-01
-1.15733576e+00 -2.84887403e-01 1.69950798e-01 -3.52686197e-01
-2.51319259e-01 -6.18463099e-01 -1.21239042e+00 -6.33863628e-01
-5.02207398e-01 2.31524482e-01 5.59684873e-01 9.43189800e-01
2.27653352e-03 1.02662213e-01 9.04453337e-01 -5.95192909e-01
-9.35555160e-01 -7.47144282e-01 -5.32103419e-01 5.87407827e-01
4.63105857e-01 -3.93657804e-01 -8.28516066e-01 2.54880130e-01] | [5.823422431945801, 7.702909469604492] |
06eaa8a4-8fb4-4c34-8781-349fc60cfdc7 | scene-domain-active-part-models-for-object | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Ren_Scene-Domain_Active_Part_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Ren_Scene-Domain_Active_Part_ICCV_2015_paper.pdf | Scene-Domain Active Part Models for Object Representation | In this paper, we are interested in enhancing the expressivity and robustness of part-based models for object representation, in the common scenario where the training data are based on 2D images. To this end, we propose scene-domain active part models (SDAPM), which reconstruct and characterize the 3D geometric statistics between object's parts in 3D scene-domain by using 2D training data in the image-domain alone. And on top of this, we explicitly model and handle occlusions in SDAPM. Together with the developed learning and inference algorithms, such a model provides rich object descriptions, including 2D object and parts localization, 3D landmark shape and camera viewpoint, which offers an effective representation to various image understanding tasks, such as object and parts detection, 3D landmark shape and viewpoint estimation from images. Experiments on the above tasks show that SDAPM outperforms previous part-based models, and thus demonstrates the potential of the proposed technique. | ['Chaohui Wang', 'Zhou Ren', 'Alan L. Yuille'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['viewpoint-estimation'] | ['computer-vision'] | [ 9.57089514e-02 6.84327334e-02 -2.30491444e-01 -3.24114680e-01
-6.10615313e-01 -4.19604719e-01 7.61609256e-01 4.16689515e-02
1.41282097e-01 2.03847930e-01 -4.75689657e-02 2.26319566e-01
-3.32090497e-01 -5.30840695e-01 -8.54265094e-01 -7.52342403e-01
3.91929448e-02 9.52510178e-01 4.61172968e-01 2.76277840e-01
3.91858488e-01 1.13599396e+00 -1.86360347e+00 -1.92530192e-02
5.52587271e-01 1.30731320e+00 5.09196520e-01 3.68424386e-01
-5.90031087e-01 6.06293797e-01 -3.23522806e-01 -1.76448762e-01
4.22844082e-01 2.29213893e-01 -5.52202702e-01 9.85609472e-01
5.85261583e-01 -3.65656435e-01 -2.56740510e-01 9.24363971e-01
1.54703066e-01 8.14188942e-02 9.84402716e-01 -1.33499992e+00
-3.12979847e-01 3.34569626e-02 -5.15389740e-01 -3.89426678e-01
4.49877352e-01 -8.97239223e-02 8.05043638e-01 -1.29159331e+00
5.72032154e-01 1.62669456e+00 6.39304161e-01 3.30864102e-01
-1.24109006e+00 -1.54198170e-01 4.53626692e-01 1.99069768e-01
-1.46751571e+00 -5.62805116e-01 1.15869391e+00 -6.43338025e-01
6.75528884e-01 2.23463744e-01 5.63319504e-01 6.94937468e-01
-1.35429621e-01 1.34885490e+00 9.19707537e-01 -5.81989884e-01
2.66860873e-01 3.13749760e-01 1.62104174e-01 6.32923126e-01
4.14034218e-01 -2.29140967e-01 -3.90711278e-01 -2.39587277e-01
1.10148942e+00 3.28601807e-01 -6.17334358e-02 -1.28044558e+00
-1.02247548e+00 5.10351837e-01 3.97580296e-01 -5.35246311e-03
-6.08296573e-01 -7.02242600e-03 7.42167991e-04 -2.04171330e-01
7.59142280e-01 8.58514383e-02 -5.03837109e-01 4.61835742e-01
-7.95260489e-01 2.28447944e-01 7.73488164e-01 1.62647212e+00
1.07145464e+00 -2.18921378e-01 -1.13203202e-03 8.79737377e-01
7.60361850e-01 7.88493574e-01 6.09107204e-02 -1.08758211e+00
8.33411664e-02 1.15818095e+00 2.31632859e-01 -8.44345629e-01
-2.03984678e-01 -1.97925627e-01 -6.06990695e-01 3.86673272e-01
9.55707058e-02 7.67518699e-01 -1.03255355e+00 1.31139195e+00
7.06489623e-01 4.16098014e-02 -1.62765250e-01 8.28173578e-01
9.82424498e-01 3.09570432e-01 -1.03770025e-01 -1.55566901e-01
1.22513628e+00 -8.52116406e-01 -6.17345035e-01 -2.80986309e-01
3.36010605e-01 -8.75346839e-01 5.54481387e-01 2.98160970e-01
-1.22891998e+00 -9.45620477e-01 -8.20491433e-01 -2.64419615e-01
-5.54631054e-01 4.18842345e-01 6.62613750e-01 4.13768202e-01
-9.23305929e-01 1.77864388e-01 -8.61830056e-01 -4.63835746e-01
6.93050325e-01 4.86559123e-01 -5.30482888e-01 -4.96518970e-01
-3.23631406e-01 9.82489645e-01 5.03851116e-01 1.56044394e-01
-1.04678714e+00 -6.65103793e-01 -1.16597867e+00 -1.91972122e-01
5.00084221e-01 -9.17762280e-01 1.00981998e+00 -5.70751727e-01
-1.18329334e+00 1.12038136e+00 -3.12606543e-01 -2.19221219e-01
3.15548152e-01 -3.75561327e-01 1.28682181e-01 3.55298579e-01
1.38967698e-02 6.02063119e-01 9.74009812e-01 -1.87337315e+00
-3.39032620e-01 -7.96825826e-01 2.40236297e-01 3.68747354e-01
2.25347608e-01 -2.92824924e-01 -7.59981096e-01 -1.02825865e-01
7.42784202e-01 -7.01374114e-01 -2.46756107e-01 7.46153235e-01
-3.95826280e-01 -2.41244927e-01 1.07976496e+00 -3.64191592e-01
3.81259441e-01 -2.19152117e+00 7.97733441e-02 1.80694148e-01
2.23936483e-01 3.99033487e-01 -1.62906006e-01 3.97177517e-01
1.82221293e-01 -2.86912829e-01 -2.87570208e-01 -8.47242534e-01
1.93175182e-01 6.86022162e-01 -1.42511457e-01 8.18369627e-01
4.80488032e-01 9.08394158e-01 -5.90656817e-01 -9.09606159e-01
9.21223342e-01 6.31879210e-01 -3.21399122e-01 5.18100321e-01
-3.89323235e-01 4.35038686e-01 -8.53833556e-01 9.22092438e-01
1.40791905e+00 -1.39875144e-01 -1.77465722e-01 -3.46959144e-01
-1.10713407e-01 1.09268889e-01 -1.48944283e+00 1.96127927e+00
-5.31841516e-01 2.58327186e-01 2.65835941e-01 -1.07075143e+00
9.54429805e-01 1.75928965e-01 8.01916361e-01 -4.25621450e-01
1.00060655e-02 1.27909914e-01 -5.12353837e-01 -4.97354627e-01
3.19671661e-01 1.83379889e-01 3.10350597e-01 1.67366296e-01
3.07644248e-01 -6.67103827e-01 4.99193817e-02 7.36550242e-02
5.63988686e-01 6.61856413e-01 6.65918589e-01 -6.40645325e-02
7.58759797e-01 1.17513239e-02 2.08371356e-01 8.18910420e-01
-1.42572829e-02 6.72897041e-01 1.09296404e-02 -1.34081826e-01
-8.75258446e-01 -1.30367482e+00 -6.00365281e-01 5.60543776e-01
5.92667937e-01 -1.52912900e-01 -4.19130594e-01 -6.44875407e-01
2.34407917e-01 4.99656171e-01 -4.47322249e-01 5.34964129e-02
-4.81136173e-01 -3.63726169e-01 -1.04796901e-01 4.69473094e-01
5.89511871e-01 -5.64715326e-01 -4.09434646e-01 -1.64410938e-02
-1.14437915e-01 -1.43238485e+00 -8.68789777e-02 2.62354404e-01
-1.12383449e+00 -1.34257424e+00 -6.02880359e-01 -8.48095834e-01
9.76184487e-01 7.51679540e-01 1.23459339e+00 -1.20704425e-02
-5.42373657e-01 1.03010774e+00 -5.16677380e-01 -5.77190161e-01
-1.57085493e-01 -5.79126894e-01 -4.60100807e-02 1.62532270e-01
2.40545094e-01 -5.63392341e-01 -5.23490191e-01 5.67439020e-01
-7.16169477e-01 3.46457809e-02 9.93602931e-01 5.03908575e-01
1.02475822e+00 5.74467368e-02 -3.79441008e-02 -8.25496018e-01
-4.42144722e-01 -2.72839785e-01 -7.49014199e-01 2.91836590e-01
-3.73797506e-01 -1.87395051e-01 1.68646917e-01 -4.83537108e-01
-1.15672553e+00 5.71257055e-01 1.18031040e-01 -7.37734497e-01
-5.38736463e-01 -3.39775607e-02 -8.10677469e-01 -4.02904302e-01
3.30758333e-01 5.70957601e-01 2.12869436e-01 -8.91104460e-01
4.40875590e-01 5.98567307e-01 3.31910729e-01 -4.77788866e-01
9.03213322e-01 9.39125001e-01 4.17161286e-01 -1.00214040e+00
-9.51462090e-01 -1.26226306e+00 -1.21963143e+00 -1.87594086e-01
5.96537113e-01 -1.11890638e+00 -4.09012616e-01 4.71640736e-01
-1.36866677e+00 3.90568711e-02 -5.04111111e-01 6.50002897e-01
-1.03140867e+00 5.66897511e-01 -1.82779729e-01 -1.32859612e+00
1.59799546e-01 -1.03814697e+00 1.71047282e+00 -4.07054648e-02
3.23758751e-01 -1.05461359e+00 -1.26411498e-01 2.84773111e-01
-5.69095947e-02 1.28639936e-01 8.18195999e-01 -6.87101543e-01
-1.22438025e+00 -3.72531056e-01 -3.07914138e-01 4.92434889e-01
-2.90085282e-02 -2.50245720e-01 -1.19785607e+00 -6.68802783e-02
3.94255549e-01 -1.61336601e-01 5.34482718e-01 6.68330073e-01
1.13658988e+00 -8.76479670e-02 -5.67542255e-01 5.26146054e-01
1.55104589e+00 -1.17193796e-02 5.29907048e-01 -1.67009145e-01
6.65137112e-01 7.94769287e-01 8.96541595e-01 5.05602241e-01
2.07220197e-01 8.95061612e-01 9.28376317e-01 -1.44183680e-01
-5.80301642e-01 -3.28925818e-01 3.60393487e-02 5.97803175e-01
-1.43334130e-02 -2.40911335e-01 -5.17373860e-01 4.14483845e-01
-1.84328663e+00 -7.09471703e-01 -4.88998234e-01 2.11180139e+00
2.56511658e-01 -4.13619056e-02 -2.15312596e-02 1.48717850e-01
7.32745409e-01 5.14181070e-02 -5.80830872e-01 3.18124801e-01
-2.31315374e-01 -9.67883021e-02 3.53602052e-01 2.48152182e-01
-1.22779262e+00 7.54307687e-01 6.31242132e+00 9.48286593e-01
-4.88418818e-01 -1.38771413e-02 1.72959805e-01 5.13441265e-01
-4.91054356e-02 2.51712382e-01 -1.03293347e+00 8.10111463e-02
2.12940127e-01 2.52228081e-01 1.67159978e-02 1.19775665e+00
6.40793815e-02 -3.64086330e-01 -1.40427208e+00 1.20829058e+00
4.31997180e-01 -1.01529229e+00 3.10769439e-01 2.96835929e-01
5.96763790e-01 -2.44864702e-01 -1.47249237e-01 1.69546679e-01
-6.27457798e-02 -5.79424679e-01 9.10950959e-01 6.80253446e-01
3.68162662e-01 -3.04357857e-01 6.39279604e-01 7.96115637e-01
-1.23250377e+00 1.34732440e-01 -8.69769931e-01 1.25678822e-01
1.79732159e-01 7.16848314e-01 -9.84206855e-01 8.67249131e-01
3.53582978e-01 9.44686174e-01 -5.92069387e-01 1.51216841e+00
-5.29567413e-02 2.39971802e-01 -4.17420745e-01 2.61297077e-01
-2.12254211e-01 -2.53684372e-01 7.32644796e-01 9.60934699e-01
1.26599535e-01 1.97308987e-01 5.24568677e-01 1.19045091e+00
2.30453670e-01 -1.38986573e-01 -6.94093823e-01 2.56843686e-01
2.99302548e-01 1.28241086e+00 -8.59033406e-01 -2.77545363e-01
-5.11517763e-01 8.19975734e-01 3.44834477e-02 3.36424768e-01
-5.81465483e-01 3.04317661e-03 5.41082799e-01 2.65386134e-01
4.76820886e-01 -5.38435042e-01 -1.90529406e-01 -1.03312182e+00
8.22628215e-02 -2.48887837e-01 3.03695388e-02 -1.03727174e+00
-1.43231881e+00 1.96427226e-01 5.78264058e-01 -1.61109555e+00
-7.34697506e-02 -1.19816685e+00 -3.61082524e-01 7.63628900e-01
-1.54799879e+00 -1.72594643e+00 -2.70983696e-01 7.13789284e-01
7.60492027e-01 1.72019109e-01 7.03708112e-01 1.45007640e-01
-7.62977675e-02 -4.29179240e-03 1.14228360e-01 5.73611669e-02
2.65307337e-01 -1.22611368e+00 1.45510405e-01 6.97399318e-01
5.54401159e-01 5.95433176e-01 5.63822389e-01 -4.97965008e-01
-1.76369476e+00 -1.01167464e+00 5.84476233e-01 -8.22290242e-01
1.69506133e-01 -6.71141028e-01 -6.43508554e-01 4.62680727e-01
-5.57112873e-01 3.13573658e-01 4.23599750e-01 -2.41593346e-02
-3.61731946e-01 2.67707799e-02 -1.23090279e+00 1.58542752e-01
1.32591629e+00 -6.82652891e-01 -6.36476398e-01 6.51383281e-01
5.20549953e-01 -4.93789971e-01 -6.66259825e-01 5.77025294e-01
3.67242426e-01 -9.62471604e-01 1.60304821e+00 -3.89022857e-01
-1.15687817e-01 -4.13077354e-01 -6.43740237e-01 -8.15373898e-01
-4.27166164e-01 -3.73913050e-02 -4.60532904e-01 1.24416792e+00
-1.99225143e-01 -2.82047212e-01 6.88460290e-01 4.10676181e-01
-5.65136552e-01 -5.73307157e-01 -9.07667458e-01 -8.48320842e-01
-4.86738086e-01 -6.27594829e-01 4.20902163e-01 3.46962363e-01
-7.34123409e-01 5.69135845e-02 1.44085344e-02 4.87978697e-01
9.03910518e-01 4.96668160e-01 1.01454997e+00 -1.61554313e+00
-1.25663489e-01 -2.26507038e-01 -9.65764642e-01 -1.62750292e+00
3.78825456e-01 -7.42895067e-01 1.87313676e-01 -1.79000556e+00
4.64372188e-01 -4.46393460e-01 -9.10028350e-03 1.61325052e-01
1.51286170e-01 4.25478399e-01 1.42198488e-01 4.46619987e-01
-8.62557948e-01 6.91119671e-01 1.39093113e+00 -2.65603483e-01
1.59709886e-01 4.37520772e-01 -3.67909729e-01 9.75905240e-01
1.03766978e-01 -2.88794696e-01 -5.11255085e-01 -3.43320668e-01
-3.63285542e-01 -1.58480480e-01 8.53858232e-01 -8.07892323e-01
1.74400136e-01 -1.77380621e-01 6.82671249e-01 -1.31937730e+00
1.00245714e+00 -1.27774811e+00 5.23764677e-02 1.74581647e-01
-9.85945091e-02 -5.83647609e-01 3.23418118e-02 8.08093548e-01
-2.00005025e-01 -4.56013054e-01 7.37279594e-01 -3.82786810e-01
-9.45112884e-01 6.28534436e-01 -8.83391723e-02 -2.26246789e-01
9.81589556e-01 -7.56721437e-01 1.28270313e-01 -1.13991305e-01
-7.90959716e-01 -1.14468122e-02 5.53981185e-01 8.89284462e-02
9.38183308e-01 -1.35587335e+00 -4.94817466e-01 4.86343533e-01
4.89626855e-01 5.61875701e-01 4.86121356e-01 9.13304031e-01
-3.38879049e-01 5.71534634e-01 -7.69603206e-03 -1.18573880e+00
-1.06647754e+00 8.91548753e-01 8.63814801e-02 2.47959286e-01
-6.05243027e-01 6.54924572e-01 7.23666251e-01 -6.33045852e-01
3.26089859e-01 -3.54895800e-01 -1.13229781e-01 -4.86502945e-02
2.81824172e-01 3.40007812e-01 6.89116418e-02 -1.02650166e+00
-4.97869879e-01 1.05001616e+00 8.76214951e-02 1.44790381e-01
1.16152346e+00 -4.80564356e-01 -1.27814412e-01 5.57468951e-01
1.21876872e+00 -5.75686693e-02 -1.41783702e+00 -6.07153714e-01
-9.80867594e-02 -8.74329209e-01 9.71197560e-02 -4.93983269e-01
-6.82832658e-01 9.71226752e-01 5.55169523e-01 -2.42719408e-02
9.51659203e-01 5.97303987e-01 1.28841251e-01 4.79446411e-01
7.47907996e-01 -6.85258567e-01 1.98414177e-01 3.36808950e-01
8.41641009e-01 -1.20369184e+00 4.09800410e-01 -8.78753364e-01
-2.18133435e-01 1.02407479e+00 4.56921607e-01 -2.38359556e-01
7.50067830e-01 -6.26014248e-02 -2.92526186e-01 -2.50963360e-01
-4.48779106e-01 -6.56943619e-01 7.62061417e-01 1.06557381e+00
-1.97749622e-02 -1.14087932e-01 2.22055793e-01 2.97470778e-01
3.53676528e-01 -4.12055224e-01 1.06476054e-01 9.05498743e-01
-5.14803290e-01 -1.09269011e+00 -5.80314517e-01 1.22849599e-01
7.00699016e-02 2.57233530e-01 -4.98919606e-01 1.10640585e+00
3.64460796e-01 7.48329759e-01 1.39572397e-01 1.28671050e-01
3.11159045e-01 -3.18343490e-02 9.34537947e-01 -8.28789473e-01
1.32016078e-01 3.15390289e-01 -2.66341090e-01 -6.34344935e-01
-9.26599026e-01 -6.63821638e-01 -9.62474227e-01 3.36407274e-01
-7.60609865e-01 -2.81819385e-02 9.97157633e-01 9.21241283e-01
2.44113848e-01 1.28113896e-01 7.49877155e-01 -1.35428202e+00
-5.98502874e-01 -8.42144012e-01 -9.75640297e-01 3.36376041e-01
3.97032917e-01 -1.17839909e+00 -4.68151242e-01 1.06963009e-01] | [7.6238861083984375, -2.7591769695281982] |
6da9c1c5-f4bf-4e62-a131-417b4a9981bf | utilization-of-domain-knowledge-to-improve | 2302.08748 | null | https://arxiv.org/abs/2302.08748v1 | https://arxiv.org/pdf/2302.08748v1.pdf | Utilization of domain knowledge to improve POMDP belief estimation | The partially observable Markov decision process (POMDP) framework is a common approach for decision making under uncertainty. Recently, multiple studies have shown that by integrating relevant domain knowledge into POMDP belief estimation, we can improve the learned policy's performance. In this study, we propose a novel method for integrating the domain knowledge into probabilistic belief update in POMDP framework using Jeffrey's rule and normalization. We show that the domain knowledge can be utilized to reduce the data requirement and improve performance for POMDP policy learning with RL. | ['Johane Takeuchi', 'Tung Nguyen'] | 2023-02-17 | null | null | null | null | ['decision-making-under-uncertainty', 'decision-making-under-uncertainty'] | ['medical', 'reasoning'] | [-2.15743527e-01 2.69610852e-01 -6.02686942e-01 -5.22323966e-01
-1.02662241e+00 -5.01095474e-01 6.53378189e-01 1.94401965e-01
-7.73202360e-01 1.38804638e+00 6.58795059e-01 -2.73354888e-01
-1.91832021e-01 -8.82882357e-01 -3.37665915e-01 -7.22094119e-01
8.46510753e-03 6.27570331e-01 6.28405690e-01 -8.04735720e-02
3.21819335e-01 4.02652532e-01 -1.02888215e+00 3.10843326e-02
4.23476666e-01 9.03577447e-01 4.38136876e-01 6.50092602e-01
-1.45323455e-01 1.31794500e+00 -6.94212854e-01 -7.69327804e-02
4.65147644e-01 -2.52657086e-01 -7.67016947e-01 -2.17321292e-01
-7.64930665e-01 -9.33042943e-01 -4.15452838e-01 1.16654253e+00
6.15349770e-01 7.55646169e-01 6.88896954e-01 -1.11102629e+00
-1.98169604e-01 8.09079051e-01 -4.40567911e-01 2.75044799e-01
5.44981897e-01 3.64621319e-02 6.79070890e-01 -2.44105570e-02
5.07813931e-01 1.97205675e+00 2.01594949e-01 5.31337142e-01
-5.55089474e-01 -5.17746091e-01 4.98870015e-01 5.95668495e-01
-1.12931502e+00 -6.96287975e-02 7.49268889e-01 2.27142721e-01
7.98574865e-01 -1.00540675e-01 7.62886345e-01 9.61333930e-01
7.98544526e-01 1.16529453e+00 1.53061450e+00 -4.23574030e-01
9.11704361e-01 -3.49146813e-01 -6.21660724e-02 3.55308890e-01
4.06957418e-01 4.34315622e-01 -2.74758726e-01 -4.68809724e-01
8.56706858e-01 1.35743514e-01 4.10289168e-01 -2.38985702e-01
-6.43405974e-01 1.00852740e+00 -2.17419520e-01 -2.63701528e-01
-7.72404253e-01 2.70289809e-01 2.51675516e-01 2.52103925e-01
7.51994327e-02 1.19760372e-01 -3.62030089e-01 -7.19312489e-01
-2.81401187e-01 6.72637880e-01 1.02191365e+00 8.40110302e-01
5.13543904e-01 -5.47789596e-02 -5.26981711e-01 1.93234295e-01
9.48588550e-01 9.39758420e-01 5.05382061e-01 -1.59102678e+00
4.36878443e-01 3.55773419e-01 9.27942038e-01 -4.56344604e-01
-3.57935607e-01 3.88251126e-01 -7.65705556e-02 5.69218278e-01
1.45382002e-01 -5.96327305e-01 -9.75628436e-01 1.54418683e+00
6.16345942e-01 -9.69796330e-02 6.01633072e-01 7.12325275e-01
2.51299441e-01 7.14649796e-01 1.99257508e-01 -4.54190344e-01
1.13538253e+00 -4.94018525e-01 -1.23995519e+00 -3.07300296e-02
-1.75141040e-02 -5.01142085e-01 5.80557227e-01 6.82802737e-01
-9.80348945e-01 -1.88237026e-01 -1.12013769e+00 1.66875049e-01
1.89524055e-01 -4.86209631e-01 8.60983849e-01 8.51402164e-01
-6.43144250e-01 6.82246387e-01 -1.44656873e+00 -3.03203110e-02
2.56533235e-01 4.63792682e-01 2.62889713e-01 -2.33238742e-01
-1.41751218e+00 1.40523803e+00 1.04728472e+00 -5.81248820e-01
-1.66293538e+00 4.07191902e-01 -6.01966083e-01 -1.13538228e-01
8.64421666e-01 -4.68775600e-01 1.81454349e+00 -1.29159326e-02
-2.43848515e+00 -2.42785156e-01 -4.16293107e-02 -9.75025356e-01
4.95951861e-01 -3.53073299e-01 -4.13472176e-01 4.06460375e-01
-1.44214645e-01 4.53213632e-01 9.73566949e-01 -9.86986220e-01
-1.06219423e+00 -3.56562853e-01 5.44790626e-01 9.76584733e-01
3.63337159e-01 6.31642193e-02 -5.96032776e-02 -9.71423909e-02
2.62946486e-01 -9.45184290e-01 -6.44928277e-01 -4.67665672e-01
1.57473713e-01 -4.46440995e-01 7.31546283e-01 -4.91554260e-01
9.34724748e-01 -1.85567784e+00 -1.63034588e-01 1.63446128e-01
-7.30528608e-02 4.45143431e-02 2.56529927e-01 5.94602525e-01
7.55015731e-01 -2.58137614e-01 1.07836770e-02 -1.80649623e-01
4.29193735e-01 1.00183344e+00 -4.23050791e-01 4.58426744e-01
-2.31794432e-01 6.04207814e-01 -9.91031468e-01 -6.28336847e-01
4.80214745e-01 6.75628632e-02 -3.36130083e-01 3.08955610e-01
-5.47858357e-01 7.21612692e-01 -1.01973081e+00 4.86424804e-01
5.74365556e-01 3.52860987e-01 4.88659084e-01 5.18673718e-01
-1.03180438e-01 5.68593204e-01 -1.74132228e+00 1.59518385e+00
-1.73575822e-02 -6.48670271e-02 -7.86905736e-02 -5.74649513e-01
8.53192031e-01 6.87349677e-01 5.30203938e-01 -3.45433891e-01
4.02423382e-01 -3.55303645e-01 7.82633480e-03 -2.95089334e-01
5.70617616e-01 -5.67219317e-01 -4.42400813e-01 8.36525619e-01
9.78448614e-02 -5.53622961e-01 -8.89508277e-02 -2.25641191e-01
1.14721107e+00 2.89021850e-01 1.11107183e+00 1.57633331e-02
1.46226615e-01 1.53195351e-01 9.98612225e-01 1.11703289e+00
-8.63756061e-01 -9.75902677e-02 4.18854177e-01 -2.96699584e-01
-5.98622143e-01 -1.20203424e+00 5.34564517e-02 1.20012176e+00
3.48574936e-01 -9.02197435e-02 -4.57169324e-01 -6.00391448e-01
-1.55808032e-02 1.13923836e+00 -3.49393457e-01 -2.06943542e-01
-2.58713841e-01 -7.82406092e-01 3.70275110e-01 5.33896148e-01
7.49841213e-01 -9.10169244e-01 -7.08869159e-01 5.85672677e-01
2.97215357e-02 -6.58312917e-01 3.07536364e-01 2.78851867e-01
-1.10265720e+00 -7.20837295e-01 -5.16998887e-01 1.66192129e-01
1.07178010e-01 2.25206494e-01 5.44114828e-01 -9.13153529e-01
6.71325922e-01 7.39426970e-01 -5.88346839e-01 -1.09049189e+00
-4.59221303e-01 -3.61532539e-01 4.78687674e-01 -6.28164530e-01
5.32067120e-01 -5.09082675e-01 -3.86218280e-01 2.03482494e-01
-8.15037668e-01 -2.22478986e-01 5.72564423e-01 5.45787573e-01
6.27621114e-01 5.63782513e-01 2.81648189e-01 -4.27032292e-01
1.17184591e+00 -4.13889080e-01 -8.55344057e-01 9.89365280e-02
-5.51866412e-01 6.42319083e-01 5.55995256e-02 -2.89175272e-01
-1.89375150e+00 4.67604175e-02 -1.08716115e-01 -5.05128391e-02
-3.47296834e-01 3.79700243e-01 -3.44238400e-01 -1.65688530e-01
3.34364146e-01 1.21818937e-01 -1.42244920e-01 -5.61591625e-01
4.44560081e-01 6.89600289e-01 1.84490457e-01 -9.75007534e-01
2.16067627e-01 8.59798133e-01 2.16802880e-01 -1.63188607e-01
-9.96038675e-01 -4.15951878e-01 -4.74751204e-01 -1.35115638e-01
8.63555908e-01 -9.72315252e-01 -1.08314073e+00 -2.09411997e-02
-9.01679754e-01 -1.56258076e-01 -4.92266119e-01 1.18060851e+00
-1.06672156e+00 5.04649162e-01 -5.78571737e-01 -1.49677002e+00
-1.07602812e-01 -1.15646923e+00 5.79907477e-01 4.09162611e-01
-9.28183421e-02 -9.41825867e-01 5.51751971e-01 9.11707133e-02
-6.56878129e-02 2.22860724e-02 4.26122636e-01 -7.20042408e-01
-4.24352556e-01 -4.00573276e-02 1.12867929e-01 2.17074275e-01
-9.94108692e-02 -5.03935814e-01 -6.81915581e-01 -2.91096449e-01
7.91255474e-01 -1.31807446e-01 5.16867936e-01 4.33189571e-01
4.34278041e-01 -3.56982499e-01 -1.33535638e-01 -1.13449402e-01
1.35104585e+00 8.94734025e-01 4.77200329e-01 4.61313665e-01
7.72777572e-02 1.43291667e-01 1.28351974e+00 1.10573208e+00
6.76081300e-01 3.61790389e-01 4.98364151e-01 8.60234857e-01
4.16741759e-01 -4.78295118e-01 6.89701021e-01 5.75164497e-01
-5.56082666e-01 -1.03857331e-01 -8.79474878e-01 1.09591261e-01
-2.39187932e+00 -1.05647159e+00 3.89948338e-01 1.95456398e+00
7.94108748e-01 4.08255309e-01 6.62623420e-02 -2.49170903e-02
5.90270460e-01 1.34977266e-01 -7.57425189e-01 -3.63807321e-01
2.30169028e-01 2.94055678e-02 9.69445765e-01 6.97829902e-01
-1.05982673e+00 1.02494287e+00 7.71265030e+00 7.57412791e-01
-4.49046850e-01 6.23850644e-01 -3.18765610e-01 -7.81159028e-02
-1.08208902e-01 3.09410453e-01 -1.22125411e+00 2.78507918e-01
1.03258646e+00 -5.29074907e-01 3.81700665e-01 1.05747187e+00
3.83483142e-01 -7.70549238e-01 -6.56912804e-01 7.11861789e-01
-3.92352313e-01 -7.18496859e-01 -9.59621519e-02 6.79783002e-02
7.99921632e-01 4.01924923e-02 -2.68365204e-01 7.44364679e-01
1.84942317e+00 -5.98821580e-01 7.72254705e-01 6.96700215e-01
-8.79102945e-02 -1.14715040e+00 9.67399895e-01 7.55936980e-01
-9.70720589e-01 -4.72373635e-01 -9.47912753e-01 -8.22735727e-01
6.21625662e-01 2.04441249e-01 -1.11505473e+00 6.53301895e-01
3.64478141e-01 3.90118688e-01 1.33321121e-01 9.65842485e-01
-9.83471572e-01 6.93403006e-01 -5.26904047e-01 -2.74900407e-01
5.92237115e-01 -1.46954075e-01 7.84788370e-01 4.96586680e-01
2.11353451e-01 5.62009931e-01 6.66742086e-01 3.09763402e-01
4.87913698e-01 -2.82237262e-01 -5.19630790e-01 -5.43668978e-02
6.01837575e-01 3.88579994e-01 -5.73281884e-01 -5.45703411e-01
-2.18279868e-01 6.31957233e-01 9.87562910e-02 2.27455243e-01
-8.26748729e-01 1.89867541e-01 5.99267542e-01 -6.60565376e-01
2.40212858e-01 -5.44056237e-01 -1.27945200e-01 -9.01198566e-01
-4.57818896e-01 -5.63532650e-01 7.17675924e-01 -6.73294067e-01
-1.03113317e+00 -1.28919393e-01 6.10891044e-01 -1.24091661e+00
-5.54832399e-01 -4.25999284e-01 -2.52509743e-01 6.37074411e-01
-1.41226053e+00 -6.22534156e-01 4.89086360e-01 7.60798931e-01
3.41662019e-01 -1.30578443e-01 9.72400546e-01 -3.00809890e-01
-1.38036445e-01 -2.91556388e-01 5.13287842e-01 -1.70147493e-01
5.04309952e-01 -1.32225013e+00 1.10031376e-02 6.27607107e-01
-2.13639095e-01 4.40131158e-01 9.85753477e-01 -1.08432400e+00
-1.44809222e+00 -6.10846817e-01 -6.83656782e-02 -4.05547231e-01
5.64213634e-01 2.73380876e-01 -5.02968907e-01 8.61868799e-01
1.43012807e-01 -6.56355500e-01 7.01414645e-01 1.94770340e-02
2.60389954e-01 2.07613364e-01 -1.59243059e+00 6.30413115e-01
6.07892096e-01 -2.43420213e-01 -1.53183687e+00 -2.95607820e-02
8.09185684e-01 -6.11947954e-01 -1.03671348e+00 3.40466231e-01
5.90135396e-01 -5.79554200e-01 8.85440767e-01 -8.27858269e-01
-1.92610279e-01 -4.02662128e-01 -5.34710824e-01 -1.28314781e+00
-3.24837387e-01 -7.26280272e-01 -5.79827487e-01 7.83897698e-01
-9.92514715e-02 -6.78001642e-01 7.17084706e-01 9.75133002e-01
9.84518602e-02 -1.00315869e-01 -1.33823121e+00 -9.39624846e-01
3.53290670e-04 -7.14787722e-01 8.45597267e-01 3.51727217e-01
2.36407235e-01 6.18814677e-02 -5.87781012e-01 3.17885995e-01
6.62333131e-01 -1.08374469e-01 5.70517719e-01 -1.10556972e+00
-5.53108692e-01 2.20597848e-01 -3.41432959e-01 -1.19287646e+00
-3.27833928e-02 -1.85389012e-01 3.22843969e-01 -1.86811042e+00
1.36097163e-01 -6.98945448e-02 -7.51489937e-01 3.72752726e-01
-1.62672624e-01 -7.48078346e-01 5.68686962e-01 3.30521166e-01
-1.10773587e+00 8.05558801e-01 1.69215965e+00 1.55745178e-01
-3.36588353e-01 4.04287815e-01 -5.47993898e-01 1.10329962e+00
1.07998121e+00 -9.66749132e-01 -5.66164970e-01 -5.68878315e-02
2.33891651e-01 7.64899850e-01 -2.56263018e-01 -1.14928007e+00
2.31809914e-01 -9.77995992e-01 2.74940014e-01 -8.98719013e-01
6.99003577e-01 -9.59415615e-01 2.09276024e-02 5.64694464e-01
-2.37483144e-01 -8.76344442e-02 1.58313721e-01 1.12958026e+00
-6.09097332e-02 -6.97502375e-01 5.45856118e-01 -6.43716753e-01
-1.11669528e+00 5.44448569e-02 -1.01691937e+00 -2.10004285e-01
1.14178193e+00 2.59004384e-01 -4.56110016e-02 -9.32016969e-01
-9.55496609e-01 3.04231524e-01 2.84688622e-01 1.37783922e-02
5.20773351e-01 -1.27659798e+00 -2.20147088e-01 -3.76977265e-01
-5.20956695e-01 -5.92695847e-02 7.42702931e-03 4.20575500e-01
-2.25630000e-01 5.09671211e-01 -6.92728996e-01 -1.79042563e-01
-8.53352070e-01 5.03064275e-01 -6.51101023e-02 -7.68750906e-01
-4.70946848e-01 4.50531155e-01 -4.37961042e-01 -4.39020723e-01
3.42947602e-01 -4.32795703e-01 -5.24852872e-01 -1.18859001e-01
8.20163071e-01 7.72880971e-01 -5.09424627e-01 -1.55505583e-01
-8.47867504e-03 -7.99420774e-02 7.46235773e-02 -1.14159489e+00
1.12258768e+00 -6.51925325e-01 2.63417125e-01 8.15903425e-01
2.89945334e-01 -2.59732813e-01 -1.83043861e+00 -3.13365430e-01
2.65138865e-01 -5.06169319e-01 1.78614035e-01 -6.82275295e-01
-7.26837739e-02 7.20826030e-01 7.80887902e-01 3.83746065e-02
6.09386861e-01 -2.41442844e-01 5.86920738e-01 1.02266538e+00
1.17667782e+00 -1.49520338e+00 -3.25499982e-01 9.18998361e-01
4.77161974e-01 -1.34418070e+00 4.56963748e-01 -1.71283185e-02
-1.20129538e+00 8.34845304e-01 1.17975116e-01 -1.45212829e-01
6.31702960e-01 3.33850920e-01 6.08722568e-02 3.35717434e-03
-8.85502517e-01 -6.03014350e-01 -1.84523046e-01 9.48627234e-01
-3.99061471e-01 6.08214617e-01 -4.31248099e-01 9.15700674e-01
-1.07888088e-01 4.50976044e-01 5.41933596e-01 1.77516973e+00
-1.29063296e+00 -1.34462583e+00 -8.60186875e-01 3.32713306e-01
-5.88814914e-01 2.52395928e-01 -2.92728364e-01 4.12995040e-01
-1.69879481e-01 1.14547431e+00 -2.50903726e-01 -8.75709802e-02
1.36707544e-01 2.25482553e-01 7.68067002e-01 -7.56272078e-01
5.07372580e-02 2.99620945e-02 1.85725093e-01 -5.33759773e-01
-6.08598709e-01 -7.08979964e-01 -1.68887818e+00 -2.50037968e-01
2.69486755e-03 2.76235729e-01 6.59819126e-01 1.37208104e+00
-1.03193084e-02 4.04677272e-01 2.82107234e-01 -6.35462046e-01
-1.27619433e+00 -1.01903284e+00 -9.16772127e-01 -4.08972800e-02
-1.53254032e-01 -1.21266508e+00 -6.19190894e-02 -3.15920115e-01] | [4.264586925506592, 2.1645302772521973] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.